andrew-healey commited on
Commit
5dd274d
·
verified ·
1 Parent(s): 59f2492

Upload folder using huggingface_hub

Browse files
lr12e-4_total_batch_size61440_baseline_seed1339/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_11/lr12e-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": 0.0012, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "12e-4_61440", "n_embd": 256}
lr12e-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
lr12e-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=2.4000e-06) (hash(x)=150724848)
5
+ 10 train 10.109632 (lr=2.6400e-05) (hash(x)=149663361)
6
+ 20 train 9.709885 (lr=5.0400e-05) (hash(x)=153483789)
7
+ 30 train 9.289991 (lr=7.4400e-05) (hash(x)=145733176)
8
+ 40 train 8.608999 (lr=9.8400e-05) (hash(x)=145287265)
9
+ 50 train 8.093102 (lr=1.2240e-04) (hash(x)=149217061)
10
+ 60 train 7.747657 (lr=1.4640e-04) (hash(x)=150157569)
11
+ 70 train 7.600236 (lr=1.7040e-04) (hash(x)=140292841)
12
+ 80 train 7.656566 (lr=1.9440e-04) (hash(x)=152157598)
13
+ 90 train 7.434501 (lr=2.1840e-04) (hash(x)=149756968)
14
+ 100 val loss 7.3974
15
+ 100 val perplexity 1631.7308
16
+ 100 train 7.354972 (lr=2.4240e-04) (hash(x)=149910534)
17
+ 110 train 7.257179 (lr=2.6640e-04) (hash(x)=159020328)
18
+ 120 train 7.189618 (lr=2.9040e-04) (hash(x)=150731273)
19
+ 130 train 6.996826 (lr=3.1440e-04) (hash(x)=156727355)
20
+ 140 train 7.020205 (lr=3.3840e-04) (hash(x)=166473281)
21
+ 150 train 6.711750 (lr=3.6240e-04) (hash(x)=142185643)
22
+ 160 train 6.706507 (lr=3.8640e-04) (hash(x)=151371145)
23
+ 170 train 6.752350 (lr=4.1040e-04) (hash(x)=151601659)
24
+ 180 train 6.592309 (lr=4.3440e-04) (hash(x)=153368917)
25
+ 190 train 6.441749 (lr=4.5840e-04) (hash(x)=145008196)
26
+ 200 val loss 6.5726
27
+ 200 val perplexity 715.2498
28
+ 200 train 6.460454 (lr=4.8240e-04) (hash(x)=148123706)
29
+ 210 train 6.363390 (lr=5.0640e-04) (hash(x)=145714762)
30
+ 220 train 6.355837 (lr=5.3040e-04) (hash(x)=150622583)
31
+ 230 train 6.135765 (lr=5.5440e-04) (hash(x)=146512648)
32
+ 240 train 6.178216 (lr=5.7840e-04) (hash(x)=151535384)
33
+ 250 train 6.120986 (lr=6.0240e-04) (hash(x)=149952383)
34
+ 260 train 6.012393 (lr=6.2640e-04) (hash(x)=148726088)
35
+ 270 train 5.977851 (lr=6.5040e-04) (hash(x)=152776695)
36
+ 280 train 6.004210 (lr=6.7440e-04) (hash(x)=157260375)
37
+ 290 train 6.057351 (lr=6.9840e-04) (hash(x)=151559106)
38
+ 300 val loss 6.1380
39
+ 300 val perplexity 463.1147
40
+ 300 train 5.857712 (lr=7.2240e-04) (hash(x)=146678221)
41
+ 310 train 5.971629 (lr=7.4640e-04) (hash(x)=156713527)
42
+ 320 train 6.011353 (lr=7.7040e-04) (hash(x)=144947971)
43
+ 330 train 6.174514 (lr=7.9440e-04) (hash(x)=158684944)
44
+ 340 train 6.044989 (lr=8.1840e-04) (hash(x)=162318756)
45
+ 350 train 5.854471 (lr=8.4240e-04) (hash(x)=156180736)
46
+ 360 train 6.023965 (lr=8.6640e-04) (hash(x)=178373797)
47
+ 370 train 5.937894 (lr=8.9040e-04) (hash(x)=154389449)
48
+ 380 train 5.779488 (lr=9.1440e-04) (hash(x)=150369080)
49
+ 390 train 5.913444 (lr=9.3840e-04) (hash(x)=159840654)
50
+ 400 val loss 5.7977
51
+ 400 val perplexity 329.5352
52
+ 400 train 5.719372 (lr=9.6240e-04) (hash(x)=151700982)
53
+ 410 train 5.689095 (lr=9.8640e-04) (hash(x)=160467274)
54
+ 420 train 5.607277 (lr=1.0104e-03) (hash(x)=148930522)
55
+ 430 train 5.817911 (lr=1.0344e-03) (hash(x)=154565303)
56
+ 440 train 5.822478 (lr=1.0584e-03) (hash(x)=148079227)
57
+ 450 train 5.507468 (lr=1.0824e-03) (hash(x)=147288467)
58
+ 460 train 5.548209 (lr=1.1064e-03) (hash(x)=145272617)
59
+ 470 train 5.571751 (lr=1.1304e-03) (hash(x)=123126729)
60
+ 480 train 5.508143 (lr=1.1544e-03) (hash(x)=156287490)
61
+ 490 train 5.629976 (lr=1.1784e-03) (hash(x)=161700145)
62
+ 500 val loss 5.5574
63
+ 500 val perplexity 259.1558
64
+ 500 train 5.666543 (lr=1.2000e-03) (hash(x)=156182087)
65
+ 510 train 5.285265 (lr=1.2000e-03) (hash(x)=131674504)
66
+ 520 train 5.604834 (lr=1.2000e-03) (hash(x)=162758254)
67
+ 530 train 5.407594 (lr=1.2000e-03) (hash(x)=133510409)
68
+ 540 train 5.392573 (lr=1.1999e-03) (hash(x)=157363914)
69
+ 550 train 5.324354 (lr=1.1999e-03) (hash(x)=147861550)
70
+ 560 train 5.277119 (lr=1.1999e-03) (hash(x)=149449280)
71
+ 570 train 5.295470 (lr=1.1998e-03) (hash(x)=143899904)
72
+ 580 train 5.257513 (lr=1.1997e-03) (hash(x)=149266312)
73
+ 590 train 5.505569 (lr=1.1997e-03) (hash(x)=164352774)
74
+ 600 val loss 5.3971
75
+ 600 val perplexity 220.7719
76
+ 600 train 5.256671 (lr=1.1996e-03) (hash(x)=149318660)
77
+ 610 train 5.145845 (lr=1.1995e-03) (hash(x)=146101781)
78
+ 620 train 5.114022 (lr=1.1994e-03) (hash(x)=137184623)
79
+ 630 train 4.987963 (lr=1.1993e-03) (hash(x)=145791873)
80
+ 640 train 5.009591 (lr=1.1992e-03) (hash(x)=143999212)
81
+ 650 train 5.109940 (lr=1.1991e-03) (hash(x)=151631103)
82
+ 660 train 4.879338 (lr=1.1990e-03) (hash(x)=138360108)
83
+ 670 train 5.270453 (lr=1.1989e-03) (hash(x)=156637679)
84
+ 680 train 4.908310 (lr=1.1987e-03) (hash(x)=143752601)
85
+ 690 train 5.048779 (lr=1.1986e-03) (hash(x)=158368903)
86
+ 700 val loss 5.2686
87
+ 700 val perplexity 194.1476
88
+ 700 train 5.251122 (lr=1.1984e-03) (hash(x)=150482428)
89
+ 710 train 5.199285 (lr=1.1983e-03) (hash(x)=158085060)
90
+ 720 train 5.280901 (lr=1.1981e-03) (hash(x)=159461069)
91
+ 730 train 5.112783 (lr=1.1979e-03) (hash(x)=148795562)
92
+ 740 train 5.135463 (lr=1.1977e-03) (hash(x)=149967984)
93
+ 750 train 5.117489 (lr=1.1976e-03) (hash(x)=144458617)
94
+ 760 train 5.147257 (lr=1.1974e-03) (hash(x)=155106202)
95
+ 770 train 4.936299 (lr=1.1971e-03) (hash(x)=144506191)
96
+ 780 train 5.079663 (lr=1.1969e-03) (hash(x)=155825158)
97
+ 790 train 5.123920 (lr=1.1967e-03) (hash(x)=158835381)
98
+ 800 val loss 5.0815
99
+ 800 val perplexity 161.0087
100
+ 800 train 5.047852 (lr=1.1965e-03) (hash(x)=143268605)
101
+ 810 train 4.973405 (lr=1.1962e-03) (hash(x)=139072755)
102
+ 820 train 5.079490 (lr=1.1960e-03) (hash(x)=179384792)
103
+ 830 train 4.870527 (lr=1.1957e-03) (hash(x)=151086962)
104
+ 840 train 5.493161 (lr=1.1955e-03) (hash(x)=142711393)
105
+ 850 train 4.969683 (lr=1.1952e-03) (hash(x)=150817011)
106
+ 860 train 4.913413 (lr=1.1949e-03) (hash(x)=138495801)
107
+ 870 train 4.899954 (lr=1.1946e-03) (hash(x)=155474016)
108
+ 880 train 4.931062 (lr=1.1944e-03) (hash(x)=151951501)
109
+ 890 train 4.796121 (lr=1.1941e-03) (hash(x)=143699524)
110
+ 900 val loss 4.9499
111
+ 900 val perplexity 141.1655
112
+ 900 train 4.935257 (lr=1.1937e-03) (hash(x)=152322423)
113
+ 910 train 4.888164 (lr=1.1934e-03) (hash(x)=151010509)
114
+ 920 train 4.685431 (lr=1.1931e-03) (hash(x)=148720342)
115
+ 930 train 4.651957 (lr=1.1928e-03) (hash(x)=125869480)
116
+ 940 train 4.880496 (lr=1.1924e-03) (hash(x)=150349503)
117
+ 950 train 4.661515 (lr=1.1921e-03) (hash(x)=143574126)
118
+ 960 train 4.864492 (lr=1.1917e-03) (hash(x)=146292562)
119
+ 970 train 4.782253 (lr=1.1914e-03) (hash(x)=162256596)
120
+ 980 train 4.783382 (lr=1.1910e-03) (hash(x)=137640259)
121
+ 990 train 4.613474 (lr=1.1906e-03) (hash(x)=141767794)
122
+ 1000 val loss 4.9009
123
+ 1000 val perplexity 134.4109
124
+ 1000 train 4.844637 (lr=1.1902e-03) (hash(x)=147904298)
125
+ 1010 train 4.787038 (lr=1.1898e-03) (hash(x)=162239186)
126
+ 1020 train 4.543619 (lr=1.1894e-03) (hash(x)=140706537)
127
+ 1030 train 4.664927 (lr=1.1890e-03) (hash(x)=155556720)
128
+ 1040 train 4.554864 (lr=1.1886e-03) (hash(x)=144746445)
129
+ 1050 train 4.600341 (lr=1.1882e-03) (hash(x)=144719400)
130
+ 1060 train 4.471046 (lr=1.1878e-03) (hash(x)=153065665)
131
+ 1070 train 4.586587 (lr=1.1873e-03) (hash(x)=136468722)
132
+ 1080 train 4.575816 (lr=1.1869e-03) (hash(x)=157674756)
133
+ 1090 train 4.764370 (lr=1.1864e-03) (hash(x)=155882884)
134
+ 1100 val loss 4.8249
135
+ 1100 val perplexity 124.5698
136
+ 1100 train 5.031359 (lr=1.1860e-03) (hash(x)=154343147)
137
+ 1110 train 4.835208 (lr=1.1855e-03) (hash(x)=158582939)
138
+ 1120 train 4.629939 (lr=1.1850e-03) (hash(x)=162479784)
139
+ 1130 train 4.825120 (lr=1.1845e-03) (hash(x)=155704345)
140
+ 1140 train 4.723613 (lr=1.1840e-03) (hash(x)=146934299)
141
+ 1150 train 4.664534 (lr=1.1835e-03) (hash(x)=154372635)
142
+ 1160 train 4.663521 (lr=1.1830e-03) (hash(x)=145536832)
143
+ 1170 train 4.951137 (lr=1.1825e-03) (hash(x)=158785901)
144
+ 1180 train 4.740629 (lr=1.1820e-03) (hash(x)=158271614)
145
+ 1190 train 4.786156 (lr=1.1815e-03) (hash(x)=157365589)
146
+ 1200 val loss 4.7733
147
+ 1200 val perplexity 118.3068
148
+ 1200 train 4.840771 (lr=1.1809e-03) (hash(x)=141843115)
149
+ 1210 train 4.714913 (lr=1.1804e-03) (hash(x)=146807973)
150
+ 1220 train 4.695413 (lr=1.1798e-03) (hash(x)=144149947)
151
+ 1230 train 4.630150 (lr=1.1793e-03) (hash(x)=147869063)
152
+ 1240 train 4.725526 (lr=1.1787e-03) (hash(x)=152217389)
153
+ 1250 train 4.719992 (lr=1.1781e-03) (hash(x)=149417679)
154
+ 1260 train 4.747976 (lr=1.1775e-03) (hash(x)=144029489)
155
+ 1270 train 4.870682 (lr=1.1770e-03) (hash(x)=154328303)
156
+ 1280 train 4.730612 (lr=1.1764e-03) (hash(x)=151625838)
157
+ 1290 train 4.588966 (lr=1.1757e-03) (hash(x)=140270541)
158
+ 1300 val loss 4.7127
159
+ 1300 val perplexity 111.3527
160
+ 1300 train 4.641655 (lr=1.1751e-03) (hash(x)=145279030)
161
+ 1310 train 4.685517 (lr=1.1745e-03) (hash(x)=140279630)
162
+ 1320 train 4.405325 (lr=1.1739e-03) (hash(x)=133642994)
163
+ 1330 train 4.569730 (lr=1.1733e-03) (hash(x)=146634776)
164
+ 1340 train 4.665359 (lr=1.1726e-03) (hash(x)=155919973)
165
+ 1350 train 4.704120 (lr=1.1720e-03) (hash(x)=146363911)
166
+ 1360 train 4.579832 (lr=1.1713e-03) (hash(x)=150613636)
167
+ 1370 train 4.573457 (lr=1.1706e-03) (hash(x)=148536643)
168
+ 1380 train 4.578975 (lr=1.1700e-03) (hash(x)=152803906)
169
+ 1390 train 4.496904 (lr=1.1693e-03) (hash(x)=150383305)
170
+ 1400 val loss 4.6996
171
+ 1400 val perplexity 109.9021
172
+ 1400 train 4.599776 (lr=1.1686e-03) (hash(x)=152507639)
173
+ 1410 train 4.363600 (lr=1.1679e-03) (hash(x)=161841408)
174
+ 1420 train 4.483188 (lr=1.1672e-03) (hash(x)=167824828)
175
+ 1430 train 4.437515 (lr=1.1665e-03) (hash(x)=147696374)
176
+ 1440 train 4.392510 (lr=1.1658e-03) (hash(x)=150085494)
177
+ 1450 train 4.346525 (lr=1.1650e-03) (hash(x)=139296191)
178
+ 1460 train 4.274680 (lr=1.1643e-03) (hash(x)=142956546)
179
+ 1470 train 4.337068 (lr=1.1636e-03) (hash(x)=155908628)
180
+ 1480 train 4.329272 (lr=1.1628e-03) (hash(x)=150893345)
181
+ 1490 train 4.349893 (lr=1.1621e-03) (hash(x)=143548902)
182
+ 1500 val loss 4.6386
183
+ 1500 val perplexity 103.3981
184
+ 1500 train 4.595102 (lr=1.1613e-03) (hash(x)=148473774)
185
+ 1510 train 4.674191 (lr=1.1606e-03) (hash(x)=161631000)
186
+ 1520 train 4.606560 (lr=1.1598e-03) (hash(x)=166217855)
187
+ 1530 train 4.563714 (lr=1.1590e-03) (hash(x)=149440046)
188
+ 1540 train 4.511814 (lr=1.1582e-03) (hash(x)=155013278)
189
+ 1550 train 4.549587 (lr=1.1574e-03) (hash(x)=154130527)
190
+ 1560 train 4.566741 (lr=1.1566e-03) (hash(x)=152138735)
191
+ 1570 train 4.473716 (lr=1.1558e-03) (hash(x)=148128305)
192
+ 1580 train 4.616622 (lr=1.1550e-03) (hash(x)=157176546)
193
+ 1590 train 4.737245 (lr=1.1541e-03) (hash(x)=152935467)
194
+ 1600 val loss 4.5777
195
+ 1600 val perplexity 97.2913
196
+ 1600 train 4.548295 (lr=1.1533e-03) (hash(x)=151117002)
197
+ 1610 train 4.544728 (lr=1.1525e-03) (hash(x)=149883384)
198
+ 1620 train 4.600081 (lr=1.1516e-03) (hash(x)=145015354)
199
+ 1630 train 4.519558 (lr=1.1508e-03) (hash(x)=146717645)
200
+ 1640 train 4.725262 (lr=1.1499e-03) (hash(x)=156372727)
201
+ 1650 train 4.643160 (lr=1.1490e-03) (hash(x)=140261555)
202
+ 1660 train 4.477713 (lr=1.1482e-03) (hash(x)=150088650)
203
+ 1670 train 4.396879 (lr=1.1473e-03) (hash(x)=135943513)
204
+ 1680 train 4.620824 (lr=1.1464e-03) (hash(x)=153529653)
205
+ 1690 train 4.680629 (lr=1.1455e-03) (hash(x)=148930360)
206
+ 1700 val loss 4.5870
207
+ 1700 val perplexity 98.1946
208
+ 1700 train 4.558865 (lr=1.1446e-03) (hash(x)=138011335)
209
+ 1710 train 4.517448 (lr=1.1437e-03) (hash(x)=141383599)
210
+ 1720 train 4.709752 (lr=1.1428e-03) (hash(x)=161449652)
211
+ 1730 train 4.487815 (lr=1.1418e-03) (hash(x)=153923561)
212
+ 1740 train 4.427644 (lr=1.1409e-03) (hash(x)=146043253)
213
+ 1750 train 4.456100 (lr=1.1400e-03) (hash(x)=155720883)
214
+ 1760 train 4.449072 (lr=1.1390e-03) (hash(x)=148222148)
215
+ 1770 train 4.506100 (lr=1.1381e-03) (hash(x)=155712380)
216
+ 1780 train 4.399938 (lr=1.1371e-03) (hash(x)=149767859)
217
+ 1790 train 4.637546 (lr=1.1361e-03) (hash(x)=167214230)
218
+ 1800 val loss 4.5498
219
+ 1800 val perplexity 94.6110
220
+ 1800 train 4.531946 (lr=1.1352e-03) (hash(x)=171180926)
221
+ 1810 train 4.552355 (lr=1.1342e-03) (hash(x)=162375834)
222
+ 1820 train 4.670647 (lr=1.1332e-03) (hash(x)=151495752)
223
+ 1830 train 4.566253 (lr=1.1322e-03) (hash(x)=146532257)
224
+ 1840 train 4.653934 (lr=1.1312e-03) (hash(x)=145694434)
225
+ 1850 train 4.534233 (lr=1.1302e-03) (hash(x)=144823488)
226
+ 1860 train 4.631394 (lr=1.1292e-03) (hash(x)=149836286)
227
+ 1870 train 4.600561 (lr=1.1282e-03) (hash(x)=143935049)
228
+ 1880 train 4.627179 (lr=1.1271e-03) (hash(x)=151786371)
229
+ 1890 train 4.518963 (lr=1.1261e-03) (hash(x)=152892567)
230
+ 1900 val loss 4.5089
231
+ 1900 val perplexity 90.8237
232
+ 1900 train 4.557005 (lr=1.1251e-03) (hash(x)=141769419)
233
+ 1910 train 4.574733 (lr=1.1240e-03) (hash(x)=156914416)
234
+ 1920 train 4.436125 (lr=1.1230e-03) (hash(x)=151573900)
235
+ 1930 train 4.523968 (lr=1.1219e-03) (hash(x)=145121701)
236
+ 1940 train 4.466547 (lr=1.1208e-03) (hash(x)=150896348)
237
+ 1950 train 4.572210 (lr=1.1198e-03) (hash(x)=168797189)
238
+ 1960 train 4.449547 (lr=1.1187e-03) (hash(x)=146711430)
239
+ 1970 train 4.306370 (lr=1.1176e-03) (hash(x)=152180946)
240
+ 1980 train 4.426885 (lr=1.1165e-03) (hash(x)=155542268)
241
+ 1990 train 4.392109 (lr=1.1154e-03) (hash(x)=144668660)
242
+ 2000 val loss 4.4779
243
+ 2000 val perplexity 88.0518
244
+ 2000 train 4.466370 (lr=1.1143e-03) (hash(x)=151963443)
245
+ 2010 train 4.462001 (lr=1.1132e-03) (hash(x)=150810352)
246
+ 2020 train 4.461997 (lr=1.1120e-03) (hash(x)=157539832)
247
+ 2030 train 4.676983 (lr=1.1109e-03) (hash(x)=137858786)
248
+ 2040 train 4.420146 (lr=1.1098e-03) (hash(x)=154270900)
249
+ 2050 train 4.273467 (lr=1.1086e-03) (hash(x)=156252579)
250
+ 2060 train 4.212379 (lr=1.1075e-03) (hash(x)=143285549)
251
+ 2070 train 4.424053 (lr=1.1063e-03) (hash(x)=150542438)
252
+ 2080 train 4.334618 (lr=1.1052e-03) (hash(x)=146290330)
253
+ 2090 train 4.433721 (lr=1.1040e-03) (hash(x)=146486756)
254
+ 2100 val loss 4.5092
255
+ 2100 val perplexity 90.8489
256
+ 2100 train 4.448768 (lr=1.1028e-03) (hash(x)=162947470)
257
+ 2110 train 4.345153 (lr=1.1017e-03) (hash(x)=150305391)
258
+ 2120 train 4.498278 (lr=1.1005e-03) (hash(x)=156452783)
259
+ 2130 train 4.473768 (lr=1.0993e-03) (hash(x)=167481875)
260
+ 2140 train 4.448520 (lr=1.0981e-03) (hash(x)=141782256)
261
+ 2150 train 4.480247 (lr=1.0969e-03) (hash(x)=145035419)
262
+ 2160 train 4.528182 (lr=1.0957e-03) (hash(x)=142000400)
263
+ 2170 train 4.494695 (lr=1.0944e-03) (hash(x)=144229380)
264
+ 2180 train 4.536742 (lr=1.0932e-03) (hash(x)=166204176)
265
+ 2190 train 4.459890 (lr=1.0920e-03) (hash(x)=150712057)
266
+ 2200 val loss 4.4380
267
+ 2200 val perplexity 84.6061
268
+ 2200 train 4.346491 (lr=1.0907e-03) (hash(x)=154954810)
269
+ 2210 train 4.559747 (lr=1.0895e-03) (hash(x)=155668760)
270
+ 2220 train 4.390372 (lr=1.0883e-03) (hash(x)=144834708)
271
+ 2230 train 4.371741 (lr=1.0870e-03) (hash(x)=159744566)
272
+ 2240 train 4.335001 (lr=1.0857e-03) (hash(x)=158543377)
273
+ 2250 train 4.348533 (lr=1.0845e-03) (hash(x)=151004281)
274
+ 2260 train 4.449752 (lr=1.0832e-03) (hash(x)=151242482)
275
+ 2270 train 4.400414 (lr=1.0819e-03) (hash(x)=152056559)
276
+ 2280 train 4.334636 (lr=1.0806e-03) (hash(x)=148864076)
277
+ 2290 train 4.459346 (lr=1.0793e-03) (hash(x)=144294586)
278
+ 2300 val loss 4.4318
279
+ 2300 val perplexity 84.0845
280
+ 2300 train 4.363368 (lr=1.0780e-03) (hash(x)=151878111)
281
+ 2310 train 4.375549 (lr=1.0767e-03) (hash(x)=154139457)
282
+ 2320 train 4.309752 (lr=1.0754e-03) (hash(x)=155571592)
283
+ 2330 train 4.348783 (lr=1.0741e-03) (hash(x)=154823764)
284
+ 2340 train 4.400339 (lr=1.0728e-03) (hash(x)=153252177)
285
+ 2350 train 4.342879 (lr=1.0715e-03) (hash(x)=136051830)
286
+ 2360 train 4.392274 (lr=1.0701e-03) (hash(x)=153814175)
287
+ 2370 train 4.379485 (lr=1.0688e-03) (hash(x)=146391078)
288
+ 2380 train 4.345505 (lr=1.0674e-03) (hash(x)=150782698)
289
+ 2390 train 4.414737 (lr=1.0661e-03) (hash(x)=142752256)
290
+ 2400 val loss 4.4448
291
+ 2400 val perplexity 85.1864
292
+ 2400 train 4.322749 (lr=1.0647e-03) (hash(x)=158661057)
293
+ 2410 train 4.397375 (lr=1.0634e-03) (hash(x)=153611303)
294
+ 2420 train 4.566748 (lr=1.0620e-03) (hash(x)=159508684)
295
+ 2430 train 4.361979 (lr=1.0606e-03) (hash(x)=149501192)
296
+ 2440 train 4.033803 (lr=1.0592e-03) (hash(x)=181177177)
297
+ 2450 train 4.485989 (lr=1.0578e-03) (hash(x)=163664732)
298
+ 2460 train 4.174714 (lr=1.0564e-03) (hash(x)=194527955)
299
+ 2470 train 4.409694 (lr=1.0550e-03) (hash(x)=158088521)
300
+ 2480 train 4.369109 (lr=1.0536e-03) (hash(x)=152865678)
301
+ 2490 train 4.349931 (lr=1.0522e-03) (hash(x)=165703871)
302
+ 2500 val loss 4.3863
303
+ 2500 val perplexity 80.3437
304
+ 2500 train 4.436466 (lr=1.0508e-03) (hash(x)=150925584)
305
+ 2510 train 4.343757 (lr=1.0494e-03) (hash(x)=144406177)
306
+ 2520 train 4.243018 (lr=1.0480e-03) (hash(x)=120133043)
307
+ 2530 train 4.433231 (lr=1.0465e-03) (hash(x)=167900247)
308
+ 2540 train 4.324368 (lr=1.0451e-03) (hash(x)=162599203)
309
+ 2550 train 4.388067 (lr=1.0437e-03) (hash(x)=149328180)
310
+ 2560 train 4.274185 (lr=1.0422e-03) (hash(x)=150049968)
311
+ 2570 train 4.417150 (lr=1.0407e-03) (hash(x)=162876159)
312
+ 2580 train 4.244152 (lr=1.0393e-03) (hash(x)=148260649)
313
+ 2590 train 4.381494 (lr=1.0378e-03) (hash(x)=148998018)
314
+ 2600 val loss 4.3784
315
+ 2600 val perplexity 79.7115
316
+ 2600 train 4.293382 (lr=1.0363e-03) (hash(x)=144515755)
317
+ 2610 train 4.411130 (lr=1.0349e-03) (hash(x)=154455224)
318
+ 2620 train 4.214087 (lr=1.0334e-03) (hash(x)=142920048)
319
+ 2630 train 4.380084 (lr=1.0319e-03) (hash(x)=143931236)
320
+ 2640 train 4.224974 (lr=1.0304e-03) (hash(x)=150983323)
321
+ 2650 train 4.325104 (lr=1.0289e-03) (hash(x)=139360405)
322
+ 2660 train 4.244003 (lr=1.0274e-03) (hash(x)=148850470)
323
+ 2670 train 4.492133 (lr=1.0259e-03) (hash(x)=149929606)
324
+ 2680 train 4.468233 (lr=1.0244e-03) (hash(x)=151673000)
325
+ 2690 train 4.341923 (lr=1.0229e-03) (hash(x)=140631266)
326
+ 2700 val loss 4.3766
327
+ 2700 val perplexity 79.5692
328
+ 2700 train 4.313663 (lr=1.0213e-03) (hash(x)=153109144)
329
+ 2710 train 4.200322 (lr=1.0198e-03) (hash(x)=135332836)
330
+ 2720 train 4.552682 (lr=1.0183e-03) (hash(x)=131611182)
331
+ 2730 train 4.386250 (lr=1.0167e-03) (hash(x)=142909388)
332
+ 2740 train 4.422916 (lr=1.0152e-03) (hash(x)=145593419)
333
+ 2750 train 4.312067 (lr=1.0136e-03) (hash(x)=158225769)
334
+ 2760 train 4.279745 (lr=1.0121e-03) (hash(x)=170509938)
335
+ 2770 train 4.293461 (lr=1.0105e-03) (hash(x)=150785805)
336
+ 2780 train 4.237528 (lr=1.0089e-03) (hash(x)=167281736)
337
+ 2790 train 4.430073 (lr=1.0074e-03) (hash(x)=161660187)
338
+ 2800 val loss 4.3471
339
+ 2800 val perplexity 77.2512
340
+ 2800 train 4.388263 (lr=1.0058e-03) (hash(x)=151152897)
341
+ 2810 train 4.344264 (lr=1.0042e-03) (hash(x)=143172067)
342
+ 2820 train 4.355720 (lr=1.0026e-03) (hash(x)=141569169)
343
+ 2830 train 4.234664 (lr=1.0010e-03) (hash(x)=159709681)
344
+ 2840 train 4.295024 (lr=9.9943e-04) (hash(x)=162552294)
345
+ 2850 train 4.427085 (lr=9.9783e-04) (hash(x)=146762690)
346
+ 2860 train 4.296045 (lr=9.9622e-04) (hash(x)=145729422)
347
+ 2870 train 4.376500 (lr=9.9461e-04) (hash(x)=144360628)
348
+ 2880 train 4.183399 (lr=9.9300e-04) (hash(x)=147737098)
349
+ 2890 train 4.291469 (lr=9.9138e-04) (hash(x)=146849057)
350
+ 2900 val loss 4.3350
351
+ 2900 val perplexity 76.3279
352
+ 2900 train 4.358004 (lr=9.8975e-04) (hash(x)=145800210)
353
+ 2910 train 4.164371 (lr=9.8812e-04) (hash(x)=147316313)
354
+ 2920 train 4.210795 (lr=9.8648e-04) (hash(x)=152595415)
355
+ 2930 train 4.374139 (lr=9.8484e-04) (hash(x)=156542647)
356
+ 2940 train 4.236506 (lr=9.8320e-04) (hash(x)=141183896)
357
+ 2950 train 4.333578 (lr=9.8155e-04) (hash(x)=156329036)
358
+ 2960 train 4.375101 (lr=9.7989e-04) (hash(x)=156717793)
359
+ 2970 train 4.239049 (lr=9.7824e-04) (hash(x)=143949931)
360
+ 2980 train 4.295738 (lr=9.7657e-04) (hash(x)=155067290)
361
+ 2990 train 4.267017 (lr=9.7490e-04) (hash(x)=148774729)
362
+ 3000 val loss 4.3519
363
+ 3000 val perplexity 77.6256
364
+ 3000 train 4.255648 (lr=9.7323e-04) (hash(x)=141997485)
365
+ 3010 train 4.228545 (lr=9.7155e-04) (hash(x)=160762953)
366
+ 3020 train 4.175551 (lr=9.6987e-04) (hash(x)=170341665)
367
+ 3030 train 4.452041 (lr=9.6819e-04) (hash(x)=150722836)
368
+ 3040 train 4.244356 (lr=9.6649e-04) (hash(x)=154243307)
369
+ 3050 train 4.297995 (lr=9.6480e-04) (hash(x)=146077234)
370
+ 3060 train 4.248845 (lr=9.6310e-04) (hash(x)=152125542)
371
+ 3070 train 4.428735 (lr=9.6140e-04) (hash(x)=144149387)
372
+ 3080 train 4.257467 (lr=9.5969e-04) (hash(x)=147508712)
373
+ 3090 train 4.255358 (lr=9.5797e-04) (hash(x)=143583369)
374
+ 3100 val loss 4.3078
375
+ 3100 val perplexity 74.2732
376
+ 3100 train 4.283609 (lr=9.5626e-04) (hash(x)=154049740)
377
+ 3110 train 4.335011 (lr=9.5454e-04) (hash(x)=147054835)
378
+ 3120 train 4.323867 (lr=9.5281e-04) (hash(x)=152468568)
379
+ 3130 train 4.279954 (lr=9.5108e-04) (hash(x)=151635194)
380
+ 3140 train 4.241763 (lr=9.4935e-04) (hash(x)=150396274)
381
+ 3150 train 4.288492 (lr=9.4761e-04) (hash(x)=141306267)
382
+ 3160 train 4.236377 (lr=9.4587e-04) (hash(x)=151110762)
383
+ 3170 train 4.313979 (lr=9.4412e-04) (hash(x)=146518875)
384
+ 3180 train 4.195427 (lr=9.4237e-04) (hash(x)=139914401)
385
+ 3190 train 4.204344 (lr=9.4061e-04) (hash(x)=161631880)
386
+ 3200 val loss 4.3195
387
+ 3200 val perplexity 75.1518
388
+ 3200 train 4.231094 (lr=9.3885e-04) (hash(x)=150471842)
389
+ 3210 train 4.303976 (lr=9.3709e-04) (hash(x)=147355138)
390
+ 3220 train 4.098399 (lr=9.3532e-04) (hash(x)=138879678)
391
+ 3230 train 4.196038 (lr=9.3355e-04) (hash(x)=123929679)
392
+ 3240 train 4.264080 (lr=9.3178e-04) (hash(x)=157263734)
393
+ 3250 train 4.274297 (lr=9.3000e-04) (hash(x)=148065539)
394
+ 3260 train 4.195098 (lr=9.2822e-04) (hash(x)=138020742)
395
+ 3270 train 4.059672 (lr=9.2643e-04) (hash(x)=142053611)
396
+ 3280 train 4.220358 (lr=9.2464e-04) (hash(x)=155444784)
397
+ 3290 train 4.048244 (lr=9.2285e-04) (hash(x)=148500315)
398
+ 3300 val loss 4.3151
399
+ 3300 val perplexity 74.8188
400
+ 3300 train 4.274778 (lr=9.2105e-04) (hash(x)=149048126)
401
+ 3310 train 4.231986 (lr=9.1925e-04) (hash(x)=141680674)
402
+ 3320 train 4.168321 (lr=9.1744e-04) (hash(x)=143310642)
403
+ 3330 train 4.281257 (lr=9.1563e-04) (hash(x)=149891496)
404
+ 3340 train 4.232554 (lr=9.1382e-04) (hash(x)=151244609)
405
+ 3350 train 4.239318 (lr=9.1200e-04) (hash(x)=157158479)
406
+ 3360 train 4.099291 (lr=9.1018e-04) (hash(x)=151709876)
407
+ 3370 train 4.253227 (lr=9.0836e-04) (hash(x)=145297090)
408
+ 3380 train 4.265821 (lr=9.0653e-04) (hash(x)=154767881)
409
+ 3390 train 4.239752 (lr=9.0470e-04) (hash(x)=152600366)
410
+ 3400 val loss 4.2891
411
+ 3400 val perplexity 72.9035
412
+ 3400 train 4.469651 (lr=9.0286e-04) (hash(x)=161261339)
413
+ 3410 train 4.236227 (lr=9.0102e-04) (hash(x)=146859162)
414
+ 3420 train 4.324354 (lr=8.9918e-04) (hash(x)=149276224)
415
+ 3430 train 4.342439 (lr=8.9734e-04) (hash(x)=155866933)
416
+ 3440 train 4.208660 (lr=8.9549e-04) (hash(x)=143191904)
417
+ 3450 train 4.422591 (lr=8.9364e-04) (hash(x)=156910661)
418
+ 3460 train 4.216209 (lr=8.9178e-04) (hash(x)=140850879)
419
+ 3470 train 4.429345 (lr=8.8992e-04) (hash(x)=162140445)
420
+ 3480 train 4.175794 (lr=8.8806e-04) (hash(x)=145607495)
421
+ 3490 train 4.197517 (lr=8.8619e-04) (hash(x)=152496121)
422
+ 3500 val loss 4.2855
423
+ 3500 val perplexity 72.6393
424
+ 3500 train 4.159913 (lr=8.8432e-04) (hash(x)=157495564)
425
+ 3510 train 4.277530 (lr=8.8245e-04) (hash(x)=147657004)
426
+ 3520 train 4.251930 (lr=8.8058e-04) (hash(x)=152978060)
427
+ 3530 train 4.190551 (lr=8.7870e-04) (hash(x)=146800180)
428
+ 3540 train 4.263475 (lr=8.7682e-04) (hash(x)=195450444)
429
+ 3550 train 4.247198 (lr=8.7493e-04) (hash(x)=151114129)
430
+ 3560 train 4.230071 (lr=8.7304e-04) (hash(x)=143833190)
431
+ 3570 train 4.036624 (lr=8.7115e-04) (hash(x)=146039971)
432
+ 3580 train 4.097275 (lr=8.6926e-04) (hash(x)=145172249)
433
+ 3590 train 4.013932 (lr=8.6736e-04) (hash(x)=159164125)
434
+ 3600 val loss 4.2868
435
+ 3600 val perplexity 72.7354
436
+ 3600 train 4.269676 (lr=8.6546e-04) (hash(x)=144352932)
437
+ 3610 train 4.224288 (lr=8.6356e-04) (hash(x)=150550670)
438
+ 3620 train 4.351537 (lr=8.6165e-04) (hash(x)=147170125)
439
+ 3630 train 4.154831 (lr=8.5974e-04) (hash(x)=139473935)
440
+ 3640 train 4.134790 (lr=8.5783e-04) (hash(x)=159294424)
441
+ 3650 train 4.272717 (lr=8.5592e-04) (hash(x)=145943278)
442
+ 3660 train 4.321851 (lr=8.5400e-04) (hash(x)=141647872)
443
+ 3670 train 4.294897 (lr=8.5208e-04) (hash(x)=147109351)
444
+ 3680 train 4.312623 (lr=8.5015e-04) (hash(x)=140683368)
445
+ 3690 train 4.230763 (lr=8.4823e-04) (hash(x)=156276284)
446
+ 3700 val loss 4.2646
447
+ 3700 val perplexity 71.1347
448
+ 3700 train 4.179705 (lr=8.4630e-04) (hash(x)=149389012)
449
+ 3710 train 4.314538 (lr=8.4437e-04) (hash(x)=142107929)
450
+ 3720 train 4.274600 (lr=8.4243e-04) (hash(x)=147828843)
451
+ 3730 train 4.347709 (lr=8.4050e-04) (hash(x)=141784679)
452
+ 3740 train 4.252450 (lr=8.3856e-04) (hash(x)=153375464)
453
+ 3750 train 4.186852 (lr=8.3662e-04) (hash(x)=153716001)
454
+ 3760 train 4.150516 (lr=8.3467e-04) (hash(x)=146848822)
455
+ 3770 train 4.111072 (lr=8.3273e-04) (hash(x)=148396782)
456
+ 3780 train 4.181536 (lr=8.3078e-04) (hash(x)=153563244)
457
+ 3790 train 4.095987 (lr=8.2882e-04) (hash(x)=160808734)
458
+ 3800 val loss 4.2598
459
+ 3800 val perplexity 70.7964
460
+ 3800 train 4.276231 (lr=8.2687e-04) (hash(x)=146607620)
461
+ 3810 train 4.256849 (lr=8.2491e-04) (hash(x)=152580677)
462
+ 3820 train 4.604735 (lr=8.2295e-04) (hash(x)=144860879)
463
+ 3830 train 4.265968 (lr=8.2099e-04) (hash(x)=162099851)
464
+ 3840 train 4.229902 (lr=8.1903e-04) (hash(x)=159506855)
465
+ 3850 train 4.146537 (lr=8.1706e-04) (hash(x)=140022686)
466
+ 3860 train 4.217778 (lr=8.1509e-04) (hash(x)=152137850)
467
+ 3870 train 4.340480 (lr=8.1312e-04) (hash(x)=142666785)
468
+ 3880 train 4.198164 (lr=8.1115e-04) (hash(x)=148285868)
469
+ 3890 train 4.153683 (lr=8.0917e-04) (hash(x)=147910067)
470
+ 3900 val loss 4.2583
471
+ 3900 val perplexity 70.6894
472
+ 3900 train 4.196803 (lr=8.0720e-04) (hash(x)=143754617)
473
+ 3910 train 4.193634 (lr=8.0522e-04) (hash(x)=154779949)
474
+ 3920 train 4.030639 (lr=8.0324e-04) (hash(x)=137824061)
475
+ 3930 train 4.011237 (lr=8.0125e-04) (hash(x)=136933373)
476
+ 3940 train 4.082316 (lr=7.9927e-04) (hash(x)=164054337)
477
+ 3950 train 4.273435 (lr=7.9728e-04) (hash(x)=155752984)
478
+ 3960 train 4.384520 (lr=7.9529e-04) (hash(x)=160957673)
479
+ 3970 train 4.259131 (lr=7.9330e-04) (hash(x)=157158690)
480
+ 3980 train 4.263556 (lr=7.9130e-04) (hash(x)=153282027)
481
+ 3990 train 4.292992 (lr=7.8931e-04) (hash(x)=152183510)
482
+ 4000 val loss 4.2331
483
+ 4000 val perplexity 68.9272
484
+ 4000 train 4.179786 (lr=7.8731e-04) (hash(x)=156930722)
485
+ 4010 train 4.177989 (lr=7.8531e-04) (hash(x)=150798740)
486
+ 4020 train 4.385306 (lr=7.8331e-04) (hash(x)=142632400)
487
+ 4030 train 4.287009 (lr=7.8131e-04) (hash(x)=173326287)
488
+ 4040 train 4.194666 (lr=7.7930e-04) (hash(x)=128797443)
489
+ 4050 train 4.179008 (lr=7.7730e-04) (hash(x)=146937792)
490
+ 4060 train 4.282561 (lr=7.7529e-04) (hash(x)=164579030)
491
+ 4070 train 4.154787 (lr=7.7328e-04) (hash(x)=149459820)
492
+ 4080 train 4.119593 (lr=7.7127e-04) (hash(x)=160159595)
493
+ 4090 train 4.061828 (lr=7.6925e-04) (hash(x)=141507853)
494
+ 4100 val loss 4.2289
495
+ 4100 val perplexity 68.6436
496
+ 4100 train 4.195910 (lr=7.6724e-04) (hash(x)=147500519)
497
+ 4110 train 4.169817 (lr=7.6522e-04) (hash(x)=153297081)
498
+ 4120 train 4.186640 (lr=7.6321e-04) (hash(x)=142560288)
499
+ 4130 train 4.197725 (lr=7.6119e-04) (hash(x)=153380118)
500
+ 4140 train 4.099753 (lr=7.5917e-04) (hash(x)=143150720)
501
+ 4150 train 4.148788 (lr=7.5714e-04) (hash(x)=143459564)
502
+ 4160 train 4.177048 (lr=7.5512e-04) (hash(x)=153955741)
503
+ 4170 train 4.287599 (lr=7.5309e-04) (hash(x)=152199253)
504
+ 4180 train 4.053465 (lr=7.5107e-04) (hash(x)=165957668)
505
+ 4190 train 4.101478 (lr=7.4904e-04) (hash(x)=144866923)
506
+ 4200 val loss 4.2386
507
+ 4200 val perplexity 69.3124
508
+ 4200 train 4.109762 (lr=7.4701e-04) (hash(x)=143232237)
509
+ 4210 train 4.032808 (lr=7.4498e-04) (hash(x)=151907163)
510
+ 4220 train 4.129560 (lr=7.4295e-04) (hash(x)=145917587)
511
+ 4230 train 3.954653 (lr=7.4092e-04) (hash(x)=147258846)
512
+ 4240 train 4.111995 (lr=7.3888e-04) (hash(x)=143449843)
513
+ 4250 train 4.073033 (lr=7.3685e-04) (hash(x)=184883715)
514
+ 4260 train 4.169010 (lr=7.3481e-04) (hash(x)=156958761)
515
+ 4270 train 4.235006 (lr=7.3278e-04) (hash(x)=168870544)
516
+ 4280 train 4.165486 (lr=7.3074e-04) (hash(x)=179446735)
517
+ 4290 train 4.205769 (lr=7.2870e-04) (hash(x)=145649386)
518
+ 4300 val loss 4.2218
519
+ 4300 val perplexity 68.1588
520
+ 4300 train 4.058278 (lr=7.2666e-04) (hash(x)=146811670)
521
+ 4310 train 4.327929 (lr=7.2462e-04) (hash(x)=151907075)
522
+ 4320 train 4.126033 (lr=7.2258e-04) (hash(x)=138628365)
523
+ 4330 train 4.185774 (lr=7.2053e-04) (hash(x)=151608240)
524
+ 4340 train 4.122470 (lr=7.1849e-04) (hash(x)=160111319)
525
+ 4350 train 4.268800 (lr=7.1645e-04) (hash(x)=144958652)
526
+ 4360 train 4.052122 (lr=7.1440e-04) (hash(x)=149176222)
527
+ 4370 train 4.163023 (lr=7.1235e-04) (hash(x)=216221371)
528
+ 4380 train 4.058805 (lr=7.1031e-04) (hash(x)=141760648)
529
+ 4390 train 4.148412 (lr=7.0826e-04) (hash(x)=151710283)
530
+ 4400 val loss 4.2119
531
+ 4400 val perplexity 67.4844
532
+ 4400 train 4.273533 (lr=7.0621e-04) (hash(x)=158418746)
533
+ 4410 train 4.118079 (lr=7.0416e-04) (hash(x)=142331301)
534
+ 4420 train 4.212812 (lr=7.0211e-04) (hash(x)=146507180)
535
+ 4430 train 4.298182 (lr=7.0006e-04) (hash(x)=166506937)
536
+ 4440 train 4.235255 (lr=6.9801e-04) (hash(x)=151209447)
537
+ 4450 train 4.203467 (lr=6.9596e-04) (hash(x)=149186822)
538
+ 4460 train 4.182313 (lr=6.9391e-04) (hash(x)=144348412)
539
+ 4470 train 4.070495 (lr=6.9185e-04) (hash(x)=156494284)
540
+ 4480 train 4.100329 (lr=6.8980e-04) (hash(x)=147238744)
541
+ 4490 train 4.047402 (lr=6.8775e-04) (hash(x)=150746185)
542
+ 4500 val loss 4.2025
543
+ 4500 val perplexity 66.8554
544
+ 4500 train 4.140866 (lr=6.8569e-04) (hash(x)=156695778)
545
+ 4510 train 3.981101 (lr=6.8364e-04) (hash(x)=135988958)
546
+ 4520 train 4.156155 (lr=6.8159e-04) (hash(x)=148272110)
547
+ 4530 train 4.070342 (lr=6.7953e-04) (hash(x)=136459794)
548
+ 4540 train 4.103515 (lr=6.7748e-04) (hash(x)=147624954)
549
+ 4550 train 4.106647 (lr=6.7542e-04) (hash(x)=154947199)
550
+ 4560 train 4.165092 (lr=6.7336e-04) (hash(x)=147869166)
551
+ 4570 train 4.022002 (lr=6.7131e-04) (hash(x)=148415463)
552
+ 4580 train 4.177087 (lr=6.6925e-04) (hash(x)=152104066)
553
+ 4590 train 4.074935 (lr=6.6720e-04) (hash(x)=147060040)
554
+ 4600 val loss 4.2002
555
+ 4600 val perplexity 66.6998
556
+ 4600 train 4.149686 (lr=6.6514e-04) (hash(x)=147791497)
557
+ 4610 train 4.348236 (lr=6.6308e-04) (hash(x)=155595556)
558
+ 4620 train 4.193893 (lr=6.6103e-04) (hash(x)=147178858)
559
+ 4630 train 4.233017 (lr=6.5897e-04) (hash(x)=152617112)
560
+ 4640 train 4.248038 (lr=6.5692e-04) (hash(x)=161766869)
561
+ 4650 train 4.227262 (lr=6.5486e-04) (hash(x)=158075561)
562
+ 4660 train 4.274863 (lr=6.5280e-04) (hash(x)=145360002)
563
+ 4670 train 4.147588 (lr=6.5075e-04) (hash(x)=144538791)
564
+ 4680 train 4.191603 (lr=6.4869e-04) (hash(x)=142711273)
565
+ 4690 train 4.344150 (lr=6.4664e-04) (hash(x)=193907362)
566
+ 4700 val loss 4.1812
567
+ 4700 val perplexity 65.4432
568
+ 4700 train 4.153625 (lr=6.4458e-04) (hash(x)=155533088)
569
+ 4710 train 4.109363 (lr=6.4252e-04) (hash(x)=148618528)
570
+ 4720 train 4.215922 (lr=6.4047e-04) (hash(x)=150502624)
571
+ 4730 train 4.090084 (lr=6.3841e-04) (hash(x)=189205847)
572
+ 4740 train 4.150286 (lr=6.3636e-04) (hash(x)=167322800)
573
+ 4750 train 4.264690 (lr=6.3431e-04) (hash(x)=162169095)
574
+ 4760 train 3.973631 (lr=6.3225e-04) (hash(x)=143459469)
575
+ 4770 train 4.046574 (lr=6.3020e-04) (hash(x)=144386593)
576
+ 4780 train 4.014859 (lr=6.2815e-04) (hash(x)=143721184)
577
+ 4790 train 4.081519 (lr=6.2609e-04) (hash(x)=146077222)
578
+ 4800 val loss 4.1867
579
+ 4800 val perplexity 65.8040
580
+ 4800 train 4.053353 (lr=6.2404e-04) (hash(x)=138350044)
581
+ 4810 train 4.063749 (lr=6.2199e-04) (hash(x)=154960112)
582
+ 4820 train 3.974273 (lr=6.1994e-04) (hash(x)=146950685)
583
+ 4830 train 3.974960 (lr=6.1789e-04) (hash(x)=140224554)
584
+ 4840 train 3.970653 (lr=6.1584e-04) (hash(x)=156640495)
585
+ 4850 train 4.178516 (lr=6.1379e-04) (hash(x)=153409150)
586
+ 4860 train 4.097592 (lr=6.1174e-04) (hash(x)=151647700)
587
+ 4870 train 4.485693 (lr=6.0969e-04) (hash(x)=155465603)
588
+ 4880 train 4.343468 (lr=6.0765e-04) (hash(x)=153404400)
589
+ 4890 train 4.217564 (lr=6.0560e-04) (hash(x)=170409281)
590
+ 4900 val loss 4.1704
591
+ 4900 val perplexity 64.7417
592
+ 4900 train 4.087573 (lr=6.0355e-04) (hash(x)=143735284)
593
+ 4910 train 4.236734 (lr=6.0151e-04) (hash(x)=150784329)
594
+ 4920 train 4.114401 (lr=5.9947e-04) (hash(x)=152353574)
595
+ 4930 train 4.067602 (lr=5.9742e-04) (hash(x)=152677008)
596
+ 4940 train 4.146633 (lr=5.9538e-04) (hash(x)=140553861)
597
+ 4950 train 4.228122 (lr=5.9334e-04) (hash(x)=152869181)
598
+ 4960 train 4.190352 (lr=5.9130e-04) (hash(x)=153011079)
599
+ 4970 train 4.133682 (lr=5.8926e-04) (hash(x)=152796617)
600
+ 4980 train 4.245240 (lr=5.8722e-04) (hash(x)=146868417)
601
+ 4990 train 4.130507 (lr=5.8519e-04) (hash(x)=140168188)
602
+ 5000 val loss 4.1612
603
+ 5000 val perplexity 64.1459
604
+ 5000 train 4.115911 (lr=5.8315e-04) (hash(x)=154976463)
605
+ 5010 train 4.170545 (lr=5.8112e-04) (hash(x)=148582690)
606
+ 5020 train 4.179060 (lr=5.7908e-04) (hash(x)=166715464)
607
+ 5030 train 4.140876 (lr=5.7705e-04) (hash(x)=150616397)
608
+ 5040 train 4.196403 (lr=5.7502e-04) (hash(x)=150775548)
609
+ 5050 train 4.177640 (lr=5.7299e-04) (hash(x)=150575367)
610
+ 5060 train 4.095939 (lr=5.7096e-04) (hash(x)=152352082)
611
+ 5070 train 4.134336 (lr=5.6893e-04) (hash(x)=150269272)
612
+ 5080 train 4.199690 (lr=5.6691e-04) (hash(x)=146499361)
613
+ 5090 train 3.885907 (lr=5.6488e-04) (hash(x)=148734454)
614
+ 5100 val loss 4.1627
615
+ 5100 val perplexity 64.2449
616
+ 5100 train 4.069824 (lr=5.6286e-04) (hash(x)=149894982)
617
+ 5110 train 4.120303 (lr=5.6083e-04) (hash(x)=154487751)
618
+ 5120 train 4.064907 (lr=5.5881e-04) (hash(x)=144169582)
619
+ 5130 train 4.046687 (lr=5.5679e-04) (hash(x)=152217630)
620
+ 5140 train 4.062883 (lr=5.5478e-04) (hash(x)=158233831)
621
+ 5150 train 4.135508 (lr=5.5276e-04) (hash(x)=142746426)
622
+ 5160 train 4.122433 (lr=5.5075e-04) (hash(x)=139184458)
623
+ 5170 train 4.201917 (lr=5.4873e-04) (hash(x)=140624857)
624
+ 5180 train 3.932074 (lr=5.4672e-04) (hash(x)=154107423)
625
+ 5190 train 3.873277 (lr=5.4471e-04) (hash(x)=144892610)
626
+ 5200 val loss 4.1630
627
+ 5200 val perplexity 64.2653
628
+ 5200 train 3.825698 (lr=5.4270e-04) (hash(x)=159326689)
629
+ 5210 train 3.995971 (lr=5.4070e-04) (hash(x)=151186017)
630
+ 5220 train 3.798286 (lr=5.3869e-04) (hash(x)=148996819)
631
+ 5230 train 3.740932 (lr=5.3669e-04) (hash(x)=150829917)
632
+ 5240 train 3.860798 (lr=5.3469e-04) (hash(x)=152758972)
633
+ 5250 train 3.829326 (lr=5.3269e-04) (hash(x)=154345903)
634
+ 5260 train 3.826759 (lr=5.3069e-04) (hash(x)=150134947)
635
+ 5270 train 3.912770 (lr=5.2870e-04) (hash(x)=149689310)
636
+ 5280 train 4.155208 (lr=5.2670e-04) (hash(x)=147662223)
637
+ 5290 train 4.220650 (lr=5.2471e-04) (hash(x)=165770608)
638
+ 5300 val loss 4.1503
639
+ 5300 val perplexity 63.4501
640
+ 5300 train 4.200786 (lr=5.2272e-04) (hash(x)=159484800)
641
+ 5310 train 4.233901 (lr=5.2073e-04) (hash(x)=152672536)
642
+ 5320 train 4.061639 (lr=5.1875e-04) (hash(x)=152728812)
643
+ 5330 train 4.001199 (lr=5.1676e-04) (hash(x)=152778685)
644
+ 5340 train 4.100103 (lr=5.1478e-04) (hash(x)=149818849)
645
+ 5350 train 4.027247 (lr=5.1280e-04) (hash(x)=168239739)
646
+ 5360 train 4.177664 (lr=5.1083e-04) (hash(x)=139859423)
647
+ 5370 train 3.997834 (lr=5.0885e-04) (hash(x)=139315327)
648
+ 5380 train 4.208892 (lr=5.0688e-04) (hash(x)=163068509)
649
+ 5390 train 4.159989 (lr=5.0491e-04) (hash(x)=147645415)
650
+ 5400 val loss 4.1302
651
+ 5400 val perplexity 62.1912
652
+ 5400 train 4.115530 (lr=5.0294e-04) (hash(x)=140385615)
653
+ 5410 train 3.975234 (lr=5.0097e-04) (hash(x)=151381862)
654
+ 5420 train 4.131387 (lr=4.9901e-04) (hash(x)=152445092)
655
+ 5430 train 4.145544 (lr=4.9705e-04) (hash(x)=142454479)
656
+ 5440 train 4.096705 (lr=4.9509e-04) (hash(x)=151308771)
657
+ 5450 train 3.949914 (lr=4.9313e-04) (hash(x)=150886070)
658
+ 5460 train 4.058845 (lr=4.9118e-04) (hash(x)=146080408)
659
+ 5470 train 4.075623 (lr=4.8922e-04) (hash(x)=149828857)
660
+ 5480 train 4.053894 (lr=4.8727e-04) (hash(x)=139922034)
661
+ 5490 train 3.995850 (lr=4.8533e-04) (hash(x)=150262276)
662
+ 5500 val loss 4.1291
663
+ 5500 val perplexity 62.1229
664
+ 5500 train 3.967076 (lr=4.8338e-04) (hash(x)=148498335)
665
+ 5510 train 4.086617 (lr=4.8144e-04) (hash(x)=157036224)
666
+ 5520 train 3.968736 (lr=4.7950e-04) (hash(x)=134093409)
667
+ 5530 train 3.995223 (lr=4.7757e-04) (hash(x)=153126250)
668
+ 5540 train 4.012202 (lr=4.7563e-04) (hash(x)=136170985)
669
+ 5550 train 3.986224 (lr=4.7370e-04) (hash(x)=152841201)
670
+ 5560 train 4.091094 (lr=4.7177e-04) (hash(x)=147439014)
671
+ 5570 train 4.008327 (lr=4.6985e-04) (hash(x)=143476132)
672
+ 5580 train 3.777795 (lr=4.6792e-04) (hash(x)=139503437)
673
+ 5590 train 3.827365 (lr=4.6600e-04) (hash(x)=131664955)
674
+ 5600 val loss 4.1382
675
+ 5600 val perplexity 62.6930
676
+ 5600 train 3.915219 (lr=4.6408e-04) (hash(x)=151907614)
677
+ 5610 train 3.801186 (lr=4.6217e-04) (hash(x)=157132674)
678
+ 5620 train 4.006526 (lr=4.6026e-04) (hash(x)=146793775)
679
+ 5630 train 3.950544 (lr=4.5835e-04) (hash(x)=147237910)
680
+ 5640 train 3.859480 (lr=4.5644e-04) (hash(x)=141120024)
681
+ 5650 train 3.843569 (lr=4.5454e-04) (hash(x)=144275645)
682
+ 5660 train 3.865237 (lr=4.5264e-04) (hash(x)=159088002)
683
+ 5670 train 3.925241 (lr=4.5074e-04) (hash(x)=147224479)
684
+ 5680 train 4.268723 (lr=4.4885e-04) (hash(x)=147208743)
685
+ 5690 train 4.069390 (lr=4.4696e-04) (hash(x)=154384701)
686
+ 5700 val loss 4.1305
687
+ 5700 val perplexity 62.2098
688
+ 5700 train 4.106432 (lr=4.4507e-04) (hash(x)=155192267)
689
+ 5710 train 4.049497 (lr=4.4318e-04) (hash(x)=148458753)
690
+ 5720 train 4.159142 (lr=4.4130e-04) (hash(x)=159947204)
691
+ 5730 train 4.142191 (lr=4.3942e-04) (hash(x)=150746366)
692
+ 5740 train 4.139050 (lr=4.3755e-04) (hash(x)=155201884)
693
+ 5750 train 4.135178 (lr=4.3568e-04) (hash(x)=156432008)
694
+ 5760 train 4.119980 (lr=4.3381e-04) (hash(x)=185977746)
695
+ 5770 train 4.117015 (lr=4.3194e-04) (hash(x)=133620000)
696
+ 5780 train 4.134962 (lr=4.3008e-04) (hash(x)=153778044)
697
+ 5790 train 3.974139 (lr=4.2822e-04) (hash(x)=147713822)
698
+ 5800 val loss 4.1080
699
+ 5800 val perplexity 60.8258
700
+ 5800 train 4.127511 (lr=4.2636e-04) (hash(x)=153132158)
701
+ 5810 train 4.086795 (lr=4.2451e-04) (hash(x)=144293895)
702
+ 5820 train 4.066003 (lr=4.2266e-04) (hash(x)=151500590)
703
+ 5830 train 4.014265 (lr=4.2082e-04) (hash(x)=139308707)
704
+ 5840 train 4.127144 (lr=4.1898e-04) (hash(x)=153591278)
705
+ 5850 train 4.108771 (lr=4.1714e-04) (hash(x)=146077037)
706
+ 5860 train 4.074887 (lr=4.1530e-04) (hash(x)=149160733)
707
+ 5870 train 4.053296 (lr=4.1347e-04) (hash(x)=138482476)
708
+ 5880 train 4.067410 (lr=4.1164e-04) (hash(x)=152173137)
709
+ 5890 train 4.042970 (lr=4.0982e-04) (hash(x)=142387057)
710
+ 5900 val loss 4.1185
711
+ 5900 val perplexity 61.4648
712
+ 5900 train 4.122308 (lr=4.0800e-04) (hash(x)=161446764)
713
+ 5910 train 3.992386 (lr=4.0618e-04) (hash(x)=144512138)
714
+ 5920 train 4.054259 (lr=4.0437e-04) (hash(x)=152555618)
715
+ 5930 train 4.011058 (lr=4.0256e-04) (hash(x)=146129367)
716
+ 5940 train 3.947918 (lr=4.0075e-04) (hash(x)=154607527)
717
+ 5950 train 4.036106 (lr=3.9895e-04) (hash(x)=143827143)
718
+ 5960 train 4.165406 (lr=3.9715e-04) (hash(x)=151168510)
719
+ 5970 train 3.951754 (lr=3.9536e-04) (hash(x)=158961574)
720
+ 5980 train 3.789775 (lr=3.9357e-04) (hash(x)=148757787)
721
+ 5990 train 4.286620 (lr=3.9178e-04) (hash(x)=161719494)
722
+ 6000 val loss 4.1174
723
+ 6000 val perplexity 61.3967
724
+ 6000 train 3.791449 (lr=3.9000e-04) (hash(x)=151512446)
725
+ 6010 train 3.905854 (lr=3.8822e-04) (hash(x)=173500735)
726
+ 6020 train 4.042132 (lr=3.8645e-04) (hash(x)=157621031)
727
+ 6030 train 3.924375 (lr=3.8468e-04) (hash(x)=142662684)
728
+ 6040 train 3.914353 (lr=3.8291e-04) (hash(x)=146180940)
729
+ 6050 train 3.834657 (lr=3.8115e-04) (hash(x)=147654412)
730
+ 6060 train 3.702378 (lr=3.7939e-04) (hash(x)=156393157)
731
+ 6070 train 4.113428 (lr=3.7763e-04) (hash(x)=156947300)
732
+ 6080 train 4.099291 (lr=3.7588e-04) (hash(x)=163108626)
733
+ 6090 train 4.197640 (lr=3.7413e-04) (hash(x)=158654713)
734
+ 6100 val loss 4.0960
735
+ 6100 val perplexity 60.0983
736
+ 6100 train 3.925329 (lr=3.7239e-04) (hash(x)=188094053)
737
+ 6110 train 4.086690 (lr=3.7065e-04) (hash(x)=145132600)
738
+ 6120 train 3.964906 (lr=3.6892e-04) (hash(x)=146999707)
739
+ 6130 train 4.192114 (lr=3.6719e-04) (hash(x)=145719530)
740
+ 6140 train 4.128633 (lr=3.6546e-04) (hash(x)=153230722)
741
+ 6150 train 4.060086 (lr=3.6374e-04) (hash(x)=147455934)
742
+ 6160 train 4.113932 (lr=3.6203e-04) (hash(x)=146596242)
743
+ 6170 train 4.205588 (lr=3.6031e-04) (hash(x)=152029855)
744
+ 6180 train 4.074909 (lr=3.5860e-04) (hash(x)=151480034)
745
+ 6190 train 4.003614 (lr=3.5690e-04) (hash(x)=147381047)
746
+ 6200 val loss 4.0801
747
+ 6200 val perplexity 59.1503
748
+ 6200 train 3.995852 (lr=3.5520e-04) (hash(x)=149389789)
749
+ 6210 train 4.046740 (lr=3.5351e-04) (hash(x)=154894138)
750
+ 6220 train 4.075162 (lr=3.5181e-04) (hash(x)=150074302)
751
+ 6230 train 3.948328 (lr=3.5013e-04) (hash(x)=142881794)
752
+ 6240 train 3.929728 (lr=3.4845e-04) (hash(x)=164747461)
753
+ 6250 train 4.096503 (lr=3.4677e-04) (hash(x)=158378292)
754
+ 6260 train 4.103336 (lr=3.4510e-04) (hash(x)=147675938)
755
+ 6270 train 4.045372 (lr=3.4343e-04) (hash(x)=148197228)
756
+ 6280 train 4.011098 (lr=3.4176e-04) (hash(x)=135296747)
757
+ 6290 train 4.073007 (lr=3.4011e-04) (hash(x)=156439346)
758
+ 6300 val loss 4.0798
759
+ 6300 val perplexity 59.1322
760
+ 6300 train 3.897714 (lr=3.3845e-04) (hash(x)=138212820)
761
+ 6310 train 4.189081 (lr=3.3680e-04) (hash(x)=165985336)
762
+ 6320 train 4.066774 (lr=3.3516e-04) (hash(x)=145921897)
763
+ 6330 train 4.269836 (lr=3.3352e-04) (hash(x)=145502689)
764
+ 6340 train 4.009390 (lr=3.3188e-04) (hash(x)=141704937)
765
+ 6350 train 4.065519 (lr=3.3025e-04) (hash(x)=161789400)
766
+ 6360 train 3.790272 (lr=3.2862e-04) (hash(x)=145457780)
767
+ 6370 train 4.039013 (lr=3.2700e-04) (hash(x)=171628530)
768
+ 6380 train 3.830574 (lr=3.2539e-04) (hash(x)=153467923)
769
+ 6390 train 4.205006 (lr=3.2378e-04) (hash(x)=158066904)
770
+ 6400 val loss 4.0861
771
+ 6400 val perplexity 59.5086
772
+ 6400 train 3.854095 (lr=3.2217e-04) (hash(x)=146535423)
773
+ 6410 train 3.717367 (lr=3.2057e-04) (hash(x)=147264155)
774
+ 6420 train 3.822229 (lr=3.1897e-04) (hash(x)=145595485)
775
+ 6430 train 3.968632 (lr=3.1738e-04) (hash(x)=146929074)
776
+ 6440 train 3.895059 (lr=3.1579e-04) (hash(x)=147290878)
777
+ 6450 train 3.992548 (lr=3.1421e-04) (hash(x)=141154205)
778
+ 6460 train 3.832183 (lr=3.1263e-04) (hash(x)=154360901)
779
+ 6470 train 4.171207 (lr=3.1106e-04) (hash(x)=146350558)
780
+ 6480 train 4.131766 (lr=3.0949e-04) (hash(x)=153927683)
781
+ 6490 train 4.051302 (lr=3.0793e-04) (hash(x)=149337047)
782
+ 6500 val loss 4.0721
783
+ 6500 val perplexity 58.6804
784
+ 6500 train 3.967958 (lr=3.0638e-04) (hash(x)=145950843)
785
+ 6510 train 4.034039 (lr=3.0482e-04) (hash(x)=142577653)
786
+ 6520 train 4.083413 (lr=3.0328e-04) (hash(x)=165060689)
787
+ 6530 train 3.938997 (lr=3.0174e-04) (hash(x)=148367505)
788
+ 6540 train 4.031820 (lr=3.0020e-04) (hash(x)=154313651)
789
+ 6550 train 3.948436 (lr=2.9867e-04) (hash(x)=147477315)
790
+ 6560 train 4.033904 (lr=2.9714e-04) (hash(x)=139825403)
791
+ 6570 train 4.050428 (lr=2.9562e-04) (hash(x)=141482518)
792
+ 6580 train 4.084535 (lr=2.9411e-04) (hash(x)=144968144)
793
+ 6590 train 4.041948 (lr=2.9260e-04) (hash(x)=154319720)
794
+ 6600 val loss 4.0684
795
+ 6600 val perplexity 58.4652
796
+ 6600 train 4.077673 (lr=2.9109e-04) (hash(x)=141162902)
797
+ 6610 train 4.049996 (lr=2.8960e-04) (hash(x)=159490449)
798
+ 6620 train 3.931417 (lr=2.8810e-04) (hash(x)=151189772)
799
+ 6630 train 3.907310 (lr=2.8661e-04) (hash(x)=148451651)
800
+ 6640 train 4.037731 (lr=2.8513e-04) (hash(x)=152815550)
801
+ 6650 train 3.883176 (lr=2.8365e-04) (hash(x)=144316094)
802
+ 6660 train 4.198475 (lr=2.8218e-04) (hash(x)=143951240)
803
+ 6670 train 3.824023 (lr=2.8072e-04) (hash(x)=143203472)
804
+ 6680 train 3.863539 (lr=2.7925e-04) (hash(x)=132063806)
805
+ 6690 train 3.957731 (lr=2.7780e-04) (hash(x)=140739750)
806
+ 6700 val loss 4.0716
807
+ 6700 val perplexity 58.6515
808
+ 6700 train 4.017274 (lr=2.7635e-04) (hash(x)=153018737)
809
+ 6710 train 3.881058 (lr=2.7490e-04) (hash(x)=140107197)
810
+ 6720 train 3.986820 (lr=2.7347e-04) (hash(x)=144396559)
811
+ 6730 train 3.833647 (lr=2.7203e-04) (hash(x)=148038991)
812
+ 6740 train 4.046393 (lr=2.7061e-04) (hash(x)=145504746)
813
+ 6750 train 4.034576 (lr=2.6918e-04) (hash(x)=119579094)
814
+ 6760 train 4.046555 (lr=2.6777e-04) (hash(x)=161725430)
815
+ 6770 train 4.064195 (lr=2.6636e-04) (hash(x)=153646966)
816
+ 6780 train 4.068097 (lr=2.6495e-04) (hash(x)=153838463)
817
+ 6790 train 4.197481 (lr=2.6355e-04) (hash(x)=157891196)
818
+ 6800 val loss 4.0539
819
+ 6800 val perplexity 57.6225
820
+ 6800 train 3.905010 (lr=2.6216e-04) (hash(x)=155640155)
821
+ 6810 train 4.102526 (lr=2.6077e-04) (hash(x)=154913809)
822
+ 6820 train 4.019461 (lr=2.5939e-04) (hash(x)=159412493)
823
+ 6830 train 3.975969 (lr=2.5801e-04) (hash(x)=144756549)
824
+ 6840 train 4.190814 (lr=2.5664e-04) (hash(x)=143824638)
825
+ 6850 train 4.521317 (lr=2.5528e-04) (hash(x)=117180982)
826
+ 6860 train 4.037671 (lr=2.5392e-04) (hash(x)=141887150)
827
+ 6870 train 4.108096 (lr=2.5257e-04) (hash(x)=140094872)
828
+ 6880 train 3.948513 (lr=2.5122e-04) (hash(x)=141479953)
829
+ 6890 train 4.007076 (lr=2.4988e-04) (hash(x)=153166840)
830
+ 6900 val loss 4.0510
831
+ 6900 val perplexity 57.4553
832
+ 6900 train 4.026999 (lr=2.4855e-04) (hash(x)=153722115)
833
+ 6910 train 4.138263 (lr=2.4722e-04) (hash(x)=149578602)
834
+ 6920 train 3.988453 (lr=2.4590e-04) (hash(x)=149925108)
835
+ 6930 train 3.931774 (lr=2.4458e-04) (hash(x)=148589489)
836
+ 6940 train 3.843216 (lr=2.4327e-04) (hash(x)=144243794)
837
+ 6950 train 3.928389 (lr=2.4196e-04) (hash(x)=157743480)
838
+ 6960 train 3.918334 (lr=2.4066e-04) (hash(x)=139858247)
839
+ 6970 train 3.850330 (lr=2.3937e-04) (hash(x)=148727536)
840
+ 6980 train 3.957868 (lr=2.3809e-04) (hash(x)=153382595)
841
+ 6990 train 4.099287 (lr=2.3681e-04) (hash(x)=147529919)
842
+ 7000 val loss 4.0460
843
+ 7000 val perplexity 57.1675
844
+ 7000 train 4.062323 (lr=2.3553e-04) (hash(x)=146953450)
845
+ 7010 train 3.977045 (lr=2.3426e-04) (hash(x)=134561015)
846
+ 7020 train 4.084570 (lr=2.3300e-04) (hash(x)=148433515)
847
+ 7030 train 3.940173 (lr=2.3175e-04) (hash(x)=140394326)
848
+ 7040 train 3.942498 (lr=2.3050e-04) (hash(x)=136740763)
849
+ 7050 train 3.783277 (lr=2.2925e-04) (hash(x)=148637938)
850
+ 7060 train 3.951658 (lr=2.2802e-04) (hash(x)=147262147)
851
+ 7070 train 3.925848 (lr=2.2679e-04) (hash(x)=140191039)
852
+ 7080 train 3.940586 (lr=2.2556e-04) (hash(x)=148223305)
853
+ 7090 train 3.854826 (lr=2.2434e-04) (hash(x)=139984988)
854
+ 7100 val loss 4.0488
855
+ 7100 val perplexity 57.3300
856
+ 7100 train 3.968258 (lr=2.2313e-04) (hash(x)=137663885)
857
+ 7110 train 3.853765 (lr=2.2193e-04) (hash(x)=150955021)
858
+ 7120 train 3.758176 (lr=2.2073e-04) (hash(x)=148585682)
859
+ 7130 train 3.965763 (lr=2.1953e-04) (hash(x)=146055548)
860
+ 7140 train 4.002175 (lr=2.1835e-04) (hash(x)=155723541)
861
+ 7150 train 4.072872 (lr=2.1717e-04) (hash(x)=140465834)
862
+ 7160 train 4.045044 (lr=2.1599e-04) (hash(x)=163246922)
863
+ 7170 train 3.903509 (lr=2.1483e-04) (hash(x)=150754166)
864
+ 7180 train 4.442712 (lr=2.1367e-04) (hash(x)=142844151)
865
+ 7190 train 4.028513 (lr=2.1251e-04) (hash(x)=145167470)
866
+ 7200 val loss 4.0327
867
+ 7200 val perplexity 56.4122
868
+ 7200 train 4.223575 (lr=2.1136e-04) (hash(x)=146172950)
869
+ 7210 train 4.044261 (lr=2.1022e-04) (hash(x)=143155287)
870
+ 7220 train 3.998927 (lr=2.0909e-04) (hash(x)=151976669)
871
+ 7230 train 4.110327 (lr=2.0796e-04) (hash(x)=145979940)
872
+ 7240 train 4.077916 (lr=2.0684e-04) (hash(x)=155656034)
873
+ 7250 train 4.078251 (lr=2.0572e-04) (hash(x)=135610136)
874
+ 7260 train 4.159351 (lr=2.0461e-04) (hash(x)=148133023)
875
+ 7270 train 4.035693 (lr=2.0351e-04) (hash(x)=148006681)
876
+ 7280 train 4.067909 (lr=2.0242e-04) (hash(x)=151397627)
877
+ 7290 train 4.029603 (lr=2.0133e-04) (hash(x)=141696871)
878
+ 7300 val loss 4.0277
879
+ 7300 val perplexity 56.1334
880
+ 7300 train 3.963065 (lr=2.0025e-04) (hash(x)=150018163)
881
+ 7310 train 4.002186 (lr=1.9917e-04) (hash(x)=154900050)
882
+ 7320 train 3.903433 (lr=1.9810e-04) (hash(x)=142608283)
883
+ 7330 train 3.997993 (lr=1.9704e-04) (hash(x)=148977837)
884
+ 7340 train 4.020750 (lr=1.9599e-04) (hash(x)=141676907)
885
+ 7350 train 3.902225 (lr=1.9494e-04) (hash(x)=147846997)
886
+ 7360 train 4.148526 (lr=1.9390e-04) (hash(x)=143252181)
887
+ 7370 train 3.968790 (lr=1.9286e-04) (hash(x)=155081717)
888
+ 7380 train 4.054010 (lr=1.9183e-04) (hash(x)=149678451)
889
+ 7390 train 3.888100 (lr=1.9081e-04) (hash(x)=159341637)
890
+ 7400 val loss 4.0277
891
+ 7400 val perplexity 56.1336
892
+ 7400 train 4.169893 (lr=1.8980e-04) (hash(x)=145351166)
893
+ 7410 train 3.984521 (lr=1.8879e-04) (hash(x)=162274009)
894
+ 7420 train 3.826013 (lr=1.8779e-04) (hash(x)=143254858)
895
+ 7430 train 3.942020 (lr=1.8679e-04) (hash(x)=143049884)
896
+ 7440 train 3.868019 (lr=1.8581e-04) (hash(x)=141768538)
897
+ 7450 train 3.894620 (lr=1.8483e-04) (hash(x)=139895310)
898
+ 7460 train 3.765445 (lr=1.8385e-04) (hash(x)=155900365)
899
+ 7470 train 3.745291 (lr=1.8289e-04) (hash(x)=145048215)
900
+ 7480 train 3.889608 (lr=1.8193e-04) (hash(x)=136992624)
901
+ 7490 train 4.001017 (lr=1.8097e-04) (hash(x)=141424789)
902
+ 7500 val loss 4.0336
903
+ 7500 val perplexity 56.4654
904
+ 7500 train 3.848607 (lr=1.8003e-04) (hash(x)=145292116)
905
+ 7510 train 3.883948 (lr=1.7909e-04) (hash(x)=144818029)
906
+ 7520 train 4.053825 (lr=1.7816e-04) (hash(x)=152903515)
907
+ 7530 train 4.063888 (lr=1.7723e-04) (hash(x)=156934724)
908
+ 7540 train 4.290950 (lr=1.7632e-04) (hash(x)=156695926)
909
+ 7550 train 4.048392 (lr=1.7540e-04) (hash(x)=149115609)
910
+ 7560 train 4.028306 (lr=1.7450e-04) (hash(x)=143974224)
911
+ 7570 train 3.985258 (lr=1.7360e-04) (hash(x)=148204424)
912
+ 7580 train 4.057468 (lr=1.7271e-04) (hash(x)=150177910)
913
+ 7590 train 4.077811 (lr=1.7183e-04) (hash(x)=140325629)
914
+ 7600 val loss 4.0183
915
+ 7600 val perplexity 55.6038
916
+ 7600 train 3.957198 (lr=1.7096e-04) (hash(x)=150235132)
917
+ 7610 train 4.082134 (lr=1.7009e-04) (hash(x)=157594837)
918
+ 7620 train 4.098681 (lr=1.6923e-04) (hash(x)=146696749)
919
+ 7630 train 4.136755 (lr=1.6837e-04) (hash(x)=148463580)
920
+ 7640 train 3.952573 (lr=1.6753e-04) (hash(x)=140271046)
921
+ 7650 train 4.010498 (lr=1.6669e-04) (hash(x)=159952319)
922
+ 7660 train 4.011227 (lr=1.6585e-04) (hash(x)=146307392)
923
+ 7670 train 3.963388 (lr=1.6503e-04) (hash(x)=145027988)
924
+ 7680 train 4.024128 (lr=1.6421e-04) (hash(x)=147729698)
925
+ 7690 train 4.017969 (lr=1.6340e-04) (hash(x)=143401210)
926
+ 7700 val loss 4.0130
927
+ 7700 val perplexity 55.3099
928
+ 7700 train 3.945021 (lr=1.6259e-04) (hash(x)=154543455)
929
+ 7710 train 4.019877 (lr=1.6180e-04) (hash(x)=151600456)
930
+ 7720 train 3.929458 (lr=1.6101e-04) (hash(x)=165913803)
931
+ 7730 train 3.808218 (lr=1.6022e-04) (hash(x)=144942059)
932
+ 7740 train 4.030307 (lr=1.5945e-04) (hash(x)=150302103)
933
+ 7750 train 3.974575 (lr=1.5868e-04) (hash(x)=155782458)
934
+ 7760 train 4.007301 (lr=1.5792e-04) (hash(x)=147824384)
935
+ 7770 train 4.024988 (lr=1.5717e-04) (hash(x)=149177657)
936
+ 7780 train 4.041616 (lr=1.5642e-04) (hash(x)=164446738)
937
+ 7790 train 3.844147 (lr=1.5568e-04) (hash(x)=146417077)
938
+ 7800 val loss 4.0164
939
+ 7800 val perplexity 55.5022
940
+ 7800 train 3.989798 (lr=1.5495e-04) (hash(x)=142456852)
941
+ 7810 train 3.696815 (lr=1.5423e-04) (hash(x)=146221807)
942
+ 7820 train 3.890409 (lr=1.5351e-04) (hash(x)=154136533)
943
+ 7830 train 3.797304 (lr=1.5280e-04) (hash(x)=147492735)
944
+ 7840 train 3.952703 (lr=1.5210e-04) (hash(x)=143481291)
945
+ 7850 train 3.781622 (lr=1.5140e-04) (hash(x)=136749307)
946
+ 7860 train 3.802867 (lr=1.5072e-04) (hash(x)=128985347)
947
+ 7870 train 3.709582 (lr=1.5004e-04) (hash(x)=152897363)
948
+ 7880 train 3.915031 (lr=1.4936e-04) (hash(x)=151460360)
949
+ 7890 train 3.736216 (lr=1.4870e-04) (hash(x)=137481738)
950
+ 7900 val loss 4.0160
951
+ 7900 val perplexity 55.4805
952
+ 7900 train 3.814426 (lr=1.4804e-04) (hash(x)=147363479)
953
+ 7910 train 4.032764 (lr=1.4739e-04) (hash(x)=151165744)
954
+ 7920 train 4.209962 (lr=1.4675e-04) (hash(x)=146214226)
955
+ 7930 train 4.018145 (lr=1.4611e-04) (hash(x)=154457502)
956
+ 7940 train 3.921236 (lr=1.4548e-04) (hash(x)=162025201)
957
+ 7950 train 3.910141 (lr=1.4486e-04) (hash(x)=156488145)
958
+ 7960 train 4.187378 (lr=1.4425e-04) (hash(x)=154011584)
959
+ 7970 train 4.037164 (lr=1.4365e-04) (hash(x)=145330320)
960
+ 7980 train 3.948156 (lr=1.4305e-04) (hash(x)=148216312)
961
+ 7990 train 3.882827 (lr=1.4246e-04) (hash(x)=156819091)
962
+ 8000 val loss 4.0058
963
+ 8000 val perplexity 54.9136
964
+ 8000 train 4.037365 (lr=1.4187e-04) (hash(x)=156122973)
965
+ 8010 train 3.978472 (lr=1.4130e-04) (hash(x)=146933088)
966
+ 8020 train 4.083376 (lr=1.4073e-04) (hash(x)=151650746)
967
+ 8030 train 4.005369 (lr=1.4017e-04) (hash(x)=140692373)
968
+ 8040 train 4.192294 (lr=1.3962e-04) (hash(x)=147850094)
969
+ 8050 train 4.008224 (lr=1.3907e-04) (hash(x)=144454109)
970
+ 8060 train 3.870044 (lr=1.3853e-04) (hash(x)=152205124)
971
+ 8070 train 4.089517 (lr=1.3800e-04) (hash(x)=148918066)
972
+ 8080 train 4.017070 (lr=1.3748e-04) (hash(x)=146481024)
973
+ 8090 train 4.055798 (lr=1.3697e-04) (hash(x)=168785549)
974
+ 8100 val loss 4.0025
975
+ 8100 val perplexity 54.7328
976
+ 8100 train 4.077525 (lr=1.3646e-04) (hash(x)=156153179)
977
+ 8110 train 4.034592 (lr=1.3596e-04) (hash(x)=150498601)
978
+ 8120 train 4.039498 (lr=1.3547e-04) (hash(x)=142002714)
979
+ 8130 train 3.927327 (lr=1.3498e-04) (hash(x)=138450483)
980
+ 8140 train 4.010201 (lr=1.3450e-04) (hash(x)=145306351)
981
+ 8150 train 3.890682 (lr=1.3403e-04) (hash(x)=172502865)
982
+ 8160 train 3.994887 (lr=1.3357e-04) (hash(x)=156689649)
983
+ 8170 train 3.912071 (lr=1.3312e-04) (hash(x)=143766932)
984
+ 8180 train 4.051481 (lr=1.3267e-04) (hash(x)=150361321)
985
+ 8190 train 3.881218 (lr=1.3223e-04) (hash(x)=152664241)
986
+ 8200 val loss 4.0026
987
+ 8200 val perplexity 54.7401
988
+ 8200 train 4.154938 (lr=1.3180e-04) (hash(x)=146430698)
989
+ 8210 train 3.842588 (lr=1.3138e-04) (hash(x)=167362078)
990
+ 8220 train 4.066050 (lr=1.3096e-04) (hash(x)=157650942)
991
+ 8230 train 3.875461 (lr=1.3055e-04) (hash(x)=142102044)
992
+ 8240 train 3.917612 (lr=1.3015e-04) (hash(x)=153405900)
993
+ 8250 train 3.865864 (lr=1.2976e-04) (hash(x)=140222909)
994
+ 8260 train 3.998036 (lr=1.2937e-04) (hash(x)=137981854)
995
+ 8270 train 3.911040 (lr=1.2900e-04) (hash(x)=147576272)
996
+ 8280 train 3.926145 (lr=1.2863e-04) (hash(x)=157988195)
997
+ 8290 train 3.977154 (lr=1.2826e-04) (hash(x)=145715681)
998
+ 8300 val loss 4.0034
999
+ 8300 val perplexity 54.7865
1000
+ 8300 train 3.925410 (lr=1.2791e-04) (hash(x)=143507257)
1001
+ 8310 train 3.982985 (lr=1.2756e-04) (hash(x)=148930064)
1002
+ 8320 train 3.913068 (lr=1.2722e-04) (hash(x)=144134805)
1003
+ 8330 train 4.023242 (lr=1.2689e-04) (hash(x)=148418079)
1004
+ 8340 train 3.753417 (lr=1.2657e-04) (hash(x)=166518807)
1005
+ 8350 train 4.060881 (lr=1.2625e-04) (hash(x)=140042036)
1006
+ 8360 train 3.929688 (lr=1.2594e-04) (hash(x)=151427074)
1007
+ 8370 train 4.017848 (lr=1.2564e-04) (hash(x)=158389185)
1008
+ 8380 train 3.921597 (lr=1.2535e-04) (hash(x)=148193308)
1009
+ 8390 train 3.990497 (lr=1.2507e-04) (hash(x)=144289008)
1010
+ 8400 val loss 3.9931
1011
+ 8400 val perplexity 54.2223
1012
+ 8400 train 3.711891 (lr=1.2479e-04) (hash(x)=166272643)
1013
+ 8410 train 3.912660 (lr=1.2452e-04) (hash(x)=162244433)
1014
+ 8420 train 3.953250 (lr=1.2426e-04) (hash(x)=149271299)
1015
+ 8430 train 4.097416 (lr=1.2400e-04) (hash(x)=154135338)
1016
+ 8440 train 4.138770 (lr=1.2376e-04) (hash(x)=149661838)
1017
+ 8450 train 3.976111 (lr=1.2352e-04) (hash(x)=142205056)
1018
+ 8460 train 3.904755 (lr=1.2329e-04) (hash(x)=166652398)
1019
+ 8470 train 4.086634 (lr=1.2307e-04) (hash(x)=149029710)
1020
+ 8480 train 3.858732 (lr=1.2285e-04) (hash(x)=154217415)
1021
+ 8490 train 3.929013 (lr=1.2264e-04) (hash(x)=151526557)
1022
+ 8500 val loss 3.9946
1023
+ 8500 val perplexity 54.3050
1024
+ 8500 train 3.920306 (lr=1.2245e-04) (hash(x)=143887848)
1025
+ 8510 train 3.907987 (lr=1.2225e-04) (hash(x)=152324645)
1026
+ 8520 train 3.645759 (lr=1.2207e-04) (hash(x)=145124959)
1027
+ 8530 train 3.964489 (lr=1.2189e-04) (hash(x)=186319340)
1028
+ 8540 train 3.979116 (lr=1.2173e-04) (hash(x)=147388339)
1029
+ 8550 train 4.010317 (lr=1.2157e-04) (hash(x)=150118928)
1030
+ 8560 train 4.010992 (lr=1.2141e-04) (hash(x)=153985286)
1031
+ 8570 train 3.946882 (lr=1.2127e-04) (hash(x)=149582799)
1032
+ 8580 train 3.950255 (lr=1.2113e-04) (hash(x)=156867870)
1033
+ 8590 train 3.840837 (lr=1.2100e-04) (hash(x)=148773298)
1034
+ 8600 val loss 3.9971
1035
+ 8600 val perplexity 54.4398
1036
+ 8600 train 3.940178 (lr=1.2088e-04) (hash(x)=156900341)
1037
+ 8610 train 4.095117 (lr=1.2077e-04) (hash(x)=156790089)
1038
+ 8620 train 3.903369 (lr=1.2066e-04) (hash(x)=154680805)
1039
+ 8630 train 3.940312 (lr=1.2056e-04) (hash(x)=138107873)
1040
+ 8640 train 4.023031 (lr=1.2047e-04) (hash(x)=142554243)
1041
+ 8650 train 3.831221 (lr=1.2039e-04) (hash(x)=134360956)
1042
+ 8660 train 3.908174 (lr=1.2032e-04) (hash(x)=128599610)
1043
+ 8670 train 3.943685 (lr=1.2025e-04) (hash(x)=159628945)
1044
+ 8680 train 3.969216 (lr=1.2019e-04) (hash(x)=143161532)
1045
+ 8690 train 3.996668 (lr=1.2014e-04) (hash(x)=140104912)
1046
+ 8700 val loss 3.9892
1047
+ 8700 val perplexity 54.0134
1048
+ 8700 train 4.183076 (lr=1.2010e-04) (hash(x)=146417632)
1049
+ 8710 train 4.152717 (lr=1.2006e-04) (hash(x)=153671695)
1050
+ 8720 train 4.026559 (lr=1.2004e-04) (hash(x)=141680750)
1051
+ 8730 train 3.901224 (lr=1.2002e-04) (hash(x)=147279231)
1052
+ 8740 train 4.190104 (lr=1.2000e-04) (hash(x)=129729371)
1053
+ 8749 val loss 3.9890
1054
+ 8749 val perplexity 53.9984
lr12e-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 sha256:6226384f5a6557247a0269adfb7fdd188d30c6c5b71ca916bc6c757ee70b7530
3
+ size 92843394
lr12e-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:ba9c4cff582a9d0ec78eca5095c00c68fa102c6e4412f8a75881395717dc0c93
3
+ size 179406214