andrew-healey commited on
Commit
a629449
·
verified ·
1 Parent(s): c4f3249

Upload folder using huggingface_hub

Browse files
lr8e-4_total_batch_size61440_baseline_seed1339/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_11/lr8e-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": 0.0008, "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_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
lr8e-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=1.6000e-06) (hash(x)=150724848)
5
+ 10 train 10.245926 (lr=1.7600e-05) (hash(x)=149663361)
6
+ 20 train 9.822400 (lr=3.3600e-05) (hash(x)=153483789)
7
+ 30 train 9.564533 (lr=4.9600e-05) (hash(x)=145733176)
8
+ 40 train 9.034535 (lr=6.5600e-05) (hash(x)=145287265)
9
+ 50 train 8.581837 (lr=8.1600e-05) (hash(x)=149217061)
10
+ 60 train 8.129409 (lr=9.7600e-05) (hash(x)=150157569)
11
+ 70 train 7.794233 (lr=1.1360e-04) (hash(x)=140292841)
12
+ 80 train 7.707448 (lr=1.2960e-04) (hash(x)=152157598)
13
+ 90 train 7.538925 (lr=1.4560e-04) (hash(x)=149756968)
14
+ 100 val loss 7.5200
15
+ 100 val perplexity 1844.5479
16
+ 100 train 7.475801 (lr=1.6160e-04) (hash(x)=149910534)
17
+ 110 train 7.385913 (lr=1.7760e-04) (hash(x)=159020328)
18
+ 120 train 7.322307 (lr=1.9360e-04) (hash(x)=150731273)
19
+ 130 train 7.142769 (lr=2.0960e-04) (hash(x)=156727355)
20
+ 140 train 7.171133 (lr=2.2560e-04) (hash(x)=166473281)
21
+ 150 train 6.873593 (lr=2.4160e-04) (hash(x)=142185643)
22
+ 160 train 6.863807 (lr=2.5760e-04) (hash(x)=151371145)
23
+ 170 train 6.893668 (lr=2.7360e-04) (hash(x)=151601659)
24
+ 180 train 6.718715 (lr=2.8960e-04) (hash(x)=153368917)
25
+ 190 train 6.572965 (lr=3.0560e-04) (hash(x)=145008196)
26
+ 200 val loss 6.7006
27
+ 200 val perplexity 812.9152
28
+ 200 train 6.599102 (lr=3.2160e-04) (hash(x)=148123706)
29
+ 210 train 6.472252 (lr=3.3760e-04) (hash(x)=145714762)
30
+ 220 train 6.468995 (lr=3.5360e-04) (hash(x)=150622583)
31
+ 230 train 6.259478 (lr=3.6960e-04) (hash(x)=146512648)
32
+ 240 train 6.294050 (lr=3.8560e-04) (hash(x)=151535384)
33
+ 250 train 6.216956 (lr=4.0160e-04) (hash(x)=149952383)
34
+ 260 train 6.123109 (lr=4.1760e-04) (hash(x)=148726088)
35
+ 270 train 6.088806 (lr=4.3360e-04) (hash(x)=152776695)
36
+ 280 train 6.128976 (lr=4.4960e-04) (hash(x)=157260375)
37
+ 290 train 6.176296 (lr=4.6560e-04) (hash(x)=151559106)
38
+ 300 val loss 6.2345
39
+ 300 val perplexity 510.0543
40
+ 300 train 5.969311 (lr=4.8160e-04) (hash(x)=146678221)
41
+ 310 train 6.073006 (lr=4.9760e-04) (hash(x)=156713527)
42
+ 320 train 6.133308 (lr=5.1360e-04) (hash(x)=144947971)
43
+ 330 train 6.297166 (lr=5.2960e-04) (hash(x)=158684944)
44
+ 340 train 6.161257 (lr=5.4560e-04) (hash(x)=162318756)
45
+ 350 train 5.966339 (lr=5.6160e-04) (hash(x)=156180736)
46
+ 360 train 6.142735 (lr=5.7760e-04) (hash(x)=178373797)
47
+ 370 train 6.044660 (lr=5.9360e-04) (hash(x)=154389449)
48
+ 380 train 5.882610 (lr=6.0960e-04) (hash(x)=150369080)
49
+ 390 train 6.010149 (lr=6.2560e-04) (hash(x)=159840654)
50
+ 400 val loss 5.8983
51
+ 400 val perplexity 364.4337
52
+ 400 train 5.810166 (lr=6.4160e-04) (hash(x)=151700982)
53
+ 410 train 5.797544 (lr=6.5760e-04) (hash(x)=160467274)
54
+ 420 train 5.707871 (lr=6.7360e-04) (hash(x)=148930522)
55
+ 430 train 5.927991 (lr=6.8960e-04) (hash(x)=154565303)
56
+ 440 train 5.887704 (lr=7.0560e-04) (hash(x)=148079227)
57
+ 450 train 5.619516 (lr=7.2160e-04) (hash(x)=147288467)
58
+ 460 train 5.638433 (lr=7.3760e-04) (hash(x)=145272617)
59
+ 470 train 5.651437 (lr=7.5360e-04) (hash(x)=123126729)
60
+ 480 train 5.613426 (lr=7.6960e-04) (hash(x)=156287490)
61
+ 490 train 5.737548 (lr=7.8560e-04) (hash(x)=161700145)
62
+ 500 val loss 5.6481
63
+ 500 val perplexity 283.7534
64
+ 500 train 5.760207 (lr=8.0000e-04) (hash(x)=156182087)
65
+ 510 train 5.364035 (lr=8.0000e-04) (hash(x)=131674504)
66
+ 520 train 5.667906 (lr=7.9999e-04) (hash(x)=162758254)
67
+ 530 train 5.490379 (lr=7.9998e-04) (hash(x)=133510409)
68
+ 540 train 5.480013 (lr=7.9996e-04) (hash(x)=157363914)
69
+ 550 train 5.413898 (lr=7.9993e-04) (hash(x)=147861550)
70
+ 560 train 5.360544 (lr=7.9991e-04) (hash(x)=149449280)
71
+ 570 train 5.363545 (lr=7.9987e-04) (hash(x)=143899904)
72
+ 580 train 5.341896 (lr=7.9983e-04) (hash(x)=149266312)
73
+ 590 train 5.577617 (lr=7.9979e-04) (hash(x)=164352774)
74
+ 600 val loss 5.4705
75
+ 600 val perplexity 237.5689
76
+ 600 train 5.338564 (lr=7.9974e-04) (hash(x)=149318660)
77
+ 610 train 5.238789 (lr=7.9968e-04) (hash(x)=146101781)
78
+ 620 train 5.189034 (lr=7.9962e-04) (hash(x)=137184623)
79
+ 630 train 5.075717 (lr=7.9956e-04) (hash(x)=145791873)
80
+ 640 train 5.097056 (lr=7.9949e-04) (hash(x)=143999212)
81
+ 650 train 5.199197 (lr=7.9941e-04) (hash(x)=151631103)
82
+ 660 train 4.962468 (lr=7.9933e-04) (hash(x)=138360108)
83
+ 670 train 5.350170 (lr=7.9925e-04) (hash(x)=156637679)
84
+ 680 train 4.987696 (lr=7.9915e-04) (hash(x)=143752601)
85
+ 690 train 5.141695 (lr=7.9906e-04) (hash(x)=158368903)
86
+ 700 val loss 5.3527
87
+ 700 val perplexity 211.1805
88
+ 700 train 5.349389 (lr=7.9896e-04) (hash(x)=150482428)
89
+ 710 train 5.290276 (lr=7.9885e-04) (hash(x)=158085060)
90
+ 720 train 5.366736 (lr=7.9874e-04) (hash(x)=159461069)
91
+ 730 train 5.202476 (lr=7.9862e-04) (hash(x)=148795562)
92
+ 740 train 5.220983 (lr=7.9850e-04) (hash(x)=149967984)
93
+ 750 train 5.208473 (lr=7.9837e-04) (hash(x)=144458617)
94
+ 760 train 5.249181 (lr=7.9824e-04) (hash(x)=155106202)
95
+ 770 train 5.044078 (lr=7.9810e-04) (hash(x)=144506191)
96
+ 780 train 5.193412 (lr=7.9796e-04) (hash(x)=155825158)
97
+ 790 train 5.238680 (lr=7.9781e-04) (hash(x)=158835381)
98
+ 800 val loss 5.1807
99
+ 800 val perplexity 177.8089
100
+ 800 train 5.125618 (lr=7.9765e-04) (hash(x)=143268605)
101
+ 810 train 5.049448 (lr=7.9749e-04) (hash(x)=139072755)
102
+ 820 train 5.186448 (lr=7.9733e-04) (hash(x)=179384792)
103
+ 830 train 4.977615 (lr=7.9716e-04) (hash(x)=151086962)
104
+ 840 train 5.597301 (lr=7.9699e-04) (hash(x)=142711393)
105
+ 850 train 5.063047 (lr=7.9681e-04) (hash(x)=150817011)
106
+ 860 train 4.995766 (lr=7.9662e-04) (hash(x)=138495801)
107
+ 870 train 5.052045 (lr=7.9643e-04) (hash(x)=155474016)
108
+ 880 train 5.058653 (lr=7.9624e-04) (hash(x)=151951501)
109
+ 890 train 4.917627 (lr=7.9604e-04) (hash(x)=143699524)
110
+ 900 val loss 5.0719
111
+ 900 val perplexity 159.4818
112
+ 900 train 5.065027 (lr=7.9583e-04) (hash(x)=152322423)
113
+ 910 train 5.011721 (lr=7.9562e-04) (hash(x)=151010509)
114
+ 920 train 4.795933 (lr=7.9541e-04) (hash(x)=148720342)
115
+ 930 train 4.768640 (lr=7.9518e-04) (hash(x)=125869480)
116
+ 940 train 5.010793 (lr=7.9496e-04) (hash(x)=150349503)
117
+ 950 train 4.785191 (lr=7.9473e-04) (hash(x)=143574126)
118
+ 960 train 4.969505 (lr=7.9449e-04) (hash(x)=146292562)
119
+ 970 train 4.912924 (lr=7.9425e-04) (hash(x)=162256596)
120
+ 980 train 4.859094 (lr=7.9400e-04) (hash(x)=137640259)
121
+ 990 train 4.734047 (lr=7.9375e-04) (hash(x)=141767794)
122
+ 1000 val loss 5.0022
123
+ 1000 val perplexity 148.7356
124
+ 1000 train 4.943435 (lr=7.9349e-04) (hash(x)=147904298)
125
+ 1010 train 4.861160 (lr=7.9323e-04) (hash(x)=162239186)
126
+ 1020 train 4.626260 (lr=7.9297e-04) (hash(x)=140706537)
127
+ 1030 train 4.750134 (lr=7.9269e-04) (hash(x)=155556720)
128
+ 1040 train 4.637603 (lr=7.9242e-04) (hash(x)=144746445)
129
+ 1050 train 4.719503 (lr=7.9213e-04) (hash(x)=144719400)
130
+ 1060 train 4.547071 (lr=7.9185e-04) (hash(x)=153065665)
131
+ 1070 train 4.686386 (lr=7.9155e-04) (hash(x)=136468722)
132
+ 1080 train 4.659657 (lr=7.9126e-04) (hash(x)=157674756)
133
+ 1090 train 4.839837 (lr=7.9095e-04) (hash(x)=155882884)
134
+ 1100 val loss 4.8999
135
+ 1100 val perplexity 134.2810
136
+ 1100 train 5.118608 (lr=7.9064e-04) (hash(x)=154343147)
137
+ 1110 train 4.914001 (lr=7.9033e-04) (hash(x)=158582939)
138
+ 1120 train 4.714669 (lr=7.9001e-04) (hash(x)=162479784)
139
+ 1130 train 4.897726 (lr=7.8969e-04) (hash(x)=155704345)
140
+ 1140 train 4.788340 (lr=7.8936e-04) (hash(x)=146934299)
141
+ 1150 train 4.736755 (lr=7.8903e-04) (hash(x)=154372635)
142
+ 1160 train 4.720962 (lr=7.8869e-04) (hash(x)=145536832)
143
+ 1170 train 5.018084 (lr=7.8835e-04) (hash(x)=158785901)
144
+ 1180 train 4.807900 (lr=7.8800e-04) (hash(x)=158271614)
145
+ 1190 train 4.850050 (lr=7.8764e-04) (hash(x)=157365589)
146
+ 1200 val loss 4.8191
147
+ 1200 val perplexity 123.8540
148
+ 1200 train 4.877159 (lr=7.8729e-04) (hash(x)=141843115)
149
+ 1210 train 4.769855 (lr=7.8692e-04) (hash(x)=146807973)
150
+ 1220 train 4.746709 (lr=7.8655e-04) (hash(x)=144149947)
151
+ 1230 train 4.683541 (lr=7.8618e-04) (hash(x)=147869063)
152
+ 1240 train 4.776947 (lr=7.8580e-04) (hash(x)=152217389)
153
+ 1250 train 4.770141 (lr=7.8542e-04) (hash(x)=149417679)
154
+ 1260 train 4.787346 (lr=7.8503e-04) (hash(x)=144029489)
155
+ 1270 train 4.917655 (lr=7.8464e-04) (hash(x)=154328303)
156
+ 1280 train 4.770100 (lr=7.8424e-04) (hash(x)=151625838)
157
+ 1290 train 4.634686 (lr=7.8383e-04) (hash(x)=140270541)
158
+ 1300 val loss 4.7562
159
+ 1300 val perplexity 116.2985
160
+ 1300 train 4.689701 (lr=7.8342e-04) (hash(x)=145279030)
161
+ 1310 train 4.722775 (lr=7.8301e-04) (hash(x)=140279630)
162
+ 1320 train 4.446221 (lr=7.8259e-04) (hash(x)=133642994)
163
+ 1330 train 4.612603 (lr=7.8217e-04) (hash(x)=146634776)
164
+ 1340 train 4.721035 (lr=7.8174e-04) (hash(x)=155919973)
165
+ 1350 train 4.745049 (lr=7.8131e-04) (hash(x)=146363911)
166
+ 1360 train 4.612625 (lr=7.8087e-04) (hash(x)=150613636)
167
+ 1370 train 4.609677 (lr=7.8042e-04) (hash(x)=148536643)
168
+ 1380 train 4.619076 (lr=7.7998e-04) (hash(x)=152803906)
169
+ 1390 train 4.535807 (lr=7.7952e-04) (hash(x)=150383305)
170
+ 1400 val loss 4.7323
171
+ 1400 val perplexity 113.5543
172
+ 1400 train 4.638844 (lr=7.7906e-04) (hash(x)=152507639)
173
+ 1410 train 4.392365 (lr=7.7860e-04) (hash(x)=161841408)
174
+ 1420 train 4.519206 (lr=7.7813e-04) (hash(x)=167824828)
175
+ 1430 train 4.476690 (lr=7.7766e-04) (hash(x)=147696374)
176
+ 1440 train 4.424263 (lr=7.7718e-04) (hash(x)=150085494)
177
+ 1450 train 4.384748 (lr=7.7670e-04) (hash(x)=139296191)
178
+ 1460 train 4.308630 (lr=7.7621e-04) (hash(x)=142956546)
179
+ 1470 train 4.372282 (lr=7.7572e-04) (hash(x)=155908628)
180
+ 1480 train 4.364110 (lr=7.7522e-04) (hash(x)=150893345)
181
+ 1490 train 4.382466 (lr=7.7472e-04) (hash(x)=143548902)
182
+ 1500 val loss 4.6704
183
+ 1500 val perplexity 106.7408
184
+ 1500 train 4.634660 (lr=7.7421e-04) (hash(x)=148473774)
185
+ 1510 train 4.708790 (lr=7.7370e-04) (hash(x)=161631000)
186
+ 1520 train 4.647066 (lr=7.7318e-04) (hash(x)=166217855)
187
+ 1530 train 4.597146 (lr=7.7266e-04) (hash(x)=149440046)
188
+ 1540 train 4.544404 (lr=7.7214e-04) (hash(x)=155013278)
189
+ 1550 train 4.579974 (lr=7.7160e-04) (hash(x)=154130527)
190
+ 1560 train 4.601164 (lr=7.7107e-04) (hash(x)=152138735)
191
+ 1570 train 4.505819 (lr=7.7053e-04) (hash(x)=148128305)
192
+ 1580 train 4.648543 (lr=7.6998e-04) (hash(x)=157176546)
193
+ 1590 train 4.759646 (lr=7.6943e-04) (hash(x)=152935467)
194
+ 1600 val loss 4.5978
195
+ 1600 val perplexity 99.2686
196
+ 1600 train 4.577094 (lr=7.6888e-04) (hash(x)=151117002)
197
+ 1610 train 4.562826 (lr=7.6832e-04) (hash(x)=149883384)
198
+ 1620 train 4.619254 (lr=7.6775e-04) (hash(x)=145015354)
199
+ 1630 train 4.542802 (lr=7.6718e-04) (hash(x)=146717645)
200
+ 1640 train 4.754841 (lr=7.6661e-04) (hash(x)=156372727)
201
+ 1650 train 4.659436 (lr=7.6603e-04) (hash(x)=140261555)
202
+ 1660 train 4.491565 (lr=7.6545e-04) (hash(x)=150088650)
203
+ 1670 train 4.413221 (lr=7.6486e-04) (hash(x)=135943513)
204
+ 1680 train 4.638790 (lr=7.6426e-04) (hash(x)=153529653)
205
+ 1690 train 4.687901 (lr=7.6367e-04) (hash(x)=148930360)
206
+ 1700 val loss 4.6019
207
+ 1700 val perplexity 99.6720
208
+ 1700 train 4.578386 (lr=7.6306e-04) (hash(x)=138011335)
209
+ 1710 train 4.546149 (lr=7.6246e-04) (hash(x)=141383599)
210
+ 1720 train 4.725561 (lr=7.6184e-04) (hash(x)=161449652)
211
+ 1730 train 4.516419 (lr=7.6123e-04) (hash(x)=153923561)
212
+ 1740 train 4.445590 (lr=7.6061e-04) (hash(x)=146043253)
213
+ 1750 train 4.475410 (lr=7.5998e-04) (hash(x)=155720883)
214
+ 1760 train 4.464515 (lr=7.5935e-04) (hash(x)=148222148)
215
+ 1770 train 4.530339 (lr=7.5872e-04) (hash(x)=155712380)
216
+ 1780 train 4.408880 (lr=7.5808e-04) (hash(x)=149767859)
217
+ 1790 train 4.639216 (lr=7.5743e-04) (hash(x)=167214230)
218
+ 1800 val loss 4.5695
219
+ 1800 val perplexity 96.4922
220
+ 1800 train 4.549855 (lr=7.5678e-04) (hash(x)=171180926)
221
+ 1810 train 4.561873 (lr=7.5613e-04) (hash(x)=162375834)
222
+ 1820 train 4.683563 (lr=7.5547e-04) (hash(x)=151495752)
223
+ 1830 train 4.583334 (lr=7.5481e-04) (hash(x)=146532257)
224
+ 1840 train 4.659523 (lr=7.5414e-04) (hash(x)=145694434)
225
+ 1850 train 4.544674 (lr=7.5347e-04) (hash(x)=144823488)
226
+ 1860 train 4.645030 (lr=7.5279e-04) (hash(x)=149836286)
227
+ 1870 train 4.613966 (lr=7.5211e-04) (hash(x)=143935049)
228
+ 1880 train 4.640106 (lr=7.5143e-04) (hash(x)=151786371)
229
+ 1890 train 4.536320 (lr=7.5074e-04) (hash(x)=152892567)
230
+ 1900 val loss 4.5268
231
+ 1900 val perplexity 92.4584
232
+ 1900 train 4.568473 (lr=7.5004e-04) (hash(x)=141769419)
233
+ 1910 train 4.592987 (lr=7.4934e-04) (hash(x)=156914416)
234
+ 1920 train 4.442967 (lr=7.4864e-04) (hash(x)=151573900)
235
+ 1930 train 4.540367 (lr=7.4793e-04) (hash(x)=145121701)
236
+ 1940 train 4.477437 (lr=7.4722e-04) (hash(x)=150896348)
237
+ 1950 train 4.586665 (lr=7.4650e-04) (hash(x)=168797189)
238
+ 1960 train 4.468393 (lr=7.4578e-04) (hash(x)=146711430)
239
+ 1970 train 4.321590 (lr=7.4505e-04) (hash(x)=152180946)
240
+ 1980 train 4.448318 (lr=7.4432e-04) (hash(x)=155542268)
241
+ 1990 train 4.413014 (lr=7.4359e-04) (hash(x)=144668660)
242
+ 2000 val loss 4.4875
243
+ 2000 val perplexity 88.8971
244
+ 2000 train 4.484056 (lr=7.4285e-04) (hash(x)=151963443)
245
+ 2010 train 4.478916 (lr=7.4211e-04) (hash(x)=150810352)
246
+ 2020 train 4.476051 (lr=7.4136e-04) (hash(x)=157539832)
247
+ 2030 train 4.695871 (lr=7.4061e-04) (hash(x)=137858786)
248
+ 2040 train 4.434136 (lr=7.3985e-04) (hash(x)=154270900)
249
+ 2050 train 4.290129 (lr=7.3909e-04) (hash(x)=156252579)
250
+ 2060 train 4.224801 (lr=7.3833e-04) (hash(x)=143285549)
251
+ 2070 train 4.433332 (lr=7.3756e-04) (hash(x)=150542438)
252
+ 2080 train 4.349329 (lr=7.3678e-04) (hash(x)=146290330)
253
+ 2090 train 4.439387 (lr=7.3600e-04) (hash(x)=146486756)
254
+ 2100 val loss 4.5148
255
+ 2100 val perplexity 91.3602
256
+ 2100 train 4.447355 (lr=7.3522e-04) (hash(x)=162947470)
257
+ 2110 train 4.346252 (lr=7.3444e-04) (hash(x)=150305391)
258
+ 2120 train 4.514754 (lr=7.3364e-04) (hash(x)=156452783)
259
+ 2130 train 4.486069 (lr=7.3285e-04) (hash(x)=167481875)
260
+ 2140 train 4.460458 (lr=7.3205e-04) (hash(x)=141782256)
261
+ 2150 train 4.489695 (lr=7.3125e-04) (hash(x)=145035419)
262
+ 2160 train 4.542157 (lr=7.3044e-04) (hash(x)=142000400)
263
+ 2170 train 4.507618 (lr=7.2963e-04) (hash(x)=144229380)
264
+ 2180 train 4.551351 (lr=7.2881e-04) (hash(x)=166204176)
265
+ 2190 train 4.471841 (lr=7.2799e-04) (hash(x)=150712057)
266
+ 2200 val loss 4.4530
267
+ 2200 val perplexity 85.8821
268
+ 2200 train 4.367837 (lr=7.2716e-04) (hash(x)=154954810)
269
+ 2210 train 4.563277 (lr=7.2634e-04) (hash(x)=155668760)
270
+ 2220 train 4.391463 (lr=7.2550e-04) (hash(x)=144834708)
271
+ 2230 train 4.371978 (lr=7.2467e-04) (hash(x)=159744566)
272
+ 2240 train 4.343382 (lr=7.2382e-04) (hash(x)=158543377)
273
+ 2250 train 4.358026 (lr=7.2298e-04) (hash(x)=151004281)
274
+ 2260 train 4.462659 (lr=7.2213e-04) (hash(x)=151242482)
275
+ 2270 train 4.404618 (lr=7.2128e-04) (hash(x)=152056559)
276
+ 2280 train 4.345529 (lr=7.2042e-04) (hash(x)=148864076)
277
+ 2290 train 4.471915 (lr=7.1956e-04) (hash(x)=144294586)
278
+ 2300 val loss 4.4409
279
+ 2300 val perplexity 84.8514
280
+ 2300 train 4.369696 (lr=7.1869e-04) (hash(x)=151878111)
281
+ 2310 train 4.382298 (lr=7.1782e-04) (hash(x)=154139457)
282
+ 2320 train 4.313839 (lr=7.1695e-04) (hash(x)=155571592)
283
+ 2330 train 4.351645 (lr=7.1607e-04) (hash(x)=154823764)
284
+ 2340 train 4.407782 (lr=7.1519e-04) (hash(x)=153252177)
285
+ 2350 train 4.345878 (lr=7.1430e-04) (hash(x)=136051830)
286
+ 2360 train 4.387623 (lr=7.1341e-04) (hash(x)=153814175)
287
+ 2370 train 4.386446 (lr=7.1252e-04) (hash(x)=146391078)
288
+ 2380 train 4.352433 (lr=7.1162e-04) (hash(x)=150782698)
289
+ 2390 train 4.426336 (lr=7.1072e-04) (hash(x)=142752256)
290
+ 2400 val loss 4.4495
291
+ 2400 val perplexity 85.5858
292
+ 2400 train 4.330253 (lr=7.0981e-04) (hash(x)=158661057)
293
+ 2410 train 4.397495 (lr=7.0890e-04) (hash(x)=153611303)
294
+ 2420 train 4.570627 (lr=7.0799e-04) (hash(x)=159508684)
295
+ 2430 train 4.364119 (lr=7.0707e-04) (hash(x)=149501192)
296
+ 2440 train 4.025926 (lr=7.0615e-04) (hash(x)=181177177)
297
+ 2450 train 4.497195 (lr=7.0523e-04) (hash(x)=163664732)
298
+ 2460 train 4.188496 (lr=7.0430e-04) (hash(x)=194527955)
299
+ 2470 train 4.411254 (lr=7.0337e-04) (hash(x)=158088521)
300
+ 2480 train 4.373590 (lr=7.0243e-04) (hash(x)=152865678)
301
+ 2490 train 4.352432 (lr=7.0149e-04) (hash(x)=165703871)
302
+ 2500 val loss 4.3975
303
+ 2500 val perplexity 81.2446
304
+ 2500 train 4.450110 (lr=7.0054e-04) (hash(x)=150925584)
305
+ 2510 train 4.351538 (lr=6.9960e-04) (hash(x)=144406177)
306
+ 2520 train 4.241173 (lr=6.9864e-04) (hash(x)=120133043)
307
+ 2530 train 4.433339 (lr=6.9769e-04) (hash(x)=167900247)
308
+ 2540 train 4.327187 (lr=6.9673e-04) (hash(x)=162599203)
309
+ 2550 train 4.391118 (lr=6.9577e-04) (hash(x)=149328180)
310
+ 2560 train 4.275068 (lr=6.9480e-04) (hash(x)=150049968)
311
+ 2570 train 4.416946 (lr=6.9383e-04) (hash(x)=162876159)
312
+ 2580 train 4.248622 (lr=6.9286e-04) (hash(x)=148260649)
313
+ 2590 train 4.390316 (lr=6.9188e-04) (hash(x)=148998018)
314
+ 2600 val loss 4.3849
315
+ 2600 val perplexity 80.2274
316
+ 2600 train 4.306809 (lr=6.9090e-04) (hash(x)=144515755)
317
+ 2610 train 4.419702 (lr=6.8991e-04) (hash(x)=154455224)
318
+ 2620 train 4.224682 (lr=6.8892e-04) (hash(x)=142920048)
319
+ 2630 train 4.378064 (lr=6.8793e-04) (hash(x)=143931236)
320
+ 2640 train 4.232391 (lr=6.8694e-04) (hash(x)=150983323)
321
+ 2650 train 4.333903 (lr=6.8594e-04) (hash(x)=139360405)
322
+ 2660 train 4.249875 (lr=6.8493e-04) (hash(x)=148850470)
323
+ 2670 train 4.498496 (lr=6.8393e-04) (hash(x)=149929606)
324
+ 2680 train 4.467369 (lr=6.8292e-04) (hash(x)=151673000)
325
+ 2690 train 4.347357 (lr=6.8190e-04) (hash(x)=140631266)
326
+ 2700 val loss 4.3795
327
+ 2700 val perplexity 79.8007
328
+ 2700 train 4.318307 (lr=6.8089e-04) (hash(x)=153109144)
329
+ 2710 train 4.203733 (lr=6.7987e-04) (hash(x)=135332836)
330
+ 2720 train 4.558447 (lr=6.7884e-04) (hash(x)=131611182)
331
+ 2730 train 4.388696 (lr=6.7782e-04) (hash(x)=142909388)
332
+ 2740 train 4.428222 (lr=6.7678e-04) (hash(x)=145593419)
333
+ 2750 train 4.315254 (lr=6.7575e-04) (hash(x)=158225769)
334
+ 2760 train 4.288288 (lr=6.7471e-04) (hash(x)=170509938)
335
+ 2770 train 4.304924 (lr=6.7367e-04) (hash(x)=150785805)
336
+ 2780 train 4.250027 (lr=6.7263e-04) (hash(x)=167281736)
337
+ 2790 train 4.435450 (lr=6.7158e-04) (hash(x)=161660187)
338
+ 2800 val loss 4.3557
339
+ 2800 val perplexity 77.9251
340
+ 2800 train 4.389541 (lr=6.7053e-04) (hash(x)=151152897)
341
+ 2810 train 4.352497 (lr=6.6947e-04) (hash(x)=143172067)
342
+ 2820 train 4.367549 (lr=6.6841e-04) (hash(x)=141569169)
343
+ 2830 train 4.245612 (lr=6.6735e-04) (hash(x)=159709681)
344
+ 2840 train 4.301286 (lr=6.6629e-04) (hash(x)=162552294)
345
+ 2850 train 4.430515 (lr=6.6522e-04) (hash(x)=146762690)
346
+ 2860 train 4.297157 (lr=6.6415e-04) (hash(x)=145729422)
347
+ 2870 train 4.374395 (lr=6.6308e-04) (hash(x)=144360628)
348
+ 2880 train 4.188087 (lr=6.6200e-04) (hash(x)=147737098)
349
+ 2890 train 4.295529 (lr=6.6092e-04) (hash(x)=146849057)
350
+ 2900 val loss 4.3367
351
+ 2900 val perplexity 76.4581
352
+ 2900 train 4.354200 (lr=6.5983e-04) (hash(x)=145800210)
353
+ 2910 train 4.169558 (lr=6.5875e-04) (hash(x)=147316313)
354
+ 2920 train 4.216280 (lr=6.5766e-04) (hash(x)=152595415)
355
+ 2930 train 4.383364 (lr=6.5656e-04) (hash(x)=156542647)
356
+ 2940 train 4.237524 (lr=6.5547e-04) (hash(x)=141183896)
357
+ 2950 train 4.340014 (lr=6.5437e-04) (hash(x)=156329036)
358
+ 2960 train 4.375528 (lr=6.5326e-04) (hash(x)=156717793)
359
+ 2970 train 4.234772 (lr=6.5216e-04) (hash(x)=143949931)
360
+ 2980 train 4.306321 (lr=6.5105e-04) (hash(x)=155067290)
361
+ 2990 train 4.267857 (lr=6.4994e-04) (hash(x)=148774729)
362
+ 3000 val loss 4.3547
363
+ 3000 val perplexity 77.8438
364
+ 3000 train 4.257525 (lr=6.4882e-04) (hash(x)=141997485)
365
+ 3010 train 4.228309 (lr=6.4770e-04) (hash(x)=160762953)
366
+ 3020 train 4.184501 (lr=6.4658e-04) (hash(x)=170341665)
367
+ 3030 train 4.451117 (lr=6.4546e-04) (hash(x)=150722836)
368
+ 3040 train 4.248068 (lr=6.4433e-04) (hash(x)=154243307)
369
+ 3050 train 4.292173 (lr=6.4320e-04) (hash(x)=146077234)
370
+ 3060 train 4.252038 (lr=6.4207e-04) (hash(x)=152125542)
371
+ 3070 train 4.433154 (lr=6.4093e-04) (hash(x)=144149387)
372
+ 3080 train 4.259621 (lr=6.3979e-04) (hash(x)=147508712)
373
+ 3090 train 4.263323 (lr=6.3865e-04) (hash(x)=143583369)
374
+ 3100 val loss 4.3086
375
+ 3100 val perplexity 74.3375
376
+ 3100 train 4.283150 (lr=6.3750e-04) (hash(x)=154049740)
377
+ 3110 train 4.341687 (lr=6.3636e-04) (hash(x)=147054835)
378
+ 3120 train 4.331788 (lr=6.3521e-04) (hash(x)=152468568)
379
+ 3130 train 4.275879 (lr=6.3405e-04) (hash(x)=151635194)
380
+ 3140 train 4.237820 (lr=6.3290e-04) (hash(x)=150396274)
381
+ 3150 train 4.285441 (lr=6.3174e-04) (hash(x)=141306267)
382
+ 3160 train 4.240428 (lr=6.3058e-04) (hash(x)=151110762)
383
+ 3170 train 4.315084 (lr=6.2941e-04) (hash(x)=146518875)
384
+ 3180 train 4.199785 (lr=6.2825e-04) (hash(x)=139914401)
385
+ 3190 train 4.211347 (lr=6.2708e-04) (hash(x)=161631880)
386
+ 3200 val loss 4.3223
387
+ 3200 val perplexity 75.3623
388
+ 3200 train 4.233214 (lr=6.2590e-04) (hash(x)=150471842)
389
+ 3210 train 4.301362 (lr=6.2473e-04) (hash(x)=147355138)
390
+ 3220 train 4.097013 (lr=6.2355e-04) (hash(x)=138879678)
391
+ 3230 train 4.205156 (lr=6.2237e-04) (hash(x)=123929679)
392
+ 3240 train 4.269939 (lr=6.2119e-04) (hash(x)=157263734)
393
+ 3250 train 4.280121 (lr=6.2000e-04) (hash(x)=148065539)
394
+ 3260 train 4.197414 (lr=6.1881e-04) (hash(x)=138020742)
395
+ 3270 train 4.060056 (lr=6.1762e-04) (hash(x)=142053611)
396
+ 3280 train 4.220547 (lr=6.1643e-04) (hash(x)=155444784)
397
+ 3290 train 4.044999 (lr=6.1523e-04) (hash(x)=148500315)
398
+ 3300 val loss 4.3126
399
+ 3300 val perplexity 74.6351
400
+ 3300 train 4.274127 (lr=6.1403e-04) (hash(x)=149048126)
401
+ 3310 train 4.230175 (lr=6.1283e-04) (hash(x)=141680674)
402
+ 3320 train 4.171342 (lr=6.1163e-04) (hash(x)=143310642)
403
+ 3330 train 4.288915 (lr=6.1042e-04) (hash(x)=149891496)
404
+ 3340 train 4.236863 (lr=6.0921e-04) (hash(x)=151244609)
405
+ 3350 train 4.244311 (lr=6.0800e-04) (hash(x)=157158479)
406
+ 3360 train 4.104171 (lr=6.0679e-04) (hash(x)=151709876)
407
+ 3370 train 4.254009 (lr=6.0557e-04) (hash(x)=145297090)
408
+ 3380 train 4.268065 (lr=6.0435e-04) (hash(x)=154767881)
409
+ 3390 train 4.241918 (lr=6.0313e-04) (hash(x)=152600366)
410
+ 3400 val loss 4.2918
411
+ 3400 val perplexity 73.0968
412
+ 3400 train 4.478643 (lr=6.0191e-04) (hash(x)=161261339)
413
+ 3410 train 4.230458 (lr=6.0068e-04) (hash(x)=146859162)
414
+ 3420 train 4.328227 (lr=5.9945e-04) (hash(x)=149276224)
415
+ 3430 train 4.334190 (lr=5.9822e-04) (hash(x)=155866933)
416
+ 3440 train 4.201931 (lr=5.9699e-04) (hash(x)=143191904)
417
+ 3450 train 4.416255 (lr=5.9576e-04) (hash(x)=156910661)
418
+ 3460 train 4.212954 (lr=5.9452e-04) (hash(x)=140850879)
419
+ 3470 train 4.433090 (lr=5.9328e-04) (hash(x)=162140445)
420
+ 3480 train 4.180176 (lr=5.9204e-04) (hash(x)=145607495)
421
+ 3490 train 4.187631 (lr=5.9080e-04) (hash(x)=152496121)
422
+ 3500 val loss 4.2844
423
+ 3500 val perplexity 72.5582
424
+ 3500 train 4.154807 (lr=5.8955e-04) (hash(x)=157495564)
425
+ 3510 train 4.284398 (lr=5.8830e-04) (hash(x)=147657004)
426
+ 3520 train 4.250596 (lr=5.8705e-04) (hash(x)=152978060)
427
+ 3530 train 4.188976 (lr=5.8580e-04) (hash(x)=146800180)
428
+ 3540 train 4.262934 (lr=5.8454e-04) (hash(x)=195450444)
429
+ 3550 train 4.244950 (lr=5.8329e-04) (hash(x)=151114129)
430
+ 3560 train 4.227644 (lr=5.8203e-04) (hash(x)=143833190)
431
+ 3570 train 4.032177 (lr=5.8077e-04) (hash(x)=146039971)
432
+ 3580 train 4.100389 (lr=5.7951e-04) (hash(x)=145172249)
433
+ 3590 train 4.020234 (lr=5.7824e-04) (hash(x)=159164125)
434
+ 3600 val loss 4.2817
435
+ 3600 val perplexity 72.3655
436
+ 3600 train 4.261386 (lr=5.7697e-04) (hash(x)=144352932)
437
+ 3610 train 4.223984 (lr=5.7571e-04) (hash(x)=150550670)
438
+ 3620 train 4.349116 (lr=5.7443e-04) (hash(x)=147170125)
439
+ 3630 train 4.148612 (lr=5.7316e-04) (hash(x)=139473935)
440
+ 3640 train 4.137799 (lr=5.7189e-04) (hash(x)=159294424)
441
+ 3650 train 4.271544 (lr=5.7061e-04) (hash(x)=145943278)
442
+ 3660 train 4.315137 (lr=5.6933e-04) (hash(x)=141647872)
443
+ 3670 train 4.294239 (lr=5.6805e-04) (hash(x)=147109351)
444
+ 3680 train 4.321830 (lr=5.6677e-04) (hash(x)=140683368)
445
+ 3690 train 4.231912 (lr=5.6549e-04) (hash(x)=156276284)
446
+ 3700 val loss 4.2653
447
+ 3700 val perplexity 71.1882
448
+ 3700 train 4.178625 (lr=5.6420e-04) (hash(x)=149389012)
449
+ 3710 train 4.312306 (lr=5.6291e-04) (hash(x)=142107929)
450
+ 3720 train 4.281901 (lr=5.6162e-04) (hash(x)=147828843)
451
+ 3730 train 4.349348 (lr=5.6033e-04) (hash(x)=141784679)
452
+ 3740 train 4.247544 (lr=5.5904e-04) (hash(x)=153375464)
453
+ 3750 train 4.191781 (lr=5.5774e-04) (hash(x)=153716001)
454
+ 3760 train 4.152761 (lr=5.5645e-04) (hash(x)=146848822)
455
+ 3770 train 4.113221 (lr=5.5515e-04) (hash(x)=148396782)
456
+ 3780 train 4.176365 (lr=5.5385e-04) (hash(x)=153563244)
457
+ 3790 train 4.089226 (lr=5.5255e-04) (hash(x)=160808734)
458
+ 3800 val loss 4.2589
459
+ 3800 val perplexity 70.7327
460
+ 3800 train 4.282885 (lr=5.5125e-04) (hash(x)=146607620)
461
+ 3810 train 4.252675 (lr=5.4994e-04) (hash(x)=152580677)
462
+ 3820 train 4.604780 (lr=5.4864e-04) (hash(x)=144860879)
463
+ 3830 train 4.264771 (lr=5.4733e-04) (hash(x)=162099851)
464
+ 3840 train 4.221808 (lr=5.4602e-04) (hash(x)=159506855)
465
+ 3850 train 4.150459 (lr=5.4471e-04) (hash(x)=140022686)
466
+ 3860 train 4.219015 (lr=5.4340e-04) (hash(x)=152137850)
467
+ 3870 train 4.344110 (lr=5.4208e-04) (hash(x)=142666785)
468
+ 3880 train 4.193684 (lr=5.4077e-04) (hash(x)=148285868)
469
+ 3890 train 4.154398 (lr=5.3945e-04) (hash(x)=147910067)
470
+ 3900 val loss 4.2583
471
+ 3900 val perplexity 70.6884
472
+ 3900 train 4.197559 (lr=5.3813e-04) (hash(x)=143754617)
473
+ 3910 train 4.187465 (lr=5.3681e-04) (hash(x)=154779949)
474
+ 3920 train 4.031582 (lr=5.3549e-04) (hash(x)=137824061)
475
+ 3930 train 4.007197 (lr=5.3417e-04) (hash(x)=136933373)
476
+ 3940 train 4.078381 (lr=5.3284e-04) (hash(x)=164054337)
477
+ 3950 train 4.270173 (lr=5.3152e-04) (hash(x)=155752984)
478
+ 3960 train 4.386187 (lr=5.3019e-04) (hash(x)=160957673)
479
+ 3970 train 4.256425 (lr=5.2886e-04) (hash(x)=157158690)
480
+ 3980 train 4.260518 (lr=5.2754e-04) (hash(x)=153282027)
481
+ 3990 train 4.293380 (lr=5.2620e-04) (hash(x)=152183510)
482
+ 4000 val loss 4.2317
483
+ 4000 val perplexity 68.8314
484
+ 4000 train 4.182569 (lr=5.2487e-04) (hash(x)=156930722)
485
+ 4010 train 4.176345 (lr=5.2354e-04) (hash(x)=150798740)
486
+ 4020 train 4.387137 (lr=5.2221e-04) (hash(x)=142632400)
487
+ 4030 train 4.284315 (lr=5.2087e-04) (hash(x)=173326287)
488
+ 4040 train 4.193781 (lr=5.1953e-04) (hash(x)=128797443)
489
+ 4050 train 4.180554 (lr=5.1820e-04) (hash(x)=146937792)
490
+ 4060 train 4.282646 (lr=5.1686e-04) (hash(x)=164579030)
491
+ 4070 train 4.153031 (lr=5.1552e-04) (hash(x)=149459820)
492
+ 4080 train 4.118611 (lr=5.1418e-04) (hash(x)=160159595)
493
+ 4090 train 4.062393 (lr=5.1284e-04) (hash(x)=141507853)
494
+ 4100 val loss 4.2298
495
+ 4100 val perplexity 68.7044
496
+ 4100 train 4.206081 (lr=5.1149e-04) (hash(x)=147500519)
497
+ 4110 train 4.173779 (lr=5.1015e-04) (hash(x)=153297081)
498
+ 4120 train 4.188906 (lr=5.0880e-04) (hash(x)=142560288)
499
+ 4130 train 4.190455 (lr=5.0746e-04) (hash(x)=153380118)
500
+ 4140 train 4.104740 (lr=5.0611e-04) (hash(x)=143150720)
501
+ 4150 train 4.147567 (lr=5.0476e-04) (hash(x)=143459564)
502
+ 4160 train 4.173014 (lr=5.0341e-04) (hash(x)=153955741)
503
+ 4170 train 4.291698 (lr=5.0206e-04) (hash(x)=152199253)
504
+ 4180 train 4.049597 (lr=5.0071e-04) (hash(x)=165957668)
505
+ 4190 train 4.092456 (lr=4.9936e-04) (hash(x)=144866923)
506
+ 4200 val loss 4.2356
507
+ 4200 val perplexity 69.1065
508
+ 4200 train 4.105317 (lr=4.9801e-04) (hash(x)=143232237)
509
+ 4210 train 4.015755 (lr=4.9665e-04) (hash(x)=151907163)
510
+ 4220 train 4.120932 (lr=4.9530e-04) (hash(x)=145917587)
511
+ 4230 train 3.956774 (lr=4.9395e-04) (hash(x)=147258846)
512
+ 4240 train 4.106093 (lr=4.9259e-04) (hash(x)=143449843)
513
+ 4250 train 4.073434 (lr=4.9123e-04) (hash(x)=184883715)
514
+ 4260 train 4.170238 (lr=4.8988e-04) (hash(x)=156958761)
515
+ 4270 train 4.231820 (lr=4.8852e-04) (hash(x)=168870544)
516
+ 4280 train 4.168087 (lr=4.8716e-04) (hash(x)=179446735)
517
+ 4290 train 4.205977 (lr=4.8580e-04) (hash(x)=145649386)
518
+ 4300 val loss 4.2210
519
+ 4300 val perplexity 68.1028
520
+ 4300 train 4.054780 (lr=4.8444e-04) (hash(x)=146811670)
521
+ 4310 train 4.328610 (lr=4.8308e-04) (hash(x)=151907075)
522
+ 4320 train 4.133140 (lr=4.8172e-04) (hash(x)=138628365)
523
+ 4330 train 4.185603 (lr=4.8036e-04) (hash(x)=151608240)
524
+ 4340 train 4.116433 (lr=4.7899e-04) (hash(x)=160111319)
525
+ 4350 train 4.266597 (lr=4.7763e-04) (hash(x)=144958652)
526
+ 4360 train 4.056381 (lr=4.7627e-04) (hash(x)=149176222)
527
+ 4370 train 4.155463 (lr=4.7490e-04) (hash(x)=216221371)
528
+ 4380 train 4.061438 (lr=4.7354e-04) (hash(x)=141760648)
529
+ 4390 train 4.143332 (lr=4.7217e-04) (hash(x)=151710283)
530
+ 4400 val loss 4.2101
531
+ 4400 val perplexity 67.3642
532
+ 4400 train 4.270244 (lr=4.7081e-04) (hash(x)=158418746)
533
+ 4410 train 4.124296 (lr=4.6944e-04) (hash(x)=142331301)
534
+ 4420 train 4.208941 (lr=4.6807e-04) (hash(x)=146507180)
535
+ 4430 train 4.294528 (lr=4.6671e-04) (hash(x)=166506937)
536
+ 4440 train 4.237477 (lr=4.6534e-04) (hash(x)=151209447)
537
+ 4450 train 4.199465 (lr=4.6397e-04) (hash(x)=149186822)
538
+ 4460 train 4.184713 (lr=4.6260e-04) (hash(x)=144348412)
539
+ 4470 train 4.067492 (lr=4.6124e-04) (hash(x)=156494284)
540
+ 4480 train 4.095995 (lr=4.5987e-04) (hash(x)=147238744)
541
+ 4490 train 4.049746 (lr=4.5850e-04) (hash(x)=150746185)
542
+ 4500 val loss 4.2015
543
+ 4500 val perplexity 66.7868
544
+ 4500 train 4.142287 (lr=4.5713e-04) (hash(x)=156695778)
545
+ 4510 train 3.981408 (lr=4.5576e-04) (hash(x)=135988958)
546
+ 4520 train 4.154624 (lr=4.5439e-04) (hash(x)=148272110)
547
+ 4530 train 4.074677 (lr=4.5302e-04) (hash(x)=136459794)
548
+ 4540 train 4.103742 (lr=4.5165e-04) (hash(x)=147624954)
549
+ 4550 train 4.099844 (lr=4.5028e-04) (hash(x)=154947199)
550
+ 4560 train 4.165448 (lr=4.4891e-04) (hash(x)=147869166)
551
+ 4570 train 4.015043 (lr=4.4754e-04) (hash(x)=148415463)
552
+ 4580 train 4.179327 (lr=4.4617e-04) (hash(x)=152104066)
553
+ 4590 train 4.077621 (lr=4.4480e-04) (hash(x)=147060040)
554
+ 4600 val loss 4.1991
555
+ 4600 val perplexity 66.6281
556
+ 4600 train 4.150793 (lr=4.4343e-04) (hash(x)=147791497)
557
+ 4610 train 4.346067 (lr=4.4206e-04) (hash(x)=155595556)
558
+ 4620 train 4.195231 (lr=4.4069e-04) (hash(x)=147178858)
559
+ 4630 train 4.247233 (lr=4.3931e-04) (hash(x)=152617112)
560
+ 4640 train 4.245648 (lr=4.3794e-04) (hash(x)=161766869)
561
+ 4650 train 4.224687 (lr=4.3657e-04) (hash(x)=158075561)
562
+ 4660 train 4.265965 (lr=4.3520e-04) (hash(x)=145360002)
563
+ 4670 train 4.147497 (lr=4.3383e-04) (hash(x)=144538791)
564
+ 4680 train 4.189404 (lr=4.3246e-04) (hash(x)=142711273)
565
+ 4690 train 4.340482 (lr=4.3109e-04) (hash(x)=193907362)
566
+ 4700 val loss 4.1799
567
+ 4700 val perplexity 65.3612
568
+ 4700 train 4.159918 (lr=4.2972e-04) (hash(x)=155533088)
569
+ 4710 train 4.108403 (lr=4.2835e-04) (hash(x)=148618528)
570
+ 4720 train 4.219090 (lr=4.2698e-04) (hash(x)=150502624)
571
+ 4730 train 4.095981 (lr=4.2561e-04) (hash(x)=189205847)
572
+ 4740 train 4.154233 (lr=4.2424e-04) (hash(x)=167322800)
573
+ 4750 train 4.266942 (lr=4.2287e-04) (hash(x)=162169095)
574
+ 4760 train 3.967456 (lr=4.2150e-04) (hash(x)=143459469)
575
+ 4770 train 4.046465 (lr=4.2013e-04) (hash(x)=144386593)
576
+ 4780 train 4.012156 (lr=4.1876e-04) (hash(x)=143721184)
577
+ 4790 train 4.084726 (lr=4.1740e-04) (hash(x)=146077222)
578
+ 4800 val loss 4.1853
579
+ 4800 val perplexity 65.7132
580
+ 4800 train 4.058820 (lr=4.1603e-04) (hash(x)=138350044)
581
+ 4810 train 4.064875 (lr=4.1466e-04) (hash(x)=154960112)
582
+ 4820 train 3.979040 (lr=4.1329e-04) (hash(x)=146950685)
583
+ 4830 train 3.978209 (lr=4.1193e-04) (hash(x)=140224554)
584
+ 4840 train 3.970171 (lr=4.1056e-04) (hash(x)=156640495)
585
+ 4850 train 4.188550 (lr=4.0919e-04) (hash(x)=153409150)
586
+ 4860 train 4.101605 (lr=4.0783e-04) (hash(x)=151647700)
587
+ 4870 train 4.497338 (lr=4.0646e-04) (hash(x)=155465603)
588
+ 4880 train 4.340479 (lr=4.0510e-04) (hash(x)=153404400)
589
+ 4890 train 4.228568 (lr=4.0373e-04) (hash(x)=170409281)
590
+ 4900 val loss 4.1704
591
+ 4900 val perplexity 64.7442
592
+ 4900 train 4.089856 (lr=4.0237e-04) (hash(x)=143735284)
593
+ 4910 train 4.225789 (lr=4.0101e-04) (hash(x)=150784329)
594
+ 4920 train 4.114635 (lr=3.9964e-04) (hash(x)=152353574)
595
+ 4930 train 4.069668 (lr=3.9828e-04) (hash(x)=152677008)
596
+ 4940 train 4.146378 (lr=3.9692e-04) (hash(x)=140553861)
597
+ 4950 train 4.230979 (lr=3.9556e-04) (hash(x)=152869181)
598
+ 4960 train 4.188179 (lr=3.9420e-04) (hash(x)=153011079)
599
+ 4970 train 4.140185 (lr=3.9284e-04) (hash(x)=152796617)
600
+ 4980 train 4.250128 (lr=3.9148e-04) (hash(x)=146868417)
601
+ 4990 train 4.133134 (lr=3.9012e-04) (hash(x)=140168188)
602
+ 5000 val loss 4.1611
603
+ 5000 val perplexity 64.1389
604
+ 5000 train 4.116722 (lr=3.8877e-04) (hash(x)=154976463)
605
+ 5010 train 4.174135 (lr=3.8741e-04) (hash(x)=148582690)
606
+ 5020 train 4.185433 (lr=3.8605e-04) (hash(x)=166715464)
607
+ 5030 train 4.146249 (lr=3.8470e-04) (hash(x)=150616397)
608
+ 5040 train 4.189446 (lr=3.8335e-04) (hash(x)=150775548)
609
+ 5050 train 4.179484 (lr=3.8199e-04) (hash(x)=150575367)
610
+ 5060 train 4.099730 (lr=3.8064e-04) (hash(x)=152352082)
611
+ 5070 train 4.135840 (lr=3.7929e-04) (hash(x)=150269272)
612
+ 5080 train 4.195846 (lr=3.7794e-04) (hash(x)=146499361)
613
+ 5090 train 3.891378 (lr=3.7659e-04) (hash(x)=148734454)
614
+ 5100 val loss 4.1624
615
+ 5100 val perplexity 64.2287
616
+ 5100 train 4.068560 (lr=3.7524e-04) (hash(x)=149894982)
617
+ 5110 train 4.119891 (lr=3.7389e-04) (hash(x)=154487751)
618
+ 5120 train 4.066651 (lr=3.7254e-04) (hash(x)=144169582)
619
+ 5130 train 4.043821 (lr=3.7120e-04) (hash(x)=152217630)
620
+ 5140 train 4.071286 (lr=3.6985e-04) (hash(x)=158233831)
621
+ 5150 train 4.131513 (lr=3.6851e-04) (hash(x)=142746426)
622
+ 5160 train 4.124256 (lr=3.6716e-04) (hash(x)=139184458)
623
+ 5170 train 4.208778 (lr=3.6582e-04) (hash(x)=140624857)
624
+ 5180 train 3.930155 (lr=3.6448e-04) (hash(x)=154107423)
625
+ 5190 train 3.871901 (lr=3.6314e-04) (hash(x)=144892610)
626
+ 5200 val loss 4.1621
627
+ 5200 val perplexity 64.2047
628
+ 5200 train 3.821012 (lr=3.6180e-04) (hash(x)=159326689)
629
+ 5210 train 3.992097 (lr=3.6047e-04) (hash(x)=151186017)
630
+ 5220 train 3.794047 (lr=3.5913e-04) (hash(x)=148996819)
631
+ 5230 train 3.741768 (lr=3.5779e-04) (hash(x)=150829917)
632
+ 5240 train 3.865284 (lr=3.5646e-04) (hash(x)=152758972)
633
+ 5250 train 3.835927 (lr=3.5513e-04) (hash(x)=154345903)
634
+ 5260 train 3.826452 (lr=3.5380e-04) (hash(x)=150134947)
635
+ 5270 train 3.924219 (lr=3.5246e-04) (hash(x)=149689310)
636
+ 5280 train 4.156047 (lr=3.5114e-04) (hash(x)=147662223)
637
+ 5290 train 4.214880 (lr=3.4981e-04) (hash(x)=165770608)
638
+ 5300 val loss 4.1495
639
+ 5300 val perplexity 63.4028
640
+ 5300 train 4.201334 (lr=3.4848e-04) (hash(x)=159484800)
641
+ 5310 train 4.231984 (lr=3.4716e-04) (hash(x)=152672536)
642
+ 5320 train 4.054946 (lr=3.4583e-04) (hash(x)=152728812)
643
+ 5330 train 3.998456 (lr=3.4451e-04) (hash(x)=152778685)
644
+ 5340 train 4.098969 (lr=3.4319e-04) (hash(x)=149818849)
645
+ 5350 train 4.026422 (lr=3.4187e-04) (hash(x)=168239739)
646
+ 5360 train 4.174008 (lr=3.4055e-04) (hash(x)=139859423)
647
+ 5370 train 4.000448 (lr=3.3923e-04) (hash(x)=139315327)
648
+ 5380 train 4.211906 (lr=3.3792e-04) (hash(x)=163068509)
649
+ 5390 train 4.166430 (lr=3.3660e-04) (hash(x)=147645415)
650
+ 5400 val loss 4.1295
651
+ 5400 val perplexity 62.1449
652
+ 5400 train 4.121242 (lr=3.3529e-04) (hash(x)=140385615)
653
+ 5410 train 3.981606 (lr=3.3398e-04) (hash(x)=151381862)
654
+ 5420 train 4.132318 (lr=3.3267e-04) (hash(x)=152445092)
655
+ 5430 train 4.141646 (lr=3.3136e-04) (hash(x)=142454479)
656
+ 5440 train 4.097904 (lr=3.3006e-04) (hash(x)=151308771)
657
+ 5450 train 3.949624 (lr=3.2875e-04) (hash(x)=150886070)
658
+ 5460 train 4.057513 (lr=3.2745e-04) (hash(x)=146080408)
659
+ 5470 train 4.080599 (lr=3.2615e-04) (hash(x)=149828857)
660
+ 5480 train 4.051802 (lr=3.2485e-04) (hash(x)=139922034)
661
+ 5490 train 4.002710 (lr=3.2355e-04) (hash(x)=150262276)
662
+ 5500 val loss 4.1296
663
+ 5500 val perplexity 62.1531
664
+ 5500 train 3.970436 (lr=3.2226e-04) (hash(x)=148498335)
665
+ 5510 train 4.093485 (lr=3.2096e-04) (hash(x)=157036224)
666
+ 5520 train 3.967459 (lr=3.1967e-04) (hash(x)=134093409)
667
+ 5530 train 3.995745 (lr=3.1838e-04) (hash(x)=153126250)
668
+ 5540 train 4.007802 (lr=3.1709e-04) (hash(x)=136170985)
669
+ 5550 train 3.985316 (lr=3.1580e-04) (hash(x)=152841201)
670
+ 5560 train 4.092659 (lr=3.1451e-04) (hash(x)=147439014)
671
+ 5570 train 4.004538 (lr=3.1323e-04) (hash(x)=143476132)
672
+ 5580 train 3.777183 (lr=3.1195e-04) (hash(x)=139503437)
673
+ 5590 train 3.821054 (lr=3.1067e-04) (hash(x)=131664955)
674
+ 5600 val loss 4.1379
675
+ 5600 val perplexity 62.6738
676
+ 5600 train 3.918692 (lr=3.0939e-04) (hash(x)=151907614)
677
+ 5610 train 3.807151 (lr=3.0811e-04) (hash(x)=157132674)
678
+ 5620 train 4.007532 (lr=3.0684e-04) (hash(x)=146793775)
679
+ 5630 train 3.952494 (lr=3.0557e-04) (hash(x)=147237910)
680
+ 5640 train 3.865453 (lr=3.0429e-04) (hash(x)=141120024)
681
+ 5650 train 3.831379 (lr=3.0303e-04) (hash(x)=144275645)
682
+ 5660 train 3.870418 (lr=3.0176e-04) (hash(x)=159088002)
683
+ 5670 train 3.927907 (lr=3.0049e-04) (hash(x)=147224479)
684
+ 5680 train 4.275138 (lr=2.9923e-04) (hash(x)=147208743)
685
+ 5690 train 4.074907 (lr=2.9797e-04) (hash(x)=154384701)
686
+ 5700 val loss 4.1309
687
+ 5700 val perplexity 62.2333
688
+ 5700 train 4.107089 (lr=2.9671e-04) (hash(x)=155192267)
689
+ 5710 train 4.051459 (lr=2.9546e-04) (hash(x)=148458753)
690
+ 5720 train 4.158017 (lr=2.9420e-04) (hash(x)=159947204)
691
+ 5730 train 4.144844 (lr=2.9295e-04) (hash(x)=150746366)
692
+ 5740 train 4.140504 (lr=2.9170e-04) (hash(x)=155201884)
693
+ 5750 train 4.139308 (lr=2.9045e-04) (hash(x)=156432008)
694
+ 5760 train 4.122699 (lr=2.8920e-04) (hash(x)=185977746)
695
+ 5770 train 4.124070 (lr=2.8796e-04) (hash(x)=133620000)
696
+ 5780 train 4.127920 (lr=2.8672e-04) (hash(x)=153778044)
697
+ 5790 train 3.978884 (lr=2.8548e-04) (hash(x)=147713822)
698
+ 5800 val loss 4.1103
699
+ 5800 val perplexity 60.9663
700
+ 5800 train 4.128989 (lr=2.8424e-04) (hash(x)=153132158)
701
+ 5810 train 4.090636 (lr=2.8301e-04) (hash(x)=144293895)
702
+ 5820 train 4.067160 (lr=2.8178e-04) (hash(x)=151500590)
703
+ 5830 train 4.016610 (lr=2.8055e-04) (hash(x)=139308707)
704
+ 5840 train 4.131730 (lr=2.7932e-04) (hash(x)=153591278)
705
+ 5850 train 4.109837 (lr=2.7809e-04) (hash(x)=146077037)
706
+ 5860 train 4.076795 (lr=2.7687e-04) (hash(x)=149160733)
707
+ 5870 train 4.054731 (lr=2.7565e-04) (hash(x)=138482476)
708
+ 5880 train 4.070661 (lr=2.7443e-04) (hash(x)=152173137)
709
+ 5890 train 4.047036 (lr=2.7321e-04) (hash(x)=142387057)
710
+ 5900 val loss 4.1189
711
+ 5900 val perplexity 61.4915
712
+ 5900 train 4.125184 (lr=2.7200e-04) (hash(x)=161446764)
713
+ 5910 train 3.992238 (lr=2.7079e-04) (hash(x)=144512138)
714
+ 5920 train 4.056687 (lr=2.6958e-04) (hash(x)=152555618)
715
+ 5930 train 4.011394 (lr=2.6837e-04) (hash(x)=146129367)
716
+ 5940 train 3.956703 (lr=2.6717e-04) (hash(x)=154607527)
717
+ 5950 train 4.040770 (lr=2.6597e-04) (hash(x)=143827143)
718
+ 5960 train 4.177609 (lr=2.6477e-04) (hash(x)=151168510)
719
+ 5970 train 3.959239 (lr=2.6357e-04) (hash(x)=158961574)
720
+ 5980 train 3.787388 (lr=2.6238e-04) (hash(x)=148757787)
721
+ 5990 train 4.294996 (lr=2.6119e-04) (hash(x)=161719494)
722
+ 6000 val loss 4.1197
723
+ 6000 val perplexity 61.5419
724
+ 6000 train 3.795982 (lr=2.6000e-04) (hash(x)=151512446)
725
+ 6010 train 3.907936 (lr=2.5881e-04) (hash(x)=173500735)
726
+ 6020 train 4.051758 (lr=2.5763e-04) (hash(x)=157621031)
727
+ 6030 train 3.935310 (lr=2.5645e-04) (hash(x)=142662684)
728
+ 6040 train 3.915016 (lr=2.5527e-04) (hash(x)=146180940)
729
+ 6050 train 3.836264 (lr=2.5410e-04) (hash(x)=147654412)
730
+ 6060 train 3.712128 (lr=2.5292e-04) (hash(x)=156393157)
731
+ 6070 train 4.106809 (lr=2.5175e-04) (hash(x)=156947300)
732
+ 6080 train 4.109121 (lr=2.5059e-04) (hash(x)=163108626)
733
+ 6090 train 4.204768 (lr=2.4942e-04) (hash(x)=158654713)
734
+ 6100 val loss 4.0992
735
+ 6100 val perplexity 60.2897
736
+ 6100 train 3.932790 (lr=2.4826e-04) (hash(x)=188094053)
737
+ 6110 train 4.087243 (lr=2.4710e-04) (hash(x)=145132600)
738
+ 6120 train 3.955223 (lr=2.4595e-04) (hash(x)=146999707)
739
+ 6130 train 4.198400 (lr=2.4479e-04) (hash(x)=145719530)
740
+ 6140 train 4.134458 (lr=2.4364e-04) (hash(x)=153230722)
741
+ 6150 train 4.068529 (lr=2.4250e-04) (hash(x)=147455934)
742
+ 6160 train 4.114559 (lr=2.4135e-04) (hash(x)=146596242)
743
+ 6170 train 4.211110 (lr=2.4021e-04) (hash(x)=152029855)
744
+ 6180 train 4.076581 (lr=2.3907e-04) (hash(x)=151480034)
745
+ 6190 train 4.007696 (lr=2.3793e-04) (hash(x)=147381047)
746
+ 6200 val loss 4.0853
747
+ 6200 val perplexity 59.4586
748
+ 6200 train 3.996751 (lr=2.3680e-04) (hash(x)=149389789)
749
+ 6210 train 4.051102 (lr=2.3567e-04) (hash(x)=154894138)
750
+ 6220 train 4.085030 (lr=2.3454e-04) (hash(x)=150074302)
751
+ 6230 train 3.949676 (lr=2.3342e-04) (hash(x)=142881794)
752
+ 6240 train 3.935044 (lr=2.3230e-04) (hash(x)=164747461)
753
+ 6250 train 4.099629 (lr=2.3118e-04) (hash(x)=158378292)
754
+ 6260 train 4.108934 (lr=2.3006e-04) (hash(x)=147675938)
755
+ 6270 train 4.050026 (lr=2.2895e-04) (hash(x)=148197228)
756
+ 6280 train 4.013082 (lr=2.2784e-04) (hash(x)=135296747)
757
+ 6290 train 4.083632 (lr=2.2674e-04) (hash(x)=156439346)
758
+ 6300 val loss 4.0846
759
+ 6300 val perplexity 59.4206
760
+ 6300 train 3.910562 (lr=2.2563e-04) (hash(x)=138212820)
761
+ 6310 train 4.199171 (lr=2.2453e-04) (hash(x)=165985336)
762
+ 6320 train 4.070144 (lr=2.2344e-04) (hash(x)=145921897)
763
+ 6330 train 4.282308 (lr=2.2234e-04) (hash(x)=145502689)
764
+ 6340 train 4.013851 (lr=2.2125e-04) (hash(x)=141704937)
765
+ 6350 train 4.063880 (lr=2.2017e-04) (hash(x)=161789400)
766
+ 6360 train 3.792254 (lr=2.1908e-04) (hash(x)=145457780)
767
+ 6370 train 4.048292 (lr=2.1800e-04) (hash(x)=171628530)
768
+ 6380 train 3.834210 (lr=2.1692e-04) (hash(x)=153467923)
769
+ 6390 train 4.208705 (lr=2.1585e-04) (hash(x)=158066904)
770
+ 6400 val loss 4.0903
771
+ 6400 val perplexity 59.7567
772
+ 6400 train 3.863652 (lr=2.1478e-04) (hash(x)=146535423)
773
+ 6410 train 3.720085 (lr=2.1371e-04) (hash(x)=147264155)
774
+ 6420 train 3.828033 (lr=2.1265e-04) (hash(x)=145595485)
775
+ 6430 train 3.972567 (lr=2.1159e-04) (hash(x)=146929074)
776
+ 6440 train 3.902391 (lr=2.1053e-04) (hash(x)=147290878)
777
+ 6450 train 3.994897 (lr=2.0947e-04) (hash(x)=141154205)
778
+ 6460 train 3.840040 (lr=2.0842e-04) (hash(x)=154360901)
779
+ 6470 train 4.177209 (lr=2.0737e-04) (hash(x)=146350558)
780
+ 6480 train 4.137039 (lr=2.0633e-04) (hash(x)=153927683)
781
+ 6490 train 4.048942 (lr=2.0529e-04) (hash(x)=149337047)
782
+ 6500 val loss 4.0774
783
+ 6500 val perplexity 58.9914
784
+ 6500 train 3.971519 (lr=2.0425e-04) (hash(x)=145950843)
785
+ 6510 train 4.037339 (lr=2.0322e-04) (hash(x)=142577653)
786
+ 6520 train 4.099168 (lr=2.0218e-04) (hash(x)=165060689)
787
+ 6530 train 3.940073 (lr=2.0116e-04) (hash(x)=148367505)
788
+ 6540 train 4.037755 (lr=2.0013e-04) (hash(x)=154313651)
789
+ 6550 train 3.954875 (lr=1.9911e-04) (hash(x)=147477315)
790
+ 6560 train 4.045574 (lr=1.9810e-04) (hash(x)=139825403)
791
+ 6570 train 4.059897 (lr=1.9708e-04) (hash(x)=141482518)
792
+ 6580 train 4.092167 (lr=1.9607e-04) (hash(x)=144968144)
793
+ 6590 train 4.044579 (lr=1.9507e-04) (hash(x)=154319720)
794
+ 6600 val loss 4.0736
795
+ 6600 val perplexity 58.7704
796
+ 6600 train 4.086018 (lr=1.9406e-04) (hash(x)=141162902)
797
+ 6610 train 4.043385 (lr=1.9306e-04) (hash(x)=159490449)
798
+ 6620 train 3.934186 (lr=1.9207e-04) (hash(x)=151189772)
799
+ 6630 train 3.912117 (lr=1.9108e-04) (hash(x)=148451651)
800
+ 6640 train 4.033502 (lr=1.9009e-04) (hash(x)=152815550)
801
+ 6650 train 3.889731 (lr=1.8910e-04) (hash(x)=144316094)
802
+ 6660 train 4.211576 (lr=1.8812e-04) (hash(x)=143951240)
803
+ 6670 train 3.824594 (lr=1.8714e-04) (hash(x)=143203472)
804
+ 6680 train 3.875219 (lr=1.8617e-04) (hash(x)=132063806)
805
+ 6690 train 3.963911 (lr=1.8520e-04) (hash(x)=140739750)
806
+ 6700 val loss 4.0769
807
+ 6700 val perplexity 58.9650
808
+ 6700 train 4.021381 (lr=1.8423e-04) (hash(x)=153018737)
809
+ 6710 train 3.887647 (lr=1.8327e-04) (hash(x)=140107197)
810
+ 6720 train 3.990199 (lr=1.8231e-04) (hash(x)=144396559)
811
+ 6730 train 3.837094 (lr=1.8136e-04) (hash(x)=148038991)
812
+ 6740 train 4.046952 (lr=1.8040e-04) (hash(x)=145504746)
813
+ 6750 train 4.036767 (lr=1.7946e-04) (hash(x)=119579094)
814
+ 6760 train 4.044994 (lr=1.7851e-04) (hash(x)=161725430)
815
+ 6770 train 4.066007 (lr=1.7757e-04) (hash(x)=153646966)
816
+ 6780 train 4.079191 (lr=1.7663e-04) (hash(x)=153838463)
817
+ 6790 train 4.198527 (lr=1.7570e-04) (hash(x)=157891196)
818
+ 6800 val loss 4.0595
819
+ 6800 val perplexity 57.9444
820
+ 6800 train 3.912669 (lr=1.7477e-04) (hash(x)=155640155)
821
+ 6810 train 4.106999 (lr=1.7385e-04) (hash(x)=154913809)
822
+ 6820 train 4.020285 (lr=1.7293e-04) (hash(x)=159412493)
823
+ 6830 train 3.983441 (lr=1.7201e-04) (hash(x)=144756549)
824
+ 6840 train 4.199935 (lr=1.7110e-04) (hash(x)=143824638)
825
+ 6850 train 4.519302 (lr=1.7019e-04) (hash(x)=117180982)
826
+ 6860 train 4.046760 (lr=1.6928e-04) (hash(x)=141887150)
827
+ 6870 train 4.110474 (lr=1.6838e-04) (hash(x)=140094872)
828
+ 6880 train 3.953631 (lr=1.6748e-04) (hash(x)=141479953)
829
+ 6890 train 4.010381 (lr=1.6659e-04) (hash(x)=153166840)
830
+ 6900 val loss 4.0575
831
+ 6900 val perplexity 57.8270
832
+ 6900 train 4.036755 (lr=1.6570e-04) (hash(x)=153722115)
833
+ 6910 train 4.143244 (lr=1.6481e-04) (hash(x)=149578602)
834
+ 6920 train 4.002674 (lr=1.6393e-04) (hash(x)=149925108)
835
+ 6930 train 3.945015 (lr=1.6305e-04) (hash(x)=148589489)
836
+ 6940 train 3.857909 (lr=1.6218e-04) (hash(x)=144243794)
837
+ 6950 train 3.941780 (lr=1.6131e-04) (hash(x)=157743480)
838
+ 6960 train 3.923009 (lr=1.6044e-04) (hash(x)=139858247)
839
+ 6970 train 3.851303 (lr=1.5958e-04) (hash(x)=148727536)
840
+ 6980 train 3.965064 (lr=1.5872e-04) (hash(x)=153382595)
841
+ 6990 train 4.115118 (lr=1.5787e-04) (hash(x)=147529919)
842
+ 7000 val loss 4.0517
843
+ 7000 val perplexity 57.4949
844
+ 7000 train 4.071392 (lr=1.5702e-04) (hash(x)=146953450)
845
+ 7010 train 3.986660 (lr=1.5618e-04) (hash(x)=134561015)
846
+ 7020 train 4.097430 (lr=1.5533e-04) (hash(x)=148433515)
847
+ 7030 train 3.946738 (lr=1.5450e-04) (hash(x)=140394326)
848
+ 7040 train 3.948417 (lr=1.5366e-04) (hash(x)=136740763)
849
+ 7050 train 3.787003 (lr=1.5284e-04) (hash(x)=148637938)
850
+ 7060 train 3.955077 (lr=1.5201e-04) (hash(x)=147262147)
851
+ 7070 train 3.934434 (lr=1.5119e-04) (hash(x)=140191039)
852
+ 7080 train 3.950030 (lr=1.5037e-04) (hash(x)=148223305)
853
+ 7090 train 3.867564 (lr=1.4956e-04) (hash(x)=139984988)
854
+ 7100 val loss 4.0560
855
+ 7100 val perplexity 57.7409
856
+ 7100 train 3.974286 (lr=1.4875e-04) (hash(x)=137663885)
857
+ 7110 train 3.858830 (lr=1.4795e-04) (hash(x)=150955021)
858
+ 7120 train 3.768010 (lr=1.4715e-04) (hash(x)=148585682)
859
+ 7130 train 3.963603 (lr=1.4636e-04) (hash(x)=146055548)
860
+ 7140 train 4.011766 (lr=1.4556e-04) (hash(x)=155723541)
861
+ 7150 train 4.085651 (lr=1.4478e-04) (hash(x)=140465834)
862
+ 7160 train 4.050817 (lr=1.4400e-04) (hash(x)=163246922)
863
+ 7170 train 3.911832 (lr=1.4322e-04) (hash(x)=150754166)
864
+ 7180 train 4.446715 (lr=1.4244e-04) (hash(x)=142844151)
865
+ 7190 train 4.034914 (lr=1.4167e-04) (hash(x)=145167470)
866
+ 7200 val loss 4.0404
867
+ 7200 val perplexity 56.8488
868
+ 7200 train 4.226603 (lr=1.4091e-04) (hash(x)=146172950)
869
+ 7210 train 4.056651 (lr=1.4015e-04) (hash(x)=143155287)
870
+ 7220 train 4.000681 (lr=1.3939e-04) (hash(x)=151976669)
871
+ 7230 train 4.115868 (lr=1.3864e-04) (hash(x)=145979940)
872
+ 7240 train 4.084537 (lr=1.3789e-04) (hash(x)=155656034)
873
+ 7250 train 4.087013 (lr=1.3715e-04) (hash(x)=135610136)
874
+ 7260 train 4.166543 (lr=1.3641e-04) (hash(x)=148133023)
875
+ 7270 train 4.036417 (lr=1.3568e-04) (hash(x)=148006681)
876
+ 7280 train 4.078573 (lr=1.3495e-04) (hash(x)=151397627)
877
+ 7290 train 4.034358 (lr=1.3422e-04) (hash(x)=141696871)
878
+ 7300 val loss 4.0361
879
+ 7300 val perplexity 56.6073
880
+ 7300 train 3.969224 (lr=1.3350e-04) (hash(x)=150018163)
881
+ 7310 train 4.009308 (lr=1.3278e-04) (hash(x)=154900050)
882
+ 7320 train 3.908590 (lr=1.3207e-04) (hash(x)=142608283)
883
+ 7330 train 4.010306 (lr=1.3136e-04) (hash(x)=148977837)
884
+ 7340 train 4.026961 (lr=1.3066e-04) (hash(x)=141676907)
885
+ 7350 train 3.913811 (lr=1.2996e-04) (hash(x)=147846997)
886
+ 7360 train 4.161057 (lr=1.2926e-04) (hash(x)=143252181)
887
+ 7370 train 3.971089 (lr=1.2857e-04) (hash(x)=155081717)
888
+ 7380 train 4.065037 (lr=1.2789e-04) (hash(x)=149678451)
889
+ 7390 train 3.895566 (lr=1.2721e-04) (hash(x)=159341637)
890
+ 7400 val loss 4.0371
891
+ 7400 val perplexity 56.6615
892
+ 7400 train 4.176729 (lr=1.2653e-04) (hash(x)=145351166)
893
+ 7410 train 3.995462 (lr=1.2586e-04) (hash(x)=162274009)
894
+ 7420 train 3.837331 (lr=1.2519e-04) (hash(x)=143254858)
895
+ 7430 train 3.953107 (lr=1.2453e-04) (hash(x)=143049884)
896
+ 7440 train 3.873784 (lr=1.2387e-04) (hash(x)=141768538)
897
+ 7450 train 3.900073 (lr=1.2322e-04) (hash(x)=139895310)
898
+ 7460 train 3.773922 (lr=1.2257e-04) (hash(x)=155900365)
899
+ 7470 train 3.757522 (lr=1.2192e-04) (hash(x)=145048215)
900
+ 7480 train 3.898866 (lr=1.2128e-04) (hash(x)=136992624)
901
+ 7490 train 4.006968 (lr=1.2065e-04) (hash(x)=141424789)
902
+ 7500 val loss 4.0419
903
+ 7500 val perplexity 56.9356
904
+ 7500 train 3.862612 (lr=1.2002e-04) (hash(x)=145292116)
905
+ 7510 train 3.896114 (lr=1.1939e-04) (hash(x)=144818029)
906
+ 7520 train 4.063148 (lr=1.1877e-04) (hash(x)=152903515)
907
+ 7530 train 4.069553 (lr=1.1816e-04) (hash(x)=156934724)
908
+ 7540 train 4.297395 (lr=1.1754e-04) (hash(x)=156695926)
909
+ 7550 train 4.058777 (lr=1.1694e-04) (hash(x)=149115609)
910
+ 7560 train 4.034388 (lr=1.1633e-04) (hash(x)=143974224)
911
+ 7570 train 3.991103 (lr=1.1574e-04) (hash(x)=148204424)
912
+ 7580 train 4.069330 (lr=1.1514e-04) (hash(x)=150177910)
913
+ 7590 train 4.080345 (lr=1.1455e-04) (hash(x)=140325629)
914
+ 7600 val loss 4.0283
915
+ 7600 val perplexity 56.1681
916
+ 7600 train 3.977276 (lr=1.1397e-04) (hash(x)=150235132)
917
+ 7610 train 4.097807 (lr=1.1339e-04) (hash(x)=157594837)
918
+ 7620 train 4.103137 (lr=1.1282e-04) (hash(x)=146696749)
919
+ 7630 train 4.141193 (lr=1.1225e-04) (hash(x)=148463580)
920
+ 7640 train 3.966796 (lr=1.1168e-04) (hash(x)=140271046)
921
+ 7650 train 4.017224 (lr=1.1112e-04) (hash(x)=159952319)
922
+ 7660 train 4.021734 (lr=1.1057e-04) (hash(x)=146307392)
923
+ 7670 train 3.967839 (lr=1.1002e-04) (hash(x)=145027988)
924
+ 7680 train 4.038776 (lr=1.0947e-04) (hash(x)=147729698)
925
+ 7690 train 4.021869 (lr=1.0893e-04) (hash(x)=143401210)
926
+ 7700 val loss 4.0228
927
+ 7700 val perplexity 55.8595
928
+ 7700 train 3.957328 (lr=1.0840e-04) (hash(x)=154543455)
929
+ 7710 train 4.028483 (lr=1.0786e-04) (hash(x)=151600456)
930
+ 7720 train 3.946132 (lr=1.0734e-04) (hash(x)=165913803)
931
+ 7730 train 3.818464 (lr=1.0682e-04) (hash(x)=144942059)
932
+ 7740 train 4.041981 (lr=1.0630e-04) (hash(x)=150302103)
933
+ 7750 train 3.983673 (lr=1.0579e-04) (hash(x)=155782458)
934
+ 7760 train 4.017528 (lr=1.0528e-04) (hash(x)=147824384)
935
+ 7770 train 4.032617 (lr=1.0478e-04) (hash(x)=149177657)
936
+ 7780 train 4.048721 (lr=1.0428e-04) (hash(x)=164446738)
937
+ 7790 train 3.856337 (lr=1.0379e-04) (hash(x)=146417077)
938
+ 7800 val loss 4.0261
939
+ 7800 val perplexity 56.0420
940
+ 7800 train 3.999558 (lr=1.0330e-04) (hash(x)=142456852)
941
+ 7810 train 3.704015 (lr=1.0282e-04) (hash(x)=146221807)
942
+ 7820 train 3.898252 (lr=1.0234e-04) (hash(x)=154136533)
943
+ 7830 train 3.808636 (lr=1.0187e-04) (hash(x)=147492735)
944
+ 7840 train 3.965074 (lr=1.0140e-04) (hash(x)=143481291)
945
+ 7850 train 3.785060 (lr=1.0094e-04) (hash(x)=136749307)
946
+ 7860 train 3.814697 (lr=1.0048e-04) (hash(x)=128985347)
947
+ 7870 train 3.719477 (lr=1.0002e-04) (hash(x)=152897363)
948
+ 7880 train 3.921330 (lr=9.9576e-05) (hash(x)=151460360)
949
+ 7890 train 3.740543 (lr=9.9133e-05) (hash(x)=137481738)
950
+ 7900 val loss 4.0266
951
+ 7900 val perplexity 56.0684
952
+ 7900 train 3.816713 (lr=9.8694e-05) (hash(x)=147363479)
953
+ 7910 train 4.038513 (lr=9.8261e-05) (hash(x)=151165744)
954
+ 7920 train 4.219281 (lr=9.7832e-05) (hash(x)=146214226)
955
+ 7930 train 4.027877 (lr=9.7408e-05) (hash(x)=154457502)
956
+ 7940 train 3.925274 (lr=9.6990e-05) (hash(x)=162025201)
957
+ 7950 train 3.923593 (lr=9.6576e-05) (hash(x)=156488145)
958
+ 7960 train 4.199000 (lr=9.6167e-05) (hash(x)=154011584)
959
+ 7970 train 4.048936 (lr=9.5764e-05) (hash(x)=145330320)
960
+ 7980 train 3.957245 (lr=9.5365e-05) (hash(x)=148216312)
961
+ 7990 train 3.896387 (lr=9.4971e-05) (hash(x)=156819091)
962
+ 8000 val loss 4.0164
963
+ 8000 val perplexity 55.5005
964
+ 8000 train 4.047329 (lr=9.4583e-05) (hash(x)=156122973)
965
+ 8010 train 3.989617 (lr=9.4199e-05) (hash(x)=146933088)
966
+ 8020 train 4.085203 (lr=9.3820e-05) (hash(x)=151650746)
967
+ 8030 train 4.013157 (lr=9.3446e-05) (hash(x)=140692373)
968
+ 8040 train 4.204532 (lr=9.3078e-05) (hash(x)=147850094)
969
+ 8050 train 4.016724 (lr=9.2714e-05) (hash(x)=144454109)
970
+ 8060 train 3.882004 (lr=9.2356e-05) (hash(x)=152205124)
971
+ 8070 train 4.095700 (lr=9.2002e-05) (hash(x)=148918066)
972
+ 8080 train 4.026139 (lr=9.1654e-05) (hash(x)=146481024)
973
+ 8090 train 4.062425 (lr=9.1310e-05) (hash(x)=168785549)
974
+ 8100 val loss 4.0143
975
+ 8100 val perplexity 55.3851
976
+ 8100 train 4.091202 (lr=9.0972e-05) (hash(x)=156153179)
977
+ 8110 train 4.043861 (lr=9.0638e-05) (hash(x)=150498601)
978
+ 8120 train 4.046705 (lr=9.0310e-05) (hash(x)=142002714)
979
+ 8130 train 3.940776 (lr=8.9987e-05) (hash(x)=138450483)
980
+ 8140 train 4.022700 (lr=8.9669e-05) (hash(x)=145306351)
981
+ 8150 train 3.906805 (lr=8.9356e-05) (hash(x)=172502865)
982
+ 8160 train 4.001176 (lr=8.9048e-05) (hash(x)=156689649)
983
+ 8170 train 3.924330 (lr=8.8745e-05) (hash(x)=143766932)
984
+ 8180 train 4.058850 (lr=8.8447e-05) (hash(x)=150361321)
985
+ 8190 train 3.890045 (lr=8.8154e-05) (hash(x)=152664241)
986
+ 8200 val loss 4.0137
987
+ 8200 val perplexity 55.3525
988
+ 8200 train 4.168352 (lr=8.7867e-05) (hash(x)=146430698)
989
+ 8210 train 3.855417 (lr=8.7584e-05) (hash(x)=167362078)
990
+ 8220 train 4.072332 (lr=8.7307e-05) (hash(x)=157650942)
991
+ 8230 train 3.879432 (lr=8.7035e-05) (hash(x)=142102044)
992
+ 8240 train 3.924657 (lr=8.6768e-05) (hash(x)=153405900)
993
+ 8250 train 3.878210 (lr=8.6506e-05) (hash(x)=140222909)
994
+ 8260 train 4.011298 (lr=8.6249e-05) (hash(x)=137981854)
995
+ 8270 train 3.928321 (lr=8.5997e-05) (hash(x)=147576272)
996
+ 8280 train 3.937930 (lr=8.5750e-05) (hash(x)=157988195)
997
+ 8290 train 3.994762 (lr=8.5509e-05) (hash(x)=145715681)
998
+ 8300 val loss 4.0149
999
+ 8300 val perplexity 55.4194
1000
+ 8300 train 3.938457 (lr=8.5273e-05) (hash(x)=143507257)
1001
+ 8310 train 3.997448 (lr=8.5041e-05) (hash(x)=148930064)
1002
+ 8320 train 3.923447 (lr=8.4815e-05) (hash(x)=144134805)
1003
+ 8330 train 4.035623 (lr=8.4594e-05) (hash(x)=148418079)
1004
+ 8340 train 3.767885 (lr=8.4379e-05) (hash(x)=166518807)
1005
+ 8350 train 4.070646 (lr=8.4168e-05) (hash(x)=140042036)
1006
+ 8360 train 3.945139 (lr=8.3963e-05) (hash(x)=151427074)
1007
+ 8370 train 4.028969 (lr=8.3762e-05) (hash(x)=158389185)
1008
+ 8380 train 3.937571 (lr=8.3567e-05) (hash(x)=148193308)
1009
+ 8390 train 3.999226 (lr=8.3377e-05) (hash(x)=144289008)
1010
+ 8400 val loss 4.0058
1011
+ 8400 val perplexity 54.9138
1012
+ 8400 train 3.717309 (lr=8.3193e-05) (hash(x)=166272643)
1013
+ 8410 train 3.928385 (lr=8.3013e-05) (hash(x)=162244433)
1014
+ 8420 train 3.962059 (lr=8.2839e-05) (hash(x)=149271299)
1015
+ 8430 train 4.105599 (lr=8.2669e-05) (hash(x)=154135338)
1016
+ 8440 train 4.151826 (lr=8.2505e-05) (hash(x)=149661838)
1017
+ 8450 train 3.985238 (lr=8.2347e-05) (hash(x)=142205056)
1018
+ 8460 train 3.919768 (lr=8.2193e-05) (hash(x)=166652398)
1019
+ 8470 train 4.095822 (lr=8.2044e-05) (hash(x)=149029710)
1020
+ 8480 train 3.863627 (lr=8.1901e-05) (hash(x)=154217415)
1021
+ 8490 train 3.945237 (lr=8.1763e-05) (hash(x)=151526557)
1022
+ 8500 val loss 4.0068
1023
+ 8500 val perplexity 54.9702
1024
+ 8500 train 3.931766 (lr=8.1630e-05) (hash(x)=143887848)
1025
+ 8510 train 3.920545 (lr=8.1502e-05) (hash(x)=152324645)
1026
+ 8520 train 3.650105 (lr=8.1380e-05) (hash(x)=145124959)
1027
+ 8530 train 3.973529 (lr=8.1263e-05) (hash(x)=186319340)
1028
+ 8540 train 3.988179 (lr=8.1150e-05) (hash(x)=147388339)
1029
+ 8550 train 4.014598 (lr=8.1044e-05) (hash(x)=150118928)
1030
+ 8560 train 4.020784 (lr=8.0942e-05) (hash(x)=153985286)
1031
+ 8570 train 3.957687 (lr=8.0845e-05) (hash(x)=149582799)
1032
+ 8580 train 3.962481 (lr=8.0754e-05) (hash(x)=156867870)
1033
+ 8590 train 3.858797 (lr=8.0668e-05) (hash(x)=148773298)
1034
+ 8600 val loss 4.0093
1035
+ 8600 val perplexity 55.1094
1036
+ 8600 train 3.951298 (lr=8.0587e-05) (hash(x)=156900341)
1037
+ 8610 train 4.110935 (lr=8.0511e-05) (hash(x)=156790089)
1038
+ 8620 train 3.912874 (lr=8.0441e-05) (hash(x)=154680805)
1039
+ 8630 train 3.950346 (lr=8.0376e-05) (hash(x)=138107873)
1040
+ 8640 train 4.042800 (lr=8.0316e-05) (hash(x)=142554243)
1041
+ 8650 train 3.841221 (lr=8.0261e-05) (hash(x)=134360956)
1042
+ 8660 train 3.923875 (lr=8.0211e-05) (hash(x)=128599610)
1043
+ 8670 train 3.954834 (lr=8.0167e-05) (hash(x)=159628945)
1044
+ 8680 train 3.986490 (lr=8.0128e-05) (hash(x)=143161532)
1045
+ 8690 train 4.012177 (lr=8.0094e-05) (hash(x)=140104912)
1046
+ 8700 val loss 4.0011
1047
+ 8700 val perplexity 54.6591
1048
+ 8700 train 4.186152 (lr=8.0065e-05) (hash(x)=146417632)
1049
+ 8710 train 4.165915 (lr=8.0042e-05) (hash(x)=153671695)
1050
+ 8720 train 4.037034 (lr=8.0023e-05) (hash(x)=141680750)
1051
+ 8730 train 3.910734 (lr=8.0010e-05) (hash(x)=147279231)
1052
+ 8740 train 4.202551 (lr=8.0003e-05) (hash(x)=129729371)
1053
+ 8749 val loss 4.0015
1054
+ 8749 val perplexity 54.6802
lr8e-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 sha256:0d6d5b631e233d2966d4ad867d741d0a78cf67edfc4c8c300a32e8736b155669
3
+ size 92843394
lr8e-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 sha256:4895d29bfae80d83fa339ba337fe8bce61c9bbf53ac457f1198d47873e128462
3
+ size 179406214