andrew-healey commited on
Commit
02030d1
·
verified ·
1 Parent(s): 034e560

Upload folder using huggingface_hub

Browse files
lr10e-4_total_batch_size61440_baseline_seed1339/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_11/lr10e-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": 0.001, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "10e-4_61440", "n_embd": 256}
lr10e-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
lr10e-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=2.0000e-06) (hash(x)=150724848)
5
+ 10 train 10.169039 (lr=2.2000e-05) (hash(x)=149663361)
6
+ 20 train 9.768238 (lr=4.2000e-05) (hash(x)=153483789)
7
+ 30 train 9.425735 (lr=6.2000e-05) (hash(x)=145733176)
8
+ 40 train 8.814079 (lr=8.2000e-05) (hash(x)=145287265)
9
+ 50 train 8.311483 (lr=1.0200e-04) (hash(x)=149217061)
10
+ 60 train 7.888088 (lr=1.2200e-04) (hash(x)=150157569)
11
+ 70 train 7.644886 (lr=1.4200e-04) (hash(x)=140292841)
12
+ 80 train 7.690776 (lr=1.6200e-04) (hash(x)=152157598)
13
+ 90 train 7.502247 (lr=1.8200e-04) (hash(x)=149756968)
14
+ 100 val loss 7.4628
15
+ 100 val perplexity 1741.9741
16
+ 100 train 7.420193 (lr=2.0200e-04) (hash(x)=149910534)
17
+ 110 train 7.332156 (lr=2.2200e-04) (hash(x)=159020328)
18
+ 120 train 7.257672 (lr=2.4200e-04) (hash(x)=150731273)
19
+ 130 train 7.072872 (lr=2.6200e-04) (hash(x)=156727355)
20
+ 140 train 7.101395 (lr=2.8200e-04) (hash(x)=166473281)
21
+ 150 train 6.781276 (lr=3.0200e-04) (hash(x)=142185643)
22
+ 160 train 6.792026 (lr=3.2200e-04) (hash(x)=151371145)
23
+ 170 train 6.813921 (lr=3.4200e-04) (hash(x)=151601659)
24
+ 180 train 6.647873 (lr=3.6200e-04) (hash(x)=153368917)
25
+ 190 train 6.502840 (lr=3.8200e-04) (hash(x)=145008196)
26
+ 200 val loss 6.6320
27
+ 200 val perplexity 759.0297
28
+ 200 train 6.526390 (lr=4.0200e-04) (hash(x)=148123706)
29
+ 210 train 6.411649 (lr=4.2200e-04) (hash(x)=145714762)
30
+ 220 train 6.417181 (lr=4.4200e-04) (hash(x)=150622583)
31
+ 230 train 6.194545 (lr=4.6200e-04) (hash(x)=146512648)
32
+ 240 train 6.237221 (lr=4.8200e-04) (hash(x)=151535384)
33
+ 250 train 6.165551 (lr=5.0200e-04) (hash(x)=149952383)
34
+ 260 train 6.074028 (lr=5.2200e-04) (hash(x)=148726088)
35
+ 270 train 6.031117 (lr=5.4200e-04) (hash(x)=152776695)
36
+ 280 train 6.060799 (lr=5.6200e-04) (hash(x)=157260375)
37
+ 290 train 6.122590 (lr=5.8200e-04) (hash(x)=151559106)
38
+ 300 val loss 6.1916
39
+ 300 val perplexity 488.6198
40
+ 300 train 5.910843 (lr=6.0200e-04) (hash(x)=146678221)
41
+ 310 train 6.019495 (lr=6.2200e-04) (hash(x)=156713527)
42
+ 320 train 6.063987 (lr=6.4200e-04) (hash(x)=144947971)
43
+ 330 train 6.241130 (lr=6.6200e-04) (hash(x)=158684944)
44
+ 340 train 6.108393 (lr=6.8200e-04) (hash(x)=162318756)
45
+ 350 train 5.915740 (lr=7.0200e-04) (hash(x)=156180736)
46
+ 360 train 6.084400 (lr=7.2200e-04) (hash(x)=178373797)
47
+ 370 train 5.980911 (lr=7.4200e-04) (hash(x)=154389449)
48
+ 380 train 5.816041 (lr=7.6200e-04) (hash(x)=150369080)
49
+ 390 train 5.955685 (lr=7.8200e-04) (hash(x)=159840654)
50
+ 400 val loss 5.8457
51
+ 400 val perplexity 345.7422
52
+ 400 train 5.763399 (lr=8.0200e-04) (hash(x)=151700982)
53
+ 410 train 5.742797 (lr=8.2200e-04) (hash(x)=160467274)
54
+ 420 train 5.653957 (lr=8.4200e-04) (hash(x)=148930522)
55
+ 430 train 5.862207 (lr=8.6200e-04) (hash(x)=154565303)
56
+ 440 train 5.836684 (lr=8.8200e-04) (hash(x)=148079227)
57
+ 450 train 5.563416 (lr=9.0200e-04) (hash(x)=147288467)
58
+ 460 train 5.586844 (lr=9.2200e-04) (hash(x)=145272617)
59
+ 470 train 5.611871 (lr=9.4200e-04) (hash(x)=123126729)
60
+ 480 train 5.560246 (lr=9.6200e-04) (hash(x)=156287490)
61
+ 490 train 5.689702 (lr=9.8200e-04) (hash(x)=161700145)
62
+ 500 val loss 5.6059
63
+ 500 val perplexity 272.0353
64
+ 500 train 5.713145 (lr=1.0000e-03) (hash(x)=156182087)
65
+ 510 train 5.324427 (lr=1.0000e-03) (hash(x)=131674504)
66
+ 520 train 5.634843 (lr=9.9999e-04) (hash(x)=162758254)
67
+ 530 train 5.441197 (lr=9.9997e-04) (hash(x)=133510409)
68
+ 540 train 5.429710 (lr=9.9995e-04) (hash(x)=157363914)
69
+ 550 train 5.363416 (lr=9.9992e-04) (hash(x)=147861550)
70
+ 560 train 5.315009 (lr=9.9988e-04) (hash(x)=149449280)
71
+ 570 train 5.336292 (lr=9.9984e-04) (hash(x)=143899904)
72
+ 580 train 5.301198 (lr=9.9979e-04) (hash(x)=149266312)
73
+ 590 train 5.539432 (lr=9.9974e-04) (hash(x)=164352774)
74
+ 600 val loss 5.4347
75
+ 600 val perplexity 229.2241
76
+ 600 train 5.300513 (lr=9.9967e-04) (hash(x)=149318660)
77
+ 610 train 5.191154 (lr=9.9961e-04) (hash(x)=146101781)
78
+ 620 train 5.149113 (lr=9.9953e-04) (hash(x)=137184623)
79
+ 630 train 5.039128 (lr=9.9945e-04) (hash(x)=145791873)
80
+ 640 train 5.066300 (lr=9.9936e-04) (hash(x)=143999212)
81
+ 650 train 5.157751 (lr=9.9927e-04) (hash(x)=151631103)
82
+ 660 train 4.931474 (lr=9.9917e-04) (hash(x)=138360108)
83
+ 670 train 5.307727 (lr=9.9906e-04) (hash(x)=156637679)
84
+ 680 train 4.954572 (lr=9.9894e-04) (hash(x)=143752601)
85
+ 690 train 5.094683 (lr=9.9882e-04) (hash(x)=158368903)
86
+ 700 val loss 5.3112
87
+ 700 val perplexity 202.5925
88
+ 700 train 5.304758 (lr=9.9870e-04) (hash(x)=150482428)
89
+ 710 train 5.262589 (lr=9.9856e-04) (hash(x)=158085060)
90
+ 720 train 5.328913 (lr=9.9842e-04) (hash(x)=159461069)
91
+ 730 train 5.161735 (lr=9.9828e-04) (hash(x)=148795562)
92
+ 740 train 5.180926 (lr=9.9812e-04) (hash(x)=149967984)
93
+ 750 train 5.173863 (lr=9.9796e-04) (hash(x)=144458617)
94
+ 760 train 5.210239 (lr=9.9780e-04) (hash(x)=155106202)
95
+ 770 train 5.002542 (lr=9.9762e-04) (hash(x)=144506191)
96
+ 780 train 5.157306 (lr=9.9744e-04) (hash(x)=155825158)
97
+ 790 train 5.201035 (lr=9.9726e-04) (hash(x)=158835381)
98
+ 800 val loss 5.1554
99
+ 800 val perplexity 173.3728
100
+ 800 train 5.105727 (lr=9.9707e-04) (hash(x)=143268605)
101
+ 810 train 5.024149 (lr=9.9687e-04) (hash(x)=139072755)
102
+ 820 train 5.148905 (lr=9.9666e-04) (hash(x)=179384792)
103
+ 830 train 4.933615 (lr=9.9645e-04) (hash(x)=151086962)
104
+ 840 train 5.559482 (lr=9.9623e-04) (hash(x)=142711393)
105
+ 850 train 5.025608 (lr=9.9601e-04) (hash(x)=150817011)
106
+ 860 train 4.964322 (lr=9.9578e-04) (hash(x)=138495801)
107
+ 870 train 4.974591 (lr=9.9554e-04) (hash(x)=155474016)
108
+ 880 train 5.003127 (lr=9.9530e-04) (hash(x)=151951501)
109
+ 890 train 4.869858 (lr=9.9505e-04) (hash(x)=143699524)
110
+ 900 val loss 5.0297
111
+ 900 val perplexity 152.8807
112
+ 900 train 5.016776 (lr=9.9479e-04) (hash(x)=152322423)
113
+ 910 train 4.951572 (lr=9.9453e-04) (hash(x)=151010509)
114
+ 920 train 4.753690 (lr=9.9426e-04) (hash(x)=148720342)
115
+ 930 train 4.713594 (lr=9.9398e-04) (hash(x)=125869480)
116
+ 940 train 4.951263 (lr=9.9370e-04) (hash(x)=150349503)
117
+ 950 train 4.728230 (lr=9.9341e-04) (hash(x)=143574126)
118
+ 960 train 4.922353 (lr=9.9311e-04) (hash(x)=146292562)
119
+ 970 train 4.855898 (lr=9.9281e-04) (hash(x)=162256596)
120
+ 980 train 4.810009 (lr=9.9250e-04) (hash(x)=137640259)
121
+ 990 train 4.673797 (lr=9.9219e-04) (hash(x)=141767794)
122
+ 1000 val loss 4.9489
123
+ 1000 val perplexity 141.0215
124
+ 1000 train 4.894795 (lr=9.9187e-04) (hash(x)=147904298)
125
+ 1010 train 4.829123 (lr=9.9154e-04) (hash(x)=162239186)
126
+ 1020 train 4.587228 (lr=9.9121e-04) (hash(x)=140706537)
127
+ 1030 train 4.707352 (lr=9.9087e-04) (hash(x)=155556720)
128
+ 1040 train 4.595198 (lr=9.9052e-04) (hash(x)=144746445)
129
+ 1050 train 4.649117 (lr=9.9017e-04) (hash(x)=144719400)
130
+ 1060 train 4.510623 (lr=9.8981e-04) (hash(x)=153065665)
131
+ 1070 train 4.635851 (lr=9.8944e-04) (hash(x)=136468722)
132
+ 1080 train 4.608644 (lr=9.8907e-04) (hash(x)=157674756)
133
+ 1090 train 4.799476 (lr=9.8869e-04) (hash(x)=155882884)
134
+ 1100 val loss 4.8567
135
+ 1100 val perplexity 128.5995
136
+ 1100 train 5.071962 (lr=9.8831e-04) (hash(x)=154343147)
137
+ 1110 train 4.858743 (lr=9.8791e-04) (hash(x)=158582939)
138
+ 1120 train 4.673484 (lr=9.8752e-04) (hash(x)=162479784)
139
+ 1130 train 4.853344 (lr=9.8711e-04) (hash(x)=155704345)
140
+ 1140 train 4.745828 (lr=9.8670e-04) (hash(x)=146934299)
141
+ 1150 train 4.696046 (lr=9.8629e-04) (hash(x)=154372635)
142
+ 1160 train 4.687279 (lr=9.8586e-04) (hash(x)=145536832)
143
+ 1170 train 4.970385 (lr=9.8543e-04) (hash(x)=158785901)
144
+ 1180 train 4.760361 (lr=9.8500e-04) (hash(x)=158271614)
145
+ 1190 train 4.820662 (lr=9.8456e-04) (hash(x)=157365589)
146
+ 1200 val loss 4.7860
147
+ 1200 val perplexity 119.8177
148
+ 1200 train 4.847801 (lr=9.8411e-04) (hash(x)=141843115)
149
+ 1210 train 4.739888 (lr=9.8365e-04) (hash(x)=146807973)
150
+ 1220 train 4.718340 (lr=9.8319e-04) (hash(x)=144149947)
151
+ 1230 train 4.649802 (lr=9.8272e-04) (hash(x)=147869063)
152
+ 1240 train 4.748072 (lr=9.8225e-04) (hash(x)=152217389)
153
+ 1250 train 4.747263 (lr=9.8177e-04) (hash(x)=149417679)
154
+ 1260 train 4.770257 (lr=9.8129e-04) (hash(x)=144029489)
155
+ 1270 train 4.879799 (lr=9.8079e-04) (hash(x)=154328303)
156
+ 1280 train 4.746103 (lr=9.8030e-04) (hash(x)=151625838)
157
+ 1290 train 4.604917 (lr=9.7979e-04) (hash(x)=140270541)
158
+ 1300 val loss 4.7262
159
+ 1300 val perplexity 112.8686
160
+ 1300 train 4.661518 (lr=9.7928e-04) (hash(x)=145279030)
161
+ 1310 train 4.689168 (lr=9.7876e-04) (hash(x)=140279630)
162
+ 1320 train 4.417997 (lr=9.7824e-04) (hash(x)=133642994)
163
+ 1330 train 4.589182 (lr=9.7771e-04) (hash(x)=146634776)
164
+ 1340 train 4.691030 (lr=9.7717e-04) (hash(x)=155919973)
165
+ 1350 train 4.716493 (lr=9.7663e-04) (hash(x)=146363911)
166
+ 1360 train 4.596271 (lr=9.7608e-04) (hash(x)=150613636)
167
+ 1370 train 4.586654 (lr=9.7553e-04) (hash(x)=148536643)
168
+ 1380 train 4.594614 (lr=9.7497e-04) (hash(x)=152803906)
169
+ 1390 train 4.521398 (lr=9.7440e-04) (hash(x)=150383305)
170
+ 1400 val loss 4.7125
171
+ 1400 val perplexity 111.3253
172
+ 1400 train 4.619091 (lr=9.7383e-04) (hash(x)=152507639)
173
+ 1410 train 4.371135 (lr=9.7325e-04) (hash(x)=161841408)
174
+ 1420 train 4.495814 (lr=9.7267e-04) (hash(x)=167824828)
175
+ 1430 train 4.454244 (lr=9.7207e-04) (hash(x)=147696374)
176
+ 1440 train 4.415960 (lr=9.7148e-04) (hash(x)=150085494)
177
+ 1450 train 4.370377 (lr=9.7087e-04) (hash(x)=139296191)
178
+ 1460 train 4.286629 (lr=9.7026e-04) (hash(x)=142956546)
179
+ 1470 train 4.353026 (lr=9.6965e-04) (hash(x)=155908628)
180
+ 1480 train 4.345649 (lr=9.6903e-04) (hash(x)=150893345)
181
+ 1490 train 4.364626 (lr=9.6840e-04) (hash(x)=143548902)
182
+ 1500 val loss 4.6534
183
+ 1500 val perplexity 104.9430
184
+ 1500 train 4.605448 (lr=9.6777e-04) (hash(x)=148473774)
185
+ 1510 train 4.693806 (lr=9.6713e-04) (hash(x)=161631000)
186
+ 1520 train 4.620327 (lr=9.6648e-04) (hash(x)=166217855)
187
+ 1530 train 4.580425 (lr=9.6583e-04) (hash(x)=149440046)
188
+ 1540 train 4.528459 (lr=9.6517e-04) (hash(x)=155013278)
189
+ 1550 train 4.551861 (lr=9.6451e-04) (hash(x)=154130527)
190
+ 1560 train 4.584275 (lr=9.6384e-04) (hash(x)=152138735)
191
+ 1570 train 4.493083 (lr=9.6316e-04) (hash(x)=148128305)
192
+ 1580 train 4.622113 (lr=9.6248e-04) (hash(x)=157176546)
193
+ 1590 train 4.739692 (lr=9.6179e-04) (hash(x)=152935467)
194
+ 1600 val loss 4.5857
195
+ 1600 val perplexity 98.0686
196
+ 1600 train 4.558611 (lr=9.6110e-04) (hash(x)=151117002)
197
+ 1610 train 4.556645 (lr=9.6040e-04) (hash(x)=149883384)
198
+ 1620 train 4.602330 (lr=9.5969e-04) (hash(x)=145015354)
199
+ 1630 train 4.536306 (lr=9.5898e-04) (hash(x)=146717645)
200
+ 1640 train 4.741371 (lr=9.5826e-04) (hash(x)=156372727)
201
+ 1650 train 4.644338 (lr=9.5754e-04) (hash(x)=140261555)
202
+ 1660 train 4.477019 (lr=9.5681e-04) (hash(x)=150088650)
203
+ 1670 train 4.409225 (lr=9.5607e-04) (hash(x)=135943513)
204
+ 1680 train 4.629479 (lr=9.5533e-04) (hash(x)=153529653)
205
+ 1690 train 4.681628 (lr=9.5458e-04) (hash(x)=148930360)
206
+ 1700 val loss 4.5904
207
+ 1700 val perplexity 98.5361
208
+ 1700 train 4.570439 (lr=9.5383e-04) (hash(x)=138011335)
209
+ 1710 train 4.522403 (lr=9.5307e-04) (hash(x)=141383599)
210
+ 1720 train 4.701284 (lr=9.5231e-04) (hash(x)=161449652)
211
+ 1730 train 4.498432 (lr=9.5153e-04) (hash(x)=153923561)
212
+ 1740 train 4.437665 (lr=9.5076e-04) (hash(x)=146043253)
213
+ 1750 train 4.465888 (lr=9.4998e-04) (hash(x)=155720883)
214
+ 1760 train 4.449500 (lr=9.4919e-04) (hash(x)=148222148)
215
+ 1770 train 4.505893 (lr=9.4839e-04) (hash(x)=155712380)
216
+ 1780 train 4.403866 (lr=9.4759e-04) (hash(x)=149767859)
217
+ 1790 train 4.628847 (lr=9.4679e-04) (hash(x)=167214230)
218
+ 1800 val loss 4.5564
219
+ 1800 val perplexity 95.2429
220
+ 1800 train 4.535213 (lr=9.4598e-04) (hash(x)=171180926)
221
+ 1810 train 4.548923 (lr=9.4516e-04) (hash(x)=162375834)
222
+ 1820 train 4.663252 (lr=9.4434e-04) (hash(x)=151495752)
223
+ 1830 train 4.568282 (lr=9.4351e-04) (hash(x)=146532257)
224
+ 1840 train 4.654027 (lr=9.4268e-04) (hash(x)=145694434)
225
+ 1850 train 4.539825 (lr=9.4184e-04) (hash(x)=144823488)
226
+ 1860 train 4.639201 (lr=9.4099e-04) (hash(x)=149836286)
227
+ 1870 train 4.603424 (lr=9.4014e-04) (hash(x)=143935049)
228
+ 1880 train 4.634578 (lr=9.3928e-04) (hash(x)=151786371)
229
+ 1890 train 4.520654 (lr=9.3842e-04) (hash(x)=152892567)
230
+ 1900 val loss 4.5127
231
+ 1900 val perplexity 91.1715
232
+ 1900 train 4.560280 (lr=9.3755e-04) (hash(x)=141769419)
233
+ 1910 train 4.579801 (lr=9.3668e-04) (hash(x)=156914416)
234
+ 1920 train 4.434911 (lr=9.3580e-04) (hash(x)=151573900)
235
+ 1930 train 4.528929 (lr=9.3491e-04) (hash(x)=145121701)
236
+ 1940 train 4.467929 (lr=9.3402e-04) (hash(x)=150896348)
237
+ 1950 train 4.576292 (lr=9.3313e-04) (hash(x)=168797189)
238
+ 1960 train 4.457506 (lr=9.3223e-04) (hash(x)=146711430)
239
+ 1970 train 4.306789 (lr=9.3132e-04) (hash(x)=152180946)
240
+ 1980 train 4.429000 (lr=9.3041e-04) (hash(x)=155542268)
241
+ 1990 train 4.402015 (lr=9.2949e-04) (hash(x)=144668660)
242
+ 2000 val loss 4.4776
243
+ 2000 val perplexity 88.0232
244
+ 2000 train 4.471121 (lr=9.2856e-04) (hash(x)=151963443)
245
+ 2010 train 4.462688 (lr=9.2763e-04) (hash(x)=150810352)
246
+ 2020 train 4.476783 (lr=9.2670e-04) (hash(x)=157539832)
247
+ 2030 train 4.685014 (lr=9.2576e-04) (hash(x)=137858786)
248
+ 2040 train 4.424544 (lr=9.2481e-04) (hash(x)=154270900)
249
+ 2050 train 4.281589 (lr=9.2386e-04) (hash(x)=156252579)
250
+ 2060 train 4.213714 (lr=9.2291e-04) (hash(x)=143285549)
251
+ 2070 train 4.431551 (lr=9.2195e-04) (hash(x)=150542438)
252
+ 2080 train 4.336223 (lr=9.2098e-04) (hash(x)=146290330)
253
+ 2090 train 4.430391 (lr=9.2001e-04) (hash(x)=146486756)
254
+ 2100 val loss 4.5100
255
+ 2100 val perplexity 90.9222
256
+ 2100 train 4.444623 (lr=9.1903e-04) (hash(x)=162947470)
257
+ 2110 train 4.343028 (lr=9.1804e-04) (hash(x)=150305391)
258
+ 2120 train 4.503345 (lr=9.1706e-04) (hash(x)=156452783)
259
+ 2130 train 4.471503 (lr=9.1606e-04) (hash(x)=167481875)
260
+ 2140 train 4.451882 (lr=9.1506e-04) (hash(x)=141782256)
261
+ 2150 train 4.479333 (lr=9.1406e-04) (hash(x)=145035419)
262
+ 2160 train 4.539510 (lr=9.1305e-04) (hash(x)=142000400)
263
+ 2170 train 4.499580 (lr=9.1203e-04) (hash(x)=144229380)
264
+ 2180 train 4.538344 (lr=9.1101e-04) (hash(x)=166204176)
265
+ 2190 train 4.466153 (lr=9.0999e-04) (hash(x)=150712057)
266
+ 2200 val loss 4.4407
267
+ 2200 val perplexity 84.8354
268
+ 2200 train 4.350904 (lr=9.0896e-04) (hash(x)=154954810)
269
+ 2210 train 4.556060 (lr=9.0792e-04) (hash(x)=155668760)
270
+ 2220 train 4.387053 (lr=9.0688e-04) (hash(x)=144834708)
271
+ 2230 train 4.372379 (lr=9.0583e-04) (hash(x)=159744566)
272
+ 2240 train 4.333399 (lr=9.0478e-04) (hash(x)=158543377)
273
+ 2250 train 4.347370 (lr=9.0372e-04) (hash(x)=151004281)
274
+ 2260 train 4.452578 (lr=9.0266e-04) (hash(x)=151242482)
275
+ 2270 train 4.407653 (lr=9.0160e-04) (hash(x)=152056559)
276
+ 2280 train 4.338912 (lr=9.0052e-04) (hash(x)=148864076)
277
+ 2290 train 4.462305 (lr=8.9945e-04) (hash(x)=144294586)
278
+ 2300 val loss 4.4286
279
+ 2300 val perplexity 83.8182
280
+ 2300 train 4.360404 (lr=8.9836e-04) (hash(x)=151878111)
281
+ 2310 train 4.373468 (lr=8.9728e-04) (hash(x)=154139457)
282
+ 2320 train 4.304017 (lr=8.9618e-04) (hash(x)=155571592)
283
+ 2330 train 4.348658 (lr=8.9509e-04) (hash(x)=154823764)
284
+ 2340 train 4.401985 (lr=8.9398e-04) (hash(x)=153252177)
285
+ 2350 train 4.333295 (lr=8.9288e-04) (hash(x)=136051830)
286
+ 2360 train 4.380826 (lr=8.9177e-04) (hash(x)=153814175)
287
+ 2370 train 4.379656 (lr=8.9065e-04) (hash(x)=146391078)
288
+ 2380 train 4.344494 (lr=8.8953e-04) (hash(x)=150782698)
289
+ 2390 train 4.415199 (lr=8.8840e-04) (hash(x)=142752256)
290
+ 2400 val loss 4.4453
291
+ 2400 val perplexity 85.2293
292
+ 2400 train 4.326324 (lr=8.8727e-04) (hash(x)=158661057)
293
+ 2410 train 4.400020 (lr=8.8613e-04) (hash(x)=153611303)
294
+ 2420 train 4.565821 (lr=8.8499e-04) (hash(x)=159508684)
295
+ 2430 train 4.362583 (lr=8.8384e-04) (hash(x)=149501192)
296
+ 2440 train 4.028908 (lr=8.8269e-04) (hash(x)=181177177)
297
+ 2450 train 4.490774 (lr=8.8153e-04) (hash(x)=163664732)
298
+ 2460 train 4.172308 (lr=8.8037e-04) (hash(x)=194527955)
299
+ 2470 train 4.407909 (lr=8.7921e-04) (hash(x)=158088521)
300
+ 2480 train 4.361482 (lr=8.7804e-04) (hash(x)=152865678)
301
+ 2490 train 4.341998 (lr=8.7686e-04) (hash(x)=165703871)
302
+ 2500 val loss 4.3880
303
+ 2500 val perplexity 80.4791
304
+ 2500 train 4.437636 (lr=8.7568e-04) (hash(x)=150925584)
305
+ 2510 train 4.341842 (lr=8.7450e-04) (hash(x)=144406177)
306
+ 2520 train 4.242796 (lr=8.7331e-04) (hash(x)=120133043)
307
+ 2530 train 4.428564 (lr=8.7211e-04) (hash(x)=167900247)
308
+ 2540 train 4.319814 (lr=8.7091e-04) (hash(x)=162599203)
309
+ 2550 train 4.384484 (lr=8.6971e-04) (hash(x)=149328180)
310
+ 2560 train 4.279142 (lr=8.6850e-04) (hash(x)=150049968)
311
+ 2570 train 4.414873 (lr=8.6729e-04) (hash(x)=162876159)
312
+ 2580 train 4.240511 (lr=8.6607e-04) (hash(x)=148260649)
313
+ 2590 train 4.378955 (lr=8.6485e-04) (hash(x)=148998018)
314
+ 2600 val loss 4.3753
315
+ 2600 val perplexity 79.4674
316
+ 2600 train 4.295646 (lr=8.6362e-04) (hash(x)=144515755)
317
+ 2610 train 4.407104 (lr=8.6239e-04) (hash(x)=154455224)
318
+ 2620 train 4.211033 (lr=8.6116e-04) (hash(x)=142920048)
319
+ 2630 train 4.371943 (lr=8.5991e-04) (hash(x)=143931236)
320
+ 2640 train 4.218086 (lr=8.5867e-04) (hash(x)=150983323)
321
+ 2650 train 4.323412 (lr=8.5742e-04) (hash(x)=139360405)
322
+ 2660 train 4.238265 (lr=8.5617e-04) (hash(x)=148850470)
323
+ 2670 train 4.488311 (lr=8.5491e-04) (hash(x)=149929606)
324
+ 2680 train 4.462916 (lr=8.5365e-04) (hash(x)=151673000)
325
+ 2690 train 4.339820 (lr=8.5238e-04) (hash(x)=140631266)
326
+ 2700 val loss 4.3727
327
+ 2700 val perplexity 79.2608
328
+ 2700 train 4.316209 (lr=8.5111e-04) (hash(x)=153109144)
329
+ 2710 train 4.193799 (lr=8.4983e-04) (hash(x)=135332836)
330
+ 2720 train 4.560100 (lr=8.4855e-04) (hash(x)=131611182)
331
+ 2730 train 4.389001 (lr=8.4727e-04) (hash(x)=142909388)
332
+ 2740 train 4.416577 (lr=8.4598e-04) (hash(x)=145593419)
333
+ 2750 train 4.301349 (lr=8.4469e-04) (hash(x)=158225769)
334
+ 2760 train 4.282790 (lr=8.4339e-04) (hash(x)=170509938)
335
+ 2770 train 4.296576 (lr=8.4209e-04) (hash(x)=150785805)
336
+ 2780 train 4.239909 (lr=8.4078e-04) (hash(x)=167281736)
337
+ 2790 train 4.426344 (lr=8.3947e-04) (hash(x)=161660187)
338
+ 2800 val loss 4.3486
339
+ 2800 val perplexity 77.3725
340
+ 2800 train 4.388307 (lr=8.3816e-04) (hash(x)=151152897)
341
+ 2810 train 4.354566 (lr=8.3684e-04) (hash(x)=143172067)
342
+ 2820 train 4.360189 (lr=8.3552e-04) (hash(x)=141569169)
343
+ 2830 train 4.238786 (lr=8.3419e-04) (hash(x)=159709681)
344
+ 2840 train 4.297164 (lr=8.3286e-04) (hash(x)=162552294)
345
+ 2850 train 4.420930 (lr=8.3153e-04) (hash(x)=146762690)
346
+ 2860 train 4.291400 (lr=8.3019e-04) (hash(x)=145729422)
347
+ 2870 train 4.368289 (lr=8.2884e-04) (hash(x)=144360628)
348
+ 2880 train 4.180243 (lr=8.2750e-04) (hash(x)=147737098)
349
+ 2890 train 4.290162 (lr=8.2615e-04) (hash(x)=146849057)
350
+ 2900 val loss 4.3312
351
+ 2900 val perplexity 76.0368
352
+ 2900 train 4.356330 (lr=8.2479e-04) (hash(x)=145800210)
353
+ 2910 train 4.157279 (lr=8.2343e-04) (hash(x)=147316313)
354
+ 2920 train 4.212349 (lr=8.2207e-04) (hash(x)=152595415)
355
+ 2930 train 4.374985 (lr=8.2070e-04) (hash(x)=156542647)
356
+ 2940 train 4.230096 (lr=8.1933e-04) (hash(x)=141183896)
357
+ 2950 train 4.326154 (lr=8.1796e-04) (hash(x)=156329036)
358
+ 2960 train 4.371807 (lr=8.1658e-04) (hash(x)=156717793)
359
+ 2970 train 4.235094 (lr=8.1520e-04) (hash(x)=143949931)
360
+ 2980 train 4.295004 (lr=8.1381e-04) (hash(x)=155067290)
361
+ 2990 train 4.271433 (lr=8.1242e-04) (hash(x)=148774729)
362
+ 3000 val loss 4.3486
363
+ 3000 val perplexity 77.3713
364
+ 3000 train 4.255401 (lr=8.1103e-04) (hash(x)=141997485)
365
+ 3010 train 4.226521 (lr=8.0963e-04) (hash(x)=160762953)
366
+ 3020 train 4.178299 (lr=8.0823e-04) (hash(x)=170341665)
367
+ 3030 train 4.448976 (lr=8.0682e-04) (hash(x)=150722836)
368
+ 3040 train 4.251589 (lr=8.0541e-04) (hash(x)=154243307)
369
+ 3050 train 4.290366 (lr=8.0400e-04) (hash(x)=146077234)
370
+ 3060 train 4.247835 (lr=8.0258e-04) (hash(x)=152125542)
371
+ 3070 train 4.425954 (lr=8.0116e-04) (hash(x)=144149387)
372
+ 3080 train 4.249808 (lr=7.9974e-04) (hash(x)=147508712)
373
+ 3090 train 4.245219 (lr=7.9831e-04) (hash(x)=143583369)
374
+ 3100 val loss 4.3018
375
+ 3100 val perplexity 73.8333
376
+ 3100 train 4.277176 (lr=7.9688e-04) (hash(x)=154049740)
377
+ 3110 train 4.332026 (lr=7.9545e-04) (hash(x)=147054835)
378
+ 3120 train 4.321158 (lr=7.9401e-04) (hash(x)=152468568)
379
+ 3130 train 4.269117 (lr=7.9257e-04) (hash(x)=151635194)
380
+ 3140 train 4.232862 (lr=7.9112e-04) (hash(x)=150396274)
381
+ 3150 train 4.293002 (lr=7.8967e-04) (hash(x)=141306267)
382
+ 3160 train 4.233846 (lr=7.8822e-04) (hash(x)=151110762)
383
+ 3170 train 4.314668 (lr=7.8677e-04) (hash(x)=146518875)
384
+ 3180 train 4.193670 (lr=7.8531e-04) (hash(x)=139914401)
385
+ 3190 train 4.204021 (lr=7.8384e-04) (hash(x)=161631880)
386
+ 3200 val loss 4.3186
387
+ 3200 val perplexity 75.0807
388
+ 3200 train 4.234648 (lr=7.8238e-04) (hash(x)=150471842)
389
+ 3210 train 4.292208 (lr=7.8091e-04) (hash(x)=147355138)
390
+ 3220 train 4.092046 (lr=7.7944e-04) (hash(x)=138879678)
391
+ 3230 train 4.186282 (lr=7.7796e-04) (hash(x)=123929679)
392
+ 3240 train 4.252555 (lr=7.7648e-04) (hash(x)=157263734)
393
+ 3250 train 4.279626 (lr=7.7500e-04) (hash(x)=148065539)
394
+ 3260 train 4.193232 (lr=7.7351e-04) (hash(x)=138020742)
395
+ 3270 train 4.056885 (lr=7.7203e-04) (hash(x)=142053611)
396
+ 3280 train 4.216621 (lr=7.7053e-04) (hash(x)=155444784)
397
+ 3290 train 4.045831 (lr=7.6904e-04) (hash(x)=148500315)
398
+ 3300 val loss 4.3095
399
+ 3300 val perplexity 74.3999
400
+ 3300 train 4.276546 (lr=7.6754e-04) (hash(x)=149048126)
401
+ 3310 train 4.225118 (lr=7.6604e-04) (hash(x)=141680674)
402
+ 3320 train 4.163364 (lr=7.6453e-04) (hash(x)=143310642)
403
+ 3330 train 4.281623 (lr=7.6303e-04) (hash(x)=149891496)
404
+ 3340 train 4.224004 (lr=7.6151e-04) (hash(x)=151244609)
405
+ 3350 train 4.233478 (lr=7.6000e-04) (hash(x)=157158479)
406
+ 3360 train 4.099353 (lr=7.5848e-04) (hash(x)=151709876)
407
+ 3370 train 4.244702 (lr=7.5696e-04) (hash(x)=145297090)
408
+ 3380 train 4.266629 (lr=7.5544e-04) (hash(x)=154767881)
409
+ 3390 train 4.237964 (lr=7.5391e-04) (hash(x)=152600366)
410
+ 3400 val loss 4.2854
411
+ 3400 val perplexity 72.6321
412
+ 3400 train 4.468420 (lr=7.5238e-04) (hash(x)=161261339)
413
+ 3410 train 4.223356 (lr=7.5085e-04) (hash(x)=146859162)
414
+ 3420 train 4.321335 (lr=7.4932e-04) (hash(x)=149276224)
415
+ 3430 train 4.335219 (lr=7.4778e-04) (hash(x)=155866933)
416
+ 3440 train 4.199460 (lr=7.4624e-04) (hash(x)=143191904)
417
+ 3450 train 4.418814 (lr=7.4470e-04) (hash(x)=156910661)
418
+ 3460 train 4.210780 (lr=7.4315e-04) (hash(x)=140850879)
419
+ 3470 train 4.429366 (lr=7.4160e-04) (hash(x)=162140445)
420
+ 3480 train 4.176719 (lr=7.4005e-04) (hash(x)=145607495)
421
+ 3490 train 4.186519 (lr=7.3849e-04) (hash(x)=152496121)
422
+ 3500 val loss 4.2784
423
+ 3500 val perplexity 72.1230
424
+ 3500 train 4.154493 (lr=7.3694e-04) (hash(x)=157495564)
425
+ 3510 train 4.277784 (lr=7.3538e-04) (hash(x)=147657004)
426
+ 3520 train 4.247444 (lr=7.3381e-04) (hash(x)=152978060)
427
+ 3530 train 4.184846 (lr=7.3225e-04) (hash(x)=146800180)
428
+ 3540 train 4.257003 (lr=7.3068e-04) (hash(x)=195450444)
429
+ 3550 train 4.246735 (lr=7.2911e-04) (hash(x)=151114129)
430
+ 3560 train 4.221786 (lr=7.2754e-04) (hash(x)=143833190)
431
+ 3570 train 4.031631 (lr=7.2596e-04) (hash(x)=146039971)
432
+ 3580 train 4.094244 (lr=7.2438e-04) (hash(x)=145172249)
433
+ 3590 train 4.014417 (lr=7.2280e-04) (hash(x)=159164125)
434
+ 3600 val loss 4.2801
435
+ 3600 val perplexity 72.2507
436
+ 3600 train 4.260633 (lr=7.2122e-04) (hash(x)=144352932)
437
+ 3610 train 4.219941 (lr=7.1963e-04) (hash(x)=150550670)
438
+ 3620 train 4.349569 (lr=7.1804e-04) (hash(x)=147170125)
439
+ 3630 train 4.147399 (lr=7.1645e-04) (hash(x)=139473935)
440
+ 3640 train 4.138421 (lr=7.1486e-04) (hash(x)=159294424)
441
+ 3650 train 4.264631 (lr=7.1326e-04) (hash(x)=145943278)
442
+ 3660 train 4.317642 (lr=7.1167e-04) (hash(x)=141647872)
443
+ 3670 train 4.288270 (lr=7.1007e-04) (hash(x)=147109351)
444
+ 3680 train 4.315784 (lr=7.0846e-04) (hash(x)=140683368)
445
+ 3690 train 4.227759 (lr=7.0686e-04) (hash(x)=156276284)
446
+ 3700 val loss 4.2595
447
+ 3700 val perplexity 70.7721
448
+ 3700 train 4.169581 (lr=7.0525e-04) (hash(x)=149389012)
449
+ 3710 train 4.305984 (lr=7.0364e-04) (hash(x)=142107929)
450
+ 3720 train 4.278129 (lr=7.0203e-04) (hash(x)=147828843)
451
+ 3730 train 4.345847 (lr=7.0041e-04) (hash(x)=141784679)
452
+ 3740 train 4.246113 (lr=6.9880e-04) (hash(x)=153375464)
453
+ 3750 train 4.180670 (lr=6.9718e-04) (hash(x)=153716001)
454
+ 3760 train 4.151164 (lr=6.9556e-04) (hash(x)=146848822)
455
+ 3770 train 4.106029 (lr=6.9394e-04) (hash(x)=148396782)
456
+ 3780 train 4.175367 (lr=6.9231e-04) (hash(x)=153563244)
457
+ 3790 train 4.089706 (lr=6.9069e-04) (hash(x)=160808734)
458
+ 3800 val loss 4.2536
459
+ 3800 val perplexity 70.3566
460
+ 3800 train 4.278991 (lr=6.8906e-04) (hash(x)=146607620)
461
+ 3810 train 4.245387 (lr=6.8743e-04) (hash(x)=152580677)
462
+ 3820 train 4.593886 (lr=6.8579e-04) (hash(x)=144860879)
463
+ 3830 train 4.261201 (lr=6.8416e-04) (hash(x)=162099851)
464
+ 3840 train 4.222744 (lr=6.8252e-04) (hash(x)=159506855)
465
+ 3850 train 4.146121 (lr=6.8088e-04) (hash(x)=140022686)
466
+ 3860 train 4.216559 (lr=6.7924e-04) (hash(x)=152137850)
467
+ 3870 train 4.335166 (lr=6.7760e-04) (hash(x)=142666785)
468
+ 3880 train 4.191841 (lr=6.7596e-04) (hash(x)=148285868)
469
+ 3890 train 4.148168 (lr=6.7431e-04) (hash(x)=147910067)
470
+ 3900 val loss 4.2526
471
+ 3900 val perplexity 70.2888
472
+ 3900 train 4.186025 (lr=6.7266e-04) (hash(x)=143754617)
473
+ 3910 train 4.177444 (lr=6.7101e-04) (hash(x)=154779949)
474
+ 3920 train 4.016380 (lr=6.6936e-04) (hash(x)=137824061)
475
+ 3930 train 4.003213 (lr=6.6771e-04) (hash(x)=136933373)
476
+ 3940 train 4.077132 (lr=6.6605e-04) (hash(x)=164054337)
477
+ 3950 train 4.247558 (lr=6.6440e-04) (hash(x)=155752984)
478
+ 3960 train 4.388119 (lr=6.6274e-04) (hash(x)=160957673)
479
+ 3970 train 4.258029 (lr=6.6108e-04) (hash(x)=157158690)
480
+ 3980 train 4.255326 (lr=6.5942e-04) (hash(x)=153282027)
481
+ 3990 train 4.288346 (lr=6.5776e-04) (hash(x)=152183510)
482
+ 4000 val loss 4.2270
483
+ 4000 val perplexity 68.5134
484
+ 4000 train 4.180590 (lr=6.5609e-04) (hash(x)=156930722)
485
+ 4010 train 4.172881 (lr=6.5443e-04) (hash(x)=150798740)
486
+ 4020 train 4.375155 (lr=6.5276e-04) (hash(x)=142632400)
487
+ 4030 train 4.284159 (lr=6.5109e-04) (hash(x)=173326287)
488
+ 4040 train 4.190619 (lr=6.4942e-04) (hash(x)=128797443)
489
+ 4050 train 4.174573 (lr=6.4775e-04) (hash(x)=146937792)
490
+ 4060 train 4.276099 (lr=6.4607e-04) (hash(x)=164579030)
491
+ 4070 train 4.145974 (lr=6.4440e-04) (hash(x)=149459820)
492
+ 4080 train 4.116166 (lr=6.4272e-04) (hash(x)=160159595)
493
+ 4090 train 4.056788 (lr=6.4104e-04) (hash(x)=141507853)
494
+ 4100 val loss 4.2242
495
+ 4100 val perplexity 68.3175
496
+ 4100 train 4.194366 (lr=6.3937e-04) (hash(x)=147500519)
497
+ 4110 train 4.166648 (lr=6.3769e-04) (hash(x)=153297081)
498
+ 4120 train 4.184754 (lr=6.3600e-04) (hash(x)=142560288)
499
+ 4130 train 4.195542 (lr=6.3432e-04) (hash(x)=153380118)
500
+ 4140 train 4.094168 (lr=6.3264e-04) (hash(x)=143150720)
501
+ 4150 train 4.140769 (lr=6.3095e-04) (hash(x)=143459564)
502
+ 4160 train 4.170622 (lr=6.2927e-04) (hash(x)=153955741)
503
+ 4170 train 4.278394 (lr=6.2758e-04) (hash(x)=152199253)
504
+ 4180 train 4.050327 (lr=6.2589e-04) (hash(x)=165957668)
505
+ 4190 train 4.083230 (lr=6.2420e-04) (hash(x)=144866923)
506
+ 4200 val loss 4.2319
507
+ 4200 val perplexity 68.8452
508
+ 4200 train 4.104544 (lr=6.2251e-04) (hash(x)=143232237)
509
+ 4210 train 4.010940 (lr=6.2082e-04) (hash(x)=151907163)
510
+ 4220 train 4.124260 (lr=6.1913e-04) (hash(x)=145917587)
511
+ 4230 train 3.952652 (lr=6.1743e-04) (hash(x)=147258846)
512
+ 4240 train 4.105657 (lr=6.1574e-04) (hash(x)=143449843)
513
+ 4250 train 4.060256 (lr=6.1404e-04) (hash(x)=184883715)
514
+ 4260 train 4.169207 (lr=6.1235e-04) (hash(x)=156958761)
515
+ 4270 train 4.229846 (lr=6.1065e-04) (hash(x)=168870544)
516
+ 4280 train 4.167910 (lr=6.0895e-04) (hash(x)=179446735)
517
+ 4290 train 4.208668 (lr=6.0725e-04) (hash(x)=145649386)
518
+ 4300 val loss 4.2156
519
+ 4300 val perplexity 67.7360
520
+ 4300 train 4.049536 (lr=6.0555e-04) (hash(x)=146811670)
521
+ 4310 train 4.318284 (lr=6.0385e-04) (hash(x)=151907075)
522
+ 4320 train 4.123010 (lr=6.0215e-04) (hash(x)=138628365)
523
+ 4330 train 4.179041 (lr=6.0044e-04) (hash(x)=151608240)
524
+ 4340 train 4.113474 (lr=5.9874e-04) (hash(x)=160111319)
525
+ 4350 train 4.262730 (lr=5.9704e-04) (hash(x)=144958652)
526
+ 4360 train 4.048933 (lr=5.9533e-04) (hash(x)=149176222)
527
+ 4370 train 4.167891 (lr=5.9363e-04) (hash(x)=216221371)
528
+ 4380 train 4.053754 (lr=5.9192e-04) (hash(x)=141760648)
529
+ 4390 train 4.141736 (lr=5.9022e-04) (hash(x)=151710283)
530
+ 4400 val loss 4.2072
531
+ 4400 val perplexity 67.1702
532
+ 4400 train 4.262396 (lr=5.8851e-04) (hash(x)=158418746)
533
+ 4410 train 4.116519 (lr=5.8680e-04) (hash(x)=142331301)
534
+ 4420 train 4.200625 (lr=5.8509e-04) (hash(x)=146507180)
535
+ 4430 train 4.288577 (lr=5.8338e-04) (hash(x)=166506937)
536
+ 4440 train 4.235051 (lr=5.8168e-04) (hash(x)=151209447)
537
+ 4450 train 4.189605 (lr=5.7997e-04) (hash(x)=149186822)
538
+ 4460 train 4.179647 (lr=5.7826e-04) (hash(x)=144348412)
539
+ 4470 train 4.061388 (lr=5.7655e-04) (hash(x)=156494284)
540
+ 4480 train 4.096772 (lr=5.7483e-04) (hash(x)=147238744)
541
+ 4490 train 4.044040 (lr=5.7312e-04) (hash(x)=150746185)
542
+ 4500 val loss 4.1989
543
+ 4500 val perplexity 66.6114
544
+ 4500 train 4.132852 (lr=5.7141e-04) (hash(x)=156695778)
545
+ 4510 train 3.977368 (lr=5.6970e-04) (hash(x)=135988958)
546
+ 4520 train 4.147857 (lr=5.6799e-04) (hash(x)=148272110)
547
+ 4530 train 4.066222 (lr=5.6628e-04) (hash(x)=136459794)
548
+ 4540 train 4.100253 (lr=5.6456e-04) (hash(x)=147624954)
549
+ 4550 train 4.101744 (lr=5.6285e-04) (hash(x)=154947199)
550
+ 4560 train 4.159940 (lr=5.6114e-04) (hash(x)=147869166)
551
+ 4570 train 4.005948 (lr=5.5942e-04) (hash(x)=148415463)
552
+ 4580 train 4.176423 (lr=5.5771e-04) (hash(x)=152104066)
553
+ 4590 train 4.076147 (lr=5.5600e-04) (hash(x)=147060040)
554
+ 4600 val loss 4.1945
555
+ 4600 val perplexity 66.3221
556
+ 4600 train 4.146423 (lr=5.5428e-04) (hash(x)=147791497)
557
+ 4610 train 4.336831 (lr=5.5257e-04) (hash(x)=155595556)
558
+ 4620 train 4.195645 (lr=5.5086e-04) (hash(x)=147178858)
559
+ 4630 train 4.240558 (lr=5.4914e-04) (hash(x)=152617112)
560
+ 4640 train 4.242253 (lr=5.4743e-04) (hash(x)=161766869)
561
+ 4650 train 4.226777 (lr=5.4572e-04) (hash(x)=158075561)
562
+ 4660 train 4.267192 (lr=5.4400e-04) (hash(x)=145360002)
563
+ 4670 train 4.145252 (lr=5.4229e-04) (hash(x)=144538791)
564
+ 4680 train 4.180242 (lr=5.4058e-04) (hash(x)=142711273)
565
+ 4690 train 4.336369 (lr=5.3886e-04) (hash(x)=193907362)
566
+ 4700 val loss 4.1767
567
+ 4700 val perplexity 65.1504
568
+ 4700 train 4.147245 (lr=5.3715e-04) (hash(x)=155533088)
569
+ 4710 train 4.107145 (lr=5.3544e-04) (hash(x)=148618528)
570
+ 4720 train 4.205025 (lr=5.3372e-04) (hash(x)=150502624)
571
+ 4730 train 4.084304 (lr=5.3201e-04) (hash(x)=189205847)
572
+ 4740 train 4.141030 (lr=5.3030e-04) (hash(x)=167322800)
573
+ 4750 train 4.257555 (lr=5.2859e-04) (hash(x)=162169095)
574
+ 4760 train 3.963910 (lr=5.2688e-04) (hash(x)=143459469)
575
+ 4770 train 4.046522 (lr=5.2517e-04) (hash(x)=144386593)
576
+ 4780 train 4.002884 (lr=5.2345e-04) (hash(x)=143721184)
577
+ 4790 train 4.080467 (lr=5.2174e-04) (hash(x)=146077222)
578
+ 4800 val loss 4.1806
579
+ 4800 val perplexity 65.4067
580
+ 4800 train 4.048178 (lr=5.2003e-04) (hash(x)=138350044)
581
+ 4810 train 4.061075 (lr=5.1832e-04) (hash(x)=154960112)
582
+ 4820 train 3.965194 (lr=5.1662e-04) (hash(x)=146950685)
583
+ 4830 train 3.966438 (lr=5.1491e-04) (hash(x)=140224554)
584
+ 4840 train 3.966837 (lr=5.1320e-04) (hash(x)=156640495)
585
+ 4850 train 4.174580 (lr=5.1149e-04) (hash(x)=153409150)
586
+ 4860 train 4.092710 (lr=5.0978e-04) (hash(x)=151647700)
587
+ 4870 train 4.488494 (lr=5.0808e-04) (hash(x)=155465603)
588
+ 4880 train 4.327782 (lr=5.0637e-04) (hash(x)=153404400)
589
+ 4890 train 4.213738 (lr=5.0467e-04) (hash(x)=170409281)
590
+ 4900 val loss 4.1646
591
+ 4900 val perplexity 64.3662
592
+ 4900 train 4.082214 (lr=5.0296e-04) (hash(x)=143735284)
593
+ 4910 train 4.218872 (lr=5.0126e-04) (hash(x)=150784329)
594
+ 4920 train 4.109700 (lr=4.9956e-04) (hash(x)=152353574)
595
+ 4930 train 4.063162 (lr=4.9785e-04) (hash(x)=152677008)
596
+ 4940 train 4.131150 (lr=4.9615e-04) (hash(x)=140553861)
597
+ 4950 train 4.227722 (lr=4.9445e-04) (hash(x)=152869181)
598
+ 4960 train 4.180067 (lr=4.9275e-04) (hash(x)=153011079)
599
+ 4970 train 4.124014 (lr=4.9105e-04) (hash(x)=152796617)
600
+ 4980 train 4.247788 (lr=4.8935e-04) (hash(x)=146868417)
601
+ 4990 train 4.129405 (lr=4.8765e-04) (hash(x)=140168188)
602
+ 5000 val loss 4.1535
603
+ 5000 val perplexity 63.6584
604
+ 5000 train 4.106555 (lr=4.8596e-04) (hash(x)=154976463)
605
+ 5010 train 4.165233 (lr=4.8426e-04) (hash(x)=148582690)
606
+ 5020 train 4.170747 (lr=4.8257e-04) (hash(x)=166715464)
607
+ 5030 train 4.138124 (lr=4.8087e-04) (hash(x)=150616397)
608
+ 5040 train 4.192421 (lr=4.7918e-04) (hash(x)=150775548)
609
+ 5050 train 4.169790 (lr=4.7749e-04) (hash(x)=150575367)
610
+ 5060 train 4.093499 (lr=4.7580e-04) (hash(x)=152352082)
611
+ 5070 train 4.132263 (lr=4.7411e-04) (hash(x)=150269272)
612
+ 5080 train 4.189360 (lr=4.7242e-04) (hash(x)=146499361)
613
+ 5090 train 3.883502 (lr=4.7073e-04) (hash(x)=148734454)
614
+ 5100 val loss 4.1571
615
+ 5100 val perplexity 63.8880
616
+ 5100 train 4.059271 (lr=4.6905e-04) (hash(x)=149894982)
617
+ 5110 train 4.108232 (lr=4.6736e-04) (hash(x)=154487751)
618
+ 5120 train 4.061627 (lr=4.6568e-04) (hash(x)=144169582)
619
+ 5130 train 4.036693 (lr=4.6400e-04) (hash(x)=152217630)
620
+ 5140 train 4.055055 (lr=4.6231e-04) (hash(x)=158233831)
621
+ 5150 train 4.126297 (lr=4.6063e-04) (hash(x)=142746426)
622
+ 5160 train 4.116474 (lr=4.5896e-04) (hash(x)=139184458)
623
+ 5170 train 4.203731 (lr=4.5728e-04) (hash(x)=140624857)
624
+ 5180 train 3.927008 (lr=4.5560e-04) (hash(x)=154107423)
625
+ 5190 train 3.860676 (lr=4.5393e-04) (hash(x)=144892610)
626
+ 5200 val loss 4.1560
627
+ 5200 val perplexity 63.8167
628
+ 5200 train 3.815462 (lr=4.5225e-04) (hash(x)=159326689)
629
+ 5210 train 3.981979 (lr=4.5058e-04) (hash(x)=151186017)
630
+ 5220 train 3.793216 (lr=4.4891e-04) (hash(x)=148996819)
631
+ 5230 train 3.739858 (lr=4.4724e-04) (hash(x)=150829917)
632
+ 5240 train 3.859729 (lr=4.4557e-04) (hash(x)=152758972)
633
+ 5250 train 3.819778 (lr=4.4391e-04) (hash(x)=154345903)
634
+ 5260 train 3.815678 (lr=4.4224e-04) (hash(x)=150134947)
635
+ 5270 train 3.912570 (lr=4.4058e-04) (hash(x)=149689310)
636
+ 5280 train 4.152149 (lr=4.3892e-04) (hash(x)=147662223)
637
+ 5290 train 4.207056 (lr=4.3726e-04) (hash(x)=165770608)
638
+ 5300 val loss 4.1458
639
+ 5300 val perplexity 63.1700
640
+ 5300 train 4.195957 (lr=4.3560e-04) (hash(x)=159484800)
641
+ 5310 train 4.226388 (lr=4.3395e-04) (hash(x)=152672536)
642
+ 5320 train 4.051078 (lr=4.3229e-04) (hash(x)=152728812)
643
+ 5330 train 3.997982 (lr=4.3064e-04) (hash(x)=152778685)
644
+ 5340 train 4.089240 (lr=4.2899e-04) (hash(x)=149818849)
645
+ 5350 train 4.021405 (lr=4.2734e-04) (hash(x)=168239739)
646
+ 5360 train 4.165332 (lr=4.2569e-04) (hash(x)=139859423)
647
+ 5370 train 3.985575 (lr=4.2404e-04) (hash(x)=139315327)
648
+ 5380 train 4.198783 (lr=4.2240e-04) (hash(x)=163068509)
649
+ 5390 train 4.158439 (lr=4.2076e-04) (hash(x)=147645415)
650
+ 5400 val loss 4.1242
651
+ 5400 val perplexity 61.8161
652
+ 5400 train 4.111883 (lr=4.1912e-04) (hash(x)=140385615)
653
+ 5410 train 3.975249 (lr=4.1748e-04) (hash(x)=151381862)
654
+ 5420 train 4.125394 (lr=4.1584e-04) (hash(x)=152445092)
655
+ 5430 train 4.139113 (lr=4.1421e-04) (hash(x)=142454479)
656
+ 5440 train 4.094466 (lr=4.1257e-04) (hash(x)=151308771)
657
+ 5450 train 3.944280 (lr=4.1094e-04) (hash(x)=150886070)
658
+ 5460 train 4.053824 (lr=4.0931e-04) (hash(x)=146080408)
659
+ 5470 train 4.073390 (lr=4.0769e-04) (hash(x)=149828857)
660
+ 5480 train 4.046815 (lr=4.0606e-04) (hash(x)=139922034)
661
+ 5490 train 3.990925 (lr=4.0444e-04) (hash(x)=150262276)
662
+ 5500 val loss 4.1243
663
+ 5500 val perplexity 61.8241
664
+ 5500 train 3.963853 (lr=4.0282e-04) (hash(x)=148498335)
665
+ 5510 train 4.076842 (lr=4.0120e-04) (hash(x)=157036224)
666
+ 5520 train 3.962531 (lr=3.9959e-04) (hash(x)=134093409)
667
+ 5530 train 3.989175 (lr=3.9797e-04) (hash(x)=153126250)
668
+ 5540 train 4.003644 (lr=3.9636e-04) (hash(x)=136170985)
669
+ 5550 train 3.976998 (lr=3.9475e-04) (hash(x)=152841201)
670
+ 5560 train 4.087109 (lr=3.9314e-04) (hash(x)=147439014)
671
+ 5570 train 3.999181 (lr=3.9154e-04) (hash(x)=143476132)
672
+ 5580 train 3.775414 (lr=3.8993e-04) (hash(x)=139503437)
673
+ 5590 train 3.820708 (lr=3.8833e-04) (hash(x)=131664955)
674
+ 5600 val loss 4.1335
675
+ 5600 val perplexity 62.3944
676
+ 5600 train 3.909602 (lr=3.8674e-04) (hash(x)=151907614)
677
+ 5610 train 3.796695 (lr=3.8514e-04) (hash(x)=157132674)
678
+ 5620 train 4.001199 (lr=3.8355e-04) (hash(x)=146793775)
679
+ 5630 train 3.942438 (lr=3.8196e-04) (hash(x)=147237910)
680
+ 5640 train 3.853781 (lr=3.8037e-04) (hash(x)=141120024)
681
+ 5650 train 3.825025 (lr=3.7878e-04) (hash(x)=144275645)
682
+ 5660 train 3.861953 (lr=3.7720e-04) (hash(x)=159088002)
683
+ 5670 train 3.917165 (lr=3.7562e-04) (hash(x)=147224479)
684
+ 5680 train 4.267145 (lr=3.7404e-04) (hash(x)=147208743)
685
+ 5690 train 4.067957 (lr=3.7246e-04) (hash(x)=154384701)
686
+ 5700 val loss 4.1267
687
+ 5700 val perplexity 61.9720
688
+ 5700 train 4.101302 (lr=3.7089e-04) (hash(x)=155192267)
689
+ 5710 train 4.051077 (lr=3.6932e-04) (hash(x)=148458753)
690
+ 5720 train 4.148335 (lr=3.6775e-04) (hash(x)=159947204)
691
+ 5730 train 4.136127 (lr=3.6619e-04) (hash(x)=150746366)
692
+ 5740 train 4.134599 (lr=3.6462e-04) (hash(x)=155201884)
693
+ 5750 train 4.133183 (lr=3.6306e-04) (hash(x)=156432008)
694
+ 5760 train 4.117596 (lr=3.6151e-04) (hash(x)=185977746)
695
+ 5770 train 4.112559 (lr=3.5995e-04) (hash(x)=133620000)
696
+ 5780 train 4.124801 (lr=3.5840e-04) (hash(x)=153778044)
697
+ 5790 train 3.969269 (lr=3.5685e-04) (hash(x)=147713822)
698
+ 5800 val loss 4.1030
699
+ 5800 val perplexity 60.5221
700
+ 5800 train 4.123096 (lr=3.5530e-04) (hash(x)=153132158)
701
+ 5810 train 4.078139 (lr=3.5376e-04) (hash(x)=144293895)
702
+ 5820 train 4.060985 (lr=3.5222e-04) (hash(x)=151500590)
703
+ 5830 train 4.014638 (lr=3.5068e-04) (hash(x)=139308707)
704
+ 5840 train 4.131819 (lr=3.4915e-04) (hash(x)=153591278)
705
+ 5850 train 4.104186 (lr=3.4762e-04) (hash(x)=146077037)
706
+ 5860 train 4.064821 (lr=3.4609e-04) (hash(x)=149160733)
707
+ 5870 train 4.048395 (lr=3.4456e-04) (hash(x)=138482476)
708
+ 5880 train 4.062891 (lr=3.4304e-04) (hash(x)=152173137)
709
+ 5890 train 4.037019 (lr=3.4152e-04) (hash(x)=142387057)
710
+ 5900 val loss 4.1137
711
+ 5900 val perplexity 61.1700
712
+ 5900 train 4.122136 (lr=3.4000e-04) (hash(x)=161446764)
713
+ 5910 train 3.986922 (lr=3.3849e-04) (hash(x)=144512138)
714
+ 5920 train 4.046385 (lr=3.3697e-04) (hash(x)=152555618)
715
+ 5930 train 4.005532 (lr=3.3547e-04) (hash(x)=146129367)
716
+ 5940 train 3.940385 (lr=3.3396e-04) (hash(x)=154607527)
717
+ 5950 train 4.038459 (lr=3.3246e-04) (hash(x)=143827143)
718
+ 5960 train 4.164495 (lr=3.3096e-04) (hash(x)=151168510)
719
+ 5970 train 3.947716 (lr=3.2947e-04) (hash(x)=158961574)
720
+ 5980 train 3.780915 (lr=3.2797e-04) (hash(x)=148757787)
721
+ 5990 train 4.290759 (lr=3.2649e-04) (hash(x)=161719494)
722
+ 6000 val loss 4.1127
723
+ 6000 val perplexity 61.1120
724
+ 6000 train 3.787089 (lr=3.2500e-04) (hash(x)=151512446)
725
+ 6010 train 3.895685 (lr=3.2352e-04) (hash(x)=173500735)
726
+ 6020 train 4.034158 (lr=3.2204e-04) (hash(x)=157621031)
727
+ 6030 train 3.924242 (lr=3.2056e-04) (hash(x)=142662684)
728
+ 6040 train 3.910200 (lr=3.1909e-04) (hash(x)=146180940)
729
+ 6050 train 3.827262 (lr=3.1762e-04) (hash(x)=147654412)
730
+ 6060 train 3.701575 (lr=3.1616e-04) (hash(x)=156393157)
731
+ 6070 train 4.102852 (lr=3.1469e-04) (hash(x)=156947300)
732
+ 6080 train 4.095179 (lr=3.1323e-04) (hash(x)=163108626)
733
+ 6090 train 4.196997 (lr=3.1178e-04) (hash(x)=158654713)
734
+ 6100 val loss 4.0920
735
+ 6100 val perplexity 59.8601
736
+ 6100 train 3.926746 (lr=3.1033e-04) (hash(x)=188094053)
737
+ 6110 train 4.081031 (lr=3.0888e-04) (hash(x)=145132600)
738
+ 6120 train 3.951181 (lr=3.0743e-04) (hash(x)=146999707)
739
+ 6130 train 4.194649 (lr=3.0599e-04) (hash(x)=145719530)
740
+ 6140 train 4.125907 (lr=3.0455e-04) (hash(x)=153230722)
741
+ 6150 train 4.063953 (lr=3.0312e-04) (hash(x)=147455934)
742
+ 6160 train 4.108177 (lr=3.0169e-04) (hash(x)=146596242)
743
+ 6170 train 4.203428 (lr=3.0026e-04) (hash(x)=152029855)
744
+ 6180 train 4.072337 (lr=2.9884e-04) (hash(x)=151480034)
745
+ 6190 train 3.997838 (lr=2.9742e-04) (hash(x)=147381047)
746
+ 6200 val loss 4.0767
747
+ 6200 val perplexity 58.9484
748
+ 6200 train 3.990994 (lr=2.9600e-04) (hash(x)=149389789)
749
+ 6210 train 4.036792 (lr=2.9459e-04) (hash(x)=154894138)
750
+ 6220 train 4.074786 (lr=2.9318e-04) (hash(x)=150074302)
751
+ 6230 train 3.941658 (lr=2.9177e-04) (hash(x)=142881794)
752
+ 6240 train 3.931683 (lr=2.9037e-04) (hash(x)=164747461)
753
+ 6250 train 4.097150 (lr=2.8897e-04) (hash(x)=158378292)
754
+ 6260 train 4.100236 (lr=2.8758e-04) (hash(x)=147675938)
755
+ 6270 train 4.040796 (lr=2.8619e-04) (hash(x)=148197228)
756
+ 6280 train 4.005805 (lr=2.8480e-04) (hash(x)=135296747)
757
+ 6290 train 4.069369 (lr=2.8342e-04) (hash(x)=156439346)
758
+ 6300 val loss 4.0763
759
+ 6300 val perplexity 58.9262
760
+ 6300 train 3.899572 (lr=2.8204e-04) (hash(x)=138212820)
761
+ 6310 train 4.183403 (lr=2.8067e-04) (hash(x)=165985336)
762
+ 6320 train 4.068029 (lr=2.7930e-04) (hash(x)=145921897)
763
+ 6330 train 4.267802 (lr=2.7793e-04) (hash(x)=145502689)
764
+ 6340 train 4.006865 (lr=2.7657e-04) (hash(x)=141704937)
765
+ 6350 train 4.056522 (lr=2.7521e-04) (hash(x)=161789400)
766
+ 6360 train 3.787497 (lr=2.7385e-04) (hash(x)=145457780)
767
+ 6370 train 4.037299 (lr=2.7250e-04) (hash(x)=171628530)
768
+ 6380 train 3.827574 (lr=2.7116e-04) (hash(x)=153467923)
769
+ 6390 train 4.199471 (lr=2.6981e-04) (hash(x)=158066904)
770
+ 6400 val loss 4.0816
771
+ 6400 val perplexity 59.2378
772
+ 6400 train 3.851231 (lr=2.6847e-04) (hash(x)=146535423)
773
+ 6410 train 3.714586 (lr=2.6714e-04) (hash(x)=147264155)
774
+ 6420 train 3.813966 (lr=2.6581e-04) (hash(x)=145595485)
775
+ 6430 train 3.963683 (lr=2.6448e-04) (hash(x)=146929074)
776
+ 6440 train 3.899956 (lr=2.6316e-04) (hash(x)=147290878)
777
+ 6450 train 3.984367 (lr=2.6184e-04) (hash(x)=141154205)
778
+ 6460 train 3.825710 (lr=2.6053e-04) (hash(x)=154360901)
779
+ 6470 train 4.171858 (lr=2.5922e-04) (hash(x)=146350558)
780
+ 6480 train 4.123651 (lr=2.5791e-04) (hash(x)=153927683)
781
+ 6490 train 4.044893 (lr=2.5661e-04) (hash(x)=149337047)
782
+ 6500 val loss 4.0674
783
+ 6500 val perplexity 58.4067
784
+ 6500 train 3.963748 (lr=2.5531e-04) (hash(x)=145950843)
785
+ 6510 train 4.028126 (lr=2.5402e-04) (hash(x)=142577653)
786
+ 6520 train 4.081803 (lr=2.5273e-04) (hash(x)=165060689)
787
+ 6530 train 3.935182 (lr=2.5145e-04) (hash(x)=148367505)
788
+ 6540 train 4.025043 (lr=2.5017e-04) (hash(x)=154313651)
789
+ 6550 train 3.947546 (lr=2.4889e-04) (hash(x)=147477315)
790
+ 6560 train 4.036665 (lr=2.4762e-04) (hash(x)=139825403)
791
+ 6570 train 4.047403 (lr=2.4635e-04) (hash(x)=141482518)
792
+ 6580 train 4.087807 (lr=2.4509e-04) (hash(x)=144968144)
793
+ 6590 train 4.041284 (lr=2.4383e-04) (hash(x)=154319720)
794
+ 6600 val loss 4.0639
795
+ 6600 val perplexity 58.2016
796
+ 6600 train 4.076058 (lr=2.4258e-04) (hash(x)=141162902)
797
+ 6610 train 4.043654 (lr=2.4133e-04) (hash(x)=159490449)
798
+ 6620 train 3.926958 (lr=2.4009e-04) (hash(x)=151189772)
799
+ 6630 train 3.905117 (lr=2.3884e-04) (hash(x)=148451651)
800
+ 6640 train 4.024920 (lr=2.3761e-04) (hash(x)=152815550)
801
+ 6650 train 3.880545 (lr=2.3638e-04) (hash(x)=144316094)
802
+ 6660 train 4.201808 (lr=2.3515e-04) (hash(x)=143951240)
803
+ 6670 train 3.817460 (lr=2.3393e-04) (hash(x)=143203472)
804
+ 6680 train 3.859293 (lr=2.3271e-04) (hash(x)=132063806)
805
+ 6690 train 3.953261 (lr=2.3150e-04) (hash(x)=140739750)
806
+ 6700 val loss 4.0679
807
+ 6700 val perplexity 58.4330
808
+ 6700 train 4.009325 (lr=2.3029e-04) (hash(x)=153018737)
809
+ 6710 train 3.879687 (lr=2.2909e-04) (hash(x)=140107197)
810
+ 6720 train 3.981326 (lr=2.2789e-04) (hash(x)=144396559)
811
+ 6730 train 3.828541 (lr=2.2669e-04) (hash(x)=148038991)
812
+ 6740 train 4.044957 (lr=2.2550e-04) (hash(x)=145504746)
813
+ 6750 train 4.023155 (lr=2.2432e-04) (hash(x)=119579094)
814
+ 6760 train 4.035737 (lr=2.2314e-04) (hash(x)=161725430)
815
+ 6770 train 4.057579 (lr=2.2196e-04) (hash(x)=153646966)
816
+ 6780 train 4.069037 (lr=2.2079e-04) (hash(x)=153838463)
817
+ 6790 train 4.191577 (lr=2.1963e-04) (hash(x)=157891196)
818
+ 6800 val loss 4.0492
819
+ 6800 val perplexity 57.3531
820
+ 6800 train 3.904472 (lr=2.1847e-04) (hash(x)=155640155)
821
+ 6810 train 4.102904 (lr=2.1731e-04) (hash(x)=154913809)
822
+ 6820 train 4.009687 (lr=2.1616e-04) (hash(x)=159412493)
823
+ 6830 train 3.974159 (lr=2.1501e-04) (hash(x)=144756549)
824
+ 6840 train 4.184587 (lr=2.1387e-04) (hash(x)=143824638)
825
+ 6850 train 4.519359 (lr=2.1273e-04) (hash(x)=117180982)
826
+ 6860 train 4.032228 (lr=2.1160e-04) (hash(x)=141887150)
827
+ 6870 train 4.103032 (lr=2.1047e-04) (hash(x)=140094872)
828
+ 6880 train 3.941839 (lr=2.0935e-04) (hash(x)=141479953)
829
+ 6890 train 4.001323 (lr=2.0823e-04) (hash(x)=153166840)
830
+ 6900 val loss 4.0474
831
+ 6900 val perplexity 57.2502
832
+ 6900 train 4.028118 (lr=2.0712e-04) (hash(x)=153722115)
833
+ 6910 train 4.133997 (lr=2.0602e-04) (hash(x)=149578602)
834
+ 6920 train 3.986024 (lr=2.0491e-04) (hash(x)=149925108)
835
+ 6930 train 3.934211 (lr=2.0382e-04) (hash(x)=148589489)
836
+ 6940 train 3.841560 (lr=2.0272e-04) (hash(x)=144243794)
837
+ 6950 train 3.924499 (lr=2.0164e-04) (hash(x)=157743480)
838
+ 6960 train 3.912309 (lr=2.0055e-04) (hash(x)=139858247)
839
+ 6970 train 3.847862 (lr=1.9948e-04) (hash(x)=148727536)
840
+ 6980 train 3.953541 (lr=1.9840e-04) (hash(x)=153382595)
841
+ 6990 train 4.103443 (lr=1.9734e-04) (hash(x)=147529919)
842
+ 7000 val loss 4.0413
843
+ 7000 val perplexity 56.9000
844
+ 7000 train 4.059993 (lr=1.9628e-04) (hash(x)=146953450)
845
+ 7010 train 3.976213 (lr=1.9522e-04) (hash(x)=134561015)
846
+ 7020 train 4.087538 (lr=1.9417e-04) (hash(x)=148433515)
847
+ 7030 train 3.931887 (lr=1.9312e-04) (hash(x)=140394326)
848
+ 7040 train 3.936596 (lr=1.9208e-04) (hash(x)=136740763)
849
+ 7050 train 3.776903 (lr=1.9104e-04) (hash(x)=148637938)
850
+ 7060 train 3.943983 (lr=1.9001e-04) (hash(x)=147262147)
851
+ 7070 train 3.920478 (lr=1.8899e-04) (hash(x)=140191039)
852
+ 7080 train 3.938794 (lr=1.8797e-04) (hash(x)=148223305)
853
+ 7090 train 3.853806 (lr=1.8695e-04) (hash(x)=139984988)
854
+ 7100 val loss 4.0452
855
+ 7100 val perplexity 57.1233
856
+ 7100 train 3.969778 (lr=1.8594e-04) (hash(x)=137663885)
857
+ 7110 train 3.850114 (lr=1.8494e-04) (hash(x)=150955021)
858
+ 7120 train 3.757265 (lr=1.8394e-04) (hash(x)=148585682)
859
+ 7130 train 3.958420 (lr=1.8294e-04) (hash(x)=146055548)
860
+ 7140 train 3.995567 (lr=1.8196e-04) (hash(x)=155723541)
861
+ 7150 train 4.075913 (lr=1.8097e-04) (hash(x)=140465834)
862
+ 7160 train 4.041488 (lr=1.7999e-04) (hash(x)=163246922)
863
+ 7170 train 3.905421 (lr=1.7902e-04) (hash(x)=150754166)
864
+ 7180 train 4.445025 (lr=1.7805e-04) (hash(x)=142844151)
865
+ 7190 train 4.024181 (lr=1.7709e-04) (hash(x)=145167470)
866
+ 7200 val loss 4.0306
867
+ 7200 val perplexity 56.2952
868
+ 7200 train 4.215508 (lr=1.7614e-04) (hash(x)=146172950)
869
+ 7210 train 4.045685 (lr=1.7519e-04) (hash(x)=143155287)
870
+ 7220 train 3.991629 (lr=1.7424e-04) (hash(x)=151976669)
871
+ 7230 train 4.102512 (lr=1.7330e-04) (hash(x)=145979940)
872
+ 7240 train 4.071609 (lr=1.7237e-04) (hash(x)=155656034)
873
+ 7250 train 4.075710 (lr=1.7144e-04) (hash(x)=135610136)
874
+ 7260 train 4.161097 (lr=1.7051e-04) (hash(x)=148133023)
875
+ 7270 train 4.027514 (lr=1.6959e-04) (hash(x)=148006681)
876
+ 7280 train 4.067182 (lr=1.6868e-04) (hash(x)=151397627)
877
+ 7290 train 4.022948 (lr=1.6777e-04) (hash(x)=141696871)
878
+ 7300 val loss 4.0257
879
+ 7300 val perplexity 56.0169
880
+ 7300 train 3.957256 (lr=1.6687e-04) (hash(x)=150018163)
881
+ 7310 train 3.997174 (lr=1.6598e-04) (hash(x)=154900050)
882
+ 7320 train 3.905186 (lr=1.6509e-04) (hash(x)=142608283)
883
+ 7330 train 3.996033 (lr=1.6420e-04) (hash(x)=148977837)
884
+ 7340 train 4.013555 (lr=1.6332e-04) (hash(x)=141676907)
885
+ 7350 train 3.903894 (lr=1.6245e-04) (hash(x)=147846997)
886
+ 7360 train 4.148656 (lr=1.6158e-04) (hash(x)=143252181)
887
+ 7370 train 3.961204 (lr=1.6072e-04) (hash(x)=155081717)
888
+ 7380 train 4.050922 (lr=1.5986e-04) (hash(x)=149678451)
889
+ 7390 train 3.886603 (lr=1.5901e-04) (hash(x)=159341637)
890
+ 7400 val loss 4.0258
891
+ 7400 val perplexity 56.0263
892
+ 7400 train 4.171954 (lr=1.5816e-04) (hash(x)=145351166)
893
+ 7410 train 3.982174 (lr=1.5732e-04) (hash(x)=162274009)
894
+ 7420 train 3.829725 (lr=1.5649e-04) (hash(x)=143254858)
895
+ 7430 train 3.944977 (lr=1.5566e-04) (hash(x)=143049884)
896
+ 7440 train 3.867578 (lr=1.5484e-04) (hash(x)=141768538)
897
+ 7450 train 3.886175 (lr=1.5402e-04) (hash(x)=139895310)
898
+ 7460 train 3.762972 (lr=1.5321e-04) (hash(x)=155900365)
899
+ 7470 train 3.740720 (lr=1.5241e-04) (hash(x)=145048215)
900
+ 7480 train 3.891387 (lr=1.5161e-04) (hash(x)=136992624)
901
+ 7490 train 3.994782 (lr=1.5081e-04) (hash(x)=141424789)
902
+ 7500 val loss 4.0308
903
+ 7500 val perplexity 56.3033
904
+ 7500 train 3.847571 (lr=1.5002e-04) (hash(x)=145292116)
905
+ 7510 train 3.880529 (lr=1.4924e-04) (hash(x)=144818029)
906
+ 7520 train 4.054898 (lr=1.4847e-04) (hash(x)=152903515)
907
+ 7530 train 4.058295 (lr=1.4769e-04) (hash(x)=156934724)
908
+ 7540 train 4.288443 (lr=1.4693e-04) (hash(x)=156695926)
909
+ 7550 train 4.047186 (lr=1.4617e-04) (hash(x)=149115609)
910
+ 7560 train 4.020660 (lr=1.4542e-04) (hash(x)=143974224)
911
+ 7570 train 3.983071 (lr=1.4467e-04) (hash(x)=148204424)
912
+ 7580 train 4.054494 (lr=1.4393e-04) (hash(x)=150177910)
913
+ 7590 train 4.071836 (lr=1.4319e-04) (hash(x)=140325629)
914
+ 7600 val loss 4.0169
915
+ 7600 val perplexity 55.5278
916
+ 7600 train 3.964485 (lr=1.4246e-04) (hash(x)=150235132)
917
+ 7610 train 4.079489 (lr=1.4174e-04) (hash(x)=157594837)
918
+ 7620 train 4.098772 (lr=1.4102e-04) (hash(x)=146696749)
919
+ 7630 train 4.128080 (lr=1.4031e-04) (hash(x)=148463580)
920
+ 7640 train 3.946570 (lr=1.3960e-04) (hash(x)=140271046)
921
+ 7650 train 4.007680 (lr=1.3890e-04) (hash(x)=159952319)
922
+ 7660 train 4.011572 (lr=1.3821e-04) (hash(x)=146307392)
923
+ 7670 train 3.959486 (lr=1.3752e-04) (hash(x)=145027988)
924
+ 7680 train 4.028000 (lr=1.3684e-04) (hash(x)=147729698)
925
+ 7690 train 4.012680 (lr=1.3616e-04) (hash(x)=143401210)
926
+ 7700 val loss 4.0114
927
+ 7700 val perplexity 55.2262
928
+ 7700 train 3.937764 (lr=1.3549e-04) (hash(x)=154543455)
929
+ 7710 train 4.016861 (lr=1.3483e-04) (hash(x)=151600456)
930
+ 7720 train 3.935154 (lr=1.3417e-04) (hash(x)=165913803)
931
+ 7730 train 3.803798 (lr=1.3352e-04) (hash(x)=144942059)
932
+ 7740 train 4.028062 (lr=1.3287e-04) (hash(x)=150302103)
933
+ 7750 train 3.972753 (lr=1.3223e-04) (hash(x)=155782458)
934
+ 7760 train 4.005678 (lr=1.3160e-04) (hash(x)=147824384)
935
+ 7770 train 4.021020 (lr=1.3097e-04) (hash(x)=149177657)
936
+ 7780 train 4.041036 (lr=1.3035e-04) (hash(x)=164446738)
937
+ 7790 train 3.841137 (lr=1.2974e-04) (hash(x)=146417077)
938
+ 7800 val loss 4.0137
939
+ 7800 val perplexity 55.3526
940
+ 7800 train 3.987621 (lr=1.2913e-04) (hash(x)=142456852)
941
+ 7810 train 3.692454 (lr=1.2852e-04) (hash(x)=146221807)
942
+ 7820 train 3.884215 (lr=1.2793e-04) (hash(x)=154136533)
943
+ 7830 train 3.793244 (lr=1.2733e-04) (hash(x)=147492735)
944
+ 7840 train 3.953902 (lr=1.2675e-04) (hash(x)=143481291)
945
+ 7850 train 3.776357 (lr=1.2617e-04) (hash(x)=136749307)
946
+ 7860 train 3.804791 (lr=1.2560e-04) (hash(x)=128985347)
947
+ 7870 train 3.706167 (lr=1.2503e-04) (hash(x)=152897363)
948
+ 7880 train 3.912717 (lr=1.2447e-04) (hash(x)=151460360)
949
+ 7890 train 3.735288 (lr=1.2392e-04) (hash(x)=137481738)
950
+ 7900 val loss 4.0150
951
+ 7900 val perplexity 55.4233
952
+ 7900 train 3.809030 (lr=1.2337e-04) (hash(x)=147363479)
953
+ 7910 train 4.032862 (lr=1.2283e-04) (hash(x)=151165744)
954
+ 7920 train 4.207320 (lr=1.2229e-04) (hash(x)=146214226)
955
+ 7930 train 4.010740 (lr=1.2176e-04) (hash(x)=154457502)
956
+ 7940 train 3.918246 (lr=1.2124e-04) (hash(x)=162025201)
957
+ 7950 train 3.907157 (lr=1.2072e-04) (hash(x)=156488145)
958
+ 7960 train 4.181965 (lr=1.2021e-04) (hash(x)=154011584)
959
+ 7970 train 4.037618 (lr=1.1970e-04) (hash(x)=145330320)
960
+ 7980 train 3.948063 (lr=1.1921e-04) (hash(x)=148216312)
961
+ 7990 train 3.882433 (lr=1.1871e-04) (hash(x)=156819091)
962
+ 8000 val loss 4.0053
963
+ 8000 val perplexity 54.8863
964
+ 8000 train 4.037604 (lr=1.1823e-04) (hash(x)=156122973)
965
+ 8010 train 3.980929 (lr=1.1775e-04) (hash(x)=146933088)
966
+ 8020 train 4.076870 (lr=1.1728e-04) (hash(x)=151650746)
967
+ 8030 train 4.001222 (lr=1.1681e-04) (hash(x)=140692373)
968
+ 8040 train 4.192969 (lr=1.1635e-04) (hash(x)=147850094)
969
+ 8050 train 4.003552 (lr=1.1589e-04) (hash(x)=144454109)
970
+ 8060 train 3.871573 (lr=1.1544e-04) (hash(x)=152205124)
971
+ 8070 train 4.087498 (lr=1.1500e-04) (hash(x)=148918066)
972
+ 8080 train 4.014620 (lr=1.1457e-04) (hash(x)=146481024)
973
+ 8090 train 4.054576 (lr=1.1414e-04) (hash(x)=168785549)
974
+ 8100 val loss 4.0026
975
+ 8100 val perplexity 54.7418
976
+ 8100 train 4.081084 (lr=1.1371e-04) (hash(x)=156153179)
977
+ 8110 train 4.035790 (lr=1.1330e-04) (hash(x)=150498601)
978
+ 8120 train 4.031190 (lr=1.1289e-04) (hash(x)=142002714)
979
+ 8130 train 3.927924 (lr=1.1248e-04) (hash(x)=138450483)
980
+ 8140 train 4.007600 (lr=1.1209e-04) (hash(x)=145306351)
981
+ 8150 train 3.888445 (lr=1.1169e-04) (hash(x)=172502865)
982
+ 8160 train 3.988620 (lr=1.1131e-04) (hash(x)=156689649)
983
+ 8170 train 3.909012 (lr=1.1093e-04) (hash(x)=143766932)
984
+ 8180 train 4.050617 (lr=1.1056e-04) (hash(x)=150361321)
985
+ 8190 train 3.878583 (lr=1.1019e-04) (hash(x)=152664241)
986
+ 8200 val loss 4.0013
987
+ 8200 val perplexity 54.6709
988
+ 8200 train 4.152580 (lr=1.0983e-04) (hash(x)=146430698)
989
+ 8210 train 3.847202 (lr=1.0948e-04) (hash(x)=167362078)
990
+ 8220 train 4.069087 (lr=1.0913e-04) (hash(x)=157650942)
991
+ 8230 train 3.872581 (lr=1.0879e-04) (hash(x)=142102044)
992
+ 8240 train 3.912508 (lr=1.0846e-04) (hash(x)=153405900)
993
+ 8250 train 3.864430 (lr=1.0813e-04) (hash(x)=140222909)
994
+ 8260 train 3.995524 (lr=1.0781e-04) (hash(x)=137981854)
995
+ 8270 train 3.908979 (lr=1.0750e-04) (hash(x)=147576272)
996
+ 8280 train 3.922947 (lr=1.0719e-04) (hash(x)=157988195)
997
+ 8290 train 3.976058 (lr=1.0689e-04) (hash(x)=145715681)
998
+ 8300 val loss 4.0026
999
+ 8300 val perplexity 54.7429
1000
+ 8300 train 3.931413 (lr=1.0659e-04) (hash(x)=143507257)
1001
+ 8310 train 3.979415 (lr=1.0630e-04) (hash(x)=148930064)
1002
+ 8320 train 3.911449 (lr=1.0602e-04) (hash(x)=144134805)
1003
+ 8330 train 4.023092 (lr=1.0574e-04) (hash(x)=148418079)
1004
+ 8340 train 3.753142 (lr=1.0547e-04) (hash(x)=166518807)
1005
+ 8350 train 4.059415 (lr=1.0521e-04) (hash(x)=140042036)
1006
+ 8360 train 3.929431 (lr=1.0495e-04) (hash(x)=151427074)
1007
+ 8370 train 4.017523 (lr=1.0470e-04) (hash(x)=158389185)
1008
+ 8380 train 3.919978 (lr=1.0446e-04) (hash(x)=148193308)
1009
+ 8390 train 3.988597 (lr=1.0422e-04) (hash(x)=144289008)
1010
+ 8400 val loss 3.9935
1011
+ 8400 val perplexity 54.2438
1012
+ 8400 train 3.703464 (lr=1.0399e-04) (hash(x)=166272643)
1013
+ 8410 train 3.916271 (lr=1.0377e-04) (hash(x)=162244433)
1014
+ 8420 train 3.948399 (lr=1.0355e-04) (hash(x)=149271299)
1015
+ 8430 train 4.097052 (lr=1.0334e-04) (hash(x)=154135338)
1016
+ 8440 train 4.136253 (lr=1.0313e-04) (hash(x)=149661838)
1017
+ 8450 train 3.980165 (lr=1.0293e-04) (hash(x)=142205056)
1018
+ 8460 train 3.907764 (lr=1.0274e-04) (hash(x)=166652398)
1019
+ 8470 train 4.080571 (lr=1.0256e-04) (hash(x)=149029710)
1020
+ 8480 train 3.847819 (lr=1.0238e-04) (hash(x)=154217415)
1021
+ 8490 train 3.931644 (lr=1.0220e-04) (hash(x)=151526557)
1022
+ 8500 val loss 3.9943
1023
+ 8500 val perplexity 54.2896
1024
+ 8500 train 3.916831 (lr=1.0204e-04) (hash(x)=143887848)
1025
+ 8510 train 3.903115 (lr=1.0188e-04) (hash(x)=152324645)
1026
+ 8520 train 3.639731 (lr=1.0172e-04) (hash(x)=145124959)
1027
+ 8530 train 3.965969 (lr=1.0158e-04) (hash(x)=186319340)
1028
+ 8540 train 3.973069 (lr=1.0144e-04) (hash(x)=147388339)
1029
+ 8550 train 4.002029 (lr=1.0130e-04) (hash(x)=150118928)
1030
+ 8560 train 4.008875 (lr=1.0118e-04) (hash(x)=153985286)
1031
+ 8570 train 3.944179 (lr=1.0106e-04) (hash(x)=149582799)
1032
+ 8580 train 3.951494 (lr=1.0094e-04) (hash(x)=156867870)
1033
+ 8590 train 3.842965 (lr=1.0083e-04) (hash(x)=148773298)
1034
+ 8600 val loss 3.9964
1035
+ 8600 val perplexity 54.4047
1036
+ 8600 train 3.939678 (lr=1.0073e-04) (hash(x)=156900341)
1037
+ 8610 train 4.098356 (lr=1.0064e-04) (hash(x)=156790089)
1038
+ 8620 train 3.895754 (lr=1.0055e-04) (hash(x)=154680805)
1039
+ 8630 train 3.936382 (lr=1.0047e-04) (hash(x)=138107873)
1040
+ 8640 train 4.027695 (lr=1.0039e-04) (hash(x)=142554243)
1041
+ 8650 train 3.836763 (lr=1.0033e-04) (hash(x)=134360956)
1042
+ 8660 train 3.913641 (lr=1.0026e-04) (hash(x)=128599610)
1043
+ 8670 train 3.937629 (lr=1.0021e-04) (hash(x)=159628945)
1044
+ 8680 train 3.965818 (lr=1.0016e-04) (hash(x)=143161532)
1045
+ 8690 train 3.998866 (lr=1.0012e-04) (hash(x)=140104912)
1046
+ 8700 val loss 3.9884
1047
+ 8700 val perplexity 53.9711
1048
+ 8700 train 4.170671 (lr=1.0008e-04) (hash(x)=146417632)
1049
+ 8710 train 4.155127 (lr=1.0005e-04) (hash(x)=153671695)
1050
+ 8720 train 4.020198 (lr=1.0003e-04) (hash(x)=141680750)
1051
+ 8730 train 3.899860 (lr=1.0001e-04) (hash(x)=147279231)
1052
+ 8740 train 4.184060 (lr=1.0000e-04) (hash(x)=129729371)
1053
+ 8749 val loss 3.9880
1054
+ 8749 val perplexity 53.9455
lr10e-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 sha256:148d8e3086af92311eb47b78b11e0cceb956142340f37cc0101a1c16b7e940e7
3
+ size 92843394
lr10e-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:b6ce7eb91123b7a439893b072383a881afd49a49077c8473b56469ab535713bd
3
+ size 179406214