andrew-healey commited on
Commit
10f34aa
·
verified ·
1 Parent(s): fb67e73

Upload folder using huggingface_hub

Browse files
lr9e-4_total_batch_size61440_baseline_seed1338/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_11/lr9e-4_total_batch_size61440_baseline_seed1338", "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": 1338, "memory_penalty_epsilon": 0.1, "selection_head_linear_combo": "none", "selection_head_linear_combo_scale": 1.0, "protection_kind": "none", "leaky_relu_alpha": null, "leaky_relu_bias": null, "use_compile": true, "use_mini_model": false, "upload_to_hf": true, "seq_len": 256, "batch_size": 120, "total_batch_size": 61440, "protection_head_scaling_factor": 1.0, "protection_head_bias": 0.0, "n_sliced_masks": null, "n_latent_masks": null, "mask_layernorm": false, "residual_attention_masks": false, "compute_base_shapes": false, "base_shapes_savefile": null, "mup": true, "disable_selection": false, "mup_enable_coord_check_logging": false, "max_lr": 0.0009, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "9e-4_61440", "n_embd": 256}
lr9e-4_total_batch_size61440_baseline_seed1338/dataloader_08749.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:953385078aa3787b69fc6857dfd48b0a2cd2f4d27c6f8892e01211aca53d07f5
3
+ size 964
lr9e-4_total_batch_size61440_baseline_seed1338/log2.txt ADDED
@@ -0,0 +1,1054 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ max_steps: 8750
2
+ 0 val loss 11.2202
3
+ 0 val perplexity 74626.3672
4
+ 0 train 11.227304 (lr=1.8000e-06) (hash(x)=150327452)
5
+ 10 train 10.126881 (lr=1.9800e-05) (hash(x)=165603215)
6
+ 20 train 9.728159 (lr=3.7800e-05) (hash(x)=142061326)
7
+ 30 train 9.335519 (lr=5.5800e-05) (hash(x)=147690730)
8
+ 40 train 8.932817 (lr=7.3800e-05) (hash(x)=160872542)
9
+ 50 train 8.428353 (lr=9.1800e-05) (hash(x)=166441190)
10
+ 60 train 7.997692 (lr=1.0980e-04) (hash(x)=140536766)
11
+ 70 train 7.700273 (lr=1.2780e-04) (hash(x)=146052944)
12
+ 80 train 7.559612 (lr=1.4580e-04) (hash(x)=140542391)
13
+ 90 train 7.430628 (lr=1.6380e-04) (hash(x)=147098660)
14
+ 100 val loss 7.4497
15
+ 100 val perplexity 1719.2618
16
+ 100 train 7.432733 (lr=1.8180e-04) (hash(x)=166780046)
17
+ 110 train 7.193934 (lr=1.9980e-04) (hash(x)=143376584)
18
+ 120 train 7.136211 (lr=2.1780e-04) (hash(x)=145337630)
19
+ 130 train 7.094437 (lr=2.3580e-04) (hash(x)=147524744)
20
+ 140 train 6.995989 (lr=2.5380e-04) (hash(x)=155587442)
21
+ 150 train 6.876152 (lr=2.7180e-04) (hash(x)=159835303)
22
+ 160 train 6.703243 (lr=2.8980e-04) (hash(x)=142171630)
23
+ 170 train 6.589501 (lr=3.0780e-04) (hash(x)=136930463)
24
+ 180 train 6.571156 (lr=3.2580e-04) (hash(x)=153579148)
25
+ 190 train 6.461697 (lr=3.4380e-04) (hash(x)=140352771)
26
+ 200 val loss 6.6338
27
+ 200 val perplexity 760.3636
28
+ 200 train 6.347960 (lr=3.6180e-04) (hash(x)=155040610)
29
+ 210 train 6.291978 (lr=3.7980e-04) (hash(x)=141273862)
30
+ 220 train 6.532618 (lr=3.9780e-04) (hash(x)=165326618)
31
+ 230 train 6.269921 (lr=4.1580e-04) (hash(x)=150122120)
32
+ 240 train 6.054725 (lr=4.3380e-04) (hash(x)=143024679)
33
+ 250 train 6.063870 (lr=4.5180e-04) (hash(x)=130190460)
34
+ 260 train 6.166791 (lr=4.6980e-04) (hash(x)=151890219)
35
+ 270 train 5.955981 (lr=4.8780e-04) (hash(x)=143685291)
36
+ 280 train 6.254799 (lr=5.0580e-04) (hash(x)=137108194)
37
+ 290 train 6.387038 (lr=5.2380e-04) (hash(x)=157920866)
38
+ 300 val loss 6.1927
39
+ 300 val perplexity 489.1858
40
+ 300 train 6.116960 (lr=5.4180e-04) (hash(x)=155504036)
41
+ 310 train 6.220977 (lr=5.5980e-04) (hash(x)=144551186)
42
+ 320 train 6.208399 (lr=5.7780e-04) (hash(x)=161180086)
43
+ 330 train 5.968534 (lr=5.9580e-04) (hash(x)=141651471)
44
+ 340 train 6.101749 (lr=6.1380e-04) (hash(x)=157490741)
45
+ 350 train 6.043244 (lr=6.3180e-04) (hash(x)=137347213)
46
+ 360 train 6.275973 (lr=6.4980e-04) (hash(x)=171273659)
47
+ 370 train 5.923140 (lr=6.6780e-04) (hash(x)=152118871)
48
+ 380 train 5.924133 (lr=6.8580e-04) (hash(x)=161338097)
49
+ 390 train 6.001586 (lr=7.0380e-04) (hash(x)=152404537)
50
+ 400 val loss 5.8507
51
+ 400 val perplexity 347.4863
52
+ 400 train 5.689188 (lr=7.2180e-04) (hash(x)=143823248)
53
+ 410 train 5.946886 (lr=7.3980e-04) (hash(x)=161686752)
54
+ 420 train 5.681756 (lr=7.5780e-04) (hash(x)=155391312)
55
+ 430 train 5.594948 (lr=7.7580e-04) (hash(x)=142145507)
56
+ 440 train 5.726826 (lr=7.9380e-04) (hash(x)=121931091)
57
+ 450 train 5.673252 (lr=8.1180e-04) (hash(x)=156260416)
58
+ 460 train 5.814106 (lr=8.2980e-04) (hash(x)=169781623)
59
+ 470 train 5.590844 (lr=8.4780e-04) (hash(x)=150131545)
60
+ 480 train 5.614183 (lr=8.6580e-04) (hash(x)=150443184)
61
+ 490 train 5.621337 (lr=8.8380e-04) (hash(x)=149368198)
62
+ 500 val loss 5.6442
63
+ 500 val perplexity 282.6544
64
+ 500 train 5.528654 (lr=9.0000e-04) (hash(x)=143734685)
65
+ 510 train 5.538619 (lr=9.0000e-04) (hash(x)=148943891)
66
+ 520 train 5.544048 (lr=8.9999e-04) (hash(x)=149070354)
67
+ 530 train 5.539348 (lr=8.9997e-04) (hash(x)=153601219)
68
+ 540 train 5.552982 (lr=8.9995e-04) (hash(x)=174970654)
69
+ 550 train 5.488515 (lr=8.9993e-04) (hash(x)=160013925)
70
+ 560 train 5.326142 (lr=8.9989e-04) (hash(x)=152096988)
71
+ 570 train 5.424340 (lr=8.9986e-04) (hash(x)=157203924)
72
+ 580 train 5.218379 (lr=8.9981e-04) (hash(x)=138582113)
73
+ 590 train 5.238603 (lr=8.9976e-04) (hash(x)=157451775)
74
+ 600 val loss 5.4632
75
+ 600 val perplexity 235.8552
76
+ 600 train 5.165352 (lr=8.9971e-04) (hash(x)=150678249)
77
+ 610 train 5.203865 (lr=8.9964e-04) (hash(x)=150248072)
78
+ 620 train 5.113957 (lr=8.9958e-04) (hash(x)=148286909)
79
+ 630 train 5.065981 (lr=8.9950e-04) (hash(x)=141197178)
80
+ 640 train 5.141225 (lr=8.9942e-04) (hash(x)=144988391)
81
+ 650 train 5.079231 (lr=8.9934e-04) (hash(x)=149073315)
82
+ 660 train 5.021560 (lr=8.9925e-04) (hash(x)=145239943)
83
+ 670 train 5.047640 (lr=8.9915e-04) (hash(x)=146657426)
84
+ 680 train 5.284650 (lr=8.9905e-04) (hash(x)=142323429)
85
+ 690 train 5.266286 (lr=8.9894e-04) (hash(x)=142724342)
86
+ 700 val loss 5.3145
87
+ 700 val perplexity 203.2642
88
+ 700 train 5.172587 (lr=8.9883e-04) (hash(x)=175802021)
89
+ 710 train 5.228343 (lr=8.9871e-04) (hash(x)=148902690)
90
+ 720 train 5.319711 (lr=8.9858e-04) (hash(x)=155151320)
91
+ 730 train 5.165679 (lr=8.9845e-04) (hash(x)=142078914)
92
+ 740 train 5.325327 (lr=8.9831e-04) (hash(x)=155634484)
93
+ 750 train 5.303681 (lr=8.9817e-04) (hash(x)=171034639)
94
+ 760 train 5.254575 (lr=8.9802e-04) (hash(x)=142470148)
95
+ 770 train 5.197950 (lr=8.9786e-04) (hash(x)=156024448)
96
+ 780 train 5.104269 (lr=8.9770e-04) (hash(x)=151369417)
97
+ 790 train 5.060380 (lr=8.9753e-04) (hash(x)=139260056)
98
+ 800 val loss 5.1570
99
+ 800 val perplexity 173.6352
100
+ 800 train 5.264471 (lr=8.9736e-04) (hash(x)=158681215)
101
+ 810 train 5.071012 (lr=8.9718e-04) (hash(x)=144884161)
102
+ 820 train 5.115275 (lr=8.9700e-04) (hash(x)=146977514)
103
+ 830 train 4.981943 (lr=8.9681e-04) (hash(x)=148983636)
104
+ 840 train 5.022918 (lr=8.9661e-04) (hash(x)=157824542)
105
+ 850 train 5.005511 (lr=8.9641e-04) (hash(x)=152116061)
106
+ 860 train 5.092813 (lr=8.9620e-04) (hash(x)=153743703)
107
+ 870 train 5.231248 (lr=8.9599e-04) (hash(x)=149913335)
108
+ 880 train 5.007698 (lr=8.9577e-04) (hash(x)=151096434)
109
+ 890 train 4.970409 (lr=8.9554e-04) (hash(x)=145848702)
110
+ 900 val loss 5.0265
111
+ 900 val perplexity 152.4018
112
+ 900 train 4.951246 (lr=8.9531e-04) (hash(x)=146108145)
113
+ 910 train 4.905765 (lr=8.9507e-04) (hash(x)=146903806)
114
+ 920 train 4.849117 (lr=8.9483e-04) (hash(x)=149225019)
115
+ 930 train 4.802539 (lr=8.9458e-04) (hash(x)=146110422)
116
+ 940 train 4.885674 (lr=8.9433e-04) (hash(x)=161333614)
117
+ 950 train 4.787793 (lr=8.9407e-04) (hash(x)=147598108)
118
+ 960 train 4.936236 (lr=8.9380e-04) (hash(x)=158133101)
119
+ 970 train 4.747996 (lr=8.9353e-04) (hash(x)=152199591)
120
+ 980 train 4.708047 (lr=8.9325e-04) (hash(x)=152359036)
121
+ 990 train 4.751160 (lr=8.9297e-04) (hash(x)=145166335)
122
+ 1000 val loss 4.9535
123
+ 1000 val perplexity 141.6736
124
+ 1000 train 4.655449 (lr=8.9268e-04) (hash(x)=154996086)
125
+ 1010 train 4.590543 (lr=8.9239e-04) (hash(x)=149406975)
126
+ 1020 train 4.704567 (lr=8.9209e-04) (hash(x)=144618270)
127
+ 1030 train 4.560729 (lr=8.9178e-04) (hash(x)=148306536)
128
+ 1040 train 4.594090 (lr=8.9147e-04) (hash(x)=151643553)
129
+ 1050 train 4.552526 (lr=8.9115e-04) (hash(x)=153396183)
130
+ 1060 train 4.548725 (lr=8.9083e-04) (hash(x)=157501787)
131
+ 1070 train 4.750874 (lr=8.9050e-04) (hash(x)=153277507)
132
+ 1080 train 4.819337 (lr=8.9016e-04) (hash(x)=156232389)
133
+ 1090 train 4.761999 (lr=8.8982e-04) (hash(x)=150673143)
134
+ 1100 val loss 4.8506
135
+ 1100 val perplexity 127.8223
136
+ 1100 train 4.836854 (lr=8.8947e-04) (hash(x)=153885445)
137
+ 1110 train 4.842272 (lr=8.8912e-04) (hash(x)=143951073)
138
+ 1120 train 5.242023 (lr=8.8876e-04) (hash(x)=151626904)
139
+ 1130 train 4.807992 (lr=8.8840e-04) (hash(x)=144013074)
140
+ 1140 train 4.737811 (lr=8.8803e-04) (hash(x)=149948988)
141
+ 1150 train 4.807371 (lr=8.8766e-04) (hash(x)=159666385)
142
+ 1160 train 4.782620 (lr=8.8728e-04) (hash(x)=146889587)
143
+ 1170 train 4.846618 (lr=8.8689e-04) (hash(x)=151746917)
144
+ 1180 train 4.852084 (lr=8.8650e-04) (hash(x)=154396948)
145
+ 1190 train 4.810452 (lr=8.8610e-04) (hash(x)=148201222)
146
+ 1200 val loss 4.7867
147
+ 1200 val perplexity 119.8991
148
+ 1200 train 4.799521 (lr=8.8570e-04) (hash(x)=142353087)
149
+ 1210 train 4.792325 (lr=8.8529e-04) (hash(x)=146779775)
150
+ 1220 train 4.857192 (lr=8.8487e-04) (hash(x)=155864740)
151
+ 1230 train 4.907359 (lr=8.8445e-04) (hash(x)=129826485)
152
+ 1240 train 4.758951 (lr=8.8403e-04) (hash(x)=108599491)
153
+ 1250 train 4.640911 (lr=8.8359e-04) (hash(x)=146491718)
154
+ 1260 train 4.728163 (lr=8.8316e-04) (hash(x)=141897382)
155
+ 1270 train 4.694450 (lr=8.8271e-04) (hash(x)=144830380)
156
+ 1280 train 4.750750 (lr=8.8227e-04) (hash(x)=157677763)
157
+ 1290 train 4.582318 (lr=8.8181e-04) (hash(x)=143611746)
158
+ 1300 val loss 4.7355
159
+ 1300 val perplexity 113.9173
160
+ 1300 train 4.614932 (lr=8.8135e-04) (hash(x)=150750353)
161
+ 1310 train 4.843960 (lr=8.8089e-04) (hash(x)=152850907)
162
+ 1320 train 4.760063 (lr=8.8042e-04) (hash(x)=148575395)
163
+ 1330 train 4.496878 (lr=8.7994e-04) (hash(x)=157113001)
164
+ 1340 train 4.554093 (lr=8.7946e-04) (hash(x)=137400723)
165
+ 1350 train 4.609486 (lr=8.7897e-04) (hash(x)=129849193)
166
+ 1360 train 4.591850 (lr=8.7848e-04) (hash(x)=144176797)
167
+ 1370 train 4.422951 (lr=8.7798e-04) (hash(x)=144550831)
168
+ 1380 train 4.398117 (lr=8.7747e-04) (hash(x)=144171821)
169
+ 1390 train 4.346493 (lr=8.7696e-04) (hash(x)=146110449)
170
+ 1400 val loss 4.6943
171
+ 1400 val perplexity 109.3189
172
+ 1400 train 4.425768 (lr=8.7645e-04) (hash(x)=152767913)
173
+ 1410 train 4.422140 (lr=8.7593e-04) (hash(x)=148110947)
174
+ 1420 train 4.308456 (lr=8.7540e-04) (hash(x)=153477905)
175
+ 1430 train 4.200732 (lr=8.7487e-04) (hash(x)=147227598)
176
+ 1440 train 4.419471 (lr=8.7433e-04) (hash(x)=146280368)
177
+ 1450 train 4.379464 (lr=8.7379e-04) (hash(x)=146531140)
178
+ 1460 train 4.225892 (lr=8.7324e-04) (hash(x)=151573873)
179
+ 1470 train 4.715286 (lr=8.7268e-04) (hash(x)=155559275)
180
+ 1480 train 4.613695 (lr=8.7212e-04) (hash(x)=140584826)
181
+ 1490 train 4.578327 (lr=8.7156e-04) (hash(x)=165070637)
182
+ 1500 val loss 4.6694
183
+ 1500 val perplexity 106.6328
184
+ 1500 train 4.685782 (lr=8.7099e-04) (hash(x)=151562048)
185
+ 1510 train 4.471927 (lr=8.7041e-04) (hash(x)=147770607)
186
+ 1520 train 4.552866 (lr=8.6983e-04) (hash(x)=156486153)
187
+ 1530 train 4.510231 (lr=8.6924e-04) (hash(x)=140244564)
188
+ 1540 train 4.564040 (lr=8.6865e-04) (hash(x)=151627549)
189
+ 1550 train 4.608399 (lr=8.6806e-04) (hash(x)=146001424)
190
+ 1560 train 4.673399 (lr=8.6745e-04) (hash(x)=184266264)
191
+ 1570 train 4.580872 (lr=8.6684e-04) (hash(x)=145002091)
192
+ 1580 train 4.470580 (lr=8.6623e-04) (hash(x)=132216826)
193
+ 1590 train 4.641253 (lr=8.6561e-04) (hash(x)=160975703)
194
+ 1600 val loss 4.6208
195
+ 1600 val perplexity 101.5803
196
+ 1600 train 4.778591 (lr=8.6499e-04) (hash(x)=166486165)
197
+ 1610 train 4.659199 (lr=8.6436e-04) (hash(x)=149601833)
198
+ 1620 train 4.549088 (lr=8.6372e-04) (hash(x)=157815188)
199
+ 1630 train 4.568474 (lr=8.6308e-04) (hash(x)=144234679)
200
+ 1640 train 4.696795 (lr=8.6243e-04) (hash(x)=160054560)
201
+ 1650 train 4.509483 (lr=8.6178e-04) (hash(x)=141061333)
202
+ 1660 train 4.635298 (lr=8.6113e-04) (hash(x)=148064219)
203
+ 1670 train 4.464472 (lr=8.6046e-04) (hash(x)=162799669)
204
+ 1680 train 4.509134 (lr=8.5980e-04) (hash(x)=147474277)
205
+ 1690 train 4.589334 (lr=8.5912e-04) (hash(x)=142369299)
206
+ 1700 val loss 4.5955
207
+ 1700 val perplexity 99.0425
208
+ 1700 train 4.356047 (lr=8.5845e-04) (hash(x)=130835396)
209
+ 1710 train 4.686548 (lr=8.5776e-04) (hash(x)=153173892)
210
+ 1720 train 4.352266 (lr=8.5707e-04) (hash(x)=148451191)
211
+ 1730 train 4.465502 (lr=8.5638e-04) (hash(x)=145047239)
212
+ 1740 train 4.378260 (lr=8.5568e-04) (hash(x)=147187109)
213
+ 1750 train 4.467139 (lr=8.5498e-04) (hash(x)=149493155)
214
+ 1760 train 4.426652 (lr=8.5427e-04) (hash(x)=147647531)
215
+ 1770 train 4.666696 (lr=8.5355e-04) (hash(x)=153677279)
216
+ 1780 train 4.487571 (lr=8.5283e-04) (hash(x)=144314015)
217
+ 1790 train 4.507946 (lr=8.5211e-04) (hash(x)=176962883)
218
+ 1800 val loss 4.5589
219
+ 1800 val perplexity 95.4752
220
+ 1800 train 4.610454 (lr=8.5138e-04) (hash(x)=158851816)
221
+ 1810 train 4.196531 (lr=8.5064e-04) (hash(x)=175415447)
222
+ 1820 train 4.491558 (lr=8.4990e-04) (hash(x)=169829733)
223
+ 1830 train 4.473262 (lr=8.4916e-04) (hash(x)=154116106)
224
+ 1840 train 4.308169 (lr=8.4841e-04) (hash(x)=150168406)
225
+ 1850 train 4.307738 (lr=8.4765e-04) (hash(x)=145782493)
226
+ 1860 train 4.291036 (lr=8.4689e-04) (hash(x)=158111664)
227
+ 1870 train 4.233665 (lr=8.4613e-04) (hash(x)=154837061)
228
+ 1880 train 4.184639 (lr=8.4535e-04) (hash(x)=151363745)
229
+ 1890 train 4.307318 (lr=8.4458e-04) (hash(x)=161283557)
230
+ 1900 val loss 4.5450
231
+ 1900 val perplexity 94.1593
232
+ 1900 train 4.552051 (lr=8.4380e-04) (hash(x)=153313879)
233
+ 1910 train 4.374538 (lr=8.4301e-04) (hash(x)=152272272)
234
+ 1920 train 4.441498 (lr=8.4222e-04) (hash(x)=153648837)
235
+ 1930 train 4.541344 (lr=8.4142e-04) (hash(x)=150161694)
236
+ 1940 train 4.413741 (lr=8.4062e-04) (hash(x)=151235093)
237
+ 1950 train 4.466205 (lr=8.3981e-04) (hash(x)=140582058)
238
+ 1960 train 4.415707 (lr=8.3900e-04) (hash(x)=125616059)
239
+ 1970 train 4.549701 (lr=8.3819e-04) (hash(x)=154353968)
240
+ 1980 train 4.560857 (lr=8.3737e-04) (hash(x)=156063143)
241
+ 1990 train 4.579793 (lr=8.3654e-04) (hash(x)=160927289)
242
+ 2000 val loss 4.5024
243
+ 2000 val perplexity 90.2300
244
+ 2000 train 4.485000 (lr=8.3571e-04) (hash(x)=158245023)
245
+ 2010 train 4.537925 (lr=8.3487e-04) (hash(x)=157096661)
246
+ 2020 train 4.756602 (lr=8.3403e-04) (hash(x)=144994302)
247
+ 2030 train 4.451793 (lr=8.3318e-04) (hash(x)=145742929)
248
+ 2040 train 4.467342 (lr=8.3233e-04) (hash(x)=156958528)
249
+ 2050 train 4.973888 (lr=8.3148e-04) (hash(x)=135313020)
250
+ 2060 train 4.609258 (lr=8.3062e-04) (hash(x)=147309332)
251
+ 2070 train 4.400607 (lr=8.2975e-04) (hash(x)=161720962)
252
+ 2080 train 4.574686 (lr=8.2888e-04) (hash(x)=167204793)
253
+ 2090 train 4.483767 (lr=8.2800e-04) (hash(x)=141325510)
254
+ 2100 val loss 4.4788
255
+ 2100 val perplexity 88.1331
256
+ 2100 train 4.302679 (lr=8.2712e-04) (hash(x)=157204896)
257
+ 2110 train 4.265213 (lr=8.2624e-04) (hash(x)=143077993)
258
+ 2120 train 4.466851 (lr=8.2535e-04) (hash(x)=159032689)
259
+ 2130 train 4.346838 (lr=8.2446e-04) (hash(x)=149513855)
260
+ 2140 train 4.283224 (lr=8.2356e-04) (hash(x)=151678267)
261
+ 2150 train 4.364764 (lr=8.2265e-04) (hash(x)=141309246)
262
+ 2160 train 4.460781 (lr=8.2174e-04) (hash(x)=153935732)
263
+ 2170 train 4.359737 (lr=8.2083e-04) (hash(x)=150668875)
264
+ 2180 train 4.365979 (lr=8.1991e-04) (hash(x)=142033123)
265
+ 2190 train 4.401786 (lr=8.1899e-04) (hash(x)=157015026)
266
+ 2200 val loss 4.4683
267
+ 2200 val perplexity 87.2076
268
+ 2200 train 4.283290 (lr=8.1806e-04) (hash(x)=137541932)
269
+ 2210 train 4.209294 (lr=8.1713e-04) (hash(x)=144424504)
270
+ 2220 train 4.129402 (lr=8.1619e-04) (hash(x)=151015810)
271
+ 2230 train 4.206590 (lr=8.1525e-04) (hash(x)=160707806)
272
+ 2240 train 4.224489 (lr=8.1430e-04) (hash(x)=151874110)
273
+ 2250 train 4.172158 (lr=8.1335e-04) (hash(x)=157341469)
274
+ 2260 train 4.007333 (lr=8.1240e-04) (hash(x)=144563224)
275
+ 2270 train 4.170427 (lr=8.1144e-04) (hash(x)=144878459)
276
+ 2280 train 4.230853 (lr=8.1047e-04) (hash(x)=158331565)
277
+ 2290 train 4.252301 (lr=8.0950e-04) (hash(x)=146875132)
278
+ 2300 val loss 4.4541
279
+ 2300 val perplexity 85.9752
280
+ 2300 train 4.395984 (lr=8.0853e-04) (hash(x)=150149692)
281
+ 2310 train 4.405484 (lr=8.0755e-04) (hash(x)=141206538)
282
+ 2320 train 4.440555 (lr=8.0657e-04) (hash(x)=136219115)
283
+ 2330 train 4.467942 (lr=8.0558e-04) (hash(x)=148496500)
284
+ 2340 train 4.449318 (lr=8.0459e-04) (hash(x)=161388511)
285
+ 2350 train 4.296736 (lr=8.0359e-04) (hash(x)=149433185)
286
+ 2360 train 4.515175 (lr=8.0259e-04) (hash(x)=154659203)
287
+ 2370 train 4.499488 (lr=8.0158e-04) (hash(x)=145696996)
288
+ 2380 train 4.313467 (lr=8.0057e-04) (hash(x)=139915095)
289
+ 2390 train 4.440209 (lr=7.9956e-04) (hash(x)=150567584)
290
+ 2400 val loss 4.4308
291
+ 2400 val perplexity 84.0020
292
+ 2400 train 4.346945 (lr=7.9854e-04) (hash(x)=151730720)
293
+ 2410 train 4.387101 (lr=7.9752e-04) (hash(x)=153136544)
294
+ 2420 train 4.348280 (lr=7.9649e-04) (hash(x)=151004677)
295
+ 2430 train 4.364992 (lr=7.9546e-04) (hash(x)=167118859)
296
+ 2440 train 4.385963 (lr=7.9442e-04) (hash(x)=145433453)
297
+ 2450 train 4.417058 (lr=7.9338e-04) (hash(x)=150913293)
298
+ 2460 train 4.377725 (lr=7.9234e-04) (hash(x)=145674639)
299
+ 2470 train 4.488893 (lr=7.9129e-04) (hash(x)=151022749)
300
+ 2480 train 4.334002 (lr=7.9023e-04) (hash(x)=136621037)
301
+ 2490 train 4.385765 (lr=7.8917e-04) (hash(x)=158100672)
302
+ 2500 val loss 4.3974
303
+ 2500 val perplexity 81.2356
304
+ 2500 train 4.407478 (lr=7.8811e-04) (hash(x)=143406752)
305
+ 2510 train 4.327102 (lr=7.8705e-04) (hash(x)=142664084)
306
+ 2520 train 4.381863 (lr=7.8598e-04) (hash(x)=132174378)
307
+ 2530 train 4.386510 (lr=7.8490e-04) (hash(x)=140760293)
308
+ 2540 train 4.337831 (lr=7.8382e-04) (hash(x)=154863398)
309
+ 2550 train 4.277137 (lr=7.8274e-04) (hash(x)=153042049)
310
+ 2560 train 4.219337 (lr=7.8165e-04) (hash(x)=145339170)
311
+ 2570 train 4.371177 (lr=7.8056e-04) (hash(x)=157656357)
312
+ 2580 train 4.351131 (lr=7.7946e-04) (hash(x)=147899462)
313
+ 2590 train 4.333192 (lr=7.7836e-04) (hash(x)=142812892)
314
+ 2600 val loss 4.4146
315
+ 2600 val perplexity 82.6528
316
+ 2600 train 4.282966 (lr=7.7726e-04) (hash(x)=157272496)
317
+ 2610 train 4.071705 (lr=7.7615e-04) (hash(x)=148923594)
318
+ 2620 train 4.004128 (lr=7.7504e-04) (hash(x)=164654807)
319
+ 2630 train 4.189145 (lr=7.7392e-04) (hash(x)=136419327)
320
+ 2640 train 4.168433 (lr=7.7280e-04) (hash(x)=153420840)
321
+ 2650 train 4.001897 (lr=7.7168e-04) (hash(x)=153693827)
322
+ 2660 train 4.159654 (lr=7.7055e-04) (hash(x)=145574165)
323
+ 2670 train 4.193202 (lr=7.6942e-04) (hash(x)=153894843)
324
+ 2680 train 4.052083 (lr=7.6828e-04) (hash(x)=132234446)
325
+ 2690 train 4.128798 (lr=7.6714e-04) (hash(x)=153430212)
326
+ 2700 val loss 4.3977
327
+ 2700 val perplexity 81.2611
328
+ 2700 train 4.301533 (lr=7.6600e-04) (hash(x)=155342327)
329
+ 2710 train 4.310565 (lr=7.6485e-04) (hash(x)=145797023)
330
+ 2720 train 4.461227 (lr=7.6370e-04) (hash(x)=149220870)
331
+ 2730 train 4.369532 (lr=7.6254e-04) (hash(x)=134069562)
332
+ 2740 train 4.374743 (lr=7.6138e-04) (hash(x)=141560499)
333
+ 2750 train 4.384855 (lr=7.6022e-04) (hash(x)=146000524)
334
+ 2760 train 4.276455 (lr=7.5905e-04) (hash(x)=164710665)
335
+ 2770 train 4.162492 (lr=7.5788e-04) (hash(x)=151199094)
336
+ 2780 train 4.365646 (lr=7.5670e-04) (hash(x)=138625404)
337
+ 2790 train 4.622402 (lr=7.5553e-04) (hash(x)=162766112)
338
+ 2800 val loss 4.3672
339
+ 2800 val perplexity 78.8237
340
+ 2800 train 4.328445 (lr=7.5434e-04) (hash(x)=140626679)
341
+ 2810 train 4.405524 (lr=7.5316e-04) (hash(x)=144248544)
342
+ 2820 train 4.333617 (lr=7.5197e-04) (hash(x)=139378481)
343
+ 2830 train 4.264656 (lr=7.5077e-04) (hash(x)=150173403)
344
+ 2840 train 4.553086 (lr=7.4958e-04) (hash(x)=160168705)
345
+ 2850 train 4.342731 (lr=7.4837e-04) (hash(x)=149480059)
346
+ 2860 train 4.385566 (lr=7.4717e-04) (hash(x)=155559096)
347
+ 2870 train 4.328866 (lr=7.4596e-04) (hash(x)=158825683)
348
+ 2880 train 4.496426 (lr=7.4475e-04) (hash(x)=152129048)
349
+ 2890 train 4.173139 (lr=7.4353e-04) (hash(x)=157866544)
350
+ 2900 val loss 4.3464
351
+ 2900 val perplexity 77.1990
352
+ 2900 train 4.299939 (lr=7.4231e-04) (hash(x)=144953350)
353
+ 2910 train 4.507321 (lr=7.4109e-04) (hash(x)=154103497)
354
+ 2920 train 4.221832 (lr=7.3986e-04) (hash(x)=144467854)
355
+ 2930 train 4.154420 (lr=7.3863e-04) (hash(x)=153784048)
356
+ 2940 train 4.451596 (lr=7.3740e-04) (hash(x)=154774534)
357
+ 2950 train 4.297555 (lr=7.3616e-04) (hash(x)=143988507)
358
+ 2960 train 4.186791 (lr=7.3492e-04) (hash(x)=159552857)
359
+ 2970 train 4.471935 (lr=7.3368e-04) (hash(x)=175539144)
360
+ 2980 train 4.354757 (lr=7.3243e-04) (hash(x)=158172686)
361
+ 2990 train 4.237168 (lr=7.3118e-04) (hash(x)=147553039)
362
+ 3000 val loss 4.3527
363
+ 3000 val perplexity 77.6864
364
+ 3000 train 4.158875 (lr=7.2992e-04) (hash(x)=172449837)
365
+ 3010 train 4.046667 (lr=7.2867e-04) (hash(x)=153285123)
366
+ 3020 train 4.228870 (lr=7.2740e-04) (hash(x)=145307252)
367
+ 3030 train 4.277246 (lr=7.2614e-04) (hash(x)=147739766)
368
+ 3040 train 3.992382 (lr=7.2487e-04) (hash(x)=154414400)
369
+ 3050 train 4.081887 (lr=7.2360e-04) (hash(x)=150696487)
370
+ 3060 train 4.090200 (lr=7.2232e-04) (hash(x)=154543441)
371
+ 3070 train 3.870064 (lr=7.2105e-04) (hash(x)=170733242)
372
+ 3080 train 4.220905 (lr=7.1977e-04) (hash(x)=161914669)
373
+ 3090 train 4.105847 (lr=7.1848e-04) (hash(x)=156125491)
374
+ 3100 val loss 4.3684
375
+ 3100 val perplexity 78.9202
376
+ 3100 train 4.002134 (lr=7.1719e-04) (hash(x)=141710086)
377
+ 3110 train 4.309513 (lr=7.1590e-04) (hash(x)=154547635)
378
+ 3120 train 4.151263 (lr=7.1461e-04) (hash(x)=151747171)
379
+ 3130 train 4.436804 (lr=7.1331e-04) (hash(x)=160311353)
380
+ 3140 train 4.547508 (lr=7.1201e-04) (hash(x)=157233821)
381
+ 3150 train 4.357860 (lr=7.1071e-04) (hash(x)=153016810)
382
+ 3160 train 4.335472 (lr=7.0940e-04) (hash(x)=157289875)
383
+ 3170 train 4.345189 (lr=7.0809e-04) (hash(x)=157459484)
384
+ 3180 train 4.254673 (lr=7.0678e-04) (hash(x)=162037435)
385
+ 3190 train 4.355122 (lr=7.0546e-04) (hash(x)=159360169)
386
+ 3200 val loss 4.3150
387
+ 3200 val perplexity 74.8120
388
+ 3200 train 4.233161 (lr=7.0414e-04) (hash(x)=151299772)
389
+ 3210 train 4.457440 (lr=7.0282e-04) (hash(x)=183865364)
390
+ 3220 train 4.208958 (lr=7.0149e-04) (hash(x)=139622592)
391
+ 3230 train 4.163774 (lr=7.0017e-04) (hash(x)=147102655)
392
+ 3240 train 4.218401 (lr=6.9883e-04) (hash(x)=161870696)
393
+ 3250 train 4.370385 (lr=6.9750e-04) (hash(x)=152692231)
394
+ 3260 train 4.116918 (lr=6.9616e-04) (hash(x)=149990134)
395
+ 3270 train 4.090548 (lr=6.9482e-04) (hash(x)=134848432)
396
+ 3280 train 4.152534 (lr=6.9348e-04) (hash(x)=138519418)
397
+ 3290 train 4.290177 (lr=6.9213e-04) (hash(x)=139301249)
398
+ 3300 val loss 4.3194
399
+ 3300 val perplexity 75.1455
400
+ 3300 train 4.125785 (lr=6.9079e-04) (hash(x)=146473110)
401
+ 3310 train 4.196344 (lr=6.8943e-04) (hash(x)=144833685)
402
+ 3320 train 4.095662 (lr=6.8808e-04) (hash(x)=148953545)
403
+ 3330 train 4.327266 (lr=6.8672e-04) (hash(x)=161302371)
404
+ 3340 train 4.258191 (lr=6.8536e-04) (hash(x)=144215411)
405
+ 3350 train 4.424178 (lr=6.8400e-04) (hash(x)=157713664)
406
+ 3360 train 4.302905 (lr=6.8263e-04) (hash(x)=150813928)
407
+ 3370 train 4.284772 (lr=6.8127e-04) (hash(x)=150889463)
408
+ 3380 train 4.328586 (lr=6.7990e-04) (hash(x)=156631928)
409
+ 3390 train 4.366568 (lr=6.7852e-04) (hash(x)=136001579)
410
+ 3400 val loss 4.2932
411
+ 3400 val perplexity 73.1971
412
+ 3400 train 4.439957 (lr=6.7715e-04) (hash(x)=153954157)
413
+ 3410 train 4.218873 (lr=6.7577e-04) (hash(x)=150012477)
414
+ 3420 train 4.581367 (lr=6.7439e-04) (hash(x)=160878899)
415
+ 3430 train 4.617033 (lr=6.7300e-04) (hash(x)=128131106)
416
+ 3440 train 4.536747 (lr=6.7162e-04) (hash(x)=149611866)
417
+ 3450 train 4.166998 (lr=6.7023e-04) (hash(x)=166942460)
418
+ 3460 train 4.321686 (lr=6.6883e-04) (hash(x)=150498414)
419
+ 3470 train 4.471635 (lr=6.6744e-04) (hash(x)=144342279)
420
+ 3480 train 4.304661 (lr=6.6604e-04) (hash(x)=148891764)
421
+ 3490 train 4.269608 (lr=6.6464e-04) (hash(x)=157308188)
422
+ 3500 val loss 4.2835
423
+ 3500 val perplexity 72.4948
424
+ 3500 train 4.171542 (lr=6.6324e-04) (hash(x)=153717336)
425
+ 3510 train 4.203552 (lr=6.6184e-04) (hash(x)=156083924)
426
+ 3520 train 4.192437 (lr=6.6043e-04) (hash(x)=144234966)
427
+ 3530 train 4.277328 (lr=6.5902e-04) (hash(x)=150184381)
428
+ 3540 train 4.159635 (lr=6.5761e-04) (hash(x)=143011198)
429
+ 3550 train 4.308262 (lr=6.5620e-04) (hash(x)=148764613)
430
+ 3560 train 4.067451 (lr=6.5478e-04) (hash(x)=150625329)
431
+ 3570 train 4.185779 (lr=6.5336e-04) (hash(x)=143981302)
432
+ 3580 train 4.230360 (lr=6.5194e-04) (hash(x)=154989703)
433
+ 3590 train 4.361461 (lr=6.5052e-04) (hash(x)=146602526)
434
+ 3600 val loss 4.2775
435
+ 3600 val perplexity 72.0628
436
+ 3600 train 4.057448 (lr=6.4910e-04) (hash(x)=144965161)
437
+ 3610 train 4.050899 (lr=6.4767e-04) (hash(x)=159452243)
438
+ 3620 train 4.044521 (lr=6.4624e-04) (hash(x)=144904300)
439
+ 3630 train 3.959744 (lr=6.4481e-04) (hash(x)=127790212)
440
+ 3640 train 4.203071 (lr=6.4337e-04) (hash(x)=150863158)
441
+ 3650 train 4.158432 (lr=6.4194e-04) (hash(x)=145691635)
442
+ 3660 train 4.137704 (lr=6.4050e-04) (hash(x)=150631264)
443
+ 3670 train 4.113419 (lr=6.3906e-04) (hash(x)=148646259)
444
+ 3680 train 4.078874 (lr=6.3762e-04) (hash(x)=139317386)
445
+ 3690 train 4.192668 (lr=6.3617e-04) (hash(x)=142206916)
446
+ 3700 val loss 4.2915
447
+ 3700 val perplexity 73.0792
448
+ 3700 train 4.193795 (lr=6.3473e-04) (hash(x)=125969741)
449
+ 3710 train 4.040253 (lr=6.3328e-04) (hash(x)=157183915)
450
+ 3720 train 4.034634 (lr=6.3183e-04) (hash(x)=150434369)
451
+ 3730 train 4.223579 (lr=6.3037e-04) (hash(x)=160911722)
452
+ 3740 train 4.123976 (lr=6.2892e-04) (hash(x)=163306921)
453
+ 3750 train 4.236686 (lr=6.2746e-04) (hash(x)=149156876)
454
+ 3760 train 4.377484 (lr=6.2600e-04) (hash(x)=156015970)
455
+ 3770 train 4.315313 (lr=6.2454e-04) (hash(x)=148222322)
456
+ 3780 train 4.424157 (lr=6.2308e-04) (hash(x)=172040174)
457
+ 3790 train 4.335230 (lr=6.2162e-04) (hash(x)=151593070)
458
+ 3800 val loss 4.2653
459
+ 3800 val perplexity 71.1889
460
+ 3800 train 4.083443 (lr=6.2015e-04) (hash(x)=155070487)
461
+ 3810 train 4.273310 (lr=6.1868e-04) (hash(x)=137073164)
462
+ 3820 train 4.172125 (lr=6.1721e-04) (hash(x)=147761510)
463
+ 3830 train 4.089846 (lr=6.1574e-04) (hash(x)=154763021)
464
+ 3840 train 4.142411 (lr=6.1427e-04) (hash(x)=149622435)
465
+ 3850 train 4.252018 (lr=6.1280e-04) (hash(x)=150779423)
466
+ 3860 train 4.217759 (lr=6.1132e-04) (hash(x)=139445120)
467
+ 3870 train 4.202183 (lr=6.0984e-04) (hash(x)=151884947)
468
+ 3880 train 4.207698 (lr=6.0836e-04) (hash(x)=139795201)
469
+ 3890 train 4.252852 (lr=6.0688e-04) (hash(x)=155589956)
470
+ 3900 val loss 4.2530
471
+ 3900 val perplexity 70.3159
472
+ 3900 train 4.177105 (lr=6.0540e-04) (hash(x)=149444644)
473
+ 3910 train 4.164174 (lr=6.0391e-04) (hash(x)=148538323)
474
+ 3920 train 4.288862 (lr=6.0243e-04) (hash(x)=144026925)
475
+ 3930 train 4.174507 (lr=6.0094e-04) (hash(x)=155311200)
476
+ 3940 train 4.154124 (lr=5.9945e-04) (hash(x)=142410034)
477
+ 3950 train 4.130711 (lr=5.9796e-04) (hash(x)=145986075)
478
+ 3960 train 4.029598 (lr=5.9647e-04) (hash(x)=156197753)
479
+ 3970 train 4.183270 (lr=5.9497e-04) (hash(x)=146981564)
480
+ 3980 train 4.496033 (lr=5.9348e-04) (hash(x)=145361976)
481
+ 3990 train 4.222556 (lr=5.9198e-04) (hash(x)=149344475)
482
+ 4000 val loss 4.2462
483
+ 4000 val perplexity 69.8369
484
+ 4000 train 4.104761 (lr=5.9048e-04) (hash(x)=151663033)
485
+ 4010 train 3.988742 (lr=5.8898e-04) (hash(x)=140915499)
486
+ 4020 train 4.205299 (lr=5.8748e-04) (hash(x)=159979134)
487
+ 4030 train 4.128789 (lr=5.8598e-04) (hash(x)=163081433)
488
+ 4040 train 4.017813 (lr=5.8448e-04) (hash(x)=144890231)
489
+ 4050 train 4.180600 (lr=5.8297e-04) (hash(x)=138242368)
490
+ 4060 train 4.075369 (lr=5.8147e-04) (hash(x)=144438406)
491
+ 4070 train 3.959241 (lr=5.7996e-04) (hash(x)=159294906)
492
+ 4080 train 4.015474 (lr=5.7845e-04) (hash(x)=143676301)
493
+ 4090 train 4.247212 (lr=5.7694e-04) (hash(x)=147875296)
494
+ 4100 val loss 4.2590
495
+ 4100 val perplexity 70.7399
496
+ 4100 train 4.307348 (lr=5.7543e-04) (hash(x)=143688282)
497
+ 4110 train 4.078314 (lr=5.7392e-04) (hash(x)=150086777)
498
+ 4120 train 4.030775 (lr=5.7240e-04) (hash(x)=154624391)
499
+ 4130 train 4.216658 (lr=5.7089e-04) (hash(x)=141163032)
500
+ 4140 train 4.241447 (lr=5.6937e-04) (hash(x)=151038180)
501
+ 4150 train 4.319368 (lr=5.6786e-04) (hash(x)=144718421)
502
+ 4160 train 4.237252 (lr=5.6634e-04) (hash(x)=154277459)
503
+ 4170 train 4.396443 (lr=5.6482e-04) (hash(x)=143765173)
504
+ 4180 train 4.264229 (lr=5.6330e-04) (hash(x)=161013425)
505
+ 4190 train 4.215936 (lr=5.6178e-04) (hash(x)=161666854)
506
+ 4200 val loss 4.2260
507
+ 4200 val perplexity 68.4457
508
+ 4200 train 4.140808 (lr=5.6026e-04) (hash(x)=163361651)
509
+ 4210 train 4.476238 (lr=5.5874e-04) (hash(x)=153491528)
510
+ 4220 train 4.268763 (lr=5.5721e-04) (hash(x)=164410385)
511
+ 4230 train 4.206676 (lr=5.5569e-04) (hash(x)=175191816)
512
+ 4240 train 4.082651 (lr=5.5416e-04) (hash(x)=133924818)
513
+ 4250 train 4.307231 (lr=5.5264e-04) (hash(x)=152932845)
514
+ 4260 train 4.181491 (lr=5.5111e-04) (hash(x)=141987318)
515
+ 4270 train 4.275719 (lr=5.4958e-04) (hash(x)=142521279)
516
+ 4280 train 4.140427 (lr=5.4805e-04) (hash(x)=147707669)
517
+ 4290 train 4.200632 (lr=5.4652e-04) (hash(x)=152245539)
518
+ 4300 val loss 4.2102
519
+ 4300 val perplexity 67.3696
520
+ 4300 train 4.097079 (lr=5.4499e-04) (hash(x)=153619361)
521
+ 4310 train 4.301521 (lr=5.4346e-04) (hash(x)=142614473)
522
+ 4320 train 4.121789 (lr=5.4193e-04) (hash(x)=142689858)
523
+ 4330 train 4.145312 (lr=5.4040e-04) (hash(x)=140837891)
524
+ 4340 train 4.133393 (lr=5.3887e-04) (hash(x)=131790760)
525
+ 4350 train 4.297874 (lr=5.3733e-04) (hash(x)=147186292)
526
+ 4360 train 4.104776 (lr=5.3580e-04) (hash(x)=148140562)
527
+ 4370 train 4.215547 (lr=5.3427e-04) (hash(x)=156409913)
528
+ 4380 train 4.132679 (lr=5.3273e-04) (hash(x)=145126034)
529
+ 4390 train 4.082170 (lr=5.3119e-04) (hash(x)=147368423)
530
+ 4400 val loss 4.2149
531
+ 4400 val perplexity 67.6866
532
+ 4400 train 4.560713 (lr=5.2966e-04) (hash(x)=168527064)
533
+ 4410 train 4.073971 (lr=5.2812e-04) (hash(x)=150152773)
534
+ 4420 train 4.156668 (lr=5.2658e-04) (hash(x)=155472304)
535
+ 4430 train 4.085621 (lr=5.2505e-04) (hash(x)=147299265)
536
+ 4440 train 4.045893 (lr=5.2351e-04) (hash(x)=140659522)
537
+ 4450 train 4.111748 (lr=5.2197e-04) (hash(x)=147169111)
538
+ 4460 train 4.173145 (lr=5.2043e-04) (hash(x)=144098557)
539
+ 4470 train 3.937320 (lr=5.1889e-04) (hash(x)=161957507)
540
+ 4480 train 3.959991 (lr=5.1735e-04) (hash(x)=154064204)
541
+ 4490 train 4.102057 (lr=5.1581e-04) (hash(x)=153999958)
542
+ 4500 val loss 4.2278
543
+ 4500 val perplexity 68.5634
544
+ 4500 train 4.254476 (lr=5.1427e-04) (hash(x)=125588037)
545
+ 4510 train 4.056396 (lr=5.1273e-04) (hash(x)=142522186)
546
+ 4520 train 3.901035 (lr=5.1119e-04) (hash(x)=152736893)
547
+ 4530 train 4.127179 (lr=5.0965e-04) (hash(x)=148091275)
548
+ 4540 train 4.247996 (lr=5.0811e-04) (hash(x)=147663019)
549
+ 4550 train 4.331471 (lr=5.0657e-04) (hash(x)=151491857)
550
+ 4560 train 4.212407 (lr=5.0502e-04) (hash(x)=161757513)
551
+ 4570 train 4.203509 (lr=5.0348e-04) (hash(x)=165448260)
552
+ 4580 train 4.339612 (lr=5.0194e-04) (hash(x)=151620292)
553
+ 4590 train 4.421757 (lr=5.0040e-04) (hash(x)=161877852)
554
+ 4600 val loss 4.1901
555
+ 4600 val perplexity 66.0276
556
+ 4600 train 4.175506 (lr=4.9886e-04) (hash(x)=143710941)
557
+ 4610 train 4.269215 (lr=4.9731e-04) (hash(x)=152244871)
558
+ 4620 train 4.214794 (lr=4.9577e-04) (hash(x)=161268167)
559
+ 4630 train 4.202896 (lr=4.9423e-04) (hash(x)=153904677)
560
+ 4640 train 4.159828 (lr=4.9269e-04) (hash(x)=152593758)
561
+ 4650 train 4.311742 (lr=4.9114e-04) (hash(x)=159047122)
562
+ 4660 train 4.377599 (lr=4.8960e-04) (hash(x)=167370244)
563
+ 4670 train 4.153496 (lr=4.8806e-04) (hash(x)=140357684)
564
+ 4680 train 4.313177 (lr=4.8652e-04) (hash(x)=140547853)
565
+ 4690 train 4.490807 (lr=4.8498e-04) (hash(x)=143000949)
566
+ 4700 val loss 4.1845
567
+ 4700 val perplexity 65.6593
568
+ 4700 train 4.126832 (lr=4.8343e-04) (hash(x)=150952742)
569
+ 4710 train 4.167521 (lr=4.8189e-04) (hash(x)=157964406)
570
+ 4720 train 4.055173 (lr=4.8035e-04) (hash(x)=147653608)
571
+ 4730 train 4.134682 (lr=4.7881e-04) (hash(x)=165545764)
572
+ 4740 train 4.123727 (lr=4.7727e-04) (hash(x)=146744631)
573
+ 4750 train 4.198169 (lr=4.7573e-04) (hash(x)=153635164)
574
+ 4760 train 4.244863 (lr=4.7419e-04) (hash(x)=153217508)
575
+ 4770 train 4.280272 (lr=4.7265e-04) (hash(x)=169622743)
576
+ 4780 train 4.120941 (lr=4.7111e-04) (hash(x)=138018297)
577
+ 4790 train 3.992206 (lr=4.6957e-04) (hash(x)=156067303)
578
+ 4800 val loss 4.1844
579
+ 4800 val perplexity 65.6528
580
+ 4800 train 4.168067 (lr=4.6803e-04) (hash(x)=145323659)
581
+ 4810 train 4.052182 (lr=4.6649e-04) (hash(x)=152587772)
582
+ 4820 train 4.280380 (lr=4.6495e-04) (hash(x)=148481485)
583
+ 4830 train 4.042943 (lr=4.6342e-04) (hash(x)=154219664)
584
+ 4840 train 3.929765 (lr=4.6188e-04) (hash(x)=156853645)
585
+ 4850 train 4.017748 (lr=4.6034e-04) (hash(x)=152329702)
586
+ 4860 train 4.007750 (lr=4.5881e-04) (hash(x)=136684779)
587
+ 4870 train 4.066093 (lr=4.5727e-04) (hash(x)=143260367)
588
+ 4880 train 4.035643 (lr=4.5573e-04) (hash(x)=150910947)
589
+ 4890 train 4.054811 (lr=4.5420e-04) (hash(x)=179249925)
590
+ 4900 val loss 4.1902
591
+ 4900 val perplexity 66.0359
592
+ 4900 train 4.213989 (lr=4.5267e-04) (hash(x)=153151397)
593
+ 4910 train 4.163309 (lr=4.5113e-04) (hash(x)=144101891)
594
+ 4920 train 4.059049 (lr=4.4960e-04) (hash(x)=144994952)
595
+ 4930 train 4.346812 (lr=4.4807e-04) (hash(x)=155390878)
596
+ 4940 train 4.124098 (lr=4.4654e-04) (hash(x)=141359221)
597
+ 4950 train 4.031269 (lr=4.4501e-04) (hash(x)=151570011)
598
+ 4960 train 4.248791 (lr=4.4348e-04) (hash(x)=146918855)
599
+ 4970 train 4.181933 (lr=4.4195e-04) (hash(x)=148647362)
600
+ 4980 train 4.136421 (lr=4.4042e-04) (hash(x)=159753562)
601
+ 4990 train 4.083972 (lr=4.3889e-04) (hash(x)=147683447)
602
+ 5000 val loss 4.1808
603
+ 5000 val perplexity 65.4169
604
+ 5000 train 4.226419 (lr=4.3736e-04) (hash(x)=143182059)
605
+ 5010 train 4.031024 (lr=4.3584e-04) (hash(x)=138657407)
606
+ 5020 train 4.144058 (lr=4.3431e-04) (hash(x)=151830497)
607
+ 5030 train 4.144313 (lr=4.3279e-04) (hash(x)=147727694)
608
+ 5040 train 3.984268 (lr=4.3126e-04) (hash(x)=151957014)
609
+ 5050 train 4.252203 (lr=4.2974e-04) (hash(x)=147828693)
610
+ 5060 train 4.303446 (lr=4.2822e-04) (hash(x)=158589700)
611
+ 5070 train 4.188478 (lr=4.2670e-04) (hash(x)=155662163)
612
+ 5080 train 4.286661 (lr=4.2518e-04) (hash(x)=157797383)
613
+ 5090 train 4.164822 (lr=4.2366e-04) (hash(x)=153848957)
614
+ 5100 val loss 4.1655
615
+ 5100 val perplexity 64.4260
616
+ 5100 train 4.116725 (lr=4.2214e-04) (hash(x)=170083586)
617
+ 5110 train 4.039621 (lr=4.2063e-04) (hash(x)=153800138)
618
+ 5120 train 4.144315 (lr=4.1911e-04) (hash(x)=153526228)
619
+ 5130 train 4.153275 (lr=4.1760e-04) (hash(x)=158650342)
620
+ 5140 train 4.145780 (lr=4.1608e-04) (hash(x)=153163034)
621
+ 5150 train 4.134039 (lr=4.1457e-04) (hash(x)=155281294)
622
+ 5160 train 4.116432 (lr=4.1306e-04) (hash(x)=147731627)
623
+ 5170 train 4.102030 (lr=4.1155e-04) (hash(x)=146388976)
624
+ 5180 train 4.311785 (lr=4.1004e-04) (hash(x)=140323065)
625
+ 5190 train 4.225274 (lr=4.0853e-04) (hash(x)=150406523)
626
+ 5200 val loss 4.1611
627
+ 5200 val perplexity 64.1429
628
+ 5200 train 4.018374 (lr=4.0703e-04) (hash(x)=149363919)
629
+ 5210 train 4.157623 (lr=4.0552e-04) (hash(x)=164390214)
630
+ 5220 train 4.116515 (lr=4.0402e-04) (hash(x)=153652984)
631
+ 5230 train 4.083348 (lr=4.0252e-04) (hash(x)=139542770)
632
+ 5240 train 3.970414 (lr=4.0102e-04) (hash(x)=140115288)
633
+ 5250 train 4.095337 (lr=3.9952e-04) (hash(x)=142158900)
634
+ 5260 train 4.025528 (lr=3.9802e-04) (hash(x)=156019449)
635
+ 5270 train 4.135943 (lr=3.9652e-04) (hash(x)=150776850)
636
+ 5280 train 4.096885 (lr=3.9503e-04) (hash(x)=151358692)
637
+ 5290 train 4.064167 (lr=3.9353e-04) (hash(x)=137112863)
638
+ 5300 val loss 4.1568
639
+ 5300 val perplexity 63.8681
640
+ 5300 train 3.995258 (lr=3.9204e-04) (hash(x)=152033784)
641
+ 5310 train 4.029696 (lr=3.9055e-04) (hash(x)=147785712)
642
+ 5320 train 4.168192 (lr=3.8906e-04) (hash(x)=156426339)
643
+ 5330 train 4.115884 (lr=3.8757e-04) (hash(x)=161907693)
644
+ 5340 train 4.186275 (lr=3.8609e-04) (hash(x)=156414094)
645
+ 5350 train 4.209031 (lr=3.8460e-04) (hash(x)=141712428)
646
+ 5360 train 4.166499 (lr=3.8312e-04) (hash(x)=156873601)
647
+ 5370 train 4.137713 (lr=3.8164e-04) (hash(x)=149863781)
648
+ 5380 train 4.145262 (lr=3.8016e-04) (hash(x)=148183971)
649
+ 5390 train 4.103902 (lr=3.7868e-04) (hash(x)=151542334)
650
+ 5400 val loss 4.1355
651
+ 5400 val perplexity 62.5211
652
+ 5400 train 4.379809 (lr=3.7720e-04) (hash(x)=154614289)
653
+ 5410 train 4.122404 (lr=3.7573e-04) (hash(x)=169124671)
654
+ 5420 train 4.147447 (lr=3.7426e-04) (hash(x)=159963738)
655
+ 5430 train 4.107211 (lr=3.7279e-04) (hash(x)=155162691)
656
+ 5440 train 4.137402 (lr=3.7132e-04) (hash(x)=135981908)
657
+ 5450 train 4.226522 (lr=3.6985e-04) (hash(x)=145055580)
658
+ 5460 train 4.087333 (lr=3.6838e-04) (hash(x)=150367825)
659
+ 5470 train 4.173470 (lr=3.6692e-04) (hash(x)=151034173)
660
+ 5480 train 4.089606 (lr=3.6546e-04) (hash(x)=155581766)
661
+ 5490 train 4.116200 (lr=3.6400e-04) (hash(x)=136973993)
662
+ 5500 val loss 4.1326
663
+ 5500 val perplexity 62.3377
664
+ 5500 train 4.288058 (lr=3.6254e-04) (hash(x)=157745174)
665
+ 5510 train 4.240202 (lr=3.6108e-04) (hash(x)=144628894)
666
+ 5520 train 4.123269 (lr=3.5963e-04) (hash(x)=135056684)
667
+ 5530 train 4.113727 (lr=3.5817e-04) (hash(x)=143418640)
668
+ 5540 train 4.089695 (lr=3.5672e-04) (hash(x)=144464984)
669
+ 5550 train 4.290135 (lr=3.5527e-04) (hash(x)=155888745)
670
+ 5560 train 3.957353 (lr=3.5383e-04) (hash(x)=149122683)
671
+ 5570 train 4.001768 (lr=3.5238e-04) (hash(x)=154573954)
672
+ 5580 train 4.102922 (lr=3.5094e-04) (hash(x)=149339361)
673
+ 5590 train 3.998308 (lr=3.4950e-04) (hash(x)=154738453)
674
+ 5600 val loss 4.1328
675
+ 5600 val perplexity 62.3515
676
+ 5600 train 4.113726 (lr=3.4806e-04) (hash(x)=147693222)
677
+ 5610 train 3.884556 (lr=3.4663e-04) (hash(x)=142140192)
678
+ 5620 train 4.022015 (lr=3.4519e-04) (hash(x)=148677209)
679
+ 5630 train 4.082187 (lr=3.4376e-04) (hash(x)=143495554)
680
+ 5640 train 4.171337 (lr=3.4233e-04) (hash(x)=153202964)
681
+ 5650 train 4.098948 (lr=3.4090e-04) (hash(x)=168724867)
682
+ 5660 train 4.066211 (lr=3.3948e-04) (hash(x)=142816746)
683
+ 5670 train 4.198560 (lr=3.3806e-04) (hash(x)=164119551)
684
+ 5680 train 4.053823 (lr=3.3664e-04) (hash(x)=157239121)
685
+ 5690 train 4.340280 (lr=3.3522e-04) (hash(x)=144662329)
686
+ 5700 val loss 4.1199
687
+ 5700 val perplexity 61.5555
688
+ 5700 train 4.050880 (lr=3.3380e-04) (hash(x)=149784627)
689
+ 5710 train 4.023172 (lr=3.3239e-04) (hash(x)=152821162)
690
+ 5720 train 4.239784 (lr=3.3098e-04) (hash(x)=158497530)
691
+ 5730 train 4.227987 (lr=3.2957e-04) (hash(x)=154583914)
692
+ 5740 train 4.044340 (lr=3.2816e-04) (hash(x)=142027172)
693
+ 5750 train 4.283438 (lr=3.2676e-04) (hash(x)=119227402)
694
+ 5760 train 4.136868 (lr=3.2536e-04) (hash(x)=153248187)
695
+ 5770 train 4.009287 (lr=3.2396e-04) (hash(x)=147951994)
696
+ 5780 train 4.022799 (lr=3.2256e-04) (hash(x)=138935284)
697
+ 5790 train 4.109351 (lr=3.2117e-04) (hash(x)=151786681)
698
+ 5800 val loss 4.1266
699
+ 5800 val perplexity 61.9672
700
+ 5800 train 4.144587 (lr=3.1977e-04) (hash(x)=158620729)
701
+ 5810 train 4.090469 (lr=3.1838e-04) (hash(x)=178115789)
702
+ 5820 train 4.015295 (lr=3.1700e-04) (hash(x)=149092790)
703
+ 5830 train 4.197200 (lr=3.1561e-04) (hash(x)=148269833)
704
+ 5840 train 4.129659 (lr=3.1423e-04) (hash(x)=163714738)
705
+ 5850 train 4.104927 (lr=3.1285e-04) (hash(x)=144111890)
706
+ 5860 train 4.066407 (lr=3.1148e-04) (hash(x)=170702475)
707
+ 5870 train 4.010305 (lr=3.1010e-04) (hash(x)=155970835)
708
+ 5880 train 4.016855 (lr=3.0873e-04) (hash(x)=146695463)
709
+ 5890 train 4.076689 (lr=3.0737e-04) (hash(x)=151603187)
710
+ 5900 val loss 4.1152
711
+ 5900 val perplexity 61.2661
712
+ 5900 train 4.093883 (lr=3.0600e-04) (hash(x)=159763910)
713
+ 5910 train 4.136925 (lr=3.0464e-04) (hash(x)=149555230)
714
+ 5920 train 4.146194 (lr=3.0328e-04) (hash(x)=138048333)
715
+ 5930 train 4.039069 (lr=3.0192e-04) (hash(x)=148668256)
716
+ 5940 train 4.176870 (lr=3.0057e-04) (hash(x)=148352543)
717
+ 5950 train 4.186347 (lr=2.9921e-04) (hash(x)=163457488)
718
+ 5960 train 4.268394 (lr=2.9787e-04) (hash(x)=155306804)
719
+ 5970 train 4.114334 (lr=2.9652e-04) (hash(x)=146576143)
720
+ 5980 train 4.123283 (lr=2.9518e-04) (hash(x)=144668956)
721
+ 5990 train 4.090063 (lr=2.9384e-04) (hash(x)=156098617)
722
+ 6000 val loss 4.1006
723
+ 6000 val perplexity 60.3761
724
+ 6000 train 4.066449 (lr=2.9250e-04) (hash(x)=147640561)
725
+ 6010 train 4.136495 (lr=2.9117e-04) (hash(x)=161223074)
726
+ 6020 train 4.123913 (lr=2.8983e-04) (hash(x)=156308454)
727
+ 6030 train 4.100221 (lr=2.8851e-04) (hash(x)=142553975)
728
+ 6040 train 4.063512 (lr=2.8718e-04) (hash(x)=142039829)
729
+ 6050 train 4.140831 (lr=2.8586e-04) (hash(x)=156329296)
730
+ 6060 train 4.068428 (lr=2.8454e-04) (hash(x)=148505537)
731
+ 6070 train 3.964927 (lr=2.8322e-04) (hash(x)=151801272)
732
+ 6080 train 4.041350 (lr=2.8191e-04) (hash(x)=159908287)
733
+ 6090 train 4.041447 (lr=2.8060e-04) (hash(x)=148714721)
734
+ 6100 val loss 4.0951
735
+ 6100 val perplexity 60.0461
736
+ 6100 train 4.109780 (lr=2.7929e-04) (hash(x)=156613394)
737
+ 6110 train 3.961176 (lr=2.7799e-04) (hash(x)=143286705)
738
+ 6120 train 4.172488 (lr=2.7669e-04) (hash(x)=150191141)
739
+ 6130 train 4.045365 (lr=2.7539e-04) (hash(x)=149855624)
740
+ 6140 train 4.035291 (lr=2.7410e-04) (hash(x)=157312372)
741
+ 6150 train 4.157441 (lr=2.7281e-04) (hash(x)=148262730)
742
+ 6160 train 4.071901 (lr=2.7152e-04) (hash(x)=147032785)
743
+ 6170 train 3.991714 (lr=2.7023e-04) (hash(x)=161213064)
744
+ 6180 train 3.870981 (lr=2.6895e-04) (hash(x)=144538373)
745
+ 6190 train 3.925468 (lr=2.6768e-04) (hash(x)=144688493)
746
+ 6200 val loss 4.0941
747
+ 6200 val perplexity 59.9849
748
+ 6200 train 3.998008 (lr=2.6640e-04) (hash(x)=186221290)
749
+ 6210 train 4.010006 (lr=2.6513e-04) (hash(x)=139529352)
750
+ 6220 train 4.175237 (lr=2.6386e-04) (hash(x)=138081469)
751
+ 6230 train 4.042695 (lr=2.6260e-04) (hash(x)=152979409)
752
+ 6240 train 4.025000 (lr=2.6133e-04) (hash(x)=157848424)
753
+ 6250 train 3.919714 (lr=2.6008e-04) (hash(x)=150770915)
754
+ 6260 train 4.011678 (lr=2.5882e-04) (hash(x)=153769379)
755
+ 6270 train 3.950395 (lr=2.5757e-04) (hash(x)=153318361)
756
+ 6280 train 3.942501 (lr=2.5632e-04) (hash(x)=143366327)
757
+ 6290 train 4.108884 (lr=2.5508e-04) (hash(x)=160684845)
758
+ 6300 val loss 4.0784
759
+ 6300 val perplexity 59.0527
760
+ 6300 train 3.992275 (lr=2.5384e-04) (hash(x)=152081419)
761
+ 6310 train 3.954185 (lr=2.5260e-04) (hash(x)=146626797)
762
+ 6320 train 4.135173 (lr=2.5137e-04) (hash(x)=155291923)
763
+ 6330 train 4.008918 (lr=2.5014e-04) (hash(x)=140920951)
764
+ 6340 train 4.084290 (lr=2.4891e-04) (hash(x)=146642359)
765
+ 6350 train 4.060712 (lr=2.4769e-04) (hash(x)=160486550)
766
+ 6360 train 4.017797 (lr=2.4647e-04) (hash(x)=151833082)
767
+ 6370 train 4.005771 (lr=2.4525e-04) (hash(x)=124740362)
768
+ 6380 train 4.157446 (lr=2.4404e-04) (hash(x)=153952434)
769
+ 6390 train 4.196740 (lr=2.4283e-04) (hash(x)=145864314)
770
+ 6400 val loss 4.0815
771
+ 6400 val perplexity 59.2348
772
+ 6400 train 4.078357 (lr=2.4163e-04) (hash(x)=154808349)
773
+ 6410 train 4.056865 (lr=2.4042e-04) (hash(x)=155037739)
774
+ 6420 train 3.991804 (lr=2.3923e-04) (hash(x)=151155588)
775
+ 6430 train 3.887644 (lr=2.3803e-04) (hash(x)=146548556)
776
+ 6440 train 4.056681 (lr=2.3684e-04) (hash(x)=152706474)
777
+ 6450 train 4.040814 (lr=2.3566e-04) (hash(x)=154207794)
778
+ 6460 train 3.918822 (lr=2.3447e-04) (hash(x)=141590515)
779
+ 6470 train 3.955927 (lr=2.3330e-04) (hash(x)=157673109)
780
+ 6480 train 3.969087 (lr=2.3212e-04) (hash(x)=155842980)
781
+ 6490 train 4.048855 (lr=2.3095e-04) (hash(x)=143595736)
782
+ 6500 val loss 4.0794
783
+ 6500 val perplexity 59.1105
784
+ 6500 train 3.992445 (lr=2.2978e-04) (hash(x)=159437208)
785
+ 6510 train 4.017516 (lr=2.2862e-04) (hash(x)=142631317)
786
+ 6520 train 3.737174 (lr=2.2746e-04) (hash(x)=139009211)
787
+ 6530 train 4.087727 (lr=2.2630e-04) (hash(x)=145451137)
788
+ 6540 train 4.036687 (lr=2.2515e-04) (hash(x)=163532779)
789
+ 6550 train 3.949154 (lr=2.2400e-04) (hash(x)=158084674)
790
+ 6560 train 3.930851 (lr=2.2286e-04) (hash(x)=142422592)
791
+ 6570 train 3.845227 (lr=2.2172e-04) (hash(x)=141089289)
792
+ 6580 train 3.948208 (lr=2.2058e-04) (hash(x)=139712979)
793
+ 6590 train 3.854307 (lr=2.1945e-04) (hash(x)=145833805)
794
+ 6600 val loss 4.0793
795
+ 6600 val perplexity 59.1041
796
+ 6600 train 3.926127 (lr=2.1832e-04) (hash(x)=157933074)
797
+ 6610 train 3.848492 (lr=2.1720e-04) (hash(x)=142534769)
798
+ 6620 train 3.902997 (lr=2.1608e-04) (hash(x)=138456044)
799
+ 6630 train 4.035989 (lr=2.1496e-04) (hash(x)=145111052)
800
+ 6640 train 3.995090 (lr=2.1385e-04) (hash(x)=166954515)
801
+ 6650 train 3.959563 (lr=2.1274e-04) (hash(x)=146889543)
802
+ 6660 train 3.991551 (lr=2.1164e-04) (hash(x)=143253726)
803
+ 6670 train 3.904911 (lr=2.1054e-04) (hash(x)=144349844)
804
+ 6680 train 4.093520 (lr=2.0944e-04) (hash(x)=139911928)
805
+ 6690 train 4.286952 (lr=2.0835e-04) (hash(x)=143135862)
806
+ 6700 val loss 4.0679
807
+ 6700 val perplexity 58.4357
808
+ 6700 train 4.000452 (lr=2.0726e-04) (hash(x)=161560240)
809
+ 6710 train 4.090009 (lr=2.0618e-04) (hash(x)=150437730)
810
+ 6720 train 4.193772 (lr=2.0510e-04) (hash(x)=153035304)
811
+ 6730 train 3.980577 (lr=2.0402e-04) (hash(x)=139323796)
812
+ 6740 train 3.985088 (lr=2.0295e-04) (hash(x)=167437923)
813
+ 6750 train 3.947841 (lr=2.0189e-04) (hash(x)=143814596)
814
+ 6760 train 4.016275 (lr=2.0083e-04) (hash(x)=143755031)
815
+ 6770 train 3.967458 (lr=1.9977e-04) (hash(x)=142014038)
816
+ 6780 train 4.029864 (lr=1.9871e-04) (hash(x)=148465062)
817
+ 6790 train 4.062628 (lr=1.9766e-04) (hash(x)=154656877)
818
+ 6800 val loss 4.0654
819
+ 6800 val perplexity 58.2886
820
+ 6800 train 4.173873 (lr=1.9662e-04) (hash(x)=155424292)
821
+ 6810 train 3.804549 (lr=1.9558e-04) (hash(x)=147358081)
822
+ 6820 train 3.926537 (lr=1.9454e-04) (hash(x)=145387291)
823
+ 6830 train 3.866898 (lr=1.9351e-04) (hash(x)=146334068)
824
+ 6840 train 3.954733 (lr=1.9248e-04) (hash(x)=139563766)
825
+ 6850 train 3.974477 (lr=1.9146e-04) (hash(x)=139095691)
826
+ 6860 train 3.911840 (lr=1.9044e-04) (hash(x)=142093999)
827
+ 6870 train 4.033524 (lr=1.8943e-04) (hash(x)=148598823)
828
+ 6880 train 4.116657 (lr=1.8842e-04) (hash(x)=146486085)
829
+ 6890 train 3.905026 (lr=1.8741e-04) (hash(x)=165683212)
830
+ 6900 val loss 4.0708
831
+ 6900 val perplexity 58.6019
832
+ 6900 train 4.104698 (lr=1.8641e-04) (hash(x)=148561470)
833
+ 6910 train 3.798106 (lr=1.8541e-04) (hash(x)=136352432)
834
+ 6920 train 4.166933 (lr=1.8442e-04) (hash(x)=147640094)
835
+ 6930 train 4.058279 (lr=1.8343e-04) (hash(x)=152343827)
836
+ 6940 train 4.036081 (lr=1.8245e-04) (hash(x)=158787971)
837
+ 6950 train 4.314651 (lr=1.8147e-04) (hash(x)=178133125)
838
+ 6960 train 4.102413 (lr=1.8050e-04) (hash(x)=167302853)
839
+ 6970 train 4.032199 (lr=1.7953e-04) (hash(x)=143729435)
840
+ 6980 train 4.132317 (lr=1.7856e-04) (hash(x)=135857879)
841
+ 6990 train 4.079530 (lr=1.7760e-04) (hash(x)=155384510)
842
+ 7000 val loss 4.0584
843
+ 7000 val perplexity 57.8821
844
+ 7000 train 4.024612 (lr=1.7665e-04) (hash(x)=141527450)
845
+ 7010 train 4.068355 (lr=1.7570e-04) (hash(x)=148648498)
846
+ 7020 train 3.994866 (lr=1.7475e-04) (hash(x)=148073842)
847
+ 7030 train 4.125118 (lr=1.7381e-04) (hash(x)=160771614)
848
+ 7040 train 3.905384 (lr=1.7287e-04) (hash(x)=148361330)
849
+ 7050 train 3.886260 (lr=1.7194e-04) (hash(x)=151212224)
850
+ 7060 train 4.078100 (lr=1.7101e-04) (hash(x)=155925366)
851
+ 7070 train 3.928343 (lr=1.7009e-04) (hash(x)=149830576)
852
+ 7080 train 3.895623 (lr=1.6917e-04) (hash(x)=144849934)
853
+ 7090 train 3.988890 (lr=1.6826e-04) (hash(x)=150997807)
854
+ 7100 val loss 4.0529
855
+ 7100 val perplexity 57.5653
856
+ 7100 train 4.024892 (lr=1.6735e-04) (hash(x)=151066339)
857
+ 7110 train 3.935751 (lr=1.6644e-04) (hash(x)=147185781)
858
+ 7120 train 3.973324 (lr=1.6554e-04) (hash(x)=141765575)
859
+ 7130 train 3.988152 (lr=1.6465e-04) (hash(x)=144204513)
860
+ 7140 train 3.824070 (lr=1.6376e-04) (hash(x)=149326253)
861
+ 7150 train 3.982937 (lr=1.6288e-04) (hash(x)=150409811)
862
+ 7160 train 3.869418 (lr=1.6200e-04) (hash(x)=144790348)
863
+ 7170 train 3.993319 (lr=1.6112e-04) (hash(x)=150500939)
864
+ 7180 train 3.815131 (lr=1.6025e-04) (hash(x)=143552103)
865
+ 7190 train 3.857952 (lr=1.5938e-04) (hash(x)=157471075)
866
+ 7200 val loss 4.0565
867
+ 7200 val perplexity 57.7697
868
+ 7200 train 3.839967 (lr=1.5852e-04) (hash(x)=155231264)
869
+ 7210 train 3.927274 (lr=1.5767e-04) (hash(x)=150350401)
870
+ 7220 train 3.940824 (lr=1.5682e-04) (hash(x)=151952594)
871
+ 7230 train 4.023512 (lr=1.5597e-04) (hash(x)=145381804)
872
+ 7240 train 3.906166 (lr=1.5513e-04) (hash(x)=142651222)
873
+ 7250 train 4.024863 (lr=1.5429e-04) (hash(x)=136115279)
874
+ 7260 train 4.124061 (lr=1.5346e-04) (hash(x)=154536751)
875
+ 7270 train 4.068368 (lr=1.5263e-04) (hash(x)=148883836)
876
+ 7280 train 3.923558 (lr=1.5181e-04) (hash(x)=150606754)
877
+ 7290 train 4.039410 (lr=1.5100e-04) (hash(x)=165845807)
878
+ 7300 val loss 4.0458
879
+ 7300 val perplexity 57.1560
880
+ 7300 train 4.119495 (lr=1.5019e-04) (hash(x)=150281149)
881
+ 7310 train 4.069716 (lr=1.4938e-04) (hash(x)=159929281)
882
+ 7320 train 3.900400 (lr=1.4858e-04) (hash(x)=143074269)
883
+ 7330 train 4.056135 (lr=1.4778e-04) (hash(x)=147369269)
884
+ 7340 train 4.106221 (lr=1.4699e-04) (hash(x)=139286601)
885
+ 7350 train 4.008608 (lr=1.4620e-04) (hash(x)=145263048)
886
+ 7360 train 3.863155 (lr=1.4542e-04) (hash(x)=144626817)
887
+ 7370 train 4.062545 (lr=1.4465e-04) (hash(x)=168989813)
888
+ 7380 train 4.041195 (lr=1.4387e-04) (hash(x)=144790656)
889
+ 7390 train 3.848826 (lr=1.4311e-04) (hash(x)=141279761)
890
+ 7400 val loss 4.0417
891
+ 7400 val perplexity 56.9226
892
+ 7400 train 3.978047 (lr=1.4235e-04) (hash(x)=148421717)
893
+ 7410 train 3.973784 (lr=1.4159e-04) (hash(x)=141241548)
894
+ 7420 train 4.049379 (lr=1.4084e-04) (hash(x)=162434228)
895
+ 7430 train 3.988369 (lr=1.4010e-04) (hash(x)=156312086)
896
+ 7440 train 3.949942 (lr=1.3936e-04) (hash(x)=139803064)
897
+ 7450 train 3.651820 (lr=1.3862e-04) (hash(x)=156361394)
898
+ 7460 train 3.955945 (lr=1.3789e-04) (hash(x)=141142183)
899
+ 7470 train 3.786820 (lr=1.3717e-04) (hash(x)=156701683)
900
+ 7480 train 3.963831 (lr=1.3645e-04) (hash(x)=148549917)
901
+ 7490 train 3.846266 (lr=1.3573e-04) (hash(x)=143002119)
902
+ 7500 val loss 4.0437
903
+ 7500 val perplexity 57.0362
904
+ 7500 train 3.852460 (lr=1.3502e-04) (hash(x)=146921118)
905
+ 7510 train 3.856812 (lr=1.3432e-04) (hash(x)=148592146)
906
+ 7520 train 4.005755 (lr=1.3362e-04) (hash(x)=148456120)
907
+ 7530 train 4.163059 (lr=1.3293e-04) (hash(x)=163865945)
908
+ 7540 train 4.054419 (lr=1.3224e-04) (hash(x)=160753039)
909
+ 7550 train 4.061856 (lr=1.3155e-04) (hash(x)=150275989)
910
+ 7560 train 4.090946 (lr=1.3088e-04) (hash(x)=141749735)
911
+ 7570 train 4.054995 (lr=1.3020e-04) (hash(x)=148449999)
912
+ 7580 train 3.931771 (lr=1.2954e-04) (hash(x)=139153889)
913
+ 7590 train 4.043928 (lr=1.2887e-04) (hash(x)=147418540)
914
+ 7600 val loss 4.0331
915
+ 7600 val perplexity 56.4353
916
+ 7600 train 4.071450 (lr=1.2822e-04) (hash(x)=150660048)
917
+ 7610 train 4.048304 (lr=1.2757e-04) (hash(x)=155369080)
918
+ 7620 train 3.951020 (lr=1.2692e-04) (hash(x)=151926309)
919
+ 7630 train 4.028627 (lr=1.2628e-04) (hash(x)=149549080)
920
+ 7640 train 3.917208 (lr=1.2564e-04) (hash(x)=140642882)
921
+ 7650 train 4.196084 (lr=1.2501e-04) (hash(x)=140218502)
922
+ 7660 train 4.027391 (lr=1.2439e-04) (hash(x)=153198276)
923
+ 7670 train 3.887356 (lr=1.2377e-04) (hash(x)=150706960)
924
+ 7680 train 4.038455 (lr=1.2316e-04) (hash(x)=152025509)
925
+ 7690 train 3.821627 (lr=1.2255e-04) (hash(x)=147288976)
926
+ 7700 val loss 4.0345
927
+ 7700 val perplexity 56.5171
928
+ 7700 train 3.933159 (lr=1.2194e-04) (hash(x)=148059852)
929
+ 7710 train 3.749440 (lr=1.2135e-04) (hash(x)=158443334)
930
+ 7720 train 3.855504 (lr=1.2076e-04) (hash(x)=143467085)
931
+ 7730 train 3.808020 (lr=1.2017e-04) (hash(x)=138439550)
932
+ 7740 train 3.942326 (lr=1.1959e-04) (hash(x)=145878152)
933
+ 7750 train 3.798401 (lr=1.1901e-04) (hash(x)=136874128)
934
+ 7760 train 3.824250 (lr=1.1844e-04) (hash(x)=148856154)
935
+ 7770 train 3.847606 (lr=1.1788e-04) (hash(x)=146982119)
936
+ 7780 train 3.948963 (lr=1.1732e-04) (hash(x)=159846779)
937
+ 7790 train 3.628645 (lr=1.1676e-04) (hash(x)=141501838)
938
+ 7800 val loss 4.0368
939
+ 7800 val perplexity 56.6428
940
+ 7800 train 3.762256 (lr=1.1621e-04) (hash(x)=148331002)
941
+ 7810 train 4.068841 (lr=1.1567e-04) (hash(x)=150692304)
942
+ 7820 train 3.939313 (lr=1.1513e-04) (hash(x)=156841183)
943
+ 7830 train 3.993525 (lr=1.1460e-04) (hash(x)=139594008)
944
+ 7840 train 3.925000 (lr=1.1407e-04) (hash(x)=149700138)
945
+ 7850 train 3.942293 (lr=1.1355e-04) (hash(x)=149327044)
946
+ 7860 train 4.108767 (lr=1.1304e-04) (hash(x)=148315418)
947
+ 7870 train 4.051887 (lr=1.1253e-04) (hash(x)=153431778)
948
+ 7880 train 3.891775 (lr=1.1202e-04) (hash(x)=145049799)
949
+ 7890 train 4.023592 (lr=1.1152e-04) (hash(x)=150265096)
950
+ 7900 val loss 4.0266
951
+ 7900 val perplexity 56.0706
952
+ 7900 train 3.853651 (lr=1.1103e-04) (hash(x)=164923883)
953
+ 7910 train 3.904125 (lr=1.1054e-04) (hash(x)=139315738)
954
+ 7920 train 3.954612 (lr=1.1006e-04) (hash(x)=153031394)
955
+ 7930 train 3.748331 (lr=1.0958e-04) (hash(x)=148835859)
956
+ 7940 train 3.882936 (lr=1.0911e-04) (hash(x)=146875217)
957
+ 7950 train 3.955657 (lr=1.0865e-04) (hash(x)=151003885)
958
+ 7960 train 3.869978 (lr=1.0819e-04) (hash(x)=141885471)
959
+ 7970 train 4.073985 (lr=1.0773e-04) (hash(x)=150541583)
960
+ 7980 train 3.775541 (lr=1.0729e-04) (hash(x)=138626492)
961
+ 7990 train 3.951455 (lr=1.0684e-04) (hash(x)=149935635)
962
+ 8000 val loss 4.0249
963
+ 8000 val perplexity 55.9754
964
+ 8000 train 3.926817 (lr=1.0641e-04) (hash(x)=143545384)
965
+ 8010 train 3.782365 (lr=1.0597e-04) (hash(x)=144689366)
966
+ 8020 train 3.963323 (lr=1.0555e-04) (hash(x)=151299532)
967
+ 8030 train 3.902810 (lr=1.0513e-04) (hash(x)=131300239)
968
+ 8040 train 3.801949 (lr=1.0471e-04) (hash(x)=146715955)
969
+ 8050 train 3.725806 (lr=1.0430e-04) (hash(x)=135376656)
970
+ 8060 train 3.813621 (lr=1.0390e-04) (hash(x)=141991433)
971
+ 8070 train 3.767188 (lr=1.0350e-04) (hash(x)=140260574)
972
+ 8080 train 3.738875 (lr=1.0311e-04) (hash(x)=151492896)
973
+ 8090 train 3.943200 (lr=1.0272e-04) (hash(x)=143333635)
974
+ 8100 val loss 4.0285
975
+ 8100 val perplexity 56.1738
976
+ 8100 train 3.820968 (lr=1.0234e-04) (hash(x)=160686959)
977
+ 8110 train 3.826267 (lr=1.0197e-04) (hash(x)=141145704)
978
+ 8120 train 4.060985 (lr=1.0160e-04) (hash(x)=146967795)
979
+ 8130 train 3.890823 (lr=1.0124e-04) (hash(x)=147844486)
980
+ 8140 train 3.946632 (lr=1.0088e-04) (hash(x)=141860136)
981
+ 8150 train 4.080942 (lr=1.0053e-04) (hash(x)=147696006)
982
+ 8160 train 3.983521 (lr=1.0018e-04) (hash(x)=151497110)
983
+ 8170 train 3.992022 (lr=9.9838e-05) (hash(x)=140881859)
984
+ 8180 train 3.932223 (lr=9.9503e-05) (hash(x)=158865574)
985
+ 8190 train 3.889981 (lr=9.9174e-05) (hash(x)=161247161)
986
+ 8200 val loss 4.0225
987
+ 8200 val perplexity 55.8408
988
+ 8200 train 3.850178 (lr=9.8850e-05) (hash(x)=156501889)
989
+ 8210 train 3.979988 (lr=9.8532e-05) (hash(x)=158384162)
990
+ 8220 train 4.022681 (lr=9.8220e-05) (hash(x)=154815062)
991
+ 8230 train 3.897854 (lr=9.7914e-05) (hash(x)=145018315)
992
+ 8240 train 3.936351 (lr=9.7614e-05) (hash(x)=161659129)
993
+ 8250 train 3.906562 (lr=9.7319e-05) (hash(x)=139863367)
994
+ 8260 train 3.859026 (lr=9.7030e-05) (hash(x)=149569663)
995
+ 8270 train 3.617335 (lr=9.6747e-05) (hash(x)=162234378)
996
+ 8280 train 4.000114 (lr=9.6469e-05) (hash(x)=144371378)
997
+ 8290 train 3.896068 (lr=9.6198e-05) (hash(x)=148120258)
998
+ 8300 val loss 4.0237
999
+ 8300 val perplexity 55.9085
1000
+ 8300 train 3.892864 (lr=9.5932e-05) (hash(x)=142716875)
1001
+ 8310 train 3.914649 (lr=9.5672e-05) (hash(x)=166789234)
1002
+ 8320 train 3.977618 (lr=9.5417e-05) (hash(x)=150613211)
1003
+ 8330 train 3.857953 (lr=9.5169e-05) (hash(x)=139895391)
1004
+ 8340 train 3.969440 (lr=9.4926e-05) (hash(x)=152011203)
1005
+ 8350 train 3.923173 (lr=9.4689e-05) (hash(x)=156010875)
1006
+ 8360 train 3.935249 (lr=9.4458e-05) (hash(x)=152988051)
1007
+ 8370 train 4.098235 (lr=9.4233e-05) (hash(x)=155144327)
1008
+ 8380 train 3.983262 (lr=9.4013e-05) (hash(x)=133642309)
1009
+ 8390 train 4.024120 (lr=9.3800e-05) (hash(x)=149745335)
1010
+ 8400 val loss 4.0155
1011
+ 8400 val perplexity 55.4504
1012
+ 8400 train 4.009270 (lr=9.3592e-05) (hash(x)=154436684)
1013
+ 8410 train 4.054511 (lr=9.3390e-05) (hash(x)=147946630)
1014
+ 8420 train 3.961271 (lr=9.3194e-05) (hash(x)=153947058)
1015
+ 8430 train 4.056878 (lr=9.3003e-05) (hash(x)=159821040)
1016
+ 8440 train 4.020013 (lr=9.2819e-05) (hash(x)=149127348)
1017
+ 8450 train 4.031244 (lr=9.2640e-05) (hash(x)=149421871)
1018
+ 8460 train 3.994771 (lr=9.2467e-05) (hash(x)=142621577)
1019
+ 8470 train 3.990736 (lr=9.2300e-05) (hash(x)=148978213)
1020
+ 8480 train 3.809310 (lr=9.2139e-05) (hash(x)=139959537)
1021
+ 8490 train 3.902503 (lr=9.1983e-05) (hash(x)=144232498)
1022
+ 8500 val loss 4.0149
1023
+ 8500 val perplexity 55.4152
1024
+ 8500 train 4.116467 (lr=9.1834e-05) (hash(x)=147965839)
1025
+ 8510 train 3.998392 (lr=9.1690e-05) (hash(x)=147679836)
1026
+ 8520 train 3.875745 (lr=9.1552e-05) (hash(x)=147486097)
1027
+ 8530 train 4.018629 (lr=9.1420e-05) (hash(x)=151655265)
1028
+ 8540 train 3.918418 (lr=9.1294e-05) (hash(x)=150866632)
1029
+ 8550 train 4.059164 (lr=9.1174e-05) (hash(x)=144355059)
1030
+ 8560 train 3.921353 (lr=9.1060e-05) (hash(x)=145518478)
1031
+ 8570 train 4.073595 (lr=9.0951e-05) (hash(x)=178973323)
1032
+ 8580 train 3.987272 (lr=9.0848e-05) (hash(x)=152773969)
1033
+ 8590 train 3.782785 (lr=9.0751e-05) (hash(x)=141805508)
1034
+ 8600 val loss 4.0183
1035
+ 8600 val perplexity 55.6088
1036
+ 8600 train 3.838658 (lr=9.0661e-05) (hash(x)=145228097)
1037
+ 8610 train 3.861017 (lr=9.0575e-05) (hash(x)=149525694)
1038
+ 8620 train 3.929534 (lr=9.0496e-05) (hash(x)=151651655)
1039
+ 8630 train 4.041018 (lr=9.0423e-05) (hash(x)=144918643)
1040
+ 8640 train 3.889800 (lr=9.0355e-05) (hash(x)=149721879)
1041
+ 8650 train 3.964548 (lr=9.0294e-05) (hash(x)=157684127)
1042
+ 8660 train 3.971960 (lr=9.0238e-05) (hash(x)=152426036)
1043
+ 8670 train 3.890494 (lr=9.0188e-05) (hash(x)=141396342)
1044
+ 8680 train 3.982955 (lr=9.0144e-05) (hash(x)=146459856)
1045
+ 8690 train 4.067213 (lr=9.0106e-05) (hash(x)=151766124)
1046
+ 8700 val loss 4.0104
1047
+ 8700 val perplexity 55.1701
1048
+ 8700 train 4.218161 (lr=9.0073e-05) (hash(x)=152910357)
1049
+ 8710 train 4.012752 (lr=9.0047e-05) (hash(x)=149724642)
1050
+ 8720 train 3.944841 (lr=9.0026e-05) (hash(x)=152483840)
1051
+ 8730 train 3.921543 (lr=9.0012e-05) (hash(x)=147289415)
1052
+ 8740 train 3.947018 (lr=9.0003e-05) (hash(x)=152254131)
1053
+ 8749 val loss 4.0098
1054
+ 8749 val perplexity 55.1366
lr9e-4_total_batch_size61440_baseline_seed1338/model_08749.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cab8593aa0d8cbcdaadf9705c7c9d91c2ae7422fee26be98a9d1521fead19edb
3
+ size 92843394
lr9e-4_total_batch_size61440_baseline_seed1338/optimizer_08749.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d00e8629084c6f86c366030452f9ac8b706628e08260caa319b660006b1fefc2
3
+ size 179406214