andrew-healey commited on
Commit
ead6859
·
verified ·
1 Parent(s): 7abd65e

Upload folder using huggingface_hub

Browse files
logs/fix_1_latent_mask/1_latent_mask_lr_35e-4_n_latent_masks_2_relu_seed_1340/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "logs/fix_1_latent_mask/1_latent_mask_lr_35e-4_n_latent_masks_2_relu_seed_1340", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 12, "n_embd": 264, "head_dim": 22, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": 4375, "warmup_steps": 250, "group": "fix_1_latent_mask", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1340, "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, "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": 32, "total_batch_size": 131072, "protection_head_scaling_factor": 1.0, "protection_head_bias": 0.0, "n_sliced_masks": null, "n_latent_masks": 2, "init_latent_masks_to_identity": true, "latent_mask_scale": null, "latent_mask_sigmoid": false, "S_layernorm": false, "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": 0.0035, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "1_latent_mask_lr_35e-4_n_latent_masks_2_relu"}
logs/fix_1_latent_mask/1_latent_mask_lr_35e-4_n_latent_masks_2_relu_seed_1340/dataloader_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6031fd3e2855a036f7a5531cc24555aabd1115f9dd6618b8b2ca6f55279ef0b2
3
+ size 964
logs/fix_1_latent_mask/1_latent_mask_lr_35e-4_n_latent_masks_2_relu_seed_1340/log2.txt ADDED
@@ -0,0 +1,529 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ max_steps: 4375
2
+ 0 val loss 11.2068
3
+ 0 val perplexity 73627.3516
4
+ 0 train 11.205775 (lr=4.8951e-06) (hash(x)=45482580)
5
+ 10 train 9.812622 (lr=5.3846e-05) (hash(x)=38414730)
6
+ 20 train 9.237215 (lr=1.0280e-04) (hash(x)=39783917)
7
+ 30 train 8.427091 (lr=1.5175e-04) (hash(x)=38395733)
8
+ 40 train 7.774748 (lr=2.0070e-04) (hash(x)=37486707)
9
+ 50 train 7.614930 (lr=2.4965e-04) (hash(x)=39897505)
10
+ 60 train 7.531648 (lr=2.9860e-04) (hash(x)=40933473)
11
+ 70 train 7.507036 (lr=3.4755e-04) (hash(x)=37690746)
12
+ 80 train 7.365547 (lr=3.9650e-04) (hash(x)=38609656)
13
+ 90 train 7.143466 (lr=4.4545e-04) (hash(x)=38356571)
14
+ 100 val loss 7.0334
15
+ 100 val perplexity 1133.8685
16
+ 100 train 7.015306 (lr=4.9441e-04) (hash(x)=35980376)
17
+ 110 train 6.898384 (lr=5.4336e-04) (hash(x)=42011042)
18
+ 120 train 6.809901 (lr=5.9231e-04) (hash(x)=40266823)
19
+ 130 train 6.661607 (lr=6.4126e-04) (hash(x)=38645447)
20
+ 140 train 6.534776 (lr=6.9021e-04) (hash(x)=36059313)
21
+ 150 train 6.488836 (lr=7.3916e-04) (hash(x)=34161947)
22
+ 160 train 6.374783 (lr=7.8811e-04) (hash(x)=42263375)
23
+ 170 train 6.422942 (lr=8.3706e-04) (hash(x)=42957725)
24
+ 180 train 6.423424 (lr=8.8601e-04) (hash(x)=40198018)
25
+ 190 train 6.158432 (lr=9.3497e-04) (hash(x)=41666215)
26
+ 200 val loss 6.1481
27
+ 200 val perplexity 467.8490
28
+ 200 train 6.107305 (lr=9.8392e-04) (hash(x)=54060482)
29
+ 210 train 6.007555 (lr=1.0329e-03) (hash(x)=42332778)
30
+ 220 train 5.963008 (lr=1.0818e-03) (hash(x)=39551486)
31
+ 230 train 5.962560 (lr=1.1308e-03) (hash(x)=38462018)
32
+ 240 train 6.012662 (lr=1.1797e-03) (hash(x)=36591442)
33
+ 250 train 5.965368 (lr=1.2287e-03) (hash(x)=40861237)
34
+ 260 train 5.898559 (lr=1.2776e-03) (hash(x)=41739369)
35
+ 270 train 5.765388 (lr=1.3266e-03) (hash(x)=39415360)
36
+ 280 train 5.752669 (lr=1.3755e-03) (hash(x)=44665934)
37
+ 290 train 5.708043 (lr=1.4245e-03) (hash(x)=37035579)
38
+ 300 val loss 5.6781
39
+ 300 val perplexity 292.3917
40
+ 300 train 5.700006 (lr=1.4734e-03) (hash(x)=38301011)
41
+ 310 train 5.537451 (lr=1.5224e-03) (hash(x)=42850980)
42
+ 320 train 5.620717 (lr=1.5713e-03) (hash(x)=37476222)
43
+ 330 train 5.532761 (lr=1.6203e-03) (hash(x)=53028205)
44
+ 340 train 5.538449 (lr=1.6692e-03) (hash(x)=41466008)
45
+ 350 train 5.450937 (lr=1.7182e-03) (hash(x)=37802865)
46
+ 360 train 5.395697 (lr=1.7671e-03) (hash(x)=42769282)
47
+ 370 train 5.427105 (lr=1.8161e-03) (hash(x)=39319256)
48
+ 380 train 5.407313 (lr=1.8650e-03) (hash(x)=42637402)
49
+ 390 train 5.383841 (lr=1.9140e-03) (hash(x)=31141514)
50
+ 400 val loss 5.3579
51
+ 400 val perplexity 212.2776
52
+ 400 train 5.437351 (lr=1.9629e-03) (hash(x)=38151157)
53
+ 410 train 5.370828 (lr=2.0119e-03) (hash(x)=46033439)
54
+ 420 train 5.345864 (lr=2.0608e-03) (hash(x)=41365246)
55
+ 430 train 5.328984 (lr=2.1098e-03) (hash(x)=42369184)
56
+ 440 train 5.322263 (lr=2.1587e-03) (hash(x)=42004840)
57
+ 450 train 5.247879 (lr=2.2077e-03) (hash(x)=37181172)
58
+ 460 train 5.338180 (lr=2.2566e-03) (hash(x)=31630797)
59
+ 470 train 5.196692 (lr=2.3056e-03) (hash(x)=42135747)
60
+ 480 train 5.123137 (lr=2.3545e-03) (hash(x)=30023651)
61
+ 490 train 5.212832 (lr=2.4035e-03) (hash(x)=40463476)
62
+ 500 val loss 5.1580
63
+ 500 val perplexity 173.8229
64
+ 500 train 5.080592 (lr=2.4524e-03) (hash(x)=37089842)
65
+ 510 train 5.093552 (lr=2.5014e-03) (hash(x)=41095974)
66
+ 520 train 5.141053 (lr=2.5503e-03) (hash(x)=41525258)
67
+ 530 train 5.108031 (lr=2.5993e-03) (hash(x)=37697291)
68
+ 540 train 5.154123 (lr=2.6483e-03) (hash(x)=43459894)
69
+ 550 train 5.000862 (lr=2.6972e-03) (hash(x)=38993956)
70
+ 560 train 5.035070 (lr=2.7462e-03) (hash(x)=40677277)
71
+ 570 train 5.103258 (lr=2.7951e-03) (hash(x)=43103127)
72
+ 580 train 5.054289 (lr=2.8441e-03) (hash(x)=41871776)
73
+ 590 train 5.007967 (lr=2.8930e-03) (hash(x)=43230967)
74
+ 600 val loss 4.9330
75
+ 600 val perplexity 138.7910
76
+ 600 train 4.977944 (lr=2.9420e-03) (hash(x)=42226172)
77
+ 610 train 5.291258 (lr=2.9909e-03) (hash(x)=43007164)
78
+ 620 train 4.861076 (lr=3.0399e-03) (hash(x)=35752341)
79
+ 630 train 4.928666 (lr=3.0888e-03) (hash(x)=39453504)
80
+ 640 train 4.863314 (lr=3.1378e-03) (hash(x)=43635712)
81
+ 650 train 4.872583 (lr=3.1867e-03) (hash(x)=41852456)
82
+ 660 train 4.888168 (lr=3.2357e-03) (hash(x)=49125742)
83
+ 670 train 4.899102 (lr=3.2846e-03) (hash(x)=34886275)
84
+ 680 train 4.979118 (lr=3.3336e-03) (hash(x)=40546482)
85
+ 690 train 4.786337 (lr=3.3825e-03) (hash(x)=40850901)
86
+ 700 val loss 4.7809
87
+ 700 val perplexity 119.2067
88
+ 700 train 4.694396 (lr=3.4315e-03) (hash(x)=37426993)
89
+ 710 train 4.925525 (lr=3.4804e-03) (hash(x)=42679743)
90
+ 720 train 4.744846 (lr=3.5000e-03) (hash(x)=44856043)
91
+ 730 train 4.741082 (lr=3.4999e-03) (hash(x)=36457471)
92
+ 740 train 4.695442 (lr=3.4996e-03) (hash(x)=47203534)
93
+ 750 train 4.717571 (lr=3.4993e-03) (hash(x)=35585657)
94
+ 760 train 4.631601 (lr=3.4988e-03) (hash(x)=37259264)
95
+ 770 train 4.540872 (lr=3.4982e-03) (hash(x)=44161997)
96
+ 780 train 4.489421 (lr=3.4975e-03) (hash(x)=41272886)
97
+ 790 train 4.496560 (lr=3.4967e-03) (hash(x)=43303662)
98
+ 800 val loss 4.6905
99
+ 800 val perplexity 108.9091
100
+ 800 train 4.403023 (lr=3.4958e-03) (hash(x)=39067231)
101
+ 810 train 4.641858 (lr=3.4948e-03) (hash(x)=44046732)
102
+ 820 train 4.628161 (lr=3.4936e-03) (hash(x)=39783962)
103
+ 830 train 4.678962 (lr=3.4923e-03) (hash(x)=40973974)
104
+ 840 train 4.656819 (lr=3.4909e-03) (hash(x)=41619968)
105
+ 850 train 4.712511 (lr=3.4894e-03) (hash(x)=36939960)
106
+ 860 train 4.583665 (lr=3.4878e-03) (hash(x)=42942377)
107
+ 870 train 4.592511 (lr=3.4861e-03) (hash(x)=39659455)
108
+ 880 train 4.607811 (lr=3.4842e-03) (hash(x)=40331986)
109
+ 890 train 4.520461 (lr=3.4823e-03) (hash(x)=38084814)
110
+ 900 val loss 4.5604
111
+ 900 val perplexity 95.6255
112
+ 900 train 4.549951 (lr=3.4802e-03) (hash(x)=30830367)
113
+ 910 train 4.463198 (lr=3.4780e-03) (hash(x)=39664356)
114
+ 920 train 4.469388 (lr=3.4757e-03) (hash(x)=39007775)
115
+ 930 train 4.491780 (lr=3.4733e-03) (hash(x)=39319254)
116
+ 940 train 4.544484 (lr=3.4707e-03) (hash(x)=36347051)
117
+ 950 train 4.497489 (lr=3.4681e-03) (hash(x)=39243577)
118
+ 960 train 4.515059 (lr=3.4653e-03) (hash(x)=45603839)
119
+ 970 train 4.326014 (lr=3.4624e-03) (hash(x)=40036075)
120
+ 980 train 4.373411 (lr=3.4594e-03) (hash(x)=43068524)
121
+ 990 train 4.361378 (lr=3.4563e-03) (hash(x)=33798472)
122
+ 1000 val loss 4.5313
123
+ 1000 val perplexity 92.8756
124
+ 1000 train 4.745913 (lr=3.4531e-03) (hash(x)=37728665)
125
+ 1010 train 4.588437 (lr=3.4498e-03) (hash(x)=43678636)
126
+ 1020 train 4.620122 (lr=3.4463e-03) (hash(x)=48464376)
127
+ 1030 train 4.581843 (lr=3.4428e-03) (hash(x)=43771461)
128
+ 1040 train 4.444706 (lr=3.4391e-03) (hash(x)=44643085)
129
+ 1050 train 4.531262 (lr=3.4353e-03) (hash(x)=39201716)
130
+ 1060 train 4.434829 (lr=3.4314e-03) (hash(x)=36396343)
131
+ 1070 train 4.424603 (lr=3.4274e-03) (hash(x)=37504111)
132
+ 1080 train 4.464855 (lr=3.4233e-03) (hash(x)=37454537)
133
+ 1090 train 4.457646 (lr=3.4191e-03) (hash(x)=36888183)
134
+ 1100 val loss 4.4468
135
+ 1100 val perplexity 85.3528
136
+ 1100 train 4.407601 (lr=3.4148e-03) (hash(x)=39549045)
137
+ 1110 train 4.592062 (lr=3.4103e-03) (hash(x)=56818420)
138
+ 1120 train 4.397832 (lr=3.4058e-03) (hash(x)=36573357)
139
+ 1130 train 4.335546 (lr=3.4011e-03) (hash(x)=38540878)
140
+ 1140 train 4.327809 (lr=3.3964e-03) (hash(x)=38645859)
141
+ 1150 train 4.537948 (lr=3.3915e-03) (hash(x)=41972503)
142
+ 1160 train 4.156209 (lr=3.3865e-03) (hash(x)=38151868)
143
+ 1170 train 4.191444 (lr=3.3814e-03) (hash(x)=42375436)
144
+ 1180 train 4.171408 (lr=3.3762e-03) (hash(x)=42868028)
145
+ 1190 train 4.419675 (lr=3.3709e-03) (hash(x)=41060868)
146
+ 1200 val loss 4.4126
147
+ 1200 val perplexity 82.4863
148
+ 1200 train 4.428247 (lr=3.3655e-03) (hash(x)=40852814)
149
+ 1210 train 4.406734 (lr=3.3600e-03) (hash(x)=43365554)
150
+ 1220 train 4.347735 (lr=3.3543e-03) (hash(x)=39121134)
151
+ 1230 train 4.435970 (lr=3.3486e-03) (hash(x)=40565426)
152
+ 1240 train 4.493879 (lr=3.3428e-03) (hash(x)=43378926)
153
+ 1250 train 4.355874 (lr=3.3368e-03) (hash(x)=35227381)
154
+ 1260 train 4.446244 (lr=3.3308e-03) (hash(x)=41267226)
155
+ 1270 train 4.355770 (lr=3.3246e-03) (hash(x)=31395565)
156
+ 1280 train 4.471599 (lr=3.3184e-03) (hash(x)=41470963)
157
+ 1290 train 4.285622 (lr=3.3120e-03) (hash(x)=40729366)
158
+ 1300 val loss 4.3484
159
+ 1300 val perplexity 77.3522
160
+ 1300 train 4.321798 (lr=3.3056e-03) (hash(x)=39916663)
161
+ 1310 train 4.374721 (lr=3.2990e-03) (hash(x)=42179962)
162
+ 1320 train 4.239704 (lr=3.2924e-03) (hash(x)=44991932)
163
+ 1330 train 4.210152 (lr=3.2856e-03) (hash(x)=38925911)
164
+ 1340 train 4.155388 (lr=3.2787e-03) (hash(x)=38565560)
165
+ 1350 train 4.209158 (lr=3.2718e-03) (hash(x)=43384902)
166
+ 1360 train 4.152629 (lr=3.2647e-03) (hash(x)=45422261)
167
+ 1370 train 4.139527 (lr=3.2576e-03) (hash(x)=40167325)
168
+ 1380 train 4.295269 (lr=3.2503e-03) (hash(x)=51258162)
169
+ 1390 train 4.353024 (lr=3.2430e-03) (hash(x)=43229106)
170
+ 1400 val loss 4.3122
171
+ 1400 val perplexity 74.6041
172
+ 1400 train 4.259473 (lr=3.2355e-03) (hash(x)=42537494)
173
+ 1410 train 4.560262 (lr=3.2280e-03) (hash(x)=51236490)
174
+ 1420 train 4.408238 (lr=3.2203e-03) (hash(x)=45243347)
175
+ 1430 train 4.247519 (lr=3.2126e-03) (hash(x)=37658566)
176
+ 1440 train 4.315698 (lr=3.2047e-03) (hash(x)=38856106)
177
+ 1450 train 4.382265 (lr=3.1968e-03) (hash(x)=36649488)
178
+ 1460 train 4.375321 (lr=3.1888e-03) (hash(x)=40445407)
179
+ 1470 train 4.308889 (lr=3.1807e-03) (hash(x)=39916387)
180
+ 1480 train 4.277141 (lr=3.1725e-03) (hash(x)=40934555)
181
+ 1490 train 4.243225 (lr=3.1642e-03) (hash(x)=42276527)
182
+ 1500 val loss 4.2946
183
+ 1500 val perplexity 73.3034
184
+ 1500 train 4.233893 (lr=3.1558e-03) (hash(x)=37973883)
185
+ 1510 train 4.292256 (lr=3.1473e-03) (hash(x)=38929286)
186
+ 1520 train 4.112845 (lr=3.1387e-03) (hash(x)=36855837)
187
+ 1530 train 4.356359 (lr=3.1301e-03) (hash(x)=42146431)
188
+ 1540 train 4.305826 (lr=3.1213e-03) (hash(x)=41898411)
189
+ 1550 train 4.212687 (lr=3.1125e-03) (hash(x)=44906272)
190
+ 1560 train 4.239693 (lr=3.1036e-03) (hash(x)=40414353)
191
+ 1570 train 4.267247 (lr=3.0945e-03) (hash(x)=38723190)
192
+ 1580 train 4.182014 (lr=3.0855e-03) (hash(x)=39065271)
193
+ 1590 train 4.180131 (lr=3.0763e-03) (hash(x)=40000886)
194
+ 1600 val loss 4.2684
195
+ 1600 val perplexity 71.4058
196
+ 1600 train 4.102335 (lr=3.0670e-03) (hash(x)=37898571)
197
+ 1610 train 4.211408 (lr=3.0576e-03) (hash(x)=40447863)
198
+ 1620 train 4.178650 (lr=3.0482e-03) (hash(x)=40119318)
199
+ 1630 train 4.111410 (lr=3.0387e-03) (hash(x)=40664074)
200
+ 1640 train 4.074966 (lr=3.0291e-03) (hash(x)=37404741)
201
+ 1650 train 4.232429 (lr=3.0194e-03) (hash(x)=39214928)
202
+ 1660 train 4.272889 (lr=3.0096e-03) (hash(x)=35005019)
203
+ 1670 train 4.332964 (lr=2.9998e-03) (hash(x)=37675832)
204
+ 1680 train 4.254737 (lr=2.9899e-03) (hash(x)=40489680)
205
+ 1690 train 4.166017 (lr=2.9799e-03) (hash(x)=45537879)
206
+ 1700 val loss 4.2323
207
+ 1700 val perplexity 68.8767
208
+ 1700 train 4.150029 (lr=2.9698e-03) (hash(x)=38866100)
209
+ 1710 train 4.216749 (lr=2.9597e-03) (hash(x)=44726254)
210
+ 1720 train 4.185137 (lr=2.9494e-03) (hash(x)=36917792)
211
+ 1730 train 4.173769 (lr=2.9391e-03) (hash(x)=42814805)
212
+ 1740 train 4.248742 (lr=2.9287e-03) (hash(x)=51812216)
213
+ 1750 train 4.120064 (lr=2.9183e-03) (hash(x)=33549014)
214
+ 1760 train 4.130404 (lr=2.9078e-03) (hash(x)=40354215)
215
+ 1770 train 4.151069 (lr=2.8972e-03) (hash(x)=40718606)
216
+ 1780 train 4.084233 (lr=2.8865e-03) (hash(x)=38274164)
217
+ 1790 train 4.267813 (lr=2.8757e-03) (hash(x)=42531471)
218
+ 1800 val loss 4.2125
219
+ 1800 val perplexity 67.5274
220
+ 1800 train 4.191724 (lr=2.8649e-03) (hash(x)=35616519)
221
+ 1810 train 4.244655 (lr=2.8540e-03) (hash(x)=33803118)
222
+ 1820 train 4.247530 (lr=2.8431e-03) (hash(x)=36973525)
223
+ 1830 train 4.189011 (lr=2.8321e-03) (hash(x)=38162549)
224
+ 1840 train 4.155122 (lr=2.8210e-03) (hash(x)=33937159)
225
+ 1850 train 4.263182 (lr=2.8098e-03) (hash(x)=39887546)
226
+ 1860 train 4.113124 (lr=2.7986e-03) (hash(x)=37818525)
227
+ 1870 train 4.151896 (lr=2.7873e-03) (hash(x)=37250478)
228
+ 1880 train 4.043389 (lr=2.7760e-03) (hash(x)=36412167)
229
+ 1890 train 4.156949 (lr=2.7646e-03) (hash(x)=40261189)
230
+ 1900 val loss 4.2104
231
+ 1900 val perplexity 67.3865
232
+ 1900 train 4.111433 (lr=2.7531e-03) (hash(x)=38654303)
233
+ 1910 train 4.085342 (lr=2.7416e-03) (hash(x)=37929515)
234
+ 1920 train 4.257605 (lr=2.7300e-03) (hash(x)=45580146)
235
+ 1930 train 4.209320 (lr=2.7183e-03) (hash(x)=32386330)
236
+ 1940 train 4.244729 (lr=2.7066e-03) (hash(x)=36331864)
237
+ 1950 train 4.215901 (lr=2.6949e-03) (hash(x)=41181727)
238
+ 1960 train 4.049536 (lr=2.6831e-03) (hash(x)=42705152)
239
+ 1970 train 4.156674 (lr=2.6712e-03) (hash(x)=40251511)
240
+ 1980 train 4.079643 (lr=2.6592e-03) (hash(x)=37525551)
241
+ 1990 train 4.149073 (lr=2.6473e-03) (hash(x)=41849618)
242
+ 2000 val loss 4.1733
243
+ 2000 val perplexity 64.9299
244
+ 2000 train 4.151613 (lr=2.6352e-03) (hash(x)=37642582)
245
+ 2010 train 4.154629 (lr=2.6231e-03) (hash(x)=42131121)
246
+ 2020 train 4.033173 (lr=2.6110e-03) (hash(x)=39000209)
247
+ 2030 train 4.128172 (lr=2.5988e-03) (hash(x)=43641355)
248
+ 2040 train 4.019113 (lr=2.5865e-03) (hash(x)=47910507)
249
+ 2050 train 4.022480 (lr=2.5742e-03) (hash(x)=36670359)
250
+ 2060 train 4.220894 (lr=2.5619e-03) (hash(x)=36477755)
251
+ 2070 train 4.310336 (lr=2.5495e-03) (hash(x)=43262487)
252
+ 2080 train 4.121815 (lr=2.5371e-03) (hash(x)=49546029)
253
+ 2090 train 4.251645 (lr=2.5246e-03) (hash(x)=39924731)
254
+ 2100 val loss 4.1639
255
+ 2100 val perplexity 64.3237
256
+ 2100 train 4.115305 (lr=2.5121e-03) (hash(x)=39921304)
257
+ 2110 train 4.062537 (lr=2.4995e-03) (hash(x)=38152788)
258
+ 2120 train 4.116332 (lr=2.4869e-03) (hash(x)=37977911)
259
+ 2130 train 4.077581 (lr=2.4743e-03) (hash(x)=34748760)
260
+ 2140 train 4.132834 (lr=2.4616e-03) (hash(x)=40509369)
261
+ 2150 train 4.103448 (lr=2.4488e-03) (hash(x)=37654262)
262
+ 2160 train 4.053531 (lr=2.4361e-03) (hash(x)=38139543)
263
+ 2170 train 3.965717 (lr=2.4233e-03) (hash(x)=42501806)
264
+ 2180 train 4.118404 (lr=2.4104e-03) (hash(x)=40085092)
265
+ 2190 train 4.057285 (lr=2.3975e-03) (hash(x)=50480193)
266
+ 2200 val loss 4.1444
267
+ 2200 val perplexity 63.0806
268
+ 2200 train 4.150290 (lr=2.3846e-03) (hash(x)=40604084)
269
+ 2210 train 4.253760 (lr=2.3717e-03) (hash(x)=41555823)
270
+ 2220 train 4.074476 (lr=2.3587e-03) (hash(x)=50441765)
271
+ 2230 train 4.179893 (lr=2.3457e-03) (hash(x)=39796580)
272
+ 2240 train 4.198711 (lr=2.3326e-03) (hash(x)=44127022)
273
+ 2250 train 4.100562 (lr=2.3196e-03) (hash(x)=37026826)
274
+ 2260 train 4.136966 (lr=2.3065e-03) (hash(x)=42133839)
275
+ 2270 train 4.098239 (lr=2.2933e-03) (hash(x)=38500664)
276
+ 2280 train 4.085431 (lr=2.2802e-03) (hash(x)=40538661)
277
+ 2290 train 4.093563 (lr=2.2670e-03) (hash(x)=51509210)
278
+ 2300 val loss 4.1201
279
+ 2300 val perplexity 61.5679
280
+ 2300 train 4.097082 (lr=2.2538e-03) (hash(x)=41952328)
281
+ 2310 train 4.018141 (lr=2.2405e-03) (hash(x)=39758123)
282
+ 2320 train 4.136131 (lr=2.2273e-03) (hash(x)=51089268)
283
+ 2330 train 4.193342 (lr=2.2140e-03) (hash(x)=39767618)
284
+ 2340 train 4.241555 (lr=2.2007e-03) (hash(x)=40409617)
285
+ 2350 train 4.069928 (lr=2.1874e-03) (hash(x)=40349634)
286
+ 2360 train 4.345746 (lr=2.1741e-03) (hash(x)=31841172)
287
+ 2370 train 4.137538 (lr=2.1607e-03) (hash(x)=42720539)
288
+ 2380 train 4.008965 (lr=2.1473e-03) (hash(x)=40998632)
289
+ 2390 train 4.111957 (lr=2.1339e-03) (hash(x)=40615413)
290
+ 2400 val loss 4.1032
291
+ 2400 val perplexity 60.5339
292
+ 2400 train 4.021462 (lr=2.1205e-03) (hash(x)=39373658)
293
+ 2410 train 4.071949 (lr=2.1071e-03) (hash(x)=35480858)
294
+ 2420 train 4.061480 (lr=2.0937e-03) (hash(x)=32241095)
295
+ 2430 train 4.049626 (lr=2.0802e-03) (hash(x)=36669715)
296
+ 2440 train 4.051574 (lr=2.0668e-03) (hash(x)=45768335)
297
+ 2450 train 3.954646 (lr=2.0533e-03) (hash(x)=39395055)
298
+ 2460 train 3.982907 (lr=2.0398e-03) (hash(x)=34899269)
299
+ 2470 train 4.171450 (lr=2.0263e-03) (hash(x)=43642420)
300
+ 2480 train 4.278519 (lr=2.0128e-03) (hash(x)=47099507)
301
+ 2490 train 4.148385 (lr=1.9993e-03) (hash(x)=38825558)
302
+ 2500 val loss 4.0916
303
+ 2500 val perplexity 59.8332
304
+ 2500 train 4.120988 (lr=1.9858e-03) (hash(x)=39833804)
305
+ 2510 train 4.219458 (lr=1.9723e-03) (hash(x)=35051654)
306
+ 2520 train 4.115241 (lr=1.9588e-03) (hash(x)=40291109)
307
+ 2530 train 4.025075 (lr=1.9453e-03) (hash(x)=36915768)
308
+ 2540 train 4.005026 (lr=1.9318e-03) (hash(x)=37016308)
309
+ 2550 train 3.973428 (lr=1.9182e-03) (hash(x)=43221777)
310
+ 2560 train 4.015605 (lr=1.9047e-03) (hash(x)=37233207)
311
+ 2570 train 3.918487 (lr=1.8912e-03) (hash(x)=37510626)
312
+ 2580 train 3.968830 (lr=1.8777e-03) (hash(x)=39624656)
313
+ 2590 train 3.822659 (lr=1.8642e-03) (hash(x)=37647501)
314
+ 2600 val loss 4.0789
315
+ 2600 val perplexity 59.0775
316
+ 2600 train 3.958989 (lr=1.8507e-03) (hash(x)=37200138)
317
+ 2610 train 3.966519 (lr=1.8372e-03) (hash(x)=43066970)
318
+ 2620 train 4.111077 (lr=1.8237e-03) (hash(x)=40503799)
319
+ 2630 train 4.045216 (lr=1.8102e-03) (hash(x)=45534660)
320
+ 2640 train 4.258453 (lr=1.7967e-03) (hash(x)=52358353)
321
+ 2650 train 4.040204 (lr=1.7832e-03) (hash(x)=41843599)
322
+ 2660 train 4.029054 (lr=1.7698e-03) (hash(x)=40519770)
323
+ 2670 train 4.106740 (lr=1.7563e-03) (hash(x)=37185602)
324
+ 2680 train 3.970518 (lr=1.7429e-03) (hash(x)=43805719)
325
+ 2690 train 3.980217 (lr=1.7295e-03) (hash(x)=39458250)
326
+ 2700 val loss 4.0570
327
+ 2700 val perplexity 57.7993
328
+ 2700 train 4.082903 (lr=1.7161e-03) (hash(x)=39149255)
329
+ 2710 train 3.923691 (lr=1.7027e-03) (hash(x)=43544861)
330
+ 2720 train 4.027200 (lr=1.6893e-03) (hash(x)=36858197)
331
+ 2730 train 3.943838 (lr=1.6759e-03) (hash(x)=37971712)
332
+ 2740 train 3.975592 (lr=1.6626e-03) (hash(x)=42416578)
333
+ 2750 train 3.916022 (lr=1.6493e-03) (hash(x)=53554622)
334
+ 2760 train 4.054662 (lr=1.6360e-03) (hash(x)=40502545)
335
+ 2770 train 4.021605 (lr=1.6227e-03) (hash(x)=36086417)
336
+ 2780 train 4.051232 (lr=1.6095e-03) (hash(x)=36242982)
337
+ 2790 train 4.123536 (lr=1.5962e-03) (hash(x)=41433780)
338
+ 2800 val loss 4.0386
339
+ 2800 val perplexity 56.7442
340
+ 2800 train 3.993366 (lr=1.5830e-03) (hash(x)=48553484)
341
+ 2810 train 3.928988 (lr=1.5698e-03) (hash(x)=39928650)
342
+ 2820 train 4.041216 (lr=1.5567e-03) (hash(x)=37891724)
343
+ 2830 train 4.014503 (lr=1.5435e-03) (hash(x)=34353412)
344
+ 2840 train 4.035425 (lr=1.5304e-03) (hash(x)=38486611)
345
+ 2850 train 3.957497 (lr=1.5174e-03) (hash(x)=32706934)
346
+ 2860 train 3.903842 (lr=1.5043e-03) (hash(x)=38693681)
347
+ 2870 train 3.972688 (lr=1.4913e-03) (hash(x)=38430800)
348
+ 2880 train 3.943854 (lr=1.4783e-03) (hash(x)=40619559)
349
+ 2890 train 3.919004 (lr=1.4654e-03) (hash(x)=42741066)
350
+ 2900 val loss 4.0372
351
+ 2900 val perplexity 56.6681
352
+ 2900 train 4.174565 (lr=1.4525e-03) (hash(x)=38216091)
353
+ 2910 train 3.973908 (lr=1.4396e-03) (hash(x)=35447832)
354
+ 2920 train 4.082532 (lr=1.4267e-03) (hash(x)=39825190)
355
+ 2930 train 4.100871 (lr=1.4139e-03) (hash(x)=37789121)
356
+ 2940 train 3.975403 (lr=1.4012e-03) (hash(x)=44499116)
357
+ 2950 train 4.058673 (lr=1.3884e-03) (hash(x)=39669860)
358
+ 2960 train 4.007223 (lr=1.3757e-03) (hash(x)=51683741)
359
+ 2970 train 3.866876 (lr=1.3631e-03) (hash(x)=36399721)
360
+ 2980 train 3.964612 (lr=1.3505e-03) (hash(x)=42629700)
361
+ 2990 train 3.851575 (lr=1.3379e-03) (hash(x)=39263773)
362
+ 3000 val loss 4.0128
363
+ 3000 val perplexity 55.3023
364
+ 3000 train 3.891414 (lr=1.3254e-03) (hash(x)=37920485)
365
+ 3010 train 3.935718 (lr=1.3129e-03) (hash(x)=40835161)
366
+ 3020 train 3.909231 (lr=1.3005e-03) (hash(x)=38381159)
367
+ 3030 train 3.895433 (lr=1.2881e-03) (hash(x)=35846270)
368
+ 3040 train 4.286858 (lr=1.2758e-03) (hash(x)=47516567)
369
+ 3050 train 4.126719 (lr=1.2635e-03) (hash(x)=35600311)
370
+ 3060 train 4.159420 (lr=1.2512e-03) (hash(x)=37775318)
371
+ 3070 train 4.015432 (lr=1.2390e-03) (hash(x)=39881333)
372
+ 3080 train 4.014544 (lr=1.2269e-03) (hash(x)=37310168)
373
+ 3090 train 3.942350 (lr=1.2148e-03) (hash(x)=43669978)
374
+ 3100 val loss 4.0025
375
+ 3100 val perplexity 54.7323
376
+ 3100 train 4.405770 (lr=1.2027e-03) (hash(x)=46356797)
377
+ 3110 train 4.041949 (lr=1.1908e-03) (hash(x)=40814232)
378
+ 3120 train 3.886579 (lr=1.1788e-03) (hash(x)=42504837)
379
+ 3130 train 4.022605 (lr=1.1669e-03) (hash(x)=39693594)
380
+ 3140 train 3.925365 (lr=1.1551e-03) (hash(x)=40277645)
381
+ 3150 train 3.932368 (lr=1.1434e-03) (hash(x)=40988003)
382
+ 3160 train 4.053073 (lr=1.1317e-03) (hash(x)=38875266)
383
+ 3170 train 4.007734 (lr=1.1200e-03) (hash(x)=45232173)
384
+ 3180 train 4.060878 (lr=1.1084e-03) (hash(x)=39213336)
385
+ 3190 train 4.119089 (lr=1.0969e-03) (hash(x)=42118576)
386
+ 3200 val loss 3.9791
387
+ 3200 val perplexity 53.4682
388
+ 3200 train 3.972818 (lr=1.0854e-03) (hash(x)=32884223)
389
+ 3210 train 4.024240 (lr=1.0740e-03) (hash(x)=41276800)
390
+ 3220 train 4.067716 (lr=1.0627e-03) (hash(x)=40284461)
391
+ 3230 train 4.049751 (lr=1.0514e-03) (hash(x)=40566734)
392
+ 3240 train 4.035659 (lr=1.0402e-03) (hash(x)=36484570)
393
+ 3250 train 3.999547 (lr=1.0290e-03) (hash(x)=41642338)
394
+ 3260 train 3.963212 (lr=1.0179e-03) (hash(x)=43883570)
395
+ 3270 train 3.903450 (lr=1.0069e-03) (hash(x)=40432560)
396
+ 3280 train 3.876330 (lr=9.9596e-04) (hash(x)=38002717)
397
+ 3290 train 3.821010 (lr=9.8508e-04) (hash(x)=41926004)
398
+ 3300 val loss 3.9716
399
+ 3300 val perplexity 53.0707
400
+ 3300 train 3.974936 (lr=9.7426e-04) (hash(x)=46015509)
401
+ 3310 train 4.015765 (lr=9.6352e-04) (hash(x)=40112249)
402
+ 3320 train 3.986632 (lr=9.5285e-04) (hash(x)=49162296)
403
+ 3330 train 4.044510 (lr=9.4225e-04) (hash(x)=46606969)
404
+ 3340 train 4.062079 (lr=9.3172e-04) (hash(x)=31128992)
405
+ 3350 train 4.036312 (lr=9.2127e-04) (hash(x)=41232534)
406
+ 3360 train 3.880438 (lr=9.1088e-04) (hash(x)=41599699)
407
+ 3370 train 4.056550 (lr=9.0058e-04) (hash(x)=40885280)
408
+ 3380 train 3.967731 (lr=8.9035e-04) (hash(x)=37169148)
409
+ 3390 train 3.873164 (lr=8.8019e-04) (hash(x)=46581889)
410
+ 3400 val loss 3.9594
411
+ 3400 val perplexity 52.4245
412
+ 3400 train 3.926034 (lr=8.7012e-04) (hash(x)=37612074)
413
+ 3410 train 3.993042 (lr=8.6012e-04) (hash(x)=38763316)
414
+ 3420 train 3.819119 (lr=8.5019e-04) (hash(x)=40562379)
415
+ 3430 train 3.919525 (lr=8.4035e-04) (hash(x)=44724867)
416
+ 3440 train 4.129735 (lr=8.3059e-04) (hash(x)=39386624)
417
+ 3450 train 4.079697 (lr=8.2091e-04) (hash(x)=38652923)
418
+ 3460 train 4.054099 (lr=8.1131e-04) (hash(x)=29587379)
419
+ 3470 train 3.962681 (lr=8.0179e-04) (hash(x)=38228776)
420
+ 3480 train 3.934667 (lr=7.9235e-04) (hash(x)=30076039)
421
+ 3490 train 3.996388 (lr=7.8300e-04) (hash(x)=38398908)
422
+ 3500 val loss 3.9452
423
+ 3500 val perplexity 51.6862
424
+ 3500 train 4.015450 (lr=7.7373e-04) (hash(x)=39259918)
425
+ 3510 train 3.911720 (lr=7.6455e-04) (hash(x)=40664091)
426
+ 3520 train 3.947352 (lr=7.5545e-04) (hash(x)=40309647)
427
+ 3530 train 3.912250 (lr=7.4644e-04) (hash(x)=38424801)
428
+ 3540 train 3.835216 (lr=7.3752e-04) (hash(x)=51322307)
429
+ 3550 train 3.899206 (lr=7.2868e-04) (hash(x)=38192628)
430
+ 3560 train 3.952981 (lr=7.1993e-04) (hash(x)=41563952)
431
+ 3570 train 3.905560 (lr=7.1127e-04) (hash(x)=39508843)
432
+ 3580 train 3.974339 (lr=7.0270e-04) (hash(x)=41260225)
433
+ 3590 train 3.955137 (lr=6.9422e-04) (hash(x)=49098107)
434
+ 3600 val loss 3.9348
435
+ 3600 val perplexity 51.1506
436
+ 3600 train 3.936163 (lr=6.8583e-04) (hash(x)=41194370)
437
+ 3610 train 4.055598 (lr=6.7753e-04) (hash(x)=40399152)
438
+ 3620 train 4.023113 (lr=6.6932e-04) (hash(x)=41625018)
439
+ 3630 train 4.042124 (lr=6.6121e-04) (hash(x)=44077942)
440
+ 3640 train 4.038013 (lr=6.5319e-04) (hash(x)=42427512)
441
+ 3650 train 4.021367 (lr=6.4526e-04) (hash(x)=40067454)
442
+ 3660 train 3.946500 (lr=6.3743e-04) (hash(x)=37969892)
443
+ 3670 train 3.916813 (lr=6.2969e-04) (hash(x)=43670246)
444
+ 3680 train 4.000622 (lr=6.2204e-04) (hash(x)=37031732)
445
+ 3690 train 3.871086 (lr=6.1450e-04) (hash(x)=39237187)
446
+ 3700 val loss 3.9280
447
+ 3700 val perplexity 50.8066
448
+ 3700 train 3.909700 (lr=6.0705e-04) (hash(x)=39521416)
449
+ 3710 train 3.807949 (lr=5.9969e-04) (hash(x)=41147425)
450
+ 3720 train 3.802788 (lr=5.9244e-04) (hash(x)=40204658)
451
+ 3730 train 3.895840 (lr=5.8528e-04) (hash(x)=39420983)
452
+ 3740 train 3.940556 (lr=5.7822e-04) (hash(x)=51155740)
453
+ 3750 train 3.929780 (lr=5.7126e-04) (hash(x)=41654187)
454
+ 3760 train 3.886291 (lr=5.6440e-04) (hash(x)=40598339)
455
+ 3770 train 3.851002 (lr=5.5764e-04) (hash(x)=44950116)
456
+ 3780 train 3.912969 (lr=5.5098e-04) (hash(x)=40660326)
457
+ 3790 train 3.958812 (lr=5.4443e-04) (hash(x)=37805851)
458
+ 3800 val loss 3.9113
459
+ 3800 val perplexity 49.9619
460
+ 3800 train 3.925322 (lr=5.3797e-04) (hash(x)=38064443)
461
+ 3810 train 3.999931 (lr=5.3162e-04) (hash(x)=39161498)
462
+ 3820 train 3.845822 (lr=5.2537e-04) (hash(x)=40626649)
463
+ 3830 train 3.919757 (lr=5.1922e-04) (hash(x)=36894771)
464
+ 3840 train 3.864330 (lr=5.1317e-04) (hash(x)=33672141)
465
+ 3850 train 3.819717 (lr=5.0723e-04) (hash(x)=41751813)
466
+ 3860 train 3.691035 (lr=5.0140e-04) (hash(x)=33784172)
467
+ 3870 train 3.623286 (lr=4.9567e-04) (hash(x)=43110493)
468
+ 3880 train 3.723695 (lr=4.9004e-04) (hash(x)=36585111)
469
+ 3890 train 3.728099 (lr=4.8452e-04) (hash(x)=40134264)
470
+ 3900 val loss 3.9178
471
+ 3900 val perplexity 50.2886
472
+ 3900 train 3.723025 (lr=4.7911e-04) (hash(x)=39815215)
473
+ 3910 train 3.862629 (lr=4.7380e-04) (hash(x)=42081557)
474
+ 3920 train 3.798627 (lr=4.6860e-04) (hash(x)=48184973)
475
+ 3930 train 3.994176 (lr=4.6351e-04) (hash(x)=42416681)
476
+ 3940 train 3.913314 (lr=4.5852e-04) (hash(x)=38587062)
477
+ 3950 train 3.914093 (lr=4.5364e-04) (hash(x)=37296617)
478
+ 3960 train 3.884839 (lr=4.4888e-04) (hash(x)=37959330)
479
+ 3970 train 3.915361 (lr=4.4422e-04) (hash(x)=35175634)
480
+ 3980 train 3.962470 (lr=4.3966e-04) (hash(x)=37219362)
481
+ 3990 train 3.988790 (lr=4.3522e-04) (hash(x)=38906132)
482
+ 4000 val loss 3.8949
483
+ 4000 val perplexity 49.1529
484
+ 4000 train 3.907897 (lr=4.3089e-04) (hash(x)=39940517)
485
+ 4010 train 3.916746 (lr=4.2667e-04) (hash(x)=39952437)
486
+ 4020 train 3.837174 (lr=4.2256e-04) (hash(x)=42000930)
487
+ 4030 train 3.848779 (lr=4.1856e-04) (hash(x)=41460672)
488
+ 4040 train 3.596305 (lr=4.1467e-04) (hash(x)=43702329)
489
+ 4050 train 3.687787 (lr=4.1089e-04) (hash(x)=41740121)
490
+ 4060 train 3.688763 (lr=4.0722e-04) (hash(x)=37066349)
491
+ 4070 train 3.528212 (lr=4.0367e-04) (hash(x)=38359869)
492
+ 4080 train 3.774959 (lr=4.0022e-04) (hash(x)=35131370)
493
+ 4090 train 3.835017 (lr=3.9689e-04) (hash(x)=39918811)
494
+ 4100 val loss 3.8943
495
+ 4100 val perplexity 49.1206
496
+ 4100 train 3.991530 (lr=3.9368e-04) (hash(x)=47036374)
497
+ 4110 train 3.918674 (lr=3.9057e-04) (hash(x)=37657523)
498
+ 4120 train 3.958759 (lr=3.8758e-04) (hash(x)=37218703)
499
+ 4130 train 3.977319 (lr=3.8470e-04) (hash(x)=41350513)
500
+ 4140 train 3.997030 (lr=3.8193e-04) (hash(x)=38283785)
501
+ 4150 train 3.976682 (lr=3.7928e-04) (hash(x)=32927892)
502
+ 4160 train 3.863143 (lr=3.7674e-04) (hash(x)=42011933)
503
+ 4170 train 3.882438 (lr=3.7432e-04) (hash(x)=41854594)
504
+ 4180 train 3.888705 (lr=3.7201e-04) (hash(x)=36737064)
505
+ 4190 train 3.803057 (lr=3.6982e-04) (hash(x)=36326176)
506
+ 4200 val loss 3.8857
507
+ 4200 val perplexity 48.7026
508
+ 4200 train 3.869851 (lr=3.6774e-04) (hash(x)=39106683)
509
+ 4210 train 3.924919 (lr=3.6577e-04) (hash(x)=39759977)
510
+ 4220 train 3.665833 (lr=3.6392e-04) (hash(x)=43666151)
511
+ 4230 train 3.615098 (lr=3.6218e-04) (hash(x)=38154367)
512
+ 4240 train 3.723154 (lr=3.6056e-04) (hash(x)=38284168)
513
+ 4250 train 3.654698 (lr=3.5906e-04) (hash(x)=38888879)
514
+ 4260 train 3.572348 (lr=3.5767e-04) (hash(x)=36846417)
515
+ 4270 train 3.979763 (lr=3.5639e-04) (hash(x)=43125375)
516
+ 4280 train 3.936189 (lr=3.5523e-04) (hash(x)=43556412)
517
+ 4290 train 3.866328 (lr=3.5419e-04) (hash(x)=48598302)
518
+ 4300 val loss 3.8806
519
+ 4300 val perplexity 48.4523
520
+ 4300 train 3.850849 (lr=3.5326e-04) (hash(x)=42764385)
521
+ 4310 train 3.906744 (lr=3.5245e-04) (hash(x)=41190028)
522
+ 4320 train 3.870403 (lr=3.5175e-04) (hash(x)=35767770)
523
+ 4330 train 3.948445 (lr=3.5117e-04) (hash(x)=36760476)
524
+ 4340 train 3.891533 (lr=3.5071e-04) (hash(x)=40618921)
525
+ 4350 train 3.991795 (lr=3.5036e-04) (hash(x)=32300164)
526
+ 4360 train 3.848344 (lr=3.5013e-04) (hash(x)=41400085)
527
+ 4370 train 3.881461 (lr=3.5001e-04) (hash(x)=40251828)
528
+ 4374 val loss 3.8816
529
+ 4374 val perplexity 48.5030
logs/fix_1_latent_mask/1_latent_mask_lr_35e-4_n_latent_masks_2_relu_seed_1340/model_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ac8e5d5b33811647139ff98489b3e069d21b42e16273ff429776be5ea06a85d7
3
+ size 97707314
logs/fix_1_latent_mask/1_latent_mask_lr_35e-4_n_latent_masks_2_relu_seed_1340/optimizer_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:66ec9a69f3b8a98ea2cab7e3f480981fbadbccc1f6f9bfd7c2662c4bd3869465
3
+ size 189136950