andrew-healey commited on
Commit
2433767
·
verified ·
1 Parent(s): 0ee702e

Upload folder using huggingface_hub

Browse files
logs/fix_1_latent_mask/1_latent_mask_lr_30e-4_n_latent_masks_2_compile_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_30e-4_n_latent_masks_2_compile_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": 0.003, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "1_latent_mask_lr_30e-4_n_latent_masks_2_compile"}
logs/fix_1_latent_mask/1_latent_mask_lr_30e-4_n_latent_masks_2_compile_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_30e-4_n_latent_masks_2_compile_seed_1340/log2.txt ADDED
@@ -0,0 +1,529 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ max_steps: 4375
2
+ 0 val loss 11.2068
3
+ 0 val perplexity 73627.0703
4
+ 0 train 11.205775 (lr=4.1958e-06) (hash(x)=45482580)
5
+ 10 train 9.850242 (lr=4.6154e-05) (hash(x)=38414730)
6
+ 20 train 9.344382 (lr=8.8112e-05) (hash(x)=39783917)
7
+ 30 train 8.604013 (lr=1.3007e-04) (hash(x)=38395733)
8
+ 40 train 7.903165 (lr=1.7203e-04) (hash(x)=37486707)
9
+ 50 train 7.630358 (lr=2.1399e-04) (hash(x)=39897505)
10
+ 60 train 7.548781 (lr=2.5594e-04) (hash(x)=40933473)
11
+ 70 train 7.542393 (lr=2.9790e-04) (hash(x)=37690746)
12
+ 80 train 7.417057 (lr=3.3986e-04) (hash(x)=38609656)
13
+ 90 train 7.193568 (lr=3.8182e-04) (hash(x)=38356571)
14
+ 100 val loss 7.0796
15
+ 100 val perplexity 1187.4360
16
+ 100 train 7.060431 (lr=4.2378e-04) (hash(x)=35980376)
17
+ 110 train 6.950359 (lr=4.6573e-04) (hash(x)=42011042)
18
+ 120 train 6.875343 (lr=5.0769e-04) (hash(x)=40266823)
19
+ 130 train 6.686852 (lr=5.4965e-04) (hash(x)=38645447)
20
+ 140 train 6.593544 (lr=5.9161e-04) (hash(x)=36059313)
21
+ 150 train 6.517693 (lr=6.3357e-04) (hash(x)=34161947)
22
+ 160 train 6.395858 (lr=6.7552e-04) (hash(x)=42263375)
23
+ 170 train 6.465276 (lr=7.1748e-04) (hash(x)=42957725)
24
+ 180 train 6.468699 (lr=7.5944e-04) (hash(x)=40198018)
25
+ 190 train 6.197411 (lr=8.0140e-04) (hash(x)=41666215)
26
+ 200 val loss 6.1951
27
+ 200 val perplexity 490.3320
28
+ 200 train 6.163189 (lr=8.4336e-04) (hash(x)=54060482)
29
+ 210 train 6.058611 (lr=8.8531e-04) (hash(x)=42332778)
30
+ 220 train 6.026582 (lr=9.2727e-04) (hash(x)=39551486)
31
+ 230 train 6.024071 (lr=9.6923e-04) (hash(x)=38462018)
32
+ 240 train 6.059446 (lr=1.0112e-03) (hash(x)=36591442)
33
+ 250 train 5.999529 (lr=1.0531e-03) (hash(x)=40861237)
34
+ 260 train 5.935446 (lr=1.0951e-03) (hash(x)=41739369)
35
+ 270 train 5.820512 (lr=1.1371e-03) (hash(x)=39415360)
36
+ 280 train 5.803187 (lr=1.1790e-03) (hash(x)=44665934)
37
+ 290 train 5.726259 (lr=1.2210e-03) (hash(x)=37035579)
38
+ 300 val loss 5.7106
39
+ 300 val perplexity 302.0634
40
+ 300 train 5.737195 (lr=1.2629e-03) (hash(x)=38301011)
41
+ 310 train 5.585602 (lr=1.3049e-03) (hash(x)=42850980)
42
+ 320 train 5.658031 (lr=1.3469e-03) (hash(x)=37476222)
43
+ 330 train 5.566215 (lr=1.3888e-03) (hash(x)=53028205)
44
+ 340 train 5.576035 (lr=1.4308e-03) (hash(x)=41466008)
45
+ 350 train 5.493243 (lr=1.4727e-03) (hash(x)=37802865)
46
+ 360 train 5.454326 (lr=1.5147e-03) (hash(x)=42769282)
47
+ 370 train 5.461638 (lr=1.5566e-03) (hash(x)=39319256)
48
+ 380 train 5.463205 (lr=1.5986e-03) (hash(x)=42637402)
49
+ 390 train 5.440033 (lr=1.6406e-03) (hash(x)=31141514)
50
+ 400 val loss 5.4314
51
+ 400 val perplexity 228.4796
52
+ 400 train 5.506195 (lr=1.6825e-03) (hash(x)=38151157)
53
+ 410 train 5.437372 (lr=1.7245e-03) (hash(x)=46033439)
54
+ 420 train 5.413689 (lr=1.7664e-03) (hash(x)=41365246)
55
+ 430 train 5.404969 (lr=1.8084e-03) (hash(x)=42369184)
56
+ 440 train 5.395553 (lr=1.8503e-03) (hash(x)=42004840)
57
+ 450 train 5.328038 (lr=1.8923e-03) (hash(x)=37181172)
58
+ 460 train 5.430197 (lr=1.9343e-03) (hash(x)=31630797)
59
+ 470 train 5.275708 (lr=1.9762e-03) (hash(x)=42135747)
60
+ 480 train 5.196821 (lr=2.0182e-03) (hash(x)=30023651)
61
+ 490 train 5.280267 (lr=2.0601e-03) (hash(x)=40463476)
62
+ 500 val loss 5.2514
63
+ 500 val perplexity 190.8298
64
+ 500 train 5.173197 (lr=2.1021e-03) (hash(x)=37089842)
65
+ 510 train 5.183855 (lr=2.1441e-03) (hash(x)=41095974)
66
+ 520 train 5.220747 (lr=2.1860e-03) (hash(x)=41525258)
67
+ 530 train 5.210732 (lr=2.2280e-03) (hash(x)=37697291)
68
+ 540 train 5.244308 (lr=2.2699e-03) (hash(x)=43459894)
69
+ 550 train 5.095397 (lr=2.3119e-03) (hash(x)=38993956)
70
+ 560 train 5.152733 (lr=2.3538e-03) (hash(x)=40677277)
71
+ 570 train 5.218308 (lr=2.3958e-03) (hash(x)=43103127)
72
+ 580 train 5.150270 (lr=2.4378e-03) (hash(x)=41871776)
73
+ 590 train 5.126070 (lr=2.4797e-03) (hash(x)=43230967)
74
+ 600 val loss 5.0617
75
+ 600 val perplexity 157.8508
76
+ 600 train 5.097387 (lr=2.5217e-03) (hash(x)=42226172)
77
+ 610 train 5.456991 (lr=2.5636e-03) (hash(x)=43007164)
78
+ 620 train 4.977196 (lr=2.6056e-03) (hash(x)=35752341)
79
+ 630 train 5.045135 (lr=2.6476e-03) (hash(x)=39453504)
80
+ 640 train 4.983343 (lr=2.6895e-03) (hash(x)=43635712)
81
+ 650 train 4.979870 (lr=2.7315e-03) (hash(x)=41852456)
82
+ 660 train 5.014956 (lr=2.7734e-03) (hash(x)=49125742)
83
+ 670 train 5.009706 (lr=2.8154e-03) (hash(x)=34886275)
84
+ 680 train 5.102986 (lr=2.8573e-03) (hash(x)=40546482)
85
+ 690 train 4.914345 (lr=2.8993e-03) (hash(x)=40850901)
86
+ 700 val loss 4.8999
87
+ 700 val perplexity 134.2703
88
+ 700 train 4.827371 (lr=2.9413e-03) (hash(x)=37426993)
89
+ 710 train 5.059704 (lr=2.9832e-03) (hash(x)=42679743)
90
+ 720 train 4.885276 (lr=3.0000e-03) (hash(x)=44856043)
91
+ 730 train 4.865795 (lr=2.9999e-03) (hash(x)=36457471)
92
+ 740 train 4.851224 (lr=2.9997e-03) (hash(x)=47203534)
93
+ 750 train 4.843387 (lr=2.9994e-03) (hash(x)=35585657)
94
+ 760 train 4.771635 (lr=2.9990e-03) (hash(x)=37259264)
95
+ 770 train 4.686066 (lr=2.9985e-03) (hash(x)=44161997)
96
+ 780 train 4.631039 (lr=2.9979e-03) (hash(x)=41272886)
97
+ 790 train 4.646680 (lr=2.9972e-03) (hash(x)=43303662)
98
+ 800 val loss 4.8438
99
+ 800 val perplexity 126.9563
100
+ 800 train 4.571098 (lr=2.9964e-03) (hash(x)=39067231)
101
+ 810 train 4.797554 (lr=2.9955e-03) (hash(x)=44046732)
102
+ 820 train 4.773737 (lr=2.9945e-03) (hash(x)=39783962)
103
+ 830 train 4.841770 (lr=2.9934e-03) (hash(x)=40973974)
104
+ 840 train 4.814116 (lr=2.9922e-03) (hash(x)=41619968)
105
+ 850 train 4.871266 (lr=2.9909e-03) (hash(x)=36939960)
106
+ 860 train 4.754478 (lr=2.9896e-03) (hash(x)=42942377)
107
+ 870 train 4.773477 (lr=2.9881e-03) (hash(x)=39659455)
108
+ 880 train 4.798032 (lr=2.9865e-03) (hash(x)=40331986)
109
+ 890 train 4.741030 (lr=2.9848e-03) (hash(x)=38084814)
110
+ 900 val loss 4.8042
111
+ 900 val perplexity 122.0189
112
+ 900 train 4.798233 (lr=2.9830e-03) (hash(x)=30830367)
113
+ 910 train 4.695802 (lr=2.9811e-03) (hash(x)=39664356)
114
+ 920 train 4.703725 (lr=2.9792e-03) (hash(x)=39007775)
115
+ 930 train 4.709078 (lr=2.9771e-03) (hash(x)=39319254)
116
+ 940 train 4.729657 (lr=2.9749e-03) (hash(x)=36347051)
117
+ 950 train 4.711105 (lr=2.9726e-03) (hash(x)=39243577)
118
+ 960 train 4.749465 (lr=2.9703e-03) (hash(x)=45603839)
119
+ 970 train 4.585362 (lr=2.9678e-03) (hash(x)=40036075)
120
+ 980 train 4.660421 (lr=2.9652e-03) (hash(x)=43068524)
121
+ 990 train 4.757687 (lr=2.9626e-03) (hash(x)=33798472)
122
+ 1000 val loss 4.9067
123
+ 1000 val perplexity 135.1982
124
+ 1000 train 5.109630 (lr=2.9598e-03) (hash(x)=37728665)
125
+ 1010 train 5.020430 (lr=2.9570e-03) (hash(x)=43678636)
126
+ 1020 train 5.015272 (lr=2.9540e-03) (hash(x)=48464376)
127
+ 1030 train 4.965438 (lr=2.9510e-03) (hash(x)=43771461)
128
+ 1040 train 4.849215 (lr=2.9478e-03) (hash(x)=44643085)
129
+ 1050 train 4.963343 (lr=2.9446e-03) (hash(x)=39201716)
130
+ 1060 train 4.824945 (lr=2.9412e-03) (hash(x)=36396343)
131
+ 1070 train 4.784394 (lr=2.9378e-03) (hash(x)=37504111)
132
+ 1080 train 4.809821 (lr=2.9343e-03) (hash(x)=37454537)
133
+ 1090 train 4.833096 (lr=2.9307e-03) (hash(x)=36888183)
134
+ 1100 val loss 4.8168
135
+ 1100 val perplexity 123.5629
136
+ 1100 train 4.769867 (lr=2.9270e-03) (hash(x)=39549045)
137
+ 1110 train 4.947967 (lr=2.9231e-03) (hash(x)=56818420)
138
+ 1120 train 4.709064 (lr=2.9192e-03) (hash(x)=36573357)
139
+ 1130 train 4.656867 (lr=2.9152e-03) (hash(x)=38540878)
140
+ 1140 train 4.635711 (lr=2.9112e-03) (hash(x)=38645859)
141
+ 1150 train 4.850067 (lr=2.9070e-03) (hash(x)=41972503)
142
+ 1160 train 4.481565 (lr=2.9027e-03) (hash(x)=38151868)
143
+ 1170 train 4.547694 (lr=2.8983e-03) (hash(x)=42375436)
144
+ 1180 train 4.516358 (lr=2.8939e-03) (hash(x)=42868028)
145
+ 1190 train 4.758000 (lr=2.8893e-03) (hash(x)=41060868)
146
+ 1200 val loss 4.7243
147
+ 1200 val perplexity 112.6548
148
+ 1200 train 4.735517 (lr=2.8847e-03) (hash(x)=40852814)
149
+ 1210 train 4.725243 (lr=2.8800e-03) (hash(x)=43365554)
150
+ 1220 train 4.738847 (lr=2.8751e-03) (hash(x)=39121134)
151
+ 1230 train 4.789391 (lr=2.8702e-03) (hash(x)=40565426)
152
+ 1240 train 4.835768 (lr=2.8652e-03) (hash(x)=43378926)
153
+ 1250 train 4.680925 (lr=2.8601e-03) (hash(x)=35227381)
154
+ 1260 train 4.808817 (lr=2.8550e-03) (hash(x)=41267226)
155
+ 1270 train 4.685859 (lr=2.8497e-03) (hash(x)=31395565)
156
+ 1280 train 4.824275 (lr=2.8443e-03) (hash(x)=41470963)
157
+ 1290 train 4.636352 (lr=2.8389e-03) (hash(x)=40729366)
158
+ 1300 val loss 4.6924
159
+ 1300 val perplexity 109.1153
160
+ 1300 train 4.648442 (lr=2.8333e-03) (hash(x)=39916663)
161
+ 1310 train 4.703752 (lr=2.8277e-03) (hash(x)=42179962)
162
+ 1320 train 4.569791 (lr=2.8220e-03) (hash(x)=44991932)
163
+ 1330 train 4.548046 (lr=2.8162e-03) (hash(x)=38925911)
164
+ 1340 train 4.537402 (lr=2.8103e-03) (hash(x)=38565560)
165
+ 1350 train 4.598592 (lr=2.8044e-03) (hash(x)=43384902)
166
+ 1360 train 4.544124 (lr=2.7983e-03) (hash(x)=45422261)
167
+ 1370 train 4.550006 (lr=2.7922e-03) (hash(x)=40167325)
168
+ 1380 train 4.687752 (lr=2.7860e-03) (hash(x)=51258162)
169
+ 1390 train 4.737079 (lr=2.7797e-03) (hash(x)=43229106)
170
+ 1400 val loss 4.6813
171
+ 1400 val perplexity 107.9079
172
+ 1400 train 4.632034 (lr=2.7733e-03) (hash(x)=42537494)
173
+ 1410 train 4.930667 (lr=2.7668e-03) (hash(x)=51236490)
174
+ 1420 train 4.787405 (lr=2.7603e-03) (hash(x)=45243347)
175
+ 1430 train 4.617177 (lr=2.7536e-03) (hash(x)=37658566)
176
+ 1440 train 4.680358 (lr=2.7469e-03) (hash(x)=38856106)
177
+ 1450 train 4.791890 (lr=2.7401e-03) (hash(x)=36649488)
178
+ 1460 train 4.771212 (lr=2.7333e-03) (hash(x)=40445407)
179
+ 1470 train 4.704154 (lr=2.7263e-03) (hash(x)=39916387)
180
+ 1480 train 4.689462 (lr=2.7193e-03) (hash(x)=40934555)
181
+ 1490 train 4.626233 (lr=2.7121e-03) (hash(x)=42276527)
182
+ 1500 val loss 4.6835
183
+ 1500 val perplexity 108.1473
184
+ 1500 train 4.627641 (lr=2.7050e-03) (hash(x)=37973883)
185
+ 1510 train 4.703391 (lr=2.6977e-03) (hash(x)=38929286)
186
+ 1520 train 4.545280 (lr=2.6903e-03) (hash(x)=36855837)
187
+ 1530 train 4.816565 (lr=2.6829e-03) (hash(x)=42146431)
188
+ 1540 train 4.768189 (lr=2.6754e-03) (hash(x)=41898411)
189
+ 1550 train 4.705527 (lr=2.6678e-03) (hash(x)=44906272)
190
+ 1560 train 4.712567 (lr=2.6602e-03) (hash(x)=40414353)
191
+ 1570 train 4.762092 (lr=2.6525e-03) (hash(x)=38723190)
192
+ 1580 train 4.712957 (lr=2.6447e-03) (hash(x)=39065271)
193
+ 1590 train 4.660449 (lr=2.6368e-03) (hash(x)=40000886)
194
+ 1600 val loss 4.7372
195
+ 1600 val perplexity 114.1168
196
+ 1600 train 4.559001 (lr=2.6289e-03) (hash(x)=37898571)
197
+ 1610 train 4.762089 (lr=2.6208e-03) (hash(x)=40447863)
198
+ 1620 train 4.683432 (lr=2.6128e-03) (hash(x)=40119318)
199
+ 1630 train 4.644299 (lr=2.6046e-03) (hash(x)=40664074)
200
+ 1640 train 4.656006 (lr=2.5964e-03) (hash(x)=37404741)
201
+ 1650 train 4.776863 (lr=2.5881e-03) (hash(x)=39214928)
202
+ 1660 train 4.830355 (lr=2.5797e-03) (hash(x)=35005019)
203
+ 1670 train 5.271607 (lr=2.5713e-03) (hash(x)=37675832)
204
+ 1680 train 5.012440 (lr=2.5628e-03) (hash(x)=40489680)
205
+ 1690 train 4.892616 (lr=2.5542e-03) (hash(x)=45537879)
206
+ 1700 val loss 4.9544
207
+ 1700 val perplexity 141.7918
208
+ 1700 train 4.868583 (lr=2.5455e-03) (hash(x)=38866100)
209
+ 1710 train 4.939192 (lr=2.5368e-03) (hash(x)=44726254)
210
+ 1720 train 4.976794 (lr=2.5281e-03) (hash(x)=36917792)
211
+ 1730 train 5.113350 (lr=2.5192e-03) (hash(x)=42814805)
212
+ 1740 train 5.140297 (lr=2.5103e-03) (hash(x)=51812216)
213
+ 1750 train 5.023253 (lr=2.5014e-03) (hash(x)=33549014)
214
+ 1760 train 5.703001 (lr=2.4924e-03) (hash(x)=40354215)
215
+ 1770 train 5.253572 (lr=2.4833e-03) (hash(x)=40718606)
216
+ 1780 train 5.356293 (lr=2.4741e-03) (hash(x)=38274164)
217
+ 1790 train 5.478086 (lr=2.4649e-03) (hash(x)=42531471)
218
+ 1800 val loss 5.3563
219
+ 1800 val perplexity 211.9458
220
+ 1800 train 5.317098 (lr=2.4556e-03) (hash(x)=35616519)
221
+ 1810 train 5.454968 (lr=2.4463e-03) (hash(x)=33803118)
222
+ 1820 train 5.498106 (lr=2.4369e-03) (hash(x)=36973525)
223
+ 1830 train 5.292574 (lr=2.4275e-03) (hash(x)=38162549)
224
+ 1840 train 5.234261 (lr=2.4180e-03) (hash(x)=33937159)
225
+ 1850 train 5.311055 (lr=2.4084e-03) (hash(x)=39887546)
226
+ 1860 train 5.207295 (lr=2.3988e-03) (hash(x)=37818525)
227
+ 1870 train 5.331614 (lr=2.3891e-03) (hash(x)=37250478)
228
+ 1880 train 5.113265 (lr=2.3794e-03) (hash(x)=36412167)
229
+ 1890 train 5.151650 (lr=2.3696e-03) (hash(x)=40261189)
230
+ 1900 val loss 5.2738
231
+ 1900 val perplexity 195.1544
232
+ 1900 train 5.115120 (lr=2.3598e-03) (hash(x)=38654303)
233
+ 1910 train 5.166984 (lr=2.3499e-03) (hash(x)=37929515)
234
+ 1920 train 5.378798 (lr=2.3400e-03) (hash(x)=45580146)
235
+ 1930 train 5.281107 (lr=2.3300e-03) (hash(x)=32386330)
236
+ 1940 train 5.410443 (lr=2.3200e-03) (hash(x)=36331864)
237
+ 1950 train 5.371981 (lr=2.3099e-03) (hash(x)=41181727)
238
+ 1960 train 5.272092 (lr=2.2998e-03) (hash(x)=42705152)
239
+ 1970 train 5.262036 (lr=2.2896e-03) (hash(x)=40251511)
240
+ 1980 train 5.247238 (lr=2.2793e-03) (hash(x)=37525551)
241
+ 1990 train 5.411318 (lr=2.2691e-03) (hash(x)=41849618)
242
+ 2000 val loss 5.3708
243
+ 2000 val perplexity 215.0336
244
+ 2000 train 5.302853 (lr=2.2588e-03) (hash(x)=37642582)
245
+ 2010 train 5.589004 (lr=2.2484e-03) (hash(x)=42131121)
246
+ 2020 train 5.557247 (lr=2.2380e-03) (hash(x)=39000209)
247
+ 2030 train 5.424429 (lr=2.2275e-03) (hash(x)=43641355)
248
+ 2040 train 5.346410 (lr=2.2170e-03) (hash(x)=47910507)
249
+ 2050 train 5.461271 (lr=2.2065e-03) (hash(x)=36670359)
250
+ 2060 train 5.810902 (lr=2.1959e-03) (hash(x)=36477755)
251
+ 2070 train 5.747754 (lr=2.1853e-03) (hash(x)=43262487)
252
+ 2080 train 5.878706 (lr=2.1746e-03) (hash(x)=49546029)
253
+ 2090 train 5.553816 (lr=2.1639e-03) (hash(x)=39924731)
254
+ 2100 val loss 5.7008
255
+ 2100 val perplexity 299.1111
256
+ 2100 train 5.656739 (lr=2.1532e-03) (hash(x)=39921304)
257
+ 2110 train 5.551340 (lr=2.1424e-03) (hash(x)=38152788)
258
+ 2120 train 5.555457 (lr=2.1316e-03) (hash(x)=37977911)
259
+ 2130 train 5.404351 (lr=2.1208e-03) (hash(x)=34748760)
260
+ 2140 train 5.487201 (lr=2.1099e-03) (hash(x)=40509369)
261
+ 2150 train 5.397923 (lr=2.0990e-03) (hash(x)=37654262)
262
+ 2160 train 5.390488 (lr=2.0881e-03) (hash(x)=38139543)
263
+ 2170 train 5.371758 (lr=2.0771e-03) (hash(x)=42501806)
264
+ 2180 train 5.537432 (lr=2.0661e-03) (hash(x)=40085092)
265
+ 2190 train 5.403194 (lr=2.0550e-03) (hash(x)=50480193)
266
+ 2200 val loss 5.6468
267
+ 2200 val perplexity 283.3873
268
+ 2200 train 5.634093 (lr=2.0440e-03) (hash(x)=40604084)
269
+ 2210 train 5.734743 (lr=2.0329e-03) (hash(x)=41555823)
270
+ 2220 train 5.405392 (lr=2.0217e-03) (hash(x)=50441765)
271
+ 2230 train 5.515377 (lr=2.0106e-03) (hash(x)=39796580)
272
+ 2240 train 5.508901 (lr=1.9994e-03) (hash(x)=44127022)
273
+ 2250 train 5.345138 (lr=1.9882e-03) (hash(x)=37026826)
274
+ 2260 train 5.322480 (lr=1.9770e-03) (hash(x)=42133839)
275
+ 2270 train 5.315242 (lr=1.9657e-03) (hash(x)=38500664)
276
+ 2280 train 5.444384 (lr=1.9544e-03) (hash(x)=40538661)
277
+ 2290 train 5.468823 (lr=1.9431e-03) (hash(x)=51509210)
278
+ 2300 val loss 5.5112
279
+ 2300 val perplexity 247.4395
280
+ 2300 train 5.429445 (lr=1.9318e-03) (hash(x)=41952328)
281
+ 2310 train 5.365610 (lr=1.9205e-03) (hash(x)=39758123)
282
+ 2320 train 5.510546 (lr=1.9091e-03) (hash(x)=51089268)
283
+ 2330 train 5.673557 (lr=1.8977e-03) (hash(x)=39767618)
284
+ 2340 train 5.690990 (lr=1.8863e-03) (hash(x)=40409617)
285
+ 2350 train 5.607941 (lr=1.8749e-03) (hash(x)=40349634)
286
+ 2360 train 5.626188 (lr=1.8635e-03) (hash(x)=31841172)
287
+ 2370 train 5.628290 (lr=1.8520e-03) (hash(x)=42720539)
288
+ 2380 train 5.637767 (lr=1.8406e-03) (hash(x)=40998632)
289
+ 2390 train 5.701478 (lr=1.8291e-03) (hash(x)=40615413)
290
+ 2400 val loss 5.7560
291
+ 2400 val perplexity 316.0739
292
+ 2400 train 5.654748 (lr=1.8176e-03) (hash(x)=39373658)
293
+ 2410 train 5.811829 (lr=1.8061e-03) (hash(x)=35480858)
294
+ 2420 train 5.781505 (lr=1.7946e-03) (hash(x)=32241095)
295
+ 2430 train 5.836713 (lr=1.7830e-03) (hash(x)=36669715)
296
+ 2440 train 5.932039 (lr=1.7715e-03) (hash(x)=45768335)
297
+ 2450 train 5.987591 (lr=1.7600e-03) (hash(x)=39395055)
298
+ 2460 train 6.033963 (lr=1.7484e-03) (hash(x)=34899269)
299
+ 2470 train 6.097768 (lr=1.7368e-03) (hash(x)=43642420)
300
+ 2480 train 6.143208 (lr=1.7253e-03) (hash(x)=47099507)
301
+ 2490 train 5.800528 (lr=1.7137e-03) (hash(x)=38825558)
302
+ 2500 val loss 5.7769
303
+ 2500 val perplexity 322.7430
304
+ 2500 train 5.772750 (lr=1.7021e-03) (hash(x)=39833804)
305
+ 2510 train 5.879582 (lr=1.6906e-03) (hash(x)=35051654)
306
+ 2520 train 5.728175 (lr=1.6790e-03) (hash(x)=40291109)
307
+ 2530 train 5.705974 (lr=1.6674e-03) (hash(x)=36915768)
308
+ 2540 train 5.805096 (lr=1.6558e-03) (hash(x)=37016308)
309
+ 2550 train 5.707369 (lr=1.6442e-03) (hash(x)=43221777)
310
+ 2560 train 5.760855 (lr=1.6326e-03) (hash(x)=37233207)
311
+ 2570 train 5.763321 (lr=1.6210e-03) (hash(x)=37510626)
312
+ 2580 train 5.748258 (lr=1.6094e-03) (hash(x)=39624656)
313
+ 2590 train 6.007481 (lr=1.5979e-03) (hash(x)=37647501)
314
+ 2600 val loss 6.0452
315
+ 2600 val perplexity 422.0990
316
+ 2600 train 5.892860 (lr=1.5863e-03) (hash(x)=37200138)
317
+ 2610 train 5.963423 (lr=1.5747e-03) (hash(x)=43066970)
318
+ 2620 train 6.082755 (lr=1.5632e-03) (hash(x)=40503799)
319
+ 2630 train 6.047740 (lr=1.5516e-03) (hash(x)=45534660)
320
+ 2640 train 6.141528 (lr=1.5400e-03) (hash(x)=52358353)
321
+ 2650 train 5.869315 (lr=1.5285e-03) (hash(x)=41843599)
322
+ 2660 train 5.899861 (lr=1.5170e-03) (hash(x)=40519770)
323
+ 2670 train 6.012674 (lr=1.5054e-03) (hash(x)=37185602)
324
+ 2680 train 5.921664 (lr=1.4939e-03) (hash(x)=43805719)
325
+ 2690 train 5.819825 (lr=1.4824e-03) (hash(x)=39458250)
326
+ 2700 val loss 5.9194
327
+ 2700 val perplexity 372.1856
328
+ 2700 train 5.882730 (lr=1.4709e-03) (hash(x)=39149255)
329
+ 2710 train 5.864868 (lr=1.4594e-03) (hash(x)=43544861)
330
+ 2720 train 6.056668 (lr=1.4480e-03) (hash(x)=36858197)
331
+ 2730 train 6.270341 (lr=1.4365e-03) (hash(x)=37971712)
332
+ 2740 train 6.136947 (lr=1.4251e-03) (hash(x)=42416578)
333
+ 2750 train 6.256564 (lr=1.4137e-03) (hash(x)=53554622)
334
+ 2760 train 6.368878 (lr=1.4023e-03) (hash(x)=40502545)
335
+ 2770 train 6.276895 (lr=1.3909e-03) (hash(x)=36086417)
336
+ 2780 train 6.155672 (lr=1.3795e-03) (hash(x)=36242982)
337
+ 2790 train 6.274453 (lr=1.3682e-03) (hash(x)=41433780)
338
+ 2800 val loss 6.1663
339
+ 2800 val perplexity 476.4402
340
+ 2800 train 6.102938 (lr=1.3569e-03) (hash(x)=48553484)
341
+ 2810 train 5.987403 (lr=1.3456e-03) (hash(x)=39928650)
342
+ 2820 train 5.988398 (lr=1.3343e-03) (hash(x)=37891724)
343
+ 2830 train 6.103116 (lr=1.3230e-03) (hash(x)=34353412)
344
+ 2840 train 6.169580 (lr=1.3118e-03) (hash(x)=38486611)
345
+ 2850 train 5.937134 (lr=1.3006e-03) (hash(x)=32706934)
346
+ 2860 train 5.960267 (lr=1.2894e-03) (hash(x)=38693681)
347
+ 2870 train 5.839033 (lr=1.2783e-03) (hash(x)=38430800)
348
+ 2880 train 5.921930 (lr=1.2671e-03) (hash(x)=40619559)
349
+ 2890 train 6.230237 (lr=1.2560e-03) (hash(x)=42741066)
350
+ 2900 val loss 6.2093
351
+ 2900 val perplexity 497.3483
352
+ 2900 train 6.324026 (lr=1.2450e-03) (hash(x)=38216091)
353
+ 2910 train 6.223581 (lr=1.2339e-03) (hash(x)=35447832)
354
+ 2920 train 6.143615 (lr=1.2229e-03) (hash(x)=39825190)
355
+ 2930 train 6.123872 (lr=1.2119e-03) (hash(x)=37789121)
356
+ 2940 train 6.111683 (lr=1.2010e-03) (hash(x)=44499116)
357
+ 2950 train 6.249564 (lr=1.1901e-03) (hash(x)=39669860)
358
+ 2960 train 6.391609 (lr=1.1792e-03) (hash(x)=51683741)
359
+ 2970 train 6.227346 (lr=1.1684e-03) (hash(x)=36399721)
360
+ 2980 train 6.193961 (lr=1.1576e-03) (hash(x)=42629700)
361
+ 2990 train 6.078774 (lr=1.1468e-03) (hash(x)=39263773)
362
+ 3000 val loss 6.1782
363
+ 3000 val perplexity 482.1354
364
+ 3000 train 6.021534 (lr=1.1361e-03) (hash(x)=37920485)
365
+ 3010 train 6.053445 (lr=1.1254e-03) (hash(x)=40835161)
366
+ 3020 train 6.039999 (lr=1.1147e-03) (hash(x)=38381159)
367
+ 3030 train 5.925543 (lr=1.1041e-03) (hash(x)=35846270)
368
+ 3040 train 6.337908 (lr=1.0935e-03) (hash(x)=47516567)
369
+ 3050 train 6.374860 (lr=1.0830e-03) (hash(x)=35600311)
370
+ 3060 train 6.356387 (lr=1.0725e-03) (hash(x)=37775318)
371
+ 3070 train 6.174574 (lr=1.0620e-03) (hash(x)=39881333)
372
+ 3080 train 6.224731 (lr=1.0516e-03) (hash(x)=37310168)
373
+ 3090 train 6.236110 (lr=1.0412e-03) (hash(x)=43669978)
374
+ 3100 val loss 6.1891
375
+ 3100 val perplexity 487.4162
376
+ 3100 train 6.529277 (lr=1.0309e-03) (hash(x)=46356797)
377
+ 3110 train 6.082632 (lr=1.0207e-03) (hash(x)=40814232)
378
+ 3120 train 6.042809 (lr=1.0104e-03) (hash(x)=42504837)
379
+ 3130 train 6.210238 (lr=1.0002e-03) (hash(x)=39693594)
380
+ 3140 train 6.245903 (lr=9.9011e-04) (hash(x)=40277645)
381
+ 3150 train 6.275350 (lr=9.8002e-04) (hash(x)=40988003)
382
+ 3160 train 6.323730 (lr=9.6999e-04) (hash(x)=38875266)
383
+ 3170 train 6.329995 (lr=9.6000e-04) (hash(x)=45232173)
384
+ 3180 train 6.319748 (lr=9.5007e-04) (hash(x)=39213336)
385
+ 3190 train 6.322706 (lr=9.4019e-04) (hash(x)=42118576)
386
+ 3200 val loss 6.2488
387
+ 3200 val perplexity 517.4070
388
+ 3200 train 6.260384 (lr=9.3036e-04) (hash(x)=32884223)
389
+ 3210 train 6.327086 (lr=9.2058e-04) (hash(x)=41276800)
390
+ 3220 train 6.333274 (lr=9.1085e-04) (hash(x)=40284461)
391
+ 3230 train 6.293327 (lr=9.0118e-04) (hash(x)=40566734)
392
+ 3240 train 6.385384 (lr=8.9157e-04) (hash(x)=36484570)
393
+ 3250 train 6.276123 (lr=8.8201e-04) (hash(x)=41642338)
394
+ 3260 train 6.260554 (lr=8.7251e-04) (hash(x)=43883570)
395
+ 3270 train 6.253153 (lr=8.6307e-04) (hash(x)=40432560)
396
+ 3280 train 6.312924 (lr=8.5368e-04) (hash(x)=38002717)
397
+ 3290 train 6.167565 (lr=8.4435e-04) (hash(x)=41926004)
398
+ 3300 val loss 6.3508
399
+ 3300 val perplexity 572.9304
400
+ 3300 train 6.380917 (lr=8.3508e-04) (hash(x)=46015509)
401
+ 3310 train 6.363791 (lr=8.2588e-04) (hash(x)=40112249)
402
+ 3320 train 6.458265 (lr=8.1673e-04) (hash(x)=49162296)
403
+ 3330 train 6.452383 (lr=8.0764e-04) (hash(x)=46606969)
404
+ 3340 train 6.479995 (lr=7.9862e-04) (hash(x)=31128992)
405
+ 3350 train 6.468321 (lr=7.8966e-04) (hash(x)=41232534)
406
+ 3360 train 6.351395 (lr=7.8076e-04) (hash(x)=41599699)
407
+ 3370 train 6.409516 (lr=7.7192e-04) (hash(x)=40885280)
408
+ 3380 train 6.450243 (lr=7.6315e-04) (hash(x)=37169148)
409
+ 3390 train 6.375486 (lr=7.5445e-04) (hash(x)=46581889)
410
+ 3400 val loss 6.3894
411
+ 3400 val perplexity 595.5137
412
+ 3400 train 6.397552 (lr=7.4581e-04) (hash(x)=37612074)
413
+ 3410 train 6.360183 (lr=7.3724e-04) (hash(x)=38763316)
414
+ 3420 train 6.220906 (lr=7.2874e-04) (hash(x)=40562379)
415
+ 3430 train 6.373963 (lr=7.2030e-04) (hash(x)=44724867)
416
+ 3440 train 6.477915 (lr=7.1193e-04) (hash(x)=39386624)
417
+ 3450 train 6.428370 (lr=7.0363e-04) (hash(x)=38652923)
418
+ 3460 train 6.413062 (lr=6.9541e-04) (hash(x)=29587379)
419
+ 3470 train 6.441362 (lr=6.8725e-04) (hash(x)=38228776)
420
+ 3480 train 6.274127 (lr=6.7916e-04) (hash(x)=30076039)
421
+ 3490 train 6.422761 (lr=6.7114e-04) (hash(x)=38398908)
422
+ 3500 val loss 6.3674
423
+ 3500 val perplexity 582.5222
424
+ 3500 train 6.395857 (lr=6.6320e-04) (hash(x)=39259918)
425
+ 3510 train 6.352352 (lr=6.5533e-04) (hash(x)=40664091)
426
+ 3520 train 6.337359 (lr=6.4753e-04) (hash(x)=40309647)
427
+ 3530 train 6.347334 (lr=6.3981e-04) (hash(x)=38424801)
428
+ 3540 train 6.358965 (lr=6.3216e-04) (hash(x)=51322307)
429
+ 3550 train 6.311920 (lr=6.2458e-04) (hash(x)=38192628)
430
+ 3560 train 6.442401 (lr=6.1708e-04) (hash(x)=41563952)
431
+ 3570 train 6.342924 (lr=6.0966e-04) (hash(x)=39508843)
432
+ 3580 train 6.418537 (lr=6.0231e-04) (hash(x)=41260225)
433
+ 3590 train 6.527983 (lr=5.9504e-04) (hash(x)=49098107)
434
+ 3600 val loss 6.4644
435
+ 3600 val perplexity 641.8986
436
+ 3600 train 6.566151 (lr=5.8785e-04) (hash(x)=41194370)
437
+ 3610 train 6.486939 (lr=5.8074e-04) (hash(x)=40399152)
438
+ 3620 train 6.564960 (lr=5.7370e-04) (hash(x)=41625018)
439
+ 3630 train 6.507775 (lr=5.6675e-04) (hash(x)=44077942)
440
+ 3640 train 6.464283 (lr=5.5987e-04) (hash(x)=42427512)
441
+ 3650 train 6.563638 (lr=5.5308e-04) (hash(x)=40067454)
442
+ 3660 train 6.442956 (lr=5.4636e-04) (hash(x)=37969892)
443
+ 3670 train 6.391196 (lr=5.3973e-04) (hash(x)=43670246)
444
+ 3680 train 6.430098 (lr=5.3318e-04) (hash(x)=37031732)
445
+ 3690 train 6.378663 (lr=5.2671e-04) (hash(x)=39237187)
446
+ 3700 val loss 6.4990
447
+ 3700 val perplexity 664.4620
448
+ 3700 train 6.441600 (lr=5.2033e-04) (hash(x)=39521416)
449
+ 3710 train 6.364771 (lr=5.1402e-04) (hash(x)=41147425)
450
+ 3720 train 6.390386 (lr=5.0780e-04) (hash(x)=40204658)
451
+ 3730 train 6.465328 (lr=5.0167e-04) (hash(x)=39420983)
452
+ 3740 train 6.638474 (lr=4.9562e-04) (hash(x)=51155740)
453
+ 3750 train 6.555363 (lr=4.8965e-04) (hash(x)=41654187)
454
+ 3760 train 6.466383 (lr=4.8377e-04) (hash(x)=40598339)
455
+ 3770 train 6.498601 (lr=4.7798e-04) (hash(x)=44950116)
456
+ 3780 train 6.509258 (lr=4.7227e-04) (hash(x)=40660326)
457
+ 3790 train 6.570079 (lr=4.6665e-04) (hash(x)=37805851)
458
+ 3800 val loss 6.5522
459
+ 3800 val perplexity 700.7584
460
+ 3800 train 6.540962 (lr=4.6112e-04) (hash(x)=38064443)
461
+ 3810 train 6.552744 (lr=4.5567e-04) (hash(x)=39161498)
462
+ 3820 train 6.440475 (lr=4.5031e-04) (hash(x)=40626649)
463
+ 3830 train 6.455749 (lr=4.4504e-04) (hash(x)=36894771)
464
+ 3840 train 6.388431 (lr=4.3986e-04) (hash(x)=33672141)
465
+ 3850 train 6.409985 (lr=4.3477e-04) (hash(x)=41751813)
466
+ 3860 train 6.346309 (lr=4.2977e-04) (hash(x)=33784172)
467
+ 3870 train 6.332170 (lr=4.2486e-04) (hash(x)=43110493)
468
+ 3880 train 6.472645 (lr=4.2004e-04) (hash(x)=36585111)
469
+ 3890 train 6.407299 (lr=4.1530e-04) (hash(x)=40134264)
470
+ 3900 val loss 6.5321
471
+ 3900 val perplexity 686.8654
472
+ 3900 train 6.451552 (lr=4.1066e-04) (hash(x)=39815215)
473
+ 3910 train 6.576622 (lr=4.0611e-04) (hash(x)=42081557)
474
+ 3920 train 6.641671 (lr=4.0166e-04) (hash(x)=48184973)
475
+ 3930 train 6.504176 (lr=3.9729e-04) (hash(x)=42416681)
476
+ 3940 train 6.576923 (lr=3.9302e-04) (hash(x)=38587062)
477
+ 3950 train 6.470510 (lr=3.8884e-04) (hash(x)=37296617)
478
+ 3960 train 6.574070 (lr=3.8475e-04) (hash(x)=37959330)
479
+ 3970 train 6.484903 (lr=3.8076e-04) (hash(x)=35175634)
480
+ 3980 train 6.593958 (lr=3.7685e-04) (hash(x)=37219362)
481
+ 3990 train 6.617995 (lr=3.7305e-04) (hash(x)=38906132)
482
+ 4000 val loss 6.5813
483
+ 4000 val perplexity 721.4675
484
+ 4000 train 6.523434 (lr=3.6933e-04) (hash(x)=39940517)
485
+ 4010 train 6.549121 (lr=3.6572e-04) (hash(x)=39952437)
486
+ 4020 train 6.549597 (lr=3.6219e-04) (hash(x)=42000930)
487
+ 4030 train 6.485825 (lr=3.5876e-04) (hash(x)=41460672)
488
+ 4040 train 6.415194 (lr=3.5543e-04) (hash(x)=43702329)
489
+ 4050 train 6.399704 (lr=3.5219e-04) (hash(x)=41740121)
490
+ 4060 train 6.397235 (lr=3.4905e-04) (hash(x)=37066349)
491
+ 4070 train 6.362606 (lr=3.4600e-04) (hash(x)=38359869)
492
+ 4080 train 6.421095 (lr=3.4305e-04) (hash(x)=35131370)
493
+ 4090 train 6.608535 (lr=3.4019e-04) (hash(x)=39918811)
494
+ 4100 val loss 6.6155
495
+ 4100 val perplexity 746.5697
496
+ 4100 train 6.702071 (lr=3.3744e-04) (hash(x)=47036374)
497
+ 4110 train 6.671317 (lr=3.3477e-04) (hash(x)=37657523)
498
+ 4120 train 6.639130 (lr=3.3221e-04) (hash(x)=37218703)
499
+ 4130 train 6.639815 (lr=3.2974e-04) (hash(x)=41350513)
500
+ 4140 train 6.676506 (lr=3.2737e-04) (hash(x)=38283785)
501
+ 4150 train 6.631254 (lr=3.2510e-04) (hash(x)=32927892)
502
+ 4160 train 6.546591 (lr=3.2292e-04) (hash(x)=42011933)
503
+ 4170 train 6.605139 (lr=3.2085e-04) (hash(x)=41854594)
504
+ 4180 train 6.620859 (lr=3.1887e-04) (hash(x)=36737064)
505
+ 4190 train 6.500279 (lr=3.1699e-04) (hash(x)=36326176)
506
+ 4200 val loss 6.6071
507
+ 4200 val perplexity 740.3132
508
+ 4200 train 6.585936 (lr=3.1520e-04) (hash(x)=39106683)
509
+ 4210 train 6.542267 (lr=3.1352e-04) (hash(x)=39759977)
510
+ 4220 train 6.485462 (lr=3.1193e-04) (hash(x)=43666151)
511
+ 4230 train 6.499436 (lr=3.1044e-04) (hash(x)=38154367)
512
+ 4240 train 6.473877 (lr=3.0905e-04) (hash(x)=38284168)
513
+ 4250 train 6.539319 (lr=3.0776e-04) (hash(x)=38888879)
514
+ 4260 train 6.399486 (lr=3.0657e-04) (hash(x)=36846417)
515
+ 4270 train 6.649404 (lr=3.0548e-04) (hash(x)=43125375)
516
+ 4280 train 6.615999 (lr=3.0449e-04) (hash(x)=43556412)
517
+ 4290 train 6.643809 (lr=3.0359e-04) (hash(x)=48598302)
518
+ 4300 val loss 6.5787
519
+ 4300 val perplexity 719.5985
520
+ 4300 train 6.588540 (lr=3.0280e-04) (hash(x)=42764385)
521
+ 4310 train 6.692759 (lr=3.0210e-04) (hash(x)=41190028)
522
+ 4320 train 6.519741 (lr=3.0150e-04) (hash(x)=35767770)
523
+ 4330 train 6.624144 (lr=3.0101e-04) (hash(x)=36760476)
524
+ 4340 train 6.658907 (lr=3.0061e-04) (hash(x)=40618921)
525
+ 4350 train 6.567091 (lr=3.0031e-04) (hash(x)=32300164)
526
+ 4360 train 6.508783 (lr=3.0011e-04) (hash(x)=41400085)
527
+ 4370 train 6.626045 (lr=3.0001e-04) (hash(x)=40251828)
528
+ 4374 val loss 6.5992
529
+ 4374 val perplexity 734.5040
logs/fix_1_latent_mask/1_latent_mask_lr_30e-4_n_latent_masks_2_compile_seed_1340/model_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:466a45b2fe97cd959dced4974582afd5e53cb4263c8c8b99f62d8bd9008d3041
3
+ size 97707314
logs/fix_1_latent_mask/1_latent_mask_lr_30e-4_n_latent_masks_2_compile_seed_1340/optimizer_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4df2bf22222a78928f04dde44aa8f4c3b6d98ebfb1d15b11b91639edd449dadf
3
+ size 189136950