| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9995388034044694, | |
| "eval_steps": 500, | |
| "global_step": 745, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0013416628233617039, | |
| "grad_norm": 38.48973609938971, | |
| "learning_rate": 0.0, | |
| "loss": 2.7745, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0026833256467234078, | |
| "grad_norm": 40.0874086838779, | |
| "learning_rate": 4.347826086956522e-07, | |
| "loss": 2.7327, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.004024988470085112, | |
| "grad_norm": 38.96060688685107, | |
| "learning_rate": 8.695652173913044e-07, | |
| "loss": 2.718, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0053666512934468155, | |
| "grad_norm": 38.56280046956328, | |
| "learning_rate": 1.3043478260869566e-06, | |
| "loss": 2.8075, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.006708314116808519, | |
| "grad_norm": 37.23315669576612, | |
| "learning_rate": 1.7391304347826088e-06, | |
| "loss": 2.6967, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.008049976940170223, | |
| "grad_norm": 33.46199452786037, | |
| "learning_rate": 2.173913043478261e-06, | |
| "loss": 2.6546, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.009391639763531927, | |
| "grad_norm": 31.355735362649718, | |
| "learning_rate": 2.6086956521739132e-06, | |
| "loss": 2.6463, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.010733302586893631, | |
| "grad_norm": 24.34325607124612, | |
| "learning_rate": 3.043478260869566e-06, | |
| "loss": 2.4834, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.012074965410255335, | |
| "grad_norm": 22.966931487164434, | |
| "learning_rate": 3.4782608695652175e-06, | |
| "loss": 2.4298, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.013416628233617039, | |
| "grad_norm": 14.52974303349036, | |
| "learning_rate": 3.91304347826087e-06, | |
| "loss": 2.1429, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.014758291056978743, | |
| "grad_norm": 13.045534352630535, | |
| "learning_rate": 4.347826086956522e-06, | |
| "loss": 2.0581, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.016099953880340447, | |
| "grad_norm": 11.707299088535546, | |
| "learning_rate": 4.782608695652174e-06, | |
| "loss": 2.0318, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.017441616703702152, | |
| "grad_norm": 13.139238478366556, | |
| "learning_rate": 5.2173913043478265e-06, | |
| "loss": 1.9436, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.018783279527063854, | |
| "grad_norm": 8.190474422197797, | |
| "learning_rate": 5.652173913043479e-06, | |
| "loss": 1.8122, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.02012494235042556, | |
| "grad_norm": 6.647921658666986, | |
| "learning_rate": 6.086956521739132e-06, | |
| "loss": 1.8019, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.021466605173787262, | |
| "grad_norm": 4.958488743145487, | |
| "learning_rate": 6.521739130434783e-06, | |
| "loss": 1.7389, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.022808267997148968, | |
| "grad_norm": 4.400330010217928, | |
| "learning_rate": 6.956521739130435e-06, | |
| "loss": 1.7066, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.02414993082051067, | |
| "grad_norm": 3.627973706802729, | |
| "learning_rate": 7.391304347826087e-06, | |
| "loss": 1.6704, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.025491593643872375, | |
| "grad_norm": 3.5769764697569753, | |
| "learning_rate": 7.82608695652174e-06, | |
| "loss": 1.6282, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.026833256467234078, | |
| "grad_norm": 3.764679416544402, | |
| "learning_rate": 8.260869565217392e-06, | |
| "loss": 1.5957, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.028174919290595783, | |
| "grad_norm": 2.871830528274261, | |
| "learning_rate": 8.695652173913044e-06, | |
| "loss": 1.5711, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.029516582113957485, | |
| "grad_norm": 2.428750661375507, | |
| "learning_rate": 9.130434782608697e-06, | |
| "loss": 1.5456, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.03085824493731919, | |
| "grad_norm": 2.256440666288964, | |
| "learning_rate": 9.565217391304349e-06, | |
| "loss": 1.4742, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.03219990776068089, | |
| "grad_norm": 2.723813414935773, | |
| "learning_rate": 1e-05, | |
| "loss": 1.4917, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0335415705840426, | |
| "grad_norm": 1.8720528042807492, | |
| "learning_rate": 9.99995279777441e-06, | |
| "loss": 1.4593, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.034883233407404304, | |
| "grad_norm": 1.7198126021417508, | |
| "learning_rate": 9.999811191988863e-06, | |
| "loss": 1.4695, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.036224896230766, | |
| "grad_norm": 1.6679472256558474, | |
| "learning_rate": 9.999575185316994e-06, | |
| "loss": 1.4246, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.03756655905412771, | |
| "grad_norm": 1.5381214409323023, | |
| "learning_rate": 9.999244782214828e-06, | |
| "loss": 1.3968, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.038908221877489414, | |
| "grad_norm": 4.591499175458982, | |
| "learning_rate": 9.998819988920665e-06, | |
| "loss": 1.3693, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.04024988470085112, | |
| "grad_norm": 1.8623520533496765, | |
| "learning_rate": 9.998300813454981e-06, | |
| "loss": 1.3941, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04159154752421282, | |
| "grad_norm": 1.7437598519757866, | |
| "learning_rate": 9.997687265620274e-06, | |
| "loss": 1.3614, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.042933210347574524, | |
| "grad_norm": 1.4366026669162075, | |
| "learning_rate": 9.996979357000869e-06, | |
| "loss": 1.3764, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.04427487317093623, | |
| "grad_norm": 1.3565401187123118, | |
| "learning_rate": 9.996177100962714e-06, | |
| "loss": 1.316, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.045616535994297935, | |
| "grad_norm": 1.2948639756531317, | |
| "learning_rate": 9.995280512653116e-06, | |
| "loss": 1.3193, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.046958198817659634, | |
| "grad_norm": 1.3385746515590846, | |
| "learning_rate": 9.99428960900046e-06, | |
| "loss": 1.3347, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.04829986164102134, | |
| "grad_norm": 1.2113204225595948, | |
| "learning_rate": 9.99320440871389e-06, | |
| "loss": 1.3017, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.049641524464383045, | |
| "grad_norm": 1.1747592623560767, | |
| "learning_rate": 9.992024932282955e-06, | |
| "loss": 1.3117, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.05098318728774475, | |
| "grad_norm": 1.1828988280330426, | |
| "learning_rate": 9.990751201977217e-06, | |
| "loss": 1.3097, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.05232485011110645, | |
| "grad_norm": 1.2617257737187395, | |
| "learning_rate": 9.98938324184584e-06, | |
| "loss": 1.2605, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.053666512934468155, | |
| "grad_norm": 1.5091324163800584, | |
| "learning_rate": 9.987921077717127e-06, | |
| "loss": 1.2597, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05500817575782986, | |
| "grad_norm": 1.2047961755289263, | |
| "learning_rate": 9.98636473719804e-06, | |
| "loss": 1.2576, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.056349838581191566, | |
| "grad_norm": 1.1654648794501645, | |
| "learning_rate": 9.984714249673676e-06, | |
| "loss": 1.2578, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.057691501404553265, | |
| "grad_norm": 1.2188610566064713, | |
| "learning_rate": 9.982969646306704e-06, | |
| "loss": 1.242, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.05903316422791497, | |
| "grad_norm": 1.1135746337829162, | |
| "learning_rate": 9.98113096003679e-06, | |
| "loss": 1.2362, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.060374827051276676, | |
| "grad_norm": 1.2295307989757311, | |
| "learning_rate": 9.979198225579968e-06, | |
| "loss": 1.2304, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.06171648987463838, | |
| "grad_norm": 1.281785874820186, | |
| "learning_rate": 9.97717147942799e-06, | |
| "loss": 1.2093, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.06305815269800008, | |
| "grad_norm": 1.455057311755706, | |
| "learning_rate": 9.97505075984762e-06, | |
| "loss": 1.2213, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.06439981552136179, | |
| "grad_norm": 1.2952480154822823, | |
| "learning_rate": 9.972836106879936e-06, | |
| "loss": 1.2143, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.06574147834472349, | |
| "grad_norm": 1.3065138725543972, | |
| "learning_rate": 9.970527562339554e-06, | |
| "loss": 1.2089, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0670831411680852, | |
| "grad_norm": 1.076622624579616, | |
| "learning_rate": 9.968125169813855e-06, | |
| "loss": 1.1908, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0684248039914469, | |
| "grad_norm": 1.096771830651398, | |
| "learning_rate": 9.965628974662145e-06, | |
| "loss": 1.1954, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.06976646681480861, | |
| "grad_norm": 1.250110050693406, | |
| "learning_rate": 9.963039024014811e-06, | |
| "loss": 1.1892, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0711081296381703, | |
| "grad_norm": 1.2945997276854586, | |
| "learning_rate": 9.96035536677243e-06, | |
| "loss": 1.1831, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.072449792461532, | |
| "grad_norm": 1.4338198138868328, | |
| "learning_rate": 9.957578053604837e-06, | |
| "loss": 1.1854, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.07379145528489371, | |
| "grad_norm": 1.485304957168343, | |
| "learning_rate": 9.954707136950176e-06, | |
| "loss": 1.1715, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.07513311810825542, | |
| "grad_norm": 1.443004775374296, | |
| "learning_rate": 9.951742671013914e-06, | |
| "loss": 1.1575, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.07647478093161712, | |
| "grad_norm": 1.5432551481625234, | |
| "learning_rate": 9.9486847117678e-06, | |
| "loss": 1.1343, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.07781644375497883, | |
| "grad_norm": 4.379703192128488, | |
| "learning_rate": 9.945533316948833e-06, | |
| "loss": 1.131, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.07915810657834053, | |
| "grad_norm": 1.3796938829655623, | |
| "learning_rate": 9.942288546058148e-06, | |
| "loss": 1.1459, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.08049976940170224, | |
| "grad_norm": 1.6863063853942253, | |
| "learning_rate": 9.938950460359912e-06, | |
| "loss": 1.1446, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.08184143222506395, | |
| "grad_norm": 1.5386826592834415, | |
| "learning_rate": 9.935519122880152e-06, | |
| "loss": 1.1422, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.08318309504842564, | |
| "grad_norm": 1.1028884753487724, | |
| "learning_rate": 9.931994598405576e-06, | |
| "loss": 1.1482, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.08452475787178734, | |
| "grad_norm": 1.5693614514575673, | |
| "learning_rate": 9.928376953482343e-06, | |
| "loss": 1.1298, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.08586642069514905, | |
| "grad_norm": 1.3233718170907751, | |
| "learning_rate": 9.924666256414812e-06, | |
| "loss": 1.1364, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.08720808351851075, | |
| "grad_norm": 1.2605390766931979, | |
| "learning_rate": 9.920862577264242e-06, | |
| "loss": 1.1229, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.08854974634187246, | |
| "grad_norm": 2.3090812645678533, | |
| "learning_rate": 9.916965987847485e-06, | |
| "loss": 1.1193, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.08989140916523417, | |
| "grad_norm": 1.154396872328338, | |
| "learning_rate": 9.912976561735617e-06, | |
| "loss": 1.0964, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.09123307198859587, | |
| "grad_norm": 1.466260207583527, | |
| "learning_rate": 9.908894374252556e-06, | |
| "loss": 1.1144, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.09257473481195758, | |
| "grad_norm": 1.304279853496226, | |
| "learning_rate": 9.904719502473635e-06, | |
| "loss": 1.1184, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.09391639763531927, | |
| "grad_norm": 1.2876032988786934, | |
| "learning_rate": 9.900452025224148e-06, | |
| "loss": 1.1078, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09525806045868097, | |
| "grad_norm": 1.2879145531472471, | |
| "learning_rate": 9.896092023077866e-06, | |
| "loss": 1.1087, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.09659972328204268, | |
| "grad_norm": 1.1885746137305104, | |
| "learning_rate": 9.891639578355511e-06, | |
| "loss": 1.1042, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.09794138610540438, | |
| "grad_norm": 1.1309624778962615, | |
| "learning_rate": 9.887094775123203e-06, | |
| "loss": 1.1103, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.09928304892876609, | |
| "grad_norm": 1.173885541691277, | |
| "learning_rate": 9.882457699190874e-06, | |
| "loss": 1.096, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.1006247117521278, | |
| "grad_norm": 1.4067525465834194, | |
| "learning_rate": 9.877728438110645e-06, | |
| "loss": 1.1021, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1019663745754895, | |
| "grad_norm": 1.4074794348566437, | |
| "learning_rate": 9.872907081175175e-06, | |
| "loss": 1.1121, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.10330803739885121, | |
| "grad_norm": 1.1488091100998024, | |
| "learning_rate": 9.867993719415974e-06, | |
| "loss": 1.0907, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.1046497002222129, | |
| "grad_norm": 1.4399599127571547, | |
| "learning_rate": 9.86298844560169e-06, | |
| "loss": 1.0786, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.1059913630455746, | |
| "grad_norm": 1.1622368595195733, | |
| "learning_rate": 9.857891354236342e-06, | |
| "loss": 1.0814, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.10733302586893631, | |
| "grad_norm": 1.2948481200374429, | |
| "learning_rate": 9.852702541557559e-06, | |
| "loss": 1.0922, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.10867468869229802, | |
| "grad_norm": 1.2974149486520998, | |
| "learning_rate": 9.847422105534739e-06, | |
| "loss": 1.0656, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.11001635151565972, | |
| "grad_norm": 1.7501728124352502, | |
| "learning_rate": 9.842050145867219e-06, | |
| "loss": 1.0734, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.11135801433902143, | |
| "grad_norm": 1.4884760574133242, | |
| "learning_rate": 9.836586763982376e-06, | |
| "loss": 1.0737, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.11269967716238313, | |
| "grad_norm": 1.213069682520622, | |
| "learning_rate": 9.831032063033726e-06, | |
| "loss": 1.0471, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.11404133998574484, | |
| "grad_norm": 1.3319329965977207, | |
| "learning_rate": 9.825386147898967e-06, | |
| "loss": 1.0828, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.11538300280910653, | |
| "grad_norm": 1.263257283427598, | |
| "learning_rate": 9.819649125178004e-06, | |
| "loss": 1.0693, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.11672466563246824, | |
| "grad_norm": 1.2275720778436565, | |
| "learning_rate": 9.813821103190932e-06, | |
| "loss": 1.0748, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.11806632845582994, | |
| "grad_norm": 1.304958073659253, | |
| "learning_rate": 9.807902191975996e-06, | |
| "loss": 1.069, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.11940799127919165, | |
| "grad_norm": 1.5011487754468782, | |
| "learning_rate": 9.801892503287507e-06, | |
| "loss": 1.0609, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.12074965410255335, | |
| "grad_norm": 1.6724656060436685, | |
| "learning_rate": 9.795792150593739e-06, | |
| "loss": 1.0757, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.12209131692591506, | |
| "grad_norm": 1.4650906803142847, | |
| "learning_rate": 9.789601249074781e-06, | |
| "loss": 1.0477, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.12343297974927676, | |
| "grad_norm": 1.3282993707440565, | |
| "learning_rate": 9.783319915620365e-06, | |
| "loss": 1.0805, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.12477464257263847, | |
| "grad_norm": 1.1332262904896697, | |
| "learning_rate": 9.776948268827658e-06, | |
| "loss": 1.0463, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.12611630539600016, | |
| "grad_norm": 1.289487333289698, | |
| "learning_rate": 9.770486428999026e-06, | |
| "loss": 1.0416, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.12745796821936187, | |
| "grad_norm": 1.1896501786756817, | |
| "learning_rate": 9.763934518139754e-06, | |
| "loss": 1.0399, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.12879963104272357, | |
| "grad_norm": 1.217899281067364, | |
| "learning_rate": 9.757292659955755e-06, | |
| "loss": 1.0539, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.13014129386608528, | |
| "grad_norm": 1.378315349430874, | |
| "learning_rate": 9.750560979851222e-06, | |
| "loss": 1.0423, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.13148295668944698, | |
| "grad_norm": 1.8521057902594815, | |
| "learning_rate": 9.743739604926268e-06, | |
| "loss": 1.0473, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.1328246195128087, | |
| "grad_norm": 1.3045953296551285, | |
| "learning_rate": 9.736828663974527e-06, | |
| "loss": 1.0264, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.1341662823361704, | |
| "grad_norm": 1.323259326309174, | |
| "learning_rate": 9.729828287480713e-06, | |
| "loss": 1.0412, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1355079451595321, | |
| "grad_norm": 1.3848917326606898, | |
| "learning_rate": 9.722738607618171e-06, | |
| "loss": 1.05, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.1368496079828938, | |
| "grad_norm": 1.4253512390195084, | |
| "learning_rate": 9.715559758246363e-06, | |
| "loss": 1.0441, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.1381912708062555, | |
| "grad_norm": 1.2292766694528439, | |
| "learning_rate": 9.70829187490836e-06, | |
| "loss": 1.0471, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.13953293362961722, | |
| "grad_norm": 1.2600308298936818, | |
| "learning_rate": 9.700935094828267e-06, | |
| "loss": 1.0427, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.14087459645297892, | |
| "grad_norm": 1.3101049105612972, | |
| "learning_rate": 9.693489556908641e-06, | |
| "loss": 1.0445, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.1422162592763406, | |
| "grad_norm": 1.440220506400761, | |
| "learning_rate": 9.68595540172787e-06, | |
| "loss": 1.0024, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.1435579220997023, | |
| "grad_norm": 1.369469800483532, | |
| "learning_rate": 9.678332771537506e-06, | |
| "loss": 1.011, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.144899584923064, | |
| "grad_norm": 1.327438350761422, | |
| "learning_rate": 9.670621810259596e-06, | |
| "loss": 1.0187, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.14624124774642572, | |
| "grad_norm": 1.326527294225068, | |
| "learning_rate": 9.662822663483952e-06, | |
| "loss": 1.0308, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.14758291056978742, | |
| "grad_norm": 1.8118538857152473, | |
| "learning_rate": 9.654935478465409e-06, | |
| "loss": 1.0047, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.14892457339314913, | |
| "grad_norm": 1.3283480891622743, | |
| "learning_rate": 9.646960404121042e-06, | |
| "loss": 1.0242, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.15026623621651083, | |
| "grad_norm": 1.190018480272337, | |
| "learning_rate": 9.638897591027355e-06, | |
| "loss": 1.0183, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.15160789903987254, | |
| "grad_norm": 1.588911478134852, | |
| "learning_rate": 9.630747191417433e-06, | |
| "loss": 1.0029, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.15294956186323425, | |
| "grad_norm": 1.4284241067617203, | |
| "learning_rate": 9.62250935917808e-06, | |
| "loss": 1.0171, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.15429122468659595, | |
| "grad_norm": 1.3440365747872154, | |
| "learning_rate": 9.614184249846903e-06, | |
| "loss": 1.0039, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.15563288750995766, | |
| "grad_norm": 1.472551706358172, | |
| "learning_rate": 9.605772020609376e-06, | |
| "loss": 1.029, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.15697455033331936, | |
| "grad_norm": 1.5434965270394372, | |
| "learning_rate": 9.597272830295877e-06, | |
| "loss": 1.0155, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.15831621315668107, | |
| "grad_norm": 1.7903643075127016, | |
| "learning_rate": 9.58868683937868e-06, | |
| "loss": 1.027, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.15965787598004277, | |
| "grad_norm": 1.5872313617712013, | |
| "learning_rate": 9.580014209968946e-06, | |
| "loss": 0.99, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.16099953880340448, | |
| "grad_norm": 1.33797132946658, | |
| "learning_rate": 9.571255105813632e-06, | |
| "loss": 1.0291, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.16234120162676619, | |
| "grad_norm": 1.3124608346564857, | |
| "learning_rate": 9.562409692292425e-06, | |
| "loss": 1.0303, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.1636828644501279, | |
| "grad_norm": 1.5149230952048165, | |
| "learning_rate": 9.553478136414606e-06, | |
| "loss": 0.9977, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.16502452727348957, | |
| "grad_norm": 1.6211734357223275, | |
| "learning_rate": 9.544460606815901e-06, | |
| "loss": 1.0082, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.16636619009685127, | |
| "grad_norm": 1.3021579126793668, | |
| "learning_rate": 9.535357273755296e-06, | |
| "loss": 1.0091, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.16770785292021298, | |
| "grad_norm": 1.3433784611947364, | |
| "learning_rate": 9.526168309111827e-06, | |
| "loss": 1.0185, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.16904951574357469, | |
| "grad_norm": 1.1341044588473663, | |
| "learning_rate": 9.516893886381324e-06, | |
| "loss": 0.9869, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.1703911785669364, | |
| "grad_norm": 1.2409232333468363, | |
| "learning_rate": 9.507534180673142e-06, | |
| "loss": 0.9972, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.1717328413902981, | |
| "grad_norm": 1.5598206606390372, | |
| "learning_rate": 9.498089368706862e-06, | |
| "loss": 0.999, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.1730745042136598, | |
| "grad_norm": 1.3223682060882507, | |
| "learning_rate": 9.488559628808939e-06, | |
| "loss": 0.9959, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.1744161670370215, | |
| "grad_norm": 1.2205385542407121, | |
| "learning_rate": 9.478945140909346e-06, | |
| "loss": 0.9827, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.1757578298603832, | |
| "grad_norm": 3.2075255695633698, | |
| "learning_rate": 9.469246086538175e-06, | |
| "loss": 1.0185, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.17709949268374492, | |
| "grad_norm": 1.3618166834541443, | |
| "learning_rate": 9.459462648822209e-06, | |
| "loss": 0.9998, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.17844115550710662, | |
| "grad_norm": 1.2950977689687375, | |
| "learning_rate": 9.449595012481455e-06, | |
| "loss": 1.0291, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.17978281833046833, | |
| "grad_norm": 1.9976980984805643, | |
| "learning_rate": 9.439643363825678e-06, | |
| "loss": 1.0147, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.18112448115383004, | |
| "grad_norm": 1.31489503517818, | |
| "learning_rate": 9.429607890750863e-06, | |
| "loss": 0.9964, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.18246614397719174, | |
| "grad_norm": 1.4436072988344344, | |
| "learning_rate": 9.419488782735671e-06, | |
| "loss": 1.0051, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.18380780680055345, | |
| "grad_norm": 1.3666027699047631, | |
| "learning_rate": 9.409286230837876e-06, | |
| "loss": 0.9503, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.18514946962391515, | |
| "grad_norm": 1.2477148872031183, | |
| "learning_rate": 9.399000427690736e-06, | |
| "loss": 0.9794, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.18649113244727683, | |
| "grad_norm": 1.5334999354972565, | |
| "learning_rate": 9.388631567499373e-06, | |
| "loss": 0.9923, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.18783279527063854, | |
| "grad_norm": 1.2107172959297416, | |
| "learning_rate": 9.378179846037096e-06, | |
| "loss": 0.9756, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.18917445809400024, | |
| "grad_norm": 1.6910026865869627, | |
| "learning_rate": 9.367645460641716e-06, | |
| "loss": 0.9793, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.19051612091736195, | |
| "grad_norm": 1.4064299287751443, | |
| "learning_rate": 9.357028610211802e-06, | |
| "loss": 0.9817, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.19185778374072365, | |
| "grad_norm": 1.166569364141333, | |
| "learning_rate": 9.346329495202944e-06, | |
| "loss": 0.9658, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.19319944656408536, | |
| "grad_norm": 1.2965118473712378, | |
| "learning_rate": 9.335548317623957e-06, | |
| "loss": 0.966, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.19454110938744706, | |
| "grad_norm": 1.3276615380747308, | |
| "learning_rate": 9.324685281033073e-06, | |
| "loss": 0.9726, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.19588277221080877, | |
| "grad_norm": 2.7946871090039465, | |
| "learning_rate": 9.313740590534093e-06, | |
| "loss": 0.9826, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.19722443503417048, | |
| "grad_norm": 1.5746974126958582, | |
| "learning_rate": 9.302714452772515e-06, | |
| "loss": 0.9967, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.19856609785753218, | |
| "grad_norm": 2.018573996099897, | |
| "learning_rate": 9.291607075931638e-06, | |
| "loss": 0.9841, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.1999077606808939, | |
| "grad_norm": 1.596645681206608, | |
| "learning_rate": 9.280418669728625e-06, | |
| "loss": 0.9542, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.2012494235042556, | |
| "grad_norm": 1.5021142172270334, | |
| "learning_rate": 9.269149445410545e-06, | |
| "loss": 0.973, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.2025910863276173, | |
| "grad_norm": 1.7243659160172629, | |
| "learning_rate": 9.257799615750384e-06, | |
| "loss": 0.9893, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.203932749150979, | |
| "grad_norm": 1.2920298384903153, | |
| "learning_rate": 9.246369395043033e-06, | |
| "loss": 0.9688, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.2052744119743407, | |
| "grad_norm": 1.6060902706519415, | |
| "learning_rate": 9.234858999101232e-06, | |
| "loss": 0.9729, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.20661607479770241, | |
| "grad_norm": 1.5359620007536603, | |
| "learning_rate": 9.223268645251504e-06, | |
| "loss": 0.9526, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.2079577376210641, | |
| "grad_norm": 1.5626700176309778, | |
| "learning_rate": 9.211598552330047e-06, | |
| "loss": 0.9872, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.2092994004444258, | |
| "grad_norm": 2.310007223880514, | |
| "learning_rate": 9.199848940678607e-06, | |
| "loss": 0.9638, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.2106410632677875, | |
| "grad_norm": 1.4873479328142536, | |
| "learning_rate": 9.188020032140308e-06, | |
| "loss": 0.9468, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.2119827260911492, | |
| "grad_norm": 2.814750480190035, | |
| "learning_rate": 9.17611205005548e-06, | |
| "loss": 0.9751, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.21332438891451092, | |
| "grad_norm": 1.384467947659544, | |
| "learning_rate": 9.164125219257419e-06, | |
| "loss": 0.9808, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.21466605173787262, | |
| "grad_norm": 1.3043220125773678, | |
| "learning_rate": 9.152059766068164e-06, | |
| "loss": 0.9891, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.21600771456123433, | |
| "grad_norm": 1.4429178434101435, | |
| "learning_rate": 9.139915918294213e-06, | |
| "loss": 0.9711, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.21734937738459603, | |
| "grad_norm": 1.379789634080121, | |
| "learning_rate": 9.127693905222223e-06, | |
| "loss": 0.9536, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.21869104020795774, | |
| "grad_norm": 1.2124592786433257, | |
| "learning_rate": 9.115393957614679e-06, | |
| "loss": 0.9667, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.22003270303131944, | |
| "grad_norm": 1.2828638178666, | |
| "learning_rate": 9.103016307705546e-06, | |
| "loss": 0.9908, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.22137436585468115, | |
| "grad_norm": 1.4214695125865153, | |
| "learning_rate": 9.09056118919587e-06, | |
| "loss": 0.9846, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.22271602867804285, | |
| "grad_norm": 1.1858223727188744, | |
| "learning_rate": 9.078028837249378e-06, | |
| "loss": 0.9632, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.22405769150140456, | |
| "grad_norm": 1.4085589772907592, | |
| "learning_rate": 9.065419488488029e-06, | |
| "loss": 0.9797, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.22539935432476627, | |
| "grad_norm": 1.251837746674831, | |
| "learning_rate": 9.052733380987555e-06, | |
| "loss": 0.9707, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.22674101714812797, | |
| "grad_norm": 1.3819999218217576, | |
| "learning_rate": 9.03997075427296e-06, | |
| "loss": 0.9679, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.22808267997148968, | |
| "grad_norm": 1.3099387375957983, | |
| "learning_rate": 9.027131849313995e-06, | |
| "loss": 0.9351, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.22942434279485138, | |
| "grad_norm": 1.4657996541663856, | |
| "learning_rate": 9.014216908520619e-06, | |
| "loss": 0.9625, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.23076600561821306, | |
| "grad_norm": 1.3466725984552, | |
| "learning_rate": 9.001226175738409e-06, | |
| "loss": 0.9554, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.23210766844157477, | |
| "grad_norm": 1.4165754552886125, | |
| "learning_rate": 8.988159896243967e-06, | |
| "loss": 0.977, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.23344933126493647, | |
| "grad_norm": 1.1408872105860128, | |
| "learning_rate": 8.975018316740278e-06, | |
| "loss": 0.9267, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.23479099408829818, | |
| "grad_norm": 1.3586993829431193, | |
| "learning_rate": 8.961801685352068e-06, | |
| "loss": 0.9571, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.23613265691165988, | |
| "grad_norm": 1.4241945569547994, | |
| "learning_rate": 8.948510251621096e-06, | |
| "loss": 0.9542, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.2374743197350216, | |
| "grad_norm": 1.3389398198577172, | |
| "learning_rate": 8.93514426650147e-06, | |
| "loss": 0.9381, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.2388159825583833, | |
| "grad_norm": 1.2225771380988635, | |
| "learning_rate": 8.921703982354886e-06, | |
| "loss": 0.964, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.240157645381745, | |
| "grad_norm": 3.592083099802136, | |
| "learning_rate": 8.908189652945871e-06, | |
| "loss": 0.9713, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.2414993082051067, | |
| "grad_norm": 2.024932112092753, | |
| "learning_rate": 8.894601533437e-06, | |
| "loss": 0.9377, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2428409710284684, | |
| "grad_norm": 1.7579130845607733, | |
| "learning_rate": 8.880939880384061e-06, | |
| "loss": 0.9618, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.24418263385183012, | |
| "grad_norm": 1.627410999442911, | |
| "learning_rate": 8.867204951731227e-06, | |
| "loss": 0.9635, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.24552429667519182, | |
| "grad_norm": 1.343115545097064, | |
| "learning_rate": 8.853397006806183e-06, | |
| "loss": 0.9241, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.24686595949855353, | |
| "grad_norm": 1.396305991368914, | |
| "learning_rate": 8.839516306315216e-06, | |
| "loss": 0.9419, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.24820762232191523, | |
| "grad_norm": 1.2089978583794718, | |
| "learning_rate": 8.825563112338308e-06, | |
| "loss": 0.9563, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.24954928514527694, | |
| "grad_norm": 1.7121754682040942, | |
| "learning_rate": 8.811537688324187e-06, | |
| "loss": 0.9508, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.25089094796863864, | |
| "grad_norm": 1.5847812839177484, | |
| "learning_rate": 8.797440299085344e-06, | |
| "loss": 0.954, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.2522326107920003, | |
| "grad_norm": 1.2931035656489251, | |
| "learning_rate": 8.783271210793034e-06, | |
| "loss": 0.9333, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.25357427361536206, | |
| "grad_norm": 1.400853466504507, | |
| "learning_rate": 8.769030690972262e-06, | |
| "loss": 0.9377, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.25491593643872373, | |
| "grad_norm": 1.4983501911170125, | |
| "learning_rate": 8.754719008496718e-06, | |
| "loss": 0.946, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.25625759926208547, | |
| "grad_norm": 1.9904680648040076, | |
| "learning_rate": 8.740336433583704e-06, | |
| "loss": 0.9496, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.25759926208544714, | |
| "grad_norm": 1.2709529254120855, | |
| "learning_rate": 8.725883237789046e-06, | |
| "loss": 0.9522, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.2589409249088089, | |
| "grad_norm": 1.4655810614703586, | |
| "learning_rate": 8.711359694001941e-06, | |
| "loss": 0.9412, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.26028258773217056, | |
| "grad_norm": 1.448365318384477, | |
| "learning_rate": 8.696766076439826e-06, | |
| "loss": 0.9422, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.2616242505555323, | |
| "grad_norm": 1.386638899065457, | |
| "learning_rate": 8.682102660643196e-06, | |
| "loss": 0.9764, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.26296591337889397, | |
| "grad_norm": 1.3761633842105907, | |
| "learning_rate": 8.667369723470393e-06, | |
| "loss": 0.9529, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.26430757620225565, | |
| "grad_norm": 1.451176717632925, | |
| "learning_rate": 8.652567543092385e-06, | |
| "loss": 0.9371, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.2656492390256174, | |
| "grad_norm": 1.4956859752239118, | |
| "learning_rate": 8.637696398987517e-06, | |
| "loss": 0.9349, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.26699090184897906, | |
| "grad_norm": 1.4970834985646684, | |
| "learning_rate": 8.622756571936229e-06, | |
| "loss": 0.9315, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.2683325646723408, | |
| "grad_norm": 1.3728792118456619, | |
| "learning_rate": 8.607748344015752e-06, | |
| "loss": 0.9278, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.26967422749570247, | |
| "grad_norm": 3.1693510901906183, | |
| "learning_rate": 8.592671998594794e-06, | |
| "loss": 0.94, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.2710158903190642, | |
| "grad_norm": 1.3498967145465, | |
| "learning_rate": 8.577527820328176e-06, | |
| "loss": 0.9244, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.2723575531424259, | |
| "grad_norm": 1.2466378588094895, | |
| "learning_rate": 8.562316095151468e-06, | |
| "loss": 0.9139, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.2736992159657876, | |
| "grad_norm": 1.2940595645923652, | |
| "learning_rate": 8.54703711027558e-06, | |
| "loss": 0.942, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.2750408787891493, | |
| "grad_norm": 1.2657171818696131, | |
| "learning_rate": 8.53169115418135e-06, | |
| "loss": 0.9243, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.276382541612511, | |
| "grad_norm": 1.5218389065576579, | |
| "learning_rate": 8.516278516614092e-06, | |
| "loss": 0.9304, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.2777242044358727, | |
| "grad_norm": 1.691480480045371, | |
| "learning_rate": 8.50079948857812e-06, | |
| "loss": 0.9268, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.27906586725923443, | |
| "grad_norm": 1.177135449105845, | |
| "learning_rate": 8.485254362331268e-06, | |
| "loss": 0.9426, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.2804075300825961, | |
| "grad_norm": 1.2706094678412008, | |
| "learning_rate": 8.469643431379355e-06, | |
| "loss": 0.9244, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.28174919290595785, | |
| "grad_norm": 1.2772319095564861, | |
| "learning_rate": 8.453966990470656e-06, | |
| "loss": 0.9281, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.2830908557293195, | |
| "grad_norm": 1.4563823965413645, | |
| "learning_rate": 8.438225335590334e-06, | |
| "loss": 0.9319, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.2844325185526812, | |
| "grad_norm": 1.3936683454684986, | |
| "learning_rate": 8.422418763954841e-06, | |
| "loss": 0.9194, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.28577418137604294, | |
| "grad_norm": 1.3349300657889855, | |
| "learning_rate": 8.406547574006326e-06, | |
| "loss": 0.9306, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.2871158441994046, | |
| "grad_norm": 1.1836954924074992, | |
| "learning_rate": 8.390612065406983e-06, | |
| "loss": 0.9157, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.28845750702276635, | |
| "grad_norm": 1.878294298469528, | |
| "learning_rate": 8.374612539033398e-06, | |
| "loss": 0.9335, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.289799169846128, | |
| "grad_norm": 1.4926770055762641, | |
| "learning_rate": 8.358549296970877e-06, | |
| "loss": 0.925, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.29114083266948976, | |
| "grad_norm": 1.2008838531823012, | |
| "learning_rate": 8.342422642507727e-06, | |
| "loss": 0.9422, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.29248249549285144, | |
| "grad_norm": 1.6229025627352722, | |
| "learning_rate": 8.32623288012954e-06, | |
| "loss": 0.9153, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.29382415831621317, | |
| "grad_norm": 1.448744835139555, | |
| "learning_rate": 8.309980315513444e-06, | |
| "loss": 0.928, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.29516582113957485, | |
| "grad_norm": 1.2474176482179575, | |
| "learning_rate": 8.29366525552233e-06, | |
| "loss": 0.9191, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.2965074839629366, | |
| "grad_norm": 1.3403479118534118, | |
| "learning_rate": 8.27728800819905e-06, | |
| "loss": 0.9255, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.29784914678629826, | |
| "grad_norm": 1.3943086416749382, | |
| "learning_rate": 8.260848882760616e-06, | |
| "loss": 0.9342, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.29919080960966, | |
| "grad_norm": 1.3055916107751189, | |
| "learning_rate": 8.24434818959235e-06, | |
| "loss": 0.9035, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.30053247243302167, | |
| "grad_norm": 1.3327493068279272, | |
| "learning_rate": 8.22778624024203e-06, | |
| "loss": 0.9261, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.3018741352563834, | |
| "grad_norm": 1.239307116444798, | |
| "learning_rate": 8.211163347414005e-06, | |
| "loss": 0.9257, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.3032157980797451, | |
| "grad_norm": 1.343985973623681, | |
| "learning_rate": 8.194479824963284e-06, | |
| "loss": 0.9225, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.3045574609031068, | |
| "grad_norm": 1.4377788560631601, | |
| "learning_rate": 8.177735987889628e-06, | |
| "loss": 0.9171, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.3058991237264685, | |
| "grad_norm": 1.592158031770944, | |
| "learning_rate": 8.160932152331587e-06, | |
| "loss": 0.9237, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.30724078654983017, | |
| "grad_norm": 2.265440536226128, | |
| "learning_rate": 8.144068635560533e-06, | |
| "loss": 0.9438, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.3085824493731919, | |
| "grad_norm": 1.2380411169498318, | |
| "learning_rate": 8.127145755974676e-06, | |
| "loss": 0.9196, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.3099241121965536, | |
| "grad_norm": 1.1816309699887972, | |
| "learning_rate": 8.11016383309305e-06, | |
| "loss": 0.9183, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.3112657750199153, | |
| "grad_norm": 1.4146315793501878, | |
| "learning_rate": 8.093123187549475e-06, | |
| "loss": 0.9292, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.312607437843277, | |
| "grad_norm": 1.560402614336781, | |
| "learning_rate": 8.07602414108651e-06, | |
| "loss": 0.9231, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.3139491006666387, | |
| "grad_norm": 1.3309792388882167, | |
| "learning_rate": 8.058867016549372e-06, | |
| "loss": 0.9266, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.3152907634900004, | |
| "grad_norm": 1.4332599646276138, | |
| "learning_rate": 8.04165213787985e-06, | |
| "loss": 0.9063, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.31663242631336214, | |
| "grad_norm": 1.203035656577196, | |
| "learning_rate": 8.024379830110176e-06, | |
| "loss": 0.8955, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.3179740891367238, | |
| "grad_norm": 1.5292641254784858, | |
| "learning_rate": 8.007050419356898e-06, | |
| "loss": 0.9173, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.31931575196008555, | |
| "grad_norm": 1.2177503203754507, | |
| "learning_rate": 7.989664232814718e-06, | |
| "loss": 0.9234, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.3206574147834472, | |
| "grad_norm": 1.3349700755826033, | |
| "learning_rate": 7.972221598750316e-06, | |
| "loss": 0.9064, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.32199907760680896, | |
| "grad_norm": 1.2473177750073925, | |
| "learning_rate": 7.95472284649615e-06, | |
| "loss": 0.9159, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.32334074043017064, | |
| "grad_norm": 1.3319677097934364, | |
| "learning_rate": 7.937168306444243e-06, | |
| "loss": 0.9067, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.32468240325353237, | |
| "grad_norm": 1.3185633256146132, | |
| "learning_rate": 7.919558310039937e-06, | |
| "loss": 0.9135, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.32602406607689405, | |
| "grad_norm": 1.380952623177733, | |
| "learning_rate": 7.90189318977564e-06, | |
| "loss": 0.9171, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.3273657289002558, | |
| "grad_norm": 1.608074562500768, | |
| "learning_rate": 7.884173279184552e-06, | |
| "loss": 0.9294, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.32870739172361746, | |
| "grad_norm": 1.278354698668591, | |
| "learning_rate": 7.866398912834357e-06, | |
| "loss": 0.9314, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.33004905454697914, | |
| "grad_norm": 1.3727186644321794, | |
| "learning_rate": 7.848570426320918e-06, | |
| "loss": 0.9162, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.33139071737034087, | |
| "grad_norm": 1.3318489778585914, | |
| "learning_rate": 7.830688156261927e-06, | |
| "loss": 0.9299, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.33273238019370255, | |
| "grad_norm": 1.2673342688345994, | |
| "learning_rate": 7.812752440290568e-06, | |
| "loss": 0.9007, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.3340740430170643, | |
| "grad_norm": 1.389895836548381, | |
| "learning_rate": 7.794763617049124e-06, | |
| "loss": 0.9088, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.33541570584042596, | |
| "grad_norm": 2.430750588903623, | |
| "learning_rate": 7.776722026182588e-06, | |
| "loss": 0.9241, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.3367573686637877, | |
| "grad_norm": 1.4249817709115309, | |
| "learning_rate": 7.758628008332262e-06, | |
| "loss": 0.9162, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.33809903148714937, | |
| "grad_norm": 1.1765685406780577, | |
| "learning_rate": 7.740481905129307e-06, | |
| "loss": 0.9104, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.3394406943105111, | |
| "grad_norm": 1.7458809897309366, | |
| "learning_rate": 7.722284059188308e-06, | |
| "loss": 0.8988, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.3407823571338728, | |
| "grad_norm": 1.3333600860522707, | |
| "learning_rate": 7.704034814100798e-06, | |
| "loss": 0.8931, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.3421240199572345, | |
| "grad_norm": 1.3898670538614775, | |
| "learning_rate": 7.685734514428767e-06, | |
| "loss": 0.9183, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.3434656827805962, | |
| "grad_norm": 1.3074770902798807, | |
| "learning_rate": 7.667383505698165e-06, | |
| "loss": 0.905, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.3448073456039579, | |
| "grad_norm": 1.4239766274564487, | |
| "learning_rate": 7.648982134392378e-06, | |
| "loss": 0.8981, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.3461490084273196, | |
| "grad_norm": 1.3976199380057532, | |
| "learning_rate": 7.630530747945672e-06, | |
| "loss": 0.8949, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.34749067125068134, | |
| "grad_norm": 1.415952263830166, | |
| "learning_rate": 7.612029694736652e-06, | |
| "loss": 0.8991, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.348832334074043, | |
| "grad_norm": 1.305041268758615, | |
| "learning_rate": 7.593479324081672e-06, | |
| "loss": 0.9239, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.3501739968974047, | |
| "grad_norm": 1.368157415364355, | |
| "learning_rate": 7.574879986228245e-06, | |
| "loss": 0.9053, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.3515156597207664, | |
| "grad_norm": 1.2265496173880226, | |
| "learning_rate": 7.556232032348429e-06, | |
| "loss": 0.9077, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.3528573225441281, | |
| "grad_norm": 1.4565310656724473, | |
| "learning_rate": 7.537535814532191e-06, | |
| "loss": 0.9086, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.35419898536748984, | |
| "grad_norm": 1.2046633053522189, | |
| "learning_rate": 7.518791685780769e-06, | |
| "loss": 0.8923, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.3555406481908515, | |
| "grad_norm": 1.2983155786441998, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.9132, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.35688231101421325, | |
| "grad_norm": 1.3513912481629218, | |
| "learning_rate": 7.481161111993642e-06, | |
| "loss": 0.9067, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.3582239738375749, | |
| "grad_norm": 1.4118077302428045, | |
| "learning_rate": 7.462275377456671e-06, | |
| "loss": 0.9017, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.35956563666093666, | |
| "grad_norm": 2.203736423265122, | |
| "learning_rate": 7.443343152968566e-06, | |
| "loss": 0.921, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.36090729948429834, | |
| "grad_norm": 1.4719213378371314, | |
| "learning_rate": 7.424364795986582e-06, | |
| "loss": 0.9042, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.36224896230766007, | |
| "grad_norm": 1.1575210226646582, | |
| "learning_rate": 7.405340664838994e-06, | |
| "loss": 0.8961, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.36359062513102175, | |
| "grad_norm": 1.1822974768717973, | |
| "learning_rate": 7.38627111871833e-06, | |
| "loss": 0.8995, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.3649322879543835, | |
| "grad_norm": 1.298422404346634, | |
| "learning_rate": 7.3671565176746025e-06, | |
| "loss": 0.9161, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.36627395077774516, | |
| "grad_norm": 1.3013885285130078, | |
| "learning_rate": 7.3479972226084925e-06, | |
| "loss": 0.9022, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.3676156136011069, | |
| "grad_norm": 1.2093347401644514, | |
| "learning_rate": 7.328793595264549e-06, | |
| "loss": 0.8891, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.36895727642446857, | |
| "grad_norm": 1.3034407062657263, | |
| "learning_rate": 7.309545998224351e-06, | |
| "loss": 0.8919, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.3702989392478303, | |
| "grad_norm": 1.1837872314851003, | |
| "learning_rate": 7.290254794899665e-06, | |
| "loss": 0.9094, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.371640602071192, | |
| "grad_norm": 1.47553224760993, | |
| "learning_rate": 7.270920349525584e-06, | |
| "loss": 0.9087, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.37298226489455366, | |
| "grad_norm": 1.3813251143214236, | |
| "learning_rate": 7.2515430271536495e-06, | |
| "loss": 0.89, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.3743239277179154, | |
| "grad_norm": 1.1890516020802695, | |
| "learning_rate": 7.232123193644957e-06, | |
| "loss": 0.9003, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.3756655905412771, | |
| "grad_norm": 1.482711276179717, | |
| "learning_rate": 7.212661215663252e-06, | |
| "loss": 0.8863, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.3770072533646388, | |
| "grad_norm": 2.470699519286606, | |
| "learning_rate": 7.193157460668005e-06, | |
| "loss": 0.8826, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.3783489161880005, | |
| "grad_norm": 1.2722870459171294, | |
| "learning_rate": 7.173612296907473e-06, | |
| "loss": 0.8938, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.3796905790113622, | |
| "grad_norm": 1.164770607468211, | |
| "learning_rate": 7.154026093411747e-06, | |
| "loss": 0.9056, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.3810322418347239, | |
| "grad_norm": 1.3066601864771532, | |
| "learning_rate": 7.134399219985786e-06, | |
| "loss": 0.9101, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.38237390465808563, | |
| "grad_norm": 1.2226710990522875, | |
| "learning_rate": 7.114732047202433e-06, | |
| "loss": 0.9036, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.3837155674814473, | |
| "grad_norm": 1.3558186731284418, | |
| "learning_rate": 7.095024946395419e-06, | |
| "loss": 0.8971, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.38505723030480904, | |
| "grad_norm": 1.3887581493685603, | |
| "learning_rate": 7.075278289652349e-06, | |
| "loss": 0.8977, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.3863988931281707, | |
| "grad_norm": 1.259821804021751, | |
| "learning_rate": 7.055492449807684e-06, | |
| "loss": 0.8793, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.38774055595153245, | |
| "grad_norm": 1.2855154210616317, | |
| "learning_rate": 7.035667800435694e-06, | |
| "loss": 0.8873, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.38908221877489413, | |
| "grad_norm": 1.3339572607637207, | |
| "learning_rate": 7.015804715843404e-06, | |
| "loss": 0.8895, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.39042388159825586, | |
| "grad_norm": 1.246283126601731, | |
| "learning_rate": 6.995903571063541e-06, | |
| "loss": 0.8918, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.39176554442161754, | |
| "grad_norm": 4.261409025409612, | |
| "learning_rate": 6.975964741847427e-06, | |
| "loss": 0.8983, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.3931072072449793, | |
| "grad_norm": 1.4742666681843175, | |
| "learning_rate": 6.955988604657914e-06, | |
| "loss": 0.8833, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.39444887006834095, | |
| "grad_norm": 1.2263954992920674, | |
| "learning_rate": 6.935975536662254e-06, | |
| "loss": 0.8831, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.39579053289170263, | |
| "grad_norm": 1.3477286754599664, | |
| "learning_rate": 6.915925915724987e-06, | |
| "loss": 0.8707, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.39713219571506436, | |
| "grad_norm": 1.1562356495785004, | |
| "learning_rate": 6.895840120400804e-06, | |
| "loss": 0.8756, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.39847385853842604, | |
| "grad_norm": 1.3759724276042458, | |
| "learning_rate": 6.875718529927404e-06, | |
| "loss": 0.8891, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.3998155213617878, | |
| "grad_norm": 1.3312094291987537, | |
| "learning_rate": 6.855561524218326e-06, | |
| "loss": 0.9229, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.40115718418514945, | |
| "grad_norm": 1.3348406885744244, | |
| "learning_rate": 6.835369483855783e-06, | |
| "loss": 0.86, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.4024988470085112, | |
| "grad_norm": 1.6527502121422286, | |
| "learning_rate": 6.815142790083473e-06, | |
| "loss": 0.894, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.40384050983187286, | |
| "grad_norm": 1.2963282343671396, | |
| "learning_rate": 6.79488182479938e-06, | |
| "loss": 0.9043, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.4051821726552346, | |
| "grad_norm": 1.2677450598520326, | |
| "learning_rate": 6.774586970548567e-06, | |
| "loss": 0.889, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.4065238354785963, | |
| "grad_norm": 1.517217039038415, | |
| "learning_rate": 6.754258610515949e-06, | |
| "loss": 0.8884, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.407865498301958, | |
| "grad_norm": 1.2413393961206562, | |
| "learning_rate": 6.73389712851906e-06, | |
| "loss": 0.8808, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.4092071611253197, | |
| "grad_norm": 1.4670436869599548, | |
| "learning_rate": 6.713502909000808e-06, | |
| "loss": 0.8642, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.4105488239486814, | |
| "grad_norm": 1.958446865127943, | |
| "learning_rate": 6.6930763370222104e-06, | |
| "loss": 0.9101, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.4118904867720431, | |
| "grad_norm": 1.2369628912481843, | |
| "learning_rate": 6.672617798255135e-06, | |
| "loss": 0.8834, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.41323214959540483, | |
| "grad_norm": 1.5043979563533736, | |
| "learning_rate": 6.652127678975002e-06, | |
| "loss": 0.8794, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.4145738124187665, | |
| "grad_norm": 1.4127332040148803, | |
| "learning_rate": 6.631606366053507e-06, | |
| "loss": 0.8861, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.4159154752421282, | |
| "grad_norm": 1.4168221736798758, | |
| "learning_rate": 6.6110542469513065e-06, | |
| "loss": 0.8786, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.4172571380654899, | |
| "grad_norm": 1.4409097127048691, | |
| "learning_rate": 6.590471709710704e-06, | |
| "loss": 0.884, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.4185988008888516, | |
| "grad_norm": 1.455604144250745, | |
| "learning_rate": 6.5698591429483286e-06, | |
| "loss": 0.8894, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.41994046371221333, | |
| "grad_norm": 1.2125504415902777, | |
| "learning_rate": 6.549216935847786e-06, | |
| "loss": 0.8735, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.421282126535575, | |
| "grad_norm": 1.3309515557805753, | |
| "learning_rate": 6.52854547815233e-06, | |
| "loss": 0.8666, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.42262378935893674, | |
| "grad_norm": 1.1540064107186283, | |
| "learning_rate": 6.507845160157476e-06, | |
| "loss": 0.8883, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.4239654521822984, | |
| "grad_norm": 1.2992399441092857, | |
| "learning_rate": 6.487116372703663e-06, | |
| "loss": 0.8889, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.42530711500566015, | |
| "grad_norm": 1.2505578379186895, | |
| "learning_rate": 6.466359507168849e-06, | |
| "loss": 0.8812, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.42664877782902183, | |
| "grad_norm": 1.4444565423946591, | |
| "learning_rate": 6.445574955461134e-06, | |
| "loss": 0.874, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.42799044065238356, | |
| "grad_norm": 1.2838136246138698, | |
| "learning_rate": 6.424763110011356e-06, | |
| "loss": 0.8904, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.42933210347574524, | |
| "grad_norm": 1.2677175813694, | |
| "learning_rate": 6.403924363765686e-06, | |
| "loss": 0.8828, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.430673766299107, | |
| "grad_norm": 1.2152928950493196, | |
| "learning_rate": 6.383059110178205e-06, | |
| "loss": 0.8744, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.43201542912246865, | |
| "grad_norm": 1.2584889553887393, | |
| "learning_rate": 6.362167743203474e-06, | |
| "loss": 0.8799, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.4333570919458304, | |
| "grad_norm": 1.2086440211957121, | |
| "learning_rate": 6.3412506572891e-06, | |
| "loss": 0.8882, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.43469875476919206, | |
| "grad_norm": 1.223278471633005, | |
| "learning_rate": 6.320308247368285e-06, | |
| "loss": 0.8703, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.4360404175925538, | |
| "grad_norm": 1.3424604902690052, | |
| "learning_rate": 6.2993409088523774e-06, | |
| "loss": 0.8921, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.4373820804159155, | |
| "grad_norm": 1.5268556690481863, | |
| "learning_rate": 6.278349037623388e-06, | |
| "loss": 0.875, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.43872374323927715, | |
| "grad_norm": 1.4120251100515266, | |
| "learning_rate": 6.2573330300265375e-06, | |
| "loss": 0.8956, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.4400654060626389, | |
| "grad_norm": 1.2646280955519513, | |
| "learning_rate": 6.236293282862756e-06, | |
| "loss": 0.885, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.44140706888600056, | |
| "grad_norm": 1.3687254991175013, | |
| "learning_rate": 6.215230193381203e-06, | |
| "loss": 0.8563, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.4427487317093623, | |
| "grad_norm": 1.2098687265389088, | |
| "learning_rate": 6.1941441592717564e-06, | |
| "loss": 0.8804, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.444090394532724, | |
| "grad_norm": 1.1748518654586857, | |
| "learning_rate": 6.173035578657513e-06, | |
| "loss": 0.8687, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.4454320573560857, | |
| "grad_norm": 1.3305767093494774, | |
| "learning_rate": 6.151904850087265e-06, | |
| "loss": 0.8805, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.4467737201794474, | |
| "grad_norm": 1.3352047002541647, | |
| "learning_rate": 6.130752372527981e-06, | |
| "loss": 0.8625, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.4481153830028091, | |
| "grad_norm": 1.4111280757375295, | |
| "learning_rate": 6.109578545357268e-06, | |
| "loss": 0.8858, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.4494570458261708, | |
| "grad_norm": 1.6323344225901677, | |
| "learning_rate": 6.088383768355829e-06, | |
| "loss": 0.8793, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.45079870864953253, | |
| "grad_norm": 1.338640903780984, | |
| "learning_rate": 6.067168441699927e-06, | |
| "loss": 0.8997, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.4521403714728942, | |
| "grad_norm": 1.2958085754511701, | |
| "learning_rate": 6.045932965953813e-06, | |
| "loss": 0.8751, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.45348203429625594, | |
| "grad_norm": 1.1674177353533752, | |
| "learning_rate": 6.024677742062176e-06, | |
| "loss": 0.8775, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.4548236971196176, | |
| "grad_norm": 1.4279216502945793, | |
| "learning_rate": 6.0034031713425636e-06, | |
| "loss": 0.879, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.45616535994297935, | |
| "grad_norm": 1.339345592536122, | |
| "learning_rate": 5.98210965547781e-06, | |
| "loss": 0.9082, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.45750702276634103, | |
| "grad_norm": 2.2820697383982416, | |
| "learning_rate": 5.960797596508453e-06, | |
| "loss": 0.8876, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.45884868558970276, | |
| "grad_norm": 1.410625566285379, | |
| "learning_rate": 5.939467396825137e-06, | |
| "loss": 0.8763, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.46019034841306444, | |
| "grad_norm": 1.3003949398928862, | |
| "learning_rate": 5.918119459161021e-06, | |
| "loss": 0.8984, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.4615320112364261, | |
| "grad_norm": 2.5162326088544202, | |
| "learning_rate": 5.896754186584173e-06, | |
| "loss": 0.8772, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.46287367405978785, | |
| "grad_norm": 1.4219922225059423, | |
| "learning_rate": 5.875371982489959e-06, | |
| "loss": 0.8738, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.46421533688314953, | |
| "grad_norm": 1.2261327989564115, | |
| "learning_rate": 5.8539732505934266e-06, | |
| "loss": 0.8534, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.46555699970651127, | |
| "grad_norm": 1.6516399440899023, | |
| "learning_rate": 5.832558394921688e-06, | |
| "loss": 0.8559, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.46689866252987294, | |
| "grad_norm": 1.3852991155430008, | |
| "learning_rate": 5.811127819806277e-06, | |
| "loss": 0.8657, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.4682403253532347, | |
| "grad_norm": 1.1404258494095492, | |
| "learning_rate": 5.7896819298755346e-06, | |
| "loss": 0.8752, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.46958198817659635, | |
| "grad_norm": 1.4748589097551792, | |
| "learning_rate": 5.768221130046954e-06, | |
| "loss": 0.8555, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.4709236509999581, | |
| "grad_norm": 1.2585550720424548, | |
| "learning_rate": 5.746745825519539e-06, | |
| "loss": 0.8676, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.47226531382331977, | |
| "grad_norm": 1.3861095318721872, | |
| "learning_rate": 5.725256421766158e-06, | |
| "loss": 0.8684, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.4736069766466815, | |
| "grad_norm": 1.3577534335228219, | |
| "learning_rate": 5.703753324525885e-06, | |
| "loss": 0.8662, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.4749486394700432, | |
| "grad_norm": 1.337863080495238, | |
| "learning_rate": 5.682236939796337e-06, | |
| "loss": 0.8625, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.4762903022934049, | |
| "grad_norm": 1.161397595087837, | |
| "learning_rate": 5.660707673826016e-06, | |
| "loss": 0.8541, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.4776319651167666, | |
| "grad_norm": 1.5176632067554763, | |
| "learning_rate": 5.639165933106626e-06, | |
| "loss": 0.8658, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.4789736279401283, | |
| "grad_norm": 1.3789249301430986, | |
| "learning_rate": 5.617612124365411e-06, | |
| "loss": 0.879, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.48031529076349, | |
| "grad_norm": 1.2478449234735498, | |
| "learning_rate": 5.596046654557467e-06, | |
| "loss": 0.8533, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.4816569535868517, | |
| "grad_norm": 1.6124536322463092, | |
| "learning_rate": 5.574469930858062e-06, | |
| "loss": 0.8686, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.4829986164102134, | |
| "grad_norm": 1.438382594629828, | |
| "learning_rate": 5.55288236065495e-06, | |
| "loss": 0.8733, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.4843402792335751, | |
| "grad_norm": 1.1901796436527716, | |
| "learning_rate": 5.5312843515406725e-06, | |
| "loss": 0.8649, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.4856819420569368, | |
| "grad_norm": 1.283431289903718, | |
| "learning_rate": 5.509676311304869e-06, | |
| "loss": 0.8855, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.4870236048802985, | |
| "grad_norm": 1.7107312518838782, | |
| "learning_rate": 5.4880586479265774e-06, | |
| "loss": 0.867, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.48836526770366023, | |
| "grad_norm": 1.347121789114676, | |
| "learning_rate": 5.466431769566525e-06, | |
| "loss": 0.864, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.4897069305270219, | |
| "grad_norm": 1.4199661131096277, | |
| "learning_rate": 5.4447960845594286e-06, | |
| "loss": 0.8907, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.49104859335038364, | |
| "grad_norm": 1.5773421101140612, | |
| "learning_rate": 5.423152001406282e-06, | |
| "loss": 0.8641, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.4923902561737453, | |
| "grad_norm": 1.195090567248642, | |
| "learning_rate": 5.401499928766644e-06, | |
| "loss": 0.8423, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.49373191899710706, | |
| "grad_norm": 1.4721305059023198, | |
| "learning_rate": 5.379840275450922e-06, | |
| "loss": 0.8463, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.49507358182046873, | |
| "grad_norm": 1.3313323366659058, | |
| "learning_rate": 5.358173450412649e-06, | |
| "loss": 0.862, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.49641524464383047, | |
| "grad_norm": 1.4016546865386463, | |
| "learning_rate": 5.3364998627407746e-06, | |
| "loss": 0.8673, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.49775690746719214, | |
| "grad_norm": 1.244400425037615, | |
| "learning_rate": 5.314819921651928e-06, | |
| "loss": 0.8651, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.4990985702905539, | |
| "grad_norm": 2.0708294070309714, | |
| "learning_rate": 5.293134036482697e-06, | |
| "loss": 0.8678, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.5004402331139156, | |
| "grad_norm": 1.1872933354198374, | |
| "learning_rate": 5.271442616681898e-06, | |
| "loss": 0.8799, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.5017818959372773, | |
| "grad_norm": 1.2079078912661831, | |
| "learning_rate": 5.249746071802849e-06, | |
| "loss": 0.851, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.503123558760639, | |
| "grad_norm": 1.2384992577919112, | |
| "learning_rate": 5.228044811495632e-06, | |
| "loss": 0.8439, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.5044652215840006, | |
| "grad_norm": 1.2984630477638008, | |
| "learning_rate": 5.2063392454993585e-06, | |
| "loss": 0.864, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.5058068844073623, | |
| "grad_norm": 1.5264026775142745, | |
| "learning_rate": 5.184629783634441e-06, | |
| "loss": 0.8546, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.5071485472307241, | |
| "grad_norm": 1.2890589238746528, | |
| "learning_rate": 5.162916835794843e-06, | |
| "loss": 0.864, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.5084902100540858, | |
| "grad_norm": 1.2335899143681852, | |
| "learning_rate": 5.141200811940352e-06, | |
| "loss": 0.8454, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.5098318728774475, | |
| "grad_norm": 1.7592621170902492, | |
| "learning_rate": 5.119482122088828e-06, | |
| "loss": 0.8735, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.5111735357008091, | |
| "grad_norm": 1.4876979529975998, | |
| "learning_rate": 5.097761176308471e-06, | |
| "loss": 0.8684, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.5125151985241709, | |
| "grad_norm": 1.2510972758949657, | |
| "learning_rate": 5.076038384710077e-06, | |
| "loss": 0.8675, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.5138568613475326, | |
| "grad_norm": 1.5218020126429133, | |
| "learning_rate": 5.054314157439286e-06, | |
| "loss": 0.8838, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.5151985241708943, | |
| "grad_norm": 1.13096936442108, | |
| "learning_rate": 5.032588904668851e-06, | |
| "loss": 0.8462, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.516540186994256, | |
| "grad_norm": 1.2661684270573001, | |
| "learning_rate": 5.010863036590885e-06, | |
| "loss": 0.8518, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.5178818498176178, | |
| "grad_norm": 1.4633526972804205, | |
| "learning_rate": 4.989136963409117e-06, | |
| "loss": 0.8618, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.5192235126409794, | |
| "grad_norm": 1.3689217775090963, | |
| "learning_rate": 4.967411095331149e-06, | |
| "loss": 0.8492, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.5205651754643411, | |
| "grad_norm": 1.2542327263166364, | |
| "learning_rate": 4.945685842560715e-06, | |
| "loss": 0.8459, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.5219068382877028, | |
| "grad_norm": 2.087404509122426, | |
| "learning_rate": 4.923961615289924e-06, | |
| "loss": 0.8578, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.5232485011110646, | |
| "grad_norm": 1.3897933176387591, | |
| "learning_rate": 4.9022388236915306e-06, | |
| "loss": 0.8824, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.5245901639344263, | |
| "grad_norm": 1.442213637108022, | |
| "learning_rate": 4.880517877911173e-06, | |
| "loss": 0.8499, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.5259318267577879, | |
| "grad_norm": 1.2960656069995036, | |
| "learning_rate": 4.858799188059651e-06, | |
| "loss": 0.8702, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.5272734895811496, | |
| "grad_norm": 1.2717303062715928, | |
| "learning_rate": 4.837083164205159e-06, | |
| "loss": 0.8632, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.5286151524045113, | |
| "grad_norm": 1.9446143798835518, | |
| "learning_rate": 4.815370216365562e-06, | |
| "loss": 0.8714, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.5299568152278731, | |
| "grad_norm": 1.3611607055274428, | |
| "learning_rate": 4.793660754500644e-06, | |
| "loss": 0.8741, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.5312984780512348, | |
| "grad_norm": 1.2670368142763155, | |
| "learning_rate": 4.771955188504371e-06, | |
| "loss": 0.8581, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.5326401408745964, | |
| "grad_norm": 2.6612537194297166, | |
| "learning_rate": 4.75025392819715e-06, | |
| "loss": 0.8883, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.5339818036979581, | |
| "grad_norm": 1.3748403692564988, | |
| "learning_rate": 4.7285573833181016e-06, | |
| "loss": 0.8618, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.5353234665213199, | |
| "grad_norm": 1.2640493254233258, | |
| "learning_rate": 4.7068659635173034e-06, | |
| "loss": 0.8719, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.5366651293446816, | |
| "grad_norm": 1.4518180026733027, | |
| "learning_rate": 4.685180078348072e-06, | |
| "loss": 0.8592, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.5380067921680433, | |
| "grad_norm": 2.3481100711317424, | |
| "learning_rate": 4.663500137259226e-06, | |
| "loss": 0.8562, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.5393484549914049, | |
| "grad_norm": 1.887175568346411, | |
| "learning_rate": 4.641826549587352e-06, | |
| "loss": 0.8595, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.5406901178147667, | |
| "grad_norm": 2.146708879532439, | |
| "learning_rate": 4.62015972454908e-06, | |
| "loss": 0.8771, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.5420317806381284, | |
| "grad_norm": 1.4057713298780292, | |
| "learning_rate": 4.598500071233358e-06, | |
| "loss": 0.8491, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.5433734434614901, | |
| "grad_norm": 1.1911750555453442, | |
| "learning_rate": 4.57684799859372e-06, | |
| "loss": 0.8583, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.5447151062848518, | |
| "grad_norm": 1.2148757866243733, | |
| "learning_rate": 4.555203915440573e-06, | |
| "loss": 0.8469, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.5460567691082135, | |
| "grad_norm": 1.4270683372088357, | |
| "learning_rate": 4.533568230433477e-06, | |
| "loss": 0.8261, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.5473984319315752, | |
| "grad_norm": 1.239545542697653, | |
| "learning_rate": 4.511941352073424e-06, | |
| "loss": 0.8526, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.5487400947549369, | |
| "grad_norm": 1.305489341681733, | |
| "learning_rate": 4.4903236886951315e-06, | |
| "loss": 0.8533, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.5500817575782986, | |
| "grad_norm": 1.1764266946265571, | |
| "learning_rate": 4.468715648459329e-06, | |
| "loss": 0.8562, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.5514234204016603, | |
| "grad_norm": 1.3351747060022305, | |
| "learning_rate": 4.447117639345052e-06, | |
| "loss": 0.8633, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.552765083225022, | |
| "grad_norm": 1.1970814495336326, | |
| "learning_rate": 4.42553006914194e-06, | |
| "loss": 0.8714, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.5541067460483837, | |
| "grad_norm": 1.4531568858696167, | |
| "learning_rate": 4.403953345442535e-06, | |
| "loss": 0.8378, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.5554484088717454, | |
| "grad_norm": 1.2558211101985586, | |
| "learning_rate": 4.382387875634592e-06, | |
| "loss": 0.8313, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.5567900716951071, | |
| "grad_norm": 1.1514810079152125, | |
| "learning_rate": 4.360834066893376e-06, | |
| "loss": 0.8403, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.5581317345184689, | |
| "grad_norm": 1.371348036969129, | |
| "learning_rate": 4.339292326173987e-06, | |
| "loss": 0.8671, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.5594733973418305, | |
| "grad_norm": 1.3230986934582532, | |
| "learning_rate": 4.317763060203665e-06, | |
| "loss": 0.8413, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.5608150601651922, | |
| "grad_norm": 1.2820607351791564, | |
| "learning_rate": 4.296246675474119e-06, | |
| "loss": 0.8583, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.5621567229885539, | |
| "grad_norm": 1.9954830950675657, | |
| "learning_rate": 4.274743578233846e-06, | |
| "loss": 0.8542, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.5634983858119157, | |
| "grad_norm": 1.3880776140709008, | |
| "learning_rate": 4.253254174480462e-06, | |
| "loss": 0.8491, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.5648400486352774, | |
| "grad_norm": 1.1952626395664954, | |
| "learning_rate": 4.231778869953047e-06, | |
| "loss": 0.8584, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.566181711458639, | |
| "grad_norm": 1.1765768602462916, | |
| "learning_rate": 4.210318070124465e-06, | |
| "loss": 0.8486, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.5675233742820007, | |
| "grad_norm": 1.1175718768610468, | |
| "learning_rate": 4.188872180193723e-06, | |
| "loss": 0.8587, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.5688650371053624, | |
| "grad_norm": 1.4073819401333307, | |
| "learning_rate": 4.167441605078314e-06, | |
| "loss": 0.851, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.5702066999287242, | |
| "grad_norm": 1.4128212039665118, | |
| "learning_rate": 4.146026749406574e-06, | |
| "loss": 0.8558, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.5715483627520859, | |
| "grad_norm": 1.4533338269101694, | |
| "learning_rate": 4.124628017510043e-06, | |
| "loss": 0.8652, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.5728900255754475, | |
| "grad_norm": 1.2071992792650619, | |
| "learning_rate": 4.10324581341583e-06, | |
| "loss": 0.8588, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.5742316883988092, | |
| "grad_norm": 1.6310164159014424, | |
| "learning_rate": 4.081880540838981e-06, | |
| "loss": 0.8455, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.575573351222171, | |
| "grad_norm": 1.3510211599555761, | |
| "learning_rate": 4.060532603174865e-06, | |
| "loss": 0.8528, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.5769150140455327, | |
| "grad_norm": 1.6069988016873833, | |
| "learning_rate": 4.039202403491548e-06, | |
| "loss": 0.8448, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.5782566768688944, | |
| "grad_norm": 1.1569790393104251, | |
| "learning_rate": 4.0178903445221905e-06, | |
| "loss": 0.8554, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.579598339692256, | |
| "grad_norm": 1.2847644678259915, | |
| "learning_rate": 3.996596828657437e-06, | |
| "loss": 0.858, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.5809400025156178, | |
| "grad_norm": 1.2947307471087732, | |
| "learning_rate": 3.975322257937825e-06, | |
| "loss": 0.8583, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.5822816653389795, | |
| "grad_norm": 1.2929072041687166, | |
| "learning_rate": 3.9540670340461875e-06, | |
| "loss": 0.8391, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.5836233281623412, | |
| "grad_norm": 1.392783182793689, | |
| "learning_rate": 3.932831558300074e-06, | |
| "loss": 0.8536, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.5849649909857029, | |
| "grad_norm": 1.069346969882725, | |
| "learning_rate": 3.911616231644172e-06, | |
| "loss": 0.8482, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.5863066538090647, | |
| "grad_norm": 1.4706657929010962, | |
| "learning_rate": 3.8904214546427355e-06, | |
| "loss": 0.8523, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.5876483166324263, | |
| "grad_norm": 1.281271506392474, | |
| "learning_rate": 3.869247627472021e-06, | |
| "loss": 0.866, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.588989979455788, | |
| "grad_norm": 1.2898365265402107, | |
| "learning_rate": 3.848095149912737e-06, | |
| "loss": 0.8668, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.5903316422791497, | |
| "grad_norm": 1.5820350165443553, | |
| "learning_rate": 3.82696442134249e-06, | |
| "loss": 0.873, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.5916733051025114, | |
| "grad_norm": 1.3476793491220833, | |
| "learning_rate": 3.8058558407282465e-06, | |
| "loss": 0.8426, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.5930149679258732, | |
| "grad_norm": 1.5205269280443248, | |
| "learning_rate": 3.7847698066187975e-06, | |
| "loss": 0.8501, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.5943566307492348, | |
| "grad_norm": 1.2926084127600388, | |
| "learning_rate": 3.7637067171372437e-06, | |
| "loss": 0.8535, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.5956982935725965, | |
| "grad_norm": 1.2933290311248173, | |
| "learning_rate": 3.742666969973463e-06, | |
| "loss": 0.8619, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.5970399563959582, | |
| "grad_norm": 1.4853553307569838, | |
| "learning_rate": 3.721650962376612e-06, | |
| "loss": 0.8542, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.59838161921932, | |
| "grad_norm": 1.2020579472445274, | |
| "learning_rate": 3.7006590911476242e-06, | |
| "loss": 0.8388, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.5997232820426817, | |
| "grad_norm": 1.3128084375138176, | |
| "learning_rate": 3.6796917526317153e-06, | |
| "loss": 0.8542, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.6010649448660433, | |
| "grad_norm": 1.1557860528616235, | |
| "learning_rate": 3.6587493427109015e-06, | |
| "loss": 0.8334, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.602406607689405, | |
| "grad_norm": 1.2001049541956863, | |
| "learning_rate": 3.6378322567965284e-06, | |
| "loss": 0.8727, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.6037482705127668, | |
| "grad_norm": 1.9097019774524748, | |
| "learning_rate": 3.6169408898217973e-06, | |
| "loss": 0.8466, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.6050899333361285, | |
| "grad_norm": 1.3409231439886813, | |
| "learning_rate": 3.596075636234315e-06, | |
| "loss": 0.8517, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.6064315961594902, | |
| "grad_norm": 1.2677304624224974, | |
| "learning_rate": 3.575236889988646e-06, | |
| "loss": 0.8387, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.6077732589828518, | |
| "grad_norm": 1.2113747508763804, | |
| "learning_rate": 3.554425044538868e-06, | |
| "loss": 0.8544, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.6091149218062136, | |
| "grad_norm": 1.1823779134452883, | |
| "learning_rate": 3.5336404928311518e-06, | |
| "loss": 0.8295, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.6104565846295753, | |
| "grad_norm": 1.5577704830610386, | |
| "learning_rate": 3.5128836272963377e-06, | |
| "loss": 0.853, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.611798247452937, | |
| "grad_norm": 1.4404082085080354, | |
| "learning_rate": 3.4921548398425246e-06, | |
| "loss": 0.8569, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.6131399102762987, | |
| "grad_norm": 1.4051477293078514, | |
| "learning_rate": 3.4714545218476727e-06, | |
| "loss": 0.8491, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.6144815730996603, | |
| "grad_norm": 1.4946853282520889, | |
| "learning_rate": 3.4507830641522144e-06, | |
| "loss": 0.8543, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.6158232359230221, | |
| "grad_norm": 1.1919040256212239, | |
| "learning_rate": 3.430140857051675e-06, | |
| "loss": 0.8544, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.6171648987463838, | |
| "grad_norm": 2.2727242203744096, | |
| "learning_rate": 3.4095282902892982e-06, | |
| "loss": 0.8431, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.6185065615697455, | |
| "grad_norm": 1.0854799673042679, | |
| "learning_rate": 3.388945753048697e-06, | |
| "loss": 0.8267, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.6198482243931072, | |
| "grad_norm": 1.1656949919338342, | |
| "learning_rate": 3.3683936339464957e-06, | |
| "loss": 0.8693, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.621189887216469, | |
| "grad_norm": 1.141589351521228, | |
| "learning_rate": 3.3478723210250006e-06, | |
| "loss": 0.8402, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.6225315500398306, | |
| "grad_norm": 1.1701905043520016, | |
| "learning_rate": 3.3273822017448683e-06, | |
| "loss": 0.8517, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.6238732128631923, | |
| "grad_norm": 1.285565681754486, | |
| "learning_rate": 3.306923662977789e-06, | |
| "loss": 0.8335, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.625214875686554, | |
| "grad_norm": 1.4128373703802648, | |
| "learning_rate": 3.2864970909991927e-06, | |
| "loss": 0.8387, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.6265565385099158, | |
| "grad_norm": 2.6789887639637544, | |
| "learning_rate": 3.2661028714809405e-06, | |
| "loss": 0.8474, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.6278982013332775, | |
| "grad_norm": 1.3319949603977097, | |
| "learning_rate": 3.2457413894840516e-06, | |
| "loss": 0.8175, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.6292398641566391, | |
| "grad_norm": 2.896720879525036, | |
| "learning_rate": 3.225413029451434e-06, | |
| "loss": 0.8453, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.6305815269800008, | |
| "grad_norm": 1.3500205291437544, | |
| "learning_rate": 3.20511817520062e-06, | |
| "loss": 0.8299, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.6319231898033626, | |
| "grad_norm": 1.287532606369544, | |
| "learning_rate": 3.184857209916528e-06, | |
| "loss": 0.8457, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.6332648526267243, | |
| "grad_norm": 1.3076164619064987, | |
| "learning_rate": 3.1646305161442183e-06, | |
| "loss": 0.8343, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.634606515450086, | |
| "grad_norm": 1.0790177080379393, | |
| "learning_rate": 3.144438475781676e-06, | |
| "loss": 0.8536, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.6359481782734476, | |
| "grad_norm": 1.1632447433989597, | |
| "learning_rate": 3.1242814700725977e-06, | |
| "loss": 0.8467, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.6372898410968093, | |
| "grad_norm": 1.1786373703258224, | |
| "learning_rate": 3.1041598795991967e-06, | |
| "loss": 0.8536, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.6386315039201711, | |
| "grad_norm": 1.2482884226487594, | |
| "learning_rate": 3.0840740842750145e-06, | |
| "loss": 0.8378, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.6399731667435328, | |
| "grad_norm": 1.243617708259521, | |
| "learning_rate": 3.064024463337747e-06, | |
| "loss": 0.8364, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.6413148295668945, | |
| "grad_norm": 1.3943485163577303, | |
| "learning_rate": 3.044011395342087e-06, | |
| "loss": 0.8596, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.6426564923902561, | |
| "grad_norm": 1.2282360805939099, | |
| "learning_rate": 3.0240352581525735e-06, | |
| "loss": 0.8506, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.6439981552136179, | |
| "grad_norm": 1.5106508743829434, | |
| "learning_rate": 3.0040964289364618e-06, | |
| "loss": 0.8479, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.6453398180369796, | |
| "grad_norm": 1.1908518888076973, | |
| "learning_rate": 2.984195284156597e-06, | |
| "loss": 0.8329, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.6466814808603413, | |
| "grad_norm": 1.256378696411481, | |
| "learning_rate": 2.964332199564309e-06, | |
| "loss": 0.8432, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.648023143683703, | |
| "grad_norm": 1.5310927556211729, | |
| "learning_rate": 2.944507550192318e-06, | |
| "loss": 0.8541, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.6493648065070647, | |
| "grad_norm": 1.2860261469481418, | |
| "learning_rate": 2.9247217103476527e-06, | |
| "loss": 0.8281, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.6507064693304264, | |
| "grad_norm": 1.2918532848488533, | |
| "learning_rate": 2.904975053604584e-06, | |
| "loss": 0.8343, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.6520481321537881, | |
| "grad_norm": 1.4077878782059026, | |
| "learning_rate": 2.885267952797569e-06, | |
| "loss": 0.8471, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.6533897949771498, | |
| "grad_norm": 1.466437435021114, | |
| "learning_rate": 2.865600780014216e-06, | |
| "loss": 0.8579, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.6547314578005116, | |
| "grad_norm": 1.4734892824238082, | |
| "learning_rate": 2.8459739065882537e-06, | |
| "loss": 0.8559, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.6560731206238732, | |
| "grad_norm": 1.2880091237191134, | |
| "learning_rate": 2.826387703092528e-06, | |
| "loss": 0.8429, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.6574147834472349, | |
| "grad_norm": 1.3265198187011118, | |
| "learning_rate": 2.8068425393319965e-06, | |
| "loss": 0.8542, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.6587564462705966, | |
| "grad_norm": 1.5509255640028885, | |
| "learning_rate": 2.7873387843367494e-06, | |
| "loss": 0.8085, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.6600981090939583, | |
| "grad_norm": 1.3255446511811233, | |
| "learning_rate": 2.7678768063550454e-06, | |
| "loss": 0.8495, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.6614397719173201, | |
| "grad_norm": 1.9005237583975607, | |
| "learning_rate": 2.748456972846353e-06, | |
| "loss": 0.8544, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.6627814347406817, | |
| "grad_norm": 1.2153008589884216, | |
| "learning_rate": 2.7290796504744184e-06, | |
| "loss": 0.8399, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.6641230975640434, | |
| "grad_norm": 1.6082631737322852, | |
| "learning_rate": 2.7097452051003375e-06, | |
| "loss": 0.8389, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.6654647603874051, | |
| "grad_norm": 1.2098630006029956, | |
| "learning_rate": 2.69045400177565e-06, | |
| "loss": 0.8547, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.6668064232107669, | |
| "grad_norm": 1.2745481004715278, | |
| "learning_rate": 2.6712064047354515e-06, | |
| "loss": 0.8372, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.6681480860341286, | |
| "grad_norm": 1.273679529878616, | |
| "learning_rate": 2.6520027773915075e-06, | |
| "loss": 0.8542, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.6694897488574902, | |
| "grad_norm": 1.2181662191670053, | |
| "learning_rate": 2.632843482325398e-06, | |
| "loss": 0.8483, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.6708314116808519, | |
| "grad_norm": 1.5349250216275356, | |
| "learning_rate": 2.6137288812816695e-06, | |
| "loss": 0.8619, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.6721730745042137, | |
| "grad_norm": 1.212279718049173, | |
| "learning_rate": 2.594659335161008e-06, | |
| "loss": 0.8269, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.6735147373275754, | |
| "grad_norm": 1.465633175883194, | |
| "learning_rate": 2.5756352040134193e-06, | |
| "loss": 0.8325, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.6748564001509371, | |
| "grad_norm": 2.167846998481404, | |
| "learning_rate": 2.5566568470314345e-06, | |
| "loss": 0.8083, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.6761980629742987, | |
| "grad_norm": 1.2655165281910223, | |
| "learning_rate": 2.5377246225433306e-06, | |
| "loss": 0.8385, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.6775397257976605, | |
| "grad_norm": 1.459299106935816, | |
| "learning_rate": 2.5188388880063595e-06, | |
| "loss": 0.8301, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.6788813886210222, | |
| "grad_norm": 1.2112725513957674, | |
| "learning_rate": 2.5000000000000015e-06, | |
| "loss": 0.8444, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.6802230514443839, | |
| "grad_norm": 4.313551779654705, | |
| "learning_rate": 2.481208314219233e-06, | |
| "loss": 0.8102, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.6815647142677456, | |
| "grad_norm": 1.2658710722451902, | |
| "learning_rate": 2.462464185467811e-06, | |
| "loss": 0.8596, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.6829063770911072, | |
| "grad_norm": 1.331719521795118, | |
| "learning_rate": 2.4437679676515735e-06, | |
| "loss": 0.8545, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.684248039914469, | |
| "grad_norm": 1.3060612955759165, | |
| "learning_rate": 2.4251200137717545e-06, | |
| "loss": 0.8422, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.6855897027378307, | |
| "grad_norm": 1.24522327945716, | |
| "learning_rate": 2.4065206759183284e-06, | |
| "loss": 0.8419, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.6869313655611924, | |
| "grad_norm": 1.4676381651479729, | |
| "learning_rate": 2.387970305263349e-06, | |
| "loss": 0.8378, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.6882730283845541, | |
| "grad_norm": 1.3237220542978192, | |
| "learning_rate": 2.3694692520543293e-06, | |
| "loss": 0.8446, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.6896146912079159, | |
| "grad_norm": 1.8196729938616845, | |
| "learning_rate": 2.3510178656076245e-06, | |
| "loss": 0.8465, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.6909563540312775, | |
| "grad_norm": 1.3428553276343433, | |
| "learning_rate": 2.3326164943018353e-06, | |
| "loss": 0.8498, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.6922980168546392, | |
| "grad_norm": 1.1519689905807404, | |
| "learning_rate": 2.3142654855712353e-06, | |
| "loss": 0.8287, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.6936396796780009, | |
| "grad_norm": 1.1369825358629937, | |
| "learning_rate": 2.295965185899205e-06, | |
| "loss": 0.8412, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.6949813425013627, | |
| "grad_norm": 1.2137229641508855, | |
| "learning_rate": 2.277715940811692e-06, | |
| "loss": 0.832, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.6963230053247244, | |
| "grad_norm": 1.1619721142855948, | |
| "learning_rate": 2.259518094870693e-06, | |
| "loss": 0.8434, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.697664668148086, | |
| "grad_norm": 1.507992647943336, | |
| "learning_rate": 2.241371991667739e-06, | |
| "loss": 0.839, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.6990063309714477, | |
| "grad_norm": 1.1435074742509186, | |
| "learning_rate": 2.2232779738174126e-06, | |
| "loss": 0.8394, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.7003479937948094, | |
| "grad_norm": 1.138116754628977, | |
| "learning_rate": 2.2052363829508776e-06, | |
| "loss": 0.8423, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.7016896566181712, | |
| "grad_norm": 1.3438972367375897, | |
| "learning_rate": 2.1872475597094323e-06, | |
| "loss": 0.8384, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.7030313194415329, | |
| "grad_norm": 1.3562891008567788, | |
| "learning_rate": 2.1693118437380727e-06, | |
| "loss": 0.8653, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.7043729822648945, | |
| "grad_norm": 1.1968097734254102, | |
| "learning_rate": 2.151429573679084e-06, | |
| "loss": 0.8191, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.7057146450882562, | |
| "grad_norm": 1.2790878945758999, | |
| "learning_rate": 2.133601087165644e-06, | |
| "loss": 0.8506, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.707056307911618, | |
| "grad_norm": 1.2144241738596728, | |
| "learning_rate": 2.1158267208154497e-06, | |
| "loss": 0.8301, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.7083979707349797, | |
| "grad_norm": 1.1550780949906838, | |
| "learning_rate": 2.098106810224362e-06, | |
| "loss": 0.8351, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.7097396335583414, | |
| "grad_norm": 1.176649100089699, | |
| "learning_rate": 2.080441689960066e-06, | |
| "loss": 0.8291, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.711081296381703, | |
| "grad_norm": 1.1813598916688428, | |
| "learning_rate": 2.0628316935557595e-06, | |
| "loss": 0.8377, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.7124229592050648, | |
| "grad_norm": 1.1562827713816028, | |
| "learning_rate": 2.0452771535038518e-06, | |
| "loss": 0.8329, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.7137646220284265, | |
| "grad_norm": 1.1879418830505444, | |
| "learning_rate": 2.0277784012496865e-06, | |
| "loss": 0.8393, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.7151062848517882, | |
| "grad_norm": 1.2459972282752854, | |
| "learning_rate": 2.010335767185283e-06, | |
| "loss": 0.8548, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.7164479476751499, | |
| "grad_norm": 1.1626070795399546, | |
| "learning_rate": 1.9929495806431024e-06, | |
| "loss": 0.8345, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.7177896104985116, | |
| "grad_norm": 1.18214175098007, | |
| "learning_rate": 1.9756201698898246e-06, | |
| "loss": 0.8295, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.7191312733218733, | |
| "grad_norm": 1.4119877204126912, | |
| "learning_rate": 1.958347862120151e-06, | |
| "loss": 0.8214, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.720472936145235, | |
| "grad_norm": 6.523297715976198, | |
| "learning_rate": 1.9411329834506286e-06, | |
| "loss": 0.8469, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.7218145989685967, | |
| "grad_norm": 2.1903775270203347, | |
| "learning_rate": 1.923975858913492e-06, | |
| "loss": 0.8287, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.7231562617919584, | |
| "grad_norm": 1.440617180275712, | |
| "learning_rate": 1.9068768124505265e-06, | |
| "loss": 0.8358, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.7244979246153201, | |
| "grad_norm": 1.7384186846110947, | |
| "learning_rate": 1.8898361669069497e-06, | |
| "loss": 0.8309, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.7258395874386818, | |
| "grad_norm": 1.3761102779704912, | |
| "learning_rate": 1.8728542440253232e-06, | |
| "loss": 0.8245, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.7271812502620435, | |
| "grad_norm": 1.3999320743351735, | |
| "learning_rate": 1.8559313644394677e-06, | |
| "loss": 0.8547, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.7285229130854052, | |
| "grad_norm": 1.2297207115367963, | |
| "learning_rate": 1.8390678476684143e-06, | |
| "loss": 0.8362, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.729864575908767, | |
| "grad_norm": 1.1671484327652857, | |
| "learning_rate": 1.8222640121103724e-06, | |
| "loss": 0.8399, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.7312062387321286, | |
| "grad_norm": 1.6018047169313065, | |
| "learning_rate": 1.8055201750367163e-06, | |
| "loss": 0.8283, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.7325479015554903, | |
| "grad_norm": 1.1524928330145296, | |
| "learning_rate": 1.7888366525859968e-06, | |
| "loss": 0.8329, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.733889564378852, | |
| "grad_norm": 1.2590599451021425, | |
| "learning_rate": 1.7722137597579698e-06, | |
| "loss": 0.8386, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.7352312272022138, | |
| "grad_norm": 1.136348429679451, | |
| "learning_rate": 1.7556518104076497e-06, | |
| "loss": 0.8466, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.7365728900255755, | |
| "grad_norm": 2.4050020145296402, | |
| "learning_rate": 1.7391511172393849e-06, | |
| "loss": 0.8203, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.7379145528489371, | |
| "grad_norm": 1.2678818194670964, | |
| "learning_rate": 1.7227119918009516e-06, | |
| "loss": 0.8404, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.7392562156722988, | |
| "grad_norm": 1.1559365595972262, | |
| "learning_rate": 1.706334744477673e-06, | |
| "loss": 0.8437, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.7405978784956606, | |
| "grad_norm": 3.1619855391356286, | |
| "learning_rate": 1.6900196844865575e-06, | |
| "loss": 0.8357, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.7419395413190223, | |
| "grad_norm": 5.971370243417492, | |
| "learning_rate": 1.6737671198704625e-06, | |
| "loss": 0.8364, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.743281204142384, | |
| "grad_norm": 1.2410121596260226, | |
| "learning_rate": 1.657577357492277e-06, | |
| "loss": 0.8379, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.7446228669657456, | |
| "grad_norm": 1.2501465011701856, | |
| "learning_rate": 1.6414507030291249e-06, | |
| "loss": 0.8317, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.7459645297891073, | |
| "grad_norm": 1.5184952502355407, | |
| "learning_rate": 1.6253874609666033e-06, | |
| "loss": 0.8341, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.7473061926124691, | |
| "grad_norm": 1.3916837970238092, | |
| "learning_rate": 1.609387934593019e-06, | |
| "loss": 0.8356, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.7486478554358308, | |
| "grad_norm": 1.319431508569105, | |
| "learning_rate": 1.5934524259936757e-06, | |
| "loss": 0.8281, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.7499895182591925, | |
| "grad_norm": 1.4978026620635356, | |
| "learning_rate": 1.5775812360451598e-06, | |
| "loss": 0.8428, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.7513311810825541, | |
| "grad_norm": 1.0810145766867532, | |
| "learning_rate": 1.5617746644096681e-06, | |
| "loss": 0.8314, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.7526728439059159, | |
| "grad_norm": 1.1830498116396289, | |
| "learning_rate": 1.5460330095293447e-06, | |
| "loss": 0.8099, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.7540145067292776, | |
| "grad_norm": 1.502144421688931, | |
| "learning_rate": 1.5303565686206452e-06, | |
| "loss": 0.8098, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.7553561695526393, | |
| "grad_norm": 1.2405881546221902, | |
| "learning_rate": 1.514745637668733e-06, | |
| "loss": 0.8353, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.756697832376001, | |
| "grad_norm": 1.3390140709621186, | |
| "learning_rate": 1.4992005114218805e-06, | |
| "loss": 0.8246, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.7580394951993628, | |
| "grad_norm": 2.161912896000941, | |
| "learning_rate": 1.4837214833859099e-06, | |
| "loss": 0.8539, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.7593811580227244, | |
| "grad_norm": 1.250899684662496, | |
| "learning_rate": 1.4683088458186512e-06, | |
| "loss": 0.8322, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.7607228208460861, | |
| "grad_norm": 3.3883226001112265, | |
| "learning_rate": 1.4529628897244214e-06, | |
| "loss": 0.8359, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.7620644836694478, | |
| "grad_norm": 1.1644103714659484, | |
| "learning_rate": 1.4376839048485335e-06, | |
| "loss": 0.8044, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.7634061464928096, | |
| "grad_norm": 1.2311518513086757, | |
| "learning_rate": 1.422472179671825e-06, | |
| "loss": 0.8472, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.7647478093161713, | |
| "grad_norm": 1.3800438741200767, | |
| "learning_rate": 1.4073280014052077e-06, | |
| "loss": 0.8649, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.7660894721395329, | |
| "grad_norm": 1.1720066594353433, | |
| "learning_rate": 1.3922516559842498e-06, | |
| "loss": 0.8275, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.7674311349628946, | |
| "grad_norm": 1.2005657900693618, | |
| "learning_rate": 1.3772434280637737e-06, | |
| "loss": 0.8471, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.7687727977862563, | |
| "grad_norm": 3.437018109942447, | |
| "learning_rate": 1.3623036010124845e-06, | |
| "loss": 0.8279, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.7701144606096181, | |
| "grad_norm": 1.452924562350273, | |
| "learning_rate": 1.347432456907617e-06, | |
| "loss": 0.8247, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.7714561234329798, | |
| "grad_norm": 1.2583863948266167, | |
| "learning_rate": 1.3326302765296096e-06, | |
| "loss": 0.85, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.7727977862563414, | |
| "grad_norm": 1.0541988421603572, | |
| "learning_rate": 1.3178973393568055e-06, | |
| "loss": 0.8352, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.7741394490797031, | |
| "grad_norm": 1.1773555696828506, | |
| "learning_rate": 1.3032339235601749e-06, | |
| "loss": 0.8114, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.7754811119030649, | |
| "grad_norm": 1.2156865144078284, | |
| "learning_rate": 1.2886403059980602e-06, | |
| "loss": 0.8215, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.7768227747264266, | |
| "grad_norm": 1.2390636354948865, | |
| "learning_rate": 1.2741167622109557e-06, | |
| "loss": 0.824, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.7781644375497883, | |
| "grad_norm": 1.1604255060152633, | |
| "learning_rate": 1.259663566416296e-06, | |
| "loss": 0.8363, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.7795061003731499, | |
| "grad_norm": 1.3478280196533168, | |
| "learning_rate": 1.2452809915032843e-06, | |
| "loss": 0.8468, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.7808477631965117, | |
| "grad_norm": 1.2550557920884333, | |
| "learning_rate": 1.2309693090277392e-06, | |
| "loss": 0.8356, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.7821894260198734, | |
| "grad_norm": 1.1678575271212497, | |
| "learning_rate": 1.2167287892069662e-06, | |
| "loss": 0.8343, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.7835310888432351, | |
| "grad_norm": 1.5962894483490684, | |
| "learning_rate": 1.202559700914656e-06, | |
| "loss": 0.8596, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.7848727516665968, | |
| "grad_norm": 1.3311037790554563, | |
| "learning_rate": 1.1884623116758121e-06, | |
| "loss": 0.8383, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.7862144144899585, | |
| "grad_norm": 1.1070638503431656, | |
| "learning_rate": 1.1744368876616913e-06, | |
| "loss": 0.8104, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.7875560773133202, | |
| "grad_norm": 1.11512318760542, | |
| "learning_rate": 1.1604836936847852e-06, | |
| "loss": 0.8591, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.7888977401366819, | |
| "grad_norm": 1.081395038448867, | |
| "learning_rate": 1.1466029931938182e-06, | |
| "loss": 0.8275, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.7902394029600436, | |
| "grad_norm": 1.1449550488318547, | |
| "learning_rate": 1.1327950482687727e-06, | |
| "loss": 0.8552, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.7915810657834053, | |
| "grad_norm": 1.751425768587215, | |
| "learning_rate": 1.1190601196159406e-06, | |
| "loss": 0.8203, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.792922728606767, | |
| "grad_norm": 1.252896211335825, | |
| "learning_rate": 1.1053984665630025e-06, | |
| "loss": 0.8237, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.7942643914301287, | |
| "grad_norm": 1.0963879029813988, | |
| "learning_rate": 1.0918103470541297e-06, | |
| "loss": 0.8415, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.7956060542534904, | |
| "grad_norm": 1.18245487613986, | |
| "learning_rate": 1.0782960176451164e-06, | |
| "loss": 0.8225, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.7969477170768521, | |
| "grad_norm": 1.2604481634762956, | |
| "learning_rate": 1.064855733498531e-06, | |
| "loss": 0.8346, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.7982893799002139, | |
| "grad_norm": 1.237220925206381, | |
| "learning_rate": 1.051489748378905e-06, | |
| "loss": 0.8392, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.7996310427235755, | |
| "grad_norm": 1.362930872327231, | |
| "learning_rate": 1.0381983146479352e-06, | |
| "loss": 0.8245, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.8009727055469372, | |
| "grad_norm": 1.2261070007069856, | |
| "learning_rate": 1.024981683259723e-06, | |
| "loss": 0.8323, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.8023143683702989, | |
| "grad_norm": 1.4659008795929862, | |
| "learning_rate": 1.0118401037560354e-06, | |
| "loss": 0.8409, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.8036560311936607, | |
| "grad_norm": 1.087120902339791, | |
| "learning_rate": 9.987738242615924e-07, | |
| "loss": 0.8215, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.8049976940170224, | |
| "grad_norm": 1.073634654619719, | |
| "learning_rate": 9.857830914793827e-07, | |
| "loss": 0.8336, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.806339356840384, | |
| "grad_norm": 1.264401682484928, | |
| "learning_rate": 9.72868150686005e-07, | |
| "loss": 0.8425, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.8076810196637457, | |
| "grad_norm": 1.3708087219205602, | |
| "learning_rate": 9.600292457270416e-07, | |
| "loss": 0.8257, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.8090226824871075, | |
| "grad_norm": 1.2988469721761515, | |
| "learning_rate": 9.472666190124457e-07, | |
| "loss": 0.8537, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.8103643453104692, | |
| "grad_norm": 3.37658394021203, | |
| "learning_rate": 9.345805115119722e-07, | |
| "loss": 0.8306, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.8117060081338309, | |
| "grad_norm": 1.3065857730324328, | |
| "learning_rate": 9.219711627506239e-07, | |
| "loss": 0.8404, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.8130476709571925, | |
| "grad_norm": 1.1113767606755425, | |
| "learning_rate": 9.094388108041302e-07, | |
| "loss": 0.8303, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.8143893337805542, | |
| "grad_norm": 1.1657800517685706, | |
| "learning_rate": 8.969836922944536e-07, | |
| "loss": 0.83, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.815730996603916, | |
| "grad_norm": 1.1596320660986, | |
| "learning_rate": 8.846060423853198e-07, | |
| "loss": 0.8283, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.8170726594272777, | |
| "grad_norm": 1.0585594827837355, | |
| "learning_rate": 8.723060947777778e-07, | |
| "loss": 0.8312, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.8184143222506394, | |
| "grad_norm": 1.1439552026420334, | |
| "learning_rate": 8.600840817057877e-07, | |
| "loss": 0.8356, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.819755985074001, | |
| "grad_norm": 1.2102667917403713, | |
| "learning_rate": 8.479402339318372e-07, | |
| "loss": 0.8339, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.8210976478973628, | |
| "grad_norm": 1.3021255138093222, | |
| "learning_rate": 8.358747807425827e-07, | |
| "loss": 0.839, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.8224393107207245, | |
| "grad_norm": 1.205768206496087, | |
| "learning_rate": 8.238879499445224e-07, | |
| "loss": 0.8382, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.8237809735440862, | |
| "grad_norm": 1.660024198173359, | |
| "learning_rate": 8.119799678596918e-07, | |
| "loss": 0.8239, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.8251226363674479, | |
| "grad_norm": 1.1733256303404405, | |
| "learning_rate": 8.001510593213946e-07, | |
| "loss": 0.8298, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.8264642991908097, | |
| "grad_norm": 1.1407167527499544, | |
| "learning_rate": 7.884014476699541e-07, | |
| "loss": 0.8316, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.8278059620141713, | |
| "grad_norm": 1.5302459706545026, | |
| "learning_rate": 7.767313547484979e-07, | |
| "loss": 0.8441, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.829147624837533, | |
| "grad_norm": 1.3006442300826015, | |
| "learning_rate": 7.651410008987698e-07, | |
| "loss": 0.8313, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.8304892876608947, | |
| "grad_norm": 1.4674777457592694, | |
| "learning_rate": 7.536306049569686e-07, | |
| "loss": 0.8397, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.8318309504842564, | |
| "grad_norm": 1.4067379527665744, | |
| "learning_rate": 7.422003842496167e-07, | |
| "loss": 0.8277, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.8331726133076182, | |
| "grad_norm": 1.3555180184110087, | |
| "learning_rate": 7.308505545894567e-07, | |
| "loss": 0.8414, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.8345142761309798, | |
| "grad_norm": 1.258513776140156, | |
| "learning_rate": 7.195813302713761e-07, | |
| "loss": 0.8151, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.8358559389543415, | |
| "grad_norm": 1.1086248000239025, | |
| "learning_rate": 7.083929240683618e-07, | |
| "loss": 0.8354, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.8371976017777032, | |
| "grad_norm": 1.1167326128771922, | |
| "learning_rate": 6.972855472274853e-07, | |
| "loss": 0.8338, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.838539264601065, | |
| "grad_norm": 1.2723332222553532, | |
| "learning_rate": 6.862594094659086e-07, | |
| "loss": 0.8481, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.8398809274244267, | |
| "grad_norm": 1.3924636024826844, | |
| "learning_rate": 6.753147189669279e-07, | |
| "loss": 0.8271, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.8412225902477883, | |
| "grad_norm": 3.1000305394961725, | |
| "learning_rate": 6.644516823760439e-07, | |
| "loss": 0.8076, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.84256425307115, | |
| "grad_norm": 1.712034530578912, | |
| "learning_rate": 6.536705047970566e-07, | |
| "loss": 0.823, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.8439059158945118, | |
| "grad_norm": 1.2254068844318322, | |
| "learning_rate": 6.429713897881984e-07, | |
| "loss": 0.8406, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.8452475787178735, | |
| "grad_norm": 1.150881741430995, | |
| "learning_rate": 6.323545393582847e-07, | |
| "loss": 0.8326, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.8465892415412352, | |
| "grad_norm": 1.2031877134089701, | |
| "learning_rate": 6.218201539629032e-07, | |
| "loss": 0.8501, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.8479309043645968, | |
| "grad_norm": 1.1331639930188349, | |
| "learning_rate": 6.113684325006286e-07, | |
| "loss": 0.8398, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.8492725671879586, | |
| "grad_norm": 1.3328337977241533, | |
| "learning_rate": 6.009995723092655e-07, | |
| "loss": 0.8447, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.8506142300113203, | |
| "grad_norm": 1.1594369234917001, | |
| "learning_rate": 5.907137691621256e-07, | |
| "loss": 0.8308, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.851955892834682, | |
| "grad_norm": 1.2166193055726704, | |
| "learning_rate": 5.805112172643296e-07, | |
| "loss": 0.8303, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.8532975556580437, | |
| "grad_norm": 1.9198110022701735, | |
| "learning_rate": 5.703921092491393e-07, | |
| "loss": 0.8467, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.8546392184814053, | |
| "grad_norm": 1.2812675851071793, | |
| "learning_rate": 5.603566361743229e-07, | |
| "loss": 0.8311, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.8559808813047671, | |
| "grad_norm": 1.2057791711394406, | |
| "learning_rate": 5.504049875185458e-07, | |
| "loss": 0.8169, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.8573225441281288, | |
| "grad_norm": 1.5775617820674916, | |
| "learning_rate": 5.405373511777939e-07, | |
| "loss": 0.8296, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.8586642069514905, | |
| "grad_norm": 1.4556909711627686, | |
| "learning_rate": 5.307539134618256e-07, | |
| "loss": 0.828, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.8600058697748522, | |
| "grad_norm": 1.1213012222168914, | |
| "learning_rate": 5.21054859090655e-07, | |
| "loss": 0.8189, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.861347532598214, | |
| "grad_norm": 1.1887977198230921, | |
| "learning_rate": 5.114403711910631e-07, | |
| "loss": 0.8434, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.8626891954215756, | |
| "grad_norm": 1.0848104797959695, | |
| "learning_rate": 5.019106312931399e-07, | |
| "loss": 0.8434, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.8640308582449373, | |
| "grad_norm": 1.5322876396151517, | |
| "learning_rate": 4.924658193268595e-07, | |
| "loss": 0.8358, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.865372521068299, | |
| "grad_norm": 3.05796115499695, | |
| "learning_rate": 4.831061136186787e-07, | |
| "loss": 0.8293, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.8667141838916608, | |
| "grad_norm": 1.4272425648005087, | |
| "learning_rate": 4.7383169088817407e-07, | |
| "loss": 0.819, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.8680558467150224, | |
| "grad_norm": 1.2840790693801114, | |
| "learning_rate": 4.646427262447034e-07, | |
| "loss": 0.8333, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.8693975095383841, | |
| "grad_norm": 1.1670977397189615, | |
| "learning_rate": 4.555393931841001e-07, | |
| "loss": 0.8308, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.8707391723617458, | |
| "grad_norm": 1.2684960800853433, | |
| "learning_rate": 4.4652186358539573e-07, | |
| "loss": 0.8235, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.8720808351851076, | |
| "grad_norm": 1.3198240704084807, | |
| "learning_rate": 4.3759030770757606e-07, | |
| "loss": 0.8321, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.8734224980084693, | |
| "grad_norm": 1.5325532454995796, | |
| "learning_rate": 4.287448941863692e-07, | |
| "loss": 0.8383, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.874764160831831, | |
| "grad_norm": 1.2440514719555422, | |
| "learning_rate": 4.1998579003105553e-07, | |
| "loss": 0.8268, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.8761058236551926, | |
| "grad_norm": 1.0964882598585846, | |
| "learning_rate": 4.1131316062131944e-07, | |
| "loss": 0.8355, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.8774474864785543, | |
| "grad_norm": 1.5114566436301768, | |
| "learning_rate": 4.0272716970412516e-07, | |
| "loss": 0.8283, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.8787891493019161, | |
| "grad_norm": 1.5190567980143286, | |
| "learning_rate": 3.9422797939062505e-07, | |
| "loss": 0.8298, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.8801308121252778, | |
| "grad_norm": 1.5755090256907556, | |
| "learning_rate": 3.858157501530974e-07, | |
| "loss": 0.8237, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.8814724749486395, | |
| "grad_norm": 1.4325123056785103, | |
| "learning_rate": 3.7749064082191976e-07, | |
| "loss": 0.8134, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.8828141377720011, | |
| "grad_norm": 1.1497375202447615, | |
| "learning_rate": 3.692528085825675e-07, | |
| "loss": 0.823, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.8841558005953629, | |
| "grad_norm": 1.1628018423072728, | |
| "learning_rate": 3.6110240897264727e-07, | |
| "loss": 0.8465, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.8854974634187246, | |
| "grad_norm": 1.1977991039877944, | |
| "learning_rate": 3.53039595878959e-07, | |
| "loss": 0.8207, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.8868391262420863, | |
| "grad_norm": 1.3418642621744674, | |
| "learning_rate": 3.450645215345921e-07, | |
| "loss": 0.8633, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.888180789065448, | |
| "grad_norm": 1.3920126895079805, | |
| "learning_rate": 3.3717733651604967e-07, | |
| "loss": 0.8308, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.8895224518888097, | |
| "grad_norm": 1.866265110213513, | |
| "learning_rate": 3.2937818974040637e-07, | |
| "loss": 0.8182, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.8908641147121714, | |
| "grad_norm": 1.1162071914685536, | |
| "learning_rate": 3.216672284624961e-07, | |
| "loss": 0.8273, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.8922057775355331, | |
| "grad_norm": 1.2563738905464021, | |
| "learning_rate": 3.140445982721324e-07, | |
| "loss": 0.8267, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.8935474403588948, | |
| "grad_norm": 1.0883224083010214, | |
| "learning_rate": 3.0651044309136016e-07, | |
| "loss": 0.8255, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.8948891031822566, | |
| "grad_norm": 1.1678392876811903, | |
| "learning_rate": 2.990649051717348e-07, | |
| "loss": 0.8203, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.8962307660056182, | |
| "grad_norm": 1.5578498393574587, | |
| "learning_rate": 2.917081250916415e-07, | |
| "loss": 0.8372, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.8975724288289799, | |
| "grad_norm": 1.2639468775994296, | |
| "learning_rate": 2.844402417536374e-07, | |
| "loss": 0.8448, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.8989140916523416, | |
| "grad_norm": 1.3022881353300302, | |
| "learning_rate": 2.772613923818301e-07, | |
| "loss": 0.8408, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.9002557544757033, | |
| "grad_norm": 1.2098380243126627, | |
| "learning_rate": 2.701717125192865e-07, | |
| "loss": 0.8056, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.9015974172990651, | |
| "grad_norm": 1.1577098434387416, | |
| "learning_rate": 2.631713360254734e-07, | |
| "loss": 0.8275, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.9029390801224267, | |
| "grad_norm": 1.2505947096110512, | |
| "learning_rate": 2.562603950737319e-07, | |
| "loss": 0.8323, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.9042807429457884, | |
| "grad_norm": 1.272901431558626, | |
| "learning_rate": 2.494390201487795e-07, | |
| "loss": 0.8286, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.9056224057691501, | |
| "grad_norm": 1.0741504442823462, | |
| "learning_rate": 2.4270734004424643e-07, | |
| "loss": 0.8268, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.9069640685925119, | |
| "grad_norm": 1.7396571960732488, | |
| "learning_rate": 2.3606548186024702e-07, | |
| "loss": 0.8365, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.9083057314158736, | |
| "grad_norm": 1.1226209969689507, | |
| "learning_rate": 2.295135710009755e-07, | |
| "loss": 0.8349, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.9096473942392352, | |
| "grad_norm": 1.465650474736467, | |
| "learning_rate": 2.2305173117234236e-07, | |
| "loss": 0.8279, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.9109890570625969, | |
| "grad_norm": 1.1481085942694036, | |
| "learning_rate": 2.166800843796357e-07, | |
| "loss": 0.8365, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.9123307198859587, | |
| "grad_norm": 1.1243474016417694, | |
| "learning_rate": 2.1039875092521978e-07, | |
| "loss": 0.806, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.9136723827093204, | |
| "grad_norm": 1.3673908785874032, | |
| "learning_rate": 2.042078494062616e-07, | |
| "loss": 0.8236, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.9150140455326821, | |
| "grad_norm": 4.451908001269186, | |
| "learning_rate": 1.9810749671249353e-07, | |
| "loss": 0.8254, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.9163557083560437, | |
| "grad_norm": 1.099328030938635, | |
| "learning_rate": 1.920978080240049e-07, | |
| "loss": 0.8267, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.9176973711794055, | |
| "grad_norm": 1.0822716090536202, | |
| "learning_rate": 1.861788968090683e-07, | |
| "loss": 0.8391, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.9190390340027672, | |
| "grad_norm": 1.3695703896869942, | |
| "learning_rate": 1.8035087482199676e-07, | |
| "loss": 0.8286, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.9203806968261289, | |
| "grad_norm": 1.2862195635452058, | |
| "learning_rate": 1.7461385210103377e-07, | |
| "loss": 0.849, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.9217223596494906, | |
| "grad_norm": 1.1086447563465651, | |
| "learning_rate": 1.68967936966275e-07, | |
| "loss": 0.835, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.9230640224728522, | |
| "grad_norm": 1.1899606187907275, | |
| "learning_rate": 1.6341323601762548e-07, | |
| "loss": 0.8393, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.924405685296214, | |
| "grad_norm": 1.1341825064809596, | |
| "learning_rate": 1.579498541327834e-07, | |
| "loss": 0.8117, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.9257473481195757, | |
| "grad_norm": 1.125774210111301, | |
| "learning_rate": 1.5257789446526172e-07, | |
| "loss": 0.826, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.9270890109429374, | |
| "grad_norm": 1.1980920240979953, | |
| "learning_rate": 1.4729745844244302e-07, | |
| "loss": 0.8334, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.9284306737662991, | |
| "grad_norm": 1.1410475304396959, | |
| "learning_rate": 1.4210864576365891e-07, | |
| "loss": 0.8238, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.9297723365896609, | |
| "grad_norm": 1.565218041447324, | |
| "learning_rate": 1.3701155439831249e-07, | |
| "loss": 0.8229, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.9311139994130225, | |
| "grad_norm": 1.211246357313267, | |
| "learning_rate": 1.320062805840261e-07, | |
| "loss": 0.825, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.9324556622363842, | |
| "grad_norm": 1.1376520906501437, | |
| "learning_rate": 1.27092918824826e-07, | |
| "loss": 0.8401, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.9337973250597459, | |
| "grad_norm": 1.1992979165042439, | |
| "learning_rate": 1.2227156188935552e-07, | |
| "loss": 0.8378, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.9351389878831077, | |
| "grad_norm": 1.368524983473057, | |
| "learning_rate": 1.1754230080912588e-07, | |
| "loss": 0.8344, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.9364806507064694, | |
| "grad_norm": 1.154202180196164, | |
| "learning_rate": 1.12905224876797e-07, | |
| "loss": 0.8372, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.937822313529831, | |
| "grad_norm": 1.3040185235533615, | |
| "learning_rate": 1.0836042164448945e-07, | |
| "loss": 0.8187, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.9391639763531927, | |
| "grad_norm": 1.2436563494742172, | |
| "learning_rate": 1.0390797692213516e-07, | |
| "loss": 0.8121, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.9405056391765545, | |
| "grad_norm": 1.407746826814002, | |
| "learning_rate": 9.954797477585376e-08, | |
| "loss": 0.84, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.9418473019999162, | |
| "grad_norm": 1.2930670232281178, | |
| "learning_rate": 9.528049752636714e-08, | |
| "loss": 0.8342, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.9431889648232779, | |
| "grad_norm": 1.272412357982192, | |
| "learning_rate": 9.110562574744519e-08, | |
| "loss": 0.8439, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.9445306276466395, | |
| "grad_norm": 1.0898984299687087, | |
| "learning_rate": 8.702343826438364e-08, | |
| "loss": 0.8377, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.9458722904700012, | |
| "grad_norm": 1.2212871617805348, | |
| "learning_rate": 8.303401215251583e-08, | |
| "loss": 0.8242, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.947213953293363, | |
| "grad_norm": 1.2286273772361522, | |
| "learning_rate": 7.913742273575886e-08, | |
| "loss": 0.8202, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.9485556161167247, | |
| "grad_norm": 1.2646180959035689, | |
| "learning_rate": 7.533374358518974e-08, | |
| "loss": 0.8264, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.9498972789400864, | |
| "grad_norm": 1.2006902919443605, | |
| "learning_rate": 7.16230465176565e-08, | |
| "loss": 0.8437, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.951238941763448, | |
| "grad_norm": 1.100152570167147, | |
| "learning_rate": 6.80054015944237e-08, | |
| "loss": 0.838, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.9525806045868098, | |
| "grad_norm": 2.4604943781028936, | |
| "learning_rate": 6.448087711984796e-08, | |
| "loss": 0.8495, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.9539222674101715, | |
| "grad_norm": 1.4080502126651815, | |
| "learning_rate": 6.104953964008897e-08, | |
| "loss": 0.8301, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.9552639302335332, | |
| "grad_norm": 1.1156687287041127, | |
| "learning_rate": 5.7711453941852736e-08, | |
| "loss": 0.8413, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.9566055930568949, | |
| "grad_norm": 1.4544622431056033, | |
| "learning_rate": 5.446668305116865e-08, | |
| "loss": 0.8342, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.9579472558802566, | |
| "grad_norm": 1.3812443826808116, | |
| "learning_rate": 5.1315288232201e-08, | |
| "loss": 0.8298, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.9592889187036183, | |
| "grad_norm": 1.272470737426276, | |
| "learning_rate": 4.825732898608826e-08, | |
| "loss": 0.8457, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.96063058152698, | |
| "grad_norm": 1.11385367991356, | |
| "learning_rate": 4.529286304982394e-08, | |
| "loss": 0.8418, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.9619722443503417, | |
| "grad_norm": 1.1417145321246553, | |
| "learning_rate": 4.2421946395164174e-08, | |
| "loss": 0.8454, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.9633139071737034, | |
| "grad_norm": 2.043018715590093, | |
| "learning_rate": 3.964463322757017e-08, | |
| "loss": 0.8201, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.9646555699970651, | |
| "grad_norm": 1.1399272549759423, | |
| "learning_rate": 3.696097598518855e-08, | |
| "loss": 0.8307, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.9659972328204268, | |
| "grad_norm": 1.8296136933263256, | |
| "learning_rate": 3.437102533785541e-08, | |
| "loss": 0.8132, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.9673388956437885, | |
| "grad_norm": 1.0854050320953994, | |
| "learning_rate": 3.187483018614601e-08, | |
| "loss": 0.8305, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.9686805584671502, | |
| "grad_norm": 1.188116758024328, | |
| "learning_rate": 2.9472437660446605e-08, | |
| "loss": 0.8298, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.970022221290512, | |
| "grad_norm": 1.5164652354075294, | |
| "learning_rate": 2.7163893120066288e-08, | |
| "loss": 0.83, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.9713638841138736, | |
| "grad_norm": 1.058567614974338, | |
| "learning_rate": 2.4949240152381536e-08, | |
| "loss": 0.8443, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.9727055469372353, | |
| "grad_norm": 1.2993064499074076, | |
| "learning_rate": 2.2828520572011902e-08, | |
| "loss": 0.8238, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.974047209760597, | |
| "grad_norm": 1.0143058955957016, | |
| "learning_rate": 2.0801774420031172e-08, | |
| "loss": 0.8162, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.9753888725839588, | |
| "grad_norm": 1.1041378492352216, | |
| "learning_rate": 1.8869039963210766e-08, | |
| "loss": 0.8465, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.9767305354073205, | |
| "grad_norm": 1.264125443357894, | |
| "learning_rate": 1.7030353693298086e-08, | |
| "loss": 0.8374, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.9780721982306821, | |
| "grad_norm": 1.0809229111040999, | |
| "learning_rate": 1.5285750326325953e-08, | |
| "loss": 0.8452, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.9794138610540438, | |
| "grad_norm": 1.2887412115766264, | |
| "learning_rate": 1.3635262801960369e-08, | |
| "loss": 0.8238, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.9807555238774056, | |
| "grad_norm": 1.0998551154972078, | |
| "learning_rate": 1.2078922282873773e-08, | |
| "loss": 0.8245, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.9820971867007673, | |
| "grad_norm": 1.4768247486368353, | |
| "learning_rate": 1.0616758154161633e-08, | |
| "loss": 0.8223, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.983438849524129, | |
| "grad_norm": 1.4165507742652732, | |
| "learning_rate": 9.248798022784001e-09, | |
| "loss": 0.8147, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.9847805123474906, | |
| "grad_norm": 1.0904365459718672, | |
| "learning_rate": 7.975067717045926e-09, | |
| "loss": 0.8377, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.9861221751708523, | |
| "grad_norm": 1.3307578082312725, | |
| "learning_rate": 6.7955912861095155e-09, | |
| "loss": 0.8409, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.9874638379942141, | |
| "grad_norm": 1.4992943559838845, | |
| "learning_rate": 5.71039099953985e-09, | |
| "loss": 0.827, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.9888055008175758, | |
| "grad_norm": 1.7051724267125627, | |
| "learning_rate": 4.719487346884211e-09, | |
| "loss": 0.8157, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.9901471636409375, | |
| "grad_norm": 1.3352668590748493, | |
| "learning_rate": 3.822899037286276e-09, | |
| "loss": 0.8275, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.9914888264642991, | |
| "grad_norm": 1.130618761765227, | |
| "learning_rate": 3.0206429991314067e-09, | |
| "loss": 0.8492, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.9928304892876609, | |
| "grad_norm": 1.309284467806641, | |
| "learning_rate": 2.3127343797269e-09, | |
| "loss": 0.8215, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.9941721521110226, | |
| "grad_norm": 2.445924082786577, | |
| "learning_rate": 1.6991865450188827e-09, | |
| "loss": 0.8045, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.9955138149343843, | |
| "grad_norm": 1.2477052455873365, | |
| "learning_rate": 1.1800110793358521e-09, | |
| "loss": 0.8161, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.996855477757746, | |
| "grad_norm": 1.1128754412911415, | |
| "learning_rate": 7.552177851732901e-10, | |
| "loss": 0.8337, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.9981971405811078, | |
| "grad_norm": 1.2593829479740748, | |
| "learning_rate": 4.2481468300603625e-10, | |
| "loss": 0.8547, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.9995388034044694, | |
| "grad_norm": 1.8056331069680909, | |
| "learning_rate": 1.8880801113951853e-10, | |
| "loss": 0.8159, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.9995388034044694, | |
| "step": 745, | |
| "total_flos": 1501298433523712.0, | |
| "train_loss": 0.24619540928194186, | |
| "train_runtime": 14818.31, | |
| "train_samples_per_second": 25.753, | |
| "train_steps_per_second": 0.05 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 746, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 25, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1501298433523712.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |