| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.3472222222222222, | |
| "eval_steps": 500, | |
| "global_step": 500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0006944444444444445, | |
| "grad_norm": 0.38245001435279846, | |
| "learning_rate": 9.090909090909091e-07, | |
| "loss": 1.5265, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.001388888888888889, | |
| "grad_norm": 0.36873823404312134, | |
| "learning_rate": 1.8181818181818183e-06, | |
| "loss": 1.5149, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0020833333333333333, | |
| "grad_norm": 0.31270918250083923, | |
| "learning_rate": 2.7272727272727272e-06, | |
| "loss": 1.478, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.002777777777777778, | |
| "grad_norm": 0.21180707216262817, | |
| "learning_rate": 3.6363636363636366e-06, | |
| "loss": 1.4297, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.003472222222222222, | |
| "grad_norm": 0.10717298090457916, | |
| "learning_rate": 4.5454545454545455e-06, | |
| "loss": 1.3724, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.004166666666666667, | |
| "grad_norm": 0.12929688394069672, | |
| "learning_rate": 5.4545454545454545e-06, | |
| "loss": 1.3816, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.004861111111111111, | |
| "grad_norm": 0.10623333603143692, | |
| "learning_rate": 6.363636363636364e-06, | |
| "loss": 1.3061, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.005555555555555556, | |
| "grad_norm": 0.07657431811094284, | |
| "learning_rate": 7.272727272727273e-06, | |
| "loss": 1.2736, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.00625, | |
| "grad_norm": 0.05140658840537071, | |
| "learning_rate": 8.181818181818183e-06, | |
| "loss": 1.2767, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.006944444444444444, | |
| "grad_norm": 0.04951579496264458, | |
| "learning_rate": 9.090909090909091e-06, | |
| "loss": 1.2659, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.007638888888888889, | |
| "grad_norm": 0.035326242446899414, | |
| "learning_rate": 1e-05, | |
| "loss": 1.1883, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.008333333333333333, | |
| "grad_norm": 0.03807157278060913, | |
| "learning_rate": 1.0909090909090909e-05, | |
| "loss": 1.2452, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.009027777777777777, | |
| "grad_norm": 0.03424696624279022, | |
| "learning_rate": 1.181818181818182e-05, | |
| "loss": 1.2394, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.009722222222222222, | |
| "grad_norm": 0.026057492941617966, | |
| "learning_rate": 1.2727272727272728e-05, | |
| "loss": 1.157, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.010416666666666666, | |
| "grad_norm": 0.024454891681671143, | |
| "learning_rate": 1.3636363636363637e-05, | |
| "loss": 1.1901, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.011111111111111112, | |
| "grad_norm": 0.02439827099442482, | |
| "learning_rate": 1.4545454545454546e-05, | |
| "loss": 1.1669, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.011805555555555555, | |
| "grad_norm": 0.023934854194521904, | |
| "learning_rate": 1.5454545454545454e-05, | |
| "loss": 1.1833, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0125, | |
| "grad_norm": 0.023052480071783066, | |
| "learning_rate": 1.6363636363636366e-05, | |
| "loss": 1.1777, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.013194444444444444, | |
| "grad_norm": 0.0216195210814476, | |
| "learning_rate": 1.7272727272727274e-05, | |
| "loss": 1.1541, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.013888888888888888, | |
| "grad_norm": 0.021930918097496033, | |
| "learning_rate": 1.8181818181818182e-05, | |
| "loss": 1.1519, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.014583333333333334, | |
| "grad_norm": 0.01956241950392723, | |
| "learning_rate": 1.9090909090909094e-05, | |
| "loss": 1.1289, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.015277777777777777, | |
| "grad_norm": 0.019718188792467117, | |
| "learning_rate": 2e-05, | |
| "loss": 1.12, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.01597222222222222, | |
| "grad_norm": 0.018330449238419533, | |
| "learning_rate": 2.090909090909091e-05, | |
| "loss": 1.151, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.016666666666666666, | |
| "grad_norm": 0.019966386258602142, | |
| "learning_rate": 2.1818181818181818e-05, | |
| "loss": 1.1717, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.017361111111111112, | |
| "grad_norm": 0.019648971036076546, | |
| "learning_rate": 2.2727272727272733e-05, | |
| "loss": 1.1188, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.018055555555555554, | |
| "grad_norm": 0.018150752410292625, | |
| "learning_rate": 2.363636363636364e-05, | |
| "loss": 1.0927, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.01875, | |
| "grad_norm": 0.02000141702592373, | |
| "learning_rate": 2.454545454545455e-05, | |
| "loss": 1.1205, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.019444444444444445, | |
| "grad_norm": 0.02152254991233349, | |
| "learning_rate": 2.5454545454545457e-05, | |
| "loss": 1.1136, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.02013888888888889, | |
| "grad_norm": 0.016942821443080902, | |
| "learning_rate": 2.6363636363636365e-05, | |
| "loss": 1.1073, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.020833333333333332, | |
| "grad_norm": 0.018401362001895905, | |
| "learning_rate": 2.7272727272727273e-05, | |
| "loss": 1.0745, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.021527777777777778, | |
| "grad_norm": 0.01856987737119198, | |
| "learning_rate": 2.8181818181818185e-05, | |
| "loss": 1.0699, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.022222222222222223, | |
| "grad_norm": 0.014833790250122547, | |
| "learning_rate": 2.9090909090909093e-05, | |
| "loss": 1.0226, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.022916666666666665, | |
| "grad_norm": 0.01955495961010456, | |
| "learning_rate": 3.0000000000000004e-05, | |
| "loss": 1.1012, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.02361111111111111, | |
| "grad_norm": 0.01699119061231613, | |
| "learning_rate": 3.090909090909091e-05, | |
| "loss": 1.0457, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.024305555555555556, | |
| "grad_norm": 0.020263031125068665, | |
| "learning_rate": 3.181818181818182e-05, | |
| "loss": 1.0581, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.025, | |
| "grad_norm": 0.015922358259558678, | |
| "learning_rate": 3.272727272727273e-05, | |
| "loss": 1.0958, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.025694444444444443, | |
| "grad_norm": 0.016595132648944855, | |
| "learning_rate": 3.363636363636364e-05, | |
| "loss": 1.0425, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.02638888888888889, | |
| "grad_norm": 0.018398325890302658, | |
| "learning_rate": 3.454545454545455e-05, | |
| "loss": 1.0893, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.027083333333333334, | |
| "grad_norm": 0.017498090863227844, | |
| "learning_rate": 3.545454545454546e-05, | |
| "loss": 1.0487, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.027777777777777776, | |
| "grad_norm": 0.016576213762164116, | |
| "learning_rate": 3.6363636363636364e-05, | |
| "loss": 1.0567, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02847222222222222, | |
| "grad_norm": 0.013903766870498657, | |
| "learning_rate": 3.7272727272727276e-05, | |
| "loss": 1.023, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.029166666666666667, | |
| "grad_norm": 0.01546618901193142, | |
| "learning_rate": 3.818181818181819e-05, | |
| "loss": 1.0456, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.029861111111111113, | |
| "grad_norm": 0.014900722540915012, | |
| "learning_rate": 3.909090909090909e-05, | |
| "loss": 1.0365, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.030555555555555555, | |
| "grad_norm": 0.015999475494027138, | |
| "learning_rate": 4e-05, | |
| "loss": 1.0486, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.03125, | |
| "grad_norm": 0.01696275919675827, | |
| "learning_rate": 3.999994935591541e-05, | |
| "loss": 1.0602, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.03194444444444444, | |
| "grad_norm": 0.016145754605531693, | |
| "learning_rate": 3.999979742391812e-05, | |
| "loss": 1.0709, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.03263888888888889, | |
| "grad_norm": 0.015007151290774345, | |
| "learning_rate": 3.999954420477757e-05, | |
| "loss": 1.0273, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.03333333333333333, | |
| "grad_norm": 0.016809619963169098, | |
| "learning_rate": 3.9999189699776166e-05, | |
| "loss": 1.062, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.034027777777777775, | |
| "grad_norm": 0.017114873975515366, | |
| "learning_rate": 3.9998733910709277e-05, | |
| "loss": 1.0624, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.034722222222222224, | |
| "grad_norm": 0.014565072022378445, | |
| "learning_rate": 3.9998176839885196e-05, | |
| "loss": 1.0487, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.035416666666666666, | |
| "grad_norm": 0.014314558357000351, | |
| "learning_rate": 3.9997518490125166e-05, | |
| "loss": 1.0673, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.03611111111111111, | |
| "grad_norm": 0.01658693701028824, | |
| "learning_rate": 3.999675886476332e-05, | |
| "loss": 1.0328, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.03680555555555556, | |
| "grad_norm": 0.014420751482248306, | |
| "learning_rate": 3.9995897967646725e-05, | |
| "loss": 0.9782, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.0375, | |
| "grad_norm": 0.017560573294758797, | |
| "learning_rate": 3.999493580313532e-05, | |
| "loss": 1.0367, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.03819444444444445, | |
| "grad_norm": 0.017693940550088882, | |
| "learning_rate": 3.9993872376101894e-05, | |
| "loss": 1.0518, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.03888888888888889, | |
| "grad_norm": 0.014515440911054611, | |
| "learning_rate": 3.9992707691932067e-05, | |
| "loss": 1.0322, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.03958333333333333, | |
| "grad_norm": 0.015049392357468605, | |
| "learning_rate": 3.999144175652428e-05, | |
| "loss": 1.0323, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.04027777777777778, | |
| "grad_norm": 0.017379263415932655, | |
| "learning_rate": 3.999007457628976e-05, | |
| "loss": 1.0391, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.04097222222222222, | |
| "grad_norm": 0.01676369458436966, | |
| "learning_rate": 3.998860615815246e-05, | |
| "loss": 1.0637, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.041666666666666664, | |
| "grad_norm": 0.020871426910161972, | |
| "learning_rate": 3.9987036509549034e-05, | |
| "loss": 1.0831, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04236111111111111, | |
| "grad_norm": 0.01579645834863186, | |
| "learning_rate": 3.998536563842884e-05, | |
| "loss": 1.0439, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.043055555555555555, | |
| "grad_norm": 0.021290535107254982, | |
| "learning_rate": 3.998359355325384e-05, | |
| "loss": 1.0859, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.04375, | |
| "grad_norm": 0.016678057610988617, | |
| "learning_rate": 3.998172026299861e-05, | |
| "loss": 1.0521, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.044444444444444446, | |
| "grad_norm": 0.016436951234936714, | |
| "learning_rate": 3.997974577715026e-05, | |
| "loss": 1.0156, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.04513888888888889, | |
| "grad_norm": 0.016367288306355476, | |
| "learning_rate": 3.9977670105708377e-05, | |
| "loss": 1.0365, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.04583333333333333, | |
| "grad_norm": 0.017032478004693985, | |
| "learning_rate": 3.997549325918501e-05, | |
| "loss": 1.068, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.04652777777777778, | |
| "grad_norm": 0.015650460496544838, | |
| "learning_rate": 3.997321524860461e-05, | |
| "loss": 1.0361, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.04722222222222222, | |
| "grad_norm": 0.014867709018290043, | |
| "learning_rate": 3.997083608550395e-05, | |
| "loss": 1.0, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.04791666666666667, | |
| "grad_norm": 0.015197121538221836, | |
| "learning_rate": 3.996835578193208e-05, | |
| "loss": 1.0488, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.04861111111111111, | |
| "grad_norm": 0.015954626724123955, | |
| "learning_rate": 3.996577435045027e-05, | |
| "loss": 1.0356, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.049305555555555554, | |
| "grad_norm": 0.01822635903954506, | |
| "learning_rate": 3.996309180413195e-05, | |
| "loss": 1.0493, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 0.014613306149840355, | |
| "learning_rate": 3.996030815656262e-05, | |
| "loss": 0.9882, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.050694444444444445, | |
| "grad_norm": 0.015572323463857174, | |
| "learning_rate": 3.995742342183982e-05, | |
| "loss": 1.0162, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.05138888888888889, | |
| "grad_norm": 0.01604314148426056, | |
| "learning_rate": 3.9954437614573015e-05, | |
| "loss": 1.0424, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.052083333333333336, | |
| "grad_norm": 0.015765074640512466, | |
| "learning_rate": 3.9951350749883555e-05, | |
| "loss": 1.0607, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05277777777777778, | |
| "grad_norm": 0.014984416775405407, | |
| "learning_rate": 3.994816284340459e-05, | |
| "loss": 1.0443, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.05347222222222222, | |
| "grad_norm": 0.016170239076018333, | |
| "learning_rate": 3.9944873911280976e-05, | |
| "loss": 1.0275, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.05416666666666667, | |
| "grad_norm": 0.014334661886096, | |
| "learning_rate": 3.99414839701692e-05, | |
| "loss": 1.0175, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.05486111111111111, | |
| "grad_norm": 0.014810354448854923, | |
| "learning_rate": 3.993799303723733e-05, | |
| "loss": 1.0015, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.05555555555555555, | |
| "grad_norm": 0.014517512172460556, | |
| "learning_rate": 3.993440113016485e-05, | |
| "loss": 1.0137, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.05625, | |
| "grad_norm": 0.01477818377315998, | |
| "learning_rate": 3.993070826714267e-05, | |
| "loss": 0.9852, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.05694444444444444, | |
| "grad_norm": 0.014672035351395607, | |
| "learning_rate": 3.9926914466872936e-05, | |
| "loss": 1.0099, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.05763888888888889, | |
| "grad_norm": 0.015323956497013569, | |
| "learning_rate": 3.9923019748569015e-05, | |
| "loss": 1.0345, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.058333333333333334, | |
| "grad_norm": 0.014713946729898453, | |
| "learning_rate": 3.991902413195535e-05, | |
| "loss": 1.0318, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.059027777777777776, | |
| "grad_norm": 0.01328709814697504, | |
| "learning_rate": 3.9914927637267366e-05, | |
| "loss": 1.0084, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.059722222222222225, | |
| "grad_norm": 0.017166944220662117, | |
| "learning_rate": 3.991073028525139e-05, | |
| "loss": 1.0036, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.06041666666666667, | |
| "grad_norm": 0.016073813661932945, | |
| "learning_rate": 3.990643209716454e-05, | |
| "loss": 1.0329, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.06111111111111111, | |
| "grad_norm": 0.01435924507677555, | |
| "learning_rate": 3.990203309477457e-05, | |
| "loss": 1.0535, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.06180555555555556, | |
| "grad_norm": 0.014585713855922222, | |
| "learning_rate": 3.989753330035985e-05, | |
| "loss": 1.0016, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0625, | |
| "grad_norm": 0.014899005182087421, | |
| "learning_rate": 3.989293273670916e-05, | |
| "loss": 1.0386, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.06319444444444444, | |
| "grad_norm": 0.015017863363027573, | |
| "learning_rate": 3.988823142712165e-05, | |
| "loss": 1.0217, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.06388888888888888, | |
| "grad_norm": 0.014558105729520321, | |
| "learning_rate": 3.9883429395406666e-05, | |
| "loss": 1.0046, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.06458333333333334, | |
| "grad_norm": 0.014639356173574924, | |
| "learning_rate": 3.987852666588364e-05, | |
| "loss": 1.0226, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.06527777777777778, | |
| "grad_norm": 0.015396458096802235, | |
| "learning_rate": 3.9873523263382015e-05, | |
| "loss": 1.0374, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.06597222222222222, | |
| "grad_norm": 0.014208097010850906, | |
| "learning_rate": 3.9868419213241064e-05, | |
| "loss": 1.0058, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.06666666666666667, | |
| "grad_norm": 0.013101368211209774, | |
| "learning_rate": 3.986321454130978e-05, | |
| "loss": 0.9642, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.06736111111111111, | |
| "grad_norm": 0.014994910918176174, | |
| "learning_rate": 3.9857909273946747e-05, | |
| "loss": 0.995, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.06805555555555555, | |
| "grad_norm": 0.01322688814252615, | |
| "learning_rate": 3.985250343802e-05, | |
| "loss": 1.0124, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.06875, | |
| "grad_norm": 0.014946015551686287, | |
| "learning_rate": 3.984699706090691e-05, | |
| "loss": 0.9908, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.06944444444444445, | |
| "grad_norm": 0.014106931164860725, | |
| "learning_rate": 3.9841390170494024e-05, | |
| "loss": 0.9985, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07013888888888889, | |
| "grad_norm": 0.014384300448000431, | |
| "learning_rate": 3.9835682795176905e-05, | |
| "loss": 0.9727, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.07083333333333333, | |
| "grad_norm": 0.014217739924788475, | |
| "learning_rate": 3.982987496386004e-05, | |
| "loss": 0.9839, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.07152777777777777, | |
| "grad_norm": 0.015070730820298195, | |
| "learning_rate": 3.982396670595668e-05, | |
| "loss": 1.0052, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.07222222222222222, | |
| "grad_norm": 0.01753557287156582, | |
| "learning_rate": 3.9817958051388624e-05, | |
| "loss": 1.0035, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.07291666666666667, | |
| "grad_norm": 0.015090793371200562, | |
| "learning_rate": 3.981184903058618e-05, | |
| "loss": 1.0413, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.07361111111111111, | |
| "grad_norm": 0.017897464334964752, | |
| "learning_rate": 3.980563967448791e-05, | |
| "loss": 0.9938, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.07430555555555556, | |
| "grad_norm": 0.015342462807893753, | |
| "learning_rate": 3.979933001454053e-05, | |
| "loss": 1.003, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.075, | |
| "grad_norm": 0.01586705632507801, | |
| "learning_rate": 3.979292008269874e-05, | |
| "loss": 1.0052, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.07569444444444444, | |
| "grad_norm": 0.013492800295352936, | |
| "learning_rate": 3.978640991142505e-05, | |
| "loss": 1.0415, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.0763888888888889, | |
| "grad_norm": 0.017706014215946198, | |
| "learning_rate": 3.9779799533689634e-05, | |
| "loss": 0.9927, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.07708333333333334, | |
| "grad_norm": 0.015358942560851574, | |
| "learning_rate": 3.9773088982970135e-05, | |
| "loss": 1.0201, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.07777777777777778, | |
| "grad_norm": 0.01901993341743946, | |
| "learning_rate": 3.9766278293251526e-05, | |
| "loss": 1.0191, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.07847222222222222, | |
| "grad_norm": 0.015617494471371174, | |
| "learning_rate": 3.9759367499025924e-05, | |
| "loss": 1.0256, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.07916666666666666, | |
| "grad_norm": 0.014908060431480408, | |
| "learning_rate": 3.9752356635292405e-05, | |
| "loss": 0.9898, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.0798611111111111, | |
| "grad_norm": 0.013952597044408321, | |
| "learning_rate": 3.974524573755686e-05, | |
| "loss": 0.9862, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.08055555555555556, | |
| "grad_norm": 0.013434095308184624, | |
| "learning_rate": 3.9738034841831776e-05, | |
| "loss": 0.9593, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.08125, | |
| "grad_norm": 0.014166664332151413, | |
| "learning_rate": 3.9730723984636064e-05, | |
| "loss": 1.0377, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.08194444444444444, | |
| "grad_norm": 0.014752853661775589, | |
| "learning_rate": 3.9723313202994904e-05, | |
| "loss": 1.0139, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.08263888888888889, | |
| "grad_norm": 0.014744486659765244, | |
| "learning_rate": 3.971580253443951e-05, | |
| "loss": 1.012, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.08333333333333333, | |
| "grad_norm": 0.01442610565572977, | |
| "learning_rate": 3.970819201700699e-05, | |
| "loss": 1.0334, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.08402777777777778, | |
| "grad_norm": 0.015393883921205997, | |
| "learning_rate": 3.970048168924009e-05, | |
| "loss": 0.9252, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.08472222222222223, | |
| "grad_norm": 0.01524246297776699, | |
| "learning_rate": 3.9692671590187093e-05, | |
| "loss": 1.0209, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.08541666666666667, | |
| "grad_norm": 0.014700529165565968, | |
| "learning_rate": 3.96847617594015e-05, | |
| "loss": 0.9943, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.08611111111111111, | |
| "grad_norm": 0.014305293560028076, | |
| "learning_rate": 3.967675223694193e-05, | |
| "loss": 0.966, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.08680555555555555, | |
| "grad_norm": 0.017354033887386322, | |
| "learning_rate": 3.966864306337189e-05, | |
| "loss": 0.9938, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.0875, | |
| "grad_norm": 0.014640097506344318, | |
| "learning_rate": 3.9660434279759536e-05, | |
| "loss": 1.0317, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.08819444444444445, | |
| "grad_norm": 0.014592519961297512, | |
| "learning_rate": 3.965212592767751e-05, | |
| "loss": 1.002, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.08888888888888889, | |
| "grad_norm": 0.01544480212032795, | |
| "learning_rate": 3.964371804920269e-05, | |
| "loss": 0.9818, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.08958333333333333, | |
| "grad_norm": 0.016679253429174423, | |
| "learning_rate": 3.9635210686916016e-05, | |
| "loss": 0.9963, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.09027777777777778, | |
| "grad_norm": 0.015196239575743675, | |
| "learning_rate": 3.962660388390224e-05, | |
| "loss": 1.017, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.09097222222222222, | |
| "grad_norm": 0.01522803958505392, | |
| "learning_rate": 3.9617897683749726e-05, | |
| "loss": 1.0141, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.09166666666666666, | |
| "grad_norm": 0.014725332148373127, | |
| "learning_rate": 3.960909213055023e-05, | |
| "loss": 1.0029, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.09236111111111112, | |
| "grad_norm": 0.015328467823565006, | |
| "learning_rate": 3.960018726889868e-05, | |
| "loss": 1.0228, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.09305555555555556, | |
| "grad_norm": 0.014145839028060436, | |
| "learning_rate": 3.959118314389291e-05, | |
| "loss": 0.9709, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.09375, | |
| "grad_norm": 0.015266870148479939, | |
| "learning_rate": 3.958207980113351e-05, | |
| "loss": 0.9938, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.09444444444444444, | |
| "grad_norm": 0.014660796150565147, | |
| "learning_rate": 3.957287728672352e-05, | |
| "loss": 0.9958, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.09513888888888888, | |
| "grad_norm": 0.015213273465633392, | |
| "learning_rate": 3.956357564726822e-05, | |
| "loss": 1.011, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.09583333333333334, | |
| "grad_norm": 0.015523552894592285, | |
| "learning_rate": 3.955417492987493e-05, | |
| "loss": 1.0297, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.09652777777777778, | |
| "grad_norm": 0.013940747827291489, | |
| "learning_rate": 3.95446751821527e-05, | |
| "loss": 1.0099, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.09722222222222222, | |
| "grad_norm": 0.015171276405453682, | |
| "learning_rate": 3.9535076452212156e-05, | |
| "loss": 0.979, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.09791666666666667, | |
| "grad_norm": 0.014734701253473759, | |
| "learning_rate": 3.952537878866517e-05, | |
| "loss": 1.0203, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.09861111111111111, | |
| "grad_norm": 0.014895283617079258, | |
| "learning_rate": 3.951558224062469e-05, | |
| "loss": 0.9803, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.09930555555555555, | |
| "grad_norm": 0.015903517603874207, | |
| "learning_rate": 3.9505686857704416e-05, | |
| "loss": 1.0109, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.018048502504825592, | |
| "learning_rate": 3.949569269001861e-05, | |
| "loss": 1.0263, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.10069444444444445, | |
| "grad_norm": 0.014940551482141018, | |
| "learning_rate": 3.948559978818184e-05, | |
| "loss": 1.0167, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.10138888888888889, | |
| "grad_norm": 0.018485594540834427, | |
| "learning_rate": 3.947540820330867e-05, | |
| "loss": 1.021, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.10208333333333333, | |
| "grad_norm": 0.015115809626877308, | |
| "learning_rate": 3.9465117987013445e-05, | |
| "loss": 1.0245, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.10277777777777777, | |
| "grad_norm": 0.016594298183918, | |
| "learning_rate": 3.9454729191410025e-05, | |
| "loss": 0.9967, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.10347222222222222, | |
| "grad_norm": 0.015151984058320522, | |
| "learning_rate": 3.944424186911152e-05, | |
| "loss": 0.9781, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.10416666666666667, | |
| "grad_norm": 0.016983836889266968, | |
| "learning_rate": 3.943365607323001e-05, | |
| "loss": 0.975, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.10486111111111111, | |
| "grad_norm": 0.015473946928977966, | |
| "learning_rate": 3.9422971857376296e-05, | |
| "loss": 1.0259, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.10555555555555556, | |
| "grad_norm": 0.014182898215949535, | |
| "learning_rate": 3.941218927565959e-05, | |
| "loss": 1.0058, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.10625, | |
| "grad_norm": 0.014923619106411934, | |
| "learning_rate": 3.9401308382687324e-05, | |
| "loss": 0.9821, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.10694444444444444, | |
| "grad_norm": 0.015497357584536076, | |
| "learning_rate": 3.9390329233564755e-05, | |
| "loss": 1.0599, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.1076388888888889, | |
| "grad_norm": 0.01431487686932087, | |
| "learning_rate": 3.9379251883894795e-05, | |
| "loss": 1.0193, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.10833333333333334, | |
| "grad_norm": 0.014660484157502651, | |
| "learning_rate": 3.9368076389777655e-05, | |
| "loss": 0.9527, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.10902777777777778, | |
| "grad_norm": 0.01643138937652111, | |
| "learning_rate": 3.9356802807810613e-05, | |
| "loss": 1.0134, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.10972222222222222, | |
| "grad_norm": 0.016061700880527496, | |
| "learning_rate": 3.934543119508769e-05, | |
| "loss": 1.0064, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.11041666666666666, | |
| "grad_norm": 0.01461365632712841, | |
| "learning_rate": 3.933396160919938e-05, | |
| "loss": 0.9953, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.1111111111111111, | |
| "grad_norm": 0.01574275828897953, | |
| "learning_rate": 3.932239410823233e-05, | |
| "loss": 1.0223, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.11180555555555556, | |
| "grad_norm": 0.015788385644555092, | |
| "learning_rate": 3.931072875076912e-05, | |
| "loss": 0.9739, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.1125, | |
| "grad_norm": 0.014799153432250023, | |
| "learning_rate": 3.929896559588786e-05, | |
| "loss": 1.0028, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.11319444444444444, | |
| "grad_norm": 0.013463880866765976, | |
| "learning_rate": 3.9287104703162005e-05, | |
| "loss": 0.9771, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.11388888888888889, | |
| "grad_norm": 0.014836735092103481, | |
| "learning_rate": 3.927514613265992e-05, | |
| "loss": 1.0159, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.11458333333333333, | |
| "grad_norm": 0.013634877279400826, | |
| "learning_rate": 3.9263089944944715e-05, | |
| "loss": 1.0431, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.11527777777777778, | |
| "grad_norm": 0.013801947236061096, | |
| "learning_rate": 3.925093620107384e-05, | |
| "loss": 0.9892, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.11597222222222223, | |
| "grad_norm": 0.015041469596326351, | |
| "learning_rate": 3.923868496259882e-05, | |
| "loss": 1.018, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.11666666666666667, | |
| "grad_norm": 0.013934548944234848, | |
| "learning_rate": 3.9226336291564935e-05, | |
| "loss": 0.9945, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.11736111111111111, | |
| "grad_norm": 0.014555184170603752, | |
| "learning_rate": 3.92138902505109e-05, | |
| "loss": 1.0101, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.11805555555555555, | |
| "grad_norm": 0.014088758267462254, | |
| "learning_rate": 3.920134690246854e-05, | |
| "loss": 0.9859, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.11875, | |
| "grad_norm": 0.013208562508225441, | |
| "learning_rate": 3.91887063109625e-05, | |
| "loss": 0.9752, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.11944444444444445, | |
| "grad_norm": 0.015783201903104782, | |
| "learning_rate": 3.91759685400099e-05, | |
| "loss": 1.0432, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.12013888888888889, | |
| "grad_norm": 0.014122221618890762, | |
| "learning_rate": 3.916313365412002e-05, | |
| "loss": 1.0079, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.12083333333333333, | |
| "grad_norm": 0.014722524210810661, | |
| "learning_rate": 3.915020171829395e-05, | |
| "loss": 1.0195, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.12152777777777778, | |
| "grad_norm": 0.016117624938488007, | |
| "learning_rate": 3.9137172798024304e-05, | |
| "loss": 0.9712, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.12222222222222222, | |
| "grad_norm": 0.015223621390759945, | |
| "learning_rate": 3.912404695929486e-05, | |
| "loss": 1.0116, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.12291666666666666, | |
| "grad_norm": 0.013804764486849308, | |
| "learning_rate": 3.9110824268580206e-05, | |
| "loss": 0.929, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.12361111111111112, | |
| "grad_norm": 0.014369015581905842, | |
| "learning_rate": 3.909750479284548e-05, | |
| "loss": 0.983, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.12430555555555556, | |
| "grad_norm": 0.015310019254684448, | |
| "learning_rate": 3.908408859954593e-05, | |
| "loss": 0.9962, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.125, | |
| "grad_norm": 0.015282726846635342, | |
| "learning_rate": 3.907057575662663e-05, | |
| "loss": 1.0367, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.12569444444444444, | |
| "grad_norm": 0.013963599689304829, | |
| "learning_rate": 3.905696633252216e-05, | |
| "loss": 0.9506, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.12638888888888888, | |
| "grad_norm": 0.015924159437417984, | |
| "learning_rate": 3.904326039615618e-05, | |
| "loss": 0.9647, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.12708333333333333, | |
| "grad_norm": 0.01227070577442646, | |
| "learning_rate": 3.902945801694117e-05, | |
| "loss": 0.9613, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.12777777777777777, | |
| "grad_norm": 0.014005225151777267, | |
| "learning_rate": 3.9015559264777994e-05, | |
| "loss": 0.9774, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.1284722222222222, | |
| "grad_norm": 0.014159591868519783, | |
| "learning_rate": 3.9001564210055624e-05, | |
| "loss": 0.9962, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.12916666666666668, | |
| "grad_norm": 0.013430262915790081, | |
| "learning_rate": 3.898747292365073e-05, | |
| "loss": 0.9928, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.12986111111111112, | |
| "grad_norm": 0.012756455689668655, | |
| "learning_rate": 3.897328547692735e-05, | |
| "loss": 0.9652, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.13055555555555556, | |
| "grad_norm": 0.015474308282136917, | |
| "learning_rate": 3.89590019417365e-05, | |
| "loss": 1.0134, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.13125, | |
| "grad_norm": 0.013962026685476303, | |
| "learning_rate": 3.8944622390415835e-05, | |
| "loss": 1.0008, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.13194444444444445, | |
| "grad_norm": 0.013678076677024364, | |
| "learning_rate": 3.893014689578928e-05, | |
| "loss": 0.9865, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.1326388888888889, | |
| "grad_norm": 0.01606924645602703, | |
| "learning_rate": 3.891557553116665e-05, | |
| "loss": 1.0097, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.13333333333333333, | |
| "grad_norm": 0.014233720488846302, | |
| "learning_rate": 3.89009083703433e-05, | |
| "loss": 1.0165, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.13402777777777777, | |
| "grad_norm": 0.019174130633473396, | |
| "learning_rate": 3.888614548759971e-05, | |
| "loss": 1.0029, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.13472222222222222, | |
| "grad_norm": 0.014648901298642159, | |
| "learning_rate": 3.8871286957701146e-05, | |
| "loss": 0.972, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.13541666666666666, | |
| "grad_norm": 0.015734290704131126, | |
| "learning_rate": 3.8856332855897286e-05, | |
| "loss": 0.9562, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.1361111111111111, | |
| "grad_norm": 0.014715912751853466, | |
| "learning_rate": 3.8841283257921794e-05, | |
| "loss": 0.9801, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.13680555555555557, | |
| "grad_norm": 0.0156076829880476, | |
| "learning_rate": 3.882613823999199e-05, | |
| "loss": 1.0438, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.1375, | |
| "grad_norm": 0.014719419181346893, | |
| "learning_rate": 3.881089787880843e-05, | |
| "loss": 1.0208, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.13819444444444445, | |
| "grad_norm": 0.015471681021153927, | |
| "learning_rate": 3.879556225155453e-05, | |
| "loss": 0.9728, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.1388888888888889, | |
| "grad_norm": 0.013889987021684647, | |
| "learning_rate": 3.878013143589617e-05, | |
| "loss": 0.9525, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.13958333333333334, | |
| "grad_norm": 0.014045029878616333, | |
| "learning_rate": 3.87646055099813e-05, | |
| "loss": 0.9568, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.14027777777777778, | |
| "grad_norm": 0.015043736435472965, | |
| "learning_rate": 3.874898455243955e-05, | |
| "loss": 0.9852, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.14097222222222222, | |
| "grad_norm": 0.014233953319489956, | |
| "learning_rate": 3.873326864238183e-05, | |
| "loss": 1.0112, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.14166666666666666, | |
| "grad_norm": 0.014034410007297993, | |
| "learning_rate": 3.871745785939993e-05, | |
| "loss": 0.959, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.1423611111111111, | |
| "grad_norm": 0.013695859350264072, | |
| "learning_rate": 3.870155228356611e-05, | |
| "loss": 0.9657, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.14305555555555555, | |
| "grad_norm": 0.01533182617276907, | |
| "learning_rate": 3.8685551995432713e-05, | |
| "loss": 1.0051, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.14375, | |
| "grad_norm": 0.014852087013423443, | |
| "learning_rate": 3.866945707603172e-05, | |
| "loss": 1.0207, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.14444444444444443, | |
| "grad_norm": 0.014271554537117481, | |
| "learning_rate": 3.8653267606874395e-05, | |
| "loss": 1.0129, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.1451388888888889, | |
| "grad_norm": 0.013892865739762783, | |
| "learning_rate": 3.863698366995079e-05, | |
| "loss": 1.0039, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.14583333333333334, | |
| "grad_norm": 0.01436774991452694, | |
| "learning_rate": 3.862060534772945e-05, | |
| "loss": 0.9811, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.14652777777777778, | |
| "grad_norm": 0.014014553278684616, | |
| "learning_rate": 3.8604132723156865e-05, | |
| "loss": 0.9575, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.14722222222222223, | |
| "grad_norm": 0.012928716838359833, | |
| "learning_rate": 3.858756587965714e-05, | |
| "loss": 0.9725, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.14791666666666667, | |
| "grad_norm": 0.014141298830509186, | |
| "learning_rate": 3.8570904901131544e-05, | |
| "loss": 0.9553, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.1486111111111111, | |
| "grad_norm": 0.014526846818625927, | |
| "learning_rate": 3.855414987195807e-05, | |
| "loss": 0.9812, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.14930555555555555, | |
| "grad_norm": 0.012949762865900993, | |
| "learning_rate": 3.853730087699103e-05, | |
| "loss": 0.9676, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 0.013820677995681763, | |
| "learning_rate": 3.852035800156062e-05, | |
| "loss": 1.0086, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.15069444444444444, | |
| "grad_norm": 0.013385163620114326, | |
| "learning_rate": 3.850332133147248e-05, | |
| "loss": 0.9818, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.15138888888888888, | |
| "grad_norm": 0.012220976874232292, | |
| "learning_rate": 3.848619095300726e-05, | |
| "loss": 0.96, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.15208333333333332, | |
| "grad_norm": 0.01417449489235878, | |
| "learning_rate": 3.8468966952920205e-05, | |
| "loss": 1.0152, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.1527777777777778, | |
| "grad_norm": 0.013405581004917622, | |
| "learning_rate": 3.845164941844068e-05, | |
| "loss": 1.0077, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.15347222222222223, | |
| "grad_norm": 0.01379991602152586, | |
| "learning_rate": 3.843423843727176e-05, | |
| "loss": 0.9917, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.15416666666666667, | |
| "grad_norm": 0.014063477516174316, | |
| "learning_rate": 3.8416734097589756e-05, | |
| "loss": 0.9679, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.15486111111111112, | |
| "grad_norm": 0.012324227020144463, | |
| "learning_rate": 3.83991364880438e-05, | |
| "loss": 0.9574, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.15555555555555556, | |
| "grad_norm": 0.01431061141192913, | |
| "learning_rate": 3.8381445697755365e-05, | |
| "loss": 0.9967, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.15625, | |
| "grad_norm": 0.014298739843070507, | |
| "learning_rate": 3.836366181631785e-05, | |
| "loss": 0.9815, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.15694444444444444, | |
| "grad_norm": 0.012113712728023529, | |
| "learning_rate": 3.8345784933796095e-05, | |
| "loss": 0.9337, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.15763888888888888, | |
| "grad_norm": 0.015284246765077114, | |
| "learning_rate": 3.832781514072593e-05, | |
| "loss": 1.0051, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.15833333333333333, | |
| "grad_norm": 0.012948422692716122, | |
| "learning_rate": 3.8309752528113725e-05, | |
| "loss": 0.9554, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.15902777777777777, | |
| "grad_norm": 0.014634348452091217, | |
| "learning_rate": 3.8291597187435926e-05, | |
| "loss": 0.9814, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.1597222222222222, | |
| "grad_norm": 0.014931918121874332, | |
| "learning_rate": 3.827334921063861e-05, | |
| "loss": 1.0053, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.16041666666666668, | |
| "grad_norm": 0.012998196296393871, | |
| "learning_rate": 3.825500869013697e-05, | |
| "loss": 0.9781, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.16111111111111112, | |
| "grad_norm": 0.015014944598078728, | |
| "learning_rate": 3.82365757188149e-05, | |
| "loss": 1.014, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.16180555555555556, | |
| "grad_norm": 0.015584563836455345, | |
| "learning_rate": 3.821805039002449e-05, | |
| "loss": 0.9788, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.1625, | |
| "grad_norm": 0.013651632703840733, | |
| "learning_rate": 3.8199432797585576e-05, | |
| "loss": 0.9905, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.16319444444444445, | |
| "grad_norm": 0.015139803290367126, | |
| "learning_rate": 3.8180723035785246e-05, | |
| "loss": 1.0362, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.1638888888888889, | |
| "grad_norm": 0.014169261790812016, | |
| "learning_rate": 3.816192119937738e-05, | |
| "loss": 0.9688, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.16458333333333333, | |
| "grad_norm": 0.013432355597615242, | |
| "learning_rate": 3.814302738358216e-05, | |
| "loss": 0.955, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.16527777777777777, | |
| "grad_norm": 0.012756401672959328, | |
| "learning_rate": 3.812404168408558e-05, | |
| "loss": 0.9817, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.16597222222222222, | |
| "grad_norm": 0.012755703181028366, | |
| "learning_rate": 3.810496419703898e-05, | |
| "loss": 0.9785, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.16666666666666666, | |
| "grad_norm": 0.014174016192555428, | |
| "learning_rate": 3.808579501905856e-05, | |
| "loss": 0.9704, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.1673611111111111, | |
| "grad_norm": 0.013185903429985046, | |
| "learning_rate": 3.806653424722484e-05, | |
| "loss": 0.9805, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.16805555555555557, | |
| "grad_norm": 0.014006533659994602, | |
| "learning_rate": 3.804718197908225e-05, | |
| "loss": 0.9846, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.16875, | |
| "grad_norm": 0.01497966144233942, | |
| "learning_rate": 3.802773831263859e-05, | |
| "loss": 0.9697, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.16944444444444445, | |
| "grad_norm": 0.0131217110902071, | |
| "learning_rate": 3.800820334636452e-05, | |
| "loss": 1.0037, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.1701388888888889, | |
| "grad_norm": 0.012915054336190224, | |
| "learning_rate": 3.798857717919308e-05, | |
| "loss": 0.9489, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.17083333333333334, | |
| "grad_norm": 0.013334513641893864, | |
| "learning_rate": 3.7968859910519215e-05, | |
| "loss": 0.9891, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.17152777777777778, | |
| "grad_norm": 0.014228935353457928, | |
| "learning_rate": 3.7949051640199216e-05, | |
| "loss": 1.0013, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.17222222222222222, | |
| "grad_norm": 0.013772614300251007, | |
| "learning_rate": 3.7929152468550245e-05, | |
| "loss": 1.0042, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.17291666666666666, | |
| "grad_norm": 0.01312232669442892, | |
| "learning_rate": 3.790916249634986e-05, | |
| "loss": 0.9374, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.1736111111111111, | |
| "grad_norm": 0.014259965158998966, | |
| "learning_rate": 3.788908182483542e-05, | |
| "loss": 0.9432, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.17430555555555555, | |
| "grad_norm": 0.014950071461498737, | |
| "learning_rate": 3.786891055570367e-05, | |
| "loss": 0.9674, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.175, | |
| "grad_norm": 0.014557240530848503, | |
| "learning_rate": 3.784864879111013e-05, | |
| "loss": 0.9811, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.17569444444444443, | |
| "grad_norm": 0.011923140846192837, | |
| "learning_rate": 3.782829663366868e-05, | |
| "loss": 0.9773, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.1763888888888889, | |
| "grad_norm": 0.012792705558240414, | |
| "learning_rate": 3.780785418645095e-05, | |
| "loss": 0.9635, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.17708333333333334, | |
| "grad_norm": 0.01456011924892664, | |
| "learning_rate": 3.7787321552985826e-05, | |
| "loss": 0.9271, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.17777777777777778, | |
| "grad_norm": 0.014588495716452599, | |
| "learning_rate": 3.776669883725897e-05, | |
| "loss": 0.9818, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.17847222222222223, | |
| "grad_norm": 0.013889756053686142, | |
| "learning_rate": 3.774598614371223e-05, | |
| "loss": 0.9894, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.17916666666666667, | |
| "grad_norm": 0.014536023139953613, | |
| "learning_rate": 3.7725183577243144e-05, | |
| "loss": 0.9675, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.1798611111111111, | |
| "grad_norm": 0.01632644794881344, | |
| "learning_rate": 3.7704291243204416e-05, | |
| "loss": 1.0065, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.18055555555555555, | |
| "grad_norm": 0.013067888095974922, | |
| "learning_rate": 3.7683309247403345e-05, | |
| "loss": 1.0002, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.18125, | |
| "grad_norm": 0.01566268689930439, | |
| "learning_rate": 3.766223769610134e-05, | |
| "loss": 0.9552, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.18194444444444444, | |
| "grad_norm": 0.01552093680948019, | |
| "learning_rate": 3.764107669601333e-05, | |
| "loss": 1.0219, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.18263888888888888, | |
| "grad_norm": 0.013710713014006615, | |
| "learning_rate": 3.761982635430729e-05, | |
| "loss": 0.9559, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.18333333333333332, | |
| "grad_norm": 0.013827464543282986, | |
| "learning_rate": 3.759848677860359e-05, | |
| "loss": 0.9741, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.1840277777777778, | |
| "grad_norm": 0.01393877249211073, | |
| "learning_rate": 3.7577058076974595e-05, | |
| "loss": 0.9841, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.18472222222222223, | |
| "grad_norm": 0.014325949363410473, | |
| "learning_rate": 3.755554035794398e-05, | |
| "loss": 1.0068, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.18541666666666667, | |
| "grad_norm": 0.013421372510492802, | |
| "learning_rate": 3.7533933730486276e-05, | |
| "loss": 0.9409, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.18611111111111112, | |
| "grad_norm": 0.015207280404865742, | |
| "learning_rate": 3.751223830402627e-05, | |
| "loss": 1.0078, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.18680555555555556, | |
| "grad_norm": 0.015149595215916634, | |
| "learning_rate": 3.749045418843844e-05, | |
| "loss": 0.9989, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.1875, | |
| "grad_norm": 0.012723335064947605, | |
| "learning_rate": 3.746858149404648e-05, | |
| "loss": 0.9656, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.18819444444444444, | |
| "grad_norm": 0.01434493437409401, | |
| "learning_rate": 3.744662033162262e-05, | |
| "loss": 0.9725, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.18888888888888888, | |
| "grad_norm": 0.01342215109616518, | |
| "learning_rate": 3.742457081238717e-05, | |
| "loss": 0.9555, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.18958333333333333, | |
| "grad_norm": 0.012995844706892967, | |
| "learning_rate": 3.740243304800791e-05, | |
| "loss": 0.9775, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.19027777777777777, | |
| "grad_norm": 0.01189506333321333, | |
| "learning_rate": 3.738020715059951e-05, | |
| "loss": 0.9443, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.1909722222222222, | |
| "grad_norm": 0.013390977866947651, | |
| "learning_rate": 3.7357893232722984e-05, | |
| "loss": 0.9675, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.19166666666666668, | |
| "grad_norm": 0.01335330493748188, | |
| "learning_rate": 3.733549140738514e-05, | |
| "loss": 0.9596, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.19236111111111112, | |
| "grad_norm": 0.012211819179356098, | |
| "learning_rate": 3.731300178803797e-05, | |
| "loss": 0.9413, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.19305555555555556, | |
| "grad_norm": 0.013362012803554535, | |
| "learning_rate": 3.7290424488578094e-05, | |
| "loss": 0.9622, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.19375, | |
| "grad_norm": 0.013488608412444592, | |
| "learning_rate": 3.726775962334617e-05, | |
| "loss": 1.0322, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.19444444444444445, | |
| "grad_norm": 0.013000058010220528, | |
| "learning_rate": 3.7245007307126346e-05, | |
| "loss": 0.9966, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.1951388888888889, | |
| "grad_norm": 0.014398259110748768, | |
| "learning_rate": 3.7222167655145636e-05, | |
| "loss": 0.9775, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.19583333333333333, | |
| "grad_norm": 0.012063449248671532, | |
| "learning_rate": 3.7199240783073365e-05, | |
| "loss": 0.9369, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.19652777777777777, | |
| "grad_norm": 0.012797530740499496, | |
| "learning_rate": 3.717622680702058e-05, | |
| "loss": 0.9591, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.19722222222222222, | |
| "grad_norm": 0.012506108731031418, | |
| "learning_rate": 3.7153125843539455e-05, | |
| "loss": 0.9781, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.19791666666666666, | |
| "grad_norm": 0.013660159893333912, | |
| "learning_rate": 3.712993800962271e-05, | |
| "loss": 0.9994, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.1986111111111111, | |
| "grad_norm": 0.013593221083283424, | |
| "learning_rate": 3.7106663422703004e-05, | |
| "loss": 0.9945, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.19930555555555557, | |
| "grad_norm": 0.013533521443605423, | |
| "learning_rate": 3.708330220065235e-05, | |
| "loss": 1.0033, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.015442555770277977, | |
| "learning_rate": 3.7059854461781514e-05, | |
| "loss": 0.9829, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.20069444444444445, | |
| "grad_norm": 0.014286418445408344, | |
| "learning_rate": 3.7036320324839444e-05, | |
| "loss": 0.9674, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.2013888888888889, | |
| "grad_norm": 0.013105183839797974, | |
| "learning_rate": 3.70126999090126e-05, | |
| "loss": 0.9562, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.20208333333333334, | |
| "grad_norm": 0.012748828157782555, | |
| "learning_rate": 3.698899333392442e-05, | |
| "loss": 0.9385, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.20277777777777778, | |
| "grad_norm": 0.013904962688684464, | |
| "learning_rate": 3.696520071963469e-05, | |
| "loss": 0.9769, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.20347222222222222, | |
| "grad_norm": 0.013301237486302853, | |
| "learning_rate": 3.6941322186638924e-05, | |
| "loss": 0.9582, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.20416666666666666, | |
| "grad_norm": 0.013818389736115932, | |
| "learning_rate": 3.691735785586777e-05, | |
| "loss": 0.9992, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.2048611111111111, | |
| "grad_norm": 0.018090086057782173, | |
| "learning_rate": 3.6893307848686376e-05, | |
| "loss": 1.0265, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.20555555555555555, | |
| "grad_norm": 0.011723886243999004, | |
| "learning_rate": 3.686917228689382e-05, | |
| "loss": 0.9447, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.20625, | |
| "grad_norm": 0.013388545252382755, | |
| "learning_rate": 3.684495129272242e-05, | |
| "loss": 0.9725, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.20694444444444443, | |
| "grad_norm": 0.013546598143875599, | |
| "learning_rate": 3.682064498883721e-05, | |
| "loss": 0.9867, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.2076388888888889, | |
| "grad_norm": 0.01278660912066698, | |
| "learning_rate": 3.679625349833523e-05, | |
| "loss": 0.9533, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.20833333333333334, | |
| "grad_norm": 0.013434414751827717, | |
| "learning_rate": 3.677177694474495e-05, | |
| "loss": 0.9587, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.20902777777777778, | |
| "grad_norm": 0.013443008065223694, | |
| "learning_rate": 3.674721545202563e-05, | |
| "loss": 0.9758, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.20972222222222223, | |
| "grad_norm": 0.012249941006302834, | |
| "learning_rate": 3.672256914456671e-05, | |
| "loss": 0.9391, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.21041666666666667, | |
| "grad_norm": 0.013929427601397038, | |
| "learning_rate": 3.669783814718716e-05, | |
| "loss": 0.966, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.2111111111111111, | |
| "grad_norm": 0.020106054842472076, | |
| "learning_rate": 3.667302258513484e-05, | |
| "loss": 0.9916, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.21180555555555555, | |
| "grad_norm": 0.017338188365101814, | |
| "learning_rate": 3.6648122584085907e-05, | |
| "loss": 0.9905, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.2125, | |
| "grad_norm": 0.016092827543616295, | |
| "learning_rate": 3.662313827014412e-05, | |
| "loss": 0.9422, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.21319444444444444, | |
| "grad_norm": 0.014228146523237228, | |
| "learning_rate": 3.659806976984026e-05, | |
| "loss": 1.0033, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.21388888888888888, | |
| "grad_norm": 0.015558873303234577, | |
| "learning_rate": 3.657291721013146e-05, | |
| "loss": 0.9775, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.21458333333333332, | |
| "grad_norm": 0.016239987686276436, | |
| "learning_rate": 3.6547680718400546e-05, | |
| "loss": 0.9863, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.2152777777777778, | |
| "grad_norm": 0.014311805367469788, | |
| "learning_rate": 3.652236042245542e-05, | |
| "loss": 1.0078, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.21597222222222223, | |
| "grad_norm": 0.014693599194288254, | |
| "learning_rate": 3.64969564505284e-05, | |
| "loss": 0.9743, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.21666666666666667, | |
| "grad_norm": 0.014121411368250847, | |
| "learning_rate": 3.647146893127559e-05, | |
| "loss": 0.9712, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.21736111111111112, | |
| "grad_norm": 0.013323220424354076, | |
| "learning_rate": 3.644589799377618e-05, | |
| "loss": 0.9912, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.21805555555555556, | |
| "grad_norm": 0.014195873402059078, | |
| "learning_rate": 3.642024376753186e-05, | |
| "loss": 0.9733, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.21875, | |
| "grad_norm": 0.01346920058131218, | |
| "learning_rate": 3.639450638246611e-05, | |
| "loss": 0.9562, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.21944444444444444, | |
| "grad_norm": 0.013232593424618244, | |
| "learning_rate": 3.636868596892353e-05, | |
| "loss": 0.9724, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.22013888888888888, | |
| "grad_norm": 0.01408706046640873, | |
| "learning_rate": 3.634278265766928e-05, | |
| "loss": 1.009, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.22083333333333333, | |
| "grad_norm": 0.013279229402542114, | |
| "learning_rate": 3.631679657988829e-05, | |
| "loss": 0.9757, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.22152777777777777, | |
| "grad_norm": 0.015104196034371853, | |
| "learning_rate": 3.6290727867184676e-05, | |
| "loss": 1.001, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.2222222222222222, | |
| "grad_norm": 0.012325095012784004, | |
| "learning_rate": 3.6264576651581036e-05, | |
| "loss": 0.9758, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.22291666666666668, | |
| "grad_norm": 0.013523522764444351, | |
| "learning_rate": 3.623834306551782e-05, | |
| "loss": 0.9482, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.22361111111111112, | |
| "grad_norm": 0.01278225053101778, | |
| "learning_rate": 3.6212027241852625e-05, | |
| "loss": 0.9631, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.22430555555555556, | |
| "grad_norm": 0.012586407363414764, | |
| "learning_rate": 3.618562931385952e-05, | |
| "loss": 1.0226, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.225, | |
| "grad_norm": 0.012502876110374928, | |
| "learning_rate": 3.6159149415228403e-05, | |
| "loss": 0.9229, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.22569444444444445, | |
| "grad_norm": 0.012781070545315742, | |
| "learning_rate": 3.613258768006429e-05, | |
| "loss": 0.9532, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.2263888888888889, | |
| "grad_norm": 0.013412564061582088, | |
| "learning_rate": 3.6105944242886674e-05, | |
| "loss": 1.0037, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.22708333333333333, | |
| "grad_norm": 0.013194269500672817, | |
| "learning_rate": 3.607921923862878e-05, | |
| "loss": 0.9241, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.22777777777777777, | |
| "grad_norm": 0.013584061525762081, | |
| "learning_rate": 3.605241280263696e-05, | |
| "loss": 0.9482, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.22847222222222222, | |
| "grad_norm": 0.013534624129533768, | |
| "learning_rate": 3.6025525070669955e-05, | |
| "loss": 0.9653, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.22916666666666666, | |
| "grad_norm": 0.015827780589461327, | |
| "learning_rate": 3.599855617889822e-05, | |
| "loss": 0.9511, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.2298611111111111, | |
| "grad_norm": 0.013115919195115566, | |
| "learning_rate": 3.5971506263903234e-05, | |
| "loss": 0.9516, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.23055555555555557, | |
| "grad_norm": 0.015311370603740215, | |
| "learning_rate": 3.594437546267682e-05, | |
| "loss": 0.9543, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.23125, | |
| "grad_norm": 0.014090127311646938, | |
| "learning_rate": 3.591716391262044e-05, | |
| "loss": 0.9914, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.23194444444444445, | |
| "grad_norm": 0.013163258321583271, | |
| "learning_rate": 3.588987175154449e-05, | |
| "loss": 0.9722, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.2326388888888889, | |
| "grad_norm": 0.012093394063413143, | |
| "learning_rate": 3.586249911766763e-05, | |
| "loss": 0.9758, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.23333333333333334, | |
| "grad_norm": 0.014174525626003742, | |
| "learning_rate": 3.583504614961605e-05, | |
| "loss": 0.9683, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.23402777777777778, | |
| "grad_norm": 0.014316610060632229, | |
| "learning_rate": 3.58075129864228e-05, | |
| "loss": 0.9814, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.23472222222222222, | |
| "grad_norm": 0.013283569365739822, | |
| "learning_rate": 3.5779899767527064e-05, | |
| "loss": 1.0012, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.23541666666666666, | |
| "grad_norm": 0.013545655645430088, | |
| "learning_rate": 3.575220663277346e-05, | |
| "loss": 0.9426, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.2361111111111111, | |
| "grad_norm": 0.01365567371249199, | |
| "learning_rate": 3.572443372241134e-05, | |
| "loss": 0.957, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.23680555555555555, | |
| "grad_norm": 0.012532095424830914, | |
| "learning_rate": 3.569658117709406e-05, | |
| "loss": 0.9439, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.2375, | |
| "grad_norm": 0.012797664850950241, | |
| "learning_rate": 3.5668649137878275e-05, | |
| "loss": 0.9805, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.23819444444444443, | |
| "grad_norm": 0.013005426153540611, | |
| "learning_rate": 3.5640637746223253e-05, | |
| "loss": 0.9515, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.2388888888888889, | |
| "grad_norm": 0.013586447574198246, | |
| "learning_rate": 3.561254714399013e-05, | |
| "loss": 0.9731, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.23958333333333334, | |
| "grad_norm": 0.013848516158759594, | |
| "learning_rate": 3.5584377473441187e-05, | |
| "loss": 0.9731, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.24027777777777778, | |
| "grad_norm": 0.014170484617352486, | |
| "learning_rate": 3.5556128877239125e-05, | |
| "loss": 0.9817, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.24097222222222223, | |
| "grad_norm": 0.014107048511505127, | |
| "learning_rate": 3.552780149844639e-05, | |
| "loss": 0.9733, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.24166666666666667, | |
| "grad_norm": 0.013129886239767075, | |
| "learning_rate": 3.5499395480524394e-05, | |
| "loss": 0.9803, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.2423611111111111, | |
| "grad_norm": 0.014618804678320885, | |
| "learning_rate": 3.5470910967332815e-05, | |
| "loss": 0.9941, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.24305555555555555, | |
| "grad_norm": 0.013775044120848179, | |
| "learning_rate": 3.544234810312886e-05, | |
| "loss": 0.9978, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.24375, | |
| "grad_norm": 0.013130244798958302, | |
| "learning_rate": 3.541370703256654e-05, | |
| "loss": 0.9978, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.24444444444444444, | |
| "grad_norm": 0.013368207029998302, | |
| "learning_rate": 3.538498790069594e-05, | |
| "loss": 0.9875, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.24513888888888888, | |
| "grad_norm": 0.014766732230782509, | |
| "learning_rate": 3.5356190852962474e-05, | |
| "loss": 0.9446, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.24583333333333332, | |
| "grad_norm": 0.012403284199535847, | |
| "learning_rate": 3.5327316035206145e-05, | |
| "loss": 0.9305, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.2465277777777778, | |
| "grad_norm": 0.013556853868067265, | |
| "learning_rate": 3.5298363593660835e-05, | |
| "loss": 1.0034, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.24722222222222223, | |
| "grad_norm": 0.013330447487533092, | |
| "learning_rate": 3.526933367495353e-05, | |
| "loss": 0.9721, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.24791666666666667, | |
| "grad_norm": 0.01214715838432312, | |
| "learning_rate": 3.5240226426103596e-05, | |
| "loss": 0.9462, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.24861111111111112, | |
| "grad_norm": 0.013268118724226952, | |
| "learning_rate": 3.521104199452203e-05, | |
| "loss": 0.9796, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.24930555555555556, | |
| "grad_norm": 0.012496302835643291, | |
| "learning_rate": 3.5181780528010715e-05, | |
| "loss": 0.9427, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 0.013434799388051033, | |
| "learning_rate": 3.515244217476166e-05, | |
| "loss": 0.9519, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.25069444444444444, | |
| "grad_norm": 0.012152577750384808, | |
| "learning_rate": 3.5123027083356285e-05, | |
| "loss": 0.9707, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.2513888888888889, | |
| "grad_norm": 0.013102750293910503, | |
| "learning_rate": 3.509353540276462e-05, | |
| "loss": 0.9594, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.2520833333333333, | |
| "grad_norm": 0.01402961928397417, | |
| "learning_rate": 3.506396728234459e-05, | |
| "loss": 0.9629, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.25277777777777777, | |
| "grad_norm": 0.013971925713121891, | |
| "learning_rate": 3.503432287184121e-05, | |
| "loss": 0.9559, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.2534722222222222, | |
| "grad_norm": 0.013591865077614784, | |
| "learning_rate": 3.500460232138591e-05, | |
| "loss": 0.9919, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.25416666666666665, | |
| "grad_norm": 0.013570788316428661, | |
| "learning_rate": 3.497480578149569e-05, | |
| "loss": 0.9537, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.2548611111111111, | |
| "grad_norm": 0.013885971158742905, | |
| "learning_rate": 3.4944933403072387e-05, | |
| "loss": 0.9678, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.25555555555555554, | |
| "grad_norm": 0.013096191920340061, | |
| "learning_rate": 3.491498533740193e-05, | |
| "loss": 0.9505, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.25625, | |
| "grad_norm": 0.013115983456373215, | |
| "learning_rate": 3.488496173615358e-05, | |
| "loss": 0.9467, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.2569444444444444, | |
| "grad_norm": 0.01301599945873022, | |
| "learning_rate": 3.485486275137909e-05, | |
| "loss": 0.9501, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.25763888888888886, | |
| "grad_norm": 0.012697902508080006, | |
| "learning_rate": 3.482468853551202e-05, | |
| "loss": 0.9749, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.25833333333333336, | |
| "grad_norm": 0.01309168990701437, | |
| "learning_rate": 3.479443924136693e-05, | |
| "loss": 0.9791, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.2590277777777778, | |
| "grad_norm": 0.012592338025569916, | |
| "learning_rate": 3.47641150221386e-05, | |
| "loss": 0.9726, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.25972222222222224, | |
| "grad_norm": 0.012716379016637802, | |
| "learning_rate": 3.473371603140125e-05, | |
| "loss": 0.9423, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.2604166666666667, | |
| "grad_norm": 0.01352920476347208, | |
| "learning_rate": 3.47032424231078e-05, | |
| "loss": 0.9829, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.2611111111111111, | |
| "grad_norm": 0.012491214089095592, | |
| "learning_rate": 3.4672694351589046e-05, | |
| "loss": 0.9633, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.26180555555555557, | |
| "grad_norm": 0.01319537591189146, | |
| "learning_rate": 3.46420719715529e-05, | |
| "loss": 0.9221, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.2625, | |
| "grad_norm": 0.013664872385561466, | |
| "learning_rate": 3.46113754380836e-05, | |
| "loss": 0.9621, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.26319444444444445, | |
| "grad_norm": 0.01348687894642353, | |
| "learning_rate": 3.4580604906640936e-05, | |
| "loss": 0.9634, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.2638888888888889, | |
| "grad_norm": 0.011980963870882988, | |
| "learning_rate": 3.454976053305943e-05, | |
| "loss": 0.9288, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.26458333333333334, | |
| "grad_norm": 0.01295421365648508, | |
| "learning_rate": 3.4518842473547614e-05, | |
| "loss": 0.9216, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.2652777777777778, | |
| "grad_norm": 0.014002040959894657, | |
| "learning_rate": 3.4487850884687155e-05, | |
| "loss": 0.9781, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.2659722222222222, | |
| "grad_norm": 0.012479842640459538, | |
| "learning_rate": 3.445678592343212e-05, | |
| "loss": 0.9305, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.26666666666666666, | |
| "grad_norm": 0.013758037239313126, | |
| "learning_rate": 3.442564774710816e-05, | |
| "loss": 0.9732, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.2673611111111111, | |
| "grad_norm": 0.01285907905548811, | |
| "learning_rate": 3.439443651341172e-05, | |
| "loss": 0.9494, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.26805555555555555, | |
| "grad_norm": 0.014414280652999878, | |
| "learning_rate": 3.436315238040924e-05, | |
| "loss": 0.9495, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.26875, | |
| "grad_norm": 0.013502138666808605, | |
| "learning_rate": 3.4331795506536336e-05, | |
| "loss": 0.924, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.26944444444444443, | |
| "grad_norm": 0.01338834036141634, | |
| "learning_rate": 3.430036605059704e-05, | |
| "loss": 0.9467, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.2701388888888889, | |
| "grad_norm": 0.014044429175555706, | |
| "learning_rate": 3.426886417176294e-05, | |
| "loss": 0.9976, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.2708333333333333, | |
| "grad_norm": 0.013052555732429028, | |
| "learning_rate": 3.423729002957244e-05, | |
| "loss": 0.9586, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.27152777777777776, | |
| "grad_norm": 0.012639213353395462, | |
| "learning_rate": 3.4205643783929875e-05, | |
| "loss": 0.9498, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.2722222222222222, | |
| "grad_norm": 0.012982922606170177, | |
| "learning_rate": 3.417392559510475e-05, | |
| "loss": 0.9844, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.27291666666666664, | |
| "grad_norm": 0.012655354104936123, | |
| "learning_rate": 3.4142135623730954e-05, | |
| "loss": 0.948, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.27361111111111114, | |
| "grad_norm": 0.013443528674542904, | |
| "learning_rate": 3.411027403080587e-05, | |
| "loss": 0.9682, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.2743055555555556, | |
| "grad_norm": 0.013121607713401318, | |
| "learning_rate": 3.407834097768962e-05, | |
| "loss": 1.0032, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.275, | |
| "grad_norm": 0.011974303051829338, | |
| "learning_rate": 3.4046336626104235e-05, | |
| "loss": 0.9615, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.27569444444444446, | |
| "grad_norm": 0.013200311921536922, | |
| "learning_rate": 3.401426113813282e-05, | |
| "loss": 0.9612, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.2763888888888889, | |
| "grad_norm": 0.011763915419578552, | |
| "learning_rate": 3.398211467621875e-05, | |
| "loss": 0.9523, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.27708333333333335, | |
| "grad_norm": 0.012212570756673813, | |
| "learning_rate": 3.394989740316484e-05, | |
| "loss": 0.9482, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.2777777777777778, | |
| "grad_norm": 0.012741866521537304, | |
| "learning_rate": 3.391760948213251e-05, | |
| "loss": 0.9787, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.27847222222222223, | |
| "grad_norm": 0.013035683892667294, | |
| "learning_rate": 3.388525107664099e-05, | |
| "loss": 0.9398, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.2791666666666667, | |
| "grad_norm": 0.013569427654147148, | |
| "learning_rate": 3.3852822350566455e-05, | |
| "loss": 0.9492, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.2798611111111111, | |
| "grad_norm": 0.012654770165681839, | |
| "learning_rate": 3.382032346814123e-05, | |
| "loss": 0.9762, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.28055555555555556, | |
| "grad_norm": 0.013426728546619415, | |
| "learning_rate": 3.378775459395292e-05, | |
| "loss": 0.9348, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.28125, | |
| "grad_norm": 0.01277459692209959, | |
| "learning_rate": 3.3755115892943616e-05, | |
| "loss": 0.9698, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.28194444444444444, | |
| "grad_norm": 0.013655406422913074, | |
| "learning_rate": 3.372240753040902e-05, | |
| "loss": 0.9792, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.2826388888888889, | |
| "grad_norm": 0.01355676632374525, | |
| "learning_rate": 3.368962967199765e-05, | |
| "loss": 0.9584, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.2833333333333333, | |
| "grad_norm": 0.013229291886091232, | |
| "learning_rate": 3.3656782483709966e-05, | |
| "loss": 0.9581, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.28402777777777777, | |
| "grad_norm": 0.01382211409509182, | |
| "learning_rate": 3.3623866131897554e-05, | |
| "loss": 0.9953, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.2847222222222222, | |
| "grad_norm": 0.012580876238644123, | |
| "learning_rate": 3.3590880783262245e-05, | |
| "loss": 0.9491, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.28541666666666665, | |
| "grad_norm": 0.014332804828882217, | |
| "learning_rate": 3.3557826604855335e-05, | |
| "loss": 0.9777, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.2861111111111111, | |
| "grad_norm": 0.014996136538684368, | |
| "learning_rate": 3.3524703764076684e-05, | |
| "loss": 0.949, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.28680555555555554, | |
| "grad_norm": 0.014224053360521793, | |
| "learning_rate": 3.3491512428673877e-05, | |
| "loss": 0.983, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.2875, | |
| "grad_norm": 0.014253957197070122, | |
| "learning_rate": 3.345825276674139e-05, | |
| "loss": 0.955, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.2881944444444444, | |
| "grad_norm": 0.0160871259868145, | |
| "learning_rate": 3.342492494671976e-05, | |
| "loss": 0.9732, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.28888888888888886, | |
| "grad_norm": 0.012912842445075512, | |
| "learning_rate": 3.339152913739466e-05, | |
| "loss": 0.9695, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.28958333333333336, | |
| "grad_norm": 0.013580222614109516, | |
| "learning_rate": 3.335806550789611e-05, | |
| "loss": 0.9974, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.2902777777777778, | |
| "grad_norm": 0.013716046698391438, | |
| "learning_rate": 3.332453422769762e-05, | |
| "loss": 0.9577, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.29097222222222224, | |
| "grad_norm": 0.013095886446535587, | |
| "learning_rate": 3.329093546661526e-05, | |
| "loss": 0.9624, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.2916666666666667, | |
| "grad_norm": 0.013227414339780807, | |
| "learning_rate": 3.3257269394806894e-05, | |
| "loss": 0.9823, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.2923611111111111, | |
| "grad_norm": 0.012513620778918266, | |
| "learning_rate": 3.322353618277126e-05, | |
| "loss": 0.9446, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.29305555555555557, | |
| "grad_norm": 0.01236687507480383, | |
| "learning_rate": 3.3189736001347125e-05, | |
| "loss": 0.9639, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.29375, | |
| "grad_norm": 0.013223248533904552, | |
| "learning_rate": 3.315586902171241e-05, | |
| "loss": 0.9436, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.29444444444444445, | |
| "grad_norm": 0.01373869925737381, | |
| "learning_rate": 3.3121935415383325e-05, | |
| "loss": 0.9751, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.2951388888888889, | |
| "grad_norm": 0.012828570790588856, | |
| "learning_rate": 3.308793535421353e-05, | |
| "loss": 0.98, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.29583333333333334, | |
| "grad_norm": 0.011885512620210648, | |
| "learning_rate": 3.305386901039321e-05, | |
| "loss": 0.9612, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.2965277777777778, | |
| "grad_norm": 0.011671421118080616, | |
| "learning_rate": 3.301973655644825e-05, | |
| "loss": 0.9269, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.2972222222222222, | |
| "grad_norm": 0.013502037152647972, | |
| "learning_rate": 3.298553816523934e-05, | |
| "loss": 0.9404, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.29791666666666666, | |
| "grad_norm": 0.013123046606779099, | |
| "learning_rate": 3.2951274009961094e-05, | |
| "loss": 0.9812, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.2986111111111111, | |
| "grad_norm": 0.012360905297100544, | |
| "learning_rate": 3.29169442641412e-05, | |
| "loss": 0.9355, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.29930555555555555, | |
| "grad_norm": 0.01276017352938652, | |
| "learning_rate": 3.288254910163951e-05, | |
| "loss": 0.9689, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 0.013017778284847736, | |
| "learning_rate": 3.284808869664716e-05, | |
| "loss": 0.9716, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.30069444444444443, | |
| "grad_norm": 0.011824820190668106, | |
| "learning_rate": 3.281356322368575e-05, | |
| "loss": 0.9633, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.3013888888888889, | |
| "grad_norm": 0.012792828492820263, | |
| "learning_rate": 3.277897285760635e-05, | |
| "loss": 0.9841, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.3020833333333333, | |
| "grad_norm": 0.012392980977892876, | |
| "learning_rate": 3.2744317773588696e-05, | |
| "loss": 0.9172, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.30277777777777776, | |
| "grad_norm": 0.013595100492238998, | |
| "learning_rate": 3.270959814714032e-05, | |
| "loss": 0.9827, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.3034722222222222, | |
| "grad_norm": 0.01175049040466547, | |
| "learning_rate": 3.267481415409557e-05, | |
| "loss": 0.9782, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.30416666666666664, | |
| "grad_norm": 0.013577048666775227, | |
| "learning_rate": 3.26399659706148e-05, | |
| "loss": 1.0069, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.30486111111111114, | |
| "grad_norm": 0.012351465411484241, | |
| "learning_rate": 3.260505377318344e-05, | |
| "loss": 0.9585, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.3055555555555556, | |
| "grad_norm": 0.013745117001235485, | |
| "learning_rate": 3.257007773861113e-05, | |
| "loss": 0.9942, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.30625, | |
| "grad_norm": 0.013106070458889008, | |
| "learning_rate": 3.253503804403079e-05, | |
| "loss": 0.9419, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.30694444444444446, | |
| "grad_norm": 0.013440150767564774, | |
| "learning_rate": 3.249993486689774e-05, | |
| "loss": 0.9428, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.3076388888888889, | |
| "grad_norm": 0.014590987004339695, | |
| "learning_rate": 3.246476838498881e-05, | |
| "loss": 0.9874, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.30833333333333335, | |
| "grad_norm": 0.013203262351453304, | |
| "learning_rate": 3.242953877640142e-05, | |
| "loss": 0.947, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.3090277777777778, | |
| "grad_norm": 0.013087602332234383, | |
| "learning_rate": 3.2394246219552724e-05, | |
| "loss": 0.9897, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.30972222222222223, | |
| "grad_norm": 0.012205411680042744, | |
| "learning_rate": 3.2358890893178617e-05, | |
| "loss": 0.9489, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.3104166666666667, | |
| "grad_norm": 0.012712490744888783, | |
| "learning_rate": 3.2323472976332926e-05, | |
| "loss": 0.9363, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.3111111111111111, | |
| "grad_norm": 0.012987968511879444, | |
| "learning_rate": 3.228799264838645e-05, | |
| "loss": 0.9568, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.31180555555555556, | |
| "grad_norm": 0.014164082705974579, | |
| "learning_rate": 3.225245008902606e-05, | |
| "loss": 0.9515, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.3125, | |
| "grad_norm": 0.014080497436225414, | |
| "learning_rate": 3.221684547825379e-05, | |
| "loss": 0.9715, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.31319444444444444, | |
| "grad_norm": 0.01266495417803526, | |
| "learning_rate": 3.218117899638594e-05, | |
| "loss": 0.9961, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.3138888888888889, | |
| "grad_norm": 0.012783526442945004, | |
| "learning_rate": 3.214545082405213e-05, | |
| "loss": 0.9667, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.3145833333333333, | |
| "grad_norm": 0.012183685787022114, | |
| "learning_rate": 3.210966114219444e-05, | |
| "loss": 0.9561, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.31527777777777777, | |
| "grad_norm": 0.014331048354506493, | |
| "learning_rate": 3.2073810132066414e-05, | |
| "loss": 1.0032, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.3159722222222222, | |
| "grad_norm": 0.012980937026441097, | |
| "learning_rate": 3.2037897975232216e-05, | |
| "loss": 0.9567, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.31666666666666665, | |
| "grad_norm": 0.01176460087299347, | |
| "learning_rate": 3.200192485356569e-05, | |
| "loss": 0.9396, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.3173611111111111, | |
| "grad_norm": 0.012090741656720638, | |
| "learning_rate": 3.1965890949249405e-05, | |
| "loss": 0.9494, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.31805555555555554, | |
| "grad_norm": 0.013988793827593327, | |
| "learning_rate": 3.192979644477378e-05, | |
| "loss": 1.0022, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.31875, | |
| "grad_norm": 0.012773574329912663, | |
| "learning_rate": 3.189364152293612e-05, | |
| "loss": 1.022, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.3194444444444444, | |
| "grad_norm": 0.013159438967704773, | |
| "learning_rate": 3.185742636683972e-05, | |
| "loss": 0.9754, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.32013888888888886, | |
| "grad_norm": 0.015010225586593151, | |
| "learning_rate": 3.1821151159892924e-05, | |
| "loss": 0.9525, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.32083333333333336, | |
| "grad_norm": 0.012587225064635277, | |
| "learning_rate": 3.1784816085808196e-05, | |
| "loss": 0.9651, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.3215277777777778, | |
| "grad_norm": 0.01335595827549696, | |
| "learning_rate": 3.17484213286012e-05, | |
| "loss": 0.9691, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.32222222222222224, | |
| "grad_norm": 0.012998990714550018, | |
| "learning_rate": 3.171196707258984e-05, | |
| "loss": 0.936, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.3229166666666667, | |
| "grad_norm": 0.013085835613310337, | |
| "learning_rate": 3.167545350239336e-05, | |
| "loss": 0.9294, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.3236111111111111, | |
| "grad_norm": 0.012886827811598778, | |
| "learning_rate": 3.16388808029314e-05, | |
| "loss": 0.9685, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.32430555555555557, | |
| "grad_norm": 0.014822276309132576, | |
| "learning_rate": 3.1602249159423054e-05, | |
| "loss": 0.9545, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.325, | |
| "grad_norm": 0.013084998354315758, | |
| "learning_rate": 3.1565558757385914e-05, | |
| "loss": 0.9638, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.32569444444444445, | |
| "grad_norm": 0.012661305256187916, | |
| "learning_rate": 3.152880978263517e-05, | |
| "loss": 0.9033, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.3263888888888889, | |
| "grad_norm": 0.0137302465736866, | |
| "learning_rate": 3.149200242128263e-05, | |
| "loss": 0.9576, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.32708333333333334, | |
| "grad_norm": 0.01353361364454031, | |
| "learning_rate": 3.145513685973583e-05, | |
| "loss": 0.9555, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.3277777777777778, | |
| "grad_norm": 0.01424756832420826, | |
| "learning_rate": 3.1418213284697e-05, | |
| "loss": 0.9583, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.3284722222222222, | |
| "grad_norm": 0.013587539084255695, | |
| "learning_rate": 3.138123188316224e-05, | |
| "loss": 0.9665, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.32916666666666666, | |
| "grad_norm": 0.013397585600614548, | |
| "learning_rate": 3.1344192842420435e-05, | |
| "loss": 0.951, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.3298611111111111, | |
| "grad_norm": 0.012381638400256634, | |
| "learning_rate": 3.130709635005245e-05, | |
| "loss": 0.96, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.33055555555555555, | |
| "grad_norm": 0.014178499579429626, | |
| "learning_rate": 3.1269942593930055e-05, | |
| "loss": 0.9891, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.33125, | |
| "grad_norm": 0.013224642723798752, | |
| "learning_rate": 3.123273176221506e-05, | |
| "loss": 0.9719, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.33194444444444443, | |
| "grad_norm": 0.012466056272387505, | |
| "learning_rate": 3.119546404335831e-05, | |
| "loss": 0.968, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.3326388888888889, | |
| "grad_norm": 0.01483081839978695, | |
| "learning_rate": 3.115813962609874e-05, | |
| "loss": 0.9601, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.013933224603533745, | |
| "learning_rate": 3.112075869946247e-05, | |
| "loss": 0.9802, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.33402777777777776, | |
| "grad_norm": 0.013764635659754276, | |
| "learning_rate": 3.108332145276177e-05, | |
| "loss": 0.9646, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.3347222222222222, | |
| "grad_norm": 0.014757770113646984, | |
| "learning_rate": 3.104582807559414e-05, | |
| "loss": 0.9592, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.33541666666666664, | |
| "grad_norm": 0.012920022010803223, | |
| "learning_rate": 3.100827875784138e-05, | |
| "loss": 0.9674, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.33611111111111114, | |
| "grad_norm": 0.013609576970338821, | |
| "learning_rate": 3.097067368966855e-05, | |
| "loss": 0.9456, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.3368055555555556, | |
| "grad_norm": 0.013191001489758492, | |
| "learning_rate": 3.093301306152308e-05, | |
| "loss": 0.9568, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.3375, | |
| "grad_norm": 0.013639998622238636, | |
| "learning_rate": 3.089529706413378e-05, | |
| "loss": 0.9746, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.33819444444444446, | |
| "grad_norm": 0.013085294514894485, | |
| "learning_rate": 3.085752588850986e-05, | |
| "loss": 0.9391, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.3388888888888889, | |
| "grad_norm": 0.01391910295933485, | |
| "learning_rate": 3.081969972593999e-05, | |
| "loss": 0.9885, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.33958333333333335, | |
| "grad_norm": 0.013547900132834911, | |
| "learning_rate": 3.0781818767991295e-05, | |
| "loss": 0.9481, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.3402777777777778, | |
| "grad_norm": 0.012131288647651672, | |
| "learning_rate": 3.074388320650843e-05, | |
| "loss": 0.9667, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.34097222222222223, | |
| "grad_norm": 0.013132079504430294, | |
| "learning_rate": 3.070589323361257e-05, | |
| "loss": 0.9915, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.3416666666666667, | |
| "grad_norm": 0.014800147153437138, | |
| "learning_rate": 3.0667849041700454e-05, | |
| "loss": 0.9737, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.3423611111111111, | |
| "grad_norm": 0.01190117746591568, | |
| "learning_rate": 3.062975082344341e-05, | |
| "loss": 0.9413, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.34305555555555556, | |
| "grad_norm": 0.011400418356060982, | |
| "learning_rate": 3.059159877178638e-05, | |
| "loss": 0.9453, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.34375, | |
| "grad_norm": 0.011752789840102196, | |
| "learning_rate": 3.055339307994693e-05, | |
| "loss": 0.9832, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.34444444444444444, | |
| "grad_norm": 0.01378630194813013, | |
| "learning_rate": 3.0515133941414294e-05, | |
| "loss": 0.9243, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.3451388888888889, | |
| "grad_norm": 0.012208312749862671, | |
| "learning_rate": 3.0476821549948376e-05, | |
| "loss": 0.9798, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.3458333333333333, | |
| "grad_norm": 0.01983562856912613, | |
| "learning_rate": 3.0438456099578775e-05, | |
| "loss": 0.9998, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.34652777777777777, | |
| "grad_norm": 0.013442217372357845, | |
| "learning_rate": 3.0400037784603805e-05, | |
| "loss": 0.9304, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.3472222222222222, | |
| "grad_norm": 0.014221866615116596, | |
| "learning_rate": 3.0361566799589498e-05, | |
| "loss": 0.9938, | |
| "step": 500 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 1440, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": false, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1598246092800000.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |