Upload folder using huggingface_hub

12_mini_head_two_masks_4_heads/args.json

@@ -1 +1 @@
-{"hellaswag": true, "attention_kind": "selective", "log_dir": "logs/repro_selective_pattern_rankings/12_mini_head_two_masks_4_heads", "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": "repro_selective_pattern_rankings", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1341, "memory_penalty_epsilon": 0.1, "selection_head_linear_combo": "two_masks", "selection_head_linear_combo_scale": 1.0, "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": 64, "total_batch_size": 131072, "protection_head_scaling_factor": 1.0, "protection_head_bias": 0.0, "n_sliced_masks": null, "n_latent_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": "12_mini_head_two_masks_4_heads"}
+{"hellaswag": true, "attention_kind": "selective", "log_dir": "logs/repro_selective_pattern_rankings/12_mini_head_two_masks_4_heads", "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": "repro_selective_pattern_rankings", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1340, "memory_penalty_epsilon": 0.1, "selection_head_linear_combo": "two_masks", "selection_head_linear_combo_scale": 1.0, "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": 64, "total_batch_size": 131072, "protection_head_scaling_factor": 1.0, "protection_head_bias": 0.0, "n_sliced_masks": null, "n_latent_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": "12_mini_head_two_masks_4_heads"}

12_mini_head_two_masks_4_heads/log2.txt

@@ -1,529 +1,529 @@
 max_steps: 4375
-0 val loss 11.2828
-0 val perplexity 79446.0234
-0 train 11.286331 (lr=4.1958e-06) (hash(x)=83426457)
-10 train 9.906208 (lr=4.6154e-05) (hash(x)=78934557)
-20 train 9.362267 (lr=8.8112e-05) (hash(x)=77027105)
-30 train 8.592892 (lr=1.3007e-04) (hash(x)=71598963)
-40 train 8.011431 (lr=1.7203e-04) (hash(x)=82386827)
-50 train 7.856654 (lr=2.1399e-04) (hash(x)=85920934)
-60 train 7.661503 (lr=2.5594e-04) (hash(x)=81616761)
-70 train 7.552011 (lr=2.9790e-04) (hash(x)=86773189)
-80 train 7.308474 (lr=3.3986e-04) (hash(x)=71973470)
-90 train 7.206971 (lr=3.8182e-04) (hash(x)=83854524)
-100 val loss 7.0813
-100 val perplexity 1189.5540
-100 train 6.997993 (lr=4.2378e-04) (hash(x)=77188949)
-110 train 6.925124 (lr=4.6573e-04) (hash(x)=72138399)
-120 train 6.788277 (lr=5.0769e-04) (hash(x)=80173123)
-130 train 6.603106 (lr=5.4965e-04) (hash(x)=71981857)
-140 train 6.586471 (lr=5.9161e-04) (hash(x)=75014146)
-150 train 6.462921 (lr=6.3357e-04) (hash(x)=83329326)
-160 train 6.314249 (lr=6.7552e-04) (hash(x)=76522382)
-170 train 6.600364 (lr=7.1748e-04) (hash(x)=88508613)
-180 train 6.357422 (lr=7.5944e-04) (hash(x)=90018376)
-190 train 6.293137 (lr=8.0140e-04) (hash(x)=79854209)
-200 val loss 6.2148
-200 val perplexity 500.1002
-200 train 6.350214 (lr=8.4336e-04) (hash(x)=78560782)
-210 train 6.190270 (lr=8.8531e-04) (hash(x)=83733723)
-220 train 6.033686 (lr=9.2727e-04) (hash(x)=79271902)
-230 train 6.007542 (lr=9.6923e-04) (hash(x)=83461326)
-240 train 5.944594 (lr=1.0112e-03) (hash(x)=90489642)
-250 train 5.903481 (lr=1.0531e-03) (hash(x)=79614895)
-260 train 5.834649 (lr=1.0951e-03) (hash(x)=88646240)
-270 train 5.824803 (lr=1.1371e-03) (hash(x)=85620399)
-280 train 5.863040 (lr=1.1790e-03) (hash(x)=89054642)
-290 train 5.616792 (lr=1.2210e-03) (hash(x)=81160339)
-300 val loss 5.7095
-300 val perplexity 301.7280
-300 train 5.692072 (lr=1.2629e-03) (hash(x)=75259104)
-310 train 5.566352 (lr=1.3049e-03) (hash(x)=84448689)
-320 train 5.478696 (lr=1.3469e-03) (hash(x)=88463193)
-330 train 5.620771 (lr=1.3888e-03) (hash(x)=76313584)
-340 train 5.602615 (lr=1.4308e-03) (hash(x)=84598604)
-350 train 5.466035 (lr=1.4727e-03) (hash(x)=80551549)
-360 train 5.556177 (lr=1.5147e-03) (hash(x)=77926207)
-370 train 5.520696 (lr=1.5566e-03) (hash(x)=78142578)
-380 train 5.450786 (lr=1.5986e-03) (hash(x)=80853941)
-390 train 5.491229 (lr=1.6406e-03) (hash(x)=73606747)
-400 val loss 5.4037
-400 val perplexity 222.2381
-400 train 5.355968 (lr=1.6825e-03) (hash(x)=75283154)
-410 train 5.462108 (lr=1.7245e-03) (hash(x)=77829445)
-420 train 5.354402 (lr=1.7664e-03) (hash(x)=92661009)
-430 train 5.237918 (lr=1.8084e-03) (hash(x)=74159185)
-440 train 5.162126 (lr=1.8503e-03) (hash(x)=77656452)
-450 train 5.158522 (lr=1.8923e-03) (hash(x)=78633015)
-460 train 5.150229 (lr=1.9343e-03) (hash(x)=79484077)
-470 train 5.162044 (lr=1.9762e-03) (hash(x)=85308160)
-480 train 5.229625 (lr=2.0182e-03) (hash(x)=86733273)
-490 train 5.129216 (lr=2.0601e-03) (hash(x)=77806270)
-500 val loss 5.1758
-500 val perplexity 176.9294
-500 train 5.161263 (lr=2.1021e-03) (hash(x)=79154473)
-510 train 5.218800 (lr=2.1441e-03) (hash(x)=82252738)
-520 train 5.114560 (lr=2.1860e-03) (hash(x)=79224242)
-530 train 5.171950 (lr=2.2280e-03) (hash(x)=77695815)
-540 train 5.053813 (lr=2.2699e-03) (hash(x)=72435526)
-550 train 5.009875 (lr=2.3119e-03) (hash(x)=78722012)
-560 train 5.050132 (lr=2.3538e-03) (hash(x)=86379154)
-570 train 4.915636 (lr=2.3958e-03) (hash(x)=75571834)
-580 train 4.914773 (lr=2.4378e-03) (hash(x)=75212674)
-590 train 4.924623 (lr=2.4797e-03) (hash(x)=83039622)
-600 val loss 4.9977
-600 val perplexity 148.0761
-600 train 4.880018 (lr=2.5217e-03) (hash(x)=75453059)
-610 train 4.804048 (lr=2.5636e-03) (hash(x)=78156260)
-620 train 4.911411 (lr=2.6056e-03) (hash(x)=73224372)
-630 train 4.882734 (lr=2.6476e-03) (hash(x)=86669689)
-640 train 4.852366 (lr=2.6895e-03) (hash(x)=76838089)
-650 train 4.891253 (lr=2.7315e-03) (hash(x)=77816509)
-660 train 4.895391 (lr=2.7734e-03) (hash(x)=78714357)
-670 train 4.791610 (lr=2.8154e-03) (hash(x)=67108441)
-680 train 4.748251 (lr=2.8573e-03) (hash(x)=74297523)
-690 train 4.863575 (lr=2.8993e-03) (hash(x)=81807227)
-700 val loss 4.8200
-700 val perplexity 123.9599
-700 train 4.686471 (lr=2.9413e-03) (hash(x)=80680202)
-710 train 4.779562 (lr=2.9832e-03) (hash(x)=81685619)
-720 train 4.747033 (lr=3.0000e-03) (hash(x)=88293912)
-730 train 4.679256 (lr=2.9999e-03) (hash(x)=81156131)
-740 train 4.633346 (lr=2.9997e-03) (hash(x)=73015886)
-750 train 4.738198 (lr=2.9994e-03) (hash(x)=78392112)
-760 train 4.695868 (lr=2.9990e-03) (hash(x)=82475322)
-770 train 4.666807 (lr=2.9985e-03) (hash(x)=73069343)
-780 train 4.866407 (lr=2.9979e-03) (hash(x)=93027554)
-790 train 4.687624 (lr=2.9972e-03) (hash(x)=73364444)
-800 val loss 4.7007
-800 val perplexity 110.0204
-800 train 4.698443 (lr=2.9964e-03) (hash(x)=78801867)
-810 train 4.680539 (lr=2.9955e-03) (hash(x)=82551837)
-820 train 4.519366 (lr=2.9945e-03) (hash(x)=78287067)
-830 train 4.562117 (lr=2.9934e-03) (hash(x)=82837338)
-840 train 4.492977 (lr=2.9922e-03) (hash(x)=72494345)
-850 train 4.556274 (lr=2.9909e-03) (hash(x)=80840474)
-860 train 4.452876 (lr=2.9896e-03) (hash(x)=77298722)
-870 train 4.477791 (lr=2.9881e-03) (hash(x)=80296151)
-880 train 4.577305 (lr=2.9865e-03) (hash(x)=72779638)
-890 train 4.571060 (lr=2.9848e-03) (hash(x)=78473759)
-900 val loss 4.6562
-900 val perplexity 105.2381
-900 train 4.551015 (lr=2.9830e-03) (hash(x)=76625599)
-910 train 4.633257 (lr=2.9811e-03) (hash(x)=82707605)
-920 train 4.562136 (lr=2.9792e-03) (hash(x)=73411530)
-930 train 4.824069 (lr=2.9771e-03) (hash(x)=84116802)
-940 train 4.593833 (lr=2.9749e-03) (hash(x)=84662792)
-950 train 4.461058 (lr=2.9726e-03) (hash(x)=77520125)
-960 train 4.600173 (lr=2.9703e-03) (hash(x)=83412876)
-970 train 4.547418 (lr=2.9678e-03) (hash(x)=76679743)
-980 train 4.555427 (lr=2.9652e-03) (hash(x)=83578850)
-990 train 4.559372 (lr=2.9626e-03) (hash(x)=73508305)
-1000 val loss 4.5499
-1000 val perplexity 94.6207
-1000 train 4.558197 (lr=2.9598e-03) (hash(x)=73551731)
-1010 train 4.397978 (lr=2.9570e-03) (hash(x)=75472577)
-1020 train 4.450787 (lr=2.9540e-03) (hash(x)=74288773)
-1030 train 4.511033 (lr=2.9510e-03) (hash(x)=74358518)
-1040 train 4.441741 (lr=2.9478e-03) (hash(x)=81742445)
-1050 train 4.296555 (lr=2.9446e-03) (hash(x)=84334093)
-1060 train 4.293860 (lr=2.9412e-03) (hash(x)=83112407)
-1070 train 4.442715 (lr=2.9378e-03) (hash(x)=75361341)
-1080 train 4.323700 (lr=2.9343e-03) (hash(x)=80067829)
-1090 train 4.321742 (lr=2.9307e-03) (hash(x)=75084510)
-1100 val loss 4.4956
-1100 val perplexity 89.6248
-1100 train 4.484448 (lr=2.9270e-03) (hash(x)=81272082)
-1110 train 4.477275 (lr=2.9231e-03) (hash(x)=79636180)
-1120 train 4.435353 (lr=2.9192e-03) (hash(x)=106787061)
-1130 train 4.519568 (lr=2.9152e-03) (hash(x)=82157433)
-1140 train 4.472306 (lr=2.9112e-03) (hash(x)=74242006)
-1150 train 4.511582 (lr=2.9070e-03) (hash(x)=77502116)
-1160 train 4.459952 (lr=2.9027e-03) (hash(x)=78658526)
-1170 train 4.474169 (lr=2.8983e-03) (hash(x)=88605653)
-1180 train 4.404093 (lr=2.8939e-03) (hash(x)=75715872)
-1190 train 4.365394 (lr=2.8893e-03) (hash(x)=85140216)
-1200 val loss 4.4185
-1200 val perplexity 82.9703
-1200 train 4.413062 (lr=2.8847e-03) (hash(x)=76843977)
-1210 train 4.409280 (lr=2.8800e-03) (hash(x)=74668022)
-1220 train 4.354440 (lr=2.8751e-03) (hash(x)=73771388)
-1230 train 4.294598 (lr=2.8702e-03) (hash(x)=79549890)
-1240 train 4.387101 (lr=2.8652e-03) (hash(x)=80768426)
-1250 train 4.264572 (lr=2.8601e-03) (hash(x)=78485732)
-1260 train 4.200811 (lr=2.8550e-03) (hash(x)=75297547)
-1270 train 4.229288 (lr=2.8497e-03) (hash(x)=80488390)
-1280 train 4.536900 (lr=2.8443e-03) (hash(x)=83986686)
-1290 train 4.488621 (lr=2.8389e-03) (hash(x)=80260787)
-1300 val loss 4.3912
-1300 val perplexity 80.7406
-1300 train 4.522745 (lr=2.8333e-03) (hash(x)=91333241)
-1310 train 4.354868 (lr=2.8277e-03) (hash(x)=76196195)
-1320 train 4.334889 (lr=2.8220e-03) (hash(x)=89902604)
-1330 train 4.403556 (lr=2.8162e-03) (hash(x)=80055610)
-1340 train 4.390299 (lr=2.8103e-03) (hash(x)=75933940)
-1350 train 4.358490 (lr=2.8044e-03) (hash(x)=82880341)
-1360 train 4.341515 (lr=2.7983e-03) (hash(x)=74279731)
-1370 train 4.328010 (lr=2.7922e-03) (hash(x)=80381849)
-1380 train 4.412302 (lr=2.7860e-03) (hash(x)=80656732)
-1390 train 4.207211 (lr=2.7797e-03) (hash(x)=75892076)
-1400 val loss 4.3670
-1400 val perplexity 78.8098
-1400 train 4.270935 (lr=2.7733e-03) (hash(x)=85316269)
-1410 train 4.237441 (lr=2.7668e-03) (hash(x)=76176954)
-1420 train 4.083758 (lr=2.7603e-03) (hash(x)=82873585)
-1430 train 4.157396 (lr=2.7536e-03) (hash(x)=76423569)
-1440 train 4.217734 (lr=2.7469e-03) (hash(x)=77452162)
-1450 train 4.280375 (lr=2.7401e-03) (hash(x)=81479795)
-1460 train 4.067278 (lr=2.7333e-03) (hash(x)=83885401)
-1470 train 4.267023 (lr=2.7263e-03) (hash(x)=88229400)
-1480 train 4.394376 (lr=2.7193e-03) (hash(x)=80388539)
-1490 train 4.349235 (lr=2.7121e-03) (hash(x)=76421482)
-1500 val loss 4.3241
-1500 val perplexity 75.4987
-1500 train 4.289557 (lr=2.7050e-03) (hash(x)=82020652)
-1510 train 4.294731 (lr=2.6977e-03) (hash(x)=80450641)
-1520 train 4.229042 (lr=2.6903e-03) (hash(x)=80425758)
-1530 train 4.291595 (lr=2.6829e-03) (hash(x)=77928744)
-1540 train 4.258931 (lr=2.6754e-03) (hash(x)=81682588)
-1550 train 4.407263 (lr=2.6678e-03) (hash(x)=76218138)
-1560 train 4.241416 (lr=2.6602e-03) (hash(x)=82961366)
-1570 train 4.111447 (lr=2.6525e-03) (hash(x)=80344260)
-1580 train 4.195801 (lr=2.6447e-03) (hash(x)=81840773)
-1590 train 4.094424 (lr=2.6368e-03) (hash(x)=79596457)
-1600 val loss 4.3060
-1600 val perplexity 74.1396
-1600 train 4.146108 (lr=2.6289e-03) (hash(x)=78425415)
-1610 train 4.279434 (lr=2.6208e-03) (hash(x)=80735884)
-1620 train 4.234827 (lr=2.6128e-03) (hash(x)=82867340)
-1630 train 4.279835 (lr=2.6046e-03) (hash(x)=81152826)
-1640 train 4.312287 (lr=2.5964e-03) (hash(x)=79292243)
-1650 train 4.207090 (lr=2.5881e-03) (hash(x)=85633419)
-1660 train 4.206246 (lr=2.5797e-03) (hash(x)=56879214)
-1670 train 4.189358 (lr=2.5713e-03) (hash(x)=78580566)
-1680 train 4.128489 (lr=2.5628e-03) (hash(x)=73904044)
-1690 train 4.156729 (lr=2.5542e-03) (hash(x)=80456616)
-1700 val loss 4.2627
-1700 val perplexity 71.0018
-1700 train 4.157961 (lr=2.5455e-03) (hash(x)=83453850)
-1710 train 4.153193 (lr=2.5368e-03) (hash(x)=81246482)
-1720 train 4.153788 (lr=2.5281e-03) (hash(x)=74302931)
-1730 train 4.176238 (lr=2.5192e-03) (hash(x)=76366637)
-1740 train 4.101587 (lr=2.5103e-03) (hash(x)=113414000)
-1750 train 4.123521 (lr=2.5014e-03) (hash(x)=80537001)
-1760 train 4.248558 (lr=2.4924e-03) (hash(x)=80610857)
-1770 train 4.314498 (lr=2.4833e-03) (hash(x)=89782327)
-1780 train 4.246590 (lr=2.4741e-03) (hash(x)=82999075)
-1790 train 4.293472 (lr=2.4649e-03) (hash(x)=82486297)
-1800 val loss 4.2564
-1800 val perplexity 70.5522
-1800 train 4.253897 (lr=2.4556e-03) (hash(x)=79159575)
-1810 train 4.246717 (lr=2.4463e-03) (hash(x)=82428296)
-1820 train 4.229787 (lr=2.4369e-03) (hash(x)=72754716)
-1830 train 4.129428 (lr=2.4275e-03) (hash(x)=66172446)
-1840 train 4.207934 (lr=2.4180e-03) (hash(x)=79104767)
-1850 train 4.210263 (lr=2.4084e-03) (hash(x)=71073869)
-1860 train 4.105842 (lr=2.3988e-03) (hash(x)=76923769)
-1870 train 4.203895 (lr=2.3891e-03) (hash(x)=77628946)
-1880 train 4.106905 (lr=2.3794e-03) (hash(x)=87651837)
-1890 train 4.209055 (lr=2.3696e-03) (hash(x)=79373565)
-1900 val loss 4.2355
-1900 val perplexity 69.0968
-1900 train 4.275647 (lr=2.3598e-03) (hash(x)=86194345)
-1910 train 4.200642 (lr=2.3499e-03) (hash(x)=80865794)
-1920 train 4.165070 (lr=2.3400e-03) (hash(x)=79368476)
-1930 train 4.172744 (lr=2.3300e-03) (hash(x)=79247868)
-1940 train 4.120127 (lr=2.3200e-03) (hash(x)=72845520)
-1950 train 4.140367 (lr=2.3099e-03) (hash(x)=83254637)
-1960 train 4.086664 (lr=2.2998e-03) (hash(x)=81975531)
-1970 train 4.173362 (lr=2.2896e-03) (hash(x)=72939168)
-1980 train 4.077135 (lr=2.2793e-03) (hash(x)=82280940)
-1990 train 4.050721 (lr=2.2691e-03) (hash(x)=78119475)
-2000 val loss 4.2136
-2000 val perplexity 67.5986
-2000 train 4.087288 (lr=2.2588e-03) (hash(x)=83046712)
-2010 train 4.146014 (lr=2.2484e-03) (hash(x)=73468195)
-2020 train 4.087158 (lr=2.2380e-03) (hash(x)=74809817)
-2030 train 4.076430 (lr=2.2275e-03) (hash(x)=76037020)
-2040 train 4.298106 (lr=2.2170e-03) (hash(x)=79058361)
-2050 train 4.187833 (lr=2.2065e-03) (hash(x)=85014027)
-2060 train 4.161737 (lr=2.1959e-03) (hash(x)=78616227)
-2070 train 4.140214 (lr=2.1853e-03) (hash(x)=75928855)
-2080 train 4.237118 (lr=2.1746e-03) (hash(x)=87553274)
-2090 train 4.142758 (lr=2.1639e-03) (hash(x)=79805596)
-2100 val loss 4.1945
-2100 val perplexity 66.3176
-2100 train 4.011424 (lr=2.1532e-03) (hash(x)=71523881)
-2110 train 4.156861 (lr=2.1424e-03) (hash(x)=78431736)
-2120 train 4.056497 (lr=2.1316e-03) (hash(x)=76846294)
-2130 train 4.080083 (lr=2.1208e-03) (hash(x)=72808367)
-2140 train 3.896003 (lr=2.1099e-03) (hash(x)=71871519)
-2150 train 4.026684 (lr=2.0990e-03) (hash(x)=79598530)
-2160 train 4.086583 (lr=2.0881e-03) (hash(x)=81709468)
-2170 train 4.058995 (lr=2.0771e-03) (hash(x)=74075345)
-2180 train 4.214379 (lr=2.0661e-03) (hash(x)=79292886)
-2190 train 4.144413 (lr=2.0550e-03) (hash(x)=84316039)
-2200 val loss 4.1830
-2200 val perplexity 65.5629
-2200 train 4.222255 (lr=2.0440e-03) (hash(x)=84083188)
-2210 train 4.140667 (lr=2.0329e-03) (hash(x)=87482474)
-2220 train 4.254497 (lr=2.0217e-03) (hash(x)=73018507)
-2230 train 4.124879 (lr=2.0106e-03) (hash(x)=80194063)
-2240 train 4.029892 (lr=1.9994e-03) (hash(x)=70761174)
-2250 train 4.088642 (lr=1.9882e-03) (hash(x)=78296069)
-2260 train 4.058685 (lr=1.9770e-03) (hash(x)=71486863)
-2270 train 4.065935 (lr=1.9657e-03) (hash(x)=73106274)
-2280 train 4.023127 (lr=1.9544e-03) (hash(x)=65724504)
-2290 train 4.141122 (lr=1.9431e-03) (hash(x)=77464207)
-2300 val loss 4.1786
-2300 val perplexity 65.2746
-2300 train 4.287225 (lr=1.9318e-03) (hash(x)=79718531)
-2310 train 4.151774 (lr=1.9205e-03) (hash(x)=79578564)
-2320 train 4.075782 (lr=1.9091e-03) (hash(x)=77538251)
-2330 train 4.189883 (lr=1.8977e-03) (hash(x)=77684942)
-2340 train 4.153886 (lr=1.8863e-03) (hash(x)=74747888)
-2350 train 4.100947 (lr=1.8749e-03) (hash(x)=88812932)
-2360 train 4.080767 (lr=1.8635e-03) (hash(x)=77049145)
-2370 train 3.920271 (lr=1.8520e-03) (hash(x)=76636845)
-2380 train 3.990987 (lr=1.8406e-03) (hash(x)=79589860)
-2390 train 4.023909 (lr=1.8291e-03) (hash(x)=73694984)
-2400 val loss 4.1424
-2400 val perplexity 62.9564
-2400 train 4.074643 (lr=1.8176e-03) (hash(x)=76483417)
-2410 train 4.068033 (lr=1.8061e-03) (hash(x)=81534661)
-2420 train 4.163315 (lr=1.7946e-03) (hash(x)=90012063)
-2430 train 4.095062 (lr=1.7830e-03) (hash(x)=78359537)
-2440 train 4.190225 (lr=1.7715e-03) (hash(x)=79027074)
-2450 train 4.186569 (lr=1.7600e-03) (hash(x)=68888678)
-2460 train 4.059834 (lr=1.7484e-03) (hash(x)=78596234)
-2470 train 4.018798 (lr=1.7368e-03) (hash(x)=78240870)
-2480 train 4.095221 (lr=1.7253e-03) (hash(x)=78765106)
-2490 train 4.114528 (lr=1.7137e-03) (hash(x)=80888122)
-2500 val loss 4.1257
-2500 val perplexity 61.9136
-2500 train 3.859569 (lr=1.7021e-03) (hash(x)=77282980)
-2510 train 3.980441 (lr=1.6906e-03) (hash(x)=73932499)
-2520 train 3.885580 (lr=1.6790e-03) (hash(x)=78258721)
-2530 train 4.055018 (lr=1.6674e-03) (hash(x)=84237884)
-2540 train 3.966310 (lr=1.6558e-03) (hash(x)=82107056)
-2550 train 4.297820 (lr=1.6442e-03) (hash(x)=77739889)
-2560 train 4.292686 (lr=1.6326e-03) (hash(x)=74186552)
-2570 train 4.121394 (lr=1.6210e-03) (hash(x)=80473388)
-2580 train 4.042549 (lr=1.6094e-03) (hash(x)=82259419)
-2590 train 4.120201 (lr=1.5979e-03) (hash(x)=78413433)
-2600 val loss 4.1073
-2600 val perplexity 60.7802
-2600 train 4.028818 (lr=1.5863e-03) (hash(x)=74502607)
-2610 train 4.034249 (lr=1.5747e-03) (hash(x)=76291308)
-2620 train 3.995912 (lr=1.5632e-03) (hash(x)=89544859)
-2630 train 3.997247 (lr=1.5516e-03) (hash(x)=72230079)
-2640 train 4.018834 (lr=1.5400e-03) (hash(x)=80712391)
-2650 train 3.970821 (lr=1.5285e-03) (hash(x)=81110141)
-2660 train 4.048407 (lr=1.5170e-03) (hash(x)=73706547)
-2670 train 3.960949 (lr=1.5054e-03) (hash(x)=75978602)
-2680 train 4.038310 (lr=1.4939e-03) (hash(x)=77181602)
-2690 train 3.952512 (lr=1.4824e-03) (hash(x)=73503073)
-2700 val loss 4.0985
-2700 val perplexity 60.2506
-2700 train 4.002494 (lr=1.4709e-03) (hash(x)=76450620)
-2710 train 4.091777 (lr=1.4594e-03) (hash(x)=82384092)
-2720 train 4.085136 (lr=1.4480e-03) (hash(x)=78992460)
-2730 train 4.084784 (lr=1.4365e-03) (hash(x)=89640628)
-2740 train 4.055937 (lr=1.4251e-03) (hash(x)=78958333)
-2750 train 4.087621 (lr=1.4137e-03) (hash(x)=75231323)
-2760 train 4.041368 (lr=1.4023e-03) (hash(x)=80141789)
-2770 train 3.959982 (lr=1.3909e-03) (hash(x)=78041972)
-2780 train 4.080272 (lr=1.3795e-03) (hash(x)=78824636)
-2790 train 3.901150 (lr=1.3682e-03) (hash(x)=88495681)
-2800 val loss 4.0804
-2800 val perplexity 59.1716
-2800 train 3.956493 (lr=1.3569e-03) (hash(x)=70894085)
-2810 train 3.895991 (lr=1.3456e-03) (hash(x)=80942795)
-2820 train 3.906981 (lr=1.3343e-03) (hash(x)=80086512)
-2830 train 3.883872 (lr=1.3230e-03) (hash(x)=76162076)
-2840 train 4.031344 (lr=1.3118e-03) (hash(x)=92909437)
-2850 train 4.167970 (lr=1.3006e-03) (hash(x)=82192377)
-2860 train 3.991079 (lr=1.2894e-03) (hash(x)=79981204)
-2870 train 4.062534 (lr=1.2783e-03) (hash(x)=86642658)
-2880 train 4.005857 (lr=1.2671e-03) (hash(x)=73934349)
-2890 train 4.109758 (lr=1.2560e-03) (hash(x)=81104933)
-2900 val loss 4.0621
-2900 val perplexity 58.0985
-2900 train 3.988972 (lr=1.2450e-03) (hash(x)=83259335)
-2910 train 4.023648 (lr=1.2339e-03) (hash(x)=76528489)
-2920 train 4.017714 (lr=1.2229e-03) (hash(x)=82547590)
-2930 train 3.862637 (lr=1.2119e-03) (hash(x)=84692438)
-2940 train 3.886798 (lr=1.2010e-03) (hash(x)=76395062)
-2950 train 3.891234 (lr=1.1901e-03) (hash(x)=79351043)
-2960 train 3.963682 (lr=1.1792e-03) (hash(x)=75184745)
-2970 train 3.903659 (lr=1.1684e-03) (hash(x)=73888960)
-2980 train 4.102394 (lr=1.1576e-03) (hash(x)=80237171)
-2990 train 4.056189 (lr=1.1468e-03) (hash(x)=80090496)
-3000 val loss 4.0599
-3000 val perplexity 57.9686
-3000 train 4.100723 (lr=1.1361e-03) (hash(x)=99715255)
-3010 train 4.056059 (lr=1.1254e-03) (hash(x)=73302575)
-3020 train 3.956450 (lr=1.1147e-03) (hash(x)=81162327)
-3030 train 3.999872 (lr=1.1041e-03) (hash(x)=78686601)
-3040 train 4.017196 (lr=1.0935e-03) (hash(x)=76186508)
-3050 train 4.144871 (lr=1.0830e-03) (hash(x)=94420278)
-3060 train 4.029797 (lr=1.0725e-03) (hash(x)=73930563)
-3070 train 4.062143 (lr=1.0620e-03) (hash(x)=78074700)
-3080 train 3.915427 (lr=1.0516e-03) (hash(x)=85521020)
-3090 train 3.967668 (lr=1.0412e-03) (hash(x)=77429090)
-3100 val loss 4.0346
-3100 val perplexity 56.5220
-3100 train 3.946910 (lr=1.0309e-03) (hash(x)=74319016)
-3110 train 3.950850 (lr=1.0207e-03) (hash(x)=81856930)
-3120 train 3.966238 (lr=1.0104e-03) (hash(x)=78854668)
-3130 train 3.803520 (lr=1.0002e-03) (hash(x)=78558831)
-3140 train 3.757164 (lr=9.9011e-04) (hash(x)=75419854)
-3150 train 3.744627 (lr=9.8002e-04) (hash(x)=72645078)
-3160 train 3.924897 (lr=9.6999e-04) (hash(x)=71954229)
-3170 train 3.853053 (lr=9.6000e-04) (hash(x)=83944062)
-3180 train 4.139953 (lr=9.5007e-04) (hash(x)=81719676)
-3190 train 4.123998 (lr=9.4019e-04) (hash(x)=78117157)
-3200 val loss 4.0186
-3200 val perplexity 55.6221
-3200 train 4.099495 (lr=9.3036e-04) (hash(x)=95062223)
-3210 train 3.990359 (lr=9.2058e-04) (hash(x)=78243969)
-3220 train 3.990211 (lr=9.1085e-04) (hash(x)=80840117)
-3230 train 4.070368 (lr=9.0118e-04) (hash(x)=89040111)
-3240 train 3.931151 (lr=8.9157e-04) (hash(x)=78018689)
-3250 train 4.017251 (lr=8.8201e-04) (hash(x)=71514679)
-3260 train 3.950819 (lr=8.7251e-04) (hash(x)=72142220)
-3270 train 3.898019 (lr=8.6307e-04) (hash(x)=81981079)
-3280 train 4.039700 (lr=8.5368e-04) (hash(x)=81221688)
-3290 train 3.924311 (lr=8.4435e-04) (hash(x)=84425616)
-3300 val loss 4.0029
-3300 val perplexity 54.7549
-3300 train 3.938393 (lr=8.3508e-04) (hash(x)=85200306)
-3310 train 3.920418 (lr=8.2588e-04) (hash(x)=90538542)
-3320 train 3.787085 (lr=8.1673e-04) (hash(x)=75045669)
-3330 train 3.688650 (lr=8.0764e-04) (hash(x)=91874178)
-3340 train 3.712383 (lr=7.9862e-04) (hash(x)=93115709)
-3350 train 3.739361 (lr=7.8966e-04) (hash(x)=80614131)
-3360 train 3.770380 (lr=7.8076e-04) (hash(x)=75966518)
-3370 train 3.928471 (lr=7.7192e-04) (hash(x)=52504673)
-3380 train 3.976027 (lr=7.6315e-04) (hash(x)=89845158)
-3390 train 4.535682 (lr=7.5445e-04) (hash(x)=84328133)
-3400 val loss 3.9968
-3400 val perplexity 54.4249
-3400 train 4.045430 (lr=7.4581e-04) (hash(x)=84410966)
-3410 train 3.988138 (lr=7.3724e-04) (hash(x)=81040312)
-3420 train 4.111779 (lr=7.2874e-04) (hash(x)=80573120)
-3430 train 3.892544 (lr=7.2030e-04) (hash(x)=77676972)
-3440 train 3.957338 (lr=7.1193e-04) (hash(x)=86131899)
-3450 train 3.975358 (lr=7.0363e-04) (hash(x)=75051359)
-3460 train 3.909394 (lr=6.9541e-04) (hash(x)=78319212)
-3470 train 3.961527 (lr=6.8725e-04) (hash(x)=75514502)
-3480 train 3.827725 (lr=6.7916e-04) (hash(x)=84577433)
-3490 train 3.995015 (lr=6.7114e-04) (hash(x)=76740859)
-3500 val loss 3.9811
-3500 val perplexity 53.5775
-3500 train 3.955020 (lr=6.6320e-04) (hash(x)=85221122)
-3510 train 3.882984 (lr=6.5533e-04) (hash(x)=78373676)
-3520 train 3.773505 (lr=6.4753e-04) (hash(x)=76713204)
-3530 train 3.673000 (lr=6.3981e-04) (hash(x)=86885732)
-3540 train 3.804622 (lr=6.3216e-04) (hash(x)=80751876)
-3550 train 3.771646 (lr=6.2458e-04) (hash(x)=80092170)
-3560 train 3.949727 (lr=6.1708e-04) (hash(x)=85501243)
-3570 train 4.046741 (lr=6.0966e-04) (hash(x)=87853218)
-3580 train 3.991044 (lr=6.0231e-04) (hash(x)=88313614)
-3590 train 3.921559 (lr=5.9504e-04) (hash(x)=75521063)
-3600 val loss 3.9756
-3600 val perplexity 53.2838
-3600 train 3.989742 (lr=5.8785e-04) (hash(x)=83721651)
-3610 train 3.878727 (lr=5.8074e-04) (hash(x)=94293884)
-3620 train 3.934310 (lr=5.7370e-04) (hash(x)=77296763)
-3630 train 4.064809 (lr=5.6675e-04) (hash(x)=77338405)
-3640 train 3.911325 (lr=5.5987e-04) (hash(x)=81223547)
-3650 train 3.951725 (lr=5.5308e-04) (hash(x)=78952410)
-3660 train 3.796055 (lr=5.4636e-04) (hash(x)=79846386)
-3670 train 3.956671 (lr=5.3973e-04) (hash(x)=80481480)
-3680 train 3.855377 (lr=5.3318e-04) (hash(x)=83821130)
-3690 train 3.785290 (lr=5.2671e-04) (hash(x)=85303368)
-3700 val loss 3.9603
-3700 val perplexity 52.4743
-3700 train 3.828219 (lr=5.2033e-04) (hash(x)=70393870)
-3710 train 3.809762 (lr=5.1402e-04) (hash(x)=81014823)
-3720 train 3.822173 (lr=5.0780e-04) (hash(x)=79155367)
-3730 train 3.800327 (lr=5.0167e-04) (hash(x)=86979929)
-3740 train 3.756806 (lr=4.9562e-04) (hash(x)=88193263)
-3750 train 4.065956 (lr=4.8965e-04) (hash(x)=89853653)
-3760 train 3.974889 (lr=4.8377e-04) (hash(x)=77692484)
-3770 train 4.048394 (lr=4.7798e-04) (hash(x)=83300185)
-3780 train 3.964279 (lr=4.7227e-04) (hash(x)=82443218)
-3790 train 3.814682 (lr=4.6665e-04) (hash(x)=90032898)
-3800 val loss 3.9469
-3800 val perplexity 51.7726
-3800 train 3.927063 (lr=4.6112e-04) (hash(x)=77605593)
-3810 train 3.986592 (lr=4.5567e-04) (hash(x)=82532796)
-3820 train 3.939422 (lr=4.5031e-04) (hash(x)=89648204)
-3830 train 4.022681 (lr=4.4504e-04) (hash(x)=83708293)
-3840 train 3.964045 (lr=4.3986e-04) (hash(x)=84336512)
-3850 train 3.925911 (lr=4.3477e-04) (hash(x)=75025285)
-3860 train 3.806261 (lr=4.2977e-04) (hash(x)=79436536)
-3870 train 3.852533 (lr=4.2486e-04) (hash(x)=90467390)
-3880 train 3.932883 (lr=4.2004e-04) (hash(x)=77292020)
-3890 train 3.822489 (lr=4.1530e-04) (hash(x)=75568927)
-3900 val loss 3.9461
-3900 val perplexity 51.7347
-3900 train 3.871303 (lr=4.1066e-04) (hash(x)=79671781)
-3910 train 3.934132 (lr=4.0611e-04) (hash(x)=84280943)
-3920 train 3.957806 (lr=4.0166e-04) (hash(x)=80901294)
-3930 train 4.035690 (lr=3.9729e-04) (hash(x)=81437584)
-3940 train 3.934170 (lr=3.9302e-04) (hash(x)=73893778)
-3950 train 3.877768 (lr=3.8884e-04) (hash(x)=74987794)
-3960 train 3.925899 (lr=3.8475e-04) (hash(x)=93721374)
-3970 train 3.936910 (lr=3.8076e-04) (hash(x)=79149678)
-3980 train 3.890261 (lr=3.7685e-04) (hash(x)=83179387)
-3990 train 3.914925 (lr=3.7305e-04) (hash(x)=89009978)
-4000 val loss 3.9333
-4000 val perplexity 51.0731
-4000 train 3.929021 (lr=3.6933e-04) (hash(x)=76483735)
-4010 train 3.816128 (lr=3.6572e-04) (hash(x)=78638850)
-4020 train 3.886682 (lr=3.6219e-04) (hash(x)=90415206)
-4030 train 3.836416 (lr=3.5876e-04) (hash(x)=90819195)
-4040 train 3.955806 (lr=3.5543e-04) (hash(x)=85938852)
-4050 train 3.981931 (lr=3.5219e-04) (hash(x)=82436789)
-4060 train 4.127342 (lr=3.4905e-04) (hash(x)=63022149)
-4070 train 3.959313 (lr=3.4600e-04) (hash(x)=83501199)
-4080 train 3.955446 (lr=3.4305e-04) (hash(x)=77093326)
-4090 train 3.829535 (lr=3.4019e-04) (hash(x)=73437559)
-4100 val loss 3.9248
-4100 val perplexity 50.6411
-4100 train 3.932780 (lr=3.3744e-04) (hash(x)=84550388)
-4110 train 3.951888 (lr=3.3477e-04) (hash(x)=86184566)
-4120 train 3.944233 (lr=3.3221e-04) (hash(x)=85644922)
-4130 train 3.887218 (lr=3.2974e-04) (hash(x)=86179801)
-4140 train 3.856702 (lr=3.2737e-04) (hash(x)=82868303)
-4150 train 3.954932 (lr=3.2510e-04) (hash(x)=96780388)
-4160 train 3.923052 (lr=3.2292e-04) (hash(x)=87223122)
-4170 train 3.902454 (lr=3.2085e-04) (hash(x)=77292786)
-4180 train 3.907773 (lr=3.1887e-04) (hash(x)=88761375)
-4190 train 3.806052 (lr=3.1699e-04) (hash(x)=80956468)
-4200 val loss 3.9218
-4200 val perplexity 50.4931
-4200 train 3.973460 (lr=3.1520e-04) (hash(x)=84725462)
-4210 train 3.959135 (lr=3.1352e-04) (hash(x)=76098113)
-4220 train 3.961047 (lr=3.1193e-04) (hash(x)=84387685)
-4230 train 3.914633 (lr=3.1044e-04) (hash(x)=76120321)
-4240 train 3.887979 (lr=3.0905e-04) (hash(x)=85974065)
-4250 train 3.814911 (lr=3.0776e-04) (hash(x)=79340644)
-4260 train 3.915934 (lr=3.0657e-04) (hash(x)=88345025)
-4270 train 3.822927 (lr=3.0548e-04) (hash(x)=86406230)
-4280 train 3.817116 (lr=3.0449e-04) (hash(x)=80608975)
-4290 train 3.868625 (lr=3.0359e-04) (hash(x)=81635225)
-4300 val loss 3.9188
-4300 val perplexity 50.3397
-4300 train 3.730527 (lr=3.0280e-04) (hash(x)=82126376)
-4310 train 3.888378 (lr=3.0210e-04) (hash(x)=83737972)
-4320 train 3.812414 (lr=3.0150e-04) (hash(x)=81132345)
-4330 train 3.877918 (lr=3.0101e-04) (hash(x)=85850404)
-4340 train 3.877436 (lr=3.0061e-04) (hash(x)=83490640)
-4350 train 3.895013 (lr=3.0031e-04) (hash(x)=77545187)
-4360 train 3.924293 (lr=3.0011e-04) (hash(x)=86412685)
-4370 train 3.825723 (lr=3.0001e-04) (hash(x)=86429640)
-4374 val loss 3.9108
-4374 val perplexity 49.9407
+0 val loss 11.2730
+0 val perplexity 78668.5547
+0 train 11.284272 (lr=4.1958e-06) (hash(x)=92476531)
+10 train 9.839178 (lr=4.6154e-05) (hash(x)=80557183)
+20 train 9.306097 (lr=8.8112e-05) (hash(x)=77833840)
+30 train 8.591415 (lr=1.3007e-04) (hash(x)=80219380)
+40 train 7.887783 (lr=1.7203e-04) (hash(x)=81242310)
+50 train 7.604287 (lr=2.1399e-04) (hash(x)=82639280)
+60 train 7.512836 (lr=2.5594e-04) (hash(x)=75412499)
+70 train 7.487408 (lr=2.9790e-04) (hash(x)=87943396)
+80 train 7.365146 (lr=3.3986e-04) (hash(x)=85407935)
+90 train 7.152790 (lr=3.8182e-04) (hash(x)=79447147)
+100 val loss 7.0772
+100 val perplexity 1184.5970
+100 train 7.033171 (lr=4.2378e-04) (hash(x)=80787570)
+110 train 6.933134 (lr=4.6573e-04) (hash(x)=79791376)
+120 train 6.827237 (lr=5.0769e-04) (hash(x)=79296900)
+130 train 6.667524 (lr=5.4965e-04) (hash(x)=79174525)
+140 train 6.580367 (lr=5.9161e-04) (hash(x)=83957686)
+150 train 6.508326 (lr=6.3357e-04) (hash(x)=77592594)
+160 train 6.376023 (lr=6.7552e-04) (hash(x)=70564809)
+170 train 6.431693 (lr=7.1748e-04) (hash(x)=72584980)
+180 train 6.443375 (lr=7.5944e-04) (hash(x)=95506562)
+190 train 6.170627 (lr=8.0140e-04) (hash(x)=73554383)
+200 val loss 6.1787
+200 val perplexity 482.3582
+200 train 6.111257 (lr=8.4336e-04) (hash(x)=84992944)
+210 train 6.027501 (lr=8.8531e-04) (hash(x)=84224321)
+220 train 5.959018 (lr=9.2727e-04) (hash(x)=91374190)
+230 train 5.971571 (lr=9.6923e-04) (hash(x)=85276683)
+240 train 6.027958 (lr=1.0112e-03) (hash(x)=81723155)
+250 train 5.952275 (lr=1.0531e-03) (hash(x)=75382743)
+260 train 5.903173 (lr=1.0951e-03) (hash(x)=79107828)
+270 train 5.768265 (lr=1.1371e-03) (hash(x)=63566375)
+280 train 5.778795 (lr=1.1790e-03) (hash(x)=72509291)
+290 train 5.701673 (lr=1.2210e-03) (hash(x)=80854173)
+300 val loss 5.7283
+300 val perplexity 307.4337
+300 train 5.711514 (lr=1.2629e-03) (hash(x)=95549669)
+310 train 5.555037 (lr=1.3049e-03) (hash(x)=82435860)
+320 train 5.622499 (lr=1.3469e-03) (hash(x)=79528901)
+330 train 5.538532 (lr=1.3888e-03) (hash(x)=89988570)
+340 train 5.552803 (lr=1.4308e-03) (hash(x)=81102378)
+350 train 5.474455 (lr=1.4727e-03) (hash(x)=75258538)
+360 train 5.432263 (lr=1.5147e-03) (hash(x)=75080686)
+370 train 5.446301 (lr=1.5566e-03) (hash(x)=87436781)
+380 train 5.425241 (lr=1.5986e-03) (hash(x)=78319003)
+390 train 5.399833 (lr=1.6406e-03) (hash(x)=80258231)
+400 val loss 5.4284
+400 val perplexity 227.7779
+400 train 5.461873 (lr=1.6825e-03) (hash(x)=93677302)
+410 train 5.406111 (lr=1.7245e-03) (hash(x)=83320730)
+420 train 5.375980 (lr=1.7664e-03) (hash(x)=88090468)
+430 train 5.364167 (lr=1.8084e-03) (hash(x)=77168020)
+440 train 5.345466 (lr=1.8503e-03) (hash(x)=75302477)
+450 train 5.289786 (lr=1.8923e-03) (hash(x)=87839104)
+460 train 5.367118 (lr=1.9343e-03) (hash(x)=84600562)
+470 train 5.223162 (lr=1.9762e-03) (hash(x)=72223964)
+480 train 5.151525 (lr=2.0182e-03) (hash(x)=85410838)
+490 train 5.226648 (lr=2.0601e-03) (hash(x)=75686864)
+500 val loss 5.2330
+500 val perplexity 187.3467
+500 train 5.118035 (lr=2.1021e-03) (hash(x)=82209970)
+510 train 5.136031 (lr=2.1441e-03) (hash(x)=78903986)
+520 train 5.189431 (lr=2.1860e-03) (hash(x)=82788367)
+530 train 5.147382 (lr=2.2280e-03) (hash(x)=79014660)
+540 train 5.172481 (lr=2.2699e-03) (hash(x)=78203384)
+550 train 5.036970 (lr=2.3119e-03) (hash(x)=74749939)
+560 train 5.071232 (lr=2.3538e-03) (hash(x)=73863798)
+570 train 5.139832 (lr=2.3958e-03) (hash(x)=81655345)
+580 train 5.060058 (lr=2.4378e-03) (hash(x)=77989048)
+590 train 5.058838 (lr=2.4797e-03) (hash(x)=75240510)
+600 val loss 5.0134
+600 val perplexity 150.4134
+600 train 5.013699 (lr=2.5217e-03) (hash(x)=75393302)
+610 train 5.375752 (lr=2.5636e-03) (hash(x)=75805578)
+620 train 4.882098 (lr=2.6056e-03) (hash(x)=76469956)
+630 train 4.950880 (lr=2.6476e-03) (hash(x)=80862833)
+640 train 4.901216 (lr=2.6895e-03) (hash(x)=69614120)
+650 train 4.888083 (lr=2.7315e-03) (hash(x)=81753733)
+660 train 4.945616 (lr=2.7734e-03) (hash(x)=80638528)
+670 train 4.906520 (lr=2.8154e-03) (hash(x)=82087325)
+680 train 5.004448 (lr=2.8573e-03) (hash(x)=83366526)
+690 train 4.814536 (lr=2.8993e-03) (hash(x)=79702542)
+700 val loss 4.8379
+700 val perplexity 126.1990
+700 train 4.745389 (lr=2.9413e-03) (hash(x)=67691907)
+710 train 4.965355 (lr=2.9832e-03) (hash(x)=78194053)
+720 train 4.763076 (lr=3.0000e-03) (hash(x)=76369890)
+730 train 4.750814 (lr=2.9999e-03) (hash(x)=76486285)
+740 train 4.718695 (lr=2.9997e-03) (hash(x)=73198975)
+750 train 4.731890 (lr=2.9994e-03) (hash(x)=72993983)
+760 train 4.641125 (lr=2.9990e-03) (hash(x)=88306535)
+770 train 4.664103 (lr=2.9985e-03) (hash(x)=87649303)
+780 train 4.510022 (lr=2.9979e-03) (hash(x)=80261986)
+790 train 4.507662 (lr=2.9972e-03) (hash(x)=84518434)
+800 val loss 4.7414
+800 val perplexity 114.5958
+800 train 4.472368 (lr=2.9964e-03) (hash(x)=77880126)
+810 train 4.687553 (lr=2.9955e-03) (hash(x)=91800941)
+820 train 4.548345 (lr=2.9945e-03) (hash(x)=77438844)
+830 train 4.743817 (lr=2.9934e-03) (hash(x)=84295123)
+840 train 4.591843 (lr=2.9922e-03) (hash(x)=82587990)
+850 train 4.691697 (lr=2.9909e-03) (hash(x)=76145087)
+860 train 4.610682 (lr=2.9896e-03) (hash(x)=80380499)
+870 train 4.594354 (lr=2.9881e-03) (hash(x)=79852152)
+880 train 4.628748 (lr=2.9865e-03) (hash(x)=80506089)
+890 train 4.529801 (lr=2.9848e-03) (hash(x)=74969638)
+900 val loss 4.6113
+900 val perplexity 100.6123
+900 train 4.579975 (lr=2.9830e-03) (hash(x)=72563497)
+910 train 4.478542 (lr=2.9811e-03) (hash(x)=79240775)
+920 train 4.529010 (lr=2.9792e-03) (hash(x)=81010967)
+930 train 4.534706 (lr=2.9771e-03) (hash(x)=76027759)
+940 train 4.506199 (lr=2.9749e-03) (hash(x)=73716762)
+950 train 4.462321 (lr=2.9726e-03) (hash(x)=79940439)
+960 train 4.497485 (lr=2.9703e-03) (hash(x)=88318293)
+970 train 4.339664 (lr=2.9678e-03) (hash(x)=80583440)
+980 train 4.335401 (lr=2.9652e-03) (hash(x)=79458899)
+990 train 4.413242 (lr=2.9626e-03) (hash(x)=79258146)
+1000 val loss 4.5700
+1000 val perplexity 96.5402
+1000 train 4.775337 (lr=2.9598e-03) (hash(x)=78128159)
+1010 train 4.597528 (lr=2.9570e-03) (hash(x)=81340506)
+1020 train 4.629194 (lr=2.9540e-03) (hash(x)=94839111)
+1030 train 4.526344 (lr=2.9510e-03) (hash(x)=79238631)
+1040 train 4.513329 (lr=2.9478e-03) (hash(x)=87577788)
+1050 train 4.520325 (lr=2.9446e-03) (hash(x)=78297110)
+1060 train 4.456205 (lr=2.9412e-03) (hash(x)=72337518)
+1070 train 4.457132 (lr=2.9378e-03) (hash(x)=72814742)
+1080 train 4.490771 (lr=2.9343e-03) (hash(x)=75722002)
+1090 train 4.435574 (lr=2.9307e-03) (hash(x)=78305141)
+1100 val loss 4.4888
+1100 val perplexity 89.0108
+1100 train 4.392165 (lr=2.9270e-03) (hash(x)=79663017)
+1110 train 4.503244 (lr=2.9231e-03) (hash(x)=104050146)
+1120 train 4.462349 (lr=2.9192e-03) (hash(x)=67379529)
+1130 train 4.364481 (lr=2.9152e-03) (hash(x)=75589492)
+1140 train 4.327545 (lr=2.9112e-03) (hash(x)=77978407)
+1150 train 4.311742 (lr=2.9070e-03) (hash(x)=79982444)
+1160 train 4.175939 (lr=2.9027e-03) (hash(x)=77038480)
+1170 train 4.183376 (lr=2.8983e-03) (hash(x)=84041157)
+1180 train 4.179930 (lr=2.8939e-03) (hash(x)=85344564)
+1190 train 4.414925 (lr=2.8893e-03) (hash(x)=82018589)
+1200 val loss 4.4525
+1200 val perplexity 85.8401
+1200 train 4.460476 (lr=2.8847e-03) (hash(x)=83172541)
+1210 train 4.411138 (lr=2.8800e-03) (hash(x)=76704756)
+1220 train 4.297914 (lr=2.8751e-03) (hash(x)=80484892)
+1230 train 4.393329 (lr=2.8702e-03) (hash(x)=80724246)
+1240 train 4.501640 (lr=2.8652e-03) (hash(x)=82844155)
+1250 train 4.381180 (lr=2.8601e-03) (hash(x)=75005983)
+1260 train 4.367898 (lr=2.8550e-03) (hash(x)=76014668)
+1270 train 4.353064 (lr=2.8497e-03) (hash(x)=74062742)
+1280 train 4.448869 (lr=2.8443e-03) (hash(x)=83893278)
+1290 train 4.305362 (lr=2.8389e-03) (hash(x)=82235715)
+1300 val loss 4.3935
+1300 val perplexity 80.9250
+1300 train 4.361597 (lr=2.8333e-03) (hash(x)=79195156)
+1310 train 4.438707 (lr=2.8277e-03) (hash(x)=79036954)
+1320 train 4.251760 (lr=2.8220e-03) (hash(x)=85008455)
+1330 train 4.219926 (lr=2.8162e-03) (hash(x)=75114081)
+1340 train 4.118965 (lr=2.8103e-03) (hash(x)=82306787)
+1350 train 4.255581 (lr=2.8044e-03) (hash(x)=85112053)
+1360 train 4.121719 (lr=2.7983e-03) (hash(x)=82963845)
+1370 train 4.096245 (lr=2.7922e-03) (hash(x)=85030982)
+1380 train 4.345950 (lr=2.7860e-03) (hash(x)=93735902)
+1390 train 4.304271 (lr=2.7797e-03) (hash(x)=89011154)
+1400 val loss 4.3487
+1400 val perplexity 77.3751
+1400 train 4.291582 (lr=2.7733e-03) (hash(x)=82085703)
+1410 train 4.605539 (lr=2.7668e-03) (hash(x)=90334911)
+1420 train 4.370450 (lr=2.7603e-03) (hash(x)=83787797)
+1430 train 4.265886 (lr=2.7536e-03) (hash(x)=82084098)
+1440 train 4.281202 (lr=2.7469e-03) (hash(x)=77539484)
+1450 train 4.378546 (lr=2.7401e-03) (hash(x)=77887047)
+1460 train 4.326289 (lr=2.7333e-03) (hash(x)=81880305)
+1470 train 4.304350 (lr=2.7263e-03) (hash(x)=75176428)
+1480 train 4.305735 (lr=2.7193e-03) (hash(x)=83835098)
+1490 train 4.260078 (lr=2.7121e-03) (hash(x)=90482123)
+1500 val loss 4.3321
+1500 val perplexity 76.1027
+1500 train 4.219032 (lr=2.7050e-03) (hash(x)=78462195)
+1510 train 4.324367 (lr=2.6977e-03) (hash(x)=78628581)
+1520 train 4.104542 (lr=2.6903e-03) (hash(x)=75543035)
+1530 train 4.364205 (lr=2.6829e-03) (hash(x)=80858373)
+1540 train 4.317539 (lr=2.6754e-03) (hash(x)=75358272)
+1550 train 4.203523 (lr=2.6678e-03) (hash(x)=81074965)
+1560 train 4.233493 (lr=2.6602e-03) (hash(x)=75919017)
+1570 train 4.263735 (lr=2.6525e-03) (hash(x)=78449050)
+1580 train 4.229527 (lr=2.6447e-03) (hash(x)=80179401)
+1590 train 4.196559 (lr=2.6368e-03) (hash(x)=78201723)
+1600 val loss 4.3011
+1600 val perplexity 73.7832
+1600 train 4.201737 (lr=2.6289e-03) (hash(x)=82628634)
+1610 train 4.183100 (lr=2.6208e-03) (hash(x)=83088672)
+1620 train 4.100358 (lr=2.6128e-03) (hash(x)=86507787)
+1630 train 4.142253 (lr=2.6046e-03) (hash(x)=73078561)
+1640 train 4.095888 (lr=2.5964e-03) (hash(x)=82166841)
+1650 train 4.200635 (lr=2.5881e-03) (hash(x)=78193565)
+1660 train 4.294629 (lr=2.5797e-03) (hash(x)=85804701)
+1670 train 4.307804 (lr=2.5713e-03) (hash(x)=70345337)
+1680 train 4.260899 (lr=2.5628e-03) (hash(x)=80320580)
+1690 train 4.248462 (lr=2.5542e-03) (hash(x)=81513796)
+1700 val loss 4.2716
+1700 val perplexity 71.6336
+1700 train 4.167795 (lr=2.5455e-03) (hash(x)=82483619)
+1710 train 4.237206 (lr=2.5368e-03) (hash(x)=74499966)
+1720 train 4.148996 (lr=2.5281e-03) (hash(x)=82891474)
+1730 train 4.145032 (lr=2.5192e-03) (hash(x)=76311864)
+1740 train 4.294483 (lr=2.5103e-03) (hash(x)=78327998)
+1750 train 4.172101 (lr=2.5014e-03) (hash(x)=81274966)
+1760 train 4.166620 (lr=2.4924e-03) (hash(x)=77505483)
+1770 train 4.107567 (lr=2.4833e-03) (hash(x)=80847695)
+1780 train 4.077849 (lr=2.4741e-03) (hash(x)=70358393)
+1790 train 4.297462 (lr=2.4649e-03) (hash(x)=86191041)
+1800 val loss 4.2534
+1800 val perplexity 70.3414
+1800 train 4.316215 (lr=2.4556e-03) (hash(x)=77210920)
+1810 train 4.206339 (lr=2.4463e-03) (hash(x)=80523396)
+1820 train 4.348994 (lr=2.4369e-03) (hash(x)=87014389)
+1830 train 4.158744 (lr=2.4275e-03) (hash(x)=73990672)
+1840 train 4.259142 (lr=2.4180e-03) (hash(x)=91802507)
+1850 train 4.234425 (lr=2.4084e-03) (hash(x)=76955322)
+1860 train 4.067959 (lr=2.3988e-03) (hash(x)=77109229)
+1870 train 4.121253 (lr=2.3891e-03) (hash(x)=90135004)
+1880 train 4.094893 (lr=2.3794e-03) (hash(x)=73788366)
+1890 train 4.113547 (lr=2.3696e-03) (hash(x)=68877898)
+1900 val loss 4.2494
+1900 val perplexity 70.0620
+1900 train 4.103993 (lr=2.3598e-03) (hash(x)=78685171)
+1910 train 4.072052 (lr=2.3499e-03) (hash(x)=77401876)
+1920 train 4.283054 (lr=2.3400e-03) (hash(x)=76083249)
+1930 train 4.355591 (lr=2.3300e-03) (hash(x)=78497419)
+1940 train 4.178007 (lr=2.3200e-03) (hash(x)=77768007)
+1950 train 4.232255 (lr=2.3099e-03) (hash(x)=78166365)
+1960 train 4.029995 (lr=2.2998e-03) (hash(x)=76255691)
+1970 train 4.117794 (lr=2.2896e-03) (hash(x)=87599964)
+1980 train 4.089890 (lr=2.2793e-03) (hash(x)=76977354)
+1990 train 4.125253 (lr=2.2691e-03) (hash(x)=75418637)
+2000 val loss 4.2124
+2000 val perplexity 67.5209
+2000 train 4.152087 (lr=2.2588e-03) (hash(x)=74505485)
+2010 train 4.127440 (lr=2.2484e-03) (hash(x)=77968902)
+2020 train 3.986981 (lr=2.2380e-03) (hash(x)=74473597)
+2030 train 4.194334 (lr=2.2275e-03) (hash(x)=74915423)
+2040 train 4.135192 (lr=2.2170e-03) (hash(x)=80532849)
+2050 train 4.030729 (lr=2.2065e-03) (hash(x)=70871513)
+2060 train 4.303420 (lr=2.1959e-03) (hash(x)=83258099)
+2070 train 4.308839 (lr=2.1853e-03) (hash(x)=79318928)
+2080 train 4.107113 (lr=2.1746e-03) (hash(x)=80046106)
+2090 train 4.122273 (lr=2.1639e-03) (hash(x)=79574366)
+2100 val loss 4.1959
+2100 val perplexity 66.4159
+2100 train 4.144666 (lr=2.1532e-03) (hash(x)=78403637)
+2110 train 4.034763 (lr=2.1424e-03) (hash(x)=83330861)
+2120 train 4.129266 (lr=2.1316e-03) (hash(x)=82018746)
+2130 train 4.058628 (lr=2.1208e-03) (hash(x)=78997551)
+2140 train 4.021439 (lr=2.1099e-03) (hash(x)=73644529)
+2150 train 4.186209 (lr=2.0990e-03) (hash(x)=80653570)
+2160 train 4.079186 (lr=2.0881e-03) (hash(x)=73393914)
+2170 train 3.980767 (lr=2.0771e-03) (hash(x)=79883496)
+2180 train 4.030167 (lr=2.0661e-03) (hash(x)=78292168)
+2190 train 4.117309 (lr=2.0550e-03) (hash(x)=78366011)
+2200 val loss 4.1828
+2200 val perplexity 65.5490
+2200 train 4.207850 (lr=2.0440e-03) (hash(x)=87877942)
+2210 train 4.159712 (lr=2.0329e-03) (hash(x)=82277941)
+2220 train 4.104120 (lr=2.0217e-03) (hash(x)=76353795)
+2230 train 4.238162 (lr=2.0106e-03) (hash(x)=84692047)
+2240 train 4.165619 (lr=1.9994e-03) (hash(x)=86440918)
+2250 train 4.130452 (lr=1.9882e-03) (hash(x)=74843751)
+2260 train 4.181947 (lr=1.9770e-03) (hash(x)=76370726)
+2270 train 4.142017 (lr=1.9657e-03) (hash(x)=81088985)
+2280 train 3.992728 (lr=1.9544e-03) (hash(x)=82596808)
+2290 train 4.033587 (lr=1.9431e-03) (hash(x)=82101185)
+2300 val loss 4.1602
+2300 val perplexity 64.0832
+2300 train 4.129260 (lr=1.9318e-03) (hash(x)=76784929)
+2310 train 4.116089 (lr=1.9205e-03) (hash(x)=79063641)
+2320 train 4.078699 (lr=1.9091e-03) (hash(x)=72997210)
+2330 train 4.137936 (lr=1.8977e-03) (hash(x)=78364076)
+2340 train 4.115185 (lr=1.8863e-03) (hash(x)=80569869)
+2350 train 4.105857 (lr=1.8749e-03) (hash(x)=82997001)
+2360 train 4.175104 (lr=1.8635e-03) (hash(x)=80217398)
+2370 train 4.189958 (lr=1.8520e-03) (hash(x)=79078550)
+2380 train 4.114212 (lr=1.8406e-03) (hash(x)=80109951)
+2390 train 4.067819 (lr=1.8291e-03) (hash(x)=81684929)
+2400 val loss 4.1422
+2400 val perplexity 62.9401
+2400 train 4.047941 (lr=1.8176e-03) (hash(x)=68751146)
+2410 train 4.088289 (lr=1.8061e-03) (hash(x)=78732720)
+2420 train 3.917746 (lr=1.7946e-03) (hash(x)=71994630)
+2430 train 4.182696 (lr=1.7830e-03) (hash(x)=77169990)
+2440 train 4.056512 (lr=1.7715e-03) (hash(x)=79305139)
+2450 train 3.906474 (lr=1.7600e-03) (hash(x)=81449659)
+2460 train 4.028008 (lr=1.7484e-03) (hash(x)=77646193)
+2470 train 4.066309 (lr=1.7368e-03) (hash(x)=82360323)
+2480 train 4.085463 (lr=1.7253e-03) (hash(x)=77493945)
+2490 train 4.203562 (lr=1.7137e-03) (hash(x)=86536175)
+2500 val loss 4.1245
+2500 val perplexity 61.8346
+2500 train 4.265385 (lr=1.7021e-03) (hash(x)=85247781)
+2510 train 4.210231 (lr=1.6906e-03) (hash(x)=68014473)
+2520 train 3.985035 (lr=1.6790e-03) (hash(x)=76322450)
+2530 train 4.018965 (lr=1.6674e-03) (hash(x)=72040116)
+2540 train 4.076411 (lr=1.6558e-03) (hash(x)=72522060)
+2550 train 4.014823 (lr=1.6442e-03) (hash(x)=75822010)
+2560 train 3.960089 (lr=1.6326e-03) (hash(x)=80473391)
+2570 train 4.042657 (lr=1.6210e-03) (hash(x)=75327173)
+2580 train 3.897619 (lr=1.6094e-03) (hash(x)=79012888)
+2590 train 3.920714 (lr=1.5979e-03) (hash(x)=90222207)
+2600 val loss 4.1143
+2600 val perplexity 61.2100
+2600 train 3.922744 (lr=1.5863e-03) (hash(x)=72605303)
+2610 train 4.032545 (lr=1.5747e-03) (hash(x)=63545259)
+2620 train 4.159969 (lr=1.5632e-03) (hash(x)=75398896)
+2630 train 4.171215 (lr=1.5516e-03) (hash(x)=76569876)
+2640 train 4.266657 (lr=1.5400e-03) (hash(x)=81208719)
+2650 train 4.058051 (lr=1.5285e-03) (hash(x)=78825850)
+2660 train 4.071661 (lr=1.5170e-03) (hash(x)=71007296)
+2670 train 4.075276 (lr=1.5054e-03) (hash(x)=85702547)
+2680 train 3.998211 (lr=1.4939e-03) (hash(x)=84007025)
+2690 train 4.035367 (lr=1.4824e-03) (hash(x)=79114893)
+2700 val loss 4.0917
+2700 val perplexity 59.8425
+2700 train 4.066069 (lr=1.4709e-03) (hash(x)=76100226)
+2710 train 3.902904 (lr=1.4594e-03) (hash(x)=95412405)
+2720 train 3.976934 (lr=1.4480e-03) (hash(x)=77968369)
+2730 train 3.984725 (lr=1.4365e-03) (hash(x)=83745566)
+2740 train 3.902402 (lr=1.4251e-03) (hash(x)=71968605)
+2750 train 3.980075 (lr=1.4137e-03) (hash(x)=78030093)
+2760 train 4.188598 (lr=1.4023e-03) (hash(x)=82034862)
+2770 train 3.998249 (lr=1.3909e-03) (hash(x)=81426905)
+2780 train 4.114886 (lr=1.3795e-03) (hash(x)=80804180)
+2790 train 4.155402 (lr=1.3682e-03) (hash(x)=78769818)
+2800 val loss 4.0747
+2800 val perplexity 58.8314
+2800 train 3.933600 (lr=1.3569e-03) (hash(x)=81546487)
+2810 train 3.981787 (lr=1.3456e-03) (hash(x)=81942293)
+2820 train 3.934145 (lr=1.3343e-03) (hash(x)=79628664)
+2830 train 4.005183 (lr=1.3230e-03) (hash(x)=78405055)
+2840 train 3.969283 (lr=1.3118e-03) (hash(x)=74847727)
+2850 train 3.923479 (lr=1.3006e-03) (hash(x)=81869030)
+2860 train 3.902036 (lr=1.2894e-03) (hash(x)=100536172)
+2870 train 3.805861 (lr=1.2783e-03) (hash(x)=75411272)
+2880 train 3.961459 (lr=1.2671e-03) (hash(x)=81382860)
+2890 train 4.106556 (lr=1.2560e-03) (hash(x)=81157461)
+2900 val loss 4.0732
+2900 val perplexity 58.7425
+2900 train 4.046466 (lr=1.2450e-03) (hash(x)=76856547)
+2910 train 4.033590 (lr=1.2339e-03) (hash(x)=76535873)
+2920 train 4.058295 (lr=1.2229e-03) (hash(x)=77761442)
+2930 train 3.957771 (lr=1.2119e-03) (hash(x)=75488021)
+2940 train 3.942060 (lr=1.2010e-03) (hash(x)=85879517)
+2950 train 4.052075 (lr=1.1901e-03) (hash(x)=79599481)
+2960 train 3.925832 (lr=1.1792e-03) (hash(x)=76820375)
+2970 train 3.918752 (lr=1.1684e-03) (hash(x)=77945910)
+2980 train 3.957390 (lr=1.1576e-03) (hash(x)=83034701)
+2990 train 3.933461 (lr=1.1468e-03) (hash(x)=75166389)
+3000 val loss 4.0497
+3000 val perplexity 57.3798
+3000 train 3.979269 (lr=1.1361e-03) (hash(x)=73461176)
+3010 train 3.847323 (lr=1.1254e-03) (hash(x)=71826799)
+3020 train 3.799285 (lr=1.1147e-03) (hash(x)=90013784)
+3030 train 3.896715 (lr=1.1041e-03) (hash(x)=68244392)
+3040 train 4.068723 (lr=1.0935e-03) (hash(x)=81477870)
+3050 train 3.984149 (lr=1.0830e-03) (hash(x)=89071978)
+3060 train 4.183552 (lr=1.0725e-03) (hash(x)=78909964)
+3070 train 4.078857 (lr=1.0620e-03) (hash(x)=72927540)
+3080 train 3.999731 (lr=1.0516e-03) (hash(x)=78576606)
+3090 train 4.086051 (lr=1.0412e-03) (hash(x)=83549010)
+3100 val loss 4.0422
+3100 val perplexity 56.9513
+3100 train 4.109275 (lr=1.0309e-03) (hash(x)=88378940)
+3110 train 3.960761 (lr=1.0207e-03) (hash(x)=75361195)
+3120 train 3.955056 (lr=1.0104e-03) (hash(x)=81548673)
+3130 train 3.935891 (lr=1.0002e-03) (hash(x)=76723731)
+3140 train 4.003508 (lr=9.9011e-04) (hash(x)=78335834)
+3150 train 3.968460 (lr=9.8002e-04) (hash(x)=79512807)
+3160 train 3.988152 (lr=9.6999e-04) (hash(x)=78022958)
+3170 train 4.079013 (lr=9.6000e-04) (hash(x)=48780915)
+3180 train 4.047294 (lr=9.5007e-04) (hash(x)=72859953)
+3190 train 4.020719 (lr=9.4019e-04) (hash(x)=70441376)
+3200 val loss 4.0134
+3200 val perplexity 55.3346
+3200 train 4.033336 (lr=9.3036e-04) (hash(x)=85361027)
+3210 train 4.108481 (lr=9.2058e-04) (hash(x)=84671249)
+3220 train 4.048403 (lr=9.1085e-04) (hash(x)=77654883)
+3230 train 3.969444 (lr=9.0118e-04) (hash(x)=74937337)
+3240 train 4.117733 (lr=8.9157e-04) (hash(x)=77182900)
+3250 train 4.031063 (lr=8.8201e-04) (hash(x)=86214664)
+3260 train 3.909645 (lr=8.7251e-04) (hash(x)=83267879)
+3270 train 3.959313 (lr=8.6307e-04) (hash(x)=71561392)
+3280 train 3.983171 (lr=8.5368e-04) (hash(x)=83098376)
+3290 train 3.869457 (lr=8.4435e-04) (hash(x)=77431692)
+3300 val loss 4.0078
+3300 val perplexity 55.0255
+3300 train 4.025945 (lr=8.3508e-04) (hash(x)=82492818)
+3310 train 4.043591 (lr=8.2588e-04) (hash(x)=80149788)
+3320 train 3.990059 (lr=8.1673e-04) (hash(x)=78903528)
+3330 train 4.045711 (lr=8.0764e-04) (hash(x)=81042090)
+3340 train 3.987937 (lr=7.9862e-04) (hash(x)=76895955)
+3350 train 4.032351 (lr=7.8966e-04) (hash(x)=76790968)
+3360 train 4.046820 (lr=7.8076e-04) (hash(x)=85235065)
+3370 train 3.991080 (lr=7.7192e-04) (hash(x)=83413227)
+3380 train 3.966572 (lr=7.6315e-04) (hash(x)=94520262)
+3390 train 3.978065 (lr=7.5445e-04) (hash(x)=86213149)
+3400 val loss 3.9961
+3400 val perplexity 54.3837
+3400 train 3.936376 (lr=7.4581e-04) (hash(x)=77028659)
+3410 train 3.884609 (lr=7.3724e-04) (hash(x)=81165199)
+3420 train 3.995416 (lr=7.2874e-04) (hash(x)=78852885)
+3430 train 3.960410 (lr=7.2030e-04) (hash(x)=82364796)
+3440 train 4.165849 (lr=7.1193e-04) (hash(x)=66884329)
+3450 train 3.986905 (lr=7.0363e-04) (hash(x)=82196927)
+3460 train 4.042554 (lr=6.9541e-04) (hash(x)=83570482)
+3470 train 3.941086 (lr=6.8725e-04) (hash(x)=86984462)
+3480 train 4.035233 (lr=6.7916e-04) (hash(x)=80446942)
+3490 train 3.979914 (lr=6.7114e-04) (hash(x)=82386020)
+3500 val loss 3.9823
+3500 val perplexity 53.6415
+3500 train 3.887445 (lr=6.6320e-04) (hash(x)=82573783)
+3510 train 3.991865 (lr=6.5533e-04) (hash(x)=86296100)
+3520 train 3.967915 (lr=6.4753e-04) (hash(x)=79959476)
+3530 train 3.986896 (lr=6.3981e-04) (hash(x)=77625970)
+3540 train 3.984396 (lr=6.3216e-04) (hash(x)=75218126)
+3550 train 3.903673 (lr=6.2458e-04) (hash(x)=85036598)
+3560 train 4.212748 (lr=6.1708e-04) (hash(x)=68457295)
+3570 train 3.903472 (lr=6.0966e-04) (hash(x)=74945892)
+3580 train 3.824989 (lr=6.0231e-04) (hash(x)=63411286)
+3590 train 3.955040 (lr=5.9504e-04) (hash(x)=81331248)
+3600 val loss 3.9723
+3600 val perplexity 53.1076
+3600 train 4.040999 (lr=5.8785e-04) (hash(x)=78629131)
+3610 train 4.033780 (lr=5.8074e-04) (hash(x)=88173496)
+3620 train 3.861711 (lr=5.7370e-04) (hash(x)=83159981)
+3630 train 4.022011 (lr=5.6675e-04) (hash(x)=73612411)
+3640 train 4.017386 (lr=5.5987e-04) (hash(x)=78561998)
+3650 train 4.082081 (lr=5.5308e-04) (hash(x)=81233526)
+3660 train 4.056116 (lr=5.4636e-04) (hash(x)=77460490)
+3670 train 3.877188 (lr=5.3973e-04) (hash(x)=79570240)
+3680 train 3.806240 (lr=5.3318e-04) (hash(x)=76916005)
+3690 train 4.038084 (lr=5.2671e-04) (hash(x)=74947748)
+3700 val loss 3.9667
+3700 val perplexity 52.8078
+3700 train 3.952035 (lr=5.2033e-04) (hash(x)=76363057)
+3710 train 3.891888 (lr=5.1402e-04) (hash(x)=74295889)
+3720 train 3.802933 (lr=5.0780e-04) (hash(x)=73447634)
+3730 train 3.901215 (lr=5.0167e-04) (hash(x)=85050263)
+3740 train 4.023206 (lr=4.9562e-04) (hash(x)=73977156)
+3750 train 3.939067 (lr=4.8965e-04) (hash(x)=79321142)
+3760 train 3.999273 (lr=4.8377e-04) (hash(x)=80904167)
+3770 train 3.910762 (lr=4.7798e-04) (hash(x)=82485963)
+3780 train 3.958925 (lr=4.7227e-04) (hash(x)=74274733)
+3790 train 3.966989 (lr=4.6665e-04) (hash(x)=95150488)
+3800 val loss 3.9488
+3800 val perplexity 51.8717
+3800 train 3.910332 (lr=4.6112e-04) (hash(x)=83818907)
+3810 train 3.951266 (lr=4.5567e-04) (hash(x)=77222344)
+3820 train 3.813209 (lr=4.5031e-04) (hash(x)=81087000)
+3830 train 3.900906 (lr=4.4504e-04) (hash(x)=81910357)
+3840 train 3.844996 (lr=4.3986e-04) (hash(x)=80293916)
+3850 train 3.953712 (lr=4.3477e-04) (hash(x)=70921646)
+3860 train 3.707465 (lr=4.2977e-04) (hash(x)=72790321)
+3870 train 3.674891 (lr=4.2486e-04) (hash(x)=78176722)
+3880 train 3.612955 (lr=4.2004e-04) (hash(x)=70542120)
+3890 train 3.719953 (lr=4.1530e-04) (hash(x)=79612492)
+3900 val loss 3.9552
+3900 val perplexity 52.2042
+3900 train 3.926685 (lr=4.1066e-04) (hash(x)=77006513)
+3910 train 3.996871 (lr=4.0611e-04) (hash(x)=85113117)
+3920 train 3.918746 (lr=4.0166e-04) (hash(x)=76822278)
+3930 train 3.926543 (lr=3.9729e-04) (hash(x)=84320111)
+3940 train 3.924459 (lr=3.9302e-04) (hash(x)=71475995)
+3950 train 3.968514 (lr=3.8884e-04) (hash(x)=79394971)
+3960 train 3.848045 (lr=3.8475e-04) (hash(x)=94369441)
+3970 train 3.960546 (lr=3.8076e-04) (hash(x)=77166806)
+3980 train 3.960797 (lr=3.7685e-04) (hash(x)=77834199)
+3990 train 3.838659 (lr=3.7305e-04) (hash(x)=77689578)
+4000 val loss 3.9337
+4000 val perplexity 51.0936
+4000 train 3.796345 (lr=3.6933e-04) (hash(x)=70746393)
+4010 train 3.814718 (lr=3.6572e-04) (hash(x)=85359486)
+4020 train 3.897910 (lr=3.6219e-04) (hash(x)=78457700)
+4030 train 3.851814 (lr=3.5876e-04) (hash(x)=78953976)
+4040 train 3.707184 (lr=3.5543e-04) (hash(x)=84086579)
+4050 train 3.655347 (lr=3.5219e-04) (hash(x)=84131543)
+4060 train 3.669833 (lr=3.4905e-04) (hash(x)=77221961)
+4070 train 3.576262 (lr=3.4600e-04) (hash(x)=87527687)
+4080 train 3.721795 (lr=3.4305e-04) (hash(x)=73788501)
+4090 train 3.912457 (lr=3.4019e-04) (hash(x)=87378361)
+4100 val loss 3.9331
+4100 val perplexity 51.0635
+4100 train 3.944263 (lr=3.3744e-04) (hash(x)=74282076)
+4110 train 3.921849 (lr=3.3477e-04) (hash(x)=86416721)
+4120 train 3.868568 (lr=3.3221e-04) (hash(x)=86325819)
+4130 train 3.887930 (lr=3.2974e-04) (hash(x)=77534282)
+4140 train 3.886406 (lr=3.2737e-04) (hash(x)=73179409)
+4150 train 3.936681 (lr=3.2510e-04) (hash(x)=91295398)
+4160 train 3.895377 (lr=3.2292e-04) (hash(x)=84833773)
+4170 train 3.902901 (lr=3.2085e-04) (hash(x)=74079521)
+4180 train 3.809912 (lr=3.1887e-04) (hash(x)=74585954)
+4190 train 3.842374 (lr=3.1699e-04) (hash(x)=74849974)
+4200 val loss 3.9263
+4200 val perplexity 50.7171
+4200 train 3.940420 (lr=3.1520e-04) (hash(x)=87782408)
+4210 train 3.993524 (lr=3.1352e-04) (hash(x)=71390249)
+4220 train 3.683071 (lr=3.1193e-04) (hash(x)=80184408)
+4230 train 3.696821 (lr=3.1044e-04) (hash(x)=78607457)
+4240 train 3.720310 (lr=3.0905e-04) (hash(x)=83027401)
+4250 train 3.577924 (lr=3.0776e-04) (hash(x)=80802729)
+4260 train 3.878965 (lr=3.0657e-04) (hash(x)=75252098)
+4270 train 3.921788 (lr=3.0548e-04) (hash(x)=79153989)
+4280 train 4.021343 (lr=3.0449e-04) (hash(x)=85833030)
+4290 train 3.970921 (lr=3.0359e-04) (hash(x)=84103039)
+4300 val loss 3.9198
+4300 val perplexity 50.3899
+4300 train 4.097700 (lr=3.0280e-04) (hash(x)=76802610)
+4310 train 3.998301 (lr=3.0210e-04) (hash(x)=77169206)
+4320 train 3.811766 (lr=3.0150e-04) (hash(x)=83226021)
+4330 train 3.803866 (lr=3.0101e-04) (hash(x)=80213024)
+4340 train 3.827402 (lr=3.0061e-04) (hash(x)=80732928)
+4350 train 3.878859 (lr=3.0031e-04) (hash(x)=81534899)
+4360 train 3.832828 (lr=3.0011e-04) (hash(x)=80854213)
+4370 train 3.907735 (lr=3.0001e-04) (hash(x)=76420919)
+4374 val loss 3.9211
+4374 val perplexity 50.4547

12_mini_head_two_masks_4_heads/model_04374.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:309e45440b003ceaef00048e164db99d73850a4b1ee1004b6cf3554a2b5615af
+oid sha256:52d6a11584690bee299b949e198c1a2f37c54b8e2adf89ef712f56cb16c692b2
 size 98537090

12_mini_head_two_masks_4_heads/optimizer_04374.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ed5891e1d1b65b2efcde8ead0bf06328e61142c6959a730b1c2198d70b90b66b
+oid sha256:818d2f1f2e45be7fe065d4cd822aee93c9cf6b9ff01168e38c934af55d84ff12
 size 190793606