andrew-healey commited on
Commit
1112bf1
·
verified ·
1 Parent(s): 2e4766e

Upload folder using huggingface_hub

Browse files
latent_masks_learnable/learnable_no_compile_seed_1339/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "latent_masks_learnable/learnable_no_compile_seed_1339", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 12, "n_embd": 264, "head_dim": 22, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": 4375, "warmup_steps": 250, "group": "latent_masks_learnable", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1339, "memory_penalty_epsilon": 0.1, "selection_head_linear_combo": "n_latent_masks", "selection_head_linear_combo_scale": 1.0, "disable_selection_head_linear_combo_bias": false, "assert_latent_matches_no_head": false, "n_latent_masks": 1, "init_latent_masks_to_identity": true, "init_latent_masks_to_inverse": false, "latent_mask_scale": null, "latent_mask_runtime_multiplier": null, "latent_mask_sigmoid": false, "S_layernorm": false, "one_head_per_latent_mask": false, "protection_kind": "none", "leaky_relu_alpha": null, "leaky_relu_bias": null, "use_compile": false, "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, "mask_layernorm": false, "residual_attention_masks": false, "compute_base_shapes": false, "base_shapes_savefile": null, "mup": true, "disable_selection": false, "mup_enable_coord_check_logging": false, "max_lr": 0.003, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "learnable_no_compile", "latent_mask_precision": "float32"}
latent_masks_learnable/learnable_no_compile_seed_1339/dataloader_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6031fd3e2855a036f7a5531cc24555aabd1115f9dd6618b8b2ca6f55279ef0b2
3
+ size 964
latent_masks_learnable/learnable_no_compile_seed_1339/log2.txt ADDED
@@ -0,0 +1,548 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ max_steps: 4375
2
+ 0 val loss 11.2160
3
+ 0 val perplexity 74309.4844
4
+ 0 hella 0.2473
5
+ 0 train 11.205509 (lr=4.1958e-06) (hash(x)=44992657)
6
+ 10 train 9.890629 (lr=4.6154e-05) (hash(x)=33468021)
7
+ 20 train 9.339613 (lr=8.8112e-05) (hash(x)=40941803)
8
+ 30 train 8.646135 (lr=1.3007e-04) (hash(x)=36715902)
9
+ 40 train 7.913006 (lr=1.7203e-04) (hash(x)=32710993)
10
+ 50 train 7.694913 (lr=2.1399e-04) (hash(x)=43839896)
11
+ 60 train 7.752412 (lr=2.5594e-04) (hash(x)=40675468)
12
+ 70 train 7.443704 (lr=2.9790e-04) (hash(x)=34592631)
13
+ 80 train 7.431669 (lr=3.3986e-04) (hash(x)=44444845)
14
+ 90 train 7.227595 (lr=3.8182e-04) (hash(x)=41965258)
15
+ 100 val loss 7.1113
16
+ 100 val perplexity 1225.7277
17
+ 100 train 6.971700 (lr=4.2378e-04) (hash(x)=41284750)
18
+ 110 train 6.922056 (lr=4.6573e-04) (hash(x)=41118734)
19
+ 120 train 6.751715 (lr=5.0769e-04) (hash(x)=37537547)
20
+ 130 train 6.583852 (lr=5.4965e-04) (hash(x)=43625179)
21
+ 140 train 6.560868 (lr=5.9161e-04) (hash(x)=41940760)
22
+ 150 train 6.611235 (lr=6.3357e-04) (hash(x)=39210431)
23
+ 160 train 6.547588 (lr=6.7552e-04) (hash(x)=41128294)
24
+ 170 train 6.424761 (lr=7.1748e-04) (hash(x)=41590227)
25
+ 180 train 6.363883 (lr=7.5944e-04) (hash(x)=38084871)
26
+ 190 train 6.266884 (lr=8.0140e-04) (hash(x)=34534333)
27
+ 200 val loss 6.1809
28
+ 200 val perplexity 483.4029
29
+ 200 train 6.215993 (lr=8.4336e-04) (hash(x)=41614299)
30
+ 210 train 6.097910 (lr=8.8531e-04) (hash(x)=35551597)
31
+ 220 train 6.021095 (lr=9.2727e-04) (hash(x)=44615432)
32
+ 230 train 6.152116 (lr=9.6923e-04) (hash(x)=37078768)
33
+ 240 train 5.841511 (lr=1.0112e-03) (hash(x)=40644078)
34
+ 250 hella 0.2426
35
+ 250 train 5.897739 (lr=1.0531e-03) (hash(x)=44636066)
36
+ 260 train 5.819059 (lr=1.0951e-03) (hash(x)=43247952)
37
+ 270 train 5.729167 (lr=1.1371e-03) (hash(x)=37989370)
38
+ 280 train 5.659914 (lr=1.1790e-03) (hash(x)=36229730)
39
+ 290 train 5.587878 (lr=1.2210e-03) (hash(x)=35032996)
40
+ 300 val loss 5.7186
41
+ 300 val perplexity 304.4671
42
+ 300 train 5.467095 (lr=1.2629e-03) (hash(x)=33868636)
43
+ 310 train 5.389896 (lr=1.3049e-03) (hash(x)=39782705)
44
+ 320 train 5.486048 (lr=1.3469e-03) (hash(x)=43525765)
45
+ 330 train 5.584064 (lr=1.3888e-03) (hash(x)=38074926)
46
+ 340 train 5.634186 (lr=1.4308e-03) (hash(x)=50125702)
47
+ 350 train 5.608922 (lr=1.4727e-03) (hash(x)=35062222)
48
+ 360 train 5.534294 (lr=1.5147e-03) (hash(x)=39671001)
49
+ 370 train 5.558186 (lr=1.5566e-03) (hash(x)=35748364)
50
+ 380 train 5.368624 (lr=1.5986e-03) (hash(x)=39962666)
51
+ 390 train 5.341872 (lr=1.6406e-03) (hash(x)=39213360)
52
+ 400 val loss 5.3633
53
+ 400 val perplexity 213.4312
54
+ 400 train 5.338474 (lr=1.6825e-03) (hash(x)=37919584)
55
+ 410 train 5.301454 (lr=1.7245e-03) (hash(x)=42001527)
56
+ 420 train 5.294563 (lr=1.7664e-03) (hash(x)=33940005)
57
+ 430 train 5.326714 (lr=1.8084e-03) (hash(x)=41749217)
58
+ 440 train 5.221928 (lr=1.8503e-03) (hash(x)=45219739)
59
+ 450 train 5.210794 (lr=1.8923e-03) (hash(x)=38515090)
60
+ 460 train 5.109742 (lr=1.9343e-03) (hash(x)=37518607)
61
+ 470 train 4.952476 (lr=1.9762e-03) (hash(x)=46830716)
62
+ 480 train 5.076671 (lr=2.0182e-03) (hash(x)=49262598)
63
+ 490 train 4.870680 (lr=2.0601e-03) (hash(x)=41370559)
64
+ 500 val loss 5.1900
65
+ 500 val perplexity 179.4669
66
+ 500 hella 0.2368
67
+ 500 train 4.964232 (lr=2.1021e-03) (hash(x)=35237459)
68
+ 510 train 5.094709 (lr=2.1441e-03) (hash(x)=35070579)
69
+ 520 train 5.122880 (lr=2.1860e-03) (hash(x)=37355110)
70
+ 530 train 5.212308 (lr=2.2280e-03) (hash(x)=39398044)
71
+ 540 train 5.140265 (lr=2.2699e-03) (hash(x)=39156676)
72
+ 550 train 5.091932 (lr=2.3119e-03) (hash(x)=38553909)
73
+ 560 train 5.129451 (lr=2.3538e-03) (hash(x)=36555371)
74
+ 570 train 5.077563 (lr=2.3958e-03) (hash(x)=42048991)
75
+ 580 train 5.026538 (lr=2.4378e-03) (hash(x)=37359399)
76
+ 590 train 5.033009 (lr=2.4797e-03) (hash(x)=39254663)
77
+ 600 val loss 4.9928
78
+ 600 val perplexity 147.3426
79
+ 600 train 5.007493 (lr=2.5217e-03) (hash(x)=38982539)
80
+ 610 train 4.925224 (lr=2.5636e-03) (hash(x)=35859145)
81
+ 620 train 4.901142 (lr=2.6056e-03) (hash(x)=35593542)
82
+ 630 train 4.801295 (lr=2.6476e-03) (hash(x)=39124418)
83
+ 640 train 4.807037 (lr=2.6895e-03) (hash(x)=39107374)
84
+ 650 train 4.735119 (lr=2.7315e-03) (hash(x)=34529151)
85
+ 660 train 4.648462 (lr=2.7734e-03) (hash(x)=39483931)
86
+ 670 train 4.647891 (lr=2.8154e-03) (hash(x)=43008010)
87
+ 680 train 4.548888 (lr=2.8573e-03) (hash(x)=33469328)
88
+ 690 train 4.593317 (lr=2.8993e-03) (hash(x)=35025253)
89
+ 700 val loss 4.8314
90
+ 700 val perplexity 125.3819
91
+ 700 train 4.562597 (lr=2.9413e-03) (hash(x)=43280188)
92
+ 710 train 4.844040 (lr=2.9832e-03) (hash(x)=40999263)
93
+ 720 train 4.834510 (lr=3.0000e-03) (hash(x)=38304742)
94
+ 730 train 4.832801 (lr=2.9999e-03) (hash(x)=34705188)
95
+ 740 train 4.785991 (lr=2.9997e-03) (hash(x)=40028964)
96
+ 750 hella 0.2534
97
+ 750 train 4.730123 (lr=2.9994e-03) (hash(x)=39051591)
98
+ 760 train 4.777149 (lr=2.9990e-03) (hash(x)=41286198)
99
+ 770 train 4.727296 (lr=2.9985e-03) (hash(x)=44512967)
100
+ 780 train 4.744031 (lr=2.9979e-03) (hash(x)=45171627)
101
+ 790 train 4.713219 (lr=2.9972e-03) (hash(x)=46345673)
102
+ 800 val loss 4.7088
103
+ 800 val perplexity 110.9171
104
+ 800 train 4.670616 (lr=2.9964e-03) (hash(x)=35529545)
105
+ 810 train 4.560767 (lr=2.9955e-03) (hash(x)=42321793)
106
+ 820 train 4.592834 (lr=2.9945e-03) (hash(x)=49327569)
107
+ 830 train 4.653664 (lr=2.9934e-03) (hash(x)=41325057)
108
+ 840 train 4.591049 (lr=2.9922e-03) (hash(x)=37305614)
109
+ 850 train 4.557293 (lr=2.9909e-03) (hash(x)=38000800)
110
+ 860 train 4.687657 (lr=2.9896e-03) (hash(x)=44047697)
111
+ 870 train 4.675812 (lr=2.9881e-03) (hash(x)=41576206)
112
+ 880 train 4.646525 (lr=2.9865e-03) (hash(x)=44685458)
113
+ 890 train 4.623953 (lr=2.9848e-03) (hash(x)=40398597)
114
+ 900 val loss 4.6057
115
+ 900 val perplexity 100.0579
116
+ 900 train 4.567166 (lr=2.9830e-03) (hash(x)=39312673)
117
+ 910 train 4.576794 (lr=2.9811e-03) (hash(x)=38881875)
118
+ 920 train 4.608295 (lr=2.9792e-03) (hash(x)=39140166)
119
+ 930 train 4.506325 (lr=2.9771e-03) (hash(x)=39075781)
120
+ 940 train 4.519354 (lr=2.9749e-03) (hash(x)=46887071)
121
+ 950 train 4.515584 (lr=2.9726e-03) (hash(x)=39740603)
122
+ 960 train 4.562313 (lr=2.9703e-03) (hash(x)=39590268)
123
+ 970 train 4.510700 (lr=2.9678e-03) (hash(x)=43580052)
124
+ 980 train 4.458999 (lr=2.9652e-03) (hash(x)=40226735)
125
+ 990 train 4.557684 (lr=2.9626e-03) (hash(x)=40723896)
126
+ 1000 val loss 4.5172
127
+ 1000 val perplexity 91.5812
128
+ 1000 hella 0.2496
129
+ 1000 train 4.518671 (lr=2.9598e-03) (hash(x)=39408050)
130
+ 1010 train 4.707730 (lr=2.9570e-03) (hash(x)=32247950)
131
+ 1020 train 4.595407 (lr=2.9540e-03) (hash(x)=39049938)
132
+ 1030 train 4.561467 (lr=2.9510e-03) (hash(x)=37453736)
133
+ 1040 train 4.465073 (lr=2.9478e-03) (hash(x)=36649666)
134
+ 1050 train 4.534705 (lr=2.9446e-03) (hash(x)=41404097)
135
+ 1060 train 4.524285 (lr=2.9412e-03) (hash(x)=40701599)
136
+ 1070 train 4.538585 (lr=2.9378e-03) (hash(x)=41787393)
137
+ 1080 train 4.426347 (lr=2.9343e-03) (hash(x)=42325465)
138
+ 1090 train 4.459751 (lr=2.9307e-03) (hash(x)=45018925)
139
+ 1100 val loss 4.4535
140
+ 1100 val perplexity 85.9300
141
+ 1100 train 4.482226 (lr=2.9270e-03) (hash(x)=53751982)
142
+ 1110 train 4.409569 (lr=2.9231e-03) (hash(x)=43118376)
143
+ 1120 train 4.434377 (lr=2.9192e-03) (hash(x)=41489942)
144
+ 1130 train 4.449849 (lr=2.9152e-03) (hash(x)=48020487)
145
+ 1140 train 4.320054 (lr=2.9112e-03) (hash(x)=43266072)
146
+ 1150 train 4.517145 (lr=2.9070e-03) (hash(x)=41790132)
147
+ 1160 train 4.479822 (lr=2.9027e-03) (hash(x)=33739090)
148
+ 1170 train 4.456712 (lr=2.8983e-03) (hash(x)=43885030)
149
+ 1180 train 4.405805 (lr=2.8939e-03) (hash(x)=39046714)
150
+ 1190 train 4.438168 (lr=2.8893e-03) (hash(x)=44387023)
151
+ 1200 val loss 4.3864
152
+ 1200 val perplexity 80.3544
153
+ 1200 train 4.300201 (lr=2.8847e-03) (hash(x)=36667141)
154
+ 1210 train 4.374377 (lr=2.8800e-03) (hash(x)=45487179)
155
+ 1220 train 4.355814 (lr=2.8751e-03) (hash(x)=43456413)
156
+ 1230 train 4.283891 (lr=2.8702e-03) (hash(x)=42473632)
157
+ 1240 train 4.369223 (lr=2.8652e-03) (hash(x)=42966315)
158
+ 1250 hella 0.2489
159
+ 1250 train 4.207786 (lr=2.8601e-03) (hash(x)=42930524)
160
+ 1260 train 4.359692 (lr=2.8550e-03) (hash(x)=43467676)
161
+ 1270 train 4.278492 (lr=2.8497e-03) (hash(x)=40872580)
162
+ 1280 train 4.356895 (lr=2.8443e-03) (hash(x)=40619157)
163
+ 1290 train 4.371532 (lr=2.8389e-03) (hash(x)=44406165)
164
+ 1300 val loss 4.3481
165
+ 1300 val perplexity 77.3328
166
+ 1300 train 4.331396 (lr=2.8333e-03) (hash(x)=40808029)
167
+ 1310 train 4.504255 (lr=2.8277e-03) (hash(x)=46125736)
168
+ 1320 train 4.270895 (lr=2.8220e-03) (hash(x)=41698487)
169
+ 1330 train 4.306168 (lr=2.8162e-03) (hash(x)=39302878)
170
+ 1340 train 4.419702 (lr=2.8103e-03) (hash(x)=41826369)
171
+ 1350 train 4.236773 (lr=2.8044e-03) (hash(x)=37061413)
172
+ 1360 train 4.356407 (lr=2.7983e-03) (hash(x)=40770974)
173
+ 1370 train 4.273082 (lr=2.7922e-03) (hash(x)=38738718)
174
+ 1380 train 4.266129 (lr=2.7860e-03) (hash(x)=34961558)
175
+ 1390 train 4.318196 (lr=2.7797e-03) (hash(x)=38559342)
176
+ 1400 val loss 4.3300
177
+ 1400 val perplexity 75.9474
178
+ 1400 train 4.270800 (lr=2.7733e-03) (hash(x)=38840020)
179
+ 1410 train 4.433578 (lr=2.7668e-03) (hash(x)=44448100)
180
+ 1420 train 4.484174 (lr=2.7603e-03) (hash(x)=39199838)
181
+ 1430 train 4.281663 (lr=2.7536e-03) (hash(x)=41669873)
182
+ 1440 train 4.406185 (lr=2.7469e-03) (hash(x)=53253836)
183
+ 1450 train 4.329373 (lr=2.7401e-03) (hash(x)=41320631)
184
+ 1460 train 4.249531 (lr=2.7333e-03) (hash(x)=48462803)
185
+ 1470 train 4.290498 (lr=2.7263e-03) (hash(x)=42356864)
186
+ 1480 train 4.340086 (lr=2.7193e-03) (hash(x)=41118068)
187
+ 1490 train 4.251418 (lr=2.7121e-03) (hash(x)=37242854)
188
+ 1500 val loss 4.2904
189
+ 1500 val perplexity 72.9977
190
+ 1500 hella 0.2495
191
+ 1500 train 4.303071 (lr=2.7050e-03) (hash(x)=40571031)
192
+ 1510 train 4.252349 (lr=2.6977e-03) (hash(x)=35384608)
193
+ 1520 train 4.283322 (lr=2.6903e-03) (hash(x)=35341056)
194
+ 1530 train 4.171772 (lr=2.6829e-03) (hash(x)=36471779)
195
+ 1540 train 4.205666 (lr=2.6754e-03) (hash(x)=40356153)
196
+ 1550 train 4.243480 (lr=2.6678e-03) (hash(x)=40133377)
197
+ 1560 train 4.177217 (lr=2.6602e-03) (hash(x)=33722160)
198
+ 1570 train 4.323147 (lr=2.6525e-03) (hash(x)=39555295)
199
+ 1580 train 4.228606 (lr=2.6447e-03) (hash(x)=35950666)
200
+ 1590 train 4.379121 (lr=2.6368e-03) (hash(x)=36528535)
201
+ 1600 val loss 4.2636
202
+ 1600 val perplexity 71.0643
203
+ 1600 train 4.267890 (lr=2.6289e-03) (hash(x)=42613017)
204
+ 1610 train 4.413244 (lr=2.6208e-03) (hash(x)=42845965)
205
+ 1620 train 4.305242 (lr=2.6128e-03) (hash(x)=36137074)
206
+ 1630 train 4.259404 (lr=2.6046e-03) (hash(x)=37560569)
207
+ 1640 train 4.353173 (lr=2.5964e-03) (hash(x)=43680614)
208
+ 1650 train 4.210605 (lr=2.5881e-03) (hash(x)=37281218)
209
+ 1660 train 4.239304 (lr=2.5797e-03) (hash(x)=36165222)
210
+ 1670 train 4.198078 (lr=2.5713e-03) (hash(x)=50222513)
211
+ 1680 train 4.085910 (lr=2.5628e-03) (hash(x)=35239173)
212
+ 1690 train 4.141810 (lr=2.5542e-03) (hash(x)=40287349)
213
+ 1700 val loss 4.2649
214
+ 1700 val perplexity 71.1567
215
+ 1700 train 4.317277 (lr=2.5455e-03) (hash(x)=37444780)
216
+ 1710 train 4.305974 (lr=2.5368e-03) (hash(x)=38962723)
217
+ 1720 train 4.305298 (lr=2.5281e-03) (hash(x)=33882305)
218
+ 1730 train 4.289834 (lr=2.5192e-03) (hash(x)=36501540)
219
+ 1740 train 4.290833 (lr=2.5103e-03) (hash(x)=38723266)
220
+ 1750 hella 0.2563
221
+ 1750 train 4.217854 (lr=2.5014e-03) (hash(x)=38772789)
222
+ 1760 train 4.291042 (lr=2.4924e-03) (hash(x)=37707898)
223
+ 1770 train 4.155387 (lr=2.4833e-03) (hash(x)=30529327)
224
+ 1780 train 4.239884 (lr=2.4741e-03) (hash(x)=37885464)
225
+ 1790 train 4.265522 (lr=2.4649e-03) (hash(x)=39013967)
226
+ 1800 val loss 4.2214
227
+ 1800 val perplexity 68.1302
228
+ 1800 train 4.291925 (lr=2.4556e-03) (hash(x)=34906955)
229
+ 1810 train 4.165086 (lr=2.4463e-03) (hash(x)=40790682)
230
+ 1820 train 4.037279 (lr=2.4369e-03) (hash(x)=39912223)
231
+ 1830 train 4.165802 (lr=2.4275e-03) (hash(x)=39304486)
232
+ 1840 train 4.140358 (lr=2.4180e-03) (hash(x)=37247975)
233
+ 1850 train 4.128611 (lr=2.4084e-03) (hash(x)=41340614)
234
+ 1860 train 4.286376 (lr=2.3988e-03) (hash(x)=39924627)
235
+ 1870 train 4.269129 (lr=2.3891e-03) (hash(x)=40235993)
236
+ 1880 train 4.154668 (lr=2.3794e-03) (hash(x)=37693476)
237
+ 1890 train 4.271746 (lr=2.3696e-03) (hash(x)=40449864)
238
+ 1900 val loss 4.1981
239
+ 1900 val perplexity 66.5564
240
+ 1900 train 4.178226 (lr=2.3598e-03) (hash(x)=37907749)
241
+ 1910 train 4.148198 (lr=2.3499e-03) (hash(x)=34547901)
242
+ 1920 train 4.179930 (lr=2.3400e-03) (hash(x)=37771788)
243
+ 1930 train 4.056910 (lr=2.3300e-03) (hash(x)=36049454)
244
+ 1940 train 4.147428 (lr=2.3200e-03) (hash(x)=29564336)
245
+ 1950 train 4.137128 (lr=2.3099e-03) (hash(x)=41689281)
246
+ 1960 train 4.056465 (lr=2.2998e-03) (hash(x)=41976699)
247
+ 1970 train 4.251668 (lr=2.2896e-03) (hash(x)=41538621)
248
+ 1980 train 4.132985 (lr=2.2793e-03) (hash(x)=36490902)
249
+ 1990 train 4.160331 (lr=2.2691e-03) (hash(x)=42739315)
250
+ 2000 val loss 4.1790
251
+ 2000 val perplexity 65.3003
252
+ 2000 hella 0.2553
253
+ 2000 train 4.181492 (lr=2.2588e-03) (hash(x)=38600074)
254
+ 2010 train 4.196176 (lr=2.2484e-03) (hash(x)=34555488)
255
+ 2020 train 4.218958 (lr=2.2380e-03) (hash(x)=37227345)
256
+ 2030 train 4.165638 (lr=2.2275e-03) (hash(x)=38172386)
257
+ 2040 train 4.277075 (lr=2.2170e-03) (hash(x)=39340786)
258
+ 2050 train 4.115651 (lr=2.2065e-03) (hash(x)=37237298)
259
+ 2060 train 4.360837 (lr=2.1959e-03) (hash(x)=35894782)
260
+ 2070 train 4.178912 (lr=2.1853e-03) (hash(x)=44161861)
261
+ 2080 train 4.194098 (lr=2.1746e-03) (hash(x)=38062558)
262
+ 2090 train 4.043975 (lr=2.1639e-03) (hash(x)=38608301)
263
+ 2100 val loss 4.1565
264
+ 2100 val perplexity 63.8478
265
+ 2100 train 4.079748 (lr=2.1532e-03) (hash(x)=47611537)
266
+ 2110 train 4.072173 (lr=2.1424e-03) (hash(x)=38076651)
267
+ 2120 train 4.086304 (lr=2.1316e-03) (hash(x)=39870893)
268
+ 2130 train 4.234252 (lr=2.1208e-03) (hash(x)=48123321)
269
+ 2140 train 4.260449 (lr=2.1099e-03) (hash(x)=44124977)
270
+ 2150 train 4.159056 (lr=2.0990e-03) (hash(x)=42181374)
271
+ 2160 train 4.188149 (lr=2.0881e-03) (hash(x)=41837878)
272
+ 2170 train 4.206254 (lr=2.0771e-03) (hash(x)=42161375)
273
+ 2180 train 4.172538 (lr=2.0661e-03) (hash(x)=41792986)
274
+ 2190 train 4.146170 (lr=2.0550e-03) (hash(x)=41872416)
275
+ 2200 val loss 4.1367
276
+ 2200 val perplexity 62.5965
277
+ 2200 train 4.119595 (lr=2.0440e-03) (hash(x)=37395985)
278
+ 2210 train 4.171786 (lr=2.0329e-03) (hash(x)=40616096)
279
+ 2220 train 4.067309 (lr=2.0217e-03) (hash(x)=40118423)
280
+ 2230 train 3.982944 (lr=2.0106e-03) (hash(x)=39641586)
281
+ 2240 train 4.077840 (lr=1.9994e-03) (hash(x)=34720116)
282
+ 2250 hella 0.2554
283
+ 2250 train 4.018040 (lr=1.9882e-03) (hash(x)=43641508)
284
+ 2260 train 4.040579 (lr=1.9770e-03) (hash(x)=31179786)
285
+ 2270 train 4.011246 (lr=1.9657e-03) (hash(x)=42721932)
286
+ 2280 train 4.341636 (lr=1.9544e-03) (hash(x)=38474505)
287
+ 2290 train 4.162724 (lr=1.9431e-03) (hash(x)=41398545)
288
+ 2300 val loss 4.1165
289
+ 2300 val perplexity 61.3472
290
+ 2300 train 4.091807 (lr=1.9318e-03) (hash(x)=44131094)
291
+ 2310 train 4.171144 (lr=1.9205e-03) (hash(x)=41667948)
292
+ 2320 train 4.250509 (lr=1.9091e-03) (hash(x)=36339232)
293
+ 2330 train 4.129472 (lr=1.8977e-03) (hash(x)=41630644)
294
+ 2340 train 4.215248 (lr=1.8863e-03) (hash(x)=41278955)
295
+ 2350 train 4.095100 (lr=1.8749e-03) (hash(x)=39491976)
296
+ 2360 train 4.150081 (lr=1.8635e-03) (hash(x)=41771238)
297
+ 2370 train 4.246689 (lr=1.8520e-03) (hash(x)=45374570)
298
+ 2380 train 4.128467 (lr=1.8406e-03) (hash(x)=41419267)
299
+ 2390 train 4.161371 (lr=1.8291e-03) (hash(x)=38869169)
300
+ 2400 val loss 4.1050
301
+ 2400 val perplexity 60.6414
302
+ 2400 train 4.058964 (lr=1.8176e-03) (hash(x)=38619293)
303
+ 2410 train 3.881442 (lr=1.8061e-03) (hash(x)=41600240)
304
+ 2420 train 4.012462 (lr=1.7946e-03) (hash(x)=40891045)
305
+ 2430 train 4.005966 (lr=1.7830e-03) (hash(x)=41023249)
306
+ 2440 train 3.971753 (lr=1.7715e-03) (hash(x)=33813452)
307
+ 2450 train 3.922422 (lr=1.7600e-03) (hash(x)=38464119)
308
+ 2460 train 3.822414 (lr=1.7484e-03) (hash(x)=40699982)
309
+ 2470 train 3.835708 (lr=1.7368e-03) (hash(x)=38254854)
310
+ 2480 train 4.116148 (lr=1.7253e-03) (hash(x)=42736069)
311
+ 2490 train 4.131783 (lr=1.7137e-03) (hash(x)=41928525)
312
+ 2500 val loss 4.0841
313
+ 2500 val perplexity 59.3911
314
+ 2500 hella 0.2532
315
+ 2500 train 4.085140 (lr=1.7021e-03) (hash(x)=45864011)
316
+ 2510 train 4.097974 (lr=1.6906e-03) (hash(x)=41380960)
317
+ 2520 train 4.095989 (lr=1.6790e-03) (hash(x)=38366545)
318
+ 2530 train 4.096006 (lr=1.6674e-03) (hash(x)=41045176)
319
+ 2540 train 4.124298 (lr=1.6558e-03) (hash(x)=43435705)
320
+ 2550 train 4.076407 (lr=1.6442e-03) (hash(x)=40693090)
321
+ 2560 train 3.995504 (lr=1.6326e-03) (hash(x)=40502478)
322
+ 2570 train 4.030334 (lr=1.6210e-03) (hash(x)=38610920)
323
+ 2580 train 3.950465 (lr=1.6094e-03) (hash(x)=38333499)
324
+ 2590 train 4.033562 (lr=1.5979e-03) (hash(x)=41982736)
325
+ 2600 val loss 4.0689
326
+ 2600 val perplexity 58.4910
327
+ 2600 train 4.009630 (lr=1.5863e-03) (hash(x)=37724702)
328
+ 2610 train 3.947030 (lr=1.5747e-03) (hash(x)=39564630)
329
+ 2620 train 3.896535 (lr=1.5632e-03) (hash(x)=38465803)
330
+ 2630 train 3.819648 (lr=1.5516e-03) (hash(x)=36955007)
331
+ 2640 train 3.851704 (lr=1.5400e-03) (hash(x)=32994568)
332
+ 2650 train 4.035601 (lr=1.5285e-03) (hash(x)=41165765)
333
+ 2660 train 4.046445 (lr=1.5170e-03) (hash(x)=42934000)
334
+ 2670 train 4.243977 (lr=1.5054e-03) (hash(x)=42726316)
335
+ 2680 train 4.130707 (lr=1.4939e-03) (hash(x)=35880318)
336
+ 2690 train 4.014536 (lr=1.4824e-03) (hash(x)=41747329)
337
+ 2700 val loss 4.0545
338
+ 2700 val perplexity 57.6592
339
+ 2700 train 4.135094 (lr=1.4709e-03) (hash(x)=40259630)
340
+ 2710 train 4.263653 (lr=1.4594e-03) (hash(x)=39895798)
341
+ 2720 train 3.993437 (lr=1.4480e-03) (hash(x)=36146683)
342
+ 2730 train 3.971633 (lr=1.4365e-03) (hash(x)=36181984)
343
+ 2740 train 3.949414 (lr=1.4251e-03) (hash(x)=43700349)
344
+ 2750 hella 0.2524
345
+ 2750 train 4.017869 (lr=1.4137e-03) (hash(x)=41610597)
346
+ 2760 train 4.000920 (lr=1.4023e-03) (hash(x)=31183639)
347
+ 2770 train 3.999709 (lr=1.3909e-03) (hash(x)=37722489)
348
+ 2780 train 3.943698 (lr=1.3795e-03) (hash(x)=47290688)
349
+ 2790 train 3.980732 (lr=1.3682e-03) (hash(x)=41205574)
350
+ 2800 val loss 4.0410
351
+ 2800 val perplexity 56.8825
352
+ 2800 train 3.830513 (lr=1.3569e-03) (hash(x)=40257962)
353
+ 2810 train 3.807233 (lr=1.3456e-03) (hash(x)=39529014)
354
+ 2820 train 3.826947 (lr=1.3343e-03) (hash(x)=42244749)
355
+ 2830 train 3.883721 (lr=1.3230e-03) (hash(x)=32820090)
356
+ 2840 train 3.836881 (lr=1.3118e-03) (hash(x)=40315769)
357
+ 2850 train 4.079078 (lr=1.3006e-03) (hash(x)=41524462)
358
+ 2860 train 4.071003 (lr=1.2894e-03) (hash(x)=38365734)
359
+ 2870 train 4.205096 (lr=1.2783e-03) (hash(x)=37682602)
360
+ 2880 train 4.054163 (lr=1.2671e-03) (hash(x)=39162991)
361
+ 2890 train 3.913689 (lr=1.2560e-03) (hash(x)=33316384)
362
+ 2900 val loss 4.0160
363
+ 2900 val perplexity 55.4785
364
+ 2900 train 3.965400 (lr=1.2450e-03) (hash(x)=37271132)
365
+ 2910 train 4.149245 (lr=1.2339e-03) (hash(x)=35586242)
366
+ 2920 train 4.043072 (lr=1.2229e-03) (hash(x)=33320586)
367
+ 2930 train 4.055053 (lr=1.2119e-03) (hash(x)=43531361)
368
+ 2940 train 3.944953 (lr=1.2010e-03) (hash(x)=37368286)
369
+ 2950 train 3.980961 (lr=1.1901e-03) (hash(x)=40363394)
370
+ 2960 train 3.945632 (lr=1.1792e-03) (hash(x)=42253792)
371
+ 2970 train 3.886268 (lr=1.1684e-03) (hash(x)=38072598)
372
+ 2980 train 3.960331 (lr=1.1576e-03) (hash(x)=41470557)
373
+ 2990 train 3.887382 (lr=1.1468e-03) (hash(x)=42600033)
374
+ 3000 val loss 4.0187
375
+ 3000 val perplexity 55.6281
376
+ 3000 hella 0.2521
377
+ 3000 train 3.841486 (lr=1.1361e-03) (hash(x)=46890983)
378
+ 3010 train 3.706032 (lr=1.1254e-03) (hash(x)=37986759)
379
+ 3020 train 3.832635 (lr=1.1147e-03) (hash(x)=36270703)
380
+ 3030 train 4.092685 (lr=1.1041e-03) (hash(x)=38228599)
381
+ 3040 train 3.948579 (lr=1.0935e-03) (hash(x)=44344296)
382
+ 3050 train 4.035525 (lr=1.0830e-03) (hash(x)=39962297)
383
+ 3060 train 4.007987 (lr=1.0725e-03) (hash(x)=39817394)
384
+ 3070 train 4.012559 (lr=1.0620e-03) (hash(x)=41763868)
385
+ 3080 train 4.094223 (lr=1.0516e-03) (hash(x)=42343051)
386
+ 3090 train 4.092932 (lr=1.0412e-03) (hash(x)=39204893)
387
+ 3100 val loss 3.9955
388
+ 3100 val perplexity 54.3521
389
+ 3100 train 3.998327 (lr=1.0309e-03) (hash(x)=39353599)
390
+ 3110 train 4.059931 (lr=1.0207e-03) (hash(x)=36118073)
391
+ 3120 train 3.767488 (lr=1.0104e-03) (hash(x)=44070271)
392
+ 3130 train 3.795837 (lr=1.0002e-03) (hash(x)=43368151)
393
+ 3140 train 3.861316 (lr=9.9011e-04) (hash(x)=40612203)
394
+ 3150 train 4.005939 (lr=9.8002e-04) (hash(x)=39598447)
395
+ 3160 train 4.053424 (lr=9.6999e-04) (hash(x)=43897682)
396
+ 3170 train 4.039975 (lr=9.6000e-04) (hash(x)=39029243)
397
+ 3180 train 4.047440 (lr=9.5007e-04) (hash(x)=38586684)
398
+ 3190 train 3.998623 (lr=9.4019e-04) (hash(x)=39299115)
399
+ 3200 val loss 3.9749
400
+ 3200 val perplexity 53.2444
401
+ 3200 train 4.029900 (lr=9.3036e-04) (hash(x)=41064949)
402
+ 3210 train 4.013452 (lr=9.2058e-04) (hash(x)=41752822)
403
+ 3220 train 4.023909 (lr=9.1085e-04) (hash(x)=37484805)
404
+ 3230 train 3.982543 (lr=9.0118e-04) (hash(x)=40514919)
405
+ 3240 train 4.036355 (lr=8.9157e-04) (hash(x)=42440343)
406
+ 3250 hella 0.2585
407
+ 3250 train 3.940191 (lr=8.8201e-04) (hash(x)=32464136)
408
+ 3260 train 3.957407 (lr=8.7251e-04) (hash(x)=47270558)
409
+ 3270 train 3.898355 (lr=8.6307e-04) (hash(x)=39307546)
410
+ 3280 train 3.933715 (lr=8.5368e-04) (hash(x)=40410299)
411
+ 3290 train 3.924451 (lr=8.4435e-04) (hash(x)=40541497)
412
+ 3300 val loss 3.9716
413
+ 3300 val perplexity 53.0714
414
+ 3300 train 3.828045 (lr=8.3508e-04) (hash(x)=38089139)
415
+ 3310 train 3.863294 (lr=8.2588e-04) (hash(x)=42006576)
416
+ 3320 train 3.953642 (lr=8.1673e-04) (hash(x)=40422556)
417
+ 3330 train 3.749929 (lr=8.0764e-04) (hash(x)=39021643)
418
+ 3340 train 4.006900 (lr=7.9862e-04) (hash(x)=40434305)
419
+ 3350 train 4.034446 (lr=7.8966e-04) (hash(x)=40833559)
420
+ 3360 train 4.031823 (lr=7.8076e-04) (hash(x)=40045231)
421
+ 3370 train 4.000900 (lr=7.7192e-04) (hash(x)=34668317)
422
+ 3380 train 3.995199 (lr=7.6315e-04) (hash(x)=42738568)
423
+ 3390 train 4.091264 (lr=7.5445e-04) (hash(x)=38980166)
424
+ 3400 val loss 3.9488
425
+ 3400 val perplexity 51.8754
426
+ 3400 train 4.001696 (lr=7.4581e-04) (hash(x)=43878176)
427
+ 3410 train 4.039756 (lr=7.3724e-04) (hash(x)=38201991)
428
+ 3420 train 4.020028 (lr=7.2874e-04) (hash(x)=34905889)
429
+ 3430 train 4.020187 (lr=7.2030e-04) (hash(x)=41680405)
430
+ 3440 train 3.981954 (lr=7.1193e-04) (hash(x)=39948184)
431
+ 3450 train 4.029565 (lr=7.0363e-04) (hash(x)=33623728)
432
+ 3460 train 3.900008 (lr=6.9541e-04) (hash(x)=41027647)
433
+ 3470 train 3.936356 (lr=6.8725e-04) (hash(x)=42779482)
434
+ 3480 train 3.953648 (lr=6.7916e-04) (hash(x)=41579288)
435
+ 3490 train 3.804503 (lr=6.7114e-04) (hash(x)=37174846)
436
+ 3500 val loss 3.9460
437
+ 3500 val perplexity 51.7255
438
+ 3500 hella 0.2570
439
+ 3500 train 3.687473 (lr=6.6320e-04) (hash(x)=37082090)
440
+ 3510 train 3.882000 (lr=6.5533e-04) (hash(x)=43131340)
441
+ 3520 train 3.795948 (lr=6.4753e-04) (hash(x)=42457250)
442
+ 3530 train 4.027563 (lr=6.3981e-04) (hash(x)=39504492)
443
+ 3540 train 3.972964 (lr=6.3216e-04) (hash(x)=36225997)
444
+ 3550 train 4.100944 (lr=6.2458e-04) (hash(x)=40912087)
445
+ 3560 train 3.940084 (lr=6.1708e-04) (hash(x)=41545925)
446
+ 3570 train 4.029422 (lr=6.0966e-04) (hash(x)=32553193)
447
+ 3580 train 3.872467 (lr=6.0231e-04) (hash(x)=38799856)
448
+ 3590 train 3.919961 (lr=5.9504e-04) (hash(x)=40772075)
449
+ 3600 val loss 3.9272
450
+ 3600 val perplexity 50.7621
451
+ 3600 train 3.958806 (lr=5.8785e-04) (hash(x)=39299903)
452
+ 3610 train 4.008116 (lr=5.8074e-04) (hash(x)=39503247)
453
+ 3620 train 3.897159 (lr=5.7370e-04) (hash(x)=38730668)
454
+ 3630 train 3.919765 (lr=5.6675e-04) (hash(x)=40503276)
455
+ 3640 train 3.933985 (lr=5.5987e-04) (hash(x)=36504573)
456
+ 3650 train 3.951691 (lr=5.5308e-04) (hash(x)=44903075)
457
+ 3660 train 3.862902 (lr=5.4636e-04) (hash(x)=33800148)
458
+ 3670 train 3.768956 (lr=5.3973e-04) (hash(x)=39404541)
459
+ 3680 train 3.713274 (lr=5.3318e-04) (hash(x)=44777432)
460
+ 3690 train 3.809445 (lr=5.2671e-04) (hash(x)=45649913)
461
+ 3700 val loss 3.9306
462
+ 3700 val perplexity 50.9397
463
+ 3700 train 3.935084 (lr=5.2033e-04) (hash(x)=33812537)
464
+ 3710 train 4.084886 (lr=5.1402e-04) (hash(x)=36161222)
465
+ 3720 train 3.960392 (lr=5.0780e-04) (hash(x)=43690616)
466
+ 3730 train 3.968457 (lr=5.0167e-04) (hash(x)=35462313)
467
+ 3740 train 4.260862 (lr=4.9562e-04) (hash(x)=44674254)
468
+ 3750 hella 0.2538
469
+ 3750 train 3.943103 (lr=4.8965e-04) (hash(x)=44276297)
470
+ 3760 train 3.987052 (lr=4.8377e-04) (hash(x)=38492178)
471
+ 3770 train 3.999232 (lr=4.7798e-04) (hash(x)=44347531)
472
+ 3780 train 3.912184 (lr=4.7227e-04) (hash(x)=33792059)
473
+ 3790 train 4.007885 (lr=4.6665e-04) (hash(x)=38549318)
474
+ 3800 val loss 3.9113
475
+ 3800 val perplexity 49.9621
476
+ 3800 train 3.885521 (lr=4.6112e-04) (hash(x)=41437448)
477
+ 3810 train 3.785176 (lr=4.5567e-04) (hash(x)=36656768)
478
+ 3820 train 3.943973 (lr=4.5031e-04) (hash(x)=46387893)
479
+ 3830 train 3.948587 (lr=4.4504e-04) (hash(x)=44312129)
480
+ 3840 train 3.869365 (lr=4.3986e-04) (hash(x)=40341896)
481
+ 3850 train 3.916968 (lr=4.3477e-04) (hash(x)=43589160)
482
+ 3860 train 3.903455 (lr=4.2977e-04) (hash(x)=40388945)
483
+ 3870 train 3.798559 (lr=4.2486e-04) (hash(x)=33336679)
484
+ 3880 train 3.891162 (lr=4.2004e-04) (hash(x)=37643311)
485
+ 3890 train 3.915297 (lr=4.1530e-04) (hash(x)=38272927)
486
+ 3900 val loss 3.9041
487
+ 3900 val perplexity 49.6076
488
+ 3900 train 3.860868 (lr=4.1066e-04) (hash(x)=41925748)
489
+ 3910 train 3.954099 (lr=4.0611e-04) (hash(x)=38377253)
490
+ 3920 train 3.957915 (lr=4.0166e-04) (hash(x)=41616611)
491
+ 3930 train 3.878563 (lr=3.9729e-04) (hash(x)=32301827)
492
+ 3940 train 3.933510 (lr=3.9302e-04) (hash(x)=47697363)
493
+ 3950 train 3.887604 (lr=3.8884e-04) (hash(x)=37867767)
494
+ 3960 train 3.886579 (lr=3.8475e-04) (hash(x)=37187295)
495
+ 3970 train 3.943573 (lr=3.8076e-04) (hash(x)=41952752)
496
+ 3980 train 3.911071 (lr=3.7685e-04) (hash(x)=38358660)
497
+ 3990 train 3.779912 (lr=3.7305e-04) (hash(x)=40207878)
498
+ 4000 val loss 3.8978
499
+ 4000 val perplexity 49.2924
500
+ 4000 hella 0.2587
501
+ 4000 train 3.870136 (lr=3.6933e-04) (hash(x)=39134015)
502
+ 4010 train 3.933059 (lr=3.6572e-04) (hash(x)=38313135)
503
+ 4020 train 3.871642 (lr=3.6219e-04) (hash(x)=40710513)
504
+ 4030 train 3.996726 (lr=3.5876e-04) (hash(x)=35530165)
505
+ 4040 train 3.922153 (lr=3.5543e-04) (hash(x)=39494066)
506
+ 4050 train 3.909729 (lr=3.5219e-04) (hash(x)=40861777)
507
+ 4060 train 3.892609 (lr=3.4905e-04) (hash(x)=36151939)
508
+ 4070 train 3.859110 (lr=3.4600e-04) (hash(x)=39008241)
509
+ 4080 train 3.942741 (lr=3.4305e-04) (hash(x)=40395746)
510
+ 4090 train 3.980504 (lr=3.4019e-04) (hash(x)=38855480)
511
+ 4100 val loss 3.8857
512
+ 4100 val perplexity 48.7008
513
+ 4100 train 3.918469 (lr=3.3744e-04) (hash(x)=45013254)
514
+ 4110 train 3.920167 (lr=3.3477e-04) (hash(x)=39934818)
515
+ 4120 train 3.923884 (lr=3.3221e-04) (hash(x)=33417705)
516
+ 4130 train 3.864084 (lr=3.2974e-04) (hash(x)=41267978)
517
+ 4140 train 3.853403 (lr=3.2737e-04) (hash(x)=38961609)
518
+ 4150 train 3.855011 (lr=3.2510e-04) (hash(x)=42173878)
519
+ 4160 train 3.821470 (lr=3.2292e-04) (hash(x)=36867405)
520
+ 4170 train 3.801349 (lr=3.2085e-04) (hash(x)=38991954)
521
+ 4180 train 3.862808 (lr=3.1887e-04) (hash(x)=39032592)
522
+ 4190 train 3.840634 (lr=3.1699e-04) (hash(x)=61329402)
523
+ 4200 val loss 3.8849
524
+ 4200 val perplexity 48.6641
525
+ 4200 train 3.852404 (lr=3.1520e-04) (hash(x)=40455200)
526
+ 4210 train 3.880306 (lr=3.1352e-04) (hash(x)=40913417)
527
+ 4220 train 3.911661 (lr=3.1193e-04) (hash(x)=38809704)
528
+ 4230 train 3.899312 (lr=3.1044e-04) (hash(x)=39560204)
529
+ 4240 train 3.899977 (lr=3.0905e-04) (hash(x)=41496549)
530
+ 4250 hella 0.2578
531
+ 4250 train 3.936690 (lr=3.0776e-04) (hash(x)=33994768)
532
+ 4260 train 3.924854 (lr=3.0657e-04) (hash(x)=39658687)
533
+ 4270 train 3.799242 (lr=3.0548e-04) (hash(x)=42583064)
534
+ 4280 train 3.867353 (lr=3.0449e-04) (hash(x)=40635811)
535
+ 4290 train 3.769593 (lr=3.0359e-04) (hash(x)=35313841)
536
+ 4300 val loss 3.8829
537
+ 4300 val perplexity 48.5669
538
+ 4300 train 3.948233 (lr=3.0280e-04) (hash(x)=39332245)
539
+ 4310 train 3.849671 (lr=3.0210e-04) (hash(x)=43732999)
540
+ 4320 train 3.852417 (lr=3.0150e-04) (hash(x)=40831835)
541
+ 4330 train 3.984606 (lr=3.0101e-04) (hash(x)=41813368)
542
+ 4340 train 3.888532 (lr=3.0061e-04) (hash(x)=36952586)
543
+ 4350 train 3.911988 (lr=3.0031e-04) (hash(x)=42470745)
544
+ 4360 train 3.839752 (lr=3.0011e-04) (hash(x)=39581495)
545
+ 4370 train 4.030051 (lr=3.0001e-04) (hash(x)=35914538)
546
+ 4374 val loss 3.8761
547
+ 4374 val perplexity 48.2358
548
+ 4374 hella 0.2562
latent_masks_learnable/learnable_no_compile_seed_1339/model_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b941788ea4b861278f09d3a45abef411a5db2e4189275866ffdee4e6859f4217
3
+ size 97706738
latent_masks_learnable/learnable_no_compile_seed_1339/optimizer_04374.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:650a63ed5e7279736fe4dba3a58690b1f80544c7e980597902c6451b5f59f048
3
+ size 189135414