andrew-healey commited on
Commit
16e6eb8
·
verified ·
1 Parent(s): 02030d1

Upload folder using huggingface_hub

Browse files
lr9e-4_total_batch_size61440_baseline_seed1339/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_11/lr9e-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": 0.0009, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "9e-4_61440", "n_embd": 256}
lr9e-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
lr9e-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=1.8000e-06) (hash(x)=150724848)
5
+ 10 train 10.204916 (lr=1.9800e-05) (hash(x)=149663361)
6
+ 20 train 9.795771 (lr=3.7800e-05) (hash(x)=153483789)
7
+ 30 train 9.495066 (lr=5.5800e-05) (hash(x)=145733176)
8
+ 40 train 8.922192 (lr=7.3800e-05) (hash(x)=145287265)
9
+ 50 train 8.440268 (lr=9.1800e-05) (hash(x)=149217061)
10
+ 60 train 7.995121 (lr=1.0980e-04) (hash(x)=150157569)
11
+ 70 train 7.708392 (lr=1.2780e-04) (hash(x)=140292841)
12
+ 80 train 7.690462 (lr=1.4580e-04) (hash(x)=152157598)
13
+ 90 train 7.511661 (lr=1.6380e-04) (hash(x)=149756968)
14
+ 100 val loss 7.4851
15
+ 100 val perplexity 1781.2697
16
+ 100 train 7.441054 (lr=1.8180e-04) (hash(x)=149910534)
17
+ 110 train 7.357661 (lr=1.9980e-04) (hash(x)=159020328)
18
+ 120 train 7.287286 (lr=2.1780e-04) (hash(x)=150731273)
19
+ 130 train 7.100514 (lr=2.3580e-04) (hash(x)=156727355)
20
+ 140 train 7.115923 (lr=2.5380e-04) (hash(x)=166473281)
21
+ 150 train 6.809099 (lr=2.7180e-04) (hash(x)=142185643)
22
+ 160 train 6.808213 (lr=2.8980e-04) (hash(x)=151371145)
23
+ 170 train 6.844340 (lr=3.0780e-04) (hash(x)=151601659)
24
+ 180 train 6.667512 (lr=3.2580e-04) (hash(x)=153368917)
25
+ 190 train 6.586907 (lr=3.4380e-04) (hash(x)=145008196)
26
+ 200 val loss 6.6496
27
+ 200 val perplexity 772.4930
28
+ 200 train 6.541251 (lr=3.6180e-04) (hash(x)=148123706)
29
+ 210 train 6.427405 (lr=3.7980e-04) (hash(x)=145714762)
30
+ 220 train 6.432698 (lr=3.9780e-04) (hash(x)=150622583)
31
+ 230 train 6.203651 (lr=4.1580e-04) (hash(x)=146512648)
32
+ 240 train 6.253048 (lr=4.3380e-04) (hash(x)=151535384)
33
+ 250 train 6.194899 (lr=4.5180e-04) (hash(x)=149952383)
34
+ 260 train 6.082754 (lr=4.6980e-04) (hash(x)=148726088)
35
+ 270 train 6.055576 (lr=4.8780e-04) (hash(x)=152776695)
36
+ 280 train 6.081209 (lr=5.0580e-04) (hash(x)=157260375)
37
+ 290 train 6.130800 (lr=5.2380e-04) (hash(x)=151559106)
38
+ 300 val loss 6.1934
39
+ 300 val perplexity 489.5092
40
+ 300 train 5.922919 (lr=5.4180e-04) (hash(x)=146678221)
41
+ 310 train 6.013123 (lr=5.5980e-04) (hash(x)=156713527)
42
+ 320 train 6.073313 (lr=5.7780e-04) (hash(x)=144947971)
43
+ 330 train 6.245952 (lr=5.9580e-04) (hash(x)=158684944)
44
+ 340 train 6.126135 (lr=6.1380e-04) (hash(x)=162318756)
45
+ 350 train 5.917670 (lr=6.3180e-04) (hash(x)=156180736)
46
+ 360 train 6.096029 (lr=6.4980e-04) (hash(x)=178373797)
47
+ 370 train 5.995826 (lr=6.6780e-04) (hash(x)=154389449)
48
+ 380 train 5.830288 (lr=6.8580e-04) (hash(x)=150369080)
49
+ 390 train 5.965586 (lr=7.0380e-04) (hash(x)=159840654)
50
+ 400 val loss 5.8562
51
+ 400 val perplexity 349.4003
52
+ 400 train 5.773590 (lr=7.2180e-04) (hash(x)=151700982)
53
+ 410 train 5.745438 (lr=7.3980e-04) (hash(x)=160467274)
54
+ 420 train 5.653216 (lr=7.5780e-04) (hash(x)=148930522)
55
+ 430 train 5.883911 (lr=7.7580e-04) (hash(x)=154565303)
56
+ 440 train 5.848239 (lr=7.9380e-04) (hash(x)=148079227)
57
+ 450 train 5.577741 (lr=8.1180e-04) (hash(x)=147288467)
58
+ 460 train 5.608313 (lr=8.2980e-04) (hash(x)=145272617)
59
+ 470 train 5.612686 (lr=8.4780e-04) (hash(x)=123126729)
60
+ 480 train 5.566161 (lr=8.6580e-04) (hash(x)=156287490)
61
+ 490 train 5.705488 (lr=8.8380e-04) (hash(x)=161700145)
62
+ 500 val loss 5.6111
63
+ 500 val perplexity 273.4426
64
+ 500 train 5.727526 (lr=9.0000e-04) (hash(x)=156182087)
65
+ 510 train 5.331793 (lr=9.0000e-04) (hash(x)=131674504)
66
+ 520 train 5.638835 (lr=8.9999e-04) (hash(x)=162758254)
67
+ 530 train 5.455791 (lr=8.9997e-04) (hash(x)=133510409)
68
+ 540 train 5.450193 (lr=8.9995e-04) (hash(x)=157363914)
69
+ 550 train 5.389903 (lr=8.9993e-04) (hash(x)=147861550)
70
+ 560 train 5.332451 (lr=8.9989e-04) (hash(x)=149449280)
71
+ 570 train 5.348060 (lr=8.9986e-04) (hash(x)=143899904)
72
+ 580 train 5.309756 (lr=8.9981e-04) (hash(x)=149266312)
73
+ 590 train 5.538180 (lr=8.9976e-04) (hash(x)=164352774)
74
+ 600 val loss 5.4445
75
+ 600 val perplexity 231.4861
76
+ 600 train 5.312519 (lr=8.9971e-04) (hash(x)=149318660)
77
+ 610 train 5.209450 (lr=8.9964e-04) (hash(x)=146101781)
78
+ 620 train 5.169048 (lr=8.9958e-04) (hash(x)=137184623)
79
+ 630 train 5.046579 (lr=8.9950e-04) (hash(x)=145791873)
80
+ 640 train 5.069399 (lr=8.9942e-04) (hash(x)=143999212)
81
+ 650 train 5.181959 (lr=8.9934e-04) (hash(x)=151631103)
82
+ 660 train 4.942088 (lr=8.9925e-04) (hash(x)=138360108)
83
+ 670 train 5.328443 (lr=8.9915e-04) (hash(x)=156637679)
84
+ 680 train 4.971367 (lr=8.9905e-04) (hash(x)=143752601)
85
+ 690 train 5.110377 (lr=8.9894e-04) (hash(x)=158368903)
86
+ 700 val loss 5.3221
87
+ 700 val perplexity 204.8193
88
+ 700 train 5.312403 (lr=8.9883e-04) (hash(x)=150482428)
89
+ 710 train 5.279535 (lr=8.9871e-04) (hash(x)=158085060)
90
+ 720 train 5.338670 (lr=8.9858e-04) (hash(x)=159461069)
91
+ 730 train 5.189444 (lr=8.9845e-04) (hash(x)=148795562)
92
+ 740 train 5.196690 (lr=8.9831e-04) (hash(x)=149967984)
93
+ 750 train 5.179222 (lr=8.9817e-04) (hash(x)=144458617)
94
+ 760 train 5.217418 (lr=8.9802e-04) (hash(x)=155106202)
95
+ 770 train 5.007403 (lr=8.9786e-04) (hash(x)=144506191)
96
+ 780 train 5.163235 (lr=8.9770e-04) (hash(x)=155825158)
97
+ 790 train 5.210736 (lr=8.9753e-04) (hash(x)=158835381)
98
+ 800 val loss 5.1603
99
+ 800 val perplexity 174.2232
100
+ 800 train 5.109372 (lr=8.9736e-04) (hash(x)=143268605)
101
+ 810 train 5.036896 (lr=8.9718e-04) (hash(x)=139072755)
102
+ 820 train 5.163919 (lr=8.9700e-04) (hash(x)=179384792)
103
+ 830 train 4.950008 (lr=8.9681e-04) (hash(x)=151086962)
104
+ 840 train 5.573259 (lr=8.9661e-04) (hash(x)=142711393)
105
+ 850 train 5.041679 (lr=8.9641e-04) (hash(x)=150817011)
106
+ 860 train 4.986570 (lr=8.9620e-04) (hash(x)=138495801)
107
+ 870 train 5.015603 (lr=8.9599e-04) (hash(x)=155474016)
108
+ 880 train 5.028063 (lr=8.9577e-04) (hash(x)=151951501)
109
+ 890 train 4.895245 (lr=8.9554e-04) (hash(x)=143699524)
110
+ 900 val loss 5.0475
111
+ 900 val perplexity 155.6355
112
+ 900 train 5.038856 (lr=8.9531e-04) (hash(x)=152322423)
113
+ 910 train 4.989079 (lr=8.9507e-04) (hash(x)=151010509)
114
+ 920 train 4.773036 (lr=8.9483e-04) (hash(x)=148720342)
115
+ 930 train 4.745556 (lr=8.9458e-04) (hash(x)=125869480)
116
+ 940 train 4.980884 (lr=8.9433e-04) (hash(x)=150349503)
117
+ 950 train 4.753446 (lr=8.9407e-04) (hash(x)=143574126)
118
+ 960 train 4.948604 (lr=8.9380e-04) (hash(x)=146292562)
119
+ 970 train 4.895460 (lr=8.9353e-04) (hash(x)=162256596)
120
+ 980 train 4.834550 (lr=8.9325e-04) (hash(x)=137640259)
121
+ 990 train 4.697695 (lr=8.9297e-04) (hash(x)=141767794)
122
+ 1000 val loss 4.9724
123
+ 1000 val perplexity 144.3767
124
+ 1000 train 4.919951 (lr=8.9268e-04) (hash(x)=147904298)
125
+ 1010 train 4.825657 (lr=8.9239e-04) (hash(x)=162239186)
126
+ 1020 train 4.597697 (lr=8.9209e-04) (hash(x)=140706537)
127
+ 1030 train 4.730639 (lr=8.9178e-04) (hash(x)=155556720)
128
+ 1040 train 4.608821 (lr=8.9147e-04) (hash(x)=144746445)
129
+ 1050 train 4.688078 (lr=8.9115e-04) (hash(x)=144719400)
130
+ 1060 train 4.517729 (lr=8.9083e-04) (hash(x)=153065665)
131
+ 1070 train 4.650485 (lr=8.9050e-04) (hash(x)=136468722)
132
+ 1080 train 4.629011 (lr=8.9016e-04) (hash(x)=157674756)
133
+ 1090 train 4.812330 (lr=8.8982e-04) (hash(x)=155882884)
134
+ 1100 val loss 4.8762
135
+ 1100 val perplexity 131.1353
136
+ 1100 train 5.090960 (lr=8.8947e-04) (hash(x)=154343147)
137
+ 1110 train 4.888655 (lr=8.8912e-04) (hash(x)=158582939)
138
+ 1120 train 4.684005 (lr=8.8876e-04) (hash(x)=162479784)
139
+ 1130 train 4.874101 (lr=8.8840e-04) (hash(x)=155704345)
140
+ 1140 train 4.762642 (lr=8.8803e-04) (hash(x)=146934299)
141
+ 1150 train 4.710603 (lr=8.8766e-04) (hash(x)=154372635)
142
+ 1160 train 4.706114 (lr=8.8728e-04) (hash(x)=145536832)
143
+ 1170 train 4.994603 (lr=8.8689e-04) (hash(x)=158785901)
144
+ 1180 train 4.772057 (lr=8.8650e-04) (hash(x)=158271614)
145
+ 1190 train 4.826585 (lr=8.8610e-04) (hash(x)=157365589)
146
+ 1200 val loss 4.7937
147
+ 1200 val perplexity 120.7470
148
+ 1200 train 4.861231 (lr=8.8570e-04) (hash(x)=141843115)
149
+ 1210 train 4.747582 (lr=8.8529e-04) (hash(x)=146807973)
150
+ 1220 train 4.728007 (lr=8.8487e-04) (hash(x)=144149947)
151
+ 1230 train 4.653369 (lr=8.8445e-04) (hash(x)=147869063)
152
+ 1240 train 4.749627 (lr=8.8403e-04) (hash(x)=152217389)
153
+ 1250 train 4.746050 (lr=8.8359e-04) (hash(x)=149417679)
154
+ 1260 train 4.762004 (lr=8.8316e-04) (hash(x)=144029489)
155
+ 1270 train 4.889368 (lr=8.8271e-04) (hash(x)=154328303)
156
+ 1280 train 4.742177 (lr=8.8227e-04) (hash(x)=151625838)
157
+ 1290 train 4.603739 (lr=8.8181e-04) (hash(x)=140270541)
158
+ 1300 val loss 4.7307
159
+ 1300 val perplexity 113.3729
160
+ 1300 train 4.658215 (lr=8.8135e-04) (hash(x)=145279030)
161
+ 1310 train 4.693948 (lr=8.8089e-04) (hash(x)=140279630)
162
+ 1320 train 4.423915 (lr=8.8042e-04) (hash(x)=133642994)
163
+ 1330 train 4.589620 (lr=8.7994e-04) (hash(x)=146634776)
164
+ 1340 train 4.691142 (lr=8.7946e-04) (hash(x)=155919973)
165
+ 1350 train 4.726771 (lr=8.7897e-04) (hash(x)=146363911)
166
+ 1360 train 4.595160 (lr=8.7848e-04) (hash(x)=150613636)
167
+ 1370 train 4.591089 (lr=8.7798e-04) (hash(x)=148536643)
168
+ 1380 train 4.594179 (lr=8.7747e-04) (hash(x)=152803906)
169
+ 1390 train 4.531226 (lr=8.7696e-04) (hash(x)=150383305)
170
+ 1400 val loss 4.7173
171
+ 1400 val perplexity 111.8608
172
+ 1400 train 4.618720 (lr=8.7645e-04) (hash(x)=152507639)
173
+ 1410 train 4.378652 (lr=8.7593e-04) (hash(x)=161841408)
174
+ 1420 train 4.494325 (lr=8.7540e-04) (hash(x)=167824828)
175
+ 1430 train 4.455928 (lr=8.7487e-04) (hash(x)=147696374)
176
+ 1440 train 4.409888 (lr=8.7433e-04) (hash(x)=150085494)
177
+ 1450 train 4.364134 (lr=8.7379e-04) (hash(x)=139296191)
178
+ 1460 train 4.288249 (lr=8.7324e-04) (hash(x)=142956546)
179
+ 1470 train 4.356631 (lr=8.7268e-04) (hash(x)=155908628)
180
+ 1480 train 4.341864 (lr=8.7212e-04) (hash(x)=150893345)
181
+ 1490 train 4.367283 (lr=8.7156e-04) (hash(x)=143548902)
182
+ 1500 val loss 4.6592
183
+ 1500 val perplexity 105.5471
184
+ 1500 train 4.619555 (lr=8.7099e-04) (hash(x)=148473774)
185
+ 1510 train 4.689001 (lr=8.7041e-04) (hash(x)=161631000)
186
+ 1520 train 4.619362 (lr=8.6983e-04) (hash(x)=166217855)
187
+ 1530 train 4.577283 (lr=8.6924e-04) (hash(x)=149440046)
188
+ 1540 train 4.521316 (lr=8.6865e-04) (hash(x)=155013278)
189
+ 1550 train 4.559105 (lr=8.6806e-04) (hash(x)=154130527)
190
+ 1560 train 4.582282 (lr=8.6745e-04) (hash(x)=152138735)
191
+ 1570 train 4.494455 (lr=8.6684e-04) (hash(x)=148128305)
192
+ 1580 train 4.632482 (lr=8.6623e-04) (hash(x)=157176546)
193
+ 1590 train 4.742635 (lr=8.6561e-04) (hash(x)=152935467)
194
+ 1600 val loss 4.5824
195
+ 1600 val perplexity 97.7482
196
+ 1600 train 4.558242 (lr=8.6499e-04) (hash(x)=151117002)
197
+ 1610 train 4.551423 (lr=8.6436e-04) (hash(x)=149883384)
198
+ 1620 train 4.603119 (lr=8.6372e-04) (hash(x)=145015354)
199
+ 1630 train 4.526357 (lr=8.6308e-04) (hash(x)=146717645)
200
+ 1640 train 4.738586 (lr=8.6243e-04) (hash(x)=156372727)
201
+ 1650 train 4.648260 (lr=8.6178e-04) (hash(x)=140261555)
202
+ 1660 train 4.473122 (lr=8.6113e-04) (hash(x)=150088650)
203
+ 1670 train 4.396767 (lr=8.6046e-04) (hash(x)=135943513)
204
+ 1680 train 4.625249 (lr=8.5980e-04) (hash(x)=153529653)
205
+ 1690 train 4.679654 (lr=8.5912e-04) (hash(x)=148930360)
206
+ 1700 val loss 4.5921
207
+ 1700 val perplexity 98.7051
208
+ 1700 train 4.570681 (lr=8.5845e-04) (hash(x)=138011335)
209
+ 1710 train 4.526221 (lr=8.5776e-04) (hash(x)=141383599)
210
+ 1720 train 4.704248 (lr=8.5707e-04) (hash(x)=161449652)
211
+ 1730 train 4.493217 (lr=8.5638e-04) (hash(x)=153923561)
212
+ 1740 train 4.432128 (lr=8.5568e-04) (hash(x)=146043253)
213
+ 1750 train 4.469324 (lr=8.5498e-04) (hash(x)=155720883)
214
+ 1760 train 4.450272 (lr=8.5427e-04) (hash(x)=148222148)
215
+ 1770 train 4.511732 (lr=8.5355e-04) (hash(x)=155712380)
216
+ 1780 train 4.403421 (lr=8.5283e-04) (hash(x)=149767859)
217
+ 1790 train 4.638729 (lr=8.5211e-04) (hash(x)=167214230)
218
+ 1800 val loss 4.5529
219
+ 1800 val perplexity 94.9042
220
+ 1800 train 4.527927 (lr=8.5138e-04) (hash(x)=171180926)
221
+ 1810 train 4.543214 (lr=8.5064e-04) (hash(x)=162375834)
222
+ 1820 train 4.660954 (lr=8.4990e-04) (hash(x)=151495752)
223
+ 1830 train 4.562450 (lr=8.4916e-04) (hash(x)=146532257)
224
+ 1840 train 4.645109 (lr=8.4841e-04) (hash(x)=145694434)
225
+ 1850 train 4.533226 (lr=8.4765e-04) (hash(x)=144823488)
226
+ 1860 train 4.635515 (lr=8.4689e-04) (hash(x)=149836286)
227
+ 1870 train 4.599837 (lr=8.4613e-04) (hash(x)=143935049)
228
+ 1880 train 4.629204 (lr=8.4535e-04) (hash(x)=151786371)
229
+ 1890 train 4.522903 (lr=8.4458e-04) (hash(x)=152892567)
230
+ 1900 val loss 4.5096
231
+ 1900 val perplexity 90.8879
232
+ 1900 train 4.555480 (lr=8.4380e-04) (hash(x)=141769419)
233
+ 1910 train 4.580752 (lr=8.4301e-04) (hash(x)=156914416)
234
+ 1920 train 4.429949 (lr=8.4222e-04) (hash(x)=151573900)
235
+ 1930 train 4.527961 (lr=8.4142e-04) (hash(x)=145121701)
236
+ 1940 train 4.468036 (lr=8.4062e-04) (hash(x)=150896348)
237
+ 1950 train 4.578573 (lr=8.3981e-04) (hash(x)=168797189)
238
+ 1960 train 4.457661 (lr=8.3900e-04) (hash(x)=146711430)
239
+ 1970 train 4.309088 (lr=8.3819e-04) (hash(x)=152180946)
240
+ 1980 train 4.432861 (lr=8.3737e-04) (hash(x)=155542268)
241
+ 1990 train 4.397873 (lr=8.3654e-04) (hash(x)=144668660)
242
+ 2000 val loss 4.4753
243
+ 2000 val perplexity 87.8243
244
+ 2000 train 4.462196 (lr=8.3571e-04) (hash(x)=151963443)
245
+ 2010 train 4.459739 (lr=8.3487e-04) (hash(x)=150810352)
246
+ 2020 train 4.465093 (lr=8.3403e-04) (hash(x)=157539832)
247
+ 2030 train 4.682232 (lr=8.3318e-04) (hash(x)=137858786)
248
+ 2040 train 4.418502 (lr=8.3233e-04) (hash(x)=154270900)
249
+ 2050 train 4.279358 (lr=8.3148e-04) (hash(x)=156252579)
250
+ 2060 train 4.214097 (lr=8.3062e-04) (hash(x)=143285549)
251
+ 2070 train 4.421713 (lr=8.2975e-04) (hash(x)=150542438)
252
+ 2080 train 4.337128 (lr=8.2888e-04) (hash(x)=146290330)
253
+ 2090 train 4.434236 (lr=8.2800e-04) (hash(x)=146486756)
254
+ 2100 val loss 4.5066
255
+ 2100 val perplexity 90.6161
256
+ 2100 train 4.434471 (lr=8.2712e-04) (hash(x)=162947470)
257
+ 2110 train 4.341206 (lr=8.2624e-04) (hash(x)=150305391)
258
+ 2120 train 4.503350 (lr=8.2535e-04) (hash(x)=156452783)
259
+ 2130 train 4.470096 (lr=8.2446e-04) (hash(x)=167481875)
260
+ 2140 train 4.452669 (lr=8.2356e-04) (hash(x)=141782256)
261
+ 2150 train 4.480330 (lr=8.2265e-04) (hash(x)=145035419)
262
+ 2160 train 4.530692 (lr=8.2174e-04) (hash(x)=142000400)
263
+ 2170 train 4.509083 (lr=8.2083e-04) (hash(x)=144229380)
264
+ 2180 train 4.534692 (lr=8.1991e-04) (hash(x)=166204176)
265
+ 2190 train 4.456527 (lr=8.1899e-04) (hash(x)=150712057)
266
+ 2200 val loss 4.4365
267
+ 2200 val perplexity 84.4746
268
+ 2200 train 4.352433 (lr=8.1806e-04) (hash(x)=154954810)
269
+ 2210 train 4.553628 (lr=8.1713e-04) (hash(x)=155668760)
270
+ 2220 train 4.385908 (lr=8.1619e-04) (hash(x)=144834708)
271
+ 2230 train 4.380559 (lr=8.1525e-04) (hash(x)=159744566)
272
+ 2240 train 4.336640 (lr=8.1430e-04) (hash(x)=158543377)
273
+ 2250 train 4.350257 (lr=8.1335e-04) (hash(x)=151004281)
274
+ 2260 train 4.452587 (lr=8.1240e-04) (hash(x)=151242482)
275
+ 2270 train 4.398960 (lr=8.1144e-04) (hash(x)=152056559)
276
+ 2280 train 4.335979 (lr=8.1047e-04) (hash(x)=148864076)
277
+ 2290 train 4.459187 (lr=8.0950e-04) (hash(x)=144294586)
278
+ 2300 val loss 4.4308
279
+ 2300 val perplexity 83.9959
280
+ 2300 train 4.364259 (lr=8.0853e-04) (hash(x)=151878111)
281
+ 2310 train 4.369933 (lr=8.0755e-04) (hash(x)=154139457)
282
+ 2320 train 4.308999 (lr=8.0657e-04) (hash(x)=155571592)
283
+ 2330 train 4.352278 (lr=8.0558e-04) (hash(x)=154823764)
284
+ 2340 train 4.394814 (lr=8.0459e-04) (hash(x)=153252177)
285
+ 2350 train 4.336636 (lr=8.0359e-04) (hash(x)=136051830)
286
+ 2360 train 4.380573 (lr=8.0259e-04) (hash(x)=153814175)
287
+ 2370 train 4.370546 (lr=8.0158e-04) (hash(x)=146391078)
288
+ 2380 train 4.340682 (lr=8.0057e-04) (hash(x)=150782698)
289
+ 2390 train 4.416305 (lr=7.9956e-04) (hash(x)=142752256)
290
+ 2400 val loss 4.4386
291
+ 2400 val perplexity 84.6603
292
+ 2400 train 4.318916 (lr=7.9854e-04) (hash(x)=158661057)
293
+ 2410 train 4.388486 (lr=7.9752e-04) (hash(x)=153611303)
294
+ 2420 train 4.565928 (lr=7.9649e-04) (hash(x)=159508684)
295
+ 2430 train 4.356451 (lr=7.9546e-04) (hash(x)=149501192)
296
+ 2440 train 4.024657 (lr=7.9442e-04) (hash(x)=181177177)
297
+ 2450 train 4.493620 (lr=7.9338e-04) (hash(x)=163664732)
298
+ 2460 train 4.178998 (lr=7.9234e-04) (hash(x)=194527955)
299
+ 2470 train 4.403702 (lr=7.9129e-04) (hash(x)=158088521)
300
+ 2480 train 4.361881 (lr=7.9023e-04) (hash(x)=152865678)
301
+ 2490 train 4.340837 (lr=7.8917e-04) (hash(x)=165703871)
302
+ 2500 val loss 4.3824
303
+ 2500 val perplexity 80.0297
304
+ 2500 train 4.429049 (lr=7.8811e-04) (hash(x)=150925584)
305
+ 2510 train 4.337191 (lr=7.8705e-04) (hash(x)=144406177)
306
+ 2520 train 4.233067 (lr=7.8598e-04) (hash(x)=120133043)
307
+ 2530 train 4.418875 (lr=7.8490e-04) (hash(x)=167900247)
308
+ 2540 train 4.319124 (lr=7.8382e-04) (hash(x)=162599203)
309
+ 2550 train 4.382001 (lr=7.8274e-04) (hash(x)=149328180)
310
+ 2560 train 4.273485 (lr=7.8165e-04) (hash(x)=150049968)
311
+ 2570 train 4.414562 (lr=7.8056e-04) (hash(x)=162876159)
312
+ 2580 train 4.236141 (lr=7.7946e-04) (hash(x)=148260649)
313
+ 2590 train 4.377519 (lr=7.7836e-04) (hash(x)=148998018)
314
+ 2600 val loss 4.3753
315
+ 2600 val perplexity 79.4598
316
+ 2600 train 4.291930 (lr=7.7726e-04) (hash(x)=144515755)
317
+ 2610 train 4.409541 (lr=7.7615e-04) (hash(x)=154455224)
318
+ 2620 train 4.205801 (lr=7.7504e-04) (hash(x)=142920048)
319
+ 2630 train 4.373425 (lr=7.7392e-04) (hash(x)=143931236)
320
+ 2640 train 4.224348 (lr=7.7280e-04) (hash(x)=150983323)
321
+ 2650 train 4.315905 (lr=7.7168e-04) (hash(x)=139360405)
322
+ 2660 train 4.232763 (lr=7.7055e-04) (hash(x)=148850470)
323
+ 2670 train 4.489038 (lr=7.6942e-04) (hash(x)=149929606)
324
+ 2680 train 4.462453 (lr=7.6828e-04) (hash(x)=151673000)
325
+ 2690 train 4.329906 (lr=7.6714e-04) (hash(x)=140631266)
326
+ 2700 val loss 4.3733
327
+ 2700 val perplexity 79.3023
328
+ 2700 train 4.309478 (lr=7.6600e-04) (hash(x)=153109144)
329
+ 2710 train 4.193210 (lr=7.6485e-04) (hash(x)=135332836)
330
+ 2720 train 4.545353 (lr=7.6370e-04) (hash(x)=131611182)
331
+ 2730 train 4.371665 (lr=7.6254e-04) (hash(x)=142909388)
332
+ 2740 train 4.415164 (lr=7.6138e-04) (hash(x)=145593419)
333
+ 2750 train 4.311321 (lr=7.6022e-04) (hash(x)=158225769)
334
+ 2760 train 4.277250 (lr=7.5905e-04) (hash(x)=170509938)
335
+ 2770 train 4.294447 (lr=7.5788e-04) (hash(x)=150785805)
336
+ 2780 train 4.240649 (lr=7.5670e-04) (hash(x)=167281736)
337
+ 2790 train 4.422886 (lr=7.5553e-04) (hash(x)=161660187)
338
+ 2800 val loss 4.3436
339
+ 2800 val perplexity 76.9851
340
+ 2800 train 4.386814 (lr=7.5434e-04) (hash(x)=151152897)
341
+ 2810 train 4.338774 (lr=7.5316e-04) (hash(x)=143172067)
342
+ 2820 train 4.352896 (lr=7.5197e-04) (hash(x)=141569169)
343
+ 2830 train 4.231173 (lr=7.5077e-04) (hash(x)=159709681)
344
+ 2840 train 4.294721 (lr=7.4958e-04) (hash(x)=162552294)
345
+ 2850 train 4.421988 (lr=7.4837e-04) (hash(x)=146762690)
346
+ 2860 train 4.286278 (lr=7.4717e-04) (hash(x)=145729422)
347
+ 2870 train 4.366582 (lr=7.4596e-04) (hash(x)=144360628)
348
+ 2880 train 4.179487 (lr=7.4475e-04) (hash(x)=147737098)
349
+ 2890 train 4.281993 (lr=7.4353e-04) (hash(x)=146849057)
350
+ 2900 val loss 4.3275
351
+ 2900 val perplexity 75.7511
352
+ 2900 train 4.345192 (lr=7.4231e-04) (hash(x)=145800210)
353
+ 2910 train 4.161344 (lr=7.4109e-04) (hash(x)=147316313)
354
+ 2920 train 4.205552 (lr=7.3986e-04) (hash(x)=152595415)
355
+ 2930 train 4.367274 (lr=7.3863e-04) (hash(x)=156542647)
356
+ 2940 train 4.230110 (lr=7.3740e-04) (hash(x)=141183896)
357
+ 2950 train 4.323563 (lr=7.3616e-04) (hash(x)=156329036)
358
+ 2960 train 4.369776 (lr=7.3492e-04) (hash(x)=156717793)
359
+ 2970 train 4.228608 (lr=7.3368e-04) (hash(x)=143949931)
360
+ 2980 train 4.288156 (lr=7.3243e-04) (hash(x)=155067290)
361
+ 2990 train 4.256852 (lr=7.3118e-04) (hash(x)=148774729)
362
+ 3000 val loss 4.3428
363
+ 3000 val perplexity 76.9229
364
+ 3000 train 4.249876 (lr=7.2992e-04) (hash(x)=141997485)
365
+ 3010 train 4.219102 (lr=7.2867e-04) (hash(x)=160762953)
366
+ 3020 train 4.173010 (lr=7.2740e-04) (hash(x)=170341665)
367
+ 3030 train 4.445178 (lr=7.2614e-04) (hash(x)=150722836)
368
+ 3040 train 4.242194 (lr=7.2487e-04) (hash(x)=154243307)
369
+ 3050 train 4.286341 (lr=7.2360e-04) (hash(x)=146077234)
370
+ 3060 train 4.240442 (lr=7.2232e-04) (hash(x)=152125542)
371
+ 3070 train 4.419087 (lr=7.2105e-04) (hash(x)=144149387)
372
+ 3080 train 4.249272 (lr=7.1977e-04) (hash(x)=147508712)
373
+ 3090 train 4.245663 (lr=7.1848e-04) (hash(x)=143583369)
374
+ 3100 val loss 4.3003
375
+ 3100 val perplexity 73.7203
376
+ 3100 train 4.277234 (lr=7.1719e-04) (hash(x)=154049740)
377
+ 3110 train 4.333055 (lr=7.1590e-04) (hash(x)=147054835)
378
+ 3120 train 4.319838 (lr=7.1461e-04) (hash(x)=152468568)
379
+ 3130 train 4.265347 (lr=7.1331e-04) (hash(x)=151635194)
380
+ 3140 train 4.225533 (lr=7.1201e-04) (hash(x)=150396274)
381
+ 3150 train 4.285470 (lr=7.1071e-04) (hash(x)=141306267)
382
+ 3160 train 4.234180 (lr=7.0940e-04) (hash(x)=151110762)
383
+ 3170 train 4.314109 (lr=7.0809e-04) (hash(x)=146518875)
384
+ 3180 train 4.194045 (lr=7.0678e-04) (hash(x)=139914401)
385
+ 3190 train 4.204399 (lr=7.0546e-04) (hash(x)=161631880)
386
+ 3200 val loss 4.3113
387
+ 3200 val perplexity 74.5376
388
+ 3200 train 4.222229 (lr=7.0414e-04) (hash(x)=150471842)
389
+ 3210 train 4.287651 (lr=7.0282e-04) (hash(x)=147355138)
390
+ 3220 train 4.088198 (lr=7.0149e-04) (hash(x)=138879678)
391
+ 3230 train 4.189610 (lr=7.0017e-04) (hash(x)=123929679)
392
+ 3240 train 4.255065 (lr=6.9883e-04) (hash(x)=157263734)
393
+ 3250 train 4.271533 (lr=6.9750e-04) (hash(x)=148065539)
394
+ 3260 train 4.186887 (lr=6.9616e-04) (hash(x)=138020742)
395
+ 3270 train 4.052160 (lr=6.9482e-04) (hash(x)=142053611)
396
+ 3280 train 4.212872 (lr=6.9348e-04) (hash(x)=155444784)
397
+ 3290 train 4.043209 (lr=6.9213e-04) (hash(x)=148500315)
398
+ 3300 val loss 4.3045
399
+ 3300 val perplexity 74.0292
400
+ 3300 train 4.270606 (lr=6.9079e-04) (hash(x)=149048126)
401
+ 3310 train 4.218890 (lr=6.8943e-04) (hash(x)=141680674)
402
+ 3320 train 4.166906 (lr=6.8808e-04) (hash(x)=143310642)
403
+ 3330 train 4.276791 (lr=6.8672e-04) (hash(x)=149891496)
404
+ 3340 train 4.224024 (lr=6.8536e-04) (hash(x)=151244609)
405
+ 3350 train 4.229612 (lr=6.8400e-04) (hash(x)=157158479)
406
+ 3360 train 4.092904 (lr=6.8263e-04) (hash(x)=151709876)
407
+ 3370 train 4.243066 (lr=6.8127e-04) (hash(x)=145297090)
408
+ 3380 train 4.252618 (lr=6.7990e-04) (hash(x)=154767881)
409
+ 3390 train 4.228580 (lr=6.7852e-04) (hash(x)=152600366)
410
+ 3400 val loss 4.2821
411
+ 3400 val perplexity 72.3948
412
+ 3400 train 4.466550 (lr=6.7715e-04) (hash(x)=161261339)
413
+ 3410 train 4.223199 (lr=6.7577e-04) (hash(x)=146859162)
414
+ 3420 train 4.317304 (lr=6.7439e-04) (hash(x)=149276224)
415
+ 3430 train 4.331322 (lr=6.7300e-04) (hash(x)=155866933)
416
+ 3440 train 4.200430 (lr=6.7162e-04) (hash(x)=143191904)
417
+ 3450 train 4.418372 (lr=6.7023e-04) (hash(x)=156910661)
418
+ 3460 train 4.202965 (lr=6.6883e-04) (hash(x)=140850879)
419
+ 3470 train 4.429811 (lr=6.6744e-04) (hash(x)=162140445)
420
+ 3480 train 4.171480 (lr=6.6604e-04) (hash(x)=145607495)
421
+ 3490 train 4.181160 (lr=6.6464e-04) (hash(x)=152496121)
422
+ 3500 val loss 4.2736
423
+ 3500 val perplexity 71.7818
424
+ 3500 train 4.150748 (lr=6.6324e-04) (hash(x)=157495564)
425
+ 3510 train 4.271947 (lr=6.6184e-04) (hash(x)=147657004)
426
+ 3520 train 4.244320 (lr=6.6043e-04) (hash(x)=152978060)
427
+ 3530 train 4.178273 (lr=6.5902e-04) (hash(x)=146800180)
428
+ 3540 train 4.258668 (lr=6.5761e-04) (hash(x)=195450444)
429
+ 3550 train 4.238681 (lr=6.5620e-04) (hash(x)=151114129)
430
+ 3560 train 4.225214 (lr=6.5478e-04) (hash(x)=143833190)
431
+ 3570 train 4.026104 (lr=6.5336e-04) (hash(x)=146039971)
432
+ 3580 train 4.085391 (lr=6.5194e-04) (hash(x)=145172249)
433
+ 3590 train 4.012236 (lr=6.5052e-04) (hash(x)=159164125)
434
+ 3600 val loss 4.2737
435
+ 3600 val perplexity 71.7832
436
+ 3600 train 4.248257 (lr=6.4910e-04) (hash(x)=144352932)
437
+ 3610 train 4.221739 (lr=6.4767e-04) (hash(x)=150550670)
438
+ 3620 train 4.336320 (lr=6.4624e-04) (hash(x)=147170125)
439
+ 3630 train 4.146653 (lr=6.4481e-04) (hash(x)=139473935)
440
+ 3640 train 4.134913 (lr=6.4337e-04) (hash(x)=159294424)
441
+ 3650 train 4.257586 (lr=6.4194e-04) (hash(x)=145943278)
442
+ 3660 train 4.310430 (lr=6.4050e-04) (hash(x)=141647872)
443
+ 3670 train 4.290152 (lr=6.3906e-04) (hash(x)=147109351)
444
+ 3680 train 4.313116 (lr=6.3762e-04) (hash(x)=140683368)
445
+ 3690 train 4.227198 (lr=6.3617e-04) (hash(x)=156276284)
446
+ 3700 val loss 4.2591
447
+ 3700 val perplexity 70.7495
448
+ 3700 train 4.173584 (lr=6.3473e-04) (hash(x)=149389012)
449
+ 3710 train 4.301973 (lr=6.3328e-04) (hash(x)=142107929)
450
+ 3720 train 4.274008 (lr=6.3183e-04) (hash(x)=147828843)
451
+ 3730 train 4.340503 (lr=6.3037e-04) (hash(x)=141784679)
452
+ 3740 train 4.235713 (lr=6.2892e-04) (hash(x)=153375464)
453
+ 3750 train 4.180956 (lr=6.2746e-04) (hash(x)=153716001)
454
+ 3760 train 4.144693 (lr=6.2600e-04) (hash(x)=146848822)
455
+ 3770 train 4.105702 (lr=6.2454e-04) (hash(x)=148396782)
456
+ 3780 train 4.178063 (lr=6.2308e-04) (hash(x)=153563244)
457
+ 3790 train 4.082102 (lr=6.2162e-04) (hash(x)=160808734)
458
+ 3800 val loss 4.2489
459
+ 3800 val perplexity 70.0277
460
+ 3800 train 4.273157 (lr=6.2015e-04) (hash(x)=146607620)
461
+ 3810 train 4.241128 (lr=6.1868e-04) (hash(x)=152580677)
462
+ 3820 train 4.595507 (lr=6.1721e-04) (hash(x)=144860879)
463
+ 3830 train 4.259708 (lr=6.1574e-04) (hash(x)=162099851)
464
+ 3840 train 4.221568 (lr=6.1427e-04) (hash(x)=159506855)
465
+ 3850 train 4.134172 (lr=6.1280e-04) (hash(x)=140022686)
466
+ 3860 train 4.218744 (lr=6.1132e-04) (hash(x)=152137850)
467
+ 3870 train 4.331231 (lr=6.0984e-04) (hash(x)=142666785)
468
+ 3880 train 4.185106 (lr=6.0836e-04) (hash(x)=148285868)
469
+ 3890 train 4.141440 (lr=6.0688e-04) (hash(x)=147910067)
470
+ 3900 val loss 4.2478
471
+ 3900 val perplexity 69.9536
472
+ 3900 train 4.186778 (lr=6.0540e-04) (hash(x)=143754617)
473
+ 3910 train 4.175729 (lr=6.0391e-04) (hash(x)=154779949)
474
+ 3920 train 4.021096 (lr=6.0243e-04) (hash(x)=137824061)
475
+ 3930 train 4.000633 (lr=6.0094e-04) (hash(x)=136933373)
476
+ 3940 train 4.072438 (lr=5.9945e-04) (hash(x)=164054337)
477
+ 3950 train 4.261671 (lr=5.9796e-04) (hash(x)=155752984)
478
+ 3960 train 4.376770 (lr=5.9647e-04) (hash(x)=160957673)
479
+ 3970 train 4.247396 (lr=5.9497e-04) (hash(x)=157158690)
480
+ 3980 train 4.253230 (lr=5.9348e-04) (hash(x)=153282027)
481
+ 3990 train 4.286444 (lr=5.9198e-04) (hash(x)=152183510)
482
+ 4000 val loss 4.2230
483
+ 4000 val perplexity 68.2362
484
+ 4000 train 4.174876 (lr=5.9048e-04) (hash(x)=156930722)
485
+ 4010 train 4.165983 (lr=5.8898e-04) (hash(x)=150798740)
486
+ 4020 train 4.381364 (lr=5.8748e-04) (hash(x)=142632400)
487
+ 4030 train 4.277005 (lr=5.8598e-04) (hash(x)=173326287)
488
+ 4040 train 4.182381 (lr=5.8448e-04) (hash(x)=128797443)
489
+ 4050 train 4.169764 (lr=5.8297e-04) (hash(x)=146937792)
490
+ 4060 train 4.279040 (lr=5.8147e-04) (hash(x)=164579030)
491
+ 4070 train 4.145548 (lr=5.7996e-04) (hash(x)=149459820)
492
+ 4080 train 4.108044 (lr=5.7845e-04) (hash(x)=160159595)
493
+ 4090 train 4.047360 (lr=5.7694e-04) (hash(x)=141507853)
494
+ 4100 val loss 4.2211
495
+ 4100 val perplexity 68.1098
496
+ 4100 train 4.189893 (lr=5.7543e-04) (hash(x)=147500519)
497
+ 4110 train 4.172020 (lr=5.7392e-04) (hash(x)=153297081)
498
+ 4120 train 4.171496 (lr=5.7240e-04) (hash(x)=142560288)
499
+ 4130 train 4.192123 (lr=5.7089e-04) (hash(x)=153380118)
500
+ 4140 train 4.091150 (lr=5.6937e-04) (hash(x)=143150720)
501
+ 4150 train 4.134495 (lr=5.6786e-04) (hash(x)=143459564)
502
+ 4160 train 4.163599 (lr=5.6634e-04) (hash(x)=153955741)
503
+ 4170 train 4.277952 (lr=5.6482e-04) (hash(x)=152199253)
504
+ 4180 train 4.039390 (lr=5.6330e-04) (hash(x)=165957668)
505
+ 4190 train 4.089068 (lr=5.6178e-04) (hash(x)=144866923)
506
+ 4200 val loss 4.2285
507
+ 4200 val perplexity 68.6133
508
+ 4200 train 4.101661 (lr=5.6026e-04) (hash(x)=143232237)
509
+ 4210 train 4.017132 (lr=5.5874e-04) (hash(x)=151907163)
510
+ 4220 train 4.107684 (lr=5.5721e-04) (hash(x)=145917587)
511
+ 4230 train 3.947146 (lr=5.5569e-04) (hash(x)=147258846)
512
+ 4240 train 4.101050 (lr=5.5416e-04) (hash(x)=143449843)
513
+ 4250 train 4.061218 (lr=5.5264e-04) (hash(x)=184883715)
514
+ 4260 train 4.160358 (lr=5.5111e-04) (hash(x)=156958761)
515
+ 4270 train 4.225371 (lr=5.4958e-04) (hash(x)=168870544)
516
+ 4280 train 4.158155 (lr=5.4805e-04) (hash(x)=179446735)
517
+ 4290 train 4.197602 (lr=5.4652e-04) (hash(x)=145649386)
518
+ 4300 val loss 4.2114
519
+ 4300 val perplexity 67.4537
520
+ 4300 train 4.045122 (lr=5.4499e-04) (hash(x)=146811670)
521
+ 4310 train 4.315369 (lr=5.4346e-04) (hash(x)=151907075)
522
+ 4320 train 4.118766 (lr=5.4193e-04) (hash(x)=138628365)
523
+ 4330 train 4.177659 (lr=5.4040e-04) (hash(x)=151608240)
524
+ 4340 train 4.108645 (lr=5.3887e-04) (hash(x)=160111319)
525
+ 4350 train 4.255198 (lr=5.3733e-04) (hash(x)=144958652)
526
+ 4360 train 4.042944 (lr=5.3580e-04) (hash(x)=149176222)
527
+ 4370 train 4.137443 (lr=5.3427e-04) (hash(x)=216221371)
528
+ 4380 train 4.049479 (lr=5.3273e-04) (hash(x)=141760648)
529
+ 4390 train 4.133317 (lr=5.3119e-04) (hash(x)=151710283)
530
+ 4400 val loss 4.2010
531
+ 4400 val perplexity 66.7534
532
+ 4400 train 4.259853 (lr=5.2966e-04) (hash(x)=158418746)
533
+ 4410 train 4.111242 (lr=5.2812e-04) (hash(x)=142331301)
534
+ 4420 train 4.201407 (lr=5.2658e-04) (hash(x)=146507180)
535
+ 4430 train 4.289933 (lr=5.2505e-04) (hash(x)=166506937)
536
+ 4440 train 4.231297 (lr=5.2351e-04) (hash(x)=151209447)
537
+ 4450 train 4.188719 (lr=5.2197e-04) (hash(x)=149186822)
538
+ 4460 train 4.173400 (lr=5.2043e-04) (hash(x)=144348412)
539
+ 4470 train 4.060436 (lr=5.1889e-04) (hash(x)=156494284)
540
+ 4480 train 4.095069 (lr=5.1735e-04) (hash(x)=147238744)
541
+ 4490 train 4.035028 (lr=5.1581e-04) (hash(x)=150746185)
542
+ 4500 val loss 4.1935
543
+ 4500 val perplexity 66.2554
544
+ 4500 train 4.130319 (lr=5.1427e-04) (hash(x)=156695778)
545
+ 4510 train 3.974437 (lr=5.1273e-04) (hash(x)=135988958)
546
+ 4520 train 4.153279 (lr=5.1119e-04) (hash(x)=148272110)
547
+ 4530 train 4.064757 (lr=5.0965e-04) (hash(x)=136459794)
548
+ 4540 train 4.100585 (lr=5.0811e-04) (hash(x)=147624954)
549
+ 4550 train 4.099239 (lr=5.0657e-04) (hash(x)=154947199)
550
+ 4560 train 4.153473 (lr=5.0502e-04) (hash(x)=147869166)
551
+ 4570 train 4.011271 (lr=5.0348e-04) (hash(x)=148415463)
552
+ 4580 train 4.169740 (lr=5.0194e-04) (hash(x)=152104066)
553
+ 4590 train 4.072468 (lr=5.0040e-04) (hash(x)=147060040)
554
+ 4600 val loss 4.1919
555
+ 4600 val perplexity 66.1451
556
+ 4600 train 4.142438 (lr=4.9886e-04) (hash(x)=147791497)
557
+ 4610 train 4.341977 (lr=4.9731e-04) (hash(x)=155595556)
558
+ 4620 train 4.191512 (lr=4.9577e-04) (hash(x)=147178858)
559
+ 4630 train 4.235603 (lr=4.9423e-04) (hash(x)=152617112)
560
+ 4640 train 4.235698 (lr=4.9269e-04) (hash(x)=161766869)
561
+ 4650 train 4.221074 (lr=4.9114e-04) (hash(x)=158075561)
562
+ 4660 train 4.258034 (lr=4.8960e-04) (hash(x)=145360002)
563
+ 4670 train 4.139149 (lr=4.8806e-04) (hash(x)=144538791)
564
+ 4680 train 4.179411 (lr=4.8652e-04) (hash(x)=142711273)
565
+ 4690 train 4.333440 (lr=4.8498e-04) (hash(x)=193907362)
566
+ 4700 val loss 4.1719
567
+ 4700 val perplexity 64.8355
568
+ 4700 train 4.152661 (lr=4.8343e-04) (hash(x)=155533088)
569
+ 4710 train 4.099715 (lr=4.8189e-04) (hash(x)=148618528)
570
+ 4720 train 4.214475 (lr=4.8035e-04) (hash(x)=150502624)
571
+ 4730 train 4.083520 (lr=4.7881e-04) (hash(x)=189205847)
572
+ 4740 train 4.134410 (lr=4.7727e-04) (hash(x)=167322800)
573
+ 4750 train 4.252748 (lr=4.7573e-04) (hash(x)=162169095)
574
+ 4760 train 3.958517 (lr=4.7419e-04) (hash(x)=143459469)
575
+ 4770 train 4.034489 (lr=4.7265e-04) (hash(x)=144386593)
576
+ 4780 train 4.003113 (lr=4.7111e-04) (hash(x)=143721184)
577
+ 4790 train 4.070515 (lr=4.6957e-04) (hash(x)=146077222)
578
+ 4800 val loss 4.1752
579
+ 4800 val perplexity 65.0499
580
+ 4800 train 4.053797 (lr=4.6803e-04) (hash(x)=138350044)
581
+ 4810 train 4.053739 (lr=4.6649e-04) (hash(x)=154960112)
582
+ 4820 train 3.966036 (lr=4.6495e-04) (hash(x)=146950685)
583
+ 4830 train 3.965119 (lr=4.6342e-04) (hash(x)=140224554)
584
+ 4840 train 3.959418 (lr=4.6188e-04) (hash(x)=156640495)
585
+ 4850 train 4.172421 (lr=4.6034e-04) (hash(x)=153409150)
586
+ 4860 train 4.090921 (lr=4.5881e-04) (hash(x)=151647700)
587
+ 4870 train 4.493041 (lr=4.5727e-04) (hash(x)=155465603)
588
+ 4880 train 4.335605 (lr=4.5573e-04) (hash(x)=153404400)
589
+ 4890 train 4.224070 (lr=4.5420e-04) (hash(x)=170409281)
590
+ 4900 val loss 4.1615
591
+ 4900 val perplexity 64.1691
592
+ 4900 train 4.078154 (lr=4.5267e-04) (hash(x)=143735284)
593
+ 4910 train 4.215267 (lr=4.5113e-04) (hash(x)=150784329)
594
+ 4920 train 4.103313 (lr=4.4960e-04) (hash(x)=152353574)
595
+ 4930 train 4.055473 (lr=4.4807e-04) (hash(x)=152677008)
596
+ 4940 train 4.134968 (lr=4.4654e-04) (hash(x)=140553861)
597
+ 4950 train 4.225443 (lr=4.4501e-04) (hash(x)=152869181)
598
+ 4960 train 4.176461 (lr=4.4348e-04) (hash(x)=153011079)
599
+ 4970 train 4.123913 (lr=4.4195e-04) (hash(x)=152796617)
600
+ 4980 train 4.238431 (lr=4.4042e-04) (hash(x)=146868417)
601
+ 4990 train 4.127503 (lr=4.3889e-04) (hash(x)=140168188)
602
+ 5000 val loss 4.1516
603
+ 5000 val perplexity 63.5338
604
+ 5000 train 4.105338 (lr=4.3736e-04) (hash(x)=154976463)
605
+ 5010 train 4.161404 (lr=4.3584e-04) (hash(x)=148582690)
606
+ 5020 train 4.169558 (lr=4.3431e-04) (hash(x)=166715464)
607
+ 5030 train 4.135546 (lr=4.3279e-04) (hash(x)=150616397)
608
+ 5040 train 4.185603 (lr=4.3126e-04) (hash(x)=150775548)
609
+ 5050 train 4.169393 (lr=4.2974e-04) (hash(x)=150575367)
610
+ 5060 train 4.096206 (lr=4.2822e-04) (hash(x)=152352082)
611
+ 5070 train 4.130648 (lr=4.2670e-04) (hash(x)=150269272)
612
+ 5080 train 4.188220 (lr=4.2518e-04) (hash(x)=146499361)
613
+ 5090 train 3.873008 (lr=4.2366e-04) (hash(x)=148734454)
614
+ 5100 val loss 4.1538
615
+ 5100 val perplexity 63.6724
616
+ 5100 train 4.055887 (lr=4.2214e-04) (hash(x)=149894982)
617
+ 5110 train 4.107533 (lr=4.2063e-04) (hash(x)=154487751)
618
+ 5120 train 4.065097 (lr=4.1911e-04) (hash(x)=144169582)
619
+ 5130 train 4.038475 (lr=4.1760e-04) (hash(x)=152217630)
620
+ 5140 train 4.061378 (lr=4.1608e-04) (hash(x)=158233831)
621
+ 5150 train 4.127509 (lr=4.1457e-04) (hash(x)=142746426)
622
+ 5160 train 4.112732 (lr=4.1306e-04) (hash(x)=139184458)
623
+ 5170 train 4.198362 (lr=4.1155e-04) (hash(x)=140624857)
624
+ 5180 train 3.922185 (lr=4.1004e-04) (hash(x)=154107423)
625
+ 5190 train 3.863898 (lr=4.0853e-04) (hash(x)=144892610)
626
+ 5200 val loss 4.1540
627
+ 5200 val perplexity 63.6878
628
+ 5200 train 3.818167 (lr=4.0703e-04) (hash(x)=159326689)
629
+ 5210 train 3.984487 (lr=4.0552e-04) (hash(x)=151186017)
630
+ 5220 train 3.789221 (lr=4.0402e-04) (hash(x)=148996819)
631
+ 5230 train 3.735144 (lr=4.0252e-04) (hash(x)=150829917)
632
+ 5240 train 3.852046 (lr=4.0102e-04) (hash(x)=152758972)
633
+ 5250 train 3.824058 (lr=3.9952e-04) (hash(x)=154345903)
634
+ 5260 train 3.816532 (lr=3.9802e-04) (hash(x)=150134947)
635
+ 5270 train 3.911081 (lr=3.9652e-04) (hash(x)=149689310)
636
+ 5280 train 4.153986 (lr=3.9503e-04) (hash(x)=147662223)
637
+ 5290 train 4.203696 (lr=3.9353e-04) (hash(x)=165770608)
638
+ 5300 val loss 4.1410
639
+ 5300 val perplexity 62.8638
640
+ 5300 train 4.191709 (lr=3.9204e-04) (hash(x)=159484800)
641
+ 5310 train 4.221487 (lr=3.9055e-04) (hash(x)=152672536)
642
+ 5320 train 4.051221 (lr=3.8906e-04) (hash(x)=152728812)
643
+ 5330 train 3.991126 (lr=3.8757e-04) (hash(x)=152778685)
644
+ 5340 train 4.087645 (lr=3.8609e-04) (hash(x)=149818849)
645
+ 5350 train 4.015409 (lr=3.8460e-04) (hash(x)=168239739)
646
+ 5360 train 4.163786 (lr=3.8312e-04) (hash(x)=139859423)
647
+ 5370 train 3.985743 (lr=3.8164e-04) (hash(x)=139315327)
648
+ 5380 train 4.198350 (lr=3.8016e-04) (hash(x)=163068509)
649
+ 5390 train 4.152156 (lr=3.7868e-04) (hash(x)=147645415)
650
+ 5400 val loss 4.1206
651
+ 5400 val perplexity 61.5977
652
+ 5400 train 4.110143 (lr=3.7720e-04) (hash(x)=140385615)
653
+ 5410 train 3.970238 (lr=3.7573e-04) (hash(x)=151381862)
654
+ 5420 train 4.118372 (lr=3.7426e-04) (hash(x)=152445092)
655
+ 5430 train 4.132278 (lr=3.7279e-04) (hash(x)=142454479)
656
+ 5440 train 4.089479 (lr=3.7132e-04) (hash(x)=151308771)
657
+ 5450 train 3.944021 (lr=3.6985e-04) (hash(x)=150886070)
658
+ 5460 train 4.047285 (lr=3.6838e-04) (hash(x)=146080408)
659
+ 5470 train 4.071188 (lr=3.6692e-04) (hash(x)=149828857)
660
+ 5480 train 4.040788 (lr=3.6546e-04) (hash(x)=139922034)
661
+ 5490 train 3.987920 (lr=3.6400e-04) (hash(x)=150262276)
662
+ 5500 val loss 4.1229
663
+ 5500 val perplexity 61.7372
664
+ 5500 train 3.964576 (lr=3.6254e-04) (hash(x)=148498335)
665
+ 5510 train 4.081041 (lr=3.6108e-04) (hash(x)=157036224)
666
+ 5520 train 3.959148 (lr=3.5963e-04) (hash(x)=134093409)
667
+ 5530 train 3.989160 (lr=3.5817e-04) (hash(x)=153126250)
668
+ 5540 train 3.998140 (lr=3.5672e-04) (hash(x)=136170985)
669
+ 5550 train 3.976689 (lr=3.5527e-04) (hash(x)=152841201)
670
+ 5560 train 4.082709 (lr=3.5383e-04) (hash(x)=147439014)
671
+ 5570 train 4.001879 (lr=3.5238e-04) (hash(x)=143476132)
672
+ 5580 train 3.769162 (lr=3.5094e-04) (hash(x)=139503437)
673
+ 5590 train 3.813571 (lr=3.4950e-04) (hash(x)=131664955)
674
+ 5600 val loss 4.1302
675
+ 5600 val perplexity 62.1909
676
+ 5600 train 3.910072 (lr=3.4806e-04) (hash(x)=151907614)
677
+ 5610 train 3.795213 (lr=3.4663e-04) (hash(x)=157132674)
678
+ 5620 train 3.992025 (lr=3.4519e-04) (hash(x)=146793775)
679
+ 5630 train 3.948164 (lr=3.4376e-04) (hash(x)=147237910)
680
+ 5640 train 3.854827 (lr=3.4233e-04) (hash(x)=141120024)
681
+ 5650 train 3.825000 (lr=3.4090e-04) (hash(x)=144275645)
682
+ 5660 train 3.859615 (lr=3.3948e-04) (hash(x)=159088002)
683
+ 5670 train 3.910187 (lr=3.3806e-04) (hash(x)=147224479)
684
+ 5680 train 4.262881 (lr=3.3664e-04) (hash(x)=147208743)
685
+ 5690 train 4.063407 (lr=3.3522e-04) (hash(x)=154384701)
686
+ 5700 val loss 4.1225
687
+ 5700 val perplexity 61.7157
688
+ 5700 train 4.098000 (lr=3.3380e-04) (hash(x)=155192267)
689
+ 5710 train 4.047134 (lr=3.3239e-04) (hash(x)=148458753)
690
+ 5720 train 4.150775 (lr=3.3098e-04) (hash(x)=159947204)
691
+ 5730 train 4.137318 (lr=3.2957e-04) (hash(x)=150746366)
692
+ 5740 train 4.138463 (lr=3.2816e-04) (hash(x)=155201884)
693
+ 5750 train 4.129984 (lr=3.2676e-04) (hash(x)=156432008)
694
+ 5760 train 4.108051 (lr=3.2536e-04) (hash(x)=185977746)
695
+ 5770 train 4.111618 (lr=3.2396e-04) (hash(x)=133620000)
696
+ 5780 train 4.115201 (lr=3.2256e-04) (hash(x)=153778044)
697
+ 5790 train 3.966661 (lr=3.2117e-04) (hash(x)=147713822)
698
+ 5800 val loss 4.1013
699
+ 5800 val perplexity 60.4209
700
+ 5800 train 4.120602 (lr=3.1977e-04) (hash(x)=153132158)
701
+ 5810 train 4.074697 (lr=3.1838e-04) (hash(x)=144293895)
702
+ 5820 train 4.059621 (lr=3.1700e-04) (hash(x)=151500590)
703
+ 5830 train 4.015221 (lr=3.1561e-04) (hash(x)=139308707)
704
+ 5840 train 4.115336 (lr=3.1423e-04) (hash(x)=153591278)
705
+ 5850 train 4.102745 (lr=3.1285e-04) (hash(x)=146077037)
706
+ 5860 train 4.070503 (lr=3.1148e-04) (hash(x)=149160733)
707
+ 5870 train 4.049407 (lr=3.1010e-04) (hash(x)=138482476)
708
+ 5880 train 4.061591 (lr=3.0873e-04) (hash(x)=152173137)
709
+ 5890 train 4.040834 (lr=3.0737e-04) (hash(x)=142387057)
710
+ 5900 val loss 4.1093
711
+ 5900 val perplexity 60.9056
712
+ 5900 train 4.107385 (lr=3.0600e-04) (hash(x)=161446764)
713
+ 5910 train 3.987519 (lr=3.0464e-04) (hash(x)=144512138)
714
+ 5920 train 4.048494 (lr=3.0328e-04) (hash(x)=152555618)
715
+ 5930 train 4.010263 (lr=3.0192e-04) (hash(x)=146129367)
716
+ 5940 train 3.935830 (lr=3.0057e-04) (hash(x)=154607527)
717
+ 5950 train 4.032087 (lr=2.9921e-04) (hash(x)=143827143)
718
+ 5960 train 4.160122 (lr=2.9787e-04) (hash(x)=151168510)
719
+ 5970 train 3.944741 (lr=2.9652e-04) (hash(x)=158961574)
720
+ 5980 train 3.781619 (lr=2.9518e-04) (hash(x)=148757787)
721
+ 5990 train 4.282948 (lr=2.9384e-04) (hash(x)=161719494)
722
+ 6000 val loss 4.1084
723
+ 6000 val perplexity 60.8520
724
+ 6000 train 3.781684 (lr=2.9250e-04) (hash(x)=151512446)
725
+ 6010 train 3.898590 (lr=2.9117e-04) (hash(x)=173500735)
726
+ 6020 train 4.037614 (lr=2.8983e-04) (hash(x)=157621031)
727
+ 6030 train 3.921065 (lr=2.8851e-04) (hash(x)=142662684)
728
+ 6040 train 3.910671 (lr=2.8718e-04) (hash(x)=146180940)
729
+ 6050 train 3.824476 (lr=2.8586e-04) (hash(x)=147654412)
730
+ 6060 train 3.703630 (lr=2.8454e-04) (hash(x)=156393157)
731
+ 6070 train 4.099494 (lr=2.8322e-04) (hash(x)=156947300)
732
+ 6080 train 4.092514 (lr=2.8191e-04) (hash(x)=163108626)
733
+ 6090 train 4.191934 (lr=2.8060e-04) (hash(x)=158654713)
734
+ 6100 val loss 4.0897
735
+ 6100 val perplexity 59.7236
736
+ 6100 train 3.930384 (lr=2.7929e-04) (hash(x)=188094053)
737
+ 6110 train 4.075995 (lr=2.7799e-04) (hash(x)=145132600)
738
+ 6120 train 3.949366 (lr=2.7669e-04) (hash(x)=146999707)
739
+ 6130 train 4.187688 (lr=2.7539e-04) (hash(x)=145719530)
740
+ 6140 train 4.122720 (lr=2.7410e-04) (hash(x)=153230722)
741
+ 6150 train 4.053794 (lr=2.7281e-04) (hash(x)=147455934)
742
+ 6160 train 4.108325 (lr=2.7152e-04) (hash(x)=146596242)
743
+ 6170 train 4.203628 (lr=2.7023e-04) (hash(x)=152029855)
744
+ 6180 train 4.068706 (lr=2.6895e-04) (hash(x)=151480034)
745
+ 6190 train 3.997958 (lr=2.6768e-04) (hash(x)=147381047)
746
+ 6200 val loss 4.0757
747
+ 6200 val perplexity 58.8899
748
+ 6200 train 3.986104 (lr=2.6640e-04) (hash(x)=149389789)
749
+ 6210 train 4.036395 (lr=2.6513e-04) (hash(x)=154894138)
750
+ 6220 train 4.071010 (lr=2.6386e-04) (hash(x)=150074302)
751
+ 6230 train 3.936634 (lr=2.6260e-04) (hash(x)=142881794)
752
+ 6240 train 3.919313 (lr=2.6133e-04) (hash(x)=164747461)
753
+ 6250 train 4.097204 (lr=2.6008e-04) (hash(x)=158378292)
754
+ 6260 train 4.093693 (lr=2.5882e-04) (hash(x)=147675938)
755
+ 6270 train 4.042212 (lr=2.5757e-04) (hash(x)=148197228)
756
+ 6280 train 4.002607 (lr=2.5632e-04) (hash(x)=135296747)
757
+ 6290 train 4.069618 (lr=2.5508e-04) (hash(x)=156439346)
758
+ 6300 val loss 4.0743
759
+ 6300 val perplexity 58.8097
760
+ 6300 train 3.894879 (lr=2.5384e-04) (hash(x)=138212820)
761
+ 6310 train 4.188173 (lr=2.5260e-04) (hash(x)=165985336)
762
+ 6320 train 4.059986 (lr=2.5137e-04) (hash(x)=145921897)
763
+ 6330 train 4.268520 (lr=2.5014e-04) (hash(x)=145502689)
764
+ 6340 train 4.007178 (lr=2.4891e-04) (hash(x)=141704937)
765
+ 6350 train 4.058394 (lr=2.4769e-04) (hash(x)=161789400)
766
+ 6360 train 3.782904 (lr=2.4647e-04) (hash(x)=145457780)
767
+ 6370 train 4.034250 (lr=2.4525e-04) (hash(x)=171628530)
768
+ 6380 train 3.825783 (lr=2.4404e-04) (hash(x)=153467923)
769
+ 6390 train 4.201286 (lr=2.4283e-04) (hash(x)=158066904)
770
+ 6400 val loss 4.0805
771
+ 6400 val perplexity 59.1748
772
+ 6400 train 3.849966 (lr=2.4163e-04) (hash(x)=146535423)
773
+ 6410 train 3.712233 (lr=2.4042e-04) (hash(x)=147264155)
774
+ 6420 train 3.814811 (lr=2.3923e-04) (hash(x)=145595485)
775
+ 6430 train 3.965225 (lr=2.3803e-04) (hash(x)=146929074)
776
+ 6440 train 3.891905 (lr=2.3684e-04) (hash(x)=147290878)
777
+ 6450 train 3.985863 (lr=2.3566e-04) (hash(x)=141154205)
778
+ 6460 train 3.826535 (lr=2.3447e-04) (hash(x)=154360901)
779
+ 6470 train 4.161309 (lr=2.3330e-04) (hash(x)=146350558)
780
+ 6480 train 4.127403 (lr=2.3212e-04) (hash(x)=153927683)
781
+ 6490 train 4.041520 (lr=2.3095e-04) (hash(x)=149337047)
782
+ 6500 val loss 4.0666
783
+ 6500 val perplexity 58.3571
784
+ 6500 train 3.959068 (lr=2.2978e-04) (hash(x)=145950843)
785
+ 6510 train 4.025646 (lr=2.2862e-04) (hash(x)=142577653)
786
+ 6520 train 4.079893 (lr=2.2746e-04) (hash(x)=165060689)
787
+ 6530 train 3.931062 (lr=2.2630e-04) (hash(x)=148367505)
788
+ 6540 train 4.026183 (lr=2.2515e-04) (hash(x)=154313651)
789
+ 6550 train 3.942679 (lr=2.2400e-04) (hash(x)=147477315)
790
+ 6560 train 4.037947 (lr=2.2286e-04) (hash(x)=139825403)
791
+ 6570 train 4.041677 (lr=2.2172e-04) (hash(x)=141482518)
792
+ 6580 train 4.086037 (lr=2.2058e-04) (hash(x)=144968144)
793
+ 6590 train 4.038290 (lr=2.1945e-04) (hash(x)=154319720)
794
+ 6600 val loss 4.0623
795
+ 6600 val perplexity 58.1095
796
+ 6600 train 4.075618 (lr=2.1832e-04) (hash(x)=141162902)
797
+ 6610 train 4.038085 (lr=2.1720e-04) (hash(x)=159490449)
798
+ 6620 train 3.925084 (lr=2.1608e-04) (hash(x)=151189772)
799
+ 6630 train 3.906769 (lr=2.1496e-04) (hash(x)=148451651)
800
+ 6640 train 4.023790 (lr=2.1385e-04) (hash(x)=152815550)
801
+ 6650 train 3.875006 (lr=2.1274e-04) (hash(x)=144316094)
802
+ 6660 train 4.201859 (lr=2.1164e-04) (hash(x)=143951240)
803
+ 6670 train 3.818467 (lr=2.1054e-04) (hash(x)=143203472)
804
+ 6680 train 3.861635 (lr=2.0944e-04) (hash(x)=132063806)
805
+ 6690 train 3.953725 (lr=2.0835e-04) (hash(x)=140739750)
806
+ 6700 val loss 4.0676
807
+ 6700 val perplexity 58.4138
808
+ 6700 train 4.016609 (lr=2.0726e-04) (hash(x)=153018737)
809
+ 6710 train 3.878596 (lr=2.0618e-04) (hash(x)=140107197)
810
+ 6720 train 3.982227 (lr=2.0510e-04) (hash(x)=144396559)
811
+ 6730 train 3.825402 (lr=2.0402e-04) (hash(x)=148038991)
812
+ 6740 train 4.044545 (lr=2.0295e-04) (hash(x)=145504746)
813
+ 6750 train 4.025566 (lr=2.0189e-04) (hash(x)=119579094)
814
+ 6760 train 4.041597 (lr=2.0083e-04) (hash(x)=161725430)
815
+ 6770 train 4.062311 (lr=1.9977e-04) (hash(x)=153646966)
816
+ 6780 train 4.065840 (lr=1.9871e-04) (hash(x)=153838463)
817
+ 6790 train 4.188358 (lr=1.9766e-04) (hash(x)=157891196)
818
+ 6800 val loss 4.0497
819
+ 6800 val perplexity 57.3800
820
+ 6800 train 3.901916 (lr=1.9662e-04) (hash(x)=155640155)
821
+ 6810 train 4.102930 (lr=1.9558e-04) (hash(x)=154913809)
822
+ 6820 train 4.015209 (lr=1.9454e-04) (hash(x)=159412493)
823
+ 6830 train 3.972904 (lr=1.9351e-04) (hash(x)=144756549)
824
+ 6840 train 4.182686 (lr=1.9248e-04) (hash(x)=143824638)
825
+ 6850 train 4.517451 (lr=1.9146e-04) (hash(x)=117180982)
826
+ 6860 train 4.031972 (lr=1.9044e-04) (hash(x)=141887150)
827
+ 6870 train 4.099271 (lr=1.8943e-04) (hash(x)=140094872)
828
+ 6880 train 3.939503 (lr=1.8842e-04) (hash(x)=141479953)
829
+ 6890 train 3.999678 (lr=1.8741e-04) (hash(x)=153166840)
830
+ 6900 val loss 4.0474
831
+ 6900 val perplexity 57.2490
832
+ 6900 train 4.028152 (lr=1.8641e-04) (hash(x)=153722115)
833
+ 6910 train 4.131654 (lr=1.8541e-04) (hash(x)=149578602)
834
+ 6920 train 3.988870 (lr=1.8442e-04) (hash(x)=149925108)
835
+ 6930 train 3.927510 (lr=1.8343e-04) (hash(x)=148589489)
836
+ 6940 train 3.845987 (lr=1.8245e-04) (hash(x)=144243794)
837
+ 6950 train 3.923504 (lr=1.8147e-04) (hash(x)=157743480)
838
+ 6960 train 3.916663 (lr=1.8050e-04) (hash(x)=139858247)
839
+ 6970 train 3.842264 (lr=1.7953e-04) (hash(x)=148727536)
840
+ 6980 train 3.953067 (lr=1.7856e-04) (hash(x)=153382595)
841
+ 6990 train 4.098304 (lr=1.7760e-04) (hash(x)=147529919)
842
+ 7000 val loss 4.0408
843
+ 7000 val perplexity 56.8704
844
+ 7000 train 4.060291 (lr=1.7665e-04) (hash(x)=146953450)
845
+ 7010 train 3.970274 (lr=1.7570e-04) (hash(x)=134561015)
846
+ 7020 train 4.081450 (lr=1.7475e-04) (hash(x)=148433515)
847
+ 7030 train 3.934173 (lr=1.7381e-04) (hash(x)=140394326)
848
+ 7040 train 3.937056 (lr=1.7287e-04) (hash(x)=136740763)
849
+ 7050 train 3.779110 (lr=1.7194e-04) (hash(x)=148637938)
850
+ 7060 train 3.948966 (lr=1.7101e-04) (hash(x)=147262147)
851
+ 7070 train 3.922874 (lr=1.7009e-04) (hash(x)=140191039)
852
+ 7080 train 3.936520 (lr=1.6917e-04) (hash(x)=148223305)
853
+ 7090 train 3.853813 (lr=1.6826e-04) (hash(x)=139984988)
854
+ 7100 val loss 4.0451
855
+ 7100 val perplexity 57.1170
856
+ 7100 train 3.962249 (lr=1.6735e-04) (hash(x)=137663885)
857
+ 7110 train 3.845660 (lr=1.6644e-04) (hash(x)=150955021)
858
+ 7120 train 3.756344 (lr=1.6554e-04) (hash(x)=148585682)
859
+ 7130 train 3.948707 (lr=1.6465e-04) (hash(x)=146055548)
860
+ 7140 train 3.994876 (lr=1.6376e-04) (hash(x)=155723541)
861
+ 7150 train 4.068661 (lr=1.6288e-04) (hash(x)=140465834)
862
+ 7160 train 4.037286 (lr=1.6200e-04) (hash(x)=163246922)
863
+ 7170 train 3.899239 (lr=1.6112e-04) (hash(x)=150754166)
864
+ 7180 train 4.442286 (lr=1.6025e-04) (hash(x)=142844151)
865
+ 7190 train 4.019622 (lr=1.5938e-04) (hash(x)=145167470)
866
+ 7200 val loss 4.0287
867
+ 7200 val perplexity 56.1898
868
+ 7200 train 4.214263 (lr=1.5852e-04) (hash(x)=146172950)
869
+ 7210 train 4.042536 (lr=1.5767e-04) (hash(x)=143155287)
870
+ 7220 train 3.993923 (lr=1.5682e-04) (hash(x)=151976669)
871
+ 7230 train 4.106172 (lr=1.5597e-04) (hash(x)=145979940)
872
+ 7240 train 4.073505 (lr=1.5513e-04) (hash(x)=155656034)
873
+ 7250 train 4.075167 (lr=1.5429e-04) (hash(x)=135610136)
874
+ 7260 train 4.158146 (lr=1.5346e-04) (hash(x)=148133023)
875
+ 7270 train 4.034868 (lr=1.5263e-04) (hash(x)=148006681)
876
+ 7280 train 4.064709 (lr=1.5181e-04) (hash(x)=151397627)
877
+ 7290 train 4.025921 (lr=1.5100e-04) (hash(x)=141696871)
878
+ 7300 val loss 4.0252
879
+ 7300 val perplexity 55.9909
880
+ 7300 train 3.951459 (lr=1.5019e-04) (hash(x)=150018163)
881
+ 7310 train 3.995695 (lr=1.4938e-04) (hash(x)=154900050)
882
+ 7320 train 3.900992 (lr=1.4858e-04) (hash(x)=142608283)
883
+ 7330 train 4.002235 (lr=1.4778e-04) (hash(x)=148977837)
884
+ 7340 train 4.009742 (lr=1.4699e-04) (hash(x)=141676907)
885
+ 7350 train 3.902120 (lr=1.4620e-04) (hash(x)=147846997)
886
+ 7360 train 4.147460 (lr=1.4542e-04) (hash(x)=143252181)
887
+ 7370 train 3.965926 (lr=1.4465e-04) (hash(x)=155081717)
888
+ 7380 train 4.051342 (lr=1.4387e-04) (hash(x)=149678451)
889
+ 7390 train 3.880734 (lr=1.4311e-04) (hash(x)=159341637)
890
+ 7400 val loss 4.0256
891
+ 7400 val perplexity 56.0113
892
+ 7400 train 4.168962 (lr=1.4235e-04) (hash(x)=145351166)
893
+ 7410 train 3.982915 (lr=1.4159e-04) (hash(x)=162274009)
894
+ 7420 train 3.823453 (lr=1.4084e-04) (hash(x)=143254858)
895
+ 7430 train 3.944281 (lr=1.4010e-04) (hash(x)=143049884)
896
+ 7440 train 3.868463 (lr=1.3936e-04) (hash(x)=141768538)
897
+ 7450 train 3.892488 (lr=1.3862e-04) (hash(x)=139895310)
898
+ 7460 train 3.763549 (lr=1.3789e-04) (hash(x)=155900365)
899
+ 7470 train 3.743206 (lr=1.3717e-04) (hash(x)=145048215)
900
+ 7480 train 3.885258 (lr=1.3645e-04) (hash(x)=136992624)
901
+ 7490 train 3.996619 (lr=1.3573e-04) (hash(x)=141424789)
902
+ 7500 val loss 4.0310
903
+ 7500 val perplexity 56.3175
904
+ 7500 train 3.846631 (lr=1.3502e-04) (hash(x)=145292116)
905
+ 7510 train 3.880993 (lr=1.3432e-04) (hash(x)=144818029)
906
+ 7520 train 4.051399 (lr=1.3362e-04) (hash(x)=152903515)
907
+ 7530 train 4.060190 (lr=1.3293e-04) (hash(x)=156934724)
908
+ 7540 train 4.282737 (lr=1.3224e-04) (hash(x)=156695926)
909
+ 7550 train 4.041656 (lr=1.3155e-04) (hash(x)=149115609)
910
+ 7560 train 4.021070 (lr=1.3088e-04) (hash(x)=143974224)
911
+ 7570 train 3.975363 (lr=1.3020e-04) (hash(x)=148204424)
912
+ 7580 train 4.061179 (lr=1.2954e-04) (hash(x)=150177910)
913
+ 7590 train 4.074142 (lr=1.2887e-04) (hash(x)=140325629)
914
+ 7600 val loss 4.0166
915
+ 7600 val perplexity 55.5144
916
+ 7600 train 3.957449 (lr=1.2822e-04) (hash(x)=150235132)
917
+ 7610 train 4.080404 (lr=1.2757e-04) (hash(x)=157594837)
918
+ 7620 train 4.095534 (lr=1.2692e-04) (hash(x)=146696749)
919
+ 7630 train 4.126130 (lr=1.2628e-04) (hash(x)=148463580)
920
+ 7640 train 3.949636 (lr=1.2564e-04) (hash(x)=140271046)
921
+ 7650 train 4.005832 (lr=1.2501e-04) (hash(x)=159952319)
922
+ 7660 train 4.004004 (lr=1.2439e-04) (hash(x)=146307392)
923
+ 7670 train 3.957639 (lr=1.2377e-04) (hash(x)=145027988)
924
+ 7680 train 4.024033 (lr=1.2316e-04) (hash(x)=147729698)
925
+ 7690 train 4.013694 (lr=1.2255e-04) (hash(x)=143401210)
926
+ 7700 val loss 4.0118
927
+ 7700 val perplexity 55.2482
928
+ 7700 train 3.939029 (lr=1.2194e-04) (hash(x)=154543455)
929
+ 7710 train 4.022234 (lr=1.2135e-04) (hash(x)=151600456)
930
+ 7720 train 3.933409 (lr=1.2076e-04) (hash(x)=165913803)
931
+ 7730 train 3.801821 (lr=1.2017e-04) (hash(x)=144942059)
932
+ 7740 train 4.027657 (lr=1.1959e-04) (hash(x)=150302103)
933
+ 7750 train 3.971071 (lr=1.1901e-04) (hash(x)=155782458)
934
+ 7760 train 4.005627 (lr=1.1844e-04) (hash(x)=147824384)
935
+ 7770 train 4.019308 (lr=1.1788e-04) (hash(x)=149177657)
936
+ 7780 train 4.037041 (lr=1.1732e-04) (hash(x)=164446738)
937
+ 7790 train 3.845381 (lr=1.1676e-04) (hash(x)=146417077)
938
+ 7800 val loss 4.0143
939
+ 7800 val perplexity 55.3869
940
+ 7800 train 3.989697 (lr=1.1621e-04) (hash(x)=142456852)
941
+ 7810 train 3.691928 (lr=1.1567e-04) (hash(x)=146221807)
942
+ 7820 train 3.887654 (lr=1.1513e-04) (hash(x)=154136533)
943
+ 7830 train 3.796869 (lr=1.1460e-04) (hash(x)=147492735)
944
+ 7840 train 3.954353 (lr=1.1407e-04) (hash(x)=143481291)
945
+ 7850 train 3.778156 (lr=1.1355e-04) (hash(x)=136749307)
946
+ 7860 train 3.803327 (lr=1.1304e-04) (hash(x)=128985347)
947
+ 7870 train 3.704331 (lr=1.1253e-04) (hash(x)=152897363)
948
+ 7880 train 3.909804 (lr=1.1202e-04) (hash(x)=151460360)
949
+ 7890 train 3.731876 (lr=1.1152e-04) (hash(x)=137481738)
950
+ 7900 val loss 4.0155
951
+ 7900 val perplexity 55.4490
952
+ 7900 train 3.807155 (lr=1.1103e-04) (hash(x)=147363479)
953
+ 7910 train 4.030183 (lr=1.1054e-04) (hash(x)=151165744)
954
+ 7920 train 4.205436 (lr=1.1006e-04) (hash(x)=146214226)
955
+ 7930 train 4.011224 (lr=1.0958e-04) (hash(x)=154457502)
956
+ 7940 train 3.916266 (lr=1.0911e-04) (hash(x)=162025201)
957
+ 7950 train 3.912090 (lr=1.0865e-04) (hash(x)=156488145)
958
+ 7960 train 4.185153 (lr=1.0819e-04) (hash(x)=154011584)
959
+ 7970 train 4.035840 (lr=1.0773e-04) (hash(x)=145330320)
960
+ 7980 train 3.938541 (lr=1.0729e-04) (hash(x)=148216312)
961
+ 7990 train 3.880335 (lr=1.0684e-04) (hash(x)=156819091)
962
+ 8000 val loss 4.0059
963
+ 8000 val perplexity 54.9192
964
+ 8000 train 4.040503 (lr=1.0641e-04) (hash(x)=156122973)
965
+ 8010 train 3.984417 (lr=1.0597e-04) (hash(x)=146933088)
966
+ 8020 train 4.079651 (lr=1.0555e-04) (hash(x)=151650746)
967
+ 8030 train 3.999864 (lr=1.0513e-04) (hash(x)=140692373)
968
+ 8040 train 4.190343 (lr=1.0471e-04) (hash(x)=147850094)
969
+ 8050 train 4.012302 (lr=1.0430e-04) (hash(x)=144454109)
970
+ 8060 train 3.869228 (lr=1.0390e-04) (hash(x)=152205124)
971
+ 8070 train 4.085998 (lr=1.0350e-04) (hash(x)=148918066)
972
+ 8080 train 4.009272 (lr=1.0311e-04) (hash(x)=146481024)
973
+ 8090 train 4.051519 (lr=1.0272e-04) (hash(x)=168785549)
974
+ 8100 val loss 4.0033
975
+ 8100 val perplexity 54.7787
976
+ 8100 train 4.083695 (lr=1.0234e-04) (hash(x)=156153179)
977
+ 8110 train 4.030242 (lr=1.0197e-04) (hash(x)=150498601)
978
+ 8120 train 4.031703 (lr=1.0160e-04) (hash(x)=142002714)
979
+ 8130 train 3.926563 (lr=1.0124e-04) (hash(x)=138450483)
980
+ 8140 train 4.010139 (lr=1.0088e-04) (hash(x)=145306351)
981
+ 8150 train 3.881377 (lr=1.0053e-04) (hash(x)=172502865)
982
+ 8160 train 3.990213 (lr=1.0018e-04) (hash(x)=156689649)
983
+ 8170 train 3.908438 (lr=9.9838e-05) (hash(x)=143766932)
984
+ 8180 train 4.048414 (lr=9.9503e-05) (hash(x)=150361321)
985
+ 8190 train 3.875005 (lr=9.9174e-05) (hash(x)=152664241)
986
+ 8200 val loss 4.0021
987
+ 8200 val perplexity 54.7122
988
+ 8200 train 4.152622 (lr=9.8850e-05) (hash(x)=146430698)
989
+ 8210 train 3.842161 (lr=9.8532e-05) (hash(x)=167362078)
990
+ 8220 train 4.068886 (lr=9.8220e-05) (hash(x)=157650942)
991
+ 8230 train 3.873731 (lr=9.7914e-05) (hash(x)=142102044)
992
+ 8240 train 3.917400 (lr=9.7614e-05) (hash(x)=153405900)
993
+ 8250 train 3.863927 (lr=9.7319e-05) (hash(x)=140222909)
994
+ 8260 train 4.001134 (lr=9.7030e-05) (hash(x)=137981854)
995
+ 8270 train 3.913814 (lr=9.6747e-05) (hash(x)=147576272)
996
+ 8280 train 3.927060 (lr=9.6469e-05) (hash(x)=157988195)
997
+ 8290 train 3.974802 (lr=9.6198e-05) (hash(x)=145715681)
998
+ 8300 val loss 4.0034
999
+ 8300 val perplexity 54.7826
1000
+ 8300 train 3.925215 (lr=9.5932e-05) (hash(x)=143507257)
1001
+ 8310 train 3.985779 (lr=9.5672e-05) (hash(x)=148930064)
1002
+ 8320 train 3.906655 (lr=9.5417e-05) (hash(x)=144134805)
1003
+ 8330 train 4.031232 (lr=9.5169e-05) (hash(x)=148418079)
1004
+ 8340 train 3.751790 (lr=9.4926e-05) (hash(x)=166518807)
1005
+ 8350 train 4.060431 (lr=9.4689e-05) (hash(x)=140042036)
1006
+ 8360 train 3.933165 (lr=9.4458e-05) (hash(x)=151427074)
1007
+ 8370 train 4.014946 (lr=9.4233e-05) (hash(x)=158389185)
1008
+ 8380 train 3.919269 (lr=9.4013e-05) (hash(x)=148193308)
1009
+ 8390 train 3.989676 (lr=9.3800e-05) (hash(x)=144289008)
1010
+ 8400 val loss 3.9945
1011
+ 8400 val perplexity 54.3011
1012
+ 8400 train 3.704267 (lr=9.3592e-05) (hash(x)=166272643)
1013
+ 8410 train 3.911696 (lr=9.3390e-05) (hash(x)=162244433)
1014
+ 8420 train 3.952242 (lr=9.3194e-05) (hash(x)=149271299)
1015
+ 8430 train 4.091085 (lr=9.3003e-05) (hash(x)=154135338)
1016
+ 8440 train 4.138370 (lr=9.2819e-05) (hash(x)=149661838)
1017
+ 8450 train 3.976594 (lr=9.2640e-05) (hash(x)=142205056)
1018
+ 8460 train 3.910329 (lr=9.2467e-05) (hash(x)=166652398)
1019
+ 8470 train 4.079313 (lr=9.2300e-05) (hash(x)=149029710)
1020
+ 8480 train 3.855407 (lr=9.2139e-05) (hash(x)=154217415)
1021
+ 8490 train 3.925338 (lr=9.1983e-05) (hash(x)=151526557)
1022
+ 8500 val loss 3.9957
1023
+ 8500 val perplexity 54.3655
1024
+ 8500 train 3.919521 (lr=9.1834e-05) (hash(x)=143887848)
1025
+ 8510 train 3.907227 (lr=9.1690e-05) (hash(x)=152324645)
1026
+ 8520 train 3.638137 (lr=9.1552e-05) (hash(x)=145124959)
1027
+ 8530 train 3.964483 (lr=9.1420e-05) (hash(x)=186319340)
1028
+ 8540 train 3.973724 (lr=9.1294e-05) (hash(x)=147388339)
1029
+ 8550 train 4.004087 (lr=9.1174e-05) (hash(x)=150118928)
1030
+ 8560 train 4.011125 (lr=9.1060e-05) (hash(x)=153985286)
1031
+ 8570 train 3.950098 (lr=9.0951e-05) (hash(x)=149582799)
1032
+ 8580 train 3.952162 (lr=9.0848e-05) (hash(x)=156867870)
1033
+ 8590 train 3.845281 (lr=9.0751e-05) (hash(x)=148773298)
1034
+ 8600 val loss 3.9972
1035
+ 8600 val perplexity 54.4466
1036
+ 8600 train 3.936340 (lr=9.0661e-05) (hash(x)=156900341)
1037
+ 8610 train 4.095030 (lr=9.0575e-05) (hash(x)=156790089)
1038
+ 8620 train 3.901847 (lr=9.0496e-05) (hash(x)=154680805)
1039
+ 8630 train 3.939135 (lr=9.0423e-05) (hash(x)=138107873)
1040
+ 8640 train 4.026122 (lr=9.0355e-05) (hash(x)=142554243)
1041
+ 8650 train 3.834553 (lr=9.0294e-05) (hash(x)=134360956)
1042
+ 8660 train 3.914890 (lr=9.0238e-05) (hash(x)=128599610)
1043
+ 8670 train 3.942565 (lr=9.0188e-05) (hash(x)=159628945)
1044
+ 8680 train 3.968983 (lr=9.0144e-05) (hash(x)=143161532)
1045
+ 8690 train 3.997907 (lr=9.0106e-05) (hash(x)=140104912)
1046
+ 8700 val loss 3.9895
1047
+ 8700 val perplexity 54.0304
1048
+ 8700 train 4.176856 (lr=9.0073e-05) (hash(x)=146417632)
1049
+ 8710 train 4.151796 (lr=9.0047e-05) (hash(x)=153671695)
1050
+ 8720 train 4.026544 (lr=9.0026e-05) (hash(x)=141680750)
1051
+ 8730 train 3.897094 (lr=9.0012e-05) (hash(x)=147279231)
1052
+ 8740 train 4.189255 (lr=9.0003e-05) (hash(x)=129729371)
1053
+ 8749 val loss 3.9895
1054
+ 8749 val perplexity 54.0295
lr9e-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 sha256:97ddbaad6a622ff8d95e555cc585fca56fa73284dae34f5f46179fcbabf67e8f
3
+ size 92843394
lr9e-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 sha256:28afecedf70868412fdb623dc4e98070e1ebbb4d3eeac7e0b3ba7d635930c5f6
3
+ size 179406214