Upload folder using huggingface_hub

att_conv_playground/f_learnable_compile_yes_weight_decay_seed_1339/args.json

@@ -0,0 +1 @@
+{"hellaswag": true, "attention_kind": "selective", "log_dir": "att_conv_playground/f_learnable_compile_yes_weight_decay_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": "att_conv_playground", "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": 2, "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, "att_conv": true, "att_conv_init": "eye", "att_conv_scale": 1.0, "att_conv_weight_decay": true, "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, "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": "f_learnable_compile_yes_weight_decay", "latent_mask_precision": "float32", "att_conv_precision": "bfloat16"}

att_conv_playground/f_learnable_compile_yes_weight_decay_seed_1339/dataloader_04374.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6031fd3e2855a036f7a5531cc24555aabd1115f9dd6618b8b2ca6f55279ef0b2
+size 964

att_conv_playground/f_learnable_compile_yes_weight_decay_seed_1339/log2.txt

@@ -0,0 +1,529 @@
+max_steps: 4375
+0 val loss 11.2970
+0 val perplexity 80580.0703
+0 train 11.313997 (lr=4.1958e-06) (hash(x)=44992657)
+10 train 9.929086 (lr=4.6154e-05) (hash(x)=33468021)
+20 train 9.376823 (lr=8.8112e-05) (hash(x)=40941803)
+30 train 8.675874 (lr=1.3007e-04) (hash(x)=36715902)
+40 train 7.942028 (lr=1.7203e-04) (hash(x)=32710993)
+50 train 7.703146 (lr=2.1399e-04) (hash(x)=43839896)
+60 train 7.752395 (lr=2.5594e-04) (hash(x)=40675468)
+70 train 7.465556 (lr=2.9790e-04) (hash(x)=34592631)
+80 train 7.426189 (lr=3.3986e-04) (hash(x)=44444845)
+90 train 7.184563 (lr=3.8182e-04) (hash(x)=41965258)
+100 val loss 7.0671
+100 val perplexity 1172.7942
+100 train 6.922498 (lr=4.2378e-04) (hash(x)=41284750)
+110 train 6.869315 (lr=4.6573e-04) (hash(x)=41118734)
+120 train 6.715339 (lr=5.0769e-04) (hash(x)=37537547)
+130 train 6.562676 (lr=5.4965e-04) (hash(x)=43625179)
+140 train 6.553432 (lr=5.9161e-04) (hash(x)=41940760)
+150 train 6.608790 (lr=6.3357e-04) (hash(x)=39210431)
+160 train 6.531977 (lr=6.7552e-04) (hash(x)=41128294)
+170 train 6.432656 (lr=7.1748e-04) (hash(x)=41590227)
+180 train 6.362526 (lr=7.5944e-04) (hash(x)=38084871)
+190 train 6.286448 (lr=8.0140e-04) (hash(x)=34534333)
+200 val loss 6.1931
+200 val perplexity 489.3398
+200 train 6.226174 (lr=8.4336e-04) (hash(x)=41614299)
+210 train 6.118168 (lr=8.8531e-04) (hash(x)=35551597)
+220 train 6.030597 (lr=9.2727e-04) (hash(x)=44615432)
+230 train 6.172233 (lr=9.6923e-04) (hash(x)=37078768)
+240 train 5.866935 (lr=1.0112e-03) (hash(x)=40644078)
+250 train 5.925122 (lr=1.0531e-03) (hash(x)=44636066)
+260 train 5.806701 (lr=1.0951e-03) (hash(x)=43247952)
+270 train 5.767631 (lr=1.1371e-03) (hash(x)=37989370)
+280 train 5.708784 (lr=1.1790e-03) (hash(x)=36229730)
+290 train 5.621446 (lr=1.2210e-03) (hash(x)=35032996)
+300 val loss 5.7315
+300 val perplexity 308.4226
+300 train 5.478877 (lr=1.2629e-03) (hash(x)=33868636)
+310 train 5.416961 (lr=1.3049e-03) (hash(x)=39782705)
+320 train 5.526266 (lr=1.3469e-03) (hash(x)=43525765)
+330 train 5.594296 (lr=1.3888e-03) (hash(x)=38074926)
+340 train 5.649379 (lr=1.4308e-03) (hash(x)=50125702)
+350 train 5.630602 (lr=1.4727e-03) (hash(x)=35062222)
+360 train 5.560308 (lr=1.5147e-03) (hash(x)=39671001)
+370 train 5.600841 (lr=1.5566e-03) (hash(x)=35748364)
+380 train 5.391724 (lr=1.5986e-03) (hash(x)=39962666)
+390 train 5.356234 (lr=1.6406e-03) (hash(x)=39213360)
+400 val loss 5.3779
+400 val perplexity 216.5712
+400 train 5.361853 (lr=1.6825e-03) (hash(x)=37919584)
+410 train 5.306700 (lr=1.7245e-03) (hash(x)=42001527)
+420 train 5.292449 (lr=1.7664e-03) (hash(x)=33940005)
+430 train 5.352309 (lr=1.8084e-03) (hash(x)=41749217)
+440 train 5.244893 (lr=1.8503e-03) (hash(x)=45219739)
+450 train 5.223666 (lr=1.8923e-03) (hash(x)=38515090)
+460 train 5.132326 (lr=1.9343e-03) (hash(x)=37518607)
+470 train 4.944524 (lr=1.9762e-03) (hash(x)=46830716)
+480 train 5.053914 (lr=2.0182e-03) (hash(x)=49262598)
+490 train 4.852821 (lr=2.0601e-03) (hash(x)=41370559)
+500 val loss 5.1668
+500 val perplexity 175.3445
+500 train 4.940689 (lr=2.1021e-03) (hash(x)=35237459)
+510 train 5.036999 (lr=2.1441e-03) (hash(x)=35070579)
+520 train 5.114145 (lr=2.1860e-03) (hash(x)=37355110)
+530 train 5.177792 (lr=2.2280e-03) (hash(x)=39398044)
+540 train 5.107660 (lr=2.2699e-03) (hash(x)=39156676)
+550 train 5.036586 (lr=2.3119e-03) (hash(x)=38553909)
+560 train 5.087225 (lr=2.3538e-03) (hash(x)=36555371)
+570 train 5.007078 (lr=2.3958e-03) (hash(x)=42048991)
+580 train 4.947815 (lr=2.4378e-03) (hash(x)=37359399)
+590 train 5.002294 (lr=2.4797e-03) (hash(x)=39254663)
+600 val loss 4.9401
+600 val perplexity 139.7876
+600 train 4.953729 (lr=2.5217e-03) (hash(x)=38982539)
+610 train 4.890610 (lr=2.5636e-03) (hash(x)=35859145)
+620 train 4.855858 (lr=2.6056e-03) (hash(x)=35593542)
+630 train 4.766426 (lr=2.6476e-03) (hash(x)=39124418)
+640 train 4.756506 (lr=2.6895e-03) (hash(x)=39107374)
+650 train 4.698158 (lr=2.7315e-03) (hash(x)=34529151)
+660 train 4.639983 (lr=2.7734e-03) (hash(x)=39483931)
+670 train 4.621902 (lr=2.8154e-03) (hash(x)=43008010)
+680 train 4.513530 (lr=2.8573e-03) (hash(x)=33469328)
+690 train 4.564831 (lr=2.8993e-03) (hash(x)=35025253)
+700 val loss 4.7976
+700 val perplexity 121.2214
+700 train 4.527308 (lr=2.9413e-03) (hash(x)=43280188)
+710 train 4.817490 (lr=2.9832e-03) (hash(x)=40999263)
+720 train 4.806442 (lr=3.0000e-03) (hash(x)=38304742)
+730 train 4.817601 (lr=2.9999e-03) (hash(x)=34705188)
+740 train 4.762948 (lr=2.9997e-03) (hash(x)=40028964)
+750 train 4.688683 (lr=2.9994e-03) (hash(x)=39051591)
+760 train 4.751065 (lr=2.9990e-03) (hash(x)=41286198)
+770 train 4.690892 (lr=2.9985e-03) (hash(x)=44512967)
+780 train 4.719403 (lr=2.9979e-03) (hash(x)=45171627)
+790 train 4.704153 (lr=2.9972e-03) (hash(x)=46345673)
+800 val loss 4.6921
+800 val perplexity 109.0816
+800 train 4.648660 (lr=2.9964e-03) (hash(x)=35529545)
+810 train 4.546840 (lr=2.9955e-03) (hash(x)=42321793)
+820 train 4.564155 (lr=2.9945e-03) (hash(x)=49327569)
+830 train 4.624649 (lr=2.9934e-03) (hash(x)=41325057)
+840 train 4.575268 (lr=2.9922e-03) (hash(x)=37305614)
+850 train 4.532531 (lr=2.9909e-03) (hash(x)=38000800)
+860 train 4.654103 (lr=2.9896e-03) (hash(x)=44047697)
+870 train 4.650467 (lr=2.9881e-03) (hash(x)=41576206)
+880 train 4.622203 (lr=2.9865e-03) (hash(x)=44685458)
+890 train 4.603188 (lr=2.9848e-03) (hash(x)=40398597)
+900 val loss 4.5837
+900 val perplexity 97.8749
+900 train 4.545106 (lr=2.9830e-03) (hash(x)=39312673)
+910 train 4.560645 (lr=2.9811e-03) (hash(x)=38881875)
+920 train 4.585423 (lr=2.9792e-03) (hash(x)=39140166)
+930 train 4.479328 (lr=2.9771e-03) (hash(x)=39075781)
+940 train 4.496046 (lr=2.9749e-03) (hash(x)=46887071)
+950 train 4.490038 (lr=2.9726e-03) (hash(x)=39740603)
+960 train 4.540222 (lr=2.9703e-03) (hash(x)=39590268)
+970 train 4.490891 (lr=2.9678e-03) (hash(x)=43580052)
+980 train 4.431120 (lr=2.9652e-03) (hash(x)=40226735)
+990 train 4.537045 (lr=2.9626e-03) (hash(x)=40723896)
+1000 val loss 4.4951
+1000 val perplexity 89.5790
+1000 train 4.498920 (lr=2.9598e-03) (hash(x)=39408050)
+1010 train 4.687536 (lr=2.9570e-03) (hash(x)=32247950)
+1020 train 4.568910 (lr=2.9540e-03) (hash(x)=39049938)
+1030 train 4.530960 (lr=2.9510e-03) (hash(x)=37453736)
+1040 train 4.438890 (lr=2.9478e-03) (hash(x)=36649666)
+1050 train 4.506237 (lr=2.9446e-03) (hash(x)=41404097)
+1060 train 4.494990 (lr=2.9412e-03) (hash(x)=40701599)
+1070 train 4.509824 (lr=2.9378e-03) (hash(x)=41787393)
+1080 train 4.401343 (lr=2.9343e-03) (hash(x)=42325465)
+1090 train 4.439435 (lr=2.9307e-03) (hash(x)=45018925)
+1100 val loss 4.4317
+1100 val perplexity 84.0731
+1100 train 4.456787 (lr=2.9270e-03) (hash(x)=53751982)
+1110 train 4.386201 (lr=2.9231e-03) (hash(x)=43118376)
+1120 train 4.413826 (lr=2.9192e-03) (hash(x)=41489942)
+1130 train 4.435517 (lr=2.9152e-03) (hash(x)=48020487)
+1140 train 4.299523 (lr=2.9112e-03) (hash(x)=43266072)
+1150 train 4.492610 (lr=2.9070e-03) (hash(x)=41790132)
+1160 train 4.460152 (lr=2.9027e-03) (hash(x)=33739090)
+1170 train 4.430157 (lr=2.8983e-03) (hash(x)=43885030)
+1180 train 4.374289 (lr=2.8939e-03) (hash(x)=39046714)
+1190 train 4.408639 (lr=2.8893e-03) (hash(x)=44387023)
+1200 val loss 4.3637
+1200 val perplexity 78.5459
+1200 train 4.279100 (lr=2.8847e-03) (hash(x)=36667141)
+1210 train 4.345467 (lr=2.8800e-03) (hash(x)=45487179)
+1220 train 4.324526 (lr=2.8751e-03) (hash(x)=43456413)
+1230 train 4.257019 (lr=2.8702e-03) (hash(x)=42473632)
+1240 train 4.336111 (lr=2.8652e-03) (hash(x)=42966315)
+1250 train 4.180969 (lr=2.8601e-03) (hash(x)=42930524)
+1260 train 4.331042 (lr=2.8550e-03) (hash(x)=43467676)
+1270 train 4.254816 (lr=2.8497e-03) (hash(x)=40872580)
+1280 train 4.325104 (lr=2.8443e-03) (hash(x)=40619157)
+1290 train 4.336909 (lr=2.8389e-03) (hash(x)=44406165)
+1300 val loss 4.3173
+1300 val perplexity 74.9860
+1300 train 4.303840 (lr=2.8333e-03) (hash(x)=40808029)
+1310 train 4.475077 (lr=2.8277e-03) (hash(x)=46125736)
+1320 train 4.237193 (lr=2.8220e-03) (hash(x)=41698487)
+1330 train 4.279537 (lr=2.8162e-03) (hash(x)=39302878)
+1340 train 4.384329 (lr=2.8103e-03) (hash(x)=41826369)
+1350 train 4.204286 (lr=2.8044e-03) (hash(x)=37061413)
+1360 train 4.333019 (lr=2.7983e-03) (hash(x)=40770974)
+1370 train 4.239984 (lr=2.7922e-03) (hash(x)=38738718)
+1380 train 4.238402 (lr=2.7860e-03) (hash(x)=34961558)
+1390 train 4.288127 (lr=2.7797e-03) (hash(x)=38559342)
+1400 val loss 4.3028
+1400 val perplexity 73.9043
+1400 train 4.246294 (lr=2.7733e-03) (hash(x)=38840020)
+1410 train 4.396265 (lr=2.7668e-03) (hash(x)=44448100)
+1420 train 4.459045 (lr=2.7603e-03) (hash(x)=39199838)
+1430 train 4.252914 (lr=2.7536e-03) (hash(x)=41669873)
+1440 train 4.374593 (lr=2.7469e-03) (hash(x)=53253836)
+1450 train 4.304939 (lr=2.7401e-03) (hash(x)=41320631)
+1460 train 4.216435 (lr=2.7333e-03) (hash(x)=48462803)
+1470 train 4.260677 (lr=2.7263e-03) (hash(x)=42356864)
+1480 train 4.311851 (lr=2.7193e-03) (hash(x)=41118068)
+1490 train 4.221905 (lr=2.7121e-03) (hash(x)=37242854)
+1500 val loss 4.2570
+1500 val perplexity 70.5998
+1500 train 4.268400 (lr=2.7050e-03) (hash(x)=40571031)
+1510 train 4.219984 (lr=2.6977e-03) (hash(x)=35384608)
+1520 train 4.251860 (lr=2.6903e-03) (hash(x)=35341056)
+1530 train 4.138052 (lr=2.6829e-03) (hash(x)=36471779)
+1540 train 4.176340 (lr=2.6754e-03) (hash(x)=40356153)
+1550 train 4.216608 (lr=2.6678e-03) (hash(x)=40133377)
+1560 train 4.149473 (lr=2.6602e-03) (hash(x)=33722160)
+1570 train 4.290572 (lr=2.6525e-03) (hash(x)=39555295)
+1580 train 4.193474 (lr=2.6447e-03) (hash(x)=35950666)
+1590 train 4.349428 (lr=2.6368e-03) (hash(x)=36528535)
+1600 val loss 4.2376
+1600 val perplexity 69.2403
+1600 train 4.238150 (lr=2.6289e-03) (hash(x)=42613017)
+1610 train 4.382337 (lr=2.6208e-03) (hash(x)=42845965)
+1620 train 4.259151 (lr=2.6128e-03) (hash(x)=36137074)
+1630 train 4.230016 (lr=2.6046e-03) (hash(x)=37560569)
+1640 train 4.329985 (lr=2.5964e-03) (hash(x)=43680614)
+1650 train 4.186232 (lr=2.5881e-03) (hash(x)=37281218)
+1660 train 4.212779 (lr=2.5797e-03) (hash(x)=36165222)
+1670 train 4.167943 (lr=2.5713e-03) (hash(x)=50222513)
+1680 train 4.054999 (lr=2.5628e-03) (hash(x)=35239173)
+1690 train 4.111139 (lr=2.5542e-03) (hash(x)=40287349)
+1700 val loss 4.2374
+1700 val perplexity 69.2278
+1700 train 4.290979 (lr=2.5455e-03) (hash(x)=37444780)
+1710 train 4.273036 (lr=2.5368e-03) (hash(x)=38962723)
+1720 train 4.283546 (lr=2.5281e-03) (hash(x)=33882305)
+1730 train 4.259590 (lr=2.5192e-03) (hash(x)=36501540)
+1740 train 4.255309 (lr=2.5103e-03) (hash(x)=38723266)
+1750 train 4.195684 (lr=2.5014e-03) (hash(x)=38772789)
+1760 train 4.265649 (lr=2.4924e-03) (hash(x)=37707898)
+1770 train 4.131569 (lr=2.4833e-03) (hash(x)=30529327)
+1780 train 4.215312 (lr=2.4741e-03) (hash(x)=37885464)
+1790 train 4.228271 (lr=2.4649e-03) (hash(x)=39013967)
+1800 val loss 4.1973
+1800 val perplexity 66.5091
+1800 train 4.266657 (lr=2.4556e-03) (hash(x)=34906955)
+1810 train 4.139020 (lr=2.4463e-03) (hash(x)=40790682)
+1820 train 4.009101 (lr=2.4369e-03) (hash(x)=39912223)
+1830 train 4.135365 (lr=2.4275e-03) (hash(x)=39304486)
+1840 train 4.111629 (lr=2.4180e-03) (hash(x)=37247975)
+1850 train 4.094750 (lr=2.4084e-03) (hash(x)=41340614)
+1860 train 4.255919 (lr=2.3988e-03) (hash(x)=39924627)
+1870 train 4.234423 (lr=2.3891e-03) (hash(x)=40235993)
+1880 train 4.126999 (lr=2.3794e-03) (hash(x)=37693476)
+1890 train 4.244117 (lr=2.3696e-03) (hash(x)=40449864)
+1900 val loss 4.1677
+1900 val perplexity 64.5647
+1900 train 4.144708 (lr=2.3598e-03) (hash(x)=37907749)
+1910 train 4.121609 (lr=2.3499e-03) (hash(x)=34547901)
+1920 train 4.155052 (lr=2.3400e-03) (hash(x)=37771788)
+1930 train 4.027755 (lr=2.3300e-03) (hash(x)=36049454)
+1940 train 4.119054 (lr=2.3200e-03) (hash(x)=29564336)
+1950 train 4.112907 (lr=2.3099e-03) (hash(x)=41689281)
+1960 train 4.022210 (lr=2.2998e-03) (hash(x)=41976699)
+1970 train 4.220210 (lr=2.2896e-03) (hash(x)=41538621)
+1980 train 4.104960 (lr=2.2793e-03) (hash(x)=36490902)
+1990 train 4.134583 (lr=2.2691e-03) (hash(x)=42739315)
+2000 val loss 4.1505
+2000 val perplexity 63.4684
+2000 train 4.148649 (lr=2.2588e-03) (hash(x)=38600074)
+2010 train 4.163734 (lr=2.2484e-03) (hash(x)=34555488)
+2020 train 4.182618 (lr=2.2380e-03) (hash(x)=37227345)
+2030 train 4.138152 (lr=2.2275e-03) (hash(x)=38172386)
+2040 train 4.247901 (lr=2.2170e-03) (hash(x)=39340786)
+2050 train 4.094316 (lr=2.2065e-03) (hash(x)=37237298)
+2060 train 4.327292 (lr=2.1959e-03) (hash(x)=35894782)
+2070 train 4.148901 (lr=2.1853e-03) (hash(x)=44161861)
+2080 train 4.169876 (lr=2.1746e-03) (hash(x)=38062558)
+2090 train 4.017947 (lr=2.1639e-03) (hash(x)=38608301)
+2100 val loss 4.1297
+2100 val perplexity 62.1602
+2100 train 4.046555 (lr=2.1532e-03) (hash(x)=47611537)
+2110 train 4.046525 (lr=2.1424e-03) (hash(x)=38076651)
+2120 train 4.055971 (lr=2.1316e-03) (hash(x)=39870893)
+2130 train 4.207125 (lr=2.1208e-03) (hash(x)=48123321)
+2140 train 4.231561 (lr=2.1099e-03) (hash(x)=44124977)
+2150 train 4.132688 (lr=2.0990e-03) (hash(x)=42181374)
+2160 train 4.163044 (lr=2.0881e-03) (hash(x)=41837878)
+2170 train 4.176198 (lr=2.0771e-03) (hash(x)=42161375)
+2180 train 4.149577 (lr=2.0661e-03) (hash(x)=41792986)
+2190 train 4.118959 (lr=2.0550e-03) (hash(x)=41872416)
+2200 val loss 4.1074
+2200 val perplexity 60.7891
+2200 train 4.093473 (lr=2.0440e-03) (hash(x)=37395985)
+2210 train 4.132019 (lr=2.0329e-03) (hash(x)=40616096)
+2220 train 4.042325 (lr=2.0217e-03) (hash(x)=40118423)
+2230 train 3.961712 (lr=2.0106e-03) (hash(x)=39641586)
+2240 train 4.049077 (lr=1.9994e-03) (hash(x)=34720116)
+2250 train 3.992643 (lr=1.9882e-03) (hash(x)=43641508)
+2260 train 4.012753 (lr=1.9770e-03) (hash(x)=31179786)
+2270 train 3.981375 (lr=1.9657e-03) (hash(x)=42721932)
+2280 train 4.306044 (lr=1.9544e-03) (hash(x)=38474505)
+2290 train 4.132988 (lr=1.9431e-03) (hash(x)=41398545)
+2300 val loss 4.0920
+2300 val perplexity 59.8599
+2300 train 4.060452 (lr=1.9318e-03) (hash(x)=44131094)
+2310 train 4.140881 (lr=1.9205e-03) (hash(x)=41667948)
+2320 train 4.221953 (lr=1.9091e-03) (hash(x)=36339232)
+2330 train 4.096367 (lr=1.8977e-03) (hash(x)=41630644)
+2340 train 4.189849 (lr=1.8863e-03) (hash(x)=41278955)
+2350 train 4.057643 (lr=1.8749e-03) (hash(x)=39491976)
+2360 train 4.123240 (lr=1.8635e-03) (hash(x)=41771238)
+2370 train 4.220885 (lr=1.8520e-03) (hash(x)=45374570)
+2380 train 4.098701 (lr=1.8406e-03) (hash(x)=41419267)
+2390 train 4.128364 (lr=1.8291e-03) (hash(x)=38869169)
+2400 val loss 4.0761
+2400 val perplexity 58.9160
+2400 train 4.035477 (lr=1.8176e-03) (hash(x)=38619293)
+2410 train 3.851847 (lr=1.8061e-03) (hash(x)=41600240)
+2420 train 3.982152 (lr=1.7946e-03) (hash(x)=40891045)
+2430 train 3.976124 (lr=1.7830e-03) (hash(x)=41023249)
+2440 train 3.948826 (lr=1.7715e-03) (hash(x)=33813452)
+2450 train 3.898032 (lr=1.7600e-03) (hash(x)=38464119)
+2460 train 3.799547 (lr=1.7484e-03) (hash(x)=40699982)
+2470 train 3.806790 (lr=1.7368e-03) (hash(x)=38254854)
+2480 train 4.091753 (lr=1.7253e-03) (hash(x)=42736069)
+2490 train 4.105101 (lr=1.7137e-03) (hash(x)=41928525)
+2500 val loss 4.0568
+2500 val perplexity 57.7872
+2500 train 4.053959 (lr=1.7021e-03) (hash(x)=45864011)
+2510 train 4.068710 (lr=1.6906e-03) (hash(x)=41380960)
+2520 train 4.063651 (lr=1.6790e-03) (hash(x)=38366545)
+2530 train 4.071119 (lr=1.6674e-03) (hash(x)=41045176)
+2540 train 4.099214 (lr=1.6558e-03) (hash(x)=43435705)
+2550 train 4.055382 (lr=1.6442e-03) (hash(x)=40693090)
+2560 train 3.972857 (lr=1.6326e-03) (hash(x)=40502478)
+2570 train 4.003119 (lr=1.6210e-03) (hash(x)=38610920)
+2580 train 3.925151 (lr=1.6094e-03) (hash(x)=38333499)
+2590 train 4.002284 (lr=1.5979e-03) (hash(x)=41982736)
+2600 val loss 4.0398
+2600 val perplexity 56.8139
+2600 train 3.978622 (lr=1.5863e-03) (hash(x)=37724702)
+2610 train 3.919917 (lr=1.5747e-03) (hash(x)=39564630)
+2620 train 3.864377 (lr=1.5632e-03) (hash(x)=38465803)
+2630 train 3.788898 (lr=1.5516e-03) (hash(x)=36955007)
+2640 train 3.821404 (lr=1.5400e-03) (hash(x)=32994568)
+2650 train 4.002573 (lr=1.5285e-03) (hash(x)=41165765)
+2660 train 4.016108 (lr=1.5170e-03) (hash(x)=42934000)
+2670 train 4.208766 (lr=1.5054e-03) (hash(x)=42726316)
+2680 train 4.101425 (lr=1.4939e-03) (hash(x)=35880318)
+2690 train 3.983535 (lr=1.4824e-03) (hash(x)=41747329)
+2700 val loss 4.0291
+2700 val perplexity 56.2084
+2700 train 4.103441 (lr=1.4709e-03) (hash(x)=40259630)
+2710 train 4.228205 (lr=1.4594e-03) (hash(x)=39895798)
+2720 train 3.969559 (lr=1.4480e-03) (hash(x)=36146683)
+2730 train 3.944956 (lr=1.4365e-03) (hash(x)=36181984)
+2740 train 3.919886 (lr=1.4251e-03) (hash(x)=43700349)
+2750 train 3.990744 (lr=1.4137e-03) (hash(x)=41610597)
+2760 train 3.972270 (lr=1.4023e-03) (hash(x)=31183639)
+2770 train 3.972181 (lr=1.3909e-03) (hash(x)=37722489)
+2780 train 3.919179 (lr=1.3795e-03) (hash(x)=47290688)
+2790 train 3.955277 (lr=1.3682e-03) (hash(x)=41205574)
+2800 val loss 4.0150
+2800 val perplexity 55.4244
+2800 train 3.806384 (lr=1.3569e-03) (hash(x)=40257962)
+2810 train 3.780862 (lr=1.3456e-03) (hash(x)=39529014)
+2820 train 3.805137 (lr=1.3343e-03) (hash(x)=42244749)
+2830 train 3.854578 (lr=1.3230e-03) (hash(x)=32820090)
+2840 train 3.811341 (lr=1.3118e-03) (hash(x)=40315769)
+2850 train 4.052773 (lr=1.3006e-03) (hash(x)=41524462)
+2860 train 4.046352 (lr=1.2894e-03) (hash(x)=38365734)
+2870 train 4.172831 (lr=1.2783e-03) (hash(x)=37682602)
+2880 train 4.029194 (lr=1.2671e-03) (hash(x)=39162991)
+2890 train 3.890569 (lr=1.2560e-03) (hash(x)=33316384)
+2900 val loss 3.9920
+2900 val perplexity 54.1638
+2900 train 3.929995 (lr=1.2450e-03) (hash(x)=37271132)
+2910 train 4.124682 (lr=1.2339e-03) (hash(x)=35586242)
+2920 train 4.016730 (lr=1.2229e-03) (hash(x)=33320586)
+2930 train 4.032073 (lr=1.2119e-03) (hash(x)=43531361)
+2940 train 3.918909 (lr=1.2010e-03) (hash(x)=37368286)
+2950 train 3.951820 (lr=1.1901e-03) (hash(x)=40363394)
+2960 train 3.917309 (lr=1.1792e-03) (hash(x)=42253792)
+2970 train 3.858671 (lr=1.1684e-03) (hash(x)=38072598)
+2980 train 3.936077 (lr=1.1576e-03) (hash(x)=41470557)
+2990 train 3.858855 (lr=1.1468e-03) (hash(x)=42600033)
+3000 val loss 3.9905
+3000 val perplexity 54.0802
+3000 train 3.811138 (lr=1.1361e-03) (hash(x)=46890983)
+3010 train 3.684202 (lr=1.1254e-03) (hash(x)=37986759)
+3020 train 3.804304 (lr=1.1147e-03) (hash(x)=36270703)
+3030 train 4.074304 (lr=1.1041e-03) (hash(x)=38228599)
+3040 train 3.918625 (lr=1.0935e-03) (hash(x)=44344296)
+3050 train 4.007771 (lr=1.0830e-03) (hash(x)=39962297)
+3060 train 3.994660 (lr=1.0725e-03) (hash(x)=39817394)
+3070 train 3.983816 (lr=1.0620e-03) (hash(x)=41763868)
+3080 train 4.067705 (lr=1.0516e-03) (hash(x)=42343051)
+3090 train 4.065829 (lr=1.0412e-03) (hash(x)=39204893)
+3100 val loss 3.9705
+3100 val perplexity 53.0124
+3100 train 3.968384 (lr=1.0309e-03) (hash(x)=39353599)
+3110 train 4.037310 (lr=1.0207e-03) (hash(x)=36118073)
+3120 train 3.742609 (lr=1.0104e-03) (hash(x)=44070271)
+3130 train 3.769604 (lr=1.0002e-03) (hash(x)=43368151)
+3140 train 3.843125 (lr=9.9011e-04) (hash(x)=40612203)
+3150 train 3.980561 (lr=9.8002e-04) (hash(x)=39598447)
+3160 train 4.031323 (lr=9.6999e-04) (hash(x)=43897682)
+3170 train 4.011442 (lr=9.6000e-04) (hash(x)=39029243)
+3180 train 4.022347 (lr=9.5007e-04) (hash(x)=38586684)
+3190 train 3.971690 (lr=9.4019e-04) (hash(x)=39299115)
+3200 val loss 3.9507
+3200 val perplexity 51.9721
+3200 train 3.996680 (lr=9.3036e-04) (hash(x)=41064949)
+3210 train 3.991051 (lr=9.2058e-04) (hash(x)=41752822)
+3220 train 3.994093 (lr=9.1085e-04) (hash(x)=37484805)
+3230 train 3.960010 (lr=9.0118e-04) (hash(x)=40514919)
+3240 train 4.009206 (lr=8.9157e-04) (hash(x)=42440343)
+3250 train 3.913702 (lr=8.8201e-04) (hash(x)=32464136)
+3260 train 3.928639 (lr=8.7251e-04) (hash(x)=47270558)
+3270 train 3.869629 (lr=8.6307e-04) (hash(x)=39307546)
+3280 train 3.907166 (lr=8.5368e-04) (hash(x)=40410299)
+3290 train 3.904580 (lr=8.4435e-04) (hash(x)=40541497)
+3300 val loss 3.9460
+3300 val perplexity 51.7279
+3300 train 3.806420 (lr=8.3508e-04) (hash(x)=38089139)
+3310 train 3.835239 (lr=8.2588e-04) (hash(x)=42006576)
+3320 train 3.930433 (lr=8.1673e-04) (hash(x)=40422556)
+3330 train 3.717797 (lr=8.0764e-04) (hash(x)=39021643)
+3340 train 3.984932 (lr=7.9862e-04) (hash(x)=40434305)
+3350 train 4.014652 (lr=7.8966e-04) (hash(x)=40833559)
+3360 train 4.006145 (lr=7.8076e-04) (hash(x)=40045231)
+3370 train 3.972114 (lr=7.7192e-04) (hash(x)=34668317)
+3380 train 3.964308 (lr=7.6315e-04) (hash(x)=42738568)
+3390 train 4.061911 (lr=7.5445e-04) (hash(x)=38980166)
+3400 val loss 3.9248
+3400 val perplexity 50.6451
+3400 train 3.972593 (lr=7.4581e-04) (hash(x)=43878176)
+3410 train 4.016264 (lr=7.3724e-04) (hash(x)=38201991)
+3420 train 3.994574 (lr=7.2874e-04) (hash(x)=34905889)
+3430 train 3.990573 (lr=7.2030e-04) (hash(x)=41680405)
+3440 train 3.954377 (lr=7.1193e-04) (hash(x)=39948184)
+3450 train 4.003587 (lr=7.0363e-04) (hash(x)=33623728)
+3460 train 3.874452 (lr=6.9541e-04) (hash(x)=41027647)
+3470 train 3.906511 (lr=6.8725e-04) (hash(x)=42779482)
+3480 train 3.927838 (lr=6.7916e-04) (hash(x)=41579288)
+3490 train 3.786224 (lr=6.7114e-04) (hash(x)=37174846)
+3500 val loss 3.9222
+3500 val perplexity 50.5113
+3500 train 3.660249 (lr=6.6320e-04) (hash(x)=37082090)
+3510 train 3.854492 (lr=6.5533e-04) (hash(x)=43131340)
+3520 train 3.769487 (lr=6.4753e-04) (hash(x)=42457250)
+3530 train 4.002453 (lr=6.3981e-04) (hash(x)=39504492)
+3540 train 3.945202 (lr=6.3216e-04) (hash(x)=36225997)
+3550 train 4.071271 (lr=6.2458e-04) (hash(x)=40912087)
+3560 train 3.914048 (lr=6.1708e-04) (hash(x)=41545925)
+3570 train 4.005684 (lr=6.0966e-04) (hash(x)=32553193)
+3580 train 3.846700 (lr=6.0231e-04) (hash(x)=38799856)
+3590 train 3.894660 (lr=5.9504e-04) (hash(x)=40772075)
+3600 val loss 3.9033
+3600 val perplexity 49.5664
+3600 train 3.936077 (lr=5.8785e-04) (hash(x)=39299903)
+3610 train 3.980863 (lr=5.8074e-04) (hash(x)=39503247)
+3620 train 3.876992 (lr=5.7370e-04) (hash(x)=38730668)
+3630 train 3.895566 (lr=5.6675e-04) (hash(x)=40503276)
+3640 train 3.904370 (lr=5.5987e-04) (hash(x)=36504573)
+3650 train 3.924896 (lr=5.5308e-04) (hash(x)=44903075)
+3660 train 3.835786 (lr=5.4636e-04) (hash(x)=33800148)
+3670 train 3.745420 (lr=5.3973e-04) (hash(x)=39404541)
+3680 train 3.686204 (lr=5.3318e-04) (hash(x)=44777432)
+3690 train 3.784354 (lr=5.2671e-04) (hash(x)=45649913)
+3700 val loss 3.9054
+3700 val perplexity 49.6675
+3700 train 3.909715 (lr=5.2033e-04) (hash(x)=33812537)
+3710 train 4.061293 (lr=5.1402e-04) (hash(x)=36161222)
+3720 train 3.936011 (lr=5.0780e-04) (hash(x)=43690616)
+3730 train 3.941621 (lr=5.0167e-04) (hash(x)=35462313)
+3740 train 4.239367 (lr=4.9562e-04) (hash(x)=44674254)
+3750 train 3.916725 (lr=4.8965e-04) (hash(x)=44276297)
+3760 train 3.963215 (lr=4.8377e-04) (hash(x)=38492178)
+3770 train 3.967733 (lr=4.7798e-04) (hash(x)=44347531)
+3780 train 3.887998 (lr=4.7227e-04) (hash(x)=33792059)
+3790 train 3.976275 (lr=4.6665e-04) (hash(x)=38549318)
+3800 val loss 3.8864
+3800 val perplexity 48.7371
+3800 train 3.860205 (lr=4.6112e-04) (hash(x)=41437448)
+3810 train 3.759775 (lr=4.5567e-04) (hash(x)=36656768)
+3820 train 3.923028 (lr=4.5031e-04) (hash(x)=46387893)
+3830 train 3.921780 (lr=4.4504e-04) (hash(x)=44312129)
+3840 train 3.838115 (lr=4.3986e-04) (hash(x)=40341896)
+3850 train 3.887269 (lr=4.3477e-04) (hash(x)=43589160)
+3860 train 3.877450 (lr=4.2977e-04) (hash(x)=40388945)
+3870 train 3.773675 (lr=4.2486e-04) (hash(x)=33336679)
+3880 train 3.869395 (lr=4.2004e-04) (hash(x)=37643311)
+3890 train 3.892237 (lr=4.1530e-04) (hash(x)=38272927)
+3900 val loss 3.8806
+3900 val perplexity 48.4538
+3900 train 3.834027 (lr=4.1066e-04) (hash(x)=41925748)
+3910 train 3.930668 (lr=4.0611e-04) (hash(x)=38377253)
+3920 train 3.931430 (lr=4.0166e-04) (hash(x)=41616611)
+3930 train 3.855855 (lr=3.9729e-04) (hash(x)=32301827)
+3940 train 3.906796 (lr=3.9302e-04) (hash(x)=47697363)
+3950 train 3.862825 (lr=3.8884e-04) (hash(x)=37867767)
+3960 train 3.861952 (lr=3.8475e-04) (hash(x)=37187295)
+3970 train 3.919327 (lr=3.8076e-04) (hash(x)=41952752)
+3980 train 3.887866 (lr=3.7685e-04) (hash(x)=38358660)
+3990 train 3.756404 (lr=3.7305e-04) (hash(x)=40207878)
+4000 val loss 3.8738
+4000 val perplexity 48.1273
+4000 train 3.845603 (lr=3.6933e-04) (hash(x)=39134015)
+4010 train 3.911463 (lr=3.6572e-04) (hash(x)=38313135)
+4020 train 3.841661 (lr=3.6219e-04) (hash(x)=40710513)
+4030 train 3.973002 (lr=3.5876e-04) (hash(x)=35530165)
+4040 train 3.895320 (lr=3.5543e-04) (hash(x)=39494066)
+4050 train 3.876886 (lr=3.5219e-04) (hash(x)=40861777)
+4060 train 3.862309 (lr=3.4905e-04) (hash(x)=36151939)
+4070 train 3.831459 (lr=3.4600e-04) (hash(x)=39008241)
+4080 train 3.917822 (lr=3.4305e-04) (hash(x)=40395746)
+4090 train 3.954761 (lr=3.4019e-04) (hash(x)=38855480)
+4100 val loss 3.8637
+4100 val perplexity 47.6393
+4100 train 3.893190 (lr=3.3744e-04) (hash(x)=45013254)
+4110 train 3.894420 (lr=3.3477e-04) (hash(x)=39934818)
+4120 train 3.900774 (lr=3.3221e-04) (hash(x)=33417705)
+4130 train 3.835951 (lr=3.2974e-04) (hash(x)=41267978)
+4140 train 3.829921 (lr=3.2737e-04) (hash(x)=38961609)
+4150 train 3.830842 (lr=3.2510e-04) (hash(x)=42173878)
+4160 train 3.795930 (lr=3.2292e-04) (hash(x)=36867405)
+4170 train 3.778615 (lr=3.2085e-04) (hash(x)=38991954)
+4180 train 3.838713 (lr=3.1887e-04) (hash(x)=39032592)
+4190 train 3.813030 (lr=3.1699e-04) (hash(x)=61329402)
+4200 val loss 3.8631
+4200 val perplexity 47.6149
+4200 train 3.825022 (lr=3.1520e-04) (hash(x)=40455200)
+4210 train 3.856199 (lr=3.1352e-04) (hash(x)=40913417)
+4220 train 3.888640 (lr=3.1193e-04) (hash(x)=38809704)
+4230 train 3.870013 (lr=3.1044e-04) (hash(x)=39560204)
+4240 train 3.874420 (lr=3.0905e-04) (hash(x)=41496549)
+4250 train 3.908419 (lr=3.0776e-04) (hash(x)=33994768)
+4260 train 3.899756 (lr=3.0657e-04) (hash(x)=39658687)
+4270 train 3.776677 (lr=3.0548e-04) (hash(x)=42583064)
+4280 train 3.840536 (lr=3.0449e-04) (hash(x)=40635811)
+4290 train 3.743476 (lr=3.0359e-04) (hash(x)=35313841)
+4300 val loss 3.8598
+4300 val perplexity 47.4547
+4300 train 3.923619 (lr=3.0280e-04) (hash(x)=39332245)
+4310 train 3.821622 (lr=3.0210e-04) (hash(x)=43732999)
+4320 train 3.824407 (lr=3.0150e-04) (hash(x)=40831835)
+4330 train 3.961749 (lr=3.0101e-04) (hash(x)=41813368)
+4340 train 3.863665 (lr=3.0061e-04) (hash(x)=36952586)
+4350 train 3.886323 (lr=3.0031e-04) (hash(x)=42470745)
+4360 train 3.813372 (lr=3.0011e-04) (hash(x)=39581495)
+4370 train 4.003078 (lr=3.0001e-04) (hash(x)=35914538)
+4374 val loss 3.8537
+4374 val perplexity 47.1668

att_conv_playground/f_learnable_compile_yes_weight_decay_seed_1339/model_04374.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f01fa9643d4705d0b10e9500c591c2fe0c1a2e3f5c2c036302e18233b90df80
+size 97744586

att_conv_playground/f_learnable_compile_yes_weight_decay_seed_1339/optimizer_04374.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0226464b179296da68daf53e7fc0f34203af303032b51ac084e7065683f485c9
+size 189212110