codeShare commited on
Commit
cfaeb00
·
verified ·
1 Parent(s): dd1023b

Upload Kromoss dual-T4 training log

Browse files
Files changed (1) hide show
  1. training_logs/kromoss_training_log.txt +1225 -0
training_logs/kromoss_training_log.txt ADDED
@@ -0,0 +1,1225 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2026-11-01 20:54:56 | INFO | ================================================================================
2
+ 2026-11-01 20:54:56 | INFO | KROMOSS DUAL-T4 SEQUENTIAL TWO-STAGE TRAINING
3
+ 2026-11-01 20:54:56 | INFO | ================================================================================
4
+ 2026-11-01 20:54:56 | INFO | PyTorch version: 2.11.0+cu128
5
+ 2026-11-01 20:54:56 | INFO | CUDA version: 12.8
6
+ 2026-11-01 20:54:56 | INFO | Dataset size: 7
7
+ 2026-11-01 20:54:56 | INFO | Repeats per image: 10
8
+ 2026-11-01 20:54:56 | INFO | Total optimizer updates: 70
9
+ 2026-11-01 20:54:56 | INFO | Checkpoint steps: [14, 35, 70]
10
+ 2026-11-01 20:54:56 | INFO | Initial learning rate: 2.00000000e-04
11
+ 2026-11-01 20:54:56 | INFO | Final learning rate per image: 1.00000000e-04
12
+ 2026-11-01 20:54:56 | INFO | Weight decay: 0.01
13
+ 2026-11-01 20:54:56 | INFO | Gradient clipping: 1.0
14
+ 2026-11-01 20:54:56 | INFO | Loss scale: 128.0
15
+ 2026-11-01 20:54:56 | INFO | LoRA rank: 16
16
+ 2026-11-01 20:54:56 | INFO | LoRA alpha: 16
17
+ 2026-11-01 20:54:56 | INFO | Trainable parameters: 37,158,912
18
+ 2026-11-01 20:54:56 | INFO | GPU0 LoRA parameters: 17,694,720
19
+ 2026-11-01 20:54:56 | INFO | GPU1 LoRA parameters: 19,464,192
20
+ 2026-11-01 20:54:56 | INFO | Continuation loaded: False
21
+ 2026-11-01 20:54:56 | INFO | HF repository: codeShare/kromoss-lora-exports
22
+ 2026-11-01 20:54:56 | INFO | HF log destination: training_logs/kromoss_training_log.txt
23
+ 2026-11-01 20:54:56 | INFO | GPU0: Tesla T4 | 14.562 GB
24
+ 2026-11-01 20:54:56 | INFO | GPU1: Tesla T4 | 14.562 GB
25
+ 2026-11-01 20:54:56 | INFO | DATASET ITEM 0: 00000000.pt | repeats=10
26
+ 2026-11-01 20:54:56 | INFO | DATASET ITEM 1: 00000001.pt | repeats=10
27
+ 2026-11-01 20:54:56 | INFO | DATASET ITEM 2: 00000002.pt | repeats=10
28
+ 2026-11-01 20:54:56 | INFO | DATASET ITEM 3: 00000003.pt | repeats=10
29
+ 2026-11-01 20:54:56 | INFO | DATASET ITEM 4: 00000004.pt | repeats=10
30
+ 2026-11-01 20:54:56 | INFO | DATASET ITEM 5: 00000005.pt | repeats=10
31
+ 2026-11-01 20:54:56 | INFO | DATASET ITEM 6: 00000006.pt | repeats=10
32
+ 2026-11-01 20:54:56 | INFO | INITIAL MODEL STATE | GPU0 | allocated=4.856 GB | reserved=4.873 GB | peak=4.856 GB | free=9.587 GB
33
+ 2026-11-01 20:54:56 | INFO | INITIAL MODEL STATE | GPU1 | allocated=4.749 GB | reserved=4.762 GB | peak=4.749 GB | free=9.698 GB
34
+ 2026-11-01 20:54:56 | INFO | ================================================================================
35
+ 2026-11-01 20:54:56 | INFO | STARTING TRAINING
36
+ 2026-11-01 20:54:56 | INFO | ================================================================================
37
+ 2026-11-01 20:54:56 | INFO | Schedule: seven embeddings, ten consecutive repeats each, one optimizer update per presentation.
38
+ 2026-11-01 20:54:56 | INFO | Learning rate: 2e-4 to 1e-4 over each embedding's ten repeats; resets for the next embedding.
39
+ 2026-11-01 20:54:56 | INFO | AdamW optimizer state is retained across all seven embeddings.
40
+ 2026-11-01 20:54:56 | INFO | No gradient accumulation and no 1F1B scheduling.
41
+ 2026-11-01 20:54:56 | INFO | --------------------------------------------------------------------------------
42
+ 2026-11-01 20:54:56 | INFO | STARTING DATASET ITEM 1/7 | file=00000000.pt
43
+ 2026-11-01 20:54:56 | INFO | LEARNING RATE RESET: 2.00000000e-04
44
+ 2026-11-01 20:55:12 | INFO | --------------------------------------------------------------------------------
45
+ 2026-11-01 20:55:12 | INFO | STEP 1/70 | dataset_item=0 | repeat=1/10
46
+ 2026-11-01 20:55:12 | INFO | LOSS: 1.35181808
47
+ 2026-11-01 20:55:12 | INFO | LR: 2.00000000e-04
48
+ 2026-11-01 20:55:12 | INFO | GLOBAL GRADIENT NORM: 0.75836104
49
+ 2026-11-01 20:55:12 | INFO | MAX ABSOLUTE GRADIENT: 2.72216797e-02
50
+ 2026-11-01 20:55:12 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
51
+ 2026-11-01 20:55:12 | INFO | MISSING GRADIENT TENSORS: 0
52
+ 2026-11-01 20:55:12 | INFO | GPU0 FORWARD: 2.735s
53
+ 2026-11-01 20:55:12 | INFO | GPU1 FORWARD: 2.313s
54
+ 2026-11-01 20:55:12 | INFO | GPU1 BACKWARD: 5.778s
55
+ 2026-11-01 20:55:12 | INFO | GPU0 BACKWARD: 5.236s
56
+ 2026-11-01 20:55:12 | INFO | UPDATE WALL TIME: 5.592s
57
+ 2026-11-01 20:55:12 | INFO | ELAPSED TRAINING TIME: 16.55s
58
+ 2026-11-01 20:55:12 | INFO | STEP 1 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.367 GB | free=3.850 GB
59
+ 2026-11-01 20:55:12 | INFO | STEP 1 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.274 GB | free=3.886 GB
60
+ 2026-11-01 20:55:28 | INFO | --------------------------------------------------------------------------------
61
+ 2026-11-01 20:55:28 | INFO | STEP 2/70 | dataset_item=0 | repeat=2/10
62
+ 2026-11-01 20:55:28 | INFO | LOSS: 1.32309210
63
+ 2026-11-01 20:55:28 | INFO | LR: 1.88888889e-04
64
+ 2026-11-01 20:55:28 | INFO | GLOBAL GRADIENT NORM: 1.07154055
65
+ 2026-11-01 20:55:28 | INFO | MAX ABSOLUTE GRADIENT: 4.19921875e-02
66
+ 2026-11-01 20:55:28 | INFO | GRADIENT CLIP COEFFICIENT: 0.93323493
67
+ 2026-11-01 20:55:28 | INFO | MISSING GRADIENT TENSORS: 0
68
+ 2026-11-01 20:55:28 | INFO | GPU0 FORWARD: 2.028s
69
+ 2026-11-01 20:55:28 | INFO | GPU1 FORWARD: 2.285s
70
+ 2026-11-01 20:55:28 | INFO | GPU1 BACKWARD: 5.827s
71
+ 2026-11-01 20:55:28 | INFO | GPU0 BACKWARD: 5.317s
72
+ 2026-11-01 20:55:28 | INFO | UPDATE WALL TIME: 5.505s
73
+ 2026-11-01 20:55:28 | INFO | ELAPSED TRAINING TIME: 32.51s
74
+ 2026-11-01 20:55:28 | INFO | STEP 2 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.499 GB | free=3.850 GB
75
+ 2026-11-01 20:55:28 | INFO | STEP 2 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.419 GB | free=3.886 GB
76
+ 2026-11-01 20:55:44 | INFO | --------------------------------------------------------------------------------
77
+ 2026-11-01 20:55:44 | INFO | STEP 3/70 | dataset_item=0 | repeat=3/10
78
+ 2026-11-01 20:55:44 | INFO | LOSS: 1.21229815
79
+ 2026-11-01 20:55:44 | INFO | LR: 1.77777778e-04
80
+ 2026-11-01 20:55:44 | INFO | GLOBAL GRADIENT NORM: 0.45528601
81
+ 2026-11-01 20:55:44 | INFO | MAX ABSOLUTE GRADIENT: 1.65100098e-02
82
+ 2026-11-01 20:55:44 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
83
+ 2026-11-01 20:55:44 | INFO | MISSING GRADIENT TENSORS: 0
84
+ 2026-11-01 20:55:44 | INFO | GPU0 FORWARD: 2.049s
85
+ 2026-11-01 20:55:44 | INFO | GPU1 FORWARD: 2.285s
86
+ 2026-11-01 20:55:44 | INFO | GPU1 BACKWARD: 5.913s
87
+ 2026-11-01 20:55:44 | INFO | GPU0 BACKWARD: 5.334s
88
+ 2026-11-01 20:55:44 | INFO | UPDATE WALL TIME: 5.519s
89
+ 2026-11-01 20:55:44 | INFO | ELAPSED TRAINING TIME: 48.59s
90
+ 2026-11-01 20:55:44 | INFO | STEP 3 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.499 GB | free=3.850 GB
91
+ 2026-11-01 20:55:44 | INFO | STEP 3 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.419 GB | free=3.886 GB
92
+ 2026-11-01 20:56:01 | INFO | --------------------------------------------------------------------------------
93
+ 2026-11-01 20:56:01 | INFO | STEP 4/70 | dataset_item=0 | repeat=4/10
94
+ 2026-11-01 20:56:01 | INFO | LOSS: 1.08296394
95
+ 2026-11-01 20:56:01 | INFO | LR: 1.66666667e-04
96
+ 2026-11-01 20:56:01 | INFO | GLOBAL GRADIENT NORM: 0.27745513
97
+ 2026-11-01 20:56:01 | INFO | MAX ABSOLUTE GRADIENT: 7.41577148e-03
98
+ 2026-11-01 20:56:01 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
99
+ 2026-11-01 20:56:01 | INFO | MISSING GRADIENT TENSORS: 0
100
+ 2026-11-01 20:56:01 | INFO | GPU0 FORWARD: 2.068s
101
+ 2026-11-01 20:56:01 | INFO | GPU1 FORWARD: 2.313s
102
+ 2026-11-01 20:56:01 | INFO | GPU1 BACKWARD: 6.074s
103
+ 2026-11-01 20:56:01 | INFO | GPU0 BACKWARD: 5.400s
104
+ 2026-11-01 20:56:01 | INFO | UPDATE WALL TIME: 5.580s
105
+ 2026-11-01 20:56:01 | INFO | ELAPSED TRAINING TIME: 64.94s
106
+ 2026-11-01 20:56:01 | INFO | STEP 4 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.499 GB | free=3.850 GB
107
+ 2026-11-01 20:56:01 | INFO | STEP 4 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.419 GB | free=3.886 GB
108
+ 2026-11-01 20:56:17 | INFO | --------------------------------------------------------------------------------
109
+ 2026-11-01 20:56:17 | INFO | STEP 5/70 | dataset_item=0 | repeat=5/10
110
+ 2026-11-01 20:56:17 | INFO | LOSS: 1.20225418
111
+ 2026-11-01 20:56:17 | INFO | LR: 1.55555556e-04
112
+ 2026-11-01 20:56:17 | INFO | GLOBAL GRADIENT NORM: 0.31596762
113
+ 2026-11-01 20:56:17 | INFO | MAX ABSOLUTE GRADIENT: 1.14669800e-02
114
+ 2026-11-01 20:56:17 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
115
+ 2026-11-01 20:56:17 | INFO | MISSING GRADIENT TENSORS: 0
116
+ 2026-11-01 20:56:17 | INFO | GPU0 FORWARD: 2.115s
117
+ 2026-11-01 20:56:17 | INFO | GPU1 FORWARD: 2.399s
118
+ 2026-11-01 20:56:17 | INFO | GPU1 BACKWARD: 6.267s
119
+ 2026-11-01 20:56:17 | INFO | GPU0 BACKWARD: 5.485s
120
+ 2026-11-01 20:56:17 | INFO | UPDATE WALL TIME: 5.664s
121
+ 2026-11-01 20:56:17 | INFO | ELAPSED TRAINING TIME: 81.70s
122
+ 2026-11-01 20:56:17 | INFO | STEP 5 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.499 GB | free=3.850 GB
123
+ 2026-11-01 20:56:17 | INFO | STEP 5 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.419 GB | free=3.886 GB
124
+ 2026-11-01 20:56:35 | INFO | --------------------------------------------------------------------------------
125
+ 2026-11-01 20:56:35 | INFO | STEP 6/70 | dataset_item=0 | repeat=6/10
126
+ 2026-11-01 20:56:35 | INFO | LOSS: 1.20414209
127
+ 2026-11-01 20:56:35 | INFO | LR: 1.44444444e-04
128
+ 2026-11-01 20:56:35 | INFO | GLOBAL GRADIENT NORM: 0.25036251
129
+ 2026-11-01 20:56:35 | INFO | MAX ABSOLUTE GRADIENT: 1.13372803e-02
130
+ 2026-11-01 20:56:35 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
131
+ 2026-11-01 20:56:35 | INFO | MISSING GRADIENT TENSORS: 0
132
+ 2026-11-01 20:56:35 | INFO | GPU0 FORWARD: 2.160s
133
+ 2026-11-01 20:56:35 | INFO | GPU1 FORWARD: 2.469s
134
+ 2026-11-01 20:56:35 | INFO | GPU1 BACKWARD: 6.458s
135
+ 2026-11-01 20:56:35 | INFO | GPU0 BACKWARD: 5.663s
136
+ 2026-11-01 20:56:35 | INFO | UPDATE WALL TIME: 5.852s
137
+ 2026-11-01 20:56:35 | INFO | ELAPSED TRAINING TIME: 98.94s
138
+ 2026-11-01 20:56:35 | INFO | STEP 6 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.499 GB | free=3.850 GB
139
+ 2026-11-01 20:56:35 | INFO | STEP 6 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.419 GB | free=3.886 GB
140
+ 2026-11-01 20:56:52 | INFO | --------------------------------------------------------------------------------
141
+ 2026-11-01 20:56:52 | INFO | STEP 7/70 | dataset_item=0 | repeat=7/10
142
+ 2026-11-01 20:56:52 | INFO | LOSS: 0.69552815
143
+ 2026-11-01 20:56:52 | INFO | LR: 1.33333333e-04
144
+ 2026-11-01 20:56:52 | INFO | GLOBAL GRADIENT NORM: 0.30514208
145
+ 2026-11-01 20:56:52 | INFO | MAX ABSOLUTE GRADIENT: 7.28225708e-03
146
+ 2026-11-01 20:56:52 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
147
+ 2026-11-01 20:56:52 | INFO | MISSING GRADIENT TENSORS: 0
148
+ 2026-11-01 20:56:52 | INFO | GPU0 FORWARD: 2.231s
149
+ 2026-11-01 20:56:52 | INFO | GPU1 FORWARD: 2.529s
150
+ 2026-11-01 20:56:52 | INFO | GPU1 BACKWARD: 6.635s
151
+ 2026-11-01 20:56:52 | INFO | GPU0 BACKWARD: 5.761s
152
+ 2026-11-01 20:56:52 | INFO | UPDATE WALL TIME: 5.944s
153
+ 2026-11-01 20:56:52 | INFO | ELAPSED TRAINING TIME: 116.61s
154
+ 2026-11-01 20:56:52 | INFO | STEP 7 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.499 GB | free=3.850 GB
155
+ 2026-11-01 20:56:52 | INFO | STEP 7 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.419 GB | free=3.886 GB
156
+ 2026-11-01 20:57:10 | INFO | --------------------------------------------------------------------------------
157
+ 2026-11-01 20:57:10 | INFO | STEP 8/70 | dataset_item=0 | repeat=8/10
158
+ 2026-11-01 20:57:10 | INFO | LOSS: 1.14074183
159
+ 2026-11-01 20:57:10 | INFO | LR: 1.22222222e-04
160
+ 2026-11-01 20:57:10 | INFO | GLOBAL GRADIENT NORM: 0.27162458
161
+ 2026-11-01 20:57:10 | INFO | MAX ABSOLUTE GRADIENT: 8.02612305e-03
162
+ 2026-11-01 20:57:10 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
163
+ 2026-11-01 20:57:10 | INFO | MISSING GRADIENT TENSORS: 0
164
+ 2026-11-01 20:57:10 | INFO | GPU0 FORWARD: 2.254s
165
+ 2026-11-01 20:57:10 | INFO | GPU1 FORWARD: 2.636s
166
+ 2026-11-01 20:57:10 | INFO | GPU1 BACKWARD: 6.881s
167
+ 2026-11-01 20:57:10 | INFO | GPU0 BACKWARD: 5.836s
168
+ 2026-11-01 20:57:10 | INFO | UPDATE WALL TIME: 6.030s
169
+ 2026-11-01 20:57:10 | INFO | ELAPSED TRAINING TIME: 134.73s
170
+ 2026-11-01 20:57:10 | INFO | STEP 8 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.499 GB | free=3.850 GB
171
+ 2026-11-01 20:57:10 | INFO | STEP 8 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.419 GB | free=3.886 GB
172
+ 2026-11-01 20:57:28 | INFO | --------------------------------------------------------------------------------
173
+ 2026-11-01 20:57:28 | INFO | STEP 9/70 | dataset_item=0 | repeat=9/10
174
+ 2026-11-01 20:57:28 | INFO | LOSS: 0.54713565
175
+ 2026-11-01 20:57:28 | INFO | LR: 1.11111111e-04
176
+ 2026-11-01 20:57:28 | INFO | GLOBAL GRADIENT NORM: 0.20800196
177
+ 2026-11-01 20:57:28 | INFO | MAX ABSOLUTE GRADIENT: 5.58853149e-03
178
+ 2026-11-01 20:57:28 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
179
+ 2026-11-01 20:57:28 | INFO | MISSING GRADIENT TENSORS: 0
180
+ 2026-11-01 20:57:28 | INFO | GPU0 FORWARD: 2.265s
181
+ 2026-11-01 20:57:28 | INFO | GPU1 FORWARD: 2.601s
182
+ 2026-11-01 20:57:28 | INFO | GPU1 BACKWARD: 6.767s
183
+ 2026-11-01 20:57:28 | INFO | GPU0 BACKWARD: 5.607s
184
+ 2026-11-01 20:57:28 | INFO | UPDATE WALL TIME: 5.791s
185
+ 2026-11-01 20:57:28 | INFO | ELAPSED TRAINING TIME: 152.47s
186
+ 2026-11-01 20:57:28 | INFO | STEP 9 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.499 GB | free=3.850 GB
187
+ 2026-11-01 20:57:28 | INFO | STEP 9 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.419 GB | free=3.886 GB
188
+ 2026-11-01 20:57:46 | INFO | --------------------------------------------------------------------------------
189
+ 2026-11-01 20:57:46 | INFO | STEP 10/70 | dataset_item=0 | repeat=10/10
190
+ 2026-11-01 20:57:46 | INFO | LOSS: 0.69199997
191
+ 2026-11-01 20:57:46 | INFO | LR: 1.00000000e-04
192
+ 2026-11-01 20:57:46 | INFO | GLOBAL GRADIENT NORM: 0.15994336
193
+ 2026-11-01 20:57:46 | INFO | MAX ABSOLUTE GRADIENT: 4.33731079e-03
194
+ 2026-11-01 20:57:46 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
195
+ 2026-11-01 20:57:46 | INFO | MISSING GRADIENT TENSORS: 0
196
+ 2026-11-01 20:57:46 | INFO | GPU0 FORWARD: 2.198s
197
+ 2026-11-01 20:57:46 | INFO | GPU1 FORWARD: 2.581s
198
+ 2026-11-01 20:57:46 | INFO | GPU1 BACKWARD: 6.719s
199
+ 2026-11-01 20:57:46 | INFO | GPU0 BACKWARD: 5.643s
200
+ 2026-11-01 20:57:46 | INFO | UPDATE WALL TIME: 5.827s
201
+ 2026-11-01 20:57:46 | INFO | ELAPSED TRAINING TIME: 170.12s
202
+ 2026-11-01 20:57:46 | INFO | STEP 10 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.499 GB | free=3.850 GB
203
+ 2026-11-01 20:57:46 | INFO | STEP 10 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.419 GB | free=3.886 GB
204
+ 2026-11-01 20:57:46 | INFO | COMPLETED DATASET ITEM 1/7
205
+ 2026-11-01 20:57:46 | INFO | --------------------------------------------------------------------------------
206
+ 2026-11-01 20:57:46 | INFO | STARTING DATASET ITEM 2/7 | file=00000001.pt
207
+ 2026-11-01 20:57:46 | INFO | LEARNING RATE RESET: 2.00000000e-04
208
+ 2026-11-01 20:58:04 | INFO | --------------------------------------------------------------------------------
209
+ 2026-11-01 20:58:04 | INFO | STEP 11/70 | dataset_item=1 | repeat=1/10
210
+ 2026-11-01 20:58:04 | INFO | LOSS: 0.57349873
211
+ 2026-11-01 20:58:04 | INFO | LR: 2.00000000e-04
212
+ 2026-11-01 20:58:04 | INFO | GLOBAL GRADIENT NORM: 0.13330216
213
+ 2026-11-01 20:58:04 | INFO | MAX ABSOLUTE GRADIENT: 4.59671021e-03
214
+ 2026-11-01 20:58:04 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
215
+ 2026-11-01 20:58:04 | INFO | MISSING GRADIENT TENSORS: 0
216
+ 2026-11-01 20:58:04 | INFO | GPU0 FORWARD: 2.212s
217
+ 2026-11-01 20:58:04 | INFO | GPU1 FORWARD: 2.638s
218
+ 2026-11-01 20:58:04 | INFO | GPU1 BACKWARD: 7.001s
219
+ 2026-11-01 20:58:04 | INFO | GPU0 BACKWARD: 5.719s
220
+ 2026-11-01 20:58:04 | INFO | UPDATE WALL TIME: 5.902s
221
+ 2026-11-01 20:58:04 | INFO | ELAPSED TRAINING TIME: 188.18s
222
+ 2026-11-01 20:58:04 | INFO | STEP 11 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.453 GB | free=3.850 GB
223
+ 2026-11-01 20:58:04 | INFO | STEP 11 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.373 GB | free=3.886 GB
224
+ 2026-11-01 20:58:22 | INFO | --------------------------------------------------------------------------------
225
+ 2026-11-01 20:58:22 | INFO | STEP 12/70 | dataset_item=1 | repeat=2/10
226
+ 2026-11-01 20:58:22 | INFO | LOSS: 1.17241204
227
+ 2026-11-01 20:58:22 | INFO | LR: 1.88888889e-04
228
+ 2026-11-01 20:58:22 | INFO | GLOBAL GRADIENT NORM: 0.19624952
229
+ 2026-11-01 20:58:22 | INFO | MAX ABSOLUTE GRADIENT: 1.09405518e-02
230
+ 2026-11-01 20:58:22 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
231
+ 2026-11-01 20:58:22 | INFO | MISSING GRADIENT TENSORS: 0
232
+ 2026-11-01 20:58:22 | INFO | GPU0 FORWARD: 2.236s
233
+ 2026-11-01 20:58:22 | INFO | GPU1 FORWARD: 2.692s
234
+ 2026-11-01 20:58:22 | INFO | GPU1 BACKWARD: 6.992s
235
+ 2026-11-01 20:58:22 | INFO | GPU0 BACKWARD: 5.705s
236
+ 2026-11-01 20:58:22 | INFO | UPDATE WALL TIME: 5.891s
237
+ 2026-11-01 20:58:22 | INFO | ELAPSED TRAINING TIME: 206.30s
238
+ 2026-11-01 20:58:22 | INFO | STEP 12 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.453 GB | free=3.850 GB
239
+ 2026-11-01 20:58:22 | INFO | STEP 12 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.373 GB | free=3.886 GB
240
+ 2026-11-01 20:58:40 | INFO | --------------------------------------------------------------------------------
241
+ 2026-11-01 20:58:40 | INFO | STEP 13/70 | dataset_item=1 | repeat=3/10
242
+ 2026-11-01 20:58:40 | INFO | LOSS: 0.53789830
243
+ 2026-11-01 20:58:40 | INFO | LR: 1.77777778e-04
244
+ 2026-11-01 20:58:40 | INFO | GLOBAL GRADIENT NORM: 0.11847581
245
+ 2026-11-01 20:58:40 | INFO | MAX ABSOLUTE GRADIENT: 9.46807861e-03
246
+ 2026-11-01 20:58:40 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
247
+ 2026-11-01 20:58:40 | INFO | MISSING GRADIENT TENSORS: 0
248
+ 2026-11-01 20:58:40 | INFO | GPU0 FORWARD: 2.211s
249
+ 2026-11-01 20:58:40 | INFO | GPU1 FORWARD: 2.699s
250
+ 2026-11-01 20:58:40 | INFO | GPU1 BACKWARD: 6.985s
251
+ 2026-11-01 20:58:40 | INFO | GPU0 BACKWARD: 5.542s
252
+ 2026-11-01 20:58:40 | INFO | UPDATE WALL TIME: 5.727s
253
+ 2026-11-01 20:58:40 | INFO | ELAPSED TRAINING TIME: 224.24s
254
+ 2026-11-01 20:58:40 | INFO | STEP 13 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.453 GB | free=3.850 GB
255
+ 2026-11-01 20:58:40 | INFO | STEP 13 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.373 GB | free=3.886 GB
256
+ 2026-11-01 20:58:58 | INFO | --------------------------------------------------------------------------------
257
+ 2026-11-01 20:58:58 | INFO | STEP 14/70 | dataset_item=1 | repeat=4/10
258
+ 2026-11-01 20:58:58 | INFO | LOSS: 0.59814781
259
+ 2026-11-01 20:58:58 | INFO | LR: 1.66666667e-04
260
+ 2026-11-01 20:58:58 | INFO | GLOBAL GRADIENT NORM: 0.08131486
261
+ 2026-11-01 20:58:58 | INFO | MAX ABSOLUTE GRADIENT: 3.25775146e-03
262
+ 2026-11-01 20:58:58 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
263
+ 2026-11-01 20:58:58 | INFO | MISSING GRADIENT TENSORS: 0
264
+ 2026-11-01 20:58:58 | INFO | GPU0 FORWARD: 2.197s
265
+ 2026-11-01 20:58:58 | INFO | GPU1 FORWARD: 2.696s
266
+ 2026-11-01 20:58:58 | INFO | GPU1 BACKWARD: 6.964s
267
+ 2026-11-01 20:58:58 | INFO | GPU0 BACKWARD: 5.607s
268
+ 2026-11-01 20:58:58 | INFO | UPDATE WALL TIME: 5.792s
269
+ 2026-11-01 20:58:58 | INFO | ELAPSED TRAINING TIME: 242.22s
270
+ 2026-11-01 20:58:58 | INFO | STEP 14 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.453 GB | free=3.850 GB
271
+ 2026-11-01 20:58:58 | INFO | STEP 14 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.373 GB | free=3.886 GB
272
+ 2026-11-01 20:58:58 | INFO | ================================================================================
273
+ 2026-11-01 20:58:58 | INFO | CHECKPOINT TRIGGERED AT OPTIMIZER STEP 14
274
+ 2026-11-01 20:59:01 | INFO | CHECKPOINT SUCCESS: kromoss_lora_step_000014.safetensors
275
+ 2026-11-01 20:59:01 | INFO | CHECKPOINT SIZE: 70.93 MB
276
+ 2026-11-01 20:59:01 | INFO | ================================================================================
277
+ 2026-11-01 20:59:19 | INFO | --------------------------------------------------------------------------------
278
+ 2026-11-01 20:59:19 | INFO | STEP 15/70 | dataset_item=1 | repeat=5/10
279
+ 2026-11-01 20:59:19 | INFO | LOSS: 1.15985024
280
+ 2026-11-01 20:59:19 | INFO | LR: 1.55555556e-04
281
+ 2026-11-01 20:59:19 | INFO | GLOBAL GRADIENT NORM: 0.22236589
282
+ 2026-11-01 20:59:19 | INFO | MAX ABSOLUTE GRADIENT: 1.53045654e-02
283
+ 2026-11-01 20:59:19 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
284
+ 2026-11-01 20:59:19 | INFO | MISSING GRADIENT TENSORS: 0
285
+ 2026-11-01 20:59:19 | INFO | GPU0 FORWARD: 2.217s
286
+ 2026-11-01 20:59:19 | INFO | GPU1 FORWARD: 2.688s
287
+ 2026-11-01 20:59:19 | INFO | GPU1 BACKWARD: 7.049s
288
+ 2026-11-01 20:59:19 | INFO | GPU0 BACKWARD: 5.674s
289
+ 2026-11-01 20:59:19 | INFO | UPDATE WALL TIME: 5.860s
290
+ 2026-11-01 20:59:19 | INFO | ELAPSED TRAINING TIME: 263.78s
291
+ 2026-11-01 20:59:19 | INFO | STEP 15 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.453 GB | free=3.850 GB
292
+ 2026-11-01 20:59:19 | INFO | STEP 15 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.373 GB | free=3.886 GB
293
+ 2026-11-01 20:59:38 | INFO | --------------------------------------------------------------------------------
294
+ 2026-11-01 20:59:38 | INFO | STEP 16/70 | dataset_item=1 | repeat=6/10
295
+ 2026-11-01 20:59:38 | INFO | LOSS: 1.06413031
296
+ 2026-11-01 20:59:38 | INFO | LR: 1.44444444e-04
297
+ 2026-11-01 20:59:38 | INFO | GLOBAL GRADIENT NORM: 0.19298101
298
+ 2026-11-01 20:59:38 | INFO | MAX ABSOLUTE GRADIENT: 1.06811523e-02
299
+ 2026-11-01 20:59:38 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
300
+ 2026-11-01 20:59:38 | INFO | MISSING GRADIENT TENSORS: 0
301
+ 2026-11-01 20:59:38 | INFO | GPU0 FORWARD: 2.221s
302
+ 2026-11-01 20:59:38 | INFO | GPU1 FORWARD: 2.773s
303
+ 2026-11-01 20:59:38 | INFO | GPU1 BACKWARD: 7.299s
304
+ 2026-11-01 20:59:38 | INFO | GPU0 BACKWARD: 5.637s
305
+ 2026-11-01 20:59:38 | INFO | UPDATE WALL TIME: 5.821s
306
+ 2026-11-01 20:59:38 | INFO | ELAPSED TRAINING TIME: 282.21s
307
+ 2026-11-01 20:59:38 | INFO | STEP 16 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.453 GB | free=3.850 GB
308
+ 2026-11-01 20:59:38 | INFO | STEP 16 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.373 GB | free=3.886 GB
309
+ 2026-11-01 20:59:56 | INFO | --------------------------------------------------------------------------------
310
+ 2026-11-01 20:59:56 | INFO | STEP 17/70 | dataset_item=1 | repeat=7/10
311
+ 2026-11-01 20:59:56 | INFO | LOSS: 1.02878594
312
+ 2026-11-01 20:59:56 | INFO | LR: 1.33333333e-04
313
+ 2026-11-01 20:59:56 | INFO | GLOBAL GRADIENT NORM: 0.29010355
314
+ 2026-11-01 20:59:56 | INFO | MAX ABSOLUTE GRADIENT: 1.40304565e-02
315
+ 2026-11-01 20:59:56 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
316
+ 2026-11-01 20:59:56 | INFO | MISSING GRADIENT TENSORS: 0
317
+ 2026-11-01 20:59:56 | INFO | GPU0 FORWARD: 2.185s
318
+ 2026-11-01 20:59:56 | INFO | GPU1 FORWARD: 2.695s
319
+ 2026-11-01 20:59:56 | INFO | GPU1 BACKWARD: 7.018s
320
+ 2026-11-01 20:59:56 | INFO | GPU0 BACKWARD: 5.568s
321
+ 2026-11-01 20:59:56 | INFO | UPDATE WALL TIME: 5.754s
322
+ 2026-11-01 20:59:56 | INFO | ELAPSED TRAINING TIME: 300.18s
323
+ 2026-11-01 20:59:56 | INFO | STEP 17 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.453 GB | free=3.850 GB
324
+ 2026-11-01 20:59:56 | INFO | STEP 17 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.373 GB | free=3.886 GB
325
+ 2026-11-01 21:00:14 | INFO | --------------------------------------------------------------------------------
326
+ 2026-11-01 21:00:14 | INFO | STEP 18/70 | dataset_item=1 | repeat=8/10
327
+ 2026-11-01 21:00:14 | INFO | LOSS: 0.78760290
328
+ 2026-11-01 21:00:14 | INFO | LR: 1.22222222e-04
329
+ 2026-11-01 21:00:14 | INFO | GLOBAL GRADIENT NORM: 0.43464107
330
+ 2026-11-01 21:00:14 | INFO | MAX ABSOLUTE GRADIENT: 2.40478516e-02
331
+ 2026-11-01 21:00:14 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
332
+ 2026-11-01 21:00:14 | INFO | MISSING GRADIENT TENSORS: 0
333
+ 2026-11-01 21:00:14 | INFO | GPU0 FORWARD: 2.171s
334
+ 2026-11-01 21:00:14 | INFO | GPU1 FORWARD: 2.693s
335
+ 2026-11-01 21:00:14 | INFO | GPU1 BACKWARD: 7.251s
336
+ 2026-11-01 21:00:14 | INFO | GPU0 BACKWARD: 5.493s
337
+ 2026-11-01 21:00:14 | INFO | UPDATE WALL TIME: 5.680s
338
+ 2026-11-01 21:00:14 | INFO | ELAPSED TRAINING TIME: 318.29s
339
+ 2026-11-01 21:00:14 | INFO | STEP 18 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.453 GB | free=3.850 GB
340
+ 2026-11-01 21:00:14 | INFO | STEP 18 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.373 GB | free=3.886 GB
341
+ 2026-11-01 21:00:32 | INFO | --------------------------------------------------------------------------------
342
+ 2026-11-01 21:00:32 | INFO | STEP 19/70 | dataset_item=1 | repeat=9/10
343
+ 2026-11-01 21:00:32 | INFO | LOSS: 0.66918015
344
+ 2026-11-01 21:00:32 | INFO | LR: 1.11111111e-04
345
+ 2026-11-01 21:00:32 | INFO | GLOBAL GRADIENT NORM: 0.09124086
346
+ 2026-11-01 21:00:32 | INFO | MAX ABSOLUTE GRADIENT: 3.74794006e-03
347
+ 2026-11-01 21:00:32 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
348
+ 2026-11-01 21:00:32 | INFO | MISSING GRADIENT TENSORS: 0
349
+ 2026-11-01 21:00:32 | INFO | GPU0 FORWARD: 2.165s
350
+ 2026-11-01 21:00:32 | INFO | GPU1 FORWARD: 2.713s
351
+ 2026-11-01 21:00:32 | INFO | GPU1 BACKWARD: 7.196s
352
+ 2026-11-01 21:00:32 | INFO | GPU0 BACKWARD: 5.513s
353
+ 2026-11-01 21:00:32 | INFO | UPDATE WALL TIME: 5.698s
354
+ 2026-11-01 21:00:32 | INFO | ELAPSED TRAINING TIME: 336.38s
355
+ 2026-11-01 21:00:32 | INFO | STEP 19 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.453 GB | free=3.850 GB
356
+ 2026-11-01 21:00:32 | INFO | STEP 19 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.373 GB | free=3.886 GB
357
+ 2026-11-01 21:00:50 | INFO | --------------------------------------------------------------------------------
358
+ 2026-11-01 21:00:50 | INFO | STEP 20/70 | dataset_item=1 | repeat=10/10
359
+ 2026-11-01 21:00:50 | INFO | LOSS: 0.67549282
360
+ 2026-11-01 21:00:50 | INFO | LR: 1.00000000e-04
361
+ 2026-11-01 21:00:50 | INFO | GLOBAL GRADIENT NORM: 0.08762595
362
+ 2026-11-01 21:00:50 | INFO | MAX ABSOLUTE GRADIENT: 3.49807739e-03
363
+ 2026-11-01 21:00:50 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
364
+ 2026-11-01 21:00:50 | INFO | MISSING GRADIENT TENSORS: 0
365
+ 2026-11-01 21:00:50 | INFO | GPU0 FORWARD: 2.168s
366
+ 2026-11-01 21:00:50 | INFO | GPU1 FORWARD: 2.774s
367
+ 2026-11-01 21:00:50 | INFO | GPU1 BACKWARD: 7.338s
368
+ 2026-11-01 21:00:50 | INFO | GPU0 BACKWARD: 5.511s
369
+ 2026-11-01 21:00:50 | INFO | UPDATE WALL TIME: 5.695s
370
+ 2026-11-01 21:00:50 | INFO | ELAPSED TRAINING TIME: 354.67s
371
+ 2026-11-01 21:00:50 | INFO | STEP 20 | GPU0 | allocated=5.070 GB | reserved=10.574 GB | peak=10.453 GB | free=3.850 GB
372
+ 2026-11-01 21:00:50 | INFO | STEP 20 | GPU1 | allocated=4.984 GB | reserved=10.539 GB | peak=10.373 GB | free=3.886 GB
373
+ 2026-11-01 21:00:51 | INFO | COMPLETED DATASET ITEM 2/7
374
+ 2026-11-01 21:00:51 | INFO | --------------------------------------------------------------------------------
375
+ 2026-11-01 21:00:51 | INFO | STARTING DATASET ITEM 3/7 | file=00000002.pt
376
+ 2026-11-01 21:00:51 | INFO | LEARNING RATE RESET: 2.00000000e-04
377
+ 2026-11-01 21:01:09 | INFO | --------------------------------------------------------------------------------
378
+ 2026-11-01 21:01:09 | INFO | STEP 21/70 | dataset_item=2 | repeat=1/10
379
+ 2026-11-01 21:01:09 | INFO | LOSS: 1.11819673
380
+ 2026-11-01 21:01:09 | INFO | LR: 2.00000000e-04
381
+ 2026-11-01 21:01:09 | INFO | GLOBAL GRADIENT NORM: 0.24811577
382
+ 2026-11-01 21:01:09 | INFO | MAX ABSOLUTE GRADIENT: 1.67083740e-02
383
+ 2026-11-01 21:01:09 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
384
+ 2026-11-01 21:01:09 | INFO | MISSING GRADIENT TENSORS: 0
385
+ 2026-11-01 21:01:09 | INFO | GPU0 FORWARD: 2.188s
386
+ 2026-11-01 21:01:09 | INFO | GPU1 FORWARD: 2.766s
387
+ 2026-11-01 21:01:09 | INFO | GPU1 BACKWARD: 7.313s
388
+ 2026-11-01 21:01:09 | INFO | GPU0 BACKWARD: 5.531s
389
+ 2026-11-01 21:01:09 | INFO | UPDATE WALL TIME: 5.714s
390
+ 2026-11-01 21:01:09 | INFO | ELAPSED TRAINING TIME: 372.97s
391
+ 2026-11-01 21:01:09 | INFO | STEP 21 | GPU0 | allocated=5.070 GB | reserved=10.613 GB | peak=10.508 GB | free=3.811 GB
392
+ 2026-11-01 21:01:09 | INFO | STEP 21 | GPU1 | allocated=4.984 GB | reserved=10.559 GB | peak=10.429 GB | free=3.866 GB
393
+ 2026-11-01 21:01:27 | INFO | --------------------------------------------------------------------------------
394
+ 2026-11-01 21:01:27 | INFO | STEP 22/70 | dataset_item=2 | repeat=2/10
395
+ 2026-11-01 21:01:27 | INFO | LOSS: 0.75084013
396
+ 2026-11-01 21:01:27 | INFO | LR: 1.88888889e-04
397
+ 2026-11-01 21:01:27 | INFO | GLOBAL GRADIENT NORM: 0.11062503
398
+ 2026-11-01 21:01:27 | INFO | MAX ABSOLUTE GRADIENT: 5.59234619e-03
399
+ 2026-11-01 21:01:27 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
400
+ 2026-11-01 21:01:27 | INFO | MISSING GRADIENT TENSORS: 0
401
+ 2026-11-01 21:01:27 | INFO | GPU0 FORWARD: 2.169s
402
+ 2026-11-01 21:01:27 | INFO | GPU1 FORWARD: 2.797s
403
+ 2026-11-01 21:01:27 | INFO | GPU1 BACKWARD: 7.322s
404
+ 2026-11-01 21:01:27 | INFO | GPU0 BACKWARD: 5.470s
405
+ 2026-11-01 21:01:27 | INFO | UPDATE WALL TIME: 5.656s
406
+ 2026-11-01 21:01:27 | INFO | ELAPSED TRAINING TIME: 391.23s
407
+ 2026-11-01 21:01:27 | INFO | STEP 22 | GPU0 | allocated=5.070 GB | reserved=10.613 GB | peak=10.508 GB | free=3.811 GB
408
+ 2026-11-01 21:01:27 | INFO | STEP 22 | GPU1 | allocated=4.984 GB | reserved=10.559 GB | peak=10.429 GB | free=3.866 GB
409
+ 2026-11-01 21:01:45 | INFO | --------------------------------------------------------------------------------
410
+ 2026-11-01 21:01:45 | INFO | STEP 23/70 | dataset_item=2 | repeat=3/10
411
+ 2026-11-01 21:01:45 | INFO | LOSS: 0.60231853
412
+ 2026-11-01 21:01:45 | INFO | LR: 1.77777778e-04
413
+ 2026-11-01 21:01:45 | INFO | GLOBAL GRADIENT NORM: 0.09068894
414
+ 2026-11-01 21:01:45 | INFO | MAX ABSOLUTE GRADIENT: 7.88879395e-03
415
+ 2026-11-01 21:01:45 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
416
+ 2026-11-01 21:01:45 | INFO | MISSING GRADIENT TENSORS: 0
417
+ 2026-11-01 21:01:45 | INFO | GPU0 FORWARD: 2.157s
418
+ 2026-11-01 21:01:45 | INFO | GPU1 FORWARD: 2.733s
419
+ 2026-11-01 21:01:45 | INFO | GPU1 BACKWARD: 7.269s
420
+ 2026-11-01 21:01:45 | INFO | GPU0 BACKWARD: 5.445s
421
+ 2026-11-01 21:01:45 | INFO | UPDATE WALL TIME: 5.658s
422
+ 2026-11-01 21:01:45 | INFO | ELAPSED TRAINING TIME: 409.37s
423
+ 2026-11-01 21:01:45 | INFO | STEP 23 | GPU0 | allocated=5.070 GB | reserved=10.613 GB | peak=10.508 GB | free=3.811 GB
424
+ 2026-11-01 21:01:45 | INFO | STEP 23 | GPU1 | allocated=4.984 GB | reserved=10.559 GB | peak=10.429 GB | free=3.866 GB
425
+ 2026-11-01 21:02:03 | INFO | --------------------------------------------------------------------------------
426
+ 2026-11-01 21:02:03 | INFO | STEP 24/70 | dataset_item=2 | repeat=4/10
427
+ 2026-11-01 21:02:03 | INFO | LOSS: 0.59630340
428
+ 2026-11-01 21:02:03 | INFO | LR: 1.66666667e-04
429
+ 2026-11-01 21:02:03 | INFO | GLOBAL GRADIENT NORM: 0.06777528
430
+ 2026-11-01 21:02:03 | INFO | MAX ABSOLUTE GRADIENT: 3.54766846e-03
431
+ 2026-11-01 21:02:03 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
432
+ 2026-11-01 21:02:03 | INFO | MISSING GRADIENT TENSORS: 0
433
+ 2026-11-01 21:02:03 | INFO | GPU0 FORWARD: 2.159s
434
+ 2026-11-01 21:02:03 | INFO | GPU1 FORWARD: 2.802s
435
+ 2026-11-01 21:02:03 | INFO | GPU1 BACKWARD: 7.354s
436
+ 2026-11-01 21:02:03 | INFO | GPU0 BACKWARD: 5.441s
437
+ 2026-11-01 21:02:03 | INFO | UPDATE WALL TIME: 5.627s
438
+ 2026-11-01 21:02:03 | INFO | ELAPSED TRAINING TIME: 427.62s
439
+ 2026-11-01 21:02:03 | INFO | STEP 24 | GPU0 | allocated=5.070 GB | reserved=10.613 GB | peak=10.508 GB | free=3.811 GB
440
+ 2026-11-01 21:02:03 | INFO | STEP 24 | GPU1 | allocated=4.984 GB | reserved=10.559 GB | peak=10.429 GB | free=3.866 GB
441
+ 2026-11-01 21:02:21 | INFO | --------------------------------------------------------------------------------
442
+ 2026-11-01 21:02:21 | INFO | STEP 25/70 | dataset_item=2 | repeat=5/10
443
+ 2026-11-01 21:02:21 | INFO | LOSS: 0.53225797
444
+ 2026-11-01 21:02:21 | INFO | LR: 1.55555556e-04
445
+ 2026-11-01 21:02:21 | INFO | GLOBAL GRADIENT NORM: 0.07344767
446
+ 2026-11-01 21:02:21 | INFO | MAX ABSOLUTE GRADIENT: 6.52313232e-03
447
+ 2026-11-01 21:02:21 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
448
+ 2026-11-01 21:02:21 | INFO | MISSING GRADIENT TENSORS: 0
449
+ 2026-11-01 21:02:21 | INFO | GPU0 FORWARD: 2.166s
450
+ 2026-11-01 21:02:21 | INFO | GPU1 FORWARD: 2.777s
451
+ 2026-11-01 21:02:21 | INFO | GPU1 BACKWARD: 7.340s
452
+ 2026-11-01 21:02:21 | INFO | GPU0 BACKWARD: 5.419s
453
+ 2026-11-01 21:02:21 | INFO | UPDATE WALL TIME: 5.606s
454
+ 2026-11-01 21:02:21 | INFO | ELAPSED TRAINING TIME: 445.83s
455
+ 2026-11-01 21:02:21 | INFO | STEP 25 | GPU0 | allocated=5.070 GB | reserved=10.613 GB | peak=10.508 GB | free=3.811 GB
456
+ 2026-11-01 21:02:21 | INFO | STEP 25 | GPU1 | allocated=4.984 GB | reserved=10.559 GB | peak=10.429 GB | free=3.866 GB
457
+ 2026-11-01 21:02:40 | INFO | --------------------------------------------------------------------------------
458
+ 2026-11-01 21:02:40 | INFO | STEP 26/70 | dataset_item=2 | repeat=6/10
459
+ 2026-11-01 21:02:40 | INFO | LOSS: 1.11425531
460
+ 2026-11-01 21:02:40 | INFO | LR: 1.44444444e-04
461
+ 2026-11-01 21:02:40 | INFO | GLOBAL GRADIENT NORM: 0.17138327
462
+ 2026-11-01 21:02:40 | INFO | MAX ABSOLUTE GRADIENT: 8.58306885e-03
463
+ 2026-11-01 21:02:40 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
464
+ 2026-11-01 21:02:40 | INFO | MISSING GRADIENT TENSORS: 0
465
+ 2026-11-01 21:02:40 | INFO | GPU0 FORWARD: 2.156s
466
+ 2026-11-01 21:02:40 | INFO | GPU1 FORWARD: 2.808s
467
+ 2026-11-01 21:02:40 | INFO | GPU1 BACKWARD: 7.260s
468
+ 2026-11-01 21:02:40 | INFO | GPU0 BACKWARD: 5.400s
469
+ 2026-11-01 21:02:40 | INFO | UPDATE WALL TIME: 5.582s
470
+ 2026-11-01 21:02:40 | INFO | ELAPSED TRAINING TIME: 463.95s
471
+ 2026-11-01 21:02:40 | INFO | STEP 26 | GPU0 | allocated=5.070 GB | reserved=10.613 GB | peak=10.508 GB | free=3.811 GB
472
+ 2026-11-01 21:02:40 | INFO | STEP 26 | GPU1 | allocated=4.984 GB | reserved=10.559 GB | peak=10.429 GB | free=3.866 GB
473
+ 2026-11-01 21:02:58 | INFO | --------------------------------------------------------------------------------
474
+ 2026-11-01 21:02:58 | INFO | STEP 27/70 | dataset_item=2 | repeat=7/10
475
+ 2026-11-01 21:02:58 | INFO | LOSS: 0.70921290
476
+ 2026-11-01 21:02:58 | INFO | LR: 1.33333333e-04
477
+ 2026-11-01 21:02:58 | INFO | GLOBAL GRADIENT NORM: 0.15583875
478
+ 2026-11-01 21:02:58 | INFO | MAX ABSOLUTE GRADIENT: 8.88061523e-03
479
+ 2026-11-01 21:02:58 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
480
+ 2026-11-01 21:02:58 | INFO | MISSING GRADIENT TENSORS: 0
481
+ 2026-11-01 21:02:58 | INFO | GPU0 FORWARD: 2.145s
482
+ 2026-11-01 21:02:58 | INFO | GPU1 FORWARD: 2.755s
483
+ 2026-11-01 21:02:58 | INFO | GPU1 BACKWARD: 7.315s
484
+ 2026-11-01 21:02:58 | INFO | GPU0 BACKWARD: 5.395s
485
+ 2026-11-01 21:02:58 | INFO | UPDATE WALL TIME: 5.576s
486
+ 2026-11-01 21:02:58 | INFO | ELAPSED TRAINING TIME: 482.06s
487
+ 2026-11-01 21:02:58 | INFO | STEP 27 | GPU0 | allocated=5.070 GB | reserved=10.613 GB | peak=10.508 GB | free=3.811 GB
488
+ 2026-11-01 21:02:58 | INFO | STEP 27 | GPU1 | allocated=4.984 GB | reserved=10.559 GB | peak=10.429 GB | free=3.866 GB
489
+ 2026-11-01 21:03:16 | INFO | --------------------------------------------------------------------------------
490
+ 2026-11-01 21:03:16 | INFO | STEP 28/70 | dataset_item=2 | repeat=8/10
491
+ 2026-11-01 21:03:16 | INFO | LOSS: 0.92250276
492
+ 2026-11-01 21:03:16 | INFO | LR: 1.22222222e-04
493
+ 2026-11-01 21:03:16 | INFO | GLOBAL GRADIENT NORM: 0.11405573
494
+ 2026-11-01 21:03:16 | INFO | MAX ABSOLUTE GRADIENT: 5.50842285e-03
495
+ 2026-11-01 21:03:16 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
496
+ 2026-11-01 21:03:16 | INFO | MISSING GRADIENT TENSORS: 0
497
+ 2026-11-01 21:03:16 | INFO | GPU0 FORWARD: 2.136s
498
+ 2026-11-01 21:03:16 | INFO | GPU1 FORWARD: 2.787s
499
+ 2026-11-01 21:03:16 | INFO | GPU1 BACKWARD: 7.358s
500
+ 2026-11-01 21:03:16 | INFO | GPU0 BACKWARD: 5.423s
501
+ 2026-11-01 21:03:16 | INFO | UPDATE WALL TIME: 5.607s
502
+ 2026-11-01 21:03:16 | INFO | ELAPSED TRAINING TIME: 500.27s
503
+ 2026-11-01 21:03:16 | INFO | STEP 28 | GPU0 | allocated=5.070 GB | reserved=10.613 GB | peak=10.508 GB | free=3.811 GB
504
+ 2026-11-01 21:03:16 | INFO | STEP 28 | GPU1 | allocated=4.984 GB | reserved=10.559 GB | peak=10.429 GB | free=3.866 GB
505
+ 2026-11-01 21:03:34 | INFO | --------------------------------------------------------------------------------
506
+ 2026-11-01 21:03:34 | INFO | STEP 29/70 | dataset_item=2 | repeat=9/10
507
+ 2026-11-01 21:03:34 | INFO | LOSS: 1.21125722
508
+ 2026-11-01 21:03:34 | INFO | LR: 1.11111111e-04
509
+ 2026-11-01 21:03:34 | INFO | GLOBAL GRADIENT NORM: 0.09722946
510
+ 2026-11-01 21:03:34 | INFO | MAX ABSOLUTE GRADIENT: 3.18908691e-03
511
+ 2026-11-01 21:03:34 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
512
+ 2026-11-01 21:03:34 | INFO | MISSING GRADIENT TENSORS: 0
513
+ 2026-11-01 21:03:34 | INFO | GPU0 FORWARD: 2.125s
514
+ 2026-11-01 21:03:34 | INFO | GPU1 FORWARD: 2.780s
515
+ 2026-11-01 21:03:34 | INFO | GPU1 BACKWARD: 7.313s
516
+ 2026-11-01 21:03:34 | INFO | GPU0 BACKWARD: 5.440s
517
+ 2026-11-01 21:03:34 | INFO | UPDATE WALL TIME: 5.623s
518
+ 2026-11-01 21:03:34 | INFO | ELAPSED TRAINING TIME: 518.42s
519
+ 2026-11-01 21:03:34 | INFO | STEP 29 | GPU0 | allocated=5.070 GB | reserved=10.613 GB | peak=10.508 GB | free=3.811 GB
520
+ 2026-11-01 21:03:34 | INFO | STEP 29 | GPU1 | allocated=4.984 GB | reserved=10.559 GB | peak=10.429 GB | free=3.866 GB
521
+ 2026-11-01 21:03:52 | INFO | --------------------------------------------------------------------------------
522
+ 2026-11-01 21:03:52 | INFO | STEP 30/70 | dataset_item=2 | repeat=10/10
523
+ 2026-11-01 21:03:52 | INFO | LOSS: 0.53835690
524
+ 2026-11-01 21:03:52 | INFO | LR: 1.00000000e-04
525
+ 2026-11-01 21:03:52 | INFO | GLOBAL GRADIENT NORM: 0.26504944
526
+ 2026-11-01 21:03:52 | INFO | MAX ABSOLUTE GRADIENT: 7.46917725e-03
527
+ 2026-11-01 21:03:52 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
528
+ 2026-11-01 21:03:52 | INFO | MISSING GRADIENT TENSORS: 0
529
+ 2026-11-01 21:03:52 | INFO | GPU0 FORWARD: 2.142s
530
+ 2026-11-01 21:03:52 | INFO | GPU1 FORWARD: 2.788s
531
+ 2026-11-01 21:03:52 | INFO | GPU1 BACKWARD: 7.437s
532
+ 2026-11-01 21:03:52 | INFO | GPU0 BACKWARD: 5.399s
533
+ 2026-11-01 21:03:52 | INFO | UPDATE WALL TIME: 5.581s
534
+ 2026-11-01 21:03:52 | INFO | ELAPSED TRAINING TIME: 536.68s
535
+ 2026-11-01 21:03:52 | INFO | STEP 30 | GPU0 | allocated=5.070 GB | reserved=10.613 GB | peak=10.508 GB | free=3.811 GB
536
+ 2026-11-01 21:03:52 | INFO | STEP 30 | GPU1 | allocated=4.984 GB | reserved=10.559 GB | peak=10.429 GB | free=3.866 GB
537
+ 2026-11-01 21:03:53 | INFO | COMPLETED DATASET ITEM 3/7
538
+ 2026-11-01 21:03:53 | INFO | --------------------------------------------------------------------------------
539
+ 2026-11-01 21:03:53 | INFO | STARTING DATASET ITEM 4/7 | file=00000003.pt
540
+ 2026-11-01 21:03:53 | INFO | LEARNING RATE RESET: 2.00000000e-04
541
+ 2026-11-01 21:04:10 | INFO | --------------------------------------------------------------------------------
542
+ 2026-11-01 21:04:10 | INFO | STEP 31/70 | dataset_item=3 | repeat=1/10
543
+ 2026-11-01 21:04:10 | INFO | LOSS: 1.17801535
544
+ 2026-11-01 21:04:10 | INFO | LR: 2.00000000e-04
545
+ 2026-11-01 21:04:10 | INFO | GLOBAL GRADIENT NORM: 0.12491994
546
+ 2026-11-01 21:04:10 | INFO | MAX ABSOLUTE GRADIENT: 8.81958008e-03
547
+ 2026-11-01 21:04:10 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
548
+ 2026-11-01 21:04:10 | INFO | MISSING GRADIENT TENSORS: 0
549
+ 2026-11-01 21:04:10 | INFO | GPU0 FORWARD: 2.118s
550
+ 2026-11-01 21:04:10 | INFO | GPU1 FORWARD: 2.765s
551
+ 2026-11-01 21:04:10 | INFO | GPU1 BACKWARD: 7.235s
552
+ 2026-11-01 21:04:10 | INFO | GPU0 BACKWARD: 5.450s
553
+ 2026-11-01 21:04:10 | INFO | UPDATE WALL TIME: 5.634s
554
+ 2026-11-01 21:04:10 | INFO | ELAPSED TRAINING TIME: 554.75s
555
+ 2026-11-01 21:04:10 | INFO | STEP 31 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.553 GB | free=3.772 GB
556
+ 2026-11-01 21:04:10 | INFO | STEP 31 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.473 GB | free=3.827 GB
557
+ 2026-11-01 21:04:29 | INFO | --------------------------------------------------------------------------------
558
+ 2026-11-01 21:04:29 | INFO | STEP 32/70 | dataset_item=3 | repeat=2/10
559
+ 2026-11-01 21:04:29 | INFO | LOSS: 1.11661386
560
+ 2026-11-01 21:04:29 | INFO | LR: 1.88888889e-04
561
+ 2026-11-01 21:04:29 | INFO | GLOBAL GRADIENT NORM: 0.14895421
562
+ 2026-11-01 21:04:29 | INFO | MAX ABSOLUTE GRADIENT: 1.44882202e-02
563
+ 2026-11-01 21:04:29 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
564
+ 2026-11-01 21:04:29 | INFO | MISSING GRADIENT TENSORS: 0
565
+ 2026-11-01 21:04:29 | INFO | GPU0 FORWARD: 2.129s
566
+ 2026-11-01 21:04:29 | INFO | GPU1 FORWARD: 2.796s
567
+ 2026-11-01 21:04:29 | INFO | GPU1 BACKWARD: 7.315s
568
+ 2026-11-01 21:04:29 | INFO | GPU0 BACKWARD: 5.419s
569
+ 2026-11-01 21:04:29 | INFO | UPDATE WALL TIME: 5.604s
570
+ 2026-11-01 21:04:29 | INFO | ELAPSED TRAINING TIME: 572.91s
571
+ 2026-11-01 21:04:29 | INFO | STEP 32 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.553 GB | free=3.772 GB
572
+ 2026-11-01 21:04:29 | INFO | STEP 32 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.473 GB | free=3.827 GB
573
+ 2026-11-01 21:04:47 | INFO | --------------------------------------------------------------------------------
574
+ 2026-11-01 21:04:47 | INFO | STEP 33/70 | dataset_item=3 | repeat=3/10
575
+ 2026-11-01 21:04:47 | INFO | LOSS: 1.07947433
576
+ 2026-11-01 21:04:47 | INFO | LR: 1.77777778e-04
577
+ 2026-11-01 21:04:47 | INFO | GLOBAL GRADIENT NORM: 0.16332199
578
+ 2026-11-01 21:04:47 | INFO | MAX ABSOLUTE GRADIENT: 1.93023682e-02
579
+ 2026-11-01 21:04:47 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
580
+ 2026-11-01 21:04:47 | INFO | MISSING GRADIENT TENSORS: 0
581
+ 2026-11-01 21:04:47 | INFO | GPU0 FORWARD: 2.135s
582
+ 2026-11-01 21:04:47 | INFO | GPU1 FORWARD: 2.812s
583
+ 2026-11-01 21:04:47 | INFO | GPU1 BACKWARD: 7.306s
584
+ 2026-11-01 21:04:47 | INFO | GPU0 BACKWARD: 5.421s
585
+ 2026-11-01 21:04:47 | INFO | UPDATE WALL TIME: 5.604s
586
+ 2026-11-01 21:04:47 | INFO | ELAPSED TRAINING TIME: 591.09s
587
+ 2026-11-01 21:04:47 | INFO | STEP 33 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.553 GB | free=3.772 GB
588
+ 2026-11-01 21:04:47 | INFO | STEP 33 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.473 GB | free=3.827 GB
589
+ 2026-11-01 21:05:05 | INFO | --------------------------------------------------------------------------------
590
+ 2026-11-01 21:05:05 | INFO | STEP 34/70 | dataset_item=3 | repeat=4/10
591
+ 2026-11-01 21:05:05 | INFO | LOSS: 0.73851019
592
+ 2026-11-01 21:05:05 | INFO | LR: 1.66666667e-04
593
+ 2026-11-01 21:05:05 | INFO | GLOBAL GRADIENT NORM: 0.07702935
594
+ 2026-11-01 21:05:05 | INFO | MAX ABSOLUTE GRADIENT: 5.06973267e-03
595
+ 2026-11-01 21:05:05 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
596
+ 2026-11-01 21:05:05 | INFO | MISSING GRADIENT TENSORS: 0
597
+ 2026-11-01 21:05:05 | INFO | GPU0 FORWARD: 2.130s
598
+ 2026-11-01 21:05:05 | INFO | GPU1 FORWARD: 2.786s
599
+ 2026-11-01 21:05:05 | INFO | GPU1 BACKWARD: 7.324s
600
+ 2026-11-01 21:05:05 | INFO | GPU0 BACKWARD: 5.402s
601
+ 2026-11-01 21:05:05 | INFO | UPDATE WALL TIME: 5.620s
602
+ 2026-11-01 21:05:05 | INFO | ELAPSED TRAINING TIME: 609.27s
603
+ 2026-11-01 21:05:05 | INFO | STEP 34 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.553 GB | free=3.772 GB
604
+ 2026-11-01 21:05:05 | INFO | STEP 34 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.473 GB | free=3.827 GB
605
+ 2026-11-01 21:05:23 | INFO | --------------------------------------------------------------------------------
606
+ 2026-11-01 21:05:23 | INFO | STEP 35/70 | dataset_item=3 | repeat=5/10
607
+ 2026-11-01 21:05:23 | INFO | LOSS: 0.71245539
608
+ 2026-11-01 21:05:23 | INFO | LR: 1.55555556e-04
609
+ 2026-11-01 21:05:23 | INFO | GLOBAL GRADIENT NORM: 0.06853743
610
+ 2026-11-01 21:05:23 | INFO | MAX ABSOLUTE GRADIENT: 5.69152832e-03
611
+ 2026-11-01 21:05:23 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
612
+ 2026-11-01 21:05:23 | INFO | MISSING GRADIENT TENSORS: 0
613
+ 2026-11-01 21:05:23 | INFO | GPU0 FORWARD: 2.127s
614
+ 2026-11-01 21:05:23 | INFO | GPU1 FORWARD: 2.797s
615
+ 2026-11-01 21:05:23 | INFO | GPU1 BACKWARD: 7.355s
616
+ 2026-11-01 21:05:23 | INFO | GPU0 BACKWARD: 5.417s
617
+ 2026-11-01 21:05:23 | INFO | UPDATE WALL TIME: 5.605s
618
+ 2026-11-01 21:05:23 | INFO | ELAPSED TRAINING TIME: 627.46s
619
+ 2026-11-01 21:05:23 | INFO | STEP 35 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.553 GB | free=3.772 GB
620
+ 2026-11-01 21:05:23 | INFO | STEP 35 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.473 GB | free=3.827 GB
621
+ 2026-11-01 21:05:23 | INFO | ================================================================================
622
+ 2026-11-01 21:05:23 | INFO | CHECKPOINT TRIGGERED AT OPTIMIZER STEP 35
623
+ 2026-11-01 21:05:27 | INFO | CHECKPOINT SUCCESS: kromoss_lora_step_000035.safetensors
624
+ 2026-11-01 21:05:27 | INFO | CHECKPOINT SIZE: 70.93 MB
625
+ 2026-11-01 21:05:27 | INFO | ================================================================================
626
+ 2026-11-01 21:05:45 | INFO | --------------------------------------------------------------------------------
627
+ 2026-11-01 21:05:45 | INFO | STEP 36/70 | dataset_item=3 | repeat=6/10
628
+ 2026-11-01 21:05:45 | INFO | LOSS: 1.08466649
629
+ 2026-11-01 21:05:45 | INFO | LR: 1.44444444e-04
630
+ 2026-11-01 21:05:45 | INFO | GLOBAL GRADIENT NORM: 0.12924485
631
+ 2026-11-01 21:05:45 | INFO | MAX ABSOLUTE GRADIENT: 1.61590576e-02
632
+ 2026-11-01 21:05:45 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
633
+ 2026-11-01 21:05:45 | INFO | MISSING GRADIENT TENSORS: 0
634
+ 2026-11-01 21:05:45 | INFO | GPU0 FORWARD: 2.169s
635
+ 2026-11-01 21:05:45 | INFO | GPU1 FORWARD: 2.742s
636
+ 2026-11-01 21:05:45 | INFO | GPU1 BACKWARD: 7.182s
637
+ 2026-11-01 21:05:45 | INFO | GPU0 BACKWARD: 5.448s
638
+ 2026-11-01 21:05:45 | INFO | UPDATE WALL TIME: 5.630s
639
+ 2026-11-01 21:05:45 | INFO | ELAPSED TRAINING TIME: 649.46s
640
+ 2026-11-01 21:05:45 | INFO | STEP 36 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.553 GB | free=3.772 GB
641
+ 2026-11-01 21:05:45 | INFO | STEP 36 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.473 GB | free=3.827 GB
642
+ 2026-11-01 21:06:04 | INFO | --------------------------------------------------------------------------------
643
+ 2026-11-01 21:06:04 | INFO | STEP 37/70 | dataset_item=3 | repeat=7/10
644
+ 2026-11-01 21:06:04 | INFO | LOSS: 0.49522993
645
+ 2026-11-01 21:06:04 | INFO | LR: 1.33333333e-04
646
+ 2026-11-01 21:06:04 | INFO | GLOBAL GRADIENT NORM: 0.08478694
647
+ 2026-11-01 21:06:04 | INFO | MAX ABSOLUTE GRADIENT: 7.45773315e-03
648
+ 2026-11-01 21:06:04 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
649
+ 2026-11-01 21:06:04 | INFO | MISSING GRADIENT TENSORS: 0
650
+ 2026-11-01 21:06:04 | INFO | GPU0 FORWARD: 2.149s
651
+ 2026-11-01 21:06:04 | INFO | GPU1 FORWARD: 2.890s
652
+ 2026-11-01 21:06:04 | INFO | GPU1 BACKWARD: 7.713s
653
+ 2026-11-01 21:06:04 | INFO | GPU0 BACKWARD: 5.429s
654
+ 2026-11-01 21:06:04 | INFO | UPDATE WALL TIME: 5.615s
655
+ 2026-11-01 21:06:04 | INFO | ELAPSED TRAINING TIME: 668.14s
656
+ 2026-11-01 21:06:04 | INFO | STEP 37 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.553 GB | free=3.772 GB
657
+ 2026-11-01 21:06:04 | INFO | STEP 37 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.473 GB | free=3.827 GB
658
+ 2026-11-01 21:06:22 | INFO | --------------------------------------------------------------------------------
659
+ 2026-11-01 21:06:22 | INFO | STEP 38/70 | dataset_item=3 | repeat=8/10
660
+ 2026-11-01 21:06:22 | INFO | LOSS: 0.96997595
661
+ 2026-11-01 21:06:22 | INFO | LR: 1.22222222e-04
662
+ 2026-11-01 21:06:22 | INFO | GLOBAL GRADIENT NORM: 0.07053143
663
+ 2026-11-01 21:06:22 | INFO | MAX ABSOLUTE GRADIENT: 2.55584717e-03
664
+ 2026-11-01 21:06:22 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
665
+ 2026-11-01 21:06:22 | INFO | MISSING GRADIENT TENSORS: 0
666
+ 2026-11-01 21:06:22 | INFO | GPU0 FORWARD: 2.124s
667
+ 2026-11-01 21:06:22 | INFO | GPU1 FORWARD: 2.794s
668
+ 2026-11-01 21:06:22 | INFO | GPU1 BACKWARD: 7.316s
669
+ 2026-11-01 21:06:22 | INFO | GPU0 BACKWARD: 5.419s
670
+ 2026-11-01 21:06:22 | INFO | UPDATE WALL TIME: 5.599s
671
+ 2026-11-01 21:06:22 | INFO | ELAPSED TRAINING TIME: 686.29s
672
+ 2026-11-01 21:06:22 | INFO | STEP 38 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.553 GB | free=3.772 GB
673
+ 2026-11-01 21:06:22 | INFO | STEP 38 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.473 GB | free=3.827 GB
674
+ 2026-11-01 21:06:40 | INFO | --------------------------------------------------------------------------------
675
+ 2026-11-01 21:06:40 | INFO | STEP 39/70 | dataset_item=3 | repeat=9/10
676
+ 2026-11-01 21:06:40 | INFO | LOSS: 0.81758428
677
+ 2026-11-01 21:06:40 | INFO | LR: 1.11111111e-04
678
+ 2026-11-01 21:06:40 | INFO | GLOBAL GRADIENT NORM: 0.05751622
679
+ 2026-11-01 21:06:40 | INFO | MAX ABSOLUTE GRADIENT: 2.10762024e-03
680
+ 2026-11-01 21:06:40 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
681
+ 2026-11-01 21:06:40 | INFO | MISSING GRADIENT TENSORS: 0
682
+ 2026-11-01 21:06:40 | INFO | GPU0 FORWARD: 2.117s
683
+ 2026-11-01 21:06:40 | INFO | GPU1 FORWARD: 2.767s
684
+ 2026-11-01 21:06:40 | INFO | GPU1 BACKWARD: 7.273s
685
+ 2026-11-01 21:06:40 | INFO | GPU0 BACKWARD: 5.429s
686
+ 2026-11-01 21:06:40 | INFO | UPDATE WALL TIME: 5.616s
687
+ 2026-11-01 21:06:40 | INFO | ELAPSED TRAINING TIME: 704.38s
688
+ 2026-11-01 21:06:40 | INFO | STEP 39 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.553 GB | free=3.772 GB
689
+ 2026-11-01 21:06:40 | INFO | STEP 39 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.473 GB | free=3.827 GB
690
+ 2026-11-01 21:06:58 | INFO | --------------------------------------------------------------------------------
691
+ 2026-11-01 21:06:58 | INFO | STEP 40/70 | dataset_item=3 | repeat=10/10
692
+ 2026-11-01 21:06:58 | INFO | LOSS: 0.55638587
693
+ 2026-11-01 21:06:58 | INFO | LR: 1.00000000e-04
694
+ 2026-11-01 21:06:58 | INFO | GLOBAL GRADIENT NORM: 0.05621507
695
+ 2026-11-01 21:06:58 | INFO | MAX ABSOLUTE GRADIENT: 2.46047974e-03
696
+ 2026-11-01 21:06:58 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
697
+ 2026-11-01 21:06:58 | INFO | MISSING GRADIENT TENSORS: 0
698
+ 2026-11-01 21:06:58 | INFO | GPU0 FORWARD: 2.123s
699
+ 2026-11-01 21:06:58 | INFO | GPU1 FORWARD: 2.782s
700
+ 2026-11-01 21:06:58 | INFO | GPU1 BACKWARD: 7.360s
701
+ 2026-11-01 21:06:58 | INFO | GPU0 BACKWARD: 5.419s
702
+ 2026-11-01 21:06:58 | INFO | UPDATE WALL TIME: 5.606s
703
+ 2026-11-01 21:06:58 | INFO | ELAPSED TRAINING TIME: 722.59s
704
+ 2026-11-01 21:06:58 | INFO | STEP 40 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.553 GB | free=3.772 GB
705
+ 2026-11-01 21:06:58 | INFO | STEP 40 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.473 GB | free=3.827 GB
706
+ 2026-11-01 21:06:59 | INFO | COMPLETED DATASET ITEM 4/7
707
+ 2026-11-01 21:06:59 | INFO | --------------------------------------------------------------------------------
708
+ 2026-11-01 21:06:59 | INFO | STARTING DATASET ITEM 5/7 | file=00000004.pt
709
+ 2026-11-01 21:06:59 | INFO | LEARNING RATE RESET: 2.00000000e-04
710
+ 2026-11-01 21:07:16 | INFO | --------------------------------------------------------------------------------
711
+ 2026-11-01 21:07:16 | INFO | STEP 41/70 | dataset_item=4 | repeat=1/10
712
+ 2026-11-01 21:07:16 | INFO | LOSS: 1.12809253
713
+ 2026-11-01 21:07:16 | INFO | LR: 2.00000000e-04
714
+ 2026-11-01 21:07:16 | INFO | GLOBAL GRADIENT NORM: 0.16674193
715
+ 2026-11-01 21:07:16 | INFO | MAX ABSOLUTE GRADIENT: 1.50070190e-02
716
+ 2026-11-01 21:07:16 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
717
+ 2026-11-01 21:07:16 | INFO | MISSING GRADIENT TENSORS: 0
718
+ 2026-11-01 21:07:16 | INFO | GPU0 FORWARD: 2.127s
719
+ 2026-11-01 21:07:16 | INFO | GPU1 FORWARD: 2.814s
720
+ 2026-11-01 21:07:16 | INFO | GPU1 BACKWARD: 7.343s
721
+ 2026-11-01 21:07:16 | INFO | GPU0 BACKWARD: 5.401s
722
+ 2026-11-01 21:07:16 | INFO | UPDATE WALL TIME: 5.585s
723
+ 2026-11-01 21:07:16 | INFO | ELAPSED TRAINING TIME: 740.78s
724
+ 2026-11-01 21:07:16 | INFO | STEP 41 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.527 GB | free=3.772 GB
725
+ 2026-11-01 21:07:16 | INFO | STEP 41 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.447 GB | free=3.827 GB
726
+ 2026-11-01 21:07:35 | INFO | --------------------------------------------------------------------------------
727
+ 2026-11-01 21:07:35 | INFO | STEP 42/70 | dataset_item=4 | repeat=2/10
728
+ 2026-11-01 21:07:35 | INFO | LOSS: 1.16119850
729
+ 2026-11-01 21:07:35 | INFO | LR: 1.88888889e-04
730
+ 2026-11-01 21:07:35 | INFO | GLOBAL GRADIENT NORM: 0.08981757
731
+ 2026-11-01 21:07:35 | INFO | MAX ABSOLUTE GRADIENT: 6.45065308e-03
732
+ 2026-11-01 21:07:35 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
733
+ 2026-11-01 21:07:35 | INFO | MISSING GRADIENT TENSORS: 0
734
+ 2026-11-01 21:07:35 | INFO | GPU0 FORWARD: 2.115s
735
+ 2026-11-01 21:07:35 | INFO | GPU1 FORWARD: 2.794s
736
+ 2026-11-01 21:07:35 | INFO | GPU1 BACKWARD: 7.328s
737
+ 2026-11-01 21:07:35 | INFO | GPU0 BACKWARD: 5.394s
738
+ 2026-11-01 21:07:35 | INFO | UPDATE WALL TIME: 5.584s
739
+ 2026-11-01 21:07:35 | INFO | ELAPSED TRAINING TIME: 758.92s
740
+ 2026-11-01 21:07:35 | INFO | STEP 42 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.527 GB | free=3.772 GB
741
+ 2026-11-01 21:07:35 | INFO | STEP 42 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.447 GB | free=3.827 GB
742
+ 2026-11-01 21:07:53 | INFO | --------------------------------------------------------------------------------
743
+ 2026-11-01 21:07:53 | INFO | STEP 43/70 | dataset_item=4 | repeat=3/10
744
+ 2026-11-01 21:07:53 | INFO | LOSS: 1.12029767
745
+ 2026-11-01 21:07:53 | INFO | LR: 1.77777778e-04
746
+ 2026-11-01 21:07:53 | INFO | GLOBAL GRADIENT NORM: 0.08757077
747
+ 2026-11-01 21:07:53 | INFO | MAX ABSOLUTE GRADIENT: 6.22558594e-03
748
+ 2026-11-01 21:07:53 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
749
+ 2026-11-01 21:07:53 | INFO | MISSING GRADIENT TENSORS: 0
750
+ 2026-11-01 21:07:53 | INFO | GPU0 FORWARD: 2.117s
751
+ 2026-11-01 21:07:53 | INFO | GPU1 FORWARD: 2.791s
752
+ 2026-11-01 21:07:53 | INFO | GPU1 BACKWARD: 7.312s
753
+ 2026-11-01 21:07:53 | INFO | GPU0 BACKWARD: 5.395s
754
+ 2026-11-01 21:07:53 | INFO | UPDATE WALL TIME: 5.575s
755
+ 2026-11-01 21:07:53 | INFO | ELAPSED TRAINING TIME: 777.05s
756
+ 2026-11-01 21:07:53 | INFO | STEP 43 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.527 GB | free=3.772 GB
757
+ 2026-11-01 21:07:53 | INFO | STEP 43 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.447 GB | free=3.827 GB
758
+ 2026-11-01 21:08:11 | INFO | --------------------------------------------------------------------------------
759
+ 2026-11-01 21:08:11 | INFO | STEP 44/70 | dataset_item=4 | repeat=4/10
760
+ 2026-11-01 21:08:11 | INFO | LOSS: 0.66419220
761
+ 2026-11-01 21:08:11 | INFO | LR: 1.66666667e-04
762
+ 2026-11-01 21:08:11 | INFO | GLOBAL GRADIENT NORM: 0.07564558
763
+ 2026-11-01 21:08:11 | INFO | MAX ABSOLUTE GRADIENT: 3.40843201e-03
764
+ 2026-11-01 21:08:11 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
765
+ 2026-11-01 21:08:11 | INFO | MISSING GRADIENT TENSORS: 0
766
+ 2026-11-01 21:08:11 | INFO | GPU0 FORWARD: 2.103s
767
+ 2026-11-01 21:08:11 | INFO | GPU1 FORWARD: 2.776s
768
+ 2026-11-01 21:08:11 | INFO | GPU1 BACKWARD: 7.334s
769
+ 2026-11-01 21:08:11 | INFO | GPU0 BACKWARD: 5.409s
770
+ 2026-11-01 21:08:11 | INFO | UPDATE WALL TIME: 5.596s
771
+ 2026-11-01 21:08:11 | INFO | ELAPSED TRAINING TIME: 795.17s
772
+ 2026-11-01 21:08:11 | INFO | STEP 44 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.527 GB | free=3.772 GB
773
+ 2026-11-01 21:08:11 | INFO | STEP 44 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.447 GB | free=3.827 GB
774
+ 2026-11-01 21:08:29 | INFO | --------------------------------------------------------------------------------
775
+ 2026-11-01 21:08:29 | INFO | STEP 45/70 | dataset_item=4 | repeat=5/10
776
+ 2026-11-01 21:08:29 | INFO | LOSS: 0.59954333
777
+ 2026-11-01 21:08:29 | INFO | LR: 1.55555556e-04
778
+ 2026-11-01 21:08:29 | INFO | GLOBAL GRADIENT NORM: 0.05642530
779
+ 2026-11-01 21:08:29 | INFO | MAX ABSOLUTE GRADIENT: 3.06510925e-03
780
+ 2026-11-01 21:08:29 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
781
+ 2026-11-01 21:08:29 | INFO | MISSING GRADIENT TENSORS: 0
782
+ 2026-11-01 21:08:29 | INFO | GPU0 FORWARD: 2.108s
783
+ 2026-11-01 21:08:29 | INFO | GPU1 FORWARD: 2.766s
784
+ 2026-11-01 21:08:29 | INFO | GPU1 BACKWARD: 7.339s
785
+ 2026-11-01 21:08:29 | INFO | GPU0 BACKWARD: 5.407s
786
+ 2026-11-01 21:08:29 | INFO | UPDATE WALL TIME: 5.589s
787
+ 2026-11-01 21:08:29 | INFO | ELAPSED TRAINING TIME: 813.28s
788
+ 2026-11-01 21:08:29 | INFO | STEP 45 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.527 GB | free=3.772 GB
789
+ 2026-11-01 21:08:29 | INFO | STEP 45 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.447 GB | free=3.827 GB
790
+ 2026-11-01 21:08:47 | INFO | --------------------------------------------------------------------------------
791
+ 2026-11-01 21:08:47 | INFO | STEP 46/70 | dataset_item=4 | repeat=6/10
792
+ 2026-11-01 21:08:47 | INFO | LOSS: 1.11565304
793
+ 2026-11-01 21:08:47 | INFO | LR: 1.44444444e-04
794
+ 2026-11-01 21:08:47 | INFO | GLOBAL GRADIENT NORM: 0.36519137
795
+ 2026-11-01 21:08:47 | INFO | MAX ABSOLUTE GRADIENT: 2.91900635e-02
796
+ 2026-11-01 21:08:47 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
797
+ 2026-11-01 21:08:47 | INFO | MISSING GRADIENT TENSORS: 0
798
+ 2026-11-01 21:08:47 | INFO | GPU0 FORWARD: 2.128s
799
+ 2026-11-01 21:08:47 | INFO | GPU1 FORWARD: 2.814s
800
+ 2026-11-01 21:08:47 | INFO | GPU1 BACKWARD: 7.346s
801
+ 2026-11-01 21:08:47 | INFO | GPU0 BACKWARD: 5.461s
802
+ 2026-11-01 21:08:47 | INFO | UPDATE WALL TIME: 5.646s
803
+ 2026-11-01 21:08:47 | INFO | ELAPSED TRAINING TIME: 831.53s
804
+ 2026-11-01 21:08:47 | INFO | STEP 46 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.527 GB | free=3.772 GB
805
+ 2026-11-01 21:08:47 | INFO | STEP 46 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.447 GB | free=3.827 GB
806
+ 2026-11-01 21:09:05 | INFO | --------------------------------------------------------------------------------
807
+ 2026-11-01 21:09:05 | INFO | STEP 47/70 | dataset_item=4 | repeat=7/10
808
+ 2026-11-01 21:09:05 | INFO | LOSS: 0.53677845
809
+ 2026-11-01 21:09:05 | INFO | LR: 1.33333333e-04
810
+ 2026-11-01 21:09:05 | INFO | GLOBAL GRADIENT NORM: 0.06391268
811
+ 2026-11-01 21:09:05 | INFO | MAX ABSOLUTE GRADIENT: 3.41606140e-03
812
+ 2026-11-01 21:09:05 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
813
+ 2026-11-01 21:09:05 | INFO | MISSING GRADIENT TENSORS: 0
814
+ 2026-11-01 21:09:05 | INFO | GPU0 FORWARD: 2.103s
815
+ 2026-11-01 21:09:05 | INFO | GPU1 FORWARD: 2.797s
816
+ 2026-11-01 21:09:05 | INFO | GPU1 BACKWARD: 7.369s
817
+ 2026-11-01 21:09:05 | INFO | GPU0 BACKWARD: 5.405s
818
+ 2026-11-01 21:09:05 | INFO | UPDATE WALL TIME: 5.594s
819
+ 2026-11-01 21:09:05 | INFO | ELAPSED TRAINING TIME: 849.72s
820
+ 2026-11-01 21:09:05 | INFO | STEP 47 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.527 GB | free=3.772 GB
821
+ 2026-11-01 21:09:05 | INFO | STEP 47 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.447 GB | free=3.827 GB
822
+ 2026-11-01 21:09:24 | INFO | --------------------------------------------------------------------------------
823
+ 2026-11-01 21:09:24 | INFO | STEP 48/70 | dataset_item=4 | repeat=8/10
824
+ 2026-11-01 21:09:24 | INFO | LOSS: 0.50866330
825
+ 2026-11-01 21:09:24 | INFO | LR: 1.22222222e-04
826
+ 2026-11-01 21:09:24 | INFO | GLOBAL GRADIENT NORM: 0.07756938
827
+ 2026-11-01 21:09:24 | INFO | MAX ABSOLUTE GRADIENT: 4.65774536e-03
828
+ 2026-11-01 21:09:24 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
829
+ 2026-11-01 21:09:24 | INFO | MISSING GRADIENT TENSORS: 0
830
+ 2026-11-01 21:09:24 | INFO | GPU0 FORWARD: 2.119s
831
+ 2026-11-01 21:09:24 | INFO | GPU1 FORWARD: 2.818s
832
+ 2026-11-01 21:09:24 | INFO | GPU1 BACKWARD: 7.446s
833
+ 2026-11-01 21:09:24 | INFO | GPU0 BACKWARD: 5.378s
834
+ 2026-11-01 21:09:24 | INFO | UPDATE WALL TIME: 5.563s
835
+ 2026-11-01 21:09:24 | INFO | ELAPSED TRAINING TIME: 867.99s
836
+ 2026-11-01 21:09:24 | INFO | STEP 48 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.527 GB | free=3.772 GB
837
+ 2026-11-01 21:09:24 | INFO | STEP 48 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.447 GB | free=3.827 GB
838
+ 2026-11-01 21:09:42 | INFO | --------------------------------------------------------------------------------
839
+ 2026-11-01 21:09:42 | INFO | STEP 49/70 | dataset_item=4 | repeat=9/10
840
+ 2026-11-01 21:09:42 | INFO | LOSS: 1.22806942
841
+ 2026-11-01 21:09:42 | INFO | LR: 1.11111111e-04
842
+ 2026-11-01 21:09:42 | INFO | GLOBAL GRADIENT NORM: 0.33487317
843
+ 2026-11-01 21:09:42 | INFO | MAX ABSOLUTE GRADIENT: 1.95159912e-02
844
+ 2026-11-01 21:09:42 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
845
+ 2026-11-01 21:09:42 | INFO | MISSING GRADIENT TENSORS: 0
846
+ 2026-11-01 21:09:42 | INFO | GPU0 FORWARD: 2.096s
847
+ 2026-11-01 21:09:42 | INFO | GPU1 FORWARD: 2.766s
848
+ 2026-11-01 21:09:42 | INFO | GPU1 BACKWARD: 7.267s
849
+ 2026-11-01 21:09:42 | INFO | GPU0 BACKWARD: 5.426s
850
+ 2026-11-01 21:09:42 | INFO | UPDATE WALL TIME: 5.612s
851
+ 2026-11-01 21:09:42 | INFO | ELAPSED TRAINING TIME: 886.04s
852
+ 2026-11-01 21:09:42 | INFO | STEP 49 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.527 GB | free=3.772 GB
853
+ 2026-11-01 21:09:42 | INFO | STEP 49 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.447 GB | free=3.827 GB
854
+ 2026-11-01 21:10:00 | INFO | --------------------------------------------------------------------------------
855
+ 2026-11-01 21:10:00 | INFO | STEP 50/70 | dataset_item=4 | repeat=10/10
856
+ 2026-11-01 21:10:00 | INFO | LOSS: 1.09903264
857
+ 2026-11-01 21:10:00 | INFO | LR: 1.00000000e-04
858
+ 2026-11-01 21:10:00 | INFO | GLOBAL GRADIENT NORM: 0.12791686
859
+ 2026-11-01 21:10:00 | INFO | MAX ABSOLUTE GRADIENT: 1.35421753e-02
860
+ 2026-11-01 21:10:00 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
861
+ 2026-11-01 21:10:00 | INFO | MISSING GRADIENT TENSORS: 0
862
+ 2026-11-01 21:10:00 | INFO | GPU0 FORWARD: 2.116s
863
+ 2026-11-01 21:10:00 | INFO | GPU1 FORWARD: 2.812s
864
+ 2026-11-01 21:10:00 | INFO | GPU1 BACKWARD: 7.410s
865
+ 2026-11-01 21:10:00 | INFO | GPU0 BACKWARD: 5.416s
866
+ 2026-11-01 21:10:00 | INFO | UPDATE WALL TIME: 5.598s
867
+ 2026-11-01 21:10:00 | INFO | ELAPSED TRAINING TIME: 904.29s
868
+ 2026-11-01 21:10:00 | INFO | STEP 50 | GPU0 | allocated=5.070 GB | reserved=10.652 GB | peak=10.527 GB | free=3.772 GB
869
+ 2026-11-01 21:10:00 | INFO | STEP 50 | GPU1 | allocated=4.984 GB | reserved=10.598 GB | peak=10.447 GB | free=3.827 GB
870
+ 2026-11-01 21:10:00 | INFO | COMPLETED DATASET ITEM 5/7
871
+ 2026-11-01 21:10:00 | INFO | --------------------------------------------------------------------------------
872
+ 2026-11-01 21:10:00 | INFO | STARTING DATASET ITEM 6/7 | file=00000005.pt
873
+ 2026-11-01 21:10:00 | INFO | LEARNING RATE RESET: 2.00000000e-04
874
+ 2026-11-01 21:10:18 | INFO | --------------------------------------------------------------------------------
875
+ 2026-11-01 21:10:18 | INFO | STEP 51/70 | dataset_item=5 | repeat=1/10
876
+ 2026-11-01 21:10:18 | INFO | LOSS: 0.54666841
877
+ 2026-11-01 21:10:18 | INFO | LR: 2.00000000e-04
878
+ 2026-11-01 21:10:18 | INFO | GLOBAL GRADIENT NORM: 0.10908916
879
+ 2026-11-01 21:10:18 | INFO | MAX ABSOLUTE GRADIENT: 5.75637817e-03
880
+ 2026-11-01 21:10:18 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
881
+ 2026-11-01 21:10:18 | INFO | MISSING GRADIENT TENSORS: 0
882
+ 2026-11-01 21:10:18 | INFO | GPU0 FORWARD: 2.145s
883
+ 2026-11-01 21:10:18 | INFO | GPU1 FORWARD: 2.827s
884
+ 2026-11-01 21:10:18 | INFO | GPU1 BACKWARD: 7.488s
885
+ 2026-11-01 21:10:18 | INFO | GPU0 BACKWARD: 5.516s
886
+ 2026-11-01 21:10:18 | INFO | UPDATE WALL TIME: 5.701s
887
+ 2026-11-01 21:10:18 | INFO | ELAPSED TRAINING TIME: 922.76s
888
+ 2026-11-01 21:10:18 | INFO | STEP 51 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.579 GB | free=3.753 GB
889
+ 2026-11-01 21:10:18 | INFO | STEP 51 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.499 GB | free=3.807 GB
890
+ 2026-11-01 21:10:37 | INFO | --------------------------------------------------------------------------------
891
+ 2026-11-01 21:10:37 | INFO | STEP 52/70 | dataset_item=5 | repeat=2/10
892
+ 2026-11-01 21:10:37 | INFO | LOSS: 0.58535677
893
+ 2026-11-01 21:10:37 | INFO | LR: 1.88888889e-04
894
+ 2026-11-01 21:10:37 | INFO | GLOBAL GRADIENT NORM: 0.09663414
895
+ 2026-11-01 21:10:37 | INFO | MAX ABSOLUTE GRADIENT: 3.99017334e-03
896
+ 2026-11-01 21:10:37 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
897
+ 2026-11-01 21:10:37 | INFO | MISSING GRADIENT TENSORS: 0
898
+ 2026-11-01 21:10:37 | INFO | GPU0 FORWARD: 2.161s
899
+ 2026-11-01 21:10:37 | INFO | GPU1 FORWARD: 2.876s
900
+ 2026-11-01 21:10:37 | INFO | GPU1 BACKWARD: 7.577s
901
+ 2026-11-01 21:10:37 | INFO | GPU0 BACKWARD: 5.504s
902
+ 2026-11-01 21:10:37 | INFO | UPDATE WALL TIME: 5.687s
903
+ 2026-11-01 21:10:37 | INFO | ELAPSED TRAINING TIME: 941.39s
904
+ 2026-11-01 21:10:37 | INFO | STEP 52 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.579 GB | free=3.753 GB
905
+ 2026-11-01 21:10:37 | INFO | STEP 52 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.499 GB | free=3.807 GB
906
+ 2026-11-01 21:10:56 | INFO | --------------------------------------------------------------------------------
907
+ 2026-11-01 21:10:56 | INFO | STEP 53/70 | dataset_item=5 | repeat=3/10
908
+ 2026-11-01 21:10:56 | INFO | LOSS: 1.40310705
909
+ 2026-11-01 21:10:56 | INFO | LR: 1.77777778e-04
910
+ 2026-11-01 21:10:56 | INFO | GLOBAL GRADIENT NORM: 2.31765136
911
+ 2026-11-01 21:10:56 | INFO | MAX ABSOLUTE GRADIENT: 1.33178711e-01
912
+ 2026-11-01 21:10:56 | INFO | GRADIENT CLIP COEFFICIENT: 0.43147109
913
+ 2026-11-01 21:10:56 | INFO | MISSING GRADIENT TENSORS: 0
914
+ 2026-11-01 21:10:56 | INFO | GPU0 FORWARD: 2.144s
915
+ 2026-11-01 21:10:56 | INFO | GPU1 FORWARD: 2.838s
916
+ 2026-11-01 21:10:56 | INFO | GPU1 BACKWARD: 7.473s
917
+ 2026-11-01 21:10:56 | INFO | GPU0 BACKWARD: 5.587s
918
+ 2026-11-01 21:10:56 | INFO | UPDATE WALL TIME: 5.775s
919
+ 2026-11-01 21:10:56 | INFO | ELAPSED TRAINING TIME: 959.94s
920
+ 2026-11-01 21:10:56 | INFO | STEP 53 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.579 GB | free=3.753 GB
921
+ 2026-11-01 21:10:56 | INFO | STEP 53 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.499 GB | free=3.807 GB
922
+ 2026-11-01 21:11:14 | INFO | --------------------------------------------------------------------------------
923
+ 2026-11-01 21:11:14 | INFO | STEP 54/70 | dataset_item=5 | repeat=4/10
924
+ 2026-11-01 21:11:14 | INFO | LOSS: 1.49102020
925
+ 2026-11-01 21:11:14 | INFO | LR: 1.66666667e-04
926
+ 2026-11-01 21:11:14 | INFO | GLOBAL GRADIENT NORM: 2.78661622
927
+ 2026-11-01 21:11:14 | INFO | MAX ABSOLUTE GRADIENT: 1.79077148e-01
928
+ 2026-11-01 21:11:14 | INFO | GRADIENT CLIP COEFFICIENT: 0.35885804
929
+ 2026-11-01 21:11:14 | INFO | MISSING GRADIENT TENSORS: 0
930
+ 2026-11-01 21:11:14 | INFO | GPU0 FORWARD: 2.146s
931
+ 2026-11-01 21:11:14 | INFO | GPU1 FORWARD: 2.813s
932
+ 2026-11-01 21:11:14 | INFO | GPU1 BACKWARD: 7.442s
933
+ 2026-11-01 21:11:14 | INFO | GPU0 BACKWARD: 5.609s
934
+ 2026-11-01 21:11:14 | INFO | UPDATE WALL TIME: 5.799s
935
+ 2026-11-01 21:11:14 | INFO | ELAPSED TRAINING TIME: 978.47s
936
+ 2026-11-01 21:11:14 | INFO | STEP 54 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.579 GB | free=3.753 GB
937
+ 2026-11-01 21:11:14 | INFO | STEP 54 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.499 GB | free=3.807 GB
938
+ 2026-11-01 21:11:33 | INFO | --------------------------------------------------------------------------------
939
+ 2026-11-01 21:11:33 | INFO | STEP 55/70 | dataset_item=5 | repeat=5/10
940
+ 2026-11-01 21:11:33 | INFO | LOSS: 0.54775751
941
+ 2026-11-01 21:11:33 | INFO | LR: 1.55555556e-04
942
+ 2026-11-01 21:11:33 | INFO | GLOBAL GRADIENT NORM: 0.12621702
943
+ 2026-11-01 21:11:33 | INFO | MAX ABSOLUTE GRADIENT: 7.01141357e-03
944
+ 2026-11-01 21:11:33 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
945
+ 2026-11-01 21:11:33 | INFO | MISSING GRADIENT TENSORS: 0
946
+ 2026-11-01 21:11:33 | INFO | GPU0 FORWARD: 2.150s
947
+ 2026-11-01 21:11:33 | INFO | GPU1 FORWARD: 2.851s
948
+ 2026-11-01 21:11:33 | INFO | GPU1 BACKWARD: 7.583s
949
+ 2026-11-01 21:11:33 | INFO | GPU0 BACKWARD: 5.516s
950
+ 2026-11-01 21:11:33 | INFO | UPDATE WALL TIME: 5.699s
951
+ 2026-11-01 21:11:33 | INFO | ELAPSED TRAINING TIME: 997.07s
952
+ 2026-11-01 21:11:33 | INFO | STEP 55 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.579 GB | free=3.753 GB
953
+ 2026-11-01 21:11:33 | INFO | STEP 55 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.499 GB | free=3.807 GB
954
+ 2026-11-01 21:11:51 | INFO | --------------------------------------------------------------------------------
955
+ 2026-11-01 21:11:51 | INFO | STEP 56/70 | dataset_item=5 | repeat=6/10
956
+ 2026-11-01 21:11:51 | INFO | LOSS: 0.55500561
957
+ 2026-11-01 21:11:51 | INFO | LR: 1.44444444e-04
958
+ 2026-11-01 21:11:51 | INFO | GLOBAL GRADIENT NORM: 0.13330906
959
+ 2026-11-01 21:11:51 | INFO | MAX ABSOLUTE GRADIENT: 9.23919678e-03
960
+ 2026-11-01 21:11:51 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
961
+ 2026-11-01 21:11:51 | INFO | MISSING GRADIENT TENSORS: 0
962
+ 2026-11-01 21:11:51 | INFO | GPU0 FORWARD: 2.150s
963
+ 2026-11-01 21:11:51 | INFO | GPU1 FORWARD: 2.865s
964
+ 2026-11-01 21:11:51 | INFO | GPU1 BACKWARD: 7.564s
965
+ 2026-11-01 21:11:51 | INFO | GPU0 BACKWARD: 5.513s
966
+ 2026-11-01 21:11:51 | INFO | UPDATE WALL TIME: 5.702s
967
+ 2026-11-01 21:11:51 | INFO | ELAPSED TRAINING TIME: 1015.66s
968
+ 2026-11-01 21:11:51 | INFO | STEP 56 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.579 GB | free=3.753 GB
969
+ 2026-11-01 21:11:51 | INFO | STEP 56 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.499 GB | free=3.807 GB
970
+ 2026-11-01 21:12:10 | INFO | --------------------------------------------------------------------------------
971
+ 2026-11-01 21:12:10 | INFO | STEP 57/70 | dataset_item=5 | repeat=7/10
972
+ 2026-11-01 21:12:10 | INFO | LOSS: 2.26420903
973
+ 2026-11-01 21:12:10 | INFO | LR: 1.33333333e-04
974
+ 2026-11-01 21:12:10 | INFO | GLOBAL GRADIENT NORM: 6.06166684
975
+ 2026-11-01 21:12:10 | INFO | MAX ABSOLUTE GRADIENT: 7.66113281e-01
976
+ 2026-11-01 21:12:10 | INFO | GRADIENT CLIP COEFFICIENT: 0.16497110
977
+ 2026-11-01 21:12:10 | INFO | MISSING GRADIENT TENSORS: 0
978
+ 2026-11-01 21:12:10 | INFO | GPU0 FORWARD: 2.133s
979
+ 2026-11-01 21:12:10 | INFO | GPU1 FORWARD: 2.833s
980
+ 2026-11-01 21:12:10 | INFO | GPU1 BACKWARD: 7.485s
981
+ 2026-11-01 21:12:10 | INFO | GPU0 BACKWARD: 5.600s
982
+ 2026-11-01 21:12:10 | INFO | UPDATE WALL TIME: 5.793s
983
+ 2026-11-01 21:12:10 | INFO | ELAPSED TRAINING TIME: 1034.23s
984
+ 2026-11-01 21:12:10 | INFO | STEP 57 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.579 GB | free=3.753 GB
985
+ 2026-11-01 21:12:10 | INFO | STEP 57 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.499 GB | free=3.807 GB
986
+ 2026-11-01 21:12:28 | INFO | --------------------------------------------------------------------------------
987
+ 2026-11-01 21:12:28 | INFO | STEP 58/70 | dataset_item=5 | repeat=8/10
988
+ 2026-11-01 21:12:28 | INFO | LOSS: 0.56494367
989
+ 2026-11-01 21:12:28 | INFO | LR: 1.22222222e-04
990
+ 2026-11-01 21:12:28 | INFO | GLOBAL GRADIENT NORM: 0.16257450
991
+ 2026-11-01 21:12:28 | INFO | MAX ABSOLUTE GRADIENT: 1.17721558e-02
992
+ 2026-11-01 21:12:28 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
993
+ 2026-11-01 21:12:28 | INFO | MISSING GRADIENT TENSORS: 0
994
+ 2026-11-01 21:12:28 | INFO | GPU0 FORWARD: 2.136s
995
+ 2026-11-01 21:12:28 | INFO | GPU1 FORWARD: 2.821s
996
+ 2026-11-01 21:12:28 | INFO | GPU1 BACKWARD: 7.464s
997
+ 2026-11-01 21:12:28 | INFO | GPU0 BACKWARD: 5.495s
998
+ 2026-11-01 21:12:28 | INFO | UPDATE WALL TIME: 5.682s
999
+ 2026-11-01 21:12:28 | INFO | ELAPSED TRAINING TIME: 1052.65s
1000
+ 2026-11-01 21:12:28 | INFO | STEP 58 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.579 GB | free=3.753 GB
1001
+ 2026-11-01 21:12:28 | INFO | STEP 58 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.499 GB | free=3.807 GB
1002
+ 2026-11-01 21:12:47 | INFO | --------------------------------------------------------------------------------
1003
+ 2026-11-01 21:12:47 | INFO | STEP 59/70 | dataset_item=5 | repeat=9/10
1004
+ 2026-11-01 21:12:47 | INFO | LOSS: 1.76634634
1005
+ 2026-11-01 21:12:47 | INFO | LR: 1.11111111e-04
1006
+ 2026-11-01 21:12:47 | INFO | GLOBAL GRADIENT NORM: 7.71759420
1007
+ 2026-11-01 21:12:47 | INFO | MAX ABSOLUTE GRADIENT: 4.01367188e-01
1008
+ 2026-11-01 21:12:47 | INFO | GRADIENT CLIP COEFFICIENT: 0.12957404
1009
+ 2026-11-01 21:12:47 | INFO | MISSING GRADIENT TENSORS: 0
1010
+ 2026-11-01 21:12:47 | INFO | GPU0 FORWARD: 2.136s
1011
+ 2026-11-01 21:12:47 | INFO | GPU1 FORWARD: 2.816s
1012
+ 2026-11-01 21:12:47 | INFO | GPU1 BACKWARD: 7.497s
1013
+ 2026-11-01 21:12:47 | INFO | GPU0 BACKWARD: 5.643s
1014
+ 2026-11-01 21:12:47 | INFO | UPDATE WALL TIME: 5.831s
1015
+ 2026-11-01 21:12:47 | INFO | ELAPSED TRAINING TIME: 1071.24s
1016
+ 2026-11-01 21:12:47 | INFO | STEP 59 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.579 GB | free=3.753 GB
1017
+ 2026-11-01 21:12:47 | INFO | STEP 59 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.499 GB | free=3.807 GB
1018
+ 2026-11-01 21:13:06 | INFO | --------------------------------------------------------------------------------
1019
+ 2026-11-01 21:13:06 | INFO | STEP 60/70 | dataset_item=5 | repeat=10/10
1020
+ 2026-11-01 21:13:06 | INFO | LOSS: 1.66516125
1021
+ 2026-11-01 21:13:06 | INFO | LR: 1.00000000e-04
1022
+ 2026-11-01 21:13:06 | INFO | GLOBAL GRADIENT NORM: 5.53647328
1023
+ 2026-11-01 21:13:06 | INFO | MAX ABSOLUTE GRADIENT: 6.27929688e-01
1024
+ 2026-11-01 21:13:06 | INFO | GRADIENT CLIP COEFFICIENT: 0.18062036
1025
+ 2026-11-01 21:13:06 | INFO | MISSING GRADIENT TENSORS: 0
1026
+ 2026-11-01 21:13:06 | INFO | GPU0 FORWARD: 2.142s
1027
+ 2026-11-01 21:13:06 | INFO | GPU1 FORWARD: 2.821s
1028
+ 2026-11-01 21:13:06 | INFO | GPU1 BACKWARD: 7.617s
1029
+ 2026-11-01 21:13:06 | INFO | GPU0 BACKWARD: 5.638s
1030
+ 2026-11-01 21:13:06 | INFO | UPDATE WALL TIME: 5.829s
1031
+ 2026-11-01 21:13:06 | INFO | ELAPSED TRAINING TIME: 1089.97s
1032
+ 2026-11-01 21:13:06 | INFO | STEP 60 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.579 GB | free=3.753 GB
1033
+ 2026-11-01 21:13:06 | INFO | STEP 60 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.499 GB | free=3.807 GB
1034
+ 2026-11-01 21:13:06 | INFO | COMPLETED DATASET ITEM 6/7
1035
+ 2026-11-01 21:13:06 | INFO | --------------------------------------------------------------------------------
1036
+ 2026-11-01 21:13:06 | INFO | STARTING DATASET ITEM 7/7 | file=00000006.pt
1037
+ 2026-11-01 21:13:06 | INFO | LEARNING RATE RESET: 2.00000000e-04
1038
+ 2026-11-01 21:13:24 | INFO | --------------------------------------------------------------------------------
1039
+ 2026-11-01 21:13:24 | INFO | STEP 61/70 | dataset_item=6 | repeat=1/10
1040
+ 2026-11-01 21:13:24 | INFO | LOSS: 1.85172176
1041
+ 2026-11-01 21:13:24 | INFO | LR: 2.00000000e-04
1042
+ 2026-11-01 21:13:24 | INFO | GLOBAL GRADIENT NORM: 4.37498812
1043
+ 2026-11-01 21:13:24 | INFO | MAX ABSOLUTE GRADIENT: 3.04687500e-01
1044
+ 2026-11-01 21:13:24 | INFO | GRADIENT CLIP COEFFICIENT: 0.22857200
1045
+ 2026-11-01 21:13:24 | INFO | MISSING GRADIENT TENSORS: 0
1046
+ 2026-11-01 21:13:24 | INFO | GPU0 FORWARD: 2.106s
1047
+ 2026-11-01 21:13:24 | INFO | GPU1 FORWARD: 2.787s
1048
+ 2026-11-01 21:13:24 | INFO | GPU1 BACKWARD: 7.383s
1049
+ 2026-11-01 21:13:24 | INFO | GPU0 BACKWARD: 5.491s
1050
+ 2026-11-01 21:13:24 | INFO | UPDATE WALL TIME: 5.679s
1051
+ 2026-11-01 21:13:24 | INFO | ELAPSED TRAINING TIME: 1108.24s
1052
+ 2026-11-01 21:13:24 | INFO | STEP 61 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.541 GB | free=3.753 GB
1053
+ 2026-11-01 21:13:24 | INFO | STEP 61 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.461 GB | free=3.807 GB
1054
+ 2026-11-01 21:13:42 | INFO | --------------------------------------------------------------------------------
1055
+ 2026-11-01 21:13:42 | INFO | STEP 62/70 | dataset_item=6 | repeat=2/10
1056
+ 2026-11-01 21:13:42 | INFO | LOSS: 1.81541848
1057
+ 2026-11-01 21:13:42 | INFO | LR: 1.88888889e-04
1058
+ 2026-11-01 21:13:42 | INFO | GLOBAL GRADIENT NORM: 4.61086091
1059
+ 2026-11-01 21:13:42 | INFO | MAX ABSOLUTE GRADIENT: 2.19604492e-01
1060
+ 2026-11-01 21:13:42 | INFO | GRADIENT CLIP COEFFICIENT: 0.21687919
1061
+ 2026-11-01 21:13:42 | INFO | MISSING GRADIENT TENSORS: 0
1062
+ 2026-11-01 21:13:42 | INFO | GPU0 FORWARD: 2.091s
1063
+ 2026-11-01 21:13:42 | INFO | GPU1 FORWARD: 2.733s
1064
+ 2026-11-01 21:13:42 | INFO | GPU1 BACKWARD: 7.300s
1065
+ 2026-11-01 21:13:42 | INFO | GPU0 BACKWARD: 5.537s
1066
+ 2026-11-01 21:13:42 | INFO | UPDATE WALL TIME: 5.725s
1067
+ 2026-11-01 21:13:42 | INFO | ELAPSED TRAINING TIME: 1126.40s
1068
+ 2026-11-01 21:13:42 | INFO | STEP 62 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.541 GB | free=3.753 GB
1069
+ 2026-11-01 21:13:42 | INFO | STEP 62 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.461 GB | free=3.807 GB
1070
+ 2026-11-01 21:14:00 | INFO | --------------------------------------------------------------------------------
1071
+ 2026-11-01 21:14:00 | INFO | STEP 63/70 | dataset_item=6 | repeat=3/10
1072
+ 2026-11-01 21:14:00 | INFO | LOSS: 1.22226369
1073
+ 2026-11-01 21:14:00 | INFO | LR: 1.77777778e-04
1074
+ 2026-11-01 21:14:00 | INFO | GLOBAL GRADIENT NORM: 1.18589308
1075
+ 2026-11-01 21:14:00 | INFO | MAX ABSOLUTE GRADIENT: 6.50024414e-02
1076
+ 2026-11-01 21:14:00 | INFO | GRADIENT CLIP COEFFICIENT: 0.84324563
1077
+ 2026-11-01 21:14:00 | INFO | MISSING GRADIENT TENSORS: 0
1078
+ 2026-11-01 21:14:00 | INFO | GPU0 FORWARD: 2.103s
1079
+ 2026-11-01 21:14:00 | INFO | GPU1 FORWARD: 2.811s
1080
+ 2026-11-01 21:14:00 | INFO | GPU1 BACKWARD: 7.431s
1081
+ 2026-11-01 21:14:00 | INFO | GPU0 BACKWARD: 5.487s
1082
+ 2026-11-01 21:14:00 | INFO | UPDATE WALL TIME: 5.675s
1083
+ 2026-11-01 21:14:00 | INFO | ELAPSED TRAINING TIME: 1144.74s
1084
+ 2026-11-01 21:14:00 | INFO | STEP 63 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.541 GB | free=3.753 GB
1085
+ 2026-11-01 21:14:00 | INFO | STEP 63 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.461 GB | free=3.807 GB
1086
+ 2026-11-01 21:14:19 | INFO | --------------------------------------------------------------------------------
1087
+ 2026-11-01 21:14:19 | INFO | STEP 64/70 | dataset_item=6 | repeat=4/10
1088
+ 2026-11-01 21:14:19 | INFO | LOSS: 0.91928631
1089
+ 2026-11-01 21:14:19 | INFO | LR: 1.66666667e-04
1090
+ 2026-11-01 21:14:19 | INFO | GLOBAL GRADIENT NORM: 0.53044180
1091
+ 2026-11-01 21:14:19 | INFO | MAX ABSOLUTE GRADIENT: 5.35278320e-02
1092
+ 2026-11-01 21:14:19 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
1093
+ 2026-11-01 21:14:19 | INFO | MISSING GRADIENT TENSORS: 0
1094
+ 2026-11-01 21:14:19 | INFO | GPU0 FORWARD: 2.116s
1095
+ 2026-11-01 21:14:19 | INFO | GPU1 FORWARD: 2.821s
1096
+ 2026-11-01 21:14:19 | INFO | GPU1 BACKWARD: 7.381s
1097
+ 2026-11-01 21:14:19 | INFO | GPU0 BACKWARD: 5.447s
1098
+ 2026-11-01 21:14:19 | INFO | UPDATE WALL TIME: 5.631s
1099
+ 2026-11-01 21:14:19 | INFO | ELAPSED TRAINING TIME: 1163.01s
1100
+ 2026-11-01 21:14:19 | INFO | STEP 64 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.541 GB | free=3.753 GB
1101
+ 2026-11-01 21:14:19 | INFO | STEP 64 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.461 GB | free=3.807 GB
1102
+ 2026-11-01 21:14:37 | INFO | --------------------------------------------------------------------------------
1103
+ 2026-11-01 21:14:37 | INFO | STEP 65/70 | dataset_item=6 | repeat=5/10
1104
+ 2026-11-01 21:14:37 | INFO | LOSS: 2.73170328
1105
+ 2026-11-01 21:14:37 | INFO | LR: 1.55555556e-04
1106
+ 2026-11-01 21:14:37 | INFO | GLOBAL GRADIENT NORM: 3.00700506
1107
+ 2026-11-01 21:14:37 | INFO | MAX ABSOLUTE GRADIENT: 1.97265625e-01
1108
+ 2026-11-01 21:14:37 | INFO | GRADIENT CLIP COEFFICIENT: 0.33255670
1109
+ 2026-11-01 21:14:37 | INFO | MISSING GRADIENT TENSORS: 0
1110
+ 2026-11-01 21:14:37 | INFO | GPU0 FORWARD: 2.103s
1111
+ 2026-11-01 21:14:37 | INFO | GPU1 FORWARD: 2.757s
1112
+ 2026-11-01 21:14:37 | INFO | GPU1 BACKWARD: 7.331s
1113
+ 2026-11-01 21:14:37 | INFO | GPU0 BACKWARD: 5.534s
1114
+ 2026-11-01 21:14:37 | INFO | UPDATE WALL TIME: 5.723s
1115
+ 2026-11-01 21:14:37 | INFO | ELAPSED TRAINING TIME: 1181.24s
1116
+ 2026-11-01 21:14:37 | INFO | STEP 65 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.541 GB | free=3.753 GB
1117
+ 2026-11-01 21:14:37 | INFO | STEP 65 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.461 GB | free=3.807 GB
1118
+ 2026-11-01 21:14:55 | INFO | --------------------------------------------------------------------------------
1119
+ 2026-11-01 21:14:55 | INFO | STEP 66/70 | dataset_item=6 | repeat=6/10
1120
+ 2026-11-01 21:14:55 | INFO | LOSS: 1.28944826
1121
+ 2026-11-01 21:14:55 | INFO | LR: 1.44444444e-04
1122
+ 2026-11-01 21:14:55 | INFO | GLOBAL GRADIENT NORM: 3.60752673
1123
+ 2026-11-01 21:14:55 | INFO | MAX ABSOLUTE GRADIENT: 2.52197266e-01
1124
+ 2026-11-01 21:14:55 | INFO | GRADIENT CLIP COEFFICIENT: 0.27719815
1125
+ 2026-11-01 21:14:55 | INFO | MISSING GRADIENT TENSORS: 0
1126
+ 2026-11-01 21:14:55 | INFO | GPU0 FORWARD: 2.086s
1127
+ 2026-11-01 21:14:55 | INFO | GPU1 FORWARD: 2.751s
1128
+ 2026-11-01 21:14:55 | INFO | GPU1 BACKWARD: 7.393s
1129
+ 2026-11-01 21:14:55 | INFO | GPU0 BACKWARD: 5.482s
1130
+ 2026-11-01 21:14:55 | INFO | UPDATE WALL TIME: 5.669s
1131
+ 2026-11-01 21:14:55 | INFO | ELAPSED TRAINING TIME: 1199.45s
1132
+ 2026-11-01 21:14:55 | INFO | STEP 66 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.541 GB | free=3.753 GB
1133
+ 2026-11-01 21:14:55 | INFO | STEP 66 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.461 GB | free=3.807 GB
1134
+ 2026-11-01 21:15:13 | INFO | --------------------------------------------------------------------------------
1135
+ 2026-11-01 21:15:13 | INFO | STEP 67/70 | dataset_item=6 | repeat=7/10
1136
+ 2026-11-01 21:15:13 | INFO | LOSS: 0.91055387
1137
+ 2026-11-01 21:15:13 | INFO | LR: 1.33333333e-04
1138
+ 2026-11-01 21:15:13 | INFO | GLOBAL GRADIENT NORM: 0.66645351
1139
+ 2026-11-01 21:15:13 | INFO | MAX ABSOLUTE GRADIENT: 5.98449707e-02
1140
+ 2026-11-01 21:15:13 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
1141
+ 2026-11-01 21:15:13 | INFO | MISSING GRADIENT TENSORS: 0
1142
+ 2026-11-01 21:15:13 | INFO | GPU0 FORWARD: 2.106s
1143
+ 2026-11-01 21:15:13 | INFO | GPU1 FORWARD: 2.793s
1144
+ 2026-11-01 21:15:13 | INFO | GPU1 BACKWARD: 7.387s
1145
+ 2026-11-01 21:15:13 | INFO | GPU0 BACKWARD: 5.502s
1146
+ 2026-11-01 21:15:13 | INFO | UPDATE WALL TIME: 5.686s
1147
+ 2026-11-01 21:15:13 | INFO | ELAPSED TRAINING TIME: 1217.73s
1148
+ 2026-11-01 21:15:13 | INFO | STEP 67 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.541 GB | free=3.753 GB
1149
+ 2026-11-01 21:15:13 | INFO | STEP 67 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.461 GB | free=3.807 GB
1150
+ 2026-11-01 21:15:32 | INFO | --------------------------------------------------------------------------------
1151
+ 2026-11-01 21:15:32 | INFO | STEP 68/70 | dataset_item=6 | repeat=8/10
1152
+ 2026-11-01 21:15:32 | INFO | LOSS: 2.69720030
1153
+ 2026-11-01 21:15:32 | INFO | LR: 1.22222222e-04
1154
+ 2026-11-01 21:15:32 | INFO | GLOBAL GRADIENT NORM: 3.34976173
1155
+ 2026-11-01 21:15:32 | INFO | MAX ABSOLUTE GRADIENT: 5.60546875e-01
1156
+ 2026-11-01 21:15:32 | INFO | GRADIENT CLIP COEFFICIENT: 0.29852861
1157
+ 2026-11-01 21:15:32 | INFO | MISSING GRADIENT TENSORS: 0
1158
+ 2026-11-01 21:15:32 | INFO | GPU0 FORWARD: 2.087s
1159
+ 2026-11-01 21:15:32 | INFO | GPU1 FORWARD: 2.747s
1160
+ 2026-11-01 21:15:32 | INFO | GPU1 BACKWARD: 7.265s
1161
+ 2026-11-01 21:15:32 | INFO | GPU0 BACKWARD: 5.514s
1162
+ 2026-11-01 21:15:32 | INFO | UPDATE WALL TIME: 5.701s
1163
+ 2026-11-01 21:15:32 | INFO | ELAPSED TRAINING TIME: 1235.86s
1164
+ 2026-11-01 21:15:32 | INFO | STEP 68 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.541 GB | free=3.753 GB
1165
+ 2026-11-01 21:15:32 | INFO | STEP 68 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.461 GB | free=3.807 GB
1166
+ 2026-11-01 21:15:50 | INFO | --------------------------------------------------------------------------------
1167
+ 2026-11-01 21:15:50 | INFO | STEP 69/70 | dataset_item=6 | repeat=9/10
1168
+ 2026-11-01 21:15:50 | INFO | LOSS: 0.68159974
1169
+ 2026-11-01 21:15:50 | INFO | LR: 1.11111111e-04
1170
+ 2026-11-01 21:15:50 | INFO | GLOBAL GRADIENT NORM: 0.37365131
1171
+ 2026-11-01 21:15:50 | INFO | MAX ABSOLUTE GRADIENT: 1.61895752e-02
1172
+ 2026-11-01 21:15:50 | INFO | GRADIENT CLIP COEFFICIENT: 1.00000000
1173
+ 2026-11-01 21:15:50 | INFO | MISSING GRADIENT TENSORS: 0
1174
+ 2026-11-01 21:15:50 | INFO | GPU0 FORWARD: 2.111s
1175
+ 2026-11-01 21:15:50 | INFO | GPU1 FORWARD: 2.778s
1176
+ 2026-11-01 21:15:50 | INFO | GPU1 BACKWARD: 7.383s
1177
+ 2026-11-01 21:15:50 | INFO | GPU0 BACKWARD: 5.453s
1178
+ 2026-11-01 21:15:50 | INFO | UPDATE WALL TIME: 5.638s
1179
+ 2026-11-01 21:15:50 | INFO | ELAPSED TRAINING TIME: 1254.09s
1180
+ 2026-11-01 21:15:50 | INFO | STEP 69 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.541 GB | free=3.753 GB
1181
+ 2026-11-01 21:15:50 | INFO | STEP 69 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.461 GB | free=3.807 GB
1182
+ 2026-11-01 21:16:08 | INFO | --------------------------------------------------------------------------------
1183
+ 2026-11-01 21:16:08 | INFO | STEP 70/70 | dataset_item=6 | repeat=10/10
1184
+ 2026-11-01 21:16:08 | INFO | LOSS: 2.05296731
1185
+ 2026-11-01 21:16:08 | INFO | LR: 1.00000000e-04
1186
+ 2026-11-01 21:16:08 | INFO | GLOBAL GRADIENT NORM: 1.90156157
1187
+ 2026-11-01 21:16:08 | INFO | MAX ABSOLUTE GRADIENT: 9.37500000e-02
1188
+ 2026-11-01 21:16:08 | INFO | GRADIENT CLIP COEFFICIENT: 0.52588330
1189
+ 2026-11-01 21:16:08 | INFO | MISSING GRADIENT TENSORS: 0
1190
+ 2026-11-01 21:16:08 | INFO | GPU0 FORWARD: 2.100s
1191
+ 2026-11-01 21:16:08 | INFO | GPU1 FORWARD: 2.744s
1192
+ 2026-11-01 21:16:08 | INFO | GPU1 BACKWARD: 7.295s
1193
+ 2026-11-01 21:16:08 | INFO | GPU0 BACKWARD: 5.496s
1194
+ 2026-11-01 21:16:08 | INFO | UPDATE WALL TIME: 5.685s
1195
+ 2026-11-01 21:16:08 | INFO | ELAPSED TRAINING TIME: 1272.22s
1196
+ 2026-11-01 21:16:08 | INFO | STEP 70 | GPU0 | allocated=5.070 GB | reserved=10.672 GB | peak=10.541 GB | free=3.753 GB
1197
+ 2026-11-01 21:16:08 | INFO | STEP 70 | GPU1 | allocated=4.984 GB | reserved=10.617 GB | peak=10.461 GB | free=3.807 GB
1198
+ 2026-11-01 21:16:08 | INFO | ================================================================================
1199
+ 2026-11-01 21:16:08 | INFO | CHECKPOINT TRIGGERED AT OPTIMIZER STEP 70
1200
+ 2026-11-01 21:16:11 | INFO | CHECKPOINT SUCCESS: kromoss_lora_step_000070.safetensors
1201
+ 2026-11-01 21:16:11 | INFO | CHECKPOINT SIZE: 70.93 MB
1202
+ 2026-11-01 21:16:11 | INFO | ================================================================================
1203
+ 2026-11-01 21:16:12 | INFO | COMPLETED DATASET ITEM 7/7
1204
+ 2026-11-01 21:16:12 | INFO | ================================================================================
1205
+ 2026-11-01 21:16:12 | INFO | ALL 70 OPTIMIZER UPDATES COMPLETED SUCCESSFULLY
1206
+ 2026-11-01 21:16:12 | INFO | ================================================================================
1207
+ 2026-11-01 21:16:12 | INFO | UPDATING CONTINUATION LORA
1208
+ 2026-11-01 21:16:13 | INFO | CONTINUATION LORA UPDATED: codeShare/kromoss-lora-exports/continue.safetensors
1209
+ 2026-11-01 21:16:13 | INFO | CONTINUATION URL: https://huggingface.co/codeShare/kromoss-lora-exports/commit/dd1023b44e5faaa4a2f0d40c064b22b5dae775ce
1210
+ 2026-11-01 21:16:13 | INFO | ================================================================================
1211
+ 2026-11-01 21:16:13 | INFO | TRAINING FINISHED SUCCESSFULLY
1212
+ 2026-11-01 21:16:13 | INFO | ================================================================================
1213
+ 2026-11-01 21:16:13 | INFO | Optimizer updates: 70
1214
+ 2026-11-01 21:16:13 | INFO | Dataset items: 7
1215
+ 2026-11-01 21:16:13 | INFO | Repeats per item: 10
1216
+ 2026-11-01 21:16:13 | INFO | Checkpoint steps: [14, 35, 70]
1217
+ 2026-11-01 21:16:13 | INFO | Initial learning rate: 2.00000000e-04
1218
+ 2026-11-01 21:16:13 | INFO | Final per-item learning rate: 1.00000000e-04
1219
+ 2026-11-01 21:16:13 | INFO | AdamW state persisted across all seven dataset items.
1220
+ 2026-11-01 21:16:13 | INFO | Learning rate reset at the beginning of each dataset item.
1221
+ 2026-11-01 21:16:13 | INFO | Final export: /kaggle/working/kromoss_lora_out/comfyui/kromoss_lora_step_000070.safetensors
1222
+ 2026-11-01 21:16:13 | INFO | Continuation file: continue.safetensors
1223
+ 2026-11-01 21:16:13 | INFO | HF repository: codeShare/kromoss-lora-exports
1224
+ 2026-11-01 21:16:13 | INFO | Total elapsed time: 1277.35s
1225
+ 2026-11-01 21:16:13 | INFO | ================================================================================