andrew-healey commited on
Commit
1e56904
·
verified ·
1 Parent(s): c357701

Upload folder using huggingface_hub

Browse files
logs/64_latent_masks_learnable/learnable_compile_seed_1339/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "logs/64_latent_masks_learnable/learnable_compile_seed_1339", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 12, "n_embd": 768, "head_dim": 64, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": 4375, "warmup_steps": 250, "group": "64_latent_masks_learnable", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1339, "memory_penalty_epsilon": 0.1, "selection_head_linear_combo": "n_latent_masks", "selection_head_linear_combo_scale": 1.0, "disable_selection_head_linear_combo_bias": false, "assert_latent_matches_no_head": false, "n_latent_masks": 1, "init_latent_masks_to_identity": true, "init_latent_masks_to_inverse": false, "latent_mask_scale": null, "latent_mask_runtime_multiplier": null, "latent_mask_sigmoid": false, "S_layernorm": false, "one_head_per_latent_mask": false, "protection_kind": "none", "leaky_relu_alpha": null, "leaky_relu_bias": null, "use_compile": true, "use_mini_model": false, "upload_to_hf": true, "seq_len": 1024, "batch_size": 4, "total_batch_size": 131072, "protection_head_scaling_factor": 1.0, "protection_head_bias": 0.0, "n_sliced_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": 0.003, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "learnable_compile", "latent_mask_precision": "float32"}
logs/64_latent_masks_learnable/learnable_compile_seed_1339/dataloader_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:994809a5ef065b3b39ff8d94ca69f209f3eedb103954655f09fa348a0918b20e
3
+ size 964
logs/64_latent_masks_learnable/learnable_compile_seed_1339/log2.txt ADDED
@@ -0,0 +1,529 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ max_steps: 4375
2
+ 0 val loss 11.0037
3
+ 0 val perplexity 60094.9570
4
+ 0 train 10.996277 (lr=4.1958e-06) (hash(x)=19499830)
5
+ 10 train 9.661660 (lr=4.6154e-05) (hash(x)=19266129)
6
+ 20 train 9.104938 (lr=8.8112e-05) (hash(x)=17425714)
7
+ 30 train 8.462620 (lr=1.3007e-04) (hash(x)=19434488)
8
+ 40 train 7.813193 (lr=1.7203e-04) (hash(x)=18289960)
9
+ 50 train 7.655207 (lr=2.1399e-04) (hash(x)=25963856)
10
+ 60 train 7.672615 (lr=2.5594e-04) (hash(x)=23798061)
11
+ 70 train 7.313420 (lr=2.9790e-04) (hash(x)=19309727)
12
+ 80 train 7.256400 (lr=3.3986e-04) (hash(x)=20092642)
13
+ 90 train 7.026623 (lr=3.8182e-04) (hash(x)=21094807)
14
+ 100 val loss 6.9371
15
+ 100 val perplexity 1029.7471
16
+ 100 train 6.781298 (lr=4.2378e-04) (hash(x)=18910383)
17
+ 110 train 6.704826 (lr=4.6573e-04) (hash(x)=17739111)
18
+ 120 train 6.576969 (lr=5.0769e-04) (hash(x)=20068667)
19
+ 130 train 6.406548 (lr=5.4965e-04) (hash(x)=19548561)
20
+ 140 train 6.396185 (lr=5.9161e-04) (hash(x)=19990057)
21
+ 150 train 6.474767 (lr=6.3357e-04) (hash(x)=17559733)
22
+ 160 train 6.394708 (lr=6.7552e-04) (hash(x)=21295815)
23
+ 170 train 6.282241 (lr=7.1748e-04) (hash(x)=21771608)
24
+ 180 train 6.208201 (lr=7.5944e-04) (hash(x)=18484596)
25
+ 190 train 6.107688 (lr=8.0140e-04) (hash(x)=18663948)
26
+ 200 val loss 6.0331
27
+ 200 val perplexity 416.9859
28
+ 200 train 6.048486 (lr=8.4336e-04) (hash(x)=19306031)
29
+ 210 train 5.942635 (lr=8.8531e-04) (hash(x)=18345886)
30
+ 220 train 5.876199 (lr=9.2727e-04) (hash(x)=18477726)
31
+ 230 train 5.972274 (lr=9.6923e-04) (hash(x)=22315538)
32
+ 240 train 5.681215 (lr=1.0112e-03) (hash(x)=16529598)
33
+ 250 train 5.765741 (lr=1.0531e-03) (hash(x)=23415380)
34
+ 260 train 5.647322 (lr=1.0951e-03) (hash(x)=21452526)
35
+ 270 train 5.577315 (lr=1.1371e-03) (hash(x)=16964237)
36
+ 280 train 5.524441 (lr=1.1790e-03) (hash(x)=20851273)
37
+ 290 train 5.470901 (lr=1.2210e-03) (hash(x)=14094390)
38
+ 300 val loss 5.5575
39
+ 300 val perplexity 259.1627
40
+ 300 train 5.298794 (lr=1.2629e-03) (hash(x)=21184674)
41
+ 310 train 5.233346 (lr=1.3049e-03) (hash(x)=18777755)
42
+ 320 train 5.317426 (lr=1.3469e-03) (hash(x)=23869658)
43
+ 330 train 5.417645 (lr=1.3888e-03) (hash(x)=18147471)
44
+ 340 train 5.514622 (lr=1.4308e-03) (hash(x)=19672778)
45
+ 350 train 5.467280 (lr=1.4727e-03) (hash(x)=21868542)
46
+ 360 train 5.391914 (lr=1.5147e-03) (hash(x)=20606890)
47
+ 370 train 5.427114 (lr=1.5566e-03) (hash(x)=22900825)
48
+ 380 train 5.246624 (lr=1.5986e-03) (hash(x)=20332562)
49
+ 390 train 5.187011 (lr=1.6406e-03) (hash(x)=16917496)
50
+ 400 val loss 5.1981
51
+ 400 val perplexity 180.9364
52
+ 400 train 5.171432 (lr=1.6825e-03) (hash(x)=20419326)
53
+ 410 train 5.136826 (lr=1.7245e-03) (hash(x)=18934133)
54
+ 420 train 5.145296 (lr=1.7664e-03) (hash(x)=17991367)
55
+ 430 train 5.179175 (lr=1.8084e-03) (hash(x)=17769433)
56
+ 440 train 5.069711 (lr=1.8503e-03) (hash(x)=21720430)
57
+ 450 train 5.059713 (lr=1.8923e-03) (hash(x)=14866747)
58
+ 460 train 4.954023 (lr=1.9343e-03) (hash(x)=22009033)
59
+ 470 train 4.766253 (lr=1.9762e-03) (hash(x)=19327881)
60
+ 480 train 4.869951 (lr=2.0182e-03) (hash(x)=19973633)
61
+ 490 train 4.700965 (lr=2.0601e-03) (hash(x)=20294459)
62
+ 500 val loss 4.9759
63
+ 500 val perplexity 144.8754
64
+ 500 train 4.754705 (lr=2.1021e-03) (hash(x)=16902401)
65
+ 510 train 4.855696 (lr=2.1441e-03) (hash(x)=20028937)
66
+ 520 train 4.907742 (lr=2.1860e-03) (hash(x)=18001903)
67
+ 530 train 4.997409 (lr=2.2280e-03) (hash(x)=21435244)
68
+ 540 train 4.912937 (lr=2.2699e-03) (hash(x)=16973801)
69
+ 550 train 4.827078 (lr=2.3119e-03) (hash(x)=21158829)
70
+ 560 train 4.904224 (lr=2.3538e-03) (hash(x)=15382082)
71
+ 570 train 4.785639 (lr=2.3958e-03) (hash(x)=18999843)
72
+ 580 train 4.737366 (lr=2.4378e-03) (hash(x)=19873877)
73
+ 590 train 4.809154 (lr=2.4797e-03) (hash(x)=22520351)
74
+ 600 val loss 4.8239
75
+ 600 val perplexity 124.4507
76
+ 600 train 4.831547 (lr=2.5217e-03) (hash(x)=18681131)
77
+ 610 train 4.733486 (lr=2.5636e-03) (hash(x)=18401442)
78
+ 620 train 4.685786 (lr=2.6056e-03) (hash(x)=21171777)
79
+ 630 train 4.579614 (lr=2.6476e-03) (hash(x)=20257940)
80
+ 640 train 4.559624 (lr=2.6895e-03) (hash(x)=21673710)
81
+ 650 train 4.460366 (lr=2.7315e-03) (hash(x)=16969297)
82
+ 660 train 4.462103 (lr=2.7734e-03) (hash(x)=18464518)
83
+ 670 train 4.439851 (lr=2.8154e-03) (hash(x)=17196302)
84
+ 680 train 4.327279 (lr=2.8573e-03) (hash(x)=18376792)
85
+ 690 train 4.342260 (lr=2.8993e-03) (hash(x)=17928860)
86
+ 700 val loss 4.5871
87
+ 700 val perplexity 98.2049
88
+ 700 train 4.311372 (lr=2.9413e-03) (hash(x)=22330314)
89
+ 710 train 4.608470 (lr=2.9832e-03) (hash(x)=17776819)
90
+ 720 train 4.618995 (lr=3.0000e-03) (hash(x)=20367943)
91
+ 730 train 4.597128 (lr=2.9999e-03) (hash(x)=20244235)
92
+ 740 train 4.524622 (lr=2.9997e-03) (hash(x)=25068819)
93
+ 750 train 4.464368 (lr=2.9994e-03) (hash(x)=20552971)
94
+ 760 train 4.557273 (lr=2.9990e-03) (hash(x)=19170549)
95
+ 770 train 4.492811 (lr=2.9985e-03) (hash(x)=20863809)
96
+ 780 train 4.480120 (lr=2.9979e-03) (hash(x)=23003114)
97
+ 790 train 4.426786 (lr=2.9972e-03) (hash(x)=21939064)
98
+ 800 val loss 4.4467
99
+ 800 val perplexity 85.3432
100
+ 800 train 4.429553 (lr=2.9964e-03) (hash(x)=19245118)
101
+ 810 train 4.321709 (lr=2.9955e-03) (hash(x)=22642355)
102
+ 820 train 4.368881 (lr=2.9945e-03) (hash(x)=23447029)
103
+ 830 train 4.391518 (lr=2.9934e-03) (hash(x)=21708229)
104
+ 840 train 4.314415 (lr=2.9922e-03) (hash(x)=18377450)
105
+ 850 train 4.332704 (lr=2.9909e-03) (hash(x)=16701034)
106
+ 860 train 4.459475 (lr=2.9896e-03) (hash(x)=21348338)
107
+ 870 train 4.410257 (lr=2.9881e-03) (hash(x)=20919606)
108
+ 880 train 4.370855 (lr=2.9865e-03) (hash(x)=22912154)
109
+ 890 train 4.303231 (lr=2.9848e-03) (hash(x)=20629816)
110
+ 900 val loss 4.3233
111
+ 900 val perplexity 75.4350
112
+ 900 train 4.328406 (lr=2.9830e-03) (hash(x)=21120493)
113
+ 910 train 4.306761 (lr=2.9811e-03) (hash(x)=19354333)
114
+ 920 train 4.331552 (lr=2.9792e-03) (hash(x)=22251784)
115
+ 930 train 4.265792 (lr=2.9771e-03) (hash(x)=17462001)
116
+ 940 train 4.221290 (lr=2.9749e-03) (hash(x)=21939624)
117
+ 950 train 4.257472 (lr=2.9726e-03) (hash(x)=20914464)
118
+ 960 train 4.257403 (lr=2.9703e-03) (hash(x)=18974331)
119
+ 970 train 4.227310 (lr=2.9678e-03) (hash(x)=21032787)
120
+ 980 train 4.204168 (lr=2.9652e-03) (hash(x)=18848940)
121
+ 990 train 4.221955 (lr=2.9626e-03) (hash(x)=20074037)
122
+ 1000 val loss 4.2171
123
+ 1000 val perplexity 67.8381
124
+ 1000 train 4.221393 (lr=2.9598e-03) (hash(x)=18791811)
125
+ 1010 train 4.407270 (lr=2.9570e-03) (hash(x)=16870030)
126
+ 1020 train 4.272988 (lr=2.9540e-03) (hash(x)=17382263)
127
+ 1030 train 4.244178 (lr=2.9510e-03) (hash(x)=19518020)
128
+ 1040 train 4.199749 (lr=2.9478e-03) (hash(x)=18479196)
129
+ 1050 train 4.202653 (lr=2.9446e-03) (hash(x)=20240434)
130
+ 1060 train 4.211992 (lr=2.9412e-03) (hash(x)=20176962)
131
+ 1070 train 4.250513 (lr=2.9378e-03) (hash(x)=22429292)
132
+ 1080 train 4.102672 (lr=2.9343e-03) (hash(x)=20635325)
133
+ 1090 train 4.147540 (lr=2.9307e-03) (hash(x)=23524462)
134
+ 1100 val loss 4.1350
135
+ 1100 val perplexity 62.4922
136
+ 1100 train 4.156718 (lr=2.9270e-03) (hash(x)=25636544)
137
+ 1110 train 4.094840 (lr=2.9231e-03) (hash(x)=21984040)
138
+ 1120 train 4.090715 (lr=2.9192e-03) (hash(x)=21955314)
139
+ 1130 train 4.147988 (lr=2.9152e-03) (hash(x)=24632576)
140
+ 1140 train 4.138317 (lr=2.9112e-03) (hash(x)=22566911)
141
+ 1150 train 4.255769 (lr=2.9070e-03) (hash(x)=19348391)
142
+ 1160 train 4.211156 (lr=2.9027e-03) (hash(x)=13676213)
143
+ 1170 train 4.167640 (lr=2.8983e-03) (hash(x)=21126914)
144
+ 1180 train 4.099019 (lr=2.8939e-03) (hash(x)=21859315)
145
+ 1190 train 4.092085 (lr=2.8893e-03) (hash(x)=22140691)
146
+ 1200 val loss 4.0646
147
+ 1200 val perplexity 58.2401
148
+ 1200 train 3.988214 (lr=2.8847e-03) (hash(x)=21010060)
149
+ 1210 train 4.063884 (lr=2.8800e-03) (hash(x)=25216648)
150
+ 1220 train 4.044487 (lr=2.8751e-03) (hash(x)=19447714)
151
+ 1230 train 3.986189 (lr=2.8702e-03) (hash(x)=20268415)
152
+ 1240 train 4.071821 (lr=2.8652e-03) (hash(x)=21742969)
153
+ 1250 train 3.945498 (lr=2.8601e-03) (hash(x)=23680287)
154
+ 1260 train 4.016034 (lr=2.8550e-03) (hash(x)=21568309)
155
+ 1270 train 3.960721 (lr=2.8497e-03) (hash(x)=20176939)
156
+ 1280 train 4.016812 (lr=2.8443e-03) (hash(x)=21657918)
157
+ 1290 train 4.039239 (lr=2.8389e-03) (hash(x)=20749696)
158
+ 1300 val loss 4.0126
159
+ 1300 val perplexity 55.2883
160
+ 1300 train 4.045858 (lr=2.8333e-03) (hash(x)=21178552)
161
+ 1310 train 4.174490 (lr=2.8277e-03) (hash(x)=24008496)
162
+ 1320 train 3.915365 (lr=2.8220e-03) (hash(x)=20529379)
163
+ 1330 train 3.990843 (lr=2.8162e-03) (hash(x)=19580783)
164
+ 1340 train 4.104467 (lr=2.8103e-03) (hash(x)=22031001)
165
+ 1350 train 3.926534 (lr=2.8044e-03) (hash(x)=19758453)
166
+ 1360 train 4.048903 (lr=2.7983e-03) (hash(x)=20524989)
167
+ 1370 train 3.913294 (lr=2.7922e-03) (hash(x)=18611857)
168
+ 1380 train 3.955461 (lr=2.7860e-03) (hash(x)=18097851)
169
+ 1390 train 4.005003 (lr=2.7797e-03) (hash(x)=18841925)
170
+ 1400 val loss 3.9855
171
+ 1400 val perplexity 53.8131
172
+ 1400 train 3.944579 (lr=2.7733e-03) (hash(x)=20959768)
173
+ 1410 train 4.034330 (lr=2.7668e-03) (hash(x)=18836875)
174
+ 1420 train 4.148649 (lr=2.7603e-03) (hash(x)=19657234)
175
+ 1430 train 3.916734 (lr=2.7536e-03) (hash(x)=21047683)
176
+ 1440 train 4.056051 (lr=2.7469e-03) (hash(x)=26743461)
177
+ 1450 train 3.980730 (lr=2.7401e-03) (hash(x)=22881324)
178
+ 1460 train 3.903520 (lr=2.7333e-03) (hash(x)=24194800)
179
+ 1470 train 3.972186 (lr=2.7263e-03) (hash(x)=23399930)
180
+ 1480 train 4.008898 (lr=2.7193e-03) (hash(x)=20569902)
181
+ 1490 train 3.912452 (lr=2.7121e-03) (hash(x)=18223528)
182
+ 1500 val loss 3.9311
183
+ 1500 val perplexity 50.9605
184
+ 1500 train 4.015514 (lr=2.7050e-03) (hash(x)=19550097)
185
+ 1510 train 3.912793 (lr=2.6977e-03) (hash(x)=19347158)
186
+ 1520 train 3.912911 (lr=2.6903e-03) (hash(x)=16014775)
187
+ 1530 train 3.855129 (lr=2.6829e-03) (hash(x)=19187039)
188
+ 1540 train 3.911889 (lr=2.6754e-03) (hash(x)=19718445)
189
+ 1550 train 3.961654 (lr=2.6678e-03) (hash(x)=20755662)
190
+ 1560 train 3.878567 (lr=2.6602e-03) (hash(x)=20026689)
191
+ 1570 train 3.913325 (lr=2.6525e-03) (hash(x)=20481658)
192
+ 1580 train 3.901427 (lr=2.6447e-03) (hash(x)=20415281)
193
+ 1590 train 4.020175 (lr=2.6368e-03) (hash(x)=21683290)
194
+ 1600 val loss 3.9092
195
+ 1600 val perplexity 49.8573
196
+ 1600 train 3.983990 (lr=2.6289e-03) (hash(x)=21287292)
197
+ 1610 train 4.039619 (lr=2.6208e-03) (hash(x)=20072672)
198
+ 1620 train 3.902715 (lr=2.6128e-03) (hash(x)=15774154)
199
+ 1630 train 3.924248 (lr=2.6046e-03) (hash(x)=19253426)
200
+ 1640 train 4.010849 (lr=2.5964e-03) (hash(x)=18235717)
201
+ 1650 train 3.910837 (lr=2.5881e-03) (hash(x)=17940066)
202
+ 1660 train 3.899940 (lr=2.5797e-03) (hash(x)=20731779)
203
+ 1670 train 3.906188 (lr=2.5713e-03) (hash(x)=21326011)
204
+ 1680 train 3.725294 (lr=2.5628e-03) (hash(x)=19569086)
205
+ 1690 train 3.756712 (lr=2.5542e-03) (hash(x)=17136153)
206
+ 1700 val loss 3.8965
207
+ 1700 val perplexity 49.2286
208
+ 1700 train 3.969625 (lr=2.5455e-03) (hash(x)=24947450)
209
+ 1710 train 3.915545 (lr=2.5368e-03) (hash(x)=20835959)
210
+ 1720 train 3.947629 (lr=2.5281e-03) (hash(x)=19803138)
211
+ 1730 train 3.912626 (lr=2.5192e-03) (hash(x)=23602127)
212
+ 1740 train 3.942135 (lr=2.5103e-03) (hash(x)=21588728)
213
+ 1750 train 3.918318 (lr=2.5014e-03) (hash(x)=22545660)
214
+ 1760 train 3.920965 (lr=2.4924e-03) (hash(x)=22464423)
215
+ 1770 train 3.787527 (lr=2.4833e-03) (hash(x)=22560125)
216
+ 1780 train 3.880257 (lr=2.4741e-03) (hash(x)=20628381)
217
+ 1790 train 3.894936 (lr=2.4649e-03) (hash(x)=19928492)
218
+ 1800 val loss 3.8517
219
+ 1800 val perplexity 47.0716
220
+ 1800 train 3.882275 (lr=2.4556e-03) (hash(x)=19855177)
221
+ 1810 train 3.844957 (lr=2.4463e-03) (hash(x)=17417243)
222
+ 1820 train 3.690928 (lr=2.4369e-03) (hash(x)=20821803)
223
+ 1830 train 3.803125 (lr=2.4275e-03) (hash(x)=24879137)
224
+ 1840 train 3.774239 (lr=2.4180e-03) (hash(x)=20844335)
225
+ 1850 train 3.802801 (lr=2.4084e-03) (hash(x)=18445411)
226
+ 1860 train 3.927529 (lr=2.3988e-03) (hash(x)=22194917)
227
+ 1870 train 3.944360 (lr=2.3891e-03) (hash(x)=21291954)
228
+ 1880 train 3.777469 (lr=2.3794e-03) (hash(x)=18252425)
229
+ 1890 train 3.883779 (lr=2.3696e-03) (hash(x)=17424764)
230
+ 1900 val loss 3.8201
231
+ 1900 val perplexity 45.6066
232
+ 1900 train 3.797578 (lr=2.3598e-03) (hash(x)=19759252)
233
+ 1910 train 3.752006 (lr=2.3499e-03) (hash(x)=17053540)
234
+ 1920 train 3.850797 (lr=2.3400e-03) (hash(x)=20059888)
235
+ 1930 train 3.698312 (lr=2.3300e-03) (hash(x)=18717513)
236
+ 1940 train 3.795895 (lr=2.3200e-03) (hash(x)=21252145)
237
+ 1950 train 3.785777 (lr=2.3099e-03) (hash(x)=17166480)
238
+ 1960 train 3.680806 (lr=2.2998e-03) (hash(x)=19791836)
239
+ 1970 train 3.836262 (lr=2.2896e-03) (hash(x)=20908476)
240
+ 1980 train 3.781847 (lr=2.2793e-03) (hash(x)=17252455)
241
+ 1990 train 3.785394 (lr=2.2691e-03) (hash(x)=19187120)
242
+ 2000 val loss 3.7923
243
+ 2000 val perplexity 44.3576
244
+ 2000 train 3.812829 (lr=2.2588e-03) (hash(x)=20996686)
245
+ 2010 train 3.820431 (lr=2.2484e-03) (hash(x)=22420273)
246
+ 2020 train 3.809748 (lr=2.2380e-03) (hash(x)=21035175)
247
+ 2030 train 3.848121 (lr=2.2275e-03) (hash(x)=21464100)
248
+ 2040 train 3.875175 (lr=2.2170e-03) (hash(x)=19332299)
249
+ 2050 train 3.760391 (lr=2.2065e-03) (hash(x)=16978105)
250
+ 2060 train 3.976656 (lr=2.1959e-03) (hash(x)=17517702)
251
+ 2070 train 3.800172 (lr=2.1853e-03) (hash(x)=19999968)
252
+ 2080 train 3.821846 (lr=2.1746e-03) (hash(x)=17974202)
253
+ 2090 train 3.661211 (lr=2.1639e-03) (hash(x)=18093857)
254
+ 2100 val loss 3.7603
255
+ 2100 val perplexity 42.9633
256
+ 2100 train 3.716931 (lr=2.1532e-03) (hash(x)=18025348)
257
+ 2110 train 3.763976 (lr=2.1424e-03) (hash(x)=18562385)
258
+ 2120 train 3.715168 (lr=2.1316e-03) (hash(x)=21660073)
259
+ 2130 train 3.837141 (lr=2.1208e-03) (hash(x)=23688255)
260
+ 2140 train 3.901382 (lr=2.1099e-03) (hash(x)=22201770)
261
+ 2150 train 3.815487 (lr=2.0990e-03) (hash(x)=19338752)
262
+ 2160 train 3.802342 (lr=2.0881e-03) (hash(x)=15027506)
263
+ 2170 train 3.784975 (lr=2.0771e-03) (hash(x)=16328757)
264
+ 2180 train 3.792895 (lr=2.0661e-03) (hash(x)=25769566)
265
+ 2190 train 3.802797 (lr=2.0550e-03) (hash(x)=21688947)
266
+ 2200 val loss 3.7382
267
+ 2200 val perplexity 42.0206
268
+ 2200 train 3.800755 (lr=2.0440e-03) (hash(x)=15500010)
269
+ 2210 train 3.745121 (lr=2.0329e-03) (hash(x)=17836120)
270
+ 2220 train 3.726338 (lr=2.0217e-03) (hash(x)=19443168)
271
+ 2230 train 3.653182 (lr=2.0106e-03) (hash(x)=17297013)
272
+ 2240 train 3.705297 (lr=1.9994e-03) (hash(x)=18650030)
273
+ 2250 train 3.671704 (lr=1.9882e-03) (hash(x)=15805613)
274
+ 2260 train 3.642856 (lr=1.9770e-03) (hash(x)=19942534)
275
+ 2270 train 3.657914 (lr=1.9657e-03) (hash(x)=22599430)
276
+ 2280 train 3.902683 (lr=1.9544e-03) (hash(x)=22218719)
277
+ 2290 train 3.803659 (lr=1.9431e-03) (hash(x)=19342048)
278
+ 2300 val loss 3.7130
279
+ 2300 val perplexity 40.9767
280
+ 2300 train 3.682651 (lr=1.9318e-03) (hash(x)=18810407)
281
+ 2310 train 3.767266 (lr=1.9205e-03) (hash(x)=22073167)
282
+ 2320 train 3.814688 (lr=1.9091e-03) (hash(x)=20883748)
283
+ 2330 train 3.630770 (lr=1.8977e-03) (hash(x)=15010975)
284
+ 2340 train 3.782107 (lr=1.8863e-03) (hash(x)=17153145)
285
+ 2350 train 3.685498 (lr=1.8749e-03) (hash(x)=25451452)
286
+ 2360 train 3.764899 (lr=1.8635e-03) (hash(x)=19036737)
287
+ 2370 train 3.829079 (lr=1.8520e-03) (hash(x)=18582711)
288
+ 2380 train 3.719662 (lr=1.8406e-03) (hash(x)=22311667)
289
+ 2390 train 3.774106 (lr=1.8291e-03) (hash(x)=23289834)
290
+ 2400 val loss 3.6998
291
+ 2400 val perplexity 40.4405
292
+ 2400 train 3.650705 (lr=1.8176e-03) (hash(x)=19343835)
293
+ 2410 train 3.525054 (lr=1.8061e-03) (hash(x)=19070868)
294
+ 2420 train 3.654564 (lr=1.7946e-03) (hash(x)=21492358)
295
+ 2430 train 3.575347 (lr=1.7830e-03) (hash(x)=21355438)
296
+ 2440 train 3.605853 (lr=1.7715e-03) (hash(x)=18688923)
297
+ 2450 train 3.442690 (lr=1.7600e-03) (hash(x)=18655016)
298
+ 2460 train 3.460123 (lr=1.7484e-03) (hash(x)=23966823)
299
+ 2470 train 3.415787 (lr=1.7368e-03) (hash(x)=19789602)
300
+ 2480 train 3.782797 (lr=1.7253e-03) (hash(x)=18904369)
301
+ 2490 train 3.768910 (lr=1.7137e-03) (hash(x)=18638893)
302
+ 2500 val loss 3.6752
303
+ 2500 val perplexity 39.4585
304
+ 2500 train 3.678448 (lr=1.7021e-03) (hash(x)=16815212)
305
+ 2510 train 3.649680 (lr=1.6906e-03) (hash(x)=24781988)
306
+ 2520 train 3.741364 (lr=1.6790e-03) (hash(x)=18249583)
307
+ 2530 train 3.709072 (lr=1.6674e-03) (hash(x)=18203346)
308
+ 2540 train 3.713552 (lr=1.6558e-03) (hash(x)=19029376)
309
+ 2550 train 3.686074 (lr=1.6442e-03) (hash(x)=21525059)
310
+ 2560 train 3.644967 (lr=1.6326e-03) (hash(x)=21419471)
311
+ 2570 train 3.699338 (lr=1.6210e-03) (hash(x)=19425528)
312
+ 2580 train 3.700691 (lr=1.6094e-03) (hash(x)=16653977)
313
+ 2590 train 3.630935 (lr=1.5979e-03) (hash(x)=23372914)
314
+ 2600 val loss 3.6499
315
+ 2600 val perplexity 38.4699
316
+ 2600 train 3.619622 (lr=1.5863e-03) (hash(x)=15676604)
317
+ 2610 train 3.606596 (lr=1.5747e-03) (hash(x)=24943464)
318
+ 2620 train 3.468441 (lr=1.5632e-03) (hash(x)=18637598)
319
+ 2630 train 3.426165 (lr=1.5516e-03) (hash(x)=20671881)
320
+ 2640 train 3.480697 (lr=1.5400e-03) (hash(x)=21040846)
321
+ 2650 train 3.525172 (lr=1.5285e-03) (hash(x)=22880651)
322
+ 2660 train 3.636086 (lr=1.5170e-03) (hash(x)=18865768)
323
+ 2670 train 3.781177 (lr=1.5054e-03) (hash(x)=21044077)
324
+ 2680 train 3.678308 (lr=1.4939e-03) (hash(x)=20980745)
325
+ 2690 train 3.650933 (lr=1.4824e-03) (hash(x)=19495821)
326
+ 2700 val loss 3.6363
327
+ 2700 val perplexity 37.9503
328
+ 2700 train 3.782370 (lr=1.4709e-03) (hash(x)=24282313)
329
+ 2710 train 3.797214 (lr=1.4594e-03) (hash(x)=16434484)
330
+ 2720 train 3.641211 (lr=1.4480e-03) (hash(x)=19401384)
331
+ 2730 train 3.550089 (lr=1.4365e-03) (hash(x)=29013935)
332
+ 2740 train 3.527739 (lr=1.4251e-03) (hash(x)=20089946)
333
+ 2750 train 3.669123 (lr=1.4137e-03) (hash(x)=19869457)
334
+ 2760 train 3.577068 (lr=1.4023e-03) (hash(x)=16709635)
335
+ 2770 train 3.577834 (lr=1.3909e-03) (hash(x)=22743096)
336
+ 2780 train 3.579457 (lr=1.3795e-03) (hash(x)=21850887)
337
+ 2790 train 3.553382 (lr=1.3682e-03) (hash(x)=21898297)
338
+ 2800 val loss 3.6126
339
+ 2800 val perplexity 37.0617
340
+ 2800 train 3.423060 (lr=1.3569e-03) (hash(x)=21413897)
341
+ 2810 train 3.439198 (lr=1.3456e-03) (hash(x)=19920845)
342
+ 2820 train 3.424116 (lr=1.3343e-03) (hash(x)=24043370)
343
+ 2830 train 3.506849 (lr=1.3230e-03) (hash(x)=18524290)
344
+ 2840 train 3.442428 (lr=1.3118e-03) (hash(x)=19600591)
345
+ 2850 train 3.614357 (lr=1.3006e-03) (hash(x)=21409231)
346
+ 2860 train 3.525144 (lr=1.2894e-03) (hash(x)=22752705)
347
+ 2870 train 3.731093 (lr=1.2783e-03) (hash(x)=19111240)
348
+ 2880 train 3.697103 (lr=1.2671e-03) (hash(x)=19723832)
349
+ 2890 train 3.505202 (lr=1.2560e-03) (hash(x)=20931007)
350
+ 2900 val loss 3.5904
351
+ 2900 val perplexity 36.2484
352
+ 2900 train 3.594578 (lr=1.2450e-03) (hash(x)=18157571)
353
+ 2910 train 3.791388 (lr=1.2339e-03) (hash(x)=15451342)
354
+ 2920 train 3.662594 (lr=1.2229e-03) (hash(x)=20840624)
355
+ 2930 train 3.686800 (lr=1.2119e-03) (hash(x)=20592314)
356
+ 2940 train 3.575172 (lr=1.2010e-03) (hash(x)=18543854)
357
+ 2950 train 3.570055 (lr=1.1901e-03) (hash(x)=18949144)
358
+ 2960 train 3.537025 (lr=1.1792e-03) (hash(x)=21711836)
359
+ 2970 train 3.519863 (lr=1.1684e-03) (hash(x)=22932108)
360
+ 2980 train 3.564512 (lr=1.1576e-03) (hash(x)=21153871)
361
+ 2990 train 3.473821 (lr=1.1468e-03) (hash(x)=20821229)
362
+ 3000 val loss 3.5815
363
+ 3000 val perplexity 35.9287
364
+ 3000 train 3.463166 (lr=1.1361e-03) (hash(x)=20308294)
365
+ 3010 train 3.355427 (lr=1.1254e-03) (hash(x)=15568285)
366
+ 3020 train 3.401875 (lr=1.1147e-03) (hash(x)=25333743)
367
+ 3030 train 3.678092 (lr=1.1041e-03) (hash(x)=21412609)
368
+ 3040 train 3.616896 (lr=1.0935e-03) (hash(x)=20306459)
369
+ 3050 train 3.637301 (lr=1.0830e-03) (hash(x)=22401325)
370
+ 3060 train 3.772789 (lr=1.0725e-03) (hash(x)=17461443)
371
+ 3070 train 3.550656 (lr=1.0620e-03) (hash(x)=21060753)
372
+ 3080 train 3.603452 (lr=1.0516e-03) (hash(x)=18276937)
373
+ 3090 train 3.658507 (lr=1.0412e-03) (hash(x)=22919536)
374
+ 3100 val loss 3.5567
375
+ 3100 val perplexity 35.0476
376
+ 3100 train 3.532186 (lr=1.0309e-03) (hash(x)=22068258)
377
+ 3110 train 3.565305 (lr=1.0207e-03) (hash(x)=22238174)
378
+ 3120 train 3.421310 (lr=1.0104e-03) (hash(x)=19841337)
379
+ 3130 train 3.443990 (lr=1.0002e-03) (hash(x)=18274815)
380
+ 3140 train 3.540902 (lr=9.9011e-04) (hash(x)=22162670)
381
+ 3150 train 3.570851 (lr=9.8002e-04) (hash(x)=22571495)
382
+ 3160 train 3.589292 (lr=9.6999e-04) (hash(x)=22642982)
383
+ 3170 train 3.604723 (lr=9.6000e-04) (hash(x)=20033229)
384
+ 3180 train 3.633376 (lr=9.5007e-04) (hash(x)=19437509)
385
+ 3190 train 3.649866 (lr=9.4019e-04) (hash(x)=16783200)
386
+ 3200 val loss 3.5366
387
+ 3200 val perplexity 34.3493
388
+ 3200 train 3.593626 (lr=9.3036e-04) (hash(x)=24078985)
389
+ 3210 train 3.548767 (lr=9.2058e-04) (hash(x)=17215036)
390
+ 3220 train 3.606020 (lr=9.1085e-04) (hash(x)=20834133)
391
+ 3230 train 3.652124 (lr=9.0118e-04) (hash(x)=22319183)
392
+ 3240 train 3.636352 (lr=8.9157e-04) (hash(x)=18488174)
393
+ 3250 train 3.548627 (lr=8.8201e-04) (hash(x)=18309688)
394
+ 3260 train 3.568924 (lr=8.7251e-04) (hash(x)=22420069)
395
+ 3270 train 3.540922 (lr=8.6307e-04) (hash(x)=20198122)
396
+ 3280 train 3.505180 (lr=8.5368e-04) (hash(x)=19296839)
397
+ 3290 train 3.627996 (lr=8.4435e-04) (hash(x)=18856940)
398
+ 3300 val loss 3.5253
399
+ 3300 val perplexity 33.9653
400
+ 3300 train 3.433718 (lr=8.3508e-04) (hash(x)=18860544)
401
+ 3310 train 3.451977 (lr=8.2588e-04) (hash(x)=18540569)
402
+ 3320 train 3.570929 (lr=8.1673e-04) (hash(x)=19524980)
403
+ 3330 train 3.281922 (lr=8.0764e-04) (hash(x)=20135353)
404
+ 3340 train 3.457626 (lr=7.9862e-04) (hash(x)=18231276)
405
+ 3350 train 3.547929 (lr=7.8966e-04) (hash(x)=21460655)
406
+ 3360 train 3.565737 (lr=7.8076e-04) (hash(x)=21427500)
407
+ 3370 train 3.502729 (lr=7.7192e-04) (hash(x)=20983682)
408
+ 3380 train 3.592497 (lr=7.6315e-04) (hash(x)=19255896)
409
+ 3390 train 3.725065 (lr=7.5445e-04) (hash(x)=22402624)
410
+ 3400 val loss 3.5035
411
+ 3400 val perplexity 33.2312
412
+ 3400 train 3.626617 (lr=7.4581e-04) (hash(x)=18912715)
413
+ 3410 train 3.618193 (lr=7.3724e-04) (hash(x)=21257398)
414
+ 3420 train 3.601127 (lr=7.2874e-04) (hash(x)=21948097)
415
+ 3430 train 3.721388 (lr=7.2030e-04) (hash(x)=20347030)
416
+ 3440 train 3.508934 (lr=7.1193e-04) (hash(x)=21481378)
417
+ 3450 train 3.515995 (lr=7.0363e-04) (hash(x)=17546482)
418
+ 3460 train 3.576174 (lr=6.9541e-04) (hash(x)=18550105)
419
+ 3470 train 3.547077 (lr=6.8725e-04) (hash(x)=24547129)
420
+ 3480 train 3.492991 (lr=6.7916e-04) (hash(x)=20098288)
421
+ 3490 train 3.360544 (lr=6.7114e-04) (hash(x)=21452276)
422
+ 3500 val loss 3.4945
423
+ 3500 val perplexity 32.9324
424
+ 3500 train 3.300024 (lr=6.6320e-04) (hash(x)=23225933)
425
+ 3510 train 3.472671 (lr=6.5533e-04) (hash(x)=21071905)
426
+ 3520 train 3.331056 (lr=6.4753e-04) (hash(x)=16906922)
427
+ 3530 train 3.616835 (lr=6.3981e-04) (hash(x)=21899151)
428
+ 3540 train 3.587606 (lr=6.3216e-04) (hash(x)=23041756)
429
+ 3550 train 3.623541 (lr=6.2458e-04) (hash(x)=21732674)
430
+ 3560 train 3.548153 (lr=6.1708e-04) (hash(x)=18473192)
431
+ 3570 train 3.581241 (lr=6.0966e-04) (hash(x)=14969150)
432
+ 3580 train 3.538009 (lr=6.0231e-04) (hash(x)=24237738)
433
+ 3590 train 3.454856 (lr=5.9504e-04) (hash(x)=17182881)
434
+ 3600 val loss 3.4756
435
+ 3600 val perplexity 32.3174
436
+ 3600 train 3.487728 (lr=5.8785e-04) (hash(x)=21408499)
437
+ 3610 train 3.465815 (lr=5.8074e-04) (hash(x)=20004592)
438
+ 3620 train 3.550044 (lr=5.7370e-04) (hash(x)=19807690)
439
+ 3630 train 3.475251 (lr=5.6675e-04) (hash(x)=15802211)
440
+ 3640 train 3.497716 (lr=5.5987e-04) (hash(x)=17076697)
441
+ 3650 train 3.549971 (lr=5.5308e-04) (hash(x)=18190785)
442
+ 3660 train 3.353331 (lr=5.4636e-04) (hash(x)=19759253)
443
+ 3670 train 3.417983 (lr=5.3973e-04) (hash(x)=17023532)
444
+ 3680 train 3.298580 (lr=5.3318e-04) (hash(x)=17433843)
445
+ 3690 train 3.294468 (lr=5.2671e-04) (hash(x)=18793933)
446
+ 3700 val loss 3.4722
447
+ 3700 val perplexity 32.2069
448
+ 3700 train 3.405239 (lr=5.2033e-04) (hash(x)=17153326)
449
+ 3710 train 3.578720 (lr=5.1402e-04) (hash(x)=18663421)
450
+ 3720 train 3.505310 (lr=5.0780e-04) (hash(x)=18726507)
451
+ 3730 train 3.489444 (lr=5.0167e-04) (hash(x)=21535668)
452
+ 3740 train 3.681874 (lr=4.9562e-04) (hash(x)=23817997)
453
+ 3750 train 3.519021 (lr=4.8965e-04) (hash(x)=16956634)
454
+ 3760 train 3.579981 (lr=4.8377e-04) (hash(x)=19896983)
455
+ 3770 train 3.557437 (lr=4.7798e-04) (hash(x)=20117113)
456
+ 3780 train 3.513931 (lr=4.7227e-04) (hash(x)=15844634)
457
+ 3790 train 3.586900 (lr=4.6665e-04) (hash(x)=20319750)
458
+ 3800 val loss 3.4546
459
+ 3800 val perplexity 31.6442
460
+ 3800 train 3.460478 (lr=4.6112e-04) (hash(x)=20435350)
461
+ 3810 train 3.477315 (lr=4.5567e-04) (hash(x)=20641047)
462
+ 3820 train 3.544650 (lr=4.5031e-04) (hash(x)=18297741)
463
+ 3830 train 3.456487 (lr=4.4504e-04) (hash(x)=17763890)
464
+ 3840 train 3.440894 (lr=4.3986e-04) (hash(x)=23487198)
465
+ 3850 train 3.415534 (lr=4.3477e-04) (hash(x)=21778602)
466
+ 3860 train 3.434304 (lr=4.2977e-04) (hash(x)=20416853)
467
+ 3870 train 3.340709 (lr=4.2486e-04) (hash(x)=20694619)
468
+ 3880 train 3.409158 (lr=4.2004e-04) (hash(x)=20581652)
469
+ 3890 train 3.487067 (lr=4.1530e-04) (hash(x)=24613563)
470
+ 3900 val loss 3.4448
471
+ 3900 val perplexity 31.3368
472
+ 3900 train 3.467971 (lr=4.1066e-04) (hash(x)=19662796)
473
+ 3910 train 3.428778 (lr=4.0611e-04) (hash(x)=23741407)
474
+ 3920 train 3.470999 (lr=4.0166e-04) (hash(x)=23031285)
475
+ 3930 train 3.506700 (lr=3.9729e-04) (hash(x)=23239084)
476
+ 3940 train 3.420606 (lr=3.9302e-04) (hash(x)=19296933)
477
+ 3950 train 3.426708 (lr=3.8884e-04) (hash(x)=20099740)
478
+ 3960 train 3.494110 (lr=3.8475e-04) (hash(x)=22043697)
479
+ 3970 train 3.470921 (lr=3.8076e-04) (hash(x)=27563931)
480
+ 3980 train 3.513916 (lr=3.7685e-04) (hash(x)=21251196)
481
+ 3990 train 3.332495 (lr=3.7305e-04) (hash(x)=22898068)
482
+ 4000 val loss 3.4350
483
+ 4000 val perplexity 31.0313
484
+ 4000 train 3.416241 (lr=3.6933e-04) (hash(x)=19449887)
485
+ 4010 train 3.423499 (lr=3.6572e-04) (hash(x)=21147640)
486
+ 4020 train 3.337980 (lr=3.6219e-04) (hash(x)=19826876)
487
+ 4030 train 3.464603 (lr=3.5876e-04) (hash(x)=23648072)
488
+ 4040 train 3.429636 (lr=3.5543e-04) (hash(x)=21568106)
489
+ 4050 train 3.406315 (lr=3.5219e-04) (hash(x)=28260637)
490
+ 4060 train 3.440016 (lr=3.4905e-04) (hash(x)=20127545)
491
+ 4070 train 3.411449 (lr=3.4600e-04) (hash(x)=20608325)
492
+ 4080 train 3.570611 (lr=3.4305e-04) (hash(x)=22853386)
493
+ 4090 train 3.444717 (lr=3.4019e-04) (hash(x)=21675157)
494
+ 4100 val loss 3.4236
495
+ 4100 val perplexity 30.6796
496
+ 4100 train 3.482323 (lr=3.3744e-04) (hash(x)=17377009)
497
+ 4110 train 3.455144 (lr=3.3477e-04) (hash(x)=22277706)
498
+ 4120 train 3.486718 (lr=3.3221e-04) (hash(x)=19781296)
499
+ 4130 train 3.475528 (lr=3.2974e-04) (hash(x)=14644070)
500
+ 4140 train 3.424757 (lr=3.2737e-04) (hash(x)=17324623)
501
+ 4150 train 3.439077 (lr=3.2510e-04) (hash(x)=18640674)
502
+ 4160 train 3.351342 (lr=3.2292e-04) (hash(x)=20754523)
503
+ 4170 train 3.325888 (lr=3.2085e-04) (hash(x)=22334132)
504
+ 4180 train 3.440910 (lr=3.1887e-04) (hash(x)=21104828)
505
+ 4190 train 3.458594 (lr=3.1699e-04) (hash(x)=19725849)
506
+ 4200 val loss 3.4197
507
+ 4200 val perplexity 30.5596
508
+ 4200 train 3.522397 (lr=3.1520e-04) (hash(x)=24693657)
509
+ 4210 train 3.409537 (lr=3.1352e-04) (hash(x)=20148669)
510
+ 4220 train 3.417567 (lr=3.1193e-04) (hash(x)=26860969)
511
+ 4230 train 3.462051 (lr=3.1044e-04) (hash(x)=20146192)
512
+ 4240 train 3.456519 (lr=3.0905e-04) (hash(x)=15710025)
513
+ 4250 train 3.480284 (lr=3.0776e-04) (hash(x)=16513041)
514
+ 4260 train 3.474449 (lr=3.0657e-04) (hash(x)=17073880)
515
+ 4270 train 3.386276 (lr=3.0548e-04) (hash(x)=21596384)
516
+ 4280 train 3.434125 (lr=3.0449e-04) (hash(x)=18548369)
517
+ 4290 train 3.333372 (lr=3.0359e-04) (hash(x)=20524519)
518
+ 4300 val loss 3.4170
519
+ 4300 val perplexity 30.4777
520
+ 4300 train 3.398649 (lr=3.0280e-04) (hash(x)=20906140)
521
+ 4310 train 3.358960 (lr=3.0210e-04) (hash(x)=21519922)
522
+ 4320 train 3.412507 (lr=3.0150e-04) (hash(x)=20493589)
523
+ 4330 train 3.491930 (lr=3.0101e-04) (hash(x)=18964922)
524
+ 4340 train 3.381837 (lr=3.0061e-04) (hash(x)=20832448)
525
+ 4350 train 3.394288 (lr=3.0031e-04) (hash(x)=18631764)
526
+ 4360 train 3.423446 (lr=3.0011e-04) (hash(x)=19542477)
527
+ 4370 train 3.468219 (lr=3.0001e-04) (hash(x)=17072278)
528
+ 4374 val loss 3.4099
529
+ 4374 val perplexity 30.2618
logs/64_latent_masks_learnable/learnable_compile_seed_1339/model_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1b57046ad46f3130c9fc3c0bb20ca6b99e2efc098c5fd3ae26707b92b4265a6d
3
+ size 555248882
logs/64_latent_masks_learnable/learnable_compile_seed_1339/optimizer_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:785f8f503a8b66386270d590375682079e9598a52d95e815a8e37581042d2730
3
+ size 1009847862