andrew-healey commited on
Commit
c4f3249
·
verified ·
1 Parent(s): f27610f

Upload folder using huggingface_hub

Browse files
lr8e-4_total_batch_size61440_baseline_seed1341/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_11/lr8e-4_total_batch_size61440_baseline_seed1341", "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": 1341, "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": 0.0008, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "8e-4_61440", "n_embd": 256}
lr8e-4_total_batch_size61440_baseline_seed1341/dataloader_08749.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:953385078aa3787b69fc6857dfd48b0a2cd2f4d27c6f8892e01211aca53d07f5
3
+ size 964
lr8e-4_total_batch_size61440_baseline_seed1341/log2.txt ADDED
@@ -0,0 +1,1054 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ max_steps: 8750
2
+ 0 val loss 11.3073
3
+ 0 val perplexity 81417.7266
4
+ 0 train 11.312937 (lr=1.6000e-06) (hash(x)=145079536)
5
+ 10 train 10.253681 (lr=1.7600e-05) (hash(x)=142935521)
6
+ 20 train 9.834095 (lr=3.3600e-05) (hash(x)=136726044)
7
+ 30 train 9.478757 (lr=4.9600e-05) (hash(x)=149725277)
8
+ 40 train 8.999757 (lr=6.5600e-05) (hash(x)=137527851)
9
+ 50 train 8.481867 (lr=8.1600e-05) (hash(x)=154745873)
10
+ 60 train 8.099227 (lr=9.7600e-05) (hash(x)=158449312)
11
+ 70 train 7.771982 (lr=1.1360e-04) (hash(x)=158867878)
12
+ 80 train 7.621725 (lr=1.2960e-04) (hash(x)=152693855)
13
+ 90 train 7.544149 (lr=1.4560e-04) (hash(x)=152089620)
14
+ 100 val loss 7.5127
15
+ 100 val perplexity 1831.1406
16
+ 100 train 7.648684 (lr=1.6160e-04) (hash(x)=155800595)
17
+ 110 train 7.528376 (lr=1.7760e-04) (hash(x)=156324852)
18
+ 120 train 7.293053 (lr=1.9360e-04) (hash(x)=148079157)
19
+ 130 train 7.322620 (lr=2.0960e-04) (hash(x)=151260070)
20
+ 140 train 7.192337 (lr=2.2560e-04) (hash(x)=161098412)
21
+ 150 train 7.091570 (lr=2.4160e-04) (hash(x)=148595389)
22
+ 160 train 6.995576 (lr=2.5760e-04) (hash(x)=162558230)
23
+ 170 train 6.718220 (lr=2.7360e-04) (hash(x)=136967959)
24
+ 180 train 6.968875 (lr=2.8960e-04) (hash(x)=160279390)
25
+ 190 train 6.660629 (lr=3.0560e-04) (hash(x)=139072589)
26
+ 200 val loss 6.7000
27
+ 200 val perplexity 812.4421
28
+ 200 train 6.522381 (lr=3.2160e-04) (hash(x)=145606733)
29
+ 210 train 6.611020 (lr=3.3760e-04) (hash(x)=151385952)
30
+ 220 train 6.468329 (lr=3.5360e-04) (hash(x)=141556767)
31
+ 230 train 6.667230 (lr=3.6960e-04) (hash(x)=159727511)
32
+ 240 train 6.278207 (lr=3.8560e-04) (hash(x)=145935602)
33
+ 250 train 6.294773 (lr=4.0160e-04) (hash(x)=138221231)
34
+ 260 train 6.352616 (lr=4.1760e-04) (hash(x)=148897684)
35
+ 270 train 6.859878 (lr=4.3360e-04) (hash(x)=155063217)
36
+ 280 train 6.186891 (lr=4.4960e-04) (hash(x)=147263890)
37
+ 290 train 6.256858 (lr=4.6560e-04) (hash(x)=151660516)
38
+ 300 val loss 6.2644
39
+ 300 val perplexity 525.5169
40
+ 300 train 6.197756 (lr=4.8160e-04) (hash(x)=150367139)
41
+ 310 train 5.935308 (lr=4.9760e-04) (hash(x)=151170832)
42
+ 320 train 6.119688 (lr=5.1360e-04) (hash(x)=138591183)
43
+ 330 train 5.854411 (lr=5.2960e-04) (hash(x)=143603420)
44
+ 340 train 5.969576 (lr=5.4560e-04) (hash(x)=140068134)
45
+ 350 train 6.033711 (lr=5.6160e-04) (hash(x)=155579314)
46
+ 360 train 5.852476 (lr=5.7760e-04) (hash(x)=148496629)
47
+ 370 train 5.948126 (lr=5.9360e-04) (hash(x)=156763563)
48
+ 380 train 5.805692 (lr=6.0960e-04) (hash(x)=160905666)
49
+ 390 train 6.205638 (lr=6.2560e-04) (hash(x)=157064388)
50
+ 400 val loss 5.9447
51
+ 400 val perplexity 381.7289
52
+ 400 train 6.228009 (lr=6.4160e-04) (hash(x)=155747374)
53
+ 410 train 5.892001 (lr=6.5760e-04) (hash(x)=147294192)
54
+ 420 train 5.913270 (lr=6.7360e-04) (hash(x)=144952473)
55
+ 430 train 5.858772 (lr=6.8960e-04) (hash(x)=160908725)
56
+ 440 train 5.924198 (lr=7.0560e-04) (hash(x)=145105827)
57
+ 450 train 5.987574 (lr=7.2160e-04) (hash(x)=159334575)
58
+ 460 train 5.795367 (lr=7.3760e-04) (hash(x)=161983315)
59
+ 470 train 5.698085 (lr=7.5360e-04) (hash(x)=138900475)
60
+ 480 train 5.710552 (lr=7.6960e-04) (hash(x)=152981338)
61
+ 490 train 5.651839 (lr=7.8560e-04) (hash(x)=143353862)
62
+ 500 val loss 5.6901
63
+ 500 val perplexity 295.9158
64
+ 500 train 5.562667 (lr=8.0000e-04) (hash(x)=140604760)
65
+ 510 train 5.474851 (lr=8.0000e-04) (hash(x)=150184398)
66
+ 520 train 5.622362 (lr=7.9999e-04) (hash(x)=141863041)
67
+ 530 train 5.615438 (lr=7.9998e-04) (hash(x)=153148776)
68
+ 540 train 5.522416 (lr=7.9996e-04) (hash(x)=144196873)
69
+ 550 train 5.535979 (lr=7.9993e-04) (hash(x)=146208052)
70
+ 560 train 5.472533 (lr=7.9991e-04) (hash(x)=139601869)
71
+ 570 train 5.604506 (lr=7.9987e-04) (hash(x)=150976466)
72
+ 580 train 5.538633 (lr=7.9983e-04) (hash(x)=150801563)
73
+ 590 train 5.681022 (lr=7.9979e-04) (hash(x)=167531956)
74
+ 600 val loss 5.5324
75
+ 600 val perplexity 252.7431
76
+ 600 train 5.427825 (lr=7.9974e-04) (hash(x)=148404734)
77
+ 610 train 5.370123 (lr=7.9968e-04) (hash(x)=146223362)
78
+ 620 train 5.349772 (lr=7.9962e-04) (hash(x)=147269386)
79
+ 630 train 5.293811 (lr=7.9956e-04) (hash(x)=149819251)
80
+ 640 train 5.606959 (lr=7.9949e-04) (hash(x)=146771854)
81
+ 650 train 5.535866 (lr=7.9941e-04) (hash(x)=155681970)
82
+ 660 train 5.255827 (lr=7.9933e-04) (hash(x)=138901204)
83
+ 670 train 5.089808 (lr=7.9925e-04) (hash(x)=143417964)
84
+ 680 train 5.251420 (lr=7.9915e-04) (hash(x)=141188344)
85
+ 690 train 5.359797 (lr=7.9906e-04) (hash(x)=148101044)
86
+ 700 val loss 5.3728
87
+ 700 val perplexity 215.4711
88
+ 700 train 5.378919 (lr=7.9896e-04) (hash(x)=148115934)
89
+ 710 train 5.431343 (lr=7.9885e-04) (hash(x)=142019907)
90
+ 720 train 5.346017 (lr=7.9874e-04) (hash(x)=144138833)
91
+ 730 train 5.417425 (lr=7.9862e-04) (hash(x)=150694253)
92
+ 740 train 5.236979 (lr=7.9850e-04) (hash(x)=157694921)
93
+ 750 train 5.330241 (lr=7.9837e-04) (hash(x)=157074034)
94
+ 760 train 5.257606 (lr=7.9824e-04) (hash(x)=157110692)
95
+ 770 train 5.183123 (lr=7.9810e-04) (hash(x)=145925624)
96
+ 780 train 5.286811 (lr=7.9796e-04) (hash(x)=156687510)
97
+ 790 train 5.253814 (lr=7.9781e-04) (hash(x)=150458321)
98
+ 800 val loss 5.2205
99
+ 800 val perplexity 185.0344
100
+ 800 train 5.134686 (lr=7.9765e-04) (hash(x)=137464699)
101
+ 810 train 5.145660 (lr=7.9749e-04) (hash(x)=158371555)
102
+ 820 train 5.147996 (lr=7.9733e-04) (hash(x)=149201877)
103
+ 830 train 5.094794 (lr=7.9716e-04) (hash(x)=143564160)
104
+ 840 train 5.219810 (lr=7.9699e-04) (hash(x)=153937595)
105
+ 850 train 5.127324 (lr=7.9681e-04) (hash(x)=166955614)
106
+ 860 train 5.125584 (lr=7.9662e-04) (hash(x)=164264614)
107
+ 870 train 4.972665 (lr=7.9643e-04) (hash(x)=139219447)
108
+ 880 train 5.168595 (lr=7.9624e-04) (hash(x)=151139580)
109
+ 890 train 5.044646 (lr=7.9604e-04) (hash(x)=147154297)
110
+ 900 val loss 5.1144
111
+ 900 val perplexity 166.4025
112
+ 900 train 4.910213 (lr=7.9583e-04) (hash(x)=143886042)
113
+ 910 train 4.866188 (lr=7.9562e-04) (hash(x)=144136925)
114
+ 920 train 5.025196 (lr=7.9541e-04) (hash(x)=139764865)
115
+ 930 train 4.995546 (lr=7.9518e-04) (hash(x)=137258208)
116
+ 940 train 5.075413 (lr=7.9496e-04) (hash(x)=151881247)
117
+ 950 train 5.004519 (lr=7.9473e-04) (hash(x)=193610391)
118
+ 960 train 4.967785 (lr=7.9449e-04) (hash(x)=149364435)
119
+ 970 train 4.982196 (lr=7.9425e-04) (hash(x)=148828909)
120
+ 980 train 4.762750 (lr=7.9400e-04) (hash(x)=143605331)
121
+ 990 train 4.777688 (lr=7.9375e-04) (hash(x)=153415979)
122
+ 1000 val loss 5.0116
123
+ 1000 val perplexity 150.1443
124
+ 1000 train 5.060415 (lr=7.9349e-04) (hash(x)=163799796)
125
+ 1010 train 5.035029 (lr=7.9323e-04) (hash(x)=145937873)
126
+ 1020 train 4.887575 (lr=7.9297e-04) (hash(x)=165459184)
127
+ 1030 train 5.070283 (lr=7.9269e-04) (hash(x)=150932001)
128
+ 1040 train 5.039569 (lr=7.9242e-04) (hash(x)=162567106)
129
+ 1050 train 5.051994 (lr=7.9213e-04) (hash(x)=154107339)
130
+ 1060 train 5.010264 (lr=7.9185e-04) (hash(x)=149509318)
131
+ 1070 train 4.826610 (lr=7.9155e-04) (hash(x)=152383589)
132
+ 1080 train 4.869904 (lr=7.9126e-04) (hash(x)=155926898)
133
+ 1090 train 4.884921 (lr=7.9095e-04) (hash(x)=150988851)
134
+ 1100 val loss 4.9136
135
+ 1100 val perplexity 136.1261
136
+ 1100 train 4.978097 (lr=7.9064e-04) (hash(x)=144592844)
137
+ 1110 train 4.926829 (lr=7.9033e-04) (hash(x)=146758273)
138
+ 1120 train 4.802599 (lr=7.9001e-04) (hash(x)=146142884)
139
+ 1130 train 4.919405 (lr=7.8969e-04) (hash(x)=152078362)
140
+ 1140 train 4.978499 (lr=7.8936e-04) (hash(x)=160186914)
141
+ 1150 train 4.917404 (lr=7.8903e-04) (hash(x)=154535861)
142
+ 1160 train 4.768675 (lr=7.8869e-04) (hash(x)=151399108)
143
+ 1170 train 4.848087 (lr=7.8835e-04) (hash(x)=146833517)
144
+ 1180 train 4.761436 (lr=7.8800e-04) (hash(x)=140850009)
145
+ 1190 train 4.713735 (lr=7.8764e-04) (hash(x)=160947364)
146
+ 1200 val loss 4.8067
147
+ 1200 val perplexity 122.3264
148
+ 1200 train 4.711884 (lr=7.8729e-04) (hash(x)=204706354)
149
+ 1210 train 4.627967 (lr=7.8692e-04) (hash(x)=145764356)
150
+ 1220 train 4.672602 (lr=7.8655e-04) (hash(x)=153116691)
151
+ 1230 train 4.695580 (lr=7.8618e-04) (hash(x)=160544763)
152
+ 1240 train 4.613328 (lr=7.8580e-04) (hash(x)=137389978)
153
+ 1250 train 4.691466 (lr=7.8542e-04) (hash(x)=143783202)
154
+ 1260 train 4.480141 (lr=7.8503e-04) (hash(x)=153470243)
155
+ 1270 train 4.639441 (lr=7.8464e-04) (hash(x)=149055354)
156
+ 1280 train 4.618998 (lr=7.8424e-04) (hash(x)=146874658)
157
+ 1290 train 4.811593 (lr=7.8383e-04) (hash(x)=152527777)
158
+ 1300 val loss 4.7526
159
+ 1300 val perplexity 115.8848
160
+ 1300 train 4.706355 (lr=7.8342e-04) (hash(x)=150862210)
161
+ 1310 train 4.709020 (lr=7.8301e-04) (hash(x)=157185054)
162
+ 1320 train 4.765158 (lr=7.8259e-04) (hash(x)=153733727)
163
+ 1330 train 4.752318 (lr=7.8217e-04) (hash(x)=163696656)
164
+ 1340 train 4.908749 (lr=7.8174e-04) (hash(x)=156931875)
165
+ 1350 train 4.676847 (lr=7.8131e-04) (hash(x)=150753426)
166
+ 1360 train 4.692238 (lr=7.8087e-04) (hash(x)=145642367)
167
+ 1370 train 4.713358 (lr=7.8042e-04) (hash(x)=159217491)
168
+ 1380 train 4.700126 (lr=7.7998e-04) (hash(x)=151499952)
169
+ 1390 train 4.808195 (lr=7.7952e-04) (hash(x)=147137197)
170
+ 1400 val loss 4.7228
171
+ 1400 val perplexity 112.4883
172
+ 1400 train 4.760168 (lr=7.7906e-04) (hash(x)=147766811)
173
+ 1410 train 4.573678 (lr=7.7860e-04) (hash(x)=155728979)
174
+ 1420 train 4.537860 (lr=7.7813e-04) (hash(x)=151858238)
175
+ 1430 train 4.608024 (lr=7.7766e-04) (hash(x)=147007501)
176
+ 1440 train 4.590528 (lr=7.7718e-04) (hash(x)=123449142)
177
+ 1450 train 4.585933 (lr=7.7670e-04) (hash(x)=151177814)
178
+ 1460 train 4.721348 (lr=7.7621e-04) (hash(x)=151804649)
179
+ 1470 train 4.682878 (lr=7.7572e-04) (hash(x)=140614506)
180
+ 1480 train 4.869624 (lr=7.7522e-04) (hash(x)=154590935)
181
+ 1490 train 4.659424 (lr=7.7472e-04) (hash(x)=155969672)
182
+ 1500 val loss 4.6581
183
+ 1500 val perplexity 105.4323
184
+ 1500 train 4.585139 (lr=7.7421e-04) (hash(x)=135925327)
185
+ 1510 train 4.574001 (lr=7.7370e-04) (hash(x)=140488340)
186
+ 1520 train 4.568420 (lr=7.7318e-04) (hash(x)=150707923)
187
+ 1530 train 4.488186 (lr=7.7266e-04) (hash(x)=141081542)
188
+ 1540 train 4.555531 (lr=7.7214e-04) (hash(x)=150136050)
189
+ 1550 train 4.403236 (lr=7.7160e-04) (hash(x)=124842353)
190
+ 1560 train 4.540962 (lr=7.7107e-04) (hash(x)=142320279)
191
+ 1570 train 4.495803 (lr=7.7053e-04) (hash(x)=137593322)
192
+ 1580 train 4.550569 (lr=7.6998e-04) (hash(x)=154607301)
193
+ 1590 train 4.424615 (lr=7.6943e-04) (hash(x)=143089856)
194
+ 1600 val loss 4.6612
195
+ 1600 val perplexity 105.7673
196
+ 1600 train 4.648029 (lr=7.6888e-04) (hash(x)=160440642)
197
+ 1610 train 4.864368 (lr=7.6832e-04) (hash(x)=150875971)
198
+ 1620 train 4.725889 (lr=7.6775e-04) (hash(x)=165443918)
199
+ 1630 train 4.541489 (lr=7.6718e-04) (hash(x)=145853246)
200
+ 1640 train 4.637426 (lr=7.6661e-04) (hash(x)=120920586)
201
+ 1650 train 4.564992 (lr=7.6603e-04) (hash(x)=146732412)
202
+ 1660 train 4.624346 (lr=7.6545e-04) (hash(x)=148900413)
203
+ 1670 train 4.857644 (lr=7.6486e-04) (hash(x)=147090451)
204
+ 1680 train 4.640466 (lr=7.6426e-04) (hash(x)=153099341)
205
+ 1690 train 4.667795 (lr=7.6367e-04) (hash(x)=148556821)
206
+ 1700 val loss 4.5850
207
+ 1700 val perplexity 98.0010
208
+ 1700 train 4.701464 (lr=7.6306e-04) (hash(x)=151184106)
209
+ 1710 train 4.609515 (lr=7.6246e-04) (hash(x)=151304861)
210
+ 1720 train 4.416762 (lr=7.6184e-04) (hash(x)=147047589)
211
+ 1730 train 4.515694 (lr=7.6123e-04) (hash(x)=157613602)
212
+ 1740 train 4.441167 (lr=7.6061e-04) (hash(x)=145631923)
213
+ 1750 train 4.400390 (lr=7.5998e-04) (hash(x)=153416694)
214
+ 1760 train 4.589398 (lr=7.5935e-04) (hash(x)=143038500)
215
+ 1770 train 4.543981 (lr=7.5872e-04) (hash(x)=163545780)
216
+ 1780 train 4.472878 (lr=7.5808e-04) (hash(x)=159712197)
217
+ 1790 train 4.384040 (lr=7.5743e-04) (hash(x)=146293430)
218
+ 1800 val loss 4.5520
219
+ 1800 val perplexity 94.8182
220
+ 1800 train 4.517808 (lr=7.5678e-04) (hash(x)=148100580)
221
+ 1810 train 4.484467 (lr=7.5613e-04) (hash(x)=137858614)
222
+ 1820 train 4.505059 (lr=7.5547e-04) (hash(x)=155765218)
223
+ 1830 train 4.425115 (lr=7.5481e-04) (hash(x)=147554787)
224
+ 1840 train 4.506282 (lr=7.5414e-04) (hash(x)=160001341)
225
+ 1850 train 4.386587 (lr=7.5347e-04) (hash(x)=153059873)
226
+ 1860 train 4.337109 (lr=7.5279e-04) (hash(x)=155656532)
227
+ 1870 train 4.400809 (lr=7.5211e-04) (hash(x)=146515528)
228
+ 1880 train 4.272449 (lr=7.5143e-04) (hash(x)=144385809)
229
+ 1890 train 4.317949 (lr=7.5074e-04) (hash(x)=150375274)
230
+ 1900 val loss 4.5734
231
+ 1900 val perplexity 96.8743
232
+ 1900 train 4.320110 (lr=7.5004e-04) (hash(x)=149434659)
233
+ 1910 train 4.413660 (lr=7.4934e-04) (hash(x)=140811482)
234
+ 1920 train 4.489527 (lr=7.4864e-04) (hash(x)=149776450)
235
+ 1930 train 4.477855 (lr=7.4793e-04) (hash(x)=153611546)
236
+ 1940 train 4.561223 (lr=7.4722e-04) (hash(x)=152628274)
237
+ 1950 train 4.485078 (lr=7.4650e-04) (hash(x)=151183924)
238
+ 1960 train 4.577679 (lr=7.4578e-04) (hash(x)=143507386)
239
+ 1970 train 4.607583 (lr=7.4505e-04) (hash(x)=180301574)
240
+ 1980 train 4.518426 (lr=7.4432e-04) (hash(x)=156383157)
241
+ 1990 train 4.561419 (lr=7.4359e-04) (hash(x)=145911523)
242
+ 2000 val loss 4.4962
243
+ 2000 val perplexity 89.6785
244
+ 2000 train 4.553709 (lr=7.4285e-04) (hash(x)=152285486)
245
+ 2010 train 4.467383 (lr=7.4211e-04) (hash(x)=154531062)
246
+ 2020 train 4.432123 (lr=7.4136e-04) (hash(x)=154403800)
247
+ 2030 train 4.517902 (lr=7.4061e-04) (hash(x)=153109813)
248
+ 2040 train 4.578059 (lr=7.3985e-04) (hash(x)=149156251)
249
+ 2050 train 4.378663 (lr=7.3909e-04) (hash(x)=147613800)
250
+ 2060 train 4.520594 (lr=7.3833e-04) (hash(x)=160283013)
251
+ 2070 train 4.481216 (lr=7.3756e-04) (hash(x)=141632717)
252
+ 2080 train 4.434110 (lr=7.3678e-04) (hash(x)=147292755)
253
+ 2090 train 4.607162 (lr=7.3600e-04) (hash(x)=146447537)
254
+ 2100 val loss 4.4652
255
+ 2100 val perplexity 86.9349
256
+ 2100 train 4.238595 (lr=7.3522e-04) (hash(x)=144294295)
257
+ 2110 train 4.789235 (lr=7.3444e-04) (hash(x)=142838110)
258
+ 2120 train 4.743314 (lr=7.3364e-04) (hash(x)=151878701)
259
+ 2130 train 4.333349 (lr=7.3285e-04) (hash(x)=143929160)
260
+ 2140 train 4.312211 (lr=7.3205e-04) (hash(x)=149368632)
261
+ 2150 train 4.386852 (lr=7.3125e-04) (hash(x)=141460003)
262
+ 2160 train 4.354455 (lr=7.3044e-04) (hash(x)=140793655)
263
+ 2170 train 4.500341 (lr=7.2963e-04) (hash(x)=146336171)
264
+ 2180 train 4.322476 (lr=7.2881e-04) (hash(x)=150200011)
265
+ 2190 train 4.369924 (lr=7.2799e-04) (hash(x)=155337561)
266
+ 2200 val loss 4.4795
267
+ 2200 val perplexity 88.1947
268
+ 2200 train 4.381548 (lr=7.2716e-04) (hash(x)=175030215)
269
+ 2210 train 4.528163 (lr=7.2634e-04) (hash(x)=139653118)
270
+ 2220 train 4.392248 (lr=7.2550e-04) (hash(x)=151690464)
271
+ 2230 train 4.130691 (lr=7.2467e-04) (hash(x)=164526288)
272
+ 2240 train 4.245984 (lr=7.2382e-04) (hash(x)=141536420)
273
+ 2250 train 4.451756 (lr=7.2298e-04) (hash(x)=167362519)
274
+ 2260 train 4.205047 (lr=7.2213e-04) (hash(x)=142323410)
275
+ 2270 train 4.618705 (lr=7.2128e-04) (hash(x)=176298114)
276
+ 2280 train 4.309641 (lr=7.2042e-04) (hash(x)=149454362)
277
+ 2290 train 4.418263 (lr=7.1956e-04) (hash(x)=172920991)
278
+ 2300 val loss 4.4613
279
+ 2300 val perplexity 86.5992
280
+ 2300 train 4.398343 (lr=7.1869e-04) (hash(x)=150831428)
281
+ 2310 train 4.264321 (lr=7.1782e-04) (hash(x)=141773102)
282
+ 2320 train 4.205518 (lr=7.1695e-04) (hash(x)=146520072)
283
+ 2330 train 4.419141 (lr=7.1607e-04) (hash(x)=174782103)
284
+ 2340 train 4.517644 (lr=7.1519e-04) (hash(x)=171102155)
285
+ 2350 train 4.354145 (lr=7.1430e-04) (hash(x)=152910404)
286
+ 2360 train 4.351776 (lr=7.1341e-04) (hash(x)=147122637)
287
+ 2370 train 4.278933 (lr=7.1252e-04) (hash(x)=152558061)
288
+ 2380 train 4.356750 (lr=7.1162e-04) (hash(x)=146957950)
289
+ 2390 train 4.399635 (lr=7.1072e-04) (hash(x)=156879460)
290
+ 2400 val loss 4.4108
291
+ 2400 val perplexity 82.3393
292
+ 2400 train 4.560460 (lr=7.0981e-04) (hash(x)=140808297)
293
+ 2410 train 4.286865 (lr=7.0890e-04) (hash(x)=164207329)
294
+ 2420 train 4.485755 (lr=7.0799e-04) (hash(x)=150119566)
295
+ 2430 train 4.341370 (lr=7.0707e-04) (hash(x)=148059202)
296
+ 2440 train 4.411613 (lr=7.0615e-04) (hash(x)=144913823)
297
+ 2450 train 4.573187 (lr=7.0523e-04) (hash(x)=155469321)
298
+ 2460 train 4.603648 (lr=7.0430e-04) (hash(x)=147971519)
299
+ 2470 train 4.318815 (lr=7.0337e-04) (hash(x)=151115240)
300
+ 2480 train 4.325778 (lr=7.0243e-04) (hash(x)=158904683)
301
+ 2490 train 4.327302 (lr=7.0149e-04) (hash(x)=153639698)
302
+ 2500 val loss 4.4121
303
+ 2500 val perplexity 82.4450
304
+ 2500 train 4.365186 (lr=7.0054e-04) (hash(x)=153160275)
305
+ 2510 train 4.189418 (lr=6.9960e-04) (hash(x)=166439539)
306
+ 2520 train 4.450213 (lr=6.9864e-04) (hash(x)=141364112)
307
+ 2530 train 4.410044 (lr=6.9769e-04) (hash(x)=147911679)
308
+ 2540 train 4.358334 (lr=6.9673e-04) (hash(x)=152607847)
309
+ 2550 train 4.298596 (lr=6.9577e-04) (hash(x)=149178874)
310
+ 2560 train 4.448351 (lr=6.9480e-04) (hash(x)=158403024)
311
+ 2570 train 4.315617 (lr=6.9383e-04) (hash(x)=148252901)
312
+ 2580 train 4.345636 (lr=6.9286e-04) (hash(x)=156933994)
313
+ 2590 train 4.293121 (lr=6.9188e-04) (hash(x)=147869995)
314
+ 2600 val loss 4.3973
315
+ 2600 val perplexity 81.2273
316
+ 2600 train 4.302682 (lr=6.9090e-04) (hash(x)=133990623)
317
+ 2610 train 4.384042 (lr=6.8991e-04) (hash(x)=155745526)
318
+ 2620 train 4.285774 (lr=6.8892e-04) (hash(x)=145640933)
319
+ 2630 train 4.363212 (lr=6.8793e-04) (hash(x)=144600130)
320
+ 2640 train 4.188065 (lr=6.8694e-04) (hash(x)=143319960)
321
+ 2650 train 4.133358 (lr=6.8594e-04) (hash(x)=144841781)
322
+ 2660 train 4.390707 (lr=6.8493e-04) (hash(x)=142306571)
323
+ 2670 train 4.313295 (lr=6.8393e-04) (hash(x)=150824939)
324
+ 2680 train 4.360540 (lr=6.8292e-04) (hash(x)=144954685)
325
+ 2690 train 4.154921 (lr=6.8190e-04) (hash(x)=144071317)
326
+ 2700 val loss 4.4084
327
+ 2700 val perplexity 82.1403
328
+ 2700 train 4.283868 (lr=6.8089e-04) (hash(x)=142860944)
329
+ 2710 train 4.284176 (lr=6.7987e-04) (hash(x)=150879676)
330
+ 2720 train 4.158360 (lr=6.7884e-04) (hash(x)=158434126)
331
+ 2730 train 4.451266 (lr=6.7782e-04) (hash(x)=149786230)
332
+ 2740 train 4.454788 (lr=6.7678e-04) (hash(x)=148769334)
333
+ 2750 train 4.381790 (lr=6.7575e-04) (hash(x)=151565702)
334
+ 2760 train 4.521334 (lr=6.7471e-04) (hash(x)=162471158)
335
+ 2770 train 4.407189 (lr=6.7367e-04) (hash(x)=168950803)
336
+ 2780 train 4.475935 (lr=6.7263e-04) (hash(x)=149815020)
337
+ 2790 train 4.403244 (lr=6.7158e-04) (hash(x)=161235555)
338
+ 2800 val loss 4.3746
339
+ 2800 val perplexity 79.4119
340
+ 2800 train 4.302833 (lr=6.7053e-04) (hash(x)=137959511)
341
+ 2810 train 4.452413 (lr=6.6947e-04) (hash(x)=148420397)
342
+ 2820 train 4.350205 (lr=6.6841e-04) (hash(x)=151862744)
343
+ 2830 train 4.433136 (lr=6.6735e-04) (hash(x)=153223303)
344
+ 2840 train 4.380392 (lr=6.6629e-04) (hash(x)=149794760)
345
+ 2850 train 4.452172 (lr=6.6522e-04) (hash(x)=153517085)
346
+ 2860 train 4.434106 (lr=6.6415e-04) (hash(x)=152044927)
347
+ 2870 train 4.619443 (lr=6.6308e-04) (hash(x)=139708749)
348
+ 2880 train 4.350974 (lr=6.6200e-04) (hash(x)=152675893)
349
+ 2890 train 4.377901 (lr=6.6092e-04) (hash(x)=153950772)
350
+ 2900 val loss 4.3427
351
+ 2900 val perplexity 76.9125
352
+ 2900 train 4.306933 (lr=6.5983e-04) (hash(x)=147009873)
353
+ 2910 train 4.401606 (lr=6.5875e-04) (hash(x)=147497701)
354
+ 2920 train 4.495472 (lr=6.5766e-04) (hash(x)=149615491)
355
+ 2930 train 4.517495 (lr=6.5656e-04) (hash(x)=152588554)
356
+ 2940 train 4.331928 (lr=6.5547e-04) (hash(x)=147080266)
357
+ 2950 train 4.567911 (lr=6.5437e-04) (hash(x)=149607682)
358
+ 2960 train 4.283422 (lr=6.5326e-04) (hash(x)=156855690)
359
+ 2970 train 4.283480 (lr=6.5216e-04) (hash(x)=147759452)
360
+ 2980 train 4.194015 (lr=6.5105e-04) (hash(x)=168906251)
361
+ 2990 train 4.346195 (lr=6.4994e-04) (hash(x)=159612982)
362
+ 3000 val loss 4.3432
363
+ 3000 val perplexity 76.9498
364
+ 3000 train 4.195820 (lr=6.4882e-04) (hash(x)=158264841)
365
+ 3010 train 4.205524 (lr=6.4770e-04) (hash(x)=151971710)
366
+ 3020 train 4.273114 (lr=6.4658e-04) (hash(x)=148961929)
367
+ 3030 train 4.066615 (lr=6.4546e-04) (hash(x)=144951447)
368
+ 3040 train 4.122841 (lr=6.4433e-04) (hash(x)=137727737)
369
+ 3050 train 4.144409 (lr=6.4320e-04) (hash(x)=145598216)
370
+ 3060 train 4.142012 (lr=6.4207e-04) (hash(x)=139250122)
371
+ 3070 train 4.203722 (lr=6.4093e-04) (hash(x)=158020889)
372
+ 3080 train 4.037512 (lr=6.3979e-04) (hash(x)=147345272)
373
+ 3090 train 4.067163 (lr=6.3865e-04) (hash(x)=148006111)
374
+ 3100 val loss 4.3466
375
+ 3100 val perplexity 77.2159
376
+ 3100 train 4.199658 (lr=6.3750e-04) (hash(x)=139232251)
377
+ 3110 train 4.018833 (lr=6.3636e-04) (hash(x)=164654684)
378
+ 3120 train 4.255218 (lr=6.3521e-04) (hash(x)=153975753)
379
+ 3130 train 4.395474 (lr=6.3405e-04) (hash(x)=154784772)
380
+ 3140 train 4.260893 (lr=6.3290e-04) (hash(x)=156671619)
381
+ 3150 train 4.244913 (lr=6.3174e-04) (hash(x)=158188012)
382
+ 3160 train 4.339773 (lr=6.3058e-04) (hash(x)=152266024)
383
+ 3170 train 4.322195 (lr=6.2941e-04) (hash(x)=145418617)
384
+ 3180 train 4.359796 (lr=6.2825e-04) (hash(x)=149947903)
385
+ 3190 train 4.239587 (lr=6.2708e-04) (hash(x)=156664225)
386
+ 3200 val loss 4.3147
387
+ 3200 val perplexity 74.7906
388
+ 3200 train 4.362389 (lr=6.2590e-04) (hash(x)=153436104)
389
+ 3210 train 4.434961 (lr=6.2473e-04) (hash(x)=163550923)
390
+ 3220 train 4.198683 (lr=6.2355e-04) (hash(x)=144583183)
391
+ 3230 train 4.268973 (lr=6.2237e-04) (hash(x)=165547680)
392
+ 3240 train 4.570931 (lr=6.2119e-04) (hash(x)=152195267)
393
+ 3250 train 4.338094 (lr=6.2000e-04) (hash(x)=150040234)
394
+ 3260 train 4.328497 (lr=6.1881e-04) (hash(x)=143788229)
395
+ 3270 train 4.183576 (lr=6.1762e-04) (hash(x)=154472495)
396
+ 3280 train 4.188994 (lr=6.1643e-04) (hash(x)=162620947)
397
+ 3290 train 4.312498 (lr=6.1523e-04) (hash(x)=157731379)
398
+ 3300 val loss 4.3095
399
+ 3300 val perplexity 74.4046
400
+ 3300 train 4.170506 (lr=6.1403e-04) (hash(x)=149681831)
401
+ 3310 train 4.240686 (lr=6.1283e-04) (hash(x)=147216223)
402
+ 3320 train 4.168376 (lr=6.1163e-04) (hash(x)=146478959)
403
+ 3330 train 4.274387 (lr=6.1042e-04) (hash(x)=161840239)
404
+ 3340 train 4.257120 (lr=6.0921e-04) (hash(x)=141748686)
405
+ 3350 train 4.076930 (lr=6.0800e-04) (hash(x)=134782096)
406
+ 3360 train 4.278628 (lr=6.0679e-04) (hash(x)=164103503)
407
+ 3370 train 4.132045 (lr=6.0557e-04) (hash(x)=144658716)
408
+ 3380 train 4.058353 (lr=6.0435e-04) (hash(x)=144343513)
409
+ 3390 train 4.143628 (lr=6.0313e-04) (hash(x)=145365816)
410
+ 3400 val loss 4.3108
411
+ 3400 val perplexity 74.4993
412
+ 3400 train 4.160998 (lr=6.0191e-04) (hash(x)=168425516)
413
+ 3410 train 4.109692 (lr=6.0068e-04) (hash(x)=139704073)
414
+ 3420 train 4.229676 (lr=5.9945e-04) (hash(x)=156592513)
415
+ 3430 train 4.225341 (lr=5.9822e-04) (hash(x)=136773063)
416
+ 3440 train 4.230970 (lr=5.9699e-04) (hash(x)=162468494)
417
+ 3450 train 4.288318 (lr=5.9576e-04) (hash(x)=143829320)
418
+ 3460 train 4.343143 (lr=5.9452e-04) (hash(x)=158562331)
419
+ 3470 train 4.335339 (lr=5.9328e-04) (hash(x)=151882487)
420
+ 3480 train 4.325079 (lr=5.9204e-04) (hash(x)=150575558)
421
+ 3490 train 4.213542 (lr=5.9080e-04) (hash(x)=147407741)
422
+ 3500 val loss 4.2707
423
+ 3500 val perplexity 71.5683
424
+ 3500 train 4.317217 (lr=5.8955e-04) (hash(x)=163104338)
425
+ 3510 train 4.246094 (lr=5.8830e-04) (hash(x)=157461421)
426
+ 3520 train 4.202423 (lr=5.8705e-04) (hash(x)=151459070)
427
+ 3530 train 4.148893 (lr=5.8580e-04) (hash(x)=150777169)
428
+ 3540 train 4.290895 (lr=5.8454e-04) (hash(x)=159138904)
429
+ 3550 train 4.207471 (lr=5.8329e-04) (hash(x)=157853732)
430
+ 3560 train 4.177369 (lr=5.8203e-04) (hash(x)=152552809)
431
+ 3570 train 4.232069 (lr=5.8077e-04) (hash(x)=145219770)
432
+ 3580 train 4.176418 (lr=5.7951e-04) (hash(x)=147325185)
433
+ 3590 train 4.271831 (lr=5.7824e-04) (hash(x)=146746257)
434
+ 3600 val loss 4.2682
435
+ 3600 val perplexity 71.3918
436
+ 3600 train 4.125894 (lr=5.7697e-04) (hash(x)=165109772)
437
+ 3610 train 4.077296 (lr=5.7571e-04) (hash(x)=142561694)
438
+ 3620 train 4.230652 (lr=5.7443e-04) (hash(x)=153512606)
439
+ 3630 train 4.079939 (lr=5.7316e-04) (hash(x)=140485474)
440
+ 3640 train 4.207884 (lr=5.7189e-04) (hash(x)=142084056)
441
+ 3650 train 4.243262 (lr=5.7061e-04) (hash(x)=143729044)
442
+ 3660 train 4.149714 (lr=5.6933e-04) (hash(x)=129606055)
443
+ 3670 train 4.183328 (lr=5.6805e-04) (hash(x)=149878209)
444
+ 3680 train 4.011010 (lr=5.6677e-04) (hash(x)=147988422)
445
+ 3690 train 4.108144 (lr=5.6549e-04) (hash(x)=134245780)
446
+ 3700 val loss 4.2891
447
+ 3700 val perplexity 72.9026
448
+ 3700 train 4.022334 (lr=5.6420e-04) (hash(x)=153420306)
449
+ 3710 train 4.198476 (lr=5.6291e-04) (hash(x)=151618359)
450
+ 3720 train 4.093060 (lr=5.6162e-04) (hash(x)=146201479)
451
+ 3730 train 4.142612 (lr=5.6033e-04) (hash(x)=173571327)
452
+ 3740 train 4.270527 (lr=5.5904e-04) (hash(x)=141793500)
453
+ 3750 train 4.361817 (lr=5.5774e-04) (hash(x)=184628568)
454
+ 3760 train 4.318676 (lr=5.5645e-04) (hash(x)=143308741)
455
+ 3770 train 4.455672 (lr=5.5515e-04) (hash(x)=151283512)
456
+ 3780 train 4.257936 (lr=5.5385e-04) (hash(x)=143336516)
457
+ 3790 train 4.272225 (lr=5.5255e-04) (hash(x)=148124486)
458
+ 3800 val loss 4.2497
459
+ 3800 val perplexity 70.0819
460
+ 3800 train 4.154808 (lr=5.5125e-04) (hash(x)=160168863)
461
+ 3810 train 4.186069 (lr=5.4994e-04) (hash(x)=155072251)
462
+ 3820 train 4.376506 (lr=5.4864e-04) (hash(x)=153837417)
463
+ 3830 train 4.444569 (lr=5.4733e-04) (hash(x)=130032011)
464
+ 3840 train 4.141962 (lr=5.4602e-04) (hash(x)=146270962)
465
+ 3850 train 4.158966 (lr=5.4471e-04) (hash(x)=146936934)
466
+ 3860 train 4.087778 (lr=5.4340e-04) (hash(x)=165785187)
467
+ 3870 train 4.289525 (lr=5.4208e-04) (hash(x)=145393482)
468
+ 3880 train 4.227941 (lr=5.4077e-04) (hash(x)=148819533)
469
+ 3890 train 4.098371 (lr=5.3945e-04) (hash(x)=146870411)
470
+ 3900 val loss 4.2445
471
+ 3900 val perplexity 69.7190
472
+ 3900 train 4.108290 (lr=5.3813e-04) (hash(x)=153906073)
473
+ 3910 train 4.213614 (lr=5.3681e-04) (hash(x)=135632500)
474
+ 3920 train 4.233675 (lr=5.3549e-04) (hash(x)=147393037)
475
+ 3930 train 4.386272 (lr=5.3417e-04) (hash(x)=150847811)
476
+ 3940 train 4.157756 (lr=5.3284e-04) (hash(x)=150253123)
477
+ 3950 train 4.036664 (lr=5.3152e-04) (hash(x)=162609448)
478
+ 3960 train 4.048598 (lr=5.3019e-04) (hash(x)=150901940)
479
+ 3970 train 4.079626 (lr=5.2886e-04) (hash(x)=149698197)
480
+ 3980 train 4.177452 (lr=5.2754e-04) (hash(x)=156055141)
481
+ 3990 train 4.142100 (lr=5.2620e-04) (hash(x)=163154074)
482
+ 4000 val loss 4.2459
483
+ 4000 val perplexity 69.8205
484
+ 4000 train 4.206126 (lr=5.2487e-04) (hash(x)=151055067)
485
+ 4010 train 4.137608 (lr=5.2354e-04) (hash(x)=149313232)
486
+ 4020 train 4.184261 (lr=5.2221e-04) (hash(x)=144350585)
487
+ 4030 train 4.188681 (lr=5.2087e-04) (hash(x)=143139724)
488
+ 4040 train 4.257005 (lr=5.1953e-04) (hash(x)=137534879)
489
+ 4050 train 4.137333 (lr=5.1820e-04) (hash(x)=147037092)
490
+ 4060 train 4.175642 (lr=5.1686e-04) (hash(x)=149915654)
491
+ 4070 train 4.115957 (lr=5.1552e-04) (hash(x)=148002990)
492
+ 4080 train 4.104807 (lr=5.1418e-04) (hash(x)=141860091)
493
+ 4090 train 4.253425 (lr=5.1284e-04) (hash(x)=152510717)
494
+ 4100 val loss 4.2308
495
+ 4100 val perplexity 68.7729
496
+ 4100 train 4.132714 (lr=5.1149e-04) (hash(x)=149629830)
497
+ 4110 train 4.121681 (lr=5.1015e-04) (hash(x)=143677674)
498
+ 4120 train 4.244156 (lr=5.0880e-04) (hash(x)=143870235)
499
+ 4130 train 4.210710 (lr=5.0746e-04) (hash(x)=156172240)
500
+ 4140 train 4.073153 (lr=5.0611e-04) (hash(x)=142768662)
501
+ 4150 train 4.115337 (lr=5.0476e-04) (hash(x)=151256841)
502
+ 4160 train 4.203154 (lr=5.0341e-04) (hash(x)=144941747)
503
+ 4170 train 4.104798 (lr=5.0206e-04) (hash(x)=143588738)
504
+ 4180 train 4.065665 (lr=5.0071e-04) (hash(x)=145117633)
505
+ 4190 train 4.233299 (lr=4.9936e-04) (hash(x)=158244222)
506
+ 4200 val loss 4.2212
507
+ 4200 val perplexity 68.1160
508
+ 4200 train 4.126451 (lr=4.9801e-04) (hash(x)=143101381)
509
+ 4210 train 4.141376 (lr=4.9665e-04) (hash(x)=148704069)
510
+ 4220 train 4.063908 (lr=4.9530e-04) (hash(x)=165334845)
511
+ 4230 train 4.101263 (lr=4.9395e-04) (hash(x)=147150527)
512
+ 4240 train 4.094983 (lr=4.9259e-04) (hash(x)=150118606)
513
+ 4250 train 4.064862 (lr=4.9123e-04) (hash(x)=142159597)
514
+ 4260 train 4.094342 (lr=4.8988e-04) (hash(x)=152185092)
515
+ 4270 train 4.109426 (lr=4.8852e-04) (hash(x)=148870173)
516
+ 4280 train 4.122981 (lr=4.8716e-04) (hash(x)=150977664)
517
+ 4290 train 4.201859 (lr=4.8580e-04) (hash(x)=165032394)
518
+ 4300 val loss 4.2210
519
+ 4300 val perplexity 68.1038
520
+ 4300 train 4.206903 (lr=4.8444e-04) (hash(x)=149712044)
521
+ 4310 train 4.131742 (lr=4.8308e-04) (hash(x)=151956500)
522
+ 4320 train 4.312198 (lr=4.8172e-04) (hash(x)=136107484)
523
+ 4330 train 4.386028 (lr=4.8036e-04) (hash(x)=160783980)
524
+ 4340 train 4.272161 (lr=4.7899e-04) (hash(x)=151772228)
525
+ 4350 train 4.073689 (lr=4.7763e-04) (hash(x)=176468048)
526
+ 4360 train 4.162971 (lr=4.7627e-04) (hash(x)=144588632)
527
+ 4370 train 4.187658 (lr=4.7490e-04) (hash(x)=156374655)
528
+ 4380 train 4.291203 (lr=4.7354e-04) (hash(x)=148502191)
529
+ 4390 train 4.380150 (lr=4.7217e-04) (hash(x)=152391280)
530
+ 4400 val loss 4.2064
531
+ 4400 val perplexity 67.1118
532
+ 4400 train 4.109535 (lr=4.7081e-04) (hash(x)=153446449)
533
+ 4410 train 4.185249 (lr=4.6944e-04) (hash(x)=167668129)
534
+ 4420 train 4.124903 (lr=4.6807e-04) (hash(x)=148598803)
535
+ 4430 train 4.191690 (lr=4.6671e-04) (hash(x)=161720292)
536
+ 4440 train 4.031214 (lr=4.6534e-04) (hash(x)=144133776)
537
+ 4450 train 4.138827 (lr=4.6397e-04) (hash(x)=151057987)
538
+ 4460 train 4.161412 (lr=4.6260e-04) (hash(x)=151971972)
539
+ 4470 train 4.021616 (lr=4.6124e-04) (hash(x)=140885903)
540
+ 4480 train 4.050742 (lr=4.5987e-04) (hash(x)=138198700)
541
+ 4490 train 4.134315 (lr=4.5850e-04) (hash(x)=150424198)
542
+ 4500 val loss 4.1964
543
+ 4500 val perplexity 66.4463
544
+ 4500 train 4.143161 (lr=4.5713e-04) (hash(x)=146086947)
545
+ 4510 train 4.127955 (lr=4.5576e-04) (hash(x)=145752336)
546
+ 4520 train 4.079544 (lr=4.5439e-04) (hash(x)=145005726)
547
+ 4530 train 4.082895 (lr=4.5302e-04) (hash(x)=144116626)
548
+ 4540 train 4.132119 (lr=4.5165e-04) (hash(x)=164507978)
549
+ 4550 train 4.017431 (lr=4.5028e-04) (hash(x)=154512699)
550
+ 4560 train 4.117078 (lr=4.4891e-04) (hash(x)=149920787)
551
+ 4570 train 4.024415 (lr=4.4754e-04) (hash(x)=155720526)
552
+ 4580 train 4.089893 (lr=4.4617e-04) (hash(x)=149918053)
553
+ 4590 train 4.234351 (lr=4.4480e-04) (hash(x)=170567219)
554
+ 4600 val loss 4.1999
555
+ 4600 val perplexity 66.6819
556
+ 4600 train 4.092412 (lr=4.4343e-04) (hash(x)=153800173)
557
+ 4610 train 4.058306 (lr=4.4206e-04) (hash(x)=151762309)
558
+ 4620 train 4.141501 (lr=4.4069e-04) (hash(x)=138620238)
559
+ 4630 train 4.073752 (lr=4.3931e-04) (hash(x)=155847676)
560
+ 4640 train 4.205337 (lr=4.3794e-04) (hash(x)=159145173)
561
+ 4650 train 4.245983 (lr=4.3657e-04) (hash(x)=163662103)
562
+ 4660 train 4.132617 (lr=4.3520e-04) (hash(x)=168862572)
563
+ 4670 train 5.297637 (lr=4.3383e-04) (hash(x)=128395338)
564
+ 4680 train 4.042050 (lr=4.3246e-04) (hash(x)=149975049)
565
+ 4690 train 4.068012 (lr=4.3109e-04) (hash(x)=156604294)
566
+ 4700 val loss 4.1834
567
+ 4700 val perplexity 65.5861
568
+ 4700 train 4.347785 (lr=4.2972e-04) (hash(x)=155962726)
569
+ 4710 train 4.069106 (lr=4.2835e-04) (hash(x)=160931268)
570
+ 4720 train 4.150998 (lr=4.2698e-04) (hash(x)=173485074)
571
+ 4730 train 4.256828 (lr=4.2561e-04) (hash(x)=146001945)
572
+ 4740 train 4.180187 (lr=4.2424e-04) (hash(x)=148305446)
573
+ 4750 train 4.139693 (lr=4.2287e-04) (hash(x)=131655842)
574
+ 4760 train 4.076835 (lr=4.2150e-04) (hash(x)=145775165)
575
+ 4770 train 4.020259 (lr=4.2013e-04) (hash(x)=146613282)
576
+ 4780 train 4.031656 (lr=4.1876e-04) (hash(x)=137882160)
577
+ 4790 train 4.181270 (lr=4.1740e-04) (hash(x)=157177531)
578
+ 4800 val loss 4.1778
579
+ 4800 val perplexity 65.2206
580
+ 4800 train 4.502291 (lr=4.1603e-04) (hash(x)=142045616)
581
+ 4810 train 4.343427 (lr=4.1466e-04) (hash(x)=153739117)
582
+ 4820 train 3.982253 (lr=4.1329e-04) (hash(x)=159336444)
583
+ 4830 train 4.177516 (lr=4.1193e-04) (hash(x)=162617526)
584
+ 4840 train 4.217037 (lr=4.1056e-04) (hash(x)=141336820)
585
+ 4850 train 4.137818 (lr=4.0919e-04) (hash(x)=146718291)
586
+ 4860 train 4.098864 (lr=4.0783e-04) (hash(x)=159542679)
587
+ 4870 train 4.090757 (lr=4.0646e-04) (hash(x)=149318029)
588
+ 4880 train 4.090718 (lr=4.0510e-04) (hash(x)=150182605)
589
+ 4890 train 4.069015 (lr=4.0373e-04) (hash(x)=139657075)
590
+ 4900 val loss 4.1695
591
+ 4900 val perplexity 64.6848
592
+ 4900 train 4.196054 (lr=4.0237e-04) (hash(x)=143418248)
593
+ 4910 train 4.268766 (lr=4.0101e-04) (hash(x)=153395021)
594
+ 4920 train 4.171443 (lr=3.9964e-04) (hash(x)=144977141)
595
+ 4930 train 4.187535 (lr=3.9828e-04) (hash(x)=143569016)
596
+ 4940 train 4.137666 (lr=3.9692e-04) (hash(x)=145702219)
597
+ 4950 train 4.133379 (lr=3.9556e-04) (hash(x)=150669399)
598
+ 4960 train 4.130672 (lr=3.9420e-04) (hash(x)=147697301)
599
+ 4970 train 4.038513 (lr=3.9284e-04) (hash(x)=158617139)
600
+ 4980 train 4.153803 (lr=3.9148e-04) (hash(x)=158937457)
601
+ 4990 train 4.056543 (lr=3.9012e-04) (hash(x)=155540364)
602
+ 5000 val loss 4.1679
603
+ 5000 val perplexity 64.5809
604
+ 5000 train 4.109276 (lr=3.8877e-04) (hash(x)=145789790)
605
+ 5010 train 4.122644 (lr=3.8741e-04) (hash(x)=142919148)
606
+ 5020 train 4.007887 (lr=3.8605e-04) (hash(x)=149129606)
607
+ 5030 train 4.246146 (lr=3.8470e-04) (hash(x)=143060789)
608
+ 5040 train 4.078556 (lr=3.8335e-04) (hash(x)=145733956)
609
+ 5050 train 3.905066 (lr=3.8199e-04) (hash(x)=139750793)
610
+ 5060 train 4.178918 (lr=3.8064e-04) (hash(x)=148027808)
611
+ 5070 train 4.005291 (lr=3.7929e-04) (hash(x)=149119770)
612
+ 5080 train 4.041074 (lr=3.7794e-04) (hash(x)=145741944)
613
+ 5090 train 4.112407 (lr=3.7659e-04) (hash(x)=144783744)
614
+ 5100 val loss 4.1701
615
+ 5100 val perplexity 64.7211
616
+ 5100 train 4.131781 (lr=3.7524e-04) (hash(x)=137795633)
617
+ 5110 train 4.026496 (lr=3.7389e-04) (hash(x)=145995538)
618
+ 5120 train 4.034956 (lr=3.7254e-04) (hash(x)=143019933)
619
+ 5130 train 4.016734 (lr=3.7120e-04) (hash(x)=143746069)
620
+ 5140 train 3.813293 (lr=3.6985e-04) (hash(x)=152212469)
621
+ 5150 train 4.108913 (lr=3.6851e-04) (hash(x)=152608669)
622
+ 5160 train 4.145670 (lr=3.6716e-04) (hash(x)=150750381)
623
+ 5170 train 4.085071 (lr=3.6582e-04) (hash(x)=133198485)
624
+ 5180 train 4.150249 (lr=3.6448e-04) (hash(x)=142233420)
625
+ 5190 train 3.991901 (lr=3.6314e-04) (hash(x)=147366941)
626
+ 5200 val loss 4.1539
627
+ 5200 val perplexity 63.6787
628
+ 5200 train 4.237551 (lr=3.6180e-04) (hash(x)=148907132)
629
+ 5210 train 4.290800 (lr=3.6047e-04) (hash(x)=154827138)
630
+ 5220 train 4.279488 (lr=3.5913e-04) (hash(x)=148717408)
631
+ 5230 train 4.442407 (lr=3.5779e-04) (hash(x)=147711017)
632
+ 5240 train 4.105790 (lr=3.5646e-04) (hash(x)=141796353)
633
+ 5250 train 4.033304 (lr=3.5513e-04) (hash(x)=143529037)
634
+ 5260 train 4.032298 (lr=3.5380e-04) (hash(x)=144773706)
635
+ 5270 train 4.091296 (lr=3.5246e-04) (hash(x)=148894329)
636
+ 5280 train 4.051103 (lr=3.5114e-04) (hash(x)=153761544)
637
+ 5290 train 4.072213 (lr=3.4981e-04) (hash(x)=150927827)
638
+ 5300 val loss 4.1521
639
+ 5300 val perplexity 63.5674
640
+ 5300 train 3.976915 (lr=3.4848e-04) (hash(x)=152343580)
641
+ 5310 train 3.923151 (lr=3.4716e-04) (hash(x)=146690249)
642
+ 5320 train 4.077559 (lr=3.4583e-04) (hash(x)=155991065)
643
+ 5330 train 4.066111 (lr=3.4451e-04) (hash(x)=151437785)
644
+ 5340 train 3.992785 (lr=3.4319e-04) (hash(x)=158840015)
645
+ 5350 train 3.994755 (lr=3.4187e-04) (hash(x)=155080378)
646
+ 5360 train 4.039775 (lr=3.4055e-04) (hash(x)=150513270)
647
+ 5370 train 3.938250 (lr=3.3923e-04) (hash(x)=147729861)
648
+ 5380 train 3.970593 (lr=3.3792e-04) (hash(x)=149870080)
649
+ 5390 train 3.800252 (lr=3.3660e-04) (hash(x)=137577816)
650
+ 5400 val loss 4.1429
651
+ 5400 val perplexity 62.9846
652
+ 5400 train 3.990262 (lr=3.3529e-04) (hash(x)=148578264)
653
+ 5410 train 4.109337 (lr=3.3398e-04) (hash(x)=149338196)
654
+ 5420 train 3.999720 (lr=3.3267e-04) (hash(x)=149581125)
655
+ 5430 train 4.042928 (lr=3.3136e-04) (hash(x)=146340931)
656
+ 5440 train 3.942032 (lr=3.3006e-04) (hash(x)=127884657)
657
+ 5450 train 4.120041 (lr=3.2875e-04) (hash(x)=142632516)
658
+ 5460 train 4.110527 (lr=3.2745e-04) (hash(x)=146519487)
659
+ 5470 train 4.191153 (lr=3.2615e-04) (hash(x)=136409211)
660
+ 5480 train 4.099573 (lr=3.2485e-04) (hash(x)=158622762)
661
+ 5490 train 4.042549 (lr=3.2355e-04) (hash(x)=154703333)
662
+ 5500 val loss 4.1295
663
+ 5500 val perplexity 62.1495
664
+ 5500 train 4.187228 (lr=3.2226e-04) (hash(x)=145635833)
665
+ 5510 train 4.165156 (lr=3.2096e-04) (hash(x)=158781769)
666
+ 5520 train 4.143455 (lr=3.1967e-04) (hash(x)=142550540)
667
+ 5530 train 4.158270 (lr=3.1838e-04) (hash(x)=152024805)
668
+ 5540 train 3.956915 (lr=3.1709e-04) (hash(x)=145812646)
669
+ 5550 train 4.033095 (lr=3.1580e-04) (hash(x)=144941659)
670
+ 5560 train 4.017865 (lr=3.1451e-04) (hash(x)=145887398)
671
+ 5570 train 4.020897 (lr=3.1323e-04) (hash(x)=144988642)
672
+ 5580 train 4.030651 (lr=3.1195e-04) (hash(x)=146366431)
673
+ 5590 train 3.960159 (lr=3.1067e-04) (hash(x)=143301433)
674
+ 5600 val loss 4.1284
675
+ 5600 val perplexity 62.0772
676
+ 5600 train 4.140479 (lr=3.0939e-04) (hash(x)=156337844)
677
+ 5610 train 4.109231 (lr=3.0811e-04) (hash(x)=146546252)
678
+ 5620 train 4.042397 (lr=3.0684e-04) (hash(x)=137610278)
679
+ 5630 train 4.065857 (lr=3.0557e-04) (hash(x)=148847402)
680
+ 5640 train 4.019013 (lr=3.0429e-04) (hash(x)=148805052)
681
+ 5650 train 3.933738 (lr=3.0303e-04) (hash(x)=144882220)
682
+ 5660 train 4.018483 (lr=3.0176e-04) (hash(x)=147894128)
683
+ 5670 train 3.930913 (lr=3.0049e-04) (hash(x)=144855036)
684
+ 5680 train 3.920979 (lr=2.9923e-04) (hash(x)=148400728)
685
+ 5690 train 3.902102 (lr=2.9797e-04) (hash(x)=138377186)
686
+ 5700 val loss 4.1297
687
+ 5700 val perplexity 62.1622
688
+ 5700 train 3.934114 (lr=2.9671e-04) (hash(x)=147168506)
689
+ 5710 train 4.064880 (lr=2.9546e-04) (hash(x)=150486217)
690
+ 5720 train 4.035311 (lr=2.9420e-04) (hash(x)=148658778)
691
+ 5730 train 3.836588 (lr=2.9295e-04) (hash(x)=148402206)
692
+ 5740 train 3.893002 (lr=2.9170e-04) (hash(x)=144432268)
693
+ 5750 train 4.256199 (lr=2.9045e-04) (hash(x)=153770308)
694
+ 5760 train 4.012364 (lr=2.8920e-04) (hash(x)=150546820)
695
+ 5770 train 4.194032 (lr=2.8796e-04) (hash(x)=140176711)
696
+ 5780 train 4.092148 (lr=2.8672e-04) (hash(x)=164276928)
697
+ 5790 train 3.977988 (lr=2.8548e-04) (hash(x)=153418756)
698
+ 5800 val loss 4.1161
699
+ 5800 val perplexity 61.3208
700
+ 5800 train 4.069331 (lr=2.8424e-04) (hash(x)=159566920)
701
+ 5810 train 4.008832 (lr=2.8301e-04) (hash(x)=161266875)
702
+ 5820 train 4.028405 (lr=2.8178e-04) (hash(x)=156298754)
703
+ 5830 train 4.164270 (lr=2.8055e-04) (hash(x)=139302200)
704
+ 5840 train 4.070086 (lr=2.7932e-04) (hash(x)=126108545)
705
+ 5850 train 3.980844 (lr=2.7809e-04) (hash(x)=147526120)
706
+ 5860 train 4.061642 (lr=2.7687e-04) (hash(x)=150242466)
707
+ 5870 train 4.036651 (lr=2.7565e-04) (hash(x)=139851741)
708
+ 5880 train 3.963077 (lr=2.7443e-04) (hash(x)=159464779)
709
+ 5890 train 4.106423 (lr=2.7321e-04) (hash(x)=148069134)
710
+ 5900 val loss 4.1103
711
+ 5900 val perplexity 60.9652
712
+ 5900 train 4.090461 (lr=2.7200e-04) (hash(x)=158273929)
713
+ 5910 train 3.797563 (lr=2.7079e-04) (hash(x)=150013727)
714
+ 5920 train 4.065010 (lr=2.6958e-04) (hash(x)=148803117)
715
+ 5930 train 4.098542 (lr=2.6837e-04) (hash(x)=156729045)
716
+ 5940 train 4.012088 (lr=2.6717e-04) (hash(x)=150475415)
717
+ 5950 train 3.891433 (lr=2.6597e-04) (hash(x)=153585864)
718
+ 5960 train 3.901314 (lr=2.6477e-04) (hash(x)=145189335)
719
+ 5970 train 3.907321 (lr=2.6357e-04) (hash(x)=145673396)
720
+ 5980 train 4.044993 (lr=2.6238e-04) (hash(x)=138413986)
721
+ 5990 train 3.835882 (lr=2.6119e-04) (hash(x)=140270894)
722
+ 6000 val loss 4.1131
723
+ 6000 val perplexity 61.1352
724
+ 6000 train 4.036602 (lr=2.6000e-04) (hash(x)=156649749)
725
+ 6010 train 4.151230 (lr=2.5881e-04) (hash(x)=154762134)
726
+ 6020 train 4.150324 (lr=2.5763e-04) (hash(x)=156088385)
727
+ 6030 train 4.003519 (lr=2.5645e-04) (hash(x)=152521323)
728
+ 6040 train 3.922624 (lr=2.5527e-04) (hash(x)=148376918)
729
+ 6050 train 4.085218 (lr=2.5410e-04) (hash(x)=155673243)
730
+ 6060 train 4.063646 (lr=2.5292e-04) (hash(x)=142344218)
731
+ 6070 train 4.198800 (lr=2.5175e-04) (hash(x)=151751923)
732
+ 6080 train 4.020770 (lr=2.5059e-04) (hash(x)=156311576)
733
+ 6090 train 3.967968 (lr=2.4942e-04) (hash(x)=145058290)
734
+ 6100 val loss 4.1012
735
+ 6100 val perplexity 60.4106
736
+ 6100 train 4.053068 (lr=2.4826e-04) (hash(x)=146812388)
737
+ 6110 train 4.107574 (lr=2.4710e-04) (hash(x)=138440385)
738
+ 6120 train 4.105769 (lr=2.4595e-04) (hash(x)=162764713)
739
+ 6130 train 4.002239 (lr=2.4479e-04) (hash(x)=156456134)
740
+ 6140 train 3.935560 (lr=2.4364e-04) (hash(x)=136402679)
741
+ 6150 train 3.926162 (lr=2.4250e-04) (hash(x)=153380357)
742
+ 6160 train 3.694425 (lr=2.4135e-04) (hash(x)=170571967)
743
+ 6170 train 3.981608 (lr=2.4021e-04) (hash(x)=150984242)
744
+ 6180 train 4.105757 (lr=2.3907e-04) (hash(x)=143187168)
745
+ 6190 train 4.096103 (lr=2.3793e-04) (hash(x)=142999298)
746
+ 6200 val loss 4.0988
747
+ 6200 val perplexity 60.2690
748
+ 6200 train 4.007848 (lr=2.3680e-04) (hash(x)=143522146)
749
+ 6210 train 4.013492 (lr=2.3567e-04) (hash(x)=141961270)
750
+ 6220 train 4.037433 (lr=2.3454e-04) (hash(x)=158572673)
751
+ 6230 train 4.003564 (lr=2.3342e-04) (hash(x)=142949296)
752
+ 6240 train 3.847000 (lr=2.3230e-04) (hash(x)=128564671)
753
+ 6250 train 3.941605 (lr=2.3118e-04) (hash(x)=147515208)
754
+ 6260 train 3.898513 (lr=2.3006e-04) (hash(x)=151825788)
755
+ 6270 train 3.869681 (lr=2.2895e-04) (hash(x)=136625971)
756
+ 6280 train 3.932375 (lr=2.2784e-04) (hash(x)=146904276)
757
+ 6290 train 3.835681 (lr=2.2674e-04) (hash(x)=151876945)
758
+ 6300 val loss 4.0975
759
+ 6300 val perplexity 60.1899
760
+ 6300 train 3.881134 (lr=2.2563e-04) (hash(x)=150124474)
761
+ 6310 train 3.846164 (lr=2.2453e-04) (hash(x)=139897420)
762
+ 6320 train 3.988628 (lr=2.2344e-04) (hash(x)=136670912)
763
+ 6330 train 3.866796 (lr=2.2234e-04) (hash(x)=140302490)
764
+ 6340 train 3.807638 (lr=2.2125e-04) (hash(x)=145564419)
765
+ 6350 train 4.142789 (lr=2.2017e-04) (hash(x)=164988308)
766
+ 6360 train 4.120636 (lr=2.1908e-04) (hash(x)=139721549)
767
+ 6370 train 4.050726 (lr=2.1800e-04) (hash(x)=154183530)
768
+ 6380 train 4.050589 (lr=2.1692e-04) (hash(x)=143895743)
769
+ 6390 train 4.093756 (lr=2.1585e-04) (hash(x)=153091571)
770
+ 6400 val loss 4.0903
771
+ 6400 val perplexity 59.7572
772
+ 6400 train 4.057827 (lr=2.1478e-04) (hash(x)=141242117)
773
+ 6410 train 4.133918 (lr=2.1371e-04) (hash(x)=146638943)
774
+ 6420 train 4.036243 (lr=2.1265e-04) (hash(x)=147429468)
775
+ 6430 train 4.133788 (lr=2.1159e-04) (hash(x)=147104391)
776
+ 6440 train 3.990261 (lr=2.1053e-04) (hash(x)=147003825)
777
+ 6450 train 3.955217 (lr=2.0947e-04) (hash(x)=150910430)
778
+ 6460 train 3.885909 (lr=2.0842e-04) (hash(x)=165853376)
779
+ 6470 train 4.065632 (lr=2.0737e-04) (hash(x)=145957630)
780
+ 6480 train 3.960941 (lr=2.0633e-04) (hash(x)=155685880)
781
+ 6490 train 4.194860 (lr=2.0529e-04) (hash(x)=150307407)
782
+ 6500 val loss 4.0780
783
+ 6500 val perplexity 59.0246
784
+ 6500 train 3.845768 (lr=2.0425e-04) (hash(x)=143529762)
785
+ 6510 train 4.108937 (lr=2.0322e-04) (hash(x)=145652949)
786
+ 6520 train 4.003870 (lr=2.0218e-04) (hash(x)=142033417)
787
+ 6530 train 3.999331 (lr=2.0116e-04) (hash(x)=131869635)
788
+ 6540 train 4.048142 (lr=2.0013e-04) (hash(x)=151253771)
789
+ 6550 train 4.087264 (lr=1.9911e-04) (hash(x)=164286071)
790
+ 6560 train 3.983271 (lr=1.9810e-04) (hash(x)=148872674)
791
+ 6570 train 3.987697 (lr=1.9708e-04) (hash(x)=147653143)
792
+ 6580 train 3.992507 (lr=1.9607e-04) (hash(x)=151684539)
793
+ 6590 train 3.999106 (lr=1.9507e-04) (hash(x)=143354218)
794
+ 6600 val loss 4.0745
795
+ 6600 val perplexity 58.8205
796
+ 6600 train 3.979779 (lr=1.9406e-04) (hash(x)=136948374)
797
+ 6610 train 3.969388 (lr=1.9306e-04) (hash(x)=149846260)
798
+ 6620 train 3.871369 (lr=1.9207e-04) (hash(x)=151615259)
799
+ 6630 train 3.815044 (lr=1.9108e-04) (hash(x)=153129395)
800
+ 6640 train 4.005975 (lr=1.9009e-04) (hash(x)=133054060)
801
+ 6650 train 4.033938 (lr=1.8910e-04) (hash(x)=149911476)
802
+ 6660 train 3.998064 (lr=1.8812e-04) (hash(x)=150548384)
803
+ 6670 train 3.861836 (lr=1.8714e-04) (hash(x)=154950888)
804
+ 6680 train 3.729674 (lr=1.8617e-04) (hash(x)=145532409)
805
+ 6690 train 3.926174 (lr=1.8520e-04) (hash(x)=151197362)
806
+ 6700 val loss 4.0800
807
+ 6700 val perplexity 59.1451
808
+ 6700 train 3.849000 (lr=1.8423e-04) (hash(x)=146268592)
809
+ 6710 train 3.823605 (lr=1.8327e-04) (hash(x)=154495757)
810
+ 6720 train 3.921066 (lr=1.8231e-04) (hash(x)=168472165)
811
+ 6730 train 3.900536 (lr=1.8136e-04) (hash(x)=150287847)
812
+ 6740 train 3.851457 (lr=1.8040e-04) (hash(x)=151465011)
813
+ 6750 train 3.790107 (lr=1.7946e-04) (hash(x)=151898013)
814
+ 6760 train 3.948223 (lr=1.7851e-04) (hash(x)=145836413)
815
+ 6770 train 4.193804 (lr=1.7757e-04) (hash(x)=146057728)
816
+ 6780 train 4.072061 (lr=1.7663e-04) (hash(x)=160113674)
817
+ 6790 train 4.091331 (lr=1.7570e-04) (hash(x)=149844509)
818
+ 6800 val loss 4.0643
819
+ 6800 val perplexity 58.2227
820
+ 6800 train 4.045891 (lr=1.7477e-04) (hash(x)=152676836)
821
+ 6810 train 4.117140 (lr=1.7385e-04) (hash(x)=147510869)
822
+ 6820 train 4.201209 (lr=1.7293e-04) (hash(x)=149778657)
823
+ 6830 train 4.119492 (lr=1.7201e-04) (hash(x)=143851885)
824
+ 6840 train 4.125612 (lr=1.7110e-04) (hash(x)=158011137)
825
+ 6850 train 4.179607 (lr=1.7019e-04) (hash(x)=171197810)
826
+ 6860 train 4.105493 (lr=1.6928e-04) (hash(x)=151171060)
827
+ 6870 train 4.050188 (lr=1.6838e-04) (hash(x)=135613468)
828
+ 6880 train 4.081071 (lr=1.6748e-04) (hash(x)=153789364)
829
+ 6890 train 4.026597 (lr=1.6659e-04) (hash(x)=128928397)
830
+ 6900 val loss 4.0528
831
+ 6900 val perplexity 57.5588
832
+ 6900 train 4.079830 (lr=1.6570e-04) (hash(x)=134657776)
833
+ 6910 train 4.023063 (lr=1.6481e-04) (hash(x)=170095944)
834
+ 6920 train 4.290233 (lr=1.6393e-04) (hash(x)=147408535)
835
+ 6930 train 4.049634 (lr=1.6305e-04) (hash(x)=141187674)
836
+ 6940 train 3.937998 (lr=1.6218e-04) (hash(x)=157752429)
837
+ 6950 train 4.137964 (lr=1.6131e-04) (hash(x)=149793159)
838
+ 6960 train 4.010489 (lr=1.6044e-04) (hash(x)=135786360)
839
+ 6970 train 3.919931 (lr=1.5958e-04) (hash(x)=144372214)
840
+ 6980 train 4.029025 (lr=1.5872e-04) (hash(x)=147592321)
841
+ 6990 train 3.964561 (lr=1.5787e-04) (hash(x)=135072395)
842
+ 7000 val loss 4.0563
843
+ 7000 val perplexity 57.7604
844
+ 7000 train 3.951340 (lr=1.5702e-04) (hash(x)=166721861)
845
+ 7010 train 4.034161 (lr=1.5618e-04) (hash(x)=157177130)
846
+ 7020 train 3.925609 (lr=1.5533e-04) (hash(x)=154916555)
847
+ 7030 train 4.058108 (lr=1.5450e-04) (hash(x)=139686863)
848
+ 7040 train 4.009642 (lr=1.5366e-04) (hash(x)=139606744)
849
+ 7050 train 4.066445 (lr=1.5284e-04) (hash(x)=149487898)
850
+ 7060 train 3.973587 (lr=1.5201e-04) (hash(x)=162536366)
851
+ 7070 train 4.000601 (lr=1.5119e-04) (hash(x)=144765016)
852
+ 7080 train 3.985364 (lr=1.5037e-04) (hash(x)=146001679)
853
+ 7090 train 3.783084 (lr=1.4956e-04) (hash(x)=153352255)
854
+ 7100 val loss 4.0540
855
+ 7100 val perplexity 57.6291
856
+ 7100 train 3.812274 (lr=1.4875e-04) (hash(x)=135496702)
857
+ 7110 train 3.988553 (lr=1.4795e-04) (hash(x)=155057574)
858
+ 7120 train 3.948895 (lr=1.4715e-04) (hash(x)=145862613)
859
+ 7130 train 3.651031 (lr=1.4636e-04) (hash(x)=160806933)
860
+ 7140 train 3.721727 (lr=1.4556e-04) (hash(x)=154582601)
861
+ 7150 train 3.820341 (lr=1.4478e-04) (hash(x)=151854329)
862
+ 7160 train 3.998087 (lr=1.4400e-04) (hash(x)=153875998)
863
+ 7170 train 3.794763 (lr=1.4322e-04) (hash(x)=143448354)
864
+ 7180 train 3.824769 (lr=1.4244e-04) (hash(x)=141619703)
865
+ 7190 train 4.083413 (lr=1.4167e-04) (hash(x)=154769375)
866
+ 7200 val loss 4.0518
867
+ 7200 val perplexity 57.5023
868
+ 7200 train 4.180781 (lr=1.4091e-04) (hash(x)=155567461)
869
+ 7210 train 4.124598 (lr=1.4015e-04) (hash(x)=160242796)
870
+ 7220 train 4.167083 (lr=1.3939e-04) (hash(x)=146362204)
871
+ 7230 train 3.890878 (lr=1.3864e-04) (hash(x)=155673641)
872
+ 7240 train 4.109086 (lr=1.3789e-04) (hash(x)=163454733)
873
+ 7250 train 4.113152 (lr=1.3715e-04) (hash(x)=166254410)
874
+ 7260 train 3.954510 (lr=1.3641e-04) (hash(x)=149576049)
875
+ 7270 train 4.009492 (lr=1.3568e-04) (hash(x)=159937867)
876
+ 7280 train 4.154032 (lr=1.3495e-04) (hash(x)=151335403)
877
+ 7290 train 3.974555 (lr=1.3422e-04) (hash(x)=148867536)
878
+ 7300 val loss 4.0377
879
+ 7300 val perplexity 56.6960
880
+ 7300 train 3.934678 (lr=1.3350e-04) (hash(x)=142803829)
881
+ 7310 train 4.095845 (lr=1.3278e-04) (hash(x)=149482208)
882
+ 7320 train 3.879230 (lr=1.3207e-04) (hash(x)=144226264)
883
+ 7330 train 3.909573 (lr=1.3136e-04) (hash(x)=189898865)
884
+ 7340 train 4.084133 (lr=1.3066e-04) (hash(x)=153884999)
885
+ 7350 train 3.999425 (lr=1.2996e-04) (hash(x)=149106483)
886
+ 7360 train 4.028568 (lr=1.2926e-04) (hash(x)=146033939)
887
+ 7370 train 4.019186 (lr=1.2857e-04) (hash(x)=133850095)
888
+ 7380 train 4.025619 (lr=1.2789e-04) (hash(x)=155008783)
889
+ 7390 train 4.034833 (lr=1.2721e-04) (hash(x)=158169816)
890
+ 7400 val loss 4.0395
891
+ 7400 val perplexity 56.7998
892
+ 7400 train 3.966940 (lr=1.2653e-04) (hash(x)=145294178)
893
+ 7410 train 3.893201 (lr=1.2586e-04) (hash(x)=156501995)
894
+ 7420 train 3.994529 (lr=1.2519e-04) (hash(x)=154883953)
895
+ 7430 train 4.005596 (lr=1.2453e-04) (hash(x)=153861637)
896
+ 7440 train 3.959023 (lr=1.2387e-04) (hash(x)=143884732)
897
+ 7450 train 4.069255 (lr=1.2322e-04) (hash(x)=147629077)
898
+ 7460 train 3.958207 (lr=1.2257e-04) (hash(x)=147092200)
899
+ 7470 train 3.829338 (lr=1.2192e-04) (hash(x)=153667163)
900
+ 7480 train 3.899948 (lr=1.2128e-04) (hash(x)=155400781)
901
+ 7490 train 4.147258 (lr=1.2065e-04) (hash(x)=158198995)
902
+ 7500 val loss 4.0404
903
+ 7500 val perplexity 56.8509
904
+ 7500 train 3.854214 (lr=1.2002e-04) (hash(x)=150573713)
905
+ 7510 train 3.915268 (lr=1.1939e-04) (hash(x)=159059956)
906
+ 7520 train 3.735042 (lr=1.1877e-04) (hash(x)=148811846)
907
+ 7530 train 3.896099 (lr=1.1816e-04) (hash(x)=143727205)
908
+ 7540 train 3.784130 (lr=1.1754e-04) (hash(x)=154980561)
909
+ 7550 train 3.900521 (lr=1.1694e-04) (hash(x)=151143108)
910
+ 7560 train 3.867799 (lr=1.1633e-04) (hash(x)=133379076)
911
+ 7570 train 3.798702 (lr=1.1574e-04) (hash(x)=148338995)
912
+ 7580 train 3.870606 (lr=1.1514e-04) (hash(x)=156733806)
913
+ 7590 train 3.991204 (lr=1.1455e-04) (hash(x)=159299389)
914
+ 7600 val loss 4.0380
915
+ 7600 val perplexity 56.7140
916
+ 7600 train 4.029341 (lr=1.1397e-04) (hash(x)=142771511)
917
+ 7610 train 4.126600 (lr=1.1339e-04) (hash(x)=157959626)
918
+ 7620 train 4.136850 (lr=1.1282e-04) (hash(x)=156257811)
919
+ 7630 train 4.179113 (lr=1.1225e-04) (hash(x)=142730196)
920
+ 7640 train 4.026977 (lr=1.1168e-04) (hash(x)=147923184)
921
+ 7650 train 3.920937 (lr=1.1112e-04) (hash(x)=154290536)
922
+ 7660 train 3.937857 (lr=1.1057e-04) (hash(x)=156433550)
923
+ 7670 train 4.081962 (lr=1.1002e-04) (hash(x)=146611945)
924
+ 7680 train 3.900638 (lr=1.0947e-04) (hash(x)=153745377)
925
+ 7690 train 4.003652 (lr=1.0893e-04) (hash(x)=150833491)
926
+ 7700 val loss 4.0282
927
+ 7700 val perplexity 56.1595
928
+ 7700 train 4.016505 (lr=1.0840e-04) (hash(x)=143602175)
929
+ 7710 train 3.901020 (lr=1.0786e-04) (hash(x)=137035204)
930
+ 7720 train 4.338457 (lr=1.0734e-04) (hash(x)=157409397)
931
+ 7730 train 3.953751 (lr=1.0682e-04) (hash(x)=180056426)
932
+ 7740 train 4.104877 (lr=1.0630e-04) (hash(x)=167791182)
933
+ 7750 train 3.899125 (lr=1.0579e-04) (hash(x)=144902753)
934
+ 7760 train 4.002109 (lr=1.0528e-04) (hash(x)=142048628)
935
+ 7770 train 4.202297 (lr=1.0478e-04) (hash(x)=150623561)
936
+ 7780 train 3.809595 (lr=1.0428e-04) (hash(x)=133068514)
937
+ 7790 train 4.035431 (lr=1.0379e-04) (hash(x)=144372942)
938
+ 7800 val loss 4.0275
939
+ 7800 val perplexity 56.1206
940
+ 7800 train 4.198331 (lr=1.0330e-04) (hash(x)=152379862)
941
+ 7810 train 3.894248 (lr=1.0282e-04) (hash(x)=149929258)
942
+ 7820 train 4.058632 (lr=1.0234e-04) (hash(x)=147738863)
943
+ 7830 train 4.068220 (lr=1.0187e-04) (hash(x)=150885420)
944
+ 7840 train 4.071206 (lr=1.0140e-04) (hash(x)=151751961)
945
+ 7850 train 3.988155 (lr=1.0094e-04) (hash(x)=141061924)
946
+ 7860 train 4.047636 (lr=1.0048e-04) (hash(x)=165946945)
947
+ 7870 train 3.936528 (lr=1.0002e-04) (hash(x)=162222193)
948
+ 7880 train 4.049668 (lr=9.9576e-05) (hash(x)=156110032)
949
+ 7890 train 3.808918 (lr=9.9133e-05) (hash(x)=155496634)
950
+ 7900 val loss 4.0257
951
+ 7900 val perplexity 56.0216
952
+ 7900 train 3.761095 (lr=9.8694e-05) (hash(x)=146655921)
953
+ 7910 train 3.741875 (lr=9.8261e-05) (hash(x)=154169360)
954
+ 7920 train 3.924557 (lr=9.7832e-05) (hash(x)=164630987)
955
+ 7930 train 3.845012 (lr=9.7408e-05) (hash(x)=157591350)
956
+ 7940 train 3.824436 (lr=9.6990e-05) (hash(x)=147986483)
957
+ 7950 train 3.652662 (lr=9.6576e-05) (hash(x)=153958875)
958
+ 7960 train 3.951481 (lr=9.6167e-05) (hash(x)=149307886)
959
+ 7970 train 3.772457 (lr=9.5764e-05) (hash(x)=143190550)
960
+ 7980 train 3.790895 (lr=9.5365e-05) (hash(x)=157802449)
961
+ 7990 train 3.810043 (lr=9.4971e-05) (hash(x)=167181278)
962
+ 8000 val loss 4.0318
963
+ 8000 val perplexity 56.3626
964
+ 8000 train 4.184242 (lr=9.4583e-05) (hash(x)=148262482)
965
+ 8010 train 4.018644 (lr=9.4199e-05) (hash(x)=151119149)
966
+ 8020 train 3.960131 (lr=9.3820e-05) (hash(x)=149748120)
967
+ 8030 train 3.948734 (lr=9.3446e-05) (hash(x)=152462863)
968
+ 8040 train 4.085291 (lr=9.3078e-05) (hash(x)=148714170)
969
+ 8050 train 4.130264 (lr=9.2714e-05) (hash(x)=150219940)
970
+ 8060 train 4.028420 (lr=9.2356e-05) (hash(x)=154859690)
971
+ 8070 train 3.985637 (lr=9.2002e-05) (hash(x)=156635675)
972
+ 8080 train 4.133367 (lr=9.1654e-05) (hash(x)=164430040)
973
+ 8090 train 3.959545 (lr=9.1310e-05) (hash(x)=149003088)
974
+ 8100 val loss 4.0155
975
+ 8100 val perplexity 55.4484
976
+ 8100 train 4.061445 (lr=9.0972e-05) (hash(x)=147683655)
977
+ 8110 train 3.922217 (lr=9.0638e-05) (hash(x)=142849380)
978
+ 8120 train 3.958202 (lr=9.0310e-05) (hash(x)=147408722)
979
+ 8130 train 4.007241 (lr=8.9987e-05) (hash(x)=154095861)
980
+ 8140 train 4.048129 (lr=8.9669e-05) (hash(x)=154770296)
981
+ 8150 train 4.044976 (lr=8.9356e-05) (hash(x)=150175912)
982
+ 8160 train 4.034694 (lr=8.9048e-05) (hash(x)=146846694)
983
+ 8170 train 3.436255 (lr=8.8745e-05) (hash(x)=161621402)
984
+ 8180 train 4.228323 (lr=8.8447e-05) (hash(x)=151058128)
985
+ 8190 train 4.158896 (lr=8.8154e-05) (hash(x)=157265995)
986
+ 8200 val loss 4.0195
987
+ 8200 val perplexity 55.6728
988
+ 8200 train 4.222792 (lr=8.7867e-05) (hash(x)=157312987)
989
+ 8210 train 4.030418 (lr=8.7584e-05) (hash(x)=158755166)
990
+ 8220 train 3.884910 (lr=8.7307e-05) (hash(x)=137031704)
991
+ 8230 train 3.853711 (lr=8.7035e-05) (hash(x)=133909075)
992
+ 8240 train 4.044965 (lr=8.6768e-05) (hash(x)=152756441)
993
+ 8250 train 3.982848 (lr=8.6506e-05) (hash(x)=161956442)
994
+ 8260 train 4.023536 (lr=8.6249e-05) (hash(x)=146994534)
995
+ 8270 train 4.080547 (lr=8.5997e-05) (hash(x)=154925653)
996
+ 8280 train 4.068771 (lr=8.5750e-05) (hash(x)=141129942)
997
+ 8290 train 3.973023 (lr=8.5509e-05) (hash(x)=146178879)
998
+ 8300 val loss 4.0179
999
+ 8300 val perplexity 55.5837
1000
+ 8300 train 3.928176 (lr=8.5273e-05) (hash(x)=141107543)
1001
+ 8310 train 3.959295 (lr=8.5041e-05) (hash(x)=153322855)
1002
+ 8320 train 4.042318 (lr=8.4815e-05) (hash(x)=151197408)
1003
+ 8330 train 4.215587 (lr=8.4594e-05) (hash(x)=155927851)
1004
+ 8340 train 4.134678 (lr=8.4379e-05) (hash(x)=162484336)
1005
+ 8350 train 4.015969 (lr=8.4168e-05) (hash(x)=144280686)
1006
+ 8360 train 3.971004 (lr=8.3963e-05) (hash(x)=147652229)
1007
+ 8370 train 4.104110 (lr=8.3762e-05) (hash(x)=150071753)
1008
+ 8380 train 3.950316 (lr=8.3567e-05) (hash(x)=146798938)
1009
+ 8390 train 3.983739 (lr=8.3377e-05) (hash(x)=149596226)
1010
+ 8400 val loss 4.0099
1011
+ 8400 val perplexity 55.1423
1012
+ 8400 train 4.063657 (lr=8.3193e-05) (hash(x)=141323024)
1013
+ 8410 train 4.352611 (lr=8.3013e-05) (hash(x)=154823144)
1014
+ 8420 train 3.967793 (lr=8.2839e-05) (hash(x)=151554301)
1015
+ 8430 train 3.985235 (lr=8.2669e-05) (hash(x)=153707717)
1016
+ 8440 train 3.984105 (lr=8.2505e-05) (hash(x)=156091926)
1017
+ 8450 train 4.014400 (lr=8.2347e-05) (hash(x)=147111043)
1018
+ 8460 train 4.130294 (lr=8.2193e-05) (hash(x)=135068736)
1019
+ 8470 train 4.041696 (lr=8.2044e-05) (hash(x)=142255415)
1020
+ 8480 train 4.053208 (lr=8.1901e-05) (hash(x)=138060402)
1021
+ 8490 train 3.942937 (lr=8.1763e-05) (hash(x)=154789940)
1022
+ 8500 val loss 4.0083
1023
+ 8500 val perplexity 55.0545
1024
+ 8500 train 3.926399 (lr=8.1630e-05) (hash(x)=150696521)
1025
+ 8510 train 3.848886 (lr=8.1502e-05) (hash(x)=139242896)
1026
+ 8520 train 3.922772 (lr=8.1380e-05) (hash(x)=148066730)
1027
+ 8530 train 4.046844 (lr=8.1263e-05) (hash(x)=148649882)
1028
+ 8540 train 3.854414 (lr=8.1150e-05) (hash(x)=148449712)
1029
+ 8550 train 4.020672 (lr=8.1044e-05) (hash(x)=157155657)
1030
+ 8560 train 3.936938 (lr=8.0942e-05) (hash(x)=139492416)
1031
+ 8570 train 3.841559 (lr=8.0845e-05) (hash(x)=157567533)
1032
+ 8580 train 3.911824 (lr=8.0754e-05) (hash(x)=143032656)
1033
+ 8590 train 4.036394 (lr=8.0668e-05) (hash(x)=150899581)
1034
+ 8600 val loss 4.0119
1035
+ 8600 val perplexity 55.2510
1036
+ 8600 train 3.925280 (lr=8.0587e-05) (hash(x)=162288191)
1037
+ 8610 train 4.066739 (lr=8.0511e-05) (hash(x)=148361485)
1038
+ 8620 train 3.970561 (lr=8.0441e-05) (hash(x)=151740051)
1039
+ 8630 train 3.924320 (lr=8.0376e-05) (hash(x)=154125251)
1040
+ 8640 train 3.989377 (lr=8.0316e-05) (hash(x)=145857675)
1041
+ 8650 train 3.857009 (lr=8.0261e-05) (hash(x)=148185775)
1042
+ 8660 train 3.899075 (lr=8.0211e-05) (hash(x)=149778450)
1043
+ 8670 train 4.178552 (lr=8.0167e-05) (hash(x)=148123274)
1044
+ 8680 train 3.953502 (lr=8.0128e-05) (hash(x)=129372639)
1045
+ 8690 train 4.027688 (lr=8.0094e-05) (hash(x)=145050683)
1046
+ 8700 val loss 4.0051
1047
+ 8700 val perplexity 54.8754
1048
+ 8700 train 3.949309 (lr=8.0065e-05) (hash(x)=152860941)
1049
+ 8710 train 4.041749 (lr=8.0042e-05) (hash(x)=156971498)
1050
+ 8720 train 4.067986 (lr=8.0023e-05) (hash(x)=155513041)
1051
+ 8730 train 4.000155 (lr=8.0010e-05) (hash(x)=167451760)
1052
+ 8740 train 4.076605 (lr=8.0003e-05) (hash(x)=158768428)
1053
+ 8749 val loss 4.0053
1054
+ 8749 val perplexity 54.8893
lr8e-4_total_batch_size61440_baseline_seed1341/model_08749.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aad9c8e924ea8c55081d196a8a742141a6e00256ee71dbb7a3838a0188bbcef7
3
+ size 92843394
lr8e-4_total_batch_size61440_baseline_seed1341/optimizer_08749.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:71737229962bda0fa8946c9f1f57abd92930cd7c1d196680e37a28fbde3d1f4e
3
+ size 179406214