andrew-healey commited on
Commit
12807b2
·
verified ·
1 Parent(s): ca5f803

Upload folder using huggingface_hub

Browse files
lr6e-4_total_batch_size61440_baseline_seed1338/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_11/lr6e-4_total_batch_size61440_baseline_seed1338", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 4, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": 8750, "warmup_steps": 500, "group": "wider_is_better_11", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1338, "memory_penalty_epsilon": 0.1, "selection_head_linear_combo": "none", "selection_head_linear_combo_scale": 1.0, "protection_kind": "none", "leaky_relu_alpha": null, "leaky_relu_bias": null, "use_compile": true, "use_mini_model": false, "upload_to_hf": true, "seq_len": 256, "batch_size": 120, "total_batch_size": 61440, "protection_head_scaling_factor": 1.0, "protection_head_bias": 0.0, "n_sliced_masks": null, "n_latent_masks": null, "mask_layernorm": false, "residual_attention_masks": false, "compute_base_shapes": false, "base_shapes_savefile": null, "mup": true, "disable_selection": false, "mup_enable_coord_check_logging": false, "max_lr": 0.0006, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "6e-4_61440", "n_embd": 256}
lr6e-4_total_batch_size61440_baseline_seed1338/dataloader_08749.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:953385078aa3787b69fc6857dfd48b0a2cd2f4d27c6f8892e01211aca53d07f5
3
+ size 964
lr6e-4_total_batch_size61440_baseline_seed1338/log2.txt ADDED
@@ -0,0 +1,1054 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ max_steps: 8750
2
+ 0 val loss 11.2202
3
+ 0 val perplexity 74626.3672
4
+ 0 train 11.227304 (lr=1.2000e-06) (hash(x)=150327452)
5
+ 10 train 10.264297 (lr=1.3200e-05) (hash(x)=165603215)
6
+ 20 train 9.822220 (lr=2.5200e-05) (hash(x)=142061326)
7
+ 30 train 9.539705 (lr=3.7200e-05) (hash(x)=147690730)
8
+ 40 train 9.265720 (lr=4.9200e-05) (hash(x)=160872542)
9
+ 50 train 8.876311 (lr=6.1200e-05) (hash(x)=166441190)
10
+ 60 train 8.467316 (lr=7.3200e-05) (hash(x)=140536766)
11
+ 70 train 8.044115 (lr=8.5200e-05) (hash(x)=146052944)
12
+ 80 train 7.776663 (lr=9.7200e-05) (hash(x)=140542391)
13
+ 90 train 7.566552 (lr=1.0920e-04) (hash(x)=147098660)
14
+ 100 val loss 7.5550
15
+ 100 val perplexity 1910.2827
16
+ 100 train 7.542656 (lr=1.2120e-04) (hash(x)=166780046)
17
+ 110 train 7.331550 (lr=1.3320e-04) (hash(x)=143376584)
18
+ 120 train 7.284158 (lr=1.4520e-04) (hash(x)=145337630)
19
+ 130 train 7.265489 (lr=1.5720e-04) (hash(x)=147524744)
20
+ 140 train 7.156878 (lr=1.6920e-04) (hash(x)=155587442)
21
+ 150 train 7.039916 (lr=1.8120e-04) (hash(x)=159835303)
22
+ 160 train 6.872100 (lr=1.9320e-04) (hash(x)=142171630)
23
+ 170 train 6.765819 (lr=2.0520e-04) (hash(x)=136930463)
24
+ 180 train 6.754886 (lr=2.1720e-04) (hash(x)=153579148)
25
+ 190 train 6.637670 (lr=2.2920e-04) (hash(x)=140352771)
26
+ 200 val loss 6.8020
27
+ 200 val perplexity 899.6688
28
+ 200 train 6.523501 (lr=2.4120e-04) (hash(x)=155040610)
29
+ 210 train 6.448275 (lr=2.5320e-04) (hash(x)=141273862)
30
+ 220 train 6.677712 (lr=2.6520e-04) (hash(x)=165326618)
31
+ 230 train 6.404750 (lr=2.7720e-04) (hash(x)=150122120)
32
+ 240 train 6.207932 (lr=2.8920e-04) (hash(x)=143024679)
33
+ 250 train 6.197723 (lr=3.0120e-04) (hash(x)=130190460)
34
+ 260 train 6.306369 (lr=3.1320e-04) (hash(x)=151890219)
35
+ 270 train 6.105533 (lr=3.2520e-04) (hash(x)=143685291)
36
+ 280 train 6.375318 (lr=3.3720e-04) (hash(x)=137108194)
37
+ 290 train 6.537086 (lr=3.4920e-04) (hash(x)=157920866)
38
+ 300 val loss 6.3194
39
+ 300 val perplexity 555.2238
40
+ 300 train 6.251268 (lr=3.6120e-04) (hash(x)=155504036)
41
+ 310 train 6.360087 (lr=3.7320e-04) (hash(x)=144551186)
42
+ 320 train 6.359428 (lr=3.8520e-04) (hash(x)=161180086)
43
+ 330 train 6.102449 (lr=3.9720e-04) (hash(x)=141651471)
44
+ 340 train 6.199506 (lr=4.0920e-04) (hash(x)=157490741)
45
+ 350 train 6.171093 (lr=4.2120e-04) (hash(x)=137347213)
46
+ 360 train 6.409575 (lr=4.3320e-04) (hash(x)=171273659)
47
+ 370 train 6.071672 (lr=4.4520e-04) (hash(x)=152118871)
48
+ 380 train 6.058847 (lr=4.5720e-04) (hash(x)=161338097)
49
+ 390 train 6.125686 (lr=4.6920e-04) (hash(x)=152404537)
50
+ 400 val loss 5.9873
51
+ 400 val perplexity 398.3297
52
+ 400 train 5.832019 (lr=4.8120e-04) (hash(x)=143823248)
53
+ 410 train 6.075671 (lr=4.9320e-04) (hash(x)=161686752)
54
+ 420 train 5.816225 (lr=5.0520e-04) (hash(x)=155391312)
55
+ 430 train 5.718140 (lr=5.1720e-04) (hash(x)=142145507)
56
+ 440 train 5.838814 (lr=5.2920e-04) (hash(x)=121931091)
57
+ 450 train 5.814079 (lr=5.4120e-04) (hash(x)=156260416)
58
+ 460 train 5.950016 (lr=5.5320e-04) (hash(x)=169781623)
59
+ 470 train 5.724694 (lr=5.6520e-04) (hash(x)=150131545)
60
+ 480 train 5.730963 (lr=5.7720e-04) (hash(x)=150443184)
61
+ 490 train 5.726679 (lr=5.8920e-04) (hash(x)=149368198)
62
+ 500 val loss 5.7502
63
+ 500 val perplexity 314.2422
64
+ 500 train 5.640978 (lr=6.0000e-04) (hash(x)=143734685)
65
+ 510 train 5.670963 (lr=6.0000e-04) (hash(x)=148943891)
66
+ 520 train 5.657520 (lr=5.9999e-04) (hash(x)=149070354)
67
+ 530 train 5.662800 (lr=5.9998e-04) (hash(x)=153601219)
68
+ 540 train 5.671170 (lr=5.9997e-04) (hash(x)=174970654)
69
+ 550 train 5.607504 (lr=5.9995e-04) (hash(x)=160013925)
70
+ 560 train 5.453632 (lr=5.9993e-04) (hash(x)=152096988)
71
+ 570 train 5.523782 (lr=5.9990e-04) (hash(x)=157203924)
72
+ 580 train 5.343158 (lr=5.9987e-04) (hash(x)=138582113)
73
+ 590 train 5.362182 (lr=5.9984e-04) (hash(x)=157451775)
74
+ 600 val loss 5.5693
75
+ 600 val perplexity 262.2405
76
+ 600 train 5.283688 (lr=5.9980e-04) (hash(x)=150678249)
77
+ 610 train 5.336588 (lr=5.9976e-04) (hash(x)=150248072)
78
+ 620 train 5.218178 (lr=5.9972e-04) (hash(x)=148286909)
79
+ 630 train 5.185419 (lr=5.9967e-04) (hash(x)=141197178)
80
+ 640 train 5.258883 (lr=5.9962e-04) (hash(x)=144988391)
81
+ 650 train 5.202114 (lr=5.9956e-04) (hash(x)=149073315)
82
+ 660 train 5.137412 (lr=5.9950e-04) (hash(x)=145239943)
83
+ 670 train 5.163773 (lr=5.9943e-04) (hash(x)=146657426)
84
+ 680 train 5.386376 (lr=5.9937e-04) (hash(x)=142323429)
85
+ 690 train 5.365347 (lr=5.9929e-04) (hash(x)=142724342)
86
+ 700 val loss 5.4224
87
+ 700 val perplexity 226.4288
88
+ 700 train 5.286738 (lr=5.9922e-04) (hash(x)=175802021)
89
+ 710 train 5.360932 (lr=5.9914e-04) (hash(x)=148902690)
90
+ 720 train 5.428729 (lr=5.9905e-04) (hash(x)=155151320)
91
+ 730 train 5.285549 (lr=5.9897e-04) (hash(x)=142078914)
92
+ 740 train 5.455531 (lr=5.9887e-04) (hash(x)=155634484)
93
+ 750 train 5.391163 (lr=5.9878e-04) (hash(x)=171034639)
94
+ 760 train 5.355544 (lr=5.9868e-04) (hash(x)=142470148)
95
+ 770 train 5.328402 (lr=5.9857e-04) (hash(x)=156024448)
96
+ 780 train 5.218045 (lr=5.9847e-04) (hash(x)=151369417)
97
+ 790 train 5.164786 (lr=5.9836e-04) (hash(x)=139260056)
98
+ 800 val loss 5.2709
99
+ 800 val perplexity 194.5971
100
+ 800 train 5.394062 (lr=5.9824e-04) (hash(x)=158681215)
101
+ 810 train 5.199815 (lr=5.9812e-04) (hash(x)=144884161)
102
+ 820 train 5.243296 (lr=5.9800e-04) (hash(x)=146977514)
103
+ 830 train 5.116415 (lr=5.9787e-04) (hash(x)=148983636)
104
+ 840 train 5.175656 (lr=5.9774e-04) (hash(x)=157824542)
105
+ 850 train 5.129094 (lr=5.9761e-04) (hash(x)=152116061)
106
+ 860 train 5.227029 (lr=5.9747e-04) (hash(x)=153743703)
107
+ 870 train 5.320691 (lr=5.9732e-04) (hash(x)=149913335)
108
+ 880 train 5.123341 (lr=5.9718e-04) (hash(x)=151096434)
109
+ 890 train 5.102977 (lr=5.9703e-04) (hash(x)=145848702)
110
+ 900 val loss 5.1598
111
+ 900 val perplexity 174.1381
112
+ 900 train 5.082558 (lr=5.9687e-04) (hash(x)=146108145)
113
+ 910 train 5.010022 (lr=5.9672e-04) (hash(x)=146903806)
114
+ 920 train 4.976546 (lr=5.9655e-04) (hash(x)=149225019)
115
+ 930 train 4.928758 (lr=5.9639e-04) (hash(x)=146110422)
116
+ 940 train 5.005662 (lr=5.9622e-04) (hash(x)=161333614)
117
+ 950 train 4.932198 (lr=5.9605e-04) (hash(x)=147598108)
118
+ 960 train 5.049557 (lr=5.9587e-04) (hash(x)=158133101)
119
+ 970 train 4.905695 (lr=5.9569e-04) (hash(x)=152199591)
120
+ 980 train 4.868150 (lr=5.9550e-04) (hash(x)=152359036)
121
+ 990 train 4.884979 (lr=5.9531e-04) (hash(x)=145166335)
122
+ 1000 val loss 5.0950
123
+ 1000 val perplexity 163.1966
124
+ 1000 train 4.792043 (lr=5.9512e-04) (hash(x)=154996086)
125
+ 1010 train 4.726423 (lr=5.9492e-04) (hash(x)=149406975)
126
+ 1020 train 4.848952 (lr=5.9472e-04) (hash(x)=144618270)
127
+ 1030 train 4.701184 (lr=5.9452e-04) (hash(x)=148306536)
128
+ 1040 train 4.733491 (lr=5.9431e-04) (hash(x)=151643553)
129
+ 1050 train 4.661354 (lr=5.9410e-04) (hash(x)=153396183)
130
+ 1060 train 4.693603 (lr=5.9388e-04) (hash(x)=157501787)
131
+ 1070 train 4.888477 (lr=5.9366e-04) (hash(x)=153277507)
132
+ 1080 train 4.959070 (lr=5.9344e-04) (hash(x)=156232389)
133
+ 1090 train 4.904087 (lr=5.9321e-04) (hash(x)=150673143)
134
+ 1100 val loss 4.9859
135
+ 1100 val perplexity 146.3392
136
+ 1100 train 4.979692 (lr=5.9298e-04) (hash(x)=153885445)
137
+ 1110 train 4.968654 (lr=5.9275e-04) (hash(x)=143951073)
138
+ 1120 train 5.354780 (lr=5.9251e-04) (hash(x)=151626904)
139
+ 1130 train 4.951041 (lr=5.9227e-04) (hash(x)=144013074)
140
+ 1140 train 4.867618 (lr=5.9202e-04) (hash(x)=149948988)
141
+ 1150 train 4.940332 (lr=5.9177e-04) (hash(x)=159666385)
142
+ 1160 train 4.890215 (lr=5.9152e-04) (hash(x)=146889587)
143
+ 1170 train 4.956844 (lr=5.9126e-04) (hash(x)=151746917)
144
+ 1180 train 4.961044 (lr=5.9100e-04) (hash(x)=154396948)
145
+ 1190 train 4.902488 (lr=5.9073e-04) (hash(x)=148201222)
146
+ 1200 val loss 4.8887
147
+ 1200 val perplexity 132.7770
148
+ 1200 train 4.907124 (lr=5.9046e-04) (hash(x)=142353087)
149
+ 1210 train 4.898867 (lr=5.9019e-04) (hash(x)=146779775)
150
+ 1220 train 4.974347 (lr=5.8992e-04) (hash(x)=155864740)
151
+ 1230 train 4.988605 (lr=5.8963e-04) (hash(x)=129826485)
152
+ 1240 train 4.819598 (lr=5.8935e-04) (hash(x)=108599491)
153
+ 1250 train 4.757872 (lr=5.8906e-04) (hash(x)=146491718)
154
+ 1260 train 4.826903 (lr=5.8877e-04) (hash(x)=141897382)
155
+ 1270 train 4.785016 (lr=5.8848e-04) (hash(x)=144830380)
156
+ 1280 train 4.840586 (lr=5.8818e-04) (hash(x)=157677763)
157
+ 1290 train 4.674917 (lr=5.8787e-04) (hash(x)=143611746)
158
+ 1300 val loss 4.8262
159
+ 1300 val perplexity 124.7353
160
+ 1300 train 4.699008 (lr=5.8757e-04) (hash(x)=150750353)
161
+ 1310 train 4.959228 (lr=5.8726e-04) (hash(x)=152850907)
162
+ 1320 train 4.832575 (lr=5.8694e-04) (hash(x)=148575395)
163
+ 1330 train 4.587888 (lr=5.8663e-04) (hash(x)=157113001)
164
+ 1340 train 4.640936 (lr=5.8630e-04) (hash(x)=137400723)
165
+ 1350 train 4.701644 (lr=5.8598e-04) (hash(x)=129849193)
166
+ 1360 train 4.675120 (lr=5.8565e-04) (hash(x)=144176797)
167
+ 1370 train 4.517898 (lr=5.8532e-04) (hash(x)=144550831)
168
+ 1380 train 4.484356 (lr=5.8498e-04) (hash(x)=144171821)
169
+ 1390 train 4.423070 (lr=5.8464e-04) (hash(x)=146110449)
170
+ 1400 val loss 4.7762
171
+ 1400 val perplexity 118.6571
172
+ 1400 train 4.505144 (lr=5.8430e-04) (hash(x)=152767913)
173
+ 1410 train 4.509851 (lr=5.8395e-04) (hash(x)=148110947)
174
+ 1420 train 4.391037 (lr=5.8360e-04) (hash(x)=153477905)
175
+ 1430 train 4.281709 (lr=5.8324e-04) (hash(x)=147227598)
176
+ 1440 train 4.490622 (lr=5.8289e-04) (hash(x)=146280368)
177
+ 1450 train 4.468853 (lr=5.8252e-04) (hash(x)=146531140)
178
+ 1460 train 4.310854 (lr=5.8216e-04) (hash(x)=151573873)
179
+ 1470 train 4.777008 (lr=5.8179e-04) (hash(x)=155559275)
180
+ 1480 train 4.657070 (lr=5.8142e-04) (hash(x)=140584826)
181
+ 1490 train 4.632737 (lr=5.8104e-04) (hash(x)=165070637)
182
+ 1500 val loss 4.7375
183
+ 1500 val perplexity 114.1522
184
+ 1500 train 4.757635 (lr=5.8066e-04) (hash(x)=151562048)
185
+ 1510 train 4.525479 (lr=5.8028e-04) (hash(x)=147770607)
186
+ 1520 train 4.616034 (lr=5.7989e-04) (hash(x)=156486153)
187
+ 1530 train 4.570763 (lr=5.7950e-04) (hash(x)=140244564)
188
+ 1540 train 4.629193 (lr=5.7910e-04) (hash(x)=151627549)
189
+ 1550 train 4.675590 (lr=5.7870e-04) (hash(x)=146001424)
190
+ 1560 train 4.744078 (lr=5.7830e-04) (hash(x)=184266264)
191
+ 1570 train 4.646249 (lr=5.7790e-04) (hash(x)=145002091)
192
+ 1580 train 4.531161 (lr=5.7749e-04) (hash(x)=132216826)
193
+ 1590 train 4.710011 (lr=5.7707e-04) (hash(x)=160975703)
194
+ 1600 val loss 4.6728
195
+ 1600 val perplexity 106.9987
196
+ 1600 train 4.820989 (lr=5.7666e-04) (hash(x)=166486165)
197
+ 1610 train 4.708908 (lr=5.7624e-04) (hash(x)=149601833)
198
+ 1620 train 4.596306 (lr=5.7581e-04) (hash(x)=157815188)
199
+ 1630 train 4.612430 (lr=5.7539e-04) (hash(x)=144234679)
200
+ 1640 train 4.756866 (lr=5.7496e-04) (hash(x)=160054560)
201
+ 1650 train 4.570097 (lr=5.7452e-04) (hash(x)=141061333)
202
+ 1660 train 4.687957 (lr=5.7408e-04) (hash(x)=148064219)
203
+ 1670 train 4.534023 (lr=5.7364e-04) (hash(x)=162799669)
204
+ 1680 train 4.561423 (lr=5.7320e-04) (hash(x)=147474277)
205
+ 1690 train 4.623002 (lr=5.7275e-04) (hash(x)=142369299)
206
+ 1700 val loss 4.6469
207
+ 1700 val perplexity 104.2572
208
+ 1700 train 4.391309 (lr=5.7230e-04) (hash(x)=130835396)
209
+ 1710 train 4.749188 (lr=5.7184e-04) (hash(x)=153173892)
210
+ 1720 train 4.395944 (lr=5.7138e-04) (hash(x)=148451191)
211
+ 1730 train 4.514690 (lr=5.7092e-04) (hash(x)=145047239)
212
+ 1740 train 4.422426 (lr=5.7045e-04) (hash(x)=147187109)
213
+ 1750 train 4.528538 (lr=5.6999e-04) (hash(x)=149493155)
214
+ 1760 train 4.479813 (lr=5.6951e-04) (hash(x)=147647531)
215
+ 1770 train 4.703120 (lr=5.6904e-04) (hash(x)=153677279)
216
+ 1780 train 4.535886 (lr=5.6856e-04) (hash(x)=144314015)
217
+ 1790 train 4.562047 (lr=5.6807e-04) (hash(x)=176962883)
218
+ 1800 val loss 4.6100
219
+ 1800 val perplexity 100.4863
220
+ 1800 train 4.658447 (lr=5.6759e-04) (hash(x)=158851816)
221
+ 1810 train 4.247943 (lr=5.6710e-04) (hash(x)=175415447)
222
+ 1820 train 4.544044 (lr=5.6660e-04) (hash(x)=169829733)
223
+ 1830 train 4.521188 (lr=5.6611e-04) (hash(x)=154116106)
224
+ 1840 train 4.360022 (lr=5.6561e-04) (hash(x)=150168406)
225
+ 1850 train 4.355367 (lr=5.6510e-04) (hash(x)=145782493)
226
+ 1860 train 4.335286 (lr=5.6459e-04) (hash(x)=158111664)
227
+ 1870 train 4.274739 (lr=5.6408e-04) (hash(x)=154837061)
228
+ 1880 train 4.237429 (lr=5.6357e-04) (hash(x)=151363745)
229
+ 1890 train 4.351537 (lr=5.6305e-04) (hash(x)=161283557)
230
+ 1900 val loss 4.5920
231
+ 1900 val perplexity 98.6896
232
+ 1900 train 4.598354 (lr=5.6253e-04) (hash(x)=153313879)
233
+ 1910 train 4.414730 (lr=5.6201e-04) (hash(x)=152272272)
234
+ 1920 train 4.486166 (lr=5.6148e-04) (hash(x)=153648837)
235
+ 1930 train 4.582623 (lr=5.6095e-04) (hash(x)=150161694)
236
+ 1940 train 4.467021 (lr=5.6041e-04) (hash(x)=151235093)
237
+ 1950 train 4.515565 (lr=5.5988e-04) (hash(x)=140582058)
238
+ 1960 train 4.447826 (lr=5.5934e-04) (hash(x)=125616059)
239
+ 1970 train 4.590308 (lr=5.5879e-04) (hash(x)=154353968)
240
+ 1980 train 4.609747 (lr=5.5824e-04) (hash(x)=156063143)
241
+ 1990 train 4.627287 (lr=5.5769e-04) (hash(x)=160927289)
242
+ 2000 val loss 4.5473
243
+ 2000 val perplexity 94.3768
244
+ 2000 train 4.528445 (lr=5.5714e-04) (hash(x)=158245023)
245
+ 2010 train 4.580523 (lr=5.5658e-04) (hash(x)=157096661)
246
+ 2020 train 4.790643 (lr=5.5602e-04) (hash(x)=144994302)
247
+ 2030 train 4.486280 (lr=5.5546e-04) (hash(x)=145742929)
248
+ 2040 train 4.496912 (lr=5.5489e-04) (hash(x)=156958528)
249
+ 2050 train 5.007847 (lr=5.5432e-04) (hash(x)=135313020)
250
+ 2060 train 4.655932 (lr=5.5374e-04) (hash(x)=147309332)
251
+ 2070 train 4.433441 (lr=5.5317e-04) (hash(x)=161720962)
252
+ 2080 train 4.622493 (lr=5.5259e-04) (hash(x)=167204793)
253
+ 2090 train 4.516514 (lr=5.5200e-04) (hash(x)=141325510)
254
+ 2100 val loss 4.5185
255
+ 2100 val perplexity 91.7006
256
+ 2100 train 4.341032 (lr=5.5142e-04) (hash(x)=157204896)
257
+ 2110 train 4.303373 (lr=5.5083e-04) (hash(x)=143077993)
258
+ 2120 train 4.509861 (lr=5.5023e-04) (hash(x)=159032689)
259
+ 2130 train 4.385473 (lr=5.4964e-04) (hash(x)=149513855)
260
+ 2140 train 4.313018 (lr=5.4904e-04) (hash(x)=151678267)
261
+ 2150 train 4.401194 (lr=5.4843e-04) (hash(x)=141309246)
262
+ 2160 train 4.500446 (lr=5.4783e-04) (hash(x)=153935732)
263
+ 2170 train 4.397508 (lr=5.4722e-04) (hash(x)=150668875)
264
+ 2180 train 4.396203 (lr=5.4661e-04) (hash(x)=142033123)
265
+ 2190 train 4.441040 (lr=5.4599e-04) (hash(x)=157015026)
266
+ 2200 val loss 4.5058
267
+ 2200 val perplexity 90.5408
268
+ 2200 train 4.330727 (lr=5.4537e-04) (hash(x)=137541932)
269
+ 2210 train 4.235078 (lr=5.4475e-04) (hash(x)=144424504)
270
+ 2220 train 4.178911 (lr=5.4413e-04) (hash(x)=151015810)
271
+ 2230 train 4.240218 (lr=5.4350e-04) (hash(x)=160707806)
272
+ 2240 train 4.250599 (lr=5.4287e-04) (hash(x)=151874110)
273
+ 2250 train 4.219031 (lr=5.4223e-04) (hash(x)=157341469)
274
+ 2260 train 4.039702 (lr=5.4160e-04) (hash(x)=144563224)
275
+ 2270 train 4.202419 (lr=5.4096e-04) (hash(x)=144878459)
276
+ 2280 train 4.274534 (lr=5.4031e-04) (hash(x)=158331565)
277
+ 2290 train 4.292605 (lr=5.3967e-04) (hash(x)=146875132)
278
+ 2300 val loss 4.4907
279
+ 2300 val perplexity 89.1855
280
+ 2300 train 4.434899 (lr=5.3902e-04) (hash(x)=150149692)
281
+ 2310 train 4.433178 (lr=5.3837e-04) (hash(x)=141206538)
282
+ 2320 train 4.472841 (lr=5.3771e-04) (hash(x)=136219115)
283
+ 2330 train 4.503714 (lr=5.3705e-04) (hash(x)=148496500)
284
+ 2340 train 4.489939 (lr=5.3639e-04) (hash(x)=161388511)
285
+ 2350 train 4.333093 (lr=5.3573e-04) (hash(x)=149433185)
286
+ 2360 train 4.552496 (lr=5.3506e-04) (hash(x)=154659203)
287
+ 2370 train 4.539382 (lr=5.3439e-04) (hash(x)=145696996)
288
+ 2380 train 4.349158 (lr=5.3372e-04) (hash(x)=139915095)
289
+ 2390 train 4.471357 (lr=5.3304e-04) (hash(x)=150567584)
290
+ 2400 val loss 4.4591
291
+ 2400 val perplexity 86.4124
292
+ 2400 train 4.378132 (lr=5.3236e-04) (hash(x)=151730720)
293
+ 2410 train 4.413832 (lr=5.3168e-04) (hash(x)=153136544)
294
+ 2420 train 4.384438 (lr=5.3099e-04) (hash(x)=151004677)
295
+ 2430 train 4.405857 (lr=5.3030e-04) (hash(x)=167118859)
296
+ 2440 train 4.415806 (lr=5.2961e-04) (hash(x)=145433453)
297
+ 2450 train 4.442924 (lr=5.2892e-04) (hash(x)=150913293)
298
+ 2460 train 4.414309 (lr=5.2822e-04) (hash(x)=145674639)
299
+ 2470 train 4.520781 (lr=5.2752e-04) (hash(x)=151022749)
300
+ 2480 train 4.369912 (lr=5.2682e-04) (hash(x)=136621037)
301
+ 2490 train 4.420648 (lr=5.2612e-04) (hash(x)=158100672)
302
+ 2500 val loss 4.4299
303
+ 2500 val perplexity 83.9196
304
+ 2500 train 4.439744 (lr=5.2541e-04) (hash(x)=143406752)
305
+ 2510 train 4.358023 (lr=5.2470e-04) (hash(x)=142664084)
306
+ 2520 train 4.418015 (lr=5.2398e-04) (hash(x)=132174378)
307
+ 2530 train 4.431797 (lr=5.2327e-04) (hash(x)=140760293)
308
+ 2540 train 4.365294 (lr=5.2255e-04) (hash(x)=154863398)
309
+ 2550 train 4.308542 (lr=5.2183e-04) (hash(x)=153042049)
310
+ 2560 train 4.241079 (lr=5.2110e-04) (hash(x)=145339170)
311
+ 2570 train 4.398878 (lr=5.2037e-04) (hash(x)=157656357)
312
+ 2580 train 4.383601 (lr=5.1964e-04) (hash(x)=147899462)
313
+ 2590 train 4.353883 (lr=5.1891e-04) (hash(x)=142812892)
314
+ 2600 val loss 4.4412
315
+ 2600 val perplexity 84.8788
316
+ 2600 train 4.311806 (lr=5.1817e-04) (hash(x)=157272496)
317
+ 2610 train 4.103051 (lr=5.1743e-04) (hash(x)=148923594)
318
+ 2620 train 4.032918 (lr=5.1669e-04) (hash(x)=164654807)
319
+ 2630 train 4.217052 (lr=5.1595e-04) (hash(x)=136419327)
320
+ 2640 train 4.185861 (lr=5.1520e-04) (hash(x)=153420840)
321
+ 2650 train 4.033188 (lr=5.1445e-04) (hash(x)=153693827)
322
+ 2660 train 4.184895 (lr=5.1370e-04) (hash(x)=145574165)
323
+ 2670 train 4.223529 (lr=5.1295e-04) (hash(x)=153894843)
324
+ 2680 train 4.073471 (lr=5.1219e-04) (hash(x)=132234446)
325
+ 2690 train 4.153461 (lr=5.1143e-04) (hash(x)=153430212)
326
+ 2700 val loss 4.4244
327
+ 2700 val perplexity 83.4648
328
+ 2700 train 4.330724 (lr=5.1067e-04) (hash(x)=155342327)
329
+ 2710 train 4.334776 (lr=5.0990e-04) (hash(x)=145797023)
330
+ 2720 train 4.485026 (lr=5.0913e-04) (hash(x)=149220870)
331
+ 2730 train 4.388978 (lr=5.0836e-04) (hash(x)=134069562)
332
+ 2740 train 4.390977 (lr=5.0759e-04) (hash(x)=141560499)
333
+ 2750 train 4.413178 (lr=5.0681e-04) (hash(x)=146000524)
334
+ 2760 train 4.308161 (lr=5.0603e-04) (hash(x)=164710665)
335
+ 2770 train 4.193572 (lr=5.0525e-04) (hash(x)=151199094)
336
+ 2780 train 4.395183 (lr=5.0447e-04) (hash(x)=138625404)
337
+ 2790 train 4.664254 (lr=5.0368e-04) (hash(x)=162766112)
338
+ 2800 val loss 4.3939
339
+ 2800 val perplexity 80.9520
340
+ 2800 train 4.363864 (lr=5.0290e-04) (hash(x)=140626679)
341
+ 2810 train 4.433792 (lr=5.0210e-04) (hash(x)=144248544)
342
+ 2820 train 4.354659 (lr=5.0131e-04) (hash(x)=139378481)
343
+ 2830 train 4.291965 (lr=5.0052e-04) (hash(x)=150173403)
344
+ 2840 train 4.580915 (lr=4.9972e-04) (hash(x)=160168705)
345
+ 2850 train 4.370169 (lr=4.9892e-04) (hash(x)=149480059)
346
+ 2860 train 4.421855 (lr=4.9811e-04) (hash(x)=155559096)
347
+ 2870 train 4.353008 (lr=4.9731e-04) (hash(x)=158825683)
348
+ 2880 train 4.523395 (lr=4.9650e-04) (hash(x)=152129048)
349
+ 2890 train 4.197937 (lr=4.9569e-04) (hash(x)=157866544)
350
+ 2900 val loss 4.3718
351
+ 2900 val perplexity 79.1841
352
+ 2900 train 4.318600 (lr=4.9487e-04) (hash(x)=144953350)
353
+ 2910 train 4.527758 (lr=4.9406e-04) (hash(x)=154103497)
354
+ 2920 train 4.241050 (lr=4.9324e-04) (hash(x)=144467854)
355
+ 2930 train 4.169162 (lr=4.9242e-04) (hash(x)=153784048)
356
+ 2940 train 4.462783 (lr=4.9160e-04) (hash(x)=154774534)
357
+ 2950 train 4.324208 (lr=4.9077e-04) (hash(x)=143988507)
358
+ 2960 train 4.198943 (lr=4.8995e-04) (hash(x)=159552857)
359
+ 2970 train 4.504108 (lr=4.8912e-04) (hash(x)=175539144)
360
+ 2980 train 4.380345 (lr=4.8829e-04) (hash(x)=158172686)
361
+ 2990 train 4.254247 (lr=4.8745e-04) (hash(x)=147553039)
362
+ 3000 val loss 4.3743
363
+ 3000 val perplexity 79.3830
364
+ 3000 train 4.185371 (lr=4.8662e-04) (hash(x)=172449837)
365
+ 3010 train 4.076173 (lr=4.8578e-04) (hash(x)=153285123)
366
+ 3020 train 4.247039 (lr=4.8494e-04) (hash(x)=145307252)
367
+ 3030 train 4.299971 (lr=4.8409e-04) (hash(x)=147739766)
368
+ 3040 train 4.017210 (lr=4.8325e-04) (hash(x)=154414400)
369
+ 3050 train 4.104315 (lr=4.8240e-04) (hash(x)=150696487)
370
+ 3060 train 4.118686 (lr=4.8155e-04) (hash(x)=154543441)
371
+ 3070 train 3.889594 (lr=4.8070e-04) (hash(x)=170733242)
372
+ 3080 train 4.248117 (lr=4.7984e-04) (hash(x)=161914669)
373
+ 3090 train 4.128709 (lr=4.7899e-04) (hash(x)=156125491)
374
+ 3100 val loss 4.3875
375
+ 3100 val perplexity 80.4404
376
+ 3100 train 4.025801 (lr=4.7813e-04) (hash(x)=141710086)
377
+ 3110 train 4.325425 (lr=4.7727e-04) (hash(x)=154547635)
378
+ 3120 train 4.175650 (lr=4.7641e-04) (hash(x)=151747171)
379
+ 3130 train 4.455506 (lr=4.7554e-04) (hash(x)=160311353)
380
+ 3140 train 4.574066 (lr=4.7467e-04) (hash(x)=157233821)
381
+ 3150 train 4.384850 (lr=4.7380e-04) (hash(x)=153016810)
382
+ 3160 train 4.353474 (lr=4.7293e-04) (hash(x)=157289875)
383
+ 3170 train 4.360764 (lr=4.7206e-04) (hash(x)=157459484)
384
+ 3180 train 4.286309 (lr=4.7118e-04) (hash(x)=162037435)
385
+ 3190 train 4.380629 (lr=4.7031e-04) (hash(x)=159360169)
386
+ 3200 val loss 4.3358
387
+ 3200 val perplexity 76.3863
388
+ 3200 train 4.254319 (lr=4.6943e-04) (hash(x)=151299772)
389
+ 3210 train 4.475201 (lr=4.6855e-04) (hash(x)=183865364)
390
+ 3220 train 4.220765 (lr=4.6766e-04) (hash(x)=139622592)
391
+ 3230 train 4.175694 (lr=4.6678e-04) (hash(x)=147102655)
392
+ 3240 train 4.239796 (lr=4.6589e-04) (hash(x)=161870696)
393
+ 3250 train 4.393182 (lr=4.6500e-04) (hash(x)=152692231)
394
+ 3260 train 4.136336 (lr=4.6411e-04) (hash(x)=149990134)
395
+ 3270 train 4.110493 (lr=4.6322e-04) (hash(x)=134848432)
396
+ 3280 train 4.165795 (lr=4.6232e-04) (hash(x)=138519418)
397
+ 3290 train 4.299252 (lr=4.6142e-04) (hash(x)=139301249)
398
+ 3300 val loss 4.3356
399
+ 3300 val perplexity 76.3679
400
+ 3300 train 4.142056 (lr=4.6052e-04) (hash(x)=146473110)
401
+ 3310 train 4.208535 (lr=4.5962e-04) (hash(x)=144833685)
402
+ 3320 train 4.112485 (lr=4.5872e-04) (hash(x)=148953545)
403
+ 3330 train 4.350647 (lr=4.5782e-04) (hash(x)=161302371)
404
+ 3340 train 4.278232 (lr=4.5691e-04) (hash(x)=144215411)
405
+ 3350 train 4.439169 (lr=4.5600e-04) (hash(x)=157713664)
406
+ 3360 train 4.318747 (lr=4.5509e-04) (hash(x)=150813928)
407
+ 3370 train 4.306203 (lr=4.5418e-04) (hash(x)=150889463)
408
+ 3380 train 4.354084 (lr=4.5326e-04) (hash(x)=156631928)
409
+ 3390 train 4.375599 (lr=4.5235e-04) (hash(x)=136001579)
410
+ 3400 val loss 4.3142
411
+ 3400 val perplexity 74.7527
412
+ 3400 train 4.465424 (lr=4.5143e-04) (hash(x)=153954157)
413
+ 3410 train 4.235224 (lr=4.5051e-04) (hash(x)=150012477)
414
+ 3420 train 4.600833 (lr=4.4959e-04) (hash(x)=160878899)
415
+ 3430 train 4.633406 (lr=4.4867e-04) (hash(x)=128131106)
416
+ 3440 train 4.564785 (lr=4.4774e-04) (hash(x)=149611866)
417
+ 3450 train 4.190346 (lr=4.4682e-04) (hash(x)=166942460)
418
+ 3460 train 4.344059 (lr=4.4589e-04) (hash(x)=150498414)
419
+ 3470 train 4.498346 (lr=4.4496e-04) (hash(x)=144342279)
420
+ 3480 train 4.316572 (lr=4.4403e-04) (hash(x)=148891764)
421
+ 3490 train 4.303015 (lr=4.4310e-04) (hash(x)=157308188)
422
+ 3500 val loss 4.3024
423
+ 3500 val perplexity 73.8781
424
+ 3500 train 4.200562 (lr=4.4216e-04) (hash(x)=153717336)
425
+ 3510 train 4.217344 (lr=4.4123e-04) (hash(x)=156083924)
426
+ 3520 train 4.211225 (lr=4.4029e-04) (hash(x)=144234966)
427
+ 3530 train 4.293066 (lr=4.3935e-04) (hash(x)=150184381)
428
+ 3540 train 4.180350 (lr=4.3841e-04) (hash(x)=143011198)
429
+ 3550 train 4.320801 (lr=4.3747e-04) (hash(x)=148764613)
430
+ 3560 train 4.084127 (lr=4.3652e-04) (hash(x)=150625329)
431
+ 3570 train 4.206750 (lr=4.3558e-04) (hash(x)=143981302)
432
+ 3580 train 4.256117 (lr=4.3463e-04) (hash(x)=154989703)
433
+ 3590 train 4.376549 (lr=4.3368e-04) (hash(x)=146602526)
434
+ 3600 val loss 4.2960
435
+ 3600 val perplexity 73.4040
436
+ 3600 train 4.076576 (lr=4.3273e-04) (hash(x)=144965161)
437
+ 3610 train 4.071461 (lr=4.3178e-04) (hash(x)=159452243)
438
+ 3620 train 4.061543 (lr=4.3083e-04) (hash(x)=144904300)
439
+ 3630 train 3.973132 (lr=4.2987e-04) (hash(x)=127790212)
440
+ 3640 train 4.210320 (lr=4.2892e-04) (hash(x)=150863158)
441
+ 3650 train 4.169933 (lr=4.2796e-04) (hash(x)=145691635)
442
+ 3660 train 4.157637 (lr=4.2700e-04) (hash(x)=150631264)
443
+ 3670 train 4.132852 (lr=4.2604e-04) (hash(x)=148646259)
444
+ 3680 train 4.098392 (lr=4.2508e-04) (hash(x)=139317386)
445
+ 3690 train 4.207949 (lr=4.2411e-04) (hash(x)=142206916)
446
+ 3700 val loss 4.3054
447
+ 3700 val perplexity 74.0952
448
+ 3700 train 4.209076 (lr=4.2315e-04) (hash(x)=125969741)
449
+ 3710 train 4.058227 (lr=4.2218e-04) (hash(x)=157183915)
450
+ 3720 train 4.053658 (lr=4.2122e-04) (hash(x)=150434369)
451
+ 3730 train 4.242596 (lr=4.2025e-04) (hash(x)=160911722)
452
+ 3740 train 4.134961 (lr=4.1928e-04) (hash(x)=163306921)
453
+ 3750 train 4.250235 (lr=4.1831e-04) (hash(x)=149156876)
454
+ 3760 train 4.390195 (lr=4.1734e-04) (hash(x)=156015970)
455
+ 3770 train 4.326524 (lr=4.1636e-04) (hash(x)=148222322)
456
+ 3780 train 4.436397 (lr=4.1539e-04) (hash(x)=172040174)
457
+ 3790 train 4.343981 (lr=4.1441e-04) (hash(x)=151593070)
458
+ 3800 val loss 4.2785
459
+ 3800 val perplexity 72.1311
460
+ 3800 train 4.096140 (lr=4.1343e-04) (hash(x)=155070487)
461
+ 3810 train 4.293722 (lr=4.1246e-04) (hash(x)=137073164)
462
+ 3820 train 4.190902 (lr=4.1148e-04) (hash(x)=147761510)
463
+ 3830 train 4.111029 (lr=4.1050e-04) (hash(x)=154763021)
464
+ 3840 train 4.154986 (lr=4.0951e-04) (hash(x)=149622435)
465
+ 3850 train 4.267877 (lr=4.0853e-04) (hash(x)=150779423)
466
+ 3860 train 4.234346 (lr=4.0755e-04) (hash(x)=139445120)
467
+ 3870 train 4.217569 (lr=4.0656e-04) (hash(x)=151884947)
468
+ 3880 train 4.226506 (lr=4.0557e-04) (hash(x)=139795201)
469
+ 3890 train 4.266909 (lr=4.0459e-04) (hash(x)=155589956)
470
+ 3900 val loss 4.2637
471
+ 3900 val perplexity 71.0733
472
+ 3900 train 4.185539 (lr=4.0360e-04) (hash(x)=149444644)
473
+ 3910 train 4.180459 (lr=4.0261e-04) (hash(x)=148538323)
474
+ 3920 train 4.302442 (lr=4.0162e-04) (hash(x)=144026925)
475
+ 3930 train 4.191366 (lr=4.0063e-04) (hash(x)=155311200)
476
+ 3940 train 4.170839 (lr=3.9963e-04) (hash(x)=142410034)
477
+ 3950 train 4.150771 (lr=3.9864e-04) (hash(x)=145986075)
478
+ 3960 train 4.050122 (lr=3.9764e-04) (hash(x)=156197753)
479
+ 3970 train 4.190085 (lr=3.9665e-04) (hash(x)=146981564)
480
+ 3980 train 4.510362 (lr=3.9565e-04) (hash(x)=145361976)
481
+ 3990 train 4.243616 (lr=3.9465e-04) (hash(x)=149344475)
482
+ 4000 val loss 4.2639
483
+ 4000 val perplexity 71.0890
484
+ 4000 train 4.125813 (lr=3.9365e-04) (hash(x)=151663033)
485
+ 4010 train 4.007957 (lr=3.9266e-04) (hash(x)=140915499)
486
+ 4020 train 4.227386 (lr=3.9165e-04) (hash(x)=159979134)
487
+ 4030 train 4.141093 (lr=3.9065e-04) (hash(x)=163081433)
488
+ 4040 train 4.031314 (lr=3.8965e-04) (hash(x)=144890231)
489
+ 4050 train 4.191522 (lr=3.8865e-04) (hash(x)=138242368)
490
+ 4060 train 4.093877 (lr=3.8764e-04) (hash(x)=144438406)
491
+ 4070 train 3.973559 (lr=3.8664e-04) (hash(x)=159294906)
492
+ 4080 train 4.029648 (lr=3.8563e-04) (hash(x)=143676301)
493
+ 4090 train 4.260283 (lr=3.8463e-04) (hash(x)=147875296)
494
+ 4100 val loss 4.2747
495
+ 4100 val perplexity 71.8606
496
+ 4100 train 4.326531 (lr=3.8362e-04) (hash(x)=143688282)
497
+ 4110 train 4.099450 (lr=3.8261e-04) (hash(x)=150086777)
498
+ 4120 train 4.047183 (lr=3.8160e-04) (hash(x)=154624391)
499
+ 4130 train 4.231260 (lr=3.8059e-04) (hash(x)=141163032)
500
+ 4140 train 4.249689 (lr=3.7958e-04) (hash(x)=151038180)
501
+ 4150 train 4.333470 (lr=3.7857e-04) (hash(x)=144718421)
502
+ 4160 train 4.253254 (lr=3.7756e-04) (hash(x)=154277459)
503
+ 4170 train 4.406524 (lr=3.7655e-04) (hash(x)=143765173)
504
+ 4180 train 4.273074 (lr=3.7553e-04) (hash(x)=161013425)
505
+ 4190 train 4.225933 (lr=3.7452e-04) (hash(x)=161666854)
506
+ 4200 val loss 4.2376
507
+ 4200 val perplexity 69.2417
508
+ 4200 train 4.151236 (lr=3.7351e-04) (hash(x)=163361651)
509
+ 4210 train 4.494137 (lr=3.7249e-04) (hash(x)=153491528)
510
+ 4220 train 4.283960 (lr=3.7148e-04) (hash(x)=164410385)
511
+ 4230 train 4.222746 (lr=3.7046e-04) (hash(x)=175191816)
512
+ 4240 train 4.094865 (lr=3.6944e-04) (hash(x)=133924818)
513
+ 4250 train 4.318819 (lr=3.6843e-04) (hash(x)=152932845)
514
+ 4260 train 4.200481 (lr=3.6741e-04) (hash(x)=141987318)
515
+ 4270 train 4.291032 (lr=3.6639e-04) (hash(x)=142521279)
516
+ 4280 train 4.152955 (lr=3.6537e-04) (hash(x)=147707669)
517
+ 4290 train 4.213024 (lr=3.6435e-04) (hash(x)=152245539)
518
+ 4300 val loss 4.2247
519
+ 4300 val perplexity 68.3535
520
+ 4300 train 4.105681 (lr=3.6333e-04) (hash(x)=153619361)
521
+ 4310 train 4.311245 (lr=3.6231e-04) (hash(x)=142614473)
522
+ 4320 train 4.140712 (lr=3.6129e-04) (hash(x)=142689858)
523
+ 4330 train 4.159927 (lr=3.6027e-04) (hash(x)=140837891)
524
+ 4340 train 4.141294 (lr=3.5925e-04) (hash(x)=131790760)
525
+ 4350 train 4.307720 (lr=3.5822e-04) (hash(x)=147186292)
526
+ 4360 train 4.118408 (lr=3.5720e-04) (hash(x)=148140562)
527
+ 4370 train 4.237746 (lr=3.5618e-04) (hash(x)=156409913)
528
+ 4380 train 4.145455 (lr=3.5515e-04) (hash(x)=145126034)
529
+ 4390 train 4.089142 (lr=3.5413e-04) (hash(x)=147368423)
530
+ 4400 val loss 4.2282
531
+ 4400 val perplexity 68.5928
532
+ 4400 train 4.578392 (lr=3.5311e-04) (hash(x)=168527064)
533
+ 4410 train 4.087142 (lr=3.5208e-04) (hash(x)=150152773)
534
+ 4420 train 4.178150 (lr=3.5106e-04) (hash(x)=155472304)
535
+ 4430 train 4.097244 (lr=3.5003e-04) (hash(x)=147299265)
536
+ 4440 train 4.058930 (lr=3.4901e-04) (hash(x)=140659522)
537
+ 4450 train 4.123826 (lr=3.4798e-04) (hash(x)=147169111)
538
+ 4460 train 4.188644 (lr=3.4695e-04) (hash(x)=144098557)
539
+ 4470 train 3.953783 (lr=3.4593e-04) (hash(x)=161957507)
540
+ 4480 train 3.977406 (lr=3.4490e-04) (hash(x)=154064204)
541
+ 4490 train 4.111873 (lr=3.4387e-04) (hash(x)=153999958)
542
+ 4500 val loss 4.2383
543
+ 4500 val perplexity 69.2910
544
+ 4500 train 4.268386 (lr=3.4285e-04) (hash(x)=125588037)
545
+ 4510 train 4.070775 (lr=3.4182e-04) (hash(x)=142522186)
546
+ 4520 train 3.913042 (lr=3.4079e-04) (hash(x)=152736893)
547
+ 4530 train 4.146557 (lr=3.3977e-04) (hash(x)=148091275)
548
+ 4540 train 4.261276 (lr=3.3874e-04) (hash(x)=147663019)
549
+ 4550 train 4.349103 (lr=3.3771e-04) (hash(x)=151491857)
550
+ 4560 train 4.226416 (lr=3.3668e-04) (hash(x)=161757513)
551
+ 4570 train 4.218224 (lr=3.3565e-04) (hash(x)=165448260)
552
+ 4580 train 4.352017 (lr=3.3463e-04) (hash(x)=151620292)
553
+ 4590 train 4.438036 (lr=3.3360e-04) (hash(x)=161877852)
554
+ 4600 val loss 4.2034
555
+ 4600 val perplexity 66.9136
556
+ 4600 train 4.185835 (lr=3.3257e-04) (hash(x)=143710941)
557
+ 4610 train 4.278987 (lr=3.3154e-04) (hash(x)=152244871)
558
+ 4620 train 4.229492 (lr=3.3051e-04) (hash(x)=161268167)
559
+ 4630 train 4.219322 (lr=3.2949e-04) (hash(x)=153904677)
560
+ 4640 train 4.165647 (lr=3.2846e-04) (hash(x)=152593758)
561
+ 4650 train 4.321199 (lr=3.2743e-04) (hash(x)=159047122)
562
+ 4660 train 4.386488 (lr=3.2640e-04) (hash(x)=167370244)
563
+ 4670 train 4.163472 (lr=3.2537e-04) (hash(x)=140357684)
564
+ 4680 train 4.319572 (lr=3.2435e-04) (hash(x)=140547853)
565
+ 4690 train 4.501917 (lr=3.2332e-04) (hash(x)=143000949)
566
+ 4700 val loss 4.1964
567
+ 4700 val perplexity 66.4450
568
+ 4700 train 4.127630 (lr=3.2229e-04) (hash(x)=150952742)
569
+ 4710 train 4.185342 (lr=3.2126e-04) (hash(x)=157964406)
570
+ 4720 train 4.078740 (lr=3.2023e-04) (hash(x)=147653608)
571
+ 4730 train 4.153180 (lr=3.1921e-04) (hash(x)=165545764)
572
+ 4740 train 4.135590 (lr=3.1818e-04) (hash(x)=146744631)
573
+ 4750 train 4.219508 (lr=3.1715e-04) (hash(x)=153635164)
574
+ 4760 train 4.262060 (lr=3.1613e-04) (hash(x)=153217508)
575
+ 4770 train 4.261677 (lr=3.1510e-04) (hash(x)=169622743)
576
+ 4780 train 4.133631 (lr=3.1407e-04) (hash(x)=138018297)
577
+ 4790 train 4.005082 (lr=3.1305e-04) (hash(x)=156067303)
578
+ 4800 val loss 4.1969
579
+ 4800 val perplexity 66.4819
580
+ 4800 train 4.187222 (lr=3.1202e-04) (hash(x)=145323659)
581
+ 4810 train 4.070339 (lr=3.1099e-04) (hash(x)=152587772)
582
+ 4820 train 4.287631 (lr=3.0997e-04) (hash(x)=148481485)
583
+ 4830 train 4.057197 (lr=3.0894e-04) (hash(x)=154219664)
584
+ 4840 train 3.946155 (lr=3.0792e-04) (hash(x)=156853645)
585
+ 4850 train 4.032663 (lr=3.0689e-04) (hash(x)=152329702)
586
+ 4860 train 4.020396 (lr=3.0587e-04) (hash(x)=136684779)
587
+ 4870 train 4.087455 (lr=3.0485e-04) (hash(x)=143260367)
588
+ 4880 train 4.055943 (lr=3.0382e-04) (hash(x)=150910947)
589
+ 4890 train 4.072757 (lr=3.0280e-04) (hash(x)=179249925)
590
+ 4900 val loss 4.2005
591
+ 4900 val perplexity 66.7193
592
+ 4900 train 4.222466 (lr=3.0178e-04) (hash(x)=153151397)
593
+ 4910 train 4.171384 (lr=3.0075e-04) (hash(x)=144101891)
594
+ 4920 train 4.075677 (lr=2.9973e-04) (hash(x)=144994952)
595
+ 4930 train 4.372343 (lr=2.9871e-04) (hash(x)=155390878)
596
+ 4940 train 4.133051 (lr=2.9769e-04) (hash(x)=141359221)
597
+ 4950 train 4.044434 (lr=2.9667e-04) (hash(x)=151570011)
598
+ 4960 train 4.257496 (lr=2.9565e-04) (hash(x)=146918855)
599
+ 4970 train 4.197004 (lr=2.9463e-04) (hash(x)=148647362)
600
+ 4980 train 4.150333 (lr=2.9361e-04) (hash(x)=159753562)
601
+ 4990 train 4.100173 (lr=2.9259e-04) (hash(x)=147683447)
602
+ 5000 val loss 4.1939
603
+ 5000 val perplexity 66.2829
604
+ 5000 train 4.240450 (lr=2.9157e-04) (hash(x)=143182059)
605
+ 5010 train 4.041979 (lr=2.9056e-04) (hash(x)=138657407)
606
+ 5020 train 4.159636 (lr=2.8954e-04) (hash(x)=151830497)
607
+ 5030 train 4.150501 (lr=2.8852e-04) (hash(x)=147727694)
608
+ 5040 train 3.995210 (lr=2.8751e-04) (hash(x)=151957014)
609
+ 5050 train 4.259698 (lr=2.8649e-04) (hash(x)=147828693)
610
+ 5060 train 4.305301 (lr=2.8548e-04) (hash(x)=158589700)
611
+ 5070 train 4.199480 (lr=2.8447e-04) (hash(x)=155662163)
612
+ 5080 train 4.302701 (lr=2.8345e-04) (hash(x)=157797383)
613
+ 5090 train 4.175357 (lr=2.8244e-04) (hash(x)=153848957)
614
+ 5100 val loss 4.1780
615
+ 5100 val perplexity 65.2353
616
+ 5100 train 4.133500 (lr=2.8143e-04) (hash(x)=170083586)
617
+ 5110 train 4.048562 (lr=2.8042e-04) (hash(x)=153800138)
618
+ 5120 train 4.160891 (lr=2.7941e-04) (hash(x)=153526228)
619
+ 5130 train 4.167576 (lr=2.7840e-04) (hash(x)=158650342)
620
+ 5140 train 4.158683 (lr=2.7739e-04) (hash(x)=153163034)
621
+ 5150 train 4.151326 (lr=2.7638e-04) (hash(x)=155281294)
622
+ 5160 train 4.130368 (lr=2.7537e-04) (hash(x)=147731627)
623
+ 5170 train 4.103950 (lr=2.7437e-04) (hash(x)=146388976)
624
+ 5180 train 4.327842 (lr=2.7336e-04) (hash(x)=140323065)
625
+ 5190 train 4.230986 (lr=2.7236e-04) (hash(x)=150406523)
626
+ 5200 val loss 4.1711
627
+ 5200 val perplexity 64.7888
628
+ 5200 train 4.030345 (lr=2.7135e-04) (hash(x)=149363919)
629
+ 5210 train 4.180054 (lr=2.7035e-04) (hash(x)=164390214)
630
+ 5220 train 4.126330 (lr=2.6935e-04) (hash(x)=153652984)
631
+ 5230 train 4.097653 (lr=2.6835e-04) (hash(x)=139542770)
632
+ 5240 train 3.997225 (lr=2.6734e-04) (hash(x)=140115288)
633
+ 5250 train 4.117519 (lr=2.6635e-04) (hash(x)=142158900)
634
+ 5260 train 4.030591 (lr=2.6535e-04) (hash(x)=156019449)
635
+ 5270 train 4.148395 (lr=2.6435e-04) (hash(x)=150776850)
636
+ 5280 train 4.106181 (lr=2.6335e-04) (hash(x)=151358692)
637
+ 5290 train 4.069046 (lr=2.6236e-04) (hash(x)=137112863)
638
+ 5300 val loss 4.1719
639
+ 5300 val perplexity 64.8381
640
+ 5300 train 3.998692 (lr=2.6136e-04) (hash(x)=152033784)
641
+ 5310 train 4.039620 (lr=2.6037e-04) (hash(x)=147785712)
642
+ 5320 train 4.186316 (lr=2.5937e-04) (hash(x)=156426339)
643
+ 5330 train 4.134931 (lr=2.5838e-04) (hash(x)=161907693)
644
+ 5340 train 4.201005 (lr=2.5739e-04) (hash(x)=156414094)
645
+ 5350 train 4.226521 (lr=2.5640e-04) (hash(x)=141712428)
646
+ 5360 train 4.178817 (lr=2.5541e-04) (hash(x)=156873601)
647
+ 5370 train 4.156247 (lr=2.5443e-04) (hash(x)=149863781)
648
+ 5380 train 4.154983 (lr=2.5344e-04) (hash(x)=148183971)
649
+ 5390 train 4.120444 (lr=2.5245e-04) (hash(x)=151542334)
650
+ 5400 val loss 4.1489
651
+ 5400 val perplexity 63.3638
652
+ 5400 train 4.392386 (lr=2.5147e-04) (hash(x)=154614289)
653
+ 5410 train 4.137814 (lr=2.5049e-04) (hash(x)=169124671)
654
+ 5420 train 4.161612 (lr=2.4950e-04) (hash(x)=159963738)
655
+ 5430 train 4.128326 (lr=2.4852e-04) (hash(x)=155162691)
656
+ 5440 train 4.146622 (lr=2.4754e-04) (hash(x)=135981908)
657
+ 5450 train 4.234727 (lr=2.4657e-04) (hash(x)=145055580)
658
+ 5460 train 4.094577 (lr=2.4559e-04) (hash(x)=150367825)
659
+ 5470 train 4.193575 (lr=2.4461e-04) (hash(x)=151034173)
660
+ 5480 train 4.105726 (lr=2.4364e-04) (hash(x)=155581766)
661
+ 5490 train 4.132806 (lr=2.4266e-04) (hash(x)=136973993)
662
+ 5500 val loss 4.1468
663
+ 5500 val perplexity 63.2302
664
+ 5500 train 4.306717 (lr=2.4169e-04) (hash(x)=157745174)
665
+ 5510 train 4.254641 (lr=2.4072e-04) (hash(x)=144628894)
666
+ 5520 train 4.132394 (lr=2.3975e-04) (hash(x)=135056684)
667
+ 5530 train 4.123548 (lr=2.3878e-04) (hash(x)=143418640)
668
+ 5540 train 4.097945 (lr=2.3782e-04) (hash(x)=144464984)
669
+ 5550 train 4.313401 (lr=2.3685e-04) (hash(x)=155888745)
670
+ 5560 train 3.968264 (lr=2.3589e-04) (hash(x)=149122683)
671
+ 5570 train 4.012491 (lr=2.3492e-04) (hash(x)=154573954)
672
+ 5580 train 4.115549 (lr=2.3396e-04) (hash(x)=149339361)
673
+ 5590 train 4.009767 (lr=2.3300e-04) (hash(x)=154738453)
674
+ 5600 val loss 4.1458
675
+ 5600 val perplexity 63.1676
676
+ 5600 train 4.122560 (lr=2.3204e-04) (hash(x)=147693222)
677
+ 5610 train 3.897916 (lr=2.3108e-04) (hash(x)=142140192)
678
+ 5620 train 4.036131 (lr=2.3013e-04) (hash(x)=148677209)
679
+ 5630 train 4.097919 (lr=2.2917e-04) (hash(x)=143495554)
680
+ 5640 train 4.187385 (lr=2.2822e-04) (hash(x)=153202964)
681
+ 5650 train 4.110124 (lr=2.2727e-04) (hash(x)=168724867)
682
+ 5660 train 4.081139 (lr=2.2632e-04) (hash(x)=142816746)
683
+ 5670 train 4.202588 (lr=2.2537e-04) (hash(x)=164119551)
684
+ 5680 train 4.070452 (lr=2.2442e-04) (hash(x)=157239121)
685
+ 5690 train 4.348936 (lr=2.2348e-04) (hash(x)=144662329)
686
+ 5700 val loss 4.1334
687
+ 5700 val perplexity 62.3916
688
+ 5700 train 4.065941 (lr=2.2253e-04) (hash(x)=149784627)
689
+ 5710 train 4.037819 (lr=2.2159e-04) (hash(x)=152821162)
690
+ 5720 train 4.254797 (lr=2.2065e-04) (hash(x)=158497530)
691
+ 5730 train 4.242295 (lr=2.1971e-04) (hash(x)=154583914)
692
+ 5740 train 4.054935 (lr=2.1877e-04) (hash(x)=142027172)
693
+ 5750 train 4.294274 (lr=2.1784e-04) (hash(x)=119227402)
694
+ 5760 train 4.150327 (lr=2.1690e-04) (hash(x)=153248187)
695
+ 5770 train 4.027086 (lr=2.1597e-04) (hash(x)=147951994)
696
+ 5780 train 4.039157 (lr=2.1504e-04) (hash(x)=138935284)
697
+ 5790 train 4.120954 (lr=2.1411e-04) (hash(x)=151786681)
698
+ 5800 val loss 4.1388
699
+ 5800 val perplexity 62.7273
700
+ 5800 train 4.160823 (lr=2.1318e-04) (hash(x)=158620729)
701
+ 5810 train 4.107616 (lr=2.1226e-04) (hash(x)=178115789)
702
+ 5820 train 4.024902 (lr=2.1133e-04) (hash(x)=149092790)
703
+ 5830 train 4.210721 (lr=2.1041e-04) (hash(x)=148269833)
704
+ 5840 train 4.144323 (lr=2.0949e-04) (hash(x)=163714738)
705
+ 5850 train 4.114985 (lr=2.0857e-04) (hash(x)=144111890)
706
+ 5860 train 4.090931 (lr=2.0765e-04) (hash(x)=170702475)
707
+ 5870 train 4.021220 (lr=2.0674e-04) (hash(x)=155970835)
708
+ 5880 train 4.033174 (lr=2.0582e-04) (hash(x)=146695463)
709
+ 5890 train 4.095443 (lr=2.0491e-04) (hash(x)=151603187)
710
+ 5900 val loss 4.1288
711
+ 5900 val perplexity 62.1056
712
+ 5900 train 4.114845 (lr=2.0400e-04) (hash(x)=159763910)
713
+ 5910 train 4.150571 (lr=2.0309e-04) (hash(x)=149555230)
714
+ 5920 train 4.158465 (lr=2.0218e-04) (hash(x)=138048333)
715
+ 5930 train 4.050347 (lr=2.0128e-04) (hash(x)=148668256)
716
+ 5940 train 4.192036 (lr=2.0038e-04) (hash(x)=148352543)
717
+ 5950 train 4.203560 (lr=1.9948e-04) (hash(x)=163457488)
718
+ 5960 train 4.282877 (lr=1.9858e-04) (hash(x)=155306804)
719
+ 5970 train 4.133232 (lr=1.9768e-04) (hash(x)=146576143)
720
+ 5980 train 4.133700 (lr=1.9678e-04) (hash(x)=144668956)
721
+ 5990 train 4.105962 (lr=1.9589e-04) (hash(x)=156098617)
722
+ 6000 val loss 4.1163
723
+ 6000 val perplexity 61.3293
724
+ 6000 train 4.083284 (lr=1.9500e-04) (hash(x)=147640561)
725
+ 6010 train 4.153471 (lr=1.9411e-04) (hash(x)=161223074)
726
+ 6020 train 4.136727 (lr=1.9322e-04) (hash(x)=156308454)
727
+ 6030 train 4.115203 (lr=1.9234e-04) (hash(x)=142553975)
728
+ 6040 train 4.075415 (lr=1.9145e-04) (hash(x)=142039829)
729
+ 6050 train 4.148516 (lr=1.9057e-04) (hash(x)=156329296)
730
+ 6060 train 4.081771 (lr=1.8969e-04) (hash(x)=148505537)
731
+ 6070 train 3.980475 (lr=1.8882e-04) (hash(x)=151801272)
732
+ 6080 train 4.062029 (lr=1.8794e-04) (hash(x)=159908287)
733
+ 6090 train 4.057708 (lr=1.8707e-04) (hash(x)=148714721)
734
+ 6100 val loss 4.1107
735
+ 6100 val perplexity 60.9872
736
+ 6100 train 4.118686 (lr=1.8620e-04) (hash(x)=156613394)
737
+ 6110 train 3.978684 (lr=1.8533e-04) (hash(x)=143286705)
738
+ 6120 train 4.188768 (lr=1.8446e-04) (hash(x)=150191141)
739
+ 6130 train 4.054944 (lr=1.8359e-04) (hash(x)=149855624)
740
+ 6140 train 4.047987 (lr=1.8273e-04) (hash(x)=157312372)
741
+ 6150 train 4.173122 (lr=1.8187e-04) (hash(x)=148262730)
742
+ 6160 train 4.084437 (lr=1.8101e-04) (hash(x)=147032785)
743
+ 6170 train 4.012024 (lr=1.8016e-04) (hash(x)=161213064)
744
+ 6180 train 3.885677 (lr=1.7930e-04) (hash(x)=144538373)
745
+ 6190 train 3.939042 (lr=1.7845e-04) (hash(x)=144688493)
746
+ 6200 val loss 4.1090
747
+ 6200 val perplexity 60.8854
748
+ 6200 train 4.017736 (lr=1.7760e-04) (hash(x)=186221290)
749
+ 6210 train 4.022638 (lr=1.7675e-04) (hash(x)=139529352)
750
+ 6220 train 4.190411 (lr=1.7591e-04) (hash(x)=138081469)
751
+ 6230 train 4.063618 (lr=1.7506e-04) (hash(x)=152979409)
752
+ 6240 train 4.037992 (lr=1.7422e-04) (hash(x)=157848424)
753
+ 6250 train 3.931513 (lr=1.7338e-04) (hash(x)=150770915)
754
+ 6260 train 4.026513 (lr=1.7255e-04) (hash(x)=153769379)
755
+ 6270 train 3.967563 (lr=1.7171e-04) (hash(x)=153318361)
756
+ 6280 train 3.958702 (lr=1.7088e-04) (hash(x)=143366327)
757
+ 6290 train 4.119064 (lr=1.7005e-04) (hash(x)=160684845)
758
+ 6300 val loss 4.0941
759
+ 6300 val perplexity 59.9859
760
+ 6300 train 4.011855 (lr=1.6923e-04) (hash(x)=152081419)
761
+ 6310 train 3.965437 (lr=1.6840e-04) (hash(x)=146626797)
762
+ 6320 train 4.154341 (lr=1.6758e-04) (hash(x)=155291923)
763
+ 6330 train 4.025539 (lr=1.6676e-04) (hash(x)=140920951)
764
+ 6340 train 4.103500 (lr=1.6594e-04) (hash(x)=146642359)
765
+ 6350 train 4.080964 (lr=1.6513e-04) (hash(x)=160486550)
766
+ 6360 train 4.027030 (lr=1.6431e-04) (hash(x)=151833082)
767
+ 6370 train 4.025888 (lr=1.6350e-04) (hash(x)=124740362)
768
+ 6380 train 4.176577 (lr=1.6269e-04) (hash(x)=153952434)
769
+ 6390 train 4.211592 (lr=1.6189e-04) (hash(x)=145864314)
770
+ 6400 val loss 4.0973
771
+ 6400 val perplexity 60.1754
772
+ 6400 train 4.101017 (lr=1.6108e-04) (hash(x)=154808349)
773
+ 6410 train 4.075342 (lr=1.6028e-04) (hash(x)=155037739)
774
+ 6420 train 4.011328 (lr=1.5948e-04) (hash(x)=151155588)
775
+ 6430 train 3.900066 (lr=1.5869e-04) (hash(x)=146548556)
776
+ 6440 train 4.067854 (lr=1.5790e-04) (hash(x)=152706474)
777
+ 6450 train 4.060566 (lr=1.5710e-04) (hash(x)=154207794)
778
+ 6460 train 3.932173 (lr=1.5632e-04) (hash(x)=141590515)
779
+ 6470 train 3.965304 (lr=1.5553e-04) (hash(x)=157673109)
780
+ 6480 train 3.978065 (lr=1.5475e-04) (hash(x)=155842980)
781
+ 6490 train 4.059033 (lr=1.5397e-04) (hash(x)=143595736)
782
+ 6500 val loss 4.0948
783
+ 6500 val perplexity 60.0293
784
+ 6500 train 4.010231 (lr=1.5319e-04) (hash(x)=159437208)
785
+ 6510 train 4.033210 (lr=1.5241e-04) (hash(x)=142631317)
786
+ 6520 train 3.751716 (lr=1.5164e-04) (hash(x)=139009211)
787
+ 6530 train 4.101348 (lr=1.5087e-04) (hash(x)=145451137)
788
+ 6540 train 4.057612 (lr=1.5010e-04) (hash(x)=163532779)
789
+ 6550 train 3.964723 (lr=1.4933e-04) (hash(x)=158084674)
790
+ 6560 train 3.941634 (lr=1.4857e-04) (hash(x)=142422592)
791
+ 6570 train 3.860327 (lr=1.4781e-04) (hash(x)=141089289)
792
+ 6580 train 3.968050 (lr=1.4705e-04) (hash(x)=139712979)
793
+ 6590 train 3.870108 (lr=1.4630e-04) (hash(x)=145833805)
794
+ 6600 val loss 4.0944
795
+ 6600 val perplexity 60.0050
796
+ 6600 train 3.943348 (lr=1.4555e-04) (hash(x)=157933074)
797
+ 6610 train 3.865143 (lr=1.4480e-04) (hash(x)=142534769)
798
+ 6620 train 3.917859 (lr=1.4405e-04) (hash(x)=138456044)
799
+ 6630 train 4.053186 (lr=1.4331e-04) (hash(x)=145111052)
800
+ 6640 train 4.014345 (lr=1.4257e-04) (hash(x)=166954515)
801
+ 6650 train 3.979203 (lr=1.4183e-04) (hash(x)=146889543)
802
+ 6660 train 4.003115 (lr=1.4109e-04) (hash(x)=143253726)
803
+ 6670 train 3.921113 (lr=1.4036e-04) (hash(x)=144349844)
804
+ 6680 train 4.112489 (lr=1.3963e-04) (hash(x)=139911928)
805
+ 6690 train 4.301867 (lr=1.3890e-04) (hash(x)=143135862)
806
+ 6700 val loss 4.0850
807
+ 6700 val perplexity 59.4434
808
+ 6700 train 4.018515 (lr=1.3817e-04) (hash(x)=161560240)
809
+ 6710 train 4.101259 (lr=1.3745e-04) (hash(x)=150437730)
810
+ 6720 train 4.204084 (lr=1.3673e-04) (hash(x)=153035304)
811
+ 6730 train 3.994936 (lr=1.3602e-04) (hash(x)=139323796)
812
+ 6740 train 4.004852 (lr=1.3530e-04) (hash(x)=167437923)
813
+ 6750 train 3.967689 (lr=1.3459e-04) (hash(x)=143814596)
814
+ 6760 train 4.031361 (lr=1.3388e-04) (hash(x)=143755031)
815
+ 6770 train 3.984771 (lr=1.3318e-04) (hash(x)=142014038)
816
+ 6780 train 4.048668 (lr=1.3248e-04) (hash(x)=148465062)
817
+ 6790 train 4.075753 (lr=1.3178e-04) (hash(x)=154656877)
818
+ 6800 val loss 4.0813
819
+ 6800 val perplexity 59.2223
820
+ 6800 train 4.186028 (lr=1.3108e-04) (hash(x)=155424292)
821
+ 6810 train 3.820129 (lr=1.3039e-04) (hash(x)=147358081)
822
+ 6820 train 3.936265 (lr=1.2970e-04) (hash(x)=145387291)
823
+ 6830 train 3.885121 (lr=1.2901e-04) (hash(x)=146334068)
824
+ 6840 train 3.962760 (lr=1.2832e-04) (hash(x)=139563766)
825
+ 6850 train 3.988779 (lr=1.2764e-04) (hash(x)=139095691)
826
+ 6860 train 3.937799 (lr=1.2696e-04) (hash(x)=142093999)
827
+ 6870 train 4.044878 (lr=1.2628e-04) (hash(x)=148598823)
828
+ 6880 train 4.131179 (lr=1.2561e-04) (hash(x)=146486085)
829
+ 6890 train 3.915892 (lr=1.2494e-04) (hash(x)=165683212)
830
+ 6900 val loss 4.0853
831
+ 6900 val perplexity 59.4604
832
+ 6900 train 4.127148 (lr=1.2427e-04) (hash(x)=148561470)
833
+ 6910 train 3.814578 (lr=1.2361e-04) (hash(x)=136352432)
834
+ 6920 train 4.185750 (lr=1.2295e-04) (hash(x)=147640094)
835
+ 6930 train 4.071712 (lr=1.2229e-04) (hash(x)=152343827)
836
+ 6940 train 4.055577 (lr=1.2163e-04) (hash(x)=158787971)
837
+ 6950 train 4.334540 (lr=1.2098e-04) (hash(x)=178133125)
838
+ 6960 train 4.124120 (lr=1.2033e-04) (hash(x)=167302853)
839
+ 6970 train 4.055647 (lr=1.1969e-04) (hash(x)=143729435)
840
+ 6980 train 4.153141 (lr=1.1904e-04) (hash(x)=135857879)
841
+ 6990 train 4.104347 (lr=1.1840e-04) (hash(x)=155384510)
842
+ 7000 val loss 4.0727
843
+ 7000 val perplexity 58.7175
844
+ 7000 train 4.035264 (lr=1.1777e-04) (hash(x)=141527450)
845
+ 7010 train 4.084986 (lr=1.1713e-04) (hash(x)=148648498)
846
+ 7020 train 4.003251 (lr=1.1650e-04) (hash(x)=148073842)
847
+ 7030 train 4.144956 (lr=1.1587e-04) (hash(x)=160771614)
848
+ 7040 train 3.917357 (lr=1.1525e-04) (hash(x)=148361330)
849
+ 7050 train 3.900904 (lr=1.1463e-04) (hash(x)=151212224)
850
+ 7060 train 4.096613 (lr=1.1401e-04) (hash(x)=155925366)
851
+ 7070 train 3.943273 (lr=1.1339e-04) (hash(x)=149830576)
852
+ 7080 train 3.912623 (lr=1.1278e-04) (hash(x)=144849934)
853
+ 7090 train 4.002218 (lr=1.1217e-04) (hash(x)=150997807)
854
+ 7100 val loss 4.0693
855
+ 7100 val perplexity 58.5131
856
+ 7100 train 4.046739 (lr=1.1157e-04) (hash(x)=151066339)
857
+ 7110 train 3.950194 (lr=1.1096e-04) (hash(x)=147185781)
858
+ 7120 train 3.989608 (lr=1.1036e-04) (hash(x)=141765575)
859
+ 7130 train 4.004335 (lr=1.0977e-04) (hash(x)=144204513)
860
+ 7140 train 3.840174 (lr=1.0917e-04) (hash(x)=149326253)
861
+ 7150 train 3.998312 (lr=1.0858e-04) (hash(x)=150409811)
862
+ 7160 train 3.887107 (lr=1.0800e-04) (hash(x)=144790348)
863
+ 7170 train 4.007476 (lr=1.0741e-04) (hash(x)=150500939)
864
+ 7180 train 3.830336 (lr=1.0683e-04) (hash(x)=143552103)
865
+ 7190 train 3.878871 (lr=1.0626e-04) (hash(x)=157471075)
866
+ 7200 val loss 4.0730
867
+ 7200 val perplexity 58.7358
868
+ 7200 train 3.858993 (lr=1.0568e-04) (hash(x)=155231264)
869
+ 7210 train 3.945701 (lr=1.0511e-04) (hash(x)=150350401)
870
+ 7220 train 3.965366 (lr=1.0454e-04) (hash(x)=151952594)
871
+ 7230 train 4.043124 (lr=1.0398e-04) (hash(x)=145381804)
872
+ 7240 train 3.924385 (lr=1.0342e-04) (hash(x)=142651222)
873
+ 7250 train 4.038084 (lr=1.0286e-04) (hash(x)=136115279)
874
+ 7260 train 4.139531 (lr=1.0231e-04) (hash(x)=154536751)
875
+ 7270 train 4.089895 (lr=1.0176e-04) (hash(x)=148883836)
876
+ 7280 train 3.946015 (lr=1.0121e-04) (hash(x)=150606754)
877
+ 7290 train 4.061311 (lr=1.0066e-04) (hash(x)=165845807)
878
+ 7300 val loss 4.0629
879
+ 7300 val perplexity 58.1432
880
+ 7300 train 4.140318 (lr=1.0012e-04) (hash(x)=150281149)
881
+ 7310 train 4.089637 (lr=9.9586e-05) (hash(x)=159929281)
882
+ 7320 train 3.912396 (lr=9.9052e-05) (hash(x)=143074269)
883
+ 7330 train 4.074787 (lr=9.8521e-05) (hash(x)=147369269)
884
+ 7340 train 4.119080 (lr=9.7993e-05) (hash(x)=139286601)
885
+ 7350 train 4.024800 (lr=9.7469e-05) (hash(x)=145263048)
886
+ 7360 train 3.882262 (lr=9.6948e-05) (hash(x)=144626817)
887
+ 7370 train 4.076595 (lr=9.6431e-05) (hash(x)=168989813)
888
+ 7380 train 4.058831 (lr=9.5917e-05) (hash(x)=144790656)
889
+ 7390 train 3.866611 (lr=9.5406e-05) (hash(x)=141279761)
890
+ 7400 val loss 4.0601
891
+ 7400 val perplexity 57.9790
892
+ 7400 train 3.996303 (lr=9.4899e-05) (hash(x)=148421717)
893
+ 7410 train 3.993309 (lr=9.4395e-05) (hash(x)=141241548)
894
+ 7420 train 4.055984 (lr=9.3894e-05) (hash(x)=162434228)
895
+ 7430 train 4.009074 (lr=9.3397e-05) (hash(x)=156312086)
896
+ 7440 train 3.964110 (lr=9.2904e-05) (hash(x)=139803064)
897
+ 7450 train 3.670104 (lr=9.2413e-05) (hash(x)=156361394)
898
+ 7460 train 3.975402 (lr=9.1927e-05) (hash(x)=141142183)
899
+ 7470 train 3.806543 (lr=9.1443e-05) (hash(x)=156701683)
900
+ 7480 train 3.977510 (lr=9.0964e-05) (hash(x)=148549917)
901
+ 7490 train 3.854869 (lr=9.0487e-05) (hash(x)=143002119)
902
+ 7500 val loss 4.0618
903
+ 7500 val perplexity 58.0814
904
+ 7500 train 3.868737 (lr=9.0014e-05) (hash(x)=146921118)
905
+ 7510 train 3.875194 (lr=8.9545e-05) (hash(x)=148592146)
906
+ 7520 train 4.017394 (lr=8.9079e-05) (hash(x)=148456120)
907
+ 7530 train 4.181747 (lr=8.8617e-05) (hash(x)=163865945)
908
+ 7540 train 4.069669 (lr=8.8158e-05) (hash(x)=160753039)
909
+ 7550 train 4.081968 (lr=8.7702e-05) (hash(x)=150275989)
910
+ 7560 train 4.105621 (lr=8.7251e-05) (hash(x)=141749735)
911
+ 7570 train 4.075472 (lr=8.6802e-05) (hash(x)=148449999)
912
+ 7580 train 3.949388 (lr=8.6357e-05) (hash(x)=139153889)
913
+ 7590 train 4.065652 (lr=8.5916e-05) (hash(x)=147418540)
914
+ 7600 val loss 4.0518
915
+ 7600 val perplexity 57.4982
916
+ 7600 train 4.093127 (lr=8.5478e-05) (hash(x)=150660048)
917
+ 7610 train 4.071154 (lr=8.5044e-05) (hash(x)=155369080)
918
+ 7620 train 3.965354 (lr=8.4613e-05) (hash(x)=151926309)
919
+ 7630 train 4.049876 (lr=8.4186e-05) (hash(x)=149549080)
920
+ 7640 train 3.934168 (lr=8.3763e-05) (hash(x)=140642882)
921
+ 7650 train 4.211053 (lr=8.3343e-05) (hash(x)=140218502)
922
+ 7660 train 4.052149 (lr=8.2926e-05) (hash(x)=153198276)
923
+ 7670 train 3.904365 (lr=8.2514e-05) (hash(x)=150706960)
924
+ 7680 train 4.057087 (lr=8.2104e-05) (hash(x)=152025509)
925
+ 7690 train 3.841843 (lr=8.1699e-05) (hash(x)=147288976)
926
+ 7700 val loss 4.0519
927
+ 7700 val perplexity 57.5041
928
+ 7700 train 3.947350 (lr=8.1297e-05) (hash(x)=148059852)
929
+ 7710 train 3.769231 (lr=8.0898e-05) (hash(x)=158443334)
930
+ 7720 train 3.876155 (lr=8.0503e-05) (hash(x)=143467085)
931
+ 7730 train 3.824585 (lr=8.0112e-05) (hash(x)=138439550)
932
+ 7740 train 3.965752 (lr=7.9725e-05) (hash(x)=145878152)
933
+ 7750 train 3.818537 (lr=7.9341e-05) (hash(x)=136874128)
934
+ 7760 train 3.840930 (lr=7.8960e-05) (hash(x)=148856154)
935
+ 7770 train 3.871827 (lr=7.8584e-05) (hash(x)=146982119)
936
+ 7780 train 3.963326 (lr=7.8211e-05) (hash(x)=159846779)
937
+ 7790 train 3.649072 (lr=7.7841e-05) (hash(x)=141501838)
938
+ 7800 val loss 4.0548
939
+ 7800 val perplexity 57.6720
940
+ 7800 train 3.776215 (lr=7.7476e-05) (hash(x)=148331002)
941
+ 7810 train 4.088730 (lr=7.7114e-05) (hash(x)=150692304)
942
+ 7820 train 3.956343 (lr=7.6755e-05) (hash(x)=156841183)
943
+ 7830 train 3.999827 (lr=7.6400e-05) (hash(x)=139594008)
944
+ 7840 train 3.940042 (lr=7.6049e-05) (hash(x)=149700138)
945
+ 7850 train 3.959618 (lr=7.5702e-05) (hash(x)=149327044)
946
+ 7860 train 4.125587 (lr=7.5358e-05) (hash(x)=148315418)
947
+ 7870 train 4.069400 (lr=7.5018e-05) (hash(x)=153431778)
948
+ 7880 train 3.913067 (lr=7.4682e-05) (hash(x)=145049799)
949
+ 7890 train 4.042965 (lr=7.4350e-05) (hash(x)=150265096)
950
+ 7900 val loss 4.0453
951
+ 7900 val perplexity 57.1296
952
+ 7900 train 3.878649 (lr=7.4021e-05) (hash(x)=164923883)
953
+ 7910 train 3.916581 (lr=7.3696e-05) (hash(x)=139315738)
954
+ 7920 train 3.979507 (lr=7.3374e-05) (hash(x)=153031394)
955
+ 7930 train 3.763025 (lr=7.3056e-05) (hash(x)=148835859)
956
+ 7940 train 3.890374 (lr=7.2742e-05) (hash(x)=146875217)
957
+ 7950 train 3.974443 (lr=7.2432e-05) (hash(x)=151003885)
958
+ 7960 train 3.888942 (lr=7.2126e-05) (hash(x)=141885471)
959
+ 7970 train 4.087327 (lr=7.1823e-05) (hash(x)=150541583)
960
+ 7980 train 3.786850 (lr=7.1524e-05) (hash(x)=138626492)
961
+ 7990 train 3.975272 (lr=7.1228e-05) (hash(x)=149935635)
962
+ 8000 val loss 4.0454
963
+ 8000 val perplexity 57.1320
964
+ 8000 train 3.946618 (lr=7.0937e-05) (hash(x)=143545384)
965
+ 8010 train 3.801200 (lr=7.0649e-05) (hash(x)=144689366)
966
+ 8020 train 3.978382 (lr=7.0365e-05) (hash(x)=151299532)
967
+ 8030 train 3.919228 (lr=7.0085e-05) (hash(x)=131300239)
968
+ 8040 train 3.821469 (lr=6.9808e-05) (hash(x)=146715955)
969
+ 8050 train 3.749524 (lr=6.9536e-05) (hash(x)=135376656)
970
+ 8060 train 3.831349 (lr=6.9267e-05) (hash(x)=141991433)
971
+ 8070 train 3.786435 (lr=6.9002e-05) (hash(x)=140260574)
972
+ 8080 train 3.759117 (lr=6.8740e-05) (hash(x)=151492896)
973
+ 8090 train 3.958464 (lr=6.8483e-05) (hash(x)=143333635)
974
+ 8100 val loss 4.0470
975
+ 8100 val perplexity 57.2254
976
+ 8100 train 3.842434 (lr=6.8229e-05) (hash(x)=160686959)
977
+ 8110 train 3.842722 (lr=6.7979e-05) (hash(x)=141145704)
978
+ 8120 train 4.083741 (lr=6.7733e-05) (hash(x)=146967795)
979
+ 8130 train 3.905297 (lr=6.7490e-05) (hash(x)=147844486)
980
+ 8140 train 3.961334 (lr=6.7252e-05) (hash(x)=141860136)
981
+ 8150 train 4.098149 (lr=6.7017e-05) (hash(x)=147696006)
982
+ 8160 train 4.001524 (lr=6.6786e-05) (hash(x)=151497110)
983
+ 8170 train 4.002586 (lr=6.6559e-05) (hash(x)=140881859)
984
+ 8180 train 3.954071 (lr=6.6335e-05) (hash(x)=158865574)
985
+ 8190 train 3.915143 (lr=6.6116e-05) (hash(x)=161247161)
986
+ 8200 val loss 4.0427
987
+ 8200 val perplexity 56.9826
988
+ 8200 train 3.872562 (lr=6.5900e-05) (hash(x)=156501889)
989
+ 8210 train 3.999889 (lr=6.5688e-05) (hash(x)=158384162)
990
+ 8220 train 4.042165 (lr=6.5480e-05) (hash(x)=154815062)
991
+ 8230 train 3.917685 (lr=6.5276e-05) (hash(x)=145018315)
992
+ 8240 train 3.959745 (lr=6.5076e-05) (hash(x)=161659129)
993
+ 8250 train 3.922977 (lr=6.4879e-05) (hash(x)=139863367)
994
+ 8260 train 3.877584 (lr=6.4687e-05) (hash(x)=149569663)
995
+ 8270 train 3.639469 (lr=6.4498e-05) (hash(x)=162234378)
996
+ 8280 train 4.017486 (lr=6.4313e-05) (hash(x)=144371378)
997
+ 8290 train 3.913353 (lr=6.4132e-05) (hash(x)=148120258)
998
+ 8300 val loss 4.0431
999
+ 8300 val perplexity 57.0041
1000
+ 8300 train 3.913517 (lr=6.3954e-05) (hash(x)=142716875)
1001
+ 8310 train 3.940189 (lr=6.3781e-05) (hash(x)=166789234)
1002
+ 8320 train 3.997204 (lr=6.3612e-05) (hash(x)=150613211)
1003
+ 8330 train 3.879534 (lr=6.3446e-05) (hash(x)=139895391)
1004
+ 8340 train 3.994856 (lr=6.3284e-05) (hash(x)=152011203)
1005
+ 8350 train 3.948846 (lr=6.3126e-05) (hash(x)=156010875)
1006
+ 8360 train 3.952508 (lr=6.2972e-05) (hash(x)=152988051)
1007
+ 8370 train 4.124910 (lr=6.2822e-05) (hash(x)=155144327)
1008
+ 8380 train 4.001295 (lr=6.2676e-05) (hash(x)=133642309)
1009
+ 8390 train 4.047203 (lr=6.2533e-05) (hash(x)=149745335)
1010
+ 8400 val loss 4.0356
1011
+ 8400 val perplexity 56.5782
1012
+ 8400 train 4.027781 (lr=6.2395e-05) (hash(x)=154436684)
1013
+ 8410 train 4.074440 (lr=6.2260e-05) (hash(x)=147946630)
1014
+ 8420 train 3.984002 (lr=6.2129e-05) (hash(x)=153947058)
1015
+ 8430 train 4.077157 (lr=6.2002e-05) (hash(x)=159821040)
1016
+ 8440 train 4.039477 (lr=6.1879e-05) (hash(x)=149127348)
1017
+ 8450 train 4.051155 (lr=6.1760e-05) (hash(x)=149421871)
1018
+ 8460 train 4.016480 (lr=6.1645e-05) (hash(x)=142621577)
1019
+ 8470 train 4.010669 (lr=6.1533e-05) (hash(x)=148978213)
1020
+ 8480 train 3.831121 (lr=6.1426e-05) (hash(x)=139959537)
1021
+ 8490 train 3.924183 (lr=6.1322e-05) (hash(x)=144232498)
1022
+ 8500 val loss 4.0348
1023
+ 8500 val perplexity 56.5314
1024
+ 8500 train 4.138939 (lr=6.1223e-05) (hash(x)=147965839)
1025
+ 8510 train 4.015815 (lr=6.1127e-05) (hash(x)=147679836)
1026
+ 8520 train 3.894574 (lr=6.1035e-05) (hash(x)=147486097)
1027
+ 8530 train 4.034348 (lr=6.0947e-05) (hash(x)=151655265)
1028
+ 8540 train 3.936261 (lr=6.0863e-05) (hash(x)=150866632)
1029
+ 8550 train 4.085984 (lr=6.0783e-05) (hash(x)=144355059)
1030
+ 8560 train 3.938924 (lr=6.0706e-05) (hash(x)=145518478)
1031
+ 8570 train 4.095529 (lr=6.0634e-05) (hash(x)=178973323)
1032
+ 8580 train 4.008992 (lr=6.0566e-05) (hash(x)=152773969)
1033
+ 8590 train 3.807635 (lr=6.0501e-05) (hash(x)=141805508)
1034
+ 8600 val loss 4.0379
1035
+ 8600 val perplexity 56.7092
1036
+ 8600 train 3.857514 (lr=6.0440e-05) (hash(x)=145228097)
1037
+ 8610 train 3.876768 (lr=6.0384e-05) (hash(x)=149525694)
1038
+ 8620 train 3.947881 (lr=6.0331e-05) (hash(x)=151651655)
1039
+ 8630 train 4.065199 (lr=6.0282e-05) (hash(x)=144918643)
1040
+ 8640 train 3.919204 (lr=6.0237e-05) (hash(x)=149721879)
1041
+ 8650 train 3.988091 (lr=6.0196e-05) (hash(x)=157684127)
1042
+ 8660 train 3.991989 (lr=6.0159e-05) (hash(x)=152426036)
1043
+ 8670 train 3.906412 (lr=6.0125e-05) (hash(x)=141396342)
1044
+ 8680 train 4.007999 (lr=6.0096e-05) (hash(x)=146459856)
1045
+ 8690 train 4.092370 (lr=6.0070e-05) (hash(x)=151766124)
1046
+ 8700 val loss 4.0300
1047
+ 8700 val perplexity 56.2583
1048
+ 8700 train 4.240069 (lr=6.0049e-05) (hash(x)=152910357)
1049
+ 8710 train 4.029060 (lr=6.0031e-05) (hash(x)=149724642)
1050
+ 8720 train 3.966117 (lr=6.0018e-05) (hash(x)=152483840)
1051
+ 8730 train 3.949208 (lr=6.0008e-05) (hash(x)=147289415)
1052
+ 8740 train 3.986015 (lr=6.0002e-05) (hash(x)=152254131)
1053
+ 8749 val loss 4.0301
1054
+ 8749 val perplexity 56.2674
lr6e-4_total_batch_size61440_baseline_seed1338/model_08749.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:02a8ed652e5388ffa39685e1c321db91dbf2ac8ffafd898415a5332de8e37eac
3
+ size 92843394
lr6e-4_total_batch_size61440_baseline_seed1338/optimizer_08749.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6af8a07a08f61fbe94e3598872d91720f9b78a06a832b888f41619f229d11b1
3
+ size 179406214