| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 29.812698412698413, |
| "eval_steps": 500, |
| "global_step": 4710, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.031746031746031744, |
| "grad_norm": 0.5545095205307007, |
| "learning_rate": 5.307855626326963e-07, |
| "loss": 3.7162, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.06349206349206349, |
| "grad_norm": 0.6163601279258728, |
| "learning_rate": 1.0615711252653927e-06, |
| "loss": 3.9388, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.09523809523809523, |
| "grad_norm": 0.5541530847549438, |
| "learning_rate": 1.5923566878980892e-06, |
| "loss": 3.9165, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.12698412698412698, |
| "grad_norm": 0.457332044839859, |
| "learning_rate": 2.1231422505307854e-06, |
| "loss": 3.7326, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.15873015873015872, |
| "grad_norm": 0.5335279107093811, |
| "learning_rate": 2.653927813163482e-06, |
| "loss": 3.8251, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.19047619047619047, |
| "grad_norm": 0.7080379724502563, |
| "learning_rate": 3.1847133757961785e-06, |
| "loss": 3.7534, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.2222222222222222, |
| "grad_norm": 0.520993709564209, |
| "learning_rate": 3.715498938428875e-06, |
| "loss": 3.898, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.25396825396825395, |
| "grad_norm": 0.5451405644416809, |
| "learning_rate": 4.246284501061571e-06, |
| "loss": 3.8951, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.2857142857142857, |
| "grad_norm": 0.6205154657363892, |
| "learning_rate": 4.777070063694268e-06, |
| "loss": 3.7666, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.31746031746031744, |
| "grad_norm": 0.7404439449310303, |
| "learning_rate": 5.307855626326964e-06, |
| "loss": 4.0258, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.3492063492063492, |
| "grad_norm": 0.6272220015525818, |
| "learning_rate": 5.838641188959661e-06, |
| "loss": 3.8464, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.38095238095238093, |
| "grad_norm": 0.7744691967964172, |
| "learning_rate": 6.369426751592357e-06, |
| "loss": 3.7299, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.4126984126984127, |
| "grad_norm": 0.8805738687515259, |
| "learning_rate": 6.900212314225053e-06, |
| "loss": 3.5008, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.4444444444444444, |
| "grad_norm": 1.0740723609924316, |
| "learning_rate": 7.43099787685775e-06, |
| "loss": 3.7552, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.47619047619047616, |
| "grad_norm": 0.965708315372467, |
| "learning_rate": 7.961783439490445e-06, |
| "loss": 3.5516, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.5079365079365079, |
| "grad_norm": 0.9812778234481812, |
| "learning_rate": 8.492569002123141e-06, |
| "loss": 3.6003, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.5396825396825397, |
| "grad_norm": 0.8831024169921875, |
| "learning_rate": 9.023354564755838e-06, |
| "loss": 3.613, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.5714285714285714, |
| "grad_norm": 0.8358364105224609, |
| "learning_rate": 9.554140127388536e-06, |
| "loss": 3.1858, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.6031746031746031, |
| "grad_norm": 1.0740444660186768, |
| "learning_rate": 1.0084925690021232e-05, |
| "loss": 3.0937, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.6349206349206349, |
| "grad_norm": 1.0987530946731567, |
| "learning_rate": 1.0615711252653929e-05, |
| "loss": 3.154, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.6666666666666666, |
| "grad_norm": 1.2300925254821777, |
| "learning_rate": 1.1146496815286625e-05, |
| "loss": 2.9414, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.6984126984126984, |
| "grad_norm": 1.2214170694351196, |
| "learning_rate": 1.1677282377919321e-05, |
| "loss": 2.9464, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.7301587301587301, |
| "grad_norm": 1.2803975343704224, |
| "learning_rate": 1.2208067940552018e-05, |
| "loss": 2.8921, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.7619047619047619, |
| "grad_norm": 1.2232719659805298, |
| "learning_rate": 1.2738853503184714e-05, |
| "loss": 2.5252, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.7936507936507936, |
| "grad_norm": 1.204835295677185, |
| "learning_rate": 1.326963906581741e-05, |
| "loss": 2.5215, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.8253968253968254, |
| "grad_norm": 1.4095579385757446, |
| "learning_rate": 1.3800424628450107e-05, |
| "loss": 2.136, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.8571428571428571, |
| "grad_norm": 1.4166598320007324, |
| "learning_rate": 1.4331210191082803e-05, |
| "loss": 2.2653, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.8888888888888888, |
| "grad_norm": 1.3040446043014526, |
| "learning_rate": 1.48619957537155e-05, |
| "loss": 2.0193, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.9206349206349206, |
| "grad_norm": 1.4114688634872437, |
| "learning_rate": 1.5392781316348196e-05, |
| "loss": 1.7935, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.9523809523809523, |
| "grad_norm": 1.8066726922988892, |
| "learning_rate": 1.592356687898089e-05, |
| "loss": 1.5731, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.9841269841269841, |
| "grad_norm": 1.4303158521652222, |
| "learning_rate": 1.6454352441613588e-05, |
| "loss": 1.6552, |
| "step": 155 |
| }, |
| { |
| "epoch": 1.0126984126984127, |
| "grad_norm": 1.6671762466430664, |
| "learning_rate": 1.6985138004246283e-05, |
| "loss": 1.6973, |
| "step": 160 |
| }, |
| { |
| "epoch": 1.0444444444444445, |
| "grad_norm": 1.5719650983810425, |
| "learning_rate": 1.751592356687898e-05, |
| "loss": 1.312, |
| "step": 165 |
| }, |
| { |
| "epoch": 1.0761904761904761, |
| "grad_norm": 1.4845054149627686, |
| "learning_rate": 1.8046709129511676e-05, |
| "loss": 1.3601, |
| "step": 170 |
| }, |
| { |
| "epoch": 1.107936507936508, |
| "grad_norm": 1.1172235012054443, |
| "learning_rate": 1.8577494692144374e-05, |
| "loss": 1.3137, |
| "step": 175 |
| }, |
| { |
| "epoch": 1.1396825396825396, |
| "grad_norm": 1.9621731042861938, |
| "learning_rate": 1.910828025477707e-05, |
| "loss": 1.1778, |
| "step": 180 |
| }, |
| { |
| "epoch": 1.1714285714285715, |
| "grad_norm": 1.7722721099853516, |
| "learning_rate": 1.963906581740977e-05, |
| "loss": 1.4534, |
| "step": 185 |
| }, |
| { |
| "epoch": 1.2031746031746031, |
| "grad_norm": 1.3677467107772827, |
| "learning_rate": 2.0169851380042464e-05, |
| "loss": 1.3356, |
| "step": 190 |
| }, |
| { |
| "epoch": 1.234920634920635, |
| "grad_norm": 1.3260482549667358, |
| "learning_rate": 2.0700636942675162e-05, |
| "loss": 1.0876, |
| "step": 195 |
| }, |
| { |
| "epoch": 1.2666666666666666, |
| "grad_norm": 1.5176818370819092, |
| "learning_rate": 2.1231422505307857e-05, |
| "loss": 1.1602, |
| "step": 200 |
| }, |
| { |
| "epoch": 1.2984126984126985, |
| "grad_norm": 1.2793077230453491, |
| "learning_rate": 2.1762208067940555e-05, |
| "loss": 1.1505, |
| "step": 205 |
| }, |
| { |
| "epoch": 1.33015873015873, |
| "grad_norm": 1.196784257888794, |
| "learning_rate": 2.229299363057325e-05, |
| "loss": 1.0664, |
| "step": 210 |
| }, |
| { |
| "epoch": 1.361904761904762, |
| "grad_norm": 1.303207516670227, |
| "learning_rate": 2.2823779193205948e-05, |
| "loss": 1.2557, |
| "step": 215 |
| }, |
| { |
| "epoch": 1.3936507936507936, |
| "grad_norm": 1.2853388786315918, |
| "learning_rate": 2.3354564755838642e-05, |
| "loss": 1.0704, |
| "step": 220 |
| }, |
| { |
| "epoch": 1.4253968253968254, |
| "grad_norm": 1.381369948387146, |
| "learning_rate": 2.388535031847134e-05, |
| "loss": 1.1371, |
| "step": 225 |
| }, |
| { |
| "epoch": 1.457142857142857, |
| "grad_norm": 1.8012712001800537, |
| "learning_rate": 2.4416135881104035e-05, |
| "loss": 1.248, |
| "step": 230 |
| }, |
| { |
| "epoch": 1.488888888888889, |
| "grad_norm": 1.7397032976150513, |
| "learning_rate": 2.4946921443736733e-05, |
| "loss": 1.2782, |
| "step": 235 |
| }, |
| { |
| "epoch": 1.5206349206349206, |
| "grad_norm": 1.4026210308074951, |
| "learning_rate": 2.5477707006369428e-05, |
| "loss": 1.154, |
| "step": 240 |
| }, |
| { |
| "epoch": 1.5523809523809524, |
| "grad_norm": 1.2906067371368408, |
| "learning_rate": 2.6008492569002126e-05, |
| "loss": 0.9141, |
| "step": 245 |
| }, |
| { |
| "epoch": 1.5841269841269843, |
| "grad_norm": 1.265598177909851, |
| "learning_rate": 2.653927813163482e-05, |
| "loss": 1.0625, |
| "step": 250 |
| }, |
| { |
| "epoch": 1.615873015873016, |
| "grad_norm": 1.6044715642929077, |
| "learning_rate": 2.707006369426752e-05, |
| "loss": 0.9624, |
| "step": 255 |
| }, |
| { |
| "epoch": 1.6476190476190475, |
| "grad_norm": 1.4612747430801392, |
| "learning_rate": 2.7600849256900213e-05, |
| "loss": 1.0413, |
| "step": 260 |
| }, |
| { |
| "epoch": 1.6793650793650794, |
| "grad_norm": 1.6222745180130005, |
| "learning_rate": 2.8131634819532908e-05, |
| "loss": 1.0929, |
| "step": 265 |
| }, |
| { |
| "epoch": 1.7111111111111112, |
| "grad_norm": 1.1456222534179688, |
| "learning_rate": 2.8662420382165606e-05, |
| "loss": 0.9957, |
| "step": 270 |
| }, |
| { |
| "epoch": 1.7428571428571429, |
| "grad_norm": 1.5746041536331177, |
| "learning_rate": 2.91932059447983e-05, |
| "loss": 1.0274, |
| "step": 275 |
| }, |
| { |
| "epoch": 1.7746031746031745, |
| "grad_norm": 1.3407832384109497, |
| "learning_rate": 2.9723991507431e-05, |
| "loss": 0.9487, |
| "step": 280 |
| }, |
| { |
| "epoch": 1.8063492063492064, |
| "grad_norm": 1.6232194900512695, |
| "learning_rate": 3.0254777070063693e-05, |
| "loss": 1.0966, |
| "step": 285 |
| }, |
| { |
| "epoch": 1.8380952380952382, |
| "grad_norm": 1.4920552968978882, |
| "learning_rate": 3.078556263269639e-05, |
| "loss": 0.9099, |
| "step": 290 |
| }, |
| { |
| "epoch": 1.8698412698412699, |
| "grad_norm": 1.2123301029205322, |
| "learning_rate": 3.1316348195329086e-05, |
| "loss": 1.0902, |
| "step": 295 |
| }, |
| { |
| "epoch": 1.9015873015873015, |
| "grad_norm": 1.2080968618392944, |
| "learning_rate": 3.184713375796178e-05, |
| "loss": 0.943, |
| "step": 300 |
| }, |
| { |
| "epoch": 1.9333333333333333, |
| "grad_norm": 1.190319299697876, |
| "learning_rate": 3.237791932059448e-05, |
| "loss": 0.7893, |
| "step": 305 |
| }, |
| { |
| "epoch": 1.9650793650793652, |
| "grad_norm": 1.5929204225540161, |
| "learning_rate": 3.2908704883227177e-05, |
| "loss": 1.0232, |
| "step": 310 |
| }, |
| { |
| "epoch": 1.9968253968253968, |
| "grad_norm": 1.0138347148895264, |
| "learning_rate": 3.343949044585987e-05, |
| "loss": 0.6693, |
| "step": 315 |
| }, |
| { |
| "epoch": 2.0253968253968253, |
| "grad_norm": 1.3012847900390625, |
| "learning_rate": 3.3970276008492566e-05, |
| "loss": 0.8355, |
| "step": 320 |
| }, |
| { |
| "epoch": 2.057142857142857, |
| "grad_norm": 1.2264782190322876, |
| "learning_rate": 3.450106157112527e-05, |
| "loss": 0.9872, |
| "step": 325 |
| }, |
| { |
| "epoch": 2.088888888888889, |
| "grad_norm": 1.139275312423706, |
| "learning_rate": 3.503184713375796e-05, |
| "loss": 0.8662, |
| "step": 330 |
| }, |
| { |
| "epoch": 2.1206349206349207, |
| "grad_norm": 1.3836581707000732, |
| "learning_rate": 3.5562632696390657e-05, |
| "loss": 0.9549, |
| "step": 335 |
| }, |
| { |
| "epoch": 2.1523809523809523, |
| "grad_norm": 1.368600845336914, |
| "learning_rate": 3.609341825902335e-05, |
| "loss": 0.9195, |
| "step": 340 |
| }, |
| { |
| "epoch": 2.1841269841269844, |
| "grad_norm": 1.8793011903762817, |
| "learning_rate": 3.662420382165605e-05, |
| "loss": 0.8505, |
| "step": 345 |
| }, |
| { |
| "epoch": 2.215873015873016, |
| "grad_norm": 1.305284023284912, |
| "learning_rate": 3.715498938428875e-05, |
| "loss": 0.7755, |
| "step": 350 |
| }, |
| { |
| "epoch": 2.2476190476190476, |
| "grad_norm": 1.7851749658584595, |
| "learning_rate": 3.768577494692145e-05, |
| "loss": 0.9242, |
| "step": 355 |
| }, |
| { |
| "epoch": 2.2793650793650793, |
| "grad_norm": 1.4341535568237305, |
| "learning_rate": 3.821656050955414e-05, |
| "loss": 0.8221, |
| "step": 360 |
| }, |
| { |
| "epoch": 2.311111111111111, |
| "grad_norm": 1.39107346534729, |
| "learning_rate": 3.874734607218684e-05, |
| "loss": 0.6999, |
| "step": 365 |
| }, |
| { |
| "epoch": 2.342857142857143, |
| "grad_norm": 1.2304264307022095, |
| "learning_rate": 3.927813163481954e-05, |
| "loss": 0.8362, |
| "step": 370 |
| }, |
| { |
| "epoch": 2.3746031746031746, |
| "grad_norm": 1.8470840454101562, |
| "learning_rate": 3.9808917197452234e-05, |
| "loss": 0.9398, |
| "step": 375 |
| }, |
| { |
| "epoch": 2.4063492063492062, |
| "grad_norm": 1.2533882856369019, |
| "learning_rate": 4.033970276008493e-05, |
| "loss": 0.7754, |
| "step": 380 |
| }, |
| { |
| "epoch": 2.4380952380952383, |
| "grad_norm": 1.5335006713867188, |
| "learning_rate": 4.087048832271762e-05, |
| "loss": 1.1124, |
| "step": 385 |
| }, |
| { |
| "epoch": 2.46984126984127, |
| "grad_norm": 1.5298357009887695, |
| "learning_rate": 4.1401273885350325e-05, |
| "loss": 1.017, |
| "step": 390 |
| }, |
| { |
| "epoch": 2.5015873015873016, |
| "grad_norm": 1.4403260946273804, |
| "learning_rate": 4.193205944798302e-05, |
| "loss": 0.8831, |
| "step": 395 |
| }, |
| { |
| "epoch": 2.533333333333333, |
| "grad_norm": 1.1528433561325073, |
| "learning_rate": 4.2462845010615714e-05, |
| "loss": 0.801, |
| "step": 400 |
| }, |
| { |
| "epoch": 2.565079365079365, |
| "grad_norm": 1.3371326923370361, |
| "learning_rate": 4.299363057324841e-05, |
| "loss": 0.8692, |
| "step": 405 |
| }, |
| { |
| "epoch": 2.596825396825397, |
| "grad_norm": 1.4064775705337524, |
| "learning_rate": 4.352441613588111e-05, |
| "loss": 0.9059, |
| "step": 410 |
| }, |
| { |
| "epoch": 2.6285714285714286, |
| "grad_norm": 1.4531422853469849, |
| "learning_rate": 4.4055201698513805e-05, |
| "loss": 0.7344, |
| "step": 415 |
| }, |
| { |
| "epoch": 2.66031746031746, |
| "grad_norm": 1.7043890953063965, |
| "learning_rate": 4.45859872611465e-05, |
| "loss": 0.8298, |
| "step": 420 |
| }, |
| { |
| "epoch": 2.6920634920634923, |
| "grad_norm": 1.5105586051940918, |
| "learning_rate": 4.5116772823779194e-05, |
| "loss": 0.7768, |
| "step": 425 |
| }, |
| { |
| "epoch": 2.723809523809524, |
| "grad_norm": 1.8101528882980347, |
| "learning_rate": 4.5647558386411895e-05, |
| "loss": 0.733, |
| "step": 430 |
| }, |
| { |
| "epoch": 2.7555555555555555, |
| "grad_norm": 1.6365174055099487, |
| "learning_rate": 4.617834394904459e-05, |
| "loss": 0.8061, |
| "step": 435 |
| }, |
| { |
| "epoch": 2.787301587301587, |
| "grad_norm": 1.7808202505111694, |
| "learning_rate": 4.6709129511677285e-05, |
| "loss": 0.8333, |
| "step": 440 |
| }, |
| { |
| "epoch": 2.819047619047619, |
| "grad_norm": 1.5223265886306763, |
| "learning_rate": 4.723991507430998e-05, |
| "loss": 0.7557, |
| "step": 445 |
| }, |
| { |
| "epoch": 2.850793650793651, |
| "grad_norm": 1.3064416646957397, |
| "learning_rate": 4.777070063694268e-05, |
| "loss": 0.8041, |
| "step": 450 |
| }, |
| { |
| "epoch": 2.8825396825396825, |
| "grad_norm": 1.8025637865066528, |
| "learning_rate": 4.8301486199575375e-05, |
| "loss": 0.9534, |
| "step": 455 |
| }, |
| { |
| "epoch": 2.914285714285714, |
| "grad_norm": 1.924846887588501, |
| "learning_rate": 4.883227176220807e-05, |
| "loss": 0.9066, |
| "step": 460 |
| }, |
| { |
| "epoch": 2.9460317460317462, |
| "grad_norm": 1.9862899780273438, |
| "learning_rate": 4.9363057324840765e-05, |
| "loss": 0.7994, |
| "step": 465 |
| }, |
| { |
| "epoch": 2.977777777777778, |
| "grad_norm": 1.9615916013717651, |
| "learning_rate": 4.9893842887473466e-05, |
| "loss": 0.7045, |
| "step": 470 |
| }, |
| { |
| "epoch": 3.0063492063492063, |
| "grad_norm": 1.519852876663208, |
| "learning_rate": 4.999989014936042e-05, |
| "loss": 0.7212, |
| "step": 475 |
| }, |
| { |
| "epoch": 3.038095238095238, |
| "grad_norm": 1.9328887462615967, |
| "learning_rate": 4.999944388279162e-05, |
| "loss": 0.6598, |
| "step": 480 |
| }, |
| { |
| "epoch": 3.06984126984127, |
| "grad_norm": 2.0340709686279297, |
| "learning_rate": 4.999865434075176e-05, |
| "loss": 0.6829, |
| "step": 485 |
| }, |
| { |
| "epoch": 3.1015873015873017, |
| "grad_norm": 1.8775280714035034, |
| "learning_rate": 4.999752153408229e-05, |
| "loss": 0.6664, |
| "step": 490 |
| }, |
| { |
| "epoch": 3.1333333333333333, |
| "grad_norm": 2.385218381881714, |
| "learning_rate": 4.999604547833814e-05, |
| "loss": 0.6836, |
| "step": 495 |
| }, |
| { |
| "epoch": 3.165079365079365, |
| "grad_norm": 2.1743783950805664, |
| "learning_rate": 4.999422619378752e-05, |
| "loss": 0.7, |
| "step": 500 |
| }, |
| { |
| "epoch": 3.196825396825397, |
| "grad_norm": 2.20786452293396, |
| "learning_rate": 4.999206370541162e-05, |
| "loss": 0.7253, |
| "step": 505 |
| }, |
| { |
| "epoch": 3.2285714285714286, |
| "grad_norm": 1.8182263374328613, |
| "learning_rate": 4.998955804290425e-05, |
| "loss": 0.6824, |
| "step": 510 |
| }, |
| { |
| "epoch": 3.2603174603174603, |
| "grad_norm": 2.2959372997283936, |
| "learning_rate": 4.9986709240671495e-05, |
| "loss": 0.601, |
| "step": 515 |
| }, |
| { |
| "epoch": 3.292063492063492, |
| "grad_norm": 2.385838031768799, |
| "learning_rate": 4.998351733783116e-05, |
| "loss": 0.7417, |
| "step": 520 |
| }, |
| { |
| "epoch": 3.323809523809524, |
| "grad_norm": 2.0416879653930664, |
| "learning_rate": 4.997998237821233e-05, |
| "loss": 0.6463, |
| "step": 525 |
| }, |
| { |
| "epoch": 3.3555555555555556, |
| "grad_norm": 2.2781031131744385, |
| "learning_rate": 4.9976104410354654e-05, |
| "loss": 0.6998, |
| "step": 530 |
| }, |
| { |
| "epoch": 3.3873015873015873, |
| "grad_norm": 2.146778106689453, |
| "learning_rate": 4.9971883487507775e-05, |
| "loss": 0.7694, |
| "step": 535 |
| }, |
| { |
| "epoch": 3.419047619047619, |
| "grad_norm": 2.1369104385375977, |
| "learning_rate": 4.9967319667630567e-05, |
| "loss": 0.6615, |
| "step": 540 |
| }, |
| { |
| "epoch": 3.450793650793651, |
| "grad_norm": 2.4529733657836914, |
| "learning_rate": 4.996241301339029e-05, |
| "loss": 0.6109, |
| "step": 545 |
| }, |
| { |
| "epoch": 3.4825396825396826, |
| "grad_norm": 2.07030987739563, |
| "learning_rate": 4.995716359216183e-05, |
| "loss": 0.7611, |
| "step": 550 |
| }, |
| { |
| "epoch": 3.5142857142857142, |
| "grad_norm": 2.4329919815063477, |
| "learning_rate": 4.995157147602669e-05, |
| "loss": 0.7515, |
| "step": 555 |
| }, |
| { |
| "epoch": 3.546031746031746, |
| "grad_norm": 2.056351900100708, |
| "learning_rate": 4.994563674177202e-05, |
| "loss": 0.6885, |
| "step": 560 |
| }, |
| { |
| "epoch": 3.5777777777777775, |
| "grad_norm": 2.3665318489074707, |
| "learning_rate": 4.993935947088958e-05, |
| "loss": 0.6271, |
| "step": 565 |
| }, |
| { |
| "epoch": 3.6095238095238096, |
| "grad_norm": 2.677706480026245, |
| "learning_rate": 4.993273974957463e-05, |
| "loss": 0.5586, |
| "step": 570 |
| }, |
| { |
| "epoch": 3.641269841269841, |
| "grad_norm": 3.422136068344116, |
| "learning_rate": 4.9925777668724685e-05, |
| "loss": 0.7552, |
| "step": 575 |
| }, |
| { |
| "epoch": 3.6730158730158733, |
| "grad_norm": 2.4525184631347656, |
| "learning_rate": 4.991847332393835e-05, |
| "loss": 0.7367, |
| "step": 580 |
| }, |
| { |
| "epoch": 3.704761904761905, |
| "grad_norm": 2.4242067337036133, |
| "learning_rate": 4.991082681551396e-05, |
| "loss": 0.7044, |
| "step": 585 |
| }, |
| { |
| "epoch": 3.7365079365079366, |
| "grad_norm": 1.8419867753982544, |
| "learning_rate": 4.9902838248448184e-05, |
| "loss": 0.5966, |
| "step": 590 |
| }, |
| { |
| "epoch": 3.768253968253968, |
| "grad_norm": 2.1394360065460205, |
| "learning_rate": 4.989450773243463e-05, |
| "loss": 0.6736, |
| "step": 595 |
| }, |
| { |
| "epoch": 3.8, |
| "grad_norm": 1.285447597503662, |
| "learning_rate": 4.9885835381862326e-05, |
| "loss": 0.5021, |
| "step": 600 |
| }, |
| { |
| "epoch": 3.831746031746032, |
| "grad_norm": 2.724978446960449, |
| "learning_rate": 4.987682131581413e-05, |
| "loss": 0.6128, |
| "step": 605 |
| }, |
| { |
| "epoch": 3.8634920634920635, |
| "grad_norm": 2.239682912826538, |
| "learning_rate": 4.986746565806508e-05, |
| "loss": 0.5457, |
| "step": 610 |
| }, |
| { |
| "epoch": 3.895238095238095, |
| "grad_norm": 2.48944091796875, |
| "learning_rate": 4.9857768537080784e-05, |
| "loss": 0.6927, |
| "step": 615 |
| }, |
| { |
| "epoch": 3.9269841269841272, |
| "grad_norm": 2.4086852073669434, |
| "learning_rate": 4.9847730086015534e-05, |
| "loss": 0.5963, |
| "step": 620 |
| }, |
| { |
| "epoch": 3.958730158730159, |
| "grad_norm": 2.0070106983184814, |
| "learning_rate": 4.9837350442710553e-05, |
| "loss": 0.5856, |
| "step": 625 |
| }, |
| { |
| "epoch": 3.9904761904761905, |
| "grad_norm": 1.9726545810699463, |
| "learning_rate": 4.98266297496921e-05, |
| "loss": 0.6208, |
| "step": 630 |
| }, |
| { |
| "epoch": 4.019047619047619, |
| "grad_norm": 2.6137828826904297, |
| "learning_rate": 4.981556815416948e-05, |
| "loss": 0.6319, |
| "step": 635 |
| }, |
| { |
| "epoch": 4.050793650793651, |
| "grad_norm": 2.3489890098571777, |
| "learning_rate": 4.9804165808033054e-05, |
| "loss": 0.5887, |
| "step": 640 |
| }, |
| { |
| "epoch": 4.082539682539682, |
| "grad_norm": 2.8010590076446533, |
| "learning_rate": 4.979242286785214e-05, |
| "loss": 0.5257, |
| "step": 645 |
| }, |
| { |
| "epoch": 4.114285714285714, |
| "grad_norm": 2.993411064147949, |
| "learning_rate": 4.978033949487284e-05, |
| "loss": 0.4545, |
| "step": 650 |
| }, |
| { |
| "epoch": 4.146031746031746, |
| "grad_norm": 2.669935703277588, |
| "learning_rate": 4.976791585501588e-05, |
| "loss": 0.5989, |
| "step": 655 |
| }, |
| { |
| "epoch": 4.177777777777778, |
| "grad_norm": 3.084409236907959, |
| "learning_rate": 4.9755152118874294e-05, |
| "loss": 0.528, |
| "step": 660 |
| }, |
| { |
| "epoch": 4.20952380952381, |
| "grad_norm": 2.797873020172119, |
| "learning_rate": 4.974204846171106e-05, |
| "loss": 0.5249, |
| "step": 665 |
| }, |
| { |
| "epoch": 4.241269841269841, |
| "grad_norm": 3.667867422103882, |
| "learning_rate": 4.9728605063456765e-05, |
| "loss": 0.5838, |
| "step": 670 |
| }, |
| { |
| "epoch": 4.273015873015873, |
| "grad_norm": 2.6918869018554688, |
| "learning_rate": 4.971482210870706e-05, |
| "loss": 0.5143, |
| "step": 675 |
| }, |
| { |
| "epoch": 4.304761904761905, |
| "grad_norm": 2.1545379161834717, |
| "learning_rate": 4.970069978672017e-05, |
| "loss": 0.5317, |
| "step": 680 |
| }, |
| { |
| "epoch": 4.336507936507936, |
| "grad_norm": 2.1043529510498047, |
| "learning_rate": 4.9686238291414275e-05, |
| "loss": 0.4815, |
| "step": 685 |
| }, |
| { |
| "epoch": 4.368253968253969, |
| "grad_norm": 2.1359753608703613, |
| "learning_rate": 4.9671437821364855e-05, |
| "loss": 0.4935, |
| "step": 690 |
| }, |
| { |
| "epoch": 4.4, |
| "grad_norm": 3.092057228088379, |
| "learning_rate": 4.965629857980197e-05, |
| "loss": 0.6831, |
| "step": 695 |
| }, |
| { |
| "epoch": 4.431746031746032, |
| "grad_norm": 2.5296835899353027, |
| "learning_rate": 4.964082077460745e-05, |
| "loss": 0.5323, |
| "step": 700 |
| }, |
| { |
| "epoch": 4.463492063492064, |
| "grad_norm": 1.6655627489089966, |
| "learning_rate": 4.962500461831207e-05, |
| "loss": 0.4553, |
| "step": 705 |
| }, |
| { |
| "epoch": 4.495238095238095, |
| "grad_norm": 2.6663475036621094, |
| "learning_rate": 4.9608850328092576e-05, |
| "loss": 0.463, |
| "step": 710 |
| }, |
| { |
| "epoch": 4.526984126984127, |
| "grad_norm": 2.3763060569763184, |
| "learning_rate": 4.959235812576879e-05, |
| "loss": 0.4861, |
| "step": 715 |
| }, |
| { |
| "epoch": 4.5587301587301585, |
| "grad_norm": 2.2217962741851807, |
| "learning_rate": 4.957552823780047e-05, |
| "loss": 0.468, |
| "step": 720 |
| }, |
| { |
| "epoch": 4.59047619047619, |
| "grad_norm": 2.8885600566864014, |
| "learning_rate": 4.9558360895284295e-05, |
| "loss": 0.4588, |
| "step": 725 |
| }, |
| { |
| "epoch": 4.622222222222222, |
| "grad_norm": 2.5661261081695557, |
| "learning_rate": 4.954085633395058e-05, |
| "loss": 0.4926, |
| "step": 730 |
| }, |
| { |
| "epoch": 4.653968253968254, |
| "grad_norm": 2.304365396499634, |
| "learning_rate": 4.952301479416015e-05, |
| "loss": 0.494, |
| "step": 735 |
| }, |
| { |
| "epoch": 4.685714285714286, |
| "grad_norm": 2.690577983856201, |
| "learning_rate": 4.9504836520900976e-05, |
| "loss": 0.5814, |
| "step": 740 |
| }, |
| { |
| "epoch": 4.717460317460318, |
| "grad_norm": 2.7180025577545166, |
| "learning_rate": 4.948632176378481e-05, |
| "loss": 0.5329, |
| "step": 745 |
| }, |
| { |
| "epoch": 4.749206349206349, |
| "grad_norm": 2.716587543487549, |
| "learning_rate": 4.9467470777043806e-05, |
| "loss": 0.5264, |
| "step": 750 |
| }, |
| { |
| "epoch": 4.780952380952381, |
| "grad_norm": 2.315419912338257, |
| "learning_rate": 4.9448283819526954e-05, |
| "loss": 0.4756, |
| "step": 755 |
| }, |
| { |
| "epoch": 4.8126984126984125, |
| "grad_norm": 2.1679515838623047, |
| "learning_rate": 4.9428761154696605e-05, |
| "loss": 0.4819, |
| "step": 760 |
| }, |
| { |
| "epoch": 4.844444444444444, |
| "grad_norm": 3.389266014099121, |
| "learning_rate": 4.9408903050624796e-05, |
| "loss": 0.5121, |
| "step": 765 |
| }, |
| { |
| "epoch": 4.876190476190477, |
| "grad_norm": 3.4317383766174316, |
| "learning_rate": 4.938870977998959e-05, |
| "loss": 0.4535, |
| "step": 770 |
| }, |
| { |
| "epoch": 4.907936507936508, |
| "grad_norm": 2.9491918087005615, |
| "learning_rate": 4.9368181620071344e-05, |
| "loss": 0.5333, |
| "step": 775 |
| }, |
| { |
| "epoch": 4.93968253968254, |
| "grad_norm": 2.516798496246338, |
| "learning_rate": 4.934731885274887e-05, |
| "loss": 0.5367, |
| "step": 780 |
| }, |
| { |
| "epoch": 4.9714285714285715, |
| "grad_norm": 3.0031046867370605, |
| "learning_rate": 4.9326121764495596e-05, |
| "loss": 0.4957, |
| "step": 785 |
| }, |
| { |
| "epoch": 5.0, |
| "grad_norm": 3.334085702896118, |
| "learning_rate": 4.9304590646375614e-05, |
| "loss": 0.5287, |
| "step": 790 |
| }, |
| { |
| "epoch": 5.031746031746032, |
| "grad_norm": 1.9608453512191772, |
| "learning_rate": 4.928272579403969e-05, |
| "loss": 0.36, |
| "step": 795 |
| }, |
| { |
| "epoch": 5.063492063492063, |
| "grad_norm": 2.328850746154785, |
| "learning_rate": 4.92605275077212e-05, |
| "loss": 0.3628, |
| "step": 800 |
| }, |
| { |
| "epoch": 5.095238095238095, |
| "grad_norm": 2.3446412086486816, |
| "learning_rate": 4.923799609223202e-05, |
| "loss": 0.3327, |
| "step": 805 |
| }, |
| { |
| "epoch": 5.1269841269841265, |
| "grad_norm": 2.476181745529175, |
| "learning_rate": 4.921513185695831e-05, |
| "loss": 0.4246, |
| "step": 810 |
| }, |
| { |
| "epoch": 5.158730158730159, |
| "grad_norm": 3.1026763916015625, |
| "learning_rate": 4.91919351158563e-05, |
| "loss": 0.5048, |
| "step": 815 |
| }, |
| { |
| "epoch": 5.190476190476191, |
| "grad_norm": 2.8165297508239746, |
| "learning_rate": 4.916840618744798e-05, |
| "loss": 0.4361, |
| "step": 820 |
| }, |
| { |
| "epoch": 5.222222222222222, |
| "grad_norm": 1.8732138872146606, |
| "learning_rate": 4.9144545394816687e-05, |
| "loss": 0.4693, |
| "step": 825 |
| }, |
| { |
| "epoch": 5.253968253968254, |
| "grad_norm": 1.7250264883041382, |
| "learning_rate": 4.91203530656027e-05, |
| "loss": 0.4076, |
| "step": 830 |
| }, |
| { |
| "epoch": 5.285714285714286, |
| "grad_norm": 2.105459690093994, |
| "learning_rate": 4.9095829531998725e-05, |
| "loss": 0.3589, |
| "step": 835 |
| }, |
| { |
| "epoch": 5.317460317460317, |
| "grad_norm": 3.6825687885284424, |
| "learning_rate": 4.9070975130745387e-05, |
| "loss": 0.5263, |
| "step": 840 |
| }, |
| { |
| "epoch": 5.349206349206349, |
| "grad_norm": 2.947052001953125, |
| "learning_rate": 4.90457902031265e-05, |
| "loss": 0.4632, |
| "step": 845 |
| }, |
| { |
| "epoch": 5.380952380952381, |
| "grad_norm": 1.9546104669570923, |
| "learning_rate": 4.902027509496448e-05, |
| "loss": 0.4348, |
| "step": 850 |
| }, |
| { |
| "epoch": 5.412698412698413, |
| "grad_norm": 2.4471983909606934, |
| "learning_rate": 4.899443015661557e-05, |
| "loss": 0.4209, |
| "step": 855 |
| }, |
| { |
| "epoch": 5.444444444444445, |
| "grad_norm": 1.827124834060669, |
| "learning_rate": 4.8968255742964975e-05, |
| "loss": 0.413, |
| "step": 860 |
| }, |
| { |
| "epoch": 5.476190476190476, |
| "grad_norm": 2.654707431793213, |
| "learning_rate": 4.894175221342207e-05, |
| "loss": 0.432, |
| "step": 865 |
| }, |
| { |
| "epoch": 5.507936507936508, |
| "grad_norm": 2.648967981338501, |
| "learning_rate": 4.8914919931915407e-05, |
| "loss": 0.4339, |
| "step": 870 |
| }, |
| { |
| "epoch": 5.5396825396825395, |
| "grad_norm": 2.874075412750244, |
| "learning_rate": 4.888775926688775e-05, |
| "loss": 0.4392, |
| "step": 875 |
| }, |
| { |
| "epoch": 5.571428571428571, |
| "grad_norm": 2.9674830436706543, |
| "learning_rate": 4.8860270591291e-05, |
| "loss": 0.4459, |
| "step": 880 |
| }, |
| { |
| "epoch": 5.603174603174603, |
| "grad_norm": 2.054748296737671, |
| "learning_rate": 4.883245428258107e-05, |
| "loss": 0.4313, |
| "step": 885 |
| }, |
| { |
| "epoch": 5.634920634920634, |
| "grad_norm": 1.9174392223358154, |
| "learning_rate": 4.880431072271272e-05, |
| "loss": 0.3906, |
| "step": 890 |
| }, |
| { |
| "epoch": 5.666666666666667, |
| "grad_norm": 2.5257787704467773, |
| "learning_rate": 4.87758402981343e-05, |
| "loss": 0.4219, |
| "step": 895 |
| }, |
| { |
| "epoch": 5.698412698412699, |
| "grad_norm": 2.6365532875061035, |
| "learning_rate": 4.8747043399782424e-05, |
| "loss": 0.3978, |
| "step": 900 |
| }, |
| { |
| "epoch": 5.73015873015873, |
| "grad_norm": 2.0583746433258057, |
| "learning_rate": 4.871792042307667e-05, |
| "loss": 0.4847, |
| "step": 905 |
| }, |
| { |
| "epoch": 5.761904761904762, |
| "grad_norm": 2.035872459411621, |
| "learning_rate": 4.868847176791406e-05, |
| "loss": 0.4675, |
| "step": 910 |
| }, |
| { |
| "epoch": 5.7936507936507935, |
| "grad_norm": 2.3722939491271973, |
| "learning_rate": 4.8658697838663625e-05, |
| "loss": 0.4586, |
| "step": 915 |
| }, |
| { |
| "epoch": 5.825396825396825, |
| "grad_norm": 1.2609732151031494, |
| "learning_rate": 4.862859904416085e-05, |
| "loss": 0.3274, |
| "step": 920 |
| }, |
| { |
| "epoch": 5.857142857142857, |
| "grad_norm": 2.3673977851867676, |
| "learning_rate": 4.8598175797702036e-05, |
| "loss": 0.4685, |
| "step": 925 |
| }, |
| { |
| "epoch": 5.888888888888889, |
| "grad_norm": 2.8414175510406494, |
| "learning_rate": 4.856742851703866e-05, |
| "loss": 0.4762, |
| "step": 930 |
| }, |
| { |
| "epoch": 5.920634920634921, |
| "grad_norm": 2.4126765727996826, |
| "learning_rate": 4.853635762437159e-05, |
| "loss": 0.4075, |
| "step": 935 |
| }, |
| { |
| "epoch": 5.9523809523809526, |
| "grad_norm": 1.8691045045852661, |
| "learning_rate": 4.8504963546345334e-05, |
| "loss": 0.4865, |
| "step": 940 |
| }, |
| { |
| "epoch": 5.984126984126984, |
| "grad_norm": 3.5297420024871826, |
| "learning_rate": 4.8473246714042155e-05, |
| "loss": 0.4623, |
| "step": 945 |
| }, |
| { |
| "epoch": 6.012698412698413, |
| "grad_norm": 2.059169054031372, |
| "learning_rate": 4.844120756297617e-05, |
| "loss": 0.4164, |
| "step": 950 |
| }, |
| { |
| "epoch": 6.044444444444444, |
| "grad_norm": 2.4746127128601074, |
| "learning_rate": 4.840884653308735e-05, |
| "loss": 0.3552, |
| "step": 955 |
| }, |
| { |
| "epoch": 6.076190476190476, |
| "grad_norm": 2.504425287246704, |
| "learning_rate": 4.8376164068735485e-05, |
| "loss": 0.3368, |
| "step": 960 |
| }, |
| { |
| "epoch": 6.1079365079365076, |
| "grad_norm": 2.062577486038208, |
| "learning_rate": 4.83431606186941e-05, |
| "loss": 0.3139, |
| "step": 965 |
| }, |
| { |
| "epoch": 6.13968253968254, |
| "grad_norm": 2.4934544563293457, |
| "learning_rate": 4.830983663614427e-05, |
| "loss": 0.3777, |
| "step": 970 |
| }, |
| { |
| "epoch": 6.171428571428572, |
| "grad_norm": 2.5747485160827637, |
| "learning_rate": 4.827619257866839e-05, |
| "loss": 0.373, |
| "step": 975 |
| }, |
| { |
| "epoch": 6.203174603174603, |
| "grad_norm": 2.449357271194458, |
| "learning_rate": 4.8242228908243946e-05, |
| "loss": 0.3936, |
| "step": 980 |
| }, |
| { |
| "epoch": 6.234920634920635, |
| "grad_norm": 2.952680826187134, |
| "learning_rate": 4.82079460912371e-05, |
| "loss": 0.407, |
| "step": 985 |
| }, |
| { |
| "epoch": 6.266666666666667, |
| "grad_norm": 2.1754496097564697, |
| "learning_rate": 4.817334459839633e-05, |
| "loss": 0.3189, |
| "step": 990 |
| }, |
| { |
| "epoch": 6.298412698412698, |
| "grad_norm": 2.8406214714050293, |
| "learning_rate": 4.8138424904845947e-05, |
| "loss": 0.3883, |
| "step": 995 |
| }, |
| { |
| "epoch": 6.33015873015873, |
| "grad_norm": 1.7533257007598877, |
| "learning_rate": 4.8103187490079604e-05, |
| "loss": 0.3131, |
| "step": 1000 |
| }, |
| { |
| "epoch": 6.3619047619047615, |
| "grad_norm": 2.4574601650238037, |
| "learning_rate": 4.806763283795366e-05, |
| "loss": 0.3606, |
| "step": 1005 |
| }, |
| { |
| "epoch": 6.393650793650794, |
| "grad_norm": 2.002281427383423, |
| "learning_rate": 4.8031761436680575e-05, |
| "loss": 0.37, |
| "step": 1010 |
| }, |
| { |
| "epoch": 6.425396825396826, |
| "grad_norm": 2.823315143585205, |
| "learning_rate": 4.79955737788222e-05, |
| "loss": 0.3791, |
| "step": 1015 |
| }, |
| { |
| "epoch": 6.457142857142857, |
| "grad_norm": 2.7891204357147217, |
| "learning_rate": 4.795907036128299e-05, |
| "loss": 0.3556, |
| "step": 1020 |
| }, |
| { |
| "epoch": 6.488888888888889, |
| "grad_norm": 2.2387146949768066, |
| "learning_rate": 4.7922251685303213e-05, |
| "loss": 0.3929, |
| "step": 1025 |
| }, |
| { |
| "epoch": 6.520634920634921, |
| "grad_norm": 2.5023891925811768, |
| "learning_rate": 4.788511825645205e-05, |
| "loss": 0.379, |
| "step": 1030 |
| }, |
| { |
| "epoch": 6.552380952380952, |
| "grad_norm": 2.2654805183410645, |
| "learning_rate": 4.7847670584620653e-05, |
| "loss": 0.3435, |
| "step": 1035 |
| }, |
| { |
| "epoch": 6.584126984126984, |
| "grad_norm": 3.3823065757751465, |
| "learning_rate": 4.7809909184015146e-05, |
| "loss": 0.4109, |
| "step": 1040 |
| }, |
| { |
| "epoch": 6.6158730158730155, |
| "grad_norm": 2.6096551418304443, |
| "learning_rate": 4.7771834573149576e-05, |
| "loss": 0.4233, |
| "step": 1045 |
| }, |
| { |
| "epoch": 6.647619047619048, |
| "grad_norm": 2.3933897018432617, |
| "learning_rate": 4.773344727483876e-05, |
| "loss": 0.3709, |
| "step": 1050 |
| }, |
| { |
| "epoch": 6.67936507936508, |
| "grad_norm": 2.189544916152954, |
| "learning_rate": 4.769474781619114e-05, |
| "loss": 0.3287, |
| "step": 1055 |
| }, |
| { |
| "epoch": 6.711111111111111, |
| "grad_norm": 2.450892686843872, |
| "learning_rate": 4.765573672860154e-05, |
| "loss": 0.4022, |
| "step": 1060 |
| }, |
| { |
| "epoch": 6.742857142857143, |
| "grad_norm": 2.4342429637908936, |
| "learning_rate": 4.761641454774386e-05, |
| "loss": 0.4029, |
| "step": 1065 |
| }, |
| { |
| "epoch": 6.7746031746031745, |
| "grad_norm": 2.2122364044189453, |
| "learning_rate": 4.75767818135637e-05, |
| "loss": 0.3322, |
| "step": 1070 |
| }, |
| { |
| "epoch": 6.806349206349206, |
| "grad_norm": 3.968445301055908, |
| "learning_rate": 4.7536839070271e-05, |
| "loss": 0.3836, |
| "step": 1075 |
| }, |
| { |
| "epoch": 6.838095238095238, |
| "grad_norm": 3.529158353805542, |
| "learning_rate": 4.749658686633251e-05, |
| "loss": 0.4745, |
| "step": 1080 |
| }, |
| { |
| "epoch": 6.86984126984127, |
| "grad_norm": 2.430727243423462, |
| "learning_rate": 4.7456025754464304e-05, |
| "loss": 0.3664, |
| "step": 1085 |
| }, |
| { |
| "epoch": 6.901587301587302, |
| "grad_norm": 2.6552302837371826, |
| "learning_rate": 4.7415156291624166e-05, |
| "loss": 0.4359, |
| "step": 1090 |
| }, |
| { |
| "epoch": 6.933333333333334, |
| "grad_norm": 2.134822130203247, |
| "learning_rate": 4.737397903900393e-05, |
| "loss": 0.3969, |
| "step": 1095 |
| }, |
| { |
| "epoch": 6.965079365079365, |
| "grad_norm": 2.5052947998046875, |
| "learning_rate": 4.7332494562021815e-05, |
| "loss": 0.4069, |
| "step": 1100 |
| }, |
| { |
| "epoch": 6.996825396825397, |
| "grad_norm": 2.1377065181732178, |
| "learning_rate": 4.729070343031463e-05, |
| "loss": 0.3853, |
| "step": 1105 |
| }, |
| { |
| "epoch": 7.025396825396825, |
| "grad_norm": 1.9704042673110962, |
| "learning_rate": 4.724860621772995e-05, |
| "loss": 0.3283, |
| "step": 1110 |
| }, |
| { |
| "epoch": 7.057142857142857, |
| "grad_norm": 2.476968765258789, |
| "learning_rate": 4.7206203502318256e-05, |
| "loss": 0.3325, |
| "step": 1115 |
| }, |
| { |
| "epoch": 7.088888888888889, |
| "grad_norm": 1.9231969118118286, |
| "learning_rate": 4.716349586632499e-05, |
| "loss": 0.2876, |
| "step": 1120 |
| }, |
| { |
| "epoch": 7.12063492063492, |
| "grad_norm": 2.6444814205169678, |
| "learning_rate": 4.712048389618254e-05, |
| "loss": 0.3005, |
| "step": 1125 |
| }, |
| { |
| "epoch": 7.152380952380953, |
| "grad_norm": 3.2589964866638184, |
| "learning_rate": 4.7077168182502216e-05, |
| "loss": 0.4023, |
| "step": 1130 |
| }, |
| { |
| "epoch": 7.184126984126984, |
| "grad_norm": 2.5481936931610107, |
| "learning_rate": 4.703354932006615e-05, |
| "loss": 0.3302, |
| "step": 1135 |
| }, |
| { |
| "epoch": 7.215873015873016, |
| "grad_norm": 1.7125908136367798, |
| "learning_rate": 4.698962790781906e-05, |
| "loss": 0.3329, |
| "step": 1140 |
| }, |
| { |
| "epoch": 7.247619047619048, |
| "grad_norm": 2.2756667137145996, |
| "learning_rate": 4.6945404548860115e-05, |
| "loss": 0.3369, |
| "step": 1145 |
| }, |
| { |
| "epoch": 7.279365079365079, |
| "grad_norm": 2.9158453941345215, |
| "learning_rate": 4.6900879850434604e-05, |
| "loss": 0.3339, |
| "step": 1150 |
| }, |
| { |
| "epoch": 7.311111111111111, |
| "grad_norm": 2.3047537803649902, |
| "learning_rate": 4.685605442392559e-05, |
| "loss": 0.3915, |
| "step": 1155 |
| }, |
| { |
| "epoch": 7.3428571428571425, |
| "grad_norm": 2.7815029621124268, |
| "learning_rate": 4.681092888484554e-05, |
| "loss": 0.3317, |
| "step": 1160 |
| }, |
| { |
| "epoch": 7.374603174603175, |
| "grad_norm": 2.2644097805023193, |
| "learning_rate": 4.676550385282787e-05, |
| "loss": 0.3314, |
| "step": 1165 |
| }, |
| { |
| "epoch": 7.406349206349207, |
| "grad_norm": 2.5144474506378174, |
| "learning_rate": 4.671977995161843e-05, |
| "loss": 0.3188, |
| "step": 1170 |
| }, |
| { |
| "epoch": 7.438095238095238, |
| "grad_norm": 3.120821714401245, |
| "learning_rate": 4.667375780906693e-05, |
| "loss": 0.3523, |
| "step": 1175 |
| }, |
| { |
| "epoch": 7.46984126984127, |
| "grad_norm": 4.47842264175415, |
| "learning_rate": 4.662743805711832e-05, |
| "loss": 0.3611, |
| "step": 1180 |
| }, |
| { |
| "epoch": 7.501587301587302, |
| "grad_norm": 1.9228928089141846, |
| "learning_rate": 4.658082133180416e-05, |
| "loss": 0.3612, |
| "step": 1185 |
| }, |
| { |
| "epoch": 7.533333333333333, |
| "grad_norm": 2.1507537364959717, |
| "learning_rate": 4.6533908273233815e-05, |
| "loss": 0.3321, |
| "step": 1190 |
| }, |
| { |
| "epoch": 7.565079365079365, |
| "grad_norm": 2.1849119663238525, |
| "learning_rate": 4.64866995255857e-05, |
| "loss": 0.2943, |
| "step": 1195 |
| }, |
| { |
| "epoch": 7.5968253968253965, |
| "grad_norm": 2.1777775287628174, |
| "learning_rate": 4.643919573709843e-05, |
| "loss": 0.353, |
| "step": 1200 |
| }, |
| { |
| "epoch": 7.628571428571428, |
| "grad_norm": 2.5231118202209473, |
| "learning_rate": 4.639139756006195e-05, |
| "loss": 0.3571, |
| "step": 1205 |
| }, |
| { |
| "epoch": 7.660317460317461, |
| "grad_norm": 1.8409479856491089, |
| "learning_rate": 4.6343305650808516e-05, |
| "loss": 0.3691, |
| "step": 1210 |
| }, |
| { |
| "epoch": 7.692063492063492, |
| "grad_norm": 1.7940895557403564, |
| "learning_rate": 4.629492066970373e-05, |
| "loss": 0.3738, |
| "step": 1215 |
| }, |
| { |
| "epoch": 7.723809523809524, |
| "grad_norm": 2.014902114868164, |
| "learning_rate": 4.6246243281137474e-05, |
| "loss": 0.361, |
| "step": 1220 |
| }, |
| { |
| "epoch": 7.7555555555555555, |
| "grad_norm": 3.4182560443878174, |
| "learning_rate": 4.6197274153514735e-05, |
| "loss": 0.3663, |
| "step": 1225 |
| }, |
| { |
| "epoch": 7.787301587301587, |
| "grad_norm": 2.518728256225586, |
| "learning_rate": 4.614801395924649e-05, |
| "loss": 0.3646, |
| "step": 1230 |
| }, |
| { |
| "epoch": 7.819047619047619, |
| "grad_norm": 2.154189109802246, |
| "learning_rate": 4.6098463374740466e-05, |
| "loss": 0.3331, |
| "step": 1235 |
| }, |
| { |
| "epoch": 7.85079365079365, |
| "grad_norm": 2.536081075668335, |
| "learning_rate": 4.604862308039177e-05, |
| "loss": 0.3742, |
| "step": 1240 |
| }, |
| { |
| "epoch": 7.882539682539683, |
| "grad_norm": 2.340764045715332, |
| "learning_rate": 4.599849376057366e-05, |
| "loss": 0.3352, |
| "step": 1245 |
| }, |
| { |
| "epoch": 7.914285714285715, |
| "grad_norm": 3.5488364696502686, |
| "learning_rate": 4.5948076103628094e-05, |
| "loss": 0.3663, |
| "step": 1250 |
| }, |
| { |
| "epoch": 7.946031746031746, |
| "grad_norm": 2.779360294342041, |
| "learning_rate": 4.589737080185625e-05, |
| "loss": 0.3362, |
| "step": 1255 |
| }, |
| { |
| "epoch": 7.977777777777778, |
| "grad_norm": 1.8792667388916016, |
| "learning_rate": 4.5846378551509097e-05, |
| "loss": 0.346, |
| "step": 1260 |
| }, |
| { |
| "epoch": 8.006349206349206, |
| "grad_norm": 2.453295946121216, |
| "learning_rate": 4.579510005277774e-05, |
| "loss": 0.3509, |
| "step": 1265 |
| }, |
| { |
| "epoch": 8.038095238095238, |
| "grad_norm": 1.9493130445480347, |
| "learning_rate": 4.574353600978388e-05, |
| "loss": 0.3062, |
| "step": 1270 |
| }, |
| { |
| "epoch": 8.06984126984127, |
| "grad_norm": 1.9360930919647217, |
| "learning_rate": 4.56916871305701e-05, |
| "loss": 0.3056, |
| "step": 1275 |
| }, |
| { |
| "epoch": 8.101587301587301, |
| "grad_norm": 1.5592070817947388, |
| "learning_rate": 4.563955412709021e-05, |
| "loss": 0.2785, |
| "step": 1280 |
| }, |
| { |
| "epoch": 8.133333333333333, |
| "grad_norm": 1.8093425035476685, |
| "learning_rate": 4.5587137715199354e-05, |
| "loss": 0.308, |
| "step": 1285 |
| }, |
| { |
| "epoch": 8.165079365079364, |
| "grad_norm": 2.2939181327819824, |
| "learning_rate": 4.5534438614644294e-05, |
| "loss": 0.3038, |
| "step": 1290 |
| }, |
| { |
| "epoch": 8.196825396825396, |
| "grad_norm": 2.4204866886138916, |
| "learning_rate": 4.548145754905346e-05, |
| "loss": 0.3375, |
| "step": 1295 |
| }, |
| { |
| "epoch": 8.228571428571428, |
| "grad_norm": 1.725534439086914, |
| "learning_rate": 4.5428195245927064e-05, |
| "loss": 0.3101, |
| "step": 1300 |
| }, |
| { |
| "epoch": 8.260317460317461, |
| "grad_norm": 1.637730360031128, |
| "learning_rate": 4.537465243662704e-05, |
| "loss": 0.2931, |
| "step": 1305 |
| }, |
| { |
| "epoch": 8.292063492063493, |
| "grad_norm": 1.3372169733047485, |
| "learning_rate": 4.532082985636709e-05, |
| "loss": 0.2763, |
| "step": 1310 |
| }, |
| { |
| "epoch": 8.323809523809524, |
| "grad_norm": 2.5993168354034424, |
| "learning_rate": 4.5266728244202494e-05, |
| "loss": 0.3458, |
| "step": 1315 |
| }, |
| { |
| "epoch": 8.355555555555556, |
| "grad_norm": 2.461862564086914, |
| "learning_rate": 4.521234834302006e-05, |
| "loss": 0.3693, |
| "step": 1320 |
| }, |
| { |
| "epoch": 8.387301587301588, |
| "grad_norm": 1.8519413471221924, |
| "learning_rate": 4.5157690899527816e-05, |
| "loss": 0.3327, |
| "step": 1325 |
| }, |
| { |
| "epoch": 8.41904761904762, |
| "grad_norm": 2.1535580158233643, |
| "learning_rate": 4.510275666424487e-05, |
| "loss": 0.3229, |
| "step": 1330 |
| }, |
| { |
| "epoch": 8.450793650793651, |
| "grad_norm": 1.6819690465927124, |
| "learning_rate": 4.5047546391491e-05, |
| "loss": 0.2925, |
| "step": 1335 |
| }, |
| { |
| "epoch": 8.482539682539683, |
| "grad_norm": 1.6538281440734863, |
| "learning_rate": 4.499206083937638e-05, |
| "loss": 0.3218, |
| "step": 1340 |
| }, |
| { |
| "epoch": 8.514285714285714, |
| "grad_norm": 1.8956862688064575, |
| "learning_rate": 4.493630076979112e-05, |
| "loss": 0.3423, |
| "step": 1345 |
| }, |
| { |
| "epoch": 8.546031746031746, |
| "grad_norm": 2.274681806564331, |
| "learning_rate": 4.48802669483948e-05, |
| "loss": 0.3152, |
| "step": 1350 |
| }, |
| { |
| "epoch": 8.577777777777778, |
| "grad_norm": 2.2956337928771973, |
| "learning_rate": 4.4823960144606014e-05, |
| "loss": 0.3417, |
| "step": 1355 |
| }, |
| { |
| "epoch": 8.60952380952381, |
| "grad_norm": 1.8650286197662354, |
| "learning_rate": 4.4767381131591734e-05, |
| "loss": 0.2896, |
| "step": 1360 |
| }, |
| { |
| "epoch": 8.64126984126984, |
| "grad_norm": 1.3998652696609497, |
| "learning_rate": 4.471053068625674e-05, |
| "loss": 0.3372, |
| "step": 1365 |
| }, |
| { |
| "epoch": 8.673015873015872, |
| "grad_norm": 2.855074167251587, |
| "learning_rate": 4.465340958923293e-05, |
| "loss": 0.332, |
| "step": 1370 |
| }, |
| { |
| "epoch": 8.704761904761904, |
| "grad_norm": 1.6865357160568237, |
| "learning_rate": 4.459601862486862e-05, |
| "loss": 0.3053, |
| "step": 1375 |
| }, |
| { |
| "epoch": 8.736507936507937, |
| "grad_norm": 2.501856803894043, |
| "learning_rate": 4.453835858121773e-05, |
| "loss": 0.3119, |
| "step": 1380 |
| }, |
| { |
| "epoch": 8.768253968253969, |
| "grad_norm": 2.4325456619262695, |
| "learning_rate": 4.4480430250029046e-05, |
| "loss": 0.3395, |
| "step": 1385 |
| }, |
| { |
| "epoch": 8.8, |
| "grad_norm": 1.4845948219299316, |
| "learning_rate": 4.4422234426735256e-05, |
| "loss": 0.3237, |
| "step": 1390 |
| }, |
| { |
| "epoch": 8.831746031746032, |
| "grad_norm": 1.3553249835968018, |
| "learning_rate": 4.436377191044208e-05, |
| "loss": 0.3387, |
| "step": 1395 |
| }, |
| { |
| "epoch": 8.863492063492064, |
| "grad_norm": 1.8338890075683594, |
| "learning_rate": 4.430504350391729e-05, |
| "loss": 0.3618, |
| "step": 1400 |
| }, |
| { |
| "epoch": 8.895238095238096, |
| "grad_norm": 2.291538953781128, |
| "learning_rate": 4.4246050013579686e-05, |
| "loss": 0.3608, |
| "step": 1405 |
| }, |
| { |
| "epoch": 8.926984126984127, |
| "grad_norm": 1.3809788227081299, |
| "learning_rate": 4.4186792249488005e-05, |
| "loss": 0.3077, |
| "step": 1410 |
| }, |
| { |
| "epoch": 8.958730158730159, |
| "grad_norm": 1.5944230556488037, |
| "learning_rate": 4.412727102532983e-05, |
| "loss": 0.3307, |
| "step": 1415 |
| }, |
| { |
| "epoch": 8.99047619047619, |
| "grad_norm": 2.2244362831115723, |
| "learning_rate": 4.4067487158410396e-05, |
| "loss": 0.3469, |
| "step": 1420 |
| }, |
| { |
| "epoch": 9.019047619047619, |
| "grad_norm": 1.444221019744873, |
| "learning_rate": 4.400744146964136e-05, |
| "loss": 0.3049, |
| "step": 1425 |
| }, |
| { |
| "epoch": 9.05079365079365, |
| "grad_norm": 1.5847752094268799, |
| "learning_rate": 4.394713478352955e-05, |
| "loss": 0.2715, |
| "step": 1430 |
| }, |
| { |
| "epoch": 9.082539682539682, |
| "grad_norm": 1.6062681674957275, |
| "learning_rate": 4.388656792816562e-05, |
| "loss": 0.2487, |
| "step": 1435 |
| }, |
| { |
| "epoch": 9.114285714285714, |
| "grad_norm": 2.099787712097168, |
| "learning_rate": 4.382574173521272e-05, |
| "loss": 0.2866, |
| "step": 1440 |
| }, |
| { |
| "epoch": 9.146031746031746, |
| "grad_norm": 1.0997334718704224, |
| "learning_rate": 4.376465703989502e-05, |
| "loss": 0.3052, |
| "step": 1445 |
| }, |
| { |
| "epoch": 9.177777777777777, |
| "grad_norm": 2.4327454566955566, |
| "learning_rate": 4.370331468098628e-05, |
| "loss": 0.3212, |
| "step": 1450 |
| }, |
| { |
| "epoch": 9.209523809523809, |
| "grad_norm": 1.4816385507583618, |
| "learning_rate": 4.364171550079833e-05, |
| "loss": 0.3046, |
| "step": 1455 |
| }, |
| { |
| "epoch": 9.24126984126984, |
| "grad_norm": 2.039186716079712, |
| "learning_rate": 4.357986034516947e-05, |
| "loss": 0.3165, |
| "step": 1460 |
| }, |
| { |
| "epoch": 9.273015873015874, |
| "grad_norm": 1.437852382659912, |
| "learning_rate": 4.3517750063452934e-05, |
| "loss": 0.3037, |
| "step": 1465 |
| }, |
| { |
| "epoch": 9.304761904761905, |
| "grad_norm": 1.818982720375061, |
| "learning_rate": 4.345538550850512e-05, |
| "loss": 0.3122, |
| "step": 1470 |
| }, |
| { |
| "epoch": 9.336507936507937, |
| "grad_norm": 1.12025785446167, |
| "learning_rate": 4.339276753667395e-05, |
| "loss": 0.2909, |
| "step": 1475 |
| }, |
| { |
| "epoch": 9.368253968253969, |
| "grad_norm": 1.6094844341278076, |
| "learning_rate": 4.3329897007787125e-05, |
| "loss": 0.2823, |
| "step": 1480 |
| }, |
| { |
| "epoch": 9.4, |
| "grad_norm": 1.916200041770935, |
| "learning_rate": 4.326677478514024e-05, |
| "loss": 0.2939, |
| "step": 1485 |
| }, |
| { |
| "epoch": 9.431746031746032, |
| "grad_norm": 1.97919499874115, |
| "learning_rate": 4.320340173548503e-05, |
| "loss": 0.2826, |
| "step": 1490 |
| }, |
| { |
| "epoch": 9.463492063492064, |
| "grad_norm": 2.0238938331604004, |
| "learning_rate": 4.313977872901737e-05, |
| "loss": 0.3273, |
| "step": 1495 |
| }, |
| { |
| "epoch": 9.495238095238095, |
| "grad_norm": 2.5840957164764404, |
| "learning_rate": 4.307590663936541e-05, |
| "loss": 0.2889, |
| "step": 1500 |
| }, |
| { |
| "epoch": 9.526984126984127, |
| "grad_norm": 2.3503904342651367, |
| "learning_rate": 4.30117863435775e-05, |
| "loss": 0.3012, |
| "step": 1505 |
| }, |
| { |
| "epoch": 9.558730158730159, |
| "grad_norm": 2.019792318344116, |
| "learning_rate": 4.294741872211024e-05, |
| "loss": 0.3267, |
| "step": 1510 |
| }, |
| { |
| "epoch": 9.59047619047619, |
| "grad_norm": 2.2713353633880615, |
| "learning_rate": 4.288280465881632e-05, |
| "loss": 0.3096, |
| "step": 1515 |
| }, |
| { |
| "epoch": 9.622222222222222, |
| "grad_norm": 2.4236693382263184, |
| "learning_rate": 4.281794504093237e-05, |
| "loss": 0.3291, |
| "step": 1520 |
| }, |
| { |
| "epoch": 9.653968253968253, |
| "grad_norm": 1.772703766822815, |
| "learning_rate": 4.275284075906686e-05, |
| "loss": 0.3117, |
| "step": 1525 |
| }, |
| { |
| "epoch": 9.685714285714285, |
| "grad_norm": 1.9665186405181885, |
| "learning_rate": 4.268749270718778e-05, |
| "loss": 0.326, |
| "step": 1530 |
| }, |
| { |
| "epoch": 9.717460317460317, |
| "grad_norm": 1.9472782611846924, |
| "learning_rate": 4.262190178261044e-05, |
| "loss": 0.2683, |
| "step": 1535 |
| }, |
| { |
| "epoch": 9.74920634920635, |
| "grad_norm": 2.0638089179992676, |
| "learning_rate": 4.255606888598508e-05, |
| "loss": 0.314, |
| "step": 1540 |
| }, |
| { |
| "epoch": 9.780952380952382, |
| "grad_norm": 2.1349925994873047, |
| "learning_rate": 4.248999492128456e-05, |
| "loss": 0.2897, |
| "step": 1545 |
| }, |
| { |
| "epoch": 9.812698412698413, |
| "grad_norm": 2.112536907196045, |
| "learning_rate": 4.242368079579192e-05, |
| "loss": 0.31, |
| "step": 1550 |
| }, |
| { |
| "epoch": 9.844444444444445, |
| "grad_norm": 1.6859878301620483, |
| "learning_rate": 4.2357127420087917e-05, |
| "loss": 0.3412, |
| "step": 1555 |
| }, |
| { |
| "epoch": 9.876190476190477, |
| "grad_norm": 1.9178651571273804, |
| "learning_rate": 4.229033570803853e-05, |
| "loss": 0.334, |
| "step": 1560 |
| }, |
| { |
| "epoch": 9.907936507936508, |
| "grad_norm": 2.562436103820801, |
| "learning_rate": 4.2223306576782426e-05, |
| "loss": 0.3379, |
| "step": 1565 |
| }, |
| { |
| "epoch": 9.93968253968254, |
| "grad_norm": 1.8472412824630737, |
| "learning_rate": 4.215604094671835e-05, |
| "loss": 0.3415, |
| "step": 1570 |
| }, |
| { |
| "epoch": 9.971428571428572, |
| "grad_norm": 1.9416279792785645, |
| "learning_rate": 4.208853974149246e-05, |
| "loss": 0.3085, |
| "step": 1575 |
| }, |
| { |
| "epoch": 10.0, |
| "grad_norm": 2.0056397914886475, |
| "learning_rate": 4.202080388798571e-05, |
| "loss": 0.3263, |
| "step": 1580 |
| }, |
| { |
| "epoch": 10.031746031746032, |
| "grad_norm": 2.195781946182251, |
| "learning_rate": 4.1952834316301065e-05, |
| "loss": 0.2867, |
| "step": 1585 |
| }, |
| { |
| "epoch": 10.063492063492063, |
| "grad_norm": 1.7489805221557617, |
| "learning_rate": 4.1884631959750766e-05, |
| "loss": 0.2589, |
| "step": 1590 |
| }, |
| { |
| "epoch": 10.095238095238095, |
| "grad_norm": 1.9361369609832764, |
| "learning_rate": 4.181619775484348e-05, |
| "loss": 0.2722, |
| "step": 1595 |
| }, |
| { |
| "epoch": 10.126984126984127, |
| "grad_norm": 2.24322509765625, |
| "learning_rate": 4.174753264127147e-05, |
| "loss": 0.2534, |
| "step": 1600 |
| }, |
| { |
| "epoch": 10.158730158730158, |
| "grad_norm": 2.4550466537475586, |
| "learning_rate": 4.167863756189767e-05, |
| "loss": 0.2777, |
| "step": 1605 |
| }, |
| { |
| "epoch": 10.19047619047619, |
| "grad_norm": 1.9439811706542969, |
| "learning_rate": 4.160951346274278e-05, |
| "loss": 0.2864, |
| "step": 1610 |
| }, |
| { |
| "epoch": 10.222222222222221, |
| "grad_norm": 1.633494257926941, |
| "learning_rate": 4.154016129297219e-05, |
| "loss": 0.2858, |
| "step": 1615 |
| }, |
| { |
| "epoch": 10.253968253968253, |
| "grad_norm": 1.69782292842865, |
| "learning_rate": 4.147058200488305e-05, |
| "loss": 0.2942, |
| "step": 1620 |
| }, |
| { |
| "epoch": 10.285714285714286, |
| "grad_norm": 1.613031268119812, |
| "learning_rate": 4.140077655389113e-05, |
| "loss": 0.2632, |
| "step": 1625 |
| }, |
| { |
| "epoch": 10.317460317460318, |
| "grad_norm": 2.0266177654266357, |
| "learning_rate": 4.1330745898517714e-05, |
| "loss": 0.3011, |
| "step": 1630 |
| }, |
| { |
| "epoch": 10.34920634920635, |
| "grad_norm": 1.8945387601852417, |
| "learning_rate": 4.1260491000376446e-05, |
| "loss": 0.2832, |
| "step": 1635 |
| }, |
| { |
| "epoch": 10.380952380952381, |
| "grad_norm": 1.7012510299682617, |
| "learning_rate": 4.119001282416009e-05, |
| "loss": 0.2718, |
| "step": 1640 |
| }, |
| { |
| "epoch": 10.412698412698413, |
| "grad_norm": 1.5538525581359863, |
| "learning_rate": 4.111931233762738e-05, |
| "loss": 0.3232, |
| "step": 1645 |
| }, |
| { |
| "epoch": 10.444444444444445, |
| "grad_norm": 2.3083150386810303, |
| "learning_rate": 4.1048390511589595e-05, |
| "loss": 0.3057, |
| "step": 1650 |
| }, |
| { |
| "epoch": 10.476190476190476, |
| "grad_norm": 1.293314814567566, |
| "learning_rate": 4.097724831989733e-05, |
| "loss": 0.2523, |
| "step": 1655 |
| }, |
| { |
| "epoch": 10.507936507936508, |
| "grad_norm": 2.517212152481079, |
| "learning_rate": 4.09058867394271e-05, |
| "loss": 0.3269, |
| "step": 1660 |
| }, |
| { |
| "epoch": 10.53968253968254, |
| "grad_norm": 2.057063102722168, |
| "learning_rate": 4.083430675006791e-05, |
| "loss": 0.2844, |
| "step": 1665 |
| }, |
| { |
| "epoch": 10.571428571428571, |
| "grad_norm": 1.5663833618164062, |
| "learning_rate": 4.0762509334707786e-05, |
| "loss": 0.3005, |
| "step": 1670 |
| }, |
| { |
| "epoch": 10.603174603174603, |
| "grad_norm": 2.5423505306243896, |
| "learning_rate": 4.069049547922035e-05, |
| "loss": 0.2802, |
| "step": 1675 |
| }, |
| { |
| "epoch": 10.634920634920634, |
| "grad_norm": 1.578316569328308, |
| "learning_rate": 4.061826617245119e-05, |
| "loss": 0.2667, |
| "step": 1680 |
| }, |
| { |
| "epoch": 10.666666666666666, |
| "grad_norm": 1.502928376197815, |
| "learning_rate": 4.0545822406204334e-05, |
| "loss": 0.3059, |
| "step": 1685 |
| }, |
| { |
| "epoch": 10.698412698412698, |
| "grad_norm": 1.2470905780792236, |
| "learning_rate": 4.047316517522864e-05, |
| "loss": 0.2879, |
| "step": 1690 |
| }, |
| { |
| "epoch": 10.73015873015873, |
| "grad_norm": 1.8238775730133057, |
| "learning_rate": 4.0400295477204105e-05, |
| "loss": 0.2923, |
| "step": 1695 |
| }, |
| { |
| "epoch": 10.761904761904763, |
| "grad_norm": 2.0516586303710938, |
| "learning_rate": 4.032721431272819e-05, |
| "loss": 0.3086, |
| "step": 1700 |
| }, |
| { |
| "epoch": 10.793650793650794, |
| "grad_norm": 1.3188791275024414, |
| "learning_rate": 4.0253922685302046e-05, |
| "loss": 0.2893, |
| "step": 1705 |
| }, |
| { |
| "epoch": 10.825396825396826, |
| "grad_norm": 1.7352266311645508, |
| "learning_rate": 4.01804216013168e-05, |
| "loss": 0.2981, |
| "step": 1710 |
| }, |
| { |
| "epoch": 10.857142857142858, |
| "grad_norm": 1.3449515104293823, |
| "learning_rate": 4.0106712070039656e-05, |
| "loss": 0.2841, |
| "step": 1715 |
| }, |
| { |
| "epoch": 10.88888888888889, |
| "grad_norm": 2.505431890487671, |
| "learning_rate": 4.00327951036001e-05, |
| "loss": 0.3034, |
| "step": 1720 |
| }, |
| { |
| "epoch": 10.920634920634921, |
| "grad_norm": 1.8870325088500977, |
| "learning_rate": 3.9958671716975966e-05, |
| "loss": 0.305, |
| "step": 1725 |
| }, |
| { |
| "epoch": 10.952380952380953, |
| "grad_norm": 2.913130044937134, |
| "learning_rate": 3.988434292797951e-05, |
| "loss": 0.3212, |
| "step": 1730 |
| }, |
| { |
| "epoch": 10.984126984126984, |
| "grad_norm": 1.7870115041732788, |
| "learning_rate": 3.980980975724344e-05, |
| "loss": 0.3108, |
| "step": 1735 |
| }, |
| { |
| "epoch": 11.012698412698413, |
| "grad_norm": 3.050985336303711, |
| "learning_rate": 3.9735073228206896e-05, |
| "loss": 0.3043, |
| "step": 1740 |
| }, |
| { |
| "epoch": 11.044444444444444, |
| "grad_norm": 1.5993611812591553, |
| "learning_rate": 3.96601343671014e-05, |
| "loss": 0.2465, |
| "step": 1745 |
| }, |
| { |
| "epoch": 11.076190476190476, |
| "grad_norm": 1.626888632774353, |
| "learning_rate": 3.9584994202936746e-05, |
| "loss": 0.2688, |
| "step": 1750 |
| }, |
| { |
| "epoch": 11.107936507936508, |
| "grad_norm": 1.7132880687713623, |
| "learning_rate": 3.950965376748689e-05, |
| "loss": 0.2458, |
| "step": 1755 |
| }, |
| { |
| "epoch": 11.13968253968254, |
| "grad_norm": 1.7764930725097656, |
| "learning_rate": 3.94341140952758e-05, |
| "loss": 0.2189, |
| "step": 1760 |
| }, |
| { |
| "epoch": 11.17142857142857, |
| "grad_norm": 2.5560712814331055, |
| "learning_rate": 3.9358376223563206e-05, |
| "loss": 0.2866, |
| "step": 1765 |
| }, |
| { |
| "epoch": 11.203174603174602, |
| "grad_norm": 1.1177359819412231, |
| "learning_rate": 3.928244119233038e-05, |
| "loss": 0.233, |
| "step": 1770 |
| }, |
| { |
| "epoch": 11.234920634920634, |
| "grad_norm": 1.584670901298523, |
| "learning_rate": 3.9206310044265866e-05, |
| "loss": 0.273, |
| "step": 1775 |
| }, |
| { |
| "epoch": 11.266666666666667, |
| "grad_norm": 1.6278687715530396, |
| "learning_rate": 3.912998382475115e-05, |
| "loss": 0.2746, |
| "step": 1780 |
| }, |
| { |
| "epoch": 11.2984126984127, |
| "grad_norm": 1.657038688659668, |
| "learning_rate": 3.905346358184629e-05, |
| "loss": 0.2885, |
| "step": 1785 |
| }, |
| { |
| "epoch": 11.33015873015873, |
| "grad_norm": 1.2840272188186646, |
| "learning_rate": 3.897675036627557e-05, |
| "loss": 0.2932, |
| "step": 1790 |
| }, |
| { |
| "epoch": 11.361904761904762, |
| "grad_norm": 1.5678766965866089, |
| "learning_rate": 3.8899845231413026e-05, |
| "loss": 0.2945, |
| "step": 1795 |
| }, |
| { |
| "epoch": 11.393650793650794, |
| "grad_norm": 1.788948655128479, |
| "learning_rate": 3.8822749233268006e-05, |
| "loss": 0.3013, |
| "step": 1800 |
| }, |
| { |
| "epoch": 11.425396825396826, |
| "grad_norm": 1.2259769439697266, |
| "learning_rate": 3.8745463430470664e-05, |
| "loss": 0.2582, |
| "step": 1805 |
| }, |
| { |
| "epoch": 11.457142857142857, |
| "grad_norm": 1.5430735349655151, |
| "learning_rate": 3.866798888425741e-05, |
| "loss": 0.275, |
| "step": 1810 |
| }, |
| { |
| "epoch": 11.488888888888889, |
| "grad_norm": 1.9102168083190918, |
| "learning_rate": 3.8590326658456376e-05, |
| "loss": 0.2909, |
| "step": 1815 |
| }, |
| { |
| "epoch": 11.52063492063492, |
| "grad_norm": 1.6118320226669312, |
| "learning_rate": 3.851247781947277e-05, |
| "loss": 0.2922, |
| "step": 1820 |
| }, |
| { |
| "epoch": 11.552380952380952, |
| "grad_norm": 1.393646478652954, |
| "learning_rate": 3.843444343627424e-05, |
| "loss": 0.2783, |
| "step": 1825 |
| }, |
| { |
| "epoch": 11.584126984126984, |
| "grad_norm": 2.522909641265869, |
| "learning_rate": 3.83562245803762e-05, |
| "loss": 0.2933, |
| "step": 1830 |
| }, |
| { |
| "epoch": 11.615873015873015, |
| "grad_norm": 2.2534332275390625, |
| "learning_rate": 3.827782232582714e-05, |
| "loss": 0.3081, |
| "step": 1835 |
| }, |
| { |
| "epoch": 11.647619047619047, |
| "grad_norm": 2.4088056087493896, |
| "learning_rate": 3.819923774919383e-05, |
| "loss": 0.276, |
| "step": 1840 |
| }, |
| { |
| "epoch": 11.679365079365079, |
| "grad_norm": 1.7626562118530273, |
| "learning_rate": 3.8120471929546576e-05, |
| "loss": 0.2697, |
| "step": 1845 |
| }, |
| { |
| "epoch": 11.71111111111111, |
| "grad_norm": 1.8656691312789917, |
| "learning_rate": 3.8041525948444414e-05, |
| "loss": 0.2979, |
| "step": 1850 |
| }, |
| { |
| "epoch": 11.742857142857144, |
| "grad_norm": 1.6537258625030518, |
| "learning_rate": 3.7962400889920185e-05, |
| "loss": 0.3042, |
| "step": 1855 |
| }, |
| { |
| "epoch": 11.774603174603175, |
| "grad_norm": 1.696975827217102, |
| "learning_rate": 3.788309784046574e-05, |
| "loss": 0.2984, |
| "step": 1860 |
| }, |
| { |
| "epoch": 11.806349206349207, |
| "grad_norm": 1.976236343383789, |
| "learning_rate": 3.780361788901696e-05, |
| "loss": 0.2711, |
| "step": 1865 |
| }, |
| { |
| "epoch": 11.838095238095239, |
| "grad_norm": 1.7781676054000854, |
| "learning_rate": 3.772396212693885e-05, |
| "loss": 0.3116, |
| "step": 1870 |
| }, |
| { |
| "epoch": 11.86984126984127, |
| "grad_norm": 1.6252080202102661, |
| "learning_rate": 3.7644131648010494e-05, |
| "loss": 0.2879, |
| "step": 1875 |
| }, |
| { |
| "epoch": 11.901587301587302, |
| "grad_norm": 1.7511012554168701, |
| "learning_rate": 3.75641275484101e-05, |
| "loss": 0.3106, |
| "step": 1880 |
| }, |
| { |
| "epoch": 11.933333333333334, |
| "grad_norm": 1.5861046314239502, |
| "learning_rate": 3.7483950926699885e-05, |
| "loss": 0.2703, |
| "step": 1885 |
| }, |
| { |
| "epoch": 11.965079365079365, |
| "grad_norm": 1.903902530670166, |
| "learning_rate": 3.740360288381105e-05, |
| "loss": 0.2808, |
| "step": 1890 |
| }, |
| { |
| "epoch": 11.996825396825397, |
| "grad_norm": 1.644740104675293, |
| "learning_rate": 3.732308452302864e-05, |
| "loss": 0.2883, |
| "step": 1895 |
| }, |
| { |
| "epoch": 12.025396825396825, |
| "grad_norm": 1.3671797513961792, |
| "learning_rate": 3.724239694997637e-05, |
| "loss": 0.2661, |
| "step": 1900 |
| }, |
| { |
| "epoch": 12.057142857142857, |
| "grad_norm": 1.1910501718521118, |
| "learning_rate": 3.716154127260147e-05, |
| "loss": 0.2352, |
| "step": 1905 |
| }, |
| { |
| "epoch": 12.088888888888889, |
| "grad_norm": 1.45137619972229, |
| "learning_rate": 3.708051860115947e-05, |
| "loss": 0.2703, |
| "step": 1910 |
| }, |
| { |
| "epoch": 12.12063492063492, |
| "grad_norm": 2.625089645385742, |
| "learning_rate": 3.699933004819895e-05, |
| "loss": 0.2524, |
| "step": 1915 |
| }, |
| { |
| "epoch": 12.152380952380952, |
| "grad_norm": 1.731602430343628, |
| "learning_rate": 3.691797672854625e-05, |
| "loss": 0.2533, |
| "step": 1920 |
| }, |
| { |
| "epoch": 12.184126984126983, |
| "grad_norm": 2.1266143321990967, |
| "learning_rate": 3.683645975929019e-05, |
| "loss": 0.2666, |
| "step": 1925 |
| }, |
| { |
| "epoch": 12.215873015873015, |
| "grad_norm": 1.7163398265838623, |
| "learning_rate": 3.675478025976671e-05, |
| "loss": 0.2838, |
| "step": 1930 |
| }, |
| { |
| "epoch": 12.247619047619047, |
| "grad_norm": 1.8726729154586792, |
| "learning_rate": 3.66729393515435e-05, |
| "loss": 0.2646, |
| "step": 1935 |
| }, |
| { |
| "epoch": 12.27936507936508, |
| "grad_norm": 1.2229722738265991, |
| "learning_rate": 3.659093815840462e-05, |
| "loss": 0.267, |
| "step": 1940 |
| }, |
| { |
| "epoch": 12.311111111111112, |
| "grad_norm": 1.341997504234314, |
| "learning_rate": 3.650877780633505e-05, |
| "loss": 0.2464, |
| "step": 1945 |
| }, |
| { |
| "epoch": 12.342857142857143, |
| "grad_norm": 1.3466174602508545, |
| "learning_rate": 3.6426459423505214e-05, |
| "loss": 0.2521, |
| "step": 1950 |
| }, |
| { |
| "epoch": 12.374603174603175, |
| "grad_norm": 1.7022134065628052, |
| "learning_rate": 3.6343984140255516e-05, |
| "loss": 0.2662, |
| "step": 1955 |
| }, |
| { |
| "epoch": 12.406349206349207, |
| "grad_norm": 2.5058987140655518, |
| "learning_rate": 3.626135308908084e-05, |
| "loss": 0.2745, |
| "step": 1960 |
| }, |
| { |
| "epoch": 12.438095238095238, |
| "grad_norm": 1.7667903900146484, |
| "learning_rate": 3.6178567404614936e-05, |
| "loss": 0.2752, |
| "step": 1965 |
| }, |
| { |
| "epoch": 12.46984126984127, |
| "grad_norm": 1.7334171533584595, |
| "learning_rate": 3.609562822361487e-05, |
| "loss": 0.2667, |
| "step": 1970 |
| }, |
| { |
| "epoch": 12.501587301587302, |
| "grad_norm": 1.7093929052352905, |
| "learning_rate": 3.601253668494546e-05, |
| "loss": 0.2829, |
| "step": 1975 |
| }, |
| { |
| "epoch": 12.533333333333333, |
| "grad_norm": 1.8272080421447754, |
| "learning_rate": 3.592929392956355e-05, |
| "loss": 0.2583, |
| "step": 1980 |
| }, |
| { |
| "epoch": 12.565079365079365, |
| "grad_norm": 1.5226343870162964, |
| "learning_rate": 3.584590110050241e-05, |
| "loss": 0.2652, |
| "step": 1985 |
| }, |
| { |
| "epoch": 12.596825396825396, |
| "grad_norm": 1.7395323514938354, |
| "learning_rate": 3.5762359342856036e-05, |
| "loss": 0.2585, |
| "step": 1990 |
| }, |
| { |
| "epoch": 12.628571428571428, |
| "grad_norm": 2.278144359588623, |
| "learning_rate": 3.567866980376337e-05, |
| "loss": 0.2862, |
| "step": 1995 |
| }, |
| { |
| "epoch": 12.66031746031746, |
| "grad_norm": 1.9335429668426514, |
| "learning_rate": 3.559483363239262e-05, |
| "loss": 0.2945, |
| "step": 2000 |
| }, |
| { |
| "epoch": 12.692063492063491, |
| "grad_norm": 1.3866902589797974, |
| "learning_rate": 3.551085197992545e-05, |
| "loss": 0.2675, |
| "step": 2005 |
| }, |
| { |
| "epoch": 12.723809523809523, |
| "grad_norm": 1.7110604047775269, |
| "learning_rate": 3.5426725999541174e-05, |
| "loss": 0.2705, |
| "step": 2010 |
| }, |
| { |
| "epoch": 12.755555555555556, |
| "grad_norm": 1.6411776542663574, |
| "learning_rate": 3.534245684640089e-05, |
| "loss": 0.2482, |
| "step": 2015 |
| }, |
| { |
| "epoch": 12.787301587301588, |
| "grad_norm": 1.4663981199264526, |
| "learning_rate": 3.525804567763167e-05, |
| "loss": 0.2838, |
| "step": 2020 |
| }, |
| { |
| "epoch": 12.81904761904762, |
| "grad_norm": 1.8676140308380127, |
| "learning_rate": 3.517349365231065e-05, |
| "loss": 0.2906, |
| "step": 2025 |
| }, |
| { |
| "epoch": 12.850793650793651, |
| "grad_norm": 1.6150639057159424, |
| "learning_rate": 3.508880193144911e-05, |
| "loss": 0.3011, |
| "step": 2030 |
| }, |
| { |
| "epoch": 12.882539682539683, |
| "grad_norm": 2.5578598976135254, |
| "learning_rate": 3.500397167797654e-05, |
| "loss": 0.2846, |
| "step": 2035 |
| }, |
| { |
| "epoch": 12.914285714285715, |
| "grad_norm": 1.3131941556930542, |
| "learning_rate": 3.491900405672466e-05, |
| "loss": 0.2509, |
| "step": 2040 |
| }, |
| { |
| "epoch": 12.946031746031746, |
| "grad_norm": 1.5713857412338257, |
| "learning_rate": 3.483390023441141e-05, |
| "loss": 0.2824, |
| "step": 2045 |
| }, |
| { |
| "epoch": 12.977777777777778, |
| "grad_norm": 1.788149356842041, |
| "learning_rate": 3.4748661379625e-05, |
| "loss": 0.2932, |
| "step": 2050 |
| }, |
| { |
| "epoch": 13.006349206349206, |
| "grad_norm": 1.080249547958374, |
| "learning_rate": 3.466328866280778e-05, |
| "loss": 0.2817, |
| "step": 2055 |
| }, |
| { |
| "epoch": 13.038095238095238, |
| "grad_norm": 1.1128981113433838, |
| "learning_rate": 3.45777832562402e-05, |
| "loss": 0.2141, |
| "step": 2060 |
| }, |
| { |
| "epoch": 13.06984126984127, |
| "grad_norm": 1.6394922733306885, |
| "learning_rate": 3.449214633402473e-05, |
| "loss": 0.216, |
| "step": 2065 |
| }, |
| { |
| "epoch": 13.101587301587301, |
| "grad_norm": 2.198416233062744, |
| "learning_rate": 3.440637907206973e-05, |
| "loss": 0.258, |
| "step": 2070 |
| }, |
| { |
| "epoch": 13.133333333333333, |
| "grad_norm": 1.5107051134109497, |
| "learning_rate": 3.4320482648073285e-05, |
| "loss": 0.2422, |
| "step": 2075 |
| }, |
| { |
| "epoch": 13.165079365079364, |
| "grad_norm": 1.7206802368164062, |
| "learning_rate": 3.4234458241507034e-05, |
| "loss": 0.2524, |
| "step": 2080 |
| }, |
| { |
| "epoch": 13.196825396825396, |
| "grad_norm": 1.4832013845443726, |
| "learning_rate": 3.4148307033600014e-05, |
| "loss": 0.27, |
| "step": 2085 |
| }, |
| { |
| "epoch": 13.228571428571428, |
| "grad_norm": 1.592344880104065, |
| "learning_rate": 3.4062030207322395e-05, |
| "loss": 0.2486, |
| "step": 2090 |
| }, |
| { |
| "epoch": 13.260317460317461, |
| "grad_norm": 1.7398722171783447, |
| "learning_rate": 3.397562894736926e-05, |
| "loss": 0.2605, |
| "step": 2095 |
| }, |
| { |
| "epoch": 13.292063492063493, |
| "grad_norm": 2.0896050930023193, |
| "learning_rate": 3.388910444014432e-05, |
| "loss": 0.2562, |
| "step": 2100 |
| }, |
| { |
| "epoch": 13.323809523809524, |
| "grad_norm": 2.7245311737060547, |
| "learning_rate": 3.380245787374365e-05, |
| "loss": 0.2573, |
| "step": 2105 |
| }, |
| { |
| "epoch": 13.355555555555556, |
| "grad_norm": 1.1951723098754883, |
| "learning_rate": 3.371569043793936e-05, |
| "loss": 0.2709, |
| "step": 2110 |
| }, |
| { |
| "epoch": 13.387301587301588, |
| "grad_norm": 1.697736144065857, |
| "learning_rate": 3.3628803324163236e-05, |
| "loss": 0.2413, |
| "step": 2115 |
| }, |
| { |
| "epoch": 13.41904761904762, |
| "grad_norm": 1.447302222251892, |
| "learning_rate": 3.354179772549042e-05, |
| "loss": 0.2685, |
| "step": 2120 |
| }, |
| { |
| "epoch": 13.450793650793651, |
| "grad_norm": 1.232428789138794, |
| "learning_rate": 3.3454674836623e-05, |
| "loss": 0.2546, |
| "step": 2125 |
| }, |
| { |
| "epoch": 13.482539682539683, |
| "grad_norm": 1.889103889465332, |
| "learning_rate": 3.336743585387362e-05, |
| "loss": 0.2412, |
| "step": 2130 |
| }, |
| { |
| "epoch": 13.514285714285714, |
| "grad_norm": 1.888352632522583, |
| "learning_rate": 3.328008197514903e-05, |
| "loss": 0.2737, |
| "step": 2135 |
| }, |
| { |
| "epoch": 13.546031746031746, |
| "grad_norm": 1.4205390214920044, |
| "learning_rate": 3.319261439993369e-05, |
| "loss": 0.2727, |
| "step": 2140 |
| }, |
| { |
| "epoch": 13.577777777777778, |
| "grad_norm": 1.442429542541504, |
| "learning_rate": 3.310503432927322e-05, |
| "loss": 0.2894, |
| "step": 2145 |
| }, |
| { |
| "epoch": 13.60952380952381, |
| "grad_norm": 1.413885474205017, |
| "learning_rate": 3.3017342965757986e-05, |
| "loss": 0.2928, |
| "step": 2150 |
| }, |
| { |
| "epoch": 13.64126984126984, |
| "grad_norm": 1.0464482307434082, |
| "learning_rate": 3.2929541513506524e-05, |
| "loss": 0.252, |
| "step": 2155 |
| }, |
| { |
| "epoch": 13.673015873015872, |
| "grad_norm": 1.6716376543045044, |
| "learning_rate": 3.284163117814906e-05, |
| "loss": 0.2614, |
| "step": 2160 |
| }, |
| { |
| "epoch": 13.704761904761904, |
| "grad_norm": 1.9366419315338135, |
| "learning_rate": 3.2753613166810914e-05, |
| "loss": 0.2963, |
| "step": 2165 |
| }, |
| { |
| "epoch": 13.736507936507937, |
| "grad_norm": 1.4641704559326172, |
| "learning_rate": 3.2665488688095944e-05, |
| "loss": 0.2654, |
| "step": 2170 |
| }, |
| { |
| "epoch": 13.768253968253969, |
| "grad_norm": 1.3404383659362793, |
| "learning_rate": 3.2577258952069934e-05, |
| "loss": 0.2601, |
| "step": 2175 |
| }, |
| { |
| "epoch": 13.8, |
| "grad_norm": 1.6471940279006958, |
| "learning_rate": 3.2488925170244015e-05, |
| "loss": 0.2775, |
| "step": 2180 |
| }, |
| { |
| "epoch": 13.831746031746032, |
| "grad_norm": 1.3665790557861328, |
| "learning_rate": 3.240048855555798e-05, |
| "loss": 0.2602, |
| "step": 2185 |
| }, |
| { |
| "epoch": 13.863492063492064, |
| "grad_norm": 1.1486542224884033, |
| "learning_rate": 3.2311950322363685e-05, |
| "loss": 0.2603, |
| "step": 2190 |
| }, |
| { |
| "epoch": 13.895238095238096, |
| "grad_norm": 1.45487642288208, |
| "learning_rate": 3.2223311686408304e-05, |
| "loss": 0.2794, |
| "step": 2195 |
| }, |
| { |
| "epoch": 13.926984126984127, |
| "grad_norm": 1.6884818077087402, |
| "learning_rate": 3.213457386481769e-05, |
| "loss": 0.2682, |
| "step": 2200 |
| }, |
| { |
| "epoch": 13.958730158730159, |
| "grad_norm": 1.9160594940185547, |
| "learning_rate": 3.204573807607967e-05, |
| "loss": 0.2619, |
| "step": 2205 |
| }, |
| { |
| "epoch": 13.99047619047619, |
| "grad_norm": 1.9444315433502197, |
| "learning_rate": 3.1956805540027244e-05, |
| "loss": 0.3029, |
| "step": 2210 |
| }, |
| { |
| "epoch": 14.019047619047619, |
| "grad_norm": 1.8739337921142578, |
| "learning_rate": 3.1867777477821915e-05, |
| "loss": 0.2298, |
| "step": 2215 |
| }, |
| { |
| "epoch": 14.05079365079365, |
| "grad_norm": 1.3971226215362549, |
| "learning_rate": 3.1778655111936866e-05, |
| "loss": 0.2351, |
| "step": 2220 |
| }, |
| { |
| "epoch": 14.082539682539682, |
| "grad_norm": 1.288387417793274, |
| "learning_rate": 3.168943966614022e-05, |
| "loss": 0.2331, |
| "step": 2225 |
| }, |
| { |
| "epoch": 14.114285714285714, |
| "grad_norm": 1.2136940956115723, |
| "learning_rate": 3.1600132365478186e-05, |
| "loss": 0.2406, |
| "step": 2230 |
| }, |
| { |
| "epoch": 14.146031746031746, |
| "grad_norm": 1.8233654499053955, |
| "learning_rate": 3.151073443625828e-05, |
| "loss": 0.2496, |
| "step": 2235 |
| }, |
| { |
| "epoch": 14.177777777777777, |
| "grad_norm": 1.7613590955734253, |
| "learning_rate": 3.142124710603245e-05, |
| "loss": 0.2523, |
| "step": 2240 |
| }, |
| { |
| "epoch": 14.209523809523809, |
| "grad_norm": 1.8933606147766113, |
| "learning_rate": 3.133167160358024e-05, |
| "loss": 0.2646, |
| "step": 2245 |
| }, |
| { |
| "epoch": 14.24126984126984, |
| "grad_norm": 2.7858712673187256, |
| "learning_rate": 3.124200915889195e-05, |
| "loss": 0.2818, |
| "step": 2250 |
| }, |
| { |
| "epoch": 14.273015873015874, |
| "grad_norm": 1.5656853914260864, |
| "learning_rate": 3.115226100315166e-05, |
| "loss": 0.2389, |
| "step": 2255 |
| }, |
| { |
| "epoch": 14.304761904761905, |
| "grad_norm": 1.606392741203308, |
| "learning_rate": 3.106242836872041e-05, |
| "loss": 0.2302, |
| "step": 2260 |
| }, |
| { |
| "epoch": 14.336507936507937, |
| "grad_norm": 2.1553008556365967, |
| "learning_rate": 3.097251248911922e-05, |
| "loss": 0.2484, |
| "step": 2265 |
| }, |
| { |
| "epoch": 14.368253968253969, |
| "grad_norm": 1.7176517248153687, |
| "learning_rate": 3.088251459901222e-05, |
| "loss": 0.268, |
| "step": 2270 |
| }, |
| { |
| "epoch": 14.4, |
| "grad_norm": 1.3635174036026, |
| "learning_rate": 3.0792435934189604e-05, |
| "loss": 0.2491, |
| "step": 2275 |
| }, |
| { |
| "epoch": 14.431746031746032, |
| "grad_norm": 1.4257878065109253, |
| "learning_rate": 3.070227773155074e-05, |
| "loss": 0.2884, |
| "step": 2280 |
| }, |
| { |
| "epoch": 14.463492063492064, |
| "grad_norm": 1.1561063528060913, |
| "learning_rate": 3.0612041229087145e-05, |
| "loss": 0.247, |
| "step": 2285 |
| }, |
| { |
| "epoch": 14.495238095238095, |
| "grad_norm": 1.6131694316864014, |
| "learning_rate": 3.0521727665865485e-05, |
| "loss": 0.2708, |
| "step": 2290 |
| }, |
| { |
| "epoch": 14.526984126984127, |
| "grad_norm": 1.7133935689926147, |
| "learning_rate": 3.0431338282010606e-05, |
| "loss": 0.2384, |
| "step": 2295 |
| }, |
| { |
| "epoch": 14.558730158730159, |
| "grad_norm": 1.3232437372207642, |
| "learning_rate": 3.0340874318688433e-05, |
| "loss": 0.2503, |
| "step": 2300 |
| }, |
| { |
| "epoch": 14.59047619047619, |
| "grad_norm": 1.5612751245498657, |
| "learning_rate": 3.0250337018088994e-05, |
| "loss": 0.2618, |
| "step": 2305 |
| }, |
| { |
| "epoch": 14.622222222222222, |
| "grad_norm": 1.279241681098938, |
| "learning_rate": 3.0159727623409313e-05, |
| "loss": 0.2535, |
| "step": 2310 |
| }, |
| { |
| "epoch": 14.653968253968253, |
| "grad_norm": 3.142040729522705, |
| "learning_rate": 3.0069047378836375e-05, |
| "loss": 0.2745, |
| "step": 2315 |
| }, |
| { |
| "epoch": 14.685714285714285, |
| "grad_norm": 1.3240234851837158, |
| "learning_rate": 2.9978297529530036e-05, |
| "loss": 0.2591, |
| "step": 2320 |
| }, |
| { |
| "epoch": 14.717460317460317, |
| "grad_norm": 1.121077537536621, |
| "learning_rate": 2.9887479321605895e-05, |
| "loss": 0.2453, |
| "step": 2325 |
| }, |
| { |
| "epoch": 14.74920634920635, |
| "grad_norm": 2.3831980228424072, |
| "learning_rate": 2.9796594002118218e-05, |
| "loss": 0.257, |
| "step": 2330 |
| }, |
| { |
| "epoch": 14.780952380952382, |
| "grad_norm": 2.175352096557617, |
| "learning_rate": 2.9705642819042805e-05, |
| "loss": 0.2434, |
| "step": 2335 |
| }, |
| { |
| "epoch": 14.812698412698413, |
| "grad_norm": 1.5452053546905518, |
| "learning_rate": 2.9614627021259846e-05, |
| "loss": 0.2683, |
| "step": 2340 |
| }, |
| { |
| "epoch": 14.844444444444445, |
| "grad_norm": 1.6765031814575195, |
| "learning_rate": 2.952354785853677e-05, |
| "loss": 0.2484, |
| "step": 2345 |
| }, |
| { |
| "epoch": 14.876190476190477, |
| "grad_norm": 1.3790463209152222, |
| "learning_rate": 2.9432406581511096e-05, |
| "loss": 0.2636, |
| "step": 2350 |
| }, |
| { |
| "epoch": 14.907936507936508, |
| "grad_norm": 1.2849820852279663, |
| "learning_rate": 2.9341204441673266e-05, |
| "loss": 0.2734, |
| "step": 2355 |
| }, |
| { |
| "epoch": 14.93968253968254, |
| "grad_norm": 1.5512354373931885, |
| "learning_rate": 2.924994269134942e-05, |
| "loss": 0.2627, |
| "step": 2360 |
| }, |
| { |
| "epoch": 14.971428571428572, |
| "grad_norm": 1.2973589897155762, |
| "learning_rate": 2.9158622583684264e-05, |
| "loss": 0.2464, |
| "step": 2365 |
| }, |
| { |
| "epoch": 15.0, |
| "grad_norm": 2.749181032180786, |
| "learning_rate": 2.906724537262381e-05, |
| "loss": 0.2622, |
| "step": 2370 |
| }, |
| { |
| "epoch": 15.031746031746032, |
| "grad_norm": 1.4415572881698608, |
| "learning_rate": 2.8975812312898152e-05, |
| "loss": 0.2395, |
| "step": 2375 |
| }, |
| { |
| "epoch": 15.063492063492063, |
| "grad_norm": 1.41774320602417, |
| "learning_rate": 2.8884324660004325e-05, |
| "loss": 0.2213, |
| "step": 2380 |
| }, |
| { |
| "epoch": 15.095238095238095, |
| "grad_norm": 1.375585675239563, |
| "learning_rate": 2.8792783670188927e-05, |
| "loss": 0.2319, |
| "step": 2385 |
| }, |
| { |
| "epoch": 15.126984126984127, |
| "grad_norm": 1.2285529375076294, |
| "learning_rate": 2.870119060043099e-05, |
| "loss": 0.2317, |
| "step": 2390 |
| }, |
| { |
| "epoch": 15.158730158730158, |
| "grad_norm": 2.5581822395324707, |
| "learning_rate": 2.8609546708424645e-05, |
| "loss": 0.2567, |
| "step": 2395 |
| }, |
| { |
| "epoch": 15.19047619047619, |
| "grad_norm": 2.0000274181365967, |
| "learning_rate": 2.8517853252561906e-05, |
| "loss": 0.2265, |
| "step": 2400 |
| }, |
| { |
| "epoch": 15.222222222222221, |
| "grad_norm": 1.6797828674316406, |
| "learning_rate": 2.8426111491915342e-05, |
| "loss": 0.243, |
| "step": 2405 |
| }, |
| { |
| "epoch": 15.253968253968253, |
| "grad_norm": 1.8280093669891357, |
| "learning_rate": 2.8334322686220826e-05, |
| "loss": 0.2442, |
| "step": 2410 |
| }, |
| { |
| "epoch": 15.285714285714286, |
| "grad_norm": 1.6335142850875854, |
| "learning_rate": 2.824248809586021e-05, |
| "loss": 0.229, |
| "step": 2415 |
| }, |
| { |
| "epoch": 15.317460317460318, |
| "grad_norm": 2.1388628482818604, |
| "learning_rate": 2.8150608981844034e-05, |
| "loss": 0.247, |
| "step": 2420 |
| }, |
| { |
| "epoch": 15.34920634920635, |
| "grad_norm": 1.5680638551712036, |
| "learning_rate": 2.805868660579422e-05, |
| "loss": 0.2269, |
| "step": 2425 |
| }, |
| { |
| "epoch": 15.380952380952381, |
| "grad_norm": 1.6333919763565063, |
| "learning_rate": 2.7966722229926712e-05, |
| "loss": 0.2426, |
| "step": 2430 |
| }, |
| { |
| "epoch": 15.412698412698413, |
| "grad_norm": 1.4023231267929077, |
| "learning_rate": 2.787471711703419e-05, |
| "loss": 0.2466, |
| "step": 2435 |
| }, |
| { |
| "epoch": 15.444444444444445, |
| "grad_norm": 1.5598480701446533, |
| "learning_rate": 2.7782672530468672e-05, |
| "loss": 0.2504, |
| "step": 2440 |
| }, |
| { |
| "epoch": 15.476190476190476, |
| "grad_norm": 1.3230295181274414, |
| "learning_rate": 2.769058973412424e-05, |
| "loss": 0.2478, |
| "step": 2445 |
| }, |
| { |
| "epoch": 15.507936507936508, |
| "grad_norm": 2.0796375274658203, |
| "learning_rate": 2.7598469992419646e-05, |
| "loss": 0.2382, |
| "step": 2450 |
| }, |
| { |
| "epoch": 15.53968253968254, |
| "grad_norm": 1.6772873401641846, |
| "learning_rate": 2.7506314570280916e-05, |
| "loss": 0.2716, |
| "step": 2455 |
| }, |
| { |
| "epoch": 15.571428571428571, |
| "grad_norm": 1.9949897527694702, |
| "learning_rate": 2.7414124733124046e-05, |
| "loss": 0.2541, |
| "step": 2460 |
| }, |
| { |
| "epoch": 15.603174603174603, |
| "grad_norm": 1.339775562286377, |
| "learning_rate": 2.7321901746837585e-05, |
| "loss": 0.238, |
| "step": 2465 |
| }, |
| { |
| "epoch": 15.634920634920634, |
| "grad_norm": 1.3870205879211426, |
| "learning_rate": 2.7229646877765264e-05, |
| "loss": 0.2377, |
| "step": 2470 |
| }, |
| { |
| "epoch": 15.666666666666666, |
| "grad_norm": 2.1485989093780518, |
| "learning_rate": 2.7137361392688613e-05, |
| "loss": 0.2748, |
| "step": 2475 |
| }, |
| { |
| "epoch": 15.698412698412698, |
| "grad_norm": 1.3165838718414307, |
| "learning_rate": 2.704504655880955e-05, |
| "loss": 0.2592, |
| "step": 2480 |
| }, |
| { |
| "epoch": 15.73015873015873, |
| "grad_norm": 1.5217639207839966, |
| "learning_rate": 2.6952703643732983e-05, |
| "loss": 0.2359, |
| "step": 2485 |
| }, |
| { |
| "epoch": 15.761904761904763, |
| "grad_norm": 1.6285779476165771, |
| "learning_rate": 2.686033391544945e-05, |
| "loss": 0.2653, |
| "step": 2490 |
| }, |
| { |
| "epoch": 15.793650793650794, |
| "grad_norm": 1.6619691848754883, |
| "learning_rate": 2.676793864231763e-05, |
| "loss": 0.2572, |
| "step": 2495 |
| }, |
| { |
| "epoch": 15.825396825396826, |
| "grad_norm": 1.6636338233947754, |
| "learning_rate": 2.6675519093046987e-05, |
| "loss": 0.2702, |
| "step": 2500 |
| }, |
| { |
| "epoch": 15.857142857142858, |
| "grad_norm": 1.6794034242630005, |
| "learning_rate": 2.6583076536680323e-05, |
| "loss": 0.2508, |
| "step": 2505 |
| }, |
| { |
| "epoch": 15.88888888888889, |
| "grad_norm": 1.669232964515686, |
| "learning_rate": 2.6490612242576344e-05, |
| "loss": 0.2601, |
| "step": 2510 |
| }, |
| { |
| "epoch": 15.920634920634921, |
| "grad_norm": 1.4551194906234741, |
| "learning_rate": 2.639812748039228e-05, |
| "loss": 0.2514, |
| "step": 2515 |
| }, |
| { |
| "epoch": 15.952380952380953, |
| "grad_norm": 1.8457551002502441, |
| "learning_rate": 2.6305623520066382e-05, |
| "loss": 0.2682, |
| "step": 2520 |
| }, |
| { |
| "epoch": 15.984126984126984, |
| "grad_norm": 1.318781852722168, |
| "learning_rate": 2.6213101631800524e-05, |
| "loss": 0.2452, |
| "step": 2525 |
| }, |
| { |
| "epoch": 16.012698412698413, |
| "grad_norm": 1.4368094205856323, |
| "learning_rate": 2.6120563086042753e-05, |
| "loss": 0.2398, |
| "step": 2530 |
| }, |
| { |
| "epoch": 16.044444444444444, |
| "grad_norm": 1.314927101135254, |
| "learning_rate": 2.602800915346986e-05, |
| "loss": 0.2321, |
| "step": 2535 |
| }, |
| { |
| "epoch": 16.076190476190476, |
| "grad_norm": 1.501521110534668, |
| "learning_rate": 2.5935441104969914e-05, |
| "loss": 0.2329, |
| "step": 2540 |
| }, |
| { |
| "epoch": 16.107936507936508, |
| "grad_norm": 1.3708770275115967, |
| "learning_rate": 2.5842860211624802e-05, |
| "loss": 0.2272, |
| "step": 2545 |
| }, |
| { |
| "epoch": 16.13968253968254, |
| "grad_norm": 1.6570358276367188, |
| "learning_rate": 2.5750267744692786e-05, |
| "loss": 0.2323, |
| "step": 2550 |
| }, |
| { |
| "epoch": 16.17142857142857, |
| "grad_norm": 1.4252843856811523, |
| "learning_rate": 2.5657664975591063e-05, |
| "loss": 0.2308, |
| "step": 2555 |
| }, |
| { |
| "epoch": 16.203174603174602, |
| "grad_norm": 1.9588584899902344, |
| "learning_rate": 2.5565053175878285e-05, |
| "loss": 0.243, |
| "step": 2560 |
| }, |
| { |
| "epoch": 16.234920634920634, |
| "grad_norm": 1.7678033113479614, |
| "learning_rate": 2.5472433617237107e-05, |
| "loss": 0.2304, |
| "step": 2565 |
| }, |
| { |
| "epoch": 16.266666666666666, |
| "grad_norm": 1.45453679561615, |
| "learning_rate": 2.5379807571456727e-05, |
| "loss": 0.231, |
| "step": 2570 |
| }, |
| { |
| "epoch": 16.298412698412697, |
| "grad_norm": 1.313704490661621, |
| "learning_rate": 2.52871763104154e-05, |
| "loss": 0.2434, |
| "step": 2575 |
| }, |
| { |
| "epoch": 16.33015873015873, |
| "grad_norm": 1.331613302230835, |
| "learning_rate": 2.519454110606304e-05, |
| "loss": 0.2477, |
| "step": 2580 |
| }, |
| { |
| "epoch": 16.36190476190476, |
| "grad_norm": 1.5135482549667358, |
| "learning_rate": 2.510190323040366e-05, |
| "loss": 0.2302, |
| "step": 2585 |
| }, |
| { |
| "epoch": 16.393650793650792, |
| "grad_norm": 1.025403618812561, |
| "learning_rate": 2.5009263955477986e-05, |
| "loss": 0.2145, |
| "step": 2590 |
| }, |
| { |
| "epoch": 16.425396825396824, |
| "grad_norm": 1.3128396272659302, |
| "learning_rate": 2.491662455334592e-05, |
| "loss": 0.237, |
| "step": 2595 |
| }, |
| { |
| "epoch": 16.457142857142856, |
| "grad_norm": 1.9846177101135254, |
| "learning_rate": 2.482398629606916e-05, |
| "loss": 0.2227, |
| "step": 2600 |
| }, |
| { |
| "epoch": 16.488888888888887, |
| "grad_norm": 1.3919912576675415, |
| "learning_rate": 2.473135045569365e-05, |
| "loss": 0.2542, |
| "step": 2605 |
| }, |
| { |
| "epoch": 16.520634920634922, |
| "grad_norm": 1.4371297359466553, |
| "learning_rate": 2.463871830423215e-05, |
| "loss": 0.2533, |
| "step": 2610 |
| }, |
| { |
| "epoch": 16.552380952380954, |
| "grad_norm": 1.2560791969299316, |
| "learning_rate": 2.4546091113646774e-05, |
| "loss": 0.2425, |
| "step": 2615 |
| }, |
| { |
| "epoch": 16.584126984126986, |
| "grad_norm": 1.1686952114105225, |
| "learning_rate": 2.445347015583151e-05, |
| "loss": 0.2196, |
| "step": 2620 |
| }, |
| { |
| "epoch": 16.615873015873017, |
| "grad_norm": 1.4156274795532227, |
| "learning_rate": 2.436085670259479e-05, |
| "loss": 0.2514, |
| "step": 2625 |
| }, |
| { |
| "epoch": 16.64761904761905, |
| "grad_norm": 1.7230234146118164, |
| "learning_rate": 2.426825202564194e-05, |
| "loss": 0.24, |
| "step": 2630 |
| }, |
| { |
| "epoch": 16.67936507936508, |
| "grad_norm": 1.6709225177764893, |
| "learning_rate": 2.417565739655783e-05, |
| "loss": 0.2523, |
| "step": 2635 |
| }, |
| { |
| "epoch": 16.711111111111112, |
| "grad_norm": 1.4194560050964355, |
| "learning_rate": 2.4083074086789332e-05, |
| "loss": 0.2464, |
| "step": 2640 |
| }, |
| { |
| "epoch": 16.742857142857144, |
| "grad_norm": 1.3121566772460938, |
| "learning_rate": 2.3990503367627896e-05, |
| "loss": 0.234, |
| "step": 2645 |
| }, |
| { |
| "epoch": 16.774603174603175, |
| "grad_norm": 1.841577410697937, |
| "learning_rate": 2.3897946510192104e-05, |
| "loss": 0.2494, |
| "step": 2650 |
| }, |
| { |
| "epoch": 16.806349206349207, |
| "grad_norm": 1.262974739074707, |
| "learning_rate": 2.3805404785410157e-05, |
| "loss": 0.2648, |
| "step": 2655 |
| }, |
| { |
| "epoch": 16.83809523809524, |
| "grad_norm": 2.1865732669830322, |
| "learning_rate": 2.3712879464002504e-05, |
| "loss": 0.2486, |
| "step": 2660 |
| }, |
| { |
| "epoch": 16.86984126984127, |
| "grad_norm": 1.3114222288131714, |
| "learning_rate": 2.362037181646433e-05, |
| "loss": 0.2617, |
| "step": 2665 |
| }, |
| { |
| "epoch": 16.901587301587302, |
| "grad_norm": 1.6503785848617554, |
| "learning_rate": 2.3527883113048154e-05, |
| "loss": 0.2588, |
| "step": 2670 |
| }, |
| { |
| "epoch": 16.933333333333334, |
| "grad_norm": 1.4067734479904175, |
| "learning_rate": 2.343541462374634e-05, |
| "loss": 0.2449, |
| "step": 2675 |
| }, |
| { |
| "epoch": 16.965079365079365, |
| "grad_norm": 1.4102448225021362, |
| "learning_rate": 2.3342967618273712e-05, |
| "loss": 0.2418, |
| "step": 2680 |
| }, |
| { |
| "epoch": 16.996825396825397, |
| "grad_norm": 1.4610909223556519, |
| "learning_rate": 2.3250543366050074e-05, |
| "loss": 0.2445, |
| "step": 2685 |
| }, |
| { |
| "epoch": 17.025396825396825, |
| "grad_norm": 1.2868993282318115, |
| "learning_rate": 2.3158143136182812e-05, |
| "loss": 0.231, |
| "step": 2690 |
| }, |
| { |
| "epoch": 17.057142857142857, |
| "grad_norm": 1.2340950965881348, |
| "learning_rate": 2.3065768197449456e-05, |
| "loss": 0.226, |
| "step": 2695 |
| }, |
| { |
| "epoch": 17.08888888888889, |
| "grad_norm": 1.2958239316940308, |
| "learning_rate": 2.2973419818280225e-05, |
| "loss": 0.2279, |
| "step": 2700 |
| }, |
| { |
| "epoch": 17.12063492063492, |
| "grad_norm": 1.271338701248169, |
| "learning_rate": 2.288109926674068e-05, |
| "loss": 0.2216, |
| "step": 2705 |
| }, |
| { |
| "epoch": 17.152380952380952, |
| "grad_norm": 1.2706894874572754, |
| "learning_rate": 2.2788807810514237e-05, |
| "loss": 0.2261, |
| "step": 2710 |
| }, |
| { |
| "epoch": 17.184126984126983, |
| "grad_norm": 1.331217646598816, |
| "learning_rate": 2.2696546716884835e-05, |
| "loss": 0.2149, |
| "step": 2715 |
| }, |
| { |
| "epoch": 17.215873015873015, |
| "grad_norm": 1.74607515335083, |
| "learning_rate": 2.260431725271944e-05, |
| "loss": 0.245, |
| "step": 2720 |
| }, |
| { |
| "epoch": 17.247619047619047, |
| "grad_norm": 1.0970765352249146, |
| "learning_rate": 2.2512120684450745e-05, |
| "loss": 0.2187, |
| "step": 2725 |
| }, |
| { |
| "epoch": 17.27936507936508, |
| "grad_norm": 1.699119210243225, |
| "learning_rate": 2.241995827805974e-05, |
| "loss": 0.2111, |
| "step": 2730 |
| }, |
| { |
| "epoch": 17.31111111111111, |
| "grad_norm": 2.3342185020446777, |
| "learning_rate": 2.23278312990583e-05, |
| "loss": 0.2292, |
| "step": 2735 |
| }, |
| { |
| "epoch": 17.34285714285714, |
| "grad_norm": 1.6536673307418823, |
| "learning_rate": 2.2235741012471866e-05, |
| "loss": 0.23, |
| "step": 2740 |
| }, |
| { |
| "epoch": 17.374603174603173, |
| "grad_norm": 2.414911985397339, |
| "learning_rate": 2.2143688682822e-05, |
| "loss": 0.2433, |
| "step": 2745 |
| }, |
| { |
| "epoch": 17.406349206349205, |
| "grad_norm": 1.3330169916152954, |
| "learning_rate": 2.205167557410912e-05, |
| "loss": 0.224, |
| "step": 2750 |
| }, |
| { |
| "epoch": 17.438095238095237, |
| "grad_norm": 1.5514572858810425, |
| "learning_rate": 2.195970294979504e-05, |
| "loss": 0.2501, |
| "step": 2755 |
| }, |
| { |
| "epoch": 17.469841269841268, |
| "grad_norm": 1.316927433013916, |
| "learning_rate": 2.1867772072785708e-05, |
| "loss": 0.2165, |
| "step": 2760 |
| }, |
| { |
| "epoch": 17.5015873015873, |
| "grad_norm": 1.425007700920105, |
| "learning_rate": 2.1775884205413793e-05, |
| "loss": 0.253, |
| "step": 2765 |
| }, |
| { |
| "epoch": 17.533333333333335, |
| "grad_norm": 2.77079701423645, |
| "learning_rate": 2.168404060942141e-05, |
| "loss": 0.2637, |
| "step": 2770 |
| }, |
| { |
| "epoch": 17.565079365079367, |
| "grad_norm": 1.365783452987671, |
| "learning_rate": 2.1592242545942755e-05, |
| "loss": 0.2303, |
| "step": 2775 |
| }, |
| { |
| "epoch": 17.5968253968254, |
| "grad_norm": 1.606326699256897, |
| "learning_rate": 2.150049127548681e-05, |
| "loss": 0.2419, |
| "step": 2780 |
| }, |
| { |
| "epoch": 17.62857142857143, |
| "grad_norm": 1.3090287446975708, |
| "learning_rate": 2.1408788057920033e-05, |
| "loss": 0.2346, |
| "step": 2785 |
| }, |
| { |
| "epoch": 17.66031746031746, |
| "grad_norm": 1.6885017156600952, |
| "learning_rate": 2.131713415244902e-05, |
| "loss": 0.2238, |
| "step": 2790 |
| }, |
| { |
| "epoch": 17.692063492063493, |
| "grad_norm": 1.3142653703689575, |
| "learning_rate": 2.122553081760329e-05, |
| "loss": 0.2423, |
| "step": 2795 |
| }, |
| { |
| "epoch": 17.723809523809525, |
| "grad_norm": 1.4189752340316772, |
| "learning_rate": 2.1133979311217923e-05, |
| "loss": 0.2411, |
| "step": 2800 |
| }, |
| { |
| "epoch": 17.755555555555556, |
| "grad_norm": 1.598166584968567, |
| "learning_rate": 2.1042480890416368e-05, |
| "loss": 0.2538, |
| "step": 2805 |
| }, |
| { |
| "epoch": 17.787301587301588, |
| "grad_norm": 1.8780243396759033, |
| "learning_rate": 2.0951036811593087e-05, |
| "loss": 0.2284, |
| "step": 2810 |
| }, |
| { |
| "epoch": 17.81904761904762, |
| "grad_norm": 1.5214310884475708, |
| "learning_rate": 2.0859648330396386e-05, |
| "loss": 0.236, |
| "step": 2815 |
| }, |
| { |
| "epoch": 17.85079365079365, |
| "grad_norm": 1.4409775733947754, |
| "learning_rate": 2.0768316701711153e-05, |
| "loss": 0.2438, |
| "step": 2820 |
| }, |
| { |
| "epoch": 17.882539682539683, |
| "grad_norm": 1.4400851726531982, |
| "learning_rate": 2.067704317964158e-05, |
| "loss": 0.2594, |
| "step": 2825 |
| }, |
| { |
| "epoch": 17.914285714285715, |
| "grad_norm": 1.6254900693893433, |
| "learning_rate": 2.0585829017494017e-05, |
| "loss": 0.2506, |
| "step": 2830 |
| }, |
| { |
| "epoch": 17.946031746031746, |
| "grad_norm": 1.6710137128829956, |
| "learning_rate": 2.049467546775968e-05, |
| "loss": 0.2627, |
| "step": 2835 |
| }, |
| { |
| "epoch": 17.977777777777778, |
| "grad_norm": 1.5946308374404907, |
| "learning_rate": 2.0403583782097547e-05, |
| "loss": 0.246, |
| "step": 2840 |
| }, |
| { |
| "epoch": 18.006349206349206, |
| "grad_norm": 1.2540661096572876, |
| "learning_rate": 2.0312555211317074e-05, |
| "loss": 0.2539, |
| "step": 2845 |
| }, |
| { |
| "epoch": 18.038095238095238, |
| "grad_norm": 1.0493476390838623, |
| "learning_rate": 2.0221591005361104e-05, |
| "loss": 0.226, |
| "step": 2850 |
| }, |
| { |
| "epoch": 18.06984126984127, |
| "grad_norm": 1.243975043296814, |
| "learning_rate": 2.0130692413288627e-05, |
| "loss": 0.2185, |
| "step": 2855 |
| }, |
| { |
| "epoch": 18.1015873015873, |
| "grad_norm": 1.2368860244750977, |
| "learning_rate": 2.00398606832577e-05, |
| "loss": 0.2237, |
| "step": 2860 |
| }, |
| { |
| "epoch": 18.133333333333333, |
| "grad_norm": 1.3286014795303345, |
| "learning_rate": 1.9949097062508267e-05, |
| "loss": 0.2216, |
| "step": 2865 |
| }, |
| { |
| "epoch": 18.165079365079364, |
| "grad_norm": 1.6492300033569336, |
| "learning_rate": 1.985840279734503e-05, |
| "loss": 0.2363, |
| "step": 2870 |
| }, |
| { |
| "epoch": 18.196825396825396, |
| "grad_norm": 1.306085228919983, |
| "learning_rate": 1.976777913312037e-05, |
| "loss": 0.2175, |
| "step": 2875 |
| }, |
| { |
| "epoch": 18.228571428571428, |
| "grad_norm": 1.6214182376861572, |
| "learning_rate": 1.9677227314217188e-05, |
| "loss": 0.2118, |
| "step": 2880 |
| }, |
| { |
| "epoch": 18.26031746031746, |
| "grad_norm": 1.2816709280014038, |
| "learning_rate": 1.9586748584031895e-05, |
| "loss": 0.2067, |
| "step": 2885 |
| }, |
| { |
| "epoch": 18.29206349206349, |
| "grad_norm": 2.00748610496521, |
| "learning_rate": 1.949634418495726e-05, |
| "loss": 0.2097, |
| "step": 2890 |
| }, |
| { |
| "epoch": 18.323809523809523, |
| "grad_norm": 1.5324851274490356, |
| "learning_rate": 1.940601535836542e-05, |
| "loss": 0.2364, |
| "step": 2895 |
| }, |
| { |
| "epoch": 18.355555555555554, |
| "grad_norm": 1.6603834629058838, |
| "learning_rate": 1.9315763344590765e-05, |
| "loss": 0.2382, |
| "step": 2900 |
| }, |
| { |
| "epoch": 18.387301587301586, |
| "grad_norm": 1.5497424602508545, |
| "learning_rate": 1.9225589382912978e-05, |
| "loss": 0.2385, |
| "step": 2905 |
| }, |
| { |
| "epoch": 18.419047619047618, |
| "grad_norm": 1.3442565202713013, |
| "learning_rate": 1.9135494711539975e-05, |
| "loss": 0.2324, |
| "step": 2910 |
| }, |
| { |
| "epoch": 18.45079365079365, |
| "grad_norm": 1.5078067779541016, |
| "learning_rate": 1.9045480567590894e-05, |
| "loss": 0.229, |
| "step": 2915 |
| }, |
| { |
| "epoch": 18.48253968253968, |
| "grad_norm": 1.402087926864624, |
| "learning_rate": 1.895554818707915e-05, |
| "loss": 0.2173, |
| "step": 2920 |
| }, |
| { |
| "epoch": 18.514285714285712, |
| "grad_norm": 1.2340173721313477, |
| "learning_rate": 1.88656988048954e-05, |
| "loss": 0.2435, |
| "step": 2925 |
| }, |
| { |
| "epoch": 18.546031746031748, |
| "grad_norm": 1.2508577108383179, |
| "learning_rate": 1.8775933654790673e-05, |
| "loss": 0.2316, |
| "step": 2930 |
| }, |
| { |
| "epoch": 18.57777777777778, |
| "grad_norm": 1.0763332843780518, |
| "learning_rate": 1.8686253969359334e-05, |
| "loss": 0.2208, |
| "step": 2935 |
| }, |
| { |
| "epoch": 18.60952380952381, |
| "grad_norm": 1.2881734371185303, |
| "learning_rate": 1.8596660980022258e-05, |
| "loss": 0.2163, |
| "step": 2940 |
| }, |
| { |
| "epoch": 18.641269841269843, |
| "grad_norm": 1.675923466682434, |
| "learning_rate": 1.8507155917009793e-05, |
| "loss": 0.2536, |
| "step": 2945 |
| }, |
| { |
| "epoch": 18.673015873015874, |
| "grad_norm": 1.1991617679595947, |
| "learning_rate": 1.8417740009345017e-05, |
| "loss": 0.2406, |
| "step": 2950 |
| }, |
| { |
| "epoch": 18.704761904761906, |
| "grad_norm": 1.3702925443649292, |
| "learning_rate": 1.8328414484826745e-05, |
| "loss": 0.2415, |
| "step": 2955 |
| }, |
| { |
| "epoch": 18.736507936507937, |
| "grad_norm": 1.0503740310668945, |
| "learning_rate": 1.8239180570012718e-05, |
| "loss": 0.222, |
| "step": 2960 |
| }, |
| { |
| "epoch": 18.76825396825397, |
| "grad_norm": 1.5632826089859009, |
| "learning_rate": 1.8150039490202772e-05, |
| "loss": 0.2494, |
| "step": 2965 |
| }, |
| { |
| "epoch": 18.8, |
| "grad_norm": 1.1418730020523071, |
| "learning_rate": 1.806099246942196e-05, |
| "loss": 0.2481, |
| "step": 2970 |
| }, |
| { |
| "epoch": 18.831746031746032, |
| "grad_norm": 1.4492175579071045, |
| "learning_rate": 1.7972040730403818e-05, |
| "loss": 0.2314, |
| "step": 2975 |
| }, |
| { |
| "epoch": 18.863492063492064, |
| "grad_norm": 1.4045511484146118, |
| "learning_rate": 1.7883185494573513e-05, |
| "loss": 0.2473, |
| "step": 2980 |
| }, |
| { |
| "epoch": 18.895238095238096, |
| "grad_norm": 1.394841194152832, |
| "learning_rate": 1.7794427982031104e-05, |
| "loss": 0.2307, |
| "step": 2985 |
| }, |
| { |
| "epoch": 18.926984126984127, |
| "grad_norm": 1.3118870258331299, |
| "learning_rate": 1.7705769411534794e-05, |
| "loss": 0.2373, |
| "step": 2990 |
| }, |
| { |
| "epoch": 18.95873015873016, |
| "grad_norm": 1.7034424543380737, |
| "learning_rate": 1.7617211000484142e-05, |
| "loss": 0.2396, |
| "step": 2995 |
| }, |
| { |
| "epoch": 18.99047619047619, |
| "grad_norm": 1.27837336063385, |
| "learning_rate": 1.7528753964903422e-05, |
| "loss": 0.2447, |
| "step": 3000 |
| }, |
| { |
| "epoch": 19.01904761904762, |
| "grad_norm": 0.9308731555938721, |
| "learning_rate": 1.7440399519424866e-05, |
| "loss": 0.2329, |
| "step": 3005 |
| }, |
| { |
| "epoch": 19.05079365079365, |
| "grad_norm": 1.1904468536376953, |
| "learning_rate": 1.735214887727204e-05, |
| "loss": 0.2075, |
| "step": 3010 |
| }, |
| { |
| "epoch": 19.082539682539682, |
| "grad_norm": 1.3396952152252197, |
| "learning_rate": 1.7264003250243102e-05, |
| "loss": 0.2144, |
| "step": 3015 |
| }, |
| { |
| "epoch": 19.114285714285714, |
| "grad_norm": 0.9192984700202942, |
| "learning_rate": 1.717596384869426e-05, |
| "loss": 0.2147, |
| "step": 3020 |
| }, |
| { |
| "epoch": 19.146031746031746, |
| "grad_norm": 1.3901828527450562, |
| "learning_rate": 1.7088031881523076e-05, |
| "loss": 0.2055, |
| "step": 3025 |
| }, |
| { |
| "epoch": 19.177777777777777, |
| "grad_norm": 1.410027265548706, |
| "learning_rate": 1.7000208556151915e-05, |
| "loss": 0.2185, |
| "step": 3030 |
| }, |
| { |
| "epoch": 19.20952380952381, |
| "grad_norm": 1.863762378692627, |
| "learning_rate": 1.6912495078511342e-05, |
| "loss": 0.2365, |
| "step": 3035 |
| }, |
| { |
| "epoch": 19.24126984126984, |
| "grad_norm": 1.281455636024475, |
| "learning_rate": 1.682489265302355e-05, |
| "loss": 0.2189, |
| "step": 3040 |
| }, |
| { |
| "epoch": 19.273015873015872, |
| "grad_norm": 1.5454927682876587, |
| "learning_rate": 1.6737402482585863e-05, |
| "loss": 0.2234, |
| "step": 3045 |
| }, |
| { |
| "epoch": 19.304761904761904, |
| "grad_norm": 1.396296501159668, |
| "learning_rate": 1.6650025768554172e-05, |
| "loss": 0.2196, |
| "step": 3050 |
| }, |
| { |
| "epoch": 19.336507936507935, |
| "grad_norm": 1.1868641376495361, |
| "learning_rate": 1.6562763710726493e-05, |
| "loss": 0.215, |
| "step": 3055 |
| }, |
| { |
| "epoch": 19.368253968253967, |
| "grad_norm": 1.503721833229065, |
| "learning_rate": 1.6475617507326418e-05, |
| "loss": 0.2164, |
| "step": 3060 |
| }, |
| { |
| "epoch": 19.4, |
| "grad_norm": 1.7851678133010864, |
| "learning_rate": 1.638858835498674e-05, |
| "loss": 0.2489, |
| "step": 3065 |
| }, |
| { |
| "epoch": 19.43174603174603, |
| "grad_norm": 1.8271689414978027, |
| "learning_rate": 1.6301677448732967e-05, |
| "loss": 0.2379, |
| "step": 3070 |
| }, |
| { |
| "epoch": 19.463492063492062, |
| "grad_norm": 1.2250396013259888, |
| "learning_rate": 1.6214885981966937e-05, |
| "loss": 0.2355, |
| "step": 3075 |
| }, |
| { |
| "epoch": 19.495238095238093, |
| "grad_norm": 1.5733988285064697, |
| "learning_rate": 1.6128215146450438e-05, |
| "loss": 0.2168, |
| "step": 3080 |
| }, |
| { |
| "epoch": 19.52698412698413, |
| "grad_norm": 1.329305648803711, |
| "learning_rate": 1.60416661322888e-05, |
| "loss": 0.2241, |
| "step": 3085 |
| }, |
| { |
| "epoch": 19.55873015873016, |
| "grad_norm": 1.8745321035385132, |
| "learning_rate": 1.5955240127914618e-05, |
| "loss": 0.2346, |
| "step": 3090 |
| }, |
| { |
| "epoch": 19.590476190476192, |
| "grad_norm": 1.7080330848693848, |
| "learning_rate": 1.5868938320071376e-05, |
| "loss": 0.2414, |
| "step": 3095 |
| }, |
| { |
| "epoch": 19.622222222222224, |
| "grad_norm": 1.6266529560089111, |
| "learning_rate": 1.5782761893797197e-05, |
| "loss": 0.2379, |
| "step": 3100 |
| }, |
| { |
| "epoch": 19.653968253968255, |
| "grad_norm": 1.3323780298233032, |
| "learning_rate": 1.569671203240852e-05, |
| "loss": 0.2557, |
| "step": 3105 |
| }, |
| { |
| "epoch": 19.685714285714287, |
| "grad_norm": 1.2912445068359375, |
| "learning_rate": 1.561078991748391e-05, |
| "loss": 0.2481, |
| "step": 3110 |
| }, |
| { |
| "epoch": 19.71746031746032, |
| "grad_norm": 1.3497467041015625, |
| "learning_rate": 1.552499672884778e-05, |
| "loss": 0.2264, |
| "step": 3115 |
| }, |
| { |
| "epoch": 19.74920634920635, |
| "grad_norm": 1.301652431488037, |
| "learning_rate": 1.5439333644554227e-05, |
| "loss": 0.2155, |
| "step": 3120 |
| }, |
| { |
| "epoch": 19.78095238095238, |
| "grad_norm": 1.579917311668396, |
| "learning_rate": 1.5353801840870862e-05, |
| "loss": 0.2211, |
| "step": 3125 |
| }, |
| { |
| "epoch": 19.812698412698413, |
| "grad_norm": 1.1716071367263794, |
| "learning_rate": 1.5268402492262586e-05, |
| "loss": 0.2202, |
| "step": 3130 |
| }, |
| { |
| "epoch": 19.844444444444445, |
| "grad_norm": 1.1521406173706055, |
| "learning_rate": 1.5183136771375579e-05, |
| "loss": 0.2097, |
| "step": 3135 |
| }, |
| { |
| "epoch": 19.876190476190477, |
| "grad_norm": 2.2419068813323975, |
| "learning_rate": 1.509800584902108e-05, |
| "loss": 0.2342, |
| "step": 3140 |
| }, |
| { |
| "epoch": 19.90793650793651, |
| "grad_norm": 1.8932515382766724, |
| "learning_rate": 1.5013010894159415e-05, |
| "loss": 0.2294, |
| "step": 3145 |
| }, |
| { |
| "epoch": 19.93968253968254, |
| "grad_norm": 1.2340586185455322, |
| "learning_rate": 1.4928153073883843e-05, |
| "loss": 0.2099, |
| "step": 3150 |
| }, |
| { |
| "epoch": 19.97142857142857, |
| "grad_norm": 1.4277369976043701, |
| "learning_rate": 1.4843433553404649e-05, |
| "loss": 0.2561, |
| "step": 3155 |
| }, |
| { |
| "epoch": 20.0, |
| "grad_norm": 1.2759459018707275, |
| "learning_rate": 1.4758853496033009e-05, |
| "loss": 0.2302, |
| "step": 3160 |
| }, |
| { |
| "epoch": 20.03174603174603, |
| "grad_norm": 1.4725241661071777, |
| "learning_rate": 1.4674414063165137e-05, |
| "loss": 0.195, |
| "step": 3165 |
| }, |
| { |
| "epoch": 20.063492063492063, |
| "grad_norm": 1.3785438537597656, |
| "learning_rate": 1.4590116414266276e-05, |
| "loss": 0.2214, |
| "step": 3170 |
| }, |
| { |
| "epoch": 20.095238095238095, |
| "grad_norm": 1.3412481546401978, |
| "learning_rate": 1.4505961706854754e-05, |
| "loss": 0.2176, |
| "step": 3175 |
| }, |
| { |
| "epoch": 20.126984126984127, |
| "grad_norm": 1.5986071825027466, |
| "learning_rate": 1.4421951096486171e-05, |
| "loss": 0.2151, |
| "step": 3180 |
| }, |
| { |
| "epoch": 20.158730158730158, |
| "grad_norm": 1.5248156785964966, |
| "learning_rate": 1.4338085736737462e-05, |
| "loss": 0.2219, |
| "step": 3185 |
| }, |
| { |
| "epoch": 20.19047619047619, |
| "grad_norm": 1.3911489248275757, |
| "learning_rate": 1.4254366779191067e-05, |
| "loss": 0.2111, |
| "step": 3190 |
| }, |
| { |
| "epoch": 20.22222222222222, |
| "grad_norm": 1.5443161725997925, |
| "learning_rate": 1.4170795373419148e-05, |
| "loss": 0.2127, |
| "step": 3195 |
| }, |
| { |
| "epoch": 20.253968253968253, |
| "grad_norm": 1.4181116819381714, |
| "learning_rate": 1.4087372666967768e-05, |
| "loss": 0.2053, |
| "step": 3200 |
| }, |
| { |
| "epoch": 20.285714285714285, |
| "grad_norm": 1.094240665435791, |
| "learning_rate": 1.4004099805341187e-05, |
| "loss": 0.2088, |
| "step": 3205 |
| }, |
| { |
| "epoch": 20.317460317460316, |
| "grad_norm": 1.1608216762542725, |
| "learning_rate": 1.392097793198605e-05, |
| "loss": 0.1988, |
| "step": 3210 |
| }, |
| { |
| "epoch": 20.349206349206348, |
| "grad_norm": 1.3101439476013184, |
| "learning_rate": 1.383800818827577e-05, |
| "loss": 0.2301, |
| "step": 3215 |
| }, |
| { |
| "epoch": 20.38095238095238, |
| "grad_norm": 1.244240403175354, |
| "learning_rate": 1.3755191713494797e-05, |
| "loss": 0.2194, |
| "step": 3220 |
| }, |
| { |
| "epoch": 20.41269841269841, |
| "grad_norm": 2.0739810466766357, |
| "learning_rate": 1.3672529644823004e-05, |
| "loss": 0.2298, |
| "step": 3225 |
| }, |
| { |
| "epoch": 20.444444444444443, |
| "grad_norm": 1.6282612085342407, |
| "learning_rate": 1.359002311732005e-05, |
| "loss": 0.2387, |
| "step": 3230 |
| }, |
| { |
| "epoch": 20.476190476190474, |
| "grad_norm": 1.4091988801956177, |
| "learning_rate": 1.3507673263909856e-05, |
| "loss": 0.202, |
| "step": 3235 |
| }, |
| { |
| "epoch": 20.507936507936506, |
| "grad_norm": 1.544467806816101, |
| "learning_rate": 1.3425481215364922e-05, |
| "loss": 0.2422, |
| "step": 3240 |
| }, |
| { |
| "epoch": 20.53968253968254, |
| "grad_norm": 1.7039313316345215, |
| "learning_rate": 1.334344810029095e-05, |
| "loss": 0.1986, |
| "step": 3245 |
| }, |
| { |
| "epoch": 20.571428571428573, |
| "grad_norm": 1.4984982013702393, |
| "learning_rate": 1.3261575045111235e-05, |
| "loss": 0.2303, |
| "step": 3250 |
| }, |
| { |
| "epoch": 20.603174603174605, |
| "grad_norm": 1.1875499486923218, |
| "learning_rate": 1.3179863174051238e-05, |
| "loss": 0.2273, |
| "step": 3255 |
| }, |
| { |
| "epoch": 20.634920634920636, |
| "grad_norm": 1.856332778930664, |
| "learning_rate": 1.3098313609123192e-05, |
| "loss": 0.2323, |
| "step": 3260 |
| }, |
| { |
| "epoch": 20.666666666666668, |
| "grad_norm": 1.6271690130233765, |
| "learning_rate": 1.301692747011058e-05, |
| "loss": 0.2387, |
| "step": 3265 |
| }, |
| { |
| "epoch": 20.6984126984127, |
| "grad_norm": 1.8108351230621338, |
| "learning_rate": 1.2935705874552894e-05, |
| "loss": 0.2224, |
| "step": 3270 |
| }, |
| { |
| "epoch": 20.73015873015873, |
| "grad_norm": 1.3620928525924683, |
| "learning_rate": 1.2854649937730196e-05, |
| "loss": 0.2237, |
| "step": 3275 |
| }, |
| { |
| "epoch": 20.761904761904763, |
| "grad_norm": 1.5440479516983032, |
| "learning_rate": 1.2773760772647839e-05, |
| "loss": 0.2319, |
| "step": 3280 |
| }, |
| { |
| "epoch": 20.793650793650794, |
| "grad_norm": 1.2190790176391602, |
| "learning_rate": 1.269303949002118e-05, |
| "loss": 0.2169, |
| "step": 3285 |
| }, |
| { |
| "epoch": 20.825396825396826, |
| "grad_norm": 1.42155122756958, |
| "learning_rate": 1.261248719826032e-05, |
| "loss": 0.2319, |
| "step": 3290 |
| }, |
| { |
| "epoch": 20.857142857142858, |
| "grad_norm": 1.7695163488388062, |
| "learning_rate": 1.2532105003454919e-05, |
| "loss": 0.2421, |
| "step": 3295 |
| }, |
| { |
| "epoch": 20.88888888888889, |
| "grad_norm": 1.3969519138336182, |
| "learning_rate": 1.245189400935895e-05, |
| "loss": 0.225, |
| "step": 3300 |
| }, |
| { |
| "epoch": 20.92063492063492, |
| "grad_norm": 1.1847434043884277, |
| "learning_rate": 1.2371855317375583e-05, |
| "loss": 0.215, |
| "step": 3305 |
| }, |
| { |
| "epoch": 20.952380952380953, |
| "grad_norm": 1.7256789207458496, |
| "learning_rate": 1.2291990026542058e-05, |
| "loss": 0.2446, |
| "step": 3310 |
| }, |
| { |
| "epoch": 20.984126984126984, |
| "grad_norm": 1.1966052055358887, |
| "learning_rate": 1.2212299233514582e-05, |
| "loss": 0.2108, |
| "step": 3315 |
| }, |
| { |
| "epoch": 21.012698412698413, |
| "grad_norm": 1.0815582275390625, |
| "learning_rate": 1.2132784032553263e-05, |
| "loss": 0.2431, |
| "step": 3320 |
| }, |
| { |
| "epoch": 21.044444444444444, |
| "grad_norm": 1.4132349491119385, |
| "learning_rate": 1.2053445515507147e-05, |
| "loss": 0.2186, |
| "step": 3325 |
| }, |
| { |
| "epoch": 21.076190476190476, |
| "grad_norm": 1.0495672225952148, |
| "learning_rate": 1.1974284771799096e-05, |
| "loss": 0.197, |
| "step": 3330 |
| }, |
| { |
| "epoch": 21.107936507936508, |
| "grad_norm": 1.072950005531311, |
| "learning_rate": 1.1895302888410986e-05, |
| "loss": 0.1922, |
| "step": 3335 |
| }, |
| { |
| "epoch": 21.13968253968254, |
| "grad_norm": 1.2509602308273315, |
| "learning_rate": 1.1816500949868659e-05, |
| "loss": 0.2067, |
| "step": 3340 |
| }, |
| { |
| "epoch": 21.17142857142857, |
| "grad_norm": 1.460114598274231, |
| "learning_rate": 1.1737880038227082e-05, |
| "loss": 0.2136, |
| "step": 3345 |
| }, |
| { |
| "epoch": 21.203174603174602, |
| "grad_norm": 1.2986267805099487, |
| "learning_rate": 1.1659441233055513e-05, |
| "loss": 0.2126, |
| "step": 3350 |
| }, |
| { |
| "epoch": 21.234920634920634, |
| "grad_norm": 1.159629464149475, |
| "learning_rate": 1.1581185611422585e-05, |
| "loss": 0.2057, |
| "step": 3355 |
| }, |
| { |
| "epoch": 21.266666666666666, |
| "grad_norm": 1.652693748474121, |
| "learning_rate": 1.1503114247881648e-05, |
| "loss": 0.2304, |
| "step": 3360 |
| }, |
| { |
| "epoch": 21.298412698412697, |
| "grad_norm": 1.2332572937011719, |
| "learning_rate": 1.14252282144559e-05, |
| "loss": 0.2093, |
| "step": 3365 |
| }, |
| { |
| "epoch": 21.33015873015873, |
| "grad_norm": 1.5902972221374512, |
| "learning_rate": 1.1347528580623726e-05, |
| "loss": 0.235, |
| "step": 3370 |
| }, |
| { |
| "epoch": 21.36190476190476, |
| "grad_norm": 1.738487720489502, |
| "learning_rate": 1.1270016413303997e-05, |
| "loss": 0.2146, |
| "step": 3375 |
| }, |
| { |
| "epoch": 21.393650793650792, |
| "grad_norm": 1.8629635572433472, |
| "learning_rate": 1.1192692776841404e-05, |
| "loss": 0.2254, |
| "step": 3380 |
| }, |
| { |
| "epoch": 21.425396825396824, |
| "grad_norm": 1.3286288976669312, |
| "learning_rate": 1.1115558732991898e-05, |
| "loss": 0.2119, |
| "step": 3385 |
| }, |
| { |
| "epoch": 21.457142857142856, |
| "grad_norm": 1.8470025062561035, |
| "learning_rate": 1.103861534090804e-05, |
| "loss": 0.2332, |
| "step": 3390 |
| }, |
| { |
| "epoch": 21.488888888888887, |
| "grad_norm": 1.7377171516418457, |
| "learning_rate": 1.0961863657124496e-05, |
| "loss": 0.2281, |
| "step": 3395 |
| }, |
| { |
| "epoch": 21.520634920634922, |
| "grad_norm": 1.1695598363876343, |
| "learning_rate": 1.0885304735543528e-05, |
| "loss": 0.2147, |
| "step": 3400 |
| }, |
| { |
| "epoch": 21.552380952380954, |
| "grad_norm": 1.0073562860488892, |
| "learning_rate": 1.0808939627420514e-05, |
| "loss": 0.222, |
| "step": 3405 |
| }, |
| { |
| "epoch": 21.584126984126986, |
| "grad_norm": 1.742859125137329, |
| "learning_rate": 1.0732769381349508e-05, |
| "loss": 0.2205, |
| "step": 3410 |
| }, |
| { |
| "epoch": 21.615873015873017, |
| "grad_norm": 1.1256352663040161, |
| "learning_rate": 1.0656795043248885e-05, |
| "loss": 0.2205, |
| "step": 3415 |
| }, |
| { |
| "epoch": 21.64761904761905, |
| "grad_norm": 1.3917202949523926, |
| "learning_rate": 1.0581017656346904e-05, |
| "loss": 0.2242, |
| "step": 3420 |
| }, |
| { |
| "epoch": 21.67936507936508, |
| "grad_norm": 1.3438773155212402, |
| "learning_rate": 1.0505438261167439e-05, |
| "loss": 0.2101, |
| "step": 3425 |
| }, |
| { |
| "epoch": 21.711111111111112, |
| "grad_norm": 1.573519229888916, |
| "learning_rate": 1.0430057895515678e-05, |
| "loss": 0.2339, |
| "step": 3430 |
| }, |
| { |
| "epoch": 21.742857142857144, |
| "grad_norm": 1.392638087272644, |
| "learning_rate": 1.0354877594463852e-05, |
| "loss": 0.2101, |
| "step": 3435 |
| }, |
| { |
| "epoch": 21.774603174603175, |
| "grad_norm": 1.6524523496627808, |
| "learning_rate": 1.027989839033709e-05, |
| "loss": 0.2178, |
| "step": 3440 |
| }, |
| { |
| "epoch": 21.806349206349207, |
| "grad_norm": 1.2898451089859009, |
| "learning_rate": 1.0205121312699117e-05, |
| "loss": 0.2267, |
| "step": 3445 |
| }, |
| { |
| "epoch": 21.83809523809524, |
| "grad_norm": 1.5442014932632446, |
| "learning_rate": 1.0130547388338268e-05, |
| "loss": 0.2191, |
| "step": 3450 |
| }, |
| { |
| "epoch": 21.86984126984127, |
| "grad_norm": 1.7914522886276245, |
| "learning_rate": 1.0056177641253266e-05, |
| "loss": 0.2292, |
| "step": 3455 |
| }, |
| { |
| "epoch": 21.901587301587302, |
| "grad_norm": 1.4891278743743896, |
| "learning_rate": 9.982013092639225e-06, |
| "loss": 0.2378, |
| "step": 3460 |
| }, |
| { |
| "epoch": 21.933333333333334, |
| "grad_norm": 1.6321289539337158, |
| "learning_rate": 9.908054760873633e-06, |
| "loss": 0.235, |
| "step": 3465 |
| }, |
| { |
| "epoch": 21.965079365079365, |
| "grad_norm": 1.5167237520217896, |
| "learning_rate": 9.834303661502294e-06, |
| "loss": 0.2181, |
| "step": 3470 |
| }, |
| { |
| "epoch": 21.996825396825397, |
| "grad_norm": 1.4538085460662842, |
| "learning_rate": 9.7607608072255e-06, |
| "loss": 0.2126, |
| "step": 3475 |
| }, |
| { |
| "epoch": 22.025396825396825, |
| "grad_norm": 1.2877874374389648, |
| "learning_rate": 9.687427207884017e-06, |
| "loss": 0.213, |
| "step": 3480 |
| }, |
| { |
| "epoch": 22.057142857142857, |
| "grad_norm": 1.4707602262496948, |
| "learning_rate": 9.614303870445287e-06, |
| "loss": 0.2183, |
| "step": 3485 |
| }, |
| { |
| "epoch": 22.08888888888889, |
| "grad_norm": 0.9957223534584045, |
| "learning_rate": 9.541391798989562e-06, |
| "loss": 0.188, |
| "step": 3490 |
| }, |
| { |
| "epoch": 22.12063492063492, |
| "grad_norm": 1.6289327144622803, |
| "learning_rate": 9.468691994696147e-06, |
| "loss": 0.2183, |
| "step": 3495 |
| }, |
| { |
| "epoch": 22.152380952380952, |
| "grad_norm": 1.2597798109054565, |
| "learning_rate": 9.396205455829615e-06, |
| "loss": 0.2266, |
| "step": 3500 |
| }, |
| { |
| "epoch": 22.184126984126983, |
| "grad_norm": 1.234129548072815, |
| "learning_rate": 9.323933177726163e-06, |
| "loss": 0.2117, |
| "step": 3505 |
| }, |
| { |
| "epoch": 22.215873015873015, |
| "grad_norm": 1.3455181121826172, |
| "learning_rate": 9.251876152779863e-06, |
| "loss": 0.2159, |
| "step": 3510 |
| }, |
| { |
| "epoch": 22.247619047619047, |
| "grad_norm": 1.6391496658325195, |
| "learning_rate": 9.180035370429096e-06, |
| "loss": 0.2142, |
| "step": 3515 |
| }, |
| { |
| "epoch": 22.27936507936508, |
| "grad_norm": 1.925265908241272, |
| "learning_rate": 9.108411817142943e-06, |
| "loss": 0.2173, |
| "step": 3520 |
| }, |
| { |
| "epoch": 22.31111111111111, |
| "grad_norm": 1.912740707397461, |
| "learning_rate": 9.037006476407628e-06, |
| "loss": 0.2129, |
| "step": 3525 |
| }, |
| { |
| "epoch": 22.34285714285714, |
| "grad_norm": 1.6799534559249878, |
| "learning_rate": 8.965820328713071e-06, |
| "loss": 0.2207, |
| "step": 3530 |
| }, |
| { |
| "epoch": 22.374603174603173, |
| "grad_norm": 1.5147929191589355, |
| "learning_rate": 8.894854351539311e-06, |
| "loss": 0.2179, |
| "step": 3535 |
| }, |
| { |
| "epoch": 22.406349206349205, |
| "grad_norm": 1.78584623336792, |
| "learning_rate": 8.824109519343227e-06, |
| "loss": 0.2105, |
| "step": 3540 |
| }, |
| { |
| "epoch": 22.438095238095237, |
| "grad_norm": 1.2689015865325928, |
| "learning_rate": 8.753586803545041e-06, |
| "loss": 0.2106, |
| "step": 3545 |
| }, |
| { |
| "epoch": 22.469841269841268, |
| "grad_norm": 1.0263334512710571, |
| "learning_rate": 8.683287172515022e-06, |
| "loss": 0.2134, |
| "step": 3550 |
| }, |
| { |
| "epoch": 22.5015873015873, |
| "grad_norm": 1.3135781288146973, |
| "learning_rate": 8.61321159156023e-06, |
| "loss": 0.1915, |
| "step": 3555 |
| }, |
| { |
| "epoch": 22.533333333333335, |
| "grad_norm": 1.3863434791564941, |
| "learning_rate": 8.543361022911156e-06, |
| "loss": 0.211, |
| "step": 3560 |
| }, |
| { |
| "epoch": 22.565079365079367, |
| "grad_norm": 1.4205353260040283, |
| "learning_rate": 8.473736425708637e-06, |
| "loss": 0.2155, |
| "step": 3565 |
| }, |
| { |
| "epoch": 22.5968253968254, |
| "grad_norm": 1.7269377708435059, |
| "learning_rate": 8.404338755990587e-06, |
| "loss": 0.2239, |
| "step": 3570 |
| }, |
| { |
| "epoch": 22.62857142857143, |
| "grad_norm": 1.5212743282318115, |
| "learning_rate": 8.335168966678913e-06, |
| "loss": 0.2188, |
| "step": 3575 |
| }, |
| { |
| "epoch": 22.66031746031746, |
| "grad_norm": 1.6680513620376587, |
| "learning_rate": 8.266228007566426e-06, |
| "loss": 0.2295, |
| "step": 3580 |
| }, |
| { |
| "epoch": 22.692063492063493, |
| "grad_norm": 1.8336963653564453, |
| "learning_rate": 8.197516825303792e-06, |
| "loss": 0.209, |
| "step": 3585 |
| }, |
| { |
| "epoch": 22.723809523809525, |
| "grad_norm": 1.3112249374389648, |
| "learning_rate": 8.129036363386524e-06, |
| "loss": 0.2091, |
| "step": 3590 |
| }, |
| { |
| "epoch": 22.755555555555556, |
| "grad_norm": 1.7051877975463867, |
| "learning_rate": 8.060787562142072e-06, |
| "loss": 0.2216, |
| "step": 3595 |
| }, |
| { |
| "epoch": 22.787301587301588, |
| "grad_norm": 1.1862343549728394, |
| "learning_rate": 7.992771358716852e-06, |
| "loss": 0.2059, |
| "step": 3600 |
| }, |
| { |
| "epoch": 22.81904761904762, |
| "grad_norm": 1.2786476612091064, |
| "learning_rate": 7.92498868706341e-06, |
| "loss": 0.2009, |
| "step": 3605 |
| }, |
| { |
| "epoch": 22.85079365079365, |
| "grad_norm": 1.6637458801269531, |
| "learning_rate": 7.857440477927588e-06, |
| "loss": 0.2168, |
| "step": 3610 |
| }, |
| { |
| "epoch": 22.882539682539683, |
| "grad_norm": 1.3283144235610962, |
| "learning_rate": 7.790127658835747e-06, |
| "loss": 0.2254, |
| "step": 3615 |
| }, |
| { |
| "epoch": 22.914285714285715, |
| "grad_norm": 1.3773967027664185, |
| "learning_rate": 7.723051154082059e-06, |
| "loss": 0.2135, |
| "step": 3620 |
| }, |
| { |
| "epoch": 22.946031746031746, |
| "grad_norm": 1.679351806640625, |
| "learning_rate": 7.656211884715728e-06, |
| "loss": 0.2267, |
| "step": 3625 |
| }, |
| { |
| "epoch": 22.977777777777778, |
| "grad_norm": 1.2563788890838623, |
| "learning_rate": 7.58961076852846e-06, |
| "loss": 0.245, |
| "step": 3630 |
| }, |
| { |
| "epoch": 23.006349206349206, |
| "grad_norm": 1.3150761127471924, |
| "learning_rate": 7.523248720041778e-06, |
| "loss": 0.2309, |
| "step": 3635 |
| }, |
| { |
| "epoch": 23.038095238095238, |
| "grad_norm": 1.3584800958633423, |
| "learning_rate": 7.457126650494478e-06, |
| "loss": 0.2081, |
| "step": 3640 |
| }, |
| { |
| "epoch": 23.06984126984127, |
| "grad_norm": 1.5431658029556274, |
| "learning_rate": 7.391245467830163e-06, |
| "loss": 0.2191, |
| "step": 3645 |
| }, |
| { |
| "epoch": 23.1015873015873, |
| "grad_norm": 1.5318843126296997, |
| "learning_rate": 7.325606076684688e-06, |
| "loss": 0.2175, |
| "step": 3650 |
| }, |
| { |
| "epoch": 23.133333333333333, |
| "grad_norm": 1.3879364728927612, |
| "learning_rate": 7.260209378373842e-06, |
| "loss": 0.2008, |
| "step": 3655 |
| }, |
| { |
| "epoch": 23.165079365079364, |
| "grad_norm": 1.4865491390228271, |
| "learning_rate": 7.195056270880887e-06, |
| "loss": 0.2231, |
| "step": 3660 |
| }, |
| { |
| "epoch": 23.196825396825396, |
| "grad_norm": 1.495409369468689, |
| "learning_rate": 7.1301476488442705e-06, |
| "loss": 0.2079, |
| "step": 3665 |
| }, |
| { |
| "epoch": 23.228571428571428, |
| "grad_norm": 1.1792328357696533, |
| "learning_rate": 7.065484403545328e-06, |
| "loss": 0.2102, |
| "step": 3670 |
| }, |
| { |
| "epoch": 23.26031746031746, |
| "grad_norm": 1.2667046785354614, |
| "learning_rate": 7.001067422896063e-06, |
| "loss": 0.1998, |
| "step": 3675 |
| }, |
| { |
| "epoch": 23.29206349206349, |
| "grad_norm": 2.0349957942962646, |
| "learning_rate": 6.936897591426916e-06, |
| "loss": 0.2117, |
| "step": 3680 |
| }, |
| { |
| "epoch": 23.323809523809523, |
| "grad_norm": 1.3733776807785034, |
| "learning_rate": 6.872975790274658e-06, |
| "loss": 0.2222, |
| "step": 3685 |
| }, |
| { |
| "epoch": 23.355555555555554, |
| "grad_norm": 1.6063858270645142, |
| "learning_rate": 6.809302897170266e-06, |
| "loss": 0.2026, |
| "step": 3690 |
| }, |
| { |
| "epoch": 23.387301587301586, |
| "grad_norm": 1.2825813293457031, |
| "learning_rate": 6.7458797864268786e-06, |
| "loss": 0.2246, |
| "step": 3695 |
| }, |
| { |
| "epoch": 23.419047619047618, |
| "grad_norm": 1.3881940841674805, |
| "learning_rate": 6.68270732892782e-06, |
| "loss": 0.2086, |
| "step": 3700 |
| }, |
| { |
| "epoch": 23.45079365079365, |
| "grad_norm": 1.478417158126831, |
| "learning_rate": 6.619786392114557e-06, |
| "loss": 0.2055, |
| "step": 3705 |
| }, |
| { |
| "epoch": 23.48253968253968, |
| "grad_norm": 1.8440154790878296, |
| "learning_rate": 6.557117839974902e-06, |
| "loss": 0.2077, |
| "step": 3710 |
| }, |
| { |
| "epoch": 23.514285714285712, |
| "grad_norm": 1.771545648574829, |
| "learning_rate": 6.494702533031064e-06, |
| "loss": 0.2204, |
| "step": 3715 |
| }, |
| { |
| "epoch": 23.546031746031748, |
| "grad_norm": 1.4290744066238403, |
| "learning_rate": 6.432541328327848e-06, |
| "loss": 0.2161, |
| "step": 3720 |
| }, |
| { |
| "epoch": 23.57777777777778, |
| "grad_norm": 1.8629150390625, |
| "learning_rate": 6.370635079420942e-06, |
| "loss": 0.2194, |
| "step": 3725 |
| }, |
| { |
| "epoch": 23.60952380952381, |
| "grad_norm": 1.5219929218292236, |
| "learning_rate": 6.308984636365092e-06, |
| "loss": 0.2091, |
| "step": 3730 |
| }, |
| { |
| "epoch": 23.641269841269843, |
| "grad_norm": 1.8589943647384644, |
| "learning_rate": 6.247590845702553e-06, |
| "loss": 0.2114, |
| "step": 3735 |
| }, |
| { |
| "epoch": 23.673015873015874, |
| "grad_norm": 1.3381997346878052, |
| "learning_rate": 6.186454550451359e-06, |
| "loss": 0.2076, |
| "step": 3740 |
| }, |
| { |
| "epoch": 23.704761904761906, |
| "grad_norm": 1.4127261638641357, |
| "learning_rate": 6.1255765900938126e-06, |
| "loss": 0.2039, |
| "step": 3745 |
| }, |
| { |
| "epoch": 23.736507936507937, |
| "grad_norm": 1.8508237600326538, |
| "learning_rate": 6.064957800564924e-06, |
| "loss": 0.2193, |
| "step": 3750 |
| }, |
| { |
| "epoch": 23.76825396825397, |
| "grad_norm": 1.652459979057312, |
| "learning_rate": 6.004599014240952e-06, |
| "loss": 0.2124, |
| "step": 3755 |
| }, |
| { |
| "epoch": 23.8, |
| "grad_norm": 1.456562876701355, |
| "learning_rate": 5.944501059927962e-06, |
| "loss": 0.2199, |
| "step": 3760 |
| }, |
| { |
| "epoch": 23.831746031746032, |
| "grad_norm": 1.305023193359375, |
| "learning_rate": 5.884664762850467e-06, |
| "loss": 0.1943, |
| "step": 3765 |
| }, |
| { |
| "epoch": 23.863492063492064, |
| "grad_norm": 1.977409839630127, |
| "learning_rate": 5.825090944640055e-06, |
| "loss": 0.2137, |
| "step": 3770 |
| }, |
| { |
| "epoch": 23.895238095238096, |
| "grad_norm": 1.4465928077697754, |
| "learning_rate": 5.7657804233241465e-06, |
| "loss": 0.2099, |
| "step": 3775 |
| }, |
| { |
| "epoch": 23.926984126984127, |
| "grad_norm": 1.4489303827285767, |
| "learning_rate": 5.706734013314746e-06, |
| "loss": 0.2229, |
| "step": 3780 |
| }, |
| { |
| "epoch": 23.95873015873016, |
| "grad_norm": 1.493857979774475, |
| "learning_rate": 5.647952525397246e-06, |
| "loss": 0.2101, |
| "step": 3785 |
| }, |
| { |
| "epoch": 23.99047619047619, |
| "grad_norm": 1.41466224193573, |
| "learning_rate": 5.589436766719342e-06, |
| "loss": 0.2145, |
| "step": 3790 |
| }, |
| { |
| "epoch": 24.01904761904762, |
| "grad_norm": 1.3483541011810303, |
| "learning_rate": 5.531187540779864e-06, |
| "loss": 0.2092, |
| "step": 3795 |
| }, |
| { |
| "epoch": 24.05079365079365, |
| "grad_norm": 1.5213457345962524, |
| "learning_rate": 5.4732056474178475e-06, |
| "loss": 0.2004, |
| "step": 3800 |
| }, |
| { |
| "epoch": 24.082539682539682, |
| "grad_norm": 1.5124719142913818, |
| "learning_rate": 5.415491882801463e-06, |
| "loss": 0.2066, |
| "step": 3805 |
| }, |
| { |
| "epoch": 24.114285714285714, |
| "grad_norm": 1.1013808250427246, |
| "learning_rate": 5.358047039417122e-06, |
| "loss": 0.2047, |
| "step": 3810 |
| }, |
| { |
| "epoch": 24.146031746031746, |
| "grad_norm": 1.0998328924179077, |
| "learning_rate": 5.300871906058619e-06, |
| "loss": 0.1935, |
| "step": 3815 |
| }, |
| { |
| "epoch": 24.177777777777777, |
| "grad_norm": 1.4137228727340698, |
| "learning_rate": 5.243967267816225e-06, |
| "loss": 0.2153, |
| "step": 3820 |
| }, |
| { |
| "epoch": 24.20952380952381, |
| "grad_norm": 1.3818597793579102, |
| "learning_rate": 5.187333906065999e-06, |
| "loss": 0.2121, |
| "step": 3825 |
| }, |
| { |
| "epoch": 24.24126984126984, |
| "grad_norm": 1.4308501482009888, |
| "learning_rate": 5.1309725984589966e-06, |
| "loss": 0.1923, |
| "step": 3830 |
| }, |
| { |
| "epoch": 24.273015873015872, |
| "grad_norm": 1.8435720205307007, |
| "learning_rate": 5.0748841189106024e-06, |
| "loss": 0.2014, |
| "step": 3835 |
| }, |
| { |
| "epoch": 24.304761904761904, |
| "grad_norm": 1.149243950843811, |
| "learning_rate": 5.019069237589921e-06, |
| "loss": 0.1993, |
| "step": 3840 |
| }, |
| { |
| "epoch": 24.336507936507935, |
| "grad_norm": 1.1767566204071045, |
| "learning_rate": 4.963528720909192e-06, |
| "loss": 0.2073, |
| "step": 3845 |
| }, |
| { |
| "epoch": 24.368253968253967, |
| "grad_norm": 1.269392967224121, |
| "learning_rate": 4.908263331513246e-06, |
| "loss": 0.2054, |
| "step": 3850 |
| }, |
| { |
| "epoch": 24.4, |
| "grad_norm": 1.4679259061813354, |
| "learning_rate": 4.853273828269089e-06, |
| "loss": 0.2159, |
| "step": 3855 |
| }, |
| { |
| "epoch": 24.43174603174603, |
| "grad_norm": 1.4625903367996216, |
| "learning_rate": 4.798560966255411e-06, |
| "loss": 0.2054, |
| "step": 3860 |
| }, |
| { |
| "epoch": 24.463492063492062, |
| "grad_norm": 1.7532594203948975, |
| "learning_rate": 4.744125496752264e-06, |
| "loss": 0.2126, |
| "step": 3865 |
| }, |
| { |
| "epoch": 24.495238095238093, |
| "grad_norm": 1.8333017826080322, |
| "learning_rate": 4.6899681672307346e-06, |
| "loss": 0.1991, |
| "step": 3870 |
| }, |
| { |
| "epoch": 24.52698412698413, |
| "grad_norm": 1.4345580339431763, |
| "learning_rate": 4.636089721342668e-06, |
| "loss": 0.206, |
| "step": 3875 |
| }, |
| { |
| "epoch": 24.55873015873016, |
| "grad_norm": 1.5848816633224487, |
| "learning_rate": 4.582490898910499e-06, |
| "loss": 0.2335, |
| "step": 3880 |
| }, |
| { |
| "epoch": 24.590476190476192, |
| "grad_norm": 1.1518175601959229, |
| "learning_rate": 4.529172435917012e-06, |
| "loss": 0.2007, |
| "step": 3885 |
| }, |
| { |
| "epoch": 24.622222222222224, |
| "grad_norm": 1.582337498664856, |
| "learning_rate": 4.476135064495335e-06, |
| "loss": 0.2242, |
| "step": 3890 |
| }, |
| { |
| "epoch": 24.653968253968255, |
| "grad_norm": 1.1616642475128174, |
| "learning_rate": 4.423379512918807e-06, |
| "loss": 0.2091, |
| "step": 3895 |
| }, |
| { |
| "epoch": 24.685714285714287, |
| "grad_norm": 1.5428528785705566, |
| "learning_rate": 4.370906505591007e-06, |
| "loss": 0.2076, |
| "step": 3900 |
| }, |
| { |
| "epoch": 24.71746031746032, |
| "grad_norm": 1.3769774436950684, |
| "learning_rate": 4.318716763035835e-06, |
| "loss": 0.2298, |
| "step": 3905 |
| }, |
| { |
| "epoch": 24.74920634920635, |
| "grad_norm": 1.6571112871170044, |
| "learning_rate": 4.2668110018875465e-06, |
| "loss": 0.2131, |
| "step": 3910 |
| }, |
| { |
| "epoch": 24.78095238095238, |
| "grad_norm": 1.7125977277755737, |
| "learning_rate": 4.215189934881001e-06, |
| "loss": 0.2197, |
| "step": 3915 |
| }, |
| { |
| "epoch": 24.812698412698413, |
| "grad_norm": 1.543136715888977, |
| "learning_rate": 4.163854270841797e-06, |
| "loss": 0.2095, |
| "step": 3920 |
| }, |
| { |
| "epoch": 24.844444444444445, |
| "grad_norm": 1.397762417793274, |
| "learning_rate": 4.112804714676594e-06, |
| "loss": 0.227, |
| "step": 3925 |
| }, |
| { |
| "epoch": 24.876190476190477, |
| "grad_norm": 2.147937059402466, |
| "learning_rate": 4.062041967363395e-06, |
| "loss": 0.2072, |
| "step": 3930 |
| }, |
| { |
| "epoch": 24.90793650793651, |
| "grad_norm": 1.2324985265731812, |
| "learning_rate": 4.011566725941942e-06, |
| "loss": 0.2116, |
| "step": 3935 |
| }, |
| { |
| "epoch": 24.93968253968254, |
| "grad_norm": 1.3976517915725708, |
| "learning_rate": 3.961379683504157e-06, |
| "loss": 0.2276, |
| "step": 3940 |
| }, |
| { |
| "epoch": 24.97142857142857, |
| "grad_norm": 1.4638493061065674, |
| "learning_rate": 3.911481529184588e-06, |
| "loss": 0.2074, |
| "step": 3945 |
| }, |
| { |
| "epoch": 25.0, |
| "grad_norm": 2.1439075469970703, |
| "learning_rate": 3.861872948150974e-06, |
| "loss": 0.1959, |
| "step": 3950 |
| }, |
| { |
| "epoch": 25.03174603174603, |
| "grad_norm": 1.3343623876571655, |
| "learning_rate": 3.812554621594827e-06, |
| "loss": 0.1867, |
| "step": 3955 |
| }, |
| { |
| "epoch": 25.063492063492063, |
| "grad_norm": 1.3594505786895752, |
| "learning_rate": 3.7635272267220858e-06, |
| "loss": 0.2108, |
| "step": 3960 |
| }, |
| { |
| "epoch": 25.095238095238095, |
| "grad_norm": 1.4568570852279663, |
| "learning_rate": 3.7147914367437977e-06, |
| "loss": 0.1973, |
| "step": 3965 |
| }, |
| { |
| "epoch": 25.126984126984127, |
| "grad_norm": 1.5874037742614746, |
| "learning_rate": 3.6663479208669223e-06, |
| "loss": 0.2169, |
| "step": 3970 |
| }, |
| { |
| "epoch": 25.158730158730158, |
| "grad_norm": 1.4729677438735962, |
| "learning_rate": 3.6181973442850597e-06, |
| "loss": 0.2117, |
| "step": 3975 |
| }, |
| { |
| "epoch": 25.19047619047619, |
| "grad_norm": 1.549204707145691, |
| "learning_rate": 3.5703403681694115e-06, |
| "loss": 0.2142, |
| "step": 3980 |
| }, |
| { |
| "epoch": 25.22222222222222, |
| "grad_norm": 1.3857277631759644, |
| "learning_rate": 3.5227776496596283e-06, |
| "loss": 0.2166, |
| "step": 3985 |
| }, |
| { |
| "epoch": 25.253968253968253, |
| "grad_norm": 1.282907247543335, |
| "learning_rate": 3.4755098418548155e-06, |
| "loss": 0.2137, |
| "step": 3990 |
| }, |
| { |
| "epoch": 25.285714285714285, |
| "grad_norm": 1.4106628894805908, |
| "learning_rate": 3.42853759380459e-06, |
| "loss": 0.1962, |
| "step": 3995 |
| }, |
| { |
| "epoch": 25.317460317460316, |
| "grad_norm": 1.1567012071609497, |
| "learning_rate": 3.3818615505000982e-06, |
| "loss": 0.184, |
| "step": 4000 |
| }, |
| { |
| "epoch": 25.349206349206348, |
| "grad_norm": 1.5095430612564087, |
| "learning_rate": 3.3354823528652463e-06, |
| "loss": 0.2084, |
| "step": 4005 |
| }, |
| { |
| "epoch": 25.38095238095238, |
| "grad_norm": 1.6294128894805908, |
| "learning_rate": 3.289400637747833e-06, |
| "loss": 0.2137, |
| "step": 4010 |
| }, |
| { |
| "epoch": 25.41269841269841, |
| "grad_norm": 1.0825536251068115, |
| "learning_rate": 3.243617037910826e-06, |
| "loss": 0.1874, |
| "step": 4015 |
| }, |
| { |
| "epoch": 25.444444444444443, |
| "grad_norm": 1.886170744895935, |
| "learning_rate": 3.1981321820236885e-06, |
| "loss": 0.2096, |
| "step": 4020 |
| }, |
| { |
| "epoch": 25.476190476190474, |
| "grad_norm": 1.4639593362808228, |
| "learning_rate": 3.152946694653719e-06, |
| "loss": 0.2162, |
| "step": 4025 |
| }, |
| { |
| "epoch": 25.507936507936506, |
| "grad_norm": 1.3187980651855469, |
| "learning_rate": 3.1080611962575084e-06, |
| "loss": 0.2045, |
| "step": 4030 |
| }, |
| { |
| "epoch": 25.53968253968254, |
| "grad_norm": 1.6470718383789062, |
| "learning_rate": 3.0634763031723882e-06, |
| "loss": 0.2056, |
| "step": 4035 |
| }, |
| { |
| "epoch": 25.571428571428573, |
| "grad_norm": 1.702276587486267, |
| "learning_rate": 3.0191926276079806e-06, |
| "loss": 0.2215, |
| "step": 4040 |
| }, |
| { |
| "epoch": 25.603174603174605, |
| "grad_norm": 1.4153060913085938, |
| "learning_rate": 2.9752107776377963e-06, |
| "loss": 0.2112, |
| "step": 4045 |
| }, |
| { |
| "epoch": 25.634920634920636, |
| "grad_norm": 1.5753271579742432, |
| "learning_rate": 2.931531357190881e-06, |
| "loss": 0.2069, |
| "step": 4050 |
| }, |
| { |
| "epoch": 25.666666666666668, |
| "grad_norm": 1.399776577949524, |
| "learning_rate": 2.888154966043516e-06, |
| "loss": 0.2175, |
| "step": 4055 |
| }, |
| { |
| "epoch": 25.6984126984127, |
| "grad_norm": 1.9482572078704834, |
| "learning_rate": 2.845082199811011e-06, |
| "loss": 0.1971, |
| "step": 4060 |
| }, |
| { |
| "epoch": 25.73015873015873, |
| "grad_norm": 1.8254753351211548, |
| "learning_rate": 2.802313649939467e-06, |
| "loss": 0.2204, |
| "step": 4065 |
| }, |
| { |
| "epoch": 25.761904761904763, |
| "grad_norm": 2.1758205890655518, |
| "learning_rate": 2.7598499036977253e-06, |
| "loss": 0.2213, |
| "step": 4070 |
| }, |
| { |
| "epoch": 25.793650793650794, |
| "grad_norm": 1.556915044784546, |
| "learning_rate": 2.717691544169254e-06, |
| "loss": 0.209, |
| "step": 4075 |
| }, |
| { |
| "epoch": 25.825396825396826, |
| "grad_norm": 1.5348148345947266, |
| "learning_rate": 2.675839150244153e-06, |
| "loss": 0.203, |
| "step": 4080 |
| }, |
| { |
| "epoch": 25.857142857142858, |
| "grad_norm": 1.1136623620986938, |
| "learning_rate": 2.634293296611229e-06, |
| "loss": 0.1943, |
| "step": 4085 |
| }, |
| { |
| "epoch": 25.88888888888889, |
| "grad_norm": 1.4910120964050293, |
| "learning_rate": 2.5930545537500515e-06, |
| "loss": 0.1954, |
| "step": 4090 |
| }, |
| { |
| "epoch": 25.92063492063492, |
| "grad_norm": 1.253962755203247, |
| "learning_rate": 2.5521234879231887e-06, |
| "loss": 0.2049, |
| "step": 4095 |
| }, |
| { |
| "epoch": 25.952380952380953, |
| "grad_norm": 1.3038915395736694, |
| "learning_rate": 2.5115006611683755e-06, |
| "loss": 0.208, |
| "step": 4100 |
| }, |
| { |
| "epoch": 25.984126984126984, |
| "grad_norm": 1.5659282207489014, |
| "learning_rate": 2.4711866312908174e-06, |
| "loss": 0.2146, |
| "step": 4105 |
| }, |
| { |
| "epoch": 26.012698412698413, |
| "grad_norm": 1.5508460998535156, |
| "learning_rate": 2.4311819518555295e-06, |
| "loss": 0.1946, |
| "step": 4110 |
| }, |
| { |
| "epoch": 26.044444444444444, |
| "grad_norm": 1.6189981698989868, |
| "learning_rate": 2.391487172179738e-06, |
| "loss": 0.2284, |
| "step": 4115 |
| }, |
| { |
| "epoch": 26.076190476190476, |
| "grad_norm": 1.6239879131317139, |
| "learning_rate": 2.3521028373253457e-06, |
| "loss": 0.2087, |
| "step": 4120 |
| }, |
| { |
| "epoch": 26.107936507936508, |
| "grad_norm": 1.3496952056884766, |
| "learning_rate": 2.3130294880914173e-06, |
| "loss": 0.1975, |
| "step": 4125 |
| }, |
| { |
| "epoch": 26.13968253968254, |
| "grad_norm": 1.1777641773223877, |
| "learning_rate": 2.2742676610067846e-06, |
| "loss": 0.1851, |
| "step": 4130 |
| }, |
| { |
| "epoch": 26.17142857142857, |
| "grad_norm": 1.4527008533477783, |
| "learning_rate": 2.235817888322664e-06, |
| "loss": 0.1768, |
| "step": 4135 |
| }, |
| { |
| "epoch": 26.203174603174602, |
| "grad_norm": 1.4031193256378174, |
| "learning_rate": 2.1976806980053556e-06, |
| "loss": 0.2125, |
| "step": 4140 |
| }, |
| { |
| "epoch": 26.234920634920634, |
| "grad_norm": 1.6796773672103882, |
| "learning_rate": 2.1598566137289806e-06, |
| "loss": 0.2175, |
| "step": 4145 |
| }, |
| { |
| "epoch": 26.266666666666666, |
| "grad_norm": 1.3337997198104858, |
| "learning_rate": 2.1223461548683184e-06, |
| "loss": 0.2081, |
| "step": 4150 |
| }, |
| { |
| "epoch": 26.298412698412697, |
| "grad_norm": 1.5123558044433594, |
| "learning_rate": 2.0851498364916345e-06, |
| "loss": 0.2022, |
| "step": 4155 |
| }, |
| { |
| "epoch": 26.33015873015873, |
| "grad_norm": 1.5714031457901, |
| "learning_rate": 2.04826816935364e-06, |
| "loss": 0.2024, |
| "step": 4160 |
| }, |
| { |
| "epoch": 26.36190476190476, |
| "grad_norm": 1.4283126592636108, |
| "learning_rate": 2.0117016598884765e-06, |
| "loss": 0.1725, |
| "step": 4165 |
| }, |
| { |
| "epoch": 26.393650793650792, |
| "grad_norm": 1.5229535102844238, |
| "learning_rate": 1.975450810202725e-06, |
| "loss": 0.2028, |
| "step": 4170 |
| }, |
| { |
| "epoch": 26.425396825396824, |
| "grad_norm": 1.0714068412780762, |
| "learning_rate": 1.9395161180685796e-06, |
| "loss": 0.1764, |
| "step": 4175 |
| }, |
| { |
| "epoch": 26.457142857142856, |
| "grad_norm": 1.3268301486968994, |
| "learning_rate": 1.9038980769169323e-06, |
| "loss": 0.1962, |
| "step": 4180 |
| }, |
| { |
| "epoch": 26.488888888888887, |
| "grad_norm": 1.5213900804519653, |
| "learning_rate": 1.8685971758306691e-06, |
| "loss": 0.2068, |
| "step": 4185 |
| }, |
| { |
| "epoch": 26.520634920634922, |
| "grad_norm": 1.774190902709961, |
| "learning_rate": 1.833613899537906e-06, |
| "loss": 0.1951, |
| "step": 4190 |
| }, |
| { |
| "epoch": 26.552380952380954, |
| "grad_norm": 1.5567339658737183, |
| "learning_rate": 1.798948728405353e-06, |
| "loss": 0.1967, |
| "step": 4195 |
| }, |
| { |
| "epoch": 26.584126984126986, |
| "grad_norm": 1.5419505834579468, |
| "learning_rate": 1.7646021384317201e-06, |
| "loss": 0.2065, |
| "step": 4200 |
| }, |
| { |
| "epoch": 26.615873015873017, |
| "grad_norm": 1.3854479789733887, |
| "learning_rate": 1.730574601241175e-06, |
| "loss": 0.2121, |
| "step": 4205 |
| }, |
| { |
| "epoch": 26.64761904761905, |
| "grad_norm": 1.505738615989685, |
| "learning_rate": 1.6968665840768588e-06, |
| "loss": 0.21, |
| "step": 4210 |
| }, |
| { |
| "epoch": 26.67936507936508, |
| "grad_norm": 1.7908788919448853, |
| "learning_rate": 1.6634785497944922e-06, |
| "loss": 0.2267, |
| "step": 4215 |
| }, |
| { |
| "epoch": 26.711111111111112, |
| "grad_norm": 1.1521767377853394, |
| "learning_rate": 1.630410956855996e-06, |
| "loss": 0.2066, |
| "step": 4220 |
| }, |
| { |
| "epoch": 26.742857142857144, |
| "grad_norm": 2.14155650138855, |
| "learning_rate": 1.5976642593232145e-06, |
| "loss": 0.2179, |
| "step": 4225 |
| }, |
| { |
| "epoch": 26.774603174603175, |
| "grad_norm": 1.6598525047302246, |
| "learning_rate": 1.5652389068516765e-06, |
| "loss": 0.2235, |
| "step": 4230 |
| }, |
| { |
| "epoch": 26.806349206349207, |
| "grad_norm": 1.3785622119903564, |
| "learning_rate": 1.5331353446844027e-06, |
| "loss": 0.2074, |
| "step": 4235 |
| }, |
| { |
| "epoch": 26.83809523809524, |
| "grad_norm": 1.594824194908142, |
| "learning_rate": 1.5013540136458227e-06, |
| "loss": 0.2138, |
| "step": 4240 |
| }, |
| { |
| "epoch": 26.86984126984127, |
| "grad_norm": 1.6035126447677612, |
| "learning_rate": 1.4698953501356972e-06, |
| "loss": 0.2203, |
| "step": 4245 |
| }, |
| { |
| "epoch": 26.901587301587302, |
| "grad_norm": 1.6783777475357056, |
| "learning_rate": 1.438759786123134e-06, |
| "loss": 0.1916, |
| "step": 4250 |
| }, |
| { |
| "epoch": 26.933333333333334, |
| "grad_norm": 1.8125184774398804, |
| "learning_rate": 1.4079477491406701e-06, |
| "loss": 0.216, |
| "step": 4255 |
| }, |
| { |
| "epoch": 26.965079365079365, |
| "grad_norm": 1.9151347875595093, |
| "learning_rate": 1.3774596622783604e-06, |
| "loss": 0.2108, |
| "step": 4260 |
| }, |
| { |
| "epoch": 26.996825396825397, |
| "grad_norm": 1.611247181892395, |
| "learning_rate": 1.3472959441780286e-06, |
| "loss": 0.203, |
| "step": 4265 |
| }, |
| { |
| "epoch": 27.025396825396825, |
| "grad_norm": 1.6793745756149292, |
| "learning_rate": 1.3174570090274618e-06, |
| "loss": 0.2025, |
| "step": 4270 |
| }, |
| { |
| "epoch": 27.057142857142857, |
| "grad_norm": 1.7114115953445435, |
| "learning_rate": 1.2879432665547558e-06, |
| "loss": 0.2021, |
| "step": 4275 |
| }, |
| { |
| "epoch": 27.08888888888889, |
| "grad_norm": 1.5666910409927368, |
| "learning_rate": 1.2587551220226812e-06, |
| "loss": 0.2035, |
| "step": 4280 |
| }, |
| { |
| "epoch": 27.12063492063492, |
| "grad_norm": 1.398741602897644, |
| "learning_rate": 1.2298929762231098e-06, |
| "loss": 0.2006, |
| "step": 4285 |
| }, |
| { |
| "epoch": 27.152380952380952, |
| "grad_norm": 1.455268144607544, |
| "learning_rate": 1.201357225471536e-06, |
| "loss": 0.2049, |
| "step": 4290 |
| }, |
| { |
| "epoch": 27.184126984126983, |
| "grad_norm": 1.4031116962432861, |
| "learning_rate": 1.1731482616016004e-06, |
| "loss": 0.2034, |
| "step": 4295 |
| }, |
| { |
| "epoch": 27.215873015873015, |
| "grad_norm": 1.7742083072662354, |
| "learning_rate": 1.1452664719597357e-06, |
| "loss": 0.2065, |
| "step": 4300 |
| }, |
| { |
| "epoch": 27.247619047619047, |
| "grad_norm": 1.729286551475525, |
| "learning_rate": 1.1177122393998374e-06, |
| "loss": 0.2125, |
| "step": 4305 |
| }, |
| { |
| "epoch": 27.27936507936508, |
| "grad_norm": 1.7302000522613525, |
| "learning_rate": 1.0904859422780079e-06, |
| "loss": 0.2179, |
| "step": 4310 |
| }, |
| { |
| "epoch": 27.31111111111111, |
| "grad_norm": 1.6988170146942139, |
| "learning_rate": 1.0635879544473592e-06, |
| "loss": 0.2039, |
| "step": 4315 |
| }, |
| { |
| "epoch": 27.34285714285714, |
| "grad_norm": 1.6711211204528809, |
| "learning_rate": 1.0370186452528935e-06, |
| "loss": 0.2083, |
| "step": 4320 |
| }, |
| { |
| "epoch": 27.374603174603173, |
| "grad_norm": 1.3255887031555176, |
| "learning_rate": 1.0107783795264002e-06, |
| "loss": 0.2027, |
| "step": 4325 |
| }, |
| { |
| "epoch": 27.406349206349205, |
| "grad_norm": 1.5709261894226074, |
| "learning_rate": 9.848675175814865e-07, |
| "loss": 0.2083, |
| "step": 4330 |
| }, |
| { |
| "epoch": 27.438095238095237, |
| "grad_norm": 1.546505093574524, |
| "learning_rate": 9.592864152085963e-07, |
| "loss": 0.2096, |
| "step": 4335 |
| }, |
| { |
| "epoch": 27.469841269841268, |
| "grad_norm": 1.689866304397583, |
| "learning_rate": 9.340354236701404e-07, |
| "loss": 0.2065, |
| "step": 4340 |
| }, |
| { |
| "epoch": 27.5015873015873, |
| "grad_norm": 1.6024067401885986, |
| "learning_rate": 9.091148896956836e-07, |
| "loss": 0.1989, |
| "step": 4345 |
| }, |
| { |
| "epoch": 27.533333333333335, |
| "grad_norm": 1.393102765083313, |
| "learning_rate": 8.845251554771422e-07, |
| "loss": 0.1883, |
| "step": 4350 |
| }, |
| { |
| "epoch": 27.565079365079367, |
| "grad_norm": 1.697074055671692, |
| "learning_rate": 8.602665586641411e-07, |
| "loss": 0.2219, |
| "step": 4355 |
| }, |
| { |
| "epoch": 27.5968253968254, |
| "grad_norm": 1.2590184211730957, |
| "learning_rate": 8.363394323593427e-07, |
| "loss": 0.1884, |
| "step": 4360 |
| }, |
| { |
| "epoch": 27.62857142857143, |
| "grad_norm": 1.6093244552612305, |
| "learning_rate": 8.127441051138662e-07, |
| "loss": 0.2005, |
| "step": 4365 |
| }, |
| { |
| "epoch": 27.66031746031746, |
| "grad_norm": 1.4474371671676636, |
| "learning_rate": 7.894809009228177e-07, |
| "loss": 0.2055, |
| "step": 4370 |
| }, |
| { |
| "epoch": 27.692063492063493, |
| "grad_norm": 2.0293591022491455, |
| "learning_rate": 7.665501392207813e-07, |
| "loss": 0.1919, |
| "step": 4375 |
| }, |
| { |
| "epoch": 27.723809523809525, |
| "grad_norm": 1.2634299993515015, |
| "learning_rate": 7.439521348774959e-07, |
| "loss": 0.19, |
| "step": 4380 |
| }, |
| { |
| "epoch": 27.755555555555556, |
| "grad_norm": 1.4110088348388672, |
| "learning_rate": 7.216871981934797e-07, |
| "loss": 0.1945, |
| "step": 4385 |
| }, |
| { |
| "epoch": 27.787301587301588, |
| "grad_norm": 1.2176947593688965, |
| "learning_rate": 6.997556348958018e-07, |
| "loss": 0.1926, |
| "step": 4390 |
| }, |
| { |
| "epoch": 27.81904761904762, |
| "grad_norm": 1.592971920967102, |
| "learning_rate": 6.781577461338673e-07, |
| "loss": 0.2062, |
| "step": 4395 |
| }, |
| { |
| "epoch": 27.85079365079365, |
| "grad_norm": 1.8953907489776611, |
| "learning_rate": 6.56893828475294e-07, |
| "loss": 0.2241, |
| "step": 4400 |
| }, |
| { |
| "epoch": 27.882539682539683, |
| "grad_norm": 1.6328877210617065, |
| "learning_rate": 6.359641739018263e-07, |
| "loss": 0.1939, |
| "step": 4405 |
| }, |
| { |
| "epoch": 27.914285714285715, |
| "grad_norm": 1.7055280208587646, |
| "learning_rate": 6.153690698053438e-07, |
| "loss": 0.1957, |
| "step": 4410 |
| }, |
| { |
| "epoch": 27.946031746031746, |
| "grad_norm": 1.6427103281021118, |
| "learning_rate": 5.951087989838983e-07, |
| "loss": 0.2129, |
| "step": 4415 |
| }, |
| { |
| "epoch": 27.977777777777778, |
| "grad_norm": 1.6233084201812744, |
| "learning_rate": 5.751836396378357e-07, |
| "loss": 0.2038, |
| "step": 4420 |
| }, |
| { |
| "epoch": 28.006349206349206, |
| "grad_norm": 1.4565635919570923, |
| "learning_rate": 5.555938653659859e-07, |
| "loss": 0.1861, |
| "step": 4425 |
| }, |
| { |
| "epoch": 28.038095238095238, |
| "grad_norm": 1.4828596115112305, |
| "learning_rate": 5.363397451618845e-07, |
| "loss": 0.2128, |
| "step": 4430 |
| }, |
| { |
| "epoch": 28.06984126984127, |
| "grad_norm": 1.4575839042663574, |
| "learning_rate": 5.174215434101099e-07, |
| "loss": 0.2048, |
| "step": 4435 |
| }, |
| { |
| "epoch": 28.1015873015873, |
| "grad_norm": 1.6370646953582764, |
| "learning_rate": 4.988395198826157e-07, |
| "loss": 0.1985, |
| "step": 4440 |
| }, |
| { |
| "epoch": 28.133333333333333, |
| "grad_norm": 1.1092596054077148, |
| "learning_rate": 4.805939297351986e-07, |
| "loss": 0.198, |
| "step": 4445 |
| }, |
| { |
| "epoch": 28.165079365079364, |
| "grad_norm": 1.5861098766326904, |
| "learning_rate": 4.6268502350397536e-07, |
| "loss": 0.2005, |
| "step": 4450 |
| }, |
| { |
| "epoch": 28.196825396825396, |
| "grad_norm": 1.5998700857162476, |
| "learning_rate": 4.45113047101936e-07, |
| "loss": 0.2022, |
| "step": 4455 |
| }, |
| { |
| "epoch": 28.228571428571428, |
| "grad_norm": 1.7914454936981201, |
| "learning_rate": 4.278782418155991e-07, |
| "loss": 0.2003, |
| "step": 4460 |
| }, |
| { |
| "epoch": 28.26031746031746, |
| "grad_norm": 1.6202130317687988, |
| "learning_rate": 4.1098084430165605e-07, |
| "loss": 0.2124, |
| "step": 4465 |
| }, |
| { |
| "epoch": 28.29206349206349, |
| "grad_norm": 1.7337385416030884, |
| "learning_rate": 3.944210865837572e-07, |
| "loss": 0.2066, |
| "step": 4470 |
| }, |
| { |
| "epoch": 28.323809523809523, |
| "grad_norm": 1.4690618515014648, |
| "learning_rate": 3.781991960493031e-07, |
| "loss": 0.2127, |
| "step": 4475 |
| }, |
| { |
| "epoch": 28.355555555555554, |
| "grad_norm": 1.5113608837127686, |
| "learning_rate": 3.623153954463332e-07, |
| "loss": 0.1946, |
| "step": 4480 |
| }, |
| { |
| "epoch": 28.387301587301586, |
| "grad_norm": 1.5071619749069214, |
| "learning_rate": 3.467699028804672e-07, |
| "loss": 0.2011, |
| "step": 4485 |
| }, |
| { |
| "epoch": 28.419047619047618, |
| "grad_norm": 1.3782387971878052, |
| "learning_rate": 3.315629318119018e-07, |
| "loss": 0.1904, |
| "step": 4490 |
| }, |
| { |
| "epoch": 28.45079365079365, |
| "grad_norm": 1.6402249336242676, |
| "learning_rate": 3.1669469105248537e-07, |
| "loss": 0.2077, |
| "step": 4495 |
| }, |
| { |
| "epoch": 28.48253968253968, |
| "grad_norm": 1.9802137613296509, |
| "learning_rate": 3.0216538476286196e-07, |
| "loss": 0.218, |
| "step": 4500 |
| }, |
| { |
| "epoch": 28.514285714285712, |
| "grad_norm": 1.4524375200271606, |
| "learning_rate": 2.8797521244964263e-07, |
| "loss": 0.2009, |
| "step": 4505 |
| }, |
| { |
| "epoch": 28.546031746031748, |
| "grad_norm": 1.1066787242889404, |
| "learning_rate": 2.74124368962686e-07, |
| "loss": 0.1689, |
| "step": 4510 |
| }, |
| { |
| "epoch": 28.57777777777778, |
| "grad_norm": 1.4385684728622437, |
| "learning_rate": 2.6061304449241655e-07, |
| "loss": 0.2104, |
| "step": 4515 |
| }, |
| { |
| "epoch": 28.60952380952381, |
| "grad_norm": 1.6011552810668945, |
| "learning_rate": 2.4744142456720744e-07, |
| "loss": 0.2036, |
| "step": 4520 |
| }, |
| { |
| "epoch": 28.641269841269843, |
| "grad_norm": 1.4969830513000488, |
| "learning_rate": 2.346096900508521e-07, |
| "loss": 0.2049, |
| "step": 4525 |
| }, |
| { |
| "epoch": 28.673015873015874, |
| "grad_norm": 1.191385269165039, |
| "learning_rate": 2.2211801714004942e-07, |
| "loss": 0.1983, |
| "step": 4530 |
| }, |
| { |
| "epoch": 28.704761904761906, |
| "grad_norm": 1.4603822231292725, |
| "learning_rate": 2.0996657736201687e-07, |
| "loss": 0.1951, |
| "step": 4535 |
| }, |
| { |
| "epoch": 28.736507936507937, |
| "grad_norm": 1.7533866167068481, |
| "learning_rate": 1.9815553757211446e-07, |
| "loss": 0.1953, |
| "step": 4540 |
| }, |
| { |
| "epoch": 28.76825396825397, |
| "grad_norm": 1.6291053295135498, |
| "learning_rate": 1.8668505995155515e-07, |
| "loss": 0.2007, |
| "step": 4545 |
| }, |
| { |
| "epoch": 28.8, |
| "grad_norm": 1.4532676935195923, |
| "learning_rate": 1.7555530200519255e-07, |
| "loss": 0.1982, |
| "step": 4550 |
| }, |
| { |
| "epoch": 28.831746031746032, |
| "grad_norm": 2.099191427230835, |
| "learning_rate": 1.6476641655933933e-07, |
| "loss": 0.2065, |
| "step": 4555 |
| }, |
| { |
| "epoch": 28.863492063492064, |
| "grad_norm": 1.7901203632354736, |
| "learning_rate": 1.5431855175968014e-07, |
| "loss": 0.1942, |
| "step": 4560 |
| }, |
| { |
| "epoch": 28.895238095238096, |
| "grad_norm": 1.4106049537658691, |
| "learning_rate": 1.4421185106923696e-07, |
| "loss": 0.1987, |
| "step": 4565 |
| }, |
| { |
| "epoch": 28.926984126984127, |
| "grad_norm": 1.347052812576294, |
| "learning_rate": 1.3444645326639026e-07, |
| "loss": 0.2083, |
| "step": 4570 |
| }, |
| { |
| "epoch": 28.95873015873016, |
| "grad_norm": 1.6261069774627686, |
| "learning_rate": 1.250224924429888e-07, |
| "loss": 0.2209, |
| "step": 4575 |
| }, |
| { |
| "epoch": 28.99047619047619, |
| "grad_norm": 2.0656380653381348, |
| "learning_rate": 1.1594009800248995e-07, |
| "loss": 0.1929, |
| "step": 4580 |
| }, |
| { |
| "epoch": 29.01904761904762, |
| "grad_norm": 1.560865879058838, |
| "learning_rate": 1.0719939465820006e-07, |
| "loss": 0.2078, |
| "step": 4585 |
| }, |
| { |
| "epoch": 29.05079365079365, |
| "grad_norm": 1.716450810432434, |
| "learning_rate": 9.880050243155359e-08, |
| "loss": 0.2133, |
| "step": 4590 |
| }, |
| { |
| "epoch": 29.082539682539682, |
| "grad_norm": 1.6580930948257446, |
| "learning_rate": 9.074353665045887e-08, |
| "loss": 0.2028, |
| "step": 4595 |
| }, |
| { |
| "epoch": 29.114285714285714, |
| "grad_norm": 1.5782244205474854, |
| "learning_rate": 8.302860794772993e-08, |
| "loss": 0.1999, |
| "step": 4600 |
| }, |
| { |
| "epoch": 29.146031746031746, |
| "grad_norm": 1.5206737518310547, |
| "learning_rate": 7.565582225955158e-08, |
| "loss": 0.2062, |
| "step": 4605 |
| }, |
| { |
| "epoch": 29.177777777777777, |
| "grad_norm": 1.3276293277740479, |
| "learning_rate": 6.862528082403341e-08, |
| "loss": 0.2015, |
| "step": 4610 |
| }, |
| { |
| "epoch": 29.20952380952381, |
| "grad_norm": 1.330771565437317, |
| "learning_rate": 6.193708017982747e-08, |
| "loss": 0.2095, |
| "step": 4615 |
| }, |
| { |
| "epoch": 29.24126984126984, |
| "grad_norm": 1.4298193454742432, |
| "learning_rate": 5.5591312164776646e-08, |
| "loss": 0.1998, |
| "step": 4620 |
| }, |
| { |
| "epoch": 29.273015873015872, |
| "grad_norm": 1.6888970136642456, |
| "learning_rate": 4.958806391468507e-08, |
| "loss": 0.1991, |
| "step": 4625 |
| }, |
| { |
| "epoch": 29.304761904761904, |
| "grad_norm": 1.7637996673583984, |
| "learning_rate": 4.392741786209964e-08, |
| "loss": 0.195, |
| "step": 4630 |
| }, |
| { |
| "epoch": 29.336507936507935, |
| "grad_norm": 1.4855365753173828, |
| "learning_rate": 3.860945173518593e-08, |
| "loss": 0.2016, |
| "step": 4635 |
| }, |
| { |
| "epoch": 29.368253968253967, |
| "grad_norm": 1.2792131900787354, |
| "learning_rate": 3.363423855666237e-08, |
| "loss": 0.186, |
| "step": 4640 |
| }, |
| { |
| "epoch": 29.4, |
| "grad_norm": 1.4719171524047852, |
| "learning_rate": 2.9001846642795484e-08, |
| "loss": 0.2052, |
| "step": 4645 |
| }, |
| { |
| "epoch": 29.43174603174603, |
| "grad_norm": 1.4484893083572388, |
| "learning_rate": 2.4712339602461774e-08, |
| "loss": 0.2086, |
| "step": 4650 |
| }, |
| { |
| "epoch": 29.463492063492062, |
| "grad_norm": 1.7089500427246094, |
| "learning_rate": 2.0765776336270638e-08, |
| "loss": 0.1959, |
| "step": 4655 |
| }, |
| { |
| "epoch": 29.495238095238093, |
| "grad_norm": 1.748536229133606, |
| "learning_rate": 1.716221103576221e-08, |
| "loss": 0.209, |
| "step": 4660 |
| }, |
| { |
| "epoch": 29.52698412698413, |
| "grad_norm": 1.675736904144287, |
| "learning_rate": 1.3901693182660768e-08, |
| "loss": 0.1923, |
| "step": 4665 |
| }, |
| { |
| "epoch": 29.55873015873016, |
| "grad_norm": 1.427011251449585, |
| "learning_rate": 1.0984267548194704e-08, |
| "loss": 0.1998, |
| "step": 4670 |
| }, |
| { |
| "epoch": 29.590476190476192, |
| "grad_norm": 1.8168760538101196, |
| "learning_rate": 8.409974192483127e-09, |
| "loss": 0.201, |
| "step": 4675 |
| }, |
| { |
| "epoch": 29.622222222222224, |
| "grad_norm": 1.398505687713623, |
| "learning_rate": 6.178848463980758e-09, |
| "loss": 0.1966, |
| "step": 4680 |
| }, |
| { |
| "epoch": 29.653968253968255, |
| "grad_norm": 1.3066248893737793, |
| "learning_rate": 4.290920999000525e-09, |
| "loss": 0.2011, |
| "step": 4685 |
| }, |
| { |
| "epoch": 29.685714285714287, |
| "grad_norm": 1.4169343709945679, |
| "learning_rate": 2.746217721283362e-09, |
| "loss": 0.2126, |
| "step": 4690 |
| }, |
| { |
| "epoch": 29.71746031746032, |
| "grad_norm": 1.8251289129257202, |
| "learning_rate": 1.544759841654031e-09, |
| "loss": 0.1973, |
| "step": 4695 |
| }, |
| { |
| "epoch": 29.74920634920635, |
| "grad_norm": 1.8100301027297974, |
| "learning_rate": 6.865638577185918e-10, |
| "loss": 0.1914, |
| "step": 4700 |
| }, |
| { |
| "epoch": 29.78095238095238, |
| "grad_norm": 1.5966072082519531, |
| "learning_rate": 1.716415536451299e-10, |
| "loss": 0.2116, |
| "step": 4705 |
| }, |
| { |
| "epoch": 29.812698412698413, |
| "grad_norm": 1.8674540519714355, |
| "learning_rate": 0.0, |
| "loss": 0.1886, |
| "step": 4710 |
| }, |
| { |
| "epoch": 29.812698412698413, |
| "step": 4710, |
| "total_flos": 1.011586335464489e+17, |
| "train_loss": 0.4251327198149039, |
| "train_runtime": 14594.4914, |
| "train_samples_per_second": 2.59, |
| "train_steps_per_second": 0.323 |
| } |
| ], |
| "logging_steps": 5, |
| "max_steps": 4710, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 30, |
| "save_steps": 157, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1.011586335464489e+17, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
