andrew-healey commited on
Commit
f27610f
·
verified ·
1 Parent(s): efae75f

Upload folder using huggingface_hub

Browse files
lr8e-4_total_batch_size61440_baseline_seed1338/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_11/lr8e-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": 0.0008, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "8e-4_61440", "n_embd": 256}
lr8e-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
lr8e-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=1.6000e-06) (hash(x)=150327452)
5
+ 10 train 10.164598 (lr=1.7600e-05) (hash(x)=165603215)
6
+ 20 train 9.758929 (lr=3.3600e-05) (hash(x)=142061326)
7
+ 30 train 9.403145 (lr=4.9600e-05) (hash(x)=147690730)
8
+ 40 train 9.040249 (lr=6.5600e-05) (hash(x)=160872542)
9
+ 50 train 8.565110 (lr=8.1600e-05) (hash(x)=166441190)
10
+ 60 train 8.124584 (lr=9.7600e-05) (hash(x)=140536766)
11
+ 70 train 7.777061 (lr=1.1360e-04) (hash(x)=146052944)
12
+ 80 train 7.601630 (lr=1.2960e-04) (hash(x)=140542391)
13
+ 90 train 7.461853 (lr=1.4560e-04) (hash(x)=147098660)
14
+ 100 val loss 7.4791
15
+ 100 val perplexity 1770.6570
16
+ 100 train 7.449233 (lr=1.6160e-04) (hash(x)=166780046)
17
+ 110 train 7.235006 (lr=1.7760e-04) (hash(x)=143376584)
18
+ 120 train 7.172765 (lr=1.9360e-04) (hash(x)=145337630)
19
+ 130 train 7.142758 (lr=2.0960e-04) (hash(x)=147524744)
20
+ 140 train 7.040697 (lr=2.2560e-04) (hash(x)=155587442)
21
+ 150 train 6.926411 (lr=2.4160e-04) (hash(x)=159835303)
22
+ 160 train 6.740836 (lr=2.5760e-04) (hash(x)=142171630)
23
+ 170 train 6.644814 (lr=2.7360e-04) (hash(x)=136930463)
24
+ 180 train 6.621518 (lr=2.8960e-04) (hash(x)=153579148)
25
+ 190 train 6.500198 (lr=3.0560e-04) (hash(x)=140352771)
26
+ 200 val loss 6.7028
27
+ 200 val perplexity 814.6483
28
+ 200 train 6.425660 (lr=3.2160e-04) (hash(x)=155040610)
29
+ 210 train 6.344302 (lr=3.3760e-04) (hash(x)=141273862)
30
+ 220 train 6.560600 (lr=3.5360e-04) (hash(x)=165326618)
31
+ 230 train 6.305447 (lr=3.6960e-04) (hash(x)=150122120)
32
+ 240 train 6.105167 (lr=3.8560e-04) (hash(x)=143024679)
33
+ 250 train 6.091118 (lr=4.0160e-04) (hash(x)=130190460)
34
+ 260 train 6.205239 (lr=4.1760e-04) (hash(x)=151890219)
35
+ 270 train 6.003814 (lr=4.3360e-04) (hash(x)=143685291)
36
+ 280 train 6.308943 (lr=4.4960e-04) (hash(x)=137108194)
37
+ 290 train 6.437833 (lr=4.6560e-04) (hash(x)=157920866)
38
+ 300 val loss 6.2329
39
+ 300 val perplexity 509.2484
40
+ 300 train 6.159818 (lr=4.8160e-04) (hash(x)=155504036)
41
+ 310 train 6.257793 (lr=4.9760e-04) (hash(x)=144551186)
42
+ 320 train 6.260306 (lr=5.1360e-04) (hash(x)=161180086)
43
+ 330 train 6.004916 (lr=5.2960e-04) (hash(x)=141651471)
44
+ 340 train 6.123817 (lr=5.4560e-04) (hash(x)=157490741)
45
+ 350 train 6.079549 (lr=5.6160e-04) (hash(x)=137347213)
46
+ 360 train 6.338968 (lr=5.7760e-04) (hash(x)=171273659)
47
+ 370 train 5.970878 (lr=5.9360e-04) (hash(x)=152118871)
48
+ 380 train 5.955096 (lr=6.0960e-04) (hash(x)=161338097)
49
+ 390 train 6.035947 (lr=6.2560e-04) (hash(x)=152404537)
50
+ 400 val loss 5.8921
51
+ 400 val perplexity 362.1648
52
+ 400 train 5.736683 (lr=6.4160e-04) (hash(x)=143823248)
53
+ 410 train 6.003430 (lr=6.5760e-04) (hash(x)=161686752)
54
+ 420 train 5.718559 (lr=6.7360e-04) (hash(x)=155391312)
55
+ 430 train 5.624483 (lr=6.8960e-04) (hash(x)=142145507)
56
+ 440 train 5.762999 (lr=7.0560e-04) (hash(x)=121931091)
57
+ 450 train 5.709932 (lr=7.2160e-04) (hash(x)=156260416)
58
+ 460 train 5.857639 (lr=7.3760e-04) (hash(x)=169781623)
59
+ 470 train 5.630699 (lr=7.5360e-04) (hash(x)=150131545)
60
+ 480 train 5.656713 (lr=7.6960e-04) (hash(x)=150443184)
61
+ 490 train 5.652027 (lr=7.8560e-04) (hash(x)=149368198)
62
+ 500 val loss 5.6768
63
+ 500 val perplexity 292.0229
64
+ 500 train 5.565771 (lr=8.0000e-04) (hash(x)=143734685)
65
+ 510 train 5.585708 (lr=8.0000e-04) (hash(x)=148943891)
66
+ 520 train 5.599650 (lr=7.9999e-04) (hash(x)=149070354)
67
+ 530 train 5.578380 (lr=7.9998e-04) (hash(x)=153601219)
68
+ 540 train 5.592160 (lr=7.9996e-04) (hash(x)=174970654)
69
+ 550 train 5.517368 (lr=7.9993e-04) (hash(x)=160013925)
70
+ 560 train 5.368655 (lr=7.9991e-04) (hash(x)=152096988)
71
+ 570 train 5.480852 (lr=7.9987e-04) (hash(x)=157203924)
72
+ 580 train 5.251802 (lr=7.9983e-04) (hash(x)=138582113)
73
+ 590 train 5.275807 (lr=7.9979e-04) (hash(x)=157451775)
74
+ 600 val loss 5.4965
75
+ 600 val perplexity 243.8253
76
+ 600 train 5.205692 (lr=7.9974e-04) (hash(x)=150678249)
77
+ 610 train 5.248729 (lr=7.9968e-04) (hash(x)=150248072)
78
+ 620 train 5.139501 (lr=7.9962e-04) (hash(x)=148286909)
79
+ 630 train 5.107310 (lr=7.9956e-04) (hash(x)=141197178)
80
+ 640 train 5.186721 (lr=7.9949e-04) (hash(x)=144988391)
81
+ 650 train 5.113807 (lr=7.9941e-04) (hash(x)=149073315)
82
+ 660 train 5.055437 (lr=7.9933e-04) (hash(x)=145239943)
83
+ 670 train 5.081699 (lr=7.9925e-04) (hash(x)=146657426)
84
+ 680 train 5.309408 (lr=7.9915e-04) (hash(x)=142323429)
85
+ 690 train 5.283567 (lr=7.9906e-04) (hash(x)=142724342)
86
+ 700 val loss 5.3447
87
+ 700 val perplexity 209.4915
88
+ 700 train 5.196348 (lr=7.9896e-04) (hash(x)=175802021)
89
+ 710 train 5.263361 (lr=7.9885e-04) (hash(x)=148902690)
90
+ 720 train 5.346172 (lr=7.9874e-04) (hash(x)=155151320)
91
+ 730 train 5.203635 (lr=7.9862e-04) (hash(x)=142078914)
92
+ 740 train 5.354824 (lr=7.9850e-04) (hash(x)=155634484)
93
+ 750 train 5.324068 (lr=7.9837e-04) (hash(x)=171034639)
94
+ 760 train 5.283617 (lr=7.9824e-04) (hash(x)=142470148)
95
+ 770 train 5.243272 (lr=7.9810e-04) (hash(x)=156024448)
96
+ 780 train 5.135063 (lr=7.9796e-04) (hash(x)=151369417)
97
+ 790 train 5.090515 (lr=7.9781e-04) (hash(x)=139260056)
98
+ 800 val loss 5.1868
99
+ 800 val perplexity 178.9006
100
+ 800 train 5.312267 (lr=7.9765e-04) (hash(x)=158681215)
101
+ 810 train 5.100136 (lr=7.9749e-04) (hash(x)=144884161)
102
+ 820 train 5.157301 (lr=7.9733e-04) (hash(x)=146977514)
103
+ 830 train 5.023214 (lr=7.9716e-04) (hash(x)=148983636)
104
+ 840 train 5.077933 (lr=7.9699e-04) (hash(x)=157824542)
105
+ 850 train 5.045485 (lr=7.9681e-04) (hash(x)=152116061)
106
+ 860 train 5.138885 (lr=7.9662e-04) (hash(x)=153743703)
107
+ 870 train 5.252891 (lr=7.9643e-04) (hash(x)=149913335)
108
+ 880 train 5.038424 (lr=7.9624e-04) (hash(x)=151096434)
109
+ 890 train 5.017186 (lr=7.9604e-04) (hash(x)=145848702)
110
+ 900 val loss 5.0735
111
+ 900 val perplexity 159.7330
112
+ 900 train 4.996807 (lr=7.9583e-04) (hash(x)=146108145)
113
+ 910 train 4.942824 (lr=7.9562e-04) (hash(x)=146903806)
114
+ 920 train 4.898376 (lr=7.9541e-04) (hash(x)=149225019)
115
+ 930 train 4.847265 (lr=7.9518e-04) (hash(x)=146110422)
116
+ 940 train 4.920490 (lr=7.9496e-04) (hash(x)=161333614)
117
+ 950 train 4.831565 (lr=7.9473e-04) (hash(x)=147598108)
118
+ 960 train 4.962214 (lr=7.9449e-04) (hash(x)=158133101)
119
+ 970 train 4.794880 (lr=7.9425e-04) (hash(x)=152199591)
120
+ 980 train 4.759202 (lr=7.9400e-04) (hash(x)=152359036)
121
+ 990 train 4.792173 (lr=7.9375e-04) (hash(x)=145166335)
122
+ 1000 val loss 4.9927
123
+ 1000 val perplexity 147.3398
124
+ 1000 train 4.694622 (lr=7.9349e-04) (hash(x)=154996086)
125
+ 1010 train 4.622330 (lr=7.9323e-04) (hash(x)=149406975)
126
+ 1020 train 4.744334 (lr=7.9297e-04) (hash(x)=144618270)
127
+ 1030 train 4.594065 (lr=7.9269e-04) (hash(x)=148306536)
128
+ 1040 train 4.630183 (lr=7.9242e-04) (hash(x)=151643553)
129
+ 1050 train 4.571986 (lr=7.9213e-04) (hash(x)=153396183)
130
+ 1060 train 4.587943 (lr=7.9185e-04) (hash(x)=157501787)
131
+ 1070 train 4.781093 (lr=7.9155e-04) (hash(x)=153277507)
132
+ 1080 train 4.852253 (lr=7.9126e-04) (hash(x)=156232389)
133
+ 1090 train 4.800875 (lr=7.9095e-04) (hash(x)=150673143)
134
+ 1100 val loss 4.8813
135
+ 1100 val perplexity 131.8062
136
+ 1100 train 4.868817 (lr=7.9064e-04) (hash(x)=153885445)
137
+ 1110 train 4.872423 (lr=7.9033e-04) (hash(x)=143951073)
138
+ 1120 train 5.274862 (lr=7.9001e-04) (hash(x)=151626904)
139
+ 1130 train 4.844402 (lr=7.8969e-04) (hash(x)=144013074)
140
+ 1140 train 4.769439 (lr=7.8936e-04) (hash(x)=149948988)
141
+ 1150 train 4.838509 (lr=7.8903e-04) (hash(x)=159666385)
142
+ 1160 train 4.803688 (lr=7.8869e-04) (hash(x)=146889587)
143
+ 1170 train 4.863293 (lr=7.8835e-04) (hash(x)=151746917)
144
+ 1180 train 4.880298 (lr=7.8800e-04) (hash(x)=154396948)
145
+ 1190 train 4.824942 (lr=7.8764e-04) (hash(x)=148201222)
146
+ 1200 val loss 4.8075
147
+ 1200 val perplexity 122.4259
148
+ 1200 train 4.819984 (lr=7.8729e-04) (hash(x)=142353087)
149
+ 1210 train 4.820332 (lr=7.8692e-04) (hash(x)=146779775)
150
+ 1220 train 4.881454 (lr=7.8655e-04) (hash(x)=155864740)
151
+ 1230 train 4.925190 (lr=7.8618e-04) (hash(x)=129826485)
152
+ 1240 train 4.768026 (lr=7.8580e-04) (hash(x)=108599491)
153
+ 1250 train 4.669080 (lr=7.8542e-04) (hash(x)=146491718)
154
+ 1260 train 4.743132 (lr=7.8503e-04) (hash(x)=141897382)
155
+ 1270 train 4.721516 (lr=7.8464e-04) (hash(x)=144830380)
156
+ 1280 train 4.769186 (lr=7.8424e-04) (hash(x)=157677763)
157
+ 1290 train 4.599349 (lr=7.8383e-04) (hash(x)=143611746)
158
+ 1300 val loss 4.7537
159
+ 1300 val perplexity 116.0140
160
+ 1300 train 4.626343 (lr=7.8342e-04) (hash(x)=150750353)
161
+ 1310 train 4.864161 (lr=7.8301e-04) (hash(x)=152850907)
162
+ 1320 train 4.771248 (lr=7.8259e-04) (hash(x)=148575395)
163
+ 1330 train 4.522063 (lr=7.8217e-04) (hash(x)=157113001)
164
+ 1340 train 4.574368 (lr=7.8174e-04) (hash(x)=137400723)
165
+ 1350 train 4.629392 (lr=7.8131e-04) (hash(x)=129849193)
166
+ 1360 train 4.607620 (lr=7.8087e-04) (hash(x)=144176797)
167
+ 1370 train 4.447096 (lr=7.8042e-04) (hash(x)=144550831)
168
+ 1380 train 4.413111 (lr=7.7998e-04) (hash(x)=144171821)
169
+ 1390 train 4.356477 (lr=7.7952e-04) (hash(x)=146110449)
170
+ 1400 val loss 4.7103
171
+ 1400 val perplexity 111.0829
172
+ 1400 train 4.436737 (lr=7.7906e-04) (hash(x)=152767913)
173
+ 1410 train 4.437268 (lr=7.7860e-04) (hash(x)=148110947)
174
+ 1420 train 4.319122 (lr=7.7813e-04) (hash(x)=153477905)
175
+ 1430 train 4.212344 (lr=7.7766e-04) (hash(x)=147227598)
176
+ 1440 train 4.432220 (lr=7.7718e-04) (hash(x)=146280368)
177
+ 1450 train 4.402471 (lr=7.7670e-04) (hash(x)=146531140)
178
+ 1460 train 4.242182 (lr=7.7621e-04) (hash(x)=151573873)
179
+ 1470 train 4.729385 (lr=7.7572e-04) (hash(x)=155559275)
180
+ 1480 train 4.621179 (lr=7.7522e-04) (hash(x)=140584826)
181
+ 1490 train 4.583060 (lr=7.7472e-04) (hash(x)=165070637)
182
+ 1500 val loss 4.6812
183
+ 1500 val perplexity 107.8981
184
+ 1500 train 4.700862 (lr=7.7421e-04) (hash(x)=151562048)
185
+ 1510 train 4.474503 (lr=7.7370e-04) (hash(x)=147770607)
186
+ 1520 train 4.558227 (lr=7.7318e-04) (hash(x)=156486153)
187
+ 1530 train 4.519832 (lr=7.7266e-04) (hash(x)=140244564)
188
+ 1540 train 4.573845 (lr=7.7214e-04) (hash(x)=151627549)
189
+ 1550 train 4.614678 (lr=7.7160e-04) (hash(x)=146001424)
190
+ 1560 train 4.683183 (lr=7.7107e-04) (hash(x)=184266264)
191
+ 1570 train 4.596179 (lr=7.7053e-04) (hash(x)=145002091)
192
+ 1580 train 4.485724 (lr=7.6998e-04) (hash(x)=132216826)
193
+ 1590 train 4.657141 (lr=7.6943e-04) (hash(x)=160975703)
194
+ 1600 val loss 4.6271
195
+ 1600 val perplexity 102.2208
196
+ 1600 train 4.780918 (lr=7.6888e-04) (hash(x)=166486165)
197
+ 1610 train 4.665181 (lr=7.6832e-04) (hash(x)=149601833)
198
+ 1620 train 4.554275 (lr=7.6775e-04) (hash(x)=157815188)
199
+ 1630 train 4.571677 (lr=7.6718e-04) (hash(x)=144234679)
200
+ 1640 train 4.702630 (lr=7.6661e-04) (hash(x)=160054560)
201
+ 1650 train 4.520803 (lr=7.6603e-04) (hash(x)=141061333)
202
+ 1660 train 4.645567 (lr=7.6545e-04) (hash(x)=148064219)
203
+ 1670 train 4.485071 (lr=7.6486e-04) (hash(x)=162799669)
204
+ 1680 train 4.526896 (lr=7.6426e-04) (hash(x)=147474277)
205
+ 1690 train 4.590459 (lr=7.6367e-04) (hash(x)=142369299)
206
+ 1700 val loss 4.6088
207
+ 1700 val perplexity 100.3609
208
+ 1700 train 4.366702 (lr=7.6306e-04) (hash(x)=130835396)
209
+ 1710 train 4.704501 (lr=7.6246e-04) (hash(x)=153173892)
210
+ 1720 train 4.361197 (lr=7.6184e-04) (hash(x)=148451191)
211
+ 1730 train 4.472566 (lr=7.6123e-04) (hash(x)=145047239)
212
+ 1740 train 4.379878 (lr=7.6061e-04) (hash(x)=147187109)
213
+ 1750 train 4.476208 (lr=7.5998e-04) (hash(x)=149493155)
214
+ 1760 train 4.432001 (lr=7.5935e-04) (hash(x)=147647531)
215
+ 1770 train 4.682613 (lr=7.5872e-04) (hash(x)=153677279)
216
+ 1780 train 4.495909 (lr=7.5808e-04) (hash(x)=144314015)
217
+ 1790 train 4.515595 (lr=7.5743e-04) (hash(x)=176962883)
218
+ 1800 val loss 4.5691
219
+ 1800 val perplexity 96.4547
220
+ 1800 train 4.618306 (lr=7.5678e-04) (hash(x)=158851816)
221
+ 1810 train 4.201349 (lr=7.5613e-04) (hash(x)=175415447)
222
+ 1820 train 4.506949 (lr=7.5547e-04) (hash(x)=169829733)
223
+ 1830 train 4.484717 (lr=7.5481e-04) (hash(x)=154116106)
224
+ 1840 train 4.321154 (lr=7.5414e-04) (hash(x)=150168406)
225
+ 1850 train 4.316216 (lr=7.5347e-04) (hash(x)=145782493)
226
+ 1860 train 4.294943 (lr=7.5279e-04) (hash(x)=158111664)
227
+ 1870 train 4.247147 (lr=7.5211e-04) (hash(x)=154837061)
228
+ 1880 train 4.192833 (lr=7.5143e-04) (hash(x)=151363745)
229
+ 1890 train 4.308922 (lr=7.5074e-04) (hash(x)=161283557)
230
+ 1900 val loss 4.5526
231
+ 1900 val perplexity 94.8755
232
+ 1900 train 4.556006 (lr=7.5004e-04) (hash(x)=153313879)
233
+ 1910 train 4.377191 (lr=7.4934e-04) (hash(x)=152272272)
234
+ 1920 train 4.449458 (lr=7.4864e-04) (hash(x)=153648837)
235
+ 1930 train 4.545497 (lr=7.4793e-04) (hash(x)=150161694)
236
+ 1940 train 4.421068 (lr=7.4722e-04) (hash(x)=151235093)
237
+ 1950 train 4.476758 (lr=7.4650e-04) (hash(x)=140582058)
238
+ 1960 train 4.420548 (lr=7.4578e-04) (hash(x)=125616059)
239
+ 1970 train 4.548450 (lr=7.4505e-04) (hash(x)=154353968)
240
+ 1980 train 4.573178 (lr=7.4432e-04) (hash(x)=156063143)
241
+ 1990 train 4.591631 (lr=7.4359e-04) (hash(x)=160927289)
242
+ 2000 val loss 4.5043
243
+ 2000 val perplexity 90.4035
244
+ 2000 train 4.485631 (lr=7.4285e-04) (hash(x)=158245023)
245
+ 2010 train 4.544114 (lr=7.4211e-04) (hash(x)=157096661)
246
+ 2020 train 4.754180 (lr=7.4136e-04) (hash(x)=144994302)
247
+ 2030 train 4.456479 (lr=7.4061e-04) (hash(x)=145742929)
248
+ 2040 train 4.469091 (lr=7.3985e-04) (hash(x)=156958528)
249
+ 2050 train 4.981962 (lr=7.3909e-04) (hash(x)=135313020)
250
+ 2060 train 4.623191 (lr=7.3833e-04) (hash(x)=147309332)
251
+ 2070 train 4.408035 (lr=7.3756e-04) (hash(x)=161720962)
252
+ 2080 train 4.588979 (lr=7.3678e-04) (hash(x)=167204793)
253
+ 2090 train 4.493253 (lr=7.3600e-04) (hash(x)=141325510)
254
+ 2100 val loss 4.4855
255
+ 2100 val perplexity 88.7196
256
+ 2100 train 4.309488 (lr=7.3522e-04) (hash(x)=157204896)
257
+ 2110 train 4.271603 (lr=7.3444e-04) (hash(x)=143077993)
258
+ 2120 train 4.474487 (lr=7.3364e-04) (hash(x)=159032689)
259
+ 2130 train 4.359325 (lr=7.3285e-04) (hash(x)=149513855)
260
+ 2140 train 4.275223 (lr=7.3205e-04) (hash(x)=151678267)
261
+ 2150 train 4.375750 (lr=7.3125e-04) (hash(x)=141309246)
262
+ 2160 train 4.467722 (lr=7.3044e-04) (hash(x)=153935732)
263
+ 2170 train 4.363901 (lr=7.2963e-04) (hash(x)=150668875)
264
+ 2180 train 4.371768 (lr=7.2881e-04) (hash(x)=142033123)
265
+ 2190 train 4.408647 (lr=7.2799e-04) (hash(x)=157015026)
266
+ 2200 val loss 4.4703
267
+ 2200 val perplexity 87.3810
268
+ 2200 train 4.290910 (lr=7.2716e-04) (hash(x)=137541932)
269
+ 2210 train 4.209034 (lr=7.2634e-04) (hash(x)=144424504)
270
+ 2220 train 4.140424 (lr=7.2550e-04) (hash(x)=151015810)
271
+ 2230 train 4.213646 (lr=7.2467e-04) (hash(x)=160707806)
272
+ 2240 train 4.224332 (lr=7.2382e-04) (hash(x)=151874110)
273
+ 2250 train 4.192152 (lr=7.2298e-04) (hash(x)=157341469)
274
+ 2260 train 4.016766 (lr=7.2213e-04) (hash(x)=144563224)
275
+ 2270 train 4.172267 (lr=7.2128e-04) (hash(x)=144878459)
276
+ 2280 train 4.241380 (lr=7.2042e-04) (hash(x)=158331565)
277
+ 2290 train 4.265651 (lr=7.1956e-04) (hash(x)=146875132)
278
+ 2300 val loss 4.4559
279
+ 2300 val perplexity 86.1349
280
+ 2300 train 4.397877 (lr=7.1869e-04) (hash(x)=150149692)
281
+ 2310 train 4.409834 (lr=7.1782e-04) (hash(x)=141206538)
282
+ 2320 train 4.440827 (lr=7.1695e-04) (hash(x)=136219115)
283
+ 2330 train 4.471018 (lr=7.1607e-04) (hash(x)=148496500)
284
+ 2340 train 4.452251 (lr=7.1519e-04) (hash(x)=161388511)
285
+ 2350 train 4.308610 (lr=7.1430e-04) (hash(x)=149433185)
286
+ 2360 train 4.524404 (lr=7.1341e-04) (hash(x)=154659203)
287
+ 2370 train 4.507723 (lr=7.1252e-04) (hash(x)=145696996)
288
+ 2380 train 4.316551 (lr=7.1162e-04) (hash(x)=139915095)
289
+ 2390 train 4.451035 (lr=7.1072e-04) (hash(x)=150567584)
290
+ 2400 val loss 4.4325
291
+ 2400 val perplexity 84.1413
292
+ 2400 train 4.344212 (lr=7.0981e-04) (hash(x)=151730720)
293
+ 2410 train 4.390540 (lr=7.0890e-04) (hash(x)=153136544)
294
+ 2420 train 4.356565 (lr=7.0799e-04) (hash(x)=151004677)
295
+ 2430 train 4.371421 (lr=7.0707e-04) (hash(x)=167118859)
296
+ 2440 train 4.393690 (lr=7.0615e-04) (hash(x)=145433453)
297
+ 2450 train 4.427093 (lr=7.0523e-04) (hash(x)=150913293)
298
+ 2460 train 4.382278 (lr=7.0430e-04) (hash(x)=145674639)
299
+ 2470 train 4.493797 (lr=7.0337e-04) (hash(x)=151022749)
300
+ 2480 train 4.335649 (lr=7.0243e-04) (hash(x)=136621037)
301
+ 2490 train 4.397604 (lr=7.0149e-04) (hash(x)=158100672)
302
+ 2500 val loss 4.4037
303
+ 2500 val perplexity 81.7554
304
+ 2500 train 4.417949 (lr=7.0054e-04) (hash(x)=143406752)
305
+ 2510 train 4.332612 (lr=6.9960e-04) (hash(x)=142664084)
306
+ 2520 train 4.386699 (lr=6.9864e-04) (hash(x)=132174378)
307
+ 2530 train 4.404899 (lr=6.9769e-04) (hash(x)=140760293)
308
+ 2540 train 4.346553 (lr=6.9673e-04) (hash(x)=154863398)
309
+ 2550 train 4.286832 (lr=6.9577e-04) (hash(x)=153042049)
310
+ 2560 train 4.222306 (lr=6.9480e-04) (hash(x)=145339170)
311
+ 2570 train 4.372819 (lr=6.9383e-04) (hash(x)=157656357)
312
+ 2580 train 4.361166 (lr=6.9286e-04) (hash(x)=147899462)
313
+ 2590 train 4.333494 (lr=6.9188e-04) (hash(x)=142812892)
314
+ 2600 val loss 4.4198
315
+ 2600 val perplexity 83.0837
316
+ 2600 train 4.293823 (lr=6.9090e-04) (hash(x)=157272496)
317
+ 2610 train 4.073718 (lr=6.8991e-04) (hash(x)=148923594)
318
+ 2620 train 4.005480 (lr=6.8892e-04) (hash(x)=164654807)
319
+ 2630 train 4.199981 (lr=6.8793e-04) (hash(x)=136419327)
320
+ 2640 train 4.168241 (lr=6.8694e-04) (hash(x)=153420840)
321
+ 2650 train 4.006059 (lr=6.8594e-04) (hash(x)=153693827)
322
+ 2660 train 4.165657 (lr=6.8493e-04) (hash(x)=145574165)
323
+ 2670 train 4.198115 (lr=6.8393e-04) (hash(x)=153894843)
324
+ 2680 train 4.053860 (lr=6.8292e-04) (hash(x)=132234446)
325
+ 2690 train 4.134620 (lr=6.8190e-04) (hash(x)=153430212)
326
+ 2700 val loss 4.4027
327
+ 2700 val perplexity 81.6705
328
+ 2700 train 4.304510 (lr=6.8089e-04) (hash(x)=155342327)
329
+ 2710 train 4.311378 (lr=6.7987e-04) (hash(x)=145797023)
330
+ 2720 train 4.469293 (lr=6.7884e-04) (hash(x)=149220870)
331
+ 2730 train 4.368367 (lr=6.7782e-04) (hash(x)=134069562)
332
+ 2740 train 4.382558 (lr=6.7678e-04) (hash(x)=141560499)
333
+ 2750 train 4.386294 (lr=6.7575e-04) (hash(x)=146000524)
334
+ 2760 train 4.281555 (lr=6.7471e-04) (hash(x)=164710665)
335
+ 2770 train 4.174331 (lr=6.7367e-04) (hash(x)=151199094)
336
+ 2780 train 4.369495 (lr=6.7263e-04) (hash(x)=138625404)
337
+ 2790 train 4.632251 (lr=6.7158e-04) (hash(x)=162766112)
338
+ 2800 val loss 4.3701
339
+ 2800 val perplexity 79.0526
340
+ 2800 train 4.335091 (lr=6.7053e-04) (hash(x)=140626679)
341
+ 2810 train 4.412583 (lr=6.6947e-04) (hash(x)=144248544)
342
+ 2820 train 4.338953 (lr=6.6841e-04) (hash(x)=139378481)
343
+ 2830 train 4.267291 (lr=6.6735e-04) (hash(x)=150173403)
344
+ 2840 train 4.558203 (lr=6.6629e-04) (hash(x)=160168705)
345
+ 2850 train 4.348989 (lr=6.6522e-04) (hash(x)=149480059)
346
+ 2860 train 4.390639 (lr=6.6415e-04) (hash(x)=155559096)
347
+ 2870 train 4.328837 (lr=6.6308e-04) (hash(x)=158825683)
348
+ 2880 train 4.496735 (lr=6.6200e-04) (hash(x)=152129048)
349
+ 2890 train 4.172532 (lr=6.6092e-04) (hash(x)=157866544)
350
+ 2900 val loss 4.3486
351
+ 2900 val perplexity 77.3663
352
+ 2900 train 4.300123 (lr=6.5983e-04) (hash(x)=144953350)
353
+ 2910 train 4.504336 (lr=6.5875e-04) (hash(x)=154103497)
354
+ 2920 train 4.219032 (lr=6.5766e-04) (hash(x)=144467854)
355
+ 2930 train 4.151663 (lr=6.5656e-04) (hash(x)=153784048)
356
+ 2940 train 4.458687 (lr=6.5547e-04) (hash(x)=154774534)
357
+ 2950 train 4.301578 (lr=6.5437e-04) (hash(x)=143988507)
358
+ 2960 train 4.182427 (lr=6.5326e-04) (hash(x)=159552857)
359
+ 2970 train 4.482988 (lr=6.5216e-04) (hash(x)=175539144)
360
+ 2980 train 4.358639 (lr=6.5105e-04) (hash(x)=158172686)
361
+ 2990 train 4.236914 (lr=6.4994e-04) (hash(x)=147553039)
362
+ 3000 val loss 4.3545
363
+ 3000 val perplexity 77.8254
364
+ 3000 train 4.161093 (lr=6.4882e-04) (hash(x)=172449837)
365
+ 3010 train 4.056998 (lr=6.4770e-04) (hash(x)=153285123)
366
+ 3020 train 4.233723 (lr=6.4658e-04) (hash(x)=145307252)
367
+ 3030 train 4.284524 (lr=6.4546e-04) (hash(x)=147739766)
368
+ 3040 train 3.998755 (lr=6.4433e-04) (hash(x)=154414400)
369
+ 3050 train 4.090173 (lr=6.4320e-04) (hash(x)=150696487)
370
+ 3060 train 4.095787 (lr=6.4207e-04) (hash(x)=154543441)
371
+ 3070 train 3.874305 (lr=6.4093e-04) (hash(x)=170733242)
372
+ 3080 train 4.225609 (lr=6.3979e-04) (hash(x)=161914669)
373
+ 3090 train 4.106264 (lr=6.3865e-04) (hash(x)=156125491)
374
+ 3100 val loss 4.3722
375
+ 3100 val perplexity 79.2192
376
+ 3100 train 4.007045 (lr=6.3750e-04) (hash(x)=141710086)
377
+ 3110 train 4.308543 (lr=6.3636e-04) (hash(x)=154547635)
378
+ 3120 train 4.153024 (lr=6.3521e-04) (hash(x)=151747171)
379
+ 3130 train 4.439054 (lr=6.3405e-04) (hash(x)=160311353)
380
+ 3140 train 4.550426 (lr=6.3290e-04) (hash(x)=157233821)
381
+ 3150 train 4.355938 (lr=6.3174e-04) (hash(x)=153016810)
382
+ 3160 train 4.337661 (lr=6.3058e-04) (hash(x)=157289875)
383
+ 3170 train 4.341450 (lr=6.2941e-04) (hash(x)=157459484)
384
+ 3180 train 4.259631 (lr=6.2825e-04) (hash(x)=162037435)
385
+ 3190 train 4.353900 (lr=6.2708e-04) (hash(x)=159360169)
386
+ 3200 val loss 4.3205
387
+ 3200 val perplexity 75.2235
388
+ 3200 train 4.235805 (lr=6.2590e-04) (hash(x)=151299772)
389
+ 3210 train 4.471615 (lr=6.2473e-04) (hash(x)=183865364)
390
+ 3220 train 4.208032 (lr=6.2355e-04) (hash(x)=139622592)
391
+ 3230 train 4.163030 (lr=6.2237e-04) (hash(x)=147102655)
392
+ 3240 train 4.226191 (lr=6.2119e-04) (hash(x)=161870696)
393
+ 3250 train 4.375998 (lr=6.2000e-04) (hash(x)=152692231)
394
+ 3260 train 4.112613 (lr=6.1881e-04) (hash(x)=149990134)
395
+ 3270 train 4.096650 (lr=6.1762e-04) (hash(x)=134848432)
396
+ 3280 train 4.154189 (lr=6.1643e-04) (hash(x)=138519418)
397
+ 3290 train 4.290042 (lr=6.1523e-04) (hash(x)=139301249)
398
+ 3300 val loss 4.3214
399
+ 3300 val perplexity 75.2937
400
+ 3300 train 4.130755 (lr=6.1403e-04) (hash(x)=146473110)
401
+ 3310 train 4.186718 (lr=6.1283e-04) (hash(x)=144833685)
402
+ 3320 train 4.102705 (lr=6.1163e-04) (hash(x)=148953545)
403
+ 3330 train 4.329613 (lr=6.1042e-04) (hash(x)=161302371)
404
+ 3340 train 4.258509 (lr=6.0921e-04) (hash(x)=144215411)
405
+ 3350 train 4.422055 (lr=6.0800e-04) (hash(x)=157713664)
406
+ 3360 train 4.305623 (lr=6.0679e-04) (hash(x)=150813928)
407
+ 3370 train 4.284945 (lr=6.0557e-04) (hash(x)=150889463)
408
+ 3380 train 4.328354 (lr=6.0435e-04) (hash(x)=156631928)
409
+ 3390 train 4.370599 (lr=6.0313e-04) (hash(x)=136001579)
410
+ 3400 val loss 4.2952
411
+ 3400 val perplexity 73.3504
412
+ 3400 train 4.439983 (lr=6.0191e-04) (hash(x)=153954157)
413
+ 3410 train 4.223451 (lr=6.0068e-04) (hash(x)=150012477)
414
+ 3420 train 4.576220 (lr=5.9945e-04) (hash(x)=160878899)
415
+ 3430 train 4.614777 (lr=5.9822e-04) (hash(x)=128131106)
416
+ 3440 train 4.539651 (lr=5.9699e-04) (hash(x)=149611866)
417
+ 3450 train 4.171678 (lr=5.9576e-04) (hash(x)=166942460)
418
+ 3460 train 4.327583 (lr=5.9452e-04) (hash(x)=150498414)
419
+ 3470 train 4.475383 (lr=5.9328e-04) (hash(x)=144342279)
420
+ 3480 train 4.299860 (lr=5.9204e-04) (hash(x)=148891764)
421
+ 3490 train 4.290066 (lr=5.9080e-04) (hash(x)=157308188)
422
+ 3500 val loss 4.2854
423
+ 3500 val perplexity 72.6349
424
+ 3500 train 4.174261 (lr=5.8955e-04) (hash(x)=153717336)
425
+ 3510 train 4.202211 (lr=5.8830e-04) (hash(x)=156083924)
426
+ 3520 train 4.191986 (lr=5.8705e-04) (hash(x)=144234966)
427
+ 3530 train 4.275837 (lr=5.8580e-04) (hash(x)=150184381)
428
+ 3540 train 4.169348 (lr=5.8454e-04) (hash(x)=143011198)
429
+ 3550 train 4.310424 (lr=5.8329e-04) (hash(x)=148764613)
430
+ 3560 train 4.064732 (lr=5.8203e-04) (hash(x)=150625329)
431
+ 3570 train 4.187873 (lr=5.8077e-04) (hash(x)=143981302)
432
+ 3580 train 4.232207 (lr=5.7951e-04) (hash(x)=154989703)
433
+ 3590 train 4.364455 (lr=5.7824e-04) (hash(x)=146602526)
434
+ 3600 val loss 4.2804
435
+ 3600 val perplexity 72.2698
436
+ 3600 train 4.060957 (lr=5.7697e-04) (hash(x)=144965161)
437
+ 3610 train 4.059373 (lr=5.7571e-04) (hash(x)=159452243)
438
+ 3620 train 4.046498 (lr=5.7443e-04) (hash(x)=144904300)
439
+ 3630 train 3.961258 (lr=5.7316e-04) (hash(x)=127790212)
440
+ 3640 train 4.200054 (lr=5.7189e-04) (hash(x)=150863158)
441
+ 3650 train 4.151321 (lr=5.7061e-04) (hash(x)=145691635)
442
+ 3660 train 4.139456 (lr=5.6933e-04) (hash(x)=150631264)
443
+ 3670 train 4.116817 (lr=5.6805e-04) (hash(x)=148646259)
444
+ 3680 train 4.085281 (lr=5.6677e-04) (hash(x)=139317386)
445
+ 3690 train 4.190476 (lr=5.6549e-04) (hash(x)=142206916)
446
+ 3700 val loss 4.2912
447
+ 3700 val perplexity 73.0525
448
+ 3700 train 4.192861 (lr=5.6420e-04) (hash(x)=125969741)
449
+ 3710 train 4.042929 (lr=5.6291e-04) (hash(x)=157183915)
450
+ 3720 train 4.033793 (lr=5.6162e-04) (hash(x)=150434369)
451
+ 3730 train 4.231685 (lr=5.6033e-04) (hash(x)=160911722)
452
+ 3740 train 4.128427 (lr=5.5904e-04) (hash(x)=163306921)
453
+ 3750 train 4.236143 (lr=5.5774e-04) (hash(x)=149156876)
454
+ 3760 train 4.373276 (lr=5.5645e-04) (hash(x)=156015970)
455
+ 3770 train 4.318218 (lr=5.5515e-04) (hash(x)=148222322)
456
+ 3780 train 4.429107 (lr=5.5385e-04) (hash(x)=172040174)
457
+ 3790 train 4.329562 (lr=5.5255e-04) (hash(x)=151593070)
458
+ 3800 val loss 4.2646
459
+ 3800 val perplexity 71.1370
460
+ 3800 train 4.081344 (lr=5.5125e-04) (hash(x)=155070487)
461
+ 3810 train 4.275709 (lr=5.4994e-04) (hash(x)=137073164)
462
+ 3820 train 4.176977 (lr=5.4864e-04) (hash(x)=147761510)
463
+ 3830 train 4.100427 (lr=5.4733e-04) (hash(x)=154763021)
464
+ 3840 train 4.144835 (lr=5.4602e-04) (hash(x)=149622435)
465
+ 3850 train 4.256602 (lr=5.4471e-04) (hash(x)=150779423)
466
+ 3860 train 4.221241 (lr=5.4340e-04) (hash(x)=139445120)
467
+ 3870 train 4.205736 (lr=5.4208e-04) (hash(x)=151884947)
468
+ 3880 train 4.211124 (lr=5.4077e-04) (hash(x)=139795201)
469
+ 3890 train 4.249043 (lr=5.3945e-04) (hash(x)=155589956)
470
+ 3900 val loss 4.2527
471
+ 3900 val perplexity 70.2927
472
+ 3900 train 4.170682 (lr=5.3813e-04) (hash(x)=149444644)
473
+ 3910 train 4.171191 (lr=5.3681e-04) (hash(x)=148538323)
474
+ 3920 train 4.292335 (lr=5.3549e-04) (hash(x)=144026925)
475
+ 3930 train 4.178129 (lr=5.3417e-04) (hash(x)=155311200)
476
+ 3940 train 4.154992 (lr=5.3284e-04) (hash(x)=142410034)
477
+ 3950 train 4.139569 (lr=5.3152e-04) (hash(x)=145986075)
478
+ 3960 train 4.033472 (lr=5.3019e-04) (hash(x)=156197753)
479
+ 3970 train 4.187565 (lr=5.2886e-04) (hash(x)=146981564)
480
+ 3980 train 4.495646 (lr=5.2754e-04) (hash(x)=145361976)
481
+ 3990 train 4.226357 (lr=5.2620e-04) (hash(x)=149344475)
482
+ 4000 val loss 4.2466
483
+ 4000 val perplexity 69.8657
484
+ 4000 train 4.110120 (lr=5.2487e-04) (hash(x)=151663033)
485
+ 4010 train 3.992573 (lr=5.2354e-04) (hash(x)=140915499)
486
+ 4020 train 4.202984 (lr=5.2221e-04) (hash(x)=159979134)
487
+ 4030 train 4.130474 (lr=5.2087e-04) (hash(x)=163081433)
488
+ 4040 train 4.013659 (lr=5.1953e-04) (hash(x)=144890231)
489
+ 4050 train 4.180538 (lr=5.1820e-04) (hash(x)=138242368)
490
+ 4060 train 4.077572 (lr=5.1686e-04) (hash(x)=144438406)
491
+ 4070 train 3.966937 (lr=5.1552e-04) (hash(x)=159294906)
492
+ 4080 train 4.019980 (lr=5.1418e-04) (hash(x)=143676301)
493
+ 4090 train 4.249904 (lr=5.1284e-04) (hash(x)=147875296)
494
+ 4100 val loss 4.2614
495
+ 4100 val perplexity 70.9083
496
+ 4100 train 4.311615 (lr=5.1149e-04) (hash(x)=143688282)
497
+ 4110 train 4.084391 (lr=5.1015e-04) (hash(x)=150086777)
498
+ 4120 train 4.033484 (lr=5.0880e-04) (hash(x)=154624391)
499
+ 4130 train 4.216310 (lr=5.0746e-04) (hash(x)=141163032)
500
+ 4140 train 4.243912 (lr=5.0611e-04) (hash(x)=151038180)
501
+ 4150 train 4.321226 (lr=5.0476e-04) (hash(x)=144718421)
502
+ 4160 train 4.238781 (lr=5.0341e-04) (hash(x)=154277459)
503
+ 4170 train 4.391020 (lr=5.0206e-04) (hash(x)=143765173)
504
+ 4180 train 4.263358 (lr=5.0071e-04) (hash(x)=161013425)
505
+ 4190 train 4.217359 (lr=4.9936e-04) (hash(x)=161666854)
506
+ 4200 val loss 4.2251
507
+ 4200 val perplexity 68.3831
508
+ 4200 train 4.135858 (lr=4.9801e-04) (hash(x)=163361651)
509
+ 4210 train 4.478011 (lr=4.9665e-04) (hash(x)=153491528)
510
+ 4220 train 4.267887 (lr=4.9530e-04) (hash(x)=164410385)
511
+ 4230 train 4.211841 (lr=4.9395e-04) (hash(x)=175191816)
512
+ 4240 train 4.081883 (lr=4.9259e-04) (hash(x)=133924818)
513
+ 4250 train 4.310592 (lr=4.9123e-04) (hash(x)=152932845)
514
+ 4260 train 4.179938 (lr=4.8988e-04) (hash(x)=141987318)
515
+ 4270 train 4.276831 (lr=4.8852e-04) (hash(x)=142521279)
516
+ 4280 train 4.137108 (lr=4.8716e-04) (hash(x)=147707669)
517
+ 4290 train 4.200391 (lr=4.8580e-04) (hash(x)=152245539)
518
+ 4300 val loss 4.2116
519
+ 4300 val perplexity 67.4610
520
+ 4300 train 4.087514 (lr=4.8444e-04) (hash(x)=153619361)
521
+ 4310 train 4.300503 (lr=4.8308e-04) (hash(x)=142614473)
522
+ 4320 train 4.128212 (lr=4.8172e-04) (hash(x)=142689858)
523
+ 4330 train 4.150764 (lr=4.8036e-04) (hash(x)=140837891)
524
+ 4340 train 4.131648 (lr=4.7899e-04) (hash(x)=131790760)
525
+ 4350 train 4.302588 (lr=4.7763e-04) (hash(x)=147186292)
526
+ 4360 train 4.107545 (lr=4.7627e-04) (hash(x)=148140562)
527
+ 4370 train 4.218799 (lr=4.7490e-04) (hash(x)=156409913)
528
+ 4380 train 4.132556 (lr=4.7354e-04) (hash(x)=145126034)
529
+ 4390 train 4.082724 (lr=4.7217e-04) (hash(x)=147368423)
530
+ 4400 val loss 4.2156
531
+ 4400 val perplexity 67.7378
532
+ 4400 train 4.552946 (lr=4.7081e-04) (hash(x)=168527064)
533
+ 4410 train 4.074948 (lr=4.6944e-04) (hash(x)=150152773)
534
+ 4420 train 4.158820 (lr=4.6807e-04) (hash(x)=155472304)
535
+ 4430 train 4.090458 (lr=4.6671e-04) (hash(x)=147299265)
536
+ 4440 train 4.053125 (lr=4.6534e-04) (hash(x)=140659522)
537
+ 4450 train 4.109721 (lr=4.6397e-04) (hash(x)=147169111)
538
+ 4460 train 4.176729 (lr=4.6260e-04) (hash(x)=144098557)
539
+ 4470 train 3.942524 (lr=4.6124e-04) (hash(x)=161957507)
540
+ 4480 train 3.966285 (lr=4.5987e-04) (hash(x)=154064204)
541
+ 4490 train 4.097857 (lr=4.5850e-04) (hash(x)=153999958)
542
+ 4500 val loss 4.2278
543
+ 4500 val perplexity 68.5674
544
+ 4500 train 4.257614 (lr=4.5713e-04) (hash(x)=125588037)
545
+ 4510 train 4.060706 (lr=4.5576e-04) (hash(x)=142522186)
546
+ 4520 train 3.902825 (lr=4.5439e-04) (hash(x)=152736893)
547
+ 4530 train 4.134715 (lr=4.5302e-04) (hash(x)=148091275)
548
+ 4540 train 4.245779 (lr=4.5165e-04) (hash(x)=147663019)
549
+ 4550 train 4.329647 (lr=4.5028e-04) (hash(x)=151491857)
550
+ 4560 train 4.213305 (lr=4.4891e-04) (hash(x)=161757513)
551
+ 4570 train 4.208701 (lr=4.4754e-04) (hash(x)=165448260)
552
+ 4580 train 4.350125 (lr=4.4617e-04) (hash(x)=151620292)
553
+ 4590 train 4.425686 (lr=4.4480e-04) (hash(x)=161877852)
554
+ 4600 val loss 4.1912
555
+ 4600 val perplexity 66.0995
556
+ 4600 train 4.169526 (lr=4.4343e-04) (hash(x)=143710941)
557
+ 4610 train 4.260080 (lr=4.4206e-04) (hash(x)=152244871)
558
+ 4620 train 4.217620 (lr=4.4069e-04) (hash(x)=161268167)
559
+ 4630 train 4.209453 (lr=4.3931e-04) (hash(x)=153904677)
560
+ 4640 train 4.157935 (lr=4.3794e-04) (hash(x)=152593758)
561
+ 4650 train 4.314240 (lr=4.3657e-04) (hash(x)=159047122)
562
+ 4660 train 4.381655 (lr=4.3520e-04) (hash(x)=167370244)
563
+ 4670 train 4.151340 (lr=4.3383e-04) (hash(x)=140357684)
564
+ 4680 train 4.316838 (lr=4.3246e-04) (hash(x)=140547853)
565
+ 4690 train 4.486212 (lr=4.3109e-04) (hash(x)=143000949)
566
+ 4700 val loss 4.1886
567
+ 4700 val perplexity 65.9293
568
+ 4700 train 4.128787 (lr=4.2972e-04) (hash(x)=150952742)
569
+ 4710 train 4.172995 (lr=4.2835e-04) (hash(x)=157964406)
570
+ 4720 train 4.068425 (lr=4.2698e-04) (hash(x)=147653608)
571
+ 4730 train 4.142352 (lr=4.2561e-04) (hash(x)=165545764)
572
+ 4740 train 4.119741 (lr=4.2424e-04) (hash(x)=146744631)
573
+ 4750 train 4.202999 (lr=4.2287e-04) (hash(x)=153635164)
574
+ 4760 train 4.247137 (lr=4.2150e-04) (hash(x)=153217508)
575
+ 4770 train 4.194675 (lr=4.2013e-04) (hash(x)=169622743)
576
+ 4780 train 4.125592 (lr=4.1876e-04) (hash(x)=138018297)
577
+ 4790 train 3.999892 (lr=4.1740e-04) (hash(x)=156067303)
578
+ 4800 val loss 4.1862
579
+ 4800 val perplexity 65.7735
580
+ 4800 train 4.170980 (lr=4.1603e-04) (hash(x)=145323659)
581
+ 4810 train 4.058845 (lr=4.1466e-04) (hash(x)=152587772)
582
+ 4820 train 4.279411 (lr=4.1329e-04) (hash(x)=148481485)
583
+ 4830 train 4.045092 (lr=4.1193e-04) (hash(x)=154219664)
584
+ 4840 train 3.930212 (lr=4.1056e-04) (hash(x)=156853645)
585
+ 4850 train 4.017058 (lr=4.0919e-04) (hash(x)=152329702)
586
+ 4860 train 4.009953 (lr=4.0783e-04) (hash(x)=136684779)
587
+ 4870 train 4.072377 (lr=4.0646e-04) (hash(x)=143260367)
588
+ 4880 train 4.043165 (lr=4.0510e-04) (hash(x)=150910947)
589
+ 4890 train 4.058242 (lr=4.0373e-04) (hash(x)=179249925)
590
+ 4900 val loss 4.1905
591
+ 4900 val perplexity 66.0537
592
+ 4900 train 4.214311 (lr=4.0237e-04) (hash(x)=153151397)
593
+ 4910 train 4.166713 (lr=4.0101e-04) (hash(x)=144101891)
594
+ 4920 train 4.061687 (lr=3.9964e-04) (hash(x)=144994952)
595
+ 4930 train 4.355833 (lr=3.9828e-04) (hash(x)=155390878)
596
+ 4940 train 4.123945 (lr=3.9692e-04) (hash(x)=141359221)
597
+ 4950 train 4.030979 (lr=3.9556e-04) (hash(x)=151570011)
598
+ 4960 train 4.249825 (lr=3.9420e-04) (hash(x)=146918855)
599
+ 4970 train 4.182886 (lr=3.9284e-04) (hash(x)=148647362)
600
+ 4980 train 4.136347 (lr=3.9148e-04) (hash(x)=159753562)
601
+ 4990 train 4.092875 (lr=3.9012e-04) (hash(x)=147683447)
602
+ 5000 val loss 4.1807
603
+ 5000 val perplexity 65.4145
604
+ 5000 train 4.227891 (lr=3.8877e-04) (hash(x)=143182059)
605
+ 5010 train 4.034374 (lr=3.8741e-04) (hash(x)=138657407)
606
+ 5020 train 4.141760 (lr=3.8605e-04) (hash(x)=151830497)
607
+ 5030 train 4.141773 (lr=3.8470e-04) (hash(x)=147727694)
608
+ 5040 train 3.984780 (lr=3.8335e-04) (hash(x)=151957014)
609
+ 5050 train 4.250139 (lr=3.8199e-04) (hash(x)=147828693)
610
+ 5060 train 4.307155 (lr=3.8064e-04) (hash(x)=158589700)
611
+ 5070 train 4.185586 (lr=3.7929e-04) (hash(x)=155662163)
612
+ 5080 train 4.292055 (lr=3.7794e-04) (hash(x)=157797383)
613
+ 5090 train 4.162642 (lr=3.7659e-04) (hash(x)=153848957)
614
+ 5100 val loss 4.1689
615
+ 5100 val perplexity 64.6423
616
+ 5100 train 4.120473 (lr=3.7524e-04) (hash(x)=170083586)
617
+ 5110 train 4.043426 (lr=3.7389e-04) (hash(x)=153800138)
618
+ 5120 train 4.145282 (lr=3.7254e-04) (hash(x)=153526228)
619
+ 5130 train 4.160254 (lr=3.7120e-04) (hash(x)=158650342)
620
+ 5140 train 4.144539 (lr=3.6985e-04) (hash(x)=153163034)
621
+ 5150 train 4.140269 (lr=3.6851e-04) (hash(x)=155281294)
622
+ 5160 train 4.114711 (lr=3.6716e-04) (hash(x)=147731627)
623
+ 5170 train 4.100759 (lr=3.6582e-04) (hash(x)=146388976)
624
+ 5180 train 4.319185 (lr=3.6448e-04) (hash(x)=140323065)
625
+ 5190 train 4.222426 (lr=3.6314e-04) (hash(x)=150406523)
626
+ 5200 val loss 4.1610
627
+ 5200 val perplexity 64.1329
628
+ 5200 train 4.018673 (lr=3.6180e-04) (hash(x)=149363919)
629
+ 5210 train 4.169579 (lr=3.6047e-04) (hash(x)=164390214)
630
+ 5220 train 4.122836 (lr=3.5913e-04) (hash(x)=153652984)
631
+ 5230 train 4.086530 (lr=3.5779e-04) (hash(x)=139542770)
632
+ 5240 train 3.977182 (lr=3.5646e-04) (hash(x)=140115288)
633
+ 5250 train 4.101837 (lr=3.5513e-04) (hash(x)=142158900)
634
+ 5260 train 4.026901 (lr=3.5380e-04) (hash(x)=156019449)
635
+ 5270 train 4.139410 (lr=3.5246e-04) (hash(x)=150776850)
636
+ 5280 train 4.096313 (lr=3.5114e-04) (hash(x)=151358692)
637
+ 5290 train 4.065066 (lr=3.4981e-04) (hash(x)=137112863)
638
+ 5300 val loss 4.1609
639
+ 5300 val perplexity 64.1288
640
+ 5300 train 3.990110 (lr=3.4848e-04) (hash(x)=152033784)
641
+ 5310 train 4.034427 (lr=3.4716e-04) (hash(x)=147785712)
642
+ 5320 train 4.171314 (lr=3.4583e-04) (hash(x)=156426339)
643
+ 5330 train 4.126149 (lr=3.4451e-04) (hash(x)=161907693)
644
+ 5340 train 4.185738 (lr=3.4319e-04) (hash(x)=156414094)
645
+ 5350 train 4.214840 (lr=3.4187e-04) (hash(x)=141712428)
646
+ 5360 train 4.163688 (lr=3.4055e-04) (hash(x)=156873601)
647
+ 5370 train 4.149635 (lr=3.3923e-04) (hash(x)=149863781)
648
+ 5380 train 4.142146 (lr=3.3792e-04) (hash(x)=148183971)
649
+ 5390 train 4.105454 (lr=3.3660e-04) (hash(x)=151542334)
650
+ 5400 val loss 4.1388
651
+ 5400 val perplexity 62.7292
652
+ 5400 train 4.382346 (lr=3.3529e-04) (hash(x)=154614289)
653
+ 5410 train 4.123094 (lr=3.3398e-04) (hash(x)=169124671)
654
+ 5420 train 4.151130 (lr=3.3267e-04) (hash(x)=159963738)
655
+ 5430 train 4.112948 (lr=3.3136e-04) (hash(x)=155162691)
656
+ 5440 train 4.138331 (lr=3.3006e-04) (hash(x)=135981908)
657
+ 5450 train 4.227556 (lr=3.2875e-04) (hash(x)=145055580)
658
+ 5460 train 4.089634 (lr=3.2745e-04) (hash(x)=150367825)
659
+ 5470 train 4.172388 (lr=3.2615e-04) (hash(x)=151034173)
660
+ 5480 train 4.095665 (lr=3.2485e-04) (hash(x)=155581766)
661
+ 5490 train 4.116316 (lr=3.2355e-04) (hash(x)=136973993)
662
+ 5500 val loss 4.1331
663
+ 5500 val perplexity 62.3713
664
+ 5500 train 4.294282 (lr=3.2226e-04) (hash(x)=157745174)
665
+ 5510 train 4.240862 (lr=3.2096e-04) (hash(x)=144628894)
666
+ 5520 train 4.124760 (lr=3.1967e-04) (hash(x)=135056684)
667
+ 5530 train 4.113754 (lr=3.1838e-04) (hash(x)=143418640)
668
+ 5540 train 4.088257 (lr=3.1709e-04) (hash(x)=144464984)
669
+ 5550 train 4.299384 (lr=3.1580e-04) (hash(x)=155888745)
670
+ 5560 train 3.956552 (lr=3.1451e-04) (hash(x)=149122683)
671
+ 5570 train 4.002886 (lr=3.1323e-04) (hash(x)=154573954)
672
+ 5580 train 4.107324 (lr=3.1195e-04) (hash(x)=149339361)
673
+ 5590 train 4.002980 (lr=3.1067e-04) (hash(x)=154738453)
674
+ 5600 val loss 4.1351
675
+ 5600 val perplexity 62.4984
676
+ 5600 train 4.114351 (lr=3.0939e-04) (hash(x)=147693222)
677
+ 5610 train 3.888367 (lr=3.0811e-04) (hash(x)=142140192)
678
+ 5620 train 4.025737 (lr=3.0684e-04) (hash(x)=148677209)
679
+ 5630 train 4.086386 (lr=3.0557e-04) (hash(x)=143495554)
680
+ 5640 train 4.172348 (lr=3.0429e-04) (hash(x)=153202964)
681
+ 5650 train 4.099509 (lr=3.0303e-04) (hash(x)=168724867)
682
+ 5660 train 4.068596 (lr=3.0176e-04) (hash(x)=142816746)
683
+ 5670 train 4.197282 (lr=3.0049e-04) (hash(x)=164119551)
684
+ 5680 train 4.061250 (lr=2.9923e-04) (hash(x)=157239121)
685
+ 5690 train 4.338422 (lr=2.9797e-04) (hash(x)=144662329)
686
+ 5700 val loss 4.1232
687
+ 5700 val perplexity 61.7578
688
+ 5700 train 4.055151 (lr=2.9671e-04) (hash(x)=149784627)
689
+ 5710 train 4.026082 (lr=2.9546e-04) (hash(x)=152821162)
690
+ 5720 train 4.247062 (lr=2.9420e-04) (hash(x)=158497530)
691
+ 5730 train 4.233568 (lr=2.9295e-04) (hash(x)=154583914)
692
+ 5740 train 4.052171 (lr=2.9170e-04) (hash(x)=142027172)
693
+ 5750 train 4.282454 (lr=2.9045e-04) (hash(x)=119227402)
694
+ 5760 train 4.138756 (lr=2.8920e-04) (hash(x)=153248187)
695
+ 5770 train 4.012252 (lr=2.8796e-04) (hash(x)=147951994)
696
+ 5780 train 4.024483 (lr=2.8672e-04) (hash(x)=138935284)
697
+ 5790 train 4.111892 (lr=2.8548e-04) (hash(x)=151786681)
698
+ 5800 val loss 4.1281
699
+ 5800 val perplexity 62.0595
700
+ 5800 train 4.143296 (lr=2.8424e-04) (hash(x)=158620729)
701
+ 5810 train 4.095777 (lr=2.8301e-04) (hash(x)=178115789)
702
+ 5820 train 4.013129 (lr=2.8178e-04) (hash(x)=149092790)
703
+ 5830 train 4.194000 (lr=2.8055e-04) (hash(x)=148269833)
704
+ 5840 train 4.129766 (lr=2.7932e-04) (hash(x)=163714738)
705
+ 5850 train 4.107069 (lr=2.7809e-04) (hash(x)=144111890)
706
+ 5860 train 4.072674 (lr=2.7687e-04) (hash(x)=170702475)
707
+ 5870 train 4.015656 (lr=2.7565e-04) (hash(x)=155970835)
708
+ 5880 train 4.016105 (lr=2.7443e-04) (hash(x)=146695463)
709
+ 5890 train 4.078478 (lr=2.7321e-04) (hash(x)=151603187)
710
+ 5900 val loss 4.1171
711
+ 5900 val perplexity 61.3832
712
+ 5900 train 4.094788 (lr=2.7200e-04) (hash(x)=159763910)
713
+ 5910 train 4.133310 (lr=2.7079e-04) (hash(x)=149555230)
714
+ 5920 train 4.145046 (lr=2.6958e-04) (hash(x)=138048333)
715
+ 5930 train 4.035842 (lr=2.6837e-04) (hash(x)=148668256)
716
+ 5940 train 4.181791 (lr=2.6717e-04) (hash(x)=148352543)
717
+ 5950 train 4.191329 (lr=2.6597e-04) (hash(x)=163457488)
718
+ 5960 train 4.260097 (lr=2.6477e-04) (hash(x)=155306804)
719
+ 5970 train 4.123145 (lr=2.6357e-04) (hash(x)=146576143)
720
+ 5980 train 4.121843 (lr=2.6238e-04) (hash(x)=144668956)
721
+ 5990 train 4.095946 (lr=2.6119e-04) (hash(x)=156098617)
722
+ 6000 val loss 4.1045
723
+ 6000 val perplexity 60.6102
724
+ 6000 train 4.071487 (lr=2.6000e-04) (hash(x)=147640561)
725
+ 6010 train 4.144564 (lr=2.5881e-04) (hash(x)=161223074)
726
+ 6020 train 4.124915 (lr=2.5763e-04) (hash(x)=156308454)
727
+ 6030 train 4.098089 (lr=2.5645e-04) (hash(x)=142553975)
728
+ 6040 train 4.066286 (lr=2.5527e-04) (hash(x)=142039829)
729
+ 6050 train 4.137825 (lr=2.5410e-04) (hash(x)=156329296)
730
+ 6060 train 4.076353 (lr=2.5292e-04) (hash(x)=148505537)
731
+ 6070 train 3.961073 (lr=2.5175e-04) (hash(x)=151801272)
732
+ 6080 train 4.049527 (lr=2.5059e-04) (hash(x)=159908287)
733
+ 6090 train 4.036864 (lr=2.4942e-04) (hash(x)=148714721)
734
+ 6100 val loss 4.0978
735
+ 6100 val perplexity 60.2088
736
+ 6100 train 4.107800 (lr=2.4826e-04) (hash(x)=156613394)
737
+ 6110 train 3.967981 (lr=2.4710e-04) (hash(x)=143286705)
738
+ 6120 train 4.170110 (lr=2.4595e-04) (hash(x)=150191141)
739
+ 6130 train 4.048105 (lr=2.4479e-04) (hash(x)=149855624)
740
+ 6140 train 4.040743 (lr=2.4364e-04) (hash(x)=157312372)
741
+ 6150 train 4.161436 (lr=2.4250e-04) (hash(x)=148262730)
742
+ 6160 train 4.071573 (lr=2.4135e-04) (hash(x)=147032785)
743
+ 6170 train 4.000295 (lr=2.4021e-04) (hash(x)=161213064)
744
+ 6180 train 3.873945 (lr=2.3907e-04) (hash(x)=144538373)
745
+ 6190 train 3.925917 (lr=2.3793e-04) (hash(x)=144688493)
746
+ 6200 val loss 4.0964
747
+ 6200 val perplexity 60.1215
748
+ 6200 train 4.000531 (lr=2.3680e-04) (hash(x)=186221290)
749
+ 6210 train 4.009966 (lr=2.3567e-04) (hash(x)=139529352)
750
+ 6220 train 4.174840 (lr=2.3454e-04) (hash(x)=138081469)
751
+ 6230 train 4.051074 (lr=2.3342e-04) (hash(x)=152979409)
752
+ 6240 train 4.023096 (lr=2.3230e-04) (hash(x)=157848424)
753
+ 6250 train 3.919580 (lr=2.3118e-04) (hash(x)=150770915)
754
+ 6260 train 4.014743 (lr=2.3006e-04) (hash(x)=153769379)
755
+ 6270 train 3.952215 (lr=2.2895e-04) (hash(x)=153318361)
756
+ 6280 train 3.943378 (lr=2.2784e-04) (hash(x)=143366327)
757
+ 6290 train 4.112032 (lr=2.2674e-04) (hash(x)=160684845)
758
+ 6300 val loss 4.0823
759
+ 6300 val perplexity 59.2787
760
+ 6300 train 3.995546 (lr=2.2563e-04) (hash(x)=152081419)
761
+ 6310 train 3.953910 (lr=2.2453e-04) (hash(x)=146626797)
762
+ 6320 train 4.138268 (lr=2.2344e-04) (hash(x)=155291923)
763
+ 6330 train 4.011633 (lr=2.2234e-04) (hash(x)=140920951)
764
+ 6340 train 4.090152 (lr=2.2125e-04) (hash(x)=146642359)
765
+ 6350 train 4.061490 (lr=2.2017e-04) (hash(x)=160486550)
766
+ 6360 train 4.018549 (lr=2.1908e-04) (hash(x)=151833082)
767
+ 6370 train 4.013324 (lr=2.1800e-04) (hash(x)=124740362)
768
+ 6380 train 4.165404 (lr=2.1692e-04) (hash(x)=153952434)
769
+ 6390 train 4.207119 (lr=2.1585e-04) (hash(x)=145864314)
770
+ 6400 val loss 4.0850
771
+ 6400 val perplexity 59.4414
772
+ 6400 train 4.083162 (lr=2.1478e-04) (hash(x)=154808349)
773
+ 6410 train 4.060822 (lr=2.1371e-04) (hash(x)=155037739)
774
+ 6420 train 3.995215 (lr=2.1265e-04) (hash(x)=151155588)
775
+ 6430 train 3.886220 (lr=2.1159e-04) (hash(x)=146548556)
776
+ 6440 train 4.055191 (lr=2.1053e-04) (hash(x)=152706474)
777
+ 6450 train 4.048583 (lr=2.0947e-04) (hash(x)=154207794)
778
+ 6460 train 3.917463 (lr=2.0842e-04) (hash(x)=141590515)
779
+ 6470 train 3.960523 (lr=2.0737e-04) (hash(x)=157673109)
780
+ 6480 train 3.967658 (lr=2.0633e-04) (hash(x)=155842980)
781
+ 6490 train 4.048142 (lr=2.0529e-04) (hash(x)=143595736)
782
+ 6500 val loss 4.0836
783
+ 6500 val perplexity 59.3594
784
+ 6500 train 4.002976 (lr=2.0425e-04) (hash(x)=159437208)
785
+ 6510 train 4.020519 (lr=2.0322e-04) (hash(x)=142631317)
786
+ 6520 train 3.741516 (lr=2.0218e-04) (hash(x)=139009211)
787
+ 6530 train 4.087826 (lr=2.0116e-04) (hash(x)=145451137)
788
+ 6540 train 4.042386 (lr=2.0013e-04) (hash(x)=163532779)
789
+ 6550 train 3.953823 (lr=1.9911e-04) (hash(x)=158084674)
790
+ 6560 train 3.931937 (lr=1.9810e-04) (hash(x)=142422592)
791
+ 6570 train 3.843050 (lr=1.9708e-04) (hash(x)=141089289)
792
+ 6580 train 3.950442 (lr=1.9607e-04) (hash(x)=139712979)
793
+ 6590 train 3.861105 (lr=1.9507e-04) (hash(x)=145833805)
794
+ 6600 val loss 4.0824
795
+ 6600 val perplexity 59.2886
796
+ 6600 train 3.931682 (lr=1.9406e-04) (hash(x)=157933074)
797
+ 6610 train 3.852073 (lr=1.9306e-04) (hash(x)=142534769)
798
+ 6620 train 3.906833 (lr=1.9207e-04) (hash(x)=138456044)
799
+ 6630 train 4.040647 (lr=1.9108e-04) (hash(x)=145111052)
800
+ 6640 train 3.997754 (lr=1.9009e-04) (hash(x)=166954515)
801
+ 6650 train 3.959227 (lr=1.8910e-04) (hash(x)=146889543)
802
+ 6660 train 3.998843 (lr=1.8812e-04) (hash(x)=143253726)
803
+ 6670 train 3.908363 (lr=1.8714e-04) (hash(x)=144349844)
804
+ 6680 train 4.096266 (lr=1.8617e-04) (hash(x)=139911928)
805
+ 6690 train 4.284996 (lr=1.8520e-04) (hash(x)=143135862)
806
+ 6700 val loss 4.0709
807
+ 6700 val perplexity 58.6093
808
+ 6700 train 4.003819 (lr=1.8423e-04) (hash(x)=161560240)
809
+ 6710 train 4.095499 (lr=1.8327e-04) (hash(x)=150437730)
810
+ 6720 train 4.195431 (lr=1.8231e-04) (hash(x)=153035304)
811
+ 6730 train 3.979951 (lr=1.8136e-04) (hash(x)=139323796)
812
+ 6740 train 3.988512 (lr=1.8040e-04) (hash(x)=167437923)
813
+ 6750 train 3.956909 (lr=1.7946e-04) (hash(x)=143814596)
814
+ 6760 train 4.022171 (lr=1.7851e-04) (hash(x)=143755031)
815
+ 6770 train 3.975881 (lr=1.7757e-04) (hash(x)=142014038)
816
+ 6780 train 4.027085 (lr=1.7663e-04) (hash(x)=148465062)
817
+ 6790 train 4.066669 (lr=1.7570e-04) (hash(x)=154656877)
818
+ 6800 val loss 4.0691
819
+ 6800 val perplexity 58.5059
820
+ 6800 train 4.171014 (lr=1.7477e-04) (hash(x)=155424292)
821
+ 6810 train 3.809006 (lr=1.7385e-04) (hash(x)=147358081)
822
+ 6820 train 3.924278 (lr=1.7293e-04) (hash(x)=145387291)
823
+ 6830 train 3.879535 (lr=1.7201e-04) (hash(x)=146334068)
824
+ 6840 train 3.950760 (lr=1.7110e-04) (hash(x)=139563766)
825
+ 6850 train 3.975258 (lr=1.7019e-04) (hash(x)=139095691)
826
+ 6860 train 3.919004 (lr=1.6928e-04) (hash(x)=142093999)
827
+ 6870 train 4.033529 (lr=1.6838e-04) (hash(x)=148598823)
828
+ 6880 train 4.117742 (lr=1.6748e-04) (hash(x)=146486085)
829
+ 6890 train 3.901623 (lr=1.6659e-04) (hash(x)=165683212)
830
+ 6900 val loss 4.0729
831
+ 6900 val perplexity 58.7262
832
+ 6900 train 4.111107 (lr=1.6570e-04) (hash(x)=148561470)
833
+ 6910 train 3.797852 (lr=1.6481e-04) (hash(x)=136352432)
834
+ 6920 train 4.176081 (lr=1.6393e-04) (hash(x)=147640094)
835
+ 6930 train 4.057874 (lr=1.6305e-04) (hash(x)=152343827)
836
+ 6940 train 4.040170 (lr=1.6218e-04) (hash(x)=158787971)
837
+ 6950 train 4.321411 (lr=1.6131e-04) (hash(x)=178133125)
838
+ 6960 train 4.102084 (lr=1.6044e-04) (hash(x)=167302853)
839
+ 6970 train 4.037830 (lr=1.5958e-04) (hash(x)=143729435)
840
+ 6980 train 4.143796 (lr=1.5872e-04) (hash(x)=135857879)
841
+ 6990 train 4.084823 (lr=1.5787e-04) (hash(x)=155384510)
842
+ 7000 val loss 4.0603
843
+ 7000 val perplexity 57.9915
844
+ 7000 train 4.029609 (lr=1.5702e-04) (hash(x)=141527450)
845
+ 7010 train 4.072048 (lr=1.5618e-04) (hash(x)=148648498)
846
+ 7020 train 3.996335 (lr=1.5533e-04) (hash(x)=148073842)
847
+ 7030 train 4.131407 (lr=1.5450e-04) (hash(x)=160771614)
848
+ 7040 train 3.907187 (lr=1.5366e-04) (hash(x)=148361330)
849
+ 7050 train 3.884920 (lr=1.5284e-04) (hash(x)=151212224)
850
+ 7060 train 4.080006 (lr=1.5201e-04) (hash(x)=155925366)
851
+ 7070 train 3.931046 (lr=1.5119e-04) (hash(x)=149830576)
852
+ 7080 train 3.899948 (lr=1.5037e-04) (hash(x)=144849934)
853
+ 7090 train 3.992213 (lr=1.4956e-04) (hash(x)=150997807)
854
+ 7100 val loss 4.0555
855
+ 7100 val perplexity 57.7145
856
+ 7100 train 4.034127 (lr=1.4875e-04) (hash(x)=151066339)
857
+ 7110 train 3.933766 (lr=1.4795e-04) (hash(x)=147185781)
858
+ 7120 train 3.975846 (lr=1.4715e-04) (hash(x)=141765575)
859
+ 7130 train 3.991612 (lr=1.4636e-04) (hash(x)=144204513)
860
+ 7140 train 3.827074 (lr=1.4556e-04) (hash(x)=149326253)
861
+ 7150 train 3.988863 (lr=1.4478e-04) (hash(x)=150409811)
862
+ 7160 train 3.875143 (lr=1.4400e-04) (hash(x)=144790348)
863
+ 7170 train 3.997903 (lr=1.4322e-04) (hash(x)=150500939)
864
+ 7180 train 3.819256 (lr=1.4244e-04) (hash(x)=143552103)
865
+ 7190 train 3.866680 (lr=1.4167e-04) (hash(x)=157471075)
866
+ 7200 val loss 4.0603
867
+ 7200 val perplexity 57.9906
868
+ 7200 train 3.845976 (lr=1.4091e-04) (hash(x)=155231264)
869
+ 7210 train 3.930204 (lr=1.4015e-04) (hash(x)=150350401)
870
+ 7220 train 3.955145 (lr=1.3939e-04) (hash(x)=151952594)
871
+ 7230 train 4.030438 (lr=1.3864e-04) (hash(x)=145381804)
872
+ 7240 train 3.909897 (lr=1.3789e-04) (hash(x)=142651222)
873
+ 7250 train 4.023250 (lr=1.3715e-04) (hash(x)=136115279)
874
+ 7260 train 4.128131 (lr=1.3641e-04) (hash(x)=154536751)
875
+ 7270 train 4.072914 (lr=1.3568e-04) (hash(x)=148883836)
876
+ 7280 train 3.930397 (lr=1.3495e-04) (hash(x)=150606754)
877
+ 7290 train 4.047969 (lr=1.3422e-04) (hash(x)=165845807)
878
+ 7300 val loss 4.0496
879
+ 7300 val perplexity 57.3754
880
+ 7300 train 4.122186 (lr=1.3350e-04) (hash(x)=150281149)
881
+ 7310 train 4.070421 (lr=1.3278e-04) (hash(x)=159929281)
882
+ 7320 train 3.896353 (lr=1.3207e-04) (hash(x)=143074269)
883
+ 7330 train 4.060897 (lr=1.3136e-04) (hash(x)=147369269)
884
+ 7340 train 4.110459 (lr=1.3066e-04) (hash(x)=139286601)
885
+ 7350 train 4.016146 (lr=1.2996e-04) (hash(x)=145263048)
886
+ 7360 train 3.864773 (lr=1.2926e-04) (hash(x)=144626817)
887
+ 7370 train 4.061992 (lr=1.2857e-04) (hash(x)=168989813)
888
+ 7380 train 4.045014 (lr=1.2789e-04) (hash(x)=144790656)
889
+ 7390 train 3.847455 (lr=1.2721e-04) (hash(x)=141279761)
890
+ 7400 val loss 4.0458
891
+ 7400 val perplexity 57.1575
892
+ 7400 train 3.977643 (lr=1.2653e-04) (hash(x)=148421717)
893
+ 7410 train 3.978487 (lr=1.2586e-04) (hash(x)=141241548)
894
+ 7420 train 4.047942 (lr=1.2519e-04) (hash(x)=162434228)
895
+ 7430 train 3.998271 (lr=1.2453e-04) (hash(x)=156312086)
896
+ 7440 train 3.950279 (lr=1.2387e-04) (hash(x)=139803064)
897
+ 7450 train 3.655139 (lr=1.2322e-04) (hash(x)=156361394)
898
+ 7460 train 3.964633 (lr=1.2257e-04) (hash(x)=141142183)
899
+ 7470 train 3.794017 (lr=1.2192e-04) (hash(x)=156701683)
900
+ 7480 train 3.968296 (lr=1.2128e-04) (hash(x)=148549917)
901
+ 7490 train 3.845934 (lr=1.2065e-04) (hash(x)=143002119)
902
+ 7500 val loss 4.0474
903
+ 7500 val perplexity 57.2499
904
+ 7500 train 3.854910 (lr=1.2002e-04) (hash(x)=146921118)
905
+ 7510 train 3.862853 (lr=1.1939e-04) (hash(x)=148592146)
906
+ 7520 train 4.008068 (lr=1.1877e-04) (hash(x)=148456120)
907
+ 7530 train 4.160484 (lr=1.1816e-04) (hash(x)=163865945)
908
+ 7540 train 4.053192 (lr=1.1754e-04) (hash(x)=160753039)
909
+ 7550 train 4.067111 (lr=1.1694e-04) (hash(x)=150275989)
910
+ 7560 train 4.093420 (lr=1.1633e-04) (hash(x)=141749735)
911
+ 7570 train 4.056247 (lr=1.1574e-04) (hash(x)=148449999)
912
+ 7580 train 3.932683 (lr=1.1514e-04) (hash(x)=139153889)
913
+ 7590 train 4.048226 (lr=1.1455e-04) (hash(x)=147418540)
914
+ 7600 val loss 4.0370
915
+ 7600 val perplexity 56.6569
916
+ 7600 train 4.080087 (lr=1.1397e-04) (hash(x)=150660048)
917
+ 7610 train 4.054899 (lr=1.1339e-04) (hash(x)=155369080)
918
+ 7620 train 3.952520 (lr=1.1282e-04) (hash(x)=151926309)
919
+ 7630 train 4.033431 (lr=1.1225e-04) (hash(x)=149549080)
920
+ 7640 train 3.923625 (lr=1.1168e-04) (hash(x)=140642882)
921
+ 7650 train 4.200468 (lr=1.1112e-04) (hash(x)=140218502)
922
+ 7660 train 4.038733 (lr=1.1057e-04) (hash(x)=153198276)
923
+ 7670 train 3.893281 (lr=1.1002e-04) (hash(x)=150706960)
924
+ 7680 train 4.041800 (lr=1.0947e-04) (hash(x)=152025509)
925
+ 7690 train 3.828660 (lr=1.0893e-04) (hash(x)=147288976)
926
+ 7700 val loss 4.0377
927
+ 7700 val perplexity 56.6975
928
+ 7700 train 3.937320 (lr=1.0840e-04) (hash(x)=148059852)
929
+ 7710 train 3.751970 (lr=1.0786e-04) (hash(x)=158443334)
930
+ 7720 train 3.869091 (lr=1.0734e-04) (hash(x)=143467085)
931
+ 7730 train 3.817537 (lr=1.0682e-04) (hash(x)=138439550)
932
+ 7740 train 3.942839 (lr=1.0630e-04) (hash(x)=145878152)
933
+ 7750 train 3.806863 (lr=1.0579e-04) (hash(x)=136874128)
934
+ 7760 train 3.826232 (lr=1.0528e-04) (hash(x)=148856154)
935
+ 7770 train 3.854548 (lr=1.0478e-04) (hash(x)=146982119)
936
+ 7780 train 3.951974 (lr=1.0428e-04) (hash(x)=159846779)
937
+ 7790 train 3.632563 (lr=1.0379e-04) (hash(x)=141501838)
938
+ 7800 val loss 4.0401
939
+ 7800 val perplexity 56.8312
940
+ 7800 train 3.769263 (lr=1.0330e-04) (hash(x)=148331002)
941
+ 7810 train 4.074198 (lr=1.0282e-04) (hash(x)=150692304)
942
+ 7820 train 3.942809 (lr=1.0234e-04) (hash(x)=156841183)
943
+ 7830 train 3.993902 (lr=1.0187e-04) (hash(x)=139594008)
944
+ 7840 train 3.922303 (lr=1.0140e-04) (hash(x)=149700138)
945
+ 7850 train 3.947139 (lr=1.0094e-04) (hash(x)=149327044)
946
+ 7860 train 4.115314 (lr=1.0048e-04) (hash(x)=148315418)
947
+ 7870 train 4.054947 (lr=1.0002e-04) (hash(x)=153431778)
948
+ 7880 train 3.898095 (lr=9.9576e-05) (hash(x)=145049799)
949
+ 7890 train 4.026278 (lr=9.9133e-05) (hash(x)=150265096)
950
+ 7900 val loss 4.0305
951
+ 7900 val perplexity 56.2887
952
+ 7900 train 3.859587 (lr=9.8694e-05) (hash(x)=164923883)
953
+ 7910 train 3.903379 (lr=9.8261e-05) (hash(x)=139315738)
954
+ 7920 train 3.964164 (lr=9.7832e-05) (hash(x)=153031394)
955
+ 7930 train 3.751105 (lr=9.7408e-05) (hash(x)=148835859)
956
+ 7940 train 3.883246 (lr=9.6990e-05) (hash(x)=146875217)
957
+ 7950 train 3.962027 (lr=9.6576e-05) (hash(x)=151003885)
958
+ 7960 train 3.873352 (lr=9.6167e-05) (hash(x)=141885471)
959
+ 7970 train 4.078067 (lr=9.5764e-05) (hash(x)=150541583)
960
+ 7980 train 3.776524 (lr=9.5365e-05) (hash(x)=138626492)
961
+ 7990 train 3.967917 (lr=9.4971e-05) (hash(x)=149935635)
962
+ 8000 val loss 4.0307
963
+ 8000 val perplexity 56.2988
964
+ 8000 train 3.930342 (lr=9.4583e-05) (hash(x)=143545384)
965
+ 8010 train 3.793429 (lr=9.4199e-05) (hash(x)=144689366)
966
+ 8020 train 3.968970 (lr=9.3820e-05) (hash(x)=151299532)
967
+ 8030 train 3.904565 (lr=9.3446e-05) (hash(x)=131300239)
968
+ 8040 train 3.805311 (lr=9.3078e-05) (hash(x)=146715955)
969
+ 8050 train 3.738265 (lr=9.2714e-05) (hash(x)=135376656)
970
+ 8060 train 3.818822 (lr=9.2356e-05) (hash(x)=141991433)
971
+ 8070 train 3.775491 (lr=9.2002e-05) (hash(x)=140260574)
972
+ 8080 train 3.741298 (lr=9.1654e-05) (hash(x)=151492896)
973
+ 8090 train 3.943421 (lr=9.1310e-05) (hash(x)=143333635)
974
+ 8100 val loss 4.0329
975
+ 8100 val perplexity 56.4217
976
+ 8100 train 3.828446 (lr=9.0972e-05) (hash(x)=160686959)
977
+ 8110 train 3.824474 (lr=9.0638e-05) (hash(x)=141145704)
978
+ 8120 train 4.061954 (lr=9.0310e-05) (hash(x)=146967795)
979
+ 8130 train 3.894828 (lr=8.9987e-05) (hash(x)=147844486)
980
+ 8140 train 3.951035 (lr=8.9669e-05) (hash(x)=141860136)
981
+ 8150 train 4.078087 (lr=8.9356e-05) (hash(x)=147696006)
982
+ 8160 train 3.986072 (lr=8.9048e-05) (hash(x)=151497110)
983
+ 8170 train 3.994370 (lr=8.8745e-05) (hash(x)=140881859)
984
+ 8180 train 3.938019 (lr=8.8447e-05) (hash(x)=158865574)
985
+ 8190 train 3.896375 (lr=8.8154e-05) (hash(x)=161247161)
986
+ 8200 val loss 4.0270
987
+ 8200 val perplexity 56.0899
988
+ 8200 train 3.857462 (lr=8.7867e-05) (hash(x)=156501889)
989
+ 8210 train 3.983042 (lr=8.7584e-05) (hash(x)=158384162)
990
+ 8220 train 4.030195 (lr=8.7307e-05) (hash(x)=154815062)
991
+ 8230 train 3.903517 (lr=8.7035e-05) (hash(x)=145018315)
992
+ 8240 train 3.937562 (lr=8.6768e-05) (hash(x)=161659129)
993
+ 8250 train 3.914047 (lr=8.6506e-05) (hash(x)=139863367)
994
+ 8260 train 3.872205 (lr=8.6249e-05) (hash(x)=149569663)
995
+ 8270 train 3.622681 (lr=8.5997e-05) (hash(x)=162234378)
996
+ 8280 train 4.005273 (lr=8.5750e-05) (hash(x)=144371378)
997
+ 8290 train 3.900044 (lr=8.5509e-05) (hash(x)=148120258)
998
+ 8300 val loss 4.0278
999
+ 8300 val perplexity 56.1347
1000
+ 8300 train 3.898458 (lr=8.5273e-05) (hash(x)=142716875)
1001
+ 8310 train 3.927238 (lr=8.5041e-05) (hash(x)=166789234)
1002
+ 8320 train 3.983576 (lr=8.4815e-05) (hash(x)=150613211)
1003
+ 8330 train 3.867534 (lr=8.4594e-05) (hash(x)=139895391)
1004
+ 8340 train 3.979063 (lr=8.4379e-05) (hash(x)=152011203)
1005
+ 8350 train 3.929983 (lr=8.4168e-05) (hash(x)=156010875)
1006
+ 8360 train 3.935096 (lr=8.3963e-05) (hash(x)=152988051)
1007
+ 8370 train 4.103879 (lr=8.3762e-05) (hash(x)=155144327)
1008
+ 8380 train 3.986644 (lr=8.3567e-05) (hash(x)=133642309)
1009
+ 8390 train 4.035217 (lr=8.3377e-05) (hash(x)=149745335)
1010
+ 8400 val loss 4.0208
1011
+ 8400 val perplexity 55.7441
1012
+ 8400 train 4.012109 (lr=8.3193e-05) (hash(x)=154436684)
1013
+ 8410 train 4.058471 (lr=8.3013e-05) (hash(x)=147946630)
1014
+ 8420 train 3.965513 (lr=8.2839e-05) (hash(x)=153947058)
1015
+ 8430 train 4.058724 (lr=8.2669e-05) (hash(x)=159821040)
1016
+ 8440 train 4.026165 (lr=8.2505e-05) (hash(x)=149127348)
1017
+ 8450 train 4.031816 (lr=8.2347e-05) (hash(x)=149421871)
1018
+ 8460 train 3.999946 (lr=8.2193e-05) (hash(x)=142621577)
1019
+ 8470 train 3.994030 (lr=8.2044e-05) (hash(x)=148978213)
1020
+ 8480 train 3.810085 (lr=8.1901e-05) (hash(x)=139959537)
1021
+ 8490 train 3.901643 (lr=8.1763e-05) (hash(x)=144232498)
1022
+ 8500 val loss 4.0193
1023
+ 8500 val perplexity 55.6622
1024
+ 8500 train 4.119896 (lr=8.1630e-05) (hash(x)=147965839)
1025
+ 8510 train 4.002148 (lr=8.1502e-05) (hash(x)=147679836)
1026
+ 8520 train 3.885429 (lr=8.1380e-05) (hash(x)=147486097)
1027
+ 8530 train 4.019017 (lr=8.1263e-05) (hash(x)=151655265)
1028
+ 8540 train 3.923688 (lr=8.1150e-05) (hash(x)=150866632)
1029
+ 8550 train 4.064749 (lr=8.1044e-05) (hash(x)=144355059)
1030
+ 8560 train 3.929752 (lr=8.0942e-05) (hash(x)=145518478)
1031
+ 8570 train 4.081664 (lr=8.0845e-05) (hash(x)=178973323)
1032
+ 8580 train 3.996408 (lr=8.0754e-05) (hash(x)=152773969)
1033
+ 8590 train 3.793073 (lr=8.0668e-05) (hash(x)=141805508)
1034
+ 8600 val loss 4.0230
1035
+ 8600 val perplexity 55.8669
1036
+ 8600 train 3.843736 (lr=8.0587e-05) (hash(x)=145228097)
1037
+ 8610 train 3.864010 (lr=8.0511e-05) (hash(x)=149525694)
1038
+ 8620 train 3.936628 (lr=8.0441e-05) (hash(x)=151651655)
1039
+ 8630 train 4.055536 (lr=8.0376e-05) (hash(x)=144918643)
1040
+ 8640 train 3.896891 (lr=8.0316e-05) (hash(x)=149721879)
1041
+ 8650 train 3.974156 (lr=8.0261e-05) (hash(x)=157684127)
1042
+ 8660 train 3.976418 (lr=8.0211e-05) (hash(x)=152426036)
1043
+ 8670 train 3.890220 (lr=8.0167e-05) (hash(x)=141396342)
1044
+ 8680 train 3.988662 (lr=8.0128e-05) (hash(x)=146459856)
1045
+ 8690 train 4.078502 (lr=8.0094e-05) (hash(x)=151766124)
1046
+ 8700 val loss 4.0148
1047
+ 8700 val perplexity 55.4135
1048
+ 8700 train 4.221764 (lr=8.0065e-05) (hash(x)=152910357)
1049
+ 8710 train 4.012555 (lr=8.0042e-05) (hash(x)=149724642)
1050
+ 8720 train 3.949025 (lr=8.0023e-05) (hash(x)=152483840)
1051
+ 8730 train 3.935750 (lr=8.0010e-05) (hash(x)=147289415)
1052
+ 8740 train 3.958358 (lr=8.0003e-05) (hash(x)=152254131)
1053
+ 8749 val loss 4.0148
1054
+ 8749 val perplexity 55.4111
lr8e-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 sha256:1be4e422c15dbfef9b089fb8915bb38f3b18e779ac4d5cc871285d634a406477
3
+ size 92843394
lr8e-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:ac81448fd39c457d16483a7cce8d742ef6c602b48f0d82799f412d3940b73461
3
+ size 179406214