andrew-healey commited on
Commit
8829fab
·
verified ·
1 Parent(s): 723aedc

Upload folder using huggingface_hub

Browse files
lr2e-4_total_batch_size61440_baseline_seed1339/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_11/lr2e-4_total_batch_size61440_baseline_seed1339", "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": 1339, "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": 0.0002, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "2e-4_61440", "n_embd": 256}
lr2e-4_total_batch_size61440_baseline_seed1339/dataloader_08749.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:953385078aa3787b69fc6857dfd48b0a2cd2f4d27c6f8892e01211aca53d07f5
3
+ size 964
lr2e-4_total_batch_size61440_baseline_seed1339/log2.txt ADDED
@@ -0,0 +1,1054 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ max_steps: 8750
2
+ 0 val loss 11.2592
3
+ 0 val perplexity 77590.4141
4
+ 0 train 11.272774 (lr=4.0000e-07) (hash(x)=150724848)
5
+ 10 train 10.749363 (lr=4.4000e-06) (hash(x)=149663361)
6
+ 20 train 10.211926 (lr=8.4000e-06) (hash(x)=153483789)
7
+ 30 train 10.005648 (lr=1.2400e-05) (hash(x)=145733176)
8
+ 40 train 9.770498 (lr=1.6400e-05) (hash(x)=145287265)
9
+ 50 train 9.663364 (lr=2.0400e-05) (hash(x)=149217061)
10
+ 60 train 9.476939 (lr=2.4400e-05) (hash(x)=150157569)
11
+ 70 train 9.253351 (lr=2.8400e-05) (hash(x)=140292841)
12
+ 80 train 9.079046 (lr=3.2400e-05) (hash(x)=152157598)
13
+ 90 train 8.799547 (lr=3.6400e-05) (hash(x)=149756968)
14
+ 100 val loss 8.5346
15
+ 100 val perplexity 5087.6470
16
+ 100 train 8.538797 (lr=4.0400e-05) (hash(x)=149910534)
17
+ 110 train 8.288116 (lr=4.4400e-05) (hash(x)=159020328)
18
+ 120 train 8.094198 (lr=4.8400e-05) (hash(x)=150731273)
19
+ 130 train 7.838605 (lr=5.2400e-05) (hash(x)=156727355)
20
+ 140 train 7.792005 (lr=5.6400e-05) (hash(x)=166473281)
21
+ 150 train 7.493546 (lr=6.0400e-05) (hash(x)=142185643)
22
+ 160 train 7.459000 (lr=6.4400e-05) (hash(x)=151371145)
23
+ 170 train 7.471778 (lr=6.8400e-05) (hash(x)=151601659)
24
+ 180 train 7.321662 (lr=7.2400e-05) (hash(x)=153368917)
25
+ 190 train 7.185073 (lr=7.6400e-05) (hash(x)=145008196)
26
+ 200 val loss 7.3137
27
+ 200 val perplexity 1500.7821
28
+ 200 train 7.249612 (lr=8.0400e-05) (hash(x)=148123706)
29
+ 210 train 7.115513 (lr=8.4400e-05) (hash(x)=145714762)
30
+ 220 train 7.131500 (lr=8.8400e-05) (hash(x)=150622583)
31
+ 230 train 6.932737 (lr=9.2400e-05) (hash(x)=146512648)
32
+ 240 train 6.952878 (lr=9.6400e-05) (hash(x)=151535384)
33
+ 250 train 6.846394 (lr=1.0040e-04) (hash(x)=149952383)
34
+ 260 train 6.759600 (lr=1.0440e-04) (hash(x)=148726088)
35
+ 270 train 6.749523 (lr=1.0840e-04) (hash(x)=152776695)
36
+ 280 train 6.774832 (lr=1.1240e-04) (hash(x)=157260375)
37
+ 290 train 6.822526 (lr=1.1640e-04) (hash(x)=151559106)
38
+ 300 val loss 6.8252
39
+ 300 val perplexity 920.7277
40
+ 300 train 6.582307 (lr=1.2040e-04) (hash(x)=146678221)
41
+ 310 train 6.681046 (lr=1.2440e-04) (hash(x)=156713527)
42
+ 320 train 6.717107 (lr=1.2840e-04) (hash(x)=144947971)
43
+ 330 train 6.845701 (lr=1.3240e-04) (hash(x)=158684944)
44
+ 340 train 6.726373 (lr=1.3640e-04) (hash(x)=162318756)
45
+ 350 train 6.534189 (lr=1.4040e-04) (hash(x)=156180736)
46
+ 360 train 6.727784 (lr=1.4440e-04) (hash(x)=178373797)
47
+ 370 train 6.587618 (lr=1.4840e-04) (hash(x)=154389449)
48
+ 380 train 6.394759 (lr=1.5240e-04) (hash(x)=150369080)
49
+ 390 train 6.567123 (lr=1.5640e-04) (hash(x)=159840654)
50
+ 400 val loss 6.4202
51
+ 400 val perplexity 614.1566
52
+ 400 train 6.294370 (lr=1.6040e-04) (hash(x)=151700982)
53
+ 410 train 6.316737 (lr=1.6440e-04) (hash(x)=160467274)
54
+ 420 train 6.255429 (lr=1.6840e-04) (hash(x)=148930522)
55
+ 430 train 6.450225 (lr=1.7240e-04) (hash(x)=154565303)
56
+ 440 train 6.341495 (lr=1.7640e-04) (hash(x)=148079227)
57
+ 450 train 6.138906 (lr=1.8040e-04) (hash(x)=147288467)
58
+ 460 train 6.124635 (lr=1.8440e-04) (hash(x)=145272617)
59
+ 470 train 6.106531 (lr=1.8840e-04) (hash(x)=123126729)
60
+ 480 train 6.154691 (lr=1.9240e-04) (hash(x)=156287490)
61
+ 490 train 6.267242 (lr=1.9640e-04) (hash(x)=161700145)
62
+ 500 val loss 6.1721
63
+ 500 val perplexity 479.1836
64
+ 500 train 6.267108 (lr=2.0000e-04) (hash(x)=156182087)
65
+ 510 train 5.863787 (lr=2.0000e-04) (hash(x)=131674504)
66
+ 520 train 6.157838 (lr=2.0000e-04) (hash(x)=162758254)
67
+ 530 train 5.971799 (lr=1.9999e-04) (hash(x)=133510409)
68
+ 540 train 6.012506 (lr=1.9999e-04) (hash(x)=157363914)
69
+ 550 train 5.936107 (lr=1.9998e-04) (hash(x)=147861550)
70
+ 560 train 5.874501 (lr=1.9998e-04) (hash(x)=149449280)
71
+ 570 train 5.856480 (lr=1.9997e-04) (hash(x)=143899904)
72
+ 580 train 5.842281 (lr=1.9996e-04) (hash(x)=149266312)
73
+ 590 train 6.031716 (lr=1.9995e-04) (hash(x)=164352774)
74
+ 600 val loss 5.9533
75
+ 600 val perplexity 385.0044
76
+ 600 train 5.869205 (lr=1.9993e-04) (hash(x)=149318660)
77
+ 610 train 5.747165 (lr=1.9992e-04) (hash(x)=146101781)
78
+ 620 train 5.676724 (lr=1.9991e-04) (hash(x)=137184623)
79
+ 630 train 5.602361 (lr=1.9989e-04) (hash(x)=145791873)
80
+ 640 train 5.615479 (lr=1.9987e-04) (hash(x)=143999212)
81
+ 650 train 5.694413 (lr=1.9985e-04) (hash(x)=151631103)
82
+ 660 train 5.432843 (lr=1.9983e-04) (hash(x)=138360108)
83
+ 670 train 5.816114 (lr=1.9981e-04) (hash(x)=156637679)
84
+ 680 train 5.457730 (lr=1.9979e-04) (hash(x)=143752601)
85
+ 690 train 5.645088 (lr=1.9976e-04) (hash(x)=158368903)
86
+ 700 val loss 5.8279
87
+ 700 val perplexity 339.6353
88
+ 700 train 5.832450 (lr=1.9974e-04) (hash(x)=150482428)
89
+ 710 train 5.789883 (lr=1.9971e-04) (hash(x)=158085060)
90
+ 720 train 5.857641 (lr=1.9968e-04) (hash(x)=159461069)
91
+ 730 train 5.680449 (lr=1.9966e-04) (hash(x)=148795562)
92
+ 740 train 5.690934 (lr=1.9962e-04) (hash(x)=149967984)
93
+ 750 train 5.668358 (lr=1.9959e-04) (hash(x)=144458617)
94
+ 760 train 5.747774 (lr=1.9956e-04) (hash(x)=155106202)
95
+ 770 train 5.582382 (lr=1.9952e-04) (hash(x)=144506191)
96
+ 780 train 5.683739 (lr=1.9949e-04) (hash(x)=155825158)
97
+ 790 train 5.719213 (lr=1.9945e-04) (hash(x)=158835381)
98
+ 800 val loss 5.6590
99
+ 800 val perplexity 286.8607
100
+ 800 train 5.588141 (lr=1.9941e-04) (hash(x)=143268605)
101
+ 810 train 5.484612 (lr=1.9937e-04) (hash(x)=139072755)
102
+ 820 train 5.726779 (lr=1.9933e-04) (hash(x)=179384792)
103
+ 830 train 5.448969 (lr=1.9929e-04) (hash(x)=151086962)
104
+ 840 train 5.966517 (lr=1.9925e-04) (hash(x)=142711393)
105
+ 850 train 5.544554 (lr=1.9920e-04) (hash(x)=150817011)
106
+ 860 train 5.439733 (lr=1.9916e-04) (hash(x)=138495801)
107
+ 870 train 5.503563 (lr=1.9911e-04) (hash(x)=155474016)
108
+ 880 train 5.536631 (lr=1.9906e-04) (hash(x)=151951501)
109
+ 890 train 5.391989 (lr=1.9901e-04) (hash(x)=143699524)
110
+ 900 val loss 5.5499
111
+ 900 val perplexity 257.2040
112
+ 900 train 5.550200 (lr=1.9896e-04) (hash(x)=152322423)
113
+ 910 train 5.506156 (lr=1.9891e-04) (hash(x)=151010509)
114
+ 920 train 5.307556 (lr=1.9885e-04) (hash(x)=148720342)
115
+ 930 train 5.178290 (lr=1.9880e-04) (hash(x)=125869480)
116
+ 940 train 5.496110 (lr=1.9874e-04) (hash(x)=150349503)
117
+ 950 train 5.297901 (lr=1.9868e-04) (hash(x)=143574126)
118
+ 960 train 5.436388 (lr=1.9862e-04) (hash(x)=146292562)
119
+ 970 train 5.403831 (lr=1.9856e-04) (hash(x)=162256596)
120
+ 980 train 5.322253 (lr=1.9850e-04) (hash(x)=137640259)
121
+ 990 train 5.284953 (lr=1.9844e-04) (hash(x)=141767794)
122
+ 1000 val loss 5.4808
123
+ 1000 val perplexity 240.0408
124
+ 1000 train 5.405114 (lr=1.9837e-04) (hash(x)=147904298)
125
+ 1010 train 5.352182 (lr=1.9831e-04) (hash(x)=162239186)
126
+ 1020 train 5.171601 (lr=1.9824e-04) (hash(x)=140706537)
127
+ 1030 train 5.240733 (lr=1.9817e-04) (hash(x)=155556720)
128
+ 1040 train 5.160347 (lr=1.9810e-04) (hash(x)=144746445)
129
+ 1050 train 5.325778 (lr=1.9803e-04) (hash(x)=144719400)
130
+ 1060 train 5.069970 (lr=1.9796e-04) (hash(x)=153065665)
131
+ 1070 train 5.209957 (lr=1.9789e-04) (hash(x)=136468722)
132
+ 1080 train 5.243615 (lr=1.9781e-04) (hash(x)=157674756)
133
+ 1090 train 5.336040 (lr=1.9774e-04) (hash(x)=155882884)
134
+ 1100 val loss 5.4028
135
+ 1100 val perplexity 222.0198
136
+ 1100 train 5.641511 (lr=1.9766e-04) (hash(x)=154343147)
137
+ 1110 train 5.376168 (lr=1.9758e-04) (hash(x)=158582939)
138
+ 1120 train 5.265318 (lr=1.9750e-04) (hash(x)=162479784)
139
+ 1130 train 5.410789 (lr=1.9742e-04) (hash(x)=155704345)
140
+ 1140 train 5.307615 (lr=1.9734e-04) (hash(x)=146934299)
141
+ 1150 train 5.281294 (lr=1.9726e-04) (hash(x)=154372635)
142
+ 1160 train 5.235214 (lr=1.9717e-04) (hash(x)=145536832)
143
+ 1170 train 5.540393 (lr=1.9709e-04) (hash(x)=158785901)
144
+ 1180 train 5.374144 (lr=1.9700e-04) (hash(x)=158271614)
145
+ 1190 train 5.374232 (lr=1.9691e-04) (hash(x)=157365589)
146
+ 1200 val loss 5.3197
147
+ 1200 val perplexity 204.3141
148
+ 1200 train 5.342470 (lr=1.9682e-04) (hash(x)=141843115)
149
+ 1210 train 5.300400 (lr=1.9673e-04) (hash(x)=146807973)
150
+ 1220 train 5.233961 (lr=1.9664e-04) (hash(x)=144149947)
151
+ 1230 train 5.211915 (lr=1.9654e-04) (hash(x)=147869063)
152
+ 1240 train 5.245860 (lr=1.9645e-04) (hash(x)=152217389)
153
+ 1250 train 5.209893 (lr=1.9635e-04) (hash(x)=149417679)
154
+ 1260 train 5.291643 (lr=1.9626e-04) (hash(x)=144029489)
155
+ 1270 train 5.383012 (lr=1.9616e-04) (hash(x)=154328303)
156
+ 1280 train 5.256554 (lr=1.9606e-04) (hash(x)=151625838)
157
+ 1290 train 5.169509 (lr=1.9596e-04) (hash(x)=140270541)
158
+ 1300 val loss 5.2514
159
+ 1300 val perplexity 190.8285
160
+ 1300 train 5.190781 (lr=1.9586e-04) (hash(x)=145279030)
161
+ 1310 train 5.231546 (lr=1.9575e-04) (hash(x)=140279630)
162
+ 1320 train 4.921338 (lr=1.9565e-04) (hash(x)=133642994)
163
+ 1330 train 5.131127 (lr=1.9554e-04) (hash(x)=146634776)
164
+ 1340 train 5.250562 (lr=1.9543e-04) (hash(x)=155919973)
165
+ 1350 train 5.235881 (lr=1.9533e-04) (hash(x)=146363911)
166
+ 1360 train 5.107622 (lr=1.9522e-04) (hash(x)=150613636)
167
+ 1370 train 5.082951 (lr=1.9511e-04) (hash(x)=148536643)
168
+ 1380 train 5.100183 (lr=1.9499e-04) (hash(x)=152803906)
169
+ 1390 train 5.199052 (lr=1.9488e-04) (hash(x)=150383305)
170
+ 1400 val loss 5.2120
171
+ 1400 val perplexity 183.4602
172
+ 1400 train 5.120304 (lr=1.9477e-04) (hash(x)=152507639)
173
+ 1410 train 4.897914 (lr=1.9465e-04) (hash(x)=161841408)
174
+ 1420 train 5.100336 (lr=1.9453e-04) (hash(x)=167824828)
175
+ 1430 train 4.989841 (lr=1.9441e-04) (hash(x)=147696374)
176
+ 1440 train 4.963121 (lr=1.9430e-04) (hash(x)=150085494)
177
+ 1450 train 4.905159 (lr=1.9417e-04) (hash(x)=139296191)
178
+ 1460 train 4.839015 (lr=1.9405e-04) (hash(x)=142956546)
179
+ 1470 train 4.989805 (lr=1.9393e-04) (hash(x)=155908628)
180
+ 1480 train 4.866210 (lr=1.9381e-04) (hash(x)=150893345)
181
+ 1490 train 4.856716 (lr=1.9368e-04) (hash(x)=143548902)
182
+ 1500 val loss 5.1558
183
+ 1500 val perplexity 173.4389
184
+ 1500 train 5.187217 (lr=1.9355e-04) (hash(x)=148473774)
185
+ 1510 train 5.198489 (lr=1.9343e-04) (hash(x)=161631000)
186
+ 1520 train 5.177841 (lr=1.9330e-04) (hash(x)=166217855)
187
+ 1530 train 5.062715 (lr=1.9317e-04) (hash(x)=149440046)
188
+ 1540 train 5.038724 (lr=1.9303e-04) (hash(x)=155013278)
189
+ 1550 train 5.006696 (lr=1.9290e-04) (hash(x)=154130527)
190
+ 1560 train 5.089114 (lr=1.9277e-04) (hash(x)=152138735)
191
+ 1570 train 5.010033 (lr=1.9263e-04) (hash(x)=148128305)
192
+ 1580 train 5.151905 (lr=1.9250e-04) (hash(x)=157176546)
193
+ 1590 train 5.207416 (lr=1.9236e-04) (hash(x)=152935467)
194
+ 1600 val loss 5.0873
195
+ 1600 val perplexity 161.9577
196
+ 1600 train 5.087369 (lr=1.9222e-04) (hash(x)=151117002)
197
+ 1610 train 5.019836 (lr=1.9208e-04) (hash(x)=149883384)
198
+ 1620 train 5.124061 (lr=1.9194e-04) (hash(x)=145015354)
199
+ 1630 train 4.990678 (lr=1.9180e-04) (hash(x)=146717645)
200
+ 1640 train 5.225142 (lr=1.9165e-04) (hash(x)=156372727)
201
+ 1650 train 5.092177 (lr=1.9151e-04) (hash(x)=140261555)
202
+ 1660 train 4.979393 (lr=1.9136e-04) (hash(x)=150088650)
203
+ 1670 train 4.854083 (lr=1.9121e-04) (hash(x)=135943513)
204
+ 1680 train 5.097488 (lr=1.9107e-04) (hash(x)=153529653)
205
+ 1690 train 5.129160 (lr=1.9092e-04) (hash(x)=148930360)
206
+ 1700 val loss 5.0606
207
+ 1700 val perplexity 157.6870
208
+ 1700 train 5.053535 (lr=1.9077e-04) (hash(x)=138011335)
209
+ 1710 train 4.981909 (lr=1.9061e-04) (hash(x)=141383599)
210
+ 1720 train 5.118598 (lr=1.9046e-04) (hash(x)=161449652)
211
+ 1730 train 4.959565 (lr=1.9031e-04) (hash(x)=153923561)
212
+ 1740 train 4.894742 (lr=1.9015e-04) (hash(x)=146043253)
213
+ 1750 train 4.947501 (lr=1.9000e-04) (hash(x)=155720883)
214
+ 1760 train 4.878337 (lr=1.8984e-04) (hash(x)=148222148)
215
+ 1770 train 5.008194 (lr=1.8968e-04) (hash(x)=155712380)
216
+ 1780 train 4.827558 (lr=1.8952e-04) (hash(x)=149767859)
217
+ 1790 train 5.032566 (lr=1.8936e-04) (hash(x)=167214230)
218
+ 1800 val loss 5.0072
219
+ 1800 val perplexity 149.4821
220
+ 1800 train 5.034396 (lr=1.8920e-04) (hash(x)=171180926)
221
+ 1810 train 4.992330 (lr=1.8903e-04) (hash(x)=162375834)
222
+ 1820 train 5.120585 (lr=1.8887e-04) (hash(x)=151495752)
223
+ 1830 train 5.002681 (lr=1.8870e-04) (hash(x)=146532257)
224
+ 1840 train 5.052368 (lr=1.8854e-04) (hash(x)=145694434)
225
+ 1850 train 4.977033 (lr=1.8837e-04) (hash(x)=144823488)
226
+ 1860 train 5.054308 (lr=1.8820e-04) (hash(x)=149836286)
227
+ 1870 train 5.007481 (lr=1.8803e-04) (hash(x)=143935049)
228
+ 1880 train 5.040244 (lr=1.8786e-04) (hash(x)=151786371)
229
+ 1890 train 4.963921 (lr=1.8768e-04) (hash(x)=152892567)
230
+ 1900 val loss 4.9548
231
+ 1900 val perplexity 141.8484
232
+ 1900 train 4.968696 (lr=1.8751e-04) (hash(x)=141769419)
233
+ 1910 train 5.009535 (lr=1.8734e-04) (hash(x)=156914416)
234
+ 1920 train 4.809900 (lr=1.8716e-04) (hash(x)=151573900)
235
+ 1930 train 4.933997 (lr=1.8698e-04) (hash(x)=145121701)
236
+ 1940 train 4.873201 (lr=1.8680e-04) (hash(x)=150896348)
237
+ 1950 train 5.034765 (lr=1.8663e-04) (hash(x)=168797189)
238
+ 1960 train 4.859721 (lr=1.8645e-04) (hash(x)=146711430)
239
+ 1970 train 4.726253 (lr=1.8626e-04) (hash(x)=152180946)
240
+ 1980 train 4.846286 (lr=1.8608e-04) (hash(x)=155542268)
241
+ 1990 train 4.810067 (lr=1.8590e-04) (hash(x)=144668660)
242
+ 2000 val loss 4.8965
243
+ 2000 val perplexity 133.8258
244
+ 2000 train 4.896779 (lr=1.8571e-04) (hash(x)=151963443)
245
+ 2010 train 4.878860 (lr=1.8553e-04) (hash(x)=150810352)
246
+ 2020 train 4.888221 (lr=1.8534e-04) (hash(x)=157539832)
247
+ 2030 train 5.029580 (lr=1.8515e-04) (hash(x)=137858786)
248
+ 2040 train 4.800919 (lr=1.8496e-04) (hash(x)=154270900)
249
+ 2050 train 4.654102 (lr=1.8477e-04) (hash(x)=156252579)
250
+ 2060 train 4.592618 (lr=1.8458e-04) (hash(x)=143285549)
251
+ 2070 train 4.812717 (lr=1.8439e-04) (hash(x)=150542438)
252
+ 2080 train 4.767097 (lr=1.8420e-04) (hash(x)=146290330)
253
+ 2090 train 4.784722 (lr=1.8400e-04) (hash(x)=146486756)
254
+ 2100 val loss 4.8826
255
+ 2100 val perplexity 131.9714
256
+ 2100 train 4.854304 (lr=1.8381e-04) (hash(x)=162947470)
257
+ 2110 train 4.727439 (lr=1.8361e-04) (hash(x)=150305391)
258
+ 2120 train 4.870378 (lr=1.8341e-04) (hash(x)=156452783)
259
+ 2130 train 4.873018 (lr=1.8321e-04) (hash(x)=167481875)
260
+ 2140 train 4.824937 (lr=1.8301e-04) (hash(x)=141782256)
261
+ 2150 train 4.821207 (lr=1.8281e-04) (hash(x)=145035419)
262
+ 2160 train 4.900690 (lr=1.8261e-04) (hash(x)=142000400)
263
+ 2170 train 4.861797 (lr=1.8241e-04) (hash(x)=144229380)
264
+ 2180 train 4.902711 (lr=1.8220e-04) (hash(x)=166204176)
265
+ 2190 train 4.838762 (lr=1.8200e-04) (hash(x)=150712057)
266
+ 2200 val loss 4.8088
267
+ 2200 val perplexity 122.5886
268
+ 2200 train 4.753928 (lr=1.8179e-04) (hash(x)=154954810)
269
+ 2210 train 4.917011 (lr=1.8158e-04) (hash(x)=155668760)
270
+ 2220 train 4.711287 (lr=1.8138e-04) (hash(x)=144834708)
271
+ 2230 train 4.761308 (lr=1.8117e-04) (hash(x)=159744566)
272
+ 2240 train 4.702114 (lr=1.8096e-04) (hash(x)=158543377)
273
+ 2250 train 4.698808 (lr=1.8074e-04) (hash(x)=151004281)
274
+ 2260 train 4.802685 (lr=1.8053e-04) (hash(x)=151242482)
275
+ 2270 train 4.753480 (lr=1.8032e-04) (hash(x)=152056559)
276
+ 2280 train 4.680284 (lr=1.8010e-04) (hash(x)=148864076)
277
+ 2290 train 4.784821 (lr=1.7989e-04) (hash(x)=144294586)
278
+ 2300 val loss 4.7792
279
+ 2300 val perplexity 119.0150
280
+ 2300 train 4.726087 (lr=1.7967e-04) (hash(x)=151878111)
281
+ 2310 train 4.703807 (lr=1.7946e-04) (hash(x)=154139457)
282
+ 2320 train 4.633800 (lr=1.7924e-04) (hash(x)=155571592)
283
+ 2330 train 4.674031 (lr=1.7902e-04) (hash(x)=154823764)
284
+ 2340 train 4.742133 (lr=1.7880e-04) (hash(x)=153252177)
285
+ 2350 train 4.659034 (lr=1.7858e-04) (hash(x)=136051830)
286
+ 2360 train 4.712145 (lr=1.7835e-04) (hash(x)=153814175)
287
+ 2370 train 4.734082 (lr=1.7813e-04) (hash(x)=146391078)
288
+ 2380 train 4.688128 (lr=1.7791e-04) (hash(x)=150782698)
289
+ 2390 train 4.738202 (lr=1.7768e-04) (hash(x)=142752256)
290
+ 2400 val loss 4.7677
291
+ 2400 val perplexity 117.6490
292
+ 2400 train 4.636341 (lr=1.7745e-04) (hash(x)=158661057)
293
+ 2410 train 4.697392 (lr=1.7723e-04) (hash(x)=153611303)
294
+ 2420 train 4.860507 (lr=1.7700e-04) (hash(x)=159508684)
295
+ 2430 train 4.668229 (lr=1.7677e-04) (hash(x)=149501192)
296
+ 2440 train 4.340519 (lr=1.7654e-04) (hash(x)=181177177)
297
+ 2450 train 4.824966 (lr=1.7631e-04) (hash(x)=163664732)
298
+ 2460 train 4.619841 (lr=1.7607e-04) (hash(x)=194527955)
299
+ 2470 train 4.740547 (lr=1.7584e-04) (hash(x)=158088521)
300
+ 2480 train 4.685566 (lr=1.7561e-04) (hash(x)=152865678)
301
+ 2490 train 4.683333 (lr=1.7537e-04) (hash(x)=165703871)
302
+ 2500 val loss 4.7055
303
+ 2500 val perplexity 110.5492
304
+ 2500 train 4.751377 (lr=1.7514e-04) (hash(x)=150925584)
305
+ 2510 train 4.655087 (lr=1.7490e-04) (hash(x)=144406177)
306
+ 2520 train 4.495244 (lr=1.7466e-04) (hash(x)=120133043)
307
+ 2530 train 4.752213 (lr=1.7442e-04) (hash(x)=167900247)
308
+ 2540 train 4.641245 (lr=1.7418e-04) (hash(x)=162599203)
309
+ 2550 train 4.655796 (lr=1.7394e-04) (hash(x)=149328180)
310
+ 2560 train 4.590129 (lr=1.7370e-04) (hash(x)=150049968)
311
+ 2570 train 4.710280 (lr=1.7346e-04) (hash(x)=162876159)
312
+ 2580 train 4.548274 (lr=1.7321e-04) (hash(x)=148260649)
313
+ 2590 train 4.670132 (lr=1.7297e-04) (hash(x)=148998018)
314
+ 2600 val loss 4.6820
315
+ 2600 val perplexity 107.9882
316
+ 2600 train 4.610139 (lr=1.7272e-04) (hash(x)=144515755)
317
+ 2610 train 4.724717 (lr=1.7248e-04) (hash(x)=154455224)
318
+ 2620 train 4.556958 (lr=1.7223e-04) (hash(x)=142920048)
319
+ 2630 train 4.667206 (lr=1.7198e-04) (hash(x)=143931236)
320
+ 2640 train 4.523838 (lr=1.7173e-04) (hash(x)=150983323)
321
+ 2650 train 4.609798 (lr=1.7148e-04) (hash(x)=139360405)
322
+ 2660 train 4.538650 (lr=1.7123e-04) (hash(x)=148850470)
323
+ 2670 train 4.777700 (lr=1.7098e-04) (hash(x)=149929606)
324
+ 2680 train 4.745118 (lr=1.7073e-04) (hash(x)=151673000)
325
+ 2690 train 4.606165 (lr=1.7048e-04) (hash(x)=140631266)
326
+ 2700 val loss 4.6652
327
+ 2700 val perplexity 106.1821
328
+ 2700 train 4.608135 (lr=1.7022e-04) (hash(x)=153109144)
329
+ 2710 train 4.466459 (lr=1.6997e-04) (hash(x)=135332836)
330
+ 2720 train 4.825546 (lr=1.6971e-04) (hash(x)=131611182)
331
+ 2730 train 4.679884 (lr=1.6945e-04) (hash(x)=142909388)
332
+ 2740 train 4.697910 (lr=1.6920e-04) (hash(x)=145593419)
333
+ 2750 train 4.599571 (lr=1.6894e-04) (hash(x)=158225769)
334
+ 2760 train 4.617232 (lr=1.6868e-04) (hash(x)=170509938)
335
+ 2770 train 4.590373 (lr=1.6842e-04) (hash(x)=150785805)
336
+ 2780 train 4.552293 (lr=1.6816e-04) (hash(x)=167281736)
337
+ 2790 train 4.738145 (lr=1.6789e-04) (hash(x)=161660187)
338
+ 2800 val loss 4.6272
339
+ 2800 val perplexity 102.2262
340
+ 2800 train 4.644163 (lr=1.6763e-04) (hash(x)=151152897)
341
+ 2810 train 4.616184 (lr=1.6737e-04) (hash(x)=143172067)
342
+ 2820 train 4.629508 (lr=1.6710e-04) (hash(x)=141569169)
343
+ 2830 train 4.511320 (lr=1.6684e-04) (hash(x)=159709681)
344
+ 2840 train 4.600198 (lr=1.6657e-04) (hash(x)=162552294)
345
+ 2850 train 4.682267 (lr=1.6631e-04) (hash(x)=146762690)
346
+ 2860 train 4.564099 (lr=1.6604e-04) (hash(x)=145729422)
347
+ 2870 train 4.618160 (lr=1.6577e-04) (hash(x)=144360628)
348
+ 2880 train 4.443743 (lr=1.6550e-04) (hash(x)=147737098)
349
+ 2890 train 4.556001 (lr=1.6523e-04) (hash(x)=146849057)
350
+ 2900 val loss 4.6070
351
+ 2900 val perplexity 100.1817
352
+ 2900 train 4.605151 (lr=1.6496e-04) (hash(x)=145800210)
353
+ 2910 train 4.423512 (lr=1.6469e-04) (hash(x)=147316313)
354
+ 2920 train 4.475365 (lr=1.6441e-04) (hash(x)=152595415)
355
+ 2930 train 4.635031 (lr=1.6414e-04) (hash(x)=156542647)
356
+ 2940 train 4.486027 (lr=1.6387e-04) (hash(x)=141183896)
357
+ 2950 train 4.603455 (lr=1.6359e-04) (hash(x)=156329036)
358
+ 2960 train 4.626204 (lr=1.6332e-04) (hash(x)=156717793)
359
+ 2970 train 4.476183 (lr=1.6304e-04) (hash(x)=143949931)
360
+ 2980 train 4.571628 (lr=1.6276e-04) (hash(x)=155067290)
361
+ 2990 train 4.515110 (lr=1.6248e-04) (hash(x)=148774729)
362
+ 3000 val loss 4.6067
363
+ 3000 val perplexity 100.1577
364
+ 3000 train 4.494736 (lr=1.6221e-04) (hash(x)=141997485)
365
+ 3010 train 4.472013 (lr=1.6193e-04) (hash(x)=160762953)
366
+ 3020 train 4.468177 (lr=1.6165e-04) (hash(x)=170341665)
367
+ 3030 train 4.708908 (lr=1.6136e-04) (hash(x)=150722836)
368
+ 3040 train 4.508104 (lr=1.6108e-04) (hash(x)=154243307)
369
+ 3050 train 4.535357 (lr=1.6080e-04) (hash(x)=146077234)
370
+ 3060 train 4.502595 (lr=1.6052e-04) (hash(x)=152125542)
371
+ 3070 train 4.676445 (lr=1.6023e-04) (hash(x)=144149387)
372
+ 3080 train 4.518022 (lr=1.5995e-04) (hash(x)=147508712)
373
+ 3090 train 4.504861 (lr=1.5966e-04) (hash(x)=143583369)
374
+ 3100 val loss 4.5625
375
+ 3100 val perplexity 95.8224
376
+ 3100 train 4.524681 (lr=1.5938e-04) (hash(x)=154049740)
377
+ 3110 train 4.599205 (lr=1.5909e-04) (hash(x)=147054835)
378
+ 3120 train 4.590798 (lr=1.5880e-04) (hash(x)=152468568)
379
+ 3130 train 4.520497 (lr=1.5851e-04) (hash(x)=151635194)
380
+ 3140 train 4.498206 (lr=1.5822e-04) (hash(x)=150396274)
381
+ 3150 train 4.512784 (lr=1.5793e-04) (hash(x)=141306267)
382
+ 3160 train 4.480770 (lr=1.5764e-04) (hash(x)=151110762)
383
+ 3170 train 4.555854 (lr=1.5735e-04) (hash(x)=146518875)
384
+ 3180 train 4.447691 (lr=1.5706e-04) (hash(x)=139914401)
385
+ 3190 train 4.458163 (lr=1.5677e-04) (hash(x)=161631880)
386
+ 3200 val loss 4.5678
387
+ 3200 val perplexity 96.3325
388
+ 3200 train 4.464018 (lr=1.5648e-04) (hash(x)=150471842)
389
+ 3210 train 4.544499 (lr=1.5618e-04) (hash(x)=147355138)
390
+ 3220 train 4.327750 (lr=1.5589e-04) (hash(x)=138879678)
391
+ 3230 train 4.413097 (lr=1.5559e-04) (hash(x)=123929679)
392
+ 3240 train 4.517000 (lr=1.5530e-04) (hash(x)=157263734)
393
+ 3250 train 4.509780 (lr=1.5500e-04) (hash(x)=148065539)
394
+ 3260 train 4.433766 (lr=1.5470e-04) (hash(x)=138020742)
395
+ 3270 train 4.299691 (lr=1.5441e-04) (hash(x)=142053611)
396
+ 3280 train 4.452711 (lr=1.5411e-04) (hash(x)=155444784)
397
+ 3290 train 4.267416 (lr=1.5381e-04) (hash(x)=148500315)
398
+ 3300 val loss 4.5479
399
+ 3300 val perplexity 94.4349
400
+ 3300 train 4.504673 (lr=1.5351e-04) (hash(x)=149048126)
401
+ 3310 train 4.459144 (lr=1.5321e-04) (hash(x)=141680674)
402
+ 3320 train 4.397059 (lr=1.5291e-04) (hash(x)=143310642)
403
+ 3330 train 4.514332 (lr=1.5261e-04) (hash(x)=149891496)
404
+ 3340 train 4.474998 (lr=1.5230e-04) (hash(x)=151244609)
405
+ 3350 train 4.485167 (lr=1.5200e-04) (hash(x)=157158479)
406
+ 3360 train 4.353802 (lr=1.5170e-04) (hash(x)=151709876)
407
+ 3370 train 4.487107 (lr=1.5139e-04) (hash(x)=145297090)
408
+ 3380 train 4.499591 (lr=1.5109e-04) (hash(x)=154767881)
409
+ 3390 train 4.483547 (lr=1.5078e-04) (hash(x)=152600366)
410
+ 3400 val loss 4.5264
411
+ 3400 val perplexity 92.4273
412
+ 3400 train 4.699473 (lr=1.5048e-04) (hash(x)=161261339)
413
+ 3410 train 4.450081 (lr=1.5017e-04) (hash(x)=146859162)
414
+ 3420 train 4.540144 (lr=1.4986e-04) (hash(x)=149276224)
415
+ 3430 train 4.581402 (lr=1.4956e-04) (hash(x)=155866933)
416
+ 3440 train 4.415562 (lr=1.4925e-04) (hash(x)=143191904)
417
+ 3450 train 4.645755 (lr=1.4894e-04) (hash(x)=156910661)
418
+ 3460 train 4.412221 (lr=1.4863e-04) (hash(x)=140850879)
419
+ 3470 train 4.655238 (lr=1.4832e-04) (hash(x)=162140445)
420
+ 3480 train 4.400720 (lr=1.4801e-04) (hash(x)=145607495)
421
+ 3490 train 4.413011 (lr=1.4770e-04) (hash(x)=152496121)
422
+ 3500 val loss 4.5130
423
+ 3500 val perplexity 91.1984
424
+ 3500 train 4.386954 (lr=1.4739e-04) (hash(x)=157495564)
425
+ 3510 train 4.499444 (lr=1.4708e-04) (hash(x)=147657004)
426
+ 3520 train 4.468650 (lr=1.4676e-04) (hash(x)=152978060)
427
+ 3530 train 4.407806 (lr=1.4645e-04) (hash(x)=146800180)
428
+ 3540 train 4.507030 (lr=1.4614e-04) (hash(x)=195450444)
429
+ 3550 train 4.449342 (lr=1.4582e-04) (hash(x)=151114129)
430
+ 3560 train 4.439520 (lr=1.4551e-04) (hash(x)=143833190)
431
+ 3570 train 4.252267 (lr=1.4519e-04) (hash(x)=146039971)
432
+ 3580 train 4.321498 (lr=1.4488e-04) (hash(x)=145172249)
433
+ 3590 train 4.241528 (lr=1.4456e-04) (hash(x)=159164125)
434
+ 3600 val loss 4.5043
435
+ 3600 val perplexity 90.4028
436
+ 3600 train 4.475197 (lr=1.4424e-04) (hash(x)=144352932)
437
+ 3610 train 4.468740 (lr=1.4393e-04) (hash(x)=150550670)
438
+ 3620 train 4.567307 (lr=1.4361e-04) (hash(x)=147170125)
439
+ 3630 train 4.356416 (lr=1.4329e-04) (hash(x)=139473935)
440
+ 3640 train 4.386318 (lr=1.4297e-04) (hash(x)=159294424)
441
+ 3650 train 4.501210 (lr=1.4265e-04) (hash(x)=145943278)
442
+ 3660 train 4.524951 (lr=1.4233e-04) (hash(x)=141647872)
443
+ 3670 train 4.498155 (lr=1.4201e-04) (hash(x)=147109351)
444
+ 3680 train 4.523514 (lr=1.4169e-04) (hash(x)=140683368)
445
+ 3690 train 4.450881 (lr=1.4137e-04) (hash(x)=156276284)
446
+ 3700 val loss 4.4844
447
+ 3700 val perplexity 88.6258
448
+ 3700 train 4.411953 (lr=1.4105e-04) (hash(x)=149389012)
449
+ 3710 train 4.512041 (lr=1.4073e-04) (hash(x)=142107929)
450
+ 3720 train 4.507792 (lr=1.4041e-04) (hash(x)=147828843)
451
+ 3730 train 4.583274 (lr=1.4008e-04) (hash(x)=141784679)
452
+ 3740 train 4.454018 (lr=1.3976e-04) (hash(x)=153375464)
453
+ 3750 train 4.405389 (lr=1.3944e-04) (hash(x)=153716001)
454
+ 3760 train 4.374485 (lr=1.3911e-04) (hash(x)=146848822)
455
+ 3770 train 4.321904 (lr=1.3879e-04) (hash(x)=148396782)
456
+ 3780 train 4.389256 (lr=1.3846e-04) (hash(x)=153563244)
457
+ 3790 train 4.311008 (lr=1.3814e-04) (hash(x)=160808734)
458
+ 3800 val loss 4.4708
459
+ 3800 val perplexity 87.4271
460
+ 3800 train 4.473656 (lr=1.3781e-04) (hash(x)=146607620)
461
+ 3810 train 4.462682 (lr=1.3749e-04) (hash(x)=152580677)
462
+ 3820 train 4.774294 (lr=1.3716e-04) (hash(x)=144860879)
463
+ 3830 train 4.474754 (lr=1.3683e-04) (hash(x)=162099851)
464
+ 3840 train 4.429060 (lr=1.3650e-04) (hash(x)=159506855)
465
+ 3850 train 4.339765 (lr=1.3618e-04) (hash(x)=140022686)
466
+ 3860 train 4.440007 (lr=1.3585e-04) (hash(x)=152137850)
467
+ 3870 train 4.523821 (lr=1.3552e-04) (hash(x)=142666785)
468
+ 3880 train 4.406719 (lr=1.3519e-04) (hash(x)=148285868)
469
+ 3890 train 4.374085 (lr=1.3486e-04) (hash(x)=147910067)
470
+ 3900 val loss 4.4646
471
+ 3900 val perplexity 86.8851
472
+ 3900 train 4.400462 (lr=1.3453e-04) (hash(x)=143754617)
473
+ 3910 train 4.403353 (lr=1.3420e-04) (hash(x)=154779949)
474
+ 3920 train 4.231364 (lr=1.3387e-04) (hash(x)=137824061)
475
+ 3930 train 4.194984 (lr=1.3354e-04) (hash(x)=136933373)
476
+ 3940 train 4.276679 (lr=1.3321e-04) (hash(x)=164054337)
477
+ 3950 train 4.564040 (lr=1.3288e-04) (hash(x)=155752984)
478
+ 3960 train 4.607040 (lr=1.3255e-04) (hash(x)=160957673)
479
+ 3970 train 4.474471 (lr=1.3222e-04) (hash(x)=157158690)
480
+ 3980 train 4.479353 (lr=1.3188e-04) (hash(x)=153282027)
481
+ 3990 train 4.506903 (lr=1.3155e-04) (hash(x)=152183510)
482
+ 4000 val loss 4.4415
483
+ 4000 val perplexity 84.8986
484
+ 4000 train 4.376513 (lr=1.3122e-04) (hash(x)=156930722)
485
+ 4010 train 4.380332 (lr=1.3089e-04) (hash(x)=150798740)
486
+ 4020 train 4.575028 (lr=1.3055e-04) (hash(x)=142632400)
487
+ 4030 train 4.500770 (lr=1.3022e-04) (hash(x)=173326287)
488
+ 4040 train 4.385245 (lr=1.2988e-04) (hash(x)=128797443)
489
+ 4050 train 4.375233 (lr=1.2955e-04) (hash(x)=146937792)
490
+ 4060 train 4.493315 (lr=1.2921e-04) (hash(x)=164579030)
491
+ 4070 train 4.362973 (lr=1.2888e-04) (hash(x)=149459820)
492
+ 4080 train 4.327494 (lr=1.2854e-04) (hash(x)=160159595)
493
+ 4090 train 4.260346 (lr=1.2821e-04) (hash(x)=141507853)
494
+ 4100 val loss 4.4336
495
+ 4100 val perplexity 84.2360
496
+ 4100 train 4.380561 (lr=1.2787e-04) (hash(x)=147500519)
497
+ 4110 train 4.375722 (lr=1.2754e-04) (hash(x)=153297081)
498
+ 4120 train 4.377880 (lr=1.2720e-04) (hash(x)=142560288)
499
+ 4130 train 4.409745 (lr=1.2686e-04) (hash(x)=153380118)
500
+ 4140 train 4.306442 (lr=1.2653e-04) (hash(x)=143150720)
501
+ 4150 train 4.351232 (lr=1.2619e-04) (hash(x)=143459564)
502
+ 4160 train 4.372649 (lr=1.2585e-04) (hash(x)=153955741)
503
+ 4170 train 4.479025 (lr=1.2552e-04) (hash(x)=152199253)
504
+ 4180 train 4.259434 (lr=1.2518e-04) (hash(x)=165957668)
505
+ 4190 train 4.295711 (lr=1.2484e-04) (hash(x)=144866923)
506
+ 4200 val loss 4.4368
507
+ 4200 val perplexity 84.5080
508
+ 4200 train 4.296550 (lr=1.2450e-04) (hash(x)=143232237)
509
+ 4210 train 4.216445 (lr=1.2416e-04) (hash(x)=151907163)
510
+ 4220 train 4.330945 (lr=1.2383e-04) (hash(x)=145917587)
511
+ 4230 train 4.143333 (lr=1.2349e-04) (hash(x)=147258846)
512
+ 4240 train 4.299682 (lr=1.2315e-04) (hash(x)=143449843)
513
+ 4250 train 4.271683 (lr=1.2281e-04) (hash(x)=184883715)
514
+ 4260 train 4.363529 (lr=1.2247e-04) (hash(x)=156958761)
515
+ 4270 train 4.454966 (lr=1.2213e-04) (hash(x)=168870544)
516
+ 4280 train 4.376136 (lr=1.2179e-04) (hash(x)=179446735)
517
+ 4290 train 4.394639 (lr=1.2145e-04) (hash(x)=145649386)
518
+ 4300 val loss 4.4171
519
+ 4300 val perplexity 82.8588
520
+ 4300 train 4.254510 (lr=1.2111e-04) (hash(x)=146811670)
521
+ 4310 train 4.520589 (lr=1.2077e-04) (hash(x)=151907075)
522
+ 4320 train 4.335262 (lr=1.2043e-04) (hash(x)=138628365)
523
+ 4330 train 4.389732 (lr=1.2009e-04) (hash(x)=151608240)
524
+ 4340 train 4.301316 (lr=1.1975e-04) (hash(x)=160111319)
525
+ 4350 train 4.451436 (lr=1.1941e-04) (hash(x)=144958652)
526
+ 4360 train 4.255314 (lr=1.1907e-04) (hash(x)=149176222)
527
+ 4370 train 4.311791 (lr=1.1873e-04) (hash(x)=216221371)
528
+ 4380 train 4.254924 (lr=1.1838e-04) (hash(x)=141760648)
529
+ 4390 train 4.333160 (lr=1.1804e-04) (hash(x)=151710283)
530
+ 4400 val loss 4.4064
531
+ 4400 val perplexity 81.9706
532
+ 4400 train 4.459096 (lr=1.1770e-04) (hash(x)=158418746)
533
+ 4410 train 4.306726 (lr=1.1736e-04) (hash(x)=142331301)
534
+ 4420 train 4.389251 (lr=1.1702e-04) (hash(x)=146507180)
535
+ 4430 train 4.487090 (lr=1.1668e-04) (hash(x)=166506937)
536
+ 4440 train 4.420139 (lr=1.1634e-04) (hash(x)=151209447)
537
+ 4450 train 4.392083 (lr=1.1599e-04) (hash(x)=149186822)
538
+ 4460 train 4.358459 (lr=1.1565e-04) (hash(x)=144348412)
539
+ 4470 train 4.265626 (lr=1.1531e-04) (hash(x)=156494284)
540
+ 4480 train 4.299117 (lr=1.1497e-04) (hash(x)=147238744)
541
+ 4490 train 4.239520 (lr=1.1462e-04) (hash(x)=150746185)
542
+ 4500 val loss 4.4016
543
+ 4500 val perplexity 81.5782
544
+ 4500 train 4.338185 (lr=1.1428e-04) (hash(x)=156695778)
545
+ 4510 train 4.168920 (lr=1.1394e-04) (hash(x)=135988958)
546
+ 4520 train 4.337834 (lr=1.1360e-04) (hash(x)=148272110)
547
+ 4530 train 4.256871 (lr=1.1326e-04) (hash(x)=136459794)
548
+ 4540 train 4.302117 (lr=1.1291e-04) (hash(x)=147624954)
549
+ 4550 train 4.299829 (lr=1.1257e-04) (hash(x)=154947199)
550
+ 4560 train 4.344871 (lr=1.1223e-04) (hash(x)=147869166)
551
+ 4570 train 4.217010 (lr=1.1188e-04) (hash(x)=148415463)
552
+ 4580 train 4.381991 (lr=1.1154e-04) (hash(x)=152104066)
553
+ 4590 train 4.273722 (lr=1.1120e-04) (hash(x)=147060040)
554
+ 4600 val loss 4.3944
555
+ 4600 val perplexity 80.9945
556
+ 4600 train 4.348307 (lr=1.1086e-04) (hash(x)=147791497)
557
+ 4610 train 4.536617 (lr=1.1051e-04) (hash(x)=155595556)
558
+ 4620 train 4.387483 (lr=1.1017e-04) (hash(x)=147178858)
559
+ 4630 train 4.434367 (lr=1.0983e-04) (hash(x)=152617112)
560
+ 4640 train 4.449227 (lr=1.0949e-04) (hash(x)=161766869)
561
+ 4650 train 4.444243 (lr=1.0914e-04) (hash(x)=158075561)
562
+ 4660 train 4.450589 (lr=1.0880e-04) (hash(x)=145360002)
563
+ 4670 train 4.341477 (lr=1.0846e-04) (hash(x)=144538791)
564
+ 4680 train 4.346433 (lr=1.0812e-04) (hash(x)=142711273)
565
+ 4690 train 4.563562 (lr=1.0777e-04) (hash(x)=193907362)
566
+ 4700 val loss 4.3761
567
+ 4700 val perplexity 79.5294
568
+ 4700 train 4.337259 (lr=1.0743e-04) (hash(x)=155533088)
569
+ 4710 train 4.290280 (lr=1.0709e-04) (hash(x)=148618528)
570
+ 4720 train 4.405389 (lr=1.0674e-04) (hash(x)=150502624)
571
+ 4730 train 4.299401 (lr=1.0640e-04) (hash(x)=189205847)
572
+ 4740 train 4.345629 (lr=1.0606e-04) (hash(x)=167322800)
573
+ 4750 train 4.453078 (lr=1.0572e-04) (hash(x)=162169095)
574
+ 4760 train 4.158504 (lr=1.0538e-04) (hash(x)=143459469)
575
+ 4770 train 4.236567 (lr=1.0503e-04) (hash(x)=144386593)
576
+ 4780 train 4.195615 (lr=1.0469e-04) (hash(x)=143721184)
577
+ 4790 train 4.280142 (lr=1.0435e-04) (hash(x)=146077222)
578
+ 4800 val loss 4.3766
579
+ 4800 val perplexity 79.5687
580
+ 4800 train 4.244648 (lr=1.0401e-04) (hash(x)=138350044)
581
+ 4810 train 4.248234 (lr=1.0366e-04) (hash(x)=154960112)
582
+ 4820 train 4.153278 (lr=1.0332e-04) (hash(x)=146950685)
583
+ 4830 train 4.164750 (lr=1.0298e-04) (hash(x)=140224554)
584
+ 4840 train 4.160818 (lr=1.0264e-04) (hash(x)=156640495)
585
+ 4850 train 4.360028 (lr=1.0230e-04) (hash(x)=153409150)
586
+ 4860 train 4.289172 (lr=1.0196e-04) (hash(x)=151647700)
587
+ 4870 train 4.685504 (lr=1.0162e-04) (hash(x)=155465603)
588
+ 4880 train 4.520584 (lr=1.0127e-04) (hash(x)=153404400)
589
+ 4890 train 4.429728 (lr=1.0093e-04) (hash(x)=170409281)
590
+ 4900 val loss 4.3620
591
+ 4900 val perplexity 78.4139
592
+ 4900 train 4.282740 (lr=1.0059e-04) (hash(x)=143735284)
593
+ 4910 train 4.413850 (lr=1.0025e-04) (hash(x)=150784329)
594
+ 4920 train 4.307442 (lr=9.9911e-05) (hash(x)=152353574)
595
+ 4930 train 4.267021 (lr=9.9571e-05) (hash(x)=152677008)
596
+ 4940 train 4.336789 (lr=9.9230e-05) (hash(x)=140553861)
597
+ 4950 train 4.417282 (lr=9.8890e-05) (hash(x)=152869181)
598
+ 4960 train 4.386991 (lr=9.8550e-05) (hash(x)=153011079)
599
+ 4970 train 4.335236 (lr=9.8210e-05) (hash(x)=152796617)
600
+ 4980 train 4.427298 (lr=9.7870e-05) (hash(x)=146868417)
601
+ 4990 train 4.309089 (lr=9.7531e-05) (hash(x)=140168188)
602
+ 5000 val loss 4.3515
603
+ 5000 val perplexity 77.5964
604
+ 5000 train 4.302033 (lr=9.7192e-05) (hash(x)=154976463)
605
+ 5010 train 4.356619 (lr=9.6853e-05) (hash(x)=148582690)
606
+ 5020 train 4.368586 (lr=9.6514e-05) (hash(x)=166715464)
607
+ 5030 train 4.327526 (lr=9.6175e-05) (hash(x)=150616397)
608
+ 5040 train 4.379251 (lr=9.5836e-05) (hash(x)=150775548)
609
+ 5050 train 4.360279 (lr=9.5498e-05) (hash(x)=150575367)
610
+ 5060 train 4.290219 (lr=9.5160e-05) (hash(x)=152352082)
611
+ 5070 train 4.340550 (lr=9.4822e-05) (hash(x)=150269272)
612
+ 5080 train 4.369705 (lr=9.4484e-05) (hash(x)=146499361)
613
+ 5090 train 4.061824 (lr=9.4147e-05) (hash(x)=148734454)
614
+ 5100 val loss 4.3528
615
+ 5100 val perplexity 77.6995
616
+ 5100 train 4.256035 (lr=9.3809e-05) (hash(x)=149894982)
617
+ 5110 train 4.297043 (lr=9.3472e-05) (hash(x)=154487751)
618
+ 5120 train 4.243855 (lr=9.3136e-05) (hash(x)=144169582)
619
+ 5130 train 4.219214 (lr=9.2799e-05) (hash(x)=152217630)
620
+ 5140 train 4.288573 (lr=9.2463e-05) (hash(x)=158233831)
621
+ 5150 train 4.306287 (lr=9.2127e-05) (hash(x)=142746426)
622
+ 5160 train 4.299368 (lr=9.1791e-05) (hash(x)=139184458)
623
+ 5170 train 4.376637 (lr=9.1456e-05) (hash(x)=140624857)
624
+ 5180 train 4.118787 (lr=9.1120e-05) (hash(x)=154107423)
625
+ 5190 train 4.076274 (lr=9.0785e-05) (hash(x)=144892610)
626
+ 5200 val loss 4.3490
627
+ 5200 val perplexity 77.4032
628
+ 5200 train 4.015386 (lr=9.0451e-05) (hash(x)=159326689)
629
+ 5210 train 4.180365 (lr=9.0116e-05) (hash(x)=151186017)
630
+ 5220 train 3.994715 (lr=8.9782e-05) (hash(x)=148996819)
631
+ 5230 train 3.963748 (lr=8.9448e-05) (hash(x)=150829917)
632
+ 5240 train 4.068003 (lr=8.9115e-05) (hash(x)=152758972)
633
+ 5250 train 4.031098 (lr=8.8782e-05) (hash(x)=154345903)
634
+ 5260 train 4.028433 (lr=8.8449e-05) (hash(x)=150134947)
635
+ 5270 train 4.113491 (lr=8.8116e-05) (hash(x)=149689310)
636
+ 5280 train 4.335382 (lr=8.7784e-05) (hash(x)=147662223)
637
+ 5290 train 4.413094 (lr=8.7452e-05) (hash(x)=165770608)
638
+ 5300 val loss 4.3397
639
+ 5300 val perplexity 76.6858
640
+ 5300 train 4.384235 (lr=8.7120e-05) (hash(x)=159484800)
641
+ 5310 train 4.413448 (lr=8.6789e-05) (hash(x)=152672536)
642
+ 5320 train 4.242175 (lr=8.6458e-05) (hash(x)=152728812)
643
+ 5330 train 4.191294 (lr=8.6127e-05) (hash(x)=152778685)
644
+ 5340 train 4.289635 (lr=8.5797e-05) (hash(x)=149818849)
645
+ 5350 train 4.207667 (lr=8.5467e-05) (hash(x)=168239739)
646
+ 5360 train 4.353477 (lr=8.5138e-05) (hash(x)=139859423)
647
+ 5370 train 4.193311 (lr=8.4809e-05) (hash(x)=139315327)
648
+ 5380 train 4.384632 (lr=8.4480e-05) (hash(x)=163068509)
649
+ 5390 train 4.345930 (lr=8.4151e-05) (hash(x)=147645415)
650
+ 5400 val loss 4.3220
651
+ 5400 val perplexity 75.3362
652
+ 5400 train 4.306064 (lr=8.3823e-05) (hash(x)=140385615)
653
+ 5410 train 4.169956 (lr=8.3495e-05) (hash(x)=151381862)
654
+ 5420 train 4.334346 (lr=8.3168e-05) (hash(x)=152445092)
655
+ 5430 train 4.323414 (lr=8.2841e-05) (hash(x)=142454479)
656
+ 5440 train 4.279739 (lr=8.2515e-05) (hash(x)=151308771)
657
+ 5450 train 4.140946 (lr=8.2188e-05) (hash(x)=150886070)
658
+ 5460 train 4.239882 (lr=8.1863e-05) (hash(x)=146080408)
659
+ 5470 train 4.267916 (lr=8.1537e-05) (hash(x)=149828857)
660
+ 5480 train 4.233772 (lr=8.1212e-05) (hash(x)=139922034)
661
+ 5490 train 4.187518 (lr=8.0888e-05) (hash(x)=150262276)
662
+ 5500 val loss 4.3213
663
+ 5500 val perplexity 75.2872
664
+ 5500 train 4.160833 (lr=8.0564e-05) (hash(x)=148498335)
665
+ 5510 train 4.272779 (lr=8.0240e-05) (hash(x)=157036224)
666
+ 5520 train 4.150296 (lr=7.9917e-05) (hash(x)=134093409)
667
+ 5530 train 4.186122 (lr=7.9594e-05) (hash(x)=153126250)
668
+ 5540 train 4.180004 (lr=7.9272e-05) (hash(x)=136170985)
669
+ 5550 train 4.170107 (lr=7.8950e-05) (hash(x)=152841201)
670
+ 5560 train 4.271146 (lr=7.8629e-05) (hash(x)=147439014)
671
+ 5570 train 4.186988 (lr=7.8308e-05) (hash(x)=143476132)
672
+ 5580 train 3.966444 (lr=7.7987e-05) (hash(x)=139503437)
673
+ 5590 train 4.020775 (lr=7.7667e-05) (hash(x)=131664955)
674
+ 5600 val loss 4.3251
675
+ 5600 val perplexity 75.5734
676
+ 5600 train 4.101376 (lr=7.7347e-05) (hash(x)=151907614)
677
+ 5610 train 4.000906 (lr=7.7028e-05) (hash(x)=157132674)
678
+ 5620 train 4.193883 (lr=7.6710e-05) (hash(x)=146793775)
679
+ 5630 train 4.137687 (lr=7.6391e-05) (hash(x)=147237910)
680
+ 5640 train 4.042099 (lr=7.6074e-05) (hash(x)=141120024)
681
+ 5650 train 4.034603 (lr=7.5756e-05) (hash(x)=144275645)
682
+ 5660 train 4.076093 (lr=7.5440e-05) (hash(x)=159088002)
683
+ 5670 train 4.118015 (lr=7.5124e-05) (hash(x)=147224479)
684
+ 5680 train 4.441064 (lr=7.4808e-05) (hash(x)=147208743)
685
+ 5690 train 4.265195 (lr=7.4493e-05) (hash(x)=154384701)
686
+ 5700 val loss 4.3196
687
+ 5700 val perplexity 75.1609
688
+ 5700 train 4.298420 (lr=7.4178e-05) (hash(x)=155192267)
689
+ 5710 train 4.238029 (lr=7.3864e-05) (hash(x)=148458753)
690
+ 5720 train 4.343795 (lr=7.3550e-05) (hash(x)=159947204)
691
+ 5730 train 4.345667 (lr=7.3237e-05) (hash(x)=150746366)
692
+ 5740 train 4.323131 (lr=7.2925e-05) (hash(x)=155201884)
693
+ 5750 train 4.327418 (lr=7.2613e-05) (hash(x)=156432008)
694
+ 5760 train 4.328330 (lr=7.2301e-05) (hash(x)=185977746)
695
+ 5770 train 4.306745 (lr=7.1990e-05) (hash(x)=133620000)
696
+ 5780 train 4.316488 (lr=7.1680e-05) (hash(x)=153778044)
697
+ 5790 train 4.164621 (lr=7.1370e-05) (hash(x)=147713822)
698
+ 5800 val loss 4.3013
699
+ 5800 val perplexity 73.7942
700
+ 5800 train 4.328545 (lr=7.1061e-05) (hash(x)=153132158)
701
+ 5810 train 4.273367 (lr=7.0752e-05) (hash(x)=144293895)
702
+ 5820 train 4.245905 (lr=7.0444e-05) (hash(x)=151500590)
703
+ 5830 train 4.196019 (lr=7.0136e-05) (hash(x)=139308707)
704
+ 5840 train 4.312316 (lr=6.9829e-05) (hash(x)=153591278)
705
+ 5850 train 4.298543 (lr=6.9523e-05) (hash(x)=146077037)
706
+ 5860 train 4.255228 (lr=6.9217e-05) (hash(x)=149160733)
707
+ 5870 train 4.241347 (lr=6.8912e-05) (hash(x)=138482476)
708
+ 5880 train 4.255463 (lr=6.8607e-05) (hash(x)=152173137)
709
+ 5890 train 4.232878 (lr=6.8303e-05) (hash(x)=142387057)
710
+ 5900 val loss 4.3065
711
+ 5900 val perplexity 74.1818
712
+ 5900 train 4.309405 (lr=6.8000e-05) (hash(x)=161446764)
713
+ 5910 train 4.167171 (lr=6.7697e-05) (hash(x)=144512138)
714
+ 5920 train 4.234548 (lr=6.7395e-05) (hash(x)=152555618)
715
+ 5930 train 4.182187 (lr=6.7093e-05) (hash(x)=146129367)
716
+ 5940 train 4.132245 (lr=6.6792e-05) (hash(x)=154607527)
717
+ 5950 train 4.228748 (lr=6.6492e-05) (hash(x)=143827143)
718
+ 5960 train 4.360529 (lr=6.6192e-05) (hash(x)=151168510)
719
+ 5970 train 4.161298 (lr=6.5893e-05) (hash(x)=158961574)
720
+ 5980 train 3.982564 (lr=6.5595e-05) (hash(x)=148757787)
721
+ 5990 train 4.509342 (lr=6.5297e-05) (hash(x)=161719494)
722
+ 6000 val loss 4.3046
723
+ 6000 val perplexity 74.0387
724
+ 6000 train 3.979873 (lr=6.5000e-05) (hash(x)=151512446)
725
+ 6010 train 4.116975 (lr=6.4704e-05) (hash(x)=173500735)
726
+ 6020 train 4.224407 (lr=6.4408e-05) (hash(x)=157621031)
727
+ 6030 train 4.100847 (lr=6.4113e-05) (hash(x)=142662684)
728
+ 6040 train 4.103895 (lr=6.3818e-05) (hash(x)=146180940)
729
+ 6050 train 4.030765 (lr=6.3524e-05) (hash(x)=147654412)
730
+ 6060 train 3.916599 (lr=6.3231e-05) (hash(x)=156393157)
731
+ 6070 train 4.299768 (lr=6.2939e-05) (hash(x)=156947300)
732
+ 6080 train 4.297406 (lr=6.2647e-05) (hash(x)=163108626)
733
+ 6090 train 4.399036 (lr=6.2356e-05) (hash(x)=158654713)
734
+ 6100 val loss 4.2890
735
+ 6100 val perplexity 72.8943
736
+ 6100 train 4.145221 (lr=6.2065e-05) (hash(x)=188094053)
737
+ 6110 train 4.274182 (lr=6.1776e-05) (hash(x)=145132600)
738
+ 6120 train 4.158657 (lr=6.1487e-05) (hash(x)=146999707)
739
+ 6130 train 4.376698 (lr=6.1198e-05) (hash(x)=145719530)
740
+ 6140 train 4.318669 (lr=6.0911e-05) (hash(x)=153230722)
741
+ 6150 train 4.262127 (lr=6.0624e-05) (hash(x)=147455934)
742
+ 6160 train 4.293757 (lr=6.0338e-05) (hash(x)=146596242)
743
+ 6170 train 4.401567 (lr=6.0052e-05) (hash(x)=152029855)
744
+ 6180 train 4.263116 (lr=5.9767e-05) (hash(x)=151480034)
745
+ 6190 train 4.191931 (lr=5.9483e-05) (hash(x)=147381047)
746
+ 6200 val loss 4.2793
747
+ 6200 val perplexity 72.1883
748
+ 6200 train 4.174676 (lr=5.9200e-05) (hash(x)=149389789)
749
+ 6210 train 4.223357 (lr=5.8918e-05) (hash(x)=154894138)
750
+ 6220 train 4.265825 (lr=5.8636e-05) (hash(x)=150074302)
751
+ 6230 train 4.118061 (lr=5.8355e-05) (hash(x)=142881794)
752
+ 6240 train 4.130050 (lr=5.8074e-05) (hash(x)=164747461)
753
+ 6250 train 4.286479 (lr=5.7795e-05) (hash(x)=158378292)
754
+ 6260 train 4.283480 (lr=5.7516e-05) (hash(x)=147675938)
755
+ 6270 train 4.232567 (lr=5.7238e-05) (hash(x)=148197228)
756
+ 6280 train 4.203140 (lr=5.6961e-05) (hash(x)=135296747)
757
+ 6290 train 4.284141 (lr=5.6684e-05) (hash(x)=156439346)
758
+ 6300 val loss 4.2769
759
+ 6300 val perplexity 72.0167
760
+ 6300 train 4.087555 (lr=5.6409e-05) (hash(x)=138212820)
761
+ 6310 train 4.430627 (lr=5.6134e-05) (hash(x)=165985336)
762
+ 6320 train 4.256569 (lr=5.5859e-05) (hash(x)=145921897)
763
+ 6330 train 4.463515 (lr=5.5586e-05) (hash(x)=145502689)
764
+ 6340 train 4.190090 (lr=5.5313e-05) (hash(x)=141704937)
765
+ 6350 train 4.266877 (lr=5.5042e-05) (hash(x)=161789400)
766
+ 6360 train 3.980389 (lr=5.4771e-05) (hash(x)=145457780)
767
+ 6370 train 4.248503 (lr=5.4500e-05) (hash(x)=171628530)
768
+ 6380 train 4.035338 (lr=5.4231e-05) (hash(x)=153467923)
769
+ 6390 train 4.399091 (lr=5.3963e-05) (hash(x)=158066904)
770
+ 6400 val loss 4.2793
771
+ 6400 val perplexity 72.1874
772
+ 6400 train 4.052232 (lr=5.3695e-05) (hash(x)=146535423)
773
+ 6410 train 3.916746 (lr=5.3428e-05) (hash(x)=147264155)
774
+ 6420 train 4.018277 (lr=5.3162e-05) (hash(x)=145595485)
775
+ 6430 train 4.159811 (lr=5.2896e-05) (hash(x)=146929074)
776
+ 6440 train 4.081059 (lr=5.2632e-05) (hash(x)=147290878)
777
+ 6450 train 4.190943 (lr=5.2368e-05) (hash(x)=141154205)
778
+ 6460 train 4.038716 (lr=5.2105e-05) (hash(x)=154360901)
779
+ 6470 train 4.354444 (lr=5.1843e-05) (hash(x)=146350558)
780
+ 6480 train 4.328847 (lr=5.1582e-05) (hash(x)=153927683)
781
+ 6490 train 4.237471 (lr=5.1322e-05) (hash(x)=149337047)
782
+ 6500 val loss 4.2691
783
+ 6500 val perplexity 71.4602
784
+ 6500 train 4.150597 (lr=5.1063e-05) (hash(x)=145950843)
785
+ 6510 train 4.225127 (lr=5.0804e-05) (hash(x)=142577653)
786
+ 6520 train 4.290921 (lr=5.0546e-05) (hash(x)=165060689)
787
+ 6530 train 4.123631 (lr=5.0289e-05) (hash(x)=148367505)
788
+ 6540 train 4.228392 (lr=5.0033e-05) (hash(x)=154313651)
789
+ 6550 train 4.147989 (lr=4.9778e-05) (hash(x)=147477315)
790
+ 6560 train 4.223585 (lr=4.9524e-05) (hash(x)=139825403)
791
+ 6570 train 4.234167 (lr=4.9271e-05) (hash(x)=141482518)
792
+ 6580 train 4.282204 (lr=4.9018e-05) (hash(x)=144968144)
793
+ 6590 train 4.222837 (lr=4.8767e-05) (hash(x)=154319720)
794
+ 6600 val loss 4.2667
795
+ 6600 val perplexity 71.2895
796
+ 6600 train 4.261629 (lr=4.8516e-05) (hash(x)=141162902)
797
+ 6610 train 4.246824 (lr=4.8266e-05) (hash(x)=159490449)
798
+ 6620 train 4.124768 (lr=4.8017e-05) (hash(x)=151189772)
799
+ 6630 train 4.104190 (lr=4.7769e-05) (hash(x)=148451651)
800
+ 6640 train 4.214663 (lr=4.7522e-05) (hash(x)=152815550)
801
+ 6650 train 4.060166 (lr=4.7276e-05) (hash(x)=144316094)
802
+ 6660 train 4.396213 (lr=4.7030e-05) (hash(x)=143951240)
803
+ 6670 train 4.015669 (lr=4.6786e-05) (hash(x)=143203472)
804
+ 6680 train 4.059124 (lr=4.6542e-05) (hash(x)=132063806)
805
+ 6690 train 4.146520 (lr=4.6300e-05) (hash(x)=140739750)
806
+ 6700 val loss 4.2690
807
+ 6700 val perplexity 71.4494
808
+ 6700 train 4.214152 (lr=4.6058e-05) (hash(x)=153018737)
809
+ 6710 train 4.075549 (lr=4.5817e-05) (hash(x)=140107197)
810
+ 6720 train 4.177268 (lr=4.5578e-05) (hash(x)=144396559)
811
+ 6730 train 4.026475 (lr=4.5339e-05) (hash(x)=148038991)
812
+ 6740 train 4.247354 (lr=4.5101e-05) (hash(x)=145504746)
813
+ 6750 train 4.219265 (lr=4.4864e-05) (hash(x)=119579094)
814
+ 6760 train 4.263727 (lr=4.4628e-05) (hash(x)=161725430)
815
+ 6770 train 4.263690 (lr=4.4393e-05) (hash(x)=153646966)
816
+ 6780 train 4.276333 (lr=4.4159e-05) (hash(x)=153838463)
817
+ 6790 train 4.407363 (lr=4.3926e-05) (hash(x)=157891196)
818
+ 6800 val loss 4.2559
819
+ 6800 val perplexity 70.5173
820
+ 6800 train 4.119740 (lr=4.3693e-05) (hash(x)=155640155)
821
+ 6810 train 4.301554 (lr=4.3462e-05) (hash(x)=154913809)
822
+ 6820 train 4.228003 (lr=4.3232e-05) (hash(x)=159412493)
823
+ 6830 train 4.179027 (lr=4.3002e-05) (hash(x)=144756549)
824
+ 6840 train 4.379315 (lr=4.2774e-05) (hash(x)=143824638)
825
+ 6850 train 4.706836 (lr=4.2547e-05) (hash(x)=117180982)
826
+ 6860 train 4.221808 (lr=4.2320e-05) (hash(x)=141887150)
827
+ 6870 train 4.293876 (lr=4.2095e-05) (hash(x)=140094872)
828
+ 6880 train 4.143913 (lr=4.1870e-05) (hash(x)=141479953)
829
+ 6890 train 4.200525 (lr=4.1647e-05) (hash(x)=153166840)
830
+ 6900 val loss 4.2544
831
+ 6900 val perplexity 70.4146
832
+ 6900 train 4.220186 (lr=4.1425e-05) (hash(x)=153722115)
833
+ 6910 train 4.340835 (lr=4.1203e-05) (hash(x)=149578602)
834
+ 6920 train 4.179356 (lr=4.0983e-05) (hash(x)=149925108)
835
+ 6930 train 4.143062 (lr=4.0763e-05) (hash(x)=148589489)
836
+ 6940 train 4.043036 (lr=4.0545e-05) (hash(x)=144243794)
837
+ 6950 train 4.121499 (lr=4.0327e-05) (hash(x)=157743480)
838
+ 6960 train 4.121491 (lr=4.0111e-05) (hash(x)=139858247)
839
+ 6970 train 4.044151 (lr=3.9895e-05) (hash(x)=148727536)
840
+ 6980 train 4.151014 (lr=3.9681e-05) (hash(x)=153382595)
841
+ 6990 train 4.287755 (lr=3.9468e-05) (hash(x)=147529919)
842
+ 7000 val loss 4.2495
843
+ 7000 val perplexity 70.0724
844
+ 7000 train 4.252787 (lr=3.9255e-05) (hash(x)=146953450)
845
+ 7010 train 4.178947 (lr=3.9044e-05) (hash(x)=134561015)
846
+ 7020 train 4.275884 (lr=3.8834e-05) (hash(x)=148433515)
847
+ 7030 train 4.133454 (lr=3.8624e-05) (hash(x)=140394326)
848
+ 7040 train 4.136540 (lr=3.8416e-05) (hash(x)=136740763)
849
+ 7050 train 3.979248 (lr=3.8209e-05) (hash(x)=148637938)
850
+ 7060 train 4.144341 (lr=3.8003e-05) (hash(x)=147262147)
851
+ 7070 train 4.115367 (lr=3.7798e-05) (hash(x)=140191039)
852
+ 7080 train 4.138793 (lr=3.7593e-05) (hash(x)=148223305)
853
+ 7090 train 4.052055 (lr=3.7390e-05) (hash(x)=139984988)
854
+ 7100 val loss 4.2512
855
+ 7100 val perplexity 70.1904
856
+ 7100 train 4.155723 (lr=3.7188e-05) (hash(x)=137663885)
857
+ 7110 train 4.070774 (lr=3.6988e-05) (hash(x)=150955021)
858
+ 7120 train 3.967480 (lr=3.6788e-05) (hash(x)=148585682)
859
+ 7130 train 4.171380 (lr=3.6589e-05) (hash(x)=146055548)
860
+ 7140 train 4.198234 (lr=3.6391e-05) (hash(x)=155723541)
861
+ 7150 train 4.259448 (lr=3.6195e-05) (hash(x)=140465834)
862
+ 7160 train 4.258598 (lr=3.5999e-05) (hash(x)=163246922)
863
+ 7170 train 4.105273 (lr=3.5804e-05) (hash(x)=150754166)
864
+ 7180 train 4.630677 (lr=3.5611e-05) (hash(x)=142844151)
865
+ 7190 train 4.235976 (lr=3.5419e-05) (hash(x)=145167470)
866
+ 7200 val loss 4.2394
867
+ 7200 val perplexity 69.3656
868
+ 7200 train 4.418829 (lr=3.5227e-05) (hash(x)=146172950)
869
+ 7210 train 4.244604 (lr=3.5037e-05) (hash(x)=143155287)
870
+ 7220 train 4.198456 (lr=3.4848e-05) (hash(x)=151976669)
871
+ 7230 train 4.304687 (lr=3.4660e-05) (hash(x)=145979940)
872
+ 7240 train 4.288617 (lr=3.4473e-05) (hash(x)=155656034)
873
+ 7250 train 4.274027 (lr=3.4287e-05) (hash(x)=135610136)
874
+ 7260 train 4.359846 (lr=3.4102e-05) (hash(x)=148133023)
875
+ 7270 train 4.229548 (lr=3.3919e-05) (hash(x)=148006681)
876
+ 7280 train 4.265179 (lr=3.3736e-05) (hash(x)=151397627)
877
+ 7290 train 4.233287 (lr=3.3555e-05) (hash(x)=141696871)
878
+ 7300 val loss 4.2366
879
+ 7300 val perplexity 69.1727
880
+ 7300 train 4.163432 (lr=3.3375e-05) (hash(x)=150018163)
881
+ 7310 train 4.206165 (lr=3.3195e-05) (hash(x)=154900050)
882
+ 7320 train 4.097047 (lr=3.3017e-05) (hash(x)=142608283)
883
+ 7330 train 4.190380 (lr=3.2840e-05) (hash(x)=148977837)
884
+ 7340 train 4.202891 (lr=3.2664e-05) (hash(x)=141676907)
885
+ 7350 train 4.104062 (lr=3.2490e-05) (hash(x)=147846997)
886
+ 7360 train 4.332650 (lr=3.2316e-05) (hash(x)=143252181)
887
+ 7370 train 4.168971 (lr=3.2144e-05) (hash(x)=155081717)
888
+ 7380 train 4.240166 (lr=3.1972e-05) (hash(x)=149678451)
889
+ 7390 train 4.094159 (lr=3.1802e-05) (hash(x)=159341637)
890
+ 7400 val loss 4.2367
891
+ 7400 val perplexity 69.1786
892
+ 7400 train 4.368141 (lr=3.1633e-05) (hash(x)=145351166)
893
+ 7410 train 4.201197 (lr=3.1465e-05) (hash(x)=162274009)
894
+ 7420 train 4.029238 (lr=3.1298e-05) (hash(x)=143254858)
895
+ 7430 train 4.138753 (lr=3.1132e-05) (hash(x)=143049884)
896
+ 7440 train 4.060177 (lr=3.0968e-05) (hash(x)=141768538)
897
+ 7450 train 4.085445 (lr=3.0804e-05) (hash(x)=139895310)
898
+ 7460 train 3.972209 (lr=3.0642e-05) (hash(x)=155900365)
899
+ 7470 train 3.948389 (lr=3.0481e-05) (hash(x)=145048215)
900
+ 7480 train 4.078039 (lr=3.0321e-05) (hash(x)=136992624)
901
+ 7490 train 4.210022 (lr=3.0162e-05) (hash(x)=141424789)
902
+ 7500 val loss 4.2389
903
+ 7500 val perplexity 69.3324
904
+ 7500 train 4.042660 (lr=3.0005e-05) (hash(x)=145292116)
905
+ 7510 train 4.090287 (lr=2.9848e-05) (hash(x)=144818029)
906
+ 7520 train 4.255069 (lr=2.9693e-05) (hash(x)=152903515)
907
+ 7530 train 4.273150 (lr=2.9539e-05) (hash(x)=156934724)
908
+ 7540 train 4.521163 (lr=2.9386e-05) (hash(x)=156695926)
909
+ 7550 train 4.252291 (lr=2.9234e-05) (hash(x)=149115609)
910
+ 7560 train 4.243062 (lr=2.9084e-05) (hash(x)=143974224)
911
+ 7570 train 4.183311 (lr=2.8934e-05) (hash(x)=148204424)
912
+ 7580 train 4.261860 (lr=2.8786e-05) (hash(x)=150177910)
913
+ 7590 train 4.281188 (lr=2.8639e-05) (hash(x)=140325629)
914
+ 7600 val loss 4.2289
915
+ 7600 val perplexity 68.6425
916
+ 7600 train 4.168799 (lr=2.8493e-05) (hash(x)=150235132)
917
+ 7610 train 4.288196 (lr=2.8348e-05) (hash(x)=157594837)
918
+ 7620 train 4.291723 (lr=2.8204e-05) (hash(x)=146696749)
919
+ 7630 train 4.340945 (lr=2.8062e-05) (hash(x)=148463580)
920
+ 7640 train 4.168705 (lr=2.7921e-05) (hash(x)=140271046)
921
+ 7650 train 4.228643 (lr=2.7781e-05) (hash(x)=159952319)
922
+ 7660 train 4.215581 (lr=2.7642e-05) (hash(x)=146307392)
923
+ 7670 train 4.154178 (lr=2.7505e-05) (hash(x)=145027988)
924
+ 7680 train 4.237452 (lr=2.7368e-05) (hash(x)=147729698)
925
+ 7690 train 4.208883 (lr=2.7233e-05) (hash(x)=143401210)
926
+ 7700 val loss 4.2247
927
+ 7700 val perplexity 68.3554
928
+ 7700 train 4.150856 (lr=2.7099e-05) (hash(x)=154543455)
929
+ 7710 train 4.221950 (lr=2.6966e-05) (hash(x)=151600456)
930
+ 7720 train 4.135848 (lr=2.6834e-05) (hash(x)=165913803)
931
+ 7730 train 4.007513 (lr=2.6704e-05) (hash(x)=144942059)
932
+ 7740 train 4.244709 (lr=2.6575e-05) (hash(x)=150302103)
933
+ 7750 train 4.185214 (lr=2.6447e-05) (hash(x)=155782458)
934
+ 7760 train 4.205628 (lr=2.6320e-05) (hash(x)=147824384)
935
+ 7770 train 4.226006 (lr=2.6195e-05) (hash(x)=149177657)
936
+ 7780 train 4.262419 (lr=2.6070e-05) (hash(x)=164446738)
937
+ 7790 train 4.042237 (lr=2.5947e-05) (hash(x)=146417077)
938
+ 7800 val loss 4.2273
939
+ 7800 val perplexity 68.5324
940
+ 7800 train 4.182542 (lr=2.5825e-05) (hash(x)=142456852)
941
+ 7810 train 3.904720 (lr=2.5705e-05) (hash(x)=146221807)
942
+ 7820 train 4.109955 (lr=2.5585e-05) (hash(x)=154136533)
943
+ 7830 train 4.003256 (lr=2.5467e-05) (hash(x)=147492735)
944
+ 7840 train 4.149416 (lr=2.5350e-05) (hash(x)=143481291)
945
+ 7850 train 3.979809 (lr=2.5234e-05) (hash(x)=136749307)
946
+ 7860 train 4.008675 (lr=2.5119e-05) (hash(x)=128985347)
947
+ 7870 train 3.922668 (lr=2.5006e-05) (hash(x)=152897363)
948
+ 7880 train 4.103605 (lr=2.4894e-05) (hash(x)=151460360)
949
+ 7890 train 3.933238 (lr=2.4783e-05) (hash(x)=137481738)
950
+ 7900 val loss 4.2268
951
+ 7900 val perplexity 68.5010
952
+ 7900 train 4.014137 (lr=2.4674e-05) (hash(x)=147363479)
953
+ 7910 train 4.240106 (lr=2.4565e-05) (hash(x)=151165744)
954
+ 7920 train 4.399516 (lr=2.4458e-05) (hash(x)=146214226)
955
+ 7930 train 4.217457 (lr=2.4352e-05) (hash(x)=154457502)
956
+ 7940 train 4.146580 (lr=2.4247e-05) (hash(x)=162025201)
957
+ 7950 train 4.141949 (lr=2.4144e-05) (hash(x)=156488145)
958
+ 7960 train 4.396036 (lr=2.4042e-05) (hash(x)=154011584)
959
+ 7970 train 4.249054 (lr=2.3941e-05) (hash(x)=145330320)
960
+ 7980 train 4.170424 (lr=2.3841e-05) (hash(x)=148216312)
961
+ 7990 train 4.096125 (lr=2.3743e-05) (hash(x)=156819091)
962
+ 8000 val loss 4.2200
963
+ 8000 val perplexity 68.0321
964
+ 8000 train 4.257406 (lr=2.3646e-05) (hash(x)=156122973)
965
+ 8010 train 4.185824 (lr=2.3550e-05) (hash(x)=146933088)
966
+ 8020 train 4.274902 (lr=2.3455e-05) (hash(x)=151650746)
967
+ 8030 train 4.205957 (lr=2.3362e-05) (hash(x)=140692373)
968
+ 8040 train 4.396065 (lr=2.3269e-05) (hash(x)=147850094)
969
+ 8050 train 4.209576 (lr=2.3179e-05) (hash(x)=144454109)
970
+ 8060 train 4.083563 (lr=2.3089e-05) (hash(x)=152205124)
971
+ 8070 train 4.286815 (lr=2.3001e-05) (hash(x)=148918066)
972
+ 8080 train 4.231596 (lr=2.2913e-05) (hash(x)=146481024)
973
+ 8090 train 4.270984 (lr=2.2828e-05) (hash(x)=168785549)
974
+ 8100 val loss 4.2177
975
+ 8100 val perplexity 67.8789
976
+ 8100 train 4.283688 (lr=2.2743e-05) (hash(x)=156153179)
977
+ 8110 train 4.234581 (lr=2.2660e-05) (hash(x)=150498601)
978
+ 8120 train 4.238067 (lr=2.2578e-05) (hash(x)=142002714)
979
+ 8130 train 4.128817 (lr=2.2497e-05) (hash(x)=138450483)
980
+ 8140 train 4.220617 (lr=2.2417e-05) (hash(x)=145306351)
981
+ 8150 train 4.118754 (lr=2.2339e-05) (hash(x)=172502865)
982
+ 8160 train 4.202406 (lr=2.2262e-05) (hash(x)=156689649)
983
+ 8170 train 4.116984 (lr=2.2186e-05) (hash(x)=143766932)
984
+ 8180 train 4.250556 (lr=2.2112e-05) (hash(x)=150361321)
985
+ 8190 train 4.087672 (lr=2.2039e-05) (hash(x)=152664241)
986
+ 8200 val loss 4.2169
987
+ 8200 val perplexity 67.8241
988
+ 8200 train 4.355350 (lr=2.1967e-05) (hash(x)=146430698)
989
+ 8210 train 4.048945 (lr=2.1896e-05) (hash(x)=167362078)
990
+ 8220 train 4.269167 (lr=2.1827e-05) (hash(x)=157650942)
991
+ 8230 train 4.073519 (lr=2.1759e-05) (hash(x)=142102044)
992
+ 8240 train 4.125787 (lr=2.1692e-05) (hash(x)=153405900)
993
+ 8250 train 4.063593 (lr=2.1626e-05) (hash(x)=140222909)
994
+ 8260 train 4.186555 (lr=2.1562e-05) (hash(x)=137981854)
995
+ 8270 train 4.116394 (lr=2.1499e-05) (hash(x)=147576272)
996
+ 8280 train 4.139251 (lr=2.1438e-05) (hash(x)=157988195)
997
+ 8290 train 4.185129 (lr=2.1377e-05) (hash(x)=145715681)
998
+ 8300 val loss 4.2173
999
+ 8300 val perplexity 67.8467
1000
+ 8300 train 4.134668 (lr=2.1318e-05) (hash(x)=143507257)
1001
+ 8310 train 4.201077 (lr=2.1260e-05) (hash(x)=148930064)
1002
+ 8320 train 4.128144 (lr=2.1204e-05) (hash(x)=144134805)
1003
+ 8330 train 4.241508 (lr=2.1149e-05) (hash(x)=148418079)
1004
+ 8340 train 3.987782 (lr=2.1095e-05) (hash(x)=166518807)
1005
+ 8350 train 4.276786 (lr=2.1042e-05) (hash(x)=140042036)
1006
+ 8360 train 4.155201 (lr=2.0991e-05) (hash(x)=151427074)
1007
+ 8370 train 4.247159 (lr=2.0941e-05) (hash(x)=158389185)
1008
+ 8380 train 4.143140 (lr=2.0892e-05) (hash(x)=148193308)
1009
+ 8390 train 4.204437 (lr=2.0844e-05) (hash(x)=144289008)
1010
+ 8400 val loss 4.2107
1011
+ 8400 val perplexity 67.4009
1012
+ 8400 train 3.919692 (lr=2.0798e-05) (hash(x)=166272643)
1013
+ 8410 train 4.124575 (lr=2.0753e-05) (hash(x)=162244433)
1014
+ 8420 train 4.165239 (lr=2.0710e-05) (hash(x)=149271299)
1015
+ 8430 train 4.309717 (lr=2.0667e-05) (hash(x)=154135338)
1016
+ 8440 train 4.344011 (lr=2.0626e-05) (hash(x)=149661838)
1017
+ 8450 train 4.187363 (lr=2.0587e-05) (hash(x)=142205056)
1018
+ 8460 train 4.128433 (lr=2.0548e-05) (hash(x)=166652398)
1019
+ 8470 train 4.293701 (lr=2.0511e-05) (hash(x)=149029710)
1020
+ 8480 train 4.056143 (lr=2.0475e-05) (hash(x)=154217415)
1021
+ 8490 train 4.144645 (lr=2.0441e-05) (hash(x)=151526557)
1022
+ 8500 val loss 4.2110
1023
+ 8500 val perplexity 67.4210
1024
+ 8500 train 4.134089 (lr=2.0408e-05) (hash(x)=143887848)
1025
+ 8510 train 4.100814 (lr=2.0376e-05) (hash(x)=152324645)
1026
+ 8520 train 3.849978 (lr=2.0345e-05) (hash(x)=145124959)
1027
+ 8530 train 4.170582 (lr=2.0316e-05) (hash(x)=186319340)
1028
+ 8540 train 4.180887 (lr=2.0288e-05) (hash(x)=147388339)
1029
+ 8550 train 4.218202 (lr=2.0261e-05) (hash(x)=150118928)
1030
+ 8560 train 4.222759 (lr=2.0235e-05) (hash(x)=153985286)
1031
+ 8570 train 4.155669 (lr=2.0211e-05) (hash(x)=149582799)
1032
+ 8580 train 4.168159 (lr=2.0189e-05) (hash(x)=156867870)
1033
+ 8590 train 4.039477 (lr=2.0167e-05) (hash(x)=148773298)
1034
+ 8600 val loss 4.2126
1035
+ 8600 val perplexity 67.5314
1036
+ 8600 train 4.167048 (lr=2.0147e-05) (hash(x)=156900341)
1037
+ 8610 train 4.310606 (lr=2.0128e-05) (hash(x)=156790089)
1038
+ 8620 train 4.111687 (lr=2.0110e-05) (hash(x)=154680805)
1039
+ 8630 train 4.151445 (lr=2.0094e-05) (hash(x)=138107873)
1040
+ 8640 train 4.244309 (lr=2.0079e-05) (hash(x)=142554243)
1041
+ 8650 train 4.048779 (lr=2.0065e-05) (hash(x)=134360956)
1042
+ 8660 train 4.123186 (lr=2.0053e-05) (hash(x)=128599610)
1043
+ 8670 train 4.156398 (lr=2.0042e-05) (hash(x)=159628945)
1044
+ 8680 train 4.186398 (lr=2.0032e-05) (hash(x)=143161532)
1045
+ 8690 train 4.223917 (lr=2.0023e-05) (hash(x)=140104912)
1046
+ 8700 val loss 4.2067
1047
+ 8700 val perplexity 67.1344
1048
+ 8700 train 4.405335 (lr=2.0016e-05) (hash(x)=146417632)
1049
+ 8710 train 4.351265 (lr=2.0010e-05) (hash(x)=153671695)
1050
+ 8720 train 4.227539 (lr=2.0006e-05) (hash(x)=141680750)
1051
+ 8730 train 4.111185 (lr=2.0003e-05) (hash(x)=147279231)
1052
+ 8740 train 4.385740 (lr=2.0001e-05) (hash(x)=129729371)
1053
+ 8749 val loss 4.2063
1054
+ 8749 val perplexity 67.1059
lr2e-4_total_batch_size61440_baseline_seed1339/model_08749.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31fcbc807a88c4f07f6c1b645a045b571bb8f18bfb3def3c80a9c5dfadcc563c
3
+ size 92843394
lr2e-4_total_batch_size61440_baseline_seed1339/optimizer_08749.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e8a7f7baeb0640108a06633c4f97fb58515259ab03dfc32cf27ae7374d8008ae
3
+ size 179406214