andrew-healey commited on
Commit
6086939
·
verified ·
1 Parent(s): 2433767

Upload folder using huggingface_hub

Browse files
logs/fix_1_latent_mask/1_latent_mask_lr_30e-4_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_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": 1, "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_compile"}
logs/fix_1_latent_mask/1_latent_mask_lr_30e-4_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_compile_seed_1340/log2.txt ADDED
@@ -0,0 +1,529 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ max_steps: 4375
2
+ 0 val loss 11.2769
3
+ 0 val perplexity 78975.0234
4
+ 0 train 11.266157 (lr=4.1958e-06) (hash(x)=45482580)
5
+ 10 train 9.936153 (lr=4.6154e-05) (hash(x)=38414730)
6
+ 20 train 9.406527 (lr=8.8112e-05) (hash(x)=39783917)
7
+ 30 train 8.674254 (lr=1.3007e-04) (hash(x)=38395733)
8
+ 40 train 7.955272 (lr=1.7203e-04) (hash(x)=37486707)
9
+ 50 train 7.635335 (lr=2.1399e-04) (hash(x)=39897505)
10
+ 60 train 7.572325 (lr=2.5594e-04) (hash(x)=40933473)
11
+ 70 train 7.547815 (lr=2.9790e-04) (hash(x)=37690746)
12
+ 80 train 7.426391 (lr=3.3986e-04) (hash(x)=38609656)
13
+ 90 train 7.214654 (lr=3.8182e-04) (hash(x)=38356571)
14
+ 100 val loss 7.1016
15
+ 100 val perplexity 1213.9561
16
+ 100 train 7.085048 (lr=4.2378e-04) (hash(x)=35980376)
17
+ 110 train 6.979500 (lr=4.6573e-04) (hash(x)=42011042)
18
+ 120 train 6.866656 (lr=5.0769e-04) (hash(x)=40266823)
19
+ 130 train 6.696860 (lr=5.4965e-04) (hash(x)=38645447)
20
+ 140 train 6.607917 (lr=5.9161e-04) (hash(x)=36059313)
21
+ 150 train 6.534482 (lr=6.3357e-04) (hash(x)=34161947)
22
+ 160 train 6.413081 (lr=6.7552e-04) (hash(x)=42263375)
23
+ 170 train 6.465937 (lr=7.1748e-04) (hash(x)=42957725)
24
+ 180 train 6.482312 (lr=7.5944e-04) (hash(x)=40198018)
25
+ 190 train 6.213732 (lr=8.0140e-04) (hash(x)=41666215)
26
+ 200 val loss 6.1945
27
+ 200 val perplexity 490.0258
28
+ 200 train 6.155314 (lr=8.4336e-04) (hash(x)=54060482)
29
+ 210 train 6.078874 (lr=8.8531e-04) (hash(x)=42332778)
30
+ 220 train 6.012352 (lr=9.2727e-04) (hash(x)=39551486)
31
+ 230 train 6.024704 (lr=9.6923e-04) (hash(x)=38462018)
32
+ 240 train 6.089528 (lr=1.0112e-03) (hash(x)=36591442)
33
+ 250 train 6.027266 (lr=1.0531e-03) (hash(x)=40861237)
34
+ 260 train 5.966507 (lr=1.0951e-03) (hash(x)=41739369)
35
+ 270 train 5.852221 (lr=1.1371e-03) (hash(x)=39415360)
36
+ 280 train 5.833349 (lr=1.1790e-03) (hash(x)=44665934)
37
+ 290 train 5.758890 (lr=1.2210e-03) (hash(x)=37035579)
38
+ 300 val loss 5.7470
39
+ 300 val perplexity 313.2615
40
+ 300 train 5.773064 (lr=1.2629e-03) (hash(x)=38301011)
41
+ 310 train 5.622420 (lr=1.3049e-03) (hash(x)=42850980)
42
+ 320 train 5.674726 (lr=1.3469e-03) (hash(x)=37476222)
43
+ 330 train 5.594656 (lr=1.3888e-03) (hash(x)=53028205)
44
+ 340 train 5.600405 (lr=1.4308e-03) (hash(x)=41466008)
45
+ 350 train 5.511488 (lr=1.4727e-03) (hash(x)=37802865)
46
+ 360 train 5.462143 (lr=1.5147e-03) (hash(x)=42769282)
47
+ 370 train 5.487390 (lr=1.5566e-03) (hash(x)=39319256)
48
+ 380 train 5.482636 (lr=1.5986e-03) (hash(x)=42637402)
49
+ 390 train 5.439856 (lr=1.6406e-03) (hash(x)=31141514)
50
+ 400 val loss 5.4306
51
+ 400 val perplexity 228.2908
52
+ 400 train 5.502704 (lr=1.6825e-03) (hash(x)=38151157)
53
+ 410 train 5.438401 (lr=1.7245e-03) (hash(x)=46033439)
54
+ 420 train 5.419750 (lr=1.7664e-03) (hash(x)=41365246)
55
+ 430 train 5.411330 (lr=1.8084e-03) (hash(x)=42369184)
56
+ 440 train 5.404766 (lr=1.8503e-03) (hash(x)=42004840)
57
+ 450 train 5.334952 (lr=1.8923e-03) (hash(x)=37181172)
58
+ 460 train 5.423909 (lr=1.9343e-03) (hash(x)=31630797)
59
+ 470 train 5.288846 (lr=1.9762e-03) (hash(x)=42135747)
60
+ 480 train 5.217771 (lr=2.0182e-03) (hash(x)=30023651)
61
+ 490 train 5.274942 (lr=2.0601e-03) (hash(x)=40463476)
62
+ 500 val loss 5.2504
63
+ 500 val perplexity 190.6495
64
+ 500 train 5.177952 (lr=2.1021e-03) (hash(x)=37089842)
65
+ 510 train 5.193227 (lr=2.1441e-03) (hash(x)=41095974)
66
+ 520 train 5.232347 (lr=2.1860e-03) (hash(x)=41525258)
67
+ 530 train 5.207566 (lr=2.2280e-03) (hash(x)=37697291)
68
+ 540 train 5.239766 (lr=2.2699e-03) (hash(x)=43459894)
69
+ 550 train 5.099840 (lr=2.3119e-03) (hash(x)=38993956)
70
+ 560 train 5.150047 (lr=2.3538e-03) (hash(x)=40677277)
71
+ 570 train 5.221459 (lr=2.3958e-03) (hash(x)=43103127)
72
+ 580 train 5.148258 (lr=2.4378e-03) (hash(x)=41871776)
73
+ 590 train 5.121481 (lr=2.4797e-03) (hash(x)=43230967)
74
+ 600 val loss 5.0525
75
+ 600 val perplexity 156.4189
76
+ 600 train 5.091735 (lr=2.5217e-03) (hash(x)=42226172)
77
+ 610 train 5.428622 (lr=2.5636e-03) (hash(x)=43007164)
78
+ 620 train 4.974985 (lr=2.6056e-03) (hash(x)=35752341)
79
+ 630 train 5.035410 (lr=2.6476e-03) (hash(x)=39453504)
80
+ 640 train 4.984005 (lr=2.6895e-03) (hash(x)=43635712)
81
+ 650 train 4.972001 (lr=2.7315e-03) (hash(x)=41852456)
82
+ 660 train 5.012881 (lr=2.7734e-03) (hash(x)=49125742)
83
+ 670 train 5.001203 (lr=2.8154e-03) (hash(x)=34886275)
84
+ 680 train 5.104885 (lr=2.8573e-03) (hash(x)=40546482)
85
+ 690 train 4.910314 (lr=2.8993e-03) (hash(x)=40850901)
86
+ 700 val loss 4.8927
87
+ 700 val perplexity 133.3100
88
+ 700 train 4.860252 (lr=2.9413e-03) (hash(x)=37426993)
89
+ 710 train 5.038372 (lr=2.9832e-03) (hash(x)=42679743)
90
+ 720 train 4.863777 (lr=3.0000e-03) (hash(x)=44856043)
91
+ 730 train 4.843232 (lr=2.9999e-03) (hash(x)=36457471)
92
+ 740 train 4.812036 (lr=2.9997e-03) (hash(x)=47203534)
93
+ 750 train 4.837779 (lr=2.9994e-03) (hash(x)=35585657)
94
+ 760 train 4.770541 (lr=2.9990e-03) (hash(x)=37259264)
95
+ 770 train 4.674052 (lr=2.9985e-03) (hash(x)=44161997)
96
+ 780 train 4.619155 (lr=2.9979e-03) (hash(x)=41272886)
97
+ 790 train 4.625469 (lr=2.9972e-03) (hash(x)=43303662)
98
+ 800 val loss 4.8247
99
+ 800 val perplexity 124.5489
100
+ 800 train 4.538507 (lr=2.9964e-03) (hash(x)=39067231)
101
+ 810 train 4.785635 (lr=2.9955e-03) (hash(x)=44046732)
102
+ 820 train 4.762691 (lr=2.9945e-03) (hash(x)=39783962)
103
+ 830 train 4.848539 (lr=2.9934e-03) (hash(x)=40973974)
104
+ 840 train 4.887047 (lr=2.9922e-03) (hash(x)=41619968)
105
+ 850 train 4.956460 (lr=2.9909e-03) (hash(x)=36939960)
106
+ 860 train 4.847009 (lr=2.9896e-03) (hash(x)=42942377)
107
+ 870 train 4.832664 (lr=2.9881e-03) (hash(x)=39659455)
108
+ 880 train 4.898205 (lr=2.9865e-03) (hash(x)=40331986)
109
+ 890 train 4.824543 (lr=2.9848e-03) (hash(x)=38084814)
110
+ 900 val loss 4.8471
111
+ 900 val perplexity 127.3738
112
+ 900 train 4.841320 (lr=2.9830e-03) (hash(x)=30830367)
113
+ 910 train 4.730321 (lr=2.9811e-03) (hash(x)=39664356)
114
+ 920 train 4.792222 (lr=2.9792e-03) (hash(x)=39007775)
115
+ 930 train 4.861274 (lr=2.9771e-03) (hash(x)=39319254)
116
+ 940 train 4.872975 (lr=2.9749e-03) (hash(x)=36347051)
117
+ 950 train 4.865705 (lr=2.9726e-03) (hash(x)=39243577)
118
+ 960 train 5.003919 (lr=2.9703e-03) (hash(x)=45603839)
119
+ 970 train 4.825254 (lr=2.9678e-03) (hash(x)=40036075)
120
+ 980 train 4.851970 (lr=2.9652e-03) (hash(x)=43068524)
121
+ 990 train 4.832294 (lr=2.9626e-03) (hash(x)=33798472)
122
+ 1000 val loss 4.9936
123
+ 1000 val perplexity 147.4617
124
+ 1000 train 5.200039 (lr=2.9598e-03) (hash(x)=37728665)
125
+ 1010 train 5.168383 (lr=2.9570e-03) (hash(x)=43678636)
126
+ 1020 train 5.176479 (lr=2.9540e-03) (hash(x)=48464376)
127
+ 1030 train 5.200836 (lr=2.9510e-03) (hash(x)=43771461)
128
+ 1040 train 5.077681 (lr=2.9478e-03) (hash(x)=44643085)
129
+ 1050 train 5.143516 (lr=2.9446e-03) (hash(x)=39201716)
130
+ 1060 train 5.108031 (lr=2.9412e-03) (hash(x)=36396343)
131
+ 1070 train 5.171178 (lr=2.9378e-03) (hash(x)=37504111)
132
+ 1080 train 5.233638 (lr=2.9343e-03) (hash(x)=37454537)
133
+ 1090 train 5.146191 (lr=2.9307e-03) (hash(x)=36888183)
134
+ 1100 val loss 5.1382
135
+ 1100 val perplexity 170.4169
136
+ 1100 train 5.086620 (lr=2.9270e-03) (hash(x)=39549045)
137
+ 1110 train 5.397051 (lr=2.9231e-03) (hash(x)=56818420)
138
+ 1120 train 5.481594 (lr=2.9192e-03) (hash(x)=36573357)
139
+ 1130 train 6.650126 (lr=2.9152e-03) (hash(x)=38540878)
140
+ 1140 train 5.311724 (lr=2.9112e-03) (hash(x)=38645859)
141
+ 1150 train 5.350887 (lr=2.9070e-03) (hash(x)=41972503)
142
+ 1160 train 4.946488 (lr=2.9027e-03) (hash(x)=38151868)
143
+ 1170 train 5.014961 (lr=2.8983e-03) (hash(x)=42375436)
144
+ 1180 train 4.988117 (lr=2.8939e-03) (hash(x)=42868028)
145
+ 1190 train 5.169252 (lr=2.8893e-03) (hash(x)=41060868)
146
+ 1200 val loss 5.2074
147
+ 1200 val perplexity 182.6203
148
+ 1200 train 5.239750 (lr=2.8847e-03) (hash(x)=40852814)
149
+ 1210 train 5.298397 (lr=2.8800e-03) (hash(x)=43365554)
150
+ 1220 train 5.273409 (lr=2.8751e-03) (hash(x)=39121134)
151
+ 1230 train 5.352452 (lr=2.8702e-03) (hash(x)=40565426)
152
+ 1240 train 5.449332 (lr=2.8652e-03) (hash(x)=43378926)
153
+ 1250 train 5.362242 (lr=2.8601e-03) (hash(x)=35227381)
154
+ 1260 train 5.438804 (lr=2.8550e-03) (hash(x)=41267226)
155
+ 1270 train 5.434634 (lr=2.8497e-03) (hash(x)=31395565)
156
+ 1280 train 5.675923 (lr=2.8443e-03) (hash(x)=41470963)
157
+ 1290 train 5.427097 (lr=2.8389e-03) (hash(x)=40729366)
158
+ 1300 val loss 5.4719
159
+ 1300 val perplexity 237.9014
160
+ 1300 train 5.413674 (lr=2.8333e-03) (hash(x)=39916663)
161
+ 1310 train 5.519399 (lr=2.8277e-03) (hash(x)=42179962)
162
+ 1320 train 5.238476 (lr=2.8220e-03) (hash(x)=44991932)
163
+ 1330 train 5.278906 (lr=2.8162e-03) (hash(x)=38925911)
164
+ 1340 train 5.278331 (lr=2.8103e-03) (hash(x)=38565560)
165
+ 1350 train 5.418691 (lr=2.8044e-03) (hash(x)=43384902)
166
+ 1360 train 5.406930 (lr=2.7983e-03) (hash(x)=45422261)
167
+ 1370 train 5.447308 (lr=2.7922e-03) (hash(x)=40167325)
168
+ 1380 train 5.703118 (lr=2.7860e-03) (hash(x)=51258162)
169
+ 1390 train 5.751878 (lr=2.7797e-03) (hash(x)=43229106)
170
+ 1400 val loss 5.7386
171
+ 1400 val perplexity 310.6315
172
+ 1400 train 5.689313 (lr=2.7733e-03) (hash(x)=42537494)
173
+ 1410 train 6.048205 (lr=2.7668e-03) (hash(x)=51236490)
174
+ 1420 train 5.968264 (lr=2.7603e-03) (hash(x)=45243347)
175
+ 1430 train 5.821448 (lr=2.7536e-03) (hash(x)=37658566)
176
+ 1440 train 5.874043 (lr=2.7469e-03) (hash(x)=38856106)
177
+ 1450 train 5.980742 (lr=2.7401e-03) (hash(x)=36649488)
178
+ 1460 train 6.060405 (lr=2.7333e-03) (hash(x)=40445407)
179
+ 1470 train 6.028110 (lr=2.7263e-03) (hash(x)=39916387)
180
+ 1480 train 6.044593 (lr=2.7193e-03) (hash(x)=40934555)
181
+ 1490 train 6.159947 (lr=2.7121e-03) (hash(x)=42276527)
182
+ 1500 val loss 6.3099
183
+ 1500 val perplexity 550.0048
184
+ 1500 train 6.252234 (lr=2.7050e-03) (hash(x)=37973883)
185
+ 1510 train 6.324541 (lr=2.6977e-03) (hash(x)=38929286)
186
+ 1520 train 5.994669 (lr=2.6903e-03) (hash(x)=36855837)
187
+ 1530 train 6.070149 (lr=2.6829e-03) (hash(x)=42146431)
188
+ 1540 train 6.112195 (lr=2.6754e-03) (hash(x)=41898411)
189
+ 1550 train 6.070686 (lr=2.6678e-03) (hash(x)=44906272)
190
+ 1560 train 6.035686 (lr=2.6602e-03) (hash(x)=40414353)
191
+ 1570 train 5.987549 (lr=2.6525e-03) (hash(x)=38723190)
192
+ 1580 train 6.090653 (lr=2.6447e-03) (hash(x)=39065271)
193
+ 1590 train 6.346761 (lr=2.6368e-03) (hash(x)=40000886)
194
+ 1600 val loss 6.5900
195
+ 1600 val perplexity 727.7643
196
+ 1600 train 6.449279 (lr=2.6289e-03) (hash(x)=37898571)
197
+ 1610 train 6.330434 (lr=2.6208e-03) (hash(x)=40447863)
198
+ 1620 train 6.208009 (lr=2.6128e-03) (hash(x)=40119318)
199
+ 1630 train 6.231786 (lr=2.6046e-03) (hash(x)=40664074)
200
+ 1640 train 6.370941 (lr=2.5964e-03) (hash(x)=37404741)
201
+ 1650 train 6.462447 (lr=2.5881e-03) (hash(x)=39214928)
202
+ 1660 train 6.621811 (lr=2.5797e-03) (hash(x)=35005019)
203
+ 1670 train 6.923933 (lr=2.5713e-03) (hash(x)=37675832)
204
+ 1680 train 6.843942 (lr=2.5628e-03) (hash(x)=40489680)
205
+ 1690 train 6.939161 (lr=2.5542e-03) (hash(x)=45537879)
206
+ 1700 val loss 7.1157
207
+ 1700 val perplexity 1231.0914
208
+ 1700 train 7.018696 (lr=2.5455e-03) (hash(x)=38866100)
209
+ 1710 train 7.228320 (lr=2.5368e-03) (hash(x)=44726254)
210
+ 1720 train 7.465189 (lr=2.5281e-03) (hash(x)=36917792)
211
+ 1730 train 7.342657 (lr=2.5192e-03) (hash(x)=42814805)
212
+ 1740 train 7.267575 (lr=2.5103e-03) (hash(x)=51812216)
213
+ 1750 train 7.100586 (lr=2.5014e-03) (hash(x)=33549014)
214
+ 1760 train 7.307338 (lr=2.4924e-03) (hash(x)=40354215)
215
+ 1770 train 7.191329 (lr=2.4833e-03) (hash(x)=40718606)
216
+ 1780 train 7.128210 (lr=2.4741e-03) (hash(x)=38274164)
217
+ 1790 train 7.235591 (lr=2.4649e-03) (hash(x)=42531471)
218
+ 1800 val loss 7.3341
219
+ 1800 val perplexity 1531.6322
220
+ 1800 train 7.260313 (lr=2.4556e-03) (hash(x)=35616519)
221
+ 1810 train 7.274873 (lr=2.4463e-03) (hash(x)=33803118)
222
+ 1820 train 7.194807 (lr=2.4369e-03) (hash(x)=36973525)
223
+ 1830 train 7.362013 (lr=2.4275e-03) (hash(x)=38162549)
224
+ 1840 train 7.271590 (lr=2.4180e-03) (hash(x)=33937159)
225
+ 1850 train 7.390532 (lr=2.4084e-03) (hash(x)=39887546)
226
+ 1860 train 7.679813 (lr=2.3988e-03) (hash(x)=37818525)
227
+ 1870 train 7.198900 (lr=2.3891e-03) (hash(x)=37250478)
228
+ 1880 train 7.338978 (lr=2.3794e-03) (hash(x)=36412167)
229
+ 1890 train 7.143837 (lr=2.3696e-03) (hash(x)=40261189)
230
+ 1900 val loss 7.2141
231
+ 1900 val perplexity 1358.4225
232
+ 1900 train 6.980299 (lr=2.3598e-03) (hash(x)=38654303)
233
+ 1910 train 7.204922 (lr=2.3499e-03) (hash(x)=37929515)
234
+ 1920 train 7.211288 (lr=2.3400e-03) (hash(x)=45580146)
235
+ 1930 train 7.226940 (lr=2.3300e-03) (hash(x)=32386330)
236
+ 1940 train 7.285474 (lr=2.3200e-03) (hash(x)=36331864)
237
+ 1950 train 7.577724 (lr=2.3099e-03) (hash(x)=41181727)
238
+ 1960 train 7.690816 (lr=2.2998e-03) (hash(x)=42705152)
239
+ 1970 train 7.538803 (lr=2.2896e-03) (hash(x)=40251511)
240
+ 1980 train 7.838525 (lr=2.2793e-03) (hash(x)=37525551)
241
+ 1990 train 7.793346 (lr=2.2691e-03) (hash(x)=41849618)
242
+ 2000 val loss 7.6526
243
+ 2000 val perplexity 2106.0215
244
+ 2000 train 7.543337 (lr=2.2588e-03) (hash(x)=37642582)
245
+ 2010 train 7.692922 (lr=2.2484e-03) (hash(x)=42131121)
246
+ 2020 train 7.784511 (lr=2.2380e-03) (hash(x)=39000209)
247
+ 2030 train 7.614171 (lr=2.2275e-03) (hash(x)=43641355)
248
+ 2040 train 7.624671 (lr=2.2170e-03) (hash(x)=47910507)
249
+ 2050 train 7.479917 (lr=2.2065e-03) (hash(x)=36670359)
250
+ 2060 train 7.562625 (lr=2.1959e-03) (hash(x)=36477755)
251
+ 2070 train 7.631532 (lr=2.1853e-03) (hash(x)=43262487)
252
+ 2080 train 7.628633 (lr=2.1746e-03) (hash(x)=49546029)
253
+ 2090 train 7.469494 (lr=2.1639e-03) (hash(x)=39924731)
254
+ 2100 val loss 7.5281
255
+ 2100 val perplexity 1859.5743
256
+ 2100 train 7.510253 (lr=2.1532e-03) (hash(x)=39921304)
257
+ 2110 train 7.455194 (lr=2.1424e-03) (hash(x)=38152788)
258
+ 2120 train 7.475663 (lr=2.1316e-03) (hash(x)=37977911)
259
+ 2130 train 7.449464 (lr=2.1208e-03) (hash(x)=34748760)
260
+ 2140 train 7.503266 (lr=2.1099e-03) (hash(x)=40509369)
261
+ 2150 train 7.491549 (lr=2.0990e-03) (hash(x)=37654262)
262
+ 2160 train 7.405478 (lr=2.0881e-03) (hash(x)=38139543)
263
+ 2170 train 7.489951 (lr=2.0771e-03) (hash(x)=42501806)
264
+ 2180 train 7.775397 (lr=2.0661e-03) (hash(x)=40085092)
265
+ 2190 train 7.753934 (lr=2.0550e-03) (hash(x)=50480193)
266
+ 2200 val loss 7.7079
267
+ 2200 val perplexity 2225.7729
268
+ 2200 train 7.678916 (lr=2.0440e-03) (hash(x)=40604084)
269
+ 2210 train 7.737811 (lr=2.0329e-03) (hash(x)=41555823)
270
+ 2220 train 7.853206 (lr=2.0217e-03) (hash(x)=50441765)
271
+ 2230 train 7.621596 (lr=2.0106e-03) (hash(x)=39796580)
272
+ 2240 train 7.713783 (lr=1.9994e-03) (hash(x)=44127022)
273
+ 2250 train 7.549945 (lr=1.9882e-03) (hash(x)=37026826)
274
+ 2260 train 7.498643 (lr=1.9770e-03) (hash(x)=42133839)
275
+ 2270 train 7.458739 (lr=1.9657e-03) (hash(x)=38500664)
276
+ 2280 train 7.426044 (lr=1.9544e-03) (hash(x)=40538661)
277
+ 2290 train 7.490203 (lr=1.9431e-03) (hash(x)=51509210)
278
+ 2300 val loss 7.5670
279
+ 2300 val perplexity 1933.3729
280
+ 2300 train 7.486565 (lr=1.9318e-03) (hash(x)=41952328)
281
+ 2310 train 7.451881 (lr=1.9205e-03) (hash(x)=39758123)
282
+ 2320 train 7.668950 (lr=1.9091e-03) (hash(x)=51089268)
283
+ 2330 train 7.593307 (lr=1.8977e-03) (hash(x)=39767618)
284
+ 2340 train 7.572250 (lr=1.8863e-03) (hash(x)=40409617)
285
+ 2350 train 7.564095 (lr=1.8749e-03) (hash(x)=40349634)
286
+ 2360 train 7.463734 (lr=1.8635e-03) (hash(x)=31841172)
287
+ 2370 train 7.602013 (lr=1.8520e-03) (hash(x)=42720539)
288
+ 2380 train 7.530239 (lr=1.8406e-03) (hash(x)=40998632)
289
+ 2390 train 7.556315 (lr=1.8291e-03) (hash(x)=40615413)
290
+ 2400 val loss 7.5685
291
+ 2400 val perplexity 1936.1388
292
+ 2400 train 7.551105 (lr=1.8176e-03) (hash(x)=39373658)
293
+ 2410 train 7.514620 (lr=1.8061e-03) (hash(x)=35480858)
294
+ 2420 train 7.549085 (lr=1.7946e-03) (hash(x)=32241095)
295
+ 2430 train 7.541862 (lr=1.7830e-03) (hash(x)=36669715)
296
+ 2440 train 7.596268 (lr=1.7715e-03) (hash(x)=45768335)
297
+ 2450 train 7.682968 (lr=1.7600e-03) (hash(x)=39395055)
298
+ 2460 train 7.656946 (lr=1.7484e-03) (hash(x)=34899269)
299
+ 2470 train 7.728311 (lr=1.7368e-03) (hash(x)=43642420)
300
+ 2480 train 7.767736 (lr=1.7253e-03) (hash(x)=47099507)
301
+ 2490 train 7.600332 (lr=1.7137e-03) (hash(x)=38825558)
302
+ 2500 val loss 7.6413
303
+ 2500 val perplexity 2082.4756
304
+ 2500 train 7.600602 (lr=1.7021e-03) (hash(x)=39833804)
305
+ 2510 train 7.620942 (lr=1.6906e-03) (hash(x)=35051654)
306
+ 2520 train 7.517879 (lr=1.6790e-03) (hash(x)=40291109)
307
+ 2530 train 7.554146 (lr=1.6674e-03) (hash(x)=36915768)
308
+ 2540 train 7.621719 (lr=1.6558e-03) (hash(x)=37016308)
309
+ 2550 train 7.603454 (lr=1.6442e-03) (hash(x)=43221777)
310
+ 2560 train 7.554169 (lr=1.6326e-03) (hash(x)=37233207)
311
+ 2570 train 7.634029 (lr=1.6210e-03) (hash(x)=37510626)
312
+ 2580 train 7.502373 (lr=1.6094e-03) (hash(x)=39624656)
313
+ 2590 train 7.478727 (lr=1.5979e-03) (hash(x)=37647501)
314
+ 2600 val loss 7.5836
315
+ 2600 val perplexity 1965.6262
316
+ 2600 train 7.463779 (lr=1.5863e-03) (hash(x)=37200138)
317
+ 2610 train 7.446686 (lr=1.5747e-03) (hash(x)=43066970)
318
+ 2620 train 7.655196 (lr=1.5632e-03) (hash(x)=40503799)
319
+ 2630 train 7.614497 (lr=1.5516e-03) (hash(x)=45534660)
320
+ 2640 train 7.716490 (lr=1.5400e-03) (hash(x)=52358353)
321
+ 2650 train 7.571498 (lr=1.5285e-03) (hash(x)=41843599)
322
+ 2660 train 7.617067 (lr=1.5170e-03) (hash(x)=40519770)
323
+ 2670 train 7.611303 (lr=1.5054e-03) (hash(x)=37185602)
324
+ 2680 train 7.574914 (lr=1.4939e-03) (hash(x)=43805719)
325
+ 2690 train 7.463801 (lr=1.4824e-03) (hash(x)=39458250)
326
+ 2700 val loss 7.5672
327
+ 2700 val perplexity 1933.8007
328
+ 2700 train 7.509787 (lr=1.4709e-03) (hash(x)=39149255)
329
+ 2710 train 7.481989 (lr=1.4594e-03) (hash(x)=43544861)
330
+ 2720 train 7.510556 (lr=1.4480e-03) (hash(x)=36858197)
331
+ 2730 train 7.515767 (lr=1.4365e-03) (hash(x)=37971712)
332
+ 2740 train 7.571373 (lr=1.4251e-03) (hash(x)=42416578)
333
+ 2750 train 7.529066 (lr=1.4137e-03) (hash(x)=53554622)
334
+ 2760 train 7.645211 (lr=1.4023e-03) (hash(x)=40502545)
335
+ 2770 train 7.578244 (lr=1.3909e-03) (hash(x)=36086417)
336
+ 2780 train 7.585574 (lr=1.3795e-03) (hash(x)=36242982)
337
+ 2790 train 7.597087 (lr=1.3682e-03) (hash(x)=41433780)
338
+ 2800 val loss 7.5658
339
+ 2800 val perplexity 1931.0686
340
+ 2800 train 7.556196 (lr=1.3569e-03) (hash(x)=48553484)
341
+ 2810 train 7.509312 (lr=1.3456e-03) (hash(x)=39928650)
342
+ 2820 train 7.469609 (lr=1.3343e-03) (hash(x)=37891724)
343
+ 2830 train 7.430135 (lr=1.3230e-03) (hash(x)=34353412)
344
+ 2840 train 7.473318 (lr=1.3118e-03) (hash(x)=38486611)
345
+ 2850 train 7.388651 (lr=1.3006e-03) (hash(x)=32706934)
346
+ 2860 train 7.461636 (lr=1.2894e-03) (hash(x)=38693681)
347
+ 2870 train 7.393054 (lr=1.2783e-03) (hash(x)=38430800)
348
+ 2880 train 7.442659 (lr=1.2671e-03) (hash(x)=40619559)
349
+ 2890 train 7.491068 (lr=1.2560e-03) (hash(x)=42741066)
350
+ 2900 val loss 7.5328
351
+ 2900 val perplexity 1868.3708
352
+ 2900 train 7.601730 (lr=1.2450e-03) (hash(x)=38216091)
353
+ 2910 train 7.645418 (lr=1.2339e-03) (hash(x)=35447832)
354
+ 2920 train 7.552339 (lr=1.2229e-03) (hash(x)=39825190)
355
+ 2930 train 7.584529 (lr=1.2119e-03) (hash(x)=37789121)
356
+ 2940 train 7.776618 (lr=1.2010e-03) (hash(x)=44499116)
357
+ 2950 train 7.713644 (lr=1.1901e-03) (hash(x)=39669860)
358
+ 2960 train 7.624273 (lr=1.1792e-03) (hash(x)=51683741)
359
+ 2970 train 7.389065 (lr=1.1684e-03) (hash(x)=36399721)
360
+ 2980 train 7.530968 (lr=1.1576e-03) (hash(x)=42629700)
361
+ 2990 train 7.443809 (lr=1.1468e-03) (hash(x)=39263773)
362
+ 3000 val loss 7.5344
363
+ 3000 val perplexity 1871.2640
364
+ 3000 train 7.384265 (lr=1.1361e-03) (hash(x)=37920485)
365
+ 3010 train 7.429597 (lr=1.1254e-03) (hash(x)=40835161)
366
+ 3020 train 7.481008 (lr=1.1147e-03) (hash(x)=38381159)
367
+ 3030 train 7.431045 (lr=1.1041e-03) (hash(x)=35846270)
368
+ 3040 train 7.536244 (lr=1.0935e-03) (hash(x)=47516567)
369
+ 3050 train 7.659072 (lr=1.0830e-03) (hash(x)=35600311)
370
+ 3060 train 7.613459 (lr=1.0725e-03) (hash(x)=37775318)
371
+ 3070 train 7.541782 (lr=1.0620e-03) (hash(x)=39881333)
372
+ 3080 train 7.425495 (lr=1.0516e-03) (hash(x)=37310168)
373
+ 3090 train 7.454187 (lr=1.0412e-03) (hash(x)=43669978)
374
+ 3100 val loss 7.4177
375
+ 3100 val perplexity 1665.1560
376
+ 3100 train 7.673238 (lr=1.0309e-03) (hash(x)=46356797)
377
+ 3110 train 7.350338 (lr=1.0207e-03) (hash(x)=40814232)
378
+ 3120 train 7.367829 (lr=1.0104e-03) (hash(x)=42504837)
379
+ 3130 train 7.460392 (lr=1.0002e-03) (hash(x)=39693594)
380
+ 3140 train 7.408485 (lr=9.9011e-04) (hash(x)=40277645)
381
+ 3150 train 7.455748 (lr=9.8002e-04) (hash(x)=40988003)
382
+ 3160 train 7.465854 (lr=9.6999e-04) (hash(x)=38875266)
383
+ 3170 train 7.529431 (lr=9.6000e-04) (hash(x)=45232173)
384
+ 3180 train 7.431017 (lr=9.5007e-04) (hash(x)=39213336)
385
+ 3190 train 7.485411 (lr=9.4019e-04) (hash(x)=42118576)
386
+ 3200 val loss 7.4590
387
+ 3200 val perplexity 1735.4666
388
+ 3200 train 7.461901 (lr=9.3036e-04) (hash(x)=32884223)
389
+ 3210 train 7.569977 (lr=9.2058e-04) (hash(x)=41276800)
390
+ 3220 train 7.542021 (lr=9.1085e-04) (hash(x)=40284461)
391
+ 3230 train 7.550803 (lr=9.0118e-04) (hash(x)=40566734)
392
+ 3240 train 7.631067 (lr=8.9157e-04) (hash(x)=36484570)
393
+ 3250 train 7.460586 (lr=8.8201e-04) (hash(x)=41642338)
394
+ 3260 train 7.435303 (lr=8.7251e-04) (hash(x)=43883570)
395
+ 3270 train 7.458158 (lr=8.6307e-04) (hash(x)=40432560)
396
+ 3280 train 7.496436 (lr=8.5368e-04) (hash(x)=38002717)
397
+ 3290 train 7.367868 (lr=8.4435e-04) (hash(x)=41926004)
398
+ 3300 val loss 7.4920
399
+ 3300 val perplexity 1793.6898
400
+ 3300 train 7.574133 (lr=8.3508e-04) (hash(x)=46015509)
401
+ 3310 train 7.482028 (lr=8.2588e-04) (hash(x)=40112249)
402
+ 3320 train 7.531399 (lr=8.1673e-04) (hash(x)=49162296)
403
+ 3330 train 7.467478 (lr=8.0764e-04) (hash(x)=46606969)
404
+ 3340 train 7.476029 (lr=7.9862e-04) (hash(x)=31128992)
405
+ 3350 train 7.474164 (lr=7.8966e-04) (hash(x)=41232534)
406
+ 3360 train 7.458097 (lr=7.8076e-04) (hash(x)=41599699)
407
+ 3370 train 7.428123 (lr=7.7192e-04) (hash(x)=40885280)
408
+ 3380 train 7.451911 (lr=7.6315e-04) (hash(x)=37169148)
409
+ 3390 train 7.449074 (lr=7.5445e-04) (hash(x)=46581889)
410
+ 3400 val loss 7.5220
411
+ 3400 val perplexity 1848.3330
412
+ 3400 train 7.593830 (lr=7.4581e-04) (hash(x)=37612074)
413
+ 3410 train 7.508540 (lr=7.3724e-04) (hash(x)=38763316)
414
+ 3420 train 7.349893 (lr=7.2874e-04) (hash(x)=40562379)
415
+ 3430 train 7.511331 (lr=7.2030e-04) (hash(x)=44724867)
416
+ 3440 train 7.540870 (lr=7.1193e-04) (hash(x)=39386624)
417
+ 3450 train 7.500563 (lr=7.0363e-04) (hash(x)=38652923)
418
+ 3460 train 7.450633 (lr=6.9541e-04) (hash(x)=29587379)
419
+ 3470 train 7.493032 (lr=6.8725e-04) (hash(x)=38228776)
420
+ 3480 train 7.350397 (lr=6.7916e-04) (hash(x)=30076039)
421
+ 3490 train 7.453751 (lr=6.7114e-04) (hash(x)=38398908)
422
+ 3500 val loss 7.4124
423
+ 3500 val perplexity 1656.4747
424
+ 3500 train 7.438497 (lr=6.6320e-04) (hash(x)=39259918)
425
+ 3510 train 7.434005 (lr=6.5533e-04) (hash(x)=40664091)
426
+ 3520 train 7.404539 (lr=6.4753e-04) (hash(x)=40309647)
427
+ 3530 train 7.412279 (lr=6.3981e-04) (hash(x)=38424801)
428
+ 3540 train 7.568933 (lr=6.3216e-04) (hash(x)=51322307)
429
+ 3550 train 7.330038 (lr=6.2458e-04) (hash(x)=38192628)
430
+ 3560 train 7.432320 (lr=6.1708e-04) (hash(x)=41563952)
431
+ 3570 train 7.350614 (lr=6.0966e-04) (hash(x)=39508843)
432
+ 3580 train 7.400346 (lr=6.0231e-04) (hash(x)=41260225)
433
+ 3590 train 7.564724 (lr=5.9504e-04) (hash(x)=49098107)
434
+ 3600 val loss 7.4197
435
+ 3600 val perplexity 1668.5808
436
+ 3600 train 7.505986 (lr=5.8785e-04) (hash(x)=41194370)
437
+ 3610 train 7.414118 (lr=5.8074e-04) (hash(x)=40399152)
438
+ 3620 train 7.512179 (lr=5.7370e-04) (hash(x)=41625018)
439
+ 3630 train 7.451025 (lr=5.6675e-04) (hash(x)=44077942)
440
+ 3640 train 7.407462 (lr=5.5987e-04) (hash(x)=42427512)
441
+ 3650 train 7.525454 (lr=5.5308e-04) (hash(x)=40067454)
442
+ 3660 train 7.416077 (lr=5.4636e-04) (hash(x)=37969892)
443
+ 3670 train 7.355086 (lr=5.3973e-04) (hash(x)=43670246)
444
+ 3680 train 7.374270 (lr=5.3318e-04) (hash(x)=37031732)
445
+ 3690 train 7.314882 (lr=5.2671e-04) (hash(x)=39237187)
446
+ 3700 val loss 7.4216
447
+ 3700 val perplexity 1671.7705
448
+ 3700 train 7.392312 (lr=5.2033e-04) (hash(x)=39521416)
449
+ 3710 train 7.347453 (lr=5.1402e-04) (hash(x)=41147425)
450
+ 3720 train 7.344787 (lr=5.0780e-04) (hash(x)=40204658)
451
+ 3730 train 7.404231 (lr=5.0167e-04) (hash(x)=39420983)
452
+ 3740 train 7.551250 (lr=4.9562e-04) (hash(x)=51155740)
453
+ 3750 train 7.493832 (lr=4.8965e-04) (hash(x)=41654187)
454
+ 3760 train 7.405545 (lr=4.8377e-04) (hash(x)=40598339)
455
+ 3770 train 7.410918 (lr=4.7798e-04) (hash(x)=44950116)
456
+ 3780 train 7.384638 (lr=4.7227e-04) (hash(x)=40660326)
457
+ 3790 train 7.475501 (lr=4.6665e-04) (hash(x)=37805851)
458
+ 3800 val loss 7.4175
459
+ 3800 val perplexity 1664.8479
460
+ 3800 train 7.402789 (lr=4.6112e-04) (hash(x)=38064443)
461
+ 3810 train 7.410206 (lr=4.5567e-04) (hash(x)=39161498)
462
+ 3820 train 7.370234 (lr=4.5031e-04) (hash(x)=40626649)
463
+ 3830 train 7.365262 (lr=4.4504e-04) (hash(x)=36894771)
464
+ 3840 train 7.291443 (lr=4.3986e-04) (hash(x)=33672141)
465
+ 3850 train 7.348138 (lr=4.3477e-04) (hash(x)=41751813)
466
+ 3860 train 7.249591 (lr=4.2977e-04) (hash(x)=33784172)
467
+ 3870 train 7.209717 (lr=4.2486e-04) (hash(x)=43110493)
468
+ 3880 train 7.284732 (lr=4.2004e-04) (hash(x)=36585111)
469
+ 3890 train 7.281297 (lr=4.1530e-04) (hash(x)=40134264)
470
+ 3900 val loss 7.4114
471
+ 3900 val perplexity 1654.6732
472
+ 3900 train 7.350392 (lr=4.1066e-04) (hash(x)=39815215)
473
+ 3910 train 7.481180 (lr=4.0611e-04) (hash(x)=42081557)
474
+ 3920 train 7.548090 (lr=4.0166e-04) (hash(x)=48184973)
475
+ 3930 train 7.360465 (lr=3.9729e-04) (hash(x)=42416681)
476
+ 3940 train 7.453910 (lr=3.9302e-04) (hash(x)=38587062)
477
+ 3950 train 7.344325 (lr=3.8884e-04) (hash(x)=37296617)
478
+ 3960 train 7.455273 (lr=3.8475e-04) (hash(x)=37959330)
479
+ 3970 train 7.279537 (lr=3.8076e-04) (hash(x)=35175634)
480
+ 3980 train 7.375590 (lr=3.7685e-04) (hash(x)=37219362)
481
+ 3990 train 7.418356 (lr=3.7305e-04) (hash(x)=38906132)
482
+ 4000 val loss 7.4194
483
+ 4000 val perplexity 1668.0804
484
+ 4000 train 7.328425 (lr=3.6933e-04) (hash(x)=39940517)
485
+ 4010 train 7.341803 (lr=3.6572e-04) (hash(x)=39952437)
486
+ 4020 train 7.349424 (lr=3.6219e-04) (hash(x)=42000930)
487
+ 4030 train 7.276704 (lr=3.5876e-04) (hash(x)=41460672)
488
+ 4040 train 7.219376 (lr=3.5543e-04) (hash(x)=43702329)
489
+ 4050 train 7.184711 (lr=3.5219e-04) (hash(x)=41740121)
490
+ 4060 train 7.250375 (lr=3.4905e-04) (hash(x)=37066349)
491
+ 4070 train 7.172188 (lr=3.4600e-04) (hash(x)=38359869)
492
+ 4080 train 7.159321 (lr=3.4305e-04) (hash(x)=35131370)
493
+ 4090 train 7.402943 (lr=3.4019e-04) (hash(x)=39918811)
494
+ 4100 val loss 7.3421
495
+ 4100 val perplexity 1543.9807
496
+ 4100 train 7.394500 (lr=3.3744e-04) (hash(x)=47036374)
497
+ 4110 train 7.359123 (lr=3.3477e-04) (hash(x)=37657523)
498
+ 4120 train 7.375312 (lr=3.3221e-04) (hash(x)=37218703)
499
+ 4130 train 7.416271 (lr=3.2974e-04) (hash(x)=41350513)
500
+ 4140 train 7.383940 (lr=3.2737e-04) (hash(x)=38283785)
501
+ 4150 train 7.333607 (lr=3.2510e-04) (hash(x)=32927892)
502
+ 4160 train 7.264695 (lr=3.2292e-04) (hash(x)=42011933)
503
+ 4170 train 7.346141 (lr=3.2085e-04) (hash(x)=41854594)
504
+ 4180 train 7.362872 (lr=3.1887e-04) (hash(x)=36737064)
505
+ 4190 train 7.235276 (lr=3.1699e-04) (hash(x)=36326176)
506
+ 4200 val loss 7.3564
507
+ 4200 val perplexity 1566.2621
508
+ 4200 train 7.339328 (lr=3.1520e-04) (hash(x)=39106683)
509
+ 4210 train 7.342415 (lr=3.1352e-04) (hash(x)=39759977)
510
+ 4220 train 7.193846 (lr=3.1193e-04) (hash(x)=43666151)
511
+ 4230 train 7.231940 (lr=3.1044e-04) (hash(x)=38154367)
512
+ 4240 train 7.196908 (lr=3.0905e-04) (hash(x)=38284168)
513
+ 4250 train 7.334686 (lr=3.0776e-04) (hash(x)=38888879)
514
+ 4260 train 7.186014 (lr=3.0657e-04) (hash(x)=36846417)
515
+ 4270 train 7.405221 (lr=3.0548e-04) (hash(x)=43125375)
516
+ 4280 train 7.340114 (lr=3.0449e-04) (hash(x)=43556412)
517
+ 4290 train 7.378199 (lr=3.0359e-04) (hash(x)=48598302)
518
+ 4300 val loss 7.3714
519
+ 4300 val perplexity 1589.8192
520
+ 4300 train 7.369385 (lr=3.0280e-04) (hash(x)=42764385)
521
+ 4310 train 7.619712 (lr=3.0210e-04) (hash(x)=41190028)
522
+ 4320 train 7.349436 (lr=3.0150e-04) (hash(x)=35767770)
523
+ 4330 train 7.401464 (lr=3.0101e-04) (hash(x)=36760476)
524
+ 4340 train 7.432033 (lr=3.0061e-04) (hash(x)=40618921)
525
+ 4350 train 7.316902 (lr=3.0031e-04) (hash(x)=32300164)
526
+ 4360 train 7.298217 (lr=3.0011e-04) (hash(x)=41400085)
527
+ 4370 train 7.409771 (lr=3.0001e-04) (hash(x)=40251828)
528
+ 4374 val loss 7.3583
529
+ 4374 val perplexity 1569.2164
logs/fix_1_latent_mask/1_latent_mask_lr_30e-4_compile_seed_1340/model_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:21dd7fbf706e7961ca030e9b3fd6ad42949c41ba890763491bebda3ec83c0f19
3
+ size 97706546
logs/fix_1_latent_mask/1_latent_mask_lr_30e-4_compile_seed_1340/optimizer_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:935981e82332b8a751fbeccffb562bb4f2217857ecf5e5258c809167ad954b0e
3
+ size 189135414