logs/training_history.jsonl

{"epoch": 0, "step": 149, "train_loss": 0.6638100527276929, "train_acc": 0.6761613884228188, "val_loss": 0.5969663418601541, "lr": 0.000149, "elapsed_s": 115.9399836063385}
{"epoch": 1, "step": 298, "train_loss": 0.5033537327443194, "train_acc": 0.8519950713087249, "val_loss": 0.4731155880339005, "lr": 0.000298, "elapsed_s": 232.38246202468872}
{"epoch": 2, "step": 447, "train_loss": 0.3110146491519557, "train_acc": 0.9030515939597316, "val_loss": 0.24350418546620536, "lr": 0.000447, "elapsed_s": 338.19464588165283}
{"epoch": 3, "step": 596, "train_loss": 0.21790639206067028, "train_acc": 0.9355075503355704, "val_loss": 0.21089388940614812, "lr": 0.0004992789550548746, "elapsed_s": 456.00061893463135}
{"epoch": 4, "step": 745, "train_loss": 0.17346532652842117, "train_acc": 0.9515126887583892, "val_loss": 0.1537043783454334, "lr": 0.0004953161807531013, "elapsed_s": 567.245138168335}
{"epoch": 5, "step": 894, "train_loss": 0.14274583941938093, "train_acc": 0.9600330327181208, "val_loss": 0.1544679962747237, "lr": 0.0004879468604440265, "elapsed_s": 676.9563376903534}
{"epoch": 6, "step": 1043, "train_loss": 0.11750165493096282, "train_acc": 0.9667444421140939, "val_loss": 0.22213954658368054, "lr": 0.00047727332707853153, "elapsed_s": 794.4324436187744}
{"epoch": 7, "step": 1192, "train_loss": 0.10584127082920715, "train_acc": 0.9699690645973155, "val_loss": 0.10747551086720299, "lr": 0.0004634437970619037, "elapsed_s": 904.3242561817169}
{"epoch": 8, "step": 1341, "train_loss": 0.08833698973359677, "train_acc": 0.9760643875838926, "val_loss": 0.15969384068250655, "lr": 0.0004466503120691616, "elapsed_s": 1015.3294868469238}
{"epoch": 9, "step": 1490, "train_loss": 0.08328283040615536, "train_acc": 0.9758546560402684, "val_loss": 0.07293711967678632, "lr": 0.000427126072287407, "elapsed_s": 1125.4256429672241}
{"epoch": 10, "step": 1639, "train_loss": 0.08347918183211513, "train_acc": 0.9757497902684564, "val_loss": 0.08363149303723784, "lr": 0.000405142198116729, "elapsed_s": 1246.7471852302551}
{"epoch": 11, "step": 1788, "train_loss": 0.07746089749888285, "train_acc": 0.9765100671140939, "val_loss": 0.1364490651032504, "lr": 0.00038100396529781167, "elapsed_s": 1368.8041446208954}
{"epoch": 12, "step": 1937, "train_loss": 0.05847198456125771, "train_acc": 0.9843618917785235, "val_loss": 0.544288030091454, "lr": 0.00035504656574657386, "elapsed_s": 1478.831687927246}
{"epoch": 13, "step": 2086, "train_loss": 0.049768463914726405, "train_acc": 0.9862888003355704, "val_loss": 0.11688621591119205, "lr": 0.0003276304529623721, "elapsed_s": 1596.107327222824}
{"epoch": 14, "step": 2235, "train_loss": 0.045146725146762476, "train_acc": 0.9874292156040269, "val_loss": 0.04456268590513398, "lr": 0.00029913633664504786, "elapsed_s": 1713.1900265216827}
{"epoch": 15, "step": 2384, "train_loss": 0.04567094507233409, "train_acc": 0.9863936661073825, "val_loss": 0.19601621710552888, "lr": 0.00026995989602731366, "elapsed_s": 1826.0041506290436}
{"epoch": 16, "step": 2533, "train_loss": 0.03973913225251556, "train_acc": 0.9889890939597316, "val_loss": 0.07772592426573528, "lr": 0.00024050628533498722, "elapsed_s": 1946.2154955863953}
{"epoch": 17, "step": 2682, "train_loss": 0.03955788589113911, "train_acc": 0.9894609899328859, "val_loss": 0.0671277295228313, "lr": 0.0002111845076741685, "elapsed_s": 2057.408088207245}
{"epoch": 18, "step": 2831, "train_loss": 0.03560613733310027, "train_acc": 0.990155725671141, "val_loss": 0.08580374415832408, "lr": 0.00018240173547152234, "elapsed_s": 2170.578821182251}
{"epoch": 19, "step": 2980, "train_loss": 0.030904719381024372, "train_acc": 0.9912961409395973, "val_loss": 0.05806329353679629, "lr": 0.00015455765633601548, "elapsed_s": 2283.02233171463}
{"epoch": 20, "step": 3129, "train_loss": 0.030267982618670735, "train_acc": 0.991440331375839, "val_loss": 0.04958753500791157, "lr": 0.0001280389228574397, "elapsed_s": 2406.581520795822}
{"epoch": 21, "step": 3278, "train_loss": 0.030025769825149703, "train_acc": 0.9917680369127517, "val_loss": 0.039248012774130876, "lr": 0.0001032137834137569, "elapsed_s": 2516.426001548767}
{"epoch": 22, "step": 3427, "train_loss": 0.027258547479224286, "train_acc": 0.9927118288590604, "val_loss": 0.03384857120890828, "lr": 8.042696854574489e-05, "elapsed_s": 2625.873127222061}
{"epoch": 23, "step": 3576, "train_loss": 0.026748527028916665, "train_acc": 0.9927904781879194, "val_loss": 0.0323622088590089, "lr": 5.999490390852946e-05, "elapsed_s": 2747.82723069191}
{"epoch": 24, "step": 3725, "train_loss": 0.02556787448116397, "train_acc": 0.9930264261744967, "val_loss": 0.02685919334432658, "lr": 4.2201316274625865e-05, "elapsed_s": 2860.2909936904907}
{"epoch": 25, "step": 3874, "train_loss": 0.025232987178352054, "train_acc": 0.9930264261744967, "val_loss": 0.027291072836255325, "lr": 2.7293293605063636e-05, "elapsed_s": 2970.7279999256134}
{"epoch": 26, "step": 4023, "train_loss": 0.024634346590946186, "train_acc": 0.9936556208053692, "val_loss": 0.0254005408076679, "lr": 1.547785389982076e-05, "elapsed_s": 3091.010478258133}
{"epoch": 27, "step": 4172, "train_loss": 0.024240581637510916, "train_acc": 0.993419672818792, "val_loss": 0.026522268282140004, "lr": 6.919070473706129e-06, "elapsed_s": 3199.4680149555206}
{"epoch": 28, "step": 4321, "train_loss": 0.023403970587173563, "train_acc": 0.993419672818792, "val_loss": 0.025312599884674828, "lr": 1.7357935771112333e-06, "elapsed_s": 3309.5373463630676}
{"epoch": 29, "step": 4470, "train_loss": 0.02398738367545525, "train_acc": 0.9935900796979866, "val_loss": 0.025535204175202286, "lr": 1e-07, "elapsed_s": 3420.383076429367}