andrew-healey commited on
Commit
b395ba9
·
verified ·
1 Parent(s): 16ddcb8

Upload folder using huggingface_hub

Browse files
lr2e-4_total_batch_size61440_baseline_seed1338/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_11/lr2e-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": 0.0002, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "2e-4_61440", "n_embd": 256}
lr2e-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
lr2e-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 74626.2266
4
+ 0 train 11.227316 (lr=4.0000e-07) (hash(x)=150327452)
5
+ 10 train 10.683886 (lr=4.4000e-06) (hash(x)=165603215)
6
+ 20 train 10.108215 (lr=8.4000e-06) (hash(x)=142061326)
7
+ 30 train 9.824213 (lr=1.2400e-05) (hash(x)=147690730)
8
+ 40 train 9.747488 (lr=1.6400e-05) (hash(x)=160872542)
9
+ 50 train 9.613020 (lr=2.0400e-05) (hash(x)=166441190)
10
+ 60 train 9.459156 (lr=2.4400e-05) (hash(x)=140536766)
11
+ 70 train 9.162580 (lr=2.8400e-05) (hash(x)=146052944)
12
+ 80 train 8.989373 (lr=3.2400e-05) (hash(x)=140542391)
13
+ 90 train 8.720427 (lr=3.6400e-05) (hash(x)=147098660)
14
+ 100 val loss 8.4932
15
+ 100 val perplexity 4881.5239
16
+ 100 train 8.540207 (lr=4.0400e-05) (hash(x)=166780046)
17
+ 110 train 8.193802 (lr=4.4400e-05) (hash(x)=143376584)
18
+ 120 train 7.965147 (lr=4.8400e-05) (hash(x)=145337630)
19
+ 130 train 7.810315 (lr=5.2400e-05) (hash(x)=147524744)
20
+ 140 train 7.635900 (lr=5.6400e-05) (hash(x)=155587442)
21
+ 150 train 7.499724 (lr=6.0400e-05) (hash(x)=159835303)
22
+ 160 train 7.328168 (lr=6.4400e-05) (hash(x)=142171630)
23
+ 170 train 7.247965 (lr=6.8400e-05) (hash(x)=136930463)
24
+ 180 train 7.249845 (lr=7.2400e-05) (hash(x)=153579148)
25
+ 190 train 7.136475 (lr=7.6400e-05) (hash(x)=140352771)
26
+ 200 val loss 7.2915
27
+ 200 val perplexity 1467.7085
28
+ 200 train 7.051206 (lr=8.0400e-05) (hash(x)=155040610)
29
+ 210 train 7.007521 (lr=8.4400e-05) (hash(x)=141273862)
30
+ 220 train 7.203588 (lr=8.8400e-05) (hash(x)=165326618)
31
+ 230 train 6.954572 (lr=9.2400e-05) (hash(x)=150122120)
32
+ 240 train 6.765862 (lr=9.6400e-05) (hash(x)=143024679)
33
+ 250 train 6.746756 (lr=1.0040e-04) (hash(x)=130190460)
34
+ 260 train 6.832635 (lr=1.0440e-04) (hash(x)=151890219)
35
+ 270 train 6.641906 (lr=1.0840e-04) (hash(x)=143685291)
36
+ 280 train 6.839006 (lr=1.1240e-04) (hash(x)=137108194)
37
+ 290 train 7.048848 (lr=1.1640e-04) (hash(x)=157920866)
38
+ 300 val loss 6.7941
39
+ 300 val perplexity 892.5493
40
+ 300 train 6.735334 (lr=1.2040e-04) (hash(x)=155504036)
41
+ 310 train 6.841365 (lr=1.2440e-04) (hash(x)=144551186)
42
+ 320 train 6.849443 (lr=1.2840e-04) (hash(x)=161180086)
43
+ 330 train 6.552999 (lr=1.3240e-04) (hash(x)=141651471)
44
+ 340 train 6.639542 (lr=1.3640e-04) (hash(x)=157490741)
45
+ 350 train 6.570763 (lr=1.4040e-04) (hash(x)=137347213)
46
+ 360 train 6.753253 (lr=1.4440e-04) (hash(x)=171273659)
47
+ 370 train 6.512650 (lr=1.4840e-04) (hash(x)=152118871)
48
+ 380 train 6.481093 (lr=1.5240e-04) (hash(x)=161338097)
49
+ 390 train 6.545945 (lr=1.5640e-04) (hash(x)=152404537)
50
+ 400 val loss 6.4258
51
+ 400 val perplexity 617.5557
52
+ 400 train 6.268042 (lr=1.6040e-04) (hash(x)=143823248)
53
+ 410 train 6.556598 (lr=1.6440e-04) (hash(x)=161686752)
54
+ 420 train 6.243070 (lr=1.6840e-04) (hash(x)=155391312)
55
+ 430 train 6.149535 (lr=1.7240e-04) (hash(x)=142145507)
56
+ 440 train 6.185111 (lr=1.7640e-04) (hash(x)=121931091)
57
+ 450 train 6.271646 (lr=1.8040e-04) (hash(x)=156260416)
58
+ 460 train 6.407983 (lr=1.8440e-04) (hash(x)=169781623)
59
+ 470 train 6.098687 (lr=1.8840e-04) (hash(x)=150131545)
60
+ 480 train 6.141150 (lr=1.9240e-04) (hash(x)=150443184)
61
+ 490 train 6.100411 (lr=1.9640e-04) (hash(x)=149368198)
62
+ 500 val loss 6.1704
63
+ 500 val perplexity 478.3535
64
+ 500 train 6.072418 (lr=2.0000e-04) (hash(x)=143734685)
65
+ 510 train 6.094096 (lr=2.0000e-04) (hash(x)=148943891)
66
+ 520 train 6.055113 (lr=2.0000e-04) (hash(x)=149070354)
67
+ 530 train 6.100227 (lr=1.9999e-04) (hash(x)=153601219)
68
+ 540 train 6.116593 (lr=1.9999e-04) (hash(x)=174970654)
69
+ 550 train 6.034959 (lr=1.9998e-04) (hash(x)=160013925)
70
+ 560 train 5.878959 (lr=1.9998e-04) (hash(x)=152096988)
71
+ 570 train 5.937550 (lr=1.9997e-04) (hash(x)=157203924)
72
+ 580 train 5.750744 (lr=1.9996e-04) (hash(x)=138582113)
73
+ 590 train 5.808489 (lr=1.9995e-04) (hash(x)=157451775)
74
+ 600 val loss 5.9783
75
+ 600 val perplexity 394.7808
76
+ 600 train 5.717074 (lr=1.9993e-04) (hash(x)=150678249)
77
+ 610 train 5.776450 (lr=1.9992e-04) (hash(x)=150248072)
78
+ 620 train 5.652534 (lr=1.9991e-04) (hash(x)=148286909)
79
+ 630 train 5.596403 (lr=1.9989e-04) (hash(x)=141197178)
80
+ 640 train 5.679282 (lr=1.9987e-04) (hash(x)=144988391)
81
+ 650 train 5.637229 (lr=1.9985e-04) (hash(x)=149073315)
82
+ 660 train 5.560102 (lr=1.9983e-04) (hash(x)=145239943)
83
+ 670 train 5.621795 (lr=1.9981e-04) (hash(x)=146657426)
84
+ 680 train 5.798547 (lr=1.9979e-04) (hash(x)=142323429)
85
+ 690 train 5.714205 (lr=1.9976e-04) (hash(x)=142724342)
86
+ 700 val loss 5.8031
87
+ 700 val perplexity 331.3297
88
+ 700 train 5.699648 (lr=1.9974e-04) (hash(x)=175802021)
89
+ 710 train 5.756840 (lr=1.9971e-04) (hash(x)=148902690)
90
+ 720 train 5.823286 (lr=1.9968e-04) (hash(x)=155151320)
91
+ 730 train 5.687770 (lr=1.9966e-04) (hash(x)=142078914)
92
+ 740 train 5.829597 (lr=1.9962e-04) (hash(x)=155634484)
93
+ 750 train 5.792685 (lr=1.9959e-04) (hash(x)=171034639)
94
+ 760 train 5.730368 (lr=1.9956e-04) (hash(x)=142470148)
95
+ 770 train 5.738827 (lr=1.9952e-04) (hash(x)=156024448)
96
+ 780 train 5.597541 (lr=1.9949e-04) (hash(x)=151369417)
97
+ 790 train 5.557663 (lr=1.9945e-04) (hash(x)=139260056)
98
+ 800 val loss 5.6633
99
+ 800 val perplexity 288.0907
100
+ 800 train 5.770459 (lr=1.9941e-04) (hash(x)=158681215)
101
+ 810 train 5.551840 (lr=1.9937e-04) (hash(x)=144884161)
102
+ 820 train 5.637030 (lr=1.9933e-04) (hash(x)=146977514)
103
+ 830 train 5.499777 (lr=1.9929e-04) (hash(x)=148983636)
104
+ 840 train 5.579484 (lr=1.9925e-04) (hash(x)=157824542)
105
+ 850 train 5.515125 (lr=1.9920e-04) (hash(x)=152116061)
106
+ 860 train 5.616893 (lr=1.9916e-04) (hash(x)=153743703)
107
+ 870 train 5.636888 (lr=1.9911e-04) (hash(x)=149913335)
108
+ 880 train 5.501937 (lr=1.9906e-04) (hash(x)=151096434)
109
+ 890 train 5.491686 (lr=1.9901e-04) (hash(x)=145848702)
110
+ 900 val loss 5.5502
111
+ 900 val perplexity 257.2924
112
+ 900 train 5.481272 (lr=1.9896e-04) (hash(x)=146108145)
113
+ 910 train 5.390782 (lr=1.9891e-04) (hash(x)=146903806)
114
+ 920 train 5.370743 (lr=1.9885e-04) (hash(x)=149225019)
115
+ 930 train 5.312906 (lr=1.9880e-04) (hash(x)=146110422)
116
+ 940 train 5.404220 (lr=1.9874e-04) (hash(x)=161333614)
117
+ 950 train 5.352971 (lr=1.9868e-04) (hash(x)=147598108)
118
+ 960 train 5.433158 (lr=1.9862e-04) (hash(x)=158133101)
119
+ 970 train 5.322838 (lr=1.9856e-04) (hash(x)=152199591)
120
+ 980 train 5.282497 (lr=1.9850e-04) (hash(x)=152359036)
121
+ 990 train 5.298548 (lr=1.9844e-04) (hash(x)=145166335)
122
+ 1000 val loss 5.4794
123
+ 1000 val perplexity 239.6945
124
+ 1000 train 5.226059 (lr=1.9837e-04) (hash(x)=154996086)
125
+ 1010 train 5.156739 (lr=1.9831e-04) (hash(x)=149406975)
126
+ 1020 train 5.259118 (lr=1.9824e-04) (hash(x)=144618270)
127
+ 1030 train 5.142270 (lr=1.9817e-04) (hash(x)=148306536)
128
+ 1040 train 5.153977 (lr=1.9810e-04) (hash(x)=151643553)
129
+ 1050 train 5.081469 (lr=1.9803e-04) (hash(x)=153396183)
130
+ 1060 train 5.130601 (lr=1.9796e-04) (hash(x)=157501787)
131
+ 1070 train 5.307559 (lr=1.9789e-04) (hash(x)=153277507)
132
+ 1080 train 5.371570 (lr=1.9781e-04) (hash(x)=156232389)
133
+ 1090 train 5.319482 (lr=1.9774e-04) (hash(x)=150673143)
134
+ 1100 val loss 5.3930
135
+ 1100 val perplexity 219.8605
136
+ 1100 train 5.421607 (lr=1.9766e-04) (hash(x)=153885445)
137
+ 1110 train 5.375536 (lr=1.9758e-04) (hash(x)=143951073)
138
+ 1120 train 5.695242 (lr=1.9750e-04) (hash(x)=151626904)
139
+ 1130 train 5.366182 (lr=1.9742e-04) (hash(x)=144013074)
140
+ 1140 train 5.270397 (lr=1.9734e-04) (hash(x)=149948988)
141
+ 1150 train 5.362445 (lr=1.9726e-04) (hash(x)=159666385)
142
+ 1160 train 5.264938 (lr=1.9717e-04) (hash(x)=146889587)
143
+ 1170 train 5.346696 (lr=1.9709e-04) (hash(x)=151746917)
144
+ 1180 train 5.348418 (lr=1.9700e-04) (hash(x)=154396948)
145
+ 1190 train 5.306416 (lr=1.9691e-04) (hash(x)=148201222)
146
+ 1200 val loss 5.3091
147
+ 1200 val perplexity 202.1602
148
+ 1200 train 5.330200 (lr=1.9682e-04) (hash(x)=142353087)
149
+ 1210 train 5.280400 (lr=1.9673e-04) (hash(x)=146779775)
150
+ 1220 train 5.414968 (lr=1.9664e-04) (hash(x)=155864740)
151
+ 1230 train 5.346854 (lr=1.9654e-04) (hash(x)=129826485)
152
+ 1240 train 5.101971 (lr=1.9645e-04) (hash(x)=108599491)
153
+ 1250 train 5.194239 (lr=1.9635e-04) (hash(x)=146491718)
154
+ 1260 train 5.247613 (lr=1.9626e-04) (hash(x)=141897382)
155
+ 1270 train 5.188968 (lr=1.9616e-04) (hash(x)=144830380)
156
+ 1280 train 5.254448 (lr=1.9606e-04) (hash(x)=157677763)
157
+ 1290 train 5.111960 (lr=1.9596e-04) (hash(x)=143611746)
158
+ 1300 val loss 5.2503
159
+ 1300 val perplexity 190.6197
160
+ 1300 train 5.126443 (lr=1.9586e-04) (hash(x)=150750353)
161
+ 1310 train 5.448357 (lr=1.9575e-04) (hash(x)=152850907)
162
+ 1320 train 5.220899 (lr=1.9565e-04) (hash(x)=148575395)
163
+ 1330 train 5.001140 (lr=1.9554e-04) (hash(x)=157113001)
164
+ 1340 train 5.086244 (lr=1.9543e-04) (hash(x)=137400723)
165
+ 1350 train 5.112762 (lr=1.9533e-04) (hash(x)=129849193)
166
+ 1360 train 5.068231 (lr=1.9522e-04) (hash(x)=144176797)
167
+ 1370 train 4.946086 (lr=1.9511e-04) (hash(x)=144550831)
168
+ 1380 train 4.949076 (lr=1.9499e-04) (hash(x)=144171821)
169
+ 1390 train 4.867982 (lr=1.9488e-04) (hash(x)=146110449)
170
+ 1400 val loss 5.2023
171
+ 1400 val perplexity 181.6878
172
+ 1400 train 4.952493 (lr=1.9477e-04) (hash(x)=152767913)
173
+ 1410 train 4.940662 (lr=1.9465e-04) (hash(x)=148110947)
174
+ 1420 train 4.852838 (lr=1.9453e-04) (hash(x)=153477905)
175
+ 1430 train 4.760102 (lr=1.9441e-04) (hash(x)=147227598)
176
+ 1440 train 4.915612 (lr=1.9430e-04) (hash(x)=146280368)
177
+ 1450 train 4.924952 (lr=1.9417e-04) (hash(x)=146531140)
178
+ 1460 train 4.832004 (lr=1.9405e-04) (hash(x)=151573873)
179
+ 1470 train 5.172921 (lr=1.9393e-04) (hash(x)=155559275)
180
+ 1480 train 5.180936 (lr=1.9381e-04) (hash(x)=140584826)
181
+ 1490 train 5.065535 (lr=1.9368e-04) (hash(x)=165070637)
182
+ 1500 val loss 5.1584
183
+ 1500 val perplexity 173.8836
184
+ 1500 train 5.182666 (lr=1.9355e-04) (hash(x)=151562048)
185
+ 1510 train 4.993468 (lr=1.9343e-04) (hash(x)=147770607)
186
+ 1520 train 5.084578 (lr=1.9330e-04) (hash(x)=156486153)
187
+ 1530 train 4.987232 (lr=1.9317e-04) (hash(x)=140244564)
188
+ 1540 train 5.042418 (lr=1.9303e-04) (hash(x)=151627549)
189
+ 1550 train 5.095022 (lr=1.9290e-04) (hash(x)=146001424)
190
+ 1560 train 5.180748 (lr=1.9277e-04) (hash(x)=184266264)
191
+ 1570 train 5.091615 (lr=1.9263e-04) (hash(x)=145002091)
192
+ 1580 train 4.957626 (lr=1.9250e-04) (hash(x)=132216826)
193
+ 1590 train 5.127022 (lr=1.9236e-04) (hash(x)=160975703)
194
+ 1600 val loss 5.0900
195
+ 1600 val perplexity 162.3964
196
+ 1600 train 5.178689 (lr=1.9222e-04) (hash(x)=166486165)
197
+ 1610 train 5.103417 (lr=1.9208e-04) (hash(x)=149601833)
198
+ 1620 train 5.005731 (lr=1.9194e-04) (hash(x)=157815188)
199
+ 1630 train 5.024209 (lr=1.9180e-04) (hash(x)=144234679)
200
+ 1640 train 5.209277 (lr=1.9165e-04) (hash(x)=160054560)
201
+ 1650 train 4.995124 (lr=1.9151e-04) (hash(x)=141061333)
202
+ 1660 train 5.101241 (lr=1.9136e-04) (hash(x)=148064219)
203
+ 1670 train 4.989864 (lr=1.9121e-04) (hash(x)=162799669)
204
+ 1680 train 4.957571 (lr=1.9107e-04) (hash(x)=147474277)
205
+ 1690 train 4.989285 (lr=1.9092e-04) (hash(x)=142369299)
206
+ 1700 val loss 5.0483
207
+ 1700 val perplexity 155.7565
208
+ 1700 train 4.739018 (lr=1.9077e-04) (hash(x)=130835396)
209
+ 1710 train 5.212411 (lr=1.9061e-04) (hash(x)=153173892)
210
+ 1720 train 4.745191 (lr=1.9046e-04) (hash(x)=148451191)
211
+ 1730 train 4.904696 (lr=1.9031e-04) (hash(x)=145047239)
212
+ 1740 train 4.801269 (lr=1.9015e-04) (hash(x)=147187109)
213
+ 1750 train 4.936886 (lr=1.9000e-04) (hash(x)=149493155)
214
+ 1760 train 4.880357 (lr=1.8984e-04) (hash(x)=147647531)
215
+ 1770 train 5.017143 (lr=1.8968e-04) (hash(x)=153677279)
216
+ 1780 train 4.896429 (lr=1.8952e-04) (hash(x)=144314015)
217
+ 1790 train 4.935747 (lr=1.8936e-04) (hash(x)=176962883)
218
+ 1800 val loss 5.0046
219
+ 1800 val perplexity 149.0968
220
+ 1800 train 5.066918 (lr=1.8920e-04) (hash(x)=158851816)
221
+ 1810 train 4.721385 (lr=1.8903e-04) (hash(x)=175415447)
222
+ 1820 train 4.955061 (lr=1.8887e-04) (hash(x)=169829733)
223
+ 1830 train 4.968624 (lr=1.8870e-04) (hash(x)=154116106)
224
+ 1840 train 4.715547 (lr=1.8854e-04) (hash(x)=150168406)
225
+ 1850 train 4.754746 (lr=1.8837e-04) (hash(x)=145782493)
226
+ 1860 train 4.731813 (lr=1.8820e-04) (hash(x)=158111664)
227
+ 1870 train 4.651647 (lr=1.8803e-04) (hash(x)=154837061)
228
+ 1880 train 4.616287 (lr=1.8786e-04) (hash(x)=151363745)
229
+ 1890 train 4.748291 (lr=1.8768e-04) (hash(x)=161283557)
230
+ 1900 val loss 4.9705
231
+ 1900 val perplexity 144.0961
232
+ 1900 train 4.967950 (lr=1.8751e-04) (hash(x)=153313879)
233
+ 1910 train 4.782467 (lr=1.8734e-04) (hash(x)=152272272)
234
+ 1920 train 4.869771 (lr=1.8716e-04) (hash(x)=153648837)
235
+ 1930 train 4.959977 (lr=1.8698e-04) (hash(x)=150161694)
236
+ 1940 train 4.849574 (lr=1.8680e-04) (hash(x)=151235093)
237
+ 1950 train 4.875148 (lr=1.8663e-04) (hash(x)=140582058)
238
+ 1960 train 4.748272 (lr=1.8645e-04) (hash(x)=125616059)
239
+ 1970 train 4.936529 (lr=1.8626e-04) (hash(x)=154353968)
240
+ 1980 train 4.962951 (lr=1.8608e-04) (hash(x)=156063143)
241
+ 1990 train 5.008716 (lr=1.8590e-04) (hash(x)=160927289)
242
+ 2000 val loss 4.9079
243
+ 2000 val perplexity 135.3550
244
+ 2000 train 4.846354 (lr=1.8571e-04) (hash(x)=158245023)
245
+ 2010 train 4.931088 (lr=1.8553e-04) (hash(x)=157096661)
246
+ 2020 train 5.066594 (lr=1.8534e-04) (hash(x)=144994302)
247
+ 2030 train 4.862819 (lr=1.8515e-04) (hash(x)=145742929)
248
+ 2040 train 4.852841 (lr=1.8496e-04) (hash(x)=156958528)
249
+ 2050 train 5.277199 (lr=1.8477e-04) (hash(x)=135313020)
250
+ 2060 train 5.006086 (lr=1.8458e-04) (hash(x)=147309332)
251
+ 2070 train 4.785357 (lr=1.8439e-04) (hash(x)=161720962)
252
+ 2080 train 4.970672 (lr=1.8420e-04) (hash(x)=167204793)
253
+ 2090 train 4.840435 (lr=1.8400e-04) (hash(x)=141325510)
254
+ 2100 val loss 4.8562
255
+ 2100 val perplexity 128.5305
256
+ 2100 train 4.667901 (lr=1.8381e-04) (hash(x)=157204896)
257
+ 2110 train 4.604472 (lr=1.8361e-04) (hash(x)=143077993)
258
+ 2120 train 4.878103 (lr=1.8341e-04) (hash(x)=159032689)
259
+ 2130 train 4.718349 (lr=1.8321e-04) (hash(x)=149513855)
260
+ 2140 train 4.657532 (lr=1.8301e-04) (hash(x)=151678267)
261
+ 2150 train 4.723930 (lr=1.8281e-04) (hash(x)=141309246)
262
+ 2160 train 4.832582 (lr=1.8261e-04) (hash(x)=153935732)
263
+ 2170 train 4.729984 (lr=1.8241e-04) (hash(x)=150668875)
264
+ 2180 train 4.673487 (lr=1.8220e-04) (hash(x)=142033123)
265
+ 2190 train 4.768703 (lr=1.8200e-04) (hash(x)=157015026)
266
+ 2200 val loss 4.8258
267
+ 2200 val perplexity 124.6825
268
+ 2200 train 4.641841 (lr=1.8179e-04) (hash(x)=137541932)
269
+ 2210 train 4.548657 (lr=1.8158e-04) (hash(x)=144424504)
270
+ 2220 train 4.492041 (lr=1.8138e-04) (hash(x)=151015810)
271
+ 2230 train 4.559648 (lr=1.8117e-04) (hash(x)=160707806)
272
+ 2240 train 4.565139 (lr=1.8096e-04) (hash(x)=151874110)
273
+ 2250 train 4.579713 (lr=1.8074e-04) (hash(x)=157341469)
274
+ 2260 train 4.367250 (lr=1.8053e-04) (hash(x)=144563224)
275
+ 2270 train 4.497512 (lr=1.8032e-04) (hash(x)=144878459)
276
+ 2280 train 4.597013 (lr=1.8010e-04) (hash(x)=158331565)
277
+ 2290 train 4.609296 (lr=1.7989e-04) (hash(x)=146875132)
278
+ 2300 val loss 4.7934
279
+ 2300 val perplexity 120.7085
280
+ 2300 train 4.743674 (lr=1.7967e-04) (hash(x)=150149692)
281
+ 2310 train 4.747735 (lr=1.7946e-04) (hash(x)=141206538)
282
+ 2320 train 4.767364 (lr=1.7924e-04) (hash(x)=136219115)
283
+ 2330 train 4.810294 (lr=1.7902e-04) (hash(x)=148496500)
284
+ 2340 train 4.778009 (lr=1.7880e-04) (hash(x)=161388511)
285
+ 2350 train 4.637798 (lr=1.7858e-04) (hash(x)=149433185)
286
+ 2360 train 4.832192 (lr=1.7835e-04) (hash(x)=154659203)
287
+ 2370 train 4.834682 (lr=1.7813e-04) (hash(x)=145696996)
288
+ 2380 train 4.632764 (lr=1.7791e-04) (hash(x)=139915095)
289
+ 2390 train 4.772333 (lr=1.7768e-04) (hash(x)=150567584)
290
+ 2400 val loss 4.7464
291
+ 2400 val perplexity 115.1710
292
+ 2400 train 4.674137 (lr=1.7745e-04) (hash(x)=151730720)
293
+ 2410 train 4.690665 (lr=1.7723e-04) (hash(x)=153136544)
294
+ 2420 train 4.683325 (lr=1.7700e-04) (hash(x)=151004677)
295
+ 2430 train 4.700076 (lr=1.7677e-04) (hash(x)=167118859)
296
+ 2440 train 4.688626 (lr=1.7654e-04) (hash(x)=145433453)
297
+ 2450 train 4.725751 (lr=1.7631e-04) (hash(x)=150913293)
298
+ 2460 train 4.699337 (lr=1.7607e-04) (hash(x)=145674639)
299
+ 2470 train 4.804339 (lr=1.7584e-04) (hash(x)=151022749)
300
+ 2480 train 4.635894 (lr=1.7561e-04) (hash(x)=136621037)
301
+ 2490 train 4.703066 (lr=1.7537e-04) (hash(x)=158100672)
302
+ 2500 val loss 4.7080
303
+ 2500 val perplexity 110.8304
304
+ 2500 train 4.677748 (lr=1.7514e-04) (hash(x)=143406752)
305
+ 2510 train 4.621717 (lr=1.7490e-04) (hash(x)=142664084)
306
+ 2520 train 4.707599 (lr=1.7466e-04) (hash(x)=132174378)
307
+ 2530 train 4.732631 (lr=1.7442e-04) (hash(x)=140760293)
308
+ 2540 train 4.616854 (lr=1.7418e-04) (hash(x)=154863398)
309
+ 2550 train 4.580019 (lr=1.7394e-04) (hash(x)=153042049)
310
+ 2560 train 4.496349 (lr=1.7370e-04) (hash(x)=145339170)
311
+ 2570 train 4.676992 (lr=1.7346e-04) (hash(x)=157656357)
312
+ 2580 train 4.640365 (lr=1.7321e-04) (hash(x)=147899462)
313
+ 2590 train 4.597972 (lr=1.7297e-04) (hash(x)=142812892)
314
+ 2600 val loss 4.7007
315
+ 2600 val perplexity 110.0189
316
+ 2600 train 4.569220 (lr=1.7272e-04) (hash(x)=157272496)
317
+ 2610 train 4.370792 (lr=1.7248e-04) (hash(x)=148923594)
318
+ 2620 train 4.269350 (lr=1.7223e-04) (hash(x)=164654807)
319
+ 2630 train 4.455981 (lr=1.7198e-04) (hash(x)=136419327)
320
+ 2640 train 4.436761 (lr=1.7173e-04) (hash(x)=153420840)
321
+ 2650 train 4.286306 (lr=1.7148e-04) (hash(x)=153693827)
322
+ 2660 train 4.447424 (lr=1.7123e-04) (hash(x)=145574165)
323
+ 2670 train 4.476869 (lr=1.7098e-04) (hash(x)=153894843)
324
+ 2680 train 4.331589 (lr=1.7073e-04) (hash(x)=132234446)
325
+ 2690 train 4.412297 (lr=1.7048e-04) (hash(x)=153430212)
326
+ 2700 val loss 4.6816
327
+ 2700 val perplexity 107.9465
328
+ 2700 train 4.590431 (lr=1.7022e-04) (hash(x)=155342327)
329
+ 2710 train 4.567213 (lr=1.6997e-04) (hash(x)=145797023)
330
+ 2720 train 4.724094 (lr=1.6971e-04) (hash(x)=149220870)
331
+ 2730 train 4.622249 (lr=1.6945e-04) (hash(x)=134069562)
332
+ 2740 train 4.628724 (lr=1.6920e-04) (hash(x)=141560499)
333
+ 2750 train 4.653552 (lr=1.6894e-04) (hash(x)=146000524)
334
+ 2760 train 4.578631 (lr=1.6868e-04) (hash(x)=164710665)
335
+ 2770 train 4.456148 (lr=1.6842e-04) (hash(x)=151199094)
336
+ 2780 train 4.633516 (lr=1.6816e-04) (hash(x)=138625404)
337
+ 2790 train 4.910608 (lr=1.6789e-04) (hash(x)=162766112)
338
+ 2800 val loss 4.6410
339
+ 2800 val perplexity 103.6496
340
+ 2800 train 4.597857 (lr=1.6763e-04) (hash(x)=140626679)
341
+ 2810 train 4.675623 (lr=1.6737e-04) (hash(x)=144248544)
342
+ 2820 train 4.574524 (lr=1.6710e-04) (hash(x)=139378481)
343
+ 2830 train 4.545200 (lr=1.6684e-04) (hash(x)=150173403)
344
+ 2840 train 4.812176 (lr=1.6657e-04) (hash(x)=160168705)
345
+ 2850 train 4.603701 (lr=1.6631e-04) (hash(x)=149480059)
346
+ 2860 train 4.684381 (lr=1.6604e-04) (hash(x)=155559096)
347
+ 2870 train 4.598352 (lr=1.6577e-04) (hash(x)=158825683)
348
+ 2880 train 4.746686 (lr=1.6550e-04) (hash(x)=152129048)
349
+ 2890 train 4.431717 (lr=1.6523e-04) (hash(x)=157866544)
350
+ 2900 val loss 4.6157
351
+ 2900 val perplexity 101.0539
352
+ 2900 train 4.550244 (lr=1.6496e-04) (hash(x)=144953350)
353
+ 2910 train 4.748907 (lr=1.6469e-04) (hash(x)=154103497)
354
+ 2920 train 4.468185 (lr=1.6441e-04) (hash(x)=144467854)
355
+ 2930 train 4.391221 (lr=1.6414e-04) (hash(x)=153784048)
356
+ 2940 train 4.692478 (lr=1.6387e-04) (hash(x)=154774534)
357
+ 2950 train 4.533800 (lr=1.6359e-04) (hash(x)=143988507)
358
+ 2960 train 4.436468 (lr=1.6332e-04) (hash(x)=159552857)
359
+ 2970 train 4.739278 (lr=1.6304e-04) (hash(x)=175539144)
360
+ 2980 train 4.620506 (lr=1.6276e-04) (hash(x)=158172686)
361
+ 2990 train 4.477989 (lr=1.6248e-04) (hash(x)=147553039)
362
+ 3000 val loss 4.6033
363
+ 3000 val perplexity 99.8119
364
+ 3000 train 4.409822 (lr=1.6221e-04) (hash(x)=172449837)
365
+ 3010 train 4.319238 (lr=1.6193e-04) (hash(x)=153285123)
366
+ 3020 train 4.525682 (lr=1.6165e-04) (hash(x)=145307252)
367
+ 3030 train 4.513878 (lr=1.6136e-04) (hash(x)=147739766)
368
+ 3040 train 4.241980 (lr=1.6108e-04) (hash(x)=154414400)
369
+ 3050 train 4.334424 (lr=1.6080e-04) (hash(x)=150696487)
370
+ 3060 train 4.364735 (lr=1.6052e-04) (hash(x)=154543441)
371
+ 3070 train 4.113924 (lr=1.6023e-04) (hash(x)=170733242)
372
+ 3080 train 4.482882 (lr=1.5995e-04) (hash(x)=161914669)
373
+ 3090 train 4.363663 (lr=1.5966e-04) (hash(x)=156125491)
374
+ 3100 val loss 4.6101
375
+ 3100 val perplexity 100.4975
376
+ 3100 train 4.263449 (lr=1.5938e-04) (hash(x)=141710086)
377
+ 3110 train 4.537390 (lr=1.5909e-04) (hash(x)=154547635)
378
+ 3120 train 4.399536 (lr=1.5880e-04) (hash(x)=151747171)
379
+ 3130 train 4.682076 (lr=1.5851e-04) (hash(x)=160311353)
380
+ 3140 train 4.767546 (lr=1.5822e-04) (hash(x)=157233821)
381
+ 3150 train 4.612141 (lr=1.5793e-04) (hash(x)=153016810)
382
+ 3160 train 4.580122 (lr=1.5764e-04) (hash(x)=157289875)
383
+ 3170 train 4.599168 (lr=1.5735e-04) (hash(x)=157459484)
384
+ 3180 train 4.527137 (lr=1.5706e-04) (hash(x)=162037435)
385
+ 3190 train 4.613904 (lr=1.5677e-04) (hash(x)=159360169)
386
+ 3200 val loss 4.5561
387
+ 3200 val perplexity 95.2097
388
+ 3200 train 4.476530 (lr=1.5648e-04) (hash(x)=151299772)
389
+ 3210 train 4.765331 (lr=1.5618e-04) (hash(x)=183865364)
390
+ 3220 train 4.447180 (lr=1.5589e-04) (hash(x)=139622592)
391
+ 3230 train 4.393746 (lr=1.5559e-04) (hash(x)=147102655)
392
+ 3240 train 4.464899 (lr=1.5530e-04) (hash(x)=161870696)
393
+ 3250 train 4.616153 (lr=1.5500e-04) (hash(x)=152692231)
394
+ 3260 train 4.341697 (lr=1.5470e-04) (hash(x)=149990134)
395
+ 3270 train 4.319035 (lr=1.5441e-04) (hash(x)=134848432)
396
+ 3280 train 4.371098 (lr=1.5411e-04) (hash(x)=138519418)
397
+ 3290 train 4.501052 (lr=1.5381e-04) (hash(x)=139301249)
398
+ 3300 val loss 4.5524
399
+ 3300 val perplexity 94.8632
400
+ 3300 train 4.369302 (lr=1.5351e-04) (hash(x)=146473110)
401
+ 3310 train 4.405541 (lr=1.5321e-04) (hash(x)=144833685)
402
+ 3320 train 4.331898 (lr=1.5291e-04) (hash(x)=148953545)
403
+ 3330 train 4.551664 (lr=1.5261e-04) (hash(x)=161302371)
404
+ 3340 train 4.484500 (lr=1.5230e-04) (hash(x)=144215411)
405
+ 3350 train 4.639852 (lr=1.5200e-04) (hash(x)=157713664)
406
+ 3360 train 4.529778 (lr=1.5170e-04) (hash(x)=150813928)
407
+ 3370 train 4.510571 (lr=1.5139e-04) (hash(x)=150889463)
408
+ 3380 train 4.585331 (lr=1.5109e-04) (hash(x)=156631928)
409
+ 3390 train 4.577444 (lr=1.5078e-04) (hash(x)=136001579)
410
+ 3400 val loss 4.5261
411
+ 3400 val perplexity 92.3991
412
+ 3400 train 4.673074 (lr=1.5048e-04) (hash(x)=153954157)
413
+ 3410 train 4.453163 (lr=1.5017e-04) (hash(x)=150012477)
414
+ 3420 train 4.828722 (lr=1.4986e-04) (hash(x)=160878899)
415
+ 3430 train 4.803936 (lr=1.4956e-04) (hash(x)=128131106)
416
+ 3440 train 4.777902 (lr=1.4925e-04) (hash(x)=149611866)
417
+ 3450 train 4.406456 (lr=1.4894e-04) (hash(x)=166942460)
418
+ 3460 train 4.548495 (lr=1.4863e-04) (hash(x)=150498414)
419
+ 3470 train 4.717762 (lr=1.4832e-04) (hash(x)=144342279)
420
+ 3480 train 4.504704 (lr=1.4801e-04) (hash(x)=148891764)
421
+ 3490 train 4.521090 (lr=1.4770e-04) (hash(x)=157308188)
422
+ 3500 val loss 4.5095
423
+ 3500 val perplexity 90.8802
424
+ 3500 train 4.405951 (lr=1.4739e-04) (hash(x)=153717336)
425
+ 3510 train 4.424365 (lr=1.4708e-04) (hash(x)=156083924)
426
+ 3520 train 4.414748 (lr=1.4676e-04) (hash(x)=144234966)
427
+ 3530 train 4.488238 (lr=1.4645e-04) (hash(x)=150184381)
428
+ 3540 train 4.367004 (lr=1.4614e-04) (hash(x)=143011198)
429
+ 3550 train 4.512517 (lr=1.4582e-04) (hash(x)=148764613)
430
+ 3560 train 4.280169 (lr=1.4551e-04) (hash(x)=150625329)
431
+ 3570 train 4.403288 (lr=1.4519e-04) (hash(x)=143981302)
432
+ 3580 train 4.442192 (lr=1.4488e-04) (hash(x)=154989703)
433
+ 3590 train 4.570028 (lr=1.4456e-04) (hash(x)=146602526)
434
+ 3600 val loss 4.4975
435
+ 3600 val perplexity 89.7965
436
+ 3600 train 4.269297 (lr=1.4424e-04) (hash(x)=144965161)
437
+ 3610 train 4.258621 (lr=1.4393e-04) (hash(x)=159452243)
438
+ 3620 train 4.255215 (lr=1.4361e-04) (hash(x)=144904300)
439
+ 3630 train 4.161673 (lr=1.4329e-04) (hash(x)=127790212)
440
+ 3640 train 4.412374 (lr=1.4297e-04) (hash(x)=150863158)
441
+ 3650 train 4.360890 (lr=1.4265e-04) (hash(x)=145691635)
442
+ 3660 train 4.335746 (lr=1.4233e-04) (hash(x)=150631264)
443
+ 3670 train 4.337727 (lr=1.4201e-04) (hash(x)=148646259)
444
+ 3680 train 4.284561 (lr=1.4169e-04) (hash(x)=139317386)
445
+ 3690 train 4.401682 (lr=1.4137e-04) (hash(x)=142206916)
446
+ 3700 val loss 4.5001
447
+ 3700 val perplexity 90.0239
448
+ 3700 train 4.382488 (lr=1.4105e-04) (hash(x)=125969741)
449
+ 3710 train 4.255900 (lr=1.4073e-04) (hash(x)=157183915)
450
+ 3720 train 4.250027 (lr=1.4041e-04) (hash(x)=150434369)
451
+ 3730 train 4.450710 (lr=1.4008e-04) (hash(x)=160911722)
452
+ 3740 train 4.341411 (lr=1.3976e-04) (hash(x)=163306921)
453
+ 3750 train 4.438487 (lr=1.3944e-04) (hash(x)=149156876)
454
+ 3760 train 4.593968 (lr=1.3911e-04) (hash(x)=156015970)
455
+ 3770 train 4.517069 (lr=1.3879e-04) (hash(x)=148222322)
456
+ 3780 train 4.629981 (lr=1.3846e-04) (hash(x)=172040174)
457
+ 3790 train 4.536961 (lr=1.3814e-04) (hash(x)=151593070)
458
+ 3800 val loss 4.4777
459
+ 3800 val perplexity 88.0276
460
+ 3800 train 4.303720 (lr=1.3781e-04) (hash(x)=155070487)
461
+ 3810 train 4.462440 (lr=1.3749e-04) (hash(x)=137073164)
462
+ 3820 train 4.399971 (lr=1.3716e-04) (hash(x)=147761510)
463
+ 3830 train 4.304118 (lr=1.3683e-04) (hash(x)=154763021)
464
+ 3840 train 4.344714 (lr=1.3650e-04) (hash(x)=149622435)
465
+ 3850 train 4.460896 (lr=1.3618e-04) (hash(x)=150779423)
466
+ 3860 train 4.418278 (lr=1.3585e-04) (hash(x)=139445120)
467
+ 3870 train 4.409332 (lr=1.3552e-04) (hash(x)=151884947)
468
+ 3880 train 4.409469 (lr=1.3519e-04) (hash(x)=139795201)
469
+ 3890 train 4.458975 (lr=1.3486e-04) (hash(x)=155589956)
470
+ 3900 val loss 4.4592
471
+ 3900 val perplexity 86.4202
472
+ 3900 train 4.364912 (lr=1.3453e-04) (hash(x)=149444644)
473
+ 3910 train 4.385828 (lr=1.3420e-04) (hash(x)=148538323)
474
+ 3920 train 4.482190 (lr=1.3387e-04) (hash(x)=144026925)
475
+ 3930 train 4.380352 (lr=1.3354e-04) (hash(x)=155311200)
476
+ 3940 train 4.342221 (lr=1.3321e-04) (hash(x)=142410034)
477
+ 3950 train 4.325383 (lr=1.3288e-04) (hash(x)=145986075)
478
+ 3960 train 4.254183 (lr=1.3255e-04) (hash(x)=156197753)
479
+ 3970 train 4.375360 (lr=1.3222e-04) (hash(x)=146981564)
480
+ 3980 train 4.688904 (lr=1.3188e-04) (hash(x)=145361976)
481
+ 3990 train 4.417999 (lr=1.3155e-04) (hash(x)=149344475)
482
+ 4000 val loss 4.4495
483
+ 4000 val perplexity 85.5842
484
+ 4000 train 4.303828 (lr=1.3122e-04) (hash(x)=151663033)
485
+ 4010 train 4.189203 (lr=1.3089e-04) (hash(x)=140915499)
486
+ 4020 train 4.421053 (lr=1.3055e-04) (hash(x)=159979134)
487
+ 4030 train 4.336899 (lr=1.3022e-04) (hash(x)=163081433)
488
+ 4040 train 4.219212 (lr=1.2988e-04) (hash(x)=144890231)
489
+ 4050 train 4.370840 (lr=1.2955e-04) (hash(x)=138242368)
490
+ 4060 train 4.276647 (lr=1.2921e-04) (hash(x)=144438406)
491
+ 4070 train 4.174635 (lr=1.2888e-04) (hash(x)=159294906)
492
+ 4080 train 4.207751 (lr=1.2854e-04) (hash(x)=143676301)
493
+ 4090 train 4.448840 (lr=1.2821e-04) (hash(x)=147875296)
494
+ 4100 val loss 4.4581
495
+ 4100 val perplexity 86.3273
496
+ 4100 train 4.495130 (lr=1.2787e-04) (hash(x)=143688282)
497
+ 4110 train 4.287563 (lr=1.2754e-04) (hash(x)=150086777)
498
+ 4120 train 4.222195 (lr=1.2720e-04) (hash(x)=154624391)
499
+ 4130 train 4.412323 (lr=1.2686e-04) (hash(x)=141163032)
500
+ 4140 train 4.429742 (lr=1.2653e-04) (hash(x)=151038180)
501
+ 4150 train 4.506675 (lr=1.2619e-04) (hash(x)=144718421)
502
+ 4160 train 4.423493 (lr=1.2585e-04) (hash(x)=154277459)
503
+ 4170 train 4.592068 (lr=1.2552e-04) (hash(x)=143765173)
504
+ 4180 train 4.452863 (lr=1.2518e-04) (hash(x)=161013425)
505
+ 4190 train 4.413038 (lr=1.2484e-04) (hash(x)=161666854)
506
+ 4200 val loss 4.4259
507
+ 4200 val perplexity 83.5852
508
+ 4200 train 4.352727 (lr=1.2450e-04) (hash(x)=163361651)
509
+ 4210 train 4.673963 (lr=1.2416e-04) (hash(x)=153491528)
510
+ 4220 train 4.467484 (lr=1.2383e-04) (hash(x)=164410385)
511
+ 4230 train 4.404262 (lr=1.2349e-04) (hash(x)=175191816)
512
+ 4240 train 4.271538 (lr=1.2315e-04) (hash(x)=133924818)
513
+ 4250 train 4.488371 (lr=1.2281e-04) (hash(x)=152932845)
514
+ 4260 train 4.370868 (lr=1.2247e-04) (hash(x)=141987318)
515
+ 4270 train 4.469886 (lr=1.2213e-04) (hash(x)=142521279)
516
+ 4280 train 4.338464 (lr=1.2179e-04) (hash(x)=147707669)
517
+ 4290 train 4.386642 (lr=1.2145e-04) (hash(x)=152245539)
518
+ 4300 val loss 4.4120
519
+ 4300 val perplexity 82.4329
520
+ 4300 train 4.286423 (lr=1.2111e-04) (hash(x)=153619361)
521
+ 4310 train 4.487903 (lr=1.2077e-04) (hash(x)=142614473)
522
+ 4320 train 4.313277 (lr=1.2043e-04) (hash(x)=142689858)
523
+ 4330 train 4.320467 (lr=1.2009e-04) (hash(x)=140837891)
524
+ 4340 train 4.328830 (lr=1.1975e-04) (hash(x)=131790760)
525
+ 4350 train 4.482335 (lr=1.1941e-04) (hash(x)=147186292)
526
+ 4360 train 4.303123 (lr=1.1907e-04) (hash(x)=148140562)
527
+ 4370 train 4.408495 (lr=1.1873e-04) (hash(x)=156409913)
528
+ 4380 train 4.324696 (lr=1.1838e-04) (hash(x)=145126034)
529
+ 4390 train 4.267291 (lr=1.1804e-04) (hash(x)=147368423)
530
+ 4400 val loss 4.4113
531
+ 4400 val perplexity 82.3739
532
+ 4400 train 4.776717 (lr=1.1770e-04) (hash(x)=168527064)
533
+ 4410 train 4.259302 (lr=1.1736e-04) (hash(x)=150152773)
534
+ 4420 train 4.358440 (lr=1.1702e-04) (hash(x)=155472304)
535
+ 4430 train 4.274035 (lr=1.1668e-04) (hash(x)=147299265)
536
+ 4440 train 4.230871 (lr=1.1634e-04) (hash(x)=140659522)
537
+ 4450 train 4.298770 (lr=1.1599e-04) (hash(x)=147169111)
538
+ 4460 train 4.361233 (lr=1.1565e-04) (hash(x)=144098557)
539
+ 4470 train 4.139716 (lr=1.1531e-04) (hash(x)=161957507)
540
+ 4480 train 4.156746 (lr=1.1497e-04) (hash(x)=154064204)
541
+ 4490 train 4.289114 (lr=1.1462e-04) (hash(x)=153999958)
542
+ 4500 val loss 4.4138
543
+ 4500 val perplexity 82.5810
544
+ 4500 train 4.407358 (lr=1.1428e-04) (hash(x)=125588037)
545
+ 4510 train 4.244913 (lr=1.1394e-04) (hash(x)=142522186)
546
+ 4520 train 4.096467 (lr=1.1360e-04) (hash(x)=152736893)
547
+ 4530 train 4.320524 (lr=1.1326e-04) (hash(x)=148091275)
548
+ 4540 train 4.452848 (lr=1.1291e-04) (hash(x)=147663019)
549
+ 4550 train 4.521071 (lr=1.1257e-04) (hash(x)=151491857)
550
+ 4560 train 4.403203 (lr=1.1223e-04) (hash(x)=161757513)
551
+ 4570 train 4.412116 (lr=1.1188e-04) (hash(x)=165448260)
552
+ 4580 train 4.549607 (lr=1.1154e-04) (hash(x)=151620292)
553
+ 4590 train 4.626790 (lr=1.1120e-04) (hash(x)=161877852)
554
+ 4600 val loss 4.3855
555
+ 4600 val perplexity 80.2796
556
+ 4600 train 4.363959 (lr=1.1086e-04) (hash(x)=143710941)
557
+ 4610 train 4.456553 (lr=1.1051e-04) (hash(x)=152244871)
558
+ 4620 train 4.419514 (lr=1.1017e-04) (hash(x)=161268167)
559
+ 4630 train 4.383834 (lr=1.0983e-04) (hash(x)=153904677)
560
+ 4640 train 4.339632 (lr=1.0949e-04) (hash(x)=152593758)
561
+ 4650 train 4.488466 (lr=1.0914e-04) (hash(x)=159047122)
562
+ 4660 train 4.560596 (lr=1.0880e-04) (hash(x)=167370244)
563
+ 4670 train 4.335510 (lr=1.0846e-04) (hash(x)=140357684)
564
+ 4680 train 4.476111 (lr=1.0812e-04) (hash(x)=140547853)
565
+ 4690 train 4.649773 (lr=1.0777e-04) (hash(x)=143000949)
566
+ 4700 val loss 4.3781
567
+ 4700 val perplexity 79.6848
568
+ 4700 train 4.302816 (lr=1.0743e-04) (hash(x)=150952742)
569
+ 4710 train 4.347490 (lr=1.0709e-04) (hash(x)=157964406)
570
+ 4720 train 4.267570 (lr=1.0674e-04) (hash(x)=147653608)
571
+ 4730 train 4.339172 (lr=1.0640e-04) (hash(x)=165545764)
572
+ 4740 train 4.300375 (lr=1.0606e-04) (hash(x)=146744631)
573
+ 4750 train 4.389025 (lr=1.0572e-04) (hash(x)=153635164)
574
+ 4760 train 4.454523 (lr=1.0538e-04) (hash(x)=153217508)
575
+ 4770 train 4.395255 (lr=1.0503e-04) (hash(x)=169622743)
576
+ 4780 train 4.299000 (lr=1.0469e-04) (hash(x)=138018297)
577
+ 4790 train 4.187404 (lr=1.0435e-04) (hash(x)=156067303)
578
+ 4800 val loss 4.3754
579
+ 4800 val perplexity 79.4706
580
+ 4800 train 4.351851 (lr=1.0401e-04) (hash(x)=145323659)
581
+ 4810 train 4.231571 (lr=1.0366e-04) (hash(x)=152587772)
582
+ 4820 train 4.456962 (lr=1.0332e-04) (hash(x)=148481485)
583
+ 4830 train 4.222940 (lr=1.0298e-04) (hash(x)=154219664)
584
+ 4840 train 4.123485 (lr=1.0264e-04) (hash(x)=156853645)
585
+ 4850 train 4.199806 (lr=1.0230e-04) (hash(x)=152329702)
586
+ 4860 train 4.184996 (lr=1.0196e-04) (hash(x)=136684779)
587
+ 4870 train 4.247592 (lr=1.0162e-04) (hash(x)=143260367)
588
+ 4880 train 4.223344 (lr=1.0127e-04) (hash(x)=150910947)
589
+ 4890 train 4.258477 (lr=1.0093e-04) (hash(x)=179249925)
590
+ 4900 val loss 4.3750
591
+ 4900 val perplexity 79.4368
592
+ 4900 train 4.395765 (lr=1.0059e-04) (hash(x)=153151397)
593
+ 4910 train 4.345817 (lr=1.0025e-04) (hash(x)=144101891)
594
+ 4920 train 4.252564 (lr=9.9911e-05) (hash(x)=144994952)
595
+ 4930 train 4.541728 (lr=9.9571e-05) (hash(x)=155390878)
596
+ 4940 train 4.301800 (lr=9.9230e-05) (hash(x)=141359221)
597
+ 4950 train 4.213285 (lr=9.8890e-05) (hash(x)=151570011)
598
+ 4960 train 4.440537 (lr=9.8550e-05) (hash(x)=146918855)
599
+ 4970 train 4.362798 (lr=9.8210e-05) (hash(x)=148647362)
600
+ 4980 train 4.326989 (lr=9.7870e-05) (hash(x)=159753562)
601
+ 4990 train 4.272650 (lr=9.7531e-05) (hash(x)=147683447)
602
+ 5000 val loss 4.3678
603
+ 5000 val perplexity 78.8724
604
+ 5000 train 4.393171 (lr=9.7192e-05) (hash(x)=143182059)
605
+ 5010 train 4.204443 (lr=9.6853e-05) (hash(x)=138657407)
606
+ 5020 train 4.317229 (lr=9.6514e-05) (hash(x)=151830497)
607
+ 5030 train 4.319872 (lr=9.6175e-05) (hash(x)=147727694)
608
+ 5040 train 4.176632 (lr=9.5836e-05) (hash(x)=151957014)
609
+ 5050 train 4.437883 (lr=9.5498e-05) (hash(x)=147828693)
610
+ 5060 train 4.472606 (lr=9.5160e-05) (hash(x)=158589700)
611
+ 5070 train 4.360204 (lr=9.4822e-05) (hash(x)=155662163)
612
+ 5080 train 4.472041 (lr=9.4484e-05) (hash(x)=157797383)
613
+ 5090 train 4.350271 (lr=9.4147e-05) (hash(x)=153848957)
614
+ 5100 val loss 4.3535
615
+ 5100 val perplexity 77.7500
616
+ 5100 train 4.314097 (lr=9.3809e-05) (hash(x)=170083586)
617
+ 5110 train 4.226809 (lr=9.3472e-05) (hash(x)=153800138)
618
+ 5120 train 4.334444 (lr=9.3136e-05) (hash(x)=153526228)
619
+ 5130 train 4.337140 (lr=9.2799e-05) (hash(x)=158650342)
620
+ 5140 train 4.323368 (lr=9.2463e-05) (hash(x)=153163034)
621
+ 5150 train 4.326264 (lr=9.2127e-05) (hash(x)=155281294)
622
+ 5160 train 4.288630 (lr=9.1791e-05) (hash(x)=147731627)
623
+ 5170 train 4.274343 (lr=9.1456e-05) (hash(x)=146388976)
624
+ 5180 train 4.484706 (lr=9.1120e-05) (hash(x)=140323065)
625
+ 5190 train 4.398393 (lr=9.0785e-05) (hash(x)=150406523)
626
+ 5200 val loss 4.3453
627
+ 5200 val perplexity 77.1166
628
+ 5200 train 4.197736 (lr=9.0451e-05) (hash(x)=149363919)
629
+ 5210 train 4.348881 (lr=9.0116e-05) (hash(x)=164390214)
630
+ 5220 train 4.301536 (lr=8.9782e-05) (hash(x)=153652984)
631
+ 5230 train 4.262619 (lr=8.9448e-05) (hash(x)=139542770)
632
+ 5240 train 4.185484 (lr=8.9115e-05) (hash(x)=140115288)
633
+ 5250 train 4.304239 (lr=8.8782e-05) (hash(x)=142158900)
634
+ 5260 train 4.203769 (lr=8.8449e-05) (hash(x)=156019449)
635
+ 5270 train 4.317973 (lr=8.8116e-05) (hash(x)=150776850)
636
+ 5280 train 4.268752 (lr=8.7784e-05) (hash(x)=151358692)
637
+ 5290 train 4.243278 (lr=8.7452e-05) (hash(x)=137112863)
638
+ 5300 val loss 4.3438
639
+ 5300 val perplexity 77.0025
640
+ 5300 train 4.160437 (lr=8.7120e-05) (hash(x)=152033784)
641
+ 5310 train 4.209918 (lr=8.6789e-05) (hash(x)=147785712)
642
+ 5320 train 4.356404 (lr=8.6458e-05) (hash(x)=156426339)
643
+ 5330 train 4.298791 (lr=8.6127e-05) (hash(x)=161907693)
644
+ 5340 train 4.376690 (lr=8.5797e-05) (hash(x)=156414094)
645
+ 5350 train 4.398882 (lr=8.5467e-05) (hash(x)=141712428)
646
+ 5360 train 4.345904 (lr=8.5138e-05) (hash(x)=156873601)
647
+ 5370 train 4.326837 (lr=8.4809e-05) (hash(x)=149863781)
648
+ 5380 train 4.326070 (lr=8.4480e-05) (hash(x)=148183971)
649
+ 5390 train 4.308631 (lr=8.4151e-05) (hash(x)=151542334)
650
+ 5400 val loss 4.3293
651
+ 5400 val perplexity 75.8910
652
+ 5400 train 4.562572 (lr=8.3823e-05) (hash(x)=154614289)
653
+ 5410 train 4.321802 (lr=8.3495e-05) (hash(x)=169124671)
654
+ 5420 train 4.347882 (lr=8.3168e-05) (hash(x)=159963738)
655
+ 5430 train 4.293843 (lr=8.2841e-05) (hash(x)=155162691)
656
+ 5440 train 4.303808 (lr=8.2515e-05) (hash(x)=135981908)
657
+ 5450 train 4.403159 (lr=8.2188e-05) (hash(x)=145055580)
658
+ 5460 train 4.264373 (lr=8.1863e-05) (hash(x)=150367825)
659
+ 5470 train 4.353034 (lr=8.1537e-05) (hash(x)=151034173)
660
+ 5480 train 4.282482 (lr=8.1212e-05) (hash(x)=155581766)
661
+ 5490 train 4.296411 (lr=8.0888e-05) (hash(x)=136973993)
662
+ 5500 val loss 4.3216
663
+ 5500 val perplexity 75.3106
664
+ 5500 train 4.474913 (lr=8.0564e-05) (hash(x)=157745174)
665
+ 5510 train 4.412661 (lr=8.0240e-05) (hash(x)=144628894)
666
+ 5520 train 4.286175 (lr=7.9917e-05) (hash(x)=135056684)
667
+ 5530 train 4.296622 (lr=7.9594e-05) (hash(x)=143418640)
668
+ 5540 train 4.263391 (lr=7.9272e-05) (hash(x)=144464984)
669
+ 5550 train 4.493005 (lr=7.8950e-05) (hash(x)=155888745)
670
+ 5560 train 4.138254 (lr=7.8629e-05) (hash(x)=149122683)
671
+ 5570 train 4.181705 (lr=7.8308e-05) (hash(x)=154573954)
672
+ 5580 train 4.271845 (lr=7.7987e-05) (hash(x)=149339361)
673
+ 5590 train 4.182251 (lr=7.7667e-05) (hash(x)=154738453)
674
+ 5600 val loss 4.3198
675
+ 5600 val perplexity 75.1747
676
+ 5600 train 4.288789 (lr=7.7347e-05) (hash(x)=147693222)
677
+ 5610 train 4.071845 (lr=7.7028e-05) (hash(x)=142140192)
678
+ 5620 train 4.192928 (lr=7.6710e-05) (hash(x)=148677209)
679
+ 5630 train 4.255360 (lr=7.6391e-05) (hash(x)=143495554)
680
+ 5640 train 4.361066 (lr=7.6074e-05) (hash(x)=153202964)
681
+ 5650 train 4.290414 (lr=7.5756e-05) (hash(x)=168724867)
682
+ 5660 train 4.257136 (lr=7.5440e-05) (hash(x)=142816746)
683
+ 5670 train 4.381474 (lr=7.5124e-05) (hash(x)=164119551)
684
+ 5680 train 4.238247 (lr=7.4808e-05) (hash(x)=157239121)
685
+ 5690 train 4.513042 (lr=7.4493e-05) (hash(x)=144662329)
686
+ 5700 val loss 4.3082
687
+ 5700 val perplexity 74.3091
688
+ 5700 train 4.254011 (lr=7.4178e-05) (hash(x)=149784627)
689
+ 5710 train 4.221472 (lr=7.3864e-05) (hash(x)=152821162)
690
+ 5720 train 4.448066 (lr=7.3550e-05) (hash(x)=158497530)
691
+ 5730 train 4.413035 (lr=7.3237e-05) (hash(x)=154583914)
692
+ 5740 train 4.223610 (lr=7.2925e-05) (hash(x)=142027172)
693
+ 5750 train 4.423906 (lr=7.2613e-05) (hash(x)=119227402)
694
+ 5760 train 4.329159 (lr=7.2301e-05) (hash(x)=153248187)
695
+ 5770 train 4.204040 (lr=7.1990e-05) (hash(x)=147951994)
696
+ 5780 train 4.202635 (lr=7.1680e-05) (hash(x)=138935284)
697
+ 5790 train 4.296416 (lr=7.1370e-05) (hash(x)=151786681)
698
+ 5800 val loss 4.3141
699
+ 5800 val perplexity 74.7481
700
+ 5800 train 4.311800 (lr=7.1061e-05) (hash(x)=158620729)
701
+ 5810 train 4.281051 (lr=7.0752e-05) (hash(x)=178115789)
702
+ 5820 train 4.202873 (lr=7.0444e-05) (hash(x)=149092790)
703
+ 5830 train 4.379194 (lr=7.0136e-05) (hash(x)=148269833)
704
+ 5840 train 4.312815 (lr=6.9829e-05) (hash(x)=163714738)
705
+ 5850 train 4.276379 (lr=6.9523e-05) (hash(x)=144111890)
706
+ 5860 train 4.275277 (lr=6.9217e-05) (hash(x)=170702475)
707
+ 5870 train 4.192323 (lr=6.8912e-05) (hash(x)=155970835)
708
+ 5880 train 4.193770 (lr=6.8607e-05) (hash(x)=146695463)
709
+ 5890 train 4.268687 (lr=6.8303e-05) (hash(x)=151603187)
710
+ 5900 val loss 4.3030
711
+ 5900 val perplexity 73.9237
712
+ 5900 train 4.286755 (lr=6.8000e-05) (hash(x)=159763910)
713
+ 5910 train 4.313234 (lr=6.7697e-05) (hash(x)=149555230)
714
+ 5920 train 4.309985 (lr=6.7395e-05) (hash(x)=138048333)
715
+ 5930 train 4.218314 (lr=6.7093e-05) (hash(x)=148668256)
716
+ 5940 train 4.355991 (lr=6.6792e-05) (hash(x)=148352543)
717
+ 5950 train 4.371619 (lr=6.6492e-05) (hash(x)=163457488)
718
+ 5960 train 4.459972 (lr=6.6192e-05) (hash(x)=155306804)
719
+ 5970 train 4.311144 (lr=6.5893e-05) (hash(x)=146576143)
720
+ 5980 train 4.308721 (lr=6.5595e-05) (hash(x)=144668956)
721
+ 5990 train 4.290262 (lr=6.5297e-05) (hash(x)=156098617)
722
+ 6000 val loss 4.2914
723
+ 6000 val perplexity 73.0668
724
+ 6000 train 4.249260 (lr=6.5000e-05) (hash(x)=147640561)
725
+ 6010 train 4.332212 (lr=6.4704e-05) (hash(x)=161223074)
726
+ 6020 train 4.314473 (lr=6.4408e-05) (hash(x)=156308454)
727
+ 6030 train 4.276718 (lr=6.4113e-05) (hash(x)=142553975)
728
+ 6040 train 4.251469 (lr=6.3818e-05) (hash(x)=142039829)
729
+ 6050 train 4.316525 (lr=6.3524e-05) (hash(x)=156329296)
730
+ 6060 train 4.254879 (lr=6.3231e-05) (hash(x)=148505537)
731
+ 6070 train 4.141137 (lr=6.2939e-05) (hash(x)=151801272)
732
+ 6080 train 4.239551 (lr=6.2647e-05) (hash(x)=159908287)
733
+ 6090 train 4.228786 (lr=6.2356e-05) (hash(x)=148714721)
734
+ 6100 val loss 4.2878
735
+ 6100 val perplexity 72.8035
736
+ 6100 train 4.291806 (lr=6.2065e-05) (hash(x)=156613394)
737
+ 6110 train 4.158760 (lr=6.1776e-05) (hash(x)=143286705)
738
+ 6120 train 4.358622 (lr=6.1487e-05) (hash(x)=150191141)
739
+ 6130 train 4.219504 (lr=6.1198e-05) (hash(x)=149855624)
740
+ 6140 train 4.221492 (lr=6.0911e-05) (hash(x)=157312372)
741
+ 6150 train 4.338254 (lr=6.0624e-05) (hash(x)=148262730)
742
+ 6160 train 4.247617 (lr=6.0338e-05) (hash(x)=147032785)
743
+ 6170 train 4.179667 (lr=6.0052e-05) (hash(x)=161213064)
744
+ 6180 train 4.046711 (lr=5.9767e-05) (hash(x)=144538373)
745
+ 6190 train 4.102021 (lr=5.9483e-05) (hash(x)=144688493)
746
+ 6200 val loss 4.2856
747
+ 6200 val perplexity 72.6446
748
+ 6200 train 4.170702 (lr=5.9200e-05) (hash(x)=186221290)
749
+ 6210 train 4.177794 (lr=5.8918e-05) (hash(x)=139529352)
750
+ 6220 train 4.351949 (lr=5.8636e-05) (hash(x)=138081469)
751
+ 6230 train 4.245850 (lr=5.8355e-05) (hash(x)=152979409)
752
+ 6240 train 4.203448 (lr=5.8074e-05) (hash(x)=157848424)
753
+ 6250 train 4.119584 (lr=5.7795e-05) (hash(x)=150770915)
754
+ 6260 train 4.211887 (lr=5.7516e-05) (hash(x)=153769379)
755
+ 6270 train 4.140826 (lr=5.7238e-05) (hash(x)=153318361)
756
+ 6280 train 4.130260 (lr=5.6961e-05) (hash(x)=143366327)
757
+ 6290 train 4.301881 (lr=5.6684e-05) (hash(x)=160684845)
758
+ 6300 val loss 4.2733
759
+ 6300 val perplexity 71.7590
760
+ 6300 train 4.184828 (lr=5.6409e-05) (hash(x)=152081419)
761
+ 6310 train 4.153626 (lr=5.6134e-05) (hash(x)=146626797)
762
+ 6320 train 4.335254 (lr=5.5859e-05) (hash(x)=155291923)
763
+ 6330 train 4.198305 (lr=5.5586e-05) (hash(x)=140920951)
764
+ 6340 train 4.275039 (lr=5.5313e-05) (hash(x)=146642359)
765
+ 6350 train 4.237655 (lr=5.5042e-05) (hash(x)=160486550)
766
+ 6360 train 4.203166 (lr=5.4771e-05) (hash(x)=151833082)
767
+ 6370 train 4.204446 (lr=5.4500e-05) (hash(x)=124740362)
768
+ 6380 train 4.365004 (lr=5.4231e-05) (hash(x)=153952434)
769
+ 6390 train 4.377242 (lr=5.3963e-05) (hash(x)=145864314)
770
+ 6400 val loss 4.2751
771
+ 6400 val perplexity 71.8861
772
+ 6400 train 4.270712 (lr=5.3695e-05) (hash(x)=154808349)
773
+ 6410 train 4.252228 (lr=5.3428e-05) (hash(x)=155037739)
774
+ 6420 train 4.185829 (lr=5.3162e-05) (hash(x)=151155588)
775
+ 6430 train 4.057166 (lr=5.2896e-05) (hash(x)=146548556)
776
+ 6440 train 4.231957 (lr=5.2632e-05) (hash(x)=152706474)
777
+ 6450 train 4.233659 (lr=5.2368e-05) (hash(x)=154207794)
778
+ 6460 train 4.092211 (lr=5.2105e-05) (hash(x)=141590515)
779
+ 6470 train 4.140141 (lr=5.1843e-05) (hash(x)=157673109)
780
+ 6480 train 4.162216 (lr=5.1582e-05) (hash(x)=155842980)
781
+ 6490 train 4.230108 (lr=5.1322e-05) (hash(x)=143595736)
782
+ 6500 val loss 4.2724
783
+ 6500 val perplexity 71.6936
784
+ 6500 train 4.186215 (lr=5.1063e-05) (hash(x)=159437208)
785
+ 6510 train 4.198236 (lr=5.0804e-05) (hash(x)=142631317)
786
+ 6520 train 3.914677 (lr=5.0546e-05) (hash(x)=139009211)
787
+ 6530 train 4.270006 (lr=5.0289e-05) (hash(x)=145451137)
788
+ 6540 train 4.228182 (lr=5.0033e-05) (hash(x)=163532779)
789
+ 6550 train 4.131004 (lr=4.9778e-05) (hash(x)=158084674)
790
+ 6560 train 4.096879 (lr=4.9524e-05) (hash(x)=142422592)
791
+ 6570 train 4.023828 (lr=4.9271e-05) (hash(x)=141089289)
792
+ 6580 train 4.123572 (lr=4.9018e-05) (hash(x)=139712979)
793
+ 6590 train 4.045252 (lr=4.8767e-05) (hash(x)=145833805)
794
+ 6600 val loss 4.2707
795
+ 6600 val perplexity 71.5721
796
+ 6600 train 4.117560 (lr=4.8516e-05) (hash(x)=157933074)
797
+ 6610 train 4.033614 (lr=4.8266e-05) (hash(x)=142534769)
798
+ 6620 train 4.093508 (lr=4.8017e-05) (hash(x)=138456044)
799
+ 6630 train 4.239148 (lr=4.7769e-05) (hash(x)=145111052)
800
+ 6640 train 4.199532 (lr=4.7522e-05) (hash(x)=166954515)
801
+ 6650 train 4.170330 (lr=4.7276e-05) (hash(x)=146889543)
802
+ 6660 train 4.174558 (lr=4.7030e-05) (hash(x)=143253726)
803
+ 6670 train 4.096807 (lr=4.6786e-05) (hash(x)=144349844)
804
+ 6680 train 4.278554 (lr=4.6542e-05) (hash(x)=139911928)
805
+ 6690 train 4.481636 (lr=4.6300e-05) (hash(x)=143135862)
806
+ 6700 val loss 4.2632
807
+ 6700 val perplexity 71.0375
808
+ 6700 train 4.205021 (lr=4.6058e-05) (hash(x)=161560240)
809
+ 6710 train 4.281558 (lr=4.5817e-05) (hash(x)=150437730)
810
+ 6720 train 4.368963 (lr=4.5578e-05) (hash(x)=153035304)
811
+ 6730 train 4.163786 (lr=4.5339e-05) (hash(x)=139323796)
812
+ 6740 train 4.181980 (lr=4.5101e-05) (hash(x)=167437923)
813
+ 6750 train 4.133231 (lr=4.4864e-05) (hash(x)=143814596)
814
+ 6760 train 4.219275 (lr=4.4628e-05) (hash(x)=143755031)
815
+ 6770 train 4.148920 (lr=4.4393e-05) (hash(x)=142014038)
816
+ 6780 train 4.217860 (lr=4.4159e-05) (hash(x)=148465062)
817
+ 6790 train 4.252775 (lr=4.3926e-05) (hash(x)=154656877)
818
+ 6800 val loss 4.2603
819
+ 6800 val perplexity 70.8338
820
+ 6800 train 4.342411 (lr=4.3693e-05) (hash(x)=155424292)
821
+ 6810 train 3.979886 (lr=4.3462e-05) (hash(x)=147358081)
822
+ 6820 train 4.113190 (lr=4.3232e-05) (hash(x)=145387291)
823
+ 6830 train 4.057794 (lr=4.3002e-05) (hash(x)=146334068)
824
+ 6840 train 4.128404 (lr=4.2774e-05) (hash(x)=139563766)
825
+ 6850 train 4.158656 (lr=4.2547e-05) (hash(x)=139095691)
826
+ 6860 train 4.107328 (lr=4.2320e-05) (hash(x)=142093999)
827
+ 6870 train 4.218301 (lr=4.2095e-05) (hash(x)=148598823)
828
+ 6880 train 4.294500 (lr=4.1870e-05) (hash(x)=146486085)
829
+ 6890 train 4.108667 (lr=4.1647e-05) (hash(x)=165683212)
830
+ 6900 val loss 4.2612
831
+ 6900 val perplexity 70.8978
832
+ 6900 train 4.301371 (lr=4.1425e-05) (hash(x)=148561470)
833
+ 6910 train 3.969691 (lr=4.1203e-05) (hash(x)=136352432)
834
+ 6920 train 4.365180 (lr=4.0983e-05) (hash(x)=147640094)
835
+ 6930 train 4.244941 (lr=4.0763e-05) (hash(x)=152343827)
836
+ 6940 train 4.223547 (lr=4.0545e-05) (hash(x)=158787971)
837
+ 6950 train 4.519065 (lr=4.0327e-05) (hash(x)=178133125)
838
+ 6960 train 4.317047 (lr=4.0111e-05) (hash(x)=167302853)
839
+ 6970 train 4.227910 (lr=3.9895e-05) (hash(x)=143729435)
840
+ 6980 train 4.324531 (lr=3.9681e-05) (hash(x)=135857879)
841
+ 6990 train 4.274819 (lr=3.9468e-05) (hash(x)=155384510)
842
+ 7000 val loss 4.2531
843
+ 7000 val perplexity 70.3226
844
+ 7000 train 4.210746 (lr=3.9255e-05) (hash(x)=141527450)
845
+ 7010 train 4.264535 (lr=3.9044e-05) (hash(x)=148648498)
846
+ 7020 train 4.178634 (lr=3.8834e-05) (hash(x)=148073842)
847
+ 7030 train 4.306561 (lr=3.8624e-05) (hash(x)=160771614)
848
+ 7040 train 4.081628 (lr=3.8416e-05) (hash(x)=148361330)
849
+ 7050 train 4.076358 (lr=3.8209e-05) (hash(x)=151212224)
850
+ 7060 train 4.274907 (lr=3.8003e-05) (hash(x)=155925366)
851
+ 7070 train 4.104435 (lr=3.7798e-05) (hash(x)=149830576)
852
+ 7080 train 4.087837 (lr=3.7593e-05) (hash(x)=144849934)
853
+ 7090 train 4.177107 (lr=3.7390e-05) (hash(x)=150997807)
854
+ 7100 val loss 4.2495
855
+ 7100 val perplexity 70.0730
856
+ 7100 train 4.215843 (lr=3.7188e-05) (hash(x)=151066339)
857
+ 7110 train 4.128480 (lr=3.6988e-05) (hash(x)=147185781)
858
+ 7120 train 4.168947 (lr=3.6788e-05) (hash(x)=141765575)
859
+ 7130 train 4.175840 (lr=3.6589e-05) (hash(x)=144204513)
860
+ 7140 train 4.008443 (lr=3.6391e-05) (hash(x)=149326253)
861
+ 7150 train 4.177330 (lr=3.6195e-05) (hash(x)=150409811)
862
+ 7160 train 4.053493 (lr=3.5999e-05) (hash(x)=144790348)
863
+ 7170 train 4.174676 (lr=3.5804e-05) (hash(x)=150500939)
864
+ 7180 train 4.000954 (lr=3.5611e-05) (hash(x)=143552103)
865
+ 7190 train 4.069571 (lr=3.5419e-05) (hash(x)=157471075)
866
+ 7200 val loss 4.2515
867
+ 7200 val perplexity 70.2112
868
+ 7200 train 4.032376 (lr=3.5227e-05) (hash(x)=155231264)
869
+ 7210 train 4.120383 (lr=3.5037e-05) (hash(x)=150350401)
870
+ 7220 train 4.135337 (lr=3.4848e-05) (hash(x)=151952594)
871
+ 7230 train 4.213009 (lr=3.4660e-05) (hash(x)=145381804)
872
+ 7240 train 4.113017 (lr=3.4473e-05) (hash(x)=142651222)
873
+ 7250 train 4.212279 (lr=3.4287e-05) (hash(x)=136115279)
874
+ 7260 train 4.327878 (lr=3.4102e-05) (hash(x)=154536751)
875
+ 7270 train 4.272999 (lr=3.3919e-05) (hash(x)=148883836)
876
+ 7280 train 4.129213 (lr=3.3736e-05) (hash(x)=150606754)
877
+ 7290 train 4.252479 (lr=3.3555e-05) (hash(x)=165845807)
878
+ 7300 val loss 4.2446
879
+ 7300 val perplexity 69.7298
880
+ 7300 train 4.323695 (lr=3.3375e-05) (hash(x)=150281149)
881
+ 7310 train 4.267867 (lr=3.3195e-05) (hash(x)=159929281)
882
+ 7320 train 4.088812 (lr=3.3017e-05) (hash(x)=143074269)
883
+ 7330 train 4.240971 (lr=3.2840e-05) (hash(x)=147369269)
884
+ 7340 train 4.287727 (lr=3.2664e-05) (hash(x)=139286601)
885
+ 7350 train 4.203074 (lr=3.2490e-05) (hash(x)=145263048)
886
+ 7360 train 4.057656 (lr=3.2316e-05) (hash(x)=144626817)
887
+ 7370 train 4.264996 (lr=3.2144e-05) (hash(x)=168989813)
888
+ 7380 train 4.225467 (lr=3.1972e-05) (hash(x)=144790656)
889
+ 7390 train 4.025857 (lr=3.1802e-05) (hash(x)=141279761)
890
+ 7400 val loss 4.2411
891
+ 7400 val perplexity 69.4830
892
+ 7400 train 4.167925 (lr=3.1633e-05) (hash(x)=148421717)
893
+ 7410 train 4.162103 (lr=3.1465e-05) (hash(x)=141241548)
894
+ 7420 train 4.231565 (lr=3.1298e-05) (hash(x)=162434228)
895
+ 7430 train 4.184918 (lr=3.1132e-05) (hash(x)=156312086)
896
+ 7440 train 4.132926 (lr=3.0968e-05) (hash(x)=139803064)
897
+ 7450 train 3.838727 (lr=3.0804e-05) (hash(x)=156361394)
898
+ 7460 train 4.141691 (lr=3.0642e-05) (hash(x)=141142183)
899
+ 7470 train 3.969291 (lr=3.0481e-05) (hash(x)=156701683)
900
+ 7480 train 4.150671 (lr=3.0321e-05) (hash(x)=148549917)
901
+ 7490 train 4.023565 (lr=3.0162e-05) (hash(x)=143002119)
902
+ 7500 val loss 4.2422
903
+ 7500 val perplexity 69.5633
904
+ 7500 train 4.035119 (lr=3.0005e-05) (hash(x)=146921118)
905
+ 7510 train 4.049643 (lr=2.9848e-05) (hash(x)=148592146)
906
+ 7520 train 4.201796 (lr=2.9693e-05) (hash(x)=148456120)
907
+ 7530 train 4.367846 (lr=2.9539e-05) (hash(x)=163865945)
908
+ 7540 train 4.261566 (lr=2.9386e-05) (hash(x)=160753039)
909
+ 7550 train 4.264345 (lr=2.9234e-05) (hash(x)=150275989)
910
+ 7560 train 4.287609 (lr=2.9084e-05) (hash(x)=141749735)
911
+ 7570 train 4.255280 (lr=2.8934e-05) (hash(x)=148449999)
912
+ 7580 train 4.131166 (lr=2.8786e-05) (hash(x)=139153889)
913
+ 7590 train 4.241937 (lr=2.8639e-05) (hash(x)=147418540)
914
+ 7600 val loss 4.2343
915
+ 7600 val perplexity 69.0153
916
+ 7600 train 4.269131 (lr=2.8493e-05) (hash(x)=150660048)
917
+ 7610 train 4.242627 (lr=2.8348e-05) (hash(x)=155369080)
918
+ 7620 train 4.141522 (lr=2.8204e-05) (hash(x)=151926309)
919
+ 7630 train 4.222584 (lr=2.8062e-05) (hash(x)=149549080)
920
+ 7640 train 4.108464 (lr=2.7921e-05) (hash(x)=140642882)
921
+ 7650 train 4.386451 (lr=2.7781e-05) (hash(x)=140218502)
922
+ 7660 train 4.230378 (lr=2.7642e-05) (hash(x)=153198276)
923
+ 7670 train 4.082559 (lr=2.7505e-05) (hash(x)=150706960)
924
+ 7680 train 4.237529 (lr=2.7368e-05) (hash(x)=152025509)
925
+ 7690 train 4.019795 (lr=2.7233e-05) (hash(x)=147288976)
926
+ 7700 val loss 4.2349
927
+ 7700 val perplexity 69.0561
928
+ 7700 train 4.127994 (lr=2.7099e-05) (hash(x)=148059852)
929
+ 7710 train 3.938552 (lr=2.6966e-05) (hash(x)=158443334)
930
+ 7720 train 4.054703 (lr=2.6834e-05) (hash(x)=143467085)
931
+ 7730 train 3.992703 (lr=2.6704e-05) (hash(x)=138439550)
932
+ 7740 train 4.135305 (lr=2.6575e-05) (hash(x)=145878152)
933
+ 7750 train 3.990562 (lr=2.6447e-05) (hash(x)=136874128)
934
+ 7760 train 4.016709 (lr=2.6320e-05) (hash(x)=148856154)
935
+ 7770 train 4.047473 (lr=2.6195e-05) (hash(x)=146982119)
936
+ 7780 train 4.143061 (lr=2.6070e-05) (hash(x)=159846779)
937
+ 7790 train 3.835675 (lr=2.5947e-05) (hash(x)=141501838)
938
+ 7800 val loss 4.2359
939
+ 7800 val perplexity 69.1239
940
+ 7800 train 3.944322 (lr=2.5825e-05) (hash(x)=148331002)
941
+ 7810 train 4.263639 (lr=2.5705e-05) (hash(x)=150692304)
942
+ 7820 train 4.148048 (lr=2.5585e-05) (hash(x)=156841183)
943
+ 7830 train 4.175803 (lr=2.5467e-05) (hash(x)=139594008)
944
+ 7840 train 4.114260 (lr=2.5350e-05) (hash(x)=149700138)
945
+ 7850 train 4.137405 (lr=2.5234e-05) (hash(x)=149327044)
946
+ 7860 train 4.290323 (lr=2.5119e-05) (hash(x)=148315418)
947
+ 7870 train 4.245211 (lr=2.5006e-05) (hash(x)=153431778)
948
+ 7880 train 4.096437 (lr=2.4894e-05) (hash(x)=145049799)
949
+ 7890 train 4.228706 (lr=2.4783e-05) (hash(x)=150265096)
950
+ 7900 val loss 4.2289
951
+ 7900 val perplexity 68.6449
952
+ 7900 train 4.077190 (lr=2.4674e-05) (hash(x)=164923883)
953
+ 7910 train 4.089196 (lr=2.4565e-05) (hash(x)=139315738)
954
+ 7920 train 4.157056 (lr=2.4458e-05) (hash(x)=153031394)
955
+ 7930 train 3.925258 (lr=2.4352e-05) (hash(x)=148835859)
956
+ 7940 train 4.074548 (lr=2.4247e-05) (hash(x)=146875217)
957
+ 7950 train 4.154520 (lr=2.4144e-05) (hash(x)=151003885)
958
+ 7960 train 4.067237 (lr=2.4042e-05) (hash(x)=141885471)
959
+ 7970 train 4.261871 (lr=2.3941e-05) (hash(x)=150541583)
960
+ 7980 train 3.957344 (lr=2.3841e-05) (hash(x)=138626492)
961
+ 7990 train 4.159150 (lr=2.3743e-05) (hash(x)=149935635)
962
+ 8000 val loss 4.2287
963
+ 8000 val perplexity 68.6285
964
+ 8000 train 4.128093 (lr=2.3646e-05) (hash(x)=143545384)
965
+ 8010 train 3.969240 (lr=2.3550e-05) (hash(x)=144689366)
966
+ 8020 train 4.162443 (lr=2.3455e-05) (hash(x)=151299532)
967
+ 8030 train 4.086966 (lr=2.3362e-05) (hash(x)=131300239)
968
+ 8040 train 3.995358 (lr=2.3269e-05) (hash(x)=146715955)
969
+ 8050 train 3.923275 (lr=2.3179e-05) (hash(x)=135376656)
970
+ 8060 train 4.001976 (lr=2.3089e-05) (hash(x)=141991433)
971
+ 8070 train 3.973842 (lr=2.3001e-05) (hash(x)=140260574)
972
+ 8080 train 3.940057 (lr=2.2913e-05) (hash(x)=151492896)
973
+ 8090 train 4.133124 (lr=2.2828e-05) (hash(x)=143333635)
974
+ 8100 val loss 4.2305
975
+ 8100 val perplexity 68.7508
976
+ 8100 train 4.017842 (lr=2.2743e-05) (hash(x)=160686959)
977
+ 8110 train 4.021292 (lr=2.2660e-05) (hash(x)=141145704)
978
+ 8120 train 4.258512 (lr=2.2578e-05) (hash(x)=146967795)
979
+ 8130 train 4.088595 (lr=2.2497e-05) (hash(x)=147844486)
980
+ 8140 train 4.136693 (lr=2.2417e-05) (hash(x)=141860136)
981
+ 8150 train 4.269496 (lr=2.2339e-05) (hash(x)=147696006)
982
+ 8160 train 4.176379 (lr=2.2262e-05) (hash(x)=151497110)
983
+ 8170 train 4.186491 (lr=2.2186e-05) (hash(x)=140881859)
984
+ 8180 train 4.133401 (lr=2.2112e-05) (hash(x)=158865574)
985
+ 8190 train 4.101454 (lr=2.2039e-05) (hash(x)=161247161)
986
+ 8200 val loss 4.2260
987
+ 8200 val perplexity 68.4445
988
+ 8200 train 4.056613 (lr=2.1967e-05) (hash(x)=156501889)
989
+ 8210 train 4.177739 (lr=2.1896e-05) (hash(x)=158384162)
990
+ 8220 train 4.218917 (lr=2.1827e-05) (hash(x)=154815062)
991
+ 8230 train 4.093947 (lr=2.1759e-05) (hash(x)=145018315)
992
+ 8240 train 4.139846 (lr=2.1692e-05) (hash(x)=161659129)
993
+ 8250 train 4.105736 (lr=2.1626e-05) (hash(x)=139863367)
994
+ 8260 train 4.070853 (lr=2.1562e-05) (hash(x)=149569663)
995
+ 8270 train 3.834073 (lr=2.1499e-05) (hash(x)=162234378)
996
+ 8280 train 4.191489 (lr=2.1438e-05) (hash(x)=144371378)
997
+ 8290 train 4.101012 (lr=2.1377e-05) (hash(x)=148120258)
998
+ 8300 val loss 4.2262
999
+ 8300 val perplexity 68.4574
1000
+ 8300 train 4.075238 (lr=2.1318e-05) (hash(x)=142716875)
1001
+ 8310 train 4.137279 (lr=2.1260e-05) (hash(x)=166789234)
1002
+ 8320 train 4.168675 (lr=2.1204e-05) (hash(x)=150613211)
1003
+ 8330 train 4.053012 (lr=2.1149e-05) (hash(x)=139895391)
1004
+ 8340 train 4.169483 (lr=2.1095e-05) (hash(x)=152011203)
1005
+ 8350 train 4.124813 (lr=2.1042e-05) (hash(x)=156010875)
1006
+ 8360 train 4.133077 (lr=2.0991e-05) (hash(x)=152988051)
1007
+ 8370 train 4.287643 (lr=2.0941e-05) (hash(x)=155144327)
1008
+ 8380 train 4.174703 (lr=2.0892e-05) (hash(x)=133642309)
1009
+ 8390 train 4.226175 (lr=2.0844e-05) (hash(x)=149745335)
1010
+ 8400 val loss 4.2195
1011
+ 8400 val perplexity 67.9998
1012
+ 8400 train 4.204172 (lr=2.0798e-05) (hash(x)=154436684)
1013
+ 8410 train 4.242471 (lr=2.0753e-05) (hash(x)=147946630)
1014
+ 8420 train 4.164583 (lr=2.0710e-05) (hash(x)=153947058)
1015
+ 8430 train 4.259352 (lr=2.0667e-05) (hash(x)=159821040)
1016
+ 8440 train 4.224552 (lr=2.0626e-05) (hash(x)=149127348)
1017
+ 8450 train 4.234138 (lr=2.0587e-05) (hash(x)=149421871)
1018
+ 8460 train 4.176043 (lr=2.0548e-05) (hash(x)=142621577)
1019
+ 8470 train 4.182089 (lr=2.0511e-05) (hash(x)=148978213)
1020
+ 8480 train 4.012268 (lr=2.0475e-05) (hash(x)=139959537)
1021
+ 8490 train 4.102921 (lr=2.0441e-05) (hash(x)=144232498)
1022
+ 8500 val loss 4.2189
1023
+ 8500 val perplexity 67.9565
1024
+ 8500 train 4.314741 (lr=2.0408e-05) (hash(x)=147965839)
1025
+ 8510 train 4.191538 (lr=2.0376e-05) (hash(x)=147679836)
1026
+ 8520 train 4.069478 (lr=2.0345e-05) (hash(x)=147486097)
1027
+ 8530 train 4.199194 (lr=2.0316e-05) (hash(x)=151655265)
1028
+ 8540 train 4.108413 (lr=2.0288e-05) (hash(x)=150866632)
1029
+ 8550 train 4.263029 (lr=2.0261e-05) (hash(x)=144355059)
1030
+ 8560 train 4.112020 (lr=2.0235e-05) (hash(x)=145518478)
1031
+ 8570 train 4.289657 (lr=2.0211e-05) (hash(x)=178973323)
1032
+ 8580 train 4.184974 (lr=2.0189e-05) (hash(x)=152773969)
1033
+ 8590 train 3.978383 (lr=2.0167e-05) (hash(x)=141805508)
1034
+ 8600 val loss 4.2210
1035
+ 8600 val perplexity 68.1044
1036
+ 8600 train 4.028108 (lr=2.0147e-05) (hash(x)=145228097)
1037
+ 8610 train 4.060266 (lr=2.0128e-05) (hash(x)=149525694)
1038
+ 8620 train 4.131866 (lr=2.0110e-05) (hash(x)=151651655)
1039
+ 8630 train 4.250931 (lr=2.0094e-05) (hash(x)=144918643)
1040
+ 8640 train 4.103380 (lr=2.0079e-05) (hash(x)=149721879)
1041
+ 8650 train 4.176823 (lr=2.0065e-05) (hash(x)=157684127)
1042
+ 8660 train 4.177157 (lr=2.0053e-05) (hash(x)=152426036)
1043
+ 8670 train 4.079011 (lr=2.0042e-05) (hash(x)=141396342)
1044
+ 8680 train 4.197928 (lr=2.0032e-05) (hash(x)=146459856)
1045
+ 8690 train 4.276552 (lr=2.0023e-05) (hash(x)=151766124)
1046
+ 8700 val loss 4.2146
1047
+ 8700 val perplexity 67.6705
1048
+ 8700 train 4.418008 (lr=2.0016e-05) (hash(x)=152910357)
1049
+ 8710 train 4.221392 (lr=2.0010e-05) (hash(x)=149724642)
1050
+ 8720 train 4.146652 (lr=2.0006e-05) (hash(x)=152483840)
1051
+ 8730 train 4.123717 (lr=2.0003e-05) (hash(x)=147289415)
1052
+ 8740 train 4.167049 (lr=2.0001e-05) (hash(x)=152254131)
1053
+ 8749 val loss 4.2154
1054
+ 8749 val perplexity 67.7183
lr2e-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 sha256:80af035c29db10c47090f16ed0228f4c53371e0dc288c0edb144b4d3270e5984
3
+ size 92843394
lr2e-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 sha256:98658f80391492412c89f5266e7814b20ba0a6badfea583718751cdad63c4c39
3
+ size 179406214