| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1932, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0005175983436853002, | |
| "grad_norm": 118.52372570307482, | |
| "learning_rate": 1.0309278350515465e-07, | |
| "loss": 3.0995, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.002587991718426501, | |
| "grad_norm": 106.06460811089278, | |
| "learning_rate": 5.154639175257732e-07, | |
| "loss": 3.2515, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.005175983436853002, | |
| "grad_norm": 45.432580654463486, | |
| "learning_rate": 1.0309278350515464e-06, | |
| "loss": 2.5474, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.007763975155279503, | |
| "grad_norm": 5.707645113932802, | |
| "learning_rate": 1.5463917525773197e-06, | |
| "loss": 1.6318, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.010351966873706004, | |
| "grad_norm": 4.4628083580213325, | |
| "learning_rate": 2.061855670103093e-06, | |
| "loss": 1.3823, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.012939958592132506, | |
| "grad_norm": 3.761000828527356, | |
| "learning_rate": 2.577319587628866e-06, | |
| "loss": 1.3176, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.015527950310559006, | |
| "grad_norm": 3.0497152788250235, | |
| "learning_rate": 3.0927835051546395e-06, | |
| "loss": 1.1812, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.018115942028985508, | |
| "grad_norm": 2.696900373411112, | |
| "learning_rate": 3.6082474226804126e-06, | |
| "loss": 1.274, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.020703933747412008, | |
| "grad_norm": 2.363119260182366, | |
| "learning_rate": 4.123711340206186e-06, | |
| "loss": 1.2032, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.023291925465838508, | |
| "grad_norm": 2.549100089774392, | |
| "learning_rate": 4.639175257731959e-06, | |
| "loss": 1.2152, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.025879917184265012, | |
| "grad_norm": 2.8330026439235567, | |
| "learning_rate": 5.154639175257732e-06, | |
| "loss": 1.1962, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.028467908902691512, | |
| "grad_norm": 2.4165498298939156, | |
| "learning_rate": 5.670103092783505e-06, | |
| "loss": 1.211, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.031055900621118012, | |
| "grad_norm": 2.409145711425535, | |
| "learning_rate": 6.185567010309279e-06, | |
| "loss": 1.2127, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.03364389233954451, | |
| "grad_norm": 2.5761976363169863, | |
| "learning_rate": 6.701030927835052e-06, | |
| "loss": 1.2057, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.036231884057971016, | |
| "grad_norm": 2.476050413527567, | |
| "learning_rate": 7.216494845360825e-06, | |
| "loss": 1.2286, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.03881987577639751, | |
| "grad_norm": 2.467699606641844, | |
| "learning_rate": 7.731958762886599e-06, | |
| "loss": 1.2134, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.041407867494824016, | |
| "grad_norm": 2.2939966741651387, | |
| "learning_rate": 8.247422680412371e-06, | |
| "loss": 1.1893, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.04399585921325052, | |
| "grad_norm": 2.762524555569291, | |
| "learning_rate": 8.762886597938146e-06, | |
| "loss": 1.2244, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.046583850931677016, | |
| "grad_norm": 2.1801186709719245, | |
| "learning_rate": 9.278350515463918e-06, | |
| "loss": 1.1464, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.04917184265010352, | |
| "grad_norm": 2.1622793216147542, | |
| "learning_rate": 9.793814432989691e-06, | |
| "loss": 1.1941, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.051759834368530024, | |
| "grad_norm": 2.327680697741413, | |
| "learning_rate": 1.0309278350515464e-05, | |
| "loss": 1.2271, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05434782608695652, | |
| "grad_norm": 2.417174235586904, | |
| "learning_rate": 1.0824742268041238e-05, | |
| "loss": 1.2522, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.056935817805383024, | |
| "grad_norm": 2.1527719549162843, | |
| "learning_rate": 1.134020618556701e-05, | |
| "loss": 1.2158, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.05952380952380952, | |
| "grad_norm": 2.089884090927776, | |
| "learning_rate": 1.1855670103092785e-05, | |
| "loss": 1.1686, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.062111801242236024, | |
| "grad_norm": 2.087619755097997, | |
| "learning_rate": 1.2371134020618558e-05, | |
| "loss": 1.1924, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.06469979296066253, | |
| "grad_norm": 2.159539140677949, | |
| "learning_rate": 1.2886597938144332e-05, | |
| "loss": 1.2169, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.06728778467908902, | |
| "grad_norm": 2.7014157342913996, | |
| "learning_rate": 1.3402061855670103e-05, | |
| "loss": 1.2805, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.06987577639751552, | |
| "grad_norm": 2.0065957039857434, | |
| "learning_rate": 1.3917525773195878e-05, | |
| "loss": 1.1845, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.07246376811594203, | |
| "grad_norm": 2.6446222341165266, | |
| "learning_rate": 1.443298969072165e-05, | |
| "loss": 1.2289, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.07505175983436853, | |
| "grad_norm": 2.115637544048349, | |
| "learning_rate": 1.4948453608247425e-05, | |
| "loss": 1.2695, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.07763975155279502, | |
| "grad_norm": 2.0387064738115095, | |
| "learning_rate": 1.5463917525773197e-05, | |
| "loss": 1.2664, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08022774327122154, | |
| "grad_norm": 2.074102510035905, | |
| "learning_rate": 1.597938144329897e-05, | |
| "loss": 1.1976, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.08281573498964803, | |
| "grad_norm": 2.871459263076559, | |
| "learning_rate": 1.6494845360824743e-05, | |
| "loss": 1.2579, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.08540372670807453, | |
| "grad_norm": 1.9920241256865008, | |
| "learning_rate": 1.7010309278350517e-05, | |
| "loss": 1.3046, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.08799171842650104, | |
| "grad_norm": 2.0573548308344707, | |
| "learning_rate": 1.752577319587629e-05, | |
| "loss": 1.2717, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.09057971014492754, | |
| "grad_norm": 1.8934736779126353, | |
| "learning_rate": 1.8041237113402062e-05, | |
| "loss": 1.2516, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.09316770186335403, | |
| "grad_norm": 1.9513462230819247, | |
| "learning_rate": 1.8556701030927837e-05, | |
| "loss": 1.275, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.09575569358178054, | |
| "grad_norm": 1.9681874455253996, | |
| "learning_rate": 1.907216494845361e-05, | |
| "loss": 1.2622, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.09834368530020704, | |
| "grad_norm": 2.168981469623358, | |
| "learning_rate": 1.9587628865979382e-05, | |
| "loss": 1.2687, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.10093167701863354, | |
| "grad_norm": 1.9357981604266234, | |
| "learning_rate": 1.999998366308358e-05, | |
| "loss": 1.2902, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.10351966873706005, | |
| "grad_norm": 1.9716824691797907, | |
| "learning_rate": 1.9999411876613578e-05, | |
| "loss": 1.2445, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.10610766045548654, | |
| "grad_norm": 1.9437244964639826, | |
| "learning_rate": 1.9998023297700656e-05, | |
| "loss": 1.2473, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.10869565217391304, | |
| "grad_norm": 2.0447501542042135, | |
| "learning_rate": 1.9995818039769564e-05, | |
| "loss": 1.3562, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.11128364389233955, | |
| "grad_norm": 2.136938650583081, | |
| "learning_rate": 1.999279628295469e-05, | |
| "loss": 1.2552, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.11387163561076605, | |
| "grad_norm": 1.9517996473885892, | |
| "learning_rate": 1.9988958274085367e-05, | |
| "loss": 1.2911, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.11645962732919254, | |
| "grad_norm": 1.7952101956790416, | |
| "learning_rate": 1.99843043266657e-05, | |
| "loss": 1.2789, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.11904761904761904, | |
| "grad_norm": 2.0601437201646275, | |
| "learning_rate": 1.9978834820848952e-05, | |
| "loss": 1.3249, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.12163561076604555, | |
| "grad_norm": 1.9179791275324676, | |
| "learning_rate": 1.99725502034065e-05, | |
| "loss": 1.2859, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.12422360248447205, | |
| "grad_norm": 2.146343845103461, | |
| "learning_rate": 1.996545098769134e-05, | |
| "loss": 1.259, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.12681159420289856, | |
| "grad_norm": 1.9311811617661367, | |
| "learning_rate": 1.995753775359615e-05, | |
| "loss": 1.2773, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.12939958592132506, | |
| "grad_norm": 1.8858776139378066, | |
| "learning_rate": 1.994881114750593e-05, | |
| "loss": 1.2574, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.13198757763975155, | |
| "grad_norm": 1.8059315541403178, | |
| "learning_rate": 1.993927188224519e-05, | |
| "loss": 1.2613, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.13457556935817805, | |
| "grad_norm": 1.7923594093741397, | |
| "learning_rate": 1.9928920737019735e-05, | |
| "loss": 1.2602, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.13716356107660455, | |
| "grad_norm": 1.774108800205075, | |
| "learning_rate": 1.991775855735301e-05, | |
| "loss": 1.3387, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.13975155279503104, | |
| "grad_norm": 1.913850446875917, | |
| "learning_rate": 1.9905786255017046e-05, | |
| "loss": 1.3221, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.14233954451345757, | |
| "grad_norm": 1.9693465712493503, | |
| "learning_rate": 1.9893004807957957e-05, | |
| "loss": 1.2983, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.14492753623188406, | |
| "grad_norm": 1.790233528375858, | |
| "learning_rate": 1.987941526021609e-05, | |
| "loss": 1.2954, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.14751552795031056, | |
| "grad_norm": 2.8906112799819574, | |
| "learning_rate": 1.9865018721840708e-05, | |
| "loss": 1.2964, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.15010351966873706, | |
| "grad_norm": 1.9226491916325386, | |
| "learning_rate": 1.9849816368799356e-05, | |
| "loss": 1.299, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.15269151138716355, | |
| "grad_norm": 1.9512005440074536, | |
| "learning_rate": 1.9833809442881776e-05, | |
| "loss": 1.323, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.15527950310559005, | |
| "grad_norm": 1.7853859665144927, | |
| "learning_rate": 1.981699925159847e-05, | |
| "loss": 1.3069, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.15786749482401657, | |
| "grad_norm": 1.7547957709103652, | |
| "learning_rate": 1.979938716807392e-05, | |
| "loss": 1.3264, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.16045548654244307, | |
| "grad_norm": 1.852378431981816, | |
| "learning_rate": 1.978097463093441e-05, | |
| "loss": 1.2946, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.16304347826086957, | |
| "grad_norm": 1.8256299629827946, | |
| "learning_rate": 1.976176314419051e-05, | |
| "loss": 1.2683, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.16563146997929606, | |
| "grad_norm": 1.8017450986442927, | |
| "learning_rate": 1.974175427711423e-05, | |
| "loss": 1.3127, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.16821946169772256, | |
| "grad_norm": 1.8676977703064013, | |
| "learning_rate": 1.9720949664110843e-05, | |
| "loss": 1.3071, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.17080745341614906, | |
| "grad_norm": 1.806177669035387, | |
| "learning_rate": 1.9699351004585354e-05, | |
| "loss": 1.2485, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.17339544513457558, | |
| "grad_norm": 1.7075013810392663, | |
| "learning_rate": 1.9676960062803714e-05, | |
| "loss": 1.2904, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.17598343685300208, | |
| "grad_norm": 1.7968766744551163, | |
| "learning_rate": 1.9653778667748695e-05, | |
| "loss": 1.3346, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.17857142857142858, | |
| "grad_norm": 2.2546697403830587, | |
| "learning_rate": 1.962980871297049e-05, | |
| "loss": 1.2952, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.18115942028985507, | |
| "grad_norm": 1.855772798318876, | |
| "learning_rate": 1.9605052156432042e-05, | |
| "loss": 1.243, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.18374741200828157, | |
| "grad_norm": 1.6468966576864474, | |
| "learning_rate": 1.9579511020349117e-05, | |
| "loss": 1.2897, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.18633540372670807, | |
| "grad_norm": 1.7882764084793803, | |
| "learning_rate": 1.9553187391025102e-05, | |
| "loss": 1.2678, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.18892339544513456, | |
| "grad_norm": 1.8772543986524286, | |
| "learning_rate": 1.952608341868061e-05, | |
| "loss": 1.3172, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.1915113871635611, | |
| "grad_norm": 1.648896837108699, | |
| "learning_rate": 1.949820131727783e-05, | |
| "loss": 1.2923, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.19409937888198758, | |
| "grad_norm": 1.7068857061711549, | |
| "learning_rate": 1.9469543364339674e-05, | |
| "loss": 1.3209, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.19668737060041408, | |
| "grad_norm": 1.6145254855564042, | |
| "learning_rate": 1.9440111900763758e-05, | |
| "loss": 1.273, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.19927536231884058, | |
| "grad_norm": 1.7464698166979602, | |
| "learning_rate": 1.940990933063118e-05, | |
| "loss": 1.3115, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.20186335403726707, | |
| "grad_norm": 2.237780229221515, | |
| "learning_rate": 1.9378938121010128e-05, | |
| "loss": 1.3093, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.20445134575569357, | |
| "grad_norm": 1.8816106738657112, | |
| "learning_rate": 1.9347200801754395e-05, | |
| "loss": 1.2999, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.2070393374741201, | |
| "grad_norm": 1.7444438072126305, | |
| "learning_rate": 1.9314699965296694e-05, | |
| "loss": 1.2824, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2096273291925466, | |
| "grad_norm": 2.048386739159661, | |
| "learning_rate": 1.9281438266436923e-05, | |
| "loss": 1.2933, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.2122153209109731, | |
| "grad_norm": 3.306515280854177, | |
| "learning_rate": 1.92474184221253e-05, | |
| "loss": 1.295, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.21480331262939958, | |
| "grad_norm": 1.5615788627848504, | |
| "learning_rate": 1.9212643211240433e-05, | |
| "loss": 1.2794, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.21739130434782608, | |
| "grad_norm": 1.6191612683725647, | |
| "learning_rate": 1.917711547436232e-05, | |
| "loss": 1.2846, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.21997929606625258, | |
| "grad_norm": 1.7354290362475042, | |
| "learning_rate": 1.9140838113540347e-05, | |
| "loss": 1.2621, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.2225672877846791, | |
| "grad_norm": 1.779070935639009, | |
| "learning_rate": 1.9103814092056205e-05, | |
| "loss": 1.3176, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.2251552795031056, | |
| "grad_norm": 1.9188634938137341, | |
| "learning_rate": 1.9066046434181854e-05, | |
| "loss": 1.3437, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.2277432712215321, | |
| "grad_norm": 1.6905846039991461, | |
| "learning_rate": 1.902753822493248e-05, | |
| "loss": 1.2722, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.2303312629399586, | |
| "grad_norm": 1.7568380824359298, | |
| "learning_rate": 1.8988292609814513e-05, | |
| "loss": 1.2728, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.2329192546583851, | |
| "grad_norm": 1.664244033715747, | |
| "learning_rate": 1.8948312794568674e-05, | |
| "loss": 1.2558, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.23550724637681159, | |
| "grad_norm": 1.6016572217912088, | |
| "learning_rate": 1.8907602044908125e-05, | |
| "loss": 1.2908, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.23809523809523808, | |
| "grad_norm": 1.7484345463876294, | |
| "learning_rate": 1.886616368625171e-05, | |
| "loss": 1.3261, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.2406832298136646, | |
| "grad_norm": 1.7212899167763012, | |
| "learning_rate": 1.8824001103452316e-05, | |
| "loss": 1.2625, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.2432712215320911, | |
| "grad_norm": 1.6482436683937527, | |
| "learning_rate": 1.8781117740520386e-05, | |
| "loss": 1.2785, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.2458592132505176, | |
| "grad_norm": 1.5757960105455753, | |
| "learning_rate": 1.8737517100342624e-05, | |
| "loss": 1.2733, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.2484472049689441, | |
| "grad_norm": 1.64845152675654, | |
| "learning_rate": 1.869320274439583e-05, | |
| "loss": 1.3033, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.2510351966873706, | |
| "grad_norm": 1.8462235169663426, | |
| "learning_rate": 1.864817829245598e-05, | |
| "loss": 1.3084, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.2536231884057971, | |
| "grad_norm": 1.7739351798022618, | |
| "learning_rate": 1.86024474223026e-05, | |
| "loss": 1.2938, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.2562111801242236, | |
| "grad_norm": 1.628868267103692, | |
| "learning_rate": 1.8556013869418282e-05, | |
| "loss": 1.2852, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.2587991718426501, | |
| "grad_norm": 1.6457493283632671, | |
| "learning_rate": 1.8508881426683618e-05, | |
| "loss": 1.2341, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.2613871635610766, | |
| "grad_norm": 1.5598114432949137, | |
| "learning_rate": 1.8461053944067324e-05, | |
| "loss": 1.3068, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.2639751552795031, | |
| "grad_norm": 1.7427944903352592, | |
| "learning_rate": 1.8412535328311813e-05, | |
| "loss": 1.2866, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.2665631469979296, | |
| "grad_norm": 1.7159655860891432, | |
| "learning_rate": 1.8363329542614033e-05, | |
| "loss": 1.2987, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.2691511387163561, | |
| "grad_norm": 1.6937281668589332, | |
| "learning_rate": 1.831344060630176e-05, | |
| "loss": 1.2861, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.2717391304347826, | |
| "grad_norm": 1.7317450117613962, | |
| "learning_rate": 1.826287259450528e-05, | |
| "loss": 1.3135, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.2743271221532091, | |
| "grad_norm": 1.736815145364657, | |
| "learning_rate": 1.8211629637824516e-05, | |
| "loss": 1.3229, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.2769151138716356, | |
| "grad_norm": 1.717368110143028, | |
| "learning_rate": 1.8159715921991612e-05, | |
| "loss": 1.3, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.2795031055900621, | |
| "grad_norm": 1.7858279813792615, | |
| "learning_rate": 1.8107135687529044e-05, | |
| "loss": 1.304, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.28209109730848864, | |
| "grad_norm": 1.6420365658498397, | |
| "learning_rate": 1.8053893229403218e-05, | |
| "loss": 1.2746, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.28467908902691513, | |
| "grad_norm": 1.781865206777271, | |
| "learning_rate": 1.7999992896673667e-05, | |
| "loss": 1.2941, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.28726708074534163, | |
| "grad_norm": 1.717576281479697, | |
| "learning_rate": 1.7945439092137763e-05, | |
| "loss": 1.2934, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.2898550724637681, | |
| "grad_norm": 1.8605583484947545, | |
| "learning_rate": 1.789023627197112e-05, | |
| "loss": 1.2609, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.2924430641821946, | |
| "grad_norm": 1.7448230138089384, | |
| "learning_rate": 1.783438894536357e-05, | |
| "loss": 1.297, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.2950310559006211, | |
| "grad_norm": 1.7954544554359166, | |
| "learning_rate": 1.7777901674150827e-05, | |
| "loss": 1.2839, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.2976190476190476, | |
| "grad_norm": 1.82466567221307, | |
| "learning_rate": 1.7720779072441897e-05, | |
| "loss": 1.2602, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.3002070393374741, | |
| "grad_norm": 1.6162229238306363, | |
| "learning_rate": 1.7663025806242126e-05, | |
| "loss": 1.2819, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.3027950310559006, | |
| "grad_norm": 1.7224353038885882, | |
| "learning_rate": 1.7604646593072107e-05, | |
| "loss": 1.2789, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.3053830227743271, | |
| "grad_norm": 1.5302928069449526, | |
| "learning_rate": 1.7545646201582304e-05, | |
| "loss": 1.2741, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.3079710144927536, | |
| "grad_norm": 1.7278169638412462, | |
| "learning_rate": 1.7486029451163552e-05, | |
| "loss": 1.2761, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.3105590062111801, | |
| "grad_norm": 7.66316697414024, | |
| "learning_rate": 1.742580121155337e-05, | |
| "loss": 1.2848, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.31314699792960665, | |
| "grad_norm": 1.7170749939899528, | |
| "learning_rate": 1.73649664024382e-05, | |
| "loss": 1.3145, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.31573498964803315, | |
| "grad_norm": 1.5882256935177697, | |
| "learning_rate": 1.730352999305152e-05, | |
| "loss": 1.2437, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.31832298136645965, | |
| "grad_norm": 1.6451039487590284, | |
| "learning_rate": 1.7241497001767967e-05, | |
| "loss": 1.3017, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.32091097308488614, | |
| "grad_norm": 1.6109005034073363, | |
| "learning_rate": 1.7178872495693397e-05, | |
| "loss": 1.264, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.32349896480331264, | |
| "grad_norm": 1.641706658754867, | |
| "learning_rate": 1.7115661590250992e-05, | |
| "loss": 1.2764, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.32608695652173914, | |
| "grad_norm": 1.8516711248065172, | |
| "learning_rate": 1.7051869448763406e-05, | |
| "loss": 1.276, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.32867494824016563, | |
| "grad_norm": 1.5631168769858184, | |
| "learning_rate": 1.6987501282031013e-05, | |
| "loss": 1.2148, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.33126293995859213, | |
| "grad_norm": 1.6459960001672511, | |
| "learning_rate": 1.692256234790624e-05, | |
| "loss": 1.2473, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.3338509316770186, | |
| "grad_norm": 2.9555007934323703, | |
| "learning_rate": 1.6857057950864134e-05, | |
| "loss": 1.2874, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.3364389233954451, | |
| "grad_norm": 1.612379368244149, | |
| "learning_rate": 1.6790993441569022e-05, | |
| "loss": 1.2373, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.3390269151138716, | |
| "grad_norm": 1.6757461678281282, | |
| "learning_rate": 1.6724374216437475e-05, | |
| "loss": 1.2801, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.3416149068322981, | |
| "grad_norm": 1.5726408014891122, | |
| "learning_rate": 1.6657205717197495e-05, | |
| "loss": 1.2065, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.3442028985507246, | |
| "grad_norm": 1.6029270709660246, | |
| "learning_rate": 1.6589493430444026e-05, | |
| "loss": 1.2928, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.34679089026915116, | |
| "grad_norm": 1.7396806928714428, | |
| "learning_rate": 1.6521242887190764e-05, | |
| "loss": 1.3018, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.34937888198757766, | |
| "grad_norm": 1.683802020776293, | |
| "learning_rate": 1.6452459662418386e-05, | |
| "loss": 1.2703, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.35196687370600416, | |
| "grad_norm": 1.6299055475198938, | |
| "learning_rate": 1.638314937461915e-05, | |
| "loss": 1.2465, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.35455486542443065, | |
| "grad_norm": 1.5793607569305301, | |
| "learning_rate": 1.6313317685337947e-05, | |
| "loss": 1.2774, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.35714285714285715, | |
| "grad_norm": 1.6509230878640204, | |
| "learning_rate": 1.6242970298709867e-05, | |
| "loss": 1.2394, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.35973084886128365, | |
| "grad_norm": 1.6564302207780428, | |
| "learning_rate": 1.6172112960994234e-05, | |
| "loss": 1.3113, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.36231884057971014, | |
| "grad_norm": 1.720021595292221, | |
| "learning_rate": 1.6100751460105244e-05, | |
| "loss": 1.2588, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.36490683229813664, | |
| "grad_norm": 1.5844840946364063, | |
| "learning_rate": 1.602889162513919e-05, | |
| "loss": 1.2807, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.36749482401656314, | |
| "grad_norm": 1.6839767345265815, | |
| "learning_rate": 1.59565393258983e-05, | |
| "loss": 1.2607, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.37008281573498963, | |
| "grad_norm": 1.7860415375986487, | |
| "learning_rate": 1.588370047241128e-05, | |
| "loss": 1.256, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.37267080745341613, | |
| "grad_norm": 1.7117135290100387, | |
| "learning_rate": 1.5810381014450557e-05, | |
| "loss": 1.2263, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.3752587991718426, | |
| "grad_norm": 1.5845047105100838, | |
| "learning_rate": 1.5736586941046285e-05, | |
| "loss": 1.242, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.3778467908902691, | |
| "grad_norm": 1.80500551409116, | |
| "learning_rate": 1.5662324279997115e-05, | |
| "loss": 1.2394, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.3804347826086957, | |
| "grad_norm": 1.5963745820078468, | |
| "learning_rate": 1.5587599097377845e-05, | |
| "loss": 1.2124, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.3830227743271222, | |
| "grad_norm": 1.7129591053368485, | |
| "learning_rate": 1.5512417497043898e-05, | |
| "loss": 1.2285, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.38561076604554867, | |
| "grad_norm": 1.947935106696502, | |
| "learning_rate": 1.5436785620132756e-05, | |
| "loss": 1.2297, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.38819875776397517, | |
| "grad_norm": 1.5854599615971694, | |
| "learning_rate": 1.5360709644562296e-05, | |
| "loss": 1.2551, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.39078674948240166, | |
| "grad_norm": 1.6101311120937332, | |
| "learning_rate": 1.5284195784526196e-05, | |
| "loss": 1.2708, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.39337474120082816, | |
| "grad_norm": 1.647707331905149, | |
| "learning_rate": 1.520725028998629e-05, | |
| "loss": 1.266, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.39596273291925466, | |
| "grad_norm": 1.9215587673185384, | |
| "learning_rate": 1.512987944616207e-05, | |
| "loss": 1.2563, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.39855072463768115, | |
| "grad_norm": 1.5923791612452982, | |
| "learning_rate": 1.5052089573017283e-05, | |
| "loss": 1.2951, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.40113871635610765, | |
| "grad_norm": 1.533522718790619, | |
| "learning_rate": 1.4973887024743686e-05, | |
| "loss": 1.2629, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.40372670807453415, | |
| "grad_norm": 1.5392604333005226, | |
| "learning_rate": 1.4895278189242017e-05, | |
| "loss": 1.1853, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.40631469979296064, | |
| "grad_norm": 1.5911615984318879, | |
| "learning_rate": 1.4816269487600184e-05, | |
| "loss": 1.2374, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.40890269151138714, | |
| "grad_norm": 1.540407306505209, | |
| "learning_rate": 1.4736867373568795e-05, | |
| "loss": 1.2208, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.4114906832298137, | |
| "grad_norm": 1.5985263367678237, | |
| "learning_rate": 1.4657078333033967e-05, | |
| "loss": 1.2569, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.4140786749482402, | |
| "grad_norm": 1.5598791416023927, | |
| "learning_rate": 1.4576908883487549e-05, | |
| "loss": 1.184, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4166666666666667, | |
| "grad_norm": 1.564166816397901, | |
| "learning_rate": 1.4496365573494724e-05, | |
| "loss": 1.278, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.4192546583850932, | |
| "grad_norm": 1.5214555732261263, | |
| "learning_rate": 1.4415454982159121e-05, | |
| "loss": 1.2756, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.4218426501035197, | |
| "grad_norm": 1.566924379219736, | |
| "learning_rate": 1.433418371858539e-05, | |
| "loss": 1.2208, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.4244306418219462, | |
| "grad_norm": 1.693241805167907, | |
| "learning_rate": 1.4252558421339355e-05, | |
| "loss": 1.2649, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.42701863354037267, | |
| "grad_norm": 1.845359674560561, | |
| "learning_rate": 1.4170585757905742e-05, | |
| "loss": 1.2263, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.42960662525879917, | |
| "grad_norm": 1.5444000254202592, | |
| "learning_rate": 1.4088272424143546e-05, | |
| "loss": 1.2764, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.43219461697722567, | |
| "grad_norm": 1.5904821648663396, | |
| "learning_rate": 1.4005625143739103e-05, | |
| "loss": 1.2467, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.43478260869565216, | |
| "grad_norm": 1.6406582849973572, | |
| "learning_rate": 1.3922650667656861e-05, | |
| "loss": 1.2466, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.43737060041407866, | |
| "grad_norm": 1.5874801423879288, | |
| "learning_rate": 1.3839355773587932e-05, | |
| "loss": 1.2818, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.43995859213250516, | |
| "grad_norm": 1.5664894322663914, | |
| "learning_rate": 1.3755747265396466e-05, | |
| "loss": 1.2569, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.44254658385093165, | |
| "grad_norm": 1.639337291197075, | |
| "learning_rate": 1.3671831972563889e-05, | |
| "loss": 1.2466, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.4451345755693582, | |
| "grad_norm": 1.4882976719575092, | |
| "learning_rate": 1.3587616749631037e-05, | |
| "loss": 1.2262, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.4477225672877847, | |
| "grad_norm": 1.5575004320774404, | |
| "learning_rate": 1.3503108475638244e-05, | |
| "loss": 1.2685, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.4503105590062112, | |
| "grad_norm": 1.5995590657708154, | |
| "learning_rate": 1.3418314053563455e-05, | |
| "loss": 1.251, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.4528985507246377, | |
| "grad_norm": 1.560631857872418, | |
| "learning_rate": 1.3333240409758328e-05, | |
| "loss": 1.2201, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.4554865424430642, | |
| "grad_norm": 1.5801386912905093, | |
| "learning_rate": 1.3247894493382507e-05, | |
| "loss": 1.2186, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.4580745341614907, | |
| "grad_norm": 1.749634243011622, | |
| "learning_rate": 1.3162283275835938e-05, | |
| "loss": 1.1815, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.4606625258799172, | |
| "grad_norm": 1.6614862019498566, | |
| "learning_rate": 1.3076413750189468e-05, | |
| "loss": 1.2601, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.4632505175983437, | |
| "grad_norm": 1.6371821901498413, | |
| "learning_rate": 1.2990292930613559e-05, | |
| "loss": 1.2368, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.4658385093167702, | |
| "grad_norm": 1.56278259362623, | |
| "learning_rate": 1.2903927851805416e-05, | |
| "loss": 1.1672, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.4684265010351967, | |
| "grad_norm": 1.4842721179109506, | |
| "learning_rate": 1.2817325568414299e-05, | |
| "loss": 1.2425, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.47101449275362317, | |
| "grad_norm": 1.5573631764732299, | |
| "learning_rate": 1.2730493154465311e-05, | |
| "loss": 1.2584, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.47360248447204967, | |
| "grad_norm": 1.5747650735756835, | |
| "learning_rate": 1.2643437702781559e-05, | |
| "loss": 1.2027, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.47619047619047616, | |
| "grad_norm": 1.7147334990616894, | |
| "learning_rate": 1.2556166324404747e-05, | |
| "loss": 1.2369, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.4787784679089027, | |
| "grad_norm": 1.6173781131907976, | |
| "learning_rate": 1.2468686148014379e-05, | |
| "loss": 1.2348, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.4813664596273292, | |
| "grad_norm": 1.6483571836594604, | |
| "learning_rate": 1.2381004319345402e-05, | |
| "loss": 1.3268, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.4839544513457557, | |
| "grad_norm": 1.5421617996607333, | |
| "learning_rate": 1.2293128000604552e-05, | |
| "loss": 1.2666, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.4865424430641822, | |
| "grad_norm": 1.618282632952526, | |
| "learning_rate": 1.2205064369885291e-05, | |
| "loss": 1.2037, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.4891304347826087, | |
| "grad_norm": 1.6786332269059123, | |
| "learning_rate": 1.2116820620581486e-05, | |
| "loss": 1.2555, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.4917184265010352, | |
| "grad_norm": 1.6174195289732147, | |
| "learning_rate": 1.2028403960799821e-05, | |
| "loss": 1.2518, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.4943064182194617, | |
| "grad_norm": 1.49372149520001, | |
| "learning_rate": 1.1939821612771008e-05, | |
| "loss": 1.2294, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.4968944099378882, | |
| "grad_norm": 1.5736827426176978, | |
| "learning_rate": 1.1851080812259839e-05, | |
| "loss": 1.216, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.4994824016563147, | |
| "grad_norm": 1.6096915420259137, | |
| "learning_rate": 1.1762188807974153e-05, | |
| "loss": 1.2033, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.5020703933747412, | |
| "grad_norm": 1.6642406229664035, | |
| "learning_rate": 1.1673152860972725e-05, | |
| "loss": 1.2276, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.5046583850931677, | |
| "grad_norm": 1.5472858925733155, | |
| "learning_rate": 1.158398024407215e-05, | |
| "loss": 1.2135, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.5072463768115942, | |
| "grad_norm": 1.6378928317247718, | |
| "learning_rate": 1.1494678241252781e-05, | |
| "loss": 1.1857, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.5098343685300207, | |
| "grad_norm": 1.669332160090618, | |
| "learning_rate": 1.1405254147063722e-05, | |
| "loss": 1.2294, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.5124223602484472, | |
| "grad_norm": 1.5329461024041375, | |
| "learning_rate": 1.1315715266027014e-05, | |
| "loss": 1.2223, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.5150103519668737, | |
| "grad_norm": 1.643890352883545, | |
| "learning_rate": 1.1226068912040935e-05, | |
| "loss": 1.223, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.5175983436853002, | |
| "grad_norm": 1.7480566033309979, | |
| "learning_rate": 1.1136322407782603e-05, | |
| "loss": 1.2288, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5201863354037267, | |
| "grad_norm": 1.5968227515636302, | |
| "learning_rate": 1.1046483084109801e-05, | |
| "loss": 1.2106, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.5227743271221532, | |
| "grad_norm": 1.6044766188232418, | |
| "learning_rate": 1.0956558279462188e-05, | |
| "loss": 1.2208, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.5253623188405797, | |
| "grad_norm": 1.6067811085133883, | |
| "learning_rate": 1.0866555339261854e-05, | |
| "loss": 1.2516, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.5279503105590062, | |
| "grad_norm": 1.7393965202791053, | |
| "learning_rate": 1.0776481615313314e-05, | |
| "loss": 1.2248, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.5305383022774327, | |
| "grad_norm": 1.5858453669945456, | |
| "learning_rate": 1.0686344465202988e-05, | |
| "loss": 1.2176, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.5331262939958592, | |
| "grad_norm": 1.5018326905549035, | |
| "learning_rate": 1.05961512516982e-05, | |
| "loss": 1.1795, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.5357142857142857, | |
| "grad_norm": 1.7095471413310692, | |
| "learning_rate": 1.050590934214576e-05, | |
| "loss": 1.2241, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.5383022774327122, | |
| "grad_norm": 1.566999738860566, | |
| "learning_rate": 1.0415626107870171e-05, | |
| "loss": 1.1825, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.5408902691511387, | |
| "grad_norm": 1.6457686223760508, | |
| "learning_rate": 1.0325308923571506e-05, | |
| "loss": 1.284, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.5434782608695652, | |
| "grad_norm": 1.6348224943577552, | |
| "learning_rate": 1.0234965166723003e-05, | |
| "loss": 1.2698, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.5460662525879917, | |
| "grad_norm": 1.5465862470654987, | |
| "learning_rate": 1.014460221696847e-05, | |
| "loss": 1.1963, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.5486542443064182, | |
| "grad_norm": 1.767248102032812, | |
| "learning_rate": 1.0054227455519459e-05, | |
| "loss": 1.1638, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.5512422360248447, | |
| "grad_norm": 2.0136126055316383, | |
| "learning_rate": 9.963848264552351e-06, | |
| "loss": 1.2307, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.5538302277432712, | |
| "grad_norm": 1.5145725671730654, | |
| "learning_rate": 9.873472026605347e-06, | |
| "loss": 1.2352, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.5564182194616977, | |
| "grad_norm": 1.5882467088516106, | |
| "learning_rate": 9.783106123975436e-06, | |
| "loss": 1.2108, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.5590062111801242, | |
| "grad_norm": 1.5181691898121104, | |
| "learning_rate": 9.692757938115376e-06, | |
| "loss": 1.1861, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.5615942028985508, | |
| "grad_norm": 1.6410339653531256, | |
| "learning_rate": 9.602434849030747e-06, | |
| "loss": 1.2465, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.5641821946169773, | |
| "grad_norm": 1.5661030436893173, | |
| "learning_rate": 9.512144234677116e-06, | |
| "loss": 1.243, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.5667701863354038, | |
| "grad_norm": 1.8049727192206169, | |
| "learning_rate": 9.421893470357397e-06, | |
| "loss": 1.2283, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.5693581780538303, | |
| "grad_norm": 1.8382407169522683, | |
| "learning_rate": 9.331689928119382e-06, | |
| "loss": 1.2307, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.5719461697722568, | |
| "grad_norm": 1.541503454678279, | |
| "learning_rate": 9.24154097615357e-06, | |
| "loss": 1.157, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.5745341614906833, | |
| "grad_norm": 1.5831300806041508, | |
| "learning_rate": 9.151453978191312e-06, | |
| "loss": 1.2318, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.5771221532091098, | |
| "grad_norm": 1.5551918280787969, | |
| "learning_rate": 9.061436292903303e-06, | |
| "loss": 1.1745, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.5797101449275363, | |
| "grad_norm": 1.4974161909157238, | |
| "learning_rate": 8.9714952732985e-06, | |
| "loss": 1.1957, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.5822981366459627, | |
| "grad_norm": 1.6338432021840628, | |
| "learning_rate": 8.88163826612349e-06, | |
| "loss": 1.2651, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.5848861283643892, | |
| "grad_norm": 1.605496912027716, | |
| "learning_rate": 8.791872611262393e-06, | |
| "loss": 1.183, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.5874741200828157, | |
| "grad_norm": 6.292512705322076, | |
| "learning_rate": 8.702205641137293e-06, | |
| "loss": 1.214, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.5900621118012422, | |
| "grad_norm": 1.5795886266757169, | |
| "learning_rate": 8.61264468010932e-06, | |
| "loss": 1.1703, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.5926501035196687, | |
| "grad_norm": 1.6798626881987566, | |
| "learning_rate": 8.523197043880334e-06, | |
| "loss": 1.1722, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.5952380952380952, | |
| "grad_norm": 1.4739551385813945, | |
| "learning_rate": 8.433870038895377e-06, | |
| "loss": 1.2068, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.5978260869565217, | |
| "grad_norm": 1.539016959394501, | |
| "learning_rate": 8.34467096174584e-06, | |
| "loss": 1.1899, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.6004140786749482, | |
| "grad_norm": 1.7032826938631798, | |
| "learning_rate": 8.255607098573454e-06, | |
| "loss": 1.2071, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.6030020703933747, | |
| "grad_norm": 1.6061844765987872, | |
| "learning_rate": 8.166685724475123e-06, | |
| "loss": 1.2302, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.6055900621118012, | |
| "grad_norm": 1.5665329666940486, | |
| "learning_rate": 8.077914102908661e-06, | |
| "loss": 1.2037, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.6081780538302277, | |
| "grad_norm": 1.6218099795936876, | |
| "learning_rate": 7.989299485099498e-06, | |
| "loss": 1.2411, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.6107660455486542, | |
| "grad_norm": 1.544888953496484, | |
| "learning_rate": 7.90084910944835e-06, | |
| "loss": 1.163, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.6133540372670807, | |
| "grad_norm": 1.5468464434079643, | |
| "learning_rate": 7.812570200939985e-06, | |
| "loss": 1.2069, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.6159420289855072, | |
| "grad_norm": 1.6191217485063485, | |
| "learning_rate": 7.724469970553028e-06, | |
| "loss": 1.212, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.6185300207039337, | |
| "grad_norm": 1.6584669157155563, | |
| "learning_rate": 7.636555614670953e-06, | |
| "loss": 1.2359, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.6211180124223602, | |
| "grad_norm": 1.6065749708841677, | |
| "learning_rate": 7.548834314494254e-06, | |
| "loss": 1.1731, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.6237060041407867, | |
| "grad_norm": 1.580536951203462, | |
| "learning_rate": 7.461313235453858e-06, | |
| "loss": 1.2095, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.6262939958592133, | |
| "grad_norm": 1.5890916003841715, | |
| "learning_rate": 7.37399952662581e-06, | |
| "loss": 1.1903, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.6288819875776398, | |
| "grad_norm": 1.6317367363623994, | |
| "learning_rate": 7.2869003201473235e-06, | |
| "loss": 1.1615, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.6314699792960663, | |
| "grad_norm": 1.5877174955747102, | |
| "learning_rate": 7.200022730634188e-06, | |
| "loss": 1.2261, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.6340579710144928, | |
| "grad_norm": 1.6863972908974845, | |
| "learning_rate": 7.113373854599621e-06, | |
| "loss": 1.1994, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.6366459627329193, | |
| "grad_norm": 1.6649535109927227, | |
| "learning_rate": 7.026960769874605e-06, | |
| "loss": 1.1562, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.6392339544513458, | |
| "grad_norm": 1.6139061664993306, | |
| "learning_rate": 6.940790535029726e-06, | |
| "loss": 1.2049, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.6418219461697723, | |
| "grad_norm": 2.0575985271226043, | |
| "learning_rate": 6.85487018879861e-06, | |
| "loss": 1.1997, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.6444099378881988, | |
| "grad_norm": 1.5693541403629991, | |
| "learning_rate": 6.76920674950297e-06, | |
| "loss": 1.1713, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.6469979296066253, | |
| "grad_norm": 1.4730612685033373, | |
| "learning_rate": 6.683807214479323e-06, | |
| "loss": 1.2526, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.6495859213250518, | |
| "grad_norm": 1.6427505615358542, | |
| "learning_rate": 6.598678559507414e-06, | |
| "loss": 1.2033, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.6521739130434783, | |
| "grad_norm": 1.4335417526682601, | |
| "learning_rate": 6.513827738240408e-06, | |
| "loss": 1.1668, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.6547619047619048, | |
| "grad_norm": 1.5130023195633595, | |
| "learning_rate": 6.429261681636889e-06, | |
| "loss": 1.2091, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.6573498964803313, | |
| "grad_norm": 1.5309037687636573, | |
| "learning_rate": 6.344987297394713e-06, | |
| "loss": 1.2039, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.6599378881987578, | |
| "grad_norm": 1.5308412500551256, | |
| "learning_rate": 6.261011469386755e-06, | |
| "loss": 1.1725, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.6625258799171843, | |
| "grad_norm": 4.125033452457996, | |
| "learning_rate": 6.177341057098607e-06, | |
| "loss": 1.2281, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.6651138716356108, | |
| "grad_norm": 1.6368311674369278, | |
| "learning_rate": 6.09398289506827e-06, | |
| "loss": 1.198, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.6677018633540373, | |
| "grad_norm": 1.543486534141238, | |
| "learning_rate": 6.010943792327875e-06, | |
| "loss": 1.1872, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.6702898550724637, | |
| "grad_norm": 1.6262983681397423, | |
| "learning_rate": 5.92823053184751e-06, | |
| "loss": 1.2178, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.6728778467908902, | |
| "grad_norm": 1.4789073947225242, | |
| "learning_rate": 5.845849869981137e-06, | |
| "loss": 1.1271, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.6754658385093167, | |
| "grad_norm": 1.7175017242675854, | |
| "learning_rate": 5.7638085359147235e-06, | |
| "loss": 1.1758, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.6780538302277432, | |
| "grad_norm": 1.621323772156704, | |
| "learning_rate": 5.682113231116557e-06, | |
| "loss": 1.1944, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.6806418219461697, | |
| "grad_norm": 1.4735479464826196, | |
| "learning_rate": 5.600770628789872e-06, | |
| "loss": 1.1943, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.6832298136645962, | |
| "grad_norm": 1.539494955374121, | |
| "learning_rate": 5.519787373327725e-06, | |
| "loss": 1.1556, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.6858178053830227, | |
| "grad_norm": 1.6129814926752055, | |
| "learning_rate": 5.439170079770269e-06, | |
| "loss": 1.2101, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.6884057971014492, | |
| "grad_norm": 1.6647287957418007, | |
| "learning_rate": 5.358925333264403e-06, | |
| "loss": 1.1943, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.6909937888198758, | |
| "grad_norm": 1.5220188441031306, | |
| "learning_rate": 5.279059688525878e-06, | |
| "loss": 1.1722, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.6935817805383023, | |
| "grad_norm": 1.6621504122467943, | |
| "learning_rate": 5.199579669303885e-06, | |
| "loss": 1.1982, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.6961697722567288, | |
| "grad_norm": 1.5286471811019702, | |
| "learning_rate": 5.1204917678481525e-06, | |
| "loss": 1.1371, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 0.6987577639751553, | |
| "grad_norm": 1.5940066135025674, | |
| "learning_rate": 5.0418024443786395e-06, | |
| "loss": 1.1547, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.7013457556935818, | |
| "grad_norm": 1.8246740368118712, | |
| "learning_rate": 4.963518126557846e-06, | |
| "loss": 1.1669, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 0.7039337474120083, | |
| "grad_norm": 1.583850157022352, | |
| "learning_rate": 4.885645208965779e-06, | |
| "loss": 1.1952, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.7065217391304348, | |
| "grad_norm": 1.826552327257342, | |
| "learning_rate": 4.8081900525775984e-06, | |
| "loss": 1.1775, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 0.7091097308488613, | |
| "grad_norm": 1.6908047939700899, | |
| "learning_rate": 4.731158984244042e-06, | |
| "loss": 1.2053, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.7116977225672878, | |
| "grad_norm": 1.577115477870209, | |
| "learning_rate": 4.654558296174617e-06, | |
| "loss": 1.1756, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.7142857142857143, | |
| "grad_norm": 1.583278280234861, | |
| "learning_rate": 4.578394245423626e-06, | |
| "loss": 1.1486, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.7168737060041408, | |
| "grad_norm": 1.5474720201156653, | |
| "learning_rate": 4.502673053379077e-06, | |
| "loss": 1.2029, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 0.7194616977225673, | |
| "grad_norm": 1.9218329295357086, | |
| "learning_rate": 4.427400905254483e-06, | |
| "loss": 1.1592, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.7220496894409938, | |
| "grad_norm": 1.5807402259350405, | |
| "learning_rate": 4.352583949583619e-06, | |
| "loss": 1.1443, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 0.7246376811594203, | |
| "grad_norm": 1.5984439783246036, | |
| "learning_rate": 4.278228297718307e-06, | |
| "loss": 1.1852, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.7272256728778468, | |
| "grad_norm": 1.562314289203796, | |
| "learning_rate": 4.204340023329215e-06, | |
| "loss": 1.1463, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 0.7298136645962733, | |
| "grad_norm": 1.6156395775351333, | |
| "learning_rate": 4.130925161909716e-06, | |
| "loss": 1.1512, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.7324016563146998, | |
| "grad_norm": 1.5829034816425451, | |
| "learning_rate": 4.057989710282897e-06, | |
| "loss": 1.1246, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 0.7349896480331263, | |
| "grad_norm": 1.523735802005734, | |
| "learning_rate": 3.985539626111708e-06, | |
| "loss": 1.1668, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.7375776397515528, | |
| "grad_norm": 1.505474487450996, | |
| "learning_rate": 3.913580827412334e-06, | |
| "loss": 1.1391, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.7401656314699793, | |
| "grad_norm": 1.6890981988605818, | |
| "learning_rate": 3.842119192070762e-06, | |
| "loss": 1.151, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.7427536231884058, | |
| "grad_norm": 1.6496446204834274, | |
| "learning_rate": 3.77116055736267e-06, | |
| "loss": 1.1906, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 0.7453416149068323, | |
| "grad_norm": 1.62048985165388, | |
| "learning_rate": 3.7007107194765945e-06, | |
| "loss": 1.1676, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.7479296066252588, | |
| "grad_norm": 1.6419977108085304, | |
| "learning_rate": 3.630775433040502e-06, | |
| "loss": 1.1397, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 0.7505175983436853, | |
| "grad_norm": 1.5143054947159678, | |
| "learning_rate": 3.561360410651713e-06, | |
| "loss": 1.1569, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.7531055900621118, | |
| "grad_norm": 1.5184069642552582, | |
| "learning_rate": 3.492471322410268e-06, | |
| "loss": 1.1661, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 0.7556935817805382, | |
| "grad_norm": 1.740302385789126, | |
| "learning_rate": 3.4241137954557792e-06, | |
| "loss": 1.2001, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.7582815734989649, | |
| "grad_norm": 1.735136651036564, | |
| "learning_rate": 3.3562934135077794e-06, | |
| "loss": 1.2071, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 0.7608695652173914, | |
| "grad_norm": 1.512017941498727, | |
| "learning_rate": 3.2890157164096315e-06, | |
| "loss": 1.1562, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.7634575569358178, | |
| "grad_norm": 1.56398858453209, | |
| "learning_rate": 3.2222861996759912e-06, | |
| "loss": 1.1755, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.7660455486542443, | |
| "grad_norm": 1.8429752138603448, | |
| "learning_rate": 3.156110314043933e-06, | |
| "loss": 1.1382, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.7686335403726708, | |
| "grad_norm": 1.531763749709719, | |
| "learning_rate": 3.0904934650276897e-06, | |
| "loss": 1.1716, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 0.7712215320910973, | |
| "grad_norm": 1.570674065613626, | |
| "learning_rate": 3.025441012477126e-06, | |
| "loss": 1.1911, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.7738095238095238, | |
| "grad_norm": 1.508664008364339, | |
| "learning_rate": 2.9609582701399266e-06, | |
| "loss": 1.1716, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 0.7763975155279503, | |
| "grad_norm": 1.4777031849165594, | |
| "learning_rate": 2.89705050522753e-06, | |
| "loss": 1.1456, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.7789855072463768, | |
| "grad_norm": 1.5683218260891874, | |
| "learning_rate": 2.833722937984893e-06, | |
| "loss": 1.1936, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 0.7815734989648033, | |
| "grad_norm": 1.558998170252945, | |
| "learning_rate": 2.7709807412640745e-06, | |
| "loss": 1.1532, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.7841614906832298, | |
| "grad_norm": 1.5325074641404255, | |
| "learning_rate": 2.70882904010171e-06, | |
| "loss": 1.1544, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 0.7867494824016563, | |
| "grad_norm": 1.587667874017625, | |
| "learning_rate": 2.6472729113003614e-06, | |
| "loss": 1.1363, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.7893374741200828, | |
| "grad_norm": 1.487145300264583, | |
| "learning_rate": 2.5863173830138212e-06, | |
| "loss": 1.1609, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.7919254658385093, | |
| "grad_norm": 1.5296852825320244, | |
| "learning_rate": 2.5259674343364104e-06, | |
| "loss": 1.1166, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.7945134575569358, | |
| "grad_norm": 1.523019173708786, | |
| "learning_rate": 2.4662279948962497e-06, | |
| "loss": 1.1443, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 0.7971014492753623, | |
| "grad_norm": 1.6276678280152375, | |
| "learning_rate": 2.4071039444526046e-06, | |
| "loss": 1.2047, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.7996894409937888, | |
| "grad_norm": 1.5518526603934675, | |
| "learning_rate": 2.3486001124972636e-06, | |
| "loss": 1.1357, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 0.8022774327122153, | |
| "grad_norm": 1.5395591809545814, | |
| "learning_rate": 2.290721277860064e-06, | |
| "loss": 1.1562, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.8048654244306418, | |
| "grad_norm": 1.5155042472429905, | |
| "learning_rate": 2.233472168318529e-06, | |
| "loss": 1.1257, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 0.8074534161490683, | |
| "grad_norm": 1.6546895809171993, | |
| "learning_rate": 2.176857460211693e-06, | |
| "loss": 1.1813, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.8100414078674948, | |
| "grad_norm": 1.8271870019910332, | |
| "learning_rate": 2.120881778058109e-06, | |
| "loss": 1.1749, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 0.8126293995859213, | |
| "grad_norm": 1.5721705885784398, | |
| "learning_rate": 2.0655496941780993e-06, | |
| "loss": 1.1332, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.8152173913043478, | |
| "grad_norm": 1.5777293602086202, | |
| "learning_rate": 2.0108657283202783e-06, | |
| "loss": 1.1468, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.8178053830227743, | |
| "grad_norm": 1.5754598191510423, | |
| "learning_rate": 1.9568343472923524e-06, | |
| "loss": 1.1793, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.8203933747412008, | |
| "grad_norm": 1.6259405743934179, | |
| "learning_rate": 1.903459964596267e-06, | |
| "loss": 1.1499, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 0.8229813664596274, | |
| "grad_norm": 1.6594307558296557, | |
| "learning_rate": 1.8507469400676736e-06, | |
| "loss": 1.2668, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.8255693581780539, | |
| "grad_norm": 1.5346716651192356, | |
| "learning_rate": 1.798699579519817e-06, | |
| "loss": 1.1057, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 0.8281573498964804, | |
| "grad_norm": 1.5898990173488856, | |
| "learning_rate": 1.7473221343918057e-06, | |
| "loss": 1.154, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.8307453416149069, | |
| "grad_norm": 1.6218963246045284, | |
| "learning_rate": 1.6966188014013574e-06, | |
| "loss": 1.203, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 1.5327256183704974, | |
| "learning_rate": 1.6465937222019745e-06, | |
| "loss": 1.0827, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.8359213250517599, | |
| "grad_norm": 1.5484614153825313, | |
| "learning_rate": 1.597250983044637e-06, | |
| "loss": 1.1512, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 0.8385093167701864, | |
| "grad_norm": 1.547251548669474, | |
| "learning_rate": 1.5485946144440357e-06, | |
| "loss": 1.1621, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.8410973084886129, | |
| "grad_norm": 1.5301077366867935, | |
| "learning_rate": 1.500628590849339e-06, | |
| "loss": 1.1, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.8436853002070394, | |
| "grad_norm": 1.5775881041342432, | |
| "learning_rate": 1.4533568303195333e-06, | |
| "loss": 1.1629, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.8462732919254659, | |
| "grad_norm": 1.6513355081710792, | |
| "learning_rate": 1.4067831942033904e-06, | |
| "loss": 1.1563, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 0.8488612836438924, | |
| "grad_norm": 1.6098056906806337, | |
| "learning_rate": 1.3609114868240537e-06, | |
| "loss": 1.1659, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.8514492753623188, | |
| "grad_norm": 1.5232813594236112, | |
| "learning_rate": 1.3157454551682824e-06, | |
| "loss": 1.1068, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 0.8540372670807453, | |
| "grad_norm": 1.596056108948684, | |
| "learning_rate": 1.2712887885803937e-06, | |
| "loss": 1.1769, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.8566252587991718, | |
| "grad_norm": 1.5685789393387255, | |
| "learning_rate": 1.227545118460889e-06, | |
| "loss": 1.1644, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 0.8592132505175983, | |
| "grad_norm": 1.6464311208376787, | |
| "learning_rate": 1.184518017969829e-06, | |
| "loss": 1.1533, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.8618012422360248, | |
| "grad_norm": 1.5274475944853032, | |
| "learning_rate": 1.1422110017349676e-06, | |
| "loss": 1.1709, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 0.8643892339544513, | |
| "grad_norm": 1.5509872198392205, | |
| "learning_rate": 1.100627525564668e-06, | |
| "loss": 1.1344, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.8669772256728778, | |
| "grad_norm": 1.504837308033305, | |
| "learning_rate": 1.0597709861656057e-06, | |
| "loss": 1.1358, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.8695652173913043, | |
| "grad_norm": 1.6991797160404174, | |
| "learning_rate": 1.0196447208653193e-06, | |
| "loss": 1.1857, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.8721532091097308, | |
| "grad_norm": 1.5275470265589042, | |
| "learning_rate": 9.802520073396016e-07, | |
| "loss": 1.1314, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 0.8747412008281573, | |
| "grad_norm": 1.6006710258152612, | |
| "learning_rate": 9.415960633447674e-07, | |
| "loss": 1.1662, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.8773291925465838, | |
| "grad_norm": 1.716257348826808, | |
| "learning_rate": 9.036800464548157e-07, | |
| "loss": 1.1509, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 0.8799171842650103, | |
| "grad_norm": 1.5117187651990462, | |
| "learning_rate": 8.665070538035037e-07, | |
| "loss": 1.1591, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.8825051759834368, | |
| "grad_norm": 1.5458076964833938, | |
| "learning_rate": 8.300801218313548e-07, | |
| "loss": 1.1849, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 0.8850931677018633, | |
| "grad_norm": 1.6346788638439136, | |
| "learning_rate": 7.944022260376416e-07, | |
| "loss": 1.1733, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.8876811594202898, | |
| "grad_norm": 1.4857294789989146, | |
| "learning_rate": 7.594762807373313e-07, | |
| "loss": 1.1091, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 0.8902691511387164, | |
| "grad_norm": 1.544124199759589, | |
| "learning_rate": 7.253051388230248e-07, | |
| "loss": 1.1374, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.8928571428571429, | |
| "grad_norm": 1.8377202213987396, | |
| "learning_rate": 6.918915915319302e-07, | |
| "loss": 1.1643, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.8954451345755694, | |
| "grad_norm": 1.5475759630185042, | |
| "learning_rate": 6.592383682178593e-07, | |
| "loss": 1.154, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.8980331262939959, | |
| "grad_norm": 1.521052252376937, | |
| "learning_rate": 6.273481361282807e-07, | |
| "loss": 1.1293, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 0.9006211180124224, | |
| "grad_norm": 1.5185508439469686, | |
| "learning_rate": 5.96223500186458e-07, | |
| "loss": 1.1744, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.9032091097308489, | |
| "grad_norm": 1.7082962342593018, | |
| "learning_rate": 5.658670027786561e-07, | |
| "loss": 1.1756, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 0.9057971014492754, | |
| "grad_norm": 1.5757735686564247, | |
| "learning_rate": 5.362811235464727e-07, | |
| "loss": 1.1849, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.9083850931677019, | |
| "grad_norm": 1.62779706084884, | |
| "learning_rate": 5.074682791842988e-07, | |
| "loss": 1.156, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 0.9109730848861284, | |
| "grad_norm": 1.5312273750679573, | |
| "learning_rate": 4.794308232419065e-07, | |
| "loss": 1.1497, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.9135610766045549, | |
| "grad_norm": 1.5488311929264333, | |
| "learning_rate": 4.521710459322015e-07, | |
| "loss": 1.1561, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 0.9161490683229814, | |
| "grad_norm": 1.6641638382041457, | |
| "learning_rate": 4.256911739441505e-07, | |
| "loss": 1.1556, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.9187370600414079, | |
| "grad_norm": 1.6032525408960623, | |
| "learning_rate": 3.999933702608949e-07, | |
| "loss": 1.1168, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.9213250517598344, | |
| "grad_norm": 1.5699960465562155, | |
| "learning_rate": 3.7507973398307584e-07, | |
| "loss": 1.1501, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.9239130434782609, | |
| "grad_norm": 1.5994478108257122, | |
| "learning_rate": 3.5095230015736113e-07, | |
| "loss": 1.1159, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 0.9265010351966874, | |
| "grad_norm": 1.5126252156346516, | |
| "learning_rate": 3.276130396102217e-07, | |
| "loss": 1.094, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.9290890269151139, | |
| "grad_norm": 1.5178216629033467, | |
| "learning_rate": 3.0506385878694255e-07, | |
| "loss": 1.1273, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 0.9316770186335404, | |
| "grad_norm": 1.5185613841579717, | |
| "learning_rate": 2.8330659959589944e-07, | |
| "loss": 1.1456, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.9342650103519669, | |
| "grad_norm": 1.6347573627683816, | |
| "learning_rate": 2.623430392581061e-07, | |
| "loss": 1.1372, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 0.9368530020703933, | |
| "grad_norm": 1.5838574431621013, | |
| "learning_rate": 2.421748901620369e-07, | |
| "loss": 1.1273, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.9394409937888198, | |
| "grad_norm": 1.6461360935258826, | |
| "learning_rate": 2.2280379972375643e-07, | |
| "loss": 1.1199, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 0.9420289855072463, | |
| "grad_norm": 1.5251560302510245, | |
| "learning_rate": 2.0423135025235298e-07, | |
| "loss": 1.1869, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.9446169772256728, | |
| "grad_norm": 1.74800052983838, | |
| "learning_rate": 1.8645905882068605e-07, | |
| "loss": 1.1277, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.9472049689440993, | |
| "grad_norm": 1.6647572293390658, | |
| "learning_rate": 1.6948837714146793e-07, | |
| "loss": 1.1026, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.9497929606625258, | |
| "grad_norm": 1.6498964813278796, | |
| "learning_rate": 1.533206914486818e-07, | |
| "loss": 1.1413, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 0.9523809523809523, | |
| "grad_norm": 1.6314577909166261, | |
| "learning_rate": 1.3795732238434557e-07, | |
| "loss": 1.1441, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.9549689440993789, | |
| "grad_norm": 1.5601815262952223, | |
| "learning_rate": 1.2339952489064056e-07, | |
| "loss": 1.2041, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 0.9575569358178054, | |
| "grad_norm": 1.5625160012256416, | |
| "learning_rate": 1.0964848810740114e-07, | |
| "loss": 1.1631, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.9601449275362319, | |
| "grad_norm": 1.7039383904038539, | |
| "learning_rate": 9.670533527498139e-08, | |
| "loss": 1.1729, | |
| "step": 1855 | |
| }, | |
| { | |
| "epoch": 0.9627329192546584, | |
| "grad_norm": 1.59975036577853, | |
| "learning_rate": 8.457112364250064e-08, | |
| "loss": 1.166, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.9653209109730849, | |
| "grad_norm": 1.601576491104471, | |
| "learning_rate": 7.324684438148933e-08, | |
| "loss": 1.152, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 0.9679089026915114, | |
| "grad_norm": 1.5522923849319805, | |
| "learning_rate": 6.273342250492254e-08, | |
| "loss": 1.1144, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.9704968944099379, | |
| "grad_norm": 1.5598616648811972, | |
| "learning_rate": 5.303171679166053e-08, | |
| "loss": 1.2196, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.9730848861283644, | |
| "grad_norm": 1.6532449106020801, | |
| "learning_rate": 4.414251971630035e-08, | |
| "loss": 1.1882, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.9756728778467909, | |
| "grad_norm": 1.5427460516528153, | |
| "learning_rate": 3.606655738444653e-08, | |
| "loss": 1.2124, | |
| "step": 1885 | |
| }, | |
| { | |
| "epoch": 0.9782608695652174, | |
| "grad_norm": 1.5671272147865478, | |
| "learning_rate": 2.8804489473395205e-08, | |
| "loss": 1.1492, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.9808488612836439, | |
| "grad_norm": 1.5788966872678778, | |
| "learning_rate": 2.235690917825384e-08, | |
| "loss": 1.1441, | |
| "step": 1895 | |
| }, | |
| { | |
| "epoch": 0.9834368530020704, | |
| "grad_norm": 1.6331717365418186, | |
| "learning_rate": 1.672434316348337e-08, | |
| "loss": 1.1179, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.9860248447204969, | |
| "grad_norm": 1.4925310986828344, | |
| "learning_rate": 1.1907251519877039e-08, | |
| "loss": 1.1612, | |
| "step": 1905 | |
| }, | |
| { | |
| "epoch": 0.9886128364389234, | |
| "grad_norm": 1.7342337869712523, | |
| "learning_rate": 7.906027726981568e-09, | |
| "loss": 1.1252, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.9912008281573499, | |
| "grad_norm": 1.6269325139131876, | |
| "learning_rate": 4.720998620955097e-09, | |
| "loss": 1.1918, | |
| "step": 1915 | |
| }, | |
| { | |
| "epoch": 0.9937888198757764, | |
| "grad_norm": 1.4814594112612938, | |
| "learning_rate": 2.3524243678685333e-09, | |
| "loss": 1.1315, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.9963768115942029, | |
| "grad_norm": 1.5876922772379496, | |
| "learning_rate": 8.004984424558881e-10, | |
| "loss": 1.1227, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.9989648033126294, | |
| "grad_norm": 1.694840692501587, | |
| "learning_rate": 6.534761230914122e-11, | |
| "loss": 1.1366, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 1.1816973686218262, | |
| "eval_runtime": 1548.7687, | |
| "eval_samples_per_second": 17.673, | |
| "eval_steps_per_second": 0.553, | |
| "step": 1932 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1932, | |
| "total_flos": 230071600152576.0, | |
| "train_loss": 1.2287210629593512, | |
| "train_runtime": 18714.7709, | |
| "train_samples_per_second": 3.303, | |
| "train_steps_per_second": 0.103 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 1932, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 230071600152576.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |