| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 471, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0021231422505307855, | |
| "grad_norm": 6.233692311689662, | |
| "learning_rate": 0.0, | |
| "loss": 1.3677, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004246284501061571, | |
| "grad_norm": 6.03613913611666, | |
| "learning_rate": 2.0833333333333333e-07, | |
| "loss": 1.4092, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.006369426751592357, | |
| "grad_norm": 5.843878249582451, | |
| "learning_rate": 4.1666666666666667e-07, | |
| "loss": 1.4005, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.008492569002123142, | |
| "grad_norm": 6.2100854445399305, | |
| "learning_rate": 6.25e-07, | |
| "loss": 1.4027, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.010615711252653927, | |
| "grad_norm": 5.700197750457959, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 1.3859, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.012738853503184714, | |
| "grad_norm": 5.708735743112818, | |
| "learning_rate": 1.0416666666666667e-06, | |
| "loss": 1.4135, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.014861995753715499, | |
| "grad_norm": 5.584947016091844, | |
| "learning_rate": 1.25e-06, | |
| "loss": 1.3083, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.016985138004246284, | |
| "grad_norm": 5.5017514099891125, | |
| "learning_rate": 1.4583333333333335e-06, | |
| "loss": 1.3761, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.01910828025477707, | |
| "grad_norm": 5.240707118749176, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 1.374, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.021231422505307854, | |
| "grad_norm": 5.159923266283754, | |
| "learning_rate": 1.8750000000000003e-06, | |
| "loss": 1.3837, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02335456475583864, | |
| "grad_norm": 4.23439034643152, | |
| "learning_rate": 2.0833333333333334e-06, | |
| "loss": 1.3278, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.025477707006369428, | |
| "grad_norm": 4.23403012426509, | |
| "learning_rate": 2.2916666666666666e-06, | |
| "loss": 1.3095, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.027600849256900213, | |
| "grad_norm": 4.033469595815775, | |
| "learning_rate": 2.5e-06, | |
| "loss": 1.3736, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.029723991507430998, | |
| "grad_norm": 3.4140671410294092, | |
| "learning_rate": 2.7083333333333334e-06, | |
| "loss": 1.2966, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.03184713375796178, | |
| "grad_norm": 3.2252832240083946, | |
| "learning_rate": 2.916666666666667e-06, | |
| "loss": 1.3211, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.03397027600849257, | |
| "grad_norm": 2.958759736658132, | |
| "learning_rate": 3.125e-06, | |
| "loss": 1.284, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.036093418259023353, | |
| "grad_norm": 3.0082219342870764, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 1.2574, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.03821656050955414, | |
| "grad_norm": 2.7607212501829186, | |
| "learning_rate": 3.5416666666666673e-06, | |
| "loss": 1.2222, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.040339702760084924, | |
| "grad_norm": 1.7794382970826705, | |
| "learning_rate": 3.7500000000000005e-06, | |
| "loss": 1.1952, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.04246284501061571, | |
| "grad_norm": 1.9000490101749923, | |
| "learning_rate": 3.958333333333333e-06, | |
| "loss": 1.0968, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.044585987261146494, | |
| "grad_norm": 2.184749467485857, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 1.2058, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.04670912951167728, | |
| "grad_norm": 2.076583497979509, | |
| "learning_rate": 4.3750000000000005e-06, | |
| "loss": 1.1726, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.04883227176220807, | |
| "grad_norm": 1.6762408968746727, | |
| "learning_rate": 4.583333333333333e-06, | |
| "loss": 1.0816, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.050955414012738856, | |
| "grad_norm": 1.655763485736625, | |
| "learning_rate": 4.791666666666668e-06, | |
| "loss": 1.0881, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.05307855626326964, | |
| "grad_norm": 1.404756822747019, | |
| "learning_rate": 5e-06, | |
| "loss": 1.111, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.055201698513800426, | |
| "grad_norm": 1.7450925608231072, | |
| "learning_rate": 4.999938256261134e-06, | |
| "loss": 1.0968, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05732484076433121, | |
| "grad_norm": 1.7455723981125268, | |
| "learning_rate": 4.999753028094368e-06, | |
| "loss": 1.135, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.059447983014861996, | |
| "grad_norm": 1.765180118972071, | |
| "learning_rate": 4.999444324649045e-06, | |
| "loss": 1.1332, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.06157112526539278, | |
| "grad_norm": 1.7025986108773807, | |
| "learning_rate": 4.99901216117357e-06, | |
| "loss": 1.1037, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.06369426751592357, | |
| "grad_norm": 1.6843781660517996, | |
| "learning_rate": 4.998456559014653e-06, | |
| "loss": 1.1036, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06581740976645435, | |
| "grad_norm": 1.4971568544476521, | |
| "learning_rate": 4.997777545616258e-06, | |
| "loss": 1.0896, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.06794055201698514, | |
| "grad_norm": 1.2471663073080574, | |
| "learning_rate": 4.996975154518245e-06, | |
| "loss": 1.0816, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.07006369426751592, | |
| "grad_norm": 1.1099200791707107, | |
| "learning_rate": 4.996049425354717e-06, | |
| "loss": 1.0408, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.07218683651804671, | |
| "grad_norm": 1.1596577803643031, | |
| "learning_rate": 4.995000403852057e-06, | |
| "loss": 1.0348, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.07430997876857749, | |
| "grad_norm": 0.9836703479205239, | |
| "learning_rate": 4.993828141826672e-06, | |
| "loss": 1.0397, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.07643312101910828, | |
| "grad_norm": 4.973840678082251, | |
| "learning_rate": 4.992532697182434e-06, | |
| "loss": 1.0692, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.07855626326963906, | |
| "grad_norm": 0.9322137357537782, | |
| "learning_rate": 4.991114133907822e-06, | |
| "loss": 1.0056, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.08067940552016985, | |
| "grad_norm": 1.0203537152829516, | |
| "learning_rate": 4.989572522072753e-06, | |
| "loss": 0.9956, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.08280254777070063, | |
| "grad_norm": 1.0385035368583873, | |
| "learning_rate": 4.9879079378251325e-06, | |
| "loss": 0.9772, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.08492569002123142, | |
| "grad_norm": 0.9097103872410344, | |
| "learning_rate": 4.986120463387084e-06, | |
| "loss": 1.0256, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0870488322717622, | |
| "grad_norm": 1.0611463152500384, | |
| "learning_rate": 4.984210187050891e-06, | |
| "loss": 0.9883, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.08917197452229299, | |
| "grad_norm": 0.779249244150645, | |
| "learning_rate": 4.982177203174636e-06, | |
| "loss": 0.9847, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.09129511677282377, | |
| "grad_norm": 0.7658008659883724, | |
| "learning_rate": 4.9800216121775404e-06, | |
| "loss": 0.9829, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.09341825902335456, | |
| "grad_norm": 0.7598772913633557, | |
| "learning_rate": 4.977743520535001e-06, | |
| "loss": 1.0496, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.09554140127388536, | |
| "grad_norm": 0.8141928977386432, | |
| "learning_rate": 4.975343040773335e-06, | |
| "loss": 1.0519, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.09766454352441614, | |
| "grad_norm": 0.8911722172701033, | |
| "learning_rate": 4.972820291464219e-06, | |
| "loss": 1.0345, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.09978768577494693, | |
| "grad_norm": 0.8358128773624875, | |
| "learning_rate": 4.970175397218832e-06, | |
| "loss": 0.9987, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.10191082802547771, | |
| "grad_norm": 0.882046205562171, | |
| "learning_rate": 4.967408488681702e-06, | |
| "loss": 1.0004, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.1040339702760085, | |
| "grad_norm": 0.8308643665449907, | |
| "learning_rate": 4.964519702524251e-06, | |
| "loss": 0.9867, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.10615711252653928, | |
| "grad_norm": 0.8727814847927396, | |
| "learning_rate": 4.9615091814380465e-06, | |
| "loss": 1.0236, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.10828025477707007, | |
| "grad_norm": 0.8128090079858757, | |
| "learning_rate": 4.958377074127751e-06, | |
| "loss": 1.0096, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.11040339702760085, | |
| "grad_norm": 0.7041201796080805, | |
| "learning_rate": 4.955123535303775e-06, | |
| "loss": 1.0161, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.11252653927813164, | |
| "grad_norm": 0.7034976999278052, | |
| "learning_rate": 4.951748725674643e-06, | |
| "loss": 0.9444, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.11464968152866242, | |
| "grad_norm": 0.7111287475587301, | |
| "learning_rate": 4.948252811939044e-06, | |
| "loss": 1.0034, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.11677282377919321, | |
| "grad_norm": 0.7112458705577068, | |
| "learning_rate": 4.944635966777607e-06, | |
| "loss": 1.0413, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.11889596602972399, | |
| "grad_norm": 0.7351439597774859, | |
| "learning_rate": 4.940898368844366e-06, | |
| "loss": 1.0663, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.12101910828025478, | |
| "grad_norm": 0.7120598412888705, | |
| "learning_rate": 4.937040202757937e-06, | |
| "loss": 1.0058, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.12314225053078556, | |
| "grad_norm": 0.8009131327695405, | |
| "learning_rate": 4.933061659092401e-06, | |
| "loss": 1.0048, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.12526539278131635, | |
| "grad_norm": 0.8355353984744548, | |
| "learning_rate": 4.928962934367887e-06, | |
| "loss": 1.0005, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.12738853503184713, | |
| "grad_norm": 0.7292500219199575, | |
| "learning_rate": 4.924744231040865e-06, | |
| "loss": 0.9948, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.12951167728237792, | |
| "grad_norm": 0.7441720822069011, | |
| "learning_rate": 4.920405757494147e-06, | |
| "loss": 0.9203, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.1316348195329087, | |
| "grad_norm": 0.7164757410132886, | |
| "learning_rate": 4.915947728026599e-06, | |
| "loss": 0.9826, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.1337579617834395, | |
| "grad_norm": 0.6688811212861865, | |
| "learning_rate": 4.911370362842544e-06, | |
| "loss": 1.0041, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.13588110403397027, | |
| "grad_norm": 0.7276933998413905, | |
| "learning_rate": 4.906673888040895e-06, | |
| "loss": 1.0254, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.13800424628450106, | |
| "grad_norm": 0.8980292377631762, | |
| "learning_rate": 4.901858535603982e-06, | |
| "loss": 1.0341, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.14012738853503184, | |
| "grad_norm": 0.7213457650159387, | |
| "learning_rate": 4.896924543386099e-06, | |
| "loss": 0.9732, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.14225053078556263, | |
| "grad_norm": 0.6575608860005288, | |
| "learning_rate": 4.891872155101746e-06, | |
| "loss": 0.9758, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.14437367303609341, | |
| "grad_norm": 0.7227626661588851, | |
| "learning_rate": 4.886701620313595e-06, | |
| "loss": 0.9736, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.1464968152866242, | |
| "grad_norm": 0.7820534177881011, | |
| "learning_rate": 4.88141319442017e-06, | |
| "loss": 0.9519, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.14861995753715498, | |
| "grad_norm": 0.7464763101124039, | |
| "learning_rate": 4.876007138643216e-06, | |
| "loss": 1.0623, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.15074309978768577, | |
| "grad_norm": 0.8031125528787094, | |
| "learning_rate": 4.870483720014814e-06, | |
| "loss": 1.0023, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.15286624203821655, | |
| "grad_norm": 0.7252756266115659, | |
| "learning_rate": 4.8648432113641765e-06, | |
| "loss": 1.025, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.15498938428874734, | |
| "grad_norm": 0.7685738411441779, | |
| "learning_rate": 4.8590858913041775e-06, | |
| "loss": 0.9701, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.15711252653927812, | |
| "grad_norm": 0.7119059019537681, | |
| "learning_rate": 4.853212044217591e-06, | |
| "loss": 0.9471, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.1592356687898089, | |
| "grad_norm": 0.670302973142834, | |
| "learning_rate": 4.847221960243042e-06, | |
| "loss": 0.9918, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1613588110403397, | |
| "grad_norm": 0.7274878719799441, | |
| "learning_rate": 4.8411159352606735e-06, | |
| "loss": 0.9114, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.16348195329087048, | |
| "grad_norm": 0.6706478978040786, | |
| "learning_rate": 4.834894270877536e-06, | |
| "loss": 0.9628, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.16560509554140126, | |
| "grad_norm": 0.7197495557043765, | |
| "learning_rate": 4.828557274412686e-06, | |
| "loss": 0.9792, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.16772823779193205, | |
| "grad_norm": 0.6740897649139851, | |
| "learning_rate": 4.822105258882007e-06, | |
| "loss": 0.978, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.16985138004246284, | |
| "grad_norm": 0.7298858435711825, | |
| "learning_rate": 4.815538542982751e-06, | |
| "loss": 1.0749, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.17197452229299362, | |
| "grad_norm": 0.7437812096942298, | |
| "learning_rate": 4.808857451077788e-06, | |
| "loss": 1.0027, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.1740976645435244, | |
| "grad_norm": 0.7793372891907893, | |
| "learning_rate": 4.802062313179595e-06, | |
| "loss": 1.056, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.1762208067940552, | |
| "grad_norm": 0.7200522013491976, | |
| "learning_rate": 4.795153464933948e-06, | |
| "loss": 0.9804, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.17834394904458598, | |
| "grad_norm": 0.6800845586517711, | |
| "learning_rate": 4.7881312476033444e-06, | |
| "loss": 1.0096, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.18046709129511676, | |
| "grad_norm": 0.6876028206701021, | |
| "learning_rate": 4.7809960080501464e-06, | |
| "loss": 0.9886, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.18259023354564755, | |
| "grad_norm": 0.7119685223963281, | |
| "learning_rate": 4.773748098719448e-06, | |
| "loss": 1.019, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.18471337579617833, | |
| "grad_norm": 0.6948951583741163, | |
| "learning_rate": 4.766387877621667e-06, | |
| "loss": 1.0177, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.18683651804670912, | |
| "grad_norm": 0.7252384478947084, | |
| "learning_rate": 4.758915708314858e-06, | |
| "loss": 0.952, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.18895966029723993, | |
| "grad_norm": 0.7132363877921689, | |
| "learning_rate": 4.751331959886758e-06, | |
| "loss": 0.9311, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.1910828025477707, | |
| "grad_norm": 0.6672381978657935, | |
| "learning_rate": 4.743637006936552e-06, | |
| "loss": 1.0019, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.1932059447983015, | |
| "grad_norm": 0.7074654491248366, | |
| "learning_rate": 4.735831229556374e-06, | |
| "loss": 0.9249, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.19532908704883228, | |
| "grad_norm": 0.8263034494745758, | |
| "learning_rate": 4.727915013312527e-06, | |
| "loss": 0.9627, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.19745222929936307, | |
| "grad_norm": 0.7564665006141713, | |
| "learning_rate": 4.719888749226442e-06, | |
| "loss": 0.9788, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.19957537154989385, | |
| "grad_norm": 0.7401712745437141, | |
| "learning_rate": 4.711752833755362e-06, | |
| "loss": 0.9595, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.20169851380042464, | |
| "grad_norm": 0.7177547311512174, | |
| "learning_rate": 4.70350766877276e-06, | |
| "loss": 1.0125, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.20382165605095542, | |
| "grad_norm": 0.8365049197183004, | |
| "learning_rate": 4.695153661548486e-06, | |
| "loss": 0.9617, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.2059447983014862, | |
| "grad_norm": 0.8275524843143772, | |
| "learning_rate": 4.686691224728652e-06, | |
| "loss": 0.9583, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.208067940552017, | |
| "grad_norm": 0.7849897218743559, | |
| "learning_rate": 4.678120776315251e-06, | |
| "loss": 0.9432, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.21019108280254778, | |
| "grad_norm": 0.7266338220692218, | |
| "learning_rate": 4.669442739645506e-06, | |
| "loss": 1.0112, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.21231422505307856, | |
| "grad_norm": 0.6786209555674868, | |
| "learning_rate": 4.660657543370958e-06, | |
| "loss": 1.0122, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.21443736730360935, | |
| "grad_norm": 0.7396143819036078, | |
| "learning_rate": 4.651765621436303e-06, | |
| "loss": 1.0491, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.21656050955414013, | |
| "grad_norm": 0.745708791916274, | |
| "learning_rate": 4.642767413057942e-06, | |
| "loss": 0.966, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.21868365180467092, | |
| "grad_norm": 0.7477607378275345, | |
| "learning_rate": 4.6336633627023e-06, | |
| "loss": 0.9129, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.2208067940552017, | |
| "grad_norm": 0.6930283267824098, | |
| "learning_rate": 4.624453920063863e-06, | |
| "loss": 0.9769, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.2229299363057325, | |
| "grad_norm": 0.6762283490594463, | |
| "learning_rate": 4.6151395400429665e-06, | |
| "loss": 0.9383, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.22505307855626328, | |
| "grad_norm": 0.7313509583041526, | |
| "learning_rate": 4.605720682723331e-06, | |
| "loss": 0.9428, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.22717622080679406, | |
| "grad_norm": 0.7156367191749474, | |
| "learning_rate": 4.596197813349328e-06, | |
| "loss": 0.9834, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.22929936305732485, | |
| "grad_norm": 0.6828464400232375, | |
| "learning_rate": 4.586571402303006e-06, | |
| "loss": 0.9504, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.23142250530785563, | |
| "grad_norm": 0.74716164026175, | |
| "learning_rate": 4.576841925080853e-06, | |
| "loss": 0.9998, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.23354564755838642, | |
| "grad_norm": 0.6856088912154495, | |
| "learning_rate": 4.56700986227031e-06, | |
| "loss": 1.0149, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2356687898089172, | |
| "grad_norm": 0.7033606189520333, | |
| "learning_rate": 4.557075699526032e-06, | |
| "loss": 0.9584, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.23779193205944799, | |
| "grad_norm": 0.7043436392355145, | |
| "learning_rate": 4.547039927545899e-06, | |
| "loss": 0.9779, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.23991507430997877, | |
| "grad_norm": 0.8202453116913221, | |
| "learning_rate": 4.536903042046778e-06, | |
| "loss": 0.9505, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.24203821656050956, | |
| "grad_norm": 0.7436505838525592, | |
| "learning_rate": 4.526665543740038e-06, | |
| "loss": 0.9561, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.24416135881104034, | |
| "grad_norm": 0.6478361281734111, | |
| "learning_rate": 4.516327938306818e-06, | |
| "loss": 1.0084, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.24628450106157113, | |
| "grad_norm": 0.7674555846640165, | |
| "learning_rate": 4.505890736373045e-06, | |
| "loss": 0.9767, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.2484076433121019, | |
| "grad_norm": 0.7184796384578326, | |
| "learning_rate": 4.495354453484216e-06, | |
| "loss": 0.9801, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.2505307855626327, | |
| "grad_norm": 0.677609956598202, | |
| "learning_rate": 4.4847196100799305e-06, | |
| "loss": 0.9825, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.2526539278131635, | |
| "grad_norm": 0.7123651122668029, | |
| "learning_rate": 4.473986731468183e-06, | |
| "loss": 1.0107, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.25477707006369427, | |
| "grad_norm": 0.7156231222559138, | |
| "learning_rate": 4.463156347799419e-06, | |
| "loss": 1.0779, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.25690021231422505, | |
| "grad_norm": 0.7717775840138449, | |
| "learning_rate": 4.452228994040341e-06, | |
| "loss": 0.9622, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.25902335456475584, | |
| "grad_norm": 0.7512072014948016, | |
| "learning_rate": 4.4412052099474916e-06, | |
| "loss": 0.9625, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.2611464968152866, | |
| "grad_norm": 0.7124270062225315, | |
| "learning_rate": 4.430085540040587e-06, | |
| "loss": 1.0169, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.2632696390658174, | |
| "grad_norm": 0.7058601891740411, | |
| "learning_rate": 4.418870533575626e-06, | |
| "loss": 0.9819, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.2653927813163482, | |
| "grad_norm": 0.6903185636770165, | |
| "learning_rate": 4.40756074451775e-06, | |
| "loss": 0.9398, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.267515923566879, | |
| "grad_norm": 0.7046682682406727, | |
| "learning_rate": 4.396156731513888e-06, | |
| "loss": 1.0341, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.26963906581740976, | |
| "grad_norm": 0.7862567299960644, | |
| "learning_rate": 4.384659057865165e-06, | |
| "loss": 0.947, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.27176220806794055, | |
| "grad_norm": 0.6195482427854749, | |
| "learning_rate": 4.373068291499065e-06, | |
| "loss": 0.8899, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.27388535031847133, | |
| "grad_norm": 0.6985398850512, | |
| "learning_rate": 4.36138500494139e-06, | |
| "loss": 0.9726, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.2760084925690021, | |
| "grad_norm": 0.668756553679393, | |
| "learning_rate": 4.349609775287977e-06, | |
| "loss": 0.9855, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2781316348195329, | |
| "grad_norm": 0.6581795928554366, | |
| "learning_rate": 4.337743184176188e-06, | |
| "loss": 0.9596, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.2802547770700637, | |
| "grad_norm": 0.6337202167292365, | |
| "learning_rate": 4.325785817756186e-06, | |
| "loss": 0.9839, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.2823779193205945, | |
| "grad_norm": 0.6680059124711757, | |
| "learning_rate": 4.313738266661979e-06, | |
| "loss": 0.9324, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.28450106157112526, | |
| "grad_norm": 0.6598271036519225, | |
| "learning_rate": 4.301601125982246e-06, | |
| "loss": 1.0066, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.28662420382165604, | |
| "grad_norm": 0.7183648150650218, | |
| "learning_rate": 4.289374995230942e-06, | |
| "loss": 1.0136, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.28874734607218683, | |
| "grad_norm": 0.742221341756798, | |
| "learning_rate": 4.277060478317687e-06, | |
| "loss": 1.0008, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.2908704883227176, | |
| "grad_norm": 0.6824304036868553, | |
| "learning_rate": 4.264658183517935e-06, | |
| "loss": 0.9381, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.2929936305732484, | |
| "grad_norm": 0.6505573173225403, | |
| "learning_rate": 4.252168723442927e-06, | |
| "loss": 0.9489, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.2951167728237792, | |
| "grad_norm": 0.6950634835305272, | |
| "learning_rate": 4.23959271500943e-06, | |
| "loss": 0.9373, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.29723991507430997, | |
| "grad_norm": 0.6872675526003413, | |
| "learning_rate": 4.226930779409271e-06, | |
| "loss": 0.9984, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.29936305732484075, | |
| "grad_norm": 0.7693317009374261, | |
| "learning_rate": 4.214183542078646e-06, | |
| "loss": 0.9391, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.30148619957537154, | |
| "grad_norm": 0.6809609221970698, | |
| "learning_rate": 4.201351632667227e-06, | |
| "loss": 1.0107, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.3036093418259023, | |
| "grad_norm": 0.6776804857260985, | |
| "learning_rate": 4.1884356850070695e-06, | |
| "loss": 1.0316, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.3057324840764331, | |
| "grad_norm": 0.6858474793512133, | |
| "learning_rate": 4.175436337081289e-06, | |
| "loss": 0.9071, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.3078556263269639, | |
| "grad_norm": 0.7262200113022307, | |
| "learning_rate": 4.162354230992562e-06, | |
| "loss": 0.9751, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.3099787685774947, | |
| "grad_norm": 0.7048851306107318, | |
| "learning_rate": 4.149190012931402e-06, | |
| "loss": 0.9222, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.31210191082802546, | |
| "grad_norm": 0.7661302534267068, | |
| "learning_rate": 4.135944333144244e-06, | |
| "loss": 0.9493, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.31422505307855625, | |
| "grad_norm": 0.6647279751401796, | |
| "learning_rate": 4.122617845901322e-06, | |
| "loss": 1.0373, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.31634819532908703, | |
| "grad_norm": 0.677884233878186, | |
| "learning_rate": 4.109211209464354e-06, | |
| "loss": 1.0299, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.3184713375796178, | |
| "grad_norm": 0.6808985503614562, | |
| "learning_rate": 4.095725086054029e-06, | |
| "loss": 0.9564, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3205944798301486, | |
| "grad_norm": 0.6844572260666089, | |
| "learning_rate": 4.0821601418172926e-06, | |
| "loss": 0.9942, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.3227176220806794, | |
| "grad_norm": 0.687636477644006, | |
| "learning_rate": 4.068517046794443e-06, | |
| "loss": 0.9925, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.3248407643312102, | |
| "grad_norm": 0.7121462034889847, | |
| "learning_rate": 4.054796474886038e-06, | |
| "loss": 1.0232, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.32696390658174096, | |
| "grad_norm": 0.6791139957154484, | |
| "learning_rate": 4.040999103819606e-06, | |
| "loss": 1.0147, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.32908704883227174, | |
| "grad_norm": 0.6317316423507299, | |
| "learning_rate": 4.0271256151161666e-06, | |
| "loss": 1.0073, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.33121019108280253, | |
| "grad_norm": 0.7021672955838567, | |
| "learning_rate": 4.013176694056571e-06, | |
| "loss": 0.9892, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.6535581508427883, | |
| "learning_rate": 3.999153029647651e-06, | |
| "loss": 1.0406, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.3354564755838641, | |
| "grad_norm": 0.6343047828611024, | |
| "learning_rate": 3.985055314588185e-06, | |
| "loss": 0.9452, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.3375796178343949, | |
| "grad_norm": 0.7092653110810674, | |
| "learning_rate": 3.970884245234684e-06, | |
| "loss": 1.0262, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.33970276008492567, | |
| "grad_norm": 0.7474243693562463, | |
| "learning_rate": 3.956640521566989e-06, | |
| "loss": 0.9584, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.34182590233545646, | |
| "grad_norm": 0.6647251802270744, | |
| "learning_rate": 3.942324847153706e-06, | |
| "loss": 0.9247, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.34394904458598724, | |
| "grad_norm": 0.6878800859406885, | |
| "learning_rate": 3.927937929117447e-06, | |
| "loss": 0.9674, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.346072186836518, | |
| "grad_norm": 0.7194337473522074, | |
| "learning_rate": 3.913480478099898e-06, | |
| "loss": 0.9727, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.3481953290870488, | |
| "grad_norm": 0.7643905012788286, | |
| "learning_rate": 3.898953208226723e-06, | |
| "loss": 1.0143, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.3503184713375796, | |
| "grad_norm": 0.6615757470184159, | |
| "learning_rate": 3.884356837072289e-06, | |
| "loss": 1.0199, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.3524416135881104, | |
| "grad_norm": 0.6594351684335442, | |
| "learning_rate": 3.869692085624218e-06, | |
| "loss": 0.929, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.35456475583864117, | |
| "grad_norm": 0.7235028424658448, | |
| "learning_rate": 3.8549596782477775e-06, | |
| "loss": 0.9624, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.35668789808917195, | |
| "grad_norm": 0.6204087859526352, | |
| "learning_rate": 3.8401603426501e-06, | |
| "loss": 1.0118, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.35881104033970274, | |
| "grad_norm": 0.6873006374808731, | |
| "learning_rate": 3.825294809844234e-06, | |
| "loss": 0.9837, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.3609341825902335, | |
| "grad_norm": 0.7107317410899462, | |
| "learning_rate": 3.8103638141130406e-06, | |
| "loss": 0.9185, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.3630573248407643, | |
| "grad_norm": 0.6549533824581767, | |
| "learning_rate": 3.7953680929729215e-06, | |
| "loss": 0.9547, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.3651804670912951, | |
| "grad_norm": 0.6914578323464019, | |
| "learning_rate": 3.7803083871373876e-06, | |
| "loss": 0.9434, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.3673036093418259, | |
| "grad_norm": 0.6752976153646614, | |
| "learning_rate": 3.7651854404804757e-06, | |
| "loss": 0.9398, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.36942675159235666, | |
| "grad_norm": 0.6817081041020392, | |
| "learning_rate": 3.7500000000000005e-06, | |
| "loss": 0.981, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.37154989384288745, | |
| "grad_norm": 0.6695314283276965, | |
| "learning_rate": 3.734752815780659e-06, | |
| "loss": 0.9526, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.37367303609341823, | |
| "grad_norm": 0.6927497018483542, | |
| "learning_rate": 3.7194446409569813e-06, | |
| "loss": 0.9904, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.37579617834394907, | |
| "grad_norm": 0.736727316466936, | |
| "learning_rate": 3.7040762316761247e-06, | |
| "loss": 0.9487, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.37791932059447986, | |
| "grad_norm": 0.7074800907113514, | |
| "learning_rate": 3.6886483470605293e-06, | |
| "loss": 0.9896, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.38004246284501064, | |
| "grad_norm": 0.7739492910066014, | |
| "learning_rate": 3.6731617491704195e-06, | |
| "loss": 0.929, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.3821656050955414, | |
| "grad_norm": 0.7085501959751563, | |
| "learning_rate": 3.6576172029661583e-06, | |
| "loss": 1.0187, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3842887473460722, | |
| "grad_norm": 0.6968943221383892, | |
| "learning_rate": 3.6420154762704685e-06, | |
| "loss": 0.9873, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.386411889596603, | |
| "grad_norm": 0.6799387175990909, | |
| "learning_rate": 3.6263573397305003e-06, | |
| "loss": 1.0038, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.3885350318471338, | |
| "grad_norm": 0.7111666054095594, | |
| "learning_rate": 3.6106435667797684e-06, | |
| "loss": 0.9405, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.39065817409766457, | |
| "grad_norm": 0.6690528674663481, | |
| "learning_rate": 3.5948749335999493e-06, | |
| "loss": 0.9196, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.39278131634819535, | |
| "grad_norm": 0.7229293413643434, | |
| "learning_rate": 3.5790522190825368e-06, | |
| "loss": 0.9907, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.39490445859872614, | |
| "grad_norm": 0.7327537558494182, | |
| "learning_rate": 3.5631762047903746e-06, | |
| "loss": 1.0238, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.3970276008492569, | |
| "grad_norm": 0.6338409213121795, | |
| "learning_rate": 3.5472476749190465e-06, | |
| "loss": 1.0138, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.3991507430997877, | |
| "grad_norm": 0.7381506943950502, | |
| "learning_rate": 3.531267416258144e-06, | |
| "loss": 1.0226, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.4012738853503185, | |
| "grad_norm": 0.7049163316810922, | |
| "learning_rate": 3.5152362181524013e-06, | |
| "loss": 0.936, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.4033970276008493, | |
| "grad_norm": 0.6826780949913842, | |
| "learning_rate": 3.4991548724627054e-06, | |
| "loss": 0.8847, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.40552016985138006, | |
| "grad_norm": 0.6878354292578048, | |
| "learning_rate": 3.4830241735269852e-06, | |
| "loss": 0.9456, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.40764331210191085, | |
| "grad_norm": 0.7027194510942952, | |
| "learning_rate": 3.4668449181209695e-06, | |
| "loss": 0.9483, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.40976645435244163, | |
| "grad_norm": 0.624138086953515, | |
| "learning_rate": 3.450617905418834e-06, | |
| "loss": 0.9822, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.4118895966029724, | |
| "grad_norm": 0.7219498766495058, | |
| "learning_rate": 3.434343936953729e-06, | |
| "loss": 0.9675, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.4140127388535032, | |
| "grad_norm": 0.6397951461970118, | |
| "learning_rate": 3.4180238165781806e-06, | |
| "loss": 0.9268, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.416135881104034, | |
| "grad_norm": 0.6918925648124641, | |
| "learning_rate": 3.4016583504243892e-06, | |
| "loss": 0.9176, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.4182590233545648, | |
| "grad_norm": 0.6466841175073968, | |
| "learning_rate": 3.38524834686441e-06, | |
| "loss": 0.9787, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.42038216560509556, | |
| "grad_norm": 0.89597220464602, | |
| "learning_rate": 3.3687946164702225e-06, | |
| "loss": 1.0087, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.42250530785562634, | |
| "grad_norm": 0.6621404227339525, | |
| "learning_rate": 3.3522979719736923e-06, | |
| "loss": 0.9418, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.42462845010615713, | |
| "grad_norm": 0.6591439083322573, | |
| "learning_rate": 3.3357592282264295e-06, | |
| "loss": 0.9602, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.4267515923566879, | |
| "grad_norm": 0.6574210701544694, | |
| "learning_rate": 3.319179202159532e-06, | |
| "loss": 0.9788, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.4288747346072187, | |
| "grad_norm": 0.6417630612223996, | |
| "learning_rate": 3.3025587127432414e-06, | |
| "loss": 0.9729, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.4309978768577495, | |
| "grad_norm": 0.6486686661277697, | |
| "learning_rate": 3.2858985809464823e-06, | |
| "loss": 0.9292, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.43312101910828027, | |
| "grad_norm": 0.707658030297379, | |
| "learning_rate": 3.2691996296963185e-06, | |
| "loss": 0.9609, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.43524416135881105, | |
| "grad_norm": 0.607355433415427, | |
| "learning_rate": 3.2524626838372976e-06, | |
| "loss": 1.0419, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.43736730360934184, | |
| "grad_norm": 0.6420567814286836, | |
| "learning_rate": 3.2356885700907096e-06, | |
| "loss": 0.95, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.4394904458598726, | |
| "grad_norm": 0.7003629480249687, | |
| "learning_rate": 3.218878117013756e-06, | |
| "loss": 0.958, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.4416135881104034, | |
| "grad_norm": 0.7685385917675293, | |
| "learning_rate": 3.202032154958615e-06, | |
| "loss": 0.9206, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.4437367303609342, | |
| "grad_norm": 0.656199294496736, | |
| "learning_rate": 3.1851515160314346e-06, | |
| "loss": 0.9196, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.445859872611465, | |
| "grad_norm": 0.6566912124054102, | |
| "learning_rate": 3.168237034051222e-06, | |
| "loss": 0.9828, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.44798301486199577, | |
| "grad_norm": 0.7017088990329898, | |
| "learning_rate": 3.151289544508664e-06, | |
| "loss": 0.9555, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.45010615711252655, | |
| "grad_norm": 0.6812153887234823, | |
| "learning_rate": 3.1343098845248564e-06, | |
| "loss": 0.9312, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.45222929936305734, | |
| "grad_norm": 0.7179386805807547, | |
| "learning_rate": 3.117298892809953e-06, | |
| "loss": 0.9316, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.4543524416135881, | |
| "grad_norm": 0.6877605246852548, | |
| "learning_rate": 3.100257409621738e-06, | |
| "loss": 0.9169, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.4564755838641189, | |
| "grad_norm": 0.6792629271222044, | |
| "learning_rate": 3.0831862767241206e-06, | |
| "loss": 0.9256, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.4585987261146497, | |
| "grad_norm": 0.7065974786302712, | |
| "learning_rate": 3.0660863373455594e-06, | |
| "loss": 0.974, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.4607218683651805, | |
| "grad_norm": 0.6680687298095636, | |
| "learning_rate": 3.0489584361374073e-06, | |
| "loss": 1.0146, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.46284501061571126, | |
| "grad_norm": 0.602133219295924, | |
| "learning_rate": 3.031803419132192e-06, | |
| "loss": 0.9495, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.46496815286624205, | |
| "grad_norm": 0.634234424637342, | |
| "learning_rate": 3.014622133701826e-06, | |
| "loss": 0.9312, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.46709129511677283, | |
| "grad_norm": 0.638306267326245, | |
| "learning_rate": 2.99741542851575e-06, | |
| "loss": 0.9509, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.4692144373673036, | |
| "grad_norm": 0.6406442189247321, | |
| "learning_rate": 2.9801841534990115e-06, | |
| "loss": 1.0191, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.4713375796178344, | |
| "grad_norm": 0.6190626628176242, | |
| "learning_rate": 2.96292915979029e-06, | |
| "loss": 0.8954, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.4734607218683652, | |
| "grad_norm": 0.6898377470965286, | |
| "learning_rate": 2.945651299699843e-06, | |
| "loss": 0.9275, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.47558386411889597, | |
| "grad_norm": 0.6613536340865172, | |
| "learning_rate": 2.9283514266674166e-06, | |
| "loss": 1.0029, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.47770700636942676, | |
| "grad_norm": 1.1441795755503248, | |
| "learning_rate": 2.9110303952200863e-06, | |
| "loss": 0.9334, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.47983014861995754, | |
| "grad_norm": 0.6957291992631914, | |
| "learning_rate": 2.893689060930045e-06, | |
| "loss": 0.9323, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.4819532908704883, | |
| "grad_norm": 0.6915777420289582, | |
| "learning_rate": 2.876328280372346e-06, | |
| "loss": 0.9663, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.4840764331210191, | |
| "grad_norm": 0.658451448474378, | |
| "learning_rate": 2.8589489110825895e-06, | |
| "loss": 0.9655, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.4861995753715499, | |
| "grad_norm": 0.7130813908114916, | |
| "learning_rate": 2.8415518115145673e-06, | |
| "loss": 0.986, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.4883227176220807, | |
| "grad_norm": 0.6409022615953691, | |
| "learning_rate": 2.824137840997858e-06, | |
| "loss": 0.9579, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.49044585987261147, | |
| "grad_norm": 0.8843352828327378, | |
| "learning_rate": 2.8067078596953793e-06, | |
| "loss": 0.9156, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.49256900212314225, | |
| "grad_norm": 0.7332731936399352, | |
| "learning_rate": 2.7892627285609035e-06, | |
| "loss": 1.0487, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.49469214437367304, | |
| "grad_norm": 0.7087669563781219, | |
| "learning_rate": 2.7718033092965267e-06, | |
| "loss": 0.9214, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.4968152866242038, | |
| "grad_norm": 0.713075469680895, | |
| "learning_rate": 2.754330464310108e-06, | |
| "loss": 0.9048, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.4989384288747346, | |
| "grad_norm": 0.7753522854989998, | |
| "learning_rate": 2.7368450566726714e-06, | |
| "loss": 0.9705, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.5010615711252654, | |
| "grad_norm": 0.6978893126662162, | |
| "learning_rate": 2.7193479500757684e-06, | |
| "loss": 0.8723, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.5031847133757962, | |
| "grad_norm": 0.7858515284602422, | |
| "learning_rate": 2.7018400087888265e-06, | |
| "loss": 0.9617, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.505307855626327, | |
| "grad_norm": 0.6654303217101978, | |
| "learning_rate": 2.684322097616448e-06, | |
| "loss": 0.99, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.5074309978768577, | |
| "grad_norm": 0.6505973880777469, | |
| "learning_rate": 2.6667950818556992e-06, | |
| "loss": 0.9332, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.5095541401273885, | |
| "grad_norm": 0.6514735895502181, | |
| "learning_rate": 2.649259827253368e-06, | |
| "loss": 1.009, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5116772823779193, | |
| "grad_norm": 0.6589778667351245, | |
| "learning_rate": 2.6317171999631992e-06, | |
| "loss": 0.9445, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.5138004246284501, | |
| "grad_norm": 0.7442921312005703, | |
| "learning_rate": 2.6141680665031116e-06, | |
| "loss": 0.9081, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.5159235668789809, | |
| "grad_norm": 0.6624301541174796, | |
| "learning_rate": 2.5966132937123963e-06, | |
| "loss": 0.9145, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.5180467091295117, | |
| "grad_norm": 0.6691500566559083, | |
| "learning_rate": 2.5790537487088975e-06, | |
| "loss": 0.992, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.5201698513800425, | |
| "grad_norm": 0.6431407890721418, | |
| "learning_rate": 2.561490298846186e-06, | |
| "loss": 1.0049, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.5222929936305732, | |
| "grad_norm": 0.6886918135171396, | |
| "learning_rate": 2.5439238116707103e-06, | |
| "loss": 0.9867, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.524416135881104, | |
| "grad_norm": 0.7161610183514595, | |
| "learning_rate": 2.5263551548789494e-06, | |
| "loss": 0.9655, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.5265392781316348, | |
| "grad_norm": 0.6337123622638601, | |
| "learning_rate": 2.5087851962745468e-06, | |
| "loss": 0.9362, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.5286624203821656, | |
| "grad_norm": 0.6805442118508325, | |
| "learning_rate": 2.4912148037254536e-06, | |
| "loss": 0.9749, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.5307855626326964, | |
| "grad_norm": 0.6850319802840648, | |
| "learning_rate": 2.4736448451210514e-06, | |
| "loss": 0.988, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5329087048832272, | |
| "grad_norm": 0.8209419171009714, | |
| "learning_rate": 2.45607618832929e-06, | |
| "loss": 0.9403, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.535031847133758, | |
| "grad_norm": 0.7609613666487252, | |
| "learning_rate": 2.4385097011538144e-06, | |
| "loss": 0.9481, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.5371549893842887, | |
| "grad_norm": 0.6602663751489974, | |
| "learning_rate": 2.420946251291103e-06, | |
| "loss": 0.9706, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.5392781316348195, | |
| "grad_norm": 0.6202006180428027, | |
| "learning_rate": 2.4033867062876054e-06, | |
| "loss": 1.0119, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.5414012738853503, | |
| "grad_norm": 0.6644803301325269, | |
| "learning_rate": 2.3858319334968892e-06, | |
| "loss": 0.9325, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.5435244161358811, | |
| "grad_norm": 1.1068638082429352, | |
| "learning_rate": 2.368282800036801e-06, | |
| "loss": 0.9264, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.5456475583864119, | |
| "grad_norm": 0.713607799195559, | |
| "learning_rate": 2.350740172746633e-06, | |
| "loss": 0.936, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.5477707006369427, | |
| "grad_norm": 0.6907351623740556, | |
| "learning_rate": 2.3332049181443016e-06, | |
| "loss": 0.9753, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.5498938428874734, | |
| "grad_norm": 0.7074500211398186, | |
| "learning_rate": 2.315677902383553e-06, | |
| "loss": 0.9484, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.5520169851380042, | |
| "grad_norm": 0.6258963310659448, | |
| "learning_rate": 2.298159991211174e-06, | |
| "loss": 0.9195, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.554140127388535, | |
| "grad_norm": 0.6940902471389512, | |
| "learning_rate": 2.280652049924232e-06, | |
| "loss": 0.9607, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.5562632696390658, | |
| "grad_norm": 0.6306810949067136, | |
| "learning_rate": 2.2631549433273294e-06, | |
| "loss": 1.0086, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.5583864118895966, | |
| "grad_norm": 0.6807622298097962, | |
| "learning_rate": 2.2456695356898915e-06, | |
| "loss": 0.8846, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.5605095541401274, | |
| "grad_norm": 0.6737682714342212, | |
| "learning_rate": 2.2281966907034737e-06, | |
| "loss": 0.9424, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.5626326963906582, | |
| "grad_norm": 0.6759877323774287, | |
| "learning_rate": 2.2107372714390973e-06, | |
| "loss": 0.9961, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.564755838641189, | |
| "grad_norm": 0.7062286704853831, | |
| "learning_rate": 2.193292140304621e-06, | |
| "loss": 0.9936, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.5668789808917197, | |
| "grad_norm": 0.6745480828446051, | |
| "learning_rate": 2.175862159002143e-06, | |
| "loss": 0.952, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.5690021231422505, | |
| "grad_norm": 0.6472133490148038, | |
| "learning_rate": 2.158448188485433e-06, | |
| "loss": 1.0431, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.5711252653927813, | |
| "grad_norm": 0.6380862328245611, | |
| "learning_rate": 2.141051088917411e-06, | |
| "loss": 0.9418, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.5732484076433121, | |
| "grad_norm": 0.6308586370084164, | |
| "learning_rate": 2.1236717196276558e-06, | |
| "loss": 0.9539, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.5753715498938429, | |
| "grad_norm": 0.6260367458835687, | |
| "learning_rate": 2.106310939069956e-06, | |
| "loss": 0.9435, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.5774946921443737, | |
| "grad_norm": 0.6674553209217928, | |
| "learning_rate": 2.088969604779914e-06, | |
| "loss": 0.999, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.5796178343949044, | |
| "grad_norm": 0.6682732504211225, | |
| "learning_rate": 2.0716485733325834e-06, | |
| "loss": 0.9189, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.5817409766454352, | |
| "grad_norm": 0.7720268519149664, | |
| "learning_rate": 2.054348700300158e-06, | |
| "loss": 0.9967, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.583864118895966, | |
| "grad_norm": 0.677559579447201, | |
| "learning_rate": 2.037070840209711e-06, | |
| "loss": 1.0036, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.5859872611464968, | |
| "grad_norm": 0.7180703759482054, | |
| "learning_rate": 2.019815846500988e-06, | |
| "loss": 0.978, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.5881104033970276, | |
| "grad_norm": 0.6652884434382036, | |
| "learning_rate": 2.0025845714842514e-06, | |
| "loss": 0.9471, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.5902335456475584, | |
| "grad_norm": 0.6670192985366462, | |
| "learning_rate": 1.9853778662981747e-06, | |
| "loss": 0.9453, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.5923566878980892, | |
| "grad_norm": 0.6816957898824048, | |
| "learning_rate": 1.968196580867808e-06, | |
| "loss": 0.9855, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.5944798301486199, | |
| "grad_norm": 0.6106556647535171, | |
| "learning_rate": 1.951041563862593e-06, | |
| "loss": 1.0012, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.5966029723991507, | |
| "grad_norm": 0.6832368297554903, | |
| "learning_rate": 1.933913662654441e-06, | |
| "loss": 0.99, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.5987261146496815, | |
| "grad_norm": 0.6967673396654995, | |
| "learning_rate": 1.91681372327588e-06, | |
| "loss": 0.9192, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.6008492569002123, | |
| "grad_norm": 0.6838616920307027, | |
| "learning_rate": 1.899742590378263e-06, | |
| "loss": 0.9807, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.6029723991507431, | |
| "grad_norm": 0.6901283920242337, | |
| "learning_rate": 1.8827011071900476e-06, | |
| "loss": 0.974, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.6050955414012739, | |
| "grad_norm": 0.719962073597759, | |
| "learning_rate": 1.865690115475144e-06, | |
| "loss": 0.9923, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.6072186836518046, | |
| "grad_norm": 0.694648374592989, | |
| "learning_rate": 1.8487104554913362e-06, | |
| "loss": 0.9311, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.6093418259023354, | |
| "grad_norm": 0.6149142322093952, | |
| "learning_rate": 1.8317629659487796e-06, | |
| "loss": 0.9273, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.6114649681528662, | |
| "grad_norm": 0.6673784545283598, | |
| "learning_rate": 1.8148484839685663e-06, | |
| "loss": 0.9731, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.613588110403397, | |
| "grad_norm": 0.6882749729992769, | |
| "learning_rate": 1.7979678450413845e-06, | |
| "loss": 0.9049, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.6157112526539278, | |
| "grad_norm": 0.6682848270425117, | |
| "learning_rate": 1.781121882986245e-06, | |
| "loss": 0.9646, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6178343949044586, | |
| "grad_norm": 0.6757546622813148, | |
| "learning_rate": 1.764311429909291e-06, | |
| "loss": 0.9679, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.6199575371549894, | |
| "grad_norm": 0.6213555164128418, | |
| "learning_rate": 1.7475373161627034e-06, | |
| "loss": 0.9799, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.6220806794055201, | |
| "grad_norm": 0.6892438575484697, | |
| "learning_rate": 1.730800370303683e-06, | |
| "loss": 0.9954, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.6242038216560509, | |
| "grad_norm": 0.7655679677743098, | |
| "learning_rate": 1.7141014190535183e-06, | |
| "loss": 0.8903, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.6263269639065817, | |
| "grad_norm": 0.6688086006810142, | |
| "learning_rate": 1.6974412872567598e-06, | |
| "loss": 0.951, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.6284501061571125, | |
| "grad_norm": 0.8034311762095041, | |
| "learning_rate": 1.6808207978404685e-06, | |
| "loss": 0.991, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.6305732484076433, | |
| "grad_norm": 0.6831580011060127, | |
| "learning_rate": 1.6642407717735711e-06, | |
| "loss": 0.9735, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.6326963906581741, | |
| "grad_norm": 0.648345056667966, | |
| "learning_rate": 1.647702028026308e-06, | |
| "loss": 0.9365, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.6348195329087049, | |
| "grad_norm": 0.6142478201259879, | |
| "learning_rate": 1.6312053835297783e-06, | |
| "loss": 0.9832, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.6369426751592356, | |
| "grad_norm": 0.6787332342297232, | |
| "learning_rate": 1.6147516531355912e-06, | |
| "loss": 0.9205, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6390658174097664, | |
| "grad_norm": 0.712325405374025, | |
| "learning_rate": 1.5983416495756116e-06, | |
| "loss": 0.9973, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.6411889596602972, | |
| "grad_norm": 0.682805933407662, | |
| "learning_rate": 1.58197618342182e-06, | |
| "loss": 0.9623, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.643312101910828, | |
| "grad_norm": 0.7161626440164035, | |
| "learning_rate": 1.565656063046272e-06, | |
| "loss": 0.9512, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.6454352441613588, | |
| "grad_norm": 0.6680248951343593, | |
| "learning_rate": 1.549382094581166e-06, | |
| "loss": 0.9152, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.6475583864118896, | |
| "grad_norm": 0.6697551511382988, | |
| "learning_rate": 1.5331550818790313e-06, | |
| "loss": 0.9358, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.6496815286624203, | |
| "grad_norm": 0.639520157757964, | |
| "learning_rate": 1.516975826473015e-06, | |
| "loss": 0.9663, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.6518046709129511, | |
| "grad_norm": 0.6798464046837605, | |
| "learning_rate": 1.5008451275372948e-06, | |
| "loss": 0.9648, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.6539278131634819, | |
| "grad_norm": 0.6181496284024336, | |
| "learning_rate": 1.4847637818475992e-06, | |
| "loss": 0.9688, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.6560509554140127, | |
| "grad_norm": 0.6388645164368109, | |
| "learning_rate": 1.4687325837418564e-06, | |
| "loss": 1.0343, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.6581740976645435, | |
| "grad_norm": 0.6279838621791385, | |
| "learning_rate": 1.4527523250809544e-06, | |
| "loss": 0.991, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.6602972399150743, | |
| "grad_norm": 0.7150277175256885, | |
| "learning_rate": 1.4368237952096258e-06, | |
| "loss": 0.9415, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.6624203821656051, | |
| "grad_norm": 0.684047421357306, | |
| "learning_rate": 1.4209477809174634e-06, | |
| "loss": 0.9117, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.6645435244161358, | |
| "grad_norm": 0.7294054910622478, | |
| "learning_rate": 1.4051250664000515e-06, | |
| "loss": 0.9571, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.6266249724475469, | |
| "learning_rate": 1.389356433220232e-06, | |
| "loss": 0.976, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.6687898089171974, | |
| "grad_norm": 0.6207832680191804, | |
| "learning_rate": 1.3736426602694999e-06, | |
| "loss": 0.918, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.6709129511677282, | |
| "grad_norm": 0.6171295975852529, | |
| "learning_rate": 1.357984523729533e-06, | |
| "loss": 0.9938, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.673036093418259, | |
| "grad_norm": 0.5944506352686731, | |
| "learning_rate": 1.3423827970338422e-06, | |
| "loss": 0.9586, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.6751592356687898, | |
| "grad_norm": 0.6534171622186304, | |
| "learning_rate": 1.3268382508295813e-06, | |
| "loss": 0.9385, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.6772823779193206, | |
| "grad_norm": 0.6440431811178177, | |
| "learning_rate": 1.3113516529394705e-06, | |
| "loss": 0.9463, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.6794055201698513, | |
| "grad_norm": 0.6583864901392601, | |
| "learning_rate": 1.2959237683238768e-06, | |
| "loss": 0.8855, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.6815286624203821, | |
| "grad_norm": 0.6070157954967158, | |
| "learning_rate": 1.2805553590430198e-06, | |
| "loss": 0.9905, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.6836518046709129, | |
| "grad_norm": 0.6610995824632437, | |
| "learning_rate": 1.2652471842193415e-06, | |
| "loss": 0.8951, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.6857749469214437, | |
| "grad_norm": 0.7159136452555714, | |
| "learning_rate": 1.2500000000000007e-06, | |
| "loss": 0.9773, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.6878980891719745, | |
| "grad_norm": 0.6625091451994695, | |
| "learning_rate": 1.2348145595195247e-06, | |
| "loss": 0.9932, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.6900212314225053, | |
| "grad_norm": 0.6836308800831725, | |
| "learning_rate": 1.2196916128626126e-06, | |
| "loss": 0.9818, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.692144373673036, | |
| "grad_norm": 0.6263773815365742, | |
| "learning_rate": 1.2046319070270793e-06, | |
| "loss": 0.9356, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.6942675159235668, | |
| "grad_norm": 0.6290073050296081, | |
| "learning_rate": 1.1896361858869598e-06, | |
| "loss": 0.9702, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.6963906581740976, | |
| "grad_norm": 0.6449334032589804, | |
| "learning_rate": 1.174705190155766e-06, | |
| "loss": 0.9049, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.6985138004246284, | |
| "grad_norm": 0.6537380733830455, | |
| "learning_rate": 1.1598396573499005e-06, | |
| "loss": 0.9279, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.7006369426751592, | |
| "grad_norm": 0.6495854492405805, | |
| "learning_rate": 1.145040321752223e-06, | |
| "loss": 0.987, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.70276008492569, | |
| "grad_norm": 0.6414908229182753, | |
| "learning_rate": 1.1303079143757831e-06, | |
| "loss": 0.9393, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.7048832271762208, | |
| "grad_norm": 0.69117546609185, | |
| "learning_rate": 1.1156431629277118e-06, | |
| "loss": 0.9622, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.7070063694267515, | |
| "grad_norm": 0.6478959458687308, | |
| "learning_rate": 1.1010467917732783e-06, | |
| "loss": 0.9571, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.7091295116772823, | |
| "grad_norm": 0.6379393935432851, | |
| "learning_rate": 1.0865195219001028e-06, | |
| "loss": 0.9481, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.7112526539278131, | |
| "grad_norm": 0.6067597577761649, | |
| "learning_rate": 1.0720620708825536e-06, | |
| "loss": 0.9375, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.7133757961783439, | |
| "grad_norm": 0.6726506283154458, | |
| "learning_rate": 1.0576751528462936e-06, | |
| "loss": 0.9726, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.7154989384288747, | |
| "grad_norm": 0.6385828023233407, | |
| "learning_rate": 1.043359478433012e-06, | |
| "loss": 0.9947, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.7176220806794055, | |
| "grad_norm": 0.6843326829486349, | |
| "learning_rate": 1.0291157547653171e-06, | |
| "loss": 0.9129, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.7197452229299363, | |
| "grad_norm": 0.6057170217437121, | |
| "learning_rate": 1.0149446854118154e-06, | |
| "loss": 0.9618, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.721868365180467, | |
| "grad_norm": 0.6530740442772679, | |
| "learning_rate": 1.0008469703523493e-06, | |
| "loss": 0.9472, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.7239915074309978, | |
| "grad_norm": 0.6342724718691047, | |
| "learning_rate": 9.868233059434286e-07, | |
| "loss": 0.9787, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.7261146496815286, | |
| "grad_norm": 0.612429745101268, | |
| "learning_rate": 9.728743848838334e-07, | |
| "loss": 0.951, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.7282377919320594, | |
| "grad_norm": 0.6079086781829004, | |
| "learning_rate": 9.590008961803942e-07, | |
| "loss": 1.0087, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.7303609341825902, | |
| "grad_norm": 0.6008775263989549, | |
| "learning_rate": 9.452035251139621e-07, | |
| "loss": 0.9733, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.732484076433121, | |
| "grad_norm": 0.6183242340808577, | |
| "learning_rate": 9.314829532055569e-07, | |
| "loss": 0.9175, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.7346072186836518, | |
| "grad_norm": 0.6337902260109236, | |
| "learning_rate": 9.178398581827086e-07, | |
| "loss": 0.9844, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.7367303609341825, | |
| "grad_norm": 0.6428669358079906, | |
| "learning_rate": 9.042749139459711e-07, | |
| "loss": 1.0276, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.7388535031847133, | |
| "grad_norm": 0.6697837382437722, | |
| "learning_rate": 8.907887905356461e-07, | |
| "loss": 0.9436, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.7409766454352441, | |
| "grad_norm": 0.6621674716731518, | |
| "learning_rate": 8.77382154098679e-07, | |
| "loss": 0.9555, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.7430997876857749, | |
| "grad_norm": 0.6660300746258937, | |
| "learning_rate": 8.64055666855757e-07, | |
| "loss": 0.9106, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7452229299363057, | |
| "grad_norm": 0.671931425419172, | |
| "learning_rate": 8.508099870685979e-07, | |
| "loss": 0.9298, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.7473460721868365, | |
| "grad_norm": 0.639197776489002, | |
| "learning_rate": 8.376457690074386e-07, | |
| "loss": 0.9655, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.7494692144373672, | |
| "grad_norm": 0.6284405737772399, | |
| "learning_rate": 8.245636629187121e-07, | |
| "loss": 0.9608, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.7515923566878981, | |
| "grad_norm": 0.6046697179997349, | |
| "learning_rate": 8.115643149929318e-07, | |
| "loss": 1.0282, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.7537154989384289, | |
| "grad_norm": 0.6557740899268497, | |
| "learning_rate": 7.986483673327724e-07, | |
| "loss": 0.9161, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.7558386411889597, | |
| "grad_norm": 0.6194099558851792, | |
| "learning_rate": 7.858164579213548e-07, | |
| "loss": 0.9719, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.7579617834394905, | |
| "grad_norm": 0.6509372982001891, | |
| "learning_rate": 7.730692205907295e-07, | |
| "loss": 0.9372, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.7600849256900213, | |
| "grad_norm": 0.6482168403153363, | |
| "learning_rate": 7.604072849905708e-07, | |
| "loss": 0.9738, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.7622080679405521, | |
| "grad_norm": 0.6752256569397955, | |
| "learning_rate": 7.478312765570747e-07, | |
| "loss": 0.9866, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.7643312101910829, | |
| "grad_norm": 0.6460966865236303, | |
| "learning_rate": 7.353418164820656e-07, | |
| "loss": 0.9503, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.7664543524416136, | |
| "grad_norm": 0.6447254205512877, | |
| "learning_rate": 7.229395216823132e-07, | |
| "loss": 0.8989, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.7685774946921444, | |
| "grad_norm": 0.6761110296588809, | |
| "learning_rate": 7.106250047690588e-07, | |
| "loss": 0.9794, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.7707006369426752, | |
| "grad_norm": 0.6147953221769132, | |
| "learning_rate": 6.983988740177553e-07, | |
| "loss": 0.9181, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.772823779193206, | |
| "grad_norm": 0.6777675753663922, | |
| "learning_rate": 6.862617333380214e-07, | |
| "loss": 0.9905, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.7749469214437368, | |
| "grad_norm": 0.5914897232209698, | |
| "learning_rate": 6.742141822438144e-07, | |
| "loss": 0.9688, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.7770700636942676, | |
| "grad_norm": 0.6260916500009547, | |
| "learning_rate": 6.622568158238126e-07, | |
| "loss": 0.897, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.7791932059447984, | |
| "grad_norm": 0.632626776061231, | |
| "learning_rate": 6.503902247120239e-07, | |
| "loss": 0.9679, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.7813163481953291, | |
| "grad_norm": 0.6560953608825832, | |
| "learning_rate": 6.386149950586099e-07, | |
| "loss": 0.9878, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.7834394904458599, | |
| "grad_norm": 0.6628696841079698, | |
| "learning_rate": 6.269317085009364e-07, | |
| "loss": 0.9181, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.7855626326963907, | |
| "grad_norm": 0.6300479712329049, | |
| "learning_rate": 6.153409421348358e-07, | |
| "loss": 0.9286, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.7876857749469215, | |
| "grad_norm": 0.6820421121103658, | |
| "learning_rate": 6.038432684861123e-07, | |
| "loss": 0.9367, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.7898089171974523, | |
| "grad_norm": 0.6370315920691889, | |
| "learning_rate": 5.92439255482252e-07, | |
| "loss": 0.9406, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.7919320594479831, | |
| "grad_norm": 0.6621812998395536, | |
| "learning_rate": 5.811294664243752e-07, | |
| "loss": 0.8829, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.7940552016985138, | |
| "grad_norm": 0.6664877663774411, | |
| "learning_rate": 5.699144599594131e-07, | |
| "loss": 0.9202, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.7961783439490446, | |
| "grad_norm": 0.6182657772931999, | |
| "learning_rate": 5.587947900525093e-07, | |
| "loss": 0.9487, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.7983014861995754, | |
| "grad_norm": 0.6208375998135864, | |
| "learning_rate": 5.4777100595966e-07, | |
| "loss": 0.9979, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.8004246284501062, | |
| "grad_norm": 0.721058744812672, | |
| "learning_rate": 5.368436522005816e-07, | |
| "loss": 0.9334, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.802547770700637, | |
| "grad_norm": 0.6596706860365731, | |
| "learning_rate": 5.26013268531817e-07, | |
| "loss": 0.9607, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.8046709129511678, | |
| "grad_norm": 0.6451230769613401, | |
| "learning_rate": 5.1528038992007e-07, | |
| "loss": 0.9963, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.8067940552016986, | |
| "grad_norm": 0.6333607561711646, | |
| "learning_rate": 5.046455465157849e-07, | |
| "loss": 0.9317, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.8089171974522293, | |
| "grad_norm": 0.619357732653376, | |
| "learning_rate": 4.941092636269554e-07, | |
| "loss": 1.0315, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.8110403397027601, | |
| "grad_norm": 0.6393177970016626, | |
| "learning_rate": 4.836720616931831e-07, | |
| "loss": 0.9902, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.8131634819532909, | |
| "grad_norm": 0.6258314508225672, | |
| "learning_rate": 4.7333445625996233e-07, | |
| "loss": 0.963, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.8152866242038217, | |
| "grad_norm": 0.6149025413970899, | |
| "learning_rate": 4.6309695795322323e-07, | |
| "loss": 0.9254, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.8174097664543525, | |
| "grad_norm": 0.6474654664009011, | |
| "learning_rate": 4.5296007245410225e-07, | |
| "loss": 0.9177, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.8195329087048833, | |
| "grad_norm": 0.663975882286753, | |
| "learning_rate": 4.429243004739692e-07, | |
| "loss": 0.8934, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.821656050955414, | |
| "grad_norm": 0.6297462844932081, | |
| "learning_rate": 4.329901377296902e-07, | |
| "loss": 0.9851, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.8237791932059448, | |
| "grad_norm": 0.6470027446979043, | |
| "learning_rate": 4.2315807491914747e-07, | |
| "loss": 0.9447, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.8259023354564756, | |
| "grad_norm": 0.6631902796897037, | |
| "learning_rate": 4.1342859769699483e-07, | |
| "loss": 1.0306, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.8280254777070064, | |
| "grad_norm": 0.6262758143393297, | |
| "learning_rate": 4.038021866506725e-07, | |
| "loss": 0.9395, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.8301486199575372, | |
| "grad_norm": 0.6072675971421893, | |
| "learning_rate": 3.9427931727666983e-07, | |
| "loss": 1.0107, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.832271762208068, | |
| "grad_norm": 0.6856116016409511, | |
| "learning_rate": 3.848604599570338e-07, | |
| "loss": 0.9383, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.8343949044585988, | |
| "grad_norm": 0.635409995447785, | |
| "learning_rate": 3.7554607993613823e-07, | |
| "loss": 0.943, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.8365180467091295, | |
| "grad_norm": 0.6455716027042272, | |
| "learning_rate": 3.6633663729770007e-07, | |
| "loss": 0.9254, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.8386411889596603, | |
| "grad_norm": 0.7061618701086447, | |
| "learning_rate": 3.572325869420587e-07, | |
| "loss": 0.9286, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.8407643312101911, | |
| "grad_norm": 0.6526609109650112, | |
| "learning_rate": 3.48234378563698e-07, | |
| "loss": 0.9486, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.8428874734607219, | |
| "grad_norm": 0.657214187790047, | |
| "learning_rate": 3.3934245662904207e-07, | |
| "loss": 0.9831, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.8450106157112527, | |
| "grad_norm": 0.6637443803485965, | |
| "learning_rate": 3.3055726035449486e-07, | |
| "loss": 1.0235, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.8471337579617835, | |
| "grad_norm": 0.6057711263679968, | |
| "learning_rate": 3.2187922368474954e-07, | |
| "loss": 0.9503, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.8492569002123143, | |
| "grad_norm": 0.629069248815946, | |
| "learning_rate": 3.133087752713479e-07, | |
| "loss": 1.0026, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.851380042462845, | |
| "grad_norm": 0.6979884576365777, | |
| "learning_rate": 3.048463384515149e-07, | |
| "loss": 0.9899, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.8535031847133758, | |
| "grad_norm": 0.6787320418702446, | |
| "learning_rate": 2.9649233122724106e-07, | |
| "loss": 0.9066, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.8556263269639066, | |
| "grad_norm": 0.6748182950146039, | |
| "learning_rate": 2.88247166244639e-07, | |
| "loss": 0.9937, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.8577494692144374, | |
| "grad_norm": 0.6112251709107203, | |
| "learning_rate": 2.8011125077355873e-07, | |
| "loss": 0.9487, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.8598726114649682, | |
| "grad_norm": 0.631590234282686, | |
| "learning_rate": 2.7208498668747393e-07, | |
| "loss": 0.926, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.861995753715499, | |
| "grad_norm": 0.6645830704461472, | |
| "learning_rate": 2.641687704436269e-07, | |
| "loss": 0.9827, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.8641188959660298, | |
| "grad_norm": 0.6442641148518611, | |
| "learning_rate": 2.563629930634487e-07, | |
| "loss": 0.9569, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.8662420382165605, | |
| "grad_norm": 0.6389819326112781, | |
| "learning_rate": 2.4866804011324296e-07, | |
| "loss": 0.9892, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.8683651804670913, | |
| "grad_norm": 0.694376173474049, | |
| "learning_rate": 2.410842916851425e-07, | |
| "loss": 0.9693, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.8704883227176221, | |
| "grad_norm": 0.6777970970974673, | |
| "learning_rate": 2.3361212237833357e-07, | |
| "loss": 0.9915, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.8726114649681529, | |
| "grad_norm": 0.6377492824234444, | |
| "learning_rate": 2.2625190128055168e-07, | |
| "loss": 0.9578, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.8747346072186837, | |
| "grad_norm": 0.705803602733728, | |
| "learning_rate": 2.1900399194985434e-07, | |
| "loss": 0.9719, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.8768577494692145, | |
| "grad_norm": 0.6395227286271804, | |
| "learning_rate": 2.118687523966559e-07, | |
| "loss": 0.9637, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.8789808917197452, | |
| "grad_norm": 0.6411571767299858, | |
| "learning_rate": 2.0484653506605223e-07, | |
| "loss": 0.9595, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.881104033970276, | |
| "grad_norm": 0.6426927564445861, | |
| "learning_rate": 1.9793768682040526e-07, | |
| "loss": 0.9981, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.8832271762208068, | |
| "grad_norm": 0.6307390400609545, | |
| "learning_rate": 1.911425489222127e-07, | |
| "loss": 0.9901, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.8853503184713376, | |
| "grad_norm": 0.6290789213902569, | |
| "learning_rate": 1.8446145701724982e-07, | |
| "loss": 0.9724, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.8874734607218684, | |
| "grad_norm": 0.6569246549564896, | |
| "learning_rate": 1.778947411179932e-07, | |
| "loss": 0.9871, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.8895966029723992, | |
| "grad_norm": 0.6687695448026262, | |
| "learning_rate": 1.7144272558731466e-07, | |
| "loss": 0.9069, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.89171974522293, | |
| "grad_norm": 0.6281132023225879, | |
| "learning_rate": 1.6510572912246476e-07, | |
| "loss": 0.914, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.8938428874734607, | |
| "grad_norm": 0.6238753091596155, | |
| "learning_rate": 1.5888406473932694e-07, | |
| "loss": 0.9244, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.8959660297239915, | |
| "grad_norm": 4.46036810790257, | |
| "learning_rate": 1.5277803975695865e-07, | |
| "loss": 0.9547, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.8980891719745223, | |
| "grad_norm": 0.600923175816901, | |
| "learning_rate": 1.4678795578240897e-07, | |
| "loss": 0.9437, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.9002123142250531, | |
| "grad_norm": 0.668587726809177, | |
| "learning_rate": 1.4091410869582267e-07, | |
| "loss": 0.9638, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.9023354564755839, | |
| "grad_norm": 0.6584076537520875, | |
| "learning_rate": 1.3515678863582432e-07, | |
| "loss": 0.9784, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.9044585987261147, | |
| "grad_norm": 0.667969952697823, | |
| "learning_rate": 1.2951627998518623e-07, | |
| "loss": 0.9596, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.9065817409766455, | |
| "grad_norm": 0.6494149342591721, | |
| "learning_rate": 1.2399286135678424e-07, | |
| "loss": 0.9505, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.9087048832271762, | |
| "grad_norm": 0.6453904770366586, | |
| "learning_rate": 1.1858680557983171e-07, | |
| "loss": 0.9491, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.910828025477707, | |
| "grad_norm": 0.5896715257061248, | |
| "learning_rate": 1.1329837968640539e-07, | |
| "loss": 0.9583, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.9129511677282378, | |
| "grad_norm": 0.6299383258809309, | |
| "learning_rate": 1.0812784489825506e-07, | |
| "loss": 0.9278, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.9150743099787686, | |
| "grad_norm": 0.6223623994432069, | |
| "learning_rate": 1.030754566139014e-07, | |
| "loss": 0.9707, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.9171974522292994, | |
| "grad_norm": 0.6426522785730759, | |
| "learning_rate": 9.814146439601741e-08, | |
| "loss": 0.9676, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.9193205944798302, | |
| "grad_norm": 0.5991827050856197, | |
| "learning_rate": 9.332611195910585e-08, | |
| "loss": 0.9621, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.921443736730361, | |
| "grad_norm": 0.6379437245945956, | |
| "learning_rate": 8.862963715745687e-08, | |
| "loss": 0.9289, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.9235668789808917, | |
| "grad_norm": 0.6154122200818485, | |
| "learning_rate": 8.405227197340216e-08, | |
| "loss": 0.8987, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.9256900212314225, | |
| "grad_norm": 0.6218251534617395, | |
| "learning_rate": 7.959424250585323e-08, | |
| "loss": 0.9971, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.9278131634819533, | |
| "grad_norm": 0.6334926602648896, | |
| "learning_rate": 7.525576895913655e-08, | |
| "loss": 0.9252, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.9299363057324841, | |
| "grad_norm": 0.6844897258602354, | |
| "learning_rate": 7.10370656321141e-08, | |
| "loss": 0.9089, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.9320594479830149, | |
| "grad_norm": 0.6358730896865131, | |
| "learning_rate": 6.69383409075991e-08, | |
| "loss": 0.9387, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.9341825902335457, | |
| "grad_norm": 0.6691049771820187, | |
| "learning_rate": 6.295979724206314e-08, | |
| "loss": 0.9779, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.9363057324840764, | |
| "grad_norm": 0.6245277020630183, | |
| "learning_rate": 5.910163115563472e-08, | |
| "loss": 0.9406, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.9384288747346072, | |
| "grad_norm": 0.5976591557512331, | |
| "learning_rate": 5.53640332223937e-08, | |
| "loss": 0.8835, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.940552016985138, | |
| "grad_norm": 0.6334583873928097, | |
| "learning_rate": 5.1747188060956296e-08, | |
| "loss": 0.9401, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.9426751592356688, | |
| "grad_norm": 0.652725927152352, | |
| "learning_rate": 4.825127432535714e-08, | |
| "loss": 0.9157, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.9447983014861996, | |
| "grad_norm": 0.6169937022982677, | |
| "learning_rate": 4.487646469622464e-08, | |
| "loss": 0.9507, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.9469214437367304, | |
| "grad_norm": 0.6900335357622487, | |
| "learning_rate": 4.1622925872249476e-08, | |
| "loss": 0.9415, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.9490445859872612, | |
| "grad_norm": 0.6863521324823155, | |
| "learning_rate": 3.849081856195341e-08, | |
| "loss": 0.9436, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.9511677282377919, | |
| "grad_norm": 0.6327941599710385, | |
| "learning_rate": 3.548029747574927e-08, | |
| "loss": 0.9609, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.9532908704883227, | |
| "grad_norm": 0.7174861358519089, | |
| "learning_rate": 3.259151131829869e-08, | |
| "loss": 0.9846, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.9554140127388535, | |
| "grad_norm": 0.6533979795497699, | |
| "learning_rate": 2.982460278116883e-08, | |
| "loss": 0.9633, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9575371549893843, | |
| "grad_norm": 0.6349445043972284, | |
| "learning_rate": 2.7179708535781945e-08, | |
| "loss": 0.967, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.9596602972399151, | |
| "grad_norm": 0.668415847583719, | |
| "learning_rate": 2.4656959226665445e-08, | |
| "loss": 0.968, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.9617834394904459, | |
| "grad_norm": 0.6890284091778482, | |
| "learning_rate": 2.2256479464999315e-08, | |
| "loss": 0.9741, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.9639065817409767, | |
| "grad_norm": 0.5908876158682079, | |
| "learning_rate": 1.99783878224602e-08, | |
| "loss": 0.9666, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.9660297239915074, | |
| "grad_norm": 0.6592959391453375, | |
| "learning_rate": 1.7822796825364142e-08, | |
| "loss": 0.9542, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.9681528662420382, | |
| "grad_norm": 0.6208149971886308, | |
| "learning_rate": 1.578981294910936e-08, | |
| "loss": 0.9273, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.970276008492569, | |
| "grad_norm": 0.6055645797730091, | |
| "learning_rate": 1.38795366129163e-08, | |
| "loss": 0.9119, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.9723991507430998, | |
| "grad_norm": 0.64150062452142, | |
| "learning_rate": 1.2092062174867414e-08, | |
| "loss": 0.9142, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.9745222929936306, | |
| "grad_norm": 0.6520718504070266, | |
| "learning_rate": 1.042747792724702e-08, | |
| "loss": 0.9492, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.9766454352441614, | |
| "grad_norm": 0.6299104202050195, | |
| "learning_rate": 8.885866092178952e-09, | |
| "loss": 0.9837, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.9787685774946921, | |
| "grad_norm": 0.6379380608023664, | |
| "learning_rate": 7.467302817566192e-09, | |
| "loss": 0.9524, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.9808917197452229, | |
| "grad_norm": 0.6559585459109882, | |
| "learning_rate": 6.171858173328604e-09, | |
| "loss": 0.9818, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.9830148619957537, | |
| "grad_norm": 0.6532382649202735, | |
| "learning_rate": 4.999596147943486e-09, | |
| "loss": 0.909, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.9851380042462845, | |
| "grad_norm": 0.6708914977227046, | |
| "learning_rate": 3.95057464528309e-09, | |
| "loss": 0.983, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.9872611464968153, | |
| "grad_norm": 0.644273751253915, | |
| "learning_rate": 3.02484548175469e-09, | |
| "loss": 1.0171, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.9893842887473461, | |
| "grad_norm": 0.7024726425096608, | |
| "learning_rate": 2.222454383742356e-09, | |
| "loss": 0.9592, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.9915074309978769, | |
| "grad_norm": 0.6291344084647575, | |
| "learning_rate": 1.5434409853473664e-09, | |
| "loss": 0.9285, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.9936305732484076, | |
| "grad_norm": 0.6141900504735813, | |
| "learning_rate": 9.878388264300543e-10, | |
| "loss": 0.9444, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.9957537154989384, | |
| "grad_norm": 0.5878235469690039, | |
| "learning_rate": 5.55675350954743e-10, | |
| "loss": 0.9222, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.9978768577494692, | |
| "grad_norm": 0.64828405546267, | |
| "learning_rate": 2.4697190563194307e-10, | |
| "loss": 0.9463, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.6112045913919388, | |
| "learning_rate": 6.174373886586038e-11, | |
| "loss": 0.949, | |
| "step": 471 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 471, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 157, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 70004199849984.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |