{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 35.0, "eval_steps": 500, "global_step": 47775, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0007326007326007326, "grad_norm": 13.9375, "learning_rate": 0.0, "loss": 3.2747, "step": 1 }, { "epoch": 0.0014652014652014652, "grad_norm": 13.6875, "learning_rate": 4.185851820845542e-08, "loss": 3.3134, "step": 2 }, { "epoch": 0.002197802197802198, "grad_norm": 13.4375, "learning_rate": 8.371703641691084e-08, "loss": 3.2735, "step": 3 }, { "epoch": 0.0029304029304029304, "grad_norm": 13.6875, "learning_rate": 1.2557555462536626e-07, "loss": 3.2729, "step": 4 }, { "epoch": 0.003663003663003663, "grad_norm": 13.4375, "learning_rate": 1.6743407283382168e-07, "loss": 3.2741, "step": 5 }, { "epoch": 0.004395604395604396, "grad_norm": 13.875, "learning_rate": 2.092925910422771e-07, "loss": 3.3034, "step": 6 }, { "epoch": 0.005128205128205128, "grad_norm": 13.4375, "learning_rate": 2.511511092507325e-07, "loss": 3.292, "step": 7 }, { "epoch": 0.005860805860805861, "grad_norm": 13.625, "learning_rate": 2.9300962745918797e-07, "loss": 3.2764, "step": 8 }, { "epoch": 0.006593406593406593, "grad_norm": 13.3125, "learning_rate": 3.3486814566764336e-07, "loss": 3.3023, "step": 9 }, { "epoch": 0.007326007326007326, "grad_norm": 13.625, "learning_rate": 3.767266638760988e-07, "loss": 3.3159, "step": 10 }, { "epoch": 0.00805860805860806, "grad_norm": 13.4375, "learning_rate": 4.185851820845542e-07, "loss": 3.2779, "step": 11 }, { "epoch": 0.008791208791208791, "grad_norm": 13.5625, "learning_rate": 4.604437002930097e-07, "loss": 3.287, "step": 12 }, { "epoch": 0.009523809523809525, "grad_norm": 13.4375, "learning_rate": 5.02302218501465e-07, "loss": 3.2874, "step": 13 }, { "epoch": 0.010256410256410256, "grad_norm": 13.6875, "learning_rate": 5.441607367099204e-07, "loss": 3.3202, "step": 14 }, { "epoch": 0.01098901098901099, "grad_norm": 13.875, "learning_rate": 5.860192549183759e-07, "loss": 3.2718, "step": 15 }, { "epoch": 0.011721611721611722, "grad_norm": 13.75, "learning_rate": 6.278777731268313e-07, "loss": 3.3141, "step": 16 }, { "epoch": 0.012454212454212455, "grad_norm": 13.5, "learning_rate": 6.697362913352867e-07, "loss": 3.2854, "step": 17 }, { "epoch": 0.013186813186813187, "grad_norm": 13.5625, "learning_rate": 7.115948095437422e-07, "loss": 3.3075, "step": 18 }, { "epoch": 0.01391941391941392, "grad_norm": 13.4375, "learning_rate": 7.534533277521976e-07, "loss": 3.3121, "step": 19 }, { "epoch": 0.014652014652014652, "grad_norm": 13.25, "learning_rate": 7.953118459606531e-07, "loss": 3.2709, "step": 20 }, { "epoch": 0.015384615384615385, "grad_norm": 13.4375, "learning_rate": 8.371703641691084e-07, "loss": 3.3017, "step": 21 }, { "epoch": 0.01611721611721612, "grad_norm": 13.5, "learning_rate": 8.790288823775639e-07, "loss": 3.2898, "step": 22 }, { "epoch": 0.01684981684981685, "grad_norm": 13.875, "learning_rate": 9.208874005860194e-07, "loss": 3.3152, "step": 23 }, { "epoch": 0.017582417582417582, "grad_norm": 13.3125, "learning_rate": 9.627459187944746e-07, "loss": 3.2612, "step": 24 }, { "epoch": 0.018315018315018316, "grad_norm": 13.625, "learning_rate": 1.00460443700293e-06, "loss": 3.3035, "step": 25 }, { "epoch": 0.01904761904761905, "grad_norm": 13.625, "learning_rate": 1.0464629552113856e-06, "loss": 3.3281, "step": 26 }, { "epoch": 0.01978021978021978, "grad_norm": 13.5, "learning_rate": 1.0883214734198409e-06, "loss": 3.2826, "step": 27 }, { "epoch": 0.020512820512820513, "grad_norm": 13.375, "learning_rate": 1.1301799916282964e-06, "loss": 3.2542, "step": 28 }, { "epoch": 0.021245421245421246, "grad_norm": 13.5, "learning_rate": 1.1720385098367519e-06, "loss": 3.2997, "step": 29 }, { "epoch": 0.02197802197802198, "grad_norm": 13.5, "learning_rate": 1.2138970280452072e-06, "loss": 3.3072, "step": 30 }, { "epoch": 0.02271062271062271, "grad_norm": 13.3125, "learning_rate": 1.2557555462536627e-06, "loss": 3.2717, "step": 31 }, { "epoch": 0.023443223443223443, "grad_norm": 13.5, "learning_rate": 1.2976140644621182e-06, "loss": 3.2921, "step": 32 }, { "epoch": 0.024175824175824177, "grad_norm": 13.875, "learning_rate": 1.3394725826705734e-06, "loss": 3.3219, "step": 33 }, { "epoch": 0.02490842490842491, "grad_norm": 13.5625, "learning_rate": 1.381331100879029e-06, "loss": 3.246, "step": 34 }, { "epoch": 0.02564102564102564, "grad_norm": 13.75, "learning_rate": 1.4231896190874844e-06, "loss": 3.3035, "step": 35 }, { "epoch": 0.026373626373626374, "grad_norm": 13.5, "learning_rate": 1.4650481372959397e-06, "loss": 3.2743, "step": 36 }, { "epoch": 0.027106227106227107, "grad_norm": 13.9375, "learning_rate": 1.5069066555043952e-06, "loss": 3.2636, "step": 37 }, { "epoch": 0.02783882783882784, "grad_norm": 13.3125, "learning_rate": 1.5487651737128507e-06, "loss": 3.2673, "step": 38 }, { "epoch": 0.02857142857142857, "grad_norm": 13.1875, "learning_rate": 1.5906236919213062e-06, "loss": 3.2827, "step": 39 }, { "epoch": 0.029304029304029304, "grad_norm": 13.25, "learning_rate": 1.6324822101297615e-06, "loss": 3.2558, "step": 40 }, { "epoch": 0.030036630036630037, "grad_norm": 13.25, "learning_rate": 1.6743407283382168e-06, "loss": 3.2757, "step": 41 }, { "epoch": 0.03076923076923077, "grad_norm": 13.75, "learning_rate": 1.7161992465466723e-06, "loss": 3.262, "step": 42 }, { "epoch": 0.0315018315018315, "grad_norm": 13.5, "learning_rate": 1.7580577647551278e-06, "loss": 3.2639, "step": 43 }, { "epoch": 0.03223443223443224, "grad_norm": 13.375, "learning_rate": 1.7999162829635833e-06, "loss": 3.2603, "step": 44 }, { "epoch": 0.03296703296703297, "grad_norm": 13.625, "learning_rate": 1.8417748011720388e-06, "loss": 3.2498, "step": 45 }, { "epoch": 0.0336996336996337, "grad_norm": 13.375, "learning_rate": 1.8836333193804939e-06, "loss": 3.2638, "step": 46 }, { "epoch": 0.034432234432234435, "grad_norm": 13.5625, "learning_rate": 1.925491837588949e-06, "loss": 3.2796, "step": 47 }, { "epoch": 0.035164835164835165, "grad_norm": 13.0, "learning_rate": 1.967350355797405e-06, "loss": 3.236, "step": 48 }, { "epoch": 0.035897435897435895, "grad_norm": 13.25, "learning_rate": 2.00920887400586e-06, "loss": 3.2425, "step": 49 }, { "epoch": 0.03663003663003663, "grad_norm": 13.375, "learning_rate": 2.051067392214316e-06, "loss": 3.243, "step": 50 }, { "epoch": 0.03736263736263736, "grad_norm": 13.5625, "learning_rate": 2.092925910422771e-06, "loss": 3.2333, "step": 51 }, { "epoch": 0.0380952380952381, "grad_norm": 13.3125, "learning_rate": 2.1347844286312265e-06, "loss": 3.2534, "step": 52 }, { "epoch": 0.03882783882783883, "grad_norm": 13.5, "learning_rate": 2.1766429468396817e-06, "loss": 3.2389, "step": 53 }, { "epoch": 0.03956043956043956, "grad_norm": 13.125, "learning_rate": 2.2185014650481375e-06, "loss": 3.2367, "step": 54 }, { "epoch": 0.040293040293040296, "grad_norm": 13.0625, "learning_rate": 2.2603599832565927e-06, "loss": 3.2026, "step": 55 }, { "epoch": 0.041025641025641026, "grad_norm": 13.1875, "learning_rate": 2.3022185014650485e-06, "loss": 3.2148, "step": 56 }, { "epoch": 0.041758241758241756, "grad_norm": 13.1875, "learning_rate": 2.3440770196735037e-06, "loss": 3.2427, "step": 57 }, { "epoch": 0.04249084249084249, "grad_norm": 12.75, "learning_rate": 2.385935537881959e-06, "loss": 3.1875, "step": 58 }, { "epoch": 0.04322344322344322, "grad_norm": 12.8125, "learning_rate": 2.4277940560904143e-06, "loss": 3.1758, "step": 59 }, { "epoch": 0.04395604395604396, "grad_norm": 13.3125, "learning_rate": 2.46965257429887e-06, "loss": 3.1817, "step": 60 }, { "epoch": 0.04468864468864469, "grad_norm": 13.3125, "learning_rate": 2.5115110925073253e-06, "loss": 3.1638, "step": 61 }, { "epoch": 0.04542124542124542, "grad_norm": 13.0, "learning_rate": 2.5533696107157806e-06, "loss": 3.1805, "step": 62 }, { "epoch": 0.046153846153846156, "grad_norm": 13.125, "learning_rate": 2.5952281289242363e-06, "loss": 3.1473, "step": 63 }, { "epoch": 0.046886446886446886, "grad_norm": 13.0625, "learning_rate": 2.6370866471326916e-06, "loss": 3.1675, "step": 64 }, { "epoch": 0.047619047619047616, "grad_norm": 12.8125, "learning_rate": 2.678945165341147e-06, "loss": 3.1479, "step": 65 }, { "epoch": 0.04835164835164835, "grad_norm": 12.625, "learning_rate": 2.7208036835496026e-06, "loss": 3.1323, "step": 66 }, { "epoch": 0.04908424908424908, "grad_norm": 13.0625, "learning_rate": 2.762662201758058e-06, "loss": 3.1314, "step": 67 }, { "epoch": 0.04981684981684982, "grad_norm": 12.6875, "learning_rate": 2.804520719966513e-06, "loss": 3.1678, "step": 68 }, { "epoch": 0.05054945054945055, "grad_norm": 12.9375, "learning_rate": 2.846379238174969e-06, "loss": 3.1161, "step": 69 }, { "epoch": 0.05128205128205128, "grad_norm": 12.875, "learning_rate": 2.888237756383424e-06, "loss": 3.1425, "step": 70 }, { "epoch": 0.05201465201465202, "grad_norm": 13.0, "learning_rate": 2.9300962745918795e-06, "loss": 3.1099, "step": 71 }, { "epoch": 0.05274725274725275, "grad_norm": 13.0, "learning_rate": 2.971954792800335e-06, "loss": 3.1457, "step": 72 }, { "epoch": 0.05347985347985348, "grad_norm": 12.8125, "learning_rate": 3.0138133110087905e-06, "loss": 3.1063, "step": 73 }, { "epoch": 0.054212454212454214, "grad_norm": 12.75, "learning_rate": 3.0556718292172457e-06, "loss": 3.0787, "step": 74 }, { "epoch": 0.054945054945054944, "grad_norm": 12.625, "learning_rate": 3.0975303474257015e-06, "loss": 3.0807, "step": 75 }, { "epoch": 0.05567765567765568, "grad_norm": 12.75, "learning_rate": 3.1393888656341567e-06, "loss": 3.0927, "step": 76 }, { "epoch": 0.05641025641025641, "grad_norm": 12.625, "learning_rate": 3.1812473838426125e-06, "loss": 3.0947, "step": 77 }, { "epoch": 0.05714285714285714, "grad_norm": 12.5, "learning_rate": 3.2231059020510677e-06, "loss": 3.0784, "step": 78 }, { "epoch": 0.05787545787545788, "grad_norm": 13.0, "learning_rate": 3.264964420259523e-06, "loss": 3.0794, "step": 79 }, { "epoch": 0.05860805860805861, "grad_norm": 12.5, "learning_rate": 3.3068229384679787e-06, "loss": 3.0724, "step": 80 }, { "epoch": 0.05934065934065934, "grad_norm": 12.375, "learning_rate": 3.3486814566764336e-06, "loss": 3.0802, "step": 81 }, { "epoch": 0.060073260073260075, "grad_norm": 12.1875, "learning_rate": 3.390539974884889e-06, "loss": 3.0672, "step": 82 }, { "epoch": 0.060805860805860805, "grad_norm": 12.4375, "learning_rate": 3.4323984930933446e-06, "loss": 3.073, "step": 83 }, { "epoch": 0.06153846153846154, "grad_norm": 12.25, "learning_rate": 3.4742570113018e-06, "loss": 3.023, "step": 84 }, { "epoch": 0.06227106227106227, "grad_norm": 12.8125, "learning_rate": 3.5161155295102556e-06, "loss": 3.0047, "step": 85 }, { "epoch": 0.063003663003663, "grad_norm": 12.4375, "learning_rate": 3.557974047718711e-06, "loss": 3.0289, "step": 86 }, { "epoch": 0.06373626373626373, "grad_norm": 12.3125, "learning_rate": 3.5998325659271666e-06, "loss": 3.0325, "step": 87 }, { "epoch": 0.06446886446886448, "grad_norm": 12.25, "learning_rate": 3.641691084135622e-06, "loss": 3.0153, "step": 88 }, { "epoch": 0.0652014652014652, "grad_norm": 12.375, "learning_rate": 3.6835496023440776e-06, "loss": 3.0218, "step": 89 }, { "epoch": 0.06593406593406594, "grad_norm": 12.1875, "learning_rate": 3.725408120552533e-06, "loss": 3.0324, "step": 90 }, { "epoch": 0.06666666666666667, "grad_norm": 12.375, "learning_rate": 3.7672666387609878e-06, "loss": 3.0192, "step": 91 }, { "epoch": 0.0673992673992674, "grad_norm": 12.25, "learning_rate": 3.809125156969443e-06, "loss": 3.0265, "step": 92 }, { "epoch": 0.06813186813186813, "grad_norm": 12.4375, "learning_rate": 3.850983675177898e-06, "loss": 2.9696, "step": 93 }, { "epoch": 0.06886446886446887, "grad_norm": 12.125, "learning_rate": 3.892842193386354e-06, "loss": 2.9875, "step": 94 }, { "epoch": 0.0695970695970696, "grad_norm": 12.1875, "learning_rate": 3.93470071159481e-06, "loss": 2.9806, "step": 95 }, { "epoch": 0.07032967032967033, "grad_norm": 12.3125, "learning_rate": 3.9765592298032655e-06, "loss": 2.9537, "step": 96 }, { "epoch": 0.07106227106227106, "grad_norm": 12.0625, "learning_rate": 4.01841774801172e-06, "loss": 2.9471, "step": 97 }, { "epoch": 0.07179487179487179, "grad_norm": 12.1875, "learning_rate": 4.060276266220176e-06, "loss": 2.9196, "step": 98 }, { "epoch": 0.07252747252747253, "grad_norm": 12.0, "learning_rate": 4.102134784428632e-06, "loss": 2.9368, "step": 99 }, { "epoch": 0.07326007326007326, "grad_norm": 12.375, "learning_rate": 4.1439933026370875e-06, "loss": 2.8943, "step": 100 }, { "epoch": 0.073992673992674, "grad_norm": 11.625, "learning_rate": 4.185851820845542e-06, "loss": 2.9095, "step": 101 }, { "epoch": 0.07472527472527472, "grad_norm": 11.6875, "learning_rate": 4.227710339053997e-06, "loss": 2.8783, "step": 102 }, { "epoch": 0.07545787545787545, "grad_norm": 11.8125, "learning_rate": 4.269568857262453e-06, "loss": 2.883, "step": 103 }, { "epoch": 0.0761904761904762, "grad_norm": 11.5, "learning_rate": 4.311427375470909e-06, "loss": 2.8648, "step": 104 }, { "epoch": 0.07692307692307693, "grad_norm": 11.5625, "learning_rate": 4.3532858936793635e-06, "loss": 2.8549, "step": 105 }, { "epoch": 0.07765567765567766, "grad_norm": 11.625, "learning_rate": 4.395144411887819e-06, "loss": 2.7849, "step": 106 }, { "epoch": 0.07838827838827839, "grad_norm": 11.125, "learning_rate": 4.437002930096275e-06, "loss": 2.8211, "step": 107 }, { "epoch": 0.07912087912087912, "grad_norm": 11.4375, "learning_rate": 4.478861448304731e-06, "loss": 2.8064, "step": 108 }, { "epoch": 0.07985347985347985, "grad_norm": 10.9375, "learning_rate": 4.5207199665131855e-06, "loss": 2.7612, "step": 109 }, { "epoch": 0.08058608058608059, "grad_norm": 10.625, "learning_rate": 4.562578484721641e-06, "loss": 2.7315, "step": 110 }, { "epoch": 0.08131868131868132, "grad_norm": 10.8125, "learning_rate": 4.604437002930097e-06, "loss": 2.7541, "step": 111 }, { "epoch": 0.08205128205128205, "grad_norm": 10.625, "learning_rate": 4.646295521138552e-06, "loss": 2.7433, "step": 112 }, { "epoch": 0.08278388278388278, "grad_norm": 10.6875, "learning_rate": 4.6881540393470075e-06, "loss": 2.7155, "step": 113 }, { "epoch": 0.08351648351648351, "grad_norm": 10.5625, "learning_rate": 4.730012557555462e-06, "loss": 2.7065, "step": 114 }, { "epoch": 0.08424908424908426, "grad_norm": 10.75, "learning_rate": 4.771871075763918e-06, "loss": 2.6902, "step": 115 }, { "epoch": 0.08498168498168499, "grad_norm": 10.5625, "learning_rate": 4.813729593972374e-06, "loss": 2.6495, "step": 116 }, { "epoch": 0.08571428571428572, "grad_norm": 10.5, "learning_rate": 4.855588112180829e-06, "loss": 2.6577, "step": 117 }, { "epoch": 0.08644688644688645, "grad_norm": 10.5, "learning_rate": 4.897446630389284e-06, "loss": 2.635, "step": 118 }, { "epoch": 0.08717948717948718, "grad_norm": 10.625, "learning_rate": 4.93930514859774e-06, "loss": 2.6112, "step": 119 }, { "epoch": 0.08791208791208792, "grad_norm": 10.125, "learning_rate": 4.981163666806196e-06, "loss": 2.6044, "step": 120 }, { "epoch": 0.08864468864468865, "grad_norm": 10.0, "learning_rate": 5.023022185014651e-06, "loss": 2.5948, "step": 121 }, { "epoch": 0.08937728937728938, "grad_norm": 10.1875, "learning_rate": 5.064880703223106e-06, "loss": 2.5742, "step": 122 }, { "epoch": 0.09010989010989011, "grad_norm": 10.0, "learning_rate": 5.106739221431561e-06, "loss": 2.5542, "step": 123 }, { "epoch": 0.09084249084249084, "grad_norm": 9.6875, "learning_rate": 5.148597739640017e-06, "loss": 2.5374, "step": 124 }, { "epoch": 0.09157509157509157, "grad_norm": 9.6875, "learning_rate": 5.190456257848473e-06, "loss": 2.5286, "step": 125 }, { "epoch": 0.09230769230769231, "grad_norm": 9.75, "learning_rate": 5.2323147760569275e-06, "loss": 2.5076, "step": 126 }, { "epoch": 0.09304029304029304, "grad_norm": 9.6875, "learning_rate": 5.274173294265383e-06, "loss": 2.5027, "step": 127 }, { "epoch": 0.09377289377289377, "grad_norm": 9.5, "learning_rate": 5.316031812473839e-06, "loss": 2.4653, "step": 128 }, { "epoch": 0.0945054945054945, "grad_norm": 9.375, "learning_rate": 5.357890330682294e-06, "loss": 2.465, "step": 129 }, { "epoch": 0.09523809523809523, "grad_norm": 9.1875, "learning_rate": 5.3997488488907495e-06, "loss": 2.4797, "step": 130 }, { "epoch": 0.09597069597069598, "grad_norm": 9.0625, "learning_rate": 5.441607367099205e-06, "loss": 2.4797, "step": 131 }, { "epoch": 0.0967032967032967, "grad_norm": 8.8125, "learning_rate": 5.483465885307661e-06, "loss": 2.4175, "step": 132 }, { "epoch": 0.09743589743589744, "grad_norm": 9.0, "learning_rate": 5.525324403516116e-06, "loss": 2.419, "step": 133 }, { "epoch": 0.09816849816849817, "grad_norm": 9.25, "learning_rate": 5.567182921724571e-06, "loss": 2.4009, "step": 134 }, { "epoch": 0.0989010989010989, "grad_norm": 8.8125, "learning_rate": 5.609041439933026e-06, "loss": 2.4101, "step": 135 }, { "epoch": 0.09963369963369964, "grad_norm": 8.375, "learning_rate": 5.650899958141482e-06, "loss": 2.3989, "step": 136 }, { "epoch": 0.10036630036630037, "grad_norm": 8.5625, "learning_rate": 5.692758476349938e-06, "loss": 2.3724, "step": 137 }, { "epoch": 0.1010989010989011, "grad_norm": 8.1875, "learning_rate": 5.734616994558393e-06, "loss": 2.3773, "step": 138 }, { "epoch": 0.10183150183150183, "grad_norm": 8.5, "learning_rate": 5.776475512766848e-06, "loss": 2.3417, "step": 139 }, { "epoch": 0.10256410256410256, "grad_norm": 8.4375, "learning_rate": 5.818334030975304e-06, "loss": 2.3405, "step": 140 }, { "epoch": 0.10329670329670329, "grad_norm": 8.5625, "learning_rate": 5.860192549183759e-06, "loss": 2.3143, "step": 141 }, { "epoch": 0.10402930402930403, "grad_norm": 8.125, "learning_rate": 5.902051067392215e-06, "loss": 2.2778, "step": 142 }, { "epoch": 0.10476190476190476, "grad_norm": 8.125, "learning_rate": 5.94390958560067e-06, "loss": 2.3031, "step": 143 }, { "epoch": 0.1054945054945055, "grad_norm": 8.1875, "learning_rate": 5.985768103809125e-06, "loss": 2.2803, "step": 144 }, { "epoch": 0.10622710622710622, "grad_norm": 7.875, "learning_rate": 6.027626622017581e-06, "loss": 2.2896, "step": 145 }, { "epoch": 0.10695970695970695, "grad_norm": 7.5, "learning_rate": 6.069485140226036e-06, "loss": 2.287, "step": 146 }, { "epoch": 0.1076923076923077, "grad_norm": 8.125, "learning_rate": 6.1113436584344915e-06, "loss": 2.2184, "step": 147 }, { "epoch": 0.10842490842490843, "grad_norm": 7.6875, "learning_rate": 6.153202176642947e-06, "loss": 2.2504, "step": 148 }, { "epoch": 0.10915750915750916, "grad_norm": 7.28125, "learning_rate": 6.195060694851403e-06, "loss": 2.2278, "step": 149 }, { "epoch": 0.10989010989010989, "grad_norm": 7.1875, "learning_rate": 6.236919213059858e-06, "loss": 2.2504, "step": 150 }, { "epoch": 0.11062271062271062, "grad_norm": 7.75, "learning_rate": 6.2787777312683135e-06, "loss": 2.1737, "step": 151 }, { "epoch": 0.11135531135531136, "grad_norm": 7.03125, "learning_rate": 6.320636249476769e-06, "loss": 2.2111, "step": 152 }, { "epoch": 0.11208791208791209, "grad_norm": 7.21875, "learning_rate": 6.362494767685225e-06, "loss": 2.1737, "step": 153 }, { "epoch": 0.11282051282051282, "grad_norm": 7.375, "learning_rate": 6.40435328589368e-06, "loss": 2.1322, "step": 154 }, { "epoch": 0.11355311355311355, "grad_norm": 6.875, "learning_rate": 6.4462118041021355e-06, "loss": 2.175, "step": 155 }, { "epoch": 0.11428571428571428, "grad_norm": 6.9375, "learning_rate": 6.488070322310591e-06, "loss": 2.131, "step": 156 }, { "epoch": 0.11501831501831501, "grad_norm": 6.875, "learning_rate": 6.529928840519046e-06, "loss": 2.1177, "step": 157 }, { "epoch": 0.11575091575091576, "grad_norm": 6.53125, "learning_rate": 6.571787358727502e-06, "loss": 2.1458, "step": 158 }, { "epoch": 0.11648351648351649, "grad_norm": 6.65625, "learning_rate": 6.6136458769359575e-06, "loss": 2.1163, "step": 159 }, { "epoch": 0.11721611721611722, "grad_norm": 6.15625, "learning_rate": 6.6555043951444115e-06, "loss": 2.0954, "step": 160 }, { "epoch": 0.11794871794871795, "grad_norm": 6.21875, "learning_rate": 6.697362913352867e-06, "loss": 2.0899, "step": 161 }, { "epoch": 0.11868131868131868, "grad_norm": 6.21875, "learning_rate": 6.739221431561323e-06, "loss": 2.0679, "step": 162 }, { "epoch": 0.11941391941391942, "grad_norm": 6.0, "learning_rate": 6.781079949769778e-06, "loss": 2.0517, "step": 163 }, { "epoch": 0.12014652014652015, "grad_norm": 5.8125, "learning_rate": 6.8229384679782335e-06, "loss": 2.0594, "step": 164 }, { "epoch": 0.12087912087912088, "grad_norm": 5.96875, "learning_rate": 6.864796986186689e-06, "loss": 2.016, "step": 165 }, { "epoch": 0.12161172161172161, "grad_norm": 5.5, "learning_rate": 6.906655504395144e-06, "loss": 2.0567, "step": 166 }, { "epoch": 0.12234432234432234, "grad_norm": 5.6875, "learning_rate": 6.9485140226036e-06, "loss": 2.0127, "step": 167 }, { "epoch": 0.12307692307692308, "grad_norm": 6.03125, "learning_rate": 6.9903725408120555e-06, "loss": 1.9597, "step": 168 }, { "epoch": 0.12380952380952381, "grad_norm": 5.375, "learning_rate": 7.032231059020511e-06, "loss": 1.9922, "step": 169 }, { "epoch": 0.12454212454212454, "grad_norm": 5.125, "learning_rate": 7.074089577228966e-06, "loss": 1.9818, "step": 170 }, { "epoch": 0.12527472527472527, "grad_norm": 5.0625, "learning_rate": 7.115948095437422e-06, "loss": 1.9556, "step": 171 }, { "epoch": 0.126007326007326, "grad_norm": 4.875, "learning_rate": 7.1578066136458775e-06, "loss": 1.9988, "step": 172 }, { "epoch": 0.12673992673992673, "grad_norm": 4.5, "learning_rate": 7.199665131854333e-06, "loss": 1.9742, "step": 173 }, { "epoch": 0.12747252747252746, "grad_norm": 4.6875, "learning_rate": 7.241523650062788e-06, "loss": 1.9489, "step": 174 }, { "epoch": 0.1282051282051282, "grad_norm": 4.65625, "learning_rate": 7.283382168271244e-06, "loss": 1.8948, "step": 175 }, { "epoch": 0.12893772893772895, "grad_norm": 4.125, "learning_rate": 7.3252406864796995e-06, "loss": 1.9719, "step": 176 }, { "epoch": 0.12967032967032968, "grad_norm": 4.21875, "learning_rate": 7.367099204688155e-06, "loss": 1.8989, "step": 177 }, { "epoch": 0.1304029304029304, "grad_norm": 4.15625, "learning_rate": 7.40895772289661e-06, "loss": 1.9178, "step": 178 }, { "epoch": 0.13113553113553114, "grad_norm": 4.53125, "learning_rate": 7.450816241105066e-06, "loss": 1.9206, "step": 179 }, { "epoch": 0.13186813186813187, "grad_norm": 3.78125, "learning_rate": 7.4926747593135215e-06, "loss": 1.8751, "step": 180 }, { "epoch": 0.1326007326007326, "grad_norm": 3.5, "learning_rate": 7.5345332775219755e-06, "loss": 1.8896, "step": 181 }, { "epoch": 0.13333333333333333, "grad_norm": 4.375, "learning_rate": 7.576391795730431e-06, "loss": 1.8365, "step": 182 }, { "epoch": 0.13406593406593406, "grad_norm": 3.203125, "learning_rate": 7.618250313938886e-06, "loss": 1.8891, "step": 183 }, { "epoch": 0.1347985347985348, "grad_norm": 3.4375, "learning_rate": 7.660108832147343e-06, "loss": 1.8561, "step": 184 }, { "epoch": 0.13553113553113552, "grad_norm": 3.140625, "learning_rate": 7.701967350355797e-06, "loss": 1.8473, "step": 185 }, { "epoch": 0.13626373626373625, "grad_norm": 3.0, "learning_rate": 7.743825868564252e-06, "loss": 1.8532, "step": 186 }, { "epoch": 0.136996336996337, "grad_norm": 3.0, "learning_rate": 7.785684386772708e-06, "loss": 1.8226, "step": 187 }, { "epoch": 0.13772893772893774, "grad_norm": 2.78125, "learning_rate": 7.827542904981164e-06, "loss": 1.8528, "step": 188 }, { "epoch": 0.13846153846153847, "grad_norm": 2.53125, "learning_rate": 7.86940142318962e-06, "loss": 1.8124, "step": 189 }, { "epoch": 0.1391941391941392, "grad_norm": 2.46875, "learning_rate": 7.911259941398075e-06, "loss": 1.8246, "step": 190 }, { "epoch": 0.13992673992673993, "grad_norm": 2.421875, "learning_rate": 7.953118459606531e-06, "loss": 1.8013, "step": 191 }, { "epoch": 0.14065934065934066, "grad_norm": 2.640625, "learning_rate": 7.994976977814985e-06, "loss": 1.8101, "step": 192 }, { "epoch": 0.1413919413919414, "grad_norm": 2.390625, "learning_rate": 8.03683549602344e-06, "loss": 1.7914, "step": 193 }, { "epoch": 0.14212454212454212, "grad_norm": 2.46875, "learning_rate": 8.078694014231896e-06, "loss": 1.7748, "step": 194 }, { "epoch": 0.14285714285714285, "grad_norm": 2.109375, "learning_rate": 8.120552532440352e-06, "loss": 1.7933, "step": 195 }, { "epoch": 0.14358974358974358, "grad_norm": 2.296875, "learning_rate": 8.162411050648808e-06, "loss": 1.7627, "step": 196 }, { "epoch": 0.1443223443223443, "grad_norm": 2.078125, "learning_rate": 8.204269568857264e-06, "loss": 1.7763, "step": 197 }, { "epoch": 0.14505494505494507, "grad_norm": 1.90625, "learning_rate": 8.24612808706572e-06, "loss": 1.7664, "step": 198 }, { "epoch": 0.1457875457875458, "grad_norm": 1.9140625, "learning_rate": 8.287986605274175e-06, "loss": 1.7701, "step": 199 }, { "epoch": 0.14652014652014653, "grad_norm": 1.8125, "learning_rate": 8.329845123482629e-06, "loss": 1.7534, "step": 200 }, { "epoch": 0.14725274725274726, "grad_norm": 1.8046875, "learning_rate": 8.371703641691085e-06, "loss": 1.7558, "step": 201 }, { "epoch": 0.147985347985348, "grad_norm": 1.8046875, "learning_rate": 8.413562159899539e-06, "loss": 1.7381, "step": 202 }, { "epoch": 0.14871794871794872, "grad_norm": 1.796875, "learning_rate": 8.455420678107994e-06, "loss": 1.7328, "step": 203 }, { "epoch": 0.14945054945054945, "grad_norm": 1.5859375, "learning_rate": 8.49727919631645e-06, "loss": 1.7403, "step": 204 }, { "epoch": 0.15018315018315018, "grad_norm": 1.703125, "learning_rate": 8.539137714524906e-06, "loss": 1.7303, "step": 205 }, { "epoch": 0.1509157509157509, "grad_norm": 1.921875, "learning_rate": 8.580996232733362e-06, "loss": 1.7088, "step": 206 }, { "epoch": 0.15164835164835164, "grad_norm": 1.515625, "learning_rate": 8.622854750941817e-06, "loss": 1.7336, "step": 207 }, { "epoch": 0.1523809523809524, "grad_norm": 1.40625, "learning_rate": 8.664713269150273e-06, "loss": 1.7239, "step": 208 }, { "epoch": 0.15311355311355312, "grad_norm": 2.015625, "learning_rate": 8.706571787358727e-06, "loss": 1.6755, "step": 209 }, { "epoch": 0.15384615384615385, "grad_norm": 1.8125, "learning_rate": 8.748430305567183e-06, "loss": 1.7158, "step": 210 }, { "epoch": 0.15457875457875458, "grad_norm": 1.46875, "learning_rate": 8.790288823775638e-06, "loss": 1.743, "step": 211 }, { "epoch": 0.15531135531135531, "grad_norm": 1.390625, "learning_rate": 8.832147341984094e-06, "loss": 1.6915, "step": 212 }, { "epoch": 0.15604395604395604, "grad_norm": 1.2734375, "learning_rate": 8.87400586019255e-06, "loss": 1.6915, "step": 213 }, { "epoch": 0.15677655677655677, "grad_norm": 1.390625, "learning_rate": 8.915864378401006e-06, "loss": 1.6793, "step": 214 }, { "epoch": 0.1575091575091575, "grad_norm": 1.546875, "learning_rate": 8.957722896609461e-06, "loss": 1.7313, "step": 215 }, { "epoch": 0.15824175824175823, "grad_norm": 1.296875, "learning_rate": 8.999581414817917e-06, "loss": 1.6951, "step": 216 }, { "epoch": 0.15897435897435896, "grad_norm": 1.4921875, "learning_rate": 9.041439933026371e-06, "loss": 1.6969, "step": 217 }, { "epoch": 0.1597069597069597, "grad_norm": 1.2890625, "learning_rate": 9.083298451234827e-06, "loss": 1.7022, "step": 218 }, { "epoch": 0.16043956043956045, "grad_norm": 1.28125, "learning_rate": 9.125156969443282e-06, "loss": 1.6808, "step": 219 }, { "epoch": 0.16117216117216118, "grad_norm": 1.5703125, "learning_rate": 9.167015487651738e-06, "loss": 1.7395, "step": 220 }, { "epoch": 0.1619047619047619, "grad_norm": 1.4609375, "learning_rate": 9.208874005860194e-06, "loss": 1.7126, "step": 221 }, { "epoch": 0.16263736263736264, "grad_norm": 1.1796875, "learning_rate": 9.25073252406865e-06, "loss": 1.7042, "step": 222 }, { "epoch": 0.16336996336996337, "grad_norm": 1.484375, "learning_rate": 9.292591042277104e-06, "loss": 1.6769, "step": 223 }, { "epoch": 0.1641025641025641, "grad_norm": 1.21875, "learning_rate": 9.33444956048556e-06, "loss": 1.6956, "step": 224 }, { "epoch": 0.16483516483516483, "grad_norm": 1.6328125, "learning_rate": 9.376308078694015e-06, "loss": 1.7085, "step": 225 }, { "epoch": 0.16556776556776556, "grad_norm": 1.5234375, "learning_rate": 9.418166596902469e-06, "loss": 1.6992, "step": 226 }, { "epoch": 0.1663003663003663, "grad_norm": 2.5, "learning_rate": 9.460025115110925e-06, "loss": 1.6658, "step": 227 }, { "epoch": 0.16703296703296702, "grad_norm": 2.09375, "learning_rate": 9.50188363331938e-06, "loss": 1.6347, "step": 228 }, { "epoch": 0.16776556776556775, "grad_norm": 2.3125, "learning_rate": 9.543742151527836e-06, "loss": 1.6427, "step": 229 }, { "epoch": 0.1684981684981685, "grad_norm": 1.234375, "learning_rate": 9.585600669736292e-06, "loss": 1.6942, "step": 230 }, { "epoch": 0.16923076923076924, "grad_norm": 1.140625, "learning_rate": 9.627459187944748e-06, "loss": 1.6979, "step": 231 }, { "epoch": 0.16996336996336997, "grad_norm": 1.2421875, "learning_rate": 9.669317706153203e-06, "loss": 1.7037, "step": 232 }, { "epoch": 0.1706959706959707, "grad_norm": 1.4921875, "learning_rate": 9.711176224361657e-06, "loss": 1.643, "step": 233 }, { "epoch": 0.17142857142857143, "grad_norm": 1.5, "learning_rate": 9.753034742570113e-06, "loss": 1.7253, "step": 234 }, { "epoch": 0.17216117216117216, "grad_norm": 1.3203125, "learning_rate": 9.794893260778569e-06, "loss": 1.6749, "step": 235 }, { "epoch": 0.1728937728937729, "grad_norm": 1.2890625, "learning_rate": 9.836751778987024e-06, "loss": 1.6929, "step": 236 }, { "epoch": 0.17362637362637362, "grad_norm": 1.1796875, "learning_rate": 9.87861029719548e-06, "loss": 1.6857, "step": 237 }, { "epoch": 0.17435897435897435, "grad_norm": 1.15625, "learning_rate": 9.920468815403936e-06, "loss": 1.669, "step": 238 }, { "epoch": 0.17509157509157508, "grad_norm": 1.296875, "learning_rate": 9.962327333612392e-06, "loss": 1.6732, "step": 239 }, { "epoch": 0.17582417582417584, "grad_norm": 1.2734375, "learning_rate": 1.0004185851820847e-05, "loss": 1.6962, "step": 240 }, { "epoch": 0.17655677655677657, "grad_norm": 1.0859375, "learning_rate": 1.0046044370029301e-05, "loss": 1.6827, "step": 241 }, { "epoch": 0.1772893772893773, "grad_norm": 1.5625, "learning_rate": 1.0087902888237757e-05, "loss": 1.6422, "step": 242 }, { "epoch": 0.17802197802197803, "grad_norm": 1.203125, "learning_rate": 1.0129761406446213e-05, "loss": 1.6693, "step": 243 }, { "epoch": 0.17875457875457876, "grad_norm": 1.109375, "learning_rate": 1.0171619924654668e-05, "loss": 1.6549, "step": 244 }, { "epoch": 0.1794871794871795, "grad_norm": 1.34375, "learning_rate": 1.0213478442863122e-05, "loss": 1.6357, "step": 245 }, { "epoch": 0.18021978021978022, "grad_norm": 1.1796875, "learning_rate": 1.0255336961071578e-05, "loss": 1.6843, "step": 246 }, { "epoch": 0.18095238095238095, "grad_norm": 1.1953125, "learning_rate": 1.0297195479280034e-05, "loss": 1.6591, "step": 247 }, { "epoch": 0.18168498168498168, "grad_norm": 1.8671875, "learning_rate": 1.033905399748849e-05, "loss": 1.6197, "step": 248 }, { "epoch": 0.1824175824175824, "grad_norm": 1.1484375, "learning_rate": 1.0380912515696945e-05, "loss": 1.6501, "step": 249 }, { "epoch": 0.18315018315018314, "grad_norm": 1.265625, "learning_rate": 1.04227710339054e-05, "loss": 1.6242, "step": 250 }, { "epoch": 0.1838827838827839, "grad_norm": 1.5546875, "learning_rate": 1.0464629552113855e-05, "loss": 1.6514, "step": 251 }, { "epoch": 0.18461538461538463, "grad_norm": 1.7890625, "learning_rate": 1.050648807032231e-05, "loss": 1.6677, "step": 252 }, { "epoch": 0.18534798534798536, "grad_norm": 1.03125, "learning_rate": 1.0548346588530766e-05, "loss": 1.6317, "step": 253 }, { "epoch": 0.18608058608058609, "grad_norm": 1.84375, "learning_rate": 1.0590205106739222e-05, "loss": 1.6847, "step": 254 }, { "epoch": 0.18681318681318682, "grad_norm": 1.0390625, "learning_rate": 1.0632063624947678e-05, "loss": 1.6503, "step": 255 }, { "epoch": 0.18754578754578755, "grad_norm": 1.0078125, "learning_rate": 1.0673922143156134e-05, "loss": 1.6534, "step": 256 }, { "epoch": 0.18827838827838828, "grad_norm": 1.4375, "learning_rate": 1.0715780661364588e-05, "loss": 1.6883, "step": 257 }, { "epoch": 0.189010989010989, "grad_norm": 1.0625, "learning_rate": 1.0757639179573043e-05, "loss": 1.6816, "step": 258 }, { "epoch": 0.18974358974358974, "grad_norm": 1.359375, "learning_rate": 1.0799497697781499e-05, "loss": 1.6134, "step": 259 }, { "epoch": 0.19047619047619047, "grad_norm": 1.109375, "learning_rate": 1.0841356215989955e-05, "loss": 1.6412, "step": 260 }, { "epoch": 0.1912087912087912, "grad_norm": 1.5, "learning_rate": 1.088321473419841e-05, "loss": 1.6122, "step": 261 }, { "epoch": 0.19194139194139195, "grad_norm": 1.390625, "learning_rate": 1.0925073252406866e-05, "loss": 1.6757, "step": 262 }, { "epoch": 0.19267399267399268, "grad_norm": 1.0, "learning_rate": 1.0966931770615322e-05, "loss": 1.6515, "step": 263 }, { "epoch": 0.1934065934065934, "grad_norm": 1.2109375, "learning_rate": 1.1008790288823778e-05, "loss": 1.6467, "step": 264 }, { "epoch": 0.19413919413919414, "grad_norm": 1.734375, "learning_rate": 1.1050648807032232e-05, "loss": 1.6457, "step": 265 }, { "epoch": 0.19487179487179487, "grad_norm": 1.59375, "learning_rate": 1.1092507325240686e-05, "loss": 1.6269, "step": 266 }, { "epoch": 0.1956043956043956, "grad_norm": 1.2421875, "learning_rate": 1.1134365843449141e-05, "loss": 1.6629, "step": 267 }, { "epoch": 0.19633699633699633, "grad_norm": 1.375, "learning_rate": 1.1176224361657597e-05, "loss": 1.6779, "step": 268 }, { "epoch": 0.19706959706959706, "grad_norm": 1.046875, "learning_rate": 1.1218082879866053e-05, "loss": 1.649, "step": 269 }, { "epoch": 0.1978021978021978, "grad_norm": 1.3359375, "learning_rate": 1.1259941398074508e-05, "loss": 1.6827, "step": 270 }, { "epoch": 0.19853479853479852, "grad_norm": 1.078125, "learning_rate": 1.1301799916282964e-05, "loss": 1.6378, "step": 271 }, { "epoch": 0.19926739926739928, "grad_norm": 2.21875, "learning_rate": 1.134365843449142e-05, "loss": 1.5804, "step": 272 }, { "epoch": 0.2, "grad_norm": 1.0625, "learning_rate": 1.1385516952699876e-05, "loss": 1.6406, "step": 273 }, { "epoch": 0.20073260073260074, "grad_norm": 1.0703125, "learning_rate": 1.142737547090833e-05, "loss": 1.6216, "step": 274 }, { "epoch": 0.20146520146520147, "grad_norm": 1.515625, "learning_rate": 1.1469233989116785e-05, "loss": 1.5926, "step": 275 }, { "epoch": 0.2021978021978022, "grad_norm": 1.359375, "learning_rate": 1.1511092507325241e-05, "loss": 1.6272, "step": 276 }, { "epoch": 0.20293040293040293, "grad_norm": 1.7578125, "learning_rate": 1.1552951025533697e-05, "loss": 1.6202, "step": 277 }, { "epoch": 0.20366300366300366, "grad_norm": 1.4453125, "learning_rate": 1.1594809543742152e-05, "loss": 1.633, "step": 278 }, { "epoch": 0.2043956043956044, "grad_norm": 1.5859375, "learning_rate": 1.1636668061950608e-05, "loss": 1.5974, "step": 279 }, { "epoch": 0.20512820512820512, "grad_norm": 1.3203125, "learning_rate": 1.1678526580159064e-05, "loss": 1.593, "step": 280 }, { "epoch": 0.20586080586080585, "grad_norm": 1.1796875, "learning_rate": 1.1720385098367518e-05, "loss": 1.6112, "step": 281 }, { "epoch": 0.20659340659340658, "grad_norm": 1.2578125, "learning_rate": 1.1762243616575974e-05, "loss": 1.6064, "step": 282 }, { "epoch": 0.20732600732600734, "grad_norm": 1.2578125, "learning_rate": 1.180410213478443e-05, "loss": 1.6386, "step": 283 }, { "epoch": 0.20805860805860807, "grad_norm": 1.0703125, "learning_rate": 1.1845960652992885e-05, "loss": 1.6085, "step": 284 }, { "epoch": 0.2087912087912088, "grad_norm": 1.234375, "learning_rate": 1.188781917120134e-05, "loss": 1.6334, "step": 285 }, { "epoch": 0.20952380952380953, "grad_norm": 1.2265625, "learning_rate": 1.1929677689409796e-05, "loss": 1.6259, "step": 286 }, { "epoch": 0.21025641025641026, "grad_norm": 1.7265625, "learning_rate": 1.197153620761825e-05, "loss": 1.6356, "step": 287 }, { "epoch": 0.210989010989011, "grad_norm": 1.0703125, "learning_rate": 1.2013394725826706e-05, "loss": 1.6126, "step": 288 }, { "epoch": 0.21172161172161172, "grad_norm": 1.0546875, "learning_rate": 1.2055253244035162e-05, "loss": 1.5987, "step": 289 }, { "epoch": 0.21245421245421245, "grad_norm": 1.0078125, "learning_rate": 1.2097111762243616e-05, "loss": 1.6075, "step": 290 }, { "epoch": 0.21318681318681318, "grad_norm": 1.34375, "learning_rate": 1.2138970280452072e-05, "loss": 1.6154, "step": 291 }, { "epoch": 0.2139194139194139, "grad_norm": 1.2265625, "learning_rate": 1.2180828798660527e-05, "loss": 1.6447, "step": 292 }, { "epoch": 0.21465201465201464, "grad_norm": 1.765625, "learning_rate": 1.2222687316868983e-05, "loss": 1.6281, "step": 293 }, { "epoch": 0.2153846153846154, "grad_norm": 1.515625, "learning_rate": 1.2264545835077439e-05, "loss": 1.6215, "step": 294 }, { "epoch": 0.21611721611721613, "grad_norm": 1.4609375, "learning_rate": 1.2306404353285894e-05, "loss": 1.6074, "step": 295 }, { "epoch": 0.21684981684981686, "grad_norm": 1.2734375, "learning_rate": 1.234826287149435e-05, "loss": 1.6019, "step": 296 }, { "epoch": 0.2175824175824176, "grad_norm": 1.046875, "learning_rate": 1.2390121389702806e-05, "loss": 1.6162, "step": 297 }, { "epoch": 0.21831501831501832, "grad_norm": 1.0390625, "learning_rate": 1.243197990791126e-05, "loss": 1.6226, "step": 298 }, { "epoch": 0.21904761904761905, "grad_norm": 1.390625, "learning_rate": 1.2473838426119716e-05, "loss": 1.6441, "step": 299 }, { "epoch": 0.21978021978021978, "grad_norm": 2.09375, "learning_rate": 1.251569694432817e-05, "loss": 1.5958, "step": 300 }, { "epoch": 0.2205128205128205, "grad_norm": 1.6328125, "learning_rate": 1.2557555462536627e-05, "loss": 1.5821, "step": 301 }, { "epoch": 0.22124542124542124, "grad_norm": 1.40625, "learning_rate": 1.2599413980745081e-05, "loss": 1.6263, "step": 302 }, { "epoch": 0.22197802197802197, "grad_norm": 1.7578125, "learning_rate": 1.2641272498953538e-05, "loss": 1.5894, "step": 303 }, { "epoch": 0.22271062271062272, "grad_norm": 1.28125, "learning_rate": 1.2683131017161992e-05, "loss": 1.5955, "step": 304 }, { "epoch": 0.22344322344322345, "grad_norm": 1.078125, "learning_rate": 1.272498953537045e-05, "loss": 1.6031, "step": 305 }, { "epoch": 0.22417582417582418, "grad_norm": 1.5390625, "learning_rate": 1.2766848053578904e-05, "loss": 1.6051, "step": 306 }, { "epoch": 0.22490842490842491, "grad_norm": 1.21875, "learning_rate": 1.280870657178736e-05, "loss": 1.5834, "step": 307 }, { "epoch": 0.22564102564102564, "grad_norm": 1.6796875, "learning_rate": 1.2850565089995814e-05, "loss": 1.5751, "step": 308 }, { "epoch": 0.22637362637362637, "grad_norm": 1.0859375, "learning_rate": 1.2892423608204271e-05, "loss": 1.5893, "step": 309 }, { "epoch": 0.2271062271062271, "grad_norm": 1.1875, "learning_rate": 1.2934282126412725e-05, "loss": 1.6025, "step": 310 }, { "epoch": 0.22783882783882783, "grad_norm": 1.28125, "learning_rate": 1.2976140644621182e-05, "loss": 1.5618, "step": 311 }, { "epoch": 0.22857142857142856, "grad_norm": 1.2734375, "learning_rate": 1.3017999162829636e-05, "loss": 1.5626, "step": 312 }, { "epoch": 0.2293040293040293, "grad_norm": 1.0703125, "learning_rate": 1.3059857681038092e-05, "loss": 1.5779, "step": 313 }, { "epoch": 0.23003663003663002, "grad_norm": 1.09375, "learning_rate": 1.3101716199246548e-05, "loss": 1.574, "step": 314 }, { "epoch": 0.23076923076923078, "grad_norm": 1.7578125, "learning_rate": 1.3143574717455004e-05, "loss": 1.6079, "step": 315 }, { "epoch": 0.2315018315018315, "grad_norm": 1.25, "learning_rate": 1.3185433235663458e-05, "loss": 1.5637, "step": 316 }, { "epoch": 0.23223443223443224, "grad_norm": 1.140625, "learning_rate": 1.3227291753871915e-05, "loss": 1.5575, "step": 317 }, { "epoch": 0.23296703296703297, "grad_norm": 1.65625, "learning_rate": 1.3269150272080369e-05, "loss": 1.5823, "step": 318 }, { "epoch": 0.2336996336996337, "grad_norm": 1.3046875, "learning_rate": 1.3311008790288823e-05, "loss": 1.593, "step": 319 }, { "epoch": 0.23443223443223443, "grad_norm": 1.3515625, "learning_rate": 1.335286730849728e-05, "loss": 1.5888, "step": 320 }, { "epoch": 0.23516483516483516, "grad_norm": 1.65625, "learning_rate": 1.3394725826705734e-05, "loss": 1.6091, "step": 321 }, { "epoch": 0.2358974358974359, "grad_norm": 1.609375, "learning_rate": 1.343658434491419e-05, "loss": 1.6055, "step": 322 }, { "epoch": 0.23663003663003662, "grad_norm": 1.0390625, "learning_rate": 1.3478442863122646e-05, "loss": 1.5758, "step": 323 }, { "epoch": 0.23736263736263735, "grad_norm": 1.046875, "learning_rate": 1.3520301381331102e-05, "loss": 1.5753, "step": 324 }, { "epoch": 0.23809523809523808, "grad_norm": 1.0, "learning_rate": 1.3562159899539556e-05, "loss": 1.5936, "step": 325 }, { "epoch": 0.23882783882783884, "grad_norm": 1.1015625, "learning_rate": 1.3604018417748013e-05, "loss": 1.5705, "step": 326 }, { "epoch": 0.23956043956043957, "grad_norm": 1.671875, "learning_rate": 1.3645876935956467e-05, "loss": 1.5734, "step": 327 }, { "epoch": 0.2402930402930403, "grad_norm": 1.0703125, "learning_rate": 1.3687735454164924e-05, "loss": 1.5946, "step": 328 }, { "epoch": 0.24102564102564103, "grad_norm": 1.2890625, "learning_rate": 1.3729593972373378e-05, "loss": 1.5878, "step": 329 }, { "epoch": 0.24175824175824176, "grad_norm": 1.140625, "learning_rate": 1.3771452490581834e-05, "loss": 1.6007, "step": 330 }, { "epoch": 0.2424908424908425, "grad_norm": 1.015625, "learning_rate": 1.3813311008790288e-05, "loss": 1.5742, "step": 331 }, { "epoch": 0.24322344322344322, "grad_norm": 1.1171875, "learning_rate": 1.3855169526998746e-05, "loss": 1.5746, "step": 332 }, { "epoch": 0.24395604395604395, "grad_norm": 1.03125, "learning_rate": 1.38970280452072e-05, "loss": 1.5869, "step": 333 }, { "epoch": 0.24468864468864468, "grad_norm": 1.3984375, "learning_rate": 1.3938886563415657e-05, "loss": 1.5367, "step": 334 }, { "epoch": 0.2454212454212454, "grad_norm": 1.28125, "learning_rate": 1.3980745081624111e-05, "loss": 1.5549, "step": 335 }, { "epoch": 0.24615384615384617, "grad_norm": 1.03125, "learning_rate": 1.4022603599832568e-05, "loss": 1.6003, "step": 336 }, { "epoch": 0.2468864468864469, "grad_norm": 2.484375, "learning_rate": 1.4064462118041022e-05, "loss": 1.5444, "step": 337 }, { "epoch": 0.24761904761904763, "grad_norm": 1.2578125, "learning_rate": 1.4106320636249478e-05, "loss": 1.5859, "step": 338 }, { "epoch": 0.24835164835164836, "grad_norm": 1.171875, "learning_rate": 1.4148179154457932e-05, "loss": 1.5927, "step": 339 }, { "epoch": 0.2490842490842491, "grad_norm": 1.5859375, "learning_rate": 1.4190037672666386e-05, "loss": 1.558, "step": 340 }, { "epoch": 0.24981684981684982, "grad_norm": 1.890625, "learning_rate": 1.4231896190874844e-05, "loss": 1.6225, "step": 341 }, { "epoch": 0.25054945054945055, "grad_norm": 1.125, "learning_rate": 1.4273754709083298e-05, "loss": 1.5577, "step": 342 }, { "epoch": 0.2512820512820513, "grad_norm": 1.5546875, "learning_rate": 1.4315613227291755e-05, "loss": 1.5562, "step": 343 }, { "epoch": 0.252014652014652, "grad_norm": 1.5, "learning_rate": 1.4357471745500209e-05, "loss": 1.5623, "step": 344 }, { "epoch": 0.25274725274725274, "grad_norm": 1.296875, "learning_rate": 1.4399330263708666e-05, "loss": 1.5888, "step": 345 }, { "epoch": 0.25347985347985347, "grad_norm": 1.4609375, "learning_rate": 1.444118878191712e-05, "loss": 1.5633, "step": 346 }, { "epoch": 0.2542124542124542, "grad_norm": 1.2109375, "learning_rate": 1.4483047300125576e-05, "loss": 1.566, "step": 347 }, { "epoch": 0.2549450549450549, "grad_norm": 2.390625, "learning_rate": 1.452490581833403e-05, "loss": 1.5016, "step": 348 }, { "epoch": 0.25567765567765566, "grad_norm": 1.453125, "learning_rate": 1.4566764336542488e-05, "loss": 1.5236, "step": 349 }, { "epoch": 0.2564102564102564, "grad_norm": 1.1484375, "learning_rate": 1.4608622854750942e-05, "loss": 1.5351, "step": 350 }, { "epoch": 0.2571428571428571, "grad_norm": 1.078125, "learning_rate": 1.4650481372959399e-05, "loss": 1.5536, "step": 351 }, { "epoch": 0.2578754578754579, "grad_norm": 1.53125, "learning_rate": 1.4692339891167853e-05, "loss": 1.5533, "step": 352 }, { "epoch": 0.25860805860805863, "grad_norm": 1.4375, "learning_rate": 1.473419840937631e-05, "loss": 1.5912, "step": 353 }, { "epoch": 0.25934065934065936, "grad_norm": 1.40625, "learning_rate": 1.4776056927584764e-05, "loss": 1.5461, "step": 354 }, { "epoch": 0.2600732600732601, "grad_norm": 1.0859375, "learning_rate": 1.481791544579322e-05, "loss": 1.5476, "step": 355 }, { "epoch": 0.2608058608058608, "grad_norm": 1.5390625, "learning_rate": 1.4859773964001674e-05, "loss": 1.513, "step": 356 }, { "epoch": 0.26153846153846155, "grad_norm": 1.5390625, "learning_rate": 1.4901632482210132e-05, "loss": 1.5286, "step": 357 }, { "epoch": 0.2622710622710623, "grad_norm": 1.1328125, "learning_rate": 1.4943491000418586e-05, "loss": 1.5737, "step": 358 }, { "epoch": 0.263003663003663, "grad_norm": 1.3125, "learning_rate": 1.4985349518627043e-05, "loss": 1.548, "step": 359 }, { "epoch": 0.26373626373626374, "grad_norm": 1.4453125, "learning_rate": 1.5027208036835497e-05, "loss": 1.5275, "step": 360 }, { "epoch": 0.2644688644688645, "grad_norm": 1.1953125, "learning_rate": 1.5069066555043951e-05, "loss": 1.5629, "step": 361 }, { "epoch": 0.2652014652014652, "grad_norm": 1.328125, "learning_rate": 1.5110925073252408e-05, "loss": 1.5367, "step": 362 }, { "epoch": 0.26593406593406593, "grad_norm": 1.2421875, "learning_rate": 1.5152783591460862e-05, "loss": 1.5687, "step": 363 }, { "epoch": 0.26666666666666666, "grad_norm": 1.234375, "learning_rate": 1.5194642109669318e-05, "loss": 1.4958, "step": 364 }, { "epoch": 0.2673992673992674, "grad_norm": 1.25, "learning_rate": 1.5236500627877772e-05, "loss": 1.5374, "step": 365 }, { "epoch": 0.2681318681318681, "grad_norm": 1.140625, "learning_rate": 1.527835914608623e-05, "loss": 1.5444, "step": 366 }, { "epoch": 0.26886446886446885, "grad_norm": 1.0234375, "learning_rate": 1.5320217664294685e-05, "loss": 1.5245, "step": 367 }, { "epoch": 0.2695970695970696, "grad_norm": 1.15625, "learning_rate": 1.536207618250314e-05, "loss": 1.531, "step": 368 }, { "epoch": 0.2703296703296703, "grad_norm": 1.1328125, "learning_rate": 1.5403934700711593e-05, "loss": 1.5261, "step": 369 }, { "epoch": 0.27106227106227104, "grad_norm": 1.1640625, "learning_rate": 1.544579321892005e-05, "loss": 1.5353, "step": 370 }, { "epoch": 0.2717948717948718, "grad_norm": 1.15625, "learning_rate": 1.5487651737128505e-05, "loss": 1.5146, "step": 371 }, { "epoch": 0.2725274725274725, "grad_norm": 1.21875, "learning_rate": 1.5529510255336962e-05, "loss": 1.5077, "step": 372 }, { "epoch": 0.27326007326007323, "grad_norm": 1.109375, "learning_rate": 1.5571368773545416e-05, "loss": 1.5622, "step": 373 }, { "epoch": 0.273992673992674, "grad_norm": 1.125, "learning_rate": 1.5613227291753874e-05, "loss": 1.5484, "step": 374 }, { "epoch": 0.27472527472527475, "grad_norm": 1.3359375, "learning_rate": 1.5655085809962328e-05, "loss": 1.5554, "step": 375 }, { "epoch": 0.2754578754578755, "grad_norm": 1.265625, "learning_rate": 1.5696944328170785e-05, "loss": 1.5033, "step": 376 }, { "epoch": 0.2761904761904762, "grad_norm": 1.2578125, "learning_rate": 1.573880284637924e-05, "loss": 1.5279, "step": 377 }, { "epoch": 0.27692307692307694, "grad_norm": 1.1875, "learning_rate": 1.5780661364587696e-05, "loss": 1.5293, "step": 378 }, { "epoch": 0.27765567765567767, "grad_norm": 1.1640625, "learning_rate": 1.582251988279615e-05, "loss": 1.5268, "step": 379 }, { "epoch": 0.2783882783882784, "grad_norm": 1.2734375, "learning_rate": 1.5864378401004608e-05, "loss": 1.5367, "step": 380 }, { "epoch": 0.27912087912087913, "grad_norm": 1.5703125, "learning_rate": 1.5906236919213062e-05, "loss": 1.4828, "step": 381 }, { "epoch": 0.27985347985347986, "grad_norm": 1.984375, "learning_rate": 1.5948095437421516e-05, "loss": 1.5289, "step": 382 }, { "epoch": 0.2805860805860806, "grad_norm": 1.2265625, "learning_rate": 1.598995395562997e-05, "loss": 1.509, "step": 383 }, { "epoch": 0.2813186813186813, "grad_norm": 1.3046875, "learning_rate": 1.6031812473838427e-05, "loss": 1.4955, "step": 384 }, { "epoch": 0.28205128205128205, "grad_norm": 1.4453125, "learning_rate": 1.607367099204688e-05, "loss": 1.5023, "step": 385 }, { "epoch": 0.2827838827838828, "grad_norm": 1.640625, "learning_rate": 1.6115529510255335e-05, "loss": 1.4896, "step": 386 }, { "epoch": 0.2835164835164835, "grad_norm": 1.7890625, "learning_rate": 1.6157388028463793e-05, "loss": 1.4884, "step": 387 }, { "epoch": 0.28424908424908424, "grad_norm": 1.9921875, "learning_rate": 1.6199246546672247e-05, "loss": 1.5339, "step": 388 }, { "epoch": 0.28498168498168497, "grad_norm": 2.15625, "learning_rate": 1.6241105064880704e-05, "loss": 1.5614, "step": 389 }, { "epoch": 0.2857142857142857, "grad_norm": 1.484375, "learning_rate": 1.6282963583089158e-05, "loss": 1.4962, "step": 390 }, { "epoch": 0.28644688644688643, "grad_norm": 1.265625, "learning_rate": 1.6324822101297616e-05, "loss": 1.4824, "step": 391 }, { "epoch": 0.28717948717948716, "grad_norm": 1.375, "learning_rate": 1.636668061950607e-05, "loss": 1.5065, "step": 392 }, { "epoch": 0.2879120879120879, "grad_norm": 1.2734375, "learning_rate": 1.6408539137714527e-05, "loss": 1.4952, "step": 393 }, { "epoch": 0.2886446886446886, "grad_norm": 1.0625, "learning_rate": 1.645039765592298e-05, "loss": 1.4883, "step": 394 }, { "epoch": 0.2893772893772894, "grad_norm": 1.78125, "learning_rate": 1.649225617413144e-05, "loss": 1.5145, "step": 395 }, { "epoch": 0.29010989010989013, "grad_norm": 1.296875, "learning_rate": 1.6534114692339892e-05, "loss": 1.4926, "step": 396 }, { "epoch": 0.29084249084249086, "grad_norm": 1.5859375, "learning_rate": 1.657597321054835e-05, "loss": 1.5025, "step": 397 }, { "epoch": 0.2915750915750916, "grad_norm": 1.1640625, "learning_rate": 1.6617831728756804e-05, "loss": 1.4813, "step": 398 }, { "epoch": 0.2923076923076923, "grad_norm": 1.296875, "learning_rate": 1.6659690246965258e-05, "loss": 1.481, "step": 399 }, { "epoch": 0.29304029304029305, "grad_norm": 1.2890625, "learning_rate": 1.6701548765173712e-05, "loss": 1.5119, "step": 400 }, { "epoch": 0.2937728937728938, "grad_norm": 1.5078125, "learning_rate": 1.674340728338217e-05, "loss": 1.4906, "step": 401 }, { "epoch": 0.2945054945054945, "grad_norm": 1.1953125, "learning_rate": 1.6785265801590623e-05, "loss": 1.4518, "step": 402 }, { "epoch": 0.29523809523809524, "grad_norm": 1.4296875, "learning_rate": 1.6827124319799077e-05, "loss": 1.4641, "step": 403 }, { "epoch": 0.295970695970696, "grad_norm": 2.0625, "learning_rate": 1.6868982838007535e-05, "loss": 1.4893, "step": 404 }, { "epoch": 0.2967032967032967, "grad_norm": 1.2421875, "learning_rate": 1.691084135621599e-05, "loss": 1.4719, "step": 405 }, { "epoch": 0.29743589743589743, "grad_norm": 1.3203125, "learning_rate": 1.6952699874424446e-05, "loss": 1.4846, "step": 406 }, { "epoch": 0.29816849816849816, "grad_norm": 1.390625, "learning_rate": 1.69945583926329e-05, "loss": 1.5131, "step": 407 }, { "epoch": 0.2989010989010989, "grad_norm": 1.8671875, "learning_rate": 1.7036416910841358e-05, "loss": 1.456, "step": 408 }, { "epoch": 0.2996336996336996, "grad_norm": 1.5859375, "learning_rate": 1.707827542904981e-05, "loss": 1.5172, "step": 409 }, { "epoch": 0.30036630036630035, "grad_norm": 1.2265625, "learning_rate": 1.712013394725827e-05, "loss": 1.4653, "step": 410 }, { "epoch": 0.3010989010989011, "grad_norm": 1.34375, "learning_rate": 1.7161992465466723e-05, "loss": 1.4627, "step": 411 }, { "epoch": 0.3018315018315018, "grad_norm": 1.671875, "learning_rate": 1.720385098367518e-05, "loss": 1.4591, "step": 412 }, { "epoch": 0.30256410256410254, "grad_norm": 1.9296875, "learning_rate": 1.7245709501883634e-05, "loss": 1.4295, "step": 413 }, { "epoch": 0.3032967032967033, "grad_norm": 1.2421875, "learning_rate": 1.7287568020092092e-05, "loss": 1.4675, "step": 414 }, { "epoch": 0.304029304029304, "grad_norm": 1.765625, "learning_rate": 1.7329426538300546e-05, "loss": 1.5205, "step": 415 }, { "epoch": 0.3047619047619048, "grad_norm": 1.6484375, "learning_rate": 1.7371285056509e-05, "loss": 1.4727, "step": 416 }, { "epoch": 0.3054945054945055, "grad_norm": 1.1171875, "learning_rate": 1.7413143574717454e-05, "loss": 1.4564, "step": 417 }, { "epoch": 0.30622710622710625, "grad_norm": 1.4140625, "learning_rate": 1.745500209292591e-05, "loss": 1.4741, "step": 418 }, { "epoch": 0.306959706959707, "grad_norm": 1.1484375, "learning_rate": 1.7496860611134365e-05, "loss": 1.4664, "step": 419 }, { "epoch": 0.3076923076923077, "grad_norm": 2.296875, "learning_rate": 1.7538719129342823e-05, "loss": 1.4326, "step": 420 }, { "epoch": 0.30842490842490844, "grad_norm": 1.40625, "learning_rate": 1.7580577647551277e-05, "loss": 1.4744, "step": 421 }, { "epoch": 0.30915750915750917, "grad_norm": 2.140625, "learning_rate": 1.7622436165759734e-05, "loss": 1.4416, "step": 422 }, { "epoch": 0.3098901098901099, "grad_norm": 1.3984375, "learning_rate": 1.7664294683968188e-05, "loss": 1.452, "step": 423 }, { "epoch": 0.31062271062271063, "grad_norm": 1.2578125, "learning_rate": 1.7706153202176642e-05, "loss": 1.4483, "step": 424 }, { "epoch": 0.31135531135531136, "grad_norm": 1.8203125, "learning_rate": 1.77480117203851e-05, "loss": 1.4558, "step": 425 }, { "epoch": 0.3120879120879121, "grad_norm": 1.4765625, "learning_rate": 1.7789870238593554e-05, "loss": 1.4622, "step": 426 }, { "epoch": 0.3128205128205128, "grad_norm": 3.078125, "learning_rate": 1.783172875680201e-05, "loss": 1.4843, "step": 427 }, { "epoch": 0.31355311355311355, "grad_norm": 1.421875, "learning_rate": 1.7873587275010465e-05, "loss": 1.452, "step": 428 }, { "epoch": 0.3142857142857143, "grad_norm": 1.5078125, "learning_rate": 1.7915445793218922e-05, "loss": 1.4185, "step": 429 }, { "epoch": 0.315018315018315, "grad_norm": 1.453125, "learning_rate": 1.7957304311427376e-05, "loss": 1.459, "step": 430 }, { "epoch": 0.31575091575091574, "grad_norm": 1.8359375, "learning_rate": 1.7999162829635834e-05, "loss": 1.4157, "step": 431 }, { "epoch": 0.31648351648351647, "grad_norm": 1.2265625, "learning_rate": 1.8041021347844288e-05, "loss": 1.4447, "step": 432 }, { "epoch": 0.3172161172161172, "grad_norm": 1.625, "learning_rate": 1.8082879866052742e-05, "loss": 1.4172, "step": 433 }, { "epoch": 0.31794871794871793, "grad_norm": 1.2890625, "learning_rate": 1.8124738384261196e-05, "loss": 1.4498, "step": 434 }, { "epoch": 0.31868131868131866, "grad_norm": 1.4921875, "learning_rate": 1.8166596902469653e-05, "loss": 1.4032, "step": 435 }, { "epoch": 0.3194139194139194, "grad_norm": 1.453125, "learning_rate": 1.8208455420678107e-05, "loss": 1.4737, "step": 436 }, { "epoch": 0.3201465201465201, "grad_norm": 1.4140625, "learning_rate": 1.8250313938886565e-05, "loss": 1.4287, "step": 437 }, { "epoch": 0.3208791208791209, "grad_norm": 1.8359375, "learning_rate": 1.829217245709502e-05, "loss": 1.4763, "step": 438 }, { "epoch": 0.32161172161172163, "grad_norm": 2.1875, "learning_rate": 1.8334030975303476e-05, "loss": 1.4565, "step": 439 }, { "epoch": 0.32234432234432236, "grad_norm": 1.8203125, "learning_rate": 1.837588949351193e-05, "loss": 1.4473, "step": 440 }, { "epoch": 0.3230769230769231, "grad_norm": 1.4375, "learning_rate": 1.8417748011720388e-05, "loss": 1.4416, "step": 441 }, { "epoch": 0.3238095238095238, "grad_norm": 1.7109375, "learning_rate": 1.845960652992884e-05, "loss": 1.3986, "step": 442 }, { "epoch": 0.32454212454212455, "grad_norm": 1.296875, "learning_rate": 1.85014650481373e-05, "loss": 1.4102, "step": 443 }, { "epoch": 0.3252747252747253, "grad_norm": 1.9296875, "learning_rate": 1.8543323566345753e-05, "loss": 1.4236, "step": 444 }, { "epoch": 0.326007326007326, "grad_norm": 1.9140625, "learning_rate": 1.8585182084554207e-05, "loss": 1.4016, "step": 445 }, { "epoch": 0.32673992673992674, "grad_norm": 1.7109375, "learning_rate": 1.8627040602762664e-05, "loss": 1.3783, "step": 446 }, { "epoch": 0.3274725274725275, "grad_norm": 1.328125, "learning_rate": 1.866889912097112e-05, "loss": 1.424, "step": 447 }, { "epoch": 0.3282051282051282, "grad_norm": 1.5234375, "learning_rate": 1.8710757639179573e-05, "loss": 1.4367, "step": 448 }, { "epoch": 0.32893772893772893, "grad_norm": 1.359375, "learning_rate": 1.875261615738803e-05, "loss": 1.4324, "step": 449 }, { "epoch": 0.32967032967032966, "grad_norm": 1.421875, "learning_rate": 1.8794474675596484e-05, "loss": 1.3876, "step": 450 }, { "epoch": 0.3304029304029304, "grad_norm": 1.25, "learning_rate": 1.8836333193804938e-05, "loss": 1.4331, "step": 451 }, { "epoch": 0.3311355311355311, "grad_norm": 1.1796875, "learning_rate": 1.8878191712013395e-05, "loss": 1.4122, "step": 452 }, { "epoch": 0.33186813186813185, "grad_norm": 1.9609375, "learning_rate": 1.892005023022185e-05, "loss": 1.3928, "step": 453 }, { "epoch": 0.3326007326007326, "grad_norm": 1.375, "learning_rate": 1.8961908748430307e-05, "loss": 1.3864, "step": 454 }, { "epoch": 0.3333333333333333, "grad_norm": 1.96875, "learning_rate": 1.900376726663876e-05, "loss": 1.4235, "step": 455 }, { "epoch": 0.33406593406593404, "grad_norm": 1.328125, "learning_rate": 1.9045625784847218e-05, "loss": 1.3984, "step": 456 }, { "epoch": 0.3347985347985348, "grad_norm": 1.9609375, "learning_rate": 1.9087484303055672e-05, "loss": 1.3639, "step": 457 }, { "epoch": 0.3355311355311355, "grad_norm": 2.078125, "learning_rate": 1.912934282126413e-05, "loss": 1.3799, "step": 458 }, { "epoch": 0.3362637362637363, "grad_norm": 1.484375, "learning_rate": 1.9171201339472584e-05, "loss": 1.3909, "step": 459 }, { "epoch": 0.336996336996337, "grad_norm": 1.265625, "learning_rate": 1.921305985768104e-05, "loss": 1.3956, "step": 460 }, { "epoch": 0.33772893772893775, "grad_norm": 2.1875, "learning_rate": 1.9254918375889495e-05, "loss": 1.458, "step": 461 }, { "epoch": 0.3384615384615385, "grad_norm": 1.7734375, "learning_rate": 1.9296776894097952e-05, "loss": 1.3949, "step": 462 }, { "epoch": 0.3391941391941392, "grad_norm": 1.375, "learning_rate": 1.9338635412306406e-05, "loss": 1.4004, "step": 463 }, { "epoch": 0.33992673992673994, "grad_norm": 1.7890625, "learning_rate": 1.938049393051486e-05, "loss": 1.4168, "step": 464 }, { "epoch": 0.34065934065934067, "grad_norm": 1.8984375, "learning_rate": 1.9422352448723315e-05, "loss": 1.3823, "step": 465 }, { "epoch": 0.3413919413919414, "grad_norm": 2.984375, "learning_rate": 1.9464210966931772e-05, "loss": 1.3461, "step": 466 }, { "epoch": 0.34212454212454213, "grad_norm": 1.2890625, "learning_rate": 1.9506069485140226e-05, "loss": 1.4025, "step": 467 }, { "epoch": 0.34285714285714286, "grad_norm": 2.09375, "learning_rate": 1.954792800334868e-05, "loss": 1.3645, "step": 468 }, { "epoch": 0.3435897435897436, "grad_norm": 2.265625, "learning_rate": 1.9589786521557137e-05, "loss": 1.3404, "step": 469 }, { "epoch": 0.3443223443223443, "grad_norm": 2.203125, "learning_rate": 1.963164503976559e-05, "loss": 1.4114, "step": 470 }, { "epoch": 0.34505494505494505, "grad_norm": 1.9765625, "learning_rate": 1.967350355797405e-05, "loss": 1.3548, "step": 471 }, { "epoch": 0.3457875457875458, "grad_norm": 1.703125, "learning_rate": 1.9715362076182503e-05, "loss": 1.357, "step": 472 }, { "epoch": 0.3465201465201465, "grad_norm": 1.3359375, "learning_rate": 1.975722059439096e-05, "loss": 1.374, "step": 473 }, { "epoch": 0.34725274725274724, "grad_norm": 1.859375, "learning_rate": 1.9799079112599414e-05, "loss": 1.3766, "step": 474 }, { "epoch": 0.34798534798534797, "grad_norm": 1.953125, "learning_rate": 1.984093763080787e-05, "loss": 1.3497, "step": 475 }, { "epoch": 0.3487179487179487, "grad_norm": 2.171875, "learning_rate": 1.9882796149016326e-05, "loss": 1.3568, "step": 476 }, { "epoch": 0.34945054945054943, "grad_norm": 2.359375, "learning_rate": 1.9924654667224783e-05, "loss": 1.3578, "step": 477 }, { "epoch": 0.35018315018315016, "grad_norm": 1.359375, "learning_rate": 1.9966513185433237e-05, "loss": 1.3768, "step": 478 }, { "epoch": 0.3509157509157509, "grad_norm": 2.046875, "learning_rate": 2.0008371703641694e-05, "loss": 1.3779, "step": 479 }, { "epoch": 0.3516483516483517, "grad_norm": 1.1875, "learning_rate": 2.005023022185015e-05, "loss": 1.3438, "step": 480 }, { "epoch": 0.3523809523809524, "grad_norm": 1.9921875, "learning_rate": 2.0092088740058603e-05, "loss": 1.3324, "step": 481 }, { "epoch": 0.35311355311355314, "grad_norm": 1.3125, "learning_rate": 2.0133947258267057e-05, "loss": 1.317, "step": 482 }, { "epoch": 0.35384615384615387, "grad_norm": 1.3671875, "learning_rate": 2.0175805776475514e-05, "loss": 1.3613, "step": 483 }, { "epoch": 0.3545787545787546, "grad_norm": 2.671875, "learning_rate": 2.0217664294683968e-05, "loss": 1.355, "step": 484 }, { "epoch": 0.3553113553113553, "grad_norm": 1.40625, "learning_rate": 2.0259522812892425e-05, "loss": 1.3634, "step": 485 }, { "epoch": 0.35604395604395606, "grad_norm": 2.25, "learning_rate": 2.030138133110088e-05, "loss": 1.3416, "step": 486 }, { "epoch": 0.3567765567765568, "grad_norm": 1.421875, "learning_rate": 2.0343239849309337e-05, "loss": 1.3255, "step": 487 }, { "epoch": 0.3575091575091575, "grad_norm": 1.703125, "learning_rate": 2.038509836751779e-05, "loss": 1.3932, "step": 488 }, { "epoch": 0.35824175824175825, "grad_norm": 1.6328125, "learning_rate": 2.0426956885726245e-05, "loss": 1.3194, "step": 489 }, { "epoch": 0.358974358974359, "grad_norm": 1.9140625, "learning_rate": 2.0468815403934702e-05, "loss": 1.3289, "step": 490 }, { "epoch": 0.3597069597069597, "grad_norm": 3.5625, "learning_rate": 2.0510673922143156e-05, "loss": 1.3108, "step": 491 }, { "epoch": 0.36043956043956044, "grad_norm": 2.015625, "learning_rate": 2.0552532440351614e-05, "loss": 1.3333, "step": 492 }, { "epoch": 0.36117216117216117, "grad_norm": 1.8046875, "learning_rate": 2.0594390958560068e-05, "loss": 1.3418, "step": 493 }, { "epoch": 0.3619047619047619, "grad_norm": 1.4296875, "learning_rate": 2.0636249476768525e-05, "loss": 1.2606, "step": 494 }, { "epoch": 0.3626373626373626, "grad_norm": 2.28125, "learning_rate": 2.067810799497698e-05, "loss": 1.3703, "step": 495 }, { "epoch": 0.36336996336996336, "grad_norm": 1.890625, "learning_rate": 2.0719966513185436e-05, "loss": 1.3386, "step": 496 }, { "epoch": 0.3641025641025641, "grad_norm": 1.3984375, "learning_rate": 2.076182503139389e-05, "loss": 1.3046, "step": 497 }, { "epoch": 0.3648351648351648, "grad_norm": 1.859375, "learning_rate": 2.0803683549602345e-05, "loss": 1.3501, "step": 498 }, { "epoch": 0.36556776556776555, "grad_norm": 1.78125, "learning_rate": 2.08455420678108e-05, "loss": 1.3194, "step": 499 }, { "epoch": 0.3663003663003663, "grad_norm": 1.5078125, "learning_rate": 2.0887400586019256e-05, "loss": 1.2773, "step": 500 }, { "epoch": 0.367032967032967, "grad_norm": 1.34375, "learning_rate": 2.092925910422771e-05, "loss": 1.3354, "step": 501 }, { "epoch": 0.3677655677655678, "grad_norm": 2.015625, "learning_rate": 2.0971117622436167e-05, "loss": 1.3762, "step": 502 }, { "epoch": 0.3684981684981685, "grad_norm": 1.625, "learning_rate": 2.101297614064462e-05, "loss": 1.3254, "step": 503 }, { "epoch": 0.36923076923076925, "grad_norm": 1.578125, "learning_rate": 2.105483465885308e-05, "loss": 1.29, "step": 504 }, { "epoch": 0.36996336996337, "grad_norm": 1.8203125, "learning_rate": 2.1096693177061533e-05, "loss": 1.2952, "step": 505 }, { "epoch": 0.3706959706959707, "grad_norm": 1.578125, "learning_rate": 2.113855169526999e-05, "loss": 1.2509, "step": 506 }, { "epoch": 0.37142857142857144, "grad_norm": 2.234375, "learning_rate": 2.1180410213478444e-05, "loss": 1.2967, "step": 507 }, { "epoch": 0.37216117216117217, "grad_norm": 2.640625, "learning_rate": 2.12222687316869e-05, "loss": 1.335, "step": 508 }, { "epoch": 0.3728937728937729, "grad_norm": 1.5625, "learning_rate": 2.1264127249895356e-05, "loss": 1.3212, "step": 509 }, { "epoch": 0.37362637362637363, "grad_norm": 1.515625, "learning_rate": 2.130598576810381e-05, "loss": 1.2908, "step": 510 }, { "epoch": 0.37435897435897436, "grad_norm": 3.09375, "learning_rate": 2.1347844286312267e-05, "loss": 1.2296, "step": 511 }, { "epoch": 0.3750915750915751, "grad_norm": 2.671875, "learning_rate": 2.138970280452072e-05, "loss": 1.2574, "step": 512 }, { "epoch": 0.3758241758241758, "grad_norm": 1.6796875, "learning_rate": 2.1431561322729175e-05, "loss": 1.3103, "step": 513 }, { "epoch": 0.37655677655677655, "grad_norm": 2.359375, "learning_rate": 2.1473419840937633e-05, "loss": 1.2947, "step": 514 }, { "epoch": 0.3772893772893773, "grad_norm": 1.359375, "learning_rate": 2.1515278359146087e-05, "loss": 1.28, "step": 515 }, { "epoch": 0.378021978021978, "grad_norm": 1.671875, "learning_rate": 2.155713687735454e-05, "loss": 1.2897, "step": 516 }, { "epoch": 0.37875457875457874, "grad_norm": 2.015625, "learning_rate": 2.1598995395562998e-05, "loss": 1.3012, "step": 517 }, { "epoch": 0.37948717948717947, "grad_norm": 1.9140625, "learning_rate": 2.1640853913771452e-05, "loss": 1.2505, "step": 518 }, { "epoch": 0.3802197802197802, "grad_norm": 1.921875, "learning_rate": 2.168271243197991e-05, "loss": 1.3242, "step": 519 }, { "epoch": 0.38095238095238093, "grad_norm": 2.109375, "learning_rate": 2.1724570950188363e-05, "loss": 1.2334, "step": 520 }, { "epoch": 0.38168498168498166, "grad_norm": 1.5546875, "learning_rate": 2.176642946839682e-05, "loss": 1.2424, "step": 521 }, { "epoch": 0.3824175824175824, "grad_norm": 1.4765625, "learning_rate": 2.1808287986605275e-05, "loss": 1.2862, "step": 522 }, { "epoch": 0.3831501831501832, "grad_norm": 1.7578125, "learning_rate": 2.1850146504813732e-05, "loss": 1.2547, "step": 523 }, { "epoch": 0.3838827838827839, "grad_norm": 1.7578125, "learning_rate": 2.1892005023022186e-05, "loss": 1.2815, "step": 524 }, { "epoch": 0.38461538461538464, "grad_norm": 1.4375, "learning_rate": 2.1933863541230644e-05, "loss": 1.2207, "step": 525 }, { "epoch": 0.38534798534798537, "grad_norm": 1.53125, "learning_rate": 2.1975722059439098e-05, "loss": 1.2369, "step": 526 }, { "epoch": 0.3860805860805861, "grad_norm": 1.34375, "learning_rate": 2.2017580577647555e-05, "loss": 1.2287, "step": 527 }, { "epoch": 0.3868131868131868, "grad_norm": 1.53125, "learning_rate": 2.205943909585601e-05, "loss": 1.2215, "step": 528 }, { "epoch": 0.38754578754578756, "grad_norm": 1.390625, "learning_rate": 2.2101297614064463e-05, "loss": 1.2366, "step": 529 }, { "epoch": 0.3882783882783883, "grad_norm": 1.859375, "learning_rate": 2.2143156132272917e-05, "loss": 1.2683, "step": 530 }, { "epoch": 0.389010989010989, "grad_norm": 1.8046875, "learning_rate": 2.218501465048137e-05, "loss": 1.212, "step": 531 }, { "epoch": 0.38974358974358975, "grad_norm": 1.765625, "learning_rate": 2.222687316868983e-05, "loss": 1.1994, "step": 532 }, { "epoch": 0.3904761904761905, "grad_norm": 1.4140625, "learning_rate": 2.2268731686898283e-05, "loss": 1.1962, "step": 533 }, { "epoch": 0.3912087912087912, "grad_norm": 2.03125, "learning_rate": 2.231059020510674e-05, "loss": 1.144, "step": 534 }, { "epoch": 0.39194139194139194, "grad_norm": 2.328125, "learning_rate": 2.2352448723315194e-05, "loss": 1.1611, "step": 535 }, { "epoch": 0.39267399267399267, "grad_norm": 1.8125, "learning_rate": 2.239430724152365e-05, "loss": 1.211, "step": 536 }, { "epoch": 0.3934065934065934, "grad_norm": 1.5859375, "learning_rate": 2.2436165759732105e-05, "loss": 1.2532, "step": 537 }, { "epoch": 0.3941391941391941, "grad_norm": 1.453125, "learning_rate": 2.2478024277940563e-05, "loss": 1.1763, "step": 538 }, { "epoch": 0.39487179487179486, "grad_norm": 1.9609375, "learning_rate": 2.2519882796149017e-05, "loss": 1.1381, "step": 539 }, { "epoch": 0.3956043956043956, "grad_norm": 1.8359375, "learning_rate": 2.2561741314357474e-05, "loss": 1.1658, "step": 540 }, { "epoch": 0.3963369963369963, "grad_norm": 1.4765625, "learning_rate": 2.2603599832565928e-05, "loss": 1.1605, "step": 541 }, { "epoch": 0.39706959706959705, "grad_norm": 1.6875, "learning_rate": 2.2645458350774386e-05, "loss": 1.1761, "step": 542 }, { "epoch": 0.3978021978021978, "grad_norm": 1.6328125, "learning_rate": 2.268731686898284e-05, "loss": 1.1002, "step": 543 }, { "epoch": 0.39853479853479856, "grad_norm": 1.4921875, "learning_rate": 2.2729175387191297e-05, "loss": 1.1255, "step": 544 }, { "epoch": 0.3992673992673993, "grad_norm": 1.921875, "learning_rate": 2.277103390539975e-05, "loss": 1.1235, "step": 545 }, { "epoch": 0.4, "grad_norm": 1.7421875, "learning_rate": 2.2812892423608205e-05, "loss": 1.1066, "step": 546 }, { "epoch": 0.40073260073260075, "grad_norm": 2.53125, "learning_rate": 2.285475094181666e-05, "loss": 1.0965, "step": 547 }, { "epoch": 0.4014652014652015, "grad_norm": 1.5078125, "learning_rate": 2.2896609460025117e-05, "loss": 1.0905, "step": 548 }, { "epoch": 0.4021978021978022, "grad_norm": 2.0625, "learning_rate": 2.293846797823357e-05, "loss": 1.0397, "step": 549 }, { "epoch": 0.40293040293040294, "grad_norm": 1.4375, "learning_rate": 2.2980326496442028e-05, "loss": 1.0002, "step": 550 }, { "epoch": 0.40366300366300367, "grad_norm": 2.203125, "learning_rate": 2.3022185014650482e-05, "loss": 1.0757, "step": 551 }, { "epoch": 0.4043956043956044, "grad_norm": 1.7109375, "learning_rate": 2.3064043532858936e-05, "loss": 1.0188, "step": 552 }, { "epoch": 0.40512820512820513, "grad_norm": 1.5859375, "learning_rate": 2.3105902051067393e-05, "loss": 1.0276, "step": 553 }, { "epoch": 0.40586080586080586, "grad_norm": 1.3671875, "learning_rate": 2.3147760569275847e-05, "loss": 1.0576, "step": 554 }, { "epoch": 0.4065934065934066, "grad_norm": 1.34375, "learning_rate": 2.3189619087484305e-05, "loss": 1.0385, "step": 555 }, { "epoch": 0.4073260073260073, "grad_norm": 1.6640625, "learning_rate": 2.323147760569276e-05, "loss": 0.9705, "step": 556 }, { "epoch": 0.40805860805860805, "grad_norm": 1.6015625, "learning_rate": 2.3273336123901216e-05, "loss": 0.9884, "step": 557 }, { "epoch": 0.4087912087912088, "grad_norm": 2.03125, "learning_rate": 2.331519464210967e-05, "loss": 0.9941, "step": 558 }, { "epoch": 0.4095238095238095, "grad_norm": 1.53125, "learning_rate": 2.3357053160318128e-05, "loss": 0.9958, "step": 559 }, { "epoch": 0.41025641025641024, "grad_norm": 1.453125, "learning_rate": 2.339891167852658e-05, "loss": 1.0317, "step": 560 }, { "epoch": 0.41098901098901097, "grad_norm": 1.640625, "learning_rate": 2.3440770196735036e-05, "loss": 0.9774, "step": 561 }, { "epoch": 0.4117216117216117, "grad_norm": 1.53125, "learning_rate": 2.3482628714943493e-05, "loss": 1.0074, "step": 562 }, { "epoch": 0.41245421245421243, "grad_norm": 1.7265625, "learning_rate": 2.3524487233151947e-05, "loss": 0.9755, "step": 563 }, { "epoch": 0.41318681318681316, "grad_norm": 1.46875, "learning_rate": 2.35663457513604e-05, "loss": 0.9996, "step": 564 }, { "epoch": 0.4139194139194139, "grad_norm": 1.3828125, "learning_rate": 2.360820426956886e-05, "loss": 0.9402, "step": 565 }, { "epoch": 0.4146520146520147, "grad_norm": 1.9296875, "learning_rate": 2.3650062787777313e-05, "loss": 0.9545, "step": 566 }, { "epoch": 0.4153846153846154, "grad_norm": 1.875, "learning_rate": 2.369192130598577e-05, "loss": 1.0163, "step": 567 }, { "epoch": 0.41611721611721614, "grad_norm": 1.609375, "learning_rate": 2.3733779824194224e-05, "loss": 0.9353, "step": 568 }, { "epoch": 0.41684981684981687, "grad_norm": 1.3671875, "learning_rate": 2.377563834240268e-05, "loss": 0.8826, "step": 569 }, { "epoch": 0.4175824175824176, "grad_norm": 1.390625, "learning_rate": 2.3817496860611135e-05, "loss": 0.9505, "step": 570 }, { "epoch": 0.4183150183150183, "grad_norm": 1.6484375, "learning_rate": 2.3859355378819593e-05, "loss": 0.9839, "step": 571 }, { "epoch": 0.41904761904761906, "grad_norm": 1.6953125, "learning_rate": 2.3901213897028047e-05, "loss": 0.9468, "step": 572 }, { "epoch": 0.4197802197802198, "grad_norm": 2.4375, "learning_rate": 2.39430724152365e-05, "loss": 0.9439, "step": 573 }, { "epoch": 0.4205128205128205, "grad_norm": 1.7734375, "learning_rate": 2.3984930933444958e-05, "loss": 0.898, "step": 574 }, { "epoch": 0.42124542124542125, "grad_norm": 1.6484375, "learning_rate": 2.4026789451653412e-05, "loss": 0.9185, "step": 575 }, { "epoch": 0.421978021978022, "grad_norm": 1.5234375, "learning_rate": 2.406864796986187e-05, "loss": 0.9116, "step": 576 }, { "epoch": 0.4227106227106227, "grad_norm": 1.703125, "learning_rate": 2.4110506488070324e-05, "loss": 0.9267, "step": 577 }, { "epoch": 0.42344322344322344, "grad_norm": 1.859375, "learning_rate": 2.4152365006278778e-05, "loss": 0.8702, "step": 578 }, { "epoch": 0.42417582417582417, "grad_norm": 1.6015625, "learning_rate": 2.4194223524487232e-05, "loss": 0.8393, "step": 579 }, { "epoch": 0.4249084249084249, "grad_norm": 1.2578125, "learning_rate": 2.423608204269569e-05, "loss": 0.8074, "step": 580 }, { "epoch": 0.4256410256410256, "grad_norm": 1.5234375, "learning_rate": 2.4277940560904143e-05, "loss": 0.9369, "step": 581 }, { "epoch": 0.42637362637362636, "grad_norm": 1.8671875, "learning_rate": 2.43197990791126e-05, "loss": 0.9256, "step": 582 }, { "epoch": 0.4271062271062271, "grad_norm": 1.9453125, "learning_rate": 2.4361657597321055e-05, "loss": 0.9056, "step": 583 }, { "epoch": 0.4278388278388278, "grad_norm": 1.75, "learning_rate": 2.4403516115529512e-05, "loss": 0.8356, "step": 584 }, { "epoch": 0.42857142857142855, "grad_norm": 1.5, "learning_rate": 2.4445374633737966e-05, "loss": 0.8328, "step": 585 }, { "epoch": 0.4293040293040293, "grad_norm": 1.7890625, "learning_rate": 2.4487233151946423e-05, "loss": 0.8395, "step": 586 }, { "epoch": 0.43003663003663006, "grad_norm": 1.671875, "learning_rate": 2.4529091670154877e-05, "loss": 0.8964, "step": 587 }, { "epoch": 0.4307692307692308, "grad_norm": 1.6640625, "learning_rate": 2.4570950188363335e-05, "loss": 0.8159, "step": 588 }, { "epoch": 0.4315018315018315, "grad_norm": 1.8046875, "learning_rate": 2.461280870657179e-05, "loss": 0.7824, "step": 589 }, { "epoch": 0.43223443223443225, "grad_norm": 2.484375, "learning_rate": 2.4654667224780246e-05, "loss": 0.8537, "step": 590 }, { "epoch": 0.432967032967033, "grad_norm": 2.046875, "learning_rate": 2.46965257429887e-05, "loss": 0.8215, "step": 591 }, { "epoch": 0.4336996336996337, "grad_norm": 1.828125, "learning_rate": 2.4738384261197158e-05, "loss": 0.807, "step": 592 }, { "epoch": 0.43443223443223444, "grad_norm": 1.796875, "learning_rate": 2.478024277940561e-05, "loss": 0.8188, "step": 593 }, { "epoch": 0.4351648351648352, "grad_norm": 1.6640625, "learning_rate": 2.4822101297614066e-05, "loss": 0.7696, "step": 594 }, { "epoch": 0.4358974358974359, "grad_norm": 1.796875, "learning_rate": 2.486395981582252e-05, "loss": 0.7906, "step": 595 }, { "epoch": 0.43663003663003663, "grad_norm": 1.65625, "learning_rate": 2.4905818334030974e-05, "loss": 0.8195, "step": 596 }, { "epoch": 0.43736263736263736, "grad_norm": 1.6484375, "learning_rate": 2.494767685223943e-05, "loss": 0.755, "step": 597 }, { "epoch": 0.4380952380952381, "grad_norm": 1.8125, "learning_rate": 2.4989535370447885e-05, "loss": 0.8013, "step": 598 }, { "epoch": 0.4388278388278388, "grad_norm": 1.8984375, "learning_rate": 2.503139388865634e-05, "loss": 0.7987, "step": 599 }, { "epoch": 0.43956043956043955, "grad_norm": 1.890625, "learning_rate": 2.50732524068648e-05, "loss": 0.8261, "step": 600 }, { "epoch": 0.4402930402930403, "grad_norm": 2.015625, "learning_rate": 2.5115110925073254e-05, "loss": 0.7727, "step": 601 }, { "epoch": 0.441025641025641, "grad_norm": 2.078125, "learning_rate": 2.5156969443281708e-05, "loss": 0.7964, "step": 602 }, { "epoch": 0.44175824175824174, "grad_norm": 1.7578125, "learning_rate": 2.5198827961490162e-05, "loss": 0.758, "step": 603 }, { "epoch": 0.4424908424908425, "grad_norm": 2.140625, "learning_rate": 2.5240686479698623e-05, "loss": 0.8261, "step": 604 }, { "epoch": 0.4432234432234432, "grad_norm": 1.8046875, "learning_rate": 2.5282544997907077e-05, "loss": 0.8109, "step": 605 }, { "epoch": 0.44395604395604393, "grad_norm": 1.90625, "learning_rate": 2.532440351611553e-05, "loss": 0.7379, "step": 606 }, { "epoch": 0.44468864468864466, "grad_norm": 2.0625, "learning_rate": 2.5366262034323985e-05, "loss": 0.7547, "step": 607 }, { "epoch": 0.44542124542124545, "grad_norm": 1.6171875, "learning_rate": 2.5408120552532442e-05, "loss": 0.7562, "step": 608 }, { "epoch": 0.4461538461538462, "grad_norm": 1.75, "learning_rate": 2.54499790707409e-05, "loss": 0.7422, "step": 609 }, { "epoch": 0.4468864468864469, "grad_norm": 1.9609375, "learning_rate": 2.5491837588949354e-05, "loss": 0.8073, "step": 610 }, { "epoch": 0.44761904761904764, "grad_norm": 2.125, "learning_rate": 2.5533696107157808e-05, "loss": 0.7814, "step": 611 }, { "epoch": 0.44835164835164837, "grad_norm": 1.6875, "learning_rate": 2.5575554625366265e-05, "loss": 0.6691, "step": 612 }, { "epoch": 0.4490842490842491, "grad_norm": 1.8046875, "learning_rate": 2.561741314357472e-05, "loss": 0.7754, "step": 613 }, { "epoch": 0.44981684981684983, "grad_norm": 2.015625, "learning_rate": 2.5659271661783173e-05, "loss": 0.7072, "step": 614 }, { "epoch": 0.45054945054945056, "grad_norm": 1.6171875, "learning_rate": 2.5701130179991627e-05, "loss": 0.6233, "step": 615 }, { "epoch": 0.4512820512820513, "grad_norm": 1.859375, "learning_rate": 2.574298869820008e-05, "loss": 0.7504, "step": 616 }, { "epoch": 0.452014652014652, "grad_norm": 1.671875, "learning_rate": 2.5784847216408542e-05, "loss": 0.6659, "step": 617 }, { "epoch": 0.45274725274725275, "grad_norm": 1.515625, "learning_rate": 2.5826705734616996e-05, "loss": 0.6922, "step": 618 }, { "epoch": 0.4534798534798535, "grad_norm": 1.8203125, "learning_rate": 2.586856425282545e-05, "loss": 0.6461, "step": 619 }, { "epoch": 0.4542124542124542, "grad_norm": 1.9453125, "learning_rate": 2.5910422771033904e-05, "loss": 0.7763, "step": 620 }, { "epoch": 0.45494505494505494, "grad_norm": 1.859375, "learning_rate": 2.5952281289242365e-05, "loss": 0.7406, "step": 621 }, { "epoch": 0.45567765567765567, "grad_norm": 1.5234375, "learning_rate": 2.599413980745082e-05, "loss": 0.676, "step": 622 }, { "epoch": 0.4564102564102564, "grad_norm": 1.65625, "learning_rate": 2.6035998325659273e-05, "loss": 0.6846, "step": 623 }, { "epoch": 0.45714285714285713, "grad_norm": 1.875, "learning_rate": 2.6077856843867727e-05, "loss": 0.7038, "step": 624 }, { "epoch": 0.45787545787545786, "grad_norm": 2.078125, "learning_rate": 2.6119715362076184e-05, "loss": 0.6811, "step": 625 }, { "epoch": 0.4586080586080586, "grad_norm": 1.7578125, "learning_rate": 2.6161573880284638e-05, "loss": 0.6556, "step": 626 }, { "epoch": 0.4593406593406593, "grad_norm": 1.6953125, "learning_rate": 2.6203432398493096e-05, "loss": 0.6138, "step": 627 }, { "epoch": 0.46007326007326005, "grad_norm": 1.5546875, "learning_rate": 2.624529091670155e-05, "loss": 0.6486, "step": 628 }, { "epoch": 0.4608058608058608, "grad_norm": 2.0, "learning_rate": 2.6287149434910007e-05, "loss": 0.643, "step": 629 }, { "epoch": 0.46153846153846156, "grad_norm": 2.109375, "learning_rate": 2.632900795311846e-05, "loss": 0.6425, "step": 630 }, { "epoch": 0.4622710622710623, "grad_norm": 1.625, "learning_rate": 2.6370866471326915e-05, "loss": 0.6142, "step": 631 }, { "epoch": 0.463003663003663, "grad_norm": 1.8359375, "learning_rate": 2.641272498953537e-05, "loss": 0.662, "step": 632 }, { "epoch": 0.46373626373626375, "grad_norm": 1.6875, "learning_rate": 2.645458350774383e-05, "loss": 0.5745, "step": 633 }, { "epoch": 0.4644688644688645, "grad_norm": 1.71875, "learning_rate": 2.6496442025952284e-05, "loss": 0.63, "step": 634 }, { "epoch": 0.4652014652014652, "grad_norm": 1.53125, "learning_rate": 2.6538300544160738e-05, "loss": 0.6147, "step": 635 }, { "epoch": 0.46593406593406594, "grad_norm": 1.9375, "learning_rate": 2.6580159062369192e-05, "loss": 0.687, "step": 636 }, { "epoch": 0.4666666666666667, "grad_norm": 1.9453125, "learning_rate": 2.6622017580577646e-05, "loss": 0.5722, "step": 637 }, { "epoch": 0.4673992673992674, "grad_norm": 1.8203125, "learning_rate": 2.6663876098786107e-05, "loss": 0.636, "step": 638 }, { "epoch": 0.46813186813186813, "grad_norm": 1.8203125, "learning_rate": 2.670573461699456e-05, "loss": 0.6986, "step": 639 }, { "epoch": 0.46886446886446886, "grad_norm": 1.8125, "learning_rate": 2.6747593135203015e-05, "loss": 0.5868, "step": 640 }, { "epoch": 0.4695970695970696, "grad_norm": 1.796875, "learning_rate": 2.678945165341147e-05, "loss": 0.6554, "step": 641 }, { "epoch": 0.4703296703296703, "grad_norm": 1.7109375, "learning_rate": 2.6831310171619926e-05, "loss": 0.6741, "step": 642 }, { "epoch": 0.47106227106227105, "grad_norm": 1.8515625, "learning_rate": 2.687316868982838e-05, "loss": 0.7199, "step": 643 }, { "epoch": 0.4717948717948718, "grad_norm": 1.8203125, "learning_rate": 2.6915027208036834e-05, "loss": 0.6546, "step": 644 }, { "epoch": 0.4725274725274725, "grad_norm": 1.859375, "learning_rate": 2.6956885726245292e-05, "loss": 0.649, "step": 645 }, { "epoch": 0.47326007326007324, "grad_norm": 1.8046875, "learning_rate": 2.699874424445375e-05, "loss": 0.6974, "step": 646 }, { "epoch": 0.473992673992674, "grad_norm": 1.7109375, "learning_rate": 2.7040602762662203e-05, "loss": 0.5746, "step": 647 }, { "epoch": 0.4747252747252747, "grad_norm": 1.65625, "learning_rate": 2.7082461280870657e-05, "loss": 0.6136, "step": 648 }, { "epoch": 0.47545787545787543, "grad_norm": 1.9453125, "learning_rate": 2.712431979907911e-05, "loss": 0.6923, "step": 649 }, { "epoch": 0.47619047619047616, "grad_norm": 1.8515625, "learning_rate": 2.7166178317287572e-05, "loss": 0.6107, "step": 650 }, { "epoch": 0.47692307692307695, "grad_norm": 1.6484375, "learning_rate": 2.7208036835496026e-05, "loss": 0.6172, "step": 651 }, { "epoch": 0.4776556776556777, "grad_norm": 1.5859375, "learning_rate": 2.724989535370448e-05, "loss": 0.6196, "step": 652 }, { "epoch": 0.4783882783882784, "grad_norm": 1.875, "learning_rate": 2.7291753871912934e-05, "loss": 0.6876, "step": 653 }, { "epoch": 0.47912087912087914, "grad_norm": 1.8515625, "learning_rate": 2.7333612390121395e-05, "loss": 0.5631, "step": 654 }, { "epoch": 0.47985347985347987, "grad_norm": 1.796875, "learning_rate": 2.737547090832985e-05, "loss": 0.5079, "step": 655 }, { "epoch": 0.4805860805860806, "grad_norm": 1.5859375, "learning_rate": 2.7417329426538303e-05, "loss": 0.5118, "step": 656 }, { "epoch": 0.48131868131868133, "grad_norm": 1.84375, "learning_rate": 2.7459187944746757e-05, "loss": 0.6268, "step": 657 }, { "epoch": 0.48205128205128206, "grad_norm": 1.8515625, "learning_rate": 2.750104646295521e-05, "loss": 0.6427, "step": 658 }, { "epoch": 0.4827838827838828, "grad_norm": 1.9140625, "learning_rate": 2.7542904981163668e-05, "loss": 0.6553, "step": 659 }, { "epoch": 0.4835164835164835, "grad_norm": 2.046875, "learning_rate": 2.7584763499372122e-05, "loss": 0.6928, "step": 660 }, { "epoch": 0.48424908424908425, "grad_norm": 1.7890625, "learning_rate": 2.7626622017580576e-05, "loss": 0.6277, "step": 661 }, { "epoch": 0.484981684981685, "grad_norm": 2.28125, "learning_rate": 2.766848053578903e-05, "loss": 0.6483, "step": 662 }, { "epoch": 0.4857142857142857, "grad_norm": 1.625, "learning_rate": 2.771033905399749e-05, "loss": 0.6063, "step": 663 }, { "epoch": 0.48644688644688644, "grad_norm": 1.4921875, "learning_rate": 2.7752197572205945e-05, "loss": 0.5515, "step": 664 }, { "epoch": 0.48717948717948717, "grad_norm": 1.7421875, "learning_rate": 2.77940560904144e-05, "loss": 0.5759, "step": 665 }, { "epoch": 0.4879120879120879, "grad_norm": 1.7578125, "learning_rate": 2.7835914608622853e-05, "loss": 0.5736, "step": 666 }, { "epoch": 0.48864468864468863, "grad_norm": 1.8828125, "learning_rate": 2.7877773126831314e-05, "loss": 0.6637, "step": 667 }, { "epoch": 0.48937728937728936, "grad_norm": 1.7109375, "learning_rate": 2.7919631645039768e-05, "loss": 0.6102, "step": 668 }, { "epoch": 0.4901098901098901, "grad_norm": 1.6796875, "learning_rate": 2.7961490163248222e-05, "loss": 0.6033, "step": 669 }, { "epoch": 0.4908424908424908, "grad_norm": 1.75, "learning_rate": 2.8003348681456676e-05, "loss": 0.6789, "step": 670 }, { "epoch": 0.49157509157509155, "grad_norm": 1.6484375, "learning_rate": 2.8045207199665137e-05, "loss": 0.6301, "step": 671 }, { "epoch": 0.49230769230769234, "grad_norm": 1.8125, "learning_rate": 2.808706571787359e-05, "loss": 0.6143, "step": 672 }, { "epoch": 0.49304029304029307, "grad_norm": 1.6171875, "learning_rate": 2.8128924236082045e-05, "loss": 0.5655, "step": 673 }, { "epoch": 0.4937728937728938, "grad_norm": 1.515625, "learning_rate": 2.81707827542905e-05, "loss": 0.5507, "step": 674 }, { "epoch": 0.4945054945054945, "grad_norm": 1.4609375, "learning_rate": 2.8212641272498956e-05, "loss": 0.53, "step": 675 }, { "epoch": 0.49523809523809526, "grad_norm": 1.421875, "learning_rate": 2.825449979070741e-05, "loss": 0.5812, "step": 676 }, { "epoch": 0.495970695970696, "grad_norm": 1.6875, "learning_rate": 2.8296358308915864e-05, "loss": 0.6324, "step": 677 }, { "epoch": 0.4967032967032967, "grad_norm": 1.4921875, "learning_rate": 2.833821682712432e-05, "loss": 0.5733, "step": 678 }, { "epoch": 0.49743589743589745, "grad_norm": 1.53125, "learning_rate": 2.8380075345332772e-05, "loss": 0.6055, "step": 679 }, { "epoch": 0.4981684981684982, "grad_norm": 1.859375, "learning_rate": 2.8421933863541233e-05, "loss": 0.629, "step": 680 }, { "epoch": 0.4989010989010989, "grad_norm": 1.875, "learning_rate": 2.8463792381749687e-05, "loss": 0.5473, "step": 681 }, { "epoch": 0.49963369963369964, "grad_norm": 1.6796875, "learning_rate": 2.850565089995814e-05, "loss": 0.5346, "step": 682 }, { "epoch": 0.5003663003663004, "grad_norm": 1.5078125, "learning_rate": 2.8547509418166595e-05, "loss": 0.5485, "step": 683 }, { "epoch": 0.5010989010989011, "grad_norm": 1.5546875, "learning_rate": 2.8589367936375056e-05, "loss": 0.5697, "step": 684 }, { "epoch": 0.5018315018315018, "grad_norm": 2.015625, "learning_rate": 2.863122645458351e-05, "loss": 0.581, "step": 685 }, { "epoch": 0.5025641025641026, "grad_norm": 1.625, "learning_rate": 2.8673084972791964e-05, "loss": 0.5339, "step": 686 }, { "epoch": 0.5032967032967033, "grad_norm": 1.5859375, "learning_rate": 2.8714943491000418e-05, "loss": 0.5906, "step": 687 }, { "epoch": 0.504029304029304, "grad_norm": 1.6171875, "learning_rate": 2.875680200920888e-05, "loss": 0.5741, "step": 688 }, { "epoch": 0.5047619047619047, "grad_norm": 1.6875, "learning_rate": 2.8798660527417333e-05, "loss": 0.4751, "step": 689 }, { "epoch": 0.5054945054945055, "grad_norm": 1.671875, "learning_rate": 2.8840519045625787e-05, "loss": 0.6095, "step": 690 }, { "epoch": 0.5062271062271062, "grad_norm": 1.3984375, "learning_rate": 2.888237756383424e-05, "loss": 0.5067, "step": 691 }, { "epoch": 0.5069597069597069, "grad_norm": 1.8515625, "learning_rate": 2.8924236082042698e-05, "loss": 0.5384, "step": 692 }, { "epoch": 0.5076923076923077, "grad_norm": 1.6796875, "learning_rate": 2.8966094600251152e-05, "loss": 0.552, "step": 693 }, { "epoch": 0.5084249084249084, "grad_norm": 1.484375, "learning_rate": 2.9007953118459606e-05, "loss": 0.5315, "step": 694 }, { "epoch": 0.5091575091575091, "grad_norm": 1.7578125, "learning_rate": 2.904981163666806e-05, "loss": 0.5577, "step": 695 }, { "epoch": 0.5098901098901099, "grad_norm": 1.4765625, "learning_rate": 2.909167015487652e-05, "loss": 0.5856, "step": 696 }, { "epoch": 0.5106227106227106, "grad_norm": 1.7421875, "learning_rate": 2.9133528673084975e-05, "loss": 0.5242, "step": 697 }, { "epoch": 0.5113553113553113, "grad_norm": 1.75, "learning_rate": 2.917538719129343e-05, "loss": 0.58, "step": 698 }, { "epoch": 0.512087912087912, "grad_norm": 1.6015625, "learning_rate": 2.9217245709501883e-05, "loss": 0.5514, "step": 699 }, { "epoch": 0.5128205128205128, "grad_norm": 1.8203125, "learning_rate": 2.9259104227710337e-05, "loss": 0.5082, "step": 700 }, { "epoch": 0.5135531135531135, "grad_norm": 1.6171875, "learning_rate": 2.9300962745918798e-05, "loss": 0.5126, "step": 701 }, { "epoch": 0.5142857142857142, "grad_norm": 1.6484375, "learning_rate": 2.9342821264127252e-05, "loss": 0.6716, "step": 702 }, { "epoch": 0.515018315018315, "grad_norm": 2.125, "learning_rate": 2.9384679782335706e-05, "loss": 0.5727, "step": 703 }, { "epoch": 0.5157509157509158, "grad_norm": 1.9140625, "learning_rate": 2.942653830054416e-05, "loss": 0.5927, "step": 704 }, { "epoch": 0.5164835164835165, "grad_norm": 2.15625, "learning_rate": 2.946839681875262e-05, "loss": 0.5573, "step": 705 }, { "epoch": 0.5172161172161173, "grad_norm": 1.53125, "learning_rate": 2.9510255336961075e-05, "loss": 0.5076, "step": 706 }, { "epoch": 0.517948717948718, "grad_norm": 1.578125, "learning_rate": 2.955211385516953e-05, "loss": 0.5296, "step": 707 }, { "epoch": 0.5186813186813187, "grad_norm": 1.703125, "learning_rate": 2.9593972373377983e-05, "loss": 0.4676, "step": 708 }, { "epoch": 0.5194139194139195, "grad_norm": 1.65625, "learning_rate": 2.963583089158644e-05, "loss": 0.5434, "step": 709 }, { "epoch": 0.5201465201465202, "grad_norm": 1.609375, "learning_rate": 2.9677689409794894e-05, "loss": 0.5455, "step": 710 }, { "epoch": 0.5208791208791209, "grad_norm": 1.6875, "learning_rate": 2.971954792800335e-05, "loss": 0.5693, "step": 711 }, { "epoch": 0.5216117216117216, "grad_norm": 1.640625, "learning_rate": 2.9761406446211802e-05, "loss": 0.5209, "step": 712 }, { "epoch": 0.5223443223443224, "grad_norm": 1.609375, "learning_rate": 2.9803264964420263e-05, "loss": 0.4515, "step": 713 }, { "epoch": 0.5230769230769231, "grad_norm": 1.546875, "learning_rate": 2.9845123482628717e-05, "loss": 0.601, "step": 714 }, { "epoch": 0.5238095238095238, "grad_norm": 1.4453125, "learning_rate": 2.988698200083717e-05, "loss": 0.5235, "step": 715 }, { "epoch": 0.5245421245421246, "grad_norm": 2.125, "learning_rate": 2.9928840519045625e-05, "loss": 0.6071, "step": 716 }, { "epoch": 0.5252747252747253, "grad_norm": 1.671875, "learning_rate": 2.9970699037254086e-05, "loss": 0.4606, "step": 717 }, { "epoch": 0.526007326007326, "grad_norm": 1.546875, "learning_rate": 3.001255755546254e-05, "loss": 0.4239, "step": 718 }, { "epoch": 0.5267399267399268, "grad_norm": 1.5, "learning_rate": 3.0054416073670994e-05, "loss": 0.5342, "step": 719 }, { "epoch": 0.5274725274725275, "grad_norm": 1.671875, "learning_rate": 3.0096274591879448e-05, "loss": 0.4767, "step": 720 }, { "epoch": 0.5282051282051282, "grad_norm": 1.4375, "learning_rate": 3.0138133110087902e-05, "loss": 0.5125, "step": 721 }, { "epoch": 0.528937728937729, "grad_norm": 1.4453125, "learning_rate": 3.0179991628296363e-05, "loss": 0.4957, "step": 722 }, { "epoch": 0.5296703296703297, "grad_norm": 1.4765625, "learning_rate": 3.0221850146504817e-05, "loss": 0.5126, "step": 723 }, { "epoch": 0.5304029304029304, "grad_norm": 1.2265625, "learning_rate": 3.026370866471327e-05, "loss": 0.4284, "step": 724 }, { "epoch": 0.5311355311355311, "grad_norm": 1.4296875, "learning_rate": 3.0305567182921725e-05, "loss": 0.5242, "step": 725 }, { "epoch": 0.5318681318681319, "grad_norm": 2.03125, "learning_rate": 3.0347425701130182e-05, "loss": 0.4725, "step": 726 }, { "epoch": 0.5326007326007326, "grad_norm": 1.734375, "learning_rate": 3.0389284219338636e-05, "loss": 0.5206, "step": 727 }, { "epoch": 0.5333333333333333, "grad_norm": 1.6328125, "learning_rate": 3.043114273754709e-05, "loss": 0.5715, "step": 728 }, { "epoch": 0.5340659340659341, "grad_norm": 1.546875, "learning_rate": 3.0473001255755544e-05, "loss": 0.554, "step": 729 }, { "epoch": 0.5347985347985348, "grad_norm": 1.4453125, "learning_rate": 3.0514859773964005e-05, "loss": 0.5189, "step": 730 }, { "epoch": 0.5355311355311355, "grad_norm": 1.390625, "learning_rate": 3.055671829217246e-05, "loss": 0.446, "step": 731 }, { "epoch": 0.5362637362637362, "grad_norm": 1.671875, "learning_rate": 3.059857681038091e-05, "loss": 0.4482, "step": 732 }, { "epoch": 0.536996336996337, "grad_norm": 1.5, "learning_rate": 3.064043532858937e-05, "loss": 0.5334, "step": 733 }, { "epoch": 0.5377289377289377, "grad_norm": 1.203125, "learning_rate": 3.068229384679783e-05, "loss": 0.4276, "step": 734 }, { "epoch": 0.5384615384615384, "grad_norm": 1.390625, "learning_rate": 3.072415236500628e-05, "loss": 0.4582, "step": 735 }, { "epoch": 0.5391941391941392, "grad_norm": 1.640625, "learning_rate": 3.0766010883214736e-05, "loss": 0.4878, "step": 736 }, { "epoch": 0.5399267399267399, "grad_norm": 2.015625, "learning_rate": 3.080786940142319e-05, "loss": 0.4606, "step": 737 }, { "epoch": 0.5406593406593406, "grad_norm": 1.625, "learning_rate": 3.084972791963165e-05, "loss": 0.4829, "step": 738 }, { "epoch": 0.5413919413919414, "grad_norm": 1.34375, "learning_rate": 3.08915864378401e-05, "loss": 0.4917, "step": 739 }, { "epoch": 0.5421245421245421, "grad_norm": 1.5, "learning_rate": 3.093344495604856e-05, "loss": 0.4134, "step": 740 }, { "epoch": 0.5428571428571428, "grad_norm": 1.71875, "learning_rate": 3.097530347425701e-05, "loss": 0.4327, "step": 741 }, { "epoch": 0.5435897435897435, "grad_norm": 1.390625, "learning_rate": 3.101716199246547e-05, "loss": 0.5026, "step": 742 }, { "epoch": 0.5443223443223443, "grad_norm": 1.390625, "learning_rate": 3.1059020510673924e-05, "loss": 0.4934, "step": 743 }, { "epoch": 0.545054945054945, "grad_norm": 1.28125, "learning_rate": 3.110087902888238e-05, "loss": 0.3994, "step": 744 }, { "epoch": 0.5457875457875457, "grad_norm": 1.4765625, "learning_rate": 3.114273754709083e-05, "loss": 0.5263, "step": 745 }, { "epoch": 0.5465201465201465, "grad_norm": 1.34375, "learning_rate": 3.118459606529929e-05, "loss": 0.4682, "step": 746 }, { "epoch": 0.5472527472527473, "grad_norm": 1.3046875, "learning_rate": 3.122645458350775e-05, "loss": 0.4967, "step": 747 }, { "epoch": 0.547985347985348, "grad_norm": 1.71875, "learning_rate": 3.1268313101716205e-05, "loss": 0.599, "step": 748 }, { "epoch": 0.5487179487179488, "grad_norm": 1.40625, "learning_rate": 3.1310171619924655e-05, "loss": 0.4815, "step": 749 }, { "epoch": 0.5494505494505495, "grad_norm": 1.46875, "learning_rate": 3.135203013813311e-05, "loss": 0.4515, "step": 750 }, { "epoch": 0.5501831501831502, "grad_norm": 1.5859375, "learning_rate": 3.139388865634157e-05, "loss": 0.4681, "step": 751 }, { "epoch": 0.550915750915751, "grad_norm": 1.3359375, "learning_rate": 3.143574717455002e-05, "loss": 0.4124, "step": 752 }, { "epoch": 0.5516483516483517, "grad_norm": 1.625, "learning_rate": 3.147760569275848e-05, "loss": 0.4081, "step": 753 }, { "epoch": 0.5523809523809524, "grad_norm": 1.4765625, "learning_rate": 3.151946421096693e-05, "loss": 0.4544, "step": 754 }, { "epoch": 0.5531135531135531, "grad_norm": 1.4296875, "learning_rate": 3.156132272917539e-05, "loss": 0.3986, "step": 755 }, { "epoch": 0.5538461538461539, "grad_norm": 1.4609375, "learning_rate": 3.1603181247383843e-05, "loss": 0.4844, "step": 756 }, { "epoch": 0.5545787545787546, "grad_norm": 1.8125, "learning_rate": 3.16450397655923e-05, "loss": 0.4502, "step": 757 }, { "epoch": 0.5553113553113553, "grad_norm": 1.5703125, "learning_rate": 3.168689828380075e-05, "loss": 0.4749, "step": 758 }, { "epoch": 0.5560439560439561, "grad_norm": 1.6171875, "learning_rate": 3.1728756802009216e-05, "loss": 0.4814, "step": 759 }, { "epoch": 0.5567765567765568, "grad_norm": 1.484375, "learning_rate": 3.1770615320217666e-05, "loss": 0.4617, "step": 760 }, { "epoch": 0.5575091575091575, "grad_norm": 1.5625, "learning_rate": 3.1812473838426124e-05, "loss": 0.4414, "step": 761 }, { "epoch": 0.5582417582417583, "grad_norm": 1.8046875, "learning_rate": 3.1854332356634574e-05, "loss": 0.4969, "step": 762 }, { "epoch": 0.558974358974359, "grad_norm": 1.5390625, "learning_rate": 3.189619087484303e-05, "loss": 0.521, "step": 763 }, { "epoch": 0.5597069597069597, "grad_norm": 1.4296875, "learning_rate": 3.193804939305149e-05, "loss": 0.4548, "step": 764 }, { "epoch": 0.5604395604395604, "grad_norm": 1.703125, "learning_rate": 3.197990791125994e-05, "loss": 0.4804, "step": 765 }, { "epoch": 0.5611721611721612, "grad_norm": 1.4765625, "learning_rate": 3.20217664294684e-05, "loss": 0.3983, "step": 766 }, { "epoch": 0.5619047619047619, "grad_norm": 1.5234375, "learning_rate": 3.2063624947676855e-05, "loss": 0.4674, "step": 767 }, { "epoch": 0.5626373626373626, "grad_norm": 1.4140625, "learning_rate": 3.210548346588531e-05, "loss": 0.4578, "step": 768 }, { "epoch": 0.5633699633699634, "grad_norm": 1.6484375, "learning_rate": 3.214734198409376e-05, "loss": 0.5139, "step": 769 }, { "epoch": 0.5641025641025641, "grad_norm": 1.828125, "learning_rate": 3.218920050230222e-05, "loss": 0.5174, "step": 770 }, { "epoch": 0.5648351648351648, "grad_norm": 1.7109375, "learning_rate": 3.223105902051067e-05, "loss": 0.4603, "step": 771 }, { "epoch": 0.5655677655677656, "grad_norm": 1.5625, "learning_rate": 3.2272917538719135e-05, "loss": 0.4957, "step": 772 }, { "epoch": 0.5663003663003663, "grad_norm": 1.7109375, "learning_rate": 3.2314776056927586e-05, "loss": 0.5171, "step": 773 }, { "epoch": 0.567032967032967, "grad_norm": 1.4921875, "learning_rate": 3.235663457513604e-05, "loss": 0.4747, "step": 774 }, { "epoch": 0.5677655677655677, "grad_norm": 1.2734375, "learning_rate": 3.2398493093344494e-05, "loss": 0.4666, "step": 775 }, { "epoch": 0.5684981684981685, "grad_norm": 1.390625, "learning_rate": 3.244035161155296e-05, "loss": 0.431, "step": 776 }, { "epoch": 0.5692307692307692, "grad_norm": 1.515625, "learning_rate": 3.248221012976141e-05, "loss": 0.4171, "step": 777 }, { "epoch": 0.5699633699633699, "grad_norm": 1.1953125, "learning_rate": 3.2524068647969866e-05, "loss": 0.3839, "step": 778 }, { "epoch": 0.5706959706959707, "grad_norm": 1.25, "learning_rate": 3.2565927166178316e-05, "loss": 0.4405, "step": 779 }, { "epoch": 0.5714285714285714, "grad_norm": 1.0859375, "learning_rate": 3.2607785684386774e-05, "loss": 0.4178, "step": 780 }, { "epoch": 0.5721611721611721, "grad_norm": 1.3828125, "learning_rate": 3.264964420259523e-05, "loss": 0.3845, "step": 781 }, { "epoch": 0.5728937728937729, "grad_norm": 1.1875, "learning_rate": 3.269150272080368e-05, "loss": 0.4949, "step": 782 }, { "epoch": 0.5736263736263736, "grad_norm": 1.578125, "learning_rate": 3.273336123901214e-05, "loss": 0.4303, "step": 783 }, { "epoch": 0.5743589743589743, "grad_norm": 1.78125, "learning_rate": 3.277521975722059e-05, "loss": 0.4505, "step": 784 }, { "epoch": 0.575091575091575, "grad_norm": 1.609375, "learning_rate": 3.2817078275429054e-05, "loss": 0.4491, "step": 785 }, { "epoch": 0.5758241758241758, "grad_norm": 1.296875, "learning_rate": 3.2858936793637505e-05, "loss": 0.4287, "step": 786 }, { "epoch": 0.5765567765567765, "grad_norm": 1.2421875, "learning_rate": 3.290079531184596e-05, "loss": 0.4357, "step": 787 }, { "epoch": 0.5772893772893772, "grad_norm": 1.4140625, "learning_rate": 3.294265383005441e-05, "loss": 0.4601, "step": 788 }, { "epoch": 0.578021978021978, "grad_norm": 1.8984375, "learning_rate": 3.298451234826288e-05, "loss": 0.4574, "step": 789 }, { "epoch": 0.5787545787545788, "grad_norm": 1.4765625, "learning_rate": 3.302637086647133e-05, "loss": 0.5251, "step": 790 }, { "epoch": 0.5794871794871795, "grad_norm": 1.2890625, "learning_rate": 3.3068229384679785e-05, "loss": 0.4296, "step": 791 }, { "epoch": 0.5802197802197803, "grad_norm": 1.53125, "learning_rate": 3.3110087902888236e-05, "loss": 0.4218, "step": 792 }, { "epoch": 0.580952380952381, "grad_norm": 1.7265625, "learning_rate": 3.31519464210967e-05, "loss": 0.4234, "step": 793 }, { "epoch": 0.5816849816849817, "grad_norm": 1.765625, "learning_rate": 3.319380493930515e-05, "loss": 0.5174, "step": 794 }, { "epoch": 0.5824175824175825, "grad_norm": 1.46875, "learning_rate": 3.323566345751361e-05, "loss": 0.4452, "step": 795 }, { "epoch": 0.5831501831501832, "grad_norm": 1.609375, "learning_rate": 3.327752197572206e-05, "loss": 0.4835, "step": 796 }, { "epoch": 0.5838827838827839, "grad_norm": 1.4765625, "learning_rate": 3.3319380493930516e-05, "loss": 0.4036, "step": 797 }, { "epoch": 0.5846153846153846, "grad_norm": 1.5625, "learning_rate": 3.336123901213897e-05, "loss": 0.4007, "step": 798 }, { "epoch": 0.5853479853479854, "grad_norm": 1.25, "learning_rate": 3.3403097530347424e-05, "loss": 0.38, "step": 799 }, { "epoch": 0.5860805860805861, "grad_norm": 1.25, "learning_rate": 3.344495604855588e-05, "loss": 0.4748, "step": 800 }, { "epoch": 0.5868131868131868, "grad_norm": 1.4375, "learning_rate": 3.348681456676434e-05, "loss": 0.4986, "step": 801 }, { "epoch": 0.5875457875457876, "grad_norm": 1.59375, "learning_rate": 3.3528673084972796e-05, "loss": 0.4713, "step": 802 }, { "epoch": 0.5882783882783883, "grad_norm": 1.7734375, "learning_rate": 3.357053160318125e-05, "loss": 0.4462, "step": 803 }, { "epoch": 0.589010989010989, "grad_norm": 1.5, "learning_rate": 3.3612390121389704e-05, "loss": 0.4533, "step": 804 }, { "epoch": 0.5897435897435898, "grad_norm": 2.09375, "learning_rate": 3.3654248639598155e-05, "loss": 0.5616, "step": 805 }, { "epoch": 0.5904761904761905, "grad_norm": 1.765625, "learning_rate": 3.369610715780662e-05, "loss": 0.4861, "step": 806 }, { "epoch": 0.5912087912087912, "grad_norm": 1.5625, "learning_rate": 3.373796567601507e-05, "loss": 0.5224, "step": 807 }, { "epoch": 0.591941391941392, "grad_norm": 1.328125, "learning_rate": 3.377982419422353e-05, "loss": 0.4682, "step": 808 }, { "epoch": 0.5926739926739927, "grad_norm": 1.3515625, "learning_rate": 3.382168271243198e-05, "loss": 0.4646, "step": 809 }, { "epoch": 0.5934065934065934, "grad_norm": 1.21875, "learning_rate": 3.386354123064044e-05, "loss": 0.3578, "step": 810 }, { "epoch": 0.5941391941391941, "grad_norm": 1.1796875, "learning_rate": 3.390539974884889e-05, "loss": 0.3756, "step": 811 }, { "epoch": 0.5948717948717949, "grad_norm": 1.5703125, "learning_rate": 3.394725826705735e-05, "loss": 0.4974, "step": 812 }, { "epoch": 0.5956043956043956, "grad_norm": 1.375, "learning_rate": 3.39891167852658e-05, "loss": 0.5055, "step": 813 }, { "epoch": 0.5963369963369963, "grad_norm": 1.3671875, "learning_rate": 3.403097530347426e-05, "loss": 0.426, "step": 814 }, { "epoch": 0.5970695970695971, "grad_norm": 1.296875, "learning_rate": 3.4072833821682715e-05, "loss": 0.4612, "step": 815 }, { "epoch": 0.5978021978021978, "grad_norm": 1.4921875, "learning_rate": 3.4114692339891166e-05, "loss": 0.4383, "step": 816 }, { "epoch": 0.5985347985347985, "grad_norm": 1.640625, "learning_rate": 3.415655085809962e-05, "loss": 0.5244, "step": 817 }, { "epoch": 0.5992673992673992, "grad_norm": 1.40625, "learning_rate": 3.419840937630808e-05, "loss": 0.4312, "step": 818 }, { "epoch": 0.6, "grad_norm": 1.421875, "learning_rate": 3.424026789451654e-05, "loss": 0.4389, "step": 819 }, { "epoch": 0.6007326007326007, "grad_norm": 1.4921875, "learning_rate": 3.428212641272499e-05, "loss": 0.3945, "step": 820 }, { "epoch": 0.6014652014652014, "grad_norm": 1.21875, "learning_rate": 3.4323984930933446e-05, "loss": 0.3903, "step": 821 }, { "epoch": 0.6021978021978022, "grad_norm": 1.546875, "learning_rate": 3.4365843449141903e-05, "loss": 0.4589, "step": 822 }, { "epoch": 0.6029304029304029, "grad_norm": 1.3125, "learning_rate": 3.440770196735036e-05, "loss": 0.436, "step": 823 }, { "epoch": 0.6036630036630036, "grad_norm": 1.265625, "learning_rate": 3.444956048555881e-05, "loss": 0.4109, "step": 824 }, { "epoch": 0.6043956043956044, "grad_norm": 1.59375, "learning_rate": 3.449141900376727e-05, "loss": 0.3869, "step": 825 }, { "epoch": 0.6051282051282051, "grad_norm": 1.78125, "learning_rate": 3.453327752197572e-05, "loss": 0.4592, "step": 826 }, { "epoch": 0.6058608058608058, "grad_norm": 1.359375, "learning_rate": 3.4575136040184184e-05, "loss": 0.4239, "step": 827 }, { "epoch": 0.6065934065934065, "grad_norm": 1.296875, "learning_rate": 3.4616994558392634e-05, "loss": 0.476, "step": 828 }, { "epoch": 0.6073260073260073, "grad_norm": 1.5234375, "learning_rate": 3.465885307660109e-05, "loss": 0.5457, "step": 829 }, { "epoch": 0.608058608058608, "grad_norm": 1.1484375, "learning_rate": 3.470071159480954e-05, "loss": 0.3655, "step": 830 }, { "epoch": 0.6087912087912087, "grad_norm": 1.28125, "learning_rate": 3.4742570113018e-05, "loss": 0.3791, "step": 831 }, { "epoch": 0.6095238095238096, "grad_norm": 1.15625, "learning_rate": 3.478442863122646e-05, "loss": 0.4398, "step": 832 }, { "epoch": 0.6102564102564103, "grad_norm": 1.3671875, "learning_rate": 3.482628714943491e-05, "loss": 0.4275, "step": 833 }, { "epoch": 0.610989010989011, "grad_norm": 1.234375, "learning_rate": 3.4868145667643365e-05, "loss": 0.4329, "step": 834 }, { "epoch": 0.6117216117216118, "grad_norm": 1.4765625, "learning_rate": 3.491000418585182e-05, "loss": 0.4184, "step": 835 }, { "epoch": 0.6124542124542125, "grad_norm": 1.4921875, "learning_rate": 3.495186270406028e-05, "loss": 0.4776, "step": 836 }, { "epoch": 0.6131868131868132, "grad_norm": 1.5, "learning_rate": 3.499372122226873e-05, "loss": 0.3512, "step": 837 }, { "epoch": 0.613919413919414, "grad_norm": 1.6171875, "learning_rate": 3.503557974047719e-05, "loss": 0.4549, "step": 838 }, { "epoch": 0.6146520146520147, "grad_norm": 1.3984375, "learning_rate": 3.5077438258685646e-05, "loss": 0.3647, "step": 839 }, { "epoch": 0.6153846153846154, "grad_norm": 1.3515625, "learning_rate": 3.51192967768941e-05, "loss": 0.3747, "step": 840 }, { "epoch": 0.6161172161172161, "grad_norm": 1.421875, "learning_rate": 3.5161155295102554e-05, "loss": 0.4084, "step": 841 }, { "epoch": 0.6168498168498169, "grad_norm": 1.5703125, "learning_rate": 3.520301381331101e-05, "loss": 0.4125, "step": 842 }, { "epoch": 0.6175824175824176, "grad_norm": 2.234375, "learning_rate": 3.524487233151947e-05, "loss": 0.3944, "step": 843 }, { "epoch": 0.6183150183150183, "grad_norm": 1.640625, "learning_rate": 3.5286730849727926e-05, "loss": 0.4986, "step": 844 }, { "epoch": 0.6190476190476191, "grad_norm": 1.5390625, "learning_rate": 3.5328589367936376e-05, "loss": 0.5309, "step": 845 }, { "epoch": 0.6197802197802198, "grad_norm": 1.140625, "learning_rate": 3.5370447886144834e-05, "loss": 0.3913, "step": 846 }, { "epoch": 0.6205128205128205, "grad_norm": 1.546875, "learning_rate": 3.5412306404353284e-05, "loss": 0.4612, "step": 847 }, { "epoch": 0.6212454212454213, "grad_norm": 1.6640625, "learning_rate": 3.545416492256174e-05, "loss": 0.4899, "step": 848 }, { "epoch": 0.621978021978022, "grad_norm": 1.0390625, "learning_rate": 3.54960234407702e-05, "loss": 0.3642, "step": 849 }, { "epoch": 0.6227106227106227, "grad_norm": 1.5859375, "learning_rate": 3.553788195897865e-05, "loss": 0.4311, "step": 850 }, { "epoch": 0.6234432234432234, "grad_norm": 1.59375, "learning_rate": 3.557974047718711e-05, "loss": 0.4603, "step": 851 }, { "epoch": 0.6241758241758242, "grad_norm": 1.1796875, "learning_rate": 3.5621598995395565e-05, "loss": 0.3846, "step": 852 }, { "epoch": 0.6249084249084249, "grad_norm": 1.40625, "learning_rate": 3.566345751360402e-05, "loss": 0.4535, "step": 853 }, { "epoch": 0.6256410256410256, "grad_norm": 1.421875, "learning_rate": 3.570531603181247e-05, "loss": 0.3759, "step": 854 }, { "epoch": 0.6263736263736264, "grad_norm": 1.4140625, "learning_rate": 3.574717455002093e-05, "loss": 0.4422, "step": 855 }, { "epoch": 0.6271062271062271, "grad_norm": 1.28125, "learning_rate": 3.578903306822939e-05, "loss": 0.4164, "step": 856 }, { "epoch": 0.6278388278388278, "grad_norm": 1.7265625, "learning_rate": 3.5830891586437845e-05, "loss": 0.4544, "step": 857 }, { "epoch": 0.6285714285714286, "grad_norm": 1.1015625, "learning_rate": 3.5872750104646296e-05, "loss": 0.4279, "step": 858 }, { "epoch": 0.6293040293040293, "grad_norm": 1.53125, "learning_rate": 3.591460862285475e-05, "loss": 0.4192, "step": 859 }, { "epoch": 0.63003663003663, "grad_norm": 1.6484375, "learning_rate": 3.595646714106321e-05, "loss": 0.4731, "step": 860 }, { "epoch": 0.6307692307692307, "grad_norm": 1.078125, "learning_rate": 3.599832565927167e-05, "loss": 0.3468, "step": 861 }, { "epoch": 0.6315018315018315, "grad_norm": 1.3046875, "learning_rate": 3.604018417748012e-05, "loss": 0.4475, "step": 862 }, { "epoch": 0.6322344322344322, "grad_norm": 2.015625, "learning_rate": 3.6082042695688576e-05, "loss": 0.4567, "step": 863 }, { "epoch": 0.6329670329670329, "grad_norm": 1.484375, "learning_rate": 3.612390121389703e-05, "loss": 0.4004, "step": 864 }, { "epoch": 0.6336996336996337, "grad_norm": 1.171875, "learning_rate": 3.6165759732105484e-05, "loss": 0.3448, "step": 865 }, { "epoch": 0.6344322344322344, "grad_norm": 1.4921875, "learning_rate": 3.620761825031394e-05, "loss": 0.4685, "step": 866 }, { "epoch": 0.6351648351648351, "grad_norm": 1.6796875, "learning_rate": 3.624947676852239e-05, "loss": 0.4531, "step": 867 }, { "epoch": 0.6358974358974359, "grad_norm": 1.203125, "learning_rate": 3.629133528673085e-05, "loss": 0.3387, "step": 868 }, { "epoch": 0.6366300366300366, "grad_norm": 1.421875, "learning_rate": 3.633319380493931e-05, "loss": 0.4322, "step": 869 }, { "epoch": 0.6373626373626373, "grad_norm": 1.484375, "learning_rate": 3.6375052323147764e-05, "loss": 0.4305, "step": 870 }, { "epoch": 0.638095238095238, "grad_norm": 1.3828125, "learning_rate": 3.6416910841356215e-05, "loss": 0.4067, "step": 871 }, { "epoch": 0.6388278388278388, "grad_norm": 1.9375, "learning_rate": 3.645876935956467e-05, "loss": 0.3912, "step": 872 }, { "epoch": 0.6395604395604395, "grad_norm": 1.4765625, "learning_rate": 3.650062787777313e-05, "loss": 0.3899, "step": 873 }, { "epoch": 0.6402930402930402, "grad_norm": 1.3515625, "learning_rate": 3.654248639598159e-05, "loss": 0.3983, "step": 874 }, { "epoch": 0.6410256410256411, "grad_norm": 1.3203125, "learning_rate": 3.658434491419004e-05, "loss": 0.3755, "step": 875 }, { "epoch": 0.6417582417582418, "grad_norm": 1.28125, "learning_rate": 3.6626203432398495e-05, "loss": 0.4345, "step": 876 }, { "epoch": 0.6424908424908425, "grad_norm": 1.6953125, "learning_rate": 3.666806195060695e-05, "loss": 0.3999, "step": 877 }, { "epoch": 0.6432234432234433, "grad_norm": 1.46875, "learning_rate": 3.67099204688154e-05, "loss": 0.4527, "step": 878 }, { "epoch": 0.643956043956044, "grad_norm": 1.546875, "learning_rate": 3.675177898702386e-05, "loss": 0.386, "step": 879 }, { "epoch": 0.6446886446886447, "grad_norm": 1.7578125, "learning_rate": 3.679363750523232e-05, "loss": 0.4485, "step": 880 }, { "epoch": 0.6454212454212455, "grad_norm": 1.640625, "learning_rate": 3.6835496023440775e-05, "loss": 0.3783, "step": 881 }, { "epoch": 0.6461538461538462, "grad_norm": 1.4140625, "learning_rate": 3.6877354541649226e-05, "loss": 0.3935, "step": 882 }, { "epoch": 0.6468864468864469, "grad_norm": 1.296875, "learning_rate": 3.691921305985768e-05, "loss": 0.3847, "step": 883 }, { "epoch": 0.6476190476190476, "grad_norm": 1.6484375, "learning_rate": 3.6961071578066134e-05, "loss": 0.4154, "step": 884 }, { "epoch": 0.6483516483516484, "grad_norm": 1.8515625, "learning_rate": 3.70029300962746e-05, "loss": 0.3996, "step": 885 }, { "epoch": 0.6490842490842491, "grad_norm": 1.359375, "learning_rate": 3.704478861448305e-05, "loss": 0.3948, "step": 886 }, { "epoch": 0.6498168498168498, "grad_norm": 1.5234375, "learning_rate": 3.7086647132691506e-05, "loss": 0.4295, "step": 887 }, { "epoch": 0.6505494505494506, "grad_norm": 1.1640625, "learning_rate": 3.712850565089996e-05, "loss": 0.3598, "step": 888 }, { "epoch": 0.6512820512820513, "grad_norm": 0.94921875, "learning_rate": 3.7170364169108414e-05, "loss": 0.3736, "step": 889 }, { "epoch": 0.652014652014652, "grad_norm": 1.1796875, "learning_rate": 3.721222268731687e-05, "loss": 0.3691, "step": 890 }, { "epoch": 0.6527472527472528, "grad_norm": 1.421875, "learning_rate": 3.725408120552533e-05, "loss": 0.4243, "step": 891 }, { "epoch": 0.6534798534798535, "grad_norm": 1.4453125, "learning_rate": 3.729593972373378e-05, "loss": 0.4798, "step": 892 }, { "epoch": 0.6542124542124542, "grad_norm": 1.59375, "learning_rate": 3.733779824194224e-05, "loss": 0.4103, "step": 893 }, { "epoch": 0.654945054945055, "grad_norm": 1.5546875, "learning_rate": 3.7379656760150694e-05, "loss": 0.4185, "step": 894 }, { "epoch": 0.6556776556776557, "grad_norm": 1.46875, "learning_rate": 3.7421515278359145e-05, "loss": 0.3964, "step": 895 }, { "epoch": 0.6564102564102564, "grad_norm": 1.4296875, "learning_rate": 3.74633737965676e-05, "loss": 0.4425, "step": 896 }, { "epoch": 0.6571428571428571, "grad_norm": 1.3125, "learning_rate": 3.750523231477606e-05, "loss": 0.3486, "step": 897 }, { "epoch": 0.6578754578754579, "grad_norm": 1.390625, "learning_rate": 3.754709083298452e-05, "loss": 0.449, "step": 898 }, { "epoch": 0.6586080586080586, "grad_norm": 1.265625, "learning_rate": 3.758894935119297e-05, "loss": 0.3437, "step": 899 }, { "epoch": 0.6593406593406593, "grad_norm": 1.3046875, "learning_rate": 3.7630807869401425e-05, "loss": 0.3831, "step": 900 }, { "epoch": 0.6600732600732601, "grad_norm": 1.2890625, "learning_rate": 3.7672666387609876e-05, "loss": 0.3864, "step": 901 }, { "epoch": 0.6608058608058608, "grad_norm": 1.265625, "learning_rate": 3.771452490581834e-05, "loss": 0.3917, "step": 902 }, { "epoch": 0.6615384615384615, "grad_norm": 1.5, "learning_rate": 3.775638342402679e-05, "loss": 0.3603, "step": 903 }, { "epoch": 0.6622710622710622, "grad_norm": 1.40625, "learning_rate": 3.779824194223525e-05, "loss": 0.3708, "step": 904 }, { "epoch": 0.663003663003663, "grad_norm": 1.515625, "learning_rate": 3.78401004604437e-05, "loss": 0.3768, "step": 905 }, { "epoch": 0.6637362637362637, "grad_norm": 1.171875, "learning_rate": 3.788195897865216e-05, "loss": 0.3533, "step": 906 }, { "epoch": 0.6644688644688644, "grad_norm": 1.328125, "learning_rate": 3.7923817496860614e-05, "loss": 0.3404, "step": 907 }, { "epoch": 0.6652014652014652, "grad_norm": 1.2734375, "learning_rate": 3.796567601506907e-05, "loss": 0.306, "step": 908 }, { "epoch": 0.6659340659340659, "grad_norm": 1.5703125, "learning_rate": 3.800753453327752e-05, "loss": 0.4156, "step": 909 }, { "epoch": 0.6666666666666666, "grad_norm": 1.25, "learning_rate": 3.804939305148598e-05, "loss": 0.3921, "step": 910 }, { "epoch": 0.6673992673992674, "grad_norm": 1.078125, "learning_rate": 3.8091251569694436e-05, "loss": 0.3611, "step": 911 }, { "epoch": 0.6681318681318681, "grad_norm": 1.328125, "learning_rate": 3.813311008790289e-05, "loss": 0.3946, "step": 912 }, { "epoch": 0.6688644688644688, "grad_norm": 1.515625, "learning_rate": 3.8174968606111344e-05, "loss": 0.4322, "step": 913 }, { "epoch": 0.6695970695970695, "grad_norm": 1.3984375, "learning_rate": 3.8216827124319795e-05, "loss": 0.4075, "step": 914 }, { "epoch": 0.6703296703296703, "grad_norm": 1.6875, "learning_rate": 3.825868564252826e-05, "loss": 0.3493, "step": 915 }, { "epoch": 0.671062271062271, "grad_norm": 1.796875, "learning_rate": 3.830054416073671e-05, "loss": 0.3906, "step": 916 }, { "epoch": 0.6717948717948717, "grad_norm": 1.6171875, "learning_rate": 3.834240267894517e-05, "loss": 0.4161, "step": 917 }, { "epoch": 0.6725274725274726, "grad_norm": 1.3671875, "learning_rate": 3.838426119715362e-05, "loss": 0.3943, "step": 918 }, { "epoch": 0.6732600732600733, "grad_norm": 1.53125, "learning_rate": 3.842611971536208e-05, "loss": 0.3877, "step": 919 }, { "epoch": 0.673992673992674, "grad_norm": 1.8984375, "learning_rate": 3.846797823357053e-05, "loss": 0.4565, "step": 920 }, { "epoch": 0.6747252747252748, "grad_norm": 1.6875, "learning_rate": 3.850983675177899e-05, "loss": 0.4098, "step": 921 }, { "epoch": 0.6754578754578755, "grad_norm": 1.1640625, "learning_rate": 3.855169526998744e-05, "loss": 0.3431, "step": 922 }, { "epoch": 0.6761904761904762, "grad_norm": 1.203125, "learning_rate": 3.8593553788195905e-05, "loss": 0.3599, "step": 923 }, { "epoch": 0.676923076923077, "grad_norm": 1.890625, "learning_rate": 3.8635412306404356e-05, "loss": 0.3813, "step": 924 }, { "epoch": 0.6776556776556777, "grad_norm": 1.5, "learning_rate": 3.867727082461281e-05, "loss": 0.3759, "step": 925 }, { "epoch": 0.6783882783882784, "grad_norm": 1.40625, "learning_rate": 3.8719129342821264e-05, "loss": 0.4188, "step": 926 }, { "epoch": 0.6791208791208792, "grad_norm": 1.0625, "learning_rate": 3.876098786102972e-05, "loss": 0.3778, "step": 927 }, { "epoch": 0.6798534798534799, "grad_norm": 1.546875, "learning_rate": 3.880284637923818e-05, "loss": 0.3558, "step": 928 }, { "epoch": 0.6805860805860806, "grad_norm": 1.5546875, "learning_rate": 3.884470489744663e-05, "loss": 0.4619, "step": 929 }, { "epoch": 0.6813186813186813, "grad_norm": 1.1953125, "learning_rate": 3.8886563415655086e-05, "loss": 0.302, "step": 930 }, { "epoch": 0.6820512820512821, "grad_norm": 1.2421875, "learning_rate": 3.8928421933863544e-05, "loss": 0.3261, "step": 931 }, { "epoch": 0.6827838827838828, "grad_norm": 1.296875, "learning_rate": 3.8970280452072e-05, "loss": 0.3708, "step": 932 }, { "epoch": 0.6835164835164835, "grad_norm": 1.203125, "learning_rate": 3.901213897028045e-05, "loss": 0.353, "step": 933 }, { "epoch": 0.6842490842490843, "grad_norm": 1.3203125, "learning_rate": 3.905399748848891e-05, "loss": 0.344, "step": 934 }, { "epoch": 0.684981684981685, "grad_norm": 1.3125, "learning_rate": 3.909585600669736e-05, "loss": 0.3726, "step": 935 }, { "epoch": 0.6857142857142857, "grad_norm": 1.328125, "learning_rate": 3.9137714524905824e-05, "loss": 0.3995, "step": 936 }, { "epoch": 0.6864468864468865, "grad_norm": 1.203125, "learning_rate": 3.9179573043114275e-05, "loss": 0.3734, "step": 937 }, { "epoch": 0.6871794871794872, "grad_norm": 1.28125, "learning_rate": 3.922143156132273e-05, "loss": 0.3406, "step": 938 }, { "epoch": 0.6879120879120879, "grad_norm": 1.578125, "learning_rate": 3.926329007953118e-05, "loss": 0.3377, "step": 939 }, { "epoch": 0.6886446886446886, "grad_norm": 1.109375, "learning_rate": 3.930514859773965e-05, "loss": 0.3389, "step": 940 }, { "epoch": 0.6893772893772894, "grad_norm": 1.7265625, "learning_rate": 3.93470071159481e-05, "loss": 0.4103, "step": 941 }, { "epoch": 0.6901098901098901, "grad_norm": 1.546875, "learning_rate": 3.9388865634156555e-05, "loss": 0.4905, "step": 942 }, { "epoch": 0.6908424908424908, "grad_norm": 1.515625, "learning_rate": 3.9430724152365006e-05, "loss": 0.4848, "step": 943 }, { "epoch": 0.6915750915750916, "grad_norm": 1.2578125, "learning_rate": 3.947258267057346e-05, "loss": 0.3939, "step": 944 }, { "epoch": 0.6923076923076923, "grad_norm": 1.3359375, "learning_rate": 3.951444118878192e-05, "loss": 0.391, "step": 945 }, { "epoch": 0.693040293040293, "grad_norm": 1.7421875, "learning_rate": 3.955629970699037e-05, "loss": 0.3847, "step": 946 }, { "epoch": 0.6937728937728938, "grad_norm": 1.7265625, "learning_rate": 3.959815822519883e-05, "loss": 0.3882, "step": 947 }, { "epoch": 0.6945054945054945, "grad_norm": 1.6640625, "learning_rate": 3.9640016743407286e-05, "loss": 0.3514, "step": 948 }, { "epoch": 0.6952380952380952, "grad_norm": 1.75, "learning_rate": 3.968187526161574e-05, "loss": 0.4052, "step": 949 }, { "epoch": 0.6959706959706959, "grad_norm": 1.3359375, "learning_rate": 3.9723733779824194e-05, "loss": 0.3727, "step": 950 }, { "epoch": 0.6967032967032967, "grad_norm": 1.5390625, "learning_rate": 3.976559229803265e-05, "loss": 0.3508, "step": 951 }, { "epoch": 0.6974358974358974, "grad_norm": 1.2890625, "learning_rate": 3.980745081624111e-05, "loss": 0.3848, "step": 952 }, { "epoch": 0.6981684981684981, "grad_norm": 1.359375, "learning_rate": 3.9849309334449566e-05, "loss": 0.3389, "step": 953 }, { "epoch": 0.6989010989010989, "grad_norm": 1.1875, "learning_rate": 3.989116785265802e-05, "loss": 0.3576, "step": 954 }, { "epoch": 0.6996336996336996, "grad_norm": 2.34375, "learning_rate": 3.9933026370866474e-05, "loss": 0.4554, "step": 955 }, { "epoch": 0.7003663003663003, "grad_norm": 1.5, "learning_rate": 3.9974884889074925e-05, "loss": 0.3938, "step": 956 }, { "epoch": 0.701098901098901, "grad_norm": 1.1875, "learning_rate": 4.001674340728339e-05, "loss": 0.3303, "step": 957 }, { "epoch": 0.7018315018315018, "grad_norm": 0.8359375, "learning_rate": 4.005860192549184e-05, "loss": 0.2822, "step": 958 }, { "epoch": 0.7025641025641025, "grad_norm": 1.265625, "learning_rate": 4.01004604437003e-05, "loss": 0.36, "step": 959 }, { "epoch": 0.7032967032967034, "grad_norm": 1.1015625, "learning_rate": 4.014231896190875e-05, "loss": 0.321, "step": 960 }, { "epoch": 0.7040293040293041, "grad_norm": 1.0859375, "learning_rate": 4.0184177480117205e-05, "loss": 0.4033, "step": 961 }, { "epoch": 0.7047619047619048, "grad_norm": 1.0625, "learning_rate": 4.022603599832566e-05, "loss": 0.3827, "step": 962 }, { "epoch": 0.7054945054945055, "grad_norm": 1.65625, "learning_rate": 4.026789451653411e-05, "loss": 0.3755, "step": 963 }, { "epoch": 0.7062271062271063, "grad_norm": 1.296875, "learning_rate": 4.030975303474257e-05, "loss": 0.361, "step": 964 }, { "epoch": 0.706959706959707, "grad_norm": 1.671875, "learning_rate": 4.035161155295103e-05, "loss": 0.3703, "step": 965 }, { "epoch": 0.7076923076923077, "grad_norm": 1.203125, "learning_rate": 4.0393470071159485e-05, "loss": 0.3844, "step": 966 }, { "epoch": 0.7084249084249085, "grad_norm": 0.95703125, "learning_rate": 4.0435328589367936e-05, "loss": 0.3421, "step": 967 }, { "epoch": 0.7091575091575092, "grad_norm": 1.1953125, "learning_rate": 4.047718710757639e-05, "loss": 0.3383, "step": 968 }, { "epoch": 0.7098901098901099, "grad_norm": 1.421875, "learning_rate": 4.051904562578485e-05, "loss": 0.3317, "step": 969 }, { "epoch": 0.7106227106227107, "grad_norm": 1.171875, "learning_rate": 4.056090414399331e-05, "loss": 0.3234, "step": 970 }, { "epoch": 0.7113553113553114, "grad_norm": 1.484375, "learning_rate": 4.060276266220176e-05, "loss": 0.435, "step": 971 }, { "epoch": 0.7120879120879121, "grad_norm": 1.890625, "learning_rate": 4.0644621180410216e-05, "loss": 0.3991, "step": 972 }, { "epoch": 0.7128205128205128, "grad_norm": 1.40625, "learning_rate": 4.0686479698618674e-05, "loss": 0.3757, "step": 973 }, { "epoch": 0.7135531135531136, "grad_norm": 1.6484375, "learning_rate": 4.072833821682713e-05, "loss": 0.3899, "step": 974 }, { "epoch": 0.7142857142857143, "grad_norm": 1.9296875, "learning_rate": 4.077019673503558e-05, "loss": 0.3695, "step": 975 }, { "epoch": 0.715018315018315, "grad_norm": 1.515625, "learning_rate": 4.081205525324404e-05, "loss": 0.4072, "step": 976 }, { "epoch": 0.7157509157509158, "grad_norm": 1.4609375, "learning_rate": 4.085391377145249e-05, "loss": 0.3844, "step": 977 }, { "epoch": 0.7164835164835165, "grad_norm": 1.03125, "learning_rate": 4.089577228966095e-05, "loss": 0.336, "step": 978 }, { "epoch": 0.7172161172161172, "grad_norm": 1.2265625, "learning_rate": 4.0937630807869404e-05, "loss": 0.4256, "step": 979 }, { "epoch": 0.717948717948718, "grad_norm": 2.0625, "learning_rate": 4.0979489326077855e-05, "loss": 0.3641, "step": 980 }, { "epoch": 0.7186813186813187, "grad_norm": 1.3515625, "learning_rate": 4.102134784428631e-05, "loss": 0.347, "step": 981 }, { "epoch": 0.7194139194139194, "grad_norm": 1.2109375, "learning_rate": 4.106320636249477e-05, "loss": 0.3772, "step": 982 }, { "epoch": 0.7201465201465201, "grad_norm": 1.2421875, "learning_rate": 4.110506488070323e-05, "loss": 0.3885, "step": 983 }, { "epoch": 0.7208791208791209, "grad_norm": 1.5078125, "learning_rate": 4.114692339891168e-05, "loss": 0.4176, "step": 984 }, { "epoch": 0.7216117216117216, "grad_norm": 1.4140625, "learning_rate": 4.1188781917120135e-05, "loss": 0.3494, "step": 985 }, { "epoch": 0.7223443223443223, "grad_norm": 1.3046875, "learning_rate": 4.123064043532859e-05, "loss": 0.4289, "step": 986 }, { "epoch": 0.7230769230769231, "grad_norm": 1.0625, "learning_rate": 4.127249895353705e-05, "loss": 0.3288, "step": 987 }, { "epoch": 0.7238095238095238, "grad_norm": 1.296875, "learning_rate": 4.13143574717455e-05, "loss": 0.4229, "step": 988 }, { "epoch": 0.7245421245421245, "grad_norm": 1.328125, "learning_rate": 4.135621598995396e-05, "loss": 0.3738, "step": 989 }, { "epoch": 0.7252747252747253, "grad_norm": 1.171875, "learning_rate": 4.1398074508162416e-05, "loss": 0.3322, "step": 990 }, { "epoch": 0.726007326007326, "grad_norm": 1.1953125, "learning_rate": 4.143993302637087e-05, "loss": 0.3651, "step": 991 }, { "epoch": 0.7267399267399267, "grad_norm": 1.1484375, "learning_rate": 4.1481791544579324e-05, "loss": 0.3512, "step": 992 }, { "epoch": 0.7274725274725274, "grad_norm": 1.1875, "learning_rate": 4.152365006278778e-05, "loss": 0.3659, "step": 993 }, { "epoch": 0.7282051282051282, "grad_norm": 1.2421875, "learning_rate": 4.156550858099624e-05, "loss": 0.3289, "step": 994 }, { "epoch": 0.7289377289377289, "grad_norm": 1.4609375, "learning_rate": 4.160736709920469e-05, "loss": 0.3508, "step": 995 }, { "epoch": 0.7296703296703296, "grad_norm": 1.171875, "learning_rate": 4.1649225617413146e-05, "loss": 0.3953, "step": 996 }, { "epoch": 0.7304029304029304, "grad_norm": 1.2109375, "learning_rate": 4.16910841356216e-05, "loss": 0.367, "step": 997 }, { "epoch": 0.7311355311355311, "grad_norm": 1.5390625, "learning_rate": 4.1732942653830054e-05, "loss": 0.4167, "step": 998 }, { "epoch": 0.7318681318681318, "grad_norm": 1.3984375, "learning_rate": 4.177480117203851e-05, "loss": 0.3509, "step": 999 }, { "epoch": 0.7326007326007326, "grad_norm": 1.296875, "learning_rate": 4.181665969024697e-05, "loss": 0.3452, "step": 1000 }, { "epoch": 0.7333333333333333, "grad_norm": 1.125, "learning_rate": 4.185851820845542e-05, "loss": 0.2966, "step": 1001 }, { "epoch": 0.734065934065934, "grad_norm": 1.2734375, "learning_rate": 4.190037672666388e-05, "loss": 0.3474, "step": 1002 }, { "epoch": 0.7347985347985349, "grad_norm": 1.3671875, "learning_rate": 4.1942235244872335e-05, "loss": 0.3514, "step": 1003 }, { "epoch": 0.7355311355311356, "grad_norm": 1.21875, "learning_rate": 4.198409376308079e-05, "loss": 0.3518, "step": 1004 }, { "epoch": 0.7362637362637363, "grad_norm": 1.203125, "learning_rate": 4.202595228128924e-05, "loss": 0.3431, "step": 1005 }, { "epoch": 0.736996336996337, "grad_norm": 0.98828125, "learning_rate": 4.20678107994977e-05, "loss": 0.3733, "step": 1006 }, { "epoch": 0.7377289377289378, "grad_norm": 1.21875, "learning_rate": 4.210966931770616e-05, "loss": 0.3748, "step": 1007 }, { "epoch": 0.7384615384615385, "grad_norm": 1.234375, "learning_rate": 4.215152783591461e-05, "loss": 0.3598, "step": 1008 }, { "epoch": 0.7391941391941392, "grad_norm": 1.2578125, "learning_rate": 4.2193386354123066e-05, "loss": 0.281, "step": 1009 }, { "epoch": 0.73992673992674, "grad_norm": 1.1640625, "learning_rate": 4.223524487233152e-05, "loss": 0.3302, "step": 1010 }, { "epoch": 0.7406593406593407, "grad_norm": 1.125, "learning_rate": 4.227710339053998e-05, "loss": 0.38, "step": 1011 }, { "epoch": 0.7413919413919414, "grad_norm": 1.4375, "learning_rate": 4.231896190874843e-05, "loss": 0.3676, "step": 1012 }, { "epoch": 0.7421245421245422, "grad_norm": 1.4765625, "learning_rate": 4.236082042695689e-05, "loss": 0.3894, "step": 1013 }, { "epoch": 0.7428571428571429, "grad_norm": 1.578125, "learning_rate": 4.240267894516534e-05, "loss": 0.3549, "step": 1014 }, { "epoch": 0.7435897435897436, "grad_norm": 1.109375, "learning_rate": 4.24445374633738e-05, "loss": 0.3683, "step": 1015 }, { "epoch": 0.7443223443223443, "grad_norm": 1.4140625, "learning_rate": 4.2486395981582254e-05, "loss": 0.3233, "step": 1016 }, { "epoch": 0.7450549450549451, "grad_norm": 1.15625, "learning_rate": 4.252825449979071e-05, "loss": 0.3442, "step": 1017 }, { "epoch": 0.7457875457875458, "grad_norm": 1.1640625, "learning_rate": 4.257011301799916e-05, "loss": 0.3606, "step": 1018 }, { "epoch": 0.7465201465201465, "grad_norm": 1.375, "learning_rate": 4.261197153620762e-05, "loss": 0.3673, "step": 1019 }, { "epoch": 0.7472527472527473, "grad_norm": 1.3203125, "learning_rate": 4.265383005441608e-05, "loss": 0.3233, "step": 1020 }, { "epoch": 0.747985347985348, "grad_norm": 1.4375, "learning_rate": 4.2695688572624534e-05, "loss": 0.4093, "step": 1021 }, { "epoch": 0.7487179487179487, "grad_norm": 1.546875, "learning_rate": 4.2737547090832985e-05, "loss": 0.3707, "step": 1022 }, { "epoch": 0.7494505494505495, "grad_norm": 1.5859375, "learning_rate": 4.277940560904144e-05, "loss": 0.3422, "step": 1023 }, { "epoch": 0.7501831501831502, "grad_norm": 1.0703125, "learning_rate": 4.28212641272499e-05, "loss": 0.3296, "step": 1024 }, { "epoch": 0.7509157509157509, "grad_norm": 1.4453125, "learning_rate": 4.286312264545835e-05, "loss": 0.3416, "step": 1025 }, { "epoch": 0.7516483516483516, "grad_norm": 1.7265625, "learning_rate": 4.290498116366681e-05, "loss": 0.322, "step": 1026 }, { "epoch": 0.7523809523809524, "grad_norm": 1.8203125, "learning_rate": 4.2946839681875265e-05, "loss": 0.3981, "step": 1027 }, { "epoch": 0.7531135531135531, "grad_norm": 1.5, "learning_rate": 4.298869820008372e-05, "loss": 0.3479, "step": 1028 }, { "epoch": 0.7538461538461538, "grad_norm": 1.5078125, "learning_rate": 4.303055671829217e-05, "loss": 0.4045, "step": 1029 }, { "epoch": 0.7545787545787546, "grad_norm": 1.1796875, "learning_rate": 4.307241523650063e-05, "loss": 0.3372, "step": 1030 }, { "epoch": 0.7553113553113553, "grad_norm": 1.4375, "learning_rate": 4.311427375470908e-05, "loss": 0.3991, "step": 1031 }, { "epoch": 0.756043956043956, "grad_norm": 1.921875, "learning_rate": 4.3156132272917545e-05, "loss": 0.3878, "step": 1032 }, { "epoch": 0.7567765567765568, "grad_norm": 1.5859375, "learning_rate": 4.3197990791125996e-05, "loss": 0.3878, "step": 1033 }, { "epoch": 0.7575091575091575, "grad_norm": 1.6015625, "learning_rate": 4.323984930933445e-05, "loss": 0.4261, "step": 1034 }, { "epoch": 0.7582417582417582, "grad_norm": 1.9296875, "learning_rate": 4.3281707827542904e-05, "loss": 0.3292, "step": 1035 }, { "epoch": 0.7589743589743589, "grad_norm": 1.9921875, "learning_rate": 4.332356634575137e-05, "loss": 0.3424, "step": 1036 }, { "epoch": 0.7597069597069597, "grad_norm": 1.3046875, "learning_rate": 4.336542486395982e-05, "loss": 0.422, "step": 1037 }, { "epoch": 0.7604395604395604, "grad_norm": 1.453125, "learning_rate": 4.3407283382168276e-05, "loss": 0.3961, "step": 1038 }, { "epoch": 0.7611721611721611, "grad_norm": 1.421875, "learning_rate": 4.344914190037673e-05, "loss": 0.3594, "step": 1039 }, { "epoch": 0.7619047619047619, "grad_norm": 1.4921875, "learning_rate": 4.3491000418585184e-05, "loss": 0.3169, "step": 1040 }, { "epoch": 0.7626373626373626, "grad_norm": 1.421875, "learning_rate": 4.353285893679364e-05, "loss": 0.3374, "step": 1041 }, { "epoch": 0.7633699633699633, "grad_norm": 1.03125, "learning_rate": 4.357471745500209e-05, "loss": 0.3506, "step": 1042 }, { "epoch": 0.764102564102564, "grad_norm": 1.2578125, "learning_rate": 4.361657597321055e-05, "loss": 0.3241, "step": 1043 }, { "epoch": 0.7648351648351648, "grad_norm": 1.40625, "learning_rate": 4.3658434491419e-05, "loss": 0.3735, "step": 1044 }, { "epoch": 0.7655677655677655, "grad_norm": 1.203125, "learning_rate": 4.3700293009627464e-05, "loss": 0.3097, "step": 1045 }, { "epoch": 0.7663003663003664, "grad_norm": 1.2421875, "learning_rate": 4.3742151527835915e-05, "loss": 0.4288, "step": 1046 }, { "epoch": 0.7670329670329671, "grad_norm": 0.9921875, "learning_rate": 4.378401004604437e-05, "loss": 0.3383, "step": 1047 }, { "epoch": 0.7677655677655678, "grad_norm": 1.0625, "learning_rate": 4.382586856425282e-05, "loss": 0.2911, "step": 1048 }, { "epoch": 0.7684981684981685, "grad_norm": 1.0546875, "learning_rate": 4.386772708246129e-05, "loss": 0.3419, "step": 1049 }, { "epoch": 0.7692307692307693, "grad_norm": 1.5703125, "learning_rate": 4.390958560066974e-05, "loss": 0.3201, "step": 1050 }, { "epoch": 0.76996336996337, "grad_norm": 1.3203125, "learning_rate": 4.3951444118878195e-05, "loss": 0.3667, "step": 1051 }, { "epoch": 0.7706959706959707, "grad_norm": 1.1640625, "learning_rate": 4.3993302637086646e-05, "loss": 0.3643, "step": 1052 }, { "epoch": 0.7714285714285715, "grad_norm": 1.140625, "learning_rate": 4.403516115529511e-05, "loss": 0.3309, "step": 1053 }, { "epoch": 0.7721611721611722, "grad_norm": 1.171875, "learning_rate": 4.407701967350356e-05, "loss": 0.3481, "step": 1054 }, { "epoch": 0.7728937728937729, "grad_norm": 1.0703125, "learning_rate": 4.411887819171202e-05, "loss": 0.373, "step": 1055 }, { "epoch": 0.7736263736263737, "grad_norm": 1.484375, "learning_rate": 4.416073670992047e-05, "loss": 0.3828, "step": 1056 }, { "epoch": 0.7743589743589744, "grad_norm": 1.140625, "learning_rate": 4.4202595228128926e-05, "loss": 0.3301, "step": 1057 }, { "epoch": 0.7750915750915751, "grad_norm": 1.3359375, "learning_rate": 4.4244453746337384e-05, "loss": 0.4091, "step": 1058 }, { "epoch": 0.7758241758241758, "grad_norm": 1.3125, "learning_rate": 4.4286312264545834e-05, "loss": 0.3545, "step": 1059 }, { "epoch": 0.7765567765567766, "grad_norm": 1.25, "learning_rate": 4.432817078275429e-05, "loss": 0.3894, "step": 1060 }, { "epoch": 0.7772893772893773, "grad_norm": 1.4453125, "learning_rate": 4.437002930096274e-05, "loss": 0.382, "step": 1061 }, { "epoch": 0.778021978021978, "grad_norm": 1.359375, "learning_rate": 4.4411887819171206e-05, "loss": 0.3906, "step": 1062 }, { "epoch": 0.7787545787545788, "grad_norm": 1.1171875, "learning_rate": 4.445374633737966e-05, "loss": 0.3607, "step": 1063 }, { "epoch": 0.7794871794871795, "grad_norm": 1.5390625, "learning_rate": 4.4495604855588114e-05, "loss": 0.4029, "step": 1064 }, { "epoch": 0.7802197802197802, "grad_norm": 1.296875, "learning_rate": 4.4537463373796565e-05, "loss": 0.3409, "step": 1065 }, { "epoch": 0.780952380952381, "grad_norm": 1.1484375, "learning_rate": 4.457932189200503e-05, "loss": 0.3531, "step": 1066 }, { "epoch": 0.7816849816849817, "grad_norm": 1.65625, "learning_rate": 4.462118041021348e-05, "loss": 0.3717, "step": 1067 }, { "epoch": 0.7824175824175824, "grad_norm": 1.4609375, "learning_rate": 4.466303892842194e-05, "loss": 0.3227, "step": 1068 }, { "epoch": 0.7831501831501831, "grad_norm": 1.90625, "learning_rate": 4.470489744663039e-05, "loss": 0.3061, "step": 1069 }, { "epoch": 0.7838827838827839, "grad_norm": 1.4296875, "learning_rate": 4.474675596483885e-05, "loss": 0.3324, "step": 1070 }, { "epoch": 0.7846153846153846, "grad_norm": 1.2421875, "learning_rate": 4.47886144830473e-05, "loss": 0.38, "step": 1071 }, { "epoch": 0.7853479853479853, "grad_norm": 1.1171875, "learning_rate": 4.483047300125576e-05, "loss": 0.3459, "step": 1072 }, { "epoch": 0.7860805860805861, "grad_norm": 1.53125, "learning_rate": 4.487233151946421e-05, "loss": 0.324, "step": 1073 }, { "epoch": 0.7868131868131868, "grad_norm": 1.34375, "learning_rate": 4.491419003767267e-05, "loss": 0.3174, "step": 1074 }, { "epoch": 0.7875457875457875, "grad_norm": 1.125, "learning_rate": 4.4956048555881126e-05, "loss": 0.2988, "step": 1075 }, { "epoch": 0.7882783882783883, "grad_norm": 1.6484375, "learning_rate": 4.4997907074089576e-05, "loss": 0.3528, "step": 1076 }, { "epoch": 0.789010989010989, "grad_norm": 1.28125, "learning_rate": 4.5039765592298034e-05, "loss": 0.3023, "step": 1077 }, { "epoch": 0.7897435897435897, "grad_norm": 1.5546875, "learning_rate": 4.508162411050649e-05, "loss": 0.3694, "step": 1078 }, { "epoch": 0.7904761904761904, "grad_norm": 1.71875, "learning_rate": 4.512348262871495e-05, "loss": 0.409, "step": 1079 }, { "epoch": 0.7912087912087912, "grad_norm": 0.96875, "learning_rate": 4.51653411469234e-05, "loss": 0.3077, "step": 1080 }, { "epoch": 0.7919413919413919, "grad_norm": 1.1484375, "learning_rate": 4.5207199665131856e-05, "loss": 0.3258, "step": 1081 }, { "epoch": 0.7926739926739926, "grad_norm": 1.578125, "learning_rate": 4.524905818334031e-05, "loss": 0.3557, "step": 1082 }, { "epoch": 0.7934065934065934, "grad_norm": 1.03125, "learning_rate": 4.529091670154877e-05, "loss": 0.2748, "step": 1083 }, { "epoch": 0.7941391941391941, "grad_norm": 1.375, "learning_rate": 4.533277521975722e-05, "loss": 0.3273, "step": 1084 }, { "epoch": 0.7948717948717948, "grad_norm": 1.265625, "learning_rate": 4.537463373796568e-05, "loss": 0.316, "step": 1085 }, { "epoch": 0.7956043956043956, "grad_norm": 1.25, "learning_rate": 4.541649225617413e-05, "loss": 0.3406, "step": 1086 }, { "epoch": 0.7963369963369963, "grad_norm": 0.9375, "learning_rate": 4.5458350774382594e-05, "loss": 0.2846, "step": 1087 }, { "epoch": 0.7970695970695971, "grad_norm": 1.2109375, "learning_rate": 4.5500209292591045e-05, "loss": 0.3359, "step": 1088 }, { "epoch": 0.7978021978021979, "grad_norm": 0.921875, "learning_rate": 4.55420678107995e-05, "loss": 0.3328, "step": 1089 }, { "epoch": 0.7985347985347986, "grad_norm": 0.98046875, "learning_rate": 4.558392632900795e-05, "loss": 0.2777, "step": 1090 }, { "epoch": 0.7992673992673993, "grad_norm": 1.0546875, "learning_rate": 4.562578484721641e-05, "loss": 0.3321, "step": 1091 }, { "epoch": 0.8, "grad_norm": 1.1015625, "learning_rate": 4.566764336542487e-05, "loss": 0.3524, "step": 1092 }, { "epoch": 0.8007326007326008, "grad_norm": 1.3671875, "learning_rate": 4.570950188363332e-05, "loss": 0.3581, "step": 1093 }, { "epoch": 0.8014652014652015, "grad_norm": 1.296875, "learning_rate": 4.5751360401841776e-05, "loss": 0.3498, "step": 1094 }, { "epoch": 0.8021978021978022, "grad_norm": 1.3359375, "learning_rate": 4.579321892005023e-05, "loss": 0.3546, "step": 1095 }, { "epoch": 0.802930402930403, "grad_norm": 1.03125, "learning_rate": 4.583507743825869e-05, "loss": 0.3145, "step": 1096 }, { "epoch": 0.8036630036630037, "grad_norm": 0.953125, "learning_rate": 4.587693595646714e-05, "loss": 0.2703, "step": 1097 }, { "epoch": 0.8043956043956044, "grad_norm": 1.4375, "learning_rate": 4.59187944746756e-05, "loss": 0.3265, "step": 1098 }, { "epoch": 0.8051282051282052, "grad_norm": 1.5234375, "learning_rate": 4.5960652992884056e-05, "loss": 0.3745, "step": 1099 }, { "epoch": 0.8058608058608059, "grad_norm": 1.109375, "learning_rate": 4.600251151109251e-05, "loss": 0.3202, "step": 1100 }, { "epoch": 0.8065934065934066, "grad_norm": 1.1796875, "learning_rate": 4.6044370029300964e-05, "loss": 0.3483, "step": 1101 }, { "epoch": 0.8073260073260073, "grad_norm": 1.28125, "learning_rate": 4.608622854750942e-05, "loss": 0.3402, "step": 1102 }, { "epoch": 0.8080586080586081, "grad_norm": 1.421875, "learning_rate": 4.612808706571787e-05, "loss": 0.3101, "step": 1103 }, { "epoch": 0.8087912087912088, "grad_norm": 1.6796875, "learning_rate": 4.6169945583926336e-05, "loss": 0.387, "step": 1104 }, { "epoch": 0.8095238095238095, "grad_norm": 1.3046875, "learning_rate": 4.621180410213479e-05, "loss": 0.3614, "step": 1105 }, { "epoch": 0.8102564102564103, "grad_norm": 2.234375, "learning_rate": 4.6253662620343244e-05, "loss": 0.4022, "step": 1106 }, { "epoch": 0.810989010989011, "grad_norm": 1.109375, "learning_rate": 4.6295521138551695e-05, "loss": 0.3138, "step": 1107 }, { "epoch": 0.8117216117216117, "grad_norm": 1.0, "learning_rate": 4.633737965676015e-05, "loss": 0.2883, "step": 1108 }, { "epoch": 0.8124542124542125, "grad_norm": 1.515625, "learning_rate": 4.637923817496861e-05, "loss": 0.3782, "step": 1109 }, { "epoch": 0.8131868131868132, "grad_norm": 1.328125, "learning_rate": 4.642109669317706e-05, "loss": 0.3092, "step": 1110 }, { "epoch": 0.8139194139194139, "grad_norm": 1.3125, "learning_rate": 4.646295521138552e-05, "loss": 0.3209, "step": 1111 }, { "epoch": 0.8146520146520146, "grad_norm": 1.21875, "learning_rate": 4.6504813729593975e-05, "loss": 0.3464, "step": 1112 }, { "epoch": 0.8153846153846154, "grad_norm": 1.453125, "learning_rate": 4.654667224780243e-05, "loss": 0.2979, "step": 1113 }, { "epoch": 0.8161172161172161, "grad_norm": 1.3046875, "learning_rate": 4.658853076601088e-05, "loss": 0.3111, "step": 1114 }, { "epoch": 0.8168498168498168, "grad_norm": 1.078125, "learning_rate": 4.663038928421934e-05, "loss": 0.2719, "step": 1115 }, { "epoch": 0.8175824175824176, "grad_norm": 1.515625, "learning_rate": 4.66722478024278e-05, "loss": 0.3838, "step": 1116 }, { "epoch": 0.8183150183150183, "grad_norm": 1.1796875, "learning_rate": 4.6714106320636255e-05, "loss": 0.3158, "step": 1117 }, { "epoch": 0.819047619047619, "grad_norm": 1.09375, "learning_rate": 4.6755964838844706e-05, "loss": 0.2809, "step": 1118 }, { "epoch": 0.8197802197802198, "grad_norm": 1.2109375, "learning_rate": 4.679782335705316e-05, "loss": 0.3394, "step": 1119 }, { "epoch": 0.8205128205128205, "grad_norm": 1.2421875, "learning_rate": 4.683968187526162e-05, "loss": 0.3513, "step": 1120 }, { "epoch": 0.8212454212454212, "grad_norm": 1.3046875, "learning_rate": 4.688154039347007e-05, "loss": 0.339, "step": 1121 }, { "epoch": 0.8219780219780219, "grad_norm": 1.2109375, "learning_rate": 4.692339891167853e-05, "loss": 0.294, "step": 1122 }, { "epoch": 0.8227106227106227, "grad_norm": 1.1328125, "learning_rate": 4.6965257429886986e-05, "loss": 0.3818, "step": 1123 }, { "epoch": 0.8234432234432234, "grad_norm": 1.8828125, "learning_rate": 4.700711594809544e-05, "loss": 0.2955, "step": 1124 }, { "epoch": 0.8241758241758241, "grad_norm": 1.2890625, "learning_rate": 4.7048974466303894e-05, "loss": 0.3802, "step": 1125 }, { "epoch": 0.8249084249084249, "grad_norm": 1.1015625, "learning_rate": 4.709083298451235e-05, "loss": 0.3151, "step": 1126 }, { "epoch": 0.8256410256410256, "grad_norm": 1.375, "learning_rate": 4.71326915027208e-05, "loss": 0.3543, "step": 1127 }, { "epoch": 0.8263736263736263, "grad_norm": 1.96875, "learning_rate": 4.717455002092926e-05, "loss": 0.3773, "step": 1128 }, { "epoch": 0.827106227106227, "grad_norm": 1.78125, "learning_rate": 4.721640853913772e-05, "loss": 0.3083, "step": 1129 }, { "epoch": 0.8278388278388278, "grad_norm": 1.78125, "learning_rate": 4.7258267057346174e-05, "loss": 0.3403, "step": 1130 }, { "epoch": 0.8285714285714286, "grad_norm": 1.6796875, "learning_rate": 4.7300125575554625e-05, "loss": 0.3107, "step": 1131 }, { "epoch": 0.8293040293040294, "grad_norm": 1.4765625, "learning_rate": 4.734198409376308e-05, "loss": 0.3338, "step": 1132 }, { "epoch": 0.8300366300366301, "grad_norm": 1.4140625, "learning_rate": 4.738384261197154e-05, "loss": 0.2542, "step": 1133 }, { "epoch": 0.8307692307692308, "grad_norm": 1.2265625, "learning_rate": 4.742570113018e-05, "loss": 0.357, "step": 1134 }, { "epoch": 0.8315018315018315, "grad_norm": 1.4296875, "learning_rate": 4.746755964838845e-05, "loss": 0.403, "step": 1135 }, { "epoch": 0.8322344322344323, "grad_norm": 1.4296875, "learning_rate": 4.7509418166596905e-05, "loss": 0.3912, "step": 1136 }, { "epoch": 0.832967032967033, "grad_norm": 1.03125, "learning_rate": 4.755127668480536e-05, "loss": 0.3077, "step": 1137 }, { "epoch": 0.8336996336996337, "grad_norm": 0.96875, "learning_rate": 4.7593135203013813e-05, "loss": 0.2936, "step": 1138 }, { "epoch": 0.8344322344322345, "grad_norm": 1.0546875, "learning_rate": 4.763499372122227e-05, "loss": 0.33, "step": 1139 }, { "epoch": 0.8351648351648352, "grad_norm": 1.015625, "learning_rate": 4.767685223943073e-05, "loss": 0.3394, "step": 1140 }, { "epoch": 0.8358974358974359, "grad_norm": 1.140625, "learning_rate": 4.7718710757639186e-05, "loss": 0.335, "step": 1141 }, { "epoch": 0.8366300366300367, "grad_norm": 1.046875, "learning_rate": 4.7760569275847636e-05, "loss": 0.2788, "step": 1142 }, { "epoch": 0.8373626373626374, "grad_norm": 1.1796875, "learning_rate": 4.7802427794056094e-05, "loss": 0.3485, "step": 1143 }, { "epoch": 0.8380952380952381, "grad_norm": 1.2890625, "learning_rate": 4.7844286312264544e-05, "loss": 0.3262, "step": 1144 }, { "epoch": 0.8388278388278388, "grad_norm": 1.328125, "learning_rate": 4.7886144830473e-05, "loss": 0.3697, "step": 1145 }, { "epoch": 0.8395604395604396, "grad_norm": 1.2109375, "learning_rate": 4.792800334868146e-05, "loss": 0.3263, "step": 1146 }, { "epoch": 0.8402930402930403, "grad_norm": 1.5703125, "learning_rate": 4.7969861866889916e-05, "loss": 0.3567, "step": 1147 }, { "epoch": 0.841025641025641, "grad_norm": 1.2890625, "learning_rate": 4.801172038509837e-05, "loss": 0.338, "step": 1148 }, { "epoch": 0.8417582417582418, "grad_norm": 1.1640625, "learning_rate": 4.8053578903306825e-05, "loss": 0.3482, "step": 1149 }, { "epoch": 0.8424908424908425, "grad_norm": 1.34375, "learning_rate": 4.809543742151528e-05, "loss": 0.2967, "step": 1150 }, { "epoch": 0.8432234432234432, "grad_norm": 1.421875, "learning_rate": 4.813729593972374e-05, "loss": 0.3051, "step": 1151 }, { "epoch": 0.843956043956044, "grad_norm": 1.4921875, "learning_rate": 4.817915445793219e-05, "loss": 0.3388, "step": 1152 }, { "epoch": 0.8446886446886447, "grad_norm": 1.171875, "learning_rate": 4.822101297614065e-05, "loss": 0.3177, "step": 1153 }, { "epoch": 0.8454212454212454, "grad_norm": 1.2890625, "learning_rate": 4.8262871494349105e-05, "loss": 0.254, "step": 1154 }, { "epoch": 0.8461538461538461, "grad_norm": 1.4375, "learning_rate": 4.8304730012557555e-05, "loss": 0.3318, "step": 1155 }, { "epoch": 0.8468864468864469, "grad_norm": 1.203125, "learning_rate": 4.834658853076601e-05, "loss": 0.2689, "step": 1156 }, { "epoch": 0.8476190476190476, "grad_norm": 0.9375, "learning_rate": 4.8388447048974463e-05, "loss": 0.3172, "step": 1157 }, { "epoch": 0.8483516483516483, "grad_norm": 1.71875, "learning_rate": 4.843030556718293e-05, "loss": 0.4396, "step": 1158 }, { "epoch": 0.8490842490842491, "grad_norm": 1.734375, "learning_rate": 4.847216408539138e-05, "loss": 0.3502, "step": 1159 }, { "epoch": 0.8498168498168498, "grad_norm": 1.5, "learning_rate": 4.8514022603599836e-05, "loss": 0.3461, "step": 1160 }, { "epoch": 0.8505494505494505, "grad_norm": 1.28125, "learning_rate": 4.8555881121808286e-05, "loss": 0.2724, "step": 1161 }, { "epoch": 0.8512820512820513, "grad_norm": 1.234375, "learning_rate": 4.859773964001675e-05, "loss": 0.3311, "step": 1162 }, { "epoch": 0.852014652014652, "grad_norm": 1.8515625, "learning_rate": 4.86395981582252e-05, "loss": 0.3366, "step": 1163 }, { "epoch": 0.8527472527472527, "grad_norm": 1.328125, "learning_rate": 4.868145667643366e-05, "loss": 0.3226, "step": 1164 }, { "epoch": 0.8534798534798534, "grad_norm": 1.1484375, "learning_rate": 4.872331519464211e-05, "loss": 0.27, "step": 1165 }, { "epoch": 0.8542124542124542, "grad_norm": 1.296875, "learning_rate": 4.8765173712850567e-05, "loss": 0.2622, "step": 1166 }, { "epoch": 0.8549450549450549, "grad_norm": 1.21875, "learning_rate": 4.8807032231059024e-05, "loss": 0.2892, "step": 1167 }, { "epoch": 0.8556776556776556, "grad_norm": 1.0703125, "learning_rate": 4.884889074926748e-05, "loss": 0.3038, "step": 1168 }, { "epoch": 0.8564102564102564, "grad_norm": 1.0546875, "learning_rate": 4.889074926747593e-05, "loss": 0.3347, "step": 1169 }, { "epoch": 0.8571428571428571, "grad_norm": 1.625, "learning_rate": 4.893260778568439e-05, "loss": 0.3224, "step": 1170 }, { "epoch": 0.8578754578754578, "grad_norm": 1.75, "learning_rate": 4.897446630389285e-05, "loss": 0.3128, "step": 1171 }, { "epoch": 0.8586080586080586, "grad_norm": 1.3515625, "learning_rate": 4.90163248221013e-05, "loss": 0.3684, "step": 1172 }, { "epoch": 0.8593406593406593, "grad_norm": 1.0, "learning_rate": 4.9058183340309755e-05, "loss": 0.2635, "step": 1173 }, { "epoch": 0.8600732600732601, "grad_norm": 0.9296875, "learning_rate": 4.9100041858518205e-05, "loss": 0.3119, "step": 1174 }, { "epoch": 0.8608058608058609, "grad_norm": 1.1328125, "learning_rate": 4.914190037672667e-05, "loss": 0.3329, "step": 1175 }, { "epoch": 0.8615384615384616, "grad_norm": 1.2265625, "learning_rate": 4.918375889493512e-05, "loss": 0.2934, "step": 1176 }, { "epoch": 0.8622710622710623, "grad_norm": 1.0234375, "learning_rate": 4.922561741314358e-05, "loss": 0.2823, "step": 1177 }, { "epoch": 0.863003663003663, "grad_norm": 1.21875, "learning_rate": 4.926747593135203e-05, "loss": 0.3052, "step": 1178 }, { "epoch": 0.8637362637362638, "grad_norm": 1.125, "learning_rate": 4.930933444956049e-05, "loss": 0.324, "step": 1179 }, { "epoch": 0.8644688644688645, "grad_norm": 1.3203125, "learning_rate": 4.935119296776894e-05, "loss": 0.32, "step": 1180 }, { "epoch": 0.8652014652014652, "grad_norm": 1.15625, "learning_rate": 4.93930514859774e-05, "loss": 0.2719, "step": 1181 }, { "epoch": 0.865934065934066, "grad_norm": 1.1640625, "learning_rate": 4.943491000418585e-05, "loss": 0.4156, "step": 1182 }, { "epoch": 0.8666666666666667, "grad_norm": 1.2734375, "learning_rate": 4.9476768522394315e-05, "loss": 0.2729, "step": 1183 }, { "epoch": 0.8673992673992674, "grad_norm": 0.95703125, "learning_rate": 4.9518627040602766e-05, "loss": 0.3141, "step": 1184 }, { "epoch": 0.8681318681318682, "grad_norm": 0.87109375, "learning_rate": 4.956048555881122e-05, "loss": 0.2737, "step": 1185 }, { "epoch": 0.8688644688644689, "grad_norm": 0.91015625, "learning_rate": 4.9602344077019674e-05, "loss": 0.2814, "step": 1186 }, { "epoch": 0.8695970695970696, "grad_norm": 1.0546875, "learning_rate": 4.964420259522813e-05, "loss": 0.3203, "step": 1187 }, { "epoch": 0.8703296703296703, "grad_norm": 1.171875, "learning_rate": 4.968606111343659e-05, "loss": 0.2805, "step": 1188 }, { "epoch": 0.8710622710622711, "grad_norm": 1.125, "learning_rate": 4.972791963164504e-05, "loss": 0.2989, "step": 1189 }, { "epoch": 0.8717948717948718, "grad_norm": 1.2109375, "learning_rate": 4.97697781498535e-05, "loss": 0.309, "step": 1190 }, { "epoch": 0.8725274725274725, "grad_norm": 1.5234375, "learning_rate": 4.981163666806195e-05, "loss": 0.3105, "step": 1191 }, { "epoch": 0.8732600732600733, "grad_norm": 1.28125, "learning_rate": 4.985349518627041e-05, "loss": 0.2981, "step": 1192 }, { "epoch": 0.873992673992674, "grad_norm": 1.8828125, "learning_rate": 4.989535370447886e-05, "loss": 0.4362, "step": 1193 }, { "epoch": 0.8747252747252747, "grad_norm": 1.0234375, "learning_rate": 4.993721222268732e-05, "loss": 0.2559, "step": 1194 }, { "epoch": 0.8754578754578755, "grad_norm": 1.546875, "learning_rate": 4.997907074089577e-05, "loss": 0.3612, "step": 1195 }, { "epoch": 0.8761904761904762, "grad_norm": 1.609375, "learning_rate": 5.0020929259104234e-05, "loss": 0.3086, "step": 1196 }, { "epoch": 0.8769230769230769, "grad_norm": 1.046875, "learning_rate": 5.006278777731268e-05, "loss": 0.2936, "step": 1197 }, { "epoch": 0.8776556776556776, "grad_norm": 0.9375, "learning_rate": 5.010464629552114e-05, "loss": 0.3253, "step": 1198 }, { "epoch": 0.8783882783882784, "grad_norm": 1.34375, "learning_rate": 5.01465048137296e-05, "loss": 0.3243, "step": 1199 }, { "epoch": 0.8791208791208791, "grad_norm": 1.0546875, "learning_rate": 5.018836333193805e-05, "loss": 0.2498, "step": 1200 }, { "epoch": 0.8798534798534798, "grad_norm": 1.375, "learning_rate": 5.023022185014651e-05, "loss": 0.3204, "step": 1201 }, { "epoch": 0.8805860805860806, "grad_norm": 1.34375, "learning_rate": 5.0272080368354965e-05, "loss": 0.3104, "step": 1202 }, { "epoch": 0.8813186813186813, "grad_norm": 1.078125, "learning_rate": 5.0313938886563416e-05, "loss": 0.3045, "step": 1203 }, { "epoch": 0.882051282051282, "grad_norm": 1.1484375, "learning_rate": 5.0355797404771873e-05, "loss": 0.3251, "step": 1204 }, { "epoch": 0.8827838827838828, "grad_norm": 1.546875, "learning_rate": 5.0397655922980324e-05, "loss": 0.3462, "step": 1205 }, { "epoch": 0.8835164835164835, "grad_norm": 1.125, "learning_rate": 5.043951444118878e-05, "loss": 0.2914, "step": 1206 }, { "epoch": 0.8842490842490842, "grad_norm": 1.1484375, "learning_rate": 5.0481372959397246e-05, "loss": 0.3039, "step": 1207 }, { "epoch": 0.884981684981685, "grad_norm": 0.99609375, "learning_rate": 5.052323147760569e-05, "loss": 0.317, "step": 1208 }, { "epoch": 0.8857142857142857, "grad_norm": 1.234375, "learning_rate": 5.0565089995814154e-05, "loss": 0.3084, "step": 1209 }, { "epoch": 0.8864468864468864, "grad_norm": 1.359375, "learning_rate": 5.06069485140226e-05, "loss": 0.2759, "step": 1210 }, { "epoch": 0.8871794871794871, "grad_norm": 1.1953125, "learning_rate": 5.064880703223106e-05, "loss": 0.2727, "step": 1211 }, { "epoch": 0.8879120879120879, "grad_norm": 0.9609375, "learning_rate": 5.069066555043952e-05, "loss": 0.2619, "step": 1212 }, { "epoch": 0.8886446886446886, "grad_norm": 1.1875, "learning_rate": 5.073252406864797e-05, "loss": 0.37, "step": 1213 }, { "epoch": 0.8893772893772893, "grad_norm": 1.203125, "learning_rate": 5.077438258685643e-05, "loss": 0.2924, "step": 1214 }, { "epoch": 0.8901098901098901, "grad_norm": 1.0390625, "learning_rate": 5.0816241105064885e-05, "loss": 0.2713, "step": 1215 }, { "epoch": 0.8908424908424909, "grad_norm": 1.640625, "learning_rate": 5.0858099623273335e-05, "loss": 0.2774, "step": 1216 }, { "epoch": 0.8915750915750916, "grad_norm": 1.1953125, "learning_rate": 5.08999581414818e-05, "loss": 0.3266, "step": 1217 }, { "epoch": 0.8923076923076924, "grad_norm": 0.97265625, "learning_rate": 5.094181665969024e-05, "loss": 0.2748, "step": 1218 }, { "epoch": 0.8930402930402931, "grad_norm": 1.671875, "learning_rate": 5.098367517789871e-05, "loss": 0.2953, "step": 1219 }, { "epoch": 0.8937728937728938, "grad_norm": 1.4765625, "learning_rate": 5.1025533696107165e-05, "loss": 0.3512, "step": 1220 }, { "epoch": 0.8945054945054945, "grad_norm": 1.390625, "learning_rate": 5.1067392214315615e-05, "loss": 0.3335, "step": 1221 }, { "epoch": 0.8952380952380953, "grad_norm": 1.140625, "learning_rate": 5.110925073252407e-05, "loss": 0.2802, "step": 1222 }, { "epoch": 0.895970695970696, "grad_norm": 1.7734375, "learning_rate": 5.115110925073253e-05, "loss": 0.3691, "step": 1223 }, { "epoch": 0.8967032967032967, "grad_norm": 1.125, "learning_rate": 5.119296776894098e-05, "loss": 0.2746, "step": 1224 }, { "epoch": 0.8974358974358975, "grad_norm": 1.265625, "learning_rate": 5.123482628714944e-05, "loss": 0.2709, "step": 1225 }, { "epoch": 0.8981684981684982, "grad_norm": 0.96484375, "learning_rate": 5.127668480535789e-05, "loss": 0.253, "step": 1226 }, { "epoch": 0.8989010989010989, "grad_norm": 1.34375, "learning_rate": 5.1318543323566346e-05, "loss": 0.3522, "step": 1227 }, { "epoch": 0.8996336996336997, "grad_norm": 1.375, "learning_rate": 5.136040184177481e-05, "loss": 0.3029, "step": 1228 }, { "epoch": 0.9003663003663004, "grad_norm": 1.3984375, "learning_rate": 5.1402260359983254e-05, "loss": 0.3352, "step": 1229 }, { "epoch": 0.9010989010989011, "grad_norm": 1.0078125, "learning_rate": 5.144411887819172e-05, "loss": 0.2644, "step": 1230 }, { "epoch": 0.9018315018315018, "grad_norm": 1.2734375, "learning_rate": 5.148597739640016e-05, "loss": 0.2963, "step": 1231 }, { "epoch": 0.9025641025641026, "grad_norm": 0.9140625, "learning_rate": 5.1527835914608627e-05, "loss": 0.2258, "step": 1232 }, { "epoch": 0.9032967032967033, "grad_norm": 1.109375, "learning_rate": 5.1569694432817084e-05, "loss": 0.2981, "step": 1233 }, { "epoch": 0.904029304029304, "grad_norm": 1.3671875, "learning_rate": 5.1611552951025535e-05, "loss": 0.2595, "step": 1234 }, { "epoch": 0.9047619047619048, "grad_norm": 1.25, "learning_rate": 5.165341146923399e-05, "loss": 0.3003, "step": 1235 }, { "epoch": 0.9054945054945055, "grad_norm": 1.0078125, "learning_rate": 5.169526998744245e-05, "loss": 0.3203, "step": 1236 }, { "epoch": 0.9062271062271062, "grad_norm": 1.171875, "learning_rate": 5.17371285056509e-05, "loss": 0.2614, "step": 1237 }, { "epoch": 0.906959706959707, "grad_norm": 1.359375, "learning_rate": 5.177898702385936e-05, "loss": 0.2788, "step": 1238 }, { "epoch": 0.9076923076923077, "grad_norm": 1.5234375, "learning_rate": 5.182084554206781e-05, "loss": 0.3135, "step": 1239 }, { "epoch": 0.9084249084249084, "grad_norm": 1.171875, "learning_rate": 5.1862704060276265e-05, "loss": 0.262, "step": 1240 }, { "epoch": 0.9091575091575091, "grad_norm": 1.1328125, "learning_rate": 5.190456257848473e-05, "loss": 0.294, "step": 1241 }, { "epoch": 0.9098901098901099, "grad_norm": 1.109375, "learning_rate": 5.1946421096693173e-05, "loss": 0.3608, "step": 1242 }, { "epoch": 0.9106227106227106, "grad_norm": 1.2578125, "learning_rate": 5.198827961490164e-05, "loss": 0.3086, "step": 1243 }, { "epoch": 0.9113553113553113, "grad_norm": 1.296875, "learning_rate": 5.2030138133110095e-05, "loss": 0.312, "step": 1244 }, { "epoch": 0.9120879120879121, "grad_norm": 1.0703125, "learning_rate": 5.2071996651318546e-05, "loss": 0.3027, "step": 1245 }, { "epoch": 0.9128205128205128, "grad_norm": 1.2421875, "learning_rate": 5.2113855169527e-05, "loss": 0.2924, "step": 1246 }, { "epoch": 0.9135531135531135, "grad_norm": 1.3515625, "learning_rate": 5.2155713687735454e-05, "loss": 0.3149, "step": 1247 }, { "epoch": 0.9142857142857143, "grad_norm": 1.1875, "learning_rate": 5.219757220594391e-05, "loss": 0.2703, "step": 1248 }, { "epoch": 0.915018315018315, "grad_norm": 1.6953125, "learning_rate": 5.223943072415237e-05, "loss": 0.2919, "step": 1249 }, { "epoch": 0.9157509157509157, "grad_norm": 0.99609375, "learning_rate": 5.228128924236082e-05, "loss": 0.2699, "step": 1250 }, { "epoch": 0.9164835164835164, "grad_norm": 1.109375, "learning_rate": 5.2323147760569277e-05, "loss": 0.3621, "step": 1251 }, { "epoch": 0.9172161172161172, "grad_norm": 0.859375, "learning_rate": 5.236500627877773e-05, "loss": 0.2806, "step": 1252 }, { "epoch": 0.9179487179487179, "grad_norm": 1.1640625, "learning_rate": 5.240686479698619e-05, "loss": 0.3275, "step": 1253 }, { "epoch": 0.9186813186813186, "grad_norm": 1.15625, "learning_rate": 5.244872331519465e-05, "loss": 0.3004, "step": 1254 }, { "epoch": 0.9194139194139194, "grad_norm": 1.46875, "learning_rate": 5.24905818334031e-05, "loss": 0.3139, "step": 1255 }, { "epoch": 0.9201465201465201, "grad_norm": 1.0, "learning_rate": 5.253244035161156e-05, "loss": 0.2941, "step": 1256 }, { "epoch": 0.9208791208791208, "grad_norm": 0.95703125, "learning_rate": 5.2574298869820014e-05, "loss": 0.2574, "step": 1257 }, { "epoch": 0.9216117216117216, "grad_norm": 1.7109375, "learning_rate": 5.2616157388028465e-05, "loss": 0.2875, "step": 1258 }, { "epoch": 0.9223443223443224, "grad_norm": 1.0390625, "learning_rate": 5.265801590623692e-05, "loss": 0.247, "step": 1259 }, { "epoch": 0.9230769230769231, "grad_norm": 1.390625, "learning_rate": 5.269987442444537e-05, "loss": 0.2798, "step": 1260 }, { "epoch": 0.9238095238095239, "grad_norm": 1.53125, "learning_rate": 5.274173294265383e-05, "loss": 0.3415, "step": 1261 }, { "epoch": 0.9245421245421246, "grad_norm": 1.421875, "learning_rate": 5.2783591460862294e-05, "loss": 0.3115, "step": 1262 }, { "epoch": 0.9252747252747253, "grad_norm": 0.921875, "learning_rate": 5.282544997907074e-05, "loss": 0.295, "step": 1263 }, { "epoch": 0.926007326007326, "grad_norm": 1.0546875, "learning_rate": 5.28673084972792e-05, "loss": 0.2606, "step": 1264 }, { "epoch": 0.9267399267399268, "grad_norm": 1.9140625, "learning_rate": 5.290916701548766e-05, "loss": 0.2461, "step": 1265 }, { "epoch": 0.9274725274725275, "grad_norm": 2.125, "learning_rate": 5.295102553369611e-05, "loss": 0.354, "step": 1266 }, { "epoch": 0.9282051282051282, "grad_norm": 1.765625, "learning_rate": 5.299288405190457e-05, "loss": 0.3229, "step": 1267 }, { "epoch": 0.928937728937729, "grad_norm": 1.640625, "learning_rate": 5.303474257011302e-05, "loss": 0.3645, "step": 1268 }, { "epoch": 0.9296703296703297, "grad_norm": 1.4296875, "learning_rate": 5.3076601088321476e-05, "loss": 0.3215, "step": 1269 }, { "epoch": 0.9304029304029304, "grad_norm": 0.921875, "learning_rate": 5.3118459606529933e-05, "loss": 0.2819, "step": 1270 }, { "epoch": 0.9311355311355312, "grad_norm": 1.15625, "learning_rate": 5.3160318124738384e-05, "loss": 0.2763, "step": 1271 }, { "epoch": 0.9318681318681319, "grad_norm": 0.95703125, "learning_rate": 5.320217664294684e-05, "loss": 0.2733, "step": 1272 }, { "epoch": 0.9326007326007326, "grad_norm": 1.328125, "learning_rate": 5.324403516115529e-05, "loss": 0.2657, "step": 1273 }, { "epoch": 0.9333333333333333, "grad_norm": 0.98046875, "learning_rate": 5.328589367936375e-05, "loss": 0.2681, "step": 1274 }, { "epoch": 0.9340659340659341, "grad_norm": 0.9765625, "learning_rate": 5.3327752197572214e-05, "loss": 0.2915, "step": 1275 }, { "epoch": 0.9347985347985348, "grad_norm": 1.078125, "learning_rate": 5.336961071578066e-05, "loss": 0.2808, "step": 1276 }, { "epoch": 0.9355311355311355, "grad_norm": 1.3203125, "learning_rate": 5.341146923398912e-05, "loss": 0.3691, "step": 1277 }, { "epoch": 0.9362637362637363, "grad_norm": 1.234375, "learning_rate": 5.345332775219758e-05, "loss": 0.2896, "step": 1278 }, { "epoch": 0.936996336996337, "grad_norm": 1.1875, "learning_rate": 5.349518627040603e-05, "loss": 0.2784, "step": 1279 }, { "epoch": 0.9377289377289377, "grad_norm": 1.3359375, "learning_rate": 5.353704478861449e-05, "loss": 0.3655, "step": 1280 }, { "epoch": 0.9384615384615385, "grad_norm": 1.375, "learning_rate": 5.357890330682294e-05, "loss": 0.3347, "step": 1281 }, { "epoch": 0.9391941391941392, "grad_norm": 1.0, "learning_rate": 5.3620761825031395e-05, "loss": 0.3195, "step": 1282 }, { "epoch": 0.9399267399267399, "grad_norm": 1.1796875, "learning_rate": 5.366262034323985e-05, "loss": 0.3038, "step": 1283 }, { "epoch": 0.9406593406593406, "grad_norm": 1.09375, "learning_rate": 5.37044788614483e-05, "loss": 0.3043, "step": 1284 }, { "epoch": 0.9413919413919414, "grad_norm": 0.984375, "learning_rate": 5.374633737965676e-05, "loss": 0.2988, "step": 1285 }, { "epoch": 0.9421245421245421, "grad_norm": 1.1328125, "learning_rate": 5.3788195897865225e-05, "loss": 0.2898, "step": 1286 }, { "epoch": 0.9428571428571428, "grad_norm": 1.109375, "learning_rate": 5.383005441607367e-05, "loss": 0.2473, "step": 1287 }, { "epoch": 0.9435897435897436, "grad_norm": 1.0625, "learning_rate": 5.387191293428213e-05, "loss": 0.2857, "step": 1288 }, { "epoch": 0.9443223443223443, "grad_norm": 1.3359375, "learning_rate": 5.3913771452490583e-05, "loss": 0.3836, "step": 1289 }, { "epoch": 0.945054945054945, "grad_norm": 1.265625, "learning_rate": 5.395562997069904e-05, "loss": 0.2896, "step": 1290 }, { "epoch": 0.9457875457875458, "grad_norm": 1.0078125, "learning_rate": 5.39974884889075e-05, "loss": 0.2711, "step": 1291 }, { "epoch": 0.9465201465201465, "grad_norm": 1.3671875, "learning_rate": 5.403934700711595e-05, "loss": 0.2667, "step": 1292 }, { "epoch": 0.9472527472527472, "grad_norm": 0.9375, "learning_rate": 5.4081205525324406e-05, "loss": 0.2703, "step": 1293 }, { "epoch": 0.947985347985348, "grad_norm": 1.2890625, "learning_rate": 5.412306404353286e-05, "loss": 0.255, "step": 1294 }, { "epoch": 0.9487179487179487, "grad_norm": 1.4453125, "learning_rate": 5.4164922561741314e-05, "loss": 0.2799, "step": 1295 }, { "epoch": 0.9494505494505494, "grad_norm": 1.2109375, "learning_rate": 5.420678107994978e-05, "loss": 0.2602, "step": 1296 }, { "epoch": 0.9501831501831501, "grad_norm": 1.515625, "learning_rate": 5.424863959815822e-05, "loss": 0.3357, "step": 1297 }, { "epoch": 0.9509157509157509, "grad_norm": 1.1328125, "learning_rate": 5.4290498116366687e-05, "loss": 0.3142, "step": 1298 }, { "epoch": 0.9516483516483516, "grad_norm": 1.25, "learning_rate": 5.4332356634575144e-05, "loss": 0.2865, "step": 1299 }, { "epoch": 0.9523809523809523, "grad_norm": 1.296875, "learning_rate": 5.4374215152783595e-05, "loss": 0.2352, "step": 1300 }, { "epoch": 0.9531135531135531, "grad_norm": 1.2421875, "learning_rate": 5.441607367099205e-05, "loss": 0.2807, "step": 1301 }, { "epoch": 0.9538461538461539, "grad_norm": 1.1484375, "learning_rate": 5.44579321892005e-05, "loss": 0.3495, "step": 1302 }, { "epoch": 0.9545787545787546, "grad_norm": 1.3984375, "learning_rate": 5.449979070740896e-05, "loss": 0.276, "step": 1303 }, { "epoch": 0.9553113553113554, "grad_norm": 1.359375, "learning_rate": 5.454164922561742e-05, "loss": 0.2772, "step": 1304 }, { "epoch": 0.9560439560439561, "grad_norm": 1.359375, "learning_rate": 5.458350774382587e-05, "loss": 0.2893, "step": 1305 }, { "epoch": 0.9567765567765568, "grad_norm": 1.0234375, "learning_rate": 5.4625366262034325e-05, "loss": 0.2829, "step": 1306 }, { "epoch": 0.9575091575091575, "grad_norm": 1.2578125, "learning_rate": 5.466722478024279e-05, "loss": 0.2758, "step": 1307 }, { "epoch": 0.9582417582417583, "grad_norm": 1.3203125, "learning_rate": 5.4709083298451233e-05, "loss": 0.2733, "step": 1308 }, { "epoch": 0.958974358974359, "grad_norm": 1.1640625, "learning_rate": 5.47509418166597e-05, "loss": 0.2833, "step": 1309 }, { "epoch": 0.9597069597069597, "grad_norm": 1.2109375, "learning_rate": 5.479280033486814e-05, "loss": 0.2634, "step": 1310 }, { "epoch": 0.9604395604395605, "grad_norm": 1.8671875, "learning_rate": 5.4834658853076606e-05, "loss": 0.3991, "step": 1311 }, { "epoch": 0.9611721611721612, "grad_norm": 1.1875, "learning_rate": 5.487651737128506e-05, "loss": 0.3042, "step": 1312 }, { "epoch": 0.9619047619047619, "grad_norm": 1.1875, "learning_rate": 5.4918375889493514e-05, "loss": 0.2495, "step": 1313 }, { "epoch": 0.9626373626373627, "grad_norm": 1.265625, "learning_rate": 5.496023440770197e-05, "loss": 0.2426, "step": 1314 }, { "epoch": 0.9633699633699634, "grad_norm": 1.359375, "learning_rate": 5.500209292591042e-05, "loss": 0.3404, "step": 1315 }, { "epoch": 0.9641025641025641, "grad_norm": 1.2578125, "learning_rate": 5.504395144411888e-05, "loss": 0.3094, "step": 1316 }, { "epoch": 0.9648351648351648, "grad_norm": 1.4765625, "learning_rate": 5.5085809962327337e-05, "loss": 0.2833, "step": 1317 }, { "epoch": 0.9655677655677656, "grad_norm": 1.2265625, "learning_rate": 5.512766848053579e-05, "loss": 0.3212, "step": 1318 }, { "epoch": 0.9663003663003663, "grad_norm": 1.1640625, "learning_rate": 5.5169526998744245e-05, "loss": 0.3082, "step": 1319 }, { "epoch": 0.967032967032967, "grad_norm": 1.578125, "learning_rate": 5.521138551695271e-05, "loss": 0.2578, "step": 1320 }, { "epoch": 0.9677655677655678, "grad_norm": 1.21875, "learning_rate": 5.525324403516115e-05, "loss": 0.3122, "step": 1321 }, { "epoch": 0.9684981684981685, "grad_norm": 0.92578125, "learning_rate": 5.529510255336962e-05, "loss": 0.2995, "step": 1322 }, { "epoch": 0.9692307692307692, "grad_norm": 1.046875, "learning_rate": 5.533696107157806e-05, "loss": 0.2577, "step": 1323 }, { "epoch": 0.96996336996337, "grad_norm": 1.046875, "learning_rate": 5.5378819589786525e-05, "loss": 0.2587, "step": 1324 }, { "epoch": 0.9706959706959707, "grad_norm": 1.4609375, "learning_rate": 5.542067810799498e-05, "loss": 0.2598, "step": 1325 }, { "epoch": 0.9714285714285714, "grad_norm": 1.140625, "learning_rate": 5.546253662620343e-05, "loss": 0.2698, "step": 1326 }, { "epoch": 0.9721611721611721, "grad_norm": 1.0, "learning_rate": 5.550439514441189e-05, "loss": 0.247, "step": 1327 }, { "epoch": 0.9728937728937729, "grad_norm": 0.98828125, "learning_rate": 5.5546253662620354e-05, "loss": 0.2785, "step": 1328 }, { "epoch": 0.9736263736263736, "grad_norm": 0.94140625, "learning_rate": 5.55881121808288e-05, "loss": 0.2608, "step": 1329 }, { "epoch": 0.9743589743589743, "grad_norm": 1.609375, "learning_rate": 5.562997069903726e-05, "loss": 0.2847, "step": 1330 }, { "epoch": 0.9750915750915751, "grad_norm": 1.1796875, "learning_rate": 5.5671829217245706e-05, "loss": 0.3204, "step": 1331 }, { "epoch": 0.9758241758241758, "grad_norm": 0.9609375, "learning_rate": 5.571368773545417e-05, "loss": 0.2767, "step": 1332 }, { "epoch": 0.9765567765567765, "grad_norm": 1.0390625, "learning_rate": 5.575554625366263e-05, "loss": 0.2684, "step": 1333 }, { "epoch": 0.9772893772893773, "grad_norm": 1.015625, "learning_rate": 5.579740477187108e-05, "loss": 0.3011, "step": 1334 }, { "epoch": 0.978021978021978, "grad_norm": 0.99609375, "learning_rate": 5.5839263290079536e-05, "loss": 0.3241, "step": 1335 }, { "epoch": 0.9787545787545787, "grad_norm": 1.328125, "learning_rate": 5.588112180828799e-05, "loss": 0.215, "step": 1336 }, { "epoch": 0.9794871794871794, "grad_norm": 1.5078125, "learning_rate": 5.5922980326496444e-05, "loss": 0.3049, "step": 1337 }, { "epoch": 0.9802197802197802, "grad_norm": 0.94921875, "learning_rate": 5.59648388447049e-05, "loss": 0.2684, "step": 1338 }, { "epoch": 0.9809523809523809, "grad_norm": 1.0625, "learning_rate": 5.600669736291335e-05, "loss": 0.2909, "step": 1339 }, { "epoch": 0.9816849816849816, "grad_norm": 1.1640625, "learning_rate": 5.604855588112181e-05, "loss": 0.3063, "step": 1340 }, { "epoch": 0.9824175824175824, "grad_norm": 1.6875, "learning_rate": 5.6090414399330274e-05, "loss": 0.2212, "step": 1341 }, { "epoch": 0.9831501831501831, "grad_norm": 1.1640625, "learning_rate": 5.613227291753872e-05, "loss": 0.2712, "step": 1342 }, { "epoch": 0.9838827838827838, "grad_norm": 1.6953125, "learning_rate": 5.617413143574718e-05, "loss": 0.2607, "step": 1343 }, { "epoch": 0.9846153846153847, "grad_norm": 1.1953125, "learning_rate": 5.6215989953955626e-05, "loss": 0.2469, "step": 1344 }, { "epoch": 0.9853479853479854, "grad_norm": 0.88671875, "learning_rate": 5.625784847216409e-05, "loss": 0.2486, "step": 1345 }, { "epoch": 0.9860805860805861, "grad_norm": 1.0390625, "learning_rate": 5.629970699037255e-05, "loss": 0.2459, "step": 1346 }, { "epoch": 0.9868131868131869, "grad_norm": 1.296875, "learning_rate": 5.6341565508581e-05, "loss": 0.2509, "step": 1347 }, { "epoch": 0.9875457875457876, "grad_norm": 2.421875, "learning_rate": 5.6383424026789455e-05, "loss": 0.3207, "step": 1348 }, { "epoch": 0.9882783882783883, "grad_norm": 1.953125, "learning_rate": 5.642528254499791e-05, "loss": 0.2985, "step": 1349 }, { "epoch": 0.989010989010989, "grad_norm": 1.2421875, "learning_rate": 5.646714106320636e-05, "loss": 0.2613, "step": 1350 }, { "epoch": 0.9897435897435898, "grad_norm": 2.03125, "learning_rate": 5.650899958141482e-05, "loss": 0.2245, "step": 1351 }, { "epoch": 0.9904761904761905, "grad_norm": 2.390625, "learning_rate": 5.655085809962327e-05, "loss": 0.3246, "step": 1352 }, { "epoch": 0.9912087912087912, "grad_norm": 2.3125, "learning_rate": 5.659271661783173e-05, "loss": 0.2523, "step": 1353 }, { "epoch": 0.991941391941392, "grad_norm": 2.359375, "learning_rate": 5.663457513604019e-05, "loss": 0.2965, "step": 1354 }, { "epoch": 0.9926739926739927, "grad_norm": 1.1328125, "learning_rate": 5.667643365424864e-05, "loss": 0.208, "step": 1355 }, { "epoch": 0.9934065934065934, "grad_norm": 1.375, "learning_rate": 5.67182921724571e-05, "loss": 0.2645, "step": 1356 }, { "epoch": 0.9941391941391942, "grad_norm": 1.9375, "learning_rate": 5.6760150690665545e-05, "loss": 0.261, "step": 1357 }, { "epoch": 0.9948717948717949, "grad_norm": 2.09375, "learning_rate": 5.680200920887401e-05, "loss": 0.3273, "step": 1358 }, { "epoch": 0.9956043956043956, "grad_norm": 1.3984375, "learning_rate": 5.6843867727082466e-05, "loss": 0.3222, "step": 1359 }, { "epoch": 0.9963369963369964, "grad_norm": 1.9453125, "learning_rate": 5.688572624529092e-05, "loss": 0.3205, "step": 1360 }, { "epoch": 0.9970695970695971, "grad_norm": 0.765625, "learning_rate": 5.6927584763499374e-05, "loss": 0.2091, "step": 1361 }, { "epoch": 0.9978021978021978, "grad_norm": 1.03125, "learning_rate": 5.696944328170783e-05, "loss": 0.2164, "step": 1362 }, { "epoch": 0.9985347985347985, "grad_norm": 1.125, "learning_rate": 5.701130179991628e-05, "loss": 0.2921, "step": 1363 }, { "epoch": 0.9992673992673993, "grad_norm": 1.53125, "learning_rate": 5.705316031812474e-05, "loss": 0.2719, "step": 1364 }, { "epoch": 1.0, "grad_norm": 1.15625, "learning_rate": 5.709501883633319e-05, "loss": 0.268, "step": 1365 }, { "epoch": 1.0007326007326007, "grad_norm": 1.3046875, "learning_rate": 5.7136877354541655e-05, "loss": 0.2627, "step": 1366 }, { "epoch": 1.0014652014652015, "grad_norm": 1.09375, "learning_rate": 5.717873587275011e-05, "loss": 0.2999, "step": 1367 }, { "epoch": 1.0021978021978022, "grad_norm": 0.9453125, "learning_rate": 5.722059439095856e-05, "loss": 0.257, "step": 1368 }, { "epoch": 1.002930402930403, "grad_norm": 1.5859375, "learning_rate": 5.726245290916702e-05, "loss": 0.237, "step": 1369 }, { "epoch": 1.0036630036630036, "grad_norm": 1.5390625, "learning_rate": 5.730431142737548e-05, "loss": 0.3161, "step": 1370 }, { "epoch": 1.0043956043956044, "grad_norm": 1.9453125, "learning_rate": 5.734616994558393e-05, "loss": 0.3146, "step": 1371 }, { "epoch": 1.005128205128205, "grad_norm": 0.83984375, "learning_rate": 5.7388028463792385e-05, "loss": 0.2534, "step": 1372 }, { "epoch": 1.0058608058608058, "grad_norm": 0.9375, "learning_rate": 5.7429886982000836e-05, "loss": 0.219, "step": 1373 }, { "epoch": 1.0065934065934066, "grad_norm": 1.1640625, "learning_rate": 5.7471745500209293e-05, "loss": 0.2665, "step": 1374 }, { "epoch": 1.0073260073260073, "grad_norm": 1.1328125, "learning_rate": 5.751360401841776e-05, "loss": 0.2869, "step": 1375 }, { "epoch": 1.008058608058608, "grad_norm": 1.171875, "learning_rate": 5.75554625366262e-05, "loss": 0.3022, "step": 1376 }, { "epoch": 1.0087912087912088, "grad_norm": 1.1328125, "learning_rate": 5.7597321054834666e-05, "loss": 0.3131, "step": 1377 }, { "epoch": 1.0095238095238095, "grad_norm": 0.89453125, "learning_rate": 5.763917957304311e-05, "loss": 0.2249, "step": 1378 }, { "epoch": 1.0102564102564102, "grad_norm": 1.0546875, "learning_rate": 5.7681038091251574e-05, "loss": 0.2697, "step": 1379 }, { "epoch": 1.010989010989011, "grad_norm": 1.0234375, "learning_rate": 5.772289660946003e-05, "loss": 0.2198, "step": 1380 }, { "epoch": 1.0117216117216117, "grad_norm": 1.2578125, "learning_rate": 5.776475512766848e-05, "loss": 0.2757, "step": 1381 }, { "epoch": 1.0124542124542124, "grad_norm": 1.0546875, "learning_rate": 5.780661364587694e-05, "loss": 0.2804, "step": 1382 }, { "epoch": 1.0131868131868131, "grad_norm": 1.453125, "learning_rate": 5.7848472164085397e-05, "loss": 0.2498, "step": 1383 }, { "epoch": 1.0139194139194139, "grad_norm": 1.140625, "learning_rate": 5.789033068229385e-05, "loss": 0.2456, "step": 1384 }, { "epoch": 1.0146520146520146, "grad_norm": 1.1640625, "learning_rate": 5.7932189200502305e-05, "loss": 0.2811, "step": 1385 }, { "epoch": 1.0153846153846153, "grad_norm": 1.1796875, "learning_rate": 5.7974047718710755e-05, "loss": 0.2649, "step": 1386 }, { "epoch": 1.016117216117216, "grad_norm": 1.2890625, "learning_rate": 5.801590623691921e-05, "loss": 0.3077, "step": 1387 }, { "epoch": 1.0168498168498168, "grad_norm": 1.21875, "learning_rate": 5.805776475512768e-05, "loss": 0.2034, "step": 1388 }, { "epoch": 1.0175824175824175, "grad_norm": 1.203125, "learning_rate": 5.809962327333612e-05, "loss": 0.2564, "step": 1389 }, { "epoch": 1.0183150183150182, "grad_norm": 0.9453125, "learning_rate": 5.8141481791544585e-05, "loss": 0.2407, "step": 1390 }, { "epoch": 1.019047619047619, "grad_norm": 1.2890625, "learning_rate": 5.818334030975304e-05, "loss": 0.2815, "step": 1391 }, { "epoch": 1.0197802197802197, "grad_norm": 0.90625, "learning_rate": 5.822519882796149e-05, "loss": 0.287, "step": 1392 }, { "epoch": 1.0205128205128204, "grad_norm": 1.0625, "learning_rate": 5.826705734616995e-05, "loss": 0.2619, "step": 1393 }, { "epoch": 1.0212454212454212, "grad_norm": 1.46875, "learning_rate": 5.83089158643784e-05, "loss": 0.2839, "step": 1394 }, { "epoch": 1.021978021978022, "grad_norm": 1.3359375, "learning_rate": 5.835077438258686e-05, "loss": 0.3079, "step": 1395 }, { "epoch": 1.0227106227106226, "grad_norm": 1.1953125, "learning_rate": 5.8392632900795316e-05, "loss": 0.2528, "step": 1396 }, { "epoch": 1.0234432234432234, "grad_norm": 1.109375, "learning_rate": 5.8434491419003766e-05, "loss": 0.3289, "step": 1397 }, { "epoch": 1.024175824175824, "grad_norm": 1.5859375, "learning_rate": 5.8476349937212224e-05, "loss": 0.2279, "step": 1398 }, { "epoch": 1.0249084249084248, "grad_norm": 1.8828125, "learning_rate": 5.8518208455420674e-05, "loss": 0.2962, "step": 1399 }, { "epoch": 1.0256410256410255, "grad_norm": 1.453125, "learning_rate": 5.856006697362913e-05, "loss": 0.3385, "step": 1400 }, { "epoch": 1.0263736263736263, "grad_norm": 1.375, "learning_rate": 5.8601925491837596e-05, "loss": 0.254, "step": 1401 }, { "epoch": 1.027106227106227, "grad_norm": 0.96875, "learning_rate": 5.864378401004605e-05, "loss": 0.2191, "step": 1402 }, { "epoch": 1.0278388278388277, "grad_norm": 1.1015625, "learning_rate": 5.8685642528254504e-05, "loss": 0.2799, "step": 1403 }, { "epoch": 1.0285714285714285, "grad_norm": 1.6796875, "learning_rate": 5.872750104646296e-05, "loss": 0.2532, "step": 1404 }, { "epoch": 1.0293040293040292, "grad_norm": 1.3125, "learning_rate": 5.876935956467141e-05, "loss": 0.2875, "step": 1405 }, { "epoch": 1.03003663003663, "grad_norm": 1.5703125, "learning_rate": 5.881121808287987e-05, "loss": 0.2826, "step": 1406 }, { "epoch": 1.0307692307692307, "grad_norm": 1.3125, "learning_rate": 5.885307660108832e-05, "loss": 0.2606, "step": 1407 }, { "epoch": 1.0315018315018314, "grad_norm": 0.9765625, "learning_rate": 5.889493511929678e-05, "loss": 0.2333, "step": 1408 }, { "epoch": 1.0322344322344323, "grad_norm": 1.03125, "learning_rate": 5.893679363750524e-05, "loss": 0.2572, "step": 1409 }, { "epoch": 1.032967032967033, "grad_norm": 1.09375, "learning_rate": 5.8978652155713686e-05, "loss": 0.2816, "step": 1410 }, { "epoch": 1.0336996336996338, "grad_norm": 1.078125, "learning_rate": 5.902051067392215e-05, "loss": 0.2319, "step": 1411 }, { "epoch": 1.0344322344322345, "grad_norm": 1.5625, "learning_rate": 5.906236919213061e-05, "loss": 0.3314, "step": 1412 }, { "epoch": 1.0351648351648353, "grad_norm": 1.46875, "learning_rate": 5.910422771033906e-05, "loss": 0.2581, "step": 1413 }, { "epoch": 1.035897435897436, "grad_norm": 1.453125, "learning_rate": 5.9146086228547515e-05, "loss": 0.2559, "step": 1414 }, { "epoch": 1.0366300366300367, "grad_norm": 1.359375, "learning_rate": 5.9187944746755966e-05, "loss": 0.2695, "step": 1415 }, { "epoch": 1.0373626373626375, "grad_norm": 1.9453125, "learning_rate": 5.922980326496442e-05, "loss": 0.282, "step": 1416 }, { "epoch": 1.0380952380952382, "grad_norm": 0.9296875, "learning_rate": 5.927166178317288e-05, "loss": 0.2143, "step": 1417 }, { "epoch": 1.038827838827839, "grad_norm": 1.390625, "learning_rate": 5.931352030138133e-05, "loss": 0.2763, "step": 1418 }, { "epoch": 1.0395604395604396, "grad_norm": 1.234375, "learning_rate": 5.935537881958979e-05, "loss": 0.3165, "step": 1419 }, { "epoch": 1.0402930402930404, "grad_norm": 1.265625, "learning_rate": 5.939723733779824e-05, "loss": 0.2742, "step": 1420 }, { "epoch": 1.041025641025641, "grad_norm": 1.953125, "learning_rate": 5.94390958560067e-05, "loss": 0.2451, "step": 1421 }, { "epoch": 1.0417582417582418, "grad_norm": 3.1875, "learning_rate": 5.948095437421516e-05, "loss": 0.3457, "step": 1422 }, { "epoch": 1.0424908424908426, "grad_norm": 2.015625, "learning_rate": 5.9522812892423605e-05, "loss": 0.3064, "step": 1423 }, { "epoch": 1.0432234432234433, "grad_norm": 1.7890625, "learning_rate": 5.956467141063207e-05, "loss": 0.2864, "step": 1424 }, { "epoch": 1.043956043956044, "grad_norm": 0.94921875, "learning_rate": 5.9606529928840526e-05, "loss": 0.246, "step": 1425 }, { "epoch": 1.0446886446886448, "grad_norm": 0.96875, "learning_rate": 5.964838844704898e-05, "loss": 0.2489, "step": 1426 }, { "epoch": 1.0454212454212455, "grad_norm": 1.53125, "learning_rate": 5.9690246965257434e-05, "loss": 0.295, "step": 1427 }, { "epoch": 1.0461538461538462, "grad_norm": 1.6171875, "learning_rate": 5.9732105483465885e-05, "loss": 0.3023, "step": 1428 }, { "epoch": 1.046886446886447, "grad_norm": 1.546875, "learning_rate": 5.977396400167434e-05, "loss": 0.2261, "step": 1429 }, { "epoch": 1.0476190476190477, "grad_norm": 0.984375, "learning_rate": 5.98158225198828e-05, "loss": 0.2493, "step": 1430 }, { "epoch": 1.0483516483516484, "grad_norm": 0.9296875, "learning_rate": 5.985768103809125e-05, "loss": 0.2881, "step": 1431 }, { "epoch": 1.0490842490842491, "grad_norm": 1.4453125, "learning_rate": 5.989953955629971e-05, "loss": 0.2781, "step": 1432 }, { "epoch": 1.0498168498168499, "grad_norm": 1.9375, "learning_rate": 5.994139807450817e-05, "loss": 0.2468, "step": 1433 }, { "epoch": 1.0505494505494506, "grad_norm": 1.890625, "learning_rate": 5.9983256592716616e-05, "loss": 0.3566, "step": 1434 }, { "epoch": 1.0512820512820513, "grad_norm": 1.4140625, "learning_rate": 6.002511511092508e-05, "loss": 0.3266, "step": 1435 }, { "epoch": 1.052014652014652, "grad_norm": 1.640625, "learning_rate": 6.0066973629133524e-05, "loss": 0.3054, "step": 1436 }, { "epoch": 1.0527472527472528, "grad_norm": 1.1328125, "learning_rate": 6.010883214734199e-05, "loss": 0.1943, "step": 1437 }, { "epoch": 1.0534798534798535, "grad_norm": 0.875, "learning_rate": 6.0150690665550445e-05, "loss": 0.2611, "step": 1438 }, { "epoch": 1.0542124542124542, "grad_norm": 0.87109375, "learning_rate": 6.0192549183758896e-05, "loss": 0.2305, "step": 1439 }, { "epoch": 1.054945054945055, "grad_norm": 0.89453125, "learning_rate": 6.0234407701967354e-05, "loss": 0.2388, "step": 1440 }, { "epoch": 1.0556776556776557, "grad_norm": 1.0390625, "learning_rate": 6.0276266220175804e-05, "loss": 0.2666, "step": 1441 }, { "epoch": 1.0564102564102564, "grad_norm": 1.1171875, "learning_rate": 6.031812473838426e-05, "loss": 0.2876, "step": 1442 }, { "epoch": 1.0571428571428572, "grad_norm": 1.0234375, "learning_rate": 6.0359983256592726e-05, "loss": 0.2498, "step": 1443 }, { "epoch": 1.057875457875458, "grad_norm": 2.015625, "learning_rate": 6.040184177480117e-05, "loss": 0.2874, "step": 1444 }, { "epoch": 1.0586080586080586, "grad_norm": 1.3203125, "learning_rate": 6.0443700293009634e-05, "loss": 0.2788, "step": 1445 }, { "epoch": 1.0593406593406594, "grad_norm": 1.2734375, "learning_rate": 6.048555881121809e-05, "loss": 0.24, "step": 1446 }, { "epoch": 1.06007326007326, "grad_norm": 0.83984375, "learning_rate": 6.052741732942654e-05, "loss": 0.21, "step": 1447 }, { "epoch": 1.0608058608058608, "grad_norm": 2.03125, "learning_rate": 6.0569275847635e-05, "loss": 0.3151, "step": 1448 }, { "epoch": 1.0615384615384615, "grad_norm": 1.546875, "learning_rate": 6.061113436584345e-05, "loss": 0.339, "step": 1449 }, { "epoch": 1.0622710622710623, "grad_norm": 1.4453125, "learning_rate": 6.065299288405191e-05, "loss": 0.2874, "step": 1450 }, { "epoch": 1.063003663003663, "grad_norm": 1.2734375, "learning_rate": 6.0694851402260365e-05, "loss": 0.2454, "step": 1451 }, { "epoch": 1.0637362637362637, "grad_norm": 1.046875, "learning_rate": 6.0736709920468815e-05, "loss": 0.258, "step": 1452 }, { "epoch": 1.0644688644688645, "grad_norm": 0.84375, "learning_rate": 6.077856843867727e-05, "loss": 0.2445, "step": 1453 }, { "epoch": 1.0652014652014652, "grad_norm": 0.9296875, "learning_rate": 6.082042695688574e-05, "loss": 0.2525, "step": 1454 }, { "epoch": 1.065934065934066, "grad_norm": 1.3828125, "learning_rate": 6.086228547509418e-05, "loss": 0.2875, "step": 1455 }, { "epoch": 1.0666666666666667, "grad_norm": 2.046875, "learning_rate": 6.0904143993302645e-05, "loss": 0.2747, "step": 1456 }, { "epoch": 1.0673992673992674, "grad_norm": 1.7265625, "learning_rate": 6.094600251151109e-05, "loss": 0.2676, "step": 1457 }, { "epoch": 1.0681318681318681, "grad_norm": 1.7265625, "learning_rate": 6.098786102971955e-05, "loss": 0.2972, "step": 1458 }, { "epoch": 1.0688644688644688, "grad_norm": 0.8984375, "learning_rate": 6.102971954792801e-05, "loss": 0.2392, "step": 1459 }, { "epoch": 1.0695970695970696, "grad_norm": 1.2734375, "learning_rate": 6.107157806613646e-05, "loss": 0.2591, "step": 1460 }, { "epoch": 1.0703296703296703, "grad_norm": 1.6640625, "learning_rate": 6.111343658434493e-05, "loss": 0.2225, "step": 1461 }, { "epoch": 1.071062271062271, "grad_norm": 1.71875, "learning_rate": 6.115529510255336e-05, "loss": 0.2532, "step": 1462 }, { "epoch": 1.0717948717948718, "grad_norm": 1.3984375, "learning_rate": 6.119715362076183e-05, "loss": 0.2803, "step": 1463 }, { "epoch": 1.0725274725274725, "grad_norm": 0.91796875, "learning_rate": 6.123901213897029e-05, "loss": 0.2308, "step": 1464 }, { "epoch": 1.0732600732600732, "grad_norm": 1.25, "learning_rate": 6.128087065717874e-05, "loss": 0.2511, "step": 1465 }, { "epoch": 1.073992673992674, "grad_norm": 1.5, "learning_rate": 6.132272917538719e-05, "loss": 0.2507, "step": 1466 }, { "epoch": 1.0747252747252747, "grad_norm": 1.0703125, "learning_rate": 6.136458769359566e-05, "loss": 0.2678, "step": 1467 }, { "epoch": 1.0754578754578754, "grad_norm": 0.9609375, "learning_rate": 6.14064462118041e-05, "loss": 0.2431, "step": 1468 }, { "epoch": 1.0761904761904761, "grad_norm": 0.96875, "learning_rate": 6.144830473001256e-05, "loss": 0.2945, "step": 1469 }, { "epoch": 1.0769230769230769, "grad_norm": 1.2734375, "learning_rate": 6.149016324822101e-05, "loss": 0.2894, "step": 1470 }, { "epoch": 1.0776556776556776, "grad_norm": 1.296875, "learning_rate": 6.153202176642947e-05, "loss": 0.3061, "step": 1471 }, { "epoch": 1.0783882783882783, "grad_norm": 0.94921875, "learning_rate": 6.157388028463794e-05, "loss": 0.2709, "step": 1472 }, { "epoch": 1.079120879120879, "grad_norm": 1.140625, "learning_rate": 6.161573880284637e-05, "loss": 0.2751, "step": 1473 }, { "epoch": 1.0798534798534798, "grad_norm": 1.1328125, "learning_rate": 6.165759732105484e-05, "loss": 0.2707, "step": 1474 }, { "epoch": 1.0805860805860805, "grad_norm": 1.5, "learning_rate": 6.16994558392633e-05, "loss": 0.2996, "step": 1475 }, { "epoch": 1.0813186813186813, "grad_norm": 1.7109375, "learning_rate": 6.174131435747175e-05, "loss": 0.2899, "step": 1476 }, { "epoch": 1.082051282051282, "grad_norm": 1.453125, "learning_rate": 6.17831728756802e-05, "loss": 0.3263, "step": 1477 }, { "epoch": 1.0827838827838827, "grad_norm": 1.015625, "learning_rate": 6.182503139388865e-05, "loss": 0.2284, "step": 1478 }, { "epoch": 1.0835164835164834, "grad_norm": 0.84765625, "learning_rate": 6.186688991209712e-05, "loss": 0.2329, "step": 1479 }, { "epoch": 1.0842490842490842, "grad_norm": 1.28125, "learning_rate": 6.190874843030557e-05, "loss": 0.3008, "step": 1480 }, { "epoch": 1.084981684981685, "grad_norm": 1.328125, "learning_rate": 6.195060694851402e-05, "loss": 0.2626, "step": 1481 }, { "epoch": 1.0857142857142856, "grad_norm": 0.87890625, "learning_rate": 6.199246546672248e-05, "loss": 0.2489, "step": 1482 }, { "epoch": 1.0864468864468864, "grad_norm": 1.1328125, "learning_rate": 6.203432398493093e-05, "loss": 0.2623, "step": 1483 }, { "epoch": 1.087179487179487, "grad_norm": 1.34375, "learning_rate": 6.207618250313938e-05, "loss": 0.2847, "step": 1484 }, { "epoch": 1.0879120879120878, "grad_norm": 1.390625, "learning_rate": 6.211804102134785e-05, "loss": 0.3003, "step": 1485 }, { "epoch": 1.0886446886446886, "grad_norm": 1.84375, "learning_rate": 6.21598995395563e-05, "loss": 0.2913, "step": 1486 }, { "epoch": 1.0893772893772893, "grad_norm": 1.9140625, "learning_rate": 6.220175805776476e-05, "loss": 0.2446, "step": 1487 }, { "epoch": 1.09010989010989, "grad_norm": 1.203125, "learning_rate": 6.224361657597321e-05, "loss": 0.2685, "step": 1488 }, { "epoch": 1.0908424908424907, "grad_norm": 1.4375, "learning_rate": 6.228547509418166e-05, "loss": 0.2356, "step": 1489 }, { "epoch": 1.0915750915750915, "grad_norm": 1.59375, "learning_rate": 6.232733361239013e-05, "loss": 0.2965, "step": 1490 }, { "epoch": 1.0923076923076924, "grad_norm": 1.9453125, "learning_rate": 6.236919213059858e-05, "loss": 0.2306, "step": 1491 }, { "epoch": 1.0930402930402932, "grad_norm": 1.265625, "learning_rate": 6.241105064880703e-05, "loss": 0.2254, "step": 1492 }, { "epoch": 1.0937728937728939, "grad_norm": 1.0390625, "learning_rate": 6.24529091670155e-05, "loss": 0.2943, "step": 1493 }, { "epoch": 1.0945054945054946, "grad_norm": 1.140625, "learning_rate": 6.249476768522394e-05, "loss": 0.2687, "step": 1494 }, { "epoch": 1.0952380952380953, "grad_norm": 1.546875, "learning_rate": 6.253662620343241e-05, "loss": 0.3096, "step": 1495 }, { "epoch": 1.095970695970696, "grad_norm": 1.5, "learning_rate": 6.257848472164086e-05, "loss": 0.2719, "step": 1496 }, { "epoch": 1.0967032967032968, "grad_norm": 1.0859375, "learning_rate": 6.262034323984931e-05, "loss": 0.2715, "step": 1497 }, { "epoch": 1.0974358974358975, "grad_norm": 1.171875, "learning_rate": 6.266220175805777e-05, "loss": 0.3447, "step": 1498 }, { "epoch": 1.0981684981684983, "grad_norm": 1.125, "learning_rate": 6.270406027626623e-05, "loss": 0.273, "step": 1499 }, { "epoch": 1.098901098901099, "grad_norm": 1.1328125, "learning_rate": 6.274591879447468e-05, "loss": 0.344, "step": 1500 }, { "epoch": 1.0996336996336997, "grad_norm": 1.296875, "learning_rate": 6.278777731268314e-05, "loss": 0.264, "step": 1501 }, { "epoch": 1.1003663003663005, "grad_norm": 1.4140625, "learning_rate": 6.282963583089159e-05, "loss": 0.2418, "step": 1502 }, { "epoch": 1.1010989010989012, "grad_norm": 1.359375, "learning_rate": 6.287149434910004e-05, "loss": 0.2155, "step": 1503 }, { "epoch": 1.101831501831502, "grad_norm": 1.2109375, "learning_rate": 6.291335286730849e-05, "loss": 0.345, "step": 1504 }, { "epoch": 1.1025641025641026, "grad_norm": 0.9140625, "learning_rate": 6.295521138551696e-05, "loss": 0.2466, "step": 1505 }, { "epoch": 1.1032967032967034, "grad_norm": 1.234375, "learning_rate": 6.299706990372542e-05, "loss": 0.2542, "step": 1506 }, { "epoch": 1.104029304029304, "grad_norm": 1.2265625, "learning_rate": 6.303892842193386e-05, "loss": 0.2353, "step": 1507 }, { "epoch": 1.1047619047619048, "grad_norm": 2.15625, "learning_rate": 6.308078694014232e-05, "loss": 0.2561, "step": 1508 }, { "epoch": 1.1054945054945056, "grad_norm": 0.921875, "learning_rate": 6.312264545835079e-05, "loss": 0.2579, "step": 1509 }, { "epoch": 1.1062271062271063, "grad_norm": 1.0625, "learning_rate": 6.316450397655924e-05, "loss": 0.2567, "step": 1510 }, { "epoch": 1.106959706959707, "grad_norm": 1.09375, "learning_rate": 6.320636249476769e-05, "loss": 0.271, "step": 1511 }, { "epoch": 1.1076923076923078, "grad_norm": 1.1875, "learning_rate": 6.324822101297614e-05, "loss": 0.2634, "step": 1512 }, { "epoch": 1.1084249084249085, "grad_norm": 1.8828125, "learning_rate": 6.32900795311846e-05, "loss": 0.297, "step": 1513 }, { "epoch": 1.1091575091575092, "grad_norm": 1.25, "learning_rate": 6.333193804939305e-05, "loss": 0.29, "step": 1514 }, { "epoch": 1.10989010989011, "grad_norm": 0.98828125, "learning_rate": 6.33737965676015e-05, "loss": 0.2511, "step": 1515 }, { "epoch": 1.1106227106227107, "grad_norm": 1.0, "learning_rate": 6.341565508580997e-05, "loss": 0.2648, "step": 1516 }, { "epoch": 1.1113553113553114, "grad_norm": 1.0625, "learning_rate": 6.345751360401843e-05, "loss": 0.2664, "step": 1517 }, { "epoch": 1.1120879120879121, "grad_norm": 1.4609375, "learning_rate": 6.349937212222687e-05, "loss": 0.2968, "step": 1518 }, { "epoch": 1.1128205128205129, "grad_norm": 1.5625, "learning_rate": 6.354123064043533e-05, "loss": 0.3463, "step": 1519 }, { "epoch": 1.1135531135531136, "grad_norm": 0.91796875, "learning_rate": 6.358308915864378e-05, "loss": 0.2643, "step": 1520 }, { "epoch": 1.1142857142857143, "grad_norm": 1.171875, "learning_rate": 6.362494767685225e-05, "loss": 0.3081, "step": 1521 }, { "epoch": 1.115018315018315, "grad_norm": 1.8984375, "learning_rate": 6.36668061950607e-05, "loss": 0.246, "step": 1522 }, { "epoch": 1.1157509157509158, "grad_norm": 1.234375, "learning_rate": 6.370866471326915e-05, "loss": 0.254, "step": 1523 }, { "epoch": 1.1164835164835165, "grad_norm": 1.4453125, "learning_rate": 6.375052323147761e-05, "loss": 0.311, "step": 1524 }, { "epoch": 1.1172161172161172, "grad_norm": 1.0859375, "learning_rate": 6.379238174968606e-05, "loss": 0.2762, "step": 1525 }, { "epoch": 1.117948717948718, "grad_norm": 0.9375, "learning_rate": 6.383424026789451e-05, "loss": 0.2309, "step": 1526 }, { "epoch": 1.1186813186813187, "grad_norm": 1.09375, "learning_rate": 6.387609878610298e-05, "loss": 0.2588, "step": 1527 }, { "epoch": 1.1194139194139194, "grad_norm": 1.65625, "learning_rate": 6.391795730431143e-05, "loss": 0.2544, "step": 1528 }, { "epoch": 1.1201465201465202, "grad_norm": 1.2890625, "learning_rate": 6.395981582251988e-05, "loss": 0.2061, "step": 1529 }, { "epoch": 1.120879120879121, "grad_norm": 1.1796875, "learning_rate": 6.400167434072834e-05, "loss": 0.2652, "step": 1530 }, { "epoch": 1.1216117216117216, "grad_norm": 1.34375, "learning_rate": 6.40435328589368e-05, "loss": 0.2063, "step": 1531 }, { "epoch": 1.1223443223443224, "grad_norm": 0.91796875, "learning_rate": 6.408539137714526e-05, "loss": 0.2267, "step": 1532 }, { "epoch": 1.123076923076923, "grad_norm": 1.5, "learning_rate": 6.412724989535371e-05, "loss": 0.234, "step": 1533 }, { "epoch": 1.1238095238095238, "grad_norm": 0.88671875, "learning_rate": 6.416910841356216e-05, "loss": 0.2054, "step": 1534 }, { "epoch": 1.1245421245421245, "grad_norm": 1.09375, "learning_rate": 6.421096693177062e-05, "loss": 0.3118, "step": 1535 }, { "epoch": 1.1252747252747253, "grad_norm": 1.03125, "learning_rate": 6.425282544997907e-05, "loss": 0.2726, "step": 1536 }, { "epoch": 1.126007326007326, "grad_norm": 0.859375, "learning_rate": 6.429468396818753e-05, "loss": 0.2629, "step": 1537 }, { "epoch": 1.1267399267399267, "grad_norm": 1.21875, "learning_rate": 6.433654248639599e-05, "loss": 0.3342, "step": 1538 }, { "epoch": 1.1274725274725275, "grad_norm": 1.078125, "learning_rate": 6.437840100460444e-05, "loss": 0.2553, "step": 1539 }, { "epoch": 1.1282051282051282, "grad_norm": 1.484375, "learning_rate": 6.44202595228129e-05, "loss": 0.3317, "step": 1540 }, { "epoch": 1.128937728937729, "grad_norm": 1.5625, "learning_rate": 6.446211804102134e-05, "loss": 0.2965, "step": 1541 }, { "epoch": 1.1296703296703297, "grad_norm": 1.59375, "learning_rate": 6.45039765592298e-05, "loss": 0.2151, "step": 1542 }, { "epoch": 1.1304029304029304, "grad_norm": 1.0625, "learning_rate": 6.454583507743827e-05, "loss": 0.2837, "step": 1543 }, { "epoch": 1.1311355311355311, "grad_norm": 0.9296875, "learning_rate": 6.458769359564672e-05, "loss": 0.2474, "step": 1544 }, { "epoch": 1.1318681318681318, "grad_norm": 1.0234375, "learning_rate": 6.462955211385517e-05, "loss": 0.2569, "step": 1545 }, { "epoch": 1.1326007326007326, "grad_norm": 1.40625, "learning_rate": 6.467141063206362e-05, "loss": 0.2814, "step": 1546 }, { "epoch": 1.1333333333333333, "grad_norm": 2.03125, "learning_rate": 6.471326915027209e-05, "loss": 0.2912, "step": 1547 }, { "epoch": 1.134065934065934, "grad_norm": 1.4296875, "learning_rate": 6.475512766848054e-05, "loss": 0.2654, "step": 1548 }, { "epoch": 1.1347985347985348, "grad_norm": 2.21875, "learning_rate": 6.479698618668899e-05, "loss": 0.2511, "step": 1549 }, { "epoch": 1.1355311355311355, "grad_norm": 0.73828125, "learning_rate": 6.483884470489745e-05, "loss": 0.2068, "step": 1550 }, { "epoch": 1.1362637362637362, "grad_norm": 1.796875, "learning_rate": 6.488070322310592e-05, "loss": 0.2832, "step": 1551 }, { "epoch": 1.136996336996337, "grad_norm": 2.1875, "learning_rate": 6.492256174131435e-05, "loss": 0.297, "step": 1552 }, { "epoch": 1.1377289377289377, "grad_norm": 1.34375, "learning_rate": 6.496442025952282e-05, "loss": 0.2761, "step": 1553 }, { "epoch": 1.1384615384615384, "grad_norm": 1.609375, "learning_rate": 6.500627877773127e-05, "loss": 0.2416, "step": 1554 }, { "epoch": 1.1391941391941391, "grad_norm": 1.09375, "learning_rate": 6.504813729593973e-05, "loss": 0.214, "step": 1555 }, { "epoch": 1.1399267399267399, "grad_norm": 1.1015625, "learning_rate": 6.508999581414818e-05, "loss": 0.3078, "step": 1556 }, { "epoch": 1.1406593406593406, "grad_norm": 0.984375, "learning_rate": 6.513185433235663e-05, "loss": 0.2094, "step": 1557 }, { "epoch": 1.1413919413919413, "grad_norm": 0.83984375, "learning_rate": 6.51737128505651e-05, "loss": 0.1874, "step": 1558 }, { "epoch": 1.142124542124542, "grad_norm": 0.9921875, "learning_rate": 6.521557136877355e-05, "loss": 0.2825, "step": 1559 }, { "epoch": 1.1428571428571428, "grad_norm": 1.0078125, "learning_rate": 6.5257429886982e-05, "loss": 0.2563, "step": 1560 }, { "epoch": 1.1435897435897435, "grad_norm": 1.234375, "learning_rate": 6.529928840519046e-05, "loss": 0.255, "step": 1561 }, { "epoch": 1.1443223443223443, "grad_norm": 1.0078125, "learning_rate": 6.534114692339891e-05, "loss": 0.2316, "step": 1562 }, { "epoch": 1.145054945054945, "grad_norm": 0.96875, "learning_rate": 6.538300544160736e-05, "loss": 0.223, "step": 1563 }, { "epoch": 1.1457875457875457, "grad_norm": 1.0, "learning_rate": 6.542486395981583e-05, "loss": 0.206, "step": 1564 }, { "epoch": 1.1465201465201464, "grad_norm": 1.203125, "learning_rate": 6.546672247802428e-05, "loss": 0.2414, "step": 1565 }, { "epoch": 1.1472527472527472, "grad_norm": 0.91015625, "learning_rate": 6.550858099623274e-05, "loss": 0.2238, "step": 1566 }, { "epoch": 1.147985347985348, "grad_norm": 1.109375, "learning_rate": 6.555043951444118e-05, "loss": 0.2517, "step": 1567 }, { "epoch": 1.1487179487179486, "grad_norm": 1.1484375, "learning_rate": 6.559229803264964e-05, "loss": 0.2414, "step": 1568 }, { "epoch": 1.1494505494505494, "grad_norm": 1.1015625, "learning_rate": 6.563415655085811e-05, "loss": 0.2297, "step": 1569 }, { "epoch": 1.15018315018315, "grad_norm": 1.234375, "learning_rate": 6.567601506906656e-05, "loss": 0.2407, "step": 1570 }, { "epoch": 1.1509157509157508, "grad_norm": 1.03125, "learning_rate": 6.571787358727501e-05, "loss": 0.2622, "step": 1571 }, { "epoch": 1.1516483516483516, "grad_norm": 0.89453125, "learning_rate": 6.575973210548347e-05, "loss": 0.2555, "step": 1572 }, { "epoch": 1.1523809523809523, "grad_norm": 0.8828125, "learning_rate": 6.580159062369192e-05, "loss": 0.2037, "step": 1573 }, { "epoch": 1.153113553113553, "grad_norm": 0.9453125, "learning_rate": 6.584344914190039e-05, "loss": 0.2208, "step": 1574 }, { "epoch": 1.1538461538461537, "grad_norm": 1.078125, "learning_rate": 6.588530766010883e-05, "loss": 0.2442, "step": 1575 }, { "epoch": 1.1545787545787545, "grad_norm": 0.89453125, "learning_rate": 6.592716617831729e-05, "loss": 0.2259, "step": 1576 }, { "epoch": 1.1553113553113552, "grad_norm": 1.0546875, "learning_rate": 6.596902469652575e-05, "loss": 0.2547, "step": 1577 }, { "epoch": 1.156043956043956, "grad_norm": 1.0703125, "learning_rate": 6.60108832147342e-05, "loss": 0.2078, "step": 1578 }, { "epoch": 1.1567765567765567, "grad_norm": 1.3984375, "learning_rate": 6.605274173294266e-05, "loss": 0.259, "step": 1579 }, { "epoch": 1.1575091575091574, "grad_norm": 1.1796875, "learning_rate": 6.609460025115112e-05, "loss": 0.2876, "step": 1580 }, { "epoch": 1.1582417582417581, "grad_norm": 1.2265625, "learning_rate": 6.613645876935957e-05, "loss": 0.3001, "step": 1581 }, { "epoch": 1.1589743589743589, "grad_norm": 0.8984375, "learning_rate": 6.617831728756802e-05, "loss": 0.2399, "step": 1582 }, { "epoch": 1.1597069597069596, "grad_norm": 2.046875, "learning_rate": 6.622017580577647e-05, "loss": 0.2981, "step": 1583 }, { "epoch": 1.1604395604395605, "grad_norm": 1.234375, "learning_rate": 6.626203432398494e-05, "loss": 0.216, "step": 1584 }, { "epoch": 1.1611721611721613, "grad_norm": 0.87890625, "learning_rate": 6.63038928421934e-05, "loss": 0.2401, "step": 1585 }, { "epoch": 1.161904761904762, "grad_norm": 0.98046875, "learning_rate": 6.634575136040184e-05, "loss": 0.2765, "step": 1586 }, { "epoch": 1.1626373626373627, "grad_norm": 1.0078125, "learning_rate": 6.63876098786103e-05, "loss": 0.2445, "step": 1587 }, { "epoch": 1.1633699633699635, "grad_norm": 0.984375, "learning_rate": 6.642946839681875e-05, "loss": 0.2072, "step": 1588 }, { "epoch": 1.1641025641025642, "grad_norm": 1.078125, "learning_rate": 6.647132691502722e-05, "loss": 0.2762, "step": 1589 }, { "epoch": 1.164835164835165, "grad_norm": 1.0859375, "learning_rate": 6.651318543323567e-05, "loss": 0.2377, "step": 1590 }, { "epoch": 1.1655677655677656, "grad_norm": 0.8515625, "learning_rate": 6.655504395144412e-05, "loss": 0.2242, "step": 1591 }, { "epoch": 1.1663003663003664, "grad_norm": 1.1640625, "learning_rate": 6.659690246965258e-05, "loss": 0.2007, "step": 1592 }, { "epoch": 1.167032967032967, "grad_norm": 1.2734375, "learning_rate": 6.663876098786103e-05, "loss": 0.2258, "step": 1593 }, { "epoch": 1.1677655677655678, "grad_norm": 0.9140625, "learning_rate": 6.668061950606948e-05, "loss": 0.2757, "step": 1594 }, { "epoch": 1.1684981684981686, "grad_norm": 0.9140625, "learning_rate": 6.672247802427795e-05, "loss": 0.2885, "step": 1595 }, { "epoch": 1.1692307692307693, "grad_norm": 0.89453125, "learning_rate": 6.67643365424864e-05, "loss": 0.2698, "step": 1596 }, { "epoch": 1.16996336996337, "grad_norm": 1.2109375, "learning_rate": 6.680619506069485e-05, "loss": 0.3154, "step": 1597 }, { "epoch": 1.1706959706959708, "grad_norm": 1.1875, "learning_rate": 6.684805357890331e-05, "loss": 0.2254, "step": 1598 }, { "epoch": 1.1714285714285715, "grad_norm": 1.03125, "learning_rate": 6.688991209711176e-05, "loss": 0.2519, "step": 1599 }, { "epoch": 1.1721611721611722, "grad_norm": 1.3046875, "learning_rate": 6.693177061532023e-05, "loss": 0.308, "step": 1600 }, { "epoch": 1.172893772893773, "grad_norm": 0.734375, "learning_rate": 6.697362913352868e-05, "loss": 0.1894, "step": 1601 }, { "epoch": 1.1736263736263737, "grad_norm": 1.25, "learning_rate": 6.701548765173713e-05, "loss": 0.2197, "step": 1602 }, { "epoch": 1.1743589743589744, "grad_norm": 1.734375, "learning_rate": 6.705734616994559e-05, "loss": 0.2738, "step": 1603 }, { "epoch": 1.1750915750915751, "grad_norm": 1.109375, "learning_rate": 6.709920468815404e-05, "loss": 0.2012, "step": 1604 }, { "epoch": 1.1758241758241759, "grad_norm": 1.0234375, "learning_rate": 6.71410632063625e-05, "loss": 0.2153, "step": 1605 }, { "epoch": 1.1765567765567766, "grad_norm": 1.265625, "learning_rate": 6.718292172457096e-05, "loss": 0.2091, "step": 1606 }, { "epoch": 1.1772893772893773, "grad_norm": 1.203125, "learning_rate": 6.722478024277941e-05, "loss": 0.2786, "step": 1607 }, { "epoch": 1.178021978021978, "grad_norm": 0.921875, "learning_rate": 6.726663876098787e-05, "loss": 0.2165, "step": 1608 }, { "epoch": 1.1787545787545788, "grad_norm": 0.92578125, "learning_rate": 6.730849727919631e-05, "loss": 0.2451, "step": 1609 }, { "epoch": 1.1794871794871795, "grad_norm": 1.2578125, "learning_rate": 6.735035579740477e-05, "loss": 0.2258, "step": 1610 }, { "epoch": 1.1802197802197802, "grad_norm": 1.265625, "learning_rate": 6.739221431561324e-05, "loss": 0.2691, "step": 1611 }, { "epoch": 1.180952380952381, "grad_norm": 1.7890625, "learning_rate": 6.743407283382169e-05, "loss": 0.2491, "step": 1612 }, { "epoch": 1.1816849816849817, "grad_norm": 1.4140625, "learning_rate": 6.747593135203014e-05, "loss": 0.283, "step": 1613 }, { "epoch": 1.1824175824175824, "grad_norm": 1.4453125, "learning_rate": 6.75177898702386e-05, "loss": 0.2212, "step": 1614 }, { "epoch": 1.1831501831501832, "grad_norm": 0.75390625, "learning_rate": 6.755964838844705e-05, "loss": 0.2183, "step": 1615 }, { "epoch": 1.183882783882784, "grad_norm": 1.203125, "learning_rate": 6.76015069066555e-05, "loss": 0.2341, "step": 1616 }, { "epoch": 1.1846153846153846, "grad_norm": 0.98046875, "learning_rate": 6.764336542486396e-05, "loss": 0.24, "step": 1617 }, { "epoch": 1.1853479853479854, "grad_norm": 1.5703125, "learning_rate": 6.768522394307242e-05, "loss": 0.2641, "step": 1618 }, { "epoch": 1.186080586080586, "grad_norm": 1.3984375, "learning_rate": 6.772708246128088e-05, "loss": 0.2595, "step": 1619 }, { "epoch": 1.1868131868131868, "grad_norm": 0.9375, "learning_rate": 6.776894097948932e-05, "loss": 0.2432, "step": 1620 }, { "epoch": 1.1875457875457875, "grad_norm": 0.8828125, "learning_rate": 6.781079949769778e-05, "loss": 0.2453, "step": 1621 }, { "epoch": 1.1882783882783883, "grad_norm": 1.515625, "learning_rate": 6.785265801590625e-05, "loss": 0.3262, "step": 1622 }, { "epoch": 1.189010989010989, "grad_norm": 1.34375, "learning_rate": 6.78945165341147e-05, "loss": 0.2537, "step": 1623 }, { "epoch": 1.1897435897435897, "grad_norm": 1.03125, "learning_rate": 6.793637505232315e-05, "loss": 0.2013, "step": 1624 }, { "epoch": 1.1904761904761905, "grad_norm": 0.984375, "learning_rate": 6.79782335705316e-05, "loss": 0.2289, "step": 1625 }, { "epoch": 1.1912087912087912, "grad_norm": 0.9140625, "learning_rate": 6.802009208874006e-05, "loss": 0.2515, "step": 1626 }, { "epoch": 1.191941391941392, "grad_norm": 2.03125, "learning_rate": 6.806195060694852e-05, "loss": 0.2207, "step": 1627 }, { "epoch": 1.1926739926739927, "grad_norm": 1.21875, "learning_rate": 6.810380912515697e-05, "loss": 0.233, "step": 1628 }, { "epoch": 1.1934065934065934, "grad_norm": 0.9765625, "learning_rate": 6.814566764336543e-05, "loss": 0.2446, "step": 1629 }, { "epoch": 1.1941391941391941, "grad_norm": 1.109375, "learning_rate": 6.818752616157388e-05, "loss": 0.216, "step": 1630 }, { "epoch": 1.1948717948717948, "grad_norm": 1.0390625, "learning_rate": 6.822938467978233e-05, "loss": 0.2608, "step": 1631 }, { "epoch": 1.1956043956043956, "grad_norm": 1.375, "learning_rate": 6.82712431979908e-05, "loss": 0.2459, "step": 1632 }, { "epoch": 1.1963369963369963, "grad_norm": 1.875, "learning_rate": 6.831310171619925e-05, "loss": 0.2688, "step": 1633 }, { "epoch": 1.197069597069597, "grad_norm": 1.546875, "learning_rate": 6.835496023440771e-05, "loss": 0.2341, "step": 1634 }, { "epoch": 1.1978021978021978, "grad_norm": 1.6328125, "learning_rate": 6.839681875261616e-05, "loss": 0.2425, "step": 1635 }, { "epoch": 1.1985347985347985, "grad_norm": 2.109375, "learning_rate": 6.843867727082461e-05, "loss": 0.2611, "step": 1636 }, { "epoch": 1.1992673992673992, "grad_norm": 1.1328125, "learning_rate": 6.848053578903308e-05, "loss": 0.2303, "step": 1637 }, { "epoch": 1.2, "grad_norm": 1.53125, "learning_rate": 6.852239430724153e-05, "loss": 0.2651, "step": 1638 }, { "epoch": 1.2007326007326007, "grad_norm": 1.1015625, "learning_rate": 6.856425282544998e-05, "loss": 0.2761, "step": 1639 }, { "epoch": 1.2014652014652014, "grad_norm": 1.7421875, "learning_rate": 6.860611134365844e-05, "loss": 0.2913, "step": 1640 }, { "epoch": 1.2021978021978021, "grad_norm": 1.9765625, "learning_rate": 6.864796986186689e-05, "loss": 0.2162, "step": 1641 }, { "epoch": 1.2029304029304029, "grad_norm": 1.78125, "learning_rate": 6.868982838007534e-05, "loss": 0.278, "step": 1642 }, { "epoch": 1.2036630036630036, "grad_norm": 1.15625, "learning_rate": 6.873168689828381e-05, "loss": 0.2561, "step": 1643 }, { "epoch": 1.2043956043956043, "grad_norm": 1.4140625, "learning_rate": 6.877354541649226e-05, "loss": 0.2241, "step": 1644 }, { "epoch": 1.205128205128205, "grad_norm": 1.0390625, "learning_rate": 6.881540393470072e-05, "loss": 0.2388, "step": 1645 }, { "epoch": 1.2058608058608058, "grad_norm": 1.6328125, "learning_rate": 6.885726245290917e-05, "loss": 0.2495, "step": 1646 }, { "epoch": 1.2065934065934065, "grad_norm": 1.421875, "learning_rate": 6.889912097111762e-05, "loss": 0.3322, "step": 1647 }, { "epoch": 1.2073260073260073, "grad_norm": 1.1796875, "learning_rate": 6.894097948932609e-05, "loss": 0.2575, "step": 1648 }, { "epoch": 1.208058608058608, "grad_norm": 1.1640625, "learning_rate": 6.898283800753454e-05, "loss": 0.3139, "step": 1649 }, { "epoch": 1.2087912087912087, "grad_norm": 1.03125, "learning_rate": 6.902469652574299e-05, "loss": 0.2382, "step": 1650 }, { "epoch": 1.2095238095238094, "grad_norm": 1.1796875, "learning_rate": 6.906655504395144e-05, "loss": 0.2214, "step": 1651 }, { "epoch": 1.2102564102564102, "grad_norm": 0.66796875, "learning_rate": 6.91084135621599e-05, "loss": 0.226, "step": 1652 }, { "epoch": 1.210989010989011, "grad_norm": 1.3203125, "learning_rate": 6.915027208036837e-05, "loss": 0.2923, "step": 1653 }, { "epoch": 1.2117216117216116, "grad_norm": 0.63671875, "learning_rate": 6.91921305985768e-05, "loss": 0.2078, "step": 1654 }, { "epoch": 1.2124542124542124, "grad_norm": 0.80078125, "learning_rate": 6.923398911678527e-05, "loss": 0.2321, "step": 1655 }, { "epoch": 1.213186813186813, "grad_norm": 0.9453125, "learning_rate": 6.927584763499373e-05, "loss": 0.222, "step": 1656 }, { "epoch": 1.2139194139194138, "grad_norm": 1.078125, "learning_rate": 6.931770615320218e-05, "loss": 0.2403, "step": 1657 }, { "epoch": 1.2146520146520146, "grad_norm": 0.91796875, "learning_rate": 6.935956467141063e-05, "loss": 0.2537, "step": 1658 }, { "epoch": 1.2153846153846155, "grad_norm": 0.75, "learning_rate": 6.940142318961908e-05, "loss": 0.2223, "step": 1659 }, { "epoch": 1.2161172161172162, "grad_norm": 1.3984375, "learning_rate": 6.944328170782755e-05, "loss": 0.207, "step": 1660 }, { "epoch": 1.216849816849817, "grad_norm": 1.3046875, "learning_rate": 6.9485140226036e-05, "loss": 0.2783, "step": 1661 }, { "epoch": 1.2175824175824177, "grad_norm": 1.0625, "learning_rate": 6.952699874424445e-05, "loss": 0.2711, "step": 1662 }, { "epoch": 1.2183150183150184, "grad_norm": 1.234375, "learning_rate": 6.956885726245291e-05, "loss": 0.2658, "step": 1663 }, { "epoch": 1.2190476190476192, "grad_norm": 0.99609375, "learning_rate": 6.961071578066138e-05, "loss": 0.2225, "step": 1664 }, { "epoch": 1.2197802197802199, "grad_norm": 0.796875, "learning_rate": 6.965257429886982e-05, "loss": 0.2415, "step": 1665 }, { "epoch": 1.2205128205128206, "grad_norm": 1.3828125, "learning_rate": 6.969443281707828e-05, "loss": 0.227, "step": 1666 }, { "epoch": 1.2212454212454213, "grad_norm": 0.90625, "learning_rate": 6.973629133528673e-05, "loss": 0.2345, "step": 1667 }, { "epoch": 1.221978021978022, "grad_norm": 0.88671875, "learning_rate": 6.97781498534952e-05, "loss": 0.2248, "step": 1668 }, { "epoch": 1.2227106227106228, "grad_norm": 1.734375, "learning_rate": 6.982000837170365e-05, "loss": 0.2863, "step": 1669 }, { "epoch": 1.2234432234432235, "grad_norm": 1.1796875, "learning_rate": 6.98618668899121e-05, "loss": 0.2464, "step": 1670 }, { "epoch": 1.2241758241758243, "grad_norm": 1.0546875, "learning_rate": 6.990372540812056e-05, "loss": 0.2796, "step": 1671 }, { "epoch": 1.224908424908425, "grad_norm": 0.9609375, "learning_rate": 6.994558392632901e-05, "loss": 0.218, "step": 1672 }, { "epoch": 1.2256410256410257, "grad_norm": 1.3203125, "learning_rate": 6.998744244453746e-05, "loss": 0.2778, "step": 1673 }, { "epoch": 1.2263736263736265, "grad_norm": 2.15625, "learning_rate": 7.002930096274593e-05, "loss": 0.3176, "step": 1674 }, { "epoch": 1.2271062271062272, "grad_norm": 1.4921875, "learning_rate": 7.007115948095438e-05, "loss": 0.2297, "step": 1675 }, { "epoch": 1.227838827838828, "grad_norm": 1.4375, "learning_rate": 7.011301799916283e-05, "loss": 0.231, "step": 1676 }, { "epoch": 1.2285714285714286, "grad_norm": 1.8515625, "learning_rate": 7.015487651737129e-05, "loss": 0.2814, "step": 1677 }, { "epoch": 1.2293040293040294, "grad_norm": 1.796875, "learning_rate": 7.019673503557974e-05, "loss": 0.3115, "step": 1678 }, { "epoch": 1.23003663003663, "grad_norm": 1.9140625, "learning_rate": 7.02385935537882e-05, "loss": 0.2596, "step": 1679 }, { "epoch": 1.2307692307692308, "grad_norm": 0.96484375, "learning_rate": 7.028045207199666e-05, "loss": 0.2413, "step": 1680 }, { "epoch": 1.2315018315018316, "grad_norm": 0.87109375, "learning_rate": 7.032231059020511e-05, "loss": 0.2448, "step": 1681 }, { "epoch": 1.2322344322344323, "grad_norm": 1.953125, "learning_rate": 7.036416910841357e-05, "loss": 0.3099, "step": 1682 }, { "epoch": 1.232967032967033, "grad_norm": 1.203125, "learning_rate": 7.040602762662202e-05, "loss": 0.2386, "step": 1683 }, { "epoch": 1.2336996336996338, "grad_norm": 2.234375, "learning_rate": 7.044788614483047e-05, "loss": 0.3663, "step": 1684 }, { "epoch": 1.2344322344322345, "grad_norm": 0.96484375, "learning_rate": 7.048974466303894e-05, "loss": 0.2409, "step": 1685 }, { "epoch": 1.2351648351648352, "grad_norm": 1.109375, "learning_rate": 7.053160318124739e-05, "loss": 0.294, "step": 1686 }, { "epoch": 1.235897435897436, "grad_norm": 0.95703125, "learning_rate": 7.057346169945585e-05, "loss": 0.1753, "step": 1687 }, { "epoch": 1.2366300366300367, "grad_norm": 1.09375, "learning_rate": 7.061532021766429e-05, "loss": 0.2509, "step": 1688 }, { "epoch": 1.2373626373626374, "grad_norm": 0.921875, "learning_rate": 7.065717873587275e-05, "loss": 0.2091, "step": 1689 }, { "epoch": 1.2380952380952381, "grad_norm": 1.2421875, "learning_rate": 7.069903725408122e-05, "loss": 0.2529, "step": 1690 }, { "epoch": 1.2388278388278389, "grad_norm": 1.5234375, "learning_rate": 7.074089577228967e-05, "loss": 0.2402, "step": 1691 }, { "epoch": 1.2395604395604396, "grad_norm": 1.7265625, "learning_rate": 7.078275429049812e-05, "loss": 0.213, "step": 1692 }, { "epoch": 1.2402930402930403, "grad_norm": 1.3515625, "learning_rate": 7.082461280870657e-05, "loss": 0.2307, "step": 1693 }, { "epoch": 1.241025641025641, "grad_norm": 0.859375, "learning_rate": 7.086647132691503e-05, "loss": 0.223, "step": 1694 }, { "epoch": 1.2417582417582418, "grad_norm": 1.8125, "learning_rate": 7.090832984512348e-05, "loss": 0.2939, "step": 1695 }, { "epoch": 1.2424908424908425, "grad_norm": 2.265625, "learning_rate": 7.095018836333193e-05, "loss": 0.2268, "step": 1696 }, { "epoch": 1.2432234432234432, "grad_norm": 1.8359375, "learning_rate": 7.09920468815404e-05, "loss": 0.2132, "step": 1697 }, { "epoch": 1.243956043956044, "grad_norm": 1.6015625, "learning_rate": 7.103390539974886e-05, "loss": 0.2167, "step": 1698 }, { "epoch": 1.2446886446886447, "grad_norm": 0.7421875, "learning_rate": 7.10757639179573e-05, "loss": 0.2074, "step": 1699 }, { "epoch": 1.2454212454212454, "grad_norm": 2.34375, "learning_rate": 7.111762243616576e-05, "loss": 0.395, "step": 1700 }, { "epoch": 1.2461538461538462, "grad_norm": 3.171875, "learning_rate": 7.115948095437421e-05, "loss": 0.2169, "step": 1701 }, { "epoch": 1.246886446886447, "grad_norm": 3.46875, "learning_rate": 7.120133947258268e-05, "loss": 0.2392, "step": 1702 }, { "epoch": 1.2476190476190476, "grad_norm": 2.1875, "learning_rate": 7.124319799079113e-05, "loss": 0.2232, "step": 1703 }, { "epoch": 1.2483516483516484, "grad_norm": 1.28125, "learning_rate": 7.128505650899958e-05, "loss": 0.2595, "step": 1704 }, { "epoch": 1.249084249084249, "grad_norm": 0.8125, "learning_rate": 7.132691502720804e-05, "loss": 0.2709, "step": 1705 }, { "epoch": 1.2498168498168498, "grad_norm": 1.1484375, "learning_rate": 7.13687735454165e-05, "loss": 0.2522, "step": 1706 }, { "epoch": 1.2505494505494505, "grad_norm": 1.0859375, "learning_rate": 7.141063206362495e-05, "loss": 0.2, "step": 1707 }, { "epoch": 1.2512820512820513, "grad_norm": 2.078125, "learning_rate": 7.145249058183341e-05, "loss": 0.3277, "step": 1708 }, { "epoch": 1.252014652014652, "grad_norm": 1.15625, "learning_rate": 7.149434910004186e-05, "loss": 0.1988, "step": 1709 }, { "epoch": 1.2527472527472527, "grad_norm": 1.2109375, "learning_rate": 7.153620761825031e-05, "loss": 0.2044, "step": 1710 }, { "epoch": 1.2534798534798535, "grad_norm": 0.80859375, "learning_rate": 7.157806613645878e-05, "loss": 0.2148, "step": 1711 }, { "epoch": 1.2542124542124542, "grad_norm": 1.5078125, "learning_rate": 7.161992465466723e-05, "loss": 0.2708, "step": 1712 }, { "epoch": 1.254945054945055, "grad_norm": 1.4296875, "learning_rate": 7.166178317287569e-05, "loss": 0.3096, "step": 1713 }, { "epoch": 1.2556776556776557, "grad_norm": 1.890625, "learning_rate": 7.170364169108413e-05, "loss": 0.275, "step": 1714 }, { "epoch": 1.2564102564102564, "grad_norm": 1.6015625, "learning_rate": 7.174550020929259e-05, "loss": 0.2207, "step": 1715 }, { "epoch": 1.2571428571428571, "grad_norm": 1.2890625, "learning_rate": 7.178735872750106e-05, "loss": 0.2669, "step": 1716 }, { "epoch": 1.2578754578754578, "grad_norm": 0.9453125, "learning_rate": 7.18292172457095e-05, "loss": 0.225, "step": 1717 }, { "epoch": 1.2586080586080586, "grad_norm": 2.25, "learning_rate": 7.187107576391796e-05, "loss": 0.2812, "step": 1718 }, { "epoch": 1.2593406593406593, "grad_norm": 2.734375, "learning_rate": 7.191293428212642e-05, "loss": 0.2763, "step": 1719 }, { "epoch": 1.26007326007326, "grad_norm": 2.78125, "learning_rate": 7.195479280033487e-05, "loss": 0.273, "step": 1720 }, { "epoch": 1.2608058608058608, "grad_norm": 3.453125, "learning_rate": 7.199665131854334e-05, "loss": 0.2816, "step": 1721 }, { "epoch": 1.2615384615384615, "grad_norm": 1.2109375, "learning_rate": 7.203850983675177e-05, "loss": 0.2699, "step": 1722 }, { "epoch": 1.2622710622710622, "grad_norm": 0.86328125, "learning_rate": 7.208036835496024e-05, "loss": 0.2183, "step": 1723 }, { "epoch": 1.263003663003663, "grad_norm": 2.28125, "learning_rate": 7.21222268731687e-05, "loss": 0.2766, "step": 1724 }, { "epoch": 1.2637362637362637, "grad_norm": 2.046875, "learning_rate": 7.216408539137715e-05, "loss": 0.2804, "step": 1725 }, { "epoch": 1.2644688644688644, "grad_norm": 2.671875, "learning_rate": 7.22059439095856e-05, "loss": 0.25, "step": 1726 }, { "epoch": 1.2652014652014651, "grad_norm": 2.296875, "learning_rate": 7.224780242779407e-05, "loss": 0.2627, "step": 1727 }, { "epoch": 1.2659340659340659, "grad_norm": 0.76171875, "learning_rate": 7.228966094600252e-05, "loss": 0.191, "step": 1728 }, { "epoch": 1.2666666666666666, "grad_norm": 1.2421875, "learning_rate": 7.233151946421097e-05, "loss": 0.2574, "step": 1729 }, { "epoch": 1.2673992673992673, "grad_norm": 2.40625, "learning_rate": 7.237337798241942e-05, "loss": 0.3177, "step": 1730 }, { "epoch": 1.268131868131868, "grad_norm": 2.25, "learning_rate": 7.241523650062788e-05, "loss": 0.2675, "step": 1731 }, { "epoch": 1.2688644688644688, "grad_norm": 1.8828125, "learning_rate": 7.245709501883635e-05, "loss": 0.312, "step": 1732 }, { "epoch": 1.2695970695970695, "grad_norm": 0.88671875, "learning_rate": 7.249895353704478e-05, "loss": 0.2483, "step": 1733 }, { "epoch": 1.2703296703296703, "grad_norm": 1.234375, "learning_rate": 7.254081205525325e-05, "loss": 0.2473, "step": 1734 }, { "epoch": 1.271062271062271, "grad_norm": 1.1015625, "learning_rate": 7.25826705734617e-05, "loss": 0.2194, "step": 1735 }, { "epoch": 1.2717948717948717, "grad_norm": 1.4609375, "learning_rate": 7.262452909167016e-05, "loss": 0.198, "step": 1736 }, { "epoch": 1.2725274725274724, "grad_norm": 1.609375, "learning_rate": 7.266638760987861e-05, "loss": 0.2462, "step": 1737 }, { "epoch": 1.2732600732600732, "grad_norm": 0.91796875, "learning_rate": 7.270824612808706e-05, "loss": 0.2045, "step": 1738 }, { "epoch": 1.273992673992674, "grad_norm": 1.5859375, "learning_rate": 7.275010464629553e-05, "loss": 0.2641, "step": 1739 }, { "epoch": 1.2747252747252746, "grad_norm": 1.8671875, "learning_rate": 7.279196316450398e-05, "loss": 0.2612, "step": 1740 }, { "epoch": 1.2754578754578754, "grad_norm": 2.03125, "learning_rate": 7.283382168271243e-05, "loss": 0.2706, "step": 1741 }, { "epoch": 1.276190476190476, "grad_norm": 1.2734375, "learning_rate": 7.28756802009209e-05, "loss": 0.2481, "step": 1742 }, { "epoch": 1.2769230769230768, "grad_norm": 0.68359375, "learning_rate": 7.291753871912934e-05, "loss": 0.2158, "step": 1743 }, { "epoch": 1.2776556776556776, "grad_norm": 1.9765625, "learning_rate": 7.29593972373378e-05, "loss": 0.2239, "step": 1744 }, { "epoch": 1.2783882783882783, "grad_norm": 1.453125, "learning_rate": 7.300125575554626e-05, "loss": 0.2851, "step": 1745 }, { "epoch": 1.279120879120879, "grad_norm": 0.7890625, "learning_rate": 7.304311427375471e-05, "loss": 0.2184, "step": 1746 }, { "epoch": 1.2798534798534797, "grad_norm": 1.171875, "learning_rate": 7.308497279196317e-05, "loss": 0.2377, "step": 1747 }, { "epoch": 1.2805860805860805, "grad_norm": 1.359375, "learning_rate": 7.312683131017162e-05, "loss": 0.2371, "step": 1748 }, { "epoch": 1.2813186813186812, "grad_norm": 0.7265625, "learning_rate": 7.316868982838008e-05, "loss": 0.2401, "step": 1749 }, { "epoch": 1.282051282051282, "grad_norm": 0.890625, "learning_rate": 7.321054834658854e-05, "loss": 0.2145, "step": 1750 }, { "epoch": 1.2827838827838827, "grad_norm": 1.3984375, "learning_rate": 7.325240686479699e-05, "loss": 0.2769, "step": 1751 }, { "epoch": 1.2835164835164834, "grad_norm": 1.21875, "learning_rate": 7.329426538300544e-05, "loss": 0.2355, "step": 1752 }, { "epoch": 1.2842490842490841, "grad_norm": 2.015625, "learning_rate": 7.33361239012139e-05, "loss": 0.2204, "step": 1753 }, { "epoch": 1.2849816849816849, "grad_norm": 1.4375, "learning_rate": 7.337798241942236e-05, "loss": 0.2422, "step": 1754 }, { "epoch": 1.2857142857142856, "grad_norm": 1.03125, "learning_rate": 7.34198409376308e-05, "loss": 0.2328, "step": 1755 }, { "epoch": 1.2864468864468863, "grad_norm": 1.140625, "learning_rate": 7.346169945583926e-05, "loss": 0.276, "step": 1756 }, { "epoch": 1.287179487179487, "grad_norm": 1.078125, "learning_rate": 7.350355797404772e-05, "loss": 0.2709, "step": 1757 }, { "epoch": 1.2879120879120878, "grad_norm": 2.09375, "learning_rate": 7.354541649225618e-05, "loss": 0.232, "step": 1758 }, { "epoch": 1.2886446886446885, "grad_norm": 1.8671875, "learning_rate": 7.358727501046464e-05, "loss": 0.1944, "step": 1759 }, { "epoch": 1.2893772893772895, "grad_norm": 0.984375, "learning_rate": 7.362913352867309e-05, "loss": 0.236, "step": 1760 }, { "epoch": 1.2901098901098902, "grad_norm": 0.87890625, "learning_rate": 7.367099204688155e-05, "loss": 0.2315, "step": 1761 }, { "epoch": 1.290842490842491, "grad_norm": 1.2578125, "learning_rate": 7.371285056509e-05, "loss": 0.223, "step": 1762 }, { "epoch": 1.2915750915750916, "grad_norm": 1.953125, "learning_rate": 7.375470908329845e-05, "loss": 0.2656, "step": 1763 }, { "epoch": 1.2923076923076924, "grad_norm": 2.109375, "learning_rate": 7.37965676015069e-05, "loss": 0.2472, "step": 1764 }, { "epoch": 1.293040293040293, "grad_norm": 1.4296875, "learning_rate": 7.383842611971537e-05, "loss": 0.2595, "step": 1765 }, { "epoch": 1.2937728937728938, "grad_norm": 1.109375, "learning_rate": 7.388028463792383e-05, "loss": 0.2906, "step": 1766 }, { "epoch": 1.2945054945054946, "grad_norm": 1.3984375, "learning_rate": 7.392214315613227e-05, "loss": 0.2213, "step": 1767 }, { "epoch": 1.2952380952380953, "grad_norm": 1.7578125, "learning_rate": 7.396400167434073e-05, "loss": 0.2499, "step": 1768 }, { "epoch": 1.295970695970696, "grad_norm": 1.1875, "learning_rate": 7.40058601925492e-05, "loss": 0.2479, "step": 1769 }, { "epoch": 1.2967032967032968, "grad_norm": 1.1171875, "learning_rate": 7.404771871075765e-05, "loss": 0.2708, "step": 1770 }, { "epoch": 1.2974358974358975, "grad_norm": 0.96484375, "learning_rate": 7.40895772289661e-05, "loss": 0.2075, "step": 1771 }, { "epoch": 1.2981684981684982, "grad_norm": 1.234375, "learning_rate": 7.413143574717455e-05, "loss": 0.2592, "step": 1772 }, { "epoch": 1.298901098901099, "grad_norm": 1.5859375, "learning_rate": 7.417329426538301e-05, "loss": 0.2462, "step": 1773 }, { "epoch": 1.2996336996336997, "grad_norm": 1.4921875, "learning_rate": 7.421515278359146e-05, "loss": 0.2289, "step": 1774 }, { "epoch": 1.3003663003663004, "grad_norm": 0.9765625, "learning_rate": 7.425701130179991e-05, "loss": 0.2387, "step": 1775 }, { "epoch": 1.3010989010989011, "grad_norm": 1.25, "learning_rate": 7.429886982000838e-05, "loss": 0.2726, "step": 1776 }, { "epoch": 1.3018315018315019, "grad_norm": 2.09375, "learning_rate": 7.434072833821683e-05, "loss": 0.2762, "step": 1777 }, { "epoch": 1.3025641025641026, "grad_norm": 2.265625, "learning_rate": 7.438258685642528e-05, "loss": 0.2131, "step": 1778 }, { "epoch": 1.3032967032967033, "grad_norm": 1.234375, "learning_rate": 7.442444537463374e-05, "loss": 0.2052, "step": 1779 }, { "epoch": 1.304029304029304, "grad_norm": 0.828125, "learning_rate": 7.44663038928422e-05, "loss": 0.2269, "step": 1780 }, { "epoch": 1.3047619047619048, "grad_norm": 1.171875, "learning_rate": 7.450816241105066e-05, "loss": 0.2047, "step": 1781 }, { "epoch": 1.3054945054945055, "grad_norm": 1.8046875, "learning_rate": 7.455002092925911e-05, "loss": 0.3093, "step": 1782 }, { "epoch": 1.3062271062271062, "grad_norm": 1.4765625, "learning_rate": 7.459187944746756e-05, "loss": 0.2663, "step": 1783 }, { "epoch": 1.306959706959707, "grad_norm": 1.375, "learning_rate": 7.463373796567602e-05, "loss": 0.2684, "step": 1784 }, { "epoch": 1.3076923076923077, "grad_norm": 1.578125, "learning_rate": 7.467559648388447e-05, "loss": 0.2117, "step": 1785 }, { "epoch": 1.3084249084249084, "grad_norm": 1.328125, "learning_rate": 7.471745500209292e-05, "loss": 0.283, "step": 1786 }, { "epoch": 1.3091575091575092, "grad_norm": 1.0234375, "learning_rate": 7.475931352030139e-05, "loss": 0.2685, "step": 1787 }, { "epoch": 1.30989010989011, "grad_norm": 2.0625, "learning_rate": 7.480117203850984e-05, "loss": 0.1975, "step": 1788 }, { "epoch": 1.3106227106227106, "grad_norm": 1.9453125, "learning_rate": 7.484303055671829e-05, "loss": 0.2758, "step": 1789 }, { "epoch": 1.3113553113553114, "grad_norm": 0.94140625, "learning_rate": 7.488488907492675e-05, "loss": 0.2625, "step": 1790 }, { "epoch": 1.312087912087912, "grad_norm": 1.125, "learning_rate": 7.49267475931352e-05, "loss": 0.2093, "step": 1791 }, { "epoch": 1.3128205128205128, "grad_norm": 1.015625, "learning_rate": 7.496860611134367e-05, "loss": 0.2482, "step": 1792 }, { "epoch": 1.3135531135531135, "grad_norm": 1.1171875, "learning_rate": 7.501046462955212e-05, "loss": 0.2498, "step": 1793 }, { "epoch": 1.3142857142857143, "grad_norm": 1.2109375, "learning_rate": 7.505232314776057e-05, "loss": 0.2267, "step": 1794 }, { "epoch": 1.315018315018315, "grad_norm": 1.0625, "learning_rate": 7.509418166596903e-05, "loss": 0.2484, "step": 1795 }, { "epoch": 1.3157509157509157, "grad_norm": 1.3984375, "learning_rate": 7.513604018417749e-05, "loss": 0.2595, "step": 1796 }, { "epoch": 1.3164835164835165, "grad_norm": 0.88671875, "learning_rate": 7.517789870238594e-05, "loss": 0.2373, "step": 1797 }, { "epoch": 1.3172161172161172, "grad_norm": 1.1328125, "learning_rate": 7.52197572205944e-05, "loss": 0.2601, "step": 1798 }, { "epoch": 1.317948717948718, "grad_norm": 0.65234375, "learning_rate": 7.526161573880285e-05, "loss": 0.2075, "step": 1799 }, { "epoch": 1.3186813186813187, "grad_norm": 0.6171875, "learning_rate": 7.530347425701131e-05, "loss": 0.1988, "step": 1800 }, { "epoch": 1.3194139194139194, "grad_norm": 0.75390625, "learning_rate": 7.534533277521975e-05, "loss": 0.1913, "step": 1801 }, { "epoch": 1.3201465201465201, "grad_norm": 1.1875, "learning_rate": 7.538719129342822e-05, "loss": 0.2841, "step": 1802 }, { "epoch": 1.3208791208791208, "grad_norm": 0.8671875, "learning_rate": 7.542904981163668e-05, "loss": 0.2249, "step": 1803 }, { "epoch": 1.3216117216117216, "grad_norm": 1.0234375, "learning_rate": 7.547090832984513e-05, "loss": 0.2509, "step": 1804 }, { "epoch": 1.3223443223443223, "grad_norm": 1.0703125, "learning_rate": 7.551276684805358e-05, "loss": 0.2828, "step": 1805 }, { "epoch": 1.323076923076923, "grad_norm": 1.171875, "learning_rate": 7.555462536626203e-05, "loss": 0.1647, "step": 1806 }, { "epoch": 1.3238095238095238, "grad_norm": 1.4609375, "learning_rate": 7.55964838844705e-05, "loss": 0.2247, "step": 1807 }, { "epoch": 1.3245421245421245, "grad_norm": 0.9609375, "learning_rate": 7.563834240267895e-05, "loss": 0.259, "step": 1808 }, { "epoch": 1.3252747252747252, "grad_norm": 1.3046875, "learning_rate": 7.56802009208874e-05, "loss": 0.2388, "step": 1809 }, { "epoch": 1.326007326007326, "grad_norm": 1.796875, "learning_rate": 7.572205943909586e-05, "loss": 0.2401, "step": 1810 }, { "epoch": 1.3267399267399267, "grad_norm": 1.4609375, "learning_rate": 7.576391795730433e-05, "loss": 0.3007, "step": 1811 }, { "epoch": 1.3274725274725274, "grad_norm": 1.015625, "learning_rate": 7.580577647551276e-05, "loss": 0.2362, "step": 1812 }, { "epoch": 1.3282051282051281, "grad_norm": 1.0234375, "learning_rate": 7.584763499372123e-05, "loss": 0.2027, "step": 1813 }, { "epoch": 1.3289377289377289, "grad_norm": 1.171875, "learning_rate": 7.588949351192968e-05, "loss": 0.2434, "step": 1814 }, { "epoch": 1.3296703296703296, "grad_norm": 1.09375, "learning_rate": 7.593135203013814e-05, "loss": 0.2023, "step": 1815 }, { "epoch": 1.3304029304029303, "grad_norm": 0.78515625, "learning_rate": 7.597321054834659e-05, "loss": 0.2429, "step": 1816 }, { "epoch": 1.331135531135531, "grad_norm": 1.0859375, "learning_rate": 7.601506906655504e-05, "loss": 0.1918, "step": 1817 }, { "epoch": 1.3318681318681318, "grad_norm": 1.0546875, "learning_rate": 7.605692758476351e-05, "loss": 0.2144, "step": 1818 }, { "epoch": 1.3326007326007325, "grad_norm": 0.81640625, "learning_rate": 7.609878610297196e-05, "loss": 0.1892, "step": 1819 }, { "epoch": 1.3333333333333333, "grad_norm": 1.3203125, "learning_rate": 7.614064462118041e-05, "loss": 0.2645, "step": 1820 }, { "epoch": 1.334065934065934, "grad_norm": 1.296875, "learning_rate": 7.618250313938887e-05, "loss": 0.2429, "step": 1821 }, { "epoch": 1.3347985347985347, "grad_norm": 1.6640625, "learning_rate": 7.622436165759732e-05, "loss": 0.247, "step": 1822 }, { "epoch": 1.3355311355311354, "grad_norm": 2.234375, "learning_rate": 7.626622017580577e-05, "loss": 0.2573, "step": 1823 }, { "epoch": 1.3362637362637364, "grad_norm": 1.140625, "learning_rate": 7.630807869401424e-05, "loss": 0.2514, "step": 1824 }, { "epoch": 1.3369963369963371, "grad_norm": 0.95703125, "learning_rate": 7.634993721222269e-05, "loss": 0.2021, "step": 1825 }, { "epoch": 1.3377289377289379, "grad_norm": 1.7890625, "learning_rate": 7.639179573043115e-05, "loss": 0.2818, "step": 1826 }, { "epoch": 1.3384615384615386, "grad_norm": 1.640625, "learning_rate": 7.643365424863959e-05, "loss": 0.2434, "step": 1827 }, { "epoch": 1.3391941391941393, "grad_norm": 0.97265625, "learning_rate": 7.647551276684805e-05, "loss": 0.2254, "step": 1828 }, { "epoch": 1.33992673992674, "grad_norm": 0.88671875, "learning_rate": 7.651737128505652e-05, "loss": 0.1735, "step": 1829 }, { "epoch": 1.3406593406593408, "grad_norm": 1.109375, "learning_rate": 7.655922980326497e-05, "loss": 0.2277, "step": 1830 }, { "epoch": 1.3413919413919415, "grad_norm": 1.0859375, "learning_rate": 7.660108832147342e-05, "loss": 0.2495, "step": 1831 }, { "epoch": 1.3421245421245422, "grad_norm": 1.875, "learning_rate": 7.664294683968188e-05, "loss": 0.2672, "step": 1832 }, { "epoch": 1.342857142857143, "grad_norm": 1.125, "learning_rate": 7.668480535789033e-05, "loss": 0.2144, "step": 1833 }, { "epoch": 1.3435897435897437, "grad_norm": 1.0625, "learning_rate": 7.67266638760988e-05, "loss": 0.2082, "step": 1834 }, { "epoch": 1.3443223443223444, "grad_norm": 1.078125, "learning_rate": 7.676852239430724e-05, "loss": 0.2655, "step": 1835 }, { "epoch": 1.3450549450549452, "grad_norm": 1.3984375, "learning_rate": 7.68103809125157e-05, "loss": 0.2446, "step": 1836 }, { "epoch": 1.345787545787546, "grad_norm": 1.140625, "learning_rate": 7.685223943072416e-05, "loss": 0.227, "step": 1837 }, { "epoch": 1.3465201465201466, "grad_norm": 1.0703125, "learning_rate": 7.689409794893261e-05, "loss": 0.2297, "step": 1838 }, { "epoch": 1.3472527472527474, "grad_norm": 1.0234375, "learning_rate": 7.693595646714107e-05, "loss": 0.2158, "step": 1839 }, { "epoch": 1.347985347985348, "grad_norm": 1.1171875, "learning_rate": 7.697781498534953e-05, "loss": 0.2345, "step": 1840 }, { "epoch": 1.3487179487179488, "grad_norm": 1.296875, "learning_rate": 7.701967350355798e-05, "loss": 0.2485, "step": 1841 }, { "epoch": 1.3494505494505495, "grad_norm": 1.3515625, "learning_rate": 7.706153202176643e-05, "loss": 0.2614, "step": 1842 }, { "epoch": 1.3501831501831503, "grad_norm": 1.4765625, "learning_rate": 7.710339053997488e-05, "loss": 0.2062, "step": 1843 }, { "epoch": 1.350915750915751, "grad_norm": 1.4453125, "learning_rate": 7.714524905818335e-05, "loss": 0.2276, "step": 1844 }, { "epoch": 1.3516483516483517, "grad_norm": 1.2109375, "learning_rate": 7.718710757639181e-05, "loss": 0.2316, "step": 1845 }, { "epoch": 1.3523809523809525, "grad_norm": 0.91015625, "learning_rate": 7.722896609460025e-05, "loss": 0.2413, "step": 1846 }, { "epoch": 1.3531135531135532, "grad_norm": 1.2421875, "learning_rate": 7.727082461280871e-05, "loss": 0.2168, "step": 1847 }, { "epoch": 1.353846153846154, "grad_norm": 1.78125, "learning_rate": 7.731268313101716e-05, "loss": 0.2563, "step": 1848 }, { "epoch": 1.3545787545787547, "grad_norm": 2.015625, "learning_rate": 7.735454164922563e-05, "loss": 0.2416, "step": 1849 }, { "epoch": 1.3553113553113554, "grad_norm": 1.0390625, "learning_rate": 7.739640016743408e-05, "loss": 0.2252, "step": 1850 }, { "epoch": 1.3560439560439561, "grad_norm": 0.9296875, "learning_rate": 7.743825868564253e-05, "loss": 0.2466, "step": 1851 }, { "epoch": 1.3567765567765568, "grad_norm": 1.0703125, "learning_rate": 7.748011720385099e-05, "loss": 0.2059, "step": 1852 }, { "epoch": 1.3575091575091576, "grad_norm": 2.1875, "learning_rate": 7.752197572205944e-05, "loss": 0.2668, "step": 1853 }, { "epoch": 1.3582417582417583, "grad_norm": 1.1796875, "learning_rate": 7.756383424026789e-05, "loss": 0.2168, "step": 1854 }, { "epoch": 1.358974358974359, "grad_norm": 1.3125, "learning_rate": 7.760569275847636e-05, "loss": 0.2793, "step": 1855 }, { "epoch": 1.3597069597069598, "grad_norm": 1.796875, "learning_rate": 7.764755127668481e-05, "loss": 0.2709, "step": 1856 }, { "epoch": 1.3604395604395605, "grad_norm": 2.0, "learning_rate": 7.768940979489326e-05, "loss": 0.2552, "step": 1857 }, { "epoch": 1.3611721611721612, "grad_norm": 1.3671875, "learning_rate": 7.773126831310172e-05, "loss": 0.2486, "step": 1858 }, { "epoch": 1.361904761904762, "grad_norm": 1.25, "learning_rate": 7.777312683131017e-05, "loss": 0.2289, "step": 1859 }, { "epoch": 1.3626373626373627, "grad_norm": 1.3828125, "learning_rate": 7.781498534951864e-05, "loss": 0.2459, "step": 1860 }, { "epoch": 1.3633699633699634, "grad_norm": 1.421875, "learning_rate": 7.785684386772709e-05, "loss": 0.234, "step": 1861 }, { "epoch": 1.3641025641025641, "grad_norm": 1.4609375, "learning_rate": 7.789870238593554e-05, "loss": 0.21, "step": 1862 }, { "epoch": 1.3648351648351649, "grad_norm": 1.7265625, "learning_rate": 7.7940560904144e-05, "loss": 0.2368, "step": 1863 }, { "epoch": 1.3655677655677656, "grad_norm": 1.0703125, "learning_rate": 7.798241942235245e-05, "loss": 0.2551, "step": 1864 }, { "epoch": 1.3663003663003663, "grad_norm": 1.0703125, "learning_rate": 7.80242779405609e-05, "loss": 0.2046, "step": 1865 }, { "epoch": 1.367032967032967, "grad_norm": 1.265625, "learning_rate": 7.806613645876937e-05, "loss": 0.2179, "step": 1866 }, { "epoch": 1.3677655677655678, "grad_norm": 1.4609375, "learning_rate": 7.810799497697782e-05, "loss": 0.2321, "step": 1867 }, { "epoch": 1.3684981684981685, "grad_norm": 1.6328125, "learning_rate": 7.814985349518628e-05, "loss": 0.2785, "step": 1868 }, { "epoch": 1.3692307692307693, "grad_norm": 1.2109375, "learning_rate": 7.819171201339472e-05, "loss": 0.2664, "step": 1869 }, { "epoch": 1.36996336996337, "grad_norm": 1.2265625, "learning_rate": 7.823357053160318e-05, "loss": 0.2625, "step": 1870 }, { "epoch": 1.3706959706959707, "grad_norm": 1.453125, "learning_rate": 7.827542904981165e-05, "loss": 0.2828, "step": 1871 }, { "epoch": 1.3714285714285714, "grad_norm": 1.4921875, "learning_rate": 7.83172875680201e-05, "loss": 0.2469, "step": 1872 }, { "epoch": 1.3721611721611722, "grad_norm": 1.671875, "learning_rate": 7.835914608622855e-05, "loss": 0.2688, "step": 1873 }, { "epoch": 1.372893772893773, "grad_norm": 1.09375, "learning_rate": 7.840100460443701e-05, "loss": 0.2105, "step": 1874 }, { "epoch": 1.3736263736263736, "grad_norm": 1.28125, "learning_rate": 7.844286312264546e-05, "loss": 0.2529, "step": 1875 }, { "epoch": 1.3743589743589744, "grad_norm": 1.0625, "learning_rate": 7.848472164085391e-05, "loss": 0.2122, "step": 1876 }, { "epoch": 1.375091575091575, "grad_norm": 0.9609375, "learning_rate": 7.852658015906237e-05, "loss": 0.2543, "step": 1877 }, { "epoch": 1.3758241758241758, "grad_norm": 1.0, "learning_rate": 7.856843867727083e-05, "loss": 0.1981, "step": 1878 }, { "epoch": 1.3765567765567766, "grad_norm": 1.3125, "learning_rate": 7.86102971954793e-05, "loss": 0.2162, "step": 1879 }, { "epoch": 1.3772893772893773, "grad_norm": 0.7734375, "learning_rate": 7.865215571368773e-05, "loss": 0.1987, "step": 1880 }, { "epoch": 1.378021978021978, "grad_norm": 1.625, "learning_rate": 7.86940142318962e-05, "loss": 0.1939, "step": 1881 }, { "epoch": 1.3787545787545787, "grad_norm": 1.3984375, "learning_rate": 7.873587275010466e-05, "loss": 0.2285, "step": 1882 }, { "epoch": 1.3794871794871795, "grad_norm": 0.97265625, "learning_rate": 7.877773126831311e-05, "loss": 0.2177, "step": 1883 }, { "epoch": 1.3802197802197802, "grad_norm": 1.078125, "learning_rate": 7.881958978652156e-05, "loss": 0.2577, "step": 1884 }, { "epoch": 1.380952380952381, "grad_norm": 1.3125, "learning_rate": 7.886144830473001e-05, "loss": 0.2282, "step": 1885 }, { "epoch": 1.3816849816849817, "grad_norm": 1.8046875, "learning_rate": 7.890330682293848e-05, "loss": 0.1844, "step": 1886 }, { "epoch": 1.3824175824175824, "grad_norm": 1.453125, "learning_rate": 7.894516534114693e-05, "loss": 0.2372, "step": 1887 }, { "epoch": 1.3831501831501831, "grad_norm": 1.1640625, "learning_rate": 7.898702385935538e-05, "loss": 0.2198, "step": 1888 }, { "epoch": 1.3838827838827839, "grad_norm": 0.765625, "learning_rate": 7.902888237756384e-05, "loss": 0.1874, "step": 1889 }, { "epoch": 1.3846153846153846, "grad_norm": 0.92578125, "learning_rate": 7.907074089577229e-05, "loss": 0.1969, "step": 1890 }, { "epoch": 1.3853479853479853, "grad_norm": 1.421875, "learning_rate": 7.911259941398074e-05, "loss": 0.2841, "step": 1891 }, { "epoch": 1.386080586080586, "grad_norm": 0.828125, "learning_rate": 7.91544579321892e-05, "loss": 0.2171, "step": 1892 }, { "epoch": 1.3868131868131868, "grad_norm": 0.76171875, "learning_rate": 7.919631645039766e-05, "loss": 0.1855, "step": 1893 }, { "epoch": 1.3875457875457875, "grad_norm": 0.96875, "learning_rate": 7.923817496860612e-05, "loss": 0.2466, "step": 1894 }, { "epoch": 1.3882783882783882, "grad_norm": 1.1640625, "learning_rate": 7.928003348681457e-05, "loss": 0.2668, "step": 1895 }, { "epoch": 1.389010989010989, "grad_norm": 1.25, "learning_rate": 7.932189200502302e-05, "loss": 0.2307, "step": 1896 }, { "epoch": 1.3897435897435897, "grad_norm": 1.1171875, "learning_rate": 7.936375052323149e-05, "loss": 0.1872, "step": 1897 }, { "epoch": 1.3904761904761904, "grad_norm": 1.296875, "learning_rate": 7.940560904143994e-05, "loss": 0.2347, "step": 1898 }, { "epoch": 1.3912087912087912, "grad_norm": 0.89453125, "learning_rate": 7.944746755964839e-05, "loss": 0.2342, "step": 1899 }, { "epoch": 1.3919413919413919, "grad_norm": 1.25, "learning_rate": 7.948932607785685e-05, "loss": 0.2809, "step": 1900 }, { "epoch": 1.3926739926739926, "grad_norm": 1.25, "learning_rate": 7.95311845960653e-05, "loss": 0.2602, "step": 1901 }, { "epoch": 1.3934065934065933, "grad_norm": 1.03125, "learning_rate": 7.957304311427375e-05, "loss": 0.2767, "step": 1902 }, { "epoch": 1.394139194139194, "grad_norm": 1.03125, "learning_rate": 7.961490163248222e-05, "loss": 0.1841, "step": 1903 }, { "epoch": 1.3948717948717948, "grad_norm": 1.0078125, "learning_rate": 7.965676015069067e-05, "loss": 0.2163, "step": 1904 }, { "epoch": 1.3956043956043955, "grad_norm": 0.75, "learning_rate": 7.969861866889913e-05, "loss": 0.163, "step": 1905 }, { "epoch": 1.3963369963369963, "grad_norm": 1.2890625, "learning_rate": 7.974047718710758e-05, "loss": 0.2827, "step": 1906 }, { "epoch": 1.397069597069597, "grad_norm": 0.875, "learning_rate": 7.978233570531603e-05, "loss": 0.1908, "step": 1907 }, { "epoch": 1.3978021978021977, "grad_norm": 1.1953125, "learning_rate": 7.98241942235245e-05, "loss": 0.2399, "step": 1908 }, { "epoch": 1.3985347985347985, "grad_norm": 0.921875, "learning_rate": 7.986605274173295e-05, "loss": 0.2016, "step": 1909 }, { "epoch": 1.3992673992673992, "grad_norm": 1.328125, "learning_rate": 7.99079112599414e-05, "loss": 0.2556, "step": 1910 }, { "epoch": 1.4, "grad_norm": 0.91796875, "learning_rate": 7.994976977814985e-05, "loss": 0.2427, "step": 1911 }, { "epoch": 1.4007326007326006, "grad_norm": 1.171875, "learning_rate": 7.999162829635831e-05, "loss": 0.2326, "step": 1912 }, { "epoch": 1.4014652014652014, "grad_norm": 0.90625, "learning_rate": 8.003348681456678e-05, "loss": 0.1976, "step": 1913 }, { "epoch": 1.402197802197802, "grad_norm": 1.34375, "learning_rate": 8.007534533277521e-05, "loss": 0.2638, "step": 1914 }, { "epoch": 1.4029304029304028, "grad_norm": 1.109375, "learning_rate": 8.011720385098368e-05, "loss": 0.226, "step": 1915 }, { "epoch": 1.4036630036630036, "grad_norm": 1.84375, "learning_rate": 8.015906236919214e-05, "loss": 0.2108, "step": 1916 }, { "epoch": 1.4043956043956043, "grad_norm": 0.85546875, "learning_rate": 8.02009208874006e-05, "loss": 0.2067, "step": 1917 }, { "epoch": 1.405128205128205, "grad_norm": 0.890625, "learning_rate": 8.024277940560904e-05, "loss": 0.2263, "step": 1918 }, { "epoch": 1.4058608058608058, "grad_norm": 1.15625, "learning_rate": 8.02846379238175e-05, "loss": 0.2079, "step": 1919 }, { "epoch": 1.4065934065934065, "grad_norm": 1.078125, "learning_rate": 8.032649644202596e-05, "loss": 0.2286, "step": 1920 }, { "epoch": 1.4073260073260072, "grad_norm": 1.515625, "learning_rate": 8.036835496023441e-05, "loss": 0.2404, "step": 1921 }, { "epoch": 1.408058608058608, "grad_norm": 0.96484375, "learning_rate": 8.041021347844286e-05, "loss": 0.2167, "step": 1922 }, { "epoch": 1.4087912087912087, "grad_norm": 1.046875, "learning_rate": 8.045207199665132e-05, "loss": 0.2004, "step": 1923 }, { "epoch": 1.4095238095238094, "grad_norm": 0.71484375, "learning_rate": 8.049393051485979e-05, "loss": 0.1764, "step": 1924 }, { "epoch": 1.4102564102564101, "grad_norm": 1.1953125, "learning_rate": 8.053578903306823e-05, "loss": 0.1945, "step": 1925 }, { "epoch": 1.4109890109890109, "grad_norm": 0.8828125, "learning_rate": 8.057764755127669e-05, "loss": 0.221, "step": 1926 }, { "epoch": 1.4117216117216116, "grad_norm": 0.96875, "learning_rate": 8.061950606948514e-05, "loss": 0.2255, "step": 1927 }, { "epoch": 1.4124542124542123, "grad_norm": 1.7421875, "learning_rate": 8.06613645876936e-05, "loss": 0.2656, "step": 1928 }, { "epoch": 1.413186813186813, "grad_norm": 1.7890625, "learning_rate": 8.070322310590206e-05, "loss": 0.2569, "step": 1929 }, { "epoch": 1.4139194139194138, "grad_norm": 1.4140625, "learning_rate": 8.07450816241105e-05, "loss": 0.2465, "step": 1930 }, { "epoch": 1.4146520146520147, "grad_norm": 1.1171875, "learning_rate": 8.078694014231897e-05, "loss": 0.2773, "step": 1931 }, { "epoch": 1.4153846153846155, "grad_norm": 1.234375, "learning_rate": 8.082879866052742e-05, "loss": 0.1916, "step": 1932 }, { "epoch": 1.4161172161172162, "grad_norm": 1.6875, "learning_rate": 8.087065717873587e-05, "loss": 0.2475, "step": 1933 }, { "epoch": 1.416849816849817, "grad_norm": 1.4453125, "learning_rate": 8.091251569694434e-05, "loss": 0.2271, "step": 1934 }, { "epoch": 1.4175824175824177, "grad_norm": 1.6015625, "learning_rate": 8.095437421515279e-05, "loss": 0.2429, "step": 1935 }, { "epoch": 1.4183150183150184, "grad_norm": 0.97265625, "learning_rate": 8.099623273336124e-05, "loss": 0.2562, "step": 1936 }, { "epoch": 1.4190476190476191, "grad_norm": 1.375, "learning_rate": 8.10380912515697e-05, "loss": 0.2772, "step": 1937 }, { "epoch": 1.4197802197802198, "grad_norm": 1.8046875, "learning_rate": 8.107994976977815e-05, "loss": 0.1638, "step": 1938 }, { "epoch": 1.4205128205128206, "grad_norm": 2.03125, "learning_rate": 8.112180828798662e-05, "loss": 0.2528, "step": 1939 }, { "epoch": 1.4212454212454213, "grad_norm": 1.3203125, "learning_rate": 8.116366680619505e-05, "loss": 0.2156, "step": 1940 }, { "epoch": 1.421978021978022, "grad_norm": 1.484375, "learning_rate": 8.120552532440352e-05, "loss": 0.2609, "step": 1941 }, { "epoch": 1.4227106227106228, "grad_norm": 1.3046875, "learning_rate": 8.124738384261198e-05, "loss": 0.1713, "step": 1942 }, { "epoch": 1.4234432234432235, "grad_norm": 1.9609375, "learning_rate": 8.128924236082043e-05, "loss": 0.293, "step": 1943 }, { "epoch": 1.4241758241758242, "grad_norm": 1.265625, "learning_rate": 8.133110087902888e-05, "loss": 0.2585, "step": 1944 }, { "epoch": 1.424908424908425, "grad_norm": 1.0234375, "learning_rate": 8.137295939723735e-05, "loss": 0.2315, "step": 1945 }, { "epoch": 1.4256410256410257, "grad_norm": 1.546875, "learning_rate": 8.14148179154458e-05, "loss": 0.2469, "step": 1946 }, { "epoch": 1.4263736263736264, "grad_norm": 2.109375, "learning_rate": 8.145667643365426e-05, "loss": 0.2201, "step": 1947 }, { "epoch": 1.4271062271062271, "grad_norm": 1.234375, "learning_rate": 8.14985349518627e-05, "loss": 0.2543, "step": 1948 }, { "epoch": 1.4278388278388279, "grad_norm": 1.3828125, "learning_rate": 8.154039347007116e-05, "loss": 0.2179, "step": 1949 }, { "epoch": 1.4285714285714286, "grad_norm": 1.046875, "learning_rate": 8.158225198827963e-05, "loss": 0.2504, "step": 1950 }, { "epoch": 1.4293040293040293, "grad_norm": 1.640625, "learning_rate": 8.162411050648808e-05, "loss": 0.2417, "step": 1951 }, { "epoch": 1.43003663003663, "grad_norm": 1.5703125, "learning_rate": 8.166596902469653e-05, "loss": 0.2304, "step": 1952 }, { "epoch": 1.4307692307692308, "grad_norm": 1.8203125, "learning_rate": 8.170782754290498e-05, "loss": 0.1925, "step": 1953 }, { "epoch": 1.4315018315018315, "grad_norm": 1.6015625, "learning_rate": 8.174968606111344e-05, "loss": 0.183, "step": 1954 }, { "epoch": 1.4322344322344323, "grad_norm": 0.9609375, "learning_rate": 8.17915445793219e-05, "loss": 0.1978, "step": 1955 }, { "epoch": 1.432967032967033, "grad_norm": 0.9296875, "learning_rate": 8.183340309753034e-05, "loss": 0.2302, "step": 1956 }, { "epoch": 1.4336996336996337, "grad_norm": 1.3984375, "learning_rate": 8.187526161573881e-05, "loss": 0.2341, "step": 1957 }, { "epoch": 1.4344322344322344, "grad_norm": 1.265625, "learning_rate": 8.191712013394727e-05, "loss": 0.2511, "step": 1958 }, { "epoch": 1.4351648351648352, "grad_norm": 0.8984375, "learning_rate": 8.195897865215571e-05, "loss": 0.2501, "step": 1959 }, { "epoch": 1.435897435897436, "grad_norm": 1.5546875, "learning_rate": 8.200083717036417e-05, "loss": 0.218, "step": 1960 }, { "epoch": 1.4366300366300366, "grad_norm": 1.640625, "learning_rate": 8.204269568857262e-05, "loss": 0.2344, "step": 1961 }, { "epoch": 1.4373626373626374, "grad_norm": 1.0859375, "learning_rate": 8.208455420678109e-05, "loss": 0.1951, "step": 1962 }, { "epoch": 1.438095238095238, "grad_norm": 1.015625, "learning_rate": 8.212641272498954e-05, "loss": 0.1912, "step": 1963 }, { "epoch": 1.4388278388278388, "grad_norm": 1.3515625, "learning_rate": 8.216827124319799e-05, "loss": 0.255, "step": 1964 }, { "epoch": 1.4395604395604396, "grad_norm": 1.7578125, "learning_rate": 8.221012976140645e-05, "loss": 0.264, "step": 1965 }, { "epoch": 1.4402930402930403, "grad_norm": 1.4921875, "learning_rate": 8.22519882796149e-05, "loss": 0.218, "step": 1966 }, { "epoch": 1.441025641025641, "grad_norm": 1.3203125, "learning_rate": 8.229384679782336e-05, "loss": 0.2599, "step": 1967 }, { "epoch": 1.4417582417582417, "grad_norm": 0.9921875, "learning_rate": 8.233570531603182e-05, "loss": 0.2529, "step": 1968 }, { "epoch": 1.4424908424908425, "grad_norm": 1.4453125, "learning_rate": 8.237756383424027e-05, "loss": 0.2504, "step": 1969 }, { "epoch": 1.4432234432234432, "grad_norm": 1.2109375, "learning_rate": 8.241942235244872e-05, "loss": 0.1995, "step": 1970 }, { "epoch": 1.443956043956044, "grad_norm": 1.0234375, "learning_rate": 8.246128087065719e-05, "loss": 0.2046, "step": 1971 }, { "epoch": 1.4446886446886447, "grad_norm": 1.0234375, "learning_rate": 8.250313938886564e-05, "loss": 0.2581, "step": 1972 }, { "epoch": 1.4454212454212454, "grad_norm": 0.9140625, "learning_rate": 8.25449979070741e-05, "loss": 0.2208, "step": 1973 }, { "epoch": 1.4461538461538461, "grad_norm": 1.2265625, "learning_rate": 8.258685642528254e-05, "loss": 0.2191, "step": 1974 }, { "epoch": 1.4468864468864469, "grad_norm": 1.1875, "learning_rate": 8.2628714943491e-05, "loss": 0.2344, "step": 1975 }, { "epoch": 1.4476190476190476, "grad_norm": 1.4140625, "learning_rate": 8.267057346169947e-05, "loss": 0.2484, "step": 1976 }, { "epoch": 1.4483516483516483, "grad_norm": 1.6328125, "learning_rate": 8.271243197990792e-05, "loss": 0.2213, "step": 1977 }, { "epoch": 1.449084249084249, "grad_norm": 1.59375, "learning_rate": 8.275429049811637e-05, "loss": 0.2845, "step": 1978 }, { "epoch": 1.4498168498168498, "grad_norm": 0.984375, "learning_rate": 8.279614901632483e-05, "loss": 0.2181, "step": 1979 }, { "epoch": 1.4505494505494505, "grad_norm": 1.3125, "learning_rate": 8.283800753453328e-05, "loss": 0.2605, "step": 1980 }, { "epoch": 1.4512820512820512, "grad_norm": 1.4921875, "learning_rate": 8.287986605274175e-05, "loss": 0.207, "step": 1981 }, { "epoch": 1.452014652014652, "grad_norm": 1.0390625, "learning_rate": 8.292172457095018e-05, "loss": 0.1999, "step": 1982 }, { "epoch": 1.4527472527472527, "grad_norm": 2.03125, "learning_rate": 8.296358308915865e-05, "loss": 0.2689, "step": 1983 }, { "epoch": 1.4534798534798534, "grad_norm": 1.0234375, "learning_rate": 8.300544160736711e-05, "loss": 0.2428, "step": 1984 }, { "epoch": 1.4542124542124542, "grad_norm": 1.6328125, "learning_rate": 8.304730012557556e-05, "loss": 0.2547, "step": 1985 }, { "epoch": 1.4549450549450549, "grad_norm": 1.1015625, "learning_rate": 8.308915864378401e-05, "loss": 0.2413, "step": 1986 }, { "epoch": 1.4556776556776556, "grad_norm": 0.87109375, "learning_rate": 8.313101716199248e-05, "loss": 0.2081, "step": 1987 }, { "epoch": 1.4564102564102563, "grad_norm": 0.98046875, "learning_rate": 8.317287568020093e-05, "loss": 0.2741, "step": 1988 }, { "epoch": 1.457142857142857, "grad_norm": 1.359375, "learning_rate": 8.321473419840938e-05, "loss": 0.1763, "step": 1989 }, { "epoch": 1.4578754578754578, "grad_norm": 1.2109375, "learning_rate": 8.325659271661783e-05, "loss": 0.217, "step": 1990 }, { "epoch": 1.4586080586080585, "grad_norm": 0.953125, "learning_rate": 8.329845123482629e-05, "loss": 0.1896, "step": 1991 }, { "epoch": 1.4593406593406593, "grad_norm": 1.1171875, "learning_rate": 8.334030975303476e-05, "loss": 0.2516, "step": 1992 }, { "epoch": 1.46007326007326, "grad_norm": 1.015625, "learning_rate": 8.33821682712432e-05, "loss": 0.2439, "step": 1993 }, { "epoch": 1.4608058608058607, "grad_norm": 0.90625, "learning_rate": 8.342402678945166e-05, "loss": 0.2197, "step": 1994 }, { "epoch": 1.4615384615384617, "grad_norm": 0.984375, "learning_rate": 8.346588530766011e-05, "loss": 0.2359, "step": 1995 }, { "epoch": 1.4622710622710624, "grad_norm": 0.8203125, "learning_rate": 8.350774382586857e-05, "loss": 0.1891, "step": 1996 }, { "epoch": 1.4630036630036631, "grad_norm": 1.453125, "learning_rate": 8.354960234407702e-05, "loss": 0.2669, "step": 1997 }, { "epoch": 1.4637362637362639, "grad_norm": 1.1328125, "learning_rate": 8.359146086228547e-05, "loss": 0.1874, "step": 1998 }, { "epoch": 1.4644688644688646, "grad_norm": 1.1953125, "learning_rate": 8.363331938049394e-05, "loss": 0.2255, "step": 1999 }, { "epoch": 1.4652014652014653, "grad_norm": 2.1875, "learning_rate": 8.367517789870239e-05, "loss": 0.1904, "step": 2000 }, { "epoch": 1.465934065934066, "grad_norm": 1.8125, "learning_rate": 8.371703641691084e-05, "loss": 0.2228, "step": 2001 }, { "epoch": 1.4666666666666668, "grad_norm": 1.0625, "learning_rate": 8.37588949351193e-05, "loss": 0.1904, "step": 2002 }, { "epoch": 1.4673992673992675, "grad_norm": 0.94921875, "learning_rate": 8.380075345332775e-05, "loss": 0.1876, "step": 2003 }, { "epoch": 1.4681318681318682, "grad_norm": 1.3203125, "learning_rate": 8.38426119715362e-05, "loss": 0.2245, "step": 2004 }, { "epoch": 1.468864468864469, "grad_norm": 1.6796875, "learning_rate": 8.388447048974467e-05, "loss": 0.2015, "step": 2005 }, { "epoch": 1.4695970695970697, "grad_norm": 1.1953125, "learning_rate": 8.392632900795312e-05, "loss": 0.1734, "step": 2006 }, { "epoch": 1.4703296703296704, "grad_norm": 0.99609375, "learning_rate": 8.396818752616158e-05, "loss": 0.1915, "step": 2007 }, { "epoch": 1.4710622710622712, "grad_norm": 1.234375, "learning_rate": 8.401004604437003e-05, "loss": 0.1886, "step": 2008 }, { "epoch": 1.471794871794872, "grad_norm": 1.2734375, "learning_rate": 8.405190456257849e-05, "loss": 0.1949, "step": 2009 }, { "epoch": 1.4725274725274726, "grad_norm": 1.125, "learning_rate": 8.409376308078695e-05, "loss": 0.211, "step": 2010 }, { "epoch": 1.4732600732600734, "grad_norm": 1.71875, "learning_rate": 8.41356215989954e-05, "loss": 0.2342, "step": 2011 }, { "epoch": 1.473992673992674, "grad_norm": 1.34375, "learning_rate": 8.417748011720385e-05, "loss": 0.2753, "step": 2012 }, { "epoch": 1.4747252747252748, "grad_norm": 1.1953125, "learning_rate": 8.421933863541232e-05, "loss": 0.1941, "step": 2013 }, { "epoch": 1.4754578754578755, "grad_norm": 1.140625, "learning_rate": 8.426119715362077e-05, "loss": 0.2327, "step": 2014 }, { "epoch": 1.4761904761904763, "grad_norm": 1.046875, "learning_rate": 8.430305567182922e-05, "loss": 0.1988, "step": 2015 }, { "epoch": 1.476923076923077, "grad_norm": 1.4296875, "learning_rate": 8.434491419003767e-05, "loss": 0.2317, "step": 2016 }, { "epoch": 1.4776556776556777, "grad_norm": 1.4453125, "learning_rate": 8.438677270824613e-05, "loss": 0.1975, "step": 2017 }, { "epoch": 1.4783882783882785, "grad_norm": 1.328125, "learning_rate": 8.44286312264546e-05, "loss": 0.2373, "step": 2018 }, { "epoch": 1.4791208791208792, "grad_norm": 1.375, "learning_rate": 8.447048974466305e-05, "loss": 0.226, "step": 2019 }, { "epoch": 1.47985347985348, "grad_norm": 1.0234375, "learning_rate": 8.45123482628715e-05, "loss": 0.2422, "step": 2020 }, { "epoch": 1.4805860805860807, "grad_norm": 1.171875, "learning_rate": 8.455420678107996e-05, "loss": 0.245, "step": 2021 }, { "epoch": 1.4813186813186814, "grad_norm": 0.94921875, "learning_rate": 8.459606529928841e-05, "loss": 0.2205, "step": 2022 }, { "epoch": 1.4820512820512821, "grad_norm": 0.9375, "learning_rate": 8.463792381749686e-05, "loss": 0.1913, "step": 2023 }, { "epoch": 1.4827838827838828, "grad_norm": 0.96875, "learning_rate": 8.467978233570531e-05, "loss": 0.2361, "step": 2024 }, { "epoch": 1.4835164835164836, "grad_norm": 1.015625, "learning_rate": 8.472164085391378e-05, "loss": 0.1857, "step": 2025 }, { "epoch": 1.4842490842490843, "grad_norm": 1.1875, "learning_rate": 8.476349937212224e-05, "loss": 0.2224, "step": 2026 }, { "epoch": 1.484981684981685, "grad_norm": 1.2578125, "learning_rate": 8.480535789033068e-05, "loss": 0.1784, "step": 2027 }, { "epoch": 1.4857142857142858, "grad_norm": 1.1328125, "learning_rate": 8.484721640853914e-05, "loss": 0.2143, "step": 2028 }, { "epoch": 1.4864468864468865, "grad_norm": 1.40625, "learning_rate": 8.48890749267476e-05, "loss": 0.2441, "step": 2029 }, { "epoch": 1.4871794871794872, "grad_norm": 0.7734375, "learning_rate": 8.493093344495606e-05, "loss": 0.2053, "step": 2030 }, { "epoch": 1.487912087912088, "grad_norm": 1.515625, "learning_rate": 8.497279196316451e-05, "loss": 0.2339, "step": 2031 }, { "epoch": 1.4886446886446887, "grad_norm": 1.1328125, "learning_rate": 8.501465048137296e-05, "loss": 0.2558, "step": 2032 }, { "epoch": 1.4893772893772894, "grad_norm": 1.09375, "learning_rate": 8.505650899958142e-05, "loss": 0.2397, "step": 2033 }, { "epoch": 1.4901098901098901, "grad_norm": 1.2265625, "learning_rate": 8.509836751778987e-05, "loss": 0.2541, "step": 2034 }, { "epoch": 1.4908424908424909, "grad_norm": 0.87890625, "learning_rate": 8.514022603599832e-05, "loss": 0.1954, "step": 2035 }, { "epoch": 1.4915750915750916, "grad_norm": 1.1640625, "learning_rate": 8.518208455420679e-05, "loss": 0.2035, "step": 2036 }, { "epoch": 1.4923076923076923, "grad_norm": 1.2578125, "learning_rate": 8.522394307241524e-05, "loss": 0.1968, "step": 2037 }, { "epoch": 1.493040293040293, "grad_norm": 1.1015625, "learning_rate": 8.526580159062369e-05, "loss": 0.247, "step": 2038 }, { "epoch": 1.4937728937728938, "grad_norm": 1.5546875, "learning_rate": 8.530766010883215e-05, "loss": 0.2357, "step": 2039 }, { "epoch": 1.4945054945054945, "grad_norm": 1.1796875, "learning_rate": 8.53495186270406e-05, "loss": 0.2169, "step": 2040 }, { "epoch": 1.4952380952380953, "grad_norm": 1.578125, "learning_rate": 8.539137714524907e-05, "loss": 0.2701, "step": 2041 }, { "epoch": 1.495970695970696, "grad_norm": 1.515625, "learning_rate": 8.543323566345752e-05, "loss": 0.1875, "step": 2042 }, { "epoch": 1.4967032967032967, "grad_norm": 1.828125, "learning_rate": 8.547509418166597e-05, "loss": 0.2109, "step": 2043 }, { "epoch": 1.4974358974358974, "grad_norm": 1.796875, "learning_rate": 8.551695269987443e-05, "loss": 0.1949, "step": 2044 }, { "epoch": 1.4981684981684982, "grad_norm": 1.4921875, "learning_rate": 8.555881121808288e-05, "loss": 0.2135, "step": 2045 }, { "epoch": 1.498901098901099, "grad_norm": 0.96484375, "learning_rate": 8.560066973629133e-05, "loss": 0.1886, "step": 2046 }, { "epoch": 1.4996336996336996, "grad_norm": 1.1015625, "learning_rate": 8.56425282544998e-05, "loss": 0.2018, "step": 2047 }, { "epoch": 1.5003663003663004, "grad_norm": 2.03125, "learning_rate": 8.568438677270825e-05, "loss": 0.2515, "step": 2048 }, { "epoch": 1.501098901098901, "grad_norm": 1.328125, "learning_rate": 8.57262452909167e-05, "loss": 0.2108, "step": 2049 }, { "epoch": 1.5018315018315018, "grad_norm": 1.109375, "learning_rate": 8.576810380912516e-05, "loss": 0.2236, "step": 2050 }, { "epoch": 1.5025641025641026, "grad_norm": 0.8515625, "learning_rate": 8.580996232733362e-05, "loss": 0.1954, "step": 2051 }, { "epoch": 1.5032967032967033, "grad_norm": 0.97265625, "learning_rate": 8.585182084554208e-05, "loss": 0.2215, "step": 2052 }, { "epoch": 1.504029304029304, "grad_norm": 1.78125, "learning_rate": 8.589367936375053e-05, "loss": 0.2492, "step": 2053 }, { "epoch": 1.5047619047619047, "grad_norm": 1.4765625, "learning_rate": 8.593553788195898e-05, "loss": 0.2607, "step": 2054 }, { "epoch": 1.5054945054945055, "grad_norm": 1.1875, "learning_rate": 8.597739640016744e-05, "loss": 0.1775, "step": 2055 }, { "epoch": 1.5062271062271062, "grad_norm": 1.0625, "learning_rate": 8.60192549183759e-05, "loss": 0.2556, "step": 2056 }, { "epoch": 1.506959706959707, "grad_norm": 1.1015625, "learning_rate": 8.606111343658435e-05, "loss": 0.1875, "step": 2057 }, { "epoch": 1.5076923076923077, "grad_norm": 1.5859375, "learning_rate": 8.61029719547928e-05, "loss": 0.2289, "step": 2058 }, { "epoch": 1.5084249084249084, "grad_norm": 1.21875, "learning_rate": 8.614483047300126e-05, "loss": 0.2174, "step": 2059 }, { "epoch": 1.5091575091575091, "grad_norm": 1.21875, "learning_rate": 8.618668899120973e-05, "loss": 0.1969, "step": 2060 }, { "epoch": 1.5098901098901099, "grad_norm": 1.375, "learning_rate": 8.622854750941816e-05, "loss": 0.2255, "step": 2061 }, { "epoch": 1.5106227106227106, "grad_norm": 1.1796875, "learning_rate": 8.627040602762663e-05, "loss": 0.2235, "step": 2062 }, { "epoch": 1.5113553113553113, "grad_norm": 1.0078125, "learning_rate": 8.631226454583509e-05, "loss": 0.1861, "step": 2063 }, { "epoch": 1.512087912087912, "grad_norm": 0.68359375, "learning_rate": 8.635412306404354e-05, "loss": 0.1699, "step": 2064 }, { "epoch": 1.5128205128205128, "grad_norm": 1.078125, "learning_rate": 8.639598158225199e-05, "loss": 0.1924, "step": 2065 }, { "epoch": 1.5135531135531135, "grad_norm": 0.72265625, "learning_rate": 8.643784010046044e-05, "loss": 0.182, "step": 2066 }, { "epoch": 1.5142857142857142, "grad_norm": 1.0625, "learning_rate": 8.64796986186689e-05, "loss": 0.2153, "step": 2067 }, { "epoch": 1.515018315018315, "grad_norm": 1.234375, "learning_rate": 8.652155713687736e-05, "loss": 0.2559, "step": 2068 }, { "epoch": 1.5157509157509157, "grad_norm": 1.09375, "learning_rate": 8.656341565508581e-05, "loss": 0.192, "step": 2069 }, { "epoch": 1.5164835164835164, "grad_norm": 1.6875, "learning_rate": 8.660527417329427e-05, "loss": 0.1991, "step": 2070 }, { "epoch": 1.5172161172161172, "grad_norm": 1.171875, "learning_rate": 8.664713269150274e-05, "loss": 0.1876, "step": 2071 }, { "epoch": 1.5179487179487179, "grad_norm": 0.8046875, "learning_rate": 8.668899120971117e-05, "loss": 0.1692, "step": 2072 }, { "epoch": 1.5186813186813186, "grad_norm": 1.28125, "learning_rate": 8.673084972791964e-05, "loss": 0.1917, "step": 2073 }, { "epoch": 1.5194139194139193, "grad_norm": 1.1953125, "learning_rate": 8.677270824612809e-05, "loss": 0.2723, "step": 2074 }, { "epoch": 1.52014652014652, "grad_norm": 1.046875, "learning_rate": 8.681456676433655e-05, "loss": 0.2344, "step": 2075 }, { "epoch": 1.5208791208791208, "grad_norm": 0.83984375, "learning_rate": 8.6856425282545e-05, "loss": 0.1838, "step": 2076 }, { "epoch": 1.5216117216117215, "grad_norm": 1.21875, "learning_rate": 8.689828380075345e-05, "loss": 0.221, "step": 2077 }, { "epoch": 1.5223443223443223, "grad_norm": 0.8515625, "learning_rate": 8.694014231896192e-05, "loss": 0.1648, "step": 2078 }, { "epoch": 1.523076923076923, "grad_norm": 1.359375, "learning_rate": 8.698200083717037e-05, "loss": 0.2119, "step": 2079 }, { "epoch": 1.5238095238095237, "grad_norm": 1.2890625, "learning_rate": 8.702385935537882e-05, "loss": 0.1958, "step": 2080 }, { "epoch": 1.5245421245421245, "grad_norm": 1.0, "learning_rate": 8.706571787358728e-05, "loss": 0.2182, "step": 2081 }, { "epoch": 1.5252747252747252, "grad_norm": 1.1953125, "learning_rate": 8.710757639179573e-05, "loss": 0.2473, "step": 2082 }, { "epoch": 1.526007326007326, "grad_norm": 1.1640625, "learning_rate": 8.714943491000418e-05, "loss": 0.1877, "step": 2083 }, { "epoch": 1.5267399267399266, "grad_norm": 1.9453125, "learning_rate": 8.719129342821265e-05, "loss": 0.1821, "step": 2084 }, { "epoch": 1.5274725274725274, "grad_norm": 1.234375, "learning_rate": 8.72331519464211e-05, "loss": 0.2073, "step": 2085 }, { "epoch": 1.528205128205128, "grad_norm": 1.078125, "learning_rate": 8.727501046462956e-05, "loss": 0.262, "step": 2086 }, { "epoch": 1.5289377289377288, "grad_norm": 1.609375, "learning_rate": 8.7316868982838e-05, "loss": 0.2134, "step": 2087 }, { "epoch": 1.5296703296703296, "grad_norm": 1.6796875, "learning_rate": 8.735872750104646e-05, "loss": 0.1904, "step": 2088 }, { "epoch": 1.5304029304029303, "grad_norm": 1.4765625, "learning_rate": 8.740058601925493e-05, "loss": 0.1696, "step": 2089 }, { "epoch": 1.531135531135531, "grad_norm": 1.0625, "learning_rate": 8.744244453746338e-05, "loss": 0.237, "step": 2090 }, { "epoch": 1.5318681318681318, "grad_norm": 1.1328125, "learning_rate": 8.748430305567183e-05, "loss": 0.1887, "step": 2091 }, { "epoch": 1.5326007326007325, "grad_norm": 2.09375, "learning_rate": 8.75261615738803e-05, "loss": 0.2216, "step": 2092 }, { "epoch": 1.5333333333333332, "grad_norm": 1.5625, "learning_rate": 8.756802009208874e-05, "loss": 0.2379, "step": 2093 }, { "epoch": 1.534065934065934, "grad_norm": 1.28125, "learning_rate": 8.760987861029721e-05, "loss": 0.2577, "step": 2094 }, { "epoch": 1.5347985347985347, "grad_norm": 0.9609375, "learning_rate": 8.765173712850565e-05, "loss": 0.2028, "step": 2095 }, { "epoch": 1.5355311355311354, "grad_norm": 1.3125, "learning_rate": 8.769359564671411e-05, "loss": 0.1958, "step": 2096 }, { "epoch": 1.5362637362637361, "grad_norm": 2.375, "learning_rate": 8.773545416492257e-05, "loss": 0.2226, "step": 2097 }, { "epoch": 1.5369963369963369, "grad_norm": 1.703125, "learning_rate": 8.777731268313103e-05, "loss": 0.1863, "step": 2098 }, { "epoch": 1.5377289377289376, "grad_norm": 1.0390625, "learning_rate": 8.781917120133948e-05, "loss": 0.2552, "step": 2099 }, { "epoch": 1.5384615384615383, "grad_norm": 1.046875, "learning_rate": 8.786102971954793e-05, "loss": 0.1611, "step": 2100 }, { "epoch": 1.539194139194139, "grad_norm": 1.1953125, "learning_rate": 8.790288823775639e-05, "loss": 0.2634, "step": 2101 }, { "epoch": 1.5399267399267398, "grad_norm": 0.87890625, "learning_rate": 8.794474675596484e-05, "loss": 0.1929, "step": 2102 }, { "epoch": 1.5406593406593405, "grad_norm": 0.93359375, "learning_rate": 8.798660527417329e-05, "loss": 0.2073, "step": 2103 }, { "epoch": 1.5413919413919412, "grad_norm": 1.0703125, "learning_rate": 8.802846379238176e-05, "loss": 0.1943, "step": 2104 }, { "epoch": 1.542124542124542, "grad_norm": 1.0625, "learning_rate": 8.807032231059022e-05, "loss": 0.1886, "step": 2105 }, { "epoch": 1.5428571428571427, "grad_norm": 0.79296875, "learning_rate": 8.811218082879866e-05, "loss": 0.1678, "step": 2106 }, { "epoch": 1.5435897435897434, "grad_norm": 1.8828125, "learning_rate": 8.815403934700712e-05, "loss": 0.2655, "step": 2107 }, { "epoch": 1.5443223443223442, "grad_norm": 1.0859375, "learning_rate": 8.819589786521557e-05, "loss": 0.1854, "step": 2108 }, { "epoch": 1.545054945054945, "grad_norm": 1.140625, "learning_rate": 8.823775638342404e-05, "loss": 0.2306, "step": 2109 }, { "epoch": 1.5457875457875456, "grad_norm": 1.0703125, "learning_rate": 8.827961490163249e-05, "loss": 0.1643, "step": 2110 }, { "epoch": 1.5465201465201464, "grad_norm": 1.2890625, "learning_rate": 8.832147341984094e-05, "loss": 0.2598, "step": 2111 }, { "epoch": 1.5472527472527473, "grad_norm": 0.97265625, "learning_rate": 8.83633319380494e-05, "loss": 0.1767, "step": 2112 }, { "epoch": 1.547985347985348, "grad_norm": 0.91796875, "learning_rate": 8.840519045625785e-05, "loss": 0.1692, "step": 2113 }, { "epoch": 1.5487179487179488, "grad_norm": 1.0703125, "learning_rate": 8.84470489744663e-05, "loss": 0.2095, "step": 2114 }, { "epoch": 1.5494505494505495, "grad_norm": 1.1015625, "learning_rate": 8.848890749267477e-05, "loss": 0.2319, "step": 2115 }, { "epoch": 1.5501831501831502, "grad_norm": 1.0390625, "learning_rate": 8.853076601088322e-05, "loss": 0.2178, "step": 2116 }, { "epoch": 1.550915750915751, "grad_norm": 1.015625, "learning_rate": 8.857262452909167e-05, "loss": 0.2074, "step": 2117 }, { "epoch": 1.5516483516483517, "grad_norm": 1.0078125, "learning_rate": 8.861448304730013e-05, "loss": 0.1933, "step": 2118 }, { "epoch": 1.5523809523809524, "grad_norm": 0.96484375, "learning_rate": 8.865634156550858e-05, "loss": 0.2317, "step": 2119 }, { "epoch": 1.5531135531135531, "grad_norm": 0.91015625, "learning_rate": 8.869820008371705e-05, "loss": 0.2091, "step": 2120 }, { "epoch": 1.5538461538461539, "grad_norm": 1.34375, "learning_rate": 8.874005860192548e-05, "loss": 0.2029, "step": 2121 }, { "epoch": 1.5545787545787546, "grad_norm": 1.421875, "learning_rate": 8.878191712013395e-05, "loss": 0.2254, "step": 2122 }, { "epoch": 1.5553113553113553, "grad_norm": 1.3515625, "learning_rate": 8.882377563834241e-05, "loss": 0.2505, "step": 2123 }, { "epoch": 1.556043956043956, "grad_norm": 1.0390625, "learning_rate": 8.886563415655086e-05, "loss": 0.2061, "step": 2124 }, { "epoch": 1.5567765567765568, "grad_norm": 1.609375, "learning_rate": 8.890749267475931e-05, "loss": 0.2702, "step": 2125 }, { "epoch": 1.5575091575091575, "grad_norm": 1.46875, "learning_rate": 8.894935119296778e-05, "loss": 0.2188, "step": 2126 }, { "epoch": 1.5582417582417583, "grad_norm": 1.78125, "learning_rate": 8.899120971117623e-05, "loss": 0.2106, "step": 2127 }, { "epoch": 1.558974358974359, "grad_norm": 1.265625, "learning_rate": 8.903306822938468e-05, "loss": 0.1957, "step": 2128 }, { "epoch": 1.5597069597069597, "grad_norm": 0.78125, "learning_rate": 8.907492674759313e-05, "loss": 0.1946, "step": 2129 }, { "epoch": 1.5604395604395604, "grad_norm": 1.0625, "learning_rate": 8.91167852658016e-05, "loss": 0.1561, "step": 2130 }, { "epoch": 1.5611721611721612, "grad_norm": 2.0625, "learning_rate": 8.915864378401006e-05, "loss": 0.2572, "step": 2131 }, { "epoch": 1.561904761904762, "grad_norm": 1.34375, "learning_rate": 8.920050230221851e-05, "loss": 0.2339, "step": 2132 }, { "epoch": 1.5626373626373626, "grad_norm": 1.15625, "learning_rate": 8.924236082042696e-05, "loss": 0.2235, "step": 2133 }, { "epoch": 1.5633699633699634, "grad_norm": 1.1171875, "learning_rate": 8.928421933863542e-05, "loss": 0.1805, "step": 2134 }, { "epoch": 1.564102564102564, "grad_norm": 1.203125, "learning_rate": 8.932607785684387e-05, "loss": 0.2349, "step": 2135 }, { "epoch": 1.5648351648351648, "grad_norm": 0.94921875, "learning_rate": 8.936793637505233e-05, "loss": 0.2106, "step": 2136 }, { "epoch": 1.5655677655677656, "grad_norm": 1.4296875, "learning_rate": 8.940979489326078e-05, "loss": 0.2089, "step": 2137 }, { "epoch": 1.5663003663003663, "grad_norm": 1.2734375, "learning_rate": 8.945165341146924e-05, "loss": 0.2365, "step": 2138 }, { "epoch": 1.567032967032967, "grad_norm": 0.82421875, "learning_rate": 8.94935119296777e-05, "loss": 0.2201, "step": 2139 }, { "epoch": 1.5677655677655677, "grad_norm": 1.015625, "learning_rate": 8.953537044788614e-05, "loss": 0.2268, "step": 2140 }, { "epoch": 1.5684981684981685, "grad_norm": 1.6484375, "learning_rate": 8.95772289660946e-05, "loss": 0.3076, "step": 2141 }, { "epoch": 1.5692307692307692, "grad_norm": 1.390625, "learning_rate": 8.961908748430306e-05, "loss": 0.2421, "step": 2142 }, { "epoch": 1.56996336996337, "grad_norm": 2.09375, "learning_rate": 8.966094600251152e-05, "loss": 0.2765, "step": 2143 }, { "epoch": 1.5706959706959707, "grad_norm": 1.3984375, "learning_rate": 8.970280452071997e-05, "loss": 0.2185, "step": 2144 }, { "epoch": 1.5714285714285714, "grad_norm": 1.578125, "learning_rate": 8.974466303892842e-05, "loss": 0.3013, "step": 2145 }, { "epoch": 1.5721611721611721, "grad_norm": 2.125, "learning_rate": 8.978652155713689e-05, "loss": 0.1877, "step": 2146 }, { "epoch": 1.5728937728937729, "grad_norm": 1.65625, "learning_rate": 8.982838007534534e-05, "loss": 0.2746, "step": 2147 }, { "epoch": 1.5736263736263736, "grad_norm": 1.5859375, "learning_rate": 8.987023859355379e-05, "loss": 0.1683, "step": 2148 }, { "epoch": 1.5743589743589743, "grad_norm": 1.03125, "learning_rate": 8.991209711176225e-05, "loss": 0.1965, "step": 2149 }, { "epoch": 1.575091575091575, "grad_norm": 1.0703125, "learning_rate": 8.99539556299707e-05, "loss": 0.2226, "step": 2150 }, { "epoch": 1.5758241758241758, "grad_norm": 1.6875, "learning_rate": 8.999581414817915e-05, "loss": 0.2207, "step": 2151 }, { "epoch": 1.5765567765567765, "grad_norm": 1.8671875, "learning_rate": 9.003767266638762e-05, "loss": 0.1723, "step": 2152 }, { "epoch": 1.5772893772893772, "grad_norm": 1.5625, "learning_rate": 9.007953118459607e-05, "loss": 0.2326, "step": 2153 }, { "epoch": 1.578021978021978, "grad_norm": 1.1953125, "learning_rate": 9.012138970280453e-05, "loss": 0.2009, "step": 2154 }, { "epoch": 1.578754578754579, "grad_norm": 1.8046875, "learning_rate": 9.016324822101298e-05, "loss": 0.2802, "step": 2155 }, { "epoch": 1.5794871794871796, "grad_norm": 2.40625, "learning_rate": 9.020510673922143e-05, "loss": 0.2019, "step": 2156 }, { "epoch": 1.5802197802197804, "grad_norm": 0.98828125, "learning_rate": 9.02469652574299e-05, "loss": 0.2058, "step": 2157 }, { "epoch": 1.580952380952381, "grad_norm": 0.90234375, "learning_rate": 9.028882377563835e-05, "loss": 0.2007, "step": 2158 }, { "epoch": 1.5816849816849818, "grad_norm": 1.5625, "learning_rate": 9.03306822938468e-05, "loss": 0.2224, "step": 2159 }, { "epoch": 1.5824175824175826, "grad_norm": 1.078125, "learning_rate": 9.037254081205526e-05, "loss": 0.1948, "step": 2160 }, { "epoch": 1.5831501831501833, "grad_norm": 1.3203125, "learning_rate": 9.041439933026371e-05, "loss": 0.2259, "step": 2161 }, { "epoch": 1.583882783882784, "grad_norm": 1.3125, "learning_rate": 9.045625784847216e-05, "loss": 0.1881, "step": 2162 }, { "epoch": 1.5846153846153848, "grad_norm": 1.6171875, "learning_rate": 9.049811636668061e-05, "loss": 0.2156, "step": 2163 }, { "epoch": 1.5853479853479855, "grad_norm": 1.328125, "learning_rate": 9.053997488488908e-05, "loss": 0.1811, "step": 2164 }, { "epoch": 1.5860805860805862, "grad_norm": 0.8515625, "learning_rate": 9.058183340309754e-05, "loss": 0.2164, "step": 2165 }, { "epoch": 1.586813186813187, "grad_norm": 1.2421875, "learning_rate": 9.0623691921306e-05, "loss": 0.2082, "step": 2166 }, { "epoch": 1.5875457875457877, "grad_norm": 0.88671875, "learning_rate": 9.066555043951444e-05, "loss": 0.1848, "step": 2167 }, { "epoch": 1.5882783882783884, "grad_norm": 1.046875, "learning_rate": 9.070740895772291e-05, "loss": 0.1928, "step": 2168 }, { "epoch": 1.5890109890109891, "grad_norm": 0.9453125, "learning_rate": 9.074926747593136e-05, "loss": 0.1938, "step": 2169 }, { "epoch": 1.5897435897435899, "grad_norm": 1.0703125, "learning_rate": 9.079112599413981e-05, "loss": 0.2023, "step": 2170 }, { "epoch": 1.5904761904761906, "grad_norm": 1.109375, "learning_rate": 9.083298451234826e-05, "loss": 0.2049, "step": 2171 }, { "epoch": 1.5912087912087913, "grad_norm": 1.0859375, "learning_rate": 9.087484303055672e-05, "loss": 0.2122, "step": 2172 }, { "epoch": 1.591941391941392, "grad_norm": 1.3359375, "learning_rate": 9.091670154876519e-05, "loss": 0.2246, "step": 2173 }, { "epoch": 1.5926739926739928, "grad_norm": 1.71875, "learning_rate": 9.095856006697363e-05, "loss": 0.1916, "step": 2174 }, { "epoch": 1.5934065934065935, "grad_norm": 0.83203125, "learning_rate": 9.100041858518209e-05, "loss": 0.2077, "step": 2175 }, { "epoch": 1.5941391941391942, "grad_norm": 0.98828125, "learning_rate": 9.104227710339055e-05, "loss": 0.2296, "step": 2176 }, { "epoch": 1.594871794871795, "grad_norm": 0.8359375, "learning_rate": 9.1084135621599e-05, "loss": 0.1967, "step": 2177 }, { "epoch": 1.5956043956043957, "grad_norm": 1.671875, "learning_rate": 9.112599413980745e-05, "loss": 0.2615, "step": 2178 }, { "epoch": 1.5963369963369964, "grad_norm": 0.9765625, "learning_rate": 9.11678526580159e-05, "loss": 0.1949, "step": 2179 }, { "epoch": 1.5970695970695972, "grad_norm": 1.4609375, "learning_rate": 9.120971117622437e-05, "loss": 0.3052, "step": 2180 }, { "epoch": 1.597802197802198, "grad_norm": 0.83984375, "learning_rate": 9.125156969443282e-05, "loss": 0.1808, "step": 2181 }, { "epoch": 1.5985347985347986, "grad_norm": 1.1640625, "learning_rate": 9.129342821264127e-05, "loss": 0.2163, "step": 2182 }, { "epoch": 1.5992673992673994, "grad_norm": 1.1328125, "learning_rate": 9.133528673084974e-05, "loss": 0.1992, "step": 2183 }, { "epoch": 1.6, "grad_norm": 1.1484375, "learning_rate": 9.137714524905819e-05, "loss": 0.2124, "step": 2184 }, { "epoch": 1.6007326007326008, "grad_norm": 1.9375, "learning_rate": 9.141900376726664e-05, "loss": 0.1912, "step": 2185 }, { "epoch": 1.6014652014652015, "grad_norm": 1.4453125, "learning_rate": 9.14608622854751e-05, "loss": 0.223, "step": 2186 }, { "epoch": 1.6021978021978023, "grad_norm": 1.125, "learning_rate": 9.150272080368355e-05, "loss": 0.2449, "step": 2187 }, { "epoch": 1.602930402930403, "grad_norm": 1.8359375, "learning_rate": 9.154457932189202e-05, "loss": 0.1743, "step": 2188 }, { "epoch": 1.6036630036630037, "grad_norm": 1.8515625, "learning_rate": 9.158643784010047e-05, "loss": 0.1965, "step": 2189 }, { "epoch": 1.6043956043956045, "grad_norm": 1.5234375, "learning_rate": 9.162829635830892e-05, "loss": 0.2191, "step": 2190 }, { "epoch": 1.6051282051282052, "grad_norm": 1.453125, "learning_rate": 9.167015487651738e-05, "loss": 0.1896, "step": 2191 }, { "epoch": 1.605860805860806, "grad_norm": 2.09375, "learning_rate": 9.171201339472583e-05, "loss": 0.2242, "step": 2192 }, { "epoch": 1.6065934065934067, "grad_norm": 1.203125, "learning_rate": 9.175387191293428e-05, "loss": 0.2072, "step": 2193 }, { "epoch": 1.6073260073260074, "grad_norm": 1.25, "learning_rate": 9.179573043114275e-05, "loss": 0.2242, "step": 2194 }, { "epoch": 1.6080586080586081, "grad_norm": 0.90625, "learning_rate": 9.18375889493512e-05, "loss": 0.2224, "step": 2195 }, { "epoch": 1.6087912087912088, "grad_norm": 1.7578125, "learning_rate": 9.187944746755965e-05, "loss": 0.2133, "step": 2196 }, { "epoch": 1.6095238095238096, "grad_norm": 2.703125, "learning_rate": 9.192130598576811e-05, "loss": 0.2169, "step": 2197 }, { "epoch": 1.6102564102564103, "grad_norm": 2.046875, "learning_rate": 9.196316450397656e-05, "loss": 0.1956, "step": 2198 }, { "epoch": 1.610989010989011, "grad_norm": 1.0703125, "learning_rate": 9.200502302218503e-05, "loss": 0.1831, "step": 2199 }, { "epoch": 1.6117216117216118, "grad_norm": 0.984375, "learning_rate": 9.204688154039346e-05, "loss": 0.2119, "step": 2200 }, { "epoch": 1.6124542124542125, "grad_norm": 1.5859375, "learning_rate": 9.208874005860193e-05, "loss": 0.2185, "step": 2201 }, { "epoch": 1.6131868131868132, "grad_norm": 1.7578125, "learning_rate": 9.213059857681039e-05, "loss": 0.2224, "step": 2202 }, { "epoch": 1.613919413919414, "grad_norm": 2.453125, "learning_rate": 9.217245709501884e-05, "loss": 0.1873, "step": 2203 }, { "epoch": 1.6146520146520147, "grad_norm": 2.140625, "learning_rate": 9.22143156132273e-05, "loss": 0.2226, "step": 2204 }, { "epoch": 1.6153846153846154, "grad_norm": 1.359375, "learning_rate": 9.225617413143574e-05, "loss": 0.2055, "step": 2205 }, { "epoch": 1.6161172161172161, "grad_norm": 1.6875, "learning_rate": 9.229803264964421e-05, "loss": 0.2529, "step": 2206 }, { "epoch": 1.6168498168498169, "grad_norm": 2.09375, "learning_rate": 9.233989116785267e-05, "loss": 0.212, "step": 2207 }, { "epoch": 1.6175824175824176, "grad_norm": 1.25, "learning_rate": 9.238174968606111e-05, "loss": 0.1891, "step": 2208 }, { "epoch": 1.6183150183150183, "grad_norm": 0.83203125, "learning_rate": 9.242360820426957e-05, "loss": 0.1989, "step": 2209 }, { "epoch": 1.619047619047619, "grad_norm": 1.59375, "learning_rate": 9.246546672247804e-05, "loss": 0.2464, "step": 2210 }, { "epoch": 1.6197802197802198, "grad_norm": 2.1875, "learning_rate": 9.250732524068649e-05, "loss": 0.2122, "step": 2211 }, { "epoch": 1.6205128205128205, "grad_norm": 1.828125, "learning_rate": 9.254918375889494e-05, "loss": 0.2249, "step": 2212 }, { "epoch": 1.6212454212454213, "grad_norm": 2.125, "learning_rate": 9.259104227710339e-05, "loss": 0.2069, "step": 2213 }, { "epoch": 1.621978021978022, "grad_norm": 0.90234375, "learning_rate": 9.263290079531185e-05, "loss": 0.2129, "step": 2214 }, { "epoch": 1.6227106227106227, "grad_norm": 1.2578125, "learning_rate": 9.26747593135203e-05, "loss": 0.1794, "step": 2215 }, { "epoch": 1.6234432234432234, "grad_norm": 1.46875, "learning_rate": 9.271661783172876e-05, "loss": 0.1865, "step": 2216 }, { "epoch": 1.6241758241758242, "grad_norm": 1.875, "learning_rate": 9.275847634993722e-05, "loss": 0.2048, "step": 2217 }, { "epoch": 1.624908424908425, "grad_norm": 1.40625, "learning_rate": 9.280033486814568e-05, "loss": 0.2453, "step": 2218 }, { "epoch": 1.6256410256410256, "grad_norm": 1.265625, "learning_rate": 9.284219338635412e-05, "loss": 0.2311, "step": 2219 }, { "epoch": 1.6263736263736264, "grad_norm": 0.95703125, "learning_rate": 9.288405190456258e-05, "loss": 0.1813, "step": 2220 }, { "epoch": 1.627106227106227, "grad_norm": 1.3828125, "learning_rate": 9.292591042277104e-05, "loss": 0.2359, "step": 2221 }, { "epoch": 1.6278388278388278, "grad_norm": 1.0703125, "learning_rate": 9.29677689409795e-05, "loss": 0.2111, "step": 2222 }, { "epoch": 1.6285714285714286, "grad_norm": 0.80078125, "learning_rate": 9.300962745918795e-05, "loss": 0.1553, "step": 2223 }, { "epoch": 1.6293040293040293, "grad_norm": 1.34375, "learning_rate": 9.30514859773964e-05, "loss": 0.1924, "step": 2224 }, { "epoch": 1.63003663003663, "grad_norm": 1.6796875, "learning_rate": 9.309334449560486e-05, "loss": 0.2041, "step": 2225 }, { "epoch": 1.6307692307692307, "grad_norm": 1.0703125, "learning_rate": 9.313520301381332e-05, "loss": 0.1921, "step": 2226 }, { "epoch": 1.6315018315018315, "grad_norm": 0.86328125, "learning_rate": 9.317706153202177e-05, "loss": 0.2193, "step": 2227 }, { "epoch": 1.6322344322344322, "grad_norm": 1.109375, "learning_rate": 9.321892005023023e-05, "loss": 0.2095, "step": 2228 }, { "epoch": 1.632967032967033, "grad_norm": 1.1328125, "learning_rate": 9.326077856843868e-05, "loss": 0.2296, "step": 2229 }, { "epoch": 1.6336996336996337, "grad_norm": 1.0703125, "learning_rate": 9.330263708664713e-05, "loss": 0.2194, "step": 2230 }, { "epoch": 1.6344322344322344, "grad_norm": 1.578125, "learning_rate": 9.33444956048556e-05, "loss": 0.2273, "step": 2231 }, { "epoch": 1.6351648351648351, "grad_norm": 1.1015625, "learning_rate": 9.338635412306405e-05, "loss": 0.1704, "step": 2232 }, { "epoch": 1.6358974358974359, "grad_norm": 1.59375, "learning_rate": 9.342821264127251e-05, "loss": 0.1921, "step": 2233 }, { "epoch": 1.6366300366300366, "grad_norm": 1.1640625, "learning_rate": 9.347007115948095e-05, "loss": 0.1964, "step": 2234 }, { "epoch": 1.6373626373626373, "grad_norm": 1.328125, "learning_rate": 9.351192967768941e-05, "loss": 0.2241, "step": 2235 }, { "epoch": 1.638095238095238, "grad_norm": 1.4140625, "learning_rate": 9.355378819589788e-05, "loss": 0.2056, "step": 2236 }, { "epoch": 1.6388278388278388, "grad_norm": 1.40625, "learning_rate": 9.359564671410633e-05, "loss": 0.2347, "step": 2237 }, { "epoch": 1.6395604395604395, "grad_norm": 1.3046875, "learning_rate": 9.363750523231478e-05, "loss": 0.1693, "step": 2238 }, { "epoch": 1.6402930402930402, "grad_norm": 0.9296875, "learning_rate": 9.367936375052324e-05, "loss": 0.1603, "step": 2239 }, { "epoch": 1.641025641025641, "grad_norm": 0.78125, "learning_rate": 9.372122226873169e-05, "loss": 0.1826, "step": 2240 }, { "epoch": 1.6417582417582417, "grad_norm": 1.1796875, "learning_rate": 9.376308078694014e-05, "loss": 0.2029, "step": 2241 }, { "epoch": 1.6424908424908424, "grad_norm": 1.109375, "learning_rate": 9.38049393051486e-05, "loss": 0.1522, "step": 2242 }, { "epoch": 1.6432234432234432, "grad_norm": 1.0703125, "learning_rate": 9.384679782335706e-05, "loss": 0.1676, "step": 2243 }, { "epoch": 1.6439560439560439, "grad_norm": 1.8203125, "learning_rate": 9.388865634156552e-05, "loss": 0.2397, "step": 2244 }, { "epoch": 1.6446886446886446, "grad_norm": 1.0703125, "learning_rate": 9.393051485977397e-05, "loss": 0.1861, "step": 2245 }, { "epoch": 1.6454212454212453, "grad_norm": 1.1640625, "learning_rate": 9.397237337798242e-05, "loss": 0.2622, "step": 2246 }, { "epoch": 1.646153846153846, "grad_norm": 2.09375, "learning_rate": 9.401423189619087e-05, "loss": 0.2575, "step": 2247 }, { "epoch": 1.6468864468864468, "grad_norm": 2.125, "learning_rate": 9.405609041439934e-05, "loss": 0.2678, "step": 2248 }, { "epoch": 1.6476190476190475, "grad_norm": 1.75, "learning_rate": 9.409794893260779e-05, "loss": 0.1728, "step": 2249 }, { "epoch": 1.6483516483516483, "grad_norm": 2.359375, "learning_rate": 9.413980745081624e-05, "loss": 0.1925, "step": 2250 }, { "epoch": 1.649084249084249, "grad_norm": 0.82421875, "learning_rate": 9.41816659690247e-05, "loss": 0.2141, "step": 2251 }, { "epoch": 1.6498168498168497, "grad_norm": 1.9375, "learning_rate": 9.422352448723317e-05, "loss": 0.245, "step": 2252 }, { "epoch": 1.6505494505494505, "grad_norm": 2.203125, "learning_rate": 9.42653830054416e-05, "loss": 0.1897, "step": 2253 }, { "epoch": 1.6512820512820512, "grad_norm": 1.734375, "learning_rate": 9.430724152365007e-05, "loss": 0.2058, "step": 2254 }, { "epoch": 1.652014652014652, "grad_norm": 0.9609375, "learning_rate": 9.434910004185852e-05, "loss": 0.1959, "step": 2255 }, { "epoch": 1.6527472527472526, "grad_norm": 1.2421875, "learning_rate": 9.439095856006698e-05, "loss": 0.218, "step": 2256 }, { "epoch": 1.6534798534798534, "grad_norm": 0.8984375, "learning_rate": 9.443281707827543e-05, "loss": 0.2121, "step": 2257 }, { "epoch": 1.654212454212454, "grad_norm": 1.59375, "learning_rate": 9.447467559648388e-05, "loss": 0.2237, "step": 2258 }, { "epoch": 1.6549450549450548, "grad_norm": 1.734375, "learning_rate": 9.451653411469235e-05, "loss": 0.2408, "step": 2259 }, { "epoch": 1.6556776556776556, "grad_norm": 0.9140625, "learning_rate": 9.45583926329008e-05, "loss": 0.1925, "step": 2260 }, { "epoch": 1.6564102564102563, "grad_norm": 1.0546875, "learning_rate": 9.460025115110925e-05, "loss": 0.1858, "step": 2261 }, { "epoch": 1.657142857142857, "grad_norm": 2.28125, "learning_rate": 9.464210966931771e-05, "loss": 0.2086, "step": 2262 }, { "epoch": 1.6578754578754578, "grad_norm": 1.0, "learning_rate": 9.468396818752617e-05, "loss": 0.2273, "step": 2263 }, { "epoch": 1.6586080586080585, "grad_norm": 1.1015625, "learning_rate": 9.472582670573462e-05, "loss": 0.2253, "step": 2264 }, { "epoch": 1.6593406593406592, "grad_norm": 1.046875, "learning_rate": 9.476768522394308e-05, "loss": 0.2048, "step": 2265 }, { "epoch": 1.66007326007326, "grad_norm": 1.03125, "learning_rate": 9.480954374215153e-05, "loss": 0.1958, "step": 2266 }, { "epoch": 1.6608058608058607, "grad_norm": 0.6796875, "learning_rate": 9.485140226036e-05, "loss": 0.1634, "step": 2267 }, { "epoch": 1.6615384615384614, "grad_norm": 0.86328125, "learning_rate": 9.489326077856843e-05, "loss": 0.1918, "step": 2268 }, { "epoch": 1.6622710622710621, "grad_norm": 1.3671875, "learning_rate": 9.49351192967769e-05, "loss": 0.2269, "step": 2269 }, { "epoch": 1.6630036630036629, "grad_norm": 0.9296875, "learning_rate": 9.497697781498536e-05, "loss": 0.2081, "step": 2270 }, { "epoch": 1.6637362637362636, "grad_norm": 0.8125, "learning_rate": 9.501883633319381e-05, "loss": 0.2851, "step": 2271 }, { "epoch": 1.6644688644688643, "grad_norm": 0.85546875, "learning_rate": 9.506069485140226e-05, "loss": 0.2118, "step": 2272 }, { "epoch": 1.665201465201465, "grad_norm": 1.9921875, "learning_rate": 9.510255336961073e-05, "loss": 0.2321, "step": 2273 }, { "epoch": 1.6659340659340658, "grad_norm": 1.6953125, "learning_rate": 9.514441188781918e-05, "loss": 0.2896, "step": 2274 }, { "epoch": 1.6666666666666665, "grad_norm": 0.96875, "learning_rate": 9.518627040602763e-05, "loss": 0.1973, "step": 2275 }, { "epoch": 1.6673992673992672, "grad_norm": 0.984375, "learning_rate": 9.522812892423608e-05, "loss": 0.1893, "step": 2276 }, { "epoch": 1.668131868131868, "grad_norm": 1.65625, "learning_rate": 9.526998744244454e-05, "loss": 0.1872, "step": 2277 }, { "epoch": 1.6688644688644687, "grad_norm": 1.484375, "learning_rate": 9.5311845960653e-05, "loss": 0.1996, "step": 2278 }, { "epoch": 1.6695970695970694, "grad_norm": 1.2109375, "learning_rate": 9.535370447886146e-05, "loss": 0.2442, "step": 2279 }, { "epoch": 1.6703296703296702, "grad_norm": 1.453125, "learning_rate": 9.539556299706991e-05, "loss": 0.2163, "step": 2280 }, { "epoch": 1.671062271062271, "grad_norm": 1.25, "learning_rate": 9.543742151527837e-05, "loss": 0.2093, "step": 2281 }, { "epoch": 1.6717948717948716, "grad_norm": 1.0625, "learning_rate": 9.547928003348682e-05, "loss": 0.1827, "step": 2282 }, { "epoch": 1.6725274725274726, "grad_norm": 1.2109375, "learning_rate": 9.552113855169527e-05, "loss": 0.2141, "step": 2283 }, { "epoch": 1.6732600732600733, "grad_norm": 1.015625, "learning_rate": 9.556299706990372e-05, "loss": 0.1584, "step": 2284 }, { "epoch": 1.673992673992674, "grad_norm": 1.0390625, "learning_rate": 9.560485558811219e-05, "loss": 0.1804, "step": 2285 }, { "epoch": 1.6747252747252748, "grad_norm": 1.75, "learning_rate": 9.564671410632065e-05, "loss": 0.2455, "step": 2286 }, { "epoch": 1.6754578754578755, "grad_norm": 1.2890625, "learning_rate": 9.568857262452909e-05, "loss": 0.2333, "step": 2287 }, { "epoch": 1.6761904761904762, "grad_norm": 1.1484375, "learning_rate": 9.573043114273755e-05, "loss": 0.1723, "step": 2288 }, { "epoch": 1.676923076923077, "grad_norm": 1.4921875, "learning_rate": 9.5772289660946e-05, "loss": 0.2324, "step": 2289 }, { "epoch": 1.6776556776556777, "grad_norm": 0.7890625, "learning_rate": 9.581414817915447e-05, "loss": 0.1567, "step": 2290 }, { "epoch": 1.6783882783882784, "grad_norm": 1.0234375, "learning_rate": 9.585600669736292e-05, "loss": 0.2107, "step": 2291 }, { "epoch": 1.6791208791208792, "grad_norm": 1.125, "learning_rate": 9.589786521557137e-05, "loss": 0.1861, "step": 2292 }, { "epoch": 1.6798534798534799, "grad_norm": 0.9921875, "learning_rate": 9.593972373377983e-05, "loss": 0.1915, "step": 2293 }, { "epoch": 1.6805860805860806, "grad_norm": 1.328125, "learning_rate": 9.598158225198828e-05, "loss": 0.1995, "step": 2294 }, { "epoch": 1.6813186813186813, "grad_norm": 1.0703125, "learning_rate": 9.602344077019673e-05, "loss": 0.1911, "step": 2295 }, { "epoch": 1.682051282051282, "grad_norm": 0.86328125, "learning_rate": 9.60652992884052e-05, "loss": 0.1917, "step": 2296 }, { "epoch": 1.6827838827838828, "grad_norm": 1.2421875, "learning_rate": 9.610715780661365e-05, "loss": 0.1811, "step": 2297 }, { "epoch": 1.6835164835164835, "grad_norm": 1.34375, "learning_rate": 9.61490163248221e-05, "loss": 0.2273, "step": 2298 }, { "epoch": 1.6842490842490843, "grad_norm": 0.94140625, "learning_rate": 9.619087484303056e-05, "loss": 0.2208, "step": 2299 }, { "epoch": 1.684981684981685, "grad_norm": 1.0546875, "learning_rate": 9.623273336123901e-05, "loss": 0.1952, "step": 2300 }, { "epoch": 1.6857142857142857, "grad_norm": 1.1484375, "learning_rate": 9.627459187944748e-05, "loss": 0.1755, "step": 2301 }, { "epoch": 1.6864468864468865, "grad_norm": 2.015625, "learning_rate": 9.631645039765593e-05, "loss": 0.2127, "step": 2302 }, { "epoch": 1.6871794871794872, "grad_norm": 1.71875, "learning_rate": 9.635830891586438e-05, "loss": 0.2587, "step": 2303 }, { "epoch": 1.687912087912088, "grad_norm": 0.85546875, "learning_rate": 9.640016743407284e-05, "loss": 0.1678, "step": 2304 }, { "epoch": 1.6886446886446886, "grad_norm": 1.375, "learning_rate": 9.64420259522813e-05, "loss": 0.2314, "step": 2305 }, { "epoch": 1.6893772893772894, "grad_norm": 1.1796875, "learning_rate": 9.648388447048975e-05, "loss": 0.1997, "step": 2306 }, { "epoch": 1.69010989010989, "grad_norm": 1.171875, "learning_rate": 9.652574298869821e-05, "loss": 0.1862, "step": 2307 }, { "epoch": 1.6908424908424908, "grad_norm": 1.6640625, "learning_rate": 9.656760150690666e-05, "loss": 0.2519, "step": 2308 }, { "epoch": 1.6915750915750916, "grad_norm": 0.9921875, "learning_rate": 9.660946002511511e-05, "loss": 0.1943, "step": 2309 }, { "epoch": 1.6923076923076923, "grad_norm": 1.0390625, "learning_rate": 9.665131854332356e-05, "loss": 0.2097, "step": 2310 }, { "epoch": 1.693040293040293, "grad_norm": 0.9609375, "learning_rate": 9.669317706153203e-05, "loss": 0.1894, "step": 2311 }, { "epoch": 1.6937728937728938, "grad_norm": 1.4375, "learning_rate": 9.673503557974049e-05, "loss": 0.2218, "step": 2312 }, { "epoch": 1.6945054945054945, "grad_norm": 1.4296875, "learning_rate": 9.677689409794893e-05, "loss": 0.2243, "step": 2313 }, { "epoch": 1.6952380952380952, "grad_norm": 1.140625, "learning_rate": 9.681875261615739e-05, "loss": 0.2216, "step": 2314 }, { "epoch": 1.695970695970696, "grad_norm": 1.2578125, "learning_rate": 9.686061113436586e-05, "loss": 0.1875, "step": 2315 }, { "epoch": 1.6967032967032967, "grad_norm": 1.078125, "learning_rate": 9.69024696525743e-05, "loss": 0.1972, "step": 2316 }, { "epoch": 1.6974358974358974, "grad_norm": 1.7421875, "learning_rate": 9.694432817078276e-05, "loss": 0.2403, "step": 2317 }, { "epoch": 1.6981684981684981, "grad_norm": 0.79296875, "learning_rate": 9.698618668899121e-05, "loss": 0.1813, "step": 2318 }, { "epoch": 1.6989010989010989, "grad_norm": 0.84375, "learning_rate": 9.702804520719967e-05, "loss": 0.1951, "step": 2319 }, { "epoch": 1.6996336996336996, "grad_norm": 1.3984375, "learning_rate": 9.706990372540814e-05, "loss": 0.2198, "step": 2320 }, { "epoch": 1.7003663003663003, "grad_norm": 1.71875, "learning_rate": 9.711176224361657e-05, "loss": 0.2049, "step": 2321 }, { "epoch": 1.701098901098901, "grad_norm": 0.89453125, "learning_rate": 9.715362076182504e-05, "loss": 0.1983, "step": 2322 }, { "epoch": 1.7018315018315018, "grad_norm": 1.0625, "learning_rate": 9.71954792800335e-05, "loss": 0.2525, "step": 2323 }, { "epoch": 1.7025641025641025, "grad_norm": 1.1796875, "learning_rate": 9.723733779824195e-05, "loss": 0.2132, "step": 2324 }, { "epoch": 1.7032967032967035, "grad_norm": 0.9921875, "learning_rate": 9.72791963164504e-05, "loss": 0.2023, "step": 2325 }, { "epoch": 1.7040293040293042, "grad_norm": 0.8359375, "learning_rate": 9.732105483465885e-05, "loss": 0.2011, "step": 2326 }, { "epoch": 1.704761904761905, "grad_norm": 0.97265625, "learning_rate": 9.736291335286732e-05, "loss": 0.2067, "step": 2327 }, { "epoch": 1.7054945054945057, "grad_norm": 1.3203125, "learning_rate": 9.740477187107577e-05, "loss": 0.237, "step": 2328 }, { "epoch": 1.7062271062271064, "grad_norm": 1.2265625, "learning_rate": 9.744663038928422e-05, "loss": 0.2516, "step": 2329 }, { "epoch": 1.7069597069597071, "grad_norm": 1.15625, "learning_rate": 9.748848890749268e-05, "loss": 0.2311, "step": 2330 }, { "epoch": 1.7076923076923078, "grad_norm": 1.5390625, "learning_rate": 9.753034742570113e-05, "loss": 0.2224, "step": 2331 }, { "epoch": 1.7084249084249086, "grad_norm": 0.94140625, "learning_rate": 9.757220594390958e-05, "loss": 0.1689, "step": 2332 }, { "epoch": 1.7091575091575093, "grad_norm": 1.0390625, "learning_rate": 9.761406446211805e-05, "loss": 0.2151, "step": 2333 }, { "epoch": 1.70989010989011, "grad_norm": 1.3359375, "learning_rate": 9.76559229803265e-05, "loss": 0.2492, "step": 2334 }, { "epoch": 1.7106227106227108, "grad_norm": 1.046875, "learning_rate": 9.769778149853496e-05, "loss": 0.1733, "step": 2335 }, { "epoch": 1.7113553113553115, "grad_norm": 1.2421875, "learning_rate": 9.773964001674341e-05, "loss": 0.2366, "step": 2336 }, { "epoch": 1.7120879120879122, "grad_norm": 1.0390625, "learning_rate": 9.778149853495186e-05, "loss": 0.2404, "step": 2337 }, { "epoch": 1.712820512820513, "grad_norm": 1.234375, "learning_rate": 9.782335705316033e-05, "loss": 0.1905, "step": 2338 }, { "epoch": 1.7135531135531137, "grad_norm": 1.2265625, "learning_rate": 9.786521557136878e-05, "loss": 0.196, "step": 2339 }, { "epoch": 1.7142857142857144, "grad_norm": 0.82421875, "learning_rate": 9.790707408957723e-05, "loss": 0.1962, "step": 2340 }, { "epoch": 1.7150183150183151, "grad_norm": 1.1875, "learning_rate": 9.79489326077857e-05, "loss": 0.2007, "step": 2341 }, { "epoch": 1.7157509157509159, "grad_norm": 1.03125, "learning_rate": 9.799079112599414e-05, "loss": 0.2386, "step": 2342 }, { "epoch": 1.7164835164835166, "grad_norm": 1.046875, "learning_rate": 9.80326496442026e-05, "loss": 0.2059, "step": 2343 }, { "epoch": 1.7172161172161173, "grad_norm": 1.3359375, "learning_rate": 9.807450816241106e-05, "loss": 0.2356, "step": 2344 }, { "epoch": 1.717948717948718, "grad_norm": 1.375, "learning_rate": 9.811636668061951e-05, "loss": 0.1548, "step": 2345 }, { "epoch": 1.7186813186813188, "grad_norm": 1.2265625, "learning_rate": 9.815822519882797e-05, "loss": 0.2029, "step": 2346 }, { "epoch": 1.7194139194139195, "grad_norm": 1.125, "learning_rate": 9.820008371703641e-05, "loss": 0.1828, "step": 2347 }, { "epoch": 1.7201465201465203, "grad_norm": 1.4375, "learning_rate": 9.824194223524488e-05, "loss": 0.2192, "step": 2348 }, { "epoch": 1.720879120879121, "grad_norm": 1.1640625, "learning_rate": 9.828380075345334e-05, "loss": 0.2011, "step": 2349 }, { "epoch": 1.7216117216117217, "grad_norm": 0.9609375, "learning_rate": 9.832565927166179e-05, "loss": 0.226, "step": 2350 }, { "epoch": 1.7223443223443224, "grad_norm": 1.0234375, "learning_rate": 9.836751778987024e-05, "loss": 0.219, "step": 2351 }, { "epoch": 1.7230769230769232, "grad_norm": 1.5078125, "learning_rate": 9.840937630807869e-05, "loss": 0.181, "step": 2352 }, { "epoch": 1.723809523809524, "grad_norm": 1.5234375, "learning_rate": 9.845123482628716e-05, "loss": 0.1841, "step": 2353 }, { "epoch": 1.7245421245421246, "grad_norm": 1.03125, "learning_rate": 9.849309334449562e-05, "loss": 0.1887, "step": 2354 }, { "epoch": 1.7252747252747254, "grad_norm": 1.4453125, "learning_rate": 9.853495186270406e-05, "loss": 0.1839, "step": 2355 }, { "epoch": 1.726007326007326, "grad_norm": 1.0078125, "learning_rate": 9.857681038091252e-05, "loss": 0.1824, "step": 2356 }, { "epoch": 1.7267399267399268, "grad_norm": 1.5546875, "learning_rate": 9.861866889912098e-05, "loss": 0.1877, "step": 2357 }, { "epoch": 1.7274725274725276, "grad_norm": 1.484375, "learning_rate": 9.866052741732944e-05, "loss": 0.1843, "step": 2358 }, { "epoch": 1.7282051282051283, "grad_norm": 1.0078125, "learning_rate": 9.870238593553789e-05, "loss": 0.2042, "step": 2359 }, { "epoch": 1.728937728937729, "grad_norm": 1.46875, "learning_rate": 9.874424445374634e-05, "loss": 0.1982, "step": 2360 }, { "epoch": 1.7296703296703297, "grad_norm": 2.109375, "learning_rate": 9.87861029719548e-05, "loss": 0.218, "step": 2361 }, { "epoch": 1.7304029304029305, "grad_norm": 1.8671875, "learning_rate": 9.882796149016325e-05, "loss": 0.1935, "step": 2362 }, { "epoch": 1.7311355311355312, "grad_norm": 1.3984375, "learning_rate": 9.88698200083717e-05, "loss": 0.1769, "step": 2363 }, { "epoch": 1.731868131868132, "grad_norm": 1.7109375, "learning_rate": 9.891167852658017e-05, "loss": 0.1814, "step": 2364 }, { "epoch": 1.7326007326007327, "grad_norm": 1.7890625, "learning_rate": 9.895353704478863e-05, "loss": 0.2192, "step": 2365 }, { "epoch": 1.7333333333333334, "grad_norm": 2.65625, "learning_rate": 9.899539556299707e-05, "loss": 0.2251, "step": 2366 }, { "epoch": 1.7340659340659341, "grad_norm": 1.8203125, "learning_rate": 9.903725408120553e-05, "loss": 0.2556, "step": 2367 }, { "epoch": 1.7347985347985349, "grad_norm": 1.0234375, "learning_rate": 9.907911259941398e-05, "loss": 0.177, "step": 2368 }, { "epoch": 1.7355311355311356, "grad_norm": 1.8203125, "learning_rate": 9.912097111762245e-05, "loss": 0.2146, "step": 2369 }, { "epoch": 1.7362637362637363, "grad_norm": 1.03125, "learning_rate": 9.91628296358309e-05, "loss": 0.1489, "step": 2370 }, { "epoch": 1.736996336996337, "grad_norm": 1.0859375, "learning_rate": 9.920468815403935e-05, "loss": 0.1797, "step": 2371 }, { "epoch": 1.7377289377289378, "grad_norm": 1.25, "learning_rate": 9.924654667224781e-05, "loss": 0.2015, "step": 2372 }, { "epoch": 1.7384615384615385, "grad_norm": 1.453125, "learning_rate": 9.928840519045626e-05, "loss": 0.1845, "step": 2373 }, { "epoch": 1.7391941391941392, "grad_norm": 0.71484375, "learning_rate": 9.933026370866471e-05, "loss": 0.1712, "step": 2374 }, { "epoch": 1.73992673992674, "grad_norm": 1.3828125, "learning_rate": 9.937212222687318e-05, "loss": 0.2195, "step": 2375 }, { "epoch": 1.7406593406593407, "grad_norm": 0.9453125, "learning_rate": 9.941398074508163e-05, "loss": 0.185, "step": 2376 }, { "epoch": 1.7413919413919414, "grad_norm": 0.875, "learning_rate": 9.945583926329008e-05, "loss": 0.218, "step": 2377 }, { "epoch": 1.7421245421245422, "grad_norm": 1.15625, "learning_rate": 9.949769778149854e-05, "loss": 0.1598, "step": 2378 }, { "epoch": 1.7428571428571429, "grad_norm": 1.2578125, "learning_rate": 9.9539556299707e-05, "loss": 0.205, "step": 2379 }, { "epoch": 1.7435897435897436, "grad_norm": 0.890625, "learning_rate": 9.958141481791546e-05, "loss": 0.2208, "step": 2380 }, { "epoch": 1.7443223443223443, "grad_norm": 1.4140625, "learning_rate": 9.96232733361239e-05, "loss": 0.2024, "step": 2381 }, { "epoch": 1.745054945054945, "grad_norm": 1.3515625, "learning_rate": 9.966513185433236e-05, "loss": 0.2132, "step": 2382 }, { "epoch": 1.7457875457875458, "grad_norm": 0.99609375, "learning_rate": 9.970699037254082e-05, "loss": 0.1831, "step": 2383 }, { "epoch": 1.7465201465201465, "grad_norm": 0.984375, "learning_rate": 9.974884889074927e-05, "loss": 0.1999, "step": 2384 }, { "epoch": 1.7472527472527473, "grad_norm": 1.1015625, "learning_rate": 9.979070740895772e-05, "loss": 0.2494, "step": 2385 }, { "epoch": 1.747985347985348, "grad_norm": 1.59375, "learning_rate": 9.983256592716619e-05, "loss": 0.2202, "step": 2386 }, { "epoch": 1.7487179487179487, "grad_norm": 1.8984375, "learning_rate": 9.987442444537464e-05, "loss": 0.2134, "step": 2387 }, { "epoch": 1.7494505494505495, "grad_norm": 1.8203125, "learning_rate": 9.991628296358309e-05, "loss": 0.1925, "step": 2388 }, { "epoch": 1.7501831501831502, "grad_norm": 1.3046875, "learning_rate": 9.995814148179154e-05, "loss": 0.2475, "step": 2389 }, { "epoch": 1.750915750915751, "grad_norm": 1.5859375, "learning_rate": 0.0001, "loss": 0.176, "step": 2390 }, { "epoch": 1.7516483516483516, "grad_norm": 1.6484375, "learning_rate": 9.999999988141463e-05, "loss": 0.1983, "step": 2391 }, { "epoch": 1.7523809523809524, "grad_norm": 1.5, "learning_rate": 9.99999995256585e-05, "loss": 0.1902, "step": 2392 }, { "epoch": 1.753113553113553, "grad_norm": 1.4140625, "learning_rate": 9.999999893273163e-05, "loss": 0.2546, "step": 2393 }, { "epoch": 1.7538461538461538, "grad_norm": 1.5703125, "learning_rate": 9.999999810263402e-05, "loss": 0.2298, "step": 2394 }, { "epoch": 1.7545787545787546, "grad_norm": 1.7890625, "learning_rate": 9.999999703536566e-05, "loss": 0.2205, "step": 2395 }, { "epoch": 1.7553113553113553, "grad_norm": 1.40625, "learning_rate": 9.999999573092656e-05, "loss": 0.1757, "step": 2396 }, { "epoch": 1.756043956043956, "grad_norm": 0.7265625, "learning_rate": 9.999999418931674e-05, "loss": 0.1436, "step": 2397 }, { "epoch": 1.7567765567765568, "grad_norm": 1.890625, "learning_rate": 9.999999241053621e-05, "loss": 0.224, "step": 2398 }, { "epoch": 1.7575091575091575, "grad_norm": 1.3203125, "learning_rate": 9.999999039458495e-05, "loss": 0.1967, "step": 2399 }, { "epoch": 1.7582417582417582, "grad_norm": 1.46875, "learning_rate": 9.999998814146298e-05, "loss": 0.196, "step": 2400 }, { "epoch": 1.758974358974359, "grad_norm": 1.609375, "learning_rate": 9.999998565117033e-05, "loss": 0.199, "step": 2401 }, { "epoch": 1.7597069597069597, "grad_norm": 1.0859375, "learning_rate": 9.999998292370699e-05, "loss": 0.2183, "step": 2402 }, { "epoch": 1.7604395604395604, "grad_norm": 1.21875, "learning_rate": 9.999997995907298e-05, "loss": 0.2277, "step": 2403 }, { "epoch": 1.7611721611721611, "grad_norm": 1.1640625, "learning_rate": 9.999997675726833e-05, "loss": 0.1736, "step": 2404 }, { "epoch": 1.7619047619047619, "grad_norm": 0.75, "learning_rate": 9.999997331829304e-05, "loss": 0.1773, "step": 2405 }, { "epoch": 1.7626373626373626, "grad_norm": 1.296875, "learning_rate": 9.999996964214712e-05, "loss": 0.2081, "step": 2406 }, { "epoch": 1.7633699633699633, "grad_norm": 0.9296875, "learning_rate": 9.999996572883059e-05, "loss": 0.183, "step": 2407 }, { "epoch": 1.764102564102564, "grad_norm": 1.5546875, "learning_rate": 9.99999615783435e-05, "loss": 0.1751, "step": 2408 }, { "epoch": 1.7648351648351648, "grad_norm": 1.078125, "learning_rate": 9.999995719068582e-05, "loss": 0.1547, "step": 2409 }, { "epoch": 1.7655677655677655, "grad_norm": 0.89453125, "learning_rate": 9.999995256585762e-05, "loss": 0.2059, "step": 2410 }, { "epoch": 1.7663003663003662, "grad_norm": 1.2578125, "learning_rate": 9.999994770385886e-05, "loss": 0.234, "step": 2411 }, { "epoch": 1.767032967032967, "grad_norm": 1.171875, "learning_rate": 9.999994260468962e-05, "loss": 0.1583, "step": 2412 }, { "epoch": 1.7677655677655677, "grad_norm": 1.21875, "learning_rate": 9.99999372683499e-05, "loss": 0.1863, "step": 2413 }, { "epoch": 1.7684981684981684, "grad_norm": 1.078125, "learning_rate": 9.999993169483976e-05, "loss": 0.1964, "step": 2414 }, { "epoch": 1.7692307692307692, "grad_norm": 1.2421875, "learning_rate": 9.999992588415917e-05, "loss": 0.1577, "step": 2415 }, { "epoch": 1.76996336996337, "grad_norm": 1.0390625, "learning_rate": 9.999991983630819e-05, "loss": 0.214, "step": 2416 }, { "epoch": 1.7706959706959706, "grad_norm": 1.484375, "learning_rate": 9.999991355128684e-05, "loss": 0.2473, "step": 2417 }, { "epoch": 1.7714285714285714, "grad_norm": 1.796875, "learning_rate": 9.999990702909516e-05, "loss": 0.2322, "step": 2418 }, { "epoch": 1.772161172161172, "grad_norm": 1.1875, "learning_rate": 9.999990026973317e-05, "loss": 0.199, "step": 2419 }, { "epoch": 1.7728937728937728, "grad_norm": 1.78125, "learning_rate": 9.999989327320091e-05, "loss": 0.1486, "step": 2420 }, { "epoch": 1.7736263736263735, "grad_norm": 1.796875, "learning_rate": 9.999988603949842e-05, "loss": 0.2239, "step": 2421 }, { "epoch": 1.7743589743589743, "grad_norm": 1.7265625, "learning_rate": 9.999987856862572e-05, "loss": 0.1855, "step": 2422 }, { "epoch": 1.775091575091575, "grad_norm": 1.1015625, "learning_rate": 9.999987086058284e-05, "loss": 0.2029, "step": 2423 }, { "epoch": 1.7758241758241757, "grad_norm": 1.625, "learning_rate": 9.999986291536986e-05, "loss": 0.2049, "step": 2424 }, { "epoch": 1.7765567765567765, "grad_norm": 1.4375, "learning_rate": 9.999985473298676e-05, "loss": 0.2136, "step": 2425 }, { "epoch": 1.7772893772893772, "grad_norm": 1.5390625, "learning_rate": 9.999984631343362e-05, "loss": 0.2286, "step": 2426 }, { "epoch": 1.778021978021978, "grad_norm": 0.99609375, "learning_rate": 9.999983765671044e-05, "loss": 0.1846, "step": 2427 }, { "epoch": 1.7787545787545787, "grad_norm": 1.328125, "learning_rate": 9.999982876281731e-05, "loss": 0.1831, "step": 2428 }, { "epoch": 1.7794871794871794, "grad_norm": 1.2109375, "learning_rate": 9.999981963175426e-05, "loss": 0.1967, "step": 2429 }, { "epoch": 1.7802197802197801, "grad_norm": 1.375, "learning_rate": 9.999981026352132e-05, "loss": 0.2503, "step": 2430 }, { "epoch": 1.7809523809523808, "grad_norm": 1.828125, "learning_rate": 9.999980065811853e-05, "loss": 0.2685, "step": 2431 }, { "epoch": 1.7816849816849816, "grad_norm": 1.25, "learning_rate": 9.999979081554594e-05, "loss": 0.1705, "step": 2432 }, { "epoch": 1.7824175824175823, "grad_norm": 1.09375, "learning_rate": 9.999978073580362e-05, "loss": 0.1904, "step": 2433 }, { "epoch": 1.783150183150183, "grad_norm": 1.15625, "learning_rate": 9.999977041889158e-05, "loss": 0.1928, "step": 2434 }, { "epoch": 1.7838827838827838, "grad_norm": 1.6796875, "learning_rate": 9.999975986480991e-05, "loss": 0.2293, "step": 2435 }, { "epoch": 1.7846153846153845, "grad_norm": 1.3671875, "learning_rate": 9.999974907355864e-05, "loss": 0.2157, "step": 2436 }, { "epoch": 1.7853479853479852, "grad_norm": 1.46875, "learning_rate": 9.999973804513781e-05, "loss": 0.195, "step": 2437 }, { "epoch": 1.786080586080586, "grad_norm": 1.703125, "learning_rate": 9.999972677954749e-05, "loss": 0.1836, "step": 2438 }, { "epoch": 1.7868131868131867, "grad_norm": 1.4453125, "learning_rate": 9.999971527678773e-05, "loss": 0.2231, "step": 2439 }, { "epoch": 1.7875457875457874, "grad_norm": 1.2734375, "learning_rate": 9.999970353685857e-05, "loss": 0.205, "step": 2440 }, { "epoch": 1.7882783882783881, "grad_norm": 1.046875, "learning_rate": 9.999969155976012e-05, "loss": 0.1873, "step": 2441 }, { "epoch": 1.7890109890109889, "grad_norm": 1.34375, "learning_rate": 9.999967934549235e-05, "loss": 0.2024, "step": 2442 }, { "epoch": 1.7897435897435896, "grad_norm": 1.8984375, "learning_rate": 9.99996668940554e-05, "loss": 0.2489, "step": 2443 }, { "epoch": 1.7904761904761903, "grad_norm": 1.453125, "learning_rate": 9.999965420544928e-05, "loss": 0.1814, "step": 2444 }, { "epoch": 1.791208791208791, "grad_norm": 2.015625, "learning_rate": 9.999964127967407e-05, "loss": 0.2646, "step": 2445 }, { "epoch": 1.7919413919413918, "grad_norm": 1.328125, "learning_rate": 9.999962811672984e-05, "loss": 0.1833, "step": 2446 }, { "epoch": 1.7926739926739925, "grad_norm": 1.078125, "learning_rate": 9.999961471661663e-05, "loss": 0.1988, "step": 2447 }, { "epoch": 1.7934065934065933, "grad_norm": 1.2265625, "learning_rate": 9.999960107933453e-05, "loss": 0.1851, "step": 2448 }, { "epoch": 1.794139194139194, "grad_norm": 1.8671875, "learning_rate": 9.999958720488357e-05, "loss": 0.2062, "step": 2449 }, { "epoch": 1.7948717948717947, "grad_norm": 0.8671875, "learning_rate": 9.999957309326385e-05, "loss": 0.1684, "step": 2450 }, { "epoch": 1.7956043956043954, "grad_norm": 1.0703125, "learning_rate": 9.999955874447543e-05, "loss": 0.2118, "step": 2451 }, { "epoch": 1.7963369963369962, "grad_norm": 1.4765625, "learning_rate": 9.999954415851838e-05, "loss": 0.1721, "step": 2452 }, { "epoch": 1.7970695970695971, "grad_norm": 1.703125, "learning_rate": 9.999952933539274e-05, "loss": 0.2202, "step": 2453 }, { "epoch": 1.7978021978021979, "grad_norm": 1.125, "learning_rate": 9.999951427509861e-05, "loss": 0.2263, "step": 2454 }, { "epoch": 1.7985347985347986, "grad_norm": 1.171875, "learning_rate": 9.999949897763608e-05, "loss": 0.1805, "step": 2455 }, { "epoch": 1.7992673992673993, "grad_norm": 1.8984375, "learning_rate": 9.99994834430052e-05, "loss": 0.2324, "step": 2456 }, { "epoch": 1.8, "grad_norm": 1.3671875, "learning_rate": 9.999946767120602e-05, "loss": 0.2021, "step": 2457 }, { "epoch": 1.8007326007326008, "grad_norm": 1.171875, "learning_rate": 9.999945166223866e-05, "loss": 0.1691, "step": 2458 }, { "epoch": 1.8014652014652015, "grad_norm": 1.0, "learning_rate": 9.999943541610317e-05, "loss": 0.1779, "step": 2459 }, { "epoch": 1.8021978021978022, "grad_norm": 1.53125, "learning_rate": 9.999941893279963e-05, "loss": 0.2448, "step": 2460 }, { "epoch": 1.802930402930403, "grad_norm": 1.078125, "learning_rate": 9.999940221232813e-05, "loss": 0.157, "step": 2461 }, { "epoch": 1.8036630036630037, "grad_norm": 1.5703125, "learning_rate": 9.999938525468874e-05, "loss": 0.2417, "step": 2462 }, { "epoch": 1.8043956043956044, "grad_norm": 1.5, "learning_rate": 9.999936805988157e-05, "loss": 0.1827, "step": 2463 }, { "epoch": 1.8051282051282052, "grad_norm": 1.734375, "learning_rate": 9.999935062790664e-05, "loss": 0.2172, "step": 2464 }, { "epoch": 1.8058608058608059, "grad_norm": 1.6484375, "learning_rate": 9.999933295876409e-05, "loss": 0.2018, "step": 2465 }, { "epoch": 1.8065934065934066, "grad_norm": 1.1953125, "learning_rate": 9.999931505245399e-05, "loss": 0.2285, "step": 2466 }, { "epoch": 1.8073260073260073, "grad_norm": 1.6796875, "learning_rate": 9.999929690897642e-05, "loss": 0.1915, "step": 2467 }, { "epoch": 1.808058608058608, "grad_norm": 1.796875, "learning_rate": 9.999927852833146e-05, "loss": 0.1937, "step": 2468 }, { "epoch": 1.8087912087912088, "grad_norm": 2.09375, "learning_rate": 9.999925991051922e-05, "loss": 0.2154, "step": 2469 }, { "epoch": 1.8095238095238095, "grad_norm": 2.21875, "learning_rate": 9.999924105553978e-05, "loss": 0.2187, "step": 2470 }, { "epoch": 1.8102564102564103, "grad_norm": 2.40625, "learning_rate": 9.999922196339321e-05, "loss": 0.167, "step": 2471 }, { "epoch": 1.810989010989011, "grad_norm": 1.421875, "learning_rate": 9.999920263407963e-05, "loss": 0.1677, "step": 2472 }, { "epoch": 1.8117216117216117, "grad_norm": 0.75, "learning_rate": 9.999918306759911e-05, "loss": 0.1597, "step": 2473 }, { "epoch": 1.8124542124542125, "grad_norm": 1.21875, "learning_rate": 9.999916326395177e-05, "loss": 0.2139, "step": 2474 }, { "epoch": 1.8131868131868132, "grad_norm": 1.578125, "learning_rate": 9.999914322313767e-05, "loss": 0.1987, "step": 2475 }, { "epoch": 1.813919413919414, "grad_norm": 1.28125, "learning_rate": 9.999912294515695e-05, "loss": 0.2083, "step": 2476 }, { "epoch": 1.8146520146520146, "grad_norm": 1.171875, "learning_rate": 9.999910243000967e-05, "loss": 0.1702, "step": 2477 }, { "epoch": 1.8153846153846154, "grad_norm": 1.2265625, "learning_rate": 9.999908167769593e-05, "loss": 0.2239, "step": 2478 }, { "epoch": 1.816117216117216, "grad_norm": 1.296875, "learning_rate": 9.999906068821583e-05, "loss": 0.1906, "step": 2479 }, { "epoch": 1.8168498168498168, "grad_norm": 1.4140625, "learning_rate": 9.999903946156951e-05, "loss": 0.2141, "step": 2480 }, { "epoch": 1.8175824175824176, "grad_norm": 1.1875, "learning_rate": 9.999901799775703e-05, "loss": 0.1735, "step": 2481 }, { "epoch": 1.8183150183150183, "grad_norm": 0.73828125, "learning_rate": 9.999899629677853e-05, "loss": 0.1565, "step": 2482 }, { "epoch": 1.819047619047619, "grad_norm": 1.3515625, "learning_rate": 9.999897435863406e-05, "loss": 0.1762, "step": 2483 }, { "epoch": 1.8197802197802198, "grad_norm": 1.1484375, "learning_rate": 9.999895218332375e-05, "loss": 0.2041, "step": 2484 }, { "epoch": 1.8205128205128205, "grad_norm": 0.83203125, "learning_rate": 9.999892977084772e-05, "loss": 0.1878, "step": 2485 }, { "epoch": 1.8212454212454212, "grad_norm": 0.91015625, "learning_rate": 9.999890712120606e-05, "loss": 0.1754, "step": 2486 }, { "epoch": 1.821978021978022, "grad_norm": 1.9765625, "learning_rate": 9.99988842343989e-05, "loss": 0.1931, "step": 2487 }, { "epoch": 1.8227106227106227, "grad_norm": 1.1640625, "learning_rate": 9.999886111042633e-05, "loss": 0.1897, "step": 2488 }, { "epoch": 1.8234432234432234, "grad_norm": 1.3203125, "learning_rate": 9.999883774928849e-05, "loss": 0.146, "step": 2489 }, { "epoch": 1.8241758241758241, "grad_norm": 1.2421875, "learning_rate": 9.999881415098543e-05, "loss": 0.1913, "step": 2490 }, { "epoch": 1.8249084249084249, "grad_norm": 0.99609375, "learning_rate": 9.999879031551733e-05, "loss": 0.202, "step": 2491 }, { "epoch": 1.8256410256410256, "grad_norm": 1.0625, "learning_rate": 9.999876624288426e-05, "loss": 0.2186, "step": 2492 }, { "epoch": 1.8263736263736263, "grad_norm": 0.78515625, "learning_rate": 9.999874193308637e-05, "loss": 0.1655, "step": 2493 }, { "epoch": 1.827106227106227, "grad_norm": 1.4296875, "learning_rate": 9.999871738612373e-05, "loss": 0.2092, "step": 2494 }, { "epoch": 1.8278388278388278, "grad_norm": 0.73046875, "learning_rate": 9.999869260199651e-05, "loss": 0.1804, "step": 2495 }, { "epoch": 1.8285714285714287, "grad_norm": 1.1875, "learning_rate": 9.99986675807048e-05, "loss": 0.2287, "step": 2496 }, { "epoch": 1.8293040293040295, "grad_norm": 1.1484375, "learning_rate": 9.999864232224872e-05, "loss": 0.2244, "step": 2497 }, { "epoch": 1.8300366300366302, "grad_norm": 1.1640625, "learning_rate": 9.999861682662839e-05, "loss": 0.2054, "step": 2498 }, { "epoch": 1.830769230769231, "grad_norm": 1.1875, "learning_rate": 9.999859109384394e-05, "loss": 0.1766, "step": 2499 }, { "epoch": 1.8315018315018317, "grad_norm": 1.7734375, "learning_rate": 9.99985651238955e-05, "loss": 0.2393, "step": 2500 }, { "epoch": 1.8322344322344324, "grad_norm": 1.4453125, "learning_rate": 9.999853891678318e-05, "loss": 0.2619, "step": 2501 }, { "epoch": 1.8329670329670331, "grad_norm": 1.2109375, "learning_rate": 9.99985124725071e-05, "loss": 0.1471, "step": 2502 }, { "epoch": 1.8336996336996338, "grad_norm": 1.0625, "learning_rate": 9.999848579106741e-05, "loss": 0.1948, "step": 2503 }, { "epoch": 1.8344322344322346, "grad_norm": 0.9765625, "learning_rate": 9.999845887246422e-05, "loss": 0.1574, "step": 2504 }, { "epoch": 1.8351648351648353, "grad_norm": 1.046875, "learning_rate": 9.999843171669768e-05, "loss": 0.2131, "step": 2505 }, { "epoch": 1.835897435897436, "grad_norm": 0.921875, "learning_rate": 9.999840432376788e-05, "loss": 0.189, "step": 2506 }, { "epoch": 1.8366300366300368, "grad_norm": 1.171875, "learning_rate": 9.999837669367499e-05, "loss": 0.1845, "step": 2507 }, { "epoch": 1.8373626373626375, "grad_norm": 1.4140625, "learning_rate": 9.999834882641913e-05, "loss": 0.2173, "step": 2508 }, { "epoch": 1.8380952380952382, "grad_norm": 1.25, "learning_rate": 9.999832072200042e-05, "loss": 0.2749, "step": 2509 }, { "epoch": 1.838827838827839, "grad_norm": 1.3125, "learning_rate": 9.999829238041902e-05, "loss": 0.2037, "step": 2510 }, { "epoch": 1.8395604395604397, "grad_norm": 1.7265625, "learning_rate": 9.999826380167504e-05, "loss": 0.2453, "step": 2511 }, { "epoch": 1.8402930402930404, "grad_norm": 2.109375, "learning_rate": 9.999823498576864e-05, "loss": 0.2066, "step": 2512 }, { "epoch": 1.8410256410256411, "grad_norm": 1.3671875, "learning_rate": 9.999820593269993e-05, "loss": 0.198, "step": 2513 }, { "epoch": 1.8417582417582419, "grad_norm": 1.1328125, "learning_rate": 9.999817664246907e-05, "loss": 0.1846, "step": 2514 }, { "epoch": 1.8424908424908426, "grad_norm": 1.4296875, "learning_rate": 9.999814711507621e-05, "loss": 0.1948, "step": 2515 }, { "epoch": 1.8432234432234433, "grad_norm": 0.73828125, "learning_rate": 9.999811735052147e-05, "loss": 0.16, "step": 2516 }, { "epoch": 1.843956043956044, "grad_norm": 1.1953125, "learning_rate": 9.9998087348805e-05, "loss": 0.156, "step": 2517 }, { "epoch": 1.8446886446886448, "grad_norm": 1.140625, "learning_rate": 9.999805710992695e-05, "loss": 0.1674, "step": 2518 }, { "epoch": 1.8454212454212455, "grad_norm": 1.2890625, "learning_rate": 9.999802663388745e-05, "loss": 0.235, "step": 2519 }, { "epoch": 1.8461538461538463, "grad_norm": 0.7890625, "learning_rate": 9.999799592068666e-05, "loss": 0.1524, "step": 2520 }, { "epoch": 1.846886446886447, "grad_norm": 1.1015625, "learning_rate": 9.999796497032471e-05, "loss": 0.1797, "step": 2521 }, { "epoch": 1.8476190476190477, "grad_norm": 0.8828125, "learning_rate": 9.999793378280179e-05, "loss": 0.1919, "step": 2522 }, { "epoch": 1.8483516483516484, "grad_norm": 1.40625, "learning_rate": 9.999790235811801e-05, "loss": 0.2153, "step": 2523 }, { "epoch": 1.8490842490842492, "grad_norm": 1.5390625, "learning_rate": 9.999787069627352e-05, "loss": 0.2098, "step": 2524 }, { "epoch": 1.84981684981685, "grad_norm": 0.9921875, "learning_rate": 9.999783879726848e-05, "loss": 0.2075, "step": 2525 }, { "epoch": 1.8505494505494506, "grad_norm": 1.1484375, "learning_rate": 9.999780666110305e-05, "loss": 0.1669, "step": 2526 }, { "epoch": 1.8512820512820514, "grad_norm": 1.6953125, "learning_rate": 9.999777428777739e-05, "loss": 0.186, "step": 2527 }, { "epoch": 1.852014652014652, "grad_norm": 0.80078125, "learning_rate": 9.999774167729163e-05, "loss": 0.1767, "step": 2528 }, { "epoch": 1.8527472527472528, "grad_norm": 0.8203125, "learning_rate": 9.999770882964594e-05, "loss": 0.1917, "step": 2529 }, { "epoch": 1.8534798534798536, "grad_norm": 1.0078125, "learning_rate": 9.99976757448405e-05, "loss": 0.1939, "step": 2530 }, { "epoch": 1.8542124542124543, "grad_norm": 1.6953125, "learning_rate": 9.999764242287542e-05, "loss": 0.229, "step": 2531 }, { "epoch": 1.854945054945055, "grad_norm": 1.3984375, "learning_rate": 9.99976088637509e-05, "loss": 0.1677, "step": 2532 }, { "epoch": 1.8556776556776557, "grad_norm": 1.6875, "learning_rate": 9.999757506746709e-05, "loss": 0.2409, "step": 2533 }, { "epoch": 1.8564102564102565, "grad_norm": 1.5234375, "learning_rate": 9.999754103402414e-05, "loss": 0.1763, "step": 2534 }, { "epoch": 1.8571428571428572, "grad_norm": 1.796875, "learning_rate": 9.999750676342221e-05, "loss": 0.1483, "step": 2535 }, { "epoch": 1.857875457875458, "grad_norm": 1.8125, "learning_rate": 9.99974722556615e-05, "loss": 0.223, "step": 2536 }, { "epoch": 1.8586080586080587, "grad_norm": 1.5546875, "learning_rate": 9.999743751074214e-05, "loss": 0.2444, "step": 2537 }, { "epoch": 1.8593406593406594, "grad_norm": 1.1328125, "learning_rate": 9.99974025286643e-05, "loss": 0.1669, "step": 2538 }, { "epoch": 1.8600732600732601, "grad_norm": 1.296875, "learning_rate": 9.999736730942818e-05, "loss": 0.2175, "step": 2539 }, { "epoch": 1.8608058608058609, "grad_norm": 1.3671875, "learning_rate": 9.999733185303392e-05, "loss": 0.2181, "step": 2540 }, { "epoch": 1.8615384615384616, "grad_norm": 1.6015625, "learning_rate": 9.999729615948166e-05, "loss": 0.1932, "step": 2541 }, { "epoch": 1.8622710622710623, "grad_norm": 0.9140625, "learning_rate": 9.999726022877164e-05, "loss": 0.2096, "step": 2542 }, { "epoch": 1.863003663003663, "grad_norm": 0.9609375, "learning_rate": 9.999722406090398e-05, "loss": 0.2202, "step": 2543 }, { "epoch": 1.8637362637362638, "grad_norm": 1.1171875, "learning_rate": 9.999718765587887e-05, "loss": 0.2403, "step": 2544 }, { "epoch": 1.8644688644688645, "grad_norm": 1.40625, "learning_rate": 9.999715101369651e-05, "loss": 0.2148, "step": 2545 }, { "epoch": 1.8652014652014652, "grad_norm": 1.0859375, "learning_rate": 9.999711413435703e-05, "loss": 0.1884, "step": 2546 }, { "epoch": 1.865934065934066, "grad_norm": 1.0625, "learning_rate": 9.999707701786063e-05, "loss": 0.2122, "step": 2547 }, { "epoch": 1.8666666666666667, "grad_norm": 1.265625, "learning_rate": 9.999703966420748e-05, "loss": 0.2084, "step": 2548 }, { "epoch": 1.8673992673992674, "grad_norm": 1.2890625, "learning_rate": 9.999700207339778e-05, "loss": 0.2211, "step": 2549 }, { "epoch": 1.8681318681318682, "grad_norm": 1.015625, "learning_rate": 9.999696424543167e-05, "loss": 0.2149, "step": 2550 }, { "epoch": 1.8688644688644689, "grad_norm": 1.578125, "learning_rate": 9.999692618030938e-05, "loss": 0.2147, "step": 2551 }, { "epoch": 1.8695970695970696, "grad_norm": 1.65625, "learning_rate": 9.999688787803105e-05, "loss": 0.2148, "step": 2552 }, { "epoch": 1.8703296703296703, "grad_norm": 1.234375, "learning_rate": 9.99968493385969e-05, "loss": 0.1939, "step": 2553 }, { "epoch": 1.871062271062271, "grad_norm": 1.375, "learning_rate": 9.999681056200708e-05, "loss": 0.175, "step": 2554 }, { "epoch": 1.8717948717948718, "grad_norm": 1.6640625, "learning_rate": 9.99967715482618e-05, "loss": 0.1815, "step": 2555 }, { "epoch": 1.8725274725274725, "grad_norm": 1.828125, "learning_rate": 9.999673229736124e-05, "loss": 0.2274, "step": 2556 }, { "epoch": 1.8732600732600733, "grad_norm": 1.3359375, "learning_rate": 9.999669280930557e-05, "loss": 0.219, "step": 2557 }, { "epoch": 1.873992673992674, "grad_norm": 1.171875, "learning_rate": 9.9996653084095e-05, "loss": 0.1808, "step": 2558 }, { "epoch": 1.8747252747252747, "grad_norm": 1.921875, "learning_rate": 9.999661312172973e-05, "loss": 0.1843, "step": 2559 }, { "epoch": 1.8754578754578755, "grad_norm": 1.359375, "learning_rate": 9.999657292220994e-05, "loss": 0.2127, "step": 2560 }, { "epoch": 1.8761904761904762, "grad_norm": 1.0703125, "learning_rate": 9.99965324855358e-05, "loss": 0.2064, "step": 2561 }, { "epoch": 1.876923076923077, "grad_norm": 0.93359375, "learning_rate": 9.999649181170757e-05, "loss": 0.1933, "step": 2562 }, { "epoch": 1.8776556776556776, "grad_norm": 1.0, "learning_rate": 9.999645090072535e-05, "loss": 0.1664, "step": 2563 }, { "epoch": 1.8783882783882784, "grad_norm": 1.234375, "learning_rate": 9.99964097525894e-05, "loss": 0.1286, "step": 2564 }, { "epoch": 1.879120879120879, "grad_norm": 1.0859375, "learning_rate": 9.99963683672999e-05, "loss": 0.2035, "step": 2565 }, { "epoch": 1.8798534798534798, "grad_norm": 2.015625, "learning_rate": 9.999632674485705e-05, "loss": 0.1929, "step": 2566 }, { "epoch": 1.8805860805860806, "grad_norm": 2.734375, "learning_rate": 9.999628488526106e-05, "loss": 0.2096, "step": 2567 }, { "epoch": 1.8813186813186813, "grad_norm": 1.8515625, "learning_rate": 9.999624278851212e-05, "loss": 0.2034, "step": 2568 }, { "epoch": 1.882051282051282, "grad_norm": 0.921875, "learning_rate": 9.999620045461043e-05, "loss": 0.1684, "step": 2569 }, { "epoch": 1.8827838827838828, "grad_norm": 1.203125, "learning_rate": 9.999615788355619e-05, "loss": 0.2152, "step": 2570 }, { "epoch": 1.8835164835164835, "grad_norm": 1.6484375, "learning_rate": 9.99961150753496e-05, "loss": 0.2328, "step": 2571 }, { "epoch": 1.8842490842490842, "grad_norm": 1.3828125, "learning_rate": 9.999607202999088e-05, "loss": 0.182, "step": 2572 }, { "epoch": 1.884981684981685, "grad_norm": 1.0625, "learning_rate": 9.999602874748024e-05, "loss": 0.1728, "step": 2573 }, { "epoch": 1.8857142857142857, "grad_norm": 0.8203125, "learning_rate": 9.999598522781787e-05, "loss": 0.1827, "step": 2574 }, { "epoch": 1.8864468864468864, "grad_norm": 1.6328125, "learning_rate": 9.9995941471004e-05, "loss": 0.2143, "step": 2575 }, { "epoch": 1.8871794871794871, "grad_norm": 2.5, "learning_rate": 9.999589747703882e-05, "loss": 0.1954, "step": 2576 }, { "epoch": 1.8879120879120879, "grad_norm": 1.984375, "learning_rate": 9.999585324592253e-05, "loss": 0.1796, "step": 2577 }, { "epoch": 1.8886446886446886, "grad_norm": 0.984375, "learning_rate": 9.999580877765537e-05, "loss": 0.2115, "step": 2578 }, { "epoch": 1.8893772893772893, "grad_norm": 0.99609375, "learning_rate": 9.999576407223754e-05, "loss": 0.2175, "step": 2579 }, { "epoch": 1.89010989010989, "grad_norm": 0.93359375, "learning_rate": 9.999571912966927e-05, "loss": 0.2126, "step": 2580 }, { "epoch": 1.8908424908424908, "grad_norm": 0.9140625, "learning_rate": 9.999567394995076e-05, "loss": 0.2031, "step": 2581 }, { "epoch": 1.8915750915750915, "grad_norm": 0.875, "learning_rate": 9.999562853308221e-05, "loss": 0.2029, "step": 2582 }, { "epoch": 1.8923076923076922, "grad_norm": 1.53125, "learning_rate": 9.999558287906386e-05, "loss": 0.2034, "step": 2583 }, { "epoch": 1.893040293040293, "grad_norm": 0.94921875, "learning_rate": 9.999553698789592e-05, "loss": 0.2318, "step": 2584 }, { "epoch": 1.8937728937728937, "grad_norm": 1.1484375, "learning_rate": 9.999549085957862e-05, "loss": 0.1921, "step": 2585 }, { "epoch": 1.8945054945054944, "grad_norm": 1.359375, "learning_rate": 9.999544449411219e-05, "loss": 0.205, "step": 2586 }, { "epoch": 1.8952380952380952, "grad_norm": 1.3046875, "learning_rate": 9.999539789149682e-05, "loss": 0.2091, "step": 2587 }, { "epoch": 1.895970695970696, "grad_norm": 1.96875, "learning_rate": 9.999535105173276e-05, "loss": 0.2367, "step": 2588 }, { "epoch": 1.8967032967032966, "grad_norm": 1.0234375, "learning_rate": 9.99953039748202e-05, "loss": 0.1872, "step": 2589 }, { "epoch": 1.8974358974358974, "grad_norm": 1.03125, "learning_rate": 9.999525666075943e-05, "loss": 0.1784, "step": 2590 }, { "epoch": 1.898168498168498, "grad_norm": 1.3046875, "learning_rate": 9.99952091095506e-05, "loss": 0.194, "step": 2591 }, { "epoch": 1.8989010989010988, "grad_norm": 1.703125, "learning_rate": 9.999516132119399e-05, "loss": 0.178, "step": 2592 }, { "epoch": 1.8996336996336995, "grad_norm": 1.890625, "learning_rate": 9.999511329568983e-05, "loss": 0.1977, "step": 2593 }, { "epoch": 1.9003663003663003, "grad_norm": 1.4140625, "learning_rate": 9.999506503303832e-05, "loss": 0.2235, "step": 2594 }, { "epoch": 1.901098901098901, "grad_norm": 1.0234375, "learning_rate": 9.99950165332397e-05, "loss": 0.2025, "step": 2595 }, { "epoch": 1.9018315018315017, "grad_norm": 1.1015625, "learning_rate": 9.99949677962942e-05, "loss": 0.2078, "step": 2596 }, { "epoch": 1.9025641025641025, "grad_norm": 1.140625, "learning_rate": 9.999491882220207e-05, "loss": 0.1545, "step": 2597 }, { "epoch": 1.9032967032967032, "grad_norm": 1.203125, "learning_rate": 9.999486961096355e-05, "loss": 0.2345, "step": 2598 }, { "epoch": 1.904029304029304, "grad_norm": 1.6796875, "learning_rate": 9.999482016257885e-05, "loss": 0.1608, "step": 2599 }, { "epoch": 1.9047619047619047, "grad_norm": 1.3828125, "learning_rate": 9.999477047704822e-05, "loss": 0.1755, "step": 2600 }, { "epoch": 1.9054945054945054, "grad_norm": 1.375, "learning_rate": 9.99947205543719e-05, "loss": 0.1913, "step": 2601 }, { "epoch": 1.9062271062271061, "grad_norm": 2.09375, "learning_rate": 9.999467039455011e-05, "loss": 0.1953, "step": 2602 }, { "epoch": 1.9069597069597068, "grad_norm": 1.359375, "learning_rate": 9.999461999758313e-05, "loss": 0.192, "step": 2603 }, { "epoch": 1.9076923076923076, "grad_norm": 1.90625, "learning_rate": 9.999456936347116e-05, "loss": 0.2145, "step": 2604 }, { "epoch": 1.9084249084249083, "grad_norm": 1.0859375, "learning_rate": 9.999451849221446e-05, "loss": 0.1751, "step": 2605 }, { "epoch": 1.909157509157509, "grad_norm": 0.921875, "learning_rate": 9.999446738381327e-05, "loss": 0.1891, "step": 2606 }, { "epoch": 1.9098901098901098, "grad_norm": 0.83203125, "learning_rate": 9.999441603826786e-05, "loss": 0.1604, "step": 2607 }, { "epoch": 1.9106227106227105, "grad_norm": 0.9453125, "learning_rate": 9.999436445557845e-05, "loss": 0.1488, "step": 2608 }, { "epoch": 1.9113553113553112, "grad_norm": 0.99609375, "learning_rate": 9.999431263574527e-05, "loss": 0.1848, "step": 2609 }, { "epoch": 1.912087912087912, "grad_norm": 0.8515625, "learning_rate": 9.999426057876861e-05, "loss": 0.1828, "step": 2610 }, { "epoch": 1.9128205128205127, "grad_norm": 0.91015625, "learning_rate": 9.99942082846487e-05, "loss": 0.1596, "step": 2611 }, { "epoch": 1.9135531135531134, "grad_norm": 0.99609375, "learning_rate": 9.99941557533858e-05, "loss": 0.2029, "step": 2612 }, { "epoch": 1.9142857142857141, "grad_norm": 1.21875, "learning_rate": 9.999410298498011e-05, "loss": 0.2164, "step": 2613 }, { "epoch": 1.9150183150183149, "grad_norm": 0.75, "learning_rate": 9.999404997943196e-05, "loss": 0.169, "step": 2614 }, { "epoch": 1.9157509157509156, "grad_norm": 1.0390625, "learning_rate": 9.999399673674155e-05, "loss": 0.1878, "step": 2615 }, { "epoch": 1.9164835164835163, "grad_norm": 0.98046875, "learning_rate": 9.999394325690917e-05, "loss": 0.2, "step": 2616 }, { "epoch": 1.917216117216117, "grad_norm": 1.046875, "learning_rate": 9.999388953993506e-05, "loss": 0.1664, "step": 2617 }, { "epoch": 1.9179487179487178, "grad_norm": 1.359375, "learning_rate": 9.999383558581948e-05, "loss": 0.2001, "step": 2618 }, { "epoch": 1.9186813186813185, "grad_norm": 1.40625, "learning_rate": 9.999378139456266e-05, "loss": 0.2103, "step": 2619 }, { "epoch": 1.9194139194139193, "grad_norm": 1.0625, "learning_rate": 9.999372696616491e-05, "loss": 0.2051, "step": 2620 }, { "epoch": 1.92014652014652, "grad_norm": 0.84375, "learning_rate": 9.999367230062647e-05, "loss": 0.1976, "step": 2621 }, { "epoch": 1.9208791208791207, "grad_norm": 1.328125, "learning_rate": 9.999361739794758e-05, "loss": 0.1917, "step": 2622 }, { "epoch": 1.9216117216117214, "grad_norm": 1.4140625, "learning_rate": 9.999356225812853e-05, "loss": 0.1972, "step": 2623 }, { "epoch": 1.9223443223443224, "grad_norm": 1.5546875, "learning_rate": 9.999350688116957e-05, "loss": 0.2067, "step": 2624 }, { "epoch": 1.9230769230769231, "grad_norm": 0.9140625, "learning_rate": 9.999345126707098e-05, "loss": 0.2006, "step": 2625 }, { "epoch": 1.9238095238095239, "grad_norm": 1.0859375, "learning_rate": 9.9993395415833e-05, "loss": 0.1956, "step": 2626 }, { "epoch": 1.9245421245421246, "grad_norm": 1.3125, "learning_rate": 9.999333932745593e-05, "loss": 0.1829, "step": 2627 }, { "epoch": 1.9252747252747253, "grad_norm": 1.1796875, "learning_rate": 9.999328300194003e-05, "loss": 0.2131, "step": 2628 }, { "epoch": 1.926007326007326, "grad_norm": 0.8515625, "learning_rate": 9.999322643928557e-05, "loss": 0.1835, "step": 2629 }, { "epoch": 1.9267399267399268, "grad_norm": 1.1953125, "learning_rate": 9.999316963949279e-05, "loss": 0.1625, "step": 2630 }, { "epoch": 1.9274725274725275, "grad_norm": 1.4453125, "learning_rate": 9.999311260256199e-05, "loss": 0.2143, "step": 2631 }, { "epoch": 1.9282051282051282, "grad_norm": 0.8828125, "learning_rate": 9.999305532849344e-05, "loss": 0.1504, "step": 2632 }, { "epoch": 1.928937728937729, "grad_norm": 0.79296875, "learning_rate": 9.999299781728743e-05, "loss": 0.1641, "step": 2633 }, { "epoch": 1.9296703296703297, "grad_norm": 1.0625, "learning_rate": 9.999294006894421e-05, "loss": 0.1943, "step": 2634 }, { "epoch": 1.9304029304029304, "grad_norm": 0.9453125, "learning_rate": 9.999288208346405e-05, "loss": 0.2019, "step": 2635 }, { "epoch": 1.9311355311355312, "grad_norm": 0.80859375, "learning_rate": 9.999282386084727e-05, "loss": 0.1729, "step": 2636 }, { "epoch": 1.9318681318681319, "grad_norm": 1.140625, "learning_rate": 9.999276540109412e-05, "loss": 0.1624, "step": 2637 }, { "epoch": 1.9326007326007326, "grad_norm": 1.2109375, "learning_rate": 9.999270670420488e-05, "loss": 0.1837, "step": 2638 }, { "epoch": 1.9333333333333333, "grad_norm": 0.59765625, "learning_rate": 9.999264777017984e-05, "loss": 0.1492, "step": 2639 }, { "epoch": 1.934065934065934, "grad_norm": 0.7578125, "learning_rate": 9.999258859901927e-05, "loss": 0.1748, "step": 2640 }, { "epoch": 1.9347985347985348, "grad_norm": 1.171875, "learning_rate": 9.999252919072346e-05, "loss": 0.2191, "step": 2641 }, { "epoch": 1.9355311355311355, "grad_norm": 1.578125, "learning_rate": 9.999246954529269e-05, "loss": 0.1897, "step": 2642 }, { "epoch": 1.9362637362637363, "grad_norm": 1.296875, "learning_rate": 9.999240966272726e-05, "loss": 0.2551, "step": 2643 }, { "epoch": 1.936996336996337, "grad_norm": 1.203125, "learning_rate": 9.999234954302744e-05, "loss": 0.1804, "step": 2644 }, { "epoch": 1.9377289377289377, "grad_norm": 1.6796875, "learning_rate": 9.999228918619353e-05, "loss": 0.2033, "step": 2645 }, { "epoch": 1.9384615384615385, "grad_norm": 1.6875, "learning_rate": 9.999222859222582e-05, "loss": 0.2045, "step": 2646 }, { "epoch": 1.9391941391941392, "grad_norm": 1.0625, "learning_rate": 9.999216776112461e-05, "loss": 0.1605, "step": 2647 }, { "epoch": 1.93992673992674, "grad_norm": 0.86328125, "learning_rate": 9.999210669289015e-05, "loss": 0.195, "step": 2648 }, { "epoch": 1.9406593406593406, "grad_norm": 1.3828125, "learning_rate": 9.999204538752276e-05, "loss": 0.1791, "step": 2649 }, { "epoch": 1.9413919413919414, "grad_norm": 0.87109375, "learning_rate": 9.999198384502276e-05, "loss": 0.1757, "step": 2650 }, { "epoch": 1.942124542124542, "grad_norm": 1.2890625, "learning_rate": 9.999192206539039e-05, "loss": 0.2758, "step": 2651 }, { "epoch": 1.9428571428571428, "grad_norm": 1.2578125, "learning_rate": 9.999186004862598e-05, "loss": 0.2321, "step": 2652 }, { "epoch": 1.9435897435897436, "grad_norm": 1.1328125, "learning_rate": 9.999179779472983e-05, "loss": 0.2114, "step": 2653 }, { "epoch": 1.9443223443223443, "grad_norm": 1.7109375, "learning_rate": 9.999173530370224e-05, "loss": 0.2301, "step": 2654 }, { "epoch": 1.945054945054945, "grad_norm": 0.97265625, "learning_rate": 9.999167257554348e-05, "loss": 0.2168, "step": 2655 }, { "epoch": 1.9457875457875458, "grad_norm": 1.3828125, "learning_rate": 9.999160961025386e-05, "loss": 0.1913, "step": 2656 }, { "epoch": 1.9465201465201465, "grad_norm": 1.3359375, "learning_rate": 9.999154640783372e-05, "loss": 0.172, "step": 2657 }, { "epoch": 1.9472527472527472, "grad_norm": 0.91015625, "learning_rate": 9.999148296828331e-05, "loss": 0.1836, "step": 2658 }, { "epoch": 1.947985347985348, "grad_norm": 1.1171875, "learning_rate": 9.999141929160298e-05, "loss": 0.2118, "step": 2659 }, { "epoch": 1.9487179487179487, "grad_norm": 1.46875, "learning_rate": 9.9991355377793e-05, "loss": 0.2186, "step": 2660 }, { "epoch": 1.9494505494505494, "grad_norm": 1.3984375, "learning_rate": 9.99912912268537e-05, "loss": 0.2409, "step": 2661 }, { "epoch": 1.9501831501831501, "grad_norm": 0.7109375, "learning_rate": 9.999122683878536e-05, "loss": 0.1724, "step": 2662 }, { "epoch": 1.9509157509157509, "grad_norm": 2.125, "learning_rate": 9.99911622135883e-05, "loss": 0.2181, "step": 2663 }, { "epoch": 1.9516483516483516, "grad_norm": 1.28125, "learning_rate": 9.999109735126285e-05, "loss": 0.1892, "step": 2664 }, { "epoch": 1.9523809523809523, "grad_norm": 1.4140625, "learning_rate": 9.999103225180931e-05, "loss": 0.1895, "step": 2665 }, { "epoch": 1.953113553113553, "grad_norm": 1.03125, "learning_rate": 9.999096691522798e-05, "loss": 0.2051, "step": 2666 }, { "epoch": 1.953846153846154, "grad_norm": 1.421875, "learning_rate": 9.99909013415192e-05, "loss": 0.2685, "step": 2667 }, { "epoch": 1.9545787545787547, "grad_norm": 1.4140625, "learning_rate": 9.999083553068324e-05, "loss": 0.1949, "step": 2668 }, { "epoch": 1.9553113553113555, "grad_norm": 0.6640625, "learning_rate": 9.999076948272044e-05, "loss": 0.1617, "step": 2669 }, { "epoch": 1.9560439560439562, "grad_norm": 0.9609375, "learning_rate": 9.999070319763113e-05, "loss": 0.1579, "step": 2670 }, { "epoch": 1.956776556776557, "grad_norm": 1.203125, "learning_rate": 9.999063667541559e-05, "loss": 0.2131, "step": 2671 }, { "epoch": 1.9575091575091577, "grad_norm": 1.703125, "learning_rate": 9.999056991607421e-05, "loss": 0.1681, "step": 2672 }, { "epoch": 1.9582417582417584, "grad_norm": 1.5234375, "learning_rate": 9.999050291960721e-05, "loss": 0.1829, "step": 2673 }, { "epoch": 1.9589743589743591, "grad_norm": 0.7890625, "learning_rate": 9.9990435686015e-05, "loss": 0.1629, "step": 2674 }, { "epoch": 1.9597069597069599, "grad_norm": 1.0703125, "learning_rate": 9.999036821529785e-05, "loss": 0.2686, "step": 2675 }, { "epoch": 1.9604395604395606, "grad_norm": 1.109375, "learning_rate": 9.999030050745611e-05, "loss": 0.1864, "step": 2676 }, { "epoch": 1.9611721611721613, "grad_norm": 0.73046875, "learning_rate": 9.999023256249008e-05, "loss": 0.1516, "step": 2677 }, { "epoch": 1.961904761904762, "grad_norm": 0.94921875, "learning_rate": 9.99901643804001e-05, "loss": 0.1861, "step": 2678 }, { "epoch": 1.9626373626373628, "grad_norm": 0.9921875, "learning_rate": 9.999009596118652e-05, "loss": 0.1873, "step": 2679 }, { "epoch": 1.9633699633699635, "grad_norm": 1.0546875, "learning_rate": 9.999002730484962e-05, "loss": 0.1677, "step": 2680 }, { "epoch": 1.9641025641025642, "grad_norm": 1.1640625, "learning_rate": 9.998995841138976e-05, "loss": 0.2098, "step": 2681 }, { "epoch": 1.964835164835165, "grad_norm": 0.93359375, "learning_rate": 9.998988928080726e-05, "loss": 0.1553, "step": 2682 }, { "epoch": 1.9655677655677657, "grad_norm": 0.703125, "learning_rate": 9.998981991310245e-05, "loss": 0.1643, "step": 2683 }, { "epoch": 1.9663003663003664, "grad_norm": 1.1953125, "learning_rate": 9.998975030827568e-05, "loss": 0.1631, "step": 2684 }, { "epoch": 1.9670329670329672, "grad_norm": 1.4921875, "learning_rate": 9.998968046632727e-05, "loss": 0.1911, "step": 2685 }, { "epoch": 1.9677655677655679, "grad_norm": 1.3359375, "learning_rate": 9.998961038725752e-05, "loss": 0.1799, "step": 2686 }, { "epoch": 1.9684981684981686, "grad_norm": 1.75, "learning_rate": 9.998954007106683e-05, "loss": 0.25, "step": 2687 }, { "epoch": 1.9692307692307693, "grad_norm": 1.1640625, "learning_rate": 9.99894695177555e-05, "loss": 0.2442, "step": 2688 }, { "epoch": 1.96996336996337, "grad_norm": 1.078125, "learning_rate": 9.998939872732388e-05, "loss": 0.2031, "step": 2689 }, { "epoch": 1.9706959706959708, "grad_norm": 1.2109375, "learning_rate": 9.998932769977231e-05, "loss": 0.1688, "step": 2690 }, { "epoch": 1.9714285714285715, "grad_norm": 1.2734375, "learning_rate": 9.998925643510112e-05, "loss": 0.1839, "step": 2691 }, { "epoch": 1.9721611721611723, "grad_norm": 0.90625, "learning_rate": 9.998918493331066e-05, "loss": 0.1573, "step": 2692 }, { "epoch": 1.972893772893773, "grad_norm": 1.3203125, "learning_rate": 9.998911319440125e-05, "loss": 0.2024, "step": 2693 }, { "epoch": 1.9736263736263737, "grad_norm": 1.0859375, "learning_rate": 9.998904121837327e-05, "loss": 0.2064, "step": 2694 }, { "epoch": 1.9743589743589745, "grad_norm": 0.96484375, "learning_rate": 9.998896900522706e-05, "loss": 0.2202, "step": 2695 }, { "epoch": 1.9750915750915752, "grad_norm": 1.0859375, "learning_rate": 9.998889655496293e-05, "loss": 0.2009, "step": 2696 }, { "epoch": 1.975824175824176, "grad_norm": 1.34375, "learning_rate": 9.998882386758125e-05, "loss": 0.187, "step": 2697 }, { "epoch": 1.9765567765567766, "grad_norm": 0.93359375, "learning_rate": 9.998875094308237e-05, "loss": 0.1636, "step": 2698 }, { "epoch": 1.9772893772893774, "grad_norm": 1.109375, "learning_rate": 9.998867778146664e-05, "loss": 0.2253, "step": 2699 }, { "epoch": 1.978021978021978, "grad_norm": 1.4375, "learning_rate": 9.99886043827344e-05, "loss": 0.1916, "step": 2700 }, { "epoch": 1.9787545787545788, "grad_norm": 1.3359375, "learning_rate": 9.998853074688603e-05, "loss": 0.1986, "step": 2701 }, { "epoch": 1.9794871794871796, "grad_norm": 0.8359375, "learning_rate": 9.998845687392187e-05, "loss": 0.1785, "step": 2702 }, { "epoch": 1.9802197802197803, "grad_norm": 1.8203125, "learning_rate": 9.998838276384223e-05, "loss": 0.2299, "step": 2703 }, { "epoch": 1.980952380952381, "grad_norm": 1.5078125, "learning_rate": 9.998830841664753e-05, "loss": 0.2077, "step": 2704 }, { "epoch": 1.9816849816849818, "grad_norm": 1.2265625, "learning_rate": 9.99882338323381e-05, "loss": 0.178, "step": 2705 }, { "epoch": 1.9824175824175825, "grad_norm": 1.078125, "learning_rate": 9.998815901091429e-05, "loss": 0.1638, "step": 2706 }, { "epoch": 1.9831501831501832, "grad_norm": 0.96484375, "learning_rate": 9.998808395237647e-05, "loss": 0.1885, "step": 2707 }, { "epoch": 1.983882783882784, "grad_norm": 1.0078125, "learning_rate": 9.998800865672497e-05, "loss": 0.2282, "step": 2708 }, { "epoch": 1.9846153846153847, "grad_norm": 1.09375, "learning_rate": 9.99879331239602e-05, "loss": 0.2106, "step": 2709 }, { "epoch": 1.9853479853479854, "grad_norm": 0.6953125, "learning_rate": 9.998785735408248e-05, "loss": 0.1704, "step": 2710 }, { "epoch": 1.9860805860805861, "grad_norm": 1.390625, "learning_rate": 9.99877813470922e-05, "loss": 0.2022, "step": 2711 }, { "epoch": 1.9868131868131869, "grad_norm": 1.0, "learning_rate": 9.998770510298971e-05, "loss": 0.2088, "step": 2712 }, { "epoch": 1.9875457875457876, "grad_norm": 1.0625, "learning_rate": 9.998762862177539e-05, "loss": 0.1918, "step": 2713 }, { "epoch": 1.9882783882783883, "grad_norm": 2.78125, "learning_rate": 9.99875519034496e-05, "loss": 0.2078, "step": 2714 }, { "epoch": 1.989010989010989, "grad_norm": 1.3125, "learning_rate": 9.998747494801267e-05, "loss": 0.1826, "step": 2715 }, { "epoch": 1.9897435897435898, "grad_norm": 1.0390625, "learning_rate": 9.998739775546503e-05, "loss": 0.1626, "step": 2716 }, { "epoch": 1.9904761904761905, "grad_norm": 0.9453125, "learning_rate": 9.998732032580701e-05, "loss": 0.2097, "step": 2717 }, { "epoch": 1.9912087912087912, "grad_norm": 1.4453125, "learning_rate": 9.9987242659039e-05, "loss": 0.2592, "step": 2718 }, { "epoch": 1.991941391941392, "grad_norm": 1.1484375, "learning_rate": 9.998716475516135e-05, "loss": 0.1738, "step": 2719 }, { "epoch": 1.9926739926739927, "grad_norm": 1.4609375, "learning_rate": 9.998708661417447e-05, "loss": 0.1923, "step": 2720 }, { "epoch": 1.9934065934065934, "grad_norm": 1.0546875, "learning_rate": 9.998700823607869e-05, "loss": 0.1902, "step": 2721 }, { "epoch": 1.9941391941391942, "grad_norm": 1.515625, "learning_rate": 9.998692962087442e-05, "loss": 0.2045, "step": 2722 }, { "epoch": 1.994871794871795, "grad_norm": 1.2578125, "learning_rate": 9.998685076856202e-05, "loss": 0.2338, "step": 2723 }, { "epoch": 1.9956043956043956, "grad_norm": 1.34375, "learning_rate": 9.998677167914188e-05, "loss": 0.1982, "step": 2724 }, { "epoch": 1.9963369963369964, "grad_norm": 1.3671875, "learning_rate": 9.998669235261437e-05, "loss": 0.1581, "step": 2725 }, { "epoch": 1.997069597069597, "grad_norm": 1.265625, "learning_rate": 9.998661278897987e-05, "loss": 0.221, "step": 2726 }, { "epoch": 1.9978021978021978, "grad_norm": 1.2109375, "learning_rate": 9.998653298823875e-05, "loss": 0.1706, "step": 2727 }, { "epoch": 1.9985347985347985, "grad_norm": 1.5078125, "learning_rate": 9.998645295039143e-05, "loss": 0.1807, "step": 2728 }, { "epoch": 1.9992673992673993, "grad_norm": 2.046875, "learning_rate": 9.998637267543825e-05, "loss": 0.1775, "step": 2729 }, { "epoch": 2.0, "grad_norm": 1.6640625, "learning_rate": 9.998629216337962e-05, "loss": 0.1811, "step": 2730 }, { "epoch": 2.0007326007326007, "grad_norm": 0.91015625, "learning_rate": 9.998621141421593e-05, "loss": 0.2318, "step": 2731 }, { "epoch": 2.0014652014652015, "grad_norm": 1.1796875, "learning_rate": 9.998613042794754e-05, "loss": 0.2144, "step": 2732 }, { "epoch": 2.002197802197802, "grad_norm": 0.94921875, "learning_rate": 9.998604920457485e-05, "loss": 0.1911, "step": 2733 }, { "epoch": 2.002930402930403, "grad_norm": 1.703125, "learning_rate": 9.998596774409826e-05, "loss": 0.2218, "step": 2734 }, { "epoch": 2.0036630036630036, "grad_norm": 1.2578125, "learning_rate": 9.998588604651816e-05, "loss": 0.2103, "step": 2735 }, { "epoch": 2.0043956043956044, "grad_norm": 1.1171875, "learning_rate": 9.998580411183492e-05, "loss": 0.1823, "step": 2736 }, { "epoch": 2.005128205128205, "grad_norm": 1.21875, "learning_rate": 9.998572194004896e-05, "loss": 0.1714, "step": 2737 }, { "epoch": 2.005860805860806, "grad_norm": 1.2578125, "learning_rate": 9.998563953116065e-05, "loss": 0.193, "step": 2738 }, { "epoch": 2.0065934065934066, "grad_norm": 0.67578125, "learning_rate": 9.998555688517041e-05, "loss": 0.1774, "step": 2739 }, { "epoch": 2.0073260073260073, "grad_norm": 0.95703125, "learning_rate": 9.998547400207861e-05, "loss": 0.1716, "step": 2740 }, { "epoch": 2.008058608058608, "grad_norm": 1.2265625, "learning_rate": 9.998539088188567e-05, "loss": 0.1616, "step": 2741 }, { "epoch": 2.0087912087912088, "grad_norm": 1.21875, "learning_rate": 9.998530752459196e-05, "loss": 0.2038, "step": 2742 }, { "epoch": 2.0095238095238095, "grad_norm": 1.6328125, "learning_rate": 9.998522393019791e-05, "loss": 0.1756, "step": 2743 }, { "epoch": 2.01025641025641, "grad_norm": 0.96875, "learning_rate": 9.998514009870392e-05, "loss": 0.1937, "step": 2744 }, { "epoch": 2.010989010989011, "grad_norm": 0.953125, "learning_rate": 9.998505603011036e-05, "loss": 0.1656, "step": 2745 }, { "epoch": 2.0117216117216117, "grad_norm": 1.90625, "learning_rate": 9.998497172441767e-05, "loss": 0.1877, "step": 2746 }, { "epoch": 2.0124542124542124, "grad_norm": 1.015625, "learning_rate": 9.99848871816262e-05, "loss": 0.157, "step": 2747 }, { "epoch": 2.013186813186813, "grad_norm": 1.0859375, "learning_rate": 9.998480240173643e-05, "loss": 0.1902, "step": 2748 }, { "epoch": 2.013919413919414, "grad_norm": 1.25, "learning_rate": 9.998471738474869e-05, "loss": 0.1898, "step": 2749 }, { "epoch": 2.0146520146520146, "grad_norm": 1.03125, "learning_rate": 9.998463213066344e-05, "loss": 0.1742, "step": 2750 }, { "epoch": 2.0153846153846153, "grad_norm": 1.3984375, "learning_rate": 9.998454663948109e-05, "loss": 0.2253, "step": 2751 }, { "epoch": 2.016117216117216, "grad_norm": 2.421875, "learning_rate": 9.998446091120201e-05, "loss": 0.2283, "step": 2752 }, { "epoch": 2.016849816849817, "grad_norm": 0.875, "learning_rate": 9.998437494582665e-05, "loss": 0.2075, "step": 2753 }, { "epoch": 2.0175824175824175, "grad_norm": 1.0, "learning_rate": 9.99842887433554e-05, "loss": 0.1438, "step": 2754 }, { "epoch": 2.0183150183150182, "grad_norm": 1.53125, "learning_rate": 9.998420230378868e-05, "loss": 0.1743, "step": 2755 }, { "epoch": 2.019047619047619, "grad_norm": 0.875, "learning_rate": 9.998411562712689e-05, "loss": 0.1807, "step": 2756 }, { "epoch": 2.0197802197802197, "grad_norm": 1.4140625, "learning_rate": 9.998402871337045e-05, "loss": 0.1992, "step": 2757 }, { "epoch": 2.0205128205128204, "grad_norm": 1.1875, "learning_rate": 9.99839415625198e-05, "loss": 0.2003, "step": 2758 }, { "epoch": 2.021245421245421, "grad_norm": 0.65234375, "learning_rate": 9.998385417457534e-05, "loss": 0.1597, "step": 2759 }, { "epoch": 2.021978021978022, "grad_norm": 0.921875, "learning_rate": 9.998376654953749e-05, "loss": 0.1622, "step": 2760 }, { "epoch": 2.0227106227106226, "grad_norm": 3.046875, "learning_rate": 9.998367868740668e-05, "loss": 0.1969, "step": 2761 }, { "epoch": 2.0234432234432234, "grad_norm": 1.84375, "learning_rate": 9.99835905881833e-05, "loss": 0.162, "step": 2762 }, { "epoch": 2.024175824175824, "grad_norm": 1.6875, "learning_rate": 9.998350225186782e-05, "loss": 0.1562, "step": 2763 }, { "epoch": 2.024908424908425, "grad_norm": 1.0625, "learning_rate": 9.998341367846061e-05, "loss": 0.2133, "step": 2764 }, { "epoch": 2.0256410256410255, "grad_norm": 1.640625, "learning_rate": 9.998332486796213e-05, "loss": 0.2207, "step": 2765 }, { "epoch": 2.0263736263736263, "grad_norm": 1.0390625, "learning_rate": 9.998323582037279e-05, "loss": 0.2002, "step": 2766 }, { "epoch": 2.027106227106227, "grad_norm": 1.3203125, "learning_rate": 9.998314653569303e-05, "loss": 0.1949, "step": 2767 }, { "epoch": 2.0278388278388277, "grad_norm": 1.2578125, "learning_rate": 9.998305701392327e-05, "loss": 0.1599, "step": 2768 }, { "epoch": 2.0285714285714285, "grad_norm": 0.84765625, "learning_rate": 9.998296725506395e-05, "loss": 0.1687, "step": 2769 }, { "epoch": 2.029304029304029, "grad_norm": 1.3125, "learning_rate": 9.998287725911547e-05, "loss": 0.1892, "step": 2770 }, { "epoch": 2.03003663003663, "grad_norm": 2.4375, "learning_rate": 9.998278702607829e-05, "loss": 0.1698, "step": 2771 }, { "epoch": 2.0307692307692307, "grad_norm": 2.421875, "learning_rate": 9.998269655595284e-05, "loss": 0.2128, "step": 2772 }, { "epoch": 2.0315018315018314, "grad_norm": 1.640625, "learning_rate": 9.998260584873954e-05, "loss": 0.1468, "step": 2773 }, { "epoch": 2.032234432234432, "grad_norm": 1.0, "learning_rate": 9.998251490443884e-05, "loss": 0.1612, "step": 2774 }, { "epoch": 2.032967032967033, "grad_norm": 1.7265625, "learning_rate": 9.998242372305116e-05, "loss": 0.2208, "step": 2775 }, { "epoch": 2.0336996336996336, "grad_norm": 2.703125, "learning_rate": 9.998233230457693e-05, "loss": 0.2107, "step": 2776 }, { "epoch": 2.0344322344322343, "grad_norm": 1.484375, "learning_rate": 9.998224064901662e-05, "loss": 0.2253, "step": 2777 }, { "epoch": 2.035164835164835, "grad_norm": 1.015625, "learning_rate": 9.998214875637064e-05, "loss": 0.1763, "step": 2778 }, { "epoch": 2.0358974358974358, "grad_norm": 1.203125, "learning_rate": 9.998205662663946e-05, "loss": 0.1812, "step": 2779 }, { "epoch": 2.0366300366300365, "grad_norm": 2.0625, "learning_rate": 9.998196425982348e-05, "loss": 0.1887, "step": 2780 }, { "epoch": 2.0373626373626372, "grad_norm": 1.3359375, "learning_rate": 9.998187165592317e-05, "loss": 0.2112, "step": 2781 }, { "epoch": 2.038095238095238, "grad_norm": 1.4765625, "learning_rate": 9.998177881493899e-05, "loss": 0.1716, "step": 2782 }, { "epoch": 2.0388278388278387, "grad_norm": 1.7421875, "learning_rate": 9.998168573687133e-05, "loss": 0.1803, "step": 2783 }, { "epoch": 2.0395604395604394, "grad_norm": 1.1953125, "learning_rate": 9.998159242172069e-05, "loss": 0.1724, "step": 2784 }, { "epoch": 2.04029304029304, "grad_norm": 1.0546875, "learning_rate": 9.998149886948751e-05, "loss": 0.1887, "step": 2785 }, { "epoch": 2.041025641025641, "grad_norm": 1.0390625, "learning_rate": 9.998140508017221e-05, "loss": 0.1474, "step": 2786 }, { "epoch": 2.0417582417582416, "grad_norm": 1.015625, "learning_rate": 9.998131105377526e-05, "loss": 0.1409, "step": 2787 }, { "epoch": 2.0424908424908423, "grad_norm": 0.8828125, "learning_rate": 9.998121679029709e-05, "loss": 0.1898, "step": 2788 }, { "epoch": 2.043223443223443, "grad_norm": 0.8359375, "learning_rate": 9.998112228973817e-05, "loss": 0.174, "step": 2789 }, { "epoch": 2.043956043956044, "grad_norm": 1.3046875, "learning_rate": 9.998102755209897e-05, "loss": 0.2056, "step": 2790 }, { "epoch": 2.0446886446886445, "grad_norm": 0.83984375, "learning_rate": 9.99809325773799e-05, "loss": 0.1737, "step": 2791 }, { "epoch": 2.0454212454212453, "grad_norm": 0.95703125, "learning_rate": 9.998083736558144e-05, "loss": 0.194, "step": 2792 }, { "epoch": 2.046153846153846, "grad_norm": 1.390625, "learning_rate": 9.998074191670406e-05, "loss": 0.188, "step": 2793 }, { "epoch": 2.0468864468864467, "grad_norm": 1.640625, "learning_rate": 9.99806462307482e-05, "loss": 0.2078, "step": 2794 }, { "epoch": 2.0476190476190474, "grad_norm": 1.5234375, "learning_rate": 9.998055030771432e-05, "loss": 0.1873, "step": 2795 }, { "epoch": 2.048351648351648, "grad_norm": 1.59375, "learning_rate": 9.998045414760288e-05, "loss": 0.2465, "step": 2796 }, { "epoch": 2.049084249084249, "grad_norm": 1.6171875, "learning_rate": 9.998035775041432e-05, "loss": 0.2347, "step": 2797 }, { "epoch": 2.0498168498168496, "grad_norm": 1.3046875, "learning_rate": 9.998026111614913e-05, "loss": 0.1744, "step": 2798 }, { "epoch": 2.0505494505494504, "grad_norm": 1.53125, "learning_rate": 9.998016424480779e-05, "loss": 0.1884, "step": 2799 }, { "epoch": 2.051282051282051, "grad_norm": 1.4765625, "learning_rate": 9.998006713639073e-05, "loss": 0.1966, "step": 2800 }, { "epoch": 2.052014652014652, "grad_norm": 1.1953125, "learning_rate": 9.997996979089841e-05, "loss": 0.1825, "step": 2801 }, { "epoch": 2.0527472527472526, "grad_norm": 2.453125, "learning_rate": 9.997987220833133e-05, "loss": 0.1673, "step": 2802 }, { "epoch": 2.0534798534798533, "grad_norm": 1.421875, "learning_rate": 9.997977438868993e-05, "loss": 0.2382, "step": 2803 }, { "epoch": 2.054212454212454, "grad_norm": 1.578125, "learning_rate": 9.997967633197468e-05, "loss": 0.1691, "step": 2804 }, { "epoch": 2.0549450549450547, "grad_norm": 2.390625, "learning_rate": 9.997957803818606e-05, "loss": 0.1711, "step": 2805 }, { "epoch": 2.0556776556776555, "grad_norm": 3.578125, "learning_rate": 9.997947950732455e-05, "loss": 0.1974, "step": 2806 }, { "epoch": 2.056410256410256, "grad_norm": 2.53125, "learning_rate": 9.997938073939062e-05, "loss": 0.1965, "step": 2807 }, { "epoch": 2.057142857142857, "grad_norm": 1.421875, "learning_rate": 9.99792817343847e-05, "loss": 0.22, "step": 2808 }, { "epoch": 2.0578754578754577, "grad_norm": 1.3828125, "learning_rate": 9.997918249230732e-05, "loss": 0.1826, "step": 2809 }, { "epoch": 2.0586080586080584, "grad_norm": 1.140625, "learning_rate": 9.997908301315893e-05, "loss": 0.1762, "step": 2810 }, { "epoch": 2.059340659340659, "grad_norm": 0.94140625, "learning_rate": 9.997898329694002e-05, "loss": 0.161, "step": 2811 }, { "epoch": 2.06007326007326, "grad_norm": 1.140625, "learning_rate": 9.997888334365104e-05, "loss": 0.1781, "step": 2812 }, { "epoch": 2.0608058608058606, "grad_norm": 0.8671875, "learning_rate": 9.99787831532925e-05, "loss": 0.1858, "step": 2813 }, { "epoch": 2.0615384615384613, "grad_norm": 0.74609375, "learning_rate": 9.997868272586485e-05, "loss": 0.1649, "step": 2814 }, { "epoch": 2.062271062271062, "grad_norm": 1.3203125, "learning_rate": 9.997858206136862e-05, "loss": 0.2002, "step": 2815 }, { "epoch": 2.063003663003663, "grad_norm": 1.2265625, "learning_rate": 9.997848115980424e-05, "loss": 0.1976, "step": 2816 }, { "epoch": 2.0637362637362635, "grad_norm": 1.3828125, "learning_rate": 9.997838002117223e-05, "loss": 0.1689, "step": 2817 }, { "epoch": 2.0644688644688647, "grad_norm": 0.734375, "learning_rate": 9.997827864547303e-05, "loss": 0.1813, "step": 2818 }, { "epoch": 2.0652014652014654, "grad_norm": 1.8828125, "learning_rate": 9.997817703270716e-05, "loss": 0.1817, "step": 2819 }, { "epoch": 2.065934065934066, "grad_norm": 1.4765625, "learning_rate": 9.997807518287511e-05, "loss": 0.2254, "step": 2820 }, { "epoch": 2.066666666666667, "grad_norm": 1.28125, "learning_rate": 9.997797309597737e-05, "loss": 0.1846, "step": 2821 }, { "epoch": 2.0673992673992676, "grad_norm": 1.2734375, "learning_rate": 9.99778707720144e-05, "loss": 0.1786, "step": 2822 }, { "epoch": 2.0681318681318683, "grad_norm": 0.8828125, "learning_rate": 9.997776821098671e-05, "loss": 0.1774, "step": 2823 }, { "epoch": 2.068864468864469, "grad_norm": 1.484375, "learning_rate": 9.99776654128948e-05, "loss": 0.1627, "step": 2824 }, { "epoch": 2.06959706959707, "grad_norm": 1.0234375, "learning_rate": 9.997756237773915e-05, "loss": 0.2241, "step": 2825 }, { "epoch": 2.0703296703296705, "grad_norm": 1.8671875, "learning_rate": 9.997745910552025e-05, "loss": 0.2067, "step": 2826 }, { "epoch": 2.0710622710622713, "grad_norm": 1.109375, "learning_rate": 9.997735559623862e-05, "loss": 0.1806, "step": 2827 }, { "epoch": 2.071794871794872, "grad_norm": 1.1953125, "learning_rate": 9.997725184989472e-05, "loss": 0.2034, "step": 2828 }, { "epoch": 2.0725274725274727, "grad_norm": 1.28125, "learning_rate": 9.997714786648907e-05, "loss": 0.1773, "step": 2829 }, { "epoch": 2.0732600732600734, "grad_norm": 0.89453125, "learning_rate": 9.997704364602215e-05, "loss": 0.2059, "step": 2830 }, { "epoch": 2.073992673992674, "grad_norm": 0.703125, "learning_rate": 9.997693918849448e-05, "loss": 0.1759, "step": 2831 }, { "epoch": 2.074725274725275, "grad_norm": 1.3515625, "learning_rate": 9.997683449390657e-05, "loss": 0.194, "step": 2832 }, { "epoch": 2.0754578754578756, "grad_norm": 1.015625, "learning_rate": 9.997672956225888e-05, "loss": 0.1504, "step": 2833 }, { "epoch": 2.0761904761904764, "grad_norm": 0.6796875, "learning_rate": 9.997662439355194e-05, "loss": 0.1625, "step": 2834 }, { "epoch": 2.076923076923077, "grad_norm": 0.921875, "learning_rate": 9.997651898778627e-05, "loss": 0.1725, "step": 2835 }, { "epoch": 2.077655677655678, "grad_norm": 1.0703125, "learning_rate": 9.997641334496232e-05, "loss": 0.2116, "step": 2836 }, { "epoch": 2.0783882783882786, "grad_norm": 1.0078125, "learning_rate": 9.997630746508065e-05, "loss": 0.229, "step": 2837 }, { "epoch": 2.0791208791208793, "grad_norm": 1.125, "learning_rate": 9.997620134814177e-05, "loss": 0.1767, "step": 2838 }, { "epoch": 2.07985347985348, "grad_norm": 1.9921875, "learning_rate": 9.997609499414614e-05, "loss": 0.2227, "step": 2839 }, { "epoch": 2.0805860805860807, "grad_norm": 1.9140625, "learning_rate": 9.99759884030943e-05, "loss": 0.2523, "step": 2840 }, { "epoch": 2.0813186813186815, "grad_norm": 1.234375, "learning_rate": 9.997588157498676e-05, "loss": 0.1778, "step": 2841 }, { "epoch": 2.082051282051282, "grad_norm": 1.3671875, "learning_rate": 9.997577450982402e-05, "loss": 0.1848, "step": 2842 }, { "epoch": 2.082783882783883, "grad_norm": 1.625, "learning_rate": 9.997566720760663e-05, "loss": 0.1857, "step": 2843 }, { "epoch": 2.0835164835164837, "grad_norm": 1.40625, "learning_rate": 9.997555966833506e-05, "loss": 0.1918, "step": 2844 }, { "epoch": 2.0842490842490844, "grad_norm": 1.71875, "learning_rate": 9.997545189200985e-05, "loss": 0.2024, "step": 2845 }, { "epoch": 2.084981684981685, "grad_norm": 1.4140625, "learning_rate": 9.99753438786315e-05, "loss": 0.2054, "step": 2846 }, { "epoch": 2.085714285714286, "grad_norm": 1.125, "learning_rate": 9.997523562820052e-05, "loss": 0.1609, "step": 2847 }, { "epoch": 2.0864468864468866, "grad_norm": 1.0703125, "learning_rate": 9.997512714071747e-05, "loss": 0.1687, "step": 2848 }, { "epoch": 2.0871794871794873, "grad_norm": 0.6328125, "learning_rate": 9.997501841618284e-05, "loss": 0.1835, "step": 2849 }, { "epoch": 2.087912087912088, "grad_norm": 1.609375, "learning_rate": 9.997490945459714e-05, "loss": 0.2192, "step": 2850 }, { "epoch": 2.0886446886446888, "grad_norm": 1.5390625, "learning_rate": 9.997480025596092e-05, "loss": 0.1726, "step": 2851 }, { "epoch": 2.0893772893772895, "grad_norm": 1.5390625, "learning_rate": 9.997469082027468e-05, "loss": 0.1521, "step": 2852 }, { "epoch": 2.0901098901098902, "grad_norm": 1.4375, "learning_rate": 9.997458114753897e-05, "loss": 0.2413, "step": 2853 }, { "epoch": 2.090842490842491, "grad_norm": 1.359375, "learning_rate": 9.997447123775429e-05, "loss": 0.282, "step": 2854 }, { "epoch": 2.0915750915750917, "grad_norm": 1.421875, "learning_rate": 9.997436109092119e-05, "loss": 0.182, "step": 2855 }, { "epoch": 2.0923076923076924, "grad_norm": 1.3203125, "learning_rate": 9.997425070704016e-05, "loss": 0.1839, "step": 2856 }, { "epoch": 2.093040293040293, "grad_norm": 0.81640625, "learning_rate": 9.997414008611178e-05, "loss": 0.1634, "step": 2857 }, { "epoch": 2.093772893772894, "grad_norm": 1.03125, "learning_rate": 9.997402922813653e-05, "loss": 0.2158, "step": 2858 }, { "epoch": 2.0945054945054946, "grad_norm": 1.359375, "learning_rate": 9.997391813311499e-05, "loss": 0.1602, "step": 2859 }, { "epoch": 2.0952380952380953, "grad_norm": 1.796875, "learning_rate": 9.997380680104766e-05, "loss": 0.1863, "step": 2860 }, { "epoch": 2.095970695970696, "grad_norm": 0.7265625, "learning_rate": 9.997369523193508e-05, "loss": 0.2162, "step": 2861 }, { "epoch": 2.096703296703297, "grad_norm": 1.015625, "learning_rate": 9.997358342577775e-05, "loss": 0.1875, "step": 2862 }, { "epoch": 2.0974358974358975, "grad_norm": 1.046875, "learning_rate": 9.997347138257628e-05, "loss": 0.2361, "step": 2863 }, { "epoch": 2.0981684981684983, "grad_norm": 0.8359375, "learning_rate": 9.997335910233116e-05, "loss": 0.1746, "step": 2864 }, { "epoch": 2.098901098901099, "grad_norm": 1.234375, "learning_rate": 9.997324658504294e-05, "loss": 0.1536, "step": 2865 }, { "epoch": 2.0996336996336997, "grad_norm": 0.84765625, "learning_rate": 9.997313383071212e-05, "loss": 0.1607, "step": 2866 }, { "epoch": 2.1003663003663005, "grad_norm": 1.2578125, "learning_rate": 9.997302083933931e-05, "loss": 0.176, "step": 2867 }, { "epoch": 2.101098901098901, "grad_norm": 1.046875, "learning_rate": 9.997290761092501e-05, "loss": 0.1533, "step": 2868 }, { "epoch": 2.101831501831502, "grad_norm": 0.86328125, "learning_rate": 9.997279414546977e-05, "loss": 0.199, "step": 2869 }, { "epoch": 2.1025641025641026, "grad_norm": 0.78515625, "learning_rate": 9.997268044297413e-05, "loss": 0.1438, "step": 2870 }, { "epoch": 2.1032967032967034, "grad_norm": 2.328125, "learning_rate": 9.997256650343864e-05, "loss": 0.1968, "step": 2871 }, { "epoch": 2.104029304029304, "grad_norm": 1.078125, "learning_rate": 9.997245232686383e-05, "loss": 0.1825, "step": 2872 }, { "epoch": 2.104761904761905, "grad_norm": 1.2265625, "learning_rate": 9.997233791325029e-05, "loss": 0.1765, "step": 2873 }, { "epoch": 2.1054945054945056, "grad_norm": 0.75390625, "learning_rate": 9.99722232625985e-05, "loss": 0.1891, "step": 2874 }, { "epoch": 2.1062271062271063, "grad_norm": 1.09375, "learning_rate": 9.997210837490907e-05, "loss": 0.1729, "step": 2875 }, { "epoch": 2.106959706959707, "grad_norm": 1.3828125, "learning_rate": 9.997199325018252e-05, "loss": 0.1626, "step": 2876 }, { "epoch": 2.1076923076923078, "grad_norm": 1.546875, "learning_rate": 9.997187788841941e-05, "loss": 0.1486, "step": 2877 }, { "epoch": 2.1084249084249085, "grad_norm": 0.97265625, "learning_rate": 9.997176228962032e-05, "loss": 0.1968, "step": 2878 }, { "epoch": 2.109157509157509, "grad_norm": 1.0859375, "learning_rate": 9.997164645378573e-05, "loss": 0.1788, "step": 2879 }, { "epoch": 2.10989010989011, "grad_norm": 0.57421875, "learning_rate": 9.997153038091628e-05, "loss": 0.1593, "step": 2880 }, { "epoch": 2.1106227106227107, "grad_norm": 1.921875, "learning_rate": 9.997141407101246e-05, "loss": 0.1637, "step": 2881 }, { "epoch": 2.1113553113553114, "grad_norm": 1.65625, "learning_rate": 9.997129752407487e-05, "loss": 0.1595, "step": 2882 }, { "epoch": 2.112087912087912, "grad_norm": 1.03125, "learning_rate": 9.997118074010406e-05, "loss": 0.1727, "step": 2883 }, { "epoch": 2.112820512820513, "grad_norm": 1.1953125, "learning_rate": 9.997106371910056e-05, "loss": 0.1875, "step": 2884 }, { "epoch": 2.1135531135531136, "grad_norm": 1.7734375, "learning_rate": 9.997094646106498e-05, "loss": 0.1768, "step": 2885 }, { "epoch": 2.1142857142857143, "grad_norm": 1.765625, "learning_rate": 9.997082896599784e-05, "loss": 0.2217, "step": 2886 }, { "epoch": 2.115018315018315, "grad_norm": 1.078125, "learning_rate": 9.997071123389972e-05, "loss": 0.1586, "step": 2887 }, { "epoch": 2.115750915750916, "grad_norm": 2.109375, "learning_rate": 9.997059326477118e-05, "loss": 0.2029, "step": 2888 }, { "epoch": 2.1164835164835165, "grad_norm": 1.6875, "learning_rate": 9.997047505861279e-05, "loss": 0.2117, "step": 2889 }, { "epoch": 2.1172161172161172, "grad_norm": 1.1953125, "learning_rate": 9.99703566154251e-05, "loss": 0.1563, "step": 2890 }, { "epoch": 2.117948717948718, "grad_norm": 1.8203125, "learning_rate": 9.99702379352087e-05, "loss": 0.1555, "step": 2891 }, { "epoch": 2.1186813186813187, "grad_norm": 1.109375, "learning_rate": 9.997011901796416e-05, "loss": 0.1901, "step": 2892 }, { "epoch": 2.1194139194139194, "grad_norm": 1.3515625, "learning_rate": 9.996999986369205e-05, "loss": 0.2455, "step": 2893 }, { "epoch": 2.12014652014652, "grad_norm": 1.0390625, "learning_rate": 9.99698804723929e-05, "loss": 0.1932, "step": 2894 }, { "epoch": 2.120879120879121, "grad_norm": 1.5859375, "learning_rate": 9.996976084406733e-05, "loss": 0.1845, "step": 2895 }, { "epoch": 2.1216117216117216, "grad_norm": 1.7578125, "learning_rate": 9.99696409787159e-05, "loss": 0.2176, "step": 2896 }, { "epoch": 2.1223443223443224, "grad_norm": 1.3515625, "learning_rate": 9.996952087633918e-05, "loss": 0.2216, "step": 2897 }, { "epoch": 2.123076923076923, "grad_norm": 2.0, "learning_rate": 9.996940053693773e-05, "loss": 0.2421, "step": 2898 }, { "epoch": 2.123809523809524, "grad_norm": 1.1015625, "learning_rate": 9.996927996051215e-05, "loss": 0.1479, "step": 2899 }, { "epoch": 2.1245421245421245, "grad_norm": 1.3671875, "learning_rate": 9.9969159147063e-05, "loss": 0.1746, "step": 2900 }, { "epoch": 2.1252747252747253, "grad_norm": 1.75, "learning_rate": 9.99690380965909e-05, "loss": 0.2727, "step": 2901 }, { "epoch": 2.126007326007326, "grad_norm": 1.984375, "learning_rate": 9.996891680909636e-05, "loss": 0.2205, "step": 2902 }, { "epoch": 2.1267399267399267, "grad_norm": 2.390625, "learning_rate": 9.996879528458004e-05, "loss": 0.2091, "step": 2903 }, { "epoch": 2.1274725274725275, "grad_norm": 1.15625, "learning_rate": 9.996867352304245e-05, "loss": 0.2064, "step": 2904 }, { "epoch": 2.128205128205128, "grad_norm": 1.9296875, "learning_rate": 9.996855152448422e-05, "loss": 0.2317, "step": 2905 }, { "epoch": 2.128937728937729, "grad_norm": 1.34375, "learning_rate": 9.996842928890593e-05, "loss": 0.2288, "step": 2906 }, { "epoch": 2.1296703296703297, "grad_norm": 1.0859375, "learning_rate": 9.996830681630813e-05, "loss": 0.2045, "step": 2907 }, { "epoch": 2.1304029304029304, "grad_norm": 2.328125, "learning_rate": 9.996818410669144e-05, "loss": 0.2068, "step": 2908 }, { "epoch": 2.131135531135531, "grad_norm": 1.3203125, "learning_rate": 9.996806116005646e-05, "loss": 0.1805, "step": 2909 }, { "epoch": 2.131868131868132, "grad_norm": 1.2734375, "learning_rate": 9.996793797640374e-05, "loss": 0.1532, "step": 2910 }, { "epoch": 2.1326007326007326, "grad_norm": 1.6796875, "learning_rate": 9.99678145557339e-05, "loss": 0.1874, "step": 2911 }, { "epoch": 2.1333333333333333, "grad_norm": 1.109375, "learning_rate": 9.996769089804751e-05, "loss": 0.2081, "step": 2912 }, { "epoch": 2.134065934065934, "grad_norm": 1.4140625, "learning_rate": 9.996756700334518e-05, "loss": 0.1983, "step": 2913 }, { "epoch": 2.1347985347985348, "grad_norm": 2.75, "learning_rate": 9.99674428716275e-05, "loss": 0.2246, "step": 2914 }, { "epoch": 2.1355311355311355, "grad_norm": 2.375, "learning_rate": 9.996731850289507e-05, "loss": 0.1755, "step": 2915 }, { "epoch": 2.1362637362637362, "grad_norm": 1.140625, "learning_rate": 9.996719389714846e-05, "loss": 0.1928, "step": 2916 }, { "epoch": 2.136996336996337, "grad_norm": 0.74609375, "learning_rate": 9.996706905438829e-05, "loss": 0.1749, "step": 2917 }, { "epoch": 2.1377289377289377, "grad_norm": 0.79296875, "learning_rate": 9.996694397461516e-05, "loss": 0.1808, "step": 2918 }, { "epoch": 2.1384615384615384, "grad_norm": 1.8046875, "learning_rate": 9.996681865782966e-05, "loss": 0.1935, "step": 2919 }, { "epoch": 2.139194139194139, "grad_norm": 1.7578125, "learning_rate": 9.996669310403239e-05, "loss": 0.1823, "step": 2920 }, { "epoch": 2.13992673992674, "grad_norm": 1.3125, "learning_rate": 9.996656731322397e-05, "loss": 0.1649, "step": 2921 }, { "epoch": 2.1406593406593406, "grad_norm": 0.97265625, "learning_rate": 9.996644128540497e-05, "loss": 0.2143, "step": 2922 }, { "epoch": 2.1413919413919413, "grad_norm": 1.1640625, "learning_rate": 9.9966315020576e-05, "loss": 0.2023, "step": 2923 }, { "epoch": 2.142124542124542, "grad_norm": 1.375, "learning_rate": 9.99661885187377e-05, "loss": 0.1989, "step": 2924 }, { "epoch": 2.142857142857143, "grad_norm": 0.95703125, "learning_rate": 9.996606177989064e-05, "loss": 0.1465, "step": 2925 }, { "epoch": 2.1435897435897435, "grad_norm": 1.4140625, "learning_rate": 9.996593480403544e-05, "loss": 0.1719, "step": 2926 }, { "epoch": 2.1443223443223443, "grad_norm": 0.8671875, "learning_rate": 9.996580759117272e-05, "loss": 0.1726, "step": 2927 }, { "epoch": 2.145054945054945, "grad_norm": 1.4765625, "learning_rate": 9.996568014130307e-05, "loss": 0.1927, "step": 2928 }, { "epoch": 2.1457875457875457, "grad_norm": 1.359375, "learning_rate": 9.99655524544271e-05, "loss": 0.1991, "step": 2929 }, { "epoch": 2.1465201465201464, "grad_norm": 0.6796875, "learning_rate": 9.996542453054544e-05, "loss": 0.1711, "step": 2930 }, { "epoch": 2.147252747252747, "grad_norm": 0.94140625, "learning_rate": 9.99652963696587e-05, "loss": 0.1472, "step": 2931 }, { "epoch": 2.147985347985348, "grad_norm": 1.03125, "learning_rate": 9.996516797176746e-05, "loss": 0.1701, "step": 2932 }, { "epoch": 2.1487179487179486, "grad_norm": 1.171875, "learning_rate": 9.996503933687238e-05, "loss": 0.2023, "step": 2933 }, { "epoch": 2.1494505494505494, "grad_norm": 0.96875, "learning_rate": 9.996491046497406e-05, "loss": 0.1755, "step": 2934 }, { "epoch": 2.15018315018315, "grad_norm": 0.94140625, "learning_rate": 9.99647813560731e-05, "loss": 0.1802, "step": 2935 }, { "epoch": 2.150915750915751, "grad_norm": 0.91015625, "learning_rate": 9.996465201017015e-05, "loss": 0.144, "step": 2936 }, { "epoch": 2.1516483516483516, "grad_norm": 1.0078125, "learning_rate": 9.99645224272658e-05, "loss": 0.2044, "step": 2937 }, { "epoch": 2.1523809523809523, "grad_norm": 1.1015625, "learning_rate": 9.99643926073607e-05, "loss": 0.1884, "step": 2938 }, { "epoch": 2.153113553113553, "grad_norm": 1.3828125, "learning_rate": 9.996426255045545e-05, "loss": 0.1986, "step": 2939 }, { "epoch": 2.1538461538461537, "grad_norm": 1.1484375, "learning_rate": 9.996413225655068e-05, "loss": 0.1653, "step": 2940 }, { "epoch": 2.1545787545787545, "grad_norm": 1.6328125, "learning_rate": 9.996400172564702e-05, "loss": 0.2038, "step": 2941 }, { "epoch": 2.155311355311355, "grad_norm": 1.046875, "learning_rate": 9.996387095774507e-05, "loss": 0.1604, "step": 2942 }, { "epoch": 2.156043956043956, "grad_norm": 2.0625, "learning_rate": 9.99637399528455e-05, "loss": 0.1752, "step": 2943 }, { "epoch": 2.1567765567765567, "grad_norm": 2.796875, "learning_rate": 9.99636087109489e-05, "loss": 0.2375, "step": 2944 }, { "epoch": 2.1575091575091574, "grad_norm": 2.640625, "learning_rate": 9.996347723205592e-05, "loss": 0.1864, "step": 2945 }, { "epoch": 2.158241758241758, "grad_norm": 0.87109375, "learning_rate": 9.996334551616718e-05, "loss": 0.1848, "step": 2946 }, { "epoch": 2.158974358974359, "grad_norm": 0.83203125, "learning_rate": 9.996321356328331e-05, "loss": 0.1337, "step": 2947 }, { "epoch": 2.1597069597069596, "grad_norm": 2.4375, "learning_rate": 9.996308137340497e-05, "loss": 0.2329, "step": 2948 }, { "epoch": 2.1604395604395603, "grad_norm": 1.96875, "learning_rate": 9.996294894653274e-05, "loss": 0.171, "step": 2949 }, { "epoch": 2.161172161172161, "grad_norm": 1.578125, "learning_rate": 9.996281628266729e-05, "loss": 0.2245, "step": 2950 }, { "epoch": 2.1619047619047618, "grad_norm": 1.6171875, "learning_rate": 9.996268338180926e-05, "loss": 0.2511, "step": 2951 }, { "epoch": 2.1626373626373625, "grad_norm": 0.859375, "learning_rate": 9.996255024395926e-05, "loss": 0.2061, "step": 2952 }, { "epoch": 2.1633699633699632, "grad_norm": 1.375, "learning_rate": 9.996241686911796e-05, "loss": 0.1837, "step": 2953 }, { "epoch": 2.164102564102564, "grad_norm": 1.9921875, "learning_rate": 9.996228325728597e-05, "loss": 0.1788, "step": 2954 }, { "epoch": 2.1648351648351647, "grad_norm": 1.2109375, "learning_rate": 9.996214940846397e-05, "loss": 0.1801, "step": 2955 }, { "epoch": 2.1655677655677654, "grad_norm": 1.265625, "learning_rate": 9.996201532265255e-05, "loss": 0.1773, "step": 2956 }, { "epoch": 2.166300366300366, "grad_norm": 1.3984375, "learning_rate": 9.996188099985238e-05, "loss": 0.2398, "step": 2957 }, { "epoch": 2.167032967032967, "grad_norm": 1.5234375, "learning_rate": 9.99617464400641e-05, "loss": 0.2162, "step": 2958 }, { "epoch": 2.1677655677655676, "grad_norm": 1.3828125, "learning_rate": 9.996161164328837e-05, "loss": 0.1762, "step": 2959 }, { "epoch": 2.1684981684981683, "grad_norm": 1.3359375, "learning_rate": 9.99614766095258e-05, "loss": 0.1774, "step": 2960 }, { "epoch": 2.169230769230769, "grad_norm": 0.8046875, "learning_rate": 9.996134133877707e-05, "loss": 0.1829, "step": 2961 }, { "epoch": 2.16996336996337, "grad_norm": 1.2578125, "learning_rate": 9.996120583104282e-05, "loss": 0.2007, "step": 2962 }, { "epoch": 2.1706959706959705, "grad_norm": 1.015625, "learning_rate": 9.99610700863237e-05, "loss": 0.1983, "step": 2963 }, { "epoch": 2.1714285714285713, "grad_norm": 1.6875, "learning_rate": 9.996093410462034e-05, "loss": 0.2752, "step": 2964 }, { "epoch": 2.172161172161172, "grad_norm": 1.0390625, "learning_rate": 9.996079788593342e-05, "loss": 0.1566, "step": 2965 }, { "epoch": 2.1728937728937727, "grad_norm": 1.5, "learning_rate": 9.996066143026357e-05, "loss": 0.17, "step": 2966 }, { "epoch": 2.1736263736263735, "grad_norm": 0.71875, "learning_rate": 9.996052473761146e-05, "loss": 0.1595, "step": 2967 }, { "epoch": 2.174358974358974, "grad_norm": 0.95703125, "learning_rate": 9.996038780797775e-05, "loss": 0.1649, "step": 2968 }, { "epoch": 2.175091575091575, "grad_norm": 1.0078125, "learning_rate": 9.996025064136306e-05, "loss": 0.1842, "step": 2969 }, { "epoch": 2.1758241758241756, "grad_norm": 0.9765625, "learning_rate": 9.996011323776809e-05, "loss": 0.1709, "step": 2970 }, { "epoch": 2.1765567765567764, "grad_norm": 1.171875, "learning_rate": 9.99599755971935e-05, "loss": 0.1563, "step": 2971 }, { "epoch": 2.177289377289377, "grad_norm": 1.4765625, "learning_rate": 9.99598377196399e-05, "loss": 0.1891, "step": 2972 }, { "epoch": 2.178021978021978, "grad_norm": 0.609375, "learning_rate": 9.995969960510797e-05, "loss": 0.1825, "step": 2973 }, { "epoch": 2.1787545787545786, "grad_norm": 1.3203125, "learning_rate": 9.995956125359842e-05, "loss": 0.1947, "step": 2974 }, { "epoch": 2.1794871794871793, "grad_norm": 1.0078125, "learning_rate": 9.995942266511185e-05, "loss": 0.1883, "step": 2975 }, { "epoch": 2.18021978021978, "grad_norm": 1.5859375, "learning_rate": 9.995928383964895e-05, "loss": 0.1926, "step": 2976 }, { "epoch": 2.1809523809523808, "grad_norm": 1.609375, "learning_rate": 9.99591447772104e-05, "loss": 0.2291, "step": 2977 }, { "epoch": 2.1816849816849815, "grad_norm": 1.0625, "learning_rate": 9.995900547779683e-05, "loss": 0.1777, "step": 2978 }, { "epoch": 2.182417582417582, "grad_norm": 0.6953125, "learning_rate": 9.995886594140896e-05, "loss": 0.1367, "step": 2979 }, { "epoch": 2.183150183150183, "grad_norm": 1.1171875, "learning_rate": 9.99587261680474e-05, "loss": 0.1899, "step": 2980 }, { "epoch": 2.183882783882784, "grad_norm": 1.0078125, "learning_rate": 9.995858615771286e-05, "loss": 0.1618, "step": 2981 }, { "epoch": 2.184615384615385, "grad_norm": 1.0234375, "learning_rate": 9.995844591040598e-05, "loss": 0.2148, "step": 2982 }, { "epoch": 2.1853479853479856, "grad_norm": 1.09375, "learning_rate": 9.995830542612747e-05, "loss": 0.1967, "step": 2983 }, { "epoch": 2.1860805860805863, "grad_norm": 1.4375, "learning_rate": 9.995816470487797e-05, "loss": 0.1553, "step": 2984 }, { "epoch": 2.186813186813187, "grad_norm": 0.9296875, "learning_rate": 9.995802374665817e-05, "loss": 0.1667, "step": 2985 }, { "epoch": 2.1875457875457878, "grad_norm": 1.25, "learning_rate": 9.995788255146875e-05, "loss": 0.148, "step": 2986 }, { "epoch": 2.1882783882783885, "grad_norm": 1.5546875, "learning_rate": 9.995774111931038e-05, "loss": 0.1673, "step": 2987 }, { "epoch": 2.1890109890109892, "grad_norm": 2.21875, "learning_rate": 9.995759945018371e-05, "loss": 0.2052, "step": 2988 }, { "epoch": 2.18974358974359, "grad_norm": 0.83203125, "learning_rate": 9.995745754408947e-05, "loss": 0.1406, "step": 2989 }, { "epoch": 2.1904761904761907, "grad_norm": 0.92578125, "learning_rate": 9.995731540102831e-05, "loss": 0.1862, "step": 2990 }, { "epoch": 2.1912087912087914, "grad_norm": 1.1328125, "learning_rate": 9.995717302100092e-05, "loss": 0.1785, "step": 2991 }, { "epoch": 2.191941391941392, "grad_norm": 1.4140625, "learning_rate": 9.995703040400797e-05, "loss": 0.191, "step": 2992 }, { "epoch": 2.192673992673993, "grad_norm": 1.0, "learning_rate": 9.995688755005015e-05, "loss": 0.1908, "step": 2993 }, { "epoch": 2.1934065934065936, "grad_norm": 1.2734375, "learning_rate": 9.995674445912815e-05, "loss": 0.1714, "step": 2994 }, { "epoch": 2.1941391941391943, "grad_norm": 2.09375, "learning_rate": 9.995660113124266e-05, "loss": 0.2352, "step": 2995 }, { "epoch": 2.194871794871795, "grad_norm": 1.0703125, "learning_rate": 9.995645756639435e-05, "loss": 0.1995, "step": 2996 }, { "epoch": 2.195604395604396, "grad_norm": 1.1796875, "learning_rate": 9.995631376458393e-05, "loss": 0.2224, "step": 2997 }, { "epoch": 2.1963369963369965, "grad_norm": 1.2890625, "learning_rate": 9.995616972581205e-05, "loss": 0.2024, "step": 2998 }, { "epoch": 2.1970695970695973, "grad_norm": 1.15625, "learning_rate": 9.995602545007944e-05, "loss": 0.1993, "step": 2999 }, { "epoch": 2.197802197802198, "grad_norm": 0.9609375, "learning_rate": 9.995588093738677e-05, "loss": 0.1686, "step": 3000 }, { "epoch": 2.1985347985347987, "grad_norm": 1.3359375, "learning_rate": 9.995573618773474e-05, "loss": 0.2088, "step": 3001 }, { "epoch": 2.1992673992673994, "grad_norm": 1.484375, "learning_rate": 9.995559120112404e-05, "loss": 0.1719, "step": 3002 }, { "epoch": 2.2, "grad_norm": 1.03125, "learning_rate": 9.995544597755538e-05, "loss": 0.185, "step": 3003 }, { "epoch": 2.200732600732601, "grad_norm": 1.8515625, "learning_rate": 9.995530051702945e-05, "loss": 0.2248, "step": 3004 }, { "epoch": 2.2014652014652016, "grad_norm": 1.484375, "learning_rate": 9.995515481954691e-05, "loss": 0.2224, "step": 3005 }, { "epoch": 2.2021978021978024, "grad_norm": 1.8515625, "learning_rate": 9.995500888510851e-05, "loss": 0.1779, "step": 3006 }, { "epoch": 2.202930402930403, "grad_norm": 1.0859375, "learning_rate": 9.995486271371493e-05, "loss": 0.1553, "step": 3007 }, { "epoch": 2.203663003663004, "grad_norm": 0.828125, "learning_rate": 9.995471630536685e-05, "loss": 0.2263, "step": 3008 }, { "epoch": 2.2043956043956046, "grad_norm": 0.93359375, "learning_rate": 9.995456966006501e-05, "loss": 0.1898, "step": 3009 }, { "epoch": 2.2051282051282053, "grad_norm": 1.2734375, "learning_rate": 9.995442277781009e-05, "loss": 0.2388, "step": 3010 }, { "epoch": 2.205860805860806, "grad_norm": 1.5078125, "learning_rate": 9.995427565860278e-05, "loss": 0.1891, "step": 3011 }, { "epoch": 2.2065934065934067, "grad_norm": 1.4609375, "learning_rate": 9.99541283024438e-05, "loss": 0.2102, "step": 3012 }, { "epoch": 2.2073260073260075, "grad_norm": 1.421875, "learning_rate": 9.995398070933387e-05, "loss": 0.234, "step": 3013 }, { "epoch": 2.208058608058608, "grad_norm": 1.4296875, "learning_rate": 9.995383287927368e-05, "loss": 0.1703, "step": 3014 }, { "epoch": 2.208791208791209, "grad_norm": 1.1796875, "learning_rate": 9.995368481226394e-05, "loss": 0.1711, "step": 3015 }, { "epoch": 2.2095238095238097, "grad_norm": 1.515625, "learning_rate": 9.995353650830536e-05, "loss": 0.2085, "step": 3016 }, { "epoch": 2.2102564102564104, "grad_norm": 1.2578125, "learning_rate": 9.995338796739864e-05, "loss": 0.208, "step": 3017 }, { "epoch": 2.210989010989011, "grad_norm": 0.98046875, "learning_rate": 9.995323918954454e-05, "loss": 0.1727, "step": 3018 }, { "epoch": 2.211721611721612, "grad_norm": 1.2265625, "learning_rate": 9.995309017474371e-05, "loss": 0.1719, "step": 3019 }, { "epoch": 2.2124542124542126, "grad_norm": 1.1484375, "learning_rate": 9.99529409229969e-05, "loss": 0.1895, "step": 3020 }, { "epoch": 2.2131868131868133, "grad_norm": 0.96484375, "learning_rate": 9.995279143430479e-05, "loss": 0.1555, "step": 3021 }, { "epoch": 2.213919413919414, "grad_norm": 1.0546875, "learning_rate": 9.995264170866814e-05, "loss": 0.1767, "step": 3022 }, { "epoch": 2.2146520146520148, "grad_norm": 1.109375, "learning_rate": 9.995249174608768e-05, "loss": 0.1828, "step": 3023 }, { "epoch": 2.2153846153846155, "grad_norm": 1.2578125, "learning_rate": 9.995234154656406e-05, "loss": 0.1987, "step": 3024 }, { "epoch": 2.2161172161172162, "grad_norm": 1.1796875, "learning_rate": 9.995219111009804e-05, "loss": 0.2205, "step": 3025 }, { "epoch": 2.216849816849817, "grad_norm": 1.171875, "learning_rate": 9.995204043669035e-05, "loss": 0.2078, "step": 3026 }, { "epoch": 2.2175824175824177, "grad_norm": 1.2734375, "learning_rate": 9.99518895263417e-05, "loss": 0.1869, "step": 3027 }, { "epoch": 2.2183150183150184, "grad_norm": 1.671875, "learning_rate": 9.995173837905281e-05, "loss": 0.1726, "step": 3028 }, { "epoch": 2.219047619047619, "grad_norm": 1.4765625, "learning_rate": 9.99515869948244e-05, "loss": 0.174, "step": 3029 }, { "epoch": 2.21978021978022, "grad_norm": 0.98046875, "learning_rate": 9.99514353736572e-05, "loss": 0.1965, "step": 3030 }, { "epoch": 2.2205128205128206, "grad_norm": 0.7890625, "learning_rate": 9.995128351555197e-05, "loss": 0.1673, "step": 3031 }, { "epoch": 2.2212454212454213, "grad_norm": 1.0390625, "learning_rate": 9.995113142050937e-05, "loss": 0.1945, "step": 3032 }, { "epoch": 2.221978021978022, "grad_norm": 1.1015625, "learning_rate": 9.995097908853018e-05, "loss": 0.1869, "step": 3033 }, { "epoch": 2.222710622710623, "grad_norm": 0.78515625, "learning_rate": 9.995082651961512e-05, "loss": 0.1761, "step": 3034 }, { "epoch": 2.2234432234432235, "grad_norm": 0.93359375, "learning_rate": 9.995067371376491e-05, "loss": 0.1468, "step": 3035 }, { "epoch": 2.2241758241758243, "grad_norm": 1.3984375, "learning_rate": 9.995052067098029e-05, "loss": 0.189, "step": 3036 }, { "epoch": 2.224908424908425, "grad_norm": 0.90625, "learning_rate": 9.995036739126202e-05, "loss": 0.1886, "step": 3037 }, { "epoch": 2.2256410256410257, "grad_norm": 0.9609375, "learning_rate": 9.995021387461076e-05, "loss": 0.1965, "step": 3038 }, { "epoch": 2.2263736263736265, "grad_norm": 1.0234375, "learning_rate": 9.995006012102733e-05, "loss": 0.194, "step": 3039 }, { "epoch": 2.227106227106227, "grad_norm": 1.484375, "learning_rate": 9.994990613051238e-05, "loss": 0.1584, "step": 3040 }, { "epoch": 2.227838827838828, "grad_norm": 1.203125, "learning_rate": 9.994975190306673e-05, "loss": 0.1825, "step": 3041 }, { "epoch": 2.2285714285714286, "grad_norm": 1.71875, "learning_rate": 9.994959743869109e-05, "loss": 0.1901, "step": 3042 }, { "epoch": 2.2293040293040294, "grad_norm": 1.7578125, "learning_rate": 9.99494427373862e-05, "loss": 0.2114, "step": 3043 }, { "epoch": 2.23003663003663, "grad_norm": 1.34375, "learning_rate": 9.994928779915277e-05, "loss": 0.1874, "step": 3044 }, { "epoch": 2.230769230769231, "grad_norm": 1.5078125, "learning_rate": 9.99491326239916e-05, "loss": 0.1851, "step": 3045 }, { "epoch": 2.2315018315018316, "grad_norm": 1.5859375, "learning_rate": 9.994897721190337e-05, "loss": 0.1672, "step": 3046 }, { "epoch": 2.2322344322344323, "grad_norm": 1.5390625, "learning_rate": 9.994882156288887e-05, "loss": 0.1849, "step": 3047 }, { "epoch": 2.232967032967033, "grad_norm": 1.2578125, "learning_rate": 9.994866567694884e-05, "loss": 0.2323, "step": 3048 }, { "epoch": 2.2336996336996338, "grad_norm": 1.359375, "learning_rate": 9.994850955408401e-05, "loss": 0.2044, "step": 3049 }, { "epoch": 2.2344322344322345, "grad_norm": 0.84765625, "learning_rate": 9.994835319429515e-05, "loss": 0.1738, "step": 3050 }, { "epoch": 2.235164835164835, "grad_norm": 1.6875, "learning_rate": 9.994819659758299e-05, "loss": 0.2522, "step": 3051 }, { "epoch": 2.235897435897436, "grad_norm": 1.28125, "learning_rate": 9.994803976394829e-05, "loss": 0.1905, "step": 3052 }, { "epoch": 2.2366300366300367, "grad_norm": 1.46875, "learning_rate": 9.994788269339179e-05, "loss": 0.1616, "step": 3053 }, { "epoch": 2.2373626373626374, "grad_norm": 1.0703125, "learning_rate": 9.994772538591427e-05, "loss": 0.1896, "step": 3054 }, { "epoch": 2.238095238095238, "grad_norm": 1.1328125, "learning_rate": 9.994756784151645e-05, "loss": 0.1911, "step": 3055 }, { "epoch": 2.238827838827839, "grad_norm": 1.1328125, "learning_rate": 9.99474100601991e-05, "loss": 0.1691, "step": 3056 }, { "epoch": 2.2395604395604396, "grad_norm": 1.578125, "learning_rate": 9.994725204196298e-05, "loss": 0.1842, "step": 3057 }, { "epoch": 2.2402930402930403, "grad_norm": 2.203125, "learning_rate": 9.994709378680884e-05, "loss": 0.2042, "step": 3058 }, { "epoch": 2.241025641025641, "grad_norm": 1.1953125, "learning_rate": 9.994693529473745e-05, "loss": 0.1285, "step": 3059 }, { "epoch": 2.241758241758242, "grad_norm": 0.95703125, "learning_rate": 9.994677656574954e-05, "loss": 0.1761, "step": 3060 }, { "epoch": 2.2424908424908425, "grad_norm": 1.25, "learning_rate": 9.994661759984592e-05, "loss": 0.2561, "step": 3061 }, { "epoch": 2.2432234432234432, "grad_norm": 1.828125, "learning_rate": 9.99464583970273e-05, "loss": 0.1838, "step": 3062 }, { "epoch": 2.243956043956044, "grad_norm": 1.171875, "learning_rate": 9.994629895729448e-05, "loss": 0.1905, "step": 3063 }, { "epoch": 2.2446886446886447, "grad_norm": 1.5625, "learning_rate": 9.99461392806482e-05, "loss": 0.1951, "step": 3064 }, { "epoch": 2.2454212454212454, "grad_norm": 1.4921875, "learning_rate": 9.994597936708924e-05, "loss": 0.2316, "step": 3065 }, { "epoch": 2.246153846153846, "grad_norm": 1.0703125, "learning_rate": 9.994581921661835e-05, "loss": 0.2232, "step": 3066 }, { "epoch": 2.246886446886447, "grad_norm": 1.625, "learning_rate": 9.994565882923632e-05, "loss": 0.1978, "step": 3067 }, { "epoch": 2.2476190476190476, "grad_norm": 2.03125, "learning_rate": 9.994549820494391e-05, "loss": 0.1769, "step": 3068 }, { "epoch": 2.2483516483516484, "grad_norm": 1.5859375, "learning_rate": 9.994533734374187e-05, "loss": 0.2008, "step": 3069 }, { "epoch": 2.249084249084249, "grad_norm": 1.1171875, "learning_rate": 9.994517624563098e-05, "loss": 0.1513, "step": 3070 }, { "epoch": 2.24981684981685, "grad_norm": 1.7109375, "learning_rate": 9.994501491061206e-05, "loss": 0.2685, "step": 3071 }, { "epoch": 2.2505494505494505, "grad_norm": 1.484375, "learning_rate": 9.994485333868581e-05, "loss": 0.1658, "step": 3072 }, { "epoch": 2.2512820512820513, "grad_norm": 1.3515625, "learning_rate": 9.994469152985305e-05, "loss": 0.1642, "step": 3073 }, { "epoch": 2.252014652014652, "grad_norm": 1.21875, "learning_rate": 9.994452948411453e-05, "loss": 0.1589, "step": 3074 }, { "epoch": 2.2527472527472527, "grad_norm": 1.03125, "learning_rate": 9.994436720147103e-05, "loss": 0.1544, "step": 3075 }, { "epoch": 2.2534798534798535, "grad_norm": 1.234375, "learning_rate": 9.994420468192336e-05, "loss": 0.1838, "step": 3076 }, { "epoch": 2.254212454212454, "grad_norm": 1.3671875, "learning_rate": 9.994404192547225e-05, "loss": 0.2093, "step": 3077 }, { "epoch": 2.254945054945055, "grad_norm": 1.296875, "learning_rate": 9.994387893211854e-05, "loss": 0.1499, "step": 3078 }, { "epoch": 2.2556776556776557, "grad_norm": 2.375, "learning_rate": 9.994371570186293e-05, "loss": 0.2042, "step": 3079 }, { "epoch": 2.2564102564102564, "grad_norm": 0.94140625, "learning_rate": 9.994355223470626e-05, "loss": 0.204, "step": 3080 }, { "epoch": 2.257142857142857, "grad_norm": 1.15625, "learning_rate": 9.994338853064932e-05, "loss": 0.1764, "step": 3081 }, { "epoch": 2.257875457875458, "grad_norm": 1.5546875, "learning_rate": 9.994322458969285e-05, "loss": 0.2128, "step": 3082 }, { "epoch": 2.2586080586080586, "grad_norm": 1.203125, "learning_rate": 9.994306041183767e-05, "loss": 0.181, "step": 3083 }, { "epoch": 2.2593406593406593, "grad_norm": 0.71484375, "learning_rate": 9.994289599708455e-05, "loss": 0.1513, "step": 3084 }, { "epoch": 2.26007326007326, "grad_norm": 1.3203125, "learning_rate": 9.99427313454343e-05, "loss": 0.194, "step": 3085 }, { "epoch": 2.2608058608058608, "grad_norm": 1.1484375, "learning_rate": 9.994256645688769e-05, "loss": 0.1789, "step": 3086 }, { "epoch": 2.2615384615384615, "grad_norm": 1.375, "learning_rate": 9.994240133144549e-05, "loss": 0.2106, "step": 3087 }, { "epoch": 2.2622710622710622, "grad_norm": 1.078125, "learning_rate": 9.994223596910854e-05, "loss": 0.2289, "step": 3088 }, { "epoch": 2.263003663003663, "grad_norm": 0.796875, "learning_rate": 9.994207036987761e-05, "loss": 0.1951, "step": 3089 }, { "epoch": 2.2637362637362637, "grad_norm": 0.90625, "learning_rate": 9.994190453375347e-05, "loss": 0.1955, "step": 3090 }, { "epoch": 2.2644688644688644, "grad_norm": 0.91015625, "learning_rate": 9.994173846073693e-05, "loss": 0.1908, "step": 3091 }, { "epoch": 2.265201465201465, "grad_norm": 0.890625, "learning_rate": 9.994157215082881e-05, "loss": 0.1784, "step": 3092 }, { "epoch": 2.265934065934066, "grad_norm": 1.7578125, "learning_rate": 9.994140560402988e-05, "loss": 0.1852, "step": 3093 }, { "epoch": 2.2666666666666666, "grad_norm": 1.5546875, "learning_rate": 9.994123882034094e-05, "loss": 0.2103, "step": 3094 }, { "epoch": 2.2673992673992673, "grad_norm": 1.1640625, "learning_rate": 9.994107179976282e-05, "loss": 0.1854, "step": 3095 }, { "epoch": 2.268131868131868, "grad_norm": 0.9296875, "learning_rate": 9.994090454229627e-05, "loss": 0.1732, "step": 3096 }, { "epoch": 2.268864468864469, "grad_norm": 1.6171875, "learning_rate": 9.994073704794211e-05, "loss": 0.1935, "step": 3097 }, { "epoch": 2.2695970695970695, "grad_norm": 1.296875, "learning_rate": 9.994056931670116e-05, "loss": 0.1906, "step": 3098 }, { "epoch": 2.2703296703296703, "grad_norm": 1.578125, "learning_rate": 9.994040134857422e-05, "loss": 0.2122, "step": 3099 }, { "epoch": 2.271062271062271, "grad_norm": 1.3125, "learning_rate": 9.994023314356207e-05, "loss": 0.1867, "step": 3100 }, { "epoch": 2.2717948717948717, "grad_norm": 1.296875, "learning_rate": 9.994006470166554e-05, "loss": 0.1917, "step": 3101 }, { "epoch": 2.2725274725274724, "grad_norm": 1.1015625, "learning_rate": 9.993989602288544e-05, "loss": 0.1874, "step": 3102 }, { "epoch": 2.273260073260073, "grad_norm": 1.1015625, "learning_rate": 9.993972710722254e-05, "loss": 0.1792, "step": 3103 }, { "epoch": 2.273992673992674, "grad_norm": 0.7265625, "learning_rate": 9.993955795467772e-05, "loss": 0.1226, "step": 3104 }, { "epoch": 2.2747252747252746, "grad_norm": 1.5, "learning_rate": 9.993938856525172e-05, "loss": 0.1826, "step": 3105 }, { "epoch": 2.2754578754578754, "grad_norm": 1.203125, "learning_rate": 9.993921893894538e-05, "loss": 0.1767, "step": 3106 }, { "epoch": 2.276190476190476, "grad_norm": 1.3046875, "learning_rate": 9.993904907575951e-05, "loss": 0.169, "step": 3107 }, { "epoch": 2.276923076923077, "grad_norm": 1.7421875, "learning_rate": 9.993887897569493e-05, "loss": 0.2445, "step": 3108 }, { "epoch": 2.2776556776556776, "grad_norm": 1.09375, "learning_rate": 9.993870863875246e-05, "loss": 0.2061, "step": 3109 }, { "epoch": 2.2783882783882783, "grad_norm": 1.3203125, "learning_rate": 9.993853806493289e-05, "loss": 0.1608, "step": 3110 }, { "epoch": 2.279120879120879, "grad_norm": 1.46875, "learning_rate": 9.993836725423708e-05, "loss": 0.2033, "step": 3111 }, { "epoch": 2.2798534798534797, "grad_norm": 1.2578125, "learning_rate": 9.99381962066658e-05, "loss": 0.1828, "step": 3112 }, { "epoch": 2.2805860805860805, "grad_norm": 1.3125, "learning_rate": 9.99380249222199e-05, "loss": 0.1938, "step": 3113 }, { "epoch": 2.281318681318681, "grad_norm": 1.21875, "learning_rate": 9.993785340090019e-05, "loss": 0.192, "step": 3114 }, { "epoch": 2.282051282051282, "grad_norm": 0.953125, "learning_rate": 9.99376816427075e-05, "loss": 0.2091, "step": 3115 }, { "epoch": 2.2827838827838827, "grad_norm": 1.1171875, "learning_rate": 9.993750964764265e-05, "loss": 0.1899, "step": 3116 }, { "epoch": 2.2835164835164834, "grad_norm": 1.8046875, "learning_rate": 9.993733741570646e-05, "loss": 0.1724, "step": 3117 }, { "epoch": 2.284249084249084, "grad_norm": 0.890625, "learning_rate": 9.993716494689976e-05, "loss": 0.1807, "step": 3118 }, { "epoch": 2.284981684981685, "grad_norm": 0.63671875, "learning_rate": 9.993699224122336e-05, "loss": 0.146, "step": 3119 }, { "epoch": 2.2857142857142856, "grad_norm": 0.83203125, "learning_rate": 9.993681929867812e-05, "loss": 0.1563, "step": 3120 }, { "epoch": 2.2864468864468863, "grad_norm": 0.9296875, "learning_rate": 9.993664611926483e-05, "loss": 0.1832, "step": 3121 }, { "epoch": 2.287179487179487, "grad_norm": 1.21875, "learning_rate": 9.993647270298434e-05, "loss": 0.1537, "step": 3122 }, { "epoch": 2.2879120879120878, "grad_norm": 1.8125, "learning_rate": 9.99362990498375e-05, "loss": 0.2079, "step": 3123 }, { "epoch": 2.2886446886446885, "grad_norm": 1.2109375, "learning_rate": 9.993612515982512e-05, "loss": 0.1924, "step": 3124 }, { "epoch": 2.2893772893772892, "grad_norm": 0.9296875, "learning_rate": 9.993595103294803e-05, "loss": 0.1555, "step": 3125 }, { "epoch": 2.29010989010989, "grad_norm": 0.73046875, "learning_rate": 9.993577666920705e-05, "loss": 0.1501, "step": 3126 }, { "epoch": 2.2908424908424907, "grad_norm": 1.1640625, "learning_rate": 9.993560206860307e-05, "loss": 0.2397, "step": 3127 }, { "epoch": 2.2915750915750914, "grad_norm": 1.0, "learning_rate": 9.993542723113686e-05, "loss": 0.2002, "step": 3128 }, { "epoch": 2.292307692307692, "grad_norm": 1.234375, "learning_rate": 9.99352521568093e-05, "loss": 0.2035, "step": 3129 }, { "epoch": 2.293040293040293, "grad_norm": 1.625, "learning_rate": 9.993507684562124e-05, "loss": 0.1734, "step": 3130 }, { "epoch": 2.2937728937728936, "grad_norm": 1.203125, "learning_rate": 9.993490129757348e-05, "loss": 0.166, "step": 3131 }, { "epoch": 2.2945054945054943, "grad_norm": 1.7421875, "learning_rate": 9.993472551266688e-05, "loss": 0.1775, "step": 3132 }, { "epoch": 2.295238095238095, "grad_norm": 1.1171875, "learning_rate": 9.993454949090228e-05, "loss": 0.1393, "step": 3133 }, { "epoch": 2.295970695970696, "grad_norm": 0.9140625, "learning_rate": 9.993437323228053e-05, "loss": 0.2272, "step": 3134 }, { "epoch": 2.2967032967032965, "grad_norm": 0.9453125, "learning_rate": 9.993419673680246e-05, "loss": 0.205, "step": 3135 }, { "epoch": 2.2974358974358973, "grad_norm": 1.0078125, "learning_rate": 9.993402000446894e-05, "loss": 0.1872, "step": 3136 }, { "epoch": 2.298168498168498, "grad_norm": 1.0625, "learning_rate": 9.993384303528078e-05, "loss": 0.2377, "step": 3137 }, { "epoch": 2.2989010989010987, "grad_norm": 0.921875, "learning_rate": 9.993366582923888e-05, "loss": 0.1988, "step": 3138 }, { "epoch": 2.2996336996336995, "grad_norm": 0.80078125, "learning_rate": 9.993348838634403e-05, "loss": 0.1882, "step": 3139 }, { "epoch": 2.3003663003663, "grad_norm": 1.2109375, "learning_rate": 9.993331070659712e-05, "loss": 0.1678, "step": 3140 }, { "epoch": 2.301098901098901, "grad_norm": 1.4453125, "learning_rate": 9.9933132789999e-05, "loss": 0.2005, "step": 3141 }, { "epoch": 2.3018315018315016, "grad_norm": 0.87109375, "learning_rate": 9.99329546365505e-05, "loss": 0.1613, "step": 3142 }, { "epoch": 2.3025641025641024, "grad_norm": 0.8125, "learning_rate": 9.993277624625249e-05, "loss": 0.1745, "step": 3143 }, { "epoch": 2.303296703296703, "grad_norm": 1.046875, "learning_rate": 9.993259761910582e-05, "loss": 0.19, "step": 3144 }, { "epoch": 2.304029304029304, "grad_norm": 1.265625, "learning_rate": 9.993241875511136e-05, "loss": 0.1741, "step": 3145 }, { "epoch": 2.3047619047619046, "grad_norm": 1.1875, "learning_rate": 9.993223965426996e-05, "loss": 0.2227, "step": 3146 }, { "epoch": 2.3054945054945053, "grad_norm": 0.9296875, "learning_rate": 9.993206031658245e-05, "loss": 0.1543, "step": 3147 }, { "epoch": 2.306227106227106, "grad_norm": 1.609375, "learning_rate": 9.99318807420497e-05, "loss": 0.1489, "step": 3148 }, { "epoch": 2.3069597069597068, "grad_norm": 1.421875, "learning_rate": 9.993170093067261e-05, "loss": 0.2117, "step": 3149 }, { "epoch": 2.3076923076923075, "grad_norm": 1.2109375, "learning_rate": 9.9931520882452e-05, "loss": 0.1746, "step": 3150 }, { "epoch": 2.308424908424908, "grad_norm": 1.1328125, "learning_rate": 9.993134059738876e-05, "loss": 0.1566, "step": 3151 }, { "epoch": 2.309157509157509, "grad_norm": 1.25, "learning_rate": 9.993116007548373e-05, "loss": 0.2135, "step": 3152 }, { "epoch": 2.3098901098901097, "grad_norm": 1.5234375, "learning_rate": 9.993097931673778e-05, "loss": 0.1696, "step": 3153 }, { "epoch": 2.3106227106227104, "grad_norm": 1.109375, "learning_rate": 9.993079832115178e-05, "loss": 0.1897, "step": 3154 }, { "epoch": 2.311355311355311, "grad_norm": 1.578125, "learning_rate": 9.993061708872661e-05, "loss": 0.2245, "step": 3155 }, { "epoch": 2.312087912087912, "grad_norm": 1.2734375, "learning_rate": 9.993043561946312e-05, "loss": 0.1751, "step": 3156 }, { "epoch": 2.3128205128205126, "grad_norm": 0.7734375, "learning_rate": 9.993025391336218e-05, "loss": 0.1697, "step": 3157 }, { "epoch": 2.3135531135531133, "grad_norm": 0.96875, "learning_rate": 9.993007197042468e-05, "loss": 0.1663, "step": 3158 }, { "epoch": 2.314285714285714, "grad_norm": 1.203125, "learning_rate": 9.992988979065148e-05, "loss": 0.1637, "step": 3159 }, { "epoch": 2.315018315018315, "grad_norm": 1.078125, "learning_rate": 9.992970737404344e-05, "loss": 0.2288, "step": 3160 }, { "epoch": 2.3157509157509155, "grad_norm": 0.97265625, "learning_rate": 9.992952472060144e-05, "loss": 0.192, "step": 3161 }, { "epoch": 2.3164835164835162, "grad_norm": 0.87109375, "learning_rate": 9.992934183032636e-05, "loss": 0.1488, "step": 3162 }, { "epoch": 2.317216117216117, "grad_norm": 0.7421875, "learning_rate": 9.992915870321909e-05, "loss": 0.1515, "step": 3163 }, { "epoch": 2.3179487179487177, "grad_norm": 1.28125, "learning_rate": 9.992897533928049e-05, "loss": 0.1716, "step": 3164 }, { "epoch": 2.3186813186813184, "grad_norm": 1.171875, "learning_rate": 9.992879173851145e-05, "loss": 0.2471, "step": 3165 }, { "epoch": 2.319413919413919, "grad_norm": 1.1484375, "learning_rate": 9.992860790091283e-05, "loss": 0.1803, "step": 3166 }, { "epoch": 2.32014652014652, "grad_norm": 0.921875, "learning_rate": 9.992842382648553e-05, "loss": 0.1514, "step": 3167 }, { "epoch": 2.320879120879121, "grad_norm": 1.0, "learning_rate": 9.992823951523041e-05, "loss": 0.1921, "step": 3168 }, { "epoch": 2.321611721611722, "grad_norm": 1.140625, "learning_rate": 9.992805496714838e-05, "loss": 0.1741, "step": 3169 }, { "epoch": 2.3223443223443225, "grad_norm": 0.79296875, "learning_rate": 9.992787018224032e-05, "loss": 0.1784, "step": 3170 }, { "epoch": 2.3230769230769233, "grad_norm": 1.6875, "learning_rate": 9.99276851605071e-05, "loss": 0.1829, "step": 3171 }, { "epoch": 2.323809523809524, "grad_norm": 0.8203125, "learning_rate": 9.992749990194962e-05, "loss": 0.1845, "step": 3172 }, { "epoch": 2.3245421245421247, "grad_norm": 0.93359375, "learning_rate": 9.992731440656875e-05, "loss": 0.1828, "step": 3173 }, { "epoch": 2.3252747252747255, "grad_norm": 1.1171875, "learning_rate": 9.992712867436541e-05, "loss": 0.173, "step": 3174 }, { "epoch": 2.326007326007326, "grad_norm": 1.3984375, "learning_rate": 9.992694270534047e-05, "loss": 0.2187, "step": 3175 }, { "epoch": 2.326739926739927, "grad_norm": 0.83203125, "learning_rate": 9.992675649949481e-05, "loss": 0.182, "step": 3176 }, { "epoch": 2.3274725274725276, "grad_norm": 1.234375, "learning_rate": 9.992657005682933e-05, "loss": 0.2071, "step": 3177 }, { "epoch": 2.3282051282051284, "grad_norm": 1.0625, "learning_rate": 9.992638337734492e-05, "loss": 0.2097, "step": 3178 }, { "epoch": 2.328937728937729, "grad_norm": 1.0546875, "learning_rate": 9.992619646104248e-05, "loss": 0.1501, "step": 3179 }, { "epoch": 2.32967032967033, "grad_norm": 1.3515625, "learning_rate": 9.992600930792294e-05, "loss": 0.2094, "step": 3180 }, { "epoch": 2.3304029304029306, "grad_norm": 0.6953125, "learning_rate": 9.992582191798714e-05, "loss": 0.1779, "step": 3181 }, { "epoch": 2.3311355311355313, "grad_norm": 1.3984375, "learning_rate": 9.992563429123602e-05, "loss": 0.1755, "step": 3182 }, { "epoch": 2.331868131868132, "grad_norm": 1.8984375, "learning_rate": 9.992544642767044e-05, "loss": 0.1943, "step": 3183 }, { "epoch": 2.3326007326007328, "grad_norm": 1.0703125, "learning_rate": 9.992525832729132e-05, "loss": 0.2144, "step": 3184 }, { "epoch": 2.3333333333333335, "grad_norm": 1.453125, "learning_rate": 9.992506999009958e-05, "loss": 0.1609, "step": 3185 }, { "epoch": 2.334065934065934, "grad_norm": 1.1640625, "learning_rate": 9.992488141609609e-05, "loss": 0.1599, "step": 3186 }, { "epoch": 2.334798534798535, "grad_norm": 1.578125, "learning_rate": 9.992469260528178e-05, "loss": 0.1815, "step": 3187 }, { "epoch": 2.3355311355311357, "grad_norm": 1.1484375, "learning_rate": 9.992450355765753e-05, "loss": 0.1737, "step": 3188 }, { "epoch": 2.3362637362637364, "grad_norm": 0.9296875, "learning_rate": 9.992431427322425e-05, "loss": 0.1886, "step": 3189 }, { "epoch": 2.336996336996337, "grad_norm": 1.5, "learning_rate": 9.992412475198287e-05, "loss": 0.2271, "step": 3190 }, { "epoch": 2.337728937728938, "grad_norm": 1.1640625, "learning_rate": 9.99239349939343e-05, "loss": 0.2193, "step": 3191 }, { "epoch": 2.3384615384615386, "grad_norm": 2.03125, "learning_rate": 9.992374499907942e-05, "loss": 0.2003, "step": 3192 }, { "epoch": 2.3391941391941393, "grad_norm": 1.15625, "learning_rate": 9.992355476741914e-05, "loss": 0.1944, "step": 3193 }, { "epoch": 2.33992673992674, "grad_norm": 1.375, "learning_rate": 9.992336429895439e-05, "loss": 0.1641, "step": 3194 }, { "epoch": 2.340659340659341, "grad_norm": 1.296875, "learning_rate": 9.992317359368607e-05, "loss": 0.1931, "step": 3195 }, { "epoch": 2.3413919413919415, "grad_norm": 1.0546875, "learning_rate": 9.992298265161512e-05, "loss": 0.185, "step": 3196 }, { "epoch": 2.3421245421245422, "grad_norm": 1.2421875, "learning_rate": 9.992279147274242e-05, "loss": 0.1905, "step": 3197 }, { "epoch": 2.342857142857143, "grad_norm": 0.859375, "learning_rate": 9.99226000570689e-05, "loss": 0.1988, "step": 3198 }, { "epoch": 2.3435897435897437, "grad_norm": 1.5546875, "learning_rate": 9.992240840459548e-05, "loss": 0.1979, "step": 3199 }, { "epoch": 2.3443223443223444, "grad_norm": 1.703125, "learning_rate": 9.992221651532311e-05, "loss": 0.2192, "step": 3200 }, { "epoch": 2.345054945054945, "grad_norm": 1.3828125, "learning_rate": 9.992202438925263e-05, "loss": 0.1664, "step": 3201 }, { "epoch": 2.345787545787546, "grad_norm": 1.2578125, "learning_rate": 9.992183202638503e-05, "loss": 0.204, "step": 3202 }, { "epoch": 2.3465201465201466, "grad_norm": 1.25, "learning_rate": 9.992163942672121e-05, "loss": 0.1598, "step": 3203 }, { "epoch": 2.3472527472527474, "grad_norm": 1.4140625, "learning_rate": 9.992144659026208e-05, "loss": 0.1861, "step": 3204 }, { "epoch": 2.347985347985348, "grad_norm": 1.046875, "learning_rate": 9.992125351700858e-05, "loss": 0.1799, "step": 3205 }, { "epoch": 2.348717948717949, "grad_norm": 1.2890625, "learning_rate": 9.992106020696164e-05, "loss": 0.1647, "step": 3206 }, { "epoch": 2.3494505494505495, "grad_norm": 0.75390625, "learning_rate": 9.992086666012217e-05, "loss": 0.1608, "step": 3207 }, { "epoch": 2.3501831501831503, "grad_norm": 0.70703125, "learning_rate": 9.992067287649109e-05, "loss": 0.1847, "step": 3208 }, { "epoch": 2.350915750915751, "grad_norm": 1.03125, "learning_rate": 9.992047885606937e-05, "loss": 0.1792, "step": 3209 }, { "epoch": 2.3516483516483517, "grad_norm": 0.80859375, "learning_rate": 9.99202845988579e-05, "loss": 0.1649, "step": 3210 }, { "epoch": 2.3523809523809525, "grad_norm": 0.9921875, "learning_rate": 9.992009010485761e-05, "loss": 0.2102, "step": 3211 }, { "epoch": 2.353113553113553, "grad_norm": 0.9765625, "learning_rate": 9.991989537406946e-05, "loss": 0.1538, "step": 3212 }, { "epoch": 2.353846153846154, "grad_norm": 0.9765625, "learning_rate": 9.991970040649436e-05, "loss": 0.1586, "step": 3213 }, { "epoch": 2.3545787545787547, "grad_norm": 1.15625, "learning_rate": 9.991950520213325e-05, "loss": 0.2253, "step": 3214 }, { "epoch": 2.3553113553113554, "grad_norm": 0.96484375, "learning_rate": 9.991930976098707e-05, "loss": 0.1916, "step": 3215 }, { "epoch": 2.356043956043956, "grad_norm": 0.7734375, "learning_rate": 9.991911408305678e-05, "loss": 0.1858, "step": 3216 }, { "epoch": 2.356776556776557, "grad_norm": 0.8125, "learning_rate": 9.991891816834325e-05, "loss": 0.1544, "step": 3217 }, { "epoch": 2.3575091575091576, "grad_norm": 0.9921875, "learning_rate": 9.991872201684747e-05, "loss": 0.1941, "step": 3218 }, { "epoch": 2.3582417582417583, "grad_norm": 0.83984375, "learning_rate": 9.991852562857039e-05, "loss": 0.1644, "step": 3219 }, { "epoch": 2.358974358974359, "grad_norm": 1.2109375, "learning_rate": 9.991832900351291e-05, "loss": 0.2001, "step": 3220 }, { "epoch": 2.3597069597069598, "grad_norm": 1.03125, "learning_rate": 9.991813214167601e-05, "loss": 0.1611, "step": 3221 }, { "epoch": 2.3604395604395605, "grad_norm": 1.1953125, "learning_rate": 9.991793504306061e-05, "loss": 0.173, "step": 3222 }, { "epoch": 2.361172161172161, "grad_norm": 1.3359375, "learning_rate": 9.991773770766764e-05, "loss": 0.2109, "step": 3223 }, { "epoch": 2.361904761904762, "grad_norm": 1.234375, "learning_rate": 9.99175401354981e-05, "loss": 0.1989, "step": 3224 }, { "epoch": 2.3626373626373627, "grad_norm": 1.0, "learning_rate": 9.991734232655289e-05, "loss": 0.1881, "step": 3225 }, { "epoch": 2.3633699633699634, "grad_norm": 1.328125, "learning_rate": 9.991714428083295e-05, "loss": 0.1843, "step": 3226 }, { "epoch": 2.364102564102564, "grad_norm": 1.0234375, "learning_rate": 9.991694599833927e-05, "loss": 0.1921, "step": 3227 }, { "epoch": 2.364835164835165, "grad_norm": 0.7421875, "learning_rate": 9.991674747907278e-05, "loss": 0.1679, "step": 3228 }, { "epoch": 2.3655677655677656, "grad_norm": 0.7421875, "learning_rate": 9.991654872303441e-05, "loss": 0.1708, "step": 3229 }, { "epoch": 2.3663003663003663, "grad_norm": 0.6796875, "learning_rate": 9.991634973022515e-05, "loss": 0.1729, "step": 3230 }, { "epoch": 2.367032967032967, "grad_norm": 1.015625, "learning_rate": 9.991615050064595e-05, "loss": 0.1866, "step": 3231 }, { "epoch": 2.367765567765568, "grad_norm": 0.89453125, "learning_rate": 9.991595103429772e-05, "loss": 0.1779, "step": 3232 }, { "epoch": 2.3684981684981685, "grad_norm": 0.76171875, "learning_rate": 9.991575133118147e-05, "loss": 0.1816, "step": 3233 }, { "epoch": 2.3692307692307693, "grad_norm": 0.90234375, "learning_rate": 9.991555139129811e-05, "loss": 0.1596, "step": 3234 }, { "epoch": 2.36996336996337, "grad_norm": 0.96484375, "learning_rate": 9.991535121464863e-05, "loss": 0.1919, "step": 3235 }, { "epoch": 2.3706959706959707, "grad_norm": 0.8671875, "learning_rate": 9.991515080123399e-05, "loss": 0.1533, "step": 3236 }, { "epoch": 2.3714285714285714, "grad_norm": 0.87109375, "learning_rate": 9.991495015105513e-05, "loss": 0.1635, "step": 3237 }, { "epoch": 2.372161172161172, "grad_norm": 1.1015625, "learning_rate": 9.991474926411304e-05, "loss": 0.1403, "step": 3238 }, { "epoch": 2.372893772893773, "grad_norm": 0.98828125, "learning_rate": 9.991454814040865e-05, "loss": 0.1724, "step": 3239 }, { "epoch": 2.3736263736263736, "grad_norm": 1.0625, "learning_rate": 9.991434677994293e-05, "loss": 0.1734, "step": 3240 }, { "epoch": 2.3743589743589744, "grad_norm": 0.828125, "learning_rate": 9.991414518271686e-05, "loss": 0.1364, "step": 3241 }, { "epoch": 2.375091575091575, "grad_norm": 0.83984375, "learning_rate": 9.991394334873141e-05, "loss": 0.1768, "step": 3242 }, { "epoch": 2.375824175824176, "grad_norm": 0.703125, "learning_rate": 9.991374127798751e-05, "loss": 0.1721, "step": 3243 }, { "epoch": 2.3765567765567766, "grad_norm": 0.6640625, "learning_rate": 9.991353897048619e-05, "loss": 0.1588, "step": 3244 }, { "epoch": 2.3772893772893773, "grad_norm": 1.5234375, "learning_rate": 9.991333642622836e-05, "loss": 0.2002, "step": 3245 }, { "epoch": 2.378021978021978, "grad_norm": 0.94140625, "learning_rate": 9.991313364521502e-05, "loss": 0.1781, "step": 3246 }, { "epoch": 2.3787545787545787, "grad_norm": 0.82421875, "learning_rate": 9.991293062744714e-05, "loss": 0.1597, "step": 3247 }, { "epoch": 2.3794871794871795, "grad_norm": 0.91796875, "learning_rate": 9.991272737292567e-05, "loss": 0.2148, "step": 3248 }, { "epoch": 2.38021978021978, "grad_norm": 1.0546875, "learning_rate": 9.991252388165162e-05, "loss": 0.178, "step": 3249 }, { "epoch": 2.380952380952381, "grad_norm": 1.3125, "learning_rate": 9.991232015362594e-05, "loss": 0.1642, "step": 3250 }, { "epoch": 2.3816849816849817, "grad_norm": 1.359375, "learning_rate": 9.991211618884961e-05, "loss": 0.1861, "step": 3251 }, { "epoch": 2.3824175824175824, "grad_norm": 1.140625, "learning_rate": 9.991191198732362e-05, "loss": 0.2212, "step": 3252 }, { "epoch": 2.383150183150183, "grad_norm": 0.984375, "learning_rate": 9.991170754904893e-05, "loss": 0.1902, "step": 3253 }, { "epoch": 2.383882783882784, "grad_norm": 0.625, "learning_rate": 9.991150287402653e-05, "loss": 0.1561, "step": 3254 }, { "epoch": 2.3846153846153846, "grad_norm": 1.640625, "learning_rate": 9.99112979622574e-05, "loss": 0.2345, "step": 3255 }, { "epoch": 2.3853479853479853, "grad_norm": 1.140625, "learning_rate": 9.991109281374252e-05, "loss": 0.2195, "step": 3256 }, { "epoch": 2.386080586080586, "grad_norm": 1.28125, "learning_rate": 9.991088742848288e-05, "loss": 0.2114, "step": 3257 }, { "epoch": 2.3868131868131868, "grad_norm": 1.0859375, "learning_rate": 9.991068180647945e-05, "loss": 0.1478, "step": 3258 }, { "epoch": 2.3875457875457875, "grad_norm": 0.80859375, "learning_rate": 9.991047594773323e-05, "loss": 0.1661, "step": 3259 }, { "epoch": 2.3882783882783882, "grad_norm": 1.1953125, "learning_rate": 9.99102698522452e-05, "loss": 0.1718, "step": 3260 }, { "epoch": 2.389010989010989, "grad_norm": 1.6484375, "learning_rate": 9.991006352001635e-05, "loss": 0.1649, "step": 3261 }, { "epoch": 2.3897435897435897, "grad_norm": 1.3046875, "learning_rate": 9.990985695104765e-05, "loss": 0.1571, "step": 3262 }, { "epoch": 2.3904761904761904, "grad_norm": 1.328125, "learning_rate": 9.990965014534012e-05, "loss": 0.145, "step": 3263 }, { "epoch": 2.391208791208791, "grad_norm": 1.1171875, "learning_rate": 9.990944310289472e-05, "loss": 0.1657, "step": 3264 }, { "epoch": 2.391941391941392, "grad_norm": 0.9375, "learning_rate": 9.990923582371248e-05, "loss": 0.1703, "step": 3265 }, { "epoch": 2.3926739926739926, "grad_norm": 1.1015625, "learning_rate": 9.990902830779435e-05, "loss": 0.1657, "step": 3266 }, { "epoch": 2.3934065934065933, "grad_norm": 1.96875, "learning_rate": 9.990882055514136e-05, "loss": 0.1606, "step": 3267 }, { "epoch": 2.394139194139194, "grad_norm": 0.9921875, "learning_rate": 9.990861256575446e-05, "loss": 0.2022, "step": 3268 }, { "epoch": 2.394871794871795, "grad_norm": 0.91796875, "learning_rate": 9.990840433963472e-05, "loss": 0.169, "step": 3269 }, { "epoch": 2.3956043956043955, "grad_norm": 1.265625, "learning_rate": 9.990819587678308e-05, "loss": 0.1692, "step": 3270 }, { "epoch": 2.3963369963369963, "grad_norm": 1.3125, "learning_rate": 9.990798717720054e-05, "loss": 0.1659, "step": 3271 }, { "epoch": 2.397069597069597, "grad_norm": 1.7734375, "learning_rate": 9.990777824088813e-05, "loss": 0.1983, "step": 3272 }, { "epoch": 2.3978021978021977, "grad_norm": 0.921875, "learning_rate": 9.990756906784683e-05, "loss": 0.147, "step": 3273 }, { "epoch": 2.3985347985347985, "grad_norm": 1.078125, "learning_rate": 9.990735965807764e-05, "loss": 0.1873, "step": 3274 }, { "epoch": 2.399267399267399, "grad_norm": 0.96484375, "learning_rate": 9.990715001158158e-05, "loss": 0.1489, "step": 3275 }, { "epoch": 2.4, "grad_norm": 0.88671875, "learning_rate": 9.990694012835964e-05, "loss": 0.1779, "step": 3276 }, { "epoch": 2.4007326007326006, "grad_norm": 0.921875, "learning_rate": 9.990673000841284e-05, "loss": 0.1677, "step": 3277 }, { "epoch": 2.4014652014652014, "grad_norm": 0.9609375, "learning_rate": 9.990651965174217e-05, "loss": 0.1683, "step": 3278 }, { "epoch": 2.402197802197802, "grad_norm": 0.96484375, "learning_rate": 9.990630905834864e-05, "loss": 0.1577, "step": 3279 }, { "epoch": 2.402930402930403, "grad_norm": 1.3984375, "learning_rate": 9.990609822823327e-05, "loss": 0.1536, "step": 3280 }, { "epoch": 2.4036630036630036, "grad_norm": 1.453125, "learning_rate": 9.990588716139707e-05, "loss": 0.2468, "step": 3281 }, { "epoch": 2.4043956043956043, "grad_norm": 1.0234375, "learning_rate": 9.990567585784103e-05, "loss": 0.1913, "step": 3282 }, { "epoch": 2.405128205128205, "grad_norm": 1.1484375, "learning_rate": 9.99054643175662e-05, "loss": 0.1947, "step": 3283 }, { "epoch": 2.4058608058608058, "grad_norm": 1.34375, "learning_rate": 9.990525254057356e-05, "loss": 0.2057, "step": 3284 }, { "epoch": 2.4065934065934065, "grad_norm": 0.84375, "learning_rate": 9.990504052686412e-05, "loss": 0.1928, "step": 3285 }, { "epoch": 2.407326007326007, "grad_norm": 0.984375, "learning_rate": 9.990482827643893e-05, "loss": 0.1606, "step": 3286 }, { "epoch": 2.408058608058608, "grad_norm": 0.94140625, "learning_rate": 9.9904615789299e-05, "loss": 0.1704, "step": 3287 }, { "epoch": 2.4087912087912087, "grad_norm": 1.140625, "learning_rate": 9.990440306544533e-05, "loss": 0.1651, "step": 3288 }, { "epoch": 2.4095238095238094, "grad_norm": 1.359375, "learning_rate": 9.990419010487894e-05, "loss": 0.1437, "step": 3289 }, { "epoch": 2.41025641025641, "grad_norm": 0.640625, "learning_rate": 9.990397690760086e-05, "loss": 0.1665, "step": 3290 }, { "epoch": 2.410989010989011, "grad_norm": 1.5859375, "learning_rate": 9.99037634736121e-05, "loss": 0.1822, "step": 3291 }, { "epoch": 2.4117216117216116, "grad_norm": 1.0, "learning_rate": 9.99035498029137e-05, "loss": 0.1458, "step": 3292 }, { "epoch": 2.4124542124542123, "grad_norm": 1.484375, "learning_rate": 9.990333589550666e-05, "loss": 0.189, "step": 3293 }, { "epoch": 2.413186813186813, "grad_norm": 1.359375, "learning_rate": 9.990312175139203e-05, "loss": 0.178, "step": 3294 }, { "epoch": 2.413919413919414, "grad_norm": 1.265625, "learning_rate": 9.990290737057084e-05, "loss": 0.1503, "step": 3295 }, { "epoch": 2.4146520146520145, "grad_norm": 0.93359375, "learning_rate": 9.990269275304409e-05, "loss": 0.1455, "step": 3296 }, { "epoch": 2.4153846153846152, "grad_norm": 1.1171875, "learning_rate": 9.990247789881281e-05, "loss": 0.1632, "step": 3297 }, { "epoch": 2.416117216117216, "grad_norm": 1.0, "learning_rate": 9.990226280787806e-05, "loss": 0.1621, "step": 3298 }, { "epoch": 2.4168498168498167, "grad_norm": 0.9609375, "learning_rate": 9.990204748024084e-05, "loss": 0.1492, "step": 3299 }, { "epoch": 2.4175824175824174, "grad_norm": 1.265625, "learning_rate": 9.99018319159022e-05, "loss": 0.2104, "step": 3300 }, { "epoch": 2.418315018315018, "grad_norm": 1.578125, "learning_rate": 9.990161611486317e-05, "loss": 0.1789, "step": 3301 }, { "epoch": 2.419047619047619, "grad_norm": 1.796875, "learning_rate": 9.990140007712477e-05, "loss": 0.1922, "step": 3302 }, { "epoch": 2.4197802197802196, "grad_norm": 0.8359375, "learning_rate": 9.990118380268805e-05, "loss": 0.1333, "step": 3303 }, { "epoch": 2.4205128205128204, "grad_norm": 1.7890625, "learning_rate": 9.990096729155403e-05, "loss": 0.1925, "step": 3304 }, { "epoch": 2.421245421245421, "grad_norm": 1.5, "learning_rate": 9.990075054372376e-05, "loss": 0.1631, "step": 3305 }, { "epoch": 2.421978021978022, "grad_norm": 0.9453125, "learning_rate": 9.990053355919828e-05, "loss": 0.1513, "step": 3306 }, { "epoch": 2.4227106227106225, "grad_norm": 1.0078125, "learning_rate": 9.990031633797864e-05, "loss": 0.1959, "step": 3307 }, { "epoch": 2.4234432234432233, "grad_norm": 0.98046875, "learning_rate": 9.990009888006587e-05, "loss": 0.1775, "step": 3308 }, { "epoch": 2.424175824175824, "grad_norm": 0.796875, "learning_rate": 9.989988118546098e-05, "loss": 0.1794, "step": 3309 }, { "epoch": 2.4249084249084247, "grad_norm": 1.0390625, "learning_rate": 9.989966325416506e-05, "loss": 0.1545, "step": 3310 }, { "epoch": 2.4256410256410255, "grad_norm": 0.734375, "learning_rate": 9.989944508617915e-05, "loss": 0.1571, "step": 3311 }, { "epoch": 2.426373626373626, "grad_norm": 1.0859375, "learning_rate": 9.989922668150427e-05, "loss": 0.1789, "step": 3312 }, { "epoch": 2.427106227106227, "grad_norm": 1.0078125, "learning_rate": 9.989900804014146e-05, "loss": 0.1419, "step": 3313 }, { "epoch": 2.4278388278388277, "grad_norm": 1.5390625, "learning_rate": 9.98987891620918e-05, "loss": 0.1753, "step": 3314 }, { "epoch": 2.4285714285714284, "grad_norm": 1.0625, "learning_rate": 9.989857004735634e-05, "loss": 0.191, "step": 3315 }, { "epoch": 2.429304029304029, "grad_norm": 3.375, "learning_rate": 9.989835069593609e-05, "loss": 0.2266, "step": 3316 }, { "epoch": 2.4300366300366303, "grad_norm": 1.5625, "learning_rate": 9.989813110783212e-05, "loss": 0.1712, "step": 3317 }, { "epoch": 2.430769230769231, "grad_norm": 0.77734375, "learning_rate": 9.989791128304552e-05, "loss": 0.1677, "step": 3318 }, { "epoch": 2.4315018315018317, "grad_norm": 1.34375, "learning_rate": 9.98976912215773e-05, "loss": 0.1944, "step": 3319 }, { "epoch": 2.4322344322344325, "grad_norm": 0.78125, "learning_rate": 9.989747092342852e-05, "loss": 0.1388, "step": 3320 }, { "epoch": 2.432967032967033, "grad_norm": 0.8046875, "learning_rate": 9.989725038860023e-05, "loss": 0.1248, "step": 3321 }, { "epoch": 2.433699633699634, "grad_norm": 0.98046875, "learning_rate": 9.989702961709351e-05, "loss": 0.1484, "step": 3322 }, { "epoch": 2.4344322344322347, "grad_norm": 0.91015625, "learning_rate": 9.98968086089094e-05, "loss": 0.1644, "step": 3323 }, { "epoch": 2.4351648351648354, "grad_norm": 1.75, "learning_rate": 9.989658736404895e-05, "loss": 0.1601, "step": 3324 }, { "epoch": 2.435897435897436, "grad_norm": 1.1328125, "learning_rate": 9.989636588251326e-05, "loss": 0.201, "step": 3325 }, { "epoch": 2.436630036630037, "grad_norm": 1.1484375, "learning_rate": 9.989614416430336e-05, "loss": 0.1925, "step": 3326 }, { "epoch": 2.4373626373626376, "grad_norm": 1.0234375, "learning_rate": 9.989592220942031e-05, "loss": 0.1983, "step": 3327 }, { "epoch": 2.4380952380952383, "grad_norm": 1.4453125, "learning_rate": 9.989570001786518e-05, "loss": 0.1599, "step": 3328 }, { "epoch": 2.438827838827839, "grad_norm": 1.03125, "learning_rate": 9.989547758963905e-05, "loss": 0.203, "step": 3329 }, { "epoch": 2.4395604395604398, "grad_norm": 1.40625, "learning_rate": 9.989525492474297e-05, "loss": 0.2004, "step": 3330 }, { "epoch": 2.4402930402930405, "grad_norm": 1.078125, "learning_rate": 9.989503202317798e-05, "loss": 0.192, "step": 3331 }, { "epoch": 2.4410256410256412, "grad_norm": 0.6796875, "learning_rate": 9.98948088849452e-05, "loss": 0.155, "step": 3332 }, { "epoch": 2.441758241758242, "grad_norm": 1.546875, "learning_rate": 9.989458551004567e-05, "loss": 0.2005, "step": 3333 }, { "epoch": 2.4424908424908427, "grad_norm": 0.89453125, "learning_rate": 9.989436189848046e-05, "loss": 0.1741, "step": 3334 }, { "epoch": 2.4432234432234434, "grad_norm": 0.88671875, "learning_rate": 9.989413805025068e-05, "loss": 0.1493, "step": 3335 }, { "epoch": 2.443956043956044, "grad_norm": 0.94140625, "learning_rate": 9.989391396535734e-05, "loss": 0.1862, "step": 3336 }, { "epoch": 2.444688644688645, "grad_norm": 0.71484375, "learning_rate": 9.989368964380154e-05, "loss": 0.1771, "step": 3337 }, { "epoch": 2.4454212454212456, "grad_norm": 1.09375, "learning_rate": 9.989346508558436e-05, "loss": 0.1771, "step": 3338 }, { "epoch": 2.4461538461538463, "grad_norm": 1.03125, "learning_rate": 9.989324029070687e-05, "loss": 0.145, "step": 3339 }, { "epoch": 2.446886446886447, "grad_norm": 1.2890625, "learning_rate": 9.989301525917016e-05, "loss": 0.1685, "step": 3340 }, { "epoch": 2.447619047619048, "grad_norm": 0.84765625, "learning_rate": 9.989278999097532e-05, "loss": 0.1916, "step": 3341 }, { "epoch": 2.4483516483516485, "grad_norm": 0.9921875, "learning_rate": 9.989256448612337e-05, "loss": 0.2323, "step": 3342 }, { "epoch": 2.4490842490842493, "grad_norm": 1.2890625, "learning_rate": 9.989233874461546e-05, "loss": 0.1688, "step": 3343 }, { "epoch": 2.44981684981685, "grad_norm": 0.78125, "learning_rate": 9.989211276645263e-05, "loss": 0.169, "step": 3344 }, { "epoch": 2.4505494505494507, "grad_norm": 0.796875, "learning_rate": 9.989188655163596e-05, "loss": 0.1984, "step": 3345 }, { "epoch": 2.4512820512820515, "grad_norm": 0.80078125, "learning_rate": 9.989166010016657e-05, "loss": 0.1829, "step": 3346 }, { "epoch": 2.452014652014652, "grad_norm": 1.4296875, "learning_rate": 9.98914334120455e-05, "loss": 0.1881, "step": 3347 }, { "epoch": 2.452747252747253, "grad_norm": 1.1328125, "learning_rate": 9.989120648727386e-05, "loss": 0.1834, "step": 3348 }, { "epoch": 2.4534798534798536, "grad_norm": 0.90234375, "learning_rate": 9.989097932585275e-05, "loss": 0.2068, "step": 3349 }, { "epoch": 2.4542124542124544, "grad_norm": 1.140625, "learning_rate": 9.989075192778323e-05, "loss": 0.2083, "step": 3350 }, { "epoch": 2.454945054945055, "grad_norm": 1.0625, "learning_rate": 9.989052429306642e-05, "loss": 0.1617, "step": 3351 }, { "epoch": 2.455677655677656, "grad_norm": 1.2890625, "learning_rate": 9.989029642170338e-05, "loss": 0.1413, "step": 3352 }, { "epoch": 2.4564102564102566, "grad_norm": 1.265625, "learning_rate": 9.989006831369524e-05, "loss": 0.1591, "step": 3353 }, { "epoch": 2.4571428571428573, "grad_norm": 1.109375, "learning_rate": 9.988983996904304e-05, "loss": 0.1837, "step": 3354 }, { "epoch": 2.457875457875458, "grad_norm": 1.4765625, "learning_rate": 9.988961138774792e-05, "loss": 0.1926, "step": 3355 }, { "epoch": 2.4586080586080588, "grad_norm": 2.171875, "learning_rate": 9.988938256981095e-05, "loss": 0.1665, "step": 3356 }, { "epoch": 2.4593406593406595, "grad_norm": 1.1953125, "learning_rate": 9.988915351523323e-05, "loss": 0.1704, "step": 3357 }, { "epoch": 2.46007326007326, "grad_norm": 1.5703125, "learning_rate": 9.988892422401587e-05, "loss": 0.1484, "step": 3358 }, { "epoch": 2.460805860805861, "grad_norm": 0.7734375, "learning_rate": 9.988869469615996e-05, "loss": 0.1713, "step": 3359 }, { "epoch": 2.4615384615384617, "grad_norm": 1.4609375, "learning_rate": 9.98884649316666e-05, "loss": 0.1664, "step": 3360 }, { "epoch": 2.4622710622710624, "grad_norm": 1.34375, "learning_rate": 9.98882349305369e-05, "loss": 0.1839, "step": 3361 }, { "epoch": 2.463003663003663, "grad_norm": 1.7734375, "learning_rate": 9.988800469277194e-05, "loss": 0.2024, "step": 3362 }, { "epoch": 2.463736263736264, "grad_norm": 0.85546875, "learning_rate": 9.988777421837284e-05, "loss": 0.2485, "step": 3363 }, { "epoch": 2.4644688644688646, "grad_norm": 1.0859375, "learning_rate": 9.98875435073407e-05, "loss": 0.1357, "step": 3364 }, { "epoch": 2.4652014652014653, "grad_norm": 1.1640625, "learning_rate": 9.988731255967663e-05, "loss": 0.1794, "step": 3365 }, { "epoch": 2.465934065934066, "grad_norm": 1.140625, "learning_rate": 9.988708137538174e-05, "loss": 0.2261, "step": 3366 }, { "epoch": 2.466666666666667, "grad_norm": 1.109375, "learning_rate": 9.988684995445712e-05, "loss": 0.1869, "step": 3367 }, { "epoch": 2.4673992673992675, "grad_norm": 1.015625, "learning_rate": 9.988661829690388e-05, "loss": 0.1505, "step": 3368 }, { "epoch": 2.4681318681318682, "grad_norm": 0.94140625, "learning_rate": 9.988638640272316e-05, "loss": 0.183, "step": 3369 }, { "epoch": 2.468864468864469, "grad_norm": 1.4296875, "learning_rate": 9.988615427191603e-05, "loss": 0.1829, "step": 3370 }, { "epoch": 2.4695970695970697, "grad_norm": 1.25, "learning_rate": 9.988592190448363e-05, "loss": 0.2122, "step": 3371 }, { "epoch": 2.4703296703296704, "grad_norm": 1.09375, "learning_rate": 9.988568930042708e-05, "loss": 0.1688, "step": 3372 }, { "epoch": 2.471062271062271, "grad_norm": 1.0390625, "learning_rate": 9.988545645974746e-05, "loss": 0.1378, "step": 3373 }, { "epoch": 2.471794871794872, "grad_norm": 1.015625, "learning_rate": 9.988522338244592e-05, "loss": 0.1625, "step": 3374 }, { "epoch": 2.4725274725274726, "grad_norm": 1.0625, "learning_rate": 9.988499006852355e-05, "loss": 0.167, "step": 3375 }, { "epoch": 2.4732600732600734, "grad_norm": 1.0859375, "learning_rate": 9.988475651798149e-05, "loss": 0.1386, "step": 3376 }, { "epoch": 2.473992673992674, "grad_norm": 1.328125, "learning_rate": 9.988452273082085e-05, "loss": 0.1853, "step": 3377 }, { "epoch": 2.474725274725275, "grad_norm": 2.015625, "learning_rate": 9.988428870704276e-05, "loss": 0.1835, "step": 3378 }, { "epoch": 2.4754578754578755, "grad_norm": 2.1875, "learning_rate": 9.98840544466483e-05, "loss": 0.26, "step": 3379 }, { "epoch": 2.4761904761904763, "grad_norm": 1.4609375, "learning_rate": 9.988381994963864e-05, "loss": 0.1683, "step": 3380 }, { "epoch": 2.476923076923077, "grad_norm": 1.265625, "learning_rate": 9.98835852160149e-05, "loss": 0.2273, "step": 3381 }, { "epoch": 2.4776556776556777, "grad_norm": 1.1875, "learning_rate": 9.988335024577817e-05, "loss": 0.2094, "step": 3382 }, { "epoch": 2.4783882783882785, "grad_norm": 1.6171875, "learning_rate": 9.988311503892961e-05, "loss": 0.195, "step": 3383 }, { "epoch": 2.479120879120879, "grad_norm": 1.6171875, "learning_rate": 9.988287959547033e-05, "loss": 0.1785, "step": 3384 }, { "epoch": 2.47985347985348, "grad_norm": 1.7421875, "learning_rate": 9.988264391540147e-05, "loss": 0.1626, "step": 3385 }, { "epoch": 2.4805860805860807, "grad_norm": 1.8359375, "learning_rate": 9.988240799872415e-05, "loss": 0.1892, "step": 3386 }, { "epoch": 2.4813186813186814, "grad_norm": 1.6328125, "learning_rate": 9.988217184543949e-05, "loss": 0.1843, "step": 3387 }, { "epoch": 2.482051282051282, "grad_norm": 1.21875, "learning_rate": 9.988193545554865e-05, "loss": 0.1552, "step": 3388 }, { "epoch": 2.482783882783883, "grad_norm": 1.0625, "learning_rate": 9.988169882905275e-05, "loss": 0.2014, "step": 3389 }, { "epoch": 2.4835164835164836, "grad_norm": 1.390625, "learning_rate": 9.988146196595289e-05, "loss": 0.2204, "step": 3390 }, { "epoch": 2.4842490842490843, "grad_norm": 0.90234375, "learning_rate": 9.988122486625026e-05, "loss": 0.1636, "step": 3391 }, { "epoch": 2.484981684981685, "grad_norm": 0.9140625, "learning_rate": 9.988098752994596e-05, "loss": 0.1738, "step": 3392 }, { "epoch": 2.4857142857142858, "grad_norm": 0.5859375, "learning_rate": 9.988074995704116e-05, "loss": 0.1417, "step": 3393 }, { "epoch": 2.4864468864468865, "grad_norm": 0.8359375, "learning_rate": 9.988051214753694e-05, "loss": 0.1751, "step": 3394 }, { "epoch": 2.4871794871794872, "grad_norm": 1.5859375, "learning_rate": 9.98802741014345e-05, "loss": 0.19, "step": 3395 }, { "epoch": 2.487912087912088, "grad_norm": 1.796875, "learning_rate": 9.988003581873494e-05, "loss": 0.2033, "step": 3396 }, { "epoch": 2.4886446886446887, "grad_norm": 0.828125, "learning_rate": 9.987979729943944e-05, "loss": 0.1865, "step": 3397 }, { "epoch": 2.4893772893772894, "grad_norm": 1.1328125, "learning_rate": 9.98795585435491e-05, "loss": 0.2456, "step": 3398 }, { "epoch": 2.49010989010989, "grad_norm": 1.3984375, "learning_rate": 9.98793195510651e-05, "loss": 0.2005, "step": 3399 }, { "epoch": 2.490842490842491, "grad_norm": 1.484375, "learning_rate": 9.987908032198856e-05, "loss": 0.1925, "step": 3400 }, { "epoch": 2.4915750915750916, "grad_norm": 1.6171875, "learning_rate": 9.987884085632064e-05, "loss": 0.1773, "step": 3401 }, { "epoch": 2.4923076923076923, "grad_norm": 0.84765625, "learning_rate": 9.987860115406249e-05, "loss": 0.1443, "step": 3402 }, { "epoch": 2.493040293040293, "grad_norm": 0.83203125, "learning_rate": 9.987836121521524e-05, "loss": 0.1701, "step": 3403 }, { "epoch": 2.493772893772894, "grad_norm": 1.4921875, "learning_rate": 9.987812103978005e-05, "loss": 0.182, "step": 3404 }, { "epoch": 2.4945054945054945, "grad_norm": 1.2109375, "learning_rate": 9.987788062775809e-05, "loss": 0.2181, "step": 3405 }, { "epoch": 2.4952380952380953, "grad_norm": 1.046875, "learning_rate": 9.987763997915048e-05, "loss": 0.1513, "step": 3406 }, { "epoch": 2.495970695970696, "grad_norm": 1.359375, "learning_rate": 9.987739909395841e-05, "loss": 0.1821, "step": 3407 }, { "epoch": 2.4967032967032967, "grad_norm": 0.59375, "learning_rate": 9.987715797218298e-05, "loss": 0.1274, "step": 3408 }, { "epoch": 2.4974358974358974, "grad_norm": 0.734375, "learning_rate": 9.987691661382539e-05, "loss": 0.1531, "step": 3409 }, { "epoch": 2.498168498168498, "grad_norm": 1.921875, "learning_rate": 9.987667501888676e-05, "loss": 0.2035, "step": 3410 }, { "epoch": 2.498901098901099, "grad_norm": 1.765625, "learning_rate": 9.987643318736832e-05, "loss": 0.1844, "step": 3411 }, { "epoch": 2.4996336996336996, "grad_norm": 1.40625, "learning_rate": 9.987619111927114e-05, "loss": 0.1957, "step": 3412 }, { "epoch": 2.5003663003663004, "grad_norm": 0.734375, "learning_rate": 9.987594881459643e-05, "loss": 0.1894, "step": 3413 }, { "epoch": 2.501098901098901, "grad_norm": 1.2578125, "learning_rate": 9.987570627334535e-05, "loss": 0.1795, "step": 3414 }, { "epoch": 2.501831501831502, "grad_norm": 2.03125, "learning_rate": 9.987546349551904e-05, "loss": 0.2048, "step": 3415 }, { "epoch": 2.5025641025641026, "grad_norm": 1.109375, "learning_rate": 9.987522048111867e-05, "loss": 0.1457, "step": 3416 }, { "epoch": 2.5032967032967033, "grad_norm": 0.828125, "learning_rate": 9.987497723014541e-05, "loss": 0.193, "step": 3417 }, { "epoch": 2.504029304029304, "grad_norm": 0.70703125, "learning_rate": 9.987473374260045e-05, "loss": 0.1655, "step": 3418 }, { "epoch": 2.5047619047619047, "grad_norm": 1.078125, "learning_rate": 9.987449001848492e-05, "loss": 0.1573, "step": 3419 }, { "epoch": 2.5054945054945055, "grad_norm": 0.61328125, "learning_rate": 9.987424605779998e-05, "loss": 0.1956, "step": 3420 }, { "epoch": 2.506227106227106, "grad_norm": 1.09375, "learning_rate": 9.987400186054685e-05, "loss": 0.1766, "step": 3421 }, { "epoch": 2.506959706959707, "grad_norm": 1.0234375, "learning_rate": 9.987375742672664e-05, "loss": 0.1837, "step": 3422 }, { "epoch": 2.5076923076923077, "grad_norm": 0.875, "learning_rate": 9.987351275634059e-05, "loss": 0.2034, "step": 3423 }, { "epoch": 2.5084249084249084, "grad_norm": 0.91015625, "learning_rate": 9.98732678493898e-05, "loss": 0.2271, "step": 3424 }, { "epoch": 2.509157509157509, "grad_norm": 0.96484375, "learning_rate": 9.987302270587547e-05, "loss": 0.1677, "step": 3425 }, { "epoch": 2.50989010989011, "grad_norm": 1.2890625, "learning_rate": 9.987277732579879e-05, "loss": 0.1417, "step": 3426 }, { "epoch": 2.5106227106227106, "grad_norm": 1.4609375, "learning_rate": 9.987253170916092e-05, "loss": 0.1953, "step": 3427 }, { "epoch": 2.5113553113553113, "grad_norm": 1.125, "learning_rate": 9.987228585596305e-05, "loss": 0.1581, "step": 3428 }, { "epoch": 2.512087912087912, "grad_norm": 0.91015625, "learning_rate": 9.987203976620635e-05, "loss": 0.1654, "step": 3429 }, { "epoch": 2.5128205128205128, "grad_norm": 1.1796875, "learning_rate": 9.9871793439892e-05, "loss": 0.1569, "step": 3430 }, { "epoch": 2.5135531135531135, "grad_norm": 0.97265625, "learning_rate": 9.987154687702119e-05, "loss": 0.1649, "step": 3431 }, { "epoch": 2.5142857142857142, "grad_norm": 1.2265625, "learning_rate": 9.987130007759506e-05, "loss": 0.1734, "step": 3432 }, { "epoch": 2.515018315018315, "grad_norm": 0.7734375, "learning_rate": 9.987105304161486e-05, "loss": 0.1388, "step": 3433 }, { "epoch": 2.5157509157509157, "grad_norm": 0.6796875, "learning_rate": 9.987080576908172e-05, "loss": 0.1499, "step": 3434 }, { "epoch": 2.5164835164835164, "grad_norm": 1.109375, "learning_rate": 9.987055825999686e-05, "loss": 0.1614, "step": 3435 }, { "epoch": 2.517216117216117, "grad_norm": 1.171875, "learning_rate": 9.987031051436142e-05, "loss": 0.1822, "step": 3436 }, { "epoch": 2.517948717948718, "grad_norm": 0.828125, "learning_rate": 9.987006253217664e-05, "loss": 0.1836, "step": 3437 }, { "epoch": 2.5186813186813186, "grad_norm": 1.453125, "learning_rate": 9.986981431344365e-05, "loss": 0.1527, "step": 3438 }, { "epoch": 2.5194139194139193, "grad_norm": 1.125, "learning_rate": 9.98695658581637e-05, "loss": 0.2019, "step": 3439 }, { "epoch": 2.52014652014652, "grad_norm": 1.1796875, "learning_rate": 9.986931716633795e-05, "loss": 0.2432, "step": 3440 }, { "epoch": 2.520879120879121, "grad_norm": 1.1171875, "learning_rate": 9.986906823796758e-05, "loss": 0.1853, "step": 3441 }, { "epoch": 2.5216117216117215, "grad_norm": 0.90625, "learning_rate": 9.986881907305381e-05, "loss": 0.1758, "step": 3442 }, { "epoch": 2.5223443223443223, "grad_norm": 0.703125, "learning_rate": 9.986856967159782e-05, "loss": 0.171, "step": 3443 }, { "epoch": 2.523076923076923, "grad_norm": 0.78515625, "learning_rate": 9.986832003360082e-05, "loss": 0.1751, "step": 3444 }, { "epoch": 2.5238095238095237, "grad_norm": 0.71875, "learning_rate": 9.986807015906397e-05, "loss": 0.127, "step": 3445 }, { "epoch": 2.5245421245421245, "grad_norm": 0.76953125, "learning_rate": 9.98678200479885e-05, "loss": 0.1298, "step": 3446 }, { "epoch": 2.525274725274725, "grad_norm": 0.69921875, "learning_rate": 9.986756970037559e-05, "loss": 0.2123, "step": 3447 }, { "epoch": 2.526007326007326, "grad_norm": 1.1640625, "learning_rate": 9.986731911622646e-05, "loss": 0.1763, "step": 3448 }, { "epoch": 2.5267399267399266, "grad_norm": 1.1953125, "learning_rate": 9.986706829554227e-05, "loss": 0.1976, "step": 3449 }, { "epoch": 2.5274725274725274, "grad_norm": 0.91015625, "learning_rate": 9.986681723832428e-05, "loss": 0.1666, "step": 3450 }, { "epoch": 2.528205128205128, "grad_norm": 0.93359375, "learning_rate": 9.986656594457366e-05, "loss": 0.2055, "step": 3451 }, { "epoch": 2.528937728937729, "grad_norm": 1.0625, "learning_rate": 9.986631441429161e-05, "loss": 0.2212, "step": 3452 }, { "epoch": 2.5296703296703296, "grad_norm": 0.640625, "learning_rate": 9.986606264747932e-05, "loss": 0.1717, "step": 3453 }, { "epoch": 2.5304029304029303, "grad_norm": 1.1171875, "learning_rate": 9.986581064413805e-05, "loss": 0.1326, "step": 3454 }, { "epoch": 2.531135531135531, "grad_norm": 1.3671875, "learning_rate": 9.986555840426897e-05, "loss": 0.2177, "step": 3455 }, { "epoch": 2.5318681318681318, "grad_norm": 1.3046875, "learning_rate": 9.986530592787328e-05, "loss": 0.2124, "step": 3456 }, { "epoch": 2.5326007326007325, "grad_norm": 0.94140625, "learning_rate": 9.986505321495221e-05, "loss": 0.1895, "step": 3457 }, { "epoch": 2.533333333333333, "grad_norm": 1.1328125, "learning_rate": 9.986480026550697e-05, "loss": 0.1942, "step": 3458 }, { "epoch": 2.534065934065934, "grad_norm": 1.015625, "learning_rate": 9.986454707953875e-05, "loss": 0.1758, "step": 3459 }, { "epoch": 2.5347985347985347, "grad_norm": 1.6171875, "learning_rate": 9.98642936570488e-05, "loss": 0.2072, "step": 3460 }, { "epoch": 2.5355311355311354, "grad_norm": 1.3203125, "learning_rate": 9.98640399980383e-05, "loss": 0.1483, "step": 3461 }, { "epoch": 2.536263736263736, "grad_norm": 0.890625, "learning_rate": 9.986378610250849e-05, "loss": 0.1821, "step": 3462 }, { "epoch": 2.536996336996337, "grad_norm": 1.421875, "learning_rate": 9.986353197046058e-05, "loss": 0.1759, "step": 3463 }, { "epoch": 2.5377289377289376, "grad_norm": 1.328125, "learning_rate": 9.986327760189576e-05, "loss": 0.2176, "step": 3464 }, { "epoch": 2.5384615384615383, "grad_norm": 0.71875, "learning_rate": 9.98630229968153e-05, "loss": 0.1474, "step": 3465 }, { "epoch": 2.539194139194139, "grad_norm": 1.203125, "learning_rate": 9.986276815522038e-05, "loss": 0.164, "step": 3466 }, { "epoch": 2.53992673992674, "grad_norm": 1.171875, "learning_rate": 9.986251307711222e-05, "loss": 0.1825, "step": 3467 }, { "epoch": 2.5406593406593405, "grad_norm": 0.93359375, "learning_rate": 9.986225776249207e-05, "loss": 0.1336, "step": 3468 }, { "epoch": 2.5413919413919412, "grad_norm": 1.1640625, "learning_rate": 9.986200221136113e-05, "loss": 0.1752, "step": 3469 }, { "epoch": 2.542124542124542, "grad_norm": 1.4765625, "learning_rate": 9.986174642372064e-05, "loss": 0.1461, "step": 3470 }, { "epoch": 2.5428571428571427, "grad_norm": 0.8515625, "learning_rate": 9.986149039957182e-05, "loss": 0.1541, "step": 3471 }, { "epoch": 2.5435897435897434, "grad_norm": 0.98046875, "learning_rate": 9.98612341389159e-05, "loss": 0.1796, "step": 3472 }, { "epoch": 2.544322344322344, "grad_norm": 0.91015625, "learning_rate": 9.986097764175409e-05, "loss": 0.1903, "step": 3473 }, { "epoch": 2.545054945054945, "grad_norm": 1.3671875, "learning_rate": 9.986072090808763e-05, "loss": 0.1409, "step": 3474 }, { "epoch": 2.5457875457875456, "grad_norm": 1.546875, "learning_rate": 9.986046393791777e-05, "loss": 0.1956, "step": 3475 }, { "epoch": 2.5465201465201464, "grad_norm": 1.1484375, "learning_rate": 9.986020673124571e-05, "loss": 0.1979, "step": 3476 }, { "epoch": 2.547252747252747, "grad_norm": 1.203125, "learning_rate": 9.985994928807269e-05, "loss": 0.1357, "step": 3477 }, { "epoch": 2.547985347985348, "grad_norm": 1.53125, "learning_rate": 9.985969160839995e-05, "loss": 0.1912, "step": 3478 }, { "epoch": 2.5487179487179485, "grad_norm": 1.8671875, "learning_rate": 9.985943369222874e-05, "loss": 0.2159, "step": 3479 }, { "epoch": 2.5494505494505493, "grad_norm": 0.9921875, "learning_rate": 9.985917553956028e-05, "loss": 0.1505, "step": 3480 }, { "epoch": 2.55018315018315, "grad_norm": 0.62109375, "learning_rate": 9.98589171503958e-05, "loss": 0.1424, "step": 3481 }, { "epoch": 2.5509157509157507, "grad_norm": 2.109375, "learning_rate": 9.985865852473652e-05, "loss": 0.2045, "step": 3482 }, { "epoch": 2.5516483516483515, "grad_norm": 1.875, "learning_rate": 9.985839966258374e-05, "loss": 0.1683, "step": 3483 }, { "epoch": 2.552380952380952, "grad_norm": 1.078125, "learning_rate": 9.985814056393864e-05, "loss": 0.2079, "step": 3484 }, { "epoch": 2.553113553113553, "grad_norm": 0.5859375, "learning_rate": 9.98578812288025e-05, "loss": 0.1692, "step": 3485 }, { "epoch": 2.5538461538461537, "grad_norm": 1.015625, "learning_rate": 9.985762165717653e-05, "loss": 0.1503, "step": 3486 }, { "epoch": 2.5545787545787544, "grad_norm": 0.9765625, "learning_rate": 9.985736184906202e-05, "loss": 0.1527, "step": 3487 }, { "epoch": 2.555311355311355, "grad_norm": 0.8828125, "learning_rate": 9.985710180446017e-05, "loss": 0.171, "step": 3488 }, { "epoch": 2.556043956043956, "grad_norm": 0.796875, "learning_rate": 9.985684152337224e-05, "loss": 0.1555, "step": 3489 }, { "epoch": 2.5567765567765566, "grad_norm": 0.9765625, "learning_rate": 9.985658100579948e-05, "loss": 0.1932, "step": 3490 }, { "epoch": 2.5575091575091573, "grad_norm": 0.75390625, "learning_rate": 9.985632025174313e-05, "loss": 0.1986, "step": 3491 }, { "epoch": 2.558241758241758, "grad_norm": 0.8203125, "learning_rate": 9.985605926120446e-05, "loss": 0.208, "step": 3492 }, { "epoch": 2.5589743589743588, "grad_norm": 1.25, "learning_rate": 9.98557980341847e-05, "loss": 0.1825, "step": 3493 }, { "epoch": 2.5597069597069595, "grad_norm": 1.21875, "learning_rate": 9.98555365706851e-05, "loss": 0.1779, "step": 3494 }, { "epoch": 2.5604395604395602, "grad_norm": 1.34375, "learning_rate": 9.985527487070695e-05, "loss": 0.1938, "step": 3495 }, { "epoch": 2.561172161172161, "grad_norm": 1.109375, "learning_rate": 9.985501293425146e-05, "loss": 0.2033, "step": 3496 }, { "epoch": 2.5619047619047617, "grad_norm": 0.8203125, "learning_rate": 9.98547507613199e-05, "loss": 0.1304, "step": 3497 }, { "epoch": 2.5626373626373624, "grad_norm": 0.72265625, "learning_rate": 9.985448835191352e-05, "loss": 0.1439, "step": 3498 }, { "epoch": 2.563369963369963, "grad_norm": 1.1015625, "learning_rate": 9.985422570603361e-05, "loss": 0.1907, "step": 3499 }, { "epoch": 2.564102564102564, "grad_norm": 1.25, "learning_rate": 9.985396282368139e-05, "loss": 0.1605, "step": 3500 }, { "epoch": 2.5648351648351646, "grad_norm": 0.83984375, "learning_rate": 9.985369970485812e-05, "loss": 0.172, "step": 3501 }, { "epoch": 2.5655677655677653, "grad_norm": 1.28125, "learning_rate": 9.985343634956508e-05, "loss": 0.1677, "step": 3502 }, { "epoch": 2.566300366300366, "grad_norm": 1.796875, "learning_rate": 9.985317275780353e-05, "loss": 0.1968, "step": 3503 }, { "epoch": 2.567032967032967, "grad_norm": 1.40625, "learning_rate": 9.985290892957472e-05, "loss": 0.2304, "step": 3504 }, { "epoch": 2.5677655677655675, "grad_norm": 0.83984375, "learning_rate": 9.985264486487993e-05, "loss": 0.1664, "step": 3505 }, { "epoch": 2.5684981684981683, "grad_norm": 1.0703125, "learning_rate": 9.985238056372041e-05, "loss": 0.1788, "step": 3506 }, { "epoch": 2.569230769230769, "grad_norm": 1.2734375, "learning_rate": 9.985211602609744e-05, "loss": 0.1907, "step": 3507 }, { "epoch": 2.5699633699633697, "grad_norm": 1.078125, "learning_rate": 9.985185125201227e-05, "loss": 0.1838, "step": 3508 }, { "epoch": 2.5706959706959704, "grad_norm": 0.890625, "learning_rate": 9.98515862414662e-05, "loss": 0.1479, "step": 3509 }, { "epoch": 2.571428571428571, "grad_norm": 0.8125, "learning_rate": 9.985132099446045e-05, "loss": 0.1797, "step": 3510 }, { "epoch": 2.572161172161172, "grad_norm": 0.85546875, "learning_rate": 9.985105551099634e-05, "loss": 0.1659, "step": 3511 }, { "epoch": 2.5728937728937726, "grad_norm": 1.3671875, "learning_rate": 9.985078979107511e-05, "loss": 0.1468, "step": 3512 }, { "epoch": 2.5736263736263734, "grad_norm": 0.875, "learning_rate": 9.985052383469806e-05, "loss": 0.1571, "step": 3513 }, { "epoch": 2.574358974358974, "grad_norm": 0.94140625, "learning_rate": 9.985025764186645e-05, "loss": 0.1733, "step": 3514 }, { "epoch": 2.575091575091575, "grad_norm": 1.296875, "learning_rate": 9.984999121258152e-05, "loss": 0.1967, "step": 3515 }, { "epoch": 2.5758241758241756, "grad_norm": 1.0, "learning_rate": 9.984972454684463e-05, "loss": 0.1679, "step": 3516 }, { "epoch": 2.5765567765567763, "grad_norm": 0.69921875, "learning_rate": 9.984945764465699e-05, "loss": 0.1652, "step": 3517 }, { "epoch": 2.577289377289377, "grad_norm": 1.6953125, "learning_rate": 9.98491905060199e-05, "loss": 0.1985, "step": 3518 }, { "epoch": 2.5780219780219777, "grad_norm": 1.7109375, "learning_rate": 9.984892313093462e-05, "loss": 0.171, "step": 3519 }, { "epoch": 2.578754578754579, "grad_norm": 2.1875, "learning_rate": 9.984865551940247e-05, "loss": 0.1806, "step": 3520 }, { "epoch": 2.5794871794871796, "grad_norm": 1.2421875, "learning_rate": 9.98483876714247e-05, "loss": 0.1624, "step": 3521 }, { "epoch": 2.5802197802197804, "grad_norm": 0.87109375, "learning_rate": 9.984811958700262e-05, "loss": 0.1354, "step": 3522 }, { "epoch": 2.580952380952381, "grad_norm": 1.6484375, "learning_rate": 9.984785126613748e-05, "loss": 0.153, "step": 3523 }, { "epoch": 2.581684981684982, "grad_norm": 1.3828125, "learning_rate": 9.98475827088306e-05, "loss": 0.2277, "step": 3524 }, { "epoch": 2.5824175824175826, "grad_norm": 0.96875, "learning_rate": 9.984731391508324e-05, "loss": 0.1985, "step": 3525 }, { "epoch": 2.5831501831501833, "grad_norm": 0.70703125, "learning_rate": 9.98470448848967e-05, "loss": 0.1335, "step": 3526 }, { "epoch": 2.583882783882784, "grad_norm": 1.0078125, "learning_rate": 9.984677561827228e-05, "loss": 0.1876, "step": 3527 }, { "epoch": 2.5846153846153848, "grad_norm": 1.2578125, "learning_rate": 9.984650611521124e-05, "loss": 0.1946, "step": 3528 }, { "epoch": 2.5853479853479855, "grad_norm": 0.97265625, "learning_rate": 9.98462363757149e-05, "loss": 0.1761, "step": 3529 }, { "epoch": 2.586080586080586, "grad_norm": 1.03125, "learning_rate": 9.984596639978455e-05, "loss": 0.1734, "step": 3530 }, { "epoch": 2.586813186813187, "grad_norm": 0.6640625, "learning_rate": 9.984569618742148e-05, "loss": 0.1756, "step": 3531 }, { "epoch": 2.5875457875457877, "grad_norm": 0.8046875, "learning_rate": 9.984542573862696e-05, "loss": 0.1989, "step": 3532 }, { "epoch": 2.5882783882783884, "grad_norm": 1.078125, "learning_rate": 9.984515505340232e-05, "loss": 0.1949, "step": 3533 }, { "epoch": 2.589010989010989, "grad_norm": 1.140625, "learning_rate": 9.984488413174885e-05, "loss": 0.2484, "step": 3534 }, { "epoch": 2.58974358974359, "grad_norm": 1.359375, "learning_rate": 9.984461297366781e-05, "loss": 0.1306, "step": 3535 }, { "epoch": 2.5904761904761906, "grad_norm": 1.234375, "learning_rate": 9.984434157916055e-05, "loss": 0.1636, "step": 3536 }, { "epoch": 2.5912087912087913, "grad_norm": 1.8125, "learning_rate": 9.984406994822836e-05, "loss": 0.1832, "step": 3537 }, { "epoch": 2.591941391941392, "grad_norm": 1.359375, "learning_rate": 9.984379808087252e-05, "loss": 0.1752, "step": 3538 }, { "epoch": 2.592673992673993, "grad_norm": 1.6484375, "learning_rate": 9.984352597709436e-05, "loss": 0.2514, "step": 3539 }, { "epoch": 2.5934065934065935, "grad_norm": 1.1015625, "learning_rate": 9.984325363689514e-05, "loss": 0.2041, "step": 3540 }, { "epoch": 2.5941391941391942, "grad_norm": 1.40625, "learning_rate": 9.984298106027621e-05, "loss": 0.1846, "step": 3541 }, { "epoch": 2.594871794871795, "grad_norm": 0.93359375, "learning_rate": 9.984270824723886e-05, "loss": 0.1885, "step": 3542 }, { "epoch": 2.5956043956043957, "grad_norm": 1.0625, "learning_rate": 9.984243519778441e-05, "loss": 0.2288, "step": 3543 }, { "epoch": 2.5963369963369964, "grad_norm": 1.015625, "learning_rate": 9.984216191191413e-05, "loss": 0.1442, "step": 3544 }, { "epoch": 2.597069597069597, "grad_norm": 0.92578125, "learning_rate": 9.984188838962937e-05, "loss": 0.1542, "step": 3545 }, { "epoch": 2.597802197802198, "grad_norm": 1.390625, "learning_rate": 9.984161463093142e-05, "loss": 0.1861, "step": 3546 }, { "epoch": 2.5985347985347986, "grad_norm": 1.4609375, "learning_rate": 9.98413406358216e-05, "loss": 0.2029, "step": 3547 }, { "epoch": 2.5992673992673994, "grad_norm": 1.109375, "learning_rate": 9.984106640430122e-05, "loss": 0.1713, "step": 3548 }, { "epoch": 2.6, "grad_norm": 0.9375, "learning_rate": 9.984079193637159e-05, "loss": 0.1838, "step": 3549 }, { "epoch": 2.600732600732601, "grad_norm": 0.6640625, "learning_rate": 9.984051723203402e-05, "loss": 0.1555, "step": 3550 }, { "epoch": 2.6014652014652015, "grad_norm": 0.82421875, "learning_rate": 9.984024229128985e-05, "loss": 0.1551, "step": 3551 }, { "epoch": 2.6021978021978023, "grad_norm": 1.1015625, "learning_rate": 9.983996711414038e-05, "loss": 0.1802, "step": 3552 }, { "epoch": 2.602930402930403, "grad_norm": 1.0703125, "learning_rate": 9.983969170058692e-05, "loss": 0.1853, "step": 3553 }, { "epoch": 2.6036630036630037, "grad_norm": 1.578125, "learning_rate": 9.983941605063081e-05, "loss": 0.2166, "step": 3554 }, { "epoch": 2.6043956043956045, "grad_norm": 1.1484375, "learning_rate": 9.983914016427335e-05, "loss": 0.1643, "step": 3555 }, { "epoch": 2.605128205128205, "grad_norm": 1.078125, "learning_rate": 9.983886404151587e-05, "loss": 0.1829, "step": 3556 }, { "epoch": 2.605860805860806, "grad_norm": 0.65234375, "learning_rate": 9.983858768235971e-05, "loss": 0.1853, "step": 3557 }, { "epoch": 2.6065934065934067, "grad_norm": 0.8984375, "learning_rate": 9.983831108680617e-05, "loss": 0.177, "step": 3558 }, { "epoch": 2.6073260073260074, "grad_norm": 0.87109375, "learning_rate": 9.983803425485659e-05, "loss": 0.1733, "step": 3559 }, { "epoch": 2.608058608058608, "grad_norm": 1.0078125, "learning_rate": 9.983775718651229e-05, "loss": 0.2053, "step": 3560 }, { "epoch": 2.608791208791209, "grad_norm": 1.140625, "learning_rate": 9.983747988177459e-05, "loss": 0.2013, "step": 3561 }, { "epoch": 2.6095238095238096, "grad_norm": 0.8359375, "learning_rate": 9.983720234064484e-05, "loss": 0.1505, "step": 3562 }, { "epoch": 2.6102564102564103, "grad_norm": 1.296875, "learning_rate": 9.983692456312435e-05, "loss": 0.1982, "step": 3563 }, { "epoch": 2.610989010989011, "grad_norm": 1.03125, "learning_rate": 9.983664654921446e-05, "loss": 0.1799, "step": 3564 }, { "epoch": 2.6117216117216118, "grad_norm": 0.7421875, "learning_rate": 9.983636829891652e-05, "loss": 0.1387, "step": 3565 }, { "epoch": 2.6124542124542125, "grad_norm": 1.25, "learning_rate": 9.983608981223181e-05, "loss": 0.1818, "step": 3566 }, { "epoch": 2.6131868131868132, "grad_norm": 1.265625, "learning_rate": 9.983581108916172e-05, "loss": 0.2296, "step": 3567 }, { "epoch": 2.613919413919414, "grad_norm": 0.7265625, "learning_rate": 9.983553212970755e-05, "loss": 0.1894, "step": 3568 }, { "epoch": 2.6146520146520147, "grad_norm": 0.74609375, "learning_rate": 9.983525293387066e-05, "loss": 0.182, "step": 3569 }, { "epoch": 2.6153846153846154, "grad_norm": 0.88671875, "learning_rate": 9.983497350165236e-05, "loss": 0.1786, "step": 3570 }, { "epoch": 2.616117216117216, "grad_norm": 0.72265625, "learning_rate": 9.983469383305402e-05, "loss": 0.1672, "step": 3571 }, { "epoch": 2.616849816849817, "grad_norm": 0.66796875, "learning_rate": 9.983441392807696e-05, "loss": 0.1497, "step": 3572 }, { "epoch": 2.6175824175824176, "grad_norm": 0.7890625, "learning_rate": 9.983413378672254e-05, "loss": 0.1343, "step": 3573 }, { "epoch": 2.6183150183150183, "grad_norm": 0.9765625, "learning_rate": 9.983385340899208e-05, "loss": 0.1998, "step": 3574 }, { "epoch": 2.619047619047619, "grad_norm": 1.28125, "learning_rate": 9.983357279488694e-05, "loss": 0.2054, "step": 3575 }, { "epoch": 2.61978021978022, "grad_norm": 0.96484375, "learning_rate": 9.983329194440846e-05, "loss": 0.1332, "step": 3576 }, { "epoch": 2.6205128205128205, "grad_norm": 1.671875, "learning_rate": 9.983301085755797e-05, "loss": 0.2036, "step": 3577 }, { "epoch": 2.6212454212454213, "grad_norm": 0.7421875, "learning_rate": 9.983272953433685e-05, "loss": 0.1754, "step": 3578 }, { "epoch": 2.621978021978022, "grad_norm": 0.91015625, "learning_rate": 9.98324479747464e-05, "loss": 0.2075, "step": 3579 }, { "epoch": 2.6227106227106227, "grad_norm": 0.93359375, "learning_rate": 9.983216617878802e-05, "loss": 0.1474, "step": 3580 }, { "epoch": 2.6234432234432234, "grad_norm": 1.1796875, "learning_rate": 9.983188414646302e-05, "loss": 0.2094, "step": 3581 }, { "epoch": 2.624175824175824, "grad_norm": 1.6015625, "learning_rate": 9.983160187777279e-05, "loss": 0.1763, "step": 3582 }, { "epoch": 2.624908424908425, "grad_norm": 1.0078125, "learning_rate": 9.983131937271863e-05, "loss": 0.1927, "step": 3583 }, { "epoch": 2.6256410256410256, "grad_norm": 0.490234375, "learning_rate": 9.983103663130194e-05, "loss": 0.1391, "step": 3584 }, { "epoch": 2.6263736263736264, "grad_norm": 1.125, "learning_rate": 9.983075365352407e-05, "loss": 0.1986, "step": 3585 }, { "epoch": 2.627106227106227, "grad_norm": 0.71484375, "learning_rate": 9.983047043938636e-05, "loss": 0.1499, "step": 3586 }, { "epoch": 2.627838827838828, "grad_norm": 1.0078125, "learning_rate": 9.983018698889018e-05, "loss": 0.1572, "step": 3587 }, { "epoch": 2.6285714285714286, "grad_norm": 0.953125, "learning_rate": 9.982990330203684e-05, "loss": 0.2062, "step": 3588 }, { "epoch": 2.6293040293040293, "grad_norm": 0.8125, "learning_rate": 9.982961937882778e-05, "loss": 0.1827, "step": 3589 }, { "epoch": 2.63003663003663, "grad_norm": 0.69921875, "learning_rate": 9.98293352192643e-05, "loss": 0.1551, "step": 3590 }, { "epoch": 2.6307692307692307, "grad_norm": 1.140625, "learning_rate": 9.982905082334779e-05, "loss": 0.183, "step": 3591 }, { "epoch": 2.6315018315018315, "grad_norm": 0.84765625, "learning_rate": 9.982876619107961e-05, "loss": 0.1656, "step": 3592 }, { "epoch": 2.632234432234432, "grad_norm": 1.4140625, "learning_rate": 9.982848132246112e-05, "loss": 0.1972, "step": 3593 }, { "epoch": 2.632967032967033, "grad_norm": 1.0625, "learning_rate": 9.982819621749366e-05, "loss": 0.1948, "step": 3594 }, { "epoch": 2.6336996336996337, "grad_norm": 0.765625, "learning_rate": 9.982791087617863e-05, "loss": 0.1497, "step": 3595 }, { "epoch": 2.6344322344322344, "grad_norm": 1.0546875, "learning_rate": 9.98276252985174e-05, "loss": 0.1669, "step": 3596 }, { "epoch": 2.635164835164835, "grad_norm": 1.65625, "learning_rate": 9.98273394845113e-05, "loss": 0.2453, "step": 3597 }, { "epoch": 2.635897435897436, "grad_norm": 0.94140625, "learning_rate": 9.982705343416174e-05, "loss": 0.1617, "step": 3598 }, { "epoch": 2.6366300366300366, "grad_norm": 1.453125, "learning_rate": 9.982676714747006e-05, "loss": 0.1848, "step": 3599 }, { "epoch": 2.6373626373626373, "grad_norm": 1.40625, "learning_rate": 9.982648062443764e-05, "loss": 0.1869, "step": 3600 }, { "epoch": 2.638095238095238, "grad_norm": 0.890625, "learning_rate": 9.98261938650659e-05, "loss": 0.1632, "step": 3601 }, { "epoch": 2.638827838827839, "grad_norm": 1.3515625, "learning_rate": 9.982590686935614e-05, "loss": 0.1672, "step": 3602 }, { "epoch": 2.6395604395604395, "grad_norm": 1.1328125, "learning_rate": 9.982561963730978e-05, "loss": 0.176, "step": 3603 }, { "epoch": 2.6402930402930402, "grad_norm": 1.3125, "learning_rate": 9.982533216892816e-05, "loss": 0.1732, "step": 3604 }, { "epoch": 2.641025641025641, "grad_norm": 1.7890625, "learning_rate": 9.982504446421271e-05, "loss": 0.1535, "step": 3605 }, { "epoch": 2.6417582417582417, "grad_norm": 2.09375, "learning_rate": 9.982475652316476e-05, "loss": 0.1805, "step": 3606 }, { "epoch": 2.6424908424908424, "grad_norm": 2.21875, "learning_rate": 9.982446834578571e-05, "loss": 0.1845, "step": 3607 }, { "epoch": 2.643223443223443, "grad_norm": 1.203125, "learning_rate": 9.982417993207694e-05, "loss": 0.1657, "step": 3608 }, { "epoch": 2.643956043956044, "grad_norm": 1.2421875, "learning_rate": 9.982389128203985e-05, "loss": 0.1139, "step": 3609 }, { "epoch": 2.6446886446886446, "grad_norm": 0.78515625, "learning_rate": 9.982360239567579e-05, "loss": 0.1646, "step": 3610 }, { "epoch": 2.6454212454212453, "grad_norm": 1.046875, "learning_rate": 9.982331327298615e-05, "loss": 0.191, "step": 3611 }, { "epoch": 2.646153846153846, "grad_norm": 1.4375, "learning_rate": 9.982302391397233e-05, "loss": 0.174, "step": 3612 }, { "epoch": 2.646886446886447, "grad_norm": 1.3046875, "learning_rate": 9.982273431863573e-05, "loss": 0.1743, "step": 3613 }, { "epoch": 2.6476190476190475, "grad_norm": 1.4921875, "learning_rate": 9.982244448697771e-05, "loss": 0.1563, "step": 3614 }, { "epoch": 2.6483516483516483, "grad_norm": 1.828125, "learning_rate": 9.982215441899966e-05, "loss": 0.1799, "step": 3615 }, { "epoch": 2.649084249084249, "grad_norm": 1.078125, "learning_rate": 9.982186411470297e-05, "loss": 0.1714, "step": 3616 }, { "epoch": 2.6498168498168497, "grad_norm": 1.0859375, "learning_rate": 9.982157357408905e-05, "loss": 0.2143, "step": 3617 }, { "epoch": 2.6505494505494505, "grad_norm": 0.74609375, "learning_rate": 9.982128279715928e-05, "loss": 0.1607, "step": 3618 }, { "epoch": 2.651282051282051, "grad_norm": 0.9140625, "learning_rate": 9.982099178391505e-05, "loss": 0.169, "step": 3619 }, { "epoch": 2.652014652014652, "grad_norm": 1.4921875, "learning_rate": 9.982070053435775e-05, "loss": 0.1899, "step": 3620 }, { "epoch": 2.6527472527472526, "grad_norm": 1.5390625, "learning_rate": 9.982040904848881e-05, "loss": 0.1907, "step": 3621 }, { "epoch": 2.6534798534798534, "grad_norm": 1.0234375, "learning_rate": 9.982011732630956e-05, "loss": 0.16, "step": 3622 }, { "epoch": 2.654212454212454, "grad_norm": 1.65625, "learning_rate": 9.981982536782147e-05, "loss": 0.2335, "step": 3623 }, { "epoch": 2.654945054945055, "grad_norm": 0.90625, "learning_rate": 9.981953317302589e-05, "loss": 0.1854, "step": 3624 }, { "epoch": 2.6556776556776556, "grad_norm": 1.3984375, "learning_rate": 9.981924074192424e-05, "loss": 0.1727, "step": 3625 }, { "epoch": 2.6564102564102563, "grad_norm": 1.484375, "learning_rate": 9.98189480745179e-05, "loss": 0.1918, "step": 3626 }, { "epoch": 2.657142857142857, "grad_norm": 1.0625, "learning_rate": 9.981865517080831e-05, "loss": 0.1628, "step": 3627 }, { "epoch": 2.6578754578754578, "grad_norm": 1.046875, "learning_rate": 9.981836203079685e-05, "loss": 0.1841, "step": 3628 }, { "epoch": 2.6586080586080585, "grad_norm": 1.3046875, "learning_rate": 9.981806865448493e-05, "loss": 0.1524, "step": 3629 }, { "epoch": 2.659340659340659, "grad_norm": 1.4921875, "learning_rate": 9.981777504187394e-05, "loss": 0.176, "step": 3630 }, { "epoch": 2.66007326007326, "grad_norm": 2.046875, "learning_rate": 9.981748119296531e-05, "loss": 0.2356, "step": 3631 }, { "epoch": 2.6608058608058607, "grad_norm": 1.0390625, "learning_rate": 9.981718710776043e-05, "loss": 0.1868, "step": 3632 }, { "epoch": 2.6615384615384614, "grad_norm": 1.4609375, "learning_rate": 9.981689278626074e-05, "loss": 0.1564, "step": 3633 }, { "epoch": 2.662271062271062, "grad_norm": 1.1171875, "learning_rate": 9.981659822846761e-05, "loss": 0.1469, "step": 3634 }, { "epoch": 2.663003663003663, "grad_norm": 1.5625, "learning_rate": 9.981630343438246e-05, "loss": 0.1614, "step": 3635 }, { "epoch": 2.6637362637362636, "grad_norm": 0.921875, "learning_rate": 9.981600840400672e-05, "loss": 0.1621, "step": 3636 }, { "epoch": 2.6644688644688643, "grad_norm": 0.8984375, "learning_rate": 9.981571313734179e-05, "loss": 0.1443, "step": 3637 }, { "epoch": 2.665201465201465, "grad_norm": 1.125, "learning_rate": 9.98154176343891e-05, "loss": 0.1728, "step": 3638 }, { "epoch": 2.665934065934066, "grad_norm": 1.046875, "learning_rate": 9.981512189515004e-05, "loss": 0.1659, "step": 3639 }, { "epoch": 2.6666666666666665, "grad_norm": 0.88671875, "learning_rate": 9.981482591962606e-05, "loss": 0.1757, "step": 3640 }, { "epoch": 2.6673992673992672, "grad_norm": 0.76953125, "learning_rate": 9.981452970781853e-05, "loss": 0.167, "step": 3641 }, { "epoch": 2.668131868131868, "grad_norm": 0.953125, "learning_rate": 9.981423325972893e-05, "loss": 0.1691, "step": 3642 }, { "epoch": 2.6688644688644687, "grad_norm": 1.4140625, "learning_rate": 9.981393657535864e-05, "loss": 0.1638, "step": 3643 }, { "epoch": 2.6695970695970694, "grad_norm": 1.1328125, "learning_rate": 9.981363965470908e-05, "loss": 0.1998, "step": 3644 }, { "epoch": 2.67032967032967, "grad_norm": 1.96875, "learning_rate": 9.98133424977817e-05, "loss": 0.1693, "step": 3645 }, { "epoch": 2.671062271062271, "grad_norm": 1.0625, "learning_rate": 9.98130451045779e-05, "loss": 0.1769, "step": 3646 }, { "epoch": 2.6717948717948716, "grad_norm": 2.140625, "learning_rate": 9.98127474750991e-05, "loss": 0.1836, "step": 3647 }, { "epoch": 2.672527472527473, "grad_norm": 1.21875, "learning_rate": 9.981244960934675e-05, "loss": 0.1956, "step": 3648 }, { "epoch": 2.6732600732600735, "grad_norm": 0.8359375, "learning_rate": 9.981215150732227e-05, "loss": 0.1925, "step": 3649 }, { "epoch": 2.6739926739926743, "grad_norm": 1.078125, "learning_rate": 9.98118531690271e-05, "loss": 0.1429, "step": 3650 }, { "epoch": 2.674725274725275, "grad_norm": 0.6953125, "learning_rate": 9.981155459446262e-05, "loss": 0.1486, "step": 3651 }, { "epoch": 2.6754578754578757, "grad_norm": 0.73046875, "learning_rate": 9.98112557836303e-05, "loss": 0.1452, "step": 3652 }, { "epoch": 2.6761904761904765, "grad_norm": 0.859375, "learning_rate": 9.981095673653159e-05, "loss": 0.1693, "step": 3653 }, { "epoch": 2.676923076923077, "grad_norm": 0.8359375, "learning_rate": 9.981065745316788e-05, "loss": 0.1701, "step": 3654 }, { "epoch": 2.677655677655678, "grad_norm": 1.15625, "learning_rate": 9.981035793354062e-05, "loss": 0.1611, "step": 3655 }, { "epoch": 2.6783882783882786, "grad_norm": 0.6328125, "learning_rate": 9.981005817765125e-05, "loss": 0.1674, "step": 3656 }, { "epoch": 2.6791208791208794, "grad_norm": 0.7734375, "learning_rate": 9.980975818550122e-05, "loss": 0.1816, "step": 3657 }, { "epoch": 2.67985347985348, "grad_norm": 0.96875, "learning_rate": 9.980945795709193e-05, "loss": 0.1489, "step": 3658 }, { "epoch": 2.680586080586081, "grad_norm": 0.953125, "learning_rate": 9.980915749242484e-05, "loss": 0.1662, "step": 3659 }, { "epoch": 2.6813186813186816, "grad_norm": 0.8671875, "learning_rate": 9.98088567915014e-05, "loss": 0.1392, "step": 3660 }, { "epoch": 2.6820512820512823, "grad_norm": 0.765625, "learning_rate": 9.980855585432305e-05, "loss": 0.1567, "step": 3661 }, { "epoch": 2.682783882783883, "grad_norm": 1.1015625, "learning_rate": 9.98082546808912e-05, "loss": 0.1654, "step": 3662 }, { "epoch": 2.6835164835164838, "grad_norm": 0.98828125, "learning_rate": 9.980795327120732e-05, "loss": 0.1926, "step": 3663 }, { "epoch": 2.6842490842490845, "grad_norm": 0.9921875, "learning_rate": 9.980765162527285e-05, "loss": 0.1522, "step": 3664 }, { "epoch": 2.684981684981685, "grad_norm": 1.1875, "learning_rate": 9.980734974308924e-05, "loss": 0.202, "step": 3665 }, { "epoch": 2.685714285714286, "grad_norm": 1.265625, "learning_rate": 9.980704762465794e-05, "loss": 0.1749, "step": 3666 }, { "epoch": 2.6864468864468867, "grad_norm": 0.828125, "learning_rate": 9.980674526998037e-05, "loss": 0.2112, "step": 3667 }, { "epoch": 2.6871794871794874, "grad_norm": 0.65625, "learning_rate": 9.9806442679058e-05, "loss": 0.1704, "step": 3668 }, { "epoch": 2.687912087912088, "grad_norm": 1.2734375, "learning_rate": 9.980613985189229e-05, "loss": 0.1956, "step": 3669 }, { "epoch": 2.688644688644689, "grad_norm": 0.87890625, "learning_rate": 9.980583678848467e-05, "loss": 0.1615, "step": 3670 }, { "epoch": 2.6893772893772896, "grad_norm": 0.546875, "learning_rate": 9.980553348883659e-05, "loss": 0.1694, "step": 3671 }, { "epoch": 2.6901098901098903, "grad_norm": 1.21875, "learning_rate": 9.980522995294952e-05, "loss": 0.1628, "step": 3672 }, { "epoch": 2.690842490842491, "grad_norm": 1.7265625, "learning_rate": 9.980492618082491e-05, "loss": 0.1863, "step": 3673 }, { "epoch": 2.691575091575092, "grad_norm": 1.1328125, "learning_rate": 9.98046221724642e-05, "loss": 0.1532, "step": 3674 }, { "epoch": 2.6923076923076925, "grad_norm": 1.09375, "learning_rate": 9.980431792786888e-05, "loss": 0.1895, "step": 3675 }, { "epoch": 2.6930402930402932, "grad_norm": 0.98828125, "learning_rate": 9.980401344704039e-05, "loss": 0.1477, "step": 3676 }, { "epoch": 2.693772893772894, "grad_norm": 1.0390625, "learning_rate": 9.980370872998017e-05, "loss": 0.1924, "step": 3677 }, { "epoch": 2.6945054945054947, "grad_norm": 1.625, "learning_rate": 9.98034037766897e-05, "loss": 0.1954, "step": 3678 }, { "epoch": 2.6952380952380954, "grad_norm": 2.390625, "learning_rate": 9.980309858717044e-05, "loss": 0.2146, "step": 3679 }, { "epoch": 2.695970695970696, "grad_norm": 1.0625, "learning_rate": 9.980279316142385e-05, "loss": 0.1769, "step": 3680 }, { "epoch": 2.696703296703297, "grad_norm": 1.046875, "learning_rate": 9.980248749945139e-05, "loss": 0.1777, "step": 3681 }, { "epoch": 2.6974358974358976, "grad_norm": 2.0625, "learning_rate": 9.980218160125452e-05, "loss": 0.208, "step": 3682 }, { "epoch": 2.6981684981684984, "grad_norm": 1.6953125, "learning_rate": 9.980187546683472e-05, "loss": 0.2063, "step": 3683 }, { "epoch": 2.698901098901099, "grad_norm": 1.6171875, "learning_rate": 9.980156909619344e-05, "loss": 0.2058, "step": 3684 }, { "epoch": 2.6996336996337, "grad_norm": 0.97265625, "learning_rate": 9.980126248933218e-05, "loss": 0.1806, "step": 3685 }, { "epoch": 2.7003663003663005, "grad_norm": 0.74609375, "learning_rate": 9.980095564625239e-05, "loss": 0.1464, "step": 3686 }, { "epoch": 2.7010989010989013, "grad_norm": 1.09375, "learning_rate": 9.98006485669555e-05, "loss": 0.195, "step": 3687 }, { "epoch": 2.701831501831502, "grad_norm": 0.78125, "learning_rate": 9.980034125144305e-05, "loss": 0.121, "step": 3688 }, { "epoch": 2.7025641025641027, "grad_norm": 0.93359375, "learning_rate": 9.980003369971648e-05, "loss": 0.1746, "step": 3689 }, { "epoch": 2.7032967032967035, "grad_norm": 1.4453125, "learning_rate": 9.979972591177725e-05, "loss": 0.1739, "step": 3690 }, { "epoch": 2.704029304029304, "grad_norm": 0.984375, "learning_rate": 9.979941788762686e-05, "loss": 0.1538, "step": 3691 }, { "epoch": 2.704761904761905, "grad_norm": 1.1640625, "learning_rate": 9.979910962726678e-05, "loss": 0.2107, "step": 3692 }, { "epoch": 2.7054945054945057, "grad_norm": 1.046875, "learning_rate": 9.979880113069846e-05, "loss": 0.187, "step": 3693 }, { "epoch": 2.7062271062271064, "grad_norm": 0.890625, "learning_rate": 9.979849239792342e-05, "loss": 0.1921, "step": 3694 }, { "epoch": 2.706959706959707, "grad_norm": 1.1640625, "learning_rate": 9.97981834289431e-05, "loss": 0.196, "step": 3695 }, { "epoch": 2.707692307692308, "grad_norm": 0.59375, "learning_rate": 9.979787422375901e-05, "loss": 0.1703, "step": 3696 }, { "epoch": 2.7084249084249086, "grad_norm": 0.78125, "learning_rate": 9.979756478237264e-05, "loss": 0.1708, "step": 3697 }, { "epoch": 2.7091575091575093, "grad_norm": 0.91015625, "learning_rate": 9.979725510478542e-05, "loss": 0.1613, "step": 3698 }, { "epoch": 2.70989010989011, "grad_norm": 0.6484375, "learning_rate": 9.979694519099888e-05, "loss": 0.1693, "step": 3699 }, { "epoch": 2.7106227106227108, "grad_norm": 1.015625, "learning_rate": 9.97966350410145e-05, "loss": 0.2054, "step": 3700 }, { "epoch": 2.7113553113553115, "grad_norm": 0.828125, "learning_rate": 9.979632465483376e-05, "loss": 0.1304, "step": 3701 }, { "epoch": 2.7120879120879122, "grad_norm": 0.9375, "learning_rate": 9.979601403245816e-05, "loss": 0.1553, "step": 3702 }, { "epoch": 2.712820512820513, "grad_norm": 0.9375, "learning_rate": 9.979570317388913e-05, "loss": 0.1894, "step": 3703 }, { "epoch": 2.7135531135531137, "grad_norm": 1.015625, "learning_rate": 9.979539207912824e-05, "loss": 0.1489, "step": 3704 }, { "epoch": 2.7142857142857144, "grad_norm": 1.0234375, "learning_rate": 9.979508074817693e-05, "loss": 0.1505, "step": 3705 }, { "epoch": 2.715018315018315, "grad_norm": 1.2578125, "learning_rate": 9.979476918103671e-05, "loss": 0.2216, "step": 3706 }, { "epoch": 2.715750915750916, "grad_norm": 0.88671875, "learning_rate": 9.979445737770904e-05, "loss": 0.1989, "step": 3707 }, { "epoch": 2.7164835164835166, "grad_norm": 1.921875, "learning_rate": 9.979414533819547e-05, "loss": 0.2249, "step": 3708 }, { "epoch": 2.7172161172161173, "grad_norm": 1.34375, "learning_rate": 9.979383306249748e-05, "loss": 0.1795, "step": 3709 }, { "epoch": 2.717948717948718, "grad_norm": 0.9921875, "learning_rate": 9.979352055061654e-05, "loss": 0.1736, "step": 3710 }, { "epoch": 2.718681318681319, "grad_norm": 0.82421875, "learning_rate": 9.979320780255416e-05, "loss": 0.1417, "step": 3711 }, { "epoch": 2.7194139194139195, "grad_norm": 0.96484375, "learning_rate": 9.979289481831185e-05, "loss": 0.168, "step": 3712 }, { "epoch": 2.7201465201465203, "grad_norm": 1.0625, "learning_rate": 9.979258159789107e-05, "loss": 0.1795, "step": 3713 }, { "epoch": 2.720879120879121, "grad_norm": 1.1015625, "learning_rate": 9.979226814129337e-05, "loss": 0.1886, "step": 3714 }, { "epoch": 2.7216117216117217, "grad_norm": 1.234375, "learning_rate": 9.979195444852024e-05, "loss": 0.1637, "step": 3715 }, { "epoch": 2.7223443223443224, "grad_norm": 1.03125, "learning_rate": 9.979164051957317e-05, "loss": 0.1805, "step": 3716 }, { "epoch": 2.723076923076923, "grad_norm": 1.0703125, "learning_rate": 9.979132635445367e-05, "loss": 0.1405, "step": 3717 }, { "epoch": 2.723809523809524, "grad_norm": 0.671875, "learning_rate": 9.979101195316323e-05, "loss": 0.1625, "step": 3718 }, { "epoch": 2.7245421245421246, "grad_norm": 1.0390625, "learning_rate": 9.979069731570337e-05, "loss": 0.1756, "step": 3719 }, { "epoch": 2.7252747252747254, "grad_norm": 1.109375, "learning_rate": 9.979038244207561e-05, "loss": 0.1999, "step": 3720 }, { "epoch": 2.726007326007326, "grad_norm": 1.0625, "learning_rate": 9.979006733228144e-05, "loss": 0.1597, "step": 3721 }, { "epoch": 2.726739926739927, "grad_norm": 1.390625, "learning_rate": 9.978975198632239e-05, "loss": 0.1389, "step": 3722 }, { "epoch": 2.7274725274725276, "grad_norm": 1.640625, "learning_rate": 9.978943640419994e-05, "loss": 0.1538, "step": 3723 }, { "epoch": 2.7282051282051283, "grad_norm": 1.8359375, "learning_rate": 9.978912058591563e-05, "loss": 0.1922, "step": 3724 }, { "epoch": 2.728937728937729, "grad_norm": 1.3984375, "learning_rate": 9.978880453147096e-05, "loss": 0.1641, "step": 3725 }, { "epoch": 2.7296703296703297, "grad_norm": 0.6875, "learning_rate": 9.978848824086745e-05, "loss": 0.1894, "step": 3726 }, { "epoch": 2.7304029304029305, "grad_norm": 1.2421875, "learning_rate": 9.97881717141066e-05, "loss": 0.1746, "step": 3727 }, { "epoch": 2.731135531135531, "grad_norm": 1.3359375, "learning_rate": 9.978785495118996e-05, "loss": 0.1971, "step": 3728 }, { "epoch": 2.731868131868132, "grad_norm": 1.0078125, "learning_rate": 9.978753795211902e-05, "loss": 0.1449, "step": 3729 }, { "epoch": 2.7326007326007327, "grad_norm": 0.9765625, "learning_rate": 9.97872207168953e-05, "loss": 0.1712, "step": 3730 }, { "epoch": 2.7333333333333334, "grad_norm": 1.1953125, "learning_rate": 9.978690324552031e-05, "loss": 0.185, "step": 3731 }, { "epoch": 2.734065934065934, "grad_norm": 1.125, "learning_rate": 9.978658553799561e-05, "loss": 0.1391, "step": 3732 }, { "epoch": 2.734798534798535, "grad_norm": 0.9765625, "learning_rate": 9.978626759432269e-05, "loss": 0.1489, "step": 3733 }, { "epoch": 2.7355311355311356, "grad_norm": 1.046875, "learning_rate": 9.978594941450308e-05, "loss": 0.1769, "step": 3734 }, { "epoch": 2.7362637362637363, "grad_norm": 0.875, "learning_rate": 9.978563099853833e-05, "loss": 0.1564, "step": 3735 }, { "epoch": 2.736996336996337, "grad_norm": 0.94921875, "learning_rate": 9.978531234642992e-05, "loss": 0.1386, "step": 3736 }, { "epoch": 2.7377289377289378, "grad_norm": 1.3046875, "learning_rate": 9.97849934581794e-05, "loss": 0.1699, "step": 3737 }, { "epoch": 2.7384615384615385, "grad_norm": 0.7109375, "learning_rate": 9.978467433378829e-05, "loss": 0.1799, "step": 3738 }, { "epoch": 2.7391941391941392, "grad_norm": 1.21875, "learning_rate": 9.978435497325814e-05, "loss": 0.1544, "step": 3739 }, { "epoch": 2.73992673992674, "grad_norm": 0.81640625, "learning_rate": 9.978403537659047e-05, "loss": 0.1934, "step": 3740 }, { "epoch": 2.7406593406593407, "grad_norm": 0.69921875, "learning_rate": 9.978371554378681e-05, "loss": 0.1353, "step": 3741 }, { "epoch": 2.7413919413919414, "grad_norm": 1.4765625, "learning_rate": 9.978339547484866e-05, "loss": 0.2017, "step": 3742 }, { "epoch": 2.742124542124542, "grad_norm": 1.7578125, "learning_rate": 9.97830751697776e-05, "loss": 0.1523, "step": 3743 }, { "epoch": 2.742857142857143, "grad_norm": 1.2890625, "learning_rate": 9.978275462857516e-05, "loss": 0.1694, "step": 3744 }, { "epoch": 2.7435897435897436, "grad_norm": 1.4609375, "learning_rate": 9.978243385124286e-05, "loss": 0.2185, "step": 3745 }, { "epoch": 2.7443223443223443, "grad_norm": 0.9921875, "learning_rate": 9.978211283778224e-05, "loss": 0.1466, "step": 3746 }, { "epoch": 2.745054945054945, "grad_norm": 1.765625, "learning_rate": 9.978179158819484e-05, "loss": 0.1891, "step": 3747 }, { "epoch": 2.745787545787546, "grad_norm": 2.0625, "learning_rate": 9.97814701024822e-05, "loss": 0.1808, "step": 3748 }, { "epoch": 2.7465201465201465, "grad_norm": 1.7890625, "learning_rate": 9.978114838064585e-05, "loss": 0.2254, "step": 3749 }, { "epoch": 2.7472527472527473, "grad_norm": 1.390625, "learning_rate": 9.978082642268733e-05, "loss": 0.1651, "step": 3750 }, { "epoch": 2.747985347985348, "grad_norm": 0.97265625, "learning_rate": 9.978050422860822e-05, "loss": 0.131, "step": 3751 }, { "epoch": 2.7487179487179487, "grad_norm": 1.171875, "learning_rate": 9.978018179841002e-05, "loss": 0.1787, "step": 3752 }, { "epoch": 2.7494505494505495, "grad_norm": 1.328125, "learning_rate": 9.977985913209429e-05, "loss": 0.1853, "step": 3753 }, { "epoch": 2.75018315018315, "grad_norm": 1.609375, "learning_rate": 9.977953622966258e-05, "loss": 0.1818, "step": 3754 }, { "epoch": 2.750915750915751, "grad_norm": 1.2265625, "learning_rate": 9.977921309111644e-05, "loss": 0.1786, "step": 3755 }, { "epoch": 2.7516483516483516, "grad_norm": 1.0859375, "learning_rate": 9.97788897164574e-05, "loss": 0.1765, "step": 3756 }, { "epoch": 2.7523809523809524, "grad_norm": 1.109375, "learning_rate": 9.977856610568705e-05, "loss": 0.1883, "step": 3757 }, { "epoch": 2.753113553113553, "grad_norm": 1.5078125, "learning_rate": 9.977824225880689e-05, "loss": 0.1993, "step": 3758 }, { "epoch": 2.753846153846154, "grad_norm": 1.4609375, "learning_rate": 9.97779181758185e-05, "loss": 0.1859, "step": 3759 }, { "epoch": 2.7545787545787546, "grad_norm": 1.234375, "learning_rate": 9.977759385672342e-05, "loss": 0.1635, "step": 3760 }, { "epoch": 2.7553113553113553, "grad_norm": 0.87890625, "learning_rate": 9.977726930152321e-05, "loss": 0.1634, "step": 3761 }, { "epoch": 2.756043956043956, "grad_norm": 1.0234375, "learning_rate": 9.977694451021943e-05, "loss": 0.1723, "step": 3762 }, { "epoch": 2.7567765567765568, "grad_norm": 0.93359375, "learning_rate": 9.977661948281365e-05, "loss": 0.1982, "step": 3763 }, { "epoch": 2.7575091575091575, "grad_norm": 1.234375, "learning_rate": 9.97762942193074e-05, "loss": 0.1801, "step": 3764 }, { "epoch": 2.758241758241758, "grad_norm": 1.1484375, "learning_rate": 9.977596871970223e-05, "loss": 0.1778, "step": 3765 }, { "epoch": 2.758974358974359, "grad_norm": 0.984375, "learning_rate": 9.977564298399973e-05, "loss": 0.2049, "step": 3766 }, { "epoch": 2.7597069597069597, "grad_norm": 1.2421875, "learning_rate": 9.977531701220146e-05, "loss": 0.2253, "step": 3767 }, { "epoch": 2.7604395604395604, "grad_norm": 0.59765625, "learning_rate": 9.977499080430896e-05, "loss": 0.1531, "step": 3768 }, { "epoch": 2.761172161172161, "grad_norm": 1.296875, "learning_rate": 9.977466436032379e-05, "loss": 0.1814, "step": 3769 }, { "epoch": 2.761904761904762, "grad_norm": 1.015625, "learning_rate": 9.977433768024754e-05, "loss": 0.154, "step": 3770 }, { "epoch": 2.7626373626373626, "grad_norm": 1.34375, "learning_rate": 9.977401076408176e-05, "loss": 0.22, "step": 3771 }, { "epoch": 2.7633699633699633, "grad_norm": 1.0625, "learning_rate": 9.977368361182802e-05, "loss": 0.1999, "step": 3772 }, { "epoch": 2.764102564102564, "grad_norm": 1.4921875, "learning_rate": 9.977335622348788e-05, "loss": 0.1866, "step": 3773 }, { "epoch": 2.764835164835165, "grad_norm": 1.3984375, "learning_rate": 9.977302859906293e-05, "loss": 0.1946, "step": 3774 }, { "epoch": 2.7655677655677655, "grad_norm": 1.28125, "learning_rate": 9.977270073855471e-05, "loss": 0.1484, "step": 3775 }, { "epoch": 2.7663003663003662, "grad_norm": 1.09375, "learning_rate": 9.977237264196482e-05, "loss": 0.1621, "step": 3776 }, { "epoch": 2.767032967032967, "grad_norm": 0.8125, "learning_rate": 9.97720443092948e-05, "loss": 0.1681, "step": 3777 }, { "epoch": 2.7677655677655677, "grad_norm": 0.9375, "learning_rate": 9.977171574054626e-05, "loss": 0.1801, "step": 3778 }, { "epoch": 2.7684981684981684, "grad_norm": 1.0859375, "learning_rate": 9.977138693572075e-05, "loss": 0.1321, "step": 3779 }, { "epoch": 2.769230769230769, "grad_norm": 1.03125, "learning_rate": 9.977105789481984e-05, "loss": 0.1508, "step": 3780 }, { "epoch": 2.76996336996337, "grad_norm": 0.7421875, "learning_rate": 9.977072861784514e-05, "loss": 0.1655, "step": 3781 }, { "epoch": 2.7706959706959706, "grad_norm": 0.99609375, "learning_rate": 9.977039910479818e-05, "loss": 0.1459, "step": 3782 }, { "epoch": 2.7714285714285714, "grad_norm": 0.609375, "learning_rate": 9.977006935568056e-05, "loss": 0.1339, "step": 3783 }, { "epoch": 2.772161172161172, "grad_norm": 1.0234375, "learning_rate": 9.976973937049388e-05, "loss": 0.17, "step": 3784 }, { "epoch": 2.772893772893773, "grad_norm": 0.6484375, "learning_rate": 9.976940914923969e-05, "loss": 0.1504, "step": 3785 }, { "epoch": 2.7736263736263735, "grad_norm": 0.9140625, "learning_rate": 9.97690786919196e-05, "loss": 0.1763, "step": 3786 }, { "epoch": 2.7743589743589743, "grad_norm": 1.3671875, "learning_rate": 9.976874799853518e-05, "loss": 0.1857, "step": 3787 }, { "epoch": 2.775091575091575, "grad_norm": 0.81640625, "learning_rate": 9.9768417069088e-05, "loss": 0.1689, "step": 3788 }, { "epoch": 2.7758241758241757, "grad_norm": 0.98828125, "learning_rate": 9.976808590357968e-05, "loss": 0.1389, "step": 3789 }, { "epoch": 2.7765567765567765, "grad_norm": 0.93359375, "learning_rate": 9.976775450201177e-05, "loss": 0.1786, "step": 3790 }, { "epoch": 2.777289377289377, "grad_norm": 0.640625, "learning_rate": 9.976742286438589e-05, "loss": 0.1594, "step": 3791 }, { "epoch": 2.778021978021978, "grad_norm": 1.21875, "learning_rate": 9.97670909907036e-05, "loss": 0.1627, "step": 3792 }, { "epoch": 2.7787545787545787, "grad_norm": 1.4140625, "learning_rate": 9.976675888096651e-05, "loss": 0.2126, "step": 3793 }, { "epoch": 2.7794871794871794, "grad_norm": 0.640625, "learning_rate": 9.97664265351762e-05, "loss": 0.1697, "step": 3794 }, { "epoch": 2.78021978021978, "grad_norm": 0.9375, "learning_rate": 9.976609395333427e-05, "loss": 0.1777, "step": 3795 }, { "epoch": 2.780952380952381, "grad_norm": 1.5390625, "learning_rate": 9.976576113544232e-05, "loss": 0.1652, "step": 3796 }, { "epoch": 2.7816849816849816, "grad_norm": 0.86328125, "learning_rate": 9.976542808150192e-05, "loss": 0.1676, "step": 3797 }, { "epoch": 2.7824175824175823, "grad_norm": 0.7578125, "learning_rate": 9.976509479151469e-05, "loss": 0.128, "step": 3798 }, { "epoch": 2.783150183150183, "grad_norm": 0.78515625, "learning_rate": 9.976476126548223e-05, "loss": 0.189, "step": 3799 }, { "epoch": 2.7838827838827838, "grad_norm": 0.9921875, "learning_rate": 9.97644275034061e-05, "loss": 0.1734, "step": 3800 }, { "epoch": 2.7846153846153845, "grad_norm": 1.546875, "learning_rate": 9.976409350528793e-05, "loss": 0.1767, "step": 3801 }, { "epoch": 2.7853479853479852, "grad_norm": 1.484375, "learning_rate": 9.976375927112934e-05, "loss": 0.1712, "step": 3802 }, { "epoch": 2.786080586080586, "grad_norm": 1.296875, "learning_rate": 9.976342480093188e-05, "loss": 0.1638, "step": 3803 }, { "epoch": 2.7868131868131867, "grad_norm": 0.9375, "learning_rate": 9.976309009469719e-05, "loss": 0.1391, "step": 3804 }, { "epoch": 2.7875457875457874, "grad_norm": 1.171875, "learning_rate": 9.976275515242686e-05, "loss": 0.1601, "step": 3805 }, { "epoch": 2.788278388278388, "grad_norm": 1.78125, "learning_rate": 9.97624199741225e-05, "loss": 0.1646, "step": 3806 }, { "epoch": 2.789010989010989, "grad_norm": 1.78125, "learning_rate": 9.976208455978573e-05, "loss": 0.1589, "step": 3807 }, { "epoch": 2.7897435897435896, "grad_norm": 0.77734375, "learning_rate": 9.976174890941813e-05, "loss": 0.1265, "step": 3808 }, { "epoch": 2.7904761904761903, "grad_norm": 0.84375, "learning_rate": 9.976141302302132e-05, "loss": 0.1591, "step": 3809 }, { "epoch": 2.791208791208791, "grad_norm": 1.34375, "learning_rate": 9.976107690059689e-05, "loss": 0.2047, "step": 3810 }, { "epoch": 2.791941391941392, "grad_norm": 1.1640625, "learning_rate": 9.97607405421465e-05, "loss": 0.1669, "step": 3811 }, { "epoch": 2.7926739926739925, "grad_norm": 1.421875, "learning_rate": 9.976040394767172e-05, "loss": 0.1822, "step": 3812 }, { "epoch": 2.7934065934065933, "grad_norm": 1.65625, "learning_rate": 9.976006711717416e-05, "loss": 0.2286, "step": 3813 }, { "epoch": 2.794139194139194, "grad_norm": 0.7265625, "learning_rate": 9.975973005065546e-05, "loss": 0.1517, "step": 3814 }, { "epoch": 2.7948717948717947, "grad_norm": 1.7265625, "learning_rate": 9.975939274811724e-05, "loss": 0.1687, "step": 3815 }, { "epoch": 2.7956043956043954, "grad_norm": 1.1875, "learning_rate": 9.975905520956106e-05, "loss": 0.1777, "step": 3816 }, { "epoch": 2.796336996336996, "grad_norm": 0.984375, "learning_rate": 9.975871743498859e-05, "loss": 0.1805, "step": 3817 }, { "epoch": 2.797069597069597, "grad_norm": 1.3046875, "learning_rate": 9.975837942440143e-05, "loss": 0.1533, "step": 3818 }, { "epoch": 2.7978021978021976, "grad_norm": 0.86328125, "learning_rate": 9.975804117780122e-05, "loss": 0.165, "step": 3819 }, { "epoch": 2.7985347985347984, "grad_norm": 0.8203125, "learning_rate": 9.975770269518955e-05, "loss": 0.1497, "step": 3820 }, { "epoch": 2.799267399267399, "grad_norm": 1.546875, "learning_rate": 9.975736397656807e-05, "loss": 0.2358, "step": 3821 }, { "epoch": 2.8, "grad_norm": 1.140625, "learning_rate": 9.975702502193839e-05, "loss": 0.1497, "step": 3822 }, { "epoch": 2.8007326007326006, "grad_norm": 0.97265625, "learning_rate": 9.975668583130212e-05, "loss": 0.1764, "step": 3823 }, { "epoch": 2.8014652014652013, "grad_norm": 0.79296875, "learning_rate": 9.975634640466091e-05, "loss": 0.1845, "step": 3824 }, { "epoch": 2.802197802197802, "grad_norm": 1.25, "learning_rate": 9.975600674201637e-05, "loss": 0.257, "step": 3825 }, { "epoch": 2.8029304029304027, "grad_norm": 1.5234375, "learning_rate": 9.975566684337012e-05, "loss": 0.2171, "step": 3826 }, { "epoch": 2.8036630036630035, "grad_norm": 1.1171875, "learning_rate": 9.97553267087238e-05, "loss": 0.1628, "step": 3827 }, { "epoch": 2.804395604395604, "grad_norm": 1.03125, "learning_rate": 9.975498633807906e-05, "loss": 0.205, "step": 3828 }, { "epoch": 2.805128205128205, "grad_norm": 0.9296875, "learning_rate": 9.97546457314375e-05, "loss": 0.18, "step": 3829 }, { "epoch": 2.8058608058608057, "grad_norm": 0.8984375, "learning_rate": 9.975430488880076e-05, "loss": 0.1926, "step": 3830 }, { "epoch": 2.8065934065934064, "grad_norm": 1.3515625, "learning_rate": 9.975396381017049e-05, "loss": 0.1564, "step": 3831 }, { "epoch": 2.807326007326007, "grad_norm": 2.03125, "learning_rate": 9.975362249554829e-05, "loss": 0.1715, "step": 3832 }, { "epoch": 2.808058608058608, "grad_norm": 1.9765625, "learning_rate": 9.975328094493583e-05, "loss": 0.1943, "step": 3833 }, { "epoch": 2.8087912087912086, "grad_norm": 1.078125, "learning_rate": 9.975293915833472e-05, "loss": 0.1743, "step": 3834 }, { "epoch": 2.8095238095238093, "grad_norm": 0.97265625, "learning_rate": 9.975259713574661e-05, "loss": 0.1579, "step": 3835 }, { "epoch": 2.81025641025641, "grad_norm": 1.078125, "learning_rate": 9.975225487717316e-05, "loss": 0.1655, "step": 3836 }, { "epoch": 2.8109890109890108, "grad_norm": 0.97265625, "learning_rate": 9.975191238261598e-05, "loss": 0.1949, "step": 3837 }, { "epoch": 2.8117216117216115, "grad_norm": 1.546875, "learning_rate": 9.975156965207672e-05, "loss": 0.1509, "step": 3838 }, { "epoch": 2.8124542124542122, "grad_norm": 1.3984375, "learning_rate": 9.9751226685557e-05, "loss": 0.1629, "step": 3839 }, { "epoch": 2.813186813186813, "grad_norm": 0.83984375, "learning_rate": 9.97508834830585e-05, "loss": 0.1763, "step": 3840 }, { "epoch": 2.8139194139194137, "grad_norm": 1.1171875, "learning_rate": 9.975054004458285e-05, "loss": 0.1984, "step": 3841 }, { "epoch": 2.8146520146520144, "grad_norm": 1.0078125, "learning_rate": 9.97501963701317e-05, "loss": 0.1633, "step": 3842 }, { "epoch": 2.815384615384615, "grad_norm": 1.3125, "learning_rate": 9.974985245970669e-05, "loss": 0.189, "step": 3843 }, { "epoch": 2.816117216117216, "grad_norm": 0.9296875, "learning_rate": 9.974950831330946e-05, "loss": 0.1584, "step": 3844 }, { "epoch": 2.8168498168498166, "grad_norm": 0.625, "learning_rate": 9.974916393094167e-05, "loss": 0.1488, "step": 3845 }, { "epoch": 2.8175824175824173, "grad_norm": 1.09375, "learning_rate": 9.974881931260499e-05, "loss": 0.1692, "step": 3846 }, { "epoch": 2.818315018315018, "grad_norm": 1.1015625, "learning_rate": 9.974847445830104e-05, "loss": 0.16, "step": 3847 }, { "epoch": 2.819047619047619, "grad_norm": 1.359375, "learning_rate": 9.974812936803146e-05, "loss": 0.2008, "step": 3848 }, { "epoch": 2.8197802197802195, "grad_norm": 0.6640625, "learning_rate": 9.974778404179794e-05, "loss": 0.161, "step": 3849 }, { "epoch": 2.8205128205128203, "grad_norm": 0.984375, "learning_rate": 9.974743847960212e-05, "loss": 0.182, "step": 3850 }, { "epoch": 2.821245421245421, "grad_norm": 0.7734375, "learning_rate": 9.974709268144565e-05, "loss": 0.1705, "step": 3851 }, { "epoch": 2.8219780219780217, "grad_norm": 0.734375, "learning_rate": 9.97467466473302e-05, "loss": 0.1469, "step": 3852 }, { "epoch": 2.8227106227106225, "grad_norm": 1.015625, "learning_rate": 9.97464003772574e-05, "loss": 0.1677, "step": 3853 }, { "epoch": 2.823443223443223, "grad_norm": 0.703125, "learning_rate": 9.974605387122896e-05, "loss": 0.1463, "step": 3854 }, { "epoch": 2.824175824175824, "grad_norm": 1.015625, "learning_rate": 9.974570712924648e-05, "loss": 0.1889, "step": 3855 }, { "epoch": 2.8249084249084246, "grad_norm": 0.91015625, "learning_rate": 9.974536015131167e-05, "loss": 0.1634, "step": 3856 }, { "epoch": 2.8256410256410254, "grad_norm": 0.73046875, "learning_rate": 9.974501293742617e-05, "loss": 0.2391, "step": 3857 }, { "epoch": 2.826373626373626, "grad_norm": 1.1015625, "learning_rate": 9.974466548759165e-05, "loss": 0.1472, "step": 3858 }, { "epoch": 2.827106227106227, "grad_norm": 1.0078125, "learning_rate": 9.974431780180976e-05, "loss": 0.1689, "step": 3859 }, { "epoch": 2.8278388278388276, "grad_norm": 0.671875, "learning_rate": 9.97439698800822e-05, "loss": 0.1279, "step": 3860 }, { "epoch": 2.8285714285714287, "grad_norm": 1.1796875, "learning_rate": 9.974362172241058e-05, "loss": 0.2279, "step": 3861 }, { "epoch": 2.8293040293040295, "grad_norm": 1.1015625, "learning_rate": 9.974327332879663e-05, "loss": 0.1839, "step": 3862 }, { "epoch": 2.83003663003663, "grad_norm": 0.80859375, "learning_rate": 9.974292469924198e-05, "loss": 0.1633, "step": 3863 }, { "epoch": 2.830769230769231, "grad_norm": 0.98828125, "learning_rate": 9.974257583374831e-05, "loss": 0.1607, "step": 3864 }, { "epoch": 2.8315018315018317, "grad_norm": 0.8828125, "learning_rate": 9.974222673231732e-05, "loss": 0.1359, "step": 3865 }, { "epoch": 2.8322344322344324, "grad_norm": 1.2421875, "learning_rate": 9.974187739495063e-05, "loss": 0.1784, "step": 3866 }, { "epoch": 2.832967032967033, "grad_norm": 0.71484375, "learning_rate": 9.974152782164995e-05, "loss": 0.1278, "step": 3867 }, { "epoch": 2.833699633699634, "grad_norm": 1.109375, "learning_rate": 9.974117801241693e-05, "loss": 0.177, "step": 3868 }, { "epoch": 2.8344322344322346, "grad_norm": 0.7109375, "learning_rate": 9.974082796725327e-05, "loss": 0.1289, "step": 3869 }, { "epoch": 2.8351648351648353, "grad_norm": 1.484375, "learning_rate": 9.974047768616065e-05, "loss": 0.224, "step": 3870 }, { "epoch": 2.835897435897436, "grad_norm": 1.0859375, "learning_rate": 9.974012716914073e-05, "loss": 0.1923, "step": 3871 }, { "epoch": 2.8366300366300368, "grad_norm": 1.0, "learning_rate": 9.97397764161952e-05, "loss": 0.1614, "step": 3872 }, { "epoch": 2.8373626373626375, "grad_norm": 0.70703125, "learning_rate": 9.973942542732573e-05, "loss": 0.1465, "step": 3873 }, { "epoch": 2.8380952380952382, "grad_norm": 1.15625, "learning_rate": 9.973907420253401e-05, "loss": 0.1657, "step": 3874 }, { "epoch": 2.838827838827839, "grad_norm": 0.98046875, "learning_rate": 9.973872274182173e-05, "loss": 0.1748, "step": 3875 }, { "epoch": 2.8395604395604397, "grad_norm": 1.2109375, "learning_rate": 9.973837104519056e-05, "loss": 0.1863, "step": 3876 }, { "epoch": 2.8402930402930404, "grad_norm": 1.5, "learning_rate": 9.973801911264217e-05, "loss": 0.1744, "step": 3877 }, { "epoch": 2.841025641025641, "grad_norm": 0.93359375, "learning_rate": 9.973766694417828e-05, "loss": 0.1402, "step": 3878 }, { "epoch": 2.841758241758242, "grad_norm": 1.171875, "learning_rate": 9.973731453980057e-05, "loss": 0.1791, "step": 3879 }, { "epoch": 2.8424908424908426, "grad_norm": 1.125, "learning_rate": 9.973696189951072e-05, "loss": 0.1407, "step": 3880 }, { "epoch": 2.8432234432234433, "grad_norm": 1.1171875, "learning_rate": 9.973660902331043e-05, "loss": 0.171, "step": 3881 }, { "epoch": 2.843956043956044, "grad_norm": 0.875, "learning_rate": 9.973625591120138e-05, "loss": 0.1617, "step": 3882 }, { "epoch": 2.844688644688645, "grad_norm": 1.15625, "learning_rate": 9.973590256318527e-05, "loss": 0.1723, "step": 3883 }, { "epoch": 2.8454212454212455, "grad_norm": 1.2578125, "learning_rate": 9.973554897926377e-05, "loss": 0.2134, "step": 3884 }, { "epoch": 2.8461538461538463, "grad_norm": 0.9375, "learning_rate": 9.97351951594386e-05, "loss": 0.1666, "step": 3885 }, { "epoch": 2.846886446886447, "grad_norm": 1.015625, "learning_rate": 9.973484110371144e-05, "loss": 0.1807, "step": 3886 }, { "epoch": 2.8476190476190477, "grad_norm": 0.99609375, "learning_rate": 9.9734486812084e-05, "loss": 0.1491, "step": 3887 }, { "epoch": 2.8483516483516484, "grad_norm": 0.68359375, "learning_rate": 9.973413228455797e-05, "loss": 0.1254, "step": 3888 }, { "epoch": 2.849084249084249, "grad_norm": 1.234375, "learning_rate": 9.973377752113506e-05, "loss": 0.2263, "step": 3889 }, { "epoch": 2.84981684981685, "grad_norm": 1.1015625, "learning_rate": 9.973342252181696e-05, "loss": 0.1633, "step": 3890 }, { "epoch": 2.8505494505494506, "grad_norm": 1.1015625, "learning_rate": 9.973306728660535e-05, "loss": 0.1917, "step": 3891 }, { "epoch": 2.8512820512820514, "grad_norm": 0.66015625, "learning_rate": 9.973271181550197e-05, "loss": 0.1325, "step": 3892 }, { "epoch": 2.852014652014652, "grad_norm": 1.4921875, "learning_rate": 9.973235610850851e-05, "loss": 0.1909, "step": 3893 }, { "epoch": 2.852747252747253, "grad_norm": 1.2734375, "learning_rate": 9.973200016562664e-05, "loss": 0.2106, "step": 3894 }, { "epoch": 2.8534798534798536, "grad_norm": 0.9453125, "learning_rate": 9.973164398685811e-05, "loss": 0.1643, "step": 3895 }, { "epoch": 2.8542124542124543, "grad_norm": 1.0078125, "learning_rate": 9.973128757220463e-05, "loss": 0.1473, "step": 3896 }, { "epoch": 2.854945054945055, "grad_norm": 0.8671875, "learning_rate": 9.973093092166787e-05, "loss": 0.178, "step": 3897 }, { "epoch": 2.8556776556776557, "grad_norm": 1.0390625, "learning_rate": 9.973057403524957e-05, "loss": 0.1584, "step": 3898 }, { "epoch": 2.8564102564102565, "grad_norm": 1.3984375, "learning_rate": 9.973021691295142e-05, "loss": 0.1449, "step": 3899 }, { "epoch": 2.857142857142857, "grad_norm": 1.1953125, "learning_rate": 9.972985955477514e-05, "loss": 0.1378, "step": 3900 }, { "epoch": 2.857875457875458, "grad_norm": 0.79296875, "learning_rate": 9.972950196072245e-05, "loss": 0.1781, "step": 3901 }, { "epoch": 2.8586080586080587, "grad_norm": 1.3359375, "learning_rate": 9.972914413079505e-05, "loss": 0.1598, "step": 3902 }, { "epoch": 2.8593406593406594, "grad_norm": 1.4921875, "learning_rate": 9.972878606499466e-05, "loss": 0.1464, "step": 3903 }, { "epoch": 2.86007326007326, "grad_norm": 0.953125, "learning_rate": 9.972842776332299e-05, "loss": 0.1911, "step": 3904 }, { "epoch": 2.860805860805861, "grad_norm": 1.609375, "learning_rate": 9.972806922578177e-05, "loss": 0.1861, "step": 3905 }, { "epoch": 2.8615384615384616, "grad_norm": 0.94140625, "learning_rate": 9.972771045237272e-05, "loss": 0.1251, "step": 3906 }, { "epoch": 2.8622710622710623, "grad_norm": 2.015625, "learning_rate": 9.972735144309753e-05, "loss": 0.1667, "step": 3907 }, { "epoch": 2.863003663003663, "grad_norm": 1.578125, "learning_rate": 9.972699219795794e-05, "loss": 0.2146, "step": 3908 }, { "epoch": 2.8637362637362638, "grad_norm": 1.125, "learning_rate": 9.972663271695567e-05, "loss": 0.1551, "step": 3909 }, { "epoch": 2.8644688644688645, "grad_norm": 1.2265625, "learning_rate": 9.972627300009246e-05, "loss": 0.1671, "step": 3910 }, { "epoch": 2.8652014652014652, "grad_norm": 1.28125, "learning_rate": 9.972591304737e-05, "loss": 0.1817, "step": 3911 }, { "epoch": 2.865934065934066, "grad_norm": 0.97265625, "learning_rate": 9.972555285879002e-05, "loss": 0.1596, "step": 3912 }, { "epoch": 2.8666666666666667, "grad_norm": 1.3984375, "learning_rate": 9.972519243435428e-05, "loss": 0.1629, "step": 3913 }, { "epoch": 2.8673992673992674, "grad_norm": 1.4921875, "learning_rate": 9.972483177406447e-05, "loss": 0.1912, "step": 3914 }, { "epoch": 2.868131868131868, "grad_norm": 1.4453125, "learning_rate": 9.972447087792234e-05, "loss": 0.2066, "step": 3915 }, { "epoch": 2.868864468864469, "grad_norm": 0.9453125, "learning_rate": 9.972410974592961e-05, "loss": 0.1833, "step": 3916 }, { "epoch": 2.8695970695970696, "grad_norm": 1.078125, "learning_rate": 9.9723748378088e-05, "loss": 0.185, "step": 3917 }, { "epoch": 2.8703296703296703, "grad_norm": 0.734375, "learning_rate": 9.972338677439926e-05, "loss": 0.1417, "step": 3918 }, { "epoch": 2.871062271062271, "grad_norm": 0.91796875, "learning_rate": 9.972302493486512e-05, "loss": 0.1601, "step": 3919 }, { "epoch": 2.871794871794872, "grad_norm": 0.66796875, "learning_rate": 9.97226628594873e-05, "loss": 0.1481, "step": 3920 }, { "epoch": 2.8725274725274725, "grad_norm": 0.94140625, "learning_rate": 9.972230054826754e-05, "loss": 0.1552, "step": 3921 }, { "epoch": 2.8732600732600733, "grad_norm": 0.7265625, "learning_rate": 9.97219380012076e-05, "loss": 0.1874, "step": 3922 }, { "epoch": 2.873992673992674, "grad_norm": 0.67578125, "learning_rate": 9.972157521830919e-05, "loss": 0.1666, "step": 3923 }, { "epoch": 2.8747252747252747, "grad_norm": 0.8828125, "learning_rate": 9.972121219957404e-05, "loss": 0.1574, "step": 3924 }, { "epoch": 2.8754578754578755, "grad_norm": 1.375, "learning_rate": 9.97208489450039e-05, "loss": 0.17, "step": 3925 }, { "epoch": 2.876190476190476, "grad_norm": 1.265625, "learning_rate": 9.972048545460054e-05, "loss": 0.1422, "step": 3926 }, { "epoch": 2.876923076923077, "grad_norm": 1.328125, "learning_rate": 9.972012172836566e-05, "loss": 0.2339, "step": 3927 }, { "epoch": 2.8776556776556776, "grad_norm": 1.421875, "learning_rate": 9.9719757766301e-05, "loss": 0.1721, "step": 3928 }, { "epoch": 2.8783882783882784, "grad_norm": 0.84375, "learning_rate": 9.971939356840836e-05, "loss": 0.1832, "step": 3929 }, { "epoch": 2.879120879120879, "grad_norm": 0.84765625, "learning_rate": 9.971902913468944e-05, "loss": 0.1465, "step": 3930 }, { "epoch": 2.87985347985348, "grad_norm": 0.8203125, "learning_rate": 9.971866446514599e-05, "loss": 0.1771, "step": 3931 }, { "epoch": 2.8805860805860806, "grad_norm": 0.796875, "learning_rate": 9.971829955977974e-05, "loss": 0.1963, "step": 3932 }, { "epoch": 2.8813186813186813, "grad_norm": 1.328125, "learning_rate": 9.971793441859248e-05, "loss": 0.1893, "step": 3933 }, { "epoch": 2.882051282051282, "grad_norm": 0.76953125, "learning_rate": 9.971756904158594e-05, "loss": 0.1618, "step": 3934 }, { "epoch": 2.8827838827838828, "grad_norm": 0.80078125, "learning_rate": 9.971720342876187e-05, "loss": 0.1308, "step": 3935 }, { "epoch": 2.8835164835164835, "grad_norm": 1.1015625, "learning_rate": 9.971683758012202e-05, "loss": 0.2004, "step": 3936 }, { "epoch": 2.884249084249084, "grad_norm": 1.2109375, "learning_rate": 9.971647149566813e-05, "loss": 0.2004, "step": 3937 }, { "epoch": 2.884981684981685, "grad_norm": 0.953125, "learning_rate": 9.971610517540198e-05, "loss": 0.1508, "step": 3938 }, { "epoch": 2.8857142857142857, "grad_norm": 1.1171875, "learning_rate": 9.97157386193253e-05, "loss": 0.1403, "step": 3939 }, { "epoch": 2.8864468864468864, "grad_norm": 1.25, "learning_rate": 9.971537182743988e-05, "loss": 0.17, "step": 3940 }, { "epoch": 2.887179487179487, "grad_norm": 1.09375, "learning_rate": 9.971500479974746e-05, "loss": 0.1436, "step": 3941 }, { "epoch": 2.887912087912088, "grad_norm": 0.91015625, "learning_rate": 9.971463753624977e-05, "loss": 0.1942, "step": 3942 }, { "epoch": 2.8886446886446886, "grad_norm": 1.5234375, "learning_rate": 9.971427003694861e-05, "loss": 0.1834, "step": 3943 }, { "epoch": 2.8893772893772893, "grad_norm": 0.90625, "learning_rate": 9.971390230184575e-05, "loss": 0.1627, "step": 3944 }, { "epoch": 2.89010989010989, "grad_norm": 0.78515625, "learning_rate": 9.971353433094288e-05, "loss": 0.1612, "step": 3945 }, { "epoch": 2.890842490842491, "grad_norm": 1.0390625, "learning_rate": 9.971316612424186e-05, "loss": 0.1756, "step": 3946 }, { "epoch": 2.8915750915750915, "grad_norm": 1.140625, "learning_rate": 9.971279768174436e-05, "loss": 0.1788, "step": 3947 }, { "epoch": 2.8923076923076922, "grad_norm": 1.328125, "learning_rate": 9.971242900345223e-05, "loss": 0.1887, "step": 3948 }, { "epoch": 2.893040293040293, "grad_norm": 1.4453125, "learning_rate": 9.971206008936717e-05, "loss": 0.1754, "step": 3949 }, { "epoch": 2.8937728937728937, "grad_norm": 1.0, "learning_rate": 9.971169093949099e-05, "loss": 0.1706, "step": 3950 }, { "epoch": 2.8945054945054944, "grad_norm": 1.1015625, "learning_rate": 9.971132155382543e-05, "loss": 0.2019, "step": 3951 }, { "epoch": 2.895238095238095, "grad_norm": 0.8828125, "learning_rate": 9.971095193237229e-05, "loss": 0.1656, "step": 3952 }, { "epoch": 2.895970695970696, "grad_norm": 0.6953125, "learning_rate": 9.971058207513332e-05, "loss": 0.1481, "step": 3953 }, { "epoch": 2.8967032967032966, "grad_norm": 1.1953125, "learning_rate": 9.971021198211028e-05, "loss": 0.158, "step": 3954 }, { "epoch": 2.8974358974358974, "grad_norm": 0.88671875, "learning_rate": 9.970984165330498e-05, "loss": 0.1488, "step": 3955 }, { "epoch": 2.898168498168498, "grad_norm": 0.8671875, "learning_rate": 9.970947108871916e-05, "loss": 0.198, "step": 3956 }, { "epoch": 2.898901098901099, "grad_norm": 1.0, "learning_rate": 9.970910028835462e-05, "loss": 0.1571, "step": 3957 }, { "epoch": 2.8996336996336995, "grad_norm": 1.0078125, "learning_rate": 9.970872925221313e-05, "loss": 0.1494, "step": 3958 }, { "epoch": 2.9003663003663003, "grad_norm": 0.8671875, "learning_rate": 9.970835798029645e-05, "loss": 0.1685, "step": 3959 }, { "epoch": 2.901098901098901, "grad_norm": 1.171875, "learning_rate": 9.970798647260639e-05, "loss": 0.1383, "step": 3960 }, { "epoch": 2.9018315018315017, "grad_norm": 1.21875, "learning_rate": 9.970761472914469e-05, "loss": 0.1355, "step": 3961 }, { "epoch": 2.9025641025641025, "grad_norm": 0.859375, "learning_rate": 9.970724274991316e-05, "loss": 0.153, "step": 3962 }, { "epoch": 2.903296703296703, "grad_norm": 1.578125, "learning_rate": 9.970687053491359e-05, "loss": 0.1736, "step": 3963 }, { "epoch": 2.904029304029304, "grad_norm": 1.2578125, "learning_rate": 9.970649808414774e-05, "loss": 0.1783, "step": 3964 }, { "epoch": 2.9047619047619047, "grad_norm": 1.046875, "learning_rate": 9.970612539761742e-05, "loss": 0.1666, "step": 3965 }, { "epoch": 2.9054945054945054, "grad_norm": 1.4609375, "learning_rate": 9.970575247532437e-05, "loss": 0.1901, "step": 3966 }, { "epoch": 2.906227106227106, "grad_norm": 1.125, "learning_rate": 9.970537931727042e-05, "loss": 0.1498, "step": 3967 }, { "epoch": 2.906959706959707, "grad_norm": 1.28125, "learning_rate": 9.970500592345736e-05, "loss": 0.1858, "step": 3968 }, { "epoch": 2.9076923076923076, "grad_norm": 1.75, "learning_rate": 9.970463229388695e-05, "loss": 0.1833, "step": 3969 }, { "epoch": 2.9084249084249083, "grad_norm": 1.015625, "learning_rate": 9.970425842856099e-05, "loss": 0.1571, "step": 3970 }, { "epoch": 2.909157509157509, "grad_norm": 1.0703125, "learning_rate": 9.970388432748128e-05, "loss": 0.1924, "step": 3971 }, { "epoch": 2.9098901098901098, "grad_norm": 1.1875, "learning_rate": 9.970350999064961e-05, "loss": 0.1713, "step": 3972 }, { "epoch": 2.9106227106227105, "grad_norm": 1.1484375, "learning_rate": 9.970313541806777e-05, "loss": 0.1922, "step": 3973 }, { "epoch": 2.9113553113553112, "grad_norm": 1.15625, "learning_rate": 9.970276060973755e-05, "loss": 0.1615, "step": 3974 }, { "epoch": 2.912087912087912, "grad_norm": 1.0390625, "learning_rate": 9.970238556566077e-05, "loss": 0.1651, "step": 3975 }, { "epoch": 2.9128205128205127, "grad_norm": 1.5390625, "learning_rate": 9.970201028583919e-05, "loss": 0.2032, "step": 3976 }, { "epoch": 2.9135531135531134, "grad_norm": 0.7578125, "learning_rate": 9.970163477027463e-05, "loss": 0.1525, "step": 3977 }, { "epoch": 2.914285714285714, "grad_norm": 0.890625, "learning_rate": 9.970125901896888e-05, "loss": 0.1605, "step": 3978 }, { "epoch": 2.915018315018315, "grad_norm": 1.15625, "learning_rate": 9.970088303192377e-05, "loss": 0.1423, "step": 3979 }, { "epoch": 2.9157509157509156, "grad_norm": 1.0625, "learning_rate": 9.970050680914105e-05, "loss": 0.1161, "step": 3980 }, { "epoch": 2.9164835164835163, "grad_norm": 1.15625, "learning_rate": 9.970013035062257e-05, "loss": 0.1921, "step": 3981 }, { "epoch": 2.917216117216117, "grad_norm": 1.2734375, "learning_rate": 9.969975365637011e-05, "loss": 0.195, "step": 3982 }, { "epoch": 2.917948717948718, "grad_norm": 1.5, "learning_rate": 9.969937672638547e-05, "loss": 0.1633, "step": 3983 }, { "epoch": 2.9186813186813185, "grad_norm": 0.98828125, "learning_rate": 9.969899956067047e-05, "loss": 0.1915, "step": 3984 }, { "epoch": 2.9194139194139193, "grad_norm": 0.92578125, "learning_rate": 9.969862215922692e-05, "loss": 0.1562, "step": 3985 }, { "epoch": 2.92014652014652, "grad_norm": 1.203125, "learning_rate": 9.969824452205661e-05, "loss": 0.1996, "step": 3986 }, { "epoch": 2.9208791208791207, "grad_norm": 1.9140625, "learning_rate": 9.969786664916136e-05, "loss": 0.1565, "step": 3987 }, { "epoch": 2.9216117216117214, "grad_norm": 1.5078125, "learning_rate": 9.9697488540543e-05, "loss": 0.1868, "step": 3988 }, { "epoch": 2.9223443223443226, "grad_norm": 0.7265625, "learning_rate": 9.96971101962033e-05, "loss": 0.1706, "step": 3989 }, { "epoch": 2.9230769230769234, "grad_norm": 0.640625, "learning_rate": 9.96967316161441e-05, "loss": 0.1082, "step": 3990 }, { "epoch": 2.923809523809524, "grad_norm": 1.2265625, "learning_rate": 9.96963528003672e-05, "loss": 0.1946, "step": 3991 }, { "epoch": 2.924542124542125, "grad_norm": 1.5078125, "learning_rate": 9.969597374887442e-05, "loss": 0.2166, "step": 3992 }, { "epoch": 2.9252747252747255, "grad_norm": 1.15625, "learning_rate": 9.969559446166761e-05, "loss": 0.1673, "step": 3993 }, { "epoch": 2.9260073260073263, "grad_norm": 1.1640625, "learning_rate": 9.969521493874852e-05, "loss": 0.196, "step": 3994 }, { "epoch": 2.926739926739927, "grad_norm": 1.2578125, "learning_rate": 9.969483518011904e-05, "loss": 0.1698, "step": 3995 }, { "epoch": 2.9274725274725277, "grad_norm": 1.578125, "learning_rate": 9.969445518578094e-05, "loss": 0.2282, "step": 3996 }, { "epoch": 2.9282051282051285, "grad_norm": 2.578125, "learning_rate": 9.969407495573605e-05, "loss": 0.1403, "step": 3997 }, { "epoch": 2.928937728937729, "grad_norm": 3.5, "learning_rate": 9.969369448998621e-05, "loss": 0.1796, "step": 3998 }, { "epoch": 2.92967032967033, "grad_norm": 2.453125, "learning_rate": 9.969331378853321e-05, "loss": 0.1753, "step": 3999 }, { "epoch": 2.9304029304029307, "grad_norm": 0.7421875, "learning_rate": 9.969293285137891e-05, "loss": 0.1704, "step": 4000 }, { "epoch": 2.9311355311355314, "grad_norm": 1.1015625, "learning_rate": 9.969255167852512e-05, "loss": 0.1598, "step": 4001 }, { "epoch": 2.931868131868132, "grad_norm": 1.21875, "learning_rate": 9.969217026997368e-05, "loss": 0.155, "step": 4002 }, { "epoch": 2.932600732600733, "grad_norm": 0.82421875, "learning_rate": 9.969178862572639e-05, "loss": 0.1926, "step": 4003 }, { "epoch": 2.9333333333333336, "grad_norm": 1.2578125, "learning_rate": 9.969140674578509e-05, "loss": 0.1361, "step": 4004 }, { "epoch": 2.9340659340659343, "grad_norm": 0.953125, "learning_rate": 9.96910246301516e-05, "loss": 0.1388, "step": 4005 }, { "epoch": 2.934798534798535, "grad_norm": 0.90625, "learning_rate": 9.969064227882777e-05, "loss": 0.1845, "step": 4006 }, { "epoch": 2.9355311355311358, "grad_norm": 1.609375, "learning_rate": 9.969025969181542e-05, "loss": 0.183, "step": 4007 }, { "epoch": 2.9362637362637365, "grad_norm": 1.15625, "learning_rate": 9.968987686911641e-05, "loss": 0.1556, "step": 4008 }, { "epoch": 2.936996336996337, "grad_norm": 1.0234375, "learning_rate": 9.968949381073254e-05, "loss": 0.1686, "step": 4009 }, { "epoch": 2.937728937728938, "grad_norm": 0.984375, "learning_rate": 9.968911051666566e-05, "loss": 0.1704, "step": 4010 }, { "epoch": 2.9384615384615387, "grad_norm": 1.109375, "learning_rate": 9.968872698691759e-05, "loss": 0.1657, "step": 4011 }, { "epoch": 2.9391941391941394, "grad_norm": 1.0078125, "learning_rate": 9.96883432214902e-05, "loss": 0.1575, "step": 4012 }, { "epoch": 2.93992673992674, "grad_norm": 2.0625, "learning_rate": 9.96879592203853e-05, "loss": 0.1654, "step": 4013 }, { "epoch": 2.940659340659341, "grad_norm": 1.078125, "learning_rate": 9.968757498360472e-05, "loss": 0.1523, "step": 4014 }, { "epoch": 2.9413919413919416, "grad_norm": 1.046875, "learning_rate": 9.968719051115035e-05, "loss": 0.1589, "step": 4015 }, { "epoch": 2.9421245421245423, "grad_norm": 1.828125, "learning_rate": 9.9686805803024e-05, "loss": 0.1754, "step": 4016 }, { "epoch": 2.942857142857143, "grad_norm": 1.5, "learning_rate": 9.968642085922751e-05, "loss": 0.1709, "step": 4017 }, { "epoch": 2.943589743589744, "grad_norm": 1.6171875, "learning_rate": 9.968603567976273e-05, "loss": 0.1798, "step": 4018 }, { "epoch": 2.9443223443223445, "grad_norm": 1.2109375, "learning_rate": 9.968565026463151e-05, "loss": 0.1617, "step": 4019 }, { "epoch": 2.9450549450549453, "grad_norm": 1.0390625, "learning_rate": 9.968526461383569e-05, "loss": 0.1704, "step": 4020 }, { "epoch": 2.945787545787546, "grad_norm": 1.4609375, "learning_rate": 9.968487872737713e-05, "loss": 0.1754, "step": 4021 }, { "epoch": 2.9465201465201467, "grad_norm": 0.8828125, "learning_rate": 9.968449260525767e-05, "loss": 0.176, "step": 4022 }, { "epoch": 2.9472527472527474, "grad_norm": 1.78125, "learning_rate": 9.968410624747913e-05, "loss": 0.233, "step": 4023 }, { "epoch": 2.947985347985348, "grad_norm": 1.015625, "learning_rate": 9.968371965404342e-05, "loss": 0.17, "step": 4024 }, { "epoch": 2.948717948717949, "grad_norm": 0.95703125, "learning_rate": 9.968333282495236e-05, "loss": 0.1667, "step": 4025 }, { "epoch": 2.9494505494505496, "grad_norm": 1.046875, "learning_rate": 9.96829457602078e-05, "loss": 0.175, "step": 4026 }, { "epoch": 2.9501831501831504, "grad_norm": 1.5625, "learning_rate": 9.96825584598116e-05, "loss": 0.1908, "step": 4027 }, { "epoch": 2.950915750915751, "grad_norm": 2.015625, "learning_rate": 9.968217092376563e-05, "loss": 0.2122, "step": 4028 }, { "epoch": 2.951648351648352, "grad_norm": 1.3515625, "learning_rate": 9.968178315207171e-05, "loss": 0.1967, "step": 4029 }, { "epoch": 2.9523809523809526, "grad_norm": 1.109375, "learning_rate": 9.968139514473174e-05, "loss": 0.1829, "step": 4030 }, { "epoch": 2.9531135531135533, "grad_norm": 0.8515625, "learning_rate": 9.968100690174755e-05, "loss": 0.1608, "step": 4031 }, { "epoch": 2.953846153846154, "grad_norm": 1.1328125, "learning_rate": 9.968061842312103e-05, "loss": 0.1737, "step": 4032 }, { "epoch": 2.9545787545787547, "grad_norm": 1.265625, "learning_rate": 9.9680229708854e-05, "loss": 0.1889, "step": 4033 }, { "epoch": 2.9553113553113555, "grad_norm": 0.8828125, "learning_rate": 9.967984075894836e-05, "loss": 0.158, "step": 4034 }, { "epoch": 2.956043956043956, "grad_norm": 0.9765625, "learning_rate": 9.967945157340595e-05, "loss": 0.1729, "step": 4035 }, { "epoch": 2.956776556776557, "grad_norm": 1.109375, "learning_rate": 9.967906215222864e-05, "loss": 0.1809, "step": 4036 }, { "epoch": 2.9575091575091577, "grad_norm": 1.15625, "learning_rate": 9.967867249541831e-05, "loss": 0.202, "step": 4037 }, { "epoch": 2.9582417582417584, "grad_norm": 1.0390625, "learning_rate": 9.96782826029768e-05, "loss": 0.2026, "step": 4038 }, { "epoch": 2.958974358974359, "grad_norm": 1.671875, "learning_rate": 9.9677892474906e-05, "loss": 0.1745, "step": 4039 }, { "epoch": 2.95970695970696, "grad_norm": 0.96875, "learning_rate": 9.967750211120778e-05, "loss": 0.1772, "step": 4040 }, { "epoch": 2.9604395604395606, "grad_norm": 1.140625, "learning_rate": 9.9677111511884e-05, "loss": 0.1728, "step": 4041 }, { "epoch": 2.9611721611721613, "grad_norm": 0.98046875, "learning_rate": 9.967672067693656e-05, "loss": 0.1935, "step": 4042 }, { "epoch": 2.961904761904762, "grad_norm": 0.921875, "learning_rate": 9.967632960636726e-05, "loss": 0.1401, "step": 4043 }, { "epoch": 2.9626373626373628, "grad_norm": 1.0078125, "learning_rate": 9.967593830017806e-05, "loss": 0.1543, "step": 4044 }, { "epoch": 2.9633699633699635, "grad_norm": 0.73046875, "learning_rate": 9.967554675837078e-05, "loss": 0.1688, "step": 4045 }, { "epoch": 2.9641025641025642, "grad_norm": 1.0703125, "learning_rate": 9.967515498094729e-05, "loss": 0.1701, "step": 4046 }, { "epoch": 2.964835164835165, "grad_norm": 0.83203125, "learning_rate": 9.967476296790954e-05, "loss": 0.142, "step": 4047 }, { "epoch": 2.9655677655677657, "grad_norm": 0.75390625, "learning_rate": 9.967437071925931e-05, "loss": 0.1955, "step": 4048 }, { "epoch": 2.9663003663003664, "grad_norm": 1.2109375, "learning_rate": 9.967397823499856e-05, "loss": 0.1751, "step": 4049 }, { "epoch": 2.967032967032967, "grad_norm": 1.3828125, "learning_rate": 9.96735855151291e-05, "loss": 0.1429, "step": 4050 }, { "epoch": 2.967765567765568, "grad_norm": 0.87109375, "learning_rate": 9.96731925596529e-05, "loss": 0.2126, "step": 4051 }, { "epoch": 2.9684981684981686, "grad_norm": 0.83984375, "learning_rate": 9.967279936857176e-05, "loss": 0.1689, "step": 4052 }, { "epoch": 2.9692307692307693, "grad_norm": 0.78515625, "learning_rate": 9.96724059418876e-05, "loss": 0.1409, "step": 4053 }, { "epoch": 2.96996336996337, "grad_norm": 0.76171875, "learning_rate": 9.967201227960229e-05, "loss": 0.1538, "step": 4054 }, { "epoch": 2.970695970695971, "grad_norm": 0.66015625, "learning_rate": 9.967161838171774e-05, "loss": 0.1297, "step": 4055 }, { "epoch": 2.9714285714285715, "grad_norm": 1.203125, "learning_rate": 9.967122424823583e-05, "loss": 0.1727, "step": 4056 }, { "epoch": 2.9721611721611723, "grad_norm": 1.0859375, "learning_rate": 9.967082987915842e-05, "loss": 0.1677, "step": 4057 }, { "epoch": 2.972893772893773, "grad_norm": 0.8828125, "learning_rate": 9.967043527448745e-05, "loss": 0.1649, "step": 4058 }, { "epoch": 2.9736263736263737, "grad_norm": 1.0703125, "learning_rate": 9.967004043422476e-05, "loss": 0.1759, "step": 4059 }, { "epoch": 2.9743589743589745, "grad_norm": 0.87890625, "learning_rate": 9.966964535837229e-05, "loss": 0.1743, "step": 4060 }, { "epoch": 2.975091575091575, "grad_norm": 1.21875, "learning_rate": 9.966925004693189e-05, "loss": 0.1787, "step": 4061 }, { "epoch": 2.975824175824176, "grad_norm": 1.3515625, "learning_rate": 9.966885449990547e-05, "loss": 0.1755, "step": 4062 }, { "epoch": 2.9765567765567766, "grad_norm": 0.98046875, "learning_rate": 9.966845871729494e-05, "loss": 0.1932, "step": 4063 }, { "epoch": 2.9772893772893774, "grad_norm": 1.46875, "learning_rate": 9.966806269910219e-05, "loss": 0.1716, "step": 4064 }, { "epoch": 2.978021978021978, "grad_norm": 1.453125, "learning_rate": 9.966766644532908e-05, "loss": 0.2164, "step": 4065 }, { "epoch": 2.978754578754579, "grad_norm": 1.34375, "learning_rate": 9.966726995597755e-05, "loss": 0.1471, "step": 4066 }, { "epoch": 2.9794871794871796, "grad_norm": 1.796875, "learning_rate": 9.966687323104951e-05, "loss": 0.1974, "step": 4067 }, { "epoch": 2.9802197802197803, "grad_norm": 1.1796875, "learning_rate": 9.966647627054682e-05, "loss": 0.1712, "step": 4068 }, { "epoch": 2.980952380952381, "grad_norm": 1.78125, "learning_rate": 9.966607907447141e-05, "loss": 0.1638, "step": 4069 }, { "epoch": 2.9816849816849818, "grad_norm": 1.4921875, "learning_rate": 9.966568164282517e-05, "loss": 0.1584, "step": 4070 }, { "epoch": 2.9824175824175825, "grad_norm": 1.09375, "learning_rate": 9.966528397561001e-05, "loss": 0.1833, "step": 4071 }, { "epoch": 2.983150183150183, "grad_norm": 1.25, "learning_rate": 9.966488607282784e-05, "loss": 0.1617, "step": 4072 }, { "epoch": 2.983882783882784, "grad_norm": 0.94140625, "learning_rate": 9.966448793448055e-05, "loss": 0.1416, "step": 4073 }, { "epoch": 2.9846153846153847, "grad_norm": 1.09375, "learning_rate": 9.966408956057006e-05, "loss": 0.1526, "step": 4074 }, { "epoch": 2.9853479853479854, "grad_norm": 1.1328125, "learning_rate": 9.966369095109828e-05, "loss": 0.1773, "step": 4075 }, { "epoch": 2.986080586080586, "grad_norm": 0.984375, "learning_rate": 9.966329210606714e-05, "loss": 0.1562, "step": 4076 }, { "epoch": 2.986813186813187, "grad_norm": 1.0703125, "learning_rate": 9.96628930254785e-05, "loss": 0.162, "step": 4077 }, { "epoch": 2.9875457875457876, "grad_norm": 0.78125, "learning_rate": 9.966249370933432e-05, "loss": 0.1888, "step": 4078 }, { "epoch": 2.9882783882783883, "grad_norm": 0.66796875, "learning_rate": 9.966209415763648e-05, "loss": 0.1384, "step": 4079 }, { "epoch": 2.989010989010989, "grad_norm": 1.203125, "learning_rate": 9.96616943703869e-05, "loss": 0.1685, "step": 4080 }, { "epoch": 2.98974358974359, "grad_norm": 1.109375, "learning_rate": 9.96612943475875e-05, "loss": 0.2229, "step": 4081 }, { "epoch": 2.9904761904761905, "grad_norm": 1.1796875, "learning_rate": 9.966089408924023e-05, "loss": 0.175, "step": 4082 }, { "epoch": 2.9912087912087912, "grad_norm": 1.140625, "learning_rate": 9.966049359534696e-05, "loss": 0.1779, "step": 4083 }, { "epoch": 2.991941391941392, "grad_norm": 0.8046875, "learning_rate": 9.966009286590963e-05, "loss": 0.1457, "step": 4084 }, { "epoch": 2.9926739926739927, "grad_norm": 0.6875, "learning_rate": 9.965969190093016e-05, "loss": 0.1553, "step": 4085 }, { "epoch": 2.9934065934065934, "grad_norm": 0.8125, "learning_rate": 9.965929070041046e-05, "loss": 0.137, "step": 4086 }, { "epoch": 2.994139194139194, "grad_norm": 0.5390625, "learning_rate": 9.965888926435246e-05, "loss": 0.1654, "step": 4087 }, { "epoch": 2.994871794871795, "grad_norm": 0.88671875, "learning_rate": 9.96584875927581e-05, "loss": 0.14, "step": 4088 }, { "epoch": 2.9956043956043956, "grad_norm": 0.87890625, "learning_rate": 9.965808568562927e-05, "loss": 0.1701, "step": 4089 }, { "epoch": 2.9963369963369964, "grad_norm": 0.9375, "learning_rate": 9.965768354296791e-05, "loss": 0.1502, "step": 4090 }, { "epoch": 2.997069597069597, "grad_norm": 0.84765625, "learning_rate": 9.965728116477598e-05, "loss": 0.204, "step": 4091 }, { "epoch": 2.997802197802198, "grad_norm": 1.1171875, "learning_rate": 9.965687855105535e-05, "loss": 0.1709, "step": 4092 }, { "epoch": 2.9985347985347985, "grad_norm": 1.1015625, "learning_rate": 9.965647570180798e-05, "loss": 0.1928, "step": 4093 }, { "epoch": 2.9992673992673993, "grad_norm": 1.546875, "learning_rate": 9.965607261703582e-05, "loss": 0.1761, "step": 4094 }, { "epoch": 3.0, "grad_norm": 0.71875, "learning_rate": 9.965566929674076e-05, "loss": 0.1412, "step": 4095 }, { "epoch": 3.0007326007326007, "grad_norm": 1.5234375, "learning_rate": 9.965526574092476e-05, "loss": 0.1635, "step": 4096 }, { "epoch": 3.0014652014652015, "grad_norm": 1.15625, "learning_rate": 9.965486194958974e-05, "loss": 0.1707, "step": 4097 }, { "epoch": 3.002197802197802, "grad_norm": 0.921875, "learning_rate": 9.965445792273764e-05, "loss": 0.187, "step": 4098 }, { "epoch": 3.002930402930403, "grad_norm": 1.421875, "learning_rate": 9.96540536603704e-05, "loss": 0.1915, "step": 4099 }, { "epoch": 3.0036630036630036, "grad_norm": 1.015625, "learning_rate": 9.965364916248993e-05, "loss": 0.1962, "step": 4100 }, { "epoch": 3.0043956043956044, "grad_norm": 1.0625, "learning_rate": 9.965324442909822e-05, "loss": 0.1866, "step": 4101 }, { "epoch": 3.005128205128205, "grad_norm": 1.1015625, "learning_rate": 9.965283946019718e-05, "loss": 0.1449, "step": 4102 }, { "epoch": 3.005860805860806, "grad_norm": 1.4609375, "learning_rate": 9.965243425578873e-05, "loss": 0.1903, "step": 4103 }, { "epoch": 3.0065934065934066, "grad_norm": 0.90234375, "learning_rate": 9.965202881587484e-05, "loss": 0.1606, "step": 4104 }, { "epoch": 3.0073260073260073, "grad_norm": 1.2890625, "learning_rate": 9.965162314045745e-05, "loss": 0.1901, "step": 4105 }, { "epoch": 3.008058608058608, "grad_norm": 1.21875, "learning_rate": 9.96512172295385e-05, "loss": 0.1913, "step": 4106 }, { "epoch": 3.0087912087912088, "grad_norm": 1.0546875, "learning_rate": 9.965081108311992e-05, "loss": 0.1566, "step": 4107 }, { "epoch": 3.0095238095238095, "grad_norm": 1.1015625, "learning_rate": 9.965040470120368e-05, "loss": 0.1846, "step": 4108 }, { "epoch": 3.01025641025641, "grad_norm": 1.234375, "learning_rate": 9.964999808379172e-05, "loss": 0.1491, "step": 4109 }, { "epoch": 3.010989010989011, "grad_norm": 1.2109375, "learning_rate": 9.964959123088598e-05, "loss": 0.1651, "step": 4110 }, { "epoch": 3.0117216117216117, "grad_norm": 0.8828125, "learning_rate": 9.964918414248839e-05, "loss": 0.183, "step": 4111 }, { "epoch": 3.0124542124542124, "grad_norm": 0.94921875, "learning_rate": 9.964877681860096e-05, "loss": 0.1745, "step": 4112 }, { "epoch": 3.013186813186813, "grad_norm": 1.0625, "learning_rate": 9.964836925922558e-05, "loss": 0.1499, "step": 4113 }, { "epoch": 3.013919413919414, "grad_norm": 1.0703125, "learning_rate": 9.964796146436425e-05, "loss": 0.1626, "step": 4114 }, { "epoch": 3.0146520146520146, "grad_norm": 1.609375, "learning_rate": 9.96475534340189e-05, "loss": 0.1757, "step": 4115 }, { "epoch": 3.0153846153846153, "grad_norm": 1.3359375, "learning_rate": 9.964714516819146e-05, "loss": 0.1611, "step": 4116 }, { "epoch": 3.016117216117216, "grad_norm": 0.78125, "learning_rate": 9.964673666688395e-05, "loss": 0.1568, "step": 4117 }, { "epoch": 3.016849816849817, "grad_norm": 0.76171875, "learning_rate": 9.964632793009827e-05, "loss": 0.1542, "step": 4118 }, { "epoch": 3.0175824175824175, "grad_norm": 0.7265625, "learning_rate": 9.964591895783641e-05, "loss": 0.1601, "step": 4119 }, { "epoch": 3.0183150183150182, "grad_norm": 1.4140625, "learning_rate": 9.96455097501003e-05, "loss": 0.206, "step": 4120 }, { "epoch": 3.019047619047619, "grad_norm": 1.0625, "learning_rate": 9.964510030689193e-05, "loss": 0.1759, "step": 4121 }, { "epoch": 3.0197802197802197, "grad_norm": 0.9140625, "learning_rate": 9.964469062821325e-05, "loss": 0.1845, "step": 4122 }, { "epoch": 3.0205128205128204, "grad_norm": 1.2578125, "learning_rate": 9.964428071406623e-05, "loss": 0.206, "step": 4123 }, { "epoch": 3.021245421245421, "grad_norm": 0.80078125, "learning_rate": 9.964387056445282e-05, "loss": 0.1488, "step": 4124 }, { "epoch": 3.021978021978022, "grad_norm": 1.9609375, "learning_rate": 9.9643460179375e-05, "loss": 0.1708, "step": 4125 }, { "epoch": 3.0227106227106226, "grad_norm": 1.375, "learning_rate": 9.964304955883472e-05, "loss": 0.1808, "step": 4126 }, { "epoch": 3.0234432234432234, "grad_norm": 0.98046875, "learning_rate": 9.964263870283397e-05, "loss": 0.1563, "step": 4127 }, { "epoch": 3.024175824175824, "grad_norm": 1.21875, "learning_rate": 9.96422276113747e-05, "loss": 0.1617, "step": 4128 }, { "epoch": 3.024908424908425, "grad_norm": 1.1953125, "learning_rate": 9.96418162844589e-05, "loss": 0.183, "step": 4129 }, { "epoch": 3.0256410256410255, "grad_norm": 1.09375, "learning_rate": 9.964140472208849e-05, "loss": 0.2047, "step": 4130 }, { "epoch": 3.0263736263736263, "grad_norm": 1.515625, "learning_rate": 9.964099292426551e-05, "loss": 0.1567, "step": 4131 }, { "epoch": 3.027106227106227, "grad_norm": 1.203125, "learning_rate": 9.964058089099189e-05, "loss": 0.1638, "step": 4132 }, { "epoch": 3.0278388278388277, "grad_norm": 0.96484375, "learning_rate": 9.964016862226962e-05, "loss": 0.1748, "step": 4133 }, { "epoch": 3.0285714285714285, "grad_norm": 0.9296875, "learning_rate": 9.963975611810065e-05, "loss": 0.1411, "step": 4134 }, { "epoch": 3.029304029304029, "grad_norm": 1.96875, "learning_rate": 9.9639343378487e-05, "loss": 0.2078, "step": 4135 }, { "epoch": 3.03003663003663, "grad_norm": 1.9375, "learning_rate": 9.963893040343062e-05, "loss": 0.1799, "step": 4136 }, { "epoch": 3.0307692307692307, "grad_norm": 1.390625, "learning_rate": 9.963851719293348e-05, "loss": 0.1772, "step": 4137 }, { "epoch": 3.0315018315018314, "grad_norm": 1.9375, "learning_rate": 9.963810374699758e-05, "loss": 0.16, "step": 4138 }, { "epoch": 3.032234432234432, "grad_norm": 0.75390625, "learning_rate": 9.963769006562492e-05, "loss": 0.1896, "step": 4139 }, { "epoch": 3.032967032967033, "grad_norm": 2.390625, "learning_rate": 9.963727614881742e-05, "loss": 0.2186, "step": 4140 }, { "epoch": 3.0336996336996336, "grad_norm": 2.421875, "learning_rate": 9.96368619965771e-05, "loss": 0.1966, "step": 4141 }, { "epoch": 3.0344322344322343, "grad_norm": 3.09375, "learning_rate": 9.963644760890596e-05, "loss": 0.1923, "step": 4142 }, { "epoch": 3.035164835164835, "grad_norm": 1.9140625, "learning_rate": 9.963603298580596e-05, "loss": 0.2029, "step": 4143 }, { "epoch": 3.0358974358974358, "grad_norm": 0.92578125, "learning_rate": 9.96356181272791e-05, "loss": 0.1776, "step": 4144 }, { "epoch": 3.0366300366300365, "grad_norm": 1.546875, "learning_rate": 9.963520303332736e-05, "loss": 0.1607, "step": 4145 }, { "epoch": 3.0373626373626372, "grad_norm": 2.171875, "learning_rate": 9.963478770395275e-05, "loss": 0.1553, "step": 4146 }, { "epoch": 3.038095238095238, "grad_norm": 1.2578125, "learning_rate": 9.963437213915723e-05, "loss": 0.1161, "step": 4147 }, { "epoch": 3.0388278388278387, "grad_norm": 0.94921875, "learning_rate": 9.96339563389428e-05, "loss": 0.1409, "step": 4148 }, { "epoch": 3.0395604395604394, "grad_norm": 0.6953125, "learning_rate": 9.963354030331145e-05, "loss": 0.1407, "step": 4149 }, { "epoch": 3.04029304029304, "grad_norm": 1.890625, "learning_rate": 9.963312403226518e-05, "loss": 0.1642, "step": 4150 }, { "epoch": 3.041025641025641, "grad_norm": 1.75, "learning_rate": 9.9632707525806e-05, "loss": 0.1808, "step": 4151 }, { "epoch": 3.0417582417582416, "grad_norm": 1.171875, "learning_rate": 9.963229078393588e-05, "loss": 0.1701, "step": 4152 }, { "epoch": 3.0424908424908423, "grad_norm": 0.953125, "learning_rate": 9.963187380665681e-05, "loss": 0.1687, "step": 4153 }, { "epoch": 3.043223443223443, "grad_norm": 0.578125, "learning_rate": 9.963145659397083e-05, "loss": 0.1532, "step": 4154 }, { "epoch": 3.043956043956044, "grad_norm": 0.95703125, "learning_rate": 9.96310391458799e-05, "loss": 0.15, "step": 4155 }, { "epoch": 3.0446886446886445, "grad_norm": 1.09375, "learning_rate": 9.963062146238602e-05, "loss": 0.1502, "step": 4156 }, { "epoch": 3.0454212454212453, "grad_norm": 1.1015625, "learning_rate": 9.963020354349123e-05, "loss": 0.1652, "step": 4157 }, { "epoch": 3.046153846153846, "grad_norm": 1.1484375, "learning_rate": 9.962978538919749e-05, "loss": 0.1588, "step": 4158 }, { "epoch": 3.0468864468864467, "grad_norm": 1.15625, "learning_rate": 9.962936699950683e-05, "loss": 0.1739, "step": 4159 }, { "epoch": 3.0476190476190474, "grad_norm": 1.5859375, "learning_rate": 9.962894837442123e-05, "loss": 0.1825, "step": 4160 }, { "epoch": 3.048351648351648, "grad_norm": 2.109375, "learning_rate": 9.962852951394273e-05, "loss": 0.2131, "step": 4161 }, { "epoch": 3.049084249084249, "grad_norm": 2.3125, "learning_rate": 9.96281104180733e-05, "loss": 0.1649, "step": 4162 }, { "epoch": 3.0498168498168496, "grad_norm": 1.03125, "learning_rate": 9.962769108681497e-05, "loss": 0.1887, "step": 4163 }, { "epoch": 3.0505494505494504, "grad_norm": 1.0078125, "learning_rate": 9.962727152016974e-05, "loss": 0.1709, "step": 4164 }, { "epoch": 3.051282051282051, "grad_norm": 0.953125, "learning_rate": 9.962685171813963e-05, "loss": 0.141, "step": 4165 }, { "epoch": 3.052014652014652, "grad_norm": 1.890625, "learning_rate": 9.962643168072664e-05, "loss": 0.184, "step": 4166 }, { "epoch": 3.0527472527472526, "grad_norm": 1.6015625, "learning_rate": 9.962601140793281e-05, "loss": 0.1831, "step": 4167 }, { "epoch": 3.0534798534798533, "grad_norm": 0.8203125, "learning_rate": 9.962559089976011e-05, "loss": 0.1875, "step": 4168 }, { "epoch": 3.054212454212454, "grad_norm": 0.68359375, "learning_rate": 9.96251701562106e-05, "loss": 0.1311, "step": 4169 }, { "epoch": 3.0549450549450547, "grad_norm": 0.96484375, "learning_rate": 9.962474917728624e-05, "loss": 0.1692, "step": 4170 }, { "epoch": 3.0556776556776555, "grad_norm": 1.296875, "learning_rate": 9.962432796298909e-05, "loss": 0.1734, "step": 4171 }, { "epoch": 3.056410256410256, "grad_norm": 1.4921875, "learning_rate": 9.962390651332117e-05, "loss": 0.1372, "step": 4172 }, { "epoch": 3.057142857142857, "grad_norm": 0.703125, "learning_rate": 9.96234848282845e-05, "loss": 0.1419, "step": 4173 }, { "epoch": 3.0578754578754577, "grad_norm": 1.03125, "learning_rate": 9.962306290788106e-05, "loss": 0.1669, "step": 4174 }, { "epoch": 3.0586080586080584, "grad_norm": 0.9765625, "learning_rate": 9.962264075211291e-05, "loss": 0.16, "step": 4175 }, { "epoch": 3.059340659340659, "grad_norm": 1.203125, "learning_rate": 9.962221836098206e-05, "loss": 0.1766, "step": 4176 }, { "epoch": 3.06007326007326, "grad_norm": 0.8203125, "learning_rate": 9.962179573449055e-05, "loss": 0.1546, "step": 4177 }, { "epoch": 3.0608058608058606, "grad_norm": 1.0859375, "learning_rate": 9.962137287264038e-05, "loss": 0.1269, "step": 4178 }, { "epoch": 3.0615384615384613, "grad_norm": 0.9296875, "learning_rate": 9.962094977543358e-05, "loss": 0.2031, "step": 4179 }, { "epoch": 3.062271062271062, "grad_norm": 1.0625, "learning_rate": 9.96205264428722e-05, "loss": 0.1785, "step": 4180 }, { "epoch": 3.063003663003663, "grad_norm": 0.94921875, "learning_rate": 9.962010287495824e-05, "loss": 0.1593, "step": 4181 }, { "epoch": 3.0637362637362635, "grad_norm": 0.9140625, "learning_rate": 9.961967907169376e-05, "loss": 0.1914, "step": 4182 }, { "epoch": 3.0644688644688647, "grad_norm": 1.34375, "learning_rate": 9.961925503308075e-05, "loss": 0.1631, "step": 4183 }, { "epoch": 3.0652014652014654, "grad_norm": 0.76171875, "learning_rate": 9.961883075912127e-05, "loss": 0.1581, "step": 4184 }, { "epoch": 3.065934065934066, "grad_norm": 0.96484375, "learning_rate": 9.961840624981734e-05, "loss": 0.1622, "step": 4185 }, { "epoch": 3.066666666666667, "grad_norm": 1.1171875, "learning_rate": 9.961798150517101e-05, "loss": 0.1573, "step": 4186 }, { "epoch": 3.0673992673992676, "grad_norm": 0.890625, "learning_rate": 9.961755652518431e-05, "loss": 0.2084, "step": 4187 }, { "epoch": 3.0681318681318683, "grad_norm": 0.6484375, "learning_rate": 9.961713130985927e-05, "loss": 0.125, "step": 4188 }, { "epoch": 3.068864468864469, "grad_norm": 1.1484375, "learning_rate": 9.961670585919793e-05, "loss": 0.1426, "step": 4189 }, { "epoch": 3.06959706959707, "grad_norm": 0.72265625, "learning_rate": 9.961628017320234e-05, "loss": 0.1392, "step": 4190 }, { "epoch": 3.0703296703296705, "grad_norm": 1.3125, "learning_rate": 9.961585425187451e-05, "loss": 0.1633, "step": 4191 }, { "epoch": 3.0710622710622713, "grad_norm": 0.80078125, "learning_rate": 9.96154280952165e-05, "loss": 0.1326, "step": 4192 }, { "epoch": 3.071794871794872, "grad_norm": 1.0390625, "learning_rate": 9.961500170323037e-05, "loss": 0.1425, "step": 4193 }, { "epoch": 3.0725274725274727, "grad_norm": 0.91796875, "learning_rate": 9.961457507591812e-05, "loss": 0.1594, "step": 4194 }, { "epoch": 3.0732600732600734, "grad_norm": 1.171875, "learning_rate": 9.961414821328182e-05, "loss": 0.1569, "step": 4195 }, { "epoch": 3.073992673992674, "grad_norm": 0.98046875, "learning_rate": 9.961372111532354e-05, "loss": 0.1521, "step": 4196 }, { "epoch": 3.074725274725275, "grad_norm": 0.62890625, "learning_rate": 9.961329378204526e-05, "loss": 0.1366, "step": 4197 }, { "epoch": 3.0754578754578756, "grad_norm": 0.75, "learning_rate": 9.961286621344908e-05, "loss": 0.1718, "step": 4198 }, { "epoch": 3.0761904761904764, "grad_norm": 1.171875, "learning_rate": 9.961243840953705e-05, "loss": 0.1805, "step": 4199 }, { "epoch": 3.076923076923077, "grad_norm": 0.89453125, "learning_rate": 9.961201037031119e-05, "loss": 0.1233, "step": 4200 }, { "epoch": 3.077655677655678, "grad_norm": 0.609375, "learning_rate": 9.961158209577357e-05, "loss": 0.1797, "step": 4201 }, { "epoch": 3.0783882783882786, "grad_norm": 1.1171875, "learning_rate": 9.961115358592624e-05, "loss": 0.1701, "step": 4202 }, { "epoch": 3.0791208791208793, "grad_norm": 1.3203125, "learning_rate": 9.961072484077126e-05, "loss": 0.2174, "step": 4203 }, { "epoch": 3.07985347985348, "grad_norm": 1.2421875, "learning_rate": 9.961029586031066e-05, "loss": 0.174, "step": 4204 }, { "epoch": 3.0805860805860807, "grad_norm": 0.88671875, "learning_rate": 9.96098666445465e-05, "loss": 0.1862, "step": 4205 }, { "epoch": 3.0813186813186815, "grad_norm": 0.9296875, "learning_rate": 9.960943719348088e-05, "loss": 0.1757, "step": 4206 }, { "epoch": 3.082051282051282, "grad_norm": 1.234375, "learning_rate": 9.960900750711581e-05, "loss": 0.1413, "step": 4207 }, { "epoch": 3.082783882783883, "grad_norm": 0.671875, "learning_rate": 9.960857758545337e-05, "loss": 0.1235, "step": 4208 }, { "epoch": 3.0835164835164837, "grad_norm": 0.75390625, "learning_rate": 9.960814742849559e-05, "loss": 0.1418, "step": 4209 }, { "epoch": 3.0842490842490844, "grad_norm": 1.734375, "learning_rate": 9.960771703624457e-05, "loss": 0.1757, "step": 4210 }, { "epoch": 3.084981684981685, "grad_norm": 1.4375, "learning_rate": 9.960728640870236e-05, "loss": 0.16, "step": 4211 }, { "epoch": 3.085714285714286, "grad_norm": 1.015625, "learning_rate": 9.960685554587101e-05, "loss": 0.1439, "step": 4212 }, { "epoch": 3.0864468864468866, "grad_norm": 2.640625, "learning_rate": 9.960642444775262e-05, "loss": 0.1421, "step": 4213 }, { "epoch": 3.0871794871794873, "grad_norm": 3.046875, "learning_rate": 9.96059931143492e-05, "loss": 0.1894, "step": 4214 }, { "epoch": 3.087912087912088, "grad_norm": 2.796875, "learning_rate": 9.960556154566285e-05, "loss": 0.2022, "step": 4215 }, { "epoch": 3.0886446886446888, "grad_norm": 3.03125, "learning_rate": 9.960512974169564e-05, "loss": 0.1793, "step": 4216 }, { "epoch": 3.0893772893772895, "grad_norm": 0.6875, "learning_rate": 9.960469770244965e-05, "loss": 0.1635, "step": 4217 }, { "epoch": 3.0901098901098902, "grad_norm": 0.83984375, "learning_rate": 9.960426542792692e-05, "loss": 0.1412, "step": 4218 }, { "epoch": 3.090842490842491, "grad_norm": 1.390625, "learning_rate": 9.960383291812953e-05, "loss": 0.1341, "step": 4219 }, { "epoch": 3.0915750915750917, "grad_norm": 2.234375, "learning_rate": 9.960340017305954e-05, "loss": 0.1778, "step": 4220 }, { "epoch": 3.0923076923076924, "grad_norm": 1.5625, "learning_rate": 9.960296719271906e-05, "loss": 0.1771, "step": 4221 }, { "epoch": 3.093040293040293, "grad_norm": 0.64453125, "learning_rate": 9.960253397711014e-05, "loss": 0.1185, "step": 4222 }, { "epoch": 3.093772893772894, "grad_norm": 1.3046875, "learning_rate": 9.960210052623488e-05, "loss": 0.2196, "step": 4223 }, { "epoch": 3.0945054945054946, "grad_norm": 1.078125, "learning_rate": 9.96016668400953e-05, "loss": 0.1519, "step": 4224 }, { "epoch": 3.0952380952380953, "grad_norm": 1.6171875, "learning_rate": 9.960123291869355e-05, "loss": 0.1488, "step": 4225 }, { "epoch": 3.095970695970696, "grad_norm": 1.1953125, "learning_rate": 9.960079876203166e-05, "loss": 0.1447, "step": 4226 }, { "epoch": 3.096703296703297, "grad_norm": 0.92578125, "learning_rate": 9.960036437011174e-05, "loss": 0.1755, "step": 4227 }, { "epoch": 3.0974358974358975, "grad_norm": 1.3515625, "learning_rate": 9.959992974293583e-05, "loss": 0.1724, "step": 4228 }, { "epoch": 3.0981684981684983, "grad_norm": 1.6015625, "learning_rate": 9.959949488050605e-05, "loss": 0.1799, "step": 4229 }, { "epoch": 3.098901098901099, "grad_norm": 2.328125, "learning_rate": 9.959905978282446e-05, "loss": 0.1591, "step": 4230 }, { "epoch": 3.0996336996336997, "grad_norm": 1.734375, "learning_rate": 9.959862444989318e-05, "loss": 0.1742, "step": 4231 }, { "epoch": 3.1003663003663005, "grad_norm": 0.9140625, "learning_rate": 9.959818888171426e-05, "loss": 0.1595, "step": 4232 }, { "epoch": 3.101098901098901, "grad_norm": 0.7265625, "learning_rate": 9.959775307828981e-05, "loss": 0.1929, "step": 4233 }, { "epoch": 3.101831501831502, "grad_norm": 1.8125, "learning_rate": 9.959731703962188e-05, "loss": 0.1865, "step": 4234 }, { "epoch": 3.1025641025641026, "grad_norm": 1.7421875, "learning_rate": 9.959688076571261e-05, "loss": 0.1677, "step": 4235 }, { "epoch": 3.1032967032967034, "grad_norm": 2.53125, "learning_rate": 9.959644425656404e-05, "loss": 0.1722, "step": 4236 }, { "epoch": 3.104029304029304, "grad_norm": 1.125, "learning_rate": 9.959600751217832e-05, "loss": 0.1956, "step": 4237 }, { "epoch": 3.104761904761905, "grad_norm": 1.0078125, "learning_rate": 9.959557053255749e-05, "loss": 0.1664, "step": 4238 }, { "epoch": 3.1054945054945056, "grad_norm": 2.875, "learning_rate": 9.959513331770369e-05, "loss": 0.2262, "step": 4239 }, { "epoch": 3.1062271062271063, "grad_norm": 2.53125, "learning_rate": 9.959469586761898e-05, "loss": 0.1866, "step": 4240 }, { "epoch": 3.106959706959707, "grad_norm": 2.453125, "learning_rate": 9.959425818230545e-05, "loss": 0.1557, "step": 4241 }, { "epoch": 3.1076923076923078, "grad_norm": 1.6484375, "learning_rate": 9.959382026176522e-05, "loss": 0.1936, "step": 4242 }, { "epoch": 3.1084249084249085, "grad_norm": 0.875, "learning_rate": 9.959338210600037e-05, "loss": 0.1781, "step": 4243 }, { "epoch": 3.109157509157509, "grad_norm": 0.97265625, "learning_rate": 9.959294371501301e-05, "loss": 0.1545, "step": 4244 }, { "epoch": 3.10989010989011, "grad_norm": 1.21875, "learning_rate": 9.959250508880527e-05, "loss": 0.1908, "step": 4245 }, { "epoch": 3.1106227106227107, "grad_norm": 0.80859375, "learning_rate": 9.95920662273792e-05, "loss": 0.1284, "step": 4246 }, { "epoch": 3.1113553113553114, "grad_norm": 1.0625, "learning_rate": 9.959162713073694e-05, "loss": 0.1756, "step": 4247 }, { "epoch": 3.112087912087912, "grad_norm": 0.73046875, "learning_rate": 9.959118779888056e-05, "loss": 0.1557, "step": 4248 }, { "epoch": 3.112820512820513, "grad_norm": 1.015625, "learning_rate": 9.959074823181219e-05, "loss": 0.1381, "step": 4249 }, { "epoch": 3.1135531135531136, "grad_norm": 1.265625, "learning_rate": 9.959030842953393e-05, "loss": 0.1897, "step": 4250 }, { "epoch": 3.1142857142857143, "grad_norm": 0.921875, "learning_rate": 9.958986839204788e-05, "loss": 0.1216, "step": 4251 }, { "epoch": 3.115018315018315, "grad_norm": 0.8125, "learning_rate": 9.958942811935617e-05, "loss": 0.1759, "step": 4252 }, { "epoch": 3.115750915750916, "grad_norm": 1.21875, "learning_rate": 9.958898761146088e-05, "loss": 0.2723, "step": 4253 }, { "epoch": 3.1164835164835165, "grad_norm": 0.765625, "learning_rate": 9.958854686836415e-05, "loss": 0.1378, "step": 4254 }, { "epoch": 3.1172161172161172, "grad_norm": 1.1484375, "learning_rate": 9.958810589006807e-05, "loss": 0.1774, "step": 4255 }, { "epoch": 3.117948717948718, "grad_norm": 1.0, "learning_rate": 9.958766467657476e-05, "loss": 0.1782, "step": 4256 }, { "epoch": 3.1186813186813187, "grad_norm": 1.3203125, "learning_rate": 9.958722322788634e-05, "loss": 0.1265, "step": 4257 }, { "epoch": 3.1194139194139194, "grad_norm": 0.828125, "learning_rate": 9.958678154400492e-05, "loss": 0.1524, "step": 4258 }, { "epoch": 3.12014652014652, "grad_norm": 1.015625, "learning_rate": 9.95863396249326e-05, "loss": 0.1753, "step": 4259 }, { "epoch": 3.120879120879121, "grad_norm": 1.703125, "learning_rate": 9.958589747067151e-05, "loss": 0.1485, "step": 4260 }, { "epoch": 3.1216117216117216, "grad_norm": 1.7734375, "learning_rate": 9.95854550812238e-05, "loss": 0.1687, "step": 4261 }, { "epoch": 3.1223443223443224, "grad_norm": 1.875, "learning_rate": 9.958501245659155e-05, "loss": 0.1681, "step": 4262 }, { "epoch": 3.123076923076923, "grad_norm": 1.2265625, "learning_rate": 9.958456959677688e-05, "loss": 0.1459, "step": 4263 }, { "epoch": 3.123809523809524, "grad_norm": 0.69140625, "learning_rate": 9.958412650178192e-05, "loss": 0.1588, "step": 4264 }, { "epoch": 3.1245421245421245, "grad_norm": 0.87109375, "learning_rate": 9.958368317160881e-05, "loss": 0.1876, "step": 4265 }, { "epoch": 3.1252747252747253, "grad_norm": 0.9453125, "learning_rate": 9.958323960625966e-05, "loss": 0.1733, "step": 4266 }, { "epoch": 3.126007326007326, "grad_norm": 1.203125, "learning_rate": 9.95827958057366e-05, "loss": 0.1491, "step": 4267 }, { "epoch": 3.1267399267399267, "grad_norm": 1.0078125, "learning_rate": 9.958235177004173e-05, "loss": 0.1289, "step": 4268 }, { "epoch": 3.1274725274725275, "grad_norm": 0.66796875, "learning_rate": 9.958190749917723e-05, "loss": 0.1604, "step": 4269 }, { "epoch": 3.128205128205128, "grad_norm": 1.0546875, "learning_rate": 9.958146299314519e-05, "loss": 0.1431, "step": 4270 }, { "epoch": 3.128937728937729, "grad_norm": 1.1328125, "learning_rate": 9.958101825194774e-05, "loss": 0.1736, "step": 4271 }, { "epoch": 3.1296703296703297, "grad_norm": 1.6015625, "learning_rate": 9.958057327558701e-05, "loss": 0.1696, "step": 4272 }, { "epoch": 3.1304029304029304, "grad_norm": 0.734375, "learning_rate": 9.958012806406516e-05, "loss": 0.1449, "step": 4273 }, { "epoch": 3.131135531135531, "grad_norm": 1.46875, "learning_rate": 9.957968261738429e-05, "loss": 0.149, "step": 4274 }, { "epoch": 3.131868131868132, "grad_norm": 1.5546875, "learning_rate": 9.957923693554657e-05, "loss": 0.1888, "step": 4275 }, { "epoch": 3.1326007326007326, "grad_norm": 0.90234375, "learning_rate": 9.957879101855411e-05, "loss": 0.1419, "step": 4276 }, { "epoch": 3.1333333333333333, "grad_norm": 0.87109375, "learning_rate": 9.957834486640902e-05, "loss": 0.1449, "step": 4277 }, { "epoch": 3.134065934065934, "grad_norm": 1.078125, "learning_rate": 9.95778984791135e-05, "loss": 0.2062, "step": 4278 }, { "epoch": 3.1347985347985348, "grad_norm": 1.0, "learning_rate": 9.957745185666966e-05, "loss": 0.1647, "step": 4279 }, { "epoch": 3.1355311355311355, "grad_norm": 1.0078125, "learning_rate": 9.957700499907961e-05, "loss": 0.1792, "step": 4280 }, { "epoch": 3.1362637362637362, "grad_norm": 0.9140625, "learning_rate": 9.957655790634554e-05, "loss": 0.1776, "step": 4281 }, { "epoch": 3.136996336996337, "grad_norm": 0.76953125, "learning_rate": 9.957611057846955e-05, "loss": 0.1536, "step": 4282 }, { "epoch": 3.1377289377289377, "grad_norm": 0.80859375, "learning_rate": 9.957566301545383e-05, "loss": 0.1499, "step": 4283 }, { "epoch": 3.1384615384615384, "grad_norm": 0.609375, "learning_rate": 9.957521521730046e-05, "loss": 0.1408, "step": 4284 }, { "epoch": 3.139194139194139, "grad_norm": 0.72265625, "learning_rate": 9.957476718401166e-05, "loss": 0.1423, "step": 4285 }, { "epoch": 3.13992673992674, "grad_norm": 0.8203125, "learning_rate": 9.957431891558953e-05, "loss": 0.1735, "step": 4286 }, { "epoch": 3.1406593406593406, "grad_norm": 1.0234375, "learning_rate": 9.957387041203623e-05, "loss": 0.1633, "step": 4287 }, { "epoch": 3.1413919413919413, "grad_norm": 1.1015625, "learning_rate": 9.957342167335391e-05, "loss": 0.1802, "step": 4288 }, { "epoch": 3.142124542124542, "grad_norm": 0.609375, "learning_rate": 9.957297269954469e-05, "loss": 0.133, "step": 4289 }, { "epoch": 3.142857142857143, "grad_norm": 0.78515625, "learning_rate": 9.957252349061078e-05, "loss": 0.1665, "step": 4290 }, { "epoch": 3.1435897435897435, "grad_norm": 0.96875, "learning_rate": 9.95720740465543e-05, "loss": 0.2037, "step": 4291 }, { "epoch": 3.1443223443223443, "grad_norm": 0.79296875, "learning_rate": 9.957162436737738e-05, "loss": 0.1631, "step": 4292 }, { "epoch": 3.145054945054945, "grad_norm": 1.515625, "learning_rate": 9.957117445308223e-05, "loss": 0.1647, "step": 4293 }, { "epoch": 3.1457875457875457, "grad_norm": 0.9140625, "learning_rate": 9.957072430367095e-05, "loss": 0.1522, "step": 4294 }, { "epoch": 3.1465201465201464, "grad_norm": 1.0390625, "learning_rate": 9.957027391914572e-05, "loss": 0.1474, "step": 4295 }, { "epoch": 3.147252747252747, "grad_norm": 1.0859375, "learning_rate": 9.956982329950871e-05, "loss": 0.1776, "step": 4296 }, { "epoch": 3.147985347985348, "grad_norm": 0.73046875, "learning_rate": 9.956937244476205e-05, "loss": 0.1518, "step": 4297 }, { "epoch": 3.1487179487179486, "grad_norm": 0.7421875, "learning_rate": 9.956892135490793e-05, "loss": 0.1576, "step": 4298 }, { "epoch": 3.1494505494505494, "grad_norm": 1.046875, "learning_rate": 9.956847002994851e-05, "loss": 0.1527, "step": 4299 }, { "epoch": 3.15018315018315, "grad_norm": 1.1015625, "learning_rate": 9.956801846988593e-05, "loss": 0.2016, "step": 4300 }, { "epoch": 3.150915750915751, "grad_norm": 1.2734375, "learning_rate": 9.956756667472238e-05, "loss": 0.1527, "step": 4301 }, { "epoch": 3.1516483516483516, "grad_norm": 0.8828125, "learning_rate": 9.956711464445999e-05, "loss": 0.1448, "step": 4302 }, { "epoch": 3.1523809523809523, "grad_norm": 0.91796875, "learning_rate": 9.956666237910097e-05, "loss": 0.1648, "step": 4303 }, { "epoch": 3.153113553113553, "grad_norm": 1.34375, "learning_rate": 9.956620987864745e-05, "loss": 0.2055, "step": 4304 }, { "epoch": 3.1538461538461537, "grad_norm": 1.2421875, "learning_rate": 9.95657571431016e-05, "loss": 0.1616, "step": 4305 }, { "epoch": 3.1545787545787545, "grad_norm": 1.390625, "learning_rate": 9.956530417246562e-05, "loss": 0.188, "step": 4306 }, { "epoch": 3.155311355311355, "grad_norm": 0.984375, "learning_rate": 9.956485096674164e-05, "loss": 0.1507, "step": 4307 }, { "epoch": 3.156043956043956, "grad_norm": 0.8515625, "learning_rate": 9.956439752593186e-05, "loss": 0.1616, "step": 4308 }, { "epoch": 3.1567765567765567, "grad_norm": 0.83203125, "learning_rate": 9.956394385003845e-05, "loss": 0.1295, "step": 4309 }, { "epoch": 3.1575091575091574, "grad_norm": 1.3984375, "learning_rate": 9.956348993906358e-05, "loss": 0.1608, "step": 4310 }, { "epoch": 3.158241758241758, "grad_norm": 1.65625, "learning_rate": 9.956303579300941e-05, "loss": 0.2272, "step": 4311 }, { "epoch": 3.158974358974359, "grad_norm": 1.1015625, "learning_rate": 9.956258141187814e-05, "loss": 0.1395, "step": 4312 }, { "epoch": 3.1597069597069596, "grad_norm": 1.125, "learning_rate": 9.956212679567192e-05, "loss": 0.1671, "step": 4313 }, { "epoch": 3.1604395604395603, "grad_norm": 1.390625, "learning_rate": 9.956167194439297e-05, "loss": 0.188, "step": 4314 }, { "epoch": 3.161172161172161, "grad_norm": 0.6640625, "learning_rate": 9.956121685804342e-05, "loss": 0.1605, "step": 4315 }, { "epoch": 3.1619047619047618, "grad_norm": 1.1484375, "learning_rate": 9.95607615366255e-05, "loss": 0.1424, "step": 4316 }, { "epoch": 3.1626373626373625, "grad_norm": 0.72265625, "learning_rate": 9.956030598014136e-05, "loss": 0.1841, "step": 4317 }, { "epoch": 3.1633699633699632, "grad_norm": 0.6953125, "learning_rate": 9.955985018859318e-05, "loss": 0.1044, "step": 4318 }, { "epoch": 3.164102564102564, "grad_norm": 1.4609375, "learning_rate": 9.955939416198315e-05, "loss": 0.1679, "step": 4319 }, { "epoch": 3.1648351648351647, "grad_norm": 1.015625, "learning_rate": 9.955893790031346e-05, "loss": 0.1573, "step": 4320 }, { "epoch": 3.1655677655677654, "grad_norm": 0.83203125, "learning_rate": 9.955848140358629e-05, "loss": 0.1627, "step": 4321 }, { "epoch": 3.166300366300366, "grad_norm": 0.7734375, "learning_rate": 9.955802467180384e-05, "loss": 0.169, "step": 4322 }, { "epoch": 3.167032967032967, "grad_norm": 0.9140625, "learning_rate": 9.955756770496828e-05, "loss": 0.207, "step": 4323 }, { "epoch": 3.1677655677655676, "grad_norm": 1.1640625, "learning_rate": 9.95571105030818e-05, "loss": 0.1718, "step": 4324 }, { "epoch": 3.1684981684981683, "grad_norm": 0.71875, "learning_rate": 9.955665306614662e-05, "loss": 0.137, "step": 4325 }, { "epoch": 3.169230769230769, "grad_norm": 1.375, "learning_rate": 9.95561953941649e-05, "loss": 0.1792, "step": 4326 }, { "epoch": 3.16996336996337, "grad_norm": 0.9453125, "learning_rate": 9.955573748713885e-05, "loss": 0.1481, "step": 4327 }, { "epoch": 3.1706959706959705, "grad_norm": 0.8671875, "learning_rate": 9.955527934507067e-05, "loss": 0.1423, "step": 4328 }, { "epoch": 3.1714285714285713, "grad_norm": 1.6953125, "learning_rate": 9.95548209679625e-05, "loss": 0.1487, "step": 4329 }, { "epoch": 3.172161172161172, "grad_norm": 0.84375, "learning_rate": 9.955436235581662e-05, "loss": 0.1426, "step": 4330 }, { "epoch": 3.1728937728937727, "grad_norm": 0.8828125, "learning_rate": 9.955390350863518e-05, "loss": 0.1704, "step": 4331 }, { "epoch": 3.1736263736263735, "grad_norm": 1.09375, "learning_rate": 9.955344442642038e-05, "loss": 0.1702, "step": 4332 }, { "epoch": 3.174358974358974, "grad_norm": 0.671875, "learning_rate": 9.955298510917441e-05, "loss": 0.1439, "step": 4333 }, { "epoch": 3.175091575091575, "grad_norm": 1.1953125, "learning_rate": 9.95525255568995e-05, "loss": 0.1879, "step": 4334 }, { "epoch": 3.1758241758241756, "grad_norm": 1.0703125, "learning_rate": 9.955206576959784e-05, "loss": 0.1623, "step": 4335 }, { "epoch": 3.1765567765567764, "grad_norm": 0.95703125, "learning_rate": 9.955160574727162e-05, "loss": 0.1426, "step": 4336 }, { "epoch": 3.177289377289377, "grad_norm": 0.765625, "learning_rate": 9.955114548992306e-05, "loss": 0.1374, "step": 4337 }, { "epoch": 3.178021978021978, "grad_norm": 1.5546875, "learning_rate": 9.955068499755437e-05, "loss": 0.1734, "step": 4338 }, { "epoch": 3.1787545787545786, "grad_norm": 1.171875, "learning_rate": 9.955022427016772e-05, "loss": 0.1827, "step": 4339 }, { "epoch": 3.1794871794871793, "grad_norm": 0.91796875, "learning_rate": 9.954976330776536e-05, "loss": 0.1588, "step": 4340 }, { "epoch": 3.18021978021978, "grad_norm": 0.9609375, "learning_rate": 9.954930211034948e-05, "loss": 0.1145, "step": 4341 }, { "epoch": 3.1809523809523808, "grad_norm": 0.73828125, "learning_rate": 9.95488406779223e-05, "loss": 0.1655, "step": 4342 }, { "epoch": 3.1816849816849815, "grad_norm": 1.1328125, "learning_rate": 9.9548379010486e-05, "loss": 0.2051, "step": 4343 }, { "epoch": 3.182417582417582, "grad_norm": 1.609375, "learning_rate": 9.954791710804283e-05, "loss": 0.1599, "step": 4344 }, { "epoch": 3.183150183150183, "grad_norm": 1.3671875, "learning_rate": 9.9547454970595e-05, "loss": 0.1555, "step": 4345 }, { "epoch": 3.183882783882784, "grad_norm": 0.80078125, "learning_rate": 9.954699259814468e-05, "loss": 0.1651, "step": 4346 }, { "epoch": 3.184615384615385, "grad_norm": 1.03125, "learning_rate": 9.954652999069415e-05, "loss": 0.1827, "step": 4347 }, { "epoch": 3.1853479853479856, "grad_norm": 0.9453125, "learning_rate": 9.954606714824558e-05, "loss": 0.1502, "step": 4348 }, { "epoch": 3.1860805860805863, "grad_norm": 0.734375, "learning_rate": 9.95456040708012e-05, "loss": 0.1665, "step": 4349 }, { "epoch": 3.186813186813187, "grad_norm": 0.89453125, "learning_rate": 9.954514075836323e-05, "loss": 0.1978, "step": 4350 }, { "epoch": 3.1875457875457878, "grad_norm": 0.8515625, "learning_rate": 9.954467721093389e-05, "loss": 0.1651, "step": 4351 }, { "epoch": 3.1882783882783885, "grad_norm": 1.015625, "learning_rate": 9.954421342851542e-05, "loss": 0.123, "step": 4352 }, { "epoch": 3.1890109890109892, "grad_norm": 0.8359375, "learning_rate": 9.954374941110998e-05, "loss": 0.1535, "step": 4353 }, { "epoch": 3.18974358974359, "grad_norm": 1.4921875, "learning_rate": 9.954328515871988e-05, "loss": 0.1498, "step": 4354 }, { "epoch": 3.1904761904761907, "grad_norm": 1.1640625, "learning_rate": 9.954282067134728e-05, "loss": 0.1886, "step": 4355 }, { "epoch": 3.1912087912087914, "grad_norm": 1.046875, "learning_rate": 9.954235594899443e-05, "loss": 0.1748, "step": 4356 }, { "epoch": 3.191941391941392, "grad_norm": 1.0, "learning_rate": 9.954189099166355e-05, "loss": 0.1819, "step": 4357 }, { "epoch": 3.192673992673993, "grad_norm": 0.82421875, "learning_rate": 9.954142579935687e-05, "loss": 0.1655, "step": 4358 }, { "epoch": 3.1934065934065936, "grad_norm": 1.015625, "learning_rate": 9.954096037207664e-05, "loss": 0.1393, "step": 4359 }, { "epoch": 3.1941391941391943, "grad_norm": 1.1484375, "learning_rate": 9.954049470982505e-05, "loss": 0.2024, "step": 4360 }, { "epoch": 3.194871794871795, "grad_norm": 0.953125, "learning_rate": 9.954002881260435e-05, "loss": 0.1335, "step": 4361 }, { "epoch": 3.195604395604396, "grad_norm": 1.359375, "learning_rate": 9.953956268041677e-05, "loss": 0.192, "step": 4362 }, { "epoch": 3.1963369963369965, "grad_norm": 1.109375, "learning_rate": 9.953909631326457e-05, "loss": 0.1457, "step": 4363 }, { "epoch": 3.1970695970695973, "grad_norm": 0.6796875, "learning_rate": 9.953862971114995e-05, "loss": 0.1342, "step": 4364 }, { "epoch": 3.197802197802198, "grad_norm": 1.109375, "learning_rate": 9.953816287407514e-05, "loss": 0.225, "step": 4365 }, { "epoch": 3.1985347985347987, "grad_norm": 1.0234375, "learning_rate": 9.953769580204241e-05, "loss": 0.1408, "step": 4366 }, { "epoch": 3.1992673992673994, "grad_norm": 1.328125, "learning_rate": 9.953722849505398e-05, "loss": 0.1527, "step": 4367 }, { "epoch": 3.2, "grad_norm": 0.65625, "learning_rate": 9.953676095311209e-05, "loss": 0.1406, "step": 4368 }, { "epoch": 3.200732600732601, "grad_norm": 1.796875, "learning_rate": 9.953629317621897e-05, "loss": 0.1821, "step": 4369 }, { "epoch": 3.2014652014652016, "grad_norm": 1.5390625, "learning_rate": 9.953582516437688e-05, "loss": 0.1489, "step": 4370 }, { "epoch": 3.2021978021978024, "grad_norm": 0.6875, "learning_rate": 9.953535691758806e-05, "loss": 0.1354, "step": 4371 }, { "epoch": 3.202930402930403, "grad_norm": 1.2421875, "learning_rate": 9.953488843585475e-05, "loss": 0.1919, "step": 4372 }, { "epoch": 3.203663003663004, "grad_norm": 0.81640625, "learning_rate": 9.953441971917917e-05, "loss": 0.1462, "step": 4373 }, { "epoch": 3.2043956043956046, "grad_norm": 1.0078125, "learning_rate": 9.95339507675636e-05, "loss": 0.1467, "step": 4374 }, { "epoch": 3.2051282051282053, "grad_norm": 1.0234375, "learning_rate": 9.95334815810103e-05, "loss": 0.1952, "step": 4375 }, { "epoch": 3.205860805860806, "grad_norm": 1.6171875, "learning_rate": 9.953301215952146e-05, "loss": 0.1805, "step": 4376 }, { "epoch": 3.2065934065934067, "grad_norm": 0.8984375, "learning_rate": 9.953254250309936e-05, "loss": 0.1611, "step": 4377 }, { "epoch": 3.2073260073260075, "grad_norm": 1.359375, "learning_rate": 9.953207261174627e-05, "loss": 0.195, "step": 4378 }, { "epoch": 3.208058608058608, "grad_norm": 0.6640625, "learning_rate": 9.953160248546444e-05, "loss": 0.1257, "step": 4379 }, { "epoch": 3.208791208791209, "grad_norm": 0.89453125, "learning_rate": 9.953113212425608e-05, "loss": 0.1475, "step": 4380 }, { "epoch": 3.2095238095238097, "grad_norm": 1.484375, "learning_rate": 9.953066152812348e-05, "loss": 0.1289, "step": 4381 }, { "epoch": 3.2102564102564104, "grad_norm": 1.765625, "learning_rate": 9.953019069706887e-05, "loss": 0.1682, "step": 4382 }, { "epoch": 3.210989010989011, "grad_norm": 1.390625, "learning_rate": 9.952971963109452e-05, "loss": 0.1694, "step": 4383 }, { "epoch": 3.211721611721612, "grad_norm": 1.3515625, "learning_rate": 9.95292483302027e-05, "loss": 0.1677, "step": 4384 }, { "epoch": 3.2124542124542126, "grad_norm": 2.203125, "learning_rate": 9.952877679439567e-05, "loss": 0.1686, "step": 4385 }, { "epoch": 3.2131868131868133, "grad_norm": 1.578125, "learning_rate": 9.952830502367566e-05, "loss": 0.1778, "step": 4386 }, { "epoch": 3.213919413919414, "grad_norm": 2.265625, "learning_rate": 9.952783301804495e-05, "loss": 0.1723, "step": 4387 }, { "epoch": 3.2146520146520148, "grad_norm": 1.65625, "learning_rate": 9.952736077750578e-05, "loss": 0.1914, "step": 4388 }, { "epoch": 3.2153846153846155, "grad_norm": 1.734375, "learning_rate": 9.952688830206044e-05, "loss": 0.158, "step": 4389 }, { "epoch": 3.2161172161172162, "grad_norm": 1.0703125, "learning_rate": 9.95264155917112e-05, "loss": 0.1656, "step": 4390 }, { "epoch": 3.216849816849817, "grad_norm": 2.703125, "learning_rate": 9.95259426464603e-05, "loss": 0.2056, "step": 4391 }, { "epoch": 3.2175824175824177, "grad_norm": 2.09375, "learning_rate": 9.952546946631e-05, "loss": 0.2025, "step": 4392 }, { "epoch": 3.2183150183150184, "grad_norm": 1.796875, "learning_rate": 9.952499605126259e-05, "loss": 0.1593, "step": 4393 }, { "epoch": 3.219047619047619, "grad_norm": 0.92578125, "learning_rate": 9.952452240132034e-05, "loss": 0.1556, "step": 4394 }, { "epoch": 3.21978021978022, "grad_norm": 0.77734375, "learning_rate": 9.952404851648551e-05, "loss": 0.1477, "step": 4395 }, { "epoch": 3.2205128205128206, "grad_norm": 1.765625, "learning_rate": 9.952357439676034e-05, "loss": 0.1796, "step": 4396 }, { "epoch": 3.2212454212454213, "grad_norm": 1.7734375, "learning_rate": 9.952310004214716e-05, "loss": 0.1769, "step": 4397 }, { "epoch": 3.221978021978022, "grad_norm": 1.71875, "learning_rate": 9.952262545264822e-05, "loss": 0.1707, "step": 4398 }, { "epoch": 3.222710622710623, "grad_norm": 1.0703125, "learning_rate": 9.952215062826577e-05, "loss": 0.1729, "step": 4399 }, { "epoch": 3.2234432234432235, "grad_norm": 1.1953125, "learning_rate": 9.952167556900212e-05, "loss": 0.2507, "step": 4400 }, { "epoch": 3.2241758241758243, "grad_norm": 0.9609375, "learning_rate": 9.952120027485952e-05, "loss": 0.1592, "step": 4401 }, { "epoch": 3.224908424908425, "grad_norm": 1.703125, "learning_rate": 9.952072474584026e-05, "loss": 0.1657, "step": 4402 }, { "epoch": 3.2256410256410257, "grad_norm": 1.5546875, "learning_rate": 9.95202489819466e-05, "loss": 0.215, "step": 4403 }, { "epoch": 3.2263736263736265, "grad_norm": 1.1640625, "learning_rate": 9.951977298318085e-05, "loss": 0.1316, "step": 4404 }, { "epoch": 3.227106227106227, "grad_norm": 0.69921875, "learning_rate": 9.951929674954527e-05, "loss": 0.1206, "step": 4405 }, { "epoch": 3.227838827838828, "grad_norm": 1.1796875, "learning_rate": 9.951882028104215e-05, "loss": 0.1239, "step": 4406 }, { "epoch": 3.2285714285714286, "grad_norm": 0.94140625, "learning_rate": 9.951834357767376e-05, "loss": 0.1705, "step": 4407 }, { "epoch": 3.2293040293040294, "grad_norm": 1.265625, "learning_rate": 9.951786663944241e-05, "loss": 0.1601, "step": 4408 }, { "epoch": 3.23003663003663, "grad_norm": 0.93359375, "learning_rate": 9.951738946635037e-05, "loss": 0.1848, "step": 4409 }, { "epoch": 3.230769230769231, "grad_norm": 1.484375, "learning_rate": 9.951691205839992e-05, "loss": 0.1724, "step": 4410 }, { "epoch": 3.2315018315018316, "grad_norm": 1.4921875, "learning_rate": 9.951643441559336e-05, "loss": 0.1621, "step": 4411 }, { "epoch": 3.2322344322344323, "grad_norm": 1.359375, "learning_rate": 9.951595653793295e-05, "loss": 0.1694, "step": 4412 }, { "epoch": 3.232967032967033, "grad_norm": 0.79296875, "learning_rate": 9.951547842542101e-05, "loss": 0.1472, "step": 4413 }, { "epoch": 3.2336996336996338, "grad_norm": 0.56640625, "learning_rate": 9.951500007805983e-05, "loss": 0.1681, "step": 4414 }, { "epoch": 3.2344322344322345, "grad_norm": 0.9765625, "learning_rate": 9.951452149585169e-05, "loss": 0.1822, "step": 4415 }, { "epoch": 3.235164835164835, "grad_norm": 0.9453125, "learning_rate": 9.951404267879889e-05, "loss": 0.1667, "step": 4416 }, { "epoch": 3.235897435897436, "grad_norm": 1.40625, "learning_rate": 9.951356362690372e-05, "loss": 0.2403, "step": 4417 }, { "epoch": 3.2366300366300367, "grad_norm": 1.1875, "learning_rate": 9.951308434016848e-05, "loss": 0.1538, "step": 4418 }, { "epoch": 3.2373626373626374, "grad_norm": 0.91015625, "learning_rate": 9.951260481859545e-05, "loss": 0.1946, "step": 4419 }, { "epoch": 3.238095238095238, "grad_norm": 1.5078125, "learning_rate": 9.951212506218696e-05, "loss": 0.175, "step": 4420 }, { "epoch": 3.238827838827839, "grad_norm": 1.2578125, "learning_rate": 9.951164507094529e-05, "loss": 0.1816, "step": 4421 }, { "epoch": 3.2395604395604396, "grad_norm": 1.2734375, "learning_rate": 9.951116484487271e-05, "loss": 0.1432, "step": 4422 }, { "epoch": 3.2402930402930403, "grad_norm": 0.828125, "learning_rate": 9.951068438397156e-05, "loss": 0.205, "step": 4423 }, { "epoch": 3.241025641025641, "grad_norm": 1.90625, "learning_rate": 9.951020368824416e-05, "loss": 0.1446, "step": 4424 }, { "epoch": 3.241758241758242, "grad_norm": 1.640625, "learning_rate": 9.950972275769276e-05, "loss": 0.1469, "step": 4425 }, { "epoch": 3.2424908424908425, "grad_norm": 0.99609375, "learning_rate": 9.950924159231969e-05, "loss": 0.1876, "step": 4426 }, { "epoch": 3.2432234432234432, "grad_norm": 1.25, "learning_rate": 9.950876019212727e-05, "loss": 0.1841, "step": 4427 }, { "epoch": 3.243956043956044, "grad_norm": 1.1171875, "learning_rate": 9.950827855711777e-05, "loss": 0.1884, "step": 4428 }, { "epoch": 3.2446886446886447, "grad_norm": 1.546875, "learning_rate": 9.950779668729354e-05, "loss": 0.1976, "step": 4429 }, { "epoch": 3.2454212454212454, "grad_norm": 1.109375, "learning_rate": 9.950731458265685e-05, "loss": 0.1751, "step": 4430 }, { "epoch": 3.246153846153846, "grad_norm": 0.71484375, "learning_rate": 9.950683224321004e-05, "loss": 0.1513, "step": 4431 }, { "epoch": 3.246886446886447, "grad_norm": 0.53515625, "learning_rate": 9.95063496689554e-05, "loss": 0.1512, "step": 4432 }, { "epoch": 3.2476190476190476, "grad_norm": 0.77734375, "learning_rate": 9.950586685989526e-05, "loss": 0.1827, "step": 4433 }, { "epoch": 3.2483516483516484, "grad_norm": 0.8828125, "learning_rate": 9.950538381603191e-05, "loss": 0.1992, "step": 4434 }, { "epoch": 3.249084249084249, "grad_norm": 1.3203125, "learning_rate": 9.95049005373677e-05, "loss": 0.1525, "step": 4435 }, { "epoch": 3.24981684981685, "grad_norm": 1.0390625, "learning_rate": 9.950441702390491e-05, "loss": 0.1651, "step": 4436 }, { "epoch": 3.2505494505494505, "grad_norm": 0.7265625, "learning_rate": 9.950393327564586e-05, "loss": 0.1958, "step": 4437 }, { "epoch": 3.2512820512820513, "grad_norm": 1.0078125, "learning_rate": 9.95034492925929e-05, "loss": 0.1394, "step": 4438 }, { "epoch": 3.252014652014652, "grad_norm": 0.75, "learning_rate": 9.950296507474832e-05, "loss": 0.1654, "step": 4439 }, { "epoch": 3.2527472527472527, "grad_norm": 1.421875, "learning_rate": 9.950248062211445e-05, "loss": 0.2037, "step": 4440 }, { "epoch": 3.2534798534798535, "grad_norm": 1.4140625, "learning_rate": 9.950199593469361e-05, "loss": 0.1675, "step": 4441 }, { "epoch": 3.254212454212454, "grad_norm": 1.2578125, "learning_rate": 9.950151101248811e-05, "loss": 0.1494, "step": 4442 }, { "epoch": 3.254945054945055, "grad_norm": 1.296875, "learning_rate": 9.95010258555003e-05, "loss": 0.1917, "step": 4443 }, { "epoch": 3.2556776556776557, "grad_norm": 1.375, "learning_rate": 9.950054046373248e-05, "loss": 0.1594, "step": 4444 }, { "epoch": 3.2564102564102564, "grad_norm": 1.203125, "learning_rate": 9.950005483718699e-05, "loss": 0.2007, "step": 4445 }, { "epoch": 3.257142857142857, "grad_norm": 1.0703125, "learning_rate": 9.949956897586613e-05, "loss": 0.1812, "step": 4446 }, { "epoch": 3.257875457875458, "grad_norm": 1.109375, "learning_rate": 9.949908287977228e-05, "loss": 0.1316, "step": 4447 }, { "epoch": 3.2586080586080586, "grad_norm": 0.7421875, "learning_rate": 9.949859654890771e-05, "loss": 0.1495, "step": 4448 }, { "epoch": 3.2593406593406593, "grad_norm": 0.99609375, "learning_rate": 9.94981099832748e-05, "loss": 0.1551, "step": 4449 }, { "epoch": 3.26007326007326, "grad_norm": 0.8984375, "learning_rate": 9.949762318287587e-05, "loss": 0.1626, "step": 4450 }, { "epoch": 3.2608058608058608, "grad_norm": 1.171875, "learning_rate": 9.949713614771322e-05, "loss": 0.1953, "step": 4451 }, { "epoch": 3.2615384615384615, "grad_norm": 1.1328125, "learning_rate": 9.94966488777892e-05, "loss": 0.1492, "step": 4452 }, { "epoch": 3.2622710622710622, "grad_norm": 1.2578125, "learning_rate": 9.949616137310617e-05, "loss": 0.1533, "step": 4453 }, { "epoch": 3.263003663003663, "grad_norm": 0.97265625, "learning_rate": 9.949567363366645e-05, "loss": 0.1953, "step": 4454 }, { "epoch": 3.2637362637362637, "grad_norm": 1.1171875, "learning_rate": 9.949518565947237e-05, "loss": 0.1984, "step": 4455 }, { "epoch": 3.2644688644688644, "grad_norm": 0.98046875, "learning_rate": 9.949469745052627e-05, "loss": 0.2045, "step": 4456 }, { "epoch": 3.265201465201465, "grad_norm": 0.88671875, "learning_rate": 9.949420900683046e-05, "loss": 0.1311, "step": 4457 }, { "epoch": 3.265934065934066, "grad_norm": 0.8984375, "learning_rate": 9.949372032838736e-05, "loss": 0.1664, "step": 4458 }, { "epoch": 3.2666666666666666, "grad_norm": 1.234375, "learning_rate": 9.949323141519922e-05, "loss": 0.1823, "step": 4459 }, { "epoch": 3.2673992673992673, "grad_norm": 1.15625, "learning_rate": 9.949274226726844e-05, "loss": 0.1645, "step": 4460 }, { "epoch": 3.268131868131868, "grad_norm": 1.109375, "learning_rate": 9.949225288459737e-05, "loss": 0.1939, "step": 4461 }, { "epoch": 3.268864468864469, "grad_norm": 0.6015625, "learning_rate": 9.94917632671883e-05, "loss": 0.1364, "step": 4462 }, { "epoch": 3.2695970695970695, "grad_norm": 1.328125, "learning_rate": 9.949127341504363e-05, "loss": 0.1968, "step": 4463 }, { "epoch": 3.2703296703296703, "grad_norm": 0.83984375, "learning_rate": 9.949078332816569e-05, "loss": 0.1546, "step": 4464 }, { "epoch": 3.271062271062271, "grad_norm": 0.94921875, "learning_rate": 9.949029300655681e-05, "loss": 0.1241, "step": 4465 }, { "epoch": 3.2717948717948717, "grad_norm": 0.95703125, "learning_rate": 9.948980245021936e-05, "loss": 0.1372, "step": 4466 }, { "epoch": 3.2725274725274724, "grad_norm": 1.46875, "learning_rate": 9.948931165915569e-05, "loss": 0.1841, "step": 4467 }, { "epoch": 3.273260073260073, "grad_norm": 0.73828125, "learning_rate": 9.948882063336812e-05, "loss": 0.1505, "step": 4468 }, { "epoch": 3.273992673992674, "grad_norm": 1.0078125, "learning_rate": 9.948832937285906e-05, "loss": 0.1788, "step": 4469 }, { "epoch": 3.2747252747252746, "grad_norm": 1.0078125, "learning_rate": 9.948783787763082e-05, "loss": 0.1593, "step": 4470 }, { "epoch": 3.2754578754578754, "grad_norm": 0.82421875, "learning_rate": 9.948734614768577e-05, "loss": 0.1667, "step": 4471 }, { "epoch": 3.276190476190476, "grad_norm": 1.3125, "learning_rate": 9.948685418302626e-05, "loss": 0.1591, "step": 4472 }, { "epoch": 3.276923076923077, "grad_norm": 0.890625, "learning_rate": 9.948636198365464e-05, "loss": 0.1673, "step": 4473 }, { "epoch": 3.2776556776556776, "grad_norm": 0.84765625, "learning_rate": 9.948586954957331e-05, "loss": 0.1421, "step": 4474 }, { "epoch": 3.2783882783882783, "grad_norm": 0.7734375, "learning_rate": 9.948537688078457e-05, "loss": 0.1652, "step": 4475 }, { "epoch": 3.279120879120879, "grad_norm": 1.0703125, "learning_rate": 9.94848839772908e-05, "loss": 0.1525, "step": 4476 }, { "epoch": 3.2798534798534797, "grad_norm": 1.6484375, "learning_rate": 9.94843908390944e-05, "loss": 0.162, "step": 4477 }, { "epoch": 3.2805860805860805, "grad_norm": 1.1640625, "learning_rate": 9.948389746619768e-05, "loss": 0.1851, "step": 4478 }, { "epoch": 3.281318681318681, "grad_norm": 1.109375, "learning_rate": 9.948340385860304e-05, "loss": 0.1736, "step": 4479 }, { "epoch": 3.282051282051282, "grad_norm": 1.0234375, "learning_rate": 9.948291001631283e-05, "loss": 0.1323, "step": 4480 }, { "epoch": 3.2827838827838827, "grad_norm": 1.0703125, "learning_rate": 9.948241593932941e-05, "loss": 0.1603, "step": 4481 }, { "epoch": 3.2835164835164834, "grad_norm": 0.99609375, "learning_rate": 9.948192162765517e-05, "loss": 0.1895, "step": 4482 }, { "epoch": 3.284249084249084, "grad_norm": 1.5, "learning_rate": 9.948142708129243e-05, "loss": 0.1625, "step": 4483 }, { "epoch": 3.284981684981685, "grad_norm": 1.0625, "learning_rate": 9.948093230024363e-05, "loss": 0.1788, "step": 4484 }, { "epoch": 3.2857142857142856, "grad_norm": 1.453125, "learning_rate": 9.94804372845111e-05, "loss": 0.1394, "step": 4485 }, { "epoch": 3.2864468864468863, "grad_norm": 1.0546875, "learning_rate": 9.94799420340972e-05, "loss": 0.2233, "step": 4486 }, { "epoch": 3.287179487179487, "grad_norm": 0.6484375, "learning_rate": 9.947944654900432e-05, "loss": 0.1773, "step": 4487 }, { "epoch": 3.2879120879120878, "grad_norm": 1.0546875, "learning_rate": 9.947895082923486e-05, "loss": 0.1362, "step": 4488 }, { "epoch": 3.2886446886446885, "grad_norm": 0.91015625, "learning_rate": 9.947845487479113e-05, "loss": 0.1582, "step": 4489 }, { "epoch": 3.2893772893772892, "grad_norm": 1.1484375, "learning_rate": 9.947795868567556e-05, "loss": 0.1893, "step": 4490 }, { "epoch": 3.29010989010989, "grad_norm": 1.09375, "learning_rate": 9.947746226189052e-05, "loss": 0.1657, "step": 4491 }, { "epoch": 3.2908424908424907, "grad_norm": 1.234375, "learning_rate": 9.947696560343836e-05, "loss": 0.1691, "step": 4492 }, { "epoch": 3.2915750915750914, "grad_norm": 0.6171875, "learning_rate": 9.947646871032149e-05, "loss": 0.1488, "step": 4493 }, { "epoch": 3.292307692307692, "grad_norm": 1.0625, "learning_rate": 9.947597158254229e-05, "loss": 0.1828, "step": 4494 }, { "epoch": 3.293040293040293, "grad_norm": 0.79296875, "learning_rate": 9.947547422010313e-05, "loss": 0.1508, "step": 4495 }, { "epoch": 3.2937728937728936, "grad_norm": 1.046875, "learning_rate": 9.947497662300639e-05, "loss": 0.1425, "step": 4496 }, { "epoch": 3.2945054945054943, "grad_norm": 1.0390625, "learning_rate": 9.947447879125445e-05, "loss": 0.1545, "step": 4497 }, { "epoch": 3.295238095238095, "grad_norm": 1.0859375, "learning_rate": 9.947398072484972e-05, "loss": 0.1404, "step": 4498 }, { "epoch": 3.295970695970696, "grad_norm": 1.328125, "learning_rate": 9.947348242379456e-05, "loss": 0.2024, "step": 4499 }, { "epoch": 3.2967032967032965, "grad_norm": 1.0625, "learning_rate": 9.947298388809139e-05, "loss": 0.1307, "step": 4500 }, { "epoch": 3.2974358974358973, "grad_norm": 1.03125, "learning_rate": 9.947248511774255e-05, "loss": 0.1335, "step": 4501 }, { "epoch": 3.298168498168498, "grad_norm": 0.8984375, "learning_rate": 9.947198611275048e-05, "loss": 0.1925, "step": 4502 }, { "epoch": 3.2989010989010987, "grad_norm": 1.3671875, "learning_rate": 9.947148687311753e-05, "loss": 0.1889, "step": 4503 }, { "epoch": 3.2996336996336995, "grad_norm": 1.1484375, "learning_rate": 9.947098739884611e-05, "loss": 0.1953, "step": 4504 }, { "epoch": 3.3003663003663, "grad_norm": 0.7890625, "learning_rate": 9.947048768993861e-05, "loss": 0.1696, "step": 4505 }, { "epoch": 3.301098901098901, "grad_norm": 1.2265625, "learning_rate": 9.946998774639746e-05, "loss": 0.1983, "step": 4506 }, { "epoch": 3.3018315018315016, "grad_norm": 1.734375, "learning_rate": 9.946948756822499e-05, "loss": 0.1568, "step": 4507 }, { "epoch": 3.3025641025641024, "grad_norm": 1.46875, "learning_rate": 9.946898715542365e-05, "loss": 0.2186, "step": 4508 }, { "epoch": 3.303296703296703, "grad_norm": 0.9140625, "learning_rate": 9.94684865079958e-05, "loss": 0.1465, "step": 4509 }, { "epoch": 3.304029304029304, "grad_norm": 1.0546875, "learning_rate": 9.946798562594389e-05, "loss": 0.1616, "step": 4510 }, { "epoch": 3.3047619047619046, "grad_norm": 1.1640625, "learning_rate": 9.946748450927024e-05, "loss": 0.1788, "step": 4511 }, { "epoch": 3.3054945054945053, "grad_norm": 1.140625, "learning_rate": 9.946698315797732e-05, "loss": 0.1657, "step": 4512 }, { "epoch": 3.306227106227106, "grad_norm": 0.91796875, "learning_rate": 9.946648157206749e-05, "loss": 0.1286, "step": 4513 }, { "epoch": 3.3069597069597068, "grad_norm": 1.3125, "learning_rate": 9.94659797515432e-05, "loss": 0.1752, "step": 4514 }, { "epoch": 3.3076923076923075, "grad_norm": 1.15625, "learning_rate": 9.94654776964068e-05, "loss": 0.1534, "step": 4515 }, { "epoch": 3.308424908424908, "grad_norm": 1.1953125, "learning_rate": 9.946497540666075e-05, "loss": 0.1659, "step": 4516 }, { "epoch": 3.309157509157509, "grad_norm": 0.890625, "learning_rate": 9.946447288230742e-05, "loss": 0.1355, "step": 4517 }, { "epoch": 3.3098901098901097, "grad_norm": 1.5390625, "learning_rate": 9.946397012334921e-05, "loss": 0.1541, "step": 4518 }, { "epoch": 3.3106227106227104, "grad_norm": 1.8828125, "learning_rate": 9.946346712978855e-05, "loss": 0.1905, "step": 4519 }, { "epoch": 3.311355311355311, "grad_norm": 1.0859375, "learning_rate": 9.946296390162786e-05, "loss": 0.1706, "step": 4520 }, { "epoch": 3.312087912087912, "grad_norm": 1.140625, "learning_rate": 9.946246043886953e-05, "loss": 0.1874, "step": 4521 }, { "epoch": 3.3128205128205126, "grad_norm": 0.875, "learning_rate": 9.946195674151597e-05, "loss": 0.1413, "step": 4522 }, { "epoch": 3.3135531135531133, "grad_norm": 1.1796875, "learning_rate": 9.946145280956963e-05, "loss": 0.154, "step": 4523 }, { "epoch": 3.314285714285714, "grad_norm": 1.40625, "learning_rate": 9.946094864303285e-05, "loss": 0.1935, "step": 4524 }, { "epoch": 3.315018315018315, "grad_norm": 1.1640625, "learning_rate": 9.946044424190812e-05, "loss": 0.1881, "step": 4525 }, { "epoch": 3.3157509157509155, "grad_norm": 0.9609375, "learning_rate": 9.945993960619784e-05, "loss": 0.1539, "step": 4526 }, { "epoch": 3.3164835164835162, "grad_norm": 1.0546875, "learning_rate": 9.945943473590439e-05, "loss": 0.1335, "step": 4527 }, { "epoch": 3.317216117216117, "grad_norm": 1.0546875, "learning_rate": 9.945892963103025e-05, "loss": 0.1764, "step": 4528 }, { "epoch": 3.3179487179487177, "grad_norm": 1.3125, "learning_rate": 9.945842429157778e-05, "loss": 0.1846, "step": 4529 }, { "epoch": 3.3186813186813184, "grad_norm": 0.99609375, "learning_rate": 9.945791871754943e-05, "loss": 0.1868, "step": 4530 }, { "epoch": 3.319413919413919, "grad_norm": 1.0703125, "learning_rate": 9.945741290894761e-05, "loss": 0.171, "step": 4531 }, { "epoch": 3.32014652014652, "grad_norm": 1.7890625, "learning_rate": 9.945690686577477e-05, "loss": 0.1492, "step": 4532 }, { "epoch": 3.320879120879121, "grad_norm": 1.6015625, "learning_rate": 9.945640058803332e-05, "loss": 0.171, "step": 4533 }, { "epoch": 3.321611721611722, "grad_norm": 1.234375, "learning_rate": 9.945589407572566e-05, "loss": 0.1678, "step": 4534 }, { "epoch": 3.3223443223443225, "grad_norm": 1.8359375, "learning_rate": 9.945538732885425e-05, "loss": 0.1609, "step": 4535 }, { "epoch": 3.3230769230769233, "grad_norm": 1.0078125, "learning_rate": 9.945488034742152e-05, "loss": 0.2018, "step": 4536 }, { "epoch": 3.323809523809524, "grad_norm": 0.9140625, "learning_rate": 9.945437313142989e-05, "loss": 0.1559, "step": 4537 }, { "epoch": 3.3245421245421247, "grad_norm": 1.3359375, "learning_rate": 9.945386568088178e-05, "loss": 0.2112, "step": 4538 }, { "epoch": 3.3252747252747255, "grad_norm": 1.75, "learning_rate": 9.945335799577963e-05, "loss": 0.1777, "step": 4539 }, { "epoch": 3.326007326007326, "grad_norm": 1.1875, "learning_rate": 9.945285007612586e-05, "loss": 0.1577, "step": 4540 }, { "epoch": 3.326739926739927, "grad_norm": 1.1328125, "learning_rate": 9.945234192192293e-05, "loss": 0.1752, "step": 4541 }, { "epoch": 3.3274725274725276, "grad_norm": 1.640625, "learning_rate": 9.945183353317325e-05, "loss": 0.1863, "step": 4542 }, { "epoch": 3.3282051282051284, "grad_norm": 0.8984375, "learning_rate": 9.945132490987928e-05, "loss": 0.1217, "step": 4543 }, { "epoch": 3.328937728937729, "grad_norm": 0.89453125, "learning_rate": 9.945081605204344e-05, "loss": 0.1644, "step": 4544 }, { "epoch": 3.32967032967033, "grad_norm": 0.78515625, "learning_rate": 9.945030695966817e-05, "loss": 0.1555, "step": 4545 }, { "epoch": 3.3304029304029306, "grad_norm": 1.2734375, "learning_rate": 9.944979763275591e-05, "loss": 0.143, "step": 4546 }, { "epoch": 3.3311355311355313, "grad_norm": 1.9375, "learning_rate": 9.944928807130908e-05, "loss": 0.2025, "step": 4547 }, { "epoch": 3.331868131868132, "grad_norm": 1.203125, "learning_rate": 9.944877827533016e-05, "loss": 0.1519, "step": 4548 }, { "epoch": 3.3326007326007328, "grad_norm": 1.1953125, "learning_rate": 9.944826824482157e-05, "loss": 0.1643, "step": 4549 }, { "epoch": 3.3333333333333335, "grad_norm": 0.9453125, "learning_rate": 9.944775797978575e-05, "loss": 0.1502, "step": 4550 }, { "epoch": 3.334065934065934, "grad_norm": 1.3203125, "learning_rate": 9.944724748022515e-05, "loss": 0.1846, "step": 4551 }, { "epoch": 3.334798534798535, "grad_norm": 0.8671875, "learning_rate": 9.944673674614225e-05, "loss": 0.1498, "step": 4552 }, { "epoch": 3.3355311355311357, "grad_norm": 0.7265625, "learning_rate": 9.944622577753944e-05, "loss": 0.1612, "step": 4553 }, { "epoch": 3.3362637362637364, "grad_norm": 1.3671875, "learning_rate": 9.94457145744192e-05, "loss": 0.1439, "step": 4554 }, { "epoch": 3.336996336996337, "grad_norm": 1.7109375, "learning_rate": 9.944520313678399e-05, "loss": 0.1799, "step": 4555 }, { "epoch": 3.337728937728938, "grad_norm": 1.7890625, "learning_rate": 9.944469146463622e-05, "loss": 0.1822, "step": 4556 }, { "epoch": 3.3384615384615386, "grad_norm": 1.4140625, "learning_rate": 9.944417955797838e-05, "loss": 0.169, "step": 4557 }, { "epoch": 3.3391941391941393, "grad_norm": 0.6875, "learning_rate": 9.94436674168129e-05, "loss": 0.1304, "step": 4558 }, { "epoch": 3.33992673992674, "grad_norm": 0.87890625, "learning_rate": 9.944315504114225e-05, "loss": 0.1358, "step": 4559 }, { "epoch": 3.340659340659341, "grad_norm": 0.8046875, "learning_rate": 9.944264243096887e-05, "loss": 0.1455, "step": 4560 }, { "epoch": 3.3413919413919415, "grad_norm": 1.09375, "learning_rate": 9.944212958629523e-05, "loss": 0.1902, "step": 4561 }, { "epoch": 3.3421245421245422, "grad_norm": 0.8671875, "learning_rate": 9.944161650712378e-05, "loss": 0.1674, "step": 4562 }, { "epoch": 3.342857142857143, "grad_norm": 1.484375, "learning_rate": 9.944110319345697e-05, "loss": 0.1878, "step": 4563 }, { "epoch": 3.3435897435897437, "grad_norm": 0.88671875, "learning_rate": 9.944058964529728e-05, "loss": 0.1238, "step": 4564 }, { "epoch": 3.3443223443223444, "grad_norm": 1.0859375, "learning_rate": 9.944007586264715e-05, "loss": 0.1735, "step": 4565 }, { "epoch": 3.345054945054945, "grad_norm": 1.0390625, "learning_rate": 9.943956184550905e-05, "loss": 0.2087, "step": 4566 }, { "epoch": 3.345787545787546, "grad_norm": 0.87109375, "learning_rate": 9.943904759388545e-05, "loss": 0.1534, "step": 4567 }, { "epoch": 3.3465201465201466, "grad_norm": 1.0703125, "learning_rate": 9.94385331077788e-05, "loss": 0.1506, "step": 4568 }, { "epoch": 3.3472527472527474, "grad_norm": 0.765625, "learning_rate": 9.943801838719158e-05, "loss": 0.1582, "step": 4569 }, { "epoch": 3.347985347985348, "grad_norm": 1.3984375, "learning_rate": 9.943750343212624e-05, "loss": 0.1365, "step": 4570 }, { "epoch": 3.348717948717949, "grad_norm": 1.25, "learning_rate": 9.943698824258527e-05, "loss": 0.191, "step": 4571 }, { "epoch": 3.3494505494505495, "grad_norm": 0.65625, "learning_rate": 9.943647281857111e-05, "loss": 0.1805, "step": 4572 }, { "epoch": 3.3501831501831503, "grad_norm": 1.2734375, "learning_rate": 9.943595716008624e-05, "loss": 0.1833, "step": 4573 }, { "epoch": 3.350915750915751, "grad_norm": 1.96875, "learning_rate": 9.943544126713314e-05, "loss": 0.1562, "step": 4574 }, { "epoch": 3.3516483516483517, "grad_norm": 1.5859375, "learning_rate": 9.943492513971429e-05, "loss": 0.1502, "step": 4575 }, { "epoch": 3.3523809523809525, "grad_norm": 1.7734375, "learning_rate": 9.943440877783215e-05, "loss": 0.1599, "step": 4576 }, { "epoch": 3.353113553113553, "grad_norm": 0.68359375, "learning_rate": 9.943389218148919e-05, "loss": 0.193, "step": 4577 }, { "epoch": 3.353846153846154, "grad_norm": 1.7734375, "learning_rate": 9.943337535068788e-05, "loss": 0.1548, "step": 4578 }, { "epoch": 3.3545787545787547, "grad_norm": 1.6640625, "learning_rate": 9.94328582854307e-05, "loss": 0.1836, "step": 4579 }, { "epoch": 3.3553113553113554, "grad_norm": 1.7734375, "learning_rate": 9.943234098572013e-05, "loss": 0.1699, "step": 4580 }, { "epoch": 3.356043956043956, "grad_norm": 2.03125, "learning_rate": 9.943182345155867e-05, "loss": 0.2137, "step": 4581 }, { "epoch": 3.356776556776557, "grad_norm": 2.0625, "learning_rate": 9.943130568294877e-05, "loss": 0.1632, "step": 4582 }, { "epoch": 3.3575091575091576, "grad_norm": 0.9140625, "learning_rate": 9.943078767989292e-05, "loss": 0.158, "step": 4583 }, { "epoch": 3.3582417582417583, "grad_norm": 1.328125, "learning_rate": 9.943026944239361e-05, "loss": 0.1747, "step": 4584 }, { "epoch": 3.358974358974359, "grad_norm": 1.40625, "learning_rate": 9.94297509704533e-05, "loss": 0.1515, "step": 4585 }, { "epoch": 3.3597069597069598, "grad_norm": 0.8203125, "learning_rate": 9.942923226407451e-05, "loss": 0.1464, "step": 4586 }, { "epoch": 3.3604395604395605, "grad_norm": 0.671875, "learning_rate": 9.94287133232597e-05, "loss": 0.1519, "step": 4587 }, { "epoch": 3.361172161172161, "grad_norm": 0.6484375, "learning_rate": 9.942819414801136e-05, "loss": 0.1525, "step": 4588 }, { "epoch": 3.361904761904762, "grad_norm": 1.359375, "learning_rate": 9.942767473833199e-05, "loss": 0.1773, "step": 4589 }, { "epoch": 3.3626373626373627, "grad_norm": 0.859375, "learning_rate": 9.942715509422404e-05, "loss": 0.1695, "step": 4590 }, { "epoch": 3.3633699633699634, "grad_norm": 0.98046875, "learning_rate": 9.942663521569006e-05, "loss": 0.1659, "step": 4591 }, { "epoch": 3.364102564102564, "grad_norm": 0.9453125, "learning_rate": 9.94261151027325e-05, "loss": 0.1462, "step": 4592 }, { "epoch": 3.364835164835165, "grad_norm": 1.078125, "learning_rate": 9.942559475535384e-05, "loss": 0.1649, "step": 4593 }, { "epoch": 3.3655677655677656, "grad_norm": 0.89453125, "learning_rate": 9.942507417355661e-05, "loss": 0.1826, "step": 4594 }, { "epoch": 3.3663003663003663, "grad_norm": 1.0625, "learning_rate": 9.942455335734328e-05, "loss": 0.157, "step": 4595 }, { "epoch": 3.367032967032967, "grad_norm": 0.7109375, "learning_rate": 9.942403230671637e-05, "loss": 0.1308, "step": 4596 }, { "epoch": 3.367765567765568, "grad_norm": 0.75, "learning_rate": 9.942351102167836e-05, "loss": 0.1637, "step": 4597 }, { "epoch": 3.3684981684981685, "grad_norm": 0.6171875, "learning_rate": 9.942298950223175e-05, "loss": 0.168, "step": 4598 }, { "epoch": 3.3692307692307693, "grad_norm": 1.1484375, "learning_rate": 9.942246774837902e-05, "loss": 0.1932, "step": 4599 }, { "epoch": 3.36996336996337, "grad_norm": 1.1171875, "learning_rate": 9.942194576012269e-05, "loss": 0.2006, "step": 4600 }, { "epoch": 3.3706959706959707, "grad_norm": 0.98046875, "learning_rate": 9.942142353746528e-05, "loss": 0.1395, "step": 4601 }, { "epoch": 3.3714285714285714, "grad_norm": 1.078125, "learning_rate": 9.942090108040926e-05, "loss": 0.2064, "step": 4602 }, { "epoch": 3.372161172161172, "grad_norm": 0.91796875, "learning_rate": 9.942037838895712e-05, "loss": 0.1894, "step": 4603 }, { "epoch": 3.372893772893773, "grad_norm": 1.3828125, "learning_rate": 9.94198554631114e-05, "loss": 0.2108, "step": 4604 }, { "epoch": 3.3736263736263736, "grad_norm": 1.21875, "learning_rate": 9.941933230287461e-05, "loss": 0.154, "step": 4605 }, { "epoch": 3.3743589743589744, "grad_norm": 1.234375, "learning_rate": 9.941880890824921e-05, "loss": 0.1871, "step": 4606 }, { "epoch": 3.375091575091575, "grad_norm": 0.67578125, "learning_rate": 9.941828527923778e-05, "loss": 0.1234, "step": 4607 }, { "epoch": 3.375824175824176, "grad_norm": 0.75, "learning_rate": 9.941776141584276e-05, "loss": 0.1451, "step": 4608 }, { "epoch": 3.3765567765567766, "grad_norm": 1.2109375, "learning_rate": 9.941723731806667e-05, "loss": 0.1443, "step": 4609 }, { "epoch": 3.3772893772893773, "grad_norm": 0.97265625, "learning_rate": 9.941671298591208e-05, "loss": 0.1336, "step": 4610 }, { "epoch": 3.378021978021978, "grad_norm": 1.5703125, "learning_rate": 9.941618841938143e-05, "loss": 0.1685, "step": 4611 }, { "epoch": 3.3787545787545787, "grad_norm": 0.9453125, "learning_rate": 9.941566361847726e-05, "loss": 0.1853, "step": 4612 }, { "epoch": 3.3794871794871795, "grad_norm": 0.62109375, "learning_rate": 9.941513858320209e-05, "loss": 0.1806, "step": 4613 }, { "epoch": 3.38021978021978, "grad_norm": 1.28125, "learning_rate": 9.941461331355846e-05, "loss": 0.1662, "step": 4614 }, { "epoch": 3.380952380952381, "grad_norm": 1.3984375, "learning_rate": 9.941408780954882e-05, "loss": 0.1801, "step": 4615 }, { "epoch": 3.3816849816849817, "grad_norm": 0.87890625, "learning_rate": 9.941356207117575e-05, "loss": 0.1632, "step": 4616 }, { "epoch": 3.3824175824175824, "grad_norm": 0.5625, "learning_rate": 9.941303609844176e-05, "loss": 0.1169, "step": 4617 }, { "epoch": 3.383150183150183, "grad_norm": 0.73828125, "learning_rate": 9.941250989134932e-05, "loss": 0.1967, "step": 4618 }, { "epoch": 3.383882783882784, "grad_norm": 0.92578125, "learning_rate": 9.941198344990102e-05, "loss": 0.188, "step": 4619 }, { "epoch": 3.3846153846153846, "grad_norm": 1.125, "learning_rate": 9.941145677409934e-05, "loss": 0.1395, "step": 4620 }, { "epoch": 3.3853479853479853, "grad_norm": 0.76171875, "learning_rate": 9.941092986394681e-05, "loss": 0.144, "step": 4621 }, { "epoch": 3.386080586080586, "grad_norm": 0.72265625, "learning_rate": 9.941040271944598e-05, "loss": 0.1476, "step": 4622 }, { "epoch": 3.3868131868131868, "grad_norm": 1.109375, "learning_rate": 9.940987534059932e-05, "loss": 0.1854, "step": 4623 }, { "epoch": 3.3875457875457875, "grad_norm": 1.1015625, "learning_rate": 9.940934772740942e-05, "loss": 0.1846, "step": 4624 }, { "epoch": 3.3882783882783882, "grad_norm": 1.875, "learning_rate": 9.940881987987877e-05, "loss": 0.1897, "step": 4625 }, { "epoch": 3.389010989010989, "grad_norm": 1.5078125, "learning_rate": 9.94082917980099e-05, "loss": 0.151, "step": 4626 }, { "epoch": 3.3897435897435897, "grad_norm": 0.6171875, "learning_rate": 9.940776348180534e-05, "loss": 0.1649, "step": 4627 }, { "epoch": 3.3904761904761904, "grad_norm": 0.9296875, "learning_rate": 9.940723493126766e-05, "loss": 0.1356, "step": 4628 }, { "epoch": 3.391208791208791, "grad_norm": 1.1484375, "learning_rate": 9.940670614639933e-05, "loss": 0.1413, "step": 4629 }, { "epoch": 3.391941391941392, "grad_norm": 1.9921875, "learning_rate": 9.940617712720294e-05, "loss": 0.1791, "step": 4630 }, { "epoch": 3.3926739926739926, "grad_norm": 1.6328125, "learning_rate": 9.9405647873681e-05, "loss": 0.1947, "step": 4631 }, { "epoch": 3.3934065934065933, "grad_norm": 0.93359375, "learning_rate": 9.940511838583603e-05, "loss": 0.164, "step": 4632 }, { "epoch": 3.394139194139194, "grad_norm": 0.93359375, "learning_rate": 9.940458866367058e-05, "loss": 0.1686, "step": 4633 }, { "epoch": 3.394871794871795, "grad_norm": 0.671875, "learning_rate": 9.940405870718722e-05, "loss": 0.1606, "step": 4634 }, { "epoch": 3.3956043956043955, "grad_norm": 1.5078125, "learning_rate": 9.940352851638843e-05, "loss": 0.154, "step": 4635 }, { "epoch": 3.3963369963369963, "grad_norm": 2.09375, "learning_rate": 9.940299809127682e-05, "loss": 0.1687, "step": 4636 }, { "epoch": 3.397069597069597, "grad_norm": 1.34375, "learning_rate": 9.940246743185487e-05, "loss": 0.1522, "step": 4637 }, { "epoch": 3.3978021978021977, "grad_norm": 1.046875, "learning_rate": 9.940193653812512e-05, "loss": 0.1649, "step": 4638 }, { "epoch": 3.3985347985347985, "grad_norm": 0.90234375, "learning_rate": 9.940140541009016e-05, "loss": 0.2215, "step": 4639 }, { "epoch": 3.399267399267399, "grad_norm": 1.2890625, "learning_rate": 9.940087404775251e-05, "loss": 0.181, "step": 4640 }, { "epoch": 3.4, "grad_norm": 0.953125, "learning_rate": 9.940034245111472e-05, "loss": 0.1798, "step": 4641 }, { "epoch": 3.4007326007326006, "grad_norm": 1.34375, "learning_rate": 9.939981062017935e-05, "loss": 0.1889, "step": 4642 }, { "epoch": 3.4014652014652014, "grad_norm": 0.80078125, "learning_rate": 9.939927855494892e-05, "loss": 0.1512, "step": 4643 }, { "epoch": 3.402197802197802, "grad_norm": 0.9453125, "learning_rate": 9.9398746255426e-05, "loss": 0.1421, "step": 4644 }, { "epoch": 3.402930402930403, "grad_norm": 1.1171875, "learning_rate": 9.939821372161313e-05, "loss": 0.1732, "step": 4645 }, { "epoch": 3.4036630036630036, "grad_norm": 1.390625, "learning_rate": 9.939768095351287e-05, "loss": 0.1504, "step": 4646 }, { "epoch": 3.4043956043956043, "grad_norm": 1.5625, "learning_rate": 9.939714795112777e-05, "loss": 0.1775, "step": 4647 }, { "epoch": 3.405128205128205, "grad_norm": 0.91015625, "learning_rate": 9.939661471446037e-05, "loss": 0.1936, "step": 4648 }, { "epoch": 3.4058608058608058, "grad_norm": 1.1171875, "learning_rate": 9.939608124351325e-05, "loss": 0.1841, "step": 4649 }, { "epoch": 3.4065934065934065, "grad_norm": 0.98046875, "learning_rate": 9.939554753828895e-05, "loss": 0.1671, "step": 4650 }, { "epoch": 3.407326007326007, "grad_norm": 1.3125, "learning_rate": 9.939501359879003e-05, "loss": 0.1628, "step": 4651 }, { "epoch": 3.408058608058608, "grad_norm": 1.2109375, "learning_rate": 9.939447942501904e-05, "loss": 0.1587, "step": 4652 }, { "epoch": 3.4087912087912087, "grad_norm": 1.203125, "learning_rate": 9.939394501697857e-05, "loss": 0.148, "step": 4653 }, { "epoch": 3.4095238095238094, "grad_norm": 1.4453125, "learning_rate": 9.939341037467115e-05, "loss": 0.1524, "step": 4654 }, { "epoch": 3.41025641025641, "grad_norm": 1.5546875, "learning_rate": 9.939287549809934e-05, "loss": 0.196, "step": 4655 }, { "epoch": 3.410989010989011, "grad_norm": 1.5078125, "learning_rate": 9.939234038726571e-05, "loss": 0.2168, "step": 4656 }, { "epoch": 3.4117216117216116, "grad_norm": 1.2265625, "learning_rate": 9.939180504217284e-05, "loss": 0.1948, "step": 4657 }, { "epoch": 3.4124542124542123, "grad_norm": 1.265625, "learning_rate": 9.939126946282328e-05, "loss": 0.2059, "step": 4658 }, { "epoch": 3.413186813186813, "grad_norm": 1.546875, "learning_rate": 9.939073364921959e-05, "loss": 0.1675, "step": 4659 }, { "epoch": 3.413919413919414, "grad_norm": 1.34375, "learning_rate": 9.939019760136437e-05, "loss": 0.17, "step": 4660 }, { "epoch": 3.4146520146520145, "grad_norm": 1.5, "learning_rate": 9.938966131926014e-05, "loss": 0.1762, "step": 4661 }, { "epoch": 3.4153846153846152, "grad_norm": 1.4453125, "learning_rate": 9.93891248029095e-05, "loss": 0.1838, "step": 4662 }, { "epoch": 3.416117216117216, "grad_norm": 1.28125, "learning_rate": 9.938858805231499e-05, "loss": 0.1805, "step": 4663 }, { "epoch": 3.4168498168498167, "grad_norm": 1.65625, "learning_rate": 9.938805106747924e-05, "loss": 0.1769, "step": 4664 }, { "epoch": 3.4175824175824174, "grad_norm": 1.4609375, "learning_rate": 9.938751384840477e-05, "loss": 0.1728, "step": 4665 }, { "epoch": 3.418315018315018, "grad_norm": 1.421875, "learning_rate": 9.938697639509417e-05, "loss": 0.1606, "step": 4666 }, { "epoch": 3.419047619047619, "grad_norm": 1.15625, "learning_rate": 9.938643870755004e-05, "loss": 0.2233, "step": 4667 }, { "epoch": 3.4197802197802196, "grad_norm": 1.71875, "learning_rate": 9.938590078577492e-05, "loss": 0.1589, "step": 4668 }, { "epoch": 3.4205128205128204, "grad_norm": 1.3046875, "learning_rate": 9.938536262977141e-05, "loss": 0.1714, "step": 4669 }, { "epoch": 3.421245421245421, "grad_norm": 1.59375, "learning_rate": 9.938482423954207e-05, "loss": 0.1617, "step": 4670 }, { "epoch": 3.421978021978022, "grad_norm": 1.7890625, "learning_rate": 9.938428561508948e-05, "loss": 0.1866, "step": 4671 }, { "epoch": 3.4227106227106225, "grad_norm": 0.77734375, "learning_rate": 9.938374675641623e-05, "loss": 0.1382, "step": 4672 }, { "epoch": 3.4234432234432233, "grad_norm": 1.21875, "learning_rate": 9.938320766352492e-05, "loss": 0.1308, "step": 4673 }, { "epoch": 3.424175824175824, "grad_norm": 2.921875, "learning_rate": 9.93826683364181e-05, "loss": 0.1729, "step": 4674 }, { "epoch": 3.4249084249084247, "grad_norm": 1.6328125, "learning_rate": 9.938212877509836e-05, "loss": 0.1346, "step": 4675 }, { "epoch": 3.4256410256410255, "grad_norm": 0.87890625, "learning_rate": 9.938158897956831e-05, "loss": 0.1472, "step": 4676 }, { "epoch": 3.426373626373626, "grad_norm": 0.78125, "learning_rate": 9.938104894983053e-05, "loss": 0.146, "step": 4677 }, { "epoch": 3.427106227106227, "grad_norm": 0.94921875, "learning_rate": 9.938050868588759e-05, "loss": 0.1544, "step": 4678 }, { "epoch": 3.4278388278388277, "grad_norm": 1.3359375, "learning_rate": 9.937996818774208e-05, "loss": 0.1451, "step": 4679 }, { "epoch": 3.4285714285714284, "grad_norm": 0.97265625, "learning_rate": 9.93794274553966e-05, "loss": 0.1708, "step": 4680 }, { "epoch": 3.429304029304029, "grad_norm": 1.2421875, "learning_rate": 9.937888648885373e-05, "loss": 0.232, "step": 4681 }, { "epoch": 3.4300366300366303, "grad_norm": 1.3671875, "learning_rate": 9.937834528811609e-05, "loss": 0.224, "step": 4682 }, { "epoch": 3.430769230769231, "grad_norm": 0.84375, "learning_rate": 9.937780385318623e-05, "loss": 0.157, "step": 4683 }, { "epoch": 3.4315018315018317, "grad_norm": 1.1015625, "learning_rate": 9.937726218406678e-05, "loss": 0.1455, "step": 4684 }, { "epoch": 3.4322344322344325, "grad_norm": 0.71484375, "learning_rate": 9.937672028076032e-05, "loss": 0.1615, "step": 4685 }, { "epoch": 3.432967032967033, "grad_norm": 0.890625, "learning_rate": 9.937617814326944e-05, "loss": 0.1512, "step": 4686 }, { "epoch": 3.433699633699634, "grad_norm": 0.83203125, "learning_rate": 9.937563577159675e-05, "loss": 0.1248, "step": 4687 }, { "epoch": 3.4344322344322347, "grad_norm": 0.90625, "learning_rate": 9.937509316574484e-05, "loss": 0.131, "step": 4688 }, { "epoch": 3.4351648351648354, "grad_norm": 1.0703125, "learning_rate": 9.937455032571631e-05, "loss": 0.1478, "step": 4689 }, { "epoch": 3.435897435897436, "grad_norm": 1.2421875, "learning_rate": 9.937400725151376e-05, "loss": 0.1528, "step": 4690 }, { "epoch": 3.436630036630037, "grad_norm": 1.0703125, "learning_rate": 9.937346394313982e-05, "loss": 0.1316, "step": 4691 }, { "epoch": 3.4373626373626376, "grad_norm": 1.359375, "learning_rate": 9.937292040059706e-05, "loss": 0.2257, "step": 4692 }, { "epoch": 3.4380952380952383, "grad_norm": 1.0546875, "learning_rate": 9.937237662388809e-05, "loss": 0.1312, "step": 4693 }, { "epoch": 3.438827838827839, "grad_norm": 1.359375, "learning_rate": 9.937183261301553e-05, "loss": 0.1508, "step": 4694 }, { "epoch": 3.4395604395604398, "grad_norm": 1.5703125, "learning_rate": 9.937128836798198e-05, "loss": 0.2096, "step": 4695 }, { "epoch": 3.4402930402930405, "grad_norm": 1.2421875, "learning_rate": 9.937074388879002e-05, "loss": 0.1526, "step": 4696 }, { "epoch": 3.4410256410256412, "grad_norm": 1.59375, "learning_rate": 9.93701991754423e-05, "loss": 0.1775, "step": 4697 }, { "epoch": 3.441758241758242, "grad_norm": 1.5859375, "learning_rate": 9.936965422794141e-05, "loss": 0.2347, "step": 4698 }, { "epoch": 3.4424908424908427, "grad_norm": 0.87890625, "learning_rate": 9.936910904628996e-05, "loss": 0.1393, "step": 4699 }, { "epoch": 3.4432234432234434, "grad_norm": 1.2421875, "learning_rate": 9.936856363049057e-05, "loss": 0.1659, "step": 4700 }, { "epoch": 3.443956043956044, "grad_norm": 1.5859375, "learning_rate": 9.936801798054585e-05, "loss": 0.1749, "step": 4701 }, { "epoch": 3.444688644688645, "grad_norm": 0.8125, "learning_rate": 9.936747209645841e-05, "loss": 0.1286, "step": 4702 }, { "epoch": 3.4454212454212456, "grad_norm": 0.8828125, "learning_rate": 9.936692597823086e-05, "loss": 0.16, "step": 4703 }, { "epoch": 3.4461538461538463, "grad_norm": 1.7421875, "learning_rate": 9.936637962586586e-05, "loss": 0.1644, "step": 4704 }, { "epoch": 3.446886446886447, "grad_norm": 1.078125, "learning_rate": 9.936583303936596e-05, "loss": 0.1649, "step": 4705 }, { "epoch": 3.447619047619048, "grad_norm": 1.1953125, "learning_rate": 9.936528621873382e-05, "loss": 0.1596, "step": 4706 }, { "epoch": 3.4483516483516485, "grad_norm": 1.3671875, "learning_rate": 9.936473916397205e-05, "loss": 0.2038, "step": 4707 }, { "epoch": 3.4490842490842493, "grad_norm": 0.99609375, "learning_rate": 9.936419187508328e-05, "loss": 0.1374, "step": 4708 }, { "epoch": 3.44981684981685, "grad_norm": 1.0234375, "learning_rate": 9.936364435207012e-05, "loss": 0.159, "step": 4709 }, { "epoch": 3.4505494505494507, "grad_norm": 1.1328125, "learning_rate": 9.93630965949352e-05, "loss": 0.1669, "step": 4710 }, { "epoch": 3.4512820512820515, "grad_norm": 1.3671875, "learning_rate": 9.936254860368115e-05, "loss": 0.1932, "step": 4711 }, { "epoch": 3.452014652014652, "grad_norm": 1.1484375, "learning_rate": 9.936200037831058e-05, "loss": 0.1633, "step": 4712 }, { "epoch": 3.452747252747253, "grad_norm": 0.9453125, "learning_rate": 9.936145191882613e-05, "loss": 0.1647, "step": 4713 }, { "epoch": 3.4534798534798536, "grad_norm": 1.265625, "learning_rate": 9.936090322523042e-05, "loss": 0.1337, "step": 4714 }, { "epoch": 3.4542124542124544, "grad_norm": 1.1484375, "learning_rate": 9.936035429752609e-05, "loss": 0.1908, "step": 4715 }, { "epoch": 3.454945054945055, "grad_norm": 1.0859375, "learning_rate": 9.935980513571576e-05, "loss": 0.1349, "step": 4716 }, { "epoch": 3.455677655677656, "grad_norm": 1.7734375, "learning_rate": 9.935925573980207e-05, "loss": 0.1599, "step": 4717 }, { "epoch": 3.4564102564102566, "grad_norm": 1.5390625, "learning_rate": 9.935870610978763e-05, "loss": 0.185, "step": 4718 }, { "epoch": 3.4571428571428573, "grad_norm": 1.4609375, "learning_rate": 9.935815624567511e-05, "loss": 0.1666, "step": 4719 }, { "epoch": 3.457875457875458, "grad_norm": 0.875, "learning_rate": 9.935760614746712e-05, "loss": 0.1382, "step": 4720 }, { "epoch": 3.4586080586080588, "grad_norm": 1.03125, "learning_rate": 9.93570558151663e-05, "loss": 0.1573, "step": 4721 }, { "epoch": 3.4593406593406595, "grad_norm": 0.99609375, "learning_rate": 9.935650524877526e-05, "loss": 0.138, "step": 4722 }, { "epoch": 3.46007326007326, "grad_norm": 0.875, "learning_rate": 9.93559544482967e-05, "loss": 0.1567, "step": 4723 }, { "epoch": 3.460805860805861, "grad_norm": 0.98046875, "learning_rate": 9.93554034137332e-05, "loss": 0.2211, "step": 4724 }, { "epoch": 3.4615384615384617, "grad_norm": 1.2109375, "learning_rate": 9.935485214508746e-05, "loss": 0.1807, "step": 4725 }, { "epoch": 3.4622710622710624, "grad_norm": 1.203125, "learning_rate": 9.935430064236204e-05, "loss": 0.145, "step": 4726 }, { "epoch": 3.463003663003663, "grad_norm": 0.94140625, "learning_rate": 9.935374890555965e-05, "loss": 0.168, "step": 4727 }, { "epoch": 3.463736263736264, "grad_norm": 1.5234375, "learning_rate": 9.935319693468291e-05, "loss": 0.1598, "step": 4728 }, { "epoch": 3.4644688644688646, "grad_norm": 0.7890625, "learning_rate": 9.935264472973445e-05, "loss": 0.1716, "step": 4729 }, { "epoch": 3.4652014652014653, "grad_norm": 0.98046875, "learning_rate": 9.935209229071696e-05, "loss": 0.156, "step": 4730 }, { "epoch": 3.465934065934066, "grad_norm": 1.6875, "learning_rate": 9.935153961763304e-05, "loss": 0.1632, "step": 4731 }, { "epoch": 3.466666666666667, "grad_norm": 1.359375, "learning_rate": 9.935098671048536e-05, "loss": 0.1721, "step": 4732 }, { "epoch": 3.4673992673992675, "grad_norm": 0.75, "learning_rate": 9.935043356927655e-05, "loss": 0.1384, "step": 4733 }, { "epoch": 3.4681318681318682, "grad_norm": 1.09375, "learning_rate": 9.93498801940093e-05, "loss": 0.1603, "step": 4734 }, { "epoch": 3.468864468864469, "grad_norm": 0.78125, "learning_rate": 9.934932658468624e-05, "loss": 0.146, "step": 4735 }, { "epoch": 3.4695970695970697, "grad_norm": 0.515625, "learning_rate": 9.934877274131001e-05, "loss": 0.1625, "step": 4736 }, { "epoch": 3.4703296703296704, "grad_norm": 1.2734375, "learning_rate": 9.934821866388329e-05, "loss": 0.1697, "step": 4737 }, { "epoch": 3.471062271062271, "grad_norm": 0.66796875, "learning_rate": 9.93476643524087e-05, "loss": 0.1392, "step": 4738 }, { "epoch": 3.471794871794872, "grad_norm": 1.2578125, "learning_rate": 9.934710980688892e-05, "loss": 0.1156, "step": 4739 }, { "epoch": 3.4725274725274726, "grad_norm": 1.765625, "learning_rate": 9.93465550273266e-05, "loss": 0.1672, "step": 4740 }, { "epoch": 3.4732600732600734, "grad_norm": 1.09375, "learning_rate": 9.93460000137244e-05, "loss": 0.1359, "step": 4741 }, { "epoch": 3.473992673992674, "grad_norm": 1.0, "learning_rate": 9.934544476608497e-05, "loss": 0.1701, "step": 4742 }, { "epoch": 3.474725274725275, "grad_norm": 0.91796875, "learning_rate": 9.934488928441099e-05, "loss": 0.1641, "step": 4743 }, { "epoch": 3.4754578754578755, "grad_norm": 2.40625, "learning_rate": 9.93443335687051e-05, "loss": 0.1846, "step": 4744 }, { "epoch": 3.4761904761904763, "grad_norm": 2.078125, "learning_rate": 9.934377761897e-05, "loss": 0.1483, "step": 4745 }, { "epoch": 3.476923076923077, "grad_norm": 1.125, "learning_rate": 9.93432214352083e-05, "loss": 0.1408, "step": 4746 }, { "epoch": 3.4776556776556777, "grad_norm": 0.95703125, "learning_rate": 9.934266501742269e-05, "loss": 0.1649, "step": 4747 }, { "epoch": 3.4783882783882785, "grad_norm": 1.265625, "learning_rate": 9.934210836561586e-05, "loss": 0.1421, "step": 4748 }, { "epoch": 3.479120879120879, "grad_norm": 1.8671875, "learning_rate": 9.934155147979045e-05, "loss": 0.171, "step": 4749 }, { "epoch": 3.47985347985348, "grad_norm": 1.6328125, "learning_rate": 9.934099435994912e-05, "loss": 0.1732, "step": 4750 }, { "epoch": 3.4805860805860807, "grad_norm": 1.0390625, "learning_rate": 9.934043700609457e-05, "loss": 0.1847, "step": 4751 }, { "epoch": 3.4813186813186814, "grad_norm": 0.6640625, "learning_rate": 9.933987941822943e-05, "loss": 0.1343, "step": 4752 }, { "epoch": 3.482051282051282, "grad_norm": 1.2734375, "learning_rate": 9.933932159635642e-05, "loss": 0.1935, "step": 4753 }, { "epoch": 3.482783882783883, "grad_norm": 0.9140625, "learning_rate": 9.933876354047816e-05, "loss": 0.1439, "step": 4754 }, { "epoch": 3.4835164835164836, "grad_norm": 1.6328125, "learning_rate": 9.933820525059738e-05, "loss": 0.1972, "step": 4755 }, { "epoch": 3.4842490842490843, "grad_norm": 1.0859375, "learning_rate": 9.933764672671671e-05, "loss": 0.1214, "step": 4756 }, { "epoch": 3.484981684981685, "grad_norm": 1.2109375, "learning_rate": 9.933708796883884e-05, "loss": 0.1722, "step": 4757 }, { "epoch": 3.4857142857142858, "grad_norm": 0.95703125, "learning_rate": 9.933652897696645e-05, "loss": 0.1786, "step": 4758 }, { "epoch": 3.4864468864468865, "grad_norm": 2.421875, "learning_rate": 9.933596975110221e-05, "loss": 0.2018, "step": 4759 }, { "epoch": 3.4871794871794872, "grad_norm": 1.6796875, "learning_rate": 9.933541029124883e-05, "loss": 0.1611, "step": 4760 }, { "epoch": 3.487912087912088, "grad_norm": 1.59375, "learning_rate": 9.933485059740896e-05, "loss": 0.1896, "step": 4761 }, { "epoch": 3.4886446886446887, "grad_norm": 0.73828125, "learning_rate": 9.933429066958528e-05, "loss": 0.1615, "step": 4762 }, { "epoch": 3.4893772893772894, "grad_norm": 1.015625, "learning_rate": 9.933373050778048e-05, "loss": 0.1893, "step": 4763 }, { "epoch": 3.49010989010989, "grad_norm": 0.9140625, "learning_rate": 9.933317011199726e-05, "loss": 0.1199, "step": 4764 }, { "epoch": 3.490842490842491, "grad_norm": 1.203125, "learning_rate": 9.933260948223828e-05, "loss": 0.1575, "step": 4765 }, { "epoch": 3.4915750915750916, "grad_norm": 0.9609375, "learning_rate": 9.933204861850625e-05, "loss": 0.161, "step": 4766 }, { "epoch": 3.4923076923076923, "grad_norm": 0.9609375, "learning_rate": 9.933148752080384e-05, "loss": 0.1805, "step": 4767 }, { "epoch": 3.493040293040293, "grad_norm": 1.4609375, "learning_rate": 9.933092618913373e-05, "loss": 0.1708, "step": 4768 }, { "epoch": 3.493772893772894, "grad_norm": 0.61328125, "learning_rate": 9.933036462349863e-05, "loss": 0.1378, "step": 4769 }, { "epoch": 3.4945054945054945, "grad_norm": 1.234375, "learning_rate": 9.932980282390123e-05, "loss": 0.1891, "step": 4770 }, { "epoch": 3.4952380952380953, "grad_norm": 0.99609375, "learning_rate": 9.932924079034422e-05, "loss": 0.2039, "step": 4771 }, { "epoch": 3.495970695970696, "grad_norm": 0.8359375, "learning_rate": 9.932867852283027e-05, "loss": 0.1748, "step": 4772 }, { "epoch": 3.4967032967032967, "grad_norm": 0.87109375, "learning_rate": 9.932811602136211e-05, "loss": 0.1956, "step": 4773 }, { "epoch": 3.4974358974358974, "grad_norm": 1.1015625, "learning_rate": 9.93275532859424e-05, "loss": 0.1996, "step": 4774 }, { "epoch": 3.498168498168498, "grad_norm": 0.83984375, "learning_rate": 9.932699031657385e-05, "loss": 0.1385, "step": 4775 }, { "epoch": 3.498901098901099, "grad_norm": 0.75, "learning_rate": 9.932642711325917e-05, "loss": 0.1468, "step": 4776 }, { "epoch": 3.4996336996336996, "grad_norm": 1.1328125, "learning_rate": 9.932586367600105e-05, "loss": 0.1742, "step": 4777 }, { "epoch": 3.5003663003663004, "grad_norm": 1.1171875, "learning_rate": 9.932530000480218e-05, "loss": 0.1374, "step": 4778 }, { "epoch": 3.501098901098901, "grad_norm": 0.76953125, "learning_rate": 9.932473609966528e-05, "loss": 0.138, "step": 4779 }, { "epoch": 3.501831501831502, "grad_norm": 1.0546875, "learning_rate": 9.932417196059305e-05, "loss": 0.1493, "step": 4780 }, { "epoch": 3.5025641025641026, "grad_norm": 1.0546875, "learning_rate": 9.932360758758816e-05, "loss": 0.2005, "step": 4781 }, { "epoch": 3.5032967032967033, "grad_norm": 1.0703125, "learning_rate": 9.932304298065336e-05, "loss": 0.2217, "step": 4782 }, { "epoch": 3.504029304029304, "grad_norm": 1.109375, "learning_rate": 9.932247813979132e-05, "loss": 0.1263, "step": 4783 }, { "epoch": 3.5047619047619047, "grad_norm": 1.0625, "learning_rate": 9.932191306500477e-05, "loss": 0.1472, "step": 4784 }, { "epoch": 3.5054945054945055, "grad_norm": 1.1171875, "learning_rate": 9.93213477562964e-05, "loss": 0.124, "step": 4785 }, { "epoch": 3.506227106227106, "grad_norm": 0.58984375, "learning_rate": 9.932078221366892e-05, "loss": 0.1356, "step": 4786 }, { "epoch": 3.506959706959707, "grad_norm": 0.65625, "learning_rate": 9.932021643712506e-05, "loss": 0.1395, "step": 4787 }, { "epoch": 3.5076923076923077, "grad_norm": 1.375, "learning_rate": 9.93196504266675e-05, "loss": 0.1342, "step": 4788 }, { "epoch": 3.5084249084249084, "grad_norm": 0.9375, "learning_rate": 9.931908418229898e-05, "loss": 0.167, "step": 4789 }, { "epoch": 3.509157509157509, "grad_norm": 0.7890625, "learning_rate": 9.93185177040222e-05, "loss": 0.1538, "step": 4790 }, { "epoch": 3.50989010989011, "grad_norm": 0.92578125, "learning_rate": 9.931795099183987e-05, "loss": 0.1206, "step": 4791 }, { "epoch": 3.5106227106227106, "grad_norm": 1.109375, "learning_rate": 9.931738404575471e-05, "loss": 0.1721, "step": 4792 }, { "epoch": 3.5113553113553113, "grad_norm": 0.73046875, "learning_rate": 9.931681686576945e-05, "loss": 0.1379, "step": 4793 }, { "epoch": 3.512087912087912, "grad_norm": 1.484375, "learning_rate": 9.93162494518868e-05, "loss": 0.1812, "step": 4794 }, { "epoch": 3.5128205128205128, "grad_norm": 0.91796875, "learning_rate": 9.931568180410946e-05, "loss": 0.1468, "step": 4795 }, { "epoch": 3.5135531135531135, "grad_norm": 0.8125, "learning_rate": 9.931511392244016e-05, "loss": 0.1805, "step": 4796 }, { "epoch": 3.5142857142857142, "grad_norm": 0.8515625, "learning_rate": 9.931454580688163e-05, "loss": 0.2046, "step": 4797 }, { "epoch": 3.515018315018315, "grad_norm": 0.8359375, "learning_rate": 9.931397745743658e-05, "loss": 0.1732, "step": 4798 }, { "epoch": 3.5157509157509157, "grad_norm": 1.1328125, "learning_rate": 9.931340887410774e-05, "loss": 0.1633, "step": 4799 }, { "epoch": 3.5164835164835164, "grad_norm": 1.203125, "learning_rate": 9.931284005689784e-05, "loss": 0.1946, "step": 4800 }, { "epoch": 3.517216117216117, "grad_norm": 0.8125, "learning_rate": 9.93122710058096e-05, "loss": 0.1755, "step": 4801 }, { "epoch": 3.517948717948718, "grad_norm": 1.203125, "learning_rate": 9.931170172084575e-05, "loss": 0.1426, "step": 4802 }, { "epoch": 3.5186813186813186, "grad_norm": 0.91796875, "learning_rate": 9.931113220200901e-05, "loss": 0.1511, "step": 4803 }, { "epoch": 3.5194139194139193, "grad_norm": 0.890625, "learning_rate": 9.93105624493021e-05, "loss": 0.135, "step": 4804 }, { "epoch": 3.52014652014652, "grad_norm": 0.83203125, "learning_rate": 9.930999246272777e-05, "loss": 0.163, "step": 4805 }, { "epoch": 3.520879120879121, "grad_norm": 0.87890625, "learning_rate": 9.930942224228873e-05, "loss": 0.1572, "step": 4806 }, { "epoch": 3.5216117216117215, "grad_norm": 1.1015625, "learning_rate": 9.930885178798775e-05, "loss": 0.1788, "step": 4807 }, { "epoch": 3.5223443223443223, "grad_norm": 0.93359375, "learning_rate": 9.930828109982754e-05, "loss": 0.1433, "step": 4808 }, { "epoch": 3.523076923076923, "grad_norm": 1.203125, "learning_rate": 9.930771017781081e-05, "loss": 0.1372, "step": 4809 }, { "epoch": 3.5238095238095237, "grad_norm": 1.421875, "learning_rate": 9.930713902194033e-05, "loss": 0.2164, "step": 4810 }, { "epoch": 3.5245421245421245, "grad_norm": 1.015625, "learning_rate": 9.930656763221882e-05, "loss": 0.1697, "step": 4811 }, { "epoch": 3.525274725274725, "grad_norm": 1.0234375, "learning_rate": 9.930599600864901e-05, "loss": 0.1516, "step": 4812 }, { "epoch": 3.526007326007326, "grad_norm": 0.9609375, "learning_rate": 9.930542415123368e-05, "loss": 0.178, "step": 4813 }, { "epoch": 3.5267399267399266, "grad_norm": 0.79296875, "learning_rate": 9.930485205997552e-05, "loss": 0.1538, "step": 4814 }, { "epoch": 3.5274725274725274, "grad_norm": 0.8984375, "learning_rate": 9.93042797348773e-05, "loss": 0.1809, "step": 4815 }, { "epoch": 3.528205128205128, "grad_norm": 0.671875, "learning_rate": 9.930370717594176e-05, "loss": 0.1485, "step": 4816 }, { "epoch": 3.528937728937729, "grad_norm": 0.796875, "learning_rate": 9.930313438317162e-05, "loss": 0.145, "step": 4817 }, { "epoch": 3.5296703296703296, "grad_norm": 0.4765625, "learning_rate": 9.930256135656965e-05, "loss": 0.1373, "step": 4818 }, { "epoch": 3.5304029304029303, "grad_norm": 0.70703125, "learning_rate": 9.930198809613859e-05, "loss": 0.173, "step": 4819 }, { "epoch": 3.531135531135531, "grad_norm": 0.734375, "learning_rate": 9.930141460188119e-05, "loss": 0.1556, "step": 4820 }, { "epoch": 3.5318681318681318, "grad_norm": 1.0546875, "learning_rate": 9.930084087380018e-05, "loss": 0.1948, "step": 4821 }, { "epoch": 3.5326007326007325, "grad_norm": 0.80078125, "learning_rate": 9.930026691189833e-05, "loss": 0.1647, "step": 4822 }, { "epoch": 3.533333333333333, "grad_norm": 1.0703125, "learning_rate": 9.929969271617838e-05, "loss": 0.1867, "step": 4823 }, { "epoch": 3.534065934065934, "grad_norm": 0.83984375, "learning_rate": 9.929911828664308e-05, "loss": 0.1142, "step": 4824 }, { "epoch": 3.5347985347985347, "grad_norm": 1.1328125, "learning_rate": 9.929854362329518e-05, "loss": 0.1947, "step": 4825 }, { "epoch": 3.5355311355311354, "grad_norm": 1.109375, "learning_rate": 9.929796872613744e-05, "loss": 0.1631, "step": 4826 }, { "epoch": 3.536263736263736, "grad_norm": 1.0078125, "learning_rate": 9.929739359517263e-05, "loss": 0.1409, "step": 4827 }, { "epoch": 3.536996336996337, "grad_norm": 1.3125, "learning_rate": 9.929681823040347e-05, "loss": 0.1637, "step": 4828 }, { "epoch": 3.5377289377289376, "grad_norm": 1.390625, "learning_rate": 9.929624263183274e-05, "loss": 0.2127, "step": 4829 }, { "epoch": 3.5384615384615383, "grad_norm": 1.546875, "learning_rate": 9.92956667994632e-05, "loss": 0.1973, "step": 4830 }, { "epoch": 3.539194139194139, "grad_norm": 1.7734375, "learning_rate": 9.929509073329757e-05, "loss": 0.1678, "step": 4831 }, { "epoch": 3.53992673992674, "grad_norm": 1.0234375, "learning_rate": 9.929451443333868e-05, "loss": 0.1887, "step": 4832 }, { "epoch": 3.5406593406593405, "grad_norm": 1.2890625, "learning_rate": 9.929393789958924e-05, "loss": 0.1383, "step": 4833 }, { "epoch": 3.5413919413919412, "grad_norm": 1.0234375, "learning_rate": 9.929336113205204e-05, "loss": 0.1832, "step": 4834 }, { "epoch": 3.542124542124542, "grad_norm": 1.21875, "learning_rate": 9.92927841307298e-05, "loss": 0.2247, "step": 4835 }, { "epoch": 3.5428571428571427, "grad_norm": 1.2265625, "learning_rate": 9.929220689562533e-05, "loss": 0.1564, "step": 4836 }, { "epoch": 3.5435897435897434, "grad_norm": 1.5390625, "learning_rate": 9.929162942674137e-05, "loss": 0.2018, "step": 4837 }, { "epoch": 3.544322344322344, "grad_norm": 1.0078125, "learning_rate": 9.92910517240807e-05, "loss": 0.1301, "step": 4838 }, { "epoch": 3.545054945054945, "grad_norm": 1.0546875, "learning_rate": 9.929047378764609e-05, "loss": 0.1415, "step": 4839 }, { "epoch": 3.5457875457875456, "grad_norm": 0.62890625, "learning_rate": 9.928989561744031e-05, "loss": 0.1352, "step": 4840 }, { "epoch": 3.5465201465201464, "grad_norm": 0.8671875, "learning_rate": 9.928931721346611e-05, "loss": 0.1444, "step": 4841 }, { "epoch": 3.547252747252747, "grad_norm": 0.91015625, "learning_rate": 9.928873857572627e-05, "loss": 0.1592, "step": 4842 }, { "epoch": 3.547985347985348, "grad_norm": 0.91015625, "learning_rate": 9.928815970422358e-05, "loss": 0.1412, "step": 4843 }, { "epoch": 3.5487179487179485, "grad_norm": 0.76171875, "learning_rate": 9.92875805989608e-05, "loss": 0.1421, "step": 4844 }, { "epoch": 3.5494505494505493, "grad_norm": 0.6953125, "learning_rate": 9.92870012599407e-05, "loss": 0.1692, "step": 4845 }, { "epoch": 3.55018315018315, "grad_norm": 0.94140625, "learning_rate": 9.928642168716606e-05, "loss": 0.1336, "step": 4846 }, { "epoch": 3.5509157509157507, "grad_norm": 1.1484375, "learning_rate": 9.928584188063965e-05, "loss": 0.1819, "step": 4847 }, { "epoch": 3.5516483516483515, "grad_norm": 0.75390625, "learning_rate": 9.928526184036427e-05, "loss": 0.1363, "step": 4848 }, { "epoch": 3.552380952380952, "grad_norm": 1.0390625, "learning_rate": 9.928468156634269e-05, "loss": 0.1556, "step": 4849 }, { "epoch": 3.553113553113553, "grad_norm": 1.09375, "learning_rate": 9.928410105857768e-05, "loss": 0.1641, "step": 4850 }, { "epoch": 3.5538461538461537, "grad_norm": 0.67578125, "learning_rate": 9.928352031707203e-05, "loss": 0.1633, "step": 4851 }, { "epoch": 3.5545787545787544, "grad_norm": 0.94921875, "learning_rate": 9.928293934182851e-05, "loss": 0.1358, "step": 4852 }, { "epoch": 3.555311355311355, "grad_norm": 1.1484375, "learning_rate": 9.928235813284992e-05, "loss": 0.1402, "step": 4853 }, { "epoch": 3.556043956043956, "grad_norm": 0.6328125, "learning_rate": 9.928177669013903e-05, "loss": 0.1397, "step": 4854 }, { "epoch": 3.5567765567765566, "grad_norm": 1.1875, "learning_rate": 9.928119501369863e-05, "loss": 0.1419, "step": 4855 }, { "epoch": 3.5575091575091573, "grad_norm": 1.5390625, "learning_rate": 9.928061310353152e-05, "loss": 0.1724, "step": 4856 }, { "epoch": 3.558241758241758, "grad_norm": 0.84375, "learning_rate": 9.928003095964049e-05, "loss": 0.1717, "step": 4857 }, { "epoch": 3.5589743589743588, "grad_norm": 0.9765625, "learning_rate": 9.92794485820283e-05, "loss": 0.148, "step": 4858 }, { "epoch": 3.5597069597069595, "grad_norm": 0.77734375, "learning_rate": 9.927886597069777e-05, "loss": 0.1224, "step": 4859 }, { "epoch": 3.5604395604395602, "grad_norm": 1.3046875, "learning_rate": 9.927828312565167e-05, "loss": 0.1898, "step": 4860 }, { "epoch": 3.561172161172161, "grad_norm": 1.0234375, "learning_rate": 9.927770004689283e-05, "loss": 0.1523, "step": 4861 }, { "epoch": 3.5619047619047617, "grad_norm": 0.8984375, "learning_rate": 9.927711673442399e-05, "loss": 0.1485, "step": 4862 }, { "epoch": 3.5626373626373624, "grad_norm": 1.09375, "learning_rate": 9.927653318824796e-05, "loss": 0.1792, "step": 4863 }, { "epoch": 3.563369963369963, "grad_norm": 1.125, "learning_rate": 9.927594940836757e-05, "loss": 0.156, "step": 4864 }, { "epoch": 3.564102564102564, "grad_norm": 1.40625, "learning_rate": 9.92753653947856e-05, "loss": 0.1408, "step": 4865 }, { "epoch": 3.5648351648351646, "grad_norm": 1.546875, "learning_rate": 9.927478114750482e-05, "loss": 0.1581, "step": 4866 }, { "epoch": 3.5655677655677653, "grad_norm": 1.1484375, "learning_rate": 9.927419666652806e-05, "loss": 0.1752, "step": 4867 }, { "epoch": 3.566300366300366, "grad_norm": 0.83984375, "learning_rate": 9.927361195185812e-05, "loss": 0.1555, "step": 4868 }, { "epoch": 3.567032967032967, "grad_norm": 0.93359375, "learning_rate": 9.92730270034978e-05, "loss": 0.1611, "step": 4869 }, { "epoch": 3.5677655677655675, "grad_norm": 0.89453125, "learning_rate": 9.927244182144988e-05, "loss": 0.1385, "step": 4870 }, { "epoch": 3.5684981684981683, "grad_norm": 0.9453125, "learning_rate": 9.927185640571718e-05, "loss": 0.1693, "step": 4871 }, { "epoch": 3.569230769230769, "grad_norm": 1.0546875, "learning_rate": 9.927127075630252e-05, "loss": 0.2204, "step": 4872 }, { "epoch": 3.5699633699633697, "grad_norm": 1.0, "learning_rate": 9.927068487320867e-05, "loss": 0.1772, "step": 4873 }, { "epoch": 3.5706959706959704, "grad_norm": 0.96484375, "learning_rate": 9.927009875643848e-05, "loss": 0.1402, "step": 4874 }, { "epoch": 3.571428571428571, "grad_norm": 0.9921875, "learning_rate": 9.926951240599471e-05, "loss": 0.177, "step": 4875 }, { "epoch": 3.572161172161172, "grad_norm": 1.3203125, "learning_rate": 9.926892582188022e-05, "loss": 0.1873, "step": 4876 }, { "epoch": 3.5728937728937726, "grad_norm": 0.9921875, "learning_rate": 9.92683390040978e-05, "loss": 0.174, "step": 4877 }, { "epoch": 3.5736263736263734, "grad_norm": 0.7265625, "learning_rate": 9.926775195265025e-05, "loss": 0.1325, "step": 4878 }, { "epoch": 3.574358974358974, "grad_norm": 0.73046875, "learning_rate": 9.926716466754038e-05, "loss": 0.1447, "step": 4879 }, { "epoch": 3.575091575091575, "grad_norm": 1.296875, "learning_rate": 9.926657714877103e-05, "loss": 0.1904, "step": 4880 }, { "epoch": 3.5758241758241756, "grad_norm": 1.828125, "learning_rate": 9.926598939634498e-05, "loss": 0.1758, "step": 4881 }, { "epoch": 3.5765567765567763, "grad_norm": 0.984375, "learning_rate": 9.926540141026509e-05, "loss": 0.1769, "step": 4882 }, { "epoch": 3.577289377289377, "grad_norm": 0.96484375, "learning_rate": 9.926481319053415e-05, "loss": 0.1498, "step": 4883 }, { "epoch": 3.5780219780219777, "grad_norm": 1.125, "learning_rate": 9.926422473715497e-05, "loss": 0.1622, "step": 4884 }, { "epoch": 3.578754578754579, "grad_norm": 1.28125, "learning_rate": 9.926363605013039e-05, "loss": 0.1562, "step": 4885 }, { "epoch": 3.5794871794871796, "grad_norm": 0.86328125, "learning_rate": 9.926304712946322e-05, "loss": 0.1436, "step": 4886 }, { "epoch": 3.5802197802197804, "grad_norm": 1.1796875, "learning_rate": 9.926245797515628e-05, "loss": 0.1725, "step": 4887 }, { "epoch": 3.580952380952381, "grad_norm": 1.4765625, "learning_rate": 9.926186858721239e-05, "loss": 0.1816, "step": 4888 }, { "epoch": 3.581684981684982, "grad_norm": 1.9375, "learning_rate": 9.926127896563439e-05, "loss": 0.1633, "step": 4889 }, { "epoch": 3.5824175824175826, "grad_norm": 1.7734375, "learning_rate": 9.926068911042509e-05, "loss": 0.2005, "step": 4890 }, { "epoch": 3.5831501831501833, "grad_norm": 1.546875, "learning_rate": 9.926009902158731e-05, "loss": 0.183, "step": 4891 }, { "epoch": 3.583882783882784, "grad_norm": 2.6875, "learning_rate": 9.92595086991239e-05, "loss": 0.211, "step": 4892 }, { "epoch": 3.5846153846153848, "grad_norm": 1.703125, "learning_rate": 9.925891814303769e-05, "loss": 0.1532, "step": 4893 }, { "epoch": 3.5853479853479855, "grad_norm": 2.25, "learning_rate": 9.92583273533315e-05, "loss": 0.204, "step": 4894 }, { "epoch": 3.586080586080586, "grad_norm": 2.046875, "learning_rate": 9.925773633000813e-05, "loss": 0.122, "step": 4895 }, { "epoch": 3.586813186813187, "grad_norm": 1.3203125, "learning_rate": 9.925714507307046e-05, "loss": 0.1394, "step": 4896 }, { "epoch": 3.5875457875457877, "grad_norm": 0.77734375, "learning_rate": 9.925655358252128e-05, "loss": 0.1503, "step": 4897 }, { "epoch": 3.5882783882783884, "grad_norm": 1.3515625, "learning_rate": 9.925596185836346e-05, "loss": 0.1493, "step": 4898 }, { "epoch": 3.589010989010989, "grad_norm": 1.71875, "learning_rate": 9.925536990059984e-05, "loss": 0.1948, "step": 4899 }, { "epoch": 3.58974358974359, "grad_norm": 1.0625, "learning_rate": 9.92547777092332e-05, "loss": 0.1788, "step": 4900 }, { "epoch": 3.5904761904761906, "grad_norm": 1.0390625, "learning_rate": 9.925418528426644e-05, "loss": 0.1757, "step": 4901 }, { "epoch": 3.5912087912087913, "grad_norm": 0.90625, "learning_rate": 9.925359262570237e-05, "loss": 0.1712, "step": 4902 }, { "epoch": 3.591941391941392, "grad_norm": 1.34375, "learning_rate": 9.925299973354382e-05, "loss": 0.1479, "step": 4903 }, { "epoch": 3.592673992673993, "grad_norm": 1.5625, "learning_rate": 9.925240660779366e-05, "loss": 0.1948, "step": 4904 }, { "epoch": 3.5934065934065935, "grad_norm": 1.6640625, "learning_rate": 9.925181324845472e-05, "loss": 0.1456, "step": 4905 }, { "epoch": 3.5941391941391942, "grad_norm": 1.5390625, "learning_rate": 9.925121965552982e-05, "loss": 0.1818, "step": 4906 }, { "epoch": 3.594871794871795, "grad_norm": 0.79296875, "learning_rate": 9.925062582902184e-05, "loss": 0.1501, "step": 4907 }, { "epoch": 3.5956043956043957, "grad_norm": 1.265625, "learning_rate": 9.92500317689336e-05, "loss": 0.1456, "step": 4908 }, { "epoch": 3.5963369963369964, "grad_norm": 2.03125, "learning_rate": 9.924943747526795e-05, "loss": 0.2045, "step": 4909 }, { "epoch": 3.597069597069597, "grad_norm": 1.1015625, "learning_rate": 9.924884294802773e-05, "loss": 0.1369, "step": 4910 }, { "epoch": 3.597802197802198, "grad_norm": 0.7734375, "learning_rate": 9.924824818721582e-05, "loss": 0.1614, "step": 4911 }, { "epoch": 3.5985347985347986, "grad_norm": 0.85546875, "learning_rate": 9.924765319283505e-05, "loss": 0.1489, "step": 4912 }, { "epoch": 3.5992673992673994, "grad_norm": 0.76953125, "learning_rate": 9.924705796488825e-05, "loss": 0.1669, "step": 4913 }, { "epoch": 3.6, "grad_norm": 1.7421875, "learning_rate": 9.924646250337832e-05, "loss": 0.1734, "step": 4914 }, { "epoch": 3.600732600732601, "grad_norm": 1.8203125, "learning_rate": 9.924586680830808e-05, "loss": 0.1606, "step": 4915 }, { "epoch": 3.6014652014652015, "grad_norm": 0.62890625, "learning_rate": 9.924527087968037e-05, "loss": 0.1287, "step": 4916 }, { "epoch": 3.6021978021978023, "grad_norm": 0.93359375, "learning_rate": 9.924467471749809e-05, "loss": 0.1537, "step": 4917 }, { "epoch": 3.602930402930403, "grad_norm": 0.79296875, "learning_rate": 9.924407832176407e-05, "loss": 0.1453, "step": 4918 }, { "epoch": 3.6036630036630037, "grad_norm": 1.53125, "learning_rate": 9.924348169248115e-05, "loss": 0.1332, "step": 4919 }, { "epoch": 3.6043956043956045, "grad_norm": 0.671875, "learning_rate": 9.924288482965221e-05, "loss": 0.1161, "step": 4920 }, { "epoch": 3.605128205128205, "grad_norm": 1.0546875, "learning_rate": 9.924228773328011e-05, "loss": 0.1829, "step": 4921 }, { "epoch": 3.605860805860806, "grad_norm": 0.8515625, "learning_rate": 9.924169040336773e-05, "loss": 0.1858, "step": 4922 }, { "epoch": 3.6065934065934067, "grad_norm": 0.6171875, "learning_rate": 9.924109283991788e-05, "loss": 0.1528, "step": 4923 }, { "epoch": 3.6073260073260074, "grad_norm": 1.046875, "learning_rate": 9.924049504293348e-05, "loss": 0.1464, "step": 4924 }, { "epoch": 3.608058608058608, "grad_norm": 1.1875, "learning_rate": 9.923989701241735e-05, "loss": 0.1667, "step": 4925 }, { "epoch": 3.608791208791209, "grad_norm": 0.82421875, "learning_rate": 9.923929874837238e-05, "loss": 0.1766, "step": 4926 }, { "epoch": 3.6095238095238096, "grad_norm": 1.0703125, "learning_rate": 9.923870025080144e-05, "loss": 0.2037, "step": 4927 }, { "epoch": 3.6102564102564103, "grad_norm": 1.2109375, "learning_rate": 9.923810151970736e-05, "loss": 0.1564, "step": 4928 }, { "epoch": 3.610989010989011, "grad_norm": 0.73046875, "learning_rate": 9.923750255509305e-05, "loss": 0.1556, "step": 4929 }, { "epoch": 3.6117216117216118, "grad_norm": 1.3125, "learning_rate": 9.923690335696136e-05, "loss": 0.1693, "step": 4930 }, { "epoch": 3.6124542124542125, "grad_norm": 1.1484375, "learning_rate": 9.923630392531518e-05, "loss": 0.1589, "step": 4931 }, { "epoch": 3.6131868131868132, "grad_norm": 1.3203125, "learning_rate": 9.923570426015736e-05, "loss": 0.1695, "step": 4932 }, { "epoch": 3.613919413919414, "grad_norm": 0.8984375, "learning_rate": 9.92351043614908e-05, "loss": 0.2138, "step": 4933 }, { "epoch": 3.6146520146520147, "grad_norm": 0.82421875, "learning_rate": 9.923450422931834e-05, "loss": 0.1319, "step": 4934 }, { "epoch": 3.6153846153846154, "grad_norm": 0.7109375, "learning_rate": 9.923390386364286e-05, "loss": 0.1598, "step": 4935 }, { "epoch": 3.616117216117216, "grad_norm": 1.71875, "learning_rate": 9.923330326446725e-05, "loss": 0.1593, "step": 4936 }, { "epoch": 3.616849816849817, "grad_norm": 1.2265625, "learning_rate": 9.92327024317944e-05, "loss": 0.1553, "step": 4937 }, { "epoch": 3.6175824175824176, "grad_norm": 0.81640625, "learning_rate": 9.923210136562718e-05, "loss": 0.1767, "step": 4938 }, { "epoch": 3.6183150183150183, "grad_norm": 1.46875, "learning_rate": 9.923150006596845e-05, "loss": 0.159, "step": 4939 }, { "epoch": 3.619047619047619, "grad_norm": 1.3515625, "learning_rate": 9.923089853282112e-05, "loss": 0.162, "step": 4940 }, { "epoch": 3.61978021978022, "grad_norm": 1.0703125, "learning_rate": 9.923029676618805e-05, "loss": 0.1588, "step": 4941 }, { "epoch": 3.6205128205128205, "grad_norm": 1.015625, "learning_rate": 9.922969476607213e-05, "loss": 0.1509, "step": 4942 }, { "epoch": 3.6212454212454213, "grad_norm": 1.640625, "learning_rate": 9.922909253247625e-05, "loss": 0.1762, "step": 4943 }, { "epoch": 3.621978021978022, "grad_norm": 1.6796875, "learning_rate": 9.922849006540328e-05, "loss": 0.1753, "step": 4944 }, { "epoch": 3.6227106227106227, "grad_norm": 1.1484375, "learning_rate": 9.922788736485613e-05, "loss": 0.1367, "step": 4945 }, { "epoch": 3.6234432234432234, "grad_norm": 0.7421875, "learning_rate": 9.922728443083768e-05, "loss": 0.1868, "step": 4946 }, { "epoch": 3.624175824175824, "grad_norm": 1.15625, "learning_rate": 9.92266812633508e-05, "loss": 0.1656, "step": 4947 }, { "epoch": 3.624908424908425, "grad_norm": 0.87890625, "learning_rate": 9.922607786239839e-05, "loss": 0.1273, "step": 4948 }, { "epoch": 3.6256410256410256, "grad_norm": 0.796875, "learning_rate": 9.922547422798335e-05, "loss": 0.1562, "step": 4949 }, { "epoch": 3.6263736263736264, "grad_norm": 1.2265625, "learning_rate": 9.922487036010858e-05, "loss": 0.1724, "step": 4950 }, { "epoch": 3.627106227106227, "grad_norm": 0.734375, "learning_rate": 9.922426625877695e-05, "loss": 0.1572, "step": 4951 }, { "epoch": 3.627838827838828, "grad_norm": 0.87890625, "learning_rate": 9.922366192399136e-05, "loss": 0.1689, "step": 4952 }, { "epoch": 3.6285714285714286, "grad_norm": 0.84765625, "learning_rate": 9.922305735575472e-05, "loss": 0.1801, "step": 4953 }, { "epoch": 3.6293040293040293, "grad_norm": 1.1328125, "learning_rate": 9.922245255406992e-05, "loss": 0.1605, "step": 4954 }, { "epoch": 3.63003663003663, "grad_norm": 1.90625, "learning_rate": 9.922184751893985e-05, "loss": 0.1132, "step": 4955 }, { "epoch": 3.6307692307692307, "grad_norm": 1.53125, "learning_rate": 9.922124225036742e-05, "loss": 0.1378, "step": 4956 }, { "epoch": 3.6315018315018315, "grad_norm": 1.2109375, "learning_rate": 9.922063674835551e-05, "loss": 0.1389, "step": 4957 }, { "epoch": 3.632234432234432, "grad_norm": 1.359375, "learning_rate": 9.922003101290705e-05, "loss": 0.1588, "step": 4958 }, { "epoch": 3.632967032967033, "grad_norm": 0.94921875, "learning_rate": 9.921942504402492e-05, "loss": 0.1497, "step": 4959 }, { "epoch": 3.6336996336996337, "grad_norm": 0.68359375, "learning_rate": 9.921881884171203e-05, "loss": 0.1377, "step": 4960 }, { "epoch": 3.6344322344322344, "grad_norm": 1.0625, "learning_rate": 9.92182124059713e-05, "loss": 0.1541, "step": 4961 }, { "epoch": 3.635164835164835, "grad_norm": 0.95703125, "learning_rate": 9.92176057368056e-05, "loss": 0.202, "step": 4962 }, { "epoch": 3.635897435897436, "grad_norm": 0.7890625, "learning_rate": 9.921699883421787e-05, "loss": 0.1389, "step": 4963 }, { "epoch": 3.6366300366300366, "grad_norm": 0.77734375, "learning_rate": 9.9216391698211e-05, "loss": 0.1181, "step": 4964 }, { "epoch": 3.6373626373626373, "grad_norm": 1.1171875, "learning_rate": 9.921578432878792e-05, "loss": 0.1919, "step": 4965 }, { "epoch": 3.638095238095238, "grad_norm": 0.6640625, "learning_rate": 9.921517672595149e-05, "loss": 0.1469, "step": 4966 }, { "epoch": 3.638827838827839, "grad_norm": 0.96484375, "learning_rate": 9.92145688897047e-05, "loss": 0.1348, "step": 4967 }, { "epoch": 3.6395604395604395, "grad_norm": 0.70703125, "learning_rate": 9.92139608200504e-05, "loss": 0.1568, "step": 4968 }, { "epoch": 3.6402930402930402, "grad_norm": 1.1328125, "learning_rate": 9.921335251699152e-05, "loss": 0.1763, "step": 4969 }, { "epoch": 3.641025641025641, "grad_norm": 1.1953125, "learning_rate": 9.921274398053096e-05, "loss": 0.1526, "step": 4970 }, { "epoch": 3.6417582417582417, "grad_norm": 1.1640625, "learning_rate": 9.921213521067168e-05, "loss": 0.1687, "step": 4971 }, { "epoch": 3.6424908424908424, "grad_norm": 1.296875, "learning_rate": 9.921152620741656e-05, "loss": 0.1932, "step": 4972 }, { "epoch": 3.643223443223443, "grad_norm": 0.875, "learning_rate": 9.921091697076852e-05, "loss": 0.1711, "step": 4973 }, { "epoch": 3.643956043956044, "grad_norm": 0.9140625, "learning_rate": 9.921030750073049e-05, "loss": 0.1583, "step": 4974 }, { "epoch": 3.6446886446886446, "grad_norm": 0.92578125, "learning_rate": 9.92096977973054e-05, "loss": 0.2018, "step": 4975 }, { "epoch": 3.6454212454212453, "grad_norm": 1.203125, "learning_rate": 9.920908786049615e-05, "loss": 0.1467, "step": 4976 }, { "epoch": 3.646153846153846, "grad_norm": 1.078125, "learning_rate": 9.920847769030566e-05, "loss": 0.1757, "step": 4977 }, { "epoch": 3.646886446886447, "grad_norm": 1.3125, "learning_rate": 9.920786728673685e-05, "loss": 0.1634, "step": 4978 }, { "epoch": 3.6476190476190475, "grad_norm": 0.9765625, "learning_rate": 9.920725664979267e-05, "loss": 0.1951, "step": 4979 }, { "epoch": 3.6483516483516483, "grad_norm": 1.046875, "learning_rate": 9.920664577947605e-05, "loss": 0.178, "step": 4980 }, { "epoch": 3.649084249084249, "grad_norm": 1.0234375, "learning_rate": 9.92060346757899e-05, "loss": 0.1761, "step": 4981 }, { "epoch": 3.6498168498168497, "grad_norm": 1.109375, "learning_rate": 9.920542333873714e-05, "loss": 0.2189, "step": 4982 }, { "epoch": 3.6505494505494505, "grad_norm": 1.234375, "learning_rate": 9.92048117683207e-05, "loss": 0.1769, "step": 4983 }, { "epoch": 3.651282051282051, "grad_norm": 1.46875, "learning_rate": 9.920419996454352e-05, "loss": 0.1536, "step": 4984 }, { "epoch": 3.652014652014652, "grad_norm": 1.2890625, "learning_rate": 9.920358792740853e-05, "loss": 0.1784, "step": 4985 }, { "epoch": 3.6527472527472526, "grad_norm": 0.8515625, "learning_rate": 9.920297565691868e-05, "loss": 0.1431, "step": 4986 }, { "epoch": 3.6534798534798534, "grad_norm": 1.1328125, "learning_rate": 9.920236315307687e-05, "loss": 0.1726, "step": 4987 }, { "epoch": 3.654212454212454, "grad_norm": 0.89453125, "learning_rate": 9.920175041588605e-05, "loss": 0.1509, "step": 4988 }, { "epoch": 3.654945054945055, "grad_norm": 1.0859375, "learning_rate": 9.920113744534916e-05, "loss": 0.1709, "step": 4989 }, { "epoch": 3.6556776556776556, "grad_norm": 1.125, "learning_rate": 9.920052424146914e-05, "loss": 0.1635, "step": 4990 }, { "epoch": 3.6564102564102563, "grad_norm": 1.0234375, "learning_rate": 9.919991080424891e-05, "loss": 0.1871, "step": 4991 }, { "epoch": 3.657142857142857, "grad_norm": 1.6171875, "learning_rate": 9.919929713369143e-05, "loss": 0.1517, "step": 4992 }, { "epoch": 3.6578754578754578, "grad_norm": 1.2109375, "learning_rate": 9.919868322979962e-05, "loss": 0.1626, "step": 4993 }, { "epoch": 3.6586080586080585, "grad_norm": 1.21875, "learning_rate": 9.919806909257644e-05, "loss": 0.1996, "step": 4994 }, { "epoch": 3.659340659340659, "grad_norm": 0.98046875, "learning_rate": 9.919745472202481e-05, "loss": 0.158, "step": 4995 }, { "epoch": 3.66007326007326, "grad_norm": 1.0546875, "learning_rate": 9.919684011814769e-05, "loss": 0.1327, "step": 4996 }, { "epoch": 3.6608058608058607, "grad_norm": 1.2734375, "learning_rate": 9.919622528094803e-05, "loss": 0.1764, "step": 4997 }, { "epoch": 3.6615384615384614, "grad_norm": 1.2421875, "learning_rate": 9.919561021042876e-05, "loss": 0.1516, "step": 4998 }, { "epoch": 3.662271062271062, "grad_norm": 0.7578125, "learning_rate": 9.919499490659285e-05, "loss": 0.1552, "step": 4999 }, { "epoch": 3.663003663003663, "grad_norm": 0.91796875, "learning_rate": 9.919437936944323e-05, "loss": 0.1729, "step": 5000 }, { "epoch": 3.6637362637362636, "grad_norm": 1.2109375, "learning_rate": 9.919376359898286e-05, "loss": 0.1697, "step": 5001 }, { "epoch": 3.6644688644688643, "grad_norm": 0.9453125, "learning_rate": 9.919314759521468e-05, "loss": 0.1672, "step": 5002 }, { "epoch": 3.665201465201465, "grad_norm": 0.98046875, "learning_rate": 9.919253135814164e-05, "loss": 0.1606, "step": 5003 }, { "epoch": 3.665934065934066, "grad_norm": 0.62109375, "learning_rate": 9.919191488776671e-05, "loss": 0.124, "step": 5004 }, { "epoch": 3.6666666666666665, "grad_norm": 1.015625, "learning_rate": 9.919129818409281e-05, "loss": 0.1466, "step": 5005 }, { "epoch": 3.6673992673992672, "grad_norm": 1.484375, "learning_rate": 9.919068124712291e-05, "loss": 0.1711, "step": 5006 }, { "epoch": 3.668131868131868, "grad_norm": 1.0234375, "learning_rate": 9.919006407686e-05, "loss": 0.1444, "step": 5007 }, { "epoch": 3.6688644688644687, "grad_norm": 0.8203125, "learning_rate": 9.9189446673307e-05, "loss": 0.2083, "step": 5008 }, { "epoch": 3.6695970695970694, "grad_norm": 1.09375, "learning_rate": 9.918882903646686e-05, "loss": 0.2104, "step": 5009 }, { "epoch": 3.67032967032967, "grad_norm": 0.71875, "learning_rate": 9.918821116634258e-05, "loss": 0.1541, "step": 5010 }, { "epoch": 3.671062271062271, "grad_norm": 0.6796875, "learning_rate": 9.918759306293708e-05, "loss": 0.1277, "step": 5011 }, { "epoch": 3.6717948717948716, "grad_norm": 1.21875, "learning_rate": 9.918697472625335e-05, "loss": 0.1875, "step": 5012 }, { "epoch": 3.672527472527473, "grad_norm": 0.93359375, "learning_rate": 9.918635615629435e-05, "loss": 0.1469, "step": 5013 }, { "epoch": 3.6732600732600735, "grad_norm": 1.1484375, "learning_rate": 9.9185737353063e-05, "loss": 0.1311, "step": 5014 }, { "epoch": 3.6739926739926743, "grad_norm": 0.98828125, "learning_rate": 9.918511831656234e-05, "loss": 0.1671, "step": 5015 }, { "epoch": 3.674725274725275, "grad_norm": 0.61328125, "learning_rate": 9.918449904679528e-05, "loss": 0.1433, "step": 5016 }, { "epoch": 3.6754578754578757, "grad_norm": 1.21875, "learning_rate": 9.91838795437648e-05, "loss": 0.1919, "step": 5017 }, { "epoch": 3.6761904761904765, "grad_norm": 0.953125, "learning_rate": 9.918325980747386e-05, "loss": 0.1994, "step": 5018 }, { "epoch": 3.676923076923077, "grad_norm": 1.0859375, "learning_rate": 9.918263983792545e-05, "loss": 0.1535, "step": 5019 }, { "epoch": 3.677655677655678, "grad_norm": 1.4453125, "learning_rate": 9.918201963512253e-05, "loss": 0.1554, "step": 5020 }, { "epoch": 3.6783882783882786, "grad_norm": 1.3125, "learning_rate": 9.918139919906809e-05, "loss": 0.1834, "step": 5021 }, { "epoch": 3.6791208791208794, "grad_norm": 1.515625, "learning_rate": 9.918077852976507e-05, "loss": 0.1536, "step": 5022 }, { "epoch": 3.67985347985348, "grad_norm": 0.60546875, "learning_rate": 9.918015762721645e-05, "loss": 0.1662, "step": 5023 }, { "epoch": 3.680586080586081, "grad_norm": 1.2890625, "learning_rate": 9.917953649142523e-05, "loss": 0.1694, "step": 5024 }, { "epoch": 3.6813186813186816, "grad_norm": 0.80859375, "learning_rate": 9.917891512239436e-05, "loss": 0.1986, "step": 5025 }, { "epoch": 3.6820512820512823, "grad_norm": 1.6171875, "learning_rate": 9.917829352012683e-05, "loss": 0.1601, "step": 5026 }, { "epoch": 3.682783882783883, "grad_norm": 2.109375, "learning_rate": 9.917767168462563e-05, "loss": 0.1695, "step": 5027 }, { "epoch": 3.6835164835164838, "grad_norm": 1.46875, "learning_rate": 9.917704961589371e-05, "loss": 0.1594, "step": 5028 }, { "epoch": 3.6842490842490845, "grad_norm": 0.80859375, "learning_rate": 9.917642731393406e-05, "loss": 0.1654, "step": 5029 }, { "epoch": 3.684981684981685, "grad_norm": 0.90234375, "learning_rate": 9.917580477874969e-05, "loss": 0.1906, "step": 5030 }, { "epoch": 3.685714285714286, "grad_norm": 1.8671875, "learning_rate": 9.917518201034354e-05, "loss": 0.1403, "step": 5031 }, { "epoch": 3.6864468864468867, "grad_norm": 2.609375, "learning_rate": 9.917455900871862e-05, "loss": 0.1868, "step": 5032 }, { "epoch": 3.6871794871794874, "grad_norm": 1.5078125, "learning_rate": 9.917393577387791e-05, "loss": 0.1382, "step": 5033 }, { "epoch": 3.687912087912088, "grad_norm": 1.3046875, "learning_rate": 9.91733123058244e-05, "loss": 0.1492, "step": 5034 }, { "epoch": 3.688644688644689, "grad_norm": 1.5078125, "learning_rate": 9.917268860456107e-05, "loss": 0.1696, "step": 5035 }, { "epoch": 3.6893772893772896, "grad_norm": 1.21875, "learning_rate": 9.91720646700909e-05, "loss": 0.1903, "step": 5036 }, { "epoch": 3.6901098901098903, "grad_norm": 1.2265625, "learning_rate": 9.91714405024169e-05, "loss": 0.1395, "step": 5037 }, { "epoch": 3.690842490842491, "grad_norm": 1.203125, "learning_rate": 9.917081610154206e-05, "loss": 0.1296, "step": 5038 }, { "epoch": 3.691575091575092, "grad_norm": 1.21875, "learning_rate": 9.917019146746935e-05, "loss": 0.1857, "step": 5039 }, { "epoch": 3.6923076923076925, "grad_norm": 1.1875, "learning_rate": 9.916956660020177e-05, "loss": 0.1471, "step": 5040 }, { "epoch": 3.6930402930402932, "grad_norm": 0.9921875, "learning_rate": 9.916894149974231e-05, "loss": 0.1277, "step": 5041 }, { "epoch": 3.693772893772894, "grad_norm": 1.203125, "learning_rate": 9.916831616609401e-05, "loss": 0.2124, "step": 5042 }, { "epoch": 3.6945054945054947, "grad_norm": 1.25, "learning_rate": 9.916769059925979e-05, "loss": 0.1808, "step": 5043 }, { "epoch": 3.6952380952380954, "grad_norm": 0.95703125, "learning_rate": 9.916706479924272e-05, "loss": 0.1861, "step": 5044 }, { "epoch": 3.695970695970696, "grad_norm": 0.94140625, "learning_rate": 9.916643876604576e-05, "loss": 0.1771, "step": 5045 }, { "epoch": 3.696703296703297, "grad_norm": 1.015625, "learning_rate": 9.91658124996719e-05, "loss": 0.1069, "step": 5046 }, { "epoch": 3.6974358974358976, "grad_norm": 1.2890625, "learning_rate": 9.916518600012417e-05, "loss": 0.1511, "step": 5047 }, { "epoch": 3.6981684981684984, "grad_norm": 0.7890625, "learning_rate": 9.916455926740557e-05, "loss": 0.1476, "step": 5048 }, { "epoch": 3.698901098901099, "grad_norm": 0.6484375, "learning_rate": 9.916393230151907e-05, "loss": 0.1463, "step": 5049 }, { "epoch": 3.6996336996337, "grad_norm": 0.65625, "learning_rate": 9.916330510246769e-05, "loss": 0.1579, "step": 5050 }, { "epoch": 3.7003663003663005, "grad_norm": 0.578125, "learning_rate": 9.916267767025445e-05, "loss": 0.152, "step": 5051 }, { "epoch": 3.7010989010989013, "grad_norm": 0.89453125, "learning_rate": 9.916205000488235e-05, "loss": 0.1353, "step": 5052 }, { "epoch": 3.701831501831502, "grad_norm": 1.25, "learning_rate": 9.916142210635441e-05, "loss": 0.1469, "step": 5053 }, { "epoch": 3.7025641025641027, "grad_norm": 0.6796875, "learning_rate": 9.916079397467358e-05, "loss": 0.1759, "step": 5054 }, { "epoch": 3.7032967032967035, "grad_norm": 0.6171875, "learning_rate": 9.916016560984294e-05, "loss": 0.1475, "step": 5055 }, { "epoch": 3.704029304029304, "grad_norm": 0.5703125, "learning_rate": 9.915953701186547e-05, "loss": 0.1467, "step": 5056 }, { "epoch": 3.704761904761905, "grad_norm": 1.3359375, "learning_rate": 9.915890818074418e-05, "loss": 0.1814, "step": 5057 }, { "epoch": 3.7054945054945057, "grad_norm": 1.1171875, "learning_rate": 9.915827911648209e-05, "loss": 0.1789, "step": 5058 }, { "epoch": 3.7062271062271064, "grad_norm": 1.0, "learning_rate": 9.915764981908221e-05, "loss": 0.135, "step": 5059 }, { "epoch": 3.706959706959707, "grad_norm": 0.9140625, "learning_rate": 9.915702028854755e-05, "loss": 0.1248, "step": 5060 }, { "epoch": 3.707692307692308, "grad_norm": 0.80078125, "learning_rate": 9.915639052488113e-05, "loss": 0.1555, "step": 5061 }, { "epoch": 3.7084249084249086, "grad_norm": 1.203125, "learning_rate": 9.9155760528086e-05, "loss": 0.1162, "step": 5062 }, { "epoch": 3.7091575091575093, "grad_norm": 1.5, "learning_rate": 9.915513029816512e-05, "loss": 0.1343, "step": 5063 }, { "epoch": 3.70989010989011, "grad_norm": 1.25, "learning_rate": 9.915449983512157e-05, "loss": 0.1445, "step": 5064 }, { "epoch": 3.7106227106227108, "grad_norm": 1.421875, "learning_rate": 9.91538691389583e-05, "loss": 0.1497, "step": 5065 }, { "epoch": 3.7113553113553115, "grad_norm": 0.93359375, "learning_rate": 9.915323820967837e-05, "loss": 0.1439, "step": 5066 }, { "epoch": 3.7120879120879122, "grad_norm": 1.0546875, "learning_rate": 9.915260704728482e-05, "loss": 0.1607, "step": 5067 }, { "epoch": 3.712820512820513, "grad_norm": 1.390625, "learning_rate": 9.915197565178067e-05, "loss": 0.1498, "step": 5068 }, { "epoch": 3.7135531135531137, "grad_norm": 0.94921875, "learning_rate": 9.915134402316892e-05, "loss": 0.1783, "step": 5069 }, { "epoch": 3.7142857142857144, "grad_norm": 0.93359375, "learning_rate": 9.915071216145262e-05, "loss": 0.1263, "step": 5070 }, { "epoch": 3.715018315018315, "grad_norm": 0.8828125, "learning_rate": 9.915008006663477e-05, "loss": 0.1848, "step": 5071 }, { "epoch": 3.715750915750916, "grad_norm": 0.90625, "learning_rate": 9.914944773871842e-05, "loss": 0.1785, "step": 5072 }, { "epoch": 3.7164835164835166, "grad_norm": 0.7890625, "learning_rate": 9.91488151777066e-05, "loss": 0.1453, "step": 5073 }, { "epoch": 3.7172161172161173, "grad_norm": 0.8203125, "learning_rate": 9.914818238360235e-05, "loss": 0.1415, "step": 5074 }, { "epoch": 3.717948717948718, "grad_norm": 0.921875, "learning_rate": 9.914754935640868e-05, "loss": 0.1282, "step": 5075 }, { "epoch": 3.718681318681319, "grad_norm": 1.359375, "learning_rate": 9.914691609612862e-05, "loss": 0.1613, "step": 5076 }, { "epoch": 3.7194139194139195, "grad_norm": 0.98046875, "learning_rate": 9.914628260276522e-05, "loss": 0.1755, "step": 5077 }, { "epoch": 3.7201465201465203, "grad_norm": 0.83984375, "learning_rate": 9.91456488763215e-05, "loss": 0.1659, "step": 5078 }, { "epoch": 3.720879120879121, "grad_norm": 0.734375, "learning_rate": 9.914501491680052e-05, "loss": 0.1608, "step": 5079 }, { "epoch": 3.7216117216117217, "grad_norm": 1.2734375, "learning_rate": 9.91443807242053e-05, "loss": 0.1636, "step": 5080 }, { "epoch": 3.7223443223443224, "grad_norm": 0.85546875, "learning_rate": 9.914374629853889e-05, "loss": 0.1628, "step": 5081 }, { "epoch": 3.723076923076923, "grad_norm": 0.640625, "learning_rate": 9.91431116398043e-05, "loss": 0.1494, "step": 5082 }, { "epoch": 3.723809523809524, "grad_norm": 1.0546875, "learning_rate": 9.914247674800462e-05, "loss": 0.1857, "step": 5083 }, { "epoch": 3.7245421245421246, "grad_norm": 1.265625, "learning_rate": 9.914184162314285e-05, "loss": 0.1913, "step": 5084 }, { "epoch": 3.7252747252747254, "grad_norm": 1.1328125, "learning_rate": 9.914120626522207e-05, "loss": 0.1467, "step": 5085 }, { "epoch": 3.726007326007326, "grad_norm": 1.40625, "learning_rate": 9.914057067424528e-05, "loss": 0.1591, "step": 5086 }, { "epoch": 3.726739926739927, "grad_norm": 0.9140625, "learning_rate": 9.913993485021555e-05, "loss": 0.1366, "step": 5087 }, { "epoch": 3.7274725274725276, "grad_norm": 1.1875, "learning_rate": 9.913929879313593e-05, "loss": 0.184, "step": 5088 }, { "epoch": 3.7282051282051283, "grad_norm": 1.3203125, "learning_rate": 9.913866250300947e-05, "loss": 0.1917, "step": 5089 }, { "epoch": 3.728937728937729, "grad_norm": 1.0625, "learning_rate": 9.913802597983921e-05, "loss": 0.14, "step": 5090 }, { "epoch": 3.7296703296703297, "grad_norm": 1.09375, "learning_rate": 9.913738922362819e-05, "loss": 0.1647, "step": 5091 }, { "epoch": 3.7304029304029305, "grad_norm": 1.4140625, "learning_rate": 9.913675223437948e-05, "loss": 0.1494, "step": 5092 }, { "epoch": 3.731135531135531, "grad_norm": 1.65625, "learning_rate": 9.913611501209613e-05, "loss": 0.1592, "step": 5093 }, { "epoch": 3.731868131868132, "grad_norm": 1.0859375, "learning_rate": 9.913547755678117e-05, "loss": 0.1427, "step": 5094 }, { "epoch": 3.7326007326007327, "grad_norm": 1.171875, "learning_rate": 9.913483986843768e-05, "loss": 0.1819, "step": 5095 }, { "epoch": 3.7333333333333334, "grad_norm": 1.2890625, "learning_rate": 9.91342019470687e-05, "loss": 0.1261, "step": 5096 }, { "epoch": 3.734065934065934, "grad_norm": 1.90625, "learning_rate": 9.913356379267731e-05, "loss": 0.1893, "step": 5097 }, { "epoch": 3.734798534798535, "grad_norm": 1.8125, "learning_rate": 9.913292540526653e-05, "loss": 0.1825, "step": 5098 }, { "epoch": 3.7355311355311356, "grad_norm": 0.7421875, "learning_rate": 9.913228678483945e-05, "loss": 0.1536, "step": 5099 }, { "epoch": 3.7362637362637363, "grad_norm": 1.015625, "learning_rate": 9.913164793139911e-05, "loss": 0.1678, "step": 5100 }, { "epoch": 3.736996336996337, "grad_norm": 0.80859375, "learning_rate": 9.913100884494859e-05, "loss": 0.1444, "step": 5101 }, { "epoch": 3.7377289377289378, "grad_norm": 0.76171875, "learning_rate": 9.913036952549094e-05, "loss": 0.163, "step": 5102 }, { "epoch": 3.7384615384615385, "grad_norm": 1.265625, "learning_rate": 9.912972997302921e-05, "loss": 0.1984, "step": 5103 }, { "epoch": 3.7391941391941392, "grad_norm": 0.76171875, "learning_rate": 9.91290901875665e-05, "loss": 0.1981, "step": 5104 }, { "epoch": 3.73992673992674, "grad_norm": 0.73046875, "learning_rate": 9.912845016910584e-05, "loss": 0.1411, "step": 5105 }, { "epoch": 3.7406593406593407, "grad_norm": 0.83984375, "learning_rate": 9.912780991765034e-05, "loss": 0.1491, "step": 5106 }, { "epoch": 3.7413919413919414, "grad_norm": 0.953125, "learning_rate": 9.912716943320301e-05, "loss": 0.1721, "step": 5107 }, { "epoch": 3.742124542124542, "grad_norm": 0.828125, "learning_rate": 9.912652871576697e-05, "loss": 0.1349, "step": 5108 }, { "epoch": 3.742857142857143, "grad_norm": 1.25, "learning_rate": 9.912588776534524e-05, "loss": 0.2062, "step": 5109 }, { "epoch": 3.7435897435897436, "grad_norm": 0.9296875, "learning_rate": 9.912524658194094e-05, "loss": 0.1437, "step": 5110 }, { "epoch": 3.7443223443223443, "grad_norm": 1.0859375, "learning_rate": 9.912460516555712e-05, "loss": 0.1574, "step": 5111 }, { "epoch": 3.745054945054945, "grad_norm": 1.0859375, "learning_rate": 9.912396351619683e-05, "loss": 0.1677, "step": 5112 }, { "epoch": 3.745787545787546, "grad_norm": 1.234375, "learning_rate": 9.912332163386321e-05, "loss": 0.1976, "step": 5113 }, { "epoch": 3.7465201465201465, "grad_norm": 1.2265625, "learning_rate": 9.912267951855926e-05, "loss": 0.2198, "step": 5114 }, { "epoch": 3.7472527472527473, "grad_norm": 0.8828125, "learning_rate": 9.91220371702881e-05, "loss": 0.1726, "step": 5115 }, { "epoch": 3.747985347985348, "grad_norm": 1.0546875, "learning_rate": 9.91213945890528e-05, "loss": 0.158, "step": 5116 }, { "epoch": 3.7487179487179487, "grad_norm": 1.5078125, "learning_rate": 9.912075177485645e-05, "loss": 0.2076, "step": 5117 }, { "epoch": 3.7494505494505495, "grad_norm": 0.8984375, "learning_rate": 9.912010872770211e-05, "loss": 0.135, "step": 5118 }, { "epoch": 3.75018315018315, "grad_norm": 1.53125, "learning_rate": 9.911946544759285e-05, "loss": 0.1748, "step": 5119 }, { "epoch": 3.750915750915751, "grad_norm": 1.046875, "learning_rate": 9.911882193453178e-05, "loss": 0.1599, "step": 5120 }, { "epoch": 3.7516483516483516, "grad_norm": 1.015625, "learning_rate": 9.911817818852198e-05, "loss": 0.1646, "step": 5121 }, { "epoch": 3.7523809523809524, "grad_norm": 1.140625, "learning_rate": 9.911753420956654e-05, "loss": 0.1547, "step": 5122 }, { "epoch": 3.753113553113553, "grad_norm": 1.0234375, "learning_rate": 9.91168899976685e-05, "loss": 0.1893, "step": 5123 }, { "epoch": 3.753846153846154, "grad_norm": 1.40625, "learning_rate": 9.9116245552831e-05, "loss": 0.1696, "step": 5124 }, { "epoch": 3.7545787545787546, "grad_norm": 2.484375, "learning_rate": 9.911560087505712e-05, "loss": 0.1671, "step": 5125 }, { "epoch": 3.7553113553113553, "grad_norm": 2.0, "learning_rate": 9.91149559643499e-05, "loss": 0.1615, "step": 5126 }, { "epoch": 3.756043956043956, "grad_norm": 1.2265625, "learning_rate": 9.911431082071251e-05, "loss": 0.147, "step": 5127 }, { "epoch": 3.7567765567765568, "grad_norm": 1.0234375, "learning_rate": 9.911366544414797e-05, "loss": 0.1751, "step": 5128 }, { "epoch": 3.7575091575091575, "grad_norm": 1.1484375, "learning_rate": 9.911301983465941e-05, "loss": 0.1856, "step": 5129 }, { "epoch": 3.758241758241758, "grad_norm": 1.859375, "learning_rate": 9.911237399224991e-05, "loss": 0.1867, "step": 5130 }, { "epoch": 3.758974358974359, "grad_norm": 1.5703125, "learning_rate": 9.911172791692257e-05, "loss": 0.156, "step": 5131 }, { "epoch": 3.7597069597069597, "grad_norm": 0.7890625, "learning_rate": 9.911108160868047e-05, "loss": 0.2022, "step": 5132 }, { "epoch": 3.7604395604395604, "grad_norm": 0.9140625, "learning_rate": 9.911043506752676e-05, "loss": 0.1377, "step": 5133 }, { "epoch": 3.761172161172161, "grad_norm": 0.68359375, "learning_rate": 9.910978829346446e-05, "loss": 0.1725, "step": 5134 }, { "epoch": 3.761904761904762, "grad_norm": 0.984375, "learning_rate": 9.910914128649672e-05, "loss": 0.1363, "step": 5135 }, { "epoch": 3.7626373626373626, "grad_norm": 1.03125, "learning_rate": 9.910849404662662e-05, "loss": 0.1398, "step": 5136 }, { "epoch": 3.7633699633699633, "grad_norm": 0.99609375, "learning_rate": 9.910784657385728e-05, "loss": 0.1367, "step": 5137 }, { "epoch": 3.764102564102564, "grad_norm": 1.53125, "learning_rate": 9.910719886819177e-05, "loss": 0.1697, "step": 5138 }, { "epoch": 3.764835164835165, "grad_norm": 0.5390625, "learning_rate": 9.910655092963324e-05, "loss": 0.1218, "step": 5139 }, { "epoch": 3.7655677655677655, "grad_norm": 1.2890625, "learning_rate": 9.910590275818476e-05, "loss": 0.1505, "step": 5140 }, { "epoch": 3.7663003663003662, "grad_norm": 0.83984375, "learning_rate": 9.910525435384942e-05, "loss": 0.1364, "step": 5141 }, { "epoch": 3.767032967032967, "grad_norm": 1.234375, "learning_rate": 9.910460571663037e-05, "loss": 0.1592, "step": 5142 }, { "epoch": 3.7677655677655677, "grad_norm": 1.46875, "learning_rate": 9.91039568465307e-05, "loss": 0.1432, "step": 5143 }, { "epoch": 3.7684981684981684, "grad_norm": 0.76953125, "learning_rate": 9.91033077435535e-05, "loss": 0.1565, "step": 5144 }, { "epoch": 3.769230769230769, "grad_norm": 0.75, "learning_rate": 9.910265840770192e-05, "loss": 0.1534, "step": 5145 }, { "epoch": 3.76996336996337, "grad_norm": 0.8671875, "learning_rate": 9.910200883897903e-05, "loss": 0.1994, "step": 5146 }, { "epoch": 3.7706959706959706, "grad_norm": 0.67578125, "learning_rate": 9.910135903738798e-05, "loss": 0.1265, "step": 5147 }, { "epoch": 3.7714285714285714, "grad_norm": 1.484375, "learning_rate": 9.910070900293185e-05, "loss": 0.1417, "step": 5148 }, { "epoch": 3.772161172161172, "grad_norm": 0.86328125, "learning_rate": 9.910005873561378e-05, "loss": 0.1434, "step": 5149 }, { "epoch": 3.772893772893773, "grad_norm": 1.25, "learning_rate": 9.909940823543686e-05, "loss": 0.1661, "step": 5150 }, { "epoch": 3.7736263736263735, "grad_norm": 0.70703125, "learning_rate": 9.909875750240423e-05, "loss": 0.1491, "step": 5151 }, { "epoch": 3.7743589743589743, "grad_norm": 1.0, "learning_rate": 9.909810653651899e-05, "loss": 0.1688, "step": 5152 }, { "epoch": 3.775091575091575, "grad_norm": 1.046875, "learning_rate": 9.909745533778427e-05, "loss": 0.1796, "step": 5153 }, { "epoch": 3.7758241758241757, "grad_norm": 0.9140625, "learning_rate": 9.909680390620319e-05, "loss": 0.1808, "step": 5154 }, { "epoch": 3.7765567765567765, "grad_norm": 0.93359375, "learning_rate": 9.909615224177888e-05, "loss": 0.171, "step": 5155 }, { "epoch": 3.777289377289377, "grad_norm": 0.91015625, "learning_rate": 9.909550034451445e-05, "loss": 0.1538, "step": 5156 }, { "epoch": 3.778021978021978, "grad_norm": 1.3125, "learning_rate": 9.909484821441301e-05, "loss": 0.1514, "step": 5157 }, { "epoch": 3.7787545787545787, "grad_norm": 0.953125, "learning_rate": 9.90941958514777e-05, "loss": 0.1227, "step": 5158 }, { "epoch": 3.7794871794871794, "grad_norm": 1.1328125, "learning_rate": 9.909354325571165e-05, "loss": 0.1818, "step": 5159 }, { "epoch": 3.78021978021978, "grad_norm": 1.0078125, "learning_rate": 9.909289042711799e-05, "loss": 0.1635, "step": 5160 }, { "epoch": 3.780952380952381, "grad_norm": 1.25, "learning_rate": 9.909223736569984e-05, "loss": 0.1307, "step": 5161 }, { "epoch": 3.7816849816849816, "grad_norm": 1.328125, "learning_rate": 9.909158407146031e-05, "loss": 0.1252, "step": 5162 }, { "epoch": 3.7824175824175823, "grad_norm": 0.96484375, "learning_rate": 9.909093054440258e-05, "loss": 0.1367, "step": 5163 }, { "epoch": 3.783150183150183, "grad_norm": 1.7734375, "learning_rate": 9.909027678452972e-05, "loss": 0.1377, "step": 5164 }, { "epoch": 3.7838827838827838, "grad_norm": 1.6328125, "learning_rate": 9.908962279184491e-05, "loss": 0.1657, "step": 5165 }, { "epoch": 3.7846153846153845, "grad_norm": 1.2578125, "learning_rate": 9.908896856635127e-05, "loss": 0.123, "step": 5166 }, { "epoch": 3.7853479853479852, "grad_norm": 1.015625, "learning_rate": 9.908831410805192e-05, "loss": 0.1647, "step": 5167 }, { "epoch": 3.786080586080586, "grad_norm": 1.3671875, "learning_rate": 9.908765941695e-05, "loss": 0.1369, "step": 5168 }, { "epoch": 3.7868131868131867, "grad_norm": 2.84375, "learning_rate": 9.908700449304867e-05, "loss": 0.1699, "step": 5169 }, { "epoch": 3.7875457875457874, "grad_norm": 2.015625, "learning_rate": 9.908634933635103e-05, "loss": 0.1493, "step": 5170 }, { "epoch": 3.788278388278388, "grad_norm": 1.9765625, "learning_rate": 9.908569394686025e-05, "loss": 0.1611, "step": 5171 }, { "epoch": 3.789010989010989, "grad_norm": 1.21875, "learning_rate": 9.908503832457948e-05, "loss": 0.1824, "step": 5172 }, { "epoch": 3.7897435897435896, "grad_norm": 0.91796875, "learning_rate": 9.908438246951183e-05, "loss": 0.1556, "step": 5173 }, { "epoch": 3.7904761904761903, "grad_norm": 1.2734375, "learning_rate": 9.908372638166044e-05, "loss": 0.1744, "step": 5174 }, { "epoch": 3.791208791208791, "grad_norm": 1.1640625, "learning_rate": 9.908307006102848e-05, "loss": 0.1308, "step": 5175 }, { "epoch": 3.791941391941392, "grad_norm": 0.76171875, "learning_rate": 9.90824135076191e-05, "loss": 0.1112, "step": 5176 }, { "epoch": 3.7926739926739925, "grad_norm": 1.0625, "learning_rate": 9.908175672143539e-05, "loss": 0.1293, "step": 5177 }, { "epoch": 3.7934065934065933, "grad_norm": 1.1171875, "learning_rate": 9.908109970248056e-05, "loss": 0.1877, "step": 5178 }, { "epoch": 3.794139194139194, "grad_norm": 1.71875, "learning_rate": 9.908044245075773e-05, "loss": 0.1584, "step": 5179 }, { "epoch": 3.7948717948717947, "grad_norm": 1.421875, "learning_rate": 9.907978496627005e-05, "loss": 0.1684, "step": 5180 }, { "epoch": 3.7956043956043954, "grad_norm": 1.3203125, "learning_rate": 9.907912724902069e-05, "loss": 0.1221, "step": 5181 }, { "epoch": 3.796336996336996, "grad_norm": 0.72265625, "learning_rate": 9.907846929901274e-05, "loss": 0.168, "step": 5182 }, { "epoch": 3.797069597069597, "grad_norm": 0.703125, "learning_rate": 9.907781111624944e-05, "loss": 0.1382, "step": 5183 }, { "epoch": 3.7978021978021976, "grad_norm": 0.94921875, "learning_rate": 9.907715270073388e-05, "loss": 0.1469, "step": 5184 }, { "epoch": 3.7985347985347984, "grad_norm": 1.2109375, "learning_rate": 9.907649405246924e-05, "loss": 0.1318, "step": 5185 }, { "epoch": 3.799267399267399, "grad_norm": 1.65625, "learning_rate": 9.907583517145867e-05, "loss": 0.1851, "step": 5186 }, { "epoch": 3.8, "grad_norm": 1.3125, "learning_rate": 9.907517605770532e-05, "loss": 0.1612, "step": 5187 }, { "epoch": 3.8007326007326006, "grad_norm": 1.125, "learning_rate": 9.907451671121236e-05, "loss": 0.1698, "step": 5188 }, { "epoch": 3.8014652014652013, "grad_norm": 1.359375, "learning_rate": 9.907385713198295e-05, "loss": 0.143, "step": 5189 }, { "epoch": 3.802197802197802, "grad_norm": 1.359375, "learning_rate": 9.907319732002025e-05, "loss": 0.1822, "step": 5190 }, { "epoch": 3.8029304029304027, "grad_norm": 1.0390625, "learning_rate": 9.907253727532741e-05, "loss": 0.1511, "step": 5191 }, { "epoch": 3.8036630036630035, "grad_norm": 0.95703125, "learning_rate": 9.907187699790758e-05, "loss": 0.1731, "step": 5192 }, { "epoch": 3.804395604395604, "grad_norm": 0.96484375, "learning_rate": 9.907121648776397e-05, "loss": 0.1809, "step": 5193 }, { "epoch": 3.805128205128205, "grad_norm": 1.2734375, "learning_rate": 9.907055574489968e-05, "loss": 0.1539, "step": 5194 }, { "epoch": 3.8058608058608057, "grad_norm": 0.875, "learning_rate": 9.906989476931796e-05, "loss": 0.1532, "step": 5195 }, { "epoch": 3.8065934065934064, "grad_norm": 0.8671875, "learning_rate": 9.906923356102192e-05, "loss": 0.1653, "step": 5196 }, { "epoch": 3.807326007326007, "grad_norm": 1.0234375, "learning_rate": 9.90685721200147e-05, "loss": 0.1631, "step": 5197 }, { "epoch": 3.808058608058608, "grad_norm": 0.9921875, "learning_rate": 9.906791044629953e-05, "loss": 0.1915, "step": 5198 }, { "epoch": 3.8087912087912086, "grad_norm": 1.3046875, "learning_rate": 9.906724853987957e-05, "loss": 0.1432, "step": 5199 }, { "epoch": 3.8095238095238093, "grad_norm": 1.1875, "learning_rate": 9.906658640075796e-05, "loss": 0.1431, "step": 5200 }, { "epoch": 3.81025641025641, "grad_norm": 0.77734375, "learning_rate": 9.906592402893791e-05, "loss": 0.1406, "step": 5201 }, { "epoch": 3.8109890109890108, "grad_norm": 0.71875, "learning_rate": 9.906526142442256e-05, "loss": 0.1383, "step": 5202 }, { "epoch": 3.8117216117216115, "grad_norm": 0.828125, "learning_rate": 9.906459858721509e-05, "loss": 0.1612, "step": 5203 }, { "epoch": 3.8124542124542122, "grad_norm": 0.97265625, "learning_rate": 9.90639355173187e-05, "loss": 0.1547, "step": 5204 }, { "epoch": 3.813186813186813, "grad_norm": 1.3125, "learning_rate": 9.906327221473656e-05, "loss": 0.1843, "step": 5205 }, { "epoch": 3.8139194139194137, "grad_norm": 1.1484375, "learning_rate": 9.906260867947184e-05, "loss": 0.1647, "step": 5206 }, { "epoch": 3.8146520146520144, "grad_norm": 1.0546875, "learning_rate": 9.90619449115277e-05, "loss": 0.1653, "step": 5207 }, { "epoch": 3.815384615384615, "grad_norm": 0.94140625, "learning_rate": 9.906128091090735e-05, "loss": 0.1589, "step": 5208 }, { "epoch": 3.816117216117216, "grad_norm": 1.4765625, "learning_rate": 9.906061667761397e-05, "loss": 0.1886, "step": 5209 }, { "epoch": 3.8168498168498166, "grad_norm": 1.3046875, "learning_rate": 9.905995221165073e-05, "loss": 0.1422, "step": 5210 }, { "epoch": 3.8175824175824173, "grad_norm": 1.015625, "learning_rate": 9.90592875130208e-05, "loss": 0.1422, "step": 5211 }, { "epoch": 3.818315018315018, "grad_norm": 1.015625, "learning_rate": 9.90586225817274e-05, "loss": 0.1534, "step": 5212 }, { "epoch": 3.819047619047619, "grad_norm": 0.7421875, "learning_rate": 9.90579574177737e-05, "loss": 0.154, "step": 5213 }, { "epoch": 3.8197802197802195, "grad_norm": 2.03125, "learning_rate": 9.905729202116288e-05, "loss": 0.1743, "step": 5214 }, { "epoch": 3.8205128205128203, "grad_norm": 1.5546875, "learning_rate": 9.905662639189813e-05, "loss": 0.2075, "step": 5215 }, { "epoch": 3.821245421245421, "grad_norm": 1.5, "learning_rate": 9.905596052998265e-05, "loss": 0.1408, "step": 5216 }, { "epoch": 3.8219780219780217, "grad_norm": 1.03125, "learning_rate": 9.905529443541961e-05, "loss": 0.153, "step": 5217 }, { "epoch": 3.8227106227106225, "grad_norm": 1.28125, "learning_rate": 9.905462810821224e-05, "loss": 0.1678, "step": 5218 }, { "epoch": 3.823443223443223, "grad_norm": 1.375, "learning_rate": 9.905396154836369e-05, "loss": 0.1832, "step": 5219 }, { "epoch": 3.824175824175824, "grad_norm": 1.296875, "learning_rate": 9.905329475587718e-05, "loss": 0.1474, "step": 5220 }, { "epoch": 3.8249084249084246, "grad_norm": 1.2109375, "learning_rate": 9.905262773075588e-05, "loss": 0.1851, "step": 5221 }, { "epoch": 3.8256410256410254, "grad_norm": 0.98046875, "learning_rate": 9.905196047300301e-05, "loss": 0.1697, "step": 5222 }, { "epoch": 3.826373626373626, "grad_norm": 0.8828125, "learning_rate": 9.905129298262176e-05, "loss": 0.146, "step": 5223 }, { "epoch": 3.827106227106227, "grad_norm": 1.421875, "learning_rate": 9.905062525961533e-05, "loss": 0.1831, "step": 5224 }, { "epoch": 3.8278388278388276, "grad_norm": 2.671875, "learning_rate": 9.904995730398692e-05, "loss": 0.1884, "step": 5225 }, { "epoch": 3.8285714285714287, "grad_norm": 1.9765625, "learning_rate": 9.904928911573972e-05, "loss": 0.1496, "step": 5226 }, { "epoch": 3.8293040293040295, "grad_norm": 1.2109375, "learning_rate": 9.904862069487694e-05, "loss": 0.1579, "step": 5227 }, { "epoch": 3.83003663003663, "grad_norm": 1.0703125, "learning_rate": 9.904795204140177e-05, "loss": 0.1627, "step": 5228 }, { "epoch": 3.830769230769231, "grad_norm": 1.2265625, "learning_rate": 9.904728315531743e-05, "loss": 0.1455, "step": 5229 }, { "epoch": 3.8315018315018317, "grad_norm": 0.95703125, "learning_rate": 9.904661403662713e-05, "loss": 0.1549, "step": 5230 }, { "epoch": 3.8322344322344324, "grad_norm": 0.98046875, "learning_rate": 9.904594468533405e-05, "loss": 0.1476, "step": 5231 }, { "epoch": 3.832967032967033, "grad_norm": 0.6484375, "learning_rate": 9.904527510144142e-05, "loss": 0.1603, "step": 5232 }, { "epoch": 3.833699633699634, "grad_norm": 0.57421875, "learning_rate": 9.904460528495246e-05, "loss": 0.1285, "step": 5233 }, { "epoch": 3.8344322344322346, "grad_norm": 1.0078125, "learning_rate": 9.904393523587034e-05, "loss": 0.1833, "step": 5234 }, { "epoch": 3.8351648351648353, "grad_norm": 0.734375, "learning_rate": 9.904326495419828e-05, "loss": 0.1737, "step": 5235 }, { "epoch": 3.835897435897436, "grad_norm": 0.8984375, "learning_rate": 9.904259443993952e-05, "loss": 0.1495, "step": 5236 }, { "epoch": 3.8366300366300368, "grad_norm": 1.3671875, "learning_rate": 9.904192369309725e-05, "loss": 0.1563, "step": 5237 }, { "epoch": 3.8373626373626375, "grad_norm": 1.609375, "learning_rate": 9.904125271367468e-05, "loss": 0.1423, "step": 5238 }, { "epoch": 3.8380952380952382, "grad_norm": 1.0, "learning_rate": 9.904058150167505e-05, "loss": 0.161, "step": 5239 }, { "epoch": 3.838827838827839, "grad_norm": 1.3671875, "learning_rate": 9.903991005710153e-05, "loss": 0.1844, "step": 5240 }, { "epoch": 3.8395604395604397, "grad_norm": 0.6875, "learning_rate": 9.90392383799574e-05, "loss": 0.15, "step": 5241 }, { "epoch": 3.8402930402930404, "grad_norm": 1.1953125, "learning_rate": 9.903856647024583e-05, "loss": 0.1471, "step": 5242 }, { "epoch": 3.841025641025641, "grad_norm": 1.34375, "learning_rate": 9.903789432797005e-05, "loss": 0.1917, "step": 5243 }, { "epoch": 3.841758241758242, "grad_norm": 0.76171875, "learning_rate": 9.903722195313328e-05, "loss": 0.1603, "step": 5244 }, { "epoch": 3.8424908424908426, "grad_norm": 1.59375, "learning_rate": 9.903654934573876e-05, "loss": 0.1567, "step": 5245 }, { "epoch": 3.8432234432234433, "grad_norm": 1.921875, "learning_rate": 9.903587650578969e-05, "loss": 0.165, "step": 5246 }, { "epoch": 3.843956043956044, "grad_norm": 1.46875, "learning_rate": 9.90352034332893e-05, "loss": 0.1507, "step": 5247 }, { "epoch": 3.844688644688645, "grad_norm": 1.78125, "learning_rate": 9.903453012824083e-05, "loss": 0.1664, "step": 5248 }, { "epoch": 3.8454212454212455, "grad_norm": 1.0859375, "learning_rate": 9.903385659064749e-05, "loss": 0.174, "step": 5249 }, { "epoch": 3.8461538461538463, "grad_norm": 0.98046875, "learning_rate": 9.90331828205125e-05, "loss": 0.1731, "step": 5250 }, { "epoch": 3.846886446886447, "grad_norm": 1.4140625, "learning_rate": 9.90325088178391e-05, "loss": 0.2011, "step": 5251 }, { "epoch": 3.8476190476190477, "grad_norm": 1.34375, "learning_rate": 9.903183458263052e-05, "loss": 0.1978, "step": 5252 }, { "epoch": 3.8483516483516484, "grad_norm": 0.76953125, "learning_rate": 9.903116011489e-05, "loss": 0.1267, "step": 5253 }, { "epoch": 3.849084249084249, "grad_norm": 0.98046875, "learning_rate": 9.903048541462075e-05, "loss": 0.1621, "step": 5254 }, { "epoch": 3.84981684981685, "grad_norm": 1.2109375, "learning_rate": 9.902981048182601e-05, "loss": 0.1471, "step": 5255 }, { "epoch": 3.8505494505494506, "grad_norm": 0.83203125, "learning_rate": 9.902913531650902e-05, "loss": 0.1469, "step": 5256 }, { "epoch": 3.8512820512820514, "grad_norm": 0.88671875, "learning_rate": 9.902845991867301e-05, "loss": 0.1677, "step": 5257 }, { "epoch": 3.852014652014652, "grad_norm": 1.0390625, "learning_rate": 9.902778428832123e-05, "loss": 0.1671, "step": 5258 }, { "epoch": 3.852747252747253, "grad_norm": 1.3125, "learning_rate": 9.90271084254569e-05, "loss": 0.1818, "step": 5259 }, { "epoch": 3.8534798534798536, "grad_norm": 0.63671875, "learning_rate": 9.902643233008324e-05, "loss": 0.1273, "step": 5260 }, { "epoch": 3.8542124542124543, "grad_norm": 1.125, "learning_rate": 9.902575600220354e-05, "loss": 0.1373, "step": 5261 }, { "epoch": 3.854945054945055, "grad_norm": 1.0078125, "learning_rate": 9.9025079441821e-05, "loss": 0.149, "step": 5262 }, { "epoch": 3.8556776556776557, "grad_norm": 1.0, "learning_rate": 9.902440264893887e-05, "loss": 0.1778, "step": 5263 }, { "epoch": 3.8564102564102565, "grad_norm": 1.5, "learning_rate": 9.90237256235604e-05, "loss": 0.1736, "step": 5264 }, { "epoch": 3.857142857142857, "grad_norm": 0.71875, "learning_rate": 9.902304836568886e-05, "loss": 0.1892, "step": 5265 }, { "epoch": 3.857875457875458, "grad_norm": 0.78125, "learning_rate": 9.902237087532743e-05, "loss": 0.1547, "step": 5266 }, { "epoch": 3.8586080586080587, "grad_norm": 2.34375, "learning_rate": 9.902169315247941e-05, "loss": 0.1478, "step": 5267 }, { "epoch": 3.8593406593406594, "grad_norm": 1.5234375, "learning_rate": 9.902101519714802e-05, "loss": 0.1371, "step": 5268 }, { "epoch": 3.86007326007326, "grad_norm": 1.4921875, "learning_rate": 9.902033700933653e-05, "loss": 0.1939, "step": 5269 }, { "epoch": 3.860805860805861, "grad_norm": 0.90234375, "learning_rate": 9.901965858904816e-05, "loss": 0.2079, "step": 5270 }, { "epoch": 3.8615384615384616, "grad_norm": 0.8828125, "learning_rate": 9.901897993628619e-05, "loss": 0.1536, "step": 5271 }, { "epoch": 3.8622710622710623, "grad_norm": 1.375, "learning_rate": 9.901830105105386e-05, "loss": 0.1713, "step": 5272 }, { "epoch": 3.863003663003663, "grad_norm": 1.625, "learning_rate": 9.90176219333544e-05, "loss": 0.2003, "step": 5273 }, { "epoch": 3.8637362637362638, "grad_norm": 1.515625, "learning_rate": 9.901694258319113e-05, "loss": 0.1752, "step": 5274 }, { "epoch": 3.8644688644688645, "grad_norm": 1.15625, "learning_rate": 9.901626300056724e-05, "loss": 0.161, "step": 5275 }, { "epoch": 3.8652014652014652, "grad_norm": 0.84765625, "learning_rate": 9.9015583185486e-05, "loss": 0.1813, "step": 5276 }, { "epoch": 3.865934065934066, "grad_norm": 1.5625, "learning_rate": 9.901490313795068e-05, "loss": 0.1463, "step": 5277 }, { "epoch": 3.8666666666666667, "grad_norm": 2.015625, "learning_rate": 9.901422285796453e-05, "loss": 0.1665, "step": 5278 }, { "epoch": 3.8673992673992674, "grad_norm": 0.9921875, "learning_rate": 9.901354234553081e-05, "loss": 0.1484, "step": 5279 }, { "epoch": 3.868131868131868, "grad_norm": 0.7109375, "learning_rate": 9.901286160065278e-05, "loss": 0.1503, "step": 5280 }, { "epoch": 3.868864468864469, "grad_norm": 1.265625, "learning_rate": 9.90121806233337e-05, "loss": 0.1666, "step": 5281 }, { "epoch": 3.8695970695970696, "grad_norm": 1.2265625, "learning_rate": 9.901149941357686e-05, "loss": 0.1645, "step": 5282 }, { "epoch": 3.8703296703296703, "grad_norm": 0.66796875, "learning_rate": 9.901081797138548e-05, "loss": 0.1461, "step": 5283 }, { "epoch": 3.871062271062271, "grad_norm": 0.9765625, "learning_rate": 9.901013629676285e-05, "loss": 0.1653, "step": 5284 }, { "epoch": 3.871794871794872, "grad_norm": 0.98046875, "learning_rate": 9.900945438971224e-05, "loss": 0.1718, "step": 5285 }, { "epoch": 3.8725274725274725, "grad_norm": 0.81640625, "learning_rate": 9.900877225023691e-05, "loss": 0.1484, "step": 5286 }, { "epoch": 3.8732600732600733, "grad_norm": 1.25, "learning_rate": 9.900808987834012e-05, "loss": 0.1878, "step": 5287 }, { "epoch": 3.873992673992674, "grad_norm": 1.1171875, "learning_rate": 9.900740727402514e-05, "loss": 0.1847, "step": 5288 }, { "epoch": 3.8747252747252747, "grad_norm": 0.94140625, "learning_rate": 9.900672443729526e-05, "loss": 0.1525, "step": 5289 }, { "epoch": 3.8754578754578755, "grad_norm": 0.9921875, "learning_rate": 9.900604136815374e-05, "loss": 0.1312, "step": 5290 }, { "epoch": 3.876190476190476, "grad_norm": 0.93359375, "learning_rate": 9.900535806660385e-05, "loss": 0.1717, "step": 5291 }, { "epoch": 3.876923076923077, "grad_norm": 0.7734375, "learning_rate": 9.900467453264888e-05, "loss": 0.1585, "step": 5292 }, { "epoch": 3.8776556776556776, "grad_norm": 0.66796875, "learning_rate": 9.900399076629206e-05, "loss": 0.1469, "step": 5293 }, { "epoch": 3.8783882783882784, "grad_norm": 0.8125, "learning_rate": 9.900330676753671e-05, "loss": 0.1393, "step": 5294 }, { "epoch": 3.879120879120879, "grad_norm": 1.2734375, "learning_rate": 9.90026225363861e-05, "loss": 0.139, "step": 5295 }, { "epoch": 3.87985347985348, "grad_norm": 1.328125, "learning_rate": 9.900193807284351e-05, "loss": 0.1758, "step": 5296 }, { "epoch": 3.8805860805860806, "grad_norm": 0.765625, "learning_rate": 9.90012533769122e-05, "loss": 0.133, "step": 5297 }, { "epoch": 3.8813186813186813, "grad_norm": 1.28125, "learning_rate": 9.900056844859548e-05, "loss": 0.1474, "step": 5298 }, { "epoch": 3.882051282051282, "grad_norm": 1.28125, "learning_rate": 9.89998832878966e-05, "loss": 0.1915, "step": 5299 }, { "epoch": 3.8827838827838828, "grad_norm": 2.578125, "learning_rate": 9.899919789481886e-05, "loss": 0.1847, "step": 5300 }, { "epoch": 3.8835164835164835, "grad_norm": 1.484375, "learning_rate": 9.899851226936553e-05, "loss": 0.1671, "step": 5301 }, { "epoch": 3.884249084249084, "grad_norm": 1.75, "learning_rate": 9.899782641153992e-05, "loss": 0.156, "step": 5302 }, { "epoch": 3.884981684981685, "grad_norm": 0.90234375, "learning_rate": 9.899714032134531e-05, "loss": 0.1437, "step": 5303 }, { "epoch": 3.8857142857142857, "grad_norm": 1.1484375, "learning_rate": 9.899645399878498e-05, "loss": 0.1507, "step": 5304 }, { "epoch": 3.8864468864468864, "grad_norm": 2.296875, "learning_rate": 9.899576744386221e-05, "loss": 0.186, "step": 5305 }, { "epoch": 3.887179487179487, "grad_norm": 2.84375, "learning_rate": 9.899508065658029e-05, "loss": 0.1709, "step": 5306 }, { "epoch": 3.887912087912088, "grad_norm": 3.296875, "learning_rate": 9.899439363694253e-05, "loss": 0.1821, "step": 5307 }, { "epoch": 3.8886446886446886, "grad_norm": 2.78125, "learning_rate": 9.89937063849522e-05, "loss": 0.1696, "step": 5308 }, { "epoch": 3.8893772893772893, "grad_norm": 1.6484375, "learning_rate": 9.899301890061262e-05, "loss": 0.1438, "step": 5309 }, { "epoch": 3.89010989010989, "grad_norm": 0.9375, "learning_rate": 9.899233118392706e-05, "loss": 0.1565, "step": 5310 }, { "epoch": 3.890842490842491, "grad_norm": 1.3671875, "learning_rate": 9.899164323489882e-05, "loss": 0.1248, "step": 5311 }, { "epoch": 3.8915750915750915, "grad_norm": 2.40625, "learning_rate": 9.89909550535312e-05, "loss": 0.1626, "step": 5312 }, { "epoch": 3.8923076923076922, "grad_norm": 2.640625, "learning_rate": 9.899026663982749e-05, "loss": 0.1934, "step": 5313 }, { "epoch": 3.893040293040293, "grad_norm": 2.765625, "learning_rate": 9.8989577993791e-05, "loss": 0.1742, "step": 5314 }, { "epoch": 3.8937728937728937, "grad_norm": 0.8125, "learning_rate": 9.898888911542502e-05, "loss": 0.1596, "step": 5315 }, { "epoch": 3.8945054945054944, "grad_norm": 1.8671875, "learning_rate": 9.898820000473286e-05, "loss": 0.1625, "step": 5316 }, { "epoch": 3.895238095238095, "grad_norm": 2.8125, "learning_rate": 9.89875106617178e-05, "loss": 0.1773, "step": 5317 }, { "epoch": 3.895970695970696, "grad_norm": 2.234375, "learning_rate": 9.898682108638319e-05, "loss": 0.168, "step": 5318 }, { "epoch": 3.8967032967032966, "grad_norm": 2.59375, "learning_rate": 9.898613127873227e-05, "loss": 0.1385, "step": 5319 }, { "epoch": 3.8974358974358974, "grad_norm": 1.0390625, "learning_rate": 9.898544123876841e-05, "loss": 0.1746, "step": 5320 }, { "epoch": 3.898168498168498, "grad_norm": 0.8984375, "learning_rate": 9.898475096649487e-05, "loss": 0.1372, "step": 5321 }, { "epoch": 3.898901098901099, "grad_norm": 1.171875, "learning_rate": 9.898406046191496e-05, "loss": 0.1421, "step": 5322 }, { "epoch": 3.8996336996336995, "grad_norm": 1.640625, "learning_rate": 9.898336972503201e-05, "loss": 0.1582, "step": 5323 }, { "epoch": 3.9003663003663003, "grad_norm": 1.953125, "learning_rate": 9.898267875584932e-05, "loss": 0.1794, "step": 5324 }, { "epoch": 3.901098901098901, "grad_norm": 1.234375, "learning_rate": 9.898198755437021e-05, "loss": 0.1723, "step": 5325 }, { "epoch": 3.9018315018315017, "grad_norm": 0.91796875, "learning_rate": 9.898129612059797e-05, "loss": 0.135, "step": 5326 }, { "epoch": 3.9025641025641025, "grad_norm": 0.67578125, "learning_rate": 9.898060445453592e-05, "loss": 0.1639, "step": 5327 }, { "epoch": 3.903296703296703, "grad_norm": 0.93359375, "learning_rate": 9.897991255618737e-05, "loss": 0.165, "step": 5328 }, { "epoch": 3.904029304029304, "grad_norm": 0.70703125, "learning_rate": 9.897922042555566e-05, "loss": 0.1226, "step": 5329 }, { "epoch": 3.9047619047619047, "grad_norm": 0.875, "learning_rate": 9.89785280626441e-05, "loss": 0.1688, "step": 5330 }, { "epoch": 3.9054945054945054, "grad_norm": 0.953125, "learning_rate": 9.8977835467456e-05, "loss": 0.1452, "step": 5331 }, { "epoch": 3.906227106227106, "grad_norm": 1.0546875, "learning_rate": 9.897714263999465e-05, "loss": 0.1736, "step": 5332 }, { "epoch": 3.906959706959707, "grad_norm": 0.95703125, "learning_rate": 9.89764495802634e-05, "loss": 0.1456, "step": 5333 }, { "epoch": 3.9076923076923076, "grad_norm": 0.9375, "learning_rate": 9.897575628826558e-05, "loss": 0.1462, "step": 5334 }, { "epoch": 3.9084249084249083, "grad_norm": 0.69140625, "learning_rate": 9.897506276400449e-05, "loss": 0.1425, "step": 5335 }, { "epoch": 3.909157509157509, "grad_norm": 1.265625, "learning_rate": 9.897436900748347e-05, "loss": 0.1744, "step": 5336 }, { "epoch": 3.9098901098901098, "grad_norm": 1.4609375, "learning_rate": 9.897367501870582e-05, "loss": 0.1264, "step": 5337 }, { "epoch": 3.9106227106227105, "grad_norm": 1.1484375, "learning_rate": 9.89729807976749e-05, "loss": 0.1505, "step": 5338 }, { "epoch": 3.9113553113553112, "grad_norm": 1.109375, "learning_rate": 9.8972286344394e-05, "loss": 0.157, "step": 5339 }, { "epoch": 3.912087912087912, "grad_norm": 1.125, "learning_rate": 9.897159165886648e-05, "loss": 0.22, "step": 5340 }, { "epoch": 3.9128205128205127, "grad_norm": 2.0625, "learning_rate": 9.897089674109563e-05, "loss": 0.2106, "step": 5341 }, { "epoch": 3.9135531135531134, "grad_norm": 1.0625, "learning_rate": 9.897020159108482e-05, "loss": 0.1465, "step": 5342 }, { "epoch": 3.914285714285714, "grad_norm": 0.76953125, "learning_rate": 9.896950620883735e-05, "loss": 0.1178, "step": 5343 }, { "epoch": 3.915018315018315, "grad_norm": 1.53125, "learning_rate": 9.896881059435656e-05, "loss": 0.1866, "step": 5344 }, { "epoch": 3.9157509157509156, "grad_norm": 0.93359375, "learning_rate": 9.896811474764581e-05, "loss": 0.1573, "step": 5345 }, { "epoch": 3.9164835164835163, "grad_norm": 0.765625, "learning_rate": 9.89674186687084e-05, "loss": 0.1601, "step": 5346 }, { "epoch": 3.917216117216117, "grad_norm": 1.5, "learning_rate": 9.896672235754767e-05, "loss": 0.1653, "step": 5347 }, { "epoch": 3.917948717948718, "grad_norm": 1.5546875, "learning_rate": 9.896602581416697e-05, "loss": 0.1809, "step": 5348 }, { "epoch": 3.9186813186813185, "grad_norm": 0.83203125, "learning_rate": 9.896532903856964e-05, "loss": 0.1774, "step": 5349 }, { "epoch": 3.9194139194139193, "grad_norm": 0.8984375, "learning_rate": 9.896463203075898e-05, "loss": 0.1661, "step": 5350 }, { "epoch": 3.92014652014652, "grad_norm": 1.890625, "learning_rate": 9.896393479073838e-05, "loss": 0.1634, "step": 5351 }, { "epoch": 3.9208791208791207, "grad_norm": 1.8046875, "learning_rate": 9.896323731851115e-05, "loss": 0.2071, "step": 5352 }, { "epoch": 3.9216117216117214, "grad_norm": 1.359375, "learning_rate": 9.896253961408064e-05, "loss": 0.1457, "step": 5353 }, { "epoch": 3.9223443223443226, "grad_norm": 0.9296875, "learning_rate": 9.896184167745019e-05, "loss": 0.1761, "step": 5354 }, { "epoch": 3.9230769230769234, "grad_norm": 1.2109375, "learning_rate": 9.896114350862316e-05, "loss": 0.1618, "step": 5355 }, { "epoch": 3.923809523809524, "grad_norm": 0.84765625, "learning_rate": 9.896044510760289e-05, "loss": 0.1694, "step": 5356 }, { "epoch": 3.924542124542125, "grad_norm": 1.1015625, "learning_rate": 9.895974647439269e-05, "loss": 0.1069, "step": 5357 }, { "epoch": 3.9252747252747255, "grad_norm": 1.734375, "learning_rate": 9.895904760899595e-05, "loss": 0.1618, "step": 5358 }, { "epoch": 3.9260073260073263, "grad_norm": 0.8828125, "learning_rate": 9.895834851141601e-05, "loss": 0.1687, "step": 5359 }, { "epoch": 3.926739926739927, "grad_norm": 0.87890625, "learning_rate": 9.895764918165621e-05, "loss": 0.1372, "step": 5360 }, { "epoch": 3.9274725274725277, "grad_norm": 1.1953125, "learning_rate": 9.89569496197199e-05, "loss": 0.1545, "step": 5361 }, { "epoch": 3.9282051282051285, "grad_norm": 1.0390625, "learning_rate": 9.895624982561044e-05, "loss": 0.1966, "step": 5362 }, { "epoch": 3.928937728937729, "grad_norm": 1.390625, "learning_rate": 9.895554979933119e-05, "loss": 0.1659, "step": 5363 }, { "epoch": 3.92967032967033, "grad_norm": 1.140625, "learning_rate": 9.895484954088549e-05, "loss": 0.1552, "step": 5364 }, { "epoch": 3.9304029304029307, "grad_norm": 1.078125, "learning_rate": 9.895414905027671e-05, "loss": 0.1191, "step": 5365 }, { "epoch": 3.9311355311355314, "grad_norm": 1.375, "learning_rate": 9.895344832750817e-05, "loss": 0.1862, "step": 5366 }, { "epoch": 3.931868131868132, "grad_norm": 1.3515625, "learning_rate": 9.895274737258327e-05, "loss": 0.2055, "step": 5367 }, { "epoch": 3.932600732600733, "grad_norm": 1.2890625, "learning_rate": 9.895204618550533e-05, "loss": 0.1495, "step": 5368 }, { "epoch": 3.9333333333333336, "grad_norm": 1.3828125, "learning_rate": 9.895134476627776e-05, "loss": 0.1438, "step": 5369 }, { "epoch": 3.9340659340659343, "grad_norm": 1.125, "learning_rate": 9.895064311490388e-05, "loss": 0.1663, "step": 5370 }, { "epoch": 3.934798534798535, "grad_norm": 0.640625, "learning_rate": 9.894994123138706e-05, "loss": 0.1718, "step": 5371 }, { "epoch": 3.9355311355311358, "grad_norm": 0.94140625, "learning_rate": 9.894923911573068e-05, "loss": 0.1897, "step": 5372 }, { "epoch": 3.9362637362637365, "grad_norm": 1.15625, "learning_rate": 9.894853676793806e-05, "loss": 0.1658, "step": 5373 }, { "epoch": 3.936996336996337, "grad_norm": 1.5859375, "learning_rate": 9.894783418801262e-05, "loss": 0.1531, "step": 5374 }, { "epoch": 3.937728937728938, "grad_norm": 0.9375, "learning_rate": 9.894713137595769e-05, "loss": 0.1364, "step": 5375 }, { "epoch": 3.9384615384615387, "grad_norm": 1.078125, "learning_rate": 9.894642833177666e-05, "loss": 0.1963, "step": 5376 }, { "epoch": 3.9391941391941394, "grad_norm": 1.8671875, "learning_rate": 9.894572505547289e-05, "loss": 0.1257, "step": 5377 }, { "epoch": 3.93992673992674, "grad_norm": 1.125, "learning_rate": 9.894502154704973e-05, "loss": 0.1676, "step": 5378 }, { "epoch": 3.940659340659341, "grad_norm": 0.84765625, "learning_rate": 9.89443178065106e-05, "loss": 0.1653, "step": 5379 }, { "epoch": 3.9413919413919416, "grad_norm": 1.375, "learning_rate": 9.89436138338588e-05, "loss": 0.1625, "step": 5380 }, { "epoch": 3.9421245421245423, "grad_norm": 0.6953125, "learning_rate": 9.894290962909777e-05, "loss": 0.1357, "step": 5381 }, { "epoch": 3.942857142857143, "grad_norm": 0.94921875, "learning_rate": 9.894220519223086e-05, "loss": 0.1487, "step": 5382 }, { "epoch": 3.943589743589744, "grad_norm": 0.9296875, "learning_rate": 9.894150052326141e-05, "loss": 0.1808, "step": 5383 }, { "epoch": 3.9443223443223445, "grad_norm": 1.0859375, "learning_rate": 9.894079562219287e-05, "loss": 0.1931, "step": 5384 }, { "epoch": 3.9450549450549453, "grad_norm": 0.99609375, "learning_rate": 9.894009048902855e-05, "loss": 0.1355, "step": 5385 }, { "epoch": 3.945787545787546, "grad_norm": 0.84375, "learning_rate": 9.893938512377187e-05, "loss": 0.1313, "step": 5386 }, { "epoch": 3.9465201465201467, "grad_norm": 0.9609375, "learning_rate": 9.89386795264262e-05, "loss": 0.1637, "step": 5387 }, { "epoch": 3.9472527472527474, "grad_norm": 0.99609375, "learning_rate": 9.893797369699489e-05, "loss": 0.1862, "step": 5388 }, { "epoch": 3.947985347985348, "grad_norm": 0.91796875, "learning_rate": 9.893726763548135e-05, "loss": 0.1871, "step": 5389 }, { "epoch": 3.948717948717949, "grad_norm": 0.71484375, "learning_rate": 9.893656134188896e-05, "loss": 0.1458, "step": 5390 }, { "epoch": 3.9494505494505496, "grad_norm": 0.67578125, "learning_rate": 9.893585481622113e-05, "loss": 0.1415, "step": 5391 }, { "epoch": 3.9501831501831504, "grad_norm": 1.15625, "learning_rate": 9.89351480584812e-05, "loss": 0.2018, "step": 5392 }, { "epoch": 3.950915750915751, "grad_norm": 1.203125, "learning_rate": 9.893444106867259e-05, "loss": 0.1379, "step": 5393 }, { "epoch": 3.951648351648352, "grad_norm": 0.91796875, "learning_rate": 9.893373384679866e-05, "loss": 0.1659, "step": 5394 }, { "epoch": 3.9523809523809526, "grad_norm": 1.1640625, "learning_rate": 9.89330263928628e-05, "loss": 0.1635, "step": 5395 }, { "epoch": 3.9531135531135533, "grad_norm": 1.140625, "learning_rate": 9.893231870686843e-05, "loss": 0.1576, "step": 5396 }, { "epoch": 3.953846153846154, "grad_norm": 0.7578125, "learning_rate": 9.89316107888189e-05, "loss": 0.1557, "step": 5397 }, { "epoch": 3.9545787545787547, "grad_norm": 0.9375, "learning_rate": 9.893090263871765e-05, "loss": 0.1666, "step": 5398 }, { "epoch": 3.9553113553113555, "grad_norm": 0.70703125, "learning_rate": 9.893019425656801e-05, "loss": 0.1534, "step": 5399 }, { "epoch": 3.956043956043956, "grad_norm": 1.1875, "learning_rate": 9.892948564237346e-05, "loss": 0.1555, "step": 5400 }, { "epoch": 3.956776556776557, "grad_norm": 1.1953125, "learning_rate": 9.892877679613733e-05, "loss": 0.1493, "step": 5401 }, { "epoch": 3.9575091575091577, "grad_norm": 0.7265625, "learning_rate": 9.8928067717863e-05, "loss": 0.1939, "step": 5402 }, { "epoch": 3.9582417582417584, "grad_norm": 1.8359375, "learning_rate": 9.892735840755393e-05, "loss": 0.154, "step": 5403 }, { "epoch": 3.958974358974359, "grad_norm": 1.2578125, "learning_rate": 9.892664886521348e-05, "loss": 0.1709, "step": 5404 }, { "epoch": 3.95970695970696, "grad_norm": 1.3203125, "learning_rate": 9.892593909084505e-05, "loss": 0.1563, "step": 5405 }, { "epoch": 3.9604395604395606, "grad_norm": 1.125, "learning_rate": 9.892522908445207e-05, "loss": 0.2022, "step": 5406 }, { "epoch": 3.9611721611721613, "grad_norm": 0.66796875, "learning_rate": 9.892451884603791e-05, "loss": 0.1512, "step": 5407 }, { "epoch": 3.961904761904762, "grad_norm": 1.3046875, "learning_rate": 9.892380837560597e-05, "loss": 0.2069, "step": 5408 }, { "epoch": 3.9626373626373628, "grad_norm": 0.9453125, "learning_rate": 9.89230976731597e-05, "loss": 0.1605, "step": 5409 }, { "epoch": 3.9633699633699635, "grad_norm": 2.203125, "learning_rate": 9.892238673870245e-05, "loss": 0.1885, "step": 5410 }, { "epoch": 3.9641025641025642, "grad_norm": 1.2109375, "learning_rate": 9.892167557223763e-05, "loss": 0.1561, "step": 5411 }, { "epoch": 3.964835164835165, "grad_norm": 0.79296875, "learning_rate": 9.89209641737687e-05, "loss": 0.1467, "step": 5412 }, { "epoch": 3.9655677655677657, "grad_norm": 1.375, "learning_rate": 9.892025254329903e-05, "loss": 0.1685, "step": 5413 }, { "epoch": 3.9663003663003664, "grad_norm": 1.0625, "learning_rate": 9.891954068083201e-05, "loss": 0.1648, "step": 5414 }, { "epoch": 3.967032967032967, "grad_norm": 0.9140625, "learning_rate": 9.89188285863711e-05, "loss": 0.1428, "step": 5415 }, { "epoch": 3.967765567765568, "grad_norm": 1.2421875, "learning_rate": 9.891811625991969e-05, "loss": 0.1557, "step": 5416 }, { "epoch": 3.9684981684981686, "grad_norm": 1.65625, "learning_rate": 9.891740370148117e-05, "loss": 0.1836, "step": 5417 }, { "epoch": 3.9692307692307693, "grad_norm": 1.265625, "learning_rate": 9.891669091105898e-05, "loss": 0.1361, "step": 5418 }, { "epoch": 3.96996336996337, "grad_norm": 1.25, "learning_rate": 9.891597788865654e-05, "loss": 0.184, "step": 5419 }, { "epoch": 3.970695970695971, "grad_norm": 1.4140625, "learning_rate": 9.891526463427724e-05, "loss": 0.1653, "step": 5420 }, { "epoch": 3.9714285714285715, "grad_norm": 1.421875, "learning_rate": 9.891455114792454e-05, "loss": 0.1863, "step": 5421 }, { "epoch": 3.9721611721611723, "grad_norm": 1.2578125, "learning_rate": 9.891383742960183e-05, "loss": 0.155, "step": 5422 }, { "epoch": 3.972893772893773, "grad_norm": 1.1875, "learning_rate": 9.89131234793125e-05, "loss": 0.2095, "step": 5423 }, { "epoch": 3.9736263736263737, "grad_norm": 1.2109375, "learning_rate": 9.891240929706002e-05, "loss": 0.1361, "step": 5424 }, { "epoch": 3.9743589743589745, "grad_norm": 1.2421875, "learning_rate": 9.891169488284781e-05, "loss": 0.1648, "step": 5425 }, { "epoch": 3.975091575091575, "grad_norm": 1.171875, "learning_rate": 9.891098023667925e-05, "loss": 0.1348, "step": 5426 }, { "epoch": 3.975824175824176, "grad_norm": 1.015625, "learning_rate": 9.89102653585578e-05, "loss": 0.1343, "step": 5427 }, { "epoch": 3.9765567765567766, "grad_norm": 0.9765625, "learning_rate": 9.890955024848688e-05, "loss": 0.1607, "step": 5428 }, { "epoch": 3.9772893772893774, "grad_norm": 1.21875, "learning_rate": 9.890883490646991e-05, "loss": 0.1776, "step": 5429 }, { "epoch": 3.978021978021978, "grad_norm": 1.5859375, "learning_rate": 9.890811933251033e-05, "loss": 0.1912, "step": 5430 }, { "epoch": 3.978754578754579, "grad_norm": 0.97265625, "learning_rate": 9.890740352661155e-05, "loss": 0.1818, "step": 5431 }, { "epoch": 3.9794871794871796, "grad_norm": 1.0390625, "learning_rate": 9.890668748877701e-05, "loss": 0.1476, "step": 5432 }, { "epoch": 3.9802197802197803, "grad_norm": 0.7890625, "learning_rate": 9.890597121901014e-05, "loss": 0.1295, "step": 5433 }, { "epoch": 3.980952380952381, "grad_norm": 0.87109375, "learning_rate": 9.890525471731438e-05, "loss": 0.1443, "step": 5434 }, { "epoch": 3.9816849816849818, "grad_norm": 1.0703125, "learning_rate": 9.890453798369313e-05, "loss": 0.1472, "step": 5435 }, { "epoch": 3.9824175824175825, "grad_norm": 0.68359375, "learning_rate": 9.890382101814987e-05, "loss": 0.1616, "step": 5436 }, { "epoch": 3.983150183150183, "grad_norm": 1.21875, "learning_rate": 9.890310382068802e-05, "loss": 0.1495, "step": 5437 }, { "epoch": 3.983882783882784, "grad_norm": 0.609375, "learning_rate": 9.890238639131099e-05, "loss": 0.1474, "step": 5438 }, { "epoch": 3.9846153846153847, "grad_norm": 1.0703125, "learning_rate": 9.890166873002225e-05, "loss": 0.1662, "step": 5439 }, { "epoch": 3.9853479853479854, "grad_norm": 1.0234375, "learning_rate": 9.890095083682523e-05, "loss": 0.1769, "step": 5440 }, { "epoch": 3.986080586080586, "grad_norm": 1.109375, "learning_rate": 9.890023271172334e-05, "loss": 0.1622, "step": 5441 }, { "epoch": 3.986813186813187, "grad_norm": 1.1640625, "learning_rate": 9.889951435472006e-05, "loss": 0.167, "step": 5442 }, { "epoch": 3.9875457875457876, "grad_norm": 1.109375, "learning_rate": 9.889879576581881e-05, "loss": 0.1821, "step": 5443 }, { "epoch": 3.9882783882783883, "grad_norm": 1.140625, "learning_rate": 9.889807694502305e-05, "loss": 0.1844, "step": 5444 }, { "epoch": 3.989010989010989, "grad_norm": 1.296875, "learning_rate": 9.889735789233624e-05, "loss": 0.2255, "step": 5445 }, { "epoch": 3.98974358974359, "grad_norm": 0.8125, "learning_rate": 9.889663860776177e-05, "loss": 0.1248, "step": 5446 }, { "epoch": 3.9904761904761905, "grad_norm": 0.91015625, "learning_rate": 9.889591909130313e-05, "loss": 0.1604, "step": 5447 }, { "epoch": 3.9912087912087912, "grad_norm": 0.7265625, "learning_rate": 9.889519934296373e-05, "loss": 0.1763, "step": 5448 }, { "epoch": 3.991941391941392, "grad_norm": 0.65234375, "learning_rate": 9.889447936274707e-05, "loss": 0.1513, "step": 5449 }, { "epoch": 3.9926739926739927, "grad_norm": 1.3984375, "learning_rate": 9.889375915065658e-05, "loss": 0.2156, "step": 5450 }, { "epoch": 3.9934065934065934, "grad_norm": 0.79296875, "learning_rate": 9.889303870669569e-05, "loss": 0.1297, "step": 5451 }, { "epoch": 3.994139194139194, "grad_norm": 1.25, "learning_rate": 9.889231803086785e-05, "loss": 0.1391, "step": 5452 }, { "epoch": 3.994871794871795, "grad_norm": 0.8359375, "learning_rate": 9.889159712317653e-05, "loss": 0.1514, "step": 5453 }, { "epoch": 3.9956043956043956, "grad_norm": 0.6484375, "learning_rate": 9.88908759836252e-05, "loss": 0.1569, "step": 5454 }, { "epoch": 3.9963369963369964, "grad_norm": 1.046875, "learning_rate": 9.889015461221729e-05, "loss": 0.1521, "step": 5455 }, { "epoch": 3.997069597069597, "grad_norm": 1.3125, "learning_rate": 9.888943300895627e-05, "loss": 0.1532, "step": 5456 }, { "epoch": 3.997802197802198, "grad_norm": 1.234375, "learning_rate": 9.888871117384558e-05, "loss": 0.1955, "step": 5457 }, { "epoch": 3.9985347985347985, "grad_norm": 2.484375, "learning_rate": 9.88879891068887e-05, "loss": 0.2063, "step": 5458 }, { "epoch": 3.9992673992673993, "grad_norm": 0.95703125, "learning_rate": 9.888726680808908e-05, "loss": 0.1629, "step": 5459 }, { "epoch": 4.0, "grad_norm": 0.8203125, "learning_rate": 9.888654427745016e-05, "loss": 0.1699, "step": 5460 }, { "epoch": 4.000732600732601, "grad_norm": 1.0390625, "learning_rate": 9.888582151497546e-05, "loss": 0.1677, "step": 5461 }, { "epoch": 4.0014652014652015, "grad_norm": 1.4765625, "learning_rate": 9.888509852066838e-05, "loss": 0.167, "step": 5462 }, { "epoch": 4.002197802197802, "grad_norm": 1.7421875, "learning_rate": 9.88843752945324e-05, "loss": 0.1495, "step": 5463 }, { "epoch": 4.002930402930403, "grad_norm": 0.85546875, "learning_rate": 9.888365183657102e-05, "loss": 0.1857, "step": 5464 }, { "epoch": 4.003663003663004, "grad_norm": 1.046875, "learning_rate": 9.888292814678765e-05, "loss": 0.1351, "step": 5465 }, { "epoch": 4.004395604395604, "grad_norm": 1.3984375, "learning_rate": 9.88822042251858e-05, "loss": 0.1446, "step": 5466 }, { "epoch": 4.005128205128205, "grad_norm": 1.140625, "learning_rate": 9.888148007176894e-05, "loss": 0.1712, "step": 5467 }, { "epoch": 4.005860805860806, "grad_norm": 0.7265625, "learning_rate": 9.888075568654053e-05, "loss": 0.139, "step": 5468 }, { "epoch": 4.006593406593407, "grad_norm": 0.55859375, "learning_rate": 9.888003106950401e-05, "loss": 0.1229, "step": 5469 }, { "epoch": 4.007326007326007, "grad_norm": 1.03125, "learning_rate": 9.88793062206629e-05, "loss": 0.1497, "step": 5470 }, { "epoch": 4.008058608058608, "grad_norm": 1.015625, "learning_rate": 9.887858114002064e-05, "loss": 0.1569, "step": 5471 }, { "epoch": 4.008791208791209, "grad_norm": 1.109375, "learning_rate": 9.887785582758073e-05, "loss": 0.1404, "step": 5472 }, { "epoch": 4.0095238095238095, "grad_norm": 0.93359375, "learning_rate": 9.887713028334662e-05, "loss": 0.171, "step": 5473 }, { "epoch": 4.01025641025641, "grad_norm": 0.9609375, "learning_rate": 9.887640450732181e-05, "loss": 0.1825, "step": 5474 }, { "epoch": 4.010989010989011, "grad_norm": 1.875, "learning_rate": 9.887567849950975e-05, "loss": 0.1648, "step": 5475 }, { "epoch": 4.011721611721612, "grad_norm": 2.390625, "learning_rate": 9.887495225991395e-05, "loss": 0.1821, "step": 5476 }, { "epoch": 4.012454212454212, "grad_norm": 1.734375, "learning_rate": 9.887422578853784e-05, "loss": 0.1433, "step": 5477 }, { "epoch": 4.013186813186813, "grad_norm": 2.109375, "learning_rate": 9.887349908538497e-05, "loss": 0.1721, "step": 5478 }, { "epoch": 4.013919413919414, "grad_norm": 0.85546875, "learning_rate": 9.887277215045876e-05, "loss": 0.1677, "step": 5479 }, { "epoch": 4.014652014652015, "grad_norm": 1.0390625, "learning_rate": 9.887204498376275e-05, "loss": 0.1752, "step": 5480 }, { "epoch": 4.015384615384615, "grad_norm": 1.4140625, "learning_rate": 9.887131758530037e-05, "loss": 0.1269, "step": 5481 }, { "epoch": 4.016117216117216, "grad_norm": 0.984375, "learning_rate": 9.887058995507511e-05, "loss": 0.1316, "step": 5482 }, { "epoch": 4.016849816849817, "grad_norm": 1.3515625, "learning_rate": 9.886986209309051e-05, "loss": 0.1826, "step": 5483 }, { "epoch": 4.0175824175824175, "grad_norm": 0.462890625, "learning_rate": 9.886913399935002e-05, "loss": 0.1355, "step": 5484 }, { "epoch": 4.018315018315018, "grad_norm": 1.1171875, "learning_rate": 9.886840567385711e-05, "loss": 0.1789, "step": 5485 }, { "epoch": 4.019047619047619, "grad_norm": 1.328125, "learning_rate": 9.886767711661531e-05, "loss": 0.1602, "step": 5486 }, { "epoch": 4.01978021978022, "grad_norm": 1.375, "learning_rate": 9.886694832762808e-05, "loss": 0.1327, "step": 5487 }, { "epoch": 4.02051282051282, "grad_norm": 1.3671875, "learning_rate": 9.886621930689892e-05, "loss": 0.1488, "step": 5488 }, { "epoch": 4.021245421245421, "grad_norm": 1.9453125, "learning_rate": 9.886549005443133e-05, "loss": 0.1499, "step": 5489 }, { "epoch": 4.021978021978022, "grad_norm": 0.73046875, "learning_rate": 9.88647605702288e-05, "loss": 0.1422, "step": 5490 }, { "epoch": 4.022710622710623, "grad_norm": 0.94921875, "learning_rate": 9.886403085429483e-05, "loss": 0.1544, "step": 5491 }, { "epoch": 4.023443223443223, "grad_norm": 2.171875, "learning_rate": 9.88633009066329e-05, "loss": 0.1707, "step": 5492 }, { "epoch": 4.024175824175824, "grad_norm": 1.1328125, "learning_rate": 9.886257072724653e-05, "loss": 0.1081, "step": 5493 }, { "epoch": 4.024908424908425, "grad_norm": 0.9453125, "learning_rate": 9.88618403161392e-05, "loss": 0.1746, "step": 5494 }, { "epoch": 4.0256410256410255, "grad_norm": 1.0, "learning_rate": 9.886110967331442e-05, "loss": 0.1934, "step": 5495 }, { "epoch": 4.026373626373626, "grad_norm": 1.3125, "learning_rate": 9.886037879877567e-05, "loss": 0.1493, "step": 5496 }, { "epoch": 4.027106227106227, "grad_norm": 1.1796875, "learning_rate": 9.88596476925265e-05, "loss": 0.1579, "step": 5497 }, { "epoch": 4.027838827838828, "grad_norm": 1.3984375, "learning_rate": 9.885891635457036e-05, "loss": 0.1705, "step": 5498 }, { "epoch": 4.0285714285714285, "grad_norm": 0.859375, "learning_rate": 9.885818478491078e-05, "loss": 0.182, "step": 5499 }, { "epoch": 4.029304029304029, "grad_norm": 1.140625, "learning_rate": 9.885745298355128e-05, "loss": 0.1961, "step": 5500 }, { "epoch": 4.03003663003663, "grad_norm": 1.6953125, "learning_rate": 9.885672095049532e-05, "loss": 0.1819, "step": 5501 }, { "epoch": 4.030769230769231, "grad_norm": 1.3046875, "learning_rate": 9.885598868574646e-05, "loss": 0.1369, "step": 5502 }, { "epoch": 4.031501831501831, "grad_norm": 1.453125, "learning_rate": 9.885525618930817e-05, "loss": 0.1802, "step": 5503 }, { "epoch": 4.032234432234432, "grad_norm": 1.0390625, "learning_rate": 9.885452346118398e-05, "loss": 0.1751, "step": 5504 }, { "epoch": 4.032967032967033, "grad_norm": 1.40625, "learning_rate": 9.88537905013774e-05, "loss": 0.1517, "step": 5505 }, { "epoch": 4.033699633699634, "grad_norm": 0.96875, "learning_rate": 9.885305730989192e-05, "loss": 0.1423, "step": 5506 }, { "epoch": 4.034432234432234, "grad_norm": 0.75, "learning_rate": 9.885232388673108e-05, "loss": 0.1559, "step": 5507 }, { "epoch": 4.035164835164835, "grad_norm": 0.84765625, "learning_rate": 9.885159023189838e-05, "loss": 0.1317, "step": 5508 }, { "epoch": 4.035897435897436, "grad_norm": 0.859375, "learning_rate": 9.885085634539733e-05, "loss": 0.1469, "step": 5509 }, { "epoch": 4.0366300366300365, "grad_norm": 1.078125, "learning_rate": 9.885012222723147e-05, "loss": 0.1419, "step": 5510 }, { "epoch": 4.037362637362637, "grad_norm": 1.65625, "learning_rate": 9.884938787740428e-05, "loss": 0.1737, "step": 5511 }, { "epoch": 4.038095238095238, "grad_norm": 0.859375, "learning_rate": 9.884865329591934e-05, "loss": 0.14, "step": 5512 }, { "epoch": 4.038827838827839, "grad_norm": 0.640625, "learning_rate": 9.88479184827801e-05, "loss": 0.1129, "step": 5513 }, { "epoch": 4.039560439560439, "grad_norm": 0.71875, "learning_rate": 9.884718343799012e-05, "loss": 0.1443, "step": 5514 }, { "epoch": 4.04029304029304, "grad_norm": 1.3125, "learning_rate": 9.884644816155291e-05, "loss": 0.1556, "step": 5515 }, { "epoch": 4.041025641025641, "grad_norm": 1.0703125, "learning_rate": 9.8845712653472e-05, "loss": 0.1618, "step": 5516 }, { "epoch": 4.041758241758242, "grad_norm": 0.8984375, "learning_rate": 9.884497691375091e-05, "loss": 0.1332, "step": 5517 }, { "epoch": 4.042490842490842, "grad_norm": 0.91015625, "learning_rate": 9.884424094239316e-05, "loss": 0.1468, "step": 5518 }, { "epoch": 4.043223443223443, "grad_norm": 0.96875, "learning_rate": 9.884350473940227e-05, "loss": 0.1866, "step": 5519 }, { "epoch": 4.043956043956044, "grad_norm": 1.578125, "learning_rate": 9.88427683047818e-05, "loss": 0.1569, "step": 5520 }, { "epoch": 4.0446886446886445, "grad_norm": 1.3046875, "learning_rate": 9.884203163853524e-05, "loss": 0.1695, "step": 5521 }, { "epoch": 4.045421245421245, "grad_norm": 1.3359375, "learning_rate": 9.884129474066614e-05, "loss": 0.1724, "step": 5522 }, { "epoch": 4.046153846153846, "grad_norm": 0.9140625, "learning_rate": 9.884055761117804e-05, "loss": 0.1371, "step": 5523 }, { "epoch": 4.046886446886447, "grad_norm": 1.03125, "learning_rate": 9.883982025007444e-05, "loss": 0.1306, "step": 5524 }, { "epoch": 4.0476190476190474, "grad_norm": 0.7578125, "learning_rate": 9.883908265735891e-05, "loss": 0.1282, "step": 5525 }, { "epoch": 4.048351648351648, "grad_norm": 0.69921875, "learning_rate": 9.883834483303496e-05, "loss": 0.1459, "step": 5526 }, { "epoch": 4.049084249084249, "grad_norm": 0.7734375, "learning_rate": 9.883760677710614e-05, "loss": 0.1605, "step": 5527 }, { "epoch": 4.04981684981685, "grad_norm": 0.9375, "learning_rate": 9.883686848957596e-05, "loss": 0.2024, "step": 5528 }, { "epoch": 4.05054945054945, "grad_norm": 0.8984375, "learning_rate": 9.883612997044799e-05, "loss": 0.1588, "step": 5529 }, { "epoch": 4.051282051282051, "grad_norm": 1.453125, "learning_rate": 9.883539121972573e-05, "loss": 0.1719, "step": 5530 }, { "epoch": 4.052014652014652, "grad_norm": 1.0625, "learning_rate": 9.883465223741277e-05, "loss": 0.1232, "step": 5531 }, { "epoch": 4.052747252747253, "grad_norm": 1.0625, "learning_rate": 9.883391302351261e-05, "loss": 0.1864, "step": 5532 }, { "epoch": 4.053479853479853, "grad_norm": 1.1640625, "learning_rate": 9.883317357802882e-05, "loss": 0.178, "step": 5533 }, { "epoch": 4.054212454212454, "grad_norm": 0.9921875, "learning_rate": 9.883243390096493e-05, "loss": 0.1479, "step": 5534 }, { "epoch": 4.054945054945055, "grad_norm": 1.359375, "learning_rate": 9.883169399232445e-05, "loss": 0.1565, "step": 5535 }, { "epoch": 4.0556776556776555, "grad_norm": 0.92578125, "learning_rate": 9.8830953852111e-05, "loss": 0.13, "step": 5536 }, { "epoch": 4.056410256410256, "grad_norm": 0.9921875, "learning_rate": 9.883021348032805e-05, "loss": 0.1618, "step": 5537 }, { "epoch": 4.057142857142857, "grad_norm": 0.87890625, "learning_rate": 9.88294728769792e-05, "loss": 0.1455, "step": 5538 }, { "epoch": 4.057875457875458, "grad_norm": 1.0078125, "learning_rate": 9.882873204206798e-05, "loss": 0.1382, "step": 5539 }, { "epoch": 4.058608058608058, "grad_norm": 0.515625, "learning_rate": 9.882799097559793e-05, "loss": 0.12, "step": 5540 }, { "epoch": 4.059340659340659, "grad_norm": 0.6328125, "learning_rate": 9.882724967757262e-05, "loss": 0.1537, "step": 5541 }, { "epoch": 4.06007326007326, "grad_norm": 0.6796875, "learning_rate": 9.882650814799558e-05, "loss": 0.1579, "step": 5542 }, { "epoch": 4.060805860805861, "grad_norm": 0.62890625, "learning_rate": 9.882576638687039e-05, "loss": 0.1207, "step": 5543 }, { "epoch": 4.061538461538461, "grad_norm": 1.0, "learning_rate": 9.882502439420058e-05, "loss": 0.1693, "step": 5544 }, { "epoch": 4.062271062271062, "grad_norm": 0.96875, "learning_rate": 9.882428216998971e-05, "loss": 0.1351, "step": 5545 }, { "epoch": 4.063003663003663, "grad_norm": 0.74609375, "learning_rate": 9.882353971424134e-05, "loss": 0.1311, "step": 5546 }, { "epoch": 4.0637362637362635, "grad_norm": 1.0546875, "learning_rate": 9.882279702695904e-05, "loss": 0.1754, "step": 5547 }, { "epoch": 4.064468864468864, "grad_norm": 0.796875, "learning_rate": 9.882205410814635e-05, "loss": 0.1292, "step": 5548 }, { "epoch": 4.065201465201465, "grad_norm": 1.6875, "learning_rate": 9.882131095780683e-05, "loss": 0.1495, "step": 5549 }, { "epoch": 4.065934065934066, "grad_norm": 1.515625, "learning_rate": 9.882056757594405e-05, "loss": 0.1582, "step": 5550 }, { "epoch": 4.066666666666666, "grad_norm": 1.40625, "learning_rate": 9.881982396256158e-05, "loss": 0.1578, "step": 5551 }, { "epoch": 4.067399267399267, "grad_norm": 1.6640625, "learning_rate": 9.881908011766296e-05, "loss": 0.2013, "step": 5552 }, { "epoch": 4.068131868131868, "grad_norm": 0.828125, "learning_rate": 9.881833604125175e-05, "loss": 0.1433, "step": 5553 }, { "epoch": 4.068864468864469, "grad_norm": 2.1875, "learning_rate": 9.881759173333155e-05, "loss": 0.1626, "step": 5554 }, { "epoch": 4.069597069597069, "grad_norm": 1.8359375, "learning_rate": 9.88168471939059e-05, "loss": 0.1555, "step": 5555 }, { "epoch": 4.07032967032967, "grad_norm": 1.8125, "learning_rate": 9.881610242297838e-05, "loss": 0.1702, "step": 5556 }, { "epoch": 4.071062271062271, "grad_norm": 1.9921875, "learning_rate": 9.881535742055255e-05, "loss": 0.1523, "step": 5557 }, { "epoch": 4.0717948717948715, "grad_norm": 0.87109375, "learning_rate": 9.881461218663197e-05, "loss": 0.1401, "step": 5558 }, { "epoch": 4.072527472527472, "grad_norm": 1.1328125, "learning_rate": 9.881386672122024e-05, "loss": 0.1627, "step": 5559 }, { "epoch": 4.073260073260073, "grad_norm": 1.515625, "learning_rate": 9.88131210243209e-05, "loss": 0.1718, "step": 5560 }, { "epoch": 4.073992673992674, "grad_norm": 0.703125, "learning_rate": 9.881237509593753e-05, "loss": 0.1433, "step": 5561 }, { "epoch": 4.0747252747252745, "grad_norm": 0.65234375, "learning_rate": 9.881162893607374e-05, "loss": 0.1543, "step": 5562 }, { "epoch": 4.075457875457875, "grad_norm": 0.953125, "learning_rate": 9.881088254473307e-05, "loss": 0.1471, "step": 5563 }, { "epoch": 4.076190476190476, "grad_norm": 0.88671875, "learning_rate": 9.881013592191909e-05, "loss": 0.1448, "step": 5564 }, { "epoch": 4.076923076923077, "grad_norm": 0.9375, "learning_rate": 9.880938906763538e-05, "loss": 0.1617, "step": 5565 }, { "epoch": 4.077655677655677, "grad_norm": 1.625, "learning_rate": 9.880864198188555e-05, "loss": 0.1704, "step": 5566 }, { "epoch": 4.078388278388278, "grad_norm": 1.90625, "learning_rate": 9.880789466467313e-05, "loss": 0.1719, "step": 5567 }, { "epoch": 4.079120879120879, "grad_norm": 1.0625, "learning_rate": 9.880714711600176e-05, "loss": 0.1683, "step": 5568 }, { "epoch": 4.07985347985348, "grad_norm": 1.09375, "learning_rate": 9.880639933587497e-05, "loss": 0.144, "step": 5569 }, { "epoch": 4.08058608058608, "grad_norm": 1.234375, "learning_rate": 9.880565132429638e-05, "loss": 0.1664, "step": 5570 }, { "epoch": 4.081318681318681, "grad_norm": 2.03125, "learning_rate": 9.880490308126954e-05, "loss": 0.1507, "step": 5571 }, { "epoch": 4.082051282051282, "grad_norm": 1.1328125, "learning_rate": 9.880415460679805e-05, "loss": 0.1634, "step": 5572 }, { "epoch": 4.0827838827838825, "grad_norm": 1.390625, "learning_rate": 9.880340590088552e-05, "loss": 0.1472, "step": 5573 }, { "epoch": 4.083516483516483, "grad_norm": 0.90234375, "learning_rate": 9.880265696353547e-05, "loss": 0.1779, "step": 5574 }, { "epoch": 4.084249084249084, "grad_norm": 0.7734375, "learning_rate": 9.880190779475159e-05, "loss": 0.1249, "step": 5575 }, { "epoch": 4.084981684981685, "grad_norm": 1.03125, "learning_rate": 9.880115839453739e-05, "loss": 0.1542, "step": 5576 }, { "epoch": 4.085714285714285, "grad_norm": 0.98046875, "learning_rate": 9.880040876289649e-05, "loss": 0.1379, "step": 5577 }, { "epoch": 4.086446886446886, "grad_norm": 1.7734375, "learning_rate": 9.879965889983244e-05, "loss": 0.1941, "step": 5578 }, { "epoch": 4.087179487179487, "grad_norm": 1.1015625, "learning_rate": 9.879890880534891e-05, "loss": 0.1703, "step": 5579 }, { "epoch": 4.087912087912088, "grad_norm": 0.69921875, "learning_rate": 9.879815847944944e-05, "loss": 0.1525, "step": 5580 }, { "epoch": 4.088644688644688, "grad_norm": 0.81640625, "learning_rate": 9.879740792213763e-05, "loss": 0.1615, "step": 5581 }, { "epoch": 4.089377289377289, "grad_norm": 1.015625, "learning_rate": 9.879665713341708e-05, "loss": 0.1722, "step": 5582 }, { "epoch": 4.09010989010989, "grad_norm": 1.3515625, "learning_rate": 9.87959061132914e-05, "loss": 0.155, "step": 5583 }, { "epoch": 4.0908424908424905, "grad_norm": 0.6640625, "learning_rate": 9.879515486176419e-05, "loss": 0.1846, "step": 5584 }, { "epoch": 4.091575091575091, "grad_norm": 0.859375, "learning_rate": 9.879440337883901e-05, "loss": 0.1357, "step": 5585 }, { "epoch": 4.092307692307692, "grad_norm": 0.84375, "learning_rate": 9.879365166451952e-05, "loss": 0.1396, "step": 5586 }, { "epoch": 4.093040293040293, "grad_norm": 0.9609375, "learning_rate": 9.879289971880927e-05, "loss": 0.1734, "step": 5587 }, { "epoch": 4.093772893772893, "grad_norm": 1.0078125, "learning_rate": 9.87921475417119e-05, "loss": 0.1491, "step": 5588 }, { "epoch": 4.094505494505494, "grad_norm": 1.1796875, "learning_rate": 9.8791395133231e-05, "loss": 0.1651, "step": 5589 }, { "epoch": 4.095238095238095, "grad_norm": 1.7109375, "learning_rate": 9.879064249337016e-05, "loss": 0.156, "step": 5590 }, { "epoch": 4.095970695970696, "grad_norm": 0.921875, "learning_rate": 9.8789889622133e-05, "loss": 0.1563, "step": 5591 }, { "epoch": 4.096703296703296, "grad_norm": 0.83984375, "learning_rate": 9.878913651952312e-05, "loss": 0.1622, "step": 5592 }, { "epoch": 4.097435897435897, "grad_norm": 1.34375, "learning_rate": 9.878838318554416e-05, "loss": 0.1794, "step": 5593 }, { "epoch": 4.098168498168498, "grad_norm": 1.1796875, "learning_rate": 9.878762962019969e-05, "loss": 0.1452, "step": 5594 }, { "epoch": 4.0989010989010985, "grad_norm": 0.87109375, "learning_rate": 9.878687582349332e-05, "loss": 0.141, "step": 5595 }, { "epoch": 4.099633699633699, "grad_norm": 0.84765625, "learning_rate": 9.878612179542868e-05, "loss": 0.1378, "step": 5596 }, { "epoch": 4.1003663003663, "grad_norm": 0.9140625, "learning_rate": 9.878536753600938e-05, "loss": 0.1625, "step": 5597 }, { "epoch": 4.101098901098901, "grad_norm": 0.8046875, "learning_rate": 9.878461304523907e-05, "loss": 0.1356, "step": 5598 }, { "epoch": 4.1018315018315015, "grad_norm": 1.234375, "learning_rate": 9.87838583231213e-05, "loss": 0.1682, "step": 5599 }, { "epoch": 4.102564102564102, "grad_norm": 1.328125, "learning_rate": 9.878310336965971e-05, "loss": 0.1916, "step": 5600 }, { "epoch": 4.103296703296703, "grad_norm": 1.0625, "learning_rate": 9.87823481848579e-05, "loss": 0.1382, "step": 5601 }, { "epoch": 4.104029304029304, "grad_norm": 1.359375, "learning_rate": 9.878159276871956e-05, "loss": 0.1874, "step": 5602 }, { "epoch": 4.104761904761904, "grad_norm": 0.96484375, "learning_rate": 9.878083712124823e-05, "loss": 0.1449, "step": 5603 }, { "epoch": 4.105494505494505, "grad_norm": 1.125, "learning_rate": 9.878008124244756e-05, "loss": 0.1634, "step": 5604 }, { "epoch": 4.106227106227106, "grad_norm": 0.88671875, "learning_rate": 9.877932513232118e-05, "loss": 0.1309, "step": 5605 }, { "epoch": 4.106959706959707, "grad_norm": 0.96484375, "learning_rate": 9.877856879087269e-05, "loss": 0.1321, "step": 5606 }, { "epoch": 4.107692307692307, "grad_norm": 1.09375, "learning_rate": 9.877781221810575e-05, "loss": 0.1192, "step": 5607 }, { "epoch": 4.108424908424908, "grad_norm": 1.28125, "learning_rate": 9.877705541402393e-05, "loss": 0.1946, "step": 5608 }, { "epoch": 4.109157509157509, "grad_norm": 0.83984375, "learning_rate": 9.87762983786309e-05, "loss": 0.1769, "step": 5609 }, { "epoch": 4.1098901098901095, "grad_norm": 0.9921875, "learning_rate": 9.87755411119303e-05, "loss": 0.1729, "step": 5610 }, { "epoch": 4.11062271062271, "grad_norm": 1.21875, "learning_rate": 9.877478361392573e-05, "loss": 0.1388, "step": 5611 }, { "epoch": 4.111355311355311, "grad_norm": 0.89453125, "learning_rate": 9.877402588462081e-05, "loss": 0.1501, "step": 5612 }, { "epoch": 4.112087912087912, "grad_norm": 0.9453125, "learning_rate": 9.877326792401919e-05, "loss": 0.1441, "step": 5613 }, { "epoch": 4.112820512820512, "grad_norm": 0.7109375, "learning_rate": 9.877250973212448e-05, "loss": 0.1339, "step": 5614 }, { "epoch": 4.113553113553113, "grad_norm": 1.75, "learning_rate": 9.877175130894036e-05, "loss": 0.1762, "step": 5615 }, { "epoch": 4.114285714285714, "grad_norm": 1.1875, "learning_rate": 9.877099265447041e-05, "loss": 0.1265, "step": 5616 }, { "epoch": 4.115018315018315, "grad_norm": 0.921875, "learning_rate": 9.87702337687183e-05, "loss": 0.1946, "step": 5617 }, { "epoch": 4.115750915750915, "grad_norm": 1.2265625, "learning_rate": 9.876947465168764e-05, "loss": 0.1694, "step": 5618 }, { "epoch": 4.116483516483516, "grad_norm": 1.015625, "learning_rate": 9.876871530338209e-05, "loss": 0.1438, "step": 5619 }, { "epoch": 4.117216117216117, "grad_norm": 0.90625, "learning_rate": 9.876795572380528e-05, "loss": 0.1556, "step": 5620 }, { "epoch": 4.1179487179487175, "grad_norm": 1.7734375, "learning_rate": 9.876719591296086e-05, "loss": 0.1822, "step": 5621 }, { "epoch": 4.118681318681318, "grad_norm": 0.9296875, "learning_rate": 9.876643587085244e-05, "loss": 0.1281, "step": 5622 }, { "epoch": 4.119413919413919, "grad_norm": 1.1171875, "learning_rate": 9.87656755974837e-05, "loss": 0.1594, "step": 5623 }, { "epoch": 4.12014652014652, "grad_norm": 1.265625, "learning_rate": 9.876491509285825e-05, "loss": 0.152, "step": 5624 }, { "epoch": 4.1208791208791204, "grad_norm": 2.09375, "learning_rate": 9.876415435697974e-05, "loss": 0.1543, "step": 5625 }, { "epoch": 4.121611721611721, "grad_norm": 1.484375, "learning_rate": 9.876339338985184e-05, "loss": 0.1435, "step": 5626 }, { "epoch": 4.122344322344322, "grad_norm": 0.953125, "learning_rate": 9.876263219147816e-05, "loss": 0.1616, "step": 5627 }, { "epoch": 4.123076923076923, "grad_norm": 0.74609375, "learning_rate": 9.876187076186238e-05, "loss": 0.1381, "step": 5628 }, { "epoch": 4.123809523809523, "grad_norm": 1.3125, "learning_rate": 9.876110910100813e-05, "loss": 0.1637, "step": 5629 }, { "epoch": 4.124542124542124, "grad_norm": 1.3671875, "learning_rate": 9.876034720891907e-05, "loss": 0.1428, "step": 5630 }, { "epoch": 4.125274725274725, "grad_norm": 0.97265625, "learning_rate": 9.875958508559883e-05, "loss": 0.1941, "step": 5631 }, { "epoch": 4.126007326007326, "grad_norm": 0.86328125, "learning_rate": 9.87588227310511e-05, "loss": 0.1238, "step": 5632 }, { "epoch": 4.126739926739926, "grad_norm": 0.8046875, "learning_rate": 9.875806014527948e-05, "loss": 0.1462, "step": 5633 }, { "epoch": 4.127472527472527, "grad_norm": 0.7421875, "learning_rate": 9.875729732828766e-05, "loss": 0.1255, "step": 5634 }, { "epoch": 4.128205128205128, "grad_norm": 1.1015625, "learning_rate": 9.875653428007929e-05, "loss": 0.1233, "step": 5635 }, { "epoch": 4.128937728937729, "grad_norm": 0.98828125, "learning_rate": 9.875577100065802e-05, "loss": 0.1852, "step": 5636 }, { "epoch": 4.12967032967033, "grad_norm": 1.0390625, "learning_rate": 9.875500749002752e-05, "loss": 0.1684, "step": 5637 }, { "epoch": 4.130402930402931, "grad_norm": 1.25, "learning_rate": 9.875424374819143e-05, "loss": 0.1137, "step": 5638 }, { "epoch": 4.131135531135532, "grad_norm": 0.98828125, "learning_rate": 9.87534797751534e-05, "loss": 0.1855, "step": 5639 }, { "epoch": 4.131868131868132, "grad_norm": 0.9765625, "learning_rate": 9.875271557091713e-05, "loss": 0.1563, "step": 5640 }, { "epoch": 4.132600732600733, "grad_norm": 1.21875, "learning_rate": 9.875195113548627e-05, "loss": 0.1438, "step": 5641 }, { "epoch": 4.133333333333334, "grad_norm": 1.2421875, "learning_rate": 9.875118646886444e-05, "loss": 0.1299, "step": 5642 }, { "epoch": 4.1340659340659345, "grad_norm": 0.71875, "learning_rate": 9.875042157105536e-05, "loss": 0.1118, "step": 5643 }, { "epoch": 4.134798534798535, "grad_norm": 0.59765625, "learning_rate": 9.874965644206265e-05, "loss": 0.1415, "step": 5644 }, { "epoch": 4.135531135531136, "grad_norm": 0.98828125, "learning_rate": 9.874889108189002e-05, "loss": 0.176, "step": 5645 }, { "epoch": 4.136263736263737, "grad_norm": 1.03125, "learning_rate": 9.87481254905411e-05, "loss": 0.1577, "step": 5646 }, { "epoch": 4.136996336996337, "grad_norm": 1.09375, "learning_rate": 9.874735966801956e-05, "loss": 0.1446, "step": 5647 }, { "epoch": 4.137728937728938, "grad_norm": 0.83984375, "learning_rate": 9.874659361432913e-05, "loss": 0.1374, "step": 5648 }, { "epoch": 4.138461538461539, "grad_norm": 0.98046875, "learning_rate": 9.874582732947337e-05, "loss": 0.1371, "step": 5649 }, { "epoch": 4.13919413919414, "grad_norm": 0.62890625, "learning_rate": 9.874506081345604e-05, "loss": 0.1589, "step": 5650 }, { "epoch": 4.13992673992674, "grad_norm": 0.87109375, "learning_rate": 9.87442940662808e-05, "loss": 0.1272, "step": 5651 }, { "epoch": 4.140659340659341, "grad_norm": 0.9765625, "learning_rate": 9.874352708795129e-05, "loss": 0.1361, "step": 5652 }, { "epoch": 4.141391941391942, "grad_norm": 0.94140625, "learning_rate": 9.874275987847122e-05, "loss": 0.1521, "step": 5653 }, { "epoch": 4.1421245421245425, "grad_norm": 1.34375, "learning_rate": 9.874199243784423e-05, "loss": 0.1708, "step": 5654 }, { "epoch": 4.142857142857143, "grad_norm": 1.1796875, "learning_rate": 9.874122476607402e-05, "loss": 0.1518, "step": 5655 }, { "epoch": 4.143589743589744, "grad_norm": 0.6484375, "learning_rate": 9.874045686316429e-05, "loss": 0.1359, "step": 5656 }, { "epoch": 4.144322344322345, "grad_norm": 0.671875, "learning_rate": 9.873968872911868e-05, "loss": 0.123, "step": 5657 }, { "epoch": 4.145054945054945, "grad_norm": 1.4296875, "learning_rate": 9.873892036394088e-05, "loss": 0.2059, "step": 5658 }, { "epoch": 4.145787545787546, "grad_norm": 1.6484375, "learning_rate": 9.873815176763458e-05, "loss": 0.1492, "step": 5659 }, { "epoch": 4.146520146520147, "grad_norm": 1.1328125, "learning_rate": 9.873738294020346e-05, "loss": 0.1684, "step": 5660 }, { "epoch": 4.147252747252748, "grad_norm": 1.015625, "learning_rate": 9.87366138816512e-05, "loss": 0.1403, "step": 5661 }, { "epoch": 4.147985347985348, "grad_norm": 0.81640625, "learning_rate": 9.87358445919815e-05, "loss": 0.1651, "step": 5662 }, { "epoch": 4.148717948717949, "grad_norm": 1.21875, "learning_rate": 9.873507507119802e-05, "loss": 0.1872, "step": 5663 }, { "epoch": 4.14945054945055, "grad_norm": 1.4375, "learning_rate": 9.873430531930446e-05, "loss": 0.171, "step": 5664 }, { "epoch": 4.1501831501831505, "grad_norm": 1.0234375, "learning_rate": 9.873353533630453e-05, "loss": 0.1623, "step": 5665 }, { "epoch": 4.150915750915751, "grad_norm": 1.3203125, "learning_rate": 9.873276512220187e-05, "loss": 0.1306, "step": 5666 }, { "epoch": 4.151648351648352, "grad_norm": 1.3828125, "learning_rate": 9.873199467700022e-05, "loss": 0.1405, "step": 5667 }, { "epoch": 4.152380952380953, "grad_norm": 1.4921875, "learning_rate": 9.873122400070324e-05, "loss": 0.1173, "step": 5668 }, { "epoch": 4.1531135531135535, "grad_norm": 1.234375, "learning_rate": 9.873045309331463e-05, "loss": 0.1529, "step": 5669 }, { "epoch": 4.153846153846154, "grad_norm": 1.015625, "learning_rate": 9.872968195483808e-05, "loss": 0.1645, "step": 5670 }, { "epoch": 4.154578754578755, "grad_norm": 0.90234375, "learning_rate": 9.87289105852773e-05, "loss": 0.1403, "step": 5671 }, { "epoch": 4.155311355311356, "grad_norm": 0.99609375, "learning_rate": 9.872813898463598e-05, "loss": 0.1782, "step": 5672 }, { "epoch": 4.156043956043956, "grad_norm": 1.375, "learning_rate": 9.872736715291781e-05, "loss": 0.1418, "step": 5673 }, { "epoch": 4.156776556776557, "grad_norm": 0.69140625, "learning_rate": 9.87265950901265e-05, "loss": 0.1366, "step": 5674 }, { "epoch": 4.157509157509158, "grad_norm": 0.82421875, "learning_rate": 9.872582279626573e-05, "loss": 0.1394, "step": 5675 }, { "epoch": 4.158241758241759, "grad_norm": 0.73046875, "learning_rate": 9.872505027133922e-05, "loss": 0.1638, "step": 5676 }, { "epoch": 4.158974358974359, "grad_norm": 0.94921875, "learning_rate": 9.872427751535068e-05, "loss": 0.1452, "step": 5677 }, { "epoch": 4.15970695970696, "grad_norm": 1.2890625, "learning_rate": 9.872350452830376e-05, "loss": 0.2221, "step": 5678 }, { "epoch": 4.160439560439561, "grad_norm": 0.65625, "learning_rate": 9.872273131020221e-05, "loss": 0.1172, "step": 5679 }, { "epoch": 4.1611721611721615, "grad_norm": 1.453125, "learning_rate": 9.872195786104972e-05, "loss": 0.1602, "step": 5680 }, { "epoch": 4.161904761904762, "grad_norm": 0.85546875, "learning_rate": 9.872118418085001e-05, "loss": 0.1276, "step": 5681 }, { "epoch": 4.162637362637363, "grad_norm": 0.71484375, "learning_rate": 9.872041026960678e-05, "loss": 0.1459, "step": 5682 }, { "epoch": 4.163369963369964, "grad_norm": 1.3984375, "learning_rate": 9.871963612732374e-05, "loss": 0.1777, "step": 5683 }, { "epoch": 4.164102564102564, "grad_norm": 1.203125, "learning_rate": 9.871886175400457e-05, "loss": 0.1464, "step": 5684 }, { "epoch": 4.164835164835165, "grad_norm": 1.1484375, "learning_rate": 9.871808714965301e-05, "loss": 0.1703, "step": 5685 }, { "epoch": 4.165567765567766, "grad_norm": 0.93359375, "learning_rate": 9.871731231427277e-05, "loss": 0.1754, "step": 5686 }, { "epoch": 4.166300366300367, "grad_norm": 1.328125, "learning_rate": 9.871653724786757e-05, "loss": 0.1727, "step": 5687 }, { "epoch": 4.167032967032967, "grad_norm": 1.3359375, "learning_rate": 9.87157619504411e-05, "loss": 0.1938, "step": 5688 }, { "epoch": 4.167765567765568, "grad_norm": 0.9765625, "learning_rate": 9.871498642199708e-05, "loss": 0.1612, "step": 5689 }, { "epoch": 4.168498168498169, "grad_norm": 1.703125, "learning_rate": 9.871421066253924e-05, "loss": 0.1664, "step": 5690 }, { "epoch": 4.1692307692307695, "grad_norm": 0.890625, "learning_rate": 9.871343467207129e-05, "loss": 0.1559, "step": 5691 }, { "epoch": 4.16996336996337, "grad_norm": 1.0703125, "learning_rate": 9.871265845059696e-05, "loss": 0.1447, "step": 5692 }, { "epoch": 4.170695970695971, "grad_norm": 0.953125, "learning_rate": 9.871188199811994e-05, "loss": 0.1602, "step": 5693 }, { "epoch": 4.171428571428572, "grad_norm": 1.25, "learning_rate": 9.871110531464395e-05, "loss": 0.1401, "step": 5694 }, { "epoch": 4.172161172161172, "grad_norm": 1.109375, "learning_rate": 9.871032840017276e-05, "loss": 0.1275, "step": 5695 }, { "epoch": 4.172893772893773, "grad_norm": 0.80859375, "learning_rate": 9.870955125471004e-05, "loss": 0.1623, "step": 5696 }, { "epoch": 4.173626373626374, "grad_norm": 1.015625, "learning_rate": 9.870877387825955e-05, "loss": 0.1414, "step": 5697 }, { "epoch": 4.174358974358975, "grad_norm": 1.015625, "learning_rate": 9.870799627082498e-05, "loss": 0.1529, "step": 5698 }, { "epoch": 4.175091575091575, "grad_norm": 0.78515625, "learning_rate": 9.870721843241009e-05, "loss": 0.1579, "step": 5699 }, { "epoch": 4.175824175824176, "grad_norm": 0.91015625, "learning_rate": 9.870644036301857e-05, "loss": 0.1503, "step": 5700 }, { "epoch": 4.176556776556777, "grad_norm": 1.296875, "learning_rate": 9.87056620626542e-05, "loss": 0.1417, "step": 5701 }, { "epoch": 4.1772893772893775, "grad_norm": 1.3984375, "learning_rate": 9.870488353132065e-05, "loss": 0.1625, "step": 5702 }, { "epoch": 4.178021978021978, "grad_norm": 2.046875, "learning_rate": 9.870410476902169e-05, "loss": 0.1484, "step": 5703 }, { "epoch": 4.178754578754579, "grad_norm": 1.2734375, "learning_rate": 9.870332577576104e-05, "loss": 0.1659, "step": 5704 }, { "epoch": 4.17948717948718, "grad_norm": 1.65625, "learning_rate": 9.870254655154245e-05, "loss": 0.1774, "step": 5705 }, { "epoch": 4.1802197802197805, "grad_norm": 1.0546875, "learning_rate": 9.87017670963696e-05, "loss": 0.1479, "step": 5706 }, { "epoch": 4.180952380952381, "grad_norm": 0.9609375, "learning_rate": 9.870098741024629e-05, "loss": 0.1818, "step": 5707 }, { "epoch": 4.181684981684982, "grad_norm": 1.3984375, "learning_rate": 9.87002074931762e-05, "loss": 0.1478, "step": 5708 }, { "epoch": 4.182417582417583, "grad_norm": 0.98046875, "learning_rate": 9.86994273451631e-05, "loss": 0.1743, "step": 5709 }, { "epoch": 4.183150183150183, "grad_norm": 1.015625, "learning_rate": 9.869864696621074e-05, "loss": 0.1514, "step": 5710 }, { "epoch": 4.183882783882784, "grad_norm": 0.86328125, "learning_rate": 9.869786635632282e-05, "loss": 0.1531, "step": 5711 }, { "epoch": 4.184615384615385, "grad_norm": 1.0078125, "learning_rate": 9.86970855155031e-05, "loss": 0.1853, "step": 5712 }, { "epoch": 4.185347985347986, "grad_norm": 1.53125, "learning_rate": 9.869630444375533e-05, "loss": 0.1956, "step": 5713 }, { "epoch": 4.186080586080586, "grad_norm": 1.1328125, "learning_rate": 9.869552314108324e-05, "loss": 0.1753, "step": 5714 }, { "epoch": 4.186813186813187, "grad_norm": 0.89453125, "learning_rate": 9.869474160749059e-05, "loss": 0.1658, "step": 5715 }, { "epoch": 4.187545787545788, "grad_norm": 1.1953125, "learning_rate": 9.869395984298108e-05, "loss": 0.1377, "step": 5716 }, { "epoch": 4.1882783882783885, "grad_norm": 1.25, "learning_rate": 9.869317784755851e-05, "loss": 0.1603, "step": 5717 }, { "epoch": 4.189010989010989, "grad_norm": 1.0703125, "learning_rate": 9.869239562122661e-05, "loss": 0.1479, "step": 5718 }, { "epoch": 4.18974358974359, "grad_norm": 1.2578125, "learning_rate": 9.869161316398913e-05, "loss": 0.1703, "step": 5719 }, { "epoch": 4.190476190476191, "grad_norm": 1.46875, "learning_rate": 9.86908304758498e-05, "loss": 0.1879, "step": 5720 }, { "epoch": 4.191208791208791, "grad_norm": 1.5625, "learning_rate": 9.869004755681236e-05, "loss": 0.139, "step": 5721 }, { "epoch": 4.191941391941392, "grad_norm": 0.7421875, "learning_rate": 9.86892644068806e-05, "loss": 0.1696, "step": 5722 }, { "epoch": 4.192673992673993, "grad_norm": 0.84765625, "learning_rate": 9.868848102605826e-05, "loss": 0.129, "step": 5723 }, { "epoch": 4.193406593406594, "grad_norm": 1.1640625, "learning_rate": 9.868769741434906e-05, "loss": 0.177, "step": 5724 }, { "epoch": 4.194139194139194, "grad_norm": 1.484375, "learning_rate": 9.868691357175683e-05, "loss": 0.1842, "step": 5725 }, { "epoch": 4.194871794871795, "grad_norm": 1.3046875, "learning_rate": 9.868612949828523e-05, "loss": 0.1845, "step": 5726 }, { "epoch": 4.195604395604396, "grad_norm": 1.1171875, "learning_rate": 9.868534519393808e-05, "loss": 0.1462, "step": 5727 }, { "epoch": 4.1963369963369965, "grad_norm": 1.6171875, "learning_rate": 9.868456065871915e-05, "loss": 0.1317, "step": 5728 }, { "epoch": 4.197069597069597, "grad_norm": 1.359375, "learning_rate": 9.868377589263214e-05, "loss": 0.1886, "step": 5729 }, { "epoch": 4.197802197802198, "grad_norm": 0.96875, "learning_rate": 9.868299089568086e-05, "loss": 0.1447, "step": 5730 }, { "epoch": 4.198534798534799, "grad_norm": 0.875, "learning_rate": 9.868220566786905e-05, "loss": 0.1251, "step": 5731 }, { "epoch": 4.1992673992673994, "grad_norm": 0.80078125, "learning_rate": 9.868142020920046e-05, "loss": 0.1482, "step": 5732 }, { "epoch": 4.2, "grad_norm": 1.0078125, "learning_rate": 9.868063451967889e-05, "loss": 0.1828, "step": 5733 }, { "epoch": 4.200732600732601, "grad_norm": 1.0234375, "learning_rate": 9.867984859930807e-05, "loss": 0.1577, "step": 5734 }, { "epoch": 4.201465201465202, "grad_norm": 0.96484375, "learning_rate": 9.867906244809179e-05, "loss": 0.1277, "step": 5735 }, { "epoch": 4.202197802197802, "grad_norm": 0.7734375, "learning_rate": 9.867827606603381e-05, "loss": 0.1543, "step": 5736 }, { "epoch": 4.202930402930403, "grad_norm": 0.7890625, "learning_rate": 9.867748945313787e-05, "loss": 0.1861, "step": 5737 }, { "epoch": 4.203663003663004, "grad_norm": 0.59765625, "learning_rate": 9.867670260940778e-05, "loss": 0.17, "step": 5738 }, { "epoch": 4.204395604395605, "grad_norm": 0.71484375, "learning_rate": 9.86759155348473e-05, "loss": 0.1575, "step": 5739 }, { "epoch": 4.205128205128205, "grad_norm": 0.75390625, "learning_rate": 9.867512822946017e-05, "loss": 0.1374, "step": 5740 }, { "epoch": 4.205860805860806, "grad_norm": 1.140625, "learning_rate": 9.86743406932502e-05, "loss": 0.1436, "step": 5741 }, { "epoch": 4.206593406593407, "grad_norm": 0.61328125, "learning_rate": 9.867355292622113e-05, "loss": 0.1567, "step": 5742 }, { "epoch": 4.2073260073260075, "grad_norm": 0.77734375, "learning_rate": 9.867276492837678e-05, "loss": 0.1185, "step": 5743 }, { "epoch": 4.208058608058608, "grad_norm": 0.75390625, "learning_rate": 9.867197669972089e-05, "loss": 0.1301, "step": 5744 }, { "epoch": 4.208791208791209, "grad_norm": 1.2734375, "learning_rate": 9.867118824025726e-05, "loss": 0.1767, "step": 5745 }, { "epoch": 4.20952380952381, "grad_norm": 0.890625, "learning_rate": 9.867039954998964e-05, "loss": 0.158, "step": 5746 }, { "epoch": 4.21025641025641, "grad_norm": 1.109375, "learning_rate": 9.86696106289218e-05, "loss": 0.1694, "step": 5747 }, { "epoch": 4.210989010989011, "grad_norm": 0.94140625, "learning_rate": 9.866882147705759e-05, "loss": 0.173, "step": 5748 }, { "epoch": 4.211721611721612, "grad_norm": 1.0546875, "learning_rate": 9.86680320944007e-05, "loss": 0.1304, "step": 5749 }, { "epoch": 4.212454212454213, "grad_norm": 1.125, "learning_rate": 9.866724248095497e-05, "loss": 0.1142, "step": 5750 }, { "epoch": 4.213186813186813, "grad_norm": 1.0078125, "learning_rate": 9.866645263672417e-05, "loss": 0.1337, "step": 5751 }, { "epoch": 4.213919413919414, "grad_norm": 0.7265625, "learning_rate": 9.86656625617121e-05, "loss": 0.1722, "step": 5752 }, { "epoch": 4.214652014652015, "grad_norm": 1.046875, "learning_rate": 9.866487225592251e-05, "loss": 0.1736, "step": 5753 }, { "epoch": 4.2153846153846155, "grad_norm": 1.1171875, "learning_rate": 9.86640817193592e-05, "loss": 0.1513, "step": 5754 }, { "epoch": 4.216117216117216, "grad_norm": 0.87890625, "learning_rate": 9.866329095202598e-05, "loss": 0.1321, "step": 5755 }, { "epoch": 4.216849816849817, "grad_norm": 1.0078125, "learning_rate": 9.866249995392662e-05, "loss": 0.1584, "step": 5756 }, { "epoch": 4.217582417582418, "grad_norm": 0.88671875, "learning_rate": 9.86617087250649e-05, "loss": 0.1821, "step": 5757 }, { "epoch": 4.218315018315018, "grad_norm": 1.03125, "learning_rate": 9.866091726544462e-05, "loss": 0.1689, "step": 5758 }, { "epoch": 4.219047619047619, "grad_norm": 0.86328125, "learning_rate": 9.86601255750696e-05, "loss": 0.1566, "step": 5759 }, { "epoch": 4.21978021978022, "grad_norm": 0.72265625, "learning_rate": 9.865933365394358e-05, "loss": 0.1303, "step": 5760 }, { "epoch": 4.220512820512821, "grad_norm": 1.4609375, "learning_rate": 9.865854150207041e-05, "loss": 0.1908, "step": 5761 }, { "epoch": 4.221245421245421, "grad_norm": 1.453125, "learning_rate": 9.865774911945383e-05, "loss": 0.203, "step": 5762 }, { "epoch": 4.221978021978022, "grad_norm": 1.0234375, "learning_rate": 9.865695650609767e-05, "loss": 0.1861, "step": 5763 }, { "epoch": 4.222710622710623, "grad_norm": 0.84765625, "learning_rate": 9.865616366200575e-05, "loss": 0.1484, "step": 5764 }, { "epoch": 4.2234432234432235, "grad_norm": 0.86328125, "learning_rate": 9.865537058718183e-05, "loss": 0.1556, "step": 5765 }, { "epoch": 4.224175824175824, "grad_norm": 0.73046875, "learning_rate": 9.86545772816297e-05, "loss": 0.1345, "step": 5766 }, { "epoch": 4.224908424908425, "grad_norm": 1.078125, "learning_rate": 9.86537837453532e-05, "loss": 0.1247, "step": 5767 }, { "epoch": 4.225641025641026, "grad_norm": 1.171875, "learning_rate": 9.865298997835612e-05, "loss": 0.1569, "step": 5768 }, { "epoch": 4.2263736263736265, "grad_norm": 0.9453125, "learning_rate": 9.865219598064224e-05, "loss": 0.1456, "step": 5769 }, { "epoch": 4.227106227106227, "grad_norm": 1.015625, "learning_rate": 9.865140175221539e-05, "loss": 0.1681, "step": 5770 }, { "epoch": 4.227838827838828, "grad_norm": 0.76171875, "learning_rate": 9.865060729307936e-05, "loss": 0.1429, "step": 5771 }, { "epoch": 4.228571428571429, "grad_norm": 1.2265625, "learning_rate": 9.864981260323798e-05, "loss": 0.1802, "step": 5772 }, { "epoch": 4.229304029304029, "grad_norm": 0.84375, "learning_rate": 9.864901768269503e-05, "loss": 0.1649, "step": 5773 }, { "epoch": 4.23003663003663, "grad_norm": 1.1640625, "learning_rate": 9.864822253145433e-05, "loss": 0.1678, "step": 5774 }, { "epoch": 4.230769230769231, "grad_norm": 0.921875, "learning_rate": 9.86474271495197e-05, "loss": 0.1392, "step": 5775 }, { "epoch": 4.231501831501832, "grad_norm": 1.5546875, "learning_rate": 9.864663153689493e-05, "loss": 0.1993, "step": 5776 }, { "epoch": 4.232234432234432, "grad_norm": 0.73046875, "learning_rate": 9.864583569358385e-05, "loss": 0.1427, "step": 5777 }, { "epoch": 4.232967032967033, "grad_norm": 0.65625, "learning_rate": 9.864503961959024e-05, "loss": 0.1166, "step": 5778 }, { "epoch": 4.233699633699634, "grad_norm": 1.0625, "learning_rate": 9.864424331491798e-05, "loss": 0.1985, "step": 5779 }, { "epoch": 4.2344322344322345, "grad_norm": 0.80078125, "learning_rate": 9.864344677957082e-05, "loss": 0.145, "step": 5780 }, { "epoch": 4.235164835164835, "grad_norm": 1.046875, "learning_rate": 9.864265001355261e-05, "loss": 0.1459, "step": 5781 }, { "epoch": 4.235897435897436, "grad_norm": 0.9140625, "learning_rate": 9.864185301686715e-05, "loss": 0.1312, "step": 5782 }, { "epoch": 4.236630036630037, "grad_norm": 0.8046875, "learning_rate": 9.864105578951829e-05, "loss": 0.1365, "step": 5783 }, { "epoch": 4.237362637362637, "grad_norm": 0.80859375, "learning_rate": 9.864025833150982e-05, "loss": 0.1228, "step": 5784 }, { "epoch": 4.238095238095238, "grad_norm": 0.73046875, "learning_rate": 9.863946064284557e-05, "loss": 0.1535, "step": 5785 }, { "epoch": 4.238827838827839, "grad_norm": 1.1875, "learning_rate": 9.863866272352933e-05, "loss": 0.2286, "step": 5786 }, { "epoch": 4.23956043956044, "grad_norm": 0.80859375, "learning_rate": 9.863786457356497e-05, "loss": 0.1487, "step": 5787 }, { "epoch": 4.24029304029304, "grad_norm": 0.94921875, "learning_rate": 9.863706619295633e-05, "loss": 0.1368, "step": 5788 }, { "epoch": 4.241025641025641, "grad_norm": 0.64453125, "learning_rate": 9.863626758170718e-05, "loss": 0.1435, "step": 5789 }, { "epoch": 4.241758241758242, "grad_norm": 0.9765625, "learning_rate": 9.863546873982136e-05, "loss": 0.1451, "step": 5790 }, { "epoch": 4.2424908424908425, "grad_norm": 0.7421875, "learning_rate": 9.863466966730272e-05, "loss": 0.1519, "step": 5791 }, { "epoch": 4.243223443223443, "grad_norm": 1.3515625, "learning_rate": 9.863387036415507e-05, "loss": 0.171, "step": 5792 }, { "epoch": 4.243956043956044, "grad_norm": 0.6328125, "learning_rate": 9.863307083038225e-05, "loss": 0.1442, "step": 5793 }, { "epoch": 4.244688644688645, "grad_norm": 1.03125, "learning_rate": 9.86322710659881e-05, "loss": 0.1311, "step": 5794 }, { "epoch": 4.245421245421245, "grad_norm": 1.1875, "learning_rate": 9.863147107097643e-05, "loss": 0.1223, "step": 5795 }, { "epoch": 4.246153846153846, "grad_norm": 1.1484375, "learning_rate": 9.863067084535105e-05, "loss": 0.1882, "step": 5796 }, { "epoch": 4.246886446886447, "grad_norm": 1.203125, "learning_rate": 9.862987038911586e-05, "loss": 0.1338, "step": 5797 }, { "epoch": 4.247619047619048, "grad_norm": 1.1953125, "learning_rate": 9.862906970227465e-05, "loss": 0.1721, "step": 5798 }, { "epoch": 4.248351648351648, "grad_norm": 1.453125, "learning_rate": 9.862826878483128e-05, "loss": 0.2011, "step": 5799 }, { "epoch": 4.249084249084249, "grad_norm": 0.99609375, "learning_rate": 9.862746763678956e-05, "loss": 0.1587, "step": 5800 }, { "epoch": 4.24981684981685, "grad_norm": 0.94921875, "learning_rate": 9.862666625815336e-05, "loss": 0.1488, "step": 5801 }, { "epoch": 4.2505494505494505, "grad_norm": 1.234375, "learning_rate": 9.862586464892648e-05, "loss": 0.1824, "step": 5802 }, { "epoch": 4.251282051282051, "grad_norm": 0.9375, "learning_rate": 9.86250628091128e-05, "loss": 0.1717, "step": 5803 }, { "epoch": 4.252014652014652, "grad_norm": 0.828125, "learning_rate": 9.862426073871614e-05, "loss": 0.1437, "step": 5804 }, { "epoch": 4.252747252747253, "grad_norm": 0.63671875, "learning_rate": 9.862345843774035e-05, "loss": 0.142, "step": 5805 }, { "epoch": 4.2534798534798535, "grad_norm": 1.2265625, "learning_rate": 9.862265590618927e-05, "loss": 0.1461, "step": 5806 }, { "epoch": 4.254212454212454, "grad_norm": 1.1484375, "learning_rate": 9.862185314406676e-05, "loss": 0.1587, "step": 5807 }, { "epoch": 4.254945054945055, "grad_norm": 0.625, "learning_rate": 9.862105015137663e-05, "loss": 0.1703, "step": 5808 }, { "epoch": 4.255677655677656, "grad_norm": 1.4453125, "learning_rate": 9.862024692812277e-05, "loss": 0.1791, "step": 5809 }, { "epoch": 4.256410256410256, "grad_norm": 0.71875, "learning_rate": 9.8619443474309e-05, "loss": 0.1093, "step": 5810 }, { "epoch": 4.257142857142857, "grad_norm": 0.953125, "learning_rate": 9.861863978993919e-05, "loss": 0.1893, "step": 5811 }, { "epoch": 4.257875457875458, "grad_norm": 1.21875, "learning_rate": 9.861783587501717e-05, "loss": 0.1353, "step": 5812 }, { "epoch": 4.258608058608059, "grad_norm": 1.2890625, "learning_rate": 9.861703172954681e-05, "loss": 0.1051, "step": 5813 }, { "epoch": 4.259340659340659, "grad_norm": 1.1328125, "learning_rate": 9.861622735353196e-05, "loss": 0.1508, "step": 5814 }, { "epoch": 4.26007326007326, "grad_norm": 0.515625, "learning_rate": 9.861542274697647e-05, "loss": 0.1208, "step": 5815 }, { "epoch": 4.260805860805861, "grad_norm": 0.93359375, "learning_rate": 9.861461790988419e-05, "loss": 0.1503, "step": 5816 }, { "epoch": 4.2615384615384615, "grad_norm": 1.4609375, "learning_rate": 9.861381284225895e-05, "loss": 0.1533, "step": 5817 }, { "epoch": 4.262271062271062, "grad_norm": 1.25, "learning_rate": 9.861300754410469e-05, "loss": 0.1587, "step": 5818 }, { "epoch": 4.263003663003663, "grad_norm": 1.015625, "learning_rate": 9.861220201542518e-05, "loss": 0.1402, "step": 5819 }, { "epoch": 4.263736263736264, "grad_norm": 1.4296875, "learning_rate": 9.861139625622433e-05, "loss": 0.1709, "step": 5820 }, { "epoch": 4.264468864468864, "grad_norm": 1.0234375, "learning_rate": 9.861059026650598e-05, "loss": 0.1638, "step": 5821 }, { "epoch": 4.265201465201465, "grad_norm": 1.71875, "learning_rate": 9.860978404627401e-05, "loss": 0.1792, "step": 5822 }, { "epoch": 4.265934065934066, "grad_norm": 2.515625, "learning_rate": 9.860897759553226e-05, "loss": 0.1655, "step": 5823 }, { "epoch": 4.266666666666667, "grad_norm": 1.703125, "learning_rate": 9.860817091428461e-05, "loss": 0.1422, "step": 5824 }, { "epoch": 4.267399267399267, "grad_norm": 1.296875, "learning_rate": 9.860736400253491e-05, "loss": 0.1765, "step": 5825 }, { "epoch": 4.268131868131868, "grad_norm": 0.91015625, "learning_rate": 9.860655686028706e-05, "loss": 0.1428, "step": 5826 }, { "epoch": 4.268864468864469, "grad_norm": 1.359375, "learning_rate": 9.86057494875449e-05, "loss": 0.1848, "step": 5827 }, { "epoch": 4.2695970695970695, "grad_norm": 1.9765625, "learning_rate": 9.860494188431229e-05, "loss": 0.1584, "step": 5828 }, { "epoch": 4.27032967032967, "grad_norm": 1.6875, "learning_rate": 9.86041340505931e-05, "loss": 0.1599, "step": 5829 }, { "epoch": 4.271062271062271, "grad_norm": 1.40625, "learning_rate": 9.860332598639123e-05, "loss": 0.1662, "step": 5830 }, { "epoch": 4.271794871794872, "grad_norm": 0.87890625, "learning_rate": 9.860251769171053e-05, "loss": 0.1379, "step": 5831 }, { "epoch": 4.2725274725274724, "grad_norm": 0.89453125, "learning_rate": 9.860170916655488e-05, "loss": 0.1535, "step": 5832 }, { "epoch": 4.273260073260073, "grad_norm": 1.71875, "learning_rate": 9.860090041092815e-05, "loss": 0.1342, "step": 5833 }, { "epoch": 4.273992673992674, "grad_norm": 1.625, "learning_rate": 9.860009142483421e-05, "loss": 0.1263, "step": 5834 }, { "epoch": 4.274725274725275, "grad_norm": 1.0703125, "learning_rate": 9.859928220827694e-05, "loss": 0.1751, "step": 5835 }, { "epoch": 4.275457875457875, "grad_norm": 0.98828125, "learning_rate": 9.859847276126023e-05, "loss": 0.1315, "step": 5836 }, { "epoch": 4.276190476190476, "grad_norm": 1.234375, "learning_rate": 9.859766308378794e-05, "loss": 0.1838, "step": 5837 }, { "epoch": 4.276923076923077, "grad_norm": 1.1640625, "learning_rate": 9.859685317586396e-05, "loss": 0.1467, "step": 5838 }, { "epoch": 4.277655677655678, "grad_norm": 1.0703125, "learning_rate": 9.859604303749217e-05, "loss": 0.1408, "step": 5839 }, { "epoch": 4.278388278388278, "grad_norm": 2.5, "learning_rate": 9.859523266867643e-05, "loss": 0.1985, "step": 5840 }, { "epoch": 4.279120879120879, "grad_norm": 1.484375, "learning_rate": 9.859442206942066e-05, "loss": 0.1691, "step": 5841 }, { "epoch": 4.27985347985348, "grad_norm": 1.171875, "learning_rate": 9.859361123972872e-05, "loss": 0.1726, "step": 5842 }, { "epoch": 4.2805860805860805, "grad_norm": 1.9921875, "learning_rate": 9.85928001796045e-05, "loss": 0.136, "step": 5843 }, { "epoch": 4.281318681318681, "grad_norm": 1.421875, "learning_rate": 9.85919888890519e-05, "loss": 0.1631, "step": 5844 }, { "epoch": 4.282051282051282, "grad_norm": 1.203125, "learning_rate": 9.859117736807477e-05, "loss": 0.1715, "step": 5845 }, { "epoch": 4.282783882783883, "grad_norm": 1.4921875, "learning_rate": 9.859036561667703e-05, "loss": 0.1478, "step": 5846 }, { "epoch": 4.283516483516483, "grad_norm": 1.0703125, "learning_rate": 9.858955363486256e-05, "loss": 0.1966, "step": 5847 }, { "epoch": 4.284249084249084, "grad_norm": 1.7265625, "learning_rate": 9.858874142263528e-05, "loss": 0.1652, "step": 5848 }, { "epoch": 4.284981684981685, "grad_norm": 0.85546875, "learning_rate": 9.858792897999902e-05, "loss": 0.1505, "step": 5849 }, { "epoch": 4.285714285714286, "grad_norm": 1.546875, "learning_rate": 9.858711630695771e-05, "loss": 0.2064, "step": 5850 }, { "epoch": 4.286446886446886, "grad_norm": 0.61328125, "learning_rate": 9.858630340351525e-05, "loss": 0.15, "step": 5851 }, { "epoch": 4.287179487179487, "grad_norm": 0.99609375, "learning_rate": 9.858549026967552e-05, "loss": 0.1595, "step": 5852 }, { "epoch": 4.287912087912088, "grad_norm": 1.03125, "learning_rate": 9.858467690544241e-05, "loss": 0.1634, "step": 5853 }, { "epoch": 4.2886446886446885, "grad_norm": 1.015625, "learning_rate": 9.858386331081985e-05, "loss": 0.1576, "step": 5854 }, { "epoch": 4.289377289377289, "grad_norm": 1.2421875, "learning_rate": 9.85830494858117e-05, "loss": 0.1542, "step": 5855 }, { "epoch": 4.29010989010989, "grad_norm": 2.078125, "learning_rate": 9.858223543042187e-05, "loss": 0.161, "step": 5856 }, { "epoch": 4.290842490842491, "grad_norm": 0.80078125, "learning_rate": 9.858142114465429e-05, "loss": 0.1723, "step": 5857 }, { "epoch": 4.291575091575091, "grad_norm": 1.1953125, "learning_rate": 9.858060662851281e-05, "loss": 0.1439, "step": 5858 }, { "epoch": 4.292307692307692, "grad_norm": 0.73828125, "learning_rate": 9.85797918820014e-05, "loss": 0.1412, "step": 5859 }, { "epoch": 4.293040293040293, "grad_norm": 0.9609375, "learning_rate": 9.857897690512387e-05, "loss": 0.1914, "step": 5860 }, { "epoch": 4.293772893772894, "grad_norm": 1.2421875, "learning_rate": 9.857816169788421e-05, "loss": 0.1541, "step": 5861 }, { "epoch": 4.294505494505494, "grad_norm": 1.8984375, "learning_rate": 9.85773462602863e-05, "loss": 0.1581, "step": 5862 }, { "epoch": 4.295238095238095, "grad_norm": 0.94921875, "learning_rate": 9.857653059233403e-05, "loss": 0.1495, "step": 5863 }, { "epoch": 4.295970695970696, "grad_norm": 0.78515625, "learning_rate": 9.85757146940313e-05, "loss": 0.1656, "step": 5864 }, { "epoch": 4.2967032967032965, "grad_norm": 1.359375, "learning_rate": 9.857489856538208e-05, "loss": 0.177, "step": 5865 }, { "epoch": 4.297435897435897, "grad_norm": 1.9453125, "learning_rate": 9.857408220639021e-05, "loss": 0.1349, "step": 5866 }, { "epoch": 4.298168498168498, "grad_norm": 1.296875, "learning_rate": 9.857326561705964e-05, "loss": 0.1682, "step": 5867 }, { "epoch": 4.298901098901099, "grad_norm": 1.03125, "learning_rate": 9.857244879739427e-05, "loss": 0.141, "step": 5868 }, { "epoch": 4.2996336996336995, "grad_norm": 1.015625, "learning_rate": 9.857163174739802e-05, "loss": 0.1445, "step": 5869 }, { "epoch": 4.3003663003663, "grad_norm": 1.203125, "learning_rate": 9.85708144670748e-05, "loss": 0.1441, "step": 5870 }, { "epoch": 4.301098901098901, "grad_norm": 1.6796875, "learning_rate": 9.856999695642853e-05, "loss": 0.2035, "step": 5871 }, { "epoch": 4.301831501831502, "grad_norm": 1.109375, "learning_rate": 9.856917921546312e-05, "loss": 0.1407, "step": 5872 }, { "epoch": 4.302564102564102, "grad_norm": 1.0234375, "learning_rate": 9.856836124418247e-05, "loss": 0.1684, "step": 5873 }, { "epoch": 4.303296703296703, "grad_norm": 1.140625, "learning_rate": 9.856754304259057e-05, "loss": 0.2156, "step": 5874 }, { "epoch": 4.304029304029304, "grad_norm": 0.953125, "learning_rate": 9.856672461069126e-05, "loss": 0.1852, "step": 5875 }, { "epoch": 4.304761904761905, "grad_norm": 1.15625, "learning_rate": 9.856590594848848e-05, "loss": 0.1736, "step": 5876 }, { "epoch": 4.305494505494505, "grad_norm": 0.671875, "learning_rate": 9.856508705598619e-05, "loss": 0.1551, "step": 5877 }, { "epoch": 4.306227106227106, "grad_norm": 1.6484375, "learning_rate": 9.85642679331883e-05, "loss": 0.1481, "step": 5878 }, { "epoch": 4.306959706959707, "grad_norm": 1.5078125, "learning_rate": 9.856344858009871e-05, "loss": 0.2328, "step": 5879 }, { "epoch": 4.3076923076923075, "grad_norm": 0.97265625, "learning_rate": 9.856262899672135e-05, "loss": 0.1568, "step": 5880 }, { "epoch": 4.308424908424908, "grad_norm": 1.2578125, "learning_rate": 9.856180918306016e-05, "loss": 0.1335, "step": 5881 }, { "epoch": 4.309157509157509, "grad_norm": 1.203125, "learning_rate": 9.856098913911906e-05, "loss": 0.1596, "step": 5882 }, { "epoch": 4.30989010989011, "grad_norm": 1.3203125, "learning_rate": 9.856016886490199e-05, "loss": 0.153, "step": 5883 }, { "epoch": 4.31062271062271, "grad_norm": 1.21875, "learning_rate": 9.855934836041288e-05, "loss": 0.1917, "step": 5884 }, { "epoch": 4.311355311355311, "grad_norm": 0.99609375, "learning_rate": 9.855852762565566e-05, "loss": 0.1925, "step": 5885 }, { "epoch": 4.312087912087912, "grad_norm": 1.3671875, "learning_rate": 9.855770666063424e-05, "loss": 0.1423, "step": 5886 }, { "epoch": 4.312820512820513, "grad_norm": 1.671875, "learning_rate": 9.85568854653526e-05, "loss": 0.1712, "step": 5887 }, { "epoch": 4.313553113553113, "grad_norm": 0.765625, "learning_rate": 9.855606403981461e-05, "loss": 0.1433, "step": 5888 }, { "epoch": 4.314285714285714, "grad_norm": 0.8515625, "learning_rate": 9.855524238402428e-05, "loss": 0.175, "step": 5889 }, { "epoch": 4.315018315018315, "grad_norm": 1.4609375, "learning_rate": 9.855442049798545e-05, "loss": 0.157, "step": 5890 }, { "epoch": 4.3157509157509155, "grad_norm": 0.84765625, "learning_rate": 9.855359838170216e-05, "loss": 0.1232, "step": 5891 }, { "epoch": 4.316483516483516, "grad_norm": 0.734375, "learning_rate": 9.855277603517831e-05, "loss": 0.1326, "step": 5892 }, { "epoch": 4.317216117216117, "grad_norm": 0.6484375, "learning_rate": 9.855195345841781e-05, "loss": 0.1042, "step": 5893 }, { "epoch": 4.317948717948718, "grad_norm": 1.28125, "learning_rate": 9.855113065142466e-05, "loss": 0.1598, "step": 5894 }, { "epoch": 4.318681318681318, "grad_norm": 1.5625, "learning_rate": 9.855030761420273e-05, "loss": 0.1864, "step": 5895 }, { "epoch": 4.319413919413919, "grad_norm": 1.3203125, "learning_rate": 9.854948434675602e-05, "loss": 0.16, "step": 5896 }, { "epoch": 4.32014652014652, "grad_norm": 1.0078125, "learning_rate": 9.854866084908846e-05, "loss": 0.1693, "step": 5897 }, { "epoch": 4.320879120879121, "grad_norm": 0.859375, "learning_rate": 9.854783712120398e-05, "loss": 0.1382, "step": 5898 }, { "epoch": 4.321611721611721, "grad_norm": 1.8203125, "learning_rate": 9.854701316310655e-05, "loss": 0.1693, "step": 5899 }, { "epoch": 4.322344322344322, "grad_norm": 0.95703125, "learning_rate": 9.854618897480011e-05, "loss": 0.1297, "step": 5900 }, { "epoch": 4.323076923076923, "grad_norm": 1.6171875, "learning_rate": 9.85453645562886e-05, "loss": 0.1987, "step": 5901 }, { "epoch": 4.3238095238095235, "grad_norm": 1.0078125, "learning_rate": 9.854453990757599e-05, "loss": 0.1615, "step": 5902 }, { "epoch": 4.324542124542124, "grad_norm": 0.875, "learning_rate": 9.85437150286662e-05, "loss": 0.1213, "step": 5903 }, { "epoch": 4.325274725274725, "grad_norm": 1.078125, "learning_rate": 9.85428899195632e-05, "loss": 0.1459, "step": 5904 }, { "epoch": 4.326007326007326, "grad_norm": 1.9296875, "learning_rate": 9.854206458027095e-05, "loss": 0.1711, "step": 5905 }, { "epoch": 4.3267399267399265, "grad_norm": 1.0234375, "learning_rate": 9.854123901079339e-05, "loss": 0.1466, "step": 5906 }, { "epoch": 4.327472527472527, "grad_norm": 0.95703125, "learning_rate": 9.85404132111345e-05, "loss": 0.1471, "step": 5907 }, { "epoch": 4.328205128205128, "grad_norm": 0.83984375, "learning_rate": 9.85395871812982e-05, "loss": 0.1344, "step": 5908 }, { "epoch": 4.328937728937729, "grad_norm": 1.3984375, "learning_rate": 9.853876092128847e-05, "loss": 0.2445, "step": 5909 }, { "epoch": 4.329670329670329, "grad_norm": 0.92578125, "learning_rate": 9.853793443110927e-05, "loss": 0.1523, "step": 5910 }, { "epoch": 4.33040293040293, "grad_norm": 1.296875, "learning_rate": 9.853710771076455e-05, "loss": 0.1705, "step": 5911 }, { "epoch": 4.331135531135531, "grad_norm": 1.5234375, "learning_rate": 9.85362807602583e-05, "loss": 0.2101, "step": 5912 }, { "epoch": 4.331868131868132, "grad_norm": 0.984375, "learning_rate": 9.853545357959445e-05, "loss": 0.1479, "step": 5913 }, { "epoch": 4.332600732600732, "grad_norm": 1.1953125, "learning_rate": 9.853462616877696e-05, "loss": 0.1555, "step": 5914 }, { "epoch": 4.333333333333333, "grad_norm": 0.71484375, "learning_rate": 9.853379852780982e-05, "loss": 0.1387, "step": 5915 }, { "epoch": 4.334065934065934, "grad_norm": 1.234375, "learning_rate": 9.853297065669697e-05, "loss": 0.1534, "step": 5916 }, { "epoch": 4.3347985347985345, "grad_norm": 0.64453125, "learning_rate": 9.85321425554424e-05, "loss": 0.1604, "step": 5917 }, { "epoch": 4.335531135531135, "grad_norm": 1.3828125, "learning_rate": 9.853131422405006e-05, "loss": 0.185, "step": 5918 }, { "epoch": 4.336263736263736, "grad_norm": 1.1875, "learning_rate": 9.853048566252393e-05, "loss": 0.2263, "step": 5919 }, { "epoch": 4.336996336996337, "grad_norm": 1.5, "learning_rate": 9.852965687086797e-05, "loss": 0.1519, "step": 5920 }, { "epoch": 4.337728937728937, "grad_norm": 1.1796875, "learning_rate": 9.852882784908617e-05, "loss": 0.1486, "step": 5921 }, { "epoch": 4.338461538461538, "grad_norm": 0.97265625, "learning_rate": 9.852799859718248e-05, "loss": 0.1448, "step": 5922 }, { "epoch": 4.339194139194139, "grad_norm": 1.03125, "learning_rate": 9.852716911516088e-05, "loss": 0.1752, "step": 5923 }, { "epoch": 4.33992673992674, "grad_norm": 0.8515625, "learning_rate": 9.852633940302535e-05, "loss": 0.1144, "step": 5924 }, { "epoch": 4.34065934065934, "grad_norm": 0.9296875, "learning_rate": 9.852550946077987e-05, "loss": 0.1567, "step": 5925 }, { "epoch": 4.341391941391941, "grad_norm": 0.68359375, "learning_rate": 9.85246792884284e-05, "loss": 0.1435, "step": 5926 }, { "epoch": 4.342124542124542, "grad_norm": 1.375, "learning_rate": 9.852384888597492e-05, "loss": 0.1571, "step": 5927 }, { "epoch": 4.3428571428571425, "grad_norm": 1.0, "learning_rate": 9.852301825342342e-05, "loss": 0.1695, "step": 5928 }, { "epoch": 4.343589743589743, "grad_norm": 1.4609375, "learning_rate": 9.852218739077789e-05, "loss": 0.124, "step": 5929 }, { "epoch": 4.344322344322344, "grad_norm": 1.1640625, "learning_rate": 9.852135629804227e-05, "loss": 0.1529, "step": 5930 }, { "epoch": 4.345054945054945, "grad_norm": 0.84765625, "learning_rate": 9.852052497522059e-05, "loss": 0.1664, "step": 5931 }, { "epoch": 4.3457875457875454, "grad_norm": 1.4765625, "learning_rate": 9.85196934223168e-05, "loss": 0.1844, "step": 5932 }, { "epoch": 4.346520146520146, "grad_norm": 1.0703125, "learning_rate": 9.851886163933488e-05, "loss": 0.1501, "step": 5933 }, { "epoch": 4.347252747252747, "grad_norm": 0.96875, "learning_rate": 9.851802962627887e-05, "loss": 0.1759, "step": 5934 }, { "epoch": 4.347985347985348, "grad_norm": 0.734375, "learning_rate": 9.851719738315269e-05, "loss": 0.1278, "step": 5935 }, { "epoch": 4.348717948717948, "grad_norm": 0.890625, "learning_rate": 9.851636490996034e-05, "loss": 0.1765, "step": 5936 }, { "epoch": 4.349450549450549, "grad_norm": 0.95703125, "learning_rate": 9.851553220670584e-05, "loss": 0.1643, "step": 5937 }, { "epoch": 4.35018315018315, "grad_norm": 1.25, "learning_rate": 9.851469927339315e-05, "loss": 0.1736, "step": 5938 }, { "epoch": 4.350915750915751, "grad_norm": 1.21875, "learning_rate": 9.851386611002629e-05, "loss": 0.1727, "step": 5939 }, { "epoch": 4.351648351648351, "grad_norm": 1.3046875, "learning_rate": 9.851303271660924e-05, "loss": 0.141, "step": 5940 }, { "epoch": 4.352380952380952, "grad_norm": 1.0234375, "learning_rate": 9.851219909314598e-05, "loss": 0.1347, "step": 5941 }, { "epoch": 4.353113553113553, "grad_norm": 1.140625, "learning_rate": 9.851136523964051e-05, "loss": 0.1346, "step": 5942 }, { "epoch": 4.3538461538461535, "grad_norm": 0.85546875, "learning_rate": 9.851053115609682e-05, "loss": 0.1441, "step": 5943 }, { "epoch": 4.354578754578754, "grad_norm": 0.80859375, "learning_rate": 9.850969684251892e-05, "loss": 0.1452, "step": 5944 }, { "epoch": 4.355311355311355, "grad_norm": 1.0703125, "learning_rate": 9.850886229891082e-05, "loss": 0.1324, "step": 5945 }, { "epoch": 4.356043956043956, "grad_norm": 1.078125, "learning_rate": 9.850802752527648e-05, "loss": 0.156, "step": 5946 }, { "epoch": 4.356776556776556, "grad_norm": 0.91796875, "learning_rate": 9.850719252161994e-05, "loss": 0.1439, "step": 5947 }, { "epoch": 4.357509157509157, "grad_norm": 0.91015625, "learning_rate": 9.850635728794516e-05, "loss": 0.1595, "step": 5948 }, { "epoch": 4.358241758241758, "grad_norm": 0.83203125, "learning_rate": 9.850552182425618e-05, "loss": 0.1621, "step": 5949 }, { "epoch": 4.358974358974359, "grad_norm": 0.75390625, "learning_rate": 9.850468613055698e-05, "loss": 0.1559, "step": 5950 }, { "epoch": 4.359706959706959, "grad_norm": 0.94921875, "learning_rate": 9.850385020685155e-05, "loss": 0.1159, "step": 5951 }, { "epoch": 4.36043956043956, "grad_norm": 0.9296875, "learning_rate": 9.850301405314393e-05, "loss": 0.1432, "step": 5952 }, { "epoch": 4.361172161172161, "grad_norm": 1.3046875, "learning_rate": 9.850217766943812e-05, "loss": 0.1919, "step": 5953 }, { "epoch": 4.3619047619047615, "grad_norm": 1.328125, "learning_rate": 9.850134105573812e-05, "loss": 0.1494, "step": 5954 }, { "epoch": 4.362637362637362, "grad_norm": 0.828125, "learning_rate": 9.850050421204792e-05, "loss": 0.1471, "step": 5955 }, { "epoch": 4.363369963369963, "grad_norm": 0.703125, "learning_rate": 9.849966713837157e-05, "loss": 0.1435, "step": 5956 }, { "epoch": 4.364102564102564, "grad_norm": 0.84375, "learning_rate": 9.849882983471303e-05, "loss": 0.156, "step": 5957 }, { "epoch": 4.364835164835164, "grad_norm": 0.796875, "learning_rate": 9.849799230107635e-05, "loss": 0.1169, "step": 5958 }, { "epoch": 4.365567765567765, "grad_norm": 1.125, "learning_rate": 9.849715453746554e-05, "loss": 0.1538, "step": 5959 }, { "epoch": 4.366300366300366, "grad_norm": 0.921875, "learning_rate": 9.849631654388461e-05, "loss": 0.1687, "step": 5960 }, { "epoch": 4.367032967032967, "grad_norm": 0.875, "learning_rate": 9.849547832033756e-05, "loss": 0.1605, "step": 5961 }, { "epoch": 4.367765567765568, "grad_norm": 0.8125, "learning_rate": 9.849463986682841e-05, "loss": 0.1169, "step": 5962 }, { "epoch": 4.368498168498169, "grad_norm": 0.67578125, "learning_rate": 9.84938011833612e-05, "loss": 0.1949, "step": 5963 }, { "epoch": 4.36923076923077, "grad_norm": 0.84375, "learning_rate": 9.849296226993992e-05, "loss": 0.1551, "step": 5964 }, { "epoch": 4.36996336996337, "grad_norm": 1.125, "learning_rate": 9.849212312656861e-05, "loss": 0.133, "step": 5965 }, { "epoch": 4.370695970695971, "grad_norm": 0.734375, "learning_rate": 9.849128375325129e-05, "loss": 0.1451, "step": 5966 }, { "epoch": 4.371428571428572, "grad_norm": 1.640625, "learning_rate": 9.849044414999197e-05, "loss": 0.205, "step": 5967 }, { "epoch": 4.372161172161173, "grad_norm": 1.4609375, "learning_rate": 9.848960431679467e-05, "loss": 0.1351, "step": 5968 }, { "epoch": 4.372893772893773, "grad_norm": 0.64453125, "learning_rate": 9.848876425366343e-05, "loss": 0.1417, "step": 5969 }, { "epoch": 4.373626373626374, "grad_norm": 0.67578125, "learning_rate": 9.848792396060228e-05, "loss": 0.1352, "step": 5970 }, { "epoch": 4.374358974358975, "grad_norm": 0.8359375, "learning_rate": 9.84870834376152e-05, "loss": 0.128, "step": 5971 }, { "epoch": 4.3750915750915755, "grad_norm": 0.8828125, "learning_rate": 9.848624268470628e-05, "loss": 0.176, "step": 5972 }, { "epoch": 4.375824175824176, "grad_norm": 0.9921875, "learning_rate": 9.84854017018795e-05, "loss": 0.189, "step": 5973 }, { "epoch": 4.376556776556777, "grad_norm": 0.8046875, "learning_rate": 9.848456048913891e-05, "loss": 0.1587, "step": 5974 }, { "epoch": 4.377289377289378, "grad_norm": 0.87109375, "learning_rate": 9.848371904648854e-05, "loss": 0.1483, "step": 5975 }, { "epoch": 4.3780219780219785, "grad_norm": 0.9375, "learning_rate": 9.848287737393244e-05, "loss": 0.156, "step": 5976 }, { "epoch": 4.378754578754579, "grad_norm": 0.703125, "learning_rate": 9.84820354714746e-05, "loss": 0.1336, "step": 5977 }, { "epoch": 4.37948717948718, "grad_norm": 0.6875, "learning_rate": 9.848119333911907e-05, "loss": 0.1455, "step": 5978 }, { "epoch": 4.380219780219781, "grad_norm": 1.15625, "learning_rate": 9.84803509768699e-05, "loss": 0.1901, "step": 5979 }, { "epoch": 4.380952380952381, "grad_norm": 1.03125, "learning_rate": 9.847950838473114e-05, "loss": 0.1907, "step": 5980 }, { "epoch": 4.381684981684982, "grad_norm": 1.0234375, "learning_rate": 9.847866556270677e-05, "loss": 0.1397, "step": 5981 }, { "epoch": 4.382417582417583, "grad_norm": 1.2890625, "learning_rate": 9.847782251080088e-05, "loss": 0.1619, "step": 5982 }, { "epoch": 4.383150183150184, "grad_norm": 2.4375, "learning_rate": 9.84769792290175e-05, "loss": 0.1664, "step": 5983 }, { "epoch": 4.383882783882784, "grad_norm": 1.875, "learning_rate": 9.847613571736065e-05, "loss": 0.1478, "step": 5984 }, { "epoch": 4.384615384615385, "grad_norm": 0.7109375, "learning_rate": 9.847529197583439e-05, "loss": 0.1246, "step": 5985 }, { "epoch": 4.385347985347986, "grad_norm": 0.93359375, "learning_rate": 9.847444800444275e-05, "loss": 0.1233, "step": 5986 }, { "epoch": 4.3860805860805865, "grad_norm": 1.3828125, "learning_rate": 9.847360380318978e-05, "loss": 0.1522, "step": 5987 }, { "epoch": 4.386813186813187, "grad_norm": 0.95703125, "learning_rate": 9.847275937207954e-05, "loss": 0.1641, "step": 5988 }, { "epoch": 4.387545787545788, "grad_norm": 2.171875, "learning_rate": 9.847191471111604e-05, "loss": 0.1824, "step": 5989 }, { "epoch": 4.388278388278389, "grad_norm": 0.81640625, "learning_rate": 9.847106982030337e-05, "loss": 0.1644, "step": 5990 }, { "epoch": 4.389010989010989, "grad_norm": 1.171875, "learning_rate": 9.847022469964554e-05, "loss": 0.1701, "step": 5991 }, { "epoch": 4.38974358974359, "grad_norm": 0.58203125, "learning_rate": 9.846937934914664e-05, "loss": 0.1358, "step": 5992 }, { "epoch": 4.390476190476191, "grad_norm": 1.15625, "learning_rate": 9.846853376881068e-05, "loss": 0.1493, "step": 5993 }, { "epoch": 4.391208791208792, "grad_norm": 0.73828125, "learning_rate": 9.846768795864172e-05, "loss": 0.141, "step": 5994 }, { "epoch": 4.391941391941392, "grad_norm": 0.92578125, "learning_rate": 9.846684191864383e-05, "loss": 0.1521, "step": 5995 }, { "epoch": 4.392673992673993, "grad_norm": 0.9609375, "learning_rate": 9.846599564882105e-05, "loss": 0.1476, "step": 5996 }, { "epoch": 4.393406593406594, "grad_norm": 1.203125, "learning_rate": 9.846514914917744e-05, "loss": 0.1461, "step": 5997 }, { "epoch": 4.3941391941391945, "grad_norm": 1.015625, "learning_rate": 9.846430241971705e-05, "loss": 0.145, "step": 5998 }, { "epoch": 4.394871794871795, "grad_norm": 1.2890625, "learning_rate": 9.846345546044395e-05, "loss": 0.1607, "step": 5999 }, { "epoch": 4.395604395604396, "grad_norm": 1.3671875, "learning_rate": 9.846260827136219e-05, "loss": 0.1578, "step": 6000 }, { "epoch": 4.396336996336997, "grad_norm": 0.82421875, "learning_rate": 9.846176085247581e-05, "loss": 0.1494, "step": 6001 }, { "epoch": 4.397069597069597, "grad_norm": 1.5078125, "learning_rate": 9.846091320378892e-05, "loss": 0.1574, "step": 6002 }, { "epoch": 4.397802197802198, "grad_norm": 1.1796875, "learning_rate": 9.846006532530552e-05, "loss": 0.1347, "step": 6003 }, { "epoch": 4.398534798534799, "grad_norm": 1.2421875, "learning_rate": 9.845921721702972e-05, "loss": 0.181, "step": 6004 }, { "epoch": 4.3992673992674, "grad_norm": 0.85546875, "learning_rate": 9.845836887896557e-05, "loss": 0.1241, "step": 6005 }, { "epoch": 4.4, "grad_norm": 0.71875, "learning_rate": 9.845752031111712e-05, "loss": 0.159, "step": 6006 }, { "epoch": 4.400732600732601, "grad_norm": 1.125, "learning_rate": 9.845667151348846e-05, "loss": 0.1742, "step": 6007 }, { "epoch": 4.401465201465202, "grad_norm": 1.078125, "learning_rate": 9.845582248608363e-05, "loss": 0.1939, "step": 6008 }, { "epoch": 4.4021978021978025, "grad_norm": 0.5, "learning_rate": 9.845497322890671e-05, "loss": 0.1469, "step": 6009 }, { "epoch": 4.402930402930403, "grad_norm": 0.9140625, "learning_rate": 9.845412374196178e-05, "loss": 0.1638, "step": 6010 }, { "epoch": 4.403663003663004, "grad_norm": 1.140625, "learning_rate": 9.84532740252529e-05, "loss": 0.1496, "step": 6011 }, { "epoch": 4.404395604395605, "grad_norm": 1.09375, "learning_rate": 9.845242407878413e-05, "loss": 0.1534, "step": 6012 }, { "epoch": 4.4051282051282055, "grad_norm": 0.59375, "learning_rate": 9.845157390255957e-05, "loss": 0.1506, "step": 6013 }, { "epoch": 4.405860805860806, "grad_norm": 1.109375, "learning_rate": 9.845072349658328e-05, "loss": 0.1586, "step": 6014 }, { "epoch": 4.406593406593407, "grad_norm": 1.1484375, "learning_rate": 9.844987286085931e-05, "loss": 0.1858, "step": 6015 }, { "epoch": 4.407326007326008, "grad_norm": 1.171875, "learning_rate": 9.844902199539177e-05, "loss": 0.1775, "step": 6016 }, { "epoch": 4.408058608058608, "grad_norm": 0.62109375, "learning_rate": 9.844817090018473e-05, "loss": 0.1278, "step": 6017 }, { "epoch": 4.408791208791209, "grad_norm": 0.86328125, "learning_rate": 9.844731957524226e-05, "loss": 0.1701, "step": 6018 }, { "epoch": 4.40952380952381, "grad_norm": 0.8984375, "learning_rate": 9.844646802056844e-05, "loss": 0.1564, "step": 6019 }, { "epoch": 4.410256410256411, "grad_norm": 0.765625, "learning_rate": 9.844561623616736e-05, "loss": 0.1462, "step": 6020 }, { "epoch": 4.410989010989011, "grad_norm": 1.0390625, "learning_rate": 9.844476422204308e-05, "loss": 0.1596, "step": 6021 }, { "epoch": 4.411721611721612, "grad_norm": 0.92578125, "learning_rate": 9.84439119781997e-05, "loss": 0.1502, "step": 6022 }, { "epoch": 4.412454212454213, "grad_norm": 1.125, "learning_rate": 9.84430595046413e-05, "loss": 0.1561, "step": 6023 }, { "epoch": 4.4131868131868135, "grad_norm": 0.90625, "learning_rate": 9.844220680137197e-05, "loss": 0.1621, "step": 6024 }, { "epoch": 4.413919413919414, "grad_norm": 0.734375, "learning_rate": 9.844135386839578e-05, "loss": 0.1347, "step": 6025 }, { "epoch": 4.414652014652015, "grad_norm": 1.40625, "learning_rate": 9.844050070571684e-05, "loss": 0.1817, "step": 6026 }, { "epoch": 4.415384615384616, "grad_norm": 1.328125, "learning_rate": 9.843964731333921e-05, "loss": 0.1774, "step": 6027 }, { "epoch": 4.416117216117216, "grad_norm": 0.7578125, "learning_rate": 9.8438793691267e-05, "loss": 0.1548, "step": 6028 }, { "epoch": 4.416849816849817, "grad_norm": 0.734375, "learning_rate": 9.843793983950428e-05, "loss": 0.1472, "step": 6029 }, { "epoch": 4.417582417582418, "grad_norm": 0.9140625, "learning_rate": 9.843708575805516e-05, "loss": 0.166, "step": 6030 }, { "epoch": 4.418315018315019, "grad_norm": 0.89453125, "learning_rate": 9.843623144692372e-05, "loss": 0.1504, "step": 6031 }, { "epoch": 4.419047619047619, "grad_norm": 1.6171875, "learning_rate": 9.843537690611407e-05, "loss": 0.1523, "step": 6032 }, { "epoch": 4.41978021978022, "grad_norm": 1.2734375, "learning_rate": 9.843452213563026e-05, "loss": 0.1482, "step": 6033 }, { "epoch": 4.420512820512821, "grad_norm": 1.2734375, "learning_rate": 9.843366713547646e-05, "loss": 0.1713, "step": 6034 }, { "epoch": 4.4212454212454215, "grad_norm": 1.0234375, "learning_rate": 9.843281190565671e-05, "loss": 0.1958, "step": 6035 }, { "epoch": 4.421978021978022, "grad_norm": 1.734375, "learning_rate": 9.843195644617513e-05, "loss": 0.1743, "step": 6036 }, { "epoch": 4.422710622710623, "grad_norm": 1.0078125, "learning_rate": 9.84311007570358e-05, "loss": 0.1662, "step": 6037 }, { "epoch": 4.423443223443224, "grad_norm": 0.99609375, "learning_rate": 9.843024483824285e-05, "loss": 0.1806, "step": 6038 }, { "epoch": 4.4241758241758244, "grad_norm": 1.1953125, "learning_rate": 9.842938868980033e-05, "loss": 0.2014, "step": 6039 }, { "epoch": 4.424908424908425, "grad_norm": 0.8125, "learning_rate": 9.842853231171241e-05, "loss": 0.111, "step": 6040 }, { "epoch": 4.425641025641026, "grad_norm": 1.078125, "learning_rate": 9.842767570398313e-05, "loss": 0.1507, "step": 6041 }, { "epoch": 4.426373626373627, "grad_norm": 0.89453125, "learning_rate": 9.842681886661664e-05, "loss": 0.1366, "step": 6042 }, { "epoch": 4.427106227106227, "grad_norm": 1.40625, "learning_rate": 9.842596179961702e-05, "loss": 0.1691, "step": 6043 }, { "epoch": 4.427838827838828, "grad_norm": 0.9453125, "learning_rate": 9.842510450298838e-05, "loss": 0.1474, "step": 6044 }, { "epoch": 4.428571428571429, "grad_norm": 1.0625, "learning_rate": 9.842424697673483e-05, "loss": 0.1897, "step": 6045 }, { "epoch": 4.4293040293040296, "grad_norm": 0.69140625, "learning_rate": 9.84233892208605e-05, "loss": 0.1423, "step": 6046 }, { "epoch": 4.43003663003663, "grad_norm": 1.609375, "learning_rate": 9.842253123536946e-05, "loss": 0.1795, "step": 6047 }, { "epoch": 4.430769230769231, "grad_norm": 0.92578125, "learning_rate": 9.842167302026586e-05, "loss": 0.1481, "step": 6048 }, { "epoch": 4.431501831501832, "grad_norm": 0.82421875, "learning_rate": 9.842081457555376e-05, "loss": 0.1372, "step": 6049 }, { "epoch": 4.4322344322344325, "grad_norm": 0.95703125, "learning_rate": 9.841995590123734e-05, "loss": 0.1659, "step": 6050 }, { "epoch": 4.432967032967033, "grad_norm": 0.97265625, "learning_rate": 9.841909699732066e-05, "loss": 0.1516, "step": 6051 }, { "epoch": 4.433699633699634, "grad_norm": 0.96875, "learning_rate": 9.841823786380787e-05, "loss": 0.1366, "step": 6052 }, { "epoch": 4.434432234432235, "grad_norm": 0.99609375, "learning_rate": 9.841737850070306e-05, "loss": 0.1577, "step": 6053 }, { "epoch": 4.435164835164835, "grad_norm": 0.90234375, "learning_rate": 9.841651890801037e-05, "loss": 0.1768, "step": 6054 }, { "epoch": 4.435897435897436, "grad_norm": 0.828125, "learning_rate": 9.84156590857339e-05, "loss": 0.1793, "step": 6055 }, { "epoch": 4.436630036630037, "grad_norm": 1.859375, "learning_rate": 9.841479903387778e-05, "loss": 0.1514, "step": 6056 }, { "epoch": 4.437362637362638, "grad_norm": 1.09375, "learning_rate": 9.841393875244612e-05, "loss": 0.1582, "step": 6057 }, { "epoch": 4.438095238095238, "grad_norm": 0.9453125, "learning_rate": 9.841307824144306e-05, "loss": 0.1447, "step": 6058 }, { "epoch": 4.438827838827839, "grad_norm": 1.3828125, "learning_rate": 9.841221750087272e-05, "loss": 0.1348, "step": 6059 }, { "epoch": 4.43956043956044, "grad_norm": 1.09375, "learning_rate": 9.84113565307392e-05, "loss": 0.1504, "step": 6060 }, { "epoch": 4.4402930402930405, "grad_norm": 1.109375, "learning_rate": 9.841049533104666e-05, "loss": 0.165, "step": 6061 }, { "epoch": 4.441025641025641, "grad_norm": 1.484375, "learning_rate": 9.840963390179919e-05, "loss": 0.1376, "step": 6062 }, { "epoch": 4.441758241758242, "grad_norm": 1.1640625, "learning_rate": 9.840877224300095e-05, "loss": 0.1618, "step": 6063 }, { "epoch": 4.442490842490843, "grad_norm": 1.0625, "learning_rate": 9.840791035465605e-05, "loss": 0.1373, "step": 6064 }, { "epoch": 4.443223443223443, "grad_norm": 0.87890625, "learning_rate": 9.840704823676863e-05, "loss": 0.1358, "step": 6065 }, { "epoch": 4.443956043956044, "grad_norm": 0.7265625, "learning_rate": 9.840618588934281e-05, "loss": 0.1571, "step": 6066 }, { "epoch": 4.444688644688645, "grad_norm": 0.8671875, "learning_rate": 9.840532331238273e-05, "loss": 0.1462, "step": 6067 }, { "epoch": 4.445421245421246, "grad_norm": 1.0859375, "learning_rate": 9.840446050589252e-05, "loss": 0.1657, "step": 6068 }, { "epoch": 4.446153846153846, "grad_norm": 0.796875, "learning_rate": 9.84035974698763e-05, "loss": 0.1288, "step": 6069 }, { "epoch": 4.446886446886447, "grad_norm": 0.8984375, "learning_rate": 9.840273420433822e-05, "loss": 0.1385, "step": 6070 }, { "epoch": 4.447619047619048, "grad_norm": 1.2265625, "learning_rate": 9.840187070928243e-05, "loss": 0.1661, "step": 6071 }, { "epoch": 4.4483516483516485, "grad_norm": 1.296875, "learning_rate": 9.840100698471304e-05, "loss": 0.1762, "step": 6072 }, { "epoch": 4.449084249084249, "grad_norm": 0.859375, "learning_rate": 9.84001430306342e-05, "loss": 0.1762, "step": 6073 }, { "epoch": 4.44981684981685, "grad_norm": 1.0703125, "learning_rate": 9.839927884705008e-05, "loss": 0.1415, "step": 6074 }, { "epoch": 4.450549450549451, "grad_norm": 1.2109375, "learning_rate": 9.839841443396476e-05, "loss": 0.1688, "step": 6075 }, { "epoch": 4.4512820512820515, "grad_norm": 0.8046875, "learning_rate": 9.839754979138242e-05, "loss": 0.1618, "step": 6076 }, { "epoch": 4.452014652014652, "grad_norm": 0.6484375, "learning_rate": 9.839668491930721e-05, "loss": 0.138, "step": 6077 }, { "epoch": 4.452747252747253, "grad_norm": 1.0625, "learning_rate": 9.839581981774323e-05, "loss": 0.1867, "step": 6078 }, { "epoch": 4.453479853479854, "grad_norm": 0.78125, "learning_rate": 9.839495448669467e-05, "loss": 0.1245, "step": 6079 }, { "epoch": 4.454212454212454, "grad_norm": 1.03125, "learning_rate": 9.839408892616566e-05, "loss": 0.1553, "step": 6080 }, { "epoch": 4.454945054945055, "grad_norm": 0.62890625, "learning_rate": 9.839322313616035e-05, "loss": 0.1351, "step": 6081 }, { "epoch": 4.455677655677656, "grad_norm": 0.890625, "learning_rate": 9.839235711668288e-05, "loss": 0.1478, "step": 6082 }, { "epoch": 4.456410256410257, "grad_norm": 0.65234375, "learning_rate": 9.839149086773741e-05, "loss": 0.1529, "step": 6083 }, { "epoch": 4.457142857142857, "grad_norm": 0.8125, "learning_rate": 9.839062438932807e-05, "loss": 0.16, "step": 6084 }, { "epoch": 4.457875457875458, "grad_norm": 1.2109375, "learning_rate": 9.838975768145905e-05, "loss": 0.1538, "step": 6085 }, { "epoch": 4.458608058608059, "grad_norm": 1.2265625, "learning_rate": 9.838889074413447e-05, "loss": 0.1365, "step": 6086 }, { "epoch": 4.4593406593406595, "grad_norm": 1.2578125, "learning_rate": 9.838802357735848e-05, "loss": 0.1861, "step": 6087 }, { "epoch": 4.46007326007326, "grad_norm": 1.6875, "learning_rate": 9.838715618113526e-05, "loss": 0.2014, "step": 6088 }, { "epoch": 4.460805860805861, "grad_norm": 1.8359375, "learning_rate": 9.838628855546896e-05, "loss": 0.1714, "step": 6089 }, { "epoch": 4.461538461538462, "grad_norm": 0.84765625, "learning_rate": 9.838542070036373e-05, "loss": 0.1509, "step": 6090 }, { "epoch": 4.462271062271062, "grad_norm": 1.4140625, "learning_rate": 9.838455261582372e-05, "loss": 0.177, "step": 6091 }, { "epoch": 4.463003663003663, "grad_norm": 0.8828125, "learning_rate": 9.838368430185312e-05, "loss": 0.1306, "step": 6092 }, { "epoch": 4.463736263736264, "grad_norm": 0.796875, "learning_rate": 9.838281575845606e-05, "loss": 0.1363, "step": 6093 }, { "epoch": 4.464468864468865, "grad_norm": 0.8203125, "learning_rate": 9.838194698563671e-05, "loss": 0.1607, "step": 6094 }, { "epoch": 4.465201465201465, "grad_norm": 1.078125, "learning_rate": 9.838107798339923e-05, "loss": 0.137, "step": 6095 }, { "epoch": 4.465934065934066, "grad_norm": 0.8984375, "learning_rate": 9.838020875174778e-05, "loss": 0.1606, "step": 6096 }, { "epoch": 4.466666666666667, "grad_norm": 1.046875, "learning_rate": 9.837933929068653e-05, "loss": 0.1439, "step": 6097 }, { "epoch": 4.4673992673992675, "grad_norm": 1.7890625, "learning_rate": 9.837846960021966e-05, "loss": 0.1397, "step": 6098 }, { "epoch": 4.468131868131868, "grad_norm": 1.296875, "learning_rate": 9.837759968035133e-05, "loss": 0.1462, "step": 6099 }, { "epoch": 4.468864468864469, "grad_norm": 1.5, "learning_rate": 9.837672953108568e-05, "loss": 0.1678, "step": 6100 }, { "epoch": 4.46959706959707, "grad_norm": 1.2890625, "learning_rate": 9.837585915242692e-05, "loss": 0.1582, "step": 6101 }, { "epoch": 4.47032967032967, "grad_norm": 1.0625, "learning_rate": 9.837498854437921e-05, "loss": 0.1394, "step": 6102 }, { "epoch": 4.471062271062271, "grad_norm": 1.015625, "learning_rate": 9.837411770694669e-05, "loss": 0.1357, "step": 6103 }, { "epoch": 4.471794871794872, "grad_norm": 0.68359375, "learning_rate": 9.837324664013357e-05, "loss": 0.1505, "step": 6104 }, { "epoch": 4.472527472527473, "grad_norm": 1.265625, "learning_rate": 9.837237534394401e-05, "loss": 0.1356, "step": 6105 }, { "epoch": 4.473260073260073, "grad_norm": 1.640625, "learning_rate": 9.837150381838217e-05, "loss": 0.1812, "step": 6106 }, { "epoch": 4.473992673992674, "grad_norm": 1.359375, "learning_rate": 9.837063206345224e-05, "loss": 0.1372, "step": 6107 }, { "epoch": 4.474725274725275, "grad_norm": 1.1328125, "learning_rate": 9.836976007915842e-05, "loss": 0.153, "step": 6108 }, { "epoch": 4.4754578754578755, "grad_norm": 0.83984375, "learning_rate": 9.836888786550486e-05, "loss": 0.1519, "step": 6109 }, { "epoch": 4.476190476190476, "grad_norm": 1.40625, "learning_rate": 9.836801542249574e-05, "loss": 0.1379, "step": 6110 }, { "epoch": 4.476923076923077, "grad_norm": 1.8828125, "learning_rate": 9.836714275013523e-05, "loss": 0.1863, "step": 6111 }, { "epoch": 4.477655677655678, "grad_norm": 2.09375, "learning_rate": 9.836626984842754e-05, "loss": 0.161, "step": 6112 }, { "epoch": 4.4783882783882785, "grad_norm": 1.1171875, "learning_rate": 9.836539671737683e-05, "loss": 0.157, "step": 6113 }, { "epoch": 4.479120879120879, "grad_norm": 1.34375, "learning_rate": 9.83645233569873e-05, "loss": 0.1294, "step": 6114 }, { "epoch": 4.47985347985348, "grad_norm": 0.97265625, "learning_rate": 9.836364976726313e-05, "loss": 0.1581, "step": 6115 }, { "epoch": 4.480586080586081, "grad_norm": 1.2890625, "learning_rate": 9.836277594820848e-05, "loss": 0.1528, "step": 6116 }, { "epoch": 4.481318681318681, "grad_norm": 0.75, "learning_rate": 9.836190189982758e-05, "loss": 0.1491, "step": 6117 }, { "epoch": 4.482051282051282, "grad_norm": 0.72265625, "learning_rate": 9.836102762212457e-05, "loss": 0.1398, "step": 6118 }, { "epoch": 4.482783882783883, "grad_norm": 0.99609375, "learning_rate": 9.83601531151037e-05, "loss": 0.1641, "step": 6119 }, { "epoch": 4.483516483516484, "grad_norm": 1.2265625, "learning_rate": 9.835927837876909e-05, "loss": 0.1492, "step": 6120 }, { "epoch": 4.484249084249084, "grad_norm": 1.1953125, "learning_rate": 9.835840341312499e-05, "loss": 0.1242, "step": 6121 }, { "epoch": 4.484981684981685, "grad_norm": 0.99609375, "learning_rate": 9.835752821817556e-05, "loss": 0.1715, "step": 6122 }, { "epoch": 4.485714285714286, "grad_norm": 0.69140625, "learning_rate": 9.835665279392501e-05, "loss": 0.1519, "step": 6123 }, { "epoch": 4.4864468864468865, "grad_norm": 1.109375, "learning_rate": 9.835577714037751e-05, "loss": 0.1753, "step": 6124 }, { "epoch": 4.487179487179487, "grad_norm": 0.796875, "learning_rate": 9.835490125753728e-05, "loss": 0.1532, "step": 6125 }, { "epoch": 4.487912087912088, "grad_norm": 0.921875, "learning_rate": 9.83540251454085e-05, "loss": 0.1783, "step": 6126 }, { "epoch": 4.488644688644689, "grad_norm": 0.8515625, "learning_rate": 9.835314880399538e-05, "loss": 0.151, "step": 6127 }, { "epoch": 4.489377289377289, "grad_norm": 1.4296875, "learning_rate": 9.835227223330214e-05, "loss": 0.1563, "step": 6128 }, { "epoch": 4.49010989010989, "grad_norm": 0.9765625, "learning_rate": 9.835139543333292e-05, "loss": 0.1556, "step": 6129 }, { "epoch": 4.490842490842491, "grad_norm": 0.63671875, "learning_rate": 9.835051840409198e-05, "loss": 0.1424, "step": 6130 }, { "epoch": 4.491575091575092, "grad_norm": 0.84375, "learning_rate": 9.83496411455835e-05, "loss": 0.1659, "step": 6131 }, { "epoch": 4.492307692307692, "grad_norm": 1.3046875, "learning_rate": 9.834876365781166e-05, "loss": 0.1497, "step": 6132 }, { "epoch": 4.493040293040293, "grad_norm": 0.94921875, "learning_rate": 9.834788594078068e-05, "loss": 0.1642, "step": 6133 }, { "epoch": 4.493772893772894, "grad_norm": 1.0234375, "learning_rate": 9.83470079944948e-05, "loss": 0.1658, "step": 6134 }, { "epoch": 4.4945054945054945, "grad_norm": 0.82421875, "learning_rate": 9.834612981895819e-05, "loss": 0.1195, "step": 6135 }, { "epoch": 4.495238095238095, "grad_norm": 1.2265625, "learning_rate": 9.834525141417505e-05, "loss": 0.1365, "step": 6136 }, { "epoch": 4.495970695970696, "grad_norm": 1.578125, "learning_rate": 9.834437278014962e-05, "loss": 0.1782, "step": 6137 }, { "epoch": 4.496703296703297, "grad_norm": 1.1640625, "learning_rate": 9.834349391688609e-05, "loss": 0.1359, "step": 6138 }, { "epoch": 4.4974358974358974, "grad_norm": 0.890625, "learning_rate": 9.834261482438865e-05, "loss": 0.1336, "step": 6139 }, { "epoch": 4.498168498168498, "grad_norm": 1.265625, "learning_rate": 9.834173550266157e-05, "loss": 0.1415, "step": 6140 }, { "epoch": 4.498901098901099, "grad_norm": 1.234375, "learning_rate": 9.8340855951709e-05, "loss": 0.163, "step": 6141 }, { "epoch": 4.4996336996337, "grad_norm": 0.8203125, "learning_rate": 9.833997617153522e-05, "loss": 0.1718, "step": 6142 }, { "epoch": 4.5003663003663, "grad_norm": 0.6796875, "learning_rate": 9.833909616214439e-05, "loss": 0.144, "step": 6143 }, { "epoch": 4.501098901098901, "grad_norm": 1.234375, "learning_rate": 9.833821592354076e-05, "loss": 0.158, "step": 6144 }, { "epoch": 4.501831501831502, "grad_norm": 1.03125, "learning_rate": 9.833733545572849e-05, "loss": 0.1669, "step": 6145 }, { "epoch": 4.5025641025641026, "grad_norm": 0.62890625, "learning_rate": 9.833645475871189e-05, "loss": 0.1599, "step": 6146 }, { "epoch": 4.503296703296703, "grad_norm": 1.015625, "learning_rate": 9.833557383249511e-05, "loss": 0.1572, "step": 6147 }, { "epoch": 4.504029304029304, "grad_norm": 0.91796875, "learning_rate": 9.833469267708238e-05, "loss": 0.1629, "step": 6148 }, { "epoch": 4.504761904761905, "grad_norm": 0.96875, "learning_rate": 9.833381129247796e-05, "loss": 0.1248, "step": 6149 }, { "epoch": 4.5054945054945055, "grad_norm": 1.09375, "learning_rate": 9.833292967868604e-05, "loss": 0.1554, "step": 6150 }, { "epoch": 4.506227106227106, "grad_norm": 1.171875, "learning_rate": 9.833204783571085e-05, "loss": 0.1569, "step": 6151 }, { "epoch": 4.506959706959707, "grad_norm": 0.8046875, "learning_rate": 9.833116576355663e-05, "loss": 0.1519, "step": 6152 }, { "epoch": 4.507692307692308, "grad_norm": 0.828125, "learning_rate": 9.833028346222757e-05, "loss": 0.1546, "step": 6153 }, { "epoch": 4.508424908424908, "grad_norm": 0.625, "learning_rate": 9.832940093172792e-05, "loss": 0.1473, "step": 6154 }, { "epoch": 4.509157509157509, "grad_norm": 0.82421875, "learning_rate": 9.832851817206192e-05, "loss": 0.1335, "step": 6155 }, { "epoch": 4.50989010989011, "grad_norm": 0.9453125, "learning_rate": 9.832763518323379e-05, "loss": 0.1806, "step": 6156 }, { "epoch": 4.510622710622711, "grad_norm": 1.3125, "learning_rate": 9.832675196524775e-05, "loss": 0.1644, "step": 6157 }, { "epoch": 4.511355311355311, "grad_norm": 1.296875, "learning_rate": 9.832586851810805e-05, "loss": 0.1824, "step": 6158 }, { "epoch": 4.512087912087912, "grad_norm": 1.0390625, "learning_rate": 9.83249848418189e-05, "loss": 0.1547, "step": 6159 }, { "epoch": 4.512820512820513, "grad_norm": 0.546875, "learning_rate": 9.832410093638455e-05, "loss": 0.1107, "step": 6160 }, { "epoch": 4.5135531135531135, "grad_norm": 1.0390625, "learning_rate": 9.832321680180925e-05, "loss": 0.1721, "step": 6161 }, { "epoch": 4.514285714285714, "grad_norm": 0.921875, "learning_rate": 9.83223324380972e-05, "loss": 0.1652, "step": 6162 }, { "epoch": 4.515018315018315, "grad_norm": 1.0078125, "learning_rate": 9.832144784525267e-05, "loss": 0.1459, "step": 6163 }, { "epoch": 4.515750915750916, "grad_norm": 1.046875, "learning_rate": 9.832056302327987e-05, "loss": 0.1741, "step": 6164 }, { "epoch": 4.516483516483516, "grad_norm": 0.9375, "learning_rate": 9.831967797218308e-05, "loss": 0.1321, "step": 6165 }, { "epoch": 4.517216117216117, "grad_norm": 1.2578125, "learning_rate": 9.831879269196649e-05, "loss": 0.1158, "step": 6166 }, { "epoch": 4.517948717948718, "grad_norm": 1.1953125, "learning_rate": 9.831790718263437e-05, "loss": 0.1867, "step": 6167 }, { "epoch": 4.518681318681319, "grad_norm": 1.0625, "learning_rate": 9.831702144419099e-05, "loss": 0.1323, "step": 6168 }, { "epoch": 4.519413919413919, "grad_norm": 0.8828125, "learning_rate": 9.831613547664051e-05, "loss": 0.1535, "step": 6169 }, { "epoch": 4.52014652014652, "grad_norm": 1.3125, "learning_rate": 9.831524927998726e-05, "loss": 0.1674, "step": 6170 }, { "epoch": 4.520879120879121, "grad_norm": 1.953125, "learning_rate": 9.831436285423545e-05, "loss": 0.1931, "step": 6171 }, { "epoch": 4.5216117216117215, "grad_norm": 2.109375, "learning_rate": 9.831347619938932e-05, "loss": 0.1447, "step": 6172 }, { "epoch": 4.522344322344322, "grad_norm": 1.0390625, "learning_rate": 9.831258931545316e-05, "loss": 0.1606, "step": 6173 }, { "epoch": 4.523076923076923, "grad_norm": 1.125, "learning_rate": 9.831170220243117e-05, "loss": 0.1446, "step": 6174 }, { "epoch": 4.523809523809524, "grad_norm": 1.0859375, "learning_rate": 9.83108148603276e-05, "loss": 0.1643, "step": 6175 }, { "epoch": 4.5245421245421245, "grad_norm": 1.3828125, "learning_rate": 9.830992728914675e-05, "loss": 0.1741, "step": 6176 }, { "epoch": 4.525274725274725, "grad_norm": 0.7890625, "learning_rate": 9.830903948889282e-05, "loss": 0.1228, "step": 6177 }, { "epoch": 4.526007326007326, "grad_norm": 0.94140625, "learning_rate": 9.830815145957011e-05, "loss": 0.1447, "step": 6178 }, { "epoch": 4.526739926739927, "grad_norm": 1.046875, "learning_rate": 9.830726320118285e-05, "loss": 0.1377, "step": 6179 }, { "epoch": 4.527472527472527, "grad_norm": 0.9453125, "learning_rate": 9.830637471373528e-05, "loss": 0.1631, "step": 6180 }, { "epoch": 4.528205128205128, "grad_norm": 1.09375, "learning_rate": 9.830548599723168e-05, "loss": 0.1472, "step": 6181 }, { "epoch": 4.528937728937729, "grad_norm": 0.88671875, "learning_rate": 9.83045970516763e-05, "loss": 0.153, "step": 6182 }, { "epoch": 4.52967032967033, "grad_norm": 1.0078125, "learning_rate": 9.83037078770734e-05, "loss": 0.1691, "step": 6183 }, { "epoch": 4.53040293040293, "grad_norm": 1.2265625, "learning_rate": 9.830281847342727e-05, "loss": 0.1502, "step": 6184 }, { "epoch": 4.531135531135531, "grad_norm": 1.0390625, "learning_rate": 9.83019288407421e-05, "loss": 0.1442, "step": 6185 }, { "epoch": 4.531868131868132, "grad_norm": 1.2109375, "learning_rate": 9.830103897902223e-05, "loss": 0.1812, "step": 6186 }, { "epoch": 4.5326007326007325, "grad_norm": 0.9609375, "learning_rate": 9.830014888827189e-05, "loss": 0.1332, "step": 6187 }, { "epoch": 4.533333333333333, "grad_norm": 1.078125, "learning_rate": 9.829925856849534e-05, "loss": 0.1862, "step": 6188 }, { "epoch": 4.534065934065934, "grad_norm": 1.390625, "learning_rate": 9.829836801969683e-05, "loss": 0.191, "step": 6189 }, { "epoch": 4.534798534798535, "grad_norm": 0.96484375, "learning_rate": 9.829747724188068e-05, "loss": 0.1927, "step": 6190 }, { "epoch": 4.535531135531135, "grad_norm": 1.3671875, "learning_rate": 9.82965862350511e-05, "loss": 0.1566, "step": 6191 }, { "epoch": 4.536263736263736, "grad_norm": 1.40625, "learning_rate": 9.82956949992124e-05, "loss": 0.1551, "step": 6192 }, { "epoch": 4.536996336996337, "grad_norm": 1.5546875, "learning_rate": 9.829480353436884e-05, "loss": 0.1855, "step": 6193 }, { "epoch": 4.537728937728938, "grad_norm": 0.69921875, "learning_rate": 9.829391184052467e-05, "loss": 0.1496, "step": 6194 }, { "epoch": 4.538461538461538, "grad_norm": 1.3359375, "learning_rate": 9.829301991768418e-05, "loss": 0.1826, "step": 6195 }, { "epoch": 4.539194139194139, "grad_norm": 1.8828125, "learning_rate": 9.829212776585165e-05, "loss": 0.1592, "step": 6196 }, { "epoch": 4.53992673992674, "grad_norm": 1.6171875, "learning_rate": 9.829123538503134e-05, "loss": 0.188, "step": 6197 }, { "epoch": 4.5406593406593405, "grad_norm": 0.8671875, "learning_rate": 9.829034277522755e-05, "loss": 0.1285, "step": 6198 }, { "epoch": 4.541391941391941, "grad_norm": 0.9296875, "learning_rate": 9.828944993644453e-05, "loss": 0.1285, "step": 6199 }, { "epoch": 4.542124542124542, "grad_norm": 0.73046875, "learning_rate": 9.828855686868657e-05, "loss": 0.128, "step": 6200 }, { "epoch": 4.542857142857143, "grad_norm": 0.8671875, "learning_rate": 9.828766357195792e-05, "loss": 0.1452, "step": 6201 }, { "epoch": 4.543589743589743, "grad_norm": 1.015625, "learning_rate": 9.828677004626291e-05, "loss": 0.1553, "step": 6202 }, { "epoch": 4.544322344322344, "grad_norm": 1.3828125, "learning_rate": 9.828587629160579e-05, "loss": 0.1407, "step": 6203 }, { "epoch": 4.545054945054945, "grad_norm": 1.1171875, "learning_rate": 9.828498230799086e-05, "loss": 0.1344, "step": 6204 }, { "epoch": 4.545787545787546, "grad_norm": 1.3984375, "learning_rate": 9.828408809542236e-05, "loss": 0.1552, "step": 6205 }, { "epoch": 4.546520146520146, "grad_norm": 1.0, "learning_rate": 9.828319365390465e-05, "loss": 0.1684, "step": 6206 }, { "epoch": 4.547252747252747, "grad_norm": 1.140625, "learning_rate": 9.828229898344196e-05, "loss": 0.1477, "step": 6207 }, { "epoch": 4.547985347985348, "grad_norm": 1.453125, "learning_rate": 9.828140408403859e-05, "loss": 0.1627, "step": 6208 }, { "epoch": 4.5487179487179485, "grad_norm": 1.453125, "learning_rate": 9.828050895569882e-05, "loss": 0.1475, "step": 6209 }, { "epoch": 4.549450549450549, "grad_norm": 1.0859375, "learning_rate": 9.827961359842695e-05, "loss": 0.1588, "step": 6210 }, { "epoch": 4.55018315018315, "grad_norm": 0.83203125, "learning_rate": 9.827871801222725e-05, "loss": 0.1391, "step": 6211 }, { "epoch": 4.550915750915751, "grad_norm": 1.3828125, "learning_rate": 9.827782219710405e-05, "loss": 0.1273, "step": 6212 }, { "epoch": 4.5516483516483515, "grad_norm": 1.0625, "learning_rate": 9.827692615306161e-05, "loss": 0.1691, "step": 6213 }, { "epoch": 4.552380952380952, "grad_norm": 1.4453125, "learning_rate": 9.827602988010424e-05, "loss": 0.1599, "step": 6214 }, { "epoch": 4.553113553113553, "grad_norm": 1.5546875, "learning_rate": 9.827513337823622e-05, "loss": 0.1265, "step": 6215 }, { "epoch": 4.553846153846154, "grad_norm": 1.96875, "learning_rate": 9.827423664746185e-05, "loss": 0.2004, "step": 6216 }, { "epoch": 4.554578754578754, "grad_norm": 1.4375, "learning_rate": 9.827333968778544e-05, "loss": 0.1546, "step": 6217 }, { "epoch": 4.555311355311355, "grad_norm": 1.390625, "learning_rate": 9.827244249921127e-05, "loss": 0.2051, "step": 6218 }, { "epoch": 4.556043956043956, "grad_norm": 2.046875, "learning_rate": 9.827154508174364e-05, "loss": 0.178, "step": 6219 }, { "epoch": 4.556776556776557, "grad_norm": 1.4453125, "learning_rate": 9.827064743538687e-05, "loss": 0.1138, "step": 6220 }, { "epoch": 4.557509157509157, "grad_norm": 1.75, "learning_rate": 9.826974956014523e-05, "loss": 0.1671, "step": 6221 }, { "epoch": 4.558241758241758, "grad_norm": 1.0859375, "learning_rate": 9.826885145602305e-05, "loss": 0.1258, "step": 6222 }, { "epoch": 4.558974358974359, "grad_norm": 1.1640625, "learning_rate": 9.82679531230246e-05, "loss": 0.1749, "step": 6223 }, { "epoch": 4.5597069597069595, "grad_norm": 1.359375, "learning_rate": 9.826705456115423e-05, "loss": 0.1702, "step": 6224 }, { "epoch": 4.56043956043956, "grad_norm": 1.3125, "learning_rate": 9.82661557704162e-05, "loss": 0.1639, "step": 6225 }, { "epoch": 4.561172161172161, "grad_norm": 1.703125, "learning_rate": 9.826525675081486e-05, "loss": 0.1475, "step": 6226 }, { "epoch": 4.561904761904762, "grad_norm": 1.0859375, "learning_rate": 9.826435750235448e-05, "loss": 0.1518, "step": 6227 }, { "epoch": 4.562637362637362, "grad_norm": 1.015625, "learning_rate": 9.82634580250394e-05, "loss": 0.1502, "step": 6228 }, { "epoch": 4.563369963369963, "grad_norm": 0.859375, "learning_rate": 9.826255831887388e-05, "loss": 0.1532, "step": 6229 }, { "epoch": 4.564102564102564, "grad_norm": 1.6953125, "learning_rate": 9.82616583838623e-05, "loss": 0.1193, "step": 6230 }, { "epoch": 4.564835164835165, "grad_norm": 1.5390625, "learning_rate": 9.826075822000891e-05, "loss": 0.1803, "step": 6231 }, { "epoch": 4.565567765567765, "grad_norm": 1.1171875, "learning_rate": 9.825985782731806e-05, "loss": 0.1549, "step": 6232 }, { "epoch": 4.566300366300366, "grad_norm": 1.78125, "learning_rate": 9.825895720579405e-05, "loss": 0.1513, "step": 6233 }, { "epoch": 4.567032967032967, "grad_norm": 0.98828125, "learning_rate": 9.82580563554412e-05, "loss": 0.1606, "step": 6234 }, { "epoch": 4.5677655677655675, "grad_norm": 1.125, "learning_rate": 9.825715527626382e-05, "loss": 0.1844, "step": 6235 }, { "epoch": 4.568498168498168, "grad_norm": 1.0, "learning_rate": 9.825625396826623e-05, "loss": 0.1115, "step": 6236 }, { "epoch": 4.569230769230769, "grad_norm": 1.0390625, "learning_rate": 9.825535243145274e-05, "loss": 0.1829, "step": 6237 }, { "epoch": 4.56996336996337, "grad_norm": 0.63671875, "learning_rate": 9.82544506658277e-05, "loss": 0.1589, "step": 6238 }, { "epoch": 4.5706959706959704, "grad_norm": 0.75390625, "learning_rate": 9.82535486713954e-05, "loss": 0.1273, "step": 6239 }, { "epoch": 4.571428571428571, "grad_norm": 1.609375, "learning_rate": 9.825264644816017e-05, "loss": 0.1791, "step": 6240 }, { "epoch": 4.572161172161172, "grad_norm": 1.1796875, "learning_rate": 9.825174399612635e-05, "loss": 0.1243, "step": 6241 }, { "epoch": 4.572893772893773, "grad_norm": 0.93359375, "learning_rate": 9.825084131529822e-05, "loss": 0.1632, "step": 6242 }, { "epoch": 4.573626373626373, "grad_norm": 0.7734375, "learning_rate": 9.824993840568018e-05, "loss": 0.1499, "step": 6243 }, { "epoch": 4.574358974358974, "grad_norm": 0.71875, "learning_rate": 9.824903526727647e-05, "loss": 0.1538, "step": 6244 }, { "epoch": 4.575091575091575, "grad_norm": 0.80859375, "learning_rate": 9.824813190009146e-05, "loss": 0.1506, "step": 6245 }, { "epoch": 4.5758241758241756, "grad_norm": 0.78515625, "learning_rate": 9.824722830412949e-05, "loss": 0.1274, "step": 6246 }, { "epoch": 4.576556776556776, "grad_norm": 0.86328125, "learning_rate": 9.824632447939489e-05, "loss": 0.1556, "step": 6247 }, { "epoch": 4.577289377289377, "grad_norm": 1.265625, "learning_rate": 9.824542042589193e-05, "loss": 0.1343, "step": 6248 }, { "epoch": 4.578021978021978, "grad_norm": 0.796875, "learning_rate": 9.824451614362502e-05, "loss": 0.1178, "step": 6249 }, { "epoch": 4.5787545787545785, "grad_norm": 1.0546875, "learning_rate": 9.824361163259844e-05, "loss": 0.1505, "step": 6250 }, { "epoch": 4.579487179487179, "grad_norm": 1.0703125, "learning_rate": 9.824270689281657e-05, "loss": 0.1778, "step": 6251 }, { "epoch": 4.58021978021978, "grad_norm": 0.96875, "learning_rate": 9.824180192428369e-05, "loss": 0.1651, "step": 6252 }, { "epoch": 4.580952380952381, "grad_norm": 0.6015625, "learning_rate": 9.824089672700417e-05, "loss": 0.1335, "step": 6253 }, { "epoch": 4.581684981684981, "grad_norm": 0.95703125, "learning_rate": 9.823999130098235e-05, "loss": 0.1694, "step": 6254 }, { "epoch": 4.582417582417582, "grad_norm": 1.34375, "learning_rate": 9.823908564622256e-05, "loss": 0.1746, "step": 6255 }, { "epoch": 4.583150183150183, "grad_norm": 1.15625, "learning_rate": 9.823817976272915e-05, "loss": 0.1613, "step": 6256 }, { "epoch": 4.583882783882784, "grad_norm": 1.0625, "learning_rate": 9.823727365050643e-05, "loss": 0.1577, "step": 6257 }, { "epoch": 4.584615384615384, "grad_norm": 0.84375, "learning_rate": 9.823636730955876e-05, "loss": 0.1879, "step": 6258 }, { "epoch": 4.585347985347985, "grad_norm": 0.83984375, "learning_rate": 9.823546073989048e-05, "loss": 0.1499, "step": 6259 }, { "epoch": 4.586080586080586, "grad_norm": 0.75, "learning_rate": 9.823455394150597e-05, "loss": 0.1734, "step": 6260 }, { "epoch": 4.5868131868131865, "grad_norm": 1.2734375, "learning_rate": 9.82336469144095e-05, "loss": 0.1813, "step": 6261 }, { "epoch": 4.587545787545787, "grad_norm": 0.91796875, "learning_rate": 9.823273965860549e-05, "loss": 0.174, "step": 6262 }, { "epoch": 4.588278388278388, "grad_norm": 0.94921875, "learning_rate": 9.823183217409824e-05, "loss": 0.1314, "step": 6263 }, { "epoch": 4.589010989010989, "grad_norm": 1.2734375, "learning_rate": 9.82309244608921e-05, "loss": 0.1748, "step": 6264 }, { "epoch": 4.589743589743589, "grad_norm": 1.1875, "learning_rate": 9.823001651899146e-05, "loss": 0.1593, "step": 6265 }, { "epoch": 4.59047619047619, "grad_norm": 1.171875, "learning_rate": 9.822910834840062e-05, "loss": 0.1655, "step": 6266 }, { "epoch": 4.591208791208791, "grad_norm": 1.1484375, "learning_rate": 9.822819994912397e-05, "loss": 0.139, "step": 6267 }, { "epoch": 4.591941391941392, "grad_norm": 1.0390625, "learning_rate": 9.822729132116583e-05, "loss": 0.1591, "step": 6268 }, { "epoch": 4.592673992673992, "grad_norm": 2.03125, "learning_rate": 9.822638246453058e-05, "loss": 0.1469, "step": 6269 }, { "epoch": 4.593406593406593, "grad_norm": 1.0859375, "learning_rate": 9.822547337922257e-05, "loss": 0.1373, "step": 6270 }, { "epoch": 4.594139194139194, "grad_norm": 0.8671875, "learning_rate": 9.822456406524613e-05, "loss": 0.1713, "step": 6271 }, { "epoch": 4.5948717948717945, "grad_norm": 1.3984375, "learning_rate": 9.822365452260564e-05, "loss": 0.1584, "step": 6272 }, { "epoch": 4.595604395604395, "grad_norm": 1.0234375, "learning_rate": 9.822274475130548e-05, "loss": 0.177, "step": 6273 }, { "epoch": 4.596336996336996, "grad_norm": 1.21875, "learning_rate": 9.822183475134995e-05, "loss": 0.1496, "step": 6274 }, { "epoch": 4.597069597069597, "grad_norm": 1.40625, "learning_rate": 9.822092452274346e-05, "loss": 0.1434, "step": 6275 }, { "epoch": 4.5978021978021975, "grad_norm": 1.0703125, "learning_rate": 9.822001406549037e-05, "loss": 0.1739, "step": 6276 }, { "epoch": 4.598534798534798, "grad_norm": 1.328125, "learning_rate": 9.821910337959501e-05, "loss": 0.1354, "step": 6277 }, { "epoch": 4.599267399267399, "grad_norm": 1.15625, "learning_rate": 9.821819246506175e-05, "loss": 0.16, "step": 6278 }, { "epoch": 4.6, "grad_norm": 0.984375, "learning_rate": 9.821728132189498e-05, "loss": 0.1578, "step": 6279 }, { "epoch": 4.6007326007326, "grad_norm": 1.109375, "learning_rate": 9.821636995009903e-05, "loss": 0.1423, "step": 6280 }, { "epoch": 4.601465201465201, "grad_norm": 1.34375, "learning_rate": 9.821545834967833e-05, "loss": 0.1935, "step": 6281 }, { "epoch": 4.602197802197802, "grad_norm": 1.0859375, "learning_rate": 9.821454652063717e-05, "loss": 0.1607, "step": 6282 }, { "epoch": 4.602930402930403, "grad_norm": 0.75390625, "learning_rate": 9.821363446297996e-05, "loss": 0.1139, "step": 6283 }, { "epoch": 4.603663003663003, "grad_norm": 1.1328125, "learning_rate": 9.821272217671107e-05, "loss": 0.1311, "step": 6284 }, { "epoch": 4.604395604395604, "grad_norm": 1.953125, "learning_rate": 9.821180966183486e-05, "loss": 0.1916, "step": 6285 }, { "epoch": 4.605128205128205, "grad_norm": 2.453125, "learning_rate": 9.821089691835572e-05, "loss": 0.1694, "step": 6286 }, { "epoch": 4.6058608058608055, "grad_norm": 2.3125, "learning_rate": 9.820998394627799e-05, "loss": 0.1936, "step": 6287 }, { "epoch": 4.606593406593406, "grad_norm": 1.375, "learning_rate": 9.820907074560608e-05, "loss": 0.1434, "step": 6288 }, { "epoch": 4.607326007326007, "grad_norm": 1.109375, "learning_rate": 9.820815731634434e-05, "loss": 0.1964, "step": 6289 }, { "epoch": 4.608058608058608, "grad_norm": 0.58984375, "learning_rate": 9.820724365849718e-05, "loss": 0.1525, "step": 6290 }, { "epoch": 4.608791208791208, "grad_norm": 1.7265625, "learning_rate": 9.820632977206893e-05, "loss": 0.178, "step": 6291 }, { "epoch": 4.609523809523809, "grad_norm": 1.703125, "learning_rate": 9.820541565706398e-05, "loss": 0.1179, "step": 6292 }, { "epoch": 4.61025641025641, "grad_norm": 1.3828125, "learning_rate": 9.820450131348675e-05, "loss": 0.1426, "step": 6293 }, { "epoch": 4.610989010989011, "grad_norm": 0.78515625, "learning_rate": 9.82035867413416e-05, "loss": 0.1456, "step": 6294 }, { "epoch": 4.611721611721611, "grad_norm": 0.8984375, "learning_rate": 9.820267194063289e-05, "loss": 0.1662, "step": 6295 }, { "epoch": 4.612454212454212, "grad_norm": 1.75, "learning_rate": 9.820175691136502e-05, "loss": 0.184, "step": 6296 }, { "epoch": 4.613186813186813, "grad_norm": 1.84375, "learning_rate": 9.820084165354237e-05, "loss": 0.158, "step": 6297 }, { "epoch": 4.6139194139194135, "grad_norm": 1.25, "learning_rate": 9.819992616716932e-05, "loss": 0.124, "step": 6298 }, { "epoch": 4.614652014652014, "grad_norm": 1.453125, "learning_rate": 9.819901045225028e-05, "loss": 0.167, "step": 6299 }, { "epoch": 4.615384615384615, "grad_norm": 1.0546875, "learning_rate": 9.819809450878963e-05, "loss": 0.1484, "step": 6300 }, { "epoch": 4.616117216117216, "grad_norm": 1.484375, "learning_rate": 9.819717833679174e-05, "loss": 0.1902, "step": 6301 }, { "epoch": 4.616849816849816, "grad_norm": 2.046875, "learning_rate": 9.8196261936261e-05, "loss": 0.1848, "step": 6302 }, { "epoch": 4.617582417582417, "grad_norm": 1.1875, "learning_rate": 9.819534530720182e-05, "loss": 0.1602, "step": 6303 }, { "epoch": 4.618315018315018, "grad_norm": 0.87890625, "learning_rate": 9.819442844961859e-05, "loss": 0.1547, "step": 6304 }, { "epoch": 4.619047619047619, "grad_norm": 0.7265625, "learning_rate": 9.81935113635157e-05, "loss": 0.1646, "step": 6305 }, { "epoch": 4.619780219780219, "grad_norm": 0.828125, "learning_rate": 9.819259404889753e-05, "loss": 0.1336, "step": 6306 }, { "epoch": 4.62051282051282, "grad_norm": 1.3671875, "learning_rate": 9.819167650576849e-05, "loss": 0.1842, "step": 6307 }, { "epoch": 4.621245421245421, "grad_norm": 1.3359375, "learning_rate": 9.819075873413297e-05, "loss": 0.1613, "step": 6308 }, { "epoch": 4.6219780219780215, "grad_norm": 1.140625, "learning_rate": 9.818984073399537e-05, "loss": 0.1586, "step": 6309 }, { "epoch": 4.622710622710622, "grad_norm": 1.2890625, "learning_rate": 9.81889225053601e-05, "loss": 0.1551, "step": 6310 }, { "epoch": 4.623443223443223, "grad_norm": 0.63671875, "learning_rate": 9.81880040482315e-05, "loss": 0.1297, "step": 6311 }, { "epoch": 4.624175824175824, "grad_norm": 0.8359375, "learning_rate": 9.818708536261406e-05, "loss": 0.1211, "step": 6312 }, { "epoch": 4.6249084249084245, "grad_norm": 1.6953125, "learning_rate": 9.818616644851214e-05, "loss": 0.1647, "step": 6313 }, { "epoch": 4.625641025641025, "grad_norm": 0.71875, "learning_rate": 9.818524730593012e-05, "loss": 0.107, "step": 6314 }, { "epoch": 4.626373626373626, "grad_norm": 1.0390625, "learning_rate": 9.818432793487246e-05, "loss": 0.1517, "step": 6315 }, { "epoch": 4.627106227106227, "grad_norm": 0.6171875, "learning_rate": 9.81834083353435e-05, "loss": 0.1369, "step": 6316 }, { "epoch": 4.627838827838827, "grad_norm": 0.859375, "learning_rate": 9.81824885073477e-05, "loss": 0.1733, "step": 6317 }, { "epoch": 4.628571428571428, "grad_norm": 0.99609375, "learning_rate": 9.81815684508894e-05, "loss": 0.1729, "step": 6318 }, { "epoch": 4.629304029304029, "grad_norm": 1.8203125, "learning_rate": 9.81806481659731e-05, "loss": 0.1569, "step": 6319 }, { "epoch": 4.63003663003663, "grad_norm": 1.203125, "learning_rate": 9.817972765260315e-05, "loss": 0.1472, "step": 6320 }, { "epoch": 4.63076923076923, "grad_norm": 0.76953125, "learning_rate": 9.817880691078397e-05, "loss": 0.1646, "step": 6321 }, { "epoch": 4.631501831501831, "grad_norm": 0.9375, "learning_rate": 9.817788594051997e-05, "loss": 0.1372, "step": 6322 }, { "epoch": 4.632234432234432, "grad_norm": 1.09375, "learning_rate": 9.817696474181557e-05, "loss": 0.1637, "step": 6323 }, { "epoch": 4.6329670329670325, "grad_norm": 1.1015625, "learning_rate": 9.817604331467519e-05, "loss": 0.126, "step": 6324 }, { "epoch": 4.633699633699633, "grad_norm": 1.65625, "learning_rate": 9.817512165910322e-05, "loss": 0.1406, "step": 6325 }, { "epoch": 4.634432234432234, "grad_norm": 1.171875, "learning_rate": 9.817419977510409e-05, "loss": 0.142, "step": 6326 }, { "epoch": 4.635164835164835, "grad_norm": 1.0859375, "learning_rate": 9.817327766268223e-05, "loss": 0.1742, "step": 6327 }, { "epoch": 4.635897435897435, "grad_norm": 1.78125, "learning_rate": 9.817235532184205e-05, "loss": 0.1565, "step": 6328 }, { "epoch": 4.636630036630036, "grad_norm": 1.1875, "learning_rate": 9.817143275258796e-05, "loss": 0.1329, "step": 6329 }, { "epoch": 4.637362637362637, "grad_norm": 0.85546875, "learning_rate": 9.817050995492439e-05, "loss": 0.1113, "step": 6330 }, { "epoch": 4.638095238095238, "grad_norm": 1.1015625, "learning_rate": 9.816958692885574e-05, "loss": 0.2248, "step": 6331 }, { "epoch": 4.638827838827838, "grad_norm": 0.84765625, "learning_rate": 9.816866367438646e-05, "loss": 0.1446, "step": 6332 }, { "epoch": 4.639560439560439, "grad_norm": 0.69921875, "learning_rate": 9.816774019152096e-05, "loss": 0.1264, "step": 6333 }, { "epoch": 4.64029304029304, "grad_norm": 0.90625, "learning_rate": 9.816681648026368e-05, "loss": 0.1279, "step": 6334 }, { "epoch": 4.641025641025641, "grad_norm": 1.1875, "learning_rate": 9.816589254061903e-05, "loss": 0.1414, "step": 6335 }, { "epoch": 4.641758241758242, "grad_norm": 1.1796875, "learning_rate": 9.816496837259144e-05, "loss": 0.122, "step": 6336 }, { "epoch": 4.642490842490843, "grad_norm": 1.15625, "learning_rate": 9.816404397618534e-05, "loss": 0.1144, "step": 6337 }, { "epoch": 4.643223443223444, "grad_norm": 2.03125, "learning_rate": 9.816311935140516e-05, "loss": 0.1598, "step": 6338 }, { "epoch": 4.643956043956044, "grad_norm": 2.09375, "learning_rate": 9.816219449825531e-05, "loss": 0.1634, "step": 6339 }, { "epoch": 4.644688644688645, "grad_norm": 1.0234375, "learning_rate": 9.816126941674027e-05, "loss": 0.1374, "step": 6340 }, { "epoch": 4.645421245421246, "grad_norm": 0.9453125, "learning_rate": 9.816034410686442e-05, "loss": 0.1492, "step": 6341 }, { "epoch": 4.6461538461538465, "grad_norm": 1.2265625, "learning_rate": 9.815941856863222e-05, "loss": 0.1767, "step": 6342 }, { "epoch": 4.646886446886447, "grad_norm": 0.94921875, "learning_rate": 9.81584928020481e-05, "loss": 0.1558, "step": 6343 }, { "epoch": 4.647619047619048, "grad_norm": 1.0703125, "learning_rate": 9.81575668071165e-05, "loss": 0.1281, "step": 6344 }, { "epoch": 4.648351648351649, "grad_norm": 1.453125, "learning_rate": 9.815664058384186e-05, "loss": 0.1413, "step": 6345 }, { "epoch": 4.649084249084249, "grad_norm": 0.92578125, "learning_rate": 9.81557141322286e-05, "loss": 0.147, "step": 6346 }, { "epoch": 4.64981684981685, "grad_norm": 1.1484375, "learning_rate": 9.815478745228117e-05, "loss": 0.1435, "step": 6347 }, { "epoch": 4.650549450549451, "grad_norm": 1.3828125, "learning_rate": 9.815386054400401e-05, "loss": 0.1759, "step": 6348 }, { "epoch": 4.651282051282052, "grad_norm": 0.6484375, "learning_rate": 9.815293340740155e-05, "loss": 0.1334, "step": 6349 }, { "epoch": 4.652014652014652, "grad_norm": 0.53125, "learning_rate": 9.815200604247826e-05, "loss": 0.1318, "step": 6350 }, { "epoch": 4.652747252747253, "grad_norm": 0.921875, "learning_rate": 9.815107844923856e-05, "loss": 0.1512, "step": 6351 }, { "epoch": 4.653479853479854, "grad_norm": 1.3984375, "learning_rate": 9.81501506276869e-05, "loss": 0.1329, "step": 6352 }, { "epoch": 4.6542124542124546, "grad_norm": 1.0, "learning_rate": 9.814922257782772e-05, "loss": 0.1375, "step": 6353 }, { "epoch": 4.654945054945055, "grad_norm": 1.28125, "learning_rate": 9.814829429966548e-05, "loss": 0.1209, "step": 6354 }, { "epoch": 4.655677655677656, "grad_norm": 0.82421875, "learning_rate": 9.814736579320461e-05, "loss": 0.1449, "step": 6355 }, { "epoch": 4.656410256410257, "grad_norm": 1.21875, "learning_rate": 9.814643705844958e-05, "loss": 0.162, "step": 6356 }, { "epoch": 4.6571428571428575, "grad_norm": 1.2890625, "learning_rate": 9.814550809540484e-05, "loss": 0.1771, "step": 6357 }, { "epoch": 4.657875457875458, "grad_norm": 1.875, "learning_rate": 9.81445789040748e-05, "loss": 0.158, "step": 6358 }, { "epoch": 4.658608058608059, "grad_norm": 1.3125, "learning_rate": 9.814364948446396e-05, "loss": 0.1618, "step": 6359 }, { "epoch": 4.65934065934066, "grad_norm": 0.94921875, "learning_rate": 9.814271983657675e-05, "loss": 0.1516, "step": 6360 }, { "epoch": 4.66007326007326, "grad_norm": 1.3984375, "learning_rate": 9.814178996041763e-05, "loss": 0.1848, "step": 6361 }, { "epoch": 4.660805860805861, "grad_norm": 1.1796875, "learning_rate": 9.814085985599106e-05, "loss": 0.1608, "step": 6362 }, { "epoch": 4.661538461538462, "grad_norm": 1.0546875, "learning_rate": 9.813992952330149e-05, "loss": 0.1411, "step": 6363 }, { "epoch": 4.662271062271063, "grad_norm": 1.578125, "learning_rate": 9.813899896235336e-05, "loss": 0.1752, "step": 6364 }, { "epoch": 4.663003663003663, "grad_norm": 1.1875, "learning_rate": 9.813806817315117e-05, "loss": 0.1684, "step": 6365 }, { "epoch": 4.663736263736264, "grad_norm": 1.078125, "learning_rate": 9.813713715569934e-05, "loss": 0.1537, "step": 6366 }, { "epoch": 4.664468864468865, "grad_norm": 1.2578125, "learning_rate": 9.813620591000236e-05, "loss": 0.138, "step": 6367 }, { "epoch": 4.6652014652014655, "grad_norm": 0.82421875, "learning_rate": 9.813527443606467e-05, "loss": 0.1327, "step": 6368 }, { "epoch": 4.665934065934066, "grad_norm": 0.99609375, "learning_rate": 9.813434273389077e-05, "loss": 0.1597, "step": 6369 }, { "epoch": 4.666666666666667, "grad_norm": 0.76171875, "learning_rate": 9.813341080348506e-05, "loss": 0.1376, "step": 6370 }, { "epoch": 4.667399267399268, "grad_norm": 0.66796875, "learning_rate": 9.813247864485205e-05, "loss": 0.1341, "step": 6371 }, { "epoch": 4.668131868131868, "grad_norm": 1.2734375, "learning_rate": 9.813154625799622e-05, "loss": 0.1715, "step": 6372 }, { "epoch": 4.668864468864469, "grad_norm": 0.921875, "learning_rate": 9.813061364292198e-05, "loss": 0.1571, "step": 6373 }, { "epoch": 4.66959706959707, "grad_norm": 1.171875, "learning_rate": 9.812968079963387e-05, "loss": 0.1261, "step": 6374 }, { "epoch": 4.670329670329671, "grad_norm": 1.59375, "learning_rate": 9.81287477281363e-05, "loss": 0.2177, "step": 6375 }, { "epoch": 4.671062271062271, "grad_norm": 0.80859375, "learning_rate": 9.812781442843378e-05, "loss": 0.1802, "step": 6376 }, { "epoch": 4.671794871794872, "grad_norm": 1.3359375, "learning_rate": 9.812688090053074e-05, "loss": 0.1392, "step": 6377 }, { "epoch": 4.672527472527473, "grad_norm": 1.1015625, "learning_rate": 9.81259471444317e-05, "loss": 0.1168, "step": 6378 }, { "epoch": 4.6732600732600735, "grad_norm": 1.3125, "learning_rate": 9.812501316014112e-05, "loss": 0.1638, "step": 6379 }, { "epoch": 4.673992673992674, "grad_norm": 0.87109375, "learning_rate": 9.812407894766344e-05, "loss": 0.1663, "step": 6380 }, { "epoch": 4.674725274725275, "grad_norm": 1.015625, "learning_rate": 9.81231445070032e-05, "loss": 0.1901, "step": 6381 }, { "epoch": 4.675457875457876, "grad_norm": 1.3203125, "learning_rate": 9.812220983816481e-05, "loss": 0.1457, "step": 6382 }, { "epoch": 4.6761904761904765, "grad_norm": 0.921875, "learning_rate": 9.81212749411528e-05, "loss": 0.1478, "step": 6383 }, { "epoch": 4.676923076923077, "grad_norm": 1.578125, "learning_rate": 9.812033981597162e-05, "loss": 0.1652, "step": 6384 }, { "epoch": 4.677655677655678, "grad_norm": 1.9140625, "learning_rate": 9.811940446262576e-05, "loss": 0.1831, "step": 6385 }, { "epoch": 4.678388278388279, "grad_norm": 2.375, "learning_rate": 9.811846888111969e-05, "loss": 0.1505, "step": 6386 }, { "epoch": 4.679120879120879, "grad_norm": 1.1640625, "learning_rate": 9.811753307145792e-05, "loss": 0.1495, "step": 6387 }, { "epoch": 4.67985347985348, "grad_norm": 1.0703125, "learning_rate": 9.811659703364491e-05, "loss": 0.1539, "step": 6388 }, { "epoch": 4.680586080586081, "grad_norm": 1.7734375, "learning_rate": 9.811566076768516e-05, "loss": 0.1664, "step": 6389 }, { "epoch": 4.681318681318682, "grad_norm": 1.2734375, "learning_rate": 9.811472427358314e-05, "loss": 0.1692, "step": 6390 }, { "epoch": 4.682051282051282, "grad_norm": 1.453125, "learning_rate": 9.811378755134334e-05, "loss": 0.1428, "step": 6391 }, { "epoch": 4.682783882783883, "grad_norm": 1.71875, "learning_rate": 9.811285060097026e-05, "loss": 0.1719, "step": 6392 }, { "epoch": 4.683516483516484, "grad_norm": 0.8203125, "learning_rate": 9.811191342246837e-05, "loss": 0.1528, "step": 6393 }, { "epoch": 4.6842490842490845, "grad_norm": 1.421875, "learning_rate": 9.81109760158422e-05, "loss": 0.1429, "step": 6394 }, { "epoch": 4.684981684981685, "grad_norm": 1.75, "learning_rate": 9.811003838109619e-05, "loss": 0.1857, "step": 6395 }, { "epoch": 4.685714285714286, "grad_norm": 1.265625, "learning_rate": 9.810910051823488e-05, "loss": 0.173, "step": 6396 }, { "epoch": 4.686446886446887, "grad_norm": 1.4296875, "learning_rate": 9.810816242726271e-05, "loss": 0.2029, "step": 6397 }, { "epoch": 4.687179487179487, "grad_norm": 0.72265625, "learning_rate": 9.810722410818421e-05, "loss": 0.1279, "step": 6398 }, { "epoch": 4.687912087912088, "grad_norm": 0.66796875, "learning_rate": 9.810628556100389e-05, "loss": 0.1172, "step": 6399 }, { "epoch": 4.688644688644689, "grad_norm": 2.03125, "learning_rate": 9.810534678572623e-05, "loss": 0.1641, "step": 6400 }, { "epoch": 4.68937728937729, "grad_norm": 2.25, "learning_rate": 9.810440778235571e-05, "loss": 0.1786, "step": 6401 }, { "epoch": 4.69010989010989, "grad_norm": 1.75, "learning_rate": 9.810346855089685e-05, "loss": 0.1857, "step": 6402 }, { "epoch": 4.690842490842491, "grad_norm": 1.0078125, "learning_rate": 9.810252909135415e-05, "loss": 0.1211, "step": 6403 }, { "epoch": 4.691575091575092, "grad_norm": 1.296875, "learning_rate": 9.81015894037321e-05, "loss": 0.1958, "step": 6404 }, { "epoch": 4.6923076923076925, "grad_norm": 1.6015625, "learning_rate": 9.810064948803521e-05, "loss": 0.1481, "step": 6405 }, { "epoch": 4.693040293040293, "grad_norm": 1.296875, "learning_rate": 9.809970934426799e-05, "loss": 0.1546, "step": 6406 }, { "epoch": 4.693772893772894, "grad_norm": 1.015625, "learning_rate": 9.809876897243492e-05, "loss": 0.151, "step": 6407 }, { "epoch": 4.694505494505495, "grad_norm": 1.09375, "learning_rate": 9.809782837254052e-05, "loss": 0.1471, "step": 6408 }, { "epoch": 4.695238095238095, "grad_norm": 0.94921875, "learning_rate": 9.809688754458931e-05, "loss": 0.1755, "step": 6409 }, { "epoch": 4.695970695970696, "grad_norm": 1.0859375, "learning_rate": 9.809594648858579e-05, "loss": 0.1422, "step": 6410 }, { "epoch": 4.696703296703297, "grad_norm": 1.375, "learning_rate": 9.809500520453446e-05, "loss": 0.1599, "step": 6411 }, { "epoch": 4.697435897435898, "grad_norm": 1.75, "learning_rate": 9.809406369243984e-05, "loss": 0.1431, "step": 6412 }, { "epoch": 4.698168498168498, "grad_norm": 1.171875, "learning_rate": 9.809312195230643e-05, "loss": 0.1798, "step": 6413 }, { "epoch": 4.698901098901099, "grad_norm": 1.1640625, "learning_rate": 9.809217998413877e-05, "loss": 0.1643, "step": 6414 }, { "epoch": 4.6996336996337, "grad_norm": 0.69140625, "learning_rate": 9.809123778794132e-05, "loss": 0.1481, "step": 6415 }, { "epoch": 4.7003663003663005, "grad_norm": 0.8515625, "learning_rate": 9.809029536371865e-05, "loss": 0.1605, "step": 6416 }, { "epoch": 4.701098901098901, "grad_norm": 1.5, "learning_rate": 9.808935271147524e-05, "loss": 0.2049, "step": 6417 }, { "epoch": 4.701831501831502, "grad_norm": 1.03125, "learning_rate": 9.808840983121562e-05, "loss": 0.166, "step": 6418 }, { "epoch": 4.702564102564103, "grad_norm": 1.28125, "learning_rate": 9.808746672294432e-05, "loss": 0.1381, "step": 6419 }, { "epoch": 4.7032967032967035, "grad_norm": 0.73828125, "learning_rate": 9.808652338666582e-05, "loss": 0.1261, "step": 6420 }, { "epoch": 4.704029304029304, "grad_norm": 1.1875, "learning_rate": 9.808557982238468e-05, "loss": 0.1589, "step": 6421 }, { "epoch": 4.704761904761905, "grad_norm": 1.0546875, "learning_rate": 9.80846360301054e-05, "loss": 0.1555, "step": 6422 }, { "epoch": 4.705494505494506, "grad_norm": 0.73828125, "learning_rate": 9.808369200983252e-05, "loss": 0.1236, "step": 6423 }, { "epoch": 4.706227106227106, "grad_norm": 0.78125, "learning_rate": 9.808274776157055e-05, "loss": 0.1462, "step": 6424 }, { "epoch": 4.706959706959707, "grad_norm": 0.6015625, "learning_rate": 9.8081803285324e-05, "loss": 0.1154, "step": 6425 }, { "epoch": 4.707692307692308, "grad_norm": 0.94921875, "learning_rate": 9.808085858109741e-05, "loss": 0.1165, "step": 6426 }, { "epoch": 4.708424908424909, "grad_norm": 0.53125, "learning_rate": 9.807991364889532e-05, "loss": 0.1319, "step": 6427 }, { "epoch": 4.709157509157509, "grad_norm": 1.1484375, "learning_rate": 9.807896848872224e-05, "loss": 0.1529, "step": 6428 }, { "epoch": 4.70989010989011, "grad_norm": 0.7734375, "learning_rate": 9.80780231005827e-05, "loss": 0.1416, "step": 6429 }, { "epoch": 4.710622710622711, "grad_norm": 0.703125, "learning_rate": 9.807707748448124e-05, "loss": 0.1377, "step": 6430 }, { "epoch": 4.7113553113553115, "grad_norm": 0.94921875, "learning_rate": 9.807613164042237e-05, "loss": 0.169, "step": 6431 }, { "epoch": 4.712087912087912, "grad_norm": 1.015625, "learning_rate": 9.807518556841065e-05, "loss": 0.1135, "step": 6432 }, { "epoch": 4.712820512820513, "grad_norm": 0.828125, "learning_rate": 9.807423926845059e-05, "loss": 0.1639, "step": 6433 }, { "epoch": 4.713553113553114, "grad_norm": 0.77734375, "learning_rate": 9.807329274054672e-05, "loss": 0.1643, "step": 6434 }, { "epoch": 4.714285714285714, "grad_norm": 1.609375, "learning_rate": 9.807234598470361e-05, "loss": 0.1453, "step": 6435 }, { "epoch": 4.715018315018315, "grad_norm": 0.8359375, "learning_rate": 9.807139900092577e-05, "loss": 0.1362, "step": 6436 }, { "epoch": 4.715750915750916, "grad_norm": 0.99609375, "learning_rate": 9.807045178921773e-05, "loss": 0.1555, "step": 6437 }, { "epoch": 4.716483516483517, "grad_norm": 1.0546875, "learning_rate": 9.806950434958404e-05, "loss": 0.1725, "step": 6438 }, { "epoch": 4.717216117216117, "grad_norm": 0.76953125, "learning_rate": 9.806855668202926e-05, "loss": 0.1144, "step": 6439 }, { "epoch": 4.717948717948718, "grad_norm": 1.375, "learning_rate": 9.806760878655788e-05, "loss": 0.1364, "step": 6440 }, { "epoch": 4.718681318681319, "grad_norm": 1.0234375, "learning_rate": 9.806666066317448e-05, "loss": 0.1548, "step": 6441 }, { "epoch": 4.7194139194139195, "grad_norm": 1.234375, "learning_rate": 9.806571231188361e-05, "loss": 0.1209, "step": 6442 }, { "epoch": 4.72014652014652, "grad_norm": 0.91796875, "learning_rate": 9.806476373268976e-05, "loss": 0.1674, "step": 6443 }, { "epoch": 4.720879120879121, "grad_norm": 1.71875, "learning_rate": 9.806381492559754e-05, "loss": 0.1452, "step": 6444 }, { "epoch": 4.721611721611722, "grad_norm": 3.015625, "learning_rate": 9.806286589061147e-05, "loss": 0.1605, "step": 6445 }, { "epoch": 4.722344322344322, "grad_norm": 1.96875, "learning_rate": 9.80619166277361e-05, "loss": 0.1571, "step": 6446 }, { "epoch": 4.723076923076923, "grad_norm": 1.859375, "learning_rate": 9.806096713697594e-05, "loss": 0.1745, "step": 6447 }, { "epoch": 4.723809523809524, "grad_norm": 1.34375, "learning_rate": 9.806001741833562e-05, "loss": 0.1365, "step": 6448 }, { "epoch": 4.724542124542125, "grad_norm": 1.2578125, "learning_rate": 9.805906747181961e-05, "loss": 0.1177, "step": 6449 }, { "epoch": 4.725274725274725, "grad_norm": 1.4140625, "learning_rate": 9.805811729743251e-05, "loss": 0.1613, "step": 6450 }, { "epoch": 4.726007326007326, "grad_norm": 1.0390625, "learning_rate": 9.805716689517887e-05, "loss": 0.143, "step": 6451 }, { "epoch": 4.726739926739927, "grad_norm": 0.8046875, "learning_rate": 9.805621626506319e-05, "loss": 0.1239, "step": 6452 }, { "epoch": 4.7274725274725276, "grad_norm": 0.69140625, "learning_rate": 9.80552654070901e-05, "loss": 0.1306, "step": 6453 }, { "epoch": 4.728205128205128, "grad_norm": 0.88671875, "learning_rate": 9.805431432126412e-05, "loss": 0.1356, "step": 6454 }, { "epoch": 4.728937728937729, "grad_norm": 1.0078125, "learning_rate": 9.805336300758979e-05, "loss": 0.1317, "step": 6455 }, { "epoch": 4.72967032967033, "grad_norm": 1.25, "learning_rate": 9.805241146607171e-05, "loss": 0.1726, "step": 6456 }, { "epoch": 4.7304029304029305, "grad_norm": 1.5703125, "learning_rate": 9.80514596967144e-05, "loss": 0.1525, "step": 6457 }, { "epoch": 4.731135531135531, "grad_norm": 0.8203125, "learning_rate": 9.805050769952245e-05, "loss": 0.1623, "step": 6458 }, { "epoch": 4.731868131868132, "grad_norm": 1.0234375, "learning_rate": 9.804955547450039e-05, "loss": 0.1604, "step": 6459 }, { "epoch": 4.732600732600733, "grad_norm": 0.98046875, "learning_rate": 9.80486030216528e-05, "loss": 0.1188, "step": 6460 }, { "epoch": 4.733333333333333, "grad_norm": 0.65234375, "learning_rate": 9.804765034098428e-05, "loss": 0.1472, "step": 6461 }, { "epoch": 4.734065934065934, "grad_norm": 1.109375, "learning_rate": 9.804669743249932e-05, "loss": 0.1836, "step": 6462 }, { "epoch": 4.734798534798535, "grad_norm": 0.765625, "learning_rate": 9.804574429620254e-05, "loss": 0.1547, "step": 6463 }, { "epoch": 4.735531135531136, "grad_norm": 0.71875, "learning_rate": 9.804479093209847e-05, "loss": 0.1375, "step": 6464 }, { "epoch": 4.736263736263736, "grad_norm": 1.078125, "learning_rate": 9.804383734019173e-05, "loss": 0.1636, "step": 6465 }, { "epoch": 4.736996336996337, "grad_norm": 1.2109375, "learning_rate": 9.804288352048684e-05, "loss": 0.1602, "step": 6466 }, { "epoch": 4.737728937728938, "grad_norm": 1.0, "learning_rate": 9.80419294729884e-05, "loss": 0.1693, "step": 6467 }, { "epoch": 4.7384615384615385, "grad_norm": 1.375, "learning_rate": 9.804097519770096e-05, "loss": 0.1922, "step": 6468 }, { "epoch": 4.739194139194139, "grad_norm": 0.98046875, "learning_rate": 9.804002069462909e-05, "loss": 0.1554, "step": 6469 }, { "epoch": 4.73992673992674, "grad_norm": 1.0546875, "learning_rate": 9.803906596377738e-05, "loss": 0.1418, "step": 6470 }, { "epoch": 4.740659340659341, "grad_norm": 1.1640625, "learning_rate": 9.803811100515041e-05, "loss": 0.1862, "step": 6471 }, { "epoch": 4.741391941391941, "grad_norm": 0.91796875, "learning_rate": 9.803715581875274e-05, "loss": 0.1557, "step": 6472 }, { "epoch": 4.742124542124542, "grad_norm": 0.53515625, "learning_rate": 9.803620040458895e-05, "loss": 0.1274, "step": 6473 }, { "epoch": 4.742857142857143, "grad_norm": 0.98828125, "learning_rate": 9.803524476266362e-05, "loss": 0.1609, "step": 6474 }, { "epoch": 4.743589743589744, "grad_norm": 0.9453125, "learning_rate": 9.803428889298133e-05, "loss": 0.1597, "step": 6475 }, { "epoch": 4.744322344322344, "grad_norm": 0.953125, "learning_rate": 9.803333279554666e-05, "loss": 0.1506, "step": 6476 }, { "epoch": 4.745054945054945, "grad_norm": 0.6953125, "learning_rate": 9.803237647036417e-05, "loss": 0.141, "step": 6477 }, { "epoch": 4.745787545787546, "grad_norm": 0.94140625, "learning_rate": 9.803141991743848e-05, "loss": 0.1922, "step": 6478 }, { "epoch": 4.7465201465201465, "grad_norm": 1.4453125, "learning_rate": 9.803046313677414e-05, "loss": 0.137, "step": 6479 }, { "epoch": 4.747252747252747, "grad_norm": 1.390625, "learning_rate": 9.802950612837576e-05, "loss": 0.1809, "step": 6480 }, { "epoch": 4.747985347985348, "grad_norm": 1.234375, "learning_rate": 9.80285488922479e-05, "loss": 0.1332, "step": 6481 }, { "epoch": 4.748717948717949, "grad_norm": 1.203125, "learning_rate": 9.802759142839517e-05, "loss": 0.133, "step": 6482 }, { "epoch": 4.7494505494505495, "grad_norm": 0.6640625, "learning_rate": 9.802663373682213e-05, "loss": 0.121, "step": 6483 }, { "epoch": 4.75018315018315, "grad_norm": 1.046875, "learning_rate": 9.802567581753342e-05, "loss": 0.1412, "step": 6484 }, { "epoch": 4.750915750915751, "grad_norm": 0.8359375, "learning_rate": 9.802471767053356e-05, "loss": 0.1428, "step": 6485 }, { "epoch": 4.751648351648352, "grad_norm": 0.80078125, "learning_rate": 9.80237592958272e-05, "loss": 0.1388, "step": 6486 }, { "epoch": 4.752380952380952, "grad_norm": 0.984375, "learning_rate": 9.802280069341891e-05, "loss": 0.1651, "step": 6487 }, { "epoch": 4.753113553113553, "grad_norm": 1.1328125, "learning_rate": 9.802184186331326e-05, "loss": 0.1448, "step": 6488 }, { "epoch": 4.753846153846154, "grad_norm": 1.203125, "learning_rate": 9.802088280551487e-05, "loss": 0.1775, "step": 6489 }, { "epoch": 4.754578754578755, "grad_norm": 1.03125, "learning_rate": 9.801992352002834e-05, "loss": 0.1615, "step": 6490 }, { "epoch": 4.755311355311355, "grad_norm": 0.9375, "learning_rate": 9.801896400685826e-05, "loss": 0.1323, "step": 6491 }, { "epoch": 4.756043956043956, "grad_norm": 1.0, "learning_rate": 9.801800426600921e-05, "loss": 0.1657, "step": 6492 }, { "epoch": 4.756776556776557, "grad_norm": 1.0859375, "learning_rate": 9.801704429748581e-05, "loss": 0.1412, "step": 6493 }, { "epoch": 4.7575091575091575, "grad_norm": 1.5390625, "learning_rate": 9.801608410129267e-05, "loss": 0.1568, "step": 6494 }, { "epoch": 4.758241758241758, "grad_norm": 1.4453125, "learning_rate": 9.801512367743434e-05, "loss": 0.1644, "step": 6495 }, { "epoch": 4.758974358974359, "grad_norm": 1.3046875, "learning_rate": 9.801416302591547e-05, "loss": 0.1857, "step": 6496 }, { "epoch": 4.75970695970696, "grad_norm": 0.5625, "learning_rate": 9.801320214674064e-05, "loss": 0.1339, "step": 6497 }, { "epoch": 4.76043956043956, "grad_norm": 0.90234375, "learning_rate": 9.801224103991448e-05, "loss": 0.1505, "step": 6498 }, { "epoch": 4.761172161172161, "grad_norm": 0.78515625, "learning_rate": 9.801127970544155e-05, "loss": 0.1302, "step": 6499 }, { "epoch": 4.761904761904762, "grad_norm": 1.34375, "learning_rate": 9.801031814332651e-05, "loss": 0.1697, "step": 6500 }, { "epoch": 4.762637362637363, "grad_norm": 1.453125, "learning_rate": 9.800935635357392e-05, "loss": 0.1319, "step": 6501 }, { "epoch": 4.763369963369963, "grad_norm": 0.97265625, "learning_rate": 9.80083943361884e-05, "loss": 0.1362, "step": 6502 }, { "epoch": 4.764102564102564, "grad_norm": 1.4921875, "learning_rate": 9.800743209117459e-05, "loss": 0.1883, "step": 6503 }, { "epoch": 4.764835164835165, "grad_norm": 1.5, "learning_rate": 9.800646961853706e-05, "loss": 0.1275, "step": 6504 }, { "epoch": 4.7655677655677655, "grad_norm": 2.046875, "learning_rate": 9.800550691828043e-05, "loss": 0.1733, "step": 6505 }, { "epoch": 4.766300366300366, "grad_norm": 1.1328125, "learning_rate": 9.800454399040936e-05, "loss": 0.1571, "step": 6506 }, { "epoch": 4.767032967032967, "grad_norm": 1.6796875, "learning_rate": 9.800358083492839e-05, "loss": 0.1311, "step": 6507 }, { "epoch": 4.767765567765568, "grad_norm": 1.7890625, "learning_rate": 9.800261745184218e-05, "loss": 0.1763, "step": 6508 }, { "epoch": 4.768498168498168, "grad_norm": 1.296875, "learning_rate": 9.800165384115534e-05, "loss": 0.1426, "step": 6509 }, { "epoch": 4.769230769230769, "grad_norm": 1.4140625, "learning_rate": 9.800069000287246e-05, "loss": 0.1522, "step": 6510 }, { "epoch": 4.76996336996337, "grad_norm": 0.76953125, "learning_rate": 9.79997259369982e-05, "loss": 0.1642, "step": 6511 }, { "epoch": 4.770695970695971, "grad_norm": 1.1640625, "learning_rate": 9.799876164353717e-05, "loss": 0.1787, "step": 6512 }, { "epoch": 4.771428571428571, "grad_norm": 1.0078125, "learning_rate": 9.799779712249397e-05, "loss": 0.1443, "step": 6513 }, { "epoch": 4.772161172161172, "grad_norm": 1.2265625, "learning_rate": 9.799683237387323e-05, "loss": 0.1654, "step": 6514 }, { "epoch": 4.772893772893773, "grad_norm": 1.0625, "learning_rate": 9.799586739767957e-05, "loss": 0.1387, "step": 6515 }, { "epoch": 4.7736263736263735, "grad_norm": 0.9296875, "learning_rate": 9.799490219391762e-05, "loss": 0.146, "step": 6516 }, { "epoch": 4.774358974358974, "grad_norm": 1.15625, "learning_rate": 9.799393676259201e-05, "loss": 0.1566, "step": 6517 }, { "epoch": 4.775091575091575, "grad_norm": 0.89453125, "learning_rate": 9.799297110370734e-05, "loss": 0.1392, "step": 6518 }, { "epoch": 4.775824175824176, "grad_norm": 1.1953125, "learning_rate": 9.799200521726827e-05, "loss": 0.1417, "step": 6519 }, { "epoch": 4.7765567765567765, "grad_norm": 0.94140625, "learning_rate": 9.799103910327942e-05, "loss": 0.2021, "step": 6520 }, { "epoch": 4.777289377289377, "grad_norm": 0.84765625, "learning_rate": 9.79900727617454e-05, "loss": 0.1403, "step": 6521 }, { "epoch": 4.778021978021978, "grad_norm": 1.375, "learning_rate": 9.798910619267084e-05, "loss": 0.166, "step": 6522 }, { "epoch": 4.778754578754579, "grad_norm": 1.1640625, "learning_rate": 9.798813939606039e-05, "loss": 0.1405, "step": 6523 }, { "epoch": 4.779487179487179, "grad_norm": 1.3828125, "learning_rate": 9.79871723719187e-05, "loss": 0.1443, "step": 6524 }, { "epoch": 4.78021978021978, "grad_norm": 2.015625, "learning_rate": 9.798620512025035e-05, "loss": 0.2032, "step": 6525 }, { "epoch": 4.780952380952381, "grad_norm": 1.3203125, "learning_rate": 9.798523764106e-05, "loss": 0.1416, "step": 6526 }, { "epoch": 4.781684981684982, "grad_norm": 1.453125, "learning_rate": 9.79842699343523e-05, "loss": 0.1502, "step": 6527 }, { "epoch": 4.782417582417582, "grad_norm": 0.65625, "learning_rate": 9.798330200013188e-05, "loss": 0.1323, "step": 6528 }, { "epoch": 4.783150183150183, "grad_norm": 1.25, "learning_rate": 9.798233383840336e-05, "loss": 0.1988, "step": 6529 }, { "epoch": 4.783882783882784, "grad_norm": 1.2265625, "learning_rate": 9.798136544917141e-05, "loss": 0.1467, "step": 6530 }, { "epoch": 4.7846153846153845, "grad_norm": 0.8671875, "learning_rate": 9.798039683244063e-05, "loss": 0.1675, "step": 6531 }, { "epoch": 4.785347985347985, "grad_norm": 0.6484375, "learning_rate": 9.797942798821567e-05, "loss": 0.1669, "step": 6532 }, { "epoch": 4.786080586080586, "grad_norm": 0.93359375, "learning_rate": 9.797845891650123e-05, "loss": 0.1487, "step": 6533 }, { "epoch": 4.786813186813187, "grad_norm": 0.8984375, "learning_rate": 9.797748961730186e-05, "loss": 0.1786, "step": 6534 }, { "epoch": 4.787545787545787, "grad_norm": 0.765625, "learning_rate": 9.797652009062229e-05, "loss": 0.1683, "step": 6535 }, { "epoch": 4.788278388278388, "grad_norm": 2.046875, "learning_rate": 9.79755503364671e-05, "loss": 0.189, "step": 6536 }, { "epoch": 4.789010989010989, "grad_norm": 1.6015625, "learning_rate": 9.797458035484096e-05, "loss": 0.1228, "step": 6537 }, { "epoch": 4.78974358974359, "grad_norm": 1.1640625, "learning_rate": 9.797361014574855e-05, "loss": 0.1574, "step": 6538 }, { "epoch": 4.79047619047619, "grad_norm": 0.90234375, "learning_rate": 9.797263970919446e-05, "loss": 0.1553, "step": 6539 }, { "epoch": 4.791208791208791, "grad_norm": 0.8984375, "learning_rate": 9.797166904518338e-05, "loss": 0.1485, "step": 6540 }, { "epoch": 4.791941391941392, "grad_norm": 1.6796875, "learning_rate": 9.797069815371995e-05, "loss": 0.1206, "step": 6541 }, { "epoch": 4.7926739926739925, "grad_norm": 1.3125, "learning_rate": 9.796972703480882e-05, "loss": 0.169, "step": 6542 }, { "epoch": 4.793406593406593, "grad_norm": 0.91015625, "learning_rate": 9.796875568845464e-05, "loss": 0.1425, "step": 6543 }, { "epoch": 4.794139194139194, "grad_norm": 0.578125, "learning_rate": 9.796778411466208e-05, "loss": 0.1289, "step": 6544 }, { "epoch": 4.794871794871795, "grad_norm": 0.84375, "learning_rate": 9.796681231343577e-05, "loss": 0.1489, "step": 6545 }, { "epoch": 4.795604395604395, "grad_norm": 1.21875, "learning_rate": 9.79658402847804e-05, "loss": 0.1528, "step": 6546 }, { "epoch": 4.796336996336996, "grad_norm": 0.578125, "learning_rate": 9.79648680287006e-05, "loss": 0.1542, "step": 6547 }, { "epoch": 4.797069597069597, "grad_norm": 1.3125, "learning_rate": 9.796389554520102e-05, "loss": 0.2105, "step": 6548 }, { "epoch": 4.797802197802198, "grad_norm": 1.4453125, "learning_rate": 9.796292283428635e-05, "loss": 0.1474, "step": 6549 }, { "epoch": 4.798534798534798, "grad_norm": 1.078125, "learning_rate": 9.796194989596122e-05, "loss": 0.1606, "step": 6550 }, { "epoch": 4.799267399267399, "grad_norm": 0.61328125, "learning_rate": 9.796097673023032e-05, "loss": 0.1292, "step": 6551 }, { "epoch": 4.8, "grad_norm": 0.91796875, "learning_rate": 9.79600033370983e-05, "loss": 0.1656, "step": 6552 }, { "epoch": 4.8007326007326006, "grad_norm": 1.53125, "learning_rate": 9.795902971656982e-05, "loss": 0.2101, "step": 6553 }, { "epoch": 4.801465201465201, "grad_norm": 0.80859375, "learning_rate": 9.795805586864955e-05, "loss": 0.1466, "step": 6554 }, { "epoch": 4.802197802197802, "grad_norm": 1.2265625, "learning_rate": 9.795708179334217e-05, "loss": 0.1922, "step": 6555 }, { "epoch": 4.802930402930403, "grad_norm": 1.0625, "learning_rate": 9.79561074906523e-05, "loss": 0.1503, "step": 6556 }, { "epoch": 4.8036630036630035, "grad_norm": 0.92578125, "learning_rate": 9.795513296058467e-05, "loss": 0.158, "step": 6557 }, { "epoch": 4.804395604395604, "grad_norm": 1.25, "learning_rate": 9.795415820314391e-05, "loss": 0.1567, "step": 6558 }, { "epoch": 4.805128205128205, "grad_norm": 0.82421875, "learning_rate": 9.79531832183347e-05, "loss": 0.1475, "step": 6559 }, { "epoch": 4.805860805860806, "grad_norm": 1.0703125, "learning_rate": 9.79522080061617e-05, "loss": 0.123, "step": 6560 }, { "epoch": 4.806593406593406, "grad_norm": 0.609375, "learning_rate": 9.795123256662961e-05, "loss": 0.1172, "step": 6561 }, { "epoch": 4.807326007326007, "grad_norm": 1.2109375, "learning_rate": 9.795025689974308e-05, "loss": 0.1896, "step": 6562 }, { "epoch": 4.808058608058608, "grad_norm": 0.875, "learning_rate": 9.79492810055068e-05, "loss": 0.1162, "step": 6563 }, { "epoch": 4.808791208791209, "grad_norm": 1.3046875, "learning_rate": 9.794830488392543e-05, "loss": 0.1524, "step": 6564 }, { "epoch": 4.809523809523809, "grad_norm": 0.9765625, "learning_rate": 9.794732853500365e-05, "loss": 0.1796, "step": 6565 }, { "epoch": 4.81025641025641, "grad_norm": 1.1328125, "learning_rate": 9.794635195874615e-05, "loss": 0.1824, "step": 6566 }, { "epoch": 4.810989010989011, "grad_norm": 0.90234375, "learning_rate": 9.79453751551576e-05, "loss": 0.1319, "step": 6567 }, { "epoch": 4.8117216117216115, "grad_norm": 0.83203125, "learning_rate": 9.794439812424269e-05, "loss": 0.1738, "step": 6568 }, { "epoch": 4.812454212454212, "grad_norm": 1.2265625, "learning_rate": 9.794342086600609e-05, "loss": 0.1924, "step": 6569 }, { "epoch": 4.813186813186813, "grad_norm": 1.2109375, "learning_rate": 9.794244338045248e-05, "loss": 0.1434, "step": 6570 }, { "epoch": 4.813919413919414, "grad_norm": 1.7421875, "learning_rate": 9.794146566758654e-05, "loss": 0.1736, "step": 6571 }, { "epoch": 4.814652014652014, "grad_norm": 1.515625, "learning_rate": 9.794048772741298e-05, "loss": 0.2103, "step": 6572 }, { "epoch": 4.815384615384615, "grad_norm": 1.25, "learning_rate": 9.793950955993645e-05, "loss": 0.1739, "step": 6573 }, { "epoch": 4.816117216117216, "grad_norm": 1.140625, "learning_rate": 9.79385311651617e-05, "loss": 0.1545, "step": 6574 }, { "epoch": 4.816849816849817, "grad_norm": 1.0703125, "learning_rate": 9.793755254309333e-05, "loss": 0.171, "step": 6575 }, { "epoch": 4.817582417582417, "grad_norm": 1.25, "learning_rate": 9.79365736937361e-05, "loss": 0.1277, "step": 6576 }, { "epoch": 4.818315018315018, "grad_norm": 1.4609375, "learning_rate": 9.793559461709466e-05, "loss": 0.1806, "step": 6577 }, { "epoch": 4.819047619047619, "grad_norm": 1.28125, "learning_rate": 9.793461531317372e-05, "loss": 0.1902, "step": 6578 }, { "epoch": 4.8197802197802195, "grad_norm": 1.0, "learning_rate": 9.793363578197797e-05, "loss": 0.1814, "step": 6579 }, { "epoch": 4.82051282051282, "grad_norm": 0.81640625, "learning_rate": 9.793265602351208e-05, "loss": 0.1417, "step": 6580 }, { "epoch": 4.821245421245421, "grad_norm": 0.765625, "learning_rate": 9.793167603778079e-05, "loss": 0.1758, "step": 6581 }, { "epoch": 4.821978021978022, "grad_norm": 0.87890625, "learning_rate": 9.793069582478874e-05, "loss": 0.1341, "step": 6582 }, { "epoch": 4.8227106227106225, "grad_norm": 1.03125, "learning_rate": 9.792971538454067e-05, "loss": 0.157, "step": 6583 }, { "epoch": 4.823443223443223, "grad_norm": 1.0, "learning_rate": 9.792873471704128e-05, "loss": 0.1325, "step": 6584 }, { "epoch": 4.824175824175824, "grad_norm": 0.8359375, "learning_rate": 9.792775382229523e-05, "loss": 0.156, "step": 6585 }, { "epoch": 4.824908424908425, "grad_norm": 0.83203125, "learning_rate": 9.792677270030723e-05, "loss": 0.1844, "step": 6586 }, { "epoch": 4.825641025641025, "grad_norm": 1.328125, "learning_rate": 9.792579135108202e-05, "loss": 0.188, "step": 6587 }, { "epoch": 4.826373626373626, "grad_norm": 1.1796875, "learning_rate": 9.792480977462427e-05, "loss": 0.1673, "step": 6588 }, { "epoch": 4.827106227106227, "grad_norm": 0.91796875, "learning_rate": 9.79238279709387e-05, "loss": 0.1177, "step": 6589 }, { "epoch": 4.827838827838828, "grad_norm": 1.453125, "learning_rate": 9.792284594002997e-05, "loss": 0.1708, "step": 6590 }, { "epoch": 4.828571428571428, "grad_norm": 1.1640625, "learning_rate": 9.792186368190284e-05, "loss": 0.1339, "step": 6591 }, { "epoch": 4.829304029304029, "grad_norm": 1.5390625, "learning_rate": 9.792088119656199e-05, "loss": 0.1644, "step": 6592 }, { "epoch": 4.83003663003663, "grad_norm": 1.1796875, "learning_rate": 9.791989848401213e-05, "loss": 0.1373, "step": 6593 }, { "epoch": 4.8307692307692305, "grad_norm": 1.921875, "learning_rate": 9.791891554425797e-05, "loss": 0.1807, "step": 6594 }, { "epoch": 4.831501831501831, "grad_norm": 1.765625, "learning_rate": 9.79179323773042e-05, "loss": 0.1638, "step": 6595 }, { "epoch": 4.832234432234432, "grad_norm": 0.875, "learning_rate": 9.791694898315558e-05, "loss": 0.1585, "step": 6596 }, { "epoch": 4.832967032967033, "grad_norm": 1.5390625, "learning_rate": 9.791596536181677e-05, "loss": 0.1564, "step": 6597 }, { "epoch": 4.833699633699633, "grad_norm": 1.7421875, "learning_rate": 9.791498151329251e-05, "loss": 0.1597, "step": 6598 }, { "epoch": 4.834432234432234, "grad_norm": 2.6875, "learning_rate": 9.791399743758752e-05, "loss": 0.2097, "step": 6599 }, { "epoch": 4.835164835164835, "grad_norm": 0.84765625, "learning_rate": 9.79130131347065e-05, "loss": 0.1357, "step": 6600 }, { "epoch": 4.835897435897436, "grad_norm": 1.25, "learning_rate": 9.791202860465415e-05, "loss": 0.1533, "step": 6601 }, { "epoch": 4.836630036630036, "grad_norm": 1.0546875, "learning_rate": 9.791104384743522e-05, "loss": 0.1484, "step": 6602 }, { "epoch": 4.837362637362637, "grad_norm": 0.82421875, "learning_rate": 9.791005886305441e-05, "loss": 0.1541, "step": 6603 }, { "epoch": 4.838095238095238, "grad_norm": 1.546875, "learning_rate": 9.790907365151646e-05, "loss": 0.1392, "step": 6604 }, { "epoch": 4.8388278388278385, "grad_norm": 1.71875, "learning_rate": 9.790808821282605e-05, "loss": 0.1384, "step": 6605 }, { "epoch": 4.839560439560439, "grad_norm": 0.99609375, "learning_rate": 9.790710254698795e-05, "loss": 0.1623, "step": 6606 }, { "epoch": 4.84029304029304, "grad_norm": 0.90625, "learning_rate": 9.790611665400684e-05, "loss": 0.1734, "step": 6607 }, { "epoch": 4.841025641025641, "grad_norm": 0.80078125, "learning_rate": 9.790513053388748e-05, "loss": 0.1456, "step": 6608 }, { "epoch": 4.841758241758241, "grad_norm": 1.1953125, "learning_rate": 9.790414418663456e-05, "loss": 0.1449, "step": 6609 }, { "epoch": 4.842490842490842, "grad_norm": 0.9609375, "learning_rate": 9.790315761225284e-05, "loss": 0.1223, "step": 6610 }, { "epoch": 4.843223443223443, "grad_norm": 1.1484375, "learning_rate": 9.7902170810747e-05, "loss": 0.1904, "step": 6611 }, { "epoch": 4.843956043956044, "grad_norm": 0.96875, "learning_rate": 9.790118378212183e-05, "loss": 0.1488, "step": 6612 }, { "epoch": 4.844688644688644, "grad_norm": 0.6171875, "learning_rate": 9.790019652638202e-05, "loss": 0.1186, "step": 6613 }, { "epoch": 4.845421245421245, "grad_norm": 0.9921875, "learning_rate": 9.789920904353233e-05, "loss": 0.1487, "step": 6614 }, { "epoch": 4.846153846153846, "grad_norm": 1.1015625, "learning_rate": 9.789822133357745e-05, "loss": 0.2059, "step": 6615 }, { "epoch": 4.8468864468864465, "grad_norm": 0.96484375, "learning_rate": 9.789723339652211e-05, "loss": 0.1365, "step": 6616 }, { "epoch": 4.847619047619047, "grad_norm": 1.109375, "learning_rate": 9.789624523237108e-05, "loss": 0.1498, "step": 6617 }, { "epoch": 4.848351648351648, "grad_norm": 0.8671875, "learning_rate": 9.789525684112909e-05, "loss": 0.1388, "step": 6618 }, { "epoch": 4.849084249084249, "grad_norm": 1.4140625, "learning_rate": 9.789426822280086e-05, "loss": 0.1526, "step": 6619 }, { "epoch": 4.8498168498168495, "grad_norm": 1.6796875, "learning_rate": 9.789327937739113e-05, "loss": 0.1526, "step": 6620 }, { "epoch": 4.85054945054945, "grad_norm": 2.390625, "learning_rate": 9.789229030490464e-05, "loss": 0.1729, "step": 6621 }, { "epoch": 4.851282051282051, "grad_norm": 1.28125, "learning_rate": 9.789130100534614e-05, "loss": 0.1299, "step": 6622 }, { "epoch": 4.852014652014652, "grad_norm": 0.65234375, "learning_rate": 9.789031147872035e-05, "loss": 0.1452, "step": 6623 }, { "epoch": 4.852747252747252, "grad_norm": 2.015625, "learning_rate": 9.788932172503202e-05, "loss": 0.186, "step": 6624 }, { "epoch": 4.853479853479853, "grad_norm": 1.5859375, "learning_rate": 9.788833174428587e-05, "loss": 0.1478, "step": 6625 }, { "epoch": 4.854212454212454, "grad_norm": 1.625, "learning_rate": 9.788734153648671e-05, "loss": 0.2033, "step": 6626 }, { "epoch": 4.854945054945055, "grad_norm": 1.6484375, "learning_rate": 9.788635110163922e-05, "loss": 0.1346, "step": 6627 }, { "epoch": 4.855677655677655, "grad_norm": 1.484375, "learning_rate": 9.788536043974816e-05, "loss": 0.1402, "step": 6628 }, { "epoch": 4.856410256410256, "grad_norm": 1.2578125, "learning_rate": 9.78843695508183e-05, "loss": 0.1441, "step": 6629 }, { "epoch": 4.857142857142857, "grad_norm": 1.484375, "learning_rate": 9.788337843485435e-05, "loss": 0.1733, "step": 6630 }, { "epoch": 4.8578754578754575, "grad_norm": 2.671875, "learning_rate": 9.78823870918611e-05, "loss": 0.1781, "step": 6631 }, { "epoch": 4.858608058608058, "grad_norm": 1.5703125, "learning_rate": 9.788139552184326e-05, "loss": 0.1914, "step": 6632 }, { "epoch": 4.859340659340659, "grad_norm": 2.546875, "learning_rate": 9.78804037248056e-05, "loss": 0.1516, "step": 6633 }, { "epoch": 4.860073260073261, "grad_norm": 1.734375, "learning_rate": 9.787941170075289e-05, "loss": 0.1448, "step": 6634 }, { "epoch": 4.860805860805861, "grad_norm": 0.7734375, "learning_rate": 9.787841944968984e-05, "loss": 0.1468, "step": 6635 }, { "epoch": 4.861538461538462, "grad_norm": 0.87109375, "learning_rate": 9.787742697162124e-05, "loss": 0.1206, "step": 6636 }, { "epoch": 4.862271062271063, "grad_norm": 2.609375, "learning_rate": 9.787643426655185e-05, "loss": 0.1726, "step": 6637 }, { "epoch": 4.8630036630036635, "grad_norm": 1.8828125, "learning_rate": 9.787544133448638e-05, "loss": 0.1414, "step": 6638 }, { "epoch": 4.863736263736264, "grad_norm": 1.7265625, "learning_rate": 9.787444817542963e-05, "loss": 0.1819, "step": 6639 }, { "epoch": 4.864468864468865, "grad_norm": 0.98046875, "learning_rate": 9.787345478938637e-05, "loss": 0.187, "step": 6640 }, { "epoch": 4.865201465201466, "grad_norm": 0.765625, "learning_rate": 9.78724611763613e-05, "loss": 0.1453, "step": 6641 }, { "epoch": 4.865934065934066, "grad_norm": 1.203125, "learning_rate": 9.787146733635925e-05, "loss": 0.12, "step": 6642 }, { "epoch": 4.866666666666667, "grad_norm": 0.796875, "learning_rate": 9.787047326938493e-05, "loss": 0.1713, "step": 6643 }, { "epoch": 4.867399267399268, "grad_norm": 1.1328125, "learning_rate": 9.786947897544312e-05, "loss": 0.1511, "step": 6644 }, { "epoch": 4.868131868131869, "grad_norm": 0.7109375, "learning_rate": 9.78684844545386e-05, "loss": 0.1557, "step": 6645 }, { "epoch": 4.868864468864469, "grad_norm": 1.0625, "learning_rate": 9.786748970667612e-05, "loss": 0.1685, "step": 6646 }, { "epoch": 4.86959706959707, "grad_norm": 1.265625, "learning_rate": 9.786649473186043e-05, "loss": 0.1261, "step": 6647 }, { "epoch": 4.870329670329671, "grad_norm": 0.87109375, "learning_rate": 9.786549953009632e-05, "loss": 0.1196, "step": 6648 }, { "epoch": 4.8710622710622715, "grad_norm": 1.578125, "learning_rate": 9.786450410138857e-05, "loss": 0.1832, "step": 6649 }, { "epoch": 4.871794871794872, "grad_norm": 1.7109375, "learning_rate": 9.786350844574191e-05, "loss": 0.1656, "step": 6650 }, { "epoch": 4.872527472527473, "grad_norm": 0.875, "learning_rate": 9.786251256316115e-05, "loss": 0.151, "step": 6651 }, { "epoch": 4.873260073260074, "grad_norm": 1.015625, "learning_rate": 9.786151645365103e-05, "loss": 0.1508, "step": 6652 }, { "epoch": 4.873992673992674, "grad_norm": 0.640625, "learning_rate": 9.786052011721636e-05, "loss": 0.1239, "step": 6653 }, { "epoch": 4.874725274725275, "grad_norm": 0.640625, "learning_rate": 9.785952355386187e-05, "loss": 0.1522, "step": 6654 }, { "epoch": 4.875457875457876, "grad_norm": 0.62890625, "learning_rate": 9.785852676359235e-05, "loss": 0.1278, "step": 6655 }, { "epoch": 4.876190476190477, "grad_norm": 1.0078125, "learning_rate": 9.785752974641258e-05, "loss": 0.1883, "step": 6656 }, { "epoch": 4.876923076923077, "grad_norm": 0.7265625, "learning_rate": 9.785653250232735e-05, "loss": 0.1594, "step": 6657 }, { "epoch": 4.877655677655678, "grad_norm": 0.96875, "learning_rate": 9.785553503134143e-05, "loss": 0.1334, "step": 6658 }, { "epoch": 4.878388278388279, "grad_norm": 0.90625, "learning_rate": 9.78545373334596e-05, "loss": 0.1551, "step": 6659 }, { "epoch": 4.8791208791208796, "grad_norm": 0.79296875, "learning_rate": 9.785353940868662e-05, "loss": 0.1465, "step": 6660 }, { "epoch": 4.87985347985348, "grad_norm": 0.74609375, "learning_rate": 9.785254125702731e-05, "loss": 0.1599, "step": 6661 }, { "epoch": 4.880586080586081, "grad_norm": 0.88671875, "learning_rate": 9.785154287848641e-05, "loss": 0.1218, "step": 6662 }, { "epoch": 4.881318681318682, "grad_norm": 0.78125, "learning_rate": 9.785054427306873e-05, "loss": 0.1154, "step": 6663 }, { "epoch": 4.8820512820512825, "grad_norm": 1.1796875, "learning_rate": 9.784954544077904e-05, "loss": 0.2112, "step": 6664 }, { "epoch": 4.882783882783883, "grad_norm": 0.98828125, "learning_rate": 9.784854638162216e-05, "loss": 0.1706, "step": 6665 }, { "epoch": 4.883516483516484, "grad_norm": 0.61328125, "learning_rate": 9.784754709560282e-05, "loss": 0.1361, "step": 6666 }, { "epoch": 4.884249084249085, "grad_norm": 0.79296875, "learning_rate": 9.784654758272585e-05, "loss": 0.1199, "step": 6667 }, { "epoch": 4.884981684981685, "grad_norm": 0.77734375, "learning_rate": 9.784554784299603e-05, "loss": 0.1627, "step": 6668 }, { "epoch": 4.885714285714286, "grad_norm": 1.1484375, "learning_rate": 9.784454787641815e-05, "loss": 0.1616, "step": 6669 }, { "epoch": 4.886446886446887, "grad_norm": 0.6328125, "learning_rate": 9.7843547682997e-05, "loss": 0.1489, "step": 6670 }, { "epoch": 4.887179487179488, "grad_norm": 0.9453125, "learning_rate": 9.784254726273735e-05, "loss": 0.1613, "step": 6671 }, { "epoch": 4.887912087912088, "grad_norm": 1.4453125, "learning_rate": 9.784154661564405e-05, "loss": 0.158, "step": 6672 }, { "epoch": 4.888644688644689, "grad_norm": 0.97265625, "learning_rate": 9.784054574172182e-05, "loss": 0.1811, "step": 6673 }, { "epoch": 4.88937728937729, "grad_norm": 1.3359375, "learning_rate": 9.783954464097552e-05, "loss": 0.1985, "step": 6674 }, { "epoch": 4.8901098901098905, "grad_norm": 1.078125, "learning_rate": 9.783854331340992e-05, "loss": 0.1894, "step": 6675 }, { "epoch": 4.890842490842491, "grad_norm": 0.89453125, "learning_rate": 9.783754175902981e-05, "loss": 0.1665, "step": 6676 }, { "epoch": 4.891575091575092, "grad_norm": 1.203125, "learning_rate": 9.783653997784002e-05, "loss": 0.1136, "step": 6677 }, { "epoch": 4.892307692307693, "grad_norm": 1.7109375, "learning_rate": 9.783553796984531e-05, "loss": 0.1401, "step": 6678 }, { "epoch": 4.893040293040293, "grad_norm": 1.21875, "learning_rate": 9.783453573505048e-05, "loss": 0.1435, "step": 6679 }, { "epoch": 4.893772893772894, "grad_norm": 1.1640625, "learning_rate": 9.783353327346038e-05, "loss": 0.1563, "step": 6680 }, { "epoch": 4.894505494505495, "grad_norm": 0.88671875, "learning_rate": 9.783253058507976e-05, "loss": 0.1759, "step": 6681 }, { "epoch": 4.895238095238096, "grad_norm": 0.9921875, "learning_rate": 9.783152766991347e-05, "loss": 0.18, "step": 6682 }, { "epoch": 4.895970695970696, "grad_norm": 1.390625, "learning_rate": 9.783052452796629e-05, "loss": 0.1735, "step": 6683 }, { "epoch": 4.896703296703297, "grad_norm": 0.89453125, "learning_rate": 9.782952115924301e-05, "loss": 0.1359, "step": 6684 }, { "epoch": 4.897435897435898, "grad_norm": 1.7109375, "learning_rate": 9.782851756374846e-05, "loss": 0.1806, "step": 6685 }, { "epoch": 4.8981684981684985, "grad_norm": 0.796875, "learning_rate": 9.782751374148746e-05, "loss": 0.1433, "step": 6686 }, { "epoch": 4.898901098901099, "grad_norm": 1.109375, "learning_rate": 9.782650969246481e-05, "loss": 0.1498, "step": 6687 }, { "epoch": 4.8996336996337, "grad_norm": 1.0390625, "learning_rate": 9.782550541668529e-05, "loss": 0.1451, "step": 6688 }, { "epoch": 4.900366300366301, "grad_norm": 0.85546875, "learning_rate": 9.782450091415375e-05, "loss": 0.1188, "step": 6689 }, { "epoch": 4.9010989010989015, "grad_norm": 0.9609375, "learning_rate": 9.7823496184875e-05, "loss": 0.1485, "step": 6690 }, { "epoch": 4.901831501831502, "grad_norm": 1.0390625, "learning_rate": 9.782249122885383e-05, "loss": 0.1726, "step": 6691 }, { "epoch": 4.902564102564103, "grad_norm": 0.98828125, "learning_rate": 9.782148604609509e-05, "loss": 0.1389, "step": 6692 }, { "epoch": 4.903296703296704, "grad_norm": 1.3984375, "learning_rate": 9.782048063660356e-05, "loss": 0.1486, "step": 6693 }, { "epoch": 4.904029304029304, "grad_norm": 1.140625, "learning_rate": 9.781947500038407e-05, "loss": 0.1536, "step": 6694 }, { "epoch": 4.904761904761905, "grad_norm": 1.09375, "learning_rate": 9.781846913744145e-05, "loss": 0.1573, "step": 6695 }, { "epoch": 4.905494505494506, "grad_norm": 1.6328125, "learning_rate": 9.781746304778051e-05, "loss": 0.1766, "step": 6696 }, { "epoch": 4.906227106227107, "grad_norm": 1.0234375, "learning_rate": 9.781645673140606e-05, "loss": 0.1413, "step": 6697 }, { "epoch": 4.906959706959707, "grad_norm": 1.171875, "learning_rate": 9.781545018832294e-05, "loss": 0.1342, "step": 6698 }, { "epoch": 4.907692307692308, "grad_norm": 0.76171875, "learning_rate": 9.781444341853597e-05, "loss": 0.1681, "step": 6699 }, { "epoch": 4.908424908424909, "grad_norm": 0.953125, "learning_rate": 9.781343642204998e-05, "loss": 0.1575, "step": 6700 }, { "epoch": 4.9091575091575095, "grad_norm": 1.125, "learning_rate": 9.781242919886975e-05, "loss": 0.1394, "step": 6701 }, { "epoch": 4.90989010989011, "grad_norm": 1.21875, "learning_rate": 9.781142174900016e-05, "loss": 0.1353, "step": 6702 }, { "epoch": 4.910622710622711, "grad_norm": 1.1015625, "learning_rate": 9.781041407244604e-05, "loss": 0.13, "step": 6703 }, { "epoch": 4.911355311355312, "grad_norm": 0.5390625, "learning_rate": 9.780940616921215e-05, "loss": 0.1317, "step": 6704 }, { "epoch": 4.912087912087912, "grad_norm": 0.61328125, "learning_rate": 9.780839803930339e-05, "loss": 0.1533, "step": 6705 }, { "epoch": 4.912820512820513, "grad_norm": 0.58984375, "learning_rate": 9.780738968272457e-05, "loss": 0.1119, "step": 6706 }, { "epoch": 4.913553113553114, "grad_norm": 0.59765625, "learning_rate": 9.780638109948051e-05, "loss": 0.1347, "step": 6707 }, { "epoch": 4.914285714285715, "grad_norm": 1.09375, "learning_rate": 9.780537228957604e-05, "loss": 0.1741, "step": 6708 }, { "epoch": 4.915018315018315, "grad_norm": 0.94140625, "learning_rate": 9.780436325301602e-05, "loss": 0.1496, "step": 6709 }, { "epoch": 4.915750915750916, "grad_norm": 1.234375, "learning_rate": 9.780335398980524e-05, "loss": 0.1653, "step": 6710 }, { "epoch": 4.916483516483517, "grad_norm": 0.94140625, "learning_rate": 9.780234449994858e-05, "loss": 0.1337, "step": 6711 }, { "epoch": 4.9172161172161175, "grad_norm": 0.90234375, "learning_rate": 9.780133478345086e-05, "loss": 0.1445, "step": 6712 }, { "epoch": 4.917948717948718, "grad_norm": 0.625, "learning_rate": 9.780032484031691e-05, "loss": 0.1384, "step": 6713 }, { "epoch": 4.918681318681319, "grad_norm": 1.1484375, "learning_rate": 9.77993146705516e-05, "loss": 0.1621, "step": 6714 }, { "epoch": 4.91941391941392, "grad_norm": 1.8671875, "learning_rate": 9.77983042741597e-05, "loss": 0.1125, "step": 6715 }, { "epoch": 4.92014652014652, "grad_norm": 1.4921875, "learning_rate": 9.779729365114612e-05, "loss": 0.1255, "step": 6716 }, { "epoch": 4.920879120879121, "grad_norm": 0.8359375, "learning_rate": 9.77962828015157e-05, "loss": 0.1165, "step": 6717 }, { "epoch": 4.921611721611722, "grad_norm": 0.8828125, "learning_rate": 9.779527172527324e-05, "loss": 0.139, "step": 6718 }, { "epoch": 4.922344322344323, "grad_norm": 1.0390625, "learning_rate": 9.779426042242361e-05, "loss": 0.1613, "step": 6719 }, { "epoch": 4.923076923076923, "grad_norm": 0.88671875, "learning_rate": 9.779324889297165e-05, "loss": 0.144, "step": 6720 }, { "epoch": 4.923809523809524, "grad_norm": 1.1875, "learning_rate": 9.779223713692223e-05, "loss": 0.1587, "step": 6721 }, { "epoch": 4.924542124542125, "grad_norm": 1.2421875, "learning_rate": 9.779122515428015e-05, "loss": 0.1282, "step": 6722 }, { "epoch": 4.9252747252747255, "grad_norm": 1.1015625, "learning_rate": 9.77902129450503e-05, "loss": 0.1419, "step": 6723 }, { "epoch": 4.926007326007326, "grad_norm": 0.671875, "learning_rate": 9.77892005092375e-05, "loss": 0.1455, "step": 6724 }, { "epoch": 4.926739926739927, "grad_norm": 0.890625, "learning_rate": 9.778818784684663e-05, "loss": 0.1662, "step": 6725 }, { "epoch": 4.927472527472528, "grad_norm": 0.921875, "learning_rate": 9.778717495788254e-05, "loss": 0.1706, "step": 6726 }, { "epoch": 4.9282051282051285, "grad_norm": 0.86328125, "learning_rate": 9.778616184235006e-05, "loss": 0.1684, "step": 6727 }, { "epoch": 4.928937728937729, "grad_norm": 1.109375, "learning_rate": 9.778514850025404e-05, "loss": 0.1882, "step": 6728 }, { "epoch": 4.92967032967033, "grad_norm": 0.890625, "learning_rate": 9.778413493159935e-05, "loss": 0.1229, "step": 6729 }, { "epoch": 4.930402930402931, "grad_norm": 1.2890625, "learning_rate": 9.778312113639087e-05, "loss": 0.1697, "step": 6730 }, { "epoch": 4.931135531135531, "grad_norm": 1.2734375, "learning_rate": 9.778210711463345e-05, "loss": 0.1909, "step": 6731 }, { "epoch": 4.931868131868132, "grad_norm": 0.8828125, "learning_rate": 9.77810928663319e-05, "loss": 0.1633, "step": 6732 }, { "epoch": 4.932600732600733, "grad_norm": 0.8203125, "learning_rate": 9.778007839149112e-05, "loss": 0.1656, "step": 6733 }, { "epoch": 4.933333333333334, "grad_norm": 0.95703125, "learning_rate": 9.777906369011597e-05, "loss": 0.1825, "step": 6734 }, { "epoch": 4.934065934065934, "grad_norm": 1.4765625, "learning_rate": 9.77780487622113e-05, "loss": 0.1982, "step": 6735 }, { "epoch": 4.934798534798535, "grad_norm": 1.2890625, "learning_rate": 9.777703360778197e-05, "loss": 0.1576, "step": 6736 }, { "epoch": 4.935531135531136, "grad_norm": 1.234375, "learning_rate": 9.777601822683285e-05, "loss": 0.1489, "step": 6737 }, { "epoch": 4.9362637362637365, "grad_norm": 0.9296875, "learning_rate": 9.777500261936882e-05, "loss": 0.1434, "step": 6738 }, { "epoch": 4.936996336996337, "grad_norm": 0.87109375, "learning_rate": 9.777398678539473e-05, "loss": 0.1625, "step": 6739 }, { "epoch": 4.937728937728938, "grad_norm": 0.69140625, "learning_rate": 9.777297072491546e-05, "loss": 0.145, "step": 6740 }, { "epoch": 4.938461538461539, "grad_norm": 0.77734375, "learning_rate": 9.777195443793587e-05, "loss": 0.1631, "step": 6741 }, { "epoch": 4.939194139194139, "grad_norm": 0.74609375, "learning_rate": 9.777093792446082e-05, "loss": 0.1476, "step": 6742 }, { "epoch": 4.93992673992674, "grad_norm": 1.125, "learning_rate": 9.776992118449517e-05, "loss": 0.1622, "step": 6743 }, { "epoch": 4.940659340659341, "grad_norm": 0.80859375, "learning_rate": 9.776890421804383e-05, "loss": 0.1464, "step": 6744 }, { "epoch": 4.941391941391942, "grad_norm": 0.66015625, "learning_rate": 9.776788702511165e-05, "loss": 0.1121, "step": 6745 }, { "epoch": 4.942124542124542, "grad_norm": 1.0078125, "learning_rate": 9.77668696057035e-05, "loss": 0.1598, "step": 6746 }, { "epoch": 4.942857142857143, "grad_norm": 0.8125, "learning_rate": 9.776585195982427e-05, "loss": 0.1581, "step": 6747 }, { "epoch": 4.943589743589744, "grad_norm": 0.80078125, "learning_rate": 9.776483408747883e-05, "loss": 0.1238, "step": 6748 }, { "epoch": 4.9443223443223445, "grad_norm": 0.81640625, "learning_rate": 9.776381598867204e-05, "loss": 0.1402, "step": 6749 }, { "epoch": 4.945054945054945, "grad_norm": 1.2890625, "learning_rate": 9.776279766340878e-05, "loss": 0.1537, "step": 6750 }, { "epoch": 4.945787545787546, "grad_norm": 1.3359375, "learning_rate": 9.776177911169397e-05, "loss": 0.1341, "step": 6751 }, { "epoch": 4.946520146520147, "grad_norm": 1.140625, "learning_rate": 9.776076033353245e-05, "loss": 0.1266, "step": 6752 }, { "epoch": 4.947252747252747, "grad_norm": 1.609375, "learning_rate": 9.775974132892911e-05, "loss": 0.1642, "step": 6753 }, { "epoch": 4.947985347985348, "grad_norm": 0.7421875, "learning_rate": 9.775872209788886e-05, "loss": 0.1438, "step": 6754 }, { "epoch": 4.948717948717949, "grad_norm": 1.3828125, "learning_rate": 9.775770264041652e-05, "loss": 0.1819, "step": 6755 }, { "epoch": 4.94945054945055, "grad_norm": 1.234375, "learning_rate": 9.775668295651703e-05, "loss": 0.155, "step": 6756 }, { "epoch": 4.95018315018315, "grad_norm": 1.40625, "learning_rate": 9.775566304619526e-05, "loss": 0.1659, "step": 6757 }, { "epoch": 4.950915750915751, "grad_norm": 1.3515625, "learning_rate": 9.775464290945609e-05, "loss": 0.1456, "step": 6758 }, { "epoch": 4.951648351648352, "grad_norm": 1.1640625, "learning_rate": 9.775362254630441e-05, "loss": 0.1722, "step": 6759 }, { "epoch": 4.9523809523809526, "grad_norm": 1.3203125, "learning_rate": 9.775260195674513e-05, "loss": 0.1452, "step": 6760 }, { "epoch": 4.953113553113553, "grad_norm": 1.9140625, "learning_rate": 9.77515811407831e-05, "loss": 0.1671, "step": 6761 }, { "epoch": 4.953846153846154, "grad_norm": 0.95703125, "learning_rate": 9.775056009842325e-05, "loss": 0.1353, "step": 6762 }, { "epoch": 4.954578754578755, "grad_norm": 1.6796875, "learning_rate": 9.774953882967045e-05, "loss": 0.1762, "step": 6763 }, { "epoch": 4.9553113553113555, "grad_norm": 1.546875, "learning_rate": 9.77485173345296e-05, "loss": 0.1571, "step": 6764 }, { "epoch": 4.956043956043956, "grad_norm": 0.72265625, "learning_rate": 9.774749561300558e-05, "loss": 0.1126, "step": 6765 }, { "epoch": 4.956776556776557, "grad_norm": 1.5390625, "learning_rate": 9.774647366510331e-05, "loss": 0.1659, "step": 6766 }, { "epoch": 4.957509157509158, "grad_norm": 1.046875, "learning_rate": 9.774545149082768e-05, "loss": 0.1382, "step": 6767 }, { "epoch": 4.958241758241758, "grad_norm": 1.046875, "learning_rate": 9.774442909018357e-05, "loss": 0.157, "step": 6768 }, { "epoch": 4.958974358974359, "grad_norm": 0.92578125, "learning_rate": 9.774340646317589e-05, "loss": 0.1417, "step": 6769 }, { "epoch": 4.95970695970696, "grad_norm": 0.87890625, "learning_rate": 9.774238360980957e-05, "loss": 0.1399, "step": 6770 }, { "epoch": 4.960439560439561, "grad_norm": 0.921875, "learning_rate": 9.774136053008944e-05, "loss": 0.1705, "step": 6771 }, { "epoch": 4.961172161172161, "grad_norm": 0.90625, "learning_rate": 9.774033722402046e-05, "loss": 0.1291, "step": 6772 }, { "epoch": 4.961904761904762, "grad_norm": 0.96875, "learning_rate": 9.773931369160751e-05, "loss": 0.1301, "step": 6773 }, { "epoch": 4.962637362637363, "grad_norm": 0.94921875, "learning_rate": 9.773828993285552e-05, "loss": 0.1625, "step": 6774 }, { "epoch": 4.9633699633699635, "grad_norm": 1.0078125, "learning_rate": 9.773726594776936e-05, "loss": 0.1503, "step": 6775 }, { "epoch": 4.964102564102564, "grad_norm": 0.94140625, "learning_rate": 9.773624173635396e-05, "loss": 0.1641, "step": 6776 }, { "epoch": 4.964835164835165, "grad_norm": 0.68359375, "learning_rate": 9.773521729861421e-05, "loss": 0.1252, "step": 6777 }, { "epoch": 4.965567765567766, "grad_norm": 1.0703125, "learning_rate": 9.773419263455503e-05, "loss": 0.1532, "step": 6778 }, { "epoch": 4.966300366300366, "grad_norm": 1.1015625, "learning_rate": 9.773316774418133e-05, "loss": 0.1689, "step": 6779 }, { "epoch": 4.967032967032967, "grad_norm": 1.4921875, "learning_rate": 9.773214262749802e-05, "loss": 0.1655, "step": 6780 }, { "epoch": 4.967765567765568, "grad_norm": 1.375, "learning_rate": 9.773111728451e-05, "loss": 0.1336, "step": 6781 }, { "epoch": 4.968498168498169, "grad_norm": 0.8125, "learning_rate": 9.773009171522221e-05, "loss": 0.129, "step": 6782 }, { "epoch": 4.969230769230769, "grad_norm": 0.9609375, "learning_rate": 9.772906591963952e-05, "loss": 0.1243, "step": 6783 }, { "epoch": 4.96996336996337, "grad_norm": 1.1171875, "learning_rate": 9.77280398977669e-05, "loss": 0.1679, "step": 6784 }, { "epoch": 4.970695970695971, "grad_norm": 1.625, "learning_rate": 9.772701364960923e-05, "loss": 0.1614, "step": 6785 }, { "epoch": 4.9714285714285715, "grad_norm": 1.328125, "learning_rate": 9.77259871751714e-05, "loss": 0.1448, "step": 6786 }, { "epoch": 4.972161172161172, "grad_norm": 0.828125, "learning_rate": 9.77249604744584e-05, "loss": 0.159, "step": 6787 }, { "epoch": 4.972893772893773, "grad_norm": 1.2734375, "learning_rate": 9.77239335474751e-05, "loss": 0.157, "step": 6788 }, { "epoch": 4.973626373626374, "grad_norm": 1.1796875, "learning_rate": 9.772290639422644e-05, "loss": 0.137, "step": 6789 }, { "epoch": 4.9743589743589745, "grad_norm": 1.3046875, "learning_rate": 9.772187901471732e-05, "loss": 0.164, "step": 6790 }, { "epoch": 4.975091575091575, "grad_norm": 1.6171875, "learning_rate": 9.772085140895268e-05, "loss": 0.1508, "step": 6791 }, { "epoch": 4.975824175824176, "grad_norm": 1.28125, "learning_rate": 9.771982357693745e-05, "loss": 0.1793, "step": 6792 }, { "epoch": 4.976556776556777, "grad_norm": 1.0234375, "learning_rate": 9.771879551867652e-05, "loss": 0.1839, "step": 6793 }, { "epoch": 4.977289377289377, "grad_norm": 0.8671875, "learning_rate": 9.771776723417485e-05, "loss": 0.1487, "step": 6794 }, { "epoch": 4.978021978021978, "grad_norm": 1.5078125, "learning_rate": 9.771673872343736e-05, "loss": 0.183, "step": 6795 }, { "epoch": 4.978754578754579, "grad_norm": 1.2265625, "learning_rate": 9.771570998646899e-05, "loss": 0.1596, "step": 6796 }, { "epoch": 4.97948717948718, "grad_norm": 0.890625, "learning_rate": 9.771468102327462e-05, "loss": 0.144, "step": 6797 }, { "epoch": 4.98021978021978, "grad_norm": 1.0234375, "learning_rate": 9.771365183385923e-05, "loss": 0.1783, "step": 6798 }, { "epoch": 4.980952380952381, "grad_norm": 1.328125, "learning_rate": 9.771262241822775e-05, "loss": 0.1807, "step": 6799 }, { "epoch": 4.981684981684982, "grad_norm": 1.3984375, "learning_rate": 9.771159277638508e-05, "loss": 0.1917, "step": 6800 }, { "epoch": 4.9824175824175825, "grad_norm": 1.328125, "learning_rate": 9.771056290833618e-05, "loss": 0.166, "step": 6801 }, { "epoch": 4.983150183150183, "grad_norm": 0.9375, "learning_rate": 9.770953281408595e-05, "loss": 0.1303, "step": 6802 }, { "epoch": 4.983882783882784, "grad_norm": 1.390625, "learning_rate": 9.770850249363938e-05, "loss": 0.1719, "step": 6803 }, { "epoch": 4.984615384615385, "grad_norm": 1.1015625, "learning_rate": 9.770747194700137e-05, "loss": 0.1334, "step": 6804 }, { "epoch": 4.985347985347985, "grad_norm": 0.96875, "learning_rate": 9.770644117417685e-05, "loss": 0.1634, "step": 6805 }, { "epoch": 4.986080586080586, "grad_norm": 0.60546875, "learning_rate": 9.770541017517079e-05, "loss": 0.1321, "step": 6806 }, { "epoch": 4.986813186813187, "grad_norm": 1.703125, "learning_rate": 9.770437894998811e-05, "loss": 0.1509, "step": 6807 }, { "epoch": 4.987545787545788, "grad_norm": 1.3671875, "learning_rate": 9.770334749863376e-05, "loss": 0.1384, "step": 6808 }, { "epoch": 4.988278388278388, "grad_norm": 1.0625, "learning_rate": 9.770231582111268e-05, "loss": 0.1118, "step": 6809 }, { "epoch": 4.989010989010989, "grad_norm": 1.0390625, "learning_rate": 9.770128391742979e-05, "loss": 0.1455, "step": 6810 }, { "epoch": 4.98974358974359, "grad_norm": 1.359375, "learning_rate": 9.770025178759006e-05, "loss": 0.1949, "step": 6811 }, { "epoch": 4.9904761904761905, "grad_norm": 0.69140625, "learning_rate": 9.769921943159844e-05, "loss": 0.1522, "step": 6812 }, { "epoch": 4.991208791208791, "grad_norm": 1.1875, "learning_rate": 9.769818684945984e-05, "loss": 0.1484, "step": 6813 }, { "epoch": 4.991941391941392, "grad_norm": 0.5703125, "learning_rate": 9.769715404117925e-05, "loss": 0.1307, "step": 6814 }, { "epoch": 4.992673992673993, "grad_norm": 1.0078125, "learning_rate": 9.76961210067616e-05, "loss": 0.1411, "step": 6815 }, { "epoch": 4.993406593406593, "grad_norm": 1.1875, "learning_rate": 9.769508774621185e-05, "loss": 0.1617, "step": 6816 }, { "epoch": 4.994139194139194, "grad_norm": 0.796875, "learning_rate": 9.769405425953492e-05, "loss": 0.1367, "step": 6817 }, { "epoch": 4.994871794871795, "grad_norm": 1.0234375, "learning_rate": 9.76930205467358e-05, "loss": 0.1277, "step": 6818 }, { "epoch": 4.995604395604396, "grad_norm": 0.9296875, "learning_rate": 9.769198660781943e-05, "loss": 0.1408, "step": 6819 }, { "epoch": 4.996336996336996, "grad_norm": 1.015625, "learning_rate": 9.769095244279074e-05, "loss": 0.1201, "step": 6820 }, { "epoch": 4.997069597069597, "grad_norm": 0.94921875, "learning_rate": 9.768991805165472e-05, "loss": 0.1297, "step": 6821 }, { "epoch": 4.997802197802198, "grad_norm": 0.80859375, "learning_rate": 9.76888834344163e-05, "loss": 0.1632, "step": 6822 }, { "epoch": 4.9985347985347985, "grad_norm": 0.5859375, "learning_rate": 9.768784859108046e-05, "loss": 0.1259, "step": 6823 }, { "epoch": 4.999267399267399, "grad_norm": 0.9609375, "learning_rate": 9.768681352165216e-05, "loss": 0.1377, "step": 6824 }, { "epoch": 5.0, "grad_norm": 0.77734375, "learning_rate": 9.768577822613632e-05, "loss": 0.1357, "step": 6825 }, { "epoch": 5.000732600732601, "grad_norm": 0.80078125, "learning_rate": 9.768474270453792e-05, "loss": 0.1389, "step": 6826 }, { "epoch": 5.0014652014652015, "grad_norm": 0.7890625, "learning_rate": 9.768370695686195e-05, "loss": 0.1102, "step": 6827 }, { "epoch": 5.002197802197802, "grad_norm": 1.15625, "learning_rate": 9.768267098311335e-05, "loss": 0.1979, "step": 6828 }, { "epoch": 5.002930402930403, "grad_norm": 1.140625, "learning_rate": 9.768163478329708e-05, "loss": 0.1208, "step": 6829 }, { "epoch": 5.003663003663004, "grad_norm": 1.40625, "learning_rate": 9.76805983574181e-05, "loss": 0.1455, "step": 6830 }, { "epoch": 5.004395604395604, "grad_norm": 0.7890625, "learning_rate": 9.767956170548139e-05, "loss": 0.1532, "step": 6831 }, { "epoch": 5.005128205128205, "grad_norm": 0.474609375, "learning_rate": 9.767852482749193e-05, "loss": 0.1127, "step": 6832 }, { "epoch": 5.005860805860806, "grad_norm": 0.6328125, "learning_rate": 9.767748772345465e-05, "loss": 0.1359, "step": 6833 }, { "epoch": 5.006593406593407, "grad_norm": 1.234375, "learning_rate": 9.767645039337454e-05, "loss": 0.1594, "step": 6834 }, { "epoch": 5.007326007326007, "grad_norm": 0.83984375, "learning_rate": 9.767541283725656e-05, "loss": 0.1618, "step": 6835 }, { "epoch": 5.008058608058608, "grad_norm": 0.703125, "learning_rate": 9.767437505510572e-05, "loss": 0.142, "step": 6836 }, { "epoch": 5.008791208791209, "grad_norm": 0.8515625, "learning_rate": 9.767333704692693e-05, "loss": 0.1638, "step": 6837 }, { "epoch": 5.0095238095238095, "grad_norm": 0.90234375, "learning_rate": 9.767229881272521e-05, "loss": 0.1551, "step": 6838 }, { "epoch": 5.01025641025641, "grad_norm": 0.71484375, "learning_rate": 9.767126035250553e-05, "loss": 0.1442, "step": 6839 }, { "epoch": 5.010989010989011, "grad_norm": 1.53125, "learning_rate": 9.767022166627284e-05, "loss": 0.1582, "step": 6840 }, { "epoch": 5.011721611721612, "grad_norm": 1.03125, "learning_rate": 9.766918275403214e-05, "loss": 0.1533, "step": 6841 }, { "epoch": 5.012454212454212, "grad_norm": 0.7578125, "learning_rate": 9.766814361578841e-05, "loss": 0.1605, "step": 6842 }, { "epoch": 5.013186813186813, "grad_norm": 0.796875, "learning_rate": 9.766710425154662e-05, "loss": 0.156, "step": 6843 }, { "epoch": 5.013919413919414, "grad_norm": 1.03125, "learning_rate": 9.766606466131173e-05, "loss": 0.165, "step": 6844 }, { "epoch": 5.014652014652015, "grad_norm": 0.88671875, "learning_rate": 9.766502484508875e-05, "loss": 0.1493, "step": 6845 }, { "epoch": 5.015384615384615, "grad_norm": 1.09375, "learning_rate": 9.766398480288266e-05, "loss": 0.143, "step": 6846 }, { "epoch": 5.016117216117216, "grad_norm": 1.109375, "learning_rate": 9.766294453469842e-05, "loss": 0.1286, "step": 6847 }, { "epoch": 5.016849816849817, "grad_norm": 1.0546875, "learning_rate": 9.766190404054105e-05, "loss": 0.132, "step": 6848 }, { "epoch": 5.0175824175824175, "grad_norm": 0.71875, "learning_rate": 9.76608633204155e-05, "loss": 0.1374, "step": 6849 }, { "epoch": 5.018315018315018, "grad_norm": 0.9453125, "learning_rate": 9.765982237432678e-05, "loss": 0.1577, "step": 6850 }, { "epoch": 5.019047619047619, "grad_norm": 0.80078125, "learning_rate": 9.765878120227987e-05, "loss": 0.1417, "step": 6851 }, { "epoch": 5.01978021978022, "grad_norm": 0.890625, "learning_rate": 9.765773980427976e-05, "loss": 0.1562, "step": 6852 }, { "epoch": 5.02051282051282, "grad_norm": 1.4140625, "learning_rate": 9.765669818033142e-05, "loss": 0.1547, "step": 6853 }, { "epoch": 5.021245421245421, "grad_norm": 0.9609375, "learning_rate": 9.765565633043988e-05, "loss": 0.143, "step": 6854 }, { "epoch": 5.021978021978022, "grad_norm": 0.99609375, "learning_rate": 9.76546142546101e-05, "loss": 0.1539, "step": 6855 }, { "epoch": 5.022710622710623, "grad_norm": 0.828125, "learning_rate": 9.765357195284708e-05, "loss": 0.1457, "step": 6856 }, { "epoch": 5.023443223443223, "grad_norm": 0.9609375, "learning_rate": 9.765252942515583e-05, "loss": 0.1364, "step": 6857 }, { "epoch": 5.024175824175824, "grad_norm": 0.875, "learning_rate": 9.765148667154133e-05, "loss": 0.1416, "step": 6858 }, { "epoch": 5.024908424908425, "grad_norm": 0.90625, "learning_rate": 9.765044369200856e-05, "loss": 0.1552, "step": 6859 }, { "epoch": 5.0256410256410255, "grad_norm": 0.70703125, "learning_rate": 9.764940048656255e-05, "loss": 0.1524, "step": 6860 }, { "epoch": 5.026373626373626, "grad_norm": 0.94921875, "learning_rate": 9.764835705520829e-05, "loss": 0.1601, "step": 6861 }, { "epoch": 5.027106227106227, "grad_norm": 1.3203125, "learning_rate": 9.764731339795076e-05, "loss": 0.1631, "step": 6862 }, { "epoch": 5.027838827838828, "grad_norm": 1.21875, "learning_rate": 9.764626951479498e-05, "loss": 0.1325, "step": 6863 }, { "epoch": 5.0285714285714285, "grad_norm": 1.1875, "learning_rate": 9.764522540574597e-05, "loss": 0.1802, "step": 6864 }, { "epoch": 5.029304029304029, "grad_norm": 0.88671875, "learning_rate": 9.764418107080867e-05, "loss": 0.1651, "step": 6865 }, { "epoch": 5.03003663003663, "grad_norm": 0.71875, "learning_rate": 9.764313650998814e-05, "loss": 0.1593, "step": 6866 }, { "epoch": 5.030769230769231, "grad_norm": 1.078125, "learning_rate": 9.764209172328936e-05, "loss": 0.2147, "step": 6867 }, { "epoch": 5.031501831501831, "grad_norm": 1.5078125, "learning_rate": 9.764104671071735e-05, "loss": 0.1669, "step": 6868 }, { "epoch": 5.032234432234432, "grad_norm": 0.953125, "learning_rate": 9.764000147227711e-05, "loss": 0.1417, "step": 6869 }, { "epoch": 5.032967032967033, "grad_norm": 1.03125, "learning_rate": 9.763895600797363e-05, "loss": 0.1436, "step": 6870 }, { "epoch": 5.033699633699634, "grad_norm": 0.703125, "learning_rate": 9.763791031781196e-05, "loss": 0.1214, "step": 6871 }, { "epoch": 5.034432234432234, "grad_norm": 1.3828125, "learning_rate": 9.763686440179707e-05, "loss": 0.1677, "step": 6872 }, { "epoch": 5.035164835164835, "grad_norm": 1.734375, "learning_rate": 9.7635818259934e-05, "loss": 0.1708, "step": 6873 }, { "epoch": 5.035897435897436, "grad_norm": 1.1484375, "learning_rate": 9.763477189222776e-05, "loss": 0.1514, "step": 6874 }, { "epoch": 5.0366300366300365, "grad_norm": 1.21875, "learning_rate": 9.763372529868333e-05, "loss": 0.199, "step": 6875 }, { "epoch": 5.037362637362637, "grad_norm": 1.328125, "learning_rate": 9.763267847930577e-05, "loss": 0.1617, "step": 6876 }, { "epoch": 5.038095238095238, "grad_norm": 0.76953125, "learning_rate": 9.763163143410006e-05, "loss": 0.1424, "step": 6877 }, { "epoch": 5.038827838827839, "grad_norm": 1.484375, "learning_rate": 9.763058416307124e-05, "loss": 0.1674, "step": 6878 }, { "epoch": 5.039560439560439, "grad_norm": 0.609375, "learning_rate": 9.762953666622431e-05, "loss": 0.1404, "step": 6879 }, { "epoch": 5.04029304029304, "grad_norm": 0.82421875, "learning_rate": 9.762848894356433e-05, "loss": 0.1785, "step": 6880 }, { "epoch": 5.041025641025641, "grad_norm": 0.97265625, "learning_rate": 9.762744099509625e-05, "loss": 0.1623, "step": 6881 }, { "epoch": 5.041758241758242, "grad_norm": 1.3125, "learning_rate": 9.762639282082515e-05, "loss": 0.1748, "step": 6882 }, { "epoch": 5.042490842490842, "grad_norm": 1.5390625, "learning_rate": 9.762534442075602e-05, "loss": 0.1476, "step": 6883 }, { "epoch": 5.043223443223443, "grad_norm": 0.9375, "learning_rate": 9.762429579489391e-05, "loss": 0.1385, "step": 6884 }, { "epoch": 5.043956043956044, "grad_norm": 1.1640625, "learning_rate": 9.76232469432438e-05, "loss": 0.1406, "step": 6885 }, { "epoch": 5.0446886446886445, "grad_norm": 1.1328125, "learning_rate": 9.762219786581077e-05, "loss": 0.1882, "step": 6886 }, { "epoch": 5.045421245421245, "grad_norm": 1.3828125, "learning_rate": 9.762114856259981e-05, "loss": 0.2042, "step": 6887 }, { "epoch": 5.046153846153846, "grad_norm": 1.171875, "learning_rate": 9.762009903361596e-05, "loss": 0.1564, "step": 6888 }, { "epoch": 5.046886446886447, "grad_norm": 0.98046875, "learning_rate": 9.761904927886426e-05, "loss": 0.1177, "step": 6889 }, { "epoch": 5.0476190476190474, "grad_norm": 1.3046875, "learning_rate": 9.76179992983497e-05, "loss": 0.118, "step": 6890 }, { "epoch": 5.048351648351648, "grad_norm": 0.96875, "learning_rate": 9.761694909207736e-05, "loss": 0.1649, "step": 6891 }, { "epoch": 5.049084249084249, "grad_norm": 0.8984375, "learning_rate": 9.761589866005225e-05, "loss": 0.1575, "step": 6892 }, { "epoch": 5.04981684981685, "grad_norm": 1.140625, "learning_rate": 9.761484800227939e-05, "loss": 0.1567, "step": 6893 }, { "epoch": 5.05054945054945, "grad_norm": 1.4921875, "learning_rate": 9.761379711876383e-05, "loss": 0.1649, "step": 6894 }, { "epoch": 5.051282051282051, "grad_norm": 1.75, "learning_rate": 9.761274600951061e-05, "loss": 0.1444, "step": 6895 }, { "epoch": 5.052014652014652, "grad_norm": 1.15625, "learning_rate": 9.761169467452474e-05, "loss": 0.1341, "step": 6896 }, { "epoch": 5.052747252747253, "grad_norm": 0.67578125, "learning_rate": 9.761064311381128e-05, "loss": 0.1676, "step": 6897 }, { "epoch": 5.053479853479853, "grad_norm": 0.85546875, "learning_rate": 9.760959132737528e-05, "loss": 0.122, "step": 6898 }, { "epoch": 5.054212454212454, "grad_norm": 1.4140625, "learning_rate": 9.760853931522176e-05, "loss": 0.123, "step": 6899 }, { "epoch": 5.054945054945055, "grad_norm": 1.8828125, "learning_rate": 9.760748707735575e-05, "loss": 0.1776, "step": 6900 }, { "epoch": 5.0556776556776555, "grad_norm": 1.3359375, "learning_rate": 9.76064346137823e-05, "loss": 0.1584, "step": 6901 }, { "epoch": 5.056410256410256, "grad_norm": 1.3046875, "learning_rate": 9.760538192450649e-05, "loss": 0.1347, "step": 6902 }, { "epoch": 5.057142857142857, "grad_norm": 1.15625, "learning_rate": 9.760432900953329e-05, "loss": 0.1694, "step": 6903 }, { "epoch": 5.057875457875458, "grad_norm": 1.3828125, "learning_rate": 9.760327586886782e-05, "loss": 0.1389, "step": 6904 }, { "epoch": 5.058608058608058, "grad_norm": 1.03125, "learning_rate": 9.760222250251509e-05, "loss": 0.1506, "step": 6905 }, { "epoch": 5.059340659340659, "grad_norm": 0.91796875, "learning_rate": 9.760116891048012e-05, "loss": 0.1356, "step": 6906 }, { "epoch": 5.06007326007326, "grad_norm": 1.609375, "learning_rate": 9.760011509276801e-05, "loss": 0.1588, "step": 6907 }, { "epoch": 5.060805860805861, "grad_norm": 0.94921875, "learning_rate": 9.759906104938378e-05, "loss": 0.1431, "step": 6908 }, { "epoch": 5.061538461538461, "grad_norm": 1.59375, "learning_rate": 9.759800678033249e-05, "loss": 0.1388, "step": 6909 }, { "epoch": 5.062271062271062, "grad_norm": 0.88671875, "learning_rate": 9.759695228561918e-05, "loss": 0.1273, "step": 6910 }, { "epoch": 5.063003663003663, "grad_norm": 0.9140625, "learning_rate": 9.759589756524892e-05, "loss": 0.1461, "step": 6911 }, { "epoch": 5.0637362637362635, "grad_norm": 0.8515625, "learning_rate": 9.759484261922675e-05, "loss": 0.1634, "step": 6912 }, { "epoch": 5.064468864468864, "grad_norm": 1.0078125, "learning_rate": 9.759378744755773e-05, "loss": 0.1783, "step": 6913 }, { "epoch": 5.065201465201465, "grad_norm": 0.96484375, "learning_rate": 9.759273205024693e-05, "loss": 0.1518, "step": 6914 }, { "epoch": 5.065934065934066, "grad_norm": 0.9765625, "learning_rate": 9.759167642729937e-05, "loss": 0.2056, "step": 6915 }, { "epoch": 5.066666666666666, "grad_norm": 1.390625, "learning_rate": 9.759062057872013e-05, "loss": 0.174, "step": 6916 }, { "epoch": 5.067399267399267, "grad_norm": 1.2421875, "learning_rate": 9.758956450451426e-05, "loss": 0.1788, "step": 6917 }, { "epoch": 5.068131868131868, "grad_norm": 1.1484375, "learning_rate": 9.758850820468684e-05, "loss": 0.1368, "step": 6918 }, { "epoch": 5.068864468864469, "grad_norm": 0.94921875, "learning_rate": 9.758745167924293e-05, "loss": 0.1431, "step": 6919 }, { "epoch": 5.069597069597069, "grad_norm": 1.03125, "learning_rate": 9.758639492818759e-05, "loss": 0.1832, "step": 6920 }, { "epoch": 5.07032967032967, "grad_norm": 0.68359375, "learning_rate": 9.758533795152585e-05, "loss": 0.1405, "step": 6921 }, { "epoch": 5.071062271062271, "grad_norm": 1.0234375, "learning_rate": 9.75842807492628e-05, "loss": 0.1593, "step": 6922 }, { "epoch": 5.0717948717948715, "grad_norm": 0.8359375, "learning_rate": 9.75832233214035e-05, "loss": 0.1229, "step": 6923 }, { "epoch": 5.072527472527472, "grad_norm": 1.3828125, "learning_rate": 9.758216566795303e-05, "loss": 0.1485, "step": 6924 }, { "epoch": 5.073260073260073, "grad_norm": 1.203125, "learning_rate": 9.758110778891646e-05, "loss": 0.1422, "step": 6925 }, { "epoch": 5.073992673992674, "grad_norm": 1.015625, "learning_rate": 9.758004968429883e-05, "loss": 0.1327, "step": 6926 }, { "epoch": 5.0747252747252745, "grad_norm": 0.5859375, "learning_rate": 9.757899135410524e-05, "loss": 0.1201, "step": 6927 }, { "epoch": 5.075457875457875, "grad_norm": 0.671875, "learning_rate": 9.757793279834073e-05, "loss": 0.1378, "step": 6928 }, { "epoch": 5.076190476190476, "grad_norm": 0.76171875, "learning_rate": 9.757687401701039e-05, "loss": 0.1526, "step": 6929 }, { "epoch": 5.076923076923077, "grad_norm": 0.99609375, "learning_rate": 9.757581501011932e-05, "loss": 0.1442, "step": 6930 }, { "epoch": 5.077655677655677, "grad_norm": 0.7578125, "learning_rate": 9.757475577767254e-05, "loss": 0.1361, "step": 6931 }, { "epoch": 5.078388278388278, "grad_norm": 1.234375, "learning_rate": 9.757369631967516e-05, "loss": 0.1527, "step": 6932 }, { "epoch": 5.079120879120879, "grad_norm": 0.93359375, "learning_rate": 9.757263663613225e-05, "loss": 0.1582, "step": 6933 }, { "epoch": 5.07985347985348, "grad_norm": 1.34375, "learning_rate": 9.757157672704888e-05, "loss": 0.1468, "step": 6934 }, { "epoch": 5.08058608058608, "grad_norm": 0.859375, "learning_rate": 9.757051659243013e-05, "loss": 0.1279, "step": 6935 }, { "epoch": 5.081318681318681, "grad_norm": 1.1953125, "learning_rate": 9.756945623228108e-05, "loss": 0.1578, "step": 6936 }, { "epoch": 5.082051282051282, "grad_norm": 0.86328125, "learning_rate": 9.756839564660681e-05, "loss": 0.1671, "step": 6937 }, { "epoch": 5.0827838827838825, "grad_norm": 0.75390625, "learning_rate": 9.756733483541243e-05, "loss": 0.1195, "step": 6938 }, { "epoch": 5.083516483516483, "grad_norm": 0.60546875, "learning_rate": 9.756627379870298e-05, "loss": 0.1418, "step": 6939 }, { "epoch": 5.084249084249084, "grad_norm": 1.140625, "learning_rate": 9.756521253648358e-05, "loss": 0.1665, "step": 6940 }, { "epoch": 5.084981684981685, "grad_norm": 1.109375, "learning_rate": 9.756415104875926e-05, "loss": 0.1823, "step": 6941 }, { "epoch": 5.085714285714285, "grad_norm": 0.8515625, "learning_rate": 9.756308933553516e-05, "loss": 0.1559, "step": 6942 }, { "epoch": 5.086446886446886, "grad_norm": 0.859375, "learning_rate": 9.756202739681635e-05, "loss": 0.1578, "step": 6943 }, { "epoch": 5.087179487179487, "grad_norm": 0.9453125, "learning_rate": 9.756096523260792e-05, "loss": 0.1331, "step": 6944 }, { "epoch": 5.087912087912088, "grad_norm": 1.0390625, "learning_rate": 9.755990284291496e-05, "loss": 0.1413, "step": 6945 }, { "epoch": 5.088644688644688, "grad_norm": 1.4375, "learning_rate": 9.755884022774256e-05, "loss": 0.1422, "step": 6946 }, { "epoch": 5.089377289377289, "grad_norm": 0.87109375, "learning_rate": 9.755777738709579e-05, "loss": 0.1842, "step": 6947 }, { "epoch": 5.09010989010989, "grad_norm": 0.81640625, "learning_rate": 9.755671432097977e-05, "loss": 0.1401, "step": 6948 }, { "epoch": 5.0908424908424905, "grad_norm": 0.82421875, "learning_rate": 9.755565102939959e-05, "loss": 0.1651, "step": 6949 }, { "epoch": 5.091575091575091, "grad_norm": 1.0, "learning_rate": 9.755458751236033e-05, "loss": 0.1404, "step": 6950 }, { "epoch": 5.092307692307692, "grad_norm": 0.6484375, "learning_rate": 9.75535237698671e-05, "loss": 0.1228, "step": 6951 }, { "epoch": 5.093040293040293, "grad_norm": 0.953125, "learning_rate": 9.755245980192499e-05, "loss": 0.1451, "step": 6952 }, { "epoch": 5.093772893772893, "grad_norm": 1.34375, "learning_rate": 9.75513956085391e-05, "loss": 0.1457, "step": 6953 }, { "epoch": 5.094505494505494, "grad_norm": 0.8125, "learning_rate": 9.755033118971451e-05, "loss": 0.1495, "step": 6954 }, { "epoch": 5.095238095238095, "grad_norm": 0.82421875, "learning_rate": 9.754926654545636e-05, "loss": 0.1358, "step": 6955 }, { "epoch": 5.095970695970696, "grad_norm": 0.87109375, "learning_rate": 9.754820167576972e-05, "loss": 0.1464, "step": 6956 }, { "epoch": 5.096703296703296, "grad_norm": 1.15625, "learning_rate": 9.75471365806597e-05, "loss": 0.135, "step": 6957 }, { "epoch": 5.097435897435897, "grad_norm": 1.0234375, "learning_rate": 9.754607126013141e-05, "loss": 0.1637, "step": 6958 }, { "epoch": 5.098168498168498, "grad_norm": 1.03125, "learning_rate": 9.754500571418995e-05, "loss": 0.1561, "step": 6959 }, { "epoch": 5.0989010989010985, "grad_norm": 1.2265625, "learning_rate": 9.754393994284041e-05, "loss": 0.1662, "step": 6960 }, { "epoch": 5.099633699633699, "grad_norm": 1.1484375, "learning_rate": 9.754287394608792e-05, "loss": 0.1719, "step": 6961 }, { "epoch": 5.1003663003663, "grad_norm": 0.95703125, "learning_rate": 9.754180772393759e-05, "loss": 0.1449, "step": 6962 }, { "epoch": 5.101098901098901, "grad_norm": 1.1953125, "learning_rate": 9.754074127639449e-05, "loss": 0.1482, "step": 6963 }, { "epoch": 5.1018315018315015, "grad_norm": 1.5390625, "learning_rate": 9.753967460346378e-05, "loss": 0.1378, "step": 6964 }, { "epoch": 5.102564102564102, "grad_norm": 1.375, "learning_rate": 9.753860770515052e-05, "loss": 0.1448, "step": 6965 }, { "epoch": 5.103296703296703, "grad_norm": 1.359375, "learning_rate": 9.753754058145987e-05, "loss": 0.1513, "step": 6966 }, { "epoch": 5.104029304029304, "grad_norm": 0.96484375, "learning_rate": 9.753647323239692e-05, "loss": 0.1254, "step": 6967 }, { "epoch": 5.104761904761904, "grad_norm": 0.671875, "learning_rate": 9.753540565796678e-05, "loss": 0.1454, "step": 6968 }, { "epoch": 5.105494505494505, "grad_norm": 0.88671875, "learning_rate": 9.753433785817457e-05, "loss": 0.143, "step": 6969 }, { "epoch": 5.106227106227106, "grad_norm": 1.828125, "learning_rate": 9.75332698330254e-05, "loss": 0.2245, "step": 6970 }, { "epoch": 5.106959706959707, "grad_norm": 0.9375, "learning_rate": 9.753220158252442e-05, "loss": 0.1659, "step": 6971 }, { "epoch": 5.107692307692307, "grad_norm": 0.9765625, "learning_rate": 9.753113310667668e-05, "loss": 0.138, "step": 6972 }, { "epoch": 5.108424908424908, "grad_norm": 1.234375, "learning_rate": 9.753006440548738e-05, "loss": 0.1384, "step": 6973 }, { "epoch": 5.109157509157509, "grad_norm": 0.8046875, "learning_rate": 9.75289954789616e-05, "loss": 0.1292, "step": 6974 }, { "epoch": 5.1098901098901095, "grad_norm": 0.91015625, "learning_rate": 9.752792632710446e-05, "loss": 0.1355, "step": 6975 }, { "epoch": 5.11062271062271, "grad_norm": 0.828125, "learning_rate": 9.752685694992107e-05, "loss": 0.1215, "step": 6976 }, { "epoch": 5.111355311355311, "grad_norm": 1.3984375, "learning_rate": 9.752578734741659e-05, "loss": 0.1346, "step": 6977 }, { "epoch": 5.112087912087912, "grad_norm": 0.7578125, "learning_rate": 9.75247175195961e-05, "loss": 0.1612, "step": 6978 }, { "epoch": 5.112820512820512, "grad_norm": 0.94140625, "learning_rate": 9.752364746646477e-05, "loss": 0.1565, "step": 6979 }, { "epoch": 5.113553113553113, "grad_norm": 1.2890625, "learning_rate": 9.752257718802771e-05, "loss": 0.1433, "step": 6980 }, { "epoch": 5.114285714285714, "grad_norm": 1.0, "learning_rate": 9.752150668429006e-05, "loss": 0.1336, "step": 6981 }, { "epoch": 5.115018315018315, "grad_norm": 1.390625, "learning_rate": 9.752043595525691e-05, "loss": 0.1674, "step": 6982 }, { "epoch": 5.115750915750915, "grad_norm": 1.4609375, "learning_rate": 9.751936500093342e-05, "loss": 0.1642, "step": 6983 }, { "epoch": 5.116483516483516, "grad_norm": 1.3046875, "learning_rate": 9.751829382132472e-05, "loss": 0.1227, "step": 6984 }, { "epoch": 5.117216117216117, "grad_norm": 1.125, "learning_rate": 9.751722241643594e-05, "loss": 0.1816, "step": 6985 }, { "epoch": 5.1179487179487175, "grad_norm": 0.89453125, "learning_rate": 9.75161507862722e-05, "loss": 0.1279, "step": 6986 }, { "epoch": 5.118681318681318, "grad_norm": 1.578125, "learning_rate": 9.751507893083866e-05, "loss": 0.1257, "step": 6987 }, { "epoch": 5.119413919413919, "grad_norm": 1.4296875, "learning_rate": 9.751400685014045e-05, "loss": 0.1645, "step": 6988 }, { "epoch": 5.12014652014652, "grad_norm": 1.265625, "learning_rate": 9.751293454418269e-05, "loss": 0.1678, "step": 6989 }, { "epoch": 5.1208791208791204, "grad_norm": 0.79296875, "learning_rate": 9.751186201297053e-05, "loss": 0.1609, "step": 6990 }, { "epoch": 5.121611721611721, "grad_norm": 0.55078125, "learning_rate": 9.75107892565091e-05, "loss": 0.0968, "step": 6991 }, { "epoch": 5.122344322344322, "grad_norm": 1.3203125, "learning_rate": 9.750971627480355e-05, "loss": 0.1364, "step": 6992 }, { "epoch": 5.123076923076923, "grad_norm": 0.9140625, "learning_rate": 9.750864306785903e-05, "loss": 0.1474, "step": 6993 }, { "epoch": 5.123809523809523, "grad_norm": 0.84375, "learning_rate": 9.750756963568065e-05, "loss": 0.1244, "step": 6994 }, { "epoch": 5.124542124542124, "grad_norm": 1.078125, "learning_rate": 9.750649597827359e-05, "loss": 0.1686, "step": 6995 }, { "epoch": 5.125274725274725, "grad_norm": 0.8203125, "learning_rate": 9.750542209564295e-05, "loss": 0.1437, "step": 6996 }, { "epoch": 5.126007326007326, "grad_norm": 1.0078125, "learning_rate": 9.750434798779392e-05, "loss": 0.1245, "step": 6997 }, { "epoch": 5.126739926739926, "grad_norm": 0.99609375, "learning_rate": 9.750327365473162e-05, "loss": 0.1722, "step": 6998 }, { "epoch": 5.127472527472527, "grad_norm": 0.97265625, "learning_rate": 9.750219909646122e-05, "loss": 0.2013, "step": 6999 }, { "epoch": 5.128205128205128, "grad_norm": 0.81640625, "learning_rate": 9.750112431298784e-05, "loss": 0.1427, "step": 7000 }, { "epoch": 5.128937728937729, "grad_norm": 1.1171875, "learning_rate": 9.750004930431666e-05, "loss": 0.132, "step": 7001 }, { "epoch": 5.12967032967033, "grad_norm": 1.3828125, "learning_rate": 9.74989740704528e-05, "loss": 0.1518, "step": 7002 }, { "epoch": 5.130402930402931, "grad_norm": 0.91015625, "learning_rate": 9.749789861140144e-05, "loss": 0.1614, "step": 7003 }, { "epoch": 5.131135531135532, "grad_norm": 0.75, "learning_rate": 9.74968229271677e-05, "loss": 0.1543, "step": 7004 }, { "epoch": 5.131868131868132, "grad_norm": 0.83203125, "learning_rate": 9.749574701775676e-05, "loss": 0.1723, "step": 7005 }, { "epoch": 5.132600732600733, "grad_norm": 0.83203125, "learning_rate": 9.749467088317377e-05, "loss": 0.1712, "step": 7006 }, { "epoch": 5.133333333333334, "grad_norm": 0.9375, "learning_rate": 9.749359452342391e-05, "loss": 0.1351, "step": 7007 }, { "epoch": 5.1340659340659345, "grad_norm": 0.76953125, "learning_rate": 9.749251793851229e-05, "loss": 0.1563, "step": 7008 }, { "epoch": 5.134798534798535, "grad_norm": 1.078125, "learning_rate": 9.749144112844408e-05, "loss": 0.1725, "step": 7009 }, { "epoch": 5.135531135531136, "grad_norm": 0.9453125, "learning_rate": 9.749036409322445e-05, "loss": 0.149, "step": 7010 }, { "epoch": 5.136263736263737, "grad_norm": 1.6015625, "learning_rate": 9.748928683285856e-05, "loss": 0.1292, "step": 7011 }, { "epoch": 5.136996336996337, "grad_norm": 0.85546875, "learning_rate": 9.748820934735159e-05, "loss": 0.1526, "step": 7012 }, { "epoch": 5.137728937728938, "grad_norm": 1.109375, "learning_rate": 9.748713163670868e-05, "loss": 0.1287, "step": 7013 }, { "epoch": 5.138461538461539, "grad_norm": 1.2109375, "learning_rate": 9.748605370093497e-05, "loss": 0.1684, "step": 7014 }, { "epoch": 5.13919413919414, "grad_norm": 1.2421875, "learning_rate": 9.748497554003568e-05, "loss": 0.148, "step": 7015 }, { "epoch": 5.13992673992674, "grad_norm": 0.5859375, "learning_rate": 9.748389715401594e-05, "loss": 0.1448, "step": 7016 }, { "epoch": 5.140659340659341, "grad_norm": 1.2421875, "learning_rate": 9.748281854288092e-05, "loss": 0.1095, "step": 7017 }, { "epoch": 5.141391941391942, "grad_norm": 0.87890625, "learning_rate": 9.748173970663581e-05, "loss": 0.1517, "step": 7018 }, { "epoch": 5.1421245421245425, "grad_norm": 0.98046875, "learning_rate": 9.748066064528575e-05, "loss": 0.1361, "step": 7019 }, { "epoch": 5.142857142857143, "grad_norm": 0.79296875, "learning_rate": 9.747958135883592e-05, "loss": 0.1363, "step": 7020 }, { "epoch": 5.143589743589744, "grad_norm": 0.9140625, "learning_rate": 9.74785018472915e-05, "loss": 0.1382, "step": 7021 }, { "epoch": 5.144322344322345, "grad_norm": 0.65625, "learning_rate": 9.747742211065764e-05, "loss": 0.1377, "step": 7022 }, { "epoch": 5.145054945054945, "grad_norm": 1.28125, "learning_rate": 9.747634214893954e-05, "loss": 0.173, "step": 7023 }, { "epoch": 5.145787545787546, "grad_norm": 1.71875, "learning_rate": 9.747526196214236e-05, "loss": 0.1595, "step": 7024 }, { "epoch": 5.146520146520147, "grad_norm": 1.765625, "learning_rate": 9.74741815502713e-05, "loss": 0.1219, "step": 7025 }, { "epoch": 5.147252747252748, "grad_norm": 1.15625, "learning_rate": 9.747310091333148e-05, "loss": 0.1646, "step": 7026 }, { "epoch": 5.147985347985348, "grad_norm": 0.84765625, "learning_rate": 9.747202005132814e-05, "loss": 0.1463, "step": 7027 }, { "epoch": 5.148717948717949, "grad_norm": 1.6328125, "learning_rate": 9.747093896426642e-05, "loss": 0.1619, "step": 7028 }, { "epoch": 5.14945054945055, "grad_norm": 1.890625, "learning_rate": 9.746985765215153e-05, "loss": 0.1402, "step": 7029 }, { "epoch": 5.1501831501831505, "grad_norm": 0.87109375, "learning_rate": 9.74687761149886e-05, "loss": 0.1328, "step": 7030 }, { "epoch": 5.150915750915751, "grad_norm": 1.578125, "learning_rate": 9.746769435278287e-05, "loss": 0.1548, "step": 7031 }, { "epoch": 5.151648351648352, "grad_norm": 0.91796875, "learning_rate": 9.746661236553948e-05, "loss": 0.1382, "step": 7032 }, { "epoch": 5.152380952380953, "grad_norm": 0.7890625, "learning_rate": 9.746553015326365e-05, "loss": 0.1521, "step": 7033 }, { "epoch": 5.1531135531135535, "grad_norm": 1.6484375, "learning_rate": 9.746444771596054e-05, "loss": 0.1483, "step": 7034 }, { "epoch": 5.153846153846154, "grad_norm": 1.53125, "learning_rate": 9.746336505363534e-05, "loss": 0.1469, "step": 7035 }, { "epoch": 5.154578754578755, "grad_norm": 0.83203125, "learning_rate": 9.746228216629326e-05, "loss": 0.1357, "step": 7036 }, { "epoch": 5.155311355311356, "grad_norm": 1.703125, "learning_rate": 9.746119905393945e-05, "loss": 0.1734, "step": 7037 }, { "epoch": 5.156043956043956, "grad_norm": 0.91015625, "learning_rate": 9.74601157165791e-05, "loss": 0.163, "step": 7038 }, { "epoch": 5.156776556776557, "grad_norm": 1.734375, "learning_rate": 9.745903215421746e-05, "loss": 0.1567, "step": 7039 }, { "epoch": 5.157509157509158, "grad_norm": 0.73046875, "learning_rate": 9.745794836685965e-05, "loss": 0.1068, "step": 7040 }, { "epoch": 5.158241758241759, "grad_norm": 2.03125, "learning_rate": 9.745686435451091e-05, "loss": 0.1443, "step": 7041 }, { "epoch": 5.158974358974359, "grad_norm": 1.2109375, "learning_rate": 9.745578011717643e-05, "loss": 0.1539, "step": 7042 }, { "epoch": 5.15970695970696, "grad_norm": 1.203125, "learning_rate": 9.745469565486137e-05, "loss": 0.1556, "step": 7043 }, { "epoch": 5.160439560439561, "grad_norm": 0.68359375, "learning_rate": 9.745361096757096e-05, "loss": 0.1341, "step": 7044 }, { "epoch": 5.1611721611721615, "grad_norm": 0.85546875, "learning_rate": 9.745252605531038e-05, "loss": 0.1279, "step": 7045 }, { "epoch": 5.161904761904762, "grad_norm": 1.984375, "learning_rate": 9.745144091808485e-05, "loss": 0.1892, "step": 7046 }, { "epoch": 5.162637362637363, "grad_norm": 1.984375, "learning_rate": 9.745035555589953e-05, "loss": 0.1716, "step": 7047 }, { "epoch": 5.163369963369964, "grad_norm": 2.328125, "learning_rate": 9.744926996875966e-05, "loss": 0.1528, "step": 7048 }, { "epoch": 5.164102564102564, "grad_norm": 1.0859375, "learning_rate": 9.744818415667042e-05, "loss": 0.1334, "step": 7049 }, { "epoch": 5.164835164835165, "grad_norm": 1.515625, "learning_rate": 9.744709811963701e-05, "loss": 0.1916, "step": 7050 }, { "epoch": 5.165567765567766, "grad_norm": 1.6015625, "learning_rate": 9.744601185766466e-05, "loss": 0.1663, "step": 7051 }, { "epoch": 5.166300366300367, "grad_norm": 1.4375, "learning_rate": 9.744492537075853e-05, "loss": 0.1317, "step": 7052 }, { "epoch": 5.167032967032967, "grad_norm": 1.921875, "learning_rate": 9.744383865892386e-05, "loss": 0.1473, "step": 7053 }, { "epoch": 5.167765567765568, "grad_norm": 1.65625, "learning_rate": 9.744275172216586e-05, "loss": 0.1951, "step": 7054 }, { "epoch": 5.168498168498169, "grad_norm": 0.765625, "learning_rate": 9.744166456048973e-05, "loss": 0.1262, "step": 7055 }, { "epoch": 5.1692307692307695, "grad_norm": 1.3984375, "learning_rate": 9.744057717390068e-05, "loss": 0.1587, "step": 7056 }, { "epoch": 5.16996336996337, "grad_norm": 1.1875, "learning_rate": 9.743948956240387e-05, "loss": 0.1385, "step": 7057 }, { "epoch": 5.170695970695971, "grad_norm": 1.0390625, "learning_rate": 9.743840172600459e-05, "loss": 0.1522, "step": 7058 }, { "epoch": 5.171428571428572, "grad_norm": 0.8046875, "learning_rate": 9.743731366470803e-05, "loss": 0.1425, "step": 7059 }, { "epoch": 5.172161172161172, "grad_norm": 0.75390625, "learning_rate": 9.743622537851938e-05, "loss": 0.1269, "step": 7060 }, { "epoch": 5.172893772893773, "grad_norm": 0.9140625, "learning_rate": 9.743513686744386e-05, "loss": 0.1675, "step": 7061 }, { "epoch": 5.173626373626374, "grad_norm": 1.203125, "learning_rate": 9.743404813148669e-05, "loss": 0.1646, "step": 7062 }, { "epoch": 5.174358974358975, "grad_norm": 1.3515625, "learning_rate": 9.74329591706531e-05, "loss": 0.145, "step": 7063 }, { "epoch": 5.175091575091575, "grad_norm": 0.953125, "learning_rate": 9.743186998494828e-05, "loss": 0.1504, "step": 7064 }, { "epoch": 5.175824175824176, "grad_norm": 0.80859375, "learning_rate": 9.743078057437747e-05, "loss": 0.1629, "step": 7065 }, { "epoch": 5.176556776556777, "grad_norm": 1.109375, "learning_rate": 9.74296909389459e-05, "loss": 0.1664, "step": 7066 }, { "epoch": 5.1772893772893775, "grad_norm": 0.94921875, "learning_rate": 9.742860107865875e-05, "loss": 0.1244, "step": 7067 }, { "epoch": 5.178021978021978, "grad_norm": 0.796875, "learning_rate": 9.742751099352127e-05, "loss": 0.1661, "step": 7068 }, { "epoch": 5.178754578754579, "grad_norm": 0.65625, "learning_rate": 9.74264206835387e-05, "loss": 0.1443, "step": 7069 }, { "epoch": 5.17948717948718, "grad_norm": 1.0078125, "learning_rate": 9.742533014871624e-05, "loss": 0.1289, "step": 7070 }, { "epoch": 5.1802197802197805, "grad_norm": 0.6015625, "learning_rate": 9.742423938905911e-05, "loss": 0.1534, "step": 7071 }, { "epoch": 5.180952380952381, "grad_norm": 0.78125, "learning_rate": 9.742314840457253e-05, "loss": 0.1637, "step": 7072 }, { "epoch": 5.181684981684982, "grad_norm": 1.0859375, "learning_rate": 9.742205719526177e-05, "loss": 0.1661, "step": 7073 }, { "epoch": 5.182417582417583, "grad_norm": 1.03125, "learning_rate": 9.7420965761132e-05, "loss": 0.1198, "step": 7074 }, { "epoch": 5.183150183150183, "grad_norm": 0.828125, "learning_rate": 9.74198741021885e-05, "loss": 0.1412, "step": 7075 }, { "epoch": 5.183882783882784, "grad_norm": 1.46875, "learning_rate": 9.741878221843648e-05, "loss": 0.1508, "step": 7076 }, { "epoch": 5.184615384615385, "grad_norm": 1.234375, "learning_rate": 9.741769010988117e-05, "loss": 0.1475, "step": 7077 }, { "epoch": 5.185347985347986, "grad_norm": 1.171875, "learning_rate": 9.74165977765278e-05, "loss": 0.1321, "step": 7078 }, { "epoch": 5.186080586080586, "grad_norm": 0.9140625, "learning_rate": 9.741550521838162e-05, "loss": 0.1334, "step": 7079 }, { "epoch": 5.186813186813187, "grad_norm": 0.66015625, "learning_rate": 9.741441243544783e-05, "loss": 0.1143, "step": 7080 }, { "epoch": 5.187545787545788, "grad_norm": 1.0546875, "learning_rate": 9.741331942773173e-05, "loss": 0.1375, "step": 7081 }, { "epoch": 5.1882783882783885, "grad_norm": 0.9765625, "learning_rate": 9.741222619523848e-05, "loss": 0.1736, "step": 7082 }, { "epoch": 5.189010989010989, "grad_norm": 0.96484375, "learning_rate": 9.741113273797337e-05, "loss": 0.1355, "step": 7083 }, { "epoch": 5.18974358974359, "grad_norm": 0.85546875, "learning_rate": 9.74100390559416e-05, "loss": 0.1219, "step": 7084 }, { "epoch": 5.190476190476191, "grad_norm": 0.67578125, "learning_rate": 9.740894514914847e-05, "loss": 0.1377, "step": 7085 }, { "epoch": 5.191208791208791, "grad_norm": 1.0, "learning_rate": 9.740785101759915e-05, "loss": 0.161, "step": 7086 }, { "epoch": 5.191941391941392, "grad_norm": 0.8359375, "learning_rate": 9.740675666129893e-05, "loss": 0.1677, "step": 7087 }, { "epoch": 5.192673992673993, "grad_norm": 1.3828125, "learning_rate": 9.740566208025306e-05, "loss": 0.1327, "step": 7088 }, { "epoch": 5.193406593406594, "grad_norm": 0.796875, "learning_rate": 9.740456727446673e-05, "loss": 0.1399, "step": 7089 }, { "epoch": 5.194139194139194, "grad_norm": 1.421875, "learning_rate": 9.740347224394525e-05, "loss": 0.1334, "step": 7090 }, { "epoch": 5.194871794871795, "grad_norm": 0.83984375, "learning_rate": 9.740237698869382e-05, "loss": 0.1554, "step": 7091 }, { "epoch": 5.195604395604396, "grad_norm": 0.83203125, "learning_rate": 9.740128150871772e-05, "loss": 0.1589, "step": 7092 }, { "epoch": 5.1963369963369965, "grad_norm": 0.8984375, "learning_rate": 9.740018580402216e-05, "loss": 0.1436, "step": 7093 }, { "epoch": 5.197069597069597, "grad_norm": 0.890625, "learning_rate": 9.739908987461244e-05, "loss": 0.1272, "step": 7094 }, { "epoch": 5.197802197802198, "grad_norm": 0.98828125, "learning_rate": 9.739799372049377e-05, "loss": 0.1228, "step": 7095 }, { "epoch": 5.198534798534799, "grad_norm": 0.64453125, "learning_rate": 9.739689734167141e-05, "loss": 0.1313, "step": 7096 }, { "epoch": 5.1992673992673994, "grad_norm": 0.98828125, "learning_rate": 9.739580073815062e-05, "loss": 0.1734, "step": 7097 }, { "epoch": 5.2, "grad_norm": 1.4765625, "learning_rate": 9.739470390993667e-05, "loss": 0.1543, "step": 7098 }, { "epoch": 5.200732600732601, "grad_norm": 0.8828125, "learning_rate": 9.739360685703478e-05, "loss": 0.1419, "step": 7099 }, { "epoch": 5.201465201465202, "grad_norm": 0.79296875, "learning_rate": 9.739250957945023e-05, "loss": 0.1404, "step": 7100 }, { "epoch": 5.202197802197802, "grad_norm": 0.93359375, "learning_rate": 9.739141207718829e-05, "loss": 0.1291, "step": 7101 }, { "epoch": 5.202930402930403, "grad_norm": 1.328125, "learning_rate": 9.73903143502542e-05, "loss": 0.1195, "step": 7102 }, { "epoch": 5.203663003663004, "grad_norm": 1.0625, "learning_rate": 9.73892163986532e-05, "loss": 0.1773, "step": 7103 }, { "epoch": 5.204395604395605, "grad_norm": 1.609375, "learning_rate": 9.738811822239059e-05, "loss": 0.1929, "step": 7104 }, { "epoch": 5.205128205128205, "grad_norm": 1.5390625, "learning_rate": 9.738701982147161e-05, "loss": 0.1241, "step": 7105 }, { "epoch": 5.205860805860806, "grad_norm": 0.89453125, "learning_rate": 9.73859211959015e-05, "loss": 0.148, "step": 7106 }, { "epoch": 5.206593406593407, "grad_norm": 1.0625, "learning_rate": 9.73848223456856e-05, "loss": 0.1661, "step": 7107 }, { "epoch": 5.2073260073260075, "grad_norm": 0.79296875, "learning_rate": 9.738372327082909e-05, "loss": 0.1598, "step": 7108 }, { "epoch": 5.208058608058608, "grad_norm": 1.125, "learning_rate": 9.738262397133726e-05, "loss": 0.1506, "step": 7109 }, { "epoch": 5.208791208791209, "grad_norm": 1.609375, "learning_rate": 9.73815244472154e-05, "loss": 0.1168, "step": 7110 }, { "epoch": 5.20952380952381, "grad_norm": 0.84375, "learning_rate": 9.738042469846878e-05, "loss": 0.1388, "step": 7111 }, { "epoch": 5.21025641025641, "grad_norm": 1.265625, "learning_rate": 9.737932472510265e-05, "loss": 0.1237, "step": 7112 }, { "epoch": 5.210989010989011, "grad_norm": 0.96484375, "learning_rate": 9.737822452712228e-05, "loss": 0.115, "step": 7113 }, { "epoch": 5.211721611721612, "grad_norm": 1.2109375, "learning_rate": 9.737712410453295e-05, "loss": 0.1655, "step": 7114 }, { "epoch": 5.212454212454213, "grad_norm": 0.8828125, "learning_rate": 9.737602345733992e-05, "loss": 0.1338, "step": 7115 }, { "epoch": 5.213186813186813, "grad_norm": 1.1953125, "learning_rate": 9.73749225855485e-05, "loss": 0.1317, "step": 7116 }, { "epoch": 5.213919413919414, "grad_norm": 1.0703125, "learning_rate": 9.737382148916392e-05, "loss": 0.135, "step": 7117 }, { "epoch": 5.214652014652015, "grad_norm": 1.0078125, "learning_rate": 9.737272016819147e-05, "loss": 0.139, "step": 7118 }, { "epoch": 5.2153846153846155, "grad_norm": 1.4375, "learning_rate": 9.737161862263643e-05, "loss": 0.145, "step": 7119 }, { "epoch": 5.216117216117216, "grad_norm": 1.1328125, "learning_rate": 9.737051685250408e-05, "loss": 0.1855, "step": 7120 }, { "epoch": 5.216849816849817, "grad_norm": 0.58203125, "learning_rate": 9.73694148577997e-05, "loss": 0.1218, "step": 7121 }, { "epoch": 5.217582417582418, "grad_norm": 1.0078125, "learning_rate": 9.736831263852857e-05, "loss": 0.1296, "step": 7122 }, { "epoch": 5.218315018315018, "grad_norm": 1.0703125, "learning_rate": 9.736721019469596e-05, "loss": 0.1508, "step": 7123 }, { "epoch": 5.219047619047619, "grad_norm": 0.90234375, "learning_rate": 9.736610752630717e-05, "loss": 0.1266, "step": 7124 }, { "epoch": 5.21978021978022, "grad_norm": 0.94921875, "learning_rate": 9.736500463336747e-05, "loss": 0.1252, "step": 7125 }, { "epoch": 5.220512820512821, "grad_norm": 0.9921875, "learning_rate": 9.736390151588215e-05, "loss": 0.1112, "step": 7126 }, { "epoch": 5.221245421245421, "grad_norm": 1.1640625, "learning_rate": 9.736279817385648e-05, "loss": 0.1495, "step": 7127 }, { "epoch": 5.221978021978022, "grad_norm": 0.91015625, "learning_rate": 9.736169460729577e-05, "loss": 0.163, "step": 7128 }, { "epoch": 5.222710622710623, "grad_norm": 0.89453125, "learning_rate": 9.736059081620532e-05, "loss": 0.1599, "step": 7129 }, { "epoch": 5.2234432234432235, "grad_norm": 0.84375, "learning_rate": 9.735948680059036e-05, "loss": 0.1326, "step": 7130 }, { "epoch": 5.224175824175824, "grad_norm": 1.09375, "learning_rate": 9.735838256045623e-05, "loss": 0.1452, "step": 7131 }, { "epoch": 5.224908424908425, "grad_norm": 1.3359375, "learning_rate": 9.735727809580822e-05, "loss": 0.1331, "step": 7132 }, { "epoch": 5.225641025641026, "grad_norm": 1.609375, "learning_rate": 9.735617340665158e-05, "loss": 0.1993, "step": 7133 }, { "epoch": 5.2263736263736265, "grad_norm": 1.109375, "learning_rate": 9.735506849299165e-05, "loss": 0.1136, "step": 7134 }, { "epoch": 5.227106227106227, "grad_norm": 0.91015625, "learning_rate": 9.73539633548337e-05, "loss": 0.1421, "step": 7135 }, { "epoch": 5.227838827838828, "grad_norm": 0.8671875, "learning_rate": 9.735285799218304e-05, "loss": 0.1783, "step": 7136 }, { "epoch": 5.228571428571429, "grad_norm": 0.98046875, "learning_rate": 9.735175240504495e-05, "loss": 0.1008, "step": 7137 }, { "epoch": 5.229304029304029, "grad_norm": 1.1171875, "learning_rate": 9.735064659342473e-05, "loss": 0.1744, "step": 7138 }, { "epoch": 5.23003663003663, "grad_norm": 1.109375, "learning_rate": 9.734954055732768e-05, "loss": 0.1635, "step": 7139 }, { "epoch": 5.230769230769231, "grad_norm": 0.87109375, "learning_rate": 9.734843429675911e-05, "loss": 0.1501, "step": 7140 }, { "epoch": 5.231501831501832, "grad_norm": 0.8125, "learning_rate": 9.73473278117243e-05, "loss": 0.1922, "step": 7141 }, { "epoch": 5.232234432234432, "grad_norm": 1.09375, "learning_rate": 9.734622110222858e-05, "loss": 0.136, "step": 7142 }, { "epoch": 5.232967032967033, "grad_norm": 0.84765625, "learning_rate": 9.734511416827723e-05, "loss": 0.1936, "step": 7143 }, { "epoch": 5.233699633699634, "grad_norm": 1.0859375, "learning_rate": 9.734400700987554e-05, "loss": 0.1326, "step": 7144 }, { "epoch": 5.2344322344322345, "grad_norm": 1.453125, "learning_rate": 9.734289962702885e-05, "loss": 0.1235, "step": 7145 }, { "epoch": 5.235164835164835, "grad_norm": 1.109375, "learning_rate": 9.734179201974248e-05, "loss": 0.1647, "step": 7146 }, { "epoch": 5.235897435897436, "grad_norm": 1.0703125, "learning_rate": 9.734068418802167e-05, "loss": 0.1751, "step": 7147 }, { "epoch": 5.236630036630037, "grad_norm": 1.1328125, "learning_rate": 9.733957613187177e-05, "loss": 0.1539, "step": 7148 }, { "epoch": 5.237362637362637, "grad_norm": 0.91796875, "learning_rate": 9.733846785129809e-05, "loss": 0.1584, "step": 7149 }, { "epoch": 5.238095238095238, "grad_norm": 0.94140625, "learning_rate": 9.733735934630593e-05, "loss": 0.1279, "step": 7150 }, { "epoch": 5.238827838827839, "grad_norm": 0.69140625, "learning_rate": 9.733625061690061e-05, "loss": 0.1688, "step": 7151 }, { "epoch": 5.23956043956044, "grad_norm": 1.3046875, "learning_rate": 9.733514166308743e-05, "loss": 0.1526, "step": 7152 }, { "epoch": 5.24029304029304, "grad_norm": 1.2421875, "learning_rate": 9.733403248487173e-05, "loss": 0.1398, "step": 7153 }, { "epoch": 5.241025641025641, "grad_norm": 0.765625, "learning_rate": 9.733292308225879e-05, "loss": 0.1506, "step": 7154 }, { "epoch": 5.241758241758242, "grad_norm": 1.15625, "learning_rate": 9.733181345525394e-05, "loss": 0.1825, "step": 7155 }, { "epoch": 5.2424908424908425, "grad_norm": 2.3125, "learning_rate": 9.733070360386252e-05, "loss": 0.1531, "step": 7156 }, { "epoch": 5.243223443223443, "grad_norm": 1.78125, "learning_rate": 9.73295935280898e-05, "loss": 0.1608, "step": 7157 }, { "epoch": 5.243956043956044, "grad_norm": 2.015625, "learning_rate": 9.732848322794114e-05, "loss": 0.1471, "step": 7158 }, { "epoch": 5.244688644688645, "grad_norm": 0.73828125, "learning_rate": 9.732737270342183e-05, "loss": 0.1254, "step": 7159 }, { "epoch": 5.245421245421245, "grad_norm": 1.0546875, "learning_rate": 9.732626195453723e-05, "loss": 0.1499, "step": 7160 }, { "epoch": 5.246153846153846, "grad_norm": 1.0703125, "learning_rate": 9.73251509812926e-05, "loss": 0.1468, "step": 7161 }, { "epoch": 5.246886446886447, "grad_norm": 0.95703125, "learning_rate": 9.732403978369333e-05, "loss": 0.1375, "step": 7162 }, { "epoch": 5.247619047619048, "grad_norm": 1.6640625, "learning_rate": 9.73229283617447e-05, "loss": 0.1444, "step": 7163 }, { "epoch": 5.248351648351648, "grad_norm": 1.828125, "learning_rate": 9.732181671545205e-05, "loss": 0.1717, "step": 7164 }, { "epoch": 5.249084249084249, "grad_norm": 1.484375, "learning_rate": 9.73207048448207e-05, "loss": 0.1648, "step": 7165 }, { "epoch": 5.24981684981685, "grad_norm": 1.2734375, "learning_rate": 9.731959274985599e-05, "loss": 0.1066, "step": 7166 }, { "epoch": 5.2505494505494505, "grad_norm": 1.9765625, "learning_rate": 9.731848043056326e-05, "loss": 0.1714, "step": 7167 }, { "epoch": 5.251282051282051, "grad_norm": 1.890625, "learning_rate": 9.731736788694778e-05, "loss": 0.1654, "step": 7168 }, { "epoch": 5.252014652014652, "grad_norm": 1.46875, "learning_rate": 9.731625511901496e-05, "loss": 0.1622, "step": 7169 }, { "epoch": 5.252747252747253, "grad_norm": 0.94921875, "learning_rate": 9.731514212677008e-05, "loss": 0.1292, "step": 7170 }, { "epoch": 5.2534798534798535, "grad_norm": 1.078125, "learning_rate": 9.731402891021847e-05, "loss": 0.1603, "step": 7171 }, { "epoch": 5.254212454212454, "grad_norm": 1.59375, "learning_rate": 9.731291546936548e-05, "loss": 0.1435, "step": 7172 }, { "epoch": 5.254945054945055, "grad_norm": 1.4609375, "learning_rate": 9.731180180421648e-05, "loss": 0.1374, "step": 7173 }, { "epoch": 5.255677655677656, "grad_norm": 1.609375, "learning_rate": 9.731068791477674e-05, "loss": 0.2142, "step": 7174 }, { "epoch": 5.256410256410256, "grad_norm": 1.6171875, "learning_rate": 9.730957380105162e-05, "loss": 0.1398, "step": 7175 }, { "epoch": 5.257142857142857, "grad_norm": 0.58984375, "learning_rate": 9.730845946304649e-05, "loss": 0.132, "step": 7176 }, { "epoch": 5.257875457875458, "grad_norm": 0.828125, "learning_rate": 9.730734490076665e-05, "loss": 0.1584, "step": 7177 }, { "epoch": 5.258608058608059, "grad_norm": 1.0390625, "learning_rate": 9.730623011421747e-05, "loss": 0.1074, "step": 7178 }, { "epoch": 5.259340659340659, "grad_norm": 1.1015625, "learning_rate": 9.730511510340427e-05, "loss": 0.1798, "step": 7179 }, { "epoch": 5.26007326007326, "grad_norm": 1.109375, "learning_rate": 9.730399986833239e-05, "loss": 0.176, "step": 7180 }, { "epoch": 5.260805860805861, "grad_norm": 1.2578125, "learning_rate": 9.730288440900718e-05, "loss": 0.1396, "step": 7181 }, { "epoch": 5.2615384615384615, "grad_norm": 1.140625, "learning_rate": 9.730176872543398e-05, "loss": 0.201, "step": 7182 }, { "epoch": 5.262271062271062, "grad_norm": 1.2109375, "learning_rate": 9.730065281761817e-05, "loss": 0.1561, "step": 7183 }, { "epoch": 5.263003663003663, "grad_norm": 1.1328125, "learning_rate": 9.729953668556505e-05, "loss": 0.1419, "step": 7184 }, { "epoch": 5.263736263736264, "grad_norm": 0.671875, "learning_rate": 9.729842032927998e-05, "loss": 0.1207, "step": 7185 }, { "epoch": 5.264468864468864, "grad_norm": 1.40625, "learning_rate": 9.729730374876833e-05, "loss": 0.1397, "step": 7186 }, { "epoch": 5.265201465201465, "grad_norm": 0.96484375, "learning_rate": 9.729618694403542e-05, "loss": 0.165, "step": 7187 }, { "epoch": 5.265934065934066, "grad_norm": 1.5546875, "learning_rate": 9.729506991508663e-05, "loss": 0.181, "step": 7188 }, { "epoch": 5.266666666666667, "grad_norm": 1.4296875, "learning_rate": 9.72939526619273e-05, "loss": 0.156, "step": 7189 }, { "epoch": 5.267399267399267, "grad_norm": 1.109375, "learning_rate": 9.729283518456277e-05, "loss": 0.15, "step": 7190 }, { "epoch": 5.268131868131868, "grad_norm": 0.7734375, "learning_rate": 9.729171748299842e-05, "loss": 0.1509, "step": 7191 }, { "epoch": 5.268864468864469, "grad_norm": 0.9453125, "learning_rate": 9.729059955723957e-05, "loss": 0.1131, "step": 7192 }, { "epoch": 5.2695970695970695, "grad_norm": 0.91796875, "learning_rate": 9.728948140729162e-05, "loss": 0.1684, "step": 7193 }, { "epoch": 5.27032967032967, "grad_norm": 0.859375, "learning_rate": 9.728836303315988e-05, "loss": 0.1308, "step": 7194 }, { "epoch": 5.271062271062271, "grad_norm": 1.6640625, "learning_rate": 9.728724443484974e-05, "loss": 0.1782, "step": 7195 }, { "epoch": 5.271794871794872, "grad_norm": 1.7890625, "learning_rate": 9.728612561236658e-05, "loss": 0.1565, "step": 7196 }, { "epoch": 5.2725274725274724, "grad_norm": 0.875, "learning_rate": 9.72850065657157e-05, "loss": 0.1597, "step": 7197 }, { "epoch": 5.273260073260073, "grad_norm": 0.75, "learning_rate": 9.72838872949025e-05, "loss": 0.1372, "step": 7198 }, { "epoch": 5.273992673992674, "grad_norm": 1.3984375, "learning_rate": 9.728276779993235e-05, "loss": 0.1514, "step": 7199 }, { "epoch": 5.274725274725275, "grad_norm": 1.0859375, "learning_rate": 9.728164808081061e-05, "loss": 0.1672, "step": 7200 }, { "epoch": 5.275457875457875, "grad_norm": 0.8984375, "learning_rate": 9.72805281375426e-05, "loss": 0.1368, "step": 7201 }, { "epoch": 5.276190476190476, "grad_norm": 0.65625, "learning_rate": 9.727940797013375e-05, "loss": 0.1444, "step": 7202 }, { "epoch": 5.276923076923077, "grad_norm": 0.796875, "learning_rate": 9.72782875785894e-05, "loss": 0.1854, "step": 7203 }, { "epoch": 5.277655677655678, "grad_norm": 0.921875, "learning_rate": 9.727716696291491e-05, "loss": 0.1442, "step": 7204 }, { "epoch": 5.278388278388278, "grad_norm": 1.03125, "learning_rate": 9.727604612311567e-05, "loss": 0.1598, "step": 7205 }, { "epoch": 5.279120879120879, "grad_norm": 0.7265625, "learning_rate": 9.727492505919703e-05, "loss": 0.1323, "step": 7206 }, { "epoch": 5.27985347985348, "grad_norm": 1.15625, "learning_rate": 9.727380377116435e-05, "loss": 0.1629, "step": 7207 }, { "epoch": 5.2805860805860805, "grad_norm": 1.0703125, "learning_rate": 9.727268225902305e-05, "loss": 0.1608, "step": 7208 }, { "epoch": 5.281318681318681, "grad_norm": 0.65625, "learning_rate": 9.727156052277846e-05, "loss": 0.1365, "step": 7209 }, { "epoch": 5.282051282051282, "grad_norm": 1.0703125, "learning_rate": 9.727043856243598e-05, "loss": 0.1308, "step": 7210 }, { "epoch": 5.282783882783883, "grad_norm": 1.1015625, "learning_rate": 9.726931637800097e-05, "loss": 0.1523, "step": 7211 }, { "epoch": 5.283516483516483, "grad_norm": 1.1953125, "learning_rate": 9.72681939694788e-05, "loss": 0.1919, "step": 7212 }, { "epoch": 5.284249084249084, "grad_norm": 1.234375, "learning_rate": 9.726707133687487e-05, "loss": 0.1691, "step": 7213 }, { "epoch": 5.284981684981685, "grad_norm": 1.1171875, "learning_rate": 9.726594848019456e-05, "loss": 0.1546, "step": 7214 }, { "epoch": 5.285714285714286, "grad_norm": 1.1171875, "learning_rate": 9.726482539944322e-05, "loss": 0.15, "step": 7215 }, { "epoch": 5.286446886446886, "grad_norm": 0.8671875, "learning_rate": 9.726370209462625e-05, "loss": 0.1417, "step": 7216 }, { "epoch": 5.287179487179487, "grad_norm": 0.94140625, "learning_rate": 9.726257856574904e-05, "loss": 0.1299, "step": 7217 }, { "epoch": 5.287912087912088, "grad_norm": 0.62109375, "learning_rate": 9.726145481281697e-05, "loss": 0.1329, "step": 7218 }, { "epoch": 5.2886446886446885, "grad_norm": 0.88671875, "learning_rate": 9.72603308358354e-05, "loss": 0.1766, "step": 7219 }, { "epoch": 5.289377289377289, "grad_norm": 1.34375, "learning_rate": 9.725920663480975e-05, "loss": 0.153, "step": 7220 }, { "epoch": 5.29010989010989, "grad_norm": 1.5703125, "learning_rate": 9.725808220974538e-05, "loss": 0.1488, "step": 7221 }, { "epoch": 5.290842490842491, "grad_norm": 0.9453125, "learning_rate": 9.725695756064771e-05, "loss": 0.1392, "step": 7222 }, { "epoch": 5.291575091575091, "grad_norm": 0.953125, "learning_rate": 9.725583268752209e-05, "loss": 0.158, "step": 7223 }, { "epoch": 5.292307692307692, "grad_norm": 1.0234375, "learning_rate": 9.725470759037392e-05, "loss": 0.1602, "step": 7224 }, { "epoch": 5.293040293040293, "grad_norm": 1.1796875, "learning_rate": 9.725358226920861e-05, "loss": 0.1768, "step": 7225 }, { "epoch": 5.293772893772894, "grad_norm": 1.109375, "learning_rate": 9.725245672403154e-05, "loss": 0.1625, "step": 7226 }, { "epoch": 5.294505494505494, "grad_norm": 1.203125, "learning_rate": 9.72513309548481e-05, "loss": 0.1422, "step": 7227 }, { "epoch": 5.295238095238095, "grad_norm": 1.1171875, "learning_rate": 9.725020496166368e-05, "loss": 0.1285, "step": 7228 }, { "epoch": 5.295970695970696, "grad_norm": 0.921875, "learning_rate": 9.724907874448369e-05, "loss": 0.1322, "step": 7229 }, { "epoch": 5.2967032967032965, "grad_norm": 0.8828125, "learning_rate": 9.724795230331352e-05, "loss": 0.1341, "step": 7230 }, { "epoch": 5.297435897435897, "grad_norm": 0.84765625, "learning_rate": 9.724682563815855e-05, "loss": 0.1309, "step": 7231 }, { "epoch": 5.298168498168498, "grad_norm": 0.921875, "learning_rate": 9.724569874902419e-05, "loss": 0.1868, "step": 7232 }, { "epoch": 5.298901098901099, "grad_norm": 0.7890625, "learning_rate": 9.724457163591585e-05, "loss": 0.1132, "step": 7233 }, { "epoch": 5.2996336996336995, "grad_norm": 1.03125, "learning_rate": 9.724344429883892e-05, "loss": 0.1419, "step": 7234 }, { "epoch": 5.3003663003663, "grad_norm": 0.90234375, "learning_rate": 9.724231673779882e-05, "loss": 0.1712, "step": 7235 }, { "epoch": 5.301098901098901, "grad_norm": 0.8046875, "learning_rate": 9.724118895280092e-05, "loss": 0.1745, "step": 7236 }, { "epoch": 5.301831501831502, "grad_norm": 0.98046875, "learning_rate": 9.724006094385063e-05, "loss": 0.145, "step": 7237 }, { "epoch": 5.302564102564102, "grad_norm": 1.453125, "learning_rate": 9.723893271095337e-05, "loss": 0.1642, "step": 7238 }, { "epoch": 5.303296703296703, "grad_norm": 0.76171875, "learning_rate": 9.723780425411456e-05, "loss": 0.1514, "step": 7239 }, { "epoch": 5.304029304029304, "grad_norm": 0.828125, "learning_rate": 9.723667557333957e-05, "loss": 0.139, "step": 7240 }, { "epoch": 5.304761904761905, "grad_norm": 0.8515625, "learning_rate": 9.723554666863381e-05, "loss": 0.1471, "step": 7241 }, { "epoch": 5.305494505494505, "grad_norm": 0.8671875, "learning_rate": 9.723441754000275e-05, "loss": 0.164, "step": 7242 }, { "epoch": 5.306227106227106, "grad_norm": 0.875, "learning_rate": 9.723328818745171e-05, "loss": 0.1467, "step": 7243 }, { "epoch": 5.306959706959707, "grad_norm": 1.078125, "learning_rate": 9.723215861098615e-05, "loss": 0.1421, "step": 7244 }, { "epoch": 5.3076923076923075, "grad_norm": 1.125, "learning_rate": 9.723102881061151e-05, "loss": 0.1288, "step": 7245 }, { "epoch": 5.308424908424908, "grad_norm": 1.1953125, "learning_rate": 9.722989878633313e-05, "loss": 0.1515, "step": 7246 }, { "epoch": 5.309157509157509, "grad_norm": 1.6796875, "learning_rate": 9.722876853815648e-05, "loss": 0.1478, "step": 7247 }, { "epoch": 5.30989010989011, "grad_norm": 0.80078125, "learning_rate": 9.722763806608696e-05, "loss": 0.1398, "step": 7248 }, { "epoch": 5.31062271062271, "grad_norm": 1.0234375, "learning_rate": 9.722650737013e-05, "loss": 0.1301, "step": 7249 }, { "epoch": 5.311355311355311, "grad_norm": 0.7890625, "learning_rate": 9.722537645029098e-05, "loss": 0.1354, "step": 7250 }, { "epoch": 5.312087912087912, "grad_norm": 1.7890625, "learning_rate": 9.722424530657536e-05, "loss": 0.1862, "step": 7251 }, { "epoch": 5.312820512820513, "grad_norm": 2.65625, "learning_rate": 9.722311393898852e-05, "loss": 0.1548, "step": 7252 }, { "epoch": 5.313553113553113, "grad_norm": 2.46875, "learning_rate": 9.722198234753593e-05, "loss": 0.1932, "step": 7253 }, { "epoch": 5.314285714285714, "grad_norm": 1.5234375, "learning_rate": 9.722085053222297e-05, "loss": 0.1572, "step": 7254 }, { "epoch": 5.315018315018315, "grad_norm": 1.1328125, "learning_rate": 9.721971849305508e-05, "loss": 0.1317, "step": 7255 }, { "epoch": 5.3157509157509155, "grad_norm": 1.78125, "learning_rate": 9.721858623003767e-05, "loss": 0.1686, "step": 7256 }, { "epoch": 5.316483516483516, "grad_norm": 3.421875, "learning_rate": 9.72174537431762e-05, "loss": 0.1853, "step": 7257 }, { "epoch": 5.317216117216117, "grad_norm": 2.5, "learning_rate": 9.721632103247606e-05, "loss": 0.1549, "step": 7258 }, { "epoch": 5.317948717948718, "grad_norm": 2.671875, "learning_rate": 9.72151880979427e-05, "loss": 0.1569, "step": 7259 }, { "epoch": 5.318681318681318, "grad_norm": 1.9609375, "learning_rate": 9.721405493958152e-05, "loss": 0.147, "step": 7260 }, { "epoch": 5.319413919413919, "grad_norm": 0.953125, "learning_rate": 9.721292155739797e-05, "loss": 0.1414, "step": 7261 }, { "epoch": 5.32014652014652, "grad_norm": 1.5390625, "learning_rate": 9.721178795139749e-05, "loss": 0.1653, "step": 7262 }, { "epoch": 5.320879120879121, "grad_norm": 2.6875, "learning_rate": 9.721065412158551e-05, "loss": 0.1878, "step": 7263 }, { "epoch": 5.321611721611721, "grad_norm": 2.546875, "learning_rate": 9.720952006796743e-05, "loss": 0.1423, "step": 7264 }, { "epoch": 5.322344322344322, "grad_norm": 2.78125, "learning_rate": 9.720838579054871e-05, "loss": 0.1572, "step": 7265 }, { "epoch": 5.323076923076923, "grad_norm": 2.078125, "learning_rate": 9.720725128933479e-05, "loss": 0.1981, "step": 7266 }, { "epoch": 5.3238095238095235, "grad_norm": 1.375, "learning_rate": 9.720611656433108e-05, "loss": 0.143, "step": 7267 }, { "epoch": 5.324542124542124, "grad_norm": 1.1953125, "learning_rate": 9.720498161554305e-05, "loss": 0.1637, "step": 7268 }, { "epoch": 5.325274725274725, "grad_norm": 1.828125, "learning_rate": 9.72038464429761e-05, "loss": 0.142, "step": 7269 }, { "epoch": 5.326007326007326, "grad_norm": 2.203125, "learning_rate": 9.720271104663572e-05, "loss": 0.1627, "step": 7270 }, { "epoch": 5.3267399267399265, "grad_norm": 2.5625, "learning_rate": 9.72015754265273e-05, "loss": 0.1253, "step": 7271 }, { "epoch": 5.327472527472527, "grad_norm": 1.0859375, "learning_rate": 9.72004395826563e-05, "loss": 0.1368, "step": 7272 }, { "epoch": 5.328205128205128, "grad_norm": 0.8125, "learning_rate": 9.719930351502816e-05, "loss": 0.1475, "step": 7273 }, { "epoch": 5.328937728937729, "grad_norm": 1.3984375, "learning_rate": 9.719816722364835e-05, "loss": 0.1702, "step": 7274 }, { "epoch": 5.329670329670329, "grad_norm": 1.625, "learning_rate": 9.719703070852227e-05, "loss": 0.1759, "step": 7275 }, { "epoch": 5.33040293040293, "grad_norm": 1.8125, "learning_rate": 9.719589396965539e-05, "loss": 0.1639, "step": 7276 }, { "epoch": 5.331135531135531, "grad_norm": 1.7890625, "learning_rate": 9.719475700705314e-05, "loss": 0.1128, "step": 7277 }, { "epoch": 5.331868131868132, "grad_norm": 0.82421875, "learning_rate": 9.719361982072101e-05, "loss": 0.1188, "step": 7278 }, { "epoch": 5.332600732600732, "grad_norm": 1.5234375, "learning_rate": 9.71924824106644e-05, "loss": 0.1806, "step": 7279 }, { "epoch": 5.333333333333333, "grad_norm": 0.70703125, "learning_rate": 9.719134477688877e-05, "loss": 0.1769, "step": 7280 }, { "epoch": 5.334065934065934, "grad_norm": 0.953125, "learning_rate": 9.719020691939959e-05, "loss": 0.1297, "step": 7281 }, { "epoch": 5.3347985347985345, "grad_norm": 1.15625, "learning_rate": 9.718906883820232e-05, "loss": 0.1206, "step": 7282 }, { "epoch": 5.335531135531135, "grad_norm": 1.4453125, "learning_rate": 9.718793053330235e-05, "loss": 0.1197, "step": 7283 }, { "epoch": 5.336263736263736, "grad_norm": 1.34375, "learning_rate": 9.718679200470521e-05, "loss": 0.1513, "step": 7284 }, { "epoch": 5.336996336996337, "grad_norm": 0.6328125, "learning_rate": 9.718565325241632e-05, "loss": 0.1288, "step": 7285 }, { "epoch": 5.337728937728937, "grad_norm": 1.0, "learning_rate": 9.718451427644114e-05, "loss": 0.1404, "step": 7286 }, { "epoch": 5.338461538461538, "grad_norm": 0.78515625, "learning_rate": 9.718337507678512e-05, "loss": 0.1579, "step": 7287 }, { "epoch": 5.339194139194139, "grad_norm": 1.0234375, "learning_rate": 9.718223565345373e-05, "loss": 0.1693, "step": 7288 }, { "epoch": 5.33992673992674, "grad_norm": 1.46875, "learning_rate": 9.718109600645243e-05, "loss": 0.1733, "step": 7289 }, { "epoch": 5.34065934065934, "grad_norm": 0.8203125, "learning_rate": 9.717995613578666e-05, "loss": 0.1621, "step": 7290 }, { "epoch": 5.341391941391941, "grad_norm": 0.87890625, "learning_rate": 9.717881604146191e-05, "loss": 0.1414, "step": 7291 }, { "epoch": 5.342124542124542, "grad_norm": 0.99609375, "learning_rate": 9.717767572348362e-05, "loss": 0.1699, "step": 7292 }, { "epoch": 5.3428571428571425, "grad_norm": 1.28125, "learning_rate": 9.717653518185727e-05, "loss": 0.1508, "step": 7293 }, { "epoch": 5.343589743589743, "grad_norm": 1.1484375, "learning_rate": 9.71753944165883e-05, "loss": 0.1491, "step": 7294 }, { "epoch": 5.344322344322344, "grad_norm": 0.99609375, "learning_rate": 9.71742534276822e-05, "loss": 0.1293, "step": 7295 }, { "epoch": 5.345054945054945, "grad_norm": 0.5859375, "learning_rate": 9.717311221514443e-05, "loss": 0.1095, "step": 7296 }, { "epoch": 5.3457875457875454, "grad_norm": 1.0859375, "learning_rate": 9.717197077898046e-05, "loss": 0.1863, "step": 7297 }, { "epoch": 5.346520146520146, "grad_norm": 0.93359375, "learning_rate": 9.717082911919578e-05, "loss": 0.1318, "step": 7298 }, { "epoch": 5.347252747252747, "grad_norm": 0.76953125, "learning_rate": 9.716968723579581e-05, "loss": 0.1591, "step": 7299 }, { "epoch": 5.347985347985348, "grad_norm": 0.5859375, "learning_rate": 9.716854512878604e-05, "loss": 0.1428, "step": 7300 }, { "epoch": 5.348717948717948, "grad_norm": 0.70703125, "learning_rate": 9.716740279817196e-05, "loss": 0.1225, "step": 7301 }, { "epoch": 5.349450549450549, "grad_norm": 0.98046875, "learning_rate": 9.716626024395904e-05, "loss": 0.1448, "step": 7302 }, { "epoch": 5.35018315018315, "grad_norm": 1.046875, "learning_rate": 9.716511746615274e-05, "loss": 0.2027, "step": 7303 }, { "epoch": 5.350915750915751, "grad_norm": 0.66015625, "learning_rate": 9.716397446475855e-05, "loss": 0.1139, "step": 7304 }, { "epoch": 5.351648351648351, "grad_norm": 1.15625, "learning_rate": 9.716283123978194e-05, "loss": 0.1769, "step": 7305 }, { "epoch": 5.352380952380952, "grad_norm": 0.8984375, "learning_rate": 9.716168779122839e-05, "loss": 0.1387, "step": 7306 }, { "epoch": 5.353113553113553, "grad_norm": 1.0625, "learning_rate": 9.716054411910336e-05, "loss": 0.147, "step": 7307 }, { "epoch": 5.3538461538461535, "grad_norm": 0.859375, "learning_rate": 9.715940022341235e-05, "loss": 0.1703, "step": 7308 }, { "epoch": 5.354578754578754, "grad_norm": 1.140625, "learning_rate": 9.715825610416083e-05, "loss": 0.1842, "step": 7309 }, { "epoch": 5.355311355311355, "grad_norm": 1.203125, "learning_rate": 9.715711176135431e-05, "loss": 0.1462, "step": 7310 }, { "epoch": 5.356043956043956, "grad_norm": 0.87890625, "learning_rate": 9.715596719499823e-05, "loss": 0.1504, "step": 7311 }, { "epoch": 5.356776556776556, "grad_norm": 0.90625, "learning_rate": 9.715482240509811e-05, "loss": 0.1214, "step": 7312 }, { "epoch": 5.357509157509157, "grad_norm": 1.5390625, "learning_rate": 9.715367739165942e-05, "loss": 0.1176, "step": 7313 }, { "epoch": 5.358241758241758, "grad_norm": 0.52734375, "learning_rate": 9.715253215468763e-05, "loss": 0.1342, "step": 7314 }, { "epoch": 5.358974358974359, "grad_norm": 0.93359375, "learning_rate": 9.715138669418826e-05, "loss": 0.1909, "step": 7315 }, { "epoch": 5.359706959706959, "grad_norm": 0.71875, "learning_rate": 9.715024101016677e-05, "loss": 0.1494, "step": 7316 }, { "epoch": 5.36043956043956, "grad_norm": 0.8984375, "learning_rate": 9.714909510262867e-05, "loss": 0.1765, "step": 7317 }, { "epoch": 5.361172161172161, "grad_norm": 0.9453125, "learning_rate": 9.714794897157944e-05, "loss": 0.1431, "step": 7318 }, { "epoch": 5.3619047619047615, "grad_norm": 1.234375, "learning_rate": 9.714680261702458e-05, "loss": 0.1332, "step": 7319 }, { "epoch": 5.362637362637362, "grad_norm": 0.90625, "learning_rate": 9.714565603896956e-05, "loss": 0.1735, "step": 7320 }, { "epoch": 5.363369963369963, "grad_norm": 1.359375, "learning_rate": 9.71445092374199e-05, "loss": 0.1422, "step": 7321 }, { "epoch": 5.364102564102564, "grad_norm": 0.6796875, "learning_rate": 9.714336221238109e-05, "loss": 0.1421, "step": 7322 }, { "epoch": 5.364835164835164, "grad_norm": 0.84765625, "learning_rate": 9.714221496385859e-05, "loss": 0.1892, "step": 7323 }, { "epoch": 5.365567765567765, "grad_norm": 0.87890625, "learning_rate": 9.714106749185795e-05, "loss": 0.1222, "step": 7324 }, { "epoch": 5.366300366300366, "grad_norm": 1.015625, "learning_rate": 9.713991979638464e-05, "loss": 0.1605, "step": 7325 }, { "epoch": 5.367032967032967, "grad_norm": 1.984375, "learning_rate": 9.713877187744417e-05, "loss": 0.1948, "step": 7326 }, { "epoch": 5.367765567765568, "grad_norm": 0.9921875, "learning_rate": 9.713762373504202e-05, "loss": 0.1365, "step": 7327 }, { "epoch": 5.368498168498169, "grad_norm": 0.9609375, "learning_rate": 9.713647536918371e-05, "loss": 0.1546, "step": 7328 }, { "epoch": 5.36923076923077, "grad_norm": 1.65625, "learning_rate": 9.713532677987475e-05, "loss": 0.16, "step": 7329 }, { "epoch": 5.36996336996337, "grad_norm": 0.65625, "learning_rate": 9.713417796712061e-05, "loss": 0.1411, "step": 7330 }, { "epoch": 5.370695970695971, "grad_norm": 1.2265625, "learning_rate": 9.713302893092683e-05, "loss": 0.1197, "step": 7331 }, { "epoch": 5.371428571428572, "grad_norm": 1.46875, "learning_rate": 9.713187967129889e-05, "loss": 0.142, "step": 7332 }, { "epoch": 5.372161172161173, "grad_norm": 1.53125, "learning_rate": 9.713073018824231e-05, "loss": 0.1635, "step": 7333 }, { "epoch": 5.372893772893773, "grad_norm": 1.15625, "learning_rate": 9.712958048176258e-05, "loss": 0.1381, "step": 7334 }, { "epoch": 5.373626373626374, "grad_norm": 0.87890625, "learning_rate": 9.712843055186525e-05, "loss": 0.1365, "step": 7335 }, { "epoch": 5.374358974358975, "grad_norm": 1.1796875, "learning_rate": 9.712728039855578e-05, "loss": 0.1362, "step": 7336 }, { "epoch": 5.3750915750915755, "grad_norm": 1.203125, "learning_rate": 9.712613002183971e-05, "loss": 0.1248, "step": 7337 }, { "epoch": 5.375824175824176, "grad_norm": 0.8359375, "learning_rate": 9.712497942172254e-05, "loss": 0.1443, "step": 7338 }, { "epoch": 5.376556776556777, "grad_norm": 1.3515625, "learning_rate": 9.712382859820979e-05, "loss": 0.1877, "step": 7339 }, { "epoch": 5.377289377289378, "grad_norm": 0.82421875, "learning_rate": 9.712267755130698e-05, "loss": 0.1406, "step": 7340 }, { "epoch": 5.3780219780219785, "grad_norm": 0.953125, "learning_rate": 9.71215262810196e-05, "loss": 0.1487, "step": 7341 }, { "epoch": 5.378754578754579, "grad_norm": 0.97265625, "learning_rate": 9.71203747873532e-05, "loss": 0.1675, "step": 7342 }, { "epoch": 5.37948717948718, "grad_norm": 1.5859375, "learning_rate": 9.711922307031327e-05, "loss": 0.1563, "step": 7343 }, { "epoch": 5.380219780219781, "grad_norm": 1.484375, "learning_rate": 9.711807112990535e-05, "loss": 0.1438, "step": 7344 }, { "epoch": 5.380952380952381, "grad_norm": 1.515625, "learning_rate": 9.711691896613494e-05, "loss": 0.146, "step": 7345 }, { "epoch": 5.381684981684982, "grad_norm": 1.3125, "learning_rate": 9.711576657900755e-05, "loss": 0.1589, "step": 7346 }, { "epoch": 5.382417582417583, "grad_norm": 0.71875, "learning_rate": 9.711461396852876e-05, "loss": 0.1201, "step": 7347 }, { "epoch": 5.383150183150184, "grad_norm": 1.3515625, "learning_rate": 9.7113461134704e-05, "loss": 0.1343, "step": 7348 }, { "epoch": 5.383882783882784, "grad_norm": 1.6640625, "learning_rate": 9.711230807753889e-05, "loss": 0.1595, "step": 7349 }, { "epoch": 5.384615384615385, "grad_norm": 2.328125, "learning_rate": 9.711115479703889e-05, "loss": 0.1615, "step": 7350 }, { "epoch": 5.385347985347986, "grad_norm": 0.93359375, "learning_rate": 9.711000129320957e-05, "loss": 0.1407, "step": 7351 }, { "epoch": 5.3860805860805865, "grad_norm": 0.875, "learning_rate": 9.710884756605641e-05, "loss": 0.1645, "step": 7352 }, { "epoch": 5.386813186813187, "grad_norm": 1.7734375, "learning_rate": 9.710769361558495e-05, "loss": 0.1295, "step": 7353 }, { "epoch": 5.387545787545788, "grad_norm": 0.9921875, "learning_rate": 9.710653944180075e-05, "loss": 0.1466, "step": 7354 }, { "epoch": 5.388278388278389, "grad_norm": 1.8203125, "learning_rate": 9.71053850447093e-05, "loss": 0.1617, "step": 7355 }, { "epoch": 5.389010989010989, "grad_norm": 0.83984375, "learning_rate": 9.710423042431617e-05, "loss": 0.1447, "step": 7356 }, { "epoch": 5.38974358974359, "grad_norm": 0.94140625, "learning_rate": 9.710307558062686e-05, "loss": 0.1523, "step": 7357 }, { "epoch": 5.390476190476191, "grad_norm": 0.796875, "learning_rate": 9.710192051364693e-05, "loss": 0.154, "step": 7358 }, { "epoch": 5.391208791208792, "grad_norm": 1.0703125, "learning_rate": 9.71007652233819e-05, "loss": 0.1063, "step": 7359 }, { "epoch": 5.391941391941392, "grad_norm": 0.94140625, "learning_rate": 9.709960970983728e-05, "loss": 0.1586, "step": 7360 }, { "epoch": 5.392673992673993, "grad_norm": 1.46875, "learning_rate": 9.709845397301865e-05, "loss": 0.1786, "step": 7361 }, { "epoch": 5.393406593406594, "grad_norm": 1.515625, "learning_rate": 9.709729801293153e-05, "loss": 0.1416, "step": 7362 }, { "epoch": 5.3941391941391945, "grad_norm": 0.765625, "learning_rate": 9.709614182958145e-05, "loss": 0.1338, "step": 7363 }, { "epoch": 5.394871794871795, "grad_norm": 1.046875, "learning_rate": 9.709498542297398e-05, "loss": 0.1274, "step": 7364 }, { "epoch": 5.395604395604396, "grad_norm": 0.83984375, "learning_rate": 9.709382879311462e-05, "loss": 0.1332, "step": 7365 }, { "epoch": 5.396336996336997, "grad_norm": 0.76953125, "learning_rate": 9.709267194000895e-05, "loss": 0.1559, "step": 7366 }, { "epoch": 5.397069597069597, "grad_norm": 1.1171875, "learning_rate": 9.709151486366247e-05, "loss": 0.1534, "step": 7367 }, { "epoch": 5.397802197802198, "grad_norm": 1.4140625, "learning_rate": 9.709035756408075e-05, "loss": 0.1517, "step": 7368 }, { "epoch": 5.398534798534799, "grad_norm": 1.3984375, "learning_rate": 9.708920004126936e-05, "loss": 0.2036, "step": 7369 }, { "epoch": 5.3992673992674, "grad_norm": 0.9765625, "learning_rate": 9.708804229523378e-05, "loss": 0.1626, "step": 7370 }, { "epoch": 5.4, "grad_norm": 1.0859375, "learning_rate": 9.708688432597962e-05, "loss": 0.1603, "step": 7371 }, { "epoch": 5.400732600732601, "grad_norm": 0.82421875, "learning_rate": 9.708572613351239e-05, "loss": 0.1478, "step": 7372 }, { "epoch": 5.401465201465202, "grad_norm": 1.0, "learning_rate": 9.708456771783766e-05, "loss": 0.1514, "step": 7373 }, { "epoch": 5.4021978021978025, "grad_norm": 0.9921875, "learning_rate": 9.708340907896098e-05, "loss": 0.1778, "step": 7374 }, { "epoch": 5.402930402930403, "grad_norm": 1.0625, "learning_rate": 9.708225021688788e-05, "loss": 0.1837, "step": 7375 }, { "epoch": 5.403663003663004, "grad_norm": 0.859375, "learning_rate": 9.708109113162394e-05, "loss": 0.1449, "step": 7376 }, { "epoch": 5.404395604395605, "grad_norm": 1.2109375, "learning_rate": 9.707993182317469e-05, "loss": 0.1299, "step": 7377 }, { "epoch": 5.4051282051282055, "grad_norm": 1.71875, "learning_rate": 9.70787722915457e-05, "loss": 0.1305, "step": 7378 }, { "epoch": 5.405860805860806, "grad_norm": 1.2421875, "learning_rate": 9.707761253674253e-05, "loss": 0.1545, "step": 7379 }, { "epoch": 5.406593406593407, "grad_norm": 1.1640625, "learning_rate": 9.707645255877071e-05, "loss": 0.1608, "step": 7380 }, { "epoch": 5.407326007326008, "grad_norm": 0.91015625, "learning_rate": 9.707529235763582e-05, "loss": 0.1463, "step": 7381 }, { "epoch": 5.408058608058608, "grad_norm": 0.72265625, "learning_rate": 9.707413193334343e-05, "loss": 0.129, "step": 7382 }, { "epoch": 5.408791208791209, "grad_norm": 1.1015625, "learning_rate": 9.707297128589908e-05, "loss": 0.1784, "step": 7383 }, { "epoch": 5.40952380952381, "grad_norm": 0.81640625, "learning_rate": 9.707181041530833e-05, "loss": 0.1202, "step": 7384 }, { "epoch": 5.410256410256411, "grad_norm": 0.953125, "learning_rate": 9.707064932157675e-05, "loss": 0.1793, "step": 7385 }, { "epoch": 5.410989010989011, "grad_norm": 1.0234375, "learning_rate": 9.70694880047099e-05, "loss": 0.1979, "step": 7386 }, { "epoch": 5.411721611721612, "grad_norm": 0.71484375, "learning_rate": 9.706832646471334e-05, "loss": 0.1088, "step": 7387 }, { "epoch": 5.412454212454213, "grad_norm": 1.2109375, "learning_rate": 9.706716470159264e-05, "loss": 0.1501, "step": 7388 }, { "epoch": 5.4131868131868135, "grad_norm": 0.88671875, "learning_rate": 9.706600271535338e-05, "loss": 0.189, "step": 7389 }, { "epoch": 5.413919413919414, "grad_norm": 0.73828125, "learning_rate": 9.70648405060011e-05, "loss": 0.1376, "step": 7390 }, { "epoch": 5.414652014652015, "grad_norm": 1.1171875, "learning_rate": 9.706367807354138e-05, "loss": 0.1927, "step": 7391 }, { "epoch": 5.415384615384616, "grad_norm": 0.91015625, "learning_rate": 9.706251541797982e-05, "loss": 0.1318, "step": 7392 }, { "epoch": 5.416117216117216, "grad_norm": 1.234375, "learning_rate": 9.706135253932194e-05, "loss": 0.1647, "step": 7393 }, { "epoch": 5.416849816849817, "grad_norm": 1.171875, "learning_rate": 9.706018943757333e-05, "loss": 0.1434, "step": 7394 }, { "epoch": 5.417582417582418, "grad_norm": 0.94921875, "learning_rate": 9.705902611273958e-05, "loss": 0.1344, "step": 7395 }, { "epoch": 5.418315018315019, "grad_norm": 0.58984375, "learning_rate": 9.705786256482626e-05, "loss": 0.1257, "step": 7396 }, { "epoch": 5.419047619047619, "grad_norm": 1.0, "learning_rate": 9.705669879383893e-05, "loss": 0.1415, "step": 7397 }, { "epoch": 5.41978021978022, "grad_norm": 0.703125, "learning_rate": 9.705553479978316e-05, "loss": 0.1472, "step": 7398 }, { "epoch": 5.420512820512821, "grad_norm": 0.74609375, "learning_rate": 9.705437058266454e-05, "loss": 0.128, "step": 7399 }, { "epoch": 5.4212454212454215, "grad_norm": 0.85546875, "learning_rate": 9.705320614248867e-05, "loss": 0.1429, "step": 7400 }, { "epoch": 5.421978021978022, "grad_norm": 0.43359375, "learning_rate": 9.705204147926108e-05, "loss": 0.1437, "step": 7401 }, { "epoch": 5.422710622710623, "grad_norm": 1.0, "learning_rate": 9.70508765929874e-05, "loss": 0.1429, "step": 7402 }, { "epoch": 5.423443223443224, "grad_norm": 0.90625, "learning_rate": 9.704971148367318e-05, "loss": 0.1436, "step": 7403 }, { "epoch": 5.4241758241758244, "grad_norm": 0.89453125, "learning_rate": 9.7048546151324e-05, "loss": 0.1391, "step": 7404 }, { "epoch": 5.424908424908425, "grad_norm": 0.84375, "learning_rate": 9.704738059594547e-05, "loss": 0.1207, "step": 7405 }, { "epoch": 5.425641025641026, "grad_norm": 0.82421875, "learning_rate": 9.704621481754314e-05, "loss": 0.1791, "step": 7406 }, { "epoch": 5.426373626373627, "grad_norm": 0.66015625, "learning_rate": 9.704504881612264e-05, "loss": 0.1557, "step": 7407 }, { "epoch": 5.427106227106227, "grad_norm": 0.84375, "learning_rate": 9.704388259168951e-05, "loss": 0.1572, "step": 7408 }, { "epoch": 5.427838827838828, "grad_norm": 1.171875, "learning_rate": 9.704271614424937e-05, "loss": 0.178, "step": 7409 }, { "epoch": 5.428571428571429, "grad_norm": 1.0078125, "learning_rate": 9.70415494738078e-05, "loss": 0.1638, "step": 7410 }, { "epoch": 5.4293040293040296, "grad_norm": 1.2734375, "learning_rate": 9.70403825803704e-05, "loss": 0.1164, "step": 7411 }, { "epoch": 5.43003663003663, "grad_norm": 0.875, "learning_rate": 9.703921546394271e-05, "loss": 0.137, "step": 7412 }, { "epoch": 5.430769230769231, "grad_norm": 1.0859375, "learning_rate": 9.703804812453039e-05, "loss": 0.138, "step": 7413 }, { "epoch": 5.431501831501832, "grad_norm": 0.984375, "learning_rate": 9.703688056213899e-05, "loss": 0.1307, "step": 7414 }, { "epoch": 5.4322344322344325, "grad_norm": 0.95703125, "learning_rate": 9.703571277677414e-05, "loss": 0.1626, "step": 7415 }, { "epoch": 5.432967032967033, "grad_norm": 1.4140625, "learning_rate": 9.703454476844138e-05, "loss": 0.1731, "step": 7416 }, { "epoch": 5.433699633699634, "grad_norm": 1.8828125, "learning_rate": 9.703337653714636e-05, "loss": 0.1668, "step": 7417 }, { "epoch": 5.434432234432235, "grad_norm": 1.2265625, "learning_rate": 9.703220808289465e-05, "loss": 0.1607, "step": 7418 }, { "epoch": 5.435164835164835, "grad_norm": 1.53125, "learning_rate": 9.703103940569187e-05, "loss": 0.1493, "step": 7419 }, { "epoch": 5.435897435897436, "grad_norm": 0.8125, "learning_rate": 9.70298705055436e-05, "loss": 0.1183, "step": 7420 }, { "epoch": 5.436630036630037, "grad_norm": 0.8671875, "learning_rate": 9.702870138245544e-05, "loss": 0.1333, "step": 7421 }, { "epoch": 5.437362637362638, "grad_norm": 1.3671875, "learning_rate": 9.7027532036433e-05, "loss": 0.1537, "step": 7422 }, { "epoch": 5.438095238095238, "grad_norm": 0.8671875, "learning_rate": 9.702636246748188e-05, "loss": 0.1525, "step": 7423 }, { "epoch": 5.438827838827839, "grad_norm": 0.8125, "learning_rate": 9.702519267560767e-05, "loss": 0.1364, "step": 7424 }, { "epoch": 5.43956043956044, "grad_norm": 1.1015625, "learning_rate": 9.702402266081601e-05, "loss": 0.1562, "step": 7425 }, { "epoch": 5.4402930402930405, "grad_norm": 0.6328125, "learning_rate": 9.702285242311249e-05, "loss": 0.1405, "step": 7426 }, { "epoch": 5.441025641025641, "grad_norm": 1.1171875, "learning_rate": 9.70216819625027e-05, "loss": 0.148, "step": 7427 }, { "epoch": 5.441758241758242, "grad_norm": 1.1328125, "learning_rate": 9.702051127899224e-05, "loss": 0.1763, "step": 7428 }, { "epoch": 5.442490842490843, "grad_norm": 1.390625, "learning_rate": 9.701934037258678e-05, "loss": 0.1431, "step": 7429 }, { "epoch": 5.443223443223443, "grad_norm": 1.0703125, "learning_rate": 9.701816924329185e-05, "loss": 0.1765, "step": 7430 }, { "epoch": 5.443956043956044, "grad_norm": 0.98828125, "learning_rate": 9.701699789111313e-05, "loss": 0.1579, "step": 7431 }, { "epoch": 5.444688644688645, "grad_norm": 1.359375, "learning_rate": 9.701582631605618e-05, "loss": 0.1676, "step": 7432 }, { "epoch": 5.445421245421246, "grad_norm": 1.4609375, "learning_rate": 9.701465451812663e-05, "loss": 0.1372, "step": 7433 }, { "epoch": 5.446153846153846, "grad_norm": 0.75390625, "learning_rate": 9.701348249733014e-05, "loss": 0.1382, "step": 7434 }, { "epoch": 5.446886446886447, "grad_norm": 1.28125, "learning_rate": 9.701231025367224e-05, "loss": 0.1502, "step": 7435 }, { "epoch": 5.447619047619048, "grad_norm": 0.68359375, "learning_rate": 9.701113778715862e-05, "loss": 0.113, "step": 7436 }, { "epoch": 5.4483516483516485, "grad_norm": 0.70703125, "learning_rate": 9.700996509779486e-05, "loss": 0.1244, "step": 7437 }, { "epoch": 5.449084249084249, "grad_norm": 0.79296875, "learning_rate": 9.700879218558661e-05, "loss": 0.1644, "step": 7438 }, { "epoch": 5.44981684981685, "grad_norm": 1.375, "learning_rate": 9.700761905053944e-05, "loss": 0.1342, "step": 7439 }, { "epoch": 5.450549450549451, "grad_norm": 1.0546875, "learning_rate": 9.700644569265901e-05, "loss": 0.1421, "step": 7440 }, { "epoch": 5.4512820512820515, "grad_norm": 1.4453125, "learning_rate": 9.700527211195094e-05, "loss": 0.1552, "step": 7441 }, { "epoch": 5.452014652014652, "grad_norm": 0.78515625, "learning_rate": 9.700409830842082e-05, "loss": 0.1399, "step": 7442 }, { "epoch": 5.452747252747253, "grad_norm": 1.515625, "learning_rate": 9.700292428207433e-05, "loss": 0.1704, "step": 7443 }, { "epoch": 5.453479853479854, "grad_norm": 0.83203125, "learning_rate": 9.700175003291706e-05, "loss": 0.16, "step": 7444 }, { "epoch": 5.454212454212454, "grad_norm": 0.84375, "learning_rate": 9.700057556095462e-05, "loss": 0.1449, "step": 7445 }, { "epoch": 5.454945054945055, "grad_norm": 0.78125, "learning_rate": 9.699940086619267e-05, "loss": 0.1376, "step": 7446 }, { "epoch": 5.455677655677656, "grad_norm": 2.203125, "learning_rate": 9.699822594863682e-05, "loss": 0.1427, "step": 7447 }, { "epoch": 5.456410256410257, "grad_norm": 1.4453125, "learning_rate": 9.69970508082927e-05, "loss": 0.1076, "step": 7448 }, { "epoch": 5.457142857142857, "grad_norm": 1.7734375, "learning_rate": 9.699587544516595e-05, "loss": 0.1789, "step": 7449 }, { "epoch": 5.457875457875458, "grad_norm": 0.8046875, "learning_rate": 9.69946998592622e-05, "loss": 0.1594, "step": 7450 }, { "epoch": 5.458608058608059, "grad_norm": 1.390625, "learning_rate": 9.699352405058709e-05, "loss": 0.1623, "step": 7451 }, { "epoch": 5.4593406593406595, "grad_norm": 0.875, "learning_rate": 9.699234801914623e-05, "loss": 0.1511, "step": 7452 }, { "epoch": 5.46007326007326, "grad_norm": 0.625, "learning_rate": 9.699117176494528e-05, "loss": 0.112, "step": 7453 }, { "epoch": 5.460805860805861, "grad_norm": 0.9296875, "learning_rate": 9.698999528798984e-05, "loss": 0.1543, "step": 7454 }, { "epoch": 5.461538461538462, "grad_norm": 1.1328125, "learning_rate": 9.69888185882856e-05, "loss": 0.141, "step": 7455 }, { "epoch": 5.462271062271062, "grad_norm": 0.94140625, "learning_rate": 9.698764166583816e-05, "loss": 0.1644, "step": 7456 }, { "epoch": 5.463003663003663, "grad_norm": 1.5, "learning_rate": 9.698646452065315e-05, "loss": 0.1265, "step": 7457 }, { "epoch": 5.463736263736264, "grad_norm": 1.359375, "learning_rate": 9.698528715273625e-05, "loss": 0.2157, "step": 7458 }, { "epoch": 5.464468864468865, "grad_norm": 0.94140625, "learning_rate": 9.698410956209307e-05, "loss": 0.1293, "step": 7459 }, { "epoch": 5.465201465201465, "grad_norm": 1.0390625, "learning_rate": 9.698293174872928e-05, "loss": 0.1487, "step": 7460 }, { "epoch": 5.465934065934066, "grad_norm": 1.71875, "learning_rate": 9.698175371265046e-05, "loss": 0.1551, "step": 7461 }, { "epoch": 5.466666666666667, "grad_norm": 1.1328125, "learning_rate": 9.698057545386235e-05, "loss": 0.1835, "step": 7462 }, { "epoch": 5.4673992673992675, "grad_norm": 0.859375, "learning_rate": 9.697939697237051e-05, "loss": 0.1507, "step": 7463 }, { "epoch": 5.468131868131868, "grad_norm": 0.8359375, "learning_rate": 9.697821826818063e-05, "loss": 0.1325, "step": 7464 }, { "epoch": 5.468864468864469, "grad_norm": 1.453125, "learning_rate": 9.697703934129834e-05, "loss": 0.1666, "step": 7465 }, { "epoch": 5.46959706959707, "grad_norm": 0.74609375, "learning_rate": 9.697586019172933e-05, "loss": 0.1356, "step": 7466 }, { "epoch": 5.47032967032967, "grad_norm": 1.125, "learning_rate": 9.697468081947919e-05, "loss": 0.1987, "step": 7467 }, { "epoch": 5.471062271062271, "grad_norm": 0.8671875, "learning_rate": 9.697350122455358e-05, "loss": 0.1358, "step": 7468 }, { "epoch": 5.471794871794872, "grad_norm": 0.640625, "learning_rate": 9.697232140695818e-05, "loss": 0.1215, "step": 7469 }, { "epoch": 5.472527472527473, "grad_norm": 1.015625, "learning_rate": 9.697114136669867e-05, "loss": 0.1367, "step": 7470 }, { "epoch": 5.473260073260073, "grad_norm": 1.078125, "learning_rate": 9.696996110378064e-05, "loss": 0.1546, "step": 7471 }, { "epoch": 5.473992673992674, "grad_norm": 1.2265625, "learning_rate": 9.696878061820976e-05, "loss": 0.1373, "step": 7472 }, { "epoch": 5.474725274725275, "grad_norm": 1.6875, "learning_rate": 9.696759990999171e-05, "loss": 0.1729, "step": 7473 }, { "epoch": 5.4754578754578755, "grad_norm": 0.890625, "learning_rate": 9.696641897913213e-05, "loss": 0.1916, "step": 7474 }, { "epoch": 5.476190476190476, "grad_norm": 1.171875, "learning_rate": 9.696523782563668e-05, "loss": 0.1258, "step": 7475 }, { "epoch": 5.476923076923077, "grad_norm": 0.8828125, "learning_rate": 9.696405644951104e-05, "loss": 0.1353, "step": 7476 }, { "epoch": 5.477655677655678, "grad_norm": 1.953125, "learning_rate": 9.696287485076083e-05, "loss": 0.1392, "step": 7477 }, { "epoch": 5.4783882783882785, "grad_norm": 1.28125, "learning_rate": 9.696169302939175e-05, "loss": 0.1882, "step": 7478 }, { "epoch": 5.479120879120879, "grad_norm": 0.94140625, "learning_rate": 9.696051098540944e-05, "loss": 0.1279, "step": 7479 }, { "epoch": 5.47985347985348, "grad_norm": 1.296875, "learning_rate": 9.695932871881957e-05, "loss": 0.1579, "step": 7480 }, { "epoch": 5.480586080586081, "grad_norm": 0.98046875, "learning_rate": 9.69581462296278e-05, "loss": 0.1269, "step": 7481 }, { "epoch": 5.481318681318681, "grad_norm": 1.3359375, "learning_rate": 9.69569635178398e-05, "loss": 0.1952, "step": 7482 }, { "epoch": 5.482051282051282, "grad_norm": 1.328125, "learning_rate": 9.695578058346124e-05, "loss": 0.1257, "step": 7483 }, { "epoch": 5.482783882783883, "grad_norm": 0.81640625, "learning_rate": 9.695459742649779e-05, "loss": 0.1447, "step": 7484 }, { "epoch": 5.483516483516484, "grad_norm": 1.2265625, "learning_rate": 9.695341404695512e-05, "loss": 0.1649, "step": 7485 }, { "epoch": 5.484249084249084, "grad_norm": 1.3046875, "learning_rate": 9.695223044483887e-05, "loss": 0.1594, "step": 7486 }, { "epoch": 5.484981684981685, "grad_norm": 1.015625, "learning_rate": 9.695104662015475e-05, "loss": 0.115, "step": 7487 }, { "epoch": 5.485714285714286, "grad_norm": 0.76171875, "learning_rate": 9.694986257290843e-05, "loss": 0.1467, "step": 7488 }, { "epoch": 5.4864468864468865, "grad_norm": 0.83203125, "learning_rate": 9.694867830310554e-05, "loss": 0.1719, "step": 7489 }, { "epoch": 5.487179487179487, "grad_norm": 0.875, "learning_rate": 9.694749381075182e-05, "loss": 0.1255, "step": 7490 }, { "epoch": 5.487912087912088, "grad_norm": 0.9375, "learning_rate": 9.694630909585287e-05, "loss": 0.1435, "step": 7491 }, { "epoch": 5.488644688644689, "grad_norm": 1.34375, "learning_rate": 9.694512415841443e-05, "loss": 0.1395, "step": 7492 }, { "epoch": 5.489377289377289, "grad_norm": 1.0703125, "learning_rate": 9.694393899844215e-05, "loss": 0.165, "step": 7493 }, { "epoch": 5.49010989010989, "grad_norm": 1.6015625, "learning_rate": 9.69427536159417e-05, "loss": 0.1455, "step": 7494 }, { "epoch": 5.490842490842491, "grad_norm": 2.09375, "learning_rate": 9.694156801091878e-05, "loss": 0.1402, "step": 7495 }, { "epoch": 5.491575091575092, "grad_norm": 2.15625, "learning_rate": 9.694038218337908e-05, "loss": 0.1388, "step": 7496 }, { "epoch": 5.492307692307692, "grad_norm": 1.390625, "learning_rate": 9.693919613332822e-05, "loss": 0.1584, "step": 7497 }, { "epoch": 5.493040293040293, "grad_norm": 1.328125, "learning_rate": 9.693800986077196e-05, "loss": 0.1598, "step": 7498 }, { "epoch": 5.493772893772894, "grad_norm": 0.84765625, "learning_rate": 9.693682336571595e-05, "loss": 0.1384, "step": 7499 }, { "epoch": 5.4945054945054945, "grad_norm": 1.609375, "learning_rate": 9.693563664816586e-05, "loss": 0.157, "step": 7500 }, { "epoch": 5.495238095238095, "grad_norm": 1.9609375, "learning_rate": 9.69344497081274e-05, "loss": 0.1453, "step": 7501 }, { "epoch": 5.495970695970696, "grad_norm": 1.4296875, "learning_rate": 9.693326254560624e-05, "loss": 0.1668, "step": 7502 }, { "epoch": 5.496703296703297, "grad_norm": 1.2578125, "learning_rate": 9.693207516060808e-05, "loss": 0.1272, "step": 7503 }, { "epoch": 5.4974358974358974, "grad_norm": 1.1796875, "learning_rate": 9.693088755313862e-05, "loss": 0.1369, "step": 7504 }, { "epoch": 5.498168498168498, "grad_norm": 1.2734375, "learning_rate": 9.692969972320353e-05, "loss": 0.1744, "step": 7505 }, { "epoch": 5.498901098901099, "grad_norm": 0.9453125, "learning_rate": 9.692851167080848e-05, "loss": 0.1245, "step": 7506 }, { "epoch": 5.4996336996337, "grad_norm": 1.5703125, "learning_rate": 9.692732339595922e-05, "loss": 0.1264, "step": 7507 }, { "epoch": 5.5003663003663, "grad_norm": 0.82421875, "learning_rate": 9.69261348986614e-05, "loss": 0.1287, "step": 7508 }, { "epoch": 5.501098901098901, "grad_norm": 0.60546875, "learning_rate": 9.692494617892073e-05, "loss": 0.1431, "step": 7509 }, { "epoch": 5.501831501831502, "grad_norm": 1.2421875, "learning_rate": 9.69237572367429e-05, "loss": 0.1828, "step": 7510 }, { "epoch": 5.5025641025641026, "grad_norm": 1.5078125, "learning_rate": 9.692256807213362e-05, "loss": 0.1161, "step": 7511 }, { "epoch": 5.503296703296703, "grad_norm": 1.25, "learning_rate": 9.692137868509857e-05, "loss": 0.1522, "step": 7512 }, { "epoch": 5.504029304029304, "grad_norm": 0.97265625, "learning_rate": 9.692018907564345e-05, "loss": 0.1534, "step": 7513 }, { "epoch": 5.504761904761905, "grad_norm": 0.734375, "learning_rate": 9.691899924377396e-05, "loss": 0.1365, "step": 7514 }, { "epoch": 5.5054945054945055, "grad_norm": 0.7421875, "learning_rate": 9.691780918949581e-05, "loss": 0.1131, "step": 7515 }, { "epoch": 5.506227106227106, "grad_norm": 0.90625, "learning_rate": 9.691661891281471e-05, "loss": 0.1537, "step": 7516 }, { "epoch": 5.506959706959707, "grad_norm": 1.1875, "learning_rate": 9.691542841373635e-05, "loss": 0.1681, "step": 7517 }, { "epoch": 5.507692307692308, "grad_norm": 1.5, "learning_rate": 9.691423769226642e-05, "loss": 0.1816, "step": 7518 }, { "epoch": 5.508424908424908, "grad_norm": 1.0546875, "learning_rate": 9.691304674841066e-05, "loss": 0.1569, "step": 7519 }, { "epoch": 5.509157509157509, "grad_norm": 0.92578125, "learning_rate": 9.691185558217472e-05, "loss": 0.1245, "step": 7520 }, { "epoch": 5.50989010989011, "grad_norm": 1.171875, "learning_rate": 9.691066419356438e-05, "loss": 0.1149, "step": 7521 }, { "epoch": 5.510622710622711, "grad_norm": 1.1171875, "learning_rate": 9.69094725825853e-05, "loss": 0.1375, "step": 7522 }, { "epoch": 5.511355311355311, "grad_norm": 0.86328125, "learning_rate": 9.69082807492432e-05, "loss": 0.1248, "step": 7523 }, { "epoch": 5.512087912087912, "grad_norm": 1.5859375, "learning_rate": 9.69070886935438e-05, "loss": 0.1813, "step": 7524 }, { "epoch": 5.512820512820513, "grad_norm": 1.5390625, "learning_rate": 9.69058964154928e-05, "loss": 0.1605, "step": 7525 }, { "epoch": 5.5135531135531135, "grad_norm": 1.765625, "learning_rate": 9.690470391509592e-05, "loss": 0.184, "step": 7526 }, { "epoch": 5.514285714285714, "grad_norm": 1.0546875, "learning_rate": 9.690351119235886e-05, "loss": 0.1646, "step": 7527 }, { "epoch": 5.515018315018315, "grad_norm": 1.3828125, "learning_rate": 9.690231824728736e-05, "loss": 0.162, "step": 7528 }, { "epoch": 5.515750915750916, "grad_norm": 0.6875, "learning_rate": 9.690112507988711e-05, "loss": 0.1241, "step": 7529 }, { "epoch": 5.516483516483516, "grad_norm": 0.75, "learning_rate": 9.689993169016383e-05, "loss": 0.12, "step": 7530 }, { "epoch": 5.517216117216117, "grad_norm": 0.890625, "learning_rate": 9.689873807812326e-05, "loss": 0.1562, "step": 7531 }, { "epoch": 5.517948717948718, "grad_norm": 0.953125, "learning_rate": 9.68975442437711e-05, "loss": 0.1252, "step": 7532 }, { "epoch": 5.518681318681319, "grad_norm": 1.203125, "learning_rate": 9.689635018711306e-05, "loss": 0.1617, "step": 7533 }, { "epoch": 5.519413919413919, "grad_norm": 0.8359375, "learning_rate": 9.689515590815489e-05, "loss": 0.1262, "step": 7534 }, { "epoch": 5.52014652014652, "grad_norm": 0.703125, "learning_rate": 9.68939614069023e-05, "loss": 0.169, "step": 7535 }, { "epoch": 5.520879120879121, "grad_norm": 0.84375, "learning_rate": 9.6892766683361e-05, "loss": 0.1534, "step": 7536 }, { "epoch": 5.5216117216117215, "grad_norm": 0.66796875, "learning_rate": 9.689157173753674e-05, "loss": 0.137, "step": 7537 }, { "epoch": 5.522344322344322, "grad_norm": 0.875, "learning_rate": 9.689037656943521e-05, "loss": 0.1409, "step": 7538 }, { "epoch": 5.523076923076923, "grad_norm": 0.9765625, "learning_rate": 9.688918117906218e-05, "loss": 0.1568, "step": 7539 }, { "epoch": 5.523809523809524, "grad_norm": 0.74609375, "learning_rate": 9.688798556642334e-05, "loss": 0.1139, "step": 7540 }, { "epoch": 5.5245421245421245, "grad_norm": 1.3984375, "learning_rate": 9.688678973152443e-05, "loss": 0.169, "step": 7541 }, { "epoch": 5.525274725274725, "grad_norm": 0.85546875, "learning_rate": 9.688559367437118e-05, "loss": 0.1331, "step": 7542 }, { "epoch": 5.526007326007326, "grad_norm": 1.7421875, "learning_rate": 9.688439739496934e-05, "loss": 0.1609, "step": 7543 }, { "epoch": 5.526739926739927, "grad_norm": 0.69921875, "learning_rate": 9.68832008933246e-05, "loss": 0.134, "step": 7544 }, { "epoch": 5.527472527472527, "grad_norm": 0.62109375, "learning_rate": 9.688200416944276e-05, "loss": 0.1394, "step": 7545 }, { "epoch": 5.528205128205128, "grad_norm": 0.6015625, "learning_rate": 9.688080722332946e-05, "loss": 0.1361, "step": 7546 }, { "epoch": 5.528937728937729, "grad_norm": 1.5859375, "learning_rate": 9.687961005499052e-05, "loss": 0.1211, "step": 7547 }, { "epoch": 5.52967032967033, "grad_norm": 0.99609375, "learning_rate": 9.687841266443163e-05, "loss": 0.1581, "step": 7548 }, { "epoch": 5.53040293040293, "grad_norm": 1.2734375, "learning_rate": 9.687721505165854e-05, "loss": 0.1392, "step": 7549 }, { "epoch": 5.531135531135531, "grad_norm": 1.2734375, "learning_rate": 9.6876017216677e-05, "loss": 0.1537, "step": 7550 }, { "epoch": 5.531868131868132, "grad_norm": 0.703125, "learning_rate": 9.687481915949274e-05, "loss": 0.1318, "step": 7551 }, { "epoch": 5.5326007326007325, "grad_norm": 1.96875, "learning_rate": 9.687362088011149e-05, "loss": 0.1595, "step": 7552 }, { "epoch": 5.533333333333333, "grad_norm": 0.9453125, "learning_rate": 9.687242237853898e-05, "loss": 0.1541, "step": 7553 }, { "epoch": 5.534065934065934, "grad_norm": 0.9453125, "learning_rate": 9.687122365478099e-05, "loss": 0.1438, "step": 7554 }, { "epoch": 5.534798534798535, "grad_norm": 0.67578125, "learning_rate": 9.687002470884326e-05, "loss": 0.1584, "step": 7555 }, { "epoch": 5.535531135531135, "grad_norm": 1.0234375, "learning_rate": 9.68688255407315e-05, "loss": 0.1421, "step": 7556 }, { "epoch": 5.536263736263736, "grad_norm": 0.77734375, "learning_rate": 9.686762615045147e-05, "loss": 0.1242, "step": 7557 }, { "epoch": 5.536996336996337, "grad_norm": 1.3046875, "learning_rate": 9.686642653800892e-05, "loss": 0.1605, "step": 7558 }, { "epoch": 5.537728937728938, "grad_norm": 1.0546875, "learning_rate": 9.68652267034096e-05, "loss": 0.1485, "step": 7559 }, { "epoch": 5.538461538461538, "grad_norm": 1.421875, "learning_rate": 9.686402664665928e-05, "loss": 0.1533, "step": 7560 }, { "epoch": 5.539194139194139, "grad_norm": 0.84375, "learning_rate": 9.686282636776368e-05, "loss": 0.1387, "step": 7561 }, { "epoch": 5.53992673992674, "grad_norm": 0.640625, "learning_rate": 9.686162586672855e-05, "loss": 0.1279, "step": 7562 }, { "epoch": 5.5406593406593405, "grad_norm": 0.63671875, "learning_rate": 9.686042514355966e-05, "loss": 0.1534, "step": 7563 }, { "epoch": 5.541391941391941, "grad_norm": 1.1015625, "learning_rate": 9.685922419826274e-05, "loss": 0.1364, "step": 7564 }, { "epoch": 5.542124542124542, "grad_norm": 0.984375, "learning_rate": 9.685802303084358e-05, "loss": 0.1212, "step": 7565 }, { "epoch": 5.542857142857143, "grad_norm": 1.1328125, "learning_rate": 9.685682164130788e-05, "loss": 0.1486, "step": 7566 }, { "epoch": 5.543589743589743, "grad_norm": 0.6796875, "learning_rate": 9.685562002966145e-05, "loss": 0.1516, "step": 7567 }, { "epoch": 5.544322344322344, "grad_norm": 1.2421875, "learning_rate": 9.685441819591004e-05, "loss": 0.1781, "step": 7568 }, { "epoch": 5.545054945054945, "grad_norm": 1.7421875, "learning_rate": 9.685321614005937e-05, "loss": 0.1888, "step": 7569 }, { "epoch": 5.545787545787546, "grad_norm": 0.9453125, "learning_rate": 9.685201386211523e-05, "loss": 0.1862, "step": 7570 }, { "epoch": 5.546520146520146, "grad_norm": 1.0625, "learning_rate": 9.685081136208339e-05, "loss": 0.1535, "step": 7571 }, { "epoch": 5.547252747252747, "grad_norm": 1.4921875, "learning_rate": 9.684960863996958e-05, "loss": 0.2157, "step": 7572 }, { "epoch": 5.547985347985348, "grad_norm": 0.58984375, "learning_rate": 9.684840569577959e-05, "loss": 0.1312, "step": 7573 }, { "epoch": 5.5487179487179485, "grad_norm": 0.81640625, "learning_rate": 9.684720252951916e-05, "loss": 0.1576, "step": 7574 }, { "epoch": 5.549450549450549, "grad_norm": 0.87109375, "learning_rate": 9.684599914119408e-05, "loss": 0.1399, "step": 7575 }, { "epoch": 5.55018315018315, "grad_norm": 1.1953125, "learning_rate": 9.68447955308101e-05, "loss": 0.1571, "step": 7576 }, { "epoch": 5.550915750915751, "grad_norm": 0.84375, "learning_rate": 9.6843591698373e-05, "loss": 0.1709, "step": 7577 }, { "epoch": 5.5516483516483515, "grad_norm": 0.7890625, "learning_rate": 9.684238764388854e-05, "loss": 0.1716, "step": 7578 }, { "epoch": 5.552380952380952, "grad_norm": 0.83984375, "learning_rate": 9.684118336736248e-05, "loss": 0.1552, "step": 7579 }, { "epoch": 5.553113553113553, "grad_norm": 0.86328125, "learning_rate": 9.683997886880059e-05, "loss": 0.1396, "step": 7580 }, { "epoch": 5.553846153846154, "grad_norm": 0.546875, "learning_rate": 9.683877414820867e-05, "loss": 0.1228, "step": 7581 }, { "epoch": 5.554578754578754, "grad_norm": 1.125, "learning_rate": 9.683756920559243e-05, "loss": 0.1773, "step": 7582 }, { "epoch": 5.555311355311355, "grad_norm": 0.85546875, "learning_rate": 9.683636404095773e-05, "loss": 0.1287, "step": 7583 }, { "epoch": 5.556043956043956, "grad_norm": 1.390625, "learning_rate": 9.683515865431026e-05, "loss": 0.1623, "step": 7584 }, { "epoch": 5.556776556776557, "grad_norm": 1.03125, "learning_rate": 9.683395304565586e-05, "loss": 0.1384, "step": 7585 }, { "epoch": 5.557509157509157, "grad_norm": 0.76171875, "learning_rate": 9.683274721500028e-05, "loss": 0.1465, "step": 7586 }, { "epoch": 5.558241758241758, "grad_norm": 0.8046875, "learning_rate": 9.683154116234929e-05, "loss": 0.1345, "step": 7587 }, { "epoch": 5.558974358974359, "grad_norm": 0.7578125, "learning_rate": 9.683033488770869e-05, "loss": 0.132, "step": 7588 }, { "epoch": 5.5597069597069595, "grad_norm": 0.64453125, "learning_rate": 9.682912839108424e-05, "loss": 0.1264, "step": 7589 }, { "epoch": 5.56043956043956, "grad_norm": 1.25, "learning_rate": 9.682792167248172e-05, "loss": 0.1791, "step": 7590 }, { "epoch": 5.561172161172161, "grad_norm": 0.7109375, "learning_rate": 9.68267147319069e-05, "loss": 0.1202, "step": 7591 }, { "epoch": 5.561904761904762, "grad_norm": 1.2265625, "learning_rate": 9.682550756936562e-05, "loss": 0.1673, "step": 7592 }, { "epoch": 5.562637362637362, "grad_norm": 0.99609375, "learning_rate": 9.68243001848636e-05, "loss": 0.1337, "step": 7593 }, { "epoch": 5.563369963369963, "grad_norm": 0.7578125, "learning_rate": 9.682309257840666e-05, "loss": 0.1269, "step": 7594 }, { "epoch": 5.564102564102564, "grad_norm": 0.82421875, "learning_rate": 9.682188475000058e-05, "loss": 0.1527, "step": 7595 }, { "epoch": 5.564835164835165, "grad_norm": 1.4140625, "learning_rate": 9.682067669965113e-05, "loss": 0.1683, "step": 7596 }, { "epoch": 5.565567765567765, "grad_norm": 1.03125, "learning_rate": 9.681946842736413e-05, "loss": 0.1598, "step": 7597 }, { "epoch": 5.566300366300366, "grad_norm": 1.125, "learning_rate": 9.681825993314534e-05, "loss": 0.1426, "step": 7598 }, { "epoch": 5.567032967032967, "grad_norm": 0.82421875, "learning_rate": 9.681705121700056e-05, "loss": 0.1511, "step": 7599 }, { "epoch": 5.5677655677655675, "grad_norm": 1.28125, "learning_rate": 9.681584227893559e-05, "loss": 0.1696, "step": 7600 }, { "epoch": 5.568498168498168, "grad_norm": 0.953125, "learning_rate": 9.681463311895621e-05, "loss": 0.1709, "step": 7601 }, { "epoch": 5.569230769230769, "grad_norm": 0.8125, "learning_rate": 9.681342373706822e-05, "loss": 0.175, "step": 7602 }, { "epoch": 5.56996336996337, "grad_norm": 1.0, "learning_rate": 9.68122141332774e-05, "loss": 0.1303, "step": 7603 }, { "epoch": 5.5706959706959704, "grad_norm": 1.6875, "learning_rate": 9.681100430758957e-05, "loss": 0.154, "step": 7604 }, { "epoch": 5.571428571428571, "grad_norm": 1.0078125, "learning_rate": 9.680979426001052e-05, "loss": 0.1577, "step": 7605 }, { "epoch": 5.572161172161172, "grad_norm": 0.953125, "learning_rate": 9.680858399054604e-05, "loss": 0.1261, "step": 7606 }, { "epoch": 5.572893772893773, "grad_norm": 0.92578125, "learning_rate": 9.680737349920194e-05, "loss": 0.1711, "step": 7607 }, { "epoch": 5.573626373626373, "grad_norm": 0.5703125, "learning_rate": 9.6806162785984e-05, "loss": 0.1339, "step": 7608 }, { "epoch": 5.574358974358974, "grad_norm": 0.6640625, "learning_rate": 9.680495185089804e-05, "loss": 0.1349, "step": 7609 }, { "epoch": 5.575091575091575, "grad_norm": 1.4609375, "learning_rate": 9.680374069394985e-05, "loss": 0.1739, "step": 7610 }, { "epoch": 5.5758241758241756, "grad_norm": 1.1171875, "learning_rate": 9.680252931514523e-05, "loss": 0.1738, "step": 7611 }, { "epoch": 5.576556776556776, "grad_norm": 1.0234375, "learning_rate": 9.680131771449001e-05, "loss": 0.1138, "step": 7612 }, { "epoch": 5.577289377289377, "grad_norm": 1.5546875, "learning_rate": 9.680010589198997e-05, "loss": 0.1616, "step": 7613 }, { "epoch": 5.578021978021978, "grad_norm": 0.74609375, "learning_rate": 9.679889384765092e-05, "loss": 0.1332, "step": 7614 }, { "epoch": 5.5787545787545785, "grad_norm": 1.671875, "learning_rate": 9.679768158147868e-05, "loss": 0.1664, "step": 7615 }, { "epoch": 5.579487179487179, "grad_norm": 1.25, "learning_rate": 9.679646909347904e-05, "loss": 0.1379, "step": 7616 }, { "epoch": 5.58021978021978, "grad_norm": 1.0234375, "learning_rate": 9.679525638365782e-05, "loss": 0.1794, "step": 7617 }, { "epoch": 5.580952380952381, "grad_norm": 1.1875, "learning_rate": 9.679404345202083e-05, "loss": 0.1511, "step": 7618 }, { "epoch": 5.581684981684981, "grad_norm": 0.75, "learning_rate": 9.679283029857386e-05, "loss": 0.1783, "step": 7619 }, { "epoch": 5.582417582417582, "grad_norm": 0.83203125, "learning_rate": 9.679161692332277e-05, "loss": 0.1597, "step": 7620 }, { "epoch": 5.583150183150183, "grad_norm": 1.3984375, "learning_rate": 9.679040332627334e-05, "loss": 0.1329, "step": 7621 }, { "epoch": 5.583882783882784, "grad_norm": 1.453125, "learning_rate": 9.678918950743139e-05, "loss": 0.1579, "step": 7622 }, { "epoch": 5.584615384615384, "grad_norm": 2.171875, "learning_rate": 9.678797546680274e-05, "loss": 0.18, "step": 7623 }, { "epoch": 5.585347985347985, "grad_norm": 0.62109375, "learning_rate": 9.678676120439321e-05, "loss": 0.1272, "step": 7624 }, { "epoch": 5.586080586080586, "grad_norm": 0.95703125, "learning_rate": 9.67855467202086e-05, "loss": 0.1734, "step": 7625 }, { "epoch": 5.5868131868131865, "grad_norm": 1.1171875, "learning_rate": 9.678433201425474e-05, "loss": 0.1241, "step": 7626 }, { "epoch": 5.587545787545787, "grad_norm": 2.453125, "learning_rate": 9.678311708653747e-05, "loss": 0.1879, "step": 7627 }, { "epoch": 5.588278388278388, "grad_norm": 2.625, "learning_rate": 9.678190193706257e-05, "loss": 0.1971, "step": 7628 }, { "epoch": 5.589010989010989, "grad_norm": 1.9765625, "learning_rate": 9.678068656583589e-05, "loss": 0.1581, "step": 7629 }, { "epoch": 5.589743589743589, "grad_norm": 2.109375, "learning_rate": 9.677947097286325e-05, "loss": 0.1769, "step": 7630 }, { "epoch": 5.59047619047619, "grad_norm": 1.2265625, "learning_rate": 9.677825515815048e-05, "loss": 0.1721, "step": 7631 }, { "epoch": 5.591208791208791, "grad_norm": 1.0546875, "learning_rate": 9.677703912170338e-05, "loss": 0.1647, "step": 7632 }, { "epoch": 5.591941391941392, "grad_norm": 0.984375, "learning_rate": 9.677582286352781e-05, "loss": 0.1345, "step": 7633 }, { "epoch": 5.592673992673992, "grad_norm": 0.8671875, "learning_rate": 9.677460638362958e-05, "loss": 0.1692, "step": 7634 }, { "epoch": 5.593406593406593, "grad_norm": 1.15625, "learning_rate": 9.67733896820145e-05, "loss": 0.1481, "step": 7635 }, { "epoch": 5.594139194139194, "grad_norm": 0.8046875, "learning_rate": 9.677217275868845e-05, "loss": 0.1251, "step": 7636 }, { "epoch": 5.5948717948717945, "grad_norm": 0.96484375, "learning_rate": 9.677095561365721e-05, "loss": 0.1653, "step": 7637 }, { "epoch": 5.595604395604395, "grad_norm": 0.8828125, "learning_rate": 9.676973824692664e-05, "loss": 0.1408, "step": 7638 }, { "epoch": 5.596336996336996, "grad_norm": 0.64453125, "learning_rate": 9.676852065850256e-05, "loss": 0.15, "step": 7639 }, { "epoch": 5.597069597069597, "grad_norm": 0.71484375, "learning_rate": 9.67673028483908e-05, "loss": 0.1639, "step": 7640 }, { "epoch": 5.5978021978021975, "grad_norm": 0.91796875, "learning_rate": 9.676608481659723e-05, "loss": 0.188, "step": 7641 }, { "epoch": 5.598534798534798, "grad_norm": 0.92578125, "learning_rate": 9.676486656312765e-05, "loss": 0.1723, "step": 7642 }, { "epoch": 5.599267399267399, "grad_norm": 1.2265625, "learning_rate": 9.676364808798788e-05, "loss": 0.1479, "step": 7643 }, { "epoch": 5.6, "grad_norm": 1.1328125, "learning_rate": 9.676242939118382e-05, "loss": 0.1682, "step": 7644 }, { "epoch": 5.6007326007326, "grad_norm": 0.78125, "learning_rate": 9.676121047272126e-05, "loss": 0.1522, "step": 7645 }, { "epoch": 5.601465201465201, "grad_norm": 0.79296875, "learning_rate": 9.675999133260605e-05, "loss": 0.1251, "step": 7646 }, { "epoch": 5.602197802197802, "grad_norm": 0.9453125, "learning_rate": 9.675877197084403e-05, "loss": 0.1285, "step": 7647 }, { "epoch": 5.602930402930403, "grad_norm": 1.0078125, "learning_rate": 9.675755238744105e-05, "loss": 0.1579, "step": 7648 }, { "epoch": 5.603663003663003, "grad_norm": 1.5, "learning_rate": 9.675633258240295e-05, "loss": 0.1363, "step": 7649 }, { "epoch": 5.604395604395604, "grad_norm": 0.70703125, "learning_rate": 9.67551125557356e-05, "loss": 0.1197, "step": 7650 }, { "epoch": 5.605128205128205, "grad_norm": 0.89453125, "learning_rate": 9.67538923074448e-05, "loss": 0.1537, "step": 7651 }, { "epoch": 5.6058608058608055, "grad_norm": 1.0859375, "learning_rate": 9.675267183753641e-05, "loss": 0.1429, "step": 7652 }, { "epoch": 5.606593406593406, "grad_norm": 1.0546875, "learning_rate": 9.67514511460163e-05, "loss": 0.1236, "step": 7653 }, { "epoch": 5.607326007326007, "grad_norm": 0.88671875, "learning_rate": 9.675023023289031e-05, "loss": 0.1547, "step": 7654 }, { "epoch": 5.608058608058608, "grad_norm": 1.125, "learning_rate": 9.674900909816426e-05, "loss": 0.1662, "step": 7655 }, { "epoch": 5.608791208791208, "grad_norm": 1.3046875, "learning_rate": 9.674778774184406e-05, "loss": 0.1639, "step": 7656 }, { "epoch": 5.609523809523809, "grad_norm": 1.09375, "learning_rate": 9.674656616393549e-05, "loss": 0.1302, "step": 7657 }, { "epoch": 5.61025641025641, "grad_norm": 0.953125, "learning_rate": 9.674534436444445e-05, "loss": 0.1558, "step": 7658 }, { "epoch": 5.610989010989011, "grad_norm": 0.94140625, "learning_rate": 9.67441223433768e-05, "loss": 0.1634, "step": 7659 }, { "epoch": 5.611721611721611, "grad_norm": 0.8671875, "learning_rate": 9.674290010073836e-05, "loss": 0.1401, "step": 7660 }, { "epoch": 5.612454212454212, "grad_norm": 1.046875, "learning_rate": 9.674167763653502e-05, "loss": 0.146, "step": 7661 }, { "epoch": 5.613186813186813, "grad_norm": 1.015625, "learning_rate": 9.674045495077262e-05, "loss": 0.1651, "step": 7662 }, { "epoch": 5.6139194139194135, "grad_norm": 0.9921875, "learning_rate": 9.673923204345702e-05, "loss": 0.1384, "step": 7663 }, { "epoch": 5.614652014652014, "grad_norm": 1.0859375, "learning_rate": 9.673800891459407e-05, "loss": 0.1517, "step": 7664 }, { "epoch": 5.615384615384615, "grad_norm": 1.015625, "learning_rate": 9.673678556418964e-05, "loss": 0.124, "step": 7665 }, { "epoch": 5.616117216117216, "grad_norm": 1.3359375, "learning_rate": 9.673556199224961e-05, "loss": 0.131, "step": 7666 }, { "epoch": 5.616849816849816, "grad_norm": 0.90234375, "learning_rate": 9.673433819877981e-05, "loss": 0.1743, "step": 7667 }, { "epoch": 5.617582417582417, "grad_norm": 0.5234375, "learning_rate": 9.673311418378612e-05, "loss": 0.1367, "step": 7668 }, { "epoch": 5.618315018315018, "grad_norm": 1.0, "learning_rate": 9.67318899472744e-05, "loss": 0.1139, "step": 7669 }, { "epoch": 5.619047619047619, "grad_norm": 0.82421875, "learning_rate": 9.673066548925052e-05, "loss": 0.1413, "step": 7670 }, { "epoch": 5.619780219780219, "grad_norm": 0.8125, "learning_rate": 9.672944080972037e-05, "loss": 0.1438, "step": 7671 }, { "epoch": 5.62051282051282, "grad_norm": 0.61328125, "learning_rate": 9.672821590868976e-05, "loss": 0.1177, "step": 7672 }, { "epoch": 5.621245421245421, "grad_norm": 0.96875, "learning_rate": 9.67269907861646e-05, "loss": 0.1542, "step": 7673 }, { "epoch": 5.6219780219780215, "grad_norm": 1.125, "learning_rate": 9.672576544215076e-05, "loss": 0.1245, "step": 7674 }, { "epoch": 5.622710622710622, "grad_norm": 0.9765625, "learning_rate": 9.672453987665409e-05, "loss": 0.1419, "step": 7675 }, { "epoch": 5.623443223443223, "grad_norm": 1.1328125, "learning_rate": 9.672331408968048e-05, "loss": 0.1637, "step": 7676 }, { "epoch": 5.624175824175824, "grad_norm": 0.75390625, "learning_rate": 9.67220880812358e-05, "loss": 0.1712, "step": 7677 }, { "epoch": 5.6249084249084245, "grad_norm": 0.875, "learning_rate": 9.672086185132591e-05, "loss": 0.1371, "step": 7678 }, { "epoch": 5.625641025641025, "grad_norm": 1.2109375, "learning_rate": 9.671963539995674e-05, "loss": 0.1758, "step": 7679 }, { "epoch": 5.626373626373626, "grad_norm": 1.0390625, "learning_rate": 9.671840872713407e-05, "loss": 0.142, "step": 7680 }, { "epoch": 5.627106227106227, "grad_norm": 1.3046875, "learning_rate": 9.671718183286386e-05, "loss": 0.1496, "step": 7681 }, { "epoch": 5.627838827838827, "grad_norm": 0.8984375, "learning_rate": 9.671595471715195e-05, "loss": 0.1663, "step": 7682 }, { "epoch": 5.628571428571428, "grad_norm": 0.96484375, "learning_rate": 9.671472738000423e-05, "loss": 0.1573, "step": 7683 }, { "epoch": 5.629304029304029, "grad_norm": 0.91015625, "learning_rate": 9.671349982142658e-05, "loss": 0.1313, "step": 7684 }, { "epoch": 5.63003663003663, "grad_norm": 1.3359375, "learning_rate": 9.67122720414249e-05, "loss": 0.1617, "step": 7685 }, { "epoch": 5.63076923076923, "grad_norm": 0.8515625, "learning_rate": 9.671104404000503e-05, "loss": 0.1308, "step": 7686 }, { "epoch": 5.631501831501831, "grad_norm": 0.6484375, "learning_rate": 9.670981581717289e-05, "loss": 0.1252, "step": 7687 }, { "epoch": 5.632234432234432, "grad_norm": 1.140625, "learning_rate": 9.670858737293434e-05, "loss": 0.1502, "step": 7688 }, { "epoch": 5.6329670329670325, "grad_norm": 0.56640625, "learning_rate": 9.670735870729529e-05, "loss": 0.1466, "step": 7689 }, { "epoch": 5.633699633699633, "grad_norm": 0.66015625, "learning_rate": 9.670612982026162e-05, "loss": 0.131, "step": 7690 }, { "epoch": 5.634432234432234, "grad_norm": 0.7890625, "learning_rate": 9.670490071183921e-05, "loss": 0.1323, "step": 7691 }, { "epoch": 5.635164835164835, "grad_norm": 1.0703125, "learning_rate": 9.670367138203395e-05, "loss": 0.1645, "step": 7692 }, { "epoch": 5.635897435897435, "grad_norm": 0.70703125, "learning_rate": 9.670244183085173e-05, "loss": 0.1221, "step": 7693 }, { "epoch": 5.636630036630036, "grad_norm": 0.70703125, "learning_rate": 9.670121205829845e-05, "loss": 0.1206, "step": 7694 }, { "epoch": 5.637362637362637, "grad_norm": 1.0078125, "learning_rate": 9.669998206437999e-05, "loss": 0.1722, "step": 7695 }, { "epoch": 5.638095238095238, "grad_norm": 0.5078125, "learning_rate": 9.669875184910225e-05, "loss": 0.1147, "step": 7696 }, { "epoch": 5.638827838827838, "grad_norm": 0.73828125, "learning_rate": 9.669752141247114e-05, "loss": 0.1369, "step": 7697 }, { "epoch": 5.639560439560439, "grad_norm": 0.8828125, "learning_rate": 9.669629075449253e-05, "loss": 0.1508, "step": 7698 }, { "epoch": 5.64029304029304, "grad_norm": 1.0859375, "learning_rate": 9.66950598751723e-05, "loss": 0.1474, "step": 7699 }, { "epoch": 5.641025641025641, "grad_norm": 1.015625, "learning_rate": 9.66938287745164e-05, "loss": 0.1457, "step": 7700 }, { "epoch": 5.641758241758242, "grad_norm": 1.0, "learning_rate": 9.66925974525307e-05, "loss": 0.114, "step": 7701 }, { "epoch": 5.642490842490843, "grad_norm": 1.140625, "learning_rate": 9.669136590922108e-05, "loss": 0.1537, "step": 7702 }, { "epoch": 5.643223443223444, "grad_norm": 1.0078125, "learning_rate": 9.669013414459349e-05, "loss": 0.1703, "step": 7703 }, { "epoch": 5.643956043956044, "grad_norm": 1.1875, "learning_rate": 9.66889021586538e-05, "loss": 0.1401, "step": 7704 }, { "epoch": 5.644688644688645, "grad_norm": 1.0390625, "learning_rate": 9.668766995140791e-05, "loss": 0.1194, "step": 7705 }, { "epoch": 5.645421245421246, "grad_norm": 0.73046875, "learning_rate": 9.668643752286173e-05, "loss": 0.1389, "step": 7706 }, { "epoch": 5.6461538461538465, "grad_norm": 1.234375, "learning_rate": 9.668520487302118e-05, "loss": 0.1803, "step": 7707 }, { "epoch": 5.646886446886447, "grad_norm": 1.0625, "learning_rate": 9.668397200189212e-05, "loss": 0.1406, "step": 7708 }, { "epoch": 5.647619047619048, "grad_norm": 1.28125, "learning_rate": 9.668273890948052e-05, "loss": 0.1601, "step": 7709 }, { "epoch": 5.648351648351649, "grad_norm": 1.4296875, "learning_rate": 9.668150559579225e-05, "loss": 0.1265, "step": 7710 }, { "epoch": 5.649084249084249, "grad_norm": 0.6796875, "learning_rate": 9.668027206083322e-05, "loss": 0.1286, "step": 7711 }, { "epoch": 5.64981684981685, "grad_norm": 1.1015625, "learning_rate": 9.667903830460935e-05, "loss": 0.1472, "step": 7712 }, { "epoch": 5.650549450549451, "grad_norm": 1.3984375, "learning_rate": 9.667780432712653e-05, "loss": 0.1263, "step": 7713 }, { "epoch": 5.651282051282052, "grad_norm": 1.1953125, "learning_rate": 9.667657012839071e-05, "loss": 0.1456, "step": 7714 }, { "epoch": 5.652014652014652, "grad_norm": 1.2578125, "learning_rate": 9.667533570840779e-05, "loss": 0.1437, "step": 7715 }, { "epoch": 5.652747252747253, "grad_norm": 0.53515625, "learning_rate": 9.667410106718368e-05, "loss": 0.1312, "step": 7716 }, { "epoch": 5.653479853479854, "grad_norm": 0.75390625, "learning_rate": 9.667286620472427e-05, "loss": 0.129, "step": 7717 }, { "epoch": 5.6542124542124546, "grad_norm": 0.74609375, "learning_rate": 9.66716311210355e-05, "loss": 0.1393, "step": 7718 }, { "epoch": 5.654945054945055, "grad_norm": 1.5859375, "learning_rate": 9.667039581612332e-05, "loss": 0.1623, "step": 7719 }, { "epoch": 5.655677655677656, "grad_norm": 1.6953125, "learning_rate": 9.666916028999359e-05, "loss": 0.1393, "step": 7720 }, { "epoch": 5.656410256410257, "grad_norm": 1.2109375, "learning_rate": 9.666792454265226e-05, "loss": 0.1119, "step": 7721 }, { "epoch": 5.6571428571428575, "grad_norm": 1.375, "learning_rate": 9.666668857410526e-05, "loss": 0.1318, "step": 7722 }, { "epoch": 5.657875457875458, "grad_norm": 1.046875, "learning_rate": 9.666545238435849e-05, "loss": 0.1563, "step": 7723 }, { "epoch": 5.658608058608059, "grad_norm": 0.6015625, "learning_rate": 9.666421597341788e-05, "loss": 0.1352, "step": 7724 }, { "epoch": 5.65934065934066, "grad_norm": 1.3671875, "learning_rate": 9.666297934128936e-05, "loss": 0.1518, "step": 7725 }, { "epoch": 5.66007326007326, "grad_norm": 1.53125, "learning_rate": 9.666174248797886e-05, "loss": 0.1706, "step": 7726 }, { "epoch": 5.660805860805861, "grad_norm": 1.3984375, "learning_rate": 9.666050541349228e-05, "loss": 0.1854, "step": 7727 }, { "epoch": 5.661538461538462, "grad_norm": 0.83203125, "learning_rate": 9.665926811783558e-05, "loss": 0.1159, "step": 7728 }, { "epoch": 5.662271062271063, "grad_norm": 1.4609375, "learning_rate": 9.665803060101468e-05, "loss": 0.18, "step": 7729 }, { "epoch": 5.663003663003663, "grad_norm": 1.2890625, "learning_rate": 9.665679286303548e-05, "loss": 0.1466, "step": 7730 }, { "epoch": 5.663736263736264, "grad_norm": 1.5546875, "learning_rate": 9.665555490390394e-05, "loss": 0.1119, "step": 7731 }, { "epoch": 5.664468864468865, "grad_norm": 0.83203125, "learning_rate": 9.665431672362599e-05, "loss": 0.128, "step": 7732 }, { "epoch": 5.6652014652014655, "grad_norm": 0.98046875, "learning_rate": 9.665307832220756e-05, "loss": 0.1624, "step": 7733 }, { "epoch": 5.665934065934066, "grad_norm": 1.2265625, "learning_rate": 9.665183969965458e-05, "loss": 0.1641, "step": 7734 }, { "epoch": 5.666666666666667, "grad_norm": 1.1484375, "learning_rate": 9.665060085597299e-05, "loss": 0.1532, "step": 7735 }, { "epoch": 5.667399267399268, "grad_norm": 1.171875, "learning_rate": 9.664936179116871e-05, "loss": 0.1997, "step": 7736 }, { "epoch": 5.668131868131868, "grad_norm": 0.70703125, "learning_rate": 9.664812250524769e-05, "loss": 0.1307, "step": 7737 }, { "epoch": 5.668864468864469, "grad_norm": 1.2578125, "learning_rate": 9.664688299821586e-05, "loss": 0.1227, "step": 7738 }, { "epoch": 5.66959706959707, "grad_norm": 1.1328125, "learning_rate": 9.664564327007918e-05, "loss": 0.157, "step": 7739 }, { "epoch": 5.670329670329671, "grad_norm": 1.4296875, "learning_rate": 9.664440332084355e-05, "loss": 0.1619, "step": 7740 }, { "epoch": 5.671062271062271, "grad_norm": 1.1953125, "learning_rate": 9.664316315051496e-05, "loss": 0.1666, "step": 7741 }, { "epoch": 5.671794871794872, "grad_norm": 1.0234375, "learning_rate": 9.664192275909933e-05, "loss": 0.158, "step": 7742 }, { "epoch": 5.672527472527473, "grad_norm": 0.75, "learning_rate": 9.664068214660257e-05, "loss": 0.1593, "step": 7743 }, { "epoch": 5.6732600732600735, "grad_norm": 1.2265625, "learning_rate": 9.663944131303068e-05, "loss": 0.147, "step": 7744 }, { "epoch": 5.673992673992674, "grad_norm": 0.8359375, "learning_rate": 9.663820025838957e-05, "loss": 0.143, "step": 7745 }, { "epoch": 5.674725274725275, "grad_norm": 1.140625, "learning_rate": 9.663695898268521e-05, "loss": 0.1425, "step": 7746 }, { "epoch": 5.675457875457876, "grad_norm": 0.7734375, "learning_rate": 9.663571748592352e-05, "loss": 0.1312, "step": 7747 }, { "epoch": 5.6761904761904765, "grad_norm": 1.1328125, "learning_rate": 9.663447576811048e-05, "loss": 0.1451, "step": 7748 }, { "epoch": 5.676923076923077, "grad_norm": 0.84375, "learning_rate": 9.663323382925199e-05, "loss": 0.1204, "step": 7749 }, { "epoch": 5.677655677655678, "grad_norm": 1.0859375, "learning_rate": 9.663199166935406e-05, "loss": 0.1481, "step": 7750 }, { "epoch": 5.678388278388279, "grad_norm": 0.7734375, "learning_rate": 9.66307492884226e-05, "loss": 0.1349, "step": 7751 }, { "epoch": 5.679120879120879, "grad_norm": 1.046875, "learning_rate": 9.662950668646359e-05, "loss": 0.1906, "step": 7752 }, { "epoch": 5.67985347985348, "grad_norm": 0.91796875, "learning_rate": 9.662826386348295e-05, "loss": 0.1263, "step": 7753 }, { "epoch": 5.680586080586081, "grad_norm": 1.015625, "learning_rate": 9.662702081948667e-05, "loss": 0.137, "step": 7754 }, { "epoch": 5.681318681318682, "grad_norm": 0.640625, "learning_rate": 9.662577755448069e-05, "loss": 0.1541, "step": 7755 }, { "epoch": 5.682051282051282, "grad_norm": 0.83203125, "learning_rate": 9.662453406847097e-05, "loss": 0.1583, "step": 7756 }, { "epoch": 5.682783882783883, "grad_norm": 1.046875, "learning_rate": 9.662329036146346e-05, "loss": 0.1306, "step": 7757 }, { "epoch": 5.683516483516484, "grad_norm": 1.0703125, "learning_rate": 9.662204643346411e-05, "loss": 0.1753, "step": 7758 }, { "epoch": 5.6842490842490845, "grad_norm": 1.1953125, "learning_rate": 9.662080228447892e-05, "loss": 0.1768, "step": 7759 }, { "epoch": 5.684981684981685, "grad_norm": 1.21875, "learning_rate": 9.66195579145138e-05, "loss": 0.1543, "step": 7760 }, { "epoch": 5.685714285714286, "grad_norm": 0.921875, "learning_rate": 9.661831332357476e-05, "loss": 0.1485, "step": 7761 }, { "epoch": 5.686446886446887, "grad_norm": 1.0390625, "learning_rate": 9.661706851166773e-05, "loss": 0.1451, "step": 7762 }, { "epoch": 5.687179487179487, "grad_norm": 0.734375, "learning_rate": 9.661582347879871e-05, "loss": 0.111, "step": 7763 }, { "epoch": 5.687912087912088, "grad_norm": 1.4921875, "learning_rate": 9.66145782249736e-05, "loss": 0.1643, "step": 7764 }, { "epoch": 5.688644688644689, "grad_norm": 0.77734375, "learning_rate": 9.661333275019844e-05, "loss": 0.1233, "step": 7765 }, { "epoch": 5.68937728937729, "grad_norm": 0.640625, "learning_rate": 9.661208705447913e-05, "loss": 0.1341, "step": 7766 }, { "epoch": 5.69010989010989, "grad_norm": 0.67578125, "learning_rate": 9.661084113782172e-05, "loss": 0.1201, "step": 7767 }, { "epoch": 5.690842490842491, "grad_norm": 0.9609375, "learning_rate": 9.66095950002321e-05, "loss": 0.1267, "step": 7768 }, { "epoch": 5.691575091575092, "grad_norm": 0.64453125, "learning_rate": 9.660834864171628e-05, "loss": 0.1136, "step": 7769 }, { "epoch": 5.6923076923076925, "grad_norm": 0.890625, "learning_rate": 9.660710206228024e-05, "loss": 0.1649, "step": 7770 }, { "epoch": 5.693040293040293, "grad_norm": 1.1328125, "learning_rate": 9.660585526192991e-05, "loss": 0.1708, "step": 7771 }, { "epoch": 5.693772893772894, "grad_norm": 1.5, "learning_rate": 9.660460824067132e-05, "loss": 0.1803, "step": 7772 }, { "epoch": 5.694505494505495, "grad_norm": 1.7890625, "learning_rate": 9.660336099851038e-05, "loss": 0.1583, "step": 7773 }, { "epoch": 5.695238095238095, "grad_norm": 2.453125, "learning_rate": 9.660211353545314e-05, "loss": 0.1521, "step": 7774 }, { "epoch": 5.695970695970696, "grad_norm": 2.921875, "learning_rate": 9.660086585150553e-05, "loss": 0.1911, "step": 7775 }, { "epoch": 5.696703296703297, "grad_norm": 1.3125, "learning_rate": 9.659961794667354e-05, "loss": 0.153, "step": 7776 }, { "epoch": 5.697435897435898, "grad_norm": 1.9921875, "learning_rate": 9.659836982096315e-05, "loss": 0.163, "step": 7777 }, { "epoch": 5.698168498168498, "grad_norm": 0.98046875, "learning_rate": 9.65971214743803e-05, "loss": 0.142, "step": 7778 }, { "epoch": 5.698901098901099, "grad_norm": 1.4609375, "learning_rate": 9.659587290693106e-05, "loss": 0.1654, "step": 7779 }, { "epoch": 5.6996336996337, "grad_norm": 2.25, "learning_rate": 9.659462411862133e-05, "loss": 0.1855, "step": 7780 }, { "epoch": 5.7003663003663005, "grad_norm": 2.703125, "learning_rate": 9.659337510945713e-05, "loss": 0.1828, "step": 7781 }, { "epoch": 5.701098901098901, "grad_norm": 2.03125, "learning_rate": 9.659212587944444e-05, "loss": 0.1655, "step": 7782 }, { "epoch": 5.701831501831502, "grad_norm": 1.4609375, "learning_rate": 9.659087642858924e-05, "loss": 0.1708, "step": 7783 }, { "epoch": 5.702564102564103, "grad_norm": 0.75390625, "learning_rate": 9.658962675689753e-05, "loss": 0.1212, "step": 7784 }, { "epoch": 5.7032967032967035, "grad_norm": 1.0859375, "learning_rate": 9.658837686437527e-05, "loss": 0.1603, "step": 7785 }, { "epoch": 5.704029304029304, "grad_norm": 0.6640625, "learning_rate": 9.658712675102848e-05, "loss": 0.1433, "step": 7786 }, { "epoch": 5.704761904761905, "grad_norm": 1.1953125, "learning_rate": 9.658587641686313e-05, "loss": 0.1251, "step": 7787 }, { "epoch": 5.705494505494506, "grad_norm": 1.3203125, "learning_rate": 9.658462586188523e-05, "loss": 0.1432, "step": 7788 }, { "epoch": 5.706227106227106, "grad_norm": 2.578125, "learning_rate": 9.658337508610075e-05, "loss": 0.1703, "step": 7789 }, { "epoch": 5.706959706959707, "grad_norm": 0.984375, "learning_rate": 9.65821240895157e-05, "loss": 0.1584, "step": 7790 }, { "epoch": 5.707692307692308, "grad_norm": 0.67578125, "learning_rate": 9.658087287213605e-05, "loss": 0.1272, "step": 7791 }, { "epoch": 5.708424908424909, "grad_norm": 1.0234375, "learning_rate": 9.657962143396783e-05, "loss": 0.1574, "step": 7792 }, { "epoch": 5.709157509157509, "grad_norm": 1.15625, "learning_rate": 9.6578369775017e-05, "loss": 0.1755, "step": 7793 }, { "epoch": 5.70989010989011, "grad_norm": 0.890625, "learning_rate": 9.657711789528958e-05, "loss": 0.1427, "step": 7794 }, { "epoch": 5.710622710622711, "grad_norm": 1.84375, "learning_rate": 9.657586579479156e-05, "loss": 0.1655, "step": 7795 }, { "epoch": 5.7113553113553115, "grad_norm": 1.203125, "learning_rate": 9.657461347352894e-05, "loss": 0.1419, "step": 7796 }, { "epoch": 5.712087912087912, "grad_norm": 1.1015625, "learning_rate": 9.657336093150773e-05, "loss": 0.1588, "step": 7797 }, { "epoch": 5.712820512820513, "grad_norm": 0.94140625, "learning_rate": 9.657210816873392e-05, "loss": 0.124, "step": 7798 }, { "epoch": 5.713553113553114, "grad_norm": 0.91015625, "learning_rate": 9.65708551852135e-05, "loss": 0.1551, "step": 7799 }, { "epoch": 5.714285714285714, "grad_norm": 1.2109375, "learning_rate": 9.656960198095252e-05, "loss": 0.1475, "step": 7800 }, { "epoch": 5.715018315018315, "grad_norm": 1.3515625, "learning_rate": 9.656834855595693e-05, "loss": 0.1478, "step": 7801 }, { "epoch": 5.715750915750916, "grad_norm": 1.1171875, "learning_rate": 9.656709491023276e-05, "loss": 0.1366, "step": 7802 }, { "epoch": 5.716483516483517, "grad_norm": 1.0390625, "learning_rate": 9.656584104378601e-05, "loss": 0.1764, "step": 7803 }, { "epoch": 5.717216117216117, "grad_norm": 0.8125, "learning_rate": 9.65645869566227e-05, "loss": 0.144, "step": 7804 }, { "epoch": 5.717948717948718, "grad_norm": 1.125, "learning_rate": 9.656333264874884e-05, "loss": 0.1451, "step": 7805 }, { "epoch": 5.718681318681319, "grad_norm": 0.8203125, "learning_rate": 9.656207812017044e-05, "loss": 0.1285, "step": 7806 }, { "epoch": 5.7194139194139195, "grad_norm": 0.84375, "learning_rate": 9.656082337089348e-05, "loss": 0.187, "step": 7807 }, { "epoch": 5.72014652014652, "grad_norm": 1.2421875, "learning_rate": 9.655956840092399e-05, "loss": 0.1602, "step": 7808 }, { "epoch": 5.720879120879121, "grad_norm": 1.0078125, "learning_rate": 9.6558313210268e-05, "loss": 0.1439, "step": 7809 }, { "epoch": 5.721611721611722, "grad_norm": 0.92578125, "learning_rate": 9.655705779893152e-05, "loss": 0.1086, "step": 7810 }, { "epoch": 5.722344322344322, "grad_norm": 1.0234375, "learning_rate": 9.655580216692055e-05, "loss": 0.1704, "step": 7811 }, { "epoch": 5.723076923076923, "grad_norm": 1.1875, "learning_rate": 9.655454631424112e-05, "loss": 0.1419, "step": 7812 }, { "epoch": 5.723809523809524, "grad_norm": 0.8125, "learning_rate": 9.655329024089922e-05, "loss": 0.125, "step": 7813 }, { "epoch": 5.724542124542125, "grad_norm": 1.4296875, "learning_rate": 9.655203394690089e-05, "loss": 0.1468, "step": 7814 }, { "epoch": 5.725274725274725, "grad_norm": 1.234375, "learning_rate": 9.655077743225218e-05, "loss": 0.2084, "step": 7815 }, { "epoch": 5.726007326007326, "grad_norm": 1.1640625, "learning_rate": 9.654952069695903e-05, "loss": 0.144, "step": 7816 }, { "epoch": 5.726739926739927, "grad_norm": 0.71484375, "learning_rate": 9.654826374102755e-05, "loss": 0.129, "step": 7817 }, { "epoch": 5.7274725274725276, "grad_norm": 0.67578125, "learning_rate": 9.65470065644637e-05, "loss": 0.1541, "step": 7818 }, { "epoch": 5.728205128205128, "grad_norm": 1.1328125, "learning_rate": 9.654574916727352e-05, "loss": 0.1404, "step": 7819 }, { "epoch": 5.728937728937729, "grad_norm": 0.83984375, "learning_rate": 9.654449154946307e-05, "loss": 0.1904, "step": 7820 }, { "epoch": 5.72967032967033, "grad_norm": 1.03125, "learning_rate": 9.654323371103832e-05, "loss": 0.1536, "step": 7821 }, { "epoch": 5.7304029304029305, "grad_norm": 1.46875, "learning_rate": 9.654197565200532e-05, "loss": 0.1708, "step": 7822 }, { "epoch": 5.731135531135531, "grad_norm": 0.7890625, "learning_rate": 9.65407173723701e-05, "loss": 0.119, "step": 7823 }, { "epoch": 5.731868131868132, "grad_norm": 1.890625, "learning_rate": 9.653945887213871e-05, "loss": 0.172, "step": 7824 }, { "epoch": 5.732600732600733, "grad_norm": 1.5390625, "learning_rate": 9.653820015131714e-05, "loss": 0.1673, "step": 7825 }, { "epoch": 5.733333333333333, "grad_norm": 1.3046875, "learning_rate": 9.653694120991144e-05, "loss": 0.2253, "step": 7826 }, { "epoch": 5.734065934065934, "grad_norm": 1.078125, "learning_rate": 9.653568204792765e-05, "loss": 0.186, "step": 7827 }, { "epoch": 5.734798534798535, "grad_norm": 1.078125, "learning_rate": 9.65344226653718e-05, "loss": 0.1464, "step": 7828 }, { "epoch": 5.735531135531136, "grad_norm": 1.84375, "learning_rate": 9.65331630622499e-05, "loss": 0.1859, "step": 7829 }, { "epoch": 5.736263736263736, "grad_norm": 0.87109375, "learning_rate": 9.653190323856802e-05, "loss": 0.136, "step": 7830 }, { "epoch": 5.736996336996337, "grad_norm": 1.328125, "learning_rate": 9.653064319433219e-05, "loss": 0.1347, "step": 7831 }, { "epoch": 5.737728937728938, "grad_norm": 1.46875, "learning_rate": 9.65293829295484e-05, "loss": 0.1596, "step": 7832 }, { "epoch": 5.7384615384615385, "grad_norm": 1.1328125, "learning_rate": 9.652812244422275e-05, "loss": 0.1551, "step": 7833 }, { "epoch": 5.739194139194139, "grad_norm": 1.6015625, "learning_rate": 9.652686173836127e-05, "loss": 0.1549, "step": 7834 }, { "epoch": 5.73992673992674, "grad_norm": 1.84375, "learning_rate": 9.652560081196997e-05, "loss": 0.1484, "step": 7835 }, { "epoch": 5.740659340659341, "grad_norm": 1.234375, "learning_rate": 9.65243396650549e-05, "loss": 0.1621, "step": 7836 }, { "epoch": 5.741391941391941, "grad_norm": 1.4765625, "learning_rate": 9.652307829762213e-05, "loss": 0.1621, "step": 7837 }, { "epoch": 5.742124542124542, "grad_norm": 0.953125, "learning_rate": 9.652181670967766e-05, "loss": 0.1637, "step": 7838 }, { "epoch": 5.742857142857143, "grad_norm": 1.4609375, "learning_rate": 9.652055490122756e-05, "loss": 0.1732, "step": 7839 }, { "epoch": 5.743589743589744, "grad_norm": 2.078125, "learning_rate": 9.651929287227789e-05, "loss": 0.1837, "step": 7840 }, { "epoch": 5.744322344322344, "grad_norm": 1.4375, "learning_rate": 9.651803062283467e-05, "loss": 0.1293, "step": 7841 }, { "epoch": 5.745054945054945, "grad_norm": 1.8359375, "learning_rate": 9.651676815290396e-05, "loss": 0.1919, "step": 7842 }, { "epoch": 5.745787545787546, "grad_norm": 1.1953125, "learning_rate": 9.651550546249182e-05, "loss": 0.1343, "step": 7843 }, { "epoch": 5.7465201465201465, "grad_norm": 0.86328125, "learning_rate": 9.651424255160427e-05, "loss": 0.1301, "step": 7844 }, { "epoch": 5.747252747252747, "grad_norm": 0.91015625, "learning_rate": 9.651297942024738e-05, "loss": 0.112, "step": 7845 }, { "epoch": 5.747985347985348, "grad_norm": 1.140625, "learning_rate": 9.651171606842722e-05, "loss": 0.1368, "step": 7846 }, { "epoch": 5.748717948717949, "grad_norm": 1.5390625, "learning_rate": 9.651045249614981e-05, "loss": 0.1536, "step": 7847 }, { "epoch": 5.7494505494505495, "grad_norm": 1.75, "learning_rate": 9.650918870342121e-05, "loss": 0.1348, "step": 7848 }, { "epoch": 5.75018315018315, "grad_norm": 0.94140625, "learning_rate": 9.650792469024748e-05, "loss": 0.1224, "step": 7849 }, { "epoch": 5.750915750915751, "grad_norm": 0.88671875, "learning_rate": 9.65066604566347e-05, "loss": 0.133, "step": 7850 }, { "epoch": 5.751648351648352, "grad_norm": 0.7265625, "learning_rate": 9.65053960025889e-05, "loss": 0.1284, "step": 7851 }, { "epoch": 5.752380952380952, "grad_norm": 1.1875, "learning_rate": 9.650413132811613e-05, "loss": 0.1213, "step": 7852 }, { "epoch": 5.753113553113553, "grad_norm": 1.4140625, "learning_rate": 9.650286643322248e-05, "loss": 0.1343, "step": 7853 }, { "epoch": 5.753846153846154, "grad_norm": 1.4609375, "learning_rate": 9.650160131791399e-05, "loss": 0.137, "step": 7854 }, { "epoch": 5.754578754578755, "grad_norm": 2.140625, "learning_rate": 9.650033598219674e-05, "loss": 0.1593, "step": 7855 }, { "epoch": 5.755311355311355, "grad_norm": 0.9765625, "learning_rate": 9.649907042607677e-05, "loss": 0.1379, "step": 7856 }, { "epoch": 5.756043956043956, "grad_norm": 0.97265625, "learning_rate": 9.649780464956014e-05, "loss": 0.1333, "step": 7857 }, { "epoch": 5.756776556776557, "grad_norm": 1.90625, "learning_rate": 9.649653865265295e-05, "loss": 0.1458, "step": 7858 }, { "epoch": 5.7575091575091575, "grad_norm": 1.78125, "learning_rate": 9.649527243536122e-05, "loss": 0.1392, "step": 7859 }, { "epoch": 5.758241758241758, "grad_norm": 1.59375, "learning_rate": 9.649400599769106e-05, "loss": 0.1306, "step": 7860 }, { "epoch": 5.758974358974359, "grad_norm": 1.09375, "learning_rate": 9.64927393396485e-05, "loss": 0.1414, "step": 7861 }, { "epoch": 5.75970695970696, "grad_norm": 1.2734375, "learning_rate": 9.649147246123965e-05, "loss": 0.1488, "step": 7862 }, { "epoch": 5.76043956043956, "grad_norm": 1.0390625, "learning_rate": 9.649020536247055e-05, "loss": 0.1579, "step": 7863 }, { "epoch": 5.761172161172161, "grad_norm": 2.171875, "learning_rate": 9.648893804334728e-05, "loss": 0.2231, "step": 7864 }, { "epoch": 5.761904761904762, "grad_norm": 1.2265625, "learning_rate": 9.648767050387588e-05, "loss": 0.1273, "step": 7865 }, { "epoch": 5.762637362637363, "grad_norm": 1.109375, "learning_rate": 9.648640274406249e-05, "loss": 0.151, "step": 7866 }, { "epoch": 5.763369963369963, "grad_norm": 0.9765625, "learning_rate": 9.648513476391315e-05, "loss": 0.149, "step": 7867 }, { "epoch": 5.764102564102564, "grad_norm": 0.79296875, "learning_rate": 9.648386656343392e-05, "loss": 0.1504, "step": 7868 }, { "epoch": 5.764835164835165, "grad_norm": 1.5234375, "learning_rate": 9.648259814263087e-05, "loss": 0.1831, "step": 7869 }, { "epoch": 5.7655677655677655, "grad_norm": 1.421875, "learning_rate": 9.648132950151012e-05, "loss": 0.1651, "step": 7870 }, { "epoch": 5.766300366300366, "grad_norm": 1.578125, "learning_rate": 9.648006064007773e-05, "loss": 0.1136, "step": 7871 }, { "epoch": 5.767032967032967, "grad_norm": 1.0546875, "learning_rate": 9.647879155833977e-05, "loss": 0.1437, "step": 7872 }, { "epoch": 5.767765567765568, "grad_norm": 0.90625, "learning_rate": 9.647752225630232e-05, "loss": 0.1318, "step": 7873 }, { "epoch": 5.768498168498168, "grad_norm": 1.2109375, "learning_rate": 9.647625273397148e-05, "loss": 0.1599, "step": 7874 }, { "epoch": 5.769230769230769, "grad_norm": 1.0546875, "learning_rate": 9.64749829913533e-05, "loss": 0.1258, "step": 7875 }, { "epoch": 5.76996336996337, "grad_norm": 1.2265625, "learning_rate": 9.64737130284539e-05, "loss": 0.1568, "step": 7876 }, { "epoch": 5.770695970695971, "grad_norm": 0.87890625, "learning_rate": 9.647244284527935e-05, "loss": 0.1439, "step": 7877 }, { "epoch": 5.771428571428571, "grad_norm": 0.68359375, "learning_rate": 9.64711724418357e-05, "loss": 0.1398, "step": 7878 }, { "epoch": 5.772161172161172, "grad_norm": 0.84375, "learning_rate": 9.646990181812911e-05, "loss": 0.12, "step": 7879 }, { "epoch": 5.772893772893773, "grad_norm": 1.2734375, "learning_rate": 9.646863097416564e-05, "loss": 0.1648, "step": 7880 }, { "epoch": 5.7736263736263735, "grad_norm": 0.98046875, "learning_rate": 9.646735990995135e-05, "loss": 0.1256, "step": 7881 }, { "epoch": 5.774358974358974, "grad_norm": 1.203125, "learning_rate": 9.646608862549234e-05, "loss": 0.1463, "step": 7882 }, { "epoch": 5.775091575091575, "grad_norm": 1.2421875, "learning_rate": 9.646481712079471e-05, "loss": 0.1366, "step": 7883 }, { "epoch": 5.775824175824176, "grad_norm": 1.5390625, "learning_rate": 9.646354539586457e-05, "loss": 0.1421, "step": 7884 }, { "epoch": 5.7765567765567765, "grad_norm": 0.77734375, "learning_rate": 9.646227345070797e-05, "loss": 0.1493, "step": 7885 }, { "epoch": 5.777289377289377, "grad_norm": 1.203125, "learning_rate": 9.646100128533105e-05, "loss": 0.1566, "step": 7886 }, { "epoch": 5.778021978021978, "grad_norm": 1.28125, "learning_rate": 9.645972889973987e-05, "loss": 0.1371, "step": 7887 }, { "epoch": 5.778754578754579, "grad_norm": 1.265625, "learning_rate": 9.645845629394056e-05, "loss": 0.1369, "step": 7888 }, { "epoch": 5.779487179487179, "grad_norm": 1.3984375, "learning_rate": 9.645718346793918e-05, "loss": 0.1639, "step": 7889 }, { "epoch": 5.78021978021978, "grad_norm": 0.7890625, "learning_rate": 9.645591042174186e-05, "loss": 0.1224, "step": 7890 }, { "epoch": 5.780952380952381, "grad_norm": 1.6015625, "learning_rate": 9.645463715535469e-05, "loss": 0.1519, "step": 7891 }, { "epoch": 5.781684981684982, "grad_norm": 1.2109375, "learning_rate": 9.645336366878375e-05, "loss": 0.146, "step": 7892 }, { "epoch": 5.782417582417582, "grad_norm": 1.0390625, "learning_rate": 9.645208996203518e-05, "loss": 0.1595, "step": 7893 }, { "epoch": 5.783150183150183, "grad_norm": 1.296875, "learning_rate": 9.645081603511504e-05, "loss": 0.1609, "step": 7894 }, { "epoch": 5.783882783882784, "grad_norm": 0.81640625, "learning_rate": 9.644954188802947e-05, "loss": 0.1368, "step": 7895 }, { "epoch": 5.7846153846153845, "grad_norm": 1.609375, "learning_rate": 9.644826752078456e-05, "loss": 0.1831, "step": 7896 }, { "epoch": 5.785347985347985, "grad_norm": 0.78515625, "learning_rate": 9.644699293338641e-05, "loss": 0.136, "step": 7897 }, { "epoch": 5.786080586080586, "grad_norm": 0.78125, "learning_rate": 9.644571812584113e-05, "loss": 0.1406, "step": 7898 }, { "epoch": 5.786813186813187, "grad_norm": 1.4140625, "learning_rate": 9.644444309815483e-05, "loss": 0.142, "step": 7899 }, { "epoch": 5.787545787545787, "grad_norm": 0.7734375, "learning_rate": 9.644316785033363e-05, "loss": 0.1369, "step": 7900 }, { "epoch": 5.788278388278388, "grad_norm": 0.71484375, "learning_rate": 9.644189238238364e-05, "loss": 0.1385, "step": 7901 }, { "epoch": 5.789010989010989, "grad_norm": 1.0859375, "learning_rate": 9.644061669431095e-05, "loss": 0.1686, "step": 7902 }, { "epoch": 5.78974358974359, "grad_norm": 0.86328125, "learning_rate": 9.643934078612168e-05, "loss": 0.1723, "step": 7903 }, { "epoch": 5.79047619047619, "grad_norm": 1.0859375, "learning_rate": 9.643806465782196e-05, "loss": 0.1425, "step": 7904 }, { "epoch": 5.791208791208791, "grad_norm": 1.0234375, "learning_rate": 9.643678830941787e-05, "loss": 0.1362, "step": 7905 }, { "epoch": 5.791941391941392, "grad_norm": 1.0078125, "learning_rate": 9.643551174091557e-05, "loss": 0.1456, "step": 7906 }, { "epoch": 5.7926739926739925, "grad_norm": 1.28125, "learning_rate": 9.643423495232115e-05, "loss": 0.1707, "step": 7907 }, { "epoch": 5.793406593406593, "grad_norm": 1.1171875, "learning_rate": 9.643295794364074e-05, "loss": 0.1515, "step": 7908 }, { "epoch": 5.794139194139194, "grad_norm": 0.8828125, "learning_rate": 9.643168071488042e-05, "loss": 0.1535, "step": 7909 }, { "epoch": 5.794871794871795, "grad_norm": 0.94140625, "learning_rate": 9.643040326604637e-05, "loss": 0.1634, "step": 7910 }, { "epoch": 5.795604395604395, "grad_norm": 1.4140625, "learning_rate": 9.642912559714466e-05, "loss": 0.1695, "step": 7911 }, { "epoch": 5.796336996336996, "grad_norm": 1.3359375, "learning_rate": 9.642784770818143e-05, "loss": 0.178, "step": 7912 }, { "epoch": 5.797069597069597, "grad_norm": 1.34375, "learning_rate": 9.642656959916281e-05, "loss": 0.1649, "step": 7913 }, { "epoch": 5.797802197802198, "grad_norm": 0.89453125, "learning_rate": 9.642529127009492e-05, "loss": 0.144, "step": 7914 }, { "epoch": 5.798534798534798, "grad_norm": 1.2421875, "learning_rate": 9.642401272098389e-05, "loss": 0.1297, "step": 7915 }, { "epoch": 5.799267399267399, "grad_norm": 1.421875, "learning_rate": 9.642273395183584e-05, "loss": 0.1402, "step": 7916 }, { "epoch": 5.8, "grad_norm": 0.9609375, "learning_rate": 9.642145496265688e-05, "loss": 0.1195, "step": 7917 }, { "epoch": 5.8007326007326006, "grad_norm": 0.80859375, "learning_rate": 9.642017575345316e-05, "loss": 0.1382, "step": 7918 }, { "epoch": 5.801465201465201, "grad_norm": 1.125, "learning_rate": 9.64188963242308e-05, "loss": 0.1563, "step": 7919 }, { "epoch": 5.802197802197802, "grad_norm": 0.94921875, "learning_rate": 9.641761667499595e-05, "loss": 0.1165, "step": 7920 }, { "epoch": 5.802930402930403, "grad_norm": 1.296875, "learning_rate": 9.641633680575472e-05, "loss": 0.1883, "step": 7921 }, { "epoch": 5.8036630036630035, "grad_norm": 1.265625, "learning_rate": 9.641505671651323e-05, "loss": 0.1454, "step": 7922 }, { "epoch": 5.804395604395604, "grad_norm": 0.94140625, "learning_rate": 9.641377640727765e-05, "loss": 0.117, "step": 7923 }, { "epoch": 5.805128205128205, "grad_norm": 1.1484375, "learning_rate": 9.641249587805407e-05, "loss": 0.1435, "step": 7924 }, { "epoch": 5.805860805860806, "grad_norm": 1.03125, "learning_rate": 9.641121512884868e-05, "loss": 0.1283, "step": 7925 }, { "epoch": 5.806593406593406, "grad_norm": 1.40625, "learning_rate": 9.640993415966755e-05, "loss": 0.1609, "step": 7926 }, { "epoch": 5.807326007326007, "grad_norm": 1.3203125, "learning_rate": 9.640865297051689e-05, "loss": 0.1386, "step": 7927 }, { "epoch": 5.808058608058608, "grad_norm": 0.625, "learning_rate": 9.64073715614028e-05, "loss": 0.1261, "step": 7928 }, { "epoch": 5.808791208791209, "grad_norm": 1.703125, "learning_rate": 9.640608993233141e-05, "loss": 0.1544, "step": 7929 }, { "epoch": 5.809523809523809, "grad_norm": 1.3984375, "learning_rate": 9.640480808330888e-05, "loss": 0.1731, "step": 7930 }, { "epoch": 5.81025641025641, "grad_norm": 1.046875, "learning_rate": 9.640352601434134e-05, "loss": 0.1319, "step": 7931 }, { "epoch": 5.810989010989011, "grad_norm": 1.4375, "learning_rate": 9.640224372543492e-05, "loss": 0.1664, "step": 7932 }, { "epoch": 5.8117216117216115, "grad_norm": 1.078125, "learning_rate": 9.640096121659582e-05, "loss": 0.1584, "step": 7933 }, { "epoch": 5.812454212454212, "grad_norm": 0.85546875, "learning_rate": 9.639967848783012e-05, "loss": 0.1625, "step": 7934 }, { "epoch": 5.813186813186813, "grad_norm": 1.109375, "learning_rate": 9.639839553914401e-05, "loss": 0.1371, "step": 7935 }, { "epoch": 5.813919413919414, "grad_norm": 0.93359375, "learning_rate": 9.639711237054361e-05, "loss": 0.1021, "step": 7936 }, { "epoch": 5.814652014652014, "grad_norm": 0.8515625, "learning_rate": 9.639582898203506e-05, "loss": 0.1207, "step": 7937 }, { "epoch": 5.815384615384615, "grad_norm": 0.9140625, "learning_rate": 9.639454537362454e-05, "loss": 0.1281, "step": 7938 }, { "epoch": 5.816117216117216, "grad_norm": 1.625, "learning_rate": 9.639326154531821e-05, "loss": 0.1832, "step": 7939 }, { "epoch": 5.816849816849817, "grad_norm": 0.84765625, "learning_rate": 9.639197749712216e-05, "loss": 0.1473, "step": 7940 }, { "epoch": 5.817582417582417, "grad_norm": 1.3359375, "learning_rate": 9.63906932290426e-05, "loss": 0.2064, "step": 7941 }, { "epoch": 5.818315018315018, "grad_norm": 1.0546875, "learning_rate": 9.638940874108566e-05, "loss": 0.1326, "step": 7942 }, { "epoch": 5.819047619047619, "grad_norm": 1.28125, "learning_rate": 9.638812403325751e-05, "loss": 0.1405, "step": 7943 }, { "epoch": 5.8197802197802195, "grad_norm": 1.078125, "learning_rate": 9.638683910556427e-05, "loss": 0.1688, "step": 7944 }, { "epoch": 5.82051282051282, "grad_norm": 0.82421875, "learning_rate": 9.638555395801216e-05, "loss": 0.1782, "step": 7945 }, { "epoch": 5.821245421245421, "grad_norm": 1.84375, "learning_rate": 9.638426859060727e-05, "loss": 0.1568, "step": 7946 }, { "epoch": 5.821978021978022, "grad_norm": 0.921875, "learning_rate": 9.638298300335577e-05, "loss": 0.1631, "step": 7947 }, { "epoch": 5.8227106227106225, "grad_norm": 1.15625, "learning_rate": 9.638169719626387e-05, "loss": 0.164, "step": 7948 }, { "epoch": 5.823443223443223, "grad_norm": 0.83203125, "learning_rate": 9.638041116933768e-05, "loss": 0.1279, "step": 7949 }, { "epoch": 5.824175824175824, "grad_norm": 1.015625, "learning_rate": 9.637912492258338e-05, "loss": 0.1275, "step": 7950 }, { "epoch": 5.824908424908425, "grad_norm": 1.3984375, "learning_rate": 9.637783845600713e-05, "loss": 0.1736, "step": 7951 }, { "epoch": 5.825641025641025, "grad_norm": 1.0390625, "learning_rate": 9.637655176961509e-05, "loss": 0.121, "step": 7952 }, { "epoch": 5.826373626373626, "grad_norm": 1.1484375, "learning_rate": 9.637526486341344e-05, "loss": 0.1704, "step": 7953 }, { "epoch": 5.827106227106227, "grad_norm": 1.1328125, "learning_rate": 9.637397773740832e-05, "loss": 0.1229, "step": 7954 }, { "epoch": 5.827838827838828, "grad_norm": 0.76953125, "learning_rate": 9.637269039160595e-05, "loss": 0.139, "step": 7955 }, { "epoch": 5.828571428571428, "grad_norm": 1.7578125, "learning_rate": 9.637140282601243e-05, "loss": 0.1687, "step": 7956 }, { "epoch": 5.829304029304029, "grad_norm": 1.3046875, "learning_rate": 9.637011504063397e-05, "loss": 0.1687, "step": 7957 }, { "epoch": 5.83003663003663, "grad_norm": 0.98828125, "learning_rate": 9.636882703547672e-05, "loss": 0.1901, "step": 7958 }, { "epoch": 5.8307692307692305, "grad_norm": 0.81640625, "learning_rate": 9.636753881054687e-05, "loss": 0.1206, "step": 7959 }, { "epoch": 5.831501831501831, "grad_norm": 1.8984375, "learning_rate": 9.636625036585059e-05, "loss": 0.1906, "step": 7960 }, { "epoch": 5.832234432234432, "grad_norm": 2.125, "learning_rate": 9.636496170139404e-05, "loss": 0.1702, "step": 7961 }, { "epoch": 5.832967032967033, "grad_norm": 1.578125, "learning_rate": 9.63636728171834e-05, "loss": 0.1482, "step": 7962 }, { "epoch": 5.833699633699633, "grad_norm": 1.2578125, "learning_rate": 9.636238371322486e-05, "loss": 0.1564, "step": 7963 }, { "epoch": 5.834432234432234, "grad_norm": 0.74609375, "learning_rate": 9.636109438952458e-05, "loss": 0.1357, "step": 7964 }, { "epoch": 5.835164835164835, "grad_norm": 0.984375, "learning_rate": 9.635980484608873e-05, "loss": 0.1366, "step": 7965 }, { "epoch": 5.835897435897436, "grad_norm": 1.40625, "learning_rate": 9.635851508292351e-05, "loss": 0.1758, "step": 7966 }, { "epoch": 5.836630036630036, "grad_norm": 1.765625, "learning_rate": 9.635722510003509e-05, "loss": 0.1516, "step": 7967 }, { "epoch": 5.837362637362637, "grad_norm": 2.03125, "learning_rate": 9.635593489742964e-05, "loss": 0.1651, "step": 7968 }, { "epoch": 5.838095238095238, "grad_norm": 0.81640625, "learning_rate": 9.635464447511336e-05, "loss": 0.1426, "step": 7969 }, { "epoch": 5.8388278388278385, "grad_norm": 1.1796875, "learning_rate": 9.635335383309241e-05, "loss": 0.1347, "step": 7970 }, { "epoch": 5.839560439560439, "grad_norm": 1.59375, "learning_rate": 9.635206297137303e-05, "loss": 0.17, "step": 7971 }, { "epoch": 5.84029304029304, "grad_norm": 1.15625, "learning_rate": 9.635077188996133e-05, "loss": 0.1374, "step": 7972 }, { "epoch": 5.841025641025641, "grad_norm": 0.69140625, "learning_rate": 9.634948058886354e-05, "loss": 0.1485, "step": 7973 }, { "epoch": 5.841758241758241, "grad_norm": 1.34375, "learning_rate": 9.634818906808582e-05, "loss": 0.1248, "step": 7974 }, { "epoch": 5.842490842490842, "grad_norm": 1.078125, "learning_rate": 9.63468973276344e-05, "loss": 0.1177, "step": 7975 }, { "epoch": 5.843223443223443, "grad_norm": 0.6328125, "learning_rate": 9.634560536751543e-05, "loss": 0.1028, "step": 7976 }, { "epoch": 5.843956043956044, "grad_norm": 0.9453125, "learning_rate": 9.634431318773513e-05, "loss": 0.1747, "step": 7977 }, { "epoch": 5.844688644688644, "grad_norm": 1.3046875, "learning_rate": 9.634302078829967e-05, "loss": 0.1517, "step": 7978 }, { "epoch": 5.845421245421245, "grad_norm": 2.03125, "learning_rate": 9.634172816921524e-05, "loss": 0.1789, "step": 7979 }, { "epoch": 5.846153846153846, "grad_norm": 1.5703125, "learning_rate": 9.634043533048805e-05, "loss": 0.1652, "step": 7980 }, { "epoch": 5.8468864468864465, "grad_norm": 0.94921875, "learning_rate": 9.633914227212427e-05, "loss": 0.1378, "step": 7981 }, { "epoch": 5.847619047619047, "grad_norm": 0.91796875, "learning_rate": 9.633784899413012e-05, "loss": 0.1566, "step": 7982 }, { "epoch": 5.848351648351648, "grad_norm": 0.5078125, "learning_rate": 9.633655549651179e-05, "loss": 0.1084, "step": 7983 }, { "epoch": 5.849084249084249, "grad_norm": 0.94140625, "learning_rate": 9.633526177927548e-05, "loss": 0.1324, "step": 7984 }, { "epoch": 5.8498168498168495, "grad_norm": 1.0390625, "learning_rate": 9.633396784242739e-05, "loss": 0.1181, "step": 7985 }, { "epoch": 5.85054945054945, "grad_norm": 0.89453125, "learning_rate": 9.633267368597369e-05, "loss": 0.1519, "step": 7986 }, { "epoch": 5.851282051282051, "grad_norm": 1.0859375, "learning_rate": 9.633137930992063e-05, "loss": 0.1579, "step": 7987 }, { "epoch": 5.852014652014652, "grad_norm": 1.21875, "learning_rate": 9.633008471427437e-05, "loss": 0.1388, "step": 7988 }, { "epoch": 5.852747252747252, "grad_norm": 0.86328125, "learning_rate": 9.632878989904114e-05, "loss": 0.153, "step": 7989 }, { "epoch": 5.853479853479853, "grad_norm": 0.75, "learning_rate": 9.632749486422713e-05, "loss": 0.1514, "step": 7990 }, { "epoch": 5.854212454212454, "grad_norm": 1.1796875, "learning_rate": 9.632619960983854e-05, "loss": 0.1788, "step": 7991 }, { "epoch": 5.854945054945055, "grad_norm": 1.1171875, "learning_rate": 9.632490413588158e-05, "loss": 0.1739, "step": 7992 }, { "epoch": 5.855677655677655, "grad_norm": 1.5546875, "learning_rate": 9.632360844236248e-05, "loss": 0.1786, "step": 7993 }, { "epoch": 5.856410256410256, "grad_norm": 1.09375, "learning_rate": 9.632231252928742e-05, "loss": 0.1163, "step": 7994 }, { "epoch": 5.857142857142857, "grad_norm": 0.796875, "learning_rate": 9.632101639666261e-05, "loss": 0.1396, "step": 7995 }, { "epoch": 5.8578754578754575, "grad_norm": 2.1875, "learning_rate": 9.631972004449427e-05, "loss": 0.121, "step": 7996 }, { "epoch": 5.858608058608058, "grad_norm": 1.90625, "learning_rate": 9.631842347278864e-05, "loss": 0.2213, "step": 7997 }, { "epoch": 5.859340659340659, "grad_norm": 2.125, "learning_rate": 9.631712668155185e-05, "loss": 0.107, "step": 7998 }, { "epoch": 5.860073260073261, "grad_norm": 1.2265625, "learning_rate": 9.63158296707902e-05, "loss": 0.1658, "step": 7999 }, { "epoch": 5.860805860805861, "grad_norm": 0.88671875, "learning_rate": 9.631453244050987e-05, "loss": 0.1093, "step": 8000 }, { "epoch": 5.861538461538462, "grad_norm": 0.875, "learning_rate": 9.631323499071705e-05, "loss": 0.1521, "step": 8001 }, { "epoch": 5.862271062271063, "grad_norm": 1.0625, "learning_rate": 9.631193732141801e-05, "loss": 0.1353, "step": 8002 }, { "epoch": 5.8630036630036635, "grad_norm": 1.8828125, "learning_rate": 9.631063943261893e-05, "loss": 0.1696, "step": 8003 }, { "epoch": 5.863736263736264, "grad_norm": 1.3671875, "learning_rate": 9.630934132432602e-05, "loss": 0.1314, "step": 8004 }, { "epoch": 5.864468864468865, "grad_norm": 1.0625, "learning_rate": 9.630804299654555e-05, "loss": 0.1439, "step": 8005 }, { "epoch": 5.865201465201466, "grad_norm": 1.1328125, "learning_rate": 9.630674444928368e-05, "loss": 0.1499, "step": 8006 }, { "epoch": 5.865934065934066, "grad_norm": 1.4453125, "learning_rate": 9.630544568254667e-05, "loss": 0.1448, "step": 8007 }, { "epoch": 5.866666666666667, "grad_norm": 1.4140625, "learning_rate": 9.630414669634075e-05, "loss": 0.1499, "step": 8008 }, { "epoch": 5.867399267399268, "grad_norm": 1.40625, "learning_rate": 9.63028474906721e-05, "loss": 0.1584, "step": 8009 }, { "epoch": 5.868131868131869, "grad_norm": 0.7578125, "learning_rate": 9.630154806554698e-05, "loss": 0.0997, "step": 8010 }, { "epoch": 5.868864468864469, "grad_norm": 1.2890625, "learning_rate": 9.63002484209716e-05, "loss": 0.1504, "step": 8011 }, { "epoch": 5.86959706959707, "grad_norm": 1.15625, "learning_rate": 9.629894855695222e-05, "loss": 0.1595, "step": 8012 }, { "epoch": 5.870329670329671, "grad_norm": 0.9921875, "learning_rate": 9.629764847349502e-05, "loss": 0.187, "step": 8013 }, { "epoch": 5.8710622710622715, "grad_norm": 1.5703125, "learning_rate": 9.629634817060628e-05, "loss": 0.1926, "step": 8014 }, { "epoch": 5.871794871794872, "grad_norm": 1.765625, "learning_rate": 9.629504764829216e-05, "loss": 0.2013, "step": 8015 }, { "epoch": 5.872527472527473, "grad_norm": 1.34375, "learning_rate": 9.629374690655896e-05, "loss": 0.1374, "step": 8016 }, { "epoch": 5.873260073260074, "grad_norm": 1.3515625, "learning_rate": 9.629244594541287e-05, "loss": 0.118, "step": 8017 }, { "epoch": 5.873992673992674, "grad_norm": 1.59375, "learning_rate": 9.629114476486016e-05, "loss": 0.1313, "step": 8018 }, { "epoch": 5.874725274725275, "grad_norm": 0.90625, "learning_rate": 9.628984336490702e-05, "loss": 0.137, "step": 8019 }, { "epoch": 5.875457875457876, "grad_norm": 0.69921875, "learning_rate": 9.628854174555972e-05, "loss": 0.1291, "step": 8020 }, { "epoch": 5.876190476190477, "grad_norm": 1.015625, "learning_rate": 9.628723990682449e-05, "loss": 0.1873, "step": 8021 }, { "epoch": 5.876923076923077, "grad_norm": 0.90625, "learning_rate": 9.628593784870757e-05, "loss": 0.1517, "step": 8022 }, { "epoch": 5.877655677655678, "grad_norm": 1.5546875, "learning_rate": 9.628463557121518e-05, "loss": 0.1431, "step": 8023 }, { "epoch": 5.878388278388279, "grad_norm": 1.4453125, "learning_rate": 9.628333307435357e-05, "loss": 0.1071, "step": 8024 }, { "epoch": 5.8791208791208796, "grad_norm": 0.8671875, "learning_rate": 9.628203035812897e-05, "loss": 0.1708, "step": 8025 }, { "epoch": 5.87985347985348, "grad_norm": 1.4921875, "learning_rate": 9.628072742254765e-05, "loss": 0.1586, "step": 8026 }, { "epoch": 5.880586080586081, "grad_norm": 2.078125, "learning_rate": 9.627942426761584e-05, "loss": 0.2015, "step": 8027 }, { "epoch": 5.881318681318682, "grad_norm": 1.2265625, "learning_rate": 9.627812089333976e-05, "loss": 0.1347, "step": 8028 }, { "epoch": 5.8820512820512825, "grad_norm": 0.796875, "learning_rate": 9.627681729972568e-05, "loss": 0.1459, "step": 8029 }, { "epoch": 5.882783882783883, "grad_norm": 0.9765625, "learning_rate": 9.627551348677985e-05, "loss": 0.1147, "step": 8030 }, { "epoch": 5.883516483516484, "grad_norm": 0.90234375, "learning_rate": 9.627420945450851e-05, "loss": 0.1431, "step": 8031 }, { "epoch": 5.884249084249085, "grad_norm": 1.6015625, "learning_rate": 9.62729052029179e-05, "loss": 0.1713, "step": 8032 }, { "epoch": 5.884981684981685, "grad_norm": 1.171875, "learning_rate": 9.627160073201426e-05, "loss": 0.1439, "step": 8033 }, { "epoch": 5.885714285714286, "grad_norm": 0.859375, "learning_rate": 9.627029604180388e-05, "loss": 0.1377, "step": 8034 }, { "epoch": 5.886446886446887, "grad_norm": 1.1328125, "learning_rate": 9.626899113229296e-05, "loss": 0.1399, "step": 8035 }, { "epoch": 5.887179487179488, "grad_norm": 0.8359375, "learning_rate": 9.626768600348781e-05, "loss": 0.1243, "step": 8036 }, { "epoch": 5.887912087912088, "grad_norm": 1.296875, "learning_rate": 9.626638065539465e-05, "loss": 0.1442, "step": 8037 }, { "epoch": 5.888644688644689, "grad_norm": 1.046875, "learning_rate": 9.62650750880197e-05, "loss": 0.1308, "step": 8038 }, { "epoch": 5.88937728937729, "grad_norm": 0.88671875, "learning_rate": 9.626376930136928e-05, "loss": 0.1486, "step": 8039 }, { "epoch": 5.8901098901098905, "grad_norm": 1.15625, "learning_rate": 9.62624632954496e-05, "loss": 0.1533, "step": 8040 }, { "epoch": 5.890842490842491, "grad_norm": 1.0234375, "learning_rate": 9.626115707026695e-05, "loss": 0.1515, "step": 8041 }, { "epoch": 5.891575091575092, "grad_norm": 1.5703125, "learning_rate": 9.625985062582759e-05, "loss": 0.1366, "step": 8042 }, { "epoch": 5.892307692307693, "grad_norm": 1.328125, "learning_rate": 9.625854396213774e-05, "loss": 0.1481, "step": 8043 }, { "epoch": 5.893040293040293, "grad_norm": 1.1953125, "learning_rate": 9.625723707920369e-05, "loss": 0.1335, "step": 8044 }, { "epoch": 5.893772893772894, "grad_norm": 0.42578125, "learning_rate": 9.62559299770317e-05, "loss": 0.1198, "step": 8045 }, { "epoch": 5.894505494505495, "grad_norm": 1.75, "learning_rate": 9.625462265562802e-05, "loss": 0.1697, "step": 8046 }, { "epoch": 5.895238095238096, "grad_norm": 1.296875, "learning_rate": 9.625331511499893e-05, "loss": 0.1063, "step": 8047 }, { "epoch": 5.895970695970696, "grad_norm": 1.5078125, "learning_rate": 9.625200735515068e-05, "loss": 0.1743, "step": 8048 }, { "epoch": 5.896703296703297, "grad_norm": 1.65625, "learning_rate": 9.625069937608955e-05, "loss": 0.1623, "step": 8049 }, { "epoch": 5.897435897435898, "grad_norm": 1.734375, "learning_rate": 9.62493911778218e-05, "loss": 0.1658, "step": 8050 }, { "epoch": 5.8981684981684985, "grad_norm": 0.84375, "learning_rate": 9.62480827603537e-05, "loss": 0.1757, "step": 8051 }, { "epoch": 5.898901098901099, "grad_norm": 1.765625, "learning_rate": 9.624677412369151e-05, "loss": 0.1314, "step": 8052 }, { "epoch": 5.8996336996337, "grad_norm": 1.515625, "learning_rate": 9.62454652678415e-05, "loss": 0.1226, "step": 8053 }, { "epoch": 5.900366300366301, "grad_norm": 2.140625, "learning_rate": 9.624415619280997e-05, "loss": 0.1635, "step": 8054 }, { "epoch": 5.9010989010989015, "grad_norm": 1.3359375, "learning_rate": 9.624284689860317e-05, "loss": 0.1871, "step": 8055 }, { "epoch": 5.901831501831502, "grad_norm": 1.2890625, "learning_rate": 9.624153738522737e-05, "loss": 0.1326, "step": 8056 }, { "epoch": 5.902564102564103, "grad_norm": 1.25, "learning_rate": 9.624022765268883e-05, "loss": 0.1897, "step": 8057 }, { "epoch": 5.903296703296704, "grad_norm": 1.53125, "learning_rate": 9.623891770099386e-05, "loss": 0.1543, "step": 8058 }, { "epoch": 5.904029304029304, "grad_norm": 1.453125, "learning_rate": 9.623760753014873e-05, "loss": 0.1814, "step": 8059 }, { "epoch": 5.904761904761905, "grad_norm": 1.125, "learning_rate": 9.623629714015968e-05, "loss": 0.1328, "step": 8060 }, { "epoch": 5.905494505494506, "grad_norm": 0.71484375, "learning_rate": 9.623498653103304e-05, "loss": 0.1407, "step": 8061 }, { "epoch": 5.906227106227107, "grad_norm": 1.2109375, "learning_rate": 9.623367570277505e-05, "loss": 0.135, "step": 8062 }, { "epoch": 5.906959706959707, "grad_norm": 1.15625, "learning_rate": 9.623236465539202e-05, "loss": 0.1728, "step": 8063 }, { "epoch": 5.907692307692308, "grad_norm": 1.2265625, "learning_rate": 9.62310533888902e-05, "loss": 0.1195, "step": 8064 }, { "epoch": 5.908424908424909, "grad_norm": 0.88671875, "learning_rate": 9.62297419032759e-05, "loss": 0.1494, "step": 8065 }, { "epoch": 5.9091575091575095, "grad_norm": 0.80078125, "learning_rate": 9.622843019855541e-05, "loss": 0.1349, "step": 8066 }, { "epoch": 5.90989010989011, "grad_norm": 0.86328125, "learning_rate": 9.622711827473499e-05, "loss": 0.1562, "step": 8067 }, { "epoch": 5.910622710622711, "grad_norm": 1.046875, "learning_rate": 9.622580613182093e-05, "loss": 0.1394, "step": 8068 }, { "epoch": 5.911355311355312, "grad_norm": 0.671875, "learning_rate": 9.622449376981951e-05, "loss": 0.1157, "step": 8069 }, { "epoch": 5.912087912087912, "grad_norm": 1.015625, "learning_rate": 9.622318118873705e-05, "loss": 0.1381, "step": 8070 }, { "epoch": 5.912820512820513, "grad_norm": 1.5625, "learning_rate": 9.622186838857981e-05, "loss": 0.1493, "step": 8071 }, { "epoch": 5.913553113553114, "grad_norm": 1.28125, "learning_rate": 9.62205553693541e-05, "loss": 0.1374, "step": 8072 }, { "epoch": 5.914285714285715, "grad_norm": 0.79296875, "learning_rate": 9.62192421310662e-05, "loss": 0.1642, "step": 8073 }, { "epoch": 5.915018315018315, "grad_norm": 0.80859375, "learning_rate": 9.621792867372237e-05, "loss": 0.137, "step": 8074 }, { "epoch": 5.915750915750916, "grad_norm": 1.234375, "learning_rate": 9.621661499732896e-05, "loss": 0.131, "step": 8075 }, { "epoch": 5.916483516483517, "grad_norm": 1.7109375, "learning_rate": 9.621530110189224e-05, "loss": 0.1416, "step": 8076 }, { "epoch": 5.9172161172161175, "grad_norm": 1.6953125, "learning_rate": 9.62139869874185e-05, "loss": 0.1592, "step": 8077 }, { "epoch": 5.917948717948718, "grad_norm": 1.484375, "learning_rate": 9.621267265391405e-05, "loss": 0.1347, "step": 8078 }, { "epoch": 5.918681318681319, "grad_norm": 1.09375, "learning_rate": 9.621135810138518e-05, "loss": 0.1543, "step": 8079 }, { "epoch": 5.91941391941392, "grad_norm": 1.0, "learning_rate": 9.621004332983817e-05, "loss": 0.1419, "step": 8080 }, { "epoch": 5.92014652014652, "grad_norm": 1.0390625, "learning_rate": 9.620872833927935e-05, "loss": 0.1285, "step": 8081 }, { "epoch": 5.920879120879121, "grad_norm": 1.171875, "learning_rate": 9.6207413129715e-05, "loss": 0.117, "step": 8082 }, { "epoch": 5.921611721611722, "grad_norm": 1.046875, "learning_rate": 9.620609770115143e-05, "loss": 0.1309, "step": 8083 }, { "epoch": 5.922344322344323, "grad_norm": 1.5078125, "learning_rate": 9.620478205359497e-05, "loss": 0.1388, "step": 8084 }, { "epoch": 5.923076923076923, "grad_norm": 0.66015625, "learning_rate": 9.620346618705187e-05, "loss": 0.1319, "step": 8085 }, { "epoch": 5.923809523809524, "grad_norm": 1.1328125, "learning_rate": 9.620215010152845e-05, "loss": 0.1179, "step": 8086 }, { "epoch": 5.924542124542125, "grad_norm": 0.97265625, "learning_rate": 9.620083379703105e-05, "loss": 0.1356, "step": 8087 }, { "epoch": 5.9252747252747255, "grad_norm": 1.6328125, "learning_rate": 9.619951727356592e-05, "loss": 0.142, "step": 8088 }, { "epoch": 5.926007326007326, "grad_norm": 0.828125, "learning_rate": 9.619820053113944e-05, "loss": 0.1618, "step": 8089 }, { "epoch": 5.926739926739927, "grad_norm": 1.6328125, "learning_rate": 9.619688356975784e-05, "loss": 0.1411, "step": 8090 }, { "epoch": 5.927472527472528, "grad_norm": 2.328125, "learning_rate": 9.619556638942749e-05, "loss": 0.1464, "step": 8091 }, { "epoch": 5.9282051282051285, "grad_norm": 1.5390625, "learning_rate": 9.619424899015467e-05, "loss": 0.1248, "step": 8092 }, { "epoch": 5.928937728937729, "grad_norm": 1.015625, "learning_rate": 9.619293137194573e-05, "loss": 0.1502, "step": 8093 }, { "epoch": 5.92967032967033, "grad_norm": 1.046875, "learning_rate": 9.619161353480693e-05, "loss": 0.1224, "step": 8094 }, { "epoch": 5.930402930402931, "grad_norm": 0.72265625, "learning_rate": 9.619029547874463e-05, "loss": 0.1567, "step": 8095 }, { "epoch": 5.931135531135531, "grad_norm": 1.25, "learning_rate": 9.618897720376511e-05, "loss": 0.145, "step": 8096 }, { "epoch": 5.931868131868132, "grad_norm": 1.2421875, "learning_rate": 9.61876587098747e-05, "loss": 0.174, "step": 8097 }, { "epoch": 5.932600732600733, "grad_norm": 1.109375, "learning_rate": 9.618633999707972e-05, "loss": 0.1123, "step": 8098 }, { "epoch": 5.933333333333334, "grad_norm": 0.828125, "learning_rate": 9.61850210653865e-05, "loss": 0.1048, "step": 8099 }, { "epoch": 5.934065934065934, "grad_norm": 1.0859375, "learning_rate": 9.618370191480134e-05, "loss": 0.1677, "step": 8100 }, { "epoch": 5.934798534798535, "grad_norm": 0.73828125, "learning_rate": 9.618238254533055e-05, "loss": 0.1408, "step": 8101 }, { "epoch": 5.935531135531136, "grad_norm": 1.203125, "learning_rate": 9.618106295698051e-05, "loss": 0.1777, "step": 8102 }, { "epoch": 5.9362637362637365, "grad_norm": 0.69921875, "learning_rate": 9.617974314975746e-05, "loss": 0.1406, "step": 8103 }, { "epoch": 5.936996336996337, "grad_norm": 0.7734375, "learning_rate": 9.617842312366778e-05, "loss": 0.1471, "step": 8104 }, { "epoch": 5.937728937728938, "grad_norm": 1.7265625, "learning_rate": 9.617710287871778e-05, "loss": 0.1456, "step": 8105 }, { "epoch": 5.938461538461539, "grad_norm": 1.1484375, "learning_rate": 9.617578241491379e-05, "loss": 0.1772, "step": 8106 }, { "epoch": 5.939194139194139, "grad_norm": 1.1328125, "learning_rate": 9.617446173226212e-05, "loss": 0.1456, "step": 8107 }, { "epoch": 5.93992673992674, "grad_norm": 0.703125, "learning_rate": 9.617314083076911e-05, "loss": 0.1511, "step": 8108 }, { "epoch": 5.940659340659341, "grad_norm": 0.875, "learning_rate": 9.61718197104411e-05, "loss": 0.1439, "step": 8109 }, { "epoch": 5.941391941391942, "grad_norm": 0.6875, "learning_rate": 9.61704983712844e-05, "loss": 0.1235, "step": 8110 }, { "epoch": 5.942124542124542, "grad_norm": 0.96875, "learning_rate": 9.616917681330535e-05, "loss": 0.1665, "step": 8111 }, { "epoch": 5.942857142857143, "grad_norm": 1.015625, "learning_rate": 9.61678550365103e-05, "loss": 0.1463, "step": 8112 }, { "epoch": 5.943589743589744, "grad_norm": 1.6796875, "learning_rate": 9.616653304090554e-05, "loss": 0.1531, "step": 8113 }, { "epoch": 5.9443223443223445, "grad_norm": 1.40625, "learning_rate": 9.616521082649744e-05, "loss": 0.164, "step": 8114 }, { "epoch": 5.945054945054945, "grad_norm": 0.72265625, "learning_rate": 9.616388839329233e-05, "loss": 0.1316, "step": 8115 }, { "epoch": 5.945787545787546, "grad_norm": 1.4765625, "learning_rate": 9.616256574129653e-05, "loss": 0.172, "step": 8116 }, { "epoch": 5.946520146520147, "grad_norm": 1.109375, "learning_rate": 9.61612428705164e-05, "loss": 0.1603, "step": 8117 }, { "epoch": 5.947252747252747, "grad_norm": 1.9296875, "learning_rate": 9.615991978095824e-05, "loss": 0.2008, "step": 8118 }, { "epoch": 5.947985347985348, "grad_norm": 1.21875, "learning_rate": 9.615859647262843e-05, "loss": 0.1259, "step": 8119 }, { "epoch": 5.948717948717949, "grad_norm": 1.8359375, "learning_rate": 9.61572729455333e-05, "loss": 0.142, "step": 8120 }, { "epoch": 5.94945054945055, "grad_norm": 0.78125, "learning_rate": 9.61559491996792e-05, "loss": 0.1454, "step": 8121 }, { "epoch": 5.95018315018315, "grad_norm": 0.57421875, "learning_rate": 9.615462523507243e-05, "loss": 0.1331, "step": 8122 }, { "epoch": 5.950915750915751, "grad_norm": 1.1953125, "learning_rate": 9.615330105171937e-05, "loss": 0.1539, "step": 8123 }, { "epoch": 5.951648351648352, "grad_norm": 1.2109375, "learning_rate": 9.615197664962637e-05, "loss": 0.1355, "step": 8124 }, { "epoch": 5.9523809523809526, "grad_norm": 1.859375, "learning_rate": 9.615065202879977e-05, "loss": 0.1559, "step": 8125 }, { "epoch": 5.953113553113553, "grad_norm": 1.171875, "learning_rate": 9.614932718924589e-05, "loss": 0.1696, "step": 8126 }, { "epoch": 5.953846153846154, "grad_norm": 1.0703125, "learning_rate": 9.61480021309711e-05, "loss": 0.1458, "step": 8127 }, { "epoch": 5.954578754578755, "grad_norm": 1.09375, "learning_rate": 9.614667685398175e-05, "loss": 0.1417, "step": 8128 }, { "epoch": 5.9553113553113555, "grad_norm": 1.0703125, "learning_rate": 9.614535135828419e-05, "loss": 0.1393, "step": 8129 }, { "epoch": 5.956043956043956, "grad_norm": 2.328125, "learning_rate": 9.614402564388478e-05, "loss": 0.1386, "step": 8130 }, { "epoch": 5.956776556776557, "grad_norm": 2.890625, "learning_rate": 9.614269971078983e-05, "loss": 0.1514, "step": 8131 }, { "epoch": 5.957509157509158, "grad_norm": 2.515625, "learning_rate": 9.614137355900575e-05, "loss": 0.1721, "step": 8132 }, { "epoch": 5.958241758241758, "grad_norm": 2.546875, "learning_rate": 9.614004718853884e-05, "loss": 0.1378, "step": 8133 }, { "epoch": 5.958974358974359, "grad_norm": 1.5078125, "learning_rate": 9.613872059939551e-05, "loss": 0.1566, "step": 8134 }, { "epoch": 5.95970695970696, "grad_norm": 1.09375, "learning_rate": 9.613739379158207e-05, "loss": 0.1317, "step": 8135 }, { "epoch": 5.960439560439561, "grad_norm": 1.546875, "learning_rate": 9.613606676510488e-05, "loss": 0.1385, "step": 8136 }, { "epoch": 5.961172161172161, "grad_norm": 2.453125, "learning_rate": 9.613473951997035e-05, "loss": 0.1712, "step": 8137 }, { "epoch": 5.961904761904762, "grad_norm": 1.90625, "learning_rate": 9.613341205618477e-05, "loss": 0.1536, "step": 8138 }, { "epoch": 5.962637362637363, "grad_norm": 1.1484375, "learning_rate": 9.613208437375455e-05, "loss": 0.1196, "step": 8139 }, { "epoch": 5.9633699633699635, "grad_norm": 0.91796875, "learning_rate": 9.613075647268602e-05, "loss": 0.1538, "step": 8140 }, { "epoch": 5.964102564102564, "grad_norm": 0.95703125, "learning_rate": 9.612942835298556e-05, "loss": 0.1449, "step": 8141 }, { "epoch": 5.964835164835165, "grad_norm": 1.2421875, "learning_rate": 9.612810001465954e-05, "loss": 0.1695, "step": 8142 }, { "epoch": 5.965567765567766, "grad_norm": 1.9296875, "learning_rate": 9.61267714577143e-05, "loss": 0.1323, "step": 8143 }, { "epoch": 5.966300366300366, "grad_norm": 1.5703125, "learning_rate": 9.612544268215624e-05, "loss": 0.1471, "step": 8144 }, { "epoch": 5.967032967032967, "grad_norm": 1.484375, "learning_rate": 9.61241136879917e-05, "loss": 0.1389, "step": 8145 }, { "epoch": 5.967765567765568, "grad_norm": 1.2265625, "learning_rate": 9.612278447522702e-05, "loss": 0.1597, "step": 8146 }, { "epoch": 5.968498168498169, "grad_norm": 1.0390625, "learning_rate": 9.612145504386864e-05, "loss": 0.1363, "step": 8147 }, { "epoch": 5.969230769230769, "grad_norm": 1.609375, "learning_rate": 9.612012539392289e-05, "loss": 0.1247, "step": 8148 }, { "epoch": 5.96996336996337, "grad_norm": 1.2890625, "learning_rate": 9.611879552539613e-05, "loss": 0.1831, "step": 8149 }, { "epoch": 5.970695970695971, "grad_norm": 1.4921875, "learning_rate": 9.611746543829475e-05, "loss": 0.1778, "step": 8150 }, { "epoch": 5.9714285714285715, "grad_norm": 0.8828125, "learning_rate": 9.611613513262511e-05, "loss": 0.1416, "step": 8151 }, { "epoch": 5.972161172161172, "grad_norm": 0.93359375, "learning_rate": 9.61148046083936e-05, "loss": 0.1164, "step": 8152 }, { "epoch": 5.972893772893773, "grad_norm": 0.875, "learning_rate": 9.611347386560659e-05, "loss": 0.133, "step": 8153 }, { "epoch": 5.973626373626374, "grad_norm": 0.80078125, "learning_rate": 9.611214290427045e-05, "loss": 0.1321, "step": 8154 }, { "epoch": 5.9743589743589745, "grad_norm": 0.90234375, "learning_rate": 9.611081172439157e-05, "loss": 0.148, "step": 8155 }, { "epoch": 5.975091575091575, "grad_norm": 0.953125, "learning_rate": 9.61094803259763e-05, "loss": 0.1404, "step": 8156 }, { "epoch": 5.975824175824176, "grad_norm": 1.015625, "learning_rate": 9.610814870903106e-05, "loss": 0.1595, "step": 8157 }, { "epoch": 5.976556776556777, "grad_norm": 0.703125, "learning_rate": 9.610681687356218e-05, "loss": 0.1304, "step": 8158 }, { "epoch": 5.977289377289377, "grad_norm": 0.91015625, "learning_rate": 9.610548481957608e-05, "loss": 0.1589, "step": 8159 }, { "epoch": 5.978021978021978, "grad_norm": 0.8671875, "learning_rate": 9.610415254707914e-05, "loss": 0.1271, "step": 8160 }, { "epoch": 5.978754578754579, "grad_norm": 0.79296875, "learning_rate": 9.610282005607771e-05, "loss": 0.1416, "step": 8161 }, { "epoch": 5.97948717948718, "grad_norm": 1.2734375, "learning_rate": 9.610148734657821e-05, "loss": 0.1476, "step": 8162 }, { "epoch": 5.98021978021978, "grad_norm": 1.015625, "learning_rate": 9.610015441858703e-05, "loss": 0.1606, "step": 8163 }, { "epoch": 5.980952380952381, "grad_norm": 0.6015625, "learning_rate": 9.609882127211053e-05, "loss": 0.1417, "step": 8164 }, { "epoch": 5.981684981684982, "grad_norm": 0.7265625, "learning_rate": 9.60974879071551e-05, "loss": 0.1594, "step": 8165 }, { "epoch": 5.9824175824175825, "grad_norm": 0.8828125, "learning_rate": 9.609615432372715e-05, "loss": 0.1659, "step": 8166 }, { "epoch": 5.983150183150183, "grad_norm": 1.5859375, "learning_rate": 9.609482052183305e-05, "loss": 0.1472, "step": 8167 }, { "epoch": 5.983882783882784, "grad_norm": 1.1953125, "learning_rate": 9.609348650147919e-05, "loss": 0.1346, "step": 8168 }, { "epoch": 5.984615384615385, "grad_norm": 0.64453125, "learning_rate": 9.609215226267198e-05, "loss": 0.1514, "step": 8169 }, { "epoch": 5.985347985347985, "grad_norm": 1.796875, "learning_rate": 9.60908178054178e-05, "loss": 0.1359, "step": 8170 }, { "epoch": 5.986080586080586, "grad_norm": 1.1875, "learning_rate": 9.608948312972303e-05, "loss": 0.1575, "step": 8171 }, { "epoch": 5.986813186813187, "grad_norm": 0.6640625, "learning_rate": 9.60881482355941e-05, "loss": 0.1441, "step": 8172 }, { "epoch": 5.987545787545788, "grad_norm": 0.83203125, "learning_rate": 9.60868131230374e-05, "loss": 0.1396, "step": 8173 }, { "epoch": 5.988278388278388, "grad_norm": 1.015625, "learning_rate": 9.608547779205927e-05, "loss": 0.1378, "step": 8174 }, { "epoch": 5.989010989010989, "grad_norm": 0.91796875, "learning_rate": 9.608414224266617e-05, "loss": 0.1527, "step": 8175 }, { "epoch": 5.98974358974359, "grad_norm": 1.1484375, "learning_rate": 9.608280647486449e-05, "loss": 0.1583, "step": 8176 }, { "epoch": 5.9904761904761905, "grad_norm": 0.84765625, "learning_rate": 9.608147048866062e-05, "loss": 0.1784, "step": 8177 }, { "epoch": 5.991208791208791, "grad_norm": 0.7265625, "learning_rate": 9.608013428406096e-05, "loss": 0.1577, "step": 8178 }, { "epoch": 5.991941391941392, "grad_norm": 0.96484375, "learning_rate": 9.60787978610719e-05, "loss": 0.1403, "step": 8179 }, { "epoch": 5.992673992673993, "grad_norm": 0.71875, "learning_rate": 9.607746121969986e-05, "loss": 0.1221, "step": 8180 }, { "epoch": 5.993406593406593, "grad_norm": 0.83203125, "learning_rate": 9.607612435995126e-05, "loss": 0.1655, "step": 8181 }, { "epoch": 5.994139194139194, "grad_norm": 1.3828125, "learning_rate": 9.607478728183247e-05, "loss": 0.145, "step": 8182 }, { "epoch": 5.994871794871795, "grad_norm": 1.0390625, "learning_rate": 9.607344998534993e-05, "loss": 0.1574, "step": 8183 }, { "epoch": 5.995604395604396, "grad_norm": 0.9453125, "learning_rate": 9.607211247051001e-05, "loss": 0.1276, "step": 8184 }, { "epoch": 5.996336996336996, "grad_norm": 1.109375, "learning_rate": 9.607077473731913e-05, "loss": 0.1648, "step": 8185 }, { "epoch": 5.997069597069597, "grad_norm": 0.7734375, "learning_rate": 9.606943678578373e-05, "loss": 0.1192, "step": 8186 }, { "epoch": 5.997802197802198, "grad_norm": 1.15625, "learning_rate": 9.606809861591021e-05, "loss": 0.1814, "step": 8187 }, { "epoch": 5.9985347985347985, "grad_norm": 1.3671875, "learning_rate": 9.606676022770495e-05, "loss": 0.1316, "step": 8188 }, { "epoch": 5.999267399267399, "grad_norm": 1.1640625, "learning_rate": 9.606542162117438e-05, "loss": 0.162, "step": 8189 }, { "epoch": 6.0, "grad_norm": 1.8046875, "learning_rate": 9.606408279632492e-05, "loss": 0.1408, "step": 8190 }, { "epoch": 6.000732600732601, "grad_norm": 1.4453125, "learning_rate": 9.6062743753163e-05, "loss": 0.174, "step": 8191 }, { "epoch": 6.0014652014652015, "grad_norm": 0.98046875, "learning_rate": 9.6061404491695e-05, "loss": 0.1561, "step": 8192 }, { "epoch": 6.002197802197802, "grad_norm": 1.078125, "learning_rate": 9.606006501192736e-05, "loss": 0.1402, "step": 8193 }, { "epoch": 6.002930402930403, "grad_norm": 1.015625, "learning_rate": 9.605872531386648e-05, "loss": 0.1441, "step": 8194 }, { "epoch": 6.003663003663004, "grad_norm": 1.03125, "learning_rate": 9.605738539751881e-05, "loss": 0.1412, "step": 8195 }, { "epoch": 6.004395604395604, "grad_norm": 2.515625, "learning_rate": 9.605604526289074e-05, "loss": 0.1274, "step": 8196 }, { "epoch": 6.005128205128205, "grad_norm": 0.96875, "learning_rate": 9.60547049099887e-05, "loss": 0.1206, "step": 8197 }, { "epoch": 6.005860805860806, "grad_norm": 0.8671875, "learning_rate": 9.605336433881912e-05, "loss": 0.1247, "step": 8198 }, { "epoch": 6.006593406593407, "grad_norm": 1.0234375, "learning_rate": 9.605202354938841e-05, "loss": 0.1586, "step": 8199 }, { "epoch": 6.007326007326007, "grad_norm": 0.87109375, "learning_rate": 9.605068254170301e-05, "loss": 0.1565, "step": 8200 }, { "epoch": 6.008058608058608, "grad_norm": 1.8359375, "learning_rate": 9.604934131576934e-05, "loss": 0.158, "step": 8201 }, { "epoch": 6.008791208791209, "grad_norm": 1.8046875, "learning_rate": 9.60479998715938e-05, "loss": 0.1681, "step": 8202 }, { "epoch": 6.0095238095238095, "grad_norm": 1.125, "learning_rate": 9.604665820918287e-05, "loss": 0.1386, "step": 8203 }, { "epoch": 6.01025641025641, "grad_norm": 0.9765625, "learning_rate": 9.604531632854293e-05, "loss": 0.1435, "step": 8204 }, { "epoch": 6.010989010989011, "grad_norm": 0.75, "learning_rate": 9.604397422968044e-05, "loss": 0.1348, "step": 8205 }, { "epoch": 6.011721611721612, "grad_norm": 0.8046875, "learning_rate": 9.60426319126018e-05, "loss": 0.1266, "step": 8206 }, { "epoch": 6.012454212454212, "grad_norm": 0.92578125, "learning_rate": 9.604128937731346e-05, "loss": 0.1357, "step": 8207 }, { "epoch": 6.013186813186813, "grad_norm": 1.234375, "learning_rate": 9.603994662382187e-05, "loss": 0.1774, "step": 8208 }, { "epoch": 6.013919413919414, "grad_norm": 1.4765625, "learning_rate": 9.603860365213344e-05, "loss": 0.1531, "step": 8209 }, { "epoch": 6.014652014652015, "grad_norm": 0.921875, "learning_rate": 9.603726046225462e-05, "loss": 0.1398, "step": 8210 }, { "epoch": 6.015384615384615, "grad_norm": 1.078125, "learning_rate": 9.603591705419183e-05, "loss": 0.1608, "step": 8211 }, { "epoch": 6.016117216117216, "grad_norm": 1.15625, "learning_rate": 9.603457342795151e-05, "loss": 0.14, "step": 8212 }, { "epoch": 6.016849816849817, "grad_norm": 1.953125, "learning_rate": 9.603322958354008e-05, "loss": 0.1662, "step": 8213 }, { "epoch": 6.0175824175824175, "grad_norm": 1.8203125, "learning_rate": 9.603188552096403e-05, "loss": 0.1737, "step": 8214 }, { "epoch": 6.018315018315018, "grad_norm": 1.203125, "learning_rate": 9.603054124022976e-05, "loss": 0.1817, "step": 8215 }, { "epoch": 6.019047619047619, "grad_norm": 1.1953125, "learning_rate": 9.602919674134371e-05, "loss": 0.1181, "step": 8216 }, { "epoch": 6.01978021978022, "grad_norm": 0.8515625, "learning_rate": 9.602785202431234e-05, "loss": 0.1216, "step": 8217 }, { "epoch": 6.02051282051282, "grad_norm": 0.91796875, "learning_rate": 9.60265070891421e-05, "loss": 0.1224, "step": 8218 }, { "epoch": 6.021245421245421, "grad_norm": 1.234375, "learning_rate": 9.60251619358394e-05, "loss": 0.1473, "step": 8219 }, { "epoch": 6.021978021978022, "grad_norm": 0.765625, "learning_rate": 9.60238165644107e-05, "loss": 0.1409, "step": 8220 }, { "epoch": 6.022710622710623, "grad_norm": 1.1484375, "learning_rate": 9.602247097486247e-05, "loss": 0.1481, "step": 8221 }, { "epoch": 6.023443223443223, "grad_norm": 0.98046875, "learning_rate": 9.602112516720113e-05, "loss": 0.1631, "step": 8222 }, { "epoch": 6.024175824175824, "grad_norm": 0.87109375, "learning_rate": 9.601977914143312e-05, "loss": 0.1414, "step": 8223 }, { "epoch": 6.024908424908425, "grad_norm": 1.8359375, "learning_rate": 9.601843289756492e-05, "loss": 0.1758, "step": 8224 }, { "epoch": 6.0256410256410255, "grad_norm": 0.9609375, "learning_rate": 9.601708643560296e-05, "loss": 0.148, "step": 8225 }, { "epoch": 6.026373626373626, "grad_norm": 0.796875, "learning_rate": 9.60157397555537e-05, "loss": 0.1664, "step": 8226 }, { "epoch": 6.027106227106227, "grad_norm": 1.0546875, "learning_rate": 9.601439285742359e-05, "loss": 0.1805, "step": 8227 }, { "epoch": 6.027838827838828, "grad_norm": 1.0859375, "learning_rate": 9.601304574121908e-05, "loss": 0.1119, "step": 8228 }, { "epoch": 6.0285714285714285, "grad_norm": 1.0859375, "learning_rate": 9.601169840694664e-05, "loss": 0.1626, "step": 8229 }, { "epoch": 6.029304029304029, "grad_norm": 1.46875, "learning_rate": 9.601035085461268e-05, "loss": 0.1422, "step": 8230 }, { "epoch": 6.03003663003663, "grad_norm": 1.5390625, "learning_rate": 9.600900308422372e-05, "loss": 0.1496, "step": 8231 }, { "epoch": 6.030769230769231, "grad_norm": 1.5546875, "learning_rate": 9.600765509578617e-05, "loss": 0.1673, "step": 8232 }, { "epoch": 6.031501831501831, "grad_norm": 0.984375, "learning_rate": 9.600630688930653e-05, "loss": 0.12, "step": 8233 }, { "epoch": 6.032234432234432, "grad_norm": 1.21875, "learning_rate": 9.60049584647912e-05, "loss": 0.1156, "step": 8234 }, { "epoch": 6.032967032967033, "grad_norm": 1.5546875, "learning_rate": 9.60036098222467e-05, "loss": 0.1218, "step": 8235 }, { "epoch": 6.033699633699634, "grad_norm": 1.765625, "learning_rate": 9.600226096167946e-05, "loss": 0.1525, "step": 8236 }, { "epoch": 6.034432234432234, "grad_norm": 1.609375, "learning_rate": 9.600091188309595e-05, "loss": 0.1599, "step": 8237 }, { "epoch": 6.035164835164835, "grad_norm": 1.3515625, "learning_rate": 9.599956258650264e-05, "loss": 0.1706, "step": 8238 }, { "epoch": 6.035897435897436, "grad_norm": 1.1640625, "learning_rate": 9.599821307190597e-05, "loss": 0.1568, "step": 8239 }, { "epoch": 6.0366300366300365, "grad_norm": 1.3671875, "learning_rate": 9.599686333931244e-05, "loss": 0.1425, "step": 8240 }, { "epoch": 6.037362637362637, "grad_norm": 1.359375, "learning_rate": 9.59955133887285e-05, "loss": 0.1335, "step": 8241 }, { "epoch": 6.038095238095238, "grad_norm": 1.2734375, "learning_rate": 9.599416322016063e-05, "loss": 0.1436, "step": 8242 }, { "epoch": 6.038827838827839, "grad_norm": 2.640625, "learning_rate": 9.599281283361527e-05, "loss": 0.169, "step": 8243 }, { "epoch": 6.039560439560439, "grad_norm": 1.28125, "learning_rate": 9.599146222909893e-05, "loss": 0.1287, "step": 8244 }, { "epoch": 6.04029304029304, "grad_norm": 1.0703125, "learning_rate": 9.599011140661803e-05, "loss": 0.1421, "step": 8245 }, { "epoch": 6.041025641025641, "grad_norm": 1.3125, "learning_rate": 9.598876036617911e-05, "loss": 0.1476, "step": 8246 }, { "epoch": 6.041758241758242, "grad_norm": 1.4296875, "learning_rate": 9.598740910778857e-05, "loss": 0.1528, "step": 8247 }, { "epoch": 6.042490842490842, "grad_norm": 1.5703125, "learning_rate": 9.598605763145294e-05, "loss": 0.1538, "step": 8248 }, { "epoch": 6.043223443223443, "grad_norm": 1.78125, "learning_rate": 9.598470593717869e-05, "loss": 0.1617, "step": 8249 }, { "epoch": 6.043956043956044, "grad_norm": 1.4921875, "learning_rate": 9.598335402497227e-05, "loss": 0.1635, "step": 8250 }, { "epoch": 6.0446886446886445, "grad_norm": 1.4140625, "learning_rate": 9.598200189484016e-05, "loss": 0.1404, "step": 8251 }, { "epoch": 6.045421245421245, "grad_norm": 1.109375, "learning_rate": 9.598064954678886e-05, "loss": 0.1543, "step": 8252 }, { "epoch": 6.046153846153846, "grad_norm": 1.5, "learning_rate": 9.597929698082483e-05, "loss": 0.1481, "step": 8253 }, { "epoch": 6.046886446886447, "grad_norm": 1.4921875, "learning_rate": 9.597794419695456e-05, "loss": 0.1346, "step": 8254 }, { "epoch": 6.0476190476190474, "grad_norm": 1.1953125, "learning_rate": 9.597659119518454e-05, "loss": 0.1353, "step": 8255 }, { "epoch": 6.048351648351648, "grad_norm": 1.1015625, "learning_rate": 9.597523797552122e-05, "loss": 0.1464, "step": 8256 }, { "epoch": 6.049084249084249, "grad_norm": 1.7578125, "learning_rate": 9.597388453797113e-05, "loss": 0.1473, "step": 8257 }, { "epoch": 6.04981684981685, "grad_norm": 1.078125, "learning_rate": 9.597253088254071e-05, "loss": 0.1366, "step": 8258 }, { "epoch": 6.05054945054945, "grad_norm": 1.4921875, "learning_rate": 9.597117700923649e-05, "loss": 0.1376, "step": 8259 }, { "epoch": 6.051282051282051, "grad_norm": 0.66796875, "learning_rate": 9.596982291806491e-05, "loss": 0.1337, "step": 8260 }, { "epoch": 6.052014652014652, "grad_norm": 2.0625, "learning_rate": 9.596846860903249e-05, "loss": 0.179, "step": 8261 }, { "epoch": 6.052747252747253, "grad_norm": 1.2890625, "learning_rate": 9.596711408214572e-05, "loss": 0.1636, "step": 8262 }, { "epoch": 6.053479853479853, "grad_norm": 1.296875, "learning_rate": 9.596575933741107e-05, "loss": 0.1728, "step": 8263 }, { "epoch": 6.054212454212454, "grad_norm": 1.1796875, "learning_rate": 9.596440437483505e-05, "loss": 0.1332, "step": 8264 }, { "epoch": 6.054945054945055, "grad_norm": 1.0234375, "learning_rate": 9.596304919442413e-05, "loss": 0.1614, "step": 8265 }, { "epoch": 6.0556776556776555, "grad_norm": 1.09375, "learning_rate": 9.596169379618482e-05, "loss": 0.1513, "step": 8266 }, { "epoch": 6.056410256410256, "grad_norm": 1.078125, "learning_rate": 9.596033818012363e-05, "loss": 0.1778, "step": 8267 }, { "epoch": 6.057142857142857, "grad_norm": 1.3828125, "learning_rate": 9.595898234624702e-05, "loss": 0.1698, "step": 8268 }, { "epoch": 6.057875457875458, "grad_norm": 1.4453125, "learning_rate": 9.59576262945615e-05, "loss": 0.1473, "step": 8269 }, { "epoch": 6.058608058608058, "grad_norm": 0.875, "learning_rate": 9.595627002507358e-05, "loss": 0.1708, "step": 8270 }, { "epoch": 6.059340659340659, "grad_norm": 1.1875, "learning_rate": 9.595491353778975e-05, "loss": 0.1502, "step": 8271 }, { "epoch": 6.06007326007326, "grad_norm": 0.74609375, "learning_rate": 9.59535568327165e-05, "loss": 0.1348, "step": 8272 }, { "epoch": 6.060805860805861, "grad_norm": 0.98046875, "learning_rate": 9.595219990986033e-05, "loss": 0.1555, "step": 8273 }, { "epoch": 6.061538461538461, "grad_norm": 1.15625, "learning_rate": 9.595084276922776e-05, "loss": 0.2042, "step": 8274 }, { "epoch": 6.062271062271062, "grad_norm": 1.2890625, "learning_rate": 9.594948541082529e-05, "loss": 0.1435, "step": 8275 }, { "epoch": 6.063003663003663, "grad_norm": 0.91015625, "learning_rate": 9.594812783465939e-05, "loss": 0.1196, "step": 8276 }, { "epoch": 6.0637362637362635, "grad_norm": 0.98828125, "learning_rate": 9.594677004073661e-05, "loss": 0.1457, "step": 8277 }, { "epoch": 6.064468864468864, "grad_norm": 1.203125, "learning_rate": 9.594541202906343e-05, "loss": 0.1439, "step": 8278 }, { "epoch": 6.065201465201465, "grad_norm": 1.28125, "learning_rate": 9.594405379964637e-05, "loss": 0.1428, "step": 8279 }, { "epoch": 6.065934065934066, "grad_norm": 0.921875, "learning_rate": 9.594269535249192e-05, "loss": 0.1224, "step": 8280 }, { "epoch": 6.066666666666666, "grad_norm": 1.4921875, "learning_rate": 9.59413366876066e-05, "loss": 0.1865, "step": 8281 }, { "epoch": 6.067399267399267, "grad_norm": 1.8359375, "learning_rate": 9.593997780499691e-05, "loss": 0.1295, "step": 8282 }, { "epoch": 6.068131868131868, "grad_norm": 0.9375, "learning_rate": 9.59386187046694e-05, "loss": 0.1142, "step": 8283 }, { "epoch": 6.068864468864469, "grad_norm": 1.4375, "learning_rate": 9.59372593866305e-05, "loss": 0.1634, "step": 8284 }, { "epoch": 6.069597069597069, "grad_norm": 0.8515625, "learning_rate": 9.59358998508868e-05, "loss": 0.1535, "step": 8285 }, { "epoch": 6.07032967032967, "grad_norm": 0.95703125, "learning_rate": 9.593454009744482e-05, "loss": 0.1496, "step": 8286 }, { "epoch": 6.071062271062271, "grad_norm": 0.625, "learning_rate": 9.593318012631101e-05, "loss": 0.1282, "step": 8287 }, { "epoch": 6.0717948717948715, "grad_norm": 0.859375, "learning_rate": 9.593181993749193e-05, "loss": 0.1166, "step": 8288 }, { "epoch": 6.072527472527472, "grad_norm": 0.7890625, "learning_rate": 9.593045953099408e-05, "loss": 0.1538, "step": 8289 }, { "epoch": 6.073260073260073, "grad_norm": 1.296875, "learning_rate": 9.592909890682398e-05, "loss": 0.134, "step": 8290 }, { "epoch": 6.073992673992674, "grad_norm": 0.73828125, "learning_rate": 9.592773806498815e-05, "loss": 0.1452, "step": 8291 }, { "epoch": 6.0747252747252745, "grad_norm": 1.0, "learning_rate": 9.592637700549312e-05, "loss": 0.1325, "step": 8292 }, { "epoch": 6.075457875457875, "grad_norm": 1.3203125, "learning_rate": 9.592501572834541e-05, "loss": 0.1856, "step": 8293 }, { "epoch": 6.076190476190476, "grad_norm": 0.7890625, "learning_rate": 9.592365423355154e-05, "loss": 0.1474, "step": 8294 }, { "epoch": 6.076923076923077, "grad_norm": 1.5703125, "learning_rate": 9.592229252111803e-05, "loss": 0.1804, "step": 8295 }, { "epoch": 6.077655677655677, "grad_norm": 0.97265625, "learning_rate": 9.592093059105142e-05, "loss": 0.1281, "step": 8296 }, { "epoch": 6.078388278388278, "grad_norm": 0.90234375, "learning_rate": 9.591956844335818e-05, "loss": 0.1268, "step": 8297 }, { "epoch": 6.079120879120879, "grad_norm": 0.86328125, "learning_rate": 9.591820607804491e-05, "loss": 0.1371, "step": 8298 }, { "epoch": 6.07985347985348, "grad_norm": 0.88671875, "learning_rate": 9.59168434951181e-05, "loss": 0.1549, "step": 8299 }, { "epoch": 6.08058608058608, "grad_norm": 0.98828125, "learning_rate": 9.591548069458427e-05, "loss": 0.1154, "step": 8300 }, { "epoch": 6.081318681318681, "grad_norm": 1.25, "learning_rate": 9.591411767644997e-05, "loss": 0.123, "step": 8301 }, { "epoch": 6.082051282051282, "grad_norm": 0.91796875, "learning_rate": 9.591275444072174e-05, "loss": 0.1552, "step": 8302 }, { "epoch": 6.0827838827838825, "grad_norm": 1.1875, "learning_rate": 9.591139098740608e-05, "loss": 0.1447, "step": 8303 }, { "epoch": 6.083516483516483, "grad_norm": 0.828125, "learning_rate": 9.591002731650954e-05, "loss": 0.1589, "step": 8304 }, { "epoch": 6.084249084249084, "grad_norm": 0.7578125, "learning_rate": 9.590866342803865e-05, "loss": 0.1377, "step": 8305 }, { "epoch": 6.084981684981685, "grad_norm": 0.859375, "learning_rate": 9.590729932199994e-05, "loss": 0.1543, "step": 8306 }, { "epoch": 6.085714285714285, "grad_norm": 0.78125, "learning_rate": 9.590593499839996e-05, "loss": 0.1823, "step": 8307 }, { "epoch": 6.086446886446886, "grad_norm": 0.796875, "learning_rate": 9.590457045724525e-05, "loss": 0.1463, "step": 8308 }, { "epoch": 6.087179487179487, "grad_norm": 1.1328125, "learning_rate": 9.590320569854233e-05, "loss": 0.1507, "step": 8309 }, { "epoch": 6.087912087912088, "grad_norm": 0.9453125, "learning_rate": 9.590184072229773e-05, "loss": 0.134, "step": 8310 }, { "epoch": 6.088644688644688, "grad_norm": 1.0703125, "learning_rate": 9.590047552851802e-05, "loss": 0.1804, "step": 8311 }, { "epoch": 6.089377289377289, "grad_norm": 1.375, "learning_rate": 9.589911011720974e-05, "loss": 0.1861, "step": 8312 }, { "epoch": 6.09010989010989, "grad_norm": 1.4296875, "learning_rate": 9.58977444883794e-05, "loss": 0.1671, "step": 8313 }, { "epoch": 6.0908424908424905, "grad_norm": 1.2578125, "learning_rate": 9.589637864203358e-05, "loss": 0.1471, "step": 8314 }, { "epoch": 6.091575091575091, "grad_norm": 0.890625, "learning_rate": 9.589501257817882e-05, "loss": 0.1547, "step": 8315 }, { "epoch": 6.092307692307692, "grad_norm": 1.4765625, "learning_rate": 9.589364629682161e-05, "loss": 0.1866, "step": 8316 }, { "epoch": 6.093040293040293, "grad_norm": 1.171875, "learning_rate": 9.589227979796855e-05, "loss": 0.1147, "step": 8317 }, { "epoch": 6.093772893772893, "grad_norm": 0.921875, "learning_rate": 9.58909130816262e-05, "loss": 0.1439, "step": 8318 }, { "epoch": 6.094505494505494, "grad_norm": 0.796875, "learning_rate": 9.588954614780107e-05, "loss": 0.1208, "step": 8319 }, { "epoch": 6.095238095238095, "grad_norm": 1.1015625, "learning_rate": 9.588817899649974e-05, "loss": 0.1211, "step": 8320 }, { "epoch": 6.095970695970696, "grad_norm": 0.65234375, "learning_rate": 9.588681162772873e-05, "loss": 0.1197, "step": 8321 }, { "epoch": 6.096703296703296, "grad_norm": 1.0, "learning_rate": 9.58854440414946e-05, "loss": 0.1571, "step": 8322 }, { "epoch": 6.097435897435897, "grad_norm": 0.796875, "learning_rate": 9.58840762378039e-05, "loss": 0.1221, "step": 8323 }, { "epoch": 6.098168498168498, "grad_norm": 1.75, "learning_rate": 9.588270821666321e-05, "loss": 0.1654, "step": 8324 }, { "epoch": 6.0989010989010985, "grad_norm": 1.078125, "learning_rate": 9.588133997807907e-05, "loss": 0.1324, "step": 8325 }, { "epoch": 6.099633699633699, "grad_norm": 1.8828125, "learning_rate": 9.587997152205803e-05, "loss": 0.1717, "step": 8326 }, { "epoch": 6.1003663003663, "grad_norm": 1.109375, "learning_rate": 9.587860284860664e-05, "loss": 0.1425, "step": 8327 }, { "epoch": 6.101098901098901, "grad_norm": 0.6640625, "learning_rate": 9.587723395773147e-05, "loss": 0.128, "step": 8328 }, { "epoch": 6.1018315018315015, "grad_norm": 1.4765625, "learning_rate": 9.587586484943908e-05, "loss": 0.1604, "step": 8329 }, { "epoch": 6.102564102564102, "grad_norm": 1.3828125, "learning_rate": 9.587449552373603e-05, "loss": 0.1213, "step": 8330 }, { "epoch": 6.103296703296703, "grad_norm": 1.5, "learning_rate": 9.587312598062886e-05, "loss": 0.1647, "step": 8331 }, { "epoch": 6.104029304029304, "grad_norm": 1.015625, "learning_rate": 9.587175622012416e-05, "loss": 0.1491, "step": 8332 }, { "epoch": 6.104761904761904, "grad_norm": 1.65625, "learning_rate": 9.587038624222848e-05, "loss": 0.1472, "step": 8333 }, { "epoch": 6.105494505494505, "grad_norm": 1.203125, "learning_rate": 9.586901604694838e-05, "loss": 0.1361, "step": 8334 }, { "epoch": 6.106227106227106, "grad_norm": 1.46875, "learning_rate": 9.586764563429045e-05, "loss": 0.1784, "step": 8335 }, { "epoch": 6.106959706959707, "grad_norm": 1.3671875, "learning_rate": 9.586627500426122e-05, "loss": 0.1467, "step": 8336 }, { "epoch": 6.107692307692307, "grad_norm": 0.8359375, "learning_rate": 9.586490415686726e-05, "loss": 0.1117, "step": 8337 }, { "epoch": 6.108424908424908, "grad_norm": 0.9921875, "learning_rate": 9.586353309211517e-05, "loss": 0.1375, "step": 8338 }, { "epoch": 6.109157509157509, "grad_norm": 0.67578125, "learning_rate": 9.586216181001151e-05, "loss": 0.1479, "step": 8339 }, { "epoch": 6.1098901098901095, "grad_norm": 0.84765625, "learning_rate": 9.586079031056284e-05, "loss": 0.1434, "step": 8340 }, { "epoch": 6.11062271062271, "grad_norm": 0.84765625, "learning_rate": 9.585941859377572e-05, "loss": 0.1532, "step": 8341 }, { "epoch": 6.111355311355311, "grad_norm": 1.4296875, "learning_rate": 9.585804665965672e-05, "loss": 0.1836, "step": 8342 }, { "epoch": 6.112087912087912, "grad_norm": 1.2890625, "learning_rate": 9.585667450821245e-05, "loss": 0.1493, "step": 8343 }, { "epoch": 6.112820512820512, "grad_norm": 1.6953125, "learning_rate": 9.585530213944946e-05, "loss": 0.1648, "step": 8344 }, { "epoch": 6.113553113553113, "grad_norm": 0.8359375, "learning_rate": 9.585392955337432e-05, "loss": 0.1493, "step": 8345 }, { "epoch": 6.114285714285714, "grad_norm": 1.1484375, "learning_rate": 9.585255674999364e-05, "loss": 0.1466, "step": 8346 }, { "epoch": 6.115018315018315, "grad_norm": 0.89453125, "learning_rate": 9.585118372931394e-05, "loss": 0.1655, "step": 8347 }, { "epoch": 6.115750915750915, "grad_norm": 0.98046875, "learning_rate": 9.584981049134184e-05, "loss": 0.1441, "step": 8348 }, { "epoch": 6.116483516483516, "grad_norm": 0.69921875, "learning_rate": 9.584843703608393e-05, "loss": 0.1356, "step": 8349 }, { "epoch": 6.117216117216117, "grad_norm": 0.57421875, "learning_rate": 9.584706336354674e-05, "loss": 0.125, "step": 8350 }, { "epoch": 6.1179487179487175, "grad_norm": 0.8359375, "learning_rate": 9.584568947373691e-05, "loss": 0.1418, "step": 8351 }, { "epoch": 6.118681318681318, "grad_norm": 0.828125, "learning_rate": 9.584431536666098e-05, "loss": 0.1481, "step": 8352 }, { "epoch": 6.119413919413919, "grad_norm": 1.375, "learning_rate": 9.584294104232556e-05, "loss": 0.1592, "step": 8353 }, { "epoch": 6.12014652014652, "grad_norm": 0.78125, "learning_rate": 9.584156650073724e-05, "loss": 0.151, "step": 8354 }, { "epoch": 6.1208791208791204, "grad_norm": 0.9765625, "learning_rate": 9.584019174190255e-05, "loss": 0.1389, "step": 8355 }, { "epoch": 6.121611721611721, "grad_norm": 0.9921875, "learning_rate": 9.583881676582813e-05, "loss": 0.1158, "step": 8356 }, { "epoch": 6.122344322344322, "grad_norm": 1.1953125, "learning_rate": 9.583744157252056e-05, "loss": 0.1188, "step": 8357 }, { "epoch": 6.123076923076923, "grad_norm": 0.62109375, "learning_rate": 9.583606616198642e-05, "loss": 0.1399, "step": 8358 }, { "epoch": 6.123809523809523, "grad_norm": 1.1484375, "learning_rate": 9.583469053423232e-05, "loss": 0.1432, "step": 8359 }, { "epoch": 6.124542124542124, "grad_norm": 1.203125, "learning_rate": 9.583331468926483e-05, "loss": 0.1368, "step": 8360 }, { "epoch": 6.125274725274725, "grad_norm": 1.140625, "learning_rate": 9.583193862709055e-05, "loss": 0.1466, "step": 8361 }, { "epoch": 6.126007326007326, "grad_norm": 1.109375, "learning_rate": 9.583056234771605e-05, "loss": 0.1548, "step": 8362 }, { "epoch": 6.126739926739926, "grad_norm": 1.5546875, "learning_rate": 9.582918585114797e-05, "loss": 0.1406, "step": 8363 }, { "epoch": 6.127472527472527, "grad_norm": 1.234375, "learning_rate": 9.582780913739287e-05, "loss": 0.1359, "step": 8364 }, { "epoch": 6.128205128205128, "grad_norm": 1.2265625, "learning_rate": 9.582643220645734e-05, "loss": 0.1459, "step": 8365 }, { "epoch": 6.128937728937729, "grad_norm": 1.546875, "learning_rate": 9.582505505834802e-05, "loss": 0.1768, "step": 8366 }, { "epoch": 6.12967032967033, "grad_norm": 0.73828125, "learning_rate": 9.582367769307145e-05, "loss": 0.1221, "step": 8367 }, { "epoch": 6.130402930402931, "grad_norm": 1.0390625, "learning_rate": 9.58223001106343e-05, "loss": 0.1698, "step": 8368 }, { "epoch": 6.131135531135532, "grad_norm": 0.94140625, "learning_rate": 9.582092231104312e-05, "loss": 0.094, "step": 8369 }, { "epoch": 6.131868131868132, "grad_norm": 1.5390625, "learning_rate": 9.58195442943045e-05, "loss": 0.1702, "step": 8370 }, { "epoch": 6.132600732600733, "grad_norm": 2.1875, "learning_rate": 9.581816606042509e-05, "loss": 0.1605, "step": 8371 }, { "epoch": 6.133333333333334, "grad_norm": 1.5546875, "learning_rate": 9.581678760941147e-05, "loss": 0.1736, "step": 8372 }, { "epoch": 6.1340659340659345, "grad_norm": 0.9375, "learning_rate": 9.581540894127024e-05, "loss": 0.1578, "step": 8373 }, { "epoch": 6.134798534798535, "grad_norm": 0.68359375, "learning_rate": 9.5814030056008e-05, "loss": 0.1248, "step": 8374 }, { "epoch": 6.135531135531136, "grad_norm": 1.0390625, "learning_rate": 9.581265095363137e-05, "loss": 0.1834, "step": 8375 }, { "epoch": 6.136263736263737, "grad_norm": 1.21875, "learning_rate": 9.581127163414695e-05, "loss": 0.1523, "step": 8376 }, { "epoch": 6.136996336996337, "grad_norm": 1.890625, "learning_rate": 9.580989209756135e-05, "loss": 0.2075, "step": 8377 }, { "epoch": 6.137728937728938, "grad_norm": 0.98046875, "learning_rate": 9.58085123438812e-05, "loss": 0.1601, "step": 8378 }, { "epoch": 6.138461538461539, "grad_norm": 0.83203125, "learning_rate": 9.580713237311308e-05, "loss": 0.1212, "step": 8379 }, { "epoch": 6.13919413919414, "grad_norm": 0.79296875, "learning_rate": 9.580575218526363e-05, "loss": 0.1484, "step": 8380 }, { "epoch": 6.13992673992674, "grad_norm": 0.9296875, "learning_rate": 9.580437178033944e-05, "loss": 0.1484, "step": 8381 }, { "epoch": 6.140659340659341, "grad_norm": 1.078125, "learning_rate": 9.580299115834712e-05, "loss": 0.1666, "step": 8382 }, { "epoch": 6.141391941391942, "grad_norm": 0.61328125, "learning_rate": 9.580161031929333e-05, "loss": 0.1514, "step": 8383 }, { "epoch": 6.1421245421245425, "grad_norm": 0.78125, "learning_rate": 9.580022926318462e-05, "loss": 0.1393, "step": 8384 }, { "epoch": 6.142857142857143, "grad_norm": 0.8359375, "learning_rate": 9.579884799002765e-05, "loss": 0.123, "step": 8385 }, { "epoch": 6.143589743589744, "grad_norm": 0.8671875, "learning_rate": 9.579746649982903e-05, "loss": 0.1309, "step": 8386 }, { "epoch": 6.144322344322345, "grad_norm": 1.5078125, "learning_rate": 9.579608479259538e-05, "loss": 0.2115, "step": 8387 }, { "epoch": 6.145054945054945, "grad_norm": 0.69140625, "learning_rate": 9.579470286833332e-05, "loss": 0.1268, "step": 8388 }, { "epoch": 6.145787545787546, "grad_norm": 1.1640625, "learning_rate": 9.579332072704946e-05, "loss": 0.1551, "step": 8389 }, { "epoch": 6.146520146520147, "grad_norm": 1.3359375, "learning_rate": 9.579193836875045e-05, "loss": 0.1417, "step": 8390 }, { "epoch": 6.147252747252748, "grad_norm": 1.390625, "learning_rate": 9.579055579344287e-05, "loss": 0.1465, "step": 8391 }, { "epoch": 6.147985347985348, "grad_norm": 1.2109375, "learning_rate": 9.57891730011334e-05, "loss": 0.1525, "step": 8392 }, { "epoch": 6.148717948717949, "grad_norm": 0.68359375, "learning_rate": 9.57877899918286e-05, "loss": 0.1267, "step": 8393 }, { "epoch": 6.14945054945055, "grad_norm": 1.0390625, "learning_rate": 9.578640676553513e-05, "loss": 0.1519, "step": 8394 }, { "epoch": 6.1501831501831505, "grad_norm": 0.69140625, "learning_rate": 9.578502332225967e-05, "loss": 0.1125, "step": 8395 }, { "epoch": 6.150915750915751, "grad_norm": 0.796875, "learning_rate": 9.578363966200875e-05, "loss": 0.1841, "step": 8396 }, { "epoch": 6.151648351648352, "grad_norm": 1.125, "learning_rate": 9.578225578478906e-05, "loss": 0.162, "step": 8397 }, { "epoch": 6.152380952380953, "grad_norm": 1.0234375, "learning_rate": 9.578087169060722e-05, "loss": 0.1401, "step": 8398 }, { "epoch": 6.1531135531135535, "grad_norm": 1.1328125, "learning_rate": 9.577948737946985e-05, "loss": 0.159, "step": 8399 }, { "epoch": 6.153846153846154, "grad_norm": 0.72265625, "learning_rate": 9.577810285138361e-05, "loss": 0.1314, "step": 8400 }, { "epoch": 6.154578754578755, "grad_norm": 0.84765625, "learning_rate": 9.57767181063551e-05, "loss": 0.1163, "step": 8401 }, { "epoch": 6.155311355311356, "grad_norm": 0.7734375, "learning_rate": 9.577533314439098e-05, "loss": 0.1749, "step": 8402 }, { "epoch": 6.156043956043956, "grad_norm": 1.265625, "learning_rate": 9.577394796549787e-05, "loss": 0.1561, "step": 8403 }, { "epoch": 6.156776556776557, "grad_norm": 0.8671875, "learning_rate": 9.577256256968241e-05, "loss": 0.164, "step": 8404 }, { "epoch": 6.157509157509158, "grad_norm": 1.0234375, "learning_rate": 9.577117695695125e-05, "loss": 0.1368, "step": 8405 }, { "epoch": 6.158241758241759, "grad_norm": 1.03125, "learning_rate": 9.5769791127311e-05, "loss": 0.1504, "step": 8406 }, { "epoch": 6.158974358974359, "grad_norm": 0.69140625, "learning_rate": 9.576840508076834e-05, "loss": 0.1419, "step": 8407 }, { "epoch": 6.15970695970696, "grad_norm": 0.609375, "learning_rate": 9.576701881732988e-05, "loss": 0.1422, "step": 8408 }, { "epoch": 6.160439560439561, "grad_norm": 0.6796875, "learning_rate": 9.576563233700227e-05, "loss": 0.1097, "step": 8409 }, { "epoch": 6.1611721611721615, "grad_norm": 0.68359375, "learning_rate": 9.576424563979217e-05, "loss": 0.1404, "step": 8410 }, { "epoch": 6.161904761904762, "grad_norm": 0.79296875, "learning_rate": 9.576285872570621e-05, "loss": 0.1294, "step": 8411 }, { "epoch": 6.162637362637363, "grad_norm": 1.125, "learning_rate": 9.576147159475102e-05, "loss": 0.1595, "step": 8412 }, { "epoch": 6.163369963369964, "grad_norm": 0.89453125, "learning_rate": 9.576008424693326e-05, "loss": 0.1362, "step": 8413 }, { "epoch": 6.164102564102564, "grad_norm": 0.99609375, "learning_rate": 9.575869668225958e-05, "loss": 0.1211, "step": 8414 }, { "epoch": 6.164835164835165, "grad_norm": 0.578125, "learning_rate": 9.575730890073664e-05, "loss": 0.124, "step": 8415 }, { "epoch": 6.165567765567766, "grad_norm": 0.73046875, "learning_rate": 9.575592090237108e-05, "loss": 0.1321, "step": 8416 }, { "epoch": 6.166300366300367, "grad_norm": 1.0546875, "learning_rate": 9.575453268716953e-05, "loss": 0.14, "step": 8417 }, { "epoch": 6.167032967032967, "grad_norm": 0.91015625, "learning_rate": 9.575314425513867e-05, "loss": 0.1369, "step": 8418 }, { "epoch": 6.167765567765568, "grad_norm": 0.6953125, "learning_rate": 9.575175560628513e-05, "loss": 0.1408, "step": 8419 }, { "epoch": 6.168498168498169, "grad_norm": 0.87890625, "learning_rate": 9.575036674061557e-05, "loss": 0.1184, "step": 8420 }, { "epoch": 6.1692307692307695, "grad_norm": 0.77734375, "learning_rate": 9.574897765813665e-05, "loss": 0.1577, "step": 8421 }, { "epoch": 6.16996336996337, "grad_norm": 0.78125, "learning_rate": 9.574758835885503e-05, "loss": 0.1224, "step": 8422 }, { "epoch": 6.170695970695971, "grad_norm": 0.69140625, "learning_rate": 9.574619884277738e-05, "loss": 0.1355, "step": 8423 }, { "epoch": 6.171428571428572, "grad_norm": 0.78515625, "learning_rate": 9.574480910991032e-05, "loss": 0.1296, "step": 8424 }, { "epoch": 6.172161172161172, "grad_norm": 0.921875, "learning_rate": 9.57434191602605e-05, "loss": 0.1321, "step": 8425 }, { "epoch": 6.172893772893773, "grad_norm": 1.0, "learning_rate": 9.574202899383463e-05, "loss": 0.1776, "step": 8426 }, { "epoch": 6.173626373626374, "grad_norm": 0.74609375, "learning_rate": 9.574063861063936e-05, "loss": 0.1322, "step": 8427 }, { "epoch": 6.174358974358975, "grad_norm": 0.95703125, "learning_rate": 9.573924801068133e-05, "loss": 0.1575, "step": 8428 }, { "epoch": 6.175091575091575, "grad_norm": 1.3203125, "learning_rate": 9.573785719396722e-05, "loss": 0.166, "step": 8429 }, { "epoch": 6.175824175824176, "grad_norm": 0.953125, "learning_rate": 9.573646616050366e-05, "loss": 0.1554, "step": 8430 }, { "epoch": 6.176556776556777, "grad_norm": 0.76953125, "learning_rate": 9.573507491029736e-05, "loss": 0.1848, "step": 8431 }, { "epoch": 6.1772893772893775, "grad_norm": 0.8046875, "learning_rate": 9.573368344335496e-05, "loss": 0.1294, "step": 8432 }, { "epoch": 6.178021978021978, "grad_norm": 1.3046875, "learning_rate": 9.573229175968313e-05, "loss": 0.1221, "step": 8433 }, { "epoch": 6.178754578754579, "grad_norm": 1.328125, "learning_rate": 9.573089985928856e-05, "loss": 0.1399, "step": 8434 }, { "epoch": 6.17948717948718, "grad_norm": 0.79296875, "learning_rate": 9.572950774217787e-05, "loss": 0.1161, "step": 8435 }, { "epoch": 6.1802197802197805, "grad_norm": 1.0078125, "learning_rate": 9.572811540835777e-05, "loss": 0.1629, "step": 8436 }, { "epoch": 6.180952380952381, "grad_norm": 1.1171875, "learning_rate": 9.572672285783493e-05, "loss": 0.1515, "step": 8437 }, { "epoch": 6.181684981684982, "grad_norm": 0.77734375, "learning_rate": 9.572533009061601e-05, "loss": 0.1134, "step": 8438 }, { "epoch": 6.182417582417583, "grad_norm": 1.0859375, "learning_rate": 9.57239371067077e-05, "loss": 0.1309, "step": 8439 }, { "epoch": 6.183150183150183, "grad_norm": 1.09375, "learning_rate": 9.572254390611664e-05, "loss": 0.1865, "step": 8440 }, { "epoch": 6.183882783882784, "grad_norm": 1.359375, "learning_rate": 9.572115048884955e-05, "loss": 0.1581, "step": 8441 }, { "epoch": 6.184615384615385, "grad_norm": 1.015625, "learning_rate": 9.571975685491306e-05, "loss": 0.1219, "step": 8442 }, { "epoch": 6.185347985347986, "grad_norm": 1.21875, "learning_rate": 9.571836300431388e-05, "loss": 0.1509, "step": 8443 }, { "epoch": 6.186080586080586, "grad_norm": 0.9453125, "learning_rate": 9.571696893705868e-05, "loss": 0.1177, "step": 8444 }, { "epoch": 6.186813186813187, "grad_norm": 1.046875, "learning_rate": 9.571557465315414e-05, "loss": 0.1513, "step": 8445 }, { "epoch": 6.187545787545788, "grad_norm": 1.4453125, "learning_rate": 9.571418015260694e-05, "loss": 0.1597, "step": 8446 }, { "epoch": 6.1882783882783885, "grad_norm": 0.7734375, "learning_rate": 9.571278543542375e-05, "loss": 0.141, "step": 8447 }, { "epoch": 6.189010989010989, "grad_norm": 0.796875, "learning_rate": 9.571139050161126e-05, "loss": 0.1424, "step": 8448 }, { "epoch": 6.18974358974359, "grad_norm": 1.71875, "learning_rate": 9.570999535117618e-05, "loss": 0.1539, "step": 8449 }, { "epoch": 6.190476190476191, "grad_norm": 0.86328125, "learning_rate": 9.570859998412515e-05, "loss": 0.1393, "step": 8450 }, { "epoch": 6.191208791208791, "grad_norm": 1.2890625, "learning_rate": 9.570720440046488e-05, "loss": 0.1243, "step": 8451 }, { "epoch": 6.191941391941392, "grad_norm": 0.97265625, "learning_rate": 9.570580860020205e-05, "loss": 0.1393, "step": 8452 }, { "epoch": 6.192673992673993, "grad_norm": 1.703125, "learning_rate": 9.570441258334336e-05, "loss": 0.1196, "step": 8453 }, { "epoch": 6.193406593406594, "grad_norm": 1.0234375, "learning_rate": 9.570301634989548e-05, "loss": 0.1092, "step": 8454 }, { "epoch": 6.194139194139194, "grad_norm": 0.8984375, "learning_rate": 9.570161989986512e-05, "loss": 0.1234, "step": 8455 }, { "epoch": 6.194871794871795, "grad_norm": 1.265625, "learning_rate": 9.570022323325895e-05, "loss": 0.1491, "step": 8456 }, { "epoch": 6.195604395604396, "grad_norm": 1.09375, "learning_rate": 9.569882635008367e-05, "loss": 0.1728, "step": 8457 }, { "epoch": 6.1963369963369965, "grad_norm": 0.61328125, "learning_rate": 9.569742925034599e-05, "loss": 0.1401, "step": 8458 }, { "epoch": 6.197069597069597, "grad_norm": 0.9921875, "learning_rate": 9.569603193405258e-05, "loss": 0.1447, "step": 8459 }, { "epoch": 6.197802197802198, "grad_norm": 0.7421875, "learning_rate": 9.569463440121015e-05, "loss": 0.1344, "step": 8460 }, { "epoch": 6.198534798534799, "grad_norm": 1.40625, "learning_rate": 9.56932366518254e-05, "loss": 0.1458, "step": 8461 }, { "epoch": 6.1992673992673994, "grad_norm": 0.7265625, "learning_rate": 9.5691838685905e-05, "loss": 0.1544, "step": 8462 }, { "epoch": 6.2, "grad_norm": 0.99609375, "learning_rate": 9.569044050345565e-05, "loss": 0.154, "step": 8463 }, { "epoch": 6.200732600732601, "grad_norm": 1.078125, "learning_rate": 9.56890421044841e-05, "loss": 0.1345, "step": 8464 }, { "epoch": 6.201465201465202, "grad_norm": 0.859375, "learning_rate": 9.568764348899699e-05, "loss": 0.1329, "step": 8465 }, { "epoch": 6.202197802197802, "grad_norm": 0.87109375, "learning_rate": 9.568624465700104e-05, "loss": 0.1295, "step": 8466 }, { "epoch": 6.202930402930403, "grad_norm": 1.078125, "learning_rate": 9.568484560850298e-05, "loss": 0.1502, "step": 8467 }, { "epoch": 6.203663003663004, "grad_norm": 0.80078125, "learning_rate": 9.568344634350949e-05, "loss": 0.1217, "step": 8468 }, { "epoch": 6.204395604395605, "grad_norm": 1.1484375, "learning_rate": 9.568204686202728e-05, "loss": 0.1402, "step": 8469 }, { "epoch": 6.205128205128205, "grad_norm": 2.234375, "learning_rate": 9.568064716406303e-05, "loss": 0.1591, "step": 8470 }, { "epoch": 6.205860805860806, "grad_norm": 1.203125, "learning_rate": 9.567924724962349e-05, "loss": 0.1526, "step": 8471 }, { "epoch": 6.206593406593407, "grad_norm": 1.359375, "learning_rate": 9.567784711871532e-05, "loss": 0.1856, "step": 8472 }, { "epoch": 6.2073260073260075, "grad_norm": 1.234375, "learning_rate": 9.567644677134526e-05, "loss": 0.165, "step": 8473 }, { "epoch": 6.208058608058608, "grad_norm": 0.55859375, "learning_rate": 9.567504620752002e-05, "loss": 0.1294, "step": 8474 }, { "epoch": 6.208791208791209, "grad_norm": 0.85546875, "learning_rate": 9.56736454272463e-05, "loss": 0.1342, "step": 8475 }, { "epoch": 6.20952380952381, "grad_norm": 0.7265625, "learning_rate": 9.567224443053082e-05, "loss": 0.115, "step": 8476 }, { "epoch": 6.21025641025641, "grad_norm": 1.109375, "learning_rate": 9.567084321738029e-05, "loss": 0.1425, "step": 8477 }, { "epoch": 6.210989010989011, "grad_norm": 1.140625, "learning_rate": 9.566944178780142e-05, "loss": 0.1512, "step": 8478 }, { "epoch": 6.211721611721612, "grad_norm": 0.94140625, "learning_rate": 9.566804014180093e-05, "loss": 0.1771, "step": 8479 }, { "epoch": 6.212454212454213, "grad_norm": 2.359375, "learning_rate": 9.566663827938551e-05, "loss": 0.1764, "step": 8480 }, { "epoch": 6.213186813186813, "grad_norm": 2.421875, "learning_rate": 9.566523620056193e-05, "loss": 0.1541, "step": 8481 }, { "epoch": 6.213919413919414, "grad_norm": 0.6875, "learning_rate": 9.566383390533685e-05, "loss": 0.1314, "step": 8482 }, { "epoch": 6.214652014652015, "grad_norm": 1.4609375, "learning_rate": 9.566243139371703e-05, "loss": 0.136, "step": 8483 }, { "epoch": 6.2153846153846155, "grad_norm": 1.4296875, "learning_rate": 9.566102866570917e-05, "loss": 0.1425, "step": 8484 }, { "epoch": 6.216117216117216, "grad_norm": 0.61328125, "learning_rate": 9.565962572132e-05, "loss": 0.1341, "step": 8485 }, { "epoch": 6.216849816849817, "grad_norm": 1.109375, "learning_rate": 9.565822256055625e-05, "loss": 0.1839, "step": 8486 }, { "epoch": 6.217582417582418, "grad_norm": 2.515625, "learning_rate": 9.56568191834246e-05, "loss": 0.1754, "step": 8487 }, { "epoch": 6.218315018315018, "grad_norm": 2.15625, "learning_rate": 9.565541558993182e-05, "loss": 0.155, "step": 8488 }, { "epoch": 6.219047619047619, "grad_norm": 1.796875, "learning_rate": 9.565401178008461e-05, "loss": 0.1623, "step": 8489 }, { "epoch": 6.21978021978022, "grad_norm": 1.46875, "learning_rate": 9.565260775388975e-05, "loss": 0.1508, "step": 8490 }, { "epoch": 6.220512820512821, "grad_norm": 0.71484375, "learning_rate": 9.565120351135388e-05, "loss": 0.1442, "step": 8491 }, { "epoch": 6.221245421245421, "grad_norm": 1.328125, "learning_rate": 9.564979905248379e-05, "loss": 0.1511, "step": 8492 }, { "epoch": 6.221978021978022, "grad_norm": 1.3515625, "learning_rate": 9.564839437728619e-05, "loss": 0.1157, "step": 8493 }, { "epoch": 6.222710622710623, "grad_norm": 1.25, "learning_rate": 9.564698948576779e-05, "loss": 0.1396, "step": 8494 }, { "epoch": 6.2234432234432235, "grad_norm": 1.078125, "learning_rate": 9.564558437793538e-05, "loss": 0.1676, "step": 8495 }, { "epoch": 6.224175824175824, "grad_norm": 0.90234375, "learning_rate": 9.564417905379563e-05, "loss": 0.1323, "step": 8496 }, { "epoch": 6.224908424908425, "grad_norm": 0.9140625, "learning_rate": 9.56427735133553e-05, "loss": 0.1313, "step": 8497 }, { "epoch": 6.225641025641026, "grad_norm": 1.703125, "learning_rate": 9.564136775662112e-05, "loss": 0.1292, "step": 8498 }, { "epoch": 6.2263736263736265, "grad_norm": 1.3203125, "learning_rate": 9.563996178359983e-05, "loss": 0.1608, "step": 8499 }, { "epoch": 6.227106227106227, "grad_norm": 1.4921875, "learning_rate": 9.563855559429818e-05, "loss": 0.1834, "step": 8500 }, { "epoch": 6.227838827838828, "grad_norm": 1.0625, "learning_rate": 9.563714918872288e-05, "loss": 0.1449, "step": 8501 }, { "epoch": 6.228571428571429, "grad_norm": 1.1640625, "learning_rate": 9.563574256688068e-05, "loss": 0.1621, "step": 8502 }, { "epoch": 6.229304029304029, "grad_norm": 1.25, "learning_rate": 9.563433572877833e-05, "loss": 0.1249, "step": 8503 }, { "epoch": 6.23003663003663, "grad_norm": 0.87109375, "learning_rate": 9.563292867442255e-05, "loss": 0.1388, "step": 8504 }, { "epoch": 6.230769230769231, "grad_norm": 1.4453125, "learning_rate": 9.56315214038201e-05, "loss": 0.1552, "step": 8505 }, { "epoch": 6.231501831501832, "grad_norm": 1.5390625, "learning_rate": 9.56301139169777e-05, "loss": 0.2011, "step": 8506 }, { "epoch": 6.232234432234432, "grad_norm": 1.203125, "learning_rate": 9.562870621390211e-05, "loss": 0.1195, "step": 8507 }, { "epoch": 6.232967032967033, "grad_norm": 1.359375, "learning_rate": 9.56272982946001e-05, "loss": 0.1373, "step": 8508 }, { "epoch": 6.233699633699634, "grad_norm": 1.09375, "learning_rate": 9.562589015907836e-05, "loss": 0.1154, "step": 8509 }, { "epoch": 6.2344322344322345, "grad_norm": 1.265625, "learning_rate": 9.562448180734369e-05, "loss": 0.1538, "step": 8510 }, { "epoch": 6.235164835164835, "grad_norm": 1.65625, "learning_rate": 9.56230732394028e-05, "loss": 0.1674, "step": 8511 }, { "epoch": 6.235897435897436, "grad_norm": 1.09375, "learning_rate": 9.562166445526246e-05, "loss": 0.126, "step": 8512 }, { "epoch": 6.236630036630037, "grad_norm": 1.375, "learning_rate": 9.562025545492941e-05, "loss": 0.1247, "step": 8513 }, { "epoch": 6.237362637362637, "grad_norm": 0.8828125, "learning_rate": 9.561884623841039e-05, "loss": 0.131, "step": 8514 }, { "epoch": 6.238095238095238, "grad_norm": 1.5390625, "learning_rate": 9.56174368057122e-05, "loss": 0.1649, "step": 8515 }, { "epoch": 6.238827838827839, "grad_norm": 0.59375, "learning_rate": 9.561602715684153e-05, "loss": 0.123, "step": 8516 }, { "epoch": 6.23956043956044, "grad_norm": 0.81640625, "learning_rate": 9.561461729180518e-05, "loss": 0.1378, "step": 8517 }, { "epoch": 6.24029304029304, "grad_norm": 1.8203125, "learning_rate": 9.561320721060987e-05, "loss": 0.1276, "step": 8518 }, { "epoch": 6.241025641025641, "grad_norm": 0.6953125, "learning_rate": 9.561179691326238e-05, "loss": 0.1542, "step": 8519 }, { "epoch": 6.241758241758242, "grad_norm": 1.359375, "learning_rate": 9.561038639976947e-05, "loss": 0.1479, "step": 8520 }, { "epoch": 6.2424908424908425, "grad_norm": 0.66015625, "learning_rate": 9.560897567013789e-05, "loss": 0.13, "step": 8521 }, { "epoch": 6.243223443223443, "grad_norm": 1.1875, "learning_rate": 9.560756472437438e-05, "loss": 0.1563, "step": 8522 }, { "epoch": 6.243956043956044, "grad_norm": 0.9296875, "learning_rate": 9.560615356248576e-05, "loss": 0.1493, "step": 8523 }, { "epoch": 6.244688644688645, "grad_norm": 1.09375, "learning_rate": 9.560474218447871e-05, "loss": 0.1377, "step": 8524 }, { "epoch": 6.245421245421245, "grad_norm": 0.80078125, "learning_rate": 9.560333059036006e-05, "loss": 0.1465, "step": 8525 }, { "epoch": 6.246153846153846, "grad_norm": 0.87890625, "learning_rate": 9.560191878013653e-05, "loss": 0.1332, "step": 8526 }, { "epoch": 6.246886446886447, "grad_norm": 1.234375, "learning_rate": 9.560050675381492e-05, "loss": 0.1456, "step": 8527 }, { "epoch": 6.247619047619048, "grad_norm": 0.73046875, "learning_rate": 9.559909451140196e-05, "loss": 0.1257, "step": 8528 }, { "epoch": 6.248351648351648, "grad_norm": 1.1171875, "learning_rate": 9.559768205290443e-05, "loss": 0.1254, "step": 8529 }, { "epoch": 6.249084249084249, "grad_norm": 0.8046875, "learning_rate": 9.559626937832912e-05, "loss": 0.1142, "step": 8530 }, { "epoch": 6.24981684981685, "grad_norm": 1.1171875, "learning_rate": 9.559485648768276e-05, "loss": 0.1455, "step": 8531 }, { "epoch": 6.2505494505494505, "grad_norm": 1.2890625, "learning_rate": 9.559344338097215e-05, "loss": 0.1726, "step": 8532 }, { "epoch": 6.251282051282051, "grad_norm": 0.88671875, "learning_rate": 9.559203005820406e-05, "loss": 0.1698, "step": 8533 }, { "epoch": 6.252014652014652, "grad_norm": 0.69921875, "learning_rate": 9.559061651938522e-05, "loss": 0.098, "step": 8534 }, { "epoch": 6.252747252747253, "grad_norm": 1.7109375, "learning_rate": 9.558920276452245e-05, "loss": 0.1276, "step": 8535 }, { "epoch": 6.2534798534798535, "grad_norm": 1.015625, "learning_rate": 9.558778879362252e-05, "loss": 0.1355, "step": 8536 }, { "epoch": 6.254212454212454, "grad_norm": 0.625, "learning_rate": 9.558637460669217e-05, "loss": 0.1418, "step": 8537 }, { "epoch": 6.254945054945055, "grad_norm": 1.140625, "learning_rate": 9.55849602037382e-05, "loss": 0.1591, "step": 8538 }, { "epoch": 6.255677655677656, "grad_norm": 0.84375, "learning_rate": 9.558354558476741e-05, "loss": 0.1311, "step": 8539 }, { "epoch": 6.256410256410256, "grad_norm": 0.96484375, "learning_rate": 9.558213074978653e-05, "loss": 0.1387, "step": 8540 }, { "epoch": 6.257142857142857, "grad_norm": 0.96484375, "learning_rate": 9.558071569880236e-05, "loss": 0.1447, "step": 8541 }, { "epoch": 6.257875457875458, "grad_norm": 1.28125, "learning_rate": 9.55793004318217e-05, "loss": 0.1528, "step": 8542 }, { "epoch": 6.258608058608059, "grad_norm": 1.90625, "learning_rate": 9.557788494885131e-05, "loss": 0.1531, "step": 8543 }, { "epoch": 6.259340659340659, "grad_norm": 1.21875, "learning_rate": 9.557646924989797e-05, "loss": 0.1345, "step": 8544 }, { "epoch": 6.26007326007326, "grad_norm": 1.2578125, "learning_rate": 9.557505333496847e-05, "loss": 0.1133, "step": 8545 }, { "epoch": 6.260805860805861, "grad_norm": 0.97265625, "learning_rate": 9.55736372040696e-05, "loss": 0.1633, "step": 8546 }, { "epoch": 6.2615384615384615, "grad_norm": 1.25, "learning_rate": 9.557222085720812e-05, "loss": 0.1677, "step": 8547 }, { "epoch": 6.262271062271062, "grad_norm": 0.8046875, "learning_rate": 9.557080429439085e-05, "loss": 0.1323, "step": 8548 }, { "epoch": 6.263003663003663, "grad_norm": 0.94140625, "learning_rate": 9.556938751562455e-05, "loss": 0.1532, "step": 8549 }, { "epoch": 6.263736263736264, "grad_norm": 0.8671875, "learning_rate": 9.556797052091602e-05, "loss": 0.1586, "step": 8550 }, { "epoch": 6.264468864468864, "grad_norm": 1.96875, "learning_rate": 9.556655331027204e-05, "loss": 0.1554, "step": 8551 }, { "epoch": 6.265201465201465, "grad_norm": 1.421875, "learning_rate": 9.556513588369943e-05, "loss": 0.1417, "step": 8552 }, { "epoch": 6.265934065934066, "grad_norm": 1.9765625, "learning_rate": 9.556371824120495e-05, "loss": 0.1625, "step": 8553 }, { "epoch": 6.266666666666667, "grad_norm": 0.85546875, "learning_rate": 9.55623003827954e-05, "loss": 0.1473, "step": 8554 }, { "epoch": 6.267399267399267, "grad_norm": 1.125, "learning_rate": 9.556088230847758e-05, "loss": 0.1337, "step": 8555 }, { "epoch": 6.268131868131868, "grad_norm": 1.078125, "learning_rate": 9.555946401825828e-05, "loss": 0.1449, "step": 8556 }, { "epoch": 6.268864468864469, "grad_norm": 0.79296875, "learning_rate": 9.555804551214429e-05, "loss": 0.1276, "step": 8557 }, { "epoch": 6.2695970695970695, "grad_norm": 0.9609375, "learning_rate": 9.555662679014242e-05, "loss": 0.1424, "step": 8558 }, { "epoch": 6.27032967032967, "grad_norm": 0.99609375, "learning_rate": 9.555520785225945e-05, "loss": 0.1601, "step": 8559 }, { "epoch": 6.271062271062271, "grad_norm": 1.203125, "learning_rate": 9.555378869850219e-05, "loss": 0.1323, "step": 8560 }, { "epoch": 6.271794871794872, "grad_norm": 0.65625, "learning_rate": 9.555236932887745e-05, "loss": 0.1286, "step": 8561 }, { "epoch": 6.2725274725274724, "grad_norm": 1.3359375, "learning_rate": 9.555094974339201e-05, "loss": 0.1483, "step": 8562 }, { "epoch": 6.273260073260073, "grad_norm": 1.2734375, "learning_rate": 9.554952994205269e-05, "loss": 0.1299, "step": 8563 }, { "epoch": 6.273992673992674, "grad_norm": 1.25, "learning_rate": 9.554810992486628e-05, "loss": 0.151, "step": 8564 }, { "epoch": 6.274725274725275, "grad_norm": 0.58984375, "learning_rate": 9.554668969183956e-05, "loss": 0.1289, "step": 8565 }, { "epoch": 6.275457875457875, "grad_norm": 1.0859375, "learning_rate": 9.55452692429794e-05, "loss": 0.1213, "step": 8566 }, { "epoch": 6.276190476190476, "grad_norm": 1.1640625, "learning_rate": 9.554384857829254e-05, "loss": 0.1626, "step": 8567 }, { "epoch": 6.276923076923077, "grad_norm": 1.28125, "learning_rate": 9.554242769778583e-05, "loss": 0.2011, "step": 8568 }, { "epoch": 6.277655677655678, "grad_norm": 1.0703125, "learning_rate": 9.554100660146605e-05, "loss": 0.1511, "step": 8569 }, { "epoch": 6.278388278388278, "grad_norm": 1.234375, "learning_rate": 9.553958528934005e-05, "loss": 0.131, "step": 8570 }, { "epoch": 6.279120879120879, "grad_norm": 0.80078125, "learning_rate": 9.553816376141459e-05, "loss": 0.1285, "step": 8571 }, { "epoch": 6.27985347985348, "grad_norm": 0.98828125, "learning_rate": 9.55367420176965e-05, "loss": 0.1371, "step": 8572 }, { "epoch": 6.2805860805860805, "grad_norm": 1.0390625, "learning_rate": 9.55353200581926e-05, "loss": 0.1769, "step": 8573 }, { "epoch": 6.281318681318681, "grad_norm": 0.8359375, "learning_rate": 9.553389788290969e-05, "loss": 0.1307, "step": 8574 }, { "epoch": 6.282051282051282, "grad_norm": 0.58203125, "learning_rate": 9.55324754918546e-05, "loss": 0.1226, "step": 8575 }, { "epoch": 6.282783882783883, "grad_norm": 1.3671875, "learning_rate": 9.553105288503414e-05, "loss": 0.1603, "step": 8576 }, { "epoch": 6.283516483516483, "grad_norm": 1.03125, "learning_rate": 9.552963006245512e-05, "loss": 0.1765, "step": 8577 }, { "epoch": 6.284249084249084, "grad_norm": 1.21875, "learning_rate": 9.552820702412435e-05, "loss": 0.1436, "step": 8578 }, { "epoch": 6.284981684981685, "grad_norm": 0.796875, "learning_rate": 9.552678377004867e-05, "loss": 0.1153, "step": 8579 }, { "epoch": 6.285714285714286, "grad_norm": 1.0, "learning_rate": 9.552536030023488e-05, "loss": 0.141, "step": 8580 }, { "epoch": 6.286446886446886, "grad_norm": 0.81640625, "learning_rate": 9.552393661468982e-05, "loss": 0.1422, "step": 8581 }, { "epoch": 6.287179487179487, "grad_norm": 1.171875, "learning_rate": 9.552251271342029e-05, "loss": 0.1163, "step": 8582 }, { "epoch": 6.287912087912088, "grad_norm": 0.9765625, "learning_rate": 9.552108859643311e-05, "loss": 0.157, "step": 8583 }, { "epoch": 6.2886446886446885, "grad_norm": 1.609375, "learning_rate": 9.551966426373513e-05, "loss": 0.1335, "step": 8584 }, { "epoch": 6.289377289377289, "grad_norm": 0.67578125, "learning_rate": 9.551823971533316e-05, "loss": 0.1333, "step": 8585 }, { "epoch": 6.29010989010989, "grad_norm": 1.0546875, "learning_rate": 9.551681495123402e-05, "loss": 0.1328, "step": 8586 }, { "epoch": 6.290842490842491, "grad_norm": 1.15625, "learning_rate": 9.551538997144454e-05, "loss": 0.1562, "step": 8587 }, { "epoch": 6.291575091575091, "grad_norm": 0.8359375, "learning_rate": 9.551396477597155e-05, "loss": 0.1761, "step": 8588 }, { "epoch": 6.292307692307692, "grad_norm": 1.7734375, "learning_rate": 9.551253936482188e-05, "loss": 0.1748, "step": 8589 }, { "epoch": 6.293040293040293, "grad_norm": 1.390625, "learning_rate": 9.551111373800236e-05, "loss": 0.1454, "step": 8590 }, { "epoch": 6.293772893772894, "grad_norm": 1.2421875, "learning_rate": 9.550968789551981e-05, "loss": 0.133, "step": 8591 }, { "epoch": 6.294505494505494, "grad_norm": 1.3125, "learning_rate": 9.550826183738108e-05, "loss": 0.1507, "step": 8592 }, { "epoch": 6.295238095238095, "grad_norm": 0.92578125, "learning_rate": 9.550683556359298e-05, "loss": 0.1261, "step": 8593 }, { "epoch": 6.295970695970696, "grad_norm": 1.015625, "learning_rate": 9.550540907416236e-05, "loss": 0.158, "step": 8594 }, { "epoch": 6.2967032967032965, "grad_norm": 0.9375, "learning_rate": 9.550398236909606e-05, "loss": 0.1424, "step": 8595 }, { "epoch": 6.297435897435897, "grad_norm": 1.34375, "learning_rate": 9.55025554484009e-05, "loss": 0.1645, "step": 8596 }, { "epoch": 6.298168498168498, "grad_norm": 1.0859375, "learning_rate": 9.550112831208373e-05, "loss": 0.1755, "step": 8597 }, { "epoch": 6.298901098901099, "grad_norm": 1.28125, "learning_rate": 9.549970096015137e-05, "loss": 0.1276, "step": 8598 }, { "epoch": 6.2996336996336995, "grad_norm": 0.87109375, "learning_rate": 9.549827339261067e-05, "loss": 0.1339, "step": 8599 }, { "epoch": 6.3003663003663, "grad_norm": 0.953125, "learning_rate": 9.549684560946848e-05, "loss": 0.1318, "step": 8600 }, { "epoch": 6.301098901098901, "grad_norm": 0.984375, "learning_rate": 9.549541761073163e-05, "loss": 0.1446, "step": 8601 }, { "epoch": 6.301831501831502, "grad_norm": 0.85546875, "learning_rate": 9.549398939640695e-05, "loss": 0.1591, "step": 8602 }, { "epoch": 6.302564102564102, "grad_norm": 1.03125, "learning_rate": 9.549256096650132e-05, "loss": 0.1585, "step": 8603 }, { "epoch": 6.303296703296703, "grad_norm": 0.859375, "learning_rate": 9.549113232102154e-05, "loss": 0.1409, "step": 8604 }, { "epoch": 6.304029304029304, "grad_norm": 0.8203125, "learning_rate": 9.548970345997448e-05, "loss": 0.1239, "step": 8605 }, { "epoch": 6.304761904761905, "grad_norm": 1.1328125, "learning_rate": 9.5488274383367e-05, "loss": 0.1373, "step": 8606 }, { "epoch": 6.305494505494505, "grad_norm": 1.546875, "learning_rate": 9.548684509120589e-05, "loss": 0.1628, "step": 8607 }, { "epoch": 6.306227106227106, "grad_norm": 1.1875, "learning_rate": 9.548541558349807e-05, "loss": 0.1594, "step": 8608 }, { "epoch": 6.306959706959707, "grad_norm": 0.90234375, "learning_rate": 9.548398586025035e-05, "loss": 0.1269, "step": 8609 }, { "epoch": 6.3076923076923075, "grad_norm": 1.890625, "learning_rate": 9.548255592146959e-05, "loss": 0.1661, "step": 8610 }, { "epoch": 6.308424908424908, "grad_norm": 1.9375, "learning_rate": 9.548112576716261e-05, "loss": 0.1348, "step": 8611 }, { "epoch": 6.309157509157509, "grad_norm": 1.125, "learning_rate": 9.547969539733631e-05, "loss": 0.1475, "step": 8612 }, { "epoch": 6.30989010989011, "grad_norm": 1.6875, "learning_rate": 9.547826481199752e-05, "loss": 0.1387, "step": 8613 }, { "epoch": 6.31062271062271, "grad_norm": 1.2421875, "learning_rate": 9.547683401115311e-05, "loss": 0.1528, "step": 8614 }, { "epoch": 6.311355311355311, "grad_norm": 0.91015625, "learning_rate": 9.54754029948099e-05, "loss": 0.1582, "step": 8615 }, { "epoch": 6.312087912087912, "grad_norm": 1.3359375, "learning_rate": 9.54739717629748e-05, "loss": 0.1504, "step": 8616 }, { "epoch": 6.312820512820513, "grad_norm": 2.015625, "learning_rate": 9.547254031565461e-05, "loss": 0.1622, "step": 8617 }, { "epoch": 6.313553113553113, "grad_norm": 1.3046875, "learning_rate": 9.547110865285624e-05, "loss": 0.1694, "step": 8618 }, { "epoch": 6.314285714285714, "grad_norm": 0.734375, "learning_rate": 9.546967677458652e-05, "loss": 0.1474, "step": 8619 }, { "epoch": 6.315018315018315, "grad_norm": 0.94140625, "learning_rate": 9.546824468085231e-05, "loss": 0.1798, "step": 8620 }, { "epoch": 6.3157509157509155, "grad_norm": 0.78515625, "learning_rate": 9.546681237166046e-05, "loss": 0.1296, "step": 8621 }, { "epoch": 6.316483516483516, "grad_norm": 1.171875, "learning_rate": 9.546537984701786e-05, "loss": 0.1571, "step": 8622 }, { "epoch": 6.317216117216117, "grad_norm": 1.2578125, "learning_rate": 9.546394710693138e-05, "loss": 0.1435, "step": 8623 }, { "epoch": 6.317948717948718, "grad_norm": 0.97265625, "learning_rate": 9.546251415140785e-05, "loss": 0.1314, "step": 8624 }, { "epoch": 6.318681318681318, "grad_norm": 1.5234375, "learning_rate": 9.546108098045418e-05, "loss": 0.1422, "step": 8625 }, { "epoch": 6.319413919413919, "grad_norm": 1.296875, "learning_rate": 9.54596475940772e-05, "loss": 0.1522, "step": 8626 }, { "epoch": 6.32014652014652, "grad_norm": 1.2265625, "learning_rate": 9.545821399228378e-05, "loss": 0.1196, "step": 8627 }, { "epoch": 6.320879120879121, "grad_norm": 1.265625, "learning_rate": 9.54567801750808e-05, "loss": 0.1415, "step": 8628 }, { "epoch": 6.321611721611721, "grad_norm": 1.5078125, "learning_rate": 9.545534614247513e-05, "loss": 0.1476, "step": 8629 }, { "epoch": 6.322344322344322, "grad_norm": 1.2578125, "learning_rate": 9.545391189447364e-05, "loss": 0.1499, "step": 8630 }, { "epoch": 6.323076923076923, "grad_norm": 1.484375, "learning_rate": 9.545247743108321e-05, "loss": 0.1365, "step": 8631 }, { "epoch": 6.3238095238095235, "grad_norm": 0.91796875, "learning_rate": 9.545104275231068e-05, "loss": 0.137, "step": 8632 }, { "epoch": 6.324542124542124, "grad_norm": 0.8046875, "learning_rate": 9.544960785816295e-05, "loss": 0.13, "step": 8633 }, { "epoch": 6.325274725274725, "grad_norm": 1.7734375, "learning_rate": 9.544817274864691e-05, "loss": 0.1717, "step": 8634 }, { "epoch": 6.326007326007326, "grad_norm": 1.75, "learning_rate": 9.544673742376942e-05, "loss": 0.1592, "step": 8635 }, { "epoch": 6.3267399267399265, "grad_norm": 0.7890625, "learning_rate": 9.544530188353734e-05, "loss": 0.1204, "step": 8636 }, { "epoch": 6.327472527472527, "grad_norm": 1.828125, "learning_rate": 9.544386612795758e-05, "loss": 0.1607, "step": 8637 }, { "epoch": 6.328205128205128, "grad_norm": 0.52734375, "learning_rate": 9.544243015703698e-05, "loss": 0.1407, "step": 8638 }, { "epoch": 6.328937728937729, "grad_norm": 0.890625, "learning_rate": 9.544099397078247e-05, "loss": 0.1445, "step": 8639 }, { "epoch": 6.329670329670329, "grad_norm": 0.8671875, "learning_rate": 9.543955756920088e-05, "loss": 0.1188, "step": 8640 }, { "epoch": 6.33040293040293, "grad_norm": 1.4765625, "learning_rate": 9.543812095229915e-05, "loss": 0.1351, "step": 8641 }, { "epoch": 6.331135531135531, "grad_norm": 1.2734375, "learning_rate": 9.543668412008412e-05, "loss": 0.1734, "step": 8642 }, { "epoch": 6.331868131868132, "grad_norm": 1.140625, "learning_rate": 9.543524707256266e-05, "loss": 0.152, "step": 8643 }, { "epoch": 6.332600732600732, "grad_norm": 0.9375, "learning_rate": 9.54338098097417e-05, "loss": 0.1381, "step": 8644 }, { "epoch": 6.333333333333333, "grad_norm": 1.296875, "learning_rate": 9.543237233162811e-05, "loss": 0.1404, "step": 8645 }, { "epoch": 6.334065934065934, "grad_norm": 0.921875, "learning_rate": 9.543093463822876e-05, "loss": 0.1617, "step": 8646 }, { "epoch": 6.3347985347985345, "grad_norm": 1.2890625, "learning_rate": 9.542949672955056e-05, "loss": 0.1595, "step": 8647 }, { "epoch": 6.335531135531135, "grad_norm": 0.80859375, "learning_rate": 9.542805860560038e-05, "loss": 0.1413, "step": 8648 }, { "epoch": 6.336263736263736, "grad_norm": 0.8828125, "learning_rate": 9.542662026638515e-05, "loss": 0.1298, "step": 8649 }, { "epoch": 6.336996336996337, "grad_norm": 0.76171875, "learning_rate": 9.542518171191171e-05, "loss": 0.1399, "step": 8650 }, { "epoch": 6.337728937728937, "grad_norm": 0.83984375, "learning_rate": 9.542374294218698e-05, "loss": 0.1366, "step": 8651 }, { "epoch": 6.338461538461538, "grad_norm": 1.15625, "learning_rate": 9.542230395721784e-05, "loss": 0.147, "step": 8652 }, { "epoch": 6.339194139194139, "grad_norm": 1.1015625, "learning_rate": 9.542086475701122e-05, "loss": 0.1506, "step": 8653 }, { "epoch": 6.33992673992674, "grad_norm": 0.87890625, "learning_rate": 9.541942534157398e-05, "loss": 0.1292, "step": 8654 }, { "epoch": 6.34065934065934, "grad_norm": 0.92578125, "learning_rate": 9.541798571091302e-05, "loss": 0.1464, "step": 8655 }, { "epoch": 6.341391941391941, "grad_norm": 1.71875, "learning_rate": 9.541654586503526e-05, "loss": 0.1446, "step": 8656 }, { "epoch": 6.342124542124542, "grad_norm": 1.0, "learning_rate": 9.541510580394757e-05, "loss": 0.1346, "step": 8657 }, { "epoch": 6.3428571428571425, "grad_norm": 1.2109375, "learning_rate": 9.541366552765687e-05, "loss": 0.1576, "step": 8658 }, { "epoch": 6.343589743589743, "grad_norm": 0.921875, "learning_rate": 9.541222503617004e-05, "loss": 0.1291, "step": 8659 }, { "epoch": 6.344322344322344, "grad_norm": 1.015625, "learning_rate": 9.541078432949401e-05, "loss": 0.1475, "step": 8660 }, { "epoch": 6.345054945054945, "grad_norm": 1.421875, "learning_rate": 9.540934340763568e-05, "loss": 0.1711, "step": 8661 }, { "epoch": 6.3457875457875454, "grad_norm": 0.984375, "learning_rate": 9.540790227060192e-05, "loss": 0.1293, "step": 8662 }, { "epoch": 6.346520146520146, "grad_norm": 2.09375, "learning_rate": 9.540646091839967e-05, "loss": 0.1214, "step": 8663 }, { "epoch": 6.347252747252747, "grad_norm": 0.95703125, "learning_rate": 9.54050193510358e-05, "loss": 0.1077, "step": 8664 }, { "epoch": 6.347985347985348, "grad_norm": 1.25, "learning_rate": 9.540357756851727e-05, "loss": 0.1649, "step": 8665 }, { "epoch": 6.348717948717948, "grad_norm": 1.2265625, "learning_rate": 9.540213557085097e-05, "loss": 0.1371, "step": 8666 }, { "epoch": 6.349450549450549, "grad_norm": 1.2890625, "learning_rate": 9.540069335804376e-05, "loss": 0.1361, "step": 8667 }, { "epoch": 6.35018315018315, "grad_norm": 1.1484375, "learning_rate": 9.539925093010261e-05, "loss": 0.1586, "step": 8668 }, { "epoch": 6.350915750915751, "grad_norm": 1.4140625, "learning_rate": 9.539780828703441e-05, "loss": 0.1528, "step": 8669 }, { "epoch": 6.351648351648351, "grad_norm": 1.0234375, "learning_rate": 9.539636542884608e-05, "loss": 0.1288, "step": 8670 }, { "epoch": 6.352380952380952, "grad_norm": 0.78515625, "learning_rate": 9.53949223555445e-05, "loss": 0.1074, "step": 8671 }, { "epoch": 6.353113553113553, "grad_norm": 1.640625, "learning_rate": 9.539347906713663e-05, "loss": 0.1473, "step": 8672 }, { "epoch": 6.3538461538461535, "grad_norm": 1.7109375, "learning_rate": 9.539203556362935e-05, "loss": 0.1529, "step": 8673 }, { "epoch": 6.354578754578754, "grad_norm": 1.265625, "learning_rate": 9.53905918450296e-05, "loss": 0.1328, "step": 8674 }, { "epoch": 6.355311355311355, "grad_norm": 0.90234375, "learning_rate": 9.538914791134428e-05, "loss": 0.1589, "step": 8675 }, { "epoch": 6.356043956043956, "grad_norm": 1.0703125, "learning_rate": 9.538770376258033e-05, "loss": 0.1363, "step": 8676 }, { "epoch": 6.356776556776556, "grad_norm": 0.79296875, "learning_rate": 9.538625939874464e-05, "loss": 0.1516, "step": 8677 }, { "epoch": 6.357509157509157, "grad_norm": 0.85546875, "learning_rate": 9.538481481984416e-05, "loss": 0.1548, "step": 8678 }, { "epoch": 6.358241758241758, "grad_norm": 1.1484375, "learning_rate": 9.538337002588578e-05, "loss": 0.2, "step": 8679 }, { "epoch": 6.358974358974359, "grad_norm": 2.234375, "learning_rate": 9.538192501687645e-05, "loss": 0.1658, "step": 8680 }, { "epoch": 6.359706959706959, "grad_norm": 2.03125, "learning_rate": 9.538047979282309e-05, "loss": 0.1611, "step": 8681 }, { "epoch": 6.36043956043956, "grad_norm": 1.078125, "learning_rate": 9.53790343537326e-05, "loss": 0.1333, "step": 8682 }, { "epoch": 6.361172161172161, "grad_norm": 0.83203125, "learning_rate": 9.537758869961196e-05, "loss": 0.1137, "step": 8683 }, { "epoch": 6.3619047619047615, "grad_norm": 1.3203125, "learning_rate": 9.537614283046802e-05, "loss": 0.1397, "step": 8684 }, { "epoch": 6.362637362637362, "grad_norm": 1.25, "learning_rate": 9.537469674630776e-05, "loss": 0.1556, "step": 8685 }, { "epoch": 6.363369963369963, "grad_norm": 0.9453125, "learning_rate": 9.53732504471381e-05, "loss": 0.1271, "step": 8686 }, { "epoch": 6.364102564102564, "grad_norm": 0.8125, "learning_rate": 9.537180393296597e-05, "loss": 0.144, "step": 8687 }, { "epoch": 6.364835164835164, "grad_norm": 0.83203125, "learning_rate": 9.537035720379828e-05, "loss": 0.1611, "step": 8688 }, { "epoch": 6.365567765567765, "grad_norm": 0.99609375, "learning_rate": 9.5368910259642e-05, "loss": 0.1314, "step": 8689 }, { "epoch": 6.366300366300366, "grad_norm": 1.1328125, "learning_rate": 9.536746310050404e-05, "loss": 0.1359, "step": 8690 }, { "epoch": 6.367032967032967, "grad_norm": 0.6328125, "learning_rate": 9.536601572639132e-05, "loss": 0.1159, "step": 8691 }, { "epoch": 6.367765567765568, "grad_norm": 1.0703125, "learning_rate": 9.53645681373108e-05, "loss": 0.1509, "step": 8692 }, { "epoch": 6.368498168498169, "grad_norm": 0.65625, "learning_rate": 9.536312033326939e-05, "loss": 0.1631, "step": 8693 }, { "epoch": 6.36923076923077, "grad_norm": 1.1875, "learning_rate": 9.536167231427407e-05, "loss": 0.1867, "step": 8694 }, { "epoch": 6.36996336996337, "grad_norm": 0.92578125, "learning_rate": 9.536022408033175e-05, "loss": 0.1284, "step": 8695 }, { "epoch": 6.370695970695971, "grad_norm": 0.95703125, "learning_rate": 9.535877563144934e-05, "loss": 0.1264, "step": 8696 }, { "epoch": 6.371428571428572, "grad_norm": 0.859375, "learning_rate": 9.535732696763384e-05, "loss": 0.1482, "step": 8697 }, { "epoch": 6.372161172161173, "grad_norm": 0.6796875, "learning_rate": 9.535587808889214e-05, "loss": 0.1303, "step": 8698 }, { "epoch": 6.372893772893773, "grad_norm": 1.171875, "learning_rate": 9.535442899523122e-05, "loss": 0.1566, "step": 8699 }, { "epoch": 6.373626373626374, "grad_norm": 0.75, "learning_rate": 9.535297968665799e-05, "loss": 0.141, "step": 8700 }, { "epoch": 6.374358974358975, "grad_norm": 0.9921875, "learning_rate": 9.535153016317942e-05, "loss": 0.1376, "step": 8701 }, { "epoch": 6.3750915750915755, "grad_norm": 0.85546875, "learning_rate": 9.535008042480244e-05, "loss": 0.1071, "step": 8702 }, { "epoch": 6.375824175824176, "grad_norm": 1.3046875, "learning_rate": 9.5348630471534e-05, "loss": 0.1165, "step": 8703 }, { "epoch": 6.376556776556777, "grad_norm": 0.87890625, "learning_rate": 9.534718030338105e-05, "loss": 0.1154, "step": 8704 }, { "epoch": 6.377289377289378, "grad_norm": 1.5, "learning_rate": 9.534572992035054e-05, "loss": 0.1378, "step": 8705 }, { "epoch": 6.3780219780219785, "grad_norm": 1.2734375, "learning_rate": 9.534427932244942e-05, "loss": 0.1269, "step": 8706 }, { "epoch": 6.378754578754579, "grad_norm": 0.73828125, "learning_rate": 9.534282850968463e-05, "loss": 0.1509, "step": 8707 }, { "epoch": 6.37948717948718, "grad_norm": 0.515625, "learning_rate": 9.534137748206314e-05, "loss": 0.1243, "step": 8708 }, { "epoch": 6.380219780219781, "grad_norm": 1.15625, "learning_rate": 9.533992623959186e-05, "loss": 0.1723, "step": 8709 }, { "epoch": 6.380952380952381, "grad_norm": 0.59765625, "learning_rate": 9.53384747822778e-05, "loss": 0.1194, "step": 8710 }, { "epoch": 6.381684981684982, "grad_norm": 1.9453125, "learning_rate": 9.533702311012788e-05, "loss": 0.1537, "step": 8711 }, { "epoch": 6.382417582417583, "grad_norm": 1.875, "learning_rate": 9.533557122314906e-05, "loss": 0.1294, "step": 8712 }, { "epoch": 6.383150183150184, "grad_norm": 1.8828125, "learning_rate": 9.53341191213483e-05, "loss": 0.1595, "step": 8713 }, { "epoch": 6.383882783882784, "grad_norm": 1.1171875, "learning_rate": 9.533266680473257e-05, "loss": 0.1319, "step": 8714 }, { "epoch": 6.384615384615385, "grad_norm": 1.3828125, "learning_rate": 9.53312142733088e-05, "loss": 0.1257, "step": 8715 }, { "epoch": 6.385347985347986, "grad_norm": 0.87890625, "learning_rate": 9.532976152708397e-05, "loss": 0.1441, "step": 8716 }, { "epoch": 6.3860805860805865, "grad_norm": 1.8515625, "learning_rate": 9.532830856606504e-05, "loss": 0.1731, "step": 8717 }, { "epoch": 6.386813186813187, "grad_norm": 0.72265625, "learning_rate": 9.532685539025897e-05, "loss": 0.1331, "step": 8718 }, { "epoch": 6.387545787545788, "grad_norm": 1.359375, "learning_rate": 9.532540199967271e-05, "loss": 0.1478, "step": 8719 }, { "epoch": 6.388278388278389, "grad_norm": 0.9609375, "learning_rate": 9.532394839431323e-05, "loss": 0.1495, "step": 8720 }, { "epoch": 6.389010989010989, "grad_norm": 1.03125, "learning_rate": 9.532249457418752e-05, "loss": 0.1516, "step": 8721 }, { "epoch": 6.38974358974359, "grad_norm": 1.2265625, "learning_rate": 9.532104053930251e-05, "loss": 0.1586, "step": 8722 }, { "epoch": 6.390476190476191, "grad_norm": 1.84375, "learning_rate": 9.531958628966517e-05, "loss": 0.1547, "step": 8723 }, { "epoch": 6.391208791208792, "grad_norm": 0.98828125, "learning_rate": 9.53181318252825e-05, "loss": 0.1249, "step": 8724 }, { "epoch": 6.391941391941392, "grad_norm": 2.21875, "learning_rate": 9.531667714616144e-05, "loss": 0.1584, "step": 8725 }, { "epoch": 6.392673992673993, "grad_norm": 0.92578125, "learning_rate": 9.531522225230897e-05, "loss": 0.1293, "step": 8726 }, { "epoch": 6.393406593406594, "grad_norm": 0.8671875, "learning_rate": 9.531376714373204e-05, "loss": 0.144, "step": 8727 }, { "epoch": 6.3941391941391945, "grad_norm": 1.3828125, "learning_rate": 9.531231182043766e-05, "loss": 0.147, "step": 8728 }, { "epoch": 6.394871794871795, "grad_norm": 1.3359375, "learning_rate": 9.531085628243278e-05, "loss": 0.1352, "step": 8729 }, { "epoch": 6.395604395604396, "grad_norm": 0.93359375, "learning_rate": 9.530940052972438e-05, "loss": 0.1175, "step": 8730 }, { "epoch": 6.396336996336997, "grad_norm": 0.6640625, "learning_rate": 9.530794456231943e-05, "loss": 0.1278, "step": 8731 }, { "epoch": 6.397069597069597, "grad_norm": 1.8359375, "learning_rate": 9.53064883802249e-05, "loss": 0.1419, "step": 8732 }, { "epoch": 6.397802197802198, "grad_norm": 1.453125, "learning_rate": 9.530503198344777e-05, "loss": 0.154, "step": 8733 }, { "epoch": 6.398534798534799, "grad_norm": 1.34375, "learning_rate": 9.530357537199505e-05, "loss": 0.1318, "step": 8734 }, { "epoch": 6.3992673992674, "grad_norm": 0.87109375, "learning_rate": 9.530211854587367e-05, "loss": 0.1339, "step": 8735 }, { "epoch": 6.4, "grad_norm": 1.2265625, "learning_rate": 9.530066150509064e-05, "loss": 0.1331, "step": 8736 }, { "epoch": 6.400732600732601, "grad_norm": 1.7421875, "learning_rate": 9.529920424965293e-05, "loss": 0.1267, "step": 8737 }, { "epoch": 6.401465201465202, "grad_norm": 1.6875, "learning_rate": 9.529774677956753e-05, "loss": 0.1553, "step": 8738 }, { "epoch": 6.4021978021978025, "grad_norm": 1.0625, "learning_rate": 9.529628909484141e-05, "loss": 0.1636, "step": 8739 }, { "epoch": 6.402930402930403, "grad_norm": 1.1796875, "learning_rate": 9.529483119548157e-05, "loss": 0.1545, "step": 8740 }, { "epoch": 6.403663003663004, "grad_norm": 0.94140625, "learning_rate": 9.5293373081495e-05, "loss": 0.1412, "step": 8741 }, { "epoch": 6.404395604395605, "grad_norm": 1.3203125, "learning_rate": 9.529191475288865e-05, "loss": 0.1449, "step": 8742 }, { "epoch": 6.4051282051282055, "grad_norm": 1.3515625, "learning_rate": 9.529045620966956e-05, "loss": 0.1307, "step": 8743 }, { "epoch": 6.405860805860806, "grad_norm": 1.421875, "learning_rate": 9.528899745184466e-05, "loss": 0.1411, "step": 8744 }, { "epoch": 6.406593406593407, "grad_norm": 1.140625, "learning_rate": 9.528753847942099e-05, "loss": 0.1821, "step": 8745 }, { "epoch": 6.407326007326008, "grad_norm": 0.875, "learning_rate": 9.528607929240552e-05, "loss": 0.1389, "step": 8746 }, { "epoch": 6.408058608058608, "grad_norm": 0.7890625, "learning_rate": 9.528461989080524e-05, "loss": 0.1492, "step": 8747 }, { "epoch": 6.408791208791209, "grad_norm": 0.75, "learning_rate": 9.528316027462715e-05, "loss": 0.1135, "step": 8748 }, { "epoch": 6.40952380952381, "grad_norm": 1.6328125, "learning_rate": 9.528170044387822e-05, "loss": 0.1335, "step": 8749 }, { "epoch": 6.410256410256411, "grad_norm": 0.91796875, "learning_rate": 9.528024039856548e-05, "loss": 0.1348, "step": 8750 }, { "epoch": 6.410989010989011, "grad_norm": 1.2265625, "learning_rate": 9.527878013869589e-05, "loss": 0.1197, "step": 8751 }, { "epoch": 6.411721611721612, "grad_norm": 1.203125, "learning_rate": 9.527731966427647e-05, "loss": 0.1672, "step": 8752 }, { "epoch": 6.412454212454213, "grad_norm": 0.84765625, "learning_rate": 9.527585897531423e-05, "loss": 0.1448, "step": 8753 }, { "epoch": 6.4131868131868135, "grad_norm": 1.015625, "learning_rate": 9.527439807181613e-05, "loss": 0.1281, "step": 8754 }, { "epoch": 6.413919413919414, "grad_norm": 0.74609375, "learning_rate": 9.52729369537892e-05, "loss": 0.1477, "step": 8755 }, { "epoch": 6.414652014652015, "grad_norm": 1.359375, "learning_rate": 9.527147562124043e-05, "loss": 0.1722, "step": 8756 }, { "epoch": 6.415384615384616, "grad_norm": 0.734375, "learning_rate": 9.527001407417683e-05, "loss": 0.1393, "step": 8757 }, { "epoch": 6.416117216117216, "grad_norm": 0.97265625, "learning_rate": 9.526855231260538e-05, "loss": 0.115, "step": 8758 }, { "epoch": 6.416849816849817, "grad_norm": 0.8828125, "learning_rate": 9.526709033653312e-05, "loss": 0.1299, "step": 8759 }, { "epoch": 6.417582417582418, "grad_norm": 1.2421875, "learning_rate": 9.526562814596701e-05, "loss": 0.1538, "step": 8760 }, { "epoch": 6.418315018315019, "grad_norm": 0.828125, "learning_rate": 9.526416574091409e-05, "loss": 0.1312, "step": 8761 }, { "epoch": 6.419047619047619, "grad_norm": 0.83984375, "learning_rate": 9.526270312138136e-05, "loss": 0.1209, "step": 8762 }, { "epoch": 6.41978021978022, "grad_norm": 1.53125, "learning_rate": 9.526124028737582e-05, "loss": 0.1373, "step": 8763 }, { "epoch": 6.420512820512821, "grad_norm": 1.59375, "learning_rate": 9.525977723890448e-05, "loss": 0.1351, "step": 8764 }, { "epoch": 6.4212454212454215, "grad_norm": 0.7890625, "learning_rate": 9.525831397597437e-05, "loss": 0.1499, "step": 8765 }, { "epoch": 6.421978021978022, "grad_norm": 0.55078125, "learning_rate": 9.525685049859248e-05, "loss": 0.1401, "step": 8766 }, { "epoch": 6.422710622710623, "grad_norm": 1.140625, "learning_rate": 9.525538680676581e-05, "loss": 0.1528, "step": 8767 }, { "epoch": 6.423443223443224, "grad_norm": 1.140625, "learning_rate": 9.52539229005014e-05, "loss": 0.1429, "step": 8768 }, { "epoch": 6.4241758241758244, "grad_norm": 1.5546875, "learning_rate": 9.525245877980625e-05, "loss": 0.2009, "step": 8769 }, { "epoch": 6.424908424908425, "grad_norm": 0.80859375, "learning_rate": 9.525099444468739e-05, "loss": 0.1486, "step": 8770 }, { "epoch": 6.425641025641026, "grad_norm": 0.9140625, "learning_rate": 9.524952989515181e-05, "loss": 0.1389, "step": 8771 }, { "epoch": 6.426373626373627, "grad_norm": 1.4296875, "learning_rate": 9.524806513120656e-05, "loss": 0.1705, "step": 8772 }, { "epoch": 6.427106227106227, "grad_norm": 1.046875, "learning_rate": 9.524660015285863e-05, "loss": 0.1271, "step": 8773 }, { "epoch": 6.427838827838828, "grad_norm": 1.1328125, "learning_rate": 9.524513496011505e-05, "loss": 0.1487, "step": 8774 }, { "epoch": 6.428571428571429, "grad_norm": 0.796875, "learning_rate": 9.524366955298284e-05, "loss": 0.1889, "step": 8775 }, { "epoch": 6.4293040293040296, "grad_norm": 0.71484375, "learning_rate": 9.524220393146901e-05, "loss": 0.1195, "step": 8776 }, { "epoch": 6.43003663003663, "grad_norm": 0.9375, "learning_rate": 9.52407380955806e-05, "loss": 0.1323, "step": 8777 }, { "epoch": 6.430769230769231, "grad_norm": 0.9453125, "learning_rate": 9.523927204532464e-05, "loss": 0.1187, "step": 8778 }, { "epoch": 6.431501831501832, "grad_norm": 0.84765625, "learning_rate": 9.523780578070812e-05, "loss": 0.1602, "step": 8779 }, { "epoch": 6.4322344322344325, "grad_norm": 1.203125, "learning_rate": 9.52363393017381e-05, "loss": 0.1424, "step": 8780 }, { "epoch": 6.432967032967033, "grad_norm": 1.0390625, "learning_rate": 9.523487260842157e-05, "loss": 0.1577, "step": 8781 }, { "epoch": 6.433699633699634, "grad_norm": 0.890625, "learning_rate": 9.523340570076561e-05, "loss": 0.1472, "step": 8782 }, { "epoch": 6.434432234432235, "grad_norm": 1.390625, "learning_rate": 9.523193857877721e-05, "loss": 0.1625, "step": 8783 }, { "epoch": 6.435164835164835, "grad_norm": 0.8984375, "learning_rate": 9.523047124246341e-05, "loss": 0.1479, "step": 8784 }, { "epoch": 6.435897435897436, "grad_norm": 1.640625, "learning_rate": 9.522900369183122e-05, "loss": 0.2239, "step": 8785 }, { "epoch": 6.436630036630037, "grad_norm": 1.0234375, "learning_rate": 9.522753592688771e-05, "loss": 0.1334, "step": 8786 }, { "epoch": 6.437362637362638, "grad_norm": 0.82421875, "learning_rate": 9.522606794763989e-05, "loss": 0.1353, "step": 8787 }, { "epoch": 6.438095238095238, "grad_norm": 0.890625, "learning_rate": 9.522459975409478e-05, "loss": 0.1478, "step": 8788 }, { "epoch": 6.438827838827839, "grad_norm": 0.8203125, "learning_rate": 9.522313134625944e-05, "loss": 0.122, "step": 8789 }, { "epoch": 6.43956043956044, "grad_norm": 0.9609375, "learning_rate": 9.52216627241409e-05, "loss": 0.1264, "step": 8790 }, { "epoch": 6.4402930402930405, "grad_norm": 1.4609375, "learning_rate": 9.52201938877462e-05, "loss": 0.129, "step": 8791 }, { "epoch": 6.441025641025641, "grad_norm": 1.0859375, "learning_rate": 9.521872483708235e-05, "loss": 0.1246, "step": 8792 }, { "epoch": 6.441758241758242, "grad_norm": 0.91015625, "learning_rate": 9.521725557215642e-05, "loss": 0.1389, "step": 8793 }, { "epoch": 6.442490842490843, "grad_norm": 0.99609375, "learning_rate": 9.521578609297544e-05, "loss": 0.1319, "step": 8794 }, { "epoch": 6.443223443223443, "grad_norm": 0.91796875, "learning_rate": 9.521431639954644e-05, "loss": 0.1241, "step": 8795 }, { "epoch": 6.443956043956044, "grad_norm": 1.171875, "learning_rate": 9.521284649187648e-05, "loss": 0.1551, "step": 8796 }, { "epoch": 6.444688644688645, "grad_norm": 0.7265625, "learning_rate": 9.521137636997259e-05, "loss": 0.1289, "step": 8797 }, { "epoch": 6.445421245421246, "grad_norm": 0.55859375, "learning_rate": 9.520990603384183e-05, "loss": 0.1199, "step": 8798 }, { "epoch": 6.446153846153846, "grad_norm": 0.8046875, "learning_rate": 9.520843548349121e-05, "loss": 0.16, "step": 8799 }, { "epoch": 6.446886446886447, "grad_norm": 1.1015625, "learning_rate": 9.520696471892783e-05, "loss": 0.1292, "step": 8800 }, { "epoch": 6.447619047619048, "grad_norm": 1.28125, "learning_rate": 9.52054937401587e-05, "loss": 0.216, "step": 8801 }, { "epoch": 6.4483516483516485, "grad_norm": 1.25, "learning_rate": 9.520402254719084e-05, "loss": 0.1803, "step": 8802 }, { "epoch": 6.449084249084249, "grad_norm": 0.921875, "learning_rate": 9.520255114003136e-05, "loss": 0.1282, "step": 8803 }, { "epoch": 6.44981684981685, "grad_norm": 1.0234375, "learning_rate": 9.520107951868727e-05, "loss": 0.1255, "step": 8804 }, { "epoch": 6.450549450549451, "grad_norm": 0.98828125, "learning_rate": 9.519960768316564e-05, "loss": 0.1451, "step": 8805 }, { "epoch": 6.4512820512820515, "grad_norm": 1.6015625, "learning_rate": 9.519813563347352e-05, "loss": 0.1527, "step": 8806 }, { "epoch": 6.452014652014652, "grad_norm": 1.2421875, "learning_rate": 9.519666336961796e-05, "loss": 0.1426, "step": 8807 }, { "epoch": 6.452747252747253, "grad_norm": 0.61328125, "learning_rate": 9.519519089160602e-05, "loss": 0.1091, "step": 8808 }, { "epoch": 6.453479853479854, "grad_norm": 1.4375, "learning_rate": 9.519371819944472e-05, "loss": 0.1274, "step": 8809 }, { "epoch": 6.454212454212454, "grad_norm": 0.90234375, "learning_rate": 9.519224529314115e-05, "loss": 0.1275, "step": 8810 }, { "epoch": 6.454945054945055, "grad_norm": 0.90234375, "learning_rate": 9.519077217270237e-05, "loss": 0.1546, "step": 8811 }, { "epoch": 6.455677655677656, "grad_norm": 1.7109375, "learning_rate": 9.518929883813544e-05, "loss": 0.17, "step": 8812 }, { "epoch": 6.456410256410257, "grad_norm": 1.1328125, "learning_rate": 9.51878252894474e-05, "loss": 0.1354, "step": 8813 }, { "epoch": 6.457142857142857, "grad_norm": 0.90234375, "learning_rate": 9.518635152664531e-05, "loss": 0.1433, "step": 8814 }, { "epoch": 6.457875457875458, "grad_norm": 0.53125, "learning_rate": 9.518487754973625e-05, "loss": 0.1573, "step": 8815 }, { "epoch": 6.458608058608059, "grad_norm": 1.1484375, "learning_rate": 9.518340335872728e-05, "loss": 0.1661, "step": 8816 }, { "epoch": 6.4593406593406595, "grad_norm": 0.92578125, "learning_rate": 9.518192895362544e-05, "loss": 0.1573, "step": 8817 }, { "epoch": 6.46007326007326, "grad_norm": 1.0703125, "learning_rate": 9.518045433443782e-05, "loss": 0.1007, "step": 8818 }, { "epoch": 6.460805860805861, "grad_norm": 1.0234375, "learning_rate": 9.517897950117147e-05, "loss": 0.1748, "step": 8819 }, { "epoch": 6.461538461538462, "grad_norm": 1.234375, "learning_rate": 9.517750445383347e-05, "loss": 0.1722, "step": 8820 }, { "epoch": 6.462271062271062, "grad_norm": 1.21875, "learning_rate": 9.517602919243087e-05, "loss": 0.1383, "step": 8821 }, { "epoch": 6.463003663003663, "grad_norm": 1.078125, "learning_rate": 9.517455371697076e-05, "loss": 0.1557, "step": 8822 }, { "epoch": 6.463736263736264, "grad_norm": 1.09375, "learning_rate": 9.517307802746018e-05, "loss": 0.1811, "step": 8823 }, { "epoch": 6.464468864468865, "grad_norm": 0.953125, "learning_rate": 9.517160212390624e-05, "loss": 0.1193, "step": 8824 }, { "epoch": 6.465201465201465, "grad_norm": 0.9765625, "learning_rate": 9.517012600631598e-05, "loss": 0.1364, "step": 8825 }, { "epoch": 6.465934065934066, "grad_norm": 1.59375, "learning_rate": 9.516864967469648e-05, "loss": 0.1523, "step": 8826 }, { "epoch": 6.466666666666667, "grad_norm": 1.8359375, "learning_rate": 9.516717312905481e-05, "loss": 0.1613, "step": 8827 }, { "epoch": 6.4673992673992675, "grad_norm": 1.6171875, "learning_rate": 9.516569636939805e-05, "loss": 0.1578, "step": 8828 }, { "epoch": 6.468131868131868, "grad_norm": 1.53125, "learning_rate": 9.516421939573328e-05, "loss": 0.1676, "step": 8829 }, { "epoch": 6.468864468864469, "grad_norm": 1.71875, "learning_rate": 9.516274220806757e-05, "loss": 0.1512, "step": 8830 }, { "epoch": 6.46959706959707, "grad_norm": 0.96484375, "learning_rate": 9.516126480640801e-05, "loss": 0.1661, "step": 8831 }, { "epoch": 6.47032967032967, "grad_norm": 0.828125, "learning_rate": 9.515978719076165e-05, "loss": 0.1294, "step": 8832 }, { "epoch": 6.471062271062271, "grad_norm": 1.1484375, "learning_rate": 9.515830936113561e-05, "loss": 0.1291, "step": 8833 }, { "epoch": 6.471794871794872, "grad_norm": 1.1484375, "learning_rate": 9.515683131753694e-05, "loss": 0.1254, "step": 8834 }, { "epoch": 6.472527472527473, "grad_norm": 0.9296875, "learning_rate": 9.515535305997274e-05, "loss": 0.1391, "step": 8835 }, { "epoch": 6.473260073260073, "grad_norm": 0.76953125, "learning_rate": 9.515387458845007e-05, "loss": 0.1197, "step": 8836 }, { "epoch": 6.473992673992674, "grad_norm": 1.078125, "learning_rate": 9.515239590297604e-05, "loss": 0.1401, "step": 8837 }, { "epoch": 6.474725274725275, "grad_norm": 0.8359375, "learning_rate": 9.515091700355772e-05, "loss": 0.1367, "step": 8838 }, { "epoch": 6.4754578754578755, "grad_norm": 1.0859375, "learning_rate": 9.514943789020219e-05, "loss": 0.1269, "step": 8839 }, { "epoch": 6.476190476190476, "grad_norm": 0.90234375, "learning_rate": 9.514795856291654e-05, "loss": 0.1155, "step": 8840 }, { "epoch": 6.476923076923077, "grad_norm": 0.85546875, "learning_rate": 9.514647902170789e-05, "loss": 0.1348, "step": 8841 }, { "epoch": 6.477655677655678, "grad_norm": 1.0, "learning_rate": 9.51449992665833e-05, "loss": 0.1207, "step": 8842 }, { "epoch": 6.4783882783882785, "grad_norm": 1.5, "learning_rate": 9.514351929754985e-05, "loss": 0.1413, "step": 8843 }, { "epoch": 6.479120879120879, "grad_norm": 1.453125, "learning_rate": 9.514203911461463e-05, "loss": 0.1565, "step": 8844 }, { "epoch": 6.47985347985348, "grad_norm": 1.015625, "learning_rate": 9.514055871778477e-05, "loss": 0.1625, "step": 8845 }, { "epoch": 6.480586080586081, "grad_norm": 0.73828125, "learning_rate": 9.513907810706732e-05, "loss": 0.1311, "step": 8846 }, { "epoch": 6.481318681318681, "grad_norm": 0.6171875, "learning_rate": 9.513759728246941e-05, "loss": 0.1365, "step": 8847 }, { "epoch": 6.482051282051282, "grad_norm": 0.85546875, "learning_rate": 9.51361162439981e-05, "loss": 0.1834, "step": 8848 }, { "epoch": 6.482783882783883, "grad_norm": 1.078125, "learning_rate": 9.513463499166052e-05, "loss": 0.1852, "step": 8849 }, { "epoch": 6.483516483516484, "grad_norm": 0.76171875, "learning_rate": 9.513315352546373e-05, "loss": 0.1571, "step": 8850 }, { "epoch": 6.484249084249084, "grad_norm": 0.8203125, "learning_rate": 9.513167184541487e-05, "loss": 0.1312, "step": 8851 }, { "epoch": 6.484981684981685, "grad_norm": 0.98046875, "learning_rate": 9.5130189951521e-05, "loss": 0.1445, "step": 8852 }, { "epoch": 6.485714285714286, "grad_norm": 0.796875, "learning_rate": 9.512870784378927e-05, "loss": 0.1478, "step": 8853 }, { "epoch": 6.4864468864468865, "grad_norm": 0.95703125, "learning_rate": 9.512722552222672e-05, "loss": 0.1609, "step": 8854 }, { "epoch": 6.487179487179487, "grad_norm": 1.0859375, "learning_rate": 9.512574298684049e-05, "loss": 0.1612, "step": 8855 }, { "epoch": 6.487912087912088, "grad_norm": 1.28125, "learning_rate": 9.512426023763768e-05, "loss": 0.1722, "step": 8856 }, { "epoch": 6.488644688644689, "grad_norm": 0.90234375, "learning_rate": 9.51227772746254e-05, "loss": 0.1434, "step": 8857 }, { "epoch": 6.489377289377289, "grad_norm": 0.7109375, "learning_rate": 9.512129409781073e-05, "loss": 0.122, "step": 8858 }, { "epoch": 6.49010989010989, "grad_norm": 0.83984375, "learning_rate": 9.51198107072008e-05, "loss": 0.15, "step": 8859 }, { "epoch": 6.490842490842491, "grad_norm": 1.0, "learning_rate": 9.51183271028027e-05, "loss": 0.1498, "step": 8860 }, { "epoch": 6.491575091575092, "grad_norm": 1.09375, "learning_rate": 9.511684328462354e-05, "loss": 0.1526, "step": 8861 }, { "epoch": 6.492307692307692, "grad_norm": 0.8671875, "learning_rate": 9.511535925267047e-05, "loss": 0.1517, "step": 8862 }, { "epoch": 6.493040293040293, "grad_norm": 1.0078125, "learning_rate": 9.511387500695054e-05, "loss": 0.1319, "step": 8863 }, { "epoch": 6.493772893772894, "grad_norm": 1.4140625, "learning_rate": 9.51123905474709e-05, "loss": 0.1291, "step": 8864 }, { "epoch": 6.4945054945054945, "grad_norm": 1.3046875, "learning_rate": 9.511090587423865e-05, "loss": 0.1585, "step": 8865 }, { "epoch": 6.495238095238095, "grad_norm": 1.3359375, "learning_rate": 9.51094209872609e-05, "loss": 0.1639, "step": 8866 }, { "epoch": 6.495970695970696, "grad_norm": 0.93359375, "learning_rate": 9.510793588654477e-05, "loss": 0.1339, "step": 8867 }, { "epoch": 6.496703296703297, "grad_norm": 1.28125, "learning_rate": 9.510645057209738e-05, "loss": 0.1688, "step": 8868 }, { "epoch": 6.4974358974358974, "grad_norm": 1.1796875, "learning_rate": 9.510496504392584e-05, "loss": 0.1208, "step": 8869 }, { "epoch": 6.498168498168498, "grad_norm": 2.09375, "learning_rate": 9.510347930203727e-05, "loss": 0.1522, "step": 8870 }, { "epoch": 6.498901098901099, "grad_norm": 1.953125, "learning_rate": 9.510199334643879e-05, "loss": 0.1495, "step": 8871 }, { "epoch": 6.4996336996337, "grad_norm": 0.95703125, "learning_rate": 9.510050717713751e-05, "loss": 0.1258, "step": 8872 }, { "epoch": 6.5003663003663, "grad_norm": 1.109375, "learning_rate": 9.509902079414057e-05, "loss": 0.1569, "step": 8873 }, { "epoch": 6.501098901098901, "grad_norm": 0.80859375, "learning_rate": 9.509753419745506e-05, "loss": 0.1604, "step": 8874 }, { "epoch": 6.501831501831502, "grad_norm": 1.1484375, "learning_rate": 9.509604738708815e-05, "loss": 0.167, "step": 8875 }, { "epoch": 6.5025641025641026, "grad_norm": 0.859375, "learning_rate": 9.509456036304692e-05, "loss": 0.1298, "step": 8876 }, { "epoch": 6.503296703296703, "grad_norm": 1.4609375, "learning_rate": 9.509307312533853e-05, "loss": 0.1294, "step": 8877 }, { "epoch": 6.504029304029304, "grad_norm": 0.95703125, "learning_rate": 9.509158567397007e-05, "loss": 0.1459, "step": 8878 }, { "epoch": 6.504761904761905, "grad_norm": 1.2265625, "learning_rate": 9.50900980089487e-05, "loss": 0.1831, "step": 8879 }, { "epoch": 6.5054945054945055, "grad_norm": 0.69140625, "learning_rate": 9.508861013028151e-05, "loss": 0.1119, "step": 8880 }, { "epoch": 6.506227106227106, "grad_norm": 0.703125, "learning_rate": 9.508712203797566e-05, "loss": 0.1137, "step": 8881 }, { "epoch": 6.506959706959707, "grad_norm": 1.3984375, "learning_rate": 9.508563373203828e-05, "loss": 0.1441, "step": 8882 }, { "epoch": 6.507692307692308, "grad_norm": 1.3515625, "learning_rate": 9.508414521247648e-05, "loss": 0.1278, "step": 8883 }, { "epoch": 6.508424908424908, "grad_norm": 0.83203125, "learning_rate": 9.508265647929742e-05, "loss": 0.1328, "step": 8884 }, { "epoch": 6.509157509157509, "grad_norm": 1.0859375, "learning_rate": 9.508116753250819e-05, "loss": 0.1535, "step": 8885 }, { "epoch": 6.50989010989011, "grad_norm": 1.3203125, "learning_rate": 9.507967837211598e-05, "loss": 0.1434, "step": 8886 }, { "epoch": 6.510622710622711, "grad_norm": 1.0703125, "learning_rate": 9.50781889981279e-05, "loss": 0.1361, "step": 8887 }, { "epoch": 6.511355311355311, "grad_norm": 2.4375, "learning_rate": 9.507669941055105e-05, "loss": 0.1591, "step": 8888 }, { "epoch": 6.512087912087912, "grad_norm": 1.3984375, "learning_rate": 9.507520960939262e-05, "loss": 0.1215, "step": 8889 }, { "epoch": 6.512820512820513, "grad_norm": 1.078125, "learning_rate": 9.507371959465973e-05, "loss": 0.1494, "step": 8890 }, { "epoch": 6.5135531135531135, "grad_norm": 0.93359375, "learning_rate": 9.507222936635952e-05, "loss": 0.133, "step": 8891 }, { "epoch": 6.514285714285714, "grad_norm": 1.078125, "learning_rate": 9.507073892449913e-05, "loss": 0.1415, "step": 8892 }, { "epoch": 6.515018315018315, "grad_norm": 1.5078125, "learning_rate": 9.506924826908569e-05, "loss": 0.1673, "step": 8893 }, { "epoch": 6.515750915750916, "grad_norm": 1.4375, "learning_rate": 9.506775740012633e-05, "loss": 0.119, "step": 8894 }, { "epoch": 6.516483516483516, "grad_norm": 1.765625, "learning_rate": 9.506626631762824e-05, "loss": 0.13, "step": 8895 }, { "epoch": 6.517216117216117, "grad_norm": 1.203125, "learning_rate": 9.506477502159854e-05, "loss": 0.1507, "step": 8896 }, { "epoch": 6.517948717948718, "grad_norm": 0.9453125, "learning_rate": 9.506328351204435e-05, "loss": 0.1282, "step": 8897 }, { "epoch": 6.518681318681319, "grad_norm": 0.75390625, "learning_rate": 9.506179178897286e-05, "loss": 0.1819, "step": 8898 }, { "epoch": 6.519413919413919, "grad_norm": 1.265625, "learning_rate": 9.506029985239118e-05, "loss": 0.1631, "step": 8899 }, { "epoch": 6.52014652014652, "grad_norm": 1.21875, "learning_rate": 9.50588077023065e-05, "loss": 0.1871, "step": 8900 }, { "epoch": 6.520879120879121, "grad_norm": 1.09375, "learning_rate": 9.505731533872591e-05, "loss": 0.1491, "step": 8901 }, { "epoch": 6.5216117216117215, "grad_norm": 1.3828125, "learning_rate": 9.505582276165662e-05, "loss": 0.156, "step": 8902 }, { "epoch": 6.522344322344322, "grad_norm": 0.7265625, "learning_rate": 9.505432997110577e-05, "loss": 0.1024, "step": 8903 }, { "epoch": 6.523076923076923, "grad_norm": 0.69921875, "learning_rate": 9.505283696708046e-05, "loss": 0.138, "step": 8904 }, { "epoch": 6.523809523809524, "grad_norm": 0.6484375, "learning_rate": 9.505134374958791e-05, "loss": 0.1421, "step": 8905 }, { "epoch": 6.5245421245421245, "grad_norm": 0.765625, "learning_rate": 9.504985031863522e-05, "loss": 0.1502, "step": 8906 }, { "epoch": 6.525274725274725, "grad_norm": 0.7734375, "learning_rate": 9.504835667422961e-05, "loss": 0.1264, "step": 8907 }, { "epoch": 6.526007326007326, "grad_norm": 0.71484375, "learning_rate": 9.504686281637817e-05, "loss": 0.1303, "step": 8908 }, { "epoch": 6.526739926739927, "grad_norm": 1.671875, "learning_rate": 9.50453687450881e-05, "loss": 0.1528, "step": 8909 }, { "epoch": 6.527472527472527, "grad_norm": 1.2734375, "learning_rate": 9.504387446036655e-05, "loss": 0.1711, "step": 8910 }, { "epoch": 6.528205128205128, "grad_norm": 0.89453125, "learning_rate": 9.504237996222067e-05, "loss": 0.1316, "step": 8911 }, { "epoch": 6.528937728937729, "grad_norm": 1.265625, "learning_rate": 9.504088525065762e-05, "loss": 0.1621, "step": 8912 }, { "epoch": 6.52967032967033, "grad_norm": 0.6875, "learning_rate": 9.503939032568456e-05, "loss": 0.1397, "step": 8913 }, { "epoch": 6.53040293040293, "grad_norm": 0.78125, "learning_rate": 9.503789518730867e-05, "loss": 0.1335, "step": 8914 }, { "epoch": 6.531135531135531, "grad_norm": 1.2109375, "learning_rate": 9.50363998355371e-05, "loss": 0.1334, "step": 8915 }, { "epoch": 6.531868131868132, "grad_norm": 0.8359375, "learning_rate": 9.503490427037703e-05, "loss": 0.1199, "step": 8916 }, { "epoch": 6.5326007326007325, "grad_norm": 0.84375, "learning_rate": 9.503340849183558e-05, "loss": 0.1378, "step": 8917 }, { "epoch": 6.533333333333333, "grad_norm": 0.95703125, "learning_rate": 9.503191249991998e-05, "loss": 0.127, "step": 8918 }, { "epoch": 6.534065934065934, "grad_norm": 0.890625, "learning_rate": 9.503041629463737e-05, "loss": 0.1282, "step": 8919 }, { "epoch": 6.534798534798535, "grad_norm": 0.859375, "learning_rate": 9.502891987599491e-05, "loss": 0.1271, "step": 8920 }, { "epoch": 6.535531135531135, "grad_norm": 2.078125, "learning_rate": 9.502742324399978e-05, "loss": 0.145, "step": 8921 }, { "epoch": 6.536263736263736, "grad_norm": 1.3125, "learning_rate": 9.502592639865915e-05, "loss": 0.1378, "step": 8922 }, { "epoch": 6.536996336996337, "grad_norm": 0.94140625, "learning_rate": 9.502442933998018e-05, "loss": 0.1087, "step": 8923 }, { "epoch": 6.537728937728938, "grad_norm": 0.81640625, "learning_rate": 9.502293206797006e-05, "loss": 0.121, "step": 8924 }, { "epoch": 6.538461538461538, "grad_norm": 0.77734375, "learning_rate": 9.502143458263593e-05, "loss": 0.1429, "step": 8925 }, { "epoch": 6.539194139194139, "grad_norm": 1.0078125, "learning_rate": 9.501993688398503e-05, "loss": 0.1181, "step": 8926 }, { "epoch": 6.53992673992674, "grad_norm": 1.1953125, "learning_rate": 9.501843897202449e-05, "loss": 0.1604, "step": 8927 }, { "epoch": 6.5406593406593405, "grad_norm": 0.40625, "learning_rate": 9.501694084676149e-05, "loss": 0.114, "step": 8928 }, { "epoch": 6.541391941391941, "grad_norm": 0.58203125, "learning_rate": 9.50154425082032e-05, "loss": 0.0967, "step": 8929 }, { "epoch": 6.542124542124542, "grad_norm": 0.65234375, "learning_rate": 9.501394395635683e-05, "loss": 0.1446, "step": 8930 }, { "epoch": 6.542857142857143, "grad_norm": 0.6640625, "learning_rate": 9.501244519122952e-05, "loss": 0.1129, "step": 8931 }, { "epoch": 6.543589743589743, "grad_norm": 1.375, "learning_rate": 9.50109462128285e-05, "loss": 0.1577, "step": 8932 }, { "epoch": 6.544322344322344, "grad_norm": 0.8046875, "learning_rate": 9.500944702116089e-05, "loss": 0.1214, "step": 8933 }, { "epoch": 6.545054945054945, "grad_norm": 0.6875, "learning_rate": 9.500794761623393e-05, "loss": 0.1387, "step": 8934 }, { "epoch": 6.545787545787546, "grad_norm": 1.0078125, "learning_rate": 9.500644799805478e-05, "loss": 0.1534, "step": 8935 }, { "epoch": 6.546520146520146, "grad_norm": 0.6484375, "learning_rate": 9.500494816663062e-05, "loss": 0.1385, "step": 8936 }, { "epoch": 6.547252747252747, "grad_norm": 1.0234375, "learning_rate": 9.500344812196866e-05, "loss": 0.1691, "step": 8937 }, { "epoch": 6.547985347985348, "grad_norm": 0.9609375, "learning_rate": 9.500194786407605e-05, "loss": 0.1313, "step": 8938 }, { "epoch": 6.5487179487179485, "grad_norm": 1.203125, "learning_rate": 9.500044739296002e-05, "loss": 0.1531, "step": 8939 }, { "epoch": 6.549450549450549, "grad_norm": 0.67578125, "learning_rate": 9.499894670862771e-05, "loss": 0.121, "step": 8940 }, { "epoch": 6.55018315018315, "grad_norm": 0.7578125, "learning_rate": 9.499744581108637e-05, "loss": 0.1142, "step": 8941 }, { "epoch": 6.550915750915751, "grad_norm": 1.15625, "learning_rate": 9.499594470034315e-05, "loss": 0.1962, "step": 8942 }, { "epoch": 6.5516483516483515, "grad_norm": 0.8984375, "learning_rate": 9.499444337640524e-05, "loss": 0.1622, "step": 8943 }, { "epoch": 6.552380952380952, "grad_norm": 0.99609375, "learning_rate": 9.499294183927985e-05, "loss": 0.165, "step": 8944 }, { "epoch": 6.553113553113553, "grad_norm": 0.82421875, "learning_rate": 9.499144008897417e-05, "loss": 0.1441, "step": 8945 }, { "epoch": 6.553846153846154, "grad_norm": 0.57421875, "learning_rate": 9.49899381254954e-05, "loss": 0.1229, "step": 8946 }, { "epoch": 6.554578754578754, "grad_norm": 0.9140625, "learning_rate": 9.498843594885073e-05, "loss": 0.1318, "step": 8947 }, { "epoch": 6.555311355311355, "grad_norm": 1.0703125, "learning_rate": 9.498693355904736e-05, "loss": 0.2097, "step": 8948 }, { "epoch": 6.556043956043956, "grad_norm": 1.203125, "learning_rate": 9.49854309560925e-05, "loss": 0.1514, "step": 8949 }, { "epoch": 6.556776556776557, "grad_norm": 0.91015625, "learning_rate": 9.49839281399933e-05, "loss": 0.1285, "step": 8950 }, { "epoch": 6.557509157509157, "grad_norm": 1.171875, "learning_rate": 9.498242511075703e-05, "loss": 0.1867, "step": 8951 }, { "epoch": 6.558241758241758, "grad_norm": 0.9921875, "learning_rate": 9.498092186839083e-05, "loss": 0.128, "step": 8952 }, { "epoch": 6.558974358974359, "grad_norm": 0.84375, "learning_rate": 9.497941841290195e-05, "loss": 0.1101, "step": 8953 }, { "epoch": 6.5597069597069595, "grad_norm": 0.75390625, "learning_rate": 9.497791474429757e-05, "loss": 0.1448, "step": 8954 }, { "epoch": 6.56043956043956, "grad_norm": 1.265625, "learning_rate": 9.49764108625849e-05, "loss": 0.1556, "step": 8955 }, { "epoch": 6.561172161172161, "grad_norm": 0.609375, "learning_rate": 9.497490676777116e-05, "loss": 0.1306, "step": 8956 }, { "epoch": 6.561904761904762, "grad_norm": 0.98046875, "learning_rate": 9.497340245986352e-05, "loss": 0.1595, "step": 8957 }, { "epoch": 6.562637362637362, "grad_norm": 0.9609375, "learning_rate": 9.497189793886921e-05, "loss": 0.1479, "step": 8958 }, { "epoch": 6.563369963369963, "grad_norm": 1.1328125, "learning_rate": 9.497039320479545e-05, "loss": 0.16, "step": 8959 }, { "epoch": 6.564102564102564, "grad_norm": 1.1484375, "learning_rate": 9.496888825764943e-05, "loss": 0.1326, "step": 8960 }, { "epoch": 6.564835164835165, "grad_norm": 0.75390625, "learning_rate": 9.496738309743836e-05, "loss": 0.1754, "step": 8961 }, { "epoch": 6.565567765567765, "grad_norm": 0.859375, "learning_rate": 9.496587772416948e-05, "loss": 0.1043, "step": 8962 }, { "epoch": 6.566300366300366, "grad_norm": 1.15625, "learning_rate": 9.496437213784998e-05, "loss": 0.1336, "step": 8963 }, { "epoch": 6.567032967032967, "grad_norm": 1.1171875, "learning_rate": 9.496286633848706e-05, "loss": 0.1255, "step": 8964 }, { "epoch": 6.5677655677655675, "grad_norm": 0.98046875, "learning_rate": 9.496136032608796e-05, "loss": 0.17, "step": 8965 }, { "epoch": 6.568498168498168, "grad_norm": 1.1171875, "learning_rate": 9.495985410065989e-05, "loss": 0.1431, "step": 8966 }, { "epoch": 6.569230769230769, "grad_norm": 0.9765625, "learning_rate": 9.495834766221005e-05, "loss": 0.1244, "step": 8967 }, { "epoch": 6.56996336996337, "grad_norm": 0.77734375, "learning_rate": 9.49568410107457e-05, "loss": 0.1641, "step": 8968 }, { "epoch": 6.5706959706959704, "grad_norm": 2.28125, "learning_rate": 9.4955334146274e-05, "loss": 0.1668, "step": 8969 }, { "epoch": 6.571428571428571, "grad_norm": 0.74609375, "learning_rate": 9.495382706880222e-05, "loss": 0.1521, "step": 8970 }, { "epoch": 6.572161172161172, "grad_norm": 0.87109375, "learning_rate": 9.495231977833757e-05, "loss": 0.1258, "step": 8971 }, { "epoch": 6.572893772893773, "grad_norm": 0.734375, "learning_rate": 9.495081227488725e-05, "loss": 0.1248, "step": 8972 }, { "epoch": 6.573626373626373, "grad_norm": 0.9296875, "learning_rate": 9.494930455845848e-05, "loss": 0.1354, "step": 8973 }, { "epoch": 6.574358974358974, "grad_norm": 1.2734375, "learning_rate": 9.49477966290585e-05, "loss": 0.127, "step": 8974 }, { "epoch": 6.575091575091575, "grad_norm": 1.0078125, "learning_rate": 9.494628848669456e-05, "loss": 0.1188, "step": 8975 }, { "epoch": 6.5758241758241756, "grad_norm": 1.0390625, "learning_rate": 9.494478013137385e-05, "loss": 0.1714, "step": 8976 }, { "epoch": 6.576556776556776, "grad_norm": 0.85546875, "learning_rate": 9.494327156310362e-05, "loss": 0.1393, "step": 8977 }, { "epoch": 6.577289377289377, "grad_norm": 1.7421875, "learning_rate": 9.494176278189107e-05, "loss": 0.1735, "step": 8978 }, { "epoch": 6.578021978021978, "grad_norm": 0.79296875, "learning_rate": 9.494025378774345e-05, "loss": 0.1526, "step": 8979 }, { "epoch": 6.5787545787545785, "grad_norm": 0.890625, "learning_rate": 9.493874458066799e-05, "loss": 0.1213, "step": 8980 }, { "epoch": 6.579487179487179, "grad_norm": 0.9140625, "learning_rate": 9.493723516067192e-05, "loss": 0.1435, "step": 8981 }, { "epoch": 6.58021978021978, "grad_norm": 0.91015625, "learning_rate": 9.493572552776246e-05, "loss": 0.1596, "step": 8982 }, { "epoch": 6.580952380952381, "grad_norm": 0.74609375, "learning_rate": 9.493421568194685e-05, "loss": 0.1004, "step": 8983 }, { "epoch": 6.581684981684981, "grad_norm": 1.5, "learning_rate": 9.493270562323232e-05, "loss": 0.1769, "step": 8984 }, { "epoch": 6.582417582417582, "grad_norm": 2.875, "learning_rate": 9.493119535162613e-05, "loss": 0.1698, "step": 8985 }, { "epoch": 6.583150183150183, "grad_norm": 1.265625, "learning_rate": 9.492968486713548e-05, "loss": 0.1696, "step": 8986 }, { "epoch": 6.583882783882784, "grad_norm": 1.7734375, "learning_rate": 9.492817416976762e-05, "loss": 0.172, "step": 8987 }, { "epoch": 6.584615384615384, "grad_norm": 1.0234375, "learning_rate": 9.492666325952981e-05, "loss": 0.1431, "step": 8988 }, { "epoch": 6.585347985347985, "grad_norm": 0.73046875, "learning_rate": 9.492515213642927e-05, "loss": 0.1339, "step": 8989 }, { "epoch": 6.586080586080586, "grad_norm": 1.2890625, "learning_rate": 9.492364080047324e-05, "loss": 0.1577, "step": 8990 }, { "epoch": 6.5868131868131865, "grad_norm": 1.1484375, "learning_rate": 9.492212925166896e-05, "loss": 0.1717, "step": 8991 }, { "epoch": 6.587545787545787, "grad_norm": 1.078125, "learning_rate": 9.492061749002368e-05, "loss": 0.1512, "step": 8992 }, { "epoch": 6.588278388278388, "grad_norm": 0.6875, "learning_rate": 9.491910551554464e-05, "loss": 0.1439, "step": 8993 }, { "epoch": 6.589010989010989, "grad_norm": 0.7265625, "learning_rate": 9.491759332823908e-05, "loss": 0.1268, "step": 8994 }, { "epoch": 6.589743589743589, "grad_norm": 1.4296875, "learning_rate": 9.491608092811425e-05, "loss": 0.1696, "step": 8995 }, { "epoch": 6.59047619047619, "grad_norm": 1.2109375, "learning_rate": 9.491456831517738e-05, "loss": 0.1428, "step": 8996 }, { "epoch": 6.591208791208791, "grad_norm": 1.0546875, "learning_rate": 9.491305548943575e-05, "loss": 0.1174, "step": 8997 }, { "epoch": 6.591941391941392, "grad_norm": 0.69140625, "learning_rate": 9.491154245089659e-05, "loss": 0.1267, "step": 8998 }, { "epoch": 6.592673992673992, "grad_norm": 0.609375, "learning_rate": 9.491002919956715e-05, "loss": 0.1576, "step": 8999 }, { "epoch": 6.593406593406593, "grad_norm": 0.6640625, "learning_rate": 9.490851573545469e-05, "loss": 0.1223, "step": 9000 }, { "epoch": 6.594139194139194, "grad_norm": 1.0703125, "learning_rate": 9.490700205856644e-05, "loss": 0.1307, "step": 9001 }, { "epoch": 6.5948717948717945, "grad_norm": 0.95703125, "learning_rate": 9.490548816890967e-05, "loss": 0.1312, "step": 9002 }, { "epoch": 6.595604395604395, "grad_norm": 1.2421875, "learning_rate": 9.490397406649163e-05, "loss": 0.1441, "step": 9003 }, { "epoch": 6.596336996336996, "grad_norm": 0.96875, "learning_rate": 9.490245975131956e-05, "loss": 0.1405, "step": 9004 }, { "epoch": 6.597069597069597, "grad_norm": 1.296875, "learning_rate": 9.490094522340075e-05, "loss": 0.1526, "step": 9005 }, { "epoch": 6.5978021978021975, "grad_norm": 0.90625, "learning_rate": 9.489943048274242e-05, "loss": 0.1538, "step": 9006 }, { "epoch": 6.598534798534798, "grad_norm": 1.015625, "learning_rate": 9.489791552935186e-05, "loss": 0.1538, "step": 9007 }, { "epoch": 6.599267399267399, "grad_norm": 0.98828125, "learning_rate": 9.489640036323629e-05, "loss": 0.1154, "step": 9008 }, { "epoch": 6.6, "grad_norm": 0.7890625, "learning_rate": 9.489488498440301e-05, "loss": 0.1327, "step": 9009 }, { "epoch": 6.6007326007326, "grad_norm": 0.99609375, "learning_rate": 9.489336939285924e-05, "loss": 0.1206, "step": 9010 }, { "epoch": 6.601465201465201, "grad_norm": 1.359375, "learning_rate": 9.489185358861229e-05, "loss": 0.1433, "step": 9011 }, { "epoch": 6.602197802197802, "grad_norm": 0.8671875, "learning_rate": 9.489033757166938e-05, "loss": 0.1348, "step": 9012 }, { "epoch": 6.602930402930403, "grad_norm": 1.0390625, "learning_rate": 9.488882134203779e-05, "loss": 0.1286, "step": 9013 }, { "epoch": 6.603663003663003, "grad_norm": 1.046875, "learning_rate": 9.48873048997248e-05, "loss": 0.1508, "step": 9014 }, { "epoch": 6.604395604395604, "grad_norm": 1.234375, "learning_rate": 9.488578824473765e-05, "loss": 0.162, "step": 9015 }, { "epoch": 6.605128205128205, "grad_norm": 1.3359375, "learning_rate": 9.488427137708363e-05, "loss": 0.1467, "step": 9016 }, { "epoch": 6.6058608058608055, "grad_norm": 1.1328125, "learning_rate": 9.488275429676997e-05, "loss": 0.1629, "step": 9017 }, { "epoch": 6.606593406593406, "grad_norm": 1.4921875, "learning_rate": 9.488123700380398e-05, "loss": 0.1201, "step": 9018 }, { "epoch": 6.607326007326007, "grad_norm": 0.75, "learning_rate": 9.487971949819292e-05, "loss": 0.1322, "step": 9019 }, { "epoch": 6.608058608058608, "grad_norm": 1.09375, "learning_rate": 9.487820177994404e-05, "loss": 0.1419, "step": 9020 }, { "epoch": 6.608791208791208, "grad_norm": 1.34375, "learning_rate": 9.487668384906465e-05, "loss": 0.1103, "step": 9021 }, { "epoch": 6.609523809523809, "grad_norm": 1.09375, "learning_rate": 9.487516570556196e-05, "loss": 0.1165, "step": 9022 }, { "epoch": 6.61025641025641, "grad_norm": 0.69140625, "learning_rate": 9.487364734944332e-05, "loss": 0.128, "step": 9023 }, { "epoch": 6.610989010989011, "grad_norm": 1.25, "learning_rate": 9.487212878071596e-05, "loss": 0.1578, "step": 9024 }, { "epoch": 6.611721611721611, "grad_norm": 0.99609375, "learning_rate": 9.487060999938715e-05, "loss": 0.1575, "step": 9025 }, { "epoch": 6.612454212454212, "grad_norm": 1.2265625, "learning_rate": 9.48690910054642e-05, "loss": 0.1154, "step": 9026 }, { "epoch": 6.613186813186813, "grad_norm": 0.78515625, "learning_rate": 9.486757179895435e-05, "loss": 0.1175, "step": 9027 }, { "epoch": 6.6139194139194135, "grad_norm": 1.109375, "learning_rate": 9.486605237986493e-05, "loss": 0.1451, "step": 9028 }, { "epoch": 6.614652014652014, "grad_norm": 1.671875, "learning_rate": 9.486453274820316e-05, "loss": 0.1124, "step": 9029 }, { "epoch": 6.615384615384615, "grad_norm": 0.63671875, "learning_rate": 9.486301290397635e-05, "loss": 0.1302, "step": 9030 }, { "epoch": 6.616117216117216, "grad_norm": 0.66015625, "learning_rate": 9.486149284719181e-05, "loss": 0.1124, "step": 9031 }, { "epoch": 6.616849816849816, "grad_norm": 1.3046875, "learning_rate": 9.485997257785676e-05, "loss": 0.1358, "step": 9032 }, { "epoch": 6.617582417582417, "grad_norm": 1.8203125, "learning_rate": 9.485845209597854e-05, "loss": 0.1788, "step": 9033 }, { "epoch": 6.618315018315018, "grad_norm": 2.25, "learning_rate": 9.485693140156441e-05, "loss": 0.1629, "step": 9034 }, { "epoch": 6.619047619047619, "grad_norm": 1.0234375, "learning_rate": 9.485541049462165e-05, "loss": 0.1425, "step": 9035 }, { "epoch": 6.619780219780219, "grad_norm": 0.78125, "learning_rate": 9.485388937515757e-05, "loss": 0.127, "step": 9036 }, { "epoch": 6.62051282051282, "grad_norm": 0.51171875, "learning_rate": 9.485236804317943e-05, "loss": 0.1301, "step": 9037 }, { "epoch": 6.621245421245421, "grad_norm": 0.76171875, "learning_rate": 9.485084649869457e-05, "loss": 0.1346, "step": 9038 }, { "epoch": 6.6219780219780215, "grad_norm": 0.83203125, "learning_rate": 9.48493247417102e-05, "loss": 0.1452, "step": 9039 }, { "epoch": 6.622710622710622, "grad_norm": 1.1015625, "learning_rate": 9.484780277223368e-05, "loss": 0.1408, "step": 9040 }, { "epoch": 6.623443223443223, "grad_norm": 1.5, "learning_rate": 9.484628059027228e-05, "loss": 0.1643, "step": 9041 }, { "epoch": 6.624175824175824, "grad_norm": 1.4609375, "learning_rate": 9.48447581958333e-05, "loss": 0.1449, "step": 9042 }, { "epoch": 6.6249084249084245, "grad_norm": 1.03125, "learning_rate": 9.484323558892399e-05, "loss": 0.1611, "step": 9043 }, { "epoch": 6.625641025641025, "grad_norm": 0.6484375, "learning_rate": 9.484171276955172e-05, "loss": 0.1476, "step": 9044 }, { "epoch": 6.626373626373626, "grad_norm": 0.73828125, "learning_rate": 9.484018973772374e-05, "loss": 0.1591, "step": 9045 }, { "epoch": 6.627106227106227, "grad_norm": 1.1953125, "learning_rate": 9.483866649344735e-05, "loss": 0.1615, "step": 9046 }, { "epoch": 6.627838827838827, "grad_norm": 1.3203125, "learning_rate": 9.483714303672985e-05, "loss": 0.122, "step": 9047 }, { "epoch": 6.628571428571428, "grad_norm": 1.2578125, "learning_rate": 9.483561936757854e-05, "loss": 0.1539, "step": 9048 }, { "epoch": 6.629304029304029, "grad_norm": 1.4140625, "learning_rate": 9.483409548600073e-05, "loss": 0.1432, "step": 9049 }, { "epoch": 6.63003663003663, "grad_norm": 1.296875, "learning_rate": 9.483257139200372e-05, "loss": 0.1381, "step": 9050 }, { "epoch": 6.63076923076923, "grad_norm": 1.15625, "learning_rate": 9.48310470855948e-05, "loss": 0.1395, "step": 9051 }, { "epoch": 6.631501831501831, "grad_norm": 1.1171875, "learning_rate": 9.482952256678127e-05, "loss": 0.1831, "step": 9052 }, { "epoch": 6.632234432234432, "grad_norm": 1.21875, "learning_rate": 9.482799783557044e-05, "loss": 0.1913, "step": 9053 }, { "epoch": 6.6329670329670325, "grad_norm": 0.98046875, "learning_rate": 9.482647289196966e-05, "loss": 0.1257, "step": 9054 }, { "epoch": 6.633699633699633, "grad_norm": 1.5859375, "learning_rate": 9.482494773598616e-05, "loss": 0.1492, "step": 9055 }, { "epoch": 6.634432234432234, "grad_norm": 0.99609375, "learning_rate": 9.482342236762729e-05, "loss": 0.1575, "step": 9056 }, { "epoch": 6.635164835164835, "grad_norm": 1.1171875, "learning_rate": 9.482189678690036e-05, "loss": 0.1379, "step": 9057 }, { "epoch": 6.635897435897435, "grad_norm": 1.71875, "learning_rate": 9.482037099381267e-05, "loss": 0.1442, "step": 9058 }, { "epoch": 6.636630036630036, "grad_norm": 0.91015625, "learning_rate": 9.481884498837154e-05, "loss": 0.1322, "step": 9059 }, { "epoch": 6.637362637362637, "grad_norm": 0.87890625, "learning_rate": 9.481731877058428e-05, "loss": 0.137, "step": 9060 }, { "epoch": 6.638095238095238, "grad_norm": 0.68359375, "learning_rate": 9.481579234045818e-05, "loss": 0.1179, "step": 9061 }, { "epoch": 6.638827838827838, "grad_norm": 0.90625, "learning_rate": 9.481426569800058e-05, "loss": 0.1502, "step": 9062 }, { "epoch": 6.639560439560439, "grad_norm": 0.6484375, "learning_rate": 9.481273884321878e-05, "loss": 0.1493, "step": 9063 }, { "epoch": 6.64029304029304, "grad_norm": 1.953125, "learning_rate": 9.481121177612012e-05, "loss": 0.1576, "step": 9064 }, { "epoch": 6.641025641025641, "grad_norm": 1.3046875, "learning_rate": 9.480968449671187e-05, "loss": 0.1442, "step": 9065 }, { "epoch": 6.641758241758242, "grad_norm": 0.97265625, "learning_rate": 9.480815700500138e-05, "loss": 0.1502, "step": 9066 }, { "epoch": 6.642490842490843, "grad_norm": 1.0625, "learning_rate": 9.4806629300996e-05, "loss": 0.1454, "step": 9067 }, { "epoch": 6.643223443223444, "grad_norm": 0.94140625, "learning_rate": 9.480510138470297e-05, "loss": 0.1329, "step": 9068 }, { "epoch": 6.643956043956044, "grad_norm": 0.8359375, "learning_rate": 9.480357325612967e-05, "loss": 0.1426, "step": 9069 }, { "epoch": 6.644688644688645, "grad_norm": 0.78125, "learning_rate": 9.480204491528341e-05, "loss": 0.1472, "step": 9070 }, { "epoch": 6.645421245421246, "grad_norm": 0.82421875, "learning_rate": 9.480051636217151e-05, "loss": 0.1345, "step": 9071 }, { "epoch": 6.6461538461538465, "grad_norm": 0.94140625, "learning_rate": 9.479898759680129e-05, "loss": 0.1633, "step": 9072 }, { "epoch": 6.646886446886447, "grad_norm": 0.87109375, "learning_rate": 9.479745861918007e-05, "loss": 0.1291, "step": 9073 }, { "epoch": 6.647619047619048, "grad_norm": 1.046875, "learning_rate": 9.47959294293152e-05, "loss": 0.1259, "step": 9074 }, { "epoch": 6.648351648351649, "grad_norm": 0.94921875, "learning_rate": 9.479440002721398e-05, "loss": 0.1308, "step": 9075 }, { "epoch": 6.649084249084249, "grad_norm": 0.58203125, "learning_rate": 9.479287041288376e-05, "loss": 0.1164, "step": 9076 }, { "epoch": 6.64981684981685, "grad_norm": 1.1640625, "learning_rate": 9.479134058633185e-05, "loss": 0.1339, "step": 9077 }, { "epoch": 6.650549450549451, "grad_norm": 0.69921875, "learning_rate": 9.478981054756559e-05, "loss": 0.1507, "step": 9078 }, { "epoch": 6.651282051282052, "grad_norm": 1.03125, "learning_rate": 9.478828029659231e-05, "loss": 0.1622, "step": 9079 }, { "epoch": 6.652014652014652, "grad_norm": 0.70703125, "learning_rate": 9.478674983341932e-05, "loss": 0.1314, "step": 9080 }, { "epoch": 6.652747252747253, "grad_norm": 0.93359375, "learning_rate": 9.4785219158054e-05, "loss": 0.1596, "step": 9081 }, { "epoch": 6.653479853479854, "grad_norm": 1.1171875, "learning_rate": 9.478368827050365e-05, "loss": 0.1418, "step": 9082 }, { "epoch": 6.6542124542124546, "grad_norm": 0.90625, "learning_rate": 9.478215717077561e-05, "loss": 0.1312, "step": 9083 }, { "epoch": 6.654945054945055, "grad_norm": 1.3125, "learning_rate": 9.478062585887721e-05, "loss": 0.1466, "step": 9084 }, { "epoch": 6.655677655677656, "grad_norm": 1.21875, "learning_rate": 9.47790943348158e-05, "loss": 0.1393, "step": 9085 }, { "epoch": 6.656410256410257, "grad_norm": 0.5390625, "learning_rate": 9.477756259859871e-05, "loss": 0.136, "step": 9086 }, { "epoch": 6.6571428571428575, "grad_norm": 1.2109375, "learning_rate": 9.477603065023329e-05, "loss": 0.1262, "step": 9087 }, { "epoch": 6.657875457875458, "grad_norm": 1.671875, "learning_rate": 9.477449848972687e-05, "loss": 0.135, "step": 9088 }, { "epoch": 6.658608058608059, "grad_norm": 1.7734375, "learning_rate": 9.47729661170868e-05, "loss": 0.1492, "step": 9089 }, { "epoch": 6.65934065934066, "grad_norm": 0.7890625, "learning_rate": 9.47714335323204e-05, "loss": 0.1692, "step": 9090 }, { "epoch": 6.66007326007326, "grad_norm": 0.9609375, "learning_rate": 9.476990073543504e-05, "loss": 0.1875, "step": 9091 }, { "epoch": 6.660805860805861, "grad_norm": 0.74609375, "learning_rate": 9.476836772643804e-05, "loss": 0.1415, "step": 9092 }, { "epoch": 6.661538461538462, "grad_norm": 1.109375, "learning_rate": 9.476683450533676e-05, "loss": 0.1239, "step": 9093 }, { "epoch": 6.662271062271063, "grad_norm": 1.1875, "learning_rate": 9.476530107213855e-05, "loss": 0.1286, "step": 9094 }, { "epoch": 6.663003663003663, "grad_norm": 1.015625, "learning_rate": 9.476376742685077e-05, "loss": 0.1939, "step": 9095 }, { "epoch": 6.663736263736264, "grad_norm": 0.9921875, "learning_rate": 9.476223356948072e-05, "loss": 0.1329, "step": 9096 }, { "epoch": 6.664468864468865, "grad_norm": 1.015625, "learning_rate": 9.47606995000358e-05, "loss": 0.1573, "step": 9097 }, { "epoch": 6.6652014652014655, "grad_norm": 0.7890625, "learning_rate": 9.475916521852333e-05, "loss": 0.1203, "step": 9098 }, { "epoch": 6.665934065934066, "grad_norm": 1.140625, "learning_rate": 9.475763072495066e-05, "loss": 0.164, "step": 9099 }, { "epoch": 6.666666666666667, "grad_norm": 1.109375, "learning_rate": 9.475609601932517e-05, "loss": 0.112, "step": 9100 }, { "epoch": 6.667399267399268, "grad_norm": 0.953125, "learning_rate": 9.475456110165419e-05, "loss": 0.1348, "step": 9101 }, { "epoch": 6.668131868131868, "grad_norm": 1.0859375, "learning_rate": 9.475302597194507e-05, "loss": 0.1657, "step": 9102 }, { "epoch": 6.668864468864469, "grad_norm": 1.0703125, "learning_rate": 9.475149063020519e-05, "loss": 0.1439, "step": 9103 }, { "epoch": 6.66959706959707, "grad_norm": 1.375, "learning_rate": 9.474995507644188e-05, "loss": 0.1281, "step": 9104 }, { "epoch": 6.670329670329671, "grad_norm": 0.68359375, "learning_rate": 9.474841931066251e-05, "loss": 0.1227, "step": 9105 }, { "epoch": 6.671062271062271, "grad_norm": 0.9375, "learning_rate": 9.474688333287443e-05, "loss": 0.1211, "step": 9106 }, { "epoch": 6.671794871794872, "grad_norm": 0.9765625, "learning_rate": 9.474534714308504e-05, "loss": 0.1543, "step": 9107 }, { "epoch": 6.672527472527473, "grad_norm": 0.82421875, "learning_rate": 9.474381074130163e-05, "loss": 0.1437, "step": 9108 }, { "epoch": 6.6732600732600735, "grad_norm": 1.4296875, "learning_rate": 9.474227412753161e-05, "loss": 0.185, "step": 9109 }, { "epoch": 6.673992673992674, "grad_norm": 1.0703125, "learning_rate": 9.474073730178235e-05, "loss": 0.1483, "step": 9110 }, { "epoch": 6.674725274725275, "grad_norm": 0.68359375, "learning_rate": 9.473920026406117e-05, "loss": 0.1465, "step": 9111 }, { "epoch": 6.675457875457876, "grad_norm": 1.0, "learning_rate": 9.473766301437547e-05, "loss": 0.1402, "step": 9112 }, { "epoch": 6.6761904761904765, "grad_norm": 0.83203125, "learning_rate": 9.47361255527326e-05, "loss": 0.1348, "step": 9113 }, { "epoch": 6.676923076923077, "grad_norm": 1.234375, "learning_rate": 9.473458787913992e-05, "loss": 0.1496, "step": 9114 }, { "epoch": 6.677655677655678, "grad_norm": 0.65234375, "learning_rate": 9.473304999360483e-05, "loss": 0.119, "step": 9115 }, { "epoch": 6.678388278388279, "grad_norm": 0.80078125, "learning_rate": 9.473151189613466e-05, "loss": 0.143, "step": 9116 }, { "epoch": 6.679120879120879, "grad_norm": 0.9765625, "learning_rate": 9.47299735867368e-05, "loss": 0.1242, "step": 9117 }, { "epoch": 6.67985347985348, "grad_norm": 0.93359375, "learning_rate": 9.472843506541863e-05, "loss": 0.1577, "step": 9118 }, { "epoch": 6.680586080586081, "grad_norm": 1.5234375, "learning_rate": 9.47268963321875e-05, "loss": 0.1564, "step": 9119 }, { "epoch": 6.681318681318682, "grad_norm": 1.2890625, "learning_rate": 9.472535738705078e-05, "loss": 0.1721, "step": 9120 }, { "epoch": 6.682051282051282, "grad_norm": 0.86328125, "learning_rate": 9.472381823001585e-05, "loss": 0.1391, "step": 9121 }, { "epoch": 6.682783882783883, "grad_norm": 0.79296875, "learning_rate": 9.472227886109011e-05, "loss": 0.1409, "step": 9122 }, { "epoch": 6.683516483516484, "grad_norm": 1.0859375, "learning_rate": 9.472073928028091e-05, "loss": 0.1392, "step": 9123 }, { "epoch": 6.6842490842490845, "grad_norm": 1.3515625, "learning_rate": 9.471919948759563e-05, "loss": 0.112, "step": 9124 }, { "epoch": 6.684981684981685, "grad_norm": 1.09375, "learning_rate": 9.471765948304166e-05, "loss": 0.1303, "step": 9125 }, { "epoch": 6.685714285714286, "grad_norm": 1.0390625, "learning_rate": 9.471611926662635e-05, "loss": 0.1422, "step": 9126 }, { "epoch": 6.686446886446887, "grad_norm": 0.90625, "learning_rate": 9.47145788383571e-05, "loss": 0.1287, "step": 9127 }, { "epoch": 6.687179487179487, "grad_norm": 0.88671875, "learning_rate": 9.47130381982413e-05, "loss": 0.1469, "step": 9128 }, { "epoch": 6.687912087912088, "grad_norm": 0.82421875, "learning_rate": 9.471149734628629e-05, "loss": 0.124, "step": 9129 }, { "epoch": 6.688644688644689, "grad_norm": 0.86328125, "learning_rate": 9.470995628249951e-05, "loss": 0.1396, "step": 9130 }, { "epoch": 6.68937728937729, "grad_norm": 0.93359375, "learning_rate": 9.470841500688831e-05, "loss": 0.1645, "step": 9131 }, { "epoch": 6.69010989010989, "grad_norm": 0.8671875, "learning_rate": 9.470687351946008e-05, "loss": 0.1505, "step": 9132 }, { "epoch": 6.690842490842491, "grad_norm": 1.3125, "learning_rate": 9.470533182022221e-05, "loss": 0.1451, "step": 9133 }, { "epoch": 6.691575091575092, "grad_norm": 1.0703125, "learning_rate": 9.470378990918207e-05, "loss": 0.1462, "step": 9134 }, { "epoch": 6.6923076923076925, "grad_norm": 0.6953125, "learning_rate": 9.470224778634708e-05, "loss": 0.1006, "step": 9135 }, { "epoch": 6.693040293040293, "grad_norm": 1.0546875, "learning_rate": 9.470070545172461e-05, "loss": 0.158, "step": 9136 }, { "epoch": 6.693772893772894, "grad_norm": 1.0859375, "learning_rate": 9.469916290532204e-05, "loss": 0.1318, "step": 9137 }, { "epoch": 6.694505494505495, "grad_norm": 0.6328125, "learning_rate": 9.469762014714676e-05, "loss": 0.1356, "step": 9138 }, { "epoch": 6.695238095238095, "grad_norm": 1.0703125, "learning_rate": 9.469607717720618e-05, "loss": 0.1526, "step": 9139 }, { "epoch": 6.695970695970696, "grad_norm": 0.7734375, "learning_rate": 9.469453399550771e-05, "loss": 0.1456, "step": 9140 }, { "epoch": 6.696703296703297, "grad_norm": 1.65625, "learning_rate": 9.46929906020587e-05, "loss": 0.1387, "step": 9141 }, { "epoch": 6.697435897435898, "grad_norm": 0.90625, "learning_rate": 9.469144699686656e-05, "loss": 0.13, "step": 9142 }, { "epoch": 6.698168498168498, "grad_norm": 0.91796875, "learning_rate": 9.46899031799387e-05, "loss": 0.1708, "step": 9143 }, { "epoch": 6.698901098901099, "grad_norm": 0.84765625, "learning_rate": 9.46883591512825e-05, "loss": 0.1557, "step": 9144 }, { "epoch": 6.6996336996337, "grad_norm": 1.1640625, "learning_rate": 9.468681491090537e-05, "loss": 0.1988, "step": 9145 }, { "epoch": 6.7003663003663005, "grad_norm": 1.109375, "learning_rate": 9.46852704588147e-05, "loss": 0.1998, "step": 9146 }, { "epoch": 6.701098901098901, "grad_norm": 1.2890625, "learning_rate": 9.468372579501789e-05, "loss": 0.1534, "step": 9147 }, { "epoch": 6.701831501831502, "grad_norm": 0.92578125, "learning_rate": 9.468218091952236e-05, "loss": 0.1421, "step": 9148 }, { "epoch": 6.702564102564103, "grad_norm": 0.94140625, "learning_rate": 9.46806358323355e-05, "loss": 0.1352, "step": 9149 }, { "epoch": 6.7032967032967035, "grad_norm": 1.1328125, "learning_rate": 9.467909053346467e-05, "loss": 0.1525, "step": 9150 }, { "epoch": 6.704029304029304, "grad_norm": 1.1640625, "learning_rate": 9.467754502291736e-05, "loss": 0.1548, "step": 9151 }, { "epoch": 6.704761904761905, "grad_norm": 0.890625, "learning_rate": 9.467599930070091e-05, "loss": 0.1351, "step": 9152 }, { "epoch": 6.705494505494506, "grad_norm": 0.72265625, "learning_rate": 9.467445336682276e-05, "loss": 0.1207, "step": 9153 }, { "epoch": 6.706227106227106, "grad_norm": 0.984375, "learning_rate": 9.467290722129027e-05, "loss": 0.1527, "step": 9154 }, { "epoch": 6.706959706959707, "grad_norm": 0.86328125, "learning_rate": 9.467136086411092e-05, "loss": 0.1376, "step": 9155 }, { "epoch": 6.707692307692308, "grad_norm": 1.3359375, "learning_rate": 9.466981429529204e-05, "loss": 0.1571, "step": 9156 }, { "epoch": 6.708424908424909, "grad_norm": 0.78515625, "learning_rate": 9.46682675148411e-05, "loss": 0.143, "step": 9157 }, { "epoch": 6.709157509157509, "grad_norm": 0.8203125, "learning_rate": 9.46667205227655e-05, "loss": 0.1254, "step": 9158 }, { "epoch": 6.70989010989011, "grad_norm": 1.109375, "learning_rate": 9.466517331907264e-05, "loss": 0.1447, "step": 9159 }, { "epoch": 6.710622710622711, "grad_norm": 1.0390625, "learning_rate": 9.466362590376992e-05, "loss": 0.1484, "step": 9160 }, { "epoch": 6.7113553113553115, "grad_norm": 1.015625, "learning_rate": 9.466207827686479e-05, "loss": 0.1514, "step": 9161 }, { "epoch": 6.712087912087912, "grad_norm": 1.546875, "learning_rate": 9.466053043836462e-05, "loss": 0.1615, "step": 9162 }, { "epoch": 6.712820512820513, "grad_norm": 1.546875, "learning_rate": 9.465898238827689e-05, "loss": 0.1415, "step": 9163 }, { "epoch": 6.713553113553114, "grad_norm": 1.28125, "learning_rate": 9.465743412660894e-05, "loss": 0.1356, "step": 9164 }, { "epoch": 6.714285714285714, "grad_norm": 0.68359375, "learning_rate": 9.465588565336825e-05, "loss": 0.1162, "step": 9165 }, { "epoch": 6.715018315018315, "grad_norm": 0.98046875, "learning_rate": 9.465433696856222e-05, "loss": 0.1532, "step": 9166 }, { "epoch": 6.715750915750916, "grad_norm": 1.2265625, "learning_rate": 9.465278807219824e-05, "loss": 0.1607, "step": 9167 }, { "epoch": 6.716483516483517, "grad_norm": 0.84375, "learning_rate": 9.465123896428378e-05, "loss": 0.1208, "step": 9168 }, { "epoch": 6.717216117216117, "grad_norm": 1.2578125, "learning_rate": 9.464968964482624e-05, "loss": 0.1729, "step": 9169 }, { "epoch": 6.717948717948718, "grad_norm": 1.25, "learning_rate": 9.464814011383304e-05, "loss": 0.1888, "step": 9170 }, { "epoch": 6.718681318681319, "grad_norm": 1.3203125, "learning_rate": 9.46465903713116e-05, "loss": 0.1652, "step": 9171 }, { "epoch": 6.7194139194139195, "grad_norm": 1.5625, "learning_rate": 9.464504041726936e-05, "loss": 0.1364, "step": 9172 }, { "epoch": 6.72014652014652, "grad_norm": 1.0546875, "learning_rate": 9.464349025171373e-05, "loss": 0.1227, "step": 9173 }, { "epoch": 6.720879120879121, "grad_norm": 0.921875, "learning_rate": 9.464193987465216e-05, "loss": 0.1398, "step": 9174 }, { "epoch": 6.721611721611722, "grad_norm": 1.265625, "learning_rate": 9.464038928609208e-05, "loss": 0.1206, "step": 9175 }, { "epoch": 6.722344322344322, "grad_norm": 1.0703125, "learning_rate": 9.463883848604088e-05, "loss": 0.1527, "step": 9176 }, { "epoch": 6.723076923076923, "grad_norm": 0.84375, "learning_rate": 9.463728747450601e-05, "loss": 0.1423, "step": 9177 }, { "epoch": 6.723809523809524, "grad_norm": 1.25, "learning_rate": 9.463573625149493e-05, "loss": 0.1651, "step": 9178 }, { "epoch": 6.724542124542125, "grad_norm": 0.77734375, "learning_rate": 9.463418481701502e-05, "loss": 0.1371, "step": 9179 }, { "epoch": 6.725274725274725, "grad_norm": 1.2890625, "learning_rate": 9.463263317107377e-05, "loss": 0.1204, "step": 9180 }, { "epoch": 6.726007326007326, "grad_norm": 1.15625, "learning_rate": 9.463108131367856e-05, "loss": 0.1533, "step": 9181 }, { "epoch": 6.726739926739927, "grad_norm": 1.1953125, "learning_rate": 9.462952924483688e-05, "loss": 0.1238, "step": 9182 }, { "epoch": 6.7274725274725276, "grad_norm": 1.0078125, "learning_rate": 9.462797696455612e-05, "loss": 0.1397, "step": 9183 }, { "epoch": 6.728205128205128, "grad_norm": 1.4921875, "learning_rate": 9.462642447284373e-05, "loss": 0.1405, "step": 9184 }, { "epoch": 6.728937728937729, "grad_norm": 0.91015625, "learning_rate": 9.462487176970718e-05, "loss": 0.1243, "step": 9185 }, { "epoch": 6.72967032967033, "grad_norm": 0.9296875, "learning_rate": 9.462331885515386e-05, "loss": 0.111, "step": 9186 }, { "epoch": 6.7304029304029305, "grad_norm": 0.6640625, "learning_rate": 9.462176572919125e-05, "loss": 0.1316, "step": 9187 }, { "epoch": 6.731135531135531, "grad_norm": 1.6875, "learning_rate": 9.462021239182678e-05, "loss": 0.1732, "step": 9188 }, { "epoch": 6.731868131868132, "grad_norm": 1.125, "learning_rate": 9.461865884306788e-05, "loss": 0.1415, "step": 9189 }, { "epoch": 6.732600732600733, "grad_norm": 1.078125, "learning_rate": 9.4617105082922e-05, "loss": 0.1476, "step": 9190 }, { "epoch": 6.733333333333333, "grad_norm": 1.2578125, "learning_rate": 9.461555111139658e-05, "loss": 0.1227, "step": 9191 }, { "epoch": 6.734065934065934, "grad_norm": 0.671875, "learning_rate": 9.461399692849909e-05, "loss": 0.1251, "step": 9192 }, { "epoch": 6.734798534798535, "grad_norm": 1.3515625, "learning_rate": 9.461244253423694e-05, "loss": 0.1862, "step": 9193 }, { "epoch": 6.735531135531136, "grad_norm": 1.3203125, "learning_rate": 9.461088792861761e-05, "loss": 0.1227, "step": 9194 }, { "epoch": 6.736263736263736, "grad_norm": 1.6484375, "learning_rate": 9.460933311164854e-05, "loss": 0.163, "step": 9195 }, { "epoch": 6.736996336996337, "grad_norm": 1.484375, "learning_rate": 9.460777808333716e-05, "loss": 0.1753, "step": 9196 }, { "epoch": 6.737728937728938, "grad_norm": 1.1640625, "learning_rate": 9.460622284369095e-05, "loss": 0.1312, "step": 9197 }, { "epoch": 6.7384615384615385, "grad_norm": 0.91796875, "learning_rate": 9.460466739271733e-05, "loss": 0.1656, "step": 9198 }, { "epoch": 6.739194139194139, "grad_norm": 1.7578125, "learning_rate": 9.460311173042378e-05, "loss": 0.1383, "step": 9199 }, { "epoch": 6.73992673992674, "grad_norm": 1.8984375, "learning_rate": 9.460155585681775e-05, "loss": 0.1464, "step": 9200 }, { "epoch": 6.740659340659341, "grad_norm": 0.98046875, "learning_rate": 9.45999997719067e-05, "loss": 0.1334, "step": 9201 }, { "epoch": 6.741391941391941, "grad_norm": 1.546875, "learning_rate": 9.459844347569802e-05, "loss": 0.1542, "step": 9202 }, { "epoch": 6.742124542124542, "grad_norm": 0.734375, "learning_rate": 9.459688696819927e-05, "loss": 0.1505, "step": 9203 }, { "epoch": 6.742857142857143, "grad_norm": 0.89453125, "learning_rate": 9.459533024941784e-05, "loss": 0.1505, "step": 9204 }, { "epoch": 6.743589743589744, "grad_norm": 1.265625, "learning_rate": 9.459377331936122e-05, "loss": 0.1589, "step": 9205 }, { "epoch": 6.744322344322344, "grad_norm": 1.140625, "learning_rate": 9.459221617803683e-05, "loss": 0.1398, "step": 9206 }, { "epoch": 6.745054945054945, "grad_norm": 0.984375, "learning_rate": 9.459065882545219e-05, "loss": 0.1522, "step": 9207 }, { "epoch": 6.745787545787546, "grad_norm": 0.7421875, "learning_rate": 9.458910126161472e-05, "loss": 0.0991, "step": 9208 }, { "epoch": 6.7465201465201465, "grad_norm": 0.66015625, "learning_rate": 9.458754348653188e-05, "loss": 0.1319, "step": 9209 }, { "epoch": 6.747252747252747, "grad_norm": 1.125, "learning_rate": 9.458598550021116e-05, "loss": 0.1565, "step": 9210 }, { "epoch": 6.747985347985348, "grad_norm": 1.046875, "learning_rate": 9.458442730266e-05, "loss": 0.1446, "step": 9211 }, { "epoch": 6.748717948717949, "grad_norm": 0.7265625, "learning_rate": 9.458286889388589e-05, "loss": 0.109, "step": 9212 }, { "epoch": 6.7494505494505495, "grad_norm": 1.828125, "learning_rate": 9.458131027389628e-05, "loss": 0.1761, "step": 9213 }, { "epoch": 6.75018315018315, "grad_norm": 1.0390625, "learning_rate": 9.457975144269862e-05, "loss": 0.1514, "step": 9214 }, { "epoch": 6.750915750915751, "grad_norm": 1.234375, "learning_rate": 9.457819240030043e-05, "loss": 0.1221, "step": 9215 }, { "epoch": 6.751648351648352, "grad_norm": 0.78515625, "learning_rate": 9.457663314670914e-05, "loss": 0.1341, "step": 9216 }, { "epoch": 6.752380952380952, "grad_norm": 1.1796875, "learning_rate": 9.457507368193224e-05, "loss": 0.145, "step": 9217 }, { "epoch": 6.753113553113553, "grad_norm": 1.203125, "learning_rate": 9.457351400597718e-05, "loss": 0.1848, "step": 9218 }, { "epoch": 6.753846153846154, "grad_norm": 1.3515625, "learning_rate": 9.457195411885144e-05, "loss": 0.1756, "step": 9219 }, { "epoch": 6.754578754578755, "grad_norm": 1.171875, "learning_rate": 9.45703940205625e-05, "loss": 0.1705, "step": 9220 }, { "epoch": 6.755311355311355, "grad_norm": 1.40625, "learning_rate": 9.456883371111786e-05, "loss": 0.1334, "step": 9221 }, { "epoch": 6.756043956043956, "grad_norm": 0.765625, "learning_rate": 9.456727319052495e-05, "loss": 0.1394, "step": 9222 }, { "epoch": 6.756776556776557, "grad_norm": 1.15625, "learning_rate": 9.456571245879128e-05, "loss": 0.1196, "step": 9223 }, { "epoch": 6.7575091575091575, "grad_norm": 1.1328125, "learning_rate": 9.456415151592429e-05, "loss": 0.1469, "step": 9224 }, { "epoch": 6.758241758241758, "grad_norm": 0.91796875, "learning_rate": 9.456259036193151e-05, "loss": 0.1414, "step": 9225 }, { "epoch": 6.758974358974359, "grad_norm": 1.03125, "learning_rate": 9.45610289968204e-05, "loss": 0.1426, "step": 9226 }, { "epoch": 6.75970695970696, "grad_norm": 0.5546875, "learning_rate": 9.455946742059843e-05, "loss": 0.107, "step": 9227 }, { "epoch": 6.76043956043956, "grad_norm": 1.3671875, "learning_rate": 9.455790563327307e-05, "loss": 0.1447, "step": 9228 }, { "epoch": 6.761172161172161, "grad_norm": 0.7890625, "learning_rate": 9.455634363485182e-05, "loss": 0.1236, "step": 9229 }, { "epoch": 6.761904761904762, "grad_norm": 1.2734375, "learning_rate": 9.455478142534219e-05, "loss": 0.1484, "step": 9230 }, { "epoch": 6.762637362637363, "grad_norm": 0.75, "learning_rate": 9.455321900475163e-05, "loss": 0.1551, "step": 9231 }, { "epoch": 6.763369963369963, "grad_norm": 1.1484375, "learning_rate": 9.455165637308763e-05, "loss": 0.1337, "step": 9232 }, { "epoch": 6.764102564102564, "grad_norm": 1.15625, "learning_rate": 9.455009353035767e-05, "loss": 0.1595, "step": 9233 }, { "epoch": 6.764835164835165, "grad_norm": 1.2109375, "learning_rate": 9.454853047656929e-05, "loss": 0.158, "step": 9234 }, { "epoch": 6.7655677655677655, "grad_norm": 1.1015625, "learning_rate": 9.454696721172991e-05, "loss": 0.185, "step": 9235 }, { "epoch": 6.766300366300366, "grad_norm": 0.53515625, "learning_rate": 9.454540373584707e-05, "loss": 0.136, "step": 9236 }, { "epoch": 6.767032967032967, "grad_norm": 1.015625, "learning_rate": 9.454384004892821e-05, "loss": 0.1195, "step": 9237 }, { "epoch": 6.767765567765568, "grad_norm": 0.89453125, "learning_rate": 9.454227615098088e-05, "loss": 0.116, "step": 9238 }, { "epoch": 6.768498168498168, "grad_norm": 1.21875, "learning_rate": 9.454071204201254e-05, "loss": 0.1483, "step": 9239 }, { "epoch": 6.769230769230769, "grad_norm": 0.96484375, "learning_rate": 9.453914772203069e-05, "loss": 0.147, "step": 9240 }, { "epoch": 6.76996336996337, "grad_norm": 1.609375, "learning_rate": 9.453758319104283e-05, "loss": 0.1286, "step": 9241 }, { "epoch": 6.770695970695971, "grad_norm": 0.75390625, "learning_rate": 9.453601844905644e-05, "loss": 0.1082, "step": 9242 }, { "epoch": 6.771428571428571, "grad_norm": 1.0234375, "learning_rate": 9.453445349607904e-05, "loss": 0.1708, "step": 9243 }, { "epoch": 6.772161172161172, "grad_norm": 1.46875, "learning_rate": 9.453288833211812e-05, "loss": 0.1511, "step": 9244 }, { "epoch": 6.772893772893773, "grad_norm": 1.484375, "learning_rate": 9.453132295718117e-05, "loss": 0.1425, "step": 9245 }, { "epoch": 6.7736263736263735, "grad_norm": 1.046875, "learning_rate": 9.452975737127572e-05, "loss": 0.1323, "step": 9246 }, { "epoch": 6.774358974358974, "grad_norm": 1.1328125, "learning_rate": 9.452819157440922e-05, "loss": 0.154, "step": 9247 }, { "epoch": 6.775091575091575, "grad_norm": 1.2734375, "learning_rate": 9.45266255665892e-05, "loss": 0.1596, "step": 9248 }, { "epoch": 6.775824175824176, "grad_norm": 1.3359375, "learning_rate": 9.452505934782319e-05, "loss": 0.1424, "step": 9249 }, { "epoch": 6.7765567765567765, "grad_norm": 1.5390625, "learning_rate": 9.452349291811864e-05, "loss": 0.1427, "step": 9250 }, { "epoch": 6.777289377289377, "grad_norm": 1.5078125, "learning_rate": 9.45219262774831e-05, "loss": 0.1785, "step": 9251 }, { "epoch": 6.778021978021978, "grad_norm": 1.4453125, "learning_rate": 9.452035942592405e-05, "loss": 0.1264, "step": 9252 }, { "epoch": 6.778754578754579, "grad_norm": 1.03125, "learning_rate": 9.451879236344902e-05, "loss": 0.1407, "step": 9253 }, { "epoch": 6.779487179487179, "grad_norm": 0.6796875, "learning_rate": 9.45172250900655e-05, "loss": 0.1176, "step": 9254 }, { "epoch": 6.78021978021978, "grad_norm": 1.265625, "learning_rate": 9.4515657605781e-05, "loss": 0.124, "step": 9255 }, { "epoch": 6.780952380952381, "grad_norm": 1.0078125, "learning_rate": 9.451408991060303e-05, "loss": 0.1082, "step": 9256 }, { "epoch": 6.781684981684982, "grad_norm": 0.734375, "learning_rate": 9.451252200453911e-05, "loss": 0.1603, "step": 9257 }, { "epoch": 6.782417582417582, "grad_norm": 0.83984375, "learning_rate": 9.451095388759676e-05, "loss": 0.122, "step": 9258 }, { "epoch": 6.783150183150183, "grad_norm": 1.0703125, "learning_rate": 9.450938555978348e-05, "loss": 0.1352, "step": 9259 }, { "epoch": 6.783882783882784, "grad_norm": 1.4453125, "learning_rate": 9.450781702110677e-05, "loss": 0.1514, "step": 9260 }, { "epoch": 6.7846153846153845, "grad_norm": 2.046875, "learning_rate": 9.450624827157418e-05, "loss": 0.1981, "step": 9261 }, { "epoch": 6.785347985347985, "grad_norm": 1.3828125, "learning_rate": 9.45046793111932e-05, "loss": 0.1585, "step": 9262 }, { "epoch": 6.786080586080586, "grad_norm": 1.0703125, "learning_rate": 9.450311013997135e-05, "loss": 0.1134, "step": 9263 }, { "epoch": 6.786813186813187, "grad_norm": 1.46875, "learning_rate": 9.450154075791616e-05, "loss": 0.1757, "step": 9264 }, { "epoch": 6.787545787545787, "grad_norm": 1.375, "learning_rate": 9.449997116503515e-05, "loss": 0.1432, "step": 9265 }, { "epoch": 6.788278388278388, "grad_norm": 1.1484375, "learning_rate": 9.449840136133582e-05, "loss": 0.1614, "step": 9266 }, { "epoch": 6.789010989010989, "grad_norm": 0.90625, "learning_rate": 9.449683134682573e-05, "loss": 0.1268, "step": 9267 }, { "epoch": 6.78974358974359, "grad_norm": 0.953125, "learning_rate": 9.449526112151236e-05, "loss": 0.171, "step": 9268 }, { "epoch": 6.79047619047619, "grad_norm": 1.109375, "learning_rate": 9.449369068540324e-05, "loss": 0.1531, "step": 9269 }, { "epoch": 6.791208791208791, "grad_norm": 1.6015625, "learning_rate": 9.449212003850593e-05, "loss": 0.1416, "step": 9270 }, { "epoch": 6.791941391941392, "grad_norm": 1.2265625, "learning_rate": 9.449054918082791e-05, "loss": 0.1126, "step": 9271 }, { "epoch": 6.7926739926739925, "grad_norm": 1.84375, "learning_rate": 9.448897811237674e-05, "loss": 0.1732, "step": 9272 }, { "epoch": 6.793406593406593, "grad_norm": 0.8671875, "learning_rate": 9.448740683315994e-05, "loss": 0.1467, "step": 9273 }, { "epoch": 6.794139194139194, "grad_norm": 1.78125, "learning_rate": 9.448583534318502e-05, "loss": 0.1351, "step": 9274 }, { "epoch": 6.794871794871795, "grad_norm": 1.3671875, "learning_rate": 9.448426364245953e-05, "loss": 0.1374, "step": 9275 }, { "epoch": 6.795604395604395, "grad_norm": 1.3359375, "learning_rate": 9.448269173099099e-05, "loss": 0.1617, "step": 9276 }, { "epoch": 6.796336996336996, "grad_norm": 0.83984375, "learning_rate": 9.448111960878694e-05, "loss": 0.1465, "step": 9277 }, { "epoch": 6.797069597069597, "grad_norm": 0.78515625, "learning_rate": 9.447954727585491e-05, "loss": 0.1307, "step": 9278 }, { "epoch": 6.797802197802198, "grad_norm": 1.109375, "learning_rate": 9.447797473220242e-05, "loss": 0.145, "step": 9279 }, { "epoch": 6.798534798534798, "grad_norm": 1.265625, "learning_rate": 9.447640197783704e-05, "loss": 0.1473, "step": 9280 }, { "epoch": 6.799267399267399, "grad_norm": 0.85546875, "learning_rate": 9.447482901276625e-05, "loss": 0.1173, "step": 9281 }, { "epoch": 6.8, "grad_norm": 1.3125, "learning_rate": 9.447325583699764e-05, "loss": 0.1264, "step": 9282 }, { "epoch": 6.8007326007326006, "grad_norm": 0.875, "learning_rate": 9.447168245053872e-05, "loss": 0.1843, "step": 9283 }, { "epoch": 6.801465201465201, "grad_norm": 0.6171875, "learning_rate": 9.447010885339705e-05, "loss": 0.1563, "step": 9284 }, { "epoch": 6.802197802197802, "grad_norm": 0.953125, "learning_rate": 9.446853504558012e-05, "loss": 0.135, "step": 9285 }, { "epoch": 6.802930402930403, "grad_norm": 1.1953125, "learning_rate": 9.446696102709552e-05, "loss": 0.1704, "step": 9286 }, { "epoch": 6.8036630036630035, "grad_norm": 1.109375, "learning_rate": 9.446538679795079e-05, "loss": 0.1492, "step": 9287 }, { "epoch": 6.804395604395604, "grad_norm": 1.0, "learning_rate": 9.446381235815345e-05, "loss": 0.1354, "step": 9288 }, { "epoch": 6.805128205128205, "grad_norm": 1.9921875, "learning_rate": 9.446223770771105e-05, "loss": 0.1614, "step": 9289 }, { "epoch": 6.805860805860806, "grad_norm": 1.640625, "learning_rate": 9.446066284663113e-05, "loss": 0.1341, "step": 9290 }, { "epoch": 6.806593406593406, "grad_norm": 0.8125, "learning_rate": 9.445908777492127e-05, "loss": 0.1197, "step": 9291 }, { "epoch": 6.807326007326007, "grad_norm": 0.81640625, "learning_rate": 9.445751249258896e-05, "loss": 0.1316, "step": 9292 }, { "epoch": 6.808058608058608, "grad_norm": 1.3046875, "learning_rate": 9.445593699964179e-05, "loss": 0.1367, "step": 9293 }, { "epoch": 6.808791208791209, "grad_norm": 0.82421875, "learning_rate": 9.445436129608729e-05, "loss": 0.1427, "step": 9294 }, { "epoch": 6.809523809523809, "grad_norm": 2.140625, "learning_rate": 9.445278538193301e-05, "loss": 0.1797, "step": 9295 }, { "epoch": 6.81025641025641, "grad_norm": 1.5234375, "learning_rate": 9.445120925718651e-05, "loss": 0.1531, "step": 9296 }, { "epoch": 6.810989010989011, "grad_norm": 0.76953125, "learning_rate": 9.444963292185536e-05, "loss": 0.1289, "step": 9297 }, { "epoch": 6.8117216117216115, "grad_norm": 0.9296875, "learning_rate": 9.444805637594707e-05, "loss": 0.0965, "step": 9298 }, { "epoch": 6.812454212454212, "grad_norm": 0.89453125, "learning_rate": 9.444647961946922e-05, "loss": 0.1096, "step": 9299 }, { "epoch": 6.813186813186813, "grad_norm": 0.9765625, "learning_rate": 9.444490265242937e-05, "loss": 0.1404, "step": 9300 }, { "epoch": 6.813919413919414, "grad_norm": 0.7578125, "learning_rate": 9.444332547483505e-05, "loss": 0.1313, "step": 9301 }, { "epoch": 6.814652014652014, "grad_norm": 0.78125, "learning_rate": 9.444174808669384e-05, "loss": 0.1403, "step": 9302 }, { "epoch": 6.815384615384615, "grad_norm": 1.1484375, "learning_rate": 9.444017048801328e-05, "loss": 0.1434, "step": 9303 }, { "epoch": 6.816117216117216, "grad_norm": 1.3125, "learning_rate": 9.443859267880094e-05, "loss": 0.1628, "step": 9304 }, { "epoch": 6.816849816849817, "grad_norm": 0.83984375, "learning_rate": 9.44370146590644e-05, "loss": 0.1528, "step": 9305 }, { "epoch": 6.817582417582417, "grad_norm": 1.0078125, "learning_rate": 9.443543642881118e-05, "loss": 0.1819, "step": 9306 }, { "epoch": 6.818315018315018, "grad_norm": 1.0, "learning_rate": 9.443385798804886e-05, "loss": 0.1362, "step": 9307 }, { "epoch": 6.819047619047619, "grad_norm": 1.8125, "learning_rate": 9.4432279336785e-05, "loss": 0.1777, "step": 9308 }, { "epoch": 6.8197802197802195, "grad_norm": 0.70703125, "learning_rate": 9.443070047502717e-05, "loss": 0.1182, "step": 9309 }, { "epoch": 6.82051282051282, "grad_norm": 1.4140625, "learning_rate": 9.442912140278294e-05, "loss": 0.1836, "step": 9310 }, { "epoch": 6.821245421245421, "grad_norm": 1.0234375, "learning_rate": 9.442754212005986e-05, "loss": 0.175, "step": 9311 }, { "epoch": 6.821978021978022, "grad_norm": 0.8125, "learning_rate": 9.442596262686553e-05, "loss": 0.1429, "step": 9312 }, { "epoch": 6.8227106227106225, "grad_norm": 1.078125, "learning_rate": 9.442438292320747e-05, "loss": 0.1864, "step": 9313 }, { "epoch": 6.823443223443223, "grad_norm": 1.1875, "learning_rate": 9.442280300909329e-05, "loss": 0.117, "step": 9314 }, { "epoch": 6.824175824175824, "grad_norm": 0.69140625, "learning_rate": 9.44212228845305e-05, "loss": 0.1475, "step": 9315 }, { "epoch": 6.824908424908425, "grad_norm": 1.3984375, "learning_rate": 9.441964254952675e-05, "loss": 0.1261, "step": 9316 }, { "epoch": 6.825641025641025, "grad_norm": 0.8515625, "learning_rate": 9.441806200408957e-05, "loss": 0.1226, "step": 9317 }, { "epoch": 6.826373626373626, "grad_norm": 0.95703125, "learning_rate": 9.441648124822653e-05, "loss": 0.1321, "step": 9318 }, { "epoch": 6.827106227106227, "grad_norm": 1.078125, "learning_rate": 9.441490028194522e-05, "loss": 0.1639, "step": 9319 }, { "epoch": 6.827838827838828, "grad_norm": 1.0390625, "learning_rate": 9.44133191052532e-05, "loss": 0.161, "step": 9320 }, { "epoch": 6.828571428571428, "grad_norm": 1.1953125, "learning_rate": 9.441173771815806e-05, "loss": 0.1037, "step": 9321 }, { "epoch": 6.829304029304029, "grad_norm": 1.328125, "learning_rate": 9.441015612066736e-05, "loss": 0.1794, "step": 9322 }, { "epoch": 6.83003663003663, "grad_norm": 0.80859375, "learning_rate": 9.440857431278869e-05, "loss": 0.1784, "step": 9323 }, { "epoch": 6.8307692307692305, "grad_norm": 1.0, "learning_rate": 9.440699229452963e-05, "loss": 0.1576, "step": 9324 }, { "epoch": 6.831501831501831, "grad_norm": 1.1484375, "learning_rate": 9.440541006589774e-05, "loss": 0.1272, "step": 9325 }, { "epoch": 6.832234432234432, "grad_norm": 1.1171875, "learning_rate": 9.440382762690063e-05, "loss": 0.1658, "step": 9326 }, { "epoch": 6.832967032967033, "grad_norm": 0.8828125, "learning_rate": 9.440224497754587e-05, "loss": 0.1291, "step": 9327 }, { "epoch": 6.833699633699633, "grad_norm": 1.7421875, "learning_rate": 9.440066211784105e-05, "loss": 0.1631, "step": 9328 }, { "epoch": 6.834432234432234, "grad_norm": 1.5703125, "learning_rate": 9.439907904779373e-05, "loss": 0.1522, "step": 9329 }, { "epoch": 6.835164835164835, "grad_norm": 1.640625, "learning_rate": 9.439749576741151e-05, "loss": 0.1624, "step": 9330 }, { "epoch": 6.835897435897436, "grad_norm": 0.84765625, "learning_rate": 9.4395912276702e-05, "loss": 0.141, "step": 9331 }, { "epoch": 6.836630036630036, "grad_norm": 0.80859375, "learning_rate": 9.439432857567274e-05, "loss": 0.1455, "step": 9332 }, { "epoch": 6.837362637362637, "grad_norm": 1.1171875, "learning_rate": 9.439274466433136e-05, "loss": 0.1767, "step": 9333 }, { "epoch": 6.838095238095238, "grad_norm": 1.1953125, "learning_rate": 9.439116054268543e-05, "loss": 0.1563, "step": 9334 }, { "epoch": 6.8388278388278385, "grad_norm": 1.1171875, "learning_rate": 9.438957621074251e-05, "loss": 0.1402, "step": 9335 }, { "epoch": 6.839560439560439, "grad_norm": 1.125, "learning_rate": 9.438799166851026e-05, "loss": 0.114, "step": 9336 }, { "epoch": 6.84029304029304, "grad_norm": 1.1171875, "learning_rate": 9.438640691599623e-05, "loss": 0.1498, "step": 9337 }, { "epoch": 6.841025641025641, "grad_norm": 1.9453125, "learning_rate": 9.4384821953208e-05, "loss": 0.1594, "step": 9338 }, { "epoch": 6.841758241758241, "grad_norm": 0.76953125, "learning_rate": 9.438323678015319e-05, "loss": 0.1333, "step": 9339 }, { "epoch": 6.842490842490842, "grad_norm": 1.0546875, "learning_rate": 9.438165139683939e-05, "loss": 0.1136, "step": 9340 }, { "epoch": 6.843223443223443, "grad_norm": 1.03125, "learning_rate": 9.438006580327418e-05, "loss": 0.1325, "step": 9341 }, { "epoch": 6.843956043956044, "grad_norm": 1.0234375, "learning_rate": 9.437847999946518e-05, "loss": 0.1467, "step": 9342 }, { "epoch": 6.844688644688644, "grad_norm": 0.7578125, "learning_rate": 9.437689398541997e-05, "loss": 0.1228, "step": 9343 }, { "epoch": 6.845421245421245, "grad_norm": 0.99609375, "learning_rate": 9.437530776114617e-05, "loss": 0.1589, "step": 9344 }, { "epoch": 6.846153846153846, "grad_norm": 1.296875, "learning_rate": 9.437372132665138e-05, "loss": 0.1286, "step": 9345 }, { "epoch": 6.8468864468864465, "grad_norm": 0.60546875, "learning_rate": 9.437213468194317e-05, "loss": 0.1335, "step": 9346 }, { "epoch": 6.847619047619047, "grad_norm": 1.3046875, "learning_rate": 9.437054782702915e-05, "loss": 0.1389, "step": 9347 }, { "epoch": 6.848351648351648, "grad_norm": 0.9609375, "learning_rate": 9.436896076191695e-05, "loss": 0.1756, "step": 9348 }, { "epoch": 6.849084249084249, "grad_norm": 0.9296875, "learning_rate": 9.436737348661415e-05, "loss": 0.1393, "step": 9349 }, { "epoch": 6.8498168498168495, "grad_norm": 1.0703125, "learning_rate": 9.436578600112838e-05, "loss": 0.1563, "step": 9350 }, { "epoch": 6.85054945054945, "grad_norm": 0.87890625, "learning_rate": 9.43641983054672e-05, "loss": 0.1481, "step": 9351 }, { "epoch": 6.851282051282051, "grad_norm": 1.296875, "learning_rate": 9.436261039963827e-05, "loss": 0.1665, "step": 9352 }, { "epoch": 6.852014652014652, "grad_norm": 0.71484375, "learning_rate": 9.436102228364916e-05, "loss": 0.1526, "step": 9353 }, { "epoch": 6.852747252747252, "grad_norm": 0.9609375, "learning_rate": 9.43594339575075e-05, "loss": 0.1475, "step": 9354 }, { "epoch": 6.853479853479853, "grad_norm": 0.83984375, "learning_rate": 9.435784542122088e-05, "loss": 0.1329, "step": 9355 }, { "epoch": 6.854212454212454, "grad_norm": 0.8828125, "learning_rate": 9.435625667479694e-05, "loss": 0.1259, "step": 9356 }, { "epoch": 6.854945054945055, "grad_norm": 0.7421875, "learning_rate": 9.435466771824327e-05, "loss": 0.1618, "step": 9357 }, { "epoch": 6.855677655677655, "grad_norm": 1.390625, "learning_rate": 9.43530785515675e-05, "loss": 0.18, "step": 9358 }, { "epoch": 6.856410256410256, "grad_norm": 1.1953125, "learning_rate": 9.435148917477723e-05, "loss": 0.1322, "step": 9359 }, { "epoch": 6.857142857142857, "grad_norm": 0.83984375, "learning_rate": 9.434989958788009e-05, "loss": 0.1263, "step": 9360 }, { "epoch": 6.8578754578754575, "grad_norm": 1.0, "learning_rate": 9.434830979088367e-05, "loss": 0.1813, "step": 9361 }, { "epoch": 6.858608058608058, "grad_norm": 0.92578125, "learning_rate": 9.43467197837956e-05, "loss": 0.1413, "step": 9362 }, { "epoch": 6.859340659340659, "grad_norm": 1.0859375, "learning_rate": 9.434512956662352e-05, "loss": 0.1551, "step": 9363 }, { "epoch": 6.860073260073261, "grad_norm": 0.80078125, "learning_rate": 9.434353913937503e-05, "loss": 0.1574, "step": 9364 }, { "epoch": 6.860805860805861, "grad_norm": 0.6484375, "learning_rate": 9.434194850205774e-05, "loss": 0.1493, "step": 9365 }, { "epoch": 6.861538461538462, "grad_norm": 0.75, "learning_rate": 9.434035765467926e-05, "loss": 0.133, "step": 9366 }, { "epoch": 6.862271062271063, "grad_norm": 1.15625, "learning_rate": 9.433876659724727e-05, "loss": 0.1706, "step": 9367 }, { "epoch": 6.8630036630036635, "grad_norm": 0.92578125, "learning_rate": 9.433717532976934e-05, "loss": 0.1252, "step": 9368 }, { "epoch": 6.863736263736264, "grad_norm": 0.94921875, "learning_rate": 9.433558385225314e-05, "loss": 0.1469, "step": 9369 }, { "epoch": 6.864468864468865, "grad_norm": 0.7734375, "learning_rate": 9.433399216470623e-05, "loss": 0.1501, "step": 9370 }, { "epoch": 6.865201465201466, "grad_norm": 0.859375, "learning_rate": 9.433240026713631e-05, "loss": 0.1403, "step": 9371 }, { "epoch": 6.865934065934066, "grad_norm": 1.1875, "learning_rate": 9.433080815955094e-05, "loss": 0.1512, "step": 9372 }, { "epoch": 6.866666666666667, "grad_norm": 0.984375, "learning_rate": 9.432921584195779e-05, "loss": 0.151, "step": 9373 }, { "epoch": 6.867399267399268, "grad_norm": 0.9140625, "learning_rate": 9.432762331436449e-05, "loss": 0.1257, "step": 9374 }, { "epoch": 6.868131868131869, "grad_norm": 0.9375, "learning_rate": 9.432603057677864e-05, "loss": 0.125, "step": 9375 }, { "epoch": 6.868864468864469, "grad_norm": 1.046875, "learning_rate": 9.43244376292079e-05, "loss": 0.1255, "step": 9376 }, { "epoch": 6.86959706959707, "grad_norm": 0.87109375, "learning_rate": 9.432284447165989e-05, "loss": 0.1507, "step": 9377 }, { "epoch": 6.870329670329671, "grad_norm": 1.046875, "learning_rate": 9.432125110414225e-05, "loss": 0.1548, "step": 9378 }, { "epoch": 6.8710622710622715, "grad_norm": 1.171875, "learning_rate": 9.431965752666261e-05, "loss": 0.1533, "step": 9379 }, { "epoch": 6.871794871794872, "grad_norm": 1.234375, "learning_rate": 9.431806373922859e-05, "loss": 0.1623, "step": 9380 }, { "epoch": 6.872527472527473, "grad_norm": 1.078125, "learning_rate": 9.431646974184787e-05, "loss": 0.1586, "step": 9381 }, { "epoch": 6.873260073260074, "grad_norm": 1.265625, "learning_rate": 9.431487553452803e-05, "loss": 0.1268, "step": 9382 }, { "epoch": 6.873992673992674, "grad_norm": 1.484375, "learning_rate": 9.431328111727677e-05, "loss": 0.1511, "step": 9383 }, { "epoch": 6.874725274725275, "grad_norm": 0.98828125, "learning_rate": 9.431168649010167e-05, "loss": 0.1535, "step": 9384 }, { "epoch": 6.875457875457876, "grad_norm": 1.203125, "learning_rate": 9.43100916530104e-05, "loss": 0.1407, "step": 9385 }, { "epoch": 6.876190476190477, "grad_norm": 1.4375, "learning_rate": 9.430849660601062e-05, "loss": 0.1429, "step": 9386 }, { "epoch": 6.876923076923077, "grad_norm": 0.828125, "learning_rate": 9.430690134910993e-05, "loss": 0.1463, "step": 9387 }, { "epoch": 6.877655677655678, "grad_norm": 0.9296875, "learning_rate": 9.430530588231602e-05, "loss": 0.1411, "step": 9388 }, { "epoch": 6.878388278388279, "grad_norm": 1.3515625, "learning_rate": 9.430371020563648e-05, "loss": 0.1304, "step": 9389 }, { "epoch": 6.8791208791208796, "grad_norm": 1.0859375, "learning_rate": 9.4302114319079e-05, "loss": 0.1701, "step": 9390 }, { "epoch": 6.87985347985348, "grad_norm": 1.1796875, "learning_rate": 9.430051822265121e-05, "loss": 0.1524, "step": 9391 }, { "epoch": 6.880586080586081, "grad_norm": 0.828125, "learning_rate": 9.429892191636075e-05, "loss": 0.1377, "step": 9392 }, { "epoch": 6.881318681318682, "grad_norm": 0.98046875, "learning_rate": 9.429732540021529e-05, "loss": 0.1678, "step": 9393 }, { "epoch": 6.8820512820512825, "grad_norm": 1.1953125, "learning_rate": 9.429572867422245e-05, "loss": 0.1555, "step": 9394 }, { "epoch": 6.882783882783883, "grad_norm": 1.3359375, "learning_rate": 9.429413173838992e-05, "loss": 0.1266, "step": 9395 }, { "epoch": 6.883516483516484, "grad_norm": 1.3359375, "learning_rate": 9.429253459272531e-05, "loss": 0.1269, "step": 9396 }, { "epoch": 6.884249084249085, "grad_norm": 0.828125, "learning_rate": 9.429093723723629e-05, "loss": 0.1465, "step": 9397 }, { "epoch": 6.884981684981685, "grad_norm": 1.171875, "learning_rate": 9.428933967193052e-05, "loss": 0.1646, "step": 9398 }, { "epoch": 6.885714285714286, "grad_norm": 1.0859375, "learning_rate": 9.428774189681564e-05, "loss": 0.122, "step": 9399 }, { "epoch": 6.886446886446887, "grad_norm": 2.15625, "learning_rate": 9.428614391189932e-05, "loss": 0.1835, "step": 9400 }, { "epoch": 6.887179487179488, "grad_norm": 1.21875, "learning_rate": 9.428454571718923e-05, "loss": 0.1366, "step": 9401 }, { "epoch": 6.887912087912088, "grad_norm": 0.890625, "learning_rate": 9.428294731269299e-05, "loss": 0.1473, "step": 9402 }, { "epoch": 6.888644688644689, "grad_norm": 1.78125, "learning_rate": 9.428134869841829e-05, "loss": 0.1836, "step": 9403 }, { "epoch": 6.88937728937729, "grad_norm": 1.53125, "learning_rate": 9.427974987437274e-05, "loss": 0.1356, "step": 9404 }, { "epoch": 6.8901098901098905, "grad_norm": 1.90625, "learning_rate": 9.427815084056407e-05, "loss": 0.1171, "step": 9405 }, { "epoch": 6.890842490842491, "grad_norm": 0.87890625, "learning_rate": 9.427655159699989e-05, "loss": 0.1382, "step": 9406 }, { "epoch": 6.891575091575092, "grad_norm": 1.078125, "learning_rate": 9.42749521436879e-05, "loss": 0.1118, "step": 9407 }, { "epoch": 6.892307692307693, "grad_norm": 0.88671875, "learning_rate": 9.427335248063573e-05, "loss": 0.1333, "step": 9408 }, { "epoch": 6.893040293040293, "grad_norm": 1.1875, "learning_rate": 9.427175260785105e-05, "loss": 0.1306, "step": 9409 }, { "epoch": 6.893772893772894, "grad_norm": 1.65625, "learning_rate": 9.427015252534155e-05, "loss": 0.1674, "step": 9410 }, { "epoch": 6.894505494505495, "grad_norm": 1.5390625, "learning_rate": 9.426855223311485e-05, "loss": 0.169, "step": 9411 }, { "epoch": 6.895238095238096, "grad_norm": 1.1953125, "learning_rate": 9.426695173117868e-05, "loss": 0.1356, "step": 9412 }, { "epoch": 6.895970695970696, "grad_norm": 1.7265625, "learning_rate": 9.426535101954067e-05, "loss": 0.1616, "step": 9413 }, { "epoch": 6.896703296703297, "grad_norm": 1.2734375, "learning_rate": 9.42637500982085e-05, "loss": 0.1847, "step": 9414 }, { "epoch": 6.897435897435898, "grad_norm": 1.578125, "learning_rate": 9.426214896718982e-05, "loss": 0.1674, "step": 9415 }, { "epoch": 6.8981684981684985, "grad_norm": 1.140625, "learning_rate": 9.426054762649234e-05, "loss": 0.146, "step": 9416 }, { "epoch": 6.898901098901099, "grad_norm": 1.3359375, "learning_rate": 9.425894607612368e-05, "loss": 0.1621, "step": 9417 }, { "epoch": 6.8996336996337, "grad_norm": 1.0078125, "learning_rate": 9.425734431609159e-05, "loss": 0.1455, "step": 9418 }, { "epoch": 6.900366300366301, "grad_norm": 0.94140625, "learning_rate": 9.425574234640367e-05, "loss": 0.1378, "step": 9419 }, { "epoch": 6.9010989010989015, "grad_norm": 1.59375, "learning_rate": 9.425414016706762e-05, "loss": 0.1758, "step": 9420 }, { "epoch": 6.901831501831502, "grad_norm": 1.5625, "learning_rate": 9.425253777809113e-05, "loss": 0.1275, "step": 9421 }, { "epoch": 6.902564102564103, "grad_norm": 2.296875, "learning_rate": 9.425093517948187e-05, "loss": 0.1523, "step": 9422 }, { "epoch": 6.903296703296704, "grad_norm": 1.796875, "learning_rate": 9.424933237124751e-05, "loss": 0.1388, "step": 9423 }, { "epoch": 6.904029304029304, "grad_norm": 1.03125, "learning_rate": 9.424772935339575e-05, "loss": 0.1166, "step": 9424 }, { "epoch": 6.904761904761905, "grad_norm": 0.7578125, "learning_rate": 9.424612612593425e-05, "loss": 0.1388, "step": 9425 }, { "epoch": 6.905494505494506, "grad_norm": 1.25, "learning_rate": 9.424452268887069e-05, "loss": 0.1543, "step": 9426 }, { "epoch": 6.906227106227107, "grad_norm": 1.2890625, "learning_rate": 9.424291904221277e-05, "loss": 0.1184, "step": 9427 }, { "epoch": 6.906959706959707, "grad_norm": 1.109375, "learning_rate": 9.424131518596815e-05, "loss": 0.1714, "step": 9428 }, { "epoch": 6.907692307692308, "grad_norm": 0.98828125, "learning_rate": 9.423971112014454e-05, "loss": 0.1371, "step": 9429 }, { "epoch": 6.908424908424909, "grad_norm": 1.1796875, "learning_rate": 9.423810684474963e-05, "loss": 0.1296, "step": 9430 }, { "epoch": 6.9091575091575095, "grad_norm": 0.6953125, "learning_rate": 9.423650235979108e-05, "loss": 0.1246, "step": 9431 }, { "epoch": 6.90989010989011, "grad_norm": 1.0234375, "learning_rate": 9.423489766527658e-05, "loss": 0.1581, "step": 9432 }, { "epoch": 6.910622710622711, "grad_norm": 1.0703125, "learning_rate": 9.423329276121385e-05, "loss": 0.1226, "step": 9433 }, { "epoch": 6.911355311355312, "grad_norm": 0.65625, "learning_rate": 9.423168764761054e-05, "loss": 0.1194, "step": 9434 }, { "epoch": 6.912087912087912, "grad_norm": 0.7265625, "learning_rate": 9.423008232447437e-05, "loss": 0.1341, "step": 9435 }, { "epoch": 6.912820512820513, "grad_norm": 0.9296875, "learning_rate": 9.422847679181299e-05, "loss": 0.1208, "step": 9436 }, { "epoch": 6.913553113553114, "grad_norm": 1.09375, "learning_rate": 9.422687104963414e-05, "loss": 0.172, "step": 9437 }, { "epoch": 6.914285714285715, "grad_norm": 1.125, "learning_rate": 9.422526509794552e-05, "loss": 0.1796, "step": 9438 }, { "epoch": 6.915018315018315, "grad_norm": 1.609375, "learning_rate": 9.422365893675478e-05, "loss": 0.1694, "step": 9439 }, { "epoch": 6.915750915750916, "grad_norm": 1.1640625, "learning_rate": 9.422205256606963e-05, "loss": 0.156, "step": 9440 }, { "epoch": 6.916483516483517, "grad_norm": 1.1484375, "learning_rate": 9.422044598589779e-05, "loss": 0.103, "step": 9441 }, { "epoch": 6.9172161172161175, "grad_norm": 1.28125, "learning_rate": 9.421883919624693e-05, "loss": 0.1497, "step": 9442 }, { "epoch": 6.917948717948718, "grad_norm": 1.1796875, "learning_rate": 9.421723219712478e-05, "loss": 0.1595, "step": 9443 }, { "epoch": 6.918681318681319, "grad_norm": 1.2890625, "learning_rate": 9.421562498853898e-05, "loss": 0.1227, "step": 9444 }, { "epoch": 6.91941391941392, "grad_norm": 0.921875, "learning_rate": 9.421401757049729e-05, "loss": 0.1452, "step": 9445 }, { "epoch": 6.92014652014652, "grad_norm": 0.65234375, "learning_rate": 9.42124099430074e-05, "loss": 0.1006, "step": 9446 }, { "epoch": 6.920879120879121, "grad_norm": 1.6171875, "learning_rate": 9.421080210607699e-05, "loss": 0.1361, "step": 9447 }, { "epoch": 6.921611721611722, "grad_norm": 1.125, "learning_rate": 9.420919405971379e-05, "loss": 0.1439, "step": 9448 }, { "epoch": 6.922344322344323, "grad_norm": 0.99609375, "learning_rate": 9.42075858039255e-05, "loss": 0.1264, "step": 9449 }, { "epoch": 6.923076923076923, "grad_norm": 1.1796875, "learning_rate": 9.42059773387198e-05, "loss": 0.1431, "step": 9450 }, { "epoch": 6.923809523809524, "grad_norm": 1.09375, "learning_rate": 9.420436866410442e-05, "loss": 0.1374, "step": 9451 }, { "epoch": 6.924542124542125, "grad_norm": 1.3125, "learning_rate": 9.420275978008706e-05, "loss": 0.161, "step": 9452 }, { "epoch": 6.9252747252747255, "grad_norm": 1.171875, "learning_rate": 9.420115068667543e-05, "loss": 0.1253, "step": 9453 }, { "epoch": 6.926007326007326, "grad_norm": 1.1015625, "learning_rate": 9.419954138387726e-05, "loss": 0.1386, "step": 9454 }, { "epoch": 6.926739926739927, "grad_norm": 1.34375, "learning_rate": 9.419793187170021e-05, "loss": 0.12, "step": 9455 }, { "epoch": 6.927472527472528, "grad_norm": 1.1875, "learning_rate": 9.419632215015204e-05, "loss": 0.133, "step": 9456 }, { "epoch": 6.9282051282051285, "grad_norm": 1.0234375, "learning_rate": 9.419471221924045e-05, "loss": 0.1541, "step": 9457 }, { "epoch": 6.928937728937729, "grad_norm": 1.3515625, "learning_rate": 9.419310207897315e-05, "loss": 0.1517, "step": 9458 }, { "epoch": 6.92967032967033, "grad_norm": 1.421875, "learning_rate": 9.419149172935785e-05, "loss": 0.1375, "step": 9459 }, { "epoch": 6.930402930402931, "grad_norm": 1.015625, "learning_rate": 9.418988117040226e-05, "loss": 0.1296, "step": 9460 }, { "epoch": 6.931135531135531, "grad_norm": 1.2421875, "learning_rate": 9.418827040211412e-05, "loss": 0.1457, "step": 9461 }, { "epoch": 6.931868131868132, "grad_norm": 0.79296875, "learning_rate": 9.418665942450114e-05, "loss": 0.1248, "step": 9462 }, { "epoch": 6.932600732600733, "grad_norm": 0.81640625, "learning_rate": 9.418504823757102e-05, "loss": 0.133, "step": 9463 }, { "epoch": 6.933333333333334, "grad_norm": 1.3515625, "learning_rate": 9.41834368413315e-05, "loss": 0.1545, "step": 9464 }, { "epoch": 6.934065934065934, "grad_norm": 1.6796875, "learning_rate": 9.418182523579028e-05, "loss": 0.1393, "step": 9465 }, { "epoch": 6.934798534798535, "grad_norm": 1.1953125, "learning_rate": 9.418021342095513e-05, "loss": 0.1336, "step": 9466 }, { "epoch": 6.935531135531136, "grad_norm": 0.984375, "learning_rate": 9.417860139683369e-05, "loss": 0.1402, "step": 9467 }, { "epoch": 6.9362637362637365, "grad_norm": 0.62109375, "learning_rate": 9.417698916343375e-05, "loss": 0.1097, "step": 9468 }, { "epoch": 6.936996336996337, "grad_norm": 1.828125, "learning_rate": 9.417537672076304e-05, "loss": 0.1598, "step": 9469 }, { "epoch": 6.937728937728938, "grad_norm": 1.0625, "learning_rate": 9.417376406882924e-05, "loss": 0.0993, "step": 9470 }, { "epoch": 6.938461538461539, "grad_norm": 1.2734375, "learning_rate": 9.41721512076401e-05, "loss": 0.1844, "step": 9471 }, { "epoch": 6.939194139194139, "grad_norm": 0.9140625, "learning_rate": 9.417053813720335e-05, "loss": 0.1278, "step": 9472 }, { "epoch": 6.93992673992674, "grad_norm": 0.6640625, "learning_rate": 9.416892485752674e-05, "loss": 0.1333, "step": 9473 }, { "epoch": 6.940659340659341, "grad_norm": 1.28125, "learning_rate": 9.416731136861794e-05, "loss": 0.1221, "step": 9474 }, { "epoch": 6.941391941391942, "grad_norm": 1.03125, "learning_rate": 9.416569767048474e-05, "loss": 0.1465, "step": 9475 }, { "epoch": 6.942124542124542, "grad_norm": 1.6015625, "learning_rate": 9.416408376313483e-05, "loss": 0.1776, "step": 9476 }, { "epoch": 6.942857142857143, "grad_norm": 0.8828125, "learning_rate": 9.416246964657597e-05, "loss": 0.1293, "step": 9477 }, { "epoch": 6.943589743589744, "grad_norm": 1.03125, "learning_rate": 9.416085532081588e-05, "loss": 0.1658, "step": 9478 }, { "epoch": 6.9443223443223445, "grad_norm": 0.75, "learning_rate": 9.41592407858623e-05, "loss": 0.1283, "step": 9479 }, { "epoch": 6.945054945054945, "grad_norm": 1.03125, "learning_rate": 9.415762604172296e-05, "loss": 0.1454, "step": 9480 }, { "epoch": 6.945787545787546, "grad_norm": 0.84765625, "learning_rate": 9.415601108840562e-05, "loss": 0.1244, "step": 9481 }, { "epoch": 6.946520146520147, "grad_norm": 1.2890625, "learning_rate": 9.415439592591798e-05, "loss": 0.1683, "step": 9482 }, { "epoch": 6.947252747252747, "grad_norm": 1.5625, "learning_rate": 9.415278055426782e-05, "loss": 0.1382, "step": 9483 }, { "epoch": 6.947985347985348, "grad_norm": 1.0859375, "learning_rate": 9.415116497346283e-05, "loss": 0.1584, "step": 9484 }, { "epoch": 6.948717948717949, "grad_norm": 1.3203125, "learning_rate": 9.414954918351081e-05, "loss": 0.137, "step": 9485 }, { "epoch": 6.94945054945055, "grad_norm": 0.984375, "learning_rate": 9.414793318441945e-05, "loss": 0.1396, "step": 9486 }, { "epoch": 6.95018315018315, "grad_norm": 0.98046875, "learning_rate": 9.41463169761965e-05, "loss": 0.1671, "step": 9487 }, { "epoch": 6.950915750915751, "grad_norm": 1.03125, "learning_rate": 9.414470055884975e-05, "loss": 0.153, "step": 9488 }, { "epoch": 6.951648351648352, "grad_norm": 0.80078125, "learning_rate": 9.41430839323869e-05, "loss": 0.1538, "step": 9489 }, { "epoch": 6.9523809523809526, "grad_norm": 1.421875, "learning_rate": 9.41414670968157e-05, "loss": 0.144, "step": 9490 }, { "epoch": 6.953113553113553, "grad_norm": 0.828125, "learning_rate": 9.413985005214392e-05, "loss": 0.1389, "step": 9491 }, { "epoch": 6.953846153846154, "grad_norm": 1.1015625, "learning_rate": 9.413823279837927e-05, "loss": 0.1358, "step": 9492 }, { "epoch": 6.954578754578755, "grad_norm": 0.9375, "learning_rate": 9.413661533552953e-05, "loss": 0.1354, "step": 9493 }, { "epoch": 6.9553113553113555, "grad_norm": 0.85546875, "learning_rate": 9.413499766360244e-05, "loss": 0.1431, "step": 9494 }, { "epoch": 6.956043956043956, "grad_norm": 1.0859375, "learning_rate": 9.413337978260577e-05, "loss": 0.1401, "step": 9495 }, { "epoch": 6.956776556776557, "grad_norm": 0.875, "learning_rate": 9.413176169254724e-05, "loss": 0.1345, "step": 9496 }, { "epoch": 6.957509157509158, "grad_norm": 0.83984375, "learning_rate": 9.413014339343461e-05, "loss": 0.1457, "step": 9497 }, { "epoch": 6.958241758241758, "grad_norm": 0.8828125, "learning_rate": 9.412852488527564e-05, "loss": 0.1467, "step": 9498 }, { "epoch": 6.958974358974359, "grad_norm": 1.0859375, "learning_rate": 9.41269061680781e-05, "loss": 0.1638, "step": 9499 }, { "epoch": 6.95970695970696, "grad_norm": 1.6953125, "learning_rate": 9.412528724184973e-05, "loss": 0.1891, "step": 9500 }, { "epoch": 6.960439560439561, "grad_norm": 0.75390625, "learning_rate": 9.412366810659826e-05, "loss": 0.1283, "step": 9501 }, { "epoch": 6.961172161172161, "grad_norm": 1.3046875, "learning_rate": 9.41220487623315e-05, "loss": 0.1369, "step": 9502 }, { "epoch": 6.961904761904762, "grad_norm": 0.796875, "learning_rate": 9.412042920905718e-05, "loss": 0.1667, "step": 9503 }, { "epoch": 6.962637362637363, "grad_norm": 0.73046875, "learning_rate": 9.411880944678306e-05, "loss": 0.0907, "step": 9504 }, { "epoch": 6.9633699633699635, "grad_norm": 0.93359375, "learning_rate": 9.411718947551691e-05, "loss": 0.1406, "step": 9505 }, { "epoch": 6.964102564102564, "grad_norm": 1.0078125, "learning_rate": 9.411556929526648e-05, "loss": 0.1421, "step": 9506 }, { "epoch": 6.964835164835165, "grad_norm": 1.2890625, "learning_rate": 9.411394890603954e-05, "loss": 0.1569, "step": 9507 }, { "epoch": 6.965567765567766, "grad_norm": 1.109375, "learning_rate": 9.411232830784387e-05, "loss": 0.1657, "step": 9508 }, { "epoch": 6.966300366300366, "grad_norm": 0.96875, "learning_rate": 9.41107075006872e-05, "loss": 0.157, "step": 9509 }, { "epoch": 6.967032967032967, "grad_norm": 0.8828125, "learning_rate": 9.410908648457733e-05, "loss": 0.1179, "step": 9510 }, { "epoch": 6.967765567765568, "grad_norm": 1.5, "learning_rate": 9.410746525952199e-05, "loss": 0.1799, "step": 9511 }, { "epoch": 6.968498168498169, "grad_norm": 1.1328125, "learning_rate": 9.410584382552898e-05, "loss": 0.1407, "step": 9512 }, { "epoch": 6.969230769230769, "grad_norm": 0.8984375, "learning_rate": 9.410422218260605e-05, "loss": 0.1368, "step": 9513 }, { "epoch": 6.96996336996337, "grad_norm": 1.15625, "learning_rate": 9.410260033076099e-05, "loss": 0.1405, "step": 9514 }, { "epoch": 6.970695970695971, "grad_norm": 0.87890625, "learning_rate": 9.410097827000156e-05, "loss": 0.1499, "step": 9515 }, { "epoch": 6.9714285714285715, "grad_norm": 1.1953125, "learning_rate": 9.40993560003355e-05, "loss": 0.1345, "step": 9516 }, { "epoch": 6.972161172161172, "grad_norm": 1.5234375, "learning_rate": 9.409773352177063e-05, "loss": 0.1705, "step": 9517 }, { "epoch": 6.972893772893773, "grad_norm": 1.3984375, "learning_rate": 9.409611083431472e-05, "loss": 0.1392, "step": 9518 }, { "epoch": 6.973626373626374, "grad_norm": 1.375, "learning_rate": 9.409448793797552e-05, "loss": 0.139, "step": 9519 }, { "epoch": 6.9743589743589745, "grad_norm": 1.1875, "learning_rate": 9.40928648327608e-05, "loss": 0.1643, "step": 9520 }, { "epoch": 6.975091575091575, "grad_norm": 0.64453125, "learning_rate": 9.409124151867838e-05, "loss": 0.139, "step": 9521 }, { "epoch": 6.975824175824176, "grad_norm": 0.92578125, "learning_rate": 9.408961799573602e-05, "loss": 0.1437, "step": 9522 }, { "epoch": 6.976556776556777, "grad_norm": 1.2265625, "learning_rate": 9.408799426394146e-05, "loss": 0.1545, "step": 9523 }, { "epoch": 6.977289377289377, "grad_norm": 1.3203125, "learning_rate": 9.408637032330253e-05, "loss": 0.1678, "step": 9524 }, { "epoch": 6.978021978021978, "grad_norm": 0.70703125, "learning_rate": 9.4084746173827e-05, "loss": 0.1214, "step": 9525 }, { "epoch": 6.978754578754579, "grad_norm": 0.98046875, "learning_rate": 9.408312181552261e-05, "loss": 0.1549, "step": 9526 }, { "epoch": 6.97948717948718, "grad_norm": 0.90625, "learning_rate": 9.408149724839721e-05, "loss": 0.1565, "step": 9527 }, { "epoch": 6.98021978021978, "grad_norm": 0.765625, "learning_rate": 9.407987247245853e-05, "loss": 0.1321, "step": 9528 }, { "epoch": 6.980952380952381, "grad_norm": 1.03125, "learning_rate": 9.407824748771438e-05, "loss": 0.1331, "step": 9529 }, { "epoch": 6.981684981684982, "grad_norm": 1.0625, "learning_rate": 9.407662229417253e-05, "loss": 0.13, "step": 9530 }, { "epoch": 6.9824175824175825, "grad_norm": 0.77734375, "learning_rate": 9.40749968918408e-05, "loss": 0.1286, "step": 9531 }, { "epoch": 6.983150183150183, "grad_norm": 1.3203125, "learning_rate": 9.407337128072695e-05, "loss": 0.144, "step": 9532 }, { "epoch": 6.983882783882784, "grad_norm": 0.84765625, "learning_rate": 9.407174546083876e-05, "loss": 0.1372, "step": 9533 }, { "epoch": 6.984615384615385, "grad_norm": 1.0703125, "learning_rate": 9.407011943218404e-05, "loss": 0.1333, "step": 9534 }, { "epoch": 6.985347985347985, "grad_norm": 1.0625, "learning_rate": 9.406849319477058e-05, "loss": 0.1926, "step": 9535 }, { "epoch": 6.986080586080586, "grad_norm": 0.828125, "learning_rate": 9.406686674860616e-05, "loss": 0.1122, "step": 9536 }, { "epoch": 6.986813186813187, "grad_norm": 1.203125, "learning_rate": 9.406524009369857e-05, "loss": 0.1346, "step": 9537 }, { "epoch": 6.987545787545788, "grad_norm": 1.265625, "learning_rate": 9.406361323005563e-05, "loss": 0.1451, "step": 9538 }, { "epoch": 6.988278388278388, "grad_norm": 1.1484375, "learning_rate": 9.406198615768511e-05, "loss": 0.1465, "step": 9539 }, { "epoch": 6.989010989010989, "grad_norm": 0.78125, "learning_rate": 9.406035887659483e-05, "loss": 0.1218, "step": 9540 }, { "epoch": 6.98974358974359, "grad_norm": 0.98828125, "learning_rate": 9.405873138679255e-05, "loss": 0.1307, "step": 9541 }, { "epoch": 6.9904761904761905, "grad_norm": 1.0546875, "learning_rate": 9.405710368828609e-05, "loss": 0.1395, "step": 9542 }, { "epoch": 6.991208791208791, "grad_norm": 1.09375, "learning_rate": 9.405547578108325e-05, "loss": 0.1566, "step": 9543 }, { "epoch": 6.991941391941392, "grad_norm": 1.125, "learning_rate": 9.405384766519184e-05, "loss": 0.1232, "step": 9544 }, { "epoch": 6.992673992673993, "grad_norm": 1.3125, "learning_rate": 9.405221934061963e-05, "loss": 0.1134, "step": 9545 }, { "epoch": 6.993406593406593, "grad_norm": 1.6328125, "learning_rate": 9.405059080737445e-05, "loss": 0.185, "step": 9546 }, { "epoch": 6.994139194139194, "grad_norm": 1.0625, "learning_rate": 9.40489620654641e-05, "loss": 0.1792, "step": 9547 }, { "epoch": 6.994871794871795, "grad_norm": 1.7890625, "learning_rate": 9.404733311489636e-05, "loss": 0.1371, "step": 9548 }, { "epoch": 6.995604395604396, "grad_norm": 1.5625, "learning_rate": 9.404570395567907e-05, "loss": 0.1712, "step": 9549 }, { "epoch": 6.996336996336996, "grad_norm": 1.4921875, "learning_rate": 9.404407458781998e-05, "loss": 0.1408, "step": 9550 }, { "epoch": 6.997069597069597, "grad_norm": 0.9453125, "learning_rate": 9.404244501132696e-05, "loss": 0.1222, "step": 9551 }, { "epoch": 6.997802197802198, "grad_norm": 1.0078125, "learning_rate": 9.404081522620782e-05, "loss": 0.1422, "step": 9552 }, { "epoch": 6.9985347985347985, "grad_norm": 0.90625, "learning_rate": 9.40391852324703e-05, "loss": 0.1278, "step": 9553 }, { "epoch": 6.999267399267399, "grad_norm": 0.51953125, "learning_rate": 9.403755503012227e-05, "loss": 0.1143, "step": 9554 }, { "epoch": 7.0, "grad_norm": 0.87109375, "learning_rate": 9.40359246191715e-05, "loss": 0.1643, "step": 9555 }, { "epoch": 7.000732600732601, "grad_norm": 0.88671875, "learning_rate": 9.403429399962585e-05, "loss": 0.115, "step": 9556 }, { "epoch": 7.0014652014652015, "grad_norm": 1.1328125, "learning_rate": 9.403266317149309e-05, "loss": 0.1596, "step": 9557 }, { "epoch": 7.002197802197802, "grad_norm": 1.2890625, "learning_rate": 9.403103213478105e-05, "loss": 0.155, "step": 9558 }, { "epoch": 7.002930402930403, "grad_norm": 1.7109375, "learning_rate": 9.402940088949755e-05, "loss": 0.1966, "step": 9559 }, { "epoch": 7.003663003663004, "grad_norm": 1.53125, "learning_rate": 9.40277694356504e-05, "loss": 0.1524, "step": 9560 }, { "epoch": 7.004395604395604, "grad_norm": 1.1015625, "learning_rate": 9.402613777324742e-05, "loss": 0.1422, "step": 9561 }, { "epoch": 7.005128205128205, "grad_norm": 1.09375, "learning_rate": 9.402450590229642e-05, "loss": 0.1339, "step": 9562 }, { "epoch": 7.005860805860806, "grad_norm": 0.86328125, "learning_rate": 9.402287382280523e-05, "loss": 0.1393, "step": 9563 }, { "epoch": 7.006593406593407, "grad_norm": 0.734375, "learning_rate": 9.402124153478165e-05, "loss": 0.1463, "step": 9564 }, { "epoch": 7.007326007326007, "grad_norm": 1.390625, "learning_rate": 9.401960903823352e-05, "loss": 0.1504, "step": 9565 }, { "epoch": 7.008058608058608, "grad_norm": 0.92578125, "learning_rate": 9.401797633316867e-05, "loss": 0.1572, "step": 9566 }, { "epoch": 7.008791208791209, "grad_norm": 1.25, "learning_rate": 9.40163434195949e-05, "loss": 0.1921, "step": 9567 }, { "epoch": 7.0095238095238095, "grad_norm": 1.3671875, "learning_rate": 9.401471029752005e-05, "loss": 0.137, "step": 9568 }, { "epoch": 7.01025641025641, "grad_norm": 1.2890625, "learning_rate": 9.401307696695192e-05, "loss": 0.153, "step": 9569 }, { "epoch": 7.010989010989011, "grad_norm": 0.921875, "learning_rate": 9.401144342789835e-05, "loss": 0.1204, "step": 9570 }, { "epoch": 7.011721611721612, "grad_norm": 0.95703125, "learning_rate": 9.400980968036721e-05, "loss": 0.1983, "step": 9571 }, { "epoch": 7.012454212454212, "grad_norm": 1.0234375, "learning_rate": 9.400817572436625e-05, "loss": 0.1377, "step": 9572 }, { "epoch": 7.013186813186813, "grad_norm": 0.99609375, "learning_rate": 9.400654155990336e-05, "loss": 0.1284, "step": 9573 }, { "epoch": 7.013919413919414, "grad_norm": 1.0234375, "learning_rate": 9.400490718698634e-05, "loss": 0.15, "step": 9574 }, { "epoch": 7.014652014652015, "grad_norm": 1.046875, "learning_rate": 9.400327260562303e-05, "loss": 0.1338, "step": 9575 }, { "epoch": 7.015384615384615, "grad_norm": 1.2109375, "learning_rate": 9.400163781582126e-05, "loss": 0.123, "step": 9576 }, { "epoch": 7.016117216117216, "grad_norm": 1.5546875, "learning_rate": 9.400000281758885e-05, "loss": 0.1508, "step": 9577 }, { "epoch": 7.016849816849817, "grad_norm": 0.89453125, "learning_rate": 9.399836761093365e-05, "loss": 0.149, "step": 9578 }, { "epoch": 7.0175824175824175, "grad_norm": 1.2421875, "learning_rate": 9.39967321958635e-05, "loss": 0.1467, "step": 9579 }, { "epoch": 7.018315018315018, "grad_norm": 1.09375, "learning_rate": 9.399509657238623e-05, "loss": 0.1451, "step": 9580 }, { "epoch": 7.019047619047619, "grad_norm": 1.125, "learning_rate": 9.399346074050966e-05, "loss": 0.1365, "step": 9581 }, { "epoch": 7.01978021978022, "grad_norm": 0.9375, "learning_rate": 9.399182470024167e-05, "loss": 0.1116, "step": 9582 }, { "epoch": 7.02051282051282, "grad_norm": 2.203125, "learning_rate": 9.399018845159004e-05, "loss": 0.1731, "step": 9583 }, { "epoch": 7.021245421245421, "grad_norm": 1.2421875, "learning_rate": 9.398855199456264e-05, "loss": 0.1307, "step": 9584 }, { "epoch": 7.021978021978022, "grad_norm": 0.98046875, "learning_rate": 9.398691532916734e-05, "loss": 0.1394, "step": 9585 }, { "epoch": 7.022710622710623, "grad_norm": 1.0078125, "learning_rate": 9.398527845541192e-05, "loss": 0.1329, "step": 9586 }, { "epoch": 7.023443223443223, "grad_norm": 0.9296875, "learning_rate": 9.398364137330428e-05, "loss": 0.147, "step": 9587 }, { "epoch": 7.024175824175824, "grad_norm": 1.515625, "learning_rate": 9.398200408285223e-05, "loss": 0.1602, "step": 9588 }, { "epoch": 7.024908424908425, "grad_norm": 1.203125, "learning_rate": 9.39803665840636e-05, "loss": 0.1107, "step": 9589 }, { "epoch": 7.0256410256410255, "grad_norm": 0.87109375, "learning_rate": 9.39787288769463e-05, "loss": 0.1364, "step": 9590 }, { "epoch": 7.026373626373626, "grad_norm": 1.28125, "learning_rate": 9.397709096150811e-05, "loss": 0.1804, "step": 9591 }, { "epoch": 7.027106227106227, "grad_norm": 0.90625, "learning_rate": 9.397545283775692e-05, "loss": 0.1668, "step": 9592 }, { "epoch": 7.027838827838828, "grad_norm": 1.359375, "learning_rate": 9.397381450570055e-05, "loss": 0.1564, "step": 9593 }, { "epoch": 7.0285714285714285, "grad_norm": 1.0234375, "learning_rate": 9.397217596534688e-05, "loss": 0.1447, "step": 9594 }, { "epoch": 7.029304029304029, "grad_norm": 0.76953125, "learning_rate": 9.397053721670374e-05, "loss": 0.1556, "step": 9595 }, { "epoch": 7.03003663003663, "grad_norm": 1.0078125, "learning_rate": 9.396889825977898e-05, "loss": 0.1547, "step": 9596 }, { "epoch": 7.030769230769231, "grad_norm": 1.25, "learning_rate": 9.396725909458046e-05, "loss": 0.1279, "step": 9597 }, { "epoch": 7.031501831501831, "grad_norm": 1.0, "learning_rate": 9.396561972111603e-05, "loss": 0.1196, "step": 9598 }, { "epoch": 7.032234432234432, "grad_norm": 1.140625, "learning_rate": 9.396398013939355e-05, "loss": 0.1602, "step": 9599 }, { "epoch": 7.032967032967033, "grad_norm": 1.125, "learning_rate": 9.396234034942087e-05, "loss": 0.1481, "step": 9600 }, { "epoch": 7.033699633699634, "grad_norm": 1.0625, "learning_rate": 9.396070035120585e-05, "loss": 0.154, "step": 9601 }, { "epoch": 7.034432234432234, "grad_norm": 1.578125, "learning_rate": 9.395906014475635e-05, "loss": 0.1436, "step": 9602 }, { "epoch": 7.035164835164835, "grad_norm": 1.3671875, "learning_rate": 9.395741973008022e-05, "loss": 0.1583, "step": 9603 }, { "epoch": 7.035897435897436, "grad_norm": 0.94921875, "learning_rate": 9.395577910718533e-05, "loss": 0.1215, "step": 9604 }, { "epoch": 7.0366300366300365, "grad_norm": 1.0703125, "learning_rate": 9.395413827607954e-05, "loss": 0.1508, "step": 9605 }, { "epoch": 7.037362637362637, "grad_norm": 1.046875, "learning_rate": 9.39524972367707e-05, "loss": 0.1443, "step": 9606 }, { "epoch": 7.038095238095238, "grad_norm": 1.6953125, "learning_rate": 9.39508559892667e-05, "loss": 0.1723, "step": 9607 }, { "epoch": 7.038827838827839, "grad_norm": 1.1953125, "learning_rate": 9.394921453357536e-05, "loss": 0.1446, "step": 9608 }, { "epoch": 7.039560439560439, "grad_norm": 1.5625, "learning_rate": 9.39475728697046e-05, "loss": 0.1293, "step": 9609 }, { "epoch": 7.04029304029304, "grad_norm": 1.7734375, "learning_rate": 9.394593099766222e-05, "loss": 0.1833, "step": 9610 }, { "epoch": 7.041025641025641, "grad_norm": 1.390625, "learning_rate": 9.394428891745614e-05, "loss": 0.1215, "step": 9611 }, { "epoch": 7.041758241758242, "grad_norm": 1.515625, "learning_rate": 9.394264662909423e-05, "loss": 0.1593, "step": 9612 }, { "epoch": 7.042490842490842, "grad_norm": 1.0390625, "learning_rate": 9.39410041325843e-05, "loss": 0.1547, "step": 9613 }, { "epoch": 7.043223443223443, "grad_norm": 1.6484375, "learning_rate": 9.393936142793428e-05, "loss": 0.1496, "step": 9614 }, { "epoch": 7.043956043956044, "grad_norm": 1.375, "learning_rate": 9.393771851515202e-05, "loss": 0.1281, "step": 9615 }, { "epoch": 7.0446886446886445, "grad_norm": 1.6171875, "learning_rate": 9.393607539424539e-05, "loss": 0.1363, "step": 9616 }, { "epoch": 7.045421245421245, "grad_norm": 1.0546875, "learning_rate": 9.393443206522225e-05, "loss": 0.1326, "step": 9617 }, { "epoch": 7.046153846153846, "grad_norm": 1.09375, "learning_rate": 9.39327885280905e-05, "loss": 0.1407, "step": 9618 }, { "epoch": 7.046886446886447, "grad_norm": 1.578125, "learning_rate": 9.393114478285799e-05, "loss": 0.1537, "step": 9619 }, { "epoch": 7.0476190476190474, "grad_norm": 1.359375, "learning_rate": 9.392950082953262e-05, "loss": 0.1288, "step": 9620 }, { "epoch": 7.048351648351648, "grad_norm": 1.4921875, "learning_rate": 9.392785666812226e-05, "loss": 0.1624, "step": 9621 }, { "epoch": 7.049084249084249, "grad_norm": 0.8046875, "learning_rate": 9.392621229863478e-05, "loss": 0.1921, "step": 9622 }, { "epoch": 7.04981684981685, "grad_norm": 0.79296875, "learning_rate": 9.392456772107805e-05, "loss": 0.1571, "step": 9623 }, { "epoch": 7.05054945054945, "grad_norm": 0.96875, "learning_rate": 9.392292293545996e-05, "loss": 0.128, "step": 9624 }, { "epoch": 7.051282051282051, "grad_norm": 0.765625, "learning_rate": 9.39212779417884e-05, "loss": 0.1293, "step": 9625 }, { "epoch": 7.052014652014652, "grad_norm": 1.359375, "learning_rate": 9.391963274007124e-05, "loss": 0.143, "step": 9626 }, { "epoch": 7.052747252747253, "grad_norm": 1.015625, "learning_rate": 9.391798733031637e-05, "loss": 0.1623, "step": 9627 }, { "epoch": 7.053479853479853, "grad_norm": 2.1875, "learning_rate": 9.391634171253167e-05, "loss": 0.1329, "step": 9628 }, { "epoch": 7.054212454212454, "grad_norm": 1.6953125, "learning_rate": 9.391469588672503e-05, "loss": 0.1368, "step": 9629 }, { "epoch": 7.054945054945055, "grad_norm": 1.9765625, "learning_rate": 9.391304985290431e-05, "loss": 0.1424, "step": 9630 }, { "epoch": 7.0556776556776555, "grad_norm": 0.84375, "learning_rate": 9.391140361107744e-05, "loss": 0.1452, "step": 9631 }, { "epoch": 7.056410256410256, "grad_norm": 1.65625, "learning_rate": 9.390975716125226e-05, "loss": 0.1863, "step": 9632 }, { "epoch": 7.057142857142857, "grad_norm": 1.109375, "learning_rate": 9.390811050343669e-05, "loss": 0.1391, "step": 9633 }, { "epoch": 7.057875457875458, "grad_norm": 2.046875, "learning_rate": 9.390646363763861e-05, "loss": 0.1851, "step": 9634 }, { "epoch": 7.058608058608058, "grad_norm": 2.3125, "learning_rate": 9.390481656386592e-05, "loss": 0.1545, "step": 9635 }, { "epoch": 7.059340659340659, "grad_norm": 1.390625, "learning_rate": 9.390316928212651e-05, "loss": 0.1389, "step": 9636 }, { "epoch": 7.06007326007326, "grad_norm": 1.6484375, "learning_rate": 9.390152179242826e-05, "loss": 0.1357, "step": 9637 }, { "epoch": 7.060805860805861, "grad_norm": 0.765625, "learning_rate": 9.389987409477908e-05, "loss": 0.1378, "step": 9638 }, { "epoch": 7.061538461538461, "grad_norm": 0.87109375, "learning_rate": 9.389822618918685e-05, "loss": 0.1498, "step": 9639 }, { "epoch": 7.062271062271062, "grad_norm": 1.484375, "learning_rate": 9.389657807565947e-05, "loss": 0.1344, "step": 9640 }, { "epoch": 7.063003663003663, "grad_norm": 1.375, "learning_rate": 9.389492975420484e-05, "loss": 0.153, "step": 9641 }, { "epoch": 7.0637362637362635, "grad_norm": 2.359375, "learning_rate": 9.389328122483086e-05, "loss": 0.1944, "step": 9642 }, { "epoch": 7.064468864468864, "grad_norm": 1.0234375, "learning_rate": 9.389163248754541e-05, "loss": 0.1446, "step": 9643 }, { "epoch": 7.065201465201465, "grad_norm": 0.76953125, "learning_rate": 9.388998354235641e-05, "loss": 0.1552, "step": 9644 }, { "epoch": 7.065934065934066, "grad_norm": 0.9296875, "learning_rate": 9.388833438927177e-05, "loss": 0.1198, "step": 9645 }, { "epoch": 7.066666666666666, "grad_norm": 1.5703125, "learning_rate": 9.388668502829936e-05, "loss": 0.1441, "step": 9646 }, { "epoch": 7.067399267399267, "grad_norm": 1.15625, "learning_rate": 9.388503545944712e-05, "loss": 0.1466, "step": 9647 }, { "epoch": 7.068131868131868, "grad_norm": 1.2734375, "learning_rate": 9.38833856827229e-05, "loss": 0.1573, "step": 9648 }, { "epoch": 7.068864468864469, "grad_norm": 0.5078125, "learning_rate": 9.388173569813466e-05, "loss": 0.0955, "step": 9649 }, { "epoch": 7.069597069597069, "grad_norm": 0.99609375, "learning_rate": 9.388008550569027e-05, "loss": 0.1925, "step": 9650 }, { "epoch": 7.07032967032967, "grad_norm": 1.0859375, "learning_rate": 9.387843510539767e-05, "loss": 0.1202, "step": 9651 }, { "epoch": 7.071062271062271, "grad_norm": 1.0859375, "learning_rate": 9.387678449726474e-05, "loss": 0.1486, "step": 9652 }, { "epoch": 7.0717948717948715, "grad_norm": 0.85546875, "learning_rate": 9.387513368129939e-05, "loss": 0.1322, "step": 9653 }, { "epoch": 7.072527472527472, "grad_norm": 0.69921875, "learning_rate": 9.387348265750953e-05, "loss": 0.1302, "step": 9654 }, { "epoch": 7.073260073260073, "grad_norm": 0.7734375, "learning_rate": 9.387183142590308e-05, "loss": 0.1219, "step": 9655 }, { "epoch": 7.073992673992674, "grad_norm": 1.046875, "learning_rate": 9.387017998648793e-05, "loss": 0.1636, "step": 9656 }, { "epoch": 7.0747252747252745, "grad_norm": 1.453125, "learning_rate": 9.386852833927205e-05, "loss": 0.1327, "step": 9657 }, { "epoch": 7.075457875457875, "grad_norm": 1.3984375, "learning_rate": 9.386687648426329e-05, "loss": 0.1585, "step": 9658 }, { "epoch": 7.076190476190476, "grad_norm": 0.9140625, "learning_rate": 9.38652244214696e-05, "loss": 0.0968, "step": 9659 }, { "epoch": 7.076923076923077, "grad_norm": 1.6796875, "learning_rate": 9.386357215089888e-05, "loss": 0.1489, "step": 9660 }, { "epoch": 7.077655677655677, "grad_norm": 0.75, "learning_rate": 9.386191967255904e-05, "loss": 0.13, "step": 9661 }, { "epoch": 7.078388278388278, "grad_norm": 1.640625, "learning_rate": 9.3860266986458e-05, "loss": 0.1631, "step": 9662 }, { "epoch": 7.079120879120879, "grad_norm": 0.8828125, "learning_rate": 9.385861409260372e-05, "loss": 0.1183, "step": 9663 }, { "epoch": 7.07985347985348, "grad_norm": 1.4375, "learning_rate": 9.385696099100406e-05, "loss": 0.1397, "step": 9664 }, { "epoch": 7.08058608058608, "grad_norm": 0.94140625, "learning_rate": 9.385530768166698e-05, "loss": 0.1366, "step": 9665 }, { "epoch": 7.081318681318681, "grad_norm": 1.0859375, "learning_rate": 9.385365416460038e-05, "loss": 0.1348, "step": 9666 }, { "epoch": 7.082051282051282, "grad_norm": 1.09375, "learning_rate": 9.385200043981219e-05, "loss": 0.0977, "step": 9667 }, { "epoch": 7.0827838827838825, "grad_norm": 1.3515625, "learning_rate": 9.385034650731034e-05, "loss": 0.1228, "step": 9668 }, { "epoch": 7.083516483516483, "grad_norm": 1.53125, "learning_rate": 9.384869236710275e-05, "loss": 0.1759, "step": 9669 }, { "epoch": 7.084249084249084, "grad_norm": 1.3984375, "learning_rate": 9.384703801919733e-05, "loss": 0.1446, "step": 9670 }, { "epoch": 7.084981684981685, "grad_norm": 1.203125, "learning_rate": 9.384538346360204e-05, "loss": 0.1416, "step": 9671 }, { "epoch": 7.085714285714285, "grad_norm": 1.234375, "learning_rate": 9.384372870032479e-05, "loss": 0.1356, "step": 9672 }, { "epoch": 7.086446886446886, "grad_norm": 1.125, "learning_rate": 9.384207372937348e-05, "loss": 0.1594, "step": 9673 }, { "epoch": 7.087179487179487, "grad_norm": 1.5234375, "learning_rate": 9.384041855075608e-05, "loss": 0.165, "step": 9674 }, { "epoch": 7.087912087912088, "grad_norm": 1.2421875, "learning_rate": 9.383876316448052e-05, "loss": 0.161, "step": 9675 }, { "epoch": 7.088644688644688, "grad_norm": 0.765625, "learning_rate": 9.383710757055472e-05, "loss": 0.1333, "step": 9676 }, { "epoch": 7.089377289377289, "grad_norm": 1.0703125, "learning_rate": 9.383545176898659e-05, "loss": 0.1218, "step": 9677 }, { "epoch": 7.09010989010989, "grad_norm": 1.0859375, "learning_rate": 9.38337957597841e-05, "loss": 0.1458, "step": 9678 }, { "epoch": 7.0908424908424905, "grad_norm": 1.1328125, "learning_rate": 9.383213954295516e-05, "loss": 0.1537, "step": 9679 }, { "epoch": 7.091575091575091, "grad_norm": 1.9296875, "learning_rate": 9.383048311850771e-05, "loss": 0.1714, "step": 9680 }, { "epoch": 7.092307692307692, "grad_norm": 0.9921875, "learning_rate": 9.382882648644971e-05, "loss": 0.1891, "step": 9681 }, { "epoch": 7.093040293040293, "grad_norm": 1.15625, "learning_rate": 9.382716964678906e-05, "loss": 0.1475, "step": 9682 }, { "epoch": 7.093772893772893, "grad_norm": 1.96875, "learning_rate": 9.382551259953372e-05, "loss": 0.1796, "step": 9683 }, { "epoch": 7.094505494505494, "grad_norm": 1.5625, "learning_rate": 9.382385534469164e-05, "loss": 0.1575, "step": 9684 }, { "epoch": 7.095238095238095, "grad_norm": 1.6953125, "learning_rate": 9.382219788227073e-05, "loss": 0.1672, "step": 9685 }, { "epoch": 7.095970695970696, "grad_norm": 1.046875, "learning_rate": 9.382054021227897e-05, "loss": 0.1255, "step": 9686 }, { "epoch": 7.096703296703296, "grad_norm": 1.0625, "learning_rate": 9.381888233472426e-05, "loss": 0.1244, "step": 9687 }, { "epoch": 7.097435897435897, "grad_norm": 1.21875, "learning_rate": 9.381722424961456e-05, "loss": 0.112, "step": 9688 }, { "epoch": 7.098168498168498, "grad_norm": 0.93359375, "learning_rate": 9.381556595695783e-05, "loss": 0.1248, "step": 9689 }, { "epoch": 7.0989010989010985, "grad_norm": 1.4765625, "learning_rate": 9.3813907456762e-05, "loss": 0.1519, "step": 9690 }, { "epoch": 7.099633699633699, "grad_norm": 1.2421875, "learning_rate": 9.381224874903503e-05, "loss": 0.1163, "step": 9691 }, { "epoch": 7.1003663003663, "grad_norm": 0.70703125, "learning_rate": 9.381058983378484e-05, "loss": 0.1193, "step": 9692 }, { "epoch": 7.101098901098901, "grad_norm": 1.7265625, "learning_rate": 9.380893071101941e-05, "loss": 0.1662, "step": 9693 }, { "epoch": 7.1018315018315015, "grad_norm": 1.046875, "learning_rate": 9.380727138074667e-05, "loss": 0.1242, "step": 9694 }, { "epoch": 7.102564102564102, "grad_norm": 0.9765625, "learning_rate": 9.380561184297459e-05, "loss": 0.1458, "step": 9695 }, { "epoch": 7.103296703296703, "grad_norm": 1.1953125, "learning_rate": 9.38039520977111e-05, "loss": 0.106, "step": 9696 }, { "epoch": 7.104029304029304, "grad_norm": 1.03125, "learning_rate": 9.380229214496414e-05, "loss": 0.1411, "step": 9697 }, { "epoch": 7.104761904761904, "grad_norm": 1.1953125, "learning_rate": 9.380063198474171e-05, "loss": 0.1365, "step": 9698 }, { "epoch": 7.105494505494505, "grad_norm": 1.15625, "learning_rate": 9.379897161705172e-05, "loss": 0.1219, "step": 9699 }, { "epoch": 7.106227106227106, "grad_norm": 2.046875, "learning_rate": 9.379731104190215e-05, "loss": 0.1443, "step": 9700 }, { "epoch": 7.106959706959707, "grad_norm": 2.453125, "learning_rate": 9.379565025930095e-05, "loss": 0.1657, "step": 9701 }, { "epoch": 7.107692307692307, "grad_norm": 2.109375, "learning_rate": 9.379398926925607e-05, "loss": 0.1425, "step": 9702 }, { "epoch": 7.108424908424908, "grad_norm": 2.125, "learning_rate": 9.379232807177549e-05, "loss": 0.159, "step": 9703 }, { "epoch": 7.109157509157509, "grad_norm": 0.83203125, "learning_rate": 9.379066666686713e-05, "loss": 0.1361, "step": 9704 }, { "epoch": 7.1098901098901095, "grad_norm": 1.0078125, "learning_rate": 9.3789005054539e-05, "loss": 0.1222, "step": 9705 }, { "epoch": 7.11062271062271, "grad_norm": 1.78125, "learning_rate": 9.378734323479902e-05, "loss": 0.1562, "step": 9706 }, { "epoch": 7.111355311355311, "grad_norm": 1.953125, "learning_rate": 9.378568120765517e-05, "loss": 0.1311, "step": 9707 }, { "epoch": 7.112087912087912, "grad_norm": 1.890625, "learning_rate": 9.378401897311542e-05, "loss": 0.1657, "step": 9708 }, { "epoch": 7.112820512820512, "grad_norm": 1.9765625, "learning_rate": 9.378235653118772e-05, "loss": 0.1763, "step": 9709 }, { "epoch": 7.113553113553113, "grad_norm": 1.4296875, "learning_rate": 9.378069388188003e-05, "loss": 0.19, "step": 9710 }, { "epoch": 7.114285714285714, "grad_norm": 0.85546875, "learning_rate": 9.377903102520035e-05, "loss": 0.1759, "step": 9711 }, { "epoch": 7.115018315018315, "grad_norm": 1.5625, "learning_rate": 9.37773679611566e-05, "loss": 0.1453, "step": 9712 }, { "epoch": 7.115750915750915, "grad_norm": 0.78515625, "learning_rate": 9.37757046897568e-05, "loss": 0.1282, "step": 9713 }, { "epoch": 7.116483516483516, "grad_norm": 1.734375, "learning_rate": 9.377404121100888e-05, "loss": 0.116, "step": 9714 }, { "epoch": 7.117216117216117, "grad_norm": 1.3125, "learning_rate": 9.377237752492081e-05, "loss": 0.1548, "step": 9715 }, { "epoch": 7.1179487179487175, "grad_norm": 0.83984375, "learning_rate": 9.377071363150058e-05, "loss": 0.1355, "step": 9716 }, { "epoch": 7.118681318681318, "grad_norm": 0.65625, "learning_rate": 9.376904953075617e-05, "loss": 0.097, "step": 9717 }, { "epoch": 7.119413919413919, "grad_norm": 1.2109375, "learning_rate": 9.376738522269554e-05, "loss": 0.1738, "step": 9718 }, { "epoch": 7.12014652014652, "grad_norm": 1.2265625, "learning_rate": 9.376572070732666e-05, "loss": 0.1226, "step": 9719 }, { "epoch": 7.1208791208791204, "grad_norm": 1.1015625, "learning_rate": 9.376405598465748e-05, "loss": 0.1466, "step": 9720 }, { "epoch": 7.121611721611721, "grad_norm": 1.0546875, "learning_rate": 9.376239105469605e-05, "loss": 0.1202, "step": 9721 }, { "epoch": 7.122344322344322, "grad_norm": 1.1328125, "learning_rate": 9.376072591745027e-05, "loss": 0.1532, "step": 9722 }, { "epoch": 7.123076923076923, "grad_norm": 1.015625, "learning_rate": 9.375906057292816e-05, "loss": 0.1149, "step": 9723 }, { "epoch": 7.123809523809523, "grad_norm": 0.61328125, "learning_rate": 9.375739502113771e-05, "loss": 0.115, "step": 9724 }, { "epoch": 7.124542124542124, "grad_norm": 1.1015625, "learning_rate": 9.375572926208685e-05, "loss": 0.1515, "step": 9725 }, { "epoch": 7.125274725274725, "grad_norm": 1.28125, "learning_rate": 9.37540632957836e-05, "loss": 0.1546, "step": 9726 }, { "epoch": 7.126007326007326, "grad_norm": 0.984375, "learning_rate": 9.375239712223594e-05, "loss": 0.156, "step": 9727 }, { "epoch": 7.126739926739926, "grad_norm": 1.609375, "learning_rate": 9.375073074145185e-05, "loss": 0.1767, "step": 9728 }, { "epoch": 7.127472527472527, "grad_norm": 1.4375, "learning_rate": 9.374906415343929e-05, "loss": 0.1682, "step": 9729 }, { "epoch": 7.128205128205128, "grad_norm": 0.93359375, "learning_rate": 9.374739735820628e-05, "loss": 0.1396, "step": 9730 }, { "epoch": 7.128937728937729, "grad_norm": 1.5234375, "learning_rate": 9.374573035576079e-05, "loss": 0.1623, "step": 9731 }, { "epoch": 7.12967032967033, "grad_norm": 1.1640625, "learning_rate": 9.374406314611082e-05, "loss": 0.1339, "step": 9732 }, { "epoch": 7.130402930402931, "grad_norm": 0.88671875, "learning_rate": 9.374239572926434e-05, "loss": 0.1246, "step": 9733 }, { "epoch": 7.131135531135532, "grad_norm": 1.03125, "learning_rate": 9.374072810522934e-05, "loss": 0.1884, "step": 9734 }, { "epoch": 7.131868131868132, "grad_norm": 1.109375, "learning_rate": 9.373906027401381e-05, "loss": 0.1213, "step": 9735 }, { "epoch": 7.132600732600733, "grad_norm": 1.859375, "learning_rate": 9.373739223562577e-05, "loss": 0.1325, "step": 9736 }, { "epoch": 7.133333333333334, "grad_norm": 0.6875, "learning_rate": 9.373572399007317e-05, "loss": 0.1061, "step": 9737 }, { "epoch": 7.1340659340659345, "grad_norm": 1.0546875, "learning_rate": 9.373405553736404e-05, "loss": 0.1491, "step": 9738 }, { "epoch": 7.134798534798535, "grad_norm": 0.72265625, "learning_rate": 9.373238687750634e-05, "loss": 0.1196, "step": 9739 }, { "epoch": 7.135531135531136, "grad_norm": 1.1328125, "learning_rate": 9.37307180105081e-05, "loss": 0.171, "step": 9740 }, { "epoch": 7.136263736263737, "grad_norm": 0.61328125, "learning_rate": 9.372904893637729e-05, "loss": 0.1081, "step": 9741 }, { "epoch": 7.136996336996337, "grad_norm": 0.859375, "learning_rate": 9.372737965512191e-05, "loss": 0.1244, "step": 9742 }, { "epoch": 7.137728937728938, "grad_norm": 1.1328125, "learning_rate": 9.372571016674997e-05, "loss": 0.1726, "step": 9743 }, { "epoch": 7.138461538461539, "grad_norm": 0.71484375, "learning_rate": 9.372404047126946e-05, "loss": 0.1383, "step": 9744 }, { "epoch": 7.13919413919414, "grad_norm": 0.91796875, "learning_rate": 9.372237056868839e-05, "loss": 0.1566, "step": 9745 }, { "epoch": 7.13992673992674, "grad_norm": 1.234375, "learning_rate": 9.372070045901474e-05, "loss": 0.133, "step": 9746 }, { "epoch": 7.140659340659341, "grad_norm": 1.2109375, "learning_rate": 9.371903014225653e-05, "loss": 0.1452, "step": 9747 }, { "epoch": 7.141391941391942, "grad_norm": 0.80859375, "learning_rate": 9.371735961842176e-05, "loss": 0.1421, "step": 9748 }, { "epoch": 7.1421245421245425, "grad_norm": 1.109375, "learning_rate": 9.371568888751843e-05, "loss": 0.1196, "step": 9749 }, { "epoch": 7.142857142857143, "grad_norm": 0.80859375, "learning_rate": 9.371401794955457e-05, "loss": 0.1159, "step": 9750 }, { "epoch": 7.143589743589744, "grad_norm": 1.140625, "learning_rate": 9.371234680453813e-05, "loss": 0.1592, "step": 9751 }, { "epoch": 7.144322344322345, "grad_norm": 0.89453125, "learning_rate": 9.371067545247718e-05, "loss": 0.1088, "step": 9752 }, { "epoch": 7.145054945054945, "grad_norm": 0.9609375, "learning_rate": 9.370900389337967e-05, "loss": 0.1406, "step": 9753 }, { "epoch": 7.145787545787546, "grad_norm": 0.7734375, "learning_rate": 9.370733212725365e-05, "loss": 0.1486, "step": 9754 }, { "epoch": 7.146520146520147, "grad_norm": 0.98046875, "learning_rate": 9.370566015410712e-05, "loss": 0.1109, "step": 9755 }, { "epoch": 7.147252747252748, "grad_norm": 1.09375, "learning_rate": 9.370398797394808e-05, "loss": 0.143, "step": 9756 }, { "epoch": 7.147985347985348, "grad_norm": 0.6875, "learning_rate": 9.370231558678457e-05, "loss": 0.157, "step": 9757 }, { "epoch": 7.148717948717949, "grad_norm": 1.4296875, "learning_rate": 9.370064299262459e-05, "loss": 0.1108, "step": 9758 }, { "epoch": 7.14945054945055, "grad_norm": 0.82421875, "learning_rate": 9.36989701914761e-05, "loss": 0.1386, "step": 9759 }, { "epoch": 7.1501831501831505, "grad_norm": 1.390625, "learning_rate": 9.369729718334721e-05, "loss": 0.1228, "step": 9760 }, { "epoch": 7.150915750915751, "grad_norm": 0.97265625, "learning_rate": 9.369562396824588e-05, "loss": 0.1469, "step": 9761 }, { "epoch": 7.151648351648352, "grad_norm": 0.9765625, "learning_rate": 9.369395054618012e-05, "loss": 0.1318, "step": 9762 }, { "epoch": 7.152380952380953, "grad_norm": 0.9296875, "learning_rate": 9.369227691715797e-05, "loss": 0.1391, "step": 9763 }, { "epoch": 7.1531135531135535, "grad_norm": 1.1171875, "learning_rate": 9.369060308118744e-05, "loss": 0.1452, "step": 9764 }, { "epoch": 7.153846153846154, "grad_norm": 0.75, "learning_rate": 9.368892903827657e-05, "loss": 0.1502, "step": 9765 }, { "epoch": 7.154578754578755, "grad_norm": 1.3984375, "learning_rate": 9.368725478843334e-05, "loss": 0.153, "step": 9766 }, { "epoch": 7.155311355311356, "grad_norm": 0.8828125, "learning_rate": 9.368558033166581e-05, "loss": 0.1596, "step": 9767 }, { "epoch": 7.156043956043956, "grad_norm": 0.73046875, "learning_rate": 9.368390566798197e-05, "loss": 0.1515, "step": 9768 }, { "epoch": 7.156776556776557, "grad_norm": 0.9296875, "learning_rate": 9.368223079738987e-05, "loss": 0.1324, "step": 9769 }, { "epoch": 7.157509157509158, "grad_norm": 0.91015625, "learning_rate": 9.368055571989753e-05, "loss": 0.1091, "step": 9770 }, { "epoch": 7.158241758241759, "grad_norm": 0.63671875, "learning_rate": 9.367888043551297e-05, "loss": 0.1193, "step": 9771 }, { "epoch": 7.158974358974359, "grad_norm": 1.1171875, "learning_rate": 9.367720494424422e-05, "loss": 0.1341, "step": 9772 }, { "epoch": 7.15970695970696, "grad_norm": 0.90625, "learning_rate": 9.36755292460993e-05, "loss": 0.1637, "step": 9773 }, { "epoch": 7.160439560439561, "grad_norm": 0.91015625, "learning_rate": 9.367385334108627e-05, "loss": 0.1605, "step": 9774 }, { "epoch": 7.1611721611721615, "grad_norm": 1.0859375, "learning_rate": 9.367217722921311e-05, "loss": 0.1362, "step": 9775 }, { "epoch": 7.161904761904762, "grad_norm": 0.92578125, "learning_rate": 9.36705009104879e-05, "loss": 0.1061, "step": 9776 }, { "epoch": 7.162637362637363, "grad_norm": 0.83203125, "learning_rate": 9.366882438491862e-05, "loss": 0.1573, "step": 9777 }, { "epoch": 7.163369963369964, "grad_norm": 0.9609375, "learning_rate": 9.366714765251335e-05, "loss": 0.1471, "step": 9778 }, { "epoch": 7.164102564102564, "grad_norm": 0.8203125, "learning_rate": 9.366547071328009e-05, "loss": 0.1214, "step": 9779 }, { "epoch": 7.164835164835165, "grad_norm": 0.92578125, "learning_rate": 9.366379356722689e-05, "loss": 0.1349, "step": 9780 }, { "epoch": 7.165567765567766, "grad_norm": 2.03125, "learning_rate": 9.366211621436181e-05, "loss": 0.152, "step": 9781 }, { "epoch": 7.166300366300367, "grad_norm": 1.390625, "learning_rate": 9.366043865469283e-05, "loss": 0.1468, "step": 9782 }, { "epoch": 7.167032967032967, "grad_norm": 1.09375, "learning_rate": 9.365876088822805e-05, "loss": 0.1211, "step": 9783 }, { "epoch": 7.167765567765568, "grad_norm": 0.9921875, "learning_rate": 9.365708291497546e-05, "loss": 0.1117, "step": 9784 }, { "epoch": 7.168498168498169, "grad_norm": 0.88671875, "learning_rate": 9.365540473494312e-05, "loss": 0.1399, "step": 9785 }, { "epoch": 7.1692307692307695, "grad_norm": 0.71484375, "learning_rate": 9.365372634813907e-05, "loss": 0.1584, "step": 9786 }, { "epoch": 7.16996336996337, "grad_norm": 0.79296875, "learning_rate": 9.365204775457135e-05, "loss": 0.1686, "step": 9787 }, { "epoch": 7.170695970695971, "grad_norm": 1.4609375, "learning_rate": 9.365036895424799e-05, "loss": 0.1739, "step": 9788 }, { "epoch": 7.171428571428572, "grad_norm": 0.64453125, "learning_rate": 9.364868994717708e-05, "loss": 0.1436, "step": 9789 }, { "epoch": 7.172161172161172, "grad_norm": 0.9921875, "learning_rate": 9.364701073336661e-05, "loss": 0.1193, "step": 9790 }, { "epoch": 7.172893772893773, "grad_norm": 1.234375, "learning_rate": 9.364533131282464e-05, "loss": 0.1193, "step": 9791 }, { "epoch": 7.173626373626374, "grad_norm": 1.125, "learning_rate": 9.364365168555925e-05, "loss": 0.1461, "step": 9792 }, { "epoch": 7.174358974358975, "grad_norm": 1.0234375, "learning_rate": 9.364197185157844e-05, "loss": 0.1494, "step": 9793 }, { "epoch": 7.175091575091575, "grad_norm": 1.171875, "learning_rate": 9.364029181089029e-05, "loss": 0.1531, "step": 9794 }, { "epoch": 7.175824175824176, "grad_norm": 1.375, "learning_rate": 9.363861156350284e-05, "loss": 0.1489, "step": 9795 }, { "epoch": 7.176556776556777, "grad_norm": 1.40625, "learning_rate": 9.363693110942414e-05, "loss": 0.1533, "step": 9796 }, { "epoch": 7.1772893772893775, "grad_norm": 0.7578125, "learning_rate": 9.363525044866224e-05, "loss": 0.1194, "step": 9797 }, { "epoch": 7.178021978021978, "grad_norm": 0.58984375, "learning_rate": 9.363356958122519e-05, "loss": 0.114, "step": 9798 }, { "epoch": 7.178754578754579, "grad_norm": 1.7421875, "learning_rate": 9.363188850712105e-05, "loss": 0.1491, "step": 9799 }, { "epoch": 7.17948717948718, "grad_norm": 1.5078125, "learning_rate": 9.363020722635787e-05, "loss": 0.1334, "step": 9800 }, { "epoch": 7.1802197802197805, "grad_norm": 1.25, "learning_rate": 9.362852573894372e-05, "loss": 0.1633, "step": 9801 }, { "epoch": 7.180952380952381, "grad_norm": 0.90234375, "learning_rate": 9.362684404488664e-05, "loss": 0.1644, "step": 9802 }, { "epoch": 7.181684981684982, "grad_norm": 0.88671875, "learning_rate": 9.362516214419469e-05, "loss": 0.1689, "step": 9803 }, { "epoch": 7.182417582417583, "grad_norm": 1.1328125, "learning_rate": 9.362348003687593e-05, "loss": 0.1416, "step": 9804 }, { "epoch": 7.183150183150183, "grad_norm": 1.625, "learning_rate": 9.362179772293842e-05, "loss": 0.1432, "step": 9805 }, { "epoch": 7.183882783882784, "grad_norm": 1.28125, "learning_rate": 9.362011520239022e-05, "loss": 0.1741, "step": 9806 }, { "epoch": 7.184615384615385, "grad_norm": 1.3828125, "learning_rate": 9.36184324752394e-05, "loss": 0.1368, "step": 9807 }, { "epoch": 7.185347985347986, "grad_norm": 1.4375, "learning_rate": 9.361674954149401e-05, "loss": 0.1234, "step": 9808 }, { "epoch": 7.186080586080586, "grad_norm": 0.8203125, "learning_rate": 9.361506640116212e-05, "loss": 0.1679, "step": 9809 }, { "epoch": 7.186813186813187, "grad_norm": 1.3046875, "learning_rate": 9.36133830542518e-05, "loss": 0.1724, "step": 9810 }, { "epoch": 7.187545787545788, "grad_norm": 0.98828125, "learning_rate": 9.361169950077112e-05, "loss": 0.1448, "step": 9811 }, { "epoch": 7.1882783882783885, "grad_norm": 0.890625, "learning_rate": 9.361001574072809e-05, "loss": 0.1373, "step": 9812 }, { "epoch": 7.189010989010989, "grad_norm": 0.8828125, "learning_rate": 9.360833177413085e-05, "loss": 0.1118, "step": 9813 }, { "epoch": 7.18974358974359, "grad_norm": 0.91796875, "learning_rate": 9.360664760098744e-05, "loss": 0.1441, "step": 9814 }, { "epoch": 7.190476190476191, "grad_norm": 1.2890625, "learning_rate": 9.360496322130595e-05, "loss": 0.1276, "step": 9815 }, { "epoch": 7.191208791208791, "grad_norm": 1.515625, "learning_rate": 9.36032786350944e-05, "loss": 0.1374, "step": 9816 }, { "epoch": 7.191941391941392, "grad_norm": 1.921875, "learning_rate": 9.360159384236092e-05, "loss": 0.1497, "step": 9817 }, { "epoch": 7.192673992673993, "grad_norm": 1.4921875, "learning_rate": 9.359990884311353e-05, "loss": 0.135, "step": 9818 }, { "epoch": 7.193406593406594, "grad_norm": 0.7578125, "learning_rate": 9.359822363736033e-05, "loss": 0.1334, "step": 9819 }, { "epoch": 7.194139194139194, "grad_norm": 0.94140625, "learning_rate": 9.35965382251094e-05, "loss": 0.1291, "step": 9820 }, { "epoch": 7.194871794871795, "grad_norm": 0.95703125, "learning_rate": 9.35948526063688e-05, "loss": 0.1698, "step": 9821 }, { "epoch": 7.195604395604396, "grad_norm": 1.0703125, "learning_rate": 9.359316678114663e-05, "loss": 0.1493, "step": 9822 }, { "epoch": 7.1963369963369965, "grad_norm": 1.34375, "learning_rate": 9.359148074945094e-05, "loss": 0.1236, "step": 9823 }, { "epoch": 7.197069597069597, "grad_norm": 1.0703125, "learning_rate": 9.358979451128982e-05, "loss": 0.1332, "step": 9824 }, { "epoch": 7.197802197802198, "grad_norm": 1.453125, "learning_rate": 9.358810806667136e-05, "loss": 0.1512, "step": 9825 }, { "epoch": 7.198534798534799, "grad_norm": 1.203125, "learning_rate": 9.358642141560361e-05, "loss": 0.1555, "step": 9826 }, { "epoch": 7.1992673992673994, "grad_norm": 1.203125, "learning_rate": 9.358473455809469e-05, "loss": 0.1206, "step": 9827 }, { "epoch": 7.2, "grad_norm": 1.0546875, "learning_rate": 9.358304749415266e-05, "loss": 0.1557, "step": 9828 }, { "epoch": 7.200732600732601, "grad_norm": 1.1875, "learning_rate": 9.358136022378559e-05, "loss": 0.1479, "step": 9829 }, { "epoch": 7.201465201465202, "grad_norm": 0.79296875, "learning_rate": 9.357967274700159e-05, "loss": 0.1315, "step": 9830 }, { "epoch": 7.202197802197802, "grad_norm": 1.0390625, "learning_rate": 9.357798506380872e-05, "loss": 0.1553, "step": 9831 }, { "epoch": 7.202930402930403, "grad_norm": 0.96484375, "learning_rate": 9.357629717421513e-05, "loss": 0.1429, "step": 9832 }, { "epoch": 7.203663003663004, "grad_norm": 1.5, "learning_rate": 9.357460907822882e-05, "loss": 0.1376, "step": 9833 }, { "epoch": 7.204395604395605, "grad_norm": 0.8828125, "learning_rate": 9.357292077585791e-05, "loss": 0.1427, "step": 9834 }, { "epoch": 7.205128205128205, "grad_norm": 1.515625, "learning_rate": 9.357123226711052e-05, "loss": 0.1583, "step": 9835 }, { "epoch": 7.205860805860806, "grad_norm": 0.79296875, "learning_rate": 9.356954355199471e-05, "loss": 0.1614, "step": 9836 }, { "epoch": 7.206593406593407, "grad_norm": 0.9375, "learning_rate": 9.356785463051857e-05, "loss": 0.1612, "step": 9837 }, { "epoch": 7.2073260073260075, "grad_norm": 0.94140625, "learning_rate": 9.356616550269021e-05, "loss": 0.1128, "step": 9838 }, { "epoch": 7.208058608058608, "grad_norm": 0.921875, "learning_rate": 9.35644761685177e-05, "loss": 0.155, "step": 9839 }, { "epoch": 7.208791208791209, "grad_norm": 0.8515625, "learning_rate": 9.356278662800916e-05, "loss": 0.1425, "step": 9840 }, { "epoch": 7.20952380952381, "grad_norm": 0.859375, "learning_rate": 9.356109688117269e-05, "loss": 0.1108, "step": 9841 }, { "epoch": 7.21025641025641, "grad_norm": 1.3359375, "learning_rate": 9.355940692801635e-05, "loss": 0.153, "step": 9842 }, { "epoch": 7.210989010989011, "grad_norm": 1.2890625, "learning_rate": 9.355771676854824e-05, "loss": 0.1237, "step": 9843 }, { "epoch": 7.211721611721612, "grad_norm": 0.84765625, "learning_rate": 9.355602640277647e-05, "loss": 0.1185, "step": 9844 }, { "epoch": 7.212454212454213, "grad_norm": 1.8671875, "learning_rate": 9.355433583070918e-05, "loss": 0.146, "step": 9845 }, { "epoch": 7.213186813186813, "grad_norm": 1.6328125, "learning_rate": 9.355264505235441e-05, "loss": 0.1378, "step": 9846 }, { "epoch": 7.213919413919414, "grad_norm": 1.2734375, "learning_rate": 9.355095406772029e-05, "loss": 0.1871, "step": 9847 }, { "epoch": 7.214652014652015, "grad_norm": 1.171875, "learning_rate": 9.354926287681492e-05, "loss": 0.1481, "step": 9848 }, { "epoch": 7.2153846153846155, "grad_norm": 1.0234375, "learning_rate": 9.35475714796464e-05, "loss": 0.1516, "step": 9849 }, { "epoch": 7.216117216117216, "grad_norm": 1.9921875, "learning_rate": 9.354587987622282e-05, "loss": 0.1564, "step": 9850 }, { "epoch": 7.216849816849817, "grad_norm": 0.94140625, "learning_rate": 9.35441880665523e-05, "loss": 0.1334, "step": 9851 }, { "epoch": 7.217582417582418, "grad_norm": 1.203125, "learning_rate": 9.354249605064296e-05, "loss": 0.1365, "step": 9852 }, { "epoch": 7.218315018315018, "grad_norm": 1.375, "learning_rate": 9.35408038285029e-05, "loss": 0.1372, "step": 9853 }, { "epoch": 7.219047619047619, "grad_norm": 0.83203125, "learning_rate": 9.353911140014018e-05, "loss": 0.1565, "step": 9854 }, { "epoch": 7.21978021978022, "grad_norm": 1.21875, "learning_rate": 9.353741876556298e-05, "loss": 0.1512, "step": 9855 }, { "epoch": 7.220512820512821, "grad_norm": 0.82421875, "learning_rate": 9.353572592477937e-05, "loss": 0.1367, "step": 9856 }, { "epoch": 7.221245421245421, "grad_norm": 0.8828125, "learning_rate": 9.353403287779747e-05, "loss": 0.142, "step": 9857 }, { "epoch": 7.221978021978022, "grad_norm": 1.46875, "learning_rate": 9.35323396246254e-05, "loss": 0.142, "step": 9858 }, { "epoch": 7.222710622710623, "grad_norm": 1.515625, "learning_rate": 9.353064616527124e-05, "loss": 0.1343, "step": 9859 }, { "epoch": 7.2234432234432235, "grad_norm": 0.921875, "learning_rate": 9.352895249974314e-05, "loss": 0.1517, "step": 9860 }, { "epoch": 7.224175824175824, "grad_norm": 1.625, "learning_rate": 9.352725862804921e-05, "loss": 0.1719, "step": 9861 }, { "epoch": 7.224908424908425, "grad_norm": 1.484375, "learning_rate": 9.352556455019755e-05, "loss": 0.1329, "step": 9862 }, { "epoch": 7.225641025641026, "grad_norm": 2.09375, "learning_rate": 9.35238702661963e-05, "loss": 0.1625, "step": 9863 }, { "epoch": 7.2263736263736265, "grad_norm": 1.1015625, "learning_rate": 9.352217577605356e-05, "loss": 0.1434, "step": 9864 }, { "epoch": 7.227106227106227, "grad_norm": 0.796875, "learning_rate": 9.352048107977744e-05, "loss": 0.121, "step": 9865 }, { "epoch": 7.227838827838828, "grad_norm": 0.80078125, "learning_rate": 9.351878617737608e-05, "loss": 0.1313, "step": 9866 }, { "epoch": 7.228571428571429, "grad_norm": 1.4765625, "learning_rate": 9.35170910688576e-05, "loss": 0.1418, "step": 9867 }, { "epoch": 7.229304029304029, "grad_norm": 1.625, "learning_rate": 9.351539575423009e-05, "loss": 0.1366, "step": 9868 }, { "epoch": 7.23003663003663, "grad_norm": 0.73828125, "learning_rate": 9.351370023350173e-05, "loss": 0.1316, "step": 9869 }, { "epoch": 7.230769230769231, "grad_norm": 1.0390625, "learning_rate": 9.35120045066806e-05, "loss": 0.1671, "step": 9870 }, { "epoch": 7.231501831501832, "grad_norm": 1.0859375, "learning_rate": 9.351030857377483e-05, "loss": 0.1863, "step": 9871 }, { "epoch": 7.232234432234432, "grad_norm": 1.1015625, "learning_rate": 9.350861243479255e-05, "loss": 0.1576, "step": 9872 }, { "epoch": 7.232967032967033, "grad_norm": 0.7578125, "learning_rate": 9.35069160897419e-05, "loss": 0.1491, "step": 9873 }, { "epoch": 7.233699633699634, "grad_norm": 0.796875, "learning_rate": 9.350521953863099e-05, "loss": 0.1001, "step": 9874 }, { "epoch": 7.2344322344322345, "grad_norm": 1.078125, "learning_rate": 9.350352278146797e-05, "loss": 0.1238, "step": 9875 }, { "epoch": 7.235164835164835, "grad_norm": 1.0546875, "learning_rate": 9.350182581826094e-05, "loss": 0.1567, "step": 9876 }, { "epoch": 7.235897435897436, "grad_norm": 0.8828125, "learning_rate": 9.350012864901804e-05, "loss": 0.1174, "step": 9877 }, { "epoch": 7.236630036630037, "grad_norm": 0.8125, "learning_rate": 9.349843127374742e-05, "loss": 0.1377, "step": 9878 }, { "epoch": 7.237362637362637, "grad_norm": 1.140625, "learning_rate": 9.34967336924572e-05, "loss": 0.1172, "step": 9879 }, { "epoch": 7.238095238095238, "grad_norm": 0.99609375, "learning_rate": 9.34950359051555e-05, "loss": 0.1528, "step": 9880 }, { "epoch": 7.238827838827839, "grad_norm": 1.4296875, "learning_rate": 9.34933379118505e-05, "loss": 0.1632, "step": 9881 }, { "epoch": 7.23956043956044, "grad_norm": 1.9140625, "learning_rate": 9.349163971255026e-05, "loss": 0.149, "step": 9882 }, { "epoch": 7.24029304029304, "grad_norm": 1.6484375, "learning_rate": 9.3489941307263e-05, "loss": 0.1516, "step": 9883 }, { "epoch": 7.241025641025641, "grad_norm": 1.2890625, "learning_rate": 9.34882426959968e-05, "loss": 0.1503, "step": 9884 }, { "epoch": 7.241758241758242, "grad_norm": 1.2421875, "learning_rate": 9.348654387875982e-05, "loss": 0.1641, "step": 9885 }, { "epoch": 7.2424908424908425, "grad_norm": 1.0703125, "learning_rate": 9.348484485556019e-05, "loss": 0.1479, "step": 9886 }, { "epoch": 7.243223443223443, "grad_norm": 1.6328125, "learning_rate": 9.348314562640606e-05, "loss": 0.2054, "step": 9887 }, { "epoch": 7.243956043956044, "grad_norm": 1.2265625, "learning_rate": 9.348144619130558e-05, "loss": 0.1395, "step": 9888 }, { "epoch": 7.244688644688645, "grad_norm": 1.03125, "learning_rate": 9.347974655026688e-05, "loss": 0.1637, "step": 9889 }, { "epoch": 7.245421245421245, "grad_norm": 0.9765625, "learning_rate": 9.347804670329809e-05, "loss": 0.1605, "step": 9890 }, { "epoch": 7.246153846153846, "grad_norm": 0.921875, "learning_rate": 9.347634665040739e-05, "loss": 0.1163, "step": 9891 }, { "epoch": 7.246886446886447, "grad_norm": 0.7890625, "learning_rate": 9.347464639160289e-05, "loss": 0.1252, "step": 9892 }, { "epoch": 7.247619047619048, "grad_norm": 0.83203125, "learning_rate": 9.347294592689275e-05, "loss": 0.1576, "step": 9893 }, { "epoch": 7.248351648351648, "grad_norm": 0.828125, "learning_rate": 9.347124525628513e-05, "loss": 0.1347, "step": 9894 }, { "epoch": 7.249084249084249, "grad_norm": 1.6015625, "learning_rate": 9.346954437978815e-05, "loss": 0.1956, "step": 9895 }, { "epoch": 7.24981684981685, "grad_norm": 1.03125, "learning_rate": 9.346784329741e-05, "loss": 0.1512, "step": 9896 }, { "epoch": 7.2505494505494505, "grad_norm": 1.15625, "learning_rate": 9.346614200915879e-05, "loss": 0.1346, "step": 9897 }, { "epoch": 7.251282051282051, "grad_norm": 1.0703125, "learning_rate": 9.346444051504271e-05, "loss": 0.1287, "step": 9898 }, { "epoch": 7.252014652014652, "grad_norm": 1.1640625, "learning_rate": 9.346273881506987e-05, "loss": 0.1189, "step": 9899 }, { "epoch": 7.252747252747253, "grad_norm": 1.5859375, "learning_rate": 9.346103690924845e-05, "loss": 0.1935, "step": 9900 }, { "epoch": 7.2534798534798535, "grad_norm": 1.640625, "learning_rate": 9.34593347975866e-05, "loss": 0.1401, "step": 9901 }, { "epoch": 7.254212454212454, "grad_norm": 1.1328125, "learning_rate": 9.345763248009247e-05, "loss": 0.1375, "step": 9902 }, { "epoch": 7.254945054945055, "grad_norm": 1.484375, "learning_rate": 9.345592995677423e-05, "loss": 0.1315, "step": 9903 }, { "epoch": 7.255677655677656, "grad_norm": 0.67578125, "learning_rate": 9.345422722764001e-05, "loss": 0.1041, "step": 9904 }, { "epoch": 7.256410256410256, "grad_norm": 0.84375, "learning_rate": 9.3452524292698e-05, "loss": 0.1309, "step": 9905 }, { "epoch": 7.257142857142857, "grad_norm": 1.0390625, "learning_rate": 9.345082115195636e-05, "loss": 0.1483, "step": 9906 }, { "epoch": 7.257875457875458, "grad_norm": 1.2734375, "learning_rate": 9.344911780542322e-05, "loss": 0.1122, "step": 9907 }, { "epoch": 7.258608058608059, "grad_norm": 1.8046875, "learning_rate": 9.344741425310674e-05, "loss": 0.1374, "step": 9908 }, { "epoch": 7.259340659340659, "grad_norm": 1.1171875, "learning_rate": 9.344571049501513e-05, "loss": 0.1503, "step": 9909 }, { "epoch": 7.26007326007326, "grad_norm": 1.1171875, "learning_rate": 9.34440065311565e-05, "loss": 0.1547, "step": 9910 }, { "epoch": 7.260805860805861, "grad_norm": 1.0859375, "learning_rate": 9.344230236153905e-05, "loss": 0.167, "step": 9911 }, { "epoch": 7.2615384615384615, "grad_norm": 1.1328125, "learning_rate": 9.344059798617093e-05, "loss": 0.1488, "step": 9912 }, { "epoch": 7.262271062271062, "grad_norm": 1.25, "learning_rate": 9.343889340506031e-05, "loss": 0.1564, "step": 9913 }, { "epoch": 7.263003663003663, "grad_norm": 1.5234375, "learning_rate": 9.343718861821536e-05, "loss": 0.1606, "step": 9914 }, { "epoch": 7.263736263736264, "grad_norm": 1.234375, "learning_rate": 9.343548362564424e-05, "loss": 0.1555, "step": 9915 }, { "epoch": 7.264468864468864, "grad_norm": 1.203125, "learning_rate": 9.343377842735512e-05, "loss": 0.1374, "step": 9916 }, { "epoch": 7.265201465201465, "grad_norm": 0.77734375, "learning_rate": 9.343207302335617e-05, "loss": 0.1496, "step": 9917 }, { "epoch": 7.265934065934066, "grad_norm": 0.90234375, "learning_rate": 9.343036741365558e-05, "loss": 0.1206, "step": 9918 }, { "epoch": 7.266666666666667, "grad_norm": 0.80859375, "learning_rate": 9.342866159826149e-05, "loss": 0.1227, "step": 9919 }, { "epoch": 7.267399267399267, "grad_norm": 0.87890625, "learning_rate": 9.342695557718209e-05, "loss": 0.1245, "step": 9920 }, { "epoch": 7.268131868131868, "grad_norm": 1.3203125, "learning_rate": 9.342524935042556e-05, "loss": 0.1341, "step": 9921 }, { "epoch": 7.268864468864469, "grad_norm": 1.0234375, "learning_rate": 9.342354291800007e-05, "loss": 0.1404, "step": 9922 }, { "epoch": 7.2695970695970695, "grad_norm": 0.9140625, "learning_rate": 9.34218362799138e-05, "loss": 0.122, "step": 9923 }, { "epoch": 7.27032967032967, "grad_norm": 0.93359375, "learning_rate": 9.342012943617491e-05, "loss": 0.1353, "step": 9924 }, { "epoch": 7.271062271062271, "grad_norm": 0.98046875, "learning_rate": 9.341842238679159e-05, "loss": 0.1096, "step": 9925 }, { "epoch": 7.271794871794872, "grad_norm": 1.0390625, "learning_rate": 9.341671513177203e-05, "loss": 0.1487, "step": 9926 }, { "epoch": 7.2725274725274724, "grad_norm": 0.71875, "learning_rate": 9.34150076711244e-05, "loss": 0.1441, "step": 9927 }, { "epoch": 7.273260073260073, "grad_norm": 0.65234375, "learning_rate": 9.341330000485688e-05, "loss": 0.1239, "step": 9928 }, { "epoch": 7.273992673992674, "grad_norm": 1.921875, "learning_rate": 9.341159213297764e-05, "loss": 0.143, "step": 9929 }, { "epoch": 7.274725274725275, "grad_norm": 0.80859375, "learning_rate": 9.340988405549488e-05, "loss": 0.1244, "step": 9930 }, { "epoch": 7.275457875457875, "grad_norm": 0.9375, "learning_rate": 9.340817577241677e-05, "loss": 0.1507, "step": 9931 }, { "epoch": 7.276190476190476, "grad_norm": 1.078125, "learning_rate": 9.340646728375152e-05, "loss": 0.1833, "step": 9932 }, { "epoch": 7.276923076923077, "grad_norm": 2.171875, "learning_rate": 9.34047585895073e-05, "loss": 0.1617, "step": 9933 }, { "epoch": 7.277655677655678, "grad_norm": 1.5234375, "learning_rate": 9.340304968969228e-05, "loss": 0.1438, "step": 9934 }, { "epoch": 7.278388278388278, "grad_norm": 2.125, "learning_rate": 9.34013405843147e-05, "loss": 0.12, "step": 9935 }, { "epoch": 7.279120879120879, "grad_norm": 1.890625, "learning_rate": 9.339963127338267e-05, "loss": 0.1448, "step": 9936 }, { "epoch": 7.27985347985348, "grad_norm": 0.9921875, "learning_rate": 9.339792175690445e-05, "loss": 0.1361, "step": 9937 }, { "epoch": 7.2805860805860805, "grad_norm": 0.53515625, "learning_rate": 9.33962120348882e-05, "loss": 0.1108, "step": 9938 }, { "epoch": 7.281318681318681, "grad_norm": 1.59375, "learning_rate": 9.339450210734211e-05, "loss": 0.1445, "step": 9939 }, { "epoch": 7.282051282051282, "grad_norm": 1.2421875, "learning_rate": 9.339279197427437e-05, "loss": 0.1538, "step": 9940 }, { "epoch": 7.282783882783883, "grad_norm": 1.2265625, "learning_rate": 9.339108163569321e-05, "loss": 0.1342, "step": 9941 }, { "epoch": 7.283516483516483, "grad_norm": 0.98046875, "learning_rate": 9.33893710916068e-05, "loss": 0.1564, "step": 9942 }, { "epoch": 7.284249084249084, "grad_norm": 0.83984375, "learning_rate": 9.338766034202328e-05, "loss": 0.1226, "step": 9943 }, { "epoch": 7.284981684981685, "grad_norm": 1.4296875, "learning_rate": 9.338594938695096e-05, "loss": 0.1793, "step": 9944 }, { "epoch": 7.285714285714286, "grad_norm": 0.73828125, "learning_rate": 9.338423822639796e-05, "loss": 0.1253, "step": 9945 }, { "epoch": 7.286446886446886, "grad_norm": 1.0390625, "learning_rate": 9.338252686037248e-05, "loss": 0.1533, "step": 9946 }, { "epoch": 7.287179487179487, "grad_norm": 1.3828125, "learning_rate": 9.338081528888275e-05, "loss": 0.1428, "step": 9947 }, { "epoch": 7.287912087912088, "grad_norm": 1.9375, "learning_rate": 9.337910351193697e-05, "loss": 0.1361, "step": 9948 }, { "epoch": 7.2886446886446885, "grad_norm": 1.625, "learning_rate": 9.33773915295433e-05, "loss": 0.1637, "step": 9949 }, { "epoch": 7.289377289377289, "grad_norm": 1.5234375, "learning_rate": 9.337567934170998e-05, "loss": 0.1711, "step": 9950 }, { "epoch": 7.29010989010989, "grad_norm": 1.0546875, "learning_rate": 9.337396694844523e-05, "loss": 0.1452, "step": 9951 }, { "epoch": 7.290842490842491, "grad_norm": 1.140625, "learning_rate": 9.33722543497572e-05, "loss": 0.1256, "step": 9952 }, { "epoch": 7.291575091575091, "grad_norm": 1.34375, "learning_rate": 9.337054154565413e-05, "loss": 0.1359, "step": 9953 }, { "epoch": 7.292307692307692, "grad_norm": 0.68359375, "learning_rate": 9.336882853614424e-05, "loss": 0.1475, "step": 9954 }, { "epoch": 7.293040293040293, "grad_norm": 0.90234375, "learning_rate": 9.33671153212357e-05, "loss": 0.1709, "step": 9955 }, { "epoch": 7.293772893772894, "grad_norm": 1.09375, "learning_rate": 9.336540190093675e-05, "loss": 0.1555, "step": 9956 }, { "epoch": 7.294505494505494, "grad_norm": 0.9375, "learning_rate": 9.336368827525558e-05, "loss": 0.1377, "step": 9957 }, { "epoch": 7.295238095238095, "grad_norm": 0.82421875, "learning_rate": 9.336197444420041e-05, "loss": 0.0965, "step": 9958 }, { "epoch": 7.295970695970696, "grad_norm": 1.0078125, "learning_rate": 9.336026040777945e-05, "loss": 0.1179, "step": 9959 }, { "epoch": 7.2967032967032965, "grad_norm": 0.78125, "learning_rate": 9.335854616600092e-05, "loss": 0.1123, "step": 9960 }, { "epoch": 7.297435897435897, "grad_norm": 0.6328125, "learning_rate": 9.335683171887302e-05, "loss": 0.1409, "step": 9961 }, { "epoch": 7.298168498168498, "grad_norm": 1.1328125, "learning_rate": 9.335511706640396e-05, "loss": 0.131, "step": 9962 }, { "epoch": 7.298901098901099, "grad_norm": 1.1171875, "learning_rate": 9.335340220860197e-05, "loss": 0.1437, "step": 9963 }, { "epoch": 7.2996336996336995, "grad_norm": 1.140625, "learning_rate": 9.335168714547528e-05, "loss": 0.1501, "step": 9964 }, { "epoch": 7.3003663003663, "grad_norm": 1.0078125, "learning_rate": 9.334997187703206e-05, "loss": 0.1288, "step": 9965 }, { "epoch": 7.301098901098901, "grad_norm": 0.96875, "learning_rate": 9.33482564032806e-05, "loss": 0.1305, "step": 9966 }, { "epoch": 7.301831501831502, "grad_norm": 1.171875, "learning_rate": 9.334654072422903e-05, "loss": 0.1411, "step": 9967 }, { "epoch": 7.302564102564102, "grad_norm": 1.09375, "learning_rate": 9.334482483988565e-05, "loss": 0.1192, "step": 9968 }, { "epoch": 7.303296703296703, "grad_norm": 1.0859375, "learning_rate": 9.334310875025864e-05, "loss": 0.1395, "step": 9969 }, { "epoch": 7.304029304029304, "grad_norm": 0.86328125, "learning_rate": 9.334139245535623e-05, "loss": 0.15, "step": 9970 }, { "epoch": 7.304761904761905, "grad_norm": 1.3359375, "learning_rate": 9.333967595518665e-05, "loss": 0.1575, "step": 9971 }, { "epoch": 7.305494505494505, "grad_norm": 0.66015625, "learning_rate": 9.333795924975811e-05, "loss": 0.107, "step": 9972 }, { "epoch": 7.306227106227106, "grad_norm": 0.6171875, "learning_rate": 9.333624233907883e-05, "loss": 0.1029, "step": 9973 }, { "epoch": 7.306959706959707, "grad_norm": 0.7734375, "learning_rate": 9.333452522315709e-05, "loss": 0.1306, "step": 9974 }, { "epoch": 7.3076923076923075, "grad_norm": 1.03125, "learning_rate": 9.333280790200106e-05, "loss": 0.1429, "step": 9975 }, { "epoch": 7.308424908424908, "grad_norm": 1.3203125, "learning_rate": 9.333109037561898e-05, "loss": 0.1413, "step": 9976 }, { "epoch": 7.309157509157509, "grad_norm": 0.73046875, "learning_rate": 9.33293726440191e-05, "loss": 0.1234, "step": 9977 }, { "epoch": 7.30989010989011, "grad_norm": 0.9296875, "learning_rate": 9.332765470720961e-05, "loss": 0.1663, "step": 9978 }, { "epoch": 7.31062271062271, "grad_norm": 1.0078125, "learning_rate": 9.332593656519879e-05, "loss": 0.1561, "step": 9979 }, { "epoch": 7.311355311355311, "grad_norm": 1.109375, "learning_rate": 9.332421821799485e-05, "loss": 0.118, "step": 9980 }, { "epoch": 7.312087912087912, "grad_norm": 1.5390625, "learning_rate": 9.332249966560601e-05, "loss": 0.1169, "step": 9981 }, { "epoch": 7.312820512820513, "grad_norm": 1.015625, "learning_rate": 9.332078090804052e-05, "loss": 0.1667, "step": 9982 }, { "epoch": 7.313553113553113, "grad_norm": 1.0234375, "learning_rate": 9.331906194530661e-05, "loss": 0.1225, "step": 9983 }, { "epoch": 7.314285714285714, "grad_norm": 1.0703125, "learning_rate": 9.331734277741252e-05, "loss": 0.1497, "step": 9984 }, { "epoch": 7.315018315018315, "grad_norm": 1.0390625, "learning_rate": 9.331562340436648e-05, "loss": 0.1134, "step": 9985 }, { "epoch": 7.3157509157509155, "grad_norm": 0.921875, "learning_rate": 9.331390382617674e-05, "loss": 0.1509, "step": 9986 }, { "epoch": 7.316483516483516, "grad_norm": 0.91015625, "learning_rate": 9.331218404285152e-05, "loss": 0.1692, "step": 9987 }, { "epoch": 7.317216117216117, "grad_norm": 1.5, "learning_rate": 9.331046405439909e-05, "loss": 0.1438, "step": 9988 }, { "epoch": 7.317948717948718, "grad_norm": 0.91015625, "learning_rate": 9.330874386082768e-05, "loss": 0.1502, "step": 9989 }, { "epoch": 7.318681318681318, "grad_norm": 1.6796875, "learning_rate": 9.33070234621455e-05, "loss": 0.1341, "step": 9990 }, { "epoch": 7.319413919413919, "grad_norm": 0.7734375, "learning_rate": 9.330530285836081e-05, "loss": 0.1468, "step": 9991 }, { "epoch": 7.32014652014652, "grad_norm": 0.828125, "learning_rate": 9.330358204948188e-05, "loss": 0.1928, "step": 9992 }, { "epoch": 7.320879120879121, "grad_norm": 1.140625, "learning_rate": 9.330186103551694e-05, "loss": 0.1474, "step": 9993 }, { "epoch": 7.321611721611721, "grad_norm": 0.9765625, "learning_rate": 9.330013981647423e-05, "loss": 0.1727, "step": 9994 }, { "epoch": 7.322344322344322, "grad_norm": 0.8515625, "learning_rate": 9.3298418392362e-05, "loss": 0.1479, "step": 9995 }, { "epoch": 7.323076923076923, "grad_norm": 0.69921875, "learning_rate": 9.32966967631885e-05, "loss": 0.1307, "step": 9996 }, { "epoch": 7.3238095238095235, "grad_norm": 1.6015625, "learning_rate": 9.329497492896197e-05, "loss": 0.1758, "step": 9997 }, { "epoch": 7.324542124542124, "grad_norm": 0.95703125, "learning_rate": 9.329325288969066e-05, "loss": 0.1553, "step": 9998 }, { "epoch": 7.325274725274725, "grad_norm": 0.90234375, "learning_rate": 9.329153064538284e-05, "loss": 0.1461, "step": 9999 }, { "epoch": 7.326007326007326, "grad_norm": 1.2734375, "learning_rate": 9.328980819604676e-05, "loss": 0.1839, "step": 10000 }, { "epoch": 7.3267399267399265, "grad_norm": 1.71875, "learning_rate": 9.328808554169064e-05, "loss": 0.1041, "step": 10001 }, { "epoch": 7.327472527472527, "grad_norm": 1.1015625, "learning_rate": 9.328636268232276e-05, "loss": 0.1441, "step": 10002 }, { "epoch": 7.328205128205128, "grad_norm": 1.1875, "learning_rate": 9.328463961795137e-05, "loss": 0.1439, "step": 10003 }, { "epoch": 7.328937728937729, "grad_norm": 0.91796875, "learning_rate": 9.328291634858473e-05, "loss": 0.1451, "step": 10004 }, { "epoch": 7.329670329670329, "grad_norm": 0.87890625, "learning_rate": 9.328119287423109e-05, "loss": 0.1266, "step": 10005 }, { "epoch": 7.33040293040293, "grad_norm": 0.90234375, "learning_rate": 9.327946919489871e-05, "loss": 0.1428, "step": 10006 }, { "epoch": 7.331135531135531, "grad_norm": 1.03125, "learning_rate": 9.327774531059585e-05, "loss": 0.1761, "step": 10007 }, { "epoch": 7.331868131868132, "grad_norm": 1.109375, "learning_rate": 9.327602122133078e-05, "loss": 0.1118, "step": 10008 }, { "epoch": 7.332600732600732, "grad_norm": 1.140625, "learning_rate": 9.327429692711174e-05, "loss": 0.1366, "step": 10009 }, { "epoch": 7.333333333333333, "grad_norm": 1.171875, "learning_rate": 9.327257242794701e-05, "loss": 0.1237, "step": 10010 }, { "epoch": 7.334065934065934, "grad_norm": 1.546875, "learning_rate": 9.327084772384484e-05, "loss": 0.1307, "step": 10011 }, { "epoch": 7.3347985347985345, "grad_norm": 0.77734375, "learning_rate": 9.326912281481349e-05, "loss": 0.1087, "step": 10012 }, { "epoch": 7.335531135531135, "grad_norm": 0.65234375, "learning_rate": 9.326739770086123e-05, "loss": 0.1231, "step": 10013 }, { "epoch": 7.336263736263736, "grad_norm": 0.87890625, "learning_rate": 9.326567238199635e-05, "loss": 0.1378, "step": 10014 }, { "epoch": 7.336996336996337, "grad_norm": 0.890625, "learning_rate": 9.326394685822707e-05, "loss": 0.1355, "step": 10015 }, { "epoch": 7.337728937728937, "grad_norm": 1.1015625, "learning_rate": 9.326222112956169e-05, "loss": 0.0993, "step": 10016 }, { "epoch": 7.338461538461538, "grad_norm": 1.4609375, "learning_rate": 9.326049519600848e-05, "loss": 0.1715, "step": 10017 }, { "epoch": 7.339194139194139, "grad_norm": 1.1796875, "learning_rate": 9.325876905757569e-05, "loss": 0.1455, "step": 10018 }, { "epoch": 7.33992673992674, "grad_norm": 0.7734375, "learning_rate": 9.32570427142716e-05, "loss": 0.1267, "step": 10019 }, { "epoch": 7.34065934065934, "grad_norm": 1.046875, "learning_rate": 9.32553161661045e-05, "loss": 0.1243, "step": 10020 }, { "epoch": 7.341391941391941, "grad_norm": 0.91796875, "learning_rate": 9.325358941308263e-05, "loss": 0.1656, "step": 10021 }, { "epoch": 7.342124542124542, "grad_norm": 0.81640625, "learning_rate": 9.325186245521428e-05, "loss": 0.1615, "step": 10022 }, { "epoch": 7.3428571428571425, "grad_norm": 0.9140625, "learning_rate": 9.325013529250771e-05, "loss": 0.115, "step": 10023 }, { "epoch": 7.343589743589743, "grad_norm": 1.5234375, "learning_rate": 9.324840792497123e-05, "loss": 0.1725, "step": 10024 }, { "epoch": 7.344322344322344, "grad_norm": 1.2734375, "learning_rate": 9.324668035261308e-05, "loss": 0.1225, "step": 10025 }, { "epoch": 7.345054945054945, "grad_norm": 0.89453125, "learning_rate": 9.324495257544156e-05, "loss": 0.1288, "step": 10026 }, { "epoch": 7.3457875457875454, "grad_norm": 0.9375, "learning_rate": 9.324322459346493e-05, "loss": 0.1531, "step": 10027 }, { "epoch": 7.346520146520146, "grad_norm": 1.140625, "learning_rate": 9.324149640669149e-05, "loss": 0.1487, "step": 10028 }, { "epoch": 7.347252747252747, "grad_norm": 0.88671875, "learning_rate": 9.32397680151295e-05, "loss": 0.1394, "step": 10029 }, { "epoch": 7.347985347985348, "grad_norm": 1.2578125, "learning_rate": 9.323803941878726e-05, "loss": 0.1696, "step": 10030 }, { "epoch": 7.348717948717948, "grad_norm": 1.046875, "learning_rate": 9.323631061767303e-05, "loss": 0.112, "step": 10031 }, { "epoch": 7.349450549450549, "grad_norm": 0.92578125, "learning_rate": 9.323458161179512e-05, "loss": 0.1537, "step": 10032 }, { "epoch": 7.35018315018315, "grad_norm": 0.5546875, "learning_rate": 9.32328524011618e-05, "loss": 0.1293, "step": 10033 }, { "epoch": 7.350915750915751, "grad_norm": 1.3671875, "learning_rate": 9.323112298578135e-05, "loss": 0.1552, "step": 10034 }, { "epoch": 7.351648351648351, "grad_norm": 1.125, "learning_rate": 9.322939336566205e-05, "loss": 0.1254, "step": 10035 }, { "epoch": 7.352380952380952, "grad_norm": 1.34375, "learning_rate": 9.322766354081221e-05, "loss": 0.1511, "step": 10036 }, { "epoch": 7.353113553113553, "grad_norm": 1.609375, "learning_rate": 9.322593351124011e-05, "loss": 0.1707, "step": 10037 }, { "epoch": 7.3538461538461535, "grad_norm": 1.0234375, "learning_rate": 9.322420327695405e-05, "loss": 0.1547, "step": 10038 }, { "epoch": 7.354578754578754, "grad_norm": 0.8828125, "learning_rate": 9.322247283796229e-05, "loss": 0.149, "step": 10039 }, { "epoch": 7.355311355311355, "grad_norm": 1.09375, "learning_rate": 9.322074219427314e-05, "loss": 0.1544, "step": 10040 }, { "epoch": 7.356043956043956, "grad_norm": 0.99609375, "learning_rate": 9.32190113458949e-05, "loss": 0.142, "step": 10041 }, { "epoch": 7.356776556776556, "grad_norm": 1.4375, "learning_rate": 9.321728029283583e-05, "loss": 0.1296, "step": 10042 }, { "epoch": 7.357509157509157, "grad_norm": 0.75, "learning_rate": 9.321554903510427e-05, "loss": 0.1243, "step": 10043 }, { "epoch": 7.358241758241758, "grad_norm": 1.1015625, "learning_rate": 9.321381757270848e-05, "loss": 0.1265, "step": 10044 }, { "epoch": 7.358974358974359, "grad_norm": 0.90234375, "learning_rate": 9.321208590565676e-05, "loss": 0.1621, "step": 10045 }, { "epoch": 7.359706959706959, "grad_norm": 1.2578125, "learning_rate": 9.321035403395743e-05, "loss": 0.1356, "step": 10046 }, { "epoch": 7.36043956043956, "grad_norm": 0.96875, "learning_rate": 9.320862195761877e-05, "loss": 0.1204, "step": 10047 }, { "epoch": 7.361172161172161, "grad_norm": 0.95703125, "learning_rate": 9.320688967664907e-05, "loss": 0.1259, "step": 10048 }, { "epoch": 7.3619047619047615, "grad_norm": 0.78515625, "learning_rate": 9.320515719105665e-05, "loss": 0.1464, "step": 10049 }, { "epoch": 7.362637362637362, "grad_norm": 0.828125, "learning_rate": 9.320342450084982e-05, "loss": 0.1497, "step": 10050 }, { "epoch": 7.363369963369963, "grad_norm": 1.3203125, "learning_rate": 9.320169160603685e-05, "loss": 0.176, "step": 10051 }, { "epoch": 7.364102564102564, "grad_norm": 1.171875, "learning_rate": 9.319995850662605e-05, "loss": 0.1921, "step": 10052 }, { "epoch": 7.364835164835164, "grad_norm": 1.078125, "learning_rate": 9.319822520262574e-05, "loss": 0.1459, "step": 10053 }, { "epoch": 7.365567765567765, "grad_norm": 1.703125, "learning_rate": 9.319649169404421e-05, "loss": 0.1745, "step": 10054 }, { "epoch": 7.366300366300366, "grad_norm": 1.578125, "learning_rate": 9.31947579808898e-05, "loss": 0.1446, "step": 10055 }, { "epoch": 7.367032967032967, "grad_norm": 0.70703125, "learning_rate": 9.319302406317074e-05, "loss": 0.1503, "step": 10056 }, { "epoch": 7.367765567765568, "grad_norm": 0.80859375, "learning_rate": 9.319128994089543e-05, "loss": 0.1361, "step": 10057 }, { "epoch": 7.368498168498169, "grad_norm": 0.828125, "learning_rate": 9.31895556140721e-05, "loss": 0.1543, "step": 10058 }, { "epoch": 7.36923076923077, "grad_norm": 1.2421875, "learning_rate": 9.318782108270911e-05, "loss": 0.1596, "step": 10059 }, { "epoch": 7.36996336996337, "grad_norm": 1.2109375, "learning_rate": 9.318608634681477e-05, "loss": 0.1045, "step": 10060 }, { "epoch": 7.370695970695971, "grad_norm": 1.5, "learning_rate": 9.318435140639737e-05, "loss": 0.1286, "step": 10061 }, { "epoch": 7.371428571428572, "grad_norm": 0.89453125, "learning_rate": 9.318261626146522e-05, "loss": 0.1327, "step": 10062 }, { "epoch": 7.372161172161173, "grad_norm": 0.9140625, "learning_rate": 9.318088091202665e-05, "loss": 0.1549, "step": 10063 }, { "epoch": 7.372893772893773, "grad_norm": 0.98828125, "learning_rate": 9.317914535808995e-05, "loss": 0.1259, "step": 10064 }, { "epoch": 7.373626373626374, "grad_norm": 1.28125, "learning_rate": 9.317740959966349e-05, "loss": 0.1442, "step": 10065 }, { "epoch": 7.374358974358975, "grad_norm": 0.90625, "learning_rate": 9.317567363675554e-05, "loss": 0.1663, "step": 10066 }, { "epoch": 7.3750915750915755, "grad_norm": 0.953125, "learning_rate": 9.31739374693744e-05, "loss": 0.113, "step": 10067 }, { "epoch": 7.375824175824176, "grad_norm": 1.1484375, "learning_rate": 9.317220109752844e-05, "loss": 0.1537, "step": 10068 }, { "epoch": 7.376556776556777, "grad_norm": 1.265625, "learning_rate": 9.317046452122596e-05, "loss": 0.152, "step": 10069 }, { "epoch": 7.377289377289378, "grad_norm": 0.9140625, "learning_rate": 9.316872774047526e-05, "loss": 0.1691, "step": 10070 }, { "epoch": 7.3780219780219785, "grad_norm": 1.3046875, "learning_rate": 9.31669907552847e-05, "loss": 0.166, "step": 10071 }, { "epoch": 7.378754578754579, "grad_norm": 0.75390625, "learning_rate": 9.316525356566255e-05, "loss": 0.1549, "step": 10072 }, { "epoch": 7.37948717948718, "grad_norm": 0.84375, "learning_rate": 9.316351617161718e-05, "loss": 0.1396, "step": 10073 }, { "epoch": 7.380219780219781, "grad_norm": 1.40625, "learning_rate": 9.316177857315691e-05, "loss": 0.173, "step": 10074 }, { "epoch": 7.380952380952381, "grad_norm": 0.87109375, "learning_rate": 9.316004077029004e-05, "loss": 0.1339, "step": 10075 }, { "epoch": 7.381684981684982, "grad_norm": 0.9609375, "learning_rate": 9.315830276302492e-05, "loss": 0.1426, "step": 10076 }, { "epoch": 7.382417582417583, "grad_norm": 1.078125, "learning_rate": 9.315656455136985e-05, "loss": 0.1608, "step": 10077 }, { "epoch": 7.383150183150184, "grad_norm": 1.0234375, "learning_rate": 9.315482613533321e-05, "loss": 0.1554, "step": 10078 }, { "epoch": 7.383882783882784, "grad_norm": 1.0703125, "learning_rate": 9.315308751492327e-05, "loss": 0.1289, "step": 10079 }, { "epoch": 7.384615384615385, "grad_norm": 1.0234375, "learning_rate": 9.315134869014836e-05, "loss": 0.1171, "step": 10080 }, { "epoch": 7.385347985347986, "grad_norm": 0.96875, "learning_rate": 9.314960966101689e-05, "loss": 0.1307, "step": 10081 }, { "epoch": 7.3860805860805865, "grad_norm": 1.25, "learning_rate": 9.31478704275371e-05, "loss": 0.1713, "step": 10082 }, { "epoch": 7.386813186813187, "grad_norm": 1.0078125, "learning_rate": 9.31461309897174e-05, "loss": 0.1769, "step": 10083 }, { "epoch": 7.387545787545788, "grad_norm": 0.93359375, "learning_rate": 9.314439134756605e-05, "loss": 0.1203, "step": 10084 }, { "epoch": 7.388278388278389, "grad_norm": 0.953125, "learning_rate": 9.314265150109142e-05, "loss": 0.133, "step": 10085 }, { "epoch": 7.389010989010989, "grad_norm": 0.81640625, "learning_rate": 9.314091145030188e-05, "loss": 0.1441, "step": 10086 }, { "epoch": 7.38974358974359, "grad_norm": 0.9453125, "learning_rate": 9.313917119520572e-05, "loss": 0.1188, "step": 10087 }, { "epoch": 7.390476190476191, "grad_norm": 1.453125, "learning_rate": 9.313743073581129e-05, "loss": 0.1293, "step": 10088 }, { "epoch": 7.391208791208792, "grad_norm": 1.5625, "learning_rate": 9.313569007212695e-05, "loss": 0.1141, "step": 10089 }, { "epoch": 7.391941391941392, "grad_norm": 1.3125, "learning_rate": 9.313394920416098e-05, "loss": 0.1146, "step": 10090 }, { "epoch": 7.392673992673993, "grad_norm": 1.3359375, "learning_rate": 9.31322081319218e-05, "loss": 0.122, "step": 10091 }, { "epoch": 7.393406593406594, "grad_norm": 1.4375, "learning_rate": 9.313046685541771e-05, "loss": 0.1353, "step": 10092 }, { "epoch": 7.3941391941391945, "grad_norm": 0.76953125, "learning_rate": 9.312872537465706e-05, "loss": 0.1141, "step": 10093 }, { "epoch": 7.394871794871795, "grad_norm": 0.91015625, "learning_rate": 9.312698368964817e-05, "loss": 0.1345, "step": 10094 }, { "epoch": 7.395604395604396, "grad_norm": 1.3046875, "learning_rate": 9.312524180039945e-05, "loss": 0.1309, "step": 10095 }, { "epoch": 7.396336996336997, "grad_norm": 0.72265625, "learning_rate": 9.312349970691918e-05, "loss": 0.102, "step": 10096 }, { "epoch": 7.397069597069597, "grad_norm": 1.0546875, "learning_rate": 9.312175740921573e-05, "loss": 0.1108, "step": 10097 }, { "epoch": 7.397802197802198, "grad_norm": 0.7890625, "learning_rate": 9.312001490729743e-05, "loss": 0.1566, "step": 10098 }, { "epoch": 7.398534798534799, "grad_norm": 0.83203125, "learning_rate": 9.311827220117268e-05, "loss": 0.1838, "step": 10099 }, { "epoch": 7.3992673992674, "grad_norm": 1.2734375, "learning_rate": 9.311652929084978e-05, "loss": 0.1604, "step": 10100 }, { "epoch": 7.4, "grad_norm": 0.875, "learning_rate": 9.31147861763371e-05, "loss": 0.1286, "step": 10101 }, { "epoch": 7.400732600732601, "grad_norm": 1.3671875, "learning_rate": 9.3113042857643e-05, "loss": 0.1357, "step": 10102 }, { "epoch": 7.401465201465202, "grad_norm": 1.078125, "learning_rate": 9.311129933477581e-05, "loss": 0.129, "step": 10103 }, { "epoch": 7.4021978021978025, "grad_norm": 1.0, "learning_rate": 9.310955560774389e-05, "loss": 0.1433, "step": 10104 }, { "epoch": 7.402930402930403, "grad_norm": 1.203125, "learning_rate": 9.310781167655563e-05, "loss": 0.1401, "step": 10105 }, { "epoch": 7.403663003663004, "grad_norm": 1.453125, "learning_rate": 9.310606754121932e-05, "loss": 0.17, "step": 10106 }, { "epoch": 7.404395604395605, "grad_norm": 1.1640625, "learning_rate": 9.310432320174336e-05, "loss": 0.148, "step": 10107 }, { "epoch": 7.4051282051282055, "grad_norm": 0.953125, "learning_rate": 9.310257865813613e-05, "loss": 0.1281, "step": 10108 }, { "epoch": 7.405860805860806, "grad_norm": 0.88671875, "learning_rate": 9.310083391040594e-05, "loss": 0.1356, "step": 10109 }, { "epoch": 7.406593406593407, "grad_norm": 0.86328125, "learning_rate": 9.309908895856118e-05, "loss": 0.1561, "step": 10110 }, { "epoch": 7.407326007326008, "grad_norm": 0.9921875, "learning_rate": 9.309734380261018e-05, "loss": 0.1414, "step": 10111 }, { "epoch": 7.408058608058608, "grad_norm": 1.234375, "learning_rate": 9.309559844256134e-05, "loss": 0.1532, "step": 10112 }, { "epoch": 7.408791208791209, "grad_norm": 0.71484375, "learning_rate": 9.309385287842301e-05, "loss": 0.1035, "step": 10113 }, { "epoch": 7.40952380952381, "grad_norm": 0.90234375, "learning_rate": 9.309210711020353e-05, "loss": 0.1325, "step": 10114 }, { "epoch": 7.410256410256411, "grad_norm": 1.03125, "learning_rate": 9.309036113791129e-05, "loss": 0.1298, "step": 10115 }, { "epoch": 7.410989010989011, "grad_norm": 0.9921875, "learning_rate": 9.308861496155465e-05, "loss": 0.1426, "step": 10116 }, { "epoch": 7.411721611721612, "grad_norm": 1.03125, "learning_rate": 9.308686858114197e-05, "loss": 0.1489, "step": 10117 }, { "epoch": 7.412454212454213, "grad_norm": 1.21875, "learning_rate": 9.308512199668163e-05, "loss": 0.1319, "step": 10118 }, { "epoch": 7.4131868131868135, "grad_norm": 1.7421875, "learning_rate": 9.308337520818199e-05, "loss": 0.2223, "step": 10119 }, { "epoch": 7.413919413919414, "grad_norm": 0.92578125, "learning_rate": 9.30816282156514e-05, "loss": 0.1204, "step": 10120 }, { "epoch": 7.414652014652015, "grad_norm": 0.73046875, "learning_rate": 9.307988101909829e-05, "loss": 0.1338, "step": 10121 }, { "epoch": 7.415384615384616, "grad_norm": 1.046875, "learning_rate": 9.307813361853097e-05, "loss": 0.1466, "step": 10122 }, { "epoch": 7.416117216117216, "grad_norm": 1.578125, "learning_rate": 9.307638601395783e-05, "loss": 0.152, "step": 10123 }, { "epoch": 7.416849816849817, "grad_norm": 0.93359375, "learning_rate": 9.307463820538726e-05, "loss": 0.1178, "step": 10124 }, { "epoch": 7.417582417582418, "grad_norm": 1.015625, "learning_rate": 9.30728901928276e-05, "loss": 0.1591, "step": 10125 }, { "epoch": 7.418315018315019, "grad_norm": 1.0703125, "learning_rate": 9.307114197628727e-05, "loss": 0.1241, "step": 10126 }, { "epoch": 7.419047619047619, "grad_norm": 1.7265625, "learning_rate": 9.306939355577463e-05, "loss": 0.1629, "step": 10127 }, { "epoch": 7.41978021978022, "grad_norm": 1.078125, "learning_rate": 9.306764493129803e-05, "loss": 0.1313, "step": 10128 }, { "epoch": 7.420512820512821, "grad_norm": 0.56640625, "learning_rate": 9.306589610286586e-05, "loss": 0.1258, "step": 10129 }, { "epoch": 7.4212454212454215, "grad_norm": 1.3515625, "learning_rate": 9.306414707048652e-05, "loss": 0.1559, "step": 10130 }, { "epoch": 7.421978021978022, "grad_norm": 1.6796875, "learning_rate": 9.306239783416838e-05, "loss": 0.1471, "step": 10131 }, { "epoch": 7.422710622710623, "grad_norm": 0.71875, "learning_rate": 9.306064839391983e-05, "loss": 0.1369, "step": 10132 }, { "epoch": 7.423443223443224, "grad_norm": 0.73046875, "learning_rate": 9.305889874974923e-05, "loss": 0.1143, "step": 10133 }, { "epoch": 7.4241758241758244, "grad_norm": 1.28125, "learning_rate": 9.305714890166498e-05, "loss": 0.156, "step": 10134 }, { "epoch": 7.424908424908425, "grad_norm": 0.80859375, "learning_rate": 9.305539884967545e-05, "loss": 0.1112, "step": 10135 }, { "epoch": 7.425641025641026, "grad_norm": 0.83984375, "learning_rate": 9.305364859378906e-05, "loss": 0.1364, "step": 10136 }, { "epoch": 7.426373626373627, "grad_norm": 0.91796875, "learning_rate": 9.305189813401414e-05, "loss": 0.1532, "step": 10137 }, { "epoch": 7.427106227106227, "grad_norm": 0.8984375, "learning_rate": 9.305014747035912e-05, "loss": 0.121, "step": 10138 }, { "epoch": 7.427838827838828, "grad_norm": 0.83203125, "learning_rate": 9.304839660283238e-05, "loss": 0.1304, "step": 10139 }, { "epoch": 7.428571428571429, "grad_norm": 1.0234375, "learning_rate": 9.30466455314423e-05, "loss": 0.1535, "step": 10140 }, { "epoch": 7.4293040293040296, "grad_norm": 1.1328125, "learning_rate": 9.304489425619726e-05, "loss": 0.1302, "step": 10141 }, { "epoch": 7.43003663003663, "grad_norm": 0.58203125, "learning_rate": 9.304314277710567e-05, "loss": 0.1239, "step": 10142 }, { "epoch": 7.430769230769231, "grad_norm": 1.1953125, "learning_rate": 9.304139109417593e-05, "loss": 0.1426, "step": 10143 }, { "epoch": 7.431501831501832, "grad_norm": 0.83984375, "learning_rate": 9.303963920741642e-05, "loss": 0.1244, "step": 10144 }, { "epoch": 7.4322344322344325, "grad_norm": 1.0546875, "learning_rate": 9.303788711683553e-05, "loss": 0.1182, "step": 10145 }, { "epoch": 7.432967032967033, "grad_norm": 1.109375, "learning_rate": 9.303613482244164e-05, "loss": 0.1474, "step": 10146 }, { "epoch": 7.433699633699634, "grad_norm": 0.85546875, "learning_rate": 9.303438232424319e-05, "loss": 0.1444, "step": 10147 }, { "epoch": 7.434432234432235, "grad_norm": 0.9375, "learning_rate": 9.303262962224854e-05, "loss": 0.1444, "step": 10148 }, { "epoch": 7.435164835164835, "grad_norm": 1.25, "learning_rate": 9.30308767164661e-05, "loss": 0.1632, "step": 10149 }, { "epoch": 7.435897435897436, "grad_norm": 0.57421875, "learning_rate": 9.302912360690427e-05, "loss": 0.1305, "step": 10150 }, { "epoch": 7.436630036630037, "grad_norm": 1.65625, "learning_rate": 9.302737029357144e-05, "loss": 0.1458, "step": 10151 }, { "epoch": 7.437362637362638, "grad_norm": 0.93359375, "learning_rate": 9.302561677647603e-05, "loss": 0.1146, "step": 10152 }, { "epoch": 7.438095238095238, "grad_norm": 1.71875, "learning_rate": 9.302386305562643e-05, "loss": 0.1657, "step": 10153 }, { "epoch": 7.438827838827839, "grad_norm": 0.8359375, "learning_rate": 9.302210913103103e-05, "loss": 0.1335, "step": 10154 }, { "epoch": 7.43956043956044, "grad_norm": 0.82421875, "learning_rate": 9.302035500269826e-05, "loss": 0.1284, "step": 10155 }, { "epoch": 7.4402930402930405, "grad_norm": 0.875, "learning_rate": 9.30186006706365e-05, "loss": 0.1228, "step": 10156 }, { "epoch": 7.441025641025641, "grad_norm": 0.79296875, "learning_rate": 9.301684613485416e-05, "loss": 0.1345, "step": 10157 }, { "epoch": 7.441758241758242, "grad_norm": 1.03125, "learning_rate": 9.301509139535967e-05, "loss": 0.1488, "step": 10158 }, { "epoch": 7.442490842490843, "grad_norm": 0.83203125, "learning_rate": 9.30133364521614e-05, "loss": 0.1457, "step": 10159 }, { "epoch": 7.443223443223443, "grad_norm": 0.6796875, "learning_rate": 9.301158130526779e-05, "loss": 0.1358, "step": 10160 }, { "epoch": 7.443956043956044, "grad_norm": 1.3125, "learning_rate": 9.300982595468726e-05, "loss": 0.1408, "step": 10161 }, { "epoch": 7.444688644688645, "grad_norm": 1.140625, "learning_rate": 9.300807040042818e-05, "loss": 0.1254, "step": 10162 }, { "epoch": 7.445421245421246, "grad_norm": 0.6953125, "learning_rate": 9.300631464249897e-05, "loss": 0.1239, "step": 10163 }, { "epoch": 7.446153846153846, "grad_norm": 1.0, "learning_rate": 9.300455868090805e-05, "loss": 0.1571, "step": 10164 }, { "epoch": 7.446886446886447, "grad_norm": 0.89453125, "learning_rate": 9.300280251566387e-05, "loss": 0.1302, "step": 10165 }, { "epoch": 7.447619047619048, "grad_norm": 1.0078125, "learning_rate": 9.300104614677477e-05, "loss": 0.1496, "step": 10166 }, { "epoch": 7.4483516483516485, "grad_norm": 1.1796875, "learning_rate": 9.299928957424923e-05, "loss": 0.1641, "step": 10167 }, { "epoch": 7.449084249084249, "grad_norm": 1.3125, "learning_rate": 9.299753279809563e-05, "loss": 0.1815, "step": 10168 }, { "epoch": 7.44981684981685, "grad_norm": 0.6953125, "learning_rate": 9.29957758183224e-05, "loss": 0.1265, "step": 10169 }, { "epoch": 7.450549450549451, "grad_norm": 0.8046875, "learning_rate": 9.299401863493798e-05, "loss": 0.1342, "step": 10170 }, { "epoch": 7.4512820512820515, "grad_norm": 1.2734375, "learning_rate": 9.299226124795074e-05, "loss": 0.1345, "step": 10171 }, { "epoch": 7.452014652014652, "grad_norm": 1.3515625, "learning_rate": 9.299050365736914e-05, "loss": 0.168, "step": 10172 }, { "epoch": 7.452747252747253, "grad_norm": 1.203125, "learning_rate": 9.298874586320158e-05, "loss": 0.1422, "step": 10173 }, { "epoch": 7.453479853479854, "grad_norm": 0.78515625, "learning_rate": 9.298698786545649e-05, "loss": 0.1306, "step": 10174 }, { "epoch": 7.454212454212454, "grad_norm": 0.73828125, "learning_rate": 9.29852296641423e-05, "loss": 0.1431, "step": 10175 }, { "epoch": 7.454945054945055, "grad_norm": 0.80859375, "learning_rate": 9.298347125926742e-05, "loss": 0.1153, "step": 10176 }, { "epoch": 7.455677655677656, "grad_norm": 0.984375, "learning_rate": 9.298171265084029e-05, "loss": 0.1298, "step": 10177 }, { "epoch": 7.456410256410257, "grad_norm": 1.890625, "learning_rate": 9.297995383886933e-05, "loss": 0.1646, "step": 10178 }, { "epoch": 7.457142857142857, "grad_norm": 1.015625, "learning_rate": 9.297819482336296e-05, "loss": 0.1232, "step": 10179 }, { "epoch": 7.457875457875458, "grad_norm": 0.7265625, "learning_rate": 9.297643560432961e-05, "loss": 0.135, "step": 10180 }, { "epoch": 7.458608058608059, "grad_norm": 1.8046875, "learning_rate": 9.297467618177773e-05, "loss": 0.1754, "step": 10181 }, { "epoch": 7.4593406593406595, "grad_norm": 0.8046875, "learning_rate": 9.297291655571573e-05, "loss": 0.1215, "step": 10182 }, { "epoch": 7.46007326007326, "grad_norm": 1.6171875, "learning_rate": 9.297115672615204e-05, "loss": 0.1337, "step": 10183 }, { "epoch": 7.460805860805861, "grad_norm": 1.1796875, "learning_rate": 9.296939669309509e-05, "loss": 0.1109, "step": 10184 }, { "epoch": 7.461538461538462, "grad_norm": 1.296875, "learning_rate": 9.29676364565533e-05, "loss": 0.1479, "step": 10185 }, { "epoch": 7.462271062271062, "grad_norm": 1.1484375, "learning_rate": 9.296587601653517e-05, "loss": 0.1694, "step": 10186 }, { "epoch": 7.463003663003663, "grad_norm": 1.765625, "learning_rate": 9.296411537304906e-05, "loss": 0.1396, "step": 10187 }, { "epoch": 7.463736263736264, "grad_norm": 2.28125, "learning_rate": 9.296235452610343e-05, "loss": 0.1975, "step": 10188 }, { "epoch": 7.464468864468865, "grad_norm": 1.46875, "learning_rate": 9.296059347570672e-05, "loss": 0.1237, "step": 10189 }, { "epoch": 7.465201465201465, "grad_norm": 0.90234375, "learning_rate": 9.295883222186739e-05, "loss": 0.1531, "step": 10190 }, { "epoch": 7.465934065934066, "grad_norm": 0.92578125, "learning_rate": 9.295707076459382e-05, "loss": 0.1322, "step": 10191 }, { "epoch": 7.466666666666667, "grad_norm": 0.94140625, "learning_rate": 9.29553091038945e-05, "loss": 0.1428, "step": 10192 }, { "epoch": 7.4673992673992675, "grad_norm": 0.6640625, "learning_rate": 9.295354723977786e-05, "loss": 0.1363, "step": 10193 }, { "epoch": 7.468131868131868, "grad_norm": 1.109375, "learning_rate": 9.295178517225234e-05, "loss": 0.1247, "step": 10194 }, { "epoch": 7.468864468864469, "grad_norm": 1.4375, "learning_rate": 9.29500229013264e-05, "loss": 0.1821, "step": 10195 }, { "epoch": 7.46959706959707, "grad_norm": 1.15625, "learning_rate": 9.294826042700844e-05, "loss": 0.1622, "step": 10196 }, { "epoch": 7.47032967032967, "grad_norm": 1.1796875, "learning_rate": 9.294649774930694e-05, "loss": 0.1212, "step": 10197 }, { "epoch": 7.471062271062271, "grad_norm": 0.8359375, "learning_rate": 9.294473486823034e-05, "loss": 0.1531, "step": 10198 }, { "epoch": 7.471794871794872, "grad_norm": 1.078125, "learning_rate": 9.294297178378707e-05, "loss": 0.1123, "step": 10199 }, { "epoch": 7.472527472527473, "grad_norm": 0.7578125, "learning_rate": 9.294120849598558e-05, "loss": 0.1303, "step": 10200 }, { "epoch": 7.473260073260073, "grad_norm": 1.0, "learning_rate": 9.293944500483433e-05, "loss": 0.1733, "step": 10201 }, { "epoch": 7.473992673992674, "grad_norm": 1.2578125, "learning_rate": 9.293768131034178e-05, "loss": 0.1384, "step": 10202 }, { "epoch": 7.474725274725275, "grad_norm": 1.3359375, "learning_rate": 9.293591741251636e-05, "loss": 0.1558, "step": 10203 }, { "epoch": 7.4754578754578755, "grad_norm": 1.109375, "learning_rate": 9.293415331136653e-05, "loss": 0.1596, "step": 10204 }, { "epoch": 7.476190476190476, "grad_norm": 0.98046875, "learning_rate": 9.293238900690074e-05, "loss": 0.1436, "step": 10205 }, { "epoch": 7.476923076923077, "grad_norm": 0.84375, "learning_rate": 9.293062449912747e-05, "loss": 0.1339, "step": 10206 }, { "epoch": 7.477655677655678, "grad_norm": 1.0390625, "learning_rate": 9.292885978805511e-05, "loss": 0.135, "step": 10207 }, { "epoch": 7.4783882783882785, "grad_norm": 1.6640625, "learning_rate": 9.29270948736922e-05, "loss": 0.1338, "step": 10208 }, { "epoch": 7.479120879120879, "grad_norm": 1.2734375, "learning_rate": 9.29253297560471e-05, "loss": 0.1595, "step": 10209 }, { "epoch": 7.47985347985348, "grad_norm": 1.09375, "learning_rate": 9.292356443512835e-05, "loss": 0.1242, "step": 10210 }, { "epoch": 7.480586080586081, "grad_norm": 1.421875, "learning_rate": 9.292179891094434e-05, "loss": 0.1282, "step": 10211 }, { "epoch": 7.481318681318681, "grad_norm": 1.9296875, "learning_rate": 9.29200331835036e-05, "loss": 0.132, "step": 10212 }, { "epoch": 7.482051282051282, "grad_norm": 1.0, "learning_rate": 9.291826725281456e-05, "loss": 0.157, "step": 10213 }, { "epoch": 7.482783882783883, "grad_norm": 1.0859375, "learning_rate": 9.291650111888566e-05, "loss": 0.1549, "step": 10214 }, { "epoch": 7.483516483516484, "grad_norm": 1.296875, "learning_rate": 9.291473478172537e-05, "loss": 0.1383, "step": 10215 }, { "epoch": 7.484249084249084, "grad_norm": 1.2734375, "learning_rate": 9.291296824134219e-05, "loss": 0.1722, "step": 10216 }, { "epoch": 7.484981684981685, "grad_norm": 1.2265625, "learning_rate": 9.291120149774453e-05, "loss": 0.1135, "step": 10217 }, { "epoch": 7.485714285714286, "grad_norm": 1.1484375, "learning_rate": 9.290943455094089e-05, "loss": 0.1609, "step": 10218 }, { "epoch": 7.4864468864468865, "grad_norm": 1.140625, "learning_rate": 9.290766740093974e-05, "loss": 0.1183, "step": 10219 }, { "epoch": 7.487179487179487, "grad_norm": 1.2734375, "learning_rate": 9.290590004774951e-05, "loss": 0.1819, "step": 10220 }, { "epoch": 7.487912087912088, "grad_norm": 1.7890625, "learning_rate": 9.290413249137871e-05, "loss": 0.1728, "step": 10221 }, { "epoch": 7.488644688644689, "grad_norm": 1.4609375, "learning_rate": 9.290236473183578e-05, "loss": 0.1536, "step": 10222 }, { "epoch": 7.489377289377289, "grad_norm": 1.6796875, "learning_rate": 9.29005967691292e-05, "loss": 0.1365, "step": 10223 }, { "epoch": 7.49010989010989, "grad_norm": 1.2265625, "learning_rate": 9.289882860326747e-05, "loss": 0.1424, "step": 10224 }, { "epoch": 7.490842490842491, "grad_norm": 1.140625, "learning_rate": 9.2897060234259e-05, "loss": 0.1245, "step": 10225 }, { "epoch": 7.491575091575092, "grad_norm": 1.1171875, "learning_rate": 9.28952916621123e-05, "loss": 0.1615, "step": 10226 }, { "epoch": 7.492307692307692, "grad_norm": 1.2265625, "learning_rate": 9.289352288683585e-05, "loss": 0.1688, "step": 10227 }, { "epoch": 7.493040293040293, "grad_norm": 1.9609375, "learning_rate": 9.289175390843812e-05, "loss": 0.1559, "step": 10228 }, { "epoch": 7.493772893772894, "grad_norm": 1.2109375, "learning_rate": 9.288998472692757e-05, "loss": 0.1424, "step": 10229 }, { "epoch": 7.4945054945054945, "grad_norm": 1.03125, "learning_rate": 9.288821534231269e-05, "loss": 0.1031, "step": 10230 }, { "epoch": 7.495238095238095, "grad_norm": 1.015625, "learning_rate": 9.288644575460195e-05, "loss": 0.1241, "step": 10231 }, { "epoch": 7.495970695970696, "grad_norm": 0.703125, "learning_rate": 9.288467596380385e-05, "loss": 0.1437, "step": 10232 }, { "epoch": 7.496703296703297, "grad_norm": 1.8125, "learning_rate": 9.288290596992684e-05, "loss": 0.1333, "step": 10233 }, { "epoch": 7.4974358974358974, "grad_norm": 1.09375, "learning_rate": 9.288113577297941e-05, "loss": 0.1395, "step": 10234 }, { "epoch": 7.498168498168498, "grad_norm": 1.28125, "learning_rate": 9.287936537297005e-05, "loss": 0.1335, "step": 10235 }, { "epoch": 7.498901098901099, "grad_norm": 1.03125, "learning_rate": 9.287759476990725e-05, "loss": 0.1698, "step": 10236 }, { "epoch": 7.4996336996337, "grad_norm": 1.484375, "learning_rate": 9.287582396379946e-05, "loss": 0.1412, "step": 10237 }, { "epoch": 7.5003663003663, "grad_norm": 0.91015625, "learning_rate": 9.28740529546552e-05, "loss": 0.1085, "step": 10238 }, { "epoch": 7.501098901098901, "grad_norm": 0.625, "learning_rate": 9.287228174248294e-05, "loss": 0.1219, "step": 10239 }, { "epoch": 7.501831501831502, "grad_norm": 1.8046875, "learning_rate": 9.287051032729118e-05, "loss": 0.1434, "step": 10240 }, { "epoch": 7.5025641025641026, "grad_norm": 2.21875, "learning_rate": 9.286873870908838e-05, "loss": 0.1613, "step": 10241 }, { "epoch": 7.503296703296703, "grad_norm": 1.1953125, "learning_rate": 9.286696688788306e-05, "loss": 0.1389, "step": 10242 }, { "epoch": 7.504029304029304, "grad_norm": 1.375, "learning_rate": 9.286519486368367e-05, "loss": 0.1854, "step": 10243 }, { "epoch": 7.504761904761905, "grad_norm": 0.63671875, "learning_rate": 9.286342263649874e-05, "loss": 0.1275, "step": 10244 }, { "epoch": 7.5054945054945055, "grad_norm": 1.328125, "learning_rate": 9.286165020633673e-05, "loss": 0.1457, "step": 10245 }, { "epoch": 7.506227106227106, "grad_norm": 0.890625, "learning_rate": 9.285987757320616e-05, "loss": 0.1087, "step": 10246 }, { "epoch": 7.506959706959707, "grad_norm": 1.265625, "learning_rate": 9.285810473711552e-05, "loss": 0.126, "step": 10247 }, { "epoch": 7.507692307692308, "grad_norm": 0.96484375, "learning_rate": 9.285633169807327e-05, "loss": 0.1228, "step": 10248 }, { "epoch": 7.508424908424908, "grad_norm": 0.9140625, "learning_rate": 9.285455845608796e-05, "loss": 0.1463, "step": 10249 }, { "epoch": 7.509157509157509, "grad_norm": 0.97265625, "learning_rate": 9.285278501116803e-05, "loss": 0.1765, "step": 10250 }, { "epoch": 7.50989010989011, "grad_norm": 0.7109375, "learning_rate": 9.285101136332202e-05, "loss": 0.1183, "step": 10251 }, { "epoch": 7.510622710622711, "grad_norm": 0.77734375, "learning_rate": 9.284923751255841e-05, "loss": 0.1149, "step": 10252 }, { "epoch": 7.511355311355311, "grad_norm": 0.83203125, "learning_rate": 9.284746345888569e-05, "loss": 0.1351, "step": 10253 }, { "epoch": 7.512087912087912, "grad_norm": 1.1015625, "learning_rate": 9.284568920231236e-05, "loss": 0.1267, "step": 10254 }, { "epoch": 7.512820512820513, "grad_norm": 0.86328125, "learning_rate": 9.284391474284697e-05, "loss": 0.1276, "step": 10255 }, { "epoch": 7.5135531135531135, "grad_norm": 1.0390625, "learning_rate": 9.284214008049796e-05, "loss": 0.162, "step": 10256 }, { "epoch": 7.514285714285714, "grad_norm": 1.34375, "learning_rate": 9.284036521527386e-05, "loss": 0.1447, "step": 10257 }, { "epoch": 7.515018315018315, "grad_norm": 1.171875, "learning_rate": 9.283859014718316e-05, "loss": 0.2085, "step": 10258 }, { "epoch": 7.515750915750916, "grad_norm": 1.484375, "learning_rate": 9.283681487623439e-05, "loss": 0.1298, "step": 10259 }, { "epoch": 7.516483516483516, "grad_norm": 1.15625, "learning_rate": 9.283503940243605e-05, "loss": 0.1326, "step": 10260 }, { "epoch": 7.517216117216117, "grad_norm": 1.3515625, "learning_rate": 9.283326372579662e-05, "loss": 0.1662, "step": 10261 }, { "epoch": 7.517948717948718, "grad_norm": 0.94140625, "learning_rate": 9.283148784632462e-05, "loss": 0.1158, "step": 10262 }, { "epoch": 7.518681318681319, "grad_norm": 0.64453125, "learning_rate": 9.282971176402859e-05, "loss": 0.1056, "step": 10263 }, { "epoch": 7.519413919413919, "grad_norm": 0.72265625, "learning_rate": 9.282793547891701e-05, "loss": 0.1207, "step": 10264 }, { "epoch": 7.52014652014652, "grad_norm": 0.75390625, "learning_rate": 9.28261589909984e-05, "loss": 0.1224, "step": 10265 }, { "epoch": 7.520879120879121, "grad_norm": 1.0859375, "learning_rate": 9.282438230028123e-05, "loss": 0.1404, "step": 10266 }, { "epoch": 7.5216117216117215, "grad_norm": 1.0625, "learning_rate": 9.282260540677408e-05, "loss": 0.1618, "step": 10267 }, { "epoch": 7.522344322344322, "grad_norm": 1.234375, "learning_rate": 9.282082831048544e-05, "loss": 0.1441, "step": 10268 }, { "epoch": 7.523076923076923, "grad_norm": 1.046875, "learning_rate": 9.281905101142381e-05, "loss": 0.1503, "step": 10269 }, { "epoch": 7.523809523809524, "grad_norm": 1.1796875, "learning_rate": 9.281727350959772e-05, "loss": 0.1521, "step": 10270 }, { "epoch": 7.5245421245421245, "grad_norm": 1.0546875, "learning_rate": 9.281549580501567e-05, "loss": 0.1618, "step": 10271 }, { "epoch": 7.525274725274725, "grad_norm": 1.171875, "learning_rate": 9.28137178976862e-05, "loss": 0.1518, "step": 10272 }, { "epoch": 7.526007326007326, "grad_norm": 1.515625, "learning_rate": 9.281193978761778e-05, "loss": 0.1574, "step": 10273 }, { "epoch": 7.526739926739927, "grad_norm": 1.2421875, "learning_rate": 9.281016147481902e-05, "loss": 0.1239, "step": 10274 }, { "epoch": 7.527472527472527, "grad_norm": 0.97265625, "learning_rate": 9.280838295929834e-05, "loss": 0.1435, "step": 10275 }, { "epoch": 7.528205128205128, "grad_norm": 0.62890625, "learning_rate": 9.280660424106435e-05, "loss": 0.1225, "step": 10276 }, { "epoch": 7.528937728937729, "grad_norm": 1.78125, "learning_rate": 9.280482532012552e-05, "loss": 0.1582, "step": 10277 }, { "epoch": 7.52967032967033, "grad_norm": 1.2578125, "learning_rate": 9.280304619649036e-05, "loss": 0.155, "step": 10278 }, { "epoch": 7.53040293040293, "grad_norm": 1.5625, "learning_rate": 9.280126687016743e-05, "loss": 0.1512, "step": 10279 }, { "epoch": 7.531135531135531, "grad_norm": 1.375, "learning_rate": 9.279948734116526e-05, "loss": 0.1936, "step": 10280 }, { "epoch": 7.531868131868132, "grad_norm": 1.34375, "learning_rate": 9.279770760949235e-05, "loss": 0.1437, "step": 10281 }, { "epoch": 7.5326007326007325, "grad_norm": 1.40625, "learning_rate": 9.279592767515724e-05, "loss": 0.1179, "step": 10282 }, { "epoch": 7.533333333333333, "grad_norm": 1.1796875, "learning_rate": 9.279414753816846e-05, "loss": 0.1665, "step": 10283 }, { "epoch": 7.534065934065934, "grad_norm": 1.2109375, "learning_rate": 9.27923671985345e-05, "loss": 0.1523, "step": 10284 }, { "epoch": 7.534798534798535, "grad_norm": 1.0546875, "learning_rate": 9.279058665626396e-05, "loss": 0.1421, "step": 10285 }, { "epoch": 7.535531135531135, "grad_norm": 0.63671875, "learning_rate": 9.278880591136532e-05, "loss": 0.1149, "step": 10286 }, { "epoch": 7.536263736263736, "grad_norm": 1.0390625, "learning_rate": 9.278702496384713e-05, "loss": 0.1146, "step": 10287 }, { "epoch": 7.536996336996337, "grad_norm": 0.96875, "learning_rate": 9.278524381371794e-05, "loss": 0.1194, "step": 10288 }, { "epoch": 7.537728937728938, "grad_norm": 0.92578125, "learning_rate": 9.278346246098625e-05, "loss": 0.1542, "step": 10289 }, { "epoch": 7.538461538461538, "grad_norm": 0.98828125, "learning_rate": 9.278168090566062e-05, "loss": 0.1402, "step": 10290 }, { "epoch": 7.539194139194139, "grad_norm": 1.1171875, "learning_rate": 9.277989914774957e-05, "loss": 0.1317, "step": 10291 }, { "epoch": 7.53992673992674, "grad_norm": 0.796875, "learning_rate": 9.277811718726164e-05, "loss": 0.1423, "step": 10292 }, { "epoch": 7.5406593406593405, "grad_norm": 1.0703125, "learning_rate": 9.277633502420537e-05, "loss": 0.1461, "step": 10293 }, { "epoch": 7.541391941391941, "grad_norm": 1.203125, "learning_rate": 9.27745526585893e-05, "loss": 0.1346, "step": 10294 }, { "epoch": 7.542124542124542, "grad_norm": 1.46875, "learning_rate": 9.277277009042199e-05, "loss": 0.1039, "step": 10295 }, { "epoch": 7.542857142857143, "grad_norm": 1.40625, "learning_rate": 9.277098731971194e-05, "loss": 0.118, "step": 10296 }, { "epoch": 7.543589743589743, "grad_norm": 0.92578125, "learning_rate": 9.276920434646773e-05, "loss": 0.1397, "step": 10297 }, { "epoch": 7.544322344322344, "grad_norm": 1.0625, "learning_rate": 9.276742117069786e-05, "loss": 0.1492, "step": 10298 }, { "epoch": 7.545054945054945, "grad_norm": 0.890625, "learning_rate": 9.276563779241094e-05, "loss": 0.1181, "step": 10299 }, { "epoch": 7.545787545787546, "grad_norm": 1.0390625, "learning_rate": 9.276385421161544e-05, "loss": 0.1306, "step": 10300 }, { "epoch": 7.546520146520146, "grad_norm": 0.75, "learning_rate": 9.276207042831995e-05, "loss": 0.1193, "step": 10301 }, { "epoch": 7.547252747252747, "grad_norm": 1.0625, "learning_rate": 9.276028644253301e-05, "loss": 0.1431, "step": 10302 }, { "epoch": 7.547985347985348, "grad_norm": 1.0390625, "learning_rate": 9.275850225426316e-05, "loss": 0.1467, "step": 10303 }, { "epoch": 7.5487179487179485, "grad_norm": 1.4609375, "learning_rate": 9.275671786351896e-05, "loss": 0.1431, "step": 10304 }, { "epoch": 7.549450549450549, "grad_norm": 0.8203125, "learning_rate": 9.275493327030895e-05, "loss": 0.1198, "step": 10305 }, { "epoch": 7.55018315018315, "grad_norm": 1.296875, "learning_rate": 9.275314847464168e-05, "loss": 0.1677, "step": 10306 }, { "epoch": 7.550915750915751, "grad_norm": 1.3984375, "learning_rate": 9.275136347652569e-05, "loss": 0.1712, "step": 10307 }, { "epoch": 7.5516483516483515, "grad_norm": 0.96484375, "learning_rate": 9.274957827596958e-05, "loss": 0.1578, "step": 10308 }, { "epoch": 7.552380952380952, "grad_norm": 0.7421875, "learning_rate": 9.274779287298185e-05, "loss": 0.1189, "step": 10309 }, { "epoch": 7.553113553113553, "grad_norm": 1.953125, "learning_rate": 9.274600726757108e-05, "loss": 0.1289, "step": 10310 }, { "epoch": 7.553846153846154, "grad_norm": 1.625, "learning_rate": 9.274422145974581e-05, "loss": 0.1372, "step": 10311 }, { "epoch": 7.554578754578754, "grad_norm": 1.2109375, "learning_rate": 9.274243544951464e-05, "loss": 0.1351, "step": 10312 }, { "epoch": 7.555311355311355, "grad_norm": 1.09375, "learning_rate": 9.274064923688605e-05, "loss": 0.1841, "step": 10313 }, { "epoch": 7.556043956043956, "grad_norm": 0.8828125, "learning_rate": 9.273886282186867e-05, "loss": 0.1192, "step": 10314 }, { "epoch": 7.556776556776557, "grad_norm": 1.75, "learning_rate": 9.273707620447103e-05, "loss": 0.1455, "step": 10315 }, { "epoch": 7.557509157509157, "grad_norm": 1.375, "learning_rate": 9.273528938470168e-05, "loss": 0.123, "step": 10316 }, { "epoch": 7.558241758241758, "grad_norm": 1.78125, "learning_rate": 9.273350236256921e-05, "loss": 0.1919, "step": 10317 }, { "epoch": 7.558974358974359, "grad_norm": 1.6953125, "learning_rate": 9.273171513808217e-05, "loss": 0.1449, "step": 10318 }, { "epoch": 7.5597069597069595, "grad_norm": 2.078125, "learning_rate": 9.27299277112491e-05, "loss": 0.1503, "step": 10319 }, { "epoch": 7.56043956043956, "grad_norm": 1.28125, "learning_rate": 9.27281400820786e-05, "loss": 0.1473, "step": 10320 }, { "epoch": 7.561172161172161, "grad_norm": 1.125, "learning_rate": 9.272635225057919e-05, "loss": 0.1373, "step": 10321 }, { "epoch": 7.561904761904762, "grad_norm": 0.90625, "learning_rate": 9.27245642167595e-05, "loss": 0.1196, "step": 10322 }, { "epoch": 7.562637362637362, "grad_norm": 1.1484375, "learning_rate": 9.272277598062803e-05, "loss": 0.1278, "step": 10323 }, { "epoch": 7.563369963369963, "grad_norm": 1.296875, "learning_rate": 9.27209875421934e-05, "loss": 0.1367, "step": 10324 }, { "epoch": 7.564102564102564, "grad_norm": 1.0703125, "learning_rate": 9.271919890146415e-05, "loss": 0.1367, "step": 10325 }, { "epoch": 7.564835164835165, "grad_norm": 0.80859375, "learning_rate": 9.271741005844887e-05, "loss": 0.1352, "step": 10326 }, { "epoch": 7.565567765567765, "grad_norm": 1.2890625, "learning_rate": 9.27156210131561e-05, "loss": 0.1368, "step": 10327 }, { "epoch": 7.566300366300366, "grad_norm": 1.8203125, "learning_rate": 9.271383176559443e-05, "loss": 0.1738, "step": 10328 }, { "epoch": 7.567032967032967, "grad_norm": 1.34375, "learning_rate": 9.271204231577246e-05, "loss": 0.1538, "step": 10329 }, { "epoch": 7.5677655677655675, "grad_norm": 0.91796875, "learning_rate": 9.271025266369872e-05, "loss": 0.1297, "step": 10330 }, { "epoch": 7.568498168498168, "grad_norm": 1.140625, "learning_rate": 9.270846280938181e-05, "loss": 0.1635, "step": 10331 }, { "epoch": 7.569230769230769, "grad_norm": 0.5390625, "learning_rate": 9.27066727528303e-05, "loss": 0.1114, "step": 10332 }, { "epoch": 7.56996336996337, "grad_norm": 1.3359375, "learning_rate": 9.270488249405275e-05, "loss": 0.1524, "step": 10333 }, { "epoch": 7.5706959706959704, "grad_norm": 1.0546875, "learning_rate": 9.270309203305775e-05, "loss": 0.1263, "step": 10334 }, { "epoch": 7.571428571428571, "grad_norm": 1.484375, "learning_rate": 9.270130136985391e-05, "loss": 0.1522, "step": 10335 }, { "epoch": 7.572161172161172, "grad_norm": 1.6796875, "learning_rate": 9.269951050444978e-05, "loss": 0.1478, "step": 10336 }, { "epoch": 7.572893772893773, "grad_norm": 0.9921875, "learning_rate": 9.269771943685392e-05, "loss": 0.124, "step": 10337 }, { "epoch": 7.573626373626373, "grad_norm": 0.97265625, "learning_rate": 9.269592816707493e-05, "loss": 0.1237, "step": 10338 }, { "epoch": 7.574358974358974, "grad_norm": 1.421875, "learning_rate": 9.269413669512143e-05, "loss": 0.1203, "step": 10339 }, { "epoch": 7.575091575091575, "grad_norm": 1.4609375, "learning_rate": 9.269234502100193e-05, "loss": 0.1363, "step": 10340 }, { "epoch": 7.5758241758241756, "grad_norm": 0.69921875, "learning_rate": 9.269055314472507e-05, "loss": 0.1233, "step": 10341 }, { "epoch": 7.576556776556776, "grad_norm": 0.9375, "learning_rate": 9.268876106629944e-05, "loss": 0.1755, "step": 10342 }, { "epoch": 7.577289377289377, "grad_norm": 0.9609375, "learning_rate": 9.268696878573357e-05, "loss": 0.1429, "step": 10343 }, { "epoch": 7.578021978021978, "grad_norm": 1.0078125, "learning_rate": 9.26851763030361e-05, "loss": 0.1136, "step": 10344 }, { "epoch": 7.5787545787545785, "grad_norm": 1.5546875, "learning_rate": 9.268338361821561e-05, "loss": 0.1133, "step": 10345 }, { "epoch": 7.579487179487179, "grad_norm": 2.421875, "learning_rate": 9.268159073128068e-05, "loss": 0.1675, "step": 10346 }, { "epoch": 7.58021978021978, "grad_norm": 1.1953125, "learning_rate": 9.26797976422399e-05, "loss": 0.1346, "step": 10347 }, { "epoch": 7.580952380952381, "grad_norm": 1.171875, "learning_rate": 9.267800435110186e-05, "loss": 0.122, "step": 10348 }, { "epoch": 7.581684981684981, "grad_norm": 0.75, "learning_rate": 9.267621085787515e-05, "loss": 0.1595, "step": 10349 }, { "epoch": 7.582417582417582, "grad_norm": 0.87890625, "learning_rate": 9.267441716256837e-05, "loss": 0.1332, "step": 10350 }, { "epoch": 7.583150183150183, "grad_norm": 1.6015625, "learning_rate": 9.26726232651901e-05, "loss": 0.1459, "step": 10351 }, { "epoch": 7.583882783882784, "grad_norm": 1.265625, "learning_rate": 9.267082916574898e-05, "loss": 0.1356, "step": 10352 }, { "epoch": 7.584615384615384, "grad_norm": 0.765625, "learning_rate": 9.266903486425355e-05, "loss": 0.1226, "step": 10353 }, { "epoch": 7.585347985347985, "grad_norm": 1.1484375, "learning_rate": 9.266724036071244e-05, "loss": 0.1405, "step": 10354 }, { "epoch": 7.586080586080586, "grad_norm": 0.65234375, "learning_rate": 9.266544565513424e-05, "loss": 0.1328, "step": 10355 }, { "epoch": 7.5868131868131865, "grad_norm": 1.3984375, "learning_rate": 9.266365074752754e-05, "loss": 0.1265, "step": 10356 }, { "epoch": 7.587545787545787, "grad_norm": 1.9765625, "learning_rate": 9.266185563790093e-05, "loss": 0.1472, "step": 10357 }, { "epoch": 7.588278388278388, "grad_norm": 1.4765625, "learning_rate": 9.266006032626305e-05, "loss": 0.1368, "step": 10358 }, { "epoch": 7.589010989010989, "grad_norm": 1.578125, "learning_rate": 9.265826481262249e-05, "loss": 0.1172, "step": 10359 }, { "epoch": 7.589743589743589, "grad_norm": 0.92578125, "learning_rate": 9.265646909698785e-05, "loss": 0.141, "step": 10360 }, { "epoch": 7.59047619047619, "grad_norm": 0.92578125, "learning_rate": 9.26546731793677e-05, "loss": 0.1228, "step": 10361 }, { "epoch": 7.591208791208791, "grad_norm": 0.98828125, "learning_rate": 9.265287705977068e-05, "loss": 0.1285, "step": 10362 }, { "epoch": 7.591941391941392, "grad_norm": 0.97265625, "learning_rate": 9.265108073820539e-05, "loss": 0.1469, "step": 10363 }, { "epoch": 7.592673992673992, "grad_norm": 1.3515625, "learning_rate": 9.264928421468044e-05, "loss": 0.1498, "step": 10364 }, { "epoch": 7.593406593406593, "grad_norm": 1.0234375, "learning_rate": 9.264748748920444e-05, "loss": 0.1289, "step": 10365 }, { "epoch": 7.594139194139194, "grad_norm": 0.89453125, "learning_rate": 9.264569056178597e-05, "loss": 0.1256, "step": 10366 }, { "epoch": 7.5948717948717945, "grad_norm": 0.66015625, "learning_rate": 9.264389343243367e-05, "loss": 0.1245, "step": 10367 }, { "epoch": 7.595604395604395, "grad_norm": 1.0546875, "learning_rate": 9.264209610115615e-05, "loss": 0.1352, "step": 10368 }, { "epoch": 7.596336996336996, "grad_norm": 1.4375, "learning_rate": 9.2640298567962e-05, "loss": 0.1525, "step": 10369 }, { "epoch": 7.597069597069597, "grad_norm": 1.4453125, "learning_rate": 9.263850083285985e-05, "loss": 0.1611, "step": 10370 }, { "epoch": 7.5978021978021975, "grad_norm": 0.76171875, "learning_rate": 9.263670289585831e-05, "loss": 0.1362, "step": 10371 }, { "epoch": 7.598534798534798, "grad_norm": 0.9609375, "learning_rate": 9.2634904756966e-05, "loss": 0.1466, "step": 10372 }, { "epoch": 7.599267399267399, "grad_norm": 0.71875, "learning_rate": 9.263310641619152e-05, "loss": 0.1159, "step": 10373 }, { "epoch": 7.6, "grad_norm": 0.90234375, "learning_rate": 9.26313078735435e-05, "loss": 0.109, "step": 10374 }, { "epoch": 7.6007326007326, "grad_norm": 1.3828125, "learning_rate": 9.262950912903057e-05, "loss": 0.167, "step": 10375 }, { "epoch": 7.601465201465201, "grad_norm": 1.3984375, "learning_rate": 9.262771018266131e-05, "loss": 0.1187, "step": 10376 }, { "epoch": 7.602197802197802, "grad_norm": 1.40625, "learning_rate": 9.262591103444436e-05, "loss": 0.1629, "step": 10377 }, { "epoch": 7.602930402930403, "grad_norm": 1.140625, "learning_rate": 9.262411168438835e-05, "loss": 0.1483, "step": 10378 }, { "epoch": 7.603663003663003, "grad_norm": 0.93359375, "learning_rate": 9.262231213250189e-05, "loss": 0.1512, "step": 10379 }, { "epoch": 7.604395604395604, "grad_norm": 1.1484375, "learning_rate": 9.262051237879361e-05, "loss": 0.1751, "step": 10380 }, { "epoch": 7.605128205128205, "grad_norm": 0.8203125, "learning_rate": 9.261871242327211e-05, "loss": 0.1176, "step": 10381 }, { "epoch": 7.6058608058608055, "grad_norm": 0.6015625, "learning_rate": 9.261691226594605e-05, "loss": 0.1325, "step": 10382 }, { "epoch": 7.606593406593406, "grad_norm": 0.98046875, "learning_rate": 9.261511190682402e-05, "loss": 0.1388, "step": 10383 }, { "epoch": 7.607326007326007, "grad_norm": 0.60546875, "learning_rate": 9.26133113459147e-05, "loss": 0.1186, "step": 10384 }, { "epoch": 7.608058608058608, "grad_norm": 0.7734375, "learning_rate": 9.261151058322664e-05, "loss": 0.1249, "step": 10385 }, { "epoch": 7.608791208791208, "grad_norm": 0.9140625, "learning_rate": 9.260970961876853e-05, "loss": 0.1488, "step": 10386 }, { "epoch": 7.609523809523809, "grad_norm": 1.34375, "learning_rate": 9.260790845254898e-05, "loss": 0.1657, "step": 10387 }, { "epoch": 7.61025641025641, "grad_norm": 0.72265625, "learning_rate": 9.26061070845766e-05, "loss": 0.1374, "step": 10388 }, { "epoch": 7.610989010989011, "grad_norm": 0.73828125, "learning_rate": 9.260430551486004e-05, "loss": 0.1286, "step": 10389 }, { "epoch": 7.611721611721611, "grad_norm": 0.71875, "learning_rate": 9.260250374340795e-05, "loss": 0.1231, "step": 10390 }, { "epoch": 7.612454212454212, "grad_norm": 0.490234375, "learning_rate": 9.260070177022892e-05, "loss": 0.1149, "step": 10391 }, { "epoch": 7.613186813186813, "grad_norm": 0.8125, "learning_rate": 9.259889959533163e-05, "loss": 0.1453, "step": 10392 }, { "epoch": 7.6139194139194135, "grad_norm": 0.97265625, "learning_rate": 9.259709721872468e-05, "loss": 0.161, "step": 10393 }, { "epoch": 7.614652014652014, "grad_norm": 1.171875, "learning_rate": 9.259529464041671e-05, "loss": 0.2021, "step": 10394 }, { "epoch": 7.615384615384615, "grad_norm": 1.1875, "learning_rate": 9.259349186041637e-05, "loss": 0.1446, "step": 10395 }, { "epoch": 7.616117216117216, "grad_norm": 1.0625, "learning_rate": 9.259168887873229e-05, "loss": 0.1491, "step": 10396 }, { "epoch": 7.616849816849816, "grad_norm": 0.76171875, "learning_rate": 9.258988569537313e-05, "loss": 0.1225, "step": 10397 }, { "epoch": 7.617582417582417, "grad_norm": 0.79296875, "learning_rate": 9.258808231034749e-05, "loss": 0.1375, "step": 10398 }, { "epoch": 7.618315018315018, "grad_norm": 1.5, "learning_rate": 9.258627872366405e-05, "loss": 0.1245, "step": 10399 }, { "epoch": 7.619047619047619, "grad_norm": 1.109375, "learning_rate": 9.258447493533141e-05, "loss": 0.1262, "step": 10400 }, { "epoch": 7.619780219780219, "grad_norm": 1.0546875, "learning_rate": 9.258267094535825e-05, "loss": 0.1519, "step": 10401 }, { "epoch": 7.62051282051282, "grad_norm": 0.73828125, "learning_rate": 9.258086675375319e-05, "loss": 0.0945, "step": 10402 }, { "epoch": 7.621245421245421, "grad_norm": 1.21875, "learning_rate": 9.257906236052489e-05, "loss": 0.1601, "step": 10403 }, { "epoch": 7.6219780219780215, "grad_norm": 0.66015625, "learning_rate": 9.257725776568197e-05, "loss": 0.1405, "step": 10404 }, { "epoch": 7.622710622710622, "grad_norm": 1.03125, "learning_rate": 9.257545296923311e-05, "loss": 0.152, "step": 10405 }, { "epoch": 7.623443223443223, "grad_norm": 1.4765625, "learning_rate": 9.257364797118694e-05, "loss": 0.1669, "step": 10406 }, { "epoch": 7.624175824175824, "grad_norm": 1.078125, "learning_rate": 9.257184277155212e-05, "loss": 0.1724, "step": 10407 }, { "epoch": 7.6249084249084245, "grad_norm": 1.1640625, "learning_rate": 9.257003737033728e-05, "loss": 0.1528, "step": 10408 }, { "epoch": 7.625641025641025, "grad_norm": 1.1015625, "learning_rate": 9.256823176755109e-05, "loss": 0.1282, "step": 10409 }, { "epoch": 7.626373626373626, "grad_norm": 0.9921875, "learning_rate": 9.256642596320219e-05, "loss": 0.1326, "step": 10410 }, { "epoch": 7.627106227106227, "grad_norm": 0.87109375, "learning_rate": 9.256461995729923e-05, "loss": 0.1472, "step": 10411 }, { "epoch": 7.627838827838827, "grad_norm": 1.046875, "learning_rate": 9.256281374985085e-05, "loss": 0.1388, "step": 10412 }, { "epoch": 7.628571428571428, "grad_norm": 0.96484375, "learning_rate": 9.256100734086573e-05, "loss": 0.1354, "step": 10413 }, { "epoch": 7.629304029304029, "grad_norm": 1.09375, "learning_rate": 9.255920073035253e-05, "loss": 0.1626, "step": 10414 }, { "epoch": 7.63003663003663, "grad_norm": 1.40625, "learning_rate": 9.25573939183199e-05, "loss": 0.1145, "step": 10415 }, { "epoch": 7.63076923076923, "grad_norm": 0.6875, "learning_rate": 9.255558690477646e-05, "loss": 0.1255, "step": 10416 }, { "epoch": 7.631501831501831, "grad_norm": 0.83984375, "learning_rate": 9.255377968973092e-05, "loss": 0.1521, "step": 10417 }, { "epoch": 7.632234432234432, "grad_norm": 1.234375, "learning_rate": 9.25519722731919e-05, "loss": 0.1406, "step": 10418 }, { "epoch": 7.6329670329670325, "grad_norm": 1.1796875, "learning_rate": 9.255016465516808e-05, "loss": 0.1497, "step": 10419 }, { "epoch": 7.633699633699633, "grad_norm": 1.0234375, "learning_rate": 9.254835683566812e-05, "loss": 0.1291, "step": 10420 }, { "epoch": 7.634432234432234, "grad_norm": 1.1875, "learning_rate": 9.254654881470068e-05, "loss": 0.1579, "step": 10421 }, { "epoch": 7.635164835164835, "grad_norm": 1.0390625, "learning_rate": 9.254474059227442e-05, "loss": 0.1332, "step": 10422 }, { "epoch": 7.635897435897435, "grad_norm": 1.1171875, "learning_rate": 9.254293216839801e-05, "loss": 0.1476, "step": 10423 }, { "epoch": 7.636630036630036, "grad_norm": 1.6640625, "learning_rate": 9.25411235430801e-05, "loss": 0.1065, "step": 10424 }, { "epoch": 7.637362637362637, "grad_norm": 1.640625, "learning_rate": 9.253931471632938e-05, "loss": 0.1321, "step": 10425 }, { "epoch": 7.638095238095238, "grad_norm": 0.99609375, "learning_rate": 9.253750568815449e-05, "loss": 0.1163, "step": 10426 }, { "epoch": 7.638827838827838, "grad_norm": 0.98046875, "learning_rate": 9.253569645856411e-05, "loss": 0.1277, "step": 10427 }, { "epoch": 7.639560439560439, "grad_norm": 0.94140625, "learning_rate": 9.253388702756692e-05, "loss": 0.1576, "step": 10428 }, { "epoch": 7.64029304029304, "grad_norm": 1.3984375, "learning_rate": 9.253207739517157e-05, "loss": 0.1363, "step": 10429 }, { "epoch": 7.641025641025641, "grad_norm": 1.53125, "learning_rate": 9.253026756138674e-05, "loss": 0.1831, "step": 10430 }, { "epoch": 7.641758241758242, "grad_norm": 1.265625, "learning_rate": 9.25284575262211e-05, "loss": 0.1665, "step": 10431 }, { "epoch": 7.642490842490843, "grad_norm": 1.734375, "learning_rate": 9.252664728968332e-05, "loss": 0.1103, "step": 10432 }, { "epoch": 7.643223443223444, "grad_norm": 0.875, "learning_rate": 9.252483685178208e-05, "loss": 0.1343, "step": 10433 }, { "epoch": 7.643956043956044, "grad_norm": 0.765625, "learning_rate": 9.252302621252605e-05, "loss": 0.1418, "step": 10434 }, { "epoch": 7.644688644688645, "grad_norm": 0.90234375, "learning_rate": 9.252121537192393e-05, "loss": 0.1378, "step": 10435 }, { "epoch": 7.645421245421246, "grad_norm": 0.70703125, "learning_rate": 9.251940432998433e-05, "loss": 0.1488, "step": 10436 }, { "epoch": 7.6461538461538465, "grad_norm": 1.0078125, "learning_rate": 9.251759308671599e-05, "loss": 0.1504, "step": 10437 }, { "epoch": 7.646886446886447, "grad_norm": 0.80078125, "learning_rate": 9.251578164212757e-05, "loss": 0.117, "step": 10438 }, { "epoch": 7.647619047619048, "grad_norm": 1.34375, "learning_rate": 9.251396999622775e-05, "loss": 0.1483, "step": 10439 }, { "epoch": 7.648351648351649, "grad_norm": 0.8984375, "learning_rate": 9.251215814902519e-05, "loss": 0.1759, "step": 10440 }, { "epoch": 7.649084249084249, "grad_norm": 1.0390625, "learning_rate": 9.251034610052861e-05, "loss": 0.1737, "step": 10441 }, { "epoch": 7.64981684981685, "grad_norm": 1.1171875, "learning_rate": 9.250853385074667e-05, "loss": 0.1203, "step": 10442 }, { "epoch": 7.650549450549451, "grad_norm": 1.3359375, "learning_rate": 9.250672139968804e-05, "loss": 0.1534, "step": 10443 }, { "epoch": 7.651282051282052, "grad_norm": 1.1015625, "learning_rate": 9.250490874736141e-05, "loss": 0.127, "step": 10444 }, { "epoch": 7.652014652014652, "grad_norm": 1.0546875, "learning_rate": 9.25030958937755e-05, "loss": 0.1201, "step": 10445 }, { "epoch": 7.652747252747253, "grad_norm": 0.85546875, "learning_rate": 9.250128283893896e-05, "loss": 0.1349, "step": 10446 }, { "epoch": 7.653479853479854, "grad_norm": 1.0, "learning_rate": 9.24994695828605e-05, "loss": 0.1437, "step": 10447 }, { "epoch": 7.6542124542124546, "grad_norm": 1.28125, "learning_rate": 9.249765612554875e-05, "loss": 0.1614, "step": 10448 }, { "epoch": 7.654945054945055, "grad_norm": 1.3203125, "learning_rate": 9.249584246701247e-05, "loss": 0.1693, "step": 10449 }, { "epoch": 7.655677655677656, "grad_norm": 1.7421875, "learning_rate": 9.249402860726033e-05, "loss": 0.1293, "step": 10450 }, { "epoch": 7.656410256410257, "grad_norm": 2.734375, "learning_rate": 9.249221454630101e-05, "loss": 0.1358, "step": 10451 }, { "epoch": 7.6571428571428575, "grad_norm": 1.53125, "learning_rate": 9.249040028414318e-05, "loss": 0.1222, "step": 10452 }, { "epoch": 7.657875457875458, "grad_norm": 0.609375, "learning_rate": 9.248858582079559e-05, "loss": 0.1141, "step": 10453 }, { "epoch": 7.658608058608059, "grad_norm": 0.86328125, "learning_rate": 9.248677115626688e-05, "loss": 0.1008, "step": 10454 }, { "epoch": 7.65934065934066, "grad_norm": 1.609375, "learning_rate": 9.248495629056577e-05, "loss": 0.1474, "step": 10455 }, { "epoch": 7.66007326007326, "grad_norm": 1.7734375, "learning_rate": 9.248314122370095e-05, "loss": 0.1723, "step": 10456 }, { "epoch": 7.660805860805861, "grad_norm": 1.25, "learning_rate": 9.248132595568113e-05, "loss": 0.1362, "step": 10457 }, { "epoch": 7.661538461538462, "grad_norm": 1.4140625, "learning_rate": 9.247951048651499e-05, "loss": 0.1432, "step": 10458 }, { "epoch": 7.662271062271063, "grad_norm": 1.0, "learning_rate": 9.247769481621124e-05, "loss": 0.1503, "step": 10459 }, { "epoch": 7.663003663003663, "grad_norm": 1.40625, "learning_rate": 9.247587894477856e-05, "loss": 0.1351, "step": 10460 }, { "epoch": 7.663736263736264, "grad_norm": 1.1796875, "learning_rate": 9.247406287222568e-05, "loss": 0.1405, "step": 10461 }, { "epoch": 7.664468864468865, "grad_norm": 1.0859375, "learning_rate": 9.247224659856128e-05, "loss": 0.1584, "step": 10462 }, { "epoch": 7.6652014652014655, "grad_norm": 1.1875, "learning_rate": 9.247043012379406e-05, "loss": 0.1255, "step": 10463 }, { "epoch": 7.665934065934066, "grad_norm": 0.9296875, "learning_rate": 9.246861344793273e-05, "loss": 0.1314, "step": 10464 }, { "epoch": 7.666666666666667, "grad_norm": 1.484375, "learning_rate": 9.2466796570986e-05, "loss": 0.1889, "step": 10465 }, { "epoch": 7.667399267399268, "grad_norm": 1.1640625, "learning_rate": 9.246497949296257e-05, "loss": 0.1553, "step": 10466 }, { "epoch": 7.668131868131868, "grad_norm": 1.0078125, "learning_rate": 9.246316221387116e-05, "loss": 0.155, "step": 10467 }, { "epoch": 7.668864468864469, "grad_norm": 0.7734375, "learning_rate": 9.246134473372045e-05, "loss": 0.1267, "step": 10468 }, { "epoch": 7.66959706959707, "grad_norm": 0.81640625, "learning_rate": 9.245952705251916e-05, "loss": 0.1272, "step": 10469 }, { "epoch": 7.670329670329671, "grad_norm": 0.9140625, "learning_rate": 9.245770917027602e-05, "loss": 0.1453, "step": 10470 }, { "epoch": 7.671062271062271, "grad_norm": 0.91015625, "learning_rate": 9.24558910869997e-05, "loss": 0.1492, "step": 10471 }, { "epoch": 7.671794871794872, "grad_norm": 1.4140625, "learning_rate": 9.245407280269895e-05, "loss": 0.1412, "step": 10472 }, { "epoch": 7.672527472527473, "grad_norm": 1.71875, "learning_rate": 9.245225431738246e-05, "loss": 0.1413, "step": 10473 }, { "epoch": 7.6732600732600735, "grad_norm": 1.421875, "learning_rate": 9.245043563105893e-05, "loss": 0.1749, "step": 10474 }, { "epoch": 7.673992673992674, "grad_norm": 1.0625, "learning_rate": 9.244861674373711e-05, "loss": 0.1571, "step": 10475 }, { "epoch": 7.674725274725275, "grad_norm": 1.1015625, "learning_rate": 9.24467976554257e-05, "loss": 0.1534, "step": 10476 }, { "epoch": 7.675457875457876, "grad_norm": 0.8359375, "learning_rate": 9.24449783661334e-05, "loss": 0.1391, "step": 10477 }, { "epoch": 7.6761904761904765, "grad_norm": 0.94140625, "learning_rate": 9.244315887586895e-05, "loss": 0.1467, "step": 10478 }, { "epoch": 7.676923076923077, "grad_norm": 1.3359375, "learning_rate": 9.244133918464104e-05, "loss": 0.1432, "step": 10479 }, { "epoch": 7.677655677655678, "grad_norm": 1.1171875, "learning_rate": 9.243951929245843e-05, "loss": 0.1356, "step": 10480 }, { "epoch": 7.678388278388279, "grad_norm": 1.59375, "learning_rate": 9.243769919932979e-05, "loss": 0.1677, "step": 10481 }, { "epoch": 7.679120879120879, "grad_norm": 0.87890625, "learning_rate": 9.243587890526387e-05, "loss": 0.1592, "step": 10482 }, { "epoch": 7.67985347985348, "grad_norm": 1.0859375, "learning_rate": 9.24340584102694e-05, "loss": 0.1342, "step": 10483 }, { "epoch": 7.680586080586081, "grad_norm": 0.625, "learning_rate": 9.243223771435508e-05, "loss": 0.1081, "step": 10484 }, { "epoch": 7.681318681318682, "grad_norm": 1.265625, "learning_rate": 9.243041681752964e-05, "loss": 0.1318, "step": 10485 }, { "epoch": 7.682051282051282, "grad_norm": 1.0234375, "learning_rate": 9.242859571980182e-05, "loss": 0.1276, "step": 10486 }, { "epoch": 7.682783882783883, "grad_norm": 0.9140625, "learning_rate": 9.242677442118033e-05, "loss": 0.1531, "step": 10487 }, { "epoch": 7.683516483516484, "grad_norm": 1.0078125, "learning_rate": 9.24249529216739e-05, "loss": 0.1683, "step": 10488 }, { "epoch": 7.6842490842490845, "grad_norm": 0.875, "learning_rate": 9.242313122129126e-05, "loss": 0.1254, "step": 10489 }, { "epoch": 7.684981684981685, "grad_norm": 0.90234375, "learning_rate": 9.242130932004114e-05, "loss": 0.1372, "step": 10490 }, { "epoch": 7.685714285714286, "grad_norm": 1.28125, "learning_rate": 9.241948721793224e-05, "loss": 0.153, "step": 10491 }, { "epoch": 7.686446886446887, "grad_norm": 1.140625, "learning_rate": 9.241766491497335e-05, "loss": 0.11, "step": 10492 }, { "epoch": 7.687179487179487, "grad_norm": 1.1015625, "learning_rate": 9.241584241117316e-05, "loss": 0.1225, "step": 10493 }, { "epoch": 7.687912087912088, "grad_norm": 0.69140625, "learning_rate": 9.24140197065404e-05, "loss": 0.1446, "step": 10494 }, { "epoch": 7.688644688644689, "grad_norm": 0.8203125, "learning_rate": 9.24121968010838e-05, "loss": 0.1235, "step": 10495 }, { "epoch": 7.68937728937729, "grad_norm": 0.97265625, "learning_rate": 9.241037369481212e-05, "loss": 0.1351, "step": 10496 }, { "epoch": 7.69010989010989, "grad_norm": 0.9609375, "learning_rate": 9.240855038773409e-05, "loss": 0.1458, "step": 10497 }, { "epoch": 7.690842490842491, "grad_norm": 1.1796875, "learning_rate": 9.240672687985841e-05, "loss": 0.1488, "step": 10498 }, { "epoch": 7.691575091575092, "grad_norm": 0.6484375, "learning_rate": 9.240490317119387e-05, "loss": 0.1033, "step": 10499 }, { "epoch": 7.6923076923076925, "grad_norm": 0.84765625, "learning_rate": 9.240307926174916e-05, "loss": 0.1092, "step": 10500 }, { "epoch": 7.693040293040293, "grad_norm": 0.84375, "learning_rate": 9.240125515153307e-05, "loss": 0.1487, "step": 10501 }, { "epoch": 7.693772893772894, "grad_norm": 2.609375, "learning_rate": 9.239943084055428e-05, "loss": 0.161, "step": 10502 }, { "epoch": 7.694505494505495, "grad_norm": 1.6171875, "learning_rate": 9.239760632882159e-05, "loss": 0.1547, "step": 10503 }, { "epoch": 7.695238095238095, "grad_norm": 1.6171875, "learning_rate": 9.23957816163437e-05, "loss": 0.1298, "step": 10504 }, { "epoch": 7.695970695970696, "grad_norm": 1.1875, "learning_rate": 9.239395670312936e-05, "loss": 0.1326, "step": 10505 }, { "epoch": 7.696703296703297, "grad_norm": 1.2890625, "learning_rate": 9.239213158918732e-05, "loss": 0.1437, "step": 10506 }, { "epoch": 7.697435897435898, "grad_norm": 1.0078125, "learning_rate": 9.239030627452634e-05, "loss": 0.131, "step": 10507 }, { "epoch": 7.698168498168498, "grad_norm": 1.90625, "learning_rate": 9.238848075915515e-05, "loss": 0.1523, "step": 10508 }, { "epoch": 7.698901098901099, "grad_norm": 2.0625, "learning_rate": 9.23866550430825e-05, "loss": 0.1327, "step": 10509 }, { "epoch": 7.6996336996337, "grad_norm": 1.8828125, "learning_rate": 9.238482912631713e-05, "loss": 0.2045, "step": 10510 }, { "epoch": 7.7003663003663005, "grad_norm": 1.1171875, "learning_rate": 9.238300300886779e-05, "loss": 0.1546, "step": 10511 }, { "epoch": 7.701098901098901, "grad_norm": 1.4609375, "learning_rate": 9.238117669074324e-05, "loss": 0.1504, "step": 10512 }, { "epoch": 7.701831501831502, "grad_norm": 1.34375, "learning_rate": 9.237935017195222e-05, "loss": 0.1547, "step": 10513 }, { "epoch": 7.702564102564103, "grad_norm": 0.6640625, "learning_rate": 9.237752345250348e-05, "loss": 0.1186, "step": 10514 }, { "epoch": 7.7032967032967035, "grad_norm": 1.0859375, "learning_rate": 9.237569653240578e-05, "loss": 0.1908, "step": 10515 }, { "epoch": 7.704029304029304, "grad_norm": 0.734375, "learning_rate": 9.237386941166787e-05, "loss": 0.1309, "step": 10516 }, { "epoch": 7.704761904761905, "grad_norm": 1.265625, "learning_rate": 9.237204209029852e-05, "loss": 0.12, "step": 10517 }, { "epoch": 7.705494505494506, "grad_norm": 1.171875, "learning_rate": 9.237021456830645e-05, "loss": 0.1627, "step": 10518 }, { "epoch": 7.706227106227106, "grad_norm": 1.125, "learning_rate": 9.236838684570045e-05, "loss": 0.1439, "step": 10519 }, { "epoch": 7.706959706959707, "grad_norm": 1.515625, "learning_rate": 9.236655892248927e-05, "loss": 0.1609, "step": 10520 }, { "epoch": 7.707692307692308, "grad_norm": 1.1171875, "learning_rate": 9.236473079868165e-05, "loss": 0.0964, "step": 10521 }, { "epoch": 7.708424908424909, "grad_norm": 1.6484375, "learning_rate": 9.236290247428635e-05, "loss": 0.1668, "step": 10522 }, { "epoch": 7.709157509157509, "grad_norm": 1.3984375, "learning_rate": 9.236107394931216e-05, "loss": 0.1179, "step": 10523 }, { "epoch": 7.70989010989011, "grad_norm": 0.625, "learning_rate": 9.235924522376782e-05, "loss": 0.1285, "step": 10524 }, { "epoch": 7.710622710622711, "grad_norm": 1.03125, "learning_rate": 9.23574162976621e-05, "loss": 0.157, "step": 10525 }, { "epoch": 7.7113553113553115, "grad_norm": 2.484375, "learning_rate": 9.235558717100375e-05, "loss": 0.1416, "step": 10526 }, { "epoch": 7.712087912087912, "grad_norm": 1.0625, "learning_rate": 9.235375784380154e-05, "loss": 0.1549, "step": 10527 }, { "epoch": 7.712820512820513, "grad_norm": 0.9375, "learning_rate": 9.235192831606423e-05, "loss": 0.109, "step": 10528 }, { "epoch": 7.713553113553114, "grad_norm": 1.1484375, "learning_rate": 9.235009858780061e-05, "loss": 0.1795, "step": 10529 }, { "epoch": 7.714285714285714, "grad_norm": 1.1640625, "learning_rate": 9.234826865901942e-05, "loss": 0.1302, "step": 10530 }, { "epoch": 7.715018315018315, "grad_norm": 1.078125, "learning_rate": 9.234643852972943e-05, "loss": 0.1304, "step": 10531 }, { "epoch": 7.715750915750916, "grad_norm": 0.8984375, "learning_rate": 9.234460819993944e-05, "loss": 0.1209, "step": 10532 }, { "epoch": 7.716483516483517, "grad_norm": 1.015625, "learning_rate": 9.234277766965818e-05, "loss": 0.125, "step": 10533 }, { "epoch": 7.717216117216117, "grad_norm": 1.359375, "learning_rate": 9.234094693889442e-05, "loss": 0.1141, "step": 10534 }, { "epoch": 7.717948717948718, "grad_norm": 1.6328125, "learning_rate": 9.233911600765696e-05, "loss": 0.1609, "step": 10535 }, { "epoch": 7.718681318681319, "grad_norm": 2.359375, "learning_rate": 9.233728487595457e-05, "loss": 0.1656, "step": 10536 }, { "epoch": 7.7194139194139195, "grad_norm": 1.390625, "learning_rate": 9.2335453543796e-05, "loss": 0.1437, "step": 10537 }, { "epoch": 7.72014652014652, "grad_norm": 0.94921875, "learning_rate": 9.233362201119003e-05, "loss": 0.1355, "step": 10538 }, { "epoch": 7.720879120879121, "grad_norm": 0.6171875, "learning_rate": 9.233179027814546e-05, "loss": 0.135, "step": 10539 }, { "epoch": 7.721611721611722, "grad_norm": 1.2890625, "learning_rate": 9.232995834467107e-05, "loss": 0.1079, "step": 10540 }, { "epoch": 7.722344322344322, "grad_norm": 1.5390625, "learning_rate": 9.232812621077558e-05, "loss": 0.146, "step": 10541 }, { "epoch": 7.723076923076923, "grad_norm": 2.0, "learning_rate": 9.232629387646783e-05, "loss": 0.1599, "step": 10542 }, { "epoch": 7.723809523809524, "grad_norm": 0.96484375, "learning_rate": 9.232446134175656e-05, "loss": 0.0953, "step": 10543 }, { "epoch": 7.724542124542125, "grad_norm": 1.125, "learning_rate": 9.232262860665058e-05, "loss": 0.1406, "step": 10544 }, { "epoch": 7.725274725274725, "grad_norm": 0.94921875, "learning_rate": 9.232079567115865e-05, "loss": 0.1159, "step": 10545 }, { "epoch": 7.726007326007326, "grad_norm": 0.96875, "learning_rate": 9.231896253528956e-05, "loss": 0.1351, "step": 10546 }, { "epoch": 7.726739926739927, "grad_norm": 0.9375, "learning_rate": 9.231712919905208e-05, "loss": 0.1234, "step": 10547 }, { "epoch": 7.7274725274725276, "grad_norm": 1.984375, "learning_rate": 9.231529566245504e-05, "loss": 0.1241, "step": 10548 }, { "epoch": 7.728205128205128, "grad_norm": 1.1796875, "learning_rate": 9.231346192550716e-05, "loss": 0.1295, "step": 10549 }, { "epoch": 7.728937728937729, "grad_norm": 1.0234375, "learning_rate": 9.231162798821726e-05, "loss": 0.1203, "step": 10550 }, { "epoch": 7.72967032967033, "grad_norm": 0.96484375, "learning_rate": 9.230979385059413e-05, "loss": 0.1233, "step": 10551 }, { "epoch": 7.7304029304029305, "grad_norm": 1.203125, "learning_rate": 9.230795951264655e-05, "loss": 0.1537, "step": 10552 }, { "epoch": 7.731135531135531, "grad_norm": 0.72265625, "learning_rate": 9.230612497438333e-05, "loss": 0.1018, "step": 10553 }, { "epoch": 7.731868131868132, "grad_norm": 1.421875, "learning_rate": 9.230429023581322e-05, "loss": 0.1426, "step": 10554 }, { "epoch": 7.732600732600733, "grad_norm": 0.87890625, "learning_rate": 9.230245529694504e-05, "loss": 0.1079, "step": 10555 }, { "epoch": 7.733333333333333, "grad_norm": 1.2734375, "learning_rate": 9.230062015778757e-05, "loss": 0.1487, "step": 10556 }, { "epoch": 7.734065934065934, "grad_norm": 0.91015625, "learning_rate": 9.229878481834962e-05, "loss": 0.1436, "step": 10557 }, { "epoch": 7.734798534798535, "grad_norm": 0.97265625, "learning_rate": 9.229694927863995e-05, "loss": 0.1535, "step": 10558 }, { "epoch": 7.735531135531136, "grad_norm": 1.3515625, "learning_rate": 9.229511353866738e-05, "loss": 0.1769, "step": 10559 }, { "epoch": 7.736263736263736, "grad_norm": 1.21875, "learning_rate": 9.229327759844071e-05, "loss": 0.1682, "step": 10560 }, { "epoch": 7.736996336996337, "grad_norm": 1.8515625, "learning_rate": 9.229144145796873e-05, "loss": 0.1242, "step": 10561 }, { "epoch": 7.737728937728938, "grad_norm": 1.375, "learning_rate": 9.228960511726021e-05, "loss": 0.1459, "step": 10562 }, { "epoch": 7.7384615384615385, "grad_norm": 1.0625, "learning_rate": 9.228776857632398e-05, "loss": 0.1259, "step": 10563 }, { "epoch": 7.739194139194139, "grad_norm": 1.328125, "learning_rate": 9.228593183516886e-05, "loss": 0.1448, "step": 10564 }, { "epoch": 7.73992673992674, "grad_norm": 0.87109375, "learning_rate": 9.22840948938036e-05, "loss": 0.1363, "step": 10565 }, { "epoch": 7.740659340659341, "grad_norm": 1.0859375, "learning_rate": 9.228225775223704e-05, "loss": 0.1708, "step": 10566 }, { "epoch": 7.741391941391941, "grad_norm": 0.9609375, "learning_rate": 9.228042041047795e-05, "loss": 0.1167, "step": 10567 }, { "epoch": 7.742124542124542, "grad_norm": 1.2265625, "learning_rate": 9.227858286853516e-05, "loss": 0.1243, "step": 10568 }, { "epoch": 7.742857142857143, "grad_norm": 0.9453125, "learning_rate": 9.227674512641746e-05, "loss": 0.1425, "step": 10569 }, { "epoch": 7.743589743589744, "grad_norm": 1.015625, "learning_rate": 9.227490718413366e-05, "loss": 0.1671, "step": 10570 }, { "epoch": 7.744322344322344, "grad_norm": 1.2421875, "learning_rate": 9.227306904169257e-05, "loss": 0.1375, "step": 10571 }, { "epoch": 7.745054945054945, "grad_norm": 1.2890625, "learning_rate": 9.2271230699103e-05, "loss": 0.1669, "step": 10572 }, { "epoch": 7.745787545787546, "grad_norm": 1.734375, "learning_rate": 9.226939215637372e-05, "loss": 0.1657, "step": 10573 }, { "epoch": 7.7465201465201465, "grad_norm": 0.828125, "learning_rate": 9.226755341351361e-05, "loss": 0.1318, "step": 10574 }, { "epoch": 7.747252747252747, "grad_norm": 0.69921875, "learning_rate": 9.226571447053142e-05, "loss": 0.1207, "step": 10575 }, { "epoch": 7.747985347985348, "grad_norm": 0.98046875, "learning_rate": 9.226387532743599e-05, "loss": 0.152, "step": 10576 }, { "epoch": 7.748717948717949, "grad_norm": 0.765625, "learning_rate": 9.22620359842361e-05, "loss": 0.1203, "step": 10577 }, { "epoch": 7.7494505494505495, "grad_norm": 0.6875, "learning_rate": 9.226019644094061e-05, "loss": 0.1301, "step": 10578 }, { "epoch": 7.75018315018315, "grad_norm": 0.84765625, "learning_rate": 9.225835669755831e-05, "loss": 0.1126, "step": 10579 }, { "epoch": 7.750915750915751, "grad_norm": 0.890625, "learning_rate": 9.2256516754098e-05, "loss": 0.1314, "step": 10580 }, { "epoch": 7.751648351648352, "grad_norm": 1.265625, "learning_rate": 9.225467661056852e-05, "loss": 0.161, "step": 10581 }, { "epoch": 7.752380952380952, "grad_norm": 0.91796875, "learning_rate": 9.225283626697866e-05, "loss": 0.1292, "step": 10582 }, { "epoch": 7.753113553113553, "grad_norm": 1.1875, "learning_rate": 9.225099572333728e-05, "loss": 0.1561, "step": 10583 }, { "epoch": 7.753846153846154, "grad_norm": 1.2578125, "learning_rate": 9.224915497965315e-05, "loss": 0.1623, "step": 10584 }, { "epoch": 7.754578754578755, "grad_norm": 0.90625, "learning_rate": 9.224731403593514e-05, "loss": 0.1624, "step": 10585 }, { "epoch": 7.755311355311355, "grad_norm": 0.7265625, "learning_rate": 9.224547289219202e-05, "loss": 0.1063, "step": 10586 }, { "epoch": 7.756043956043956, "grad_norm": 1.125, "learning_rate": 9.224363154843265e-05, "loss": 0.1585, "step": 10587 }, { "epoch": 7.756776556776557, "grad_norm": 0.70703125, "learning_rate": 9.224179000466582e-05, "loss": 0.1093, "step": 10588 }, { "epoch": 7.7575091575091575, "grad_norm": 0.8046875, "learning_rate": 9.22399482609004e-05, "loss": 0.1158, "step": 10589 }, { "epoch": 7.758241758241758, "grad_norm": 1.171875, "learning_rate": 9.223810631714517e-05, "loss": 0.1296, "step": 10590 }, { "epoch": 7.758974358974359, "grad_norm": 0.98828125, "learning_rate": 9.223626417340895e-05, "loss": 0.1421, "step": 10591 }, { "epoch": 7.75970695970696, "grad_norm": 0.83984375, "learning_rate": 9.223442182970063e-05, "loss": 0.1293, "step": 10592 }, { "epoch": 7.76043956043956, "grad_norm": 0.9609375, "learning_rate": 9.223257928602897e-05, "loss": 0.1143, "step": 10593 }, { "epoch": 7.761172161172161, "grad_norm": 1.2578125, "learning_rate": 9.223073654240282e-05, "loss": 0.1786, "step": 10594 }, { "epoch": 7.761904761904762, "grad_norm": 0.90625, "learning_rate": 9.222889359883101e-05, "loss": 0.1526, "step": 10595 }, { "epoch": 7.762637362637363, "grad_norm": 0.8125, "learning_rate": 9.222705045532238e-05, "loss": 0.1473, "step": 10596 }, { "epoch": 7.763369963369963, "grad_norm": 1.734375, "learning_rate": 9.222520711188575e-05, "loss": 0.1697, "step": 10597 }, { "epoch": 7.764102564102564, "grad_norm": 1.0546875, "learning_rate": 9.222336356852997e-05, "loss": 0.1385, "step": 10598 }, { "epoch": 7.764835164835165, "grad_norm": 0.83203125, "learning_rate": 9.222151982526383e-05, "loss": 0.1204, "step": 10599 }, { "epoch": 7.7655677655677655, "grad_norm": 0.71875, "learning_rate": 9.221967588209623e-05, "loss": 0.1433, "step": 10600 }, { "epoch": 7.766300366300366, "grad_norm": 0.58203125, "learning_rate": 9.221783173903594e-05, "loss": 0.132, "step": 10601 }, { "epoch": 7.767032967032967, "grad_norm": 0.890625, "learning_rate": 9.221598739609182e-05, "loss": 0.1228, "step": 10602 }, { "epoch": 7.767765567765568, "grad_norm": 0.66015625, "learning_rate": 9.221414285327272e-05, "loss": 0.1402, "step": 10603 }, { "epoch": 7.768498168498168, "grad_norm": 1.6875, "learning_rate": 9.221229811058746e-05, "loss": 0.1639, "step": 10604 }, { "epoch": 7.769230769230769, "grad_norm": 0.93359375, "learning_rate": 9.221045316804489e-05, "loss": 0.1479, "step": 10605 }, { "epoch": 7.76996336996337, "grad_norm": 1.1328125, "learning_rate": 9.220860802565387e-05, "loss": 0.1315, "step": 10606 }, { "epoch": 7.770695970695971, "grad_norm": 1.0234375, "learning_rate": 9.220676268342318e-05, "loss": 0.1417, "step": 10607 }, { "epoch": 7.771428571428571, "grad_norm": 0.734375, "learning_rate": 9.220491714136172e-05, "loss": 0.1804, "step": 10608 }, { "epoch": 7.772161172161172, "grad_norm": 0.94921875, "learning_rate": 9.22030713994783e-05, "loss": 0.1412, "step": 10609 }, { "epoch": 7.772893772893773, "grad_norm": 0.88671875, "learning_rate": 9.220122545778176e-05, "loss": 0.1446, "step": 10610 }, { "epoch": 7.7736263736263735, "grad_norm": 0.96484375, "learning_rate": 9.2199379316281e-05, "loss": 0.133, "step": 10611 }, { "epoch": 7.774358974358974, "grad_norm": 1.3203125, "learning_rate": 9.21975329749848e-05, "loss": 0.1474, "step": 10612 }, { "epoch": 7.775091575091575, "grad_norm": 1.6171875, "learning_rate": 9.219568643390204e-05, "loss": 0.137, "step": 10613 }, { "epoch": 7.775824175824176, "grad_norm": 1.4375, "learning_rate": 9.219383969304156e-05, "loss": 0.1467, "step": 10614 }, { "epoch": 7.7765567765567765, "grad_norm": 1.2578125, "learning_rate": 9.219199275241218e-05, "loss": 0.1378, "step": 10615 }, { "epoch": 7.777289377289377, "grad_norm": 0.76953125, "learning_rate": 9.21901456120228e-05, "loss": 0.1486, "step": 10616 }, { "epoch": 7.778021978021978, "grad_norm": 1.0390625, "learning_rate": 9.218829827188223e-05, "loss": 0.1372, "step": 10617 }, { "epoch": 7.778754578754579, "grad_norm": 0.890625, "learning_rate": 9.218645073199936e-05, "loss": 0.1056, "step": 10618 }, { "epoch": 7.779487179487179, "grad_norm": 0.89453125, "learning_rate": 9.218460299238302e-05, "loss": 0.1098, "step": 10619 }, { "epoch": 7.78021978021978, "grad_norm": 1.0078125, "learning_rate": 9.218275505304204e-05, "loss": 0.1441, "step": 10620 }, { "epoch": 7.780952380952381, "grad_norm": 0.91015625, "learning_rate": 9.21809069139853e-05, "loss": 0.1656, "step": 10621 }, { "epoch": 7.781684981684982, "grad_norm": 0.6328125, "learning_rate": 9.217905857522167e-05, "loss": 0.1292, "step": 10622 }, { "epoch": 7.782417582417582, "grad_norm": 1.0234375, "learning_rate": 9.217721003675999e-05, "loss": 0.1462, "step": 10623 }, { "epoch": 7.783150183150183, "grad_norm": 1.015625, "learning_rate": 9.21753612986091e-05, "loss": 0.1306, "step": 10624 }, { "epoch": 7.783882783882784, "grad_norm": 1.3125, "learning_rate": 9.217351236077788e-05, "loss": 0.1394, "step": 10625 }, { "epoch": 7.7846153846153845, "grad_norm": 1.1484375, "learning_rate": 9.217166322327519e-05, "loss": 0.1326, "step": 10626 }, { "epoch": 7.785347985347985, "grad_norm": 0.6640625, "learning_rate": 9.216981388610985e-05, "loss": 0.1119, "step": 10627 }, { "epoch": 7.786080586080586, "grad_norm": 0.953125, "learning_rate": 9.216796434929078e-05, "loss": 0.158, "step": 10628 }, { "epoch": 7.786813186813187, "grad_norm": 1.515625, "learning_rate": 9.216611461282681e-05, "loss": 0.1427, "step": 10629 }, { "epoch": 7.787545787545787, "grad_norm": 0.91796875, "learning_rate": 9.216426467672679e-05, "loss": 0.1381, "step": 10630 }, { "epoch": 7.788278388278388, "grad_norm": 1.0703125, "learning_rate": 9.216241454099963e-05, "loss": 0.1167, "step": 10631 }, { "epoch": 7.789010989010989, "grad_norm": 0.921875, "learning_rate": 9.216056420565416e-05, "loss": 0.1442, "step": 10632 }, { "epoch": 7.78974358974359, "grad_norm": 0.73046875, "learning_rate": 9.215871367069923e-05, "loss": 0.1559, "step": 10633 }, { "epoch": 7.79047619047619, "grad_norm": 1.0390625, "learning_rate": 9.215686293614373e-05, "loss": 0.1423, "step": 10634 }, { "epoch": 7.791208791208791, "grad_norm": 0.7109375, "learning_rate": 9.215501200199654e-05, "loss": 0.1236, "step": 10635 }, { "epoch": 7.791941391941392, "grad_norm": 0.8515625, "learning_rate": 9.215316086826651e-05, "loss": 0.1521, "step": 10636 }, { "epoch": 7.7926739926739925, "grad_norm": 1.125, "learning_rate": 9.21513095349625e-05, "loss": 0.1313, "step": 10637 }, { "epoch": 7.793406593406593, "grad_norm": 1.0546875, "learning_rate": 9.21494580020934e-05, "loss": 0.1523, "step": 10638 }, { "epoch": 7.794139194139194, "grad_norm": 1.3359375, "learning_rate": 9.214760626966808e-05, "loss": 0.1251, "step": 10639 }, { "epoch": 7.794871794871795, "grad_norm": 1.1953125, "learning_rate": 9.21457543376954e-05, "loss": 0.151, "step": 10640 }, { "epoch": 7.795604395604395, "grad_norm": 0.8515625, "learning_rate": 9.214390220618423e-05, "loss": 0.1178, "step": 10641 }, { "epoch": 7.796336996336996, "grad_norm": 0.9140625, "learning_rate": 9.214204987514347e-05, "loss": 0.1071, "step": 10642 }, { "epoch": 7.797069597069597, "grad_norm": 1.3203125, "learning_rate": 9.214019734458198e-05, "loss": 0.1376, "step": 10643 }, { "epoch": 7.797802197802198, "grad_norm": 0.8515625, "learning_rate": 9.213834461450863e-05, "loss": 0.13, "step": 10644 }, { "epoch": 7.798534798534798, "grad_norm": 0.78125, "learning_rate": 9.213649168493228e-05, "loss": 0.1354, "step": 10645 }, { "epoch": 7.799267399267399, "grad_norm": 0.75390625, "learning_rate": 9.213463855586187e-05, "loss": 0.1064, "step": 10646 }, { "epoch": 7.8, "grad_norm": 1.0, "learning_rate": 9.213278522730622e-05, "loss": 0.1179, "step": 10647 }, { "epoch": 7.8007326007326006, "grad_norm": 1.0234375, "learning_rate": 9.213093169927422e-05, "loss": 0.1674, "step": 10648 }, { "epoch": 7.801465201465201, "grad_norm": 0.80859375, "learning_rate": 9.212907797177477e-05, "loss": 0.1434, "step": 10649 }, { "epoch": 7.802197802197802, "grad_norm": 1.1640625, "learning_rate": 9.212722404481674e-05, "loss": 0.1219, "step": 10650 }, { "epoch": 7.802930402930403, "grad_norm": 1.0390625, "learning_rate": 9.2125369918409e-05, "loss": 0.1511, "step": 10651 }, { "epoch": 7.8036630036630035, "grad_norm": 0.62890625, "learning_rate": 9.212351559256047e-05, "loss": 0.1196, "step": 10652 }, { "epoch": 7.804395604395604, "grad_norm": 1.25, "learning_rate": 9.212166106728001e-05, "loss": 0.1742, "step": 10653 }, { "epoch": 7.805128205128205, "grad_norm": 0.9453125, "learning_rate": 9.211980634257651e-05, "loss": 0.1081, "step": 10654 }, { "epoch": 7.805860805860806, "grad_norm": 1.0625, "learning_rate": 9.211795141845883e-05, "loss": 0.1318, "step": 10655 }, { "epoch": 7.806593406593406, "grad_norm": 1.5078125, "learning_rate": 9.21160962949359e-05, "loss": 0.1211, "step": 10656 }, { "epoch": 7.807326007326007, "grad_norm": 1.1484375, "learning_rate": 9.21142409720166e-05, "loss": 0.1187, "step": 10657 }, { "epoch": 7.808058608058608, "grad_norm": 1.1953125, "learning_rate": 9.211238544970979e-05, "loss": 0.1545, "step": 10658 }, { "epoch": 7.808791208791209, "grad_norm": 1.0859375, "learning_rate": 9.211052972802438e-05, "loss": 0.1516, "step": 10659 }, { "epoch": 7.809523809523809, "grad_norm": 1.0546875, "learning_rate": 9.210867380696929e-05, "loss": 0.156, "step": 10660 }, { "epoch": 7.81025641025641, "grad_norm": 1.0390625, "learning_rate": 9.210681768655337e-05, "loss": 0.1649, "step": 10661 }, { "epoch": 7.810989010989011, "grad_norm": 1.03125, "learning_rate": 9.210496136678552e-05, "loss": 0.1162, "step": 10662 }, { "epoch": 7.8117216117216115, "grad_norm": 1.5703125, "learning_rate": 9.210310484767466e-05, "loss": 0.1855, "step": 10663 }, { "epoch": 7.812454212454212, "grad_norm": 0.8046875, "learning_rate": 9.210124812922965e-05, "loss": 0.1227, "step": 10664 }, { "epoch": 7.813186813186813, "grad_norm": 0.8515625, "learning_rate": 9.209939121145941e-05, "loss": 0.1233, "step": 10665 }, { "epoch": 7.813919413919414, "grad_norm": 0.9765625, "learning_rate": 9.209753409437284e-05, "loss": 0.1238, "step": 10666 }, { "epoch": 7.814652014652014, "grad_norm": 0.80859375, "learning_rate": 9.20956767779788e-05, "loss": 0.1306, "step": 10667 }, { "epoch": 7.815384615384615, "grad_norm": 0.765625, "learning_rate": 9.209381926228624e-05, "loss": 0.1651, "step": 10668 }, { "epoch": 7.816117216117216, "grad_norm": 0.95703125, "learning_rate": 9.209196154730403e-05, "loss": 0.126, "step": 10669 }, { "epoch": 7.816849816849817, "grad_norm": 0.69921875, "learning_rate": 9.20901036330411e-05, "loss": 0.1358, "step": 10670 }, { "epoch": 7.817582417582417, "grad_norm": 0.859375, "learning_rate": 9.20882455195063e-05, "loss": 0.1546, "step": 10671 }, { "epoch": 7.818315018315018, "grad_norm": 0.88671875, "learning_rate": 9.208638720670858e-05, "loss": 0.1363, "step": 10672 }, { "epoch": 7.819047619047619, "grad_norm": 0.921875, "learning_rate": 9.208452869465681e-05, "loss": 0.1544, "step": 10673 }, { "epoch": 7.8197802197802195, "grad_norm": 0.65234375, "learning_rate": 9.208266998335993e-05, "loss": 0.1309, "step": 10674 }, { "epoch": 7.82051282051282, "grad_norm": 1.21875, "learning_rate": 9.208081107282682e-05, "loss": 0.1508, "step": 10675 }, { "epoch": 7.821245421245421, "grad_norm": 1.015625, "learning_rate": 9.20789519630664e-05, "loss": 0.1406, "step": 10676 }, { "epoch": 7.821978021978022, "grad_norm": 1.25, "learning_rate": 9.207709265408757e-05, "loss": 0.1193, "step": 10677 }, { "epoch": 7.8227106227106225, "grad_norm": 0.84765625, "learning_rate": 9.207523314589924e-05, "loss": 0.1663, "step": 10678 }, { "epoch": 7.823443223443223, "grad_norm": 1.0234375, "learning_rate": 9.20733734385103e-05, "loss": 0.1277, "step": 10679 }, { "epoch": 7.824175824175824, "grad_norm": 1.0703125, "learning_rate": 9.207151353192971e-05, "loss": 0.1467, "step": 10680 }, { "epoch": 7.824908424908425, "grad_norm": 1.7109375, "learning_rate": 9.206965342616634e-05, "loss": 0.1726, "step": 10681 }, { "epoch": 7.825641025641025, "grad_norm": 1.328125, "learning_rate": 9.206779312122912e-05, "loss": 0.1224, "step": 10682 }, { "epoch": 7.826373626373626, "grad_norm": 1.515625, "learning_rate": 9.206593261712694e-05, "loss": 0.1301, "step": 10683 }, { "epoch": 7.827106227106227, "grad_norm": 1.2890625, "learning_rate": 9.206407191386875e-05, "loss": 0.1512, "step": 10684 }, { "epoch": 7.827838827838828, "grad_norm": 0.55859375, "learning_rate": 9.206221101146343e-05, "loss": 0.1397, "step": 10685 }, { "epoch": 7.828571428571428, "grad_norm": 1.3515625, "learning_rate": 9.206034990991992e-05, "loss": 0.1778, "step": 10686 }, { "epoch": 7.829304029304029, "grad_norm": 1.2109375, "learning_rate": 9.205848860924714e-05, "loss": 0.1467, "step": 10687 }, { "epoch": 7.83003663003663, "grad_norm": 1.2578125, "learning_rate": 9.205662710945399e-05, "loss": 0.1295, "step": 10688 }, { "epoch": 7.8307692307692305, "grad_norm": 1.3359375, "learning_rate": 9.20547654105494e-05, "loss": 0.146, "step": 10689 }, { "epoch": 7.831501831501831, "grad_norm": 1.1875, "learning_rate": 9.205290351254227e-05, "loss": 0.151, "step": 10690 }, { "epoch": 7.832234432234432, "grad_norm": 0.7421875, "learning_rate": 9.205104141544157e-05, "loss": 0.1432, "step": 10691 }, { "epoch": 7.832967032967033, "grad_norm": 0.95703125, "learning_rate": 9.204917911925617e-05, "loss": 0.1403, "step": 10692 }, { "epoch": 7.833699633699633, "grad_norm": 0.9609375, "learning_rate": 9.204731662399502e-05, "loss": 0.1425, "step": 10693 }, { "epoch": 7.834432234432234, "grad_norm": 0.96875, "learning_rate": 9.204545392966704e-05, "loss": 0.1771, "step": 10694 }, { "epoch": 7.835164835164835, "grad_norm": 1.2734375, "learning_rate": 9.204359103628114e-05, "loss": 0.1322, "step": 10695 }, { "epoch": 7.835897435897436, "grad_norm": 0.9609375, "learning_rate": 9.204172794384626e-05, "loss": 0.1134, "step": 10696 }, { "epoch": 7.836630036630036, "grad_norm": 1.1875, "learning_rate": 9.203986465237133e-05, "loss": 0.1164, "step": 10697 }, { "epoch": 7.837362637362637, "grad_norm": 1.015625, "learning_rate": 9.203800116186526e-05, "loss": 0.1562, "step": 10698 }, { "epoch": 7.838095238095238, "grad_norm": 1.2109375, "learning_rate": 9.2036137472337e-05, "loss": 0.1336, "step": 10699 }, { "epoch": 7.8388278388278385, "grad_norm": 1.0625, "learning_rate": 9.203427358379548e-05, "loss": 0.1471, "step": 10700 }, { "epoch": 7.839560439560439, "grad_norm": 1.5234375, "learning_rate": 9.20324094962496e-05, "loss": 0.1302, "step": 10701 }, { "epoch": 7.84029304029304, "grad_norm": 0.76171875, "learning_rate": 9.203054520970832e-05, "loss": 0.1091, "step": 10702 }, { "epoch": 7.841025641025641, "grad_norm": 0.828125, "learning_rate": 9.202868072418058e-05, "loss": 0.1489, "step": 10703 }, { "epoch": 7.841758241758241, "grad_norm": 0.7109375, "learning_rate": 9.202681603967528e-05, "loss": 0.1182, "step": 10704 }, { "epoch": 7.842490842490842, "grad_norm": 1.25, "learning_rate": 9.202495115620137e-05, "loss": 0.1422, "step": 10705 }, { "epoch": 7.843223443223443, "grad_norm": 0.77734375, "learning_rate": 9.202308607376779e-05, "loss": 0.1287, "step": 10706 }, { "epoch": 7.843956043956044, "grad_norm": 1.8046875, "learning_rate": 9.202122079238347e-05, "loss": 0.1201, "step": 10707 }, { "epoch": 7.844688644688644, "grad_norm": 0.96875, "learning_rate": 9.201935531205737e-05, "loss": 0.1176, "step": 10708 }, { "epoch": 7.845421245421245, "grad_norm": 1.4609375, "learning_rate": 9.201748963279839e-05, "loss": 0.1224, "step": 10709 }, { "epoch": 7.846153846153846, "grad_norm": 1.9921875, "learning_rate": 9.201562375461551e-05, "loss": 0.1974, "step": 10710 }, { "epoch": 7.8468864468864465, "grad_norm": 0.59375, "learning_rate": 9.201375767751762e-05, "loss": 0.114, "step": 10711 }, { "epoch": 7.847619047619047, "grad_norm": 1.171875, "learning_rate": 9.20118914015137e-05, "loss": 0.1363, "step": 10712 }, { "epoch": 7.848351648351648, "grad_norm": 1.15625, "learning_rate": 9.201002492661269e-05, "loss": 0.1191, "step": 10713 }, { "epoch": 7.849084249084249, "grad_norm": 1.34375, "learning_rate": 9.20081582528235e-05, "loss": 0.1449, "step": 10714 }, { "epoch": 7.8498168498168495, "grad_norm": 0.96484375, "learning_rate": 9.200629138015512e-05, "loss": 0.1246, "step": 10715 }, { "epoch": 7.85054945054945, "grad_norm": 0.68359375, "learning_rate": 9.200442430861646e-05, "loss": 0.1169, "step": 10716 }, { "epoch": 7.851282051282051, "grad_norm": 1.0234375, "learning_rate": 9.200255703821649e-05, "loss": 0.1231, "step": 10717 }, { "epoch": 7.852014652014652, "grad_norm": 0.81640625, "learning_rate": 9.200068956896413e-05, "loss": 0.1528, "step": 10718 }, { "epoch": 7.852747252747252, "grad_norm": 1.0625, "learning_rate": 9.199882190086836e-05, "loss": 0.1545, "step": 10719 }, { "epoch": 7.853479853479853, "grad_norm": 1.015625, "learning_rate": 9.19969540339381e-05, "loss": 0.1498, "step": 10720 }, { "epoch": 7.854212454212454, "grad_norm": 1.34375, "learning_rate": 9.199508596818231e-05, "loss": 0.1168, "step": 10721 }, { "epoch": 7.854945054945055, "grad_norm": 0.7265625, "learning_rate": 9.199321770360995e-05, "loss": 0.1221, "step": 10722 }, { "epoch": 7.855677655677655, "grad_norm": 1.046875, "learning_rate": 9.199134924022995e-05, "loss": 0.1469, "step": 10723 }, { "epoch": 7.856410256410256, "grad_norm": 0.76953125, "learning_rate": 9.198948057805127e-05, "loss": 0.118, "step": 10724 }, { "epoch": 7.857142857142857, "grad_norm": 1.1953125, "learning_rate": 9.198761171708288e-05, "loss": 0.1421, "step": 10725 }, { "epoch": 7.8578754578754575, "grad_norm": 0.75, "learning_rate": 9.198574265733374e-05, "loss": 0.108, "step": 10726 }, { "epoch": 7.858608058608058, "grad_norm": 0.73046875, "learning_rate": 9.198387339881278e-05, "loss": 0.1177, "step": 10727 }, { "epoch": 7.859340659340659, "grad_norm": 1.0, "learning_rate": 9.198200394152894e-05, "loss": 0.1608, "step": 10728 }, { "epoch": 7.860073260073261, "grad_norm": 0.65234375, "learning_rate": 9.19801342854912e-05, "loss": 0.1317, "step": 10729 }, { "epoch": 7.860805860805861, "grad_norm": 0.91796875, "learning_rate": 9.197826443070855e-05, "loss": 0.1259, "step": 10730 }, { "epoch": 7.861538461538462, "grad_norm": 0.93359375, "learning_rate": 9.197639437718991e-05, "loss": 0.1921, "step": 10731 }, { "epoch": 7.862271062271063, "grad_norm": 0.9609375, "learning_rate": 9.197452412494425e-05, "loss": 0.1423, "step": 10732 }, { "epoch": 7.8630036630036635, "grad_norm": 1.171875, "learning_rate": 9.197265367398051e-05, "loss": 0.1265, "step": 10733 }, { "epoch": 7.863736263736264, "grad_norm": 1.46875, "learning_rate": 9.19707830243077e-05, "loss": 0.152, "step": 10734 }, { "epoch": 7.864468864468865, "grad_norm": 1.578125, "learning_rate": 9.196891217593472e-05, "loss": 0.1526, "step": 10735 }, { "epoch": 7.865201465201466, "grad_norm": 1.1640625, "learning_rate": 9.19670411288706e-05, "loss": 0.132, "step": 10736 }, { "epoch": 7.865934065934066, "grad_norm": 1.0390625, "learning_rate": 9.196516988312424e-05, "loss": 0.1403, "step": 10737 }, { "epoch": 7.866666666666667, "grad_norm": 0.66015625, "learning_rate": 9.196329843870468e-05, "loss": 0.1334, "step": 10738 }, { "epoch": 7.867399267399268, "grad_norm": 1.0546875, "learning_rate": 9.196142679562081e-05, "loss": 0.137, "step": 10739 }, { "epoch": 7.868131868131869, "grad_norm": 0.75, "learning_rate": 9.195955495388167e-05, "loss": 0.1487, "step": 10740 }, { "epoch": 7.868864468864469, "grad_norm": 1.5234375, "learning_rate": 9.195768291349617e-05, "loss": 0.1359, "step": 10741 }, { "epoch": 7.86959706959707, "grad_norm": 0.87109375, "learning_rate": 9.19558106744733e-05, "loss": 0.1618, "step": 10742 }, { "epoch": 7.870329670329671, "grad_norm": 1.0859375, "learning_rate": 9.195393823682203e-05, "loss": 0.1599, "step": 10743 }, { "epoch": 7.8710622710622715, "grad_norm": 0.9765625, "learning_rate": 9.195206560055133e-05, "loss": 0.1217, "step": 10744 }, { "epoch": 7.871794871794872, "grad_norm": 0.6953125, "learning_rate": 9.195019276567019e-05, "loss": 0.1305, "step": 10745 }, { "epoch": 7.872527472527473, "grad_norm": 0.7421875, "learning_rate": 9.194831973218757e-05, "loss": 0.1318, "step": 10746 }, { "epoch": 7.873260073260074, "grad_norm": 0.76953125, "learning_rate": 9.194644650011243e-05, "loss": 0.143, "step": 10747 }, { "epoch": 7.873992673992674, "grad_norm": 0.9921875, "learning_rate": 9.194457306945377e-05, "loss": 0.1244, "step": 10748 }, { "epoch": 7.874725274725275, "grad_norm": 1.4140625, "learning_rate": 9.194269944022054e-05, "loss": 0.1457, "step": 10749 }, { "epoch": 7.875457875457876, "grad_norm": 0.8359375, "learning_rate": 9.194082561242175e-05, "loss": 0.1099, "step": 10750 }, { "epoch": 7.876190476190477, "grad_norm": 1.0234375, "learning_rate": 9.193895158606635e-05, "loss": 0.1365, "step": 10751 }, { "epoch": 7.876923076923077, "grad_norm": 0.82421875, "learning_rate": 9.193707736116334e-05, "loss": 0.1351, "step": 10752 }, { "epoch": 7.877655677655678, "grad_norm": 0.89453125, "learning_rate": 9.193520293772167e-05, "loss": 0.1532, "step": 10753 }, { "epoch": 7.878388278388279, "grad_norm": 0.85546875, "learning_rate": 9.193332831575036e-05, "loss": 0.1307, "step": 10754 }, { "epoch": 7.8791208791208796, "grad_norm": 1.125, "learning_rate": 9.193145349525835e-05, "loss": 0.1322, "step": 10755 }, { "epoch": 7.87985347985348, "grad_norm": 0.8984375, "learning_rate": 9.192957847625467e-05, "loss": 0.1515, "step": 10756 }, { "epoch": 7.880586080586081, "grad_norm": 1.4140625, "learning_rate": 9.192770325874828e-05, "loss": 0.1709, "step": 10757 }, { "epoch": 7.881318681318682, "grad_norm": 0.79296875, "learning_rate": 9.192582784274814e-05, "loss": 0.1342, "step": 10758 }, { "epoch": 7.8820512820512825, "grad_norm": 0.8984375, "learning_rate": 9.192395222826327e-05, "loss": 0.1445, "step": 10759 }, { "epoch": 7.882783882783883, "grad_norm": 0.8828125, "learning_rate": 9.192207641530267e-05, "loss": 0.1455, "step": 10760 }, { "epoch": 7.883516483516484, "grad_norm": 0.61328125, "learning_rate": 9.192020040387527e-05, "loss": 0.117, "step": 10761 }, { "epoch": 7.884249084249085, "grad_norm": 1.296875, "learning_rate": 9.191832419399012e-05, "loss": 0.1421, "step": 10762 }, { "epoch": 7.884981684981685, "grad_norm": 1.1484375, "learning_rate": 9.191644778565615e-05, "loss": 0.1306, "step": 10763 }, { "epoch": 7.885714285714286, "grad_norm": 0.734375, "learning_rate": 9.19145711788824e-05, "loss": 0.1259, "step": 10764 }, { "epoch": 7.886446886446887, "grad_norm": 1.390625, "learning_rate": 9.191269437367786e-05, "loss": 0.1625, "step": 10765 }, { "epoch": 7.887179487179488, "grad_norm": 0.69140625, "learning_rate": 9.191081737005148e-05, "loss": 0.1354, "step": 10766 }, { "epoch": 7.887912087912088, "grad_norm": 1.1484375, "learning_rate": 9.190894016801229e-05, "loss": 0.152, "step": 10767 }, { "epoch": 7.888644688644689, "grad_norm": 0.78125, "learning_rate": 9.190706276756927e-05, "loss": 0.1701, "step": 10768 }, { "epoch": 7.88937728937729, "grad_norm": 1.0859375, "learning_rate": 9.190518516873142e-05, "loss": 0.1245, "step": 10769 }, { "epoch": 7.8901098901098905, "grad_norm": 0.625, "learning_rate": 9.190330737150774e-05, "loss": 0.141, "step": 10770 }, { "epoch": 7.890842490842491, "grad_norm": 0.921875, "learning_rate": 9.190142937590723e-05, "loss": 0.1327, "step": 10771 }, { "epoch": 7.891575091575092, "grad_norm": 1.109375, "learning_rate": 9.189955118193886e-05, "loss": 0.1441, "step": 10772 }, { "epoch": 7.892307692307693, "grad_norm": 0.71484375, "learning_rate": 9.189767278961167e-05, "loss": 0.145, "step": 10773 }, { "epoch": 7.893040293040293, "grad_norm": 1.0703125, "learning_rate": 9.189579419893463e-05, "loss": 0.1369, "step": 10774 }, { "epoch": 7.893772893772894, "grad_norm": 0.6875, "learning_rate": 9.189391540991676e-05, "loss": 0.1276, "step": 10775 }, { "epoch": 7.894505494505495, "grad_norm": 1.328125, "learning_rate": 9.189203642256702e-05, "loss": 0.1164, "step": 10776 }, { "epoch": 7.895238095238096, "grad_norm": 0.95703125, "learning_rate": 9.189015723689449e-05, "loss": 0.127, "step": 10777 }, { "epoch": 7.895970695970696, "grad_norm": 0.8203125, "learning_rate": 9.18882778529081e-05, "loss": 0.1147, "step": 10778 }, { "epoch": 7.896703296703297, "grad_norm": 0.74609375, "learning_rate": 9.18863982706169e-05, "loss": 0.115, "step": 10779 }, { "epoch": 7.897435897435898, "grad_norm": 0.75, "learning_rate": 9.188451849002987e-05, "loss": 0.1169, "step": 10780 }, { "epoch": 7.8981684981684985, "grad_norm": 0.92578125, "learning_rate": 9.188263851115603e-05, "loss": 0.12, "step": 10781 }, { "epoch": 7.898901098901099, "grad_norm": 1.1171875, "learning_rate": 9.188075833400438e-05, "loss": 0.1396, "step": 10782 }, { "epoch": 7.8996336996337, "grad_norm": 1.0859375, "learning_rate": 9.187887795858393e-05, "loss": 0.1337, "step": 10783 }, { "epoch": 7.900366300366301, "grad_norm": 0.87890625, "learning_rate": 9.18769973849037e-05, "loss": 0.1393, "step": 10784 }, { "epoch": 7.9010989010989015, "grad_norm": 1.375, "learning_rate": 9.187511661297269e-05, "loss": 0.1453, "step": 10785 }, { "epoch": 7.901831501831502, "grad_norm": 1.09375, "learning_rate": 9.18732356427999e-05, "loss": 0.1439, "step": 10786 }, { "epoch": 7.902564102564103, "grad_norm": 1.3515625, "learning_rate": 9.187135447439438e-05, "loss": 0.1402, "step": 10787 }, { "epoch": 7.903296703296704, "grad_norm": 1.0859375, "learning_rate": 9.186947310776512e-05, "loss": 0.1567, "step": 10788 }, { "epoch": 7.904029304029304, "grad_norm": 0.93359375, "learning_rate": 9.18675915429211e-05, "loss": 0.1582, "step": 10789 }, { "epoch": 7.904761904761905, "grad_norm": 0.5625, "learning_rate": 9.18657097798714e-05, "loss": 0.1352, "step": 10790 }, { "epoch": 7.905494505494506, "grad_norm": 0.83984375, "learning_rate": 9.1863827818625e-05, "loss": 0.1184, "step": 10791 }, { "epoch": 7.906227106227107, "grad_norm": 1.390625, "learning_rate": 9.186194565919091e-05, "loss": 0.1723, "step": 10792 }, { "epoch": 7.906959706959707, "grad_norm": 1.078125, "learning_rate": 9.186006330157816e-05, "loss": 0.1591, "step": 10793 }, { "epoch": 7.907692307692308, "grad_norm": 0.94921875, "learning_rate": 9.185818074579579e-05, "loss": 0.1307, "step": 10794 }, { "epoch": 7.908424908424909, "grad_norm": 0.94140625, "learning_rate": 9.185629799185277e-05, "loss": 0.1575, "step": 10795 }, { "epoch": 7.9091575091575095, "grad_norm": 1.0390625, "learning_rate": 9.185441503975818e-05, "loss": 0.1198, "step": 10796 }, { "epoch": 7.90989010989011, "grad_norm": 0.439453125, "learning_rate": 9.1852531889521e-05, "loss": 0.1313, "step": 10797 }, { "epoch": 7.910622710622711, "grad_norm": 1.140625, "learning_rate": 9.185064854115026e-05, "loss": 0.1202, "step": 10798 }, { "epoch": 7.911355311355312, "grad_norm": 1.3203125, "learning_rate": 9.184876499465499e-05, "loss": 0.1274, "step": 10799 }, { "epoch": 7.912087912087912, "grad_norm": 1.0234375, "learning_rate": 9.184688125004421e-05, "loss": 0.1974, "step": 10800 }, { "epoch": 7.912820512820513, "grad_norm": 0.8828125, "learning_rate": 9.184499730732697e-05, "loss": 0.152, "step": 10801 }, { "epoch": 7.913553113553114, "grad_norm": 0.87890625, "learning_rate": 9.184311316651225e-05, "loss": 0.1365, "step": 10802 }, { "epoch": 7.914285714285715, "grad_norm": 0.70703125, "learning_rate": 9.18412288276091e-05, "loss": 0.1291, "step": 10803 }, { "epoch": 7.915018315018315, "grad_norm": 1.3203125, "learning_rate": 9.183934429062657e-05, "loss": 0.1272, "step": 10804 }, { "epoch": 7.915750915750916, "grad_norm": 0.734375, "learning_rate": 9.183745955557366e-05, "loss": 0.1259, "step": 10805 }, { "epoch": 7.916483516483517, "grad_norm": 1.5703125, "learning_rate": 9.183557462245941e-05, "loss": 0.1758, "step": 10806 }, { "epoch": 7.9172161172161175, "grad_norm": 0.921875, "learning_rate": 9.183368949129286e-05, "loss": 0.1501, "step": 10807 }, { "epoch": 7.917948717948718, "grad_norm": 0.921875, "learning_rate": 9.183180416208303e-05, "loss": 0.131, "step": 10808 }, { "epoch": 7.918681318681319, "grad_norm": 0.96484375, "learning_rate": 9.182991863483897e-05, "loss": 0.1217, "step": 10809 }, { "epoch": 7.91941391941392, "grad_norm": 1.25, "learning_rate": 9.182803290956967e-05, "loss": 0.12, "step": 10810 }, { "epoch": 7.92014652014652, "grad_norm": 0.68359375, "learning_rate": 9.182614698628424e-05, "loss": 0.123, "step": 10811 }, { "epoch": 7.920879120879121, "grad_norm": 1.2578125, "learning_rate": 9.182426086499163e-05, "loss": 0.1584, "step": 10812 }, { "epoch": 7.921611721611722, "grad_norm": 0.8203125, "learning_rate": 9.182237454570095e-05, "loss": 0.1552, "step": 10813 }, { "epoch": 7.922344322344323, "grad_norm": 1.6171875, "learning_rate": 9.18204880284212e-05, "loss": 0.1754, "step": 10814 }, { "epoch": 7.923076923076923, "grad_norm": 0.92578125, "learning_rate": 9.181860131316142e-05, "loss": 0.1146, "step": 10815 }, { "epoch": 7.923809523809524, "grad_norm": 1.1484375, "learning_rate": 9.181671439993067e-05, "loss": 0.1445, "step": 10816 }, { "epoch": 7.924542124542125, "grad_norm": 0.73828125, "learning_rate": 9.181482728873797e-05, "loss": 0.1362, "step": 10817 }, { "epoch": 7.9252747252747255, "grad_norm": 0.82421875, "learning_rate": 9.181293997959237e-05, "loss": 0.12, "step": 10818 }, { "epoch": 7.926007326007326, "grad_norm": 1.1328125, "learning_rate": 9.181105247250291e-05, "loss": 0.1359, "step": 10819 }, { "epoch": 7.926739926739927, "grad_norm": 1.8671875, "learning_rate": 9.180916476747865e-05, "loss": 0.1559, "step": 10820 }, { "epoch": 7.927472527472528, "grad_norm": 0.91796875, "learning_rate": 9.180727686452861e-05, "loss": 0.1219, "step": 10821 }, { "epoch": 7.9282051282051285, "grad_norm": 1.296875, "learning_rate": 9.180538876366185e-05, "loss": 0.193, "step": 10822 }, { "epoch": 7.928937728937729, "grad_norm": 1.1171875, "learning_rate": 9.18035004648874e-05, "loss": 0.1047, "step": 10823 }, { "epoch": 7.92967032967033, "grad_norm": 1.25, "learning_rate": 9.180161196821433e-05, "loss": 0.1421, "step": 10824 }, { "epoch": 7.930402930402931, "grad_norm": 1.4765625, "learning_rate": 9.179972327365168e-05, "loss": 0.1373, "step": 10825 }, { "epoch": 7.931135531135531, "grad_norm": 0.88671875, "learning_rate": 9.17978343812085e-05, "loss": 0.1276, "step": 10826 }, { "epoch": 7.931868131868132, "grad_norm": 1.6484375, "learning_rate": 9.179594529089383e-05, "loss": 0.1733, "step": 10827 }, { "epoch": 7.932600732600733, "grad_norm": 1.3125, "learning_rate": 9.179405600271673e-05, "loss": 0.1368, "step": 10828 }, { "epoch": 7.933333333333334, "grad_norm": 0.8671875, "learning_rate": 9.179216651668627e-05, "loss": 0.1221, "step": 10829 }, { "epoch": 7.934065934065934, "grad_norm": 1.6484375, "learning_rate": 9.179027683281146e-05, "loss": 0.1626, "step": 10830 }, { "epoch": 7.934798534798535, "grad_norm": 2.140625, "learning_rate": 9.178838695110139e-05, "loss": 0.172, "step": 10831 }, { "epoch": 7.935531135531136, "grad_norm": 1.3046875, "learning_rate": 9.178649687156511e-05, "loss": 0.1148, "step": 10832 }, { "epoch": 7.9362637362637365, "grad_norm": 1.2265625, "learning_rate": 9.178460659421166e-05, "loss": 0.1476, "step": 10833 }, { "epoch": 7.936996336996337, "grad_norm": 0.74609375, "learning_rate": 9.178271611905011e-05, "loss": 0.141, "step": 10834 }, { "epoch": 7.937728937728938, "grad_norm": 1.0625, "learning_rate": 9.178082544608952e-05, "loss": 0.1681, "step": 10835 }, { "epoch": 7.938461538461539, "grad_norm": 1.046875, "learning_rate": 9.177893457533894e-05, "loss": 0.1185, "step": 10836 }, { "epoch": 7.939194139194139, "grad_norm": 0.78515625, "learning_rate": 9.177704350680744e-05, "loss": 0.1252, "step": 10837 }, { "epoch": 7.93992673992674, "grad_norm": 0.97265625, "learning_rate": 9.177515224050405e-05, "loss": 0.1442, "step": 10838 }, { "epoch": 7.940659340659341, "grad_norm": 1.484375, "learning_rate": 9.177326077643787e-05, "loss": 0.1431, "step": 10839 }, { "epoch": 7.941391941391942, "grad_norm": 1.1328125, "learning_rate": 9.177136911461797e-05, "loss": 0.1198, "step": 10840 }, { "epoch": 7.942124542124542, "grad_norm": 0.74609375, "learning_rate": 9.176947725505337e-05, "loss": 0.1571, "step": 10841 }, { "epoch": 7.942857142857143, "grad_norm": 0.78125, "learning_rate": 9.176758519775316e-05, "loss": 0.1338, "step": 10842 }, { "epoch": 7.943589743589744, "grad_norm": 1.546875, "learning_rate": 9.17656929427264e-05, "loss": 0.1784, "step": 10843 }, { "epoch": 7.9443223443223445, "grad_norm": 1.109375, "learning_rate": 9.176380048998217e-05, "loss": 0.1601, "step": 10844 }, { "epoch": 7.945054945054945, "grad_norm": 0.734375, "learning_rate": 9.176190783952951e-05, "loss": 0.1305, "step": 10845 }, { "epoch": 7.945787545787546, "grad_norm": 1.0234375, "learning_rate": 9.176001499137751e-05, "loss": 0.1343, "step": 10846 }, { "epoch": 7.946520146520147, "grad_norm": 0.71484375, "learning_rate": 9.175812194553525e-05, "loss": 0.1193, "step": 10847 }, { "epoch": 7.947252747252747, "grad_norm": 0.88671875, "learning_rate": 9.175622870201175e-05, "loss": 0.1173, "step": 10848 }, { "epoch": 7.947985347985348, "grad_norm": 0.9140625, "learning_rate": 9.175433526081614e-05, "loss": 0.1369, "step": 10849 }, { "epoch": 7.948717948717949, "grad_norm": 1.0703125, "learning_rate": 9.175244162195745e-05, "loss": 0.1382, "step": 10850 }, { "epoch": 7.94945054945055, "grad_norm": 1.0078125, "learning_rate": 9.175054778544478e-05, "loss": 0.1302, "step": 10851 }, { "epoch": 7.95018315018315, "grad_norm": 1.5703125, "learning_rate": 9.174865375128721e-05, "loss": 0.1555, "step": 10852 }, { "epoch": 7.950915750915751, "grad_norm": 1.15625, "learning_rate": 9.174675951949377e-05, "loss": 0.1622, "step": 10853 }, { "epoch": 7.951648351648352, "grad_norm": 2.21875, "learning_rate": 9.174486509007357e-05, "loss": 0.1248, "step": 10854 }, { "epoch": 7.9523809523809526, "grad_norm": 0.98046875, "learning_rate": 9.17429704630357e-05, "loss": 0.1146, "step": 10855 }, { "epoch": 7.953113553113553, "grad_norm": 2.765625, "learning_rate": 9.17410756383892e-05, "loss": 0.1578, "step": 10856 }, { "epoch": 7.953846153846154, "grad_norm": 1.546875, "learning_rate": 9.173918061614317e-05, "loss": 0.1516, "step": 10857 }, { "epoch": 7.954578754578755, "grad_norm": 0.765625, "learning_rate": 9.173728539630669e-05, "loss": 0.1687, "step": 10858 }, { "epoch": 7.9553113553113555, "grad_norm": 1.1953125, "learning_rate": 9.173538997888885e-05, "loss": 0.1658, "step": 10859 }, { "epoch": 7.956043956043956, "grad_norm": 1.9921875, "learning_rate": 9.17334943638987e-05, "loss": 0.1736, "step": 10860 }, { "epoch": 7.956776556776557, "grad_norm": 1.9140625, "learning_rate": 9.173159855134535e-05, "loss": 0.149, "step": 10861 }, { "epoch": 7.957509157509158, "grad_norm": 2.109375, "learning_rate": 9.172970254123789e-05, "loss": 0.168, "step": 10862 }, { "epoch": 7.958241758241758, "grad_norm": 1.0546875, "learning_rate": 9.172780633358539e-05, "loss": 0.1299, "step": 10863 }, { "epoch": 7.958974358974359, "grad_norm": 0.70703125, "learning_rate": 9.172590992839692e-05, "loss": 0.1108, "step": 10864 }, { "epoch": 7.95970695970696, "grad_norm": 1.609375, "learning_rate": 9.172401332568159e-05, "loss": 0.1448, "step": 10865 }, { "epoch": 7.960439560439561, "grad_norm": 0.9765625, "learning_rate": 9.172211652544847e-05, "loss": 0.1345, "step": 10866 }, { "epoch": 7.961172161172161, "grad_norm": 1.421875, "learning_rate": 9.172021952770668e-05, "loss": 0.1677, "step": 10867 }, { "epoch": 7.961904761904762, "grad_norm": 1.3046875, "learning_rate": 9.171832233246526e-05, "loss": 0.133, "step": 10868 }, { "epoch": 7.962637362637363, "grad_norm": 0.7890625, "learning_rate": 9.171642493973333e-05, "loss": 0.1333, "step": 10869 }, { "epoch": 7.9633699633699635, "grad_norm": 0.984375, "learning_rate": 9.171452734951999e-05, "loss": 0.1112, "step": 10870 }, { "epoch": 7.964102564102564, "grad_norm": 1.578125, "learning_rate": 9.171262956183432e-05, "loss": 0.1692, "step": 10871 }, { "epoch": 7.964835164835165, "grad_norm": 1.3203125, "learning_rate": 9.171073157668542e-05, "loss": 0.1361, "step": 10872 }, { "epoch": 7.965567765567766, "grad_norm": 0.70703125, "learning_rate": 9.170883339408235e-05, "loss": 0.1176, "step": 10873 }, { "epoch": 7.966300366300366, "grad_norm": 0.7890625, "learning_rate": 9.170693501403425e-05, "loss": 0.145, "step": 10874 }, { "epoch": 7.967032967032967, "grad_norm": 1.15625, "learning_rate": 9.170503643655019e-05, "loss": 0.1388, "step": 10875 }, { "epoch": 7.967765567765568, "grad_norm": 0.73828125, "learning_rate": 9.170313766163928e-05, "loss": 0.1259, "step": 10876 }, { "epoch": 7.968498168498169, "grad_norm": 1.125, "learning_rate": 9.170123868931061e-05, "loss": 0.123, "step": 10877 }, { "epoch": 7.969230769230769, "grad_norm": 1.4453125, "learning_rate": 9.169933951957327e-05, "loss": 0.1615, "step": 10878 }, { "epoch": 7.96996336996337, "grad_norm": 1.5234375, "learning_rate": 9.169744015243636e-05, "loss": 0.13, "step": 10879 }, { "epoch": 7.970695970695971, "grad_norm": 1.0234375, "learning_rate": 9.169554058790903e-05, "loss": 0.1373, "step": 10880 }, { "epoch": 7.9714285714285715, "grad_norm": 1.203125, "learning_rate": 9.16936408260003e-05, "loss": 0.1572, "step": 10881 }, { "epoch": 7.972161172161172, "grad_norm": 0.95703125, "learning_rate": 9.169174086671933e-05, "loss": 0.1671, "step": 10882 }, { "epoch": 7.972893772893773, "grad_norm": 1.1015625, "learning_rate": 9.168984071007521e-05, "loss": 0.1512, "step": 10883 }, { "epoch": 7.973626373626374, "grad_norm": 0.90234375, "learning_rate": 9.168794035607705e-05, "loss": 0.1471, "step": 10884 }, { "epoch": 7.9743589743589745, "grad_norm": 0.89453125, "learning_rate": 9.168603980473393e-05, "loss": 0.1258, "step": 10885 }, { "epoch": 7.975091575091575, "grad_norm": 0.546875, "learning_rate": 9.1684139056055e-05, "loss": 0.1152, "step": 10886 }, { "epoch": 7.975824175824176, "grad_norm": 1.328125, "learning_rate": 9.16822381100493e-05, "loss": 0.1376, "step": 10887 }, { "epoch": 7.976556776556777, "grad_norm": 1.4453125, "learning_rate": 9.1680336966726e-05, "loss": 0.1437, "step": 10888 }, { "epoch": 7.977289377289377, "grad_norm": 1.1484375, "learning_rate": 9.16784356260942e-05, "loss": 0.1235, "step": 10889 }, { "epoch": 7.978021978021978, "grad_norm": 1.0625, "learning_rate": 9.167653408816297e-05, "loss": 0.1424, "step": 10890 }, { "epoch": 7.978754578754579, "grad_norm": 0.671875, "learning_rate": 9.167463235294146e-05, "loss": 0.1164, "step": 10891 }, { "epoch": 7.97948717948718, "grad_norm": 0.76953125, "learning_rate": 9.167273042043876e-05, "loss": 0.1206, "step": 10892 }, { "epoch": 7.98021978021978, "grad_norm": 1.609375, "learning_rate": 9.1670828290664e-05, "loss": 0.1643, "step": 10893 }, { "epoch": 7.980952380952381, "grad_norm": 1.0546875, "learning_rate": 9.166892596362631e-05, "loss": 0.1566, "step": 10894 }, { "epoch": 7.981684981684982, "grad_norm": 1.546875, "learning_rate": 9.166702343933474e-05, "loss": 0.1424, "step": 10895 }, { "epoch": 7.9824175824175825, "grad_norm": 0.94921875, "learning_rate": 9.166512071779847e-05, "loss": 0.119, "step": 10896 }, { "epoch": 7.983150183150183, "grad_norm": 0.50390625, "learning_rate": 9.16632177990266e-05, "loss": 0.1548, "step": 10897 }, { "epoch": 7.983882783882784, "grad_norm": 1.3359375, "learning_rate": 9.166131468302821e-05, "loss": 0.1254, "step": 10898 }, { "epoch": 7.984615384615385, "grad_norm": 1.2265625, "learning_rate": 9.165941136981248e-05, "loss": 0.1052, "step": 10899 }, { "epoch": 7.985347985347985, "grad_norm": 1.296875, "learning_rate": 9.165750785938849e-05, "loss": 0.1509, "step": 10900 }, { "epoch": 7.986080586080586, "grad_norm": 1.109375, "learning_rate": 9.165560415176536e-05, "loss": 0.098, "step": 10901 }, { "epoch": 7.986813186813187, "grad_norm": 1.03125, "learning_rate": 9.165370024695222e-05, "loss": 0.1491, "step": 10902 }, { "epoch": 7.987545787545788, "grad_norm": 0.92578125, "learning_rate": 9.165179614495821e-05, "loss": 0.1398, "step": 10903 }, { "epoch": 7.988278388278388, "grad_norm": 1.2265625, "learning_rate": 9.164989184579241e-05, "loss": 0.1354, "step": 10904 }, { "epoch": 7.989010989010989, "grad_norm": 0.8359375, "learning_rate": 9.164798734946399e-05, "loss": 0.1364, "step": 10905 }, { "epoch": 7.98974358974359, "grad_norm": 1.5078125, "learning_rate": 9.164608265598203e-05, "loss": 0.1581, "step": 10906 }, { "epoch": 7.9904761904761905, "grad_norm": 1.109375, "learning_rate": 9.16441777653557e-05, "loss": 0.1121, "step": 10907 }, { "epoch": 7.991208791208791, "grad_norm": 1.0078125, "learning_rate": 9.164227267759411e-05, "loss": 0.1418, "step": 10908 }, { "epoch": 7.991941391941392, "grad_norm": 0.69140625, "learning_rate": 9.164036739270638e-05, "loss": 0.1286, "step": 10909 }, { "epoch": 7.992673992673993, "grad_norm": 0.85546875, "learning_rate": 9.163846191070163e-05, "loss": 0.1227, "step": 10910 }, { "epoch": 7.993406593406593, "grad_norm": 1.140625, "learning_rate": 9.163655623158903e-05, "loss": 0.1574, "step": 10911 }, { "epoch": 7.994139194139194, "grad_norm": 1.2578125, "learning_rate": 9.163465035537768e-05, "loss": 0.1511, "step": 10912 }, { "epoch": 7.994871794871795, "grad_norm": 1.5625, "learning_rate": 9.16327442820767e-05, "loss": 0.1312, "step": 10913 }, { "epoch": 7.995604395604396, "grad_norm": 0.5703125, "learning_rate": 9.163083801169525e-05, "loss": 0.1369, "step": 10914 }, { "epoch": 7.996336996336996, "grad_norm": 1.1015625, "learning_rate": 9.162893154424246e-05, "loss": 0.1395, "step": 10915 }, { "epoch": 7.997069597069597, "grad_norm": 1.15625, "learning_rate": 9.162702487972744e-05, "loss": 0.1763, "step": 10916 }, { "epoch": 7.997802197802198, "grad_norm": 0.89453125, "learning_rate": 9.162511801815936e-05, "loss": 0.11, "step": 10917 }, { "epoch": 7.9985347985347985, "grad_norm": 0.9921875, "learning_rate": 9.162321095954732e-05, "loss": 0.1415, "step": 10918 }, { "epoch": 7.999267399267399, "grad_norm": 1.0234375, "learning_rate": 9.16213037039005e-05, "loss": 0.117, "step": 10919 }, { "epoch": 8.0, "grad_norm": 0.90625, "learning_rate": 9.1619396251228e-05, "loss": 0.1399, "step": 10920 }, { "epoch": 8.0007326007326, "grad_norm": 1.203125, "learning_rate": 9.161748860153898e-05, "loss": 0.1777, "step": 10921 }, { "epoch": 8.001465201465201, "grad_norm": 1.484375, "learning_rate": 9.161558075484255e-05, "loss": 0.1635, "step": 10922 }, { "epoch": 8.002197802197802, "grad_norm": 0.6875, "learning_rate": 9.161367271114791e-05, "loss": 0.1248, "step": 10923 }, { "epoch": 8.002930402930403, "grad_norm": 1.2265625, "learning_rate": 9.161176447046416e-05, "loss": 0.1426, "step": 10924 }, { "epoch": 8.003663003663004, "grad_norm": 0.96484375, "learning_rate": 9.160985603280043e-05, "loss": 0.145, "step": 10925 }, { "epoch": 8.004395604395604, "grad_norm": 0.62109375, "learning_rate": 9.16079473981659e-05, "loss": 0.1, "step": 10926 }, { "epoch": 8.005128205128205, "grad_norm": 1.34375, "learning_rate": 9.16060385665697e-05, "loss": 0.1684, "step": 10927 }, { "epoch": 8.005860805860806, "grad_norm": 1.265625, "learning_rate": 9.160412953802096e-05, "loss": 0.1386, "step": 10928 }, { "epoch": 8.006593406593407, "grad_norm": 1.0859375, "learning_rate": 9.160222031252886e-05, "loss": 0.1429, "step": 10929 }, { "epoch": 8.007326007326007, "grad_norm": 1.1640625, "learning_rate": 9.160031089010251e-05, "loss": 0.1446, "step": 10930 }, { "epoch": 8.008058608058608, "grad_norm": 0.890625, "learning_rate": 9.15984012707511e-05, "loss": 0.1274, "step": 10931 }, { "epoch": 8.008791208791209, "grad_norm": 1.3125, "learning_rate": 9.159649145448375e-05, "loss": 0.1412, "step": 10932 }, { "epoch": 8.00952380952381, "grad_norm": 1.4609375, "learning_rate": 9.159458144130961e-05, "loss": 0.1456, "step": 10933 }, { "epoch": 8.01025641025641, "grad_norm": 1.4296875, "learning_rate": 9.159267123123784e-05, "loss": 0.1699, "step": 10934 }, { "epoch": 8.010989010989011, "grad_norm": 1.2578125, "learning_rate": 9.159076082427761e-05, "loss": 0.1363, "step": 10935 }, { "epoch": 8.011721611721612, "grad_norm": 0.88671875, "learning_rate": 9.158885022043804e-05, "loss": 0.1543, "step": 10936 }, { "epoch": 8.012454212454212, "grad_norm": 1.1015625, "learning_rate": 9.158693941972831e-05, "loss": 0.1567, "step": 10937 }, { "epoch": 8.013186813186813, "grad_norm": 1.1875, "learning_rate": 9.158502842215756e-05, "loss": 0.129, "step": 10938 }, { "epoch": 8.013919413919414, "grad_norm": 1.765625, "learning_rate": 9.158311722773495e-05, "loss": 0.1324, "step": 10939 }, { "epoch": 8.014652014652015, "grad_norm": 1.234375, "learning_rate": 9.158120583646964e-05, "loss": 0.1393, "step": 10940 }, { "epoch": 8.015384615384615, "grad_norm": 1.0078125, "learning_rate": 9.15792942483708e-05, "loss": 0.1432, "step": 10941 }, { "epoch": 8.016117216117216, "grad_norm": 1.015625, "learning_rate": 9.157738246344754e-05, "loss": 0.1613, "step": 10942 }, { "epoch": 8.016849816849817, "grad_norm": 1.796875, "learning_rate": 9.15754704817091e-05, "loss": 0.1741, "step": 10943 }, { "epoch": 8.017582417582418, "grad_norm": 1.421875, "learning_rate": 9.157355830316456e-05, "loss": 0.1386, "step": 10944 }, { "epoch": 8.018315018315018, "grad_norm": 0.75, "learning_rate": 9.157164592782313e-05, "loss": 0.1153, "step": 10945 }, { "epoch": 8.019047619047619, "grad_norm": 0.921875, "learning_rate": 9.156973335569398e-05, "loss": 0.1244, "step": 10946 }, { "epoch": 8.01978021978022, "grad_norm": 1.0703125, "learning_rate": 9.156782058678622e-05, "loss": 0.1288, "step": 10947 }, { "epoch": 8.02051282051282, "grad_norm": 1.6015625, "learning_rate": 9.156590762110908e-05, "loss": 0.1697, "step": 10948 }, { "epoch": 8.021245421245421, "grad_norm": 0.8984375, "learning_rate": 9.156399445867169e-05, "loss": 0.1367, "step": 10949 }, { "epoch": 8.021978021978022, "grad_norm": 1.6640625, "learning_rate": 9.156208109948321e-05, "loss": 0.1735, "step": 10950 }, { "epoch": 8.022710622710623, "grad_norm": 0.94140625, "learning_rate": 9.156016754355283e-05, "loss": 0.1455, "step": 10951 }, { "epoch": 8.023443223443223, "grad_norm": 1.21875, "learning_rate": 9.155825379088972e-05, "loss": 0.1056, "step": 10952 }, { "epoch": 8.024175824175824, "grad_norm": 1.046875, "learning_rate": 9.155633984150302e-05, "loss": 0.1223, "step": 10953 }, { "epoch": 8.024908424908425, "grad_norm": 1.1171875, "learning_rate": 9.155442569540191e-05, "loss": 0.1373, "step": 10954 }, { "epoch": 8.025641025641026, "grad_norm": 1.453125, "learning_rate": 9.155251135259558e-05, "loss": 0.1329, "step": 10955 }, { "epoch": 8.026373626373626, "grad_norm": 1.2890625, "learning_rate": 9.15505968130932e-05, "loss": 0.1378, "step": 10956 }, { "epoch": 8.027106227106227, "grad_norm": 1.5546875, "learning_rate": 9.154868207690393e-05, "loss": 0.1572, "step": 10957 }, { "epoch": 8.027838827838828, "grad_norm": 2.15625, "learning_rate": 9.154676714403694e-05, "loss": 0.1473, "step": 10958 }, { "epoch": 8.028571428571428, "grad_norm": 1.703125, "learning_rate": 9.154485201450142e-05, "loss": 0.1427, "step": 10959 }, { "epoch": 8.02930402930403, "grad_norm": 0.70703125, "learning_rate": 9.154293668830653e-05, "loss": 0.1379, "step": 10960 }, { "epoch": 8.03003663003663, "grad_norm": 0.7421875, "learning_rate": 9.154102116546148e-05, "loss": 0.1457, "step": 10961 }, { "epoch": 8.03076923076923, "grad_norm": 1.3125, "learning_rate": 9.15391054459754e-05, "loss": 0.1172, "step": 10962 }, { "epoch": 8.031501831501831, "grad_norm": 0.8671875, "learning_rate": 9.153718952985751e-05, "loss": 0.1473, "step": 10963 }, { "epoch": 8.032234432234432, "grad_norm": 1.0234375, "learning_rate": 9.153527341711696e-05, "loss": 0.1409, "step": 10964 }, { "epoch": 8.032967032967033, "grad_norm": 0.99609375, "learning_rate": 9.153335710776295e-05, "loss": 0.1533, "step": 10965 }, { "epoch": 8.033699633699634, "grad_norm": 1.8359375, "learning_rate": 9.153144060180466e-05, "loss": 0.137, "step": 10966 }, { "epoch": 8.034432234432234, "grad_norm": 0.8984375, "learning_rate": 9.152952389925126e-05, "loss": 0.1084, "step": 10967 }, { "epoch": 8.035164835164835, "grad_norm": 1.1328125, "learning_rate": 9.152760700011196e-05, "loss": 0.1264, "step": 10968 }, { "epoch": 8.035897435897436, "grad_norm": 0.88671875, "learning_rate": 9.15256899043959e-05, "loss": 0.1379, "step": 10969 }, { "epoch": 8.036630036630036, "grad_norm": 0.8671875, "learning_rate": 9.152377261211232e-05, "loss": 0.121, "step": 10970 }, { "epoch": 8.037362637362637, "grad_norm": 0.58984375, "learning_rate": 9.152185512327036e-05, "loss": 0.1273, "step": 10971 }, { "epoch": 8.038095238095238, "grad_norm": 1.21875, "learning_rate": 9.151993743787922e-05, "loss": 0.1425, "step": 10972 }, { "epoch": 8.038827838827839, "grad_norm": 0.69921875, "learning_rate": 9.15180195559481e-05, "loss": 0.1042, "step": 10973 }, { "epoch": 8.03956043956044, "grad_norm": 0.73046875, "learning_rate": 9.15161014774862e-05, "loss": 0.1424, "step": 10974 }, { "epoch": 8.04029304029304, "grad_norm": 1.5625, "learning_rate": 9.151418320250267e-05, "loss": 0.1753, "step": 10975 }, { "epoch": 8.04102564102564, "grad_norm": 0.984375, "learning_rate": 9.151226473100673e-05, "loss": 0.1377, "step": 10976 }, { "epoch": 8.041758241758242, "grad_norm": 0.96484375, "learning_rate": 9.151034606300758e-05, "loss": 0.1317, "step": 10977 }, { "epoch": 8.042490842490842, "grad_norm": 0.7421875, "learning_rate": 9.150842719851439e-05, "loss": 0.1489, "step": 10978 }, { "epoch": 8.043223443223443, "grad_norm": 0.9609375, "learning_rate": 9.150650813753635e-05, "loss": 0.1141, "step": 10979 }, { "epoch": 8.043956043956044, "grad_norm": 0.79296875, "learning_rate": 9.150458888008269e-05, "loss": 0.1281, "step": 10980 }, { "epoch": 8.044688644688645, "grad_norm": 1.0703125, "learning_rate": 9.150266942616258e-05, "loss": 0.1089, "step": 10981 }, { "epoch": 8.045421245421245, "grad_norm": 0.7734375, "learning_rate": 9.15007497757852e-05, "loss": 0.1274, "step": 10982 }, { "epoch": 8.046153846153846, "grad_norm": 1.0703125, "learning_rate": 9.149882992895978e-05, "loss": 0.1205, "step": 10983 }, { "epoch": 8.046886446886447, "grad_norm": 0.71875, "learning_rate": 9.149690988569551e-05, "loss": 0.1373, "step": 10984 }, { "epoch": 8.047619047619047, "grad_norm": 0.96875, "learning_rate": 9.14949896460016e-05, "loss": 0.1465, "step": 10985 }, { "epoch": 8.048351648351648, "grad_norm": 1.125, "learning_rate": 9.149306920988721e-05, "loss": 0.1879, "step": 10986 }, { "epoch": 8.049084249084249, "grad_norm": 1.0390625, "learning_rate": 9.149114857736159e-05, "loss": 0.1631, "step": 10987 }, { "epoch": 8.04981684981685, "grad_norm": 0.96484375, "learning_rate": 9.14892277484339e-05, "loss": 0.1389, "step": 10988 }, { "epoch": 8.05054945054945, "grad_norm": 1.75, "learning_rate": 9.148730672311338e-05, "loss": 0.1723, "step": 10989 }, { "epoch": 8.051282051282051, "grad_norm": 0.9765625, "learning_rate": 9.148538550140922e-05, "loss": 0.146, "step": 10990 }, { "epoch": 8.052014652014652, "grad_norm": 1.3046875, "learning_rate": 9.148346408333062e-05, "loss": 0.1457, "step": 10991 }, { "epoch": 8.052747252747253, "grad_norm": 1.1796875, "learning_rate": 9.148154246888677e-05, "loss": 0.1936, "step": 10992 }, { "epoch": 8.053479853479853, "grad_norm": 1.3515625, "learning_rate": 9.147962065808692e-05, "loss": 0.1085, "step": 10993 }, { "epoch": 8.054212454212454, "grad_norm": 1.078125, "learning_rate": 9.147769865094026e-05, "loss": 0.1248, "step": 10994 }, { "epoch": 8.054945054945055, "grad_norm": 1.0, "learning_rate": 9.147577644745596e-05, "loss": 0.1395, "step": 10995 }, { "epoch": 8.055677655677655, "grad_norm": 1.2421875, "learning_rate": 9.14738540476433e-05, "loss": 0.1169, "step": 10996 }, { "epoch": 8.056410256410256, "grad_norm": 1.484375, "learning_rate": 9.147193145151144e-05, "loss": 0.1475, "step": 10997 }, { "epoch": 8.057142857142857, "grad_norm": 1.4140625, "learning_rate": 9.14700086590696e-05, "loss": 0.1532, "step": 10998 }, { "epoch": 8.057875457875458, "grad_norm": 1.2578125, "learning_rate": 9.146808567032699e-05, "loss": 0.1618, "step": 10999 }, { "epoch": 8.058608058608058, "grad_norm": 1.296875, "learning_rate": 9.146616248529286e-05, "loss": 0.1082, "step": 11000 }, { "epoch": 8.05934065934066, "grad_norm": 0.671875, "learning_rate": 9.146423910397638e-05, "loss": 0.1117, "step": 11001 }, { "epoch": 8.06007326007326, "grad_norm": 1.09375, "learning_rate": 9.146231552638678e-05, "loss": 0.1208, "step": 11002 }, { "epoch": 8.06080586080586, "grad_norm": 1.4375, "learning_rate": 9.14603917525333e-05, "loss": 0.1642, "step": 11003 }, { "epoch": 8.061538461538461, "grad_norm": 2.328125, "learning_rate": 9.145846778242512e-05, "loss": 0.1461, "step": 11004 }, { "epoch": 8.062271062271062, "grad_norm": 1.6328125, "learning_rate": 9.145654361607148e-05, "loss": 0.1588, "step": 11005 }, { "epoch": 8.063003663003663, "grad_norm": 1.296875, "learning_rate": 9.145461925348158e-05, "loss": 0.1349, "step": 11006 }, { "epoch": 8.063736263736264, "grad_norm": 0.90234375, "learning_rate": 9.145269469466467e-05, "loss": 0.1255, "step": 11007 }, { "epoch": 8.064468864468864, "grad_norm": 1.4609375, "learning_rate": 9.145076993962996e-05, "loss": 0.127, "step": 11008 }, { "epoch": 8.065201465201465, "grad_norm": 1.3828125, "learning_rate": 9.144884498838666e-05, "loss": 0.1343, "step": 11009 }, { "epoch": 8.065934065934066, "grad_norm": 1.2109375, "learning_rate": 9.1446919840944e-05, "loss": 0.152, "step": 11010 }, { "epoch": 8.066666666666666, "grad_norm": 1.0078125, "learning_rate": 9.14449944973112e-05, "loss": 0.1501, "step": 11011 }, { "epoch": 8.067399267399267, "grad_norm": 0.96875, "learning_rate": 9.144306895749751e-05, "loss": 0.1245, "step": 11012 }, { "epoch": 8.068131868131868, "grad_norm": 1.46875, "learning_rate": 9.144114322151213e-05, "loss": 0.1295, "step": 11013 }, { "epoch": 8.068864468864469, "grad_norm": 1.1015625, "learning_rate": 9.143921728936428e-05, "loss": 0.1178, "step": 11014 }, { "epoch": 8.06959706959707, "grad_norm": 1.4375, "learning_rate": 9.143729116106318e-05, "loss": 0.1238, "step": 11015 }, { "epoch": 8.07032967032967, "grad_norm": 1.015625, "learning_rate": 9.143536483661813e-05, "loss": 0.1279, "step": 11016 }, { "epoch": 8.07106227106227, "grad_norm": 0.458984375, "learning_rate": 9.143343831603828e-05, "loss": 0.1128, "step": 11017 }, { "epoch": 8.071794871794872, "grad_norm": 0.67578125, "learning_rate": 9.143151159933288e-05, "loss": 0.1421, "step": 11018 }, { "epoch": 8.072527472527472, "grad_norm": 0.75, "learning_rate": 9.142958468651118e-05, "loss": 0.1436, "step": 11019 }, { "epoch": 8.073260073260073, "grad_norm": 0.7109375, "learning_rate": 9.142765757758242e-05, "loss": 0.1216, "step": 11020 }, { "epoch": 8.073992673992674, "grad_norm": 1.609375, "learning_rate": 9.14257302725558e-05, "loss": 0.1625, "step": 11021 }, { "epoch": 8.074725274725274, "grad_norm": 1.0703125, "learning_rate": 9.142380277144059e-05, "loss": 0.1556, "step": 11022 }, { "epoch": 8.075457875457875, "grad_norm": 0.87890625, "learning_rate": 9.1421875074246e-05, "loss": 0.1133, "step": 11023 }, { "epoch": 8.076190476190476, "grad_norm": 1.5, "learning_rate": 9.141994718098124e-05, "loss": 0.1526, "step": 11024 }, { "epoch": 8.076923076923077, "grad_norm": 0.6796875, "learning_rate": 9.141801909165561e-05, "loss": 0.1227, "step": 11025 }, { "epoch": 8.077655677655677, "grad_norm": 1.3046875, "learning_rate": 9.141609080627832e-05, "loss": 0.1575, "step": 11026 }, { "epoch": 8.078388278388278, "grad_norm": 0.74609375, "learning_rate": 9.141416232485861e-05, "loss": 0.1335, "step": 11027 }, { "epoch": 8.079120879120879, "grad_norm": 1.09375, "learning_rate": 9.141223364740572e-05, "loss": 0.1176, "step": 11028 }, { "epoch": 8.07985347985348, "grad_norm": 1.0546875, "learning_rate": 9.141030477392885e-05, "loss": 0.1263, "step": 11029 }, { "epoch": 8.08058608058608, "grad_norm": 1.171875, "learning_rate": 9.140837570443732e-05, "loss": 0.1367, "step": 11030 }, { "epoch": 8.081318681318681, "grad_norm": 1.09375, "learning_rate": 9.140644643894031e-05, "loss": 0.1188, "step": 11031 }, { "epoch": 8.082051282051282, "grad_norm": 1.4375, "learning_rate": 9.140451697744708e-05, "loss": 0.1198, "step": 11032 }, { "epoch": 8.082783882783882, "grad_norm": 0.8515625, "learning_rate": 9.14025873199669e-05, "loss": 0.1552, "step": 11033 }, { "epoch": 8.083516483516483, "grad_norm": 1.078125, "learning_rate": 9.1400657466509e-05, "loss": 0.1123, "step": 11034 }, { "epoch": 8.084249084249084, "grad_norm": 0.8828125, "learning_rate": 9.139872741708262e-05, "loss": 0.1606, "step": 11035 }, { "epoch": 8.084981684981685, "grad_norm": 0.83984375, "learning_rate": 9.1396797171697e-05, "loss": 0.1264, "step": 11036 }, { "epoch": 8.085714285714285, "grad_norm": 0.97265625, "learning_rate": 9.13948667303614e-05, "loss": 0.1188, "step": 11037 }, { "epoch": 8.086446886446886, "grad_norm": 1.4375, "learning_rate": 9.139293609308508e-05, "loss": 0.1555, "step": 11038 }, { "epoch": 8.087179487179487, "grad_norm": 1.1875, "learning_rate": 9.139100525987728e-05, "loss": 0.1197, "step": 11039 }, { "epoch": 8.087912087912088, "grad_norm": 1.1875, "learning_rate": 9.138907423074724e-05, "loss": 0.1305, "step": 11040 }, { "epoch": 8.088644688644688, "grad_norm": 0.9296875, "learning_rate": 9.138714300570424e-05, "loss": 0.1305, "step": 11041 }, { "epoch": 8.089377289377289, "grad_norm": 0.99609375, "learning_rate": 9.138521158475748e-05, "loss": 0.1183, "step": 11042 }, { "epoch": 8.09010989010989, "grad_norm": 1.6171875, "learning_rate": 9.138327996791627e-05, "loss": 0.1489, "step": 11043 }, { "epoch": 8.09084249084249, "grad_norm": 1.28125, "learning_rate": 9.138134815518984e-05, "loss": 0.1638, "step": 11044 }, { "epoch": 8.091575091575091, "grad_norm": 1.5625, "learning_rate": 9.137941614658747e-05, "loss": 0.1637, "step": 11045 }, { "epoch": 8.092307692307692, "grad_norm": 0.80859375, "learning_rate": 9.137748394211838e-05, "loss": 0.1329, "step": 11046 }, { "epoch": 8.093040293040293, "grad_norm": 0.9609375, "learning_rate": 9.137555154179185e-05, "loss": 0.1557, "step": 11047 }, { "epoch": 8.093772893772893, "grad_norm": 1.0078125, "learning_rate": 9.137361894561712e-05, "loss": 0.1485, "step": 11048 }, { "epoch": 8.094505494505494, "grad_norm": 1.6484375, "learning_rate": 9.137168615360349e-05, "loss": 0.1342, "step": 11049 }, { "epoch": 8.095238095238095, "grad_norm": 1.9453125, "learning_rate": 9.136975316576019e-05, "loss": 0.149, "step": 11050 }, { "epoch": 8.095970695970696, "grad_norm": 1.25, "learning_rate": 9.136781998209646e-05, "loss": 0.1253, "step": 11051 }, { "epoch": 8.096703296703296, "grad_norm": 0.9375, "learning_rate": 9.136588660262159e-05, "loss": 0.1411, "step": 11052 }, { "epoch": 8.097435897435897, "grad_norm": 1.1328125, "learning_rate": 9.136395302734488e-05, "loss": 0.1287, "step": 11053 }, { "epoch": 8.098168498168498, "grad_norm": 0.83203125, "learning_rate": 9.136201925627553e-05, "loss": 0.1234, "step": 11054 }, { "epoch": 8.098901098901099, "grad_norm": 0.76953125, "learning_rate": 9.136008528942284e-05, "loss": 0.1199, "step": 11055 }, { "epoch": 8.0996336996337, "grad_norm": 1.0703125, "learning_rate": 9.135815112679606e-05, "loss": 0.1633, "step": 11056 }, { "epoch": 8.1003663003663, "grad_norm": 1.3203125, "learning_rate": 9.135621676840445e-05, "loss": 0.1137, "step": 11057 }, { "epoch": 8.1010989010989, "grad_norm": 1.296875, "learning_rate": 9.135428221425731e-05, "loss": 0.1338, "step": 11058 }, { "epoch": 8.101831501831501, "grad_norm": 0.55859375, "learning_rate": 9.13523474643639e-05, "loss": 0.1174, "step": 11059 }, { "epoch": 8.102564102564102, "grad_norm": 0.91796875, "learning_rate": 9.135041251873349e-05, "loss": 0.1577, "step": 11060 }, { "epoch": 8.103296703296703, "grad_norm": 0.98046875, "learning_rate": 9.134847737737532e-05, "loss": 0.1374, "step": 11061 }, { "epoch": 8.104029304029304, "grad_norm": 0.65625, "learning_rate": 9.13465420402987e-05, "loss": 0.1108, "step": 11062 }, { "epoch": 8.104761904761904, "grad_norm": 0.96484375, "learning_rate": 9.134460650751287e-05, "loss": 0.16, "step": 11063 }, { "epoch": 8.105494505494505, "grad_norm": 1.3984375, "learning_rate": 9.134267077902714e-05, "loss": 0.1847, "step": 11064 }, { "epoch": 8.106227106227106, "grad_norm": 1.1015625, "learning_rate": 9.134073485485075e-05, "loss": 0.144, "step": 11065 }, { "epoch": 8.106959706959707, "grad_norm": 1.1875, "learning_rate": 9.133879873499301e-05, "loss": 0.1486, "step": 11066 }, { "epoch": 8.107692307692307, "grad_norm": 1.1796875, "learning_rate": 9.133686241946316e-05, "loss": 0.1321, "step": 11067 }, { "epoch": 8.108424908424908, "grad_norm": 0.73828125, "learning_rate": 9.133492590827052e-05, "loss": 0.1167, "step": 11068 }, { "epoch": 8.109157509157509, "grad_norm": 0.86328125, "learning_rate": 9.133298920142432e-05, "loss": 0.1152, "step": 11069 }, { "epoch": 8.10989010989011, "grad_norm": 1.203125, "learning_rate": 9.133105229893388e-05, "loss": 0.1429, "step": 11070 }, { "epoch": 8.11062271062271, "grad_norm": 1.1015625, "learning_rate": 9.132911520080847e-05, "loss": 0.1588, "step": 11071 }, { "epoch": 8.111355311355311, "grad_norm": 1.390625, "learning_rate": 9.132717790705735e-05, "loss": 0.1559, "step": 11072 }, { "epoch": 8.112087912087912, "grad_norm": 1.1953125, "learning_rate": 9.132524041768982e-05, "loss": 0.1422, "step": 11073 }, { "epoch": 8.112820512820512, "grad_norm": 1.1015625, "learning_rate": 9.132330273271516e-05, "loss": 0.1484, "step": 11074 }, { "epoch": 8.113553113553113, "grad_norm": 1.2890625, "learning_rate": 9.132136485214268e-05, "loss": 0.1658, "step": 11075 }, { "epoch": 8.114285714285714, "grad_norm": 1.0859375, "learning_rate": 9.131942677598162e-05, "loss": 0.1196, "step": 11076 }, { "epoch": 8.115018315018315, "grad_norm": 1.34375, "learning_rate": 9.131748850424128e-05, "loss": 0.1434, "step": 11077 }, { "epoch": 8.115750915750915, "grad_norm": 1.6484375, "learning_rate": 9.131555003693095e-05, "loss": 0.1539, "step": 11078 }, { "epoch": 8.116483516483516, "grad_norm": 1.34375, "learning_rate": 9.131361137405995e-05, "loss": 0.1389, "step": 11079 }, { "epoch": 8.117216117216117, "grad_norm": 1.0859375, "learning_rate": 9.131167251563752e-05, "loss": 0.1529, "step": 11080 }, { "epoch": 8.117948717948718, "grad_norm": 1.109375, "learning_rate": 9.130973346167296e-05, "loss": 0.1337, "step": 11081 }, { "epoch": 8.118681318681318, "grad_norm": 0.76953125, "learning_rate": 9.130779421217558e-05, "loss": 0.1128, "step": 11082 }, { "epoch": 8.119413919413919, "grad_norm": 1.1328125, "learning_rate": 9.130585476715466e-05, "loss": 0.1419, "step": 11083 }, { "epoch": 8.12014652014652, "grad_norm": 2.375, "learning_rate": 9.13039151266195e-05, "loss": 0.1733, "step": 11084 }, { "epoch": 8.12087912087912, "grad_norm": 1.1875, "learning_rate": 9.130197529057939e-05, "loss": 0.147, "step": 11085 }, { "epoch": 8.121611721611721, "grad_norm": 1.125, "learning_rate": 9.130003525904362e-05, "loss": 0.1329, "step": 11086 }, { "epoch": 8.122344322344322, "grad_norm": 1.8671875, "learning_rate": 9.129809503202148e-05, "loss": 0.1426, "step": 11087 }, { "epoch": 8.123076923076923, "grad_norm": 1.71875, "learning_rate": 9.129615460952226e-05, "loss": 0.1334, "step": 11088 }, { "epoch": 8.123809523809523, "grad_norm": 1.640625, "learning_rate": 9.12942139915553e-05, "loss": 0.1453, "step": 11089 }, { "epoch": 8.124542124542124, "grad_norm": 1.703125, "learning_rate": 9.129227317812986e-05, "loss": 0.1167, "step": 11090 }, { "epoch": 8.125274725274725, "grad_norm": 1.25, "learning_rate": 9.129033216925524e-05, "loss": 0.1032, "step": 11091 }, { "epoch": 8.126007326007326, "grad_norm": 1.1953125, "learning_rate": 9.128839096494074e-05, "loss": 0.1217, "step": 11092 }, { "epoch": 8.126739926739926, "grad_norm": 1.8203125, "learning_rate": 9.128644956519569e-05, "loss": 0.1253, "step": 11093 }, { "epoch": 8.127472527472527, "grad_norm": 2.65625, "learning_rate": 9.128450797002936e-05, "loss": 0.1434, "step": 11094 }, { "epoch": 8.128205128205128, "grad_norm": 1.84375, "learning_rate": 9.128256617945107e-05, "loss": 0.1393, "step": 11095 }, { "epoch": 8.128937728937728, "grad_norm": 2.84375, "learning_rate": 9.128062419347011e-05, "loss": 0.1768, "step": 11096 }, { "epoch": 8.12967032967033, "grad_norm": 2.3125, "learning_rate": 9.12786820120958e-05, "loss": 0.1343, "step": 11097 }, { "epoch": 8.13040293040293, "grad_norm": 2.140625, "learning_rate": 9.127673963533743e-05, "loss": 0.1679, "step": 11098 }, { "epoch": 8.13113553113553, "grad_norm": 0.68359375, "learning_rate": 9.127479706320432e-05, "loss": 0.137, "step": 11099 }, { "epoch": 8.131868131868131, "grad_norm": 0.9453125, "learning_rate": 9.127285429570577e-05, "loss": 0.1133, "step": 11100 }, { "epoch": 8.132600732600732, "grad_norm": 2.796875, "learning_rate": 9.127091133285109e-05, "loss": 0.2309, "step": 11101 }, { "epoch": 8.133333333333333, "grad_norm": 1.96875, "learning_rate": 9.126896817464958e-05, "loss": 0.144, "step": 11102 }, { "epoch": 8.134065934065934, "grad_norm": 1.7578125, "learning_rate": 9.126702482111059e-05, "loss": 0.1579, "step": 11103 }, { "epoch": 8.134798534798534, "grad_norm": 1.796875, "learning_rate": 9.126508127224336e-05, "loss": 0.1473, "step": 11104 }, { "epoch": 8.135531135531135, "grad_norm": 0.96484375, "learning_rate": 9.126313752805726e-05, "loss": 0.1565, "step": 11105 }, { "epoch": 8.136263736263736, "grad_norm": 0.9765625, "learning_rate": 9.12611935885616e-05, "loss": 0.1426, "step": 11106 }, { "epoch": 8.136996336996337, "grad_norm": 2.078125, "learning_rate": 9.125924945376566e-05, "loss": 0.1376, "step": 11107 }, { "epoch": 8.137728937728937, "grad_norm": 0.890625, "learning_rate": 9.125730512367879e-05, "loss": 0.0972, "step": 11108 }, { "epoch": 8.138461538461538, "grad_norm": 0.828125, "learning_rate": 9.125536059831028e-05, "loss": 0.1517, "step": 11109 }, { "epoch": 8.139194139194139, "grad_norm": 1.640625, "learning_rate": 9.125341587766946e-05, "loss": 0.1214, "step": 11110 }, { "epoch": 8.13992673992674, "grad_norm": 1.3984375, "learning_rate": 9.125147096176566e-05, "loss": 0.1586, "step": 11111 }, { "epoch": 8.14065934065934, "grad_norm": 0.796875, "learning_rate": 9.124952585060818e-05, "loss": 0.124, "step": 11112 }, { "epoch": 8.141391941391941, "grad_norm": 0.69921875, "learning_rate": 9.124758054420635e-05, "loss": 0.1178, "step": 11113 }, { "epoch": 8.142124542124542, "grad_norm": 0.90625, "learning_rate": 9.124563504256946e-05, "loss": 0.1428, "step": 11114 }, { "epoch": 8.142857142857142, "grad_norm": 2.234375, "learning_rate": 9.124368934570687e-05, "loss": 0.1628, "step": 11115 }, { "epoch": 8.143589743589743, "grad_norm": 1.421875, "learning_rate": 9.124174345362789e-05, "loss": 0.1298, "step": 11116 }, { "epoch": 8.144322344322344, "grad_norm": 1.2109375, "learning_rate": 9.123979736634184e-05, "loss": 0.1461, "step": 11117 }, { "epoch": 8.145054945054945, "grad_norm": 1.8671875, "learning_rate": 9.123785108385806e-05, "loss": 0.1339, "step": 11118 }, { "epoch": 8.145787545787545, "grad_norm": 0.9140625, "learning_rate": 9.123590460618585e-05, "loss": 0.1336, "step": 11119 }, { "epoch": 8.146520146520146, "grad_norm": 1.2109375, "learning_rate": 9.123395793333454e-05, "loss": 0.1721, "step": 11120 }, { "epoch": 8.147252747252747, "grad_norm": 1.171875, "learning_rate": 9.12320110653135e-05, "loss": 0.1461, "step": 11121 }, { "epoch": 8.147985347985347, "grad_norm": 0.60546875, "learning_rate": 9.123006400213198e-05, "loss": 0.1215, "step": 11122 }, { "epoch": 8.148717948717948, "grad_norm": 0.78515625, "learning_rate": 9.122811674379939e-05, "loss": 0.1256, "step": 11123 }, { "epoch": 8.149450549450549, "grad_norm": 0.75, "learning_rate": 9.1226169290325e-05, "loss": 0.1433, "step": 11124 }, { "epoch": 8.15018315018315, "grad_norm": 0.8671875, "learning_rate": 9.122422164171816e-05, "loss": 0.1285, "step": 11125 }, { "epoch": 8.15091575091575, "grad_norm": 0.99609375, "learning_rate": 9.122227379798822e-05, "loss": 0.1322, "step": 11126 }, { "epoch": 8.151648351648351, "grad_norm": 0.80859375, "learning_rate": 9.122032575914448e-05, "loss": 0.1444, "step": 11127 }, { "epoch": 8.152380952380952, "grad_norm": 1.0625, "learning_rate": 9.121837752519632e-05, "loss": 0.1155, "step": 11128 }, { "epoch": 8.153113553113553, "grad_norm": 0.9296875, "learning_rate": 9.121642909615301e-05, "loss": 0.1303, "step": 11129 }, { "epoch": 8.153846153846153, "grad_norm": 0.75390625, "learning_rate": 9.121448047202395e-05, "loss": 0.1252, "step": 11130 }, { "epoch": 8.154578754578754, "grad_norm": 0.90234375, "learning_rate": 9.121253165281843e-05, "loss": 0.1152, "step": 11131 }, { "epoch": 8.155311355311355, "grad_norm": 0.96875, "learning_rate": 9.121058263854582e-05, "loss": 0.1509, "step": 11132 }, { "epoch": 8.156043956043955, "grad_norm": 1.078125, "learning_rate": 9.120863342921544e-05, "loss": 0.1622, "step": 11133 }, { "epoch": 8.156776556776556, "grad_norm": 0.859375, "learning_rate": 9.120668402483664e-05, "loss": 0.1635, "step": 11134 }, { "epoch": 8.157509157509157, "grad_norm": 1.3359375, "learning_rate": 9.120473442541874e-05, "loss": 0.1377, "step": 11135 }, { "epoch": 8.158241758241758, "grad_norm": 0.90625, "learning_rate": 9.12027846309711e-05, "loss": 0.1508, "step": 11136 }, { "epoch": 8.158974358974358, "grad_norm": 0.8359375, "learning_rate": 9.120083464150307e-05, "loss": 0.1176, "step": 11137 }, { "epoch": 8.15970695970696, "grad_norm": 1.2890625, "learning_rate": 9.119888445702395e-05, "loss": 0.1234, "step": 11138 }, { "epoch": 8.16043956043956, "grad_norm": 1.0703125, "learning_rate": 9.119693407754315e-05, "loss": 0.1421, "step": 11139 }, { "epoch": 8.16117216117216, "grad_norm": 1.03125, "learning_rate": 9.119498350306995e-05, "loss": 0.1349, "step": 11140 }, { "epoch": 8.161904761904761, "grad_norm": 1.0859375, "learning_rate": 9.119303273361373e-05, "loss": 0.1624, "step": 11141 }, { "epoch": 8.162637362637362, "grad_norm": 0.98828125, "learning_rate": 9.119108176918385e-05, "loss": 0.1594, "step": 11142 }, { "epoch": 8.163369963369963, "grad_norm": 1.2734375, "learning_rate": 9.118913060978964e-05, "loss": 0.2066, "step": 11143 }, { "epoch": 8.164102564102564, "grad_norm": 1.2109375, "learning_rate": 9.118717925544041e-05, "loss": 0.1663, "step": 11144 }, { "epoch": 8.164835164835164, "grad_norm": 0.99609375, "learning_rate": 9.118522770614558e-05, "loss": 0.1151, "step": 11145 }, { "epoch": 8.165567765567765, "grad_norm": 1.1328125, "learning_rate": 9.118327596191446e-05, "loss": 0.1531, "step": 11146 }, { "epoch": 8.166300366300366, "grad_norm": 0.69921875, "learning_rate": 9.118132402275641e-05, "loss": 0.1283, "step": 11147 }, { "epoch": 8.167032967032966, "grad_norm": 1.34375, "learning_rate": 9.117937188868078e-05, "loss": 0.1738, "step": 11148 }, { "epoch": 8.167765567765567, "grad_norm": 0.98828125, "learning_rate": 9.117741955969693e-05, "loss": 0.1239, "step": 11149 }, { "epoch": 8.168498168498168, "grad_norm": 1.3203125, "learning_rate": 9.11754670358142e-05, "loss": 0.129, "step": 11150 }, { "epoch": 8.169230769230769, "grad_norm": 0.70703125, "learning_rate": 9.117351431704197e-05, "loss": 0.1307, "step": 11151 }, { "epoch": 8.16996336996337, "grad_norm": 0.9296875, "learning_rate": 9.117156140338957e-05, "loss": 0.1521, "step": 11152 }, { "epoch": 8.17069597069597, "grad_norm": 1.0390625, "learning_rate": 9.116960829486638e-05, "loss": 0.1457, "step": 11153 }, { "epoch": 8.17142857142857, "grad_norm": 0.78515625, "learning_rate": 9.116765499148172e-05, "loss": 0.1375, "step": 11154 }, { "epoch": 8.172161172161172, "grad_norm": 0.78125, "learning_rate": 9.1165701493245e-05, "loss": 0.1346, "step": 11155 }, { "epoch": 8.172893772893772, "grad_norm": 1.09375, "learning_rate": 9.116374780016553e-05, "loss": 0.1697, "step": 11156 }, { "epoch": 8.173626373626373, "grad_norm": 1.0859375, "learning_rate": 9.116179391225273e-05, "loss": 0.15, "step": 11157 }, { "epoch": 8.174358974358974, "grad_norm": 1.4296875, "learning_rate": 9.115983982951591e-05, "loss": 0.1488, "step": 11158 }, { "epoch": 8.175091575091574, "grad_norm": 1.1796875, "learning_rate": 9.115788555196442e-05, "loss": 0.1845, "step": 11159 }, { "epoch": 8.175824175824175, "grad_norm": 0.8125, "learning_rate": 9.115593107960769e-05, "loss": 0.1268, "step": 11160 }, { "epoch": 8.176556776556776, "grad_norm": 1.078125, "learning_rate": 9.115397641245504e-05, "loss": 0.1279, "step": 11161 }, { "epoch": 8.177289377289377, "grad_norm": 1.3984375, "learning_rate": 9.115202155051583e-05, "loss": 0.1455, "step": 11162 }, { "epoch": 8.178021978021977, "grad_norm": 1.078125, "learning_rate": 9.115006649379946e-05, "loss": 0.1288, "step": 11163 }, { "epoch": 8.178754578754578, "grad_norm": 0.91796875, "learning_rate": 9.114811124231526e-05, "loss": 0.1483, "step": 11164 }, { "epoch": 8.179487179487179, "grad_norm": 1.2890625, "learning_rate": 9.114615579607263e-05, "loss": 0.15, "step": 11165 }, { "epoch": 8.18021978021978, "grad_norm": 1.1015625, "learning_rate": 9.114420015508091e-05, "loss": 0.1401, "step": 11166 }, { "epoch": 8.18095238095238, "grad_norm": 0.92578125, "learning_rate": 9.114224431934949e-05, "loss": 0.1391, "step": 11167 }, { "epoch": 8.181684981684981, "grad_norm": 1.03125, "learning_rate": 9.114028828888773e-05, "loss": 0.1225, "step": 11168 }, { "epoch": 8.182417582417582, "grad_norm": 1.15625, "learning_rate": 9.113833206370502e-05, "loss": 0.1675, "step": 11169 }, { "epoch": 8.183150183150182, "grad_norm": 1.09375, "learning_rate": 9.11363756438107e-05, "loss": 0.1369, "step": 11170 }, { "epoch": 8.183882783882783, "grad_norm": 0.8203125, "learning_rate": 9.113441902921418e-05, "loss": 0.1301, "step": 11171 }, { "epoch": 8.184615384615384, "grad_norm": 0.7265625, "learning_rate": 9.113246221992482e-05, "loss": 0.1228, "step": 11172 }, { "epoch": 8.185347985347985, "grad_norm": 1.515625, "learning_rate": 9.113050521595199e-05, "loss": 0.1487, "step": 11173 }, { "epoch": 8.186080586080585, "grad_norm": 0.91015625, "learning_rate": 9.112854801730508e-05, "loss": 0.1115, "step": 11174 }, { "epoch": 8.186813186813186, "grad_norm": 1.4921875, "learning_rate": 9.112659062399345e-05, "loss": 0.176, "step": 11175 }, { "epoch": 8.187545787545787, "grad_norm": 1.09375, "learning_rate": 9.11246330360265e-05, "loss": 0.1377, "step": 11176 }, { "epoch": 8.188278388278388, "grad_norm": 1.609375, "learning_rate": 9.112267525341357e-05, "loss": 0.1673, "step": 11177 }, { "epoch": 8.189010989010988, "grad_norm": 0.81640625, "learning_rate": 9.112071727616408e-05, "loss": 0.1349, "step": 11178 }, { "epoch": 8.189743589743589, "grad_norm": 1.3359375, "learning_rate": 9.11187591042874e-05, "loss": 0.1269, "step": 11179 }, { "epoch": 8.19047619047619, "grad_norm": 0.99609375, "learning_rate": 9.111680073779291e-05, "loss": 0.14, "step": 11180 }, { "epoch": 8.19120879120879, "grad_norm": 0.73828125, "learning_rate": 9.111484217669001e-05, "loss": 0.113, "step": 11181 }, { "epoch": 8.191941391941391, "grad_norm": 1.0234375, "learning_rate": 9.111288342098805e-05, "loss": 0.1585, "step": 11182 }, { "epoch": 8.192673992673992, "grad_norm": 1.015625, "learning_rate": 9.111092447069645e-05, "loss": 0.137, "step": 11183 }, { "epoch": 8.193406593406593, "grad_norm": 0.890625, "learning_rate": 9.110896532582456e-05, "loss": 0.1562, "step": 11184 }, { "epoch": 8.194139194139193, "grad_norm": 1.171875, "learning_rate": 9.11070059863818e-05, "loss": 0.1483, "step": 11185 }, { "epoch": 8.194871794871794, "grad_norm": 1.0625, "learning_rate": 9.110504645237752e-05, "loss": 0.1473, "step": 11186 }, { "epoch": 8.195604395604395, "grad_norm": 1.078125, "learning_rate": 9.110308672382116e-05, "loss": 0.1502, "step": 11187 }, { "epoch": 8.196336996336996, "grad_norm": 1.0625, "learning_rate": 9.110112680072207e-05, "loss": 0.1849, "step": 11188 }, { "epoch": 8.197069597069596, "grad_norm": 1.2109375, "learning_rate": 9.109916668308966e-05, "loss": 0.1131, "step": 11189 }, { "epoch": 8.197802197802197, "grad_norm": 1.1328125, "learning_rate": 9.109720637093331e-05, "loss": 0.1718, "step": 11190 }, { "epoch": 8.198534798534798, "grad_norm": 1.46875, "learning_rate": 9.109524586426243e-05, "loss": 0.1611, "step": 11191 }, { "epoch": 8.199267399267399, "grad_norm": 0.7734375, "learning_rate": 9.109328516308637e-05, "loss": 0.1474, "step": 11192 }, { "epoch": 8.2, "grad_norm": 1.1875, "learning_rate": 9.109132426741458e-05, "loss": 0.1155, "step": 11193 }, { "epoch": 8.2007326007326, "grad_norm": 0.91796875, "learning_rate": 9.108936317725642e-05, "loss": 0.1287, "step": 11194 }, { "epoch": 8.2014652014652, "grad_norm": 1.6640625, "learning_rate": 9.10874018926213e-05, "loss": 0.1523, "step": 11195 }, { "epoch": 8.202197802197801, "grad_norm": 1.484375, "learning_rate": 9.10854404135186e-05, "loss": 0.1718, "step": 11196 }, { "epoch": 8.202930402930402, "grad_norm": 0.8828125, "learning_rate": 9.108347873995776e-05, "loss": 0.1455, "step": 11197 }, { "epoch": 8.203663003663003, "grad_norm": 0.7578125, "learning_rate": 9.108151687194813e-05, "loss": 0.1047, "step": 11198 }, { "epoch": 8.204395604395604, "grad_norm": 1.0546875, "learning_rate": 9.107955480949914e-05, "loss": 0.1352, "step": 11199 }, { "epoch": 8.205128205128204, "grad_norm": 0.96875, "learning_rate": 9.107759255262017e-05, "loss": 0.179, "step": 11200 }, { "epoch": 8.205860805860805, "grad_norm": 1.0546875, "learning_rate": 9.107563010132065e-05, "loss": 0.1578, "step": 11201 }, { "epoch": 8.206593406593406, "grad_norm": 0.875, "learning_rate": 9.107366745560996e-05, "loss": 0.1319, "step": 11202 }, { "epoch": 8.207326007326007, "grad_norm": 1.28125, "learning_rate": 9.10717046154975e-05, "loss": 0.1498, "step": 11203 }, { "epoch": 8.208058608058607, "grad_norm": 0.65234375, "learning_rate": 9.106974158099269e-05, "loss": 0.1226, "step": 11204 }, { "epoch": 8.208791208791208, "grad_norm": 1.2421875, "learning_rate": 9.106777835210492e-05, "loss": 0.1611, "step": 11205 }, { "epoch": 8.209523809523809, "grad_norm": 1.078125, "learning_rate": 9.106581492884363e-05, "loss": 0.1467, "step": 11206 }, { "epoch": 8.21025641025641, "grad_norm": 1.296875, "learning_rate": 9.106385131121818e-05, "loss": 0.1886, "step": 11207 }, { "epoch": 8.21098901098901, "grad_norm": 0.85546875, "learning_rate": 9.106188749923801e-05, "loss": 0.1143, "step": 11208 }, { "epoch": 8.211721611721611, "grad_norm": 0.99609375, "learning_rate": 9.105992349291254e-05, "loss": 0.1391, "step": 11209 }, { "epoch": 8.212454212454212, "grad_norm": 1.078125, "learning_rate": 9.105795929225115e-05, "loss": 0.1243, "step": 11210 }, { "epoch": 8.213186813186812, "grad_norm": 1.0, "learning_rate": 9.105599489726326e-05, "loss": 0.1202, "step": 11211 }, { "epoch": 8.213919413919413, "grad_norm": 0.83203125, "learning_rate": 9.105403030795828e-05, "loss": 0.1404, "step": 11212 }, { "epoch": 8.214652014652014, "grad_norm": 1.3828125, "learning_rate": 9.105206552434561e-05, "loss": 0.1498, "step": 11213 }, { "epoch": 8.215384615384615, "grad_norm": 0.76953125, "learning_rate": 9.105010054643472e-05, "loss": 0.1336, "step": 11214 }, { "epoch": 8.216117216117215, "grad_norm": 1.390625, "learning_rate": 9.104813537423496e-05, "loss": 0.1417, "step": 11215 }, { "epoch": 8.216849816849816, "grad_norm": 1.6953125, "learning_rate": 9.10461700077558e-05, "loss": 0.1532, "step": 11216 }, { "epoch": 8.217582417582417, "grad_norm": 1.390625, "learning_rate": 9.10442044470066e-05, "loss": 0.1483, "step": 11217 }, { "epoch": 8.218315018315018, "grad_norm": 1.15625, "learning_rate": 9.104223869199682e-05, "loss": 0.1332, "step": 11218 }, { "epoch": 8.219047619047618, "grad_norm": 1.4765625, "learning_rate": 9.104027274273585e-05, "loss": 0.1632, "step": 11219 }, { "epoch": 8.219780219780219, "grad_norm": 0.953125, "learning_rate": 9.103830659923315e-05, "loss": 0.1398, "step": 11220 }, { "epoch": 8.22051282051282, "grad_norm": 1.15625, "learning_rate": 9.10363402614981e-05, "loss": 0.1544, "step": 11221 }, { "epoch": 8.22124542124542, "grad_norm": 1.703125, "learning_rate": 9.103437372954015e-05, "loss": 0.1411, "step": 11222 }, { "epoch": 8.221978021978021, "grad_norm": 1.6953125, "learning_rate": 9.103240700336869e-05, "loss": 0.1349, "step": 11223 }, { "epoch": 8.222710622710622, "grad_norm": 1.5234375, "learning_rate": 9.103044008299317e-05, "loss": 0.1377, "step": 11224 }, { "epoch": 8.223443223443223, "grad_norm": 1.453125, "learning_rate": 9.102847296842301e-05, "loss": 0.1554, "step": 11225 }, { "epoch": 8.224175824175823, "grad_norm": 1.796875, "learning_rate": 9.102650565966763e-05, "loss": 0.1245, "step": 11226 }, { "epoch": 8.224908424908424, "grad_norm": 1.296875, "learning_rate": 9.102453815673646e-05, "loss": 0.1274, "step": 11227 }, { "epoch": 8.225641025641025, "grad_norm": 1.0625, "learning_rate": 9.102257045963891e-05, "loss": 0.1633, "step": 11228 }, { "epoch": 8.226373626373626, "grad_norm": 1.109375, "learning_rate": 9.102060256838444e-05, "loss": 0.1503, "step": 11229 }, { "epoch": 8.227106227106226, "grad_norm": 1.71875, "learning_rate": 9.101863448298244e-05, "loss": 0.168, "step": 11230 }, { "epoch": 8.227838827838827, "grad_norm": 1.53125, "learning_rate": 9.101666620344238e-05, "loss": 0.1109, "step": 11231 }, { "epoch": 8.228571428571428, "grad_norm": 1.59375, "learning_rate": 9.101469772977366e-05, "loss": 0.1452, "step": 11232 }, { "epoch": 8.229304029304028, "grad_norm": 1.6953125, "learning_rate": 9.101272906198575e-05, "loss": 0.1469, "step": 11233 }, { "epoch": 8.23003663003663, "grad_norm": 1.359375, "learning_rate": 9.101076020008802e-05, "loss": 0.1558, "step": 11234 }, { "epoch": 8.23076923076923, "grad_norm": 1.2109375, "learning_rate": 9.100879114408994e-05, "loss": 0.1346, "step": 11235 }, { "epoch": 8.23150183150183, "grad_norm": 1.3515625, "learning_rate": 9.100682189400097e-05, "loss": 0.1314, "step": 11236 }, { "epoch": 8.232234432234431, "grad_norm": 1.2578125, "learning_rate": 9.100485244983052e-05, "loss": 0.1378, "step": 11237 }, { "epoch": 8.232967032967032, "grad_norm": 1.8203125, "learning_rate": 9.1002882811588e-05, "loss": 0.1503, "step": 11238 }, { "epoch": 8.233699633699633, "grad_norm": 2.078125, "learning_rate": 9.100091297928288e-05, "loss": 0.1348, "step": 11239 }, { "epoch": 8.234432234432234, "grad_norm": 1.828125, "learning_rate": 9.099894295292459e-05, "loss": 0.1323, "step": 11240 }, { "epoch": 8.235164835164834, "grad_norm": 1.578125, "learning_rate": 9.099697273252258e-05, "loss": 0.1348, "step": 11241 }, { "epoch": 8.235897435897435, "grad_norm": 1.5234375, "learning_rate": 9.099500231808627e-05, "loss": 0.1927, "step": 11242 }, { "epoch": 8.236630036630036, "grad_norm": 1.5625, "learning_rate": 9.099303170962511e-05, "loss": 0.1105, "step": 11243 }, { "epoch": 8.237362637362637, "grad_norm": 1.578125, "learning_rate": 9.099106090714856e-05, "loss": 0.1449, "step": 11244 }, { "epoch": 8.238095238095237, "grad_norm": 1.1328125, "learning_rate": 9.098908991066601e-05, "loss": 0.1578, "step": 11245 }, { "epoch": 8.238827838827838, "grad_norm": 1.3125, "learning_rate": 9.098711872018696e-05, "loss": 0.1466, "step": 11246 }, { "epoch": 8.239560439560439, "grad_norm": 1.0078125, "learning_rate": 9.098514733572084e-05, "loss": 0.1227, "step": 11247 }, { "epoch": 8.24029304029304, "grad_norm": 0.86328125, "learning_rate": 9.098317575727708e-05, "loss": 0.1456, "step": 11248 }, { "epoch": 8.24102564102564, "grad_norm": 1.09375, "learning_rate": 9.098120398486512e-05, "loss": 0.1494, "step": 11249 }, { "epoch": 8.241758241758241, "grad_norm": 1.1171875, "learning_rate": 9.097923201849445e-05, "loss": 0.1537, "step": 11250 }, { "epoch": 8.242490842490842, "grad_norm": 0.921875, "learning_rate": 9.097725985817447e-05, "loss": 0.1265, "step": 11251 }, { "epoch": 8.243223443223442, "grad_norm": 0.86328125, "learning_rate": 9.097528750391467e-05, "loss": 0.1221, "step": 11252 }, { "epoch": 8.243956043956043, "grad_norm": 0.890625, "learning_rate": 9.097331495572446e-05, "loss": 0.1383, "step": 11253 }, { "epoch": 8.244688644688644, "grad_norm": 1.25, "learning_rate": 9.09713422136133e-05, "loss": 0.1482, "step": 11254 }, { "epoch": 8.245421245421245, "grad_norm": 1.046875, "learning_rate": 9.09693692775907e-05, "loss": 0.1472, "step": 11255 }, { "epoch": 8.246153846153845, "grad_norm": 0.8515625, "learning_rate": 9.096739614766603e-05, "loss": 0.108, "step": 11256 }, { "epoch": 8.246886446886446, "grad_norm": 0.515625, "learning_rate": 9.09654228238488e-05, "loss": 0.1176, "step": 11257 }, { "epoch": 8.247619047619047, "grad_norm": 0.76953125, "learning_rate": 9.096344930614843e-05, "loss": 0.1474, "step": 11258 }, { "epoch": 8.248351648351647, "grad_norm": 0.97265625, "learning_rate": 9.09614755945744e-05, "loss": 0.1223, "step": 11259 }, { "epoch": 8.249084249084248, "grad_norm": 0.671875, "learning_rate": 9.095950168913616e-05, "loss": 0.1101, "step": 11260 }, { "epoch": 8.249816849816849, "grad_norm": 0.81640625, "learning_rate": 9.095752758984315e-05, "loss": 0.1408, "step": 11261 }, { "epoch": 8.25054945054945, "grad_norm": 0.63671875, "learning_rate": 9.095555329670485e-05, "loss": 0.1247, "step": 11262 }, { "epoch": 8.25128205128205, "grad_norm": 1.0859375, "learning_rate": 9.095357880973073e-05, "loss": 0.138, "step": 11263 }, { "epoch": 8.252014652014651, "grad_norm": 0.7265625, "learning_rate": 9.095160412893021e-05, "loss": 0.1239, "step": 11264 }, { "epoch": 8.252747252747252, "grad_norm": 0.66796875, "learning_rate": 9.094962925431281e-05, "loss": 0.1407, "step": 11265 }, { "epoch": 8.253479853479853, "grad_norm": 0.8046875, "learning_rate": 9.094765418588792e-05, "loss": 0.1009, "step": 11266 }, { "epoch": 8.254212454212453, "grad_norm": 1.0859375, "learning_rate": 9.094567892366505e-05, "loss": 0.1501, "step": 11267 }, { "epoch": 8.254945054945054, "grad_norm": 0.6328125, "learning_rate": 9.094370346765366e-05, "loss": 0.128, "step": 11268 }, { "epoch": 8.255677655677655, "grad_norm": 0.96484375, "learning_rate": 9.094172781786322e-05, "loss": 0.1324, "step": 11269 }, { "epoch": 8.256410256410255, "grad_norm": 0.78125, "learning_rate": 9.093975197430318e-05, "loss": 0.1022, "step": 11270 }, { "epoch": 8.257142857142856, "grad_norm": 0.69921875, "learning_rate": 9.0937775936983e-05, "loss": 0.1475, "step": 11271 }, { "epoch": 8.257875457875459, "grad_norm": 0.85546875, "learning_rate": 9.093579970591218e-05, "loss": 0.1222, "step": 11272 }, { "epoch": 8.25860805860806, "grad_norm": 1.0234375, "learning_rate": 9.093382328110014e-05, "loss": 0.1577, "step": 11273 }, { "epoch": 8.25934065934066, "grad_norm": 0.90234375, "learning_rate": 9.09318466625564e-05, "loss": 0.1043, "step": 11274 }, { "epoch": 8.260073260073261, "grad_norm": 1.265625, "learning_rate": 9.092986985029041e-05, "loss": 0.1461, "step": 11275 }, { "epoch": 8.260805860805862, "grad_norm": 0.9609375, "learning_rate": 9.092789284431165e-05, "loss": 0.1379, "step": 11276 }, { "epoch": 8.261538461538462, "grad_norm": 1.015625, "learning_rate": 9.092591564462956e-05, "loss": 0.142, "step": 11277 }, { "epoch": 8.262271062271063, "grad_norm": 1.125, "learning_rate": 9.092393825125365e-05, "loss": 0.1433, "step": 11278 }, { "epoch": 8.263003663003664, "grad_norm": 0.84375, "learning_rate": 9.09219606641934e-05, "loss": 0.1177, "step": 11279 }, { "epoch": 8.263736263736265, "grad_norm": 0.75, "learning_rate": 9.091998288345824e-05, "loss": 0.1281, "step": 11280 }, { "epoch": 8.264468864468865, "grad_norm": 0.75390625, "learning_rate": 9.091800490905767e-05, "loss": 0.1471, "step": 11281 }, { "epoch": 8.265201465201466, "grad_norm": 0.94921875, "learning_rate": 9.091602674100119e-05, "loss": 0.1077, "step": 11282 }, { "epoch": 8.265934065934067, "grad_norm": 1.078125, "learning_rate": 9.091404837929824e-05, "loss": 0.1389, "step": 11283 }, { "epoch": 8.266666666666667, "grad_norm": 1.25, "learning_rate": 9.091206982395833e-05, "loss": 0.1326, "step": 11284 }, { "epoch": 8.267399267399268, "grad_norm": 0.91015625, "learning_rate": 9.091009107499093e-05, "loss": 0.1215, "step": 11285 }, { "epoch": 8.268131868131869, "grad_norm": 1.390625, "learning_rate": 9.090811213240551e-05, "loss": 0.1415, "step": 11286 }, { "epoch": 8.26886446886447, "grad_norm": 0.73046875, "learning_rate": 9.090613299621156e-05, "loss": 0.1137, "step": 11287 }, { "epoch": 8.26959706959707, "grad_norm": 0.8515625, "learning_rate": 9.090415366641856e-05, "loss": 0.1467, "step": 11288 }, { "epoch": 8.270329670329671, "grad_norm": 0.68359375, "learning_rate": 9.0902174143036e-05, "loss": 0.1205, "step": 11289 }, { "epoch": 8.271062271062272, "grad_norm": 1.015625, "learning_rate": 9.090019442607336e-05, "loss": 0.1674, "step": 11290 }, { "epoch": 8.271794871794873, "grad_norm": 0.765625, "learning_rate": 9.089821451554012e-05, "loss": 0.1156, "step": 11291 }, { "epoch": 8.272527472527473, "grad_norm": 1.03125, "learning_rate": 9.089623441144579e-05, "loss": 0.1469, "step": 11292 }, { "epoch": 8.273260073260074, "grad_norm": 0.90234375, "learning_rate": 9.089425411379983e-05, "loss": 0.134, "step": 11293 }, { "epoch": 8.273992673992675, "grad_norm": 1.203125, "learning_rate": 9.089227362261173e-05, "loss": 0.157, "step": 11294 }, { "epoch": 8.274725274725276, "grad_norm": 0.8125, "learning_rate": 9.089029293789099e-05, "loss": 0.1509, "step": 11295 }, { "epoch": 8.275457875457876, "grad_norm": 0.8671875, "learning_rate": 9.088831205964711e-05, "loss": 0.1248, "step": 11296 }, { "epoch": 8.276190476190477, "grad_norm": 1.2109375, "learning_rate": 9.088633098788956e-05, "loss": 0.1615, "step": 11297 }, { "epoch": 8.276923076923078, "grad_norm": 0.98828125, "learning_rate": 9.088434972262784e-05, "loss": 0.1513, "step": 11298 }, { "epoch": 8.277655677655678, "grad_norm": 0.87109375, "learning_rate": 9.088236826387143e-05, "loss": 0.1106, "step": 11299 }, { "epoch": 8.27838827838828, "grad_norm": 1.25, "learning_rate": 9.088038661162985e-05, "loss": 0.1189, "step": 11300 }, { "epoch": 8.27912087912088, "grad_norm": 1.046875, "learning_rate": 9.087840476591259e-05, "loss": 0.1489, "step": 11301 }, { "epoch": 8.27985347985348, "grad_norm": 1.171875, "learning_rate": 9.087642272672913e-05, "loss": 0.1116, "step": 11302 }, { "epoch": 8.280586080586081, "grad_norm": 0.96875, "learning_rate": 9.087444049408897e-05, "loss": 0.1304, "step": 11303 }, { "epoch": 8.281318681318682, "grad_norm": 1.0546875, "learning_rate": 9.087245806800161e-05, "loss": 0.1113, "step": 11304 }, { "epoch": 8.282051282051283, "grad_norm": 1.046875, "learning_rate": 9.087047544847657e-05, "loss": 0.1617, "step": 11305 }, { "epoch": 8.282783882783884, "grad_norm": 1.1171875, "learning_rate": 9.08684926355233e-05, "loss": 0.1542, "step": 11306 }, { "epoch": 8.283516483516484, "grad_norm": 0.8828125, "learning_rate": 9.086650962915133e-05, "loss": 0.1055, "step": 11307 }, { "epoch": 8.284249084249085, "grad_norm": 1.1640625, "learning_rate": 9.086452642937018e-05, "loss": 0.1496, "step": 11308 }, { "epoch": 8.284981684981686, "grad_norm": 1.4609375, "learning_rate": 9.086254303618932e-05, "loss": 0.159, "step": 11309 }, { "epoch": 8.285714285714286, "grad_norm": 1.1796875, "learning_rate": 9.08605594496183e-05, "loss": 0.1532, "step": 11310 }, { "epoch": 8.286446886446887, "grad_norm": 1.359375, "learning_rate": 9.085857566966656e-05, "loss": 0.1531, "step": 11311 }, { "epoch": 8.287179487179488, "grad_norm": 1.203125, "learning_rate": 9.085659169634363e-05, "loss": 0.1293, "step": 11312 }, { "epoch": 8.287912087912089, "grad_norm": 0.67578125, "learning_rate": 9.085460752965902e-05, "loss": 0.1184, "step": 11313 }, { "epoch": 8.28864468864469, "grad_norm": 0.765625, "learning_rate": 9.085262316962226e-05, "loss": 0.1433, "step": 11314 }, { "epoch": 8.28937728937729, "grad_norm": 0.890625, "learning_rate": 9.085063861624283e-05, "loss": 0.1237, "step": 11315 }, { "epoch": 8.29010989010989, "grad_norm": 1.4765625, "learning_rate": 9.084865386953022e-05, "loss": 0.1513, "step": 11316 }, { "epoch": 8.290842490842492, "grad_norm": 1.7421875, "learning_rate": 9.084666892949398e-05, "loss": 0.1334, "step": 11317 }, { "epoch": 8.291575091575092, "grad_norm": 0.7265625, "learning_rate": 9.084468379614361e-05, "loss": 0.1206, "step": 11318 }, { "epoch": 8.292307692307693, "grad_norm": 0.93359375, "learning_rate": 9.084269846948861e-05, "loss": 0.16, "step": 11319 }, { "epoch": 8.293040293040294, "grad_norm": 1.046875, "learning_rate": 9.08407129495385e-05, "loss": 0.152, "step": 11320 }, { "epoch": 8.293772893772895, "grad_norm": 0.640625, "learning_rate": 9.08387272363028e-05, "loss": 0.1414, "step": 11321 }, { "epoch": 8.294505494505495, "grad_norm": 0.84765625, "learning_rate": 9.0836741329791e-05, "loss": 0.1322, "step": 11322 }, { "epoch": 8.295238095238096, "grad_norm": 1.390625, "learning_rate": 9.083475523001262e-05, "loss": 0.1774, "step": 11323 }, { "epoch": 8.295970695970697, "grad_norm": 1.109375, "learning_rate": 9.08327689369772e-05, "loss": 0.1578, "step": 11324 }, { "epoch": 8.296703296703297, "grad_norm": 1.0390625, "learning_rate": 9.083078245069425e-05, "loss": 0.1165, "step": 11325 }, { "epoch": 8.297435897435898, "grad_norm": 1.1953125, "learning_rate": 9.082879577117329e-05, "loss": 0.1332, "step": 11326 }, { "epoch": 8.298168498168499, "grad_norm": 1.0859375, "learning_rate": 9.082680889842382e-05, "loss": 0.1329, "step": 11327 }, { "epoch": 8.2989010989011, "grad_norm": 1.1328125, "learning_rate": 9.082482183245536e-05, "loss": 0.1519, "step": 11328 }, { "epoch": 8.2996336996337, "grad_norm": 0.9453125, "learning_rate": 9.082283457327745e-05, "loss": 0.1458, "step": 11329 }, { "epoch": 8.300366300366301, "grad_norm": 0.8125, "learning_rate": 9.08208471208996e-05, "loss": 0.1325, "step": 11330 }, { "epoch": 8.301098901098902, "grad_norm": 1.3828125, "learning_rate": 9.081885947533134e-05, "loss": 0.136, "step": 11331 }, { "epoch": 8.301831501831503, "grad_norm": 0.69140625, "learning_rate": 9.081687163658219e-05, "loss": 0.1322, "step": 11332 }, { "epoch": 8.302564102564103, "grad_norm": 1.5625, "learning_rate": 9.081488360466166e-05, "loss": 0.1622, "step": 11333 }, { "epoch": 8.303296703296704, "grad_norm": 1.125, "learning_rate": 9.081289537957929e-05, "loss": 0.1132, "step": 11334 }, { "epoch": 8.304029304029305, "grad_norm": 1.78125, "learning_rate": 9.081090696134462e-05, "loss": 0.1522, "step": 11335 }, { "epoch": 8.304761904761905, "grad_norm": 0.90625, "learning_rate": 9.080891834996716e-05, "loss": 0.1099, "step": 11336 }, { "epoch": 8.305494505494506, "grad_norm": 1.0703125, "learning_rate": 9.080692954545642e-05, "loss": 0.1438, "step": 11337 }, { "epoch": 8.306227106227107, "grad_norm": 1.328125, "learning_rate": 9.080494054782197e-05, "loss": 0.1498, "step": 11338 }, { "epoch": 8.306959706959708, "grad_norm": 1.5546875, "learning_rate": 9.08029513570733e-05, "loss": 0.1695, "step": 11339 }, { "epoch": 8.307692307692308, "grad_norm": 1.2109375, "learning_rate": 9.080096197321997e-05, "loss": 0.1565, "step": 11340 }, { "epoch": 8.308424908424909, "grad_norm": 1.2109375, "learning_rate": 9.07989723962715e-05, "loss": 0.1193, "step": 11341 }, { "epoch": 8.30915750915751, "grad_norm": 1.2578125, "learning_rate": 9.079698262623742e-05, "loss": 0.1389, "step": 11342 }, { "epoch": 8.30989010989011, "grad_norm": 1.3125, "learning_rate": 9.079499266312727e-05, "loss": 0.1436, "step": 11343 }, { "epoch": 8.310622710622711, "grad_norm": 0.93359375, "learning_rate": 9.079300250695058e-05, "loss": 0.1322, "step": 11344 }, { "epoch": 8.311355311355312, "grad_norm": 1.1953125, "learning_rate": 9.07910121577169e-05, "loss": 0.0952, "step": 11345 }, { "epoch": 8.312087912087913, "grad_norm": 0.82421875, "learning_rate": 9.078902161543573e-05, "loss": 0.1232, "step": 11346 }, { "epoch": 8.312820512820513, "grad_norm": 1.0703125, "learning_rate": 9.078703088011666e-05, "loss": 0.1416, "step": 11347 }, { "epoch": 8.313553113553114, "grad_norm": 1.25, "learning_rate": 9.078503995176917e-05, "loss": 0.1541, "step": 11348 }, { "epoch": 8.314285714285715, "grad_norm": 1.125, "learning_rate": 9.078304883040284e-05, "loss": 0.137, "step": 11349 }, { "epoch": 8.315018315018316, "grad_norm": 1.1875, "learning_rate": 9.078105751602723e-05, "loss": 0.1559, "step": 11350 }, { "epoch": 8.315750915750916, "grad_norm": 1.0390625, "learning_rate": 9.07790660086518e-05, "loss": 0.1139, "step": 11351 }, { "epoch": 8.316483516483517, "grad_norm": 1.734375, "learning_rate": 9.077707430828618e-05, "loss": 0.1622, "step": 11352 }, { "epoch": 8.317216117216118, "grad_norm": 0.8828125, "learning_rate": 9.077508241493986e-05, "loss": 0.1293, "step": 11353 }, { "epoch": 8.317948717948719, "grad_norm": 0.890625, "learning_rate": 9.07730903286224e-05, "loss": 0.1363, "step": 11354 }, { "epoch": 8.31868131868132, "grad_norm": 0.81640625, "learning_rate": 9.077109804934335e-05, "loss": 0.1, "step": 11355 }, { "epoch": 8.31941391941392, "grad_norm": 0.79296875, "learning_rate": 9.076910557711224e-05, "loss": 0.13, "step": 11356 }, { "epoch": 8.32014652014652, "grad_norm": 1.140625, "learning_rate": 9.076711291193861e-05, "loss": 0.1482, "step": 11357 }, { "epoch": 8.320879120879122, "grad_norm": 1.5625, "learning_rate": 9.076512005383204e-05, "loss": 0.132, "step": 11358 }, { "epoch": 8.321611721611722, "grad_norm": 1.34375, "learning_rate": 9.076312700280207e-05, "loss": 0.1316, "step": 11359 }, { "epoch": 8.322344322344323, "grad_norm": 1.09375, "learning_rate": 9.076113375885823e-05, "loss": 0.1389, "step": 11360 }, { "epoch": 8.323076923076924, "grad_norm": 1.796875, "learning_rate": 9.075914032201009e-05, "loss": 0.1423, "step": 11361 }, { "epoch": 8.323809523809524, "grad_norm": 1.4375, "learning_rate": 9.075714669226719e-05, "loss": 0.1718, "step": 11362 }, { "epoch": 8.324542124542125, "grad_norm": 0.82421875, "learning_rate": 9.075515286963909e-05, "loss": 0.1246, "step": 11363 }, { "epoch": 8.325274725274726, "grad_norm": 0.984375, "learning_rate": 9.075315885413531e-05, "loss": 0.1879, "step": 11364 }, { "epoch": 8.326007326007327, "grad_norm": 0.9453125, "learning_rate": 9.075116464576546e-05, "loss": 0.1464, "step": 11365 }, { "epoch": 8.326739926739927, "grad_norm": 1.40625, "learning_rate": 9.074917024453906e-05, "loss": 0.1074, "step": 11366 }, { "epoch": 8.327472527472528, "grad_norm": 0.87890625, "learning_rate": 9.074717565046567e-05, "loss": 0.1609, "step": 11367 }, { "epoch": 8.328205128205129, "grad_norm": 1.6640625, "learning_rate": 9.074518086355485e-05, "loss": 0.1835, "step": 11368 }, { "epoch": 8.32893772893773, "grad_norm": 1.2109375, "learning_rate": 9.074318588381613e-05, "loss": 0.1472, "step": 11369 }, { "epoch": 8.32967032967033, "grad_norm": 0.93359375, "learning_rate": 9.074119071125914e-05, "loss": 0.1344, "step": 11370 }, { "epoch": 8.330402930402931, "grad_norm": 1.03125, "learning_rate": 9.073919534589338e-05, "loss": 0.1375, "step": 11371 }, { "epoch": 8.331135531135532, "grad_norm": 0.96484375, "learning_rate": 9.073719978772842e-05, "loss": 0.1566, "step": 11372 }, { "epoch": 8.331868131868132, "grad_norm": 1.2109375, "learning_rate": 9.073520403677384e-05, "loss": 0.1389, "step": 11373 }, { "epoch": 8.332600732600733, "grad_norm": 0.82421875, "learning_rate": 9.073320809303918e-05, "loss": 0.1398, "step": 11374 }, { "epoch": 8.333333333333334, "grad_norm": 1.1328125, "learning_rate": 9.0731211956534e-05, "loss": 0.1331, "step": 11375 }, { "epoch": 8.334065934065935, "grad_norm": 1.1328125, "learning_rate": 9.072921562726789e-05, "loss": 0.1373, "step": 11376 }, { "epoch": 8.334798534798535, "grad_norm": 1.6640625, "learning_rate": 9.072721910525041e-05, "loss": 0.1456, "step": 11377 }, { "epoch": 8.335531135531136, "grad_norm": 0.92578125, "learning_rate": 9.072522239049108e-05, "loss": 0.1525, "step": 11378 }, { "epoch": 8.336263736263737, "grad_norm": 0.9765625, "learning_rate": 9.072322548299955e-05, "loss": 0.1418, "step": 11379 }, { "epoch": 8.336996336996338, "grad_norm": 1.375, "learning_rate": 9.072122838278531e-05, "loss": 0.1501, "step": 11380 }, { "epoch": 8.337728937728938, "grad_norm": 1.2109375, "learning_rate": 9.071923108985798e-05, "loss": 0.1096, "step": 11381 }, { "epoch": 8.338461538461539, "grad_norm": 0.5546875, "learning_rate": 9.071723360422711e-05, "loss": 0.1287, "step": 11382 }, { "epoch": 8.33919413919414, "grad_norm": 0.77734375, "learning_rate": 9.071523592590227e-05, "loss": 0.1149, "step": 11383 }, { "epoch": 8.33992673992674, "grad_norm": 0.8359375, "learning_rate": 9.071323805489302e-05, "loss": 0.1177, "step": 11384 }, { "epoch": 8.340659340659341, "grad_norm": 0.6953125, "learning_rate": 9.071123999120895e-05, "loss": 0.1191, "step": 11385 }, { "epoch": 8.341391941391942, "grad_norm": 1.5625, "learning_rate": 9.070924173485964e-05, "loss": 0.1686, "step": 11386 }, { "epoch": 8.342124542124543, "grad_norm": 1.2734375, "learning_rate": 9.070724328585465e-05, "loss": 0.1454, "step": 11387 }, { "epoch": 8.342857142857143, "grad_norm": 0.7421875, "learning_rate": 9.070524464420354e-05, "loss": 0.1163, "step": 11388 }, { "epoch": 8.343589743589744, "grad_norm": 0.796875, "learning_rate": 9.070324580991592e-05, "loss": 0.1482, "step": 11389 }, { "epoch": 8.344322344322345, "grad_norm": 0.78125, "learning_rate": 9.070124678300135e-05, "loss": 0.1523, "step": 11390 }, { "epoch": 8.345054945054946, "grad_norm": 0.875, "learning_rate": 9.069924756346942e-05, "loss": 0.1276, "step": 11391 }, { "epoch": 8.345787545787546, "grad_norm": 1.09375, "learning_rate": 9.069724815132969e-05, "loss": 0.147, "step": 11392 }, { "epoch": 8.346520146520147, "grad_norm": 1.3828125, "learning_rate": 9.069524854659174e-05, "loss": 0.1509, "step": 11393 }, { "epoch": 8.347252747252748, "grad_norm": 1.265625, "learning_rate": 9.069324874926517e-05, "loss": 0.1314, "step": 11394 }, { "epoch": 8.347985347985349, "grad_norm": 0.96484375, "learning_rate": 9.069124875935954e-05, "loss": 0.1542, "step": 11395 }, { "epoch": 8.34871794871795, "grad_norm": 1.3046875, "learning_rate": 9.068924857688444e-05, "loss": 0.1326, "step": 11396 }, { "epoch": 8.34945054945055, "grad_norm": 0.76953125, "learning_rate": 9.068724820184948e-05, "loss": 0.1204, "step": 11397 }, { "epoch": 8.35018315018315, "grad_norm": 1.140625, "learning_rate": 9.06852476342642e-05, "loss": 0.1573, "step": 11398 }, { "epoch": 8.350915750915751, "grad_norm": 1.59375, "learning_rate": 9.068324687413821e-05, "loss": 0.1708, "step": 11399 }, { "epoch": 8.351648351648352, "grad_norm": 1.1015625, "learning_rate": 9.068124592148109e-05, "loss": 0.1215, "step": 11400 }, { "epoch": 8.352380952380953, "grad_norm": 0.71875, "learning_rate": 9.067924477630245e-05, "loss": 0.1188, "step": 11401 }, { "epoch": 8.353113553113554, "grad_norm": 0.83203125, "learning_rate": 9.067724343861185e-05, "loss": 0.1443, "step": 11402 }, { "epoch": 8.353846153846154, "grad_norm": 0.7734375, "learning_rate": 9.067524190841886e-05, "loss": 0.1467, "step": 11403 }, { "epoch": 8.354578754578755, "grad_norm": 1.34375, "learning_rate": 9.067324018573314e-05, "loss": 0.165, "step": 11404 }, { "epoch": 8.355311355311356, "grad_norm": 1.0078125, "learning_rate": 9.067123827056421e-05, "loss": 0.1259, "step": 11405 }, { "epoch": 8.356043956043957, "grad_norm": 0.9296875, "learning_rate": 9.06692361629217e-05, "loss": 0.1152, "step": 11406 }, { "epoch": 8.356776556776557, "grad_norm": 1.03125, "learning_rate": 9.066723386281519e-05, "loss": 0.1353, "step": 11407 }, { "epoch": 8.357509157509158, "grad_norm": 0.921875, "learning_rate": 9.06652313702543e-05, "loss": 0.1167, "step": 11408 }, { "epoch": 8.358241758241759, "grad_norm": 1.3359375, "learning_rate": 9.066322868524855e-05, "loss": 0.154, "step": 11409 }, { "epoch": 8.35897435897436, "grad_norm": 0.8359375, "learning_rate": 9.066122580780761e-05, "loss": 0.1427, "step": 11410 }, { "epoch": 8.35970695970696, "grad_norm": 1.5078125, "learning_rate": 9.065922273794107e-05, "loss": 0.1771, "step": 11411 }, { "epoch": 8.360439560439561, "grad_norm": 1.078125, "learning_rate": 9.065721947565848e-05, "loss": 0.1498, "step": 11412 }, { "epoch": 8.361172161172162, "grad_norm": 0.83203125, "learning_rate": 9.06552160209695e-05, "loss": 0.1486, "step": 11413 }, { "epoch": 8.361904761904762, "grad_norm": 1.9140625, "learning_rate": 9.065321237388368e-05, "loss": 0.1532, "step": 11414 }, { "epoch": 8.362637362637363, "grad_norm": 1.578125, "learning_rate": 9.065120853441066e-05, "loss": 0.1396, "step": 11415 }, { "epoch": 8.363369963369964, "grad_norm": 1.140625, "learning_rate": 9.064920450255999e-05, "loss": 0.1346, "step": 11416 }, { "epoch": 8.364102564102565, "grad_norm": 0.9609375, "learning_rate": 9.064720027834132e-05, "loss": 0.1398, "step": 11417 }, { "epoch": 8.364835164835165, "grad_norm": 1.25, "learning_rate": 9.064519586176424e-05, "loss": 0.1403, "step": 11418 }, { "epoch": 8.365567765567766, "grad_norm": 1.453125, "learning_rate": 9.064319125283832e-05, "loss": 0.1506, "step": 11419 }, { "epoch": 8.366300366300367, "grad_norm": 0.95703125, "learning_rate": 9.064118645157322e-05, "loss": 0.1564, "step": 11420 }, { "epoch": 8.367032967032968, "grad_norm": 0.87109375, "learning_rate": 9.06391814579785e-05, "loss": 0.1189, "step": 11421 }, { "epoch": 8.367765567765568, "grad_norm": 0.74609375, "learning_rate": 9.063717627206378e-05, "loss": 0.1444, "step": 11422 }, { "epoch": 8.368498168498169, "grad_norm": 1.25, "learning_rate": 9.06351708938387e-05, "loss": 0.1024, "step": 11423 }, { "epoch": 8.36923076923077, "grad_norm": 0.97265625, "learning_rate": 9.06331653233128e-05, "loss": 0.1531, "step": 11424 }, { "epoch": 8.36996336996337, "grad_norm": 1.0078125, "learning_rate": 9.063115956049578e-05, "loss": 0.1034, "step": 11425 }, { "epoch": 8.370695970695971, "grad_norm": 0.890625, "learning_rate": 9.062915360539716e-05, "loss": 0.1488, "step": 11426 }, { "epoch": 8.371428571428572, "grad_norm": 0.90625, "learning_rate": 9.062714745802661e-05, "loss": 0.1495, "step": 11427 }, { "epoch": 8.372161172161173, "grad_norm": 1.0, "learning_rate": 9.062514111839372e-05, "loss": 0.1517, "step": 11428 }, { "epoch": 8.372893772893773, "grad_norm": 1.21875, "learning_rate": 9.06231345865081e-05, "loss": 0.1448, "step": 11429 }, { "epoch": 8.373626373626374, "grad_norm": 0.69921875, "learning_rate": 9.062112786237939e-05, "loss": 0.1485, "step": 11430 }, { "epoch": 8.374358974358975, "grad_norm": 0.6796875, "learning_rate": 9.061912094601718e-05, "loss": 0.1089, "step": 11431 }, { "epoch": 8.375091575091576, "grad_norm": 0.96484375, "learning_rate": 9.061711383743106e-05, "loss": 0.1682, "step": 11432 }, { "epoch": 8.375824175824176, "grad_norm": 1.1015625, "learning_rate": 9.061510653663073e-05, "loss": 0.1467, "step": 11433 }, { "epoch": 8.376556776556777, "grad_norm": 1.4453125, "learning_rate": 9.06130990436257e-05, "loss": 0.1262, "step": 11434 }, { "epoch": 8.377289377289378, "grad_norm": 1.7734375, "learning_rate": 9.061109135842568e-05, "loss": 0.1228, "step": 11435 }, { "epoch": 8.378021978021978, "grad_norm": 0.91015625, "learning_rate": 9.060908348104025e-05, "loss": 0.1495, "step": 11436 }, { "epoch": 8.37875457875458, "grad_norm": 0.96875, "learning_rate": 9.060707541147902e-05, "loss": 0.1326, "step": 11437 }, { "epoch": 8.37948717948718, "grad_norm": 0.63671875, "learning_rate": 9.060506714975164e-05, "loss": 0.1278, "step": 11438 }, { "epoch": 8.38021978021978, "grad_norm": 1.2265625, "learning_rate": 9.060305869586772e-05, "loss": 0.1461, "step": 11439 }, { "epoch": 8.380952380952381, "grad_norm": 0.8359375, "learning_rate": 9.060105004983687e-05, "loss": 0.128, "step": 11440 }, { "epoch": 8.381684981684982, "grad_norm": 1.7109375, "learning_rate": 9.059904121166873e-05, "loss": 0.126, "step": 11441 }, { "epoch": 8.382417582417583, "grad_norm": 1.59375, "learning_rate": 9.059703218137291e-05, "loss": 0.1419, "step": 11442 }, { "epoch": 8.383150183150184, "grad_norm": 0.84765625, "learning_rate": 9.059502295895906e-05, "loss": 0.1372, "step": 11443 }, { "epoch": 8.383882783882784, "grad_norm": 0.6953125, "learning_rate": 9.059301354443678e-05, "loss": 0.1282, "step": 11444 }, { "epoch": 8.384615384615385, "grad_norm": 1.0390625, "learning_rate": 9.059100393781572e-05, "loss": 0.1404, "step": 11445 }, { "epoch": 8.385347985347986, "grad_norm": 1.234375, "learning_rate": 9.05889941391055e-05, "loss": 0.1492, "step": 11446 }, { "epoch": 8.386080586080586, "grad_norm": 0.83984375, "learning_rate": 9.058698414831574e-05, "loss": 0.1383, "step": 11447 }, { "epoch": 8.386813186813187, "grad_norm": 1.0546875, "learning_rate": 9.058497396545608e-05, "loss": 0.1353, "step": 11448 }, { "epoch": 8.387545787545788, "grad_norm": 0.8984375, "learning_rate": 9.058296359053616e-05, "loss": 0.1604, "step": 11449 }, { "epoch": 8.388278388278389, "grad_norm": 0.8671875, "learning_rate": 9.05809530235656e-05, "loss": 0.1358, "step": 11450 }, { "epoch": 8.38901098901099, "grad_norm": 1.4375, "learning_rate": 9.057894226455403e-05, "loss": 0.1406, "step": 11451 }, { "epoch": 8.38974358974359, "grad_norm": 0.72265625, "learning_rate": 9.057693131351109e-05, "loss": 0.1116, "step": 11452 }, { "epoch": 8.39047619047619, "grad_norm": 1.125, "learning_rate": 9.057492017044643e-05, "loss": 0.1323, "step": 11453 }, { "epoch": 8.391208791208792, "grad_norm": 1.125, "learning_rate": 9.057290883536965e-05, "loss": 0.1607, "step": 11454 }, { "epoch": 8.391941391941392, "grad_norm": 0.8125, "learning_rate": 9.057089730829042e-05, "loss": 0.1234, "step": 11455 }, { "epoch": 8.392673992673993, "grad_norm": 1.234375, "learning_rate": 9.056888558921838e-05, "loss": 0.1576, "step": 11456 }, { "epoch": 8.393406593406594, "grad_norm": 1.1171875, "learning_rate": 9.056687367816313e-05, "loss": 0.1797, "step": 11457 }, { "epoch": 8.394139194139195, "grad_norm": 1.265625, "learning_rate": 9.056486157513437e-05, "loss": 0.1435, "step": 11458 }, { "epoch": 8.394871794871795, "grad_norm": 0.93359375, "learning_rate": 9.056284928014167e-05, "loss": 0.1388, "step": 11459 }, { "epoch": 8.395604395604396, "grad_norm": 1.1171875, "learning_rate": 9.056083679319473e-05, "loss": 0.1752, "step": 11460 }, { "epoch": 8.396336996336997, "grad_norm": 0.76171875, "learning_rate": 9.055882411430315e-05, "loss": 0.1403, "step": 11461 }, { "epoch": 8.397069597069597, "grad_norm": 0.796875, "learning_rate": 9.05568112434766e-05, "loss": 0.1347, "step": 11462 }, { "epoch": 8.397802197802198, "grad_norm": 1.1796875, "learning_rate": 9.055479818072472e-05, "loss": 0.1636, "step": 11463 }, { "epoch": 8.398534798534799, "grad_norm": 1.0, "learning_rate": 9.055278492605717e-05, "loss": 0.1409, "step": 11464 }, { "epoch": 8.3992673992674, "grad_norm": 1.390625, "learning_rate": 9.055077147948356e-05, "loss": 0.1439, "step": 11465 }, { "epoch": 8.4, "grad_norm": 0.80078125, "learning_rate": 9.054875784101356e-05, "loss": 0.1447, "step": 11466 }, { "epoch": 8.400732600732601, "grad_norm": 0.91796875, "learning_rate": 9.05467440106568e-05, "loss": 0.1385, "step": 11467 }, { "epoch": 8.401465201465202, "grad_norm": 1.203125, "learning_rate": 9.054472998842297e-05, "loss": 0.1525, "step": 11468 }, { "epoch": 8.402197802197803, "grad_norm": 1.125, "learning_rate": 9.054271577432168e-05, "loss": 0.147, "step": 11469 }, { "epoch": 8.402930402930403, "grad_norm": 1.5234375, "learning_rate": 9.05407013683626e-05, "loss": 0.1343, "step": 11470 }, { "epoch": 8.403663003663004, "grad_norm": 1.3046875, "learning_rate": 9.053868677055535e-05, "loss": 0.1736, "step": 11471 }, { "epoch": 8.404395604395605, "grad_norm": 1.671875, "learning_rate": 9.053667198090964e-05, "loss": 0.1357, "step": 11472 }, { "epoch": 8.405128205128205, "grad_norm": 1.6015625, "learning_rate": 9.053465699943508e-05, "loss": 0.1546, "step": 11473 }, { "epoch": 8.405860805860806, "grad_norm": 0.92578125, "learning_rate": 9.053264182614133e-05, "loss": 0.1071, "step": 11474 }, { "epoch": 8.406593406593407, "grad_norm": 0.78515625, "learning_rate": 9.053062646103805e-05, "loss": 0.1117, "step": 11475 }, { "epoch": 8.407326007326008, "grad_norm": 1.4765625, "learning_rate": 9.052861090413491e-05, "loss": 0.1672, "step": 11476 }, { "epoch": 8.408058608058608, "grad_norm": 1.5625, "learning_rate": 9.052659515544152e-05, "loss": 0.1339, "step": 11477 }, { "epoch": 8.40879120879121, "grad_norm": 0.91015625, "learning_rate": 9.05245792149676e-05, "loss": 0.124, "step": 11478 }, { "epoch": 8.40952380952381, "grad_norm": 1.1796875, "learning_rate": 9.05225630827228e-05, "loss": 0.1316, "step": 11479 }, { "epoch": 8.41025641025641, "grad_norm": 1.1171875, "learning_rate": 9.052054675871675e-05, "loss": 0.1049, "step": 11480 }, { "epoch": 8.410989010989011, "grad_norm": 0.6328125, "learning_rate": 9.05185302429591e-05, "loss": 0.1057, "step": 11481 }, { "epoch": 8.411721611721612, "grad_norm": 1.078125, "learning_rate": 9.051651353545954e-05, "loss": 0.1338, "step": 11482 }, { "epoch": 8.412454212454213, "grad_norm": 1.8828125, "learning_rate": 9.051449663622774e-05, "loss": 0.1339, "step": 11483 }, { "epoch": 8.413186813186813, "grad_norm": 2.1875, "learning_rate": 9.051247954527335e-05, "loss": 0.1258, "step": 11484 }, { "epoch": 8.413919413919414, "grad_norm": 2.03125, "learning_rate": 9.051046226260604e-05, "loss": 0.1384, "step": 11485 }, { "epoch": 8.414652014652015, "grad_norm": 1.59375, "learning_rate": 9.050844478823547e-05, "loss": 0.1511, "step": 11486 }, { "epoch": 8.415384615384616, "grad_norm": 1.109375, "learning_rate": 9.05064271221713e-05, "loss": 0.1481, "step": 11487 }, { "epoch": 8.416117216117216, "grad_norm": 1.03125, "learning_rate": 9.05044092644232e-05, "loss": 0.1298, "step": 11488 }, { "epoch": 8.416849816849817, "grad_norm": 1.484375, "learning_rate": 9.050239121500085e-05, "loss": 0.1317, "step": 11489 }, { "epoch": 8.417582417582418, "grad_norm": 1.5703125, "learning_rate": 9.050037297391391e-05, "loss": 0.1405, "step": 11490 }, { "epoch": 8.418315018315019, "grad_norm": 2.40625, "learning_rate": 9.049835454117204e-05, "loss": 0.1378, "step": 11491 }, { "epoch": 8.41904761904762, "grad_norm": 0.96484375, "learning_rate": 9.049633591678494e-05, "loss": 0.1481, "step": 11492 }, { "epoch": 8.41978021978022, "grad_norm": 0.80078125, "learning_rate": 9.049431710076225e-05, "loss": 0.1271, "step": 11493 }, { "epoch": 8.42051282051282, "grad_norm": 0.82421875, "learning_rate": 9.049229809311367e-05, "loss": 0.142, "step": 11494 }, { "epoch": 8.421245421245422, "grad_norm": 1.0703125, "learning_rate": 9.049027889384886e-05, "loss": 0.1104, "step": 11495 }, { "epoch": 8.421978021978022, "grad_norm": 0.765625, "learning_rate": 9.04882595029775e-05, "loss": 0.1271, "step": 11496 }, { "epoch": 8.422710622710623, "grad_norm": 1.078125, "learning_rate": 9.048623992050926e-05, "loss": 0.1486, "step": 11497 }, { "epoch": 8.423443223443224, "grad_norm": 1.0703125, "learning_rate": 9.04842201464538e-05, "loss": 0.119, "step": 11498 }, { "epoch": 8.424175824175824, "grad_norm": 1.3125, "learning_rate": 9.048220018082083e-05, "loss": 0.1513, "step": 11499 }, { "epoch": 8.424908424908425, "grad_norm": 0.84375, "learning_rate": 9.048018002362001e-05, "loss": 0.1765, "step": 11500 }, { "epoch": 8.425641025641026, "grad_norm": 1.4375, "learning_rate": 9.047815967486103e-05, "loss": 0.1392, "step": 11501 }, { "epoch": 8.426373626373627, "grad_norm": 1.296875, "learning_rate": 9.047613913455355e-05, "loss": 0.1501, "step": 11502 }, { "epoch": 8.427106227106227, "grad_norm": 1.8984375, "learning_rate": 9.047411840270728e-05, "loss": 0.1141, "step": 11503 }, { "epoch": 8.427838827838828, "grad_norm": 1.390625, "learning_rate": 9.047209747933187e-05, "loss": 0.1469, "step": 11504 }, { "epoch": 8.428571428571429, "grad_norm": 1.09375, "learning_rate": 9.047007636443702e-05, "loss": 0.166, "step": 11505 }, { "epoch": 8.42930402930403, "grad_norm": 1.359375, "learning_rate": 9.046805505803244e-05, "loss": 0.1325, "step": 11506 }, { "epoch": 8.43003663003663, "grad_norm": 1.1484375, "learning_rate": 9.046603356012775e-05, "loss": 0.1601, "step": 11507 }, { "epoch": 8.430769230769231, "grad_norm": 1.8515625, "learning_rate": 9.046401187073269e-05, "loss": 0.141, "step": 11508 }, { "epoch": 8.431501831501832, "grad_norm": 1.4140625, "learning_rate": 9.046198998985694e-05, "loss": 0.14, "step": 11509 }, { "epoch": 8.432234432234432, "grad_norm": 0.93359375, "learning_rate": 9.045996791751016e-05, "loss": 0.151, "step": 11510 }, { "epoch": 8.432967032967033, "grad_norm": 0.52734375, "learning_rate": 9.045794565370206e-05, "loss": 0.1151, "step": 11511 }, { "epoch": 8.433699633699634, "grad_norm": 0.69140625, "learning_rate": 9.045592319844232e-05, "loss": 0.1192, "step": 11512 }, { "epoch": 8.434432234432235, "grad_norm": 1.296875, "learning_rate": 9.045390055174065e-05, "loss": 0.1358, "step": 11513 }, { "epoch": 8.435164835164835, "grad_norm": 1.3515625, "learning_rate": 9.045187771360673e-05, "loss": 0.1375, "step": 11514 }, { "epoch": 8.435897435897436, "grad_norm": 1.34375, "learning_rate": 9.044985468405023e-05, "loss": 0.1416, "step": 11515 }, { "epoch": 8.436630036630037, "grad_norm": 0.9609375, "learning_rate": 9.044783146308086e-05, "loss": 0.1268, "step": 11516 }, { "epoch": 8.437362637362638, "grad_norm": 0.91796875, "learning_rate": 9.044580805070834e-05, "loss": 0.1387, "step": 11517 }, { "epoch": 8.438095238095238, "grad_norm": 1.03125, "learning_rate": 9.044378444694233e-05, "loss": 0.1867, "step": 11518 }, { "epoch": 8.438827838827839, "grad_norm": 1.03125, "learning_rate": 9.044176065179251e-05, "loss": 0.1289, "step": 11519 }, { "epoch": 8.43956043956044, "grad_norm": 0.8671875, "learning_rate": 9.043973666526864e-05, "loss": 0.1103, "step": 11520 }, { "epoch": 8.44029304029304, "grad_norm": 1.65625, "learning_rate": 9.043771248738036e-05, "loss": 0.1421, "step": 11521 }, { "epoch": 8.441025641025641, "grad_norm": 1.0859375, "learning_rate": 9.043568811813739e-05, "loss": 0.1438, "step": 11522 }, { "epoch": 8.441758241758242, "grad_norm": 1.2265625, "learning_rate": 9.043366355754944e-05, "loss": 0.1683, "step": 11523 }, { "epoch": 8.442490842490843, "grad_norm": 0.5390625, "learning_rate": 9.043163880562618e-05, "loss": 0.1363, "step": 11524 }, { "epoch": 8.443223443223443, "grad_norm": 1.203125, "learning_rate": 9.042961386237735e-05, "loss": 0.1566, "step": 11525 }, { "epoch": 8.443956043956044, "grad_norm": 0.70703125, "learning_rate": 9.042758872781261e-05, "loss": 0.1269, "step": 11526 }, { "epoch": 8.444688644688645, "grad_norm": 0.79296875, "learning_rate": 9.04255634019417e-05, "loss": 0.1647, "step": 11527 }, { "epoch": 8.445421245421246, "grad_norm": 0.80078125, "learning_rate": 9.042353788477432e-05, "loss": 0.1475, "step": 11528 }, { "epoch": 8.446153846153846, "grad_norm": 1.2109375, "learning_rate": 9.042151217632015e-05, "loss": 0.1379, "step": 11529 }, { "epoch": 8.446886446886447, "grad_norm": 0.9453125, "learning_rate": 9.041948627658892e-05, "loss": 0.1237, "step": 11530 }, { "epoch": 8.447619047619048, "grad_norm": 0.8046875, "learning_rate": 9.04174601855903e-05, "loss": 0.1605, "step": 11531 }, { "epoch": 8.448351648351649, "grad_norm": 1.109375, "learning_rate": 9.041543390333403e-05, "loss": 0.1447, "step": 11532 }, { "epoch": 8.44908424908425, "grad_norm": 0.625, "learning_rate": 9.041340742982983e-05, "loss": 0.1385, "step": 11533 }, { "epoch": 8.44981684981685, "grad_norm": 0.85546875, "learning_rate": 9.041138076508739e-05, "loss": 0.1317, "step": 11534 }, { "epoch": 8.45054945054945, "grad_norm": 0.75, "learning_rate": 9.040935390911642e-05, "loss": 0.1298, "step": 11535 }, { "epoch": 8.451282051282051, "grad_norm": 1.09375, "learning_rate": 9.040732686192662e-05, "loss": 0.098, "step": 11536 }, { "epoch": 8.452014652014652, "grad_norm": 0.984375, "learning_rate": 9.040529962352773e-05, "loss": 0.1385, "step": 11537 }, { "epoch": 8.452747252747253, "grad_norm": 1.125, "learning_rate": 9.040327219392946e-05, "loss": 0.15, "step": 11538 }, { "epoch": 8.453479853479854, "grad_norm": 1.1015625, "learning_rate": 9.04012445731415e-05, "loss": 0.1221, "step": 11539 }, { "epoch": 8.454212454212454, "grad_norm": 1.625, "learning_rate": 9.039921676117357e-05, "loss": 0.1687, "step": 11540 }, { "epoch": 8.454945054945055, "grad_norm": 0.796875, "learning_rate": 9.03971887580354e-05, "loss": 0.1376, "step": 11541 }, { "epoch": 8.455677655677656, "grad_norm": 0.875, "learning_rate": 9.039516056373671e-05, "loss": 0.1445, "step": 11542 }, { "epoch": 8.456410256410257, "grad_norm": 1.234375, "learning_rate": 9.039313217828721e-05, "loss": 0.1545, "step": 11543 }, { "epoch": 8.457142857142857, "grad_norm": 1.1796875, "learning_rate": 9.039110360169661e-05, "loss": 0.1494, "step": 11544 }, { "epoch": 8.457875457875458, "grad_norm": 0.92578125, "learning_rate": 9.038907483397463e-05, "loss": 0.1255, "step": 11545 }, { "epoch": 8.458608058608059, "grad_norm": 0.6875, "learning_rate": 9.038704587513102e-05, "loss": 0.1401, "step": 11546 }, { "epoch": 8.45934065934066, "grad_norm": 1.046875, "learning_rate": 9.038501672517547e-05, "loss": 0.128, "step": 11547 }, { "epoch": 8.46007326007326, "grad_norm": 0.93359375, "learning_rate": 9.03829873841177e-05, "loss": 0.1404, "step": 11548 }, { "epoch": 8.460805860805861, "grad_norm": 0.87109375, "learning_rate": 9.038095785196746e-05, "loss": 0.1413, "step": 11549 }, { "epoch": 8.461538461538462, "grad_norm": 0.8828125, "learning_rate": 9.037892812873444e-05, "loss": 0.137, "step": 11550 }, { "epoch": 8.462271062271062, "grad_norm": 1.40625, "learning_rate": 9.03768982144284e-05, "loss": 0.1454, "step": 11551 }, { "epoch": 8.463003663003663, "grad_norm": 0.98828125, "learning_rate": 9.037486810905906e-05, "loss": 0.1315, "step": 11552 }, { "epoch": 8.463736263736264, "grad_norm": 1.3359375, "learning_rate": 9.03728378126361e-05, "loss": 0.1242, "step": 11553 }, { "epoch": 8.464468864468865, "grad_norm": 0.67578125, "learning_rate": 9.037080732516931e-05, "loss": 0.123, "step": 11554 }, { "epoch": 8.465201465201465, "grad_norm": 1.4375, "learning_rate": 9.036877664666839e-05, "loss": 0.1354, "step": 11555 }, { "epoch": 8.465934065934066, "grad_norm": 0.8828125, "learning_rate": 9.036674577714307e-05, "loss": 0.1459, "step": 11556 }, { "epoch": 8.466666666666667, "grad_norm": 1.1640625, "learning_rate": 9.03647147166031e-05, "loss": 0.1452, "step": 11557 }, { "epoch": 8.467399267399268, "grad_norm": 0.86328125, "learning_rate": 9.036268346505818e-05, "loss": 0.116, "step": 11558 }, { "epoch": 8.468131868131868, "grad_norm": 1.6328125, "learning_rate": 9.036065202251806e-05, "loss": 0.1647, "step": 11559 }, { "epoch": 8.468864468864469, "grad_norm": 1.3125, "learning_rate": 9.035862038899246e-05, "loss": 0.1612, "step": 11560 }, { "epoch": 8.46959706959707, "grad_norm": 1.2109375, "learning_rate": 9.035658856449114e-05, "loss": 0.1181, "step": 11561 }, { "epoch": 8.47032967032967, "grad_norm": 0.94921875, "learning_rate": 9.035455654902382e-05, "loss": 0.1506, "step": 11562 }, { "epoch": 8.471062271062271, "grad_norm": 0.6796875, "learning_rate": 9.035252434260023e-05, "loss": 0.1504, "step": 11563 }, { "epoch": 8.471794871794872, "grad_norm": 1.6015625, "learning_rate": 9.03504919452301e-05, "loss": 0.1524, "step": 11564 }, { "epoch": 8.472527472527473, "grad_norm": 1.421875, "learning_rate": 9.034845935692321e-05, "loss": 0.153, "step": 11565 }, { "epoch": 8.473260073260073, "grad_norm": 1.2109375, "learning_rate": 9.034642657768925e-05, "loss": 0.1087, "step": 11566 }, { "epoch": 8.473992673992674, "grad_norm": 1.1484375, "learning_rate": 9.034439360753798e-05, "loss": 0.127, "step": 11567 }, { "epoch": 8.474725274725275, "grad_norm": 1.4921875, "learning_rate": 9.034236044647915e-05, "loss": 0.1626, "step": 11568 }, { "epoch": 8.475457875457876, "grad_norm": 0.8828125, "learning_rate": 9.034032709452248e-05, "loss": 0.1412, "step": 11569 }, { "epoch": 8.476190476190476, "grad_norm": 1.2265625, "learning_rate": 9.033829355167775e-05, "loss": 0.1483, "step": 11570 }, { "epoch": 8.476923076923077, "grad_norm": 1.0390625, "learning_rate": 9.033625981795464e-05, "loss": 0.1279, "step": 11571 }, { "epoch": 8.477655677655678, "grad_norm": 0.66796875, "learning_rate": 9.033422589336295e-05, "loss": 0.1312, "step": 11572 }, { "epoch": 8.478388278388278, "grad_norm": 0.921875, "learning_rate": 9.033219177791243e-05, "loss": 0.1258, "step": 11573 }, { "epoch": 8.47912087912088, "grad_norm": 0.546875, "learning_rate": 9.033015747161276e-05, "loss": 0.1113, "step": 11574 }, { "epoch": 8.47985347985348, "grad_norm": 1.0, "learning_rate": 9.032812297447376e-05, "loss": 0.143, "step": 11575 }, { "epoch": 8.48058608058608, "grad_norm": 0.8828125, "learning_rate": 9.032608828650513e-05, "loss": 0.1677, "step": 11576 }, { "epoch": 8.481318681318681, "grad_norm": 0.796875, "learning_rate": 9.032405340771666e-05, "loss": 0.1458, "step": 11577 }, { "epoch": 8.482051282051282, "grad_norm": 1.0234375, "learning_rate": 9.032201833811806e-05, "loss": 0.1346, "step": 11578 }, { "epoch": 8.482783882783883, "grad_norm": 0.82421875, "learning_rate": 9.03199830777191e-05, "loss": 0.0996, "step": 11579 }, { "epoch": 8.483516483516484, "grad_norm": 1.0390625, "learning_rate": 9.031794762652952e-05, "loss": 0.1097, "step": 11580 }, { "epoch": 8.484249084249084, "grad_norm": 0.84765625, "learning_rate": 9.031591198455908e-05, "loss": 0.1501, "step": 11581 }, { "epoch": 8.484981684981685, "grad_norm": 1.1015625, "learning_rate": 9.031387615181756e-05, "loss": 0.1425, "step": 11582 }, { "epoch": 8.485714285714286, "grad_norm": 1.3828125, "learning_rate": 9.031184012831468e-05, "loss": 0.126, "step": 11583 }, { "epoch": 8.486446886446886, "grad_norm": 2.40625, "learning_rate": 9.030980391406019e-05, "loss": 0.1909, "step": 11584 }, { "epoch": 8.487179487179487, "grad_norm": 1.3359375, "learning_rate": 9.030776750906387e-05, "loss": 0.1711, "step": 11585 }, { "epoch": 8.487912087912088, "grad_norm": 0.85546875, "learning_rate": 9.030573091333547e-05, "loss": 0.1538, "step": 11586 }, { "epoch": 8.488644688644689, "grad_norm": 1.40625, "learning_rate": 9.030369412688475e-05, "loss": 0.1669, "step": 11587 }, { "epoch": 8.48937728937729, "grad_norm": 1.8984375, "learning_rate": 9.030165714972145e-05, "loss": 0.1439, "step": 11588 }, { "epoch": 8.49010989010989, "grad_norm": 0.7890625, "learning_rate": 9.029961998185536e-05, "loss": 0.1266, "step": 11589 }, { "epoch": 8.49084249084249, "grad_norm": 1.0390625, "learning_rate": 9.029758262329621e-05, "loss": 0.1428, "step": 11590 }, { "epoch": 8.491575091575092, "grad_norm": 0.95703125, "learning_rate": 9.029554507405379e-05, "loss": 0.1192, "step": 11591 }, { "epoch": 8.492307692307692, "grad_norm": 1.0546875, "learning_rate": 9.029350733413783e-05, "loss": 0.1128, "step": 11592 }, { "epoch": 8.493040293040293, "grad_norm": 1.859375, "learning_rate": 9.029146940355814e-05, "loss": 0.1606, "step": 11593 }, { "epoch": 8.493772893772894, "grad_norm": 1.0625, "learning_rate": 9.028943128232444e-05, "loss": 0.154, "step": 11594 }, { "epoch": 8.494505494505495, "grad_norm": 1.390625, "learning_rate": 9.028739297044653e-05, "loss": 0.1196, "step": 11595 }, { "epoch": 8.495238095238095, "grad_norm": 0.7734375, "learning_rate": 9.028535446793415e-05, "loss": 0.1539, "step": 11596 }, { "epoch": 8.495970695970696, "grad_norm": 0.94140625, "learning_rate": 9.028331577479706e-05, "loss": 0.1258, "step": 11597 }, { "epoch": 8.496703296703297, "grad_norm": 0.90234375, "learning_rate": 9.028127689104506e-05, "loss": 0.154, "step": 11598 }, { "epoch": 8.497435897435897, "grad_norm": 0.6953125, "learning_rate": 9.02792378166879e-05, "loss": 0.1302, "step": 11599 }, { "epoch": 8.498168498168498, "grad_norm": 1.9453125, "learning_rate": 9.027719855173536e-05, "loss": 0.1364, "step": 11600 }, { "epoch": 8.498901098901099, "grad_norm": 1.1015625, "learning_rate": 9.02751590961972e-05, "loss": 0.1472, "step": 11601 }, { "epoch": 8.4996336996337, "grad_norm": 1.8203125, "learning_rate": 9.027311945008317e-05, "loss": 0.1634, "step": 11602 }, { "epoch": 8.5003663003663, "grad_norm": 0.9140625, "learning_rate": 9.02710796134031e-05, "loss": 0.1489, "step": 11603 }, { "epoch": 8.501098901098901, "grad_norm": 0.8046875, "learning_rate": 9.026903958616674e-05, "loss": 0.1515, "step": 11604 }, { "epoch": 8.501831501831502, "grad_norm": 0.984375, "learning_rate": 9.026699936838382e-05, "loss": 0.1507, "step": 11605 }, { "epoch": 8.502564102564103, "grad_norm": 1.078125, "learning_rate": 9.026495896006418e-05, "loss": 0.1641, "step": 11606 }, { "epoch": 8.503296703296703, "grad_norm": 0.77734375, "learning_rate": 9.026291836121757e-05, "loss": 0.1212, "step": 11607 }, { "epoch": 8.504029304029304, "grad_norm": 0.828125, "learning_rate": 9.026087757185373e-05, "loss": 0.1665, "step": 11608 }, { "epoch": 8.504761904761905, "grad_norm": 0.60546875, "learning_rate": 9.025883659198249e-05, "loss": 0.1252, "step": 11609 }, { "epoch": 8.505494505494505, "grad_norm": 1.0703125, "learning_rate": 9.025679542161361e-05, "loss": 0.121, "step": 11610 }, { "epoch": 8.506227106227106, "grad_norm": 0.85546875, "learning_rate": 9.025475406075689e-05, "loss": 0.1368, "step": 11611 }, { "epoch": 8.506959706959707, "grad_norm": 0.7421875, "learning_rate": 9.025271250942208e-05, "loss": 0.1127, "step": 11612 }, { "epoch": 8.507692307692308, "grad_norm": 0.75, "learning_rate": 9.025067076761898e-05, "loss": 0.1183, "step": 11613 }, { "epoch": 8.508424908424908, "grad_norm": 0.95703125, "learning_rate": 9.024862883535737e-05, "loss": 0.1489, "step": 11614 }, { "epoch": 8.50915750915751, "grad_norm": 0.875, "learning_rate": 9.024658671264702e-05, "loss": 0.1291, "step": 11615 }, { "epoch": 8.50989010989011, "grad_norm": 0.91796875, "learning_rate": 9.024454439949772e-05, "loss": 0.1143, "step": 11616 }, { "epoch": 8.51062271062271, "grad_norm": 1.171875, "learning_rate": 9.024250189591927e-05, "loss": 0.1419, "step": 11617 }, { "epoch": 8.511355311355311, "grad_norm": 0.625, "learning_rate": 9.024045920192145e-05, "loss": 0.128, "step": 11618 }, { "epoch": 8.512087912087912, "grad_norm": 1.078125, "learning_rate": 9.023841631751404e-05, "loss": 0.1396, "step": 11619 }, { "epoch": 8.512820512820513, "grad_norm": 1.1171875, "learning_rate": 9.023637324270682e-05, "loss": 0.166, "step": 11620 }, { "epoch": 8.513553113553114, "grad_norm": 0.8828125, "learning_rate": 9.023432997750961e-05, "loss": 0.1347, "step": 11621 }, { "epoch": 8.514285714285714, "grad_norm": 0.9765625, "learning_rate": 9.023228652193217e-05, "loss": 0.1229, "step": 11622 }, { "epoch": 8.515018315018315, "grad_norm": 1.125, "learning_rate": 9.023024287598431e-05, "loss": 0.137, "step": 11623 }, { "epoch": 8.515750915750916, "grad_norm": 1.453125, "learning_rate": 9.02281990396758e-05, "loss": 0.1391, "step": 11624 }, { "epoch": 8.516483516483516, "grad_norm": 1.140625, "learning_rate": 9.022615501301647e-05, "loss": 0.1395, "step": 11625 }, { "epoch": 8.517216117216117, "grad_norm": 1.3671875, "learning_rate": 9.022411079601609e-05, "loss": 0.1584, "step": 11626 }, { "epoch": 8.517948717948718, "grad_norm": 1.1171875, "learning_rate": 9.022206638868442e-05, "loss": 0.1477, "step": 11627 }, { "epoch": 8.518681318681319, "grad_norm": 1.390625, "learning_rate": 9.022002179103131e-05, "loss": 0.1349, "step": 11628 }, { "epoch": 8.51941391941392, "grad_norm": 1.7265625, "learning_rate": 9.021797700306655e-05, "loss": 0.1387, "step": 11629 }, { "epoch": 8.52014652014652, "grad_norm": 1.6640625, "learning_rate": 9.02159320247999e-05, "loss": 0.1251, "step": 11630 }, { "epoch": 8.52087912087912, "grad_norm": 1.4609375, "learning_rate": 9.02138868562412e-05, "loss": 0.1946, "step": 11631 }, { "epoch": 8.521611721611722, "grad_norm": 0.87109375, "learning_rate": 9.021184149740022e-05, "loss": 0.1737, "step": 11632 }, { "epoch": 8.522344322344322, "grad_norm": 0.90625, "learning_rate": 9.020979594828676e-05, "loss": 0.1396, "step": 11633 }, { "epoch": 8.523076923076923, "grad_norm": 1.3984375, "learning_rate": 9.020775020891067e-05, "loss": 0.1392, "step": 11634 }, { "epoch": 8.523809523809524, "grad_norm": 0.92578125, "learning_rate": 9.020570427928168e-05, "loss": 0.1248, "step": 11635 }, { "epoch": 8.524542124542124, "grad_norm": 1.0078125, "learning_rate": 9.020365815940964e-05, "loss": 0.1358, "step": 11636 }, { "epoch": 8.525274725274725, "grad_norm": 0.85546875, "learning_rate": 9.020161184930433e-05, "loss": 0.1284, "step": 11637 }, { "epoch": 8.526007326007326, "grad_norm": 0.91796875, "learning_rate": 9.019956534897558e-05, "loss": 0.1361, "step": 11638 }, { "epoch": 8.526739926739927, "grad_norm": 1.0234375, "learning_rate": 9.019751865843317e-05, "loss": 0.1281, "step": 11639 }, { "epoch": 8.527472527472527, "grad_norm": 1.0078125, "learning_rate": 9.019547177768692e-05, "loss": 0.1348, "step": 11640 }, { "epoch": 8.528205128205128, "grad_norm": 0.859375, "learning_rate": 9.019342470674663e-05, "loss": 0.1392, "step": 11641 }, { "epoch": 8.528937728937729, "grad_norm": 1.1484375, "learning_rate": 9.019137744562211e-05, "loss": 0.1384, "step": 11642 }, { "epoch": 8.52967032967033, "grad_norm": 1.28125, "learning_rate": 9.018932999432317e-05, "loss": 0.1515, "step": 11643 }, { "epoch": 8.53040293040293, "grad_norm": 1.2890625, "learning_rate": 9.018728235285962e-05, "loss": 0.1556, "step": 11644 }, { "epoch": 8.531135531135531, "grad_norm": 1.0546875, "learning_rate": 9.018523452124127e-05, "loss": 0.1602, "step": 11645 }, { "epoch": 8.531868131868132, "grad_norm": 0.76953125, "learning_rate": 9.018318649947794e-05, "loss": 0.1412, "step": 11646 }, { "epoch": 8.532600732600732, "grad_norm": 0.98046875, "learning_rate": 9.018113828757944e-05, "loss": 0.1427, "step": 11647 }, { "epoch": 8.533333333333333, "grad_norm": 0.9921875, "learning_rate": 9.017908988555557e-05, "loss": 0.1262, "step": 11648 }, { "epoch": 8.534065934065934, "grad_norm": 1.203125, "learning_rate": 9.017704129341617e-05, "loss": 0.1443, "step": 11649 }, { "epoch": 8.534798534798535, "grad_norm": 0.78515625, "learning_rate": 9.017499251117104e-05, "loss": 0.1389, "step": 11650 }, { "epoch": 8.535531135531135, "grad_norm": 1.234375, "learning_rate": 9.017294353882997e-05, "loss": 0.14, "step": 11651 }, { "epoch": 8.536263736263736, "grad_norm": 1.0, "learning_rate": 9.017089437640283e-05, "loss": 0.1142, "step": 11652 }, { "epoch": 8.536996336996337, "grad_norm": 1.25, "learning_rate": 9.01688450238994e-05, "loss": 0.1168, "step": 11653 }, { "epoch": 8.537728937728938, "grad_norm": 1.203125, "learning_rate": 9.01667954813295e-05, "loss": 0.1189, "step": 11654 }, { "epoch": 8.538461538461538, "grad_norm": 0.7890625, "learning_rate": 9.0164745748703e-05, "loss": 0.1486, "step": 11655 }, { "epoch": 8.539194139194139, "grad_norm": 0.72265625, "learning_rate": 9.016269582602964e-05, "loss": 0.1063, "step": 11656 }, { "epoch": 8.53992673992674, "grad_norm": 0.7421875, "learning_rate": 9.01606457133193e-05, "loss": 0.1272, "step": 11657 }, { "epoch": 8.54065934065934, "grad_norm": 0.90234375, "learning_rate": 9.015859541058179e-05, "loss": 0.1238, "step": 11658 }, { "epoch": 8.541391941391941, "grad_norm": 0.74609375, "learning_rate": 9.015654491782691e-05, "loss": 0.125, "step": 11659 }, { "epoch": 8.542124542124542, "grad_norm": 1.359375, "learning_rate": 9.015449423506454e-05, "loss": 0.1627, "step": 11660 }, { "epoch": 8.542857142857143, "grad_norm": 2.0, "learning_rate": 9.015244336230446e-05, "loss": 0.152, "step": 11661 }, { "epoch": 8.543589743589743, "grad_norm": 0.380859375, "learning_rate": 9.015039229955648e-05, "loss": 0.0996, "step": 11662 }, { "epoch": 8.544322344322344, "grad_norm": 0.875, "learning_rate": 9.014834104683048e-05, "loss": 0.1637, "step": 11663 }, { "epoch": 8.545054945054945, "grad_norm": 1.25, "learning_rate": 9.014628960413626e-05, "loss": 0.1507, "step": 11664 }, { "epoch": 8.545787545787546, "grad_norm": 0.6875, "learning_rate": 9.014423797148365e-05, "loss": 0.1237, "step": 11665 }, { "epoch": 8.546520146520146, "grad_norm": 1.40625, "learning_rate": 9.014218614888247e-05, "loss": 0.1302, "step": 11666 }, { "epoch": 8.547252747252747, "grad_norm": 1.15625, "learning_rate": 9.014013413634256e-05, "loss": 0.1327, "step": 11667 }, { "epoch": 8.547985347985348, "grad_norm": 1.2578125, "learning_rate": 9.013808193387377e-05, "loss": 0.142, "step": 11668 }, { "epoch": 8.548717948717949, "grad_norm": 1.609375, "learning_rate": 9.013602954148592e-05, "loss": 0.1425, "step": 11669 }, { "epoch": 8.54945054945055, "grad_norm": 1.3984375, "learning_rate": 9.013397695918884e-05, "loss": 0.1586, "step": 11670 }, { "epoch": 8.55018315018315, "grad_norm": 0.953125, "learning_rate": 9.013192418699235e-05, "loss": 0.1511, "step": 11671 }, { "epoch": 8.55091575091575, "grad_norm": 1.1953125, "learning_rate": 9.012987122490631e-05, "loss": 0.1597, "step": 11672 }, { "epoch": 8.551648351648351, "grad_norm": 1.1953125, "learning_rate": 9.012781807294055e-05, "loss": 0.1242, "step": 11673 }, { "epoch": 8.552380952380952, "grad_norm": 0.94921875, "learning_rate": 9.01257647311049e-05, "loss": 0.1265, "step": 11674 }, { "epoch": 8.553113553113553, "grad_norm": 1.203125, "learning_rate": 9.01237111994092e-05, "loss": 0.1162, "step": 11675 }, { "epoch": 8.553846153846154, "grad_norm": 1.171875, "learning_rate": 9.01216574778633e-05, "loss": 0.1511, "step": 11676 }, { "epoch": 8.554578754578754, "grad_norm": 1.9921875, "learning_rate": 9.011960356647703e-05, "loss": 0.159, "step": 11677 }, { "epoch": 8.555311355311355, "grad_norm": 1.046875, "learning_rate": 9.011754946526023e-05, "loss": 0.1238, "step": 11678 }, { "epoch": 8.556043956043956, "grad_norm": 0.9453125, "learning_rate": 9.011549517422274e-05, "loss": 0.1285, "step": 11679 }, { "epoch": 8.556776556776557, "grad_norm": 0.69140625, "learning_rate": 9.011344069337443e-05, "loss": 0.1265, "step": 11680 }, { "epoch": 8.557509157509157, "grad_norm": 1.4140625, "learning_rate": 9.01113860227251e-05, "loss": 0.1321, "step": 11681 }, { "epoch": 8.558241758241758, "grad_norm": 1.1328125, "learning_rate": 9.010933116228462e-05, "loss": 0.1538, "step": 11682 }, { "epoch": 8.558974358974359, "grad_norm": 0.63671875, "learning_rate": 9.010727611206283e-05, "loss": 0.1095, "step": 11683 }, { "epoch": 8.55970695970696, "grad_norm": 1.1796875, "learning_rate": 9.010522087206958e-05, "loss": 0.0983, "step": 11684 }, { "epoch": 8.56043956043956, "grad_norm": 0.9375, "learning_rate": 9.010316544231472e-05, "loss": 0.1247, "step": 11685 }, { "epoch": 8.561172161172161, "grad_norm": 1.484375, "learning_rate": 9.01011098228081e-05, "loss": 0.1586, "step": 11686 }, { "epoch": 8.561904761904762, "grad_norm": 0.8984375, "learning_rate": 9.009905401355954e-05, "loss": 0.1286, "step": 11687 }, { "epoch": 8.562637362637362, "grad_norm": 0.71875, "learning_rate": 9.009699801457892e-05, "loss": 0.1525, "step": 11688 }, { "epoch": 8.563369963369963, "grad_norm": 1.5234375, "learning_rate": 9.00949418258761e-05, "loss": 0.1421, "step": 11689 }, { "epoch": 8.564102564102564, "grad_norm": 1.1953125, "learning_rate": 9.009288544746091e-05, "loss": 0.1275, "step": 11690 }, { "epoch": 8.564835164835165, "grad_norm": 0.9140625, "learning_rate": 9.009082887934319e-05, "loss": 0.1414, "step": 11691 }, { "epoch": 8.565567765567765, "grad_norm": 1.1640625, "learning_rate": 9.008877212153282e-05, "loss": 0.1542, "step": 11692 }, { "epoch": 8.566300366300366, "grad_norm": 0.93359375, "learning_rate": 9.008671517403966e-05, "loss": 0.1301, "step": 11693 }, { "epoch": 8.567032967032967, "grad_norm": 1.3984375, "learning_rate": 9.008465803687355e-05, "loss": 0.1242, "step": 11694 }, { "epoch": 8.567765567765568, "grad_norm": 1.28125, "learning_rate": 9.008260071004435e-05, "loss": 0.1403, "step": 11695 }, { "epoch": 8.568498168498168, "grad_norm": 1.125, "learning_rate": 9.00805431935619e-05, "loss": 0.139, "step": 11696 }, { "epoch": 8.569230769230769, "grad_norm": 1.0859375, "learning_rate": 9.007848548743609e-05, "loss": 0.1456, "step": 11697 }, { "epoch": 8.56996336996337, "grad_norm": 1.171875, "learning_rate": 9.007642759167674e-05, "loss": 0.1634, "step": 11698 }, { "epoch": 8.57069597069597, "grad_norm": 0.9140625, "learning_rate": 9.007436950629376e-05, "loss": 0.1412, "step": 11699 }, { "epoch": 8.571428571428571, "grad_norm": 1.546875, "learning_rate": 9.007231123129697e-05, "loss": 0.187, "step": 11700 }, { "epoch": 8.572161172161172, "grad_norm": 1.2265625, "learning_rate": 9.007025276669624e-05, "loss": 0.147, "step": 11701 }, { "epoch": 8.572893772893773, "grad_norm": 1.8203125, "learning_rate": 9.006819411250146e-05, "loss": 0.1871, "step": 11702 }, { "epoch": 8.573626373626373, "grad_norm": 1.4453125, "learning_rate": 9.006613526872245e-05, "loss": 0.1658, "step": 11703 }, { "epoch": 8.574358974358974, "grad_norm": 1.265625, "learning_rate": 9.006407623536912e-05, "loss": 0.1341, "step": 11704 }, { "epoch": 8.575091575091575, "grad_norm": 1.0, "learning_rate": 9.006201701245128e-05, "loss": 0.1368, "step": 11705 }, { "epoch": 8.575824175824176, "grad_norm": 1.1953125, "learning_rate": 9.005995759997887e-05, "loss": 0.1334, "step": 11706 }, { "epoch": 8.576556776556776, "grad_norm": 1.3984375, "learning_rate": 9.005789799796168e-05, "loss": 0.1412, "step": 11707 }, { "epoch": 8.577289377289377, "grad_norm": 1.0234375, "learning_rate": 9.005583820640965e-05, "loss": 0.1342, "step": 11708 }, { "epoch": 8.578021978021978, "grad_norm": 0.96484375, "learning_rate": 9.005377822533259e-05, "loss": 0.096, "step": 11709 }, { "epoch": 8.578754578754578, "grad_norm": 0.890625, "learning_rate": 9.005171805474039e-05, "loss": 0.1303, "step": 11710 }, { "epoch": 8.57948717948718, "grad_norm": 1.34375, "learning_rate": 9.004965769464293e-05, "loss": 0.1161, "step": 11711 }, { "epoch": 8.58021978021978, "grad_norm": 0.53515625, "learning_rate": 9.004759714505008e-05, "loss": 0.1088, "step": 11712 }, { "epoch": 8.58095238095238, "grad_norm": 0.87109375, "learning_rate": 9.004553640597172e-05, "loss": 0.1124, "step": 11713 }, { "epoch": 8.581684981684981, "grad_norm": 1.09375, "learning_rate": 9.00434754774177e-05, "loss": 0.1201, "step": 11714 }, { "epoch": 8.582417582417582, "grad_norm": 0.97265625, "learning_rate": 9.004141435939791e-05, "loss": 0.1482, "step": 11715 }, { "epoch": 8.583150183150183, "grad_norm": 1.71875, "learning_rate": 9.003935305192222e-05, "loss": 0.1193, "step": 11716 }, { "epoch": 8.583882783882784, "grad_norm": 1.2265625, "learning_rate": 9.003729155500051e-05, "loss": 0.1467, "step": 11717 }, { "epoch": 8.584615384615384, "grad_norm": 1.1015625, "learning_rate": 9.003522986864266e-05, "loss": 0.1415, "step": 11718 }, { "epoch": 8.585347985347985, "grad_norm": 0.5234375, "learning_rate": 9.003316799285855e-05, "loss": 0.1406, "step": 11719 }, { "epoch": 8.586080586080586, "grad_norm": 0.5546875, "learning_rate": 9.003110592765805e-05, "loss": 0.1097, "step": 11720 }, { "epoch": 8.586813186813187, "grad_norm": 0.87890625, "learning_rate": 9.002904367305104e-05, "loss": 0.1138, "step": 11721 }, { "epoch": 8.587545787545787, "grad_norm": 1.2265625, "learning_rate": 9.002698122904742e-05, "loss": 0.1309, "step": 11722 }, { "epoch": 8.588278388278388, "grad_norm": 0.90625, "learning_rate": 9.002491859565704e-05, "loss": 0.1198, "step": 11723 }, { "epoch": 8.589010989010989, "grad_norm": 1.34375, "learning_rate": 9.002285577288981e-05, "loss": 0.1302, "step": 11724 }, { "epoch": 8.58974358974359, "grad_norm": 0.921875, "learning_rate": 9.00207927607556e-05, "loss": 0.1278, "step": 11725 }, { "epoch": 8.59047619047619, "grad_norm": 1.515625, "learning_rate": 9.00187295592643e-05, "loss": 0.1447, "step": 11726 }, { "epoch": 8.59120879120879, "grad_norm": 1.5, "learning_rate": 9.001666616842579e-05, "loss": 0.1183, "step": 11727 }, { "epoch": 8.591941391941392, "grad_norm": 1.0390625, "learning_rate": 9.001460258824995e-05, "loss": 0.1206, "step": 11728 }, { "epoch": 8.592673992673992, "grad_norm": 1.4453125, "learning_rate": 9.00125388187467e-05, "loss": 0.1721, "step": 11729 }, { "epoch": 8.593406593406593, "grad_norm": 0.83984375, "learning_rate": 9.001047485992589e-05, "loss": 0.131, "step": 11730 }, { "epoch": 8.594139194139194, "grad_norm": 1.046875, "learning_rate": 9.000841071179743e-05, "loss": 0.1289, "step": 11731 }, { "epoch": 8.594871794871795, "grad_norm": 1.34375, "learning_rate": 9.000634637437118e-05, "loss": 0.1399, "step": 11732 }, { "epoch": 8.595604395604395, "grad_norm": 0.73046875, "learning_rate": 9.000428184765708e-05, "loss": 0.1253, "step": 11733 }, { "epoch": 8.596336996336996, "grad_norm": 0.7109375, "learning_rate": 9.000221713166501e-05, "loss": 0.1188, "step": 11734 }, { "epoch": 8.597069597069597, "grad_norm": 1.1015625, "learning_rate": 9.000015222640484e-05, "loss": 0.1271, "step": 11735 }, { "epoch": 8.597802197802197, "grad_norm": 0.78515625, "learning_rate": 8.999808713188646e-05, "loss": 0.1444, "step": 11736 }, { "epoch": 8.598534798534798, "grad_norm": 0.953125, "learning_rate": 8.99960218481198e-05, "loss": 0.1304, "step": 11737 }, { "epoch": 8.599267399267399, "grad_norm": 1.03125, "learning_rate": 8.999395637511471e-05, "loss": 0.1304, "step": 11738 }, { "epoch": 8.6, "grad_norm": 1.25, "learning_rate": 8.999189071288114e-05, "loss": 0.0875, "step": 11739 }, { "epoch": 8.6007326007326, "grad_norm": 1.2578125, "learning_rate": 8.998982486142891e-05, "loss": 0.1468, "step": 11740 }, { "epoch": 8.601465201465201, "grad_norm": 1.1015625, "learning_rate": 8.998775882076801e-05, "loss": 0.1482, "step": 11741 }, { "epoch": 8.602197802197802, "grad_norm": 1.453125, "learning_rate": 8.998569259090828e-05, "loss": 0.1652, "step": 11742 }, { "epoch": 8.602930402930403, "grad_norm": 1.1328125, "learning_rate": 8.998362617185963e-05, "loss": 0.1445, "step": 11743 }, { "epoch": 8.603663003663003, "grad_norm": 0.94140625, "learning_rate": 8.998155956363195e-05, "loss": 0.1392, "step": 11744 }, { "epoch": 8.604395604395604, "grad_norm": 1.078125, "learning_rate": 8.997949276623518e-05, "loss": 0.1143, "step": 11745 }, { "epoch": 8.605128205128205, "grad_norm": 0.9140625, "learning_rate": 8.99774257796792e-05, "loss": 0.1116, "step": 11746 }, { "epoch": 8.605860805860805, "grad_norm": 0.8984375, "learning_rate": 8.997535860397391e-05, "loss": 0.1247, "step": 11747 }, { "epoch": 8.606593406593406, "grad_norm": 1.265625, "learning_rate": 8.997329123912921e-05, "loss": 0.1197, "step": 11748 }, { "epoch": 8.607326007326007, "grad_norm": 1.2578125, "learning_rate": 8.997122368515503e-05, "loss": 0.1545, "step": 11749 }, { "epoch": 8.608058608058608, "grad_norm": 0.57421875, "learning_rate": 8.996915594206124e-05, "loss": 0.0994, "step": 11750 }, { "epoch": 8.608791208791208, "grad_norm": 1.359375, "learning_rate": 8.99670880098578e-05, "loss": 0.1407, "step": 11751 }, { "epoch": 8.60952380952381, "grad_norm": 1.28125, "learning_rate": 8.996501988855455e-05, "loss": 0.1653, "step": 11752 }, { "epoch": 8.61025641025641, "grad_norm": 1.15625, "learning_rate": 8.996295157816143e-05, "loss": 0.1258, "step": 11753 }, { "epoch": 8.61098901098901, "grad_norm": 0.984375, "learning_rate": 8.996088307868838e-05, "loss": 0.1403, "step": 11754 }, { "epoch": 8.611721611721611, "grad_norm": 1.015625, "learning_rate": 8.995881439014528e-05, "loss": 0.1671, "step": 11755 }, { "epoch": 8.612454212454212, "grad_norm": 0.94921875, "learning_rate": 8.995674551254204e-05, "loss": 0.1164, "step": 11756 }, { "epoch": 8.613186813186813, "grad_norm": 1.1015625, "learning_rate": 8.995467644588857e-05, "loss": 0.1751, "step": 11757 }, { "epoch": 8.613919413919414, "grad_norm": 1.3359375, "learning_rate": 8.995260719019482e-05, "loss": 0.1461, "step": 11758 }, { "epoch": 8.614652014652014, "grad_norm": 1.2265625, "learning_rate": 8.995053774547066e-05, "loss": 0.1411, "step": 11759 }, { "epoch": 8.615384615384615, "grad_norm": 1.1796875, "learning_rate": 8.994846811172603e-05, "loss": 0.1614, "step": 11760 }, { "epoch": 8.616117216117216, "grad_norm": 1.3125, "learning_rate": 8.994639828897084e-05, "loss": 0.1106, "step": 11761 }, { "epoch": 8.616849816849816, "grad_norm": 1.0625, "learning_rate": 8.994432827721499e-05, "loss": 0.1244, "step": 11762 }, { "epoch": 8.617582417582417, "grad_norm": 0.81640625, "learning_rate": 8.994225807646841e-05, "loss": 0.1131, "step": 11763 }, { "epoch": 8.618315018315018, "grad_norm": 0.9375, "learning_rate": 8.994018768674104e-05, "loss": 0.1462, "step": 11764 }, { "epoch": 8.619047619047619, "grad_norm": 0.66015625, "learning_rate": 8.993811710804279e-05, "loss": 0.1131, "step": 11765 }, { "epoch": 8.61978021978022, "grad_norm": 1.2109375, "learning_rate": 8.993604634038356e-05, "loss": 0.1435, "step": 11766 }, { "epoch": 8.62051282051282, "grad_norm": 0.67578125, "learning_rate": 8.99339753837733e-05, "loss": 0.1308, "step": 11767 }, { "epoch": 8.62124542124542, "grad_norm": 1.09375, "learning_rate": 8.993190423822191e-05, "loss": 0.1375, "step": 11768 }, { "epoch": 8.621978021978022, "grad_norm": 1.1640625, "learning_rate": 8.992983290373932e-05, "loss": 0.1154, "step": 11769 }, { "epoch": 8.622710622710622, "grad_norm": 0.490234375, "learning_rate": 8.992776138033547e-05, "loss": 0.1293, "step": 11770 }, { "epoch": 8.623443223443223, "grad_norm": 0.8671875, "learning_rate": 8.992568966802025e-05, "loss": 0.1338, "step": 11771 }, { "epoch": 8.624175824175824, "grad_norm": 0.7109375, "learning_rate": 8.992361776680363e-05, "loss": 0.1009, "step": 11772 }, { "epoch": 8.624908424908424, "grad_norm": 1.046875, "learning_rate": 8.99215456766955e-05, "loss": 0.1168, "step": 11773 }, { "epoch": 8.625641025641025, "grad_norm": 1.2265625, "learning_rate": 8.99194733977058e-05, "loss": 0.1461, "step": 11774 }, { "epoch": 8.626373626373626, "grad_norm": 1.2109375, "learning_rate": 8.991740092984447e-05, "loss": 0.1547, "step": 11775 }, { "epoch": 8.627106227106227, "grad_norm": 0.75390625, "learning_rate": 8.991532827312144e-05, "loss": 0.1214, "step": 11776 }, { "epoch": 8.627838827838827, "grad_norm": 1.1875, "learning_rate": 8.991325542754662e-05, "loss": 0.1216, "step": 11777 }, { "epoch": 8.628571428571428, "grad_norm": 0.84375, "learning_rate": 8.991118239312996e-05, "loss": 0.1355, "step": 11778 }, { "epoch": 8.629304029304029, "grad_norm": 0.53125, "learning_rate": 8.990910916988138e-05, "loss": 0.1113, "step": 11779 }, { "epoch": 8.63003663003663, "grad_norm": 0.75390625, "learning_rate": 8.990703575781085e-05, "loss": 0.1274, "step": 11780 }, { "epoch": 8.63076923076923, "grad_norm": 0.9921875, "learning_rate": 8.990496215692824e-05, "loss": 0.13, "step": 11781 }, { "epoch": 8.631501831501831, "grad_norm": 1.0, "learning_rate": 8.990288836724353e-05, "loss": 0.1371, "step": 11782 }, { "epoch": 8.632234432234432, "grad_norm": 0.5390625, "learning_rate": 8.990081438876666e-05, "loss": 0.1158, "step": 11783 }, { "epoch": 8.632967032967032, "grad_norm": 0.81640625, "learning_rate": 8.989874022150756e-05, "loss": 0.1143, "step": 11784 }, { "epoch": 8.633699633699633, "grad_norm": 1.078125, "learning_rate": 8.989666586547614e-05, "loss": 0.1847, "step": 11785 }, { "epoch": 8.634432234432234, "grad_norm": 1.3359375, "learning_rate": 8.989459132068236e-05, "loss": 0.1284, "step": 11786 }, { "epoch": 8.635164835164835, "grad_norm": 0.77734375, "learning_rate": 8.989251658713617e-05, "loss": 0.095, "step": 11787 }, { "epoch": 8.635897435897435, "grad_norm": 1.234375, "learning_rate": 8.989044166484748e-05, "loss": 0.1427, "step": 11788 }, { "epoch": 8.636630036630036, "grad_norm": 0.6953125, "learning_rate": 8.988836655382629e-05, "loss": 0.1275, "step": 11789 }, { "epoch": 8.637362637362637, "grad_norm": 1.3125, "learning_rate": 8.988629125408248e-05, "loss": 0.1532, "step": 11790 }, { "epoch": 8.638095238095238, "grad_norm": 1.0234375, "learning_rate": 8.988421576562604e-05, "loss": 0.1277, "step": 11791 }, { "epoch": 8.638827838827838, "grad_norm": 0.87109375, "learning_rate": 8.988214008846689e-05, "loss": 0.1144, "step": 11792 }, { "epoch": 8.639560439560439, "grad_norm": 0.84765625, "learning_rate": 8.988006422261495e-05, "loss": 0.1374, "step": 11793 }, { "epoch": 8.64029304029304, "grad_norm": 0.91015625, "learning_rate": 8.987798816808022e-05, "loss": 0.1346, "step": 11794 }, { "epoch": 8.64102564102564, "grad_norm": 0.9140625, "learning_rate": 8.98759119248726e-05, "loss": 0.1127, "step": 11795 }, { "epoch": 8.641758241758241, "grad_norm": 1.46875, "learning_rate": 8.987383549300208e-05, "loss": 0.1534, "step": 11796 }, { "epoch": 8.642490842490842, "grad_norm": 0.84375, "learning_rate": 8.987175887247859e-05, "loss": 0.1459, "step": 11797 }, { "epoch": 8.643223443223443, "grad_norm": 0.984375, "learning_rate": 8.986968206331207e-05, "loss": 0.1327, "step": 11798 }, { "epoch": 8.643956043956043, "grad_norm": 0.921875, "learning_rate": 8.986760506551249e-05, "loss": 0.145, "step": 11799 }, { "epoch": 8.644688644688644, "grad_norm": 1.28125, "learning_rate": 8.986552787908977e-05, "loss": 0.1212, "step": 11800 }, { "epoch": 8.645421245421245, "grad_norm": 0.6015625, "learning_rate": 8.98634505040539e-05, "loss": 0.1259, "step": 11801 }, { "epoch": 8.646153846153846, "grad_norm": 1.4296875, "learning_rate": 8.986137294041481e-05, "loss": 0.1609, "step": 11802 }, { "epoch": 8.646886446886446, "grad_norm": 0.921875, "learning_rate": 8.985929518818246e-05, "loss": 0.132, "step": 11803 }, { "epoch": 8.647619047619047, "grad_norm": 1.125, "learning_rate": 8.98572172473668e-05, "loss": 0.1417, "step": 11804 }, { "epoch": 8.648351648351648, "grad_norm": 0.99609375, "learning_rate": 8.98551391179778e-05, "loss": 0.1181, "step": 11805 }, { "epoch": 8.649084249084249, "grad_norm": 0.84765625, "learning_rate": 8.985306080002541e-05, "loss": 0.113, "step": 11806 }, { "epoch": 8.64981684981685, "grad_norm": 1.4296875, "learning_rate": 8.98509822935196e-05, "loss": 0.1933, "step": 11807 }, { "epoch": 8.65054945054945, "grad_norm": 0.83984375, "learning_rate": 8.98489035984703e-05, "loss": 0.15, "step": 11808 }, { "epoch": 8.65128205128205, "grad_norm": 0.72265625, "learning_rate": 8.984682471488748e-05, "loss": 0.1179, "step": 11809 }, { "epoch": 8.652014652014651, "grad_norm": 1.03125, "learning_rate": 8.984474564278111e-05, "loss": 0.1531, "step": 11810 }, { "epoch": 8.652747252747252, "grad_norm": 0.5703125, "learning_rate": 8.984266638216115e-05, "loss": 0.1083, "step": 11811 }, { "epoch": 8.653479853479853, "grad_norm": 1.2734375, "learning_rate": 8.984058693303757e-05, "loss": 0.1348, "step": 11812 }, { "epoch": 8.654212454212454, "grad_norm": 1.0625, "learning_rate": 8.983850729542031e-05, "loss": 0.1488, "step": 11813 }, { "epoch": 8.654945054945054, "grad_norm": 0.9296875, "learning_rate": 8.983642746931935e-05, "loss": 0.1039, "step": 11814 }, { "epoch": 8.655677655677655, "grad_norm": 1.4296875, "learning_rate": 8.983434745474464e-05, "loss": 0.1699, "step": 11815 }, { "epoch": 8.656410256410256, "grad_norm": 1.3984375, "learning_rate": 8.983226725170619e-05, "loss": 0.1708, "step": 11816 }, { "epoch": 8.657142857142857, "grad_norm": 1.015625, "learning_rate": 8.983018686021392e-05, "loss": 0.1363, "step": 11817 }, { "epoch": 8.657875457875457, "grad_norm": 0.75390625, "learning_rate": 8.982810628027781e-05, "loss": 0.1332, "step": 11818 }, { "epoch": 8.658608058608058, "grad_norm": 1.078125, "learning_rate": 8.982602551190783e-05, "loss": 0.1407, "step": 11819 }, { "epoch": 8.659340659340659, "grad_norm": 0.984375, "learning_rate": 8.982394455511394e-05, "loss": 0.1291, "step": 11820 }, { "epoch": 8.66007326007326, "grad_norm": 1.375, "learning_rate": 8.982186340990616e-05, "loss": 0.1689, "step": 11821 }, { "epoch": 8.66080586080586, "grad_norm": 1.109375, "learning_rate": 8.981978207629439e-05, "loss": 0.1501, "step": 11822 }, { "epoch": 8.661538461538461, "grad_norm": 1.2734375, "learning_rate": 8.981770055428865e-05, "loss": 0.1622, "step": 11823 }, { "epoch": 8.662271062271062, "grad_norm": 1.0625, "learning_rate": 8.98156188438989e-05, "loss": 0.1358, "step": 11824 }, { "epoch": 8.663003663003662, "grad_norm": 0.92578125, "learning_rate": 8.981353694513513e-05, "loss": 0.1226, "step": 11825 }, { "epoch": 8.663736263736263, "grad_norm": 0.9765625, "learning_rate": 8.981145485800725e-05, "loss": 0.1213, "step": 11826 }, { "epoch": 8.664468864468864, "grad_norm": 1.8203125, "learning_rate": 8.980937258252533e-05, "loss": 0.1562, "step": 11827 }, { "epoch": 8.665201465201465, "grad_norm": 0.828125, "learning_rate": 8.980729011869929e-05, "loss": 0.1193, "step": 11828 }, { "epoch": 8.665934065934065, "grad_norm": 1.453125, "learning_rate": 8.98052074665391e-05, "loss": 0.1379, "step": 11829 }, { "epoch": 8.666666666666666, "grad_norm": 1.28125, "learning_rate": 8.980312462605478e-05, "loss": 0.1425, "step": 11830 }, { "epoch": 8.667399267399267, "grad_norm": 0.70703125, "learning_rate": 8.980104159725627e-05, "loss": 0.1441, "step": 11831 }, { "epoch": 8.668131868131868, "grad_norm": 0.52734375, "learning_rate": 8.97989583801536e-05, "loss": 0.1129, "step": 11832 }, { "epoch": 8.668864468864468, "grad_norm": 0.609375, "learning_rate": 8.979687497475669e-05, "loss": 0.1217, "step": 11833 }, { "epoch": 8.669597069597069, "grad_norm": 1.2578125, "learning_rate": 8.979479138107555e-05, "loss": 0.1206, "step": 11834 }, { "epoch": 8.67032967032967, "grad_norm": 0.73828125, "learning_rate": 8.979270759912019e-05, "loss": 0.1187, "step": 11835 }, { "epoch": 8.67106227106227, "grad_norm": 0.625, "learning_rate": 8.979062362890056e-05, "loss": 0.1435, "step": 11836 }, { "epoch": 8.671794871794871, "grad_norm": 1.109375, "learning_rate": 8.978853947042664e-05, "loss": 0.1638, "step": 11837 }, { "epoch": 8.672527472527472, "grad_norm": 0.92578125, "learning_rate": 8.978645512370845e-05, "loss": 0.1254, "step": 11838 }, { "epoch": 8.673260073260073, "grad_norm": 0.6484375, "learning_rate": 8.978437058875594e-05, "loss": 0.1247, "step": 11839 }, { "epoch": 8.673992673992673, "grad_norm": 1.015625, "learning_rate": 8.978228586557912e-05, "loss": 0.1191, "step": 11840 }, { "epoch": 8.674725274725274, "grad_norm": 1.0625, "learning_rate": 8.978020095418797e-05, "loss": 0.1456, "step": 11841 }, { "epoch": 8.675457875457875, "grad_norm": 0.66796875, "learning_rate": 8.977811585459249e-05, "loss": 0.1043, "step": 11842 }, { "epoch": 8.676190476190476, "grad_norm": 0.8671875, "learning_rate": 8.977603056680265e-05, "loss": 0.1577, "step": 11843 }, { "epoch": 8.676923076923076, "grad_norm": 0.73828125, "learning_rate": 8.977394509082846e-05, "loss": 0.1674, "step": 11844 }, { "epoch": 8.677655677655677, "grad_norm": 1.5703125, "learning_rate": 8.97718594266799e-05, "loss": 0.1404, "step": 11845 }, { "epoch": 8.678388278388278, "grad_norm": 1.0859375, "learning_rate": 8.976977357436697e-05, "loss": 0.1369, "step": 11846 }, { "epoch": 8.679120879120878, "grad_norm": 1.015625, "learning_rate": 8.976768753389968e-05, "loss": 0.1286, "step": 11847 }, { "epoch": 8.67985347985348, "grad_norm": 0.765625, "learning_rate": 8.976560130528801e-05, "loss": 0.146, "step": 11848 }, { "epoch": 8.68058608058608, "grad_norm": 0.8984375, "learning_rate": 8.976351488854195e-05, "loss": 0.1236, "step": 11849 }, { "epoch": 8.68131868131868, "grad_norm": 0.65625, "learning_rate": 8.976142828367148e-05, "loss": 0.1347, "step": 11850 }, { "epoch": 8.682051282051281, "grad_norm": 0.76171875, "learning_rate": 8.975934149068662e-05, "loss": 0.1045, "step": 11851 }, { "epoch": 8.682783882783882, "grad_norm": 0.99609375, "learning_rate": 8.975725450959739e-05, "loss": 0.1092, "step": 11852 }, { "epoch": 8.683516483516483, "grad_norm": 1.25, "learning_rate": 8.975516734041375e-05, "loss": 0.1558, "step": 11853 }, { "epoch": 8.684249084249084, "grad_norm": 0.59375, "learning_rate": 8.975307998314572e-05, "loss": 0.1259, "step": 11854 }, { "epoch": 8.684981684981684, "grad_norm": 1.0625, "learning_rate": 8.975099243780331e-05, "loss": 0.1644, "step": 11855 }, { "epoch": 8.685714285714285, "grad_norm": 1.3046875, "learning_rate": 8.97489047043965e-05, "loss": 0.1596, "step": 11856 }, { "epoch": 8.686446886446886, "grad_norm": 1.1484375, "learning_rate": 8.974681678293528e-05, "loss": 0.1297, "step": 11857 }, { "epoch": 8.687179487179487, "grad_norm": 1.5625, "learning_rate": 8.974472867342971e-05, "loss": 0.1783, "step": 11858 }, { "epoch": 8.687912087912087, "grad_norm": 1.0859375, "learning_rate": 8.974264037588976e-05, "loss": 0.1447, "step": 11859 }, { "epoch": 8.688644688644688, "grad_norm": 0.98046875, "learning_rate": 8.974055189032542e-05, "loss": 0.1201, "step": 11860 }, { "epoch": 8.689377289377289, "grad_norm": 0.79296875, "learning_rate": 8.973846321674673e-05, "loss": 0.1446, "step": 11861 }, { "epoch": 8.69010989010989, "grad_norm": 1.0859375, "learning_rate": 8.973637435516367e-05, "loss": 0.1506, "step": 11862 }, { "epoch": 8.69084249084249, "grad_norm": 1.2265625, "learning_rate": 8.973428530558626e-05, "loss": 0.1366, "step": 11863 }, { "epoch": 8.691575091575091, "grad_norm": 1.046875, "learning_rate": 8.973219606802451e-05, "loss": 0.1752, "step": 11864 }, { "epoch": 8.692307692307692, "grad_norm": 0.96875, "learning_rate": 8.973010664248844e-05, "loss": 0.126, "step": 11865 }, { "epoch": 8.693040293040292, "grad_norm": 1.3125, "learning_rate": 8.972801702898805e-05, "loss": 0.1231, "step": 11866 }, { "epoch": 8.693772893772893, "grad_norm": 0.95703125, "learning_rate": 8.972592722753335e-05, "loss": 0.1283, "step": 11867 }, { "epoch": 8.694505494505494, "grad_norm": 0.9296875, "learning_rate": 8.972383723813434e-05, "loss": 0.1257, "step": 11868 }, { "epoch": 8.695238095238095, "grad_norm": 1.34375, "learning_rate": 8.972174706080107e-05, "loss": 0.1492, "step": 11869 }, { "epoch": 8.695970695970695, "grad_norm": 1.296875, "learning_rate": 8.971965669554352e-05, "loss": 0.1448, "step": 11870 }, { "epoch": 8.696703296703296, "grad_norm": 0.6640625, "learning_rate": 8.971756614237172e-05, "loss": 0.1205, "step": 11871 }, { "epoch": 8.697435897435897, "grad_norm": 0.91796875, "learning_rate": 8.97154754012957e-05, "loss": 0.1307, "step": 11872 }, { "epoch": 8.698168498168497, "grad_norm": 1.7734375, "learning_rate": 8.971338447232546e-05, "loss": 0.1447, "step": 11873 }, { "epoch": 8.698901098901098, "grad_norm": 1.6796875, "learning_rate": 8.971129335547101e-05, "loss": 0.1172, "step": 11874 }, { "epoch": 8.699633699633699, "grad_norm": 1.234375, "learning_rate": 8.97092020507424e-05, "loss": 0.1578, "step": 11875 }, { "epoch": 8.7003663003663, "grad_norm": 0.7890625, "learning_rate": 8.97071105581496e-05, "loss": 0.1057, "step": 11876 }, { "epoch": 8.7010989010989, "grad_norm": 1.0859375, "learning_rate": 8.970501887770269e-05, "loss": 0.145, "step": 11877 }, { "epoch": 8.701831501831501, "grad_norm": 1.234375, "learning_rate": 8.970292700941165e-05, "loss": 0.1167, "step": 11878 }, { "epoch": 8.702564102564102, "grad_norm": 1.2421875, "learning_rate": 8.970083495328654e-05, "loss": 0.1206, "step": 11879 }, { "epoch": 8.703296703296703, "grad_norm": 1.546875, "learning_rate": 8.969874270933734e-05, "loss": 0.1472, "step": 11880 }, { "epoch": 8.704029304029303, "grad_norm": 0.8359375, "learning_rate": 8.96966502775741e-05, "loss": 0.1397, "step": 11881 }, { "epoch": 8.704761904761904, "grad_norm": 0.7109375, "learning_rate": 8.969455765800683e-05, "loss": 0.1375, "step": 11882 }, { "epoch": 8.705494505494505, "grad_norm": 1.8359375, "learning_rate": 8.969246485064559e-05, "loss": 0.1556, "step": 11883 }, { "epoch": 8.706227106227105, "grad_norm": 0.7890625, "learning_rate": 8.969037185550037e-05, "loss": 0.1283, "step": 11884 }, { "epoch": 8.706959706959706, "grad_norm": 0.9609375, "learning_rate": 8.968827867258121e-05, "loss": 0.1573, "step": 11885 }, { "epoch": 8.707692307692307, "grad_norm": 1.3515625, "learning_rate": 8.968618530189816e-05, "loss": 0.1626, "step": 11886 }, { "epoch": 8.708424908424908, "grad_norm": 1.015625, "learning_rate": 8.968409174346122e-05, "loss": 0.1303, "step": 11887 }, { "epoch": 8.709157509157508, "grad_norm": 0.9375, "learning_rate": 8.968199799728045e-05, "loss": 0.1599, "step": 11888 }, { "epoch": 8.70989010989011, "grad_norm": 0.62890625, "learning_rate": 8.967990406336584e-05, "loss": 0.1271, "step": 11889 }, { "epoch": 8.71062271062271, "grad_norm": 1.203125, "learning_rate": 8.967780994172746e-05, "loss": 0.1325, "step": 11890 }, { "epoch": 8.71135531135531, "grad_norm": 1.34375, "learning_rate": 8.967571563237534e-05, "loss": 0.14, "step": 11891 }, { "epoch": 8.712087912087911, "grad_norm": 1.625, "learning_rate": 8.967362113531948e-05, "loss": 0.155, "step": 11892 }, { "epoch": 8.712820512820512, "grad_norm": 1.2421875, "learning_rate": 8.967152645056996e-05, "loss": 0.1485, "step": 11893 }, { "epoch": 8.713553113553113, "grad_norm": 0.8515625, "learning_rate": 8.966943157813679e-05, "loss": 0.1622, "step": 11894 }, { "epoch": 8.714285714285714, "grad_norm": 1.125, "learning_rate": 8.966733651803003e-05, "loss": 0.1253, "step": 11895 }, { "epoch": 8.715018315018314, "grad_norm": 0.95703125, "learning_rate": 8.966524127025969e-05, "loss": 0.1361, "step": 11896 }, { "epoch": 8.715750915750915, "grad_norm": 0.875, "learning_rate": 8.966314583483582e-05, "loss": 0.1606, "step": 11897 }, { "epoch": 8.716483516483516, "grad_norm": 1.625, "learning_rate": 8.966105021176847e-05, "loss": 0.1525, "step": 11898 }, { "epoch": 8.717216117216116, "grad_norm": 1.0078125, "learning_rate": 8.965895440106766e-05, "loss": 0.1232, "step": 11899 }, { "epoch": 8.717948717948717, "grad_norm": 0.91796875, "learning_rate": 8.965685840274346e-05, "loss": 0.1853, "step": 11900 }, { "epoch": 8.718681318681318, "grad_norm": 1.0078125, "learning_rate": 8.965476221680589e-05, "loss": 0.1215, "step": 11901 }, { "epoch": 8.719413919413919, "grad_norm": 1.125, "learning_rate": 8.965266584326499e-05, "loss": 0.1339, "step": 11902 }, { "epoch": 8.72014652014652, "grad_norm": 1.1953125, "learning_rate": 8.965056928213082e-05, "loss": 0.1534, "step": 11903 }, { "epoch": 8.72087912087912, "grad_norm": 1.2578125, "learning_rate": 8.964847253341343e-05, "loss": 0.1533, "step": 11904 }, { "epoch": 8.72161172161172, "grad_norm": 1.171875, "learning_rate": 8.964637559712285e-05, "loss": 0.1322, "step": 11905 }, { "epoch": 8.722344322344322, "grad_norm": 0.94921875, "learning_rate": 8.964427847326914e-05, "loss": 0.1317, "step": 11906 }, { "epoch": 8.723076923076922, "grad_norm": 1.171875, "learning_rate": 8.964218116186233e-05, "loss": 0.1417, "step": 11907 }, { "epoch": 8.723809523809523, "grad_norm": 1.4765625, "learning_rate": 8.96400836629125e-05, "loss": 0.1543, "step": 11908 }, { "epoch": 8.724542124542124, "grad_norm": 1.21875, "learning_rate": 8.963798597642965e-05, "loss": 0.1443, "step": 11909 }, { "epoch": 8.725274725274724, "grad_norm": 0.9609375, "learning_rate": 8.96358881024239e-05, "loss": 0.1243, "step": 11910 }, { "epoch": 8.726007326007325, "grad_norm": 1.5234375, "learning_rate": 8.963379004090523e-05, "loss": 0.1524, "step": 11911 }, { "epoch": 8.726739926739926, "grad_norm": 1.0234375, "learning_rate": 8.963169179188373e-05, "loss": 0.1551, "step": 11912 }, { "epoch": 8.727472527472527, "grad_norm": 1.5703125, "learning_rate": 8.962959335536947e-05, "loss": 0.1208, "step": 11913 }, { "epoch": 8.728205128205127, "grad_norm": 1.2421875, "learning_rate": 8.962749473137246e-05, "loss": 0.1192, "step": 11914 }, { "epoch": 8.728937728937728, "grad_norm": 1.796875, "learning_rate": 8.962539591990279e-05, "loss": 0.1193, "step": 11915 }, { "epoch": 8.729670329670329, "grad_norm": 1.84375, "learning_rate": 8.96232969209705e-05, "loss": 0.1347, "step": 11916 }, { "epoch": 8.73040293040293, "grad_norm": 0.7109375, "learning_rate": 8.962119773458565e-05, "loss": 0.119, "step": 11917 }, { "epoch": 8.73113553113553, "grad_norm": 1.1953125, "learning_rate": 8.96190983607583e-05, "loss": 0.1436, "step": 11918 }, { "epoch": 8.731868131868131, "grad_norm": 0.87109375, "learning_rate": 8.961699879949852e-05, "loss": 0.1637, "step": 11919 }, { "epoch": 8.732600732600732, "grad_norm": 1.0546875, "learning_rate": 8.961489905081634e-05, "loss": 0.1191, "step": 11920 }, { "epoch": 8.733333333333333, "grad_norm": 1.1796875, "learning_rate": 8.961279911472183e-05, "loss": 0.1388, "step": 11921 }, { "epoch": 8.734065934065933, "grad_norm": 0.93359375, "learning_rate": 8.961069899122507e-05, "loss": 0.1599, "step": 11922 }, { "epoch": 8.734798534798536, "grad_norm": 0.92578125, "learning_rate": 8.960859868033613e-05, "loss": 0.142, "step": 11923 }, { "epoch": 8.735531135531136, "grad_norm": 1.0625, "learning_rate": 8.960649818206504e-05, "loss": 0.1222, "step": 11924 }, { "epoch": 8.736263736263737, "grad_norm": 0.82421875, "learning_rate": 8.960439749642188e-05, "loss": 0.156, "step": 11925 }, { "epoch": 8.736996336996338, "grad_norm": 0.88671875, "learning_rate": 8.96022966234167e-05, "loss": 0.128, "step": 11926 }, { "epoch": 8.737728937728939, "grad_norm": 1.015625, "learning_rate": 8.960019556305961e-05, "loss": 0.1263, "step": 11927 }, { "epoch": 8.73846153846154, "grad_norm": 1.203125, "learning_rate": 8.959809431536062e-05, "loss": 0.1408, "step": 11928 }, { "epoch": 8.73919413919414, "grad_norm": 0.98046875, "learning_rate": 8.959599288032985e-05, "loss": 0.1264, "step": 11929 }, { "epoch": 8.73992673992674, "grad_norm": 0.609375, "learning_rate": 8.959389125797733e-05, "loss": 0.1069, "step": 11930 }, { "epoch": 8.740659340659342, "grad_norm": 1.0234375, "learning_rate": 8.959178944831315e-05, "loss": 0.1283, "step": 11931 }, { "epoch": 8.741391941391942, "grad_norm": 1.3359375, "learning_rate": 8.958968745134737e-05, "loss": 0.1449, "step": 11932 }, { "epoch": 8.742124542124543, "grad_norm": 1.2421875, "learning_rate": 8.958758526709006e-05, "loss": 0.1591, "step": 11933 }, { "epoch": 8.742857142857144, "grad_norm": 1.15625, "learning_rate": 8.95854828955513e-05, "loss": 0.1653, "step": 11934 }, { "epoch": 8.743589743589745, "grad_norm": 0.5859375, "learning_rate": 8.958338033674115e-05, "loss": 0.1136, "step": 11935 }, { "epoch": 8.744322344322345, "grad_norm": 0.91796875, "learning_rate": 8.958127759066971e-05, "loss": 0.1177, "step": 11936 }, { "epoch": 8.745054945054946, "grad_norm": 0.75, "learning_rate": 8.957917465734703e-05, "loss": 0.1579, "step": 11937 }, { "epoch": 8.745787545787547, "grad_norm": 0.87109375, "learning_rate": 8.957707153678321e-05, "loss": 0.1109, "step": 11938 }, { "epoch": 8.746520146520147, "grad_norm": 0.73046875, "learning_rate": 8.95749682289883e-05, "loss": 0.1319, "step": 11939 }, { "epoch": 8.747252747252748, "grad_norm": 0.98828125, "learning_rate": 8.957286473397239e-05, "loss": 0.1303, "step": 11940 }, { "epoch": 8.747985347985349, "grad_norm": 1.2890625, "learning_rate": 8.957076105174556e-05, "loss": 0.1429, "step": 11941 }, { "epoch": 8.74871794871795, "grad_norm": 2.4375, "learning_rate": 8.956865718231789e-05, "loss": 0.1361, "step": 11942 }, { "epoch": 8.74945054945055, "grad_norm": 1.046875, "learning_rate": 8.956655312569946e-05, "loss": 0.1324, "step": 11943 }, { "epoch": 8.750183150183151, "grad_norm": 1.1875, "learning_rate": 8.956444888190034e-05, "loss": 0.1303, "step": 11944 }, { "epoch": 8.750915750915752, "grad_norm": 1.796875, "learning_rate": 8.956234445093064e-05, "loss": 0.1394, "step": 11945 }, { "epoch": 8.751648351648353, "grad_norm": 0.921875, "learning_rate": 8.956023983280039e-05, "loss": 0.1184, "step": 11946 }, { "epoch": 8.752380952380953, "grad_norm": 1.1328125, "learning_rate": 8.955813502751974e-05, "loss": 0.1146, "step": 11947 }, { "epoch": 8.753113553113554, "grad_norm": 0.97265625, "learning_rate": 8.955603003509874e-05, "loss": 0.151, "step": 11948 }, { "epoch": 8.753846153846155, "grad_norm": 0.53515625, "learning_rate": 8.955392485554747e-05, "loss": 0.1158, "step": 11949 }, { "epoch": 8.754578754578755, "grad_norm": 0.8515625, "learning_rate": 8.955181948887602e-05, "loss": 0.1126, "step": 11950 }, { "epoch": 8.755311355311356, "grad_norm": 0.9375, "learning_rate": 8.954971393509448e-05, "loss": 0.1193, "step": 11951 }, { "epoch": 8.756043956043957, "grad_norm": 1.0234375, "learning_rate": 8.954760819421296e-05, "loss": 0.1159, "step": 11952 }, { "epoch": 8.756776556776558, "grad_norm": 1.5859375, "learning_rate": 8.954550226624152e-05, "loss": 0.1258, "step": 11953 }, { "epoch": 8.757509157509158, "grad_norm": 1.2109375, "learning_rate": 8.954339615119027e-05, "loss": 0.1113, "step": 11954 }, { "epoch": 8.758241758241759, "grad_norm": 0.9921875, "learning_rate": 8.954128984906928e-05, "loss": 0.1761, "step": 11955 }, { "epoch": 8.75897435897436, "grad_norm": 0.609375, "learning_rate": 8.953918335988866e-05, "loss": 0.126, "step": 11956 }, { "epoch": 8.75970695970696, "grad_norm": 1.3828125, "learning_rate": 8.95370766836585e-05, "loss": 0.1534, "step": 11957 }, { "epoch": 8.760439560439561, "grad_norm": 1.0859375, "learning_rate": 8.953496982038887e-05, "loss": 0.1256, "step": 11958 }, { "epoch": 8.761172161172162, "grad_norm": 1.671875, "learning_rate": 8.953286277008992e-05, "loss": 0.1172, "step": 11959 }, { "epoch": 8.761904761904763, "grad_norm": 1.8125, "learning_rate": 8.95307555327717e-05, "loss": 0.142, "step": 11960 }, { "epoch": 8.762637362637363, "grad_norm": 1.015625, "learning_rate": 8.952864810844429e-05, "loss": 0.1897, "step": 11961 }, { "epoch": 8.763369963369964, "grad_norm": 0.91796875, "learning_rate": 8.952654049711784e-05, "loss": 0.1407, "step": 11962 }, { "epoch": 8.764102564102565, "grad_norm": 0.92578125, "learning_rate": 8.952443269880241e-05, "loss": 0.1314, "step": 11963 }, { "epoch": 8.764835164835166, "grad_norm": 1.4140625, "learning_rate": 8.952232471350812e-05, "loss": 0.1424, "step": 11964 }, { "epoch": 8.765567765567766, "grad_norm": 1.09375, "learning_rate": 8.952021654124506e-05, "loss": 0.1271, "step": 11965 }, { "epoch": 8.766300366300367, "grad_norm": 1.0703125, "learning_rate": 8.951810818202334e-05, "loss": 0.1271, "step": 11966 }, { "epoch": 8.767032967032968, "grad_norm": 0.8515625, "learning_rate": 8.951599963585306e-05, "loss": 0.116, "step": 11967 }, { "epoch": 8.767765567765569, "grad_norm": 1.21875, "learning_rate": 8.951389090274429e-05, "loss": 0.1502, "step": 11968 }, { "epoch": 8.76849816849817, "grad_norm": 1.6171875, "learning_rate": 8.951178198270719e-05, "loss": 0.1359, "step": 11969 }, { "epoch": 8.76923076923077, "grad_norm": 1.03125, "learning_rate": 8.950967287575182e-05, "loss": 0.1348, "step": 11970 }, { "epoch": 8.76996336996337, "grad_norm": 1.25, "learning_rate": 8.950756358188829e-05, "loss": 0.1372, "step": 11971 }, { "epoch": 8.770695970695972, "grad_norm": 0.9296875, "learning_rate": 8.950545410112675e-05, "loss": 0.0979, "step": 11972 }, { "epoch": 8.771428571428572, "grad_norm": 1.515625, "learning_rate": 8.950334443347724e-05, "loss": 0.1438, "step": 11973 }, { "epoch": 8.772161172161173, "grad_norm": 0.86328125, "learning_rate": 8.950123457894992e-05, "loss": 0.1444, "step": 11974 }, { "epoch": 8.772893772893774, "grad_norm": 1.1484375, "learning_rate": 8.94991245375549e-05, "loss": 0.1493, "step": 11975 }, { "epoch": 8.773626373626374, "grad_norm": 0.73828125, "learning_rate": 8.949701430930224e-05, "loss": 0.1144, "step": 11976 }, { "epoch": 8.774358974358975, "grad_norm": 0.96875, "learning_rate": 8.94949038942021e-05, "loss": 0.1195, "step": 11977 }, { "epoch": 8.775091575091576, "grad_norm": 1.34375, "learning_rate": 8.949279329226458e-05, "loss": 0.1452, "step": 11978 }, { "epoch": 8.775824175824177, "grad_norm": 0.9453125, "learning_rate": 8.949068250349977e-05, "loss": 0.1345, "step": 11979 }, { "epoch": 8.776556776556777, "grad_norm": 1.296875, "learning_rate": 8.948857152791781e-05, "loss": 0.1404, "step": 11980 }, { "epoch": 8.777289377289378, "grad_norm": 1.3203125, "learning_rate": 8.94864603655288e-05, "loss": 0.1746, "step": 11981 }, { "epoch": 8.778021978021979, "grad_norm": 0.95703125, "learning_rate": 8.948434901634286e-05, "loss": 0.1415, "step": 11982 }, { "epoch": 8.77875457875458, "grad_norm": 0.953125, "learning_rate": 8.948223748037012e-05, "loss": 0.1345, "step": 11983 }, { "epoch": 8.77948717948718, "grad_norm": 1.0703125, "learning_rate": 8.948012575762067e-05, "loss": 0.1293, "step": 11984 }, { "epoch": 8.780219780219781, "grad_norm": 1.2734375, "learning_rate": 8.947801384810464e-05, "loss": 0.1202, "step": 11985 }, { "epoch": 8.780952380952382, "grad_norm": 1.0859375, "learning_rate": 8.947590175183216e-05, "loss": 0.1393, "step": 11986 }, { "epoch": 8.781684981684982, "grad_norm": 1.0390625, "learning_rate": 8.947378946881332e-05, "loss": 0.142, "step": 11987 }, { "epoch": 8.782417582417583, "grad_norm": 0.95703125, "learning_rate": 8.947167699905828e-05, "loss": 0.1408, "step": 11988 }, { "epoch": 8.783150183150184, "grad_norm": 1.4140625, "learning_rate": 8.946956434257711e-05, "loss": 0.1673, "step": 11989 }, { "epoch": 8.783882783882785, "grad_norm": 1.3828125, "learning_rate": 8.946745149938e-05, "loss": 0.1148, "step": 11990 }, { "epoch": 8.784615384615385, "grad_norm": 1.484375, "learning_rate": 8.946533846947704e-05, "loss": 0.1129, "step": 11991 }, { "epoch": 8.785347985347986, "grad_norm": 1.9921875, "learning_rate": 8.946322525287833e-05, "loss": 0.1753, "step": 11992 }, { "epoch": 8.786080586080587, "grad_norm": 1.0078125, "learning_rate": 8.946111184959403e-05, "loss": 0.1451, "step": 11993 }, { "epoch": 8.786813186813188, "grad_norm": 0.8671875, "learning_rate": 8.945899825963425e-05, "loss": 0.0999, "step": 11994 }, { "epoch": 8.787545787545788, "grad_norm": 1.3671875, "learning_rate": 8.945688448300912e-05, "loss": 0.1326, "step": 11995 }, { "epoch": 8.788278388278389, "grad_norm": 0.9453125, "learning_rate": 8.945477051972876e-05, "loss": 0.1471, "step": 11996 }, { "epoch": 8.78901098901099, "grad_norm": 1.1953125, "learning_rate": 8.945265636980333e-05, "loss": 0.1582, "step": 11997 }, { "epoch": 8.78974358974359, "grad_norm": 1.453125, "learning_rate": 8.945054203324291e-05, "loss": 0.14, "step": 11998 }, { "epoch": 8.790476190476191, "grad_norm": 1.25, "learning_rate": 8.944842751005767e-05, "loss": 0.1259, "step": 11999 }, { "epoch": 8.791208791208792, "grad_norm": 0.8359375, "learning_rate": 8.944631280025773e-05, "loss": 0.1078, "step": 12000 }, { "epoch": 8.791941391941393, "grad_norm": 0.8671875, "learning_rate": 8.94441979038532e-05, "loss": 0.1322, "step": 12001 }, { "epoch": 8.792673992673993, "grad_norm": 1.359375, "learning_rate": 8.944208282085425e-05, "loss": 0.123, "step": 12002 }, { "epoch": 8.793406593406594, "grad_norm": 0.9921875, "learning_rate": 8.943996755127099e-05, "loss": 0.1468, "step": 12003 }, { "epoch": 8.794139194139195, "grad_norm": 1.46875, "learning_rate": 8.943785209511358e-05, "loss": 0.158, "step": 12004 }, { "epoch": 8.794871794871796, "grad_norm": 1.171875, "learning_rate": 8.943573645239211e-05, "loss": 0.1221, "step": 12005 }, { "epoch": 8.795604395604396, "grad_norm": 0.9375, "learning_rate": 8.943362062311675e-05, "loss": 0.1153, "step": 12006 }, { "epoch": 8.796336996336997, "grad_norm": 1.4765625, "learning_rate": 8.943150460729766e-05, "loss": 0.1527, "step": 12007 }, { "epoch": 8.797069597069598, "grad_norm": 1.28125, "learning_rate": 8.942938840494491e-05, "loss": 0.1191, "step": 12008 }, { "epoch": 8.797802197802199, "grad_norm": 0.77734375, "learning_rate": 8.942727201606872e-05, "loss": 0.1317, "step": 12009 }, { "epoch": 8.7985347985348, "grad_norm": 0.91796875, "learning_rate": 8.942515544067919e-05, "loss": 0.1444, "step": 12010 }, { "epoch": 8.7992673992674, "grad_norm": 1.1015625, "learning_rate": 8.942303867878643e-05, "loss": 0.124, "step": 12011 }, { "epoch": 8.8, "grad_norm": 1.2421875, "learning_rate": 8.942092173040063e-05, "loss": 0.116, "step": 12012 }, { "epoch": 8.800732600732601, "grad_norm": 1.1953125, "learning_rate": 8.941880459553193e-05, "loss": 0.1349, "step": 12013 }, { "epoch": 8.801465201465202, "grad_norm": 1.4609375, "learning_rate": 8.941668727419044e-05, "loss": 0.1321, "step": 12014 }, { "epoch": 8.802197802197803, "grad_norm": 0.92578125, "learning_rate": 8.941456976638634e-05, "loss": 0.1391, "step": 12015 }, { "epoch": 8.802930402930404, "grad_norm": 1.0546875, "learning_rate": 8.941245207212975e-05, "loss": 0.1524, "step": 12016 }, { "epoch": 8.803663003663004, "grad_norm": 1.1953125, "learning_rate": 8.941033419143083e-05, "loss": 0.1528, "step": 12017 }, { "epoch": 8.804395604395605, "grad_norm": 1.4453125, "learning_rate": 8.940821612429974e-05, "loss": 0.156, "step": 12018 }, { "epoch": 8.805128205128206, "grad_norm": 1.0, "learning_rate": 8.940609787074659e-05, "loss": 0.1301, "step": 12019 }, { "epoch": 8.805860805860807, "grad_norm": 1.1015625, "learning_rate": 8.940397943078157e-05, "loss": 0.1589, "step": 12020 }, { "epoch": 8.806593406593407, "grad_norm": 1.109375, "learning_rate": 8.94018608044148e-05, "loss": 0.1298, "step": 12021 }, { "epoch": 8.807326007326008, "grad_norm": 1.078125, "learning_rate": 8.939974199165646e-05, "loss": 0.1445, "step": 12022 }, { "epoch": 8.808058608058609, "grad_norm": 1.3125, "learning_rate": 8.939762299251668e-05, "loss": 0.1339, "step": 12023 }, { "epoch": 8.80879120879121, "grad_norm": 1.25, "learning_rate": 8.939550380700562e-05, "loss": 0.1652, "step": 12024 }, { "epoch": 8.80952380952381, "grad_norm": 1.2890625, "learning_rate": 8.939338443513343e-05, "loss": 0.1253, "step": 12025 }, { "epoch": 8.810256410256411, "grad_norm": 1.0078125, "learning_rate": 8.939126487691028e-05, "loss": 0.1559, "step": 12026 }, { "epoch": 8.810989010989012, "grad_norm": 1.0625, "learning_rate": 8.938914513234628e-05, "loss": 0.1149, "step": 12027 }, { "epoch": 8.811721611721612, "grad_norm": 1.1171875, "learning_rate": 8.938702520145166e-05, "loss": 0.1297, "step": 12028 }, { "epoch": 8.812454212454213, "grad_norm": 1.0546875, "learning_rate": 8.938490508423652e-05, "loss": 0.1464, "step": 12029 }, { "epoch": 8.813186813186814, "grad_norm": 1.0546875, "learning_rate": 8.938278478071103e-05, "loss": 0.1171, "step": 12030 }, { "epoch": 8.813919413919415, "grad_norm": 1.2890625, "learning_rate": 8.938066429088535e-05, "loss": 0.1466, "step": 12031 }, { "epoch": 8.814652014652015, "grad_norm": 1.1328125, "learning_rate": 8.937854361476965e-05, "loss": 0.1216, "step": 12032 }, { "epoch": 8.815384615384616, "grad_norm": 1.234375, "learning_rate": 8.937642275237408e-05, "loss": 0.1566, "step": 12033 }, { "epoch": 8.816117216117217, "grad_norm": 1.0390625, "learning_rate": 8.937430170370881e-05, "loss": 0.1591, "step": 12034 }, { "epoch": 8.816849816849818, "grad_norm": 0.84375, "learning_rate": 8.9372180468784e-05, "loss": 0.1258, "step": 12035 }, { "epoch": 8.817582417582418, "grad_norm": 0.984375, "learning_rate": 8.937005904760981e-05, "loss": 0.1592, "step": 12036 }, { "epoch": 8.818315018315019, "grad_norm": 1.28125, "learning_rate": 8.936793744019642e-05, "loss": 0.1328, "step": 12037 }, { "epoch": 8.81904761904762, "grad_norm": 0.78515625, "learning_rate": 8.936581564655397e-05, "loss": 0.1544, "step": 12038 }, { "epoch": 8.81978021978022, "grad_norm": 0.94140625, "learning_rate": 8.936369366669264e-05, "loss": 0.1432, "step": 12039 }, { "epoch": 8.820512820512821, "grad_norm": 1.3515625, "learning_rate": 8.936157150062257e-05, "loss": 0.1336, "step": 12040 }, { "epoch": 8.821245421245422, "grad_norm": 0.87109375, "learning_rate": 8.935944914835399e-05, "loss": 0.1353, "step": 12041 }, { "epoch": 8.821978021978023, "grad_norm": 1.046875, "learning_rate": 8.935732660989702e-05, "loss": 0.1307, "step": 12042 }, { "epoch": 8.822710622710623, "grad_norm": 0.8828125, "learning_rate": 8.935520388526184e-05, "loss": 0.132, "step": 12043 }, { "epoch": 8.823443223443224, "grad_norm": 1.421875, "learning_rate": 8.935308097445863e-05, "loss": 0.1516, "step": 12044 }, { "epoch": 8.824175824175825, "grad_norm": 1.3515625, "learning_rate": 8.935095787749755e-05, "loss": 0.1598, "step": 12045 }, { "epoch": 8.824908424908426, "grad_norm": 1.0625, "learning_rate": 8.934883459438877e-05, "loss": 0.1447, "step": 12046 }, { "epoch": 8.825641025641026, "grad_norm": 1.0546875, "learning_rate": 8.934671112514249e-05, "loss": 0.1197, "step": 12047 }, { "epoch": 8.826373626373627, "grad_norm": 0.8984375, "learning_rate": 8.934458746976885e-05, "loss": 0.1267, "step": 12048 }, { "epoch": 8.827106227106228, "grad_norm": 1.25, "learning_rate": 8.934246362827805e-05, "loss": 0.1409, "step": 12049 }, { "epoch": 8.827838827838828, "grad_norm": 1.9921875, "learning_rate": 8.934033960068024e-05, "loss": 0.1488, "step": 12050 }, { "epoch": 8.82857142857143, "grad_norm": 1.5859375, "learning_rate": 8.933821538698562e-05, "loss": 0.1347, "step": 12051 }, { "epoch": 8.82930402930403, "grad_norm": 1.2421875, "learning_rate": 8.933609098720436e-05, "loss": 0.0969, "step": 12052 }, { "epoch": 8.83003663003663, "grad_norm": 1.78125, "learning_rate": 8.933396640134662e-05, "loss": 0.1681, "step": 12053 }, { "epoch": 8.830769230769231, "grad_norm": 1.1875, "learning_rate": 8.933184162942262e-05, "loss": 0.144, "step": 12054 }, { "epoch": 8.831501831501832, "grad_norm": 1.2734375, "learning_rate": 8.932971667144253e-05, "loss": 0.1449, "step": 12055 }, { "epoch": 8.832234432234433, "grad_norm": 1.1875, "learning_rate": 8.932759152741649e-05, "loss": 0.1577, "step": 12056 }, { "epoch": 8.832967032967034, "grad_norm": 1.046875, "learning_rate": 8.932546619735474e-05, "loss": 0.1586, "step": 12057 }, { "epoch": 8.833699633699634, "grad_norm": 1.125, "learning_rate": 8.932334068126743e-05, "loss": 0.1296, "step": 12058 }, { "epoch": 8.834432234432235, "grad_norm": 0.73828125, "learning_rate": 8.932121497916474e-05, "loss": 0.1366, "step": 12059 }, { "epoch": 8.835164835164836, "grad_norm": 1.3984375, "learning_rate": 8.931908909105688e-05, "loss": 0.2049, "step": 12060 }, { "epoch": 8.835897435897436, "grad_norm": 1.0234375, "learning_rate": 8.9316963016954e-05, "loss": 0.1292, "step": 12061 }, { "epoch": 8.836630036630037, "grad_norm": 1.2265625, "learning_rate": 8.931483675686633e-05, "loss": 0.1525, "step": 12062 }, { "epoch": 8.837362637362638, "grad_norm": 1.0078125, "learning_rate": 8.931271031080402e-05, "loss": 0.1234, "step": 12063 }, { "epoch": 8.838095238095239, "grad_norm": 1.375, "learning_rate": 8.931058367877728e-05, "loss": 0.155, "step": 12064 }, { "epoch": 8.83882783882784, "grad_norm": 0.67578125, "learning_rate": 8.930845686079628e-05, "loss": 0.1212, "step": 12065 }, { "epoch": 8.83956043956044, "grad_norm": 0.875, "learning_rate": 8.930632985687123e-05, "loss": 0.1357, "step": 12066 }, { "epoch": 8.84029304029304, "grad_norm": 1.2265625, "learning_rate": 8.930420266701232e-05, "loss": 0.1496, "step": 12067 }, { "epoch": 8.841025641025642, "grad_norm": 0.82421875, "learning_rate": 8.930207529122972e-05, "loss": 0.1225, "step": 12068 }, { "epoch": 8.841758241758242, "grad_norm": 1.2421875, "learning_rate": 8.929994772953367e-05, "loss": 0.1611, "step": 12069 }, { "epoch": 8.842490842490843, "grad_norm": 0.90234375, "learning_rate": 8.92978199819343e-05, "loss": 0.1277, "step": 12070 }, { "epoch": 8.843223443223444, "grad_norm": 0.7578125, "learning_rate": 8.929569204844187e-05, "loss": 0.1533, "step": 12071 }, { "epoch": 8.843956043956045, "grad_norm": 1.171875, "learning_rate": 8.929356392906653e-05, "loss": 0.1215, "step": 12072 }, { "epoch": 8.844688644688645, "grad_norm": 0.97265625, "learning_rate": 8.929143562381848e-05, "loss": 0.1424, "step": 12073 }, { "epoch": 8.845421245421246, "grad_norm": 1.390625, "learning_rate": 8.928930713270796e-05, "loss": 0.1559, "step": 12074 }, { "epoch": 8.846153846153847, "grad_norm": 1.6953125, "learning_rate": 8.92871784557451e-05, "loss": 0.1169, "step": 12075 }, { "epoch": 8.846886446886447, "grad_norm": 0.5859375, "learning_rate": 8.928504959294016e-05, "loss": 0.1177, "step": 12076 }, { "epoch": 8.847619047619048, "grad_norm": 0.8359375, "learning_rate": 8.928292054430328e-05, "loss": 0.1473, "step": 12077 }, { "epoch": 8.848351648351649, "grad_norm": 0.921875, "learning_rate": 8.928079130984473e-05, "loss": 0.1207, "step": 12078 }, { "epoch": 8.84908424908425, "grad_norm": 0.640625, "learning_rate": 8.927866188957466e-05, "loss": 0.1628, "step": 12079 }, { "epoch": 8.84981684981685, "grad_norm": 1.5859375, "learning_rate": 8.927653228350329e-05, "loss": 0.1361, "step": 12080 }, { "epoch": 8.850549450549451, "grad_norm": 0.97265625, "learning_rate": 8.927440249164083e-05, "loss": 0.113, "step": 12081 }, { "epoch": 8.851282051282052, "grad_norm": 1.4453125, "learning_rate": 8.927227251399749e-05, "loss": 0.1394, "step": 12082 }, { "epoch": 8.852014652014653, "grad_norm": 1.484375, "learning_rate": 8.927014235058345e-05, "loss": 0.1303, "step": 12083 }, { "epoch": 8.852747252747253, "grad_norm": 0.97265625, "learning_rate": 8.926801200140894e-05, "loss": 0.1111, "step": 12084 }, { "epoch": 8.853479853479854, "grad_norm": 0.7265625, "learning_rate": 8.926588146648413e-05, "loss": 0.1275, "step": 12085 }, { "epoch": 8.854212454212455, "grad_norm": 0.921875, "learning_rate": 8.926375074581929e-05, "loss": 0.1284, "step": 12086 }, { "epoch": 8.854945054945055, "grad_norm": 0.765625, "learning_rate": 8.926161983942458e-05, "loss": 0.1308, "step": 12087 }, { "epoch": 8.855677655677656, "grad_norm": 2.15625, "learning_rate": 8.925948874731022e-05, "loss": 0.1502, "step": 12088 }, { "epoch": 8.856410256410257, "grad_norm": 0.859375, "learning_rate": 8.925735746948644e-05, "loss": 0.1222, "step": 12089 }, { "epoch": 8.857142857142858, "grad_norm": 1.0703125, "learning_rate": 8.925522600596343e-05, "loss": 0.1197, "step": 12090 }, { "epoch": 8.857875457875458, "grad_norm": 1.4453125, "learning_rate": 8.925309435675141e-05, "loss": 0.1494, "step": 12091 }, { "epoch": 8.858608058608059, "grad_norm": 1.1640625, "learning_rate": 8.92509625218606e-05, "loss": 0.1621, "step": 12092 }, { "epoch": 8.85934065934066, "grad_norm": 1.96875, "learning_rate": 8.924883050130119e-05, "loss": 0.1421, "step": 12093 }, { "epoch": 8.86007326007326, "grad_norm": 1.046875, "learning_rate": 8.924669829508342e-05, "loss": 0.1531, "step": 12094 }, { "epoch": 8.860805860805861, "grad_norm": 2.234375, "learning_rate": 8.924456590321752e-05, "loss": 0.1229, "step": 12095 }, { "epoch": 8.861538461538462, "grad_norm": 1.7734375, "learning_rate": 8.924243332571367e-05, "loss": 0.1508, "step": 12096 }, { "epoch": 8.862271062271063, "grad_norm": 1.03125, "learning_rate": 8.92403005625821e-05, "loss": 0.1597, "step": 12097 }, { "epoch": 8.863003663003663, "grad_norm": 0.59765625, "learning_rate": 8.923816761383304e-05, "loss": 0.127, "step": 12098 }, { "epoch": 8.863736263736264, "grad_norm": 1.234375, "learning_rate": 8.923603447947668e-05, "loss": 0.1019, "step": 12099 }, { "epoch": 8.864468864468865, "grad_norm": 1.3203125, "learning_rate": 8.923390115952328e-05, "loss": 0.1015, "step": 12100 }, { "epoch": 8.865201465201466, "grad_norm": 1.4296875, "learning_rate": 8.923176765398305e-05, "loss": 0.1642, "step": 12101 }, { "epoch": 8.865934065934066, "grad_norm": 0.8984375, "learning_rate": 8.92296339628662e-05, "loss": 0.1204, "step": 12102 }, { "epoch": 8.866666666666667, "grad_norm": 0.6875, "learning_rate": 8.922750008618295e-05, "loss": 0.1147, "step": 12103 }, { "epoch": 8.867399267399268, "grad_norm": 1.15625, "learning_rate": 8.922536602394355e-05, "loss": 0.1461, "step": 12104 }, { "epoch": 8.868131868131869, "grad_norm": 0.953125, "learning_rate": 8.92232317761582e-05, "loss": 0.1152, "step": 12105 }, { "epoch": 8.86886446886447, "grad_norm": 0.94140625, "learning_rate": 8.922109734283712e-05, "loss": 0.1443, "step": 12106 }, { "epoch": 8.86959706959707, "grad_norm": 1.109375, "learning_rate": 8.921896272399059e-05, "loss": 0.1504, "step": 12107 }, { "epoch": 8.87032967032967, "grad_norm": 0.94140625, "learning_rate": 8.921682791962876e-05, "loss": 0.1538, "step": 12108 }, { "epoch": 8.871062271062272, "grad_norm": 1.109375, "learning_rate": 8.921469292976191e-05, "loss": 0.1361, "step": 12109 }, { "epoch": 8.871794871794872, "grad_norm": 0.8984375, "learning_rate": 8.921255775440027e-05, "loss": 0.1277, "step": 12110 }, { "epoch": 8.872527472527473, "grad_norm": 1.15625, "learning_rate": 8.921042239355403e-05, "loss": 0.1629, "step": 12111 }, { "epoch": 8.873260073260074, "grad_norm": 1.296875, "learning_rate": 8.920828684723347e-05, "loss": 0.141, "step": 12112 }, { "epoch": 8.873992673992674, "grad_norm": 0.66796875, "learning_rate": 8.920615111544881e-05, "loss": 0.1311, "step": 12113 }, { "epoch": 8.874725274725275, "grad_norm": 0.80078125, "learning_rate": 8.920401519821027e-05, "loss": 0.1667, "step": 12114 }, { "epoch": 8.875457875457876, "grad_norm": 0.97265625, "learning_rate": 8.920187909552806e-05, "loss": 0.1358, "step": 12115 }, { "epoch": 8.876190476190477, "grad_norm": 0.70703125, "learning_rate": 8.919974280741247e-05, "loss": 0.1475, "step": 12116 }, { "epoch": 8.876923076923077, "grad_norm": 0.78125, "learning_rate": 8.91976063338737e-05, "loss": 0.1573, "step": 12117 }, { "epoch": 8.877655677655678, "grad_norm": 1.1484375, "learning_rate": 8.919546967492198e-05, "loss": 0.1409, "step": 12118 }, { "epoch": 8.878388278388279, "grad_norm": 1.2265625, "learning_rate": 8.919333283056759e-05, "loss": 0.1398, "step": 12119 }, { "epoch": 8.87912087912088, "grad_norm": 1.3671875, "learning_rate": 8.91911958008207e-05, "loss": 0.1537, "step": 12120 }, { "epoch": 8.87985347985348, "grad_norm": 1.203125, "learning_rate": 8.918905858569163e-05, "loss": 0.1069, "step": 12121 }, { "epoch": 8.880586080586081, "grad_norm": 0.8984375, "learning_rate": 8.918692118519056e-05, "loss": 0.1194, "step": 12122 }, { "epoch": 8.881318681318682, "grad_norm": 1.90625, "learning_rate": 8.918478359932776e-05, "loss": 0.1594, "step": 12123 }, { "epoch": 8.882051282051282, "grad_norm": 1.0390625, "learning_rate": 8.918264582811346e-05, "loss": 0.1535, "step": 12124 }, { "epoch": 8.882783882783883, "grad_norm": 1.359375, "learning_rate": 8.918050787155789e-05, "loss": 0.1412, "step": 12125 }, { "epoch": 8.883516483516484, "grad_norm": 1.0078125, "learning_rate": 8.917836972967132e-05, "loss": 0.1064, "step": 12126 }, { "epoch": 8.884249084249085, "grad_norm": 0.6328125, "learning_rate": 8.917623140246398e-05, "loss": 0.1123, "step": 12127 }, { "epoch": 8.884981684981685, "grad_norm": 1.2734375, "learning_rate": 8.917409288994614e-05, "loss": 0.1249, "step": 12128 }, { "epoch": 8.885714285714286, "grad_norm": 1.1015625, "learning_rate": 8.9171954192128e-05, "loss": 0.127, "step": 12129 }, { "epoch": 8.886446886446887, "grad_norm": 0.99609375, "learning_rate": 8.916981530901983e-05, "loss": 0.1491, "step": 12130 }, { "epoch": 8.887179487179488, "grad_norm": 1.75, "learning_rate": 8.916767624063189e-05, "loss": 0.1415, "step": 12131 }, { "epoch": 8.887912087912088, "grad_norm": 1.8359375, "learning_rate": 8.916553698697442e-05, "loss": 0.1185, "step": 12132 }, { "epoch": 8.888644688644689, "grad_norm": 1.515625, "learning_rate": 8.916339754805767e-05, "loss": 0.1565, "step": 12133 }, { "epoch": 8.88937728937729, "grad_norm": 1.4296875, "learning_rate": 8.916125792389189e-05, "loss": 0.128, "step": 12134 }, { "epoch": 8.89010989010989, "grad_norm": 1.109375, "learning_rate": 8.915911811448733e-05, "loss": 0.132, "step": 12135 }, { "epoch": 8.890842490842491, "grad_norm": 1.1015625, "learning_rate": 8.915697811985425e-05, "loss": 0.1614, "step": 12136 }, { "epoch": 8.891575091575092, "grad_norm": 0.75, "learning_rate": 8.915483794000289e-05, "loss": 0.123, "step": 12137 }, { "epoch": 8.892307692307693, "grad_norm": 0.83203125, "learning_rate": 8.915269757494352e-05, "loss": 0.1322, "step": 12138 }, { "epoch": 8.893040293040293, "grad_norm": 1.140625, "learning_rate": 8.915055702468637e-05, "loss": 0.1553, "step": 12139 }, { "epoch": 8.893772893772894, "grad_norm": 0.76953125, "learning_rate": 8.914841628924172e-05, "loss": 0.1194, "step": 12140 }, { "epoch": 8.894505494505495, "grad_norm": 0.796875, "learning_rate": 8.914627536861982e-05, "loss": 0.0934, "step": 12141 }, { "epoch": 8.895238095238096, "grad_norm": 1.4609375, "learning_rate": 8.914413426283094e-05, "loss": 0.1609, "step": 12142 }, { "epoch": 8.895970695970696, "grad_norm": 0.8828125, "learning_rate": 8.914199297188532e-05, "loss": 0.1475, "step": 12143 }, { "epoch": 8.896703296703297, "grad_norm": 1.4375, "learning_rate": 8.913985149579323e-05, "loss": 0.166, "step": 12144 }, { "epoch": 8.897435897435898, "grad_norm": 1.21875, "learning_rate": 8.913770983456492e-05, "loss": 0.0996, "step": 12145 }, { "epoch": 8.898168498168499, "grad_norm": 1.1484375, "learning_rate": 8.913556798821066e-05, "loss": 0.1478, "step": 12146 }, { "epoch": 8.8989010989011, "grad_norm": 1.1171875, "learning_rate": 8.91334259567407e-05, "loss": 0.1399, "step": 12147 }, { "epoch": 8.8996336996337, "grad_norm": 0.6171875, "learning_rate": 8.913128374016534e-05, "loss": 0.1193, "step": 12148 }, { "epoch": 8.9003663003663, "grad_norm": 1.265625, "learning_rate": 8.912914133849479e-05, "loss": 0.1125, "step": 12149 }, { "epoch": 8.901098901098901, "grad_norm": 1.6015625, "learning_rate": 8.912699875173935e-05, "loss": 0.1076, "step": 12150 }, { "epoch": 8.901831501831502, "grad_norm": 1.125, "learning_rate": 8.912485597990929e-05, "loss": 0.185, "step": 12151 }, { "epoch": 8.902564102564103, "grad_norm": 1.6171875, "learning_rate": 8.912271302301486e-05, "loss": 0.2115, "step": 12152 }, { "epoch": 8.903296703296704, "grad_norm": 1.6171875, "learning_rate": 8.912056988106634e-05, "loss": 0.1382, "step": 12153 }, { "epoch": 8.904029304029304, "grad_norm": 0.6171875, "learning_rate": 8.911842655407397e-05, "loss": 0.1198, "step": 12154 }, { "epoch": 8.904761904761905, "grad_norm": 0.9375, "learning_rate": 8.911628304204807e-05, "loss": 0.1192, "step": 12155 }, { "epoch": 8.905494505494506, "grad_norm": 1.703125, "learning_rate": 8.911413934499887e-05, "loss": 0.1517, "step": 12156 }, { "epoch": 8.906227106227107, "grad_norm": 0.98046875, "learning_rate": 8.911199546293663e-05, "loss": 0.1326, "step": 12157 }, { "epoch": 8.906959706959707, "grad_norm": 0.9765625, "learning_rate": 8.910985139587167e-05, "loss": 0.1132, "step": 12158 }, { "epoch": 8.907692307692308, "grad_norm": 0.8203125, "learning_rate": 8.910770714381423e-05, "loss": 0.1045, "step": 12159 }, { "epoch": 8.908424908424909, "grad_norm": 0.83984375, "learning_rate": 8.910556270677459e-05, "loss": 0.1299, "step": 12160 }, { "epoch": 8.90915750915751, "grad_norm": 0.67578125, "learning_rate": 8.910341808476302e-05, "loss": 0.1293, "step": 12161 }, { "epoch": 8.90989010989011, "grad_norm": 1.4375, "learning_rate": 8.910127327778981e-05, "loss": 0.14, "step": 12162 }, { "epoch": 8.910622710622711, "grad_norm": 0.78515625, "learning_rate": 8.909912828586522e-05, "loss": 0.1205, "step": 12163 }, { "epoch": 8.911355311355312, "grad_norm": 0.58984375, "learning_rate": 8.909698310899954e-05, "loss": 0.1188, "step": 12164 }, { "epoch": 8.912087912087912, "grad_norm": 0.7421875, "learning_rate": 8.909483774720305e-05, "loss": 0.1186, "step": 12165 }, { "epoch": 8.912820512820513, "grad_norm": 0.83984375, "learning_rate": 8.909269220048601e-05, "loss": 0.1472, "step": 12166 }, { "epoch": 8.913553113553114, "grad_norm": 1.015625, "learning_rate": 8.909054646885873e-05, "loss": 0.1937, "step": 12167 }, { "epoch": 8.914285714285715, "grad_norm": 1.171875, "learning_rate": 8.908840055233144e-05, "loss": 0.122, "step": 12168 }, { "epoch": 8.915018315018315, "grad_norm": 1.109375, "learning_rate": 8.908625445091449e-05, "loss": 0.1621, "step": 12169 }, { "epoch": 8.915750915750916, "grad_norm": 1.34375, "learning_rate": 8.908410816461811e-05, "loss": 0.113, "step": 12170 }, { "epoch": 8.916483516483517, "grad_norm": 1.1640625, "learning_rate": 8.908196169345261e-05, "loss": 0.1552, "step": 12171 }, { "epoch": 8.917216117216118, "grad_norm": 1.03125, "learning_rate": 8.907981503742826e-05, "loss": 0.1336, "step": 12172 }, { "epoch": 8.917948717948718, "grad_norm": 0.86328125, "learning_rate": 8.907766819655535e-05, "loss": 0.1504, "step": 12173 }, { "epoch": 8.918681318681319, "grad_norm": 0.75390625, "learning_rate": 8.907552117084416e-05, "loss": 0.1398, "step": 12174 }, { "epoch": 8.91941391941392, "grad_norm": 0.91796875, "learning_rate": 8.907337396030499e-05, "loss": 0.1384, "step": 12175 }, { "epoch": 8.92014652014652, "grad_norm": 1.375, "learning_rate": 8.907122656494815e-05, "loss": 0.114, "step": 12176 }, { "epoch": 8.920879120879121, "grad_norm": 1.0546875, "learning_rate": 8.906907898478388e-05, "loss": 0.149, "step": 12177 }, { "epoch": 8.921611721611722, "grad_norm": 1.1328125, "learning_rate": 8.906693121982249e-05, "loss": 0.1331, "step": 12178 }, { "epoch": 8.922344322344323, "grad_norm": 1.1875, "learning_rate": 8.906478327007427e-05, "loss": 0.1181, "step": 12179 }, { "epoch": 8.923076923076923, "grad_norm": 1.1796875, "learning_rate": 8.906263513554951e-05, "loss": 0.1317, "step": 12180 }, { "epoch": 8.923809523809524, "grad_norm": 1.7109375, "learning_rate": 8.906048681625852e-05, "loss": 0.2097, "step": 12181 }, { "epoch": 8.924542124542125, "grad_norm": 1.140625, "learning_rate": 8.905833831221157e-05, "loss": 0.1413, "step": 12182 }, { "epoch": 8.925274725274726, "grad_norm": 1.109375, "learning_rate": 8.905618962341895e-05, "loss": 0.1134, "step": 12183 }, { "epoch": 8.926007326007326, "grad_norm": 1.625, "learning_rate": 8.9054040749891e-05, "loss": 0.1451, "step": 12184 }, { "epoch": 8.926739926739927, "grad_norm": 1.3515625, "learning_rate": 8.905189169163795e-05, "loss": 0.1651, "step": 12185 }, { "epoch": 8.927472527472528, "grad_norm": 0.77734375, "learning_rate": 8.904974244867016e-05, "loss": 0.1402, "step": 12186 }, { "epoch": 8.928205128205128, "grad_norm": 1.5, "learning_rate": 8.904759302099787e-05, "loss": 0.1508, "step": 12187 }, { "epoch": 8.92893772893773, "grad_norm": 0.77734375, "learning_rate": 8.904544340863144e-05, "loss": 0.1325, "step": 12188 }, { "epoch": 8.92967032967033, "grad_norm": 0.96875, "learning_rate": 8.90432936115811e-05, "loss": 0.1232, "step": 12189 }, { "epoch": 8.93040293040293, "grad_norm": 1.15625, "learning_rate": 8.904114362985722e-05, "loss": 0.1238, "step": 12190 }, { "epoch": 8.931135531135531, "grad_norm": 1.328125, "learning_rate": 8.903899346347007e-05, "loss": 0.1187, "step": 12191 }, { "epoch": 8.931868131868132, "grad_norm": 1.5234375, "learning_rate": 8.903684311242992e-05, "loss": 0.1395, "step": 12192 }, { "epoch": 8.932600732600733, "grad_norm": 0.85546875, "learning_rate": 8.903469257674712e-05, "loss": 0.1211, "step": 12193 }, { "epoch": 8.933333333333334, "grad_norm": 1.4921875, "learning_rate": 8.903254185643197e-05, "loss": 0.1391, "step": 12194 }, { "epoch": 8.934065934065934, "grad_norm": 0.55078125, "learning_rate": 8.903039095149475e-05, "loss": 0.1229, "step": 12195 }, { "epoch": 8.934798534798535, "grad_norm": 0.9921875, "learning_rate": 8.902823986194578e-05, "loss": 0.1511, "step": 12196 }, { "epoch": 8.935531135531136, "grad_norm": 1.203125, "learning_rate": 8.902608858779537e-05, "loss": 0.1524, "step": 12197 }, { "epoch": 8.936263736263736, "grad_norm": 1.0625, "learning_rate": 8.902393712905382e-05, "loss": 0.1306, "step": 12198 }, { "epoch": 8.936996336996337, "grad_norm": 1.1015625, "learning_rate": 8.902178548573142e-05, "loss": 0.1343, "step": 12199 }, { "epoch": 8.937728937728938, "grad_norm": 1.2890625, "learning_rate": 8.901963365783852e-05, "loss": 0.113, "step": 12200 }, { "epoch": 8.938461538461539, "grad_norm": 0.88671875, "learning_rate": 8.901748164538542e-05, "loss": 0.1189, "step": 12201 }, { "epoch": 8.93919413919414, "grad_norm": 0.9140625, "learning_rate": 8.901532944838241e-05, "loss": 0.1314, "step": 12202 }, { "epoch": 8.93992673992674, "grad_norm": 1.53125, "learning_rate": 8.901317706683981e-05, "loss": 0.1656, "step": 12203 }, { "epoch": 8.94065934065934, "grad_norm": 1.4921875, "learning_rate": 8.901102450076793e-05, "loss": 0.1518, "step": 12204 }, { "epoch": 8.941391941391942, "grad_norm": 1.7109375, "learning_rate": 8.90088717501771e-05, "loss": 0.1517, "step": 12205 }, { "epoch": 8.942124542124542, "grad_norm": 0.78515625, "learning_rate": 8.90067188150776e-05, "loss": 0.1311, "step": 12206 }, { "epoch": 8.942857142857143, "grad_norm": 1.59375, "learning_rate": 8.90045656954798e-05, "loss": 0.1598, "step": 12207 }, { "epoch": 8.943589743589744, "grad_norm": 0.94140625, "learning_rate": 8.900241239139397e-05, "loss": 0.1637, "step": 12208 }, { "epoch": 8.944322344322345, "grad_norm": 0.9609375, "learning_rate": 8.900025890283044e-05, "loss": 0.1226, "step": 12209 }, { "epoch": 8.945054945054945, "grad_norm": 1.15625, "learning_rate": 8.899810522979956e-05, "loss": 0.1544, "step": 12210 }, { "epoch": 8.945787545787546, "grad_norm": 1.1328125, "learning_rate": 8.899595137231159e-05, "loss": 0.1359, "step": 12211 }, { "epoch": 8.946520146520147, "grad_norm": 0.8515625, "learning_rate": 8.89937973303769e-05, "loss": 0.1285, "step": 12212 }, { "epoch": 8.947252747252747, "grad_norm": 1.1015625, "learning_rate": 8.899164310400576e-05, "loss": 0.1519, "step": 12213 }, { "epoch": 8.947985347985348, "grad_norm": 0.8984375, "learning_rate": 8.898948869320854e-05, "loss": 0.153, "step": 12214 }, { "epoch": 8.948717948717949, "grad_norm": 0.71484375, "learning_rate": 8.898733409799556e-05, "loss": 0.1359, "step": 12215 }, { "epoch": 8.94945054945055, "grad_norm": 1.0625, "learning_rate": 8.898517931837711e-05, "loss": 0.0931, "step": 12216 }, { "epoch": 8.95018315018315, "grad_norm": 1.0859375, "learning_rate": 8.898302435436354e-05, "loss": 0.1593, "step": 12217 }, { "epoch": 8.950915750915751, "grad_norm": 0.8359375, "learning_rate": 8.898086920596516e-05, "loss": 0.1356, "step": 12218 }, { "epoch": 8.951648351648352, "grad_norm": 1.3359375, "learning_rate": 8.89787138731923e-05, "loss": 0.1991, "step": 12219 }, { "epoch": 8.952380952380953, "grad_norm": 1.1953125, "learning_rate": 8.897655835605529e-05, "loss": 0.1429, "step": 12220 }, { "epoch": 8.953113553113553, "grad_norm": 0.921875, "learning_rate": 8.897440265456445e-05, "loss": 0.1415, "step": 12221 }, { "epoch": 8.953846153846154, "grad_norm": 0.79296875, "learning_rate": 8.897224676873014e-05, "loss": 0.1028, "step": 12222 }, { "epoch": 8.954578754578755, "grad_norm": 1.359375, "learning_rate": 8.897009069856266e-05, "loss": 0.1536, "step": 12223 }, { "epoch": 8.955311355311355, "grad_norm": 0.9140625, "learning_rate": 8.896793444407235e-05, "loss": 0.1384, "step": 12224 }, { "epoch": 8.956043956043956, "grad_norm": 0.88671875, "learning_rate": 8.896577800526951e-05, "loss": 0.1119, "step": 12225 }, { "epoch": 8.956776556776557, "grad_norm": 1.046875, "learning_rate": 8.896362138216452e-05, "loss": 0.1241, "step": 12226 }, { "epoch": 8.957509157509158, "grad_norm": 1.1640625, "learning_rate": 8.896146457476771e-05, "loss": 0.1901, "step": 12227 }, { "epoch": 8.958241758241758, "grad_norm": 1.078125, "learning_rate": 8.895930758308938e-05, "loss": 0.1464, "step": 12228 }, { "epoch": 8.95897435897436, "grad_norm": 0.8671875, "learning_rate": 8.895715040713986e-05, "loss": 0.1277, "step": 12229 }, { "epoch": 8.95970695970696, "grad_norm": 0.89453125, "learning_rate": 8.895499304692953e-05, "loss": 0.1278, "step": 12230 }, { "epoch": 8.96043956043956, "grad_norm": 0.95703125, "learning_rate": 8.895283550246871e-05, "loss": 0.1225, "step": 12231 }, { "epoch": 8.961172161172161, "grad_norm": 1.0625, "learning_rate": 8.895067777376773e-05, "loss": 0.1115, "step": 12232 }, { "epoch": 8.961904761904762, "grad_norm": 1.7109375, "learning_rate": 8.894851986083692e-05, "loss": 0.1328, "step": 12233 }, { "epoch": 8.962637362637363, "grad_norm": 1.2890625, "learning_rate": 8.894636176368663e-05, "loss": 0.1474, "step": 12234 }, { "epoch": 8.963369963369964, "grad_norm": 1.40625, "learning_rate": 8.894420348232723e-05, "loss": 0.12, "step": 12235 }, { "epoch": 8.964102564102564, "grad_norm": 1.6328125, "learning_rate": 8.894204501676899e-05, "loss": 0.1116, "step": 12236 }, { "epoch": 8.964835164835165, "grad_norm": 1.6328125, "learning_rate": 8.893988636702231e-05, "loss": 0.1481, "step": 12237 }, { "epoch": 8.965567765567766, "grad_norm": 1.125, "learning_rate": 8.893772753309752e-05, "loss": 0.1323, "step": 12238 }, { "epoch": 8.966300366300366, "grad_norm": 1.0078125, "learning_rate": 8.893556851500496e-05, "loss": 0.158, "step": 12239 }, { "epoch": 8.967032967032967, "grad_norm": 0.828125, "learning_rate": 8.893340931275495e-05, "loss": 0.1484, "step": 12240 }, { "epoch": 8.967765567765568, "grad_norm": 0.7265625, "learning_rate": 8.893124992635788e-05, "loss": 0.1122, "step": 12241 }, { "epoch": 8.968498168498169, "grad_norm": 1.3125, "learning_rate": 8.892909035582409e-05, "loss": 0.1366, "step": 12242 }, { "epoch": 8.96923076923077, "grad_norm": 1.6796875, "learning_rate": 8.892693060116389e-05, "loss": 0.1432, "step": 12243 }, { "epoch": 8.96996336996337, "grad_norm": 1.5546875, "learning_rate": 8.892477066238763e-05, "loss": 0.1051, "step": 12244 }, { "epoch": 8.97069597069597, "grad_norm": 0.9921875, "learning_rate": 8.892261053950571e-05, "loss": 0.1197, "step": 12245 }, { "epoch": 8.971428571428572, "grad_norm": 1.1796875, "learning_rate": 8.892045023252845e-05, "loss": 0.1264, "step": 12246 }, { "epoch": 8.972161172161172, "grad_norm": 1.1328125, "learning_rate": 8.89182897414662e-05, "loss": 0.1826, "step": 12247 }, { "epoch": 8.972893772893773, "grad_norm": 0.88671875, "learning_rate": 8.891612906632931e-05, "loss": 0.1077, "step": 12248 }, { "epoch": 8.973626373626374, "grad_norm": 1.1171875, "learning_rate": 8.891396820712813e-05, "loss": 0.1729, "step": 12249 }, { "epoch": 8.974358974358974, "grad_norm": 0.79296875, "learning_rate": 8.891180716387302e-05, "loss": 0.1564, "step": 12250 }, { "epoch": 8.975091575091575, "grad_norm": 0.9453125, "learning_rate": 8.890964593657433e-05, "loss": 0.1379, "step": 12251 }, { "epoch": 8.975824175824176, "grad_norm": 1.0078125, "learning_rate": 8.890748452524242e-05, "loss": 0.1651, "step": 12252 }, { "epoch": 8.976556776556777, "grad_norm": 1.2578125, "learning_rate": 8.890532292988764e-05, "loss": 0.1335, "step": 12253 }, { "epoch": 8.977289377289377, "grad_norm": 0.953125, "learning_rate": 8.890316115052034e-05, "loss": 0.1618, "step": 12254 }, { "epoch": 8.978021978021978, "grad_norm": 1.203125, "learning_rate": 8.890099918715091e-05, "loss": 0.145, "step": 12255 }, { "epoch": 8.978754578754579, "grad_norm": 0.96484375, "learning_rate": 8.889883703978967e-05, "loss": 0.1207, "step": 12256 }, { "epoch": 8.97948717948718, "grad_norm": 0.98046875, "learning_rate": 8.8896674708447e-05, "loss": 0.1274, "step": 12257 }, { "epoch": 8.98021978021978, "grad_norm": 0.890625, "learning_rate": 8.889451219313326e-05, "loss": 0.1443, "step": 12258 }, { "epoch": 8.980952380952381, "grad_norm": 1.2265625, "learning_rate": 8.889234949385881e-05, "loss": 0.1477, "step": 12259 }, { "epoch": 8.981684981684982, "grad_norm": 1.296875, "learning_rate": 8.889018661063401e-05, "loss": 0.1801, "step": 12260 }, { "epoch": 8.982417582417582, "grad_norm": 1.28125, "learning_rate": 8.888802354346921e-05, "loss": 0.1432, "step": 12261 }, { "epoch": 8.983150183150183, "grad_norm": 0.95703125, "learning_rate": 8.888586029237481e-05, "loss": 0.1242, "step": 12262 }, { "epoch": 8.983882783882784, "grad_norm": 1.015625, "learning_rate": 8.888369685736115e-05, "loss": 0.123, "step": 12263 }, { "epoch": 8.984615384615385, "grad_norm": 0.9140625, "learning_rate": 8.888153323843857e-05, "loss": 0.1493, "step": 12264 }, { "epoch": 8.985347985347985, "grad_norm": 1.296875, "learning_rate": 8.887936943561749e-05, "loss": 0.1231, "step": 12265 }, { "epoch": 8.986080586080586, "grad_norm": 1.2734375, "learning_rate": 8.887720544890823e-05, "loss": 0.1238, "step": 12266 }, { "epoch": 8.986813186813187, "grad_norm": 1.1796875, "learning_rate": 8.887504127832121e-05, "loss": 0.1876, "step": 12267 }, { "epoch": 8.987545787545788, "grad_norm": 0.86328125, "learning_rate": 8.887287692386675e-05, "loss": 0.1292, "step": 12268 }, { "epoch": 8.988278388278388, "grad_norm": 1.2265625, "learning_rate": 8.887071238555524e-05, "loss": 0.1375, "step": 12269 }, { "epoch": 8.989010989010989, "grad_norm": 1.6015625, "learning_rate": 8.886854766339706e-05, "loss": 0.173, "step": 12270 }, { "epoch": 8.98974358974359, "grad_norm": 1.109375, "learning_rate": 8.886638275740255e-05, "loss": 0.1206, "step": 12271 }, { "epoch": 8.99047619047619, "grad_norm": 0.90234375, "learning_rate": 8.886421766758213e-05, "loss": 0.1644, "step": 12272 }, { "epoch": 8.991208791208791, "grad_norm": 0.73828125, "learning_rate": 8.886205239394615e-05, "loss": 0.1231, "step": 12273 }, { "epoch": 8.991941391941392, "grad_norm": 1.125, "learning_rate": 8.885988693650496e-05, "loss": 0.1436, "step": 12274 }, { "epoch": 8.992673992673993, "grad_norm": 1.03125, "learning_rate": 8.885772129526899e-05, "loss": 0.1519, "step": 12275 }, { "epoch": 8.993406593406593, "grad_norm": 1.140625, "learning_rate": 8.885555547024856e-05, "loss": 0.169, "step": 12276 }, { "epoch": 8.994139194139194, "grad_norm": 0.89453125, "learning_rate": 8.885338946145408e-05, "loss": 0.1467, "step": 12277 }, { "epoch": 8.994871794871795, "grad_norm": 1.125, "learning_rate": 8.885122326889593e-05, "loss": 0.1418, "step": 12278 }, { "epoch": 8.995604395604396, "grad_norm": 0.81640625, "learning_rate": 8.884905689258446e-05, "loss": 0.1106, "step": 12279 }, { "epoch": 8.996336996336996, "grad_norm": 1.5234375, "learning_rate": 8.88468903325301e-05, "loss": 0.1868, "step": 12280 }, { "epoch": 8.997069597069597, "grad_norm": 0.60546875, "learning_rate": 8.884472358874315e-05, "loss": 0.134, "step": 12281 }, { "epoch": 8.997802197802198, "grad_norm": 0.88671875, "learning_rate": 8.884255666123409e-05, "loss": 0.1497, "step": 12282 }, { "epoch": 8.998534798534799, "grad_norm": 0.87890625, "learning_rate": 8.884038955001323e-05, "loss": 0.1314, "step": 12283 }, { "epoch": 8.9992673992674, "grad_norm": 0.6328125, "learning_rate": 8.883822225509099e-05, "loss": 0.1381, "step": 12284 }, { "epoch": 9.0, "grad_norm": 0.96875, "learning_rate": 8.883605477647773e-05, "loss": 0.1657, "step": 12285 }, { "epoch": 9.0007326007326, "grad_norm": 1.0, "learning_rate": 8.883388711418385e-05, "loss": 0.1206, "step": 12286 }, { "epoch": 9.001465201465201, "grad_norm": 1.25, "learning_rate": 8.883171926821974e-05, "loss": 0.1158, "step": 12287 }, { "epoch": 9.002197802197802, "grad_norm": 1.1171875, "learning_rate": 8.882955123859577e-05, "loss": 0.134, "step": 12288 }, { "epoch": 9.002930402930403, "grad_norm": 1.3203125, "learning_rate": 8.882738302532234e-05, "loss": 0.1448, "step": 12289 }, { "epoch": 9.003663003663004, "grad_norm": 0.78125, "learning_rate": 8.882521462840983e-05, "loss": 0.13, "step": 12290 }, { "epoch": 9.004395604395604, "grad_norm": 1.2265625, "learning_rate": 8.882304604786866e-05, "loss": 0.1059, "step": 12291 }, { "epoch": 9.005128205128205, "grad_norm": 0.9453125, "learning_rate": 8.882087728370917e-05, "loss": 0.1551, "step": 12292 }, { "epoch": 9.005860805860806, "grad_norm": 1.140625, "learning_rate": 8.881870833594178e-05, "loss": 0.161, "step": 12293 }, { "epoch": 9.006593406593407, "grad_norm": 1.15625, "learning_rate": 8.88165392045769e-05, "loss": 0.1146, "step": 12294 }, { "epoch": 9.007326007326007, "grad_norm": 1.546875, "learning_rate": 8.881436988962488e-05, "loss": 0.1561, "step": 12295 }, { "epoch": 9.008058608058608, "grad_norm": 1.296875, "learning_rate": 8.881220039109615e-05, "loss": 0.1447, "step": 12296 }, { "epoch": 9.008791208791209, "grad_norm": 1.1875, "learning_rate": 8.881003070900107e-05, "loss": 0.1483, "step": 12297 }, { "epoch": 9.00952380952381, "grad_norm": 0.91796875, "learning_rate": 8.880786084335007e-05, "loss": 0.1053, "step": 12298 }, { "epoch": 9.01025641025641, "grad_norm": 0.96875, "learning_rate": 8.880569079415354e-05, "loss": 0.1305, "step": 12299 }, { "epoch": 9.010989010989011, "grad_norm": 1.265625, "learning_rate": 8.880352056142187e-05, "loss": 0.1299, "step": 12300 }, { "epoch": 9.011721611721612, "grad_norm": 0.62890625, "learning_rate": 8.880135014516544e-05, "loss": 0.1214, "step": 12301 }, { "epoch": 9.012454212454212, "grad_norm": 0.7890625, "learning_rate": 8.879917954539468e-05, "loss": 0.1199, "step": 12302 }, { "epoch": 9.013186813186813, "grad_norm": 1.515625, "learning_rate": 8.879700876211999e-05, "loss": 0.142, "step": 12303 }, { "epoch": 9.013919413919414, "grad_norm": 0.890625, "learning_rate": 8.879483779535173e-05, "loss": 0.1174, "step": 12304 }, { "epoch": 9.014652014652015, "grad_norm": 1.09375, "learning_rate": 8.879266664510036e-05, "loss": 0.1548, "step": 12305 }, { "epoch": 9.015384615384615, "grad_norm": 1.125, "learning_rate": 8.879049531137623e-05, "loss": 0.1273, "step": 12306 }, { "epoch": 9.016117216117216, "grad_norm": 0.9296875, "learning_rate": 8.878832379418979e-05, "loss": 0.1356, "step": 12307 }, { "epoch": 9.016849816849817, "grad_norm": 1.2265625, "learning_rate": 8.87861520935514e-05, "loss": 0.184, "step": 12308 }, { "epoch": 9.017582417582418, "grad_norm": 1.2421875, "learning_rate": 8.87839802094715e-05, "loss": 0.1528, "step": 12309 }, { "epoch": 9.018315018315018, "grad_norm": 0.953125, "learning_rate": 8.878180814196047e-05, "loss": 0.1766, "step": 12310 }, { "epoch": 9.019047619047619, "grad_norm": 1.140625, "learning_rate": 8.877963589102876e-05, "loss": 0.1333, "step": 12311 }, { "epoch": 9.01978021978022, "grad_norm": 0.8203125, "learning_rate": 8.877746345668673e-05, "loss": 0.1371, "step": 12312 }, { "epoch": 9.02051282051282, "grad_norm": 0.79296875, "learning_rate": 8.877529083894481e-05, "loss": 0.1103, "step": 12313 }, { "epoch": 9.021245421245421, "grad_norm": 1.1953125, "learning_rate": 8.877311803781338e-05, "loss": 0.1463, "step": 12314 }, { "epoch": 9.021978021978022, "grad_norm": 0.890625, "learning_rate": 8.87709450533029e-05, "loss": 0.1233, "step": 12315 }, { "epoch": 9.022710622710623, "grad_norm": 0.85546875, "learning_rate": 8.876877188542376e-05, "loss": 0.1327, "step": 12316 }, { "epoch": 9.023443223443223, "grad_norm": 0.8828125, "learning_rate": 8.876659853418638e-05, "loss": 0.1366, "step": 12317 }, { "epoch": 9.024175824175824, "grad_norm": 0.84765625, "learning_rate": 8.876442499960114e-05, "loss": 0.1265, "step": 12318 }, { "epoch": 9.024908424908425, "grad_norm": 1.28125, "learning_rate": 8.87622512816785e-05, "loss": 0.1648, "step": 12319 }, { "epoch": 9.025641025641026, "grad_norm": 0.94140625, "learning_rate": 8.876007738042884e-05, "loss": 0.1234, "step": 12320 }, { "epoch": 9.026373626373626, "grad_norm": 1.0078125, "learning_rate": 8.875790329586259e-05, "loss": 0.1337, "step": 12321 }, { "epoch": 9.027106227106227, "grad_norm": 0.70703125, "learning_rate": 8.875572902799017e-05, "loss": 0.1226, "step": 12322 }, { "epoch": 9.027838827838828, "grad_norm": 0.98828125, "learning_rate": 8.8753554576822e-05, "loss": 0.1298, "step": 12323 }, { "epoch": 9.028571428571428, "grad_norm": 1.4609375, "learning_rate": 8.875137994236848e-05, "loss": 0.1203, "step": 12324 }, { "epoch": 9.02930402930403, "grad_norm": 0.87890625, "learning_rate": 8.874920512464004e-05, "loss": 0.1179, "step": 12325 }, { "epoch": 9.03003663003663, "grad_norm": 1.5703125, "learning_rate": 8.874703012364712e-05, "loss": 0.1522, "step": 12326 }, { "epoch": 9.03076923076923, "grad_norm": 0.9921875, "learning_rate": 8.874485493940011e-05, "loss": 0.1358, "step": 12327 }, { "epoch": 9.031501831501831, "grad_norm": 0.82421875, "learning_rate": 8.874267957190945e-05, "loss": 0.135, "step": 12328 }, { "epoch": 9.032234432234432, "grad_norm": 0.9140625, "learning_rate": 8.874050402118554e-05, "loss": 0.1542, "step": 12329 }, { "epoch": 9.032967032967033, "grad_norm": 0.84375, "learning_rate": 8.873832828723885e-05, "loss": 0.1217, "step": 12330 }, { "epoch": 9.033699633699634, "grad_norm": 1.46875, "learning_rate": 8.873615237007975e-05, "loss": 0.1637, "step": 12331 }, { "epoch": 9.034432234432234, "grad_norm": 1.390625, "learning_rate": 8.87339762697187e-05, "loss": 0.1198, "step": 12332 }, { "epoch": 9.035164835164835, "grad_norm": 1.2421875, "learning_rate": 8.873179998616614e-05, "loss": 0.1497, "step": 12333 }, { "epoch": 9.035897435897436, "grad_norm": 0.953125, "learning_rate": 8.872962351943245e-05, "loss": 0.1354, "step": 12334 }, { "epoch": 9.036630036630036, "grad_norm": 1.2265625, "learning_rate": 8.87274468695281e-05, "loss": 0.1413, "step": 12335 }, { "epoch": 9.037362637362637, "grad_norm": 1.6015625, "learning_rate": 8.872527003646349e-05, "loss": 0.1194, "step": 12336 }, { "epoch": 9.038095238095238, "grad_norm": 0.9921875, "learning_rate": 8.872309302024907e-05, "loss": 0.1256, "step": 12337 }, { "epoch": 9.038827838827839, "grad_norm": 0.70703125, "learning_rate": 8.872091582089527e-05, "loss": 0.1179, "step": 12338 }, { "epoch": 9.03956043956044, "grad_norm": 1.2421875, "learning_rate": 8.87187384384125e-05, "loss": 0.1767, "step": 12339 }, { "epoch": 9.04029304029304, "grad_norm": 1.015625, "learning_rate": 8.871656087281123e-05, "loss": 0.1346, "step": 12340 }, { "epoch": 9.04102564102564, "grad_norm": 0.95703125, "learning_rate": 8.871438312410184e-05, "loss": 0.1616, "step": 12341 }, { "epoch": 9.041758241758242, "grad_norm": 0.69921875, "learning_rate": 8.871220519229482e-05, "loss": 0.1192, "step": 12342 }, { "epoch": 9.042490842490842, "grad_norm": 0.87890625, "learning_rate": 8.871002707740057e-05, "loss": 0.1259, "step": 12343 }, { "epoch": 9.043223443223443, "grad_norm": 1.0234375, "learning_rate": 8.870784877942954e-05, "loss": 0.1254, "step": 12344 }, { "epoch": 9.043956043956044, "grad_norm": 0.95703125, "learning_rate": 8.870567029839216e-05, "loss": 0.115, "step": 12345 }, { "epoch": 9.044688644688645, "grad_norm": 0.80859375, "learning_rate": 8.870349163429887e-05, "loss": 0.1129, "step": 12346 }, { "epoch": 9.045421245421245, "grad_norm": 0.9921875, "learning_rate": 8.870131278716012e-05, "loss": 0.1635, "step": 12347 }, { "epoch": 9.046153846153846, "grad_norm": 1.2578125, "learning_rate": 8.869913375698633e-05, "loss": 0.1148, "step": 12348 }, { "epoch": 9.046886446886447, "grad_norm": 1.296875, "learning_rate": 8.869695454378796e-05, "loss": 0.1133, "step": 12349 }, { "epoch": 9.047619047619047, "grad_norm": 1.015625, "learning_rate": 8.869477514757543e-05, "loss": 0.1181, "step": 12350 }, { "epoch": 9.048351648351648, "grad_norm": 1.1796875, "learning_rate": 8.869259556835921e-05, "loss": 0.1357, "step": 12351 }, { "epoch": 9.049084249084249, "grad_norm": 0.91796875, "learning_rate": 8.86904158061497e-05, "loss": 0.124, "step": 12352 }, { "epoch": 9.04981684981685, "grad_norm": 1.984375, "learning_rate": 8.86882358609574e-05, "loss": 0.1611, "step": 12353 }, { "epoch": 9.05054945054945, "grad_norm": 0.82421875, "learning_rate": 8.86860557327927e-05, "loss": 0.1204, "step": 12354 }, { "epoch": 9.051282051282051, "grad_norm": 0.92578125, "learning_rate": 8.868387542166608e-05, "loss": 0.1367, "step": 12355 }, { "epoch": 9.052014652014652, "grad_norm": 1.3046875, "learning_rate": 8.868169492758798e-05, "loss": 0.1299, "step": 12356 }, { "epoch": 9.052747252747253, "grad_norm": 0.6796875, "learning_rate": 8.867951425056885e-05, "loss": 0.1507, "step": 12357 }, { "epoch": 9.053479853479853, "grad_norm": 1.234375, "learning_rate": 8.867733339061914e-05, "loss": 0.1275, "step": 12358 }, { "epoch": 9.054212454212454, "grad_norm": 0.8671875, "learning_rate": 8.867515234774928e-05, "loss": 0.1339, "step": 12359 }, { "epoch": 9.054945054945055, "grad_norm": 0.96484375, "learning_rate": 8.867297112196974e-05, "loss": 0.1307, "step": 12360 }, { "epoch": 9.055677655677655, "grad_norm": 0.671875, "learning_rate": 8.867078971329095e-05, "loss": 0.1152, "step": 12361 }, { "epoch": 9.056410256410256, "grad_norm": 1.2734375, "learning_rate": 8.86686081217234e-05, "loss": 0.1156, "step": 12362 }, { "epoch": 9.057142857142857, "grad_norm": 0.875, "learning_rate": 8.86664263472775e-05, "loss": 0.1125, "step": 12363 }, { "epoch": 9.057875457875458, "grad_norm": 0.79296875, "learning_rate": 8.866424438996373e-05, "loss": 0.1179, "step": 12364 }, { "epoch": 9.058608058608058, "grad_norm": 1.125, "learning_rate": 8.866206224979252e-05, "loss": 0.143, "step": 12365 }, { "epoch": 9.05934065934066, "grad_norm": 1.3671875, "learning_rate": 8.865987992677437e-05, "loss": 0.1577, "step": 12366 }, { "epoch": 9.06007326007326, "grad_norm": 1.203125, "learning_rate": 8.865769742091968e-05, "loss": 0.1334, "step": 12367 }, { "epoch": 9.06080586080586, "grad_norm": 0.9453125, "learning_rate": 8.865551473223897e-05, "loss": 0.1407, "step": 12368 }, { "epoch": 9.061538461538461, "grad_norm": 0.75390625, "learning_rate": 8.865333186074264e-05, "loss": 0.1278, "step": 12369 }, { "epoch": 9.062271062271062, "grad_norm": 0.87890625, "learning_rate": 8.865114880644117e-05, "loss": 0.1178, "step": 12370 }, { "epoch": 9.063003663003663, "grad_norm": 0.94140625, "learning_rate": 8.864896556934502e-05, "loss": 0.1481, "step": 12371 }, { "epoch": 9.063736263736264, "grad_norm": 1.421875, "learning_rate": 8.864678214946465e-05, "loss": 0.0924, "step": 12372 }, { "epoch": 9.064468864468864, "grad_norm": 0.9140625, "learning_rate": 8.864459854681054e-05, "loss": 0.1198, "step": 12373 }, { "epoch": 9.065201465201465, "grad_norm": 0.8203125, "learning_rate": 8.864241476139313e-05, "loss": 0.1463, "step": 12374 }, { "epoch": 9.065934065934066, "grad_norm": 1.1953125, "learning_rate": 8.86402307932229e-05, "loss": 0.1633, "step": 12375 }, { "epoch": 9.066666666666666, "grad_norm": 0.89453125, "learning_rate": 8.86380466423103e-05, "loss": 0.1201, "step": 12376 }, { "epoch": 9.067399267399267, "grad_norm": 1.171875, "learning_rate": 8.863586230866579e-05, "loss": 0.1713, "step": 12377 }, { "epoch": 9.068131868131868, "grad_norm": 0.94140625, "learning_rate": 8.863367779229984e-05, "loss": 0.1159, "step": 12378 }, { "epoch": 9.068864468864469, "grad_norm": 1.265625, "learning_rate": 8.863149309322292e-05, "loss": 0.1402, "step": 12379 }, { "epoch": 9.06959706959707, "grad_norm": 1.09375, "learning_rate": 8.862930821144551e-05, "loss": 0.162, "step": 12380 }, { "epoch": 9.07032967032967, "grad_norm": 1.2265625, "learning_rate": 8.862712314697807e-05, "loss": 0.1488, "step": 12381 }, { "epoch": 9.07106227106227, "grad_norm": 1.0546875, "learning_rate": 8.862493789983105e-05, "loss": 0.1382, "step": 12382 }, { "epoch": 9.071794871794872, "grad_norm": 1.6171875, "learning_rate": 8.862275247001495e-05, "loss": 0.141, "step": 12383 }, { "epoch": 9.072527472527472, "grad_norm": 0.58984375, "learning_rate": 8.862056685754024e-05, "loss": 0.1277, "step": 12384 }, { "epoch": 9.073260073260073, "grad_norm": 0.8984375, "learning_rate": 8.861838106241735e-05, "loss": 0.1333, "step": 12385 }, { "epoch": 9.073992673992674, "grad_norm": 1.4375, "learning_rate": 8.861619508465681e-05, "loss": 0.1259, "step": 12386 }, { "epoch": 9.074725274725274, "grad_norm": 1.28125, "learning_rate": 8.861400892426904e-05, "loss": 0.1286, "step": 12387 }, { "epoch": 9.075457875457875, "grad_norm": 1.703125, "learning_rate": 8.861182258126456e-05, "loss": 0.1932, "step": 12388 }, { "epoch": 9.076190476190476, "grad_norm": 0.90625, "learning_rate": 8.860963605565382e-05, "loss": 0.1182, "step": 12389 }, { "epoch": 9.076923076923077, "grad_norm": 0.73046875, "learning_rate": 8.86074493474473e-05, "loss": 0.155, "step": 12390 }, { "epoch": 9.077655677655677, "grad_norm": 1.0859375, "learning_rate": 8.860526245665549e-05, "loss": 0.1336, "step": 12391 }, { "epoch": 9.078388278388278, "grad_norm": 1.4765625, "learning_rate": 8.860307538328885e-05, "loss": 0.1344, "step": 12392 }, { "epoch": 9.079120879120879, "grad_norm": 1.5390625, "learning_rate": 8.860088812735787e-05, "loss": 0.1546, "step": 12393 }, { "epoch": 9.07985347985348, "grad_norm": 1.2890625, "learning_rate": 8.859870068887303e-05, "loss": 0.138, "step": 12394 }, { "epoch": 9.08058608058608, "grad_norm": 1.40625, "learning_rate": 8.85965130678448e-05, "loss": 0.1188, "step": 12395 }, { "epoch": 9.081318681318681, "grad_norm": 1.3984375, "learning_rate": 8.85943252642837e-05, "loss": 0.1436, "step": 12396 }, { "epoch": 9.082051282051282, "grad_norm": 1.1171875, "learning_rate": 8.859213727820014e-05, "loss": 0.1422, "step": 12397 }, { "epoch": 9.082783882783882, "grad_norm": 1.28125, "learning_rate": 8.858994910960466e-05, "loss": 0.1644, "step": 12398 }, { "epoch": 9.083516483516483, "grad_norm": 1.109375, "learning_rate": 8.858776075850774e-05, "loss": 0.1523, "step": 12399 }, { "epoch": 9.084249084249084, "grad_norm": 0.9296875, "learning_rate": 8.858557222491984e-05, "loss": 0.1391, "step": 12400 }, { "epoch": 9.084981684981685, "grad_norm": 0.8203125, "learning_rate": 8.858338350885147e-05, "loss": 0.1213, "step": 12401 }, { "epoch": 9.085714285714285, "grad_norm": 1.4140625, "learning_rate": 8.858119461031312e-05, "loss": 0.147, "step": 12402 }, { "epoch": 9.086446886446886, "grad_norm": 1.328125, "learning_rate": 8.857900552931524e-05, "loss": 0.1705, "step": 12403 }, { "epoch": 9.087179487179487, "grad_norm": 0.96484375, "learning_rate": 8.857681626586836e-05, "loss": 0.1267, "step": 12404 }, { "epoch": 9.087912087912088, "grad_norm": 0.84375, "learning_rate": 8.857462681998296e-05, "loss": 0.1253, "step": 12405 }, { "epoch": 9.088644688644688, "grad_norm": 1.015625, "learning_rate": 8.85724371916695e-05, "loss": 0.1267, "step": 12406 }, { "epoch": 9.089377289377289, "grad_norm": 0.77734375, "learning_rate": 8.857024738093852e-05, "loss": 0.1144, "step": 12407 }, { "epoch": 9.09010989010989, "grad_norm": 0.984375, "learning_rate": 8.856805738780047e-05, "loss": 0.1296, "step": 12408 }, { "epoch": 9.09084249084249, "grad_norm": 0.8515625, "learning_rate": 8.856586721226587e-05, "loss": 0.1123, "step": 12409 }, { "epoch": 9.091575091575091, "grad_norm": 1.3984375, "learning_rate": 8.85636768543452e-05, "loss": 0.1442, "step": 12410 }, { "epoch": 9.092307692307692, "grad_norm": 1.296875, "learning_rate": 8.856148631404896e-05, "loss": 0.1597, "step": 12411 }, { "epoch": 9.093040293040293, "grad_norm": 0.73828125, "learning_rate": 8.855929559138765e-05, "loss": 0.1394, "step": 12412 }, { "epoch": 9.093772893772893, "grad_norm": 1.2890625, "learning_rate": 8.855710468637174e-05, "loss": 0.1427, "step": 12413 }, { "epoch": 9.094505494505494, "grad_norm": 1.375, "learning_rate": 8.855491359901177e-05, "loss": 0.1373, "step": 12414 }, { "epoch": 9.095238095238095, "grad_norm": 0.87890625, "learning_rate": 8.855272232931821e-05, "loss": 0.1232, "step": 12415 }, { "epoch": 9.095970695970696, "grad_norm": 0.91796875, "learning_rate": 8.855053087730156e-05, "loss": 0.122, "step": 12416 }, { "epoch": 9.096703296703296, "grad_norm": 0.9140625, "learning_rate": 8.854833924297233e-05, "loss": 0.1106, "step": 12417 }, { "epoch": 9.097435897435897, "grad_norm": 1.984375, "learning_rate": 8.854614742634101e-05, "loss": 0.1822, "step": 12418 }, { "epoch": 9.098168498168498, "grad_norm": 0.875, "learning_rate": 8.854395542741813e-05, "loss": 0.1159, "step": 12419 }, { "epoch": 9.098901098901099, "grad_norm": 1.03125, "learning_rate": 8.854176324621413e-05, "loss": 0.103, "step": 12420 }, { "epoch": 9.0996336996337, "grad_norm": 0.83984375, "learning_rate": 8.853957088273959e-05, "loss": 0.1578, "step": 12421 }, { "epoch": 9.1003663003663, "grad_norm": 1.15625, "learning_rate": 8.853737833700496e-05, "loss": 0.1065, "step": 12422 }, { "epoch": 9.1010989010989, "grad_norm": 1.5859375, "learning_rate": 8.853518560902078e-05, "loss": 0.1767, "step": 12423 }, { "epoch": 9.101831501831501, "grad_norm": 0.88671875, "learning_rate": 8.853299269879753e-05, "loss": 0.1213, "step": 12424 }, { "epoch": 9.102564102564102, "grad_norm": 0.6953125, "learning_rate": 8.853079960634573e-05, "loss": 0.1057, "step": 12425 }, { "epoch": 9.103296703296703, "grad_norm": 1.3125, "learning_rate": 8.852860633167588e-05, "loss": 0.1375, "step": 12426 }, { "epoch": 9.104029304029304, "grad_norm": 1.0390625, "learning_rate": 8.852641287479849e-05, "loss": 0.1347, "step": 12427 }, { "epoch": 9.104761904761904, "grad_norm": 0.734375, "learning_rate": 8.852421923572407e-05, "loss": 0.1337, "step": 12428 }, { "epoch": 9.105494505494505, "grad_norm": 1.234375, "learning_rate": 8.852202541446315e-05, "loss": 0.1513, "step": 12429 }, { "epoch": 9.106227106227106, "grad_norm": 1.3203125, "learning_rate": 8.85198314110262e-05, "loss": 0.1236, "step": 12430 }, { "epoch": 9.106959706959707, "grad_norm": 0.95703125, "learning_rate": 8.851763722542377e-05, "loss": 0.1422, "step": 12431 }, { "epoch": 9.107692307692307, "grad_norm": 1.3125, "learning_rate": 8.851544285766637e-05, "loss": 0.1277, "step": 12432 }, { "epoch": 9.108424908424908, "grad_norm": 0.63671875, "learning_rate": 8.851324830776448e-05, "loss": 0.1324, "step": 12433 }, { "epoch": 9.109157509157509, "grad_norm": 1.09375, "learning_rate": 8.851105357572866e-05, "loss": 0.123, "step": 12434 }, { "epoch": 9.10989010989011, "grad_norm": 1.0078125, "learning_rate": 8.85088586615694e-05, "loss": 0.146, "step": 12435 }, { "epoch": 9.11062271062271, "grad_norm": 1.1328125, "learning_rate": 8.850666356529722e-05, "loss": 0.126, "step": 12436 }, { "epoch": 9.111355311355311, "grad_norm": 1.078125, "learning_rate": 8.850446828692263e-05, "loss": 0.137, "step": 12437 }, { "epoch": 9.112087912087912, "grad_norm": 1.0625, "learning_rate": 8.850227282645614e-05, "loss": 0.1576, "step": 12438 }, { "epoch": 9.112820512820512, "grad_norm": 1.03125, "learning_rate": 8.850007718390832e-05, "loss": 0.133, "step": 12439 }, { "epoch": 9.113553113553113, "grad_norm": 0.82421875, "learning_rate": 8.849788135928964e-05, "loss": 0.1318, "step": 12440 }, { "epoch": 9.114285714285714, "grad_norm": 1.1015625, "learning_rate": 8.849568535261064e-05, "loss": 0.1092, "step": 12441 }, { "epoch": 9.115018315018315, "grad_norm": 0.984375, "learning_rate": 8.849348916388184e-05, "loss": 0.125, "step": 12442 }, { "epoch": 9.115750915750915, "grad_norm": 0.765625, "learning_rate": 8.849129279311376e-05, "loss": 0.1321, "step": 12443 }, { "epoch": 9.116483516483516, "grad_norm": 1.4375, "learning_rate": 8.848909624031692e-05, "loss": 0.1077, "step": 12444 }, { "epoch": 9.117216117216117, "grad_norm": 0.75390625, "learning_rate": 8.848689950550186e-05, "loss": 0.1233, "step": 12445 }, { "epoch": 9.117948717948718, "grad_norm": 0.8046875, "learning_rate": 8.848470258867908e-05, "loss": 0.1193, "step": 12446 }, { "epoch": 9.118681318681318, "grad_norm": 1.1796875, "learning_rate": 8.848250548985912e-05, "loss": 0.1056, "step": 12447 }, { "epoch": 9.119413919413919, "grad_norm": 0.9921875, "learning_rate": 8.848030820905251e-05, "loss": 0.1558, "step": 12448 }, { "epoch": 9.12014652014652, "grad_norm": 1.1953125, "learning_rate": 8.84781107462698e-05, "loss": 0.1244, "step": 12449 }, { "epoch": 9.12087912087912, "grad_norm": 0.94140625, "learning_rate": 8.847591310152146e-05, "loss": 0.1483, "step": 12450 }, { "epoch": 9.121611721611721, "grad_norm": 1.0703125, "learning_rate": 8.847371527481807e-05, "loss": 0.1621, "step": 12451 }, { "epoch": 9.122344322344322, "grad_norm": 0.71484375, "learning_rate": 8.847151726617014e-05, "loss": 0.1102, "step": 12452 }, { "epoch": 9.123076923076923, "grad_norm": 1.4609375, "learning_rate": 8.84693190755882e-05, "loss": 0.1186, "step": 12453 }, { "epoch": 9.123809523809523, "grad_norm": 1.8671875, "learning_rate": 8.84671207030828e-05, "loss": 0.1732, "step": 12454 }, { "epoch": 9.124542124542124, "grad_norm": 1.375, "learning_rate": 8.846492214866445e-05, "loss": 0.1611, "step": 12455 }, { "epoch": 9.125274725274725, "grad_norm": 1.015625, "learning_rate": 8.84627234123437e-05, "loss": 0.1178, "step": 12456 }, { "epoch": 9.126007326007326, "grad_norm": 1.234375, "learning_rate": 8.846052449413109e-05, "loss": 0.1864, "step": 12457 }, { "epoch": 9.126739926739926, "grad_norm": 0.92578125, "learning_rate": 8.845832539403713e-05, "loss": 0.1354, "step": 12458 }, { "epoch": 9.127472527472527, "grad_norm": 0.98828125, "learning_rate": 8.845612611207237e-05, "loss": 0.1595, "step": 12459 }, { "epoch": 9.128205128205128, "grad_norm": 1.1875, "learning_rate": 8.845392664824737e-05, "loss": 0.18, "step": 12460 }, { "epoch": 9.128937728937728, "grad_norm": 1.671875, "learning_rate": 8.845172700257263e-05, "loss": 0.1655, "step": 12461 }, { "epoch": 9.12967032967033, "grad_norm": 1.1015625, "learning_rate": 8.84495271750587e-05, "loss": 0.0981, "step": 12462 }, { "epoch": 9.13040293040293, "grad_norm": 1.1796875, "learning_rate": 8.844732716571614e-05, "loss": 0.1444, "step": 12463 }, { "epoch": 9.13113553113553, "grad_norm": 1.953125, "learning_rate": 8.844512697455547e-05, "loss": 0.1368, "step": 12464 }, { "epoch": 9.131868131868131, "grad_norm": 1.03125, "learning_rate": 8.844292660158727e-05, "loss": 0.121, "step": 12465 }, { "epoch": 9.132600732600732, "grad_norm": 1.109375, "learning_rate": 8.844072604682202e-05, "loss": 0.1396, "step": 12466 }, { "epoch": 9.133333333333333, "grad_norm": 1.0703125, "learning_rate": 8.843852531027031e-05, "loss": 0.1421, "step": 12467 }, { "epoch": 9.134065934065934, "grad_norm": 1.171875, "learning_rate": 8.843632439194267e-05, "loss": 0.1374, "step": 12468 }, { "epoch": 9.134798534798534, "grad_norm": 2.0, "learning_rate": 8.843412329184963e-05, "loss": 0.1283, "step": 12469 }, { "epoch": 9.135531135531135, "grad_norm": 1.2265625, "learning_rate": 8.843192201000176e-05, "loss": 0.131, "step": 12470 }, { "epoch": 9.136263736263736, "grad_norm": 1.796875, "learning_rate": 8.842972054640961e-05, "loss": 0.1554, "step": 12471 }, { "epoch": 9.136996336996337, "grad_norm": 1.0859375, "learning_rate": 8.842751890108372e-05, "loss": 0.1377, "step": 12472 }, { "epoch": 9.137728937728937, "grad_norm": 1.1484375, "learning_rate": 8.842531707403463e-05, "loss": 0.1484, "step": 12473 }, { "epoch": 9.138461538461538, "grad_norm": 0.6171875, "learning_rate": 8.842311506527289e-05, "loss": 0.1051, "step": 12474 }, { "epoch": 9.139194139194139, "grad_norm": 1.3359375, "learning_rate": 8.842091287480905e-05, "loss": 0.1369, "step": 12475 }, { "epoch": 9.13992673992674, "grad_norm": 1.0, "learning_rate": 8.841871050265369e-05, "loss": 0.1267, "step": 12476 }, { "epoch": 9.14065934065934, "grad_norm": 0.734375, "learning_rate": 8.841650794881733e-05, "loss": 0.117, "step": 12477 }, { "epoch": 9.141391941391941, "grad_norm": 0.80859375, "learning_rate": 8.841430521331053e-05, "loss": 0.1631, "step": 12478 }, { "epoch": 9.142124542124542, "grad_norm": 0.6640625, "learning_rate": 8.841210229614385e-05, "loss": 0.1304, "step": 12479 }, { "epoch": 9.142857142857142, "grad_norm": 1.109375, "learning_rate": 8.840989919732784e-05, "loss": 0.136, "step": 12480 }, { "epoch": 9.143589743589743, "grad_norm": 0.84375, "learning_rate": 8.840769591687307e-05, "loss": 0.1149, "step": 12481 }, { "epoch": 9.144322344322344, "grad_norm": 0.88671875, "learning_rate": 8.840549245479007e-05, "loss": 0.1235, "step": 12482 }, { "epoch": 9.145054945054945, "grad_norm": 1.328125, "learning_rate": 8.840328881108941e-05, "loss": 0.1475, "step": 12483 }, { "epoch": 9.145787545787545, "grad_norm": 0.92578125, "learning_rate": 8.840108498578167e-05, "loss": 0.1163, "step": 12484 }, { "epoch": 9.146520146520146, "grad_norm": 0.5625, "learning_rate": 8.839888097887738e-05, "loss": 0.1085, "step": 12485 }, { "epoch": 9.147252747252747, "grad_norm": 0.921875, "learning_rate": 8.83966767903871e-05, "loss": 0.1509, "step": 12486 }, { "epoch": 9.147985347985347, "grad_norm": 1.390625, "learning_rate": 8.83944724203214e-05, "loss": 0.1466, "step": 12487 }, { "epoch": 9.148717948717948, "grad_norm": 1.3125, "learning_rate": 8.839226786869084e-05, "loss": 0.1195, "step": 12488 }, { "epoch": 9.149450549450549, "grad_norm": 1.03125, "learning_rate": 8.8390063135506e-05, "loss": 0.1416, "step": 12489 }, { "epoch": 9.15018315018315, "grad_norm": 1.2578125, "learning_rate": 8.838785822077742e-05, "loss": 0.1247, "step": 12490 }, { "epoch": 9.15091575091575, "grad_norm": 1.53125, "learning_rate": 8.83856531245157e-05, "loss": 0.1422, "step": 12491 }, { "epoch": 9.151648351648351, "grad_norm": 0.6328125, "learning_rate": 8.838344784673134e-05, "loss": 0.1209, "step": 12492 }, { "epoch": 9.152380952380952, "grad_norm": 0.8203125, "learning_rate": 8.838124238743496e-05, "loss": 0.1083, "step": 12493 }, { "epoch": 9.153113553113553, "grad_norm": 1.5859375, "learning_rate": 8.837903674663712e-05, "loss": 0.1482, "step": 12494 }, { "epoch": 9.153846153846153, "grad_norm": 1.8203125, "learning_rate": 8.837683092434838e-05, "loss": 0.1482, "step": 12495 }, { "epoch": 9.154578754578754, "grad_norm": 0.6796875, "learning_rate": 8.837462492057929e-05, "loss": 0.1257, "step": 12496 }, { "epoch": 9.155311355311355, "grad_norm": 1.6328125, "learning_rate": 8.837241873534047e-05, "loss": 0.1594, "step": 12497 }, { "epoch": 9.156043956043955, "grad_norm": 1.703125, "learning_rate": 8.837021236864243e-05, "loss": 0.1534, "step": 12498 }, { "epoch": 9.156776556776556, "grad_norm": 1.2734375, "learning_rate": 8.836800582049578e-05, "loss": 0.1206, "step": 12499 }, { "epoch": 9.157509157509157, "grad_norm": 1.03125, "learning_rate": 8.836579909091109e-05, "loss": 0.1731, "step": 12500 }, { "epoch": 9.158241758241758, "grad_norm": 0.77734375, "learning_rate": 8.836359217989892e-05, "loss": 0.1384, "step": 12501 }, { "epoch": 9.158974358974358, "grad_norm": 1.125, "learning_rate": 8.836138508746985e-05, "loss": 0.1417, "step": 12502 }, { "epoch": 9.15970695970696, "grad_norm": 0.72265625, "learning_rate": 8.835917781363445e-05, "loss": 0.1189, "step": 12503 }, { "epoch": 9.16043956043956, "grad_norm": 0.70703125, "learning_rate": 8.83569703584033e-05, "loss": 0.1027, "step": 12504 }, { "epoch": 9.16117216117216, "grad_norm": 1.1953125, "learning_rate": 8.835476272178699e-05, "loss": 0.1382, "step": 12505 }, { "epoch": 9.161904761904761, "grad_norm": 1.4453125, "learning_rate": 8.835255490379609e-05, "loss": 0.1296, "step": 12506 }, { "epoch": 9.162637362637362, "grad_norm": 0.8359375, "learning_rate": 8.835034690444113e-05, "loss": 0.1189, "step": 12507 }, { "epoch": 9.163369963369963, "grad_norm": 0.7265625, "learning_rate": 8.834813872373278e-05, "loss": 0.1112, "step": 12508 }, { "epoch": 9.164102564102564, "grad_norm": 1.3203125, "learning_rate": 8.834593036168156e-05, "loss": 0.1533, "step": 12509 }, { "epoch": 9.164835164835164, "grad_norm": 1.34375, "learning_rate": 8.834372181829806e-05, "loss": 0.1356, "step": 12510 }, { "epoch": 9.165567765567765, "grad_norm": 1.5078125, "learning_rate": 8.834151309359287e-05, "loss": 0.1535, "step": 12511 }, { "epoch": 9.166300366300366, "grad_norm": 1.0859375, "learning_rate": 8.833930418757656e-05, "loss": 0.144, "step": 12512 }, { "epoch": 9.167032967032966, "grad_norm": 0.9765625, "learning_rate": 8.833709510025973e-05, "loss": 0.1217, "step": 12513 }, { "epoch": 9.167765567765567, "grad_norm": 1.3359375, "learning_rate": 8.833488583165294e-05, "loss": 0.1353, "step": 12514 }, { "epoch": 9.168498168498168, "grad_norm": 0.73046875, "learning_rate": 8.83326763817668e-05, "loss": 0.139, "step": 12515 }, { "epoch": 9.169230769230769, "grad_norm": 1.453125, "learning_rate": 8.833046675061189e-05, "loss": 0.1199, "step": 12516 }, { "epoch": 9.16996336996337, "grad_norm": 1.25, "learning_rate": 8.832825693819881e-05, "loss": 0.1328, "step": 12517 }, { "epoch": 9.17069597069597, "grad_norm": 0.65625, "learning_rate": 8.832604694453811e-05, "loss": 0.1252, "step": 12518 }, { "epoch": 9.17142857142857, "grad_norm": 1.25, "learning_rate": 8.832383676964042e-05, "loss": 0.1546, "step": 12519 }, { "epoch": 9.172161172161172, "grad_norm": 1.2578125, "learning_rate": 8.832162641351628e-05, "loss": 0.1444, "step": 12520 }, { "epoch": 9.172893772893772, "grad_norm": 0.8046875, "learning_rate": 8.831941587617634e-05, "loss": 0.1198, "step": 12521 }, { "epoch": 9.173626373626373, "grad_norm": 1.4609375, "learning_rate": 8.831720515763116e-05, "loss": 0.164, "step": 12522 }, { "epoch": 9.174358974358974, "grad_norm": 1.21875, "learning_rate": 8.831499425789133e-05, "loss": 0.1589, "step": 12523 }, { "epoch": 9.175091575091574, "grad_norm": 0.89453125, "learning_rate": 8.831278317696746e-05, "loss": 0.1304, "step": 12524 }, { "epoch": 9.175824175824175, "grad_norm": 1.2578125, "learning_rate": 8.831057191487012e-05, "loss": 0.1179, "step": 12525 }, { "epoch": 9.176556776556776, "grad_norm": 0.671875, "learning_rate": 8.830836047160993e-05, "loss": 0.1174, "step": 12526 }, { "epoch": 9.177289377289377, "grad_norm": 1.2109375, "learning_rate": 8.830614884719746e-05, "loss": 0.1437, "step": 12527 }, { "epoch": 9.178021978021977, "grad_norm": 0.87890625, "learning_rate": 8.830393704164332e-05, "loss": 0.1398, "step": 12528 }, { "epoch": 9.178754578754578, "grad_norm": 1.0, "learning_rate": 8.830172505495811e-05, "loss": 0.1251, "step": 12529 }, { "epoch": 9.179487179487179, "grad_norm": 0.76953125, "learning_rate": 8.829951288715243e-05, "loss": 0.1338, "step": 12530 }, { "epoch": 9.18021978021978, "grad_norm": 1.6328125, "learning_rate": 8.829730053823689e-05, "loss": 0.1428, "step": 12531 }, { "epoch": 9.18095238095238, "grad_norm": 1.765625, "learning_rate": 8.829508800822206e-05, "loss": 0.1494, "step": 12532 }, { "epoch": 9.181684981684981, "grad_norm": 0.90625, "learning_rate": 8.829287529711853e-05, "loss": 0.1398, "step": 12533 }, { "epoch": 9.182417582417582, "grad_norm": 0.7578125, "learning_rate": 8.829066240493696e-05, "loss": 0.1376, "step": 12534 }, { "epoch": 9.183150183150182, "grad_norm": 1.4296875, "learning_rate": 8.828844933168791e-05, "loss": 0.1087, "step": 12535 }, { "epoch": 9.183882783882783, "grad_norm": 0.74609375, "learning_rate": 8.8286236077382e-05, "loss": 0.0968, "step": 12536 }, { "epoch": 9.184615384615384, "grad_norm": 0.9140625, "learning_rate": 8.828402264202982e-05, "loss": 0.1385, "step": 12537 }, { "epoch": 9.185347985347985, "grad_norm": 1.0, "learning_rate": 8.828180902564199e-05, "loss": 0.154, "step": 12538 }, { "epoch": 9.186080586080585, "grad_norm": 0.73828125, "learning_rate": 8.827959522822912e-05, "loss": 0.1268, "step": 12539 }, { "epoch": 9.186813186813186, "grad_norm": 1.09375, "learning_rate": 8.827738124980178e-05, "loss": 0.1452, "step": 12540 }, { "epoch": 9.187545787545787, "grad_norm": 0.6171875, "learning_rate": 8.827516709037062e-05, "loss": 0.0999, "step": 12541 }, { "epoch": 9.188278388278388, "grad_norm": 0.89453125, "learning_rate": 8.827295274994624e-05, "loss": 0.1309, "step": 12542 }, { "epoch": 9.189010989010988, "grad_norm": 0.73828125, "learning_rate": 8.827073822853925e-05, "loss": 0.1082, "step": 12543 }, { "epoch": 9.189743589743589, "grad_norm": 0.90234375, "learning_rate": 8.826852352616024e-05, "loss": 0.1256, "step": 12544 }, { "epoch": 9.19047619047619, "grad_norm": 0.85546875, "learning_rate": 8.826630864281982e-05, "loss": 0.1295, "step": 12545 }, { "epoch": 9.19120879120879, "grad_norm": 0.78515625, "learning_rate": 8.826409357852864e-05, "loss": 0.1125, "step": 12546 }, { "epoch": 9.191941391941391, "grad_norm": 0.8984375, "learning_rate": 8.826187833329729e-05, "loss": 0.1582, "step": 12547 }, { "epoch": 9.192673992673992, "grad_norm": 0.82421875, "learning_rate": 8.825966290713638e-05, "loss": 0.1496, "step": 12548 }, { "epoch": 9.193406593406593, "grad_norm": 0.68359375, "learning_rate": 8.825744730005652e-05, "loss": 0.1698, "step": 12549 }, { "epoch": 9.194139194139193, "grad_norm": 1.109375, "learning_rate": 8.825523151206835e-05, "loss": 0.127, "step": 12550 }, { "epoch": 9.194871794871794, "grad_norm": 1.0625, "learning_rate": 8.825301554318247e-05, "loss": 0.1477, "step": 12551 }, { "epoch": 9.195604395604395, "grad_norm": 0.5703125, "learning_rate": 8.825079939340949e-05, "loss": 0.12, "step": 12552 }, { "epoch": 9.196336996336996, "grad_norm": 0.87890625, "learning_rate": 8.824858306276003e-05, "loss": 0.1133, "step": 12553 }, { "epoch": 9.197069597069596, "grad_norm": 0.796875, "learning_rate": 8.824636655124474e-05, "loss": 0.0935, "step": 12554 }, { "epoch": 9.197802197802197, "grad_norm": 0.87890625, "learning_rate": 8.82441498588742e-05, "loss": 0.1288, "step": 12555 }, { "epoch": 9.198534798534798, "grad_norm": 0.9609375, "learning_rate": 8.824193298565905e-05, "loss": 0.1334, "step": 12556 }, { "epoch": 9.199267399267399, "grad_norm": 0.98828125, "learning_rate": 8.82397159316099e-05, "loss": 0.1354, "step": 12557 }, { "epoch": 9.2, "grad_norm": 0.78515625, "learning_rate": 8.82374986967374e-05, "loss": 0.1181, "step": 12558 }, { "epoch": 9.2007326007326, "grad_norm": 1.0078125, "learning_rate": 8.823528128105215e-05, "loss": 0.1729, "step": 12559 }, { "epoch": 9.2014652014652, "grad_norm": 0.8984375, "learning_rate": 8.823306368456479e-05, "loss": 0.1303, "step": 12560 }, { "epoch": 9.202197802197801, "grad_norm": 0.79296875, "learning_rate": 8.823084590728591e-05, "loss": 0.128, "step": 12561 }, { "epoch": 9.202930402930402, "grad_norm": 1.0546875, "learning_rate": 8.822862794922618e-05, "loss": 0.1028, "step": 12562 }, { "epoch": 9.203663003663003, "grad_norm": 0.9375, "learning_rate": 8.82264098103962e-05, "loss": 0.1252, "step": 12563 }, { "epoch": 9.204395604395604, "grad_norm": 0.87890625, "learning_rate": 8.822419149080661e-05, "loss": 0.1479, "step": 12564 }, { "epoch": 9.205128205128204, "grad_norm": 0.7421875, "learning_rate": 8.822197299046803e-05, "loss": 0.1412, "step": 12565 }, { "epoch": 9.205860805860805, "grad_norm": 0.55859375, "learning_rate": 8.82197543093911e-05, "loss": 0.1241, "step": 12566 }, { "epoch": 9.206593406593406, "grad_norm": 1.15625, "learning_rate": 8.821753544758644e-05, "loss": 0.1526, "step": 12567 }, { "epoch": 9.207326007326007, "grad_norm": 1.265625, "learning_rate": 8.821531640506469e-05, "loss": 0.1604, "step": 12568 }, { "epoch": 9.208058608058607, "grad_norm": 0.98828125, "learning_rate": 8.821309718183646e-05, "loss": 0.1282, "step": 12569 }, { "epoch": 9.208791208791208, "grad_norm": 1.0, "learning_rate": 8.821087777791242e-05, "loss": 0.1496, "step": 12570 }, { "epoch": 9.209523809523809, "grad_norm": 0.76953125, "learning_rate": 8.820865819330318e-05, "loss": 0.1193, "step": 12571 }, { "epoch": 9.21025641025641, "grad_norm": 1.1015625, "learning_rate": 8.820643842801938e-05, "loss": 0.1244, "step": 12572 }, { "epoch": 9.21098901098901, "grad_norm": 0.96875, "learning_rate": 8.820421848207167e-05, "loss": 0.12, "step": 12573 }, { "epoch": 9.211721611721611, "grad_norm": 1.4453125, "learning_rate": 8.820199835547065e-05, "loss": 0.1488, "step": 12574 }, { "epoch": 9.212454212454212, "grad_norm": 1.0, "learning_rate": 8.819977804822699e-05, "loss": 0.1231, "step": 12575 }, { "epoch": 9.213186813186812, "grad_norm": 1.0234375, "learning_rate": 8.819755756035132e-05, "loss": 0.1382, "step": 12576 }, { "epoch": 9.213919413919413, "grad_norm": 0.9609375, "learning_rate": 8.819533689185426e-05, "loss": 0.1329, "step": 12577 }, { "epoch": 9.214652014652014, "grad_norm": 0.921875, "learning_rate": 8.819311604274647e-05, "loss": 0.1278, "step": 12578 }, { "epoch": 9.215384615384615, "grad_norm": 1.15625, "learning_rate": 8.81908950130386e-05, "loss": 0.1385, "step": 12579 }, { "epoch": 9.216117216117215, "grad_norm": 0.83203125, "learning_rate": 8.818867380274127e-05, "loss": 0.1228, "step": 12580 }, { "epoch": 9.216849816849816, "grad_norm": 0.92578125, "learning_rate": 8.818645241186513e-05, "loss": 0.1214, "step": 12581 }, { "epoch": 9.217582417582417, "grad_norm": 0.99609375, "learning_rate": 8.818423084042083e-05, "loss": 0.1577, "step": 12582 }, { "epoch": 9.218315018315018, "grad_norm": 1.1328125, "learning_rate": 8.818200908841901e-05, "loss": 0.1258, "step": 12583 }, { "epoch": 9.219047619047618, "grad_norm": 1.28125, "learning_rate": 8.817978715587032e-05, "loss": 0.1132, "step": 12584 }, { "epoch": 9.219780219780219, "grad_norm": 1.0859375, "learning_rate": 8.817756504278538e-05, "loss": 0.1418, "step": 12585 }, { "epoch": 9.22051282051282, "grad_norm": 1.2265625, "learning_rate": 8.817534274917487e-05, "loss": 0.1294, "step": 12586 }, { "epoch": 9.22124542124542, "grad_norm": 0.58203125, "learning_rate": 8.817312027504944e-05, "loss": 0.1112, "step": 12587 }, { "epoch": 9.221978021978021, "grad_norm": 0.69921875, "learning_rate": 8.81708976204197e-05, "loss": 0.119, "step": 12588 }, { "epoch": 9.222710622710622, "grad_norm": 0.921875, "learning_rate": 8.816867478529634e-05, "loss": 0.1209, "step": 12589 }, { "epoch": 9.223443223443223, "grad_norm": 0.61328125, "learning_rate": 8.816645176968997e-05, "loss": 0.1134, "step": 12590 }, { "epoch": 9.224175824175823, "grad_norm": 1.140625, "learning_rate": 8.816422857361127e-05, "loss": 0.1493, "step": 12591 }, { "epoch": 9.224908424908424, "grad_norm": 1.3515625, "learning_rate": 8.816200519707091e-05, "loss": 0.1199, "step": 12592 }, { "epoch": 9.225641025641025, "grad_norm": 1.1953125, "learning_rate": 8.81597816400795e-05, "loss": 0.1502, "step": 12593 }, { "epoch": 9.226373626373626, "grad_norm": 0.81640625, "learning_rate": 8.815755790264772e-05, "loss": 0.0836, "step": 12594 }, { "epoch": 9.227106227106226, "grad_norm": 1.0703125, "learning_rate": 8.815533398478622e-05, "loss": 0.1417, "step": 12595 }, { "epoch": 9.227838827838827, "grad_norm": 1.6796875, "learning_rate": 8.815310988650565e-05, "loss": 0.1657, "step": 12596 }, { "epoch": 9.228571428571428, "grad_norm": 1.4140625, "learning_rate": 8.815088560781666e-05, "loss": 0.1465, "step": 12597 }, { "epoch": 9.229304029304028, "grad_norm": 0.9296875, "learning_rate": 8.814866114872993e-05, "loss": 0.158, "step": 12598 }, { "epoch": 9.23003663003663, "grad_norm": 0.85546875, "learning_rate": 8.814643650925609e-05, "loss": 0.1315, "step": 12599 }, { "epoch": 9.23076923076923, "grad_norm": 0.7578125, "learning_rate": 8.814421168940585e-05, "loss": 0.1325, "step": 12600 }, { "epoch": 9.23150183150183, "grad_norm": 1.40625, "learning_rate": 8.814198668918979e-05, "loss": 0.1504, "step": 12601 }, { "epoch": 9.232234432234431, "grad_norm": 0.67578125, "learning_rate": 8.813976150861863e-05, "loss": 0.0985, "step": 12602 }, { "epoch": 9.232967032967032, "grad_norm": 0.95703125, "learning_rate": 8.813753614770304e-05, "loss": 0.1423, "step": 12603 }, { "epoch": 9.233699633699633, "grad_norm": 0.62890625, "learning_rate": 8.813531060645364e-05, "loss": 0.105, "step": 12604 }, { "epoch": 9.234432234432234, "grad_norm": 0.75, "learning_rate": 8.81330848848811e-05, "loss": 0.1047, "step": 12605 }, { "epoch": 9.235164835164834, "grad_norm": 1.4453125, "learning_rate": 8.813085898299613e-05, "loss": 0.1417, "step": 12606 }, { "epoch": 9.235897435897435, "grad_norm": 1.546875, "learning_rate": 8.812863290080931e-05, "loss": 0.1427, "step": 12607 }, { "epoch": 9.236630036630036, "grad_norm": 1.65625, "learning_rate": 8.81264066383314e-05, "loss": 0.1483, "step": 12608 }, { "epoch": 9.237362637362637, "grad_norm": 1.171875, "learning_rate": 8.812418019557301e-05, "loss": 0.137, "step": 12609 }, { "epoch": 9.238095238095237, "grad_norm": 0.90234375, "learning_rate": 8.812195357254481e-05, "loss": 0.1366, "step": 12610 }, { "epoch": 9.238827838827838, "grad_norm": 0.90625, "learning_rate": 8.81197267692575e-05, "loss": 0.1566, "step": 12611 }, { "epoch": 9.239560439560439, "grad_norm": 1.0703125, "learning_rate": 8.811749978572173e-05, "loss": 0.171, "step": 12612 }, { "epoch": 9.24029304029304, "grad_norm": 1.5546875, "learning_rate": 8.811527262194815e-05, "loss": 0.1489, "step": 12613 }, { "epoch": 9.24102564102564, "grad_norm": 0.7890625, "learning_rate": 8.811304527794748e-05, "loss": 0.1478, "step": 12614 }, { "epoch": 9.241758241758241, "grad_norm": 1.1171875, "learning_rate": 8.811081775373033e-05, "loss": 0.1403, "step": 12615 }, { "epoch": 9.242490842490842, "grad_norm": 0.75390625, "learning_rate": 8.810859004930742e-05, "loss": 0.1412, "step": 12616 }, { "epoch": 9.243223443223442, "grad_norm": 1.15625, "learning_rate": 8.81063621646894e-05, "loss": 0.1153, "step": 12617 }, { "epoch": 9.243956043956043, "grad_norm": 1.9765625, "learning_rate": 8.810413409988697e-05, "loss": 0.1497, "step": 12618 }, { "epoch": 9.244688644688644, "grad_norm": 1.640625, "learning_rate": 8.810190585491078e-05, "loss": 0.1269, "step": 12619 }, { "epoch": 9.245421245421245, "grad_norm": 1.1484375, "learning_rate": 8.809967742977151e-05, "loss": 0.1289, "step": 12620 }, { "epoch": 9.246153846153845, "grad_norm": 1.359375, "learning_rate": 8.809744882447984e-05, "loss": 0.1425, "step": 12621 }, { "epoch": 9.246886446886446, "grad_norm": 0.80859375, "learning_rate": 8.809522003904646e-05, "loss": 0.1415, "step": 12622 }, { "epoch": 9.247619047619047, "grad_norm": 1.546875, "learning_rate": 8.809299107348204e-05, "loss": 0.1592, "step": 12623 }, { "epoch": 9.248351648351647, "grad_norm": 1.1171875, "learning_rate": 8.809076192779725e-05, "loss": 0.1449, "step": 12624 }, { "epoch": 9.249084249084248, "grad_norm": 1.4453125, "learning_rate": 8.80885326020028e-05, "loss": 0.1318, "step": 12625 }, { "epoch": 9.249816849816849, "grad_norm": 1.0703125, "learning_rate": 8.808630309610931e-05, "loss": 0.1375, "step": 12626 }, { "epoch": 9.25054945054945, "grad_norm": 0.71875, "learning_rate": 8.808407341012754e-05, "loss": 0.1346, "step": 12627 }, { "epoch": 9.25128205128205, "grad_norm": 1.1875, "learning_rate": 8.808184354406812e-05, "loss": 0.1691, "step": 12628 }, { "epoch": 9.252014652014651, "grad_norm": 1.1640625, "learning_rate": 8.807961349794176e-05, "loss": 0.1307, "step": 12629 }, { "epoch": 9.252747252747252, "grad_norm": 0.99609375, "learning_rate": 8.807738327175913e-05, "loss": 0.142, "step": 12630 }, { "epoch": 9.253479853479853, "grad_norm": 1.03125, "learning_rate": 8.807515286553093e-05, "loss": 0.1332, "step": 12631 }, { "epoch": 9.254212454212453, "grad_norm": 1.2265625, "learning_rate": 8.807292227926783e-05, "loss": 0.169, "step": 12632 }, { "epoch": 9.254945054945054, "grad_norm": 1.1015625, "learning_rate": 8.807069151298052e-05, "loss": 0.1487, "step": 12633 }, { "epoch": 9.255677655677655, "grad_norm": 1.15625, "learning_rate": 8.80684605666797e-05, "loss": 0.1609, "step": 12634 }, { "epoch": 9.256410256410255, "grad_norm": 0.90234375, "learning_rate": 8.806622944037603e-05, "loss": 0.1166, "step": 12635 }, { "epoch": 9.257142857142856, "grad_norm": 0.8671875, "learning_rate": 8.806399813408024e-05, "loss": 0.0887, "step": 12636 }, { "epoch": 9.257875457875459, "grad_norm": 0.8828125, "learning_rate": 8.806176664780302e-05, "loss": 0.1297, "step": 12637 }, { "epoch": 9.25860805860806, "grad_norm": 1.1015625, "learning_rate": 8.805953498155502e-05, "loss": 0.1277, "step": 12638 }, { "epoch": 9.25934065934066, "grad_norm": 0.90234375, "learning_rate": 8.805730313534698e-05, "loss": 0.163, "step": 12639 }, { "epoch": 9.260073260073261, "grad_norm": 1.3359375, "learning_rate": 8.805507110918955e-05, "loss": 0.1135, "step": 12640 }, { "epoch": 9.260805860805862, "grad_norm": 2.234375, "learning_rate": 8.805283890309347e-05, "loss": 0.1516, "step": 12641 }, { "epoch": 9.261538461538462, "grad_norm": 1.265625, "learning_rate": 8.80506065170694e-05, "loss": 0.1409, "step": 12642 }, { "epoch": 9.262271062271063, "grad_norm": 1.5078125, "learning_rate": 8.804837395112803e-05, "loss": 0.1365, "step": 12643 }, { "epoch": 9.263003663003664, "grad_norm": 0.71484375, "learning_rate": 8.804614120528011e-05, "loss": 0.1264, "step": 12644 }, { "epoch": 9.263736263736265, "grad_norm": 1.015625, "learning_rate": 8.804390827953627e-05, "loss": 0.1423, "step": 12645 }, { "epoch": 9.264468864468865, "grad_norm": 1.3046875, "learning_rate": 8.804167517390724e-05, "loss": 0.1285, "step": 12646 }, { "epoch": 9.265201465201466, "grad_norm": 0.828125, "learning_rate": 8.803944188840375e-05, "loss": 0.1471, "step": 12647 }, { "epoch": 9.265934065934067, "grad_norm": 1.0546875, "learning_rate": 8.803720842303645e-05, "loss": 0.136, "step": 12648 }, { "epoch": 9.266666666666667, "grad_norm": 1.3125, "learning_rate": 8.803497477781605e-05, "loss": 0.1219, "step": 12649 }, { "epoch": 9.267399267399268, "grad_norm": 0.83203125, "learning_rate": 8.803274095275328e-05, "loss": 0.1105, "step": 12650 }, { "epoch": 9.268131868131869, "grad_norm": 0.796875, "learning_rate": 8.803050694785882e-05, "loss": 0.1322, "step": 12651 }, { "epoch": 9.26886446886447, "grad_norm": 1.1640625, "learning_rate": 8.802827276314338e-05, "loss": 0.1179, "step": 12652 }, { "epoch": 9.26959706959707, "grad_norm": 1.2109375, "learning_rate": 8.802603839861767e-05, "loss": 0.1316, "step": 12653 }, { "epoch": 9.270329670329671, "grad_norm": 1.4765625, "learning_rate": 8.80238038542924e-05, "loss": 0.163, "step": 12654 }, { "epoch": 9.271062271062272, "grad_norm": 1.015625, "learning_rate": 8.802156913017825e-05, "loss": 0.1233, "step": 12655 }, { "epoch": 9.271794871794873, "grad_norm": 1.9921875, "learning_rate": 8.801933422628596e-05, "loss": 0.1316, "step": 12656 }, { "epoch": 9.272527472527473, "grad_norm": 1.2265625, "learning_rate": 8.801709914262622e-05, "loss": 0.1282, "step": 12657 }, { "epoch": 9.273260073260074, "grad_norm": 1.546875, "learning_rate": 8.801486387920975e-05, "loss": 0.1875, "step": 12658 }, { "epoch": 9.273992673992675, "grad_norm": 2.109375, "learning_rate": 8.801262843604722e-05, "loss": 0.1328, "step": 12659 }, { "epoch": 9.274725274725276, "grad_norm": 1.09375, "learning_rate": 8.80103928131494e-05, "loss": 0.133, "step": 12660 }, { "epoch": 9.275457875457876, "grad_norm": 1.875, "learning_rate": 8.800815701052696e-05, "loss": 0.129, "step": 12661 }, { "epoch": 9.276190476190477, "grad_norm": 0.55078125, "learning_rate": 8.800592102819064e-05, "loss": 0.0904, "step": 12662 }, { "epoch": 9.276923076923078, "grad_norm": 1.1328125, "learning_rate": 8.800368486615113e-05, "loss": 0.0972, "step": 12663 }, { "epoch": 9.277655677655678, "grad_norm": 1.203125, "learning_rate": 8.800144852441916e-05, "loss": 0.1442, "step": 12664 }, { "epoch": 9.27838827838828, "grad_norm": 0.9609375, "learning_rate": 8.799921200300542e-05, "loss": 0.1196, "step": 12665 }, { "epoch": 9.27912087912088, "grad_norm": 1.46875, "learning_rate": 8.799697530192068e-05, "loss": 0.1363, "step": 12666 }, { "epoch": 9.27985347985348, "grad_norm": 1.484375, "learning_rate": 8.799473842117559e-05, "loss": 0.1563, "step": 12667 }, { "epoch": 9.280586080586081, "grad_norm": 1.140625, "learning_rate": 8.799250136078092e-05, "loss": 0.0988, "step": 12668 }, { "epoch": 9.281318681318682, "grad_norm": 0.88671875, "learning_rate": 8.799026412074737e-05, "loss": 0.1375, "step": 12669 }, { "epoch": 9.282051282051283, "grad_norm": 1.390625, "learning_rate": 8.798802670108565e-05, "loss": 0.1467, "step": 12670 }, { "epoch": 9.282783882783884, "grad_norm": 0.80078125, "learning_rate": 8.798578910180649e-05, "loss": 0.1442, "step": 12671 }, { "epoch": 9.283516483516484, "grad_norm": 1.5546875, "learning_rate": 8.798355132292059e-05, "loss": 0.1275, "step": 12672 }, { "epoch": 9.284249084249085, "grad_norm": 0.828125, "learning_rate": 8.798131336443871e-05, "loss": 0.1262, "step": 12673 }, { "epoch": 9.284981684981686, "grad_norm": 1.390625, "learning_rate": 8.797907522637154e-05, "loss": 0.1416, "step": 12674 }, { "epoch": 9.285714285714286, "grad_norm": 1.203125, "learning_rate": 8.797683690872983e-05, "loss": 0.141, "step": 12675 }, { "epoch": 9.286446886446887, "grad_norm": 0.765625, "learning_rate": 8.79745984115243e-05, "loss": 0.1422, "step": 12676 }, { "epoch": 9.287179487179488, "grad_norm": 0.84375, "learning_rate": 8.797235973476565e-05, "loss": 0.1028, "step": 12677 }, { "epoch": 9.287912087912089, "grad_norm": 0.93359375, "learning_rate": 8.797012087846461e-05, "loss": 0.1484, "step": 12678 }, { "epoch": 9.28864468864469, "grad_norm": 1.0234375, "learning_rate": 8.796788184263196e-05, "loss": 0.1473, "step": 12679 }, { "epoch": 9.28937728937729, "grad_norm": 0.9921875, "learning_rate": 8.796564262727835e-05, "loss": 0.1222, "step": 12680 }, { "epoch": 9.29010989010989, "grad_norm": 0.9765625, "learning_rate": 8.796340323241457e-05, "loss": 0.1202, "step": 12681 }, { "epoch": 9.290842490842492, "grad_norm": 1.296875, "learning_rate": 8.796116365805132e-05, "loss": 0.1326, "step": 12682 }, { "epoch": 9.291575091575092, "grad_norm": 1.046875, "learning_rate": 8.795892390419934e-05, "loss": 0.1617, "step": 12683 }, { "epoch": 9.292307692307693, "grad_norm": 1.0390625, "learning_rate": 8.795668397086934e-05, "loss": 0.1536, "step": 12684 }, { "epoch": 9.293040293040294, "grad_norm": 1.484375, "learning_rate": 8.79544438580721e-05, "loss": 0.1276, "step": 12685 }, { "epoch": 9.293772893772895, "grad_norm": 1.3203125, "learning_rate": 8.79522035658183e-05, "loss": 0.1203, "step": 12686 }, { "epoch": 9.294505494505495, "grad_norm": 1.2890625, "learning_rate": 8.794996309411872e-05, "loss": 0.1392, "step": 12687 }, { "epoch": 9.295238095238096, "grad_norm": 1.015625, "learning_rate": 8.794772244298406e-05, "loss": 0.1512, "step": 12688 }, { "epoch": 9.295970695970697, "grad_norm": 1.296875, "learning_rate": 8.794548161242507e-05, "loss": 0.1362, "step": 12689 }, { "epoch": 9.296703296703297, "grad_norm": 1.3828125, "learning_rate": 8.794324060245248e-05, "loss": 0.1403, "step": 12690 }, { "epoch": 9.297435897435898, "grad_norm": 1.734375, "learning_rate": 8.794099941307704e-05, "loss": 0.1664, "step": 12691 }, { "epoch": 9.298168498168499, "grad_norm": 0.76953125, "learning_rate": 8.793875804430948e-05, "loss": 0.0895, "step": 12692 }, { "epoch": 9.2989010989011, "grad_norm": 1.09375, "learning_rate": 8.793651649616053e-05, "loss": 0.1322, "step": 12693 }, { "epoch": 9.2996336996337, "grad_norm": 1.0703125, "learning_rate": 8.793427476864095e-05, "loss": 0.1359, "step": 12694 }, { "epoch": 9.300366300366301, "grad_norm": 1.2890625, "learning_rate": 8.793203286176147e-05, "loss": 0.1419, "step": 12695 }, { "epoch": 9.301098901098902, "grad_norm": 1.7421875, "learning_rate": 8.792979077553283e-05, "loss": 0.1434, "step": 12696 }, { "epoch": 9.301831501831503, "grad_norm": 1.09375, "learning_rate": 8.792754850996577e-05, "loss": 0.1153, "step": 12697 }, { "epoch": 9.302564102564103, "grad_norm": 1.7421875, "learning_rate": 8.792530606507104e-05, "loss": 0.1789, "step": 12698 }, { "epoch": 9.303296703296704, "grad_norm": 1.5, "learning_rate": 8.79230634408594e-05, "loss": 0.1375, "step": 12699 }, { "epoch": 9.304029304029305, "grad_norm": 0.96484375, "learning_rate": 8.792082063734157e-05, "loss": 0.1346, "step": 12700 }, { "epoch": 9.304761904761905, "grad_norm": 1.1484375, "learning_rate": 8.791857765452829e-05, "loss": 0.1376, "step": 12701 }, { "epoch": 9.305494505494506, "grad_norm": 0.953125, "learning_rate": 8.791633449243033e-05, "loss": 0.1413, "step": 12702 }, { "epoch": 9.306227106227107, "grad_norm": 0.74609375, "learning_rate": 8.791409115105843e-05, "loss": 0.1249, "step": 12703 }, { "epoch": 9.306959706959708, "grad_norm": 0.671875, "learning_rate": 8.791184763042332e-05, "loss": 0.1024, "step": 12704 }, { "epoch": 9.307692307692308, "grad_norm": 1.171875, "learning_rate": 8.790960393053579e-05, "loss": 0.1183, "step": 12705 }, { "epoch": 9.308424908424909, "grad_norm": 1.0625, "learning_rate": 8.790736005140655e-05, "loss": 0.1517, "step": 12706 }, { "epoch": 9.30915750915751, "grad_norm": 0.9453125, "learning_rate": 8.790511599304638e-05, "loss": 0.149, "step": 12707 }, { "epoch": 9.30989010989011, "grad_norm": 1.0859375, "learning_rate": 8.790287175546602e-05, "loss": 0.121, "step": 12708 }, { "epoch": 9.310622710622711, "grad_norm": 1.3125, "learning_rate": 8.790062733867623e-05, "loss": 0.1815, "step": 12709 }, { "epoch": 9.311355311355312, "grad_norm": 0.859375, "learning_rate": 8.789838274268774e-05, "loss": 0.1118, "step": 12710 }, { "epoch": 9.312087912087913, "grad_norm": 0.67578125, "learning_rate": 8.789613796751133e-05, "loss": 0.1348, "step": 12711 }, { "epoch": 9.312820512820513, "grad_norm": 1.1484375, "learning_rate": 8.789389301315772e-05, "loss": 0.1453, "step": 12712 }, { "epoch": 9.313553113553114, "grad_norm": 1.03125, "learning_rate": 8.789164787963772e-05, "loss": 0.1251, "step": 12713 }, { "epoch": 9.314285714285715, "grad_norm": 1.140625, "learning_rate": 8.788940256696206e-05, "loss": 0.1643, "step": 12714 }, { "epoch": 9.315018315018316, "grad_norm": 1.1328125, "learning_rate": 8.78871570751415e-05, "loss": 0.1226, "step": 12715 }, { "epoch": 9.315750915750916, "grad_norm": 0.98046875, "learning_rate": 8.788491140418678e-05, "loss": 0.1114, "step": 12716 }, { "epoch": 9.316483516483517, "grad_norm": 0.9765625, "learning_rate": 8.78826655541087e-05, "loss": 0.1252, "step": 12717 }, { "epoch": 9.317216117216118, "grad_norm": 1.359375, "learning_rate": 8.788041952491798e-05, "loss": 0.164, "step": 12718 }, { "epoch": 9.317948717948719, "grad_norm": 0.98828125, "learning_rate": 8.787817331662542e-05, "loss": 0.148, "step": 12719 }, { "epoch": 9.31868131868132, "grad_norm": 1.046875, "learning_rate": 8.787592692924173e-05, "loss": 0.1431, "step": 12720 }, { "epoch": 9.31941391941392, "grad_norm": 0.59765625, "learning_rate": 8.787368036277772e-05, "loss": 0.1182, "step": 12721 }, { "epoch": 9.32014652014652, "grad_norm": 1.171875, "learning_rate": 8.787143361724415e-05, "loss": 0.1563, "step": 12722 }, { "epoch": 9.320879120879122, "grad_norm": 0.80078125, "learning_rate": 8.786918669265176e-05, "loss": 0.1492, "step": 12723 }, { "epoch": 9.321611721611722, "grad_norm": 1.2421875, "learning_rate": 8.786693958901132e-05, "loss": 0.1319, "step": 12724 }, { "epoch": 9.322344322344323, "grad_norm": 1.0703125, "learning_rate": 8.786469230633362e-05, "loss": 0.1444, "step": 12725 }, { "epoch": 9.323076923076924, "grad_norm": 0.91796875, "learning_rate": 8.786244484462941e-05, "loss": 0.1556, "step": 12726 }, { "epoch": 9.323809523809524, "grad_norm": 1.09375, "learning_rate": 8.786019720390947e-05, "loss": 0.1356, "step": 12727 }, { "epoch": 9.324542124542125, "grad_norm": 1.0546875, "learning_rate": 8.785794938418454e-05, "loss": 0.1316, "step": 12728 }, { "epoch": 9.325274725274726, "grad_norm": 1.453125, "learning_rate": 8.785570138546543e-05, "loss": 0.1408, "step": 12729 }, { "epoch": 9.326007326007327, "grad_norm": 1.3203125, "learning_rate": 8.785345320776289e-05, "loss": 0.1358, "step": 12730 }, { "epoch": 9.326739926739927, "grad_norm": 1.6484375, "learning_rate": 8.785120485108768e-05, "loss": 0.141, "step": 12731 }, { "epoch": 9.327472527472528, "grad_norm": 1.390625, "learning_rate": 8.784895631545059e-05, "loss": 0.1561, "step": 12732 }, { "epoch": 9.328205128205129, "grad_norm": 1.40625, "learning_rate": 8.78467076008624e-05, "loss": 0.1812, "step": 12733 }, { "epoch": 9.32893772893773, "grad_norm": 0.9609375, "learning_rate": 8.784445870733387e-05, "loss": 0.1103, "step": 12734 }, { "epoch": 9.32967032967033, "grad_norm": 1.7890625, "learning_rate": 8.784220963487576e-05, "loss": 0.1394, "step": 12735 }, { "epoch": 9.330402930402931, "grad_norm": 1.015625, "learning_rate": 8.783996038349889e-05, "loss": 0.1401, "step": 12736 }, { "epoch": 9.331135531135532, "grad_norm": 1.2890625, "learning_rate": 8.7837710953214e-05, "loss": 0.1121, "step": 12737 }, { "epoch": 9.331868131868132, "grad_norm": 1.3125, "learning_rate": 8.783546134403188e-05, "loss": 0.1558, "step": 12738 }, { "epoch": 9.332600732600733, "grad_norm": 1.046875, "learning_rate": 8.783321155596331e-05, "loss": 0.1496, "step": 12739 }, { "epoch": 9.333333333333334, "grad_norm": 1.0390625, "learning_rate": 8.783096158901908e-05, "loss": 0.1343, "step": 12740 }, { "epoch": 9.334065934065935, "grad_norm": 0.8984375, "learning_rate": 8.782871144320994e-05, "loss": 0.1353, "step": 12741 }, { "epoch": 9.334798534798535, "grad_norm": 0.76953125, "learning_rate": 8.78264611185467e-05, "loss": 0.1354, "step": 12742 }, { "epoch": 9.335531135531136, "grad_norm": 1.1875, "learning_rate": 8.782421061504013e-05, "loss": 0.1336, "step": 12743 }, { "epoch": 9.336263736263737, "grad_norm": 1.03125, "learning_rate": 8.782195993270101e-05, "loss": 0.1514, "step": 12744 }, { "epoch": 9.336996336996338, "grad_norm": 0.96875, "learning_rate": 8.781970907154012e-05, "loss": 0.1584, "step": 12745 }, { "epoch": 9.337728937728938, "grad_norm": 1.1953125, "learning_rate": 8.781745803156826e-05, "loss": 0.1491, "step": 12746 }, { "epoch": 9.338461538461539, "grad_norm": 1.0078125, "learning_rate": 8.781520681279622e-05, "loss": 0.1158, "step": 12747 }, { "epoch": 9.33919413919414, "grad_norm": 1.265625, "learning_rate": 8.781295541523477e-05, "loss": 0.1637, "step": 12748 }, { "epoch": 9.33992673992674, "grad_norm": 1.3125, "learning_rate": 8.781070383889468e-05, "loss": 0.1261, "step": 12749 }, { "epoch": 9.340659340659341, "grad_norm": 0.828125, "learning_rate": 8.780845208378678e-05, "loss": 0.1092, "step": 12750 }, { "epoch": 9.341391941391942, "grad_norm": 1.03125, "learning_rate": 8.780620014992183e-05, "loss": 0.13, "step": 12751 }, { "epoch": 9.342124542124543, "grad_norm": 1.1640625, "learning_rate": 8.780394803731063e-05, "loss": 0.1277, "step": 12752 }, { "epoch": 9.342857142857143, "grad_norm": 1.1484375, "learning_rate": 8.780169574596397e-05, "loss": 0.1635, "step": 12753 }, { "epoch": 9.343589743589744, "grad_norm": 0.90234375, "learning_rate": 8.779944327589264e-05, "loss": 0.1471, "step": 12754 }, { "epoch": 9.344322344322345, "grad_norm": 1.1328125, "learning_rate": 8.779719062710746e-05, "loss": 0.0921, "step": 12755 }, { "epoch": 9.345054945054946, "grad_norm": 0.9609375, "learning_rate": 8.779493779961915e-05, "loss": 0.1586, "step": 12756 }, { "epoch": 9.345787545787546, "grad_norm": 1.140625, "learning_rate": 8.779268479343858e-05, "loss": 0.1376, "step": 12757 }, { "epoch": 9.346520146520147, "grad_norm": 1.140625, "learning_rate": 8.779043160857649e-05, "loss": 0.1819, "step": 12758 }, { "epoch": 9.347252747252748, "grad_norm": 1.640625, "learning_rate": 8.77881782450437e-05, "loss": 0.1644, "step": 12759 }, { "epoch": 9.347985347985349, "grad_norm": 1.1875, "learning_rate": 8.778592470285105e-05, "loss": 0.1445, "step": 12760 }, { "epoch": 9.34871794871795, "grad_norm": 1.609375, "learning_rate": 8.778367098200926e-05, "loss": 0.12, "step": 12761 }, { "epoch": 9.34945054945055, "grad_norm": 1.375, "learning_rate": 8.778141708252918e-05, "loss": 0.1414, "step": 12762 }, { "epoch": 9.35018315018315, "grad_norm": 1.0546875, "learning_rate": 8.777916300442157e-05, "loss": 0.1534, "step": 12763 }, { "epoch": 9.350915750915751, "grad_norm": 1.2578125, "learning_rate": 8.777690874769726e-05, "loss": 0.1344, "step": 12764 }, { "epoch": 9.351648351648352, "grad_norm": 0.921875, "learning_rate": 8.777465431236704e-05, "loss": 0.1216, "step": 12765 }, { "epoch": 9.352380952380953, "grad_norm": 0.73828125, "learning_rate": 8.777239969844173e-05, "loss": 0.1418, "step": 12766 }, { "epoch": 9.353113553113554, "grad_norm": 0.78125, "learning_rate": 8.777014490593211e-05, "loss": 0.1106, "step": 12767 }, { "epoch": 9.353846153846154, "grad_norm": 1.265625, "learning_rate": 8.776788993484897e-05, "loss": 0.1358, "step": 12768 }, { "epoch": 9.354578754578755, "grad_norm": 1.5703125, "learning_rate": 8.776563478520317e-05, "loss": 0.1375, "step": 12769 }, { "epoch": 9.355311355311356, "grad_norm": 1.140625, "learning_rate": 8.776337945700547e-05, "loss": 0.1152, "step": 12770 }, { "epoch": 9.356043956043957, "grad_norm": 0.91015625, "learning_rate": 8.776112395026668e-05, "loss": 0.1193, "step": 12771 }, { "epoch": 9.356776556776557, "grad_norm": 1.390625, "learning_rate": 8.77588682649976e-05, "loss": 0.1187, "step": 12772 }, { "epoch": 9.357509157509158, "grad_norm": 1.8359375, "learning_rate": 8.775661240120907e-05, "loss": 0.1527, "step": 12773 }, { "epoch": 9.358241758241759, "grad_norm": 1.1484375, "learning_rate": 8.775435635891188e-05, "loss": 0.139, "step": 12774 }, { "epoch": 9.35897435897436, "grad_norm": 1.4609375, "learning_rate": 8.775210013811683e-05, "loss": 0.1838, "step": 12775 }, { "epoch": 9.35970695970696, "grad_norm": 0.96484375, "learning_rate": 8.774984373883474e-05, "loss": 0.1512, "step": 12776 }, { "epoch": 9.360439560439561, "grad_norm": 1.140625, "learning_rate": 8.774758716107643e-05, "loss": 0.1214, "step": 12777 }, { "epoch": 9.361172161172162, "grad_norm": 1.5859375, "learning_rate": 8.774533040485268e-05, "loss": 0.1539, "step": 12778 }, { "epoch": 9.361904761904762, "grad_norm": 1.3828125, "learning_rate": 8.774307347017435e-05, "loss": 0.1365, "step": 12779 }, { "epoch": 9.362637362637363, "grad_norm": 0.8671875, "learning_rate": 8.774081635705223e-05, "loss": 0.1453, "step": 12780 }, { "epoch": 9.363369963369964, "grad_norm": 1.171875, "learning_rate": 8.773855906549712e-05, "loss": 0.1327, "step": 12781 }, { "epoch": 9.364102564102565, "grad_norm": 0.72265625, "learning_rate": 8.773630159551984e-05, "loss": 0.1135, "step": 12782 }, { "epoch": 9.364835164835165, "grad_norm": 1.90625, "learning_rate": 8.773404394713122e-05, "loss": 0.1532, "step": 12783 }, { "epoch": 9.365567765567766, "grad_norm": 1.1640625, "learning_rate": 8.77317861203421e-05, "loss": 0.1355, "step": 12784 }, { "epoch": 9.366300366300367, "grad_norm": 1.4453125, "learning_rate": 8.772952811516324e-05, "loss": 0.1569, "step": 12785 }, { "epoch": 9.367032967032968, "grad_norm": 2.046875, "learning_rate": 8.772726993160549e-05, "loss": 0.1284, "step": 12786 }, { "epoch": 9.367765567765568, "grad_norm": 1.0390625, "learning_rate": 8.772501156967968e-05, "loss": 0.146, "step": 12787 }, { "epoch": 9.368498168498169, "grad_norm": 0.83203125, "learning_rate": 8.772275302939662e-05, "loss": 0.1816, "step": 12788 }, { "epoch": 9.36923076923077, "grad_norm": 1.734375, "learning_rate": 8.772049431076713e-05, "loss": 0.1205, "step": 12789 }, { "epoch": 9.36996336996337, "grad_norm": 1.6953125, "learning_rate": 8.771823541380202e-05, "loss": 0.1715, "step": 12790 }, { "epoch": 9.370695970695971, "grad_norm": 1.3359375, "learning_rate": 8.771597633851214e-05, "loss": 0.1568, "step": 12791 }, { "epoch": 9.371428571428572, "grad_norm": 1.3046875, "learning_rate": 8.77137170849083e-05, "loss": 0.1451, "step": 12792 }, { "epoch": 9.372161172161173, "grad_norm": 1.125, "learning_rate": 8.771145765300133e-05, "loss": 0.1201, "step": 12793 }, { "epoch": 9.372893772893773, "grad_norm": 1.0625, "learning_rate": 8.770919804280204e-05, "loss": 0.1059, "step": 12794 }, { "epoch": 9.373626373626374, "grad_norm": 1.2109375, "learning_rate": 8.770693825432127e-05, "loss": 0.1316, "step": 12795 }, { "epoch": 9.374358974358975, "grad_norm": 1.0703125, "learning_rate": 8.770467828756984e-05, "loss": 0.1263, "step": 12796 }, { "epoch": 9.375091575091576, "grad_norm": 1.1640625, "learning_rate": 8.770241814255859e-05, "loss": 0.1488, "step": 12797 }, { "epoch": 9.375824175824176, "grad_norm": 1.3046875, "learning_rate": 8.770015781929834e-05, "loss": 0.1402, "step": 12798 }, { "epoch": 9.376556776556777, "grad_norm": 1.2265625, "learning_rate": 8.769789731779991e-05, "loss": 0.153, "step": 12799 }, { "epoch": 9.377289377289378, "grad_norm": 0.65625, "learning_rate": 8.769563663807415e-05, "loss": 0.1132, "step": 12800 }, { "epoch": 9.378021978021978, "grad_norm": 1.25, "learning_rate": 8.76933757801319e-05, "loss": 0.1722, "step": 12801 }, { "epoch": 9.37875457875458, "grad_norm": 0.99609375, "learning_rate": 8.769111474398397e-05, "loss": 0.1446, "step": 12802 }, { "epoch": 9.37948717948718, "grad_norm": 0.9375, "learning_rate": 8.768885352964118e-05, "loss": 0.1153, "step": 12803 }, { "epoch": 9.38021978021978, "grad_norm": 0.80078125, "learning_rate": 8.768659213711441e-05, "loss": 0.1238, "step": 12804 }, { "epoch": 9.380952380952381, "grad_norm": 0.95703125, "learning_rate": 8.768433056641446e-05, "loss": 0.1107, "step": 12805 }, { "epoch": 9.381684981684982, "grad_norm": 0.82421875, "learning_rate": 8.768206881755217e-05, "loss": 0.1423, "step": 12806 }, { "epoch": 9.382417582417583, "grad_norm": 0.8984375, "learning_rate": 8.767980689053838e-05, "loss": 0.1385, "step": 12807 }, { "epoch": 9.383150183150184, "grad_norm": 1.015625, "learning_rate": 8.767754478538395e-05, "loss": 0.1177, "step": 12808 }, { "epoch": 9.383882783882784, "grad_norm": 1.109375, "learning_rate": 8.767528250209969e-05, "loss": 0.1121, "step": 12809 }, { "epoch": 9.384615384615385, "grad_norm": 2.125, "learning_rate": 8.767302004069642e-05, "loss": 0.1434, "step": 12810 }, { "epoch": 9.385347985347986, "grad_norm": 1.5390625, "learning_rate": 8.767075740118504e-05, "loss": 0.1545, "step": 12811 }, { "epoch": 9.386080586080586, "grad_norm": 1.5546875, "learning_rate": 8.766849458357635e-05, "loss": 0.1666, "step": 12812 }, { "epoch": 9.386813186813187, "grad_norm": 1.5546875, "learning_rate": 8.766623158788121e-05, "loss": 0.1593, "step": 12813 }, { "epoch": 9.387545787545788, "grad_norm": 0.80859375, "learning_rate": 8.766396841411044e-05, "loss": 0.1193, "step": 12814 }, { "epoch": 9.388278388278389, "grad_norm": 1.265625, "learning_rate": 8.766170506227491e-05, "loss": 0.1652, "step": 12815 }, { "epoch": 9.38901098901099, "grad_norm": 2.0625, "learning_rate": 8.765944153238542e-05, "loss": 0.1447, "step": 12816 }, { "epoch": 9.38974358974359, "grad_norm": 1.90625, "learning_rate": 8.765717782445288e-05, "loss": 0.1594, "step": 12817 }, { "epoch": 9.39047619047619, "grad_norm": 1.1796875, "learning_rate": 8.765491393848809e-05, "loss": 0.1198, "step": 12818 }, { "epoch": 9.391208791208792, "grad_norm": 2.15625, "learning_rate": 8.765264987450192e-05, "loss": 0.1473, "step": 12819 }, { "epoch": 9.391941391941392, "grad_norm": 0.91796875, "learning_rate": 8.765038563250519e-05, "loss": 0.1262, "step": 12820 }, { "epoch": 9.392673992673993, "grad_norm": 1.1640625, "learning_rate": 8.764812121250878e-05, "loss": 0.1134, "step": 12821 }, { "epoch": 9.393406593406594, "grad_norm": 1.125, "learning_rate": 8.764585661452354e-05, "loss": 0.1646, "step": 12822 }, { "epoch": 9.394139194139195, "grad_norm": 2.390625, "learning_rate": 8.764359183856028e-05, "loss": 0.1686, "step": 12823 }, { "epoch": 9.394871794871795, "grad_norm": 2.21875, "learning_rate": 8.764132688462988e-05, "loss": 0.1614, "step": 12824 }, { "epoch": 9.395604395604396, "grad_norm": 1.5703125, "learning_rate": 8.76390617527432e-05, "loss": 0.1686, "step": 12825 }, { "epoch": 9.396336996336997, "grad_norm": 1.3671875, "learning_rate": 8.763679644291108e-05, "loss": 0.1372, "step": 12826 }, { "epoch": 9.397069597069597, "grad_norm": 0.98828125, "learning_rate": 8.763453095514439e-05, "loss": 0.1182, "step": 12827 }, { "epoch": 9.397802197802198, "grad_norm": 1.578125, "learning_rate": 8.763226528945397e-05, "loss": 0.1441, "step": 12828 }, { "epoch": 9.398534798534799, "grad_norm": 2.5625, "learning_rate": 8.762999944585065e-05, "loss": 0.1677, "step": 12829 }, { "epoch": 9.3992673992674, "grad_norm": 2.125, "learning_rate": 8.762773342434534e-05, "loss": 0.165, "step": 12830 }, { "epoch": 9.4, "grad_norm": 2.21875, "learning_rate": 8.762546722494886e-05, "loss": 0.1351, "step": 12831 }, { "epoch": 9.400732600732601, "grad_norm": 1.9453125, "learning_rate": 8.76232008476721e-05, "loss": 0.122, "step": 12832 }, { "epoch": 9.401465201465202, "grad_norm": 1.859375, "learning_rate": 8.762093429252586e-05, "loss": 0.141, "step": 12833 }, { "epoch": 9.402197802197803, "grad_norm": 1.0390625, "learning_rate": 8.761866755952106e-05, "loss": 0.1248, "step": 12834 }, { "epoch": 9.402930402930403, "grad_norm": 1.4609375, "learning_rate": 8.761640064866853e-05, "loss": 0.1618, "step": 12835 }, { "epoch": 9.403663003663004, "grad_norm": 1.2734375, "learning_rate": 8.761413355997915e-05, "loss": 0.1552, "step": 12836 }, { "epoch": 9.404395604395605, "grad_norm": 1.765625, "learning_rate": 8.761186629346377e-05, "loss": 0.1349, "step": 12837 }, { "epoch": 9.405128205128205, "grad_norm": 2.640625, "learning_rate": 8.760959884913324e-05, "loss": 0.1359, "step": 12838 }, { "epoch": 9.405860805860806, "grad_norm": 2.28125, "learning_rate": 8.760733122699846e-05, "loss": 0.1474, "step": 12839 }, { "epoch": 9.406593406593407, "grad_norm": 1.265625, "learning_rate": 8.760506342707028e-05, "loss": 0.1215, "step": 12840 }, { "epoch": 9.407326007326008, "grad_norm": 0.7734375, "learning_rate": 8.760279544935956e-05, "loss": 0.1344, "step": 12841 }, { "epoch": 9.408058608058608, "grad_norm": 1.9375, "learning_rate": 8.760052729387714e-05, "loss": 0.1262, "step": 12842 }, { "epoch": 9.40879120879121, "grad_norm": 2.1875, "learning_rate": 8.759825896063393e-05, "loss": 0.1111, "step": 12843 }, { "epoch": 9.40952380952381, "grad_norm": 2.0625, "learning_rate": 8.759599044964079e-05, "loss": 0.1182, "step": 12844 }, { "epoch": 9.41025641025641, "grad_norm": 1.171875, "learning_rate": 8.759372176090857e-05, "loss": 0.149, "step": 12845 }, { "epoch": 9.410989010989011, "grad_norm": 1.25, "learning_rate": 8.759145289444815e-05, "loss": 0.1612, "step": 12846 }, { "epoch": 9.411721611721612, "grad_norm": 0.73046875, "learning_rate": 8.758918385027041e-05, "loss": 0.1303, "step": 12847 }, { "epoch": 9.412454212454213, "grad_norm": 1.4296875, "learning_rate": 8.758691462838622e-05, "loss": 0.1435, "step": 12848 }, { "epoch": 9.413186813186813, "grad_norm": 1.1875, "learning_rate": 8.758464522880643e-05, "loss": 0.1486, "step": 12849 }, { "epoch": 9.413919413919414, "grad_norm": 1.53125, "learning_rate": 8.758237565154195e-05, "loss": 0.1366, "step": 12850 }, { "epoch": 9.414652014652015, "grad_norm": 1.4296875, "learning_rate": 8.758010589660363e-05, "loss": 0.1342, "step": 12851 }, { "epoch": 9.415384615384616, "grad_norm": 1.015625, "learning_rate": 8.757783596400233e-05, "loss": 0.1108, "step": 12852 }, { "epoch": 9.416117216117216, "grad_norm": 0.7578125, "learning_rate": 8.757556585374899e-05, "loss": 0.1184, "step": 12853 }, { "epoch": 9.416849816849817, "grad_norm": 0.79296875, "learning_rate": 8.75732955658544e-05, "loss": 0.1469, "step": 12854 }, { "epoch": 9.417582417582418, "grad_norm": 1.1015625, "learning_rate": 8.75710251003295e-05, "loss": 0.1437, "step": 12855 }, { "epoch": 9.418315018315019, "grad_norm": 2.109375, "learning_rate": 8.756875445718515e-05, "loss": 0.1648, "step": 12856 }, { "epoch": 9.41904761904762, "grad_norm": 1.140625, "learning_rate": 8.756648363643222e-05, "loss": 0.13, "step": 12857 }, { "epoch": 9.41978021978022, "grad_norm": 1.828125, "learning_rate": 8.756421263808158e-05, "loss": 0.1489, "step": 12858 }, { "epoch": 9.42051282051282, "grad_norm": 0.83984375, "learning_rate": 8.756194146214418e-05, "loss": 0.1564, "step": 12859 }, { "epoch": 9.421245421245422, "grad_norm": 0.94140625, "learning_rate": 8.755967010863081e-05, "loss": 0.13, "step": 12860 }, { "epoch": 9.421978021978022, "grad_norm": 0.8515625, "learning_rate": 8.755739857755242e-05, "loss": 0.1533, "step": 12861 }, { "epoch": 9.422710622710623, "grad_norm": 0.88671875, "learning_rate": 8.755512686891985e-05, "loss": 0.1294, "step": 12862 }, { "epoch": 9.423443223443224, "grad_norm": 0.84765625, "learning_rate": 8.7552854982744e-05, "loss": 0.1426, "step": 12863 }, { "epoch": 9.424175824175824, "grad_norm": 0.9296875, "learning_rate": 8.755058291903578e-05, "loss": 0.1164, "step": 12864 }, { "epoch": 9.424908424908425, "grad_norm": 1.5390625, "learning_rate": 8.754831067780605e-05, "loss": 0.1508, "step": 12865 }, { "epoch": 9.425641025641026, "grad_norm": 0.97265625, "learning_rate": 8.75460382590657e-05, "loss": 0.1392, "step": 12866 }, { "epoch": 9.426373626373627, "grad_norm": 1.5390625, "learning_rate": 8.75437656628256e-05, "loss": 0.1386, "step": 12867 }, { "epoch": 9.427106227106227, "grad_norm": 0.8203125, "learning_rate": 8.754149288909666e-05, "loss": 0.1268, "step": 12868 }, { "epoch": 9.427838827838828, "grad_norm": 1.0546875, "learning_rate": 8.753921993788978e-05, "loss": 0.1187, "step": 12869 }, { "epoch": 9.428571428571429, "grad_norm": 1.3203125, "learning_rate": 8.753694680921585e-05, "loss": 0.1282, "step": 12870 }, { "epoch": 9.42930402930403, "grad_norm": 0.8828125, "learning_rate": 8.753467350308573e-05, "loss": 0.1622, "step": 12871 }, { "epoch": 9.43003663003663, "grad_norm": 0.73046875, "learning_rate": 8.753240001951034e-05, "loss": 0.1254, "step": 12872 }, { "epoch": 9.430769230769231, "grad_norm": 1.0703125, "learning_rate": 8.753012635850056e-05, "loss": 0.1421, "step": 12873 }, { "epoch": 9.431501831501832, "grad_norm": 0.98046875, "learning_rate": 8.75278525200673e-05, "loss": 0.1434, "step": 12874 }, { "epoch": 9.432234432234432, "grad_norm": 1.015625, "learning_rate": 8.752557850422143e-05, "loss": 0.1182, "step": 12875 }, { "epoch": 9.432967032967033, "grad_norm": 1.109375, "learning_rate": 8.752330431097385e-05, "loss": 0.1406, "step": 12876 }, { "epoch": 9.433699633699634, "grad_norm": 1.3359375, "learning_rate": 8.752102994033548e-05, "loss": 0.1368, "step": 12877 }, { "epoch": 9.434432234432235, "grad_norm": 0.91796875, "learning_rate": 8.75187553923172e-05, "loss": 0.1134, "step": 12878 }, { "epoch": 9.435164835164835, "grad_norm": 1.0, "learning_rate": 8.751648066692992e-05, "loss": 0.1366, "step": 12879 }, { "epoch": 9.435897435897436, "grad_norm": 1.1796875, "learning_rate": 8.751420576418452e-05, "loss": 0.147, "step": 12880 }, { "epoch": 9.436630036630037, "grad_norm": 0.953125, "learning_rate": 8.751193068409191e-05, "loss": 0.1389, "step": 12881 }, { "epoch": 9.437362637362638, "grad_norm": 0.984375, "learning_rate": 8.750965542666297e-05, "loss": 0.115, "step": 12882 }, { "epoch": 9.438095238095238, "grad_norm": 0.83203125, "learning_rate": 8.750737999190865e-05, "loss": 0.1223, "step": 12883 }, { "epoch": 9.438827838827839, "grad_norm": 1.3515625, "learning_rate": 8.750510437983982e-05, "loss": 0.1387, "step": 12884 }, { "epoch": 9.43956043956044, "grad_norm": 0.94140625, "learning_rate": 8.750282859046736e-05, "loss": 0.1342, "step": 12885 }, { "epoch": 9.44029304029304, "grad_norm": 0.921875, "learning_rate": 8.750055262380223e-05, "loss": 0.1087, "step": 12886 }, { "epoch": 9.441025641025641, "grad_norm": 1.140625, "learning_rate": 8.749827647985531e-05, "loss": 0.1093, "step": 12887 }, { "epoch": 9.441758241758242, "grad_norm": 1.0234375, "learning_rate": 8.74960001586375e-05, "loss": 0.1204, "step": 12888 }, { "epoch": 9.442490842490843, "grad_norm": 1.5546875, "learning_rate": 8.749372366015968e-05, "loss": 0.1348, "step": 12889 }, { "epoch": 9.443223443223443, "grad_norm": 1.0234375, "learning_rate": 8.749144698443282e-05, "loss": 0.1208, "step": 12890 }, { "epoch": 9.443956043956044, "grad_norm": 1.234375, "learning_rate": 8.748917013146778e-05, "loss": 0.1745, "step": 12891 }, { "epoch": 9.444688644688645, "grad_norm": 1.703125, "learning_rate": 8.748689310127547e-05, "loss": 0.1435, "step": 12892 }, { "epoch": 9.445421245421246, "grad_norm": 1.3515625, "learning_rate": 8.748461589386682e-05, "loss": 0.1576, "step": 12893 }, { "epoch": 9.446153846153846, "grad_norm": 1.40625, "learning_rate": 8.748233850925273e-05, "loss": 0.1242, "step": 12894 }, { "epoch": 9.446886446886447, "grad_norm": 1.0625, "learning_rate": 8.748006094744414e-05, "loss": 0.1196, "step": 12895 }, { "epoch": 9.447619047619048, "grad_norm": 1.1875, "learning_rate": 8.747778320845191e-05, "loss": 0.1459, "step": 12896 }, { "epoch": 9.448351648351649, "grad_norm": 0.8515625, "learning_rate": 8.7475505292287e-05, "loss": 0.1281, "step": 12897 }, { "epoch": 9.44908424908425, "grad_norm": 1.1796875, "learning_rate": 8.74732271989603e-05, "loss": 0.1431, "step": 12898 }, { "epoch": 9.44981684981685, "grad_norm": 1.9453125, "learning_rate": 8.747094892848272e-05, "loss": 0.1326, "step": 12899 }, { "epoch": 9.45054945054945, "grad_norm": 1.09375, "learning_rate": 8.74686704808652e-05, "loss": 0.147, "step": 12900 }, { "epoch": 9.451282051282051, "grad_norm": 1.28125, "learning_rate": 8.746639185611864e-05, "loss": 0.1236, "step": 12901 }, { "epoch": 9.452014652014652, "grad_norm": 1.1171875, "learning_rate": 8.746411305425398e-05, "loss": 0.1524, "step": 12902 }, { "epoch": 9.452747252747253, "grad_norm": 1.4921875, "learning_rate": 8.746183407528209e-05, "loss": 0.1213, "step": 12903 }, { "epoch": 9.453479853479854, "grad_norm": 1.0703125, "learning_rate": 8.745955491921392e-05, "loss": 0.1489, "step": 12904 }, { "epoch": 9.454212454212454, "grad_norm": 1.1953125, "learning_rate": 8.74572755860604e-05, "loss": 0.1456, "step": 12905 }, { "epoch": 9.454945054945055, "grad_norm": 0.62109375, "learning_rate": 8.745499607583245e-05, "loss": 0.1405, "step": 12906 }, { "epoch": 9.455677655677656, "grad_norm": 0.54296875, "learning_rate": 8.745271638854098e-05, "loss": 0.1271, "step": 12907 }, { "epoch": 9.456410256410257, "grad_norm": 1.0, "learning_rate": 8.74504365241969e-05, "loss": 0.13, "step": 12908 }, { "epoch": 9.457142857142857, "grad_norm": 0.875, "learning_rate": 8.744815648281116e-05, "loss": 0.1314, "step": 12909 }, { "epoch": 9.457875457875458, "grad_norm": 1.21875, "learning_rate": 8.744587626439468e-05, "loss": 0.1319, "step": 12910 }, { "epoch": 9.458608058608059, "grad_norm": 0.8828125, "learning_rate": 8.744359586895837e-05, "loss": 0.1305, "step": 12911 }, { "epoch": 9.45934065934066, "grad_norm": 0.94140625, "learning_rate": 8.744131529651315e-05, "loss": 0.1609, "step": 12912 }, { "epoch": 9.46007326007326, "grad_norm": 0.96875, "learning_rate": 8.743903454706997e-05, "loss": 0.1446, "step": 12913 }, { "epoch": 9.460805860805861, "grad_norm": 0.82421875, "learning_rate": 8.743675362063975e-05, "loss": 0.1291, "step": 12914 }, { "epoch": 9.461538461538462, "grad_norm": 1.171875, "learning_rate": 8.743447251723343e-05, "loss": 0.1615, "step": 12915 }, { "epoch": 9.462271062271062, "grad_norm": 1.140625, "learning_rate": 8.743219123686192e-05, "loss": 0.122, "step": 12916 }, { "epoch": 9.463003663003663, "grad_norm": 1.046875, "learning_rate": 8.742990977953615e-05, "loss": 0.1405, "step": 12917 }, { "epoch": 9.463736263736264, "grad_norm": 0.8125, "learning_rate": 8.742762814526707e-05, "loss": 0.1387, "step": 12918 }, { "epoch": 9.464468864468865, "grad_norm": 1.125, "learning_rate": 8.742534633406558e-05, "loss": 0.164, "step": 12919 }, { "epoch": 9.465201465201465, "grad_norm": 1.5234375, "learning_rate": 8.742306434594267e-05, "loss": 0.1188, "step": 12920 }, { "epoch": 9.465934065934066, "grad_norm": 1.59375, "learning_rate": 8.742078218090921e-05, "loss": 0.1489, "step": 12921 }, { "epoch": 9.466666666666667, "grad_norm": 1.4375, "learning_rate": 8.741849983897618e-05, "loss": 0.1278, "step": 12922 }, { "epoch": 9.467399267399268, "grad_norm": 1.0625, "learning_rate": 8.741621732015448e-05, "loss": 0.1438, "step": 12923 }, { "epoch": 9.468131868131868, "grad_norm": 1.3515625, "learning_rate": 8.741393462445507e-05, "loss": 0.1309, "step": 12924 }, { "epoch": 9.468864468864469, "grad_norm": 1.6328125, "learning_rate": 8.741165175188887e-05, "loss": 0.1573, "step": 12925 }, { "epoch": 9.46959706959707, "grad_norm": 2.421875, "learning_rate": 8.740936870246686e-05, "loss": 0.1386, "step": 12926 }, { "epoch": 9.47032967032967, "grad_norm": 2.265625, "learning_rate": 8.740708547619993e-05, "loss": 0.1727, "step": 12927 }, { "epoch": 9.471062271062271, "grad_norm": 1.625, "learning_rate": 8.740480207309904e-05, "loss": 0.1486, "step": 12928 }, { "epoch": 9.471794871794872, "grad_norm": 1.015625, "learning_rate": 8.740251849317513e-05, "loss": 0.1298, "step": 12929 }, { "epoch": 9.472527472527473, "grad_norm": 0.609375, "learning_rate": 8.740023473643912e-05, "loss": 0.1286, "step": 12930 }, { "epoch": 9.473260073260073, "grad_norm": 1.0078125, "learning_rate": 8.739795080290199e-05, "loss": 0.0993, "step": 12931 }, { "epoch": 9.473992673992674, "grad_norm": 1.328125, "learning_rate": 8.739566669257467e-05, "loss": 0.1455, "step": 12932 }, { "epoch": 9.474725274725275, "grad_norm": 1.5625, "learning_rate": 8.73933824054681e-05, "loss": 0.1543, "step": 12933 }, { "epoch": 9.475457875457876, "grad_norm": 0.83203125, "learning_rate": 8.739109794159321e-05, "loss": 0.1231, "step": 12934 }, { "epoch": 9.476190476190476, "grad_norm": 1.046875, "learning_rate": 8.738881330096096e-05, "loss": 0.1019, "step": 12935 }, { "epoch": 9.476923076923077, "grad_norm": 0.76171875, "learning_rate": 8.73865284835823e-05, "loss": 0.1257, "step": 12936 }, { "epoch": 9.477655677655678, "grad_norm": 0.921875, "learning_rate": 8.738424348946816e-05, "loss": 0.1125, "step": 12937 }, { "epoch": 9.478388278388278, "grad_norm": 1.265625, "learning_rate": 8.738195831862953e-05, "loss": 0.1273, "step": 12938 }, { "epoch": 9.47912087912088, "grad_norm": 0.65625, "learning_rate": 8.73796729710773e-05, "loss": 0.1451, "step": 12939 }, { "epoch": 9.47985347985348, "grad_norm": 0.8671875, "learning_rate": 8.737738744682247e-05, "loss": 0.1094, "step": 12940 }, { "epoch": 9.48058608058608, "grad_norm": 0.83203125, "learning_rate": 8.737510174587596e-05, "loss": 0.1201, "step": 12941 }, { "epoch": 9.481318681318681, "grad_norm": 1.0078125, "learning_rate": 8.737281586824874e-05, "loss": 0.1108, "step": 12942 }, { "epoch": 9.482051282051282, "grad_norm": 1.1953125, "learning_rate": 8.737052981395174e-05, "loss": 0.1111, "step": 12943 }, { "epoch": 9.482783882783883, "grad_norm": 1.59375, "learning_rate": 8.736824358299594e-05, "loss": 0.1702, "step": 12944 }, { "epoch": 9.483516483516484, "grad_norm": 1.109375, "learning_rate": 8.736595717539228e-05, "loss": 0.1424, "step": 12945 }, { "epoch": 9.484249084249084, "grad_norm": 1.4453125, "learning_rate": 8.736367059115174e-05, "loss": 0.153, "step": 12946 }, { "epoch": 9.484981684981685, "grad_norm": 1.03125, "learning_rate": 8.736138383028522e-05, "loss": 0.1234, "step": 12947 }, { "epoch": 9.485714285714286, "grad_norm": 1.375, "learning_rate": 8.735909689280372e-05, "loss": 0.1304, "step": 12948 }, { "epoch": 9.486446886446886, "grad_norm": 0.6875, "learning_rate": 8.735680977871817e-05, "loss": 0.1204, "step": 12949 }, { "epoch": 9.487179487179487, "grad_norm": 0.77734375, "learning_rate": 8.735452248803957e-05, "loss": 0.1035, "step": 12950 }, { "epoch": 9.487912087912088, "grad_norm": 0.62890625, "learning_rate": 8.735223502077885e-05, "loss": 0.1398, "step": 12951 }, { "epoch": 9.488644688644689, "grad_norm": 0.828125, "learning_rate": 8.734994737694696e-05, "loss": 0.1355, "step": 12952 }, { "epoch": 9.48937728937729, "grad_norm": 1.265625, "learning_rate": 8.734765955655488e-05, "loss": 0.1207, "step": 12953 }, { "epoch": 9.49010989010989, "grad_norm": 0.86328125, "learning_rate": 8.734537155961359e-05, "loss": 0.1238, "step": 12954 }, { "epoch": 9.49084249084249, "grad_norm": 1.203125, "learning_rate": 8.734308338613401e-05, "loss": 0.1294, "step": 12955 }, { "epoch": 9.491575091575092, "grad_norm": 0.8125, "learning_rate": 8.734079503612712e-05, "loss": 0.1006, "step": 12956 }, { "epoch": 9.492307692307692, "grad_norm": 1.4921875, "learning_rate": 8.73385065096039e-05, "loss": 0.113, "step": 12957 }, { "epoch": 9.493040293040293, "grad_norm": 0.9921875, "learning_rate": 8.733621780657528e-05, "loss": 0.1169, "step": 12958 }, { "epoch": 9.493772893772894, "grad_norm": 0.734375, "learning_rate": 8.733392892705226e-05, "loss": 0.1218, "step": 12959 }, { "epoch": 9.494505494505495, "grad_norm": 1.0234375, "learning_rate": 8.73316398710458e-05, "loss": 0.115, "step": 12960 }, { "epoch": 9.495238095238095, "grad_norm": 0.76171875, "learning_rate": 8.732935063856685e-05, "loss": 0.1209, "step": 12961 }, { "epoch": 9.495970695970696, "grad_norm": 1.1484375, "learning_rate": 8.732706122962639e-05, "loss": 0.1466, "step": 12962 }, { "epoch": 9.496703296703297, "grad_norm": 1.3984375, "learning_rate": 8.732477164423541e-05, "loss": 0.1846, "step": 12963 }, { "epoch": 9.497435897435897, "grad_norm": 1.0546875, "learning_rate": 8.732248188240484e-05, "loss": 0.1324, "step": 12964 }, { "epoch": 9.498168498168498, "grad_norm": 0.98046875, "learning_rate": 8.732019194414569e-05, "loss": 0.1172, "step": 12965 }, { "epoch": 9.498901098901099, "grad_norm": 1.1171875, "learning_rate": 8.731790182946892e-05, "loss": 0.1295, "step": 12966 }, { "epoch": 9.4996336996337, "grad_norm": 0.9453125, "learning_rate": 8.731561153838548e-05, "loss": 0.1361, "step": 12967 }, { "epoch": 9.5003663003663, "grad_norm": 1.4296875, "learning_rate": 8.731332107090635e-05, "loss": 0.1701, "step": 12968 }, { "epoch": 9.501098901098901, "grad_norm": 1.2734375, "learning_rate": 8.731103042704251e-05, "loss": 0.1313, "step": 12969 }, { "epoch": 9.501831501831502, "grad_norm": 1.7109375, "learning_rate": 8.730873960680494e-05, "loss": 0.1312, "step": 12970 }, { "epoch": 9.502564102564103, "grad_norm": 1.40625, "learning_rate": 8.730644861020464e-05, "loss": 0.1605, "step": 12971 }, { "epoch": 9.503296703296703, "grad_norm": 0.8125, "learning_rate": 8.730415743725256e-05, "loss": 0.0978, "step": 12972 }, { "epoch": 9.504029304029304, "grad_norm": 1.2578125, "learning_rate": 8.730186608795966e-05, "loss": 0.1462, "step": 12973 }, { "epoch": 9.504761904761905, "grad_norm": 1.0234375, "learning_rate": 8.729957456233695e-05, "loss": 0.1037, "step": 12974 }, { "epoch": 9.505494505494505, "grad_norm": 0.8671875, "learning_rate": 8.729728286039539e-05, "loss": 0.1419, "step": 12975 }, { "epoch": 9.506227106227106, "grad_norm": 0.70703125, "learning_rate": 8.729499098214596e-05, "loss": 0.1341, "step": 12976 }, { "epoch": 9.506959706959707, "grad_norm": 0.82421875, "learning_rate": 8.729269892759966e-05, "loss": 0.1215, "step": 12977 }, { "epoch": 9.507692307692308, "grad_norm": 1.2734375, "learning_rate": 8.729040669676746e-05, "loss": 0.1784, "step": 12978 }, { "epoch": 9.508424908424908, "grad_norm": 1.078125, "learning_rate": 8.728811428966037e-05, "loss": 0.1312, "step": 12979 }, { "epoch": 9.50915750915751, "grad_norm": 1.4609375, "learning_rate": 8.72858217062893e-05, "loss": 0.1108, "step": 12980 }, { "epoch": 9.50989010989011, "grad_norm": 1.109375, "learning_rate": 8.728352894666533e-05, "loss": 0.1163, "step": 12981 }, { "epoch": 9.51062271062271, "grad_norm": 1.140625, "learning_rate": 8.728123601079936e-05, "loss": 0.1461, "step": 12982 }, { "epoch": 9.511355311355311, "grad_norm": 1.0703125, "learning_rate": 8.727894289870244e-05, "loss": 0.1404, "step": 12983 }, { "epoch": 9.512087912087912, "grad_norm": 1.625, "learning_rate": 8.727664961038551e-05, "loss": 0.1201, "step": 12984 }, { "epoch": 9.512820512820513, "grad_norm": 1.34375, "learning_rate": 8.72743561458596e-05, "loss": 0.1484, "step": 12985 }, { "epoch": 9.513553113553114, "grad_norm": 0.796875, "learning_rate": 8.727206250513566e-05, "loss": 0.1283, "step": 12986 }, { "epoch": 9.514285714285714, "grad_norm": 1.1484375, "learning_rate": 8.726976868822469e-05, "loss": 0.1374, "step": 12987 }, { "epoch": 9.515018315018315, "grad_norm": 0.90625, "learning_rate": 8.72674746951377e-05, "loss": 0.1376, "step": 12988 }, { "epoch": 9.515750915750916, "grad_norm": 1.015625, "learning_rate": 8.726518052588568e-05, "loss": 0.1302, "step": 12989 }, { "epoch": 9.516483516483516, "grad_norm": 1.5546875, "learning_rate": 8.72628861804796e-05, "loss": 0.1412, "step": 12990 }, { "epoch": 9.517216117216117, "grad_norm": 1.78125, "learning_rate": 8.726059165893044e-05, "loss": 0.1322, "step": 12991 }, { "epoch": 9.517948717948718, "grad_norm": 1.25, "learning_rate": 8.725829696124926e-05, "loss": 0.1365, "step": 12992 }, { "epoch": 9.518681318681319, "grad_norm": 0.8046875, "learning_rate": 8.725600208744699e-05, "loss": 0.1423, "step": 12993 }, { "epoch": 9.51941391941392, "grad_norm": 1.0703125, "learning_rate": 8.725370703753467e-05, "loss": 0.1568, "step": 12994 }, { "epoch": 9.52014652014652, "grad_norm": 0.8515625, "learning_rate": 8.725141181152323e-05, "loss": 0.105, "step": 12995 }, { "epoch": 9.52087912087912, "grad_norm": 1.5703125, "learning_rate": 8.724911640942374e-05, "loss": 0.1336, "step": 12996 }, { "epoch": 9.521611721611722, "grad_norm": 2.125, "learning_rate": 8.724682083124717e-05, "loss": 0.1349, "step": 12997 }, { "epoch": 9.522344322344322, "grad_norm": 1.6796875, "learning_rate": 8.724452507700451e-05, "loss": 0.1595, "step": 12998 }, { "epoch": 9.523076923076923, "grad_norm": 1.296875, "learning_rate": 8.724222914670677e-05, "loss": 0.12, "step": 12999 }, { "epoch": 9.523809523809524, "grad_norm": 1.609375, "learning_rate": 8.723993304036495e-05, "loss": 0.1387, "step": 13000 }, { "epoch": 9.524542124542124, "grad_norm": 1.390625, "learning_rate": 8.723763675799005e-05, "loss": 0.1296, "step": 13001 }, { "epoch": 9.525274725274725, "grad_norm": 1.6171875, "learning_rate": 8.723534029959307e-05, "loss": 0.1753, "step": 13002 }, { "epoch": 9.526007326007326, "grad_norm": 1.0234375, "learning_rate": 8.723304366518502e-05, "loss": 0.1183, "step": 13003 }, { "epoch": 9.526739926739927, "grad_norm": 0.79296875, "learning_rate": 8.723074685477689e-05, "loss": 0.1233, "step": 13004 }, { "epoch": 9.527472527472527, "grad_norm": 1.3515625, "learning_rate": 8.72284498683797e-05, "loss": 0.133, "step": 13005 }, { "epoch": 9.528205128205128, "grad_norm": 0.73046875, "learning_rate": 8.722615270600445e-05, "loss": 0.1464, "step": 13006 }, { "epoch": 9.528937728937729, "grad_norm": 0.94921875, "learning_rate": 8.722385536766216e-05, "loss": 0.1497, "step": 13007 }, { "epoch": 9.52967032967033, "grad_norm": 0.76171875, "learning_rate": 8.72215578533638e-05, "loss": 0.1414, "step": 13008 }, { "epoch": 9.53040293040293, "grad_norm": 1.5, "learning_rate": 8.721926016312041e-05, "loss": 0.128, "step": 13009 }, { "epoch": 9.531135531135531, "grad_norm": 1.3359375, "learning_rate": 8.721696229694299e-05, "loss": 0.1213, "step": 13010 }, { "epoch": 9.531868131868132, "grad_norm": 1.0546875, "learning_rate": 8.721466425484256e-05, "loss": 0.1326, "step": 13011 }, { "epoch": 9.532600732600732, "grad_norm": 0.98828125, "learning_rate": 8.721236603683011e-05, "loss": 0.1113, "step": 13012 }, { "epoch": 9.533333333333333, "grad_norm": 0.9921875, "learning_rate": 8.721006764291666e-05, "loss": 0.1444, "step": 13013 }, { "epoch": 9.534065934065934, "grad_norm": 0.8359375, "learning_rate": 8.720776907311324e-05, "loss": 0.1293, "step": 13014 }, { "epoch": 9.534798534798535, "grad_norm": 1.3125, "learning_rate": 8.720547032743081e-05, "loss": 0.1478, "step": 13015 }, { "epoch": 9.535531135531135, "grad_norm": 0.84375, "learning_rate": 8.720317140588045e-05, "loss": 0.1544, "step": 13016 }, { "epoch": 9.536263736263736, "grad_norm": 1.3046875, "learning_rate": 8.720087230847315e-05, "loss": 0.1333, "step": 13017 }, { "epoch": 9.536996336996337, "grad_norm": 1.515625, "learning_rate": 8.719857303521992e-05, "loss": 0.1665, "step": 13018 }, { "epoch": 9.537728937728938, "grad_norm": 1.1640625, "learning_rate": 8.719627358613179e-05, "loss": 0.1456, "step": 13019 }, { "epoch": 9.538461538461538, "grad_norm": 0.9375, "learning_rate": 8.719397396121976e-05, "loss": 0.1501, "step": 13020 }, { "epoch": 9.539194139194139, "grad_norm": 1.4140625, "learning_rate": 8.719167416049483e-05, "loss": 0.1435, "step": 13021 }, { "epoch": 9.53992673992674, "grad_norm": 0.98046875, "learning_rate": 8.718937418396807e-05, "loss": 0.1024, "step": 13022 }, { "epoch": 9.54065934065934, "grad_norm": 1.421875, "learning_rate": 8.718707403165047e-05, "loss": 0.1369, "step": 13023 }, { "epoch": 9.541391941391941, "grad_norm": 1.609375, "learning_rate": 8.718477370355306e-05, "loss": 0.1606, "step": 13024 }, { "epoch": 9.542124542124542, "grad_norm": 1.2109375, "learning_rate": 8.718247319968684e-05, "loss": 0.1303, "step": 13025 }, { "epoch": 9.542857142857143, "grad_norm": 1.078125, "learning_rate": 8.718017252006285e-05, "loss": 0.1206, "step": 13026 }, { "epoch": 9.543589743589743, "grad_norm": 1.0234375, "learning_rate": 8.717787166469212e-05, "loss": 0.109, "step": 13027 }, { "epoch": 9.544322344322344, "grad_norm": 1.421875, "learning_rate": 8.717557063358566e-05, "loss": 0.1376, "step": 13028 }, { "epoch": 9.545054945054945, "grad_norm": 0.95703125, "learning_rate": 8.717326942675452e-05, "loss": 0.1493, "step": 13029 }, { "epoch": 9.545787545787546, "grad_norm": 1.046875, "learning_rate": 8.71709680442097e-05, "loss": 0.1149, "step": 13030 }, { "epoch": 9.546520146520146, "grad_norm": 1.015625, "learning_rate": 8.716866648596222e-05, "loss": 0.1064, "step": 13031 }, { "epoch": 9.547252747252747, "grad_norm": 1.4609375, "learning_rate": 8.716636475202314e-05, "loss": 0.1645, "step": 13032 }, { "epoch": 9.547985347985348, "grad_norm": 1.390625, "learning_rate": 8.716406284240344e-05, "loss": 0.1302, "step": 13033 }, { "epoch": 9.548717948717949, "grad_norm": 1.125, "learning_rate": 8.71617607571142e-05, "loss": 0.1264, "step": 13034 }, { "epoch": 9.54945054945055, "grad_norm": 0.99609375, "learning_rate": 8.715945849616645e-05, "loss": 0.1103, "step": 13035 }, { "epoch": 9.55018315018315, "grad_norm": 0.953125, "learning_rate": 8.715715605957117e-05, "loss": 0.1462, "step": 13036 }, { "epoch": 9.55091575091575, "grad_norm": 0.92578125, "learning_rate": 8.715485344733942e-05, "loss": 0.1176, "step": 13037 }, { "epoch": 9.551648351648351, "grad_norm": 1.0625, "learning_rate": 8.715255065948224e-05, "loss": 0.1397, "step": 13038 }, { "epoch": 9.552380952380952, "grad_norm": 1.03125, "learning_rate": 8.715024769601067e-05, "loss": 0.1129, "step": 13039 }, { "epoch": 9.553113553113553, "grad_norm": 0.52734375, "learning_rate": 8.714794455693572e-05, "loss": 0.1273, "step": 13040 }, { "epoch": 9.553846153846154, "grad_norm": 0.7109375, "learning_rate": 8.714564124226844e-05, "loss": 0.106, "step": 13041 }, { "epoch": 9.554578754578754, "grad_norm": 0.91796875, "learning_rate": 8.714333775201985e-05, "loss": 0.1195, "step": 13042 }, { "epoch": 9.555311355311355, "grad_norm": 1.28125, "learning_rate": 8.714103408620102e-05, "loss": 0.1557, "step": 13043 }, { "epoch": 9.556043956043956, "grad_norm": 0.80859375, "learning_rate": 8.713873024482296e-05, "loss": 0.1223, "step": 13044 }, { "epoch": 9.556776556776557, "grad_norm": 0.89453125, "learning_rate": 8.71364262278967e-05, "loss": 0.1398, "step": 13045 }, { "epoch": 9.557509157509157, "grad_norm": 1.1796875, "learning_rate": 8.713412203543332e-05, "loss": 0.1194, "step": 13046 }, { "epoch": 9.558241758241758, "grad_norm": 0.89453125, "learning_rate": 8.713181766744382e-05, "loss": 0.1279, "step": 13047 }, { "epoch": 9.558974358974359, "grad_norm": 0.58984375, "learning_rate": 8.712951312393925e-05, "loss": 0.1141, "step": 13048 }, { "epoch": 9.55970695970696, "grad_norm": 0.85546875, "learning_rate": 8.712720840493068e-05, "loss": 0.1303, "step": 13049 }, { "epoch": 9.56043956043956, "grad_norm": 1.125, "learning_rate": 8.712490351042913e-05, "loss": 0.1118, "step": 13050 }, { "epoch": 9.561172161172161, "grad_norm": 0.9375, "learning_rate": 8.712259844044562e-05, "loss": 0.1461, "step": 13051 }, { "epoch": 9.561904761904762, "grad_norm": 0.7890625, "learning_rate": 8.712029319499123e-05, "loss": 0.1388, "step": 13052 }, { "epoch": 9.562637362637362, "grad_norm": 0.76953125, "learning_rate": 8.711798777407701e-05, "loss": 0.1276, "step": 13053 }, { "epoch": 9.563369963369963, "grad_norm": 0.6015625, "learning_rate": 8.711568217771397e-05, "loss": 0.1185, "step": 13054 }, { "epoch": 9.564102564102564, "grad_norm": 1.109375, "learning_rate": 8.711337640591317e-05, "loss": 0.1532, "step": 13055 }, { "epoch": 9.564835164835165, "grad_norm": 1.0546875, "learning_rate": 8.711107045868568e-05, "loss": 0.1513, "step": 13056 }, { "epoch": 9.565567765567765, "grad_norm": 1.2265625, "learning_rate": 8.710876433604253e-05, "loss": 0.1739, "step": 13057 }, { "epoch": 9.566300366300366, "grad_norm": 0.73828125, "learning_rate": 8.710645803799477e-05, "loss": 0.1025, "step": 13058 }, { "epoch": 9.567032967032967, "grad_norm": 1.0546875, "learning_rate": 8.710415156455345e-05, "loss": 0.1409, "step": 13059 }, { "epoch": 9.567765567765568, "grad_norm": 1.0703125, "learning_rate": 8.710184491572963e-05, "loss": 0.1079, "step": 13060 }, { "epoch": 9.568498168498168, "grad_norm": 0.98046875, "learning_rate": 8.709953809153434e-05, "loss": 0.0918, "step": 13061 }, { "epoch": 9.569230769230769, "grad_norm": 0.9140625, "learning_rate": 8.709723109197868e-05, "loss": 0.1186, "step": 13062 }, { "epoch": 9.56996336996337, "grad_norm": 1.140625, "learning_rate": 8.709492391707364e-05, "loss": 0.1412, "step": 13063 }, { "epoch": 9.57069597069597, "grad_norm": 1.1171875, "learning_rate": 8.709261656683032e-05, "loss": 0.161, "step": 13064 }, { "epoch": 9.571428571428571, "grad_norm": 1.0859375, "learning_rate": 8.709030904125975e-05, "loss": 0.1485, "step": 13065 }, { "epoch": 9.572161172161172, "grad_norm": 0.76953125, "learning_rate": 8.7088001340373e-05, "loss": 0.1252, "step": 13066 }, { "epoch": 9.572893772893773, "grad_norm": 0.65625, "learning_rate": 8.708569346418115e-05, "loss": 0.1388, "step": 13067 }, { "epoch": 9.573626373626373, "grad_norm": 0.91796875, "learning_rate": 8.70833854126952e-05, "loss": 0.137, "step": 13068 }, { "epoch": 9.574358974358974, "grad_norm": 0.7421875, "learning_rate": 8.708107718592626e-05, "loss": 0.1346, "step": 13069 }, { "epoch": 9.575091575091575, "grad_norm": 0.63671875, "learning_rate": 8.707876878388537e-05, "loss": 0.1006, "step": 13070 }, { "epoch": 9.575824175824176, "grad_norm": 0.75, "learning_rate": 8.70764602065836e-05, "loss": 0.1368, "step": 13071 }, { "epoch": 9.576556776556776, "grad_norm": 0.8671875, "learning_rate": 8.707415145403199e-05, "loss": 0.1424, "step": 13072 }, { "epoch": 9.577289377289377, "grad_norm": 0.88671875, "learning_rate": 8.70718425262416e-05, "loss": 0.1338, "step": 13073 }, { "epoch": 9.578021978021978, "grad_norm": 0.7578125, "learning_rate": 8.706953342322351e-05, "loss": 0.1228, "step": 13074 }, { "epoch": 9.578754578754578, "grad_norm": 0.7890625, "learning_rate": 8.70672241449888e-05, "loss": 0.149, "step": 13075 }, { "epoch": 9.57948717948718, "grad_norm": 0.66796875, "learning_rate": 8.706491469154849e-05, "loss": 0.1079, "step": 13076 }, { "epoch": 9.58021978021978, "grad_norm": 0.72265625, "learning_rate": 8.70626050629137e-05, "loss": 0.1334, "step": 13077 }, { "epoch": 9.58095238095238, "grad_norm": 1.390625, "learning_rate": 8.706029525909545e-05, "loss": 0.1356, "step": 13078 }, { "epoch": 9.581684981684981, "grad_norm": 0.9453125, "learning_rate": 8.705798528010481e-05, "loss": 0.1264, "step": 13079 }, { "epoch": 9.582417582417582, "grad_norm": 0.90234375, "learning_rate": 8.705567512595289e-05, "loss": 0.1529, "step": 13080 }, { "epoch": 9.583150183150183, "grad_norm": 1.2109375, "learning_rate": 8.70533647966507e-05, "loss": 0.1612, "step": 13081 }, { "epoch": 9.583882783882784, "grad_norm": 1.09375, "learning_rate": 8.705105429220936e-05, "loss": 0.1181, "step": 13082 }, { "epoch": 9.584615384615384, "grad_norm": 0.80859375, "learning_rate": 8.704874361263991e-05, "loss": 0.134, "step": 13083 }, { "epoch": 9.585347985347985, "grad_norm": 1.2578125, "learning_rate": 8.704643275795345e-05, "loss": 0.1559, "step": 13084 }, { "epoch": 9.586080586080586, "grad_norm": 1.28125, "learning_rate": 8.704412172816103e-05, "loss": 0.1497, "step": 13085 }, { "epoch": 9.586813186813187, "grad_norm": 0.6953125, "learning_rate": 8.70418105232737e-05, "loss": 0.1201, "step": 13086 }, { "epoch": 9.587545787545787, "grad_norm": 0.828125, "learning_rate": 8.703949914330257e-05, "loss": 0.1326, "step": 13087 }, { "epoch": 9.588278388278388, "grad_norm": 1.625, "learning_rate": 8.70371875882587e-05, "loss": 0.1243, "step": 13088 }, { "epoch": 9.589010989010989, "grad_norm": 0.93359375, "learning_rate": 8.70348758581532e-05, "loss": 0.1234, "step": 13089 }, { "epoch": 9.58974358974359, "grad_norm": 1.1328125, "learning_rate": 8.703256395299708e-05, "loss": 0.1589, "step": 13090 }, { "epoch": 9.59047619047619, "grad_norm": 1.0859375, "learning_rate": 8.703025187280148e-05, "loss": 0.126, "step": 13091 }, { "epoch": 9.59120879120879, "grad_norm": 0.76953125, "learning_rate": 8.702793961757743e-05, "loss": 0.1225, "step": 13092 }, { "epoch": 9.591941391941392, "grad_norm": 0.8515625, "learning_rate": 8.702562718733604e-05, "loss": 0.1115, "step": 13093 }, { "epoch": 9.592673992673992, "grad_norm": 1.5625, "learning_rate": 8.702331458208839e-05, "loss": 0.1401, "step": 13094 }, { "epoch": 9.593406593406593, "grad_norm": 1.0390625, "learning_rate": 8.702100180184553e-05, "loss": 0.1275, "step": 13095 }, { "epoch": 9.594139194139194, "grad_norm": 0.6796875, "learning_rate": 8.70186888466186e-05, "loss": 0.1137, "step": 13096 }, { "epoch": 9.594871794871795, "grad_norm": 1.0078125, "learning_rate": 8.701637571641858e-05, "loss": 0.1438, "step": 13097 }, { "epoch": 9.595604395604395, "grad_norm": 0.7734375, "learning_rate": 8.701406241125665e-05, "loss": 0.1266, "step": 13098 }, { "epoch": 9.596336996336996, "grad_norm": 0.98828125, "learning_rate": 8.701174893114388e-05, "loss": 0.1524, "step": 13099 }, { "epoch": 9.597069597069597, "grad_norm": 1.65625, "learning_rate": 8.70094352760913e-05, "loss": 0.1255, "step": 13100 }, { "epoch": 9.597802197802197, "grad_norm": 1.3203125, "learning_rate": 8.700712144611005e-05, "loss": 0.1251, "step": 13101 }, { "epoch": 9.598534798534798, "grad_norm": 1.0234375, "learning_rate": 8.700480744121119e-05, "loss": 0.1518, "step": 13102 }, { "epoch": 9.599267399267399, "grad_norm": 1.9765625, "learning_rate": 8.700249326140583e-05, "loss": 0.1666, "step": 13103 }, { "epoch": 9.6, "grad_norm": 1.2265625, "learning_rate": 8.700017890670502e-05, "loss": 0.1381, "step": 13104 }, { "epoch": 9.6007326007326, "grad_norm": 1.546875, "learning_rate": 8.699786437711988e-05, "loss": 0.1244, "step": 13105 }, { "epoch": 9.601465201465201, "grad_norm": 0.85546875, "learning_rate": 8.699554967266149e-05, "loss": 0.1098, "step": 13106 }, { "epoch": 9.602197802197802, "grad_norm": 0.703125, "learning_rate": 8.699323479334094e-05, "loss": 0.0992, "step": 13107 }, { "epoch": 9.602930402930403, "grad_norm": 0.9375, "learning_rate": 8.699091973916933e-05, "loss": 0.1413, "step": 13108 }, { "epoch": 9.603663003663003, "grad_norm": 1.515625, "learning_rate": 8.698860451015774e-05, "loss": 0.1085, "step": 13109 }, { "epoch": 9.604395604395604, "grad_norm": 0.73828125, "learning_rate": 8.698628910631725e-05, "loss": 0.1115, "step": 13110 }, { "epoch": 9.605128205128205, "grad_norm": 1.5546875, "learning_rate": 8.698397352765898e-05, "loss": 0.1289, "step": 13111 }, { "epoch": 9.605860805860805, "grad_norm": 1.6171875, "learning_rate": 8.698165777419403e-05, "loss": 0.1437, "step": 13112 }, { "epoch": 9.606593406593406, "grad_norm": 1.203125, "learning_rate": 8.697934184593347e-05, "loss": 0.1532, "step": 13113 }, { "epoch": 9.607326007326007, "grad_norm": 1.1875, "learning_rate": 8.697702574288842e-05, "loss": 0.1193, "step": 13114 }, { "epoch": 9.608058608058608, "grad_norm": 0.83984375, "learning_rate": 8.697470946506994e-05, "loss": 0.1671, "step": 13115 }, { "epoch": 9.608791208791208, "grad_norm": 1.3125, "learning_rate": 8.697239301248918e-05, "loss": 0.1491, "step": 13116 }, { "epoch": 9.60952380952381, "grad_norm": 1.140625, "learning_rate": 8.697007638515721e-05, "loss": 0.1151, "step": 13117 }, { "epoch": 9.61025641025641, "grad_norm": 2.625, "learning_rate": 8.69677595830851e-05, "loss": 0.1627, "step": 13118 }, { "epoch": 9.61098901098901, "grad_norm": 2.671875, "learning_rate": 8.696544260628403e-05, "loss": 0.1499, "step": 13119 }, { "epoch": 9.611721611721611, "grad_norm": 1.6171875, "learning_rate": 8.696312545476503e-05, "loss": 0.1469, "step": 13120 }, { "epoch": 9.612454212454212, "grad_norm": 1.390625, "learning_rate": 8.696080812853922e-05, "loss": 0.2245, "step": 13121 }, { "epoch": 9.613186813186813, "grad_norm": 1.1484375, "learning_rate": 8.695849062761769e-05, "loss": 0.1327, "step": 13122 }, { "epoch": 9.613919413919414, "grad_norm": 1.3203125, "learning_rate": 8.695617295201158e-05, "loss": 0.1341, "step": 13123 }, { "epoch": 9.614652014652014, "grad_norm": 0.83203125, "learning_rate": 8.6953855101732e-05, "loss": 0.1469, "step": 13124 }, { "epoch": 9.615384615384615, "grad_norm": 1.125, "learning_rate": 8.695153707679003e-05, "loss": 0.1234, "step": 13125 }, { "epoch": 9.616117216117216, "grad_norm": 0.796875, "learning_rate": 8.694921887719675e-05, "loss": 0.1306, "step": 13126 }, { "epoch": 9.616849816849816, "grad_norm": 1.03125, "learning_rate": 8.694690050296333e-05, "loss": 0.1345, "step": 13127 }, { "epoch": 9.617582417582417, "grad_norm": 1.0859375, "learning_rate": 8.694458195410082e-05, "loss": 0.1609, "step": 13128 }, { "epoch": 9.618315018315018, "grad_norm": 0.859375, "learning_rate": 8.694226323062036e-05, "loss": 0.1086, "step": 13129 }, { "epoch": 9.619047619047619, "grad_norm": 0.88671875, "learning_rate": 8.693994433253304e-05, "loss": 0.1236, "step": 13130 }, { "epoch": 9.61978021978022, "grad_norm": 0.7890625, "learning_rate": 8.693762525985002e-05, "loss": 0.1385, "step": 13131 }, { "epoch": 9.62051282051282, "grad_norm": 1.234375, "learning_rate": 8.693530601258235e-05, "loss": 0.1231, "step": 13132 }, { "epoch": 9.62124542124542, "grad_norm": 1.390625, "learning_rate": 8.693298659074118e-05, "loss": 0.169, "step": 13133 }, { "epoch": 9.621978021978022, "grad_norm": 1.28125, "learning_rate": 8.693066699433759e-05, "loss": 0.1164, "step": 13134 }, { "epoch": 9.622710622710622, "grad_norm": 0.9296875, "learning_rate": 8.692834722338274e-05, "loss": 0.1202, "step": 13135 }, { "epoch": 9.623443223443223, "grad_norm": 1.609375, "learning_rate": 8.69260272778877e-05, "loss": 0.1291, "step": 13136 }, { "epoch": 9.624175824175824, "grad_norm": 1.03125, "learning_rate": 8.692370715786362e-05, "loss": 0.1508, "step": 13137 }, { "epoch": 9.624908424908424, "grad_norm": 1.3125, "learning_rate": 8.69213868633216e-05, "loss": 0.1417, "step": 13138 }, { "epoch": 9.625641025641025, "grad_norm": 0.93359375, "learning_rate": 8.691906639427277e-05, "loss": 0.1219, "step": 13139 }, { "epoch": 9.626373626373626, "grad_norm": 1.28125, "learning_rate": 8.69167457507282e-05, "loss": 0.1176, "step": 13140 }, { "epoch": 9.627106227106227, "grad_norm": 1.3046875, "learning_rate": 8.691442493269908e-05, "loss": 0.1209, "step": 13141 }, { "epoch": 9.627838827838827, "grad_norm": 0.609375, "learning_rate": 8.69121039401965e-05, "loss": 0.1131, "step": 13142 }, { "epoch": 9.628571428571428, "grad_norm": 1.203125, "learning_rate": 8.690978277323155e-05, "loss": 0.0996, "step": 13143 }, { "epoch": 9.629304029304029, "grad_norm": 1.0, "learning_rate": 8.690746143181538e-05, "loss": 0.1334, "step": 13144 }, { "epoch": 9.63003663003663, "grad_norm": 1.328125, "learning_rate": 8.690513991595912e-05, "loss": 0.159, "step": 13145 }, { "epoch": 9.63076923076923, "grad_norm": 1.6640625, "learning_rate": 8.690281822567389e-05, "loss": 0.1253, "step": 13146 }, { "epoch": 9.631501831501831, "grad_norm": 1.90625, "learning_rate": 8.690049636097081e-05, "loss": 0.1599, "step": 13147 }, { "epoch": 9.632234432234432, "grad_norm": 1.125, "learning_rate": 8.689817432186099e-05, "loss": 0.1491, "step": 13148 }, { "epoch": 9.632967032967032, "grad_norm": 0.9765625, "learning_rate": 8.689585210835557e-05, "loss": 0.139, "step": 13149 }, { "epoch": 9.633699633699633, "grad_norm": 0.8984375, "learning_rate": 8.689352972046568e-05, "loss": 0.1122, "step": 13150 }, { "epoch": 9.634432234432234, "grad_norm": 1.765625, "learning_rate": 8.689120715820244e-05, "loss": 0.1516, "step": 13151 }, { "epoch": 9.635164835164835, "grad_norm": 2.296875, "learning_rate": 8.688888442157698e-05, "loss": 0.1395, "step": 13152 }, { "epoch": 9.635897435897435, "grad_norm": 1.671875, "learning_rate": 8.688656151060043e-05, "loss": 0.1338, "step": 13153 }, { "epoch": 9.636630036630036, "grad_norm": 1.328125, "learning_rate": 8.688423842528391e-05, "loss": 0.1206, "step": 13154 }, { "epoch": 9.637362637362637, "grad_norm": 1.046875, "learning_rate": 8.688191516563857e-05, "loss": 0.1434, "step": 13155 }, { "epoch": 9.638095238095238, "grad_norm": 0.64453125, "learning_rate": 8.687959173167552e-05, "loss": 0.1255, "step": 13156 }, { "epoch": 9.638827838827838, "grad_norm": 1.2109375, "learning_rate": 8.687726812340591e-05, "loss": 0.1248, "step": 13157 }, { "epoch": 9.639560439560439, "grad_norm": 1.015625, "learning_rate": 8.687494434084089e-05, "loss": 0.1506, "step": 13158 }, { "epoch": 9.64029304029304, "grad_norm": 0.98828125, "learning_rate": 8.687262038399154e-05, "loss": 0.12, "step": 13159 }, { "epoch": 9.64102564102564, "grad_norm": 1.7421875, "learning_rate": 8.687029625286904e-05, "loss": 0.1446, "step": 13160 }, { "epoch": 9.641758241758241, "grad_norm": 1.3203125, "learning_rate": 8.68679719474845e-05, "loss": 0.1313, "step": 13161 }, { "epoch": 9.642490842490842, "grad_norm": 1.0703125, "learning_rate": 8.686564746784906e-05, "loss": 0.1555, "step": 13162 }, { "epoch": 9.643223443223443, "grad_norm": 1.234375, "learning_rate": 8.686332281397387e-05, "loss": 0.131, "step": 13163 }, { "epoch": 9.643956043956043, "grad_norm": 1.8203125, "learning_rate": 8.686099798587007e-05, "loss": 0.1238, "step": 13164 }, { "epoch": 9.644688644688644, "grad_norm": 1.2890625, "learning_rate": 8.68586729835488e-05, "loss": 0.1496, "step": 13165 }, { "epoch": 9.645421245421245, "grad_norm": 0.57421875, "learning_rate": 8.685634780702119e-05, "loss": 0.1064, "step": 13166 }, { "epoch": 9.646153846153846, "grad_norm": 1.1640625, "learning_rate": 8.685402245629838e-05, "loss": 0.1589, "step": 13167 }, { "epoch": 9.646886446886446, "grad_norm": 0.7109375, "learning_rate": 8.685169693139152e-05, "loss": 0.1178, "step": 13168 }, { "epoch": 9.647619047619047, "grad_norm": 0.9921875, "learning_rate": 8.684937123231172e-05, "loss": 0.0977, "step": 13169 }, { "epoch": 9.648351648351648, "grad_norm": 1.09375, "learning_rate": 8.684704535907017e-05, "loss": 0.1428, "step": 13170 }, { "epoch": 9.649084249084249, "grad_norm": 1.4140625, "learning_rate": 8.684471931167801e-05, "loss": 0.1507, "step": 13171 }, { "epoch": 9.64981684981685, "grad_norm": 1.265625, "learning_rate": 8.684239309014634e-05, "loss": 0.1433, "step": 13172 }, { "epoch": 9.65054945054945, "grad_norm": 0.859375, "learning_rate": 8.684006669448633e-05, "loss": 0.1032, "step": 13173 }, { "epoch": 9.65128205128205, "grad_norm": 1.171875, "learning_rate": 8.683774012470916e-05, "loss": 0.152, "step": 13174 }, { "epoch": 9.652014652014651, "grad_norm": 0.984375, "learning_rate": 8.683541338082593e-05, "loss": 0.125, "step": 13175 }, { "epoch": 9.652747252747252, "grad_norm": 0.8828125, "learning_rate": 8.68330864628478e-05, "loss": 0.1215, "step": 13176 }, { "epoch": 9.653479853479853, "grad_norm": 0.953125, "learning_rate": 8.683075937078595e-05, "loss": 0.1334, "step": 13177 }, { "epoch": 9.654212454212454, "grad_norm": 0.734375, "learning_rate": 8.682843210465147e-05, "loss": 0.127, "step": 13178 }, { "epoch": 9.654945054945054, "grad_norm": 0.9453125, "learning_rate": 8.682610466445558e-05, "loss": 0.109, "step": 13179 }, { "epoch": 9.655677655677655, "grad_norm": 1.03125, "learning_rate": 8.682377705020938e-05, "loss": 0.1176, "step": 13180 }, { "epoch": 9.656410256410256, "grad_norm": 0.91015625, "learning_rate": 8.682144926192404e-05, "loss": 0.1456, "step": 13181 }, { "epoch": 9.657142857142857, "grad_norm": 0.83203125, "learning_rate": 8.681912129961072e-05, "loss": 0.1376, "step": 13182 }, { "epoch": 9.657875457875457, "grad_norm": 1.421875, "learning_rate": 8.681679316328057e-05, "loss": 0.1362, "step": 13183 }, { "epoch": 9.658608058608058, "grad_norm": 0.6328125, "learning_rate": 8.681446485294474e-05, "loss": 0.1149, "step": 13184 }, { "epoch": 9.659340659340659, "grad_norm": 1.0078125, "learning_rate": 8.681213636861437e-05, "loss": 0.1157, "step": 13185 }, { "epoch": 9.66007326007326, "grad_norm": 0.953125, "learning_rate": 8.680980771030065e-05, "loss": 0.1098, "step": 13186 }, { "epoch": 9.66080586080586, "grad_norm": 1.1640625, "learning_rate": 8.680747887801471e-05, "loss": 0.1176, "step": 13187 }, { "epoch": 9.661538461538461, "grad_norm": 1.109375, "learning_rate": 8.680514987176774e-05, "loss": 0.1377, "step": 13188 }, { "epoch": 9.662271062271062, "grad_norm": 1.34375, "learning_rate": 8.680282069157085e-05, "loss": 0.1321, "step": 13189 }, { "epoch": 9.663003663003662, "grad_norm": 0.73046875, "learning_rate": 8.680049133743525e-05, "loss": 0.1416, "step": 13190 }, { "epoch": 9.663736263736263, "grad_norm": 0.98828125, "learning_rate": 8.679816180937209e-05, "loss": 0.1494, "step": 13191 }, { "epoch": 9.664468864468864, "grad_norm": 1.4453125, "learning_rate": 8.679583210739251e-05, "loss": 0.1623, "step": 13192 }, { "epoch": 9.665201465201465, "grad_norm": 1.1328125, "learning_rate": 8.679350223150768e-05, "loss": 0.1309, "step": 13193 }, { "epoch": 9.665934065934065, "grad_norm": 1.8125, "learning_rate": 8.679117218172876e-05, "loss": 0.1581, "step": 13194 }, { "epoch": 9.666666666666666, "grad_norm": 0.99609375, "learning_rate": 8.678884195806692e-05, "loss": 0.1113, "step": 13195 }, { "epoch": 9.667399267399267, "grad_norm": 0.87109375, "learning_rate": 8.678651156053332e-05, "loss": 0.152, "step": 13196 }, { "epoch": 9.668131868131868, "grad_norm": 1.03125, "learning_rate": 8.678418098913915e-05, "loss": 0.1544, "step": 13197 }, { "epoch": 9.668864468864468, "grad_norm": 1.21875, "learning_rate": 8.678185024389555e-05, "loss": 0.1375, "step": 13198 }, { "epoch": 9.669597069597069, "grad_norm": 1.7265625, "learning_rate": 8.677951932481369e-05, "loss": 0.1807, "step": 13199 }, { "epoch": 9.67032967032967, "grad_norm": 1.03125, "learning_rate": 8.677718823190477e-05, "loss": 0.1254, "step": 13200 }, { "epoch": 9.67106227106227, "grad_norm": 1.421875, "learning_rate": 8.67748569651799e-05, "loss": 0.1341, "step": 13201 }, { "epoch": 9.671794871794871, "grad_norm": 1.453125, "learning_rate": 8.677252552465029e-05, "loss": 0.1645, "step": 13202 }, { "epoch": 9.672527472527472, "grad_norm": 1.4453125, "learning_rate": 8.67701939103271e-05, "loss": 0.1482, "step": 13203 }, { "epoch": 9.673260073260073, "grad_norm": 0.69921875, "learning_rate": 8.676786212222152e-05, "loss": 0.146, "step": 13204 }, { "epoch": 9.673992673992673, "grad_norm": 1.2578125, "learning_rate": 8.676553016034469e-05, "loss": 0.1161, "step": 13205 }, { "epoch": 9.674725274725274, "grad_norm": 0.69140625, "learning_rate": 8.676319802470782e-05, "loss": 0.1364, "step": 13206 }, { "epoch": 9.675457875457875, "grad_norm": 0.66015625, "learning_rate": 8.676086571532202e-05, "loss": 0.1429, "step": 13207 }, { "epoch": 9.676190476190476, "grad_norm": 0.84375, "learning_rate": 8.675853323219855e-05, "loss": 0.1282, "step": 13208 }, { "epoch": 9.676923076923076, "grad_norm": 0.9140625, "learning_rate": 8.675620057534853e-05, "loss": 0.1372, "step": 13209 }, { "epoch": 9.677655677655677, "grad_norm": 1.1953125, "learning_rate": 8.675386774478316e-05, "loss": 0.1253, "step": 13210 }, { "epoch": 9.678388278388278, "grad_norm": 0.7578125, "learning_rate": 8.67515347405136e-05, "loss": 0.1286, "step": 13211 }, { "epoch": 9.679120879120878, "grad_norm": 0.74609375, "learning_rate": 8.674920156255103e-05, "loss": 0.1418, "step": 13212 }, { "epoch": 9.67985347985348, "grad_norm": 0.984375, "learning_rate": 8.674686821090664e-05, "loss": 0.1392, "step": 13213 }, { "epoch": 9.68058608058608, "grad_norm": 1.1015625, "learning_rate": 8.674453468559161e-05, "loss": 0.1342, "step": 13214 }, { "epoch": 9.68131868131868, "grad_norm": 0.58984375, "learning_rate": 8.674220098661712e-05, "loss": 0.1196, "step": 13215 }, { "epoch": 9.682051282051281, "grad_norm": 0.9375, "learning_rate": 8.673986711399436e-05, "loss": 0.1226, "step": 13216 }, { "epoch": 9.682783882783882, "grad_norm": 1.4921875, "learning_rate": 8.673753306773446e-05, "loss": 0.1328, "step": 13217 }, { "epoch": 9.683516483516483, "grad_norm": 1.1015625, "learning_rate": 8.673519884784867e-05, "loss": 0.1353, "step": 13218 }, { "epoch": 9.684249084249084, "grad_norm": 0.7890625, "learning_rate": 8.673286445434815e-05, "loss": 0.1487, "step": 13219 }, { "epoch": 9.684981684981684, "grad_norm": 0.91796875, "learning_rate": 8.673052988724406e-05, "loss": 0.1481, "step": 13220 }, { "epoch": 9.685714285714285, "grad_norm": 0.9375, "learning_rate": 8.672819514654764e-05, "loss": 0.1269, "step": 13221 }, { "epoch": 9.686446886446886, "grad_norm": 1.2265625, "learning_rate": 8.672586023227001e-05, "loss": 0.1678, "step": 13222 }, { "epoch": 9.687179487179487, "grad_norm": 1.015625, "learning_rate": 8.672352514442243e-05, "loss": 0.116, "step": 13223 }, { "epoch": 9.687912087912087, "grad_norm": 1.1796875, "learning_rate": 8.672118988301602e-05, "loss": 0.1199, "step": 13224 }, { "epoch": 9.688644688644688, "grad_norm": 0.84375, "learning_rate": 8.6718854448062e-05, "loss": 0.1263, "step": 13225 }, { "epoch": 9.689377289377289, "grad_norm": 1.109375, "learning_rate": 8.671651883957157e-05, "loss": 0.1435, "step": 13226 }, { "epoch": 9.69010989010989, "grad_norm": 1.09375, "learning_rate": 8.67141830575559e-05, "loss": 0.127, "step": 13227 }, { "epoch": 9.69084249084249, "grad_norm": 1.5390625, "learning_rate": 8.67118471020262e-05, "loss": 0.1491, "step": 13228 }, { "epoch": 9.691575091575091, "grad_norm": 1.3125, "learning_rate": 8.670951097299364e-05, "loss": 0.1571, "step": 13229 }, { "epoch": 9.692307692307692, "grad_norm": 0.59375, "learning_rate": 8.670717467046942e-05, "loss": 0.1134, "step": 13230 }, { "epoch": 9.693040293040292, "grad_norm": 1.03125, "learning_rate": 8.670483819446475e-05, "loss": 0.1279, "step": 13231 }, { "epoch": 9.693772893772893, "grad_norm": 1.1640625, "learning_rate": 8.670250154499084e-05, "loss": 0.15, "step": 13232 }, { "epoch": 9.694505494505494, "grad_norm": 0.921875, "learning_rate": 8.670016472205884e-05, "loss": 0.1285, "step": 13233 }, { "epoch": 9.695238095238095, "grad_norm": 1.1953125, "learning_rate": 8.669782772567996e-05, "loss": 0.1801, "step": 13234 }, { "epoch": 9.695970695970695, "grad_norm": 1.1796875, "learning_rate": 8.66954905558654e-05, "loss": 0.1455, "step": 13235 }, { "epoch": 9.696703296703296, "grad_norm": 1.0625, "learning_rate": 8.669315321262637e-05, "loss": 0.1502, "step": 13236 }, { "epoch": 9.697435897435897, "grad_norm": 1.53125, "learning_rate": 8.669081569597408e-05, "loss": 0.1276, "step": 13237 }, { "epoch": 9.698168498168497, "grad_norm": 0.6875, "learning_rate": 8.66884780059197e-05, "loss": 0.1189, "step": 13238 }, { "epoch": 9.698901098901098, "grad_norm": 1.1171875, "learning_rate": 8.668614014247442e-05, "loss": 0.1249, "step": 13239 }, { "epoch": 9.699633699633699, "grad_norm": 0.671875, "learning_rate": 8.668380210564949e-05, "loss": 0.1238, "step": 13240 }, { "epoch": 9.7003663003663, "grad_norm": 0.79296875, "learning_rate": 8.668146389545608e-05, "loss": 0.1384, "step": 13241 }, { "epoch": 9.7010989010989, "grad_norm": 0.95703125, "learning_rate": 8.66791255119054e-05, "loss": 0.173, "step": 13242 }, { "epoch": 9.701831501831501, "grad_norm": 0.88671875, "learning_rate": 8.667678695500864e-05, "loss": 0.114, "step": 13243 }, { "epoch": 9.702564102564102, "grad_norm": 1.015625, "learning_rate": 8.667444822477703e-05, "loss": 0.1506, "step": 13244 }, { "epoch": 9.703296703296703, "grad_norm": 0.8203125, "learning_rate": 8.667210932122175e-05, "loss": 0.1486, "step": 13245 }, { "epoch": 9.704029304029303, "grad_norm": 0.94921875, "learning_rate": 8.666977024435404e-05, "loss": 0.1575, "step": 13246 }, { "epoch": 9.704761904761904, "grad_norm": 0.83203125, "learning_rate": 8.666743099418509e-05, "loss": 0.1449, "step": 13247 }, { "epoch": 9.705494505494505, "grad_norm": 1.3359375, "learning_rate": 8.666509157072607e-05, "loss": 0.112, "step": 13248 }, { "epoch": 9.706227106227105, "grad_norm": 0.95703125, "learning_rate": 8.666275197398825e-05, "loss": 0.1279, "step": 13249 }, { "epoch": 9.706959706959706, "grad_norm": 0.83203125, "learning_rate": 8.666041220398282e-05, "loss": 0.1369, "step": 13250 }, { "epoch": 9.707692307692307, "grad_norm": 0.99609375, "learning_rate": 8.665807226072096e-05, "loss": 0.117, "step": 13251 }, { "epoch": 9.708424908424908, "grad_norm": 1.203125, "learning_rate": 8.665573214421392e-05, "loss": 0.1467, "step": 13252 }, { "epoch": 9.709157509157508, "grad_norm": 1.4453125, "learning_rate": 8.66533918544729e-05, "loss": 0.1307, "step": 13253 }, { "epoch": 9.70989010989011, "grad_norm": 0.9375, "learning_rate": 8.665105139150911e-05, "loss": 0.1513, "step": 13254 }, { "epoch": 9.71062271062271, "grad_norm": 1.15625, "learning_rate": 8.664871075533377e-05, "loss": 0.1661, "step": 13255 }, { "epoch": 9.71135531135531, "grad_norm": 1.453125, "learning_rate": 8.664636994595807e-05, "loss": 0.1285, "step": 13256 }, { "epoch": 9.712087912087911, "grad_norm": 1.609375, "learning_rate": 8.664402896339328e-05, "loss": 0.134, "step": 13257 }, { "epoch": 9.712820512820512, "grad_norm": 1.359375, "learning_rate": 8.664168780765056e-05, "loss": 0.1479, "step": 13258 }, { "epoch": 9.713553113553113, "grad_norm": 1.3046875, "learning_rate": 8.663934647874115e-05, "loss": 0.1679, "step": 13259 }, { "epoch": 9.714285714285714, "grad_norm": 1.125, "learning_rate": 8.663700497667625e-05, "loss": 0.1585, "step": 13260 }, { "epoch": 9.715018315018314, "grad_norm": 0.85546875, "learning_rate": 8.663466330146712e-05, "loss": 0.1201, "step": 13261 }, { "epoch": 9.715750915750915, "grad_norm": 0.91796875, "learning_rate": 8.663232145312495e-05, "loss": 0.1331, "step": 13262 }, { "epoch": 9.716483516483516, "grad_norm": 1.25, "learning_rate": 8.662997943166096e-05, "loss": 0.1314, "step": 13263 }, { "epoch": 9.717216117216116, "grad_norm": 1.109375, "learning_rate": 8.662763723708639e-05, "loss": 0.0969, "step": 13264 }, { "epoch": 9.717948717948717, "grad_norm": 1.09375, "learning_rate": 8.662529486941245e-05, "loss": 0.1247, "step": 13265 }, { "epoch": 9.718681318681318, "grad_norm": 0.9296875, "learning_rate": 8.662295232865036e-05, "loss": 0.157, "step": 13266 }, { "epoch": 9.719413919413919, "grad_norm": 0.92578125, "learning_rate": 8.662060961481134e-05, "loss": 0.1175, "step": 13267 }, { "epoch": 9.72014652014652, "grad_norm": 1.234375, "learning_rate": 8.661826672790664e-05, "loss": 0.1326, "step": 13268 }, { "epoch": 9.72087912087912, "grad_norm": 1.6875, "learning_rate": 8.661592366794744e-05, "loss": 0.1474, "step": 13269 }, { "epoch": 9.72161172161172, "grad_norm": 1.046875, "learning_rate": 8.661358043494501e-05, "loss": 0.1706, "step": 13270 }, { "epoch": 9.722344322344322, "grad_norm": 1.171875, "learning_rate": 8.661123702891055e-05, "loss": 0.1538, "step": 13271 }, { "epoch": 9.723076923076922, "grad_norm": 1.1015625, "learning_rate": 8.660889344985531e-05, "loss": 0.1459, "step": 13272 }, { "epoch": 9.723809523809523, "grad_norm": 0.87890625, "learning_rate": 8.66065496977905e-05, "loss": 0.1592, "step": 13273 }, { "epoch": 9.724542124542124, "grad_norm": 0.84765625, "learning_rate": 8.660420577272736e-05, "loss": 0.1044, "step": 13274 }, { "epoch": 9.725274725274724, "grad_norm": 0.8828125, "learning_rate": 8.660186167467713e-05, "loss": 0.1174, "step": 13275 }, { "epoch": 9.726007326007325, "grad_norm": 1.5546875, "learning_rate": 8.659951740365103e-05, "loss": 0.1627, "step": 13276 }, { "epoch": 9.726739926739926, "grad_norm": 1.484375, "learning_rate": 8.659717295966026e-05, "loss": 0.1369, "step": 13277 }, { "epoch": 9.727472527472527, "grad_norm": 2.140625, "learning_rate": 8.659482834271612e-05, "loss": 0.1603, "step": 13278 }, { "epoch": 9.728205128205127, "grad_norm": 0.78515625, "learning_rate": 8.659248355282979e-05, "loss": 0.1175, "step": 13279 }, { "epoch": 9.728937728937728, "grad_norm": 1.2578125, "learning_rate": 8.659013859001252e-05, "loss": 0.1446, "step": 13280 }, { "epoch": 9.729670329670329, "grad_norm": 0.82421875, "learning_rate": 8.658779345427557e-05, "loss": 0.1392, "step": 13281 }, { "epoch": 9.73040293040293, "grad_norm": 1.5546875, "learning_rate": 8.658544814563013e-05, "loss": 0.1128, "step": 13282 }, { "epoch": 9.73113553113553, "grad_norm": 1.1640625, "learning_rate": 8.658310266408748e-05, "loss": 0.163, "step": 13283 }, { "epoch": 9.731868131868131, "grad_norm": 0.859375, "learning_rate": 8.658075700965882e-05, "loss": 0.1477, "step": 13284 }, { "epoch": 9.732600732600732, "grad_norm": 0.828125, "learning_rate": 8.657841118235541e-05, "loss": 0.12, "step": 13285 }, { "epoch": 9.733333333333333, "grad_norm": 0.671875, "learning_rate": 8.657606518218851e-05, "loss": 0.119, "step": 13286 }, { "epoch": 9.734065934065933, "grad_norm": 1.1875, "learning_rate": 8.657371900916931e-05, "loss": 0.1289, "step": 13287 }, { "epoch": 9.734798534798536, "grad_norm": 0.921875, "learning_rate": 8.657137266330912e-05, "loss": 0.1586, "step": 13288 }, { "epoch": 9.735531135531136, "grad_norm": 1.0234375, "learning_rate": 8.656902614461911e-05, "loss": 0.1271, "step": 13289 }, { "epoch": 9.736263736263737, "grad_norm": 0.71484375, "learning_rate": 8.656667945311055e-05, "loss": 0.1293, "step": 13290 }, { "epoch": 9.736996336996338, "grad_norm": 1.2265625, "learning_rate": 8.65643325887947e-05, "loss": 0.1298, "step": 13291 }, { "epoch": 9.737728937728939, "grad_norm": 0.9375, "learning_rate": 8.65619855516828e-05, "loss": 0.1488, "step": 13292 }, { "epoch": 9.73846153846154, "grad_norm": 0.98046875, "learning_rate": 8.655963834178608e-05, "loss": 0.1673, "step": 13293 }, { "epoch": 9.73919413919414, "grad_norm": 1.171875, "learning_rate": 8.655729095911581e-05, "loss": 0.1299, "step": 13294 }, { "epoch": 9.73992673992674, "grad_norm": 1.109375, "learning_rate": 8.65549434036832e-05, "loss": 0.1496, "step": 13295 }, { "epoch": 9.740659340659342, "grad_norm": 0.9765625, "learning_rate": 8.655259567549953e-05, "loss": 0.1372, "step": 13296 }, { "epoch": 9.741391941391942, "grad_norm": 1.2109375, "learning_rate": 8.655024777457606e-05, "loss": 0.1573, "step": 13297 }, { "epoch": 9.742124542124543, "grad_norm": 1.5234375, "learning_rate": 8.6547899700924e-05, "loss": 0.1197, "step": 13298 }, { "epoch": 9.742857142857144, "grad_norm": 1.3984375, "learning_rate": 8.65455514545546e-05, "loss": 0.1462, "step": 13299 }, { "epoch": 9.743589743589745, "grad_norm": 1.171875, "learning_rate": 8.654320303547916e-05, "loss": 0.1407, "step": 13300 }, { "epoch": 9.744322344322345, "grad_norm": 1.234375, "learning_rate": 8.65408544437089e-05, "loss": 0.1122, "step": 13301 }, { "epoch": 9.745054945054946, "grad_norm": 1.0625, "learning_rate": 8.653850567925505e-05, "loss": 0.1426, "step": 13302 }, { "epoch": 9.745787545787547, "grad_norm": 0.76171875, "learning_rate": 8.653615674212892e-05, "loss": 0.1186, "step": 13303 }, { "epoch": 9.746520146520147, "grad_norm": 0.86328125, "learning_rate": 8.653380763234171e-05, "loss": 0.0954, "step": 13304 }, { "epoch": 9.747252747252748, "grad_norm": 0.7421875, "learning_rate": 8.65314583499047e-05, "loss": 0.129, "step": 13305 }, { "epoch": 9.747985347985349, "grad_norm": 0.9453125, "learning_rate": 8.652910889482915e-05, "loss": 0.1177, "step": 13306 }, { "epoch": 9.74871794871795, "grad_norm": 0.90234375, "learning_rate": 8.652675926712632e-05, "loss": 0.1286, "step": 13307 }, { "epoch": 9.74945054945055, "grad_norm": 0.92578125, "learning_rate": 8.652440946680745e-05, "loss": 0.1269, "step": 13308 }, { "epoch": 9.750183150183151, "grad_norm": 1.046875, "learning_rate": 8.652205949388381e-05, "loss": 0.1258, "step": 13309 }, { "epoch": 9.750915750915752, "grad_norm": 0.78125, "learning_rate": 8.651970934836666e-05, "loss": 0.1422, "step": 13310 }, { "epoch": 9.751648351648353, "grad_norm": 0.87109375, "learning_rate": 8.651735903026725e-05, "loss": 0.1278, "step": 13311 }, { "epoch": 9.752380952380953, "grad_norm": 0.71875, "learning_rate": 8.651500853959685e-05, "loss": 0.1269, "step": 13312 }, { "epoch": 9.753113553113554, "grad_norm": 0.703125, "learning_rate": 8.651265787636672e-05, "loss": 0.1345, "step": 13313 }, { "epoch": 9.753846153846155, "grad_norm": 0.921875, "learning_rate": 8.651030704058813e-05, "loss": 0.1105, "step": 13314 }, { "epoch": 9.754578754578755, "grad_norm": 0.71875, "learning_rate": 8.650795603227235e-05, "loss": 0.1377, "step": 13315 }, { "epoch": 9.755311355311356, "grad_norm": 0.6171875, "learning_rate": 8.65056048514306e-05, "loss": 0.1301, "step": 13316 }, { "epoch": 9.756043956043957, "grad_norm": 1.3203125, "learning_rate": 8.650325349807421e-05, "loss": 0.1353, "step": 13317 }, { "epoch": 9.756776556776558, "grad_norm": 1.921875, "learning_rate": 8.65009019722144e-05, "loss": 0.1347, "step": 13318 }, { "epoch": 9.757509157509158, "grad_norm": 1.3359375, "learning_rate": 8.649855027386246e-05, "loss": 0.161, "step": 13319 }, { "epoch": 9.758241758241759, "grad_norm": 1.328125, "learning_rate": 8.649619840302963e-05, "loss": 0.1563, "step": 13320 }, { "epoch": 9.75897435897436, "grad_norm": 1.1640625, "learning_rate": 8.64938463597272e-05, "loss": 0.1349, "step": 13321 }, { "epoch": 9.75970695970696, "grad_norm": 0.9296875, "learning_rate": 8.649149414396645e-05, "loss": 0.1271, "step": 13322 }, { "epoch": 9.760439560439561, "grad_norm": 1.4375, "learning_rate": 8.648914175575861e-05, "loss": 0.184, "step": 13323 }, { "epoch": 9.761172161172162, "grad_norm": 0.61328125, "learning_rate": 8.648678919511502e-05, "loss": 0.1206, "step": 13324 }, { "epoch": 9.761904761904763, "grad_norm": 1.3828125, "learning_rate": 8.648443646204686e-05, "loss": 0.1572, "step": 13325 }, { "epoch": 9.762637362637363, "grad_norm": 1.015625, "learning_rate": 8.648208355656549e-05, "loss": 0.1379, "step": 13326 }, { "epoch": 9.763369963369964, "grad_norm": 1.3125, "learning_rate": 8.647973047868214e-05, "loss": 0.1465, "step": 13327 }, { "epoch": 9.764102564102565, "grad_norm": 1.2734375, "learning_rate": 8.647737722840808e-05, "loss": 0.1684, "step": 13328 }, { "epoch": 9.764835164835166, "grad_norm": 0.71875, "learning_rate": 8.647502380575458e-05, "loss": 0.1095, "step": 13329 }, { "epoch": 9.765567765567766, "grad_norm": 0.6484375, "learning_rate": 8.647267021073295e-05, "loss": 0.1236, "step": 13330 }, { "epoch": 9.766300366300367, "grad_norm": 0.8828125, "learning_rate": 8.647031644335445e-05, "loss": 0.1188, "step": 13331 }, { "epoch": 9.767032967032968, "grad_norm": 0.98046875, "learning_rate": 8.646796250363034e-05, "loss": 0.1099, "step": 13332 }, { "epoch": 9.767765567765569, "grad_norm": 1.2578125, "learning_rate": 8.646560839157194e-05, "loss": 0.1676, "step": 13333 }, { "epoch": 9.76849816849817, "grad_norm": 1.0234375, "learning_rate": 8.646325410719048e-05, "loss": 0.1461, "step": 13334 }, { "epoch": 9.76923076923077, "grad_norm": 1.2578125, "learning_rate": 8.646089965049727e-05, "loss": 0.1371, "step": 13335 }, { "epoch": 9.76996336996337, "grad_norm": 1.4375, "learning_rate": 8.645854502150357e-05, "loss": 0.1564, "step": 13336 }, { "epoch": 9.770695970695972, "grad_norm": 0.98046875, "learning_rate": 8.64561902202207e-05, "loss": 0.1284, "step": 13337 }, { "epoch": 9.771428571428572, "grad_norm": 1.125, "learning_rate": 8.645383524665993e-05, "loss": 0.1411, "step": 13338 }, { "epoch": 9.772161172161173, "grad_norm": 1.0390625, "learning_rate": 8.64514801008325e-05, "loss": 0.1268, "step": 13339 }, { "epoch": 9.772893772893774, "grad_norm": 1.125, "learning_rate": 8.644912478274975e-05, "loss": 0.1264, "step": 13340 }, { "epoch": 9.773626373626374, "grad_norm": 1.0546875, "learning_rate": 8.644676929242294e-05, "loss": 0.1371, "step": 13341 }, { "epoch": 9.774358974358975, "grad_norm": 1.9921875, "learning_rate": 8.644441362986334e-05, "loss": 0.1669, "step": 13342 }, { "epoch": 9.775091575091576, "grad_norm": 1.1484375, "learning_rate": 8.644205779508226e-05, "loss": 0.1142, "step": 13343 }, { "epoch": 9.775824175824177, "grad_norm": 1.7109375, "learning_rate": 8.6439701788091e-05, "loss": 0.1236, "step": 13344 }, { "epoch": 9.776556776556777, "grad_norm": 1.53125, "learning_rate": 8.64373456089008e-05, "loss": 0.144, "step": 13345 }, { "epoch": 9.777289377289378, "grad_norm": 1.046875, "learning_rate": 8.643498925752302e-05, "loss": 0.1403, "step": 13346 }, { "epoch": 9.778021978021979, "grad_norm": 0.6640625, "learning_rate": 8.643263273396887e-05, "loss": 0.0927, "step": 13347 }, { "epoch": 9.77875457875458, "grad_norm": 0.796875, "learning_rate": 8.64302760382497e-05, "loss": 0.1096, "step": 13348 }, { "epoch": 9.77948717948718, "grad_norm": 0.84375, "learning_rate": 8.642791917037679e-05, "loss": 0.128, "step": 13349 }, { "epoch": 9.780219780219781, "grad_norm": 1.078125, "learning_rate": 8.642556213036142e-05, "loss": 0.1094, "step": 13350 }, { "epoch": 9.780952380952382, "grad_norm": 1.15625, "learning_rate": 8.642320491821488e-05, "loss": 0.1528, "step": 13351 }, { "epoch": 9.781684981684982, "grad_norm": 0.80078125, "learning_rate": 8.642084753394848e-05, "loss": 0.1154, "step": 13352 }, { "epoch": 9.782417582417583, "grad_norm": 0.8046875, "learning_rate": 8.64184899775735e-05, "loss": 0.1116, "step": 13353 }, { "epoch": 9.783150183150184, "grad_norm": 0.8203125, "learning_rate": 8.641613224910126e-05, "loss": 0.1534, "step": 13354 }, { "epoch": 9.783882783882785, "grad_norm": 0.7265625, "learning_rate": 8.641377434854302e-05, "loss": 0.1115, "step": 13355 }, { "epoch": 9.784615384615385, "grad_norm": 1.15625, "learning_rate": 8.64114162759101e-05, "loss": 0.1527, "step": 13356 }, { "epoch": 9.785347985347986, "grad_norm": 1.40625, "learning_rate": 8.640905803121379e-05, "loss": 0.1219, "step": 13357 }, { "epoch": 9.786080586080587, "grad_norm": 0.9609375, "learning_rate": 8.640669961446541e-05, "loss": 0.1253, "step": 13358 }, { "epoch": 9.786813186813188, "grad_norm": 0.7578125, "learning_rate": 8.640434102567625e-05, "loss": 0.1125, "step": 13359 }, { "epoch": 9.787545787545788, "grad_norm": 1.0, "learning_rate": 8.640198226485759e-05, "loss": 0.1248, "step": 13360 }, { "epoch": 9.788278388278389, "grad_norm": 1.015625, "learning_rate": 8.639962333202075e-05, "loss": 0.1616, "step": 13361 }, { "epoch": 9.78901098901099, "grad_norm": 0.90234375, "learning_rate": 8.639726422717704e-05, "loss": 0.1505, "step": 13362 }, { "epoch": 9.78974358974359, "grad_norm": 1.9609375, "learning_rate": 8.639490495033774e-05, "loss": 0.1487, "step": 13363 }, { "epoch": 9.790476190476191, "grad_norm": 1.5546875, "learning_rate": 8.639254550151418e-05, "loss": 0.1407, "step": 13364 }, { "epoch": 9.791208791208792, "grad_norm": 1.6328125, "learning_rate": 8.639018588071763e-05, "loss": 0.1147, "step": 13365 }, { "epoch": 9.791941391941393, "grad_norm": 0.8515625, "learning_rate": 8.638782608795943e-05, "loss": 0.1281, "step": 13366 }, { "epoch": 9.792673992673993, "grad_norm": 0.9765625, "learning_rate": 8.638546612325089e-05, "loss": 0.139, "step": 13367 }, { "epoch": 9.793406593406594, "grad_norm": 0.94140625, "learning_rate": 8.638310598660327e-05, "loss": 0.1223, "step": 13368 }, { "epoch": 9.794139194139195, "grad_norm": 1.328125, "learning_rate": 8.638074567802792e-05, "loss": 0.1111, "step": 13369 }, { "epoch": 9.794871794871796, "grad_norm": 0.90234375, "learning_rate": 8.637838519753615e-05, "loss": 0.1388, "step": 13370 }, { "epoch": 9.795604395604396, "grad_norm": 1.171875, "learning_rate": 8.637602454513925e-05, "loss": 0.1392, "step": 13371 }, { "epoch": 9.796336996336997, "grad_norm": 1.109375, "learning_rate": 8.637366372084854e-05, "loss": 0.1262, "step": 13372 }, { "epoch": 9.797069597069598, "grad_norm": 0.671875, "learning_rate": 8.637130272467534e-05, "loss": 0.115, "step": 13373 }, { "epoch": 9.797802197802199, "grad_norm": 0.953125, "learning_rate": 8.636894155663094e-05, "loss": 0.1118, "step": 13374 }, { "epoch": 9.7985347985348, "grad_norm": 1.171875, "learning_rate": 8.636658021672668e-05, "loss": 0.1214, "step": 13375 }, { "epoch": 9.7992673992674, "grad_norm": 2.140625, "learning_rate": 8.636421870497384e-05, "loss": 0.1781, "step": 13376 }, { "epoch": 9.8, "grad_norm": 1.2734375, "learning_rate": 8.636185702138377e-05, "loss": 0.1476, "step": 13377 }, { "epoch": 9.800732600732601, "grad_norm": 0.625, "learning_rate": 8.635949516596778e-05, "loss": 0.1243, "step": 13378 }, { "epoch": 9.801465201465202, "grad_norm": 0.9921875, "learning_rate": 8.635713313873715e-05, "loss": 0.108, "step": 13379 }, { "epoch": 9.802197802197803, "grad_norm": 0.7734375, "learning_rate": 8.635477093970325e-05, "loss": 0.1314, "step": 13380 }, { "epoch": 9.802930402930404, "grad_norm": 0.73828125, "learning_rate": 8.635240856887735e-05, "loss": 0.1244, "step": 13381 }, { "epoch": 9.803663003663004, "grad_norm": 0.5703125, "learning_rate": 8.635004602627079e-05, "loss": 0.1297, "step": 13382 }, { "epoch": 9.804395604395605, "grad_norm": 1.875, "learning_rate": 8.63476833118949e-05, "loss": 0.1707, "step": 13383 }, { "epoch": 9.805128205128206, "grad_norm": 1.3359375, "learning_rate": 8.6345320425761e-05, "loss": 0.1531, "step": 13384 }, { "epoch": 9.805860805860807, "grad_norm": 1.3046875, "learning_rate": 8.634295736788039e-05, "loss": 0.1092, "step": 13385 }, { "epoch": 9.806593406593407, "grad_norm": 0.7109375, "learning_rate": 8.634059413826442e-05, "loss": 0.1211, "step": 13386 }, { "epoch": 9.807326007326008, "grad_norm": 1.4765625, "learning_rate": 8.633823073692437e-05, "loss": 0.1532, "step": 13387 }, { "epoch": 9.808058608058609, "grad_norm": 0.94921875, "learning_rate": 8.633586716387161e-05, "loss": 0.1343, "step": 13388 }, { "epoch": 9.80879120879121, "grad_norm": 1.1171875, "learning_rate": 8.633350341911744e-05, "loss": 0.1354, "step": 13389 }, { "epoch": 9.80952380952381, "grad_norm": 0.61328125, "learning_rate": 8.633113950267319e-05, "loss": 0.1252, "step": 13390 }, { "epoch": 9.810256410256411, "grad_norm": 0.85546875, "learning_rate": 8.632877541455021e-05, "loss": 0.1115, "step": 13391 }, { "epoch": 9.810989010989012, "grad_norm": 1.2890625, "learning_rate": 8.632641115475979e-05, "loss": 0.142, "step": 13392 }, { "epoch": 9.811721611721612, "grad_norm": 0.85546875, "learning_rate": 8.632404672331327e-05, "loss": 0.1024, "step": 13393 }, { "epoch": 9.812454212454213, "grad_norm": 0.953125, "learning_rate": 8.632168212022199e-05, "loss": 0.1325, "step": 13394 }, { "epoch": 9.813186813186814, "grad_norm": 0.71875, "learning_rate": 8.631931734549726e-05, "loss": 0.1279, "step": 13395 }, { "epoch": 9.813919413919415, "grad_norm": 1.140625, "learning_rate": 8.631695239915043e-05, "loss": 0.1032, "step": 13396 }, { "epoch": 9.814652014652015, "grad_norm": 0.63671875, "learning_rate": 8.631458728119282e-05, "loss": 0.1295, "step": 13397 }, { "epoch": 9.815384615384616, "grad_norm": 0.9375, "learning_rate": 8.631222199163576e-05, "loss": 0.1345, "step": 13398 }, { "epoch": 9.816117216117217, "grad_norm": 1.0859375, "learning_rate": 8.630985653049061e-05, "loss": 0.174, "step": 13399 }, { "epoch": 9.816849816849818, "grad_norm": 0.9921875, "learning_rate": 8.630749089776865e-05, "loss": 0.1134, "step": 13400 }, { "epoch": 9.817582417582418, "grad_norm": 1.0078125, "learning_rate": 8.630512509348128e-05, "loss": 0.1271, "step": 13401 }, { "epoch": 9.818315018315019, "grad_norm": 1.1640625, "learning_rate": 8.630275911763976e-05, "loss": 0.1397, "step": 13402 }, { "epoch": 9.81904761904762, "grad_norm": 1.1953125, "learning_rate": 8.630039297025551e-05, "loss": 0.1397, "step": 13403 }, { "epoch": 9.81978021978022, "grad_norm": 0.6796875, "learning_rate": 8.62980266513398e-05, "loss": 0.1221, "step": 13404 }, { "epoch": 9.820512820512821, "grad_norm": 0.8359375, "learning_rate": 8.6295660160904e-05, "loss": 0.1392, "step": 13405 }, { "epoch": 9.821245421245422, "grad_norm": 1.2578125, "learning_rate": 8.629329349895944e-05, "loss": 0.1412, "step": 13406 }, { "epoch": 9.821978021978023, "grad_norm": 1.0, "learning_rate": 8.629092666551746e-05, "loss": 0.154, "step": 13407 }, { "epoch": 9.822710622710623, "grad_norm": 0.75390625, "learning_rate": 8.628855966058941e-05, "loss": 0.1215, "step": 13408 }, { "epoch": 9.823443223443224, "grad_norm": 0.859375, "learning_rate": 8.628619248418661e-05, "loss": 0.1284, "step": 13409 }, { "epoch": 9.824175824175825, "grad_norm": 1.1171875, "learning_rate": 8.628382513632042e-05, "loss": 0.1615, "step": 13410 }, { "epoch": 9.824908424908426, "grad_norm": 0.94140625, "learning_rate": 8.628145761700218e-05, "loss": 0.1327, "step": 13411 }, { "epoch": 9.825641025641026, "grad_norm": 1.0625, "learning_rate": 8.627908992624321e-05, "loss": 0.1157, "step": 13412 }, { "epoch": 9.826373626373627, "grad_norm": 1.3828125, "learning_rate": 8.62767220640549e-05, "loss": 0.1386, "step": 13413 }, { "epoch": 9.827106227106228, "grad_norm": 1.0234375, "learning_rate": 8.627435403044856e-05, "loss": 0.1731, "step": 13414 }, { "epoch": 9.827838827838828, "grad_norm": 0.96875, "learning_rate": 8.627198582543554e-05, "loss": 0.1222, "step": 13415 }, { "epoch": 9.82857142857143, "grad_norm": 0.87890625, "learning_rate": 8.62696174490272e-05, "loss": 0.1403, "step": 13416 }, { "epoch": 9.82930402930403, "grad_norm": 0.9921875, "learning_rate": 8.626724890123489e-05, "loss": 0.1218, "step": 13417 }, { "epoch": 9.83003663003663, "grad_norm": 0.8203125, "learning_rate": 8.626488018206993e-05, "loss": 0.0927, "step": 13418 }, { "epoch": 9.830769230769231, "grad_norm": 0.7734375, "learning_rate": 8.626251129154368e-05, "loss": 0.1212, "step": 13419 }, { "epoch": 9.831501831501832, "grad_norm": 0.99609375, "learning_rate": 8.626014222966753e-05, "loss": 0.1302, "step": 13420 }, { "epoch": 9.832234432234433, "grad_norm": 0.9296875, "learning_rate": 8.625777299645277e-05, "loss": 0.1401, "step": 13421 }, { "epoch": 9.832967032967034, "grad_norm": 1.1328125, "learning_rate": 8.625540359191081e-05, "loss": 0.1747, "step": 13422 }, { "epoch": 9.833699633699634, "grad_norm": 1.4375, "learning_rate": 8.625303401605296e-05, "loss": 0.158, "step": 13423 }, { "epoch": 9.834432234432235, "grad_norm": 1.234375, "learning_rate": 8.625066426889058e-05, "loss": 0.1428, "step": 13424 }, { "epoch": 9.835164835164836, "grad_norm": 1.046875, "learning_rate": 8.624829435043505e-05, "loss": 0.1196, "step": 13425 }, { "epoch": 9.835897435897436, "grad_norm": 1.71875, "learning_rate": 8.624592426069768e-05, "loss": 0.1428, "step": 13426 }, { "epoch": 9.836630036630037, "grad_norm": 1.015625, "learning_rate": 8.624355399968987e-05, "loss": 0.1696, "step": 13427 }, { "epoch": 9.837362637362638, "grad_norm": 1.3125, "learning_rate": 8.624118356742294e-05, "loss": 0.1414, "step": 13428 }, { "epoch": 9.838095238095239, "grad_norm": 1.0625, "learning_rate": 8.623881296390829e-05, "loss": 0.1269, "step": 13429 }, { "epoch": 9.83882783882784, "grad_norm": 1.3671875, "learning_rate": 8.623644218915726e-05, "loss": 0.117, "step": 13430 }, { "epoch": 9.83956043956044, "grad_norm": 1.3046875, "learning_rate": 8.623407124318117e-05, "loss": 0.1653, "step": 13431 }, { "epoch": 9.84029304029304, "grad_norm": 1.265625, "learning_rate": 8.623170012599142e-05, "loss": 0.1692, "step": 13432 }, { "epoch": 9.841025641025642, "grad_norm": 0.90234375, "learning_rate": 8.62293288375994e-05, "loss": 0.1188, "step": 13433 }, { "epoch": 9.841758241758242, "grad_norm": 0.54296875, "learning_rate": 8.622695737801642e-05, "loss": 0.1237, "step": 13434 }, { "epoch": 9.842490842490843, "grad_norm": 0.74609375, "learning_rate": 8.622458574725383e-05, "loss": 0.1257, "step": 13435 }, { "epoch": 9.843223443223444, "grad_norm": 1.1796875, "learning_rate": 8.622221394532306e-05, "loss": 0.1328, "step": 13436 }, { "epoch": 9.843956043956045, "grad_norm": 0.91796875, "learning_rate": 8.621984197223543e-05, "loss": 0.1216, "step": 13437 }, { "epoch": 9.844688644688645, "grad_norm": 1.234375, "learning_rate": 8.621746982800229e-05, "loss": 0.1521, "step": 13438 }, { "epoch": 9.845421245421246, "grad_norm": 0.87109375, "learning_rate": 8.621509751263504e-05, "loss": 0.156, "step": 13439 }, { "epoch": 9.846153846153847, "grad_norm": 1.1953125, "learning_rate": 8.621272502614503e-05, "loss": 0.1159, "step": 13440 }, { "epoch": 9.846886446886447, "grad_norm": 0.8984375, "learning_rate": 8.621035236854362e-05, "loss": 0.1253, "step": 13441 }, { "epoch": 9.847619047619048, "grad_norm": 0.890625, "learning_rate": 8.620797953984222e-05, "loss": 0.1464, "step": 13442 }, { "epoch": 9.848351648351649, "grad_norm": 1.0859375, "learning_rate": 8.620560654005216e-05, "loss": 0.1411, "step": 13443 }, { "epoch": 9.84908424908425, "grad_norm": 0.96875, "learning_rate": 8.62032333691848e-05, "loss": 0.1612, "step": 13444 }, { "epoch": 9.84981684981685, "grad_norm": 1.25, "learning_rate": 8.620086002725153e-05, "loss": 0.1468, "step": 13445 }, { "epoch": 9.850549450549451, "grad_norm": 1.1328125, "learning_rate": 8.619848651426372e-05, "loss": 0.1445, "step": 13446 }, { "epoch": 9.851282051282052, "grad_norm": 0.8828125, "learning_rate": 8.619611283023275e-05, "loss": 0.144, "step": 13447 }, { "epoch": 9.852014652014653, "grad_norm": 0.9375, "learning_rate": 8.619373897516999e-05, "loss": 0.1361, "step": 13448 }, { "epoch": 9.852747252747253, "grad_norm": 1.0546875, "learning_rate": 8.619136494908681e-05, "loss": 0.1451, "step": 13449 }, { "epoch": 9.853479853479854, "grad_norm": 1.390625, "learning_rate": 8.618899075199457e-05, "loss": 0.1248, "step": 13450 }, { "epoch": 9.854212454212455, "grad_norm": 1.5078125, "learning_rate": 8.618661638390466e-05, "loss": 0.1654, "step": 13451 }, { "epoch": 9.854945054945055, "grad_norm": 0.82421875, "learning_rate": 8.618424184482847e-05, "loss": 0.1096, "step": 13452 }, { "epoch": 9.855677655677656, "grad_norm": 1.125, "learning_rate": 8.618186713477733e-05, "loss": 0.1377, "step": 13453 }, { "epoch": 9.856410256410257, "grad_norm": 1.3828125, "learning_rate": 8.61794922537627e-05, "loss": 0.1417, "step": 13454 }, { "epoch": 9.857142857142858, "grad_norm": 0.99609375, "learning_rate": 8.617711720179587e-05, "loss": 0.1274, "step": 13455 }, { "epoch": 9.857875457875458, "grad_norm": 1.2265625, "learning_rate": 8.617474197888829e-05, "loss": 0.1287, "step": 13456 }, { "epoch": 9.858608058608059, "grad_norm": 0.72265625, "learning_rate": 8.617236658505128e-05, "loss": 0.1312, "step": 13457 }, { "epoch": 9.85934065934066, "grad_norm": 1.1875, "learning_rate": 8.616999102029626e-05, "loss": 0.1374, "step": 13458 }, { "epoch": 9.86007326007326, "grad_norm": 1.1953125, "learning_rate": 8.616761528463462e-05, "loss": 0.1707, "step": 13459 }, { "epoch": 9.860805860805861, "grad_norm": 0.69140625, "learning_rate": 8.616523937807772e-05, "loss": 0.134, "step": 13460 }, { "epoch": 9.861538461538462, "grad_norm": 0.984375, "learning_rate": 8.616286330063694e-05, "loss": 0.1751, "step": 13461 }, { "epoch": 9.862271062271063, "grad_norm": 1.1484375, "learning_rate": 8.616048705232369e-05, "loss": 0.1264, "step": 13462 }, { "epoch": 9.863003663003663, "grad_norm": 0.875, "learning_rate": 8.615811063314931e-05, "loss": 0.1063, "step": 13463 }, { "epoch": 9.863736263736264, "grad_norm": 0.97265625, "learning_rate": 8.615573404312524e-05, "loss": 0.1261, "step": 13464 }, { "epoch": 9.864468864468865, "grad_norm": 0.6484375, "learning_rate": 8.615335728226284e-05, "loss": 0.1225, "step": 13465 }, { "epoch": 9.865201465201466, "grad_norm": 1.5234375, "learning_rate": 8.61509803505735e-05, "loss": 0.1548, "step": 13466 }, { "epoch": 9.865934065934066, "grad_norm": 1.390625, "learning_rate": 8.614860324806862e-05, "loss": 0.1487, "step": 13467 }, { "epoch": 9.866666666666667, "grad_norm": 0.87109375, "learning_rate": 8.614622597475956e-05, "loss": 0.1404, "step": 13468 }, { "epoch": 9.867399267399268, "grad_norm": 0.859375, "learning_rate": 8.614384853065773e-05, "loss": 0.1061, "step": 13469 }, { "epoch": 9.868131868131869, "grad_norm": 0.99609375, "learning_rate": 8.614147091577453e-05, "loss": 0.1311, "step": 13470 }, { "epoch": 9.86886446886447, "grad_norm": 0.640625, "learning_rate": 8.613909313012134e-05, "loss": 0.1017, "step": 13471 }, { "epoch": 9.86959706959707, "grad_norm": 1.59375, "learning_rate": 8.613671517370956e-05, "loss": 0.137, "step": 13472 }, { "epoch": 9.87032967032967, "grad_norm": 1.25, "learning_rate": 8.613433704655058e-05, "loss": 0.1121, "step": 13473 }, { "epoch": 9.871062271062272, "grad_norm": 0.8515625, "learning_rate": 8.613195874865579e-05, "loss": 0.1212, "step": 13474 }, { "epoch": 9.871794871794872, "grad_norm": 0.62109375, "learning_rate": 8.612958028003657e-05, "loss": 0.1242, "step": 13475 }, { "epoch": 9.872527472527473, "grad_norm": 1.90625, "learning_rate": 8.612720164070437e-05, "loss": 0.178, "step": 13476 }, { "epoch": 9.873260073260074, "grad_norm": 2.015625, "learning_rate": 8.612482283067052e-05, "loss": 0.1692, "step": 13477 }, { "epoch": 9.873992673992674, "grad_norm": 0.96484375, "learning_rate": 8.612244384994646e-05, "loss": 0.1265, "step": 13478 }, { "epoch": 9.874725274725275, "grad_norm": 1.078125, "learning_rate": 8.61200646985436e-05, "loss": 0.1208, "step": 13479 }, { "epoch": 9.875457875457876, "grad_norm": 1.25, "learning_rate": 8.611768537647328e-05, "loss": 0.116, "step": 13480 }, { "epoch": 9.876190476190477, "grad_norm": 1.1328125, "learning_rate": 8.611530588374696e-05, "loss": 0.159, "step": 13481 }, { "epoch": 9.876923076923077, "grad_norm": 1.90625, "learning_rate": 8.611292622037601e-05, "loss": 0.1558, "step": 13482 }, { "epoch": 9.877655677655678, "grad_norm": 0.91015625, "learning_rate": 8.611054638637183e-05, "loss": 0.127, "step": 13483 }, { "epoch": 9.878388278388279, "grad_norm": 1.734375, "learning_rate": 8.610816638174586e-05, "loss": 0.1679, "step": 13484 }, { "epoch": 9.87912087912088, "grad_norm": 0.99609375, "learning_rate": 8.610578620650946e-05, "loss": 0.1492, "step": 13485 }, { "epoch": 9.87985347985348, "grad_norm": 0.9140625, "learning_rate": 8.610340586067403e-05, "loss": 0.1074, "step": 13486 }, { "epoch": 9.880586080586081, "grad_norm": 1.0078125, "learning_rate": 8.610102534425102e-05, "loss": 0.1452, "step": 13487 }, { "epoch": 9.881318681318682, "grad_norm": 1.359375, "learning_rate": 8.609864465725179e-05, "loss": 0.214, "step": 13488 }, { "epoch": 9.882051282051282, "grad_norm": 1.8515625, "learning_rate": 8.609626379968779e-05, "loss": 0.1879, "step": 13489 }, { "epoch": 9.882783882783883, "grad_norm": 1.265625, "learning_rate": 8.609388277157039e-05, "loss": 0.1307, "step": 13490 }, { "epoch": 9.883516483516484, "grad_norm": 1.0, "learning_rate": 8.609150157291101e-05, "loss": 0.1671, "step": 13491 }, { "epoch": 9.884249084249085, "grad_norm": 0.86328125, "learning_rate": 8.608912020372108e-05, "loss": 0.1254, "step": 13492 }, { "epoch": 9.884981684981685, "grad_norm": 1.2890625, "learning_rate": 8.608673866401198e-05, "loss": 0.1079, "step": 13493 }, { "epoch": 9.885714285714286, "grad_norm": 1.0078125, "learning_rate": 8.608435695379513e-05, "loss": 0.1232, "step": 13494 }, { "epoch": 9.886446886446887, "grad_norm": 0.75390625, "learning_rate": 8.608197507308196e-05, "loss": 0.1331, "step": 13495 }, { "epoch": 9.887179487179488, "grad_norm": 0.8828125, "learning_rate": 8.607959302188385e-05, "loss": 0.1662, "step": 13496 }, { "epoch": 9.887912087912088, "grad_norm": 0.81640625, "learning_rate": 8.607721080021222e-05, "loss": 0.115, "step": 13497 }, { "epoch": 9.888644688644689, "grad_norm": 0.8125, "learning_rate": 8.607482840807852e-05, "loss": 0.1404, "step": 13498 }, { "epoch": 9.88937728937729, "grad_norm": 1.140625, "learning_rate": 8.607244584549413e-05, "loss": 0.1128, "step": 13499 }, { "epoch": 9.89010989010989, "grad_norm": 0.89453125, "learning_rate": 8.607006311247047e-05, "loss": 0.1328, "step": 13500 }, { "epoch": 9.890842490842491, "grad_norm": 1.3125, "learning_rate": 8.606768020901895e-05, "loss": 0.1608, "step": 13501 }, { "epoch": 9.891575091575092, "grad_norm": 0.98828125, "learning_rate": 8.606529713515101e-05, "loss": 0.1138, "step": 13502 }, { "epoch": 9.892307692307693, "grad_norm": 1.75, "learning_rate": 8.606291389087806e-05, "loss": 0.135, "step": 13503 }, { "epoch": 9.893040293040293, "grad_norm": 1.859375, "learning_rate": 8.606053047621151e-05, "loss": 0.1664, "step": 13504 }, { "epoch": 9.893772893772894, "grad_norm": 1.25, "learning_rate": 8.605814689116279e-05, "loss": 0.1399, "step": 13505 }, { "epoch": 9.894505494505495, "grad_norm": 1.9140625, "learning_rate": 8.605576313574331e-05, "loss": 0.1901, "step": 13506 }, { "epoch": 9.895238095238096, "grad_norm": 0.76953125, "learning_rate": 8.60533792099645e-05, "loss": 0.1297, "step": 13507 }, { "epoch": 9.895970695970696, "grad_norm": 0.88671875, "learning_rate": 8.605099511383779e-05, "loss": 0.1317, "step": 13508 }, { "epoch": 9.896703296703297, "grad_norm": 1.5390625, "learning_rate": 8.604861084737455e-05, "loss": 0.1765, "step": 13509 }, { "epoch": 9.897435897435898, "grad_norm": 1.1953125, "learning_rate": 8.604622641058628e-05, "loss": 0.133, "step": 13510 }, { "epoch": 9.898168498168499, "grad_norm": 1.0859375, "learning_rate": 8.604384180348437e-05, "loss": 0.1076, "step": 13511 }, { "epoch": 9.8989010989011, "grad_norm": 0.7890625, "learning_rate": 8.604145702608024e-05, "loss": 0.1527, "step": 13512 }, { "epoch": 9.8996336996337, "grad_norm": 0.9609375, "learning_rate": 8.603907207838533e-05, "loss": 0.1143, "step": 13513 }, { "epoch": 9.9003663003663, "grad_norm": 1.53125, "learning_rate": 8.603668696041104e-05, "loss": 0.1195, "step": 13514 }, { "epoch": 9.901098901098901, "grad_norm": 1.609375, "learning_rate": 8.603430167216882e-05, "loss": 0.1749, "step": 13515 }, { "epoch": 9.901831501831502, "grad_norm": 1.21875, "learning_rate": 8.603191621367012e-05, "loss": 0.1279, "step": 13516 }, { "epoch": 9.902564102564103, "grad_norm": 1.171875, "learning_rate": 8.602953058492632e-05, "loss": 0.1361, "step": 13517 }, { "epoch": 9.903296703296704, "grad_norm": 1.4140625, "learning_rate": 8.602714478594889e-05, "loss": 0.1211, "step": 13518 }, { "epoch": 9.904029304029304, "grad_norm": 0.9453125, "learning_rate": 8.602475881674924e-05, "loss": 0.1193, "step": 13519 }, { "epoch": 9.904761904761905, "grad_norm": 1.9140625, "learning_rate": 8.60223726773388e-05, "loss": 0.1365, "step": 13520 }, { "epoch": 9.905494505494506, "grad_norm": 1.25, "learning_rate": 8.601998636772902e-05, "loss": 0.1126, "step": 13521 }, { "epoch": 9.906227106227107, "grad_norm": 1.2109375, "learning_rate": 8.601759988793134e-05, "loss": 0.1237, "step": 13522 }, { "epoch": 9.906959706959707, "grad_norm": 2.234375, "learning_rate": 8.601521323795715e-05, "loss": 0.146, "step": 13523 }, { "epoch": 9.907692307692308, "grad_norm": 0.74609375, "learning_rate": 8.601282641781794e-05, "loss": 0.1143, "step": 13524 }, { "epoch": 9.908424908424909, "grad_norm": 0.953125, "learning_rate": 8.60104394275251e-05, "loss": 0.1305, "step": 13525 }, { "epoch": 9.90915750915751, "grad_norm": 0.97265625, "learning_rate": 8.60080522670901e-05, "loss": 0.1346, "step": 13526 }, { "epoch": 9.90989010989011, "grad_norm": 1.2109375, "learning_rate": 8.600566493652437e-05, "loss": 0.117, "step": 13527 }, { "epoch": 9.910622710622711, "grad_norm": 1.53125, "learning_rate": 8.600327743583933e-05, "loss": 0.1415, "step": 13528 }, { "epoch": 9.911355311355312, "grad_norm": 0.8828125, "learning_rate": 8.600088976504642e-05, "loss": 0.1412, "step": 13529 }, { "epoch": 9.912087912087912, "grad_norm": 1.3671875, "learning_rate": 8.599850192415713e-05, "loss": 0.1398, "step": 13530 }, { "epoch": 9.912820512820513, "grad_norm": 1.6875, "learning_rate": 8.599611391318284e-05, "loss": 0.1788, "step": 13531 }, { "epoch": 9.913553113553114, "grad_norm": 0.921875, "learning_rate": 8.599372573213502e-05, "loss": 0.1144, "step": 13532 }, { "epoch": 9.914285714285715, "grad_norm": 1.3828125, "learning_rate": 8.599133738102509e-05, "loss": 0.1658, "step": 13533 }, { "epoch": 9.915018315018315, "grad_norm": 0.6484375, "learning_rate": 8.598894885986454e-05, "loss": 0.1083, "step": 13534 }, { "epoch": 9.915750915750916, "grad_norm": 2.1875, "learning_rate": 8.598656016866479e-05, "loss": 0.1646, "step": 13535 }, { "epoch": 9.916483516483517, "grad_norm": 1.5546875, "learning_rate": 8.598417130743725e-05, "loss": 0.1481, "step": 13536 }, { "epoch": 9.917216117216118, "grad_norm": 0.984375, "learning_rate": 8.598178227619342e-05, "loss": 0.1234, "step": 13537 }, { "epoch": 9.917948717948718, "grad_norm": 1.2734375, "learning_rate": 8.597939307494472e-05, "loss": 0.1735, "step": 13538 }, { "epoch": 9.918681318681319, "grad_norm": 1.4140625, "learning_rate": 8.597700370370261e-05, "loss": 0.1319, "step": 13539 }, { "epoch": 9.91941391941392, "grad_norm": 1.28125, "learning_rate": 8.597461416247851e-05, "loss": 0.1478, "step": 13540 }, { "epoch": 9.92014652014652, "grad_norm": 1.109375, "learning_rate": 8.597222445128388e-05, "loss": 0.1425, "step": 13541 }, { "epoch": 9.920879120879121, "grad_norm": 0.8671875, "learning_rate": 8.59698345701302e-05, "loss": 0.135, "step": 13542 }, { "epoch": 9.921611721611722, "grad_norm": 0.72265625, "learning_rate": 8.59674445190289e-05, "loss": 0.1317, "step": 13543 }, { "epoch": 9.922344322344323, "grad_norm": 1.25, "learning_rate": 8.59650542979914e-05, "loss": 0.1493, "step": 13544 }, { "epoch": 9.923076923076923, "grad_norm": 1.0234375, "learning_rate": 8.596266390702921e-05, "loss": 0.1554, "step": 13545 }, { "epoch": 9.923809523809524, "grad_norm": 0.92578125, "learning_rate": 8.596027334615375e-05, "loss": 0.1376, "step": 13546 }, { "epoch": 9.924542124542125, "grad_norm": 0.890625, "learning_rate": 8.595788261537649e-05, "loss": 0.1441, "step": 13547 }, { "epoch": 9.925274725274726, "grad_norm": 1.453125, "learning_rate": 8.595549171470885e-05, "loss": 0.1233, "step": 13548 }, { "epoch": 9.926007326007326, "grad_norm": 0.75390625, "learning_rate": 8.595310064416232e-05, "loss": 0.1021, "step": 13549 }, { "epoch": 9.926739926739927, "grad_norm": 1.3359375, "learning_rate": 8.595070940374835e-05, "loss": 0.1753, "step": 13550 }, { "epoch": 9.927472527472528, "grad_norm": 1.453125, "learning_rate": 8.59483179934784e-05, "loss": 0.1391, "step": 13551 }, { "epoch": 9.928205128205128, "grad_norm": 1.0078125, "learning_rate": 8.594592641336391e-05, "loss": 0.1504, "step": 13552 }, { "epoch": 9.92893772893773, "grad_norm": 1.0859375, "learning_rate": 8.594353466341635e-05, "loss": 0.125, "step": 13553 }, { "epoch": 9.92967032967033, "grad_norm": 0.8125, "learning_rate": 8.594114274364718e-05, "loss": 0.1371, "step": 13554 }, { "epoch": 9.93040293040293, "grad_norm": 1.015625, "learning_rate": 8.593875065406787e-05, "loss": 0.1419, "step": 13555 }, { "epoch": 9.931135531135531, "grad_norm": 1.1015625, "learning_rate": 8.593635839468986e-05, "loss": 0.1312, "step": 13556 }, { "epoch": 9.931868131868132, "grad_norm": 0.90234375, "learning_rate": 8.593396596552463e-05, "loss": 0.1035, "step": 13557 }, { "epoch": 9.932600732600733, "grad_norm": 1.09375, "learning_rate": 8.593157336658365e-05, "loss": 0.1249, "step": 13558 }, { "epoch": 9.933333333333334, "grad_norm": 1.4140625, "learning_rate": 8.592918059787835e-05, "loss": 0.1328, "step": 13559 }, { "epoch": 9.934065934065934, "grad_norm": 0.953125, "learning_rate": 8.592678765942022e-05, "loss": 0.1772, "step": 13560 }, { "epoch": 9.934798534798535, "grad_norm": 0.9453125, "learning_rate": 8.592439455122073e-05, "loss": 0.1033, "step": 13561 }, { "epoch": 9.935531135531136, "grad_norm": 0.8515625, "learning_rate": 8.592200127329134e-05, "loss": 0.13, "step": 13562 }, { "epoch": 9.936263736263736, "grad_norm": 1.1875, "learning_rate": 8.591960782564348e-05, "loss": 0.1262, "step": 13563 }, { "epoch": 9.936996336996337, "grad_norm": 0.83984375, "learning_rate": 8.591721420828869e-05, "loss": 0.1144, "step": 13564 }, { "epoch": 9.937728937728938, "grad_norm": 0.80078125, "learning_rate": 8.591482042123838e-05, "loss": 0.1275, "step": 13565 }, { "epoch": 9.938461538461539, "grad_norm": 1.125, "learning_rate": 8.591242646450405e-05, "loss": 0.1338, "step": 13566 }, { "epoch": 9.93919413919414, "grad_norm": 0.8046875, "learning_rate": 8.591003233809715e-05, "loss": 0.1021, "step": 13567 }, { "epoch": 9.93992673992674, "grad_norm": 0.89453125, "learning_rate": 8.590763804202915e-05, "loss": 0.136, "step": 13568 }, { "epoch": 9.94065934065934, "grad_norm": 0.9609375, "learning_rate": 8.590524357631157e-05, "loss": 0.1218, "step": 13569 }, { "epoch": 9.941391941391942, "grad_norm": 1.390625, "learning_rate": 8.590284894095581e-05, "loss": 0.1294, "step": 13570 }, { "epoch": 9.942124542124542, "grad_norm": 1.4453125, "learning_rate": 8.59004541359734e-05, "loss": 0.1567, "step": 13571 }, { "epoch": 9.942857142857143, "grad_norm": 1.2421875, "learning_rate": 8.589805916137577e-05, "loss": 0.1501, "step": 13572 }, { "epoch": 9.943589743589744, "grad_norm": 1.1328125, "learning_rate": 8.589566401717445e-05, "loss": 0.1386, "step": 13573 }, { "epoch": 9.944322344322345, "grad_norm": 1.09375, "learning_rate": 8.589326870338085e-05, "loss": 0.1258, "step": 13574 }, { "epoch": 9.945054945054945, "grad_norm": 0.8125, "learning_rate": 8.58908732200065e-05, "loss": 0.1454, "step": 13575 }, { "epoch": 9.945787545787546, "grad_norm": 0.89453125, "learning_rate": 8.588847756706287e-05, "loss": 0.1548, "step": 13576 }, { "epoch": 9.946520146520147, "grad_norm": 1.09375, "learning_rate": 8.588608174456141e-05, "loss": 0.1349, "step": 13577 }, { "epoch": 9.947252747252747, "grad_norm": 1.046875, "learning_rate": 8.588368575251362e-05, "loss": 0.1333, "step": 13578 }, { "epoch": 9.947985347985348, "grad_norm": 1.375, "learning_rate": 8.588128959093097e-05, "loss": 0.1397, "step": 13579 }, { "epoch": 9.948717948717949, "grad_norm": 0.98828125, "learning_rate": 8.587889325982495e-05, "loss": 0.1228, "step": 13580 }, { "epoch": 9.94945054945055, "grad_norm": 1.7890625, "learning_rate": 8.587649675920705e-05, "loss": 0.1709, "step": 13581 }, { "epoch": 9.95018315018315, "grad_norm": 1.203125, "learning_rate": 8.587410008908872e-05, "loss": 0.1423, "step": 13582 }, { "epoch": 9.950915750915751, "grad_norm": 0.91796875, "learning_rate": 8.587170324948147e-05, "loss": 0.1243, "step": 13583 }, { "epoch": 9.951648351648352, "grad_norm": 1.0703125, "learning_rate": 8.58693062403968e-05, "loss": 0.1209, "step": 13584 }, { "epoch": 9.952380952380953, "grad_norm": 0.7890625, "learning_rate": 8.586690906184615e-05, "loss": 0.1419, "step": 13585 }, { "epoch": 9.953113553113553, "grad_norm": 0.84375, "learning_rate": 8.586451171384102e-05, "loss": 0.1415, "step": 13586 }, { "epoch": 9.953846153846154, "grad_norm": 0.8984375, "learning_rate": 8.586211419639291e-05, "loss": 0.1229, "step": 13587 }, { "epoch": 9.954578754578755, "grad_norm": 0.78515625, "learning_rate": 8.585971650951332e-05, "loss": 0.1377, "step": 13588 }, { "epoch": 9.955311355311355, "grad_norm": 1.3671875, "learning_rate": 8.585731865321371e-05, "loss": 0.1448, "step": 13589 }, { "epoch": 9.956043956043956, "grad_norm": 0.78515625, "learning_rate": 8.585492062750558e-05, "loss": 0.1142, "step": 13590 }, { "epoch": 9.956776556776557, "grad_norm": 0.92578125, "learning_rate": 8.585252243240042e-05, "loss": 0.1263, "step": 13591 }, { "epoch": 9.957509157509158, "grad_norm": 0.85546875, "learning_rate": 8.585012406790973e-05, "loss": 0.1389, "step": 13592 }, { "epoch": 9.958241758241758, "grad_norm": 0.91015625, "learning_rate": 8.584772553404496e-05, "loss": 0.1674, "step": 13593 }, { "epoch": 9.95897435897436, "grad_norm": 1.1015625, "learning_rate": 8.584532683081766e-05, "loss": 0.1284, "step": 13594 }, { "epoch": 9.95970695970696, "grad_norm": 0.80078125, "learning_rate": 8.58429279582393e-05, "loss": 0.1317, "step": 13595 }, { "epoch": 9.96043956043956, "grad_norm": 0.69921875, "learning_rate": 8.584052891632135e-05, "loss": 0.1223, "step": 13596 }, { "epoch": 9.961172161172161, "grad_norm": 0.70703125, "learning_rate": 8.583812970507533e-05, "loss": 0.1029, "step": 13597 }, { "epoch": 9.961904761904762, "grad_norm": 1.4453125, "learning_rate": 8.583573032451275e-05, "loss": 0.1095, "step": 13598 }, { "epoch": 9.962637362637363, "grad_norm": 1.5, "learning_rate": 8.583333077464506e-05, "loss": 0.1489, "step": 13599 }, { "epoch": 9.963369963369964, "grad_norm": 0.7109375, "learning_rate": 8.583093105548378e-05, "loss": 0.1394, "step": 13600 }, { "epoch": 9.964102564102564, "grad_norm": 1.2109375, "learning_rate": 8.582853116704042e-05, "loss": 0.1431, "step": 13601 }, { "epoch": 9.964835164835165, "grad_norm": 1.0078125, "learning_rate": 8.58261311093265e-05, "loss": 0.1024, "step": 13602 }, { "epoch": 9.965567765567766, "grad_norm": 0.796875, "learning_rate": 8.582373088235345e-05, "loss": 0.1408, "step": 13603 }, { "epoch": 9.966300366300366, "grad_norm": 1.1875, "learning_rate": 8.582133048613281e-05, "loss": 0.1519, "step": 13604 }, { "epoch": 9.967032967032967, "grad_norm": 0.9375, "learning_rate": 8.581892992067608e-05, "loss": 0.1224, "step": 13605 }, { "epoch": 9.967765567765568, "grad_norm": 0.76171875, "learning_rate": 8.581652918599478e-05, "loss": 0.1153, "step": 13606 }, { "epoch": 9.968498168498169, "grad_norm": 1.078125, "learning_rate": 8.58141282821004e-05, "loss": 0.1382, "step": 13607 }, { "epoch": 9.96923076923077, "grad_norm": 1.4453125, "learning_rate": 8.581172720900442e-05, "loss": 0.2025, "step": 13608 }, { "epoch": 9.96996336996337, "grad_norm": 1.1640625, "learning_rate": 8.580932596671837e-05, "loss": 0.1121, "step": 13609 }, { "epoch": 9.97069597069597, "grad_norm": 0.91015625, "learning_rate": 8.580692455525375e-05, "loss": 0.1131, "step": 13610 }, { "epoch": 9.971428571428572, "grad_norm": 1.0859375, "learning_rate": 8.580452297462206e-05, "loss": 0.1259, "step": 13611 }, { "epoch": 9.972161172161172, "grad_norm": 1.0078125, "learning_rate": 8.580212122483482e-05, "loss": 0.1419, "step": 13612 }, { "epoch": 9.972893772893773, "grad_norm": 0.890625, "learning_rate": 8.579971930590352e-05, "loss": 0.144, "step": 13613 }, { "epoch": 9.973626373626374, "grad_norm": 1.2265625, "learning_rate": 8.579731721783966e-05, "loss": 0.0988, "step": 13614 }, { "epoch": 9.974358974358974, "grad_norm": 1.03125, "learning_rate": 8.579491496065479e-05, "loss": 0.1425, "step": 13615 }, { "epoch": 9.975091575091575, "grad_norm": 0.8828125, "learning_rate": 8.57925125343604e-05, "loss": 0.1015, "step": 13616 }, { "epoch": 9.975824175824176, "grad_norm": 0.66796875, "learning_rate": 8.5790109938968e-05, "loss": 0.1181, "step": 13617 }, { "epoch": 9.976556776556777, "grad_norm": 0.78125, "learning_rate": 8.578770717448906e-05, "loss": 0.121, "step": 13618 }, { "epoch": 9.977289377289377, "grad_norm": 1.453125, "learning_rate": 8.578530424093514e-05, "loss": 0.1644, "step": 13619 }, { "epoch": 9.978021978021978, "grad_norm": 0.8828125, "learning_rate": 8.578290113831779e-05, "loss": 0.1249, "step": 13620 }, { "epoch": 9.978754578754579, "grad_norm": 1.0625, "learning_rate": 8.578049786664843e-05, "loss": 0.1285, "step": 13621 }, { "epoch": 9.97948717948718, "grad_norm": 0.91015625, "learning_rate": 8.577809442593864e-05, "loss": 0.1522, "step": 13622 }, { "epoch": 9.98021978021978, "grad_norm": 0.96875, "learning_rate": 8.577569081619992e-05, "loss": 0.1476, "step": 13623 }, { "epoch": 9.980952380952381, "grad_norm": 0.91015625, "learning_rate": 8.577328703744379e-05, "loss": 0.1325, "step": 13624 }, { "epoch": 9.981684981684982, "grad_norm": 0.84375, "learning_rate": 8.577088308968176e-05, "loss": 0.1404, "step": 13625 }, { "epoch": 9.982417582417582, "grad_norm": 0.9921875, "learning_rate": 8.576847897292534e-05, "loss": 0.1286, "step": 13626 }, { "epoch": 9.983150183150183, "grad_norm": 0.796875, "learning_rate": 8.576607468718606e-05, "loss": 0.112, "step": 13627 }, { "epoch": 9.983882783882784, "grad_norm": 1.6484375, "learning_rate": 8.576367023247545e-05, "loss": 0.1472, "step": 13628 }, { "epoch": 9.984615384615385, "grad_norm": 1.5, "learning_rate": 8.576126560880502e-05, "loss": 0.0976, "step": 13629 }, { "epoch": 9.985347985347985, "grad_norm": 1.1015625, "learning_rate": 8.575886081618628e-05, "loss": 0.1502, "step": 13630 }, { "epoch": 9.986080586080586, "grad_norm": 0.87890625, "learning_rate": 8.575645585463077e-05, "loss": 0.1294, "step": 13631 }, { "epoch": 9.986813186813187, "grad_norm": 1.171875, "learning_rate": 8.575405072415e-05, "loss": 0.1595, "step": 13632 }, { "epoch": 9.987545787545788, "grad_norm": 1.0078125, "learning_rate": 8.57516454247555e-05, "loss": 0.14, "step": 13633 }, { "epoch": 9.988278388278388, "grad_norm": 1.0625, "learning_rate": 8.574923995645879e-05, "loss": 0.1496, "step": 13634 }, { "epoch": 9.989010989010989, "grad_norm": 0.86328125, "learning_rate": 8.574683431927141e-05, "loss": 0.1293, "step": 13635 }, { "epoch": 9.98974358974359, "grad_norm": 1.28125, "learning_rate": 8.574442851320487e-05, "loss": 0.145, "step": 13636 }, { "epoch": 9.99047619047619, "grad_norm": 2.109375, "learning_rate": 8.574202253827072e-05, "loss": 0.1791, "step": 13637 }, { "epoch": 9.991208791208791, "grad_norm": 0.93359375, "learning_rate": 8.573961639448046e-05, "loss": 0.1186, "step": 13638 }, { "epoch": 9.991941391941392, "grad_norm": 1.6640625, "learning_rate": 8.573721008184563e-05, "loss": 0.1099, "step": 13639 }, { "epoch": 9.992673992673993, "grad_norm": 0.8984375, "learning_rate": 8.573480360037776e-05, "loss": 0.111, "step": 13640 }, { "epoch": 9.993406593406593, "grad_norm": 1.796875, "learning_rate": 8.573239695008837e-05, "loss": 0.1503, "step": 13641 }, { "epoch": 9.994139194139194, "grad_norm": 0.515625, "learning_rate": 8.5729990130989e-05, "loss": 0.1114, "step": 13642 }, { "epoch": 9.994871794871795, "grad_norm": 0.96875, "learning_rate": 8.572758314309119e-05, "loss": 0.127, "step": 13643 }, { "epoch": 9.995604395604396, "grad_norm": 2.015625, "learning_rate": 8.572517598640646e-05, "loss": 0.1636, "step": 13644 }, { "epoch": 9.996336996336996, "grad_norm": 1.0859375, "learning_rate": 8.572276866094635e-05, "loss": 0.1068, "step": 13645 }, { "epoch": 9.997069597069597, "grad_norm": 0.828125, "learning_rate": 8.57203611667224e-05, "loss": 0.1497, "step": 13646 }, { "epoch": 9.997802197802198, "grad_norm": 0.92578125, "learning_rate": 8.571795350374613e-05, "loss": 0.1465, "step": 13647 }, { "epoch": 9.998534798534799, "grad_norm": 0.89453125, "learning_rate": 8.571554567202909e-05, "loss": 0.1223, "step": 13648 }, { "epoch": 9.9992673992674, "grad_norm": 0.85546875, "learning_rate": 8.57131376715828e-05, "loss": 0.1022, "step": 13649 }, { "epoch": 10.0, "grad_norm": 0.80859375, "learning_rate": 8.57107295024188e-05, "loss": 0.1239, "step": 13650 }, { "epoch": 10.0007326007326, "grad_norm": 0.875, "learning_rate": 8.570832116454866e-05, "loss": 0.1445, "step": 13651 }, { "epoch": 10.001465201465201, "grad_norm": 0.99609375, "learning_rate": 8.570591265798388e-05, "loss": 0.1572, "step": 13652 }, { "epoch": 10.002197802197802, "grad_norm": 1.28125, "learning_rate": 8.570350398273603e-05, "loss": 0.1056, "step": 13653 }, { "epoch": 10.002930402930403, "grad_norm": 1.0859375, "learning_rate": 8.570109513881661e-05, "loss": 0.1122, "step": 13654 }, { "epoch": 10.003663003663004, "grad_norm": 1.0390625, "learning_rate": 8.569868612623721e-05, "loss": 0.1256, "step": 13655 }, { "epoch": 10.004395604395604, "grad_norm": 1.4765625, "learning_rate": 8.569627694500934e-05, "loss": 0.1678, "step": 13656 }, { "epoch": 10.005128205128205, "grad_norm": 0.98828125, "learning_rate": 8.569386759514455e-05, "loss": 0.1114, "step": 13657 }, { "epoch": 10.005860805860806, "grad_norm": 0.6640625, "learning_rate": 8.569145807665438e-05, "loss": 0.1099, "step": 13658 }, { "epoch": 10.006593406593407, "grad_norm": 0.8203125, "learning_rate": 8.568904838955041e-05, "loss": 0.1143, "step": 13659 }, { "epoch": 10.007326007326007, "grad_norm": 1.09375, "learning_rate": 8.568663853384415e-05, "loss": 0.1481, "step": 13660 }, { "epoch": 10.008058608058608, "grad_norm": 1.265625, "learning_rate": 8.568422850954714e-05, "loss": 0.1273, "step": 13661 }, { "epoch": 10.008791208791209, "grad_norm": 1.0078125, "learning_rate": 8.568181831667093e-05, "loss": 0.1523, "step": 13662 }, { "epoch": 10.00952380952381, "grad_norm": 1.0546875, "learning_rate": 8.567940795522711e-05, "loss": 0.1109, "step": 13663 }, { "epoch": 10.01025641025641, "grad_norm": 0.93359375, "learning_rate": 8.567699742522717e-05, "loss": 0.1056, "step": 13664 }, { "epoch": 10.010989010989011, "grad_norm": 1.0625, "learning_rate": 8.567458672668271e-05, "loss": 0.1548, "step": 13665 }, { "epoch": 10.011721611721612, "grad_norm": 1.7109375, "learning_rate": 8.567217585960524e-05, "loss": 0.1261, "step": 13666 }, { "epoch": 10.012454212454212, "grad_norm": 1.640625, "learning_rate": 8.566976482400633e-05, "loss": 0.1439, "step": 13667 }, { "epoch": 10.013186813186813, "grad_norm": 0.64453125, "learning_rate": 8.566735361989756e-05, "loss": 0.1092, "step": 13668 }, { "epoch": 10.013919413919414, "grad_norm": 1.1875, "learning_rate": 8.566494224729042e-05, "loss": 0.1314, "step": 13669 }, { "epoch": 10.014652014652015, "grad_norm": 1.046875, "learning_rate": 8.56625307061965e-05, "loss": 0.1248, "step": 13670 }, { "epoch": 10.015384615384615, "grad_norm": 1.2265625, "learning_rate": 8.566011899662736e-05, "loss": 0.1822, "step": 13671 }, { "epoch": 10.016117216117216, "grad_norm": 1.359375, "learning_rate": 8.565770711859456e-05, "loss": 0.176, "step": 13672 }, { "epoch": 10.016849816849817, "grad_norm": 0.91796875, "learning_rate": 8.565529507210964e-05, "loss": 0.1155, "step": 13673 }, { "epoch": 10.017582417582418, "grad_norm": 0.9453125, "learning_rate": 8.565288285718414e-05, "loss": 0.172, "step": 13674 }, { "epoch": 10.018315018315018, "grad_norm": 1.1015625, "learning_rate": 8.565047047382968e-05, "loss": 0.128, "step": 13675 }, { "epoch": 10.019047619047619, "grad_norm": 0.7265625, "learning_rate": 8.564805792205772e-05, "loss": 0.1211, "step": 13676 }, { "epoch": 10.01978021978022, "grad_norm": 1.125, "learning_rate": 8.564564520187992e-05, "loss": 0.1322, "step": 13677 }, { "epoch": 10.02051282051282, "grad_norm": 1.671875, "learning_rate": 8.564323231330777e-05, "loss": 0.1423, "step": 13678 }, { "epoch": 10.021245421245421, "grad_norm": 1.4609375, "learning_rate": 8.564081925635286e-05, "loss": 0.1656, "step": 13679 }, { "epoch": 10.021978021978022, "grad_norm": 1.328125, "learning_rate": 8.563840603102678e-05, "loss": 0.1341, "step": 13680 }, { "epoch": 10.022710622710623, "grad_norm": 1.265625, "learning_rate": 8.563599263734103e-05, "loss": 0.1475, "step": 13681 }, { "epoch": 10.023443223443223, "grad_norm": 0.97265625, "learning_rate": 8.563357907530721e-05, "loss": 0.1393, "step": 13682 }, { "epoch": 10.024175824175824, "grad_norm": 0.96875, "learning_rate": 8.563116534493689e-05, "loss": 0.122, "step": 13683 }, { "epoch": 10.024908424908425, "grad_norm": 1.46875, "learning_rate": 8.562875144624162e-05, "loss": 0.151, "step": 13684 }, { "epoch": 10.025641025641026, "grad_norm": 0.87890625, "learning_rate": 8.562633737923297e-05, "loss": 0.1284, "step": 13685 }, { "epoch": 10.026373626373626, "grad_norm": 1.140625, "learning_rate": 8.56239231439225e-05, "loss": 0.1267, "step": 13686 }, { "epoch": 10.027106227106227, "grad_norm": 1.1796875, "learning_rate": 8.562150874032179e-05, "loss": 0.1079, "step": 13687 }, { "epoch": 10.027838827838828, "grad_norm": 0.83203125, "learning_rate": 8.56190941684424e-05, "loss": 0.1404, "step": 13688 }, { "epoch": 10.028571428571428, "grad_norm": 1.140625, "learning_rate": 8.561667942829589e-05, "loss": 0.1517, "step": 13689 }, { "epoch": 10.02930402930403, "grad_norm": 0.94921875, "learning_rate": 8.561426451989385e-05, "loss": 0.1392, "step": 13690 }, { "epoch": 10.03003663003663, "grad_norm": 1.140625, "learning_rate": 8.561184944324784e-05, "loss": 0.1373, "step": 13691 }, { "epoch": 10.03076923076923, "grad_norm": 0.91015625, "learning_rate": 8.560943419836944e-05, "loss": 0.1249, "step": 13692 }, { "epoch": 10.031501831501831, "grad_norm": 1.203125, "learning_rate": 8.560701878527021e-05, "loss": 0.1527, "step": 13693 }, { "epoch": 10.032234432234432, "grad_norm": 1.015625, "learning_rate": 8.560460320396173e-05, "loss": 0.1362, "step": 13694 }, { "epoch": 10.032967032967033, "grad_norm": 1.09375, "learning_rate": 8.560218745445556e-05, "loss": 0.1266, "step": 13695 }, { "epoch": 10.033699633699634, "grad_norm": 0.89453125, "learning_rate": 8.55997715367633e-05, "loss": 0.1114, "step": 13696 }, { "epoch": 10.034432234432234, "grad_norm": 1.375, "learning_rate": 8.559735545089653e-05, "loss": 0.1543, "step": 13697 }, { "epoch": 10.035164835164835, "grad_norm": 1.59375, "learning_rate": 8.559493919686677e-05, "loss": 0.1587, "step": 13698 }, { "epoch": 10.035897435897436, "grad_norm": 0.546875, "learning_rate": 8.559252277468566e-05, "loss": 0.0997, "step": 13699 }, { "epoch": 10.036630036630036, "grad_norm": 1.78125, "learning_rate": 8.559010618436474e-05, "loss": 0.1589, "step": 13700 }, { "epoch": 10.037362637362637, "grad_norm": 1.015625, "learning_rate": 8.558768942591561e-05, "loss": 0.1401, "step": 13701 }, { "epoch": 10.038095238095238, "grad_norm": 0.890625, "learning_rate": 8.558527249934984e-05, "loss": 0.1042, "step": 13702 }, { "epoch": 10.038827838827839, "grad_norm": 0.55859375, "learning_rate": 8.558285540467901e-05, "loss": 0.0849, "step": 13703 }, { "epoch": 10.03956043956044, "grad_norm": 1.0546875, "learning_rate": 8.55804381419147e-05, "loss": 0.138, "step": 13704 }, { "epoch": 10.04029304029304, "grad_norm": 1.0078125, "learning_rate": 8.55780207110685e-05, "loss": 0.1549, "step": 13705 }, { "epoch": 10.04102564102564, "grad_norm": 1.09375, "learning_rate": 8.557560311215198e-05, "loss": 0.1182, "step": 13706 }, { "epoch": 10.041758241758242, "grad_norm": 1.1640625, "learning_rate": 8.557318534517675e-05, "loss": 0.132, "step": 13707 }, { "epoch": 10.042490842490842, "grad_norm": 0.9921875, "learning_rate": 8.557076741015435e-05, "loss": 0.1514, "step": 13708 }, { "epoch": 10.043223443223443, "grad_norm": 1.140625, "learning_rate": 8.556834930709642e-05, "loss": 0.1337, "step": 13709 }, { "epoch": 10.043956043956044, "grad_norm": 1.046875, "learning_rate": 8.556593103601449e-05, "loss": 0.1405, "step": 13710 }, { "epoch": 10.044688644688645, "grad_norm": 0.83984375, "learning_rate": 8.556351259692019e-05, "loss": 0.1295, "step": 13711 }, { "epoch": 10.045421245421245, "grad_norm": 0.92578125, "learning_rate": 8.556109398982507e-05, "loss": 0.1152, "step": 13712 }, { "epoch": 10.046153846153846, "grad_norm": 1.1640625, "learning_rate": 8.555867521474075e-05, "loss": 0.1275, "step": 13713 }, { "epoch": 10.046886446886447, "grad_norm": 1.046875, "learning_rate": 8.555625627167881e-05, "loss": 0.1304, "step": 13714 }, { "epoch": 10.047619047619047, "grad_norm": 0.98828125, "learning_rate": 8.555383716065085e-05, "loss": 0.136, "step": 13715 }, { "epoch": 10.048351648351648, "grad_norm": 1.0703125, "learning_rate": 8.555141788166843e-05, "loss": 0.1312, "step": 13716 }, { "epoch": 10.049084249084249, "grad_norm": 0.9609375, "learning_rate": 8.554899843474317e-05, "loss": 0.1158, "step": 13717 }, { "epoch": 10.04981684981685, "grad_norm": 0.95703125, "learning_rate": 8.554657881988664e-05, "loss": 0.1233, "step": 13718 }, { "epoch": 10.05054945054945, "grad_norm": 0.63671875, "learning_rate": 8.554415903711046e-05, "loss": 0.1142, "step": 13719 }, { "epoch": 10.051282051282051, "grad_norm": 1.1015625, "learning_rate": 8.554173908642622e-05, "loss": 0.1417, "step": 13720 }, { "epoch": 10.052014652014652, "grad_norm": 0.921875, "learning_rate": 8.553931896784546e-05, "loss": 0.1255, "step": 13721 }, { "epoch": 10.052747252747253, "grad_norm": 1.109375, "learning_rate": 8.553689868137986e-05, "loss": 0.1484, "step": 13722 }, { "epoch": 10.053479853479853, "grad_norm": 0.96875, "learning_rate": 8.553447822704096e-05, "loss": 0.1329, "step": 13723 }, { "epoch": 10.054212454212454, "grad_norm": 0.87109375, "learning_rate": 8.553205760484039e-05, "loss": 0.1145, "step": 13724 }, { "epoch": 10.054945054945055, "grad_norm": 1.3515625, "learning_rate": 8.552963681478973e-05, "loss": 0.1531, "step": 13725 }, { "epoch": 10.055677655677655, "grad_norm": 0.66796875, "learning_rate": 8.552721585690056e-05, "loss": 0.0999, "step": 13726 }, { "epoch": 10.056410256410256, "grad_norm": 0.84765625, "learning_rate": 8.552479473118451e-05, "loss": 0.1376, "step": 13727 }, { "epoch": 10.057142857142857, "grad_norm": 0.8828125, "learning_rate": 8.552237343765317e-05, "loss": 0.1333, "step": 13728 }, { "epoch": 10.057875457875458, "grad_norm": 1.546875, "learning_rate": 8.551995197631815e-05, "loss": 0.142, "step": 13729 }, { "epoch": 10.058608058608058, "grad_norm": 1.6015625, "learning_rate": 8.551753034719104e-05, "loss": 0.1841, "step": 13730 }, { "epoch": 10.05934065934066, "grad_norm": 1.0859375, "learning_rate": 8.551510855028344e-05, "loss": 0.1258, "step": 13731 }, { "epoch": 10.06007326007326, "grad_norm": 1.3515625, "learning_rate": 8.551268658560698e-05, "loss": 0.1304, "step": 13732 }, { "epoch": 10.06080586080586, "grad_norm": 1.234375, "learning_rate": 8.551026445317321e-05, "loss": 0.1301, "step": 13733 }, { "epoch": 10.061538461538461, "grad_norm": 0.9375, "learning_rate": 8.55078421529938e-05, "loss": 0.0982, "step": 13734 }, { "epoch": 10.062271062271062, "grad_norm": 0.81640625, "learning_rate": 8.550541968508031e-05, "loss": 0.1135, "step": 13735 }, { "epoch": 10.063003663003663, "grad_norm": 1.4921875, "learning_rate": 8.550299704944437e-05, "loss": 0.1441, "step": 13736 }, { "epoch": 10.063736263736264, "grad_norm": 1.1484375, "learning_rate": 8.550057424609758e-05, "loss": 0.1269, "step": 13737 }, { "epoch": 10.064468864468864, "grad_norm": 1.1796875, "learning_rate": 8.549815127505155e-05, "loss": 0.139, "step": 13738 }, { "epoch": 10.065201465201465, "grad_norm": 1.421875, "learning_rate": 8.549572813631789e-05, "loss": 0.1431, "step": 13739 }, { "epoch": 10.065934065934066, "grad_norm": 0.90625, "learning_rate": 8.54933048299082e-05, "loss": 0.1174, "step": 13740 }, { "epoch": 10.066666666666666, "grad_norm": 1.1796875, "learning_rate": 8.54908813558341e-05, "loss": 0.1641, "step": 13741 }, { "epoch": 10.067399267399267, "grad_norm": 0.51953125, "learning_rate": 8.548845771410721e-05, "loss": 0.1003, "step": 13742 }, { "epoch": 10.068131868131868, "grad_norm": 0.71484375, "learning_rate": 8.548603390473912e-05, "loss": 0.1204, "step": 13743 }, { "epoch": 10.068864468864469, "grad_norm": 1.25, "learning_rate": 8.548360992774146e-05, "loss": 0.131, "step": 13744 }, { "epoch": 10.06959706959707, "grad_norm": 1.0, "learning_rate": 8.548118578312585e-05, "loss": 0.1435, "step": 13745 }, { "epoch": 10.07032967032967, "grad_norm": 1.6328125, "learning_rate": 8.54787614709039e-05, "loss": 0.1514, "step": 13746 }, { "epoch": 10.07106227106227, "grad_norm": 1.7734375, "learning_rate": 8.547633699108721e-05, "loss": 0.1311, "step": 13747 }, { "epoch": 10.071794871794872, "grad_norm": 1.4140625, "learning_rate": 8.547391234368739e-05, "loss": 0.1212, "step": 13748 }, { "epoch": 10.072527472527472, "grad_norm": 1.21875, "learning_rate": 8.547148752871608e-05, "loss": 0.1398, "step": 13749 }, { "epoch": 10.073260073260073, "grad_norm": 1.1328125, "learning_rate": 8.546906254618492e-05, "loss": 0.1445, "step": 13750 }, { "epoch": 10.073992673992674, "grad_norm": 1.25, "learning_rate": 8.546663739610549e-05, "loss": 0.113, "step": 13751 }, { "epoch": 10.074725274725274, "grad_norm": 1.2265625, "learning_rate": 8.546421207848942e-05, "loss": 0.1348, "step": 13752 }, { "epoch": 10.075457875457875, "grad_norm": 1.8125, "learning_rate": 8.546178659334833e-05, "loss": 0.1584, "step": 13753 }, { "epoch": 10.076190476190476, "grad_norm": 1.3359375, "learning_rate": 8.545936094069383e-05, "loss": 0.1456, "step": 13754 }, { "epoch": 10.076923076923077, "grad_norm": 1.3671875, "learning_rate": 8.545693512053758e-05, "loss": 0.1281, "step": 13755 }, { "epoch": 10.077655677655677, "grad_norm": 0.85546875, "learning_rate": 8.545450913289117e-05, "loss": 0.1304, "step": 13756 }, { "epoch": 10.078388278388278, "grad_norm": 1.4140625, "learning_rate": 8.545208297776621e-05, "loss": 0.1544, "step": 13757 }, { "epoch": 10.079120879120879, "grad_norm": 1.1328125, "learning_rate": 8.544965665517438e-05, "loss": 0.1313, "step": 13758 }, { "epoch": 10.07985347985348, "grad_norm": 1.5703125, "learning_rate": 8.544723016512726e-05, "loss": 0.1447, "step": 13759 }, { "epoch": 10.08058608058608, "grad_norm": 1.1171875, "learning_rate": 8.544480350763649e-05, "loss": 0.1277, "step": 13760 }, { "epoch": 10.081318681318681, "grad_norm": 1.109375, "learning_rate": 8.54423766827137e-05, "loss": 0.1517, "step": 13761 }, { "epoch": 10.082051282051282, "grad_norm": 0.96875, "learning_rate": 8.543994969037049e-05, "loss": 0.0991, "step": 13762 }, { "epoch": 10.082783882783882, "grad_norm": 1.171875, "learning_rate": 8.543752253061853e-05, "loss": 0.1106, "step": 13763 }, { "epoch": 10.083516483516483, "grad_norm": 1.09375, "learning_rate": 8.54350952034694e-05, "loss": 0.1173, "step": 13764 }, { "epoch": 10.084249084249084, "grad_norm": 0.6171875, "learning_rate": 8.54326677089348e-05, "loss": 0.1467, "step": 13765 }, { "epoch": 10.084981684981685, "grad_norm": 1.6875, "learning_rate": 8.54302400470263e-05, "loss": 0.1327, "step": 13766 }, { "epoch": 10.085714285714285, "grad_norm": 1.6875, "learning_rate": 8.542781221775556e-05, "loss": 0.1468, "step": 13767 }, { "epoch": 10.086446886446886, "grad_norm": 0.88671875, "learning_rate": 8.54253842211342e-05, "loss": 0.1249, "step": 13768 }, { "epoch": 10.087179487179487, "grad_norm": 0.76953125, "learning_rate": 8.542295605717387e-05, "loss": 0.1239, "step": 13769 }, { "epoch": 10.087912087912088, "grad_norm": 0.859375, "learning_rate": 8.542052772588618e-05, "loss": 0.1463, "step": 13770 }, { "epoch": 10.088644688644688, "grad_norm": 1.21875, "learning_rate": 8.541809922728279e-05, "loss": 0.1253, "step": 13771 }, { "epoch": 10.089377289377289, "grad_norm": 1.03125, "learning_rate": 8.541567056137531e-05, "loss": 0.1365, "step": 13772 }, { "epoch": 10.09010989010989, "grad_norm": 0.91015625, "learning_rate": 8.54132417281754e-05, "loss": 0.1032, "step": 13773 }, { "epoch": 10.09084249084249, "grad_norm": 0.87890625, "learning_rate": 8.541081272769468e-05, "loss": 0.1439, "step": 13774 }, { "epoch": 10.091575091575091, "grad_norm": 1.2265625, "learning_rate": 8.54083835599448e-05, "loss": 0.1256, "step": 13775 }, { "epoch": 10.092307692307692, "grad_norm": 1.2109375, "learning_rate": 8.54059542249374e-05, "loss": 0.1499, "step": 13776 }, { "epoch": 10.093040293040293, "grad_norm": 1.265625, "learning_rate": 8.540352472268411e-05, "loss": 0.1137, "step": 13777 }, { "epoch": 10.093772893772893, "grad_norm": 1.5703125, "learning_rate": 8.540109505319659e-05, "loss": 0.1282, "step": 13778 }, { "epoch": 10.094505494505494, "grad_norm": 0.82421875, "learning_rate": 8.539866521648645e-05, "loss": 0.1505, "step": 13779 }, { "epoch": 10.095238095238095, "grad_norm": 0.76953125, "learning_rate": 8.539623521256535e-05, "loss": 0.1365, "step": 13780 }, { "epoch": 10.095970695970696, "grad_norm": 1.6171875, "learning_rate": 8.539380504144492e-05, "loss": 0.1196, "step": 13781 }, { "epoch": 10.096703296703296, "grad_norm": 0.99609375, "learning_rate": 8.539137470313684e-05, "loss": 0.129, "step": 13782 }, { "epoch": 10.097435897435897, "grad_norm": 1.0078125, "learning_rate": 8.538894419765271e-05, "loss": 0.1419, "step": 13783 }, { "epoch": 10.098168498168498, "grad_norm": 0.83984375, "learning_rate": 8.53865135250042e-05, "loss": 0.1407, "step": 13784 }, { "epoch": 10.098901098901099, "grad_norm": 1.328125, "learning_rate": 8.538408268520294e-05, "loss": 0.1554, "step": 13785 }, { "epoch": 10.0996336996337, "grad_norm": 0.82421875, "learning_rate": 8.538165167826062e-05, "loss": 0.1236, "step": 13786 }, { "epoch": 10.1003663003663, "grad_norm": 1.21875, "learning_rate": 8.537922050418883e-05, "loss": 0.1087, "step": 13787 }, { "epoch": 10.1010989010989, "grad_norm": 0.8828125, "learning_rate": 8.537678916299925e-05, "loss": 0.1441, "step": 13788 }, { "epoch": 10.101831501831501, "grad_norm": 0.87890625, "learning_rate": 8.537435765470353e-05, "loss": 0.1133, "step": 13789 }, { "epoch": 10.102564102564102, "grad_norm": 1.4921875, "learning_rate": 8.53719259793133e-05, "loss": 0.1353, "step": 13790 }, { "epoch": 10.103296703296703, "grad_norm": 0.7265625, "learning_rate": 8.536949413684024e-05, "loss": 0.1299, "step": 13791 }, { "epoch": 10.104029304029304, "grad_norm": 1.1796875, "learning_rate": 8.536706212729598e-05, "loss": 0.1426, "step": 13792 }, { "epoch": 10.104761904761904, "grad_norm": 1.0859375, "learning_rate": 8.536462995069217e-05, "loss": 0.1651, "step": 13793 }, { "epoch": 10.105494505494505, "grad_norm": 0.88671875, "learning_rate": 8.53621976070405e-05, "loss": 0.1293, "step": 13794 }, { "epoch": 10.106227106227106, "grad_norm": 1.234375, "learning_rate": 8.535976509635257e-05, "loss": 0.1558, "step": 13795 }, { "epoch": 10.106959706959707, "grad_norm": 0.97265625, "learning_rate": 8.535733241864008e-05, "loss": 0.1483, "step": 13796 }, { "epoch": 10.107692307692307, "grad_norm": 0.640625, "learning_rate": 8.535489957391465e-05, "loss": 0.123, "step": 13797 }, { "epoch": 10.108424908424908, "grad_norm": 0.78515625, "learning_rate": 8.535246656218796e-05, "loss": 0.1126, "step": 13798 }, { "epoch": 10.109157509157509, "grad_norm": 1.2109375, "learning_rate": 8.535003338347167e-05, "loss": 0.1233, "step": 13799 }, { "epoch": 10.10989010989011, "grad_norm": 0.7265625, "learning_rate": 8.534760003777743e-05, "loss": 0.1549, "step": 13800 }, { "epoch": 10.11062271062271, "grad_norm": 1.2421875, "learning_rate": 8.534516652511689e-05, "loss": 0.1436, "step": 13801 }, { "epoch": 10.111355311355311, "grad_norm": 0.921875, "learning_rate": 8.534273284550171e-05, "loss": 0.1046, "step": 13802 }, { "epoch": 10.112087912087912, "grad_norm": 1.78125, "learning_rate": 8.534029899894357e-05, "loss": 0.1352, "step": 13803 }, { "epoch": 10.112820512820512, "grad_norm": 1.109375, "learning_rate": 8.533786498545413e-05, "loss": 0.1315, "step": 13804 }, { "epoch": 10.113553113553113, "grad_norm": 1.078125, "learning_rate": 8.533543080504501e-05, "loss": 0.1501, "step": 13805 }, { "epoch": 10.114285714285714, "grad_norm": 1.546875, "learning_rate": 8.533299645772794e-05, "loss": 0.1054, "step": 13806 }, { "epoch": 10.115018315018315, "grad_norm": 1.1484375, "learning_rate": 8.533056194351455e-05, "loss": 0.1605, "step": 13807 }, { "epoch": 10.115750915750915, "grad_norm": 1.0546875, "learning_rate": 8.532812726241649e-05, "loss": 0.1544, "step": 13808 }, { "epoch": 10.116483516483516, "grad_norm": 0.7265625, "learning_rate": 8.532569241444543e-05, "loss": 0.1211, "step": 13809 }, { "epoch": 10.117216117216117, "grad_norm": 0.80859375, "learning_rate": 8.532325739961304e-05, "loss": 0.1199, "step": 13810 }, { "epoch": 10.117948717948718, "grad_norm": 0.89453125, "learning_rate": 8.532082221793102e-05, "loss": 0.1409, "step": 13811 }, { "epoch": 10.118681318681318, "grad_norm": 0.7890625, "learning_rate": 8.531838686941098e-05, "loss": 0.0992, "step": 13812 }, { "epoch": 10.119413919413919, "grad_norm": 1.140625, "learning_rate": 8.531595135406465e-05, "loss": 0.107, "step": 13813 }, { "epoch": 10.12014652014652, "grad_norm": 1.0390625, "learning_rate": 8.531351567190366e-05, "loss": 0.1467, "step": 13814 }, { "epoch": 10.12087912087912, "grad_norm": 1.0703125, "learning_rate": 8.531107982293966e-05, "loss": 0.12, "step": 13815 }, { "epoch": 10.121611721611721, "grad_norm": 0.87890625, "learning_rate": 8.530864380718438e-05, "loss": 0.1184, "step": 13816 }, { "epoch": 10.122344322344322, "grad_norm": 0.71484375, "learning_rate": 8.530620762464943e-05, "loss": 0.1352, "step": 13817 }, { "epoch": 10.123076923076923, "grad_norm": 1.34375, "learning_rate": 8.530377127534655e-05, "loss": 0.1284, "step": 13818 }, { "epoch": 10.123809523809523, "grad_norm": 1.1015625, "learning_rate": 8.530133475928735e-05, "loss": 0.149, "step": 13819 }, { "epoch": 10.124542124542124, "grad_norm": 0.80859375, "learning_rate": 8.529889807648355e-05, "loss": 0.1102, "step": 13820 }, { "epoch": 10.125274725274725, "grad_norm": 0.95703125, "learning_rate": 8.529646122694678e-05, "loss": 0.1328, "step": 13821 }, { "epoch": 10.126007326007326, "grad_norm": 1.0390625, "learning_rate": 8.529402421068877e-05, "loss": 0.1098, "step": 13822 }, { "epoch": 10.126739926739926, "grad_norm": 0.81640625, "learning_rate": 8.529158702772116e-05, "loss": 0.1181, "step": 13823 }, { "epoch": 10.127472527472527, "grad_norm": 1.4140625, "learning_rate": 8.52891496780556e-05, "loss": 0.113, "step": 13824 }, { "epoch": 10.128205128205128, "grad_norm": 1.25, "learning_rate": 8.528671216170385e-05, "loss": 0.1363, "step": 13825 }, { "epoch": 10.128937728937728, "grad_norm": 1.0390625, "learning_rate": 8.528427447867753e-05, "loss": 0.1279, "step": 13826 }, { "epoch": 10.12967032967033, "grad_norm": 1.5546875, "learning_rate": 8.528183662898832e-05, "loss": 0.1417, "step": 13827 }, { "epoch": 10.13040293040293, "grad_norm": 0.60546875, "learning_rate": 8.527939861264792e-05, "loss": 0.1153, "step": 13828 }, { "epoch": 10.13113553113553, "grad_norm": 0.94921875, "learning_rate": 8.527696042966801e-05, "loss": 0.1236, "step": 13829 }, { "epoch": 10.131868131868131, "grad_norm": 1.609375, "learning_rate": 8.527452208006025e-05, "loss": 0.1706, "step": 13830 }, { "epoch": 10.132600732600732, "grad_norm": 1.296875, "learning_rate": 8.527208356383634e-05, "loss": 0.12, "step": 13831 }, { "epoch": 10.133333333333333, "grad_norm": 0.73828125, "learning_rate": 8.526964488100798e-05, "loss": 0.1351, "step": 13832 }, { "epoch": 10.134065934065934, "grad_norm": 1.2421875, "learning_rate": 8.526720603158682e-05, "loss": 0.1304, "step": 13833 }, { "epoch": 10.134798534798534, "grad_norm": 1.015625, "learning_rate": 8.526476701558458e-05, "loss": 0.1421, "step": 13834 }, { "epoch": 10.135531135531135, "grad_norm": 0.75, "learning_rate": 8.526232783301292e-05, "loss": 0.1111, "step": 13835 }, { "epoch": 10.136263736263736, "grad_norm": 1.5703125, "learning_rate": 8.525988848388353e-05, "loss": 0.1218, "step": 13836 }, { "epoch": 10.136996336996337, "grad_norm": 1.65625, "learning_rate": 8.525744896820812e-05, "loss": 0.1142, "step": 13837 }, { "epoch": 10.137728937728937, "grad_norm": 0.97265625, "learning_rate": 8.525500928599833e-05, "loss": 0.1346, "step": 13838 }, { "epoch": 10.138461538461538, "grad_norm": 0.95703125, "learning_rate": 8.525256943726591e-05, "loss": 0.1182, "step": 13839 }, { "epoch": 10.139194139194139, "grad_norm": 1.296875, "learning_rate": 8.525012942202252e-05, "loss": 0.1386, "step": 13840 }, { "epoch": 10.13992673992674, "grad_norm": 1.1171875, "learning_rate": 8.524768924027985e-05, "loss": 0.1167, "step": 13841 }, { "epoch": 10.14065934065934, "grad_norm": 1.0703125, "learning_rate": 8.524524889204958e-05, "loss": 0.1277, "step": 13842 }, { "epoch": 10.141391941391941, "grad_norm": 1.7734375, "learning_rate": 8.524280837734343e-05, "loss": 0.1351, "step": 13843 }, { "epoch": 10.142124542124542, "grad_norm": 1.1953125, "learning_rate": 8.524036769617307e-05, "loss": 0.1275, "step": 13844 }, { "epoch": 10.142857142857142, "grad_norm": 0.90625, "learning_rate": 8.52379268485502e-05, "loss": 0.1315, "step": 13845 }, { "epoch": 10.143589743589743, "grad_norm": 1.4765625, "learning_rate": 8.523548583448654e-05, "loss": 0.1487, "step": 13846 }, { "epoch": 10.144322344322344, "grad_norm": 1.0234375, "learning_rate": 8.523304465399376e-05, "loss": 0.1425, "step": 13847 }, { "epoch": 10.145054945054945, "grad_norm": 1.0234375, "learning_rate": 8.523060330708354e-05, "loss": 0.1371, "step": 13848 }, { "epoch": 10.145787545787545, "grad_norm": 0.73828125, "learning_rate": 8.522816179376762e-05, "loss": 0.1191, "step": 13849 }, { "epoch": 10.146520146520146, "grad_norm": 0.7734375, "learning_rate": 8.522572011405766e-05, "loss": 0.1077, "step": 13850 }, { "epoch": 10.147252747252747, "grad_norm": 1.1640625, "learning_rate": 8.522327826796539e-05, "loss": 0.1432, "step": 13851 }, { "epoch": 10.147985347985347, "grad_norm": 1.171875, "learning_rate": 8.52208362555025e-05, "loss": 0.1312, "step": 13852 }, { "epoch": 10.148717948717948, "grad_norm": 1.140625, "learning_rate": 8.521839407668065e-05, "loss": 0.1166, "step": 13853 }, { "epoch": 10.149450549450549, "grad_norm": 0.96484375, "learning_rate": 8.521595173151161e-05, "loss": 0.1431, "step": 13854 }, { "epoch": 10.15018315018315, "grad_norm": 0.83984375, "learning_rate": 8.521350922000703e-05, "loss": 0.1246, "step": 13855 }, { "epoch": 10.15091575091575, "grad_norm": 0.96484375, "learning_rate": 8.521106654217865e-05, "loss": 0.1155, "step": 13856 }, { "epoch": 10.151648351648351, "grad_norm": 1.078125, "learning_rate": 8.520862369803815e-05, "loss": 0.1422, "step": 13857 }, { "epoch": 10.152380952380952, "grad_norm": 0.70703125, "learning_rate": 8.520618068759722e-05, "loss": 0.0933, "step": 13858 }, { "epoch": 10.153113553113553, "grad_norm": 1.453125, "learning_rate": 8.520373751086761e-05, "loss": 0.1438, "step": 13859 }, { "epoch": 10.153846153846153, "grad_norm": 1.0546875, "learning_rate": 8.5201294167861e-05, "loss": 0.1215, "step": 13860 }, { "epoch": 10.154578754578754, "grad_norm": 0.83203125, "learning_rate": 8.51988506585891e-05, "loss": 0.0983, "step": 13861 }, { "epoch": 10.155311355311355, "grad_norm": 1.265625, "learning_rate": 8.51964069830636e-05, "loss": 0.1301, "step": 13862 }, { "epoch": 10.156043956043955, "grad_norm": 1.375, "learning_rate": 8.519396314129622e-05, "loss": 0.1347, "step": 13863 }, { "epoch": 10.156776556776556, "grad_norm": 0.953125, "learning_rate": 8.519151913329869e-05, "loss": 0.0923, "step": 13864 }, { "epoch": 10.157509157509157, "grad_norm": 1.421875, "learning_rate": 8.51890749590827e-05, "loss": 0.15, "step": 13865 }, { "epoch": 10.158241758241758, "grad_norm": 1.359375, "learning_rate": 8.518663061865995e-05, "loss": 0.1724, "step": 13866 }, { "epoch": 10.158974358974358, "grad_norm": 0.85546875, "learning_rate": 8.518418611204216e-05, "loss": 0.0991, "step": 13867 }, { "epoch": 10.15970695970696, "grad_norm": 0.921875, "learning_rate": 8.518174143924107e-05, "loss": 0.136, "step": 13868 }, { "epoch": 10.16043956043956, "grad_norm": 1.2109375, "learning_rate": 8.517929660026837e-05, "loss": 0.1288, "step": 13869 }, { "epoch": 10.16117216117216, "grad_norm": 0.92578125, "learning_rate": 8.517685159513577e-05, "loss": 0.0941, "step": 13870 }, { "epoch": 10.161904761904761, "grad_norm": 0.96484375, "learning_rate": 8.517440642385499e-05, "loss": 0.143, "step": 13871 }, { "epoch": 10.162637362637362, "grad_norm": 0.98046875, "learning_rate": 8.517196108643776e-05, "loss": 0.1679, "step": 13872 }, { "epoch": 10.163369963369963, "grad_norm": 1.71875, "learning_rate": 8.516951558289576e-05, "loss": 0.1217, "step": 13873 }, { "epoch": 10.164102564102564, "grad_norm": 1.0390625, "learning_rate": 8.516706991324074e-05, "loss": 0.1464, "step": 13874 }, { "epoch": 10.164835164835164, "grad_norm": 0.75390625, "learning_rate": 8.51646240774844e-05, "loss": 0.118, "step": 13875 }, { "epoch": 10.165567765567765, "grad_norm": 0.96875, "learning_rate": 8.516217807563847e-05, "loss": 0.124, "step": 13876 }, { "epoch": 10.166300366300366, "grad_norm": 0.8984375, "learning_rate": 8.51597319077147e-05, "loss": 0.1166, "step": 13877 }, { "epoch": 10.167032967032966, "grad_norm": 0.84375, "learning_rate": 8.515728557372473e-05, "loss": 0.1163, "step": 13878 }, { "epoch": 10.167765567765567, "grad_norm": 1.046875, "learning_rate": 8.515483907368033e-05, "loss": 0.1326, "step": 13879 }, { "epoch": 10.168498168498168, "grad_norm": 1.484375, "learning_rate": 8.515239240759325e-05, "loss": 0.119, "step": 13880 }, { "epoch": 10.169230769230769, "grad_norm": 1.3828125, "learning_rate": 8.514994557547516e-05, "loss": 0.1558, "step": 13881 }, { "epoch": 10.16996336996337, "grad_norm": 1.109375, "learning_rate": 8.514749857733781e-05, "loss": 0.1182, "step": 13882 }, { "epoch": 10.17069597069597, "grad_norm": 2.03125, "learning_rate": 8.514505141319292e-05, "loss": 0.1429, "step": 13883 }, { "epoch": 10.17142857142857, "grad_norm": 0.68359375, "learning_rate": 8.514260408305222e-05, "loss": 0.1236, "step": 13884 }, { "epoch": 10.172161172161172, "grad_norm": 1.40625, "learning_rate": 8.514015658692743e-05, "loss": 0.1613, "step": 13885 }, { "epoch": 10.172893772893772, "grad_norm": 0.7890625, "learning_rate": 8.513770892483028e-05, "loss": 0.1234, "step": 13886 }, { "epoch": 10.173626373626373, "grad_norm": 1.1328125, "learning_rate": 8.513526109677248e-05, "loss": 0.147, "step": 13887 }, { "epoch": 10.174358974358974, "grad_norm": 1.125, "learning_rate": 8.513281310276579e-05, "loss": 0.1713, "step": 13888 }, { "epoch": 10.175091575091574, "grad_norm": 2.03125, "learning_rate": 8.513036494282193e-05, "loss": 0.1393, "step": 13889 }, { "epoch": 10.175824175824175, "grad_norm": 0.8125, "learning_rate": 8.512791661695261e-05, "loss": 0.1321, "step": 13890 }, { "epoch": 10.176556776556776, "grad_norm": 0.62109375, "learning_rate": 8.512546812516958e-05, "loss": 0.1316, "step": 13891 }, { "epoch": 10.177289377289377, "grad_norm": 1.6953125, "learning_rate": 8.512301946748455e-05, "loss": 0.153, "step": 13892 }, { "epoch": 10.178021978021977, "grad_norm": 0.90625, "learning_rate": 8.51205706439093e-05, "loss": 0.1334, "step": 13893 }, { "epoch": 10.178754578754578, "grad_norm": 0.69921875, "learning_rate": 8.511812165445549e-05, "loss": 0.1242, "step": 13894 }, { "epoch": 10.179487179487179, "grad_norm": 0.953125, "learning_rate": 8.511567249913493e-05, "loss": 0.1472, "step": 13895 }, { "epoch": 10.18021978021978, "grad_norm": 1.1484375, "learning_rate": 8.51132231779593e-05, "loss": 0.1506, "step": 13896 }, { "epoch": 10.18095238095238, "grad_norm": 0.98046875, "learning_rate": 8.511077369094037e-05, "loss": 0.1352, "step": 13897 }, { "epoch": 10.181684981684981, "grad_norm": 1.15625, "learning_rate": 8.510832403808984e-05, "loss": 0.1388, "step": 13898 }, { "epoch": 10.182417582417582, "grad_norm": 1.9453125, "learning_rate": 8.510587421941948e-05, "loss": 0.1628, "step": 13899 }, { "epoch": 10.183150183150182, "grad_norm": 0.98046875, "learning_rate": 8.510342423494101e-05, "loss": 0.157, "step": 13900 }, { "epoch": 10.183882783882783, "grad_norm": 1.1640625, "learning_rate": 8.510097408466618e-05, "loss": 0.1186, "step": 13901 }, { "epoch": 10.184615384615384, "grad_norm": 0.87109375, "learning_rate": 8.509852376860672e-05, "loss": 0.1207, "step": 13902 }, { "epoch": 10.185347985347985, "grad_norm": 1.3359375, "learning_rate": 8.509607328677438e-05, "loss": 0.1373, "step": 13903 }, { "epoch": 10.186080586080585, "grad_norm": 0.8984375, "learning_rate": 8.50936226391809e-05, "loss": 0.1507, "step": 13904 }, { "epoch": 10.186813186813186, "grad_norm": 1.046875, "learning_rate": 8.509117182583802e-05, "loss": 0.1445, "step": 13905 }, { "epoch": 10.187545787545787, "grad_norm": 1.125, "learning_rate": 8.508872084675747e-05, "loss": 0.1345, "step": 13906 }, { "epoch": 10.188278388278388, "grad_norm": 1.6875, "learning_rate": 8.508626970195102e-05, "loss": 0.1263, "step": 13907 }, { "epoch": 10.189010989010988, "grad_norm": 1.5859375, "learning_rate": 8.508381839143037e-05, "loss": 0.1304, "step": 13908 }, { "epoch": 10.189743589743589, "grad_norm": 1.3203125, "learning_rate": 8.508136691520732e-05, "loss": 0.1558, "step": 13909 }, { "epoch": 10.19047619047619, "grad_norm": 1.6015625, "learning_rate": 8.507891527329356e-05, "loss": 0.1403, "step": 13910 }, { "epoch": 10.19120879120879, "grad_norm": 1.046875, "learning_rate": 8.50764634657009e-05, "loss": 0.1631, "step": 13911 }, { "epoch": 10.191941391941391, "grad_norm": 0.83984375, "learning_rate": 8.507401149244102e-05, "loss": 0.1606, "step": 13912 }, { "epoch": 10.192673992673992, "grad_norm": 1.484375, "learning_rate": 8.507155935352571e-05, "loss": 0.1451, "step": 13913 }, { "epoch": 10.193406593406593, "grad_norm": 1.0078125, "learning_rate": 8.506910704896672e-05, "loss": 0.1436, "step": 13914 }, { "epoch": 10.194139194139193, "grad_norm": 1.265625, "learning_rate": 8.506665457877579e-05, "loss": 0.1592, "step": 13915 }, { "epoch": 10.194871794871794, "grad_norm": 1.1328125, "learning_rate": 8.506420194296467e-05, "loss": 0.1089, "step": 13916 }, { "epoch": 10.195604395604395, "grad_norm": 0.98828125, "learning_rate": 8.506174914154511e-05, "loss": 0.1362, "step": 13917 }, { "epoch": 10.196336996336996, "grad_norm": 0.98828125, "learning_rate": 8.505929617452886e-05, "loss": 0.157, "step": 13918 }, { "epoch": 10.197069597069596, "grad_norm": 1.046875, "learning_rate": 8.505684304192768e-05, "loss": 0.1756, "step": 13919 }, { "epoch": 10.197802197802197, "grad_norm": 0.7734375, "learning_rate": 8.505438974375333e-05, "loss": 0.1256, "step": 13920 }, { "epoch": 10.198534798534798, "grad_norm": 1.2734375, "learning_rate": 8.505193628001753e-05, "loss": 0.1258, "step": 13921 }, { "epoch": 10.199267399267399, "grad_norm": 1.171875, "learning_rate": 8.504948265073208e-05, "loss": 0.1425, "step": 13922 }, { "epoch": 10.2, "grad_norm": 0.72265625, "learning_rate": 8.504702885590871e-05, "loss": 0.0981, "step": 13923 }, { "epoch": 10.2007326007326, "grad_norm": 0.85546875, "learning_rate": 8.50445748955592e-05, "loss": 0.1361, "step": 13924 }, { "epoch": 10.2014652014652, "grad_norm": 1.265625, "learning_rate": 8.504212076969526e-05, "loss": 0.156, "step": 13925 }, { "epoch": 10.202197802197801, "grad_norm": 0.9375, "learning_rate": 8.50396664783287e-05, "loss": 0.1484, "step": 13926 }, { "epoch": 10.202930402930402, "grad_norm": 1.3671875, "learning_rate": 8.503721202147125e-05, "loss": 0.161, "step": 13927 }, { "epoch": 10.203663003663003, "grad_norm": 0.79296875, "learning_rate": 8.503475739913467e-05, "loss": 0.1203, "step": 13928 }, { "epoch": 10.204395604395604, "grad_norm": 1.0546875, "learning_rate": 8.503230261133076e-05, "loss": 0.1207, "step": 13929 }, { "epoch": 10.205128205128204, "grad_norm": 1.1640625, "learning_rate": 8.502984765807125e-05, "loss": 0.1509, "step": 13930 }, { "epoch": 10.205860805860805, "grad_norm": 0.82421875, "learning_rate": 8.502739253936788e-05, "loss": 0.1397, "step": 13931 }, { "epoch": 10.206593406593406, "grad_norm": 0.80859375, "learning_rate": 8.502493725523247e-05, "loss": 0.1093, "step": 13932 }, { "epoch": 10.207326007326007, "grad_norm": 0.78515625, "learning_rate": 8.502248180567672e-05, "loss": 0.1182, "step": 13933 }, { "epoch": 10.208058608058607, "grad_norm": 0.78515625, "learning_rate": 8.502002619071244e-05, "loss": 0.1151, "step": 13934 }, { "epoch": 10.208791208791208, "grad_norm": 1.0, "learning_rate": 8.501757041035137e-05, "loss": 0.1422, "step": 13935 }, { "epoch": 10.209523809523809, "grad_norm": 1.1640625, "learning_rate": 8.501511446460531e-05, "loss": 0.1112, "step": 13936 }, { "epoch": 10.21025641025641, "grad_norm": 1.21875, "learning_rate": 8.5012658353486e-05, "loss": 0.1149, "step": 13937 }, { "epoch": 10.21098901098901, "grad_norm": 0.94921875, "learning_rate": 8.501020207700522e-05, "loss": 0.1201, "step": 13938 }, { "epoch": 10.211721611721611, "grad_norm": 1.8515625, "learning_rate": 8.500774563517472e-05, "loss": 0.1354, "step": 13939 }, { "epoch": 10.212454212454212, "grad_norm": 1.0390625, "learning_rate": 8.500528902800628e-05, "loss": 0.132, "step": 13940 }, { "epoch": 10.213186813186812, "grad_norm": 1.7578125, "learning_rate": 8.500283225551168e-05, "loss": 0.1603, "step": 13941 }, { "epoch": 10.213919413919413, "grad_norm": 1.4609375, "learning_rate": 8.500037531770269e-05, "loss": 0.1347, "step": 13942 }, { "epoch": 10.214652014652014, "grad_norm": 0.796875, "learning_rate": 8.499791821459106e-05, "loss": 0.1115, "step": 13943 }, { "epoch": 10.215384615384615, "grad_norm": 0.92578125, "learning_rate": 8.499546094618859e-05, "loss": 0.1446, "step": 13944 }, { "epoch": 10.216117216117215, "grad_norm": 1.75, "learning_rate": 8.499300351250703e-05, "loss": 0.1214, "step": 13945 }, { "epoch": 10.216849816849816, "grad_norm": 1.625, "learning_rate": 8.499054591355819e-05, "loss": 0.1562, "step": 13946 }, { "epoch": 10.217582417582417, "grad_norm": 1.40625, "learning_rate": 8.49880881493538e-05, "loss": 0.1167, "step": 13947 }, { "epoch": 10.218315018315018, "grad_norm": 1.625, "learning_rate": 8.498563021990568e-05, "loss": 0.1268, "step": 13948 }, { "epoch": 10.219047619047618, "grad_norm": 2.015625, "learning_rate": 8.498317212522557e-05, "loss": 0.1613, "step": 13949 }, { "epoch": 10.219780219780219, "grad_norm": 0.8359375, "learning_rate": 8.498071386532525e-05, "loss": 0.1181, "step": 13950 }, { "epoch": 10.22051282051282, "grad_norm": 1.3515625, "learning_rate": 8.497825544021653e-05, "loss": 0.1399, "step": 13951 }, { "epoch": 10.22124542124542, "grad_norm": 1.5234375, "learning_rate": 8.497579684991119e-05, "loss": 0.1386, "step": 13952 }, { "epoch": 10.221978021978021, "grad_norm": 1.21875, "learning_rate": 8.497333809442094e-05, "loss": 0.141, "step": 13953 }, { "epoch": 10.222710622710622, "grad_norm": 1.75, "learning_rate": 8.497087917375763e-05, "loss": 0.1275, "step": 13954 }, { "epoch": 10.223443223443223, "grad_norm": 0.875, "learning_rate": 8.496842008793304e-05, "loss": 0.1286, "step": 13955 }, { "epoch": 10.224175824175823, "grad_norm": 1.1484375, "learning_rate": 8.496596083695893e-05, "loss": 0.1095, "step": 13956 }, { "epoch": 10.224908424908424, "grad_norm": 0.7421875, "learning_rate": 8.496350142084707e-05, "loss": 0.1211, "step": 13957 }, { "epoch": 10.225641025641025, "grad_norm": 1.15625, "learning_rate": 8.496104183960925e-05, "loss": 0.1195, "step": 13958 }, { "epoch": 10.226373626373626, "grad_norm": 1.5859375, "learning_rate": 8.49585820932573e-05, "loss": 0.1584, "step": 13959 }, { "epoch": 10.227106227106226, "grad_norm": 0.875, "learning_rate": 8.495612218180296e-05, "loss": 0.1402, "step": 13960 }, { "epoch": 10.227838827838827, "grad_norm": 1.2265625, "learning_rate": 8.495366210525802e-05, "loss": 0.1278, "step": 13961 }, { "epoch": 10.228571428571428, "grad_norm": 1.390625, "learning_rate": 8.495120186363427e-05, "loss": 0.1439, "step": 13962 }, { "epoch": 10.229304029304028, "grad_norm": 1.2734375, "learning_rate": 8.494874145694352e-05, "loss": 0.1256, "step": 13963 }, { "epoch": 10.23003663003663, "grad_norm": 1.359375, "learning_rate": 8.494628088519753e-05, "loss": 0.1347, "step": 13964 }, { "epoch": 10.23076923076923, "grad_norm": 1.1796875, "learning_rate": 8.494382014840808e-05, "loss": 0.1309, "step": 13965 }, { "epoch": 10.23150183150183, "grad_norm": 0.98828125, "learning_rate": 8.494135924658701e-05, "loss": 0.1581, "step": 13966 }, { "epoch": 10.232234432234431, "grad_norm": 0.76171875, "learning_rate": 8.493889817974608e-05, "loss": 0.1257, "step": 13967 }, { "epoch": 10.232967032967032, "grad_norm": 0.5859375, "learning_rate": 8.493643694789707e-05, "loss": 0.1166, "step": 13968 }, { "epoch": 10.233699633699633, "grad_norm": 1.125, "learning_rate": 8.49339755510518e-05, "loss": 0.119, "step": 13969 }, { "epoch": 10.234432234432234, "grad_norm": 0.94921875, "learning_rate": 8.493151398922204e-05, "loss": 0.1498, "step": 13970 }, { "epoch": 10.235164835164834, "grad_norm": 1.5546875, "learning_rate": 8.49290522624196e-05, "loss": 0.153, "step": 13971 }, { "epoch": 10.235897435897435, "grad_norm": 1.9140625, "learning_rate": 8.492659037065626e-05, "loss": 0.1179, "step": 13972 }, { "epoch": 10.236630036630036, "grad_norm": 1.1015625, "learning_rate": 8.492412831394382e-05, "loss": 0.1227, "step": 13973 }, { "epoch": 10.237362637362637, "grad_norm": 0.953125, "learning_rate": 8.49216660922941e-05, "loss": 0.127, "step": 13974 }, { "epoch": 10.238095238095237, "grad_norm": 1.2890625, "learning_rate": 8.491920370571886e-05, "loss": 0.1472, "step": 13975 }, { "epoch": 10.238827838827838, "grad_norm": 1.5390625, "learning_rate": 8.491674115422992e-05, "loss": 0.1734, "step": 13976 }, { "epoch": 10.239560439560439, "grad_norm": 0.98046875, "learning_rate": 8.491427843783906e-05, "loss": 0.1492, "step": 13977 }, { "epoch": 10.24029304029304, "grad_norm": 1.1171875, "learning_rate": 8.491181555655812e-05, "loss": 0.1381, "step": 13978 }, { "epoch": 10.24102564102564, "grad_norm": 1.03125, "learning_rate": 8.490935251039886e-05, "loss": 0.122, "step": 13979 }, { "epoch": 10.241758241758241, "grad_norm": 1.171875, "learning_rate": 8.49068892993731e-05, "loss": 0.135, "step": 13980 }, { "epoch": 10.242490842490842, "grad_norm": 1.015625, "learning_rate": 8.490442592349264e-05, "loss": 0.1574, "step": 13981 }, { "epoch": 10.243223443223442, "grad_norm": 0.94140625, "learning_rate": 8.49019623827693e-05, "loss": 0.1348, "step": 13982 }, { "epoch": 10.243956043956043, "grad_norm": 1.0703125, "learning_rate": 8.489949867721484e-05, "loss": 0.1453, "step": 13983 }, { "epoch": 10.244688644688644, "grad_norm": 0.875, "learning_rate": 8.48970348068411e-05, "loss": 0.1437, "step": 13984 }, { "epoch": 10.245421245421245, "grad_norm": 0.7578125, "learning_rate": 8.489457077165987e-05, "loss": 0.1229, "step": 13985 }, { "epoch": 10.246153846153845, "grad_norm": 0.98828125, "learning_rate": 8.489210657168296e-05, "loss": 0.1401, "step": 13986 }, { "epoch": 10.246886446886446, "grad_norm": 1.0546875, "learning_rate": 8.488964220692217e-05, "loss": 0.1305, "step": 13987 }, { "epoch": 10.247619047619047, "grad_norm": 1.5546875, "learning_rate": 8.488717767738931e-05, "loss": 0.158, "step": 13988 }, { "epoch": 10.248351648351647, "grad_norm": 0.859375, "learning_rate": 8.488471298309621e-05, "loss": 0.1346, "step": 13989 }, { "epoch": 10.249084249084248, "grad_norm": 1.25, "learning_rate": 8.488224812405466e-05, "loss": 0.1461, "step": 13990 }, { "epoch": 10.249816849816849, "grad_norm": 1.078125, "learning_rate": 8.487978310027648e-05, "loss": 0.129, "step": 13991 }, { "epoch": 10.25054945054945, "grad_norm": 1.15625, "learning_rate": 8.487731791177345e-05, "loss": 0.1371, "step": 13992 }, { "epoch": 10.25128205128205, "grad_norm": 1.3203125, "learning_rate": 8.487485255855743e-05, "loss": 0.1175, "step": 13993 }, { "epoch": 10.252014652014651, "grad_norm": 1.5078125, "learning_rate": 8.487238704064019e-05, "loss": 0.1471, "step": 13994 }, { "epoch": 10.252747252747252, "grad_norm": 1.765625, "learning_rate": 8.486992135803357e-05, "loss": 0.1193, "step": 13995 }, { "epoch": 10.253479853479853, "grad_norm": 1.484375, "learning_rate": 8.486745551074936e-05, "loss": 0.1467, "step": 13996 }, { "epoch": 10.254212454212453, "grad_norm": 1.1171875, "learning_rate": 8.486498949879939e-05, "loss": 0.1115, "step": 13997 }, { "epoch": 10.254945054945054, "grad_norm": 0.87890625, "learning_rate": 8.48625233221955e-05, "loss": 0.1453, "step": 13998 }, { "epoch": 10.255677655677655, "grad_norm": 1.625, "learning_rate": 8.486005698094944e-05, "loss": 0.1367, "step": 13999 }, { "epoch": 10.256410256410255, "grad_norm": 0.66796875, "learning_rate": 8.48575904750731e-05, "loss": 0.1304, "step": 14000 }, { "epoch": 10.257142857142856, "grad_norm": 1.6171875, "learning_rate": 8.485512380457826e-05, "loss": 0.1365, "step": 14001 }, { "epoch": 10.257875457875459, "grad_norm": 1.71875, "learning_rate": 8.485265696947674e-05, "loss": 0.1262, "step": 14002 }, { "epoch": 10.25860805860806, "grad_norm": 1.609375, "learning_rate": 8.485018996978035e-05, "loss": 0.1398, "step": 14003 }, { "epoch": 10.25934065934066, "grad_norm": 1.5, "learning_rate": 8.484772280550094e-05, "loss": 0.1408, "step": 14004 }, { "epoch": 10.260073260073261, "grad_norm": 0.5703125, "learning_rate": 8.484525547665031e-05, "loss": 0.1148, "step": 14005 }, { "epoch": 10.260805860805862, "grad_norm": 0.96484375, "learning_rate": 8.484278798324027e-05, "loss": 0.107, "step": 14006 }, { "epoch": 10.261538461538462, "grad_norm": 1.1796875, "learning_rate": 8.484032032528269e-05, "loss": 0.1297, "step": 14007 }, { "epoch": 10.262271062271063, "grad_norm": 1.25, "learning_rate": 8.483785250278934e-05, "loss": 0.139, "step": 14008 }, { "epoch": 10.263003663003664, "grad_norm": 1.375, "learning_rate": 8.483538451577207e-05, "loss": 0.1215, "step": 14009 }, { "epoch": 10.263736263736265, "grad_norm": 0.94140625, "learning_rate": 8.48329163642427e-05, "loss": 0.1445, "step": 14010 }, { "epoch": 10.264468864468865, "grad_norm": 1.125, "learning_rate": 8.483044804821306e-05, "loss": 0.1501, "step": 14011 }, { "epoch": 10.265201465201466, "grad_norm": 0.94921875, "learning_rate": 8.482797956769498e-05, "loss": 0.1505, "step": 14012 }, { "epoch": 10.265934065934067, "grad_norm": 1.28125, "learning_rate": 8.482551092270027e-05, "loss": 0.1271, "step": 14013 }, { "epoch": 10.266666666666667, "grad_norm": 2.109375, "learning_rate": 8.482304211324078e-05, "loss": 0.1646, "step": 14014 }, { "epoch": 10.267399267399268, "grad_norm": 1.09375, "learning_rate": 8.482057313932832e-05, "loss": 0.1302, "step": 14015 }, { "epoch": 10.268131868131869, "grad_norm": 1.859375, "learning_rate": 8.481810400097473e-05, "loss": 0.1261, "step": 14016 }, { "epoch": 10.26886446886447, "grad_norm": 1.03125, "learning_rate": 8.481563469819182e-05, "loss": 0.1246, "step": 14017 }, { "epoch": 10.26959706959707, "grad_norm": 1.5390625, "learning_rate": 8.481316523099148e-05, "loss": 0.1726, "step": 14018 }, { "epoch": 10.270329670329671, "grad_norm": 1.28125, "learning_rate": 8.481069559938546e-05, "loss": 0.1318, "step": 14019 }, { "epoch": 10.271062271062272, "grad_norm": 1.359375, "learning_rate": 8.480822580338564e-05, "loss": 0.1598, "step": 14020 }, { "epoch": 10.271794871794873, "grad_norm": 1.0625, "learning_rate": 8.480575584300388e-05, "loss": 0.1206, "step": 14021 }, { "epoch": 10.272527472527473, "grad_norm": 1.03125, "learning_rate": 8.480328571825196e-05, "loss": 0.1117, "step": 14022 }, { "epoch": 10.273260073260074, "grad_norm": 0.6953125, "learning_rate": 8.480081542914173e-05, "loss": 0.1366, "step": 14023 }, { "epoch": 10.273992673992675, "grad_norm": 1.0234375, "learning_rate": 8.479834497568505e-05, "loss": 0.1226, "step": 14024 }, { "epoch": 10.274725274725276, "grad_norm": 0.84765625, "learning_rate": 8.479587435789371e-05, "loss": 0.1299, "step": 14025 }, { "epoch": 10.275457875457876, "grad_norm": 0.94140625, "learning_rate": 8.47934035757796e-05, "loss": 0.1187, "step": 14026 }, { "epoch": 10.276190476190477, "grad_norm": 1.0078125, "learning_rate": 8.479093262935454e-05, "loss": 0.1454, "step": 14027 }, { "epoch": 10.276923076923078, "grad_norm": 0.73046875, "learning_rate": 8.478846151863036e-05, "loss": 0.1097, "step": 14028 }, { "epoch": 10.277655677655678, "grad_norm": 1.1640625, "learning_rate": 8.47859902436189e-05, "loss": 0.1557, "step": 14029 }, { "epoch": 10.27838827838828, "grad_norm": 1.453125, "learning_rate": 8.478351880433199e-05, "loss": 0.1419, "step": 14030 }, { "epoch": 10.27912087912088, "grad_norm": 0.78125, "learning_rate": 8.478104720078152e-05, "loss": 0.1259, "step": 14031 }, { "epoch": 10.27985347985348, "grad_norm": 1.1875, "learning_rate": 8.477857543297928e-05, "loss": 0.1395, "step": 14032 }, { "epoch": 10.280586080586081, "grad_norm": 0.9921875, "learning_rate": 8.477610350093714e-05, "loss": 0.1203, "step": 14033 }, { "epoch": 10.281318681318682, "grad_norm": 1.578125, "learning_rate": 8.477363140466694e-05, "loss": 0.1745, "step": 14034 }, { "epoch": 10.282051282051283, "grad_norm": 1.609375, "learning_rate": 8.47711591441805e-05, "loss": 0.1487, "step": 14035 }, { "epoch": 10.282783882783884, "grad_norm": 0.9921875, "learning_rate": 8.47686867194897e-05, "loss": 0.1363, "step": 14036 }, { "epoch": 10.283516483516484, "grad_norm": 1.3203125, "learning_rate": 8.476621413060637e-05, "loss": 0.1086, "step": 14037 }, { "epoch": 10.284249084249085, "grad_norm": 1.0390625, "learning_rate": 8.476374137754235e-05, "loss": 0.1326, "step": 14038 }, { "epoch": 10.284981684981686, "grad_norm": 1.7734375, "learning_rate": 8.476126846030954e-05, "loss": 0.1359, "step": 14039 }, { "epoch": 10.285714285714286, "grad_norm": 1.0625, "learning_rate": 8.475879537891969e-05, "loss": 0.1308, "step": 14040 }, { "epoch": 10.286446886446887, "grad_norm": 1.0546875, "learning_rate": 8.475632213338473e-05, "loss": 0.1346, "step": 14041 }, { "epoch": 10.287179487179488, "grad_norm": 1.0078125, "learning_rate": 8.475384872371649e-05, "loss": 0.1046, "step": 14042 }, { "epoch": 10.287912087912089, "grad_norm": 1.03125, "learning_rate": 8.47513751499268e-05, "loss": 0.1255, "step": 14043 }, { "epoch": 10.28864468864469, "grad_norm": 0.9609375, "learning_rate": 8.474890141202754e-05, "loss": 0.1333, "step": 14044 }, { "epoch": 10.28937728937729, "grad_norm": 2.25, "learning_rate": 8.474642751003054e-05, "loss": 0.1601, "step": 14045 }, { "epoch": 10.29010989010989, "grad_norm": 2.25, "learning_rate": 8.474395344394768e-05, "loss": 0.1461, "step": 14046 }, { "epoch": 10.290842490842492, "grad_norm": 1.8125, "learning_rate": 8.474147921379078e-05, "loss": 0.1504, "step": 14047 }, { "epoch": 10.291575091575092, "grad_norm": 1.78125, "learning_rate": 8.473900481957172e-05, "loss": 0.1289, "step": 14048 }, { "epoch": 10.292307692307693, "grad_norm": 1.3828125, "learning_rate": 8.473653026130234e-05, "loss": 0.1462, "step": 14049 }, { "epoch": 10.293040293040294, "grad_norm": 1.3046875, "learning_rate": 8.473405553899451e-05, "loss": 0.1368, "step": 14050 }, { "epoch": 10.293772893772895, "grad_norm": 0.8125, "learning_rate": 8.47315806526601e-05, "loss": 0.108, "step": 14051 }, { "epoch": 10.294505494505495, "grad_norm": 1.046875, "learning_rate": 8.472910560231093e-05, "loss": 0.1012, "step": 14052 }, { "epoch": 10.295238095238096, "grad_norm": 1.5859375, "learning_rate": 8.472663038795888e-05, "loss": 0.1617, "step": 14053 }, { "epoch": 10.295970695970697, "grad_norm": 0.75, "learning_rate": 8.47241550096158e-05, "loss": 0.1372, "step": 14054 }, { "epoch": 10.296703296703297, "grad_norm": 0.8984375, "learning_rate": 8.472167946729356e-05, "loss": 0.1174, "step": 14055 }, { "epoch": 10.297435897435898, "grad_norm": 1.0703125, "learning_rate": 8.471920376100403e-05, "loss": 0.1174, "step": 14056 }, { "epoch": 10.298168498168499, "grad_norm": 1.21875, "learning_rate": 8.471672789075907e-05, "loss": 0.1486, "step": 14057 }, { "epoch": 10.2989010989011, "grad_norm": 1.171875, "learning_rate": 8.471425185657053e-05, "loss": 0.1006, "step": 14058 }, { "epoch": 10.2996336996337, "grad_norm": 1.6484375, "learning_rate": 8.471177565845025e-05, "loss": 0.1305, "step": 14059 }, { "epoch": 10.300366300366301, "grad_norm": 1.28125, "learning_rate": 8.470929929641014e-05, "loss": 0.1291, "step": 14060 }, { "epoch": 10.301098901098902, "grad_norm": 1.359375, "learning_rate": 8.470682277046206e-05, "loss": 0.1623, "step": 14061 }, { "epoch": 10.301831501831503, "grad_norm": 1.15625, "learning_rate": 8.470434608061785e-05, "loss": 0.1471, "step": 14062 }, { "epoch": 10.302564102564103, "grad_norm": 1.25, "learning_rate": 8.47018692268894e-05, "loss": 0.128, "step": 14063 }, { "epoch": 10.303296703296704, "grad_norm": 1.265625, "learning_rate": 8.469939220928855e-05, "loss": 0.1272, "step": 14064 }, { "epoch": 10.304029304029305, "grad_norm": 1.703125, "learning_rate": 8.46969150278272e-05, "loss": 0.1425, "step": 14065 }, { "epoch": 10.304761904761905, "grad_norm": 1.046875, "learning_rate": 8.46944376825172e-05, "loss": 0.1507, "step": 14066 }, { "epoch": 10.305494505494506, "grad_norm": 1.1484375, "learning_rate": 8.469196017337041e-05, "loss": 0.1259, "step": 14067 }, { "epoch": 10.306227106227107, "grad_norm": 1.3125, "learning_rate": 8.468948250039873e-05, "loss": 0.1161, "step": 14068 }, { "epoch": 10.306959706959708, "grad_norm": 0.90234375, "learning_rate": 8.4687004663614e-05, "loss": 0.112, "step": 14069 }, { "epoch": 10.307692307692308, "grad_norm": 1.0234375, "learning_rate": 8.468452666302813e-05, "loss": 0.125, "step": 14070 }, { "epoch": 10.308424908424909, "grad_norm": 1.1796875, "learning_rate": 8.468204849865294e-05, "loss": 0.1652, "step": 14071 }, { "epoch": 10.30915750915751, "grad_norm": 1.7734375, "learning_rate": 8.467957017050038e-05, "loss": 0.1301, "step": 14072 }, { "epoch": 10.30989010989011, "grad_norm": 0.59375, "learning_rate": 8.467709167858225e-05, "loss": 0.0906, "step": 14073 }, { "epoch": 10.310622710622711, "grad_norm": 1.1796875, "learning_rate": 8.467461302291045e-05, "loss": 0.1217, "step": 14074 }, { "epoch": 10.311355311355312, "grad_norm": 1.1328125, "learning_rate": 8.467213420349687e-05, "loss": 0.1406, "step": 14075 }, { "epoch": 10.312087912087913, "grad_norm": 0.875, "learning_rate": 8.466965522035337e-05, "loss": 0.1133, "step": 14076 }, { "epoch": 10.312820512820513, "grad_norm": 1.234375, "learning_rate": 8.466717607349183e-05, "loss": 0.0957, "step": 14077 }, { "epoch": 10.313553113553114, "grad_norm": 1.2109375, "learning_rate": 8.466469676292415e-05, "loss": 0.1196, "step": 14078 }, { "epoch": 10.314285714285715, "grad_norm": 1.125, "learning_rate": 8.466221728866218e-05, "loss": 0.1204, "step": 14079 }, { "epoch": 10.315018315018316, "grad_norm": 0.62890625, "learning_rate": 8.465973765071781e-05, "loss": 0.0986, "step": 14080 }, { "epoch": 10.315750915750916, "grad_norm": 0.91015625, "learning_rate": 8.465725784910294e-05, "loss": 0.1312, "step": 14081 }, { "epoch": 10.316483516483517, "grad_norm": 1.0390625, "learning_rate": 8.46547778838294e-05, "loss": 0.1162, "step": 14082 }, { "epoch": 10.317216117216118, "grad_norm": 0.93359375, "learning_rate": 8.465229775490916e-05, "loss": 0.1249, "step": 14083 }, { "epoch": 10.317948717948719, "grad_norm": 1.1640625, "learning_rate": 8.4649817462354e-05, "loss": 0.1438, "step": 14084 }, { "epoch": 10.31868131868132, "grad_norm": 0.98046875, "learning_rate": 8.464733700617589e-05, "loss": 0.1378, "step": 14085 }, { "epoch": 10.31941391941392, "grad_norm": 1.7578125, "learning_rate": 8.464485638638666e-05, "loss": 0.1608, "step": 14086 }, { "epoch": 10.32014652014652, "grad_norm": 1.515625, "learning_rate": 8.464237560299821e-05, "loss": 0.1525, "step": 14087 }, { "epoch": 10.320879120879122, "grad_norm": 1.1875, "learning_rate": 8.463989465602244e-05, "loss": 0.133, "step": 14088 }, { "epoch": 10.321611721611722, "grad_norm": 1.109375, "learning_rate": 8.463741354547123e-05, "loss": 0.1358, "step": 14089 }, { "epoch": 10.322344322344323, "grad_norm": 0.91796875, "learning_rate": 8.463493227135646e-05, "loss": 0.1435, "step": 14090 }, { "epoch": 10.323076923076924, "grad_norm": 0.8359375, "learning_rate": 8.463245083369003e-05, "loss": 0.1372, "step": 14091 }, { "epoch": 10.323809523809524, "grad_norm": 1.1015625, "learning_rate": 8.46299692324838e-05, "loss": 0.114, "step": 14092 }, { "epoch": 10.324542124542125, "grad_norm": 0.8828125, "learning_rate": 8.462748746774969e-05, "loss": 0.1508, "step": 14093 }, { "epoch": 10.325274725274726, "grad_norm": 1.3671875, "learning_rate": 8.462500553949962e-05, "loss": 0.1167, "step": 14094 }, { "epoch": 10.326007326007327, "grad_norm": 1.0390625, "learning_rate": 8.462252344774541e-05, "loss": 0.1039, "step": 14095 }, { "epoch": 10.326739926739927, "grad_norm": 0.859375, "learning_rate": 8.4620041192499e-05, "loss": 0.1553, "step": 14096 }, { "epoch": 10.327472527472528, "grad_norm": 1.5859375, "learning_rate": 8.461755877377227e-05, "loss": 0.1206, "step": 14097 }, { "epoch": 10.328205128205129, "grad_norm": 1.1875, "learning_rate": 8.461507619157711e-05, "loss": 0.1315, "step": 14098 }, { "epoch": 10.32893772893773, "grad_norm": 0.80078125, "learning_rate": 8.461259344592544e-05, "loss": 0.1331, "step": 14099 }, { "epoch": 10.32967032967033, "grad_norm": 1.4765625, "learning_rate": 8.461011053682911e-05, "loss": 0.1752, "step": 14100 }, { "epoch": 10.330402930402931, "grad_norm": 1.09375, "learning_rate": 8.460762746430005e-05, "loss": 0.1201, "step": 14101 }, { "epoch": 10.331135531135532, "grad_norm": 0.83203125, "learning_rate": 8.460514422835016e-05, "loss": 0.1231, "step": 14102 }, { "epoch": 10.331868131868132, "grad_norm": 1.6328125, "learning_rate": 8.460266082899133e-05, "loss": 0.1336, "step": 14103 }, { "epoch": 10.332600732600733, "grad_norm": 1.0078125, "learning_rate": 8.460017726623546e-05, "loss": 0.0988, "step": 14104 }, { "epoch": 10.333333333333334, "grad_norm": 1.0625, "learning_rate": 8.459769354009442e-05, "loss": 0.1175, "step": 14105 }, { "epoch": 10.334065934065935, "grad_norm": 1.3984375, "learning_rate": 8.459520965058015e-05, "loss": 0.1326, "step": 14106 }, { "epoch": 10.334798534798535, "grad_norm": 0.7734375, "learning_rate": 8.459272559770455e-05, "loss": 0.1141, "step": 14107 }, { "epoch": 10.335531135531136, "grad_norm": 0.91015625, "learning_rate": 8.459024138147949e-05, "loss": 0.1176, "step": 14108 }, { "epoch": 10.336263736263737, "grad_norm": 1.359375, "learning_rate": 8.458775700191693e-05, "loss": 0.1273, "step": 14109 }, { "epoch": 10.336996336996338, "grad_norm": 1.625, "learning_rate": 8.458527245902869e-05, "loss": 0.149, "step": 14110 }, { "epoch": 10.337728937728938, "grad_norm": 0.8046875, "learning_rate": 8.458278775282675e-05, "loss": 0.1273, "step": 14111 }, { "epoch": 10.338461538461539, "grad_norm": 1.71875, "learning_rate": 8.458030288332298e-05, "loss": 0.1477, "step": 14112 }, { "epoch": 10.33919413919414, "grad_norm": 0.75390625, "learning_rate": 8.457781785052928e-05, "loss": 0.1459, "step": 14113 }, { "epoch": 10.33992673992674, "grad_norm": 0.984375, "learning_rate": 8.457533265445758e-05, "loss": 0.1102, "step": 14114 }, { "epoch": 10.340659340659341, "grad_norm": 1.328125, "learning_rate": 8.457284729511976e-05, "loss": 0.1404, "step": 14115 }, { "epoch": 10.341391941391942, "grad_norm": 0.625, "learning_rate": 8.457036177252775e-05, "loss": 0.1168, "step": 14116 }, { "epoch": 10.342124542124543, "grad_norm": 1.078125, "learning_rate": 8.456787608669346e-05, "loss": 0.1593, "step": 14117 }, { "epoch": 10.342857142857143, "grad_norm": 1.2734375, "learning_rate": 8.456539023762878e-05, "loss": 0.1293, "step": 14118 }, { "epoch": 10.343589743589744, "grad_norm": 0.78125, "learning_rate": 8.456290422534564e-05, "loss": 0.1054, "step": 14119 }, { "epoch": 10.344322344322345, "grad_norm": 0.87109375, "learning_rate": 8.456041804985594e-05, "loss": 0.1256, "step": 14120 }, { "epoch": 10.345054945054946, "grad_norm": 0.8125, "learning_rate": 8.45579317111716e-05, "loss": 0.1296, "step": 14121 }, { "epoch": 10.345787545787546, "grad_norm": 1.1328125, "learning_rate": 8.455544520930453e-05, "loss": 0.1358, "step": 14122 }, { "epoch": 10.346520146520147, "grad_norm": 0.60546875, "learning_rate": 8.455295854426664e-05, "loss": 0.1347, "step": 14123 }, { "epoch": 10.347252747252748, "grad_norm": 1.0078125, "learning_rate": 8.455047171606985e-05, "loss": 0.1312, "step": 14124 }, { "epoch": 10.347985347985349, "grad_norm": 0.8359375, "learning_rate": 8.454798472472607e-05, "loss": 0.121, "step": 14125 }, { "epoch": 10.34871794871795, "grad_norm": 1.5859375, "learning_rate": 8.454549757024721e-05, "loss": 0.1537, "step": 14126 }, { "epoch": 10.34945054945055, "grad_norm": 1.0234375, "learning_rate": 8.45430102526452e-05, "loss": 0.1342, "step": 14127 }, { "epoch": 10.35018315018315, "grad_norm": 0.9453125, "learning_rate": 8.454052277193194e-05, "loss": 0.1638, "step": 14128 }, { "epoch": 10.350915750915751, "grad_norm": 1.1015625, "learning_rate": 8.453803512811939e-05, "loss": 0.1281, "step": 14129 }, { "epoch": 10.351648351648352, "grad_norm": 1.2578125, "learning_rate": 8.453554732121942e-05, "loss": 0.1678, "step": 14130 }, { "epoch": 10.352380952380953, "grad_norm": 0.80859375, "learning_rate": 8.453305935124397e-05, "loss": 0.1117, "step": 14131 }, { "epoch": 10.353113553113554, "grad_norm": 1.296875, "learning_rate": 8.453057121820496e-05, "loss": 0.1266, "step": 14132 }, { "epoch": 10.353846153846154, "grad_norm": 0.890625, "learning_rate": 8.452808292211431e-05, "loss": 0.136, "step": 14133 }, { "epoch": 10.354578754578755, "grad_norm": 0.98046875, "learning_rate": 8.452559446298396e-05, "loss": 0.171, "step": 14134 }, { "epoch": 10.355311355311356, "grad_norm": 1.265625, "learning_rate": 8.452310584082581e-05, "loss": 0.1884, "step": 14135 }, { "epoch": 10.356043956043957, "grad_norm": 1.0234375, "learning_rate": 8.452061705565178e-05, "loss": 0.1315, "step": 14136 }, { "epoch": 10.356776556776557, "grad_norm": 1.0390625, "learning_rate": 8.451812810747381e-05, "loss": 0.1256, "step": 14137 }, { "epoch": 10.357509157509158, "grad_norm": 1.2578125, "learning_rate": 8.451563899630383e-05, "loss": 0.1381, "step": 14138 }, { "epoch": 10.358241758241759, "grad_norm": 0.6953125, "learning_rate": 8.451314972215375e-05, "loss": 0.1309, "step": 14139 }, { "epoch": 10.35897435897436, "grad_norm": 0.74609375, "learning_rate": 8.45106602850355e-05, "loss": 0.1145, "step": 14140 }, { "epoch": 10.35970695970696, "grad_norm": 1.2734375, "learning_rate": 8.450817068496102e-05, "loss": 0.1304, "step": 14141 }, { "epoch": 10.360439560439561, "grad_norm": 0.87890625, "learning_rate": 8.450568092194222e-05, "loss": 0.1404, "step": 14142 }, { "epoch": 10.361172161172162, "grad_norm": 1.171875, "learning_rate": 8.450319099599106e-05, "loss": 0.1649, "step": 14143 }, { "epoch": 10.361904761904762, "grad_norm": 1.25, "learning_rate": 8.450070090711942e-05, "loss": 0.1429, "step": 14144 }, { "epoch": 10.362637362637363, "grad_norm": 1.015625, "learning_rate": 8.449821065533929e-05, "loss": 0.1545, "step": 14145 }, { "epoch": 10.363369963369964, "grad_norm": 0.625, "learning_rate": 8.449572024066256e-05, "loss": 0.1065, "step": 14146 }, { "epoch": 10.364102564102565, "grad_norm": 0.96875, "learning_rate": 8.449322966310116e-05, "loss": 0.1383, "step": 14147 }, { "epoch": 10.364835164835165, "grad_norm": 0.8828125, "learning_rate": 8.449073892266706e-05, "loss": 0.1109, "step": 14148 }, { "epoch": 10.365567765567766, "grad_norm": 0.83984375, "learning_rate": 8.448824801937215e-05, "loss": 0.1341, "step": 14149 }, { "epoch": 10.366300366300367, "grad_norm": 0.984375, "learning_rate": 8.448575695322838e-05, "loss": 0.1536, "step": 14150 }, { "epoch": 10.367032967032968, "grad_norm": 0.97265625, "learning_rate": 8.448326572424772e-05, "loss": 0.1426, "step": 14151 }, { "epoch": 10.367765567765568, "grad_norm": 1.796875, "learning_rate": 8.448077433244207e-05, "loss": 0.1228, "step": 14152 }, { "epoch": 10.368498168498169, "grad_norm": 1.0078125, "learning_rate": 8.447828277782337e-05, "loss": 0.1246, "step": 14153 }, { "epoch": 10.36923076923077, "grad_norm": 0.80859375, "learning_rate": 8.447579106040356e-05, "loss": 0.1148, "step": 14154 }, { "epoch": 10.36996336996337, "grad_norm": 1.046875, "learning_rate": 8.447329918019459e-05, "loss": 0.1316, "step": 14155 }, { "epoch": 10.370695970695971, "grad_norm": 1.34375, "learning_rate": 8.447080713720838e-05, "loss": 0.1159, "step": 14156 }, { "epoch": 10.371428571428572, "grad_norm": 1.0390625, "learning_rate": 8.44683149314569e-05, "loss": 0.1197, "step": 14157 }, { "epoch": 10.372161172161173, "grad_norm": 1.1328125, "learning_rate": 8.446582256295206e-05, "loss": 0.1364, "step": 14158 }, { "epoch": 10.372893772893773, "grad_norm": 1.265625, "learning_rate": 8.44633300317058e-05, "loss": 0.166, "step": 14159 }, { "epoch": 10.373626373626374, "grad_norm": 1.1953125, "learning_rate": 8.446083733773009e-05, "loss": 0.1282, "step": 14160 }, { "epoch": 10.374358974358975, "grad_norm": 1.546875, "learning_rate": 8.445834448103687e-05, "loss": 0.1278, "step": 14161 }, { "epoch": 10.375091575091576, "grad_norm": 1.375, "learning_rate": 8.445585146163804e-05, "loss": 0.1754, "step": 14162 }, { "epoch": 10.375824175824176, "grad_norm": 1.28125, "learning_rate": 8.445335827954561e-05, "loss": 0.1431, "step": 14163 }, { "epoch": 10.376556776556777, "grad_norm": 0.92578125, "learning_rate": 8.445086493477149e-05, "loss": 0.1618, "step": 14164 }, { "epoch": 10.377289377289378, "grad_norm": 2.0625, "learning_rate": 8.444837142732762e-05, "loss": 0.1461, "step": 14165 }, { "epoch": 10.378021978021978, "grad_norm": 2.3125, "learning_rate": 8.444587775722598e-05, "loss": 0.1297, "step": 14166 }, { "epoch": 10.37875457875458, "grad_norm": 0.94140625, "learning_rate": 8.444338392447848e-05, "loss": 0.1134, "step": 14167 }, { "epoch": 10.37948717948718, "grad_norm": 0.953125, "learning_rate": 8.444088992909707e-05, "loss": 0.1751, "step": 14168 }, { "epoch": 10.38021978021978, "grad_norm": 0.8203125, "learning_rate": 8.443839577109373e-05, "loss": 0.1362, "step": 14169 }, { "epoch": 10.380952380952381, "grad_norm": 0.96875, "learning_rate": 8.44359014504804e-05, "loss": 0.1156, "step": 14170 }, { "epoch": 10.381684981684982, "grad_norm": 0.56640625, "learning_rate": 8.443340696726903e-05, "loss": 0.106, "step": 14171 }, { "epoch": 10.382417582417583, "grad_norm": 1.6875, "learning_rate": 8.443091232147156e-05, "loss": 0.1587, "step": 14172 }, { "epoch": 10.383150183150184, "grad_norm": 0.96484375, "learning_rate": 8.442841751309994e-05, "loss": 0.122, "step": 14173 }, { "epoch": 10.383882783882784, "grad_norm": 1.3125, "learning_rate": 8.442592254216615e-05, "loss": 0.1536, "step": 14174 }, { "epoch": 10.384615384615385, "grad_norm": 1.1328125, "learning_rate": 8.442342740868211e-05, "loss": 0.1455, "step": 14175 }, { "epoch": 10.385347985347986, "grad_norm": 0.72265625, "learning_rate": 8.44209321126598e-05, "loss": 0.1084, "step": 14176 }, { "epoch": 10.386080586080586, "grad_norm": 0.92578125, "learning_rate": 8.441843665411117e-05, "loss": 0.1026, "step": 14177 }, { "epoch": 10.386813186813187, "grad_norm": 0.60546875, "learning_rate": 8.441594103304817e-05, "loss": 0.0956, "step": 14178 }, { "epoch": 10.387545787545788, "grad_norm": 1.2421875, "learning_rate": 8.441344524948278e-05, "loss": 0.1115, "step": 14179 }, { "epoch": 10.388278388278389, "grad_norm": 1.28125, "learning_rate": 8.441094930342692e-05, "loss": 0.1375, "step": 14180 }, { "epoch": 10.38901098901099, "grad_norm": 1.8828125, "learning_rate": 8.440845319489259e-05, "loss": 0.1587, "step": 14181 }, { "epoch": 10.38974358974359, "grad_norm": 1.109375, "learning_rate": 8.440595692389172e-05, "loss": 0.1536, "step": 14182 }, { "epoch": 10.39047619047619, "grad_norm": 1.3359375, "learning_rate": 8.440346049043628e-05, "loss": 0.1295, "step": 14183 }, { "epoch": 10.391208791208792, "grad_norm": 0.80859375, "learning_rate": 8.440096389453822e-05, "loss": 0.1173, "step": 14184 }, { "epoch": 10.391941391941392, "grad_norm": 0.99609375, "learning_rate": 8.439846713620953e-05, "loss": 0.1174, "step": 14185 }, { "epoch": 10.392673992673993, "grad_norm": 1.359375, "learning_rate": 8.439597021546214e-05, "loss": 0.151, "step": 14186 }, { "epoch": 10.393406593406594, "grad_norm": 1.109375, "learning_rate": 8.439347313230805e-05, "loss": 0.1038, "step": 14187 }, { "epoch": 10.394139194139195, "grad_norm": 0.76171875, "learning_rate": 8.439097588675919e-05, "loss": 0.0956, "step": 14188 }, { "epoch": 10.394871794871795, "grad_norm": 1.2421875, "learning_rate": 8.438847847882754e-05, "loss": 0.1211, "step": 14189 }, { "epoch": 10.395604395604396, "grad_norm": 1.2734375, "learning_rate": 8.438598090852506e-05, "loss": 0.1081, "step": 14190 }, { "epoch": 10.396336996336997, "grad_norm": 1.3046875, "learning_rate": 8.438348317586373e-05, "loss": 0.1103, "step": 14191 }, { "epoch": 10.397069597069597, "grad_norm": 1.1796875, "learning_rate": 8.438098528085551e-05, "loss": 0.1378, "step": 14192 }, { "epoch": 10.397802197802198, "grad_norm": 1.078125, "learning_rate": 8.437848722351237e-05, "loss": 0.1228, "step": 14193 }, { "epoch": 10.398534798534799, "grad_norm": 1.046875, "learning_rate": 8.437598900384626e-05, "loss": 0.1235, "step": 14194 }, { "epoch": 10.3992673992674, "grad_norm": 0.79296875, "learning_rate": 8.437349062186918e-05, "loss": 0.1331, "step": 14195 }, { "epoch": 10.4, "grad_norm": 0.96484375, "learning_rate": 8.437099207759306e-05, "loss": 0.0944, "step": 14196 }, { "epoch": 10.400732600732601, "grad_norm": 1.234375, "learning_rate": 8.436849337102992e-05, "loss": 0.1733, "step": 14197 }, { "epoch": 10.401465201465202, "grad_norm": 1.296875, "learning_rate": 8.436599450219171e-05, "loss": 0.132, "step": 14198 }, { "epoch": 10.402197802197803, "grad_norm": 1.078125, "learning_rate": 8.43634954710904e-05, "loss": 0.1322, "step": 14199 }, { "epoch": 10.402930402930403, "grad_norm": 0.83203125, "learning_rate": 8.436099627773799e-05, "loss": 0.1077, "step": 14200 }, { "epoch": 10.403663003663004, "grad_norm": 0.97265625, "learning_rate": 8.43584969221464e-05, "loss": 0.1626, "step": 14201 }, { "epoch": 10.404395604395605, "grad_norm": 1.328125, "learning_rate": 8.435599740432764e-05, "loss": 0.1438, "step": 14202 }, { "epoch": 10.405128205128205, "grad_norm": 0.859375, "learning_rate": 8.435349772429369e-05, "loss": 0.1132, "step": 14203 }, { "epoch": 10.405860805860806, "grad_norm": 1.1328125, "learning_rate": 8.435099788205652e-05, "loss": 0.1205, "step": 14204 }, { "epoch": 10.406593406593407, "grad_norm": 1.171875, "learning_rate": 8.43484978776281e-05, "loss": 0.1431, "step": 14205 }, { "epoch": 10.407326007326008, "grad_norm": 0.80859375, "learning_rate": 8.434599771102042e-05, "loss": 0.1411, "step": 14206 }, { "epoch": 10.408058608058608, "grad_norm": 1.40625, "learning_rate": 8.434349738224545e-05, "loss": 0.1318, "step": 14207 }, { "epoch": 10.40879120879121, "grad_norm": 0.921875, "learning_rate": 8.434099689131518e-05, "loss": 0.1189, "step": 14208 }, { "epoch": 10.40952380952381, "grad_norm": 0.8359375, "learning_rate": 8.433849623824157e-05, "loss": 0.1478, "step": 14209 }, { "epoch": 10.41025641025641, "grad_norm": 0.87109375, "learning_rate": 8.433599542303664e-05, "loss": 0.16, "step": 14210 }, { "epoch": 10.410989010989011, "grad_norm": 0.984375, "learning_rate": 8.433349444571233e-05, "loss": 0.116, "step": 14211 }, { "epoch": 10.411721611721612, "grad_norm": 0.6640625, "learning_rate": 8.433099330628065e-05, "loss": 0.1337, "step": 14212 }, { "epoch": 10.412454212454213, "grad_norm": 1.1484375, "learning_rate": 8.432849200475357e-05, "loss": 0.1694, "step": 14213 }, { "epoch": 10.413186813186813, "grad_norm": 1.5546875, "learning_rate": 8.432599054114309e-05, "loss": 0.1857, "step": 14214 }, { "epoch": 10.413919413919414, "grad_norm": 1.1015625, "learning_rate": 8.432348891546118e-05, "loss": 0.1606, "step": 14215 }, { "epoch": 10.414652014652015, "grad_norm": 0.8671875, "learning_rate": 8.432098712771981e-05, "loss": 0.1201, "step": 14216 }, { "epoch": 10.415384615384616, "grad_norm": 1.1171875, "learning_rate": 8.431848517793101e-05, "loss": 0.1476, "step": 14217 }, { "epoch": 10.416117216117216, "grad_norm": 0.98046875, "learning_rate": 8.431598306610674e-05, "loss": 0.1332, "step": 14218 }, { "epoch": 10.416849816849817, "grad_norm": 0.67578125, "learning_rate": 8.431348079225901e-05, "loss": 0.1097, "step": 14219 }, { "epoch": 10.417582417582418, "grad_norm": 0.80859375, "learning_rate": 8.431097835639976e-05, "loss": 0.1335, "step": 14220 }, { "epoch": 10.418315018315019, "grad_norm": 1.4140625, "learning_rate": 8.430847575854104e-05, "loss": 0.1333, "step": 14221 }, { "epoch": 10.41904761904762, "grad_norm": 0.67578125, "learning_rate": 8.43059729986948e-05, "loss": 0.1155, "step": 14222 }, { "epoch": 10.41978021978022, "grad_norm": 1.5078125, "learning_rate": 8.430347007687305e-05, "loss": 0.1775, "step": 14223 }, { "epoch": 10.42051282051282, "grad_norm": 1.3359375, "learning_rate": 8.430096699308778e-05, "loss": 0.1327, "step": 14224 }, { "epoch": 10.421245421245422, "grad_norm": 0.625, "learning_rate": 8.429846374735097e-05, "loss": 0.13, "step": 14225 }, { "epoch": 10.421978021978022, "grad_norm": 0.74609375, "learning_rate": 8.429596033967464e-05, "loss": 0.1014, "step": 14226 }, { "epoch": 10.422710622710623, "grad_norm": 1.1484375, "learning_rate": 8.429345677007078e-05, "loss": 0.1619, "step": 14227 }, { "epoch": 10.423443223443224, "grad_norm": 0.8984375, "learning_rate": 8.429095303855135e-05, "loss": 0.1017, "step": 14228 }, { "epoch": 10.424175824175824, "grad_norm": 0.97265625, "learning_rate": 8.428844914512837e-05, "loss": 0.1027, "step": 14229 }, { "epoch": 10.424908424908425, "grad_norm": 1.2890625, "learning_rate": 8.428594508981386e-05, "loss": 0.1215, "step": 14230 }, { "epoch": 10.425641025641026, "grad_norm": 1.5859375, "learning_rate": 8.428344087261978e-05, "loss": 0.1306, "step": 14231 }, { "epoch": 10.426373626373627, "grad_norm": 0.96875, "learning_rate": 8.428093649355816e-05, "loss": 0.1418, "step": 14232 }, { "epoch": 10.427106227106227, "grad_norm": 1.28125, "learning_rate": 8.427843195264097e-05, "loss": 0.1284, "step": 14233 }, { "epoch": 10.427838827838828, "grad_norm": 0.7890625, "learning_rate": 8.427592724988022e-05, "loss": 0.1363, "step": 14234 }, { "epoch": 10.428571428571429, "grad_norm": 0.859375, "learning_rate": 8.427342238528793e-05, "loss": 0.091, "step": 14235 }, { "epoch": 10.42930402930403, "grad_norm": 1.0859375, "learning_rate": 8.427091735887608e-05, "loss": 0.1318, "step": 14236 }, { "epoch": 10.43003663003663, "grad_norm": 1.015625, "learning_rate": 8.426841217065667e-05, "loss": 0.1641, "step": 14237 }, { "epoch": 10.430769230769231, "grad_norm": 1.0625, "learning_rate": 8.426590682064171e-05, "loss": 0.152, "step": 14238 }, { "epoch": 10.431501831501832, "grad_norm": 1.4921875, "learning_rate": 8.426340130884323e-05, "loss": 0.1089, "step": 14239 }, { "epoch": 10.432234432234432, "grad_norm": 1.7578125, "learning_rate": 8.426089563527319e-05, "loss": 0.1215, "step": 14240 }, { "epoch": 10.432967032967033, "grad_norm": 1.453125, "learning_rate": 8.42583897999436e-05, "loss": 0.1194, "step": 14241 }, { "epoch": 10.433699633699634, "grad_norm": 1.1953125, "learning_rate": 8.425588380286652e-05, "loss": 0.1423, "step": 14242 }, { "epoch": 10.434432234432235, "grad_norm": 0.90234375, "learning_rate": 8.42533776440539e-05, "loss": 0.1438, "step": 14243 }, { "epoch": 10.435164835164835, "grad_norm": 1.6953125, "learning_rate": 8.425087132351776e-05, "loss": 0.1493, "step": 14244 }, { "epoch": 10.435897435897436, "grad_norm": 1.015625, "learning_rate": 8.424836484127015e-05, "loss": 0.145, "step": 14245 }, { "epoch": 10.436630036630037, "grad_norm": 1.53125, "learning_rate": 8.4245858197323e-05, "loss": 0.1544, "step": 14246 }, { "epoch": 10.437362637362638, "grad_norm": 0.90625, "learning_rate": 8.424335139168838e-05, "loss": 0.1232, "step": 14247 }, { "epoch": 10.438095238095238, "grad_norm": 1.0234375, "learning_rate": 8.42408444243783e-05, "loss": 0.1434, "step": 14248 }, { "epoch": 10.438827838827839, "grad_norm": 1.078125, "learning_rate": 8.423833729540474e-05, "loss": 0.1322, "step": 14249 }, { "epoch": 10.43956043956044, "grad_norm": 0.96484375, "learning_rate": 8.423583000477975e-05, "loss": 0.1235, "step": 14250 }, { "epoch": 10.44029304029304, "grad_norm": 1.2578125, "learning_rate": 8.423332255251531e-05, "loss": 0.1603, "step": 14251 }, { "epoch": 10.441025641025641, "grad_norm": 2.125, "learning_rate": 8.423081493862346e-05, "loss": 0.1636, "step": 14252 }, { "epoch": 10.441758241758242, "grad_norm": 1.015625, "learning_rate": 8.422830716311618e-05, "loss": 0.1183, "step": 14253 }, { "epoch": 10.442490842490843, "grad_norm": 0.9609375, "learning_rate": 8.422579922600551e-05, "loss": 0.1393, "step": 14254 }, { "epoch": 10.443223443223443, "grad_norm": 1.0625, "learning_rate": 8.422329112730351e-05, "loss": 0.1609, "step": 14255 }, { "epoch": 10.443956043956044, "grad_norm": 0.7265625, "learning_rate": 8.422078286702211e-05, "loss": 0.1249, "step": 14256 }, { "epoch": 10.444688644688645, "grad_norm": 1.1484375, "learning_rate": 8.421827444517337e-05, "loss": 0.1125, "step": 14257 }, { "epoch": 10.445421245421246, "grad_norm": 1.3828125, "learning_rate": 8.421576586176933e-05, "loss": 0.1511, "step": 14258 }, { "epoch": 10.446153846153846, "grad_norm": 0.69140625, "learning_rate": 8.421325711682196e-05, "loss": 0.1062, "step": 14259 }, { "epoch": 10.446886446886447, "grad_norm": 0.70703125, "learning_rate": 8.421074821034333e-05, "loss": 0.1172, "step": 14260 }, { "epoch": 10.447619047619048, "grad_norm": 0.8125, "learning_rate": 8.420823914234543e-05, "loss": 0.1824, "step": 14261 }, { "epoch": 10.448351648351649, "grad_norm": 1.046875, "learning_rate": 8.420572991284029e-05, "loss": 0.1116, "step": 14262 }, { "epoch": 10.44908424908425, "grad_norm": 0.59375, "learning_rate": 8.420322052183993e-05, "loss": 0.113, "step": 14263 }, { "epoch": 10.44981684981685, "grad_norm": 0.98828125, "learning_rate": 8.420071096935637e-05, "loss": 0.1317, "step": 14264 }, { "epoch": 10.45054945054945, "grad_norm": 0.9453125, "learning_rate": 8.419820125540166e-05, "loss": 0.1241, "step": 14265 }, { "epoch": 10.451282051282051, "grad_norm": 0.8984375, "learning_rate": 8.41956913799878e-05, "loss": 0.1118, "step": 14266 }, { "epoch": 10.452014652014652, "grad_norm": 1.25, "learning_rate": 8.419318134312682e-05, "loss": 0.1098, "step": 14267 }, { "epoch": 10.452747252747253, "grad_norm": 0.9375, "learning_rate": 8.419067114483074e-05, "loss": 0.1333, "step": 14268 }, { "epoch": 10.453479853479854, "grad_norm": 0.84765625, "learning_rate": 8.41881607851116e-05, "loss": 0.1057, "step": 14269 }, { "epoch": 10.454212454212454, "grad_norm": 0.94921875, "learning_rate": 8.418565026398142e-05, "loss": 0.1355, "step": 14270 }, { "epoch": 10.454945054945055, "grad_norm": 1.4296875, "learning_rate": 8.418313958145224e-05, "loss": 0.1221, "step": 14271 }, { "epoch": 10.455677655677656, "grad_norm": 1.2890625, "learning_rate": 8.418062873753607e-05, "loss": 0.1136, "step": 14272 }, { "epoch": 10.456410256410257, "grad_norm": 0.98828125, "learning_rate": 8.417811773224495e-05, "loss": 0.1562, "step": 14273 }, { "epoch": 10.457142857142857, "grad_norm": 0.984375, "learning_rate": 8.417560656559091e-05, "loss": 0.1253, "step": 14274 }, { "epoch": 10.457875457875458, "grad_norm": 0.640625, "learning_rate": 8.4173095237586e-05, "loss": 0.1177, "step": 14275 }, { "epoch": 10.458608058608059, "grad_norm": 1.296875, "learning_rate": 8.417058374824224e-05, "loss": 0.1637, "step": 14276 }, { "epoch": 10.45934065934066, "grad_norm": 1.546875, "learning_rate": 8.416807209757163e-05, "loss": 0.1781, "step": 14277 }, { "epoch": 10.46007326007326, "grad_norm": 1.0546875, "learning_rate": 8.416556028558627e-05, "loss": 0.1081, "step": 14278 }, { "epoch": 10.460805860805861, "grad_norm": 1.0390625, "learning_rate": 8.416304831229813e-05, "loss": 0.1416, "step": 14279 }, { "epoch": 10.461538461538462, "grad_norm": 1.0234375, "learning_rate": 8.416053617771928e-05, "loss": 0.1311, "step": 14280 }, { "epoch": 10.462271062271062, "grad_norm": 1.5078125, "learning_rate": 8.415802388186176e-05, "loss": 0.1148, "step": 14281 }, { "epoch": 10.463003663003663, "grad_norm": 1.40625, "learning_rate": 8.415551142473758e-05, "loss": 0.1255, "step": 14282 }, { "epoch": 10.463736263736264, "grad_norm": 0.90234375, "learning_rate": 8.415299880635883e-05, "loss": 0.1217, "step": 14283 }, { "epoch": 10.464468864468865, "grad_norm": 0.84765625, "learning_rate": 8.41504860267375e-05, "loss": 0.1407, "step": 14284 }, { "epoch": 10.465201465201465, "grad_norm": 0.6953125, "learning_rate": 8.414797308588564e-05, "loss": 0.1328, "step": 14285 }, { "epoch": 10.465934065934066, "grad_norm": 0.9296875, "learning_rate": 8.41454599838153e-05, "loss": 0.1425, "step": 14286 }, { "epoch": 10.466666666666667, "grad_norm": 1.7578125, "learning_rate": 8.41429467205385e-05, "loss": 0.1533, "step": 14287 }, { "epoch": 10.467399267399268, "grad_norm": 0.85546875, "learning_rate": 8.414043329606732e-05, "loss": 0.1013, "step": 14288 }, { "epoch": 10.468131868131868, "grad_norm": 0.8125, "learning_rate": 8.413791971041377e-05, "loss": 0.1192, "step": 14289 }, { "epoch": 10.468864468864469, "grad_norm": 0.96875, "learning_rate": 8.41354059635899e-05, "loss": 0.1439, "step": 14290 }, { "epoch": 10.46959706959707, "grad_norm": 1.0859375, "learning_rate": 8.413289205560777e-05, "loss": 0.1203, "step": 14291 }, { "epoch": 10.47032967032967, "grad_norm": 0.96484375, "learning_rate": 8.41303779864794e-05, "loss": 0.1241, "step": 14292 }, { "epoch": 10.471062271062271, "grad_norm": 0.7890625, "learning_rate": 8.412786375621686e-05, "loss": 0.1461, "step": 14293 }, { "epoch": 10.471794871794872, "grad_norm": 1.3359375, "learning_rate": 8.412534936483217e-05, "loss": 0.1212, "step": 14294 }, { "epoch": 10.472527472527473, "grad_norm": 1.234375, "learning_rate": 8.41228348123374e-05, "loss": 0.1253, "step": 14295 }, { "epoch": 10.473260073260073, "grad_norm": 0.80859375, "learning_rate": 8.41203200987446e-05, "loss": 0.1433, "step": 14296 }, { "epoch": 10.473992673992674, "grad_norm": 0.76953125, "learning_rate": 8.411780522406582e-05, "loss": 0.1122, "step": 14297 }, { "epoch": 10.474725274725275, "grad_norm": 2.296875, "learning_rate": 8.411529018831307e-05, "loss": 0.1512, "step": 14298 }, { "epoch": 10.475457875457876, "grad_norm": 1.109375, "learning_rate": 8.411277499149845e-05, "loss": 0.1406, "step": 14299 }, { "epoch": 10.476190476190476, "grad_norm": 0.81640625, "learning_rate": 8.411025963363397e-05, "loss": 0.108, "step": 14300 }, { "epoch": 10.476923076923077, "grad_norm": 1.0, "learning_rate": 8.410774411473172e-05, "loss": 0.1314, "step": 14301 }, { "epoch": 10.477655677655678, "grad_norm": 1.03125, "learning_rate": 8.410522843480374e-05, "loss": 0.1198, "step": 14302 }, { "epoch": 10.478388278388278, "grad_norm": 1.515625, "learning_rate": 8.410271259386208e-05, "loss": 0.1432, "step": 14303 }, { "epoch": 10.47912087912088, "grad_norm": 1.1171875, "learning_rate": 8.410019659191878e-05, "loss": 0.0886, "step": 14304 }, { "epoch": 10.47985347985348, "grad_norm": 0.78125, "learning_rate": 8.40976804289859e-05, "loss": 0.1256, "step": 14305 }, { "epoch": 10.48058608058608, "grad_norm": 0.91015625, "learning_rate": 8.409516410507552e-05, "loss": 0.1326, "step": 14306 }, { "epoch": 10.481318681318681, "grad_norm": 0.96875, "learning_rate": 8.409264762019966e-05, "loss": 0.1057, "step": 14307 }, { "epoch": 10.482051282051282, "grad_norm": 0.984375, "learning_rate": 8.409013097437042e-05, "loss": 0.1616, "step": 14308 }, { "epoch": 10.482783882783883, "grad_norm": 1.6953125, "learning_rate": 8.408761416759983e-05, "loss": 0.1262, "step": 14309 }, { "epoch": 10.483516483516484, "grad_norm": 1.9375, "learning_rate": 8.408509719989995e-05, "loss": 0.1581, "step": 14310 }, { "epoch": 10.484249084249084, "grad_norm": 1.046875, "learning_rate": 8.408258007128284e-05, "loss": 0.1328, "step": 14311 }, { "epoch": 10.484981684981685, "grad_norm": 1.1796875, "learning_rate": 8.408006278176057e-05, "loss": 0.1239, "step": 14312 }, { "epoch": 10.485714285714286, "grad_norm": 1.1796875, "learning_rate": 8.40775453313452e-05, "loss": 0.1207, "step": 14313 }, { "epoch": 10.486446886446886, "grad_norm": 1.5, "learning_rate": 8.407502772004879e-05, "loss": 0.1538, "step": 14314 }, { "epoch": 10.487179487179487, "grad_norm": 1.2890625, "learning_rate": 8.407250994788339e-05, "loss": 0.1597, "step": 14315 }, { "epoch": 10.487912087912088, "grad_norm": 1.390625, "learning_rate": 8.406999201486108e-05, "loss": 0.1512, "step": 14316 }, { "epoch": 10.488644688644689, "grad_norm": 1.5859375, "learning_rate": 8.406747392099391e-05, "loss": 0.1219, "step": 14317 }, { "epoch": 10.48937728937729, "grad_norm": 1.2109375, "learning_rate": 8.406495566629397e-05, "loss": 0.1429, "step": 14318 }, { "epoch": 10.49010989010989, "grad_norm": 1.046875, "learning_rate": 8.406243725077329e-05, "loss": 0.1353, "step": 14319 }, { "epoch": 10.49084249084249, "grad_norm": 1.0859375, "learning_rate": 8.405991867444397e-05, "loss": 0.129, "step": 14320 }, { "epoch": 10.491575091575092, "grad_norm": 1.203125, "learning_rate": 8.405739993731804e-05, "loss": 0.1199, "step": 14321 }, { "epoch": 10.492307692307692, "grad_norm": 1.390625, "learning_rate": 8.405488103940761e-05, "loss": 0.1377, "step": 14322 }, { "epoch": 10.493040293040293, "grad_norm": 1.8828125, "learning_rate": 8.405236198072472e-05, "loss": 0.1716, "step": 14323 }, { "epoch": 10.493772893772894, "grad_norm": 1.1015625, "learning_rate": 8.404984276128146e-05, "loss": 0.1242, "step": 14324 }, { "epoch": 10.494505494505495, "grad_norm": 1.453125, "learning_rate": 8.40473233810899e-05, "loss": 0.1729, "step": 14325 }, { "epoch": 10.495238095238095, "grad_norm": 1.328125, "learning_rate": 8.404480384016207e-05, "loss": 0.1231, "step": 14326 }, { "epoch": 10.495970695970696, "grad_norm": 1.0, "learning_rate": 8.40422841385101e-05, "loss": 0.1371, "step": 14327 }, { "epoch": 10.496703296703297, "grad_norm": 1.703125, "learning_rate": 8.403976427614603e-05, "loss": 0.1431, "step": 14328 }, { "epoch": 10.497435897435897, "grad_norm": 1.1171875, "learning_rate": 8.403724425308192e-05, "loss": 0.1594, "step": 14329 }, { "epoch": 10.498168498168498, "grad_norm": 1.15625, "learning_rate": 8.403472406932987e-05, "loss": 0.1101, "step": 14330 }, { "epoch": 10.498901098901099, "grad_norm": 1.0078125, "learning_rate": 8.403220372490196e-05, "loss": 0.1127, "step": 14331 }, { "epoch": 10.4996336996337, "grad_norm": 0.84375, "learning_rate": 8.402968321981024e-05, "loss": 0.1303, "step": 14332 }, { "epoch": 10.5003663003663, "grad_norm": 1.125, "learning_rate": 8.402716255406682e-05, "loss": 0.1408, "step": 14333 }, { "epoch": 10.501098901098901, "grad_norm": 0.921875, "learning_rate": 8.402464172768374e-05, "loss": 0.1255, "step": 14334 }, { "epoch": 10.501831501831502, "grad_norm": 0.7578125, "learning_rate": 8.402212074067309e-05, "loss": 0.1307, "step": 14335 }, { "epoch": 10.502564102564103, "grad_norm": 1.0859375, "learning_rate": 8.401959959304699e-05, "loss": 0.113, "step": 14336 }, { "epoch": 10.503296703296703, "grad_norm": 0.76171875, "learning_rate": 8.401707828481744e-05, "loss": 0.1185, "step": 14337 }, { "epoch": 10.504029304029304, "grad_norm": 1.03125, "learning_rate": 8.40145568159966e-05, "loss": 0.1417, "step": 14338 }, { "epoch": 10.504761904761905, "grad_norm": 0.96484375, "learning_rate": 8.40120351865965e-05, "loss": 0.1628, "step": 14339 }, { "epoch": 10.505494505494505, "grad_norm": 0.9296875, "learning_rate": 8.400951339662923e-05, "loss": 0.1123, "step": 14340 }, { "epoch": 10.506227106227106, "grad_norm": 1.203125, "learning_rate": 8.40069914461069e-05, "loss": 0.1448, "step": 14341 }, { "epoch": 10.506959706959707, "grad_norm": 1.3984375, "learning_rate": 8.400446933504157e-05, "loss": 0.1597, "step": 14342 }, { "epoch": 10.507692307692308, "grad_norm": 1.5859375, "learning_rate": 8.400194706344534e-05, "loss": 0.1445, "step": 14343 }, { "epoch": 10.508424908424908, "grad_norm": 1.6953125, "learning_rate": 8.399942463133025e-05, "loss": 0.1323, "step": 14344 }, { "epoch": 10.50915750915751, "grad_norm": 0.79296875, "learning_rate": 8.399690203870844e-05, "loss": 0.1215, "step": 14345 }, { "epoch": 10.50989010989011, "grad_norm": 0.84375, "learning_rate": 8.399437928559197e-05, "loss": 0.1217, "step": 14346 }, { "epoch": 10.51062271062271, "grad_norm": 1.890625, "learning_rate": 8.399185637199295e-05, "loss": 0.1347, "step": 14347 }, { "epoch": 10.511355311355311, "grad_norm": 1.234375, "learning_rate": 8.398933329792344e-05, "loss": 0.1425, "step": 14348 }, { "epoch": 10.512087912087912, "grad_norm": 0.94140625, "learning_rate": 8.398681006339553e-05, "loss": 0.1109, "step": 14349 }, { "epoch": 10.512820512820513, "grad_norm": 1.4375, "learning_rate": 8.398428666842133e-05, "loss": 0.1084, "step": 14350 }, { "epoch": 10.513553113553114, "grad_norm": 0.90234375, "learning_rate": 8.398176311301293e-05, "loss": 0.1433, "step": 14351 }, { "epoch": 10.514285714285714, "grad_norm": 1.1328125, "learning_rate": 8.397923939718239e-05, "loss": 0.0846, "step": 14352 }, { "epoch": 10.515018315018315, "grad_norm": 0.92578125, "learning_rate": 8.397671552094184e-05, "loss": 0.1668, "step": 14353 }, { "epoch": 10.515750915750916, "grad_norm": 0.8203125, "learning_rate": 8.397419148430337e-05, "loss": 0.1033, "step": 14354 }, { "epoch": 10.516483516483516, "grad_norm": 1.046875, "learning_rate": 8.397166728727902e-05, "loss": 0.1368, "step": 14355 }, { "epoch": 10.517216117216117, "grad_norm": 1.25, "learning_rate": 8.396914292988095e-05, "loss": 0.1477, "step": 14356 }, { "epoch": 10.517948717948718, "grad_norm": 1.3203125, "learning_rate": 8.396661841212122e-05, "loss": 0.1284, "step": 14357 }, { "epoch": 10.518681318681319, "grad_norm": 1.40625, "learning_rate": 8.396409373401194e-05, "loss": 0.1467, "step": 14358 }, { "epoch": 10.51941391941392, "grad_norm": 1.03125, "learning_rate": 8.39615688955652e-05, "loss": 0.1438, "step": 14359 }, { "epoch": 10.52014652014652, "grad_norm": 1.2109375, "learning_rate": 8.39590438967931e-05, "loss": 0.1568, "step": 14360 }, { "epoch": 10.52087912087912, "grad_norm": 1.734375, "learning_rate": 8.395651873770774e-05, "loss": 0.1575, "step": 14361 }, { "epoch": 10.521611721611722, "grad_norm": 0.921875, "learning_rate": 8.395399341832121e-05, "loss": 0.1431, "step": 14362 }, { "epoch": 10.522344322344322, "grad_norm": 1.09375, "learning_rate": 8.39514679386456e-05, "loss": 0.1364, "step": 14363 }, { "epoch": 10.523076923076923, "grad_norm": 0.77734375, "learning_rate": 8.394894229869306e-05, "loss": 0.1372, "step": 14364 }, { "epoch": 10.523809523809524, "grad_norm": 1.109375, "learning_rate": 8.394641649847564e-05, "loss": 0.1479, "step": 14365 }, { "epoch": 10.524542124542124, "grad_norm": 1.4921875, "learning_rate": 8.394389053800546e-05, "loss": 0.1319, "step": 14366 }, { "epoch": 10.525274725274725, "grad_norm": 1.171875, "learning_rate": 8.394136441729463e-05, "loss": 0.1163, "step": 14367 }, { "epoch": 10.526007326007326, "grad_norm": 1.421875, "learning_rate": 8.393883813635522e-05, "loss": 0.1305, "step": 14368 }, { "epoch": 10.526739926739927, "grad_norm": 0.80078125, "learning_rate": 8.393631169519937e-05, "loss": 0.1355, "step": 14369 }, { "epoch": 10.527472527472527, "grad_norm": 1.09375, "learning_rate": 8.393378509383918e-05, "loss": 0.1814, "step": 14370 }, { "epoch": 10.528205128205128, "grad_norm": 1.03125, "learning_rate": 8.393125833228676e-05, "loss": 0.1302, "step": 14371 }, { "epoch": 10.528937728937729, "grad_norm": 0.93359375, "learning_rate": 8.392873141055418e-05, "loss": 0.116, "step": 14372 }, { "epoch": 10.52967032967033, "grad_norm": 1.1015625, "learning_rate": 8.392620432865359e-05, "loss": 0.1098, "step": 14373 }, { "epoch": 10.53040293040293, "grad_norm": 1.0703125, "learning_rate": 8.392367708659709e-05, "loss": 0.1445, "step": 14374 }, { "epoch": 10.531135531135531, "grad_norm": 1.2734375, "learning_rate": 8.392114968439677e-05, "loss": 0.1488, "step": 14375 }, { "epoch": 10.531868131868132, "grad_norm": 1.2265625, "learning_rate": 8.391862212206475e-05, "loss": 0.1742, "step": 14376 }, { "epoch": 10.532600732600732, "grad_norm": 1.0390625, "learning_rate": 8.391609439961314e-05, "loss": 0.1256, "step": 14377 }, { "epoch": 10.533333333333333, "grad_norm": 1.046875, "learning_rate": 8.391356651705406e-05, "loss": 0.1338, "step": 14378 }, { "epoch": 10.534065934065934, "grad_norm": 1.9765625, "learning_rate": 8.39110384743996e-05, "loss": 0.139, "step": 14379 }, { "epoch": 10.534798534798535, "grad_norm": 1.3984375, "learning_rate": 8.39085102716619e-05, "loss": 0.1112, "step": 14380 }, { "epoch": 10.535531135531135, "grad_norm": 1.453125, "learning_rate": 8.390598190885306e-05, "loss": 0.1472, "step": 14381 }, { "epoch": 10.536263736263736, "grad_norm": 0.9609375, "learning_rate": 8.390345338598518e-05, "loss": 0.1511, "step": 14382 }, { "epoch": 10.536996336996337, "grad_norm": 1.5390625, "learning_rate": 8.390092470307039e-05, "loss": 0.13, "step": 14383 }, { "epoch": 10.537728937728938, "grad_norm": 0.8046875, "learning_rate": 8.389839586012082e-05, "loss": 0.119, "step": 14384 }, { "epoch": 10.538461538461538, "grad_norm": 0.86328125, "learning_rate": 8.389586685714857e-05, "loss": 0.157, "step": 14385 }, { "epoch": 10.539194139194139, "grad_norm": 1.109375, "learning_rate": 8.389333769416575e-05, "loss": 0.1577, "step": 14386 }, { "epoch": 10.53992673992674, "grad_norm": 0.9296875, "learning_rate": 8.38908083711845e-05, "loss": 0.1341, "step": 14387 }, { "epoch": 10.54065934065934, "grad_norm": 1.2578125, "learning_rate": 8.38882788882169e-05, "loss": 0.1494, "step": 14388 }, { "epoch": 10.541391941391941, "grad_norm": 0.921875, "learning_rate": 8.388574924527512e-05, "loss": 0.1313, "step": 14389 }, { "epoch": 10.542124542124542, "grad_norm": 1.15625, "learning_rate": 8.388321944237124e-05, "loss": 0.1336, "step": 14390 }, { "epoch": 10.542857142857143, "grad_norm": 1.0, "learning_rate": 8.388068947951741e-05, "loss": 0.1138, "step": 14391 }, { "epoch": 10.543589743589743, "grad_norm": 1.203125, "learning_rate": 8.387815935672573e-05, "loss": 0.1216, "step": 14392 }, { "epoch": 10.544322344322344, "grad_norm": 1.34375, "learning_rate": 8.387562907400833e-05, "loss": 0.1202, "step": 14393 }, { "epoch": 10.545054945054945, "grad_norm": 0.6171875, "learning_rate": 8.387309863137733e-05, "loss": 0.1424, "step": 14394 }, { "epoch": 10.545787545787546, "grad_norm": 1.40625, "learning_rate": 8.387056802884488e-05, "loss": 0.1341, "step": 14395 }, { "epoch": 10.546520146520146, "grad_norm": 2.078125, "learning_rate": 8.386803726642308e-05, "loss": 0.1551, "step": 14396 }, { "epoch": 10.547252747252747, "grad_norm": 0.734375, "learning_rate": 8.386550634412403e-05, "loss": 0.1411, "step": 14397 }, { "epoch": 10.547985347985348, "grad_norm": 1.1015625, "learning_rate": 8.386297526195993e-05, "loss": 0.1407, "step": 14398 }, { "epoch": 10.548717948717949, "grad_norm": 1.359375, "learning_rate": 8.386044401994282e-05, "loss": 0.1639, "step": 14399 }, { "epoch": 10.54945054945055, "grad_norm": 0.95703125, "learning_rate": 8.38579126180849e-05, "loss": 0.1124, "step": 14400 }, { "epoch": 10.55018315018315, "grad_norm": 0.8203125, "learning_rate": 8.385538105639824e-05, "loss": 0.1279, "step": 14401 }, { "epoch": 10.55091575091575, "grad_norm": 0.80078125, "learning_rate": 8.385284933489504e-05, "loss": 0.1023, "step": 14402 }, { "epoch": 10.551648351648351, "grad_norm": 0.91015625, "learning_rate": 8.385031745358736e-05, "loss": 0.1266, "step": 14403 }, { "epoch": 10.552380952380952, "grad_norm": 1.8125, "learning_rate": 8.384778541248737e-05, "loss": 0.1414, "step": 14404 }, { "epoch": 10.553113553113553, "grad_norm": 1.9609375, "learning_rate": 8.38452532116072e-05, "loss": 0.1566, "step": 14405 }, { "epoch": 10.553846153846154, "grad_norm": 1.5625, "learning_rate": 8.384272085095896e-05, "loss": 0.1672, "step": 14406 }, { "epoch": 10.554578754578754, "grad_norm": 1.1171875, "learning_rate": 8.38401883305548e-05, "loss": 0.1107, "step": 14407 }, { "epoch": 10.555311355311355, "grad_norm": 1.3359375, "learning_rate": 8.383765565040687e-05, "loss": 0.128, "step": 14408 }, { "epoch": 10.556043956043956, "grad_norm": 1.3125, "learning_rate": 8.383512281052728e-05, "loss": 0.1366, "step": 14409 }, { "epoch": 10.556776556776557, "grad_norm": 0.9765625, "learning_rate": 8.383258981092817e-05, "loss": 0.1465, "step": 14410 }, { "epoch": 10.557509157509157, "grad_norm": 1.6171875, "learning_rate": 8.383005665162167e-05, "loss": 0.1532, "step": 14411 }, { "epoch": 10.558241758241758, "grad_norm": 1.203125, "learning_rate": 8.382752333261994e-05, "loss": 0.1201, "step": 14412 }, { "epoch": 10.558974358974359, "grad_norm": 0.84765625, "learning_rate": 8.382498985393511e-05, "loss": 0.1569, "step": 14413 }, { "epoch": 10.55970695970696, "grad_norm": 1.453125, "learning_rate": 8.382245621557931e-05, "loss": 0.1154, "step": 14414 }, { "epoch": 10.56043956043956, "grad_norm": 1.0703125, "learning_rate": 8.381992241756467e-05, "loss": 0.1086, "step": 14415 }, { "epoch": 10.561172161172161, "grad_norm": 1.2890625, "learning_rate": 8.381738845990335e-05, "loss": 0.1469, "step": 14416 }, { "epoch": 10.561904761904762, "grad_norm": 1.390625, "learning_rate": 8.38148543426075e-05, "loss": 0.1334, "step": 14417 }, { "epoch": 10.562637362637362, "grad_norm": 1.109375, "learning_rate": 8.381232006568922e-05, "loss": 0.1195, "step": 14418 }, { "epoch": 10.563369963369963, "grad_norm": 0.703125, "learning_rate": 8.380978562916071e-05, "loss": 0.1152, "step": 14419 }, { "epoch": 10.564102564102564, "grad_norm": 1.09375, "learning_rate": 8.380725103303405e-05, "loss": 0.1246, "step": 14420 }, { "epoch": 10.564835164835165, "grad_norm": 1.1171875, "learning_rate": 8.380471627732144e-05, "loss": 0.1145, "step": 14421 }, { "epoch": 10.565567765567765, "grad_norm": 2.078125, "learning_rate": 8.380218136203498e-05, "loss": 0.1599, "step": 14422 }, { "epoch": 10.566300366300366, "grad_norm": 1.6640625, "learning_rate": 8.379964628718686e-05, "loss": 0.1249, "step": 14423 }, { "epoch": 10.567032967032967, "grad_norm": 1.0546875, "learning_rate": 8.37971110527892e-05, "loss": 0.1315, "step": 14424 }, { "epoch": 10.567765567765568, "grad_norm": 1.0, "learning_rate": 8.379457565885413e-05, "loss": 0.1047, "step": 14425 }, { "epoch": 10.568498168498168, "grad_norm": 1.1796875, "learning_rate": 8.379204010539385e-05, "loss": 0.097, "step": 14426 }, { "epoch": 10.569230769230769, "grad_norm": 1.140625, "learning_rate": 8.378950439242044e-05, "loss": 0.1364, "step": 14427 }, { "epoch": 10.56996336996337, "grad_norm": 0.8046875, "learning_rate": 8.37869685199461e-05, "loss": 0.1092, "step": 14428 }, { "epoch": 10.57069597069597, "grad_norm": 0.93359375, "learning_rate": 8.378443248798298e-05, "loss": 0.1368, "step": 14429 }, { "epoch": 10.571428571428571, "grad_norm": 1.171875, "learning_rate": 8.37818962965432e-05, "loss": 0.1788, "step": 14430 }, { "epoch": 10.572161172161172, "grad_norm": 1.0390625, "learning_rate": 8.377935994563894e-05, "loss": 0.132, "step": 14431 }, { "epoch": 10.572893772893773, "grad_norm": 1.03125, "learning_rate": 8.377682343528234e-05, "loss": 0.1131, "step": 14432 }, { "epoch": 10.573626373626373, "grad_norm": 1.1796875, "learning_rate": 8.377428676548555e-05, "loss": 0.1312, "step": 14433 }, { "epoch": 10.574358974358974, "grad_norm": 1.078125, "learning_rate": 8.377174993626072e-05, "loss": 0.1161, "step": 14434 }, { "epoch": 10.575091575091575, "grad_norm": 1.078125, "learning_rate": 8.376921294762003e-05, "loss": 0.1379, "step": 14435 }, { "epoch": 10.575824175824176, "grad_norm": 1.046875, "learning_rate": 8.376667579957563e-05, "loss": 0.1404, "step": 14436 }, { "epoch": 10.576556776556776, "grad_norm": 1.375, "learning_rate": 8.376413849213964e-05, "loss": 0.1379, "step": 14437 }, { "epoch": 10.577289377289377, "grad_norm": 1.09375, "learning_rate": 8.376160102532425e-05, "loss": 0.1257, "step": 14438 }, { "epoch": 10.578021978021978, "grad_norm": 0.92578125, "learning_rate": 8.375906339914162e-05, "loss": 0.1266, "step": 14439 }, { "epoch": 10.578754578754578, "grad_norm": 1.0859375, "learning_rate": 8.37565256136039e-05, "loss": 0.135, "step": 14440 }, { "epoch": 10.57948717948718, "grad_norm": 1.8203125, "learning_rate": 8.375398766872322e-05, "loss": 0.1296, "step": 14441 }, { "epoch": 10.58021978021978, "grad_norm": 1.609375, "learning_rate": 8.375144956451179e-05, "loss": 0.1164, "step": 14442 }, { "epoch": 10.58095238095238, "grad_norm": 0.92578125, "learning_rate": 8.374891130098176e-05, "loss": 0.1342, "step": 14443 }, { "epoch": 10.581684981684981, "grad_norm": 0.98046875, "learning_rate": 8.374637287814527e-05, "loss": 0.1104, "step": 14444 }, { "epoch": 10.582417582417582, "grad_norm": 1.6875, "learning_rate": 8.374383429601449e-05, "loss": 0.1569, "step": 14445 }, { "epoch": 10.583150183150183, "grad_norm": 1.7890625, "learning_rate": 8.37412955546016e-05, "loss": 0.1878, "step": 14446 }, { "epoch": 10.583882783882784, "grad_norm": 1.0859375, "learning_rate": 8.373875665391872e-05, "loss": 0.1222, "step": 14447 }, { "epoch": 10.584615384615384, "grad_norm": 1.015625, "learning_rate": 8.373621759397807e-05, "loss": 0.1662, "step": 14448 }, { "epoch": 10.585347985347985, "grad_norm": 1.09375, "learning_rate": 8.373367837479179e-05, "loss": 0.176, "step": 14449 }, { "epoch": 10.586080586080586, "grad_norm": 1.234375, "learning_rate": 8.373113899637204e-05, "loss": 0.1535, "step": 14450 }, { "epoch": 10.586813186813187, "grad_norm": 0.71875, "learning_rate": 8.3728599458731e-05, "loss": 0.1056, "step": 14451 }, { "epoch": 10.587545787545787, "grad_norm": 1.234375, "learning_rate": 8.372605976188082e-05, "loss": 0.1234, "step": 14452 }, { "epoch": 10.588278388278388, "grad_norm": 1.2578125, "learning_rate": 8.372351990583369e-05, "loss": 0.1378, "step": 14453 }, { "epoch": 10.589010989010989, "grad_norm": 1.3359375, "learning_rate": 8.372097989060177e-05, "loss": 0.1383, "step": 14454 }, { "epoch": 10.58974358974359, "grad_norm": 1.0625, "learning_rate": 8.371843971619724e-05, "loss": 0.1224, "step": 14455 }, { "epoch": 10.59047619047619, "grad_norm": 1.0234375, "learning_rate": 8.371589938263225e-05, "loss": 0.1176, "step": 14456 }, { "epoch": 10.59120879120879, "grad_norm": 1.3125, "learning_rate": 8.371335888991897e-05, "loss": 0.1353, "step": 14457 }, { "epoch": 10.591941391941392, "grad_norm": 1.0, "learning_rate": 8.37108182380696e-05, "loss": 0.1055, "step": 14458 }, { "epoch": 10.592673992673992, "grad_norm": 0.69921875, "learning_rate": 8.370827742709629e-05, "loss": 0.1369, "step": 14459 }, { "epoch": 10.593406593406593, "grad_norm": 0.765625, "learning_rate": 8.370573645701123e-05, "loss": 0.1245, "step": 14460 }, { "epoch": 10.594139194139194, "grad_norm": 0.87890625, "learning_rate": 8.370319532782658e-05, "loss": 0.13, "step": 14461 }, { "epoch": 10.594871794871795, "grad_norm": 0.94140625, "learning_rate": 8.370065403955452e-05, "loss": 0.1479, "step": 14462 }, { "epoch": 10.595604395604395, "grad_norm": 1.078125, "learning_rate": 8.369811259220723e-05, "loss": 0.1517, "step": 14463 }, { "epoch": 10.596336996336996, "grad_norm": 1.125, "learning_rate": 8.369557098579688e-05, "loss": 0.1431, "step": 14464 }, { "epoch": 10.597069597069597, "grad_norm": 1.078125, "learning_rate": 8.369302922033565e-05, "loss": 0.137, "step": 14465 }, { "epoch": 10.597802197802197, "grad_norm": 1.015625, "learning_rate": 8.369048729583573e-05, "loss": 0.1271, "step": 14466 }, { "epoch": 10.598534798534798, "grad_norm": 0.9375, "learning_rate": 8.368794521230929e-05, "loss": 0.1277, "step": 14467 }, { "epoch": 10.599267399267399, "grad_norm": 0.984375, "learning_rate": 8.36854029697685e-05, "loss": 0.1163, "step": 14468 }, { "epoch": 10.6, "grad_norm": 1.0546875, "learning_rate": 8.368286056822556e-05, "loss": 0.1478, "step": 14469 }, { "epoch": 10.6007326007326, "grad_norm": 1.109375, "learning_rate": 8.368031800769263e-05, "loss": 0.1251, "step": 14470 }, { "epoch": 10.601465201465201, "grad_norm": 1.1953125, "learning_rate": 8.36777752881819e-05, "loss": 0.1572, "step": 14471 }, { "epoch": 10.602197802197802, "grad_norm": 0.9765625, "learning_rate": 8.367523240970558e-05, "loss": 0.1348, "step": 14472 }, { "epoch": 10.602930402930403, "grad_norm": 1.1953125, "learning_rate": 8.367268937227581e-05, "loss": 0.1296, "step": 14473 }, { "epoch": 10.603663003663003, "grad_norm": 0.76171875, "learning_rate": 8.367014617590481e-05, "loss": 0.1384, "step": 14474 }, { "epoch": 10.604395604395604, "grad_norm": 0.6328125, "learning_rate": 8.366760282060472e-05, "loss": 0.1214, "step": 14475 }, { "epoch": 10.605128205128205, "grad_norm": 0.8046875, "learning_rate": 8.366505930638781e-05, "loss": 0.1357, "step": 14476 }, { "epoch": 10.605860805860805, "grad_norm": 1.0625, "learning_rate": 8.366251563326617e-05, "loss": 0.1319, "step": 14477 }, { "epoch": 10.606593406593406, "grad_norm": 1.2578125, "learning_rate": 8.365997180125205e-05, "loss": 0.1517, "step": 14478 }, { "epoch": 10.607326007326007, "grad_norm": 0.890625, "learning_rate": 8.365742781035762e-05, "loss": 0.1524, "step": 14479 }, { "epoch": 10.608058608058608, "grad_norm": 1.03125, "learning_rate": 8.365488366059505e-05, "loss": 0.1244, "step": 14480 }, { "epoch": 10.608791208791208, "grad_norm": 1.109375, "learning_rate": 8.365233935197658e-05, "loss": 0.1685, "step": 14481 }, { "epoch": 10.60952380952381, "grad_norm": 1.09375, "learning_rate": 8.364979488451436e-05, "loss": 0.1518, "step": 14482 }, { "epoch": 10.61025641025641, "grad_norm": 0.72265625, "learning_rate": 8.364725025822057e-05, "loss": 0.1278, "step": 14483 }, { "epoch": 10.61098901098901, "grad_norm": 0.96875, "learning_rate": 8.364470547310742e-05, "loss": 0.1212, "step": 14484 }, { "epoch": 10.611721611721611, "grad_norm": 0.69140625, "learning_rate": 8.364216052918713e-05, "loss": 0.1218, "step": 14485 }, { "epoch": 10.612454212454212, "grad_norm": 0.8671875, "learning_rate": 8.363961542647185e-05, "loss": 0.1015, "step": 14486 }, { "epoch": 10.613186813186813, "grad_norm": 1.078125, "learning_rate": 8.36370701649738e-05, "loss": 0.1349, "step": 14487 }, { "epoch": 10.613919413919414, "grad_norm": 1.328125, "learning_rate": 8.363452474470517e-05, "loss": 0.1322, "step": 14488 }, { "epoch": 10.614652014652014, "grad_norm": 1.2421875, "learning_rate": 8.363197916567817e-05, "loss": 0.1266, "step": 14489 }, { "epoch": 10.615384615384615, "grad_norm": 1.640625, "learning_rate": 8.362943342790494e-05, "loss": 0.1689, "step": 14490 }, { "epoch": 10.616117216117216, "grad_norm": 0.91796875, "learning_rate": 8.362688753139777e-05, "loss": 0.1096, "step": 14491 }, { "epoch": 10.616849816849816, "grad_norm": 0.78125, "learning_rate": 8.362434147616876e-05, "loss": 0.1172, "step": 14492 }, { "epoch": 10.617582417582417, "grad_norm": 1.4296875, "learning_rate": 8.362179526223019e-05, "loss": 0.1471, "step": 14493 }, { "epoch": 10.618315018315018, "grad_norm": 0.859375, "learning_rate": 8.361924888959421e-05, "loss": 0.1423, "step": 14494 }, { "epoch": 10.619047619047619, "grad_norm": 1.09375, "learning_rate": 8.361670235827304e-05, "loss": 0.1159, "step": 14495 }, { "epoch": 10.61978021978022, "grad_norm": 0.87109375, "learning_rate": 8.361415566827889e-05, "loss": 0.1097, "step": 14496 }, { "epoch": 10.62051282051282, "grad_norm": 0.83203125, "learning_rate": 8.361160881962393e-05, "loss": 0.1012, "step": 14497 }, { "epoch": 10.62124542124542, "grad_norm": 1.046875, "learning_rate": 8.360906181232039e-05, "loss": 0.1091, "step": 14498 }, { "epoch": 10.621978021978022, "grad_norm": 0.95703125, "learning_rate": 8.360651464638047e-05, "loss": 0.1407, "step": 14499 }, { "epoch": 10.622710622710622, "grad_norm": 1.40625, "learning_rate": 8.360396732181636e-05, "loss": 0.115, "step": 14500 }, { "epoch": 10.623443223443223, "grad_norm": 1.375, "learning_rate": 8.360141983864027e-05, "loss": 0.1644, "step": 14501 }, { "epoch": 10.624175824175824, "grad_norm": 1.359375, "learning_rate": 8.359887219686444e-05, "loss": 0.1235, "step": 14502 }, { "epoch": 10.624908424908424, "grad_norm": 1.5546875, "learning_rate": 8.359632439650103e-05, "loss": 0.145, "step": 14503 }, { "epoch": 10.625641025641025, "grad_norm": 1.3359375, "learning_rate": 8.359377643756227e-05, "loss": 0.1235, "step": 14504 }, { "epoch": 10.626373626373626, "grad_norm": 1.921875, "learning_rate": 8.359122832006035e-05, "loss": 0.1316, "step": 14505 }, { "epoch": 10.627106227106227, "grad_norm": 1.140625, "learning_rate": 8.358868004400749e-05, "loss": 0.1175, "step": 14506 }, { "epoch": 10.627838827838827, "grad_norm": 2.734375, "learning_rate": 8.358613160941592e-05, "loss": 0.1473, "step": 14507 }, { "epoch": 10.628571428571428, "grad_norm": 1.5234375, "learning_rate": 8.358358301629781e-05, "loss": 0.143, "step": 14508 }, { "epoch": 10.629304029304029, "grad_norm": 2.03125, "learning_rate": 8.358103426466542e-05, "loss": 0.2117, "step": 14509 }, { "epoch": 10.63003663003663, "grad_norm": 0.83203125, "learning_rate": 8.357848535453092e-05, "loss": 0.1236, "step": 14510 }, { "epoch": 10.63076923076923, "grad_norm": 1.015625, "learning_rate": 8.357593628590652e-05, "loss": 0.1307, "step": 14511 }, { "epoch": 10.631501831501831, "grad_norm": 1.1796875, "learning_rate": 8.357338705880449e-05, "loss": 0.1191, "step": 14512 }, { "epoch": 10.632234432234432, "grad_norm": 0.75390625, "learning_rate": 8.3570837673237e-05, "loss": 0.1428, "step": 14513 }, { "epoch": 10.632967032967032, "grad_norm": 1.296875, "learning_rate": 8.356828812921623e-05, "loss": 0.1215, "step": 14514 }, { "epoch": 10.633699633699633, "grad_norm": 1.015625, "learning_rate": 8.356573842675447e-05, "loss": 0.1306, "step": 14515 }, { "epoch": 10.634432234432234, "grad_norm": 1.0703125, "learning_rate": 8.356318856586388e-05, "loss": 0.1329, "step": 14516 }, { "epoch": 10.635164835164835, "grad_norm": 1.2265625, "learning_rate": 8.356063854655671e-05, "loss": 0.1405, "step": 14517 }, { "epoch": 10.635897435897435, "grad_norm": 0.8203125, "learning_rate": 8.355808836884519e-05, "loss": 0.1248, "step": 14518 }, { "epoch": 10.636630036630036, "grad_norm": 1.1328125, "learning_rate": 8.355553803274148e-05, "loss": 0.1225, "step": 14519 }, { "epoch": 10.637362637362637, "grad_norm": 1.2265625, "learning_rate": 8.355298753825786e-05, "loss": 0.1297, "step": 14520 }, { "epoch": 10.638095238095238, "grad_norm": 1.015625, "learning_rate": 8.355043688540653e-05, "loss": 0.1383, "step": 14521 }, { "epoch": 10.638827838827838, "grad_norm": 0.85546875, "learning_rate": 8.35478860741997e-05, "loss": 0.1508, "step": 14522 }, { "epoch": 10.639560439560439, "grad_norm": 2.15625, "learning_rate": 8.35453351046496e-05, "loss": 0.1499, "step": 14523 }, { "epoch": 10.64029304029304, "grad_norm": 1.3984375, "learning_rate": 8.354278397676843e-05, "loss": 0.1451, "step": 14524 }, { "epoch": 10.64102564102564, "grad_norm": 1.6640625, "learning_rate": 8.354023269056845e-05, "loss": 0.1278, "step": 14525 }, { "epoch": 10.641758241758241, "grad_norm": 1.1171875, "learning_rate": 8.353768124606188e-05, "loss": 0.1186, "step": 14526 }, { "epoch": 10.642490842490842, "grad_norm": 0.7421875, "learning_rate": 8.353512964326092e-05, "loss": 0.1052, "step": 14527 }, { "epoch": 10.643223443223443, "grad_norm": 1.125, "learning_rate": 8.353257788217782e-05, "loss": 0.1398, "step": 14528 }, { "epoch": 10.643956043956043, "grad_norm": 1.8671875, "learning_rate": 8.353002596282478e-05, "loss": 0.1364, "step": 14529 }, { "epoch": 10.644688644688644, "grad_norm": 1.8125, "learning_rate": 8.352747388521406e-05, "loss": 0.1574, "step": 14530 }, { "epoch": 10.645421245421245, "grad_norm": 1.4609375, "learning_rate": 8.352492164935786e-05, "loss": 0.1346, "step": 14531 }, { "epoch": 10.646153846153846, "grad_norm": 1.328125, "learning_rate": 8.352236925526842e-05, "loss": 0.1529, "step": 14532 }, { "epoch": 10.646886446886446, "grad_norm": 0.83203125, "learning_rate": 8.351981670295797e-05, "loss": 0.1329, "step": 14533 }, { "epoch": 10.647619047619047, "grad_norm": 0.8046875, "learning_rate": 8.351726399243873e-05, "loss": 0.1451, "step": 14534 }, { "epoch": 10.648351648351648, "grad_norm": 1.4921875, "learning_rate": 8.351471112372295e-05, "loss": 0.119, "step": 14535 }, { "epoch": 10.649084249084249, "grad_norm": 0.98046875, "learning_rate": 8.351215809682285e-05, "loss": 0.1439, "step": 14536 }, { "epoch": 10.64981684981685, "grad_norm": 0.9921875, "learning_rate": 8.350960491175064e-05, "loss": 0.1681, "step": 14537 }, { "epoch": 10.65054945054945, "grad_norm": 1.453125, "learning_rate": 8.350705156851862e-05, "loss": 0.0993, "step": 14538 }, { "epoch": 10.65128205128205, "grad_norm": 1.5390625, "learning_rate": 8.350449806713896e-05, "loss": 0.1551, "step": 14539 }, { "epoch": 10.652014652014651, "grad_norm": 1.1796875, "learning_rate": 8.35019444076239e-05, "loss": 0.1176, "step": 14540 }, { "epoch": 10.652747252747252, "grad_norm": 1.078125, "learning_rate": 8.349939058998571e-05, "loss": 0.0835, "step": 14541 }, { "epoch": 10.653479853479853, "grad_norm": 1.4609375, "learning_rate": 8.349683661423659e-05, "loss": 0.1398, "step": 14542 }, { "epoch": 10.654212454212454, "grad_norm": 1.765625, "learning_rate": 8.349428248038881e-05, "loss": 0.1521, "step": 14543 }, { "epoch": 10.654945054945054, "grad_norm": 1.484375, "learning_rate": 8.34917281884546e-05, "loss": 0.1194, "step": 14544 }, { "epoch": 10.655677655677655, "grad_norm": 1.0078125, "learning_rate": 8.348917373844616e-05, "loss": 0.1186, "step": 14545 }, { "epoch": 10.656410256410256, "grad_norm": 1.2421875, "learning_rate": 8.348661913037577e-05, "loss": 0.1017, "step": 14546 }, { "epoch": 10.657142857142857, "grad_norm": 0.67578125, "learning_rate": 8.348406436425568e-05, "loss": 0.1537, "step": 14547 }, { "epoch": 10.657875457875457, "grad_norm": 1.4609375, "learning_rate": 8.348150944009808e-05, "loss": 0.1359, "step": 14548 }, { "epoch": 10.658608058608058, "grad_norm": 1.1484375, "learning_rate": 8.347895435791526e-05, "loss": 0.1722, "step": 14549 }, { "epoch": 10.659340659340659, "grad_norm": 1.5546875, "learning_rate": 8.347639911771944e-05, "loss": 0.1404, "step": 14550 }, { "epoch": 10.66007326007326, "grad_norm": 0.79296875, "learning_rate": 8.347384371952285e-05, "loss": 0.1256, "step": 14551 }, { "epoch": 10.66080586080586, "grad_norm": 1.109375, "learning_rate": 8.347128816333778e-05, "loss": 0.125, "step": 14552 }, { "epoch": 10.661538461538461, "grad_norm": 1.4609375, "learning_rate": 8.346873244917642e-05, "loss": 0.1564, "step": 14553 }, { "epoch": 10.662271062271062, "grad_norm": 1.1328125, "learning_rate": 8.346617657705105e-05, "loss": 0.1143, "step": 14554 }, { "epoch": 10.663003663003662, "grad_norm": 1.3671875, "learning_rate": 8.346362054697391e-05, "loss": 0.1203, "step": 14555 }, { "epoch": 10.663736263736263, "grad_norm": 0.89453125, "learning_rate": 8.346106435895723e-05, "loss": 0.1246, "step": 14556 }, { "epoch": 10.664468864468864, "grad_norm": 1.0859375, "learning_rate": 8.345850801301327e-05, "loss": 0.1185, "step": 14557 }, { "epoch": 10.665201465201465, "grad_norm": 0.62109375, "learning_rate": 8.34559515091543e-05, "loss": 0.1127, "step": 14558 }, { "epoch": 10.665934065934065, "grad_norm": 1.0703125, "learning_rate": 8.345339484739254e-05, "loss": 0.1273, "step": 14559 }, { "epoch": 10.666666666666666, "grad_norm": 1.15625, "learning_rate": 8.345083802774024e-05, "loss": 0.1036, "step": 14560 }, { "epoch": 10.667399267399267, "grad_norm": 1.703125, "learning_rate": 8.344828105020965e-05, "loss": 0.1641, "step": 14561 }, { "epoch": 10.668131868131868, "grad_norm": 1.1796875, "learning_rate": 8.344572391481303e-05, "loss": 0.12, "step": 14562 }, { "epoch": 10.668864468864468, "grad_norm": 0.91015625, "learning_rate": 8.344316662156264e-05, "loss": 0.139, "step": 14563 }, { "epoch": 10.669597069597069, "grad_norm": 1.1953125, "learning_rate": 8.344060917047073e-05, "loss": 0.1601, "step": 14564 }, { "epoch": 10.67032967032967, "grad_norm": 1.640625, "learning_rate": 8.343805156154954e-05, "loss": 0.1582, "step": 14565 }, { "epoch": 10.67106227106227, "grad_norm": 1.78125, "learning_rate": 8.343549379481133e-05, "loss": 0.1993, "step": 14566 }, { "epoch": 10.671794871794871, "grad_norm": 0.87890625, "learning_rate": 8.343293587026836e-05, "loss": 0.134, "step": 14567 }, { "epoch": 10.672527472527472, "grad_norm": 1.1796875, "learning_rate": 8.343037778793289e-05, "loss": 0.1474, "step": 14568 }, { "epoch": 10.673260073260073, "grad_norm": 0.859375, "learning_rate": 8.342781954781716e-05, "loss": 0.0956, "step": 14569 }, { "epoch": 10.673992673992673, "grad_norm": 1.1796875, "learning_rate": 8.342526114993345e-05, "loss": 0.1718, "step": 14570 }, { "epoch": 10.674725274725274, "grad_norm": 0.95703125, "learning_rate": 8.342270259429398e-05, "loss": 0.1588, "step": 14571 }, { "epoch": 10.675457875457875, "grad_norm": 0.953125, "learning_rate": 8.342014388091106e-05, "loss": 0.1526, "step": 14572 }, { "epoch": 10.676190476190476, "grad_norm": 1.109375, "learning_rate": 8.341758500979692e-05, "loss": 0.1625, "step": 14573 }, { "epoch": 10.676923076923076, "grad_norm": 1.0859375, "learning_rate": 8.34150259809638e-05, "loss": 0.1611, "step": 14574 }, { "epoch": 10.677655677655677, "grad_norm": 1.640625, "learning_rate": 8.3412466794424e-05, "loss": 0.1407, "step": 14575 }, { "epoch": 10.678388278388278, "grad_norm": 1.3671875, "learning_rate": 8.340990745018977e-05, "loss": 0.1357, "step": 14576 }, { "epoch": 10.679120879120878, "grad_norm": 1.2109375, "learning_rate": 8.340734794827338e-05, "loss": 0.1662, "step": 14577 }, { "epoch": 10.67985347985348, "grad_norm": 1.40625, "learning_rate": 8.340478828868708e-05, "loss": 0.1259, "step": 14578 }, { "epoch": 10.68058608058608, "grad_norm": 0.953125, "learning_rate": 8.340222847144314e-05, "loss": 0.1487, "step": 14579 }, { "epoch": 10.68131868131868, "grad_norm": 1.7265625, "learning_rate": 8.33996684965538e-05, "loss": 0.119, "step": 14580 }, { "epoch": 10.682051282051281, "grad_norm": 0.84375, "learning_rate": 8.339710836403135e-05, "loss": 0.1246, "step": 14581 }, { "epoch": 10.682783882783882, "grad_norm": 1.0859375, "learning_rate": 8.339454807388807e-05, "loss": 0.1525, "step": 14582 }, { "epoch": 10.683516483516483, "grad_norm": 0.78125, "learning_rate": 8.339198762613622e-05, "loss": 0.1441, "step": 14583 }, { "epoch": 10.684249084249084, "grad_norm": 0.78515625, "learning_rate": 8.338942702078803e-05, "loss": 0.1372, "step": 14584 }, { "epoch": 10.684981684981684, "grad_norm": 0.9375, "learning_rate": 8.338686625785582e-05, "loss": 0.1514, "step": 14585 }, { "epoch": 10.685714285714285, "grad_norm": 0.81640625, "learning_rate": 8.338430533735184e-05, "loss": 0.1233, "step": 14586 }, { "epoch": 10.686446886446886, "grad_norm": 0.78515625, "learning_rate": 8.338174425928835e-05, "loss": 0.1351, "step": 14587 }, { "epoch": 10.687179487179487, "grad_norm": 0.84765625, "learning_rate": 8.337918302367762e-05, "loss": 0.1075, "step": 14588 }, { "epoch": 10.687912087912087, "grad_norm": 0.859375, "learning_rate": 8.337662163053193e-05, "loss": 0.1178, "step": 14589 }, { "epoch": 10.688644688644688, "grad_norm": 0.8359375, "learning_rate": 8.337406007986357e-05, "loss": 0.1451, "step": 14590 }, { "epoch": 10.689377289377289, "grad_norm": 1.6015625, "learning_rate": 8.33714983716848e-05, "loss": 0.1082, "step": 14591 }, { "epoch": 10.69010989010989, "grad_norm": 1.59375, "learning_rate": 8.336893650600787e-05, "loss": 0.1375, "step": 14592 }, { "epoch": 10.69084249084249, "grad_norm": 1.875, "learning_rate": 8.336637448284508e-05, "loss": 0.1492, "step": 14593 }, { "epoch": 10.691575091575091, "grad_norm": 1.5390625, "learning_rate": 8.336381230220871e-05, "loss": 0.1468, "step": 14594 }, { "epoch": 10.692307692307692, "grad_norm": 1.109375, "learning_rate": 8.3361249964111e-05, "loss": 0.1265, "step": 14595 }, { "epoch": 10.693040293040292, "grad_norm": 0.96875, "learning_rate": 8.33586874685643e-05, "loss": 0.1252, "step": 14596 }, { "epoch": 10.693772893772893, "grad_norm": 1.4296875, "learning_rate": 8.335612481558081e-05, "loss": 0.1332, "step": 14597 }, { "epoch": 10.694505494505494, "grad_norm": 1.234375, "learning_rate": 8.335356200517284e-05, "loss": 0.1136, "step": 14598 }, { "epoch": 10.695238095238095, "grad_norm": 1.8671875, "learning_rate": 8.335099903735269e-05, "loss": 0.13, "step": 14599 }, { "epoch": 10.695970695970695, "grad_norm": 0.67578125, "learning_rate": 8.33484359121326e-05, "loss": 0.1146, "step": 14600 }, { "epoch": 10.696703296703296, "grad_norm": 1.0234375, "learning_rate": 8.334587262952488e-05, "loss": 0.124, "step": 14601 }, { "epoch": 10.697435897435897, "grad_norm": 1.109375, "learning_rate": 8.33433091895418e-05, "loss": 0.1218, "step": 14602 }, { "epoch": 10.698168498168497, "grad_norm": 0.86328125, "learning_rate": 8.334074559219562e-05, "loss": 0.1281, "step": 14603 }, { "epoch": 10.698901098901098, "grad_norm": 0.75390625, "learning_rate": 8.333818183749868e-05, "loss": 0.111, "step": 14604 }, { "epoch": 10.699633699633699, "grad_norm": 1.0703125, "learning_rate": 8.333561792546323e-05, "loss": 0.1399, "step": 14605 }, { "epoch": 10.7003663003663, "grad_norm": 1.03125, "learning_rate": 8.333305385610154e-05, "loss": 0.1481, "step": 14606 }, { "epoch": 10.7010989010989, "grad_norm": 1.125, "learning_rate": 8.333048962942592e-05, "loss": 0.1373, "step": 14607 }, { "epoch": 10.701831501831501, "grad_norm": 1.21875, "learning_rate": 8.332792524544863e-05, "loss": 0.1233, "step": 14608 }, { "epoch": 10.702564102564102, "grad_norm": 1.21875, "learning_rate": 8.332536070418198e-05, "loss": 0.1383, "step": 14609 }, { "epoch": 10.703296703296703, "grad_norm": 1.171875, "learning_rate": 8.332279600563826e-05, "loss": 0.1258, "step": 14610 }, { "epoch": 10.704029304029303, "grad_norm": 0.828125, "learning_rate": 8.332023114982974e-05, "loss": 0.0993, "step": 14611 }, { "epoch": 10.704761904761904, "grad_norm": 1.375, "learning_rate": 8.331766613676874e-05, "loss": 0.1634, "step": 14612 }, { "epoch": 10.705494505494505, "grad_norm": 1.109375, "learning_rate": 8.33151009664675e-05, "loss": 0.1395, "step": 14613 }, { "epoch": 10.706227106227105, "grad_norm": 0.77734375, "learning_rate": 8.331253563893835e-05, "loss": 0.1345, "step": 14614 }, { "epoch": 10.706959706959706, "grad_norm": 0.84765625, "learning_rate": 8.330997015419356e-05, "loss": 0.1059, "step": 14615 }, { "epoch": 10.707692307692307, "grad_norm": 1.40625, "learning_rate": 8.330740451224547e-05, "loss": 0.1268, "step": 14616 }, { "epoch": 10.708424908424908, "grad_norm": 0.859375, "learning_rate": 8.330483871310629e-05, "loss": 0.1247, "step": 14617 }, { "epoch": 10.709157509157508, "grad_norm": 1.7734375, "learning_rate": 8.33022727567884e-05, "loss": 0.1393, "step": 14618 }, { "epoch": 10.70989010989011, "grad_norm": 1.3828125, "learning_rate": 8.329970664330401e-05, "loss": 0.1211, "step": 14619 }, { "epoch": 10.71062271062271, "grad_norm": 1.2578125, "learning_rate": 8.329714037266548e-05, "loss": 0.1749, "step": 14620 }, { "epoch": 10.71135531135531, "grad_norm": 1.25, "learning_rate": 8.329457394488507e-05, "loss": 0.1523, "step": 14621 }, { "epoch": 10.712087912087911, "grad_norm": 0.921875, "learning_rate": 8.329200735997511e-05, "loss": 0.1617, "step": 14622 }, { "epoch": 10.712820512820512, "grad_norm": 1.765625, "learning_rate": 8.328944061794786e-05, "loss": 0.1434, "step": 14623 }, { "epoch": 10.713553113553113, "grad_norm": 1.9375, "learning_rate": 8.328687371881566e-05, "loss": 0.1207, "step": 14624 }, { "epoch": 10.714285714285714, "grad_norm": 1.578125, "learning_rate": 8.328430666259077e-05, "loss": 0.1255, "step": 14625 }, { "epoch": 10.715018315018314, "grad_norm": 1.1484375, "learning_rate": 8.32817394492855e-05, "loss": 0.1393, "step": 14626 }, { "epoch": 10.715750915750915, "grad_norm": 0.9453125, "learning_rate": 8.327917207891215e-05, "loss": 0.1169, "step": 14627 }, { "epoch": 10.716483516483516, "grad_norm": 1.03125, "learning_rate": 8.327660455148305e-05, "loss": 0.1243, "step": 14628 }, { "epoch": 10.717216117216116, "grad_norm": 0.9453125, "learning_rate": 8.327403686701045e-05, "loss": 0.131, "step": 14629 }, { "epoch": 10.717948717948717, "grad_norm": 1.9375, "learning_rate": 8.327146902550672e-05, "loss": 0.1681, "step": 14630 }, { "epoch": 10.718681318681318, "grad_norm": 1.2578125, "learning_rate": 8.326890102698408e-05, "loss": 0.1276, "step": 14631 }, { "epoch": 10.719413919413919, "grad_norm": 1.5625, "learning_rate": 8.32663328714549e-05, "loss": 0.1606, "step": 14632 }, { "epoch": 10.72014652014652, "grad_norm": 1.046875, "learning_rate": 8.326376455893146e-05, "loss": 0.1431, "step": 14633 }, { "epoch": 10.72087912087912, "grad_norm": 1.4921875, "learning_rate": 8.326119608942606e-05, "loss": 0.1425, "step": 14634 }, { "epoch": 10.72161172161172, "grad_norm": 2.078125, "learning_rate": 8.325862746295102e-05, "loss": 0.1465, "step": 14635 }, { "epoch": 10.722344322344322, "grad_norm": 1.390625, "learning_rate": 8.325605867951863e-05, "loss": 0.1214, "step": 14636 }, { "epoch": 10.723076923076922, "grad_norm": 2.125, "learning_rate": 8.325348973914123e-05, "loss": 0.1485, "step": 14637 }, { "epoch": 10.723809523809523, "grad_norm": 0.95703125, "learning_rate": 8.32509206418311e-05, "loss": 0.1366, "step": 14638 }, { "epoch": 10.724542124542124, "grad_norm": 1.0859375, "learning_rate": 8.324835138760056e-05, "loss": 0.1372, "step": 14639 }, { "epoch": 10.725274725274724, "grad_norm": 1.2421875, "learning_rate": 8.324578197646193e-05, "loss": 0.1226, "step": 14640 }, { "epoch": 10.726007326007325, "grad_norm": 0.984375, "learning_rate": 8.324321240842748e-05, "loss": 0.126, "step": 14641 }, { "epoch": 10.726739926739926, "grad_norm": 1.140625, "learning_rate": 8.324064268350957e-05, "loss": 0.1284, "step": 14642 }, { "epoch": 10.727472527472527, "grad_norm": 0.765625, "learning_rate": 8.32380728017205e-05, "loss": 0.1187, "step": 14643 }, { "epoch": 10.728205128205127, "grad_norm": 0.921875, "learning_rate": 8.323550276307255e-05, "loss": 0.1409, "step": 14644 }, { "epoch": 10.728937728937728, "grad_norm": 1.40625, "learning_rate": 8.323293256757807e-05, "loss": 0.1327, "step": 14645 }, { "epoch": 10.729670329670329, "grad_norm": 0.85546875, "learning_rate": 8.323036221524939e-05, "loss": 0.1259, "step": 14646 }, { "epoch": 10.73040293040293, "grad_norm": 1.3984375, "learning_rate": 8.322779170609877e-05, "loss": 0.1552, "step": 14647 }, { "epoch": 10.73113553113553, "grad_norm": 1.0859375, "learning_rate": 8.322522104013856e-05, "loss": 0.1331, "step": 14648 }, { "epoch": 10.731868131868131, "grad_norm": 1.6015625, "learning_rate": 8.322265021738109e-05, "loss": 0.1401, "step": 14649 }, { "epoch": 10.732600732600732, "grad_norm": 1.09375, "learning_rate": 8.322007923783863e-05, "loss": 0.1319, "step": 14650 }, { "epoch": 10.733333333333333, "grad_norm": 1.25, "learning_rate": 8.321750810152356e-05, "loss": 0.1161, "step": 14651 }, { "epoch": 10.734065934065933, "grad_norm": 1.046875, "learning_rate": 8.321493680844815e-05, "loss": 0.1368, "step": 14652 }, { "epoch": 10.734798534798536, "grad_norm": 0.875, "learning_rate": 8.321236535862474e-05, "loss": 0.1263, "step": 14653 }, { "epoch": 10.735531135531136, "grad_norm": 0.8125, "learning_rate": 8.320979375206566e-05, "loss": 0.1232, "step": 14654 }, { "epoch": 10.736263736263737, "grad_norm": 0.90625, "learning_rate": 8.320722198878321e-05, "loss": 0.1123, "step": 14655 }, { "epoch": 10.736996336996338, "grad_norm": 3.0, "learning_rate": 8.320465006878973e-05, "loss": 0.2239, "step": 14656 }, { "epoch": 10.737728937728939, "grad_norm": 1.484375, "learning_rate": 8.320207799209754e-05, "loss": 0.16, "step": 14657 }, { "epoch": 10.73846153846154, "grad_norm": 1.5546875, "learning_rate": 8.319950575871894e-05, "loss": 0.1301, "step": 14658 }, { "epoch": 10.73919413919414, "grad_norm": 1.1484375, "learning_rate": 8.319693336866629e-05, "loss": 0.148, "step": 14659 }, { "epoch": 10.73992673992674, "grad_norm": 0.98828125, "learning_rate": 8.319436082195188e-05, "loss": 0.1517, "step": 14660 }, { "epoch": 10.740659340659342, "grad_norm": 0.86328125, "learning_rate": 8.319178811858807e-05, "loss": 0.1105, "step": 14661 }, { "epoch": 10.741391941391942, "grad_norm": 1.15625, "learning_rate": 8.318921525858717e-05, "loss": 0.136, "step": 14662 }, { "epoch": 10.742124542124543, "grad_norm": 1.5546875, "learning_rate": 8.318664224196152e-05, "loss": 0.1321, "step": 14663 }, { "epoch": 10.742857142857144, "grad_norm": 1.5, "learning_rate": 8.318406906872342e-05, "loss": 0.1254, "step": 14664 }, { "epoch": 10.743589743589745, "grad_norm": 1.453125, "learning_rate": 8.31814957388852e-05, "loss": 0.1331, "step": 14665 }, { "epoch": 10.744322344322345, "grad_norm": 0.69921875, "learning_rate": 8.317892225245924e-05, "loss": 0.1025, "step": 14666 }, { "epoch": 10.745054945054946, "grad_norm": 1.390625, "learning_rate": 8.31763486094578e-05, "loss": 0.1316, "step": 14667 }, { "epoch": 10.745787545787547, "grad_norm": 1.296875, "learning_rate": 8.317377480989329e-05, "loss": 0.1268, "step": 14668 }, { "epoch": 10.746520146520147, "grad_norm": 1.5625, "learning_rate": 8.317120085377795e-05, "loss": 0.1227, "step": 14669 }, { "epoch": 10.747252747252748, "grad_norm": 0.984375, "learning_rate": 8.31686267411242e-05, "loss": 0.1334, "step": 14670 }, { "epoch": 10.747985347985349, "grad_norm": 1.265625, "learning_rate": 8.316605247194432e-05, "loss": 0.1218, "step": 14671 }, { "epoch": 10.74871794871795, "grad_norm": 1.0234375, "learning_rate": 8.316347804625066e-05, "loss": 0.1269, "step": 14672 }, { "epoch": 10.74945054945055, "grad_norm": 1.0, "learning_rate": 8.316090346405555e-05, "loss": 0.1277, "step": 14673 }, { "epoch": 10.750183150183151, "grad_norm": 1.0078125, "learning_rate": 8.315832872537134e-05, "loss": 0.13, "step": 14674 }, { "epoch": 10.750915750915752, "grad_norm": 0.8203125, "learning_rate": 8.315575383021034e-05, "loss": 0.0977, "step": 14675 }, { "epoch": 10.751648351648353, "grad_norm": 0.87890625, "learning_rate": 8.315317877858492e-05, "loss": 0.1318, "step": 14676 }, { "epoch": 10.752380952380953, "grad_norm": 1.0859375, "learning_rate": 8.315060357050739e-05, "loss": 0.1503, "step": 14677 }, { "epoch": 10.753113553113554, "grad_norm": 1.15625, "learning_rate": 8.31480282059901e-05, "loss": 0.1093, "step": 14678 }, { "epoch": 10.753846153846155, "grad_norm": 1.6953125, "learning_rate": 8.314545268504539e-05, "loss": 0.1228, "step": 14679 }, { "epoch": 10.754578754578755, "grad_norm": 1.15625, "learning_rate": 8.31428770076856e-05, "loss": 0.1222, "step": 14680 }, { "epoch": 10.755311355311356, "grad_norm": 0.97265625, "learning_rate": 8.314030117392307e-05, "loss": 0.1306, "step": 14681 }, { "epoch": 10.756043956043957, "grad_norm": 0.98046875, "learning_rate": 8.313772518377017e-05, "loss": 0.1221, "step": 14682 }, { "epoch": 10.756776556776558, "grad_norm": 1.2421875, "learning_rate": 8.313514903723918e-05, "loss": 0.141, "step": 14683 }, { "epoch": 10.757509157509158, "grad_norm": 0.609375, "learning_rate": 8.313257273434249e-05, "loss": 0.1066, "step": 14684 }, { "epoch": 10.758241758241759, "grad_norm": 1.0703125, "learning_rate": 8.312999627509244e-05, "loss": 0.1245, "step": 14685 }, { "epoch": 10.75897435897436, "grad_norm": 1.421875, "learning_rate": 8.312741965950134e-05, "loss": 0.1294, "step": 14686 }, { "epoch": 10.75970695970696, "grad_norm": 1.1953125, "learning_rate": 8.312484288758158e-05, "loss": 0.1463, "step": 14687 }, { "epoch": 10.760439560439561, "grad_norm": 1.1796875, "learning_rate": 8.312226595934547e-05, "loss": 0.1201, "step": 14688 }, { "epoch": 10.761172161172162, "grad_norm": 1.6484375, "learning_rate": 8.311968887480539e-05, "loss": 0.1478, "step": 14689 }, { "epoch": 10.761904761904763, "grad_norm": 0.91015625, "learning_rate": 8.311711163397368e-05, "loss": 0.1435, "step": 14690 }, { "epoch": 10.762637362637363, "grad_norm": 1.1171875, "learning_rate": 8.311453423686267e-05, "loss": 0.143, "step": 14691 }, { "epoch": 10.763369963369964, "grad_norm": 1.46875, "learning_rate": 8.311195668348472e-05, "loss": 0.1419, "step": 14692 }, { "epoch": 10.764102564102565, "grad_norm": 1.015625, "learning_rate": 8.310937897385218e-05, "loss": 0.1521, "step": 14693 }, { "epoch": 10.764835164835166, "grad_norm": 1.0859375, "learning_rate": 8.31068011079774e-05, "loss": 0.1416, "step": 14694 }, { "epoch": 10.765567765567766, "grad_norm": 0.703125, "learning_rate": 8.310422308587273e-05, "loss": 0.1175, "step": 14695 }, { "epoch": 10.766300366300367, "grad_norm": 1.640625, "learning_rate": 8.310164490755053e-05, "loss": 0.1588, "step": 14696 }, { "epoch": 10.767032967032968, "grad_norm": 1.2421875, "learning_rate": 8.309906657302313e-05, "loss": 0.1231, "step": 14697 }, { "epoch": 10.767765567765569, "grad_norm": 0.90625, "learning_rate": 8.309648808230293e-05, "loss": 0.1132, "step": 14698 }, { "epoch": 10.76849816849817, "grad_norm": 1.21875, "learning_rate": 8.309390943540224e-05, "loss": 0.1599, "step": 14699 }, { "epoch": 10.76923076923077, "grad_norm": 1.1875, "learning_rate": 8.309133063233344e-05, "loss": 0.115, "step": 14700 }, { "epoch": 10.76996336996337, "grad_norm": 1.0078125, "learning_rate": 8.308875167310885e-05, "loss": 0.167, "step": 14701 }, { "epoch": 10.770695970695972, "grad_norm": 0.96875, "learning_rate": 8.308617255774087e-05, "loss": 0.1667, "step": 14702 }, { "epoch": 10.771428571428572, "grad_norm": 0.87890625, "learning_rate": 8.308359328624183e-05, "loss": 0.1814, "step": 14703 }, { "epoch": 10.772161172161173, "grad_norm": 0.90234375, "learning_rate": 8.308101385862411e-05, "loss": 0.1293, "step": 14704 }, { "epoch": 10.772893772893774, "grad_norm": 1.3984375, "learning_rate": 8.307843427490005e-05, "loss": 0.1653, "step": 14705 }, { "epoch": 10.773626373626374, "grad_norm": 1.046875, "learning_rate": 8.307585453508202e-05, "loss": 0.1095, "step": 14706 }, { "epoch": 10.774358974358975, "grad_norm": 1.0234375, "learning_rate": 8.307327463918238e-05, "loss": 0.1663, "step": 14707 }, { "epoch": 10.775091575091576, "grad_norm": 1.5, "learning_rate": 8.307069458721349e-05, "loss": 0.1241, "step": 14708 }, { "epoch": 10.775824175824177, "grad_norm": 1.125, "learning_rate": 8.30681143791877e-05, "loss": 0.1438, "step": 14709 }, { "epoch": 10.776556776556777, "grad_norm": 1.0859375, "learning_rate": 8.306553401511737e-05, "loss": 0.1156, "step": 14710 }, { "epoch": 10.777289377289378, "grad_norm": 1.1015625, "learning_rate": 8.30629534950149e-05, "loss": 0.1333, "step": 14711 }, { "epoch": 10.778021978021979, "grad_norm": 0.79296875, "learning_rate": 8.306037281889263e-05, "loss": 0.1369, "step": 14712 }, { "epoch": 10.77875457875458, "grad_norm": 1.2890625, "learning_rate": 8.305779198676293e-05, "loss": 0.17, "step": 14713 }, { "epoch": 10.77948717948718, "grad_norm": 1.046875, "learning_rate": 8.305521099863813e-05, "loss": 0.1435, "step": 14714 }, { "epoch": 10.780219780219781, "grad_norm": 1.0625, "learning_rate": 8.305262985453065e-05, "loss": 0.1351, "step": 14715 }, { "epoch": 10.780952380952382, "grad_norm": 1.8671875, "learning_rate": 8.305004855445282e-05, "loss": 0.1504, "step": 14716 }, { "epoch": 10.781684981684982, "grad_norm": 2.109375, "learning_rate": 8.304746709841703e-05, "loss": 0.1186, "step": 14717 }, { "epoch": 10.782417582417583, "grad_norm": 1.015625, "learning_rate": 8.304488548643564e-05, "loss": 0.1376, "step": 14718 }, { "epoch": 10.783150183150184, "grad_norm": 1.296875, "learning_rate": 8.304230371852103e-05, "loss": 0.1269, "step": 14719 }, { "epoch": 10.783882783882785, "grad_norm": 0.96484375, "learning_rate": 8.303972179468556e-05, "loss": 0.1213, "step": 14720 }, { "epoch": 10.784615384615385, "grad_norm": 0.83984375, "learning_rate": 8.303713971494159e-05, "loss": 0.1177, "step": 14721 }, { "epoch": 10.785347985347986, "grad_norm": 1.0546875, "learning_rate": 8.30345574793015e-05, "loss": 0.1231, "step": 14722 }, { "epoch": 10.786080586080587, "grad_norm": 1.0234375, "learning_rate": 8.303197508777766e-05, "loss": 0.1444, "step": 14723 }, { "epoch": 10.786813186813188, "grad_norm": 0.9609375, "learning_rate": 8.302939254038246e-05, "loss": 0.152, "step": 14724 }, { "epoch": 10.787545787545788, "grad_norm": 1.4375, "learning_rate": 8.302680983712827e-05, "loss": 0.1369, "step": 14725 }, { "epoch": 10.788278388278389, "grad_norm": 1.4609375, "learning_rate": 8.302422697802745e-05, "loss": 0.1661, "step": 14726 }, { "epoch": 10.78901098901099, "grad_norm": 1.4140625, "learning_rate": 8.302164396309237e-05, "loss": 0.1139, "step": 14727 }, { "epoch": 10.78974358974359, "grad_norm": 1.3515625, "learning_rate": 8.301906079233543e-05, "loss": 0.1353, "step": 14728 }, { "epoch": 10.790476190476191, "grad_norm": 0.98046875, "learning_rate": 8.301647746576898e-05, "loss": 0.1406, "step": 14729 }, { "epoch": 10.791208791208792, "grad_norm": 1.6796875, "learning_rate": 8.301389398340541e-05, "loss": 0.1441, "step": 14730 }, { "epoch": 10.791941391941393, "grad_norm": 0.921875, "learning_rate": 8.301131034525712e-05, "loss": 0.15, "step": 14731 }, { "epoch": 10.792673992673993, "grad_norm": 1.0703125, "learning_rate": 8.300872655133646e-05, "loss": 0.1279, "step": 14732 }, { "epoch": 10.793406593406594, "grad_norm": 0.7890625, "learning_rate": 8.300614260165582e-05, "loss": 0.1198, "step": 14733 }, { "epoch": 10.794139194139195, "grad_norm": 0.67578125, "learning_rate": 8.300355849622758e-05, "loss": 0.1251, "step": 14734 }, { "epoch": 10.794871794871796, "grad_norm": 1.2890625, "learning_rate": 8.300097423506412e-05, "loss": 0.1183, "step": 14735 }, { "epoch": 10.795604395604396, "grad_norm": 1.328125, "learning_rate": 8.299838981817783e-05, "loss": 0.1457, "step": 14736 }, { "epoch": 10.796336996336997, "grad_norm": 1.5703125, "learning_rate": 8.299580524558107e-05, "loss": 0.1758, "step": 14737 }, { "epoch": 10.797069597069598, "grad_norm": 1.1875, "learning_rate": 8.299322051728624e-05, "loss": 0.1524, "step": 14738 }, { "epoch": 10.797802197802199, "grad_norm": 1.1015625, "learning_rate": 8.299063563330573e-05, "loss": 0.1345, "step": 14739 }, { "epoch": 10.7985347985348, "grad_norm": 1.5234375, "learning_rate": 8.298805059365192e-05, "loss": 0.1358, "step": 14740 }, { "epoch": 10.7992673992674, "grad_norm": 0.8515625, "learning_rate": 8.298546539833721e-05, "loss": 0.1057, "step": 14741 }, { "epoch": 10.8, "grad_norm": 1.09375, "learning_rate": 8.298288004737396e-05, "loss": 0.1352, "step": 14742 }, { "epoch": 10.800732600732601, "grad_norm": 0.8984375, "learning_rate": 8.298029454077455e-05, "loss": 0.1275, "step": 14743 }, { "epoch": 10.801465201465202, "grad_norm": 1.09375, "learning_rate": 8.297770887855141e-05, "loss": 0.1295, "step": 14744 }, { "epoch": 10.802197802197803, "grad_norm": 1.1875, "learning_rate": 8.297512306071689e-05, "loss": 0.1111, "step": 14745 }, { "epoch": 10.802930402930404, "grad_norm": 1.3515625, "learning_rate": 8.29725370872834e-05, "loss": 0.133, "step": 14746 }, { "epoch": 10.803663003663004, "grad_norm": 0.93359375, "learning_rate": 8.296995095826335e-05, "loss": 0.1383, "step": 14747 }, { "epoch": 10.804395604395605, "grad_norm": 1.3203125, "learning_rate": 8.296736467366906e-05, "loss": 0.1372, "step": 14748 }, { "epoch": 10.805128205128206, "grad_norm": 0.88671875, "learning_rate": 8.296477823351298e-05, "loss": 0.0986, "step": 14749 }, { "epoch": 10.805860805860807, "grad_norm": 1.1875, "learning_rate": 8.296219163780751e-05, "loss": 0.1591, "step": 14750 }, { "epoch": 10.806593406593407, "grad_norm": 0.69140625, "learning_rate": 8.295960488656501e-05, "loss": 0.1165, "step": 14751 }, { "epoch": 10.807326007326008, "grad_norm": 1.0078125, "learning_rate": 8.295701797979789e-05, "loss": 0.1603, "step": 14752 }, { "epoch": 10.808058608058609, "grad_norm": 0.68359375, "learning_rate": 8.295443091751852e-05, "loss": 0.1085, "step": 14753 }, { "epoch": 10.80879120879121, "grad_norm": 1.03125, "learning_rate": 8.295184369973936e-05, "loss": 0.0984, "step": 14754 }, { "epoch": 10.80952380952381, "grad_norm": 1.6796875, "learning_rate": 8.294925632647274e-05, "loss": 0.1759, "step": 14755 }, { "epoch": 10.810256410256411, "grad_norm": 0.859375, "learning_rate": 8.294666879773108e-05, "loss": 0.1341, "step": 14756 }, { "epoch": 10.810989010989012, "grad_norm": 0.84375, "learning_rate": 8.294408111352678e-05, "loss": 0.1468, "step": 14757 }, { "epoch": 10.811721611721612, "grad_norm": 1.71875, "learning_rate": 8.294149327387223e-05, "loss": 0.1434, "step": 14758 }, { "epoch": 10.812454212454213, "grad_norm": 1.0703125, "learning_rate": 8.293890527877985e-05, "loss": 0.1006, "step": 14759 }, { "epoch": 10.813186813186814, "grad_norm": 0.625, "learning_rate": 8.293631712826203e-05, "loss": 0.1237, "step": 14760 }, { "epoch": 10.813919413919415, "grad_norm": 0.75390625, "learning_rate": 8.293372882233114e-05, "loss": 0.1367, "step": 14761 }, { "epoch": 10.814652014652015, "grad_norm": 1.4140625, "learning_rate": 8.293114036099962e-05, "loss": 0.1511, "step": 14762 }, { "epoch": 10.815384615384616, "grad_norm": 0.95703125, "learning_rate": 8.292855174427987e-05, "loss": 0.1421, "step": 14763 }, { "epoch": 10.816117216117217, "grad_norm": 0.7890625, "learning_rate": 8.292596297218428e-05, "loss": 0.1213, "step": 14764 }, { "epoch": 10.816849816849818, "grad_norm": 1.1484375, "learning_rate": 8.292337404472524e-05, "loss": 0.1384, "step": 14765 }, { "epoch": 10.817582417582418, "grad_norm": 1.0546875, "learning_rate": 8.292078496191518e-05, "loss": 0.0999, "step": 14766 }, { "epoch": 10.818315018315019, "grad_norm": 1.0546875, "learning_rate": 8.291819572376652e-05, "loss": 0.1253, "step": 14767 }, { "epoch": 10.81904761904762, "grad_norm": 1.546875, "learning_rate": 8.291560633029161e-05, "loss": 0.1248, "step": 14768 }, { "epoch": 10.81978021978022, "grad_norm": 0.6796875, "learning_rate": 8.29130167815029e-05, "loss": 0.1199, "step": 14769 }, { "epoch": 10.820512820512821, "grad_norm": 1.484375, "learning_rate": 8.29104270774128e-05, "loss": 0.1434, "step": 14770 }, { "epoch": 10.821245421245422, "grad_norm": 1.2265625, "learning_rate": 8.290783721803368e-05, "loss": 0.1294, "step": 14771 }, { "epoch": 10.821978021978023, "grad_norm": 0.95703125, "learning_rate": 8.290524720337799e-05, "loss": 0.1361, "step": 14772 }, { "epoch": 10.822710622710623, "grad_norm": 0.9765625, "learning_rate": 8.29026570334581e-05, "loss": 0.1429, "step": 14773 }, { "epoch": 10.823443223443224, "grad_norm": 0.9609375, "learning_rate": 8.290006670828647e-05, "loss": 0.1405, "step": 14774 }, { "epoch": 10.824175824175825, "grad_norm": 1.28125, "learning_rate": 8.289747622787548e-05, "loss": 0.1515, "step": 14775 }, { "epoch": 10.824908424908426, "grad_norm": 1.890625, "learning_rate": 8.289488559223755e-05, "loss": 0.126, "step": 14776 }, { "epoch": 10.825641025641026, "grad_norm": 1.0078125, "learning_rate": 8.289229480138508e-05, "loss": 0.1228, "step": 14777 }, { "epoch": 10.826373626373627, "grad_norm": 0.88671875, "learning_rate": 8.288970385533048e-05, "loss": 0.1393, "step": 14778 }, { "epoch": 10.827106227106228, "grad_norm": 0.7890625, "learning_rate": 8.288711275408619e-05, "loss": 0.1171, "step": 14779 }, { "epoch": 10.827838827838828, "grad_norm": 0.625, "learning_rate": 8.288452149766462e-05, "loss": 0.1417, "step": 14780 }, { "epoch": 10.82857142857143, "grad_norm": 1.40625, "learning_rate": 8.288193008607815e-05, "loss": 0.1548, "step": 14781 }, { "epoch": 10.82930402930403, "grad_norm": 0.77734375, "learning_rate": 8.287933851933922e-05, "loss": 0.1162, "step": 14782 }, { "epoch": 10.83003663003663, "grad_norm": 1.65625, "learning_rate": 8.287674679746028e-05, "loss": 0.1266, "step": 14783 }, { "epoch": 10.830769230769231, "grad_norm": 0.88671875, "learning_rate": 8.287415492045372e-05, "loss": 0.1126, "step": 14784 }, { "epoch": 10.831501831501832, "grad_norm": 1.53125, "learning_rate": 8.287156288833193e-05, "loss": 0.1282, "step": 14785 }, { "epoch": 10.832234432234433, "grad_norm": 1.0546875, "learning_rate": 8.286897070110736e-05, "loss": 0.1283, "step": 14786 }, { "epoch": 10.832967032967034, "grad_norm": 1.09375, "learning_rate": 8.286637835879245e-05, "loss": 0.1388, "step": 14787 }, { "epoch": 10.833699633699634, "grad_norm": 0.98828125, "learning_rate": 8.286378586139956e-05, "loss": 0.1242, "step": 14788 }, { "epoch": 10.834432234432235, "grad_norm": 0.96484375, "learning_rate": 8.286119320894116e-05, "loss": 0.1484, "step": 14789 }, { "epoch": 10.835164835164836, "grad_norm": 1.0390625, "learning_rate": 8.285860040142966e-05, "loss": 0.1066, "step": 14790 }, { "epoch": 10.835897435897436, "grad_norm": 1.28125, "learning_rate": 8.28560074388775e-05, "loss": 0.1418, "step": 14791 }, { "epoch": 10.836630036630037, "grad_norm": 0.859375, "learning_rate": 8.285341432129707e-05, "loss": 0.1197, "step": 14792 }, { "epoch": 10.837362637362638, "grad_norm": 1.015625, "learning_rate": 8.285082104870081e-05, "loss": 0.1044, "step": 14793 }, { "epoch": 10.838095238095239, "grad_norm": 0.98046875, "learning_rate": 8.284822762110116e-05, "loss": 0.1344, "step": 14794 }, { "epoch": 10.83882783882784, "grad_norm": 1.03125, "learning_rate": 8.284563403851051e-05, "loss": 0.1246, "step": 14795 }, { "epoch": 10.83956043956044, "grad_norm": 0.93359375, "learning_rate": 8.284304030094133e-05, "loss": 0.1145, "step": 14796 }, { "epoch": 10.84029304029304, "grad_norm": 0.80078125, "learning_rate": 8.284044640840601e-05, "loss": 0.1197, "step": 14797 }, { "epoch": 10.841025641025642, "grad_norm": 0.703125, "learning_rate": 8.283785236091702e-05, "loss": 0.1149, "step": 14798 }, { "epoch": 10.841758241758242, "grad_norm": 1.171875, "learning_rate": 8.283525815848674e-05, "loss": 0.1554, "step": 14799 }, { "epoch": 10.842490842490843, "grad_norm": 0.65625, "learning_rate": 8.283266380112763e-05, "loss": 0.1273, "step": 14800 }, { "epoch": 10.843223443223444, "grad_norm": 0.85546875, "learning_rate": 8.283006928885211e-05, "loss": 0.1171, "step": 14801 }, { "epoch": 10.843956043956045, "grad_norm": 0.89453125, "learning_rate": 8.282747462167262e-05, "loss": 0.1422, "step": 14802 }, { "epoch": 10.844688644688645, "grad_norm": 0.82421875, "learning_rate": 8.282487979960159e-05, "loss": 0.1124, "step": 14803 }, { "epoch": 10.845421245421246, "grad_norm": 1.3125, "learning_rate": 8.282228482265144e-05, "loss": 0.1147, "step": 14804 }, { "epoch": 10.846153846153847, "grad_norm": 1.125, "learning_rate": 8.281968969083463e-05, "loss": 0.1456, "step": 14805 }, { "epoch": 10.846886446886447, "grad_norm": 1.5625, "learning_rate": 8.281709440416355e-05, "loss": 0.126, "step": 14806 }, { "epoch": 10.847619047619048, "grad_norm": 1.3828125, "learning_rate": 8.281449896265068e-05, "loss": 0.1533, "step": 14807 }, { "epoch": 10.848351648351649, "grad_norm": 1.1171875, "learning_rate": 8.281190336630842e-05, "loss": 0.1157, "step": 14808 }, { "epoch": 10.84908424908425, "grad_norm": 1.1171875, "learning_rate": 8.280930761514924e-05, "loss": 0.1294, "step": 14809 }, { "epoch": 10.84981684981685, "grad_norm": 1.7890625, "learning_rate": 8.280671170918554e-05, "loss": 0.122, "step": 14810 }, { "epoch": 10.850549450549451, "grad_norm": 0.8515625, "learning_rate": 8.280411564842983e-05, "loss": 0.1169, "step": 14811 }, { "epoch": 10.851282051282052, "grad_norm": 0.8671875, "learning_rate": 8.280151943289445e-05, "loss": 0.155, "step": 14812 }, { "epoch": 10.852014652014653, "grad_norm": 1.3203125, "learning_rate": 8.27989230625919e-05, "loss": 0.1531, "step": 14813 }, { "epoch": 10.852747252747253, "grad_norm": 1.2578125, "learning_rate": 8.27963265375346e-05, "loss": 0.116, "step": 14814 }, { "epoch": 10.853479853479854, "grad_norm": 1.6015625, "learning_rate": 8.279372985773502e-05, "loss": 0.1357, "step": 14815 }, { "epoch": 10.854212454212455, "grad_norm": 1.984375, "learning_rate": 8.279113302320554e-05, "loss": 0.1447, "step": 14816 }, { "epoch": 10.854945054945055, "grad_norm": 1.2734375, "learning_rate": 8.278853603395867e-05, "loss": 0.1453, "step": 14817 }, { "epoch": 10.855677655677656, "grad_norm": 0.85546875, "learning_rate": 8.278593889000683e-05, "loss": 0.1185, "step": 14818 }, { "epoch": 10.856410256410257, "grad_norm": 0.8828125, "learning_rate": 8.278334159136242e-05, "loss": 0.1316, "step": 14819 }, { "epoch": 10.857142857142858, "grad_norm": 0.9921875, "learning_rate": 8.278074413803796e-05, "loss": 0.1515, "step": 14820 }, { "epoch": 10.857875457875458, "grad_norm": 1.75, "learning_rate": 8.277814653004585e-05, "loss": 0.1402, "step": 14821 }, { "epoch": 10.858608058608059, "grad_norm": 1.1953125, "learning_rate": 8.277554876739854e-05, "loss": 0.1067, "step": 14822 }, { "epoch": 10.85934065934066, "grad_norm": 0.63671875, "learning_rate": 8.277295085010849e-05, "loss": 0.0943, "step": 14823 }, { "epoch": 10.86007326007326, "grad_norm": 1.1328125, "learning_rate": 8.277035277818813e-05, "loss": 0.1677, "step": 14824 }, { "epoch": 10.860805860805861, "grad_norm": 0.8125, "learning_rate": 8.276775455164991e-05, "loss": 0.1204, "step": 14825 }, { "epoch": 10.861538461538462, "grad_norm": 1.2890625, "learning_rate": 8.276515617050628e-05, "loss": 0.1545, "step": 14826 }, { "epoch": 10.862271062271063, "grad_norm": 0.640625, "learning_rate": 8.27625576347697e-05, "loss": 0.106, "step": 14827 }, { "epoch": 10.863003663003663, "grad_norm": 1.15625, "learning_rate": 8.275995894445262e-05, "loss": 0.1168, "step": 14828 }, { "epoch": 10.863736263736264, "grad_norm": 1.1875, "learning_rate": 8.27573600995675e-05, "loss": 0.121, "step": 14829 }, { "epoch": 10.864468864468865, "grad_norm": 1.1484375, "learning_rate": 8.275476110012677e-05, "loss": 0.1316, "step": 14830 }, { "epoch": 10.865201465201466, "grad_norm": 1.1484375, "learning_rate": 8.275216194614287e-05, "loss": 0.1143, "step": 14831 }, { "epoch": 10.865934065934066, "grad_norm": 1.234375, "learning_rate": 8.274956263762829e-05, "loss": 0.1242, "step": 14832 }, { "epoch": 10.866666666666667, "grad_norm": 0.796875, "learning_rate": 8.274696317459547e-05, "loss": 0.1358, "step": 14833 }, { "epoch": 10.867399267399268, "grad_norm": 1.53125, "learning_rate": 8.274436355705686e-05, "loss": 0.1176, "step": 14834 }, { "epoch": 10.868131868131869, "grad_norm": 1.3984375, "learning_rate": 8.274176378502492e-05, "loss": 0.1371, "step": 14835 }, { "epoch": 10.86886446886447, "grad_norm": 1.8125, "learning_rate": 8.273916385851211e-05, "loss": 0.1344, "step": 14836 }, { "epoch": 10.86959706959707, "grad_norm": 1.046875, "learning_rate": 8.273656377753087e-05, "loss": 0.1153, "step": 14837 }, { "epoch": 10.87032967032967, "grad_norm": 1.0625, "learning_rate": 8.273396354209368e-05, "loss": 0.1422, "step": 14838 }, { "epoch": 10.871062271062272, "grad_norm": 1.203125, "learning_rate": 8.273136315221299e-05, "loss": 0.151, "step": 14839 }, { "epoch": 10.871794871794872, "grad_norm": 0.90625, "learning_rate": 8.272876260790125e-05, "loss": 0.1475, "step": 14840 }, { "epoch": 10.872527472527473, "grad_norm": 1.0078125, "learning_rate": 8.272616190917095e-05, "loss": 0.1489, "step": 14841 }, { "epoch": 10.873260073260074, "grad_norm": 1.6171875, "learning_rate": 8.272356105603452e-05, "loss": 0.1195, "step": 14842 }, { "epoch": 10.873992673992674, "grad_norm": 2.21875, "learning_rate": 8.272096004850442e-05, "loss": 0.1466, "step": 14843 }, { "epoch": 10.874725274725275, "grad_norm": 1.1015625, "learning_rate": 8.271835888659312e-05, "loss": 0.1394, "step": 14844 }, { "epoch": 10.875457875457876, "grad_norm": 0.77734375, "learning_rate": 8.271575757031309e-05, "loss": 0.1246, "step": 14845 }, { "epoch": 10.876190476190477, "grad_norm": 0.87890625, "learning_rate": 8.27131560996768e-05, "loss": 0.0971, "step": 14846 }, { "epoch": 10.876923076923077, "grad_norm": 0.8984375, "learning_rate": 8.27105544746967e-05, "loss": 0.1206, "step": 14847 }, { "epoch": 10.877655677655678, "grad_norm": 1.078125, "learning_rate": 8.270795269538525e-05, "loss": 0.1209, "step": 14848 }, { "epoch": 10.878388278388279, "grad_norm": 1.5234375, "learning_rate": 8.270535076175496e-05, "loss": 0.1393, "step": 14849 }, { "epoch": 10.87912087912088, "grad_norm": 1.9453125, "learning_rate": 8.270274867381821e-05, "loss": 0.1554, "step": 14850 }, { "epoch": 10.87985347985348, "grad_norm": 1.9140625, "learning_rate": 8.270014643158755e-05, "loss": 0.1565, "step": 14851 }, { "epoch": 10.880586080586081, "grad_norm": 1.140625, "learning_rate": 8.26975440350754e-05, "loss": 0.1402, "step": 14852 }, { "epoch": 10.881318681318682, "grad_norm": 0.7109375, "learning_rate": 8.269494148429427e-05, "loss": 0.0959, "step": 14853 }, { "epoch": 10.882051282051282, "grad_norm": 1.2734375, "learning_rate": 8.26923387792566e-05, "loss": 0.1375, "step": 14854 }, { "epoch": 10.882783882783883, "grad_norm": 1.0546875, "learning_rate": 8.268973591997486e-05, "loss": 0.1213, "step": 14855 }, { "epoch": 10.883516483516484, "grad_norm": 0.8984375, "learning_rate": 8.268713290646153e-05, "loss": 0.144, "step": 14856 }, { "epoch": 10.884249084249085, "grad_norm": 1.25, "learning_rate": 8.268452973872906e-05, "loss": 0.1689, "step": 14857 }, { "epoch": 10.884981684981685, "grad_norm": 1.6328125, "learning_rate": 8.268192641678998e-05, "loss": 0.1568, "step": 14858 }, { "epoch": 10.885714285714286, "grad_norm": 1.21875, "learning_rate": 8.26793229406567e-05, "loss": 0.1558, "step": 14859 }, { "epoch": 10.886446886446887, "grad_norm": 1.0078125, "learning_rate": 8.267671931034174e-05, "loss": 0.1251, "step": 14860 }, { "epoch": 10.887179487179488, "grad_norm": 1.1328125, "learning_rate": 8.267411552585753e-05, "loss": 0.1331, "step": 14861 }, { "epoch": 10.887912087912088, "grad_norm": 0.859375, "learning_rate": 8.26715115872166e-05, "loss": 0.1226, "step": 14862 }, { "epoch": 10.888644688644689, "grad_norm": 0.8984375, "learning_rate": 8.266890749443137e-05, "loss": 0.1226, "step": 14863 }, { "epoch": 10.88937728937729, "grad_norm": 1.09375, "learning_rate": 8.266630324751435e-05, "loss": 0.1232, "step": 14864 }, { "epoch": 10.89010989010989, "grad_norm": 0.6484375, "learning_rate": 8.266369884647802e-05, "loss": 0.1271, "step": 14865 }, { "epoch": 10.890842490842491, "grad_norm": 0.7578125, "learning_rate": 8.266109429133484e-05, "loss": 0.1329, "step": 14866 }, { "epoch": 10.891575091575092, "grad_norm": 1.3828125, "learning_rate": 8.265848958209731e-05, "loss": 0.1119, "step": 14867 }, { "epoch": 10.892307692307693, "grad_norm": 0.7734375, "learning_rate": 8.26558847187779e-05, "loss": 0.1211, "step": 14868 }, { "epoch": 10.893040293040293, "grad_norm": 1.1640625, "learning_rate": 8.265327970138908e-05, "loss": 0.1145, "step": 14869 }, { "epoch": 10.893772893772894, "grad_norm": 1.1796875, "learning_rate": 8.265067452994335e-05, "loss": 0.1205, "step": 14870 }, { "epoch": 10.894505494505495, "grad_norm": 1.0625, "learning_rate": 8.264806920445319e-05, "loss": 0.133, "step": 14871 }, { "epoch": 10.895238095238096, "grad_norm": 0.79296875, "learning_rate": 8.264546372493106e-05, "loss": 0.1158, "step": 14872 }, { "epoch": 10.895970695970696, "grad_norm": 1.09375, "learning_rate": 8.264285809138946e-05, "loss": 0.1449, "step": 14873 }, { "epoch": 10.896703296703297, "grad_norm": 1.65625, "learning_rate": 8.26402523038409e-05, "loss": 0.1437, "step": 14874 }, { "epoch": 10.897435897435898, "grad_norm": 0.87890625, "learning_rate": 8.263764636229782e-05, "loss": 0.1115, "step": 14875 }, { "epoch": 10.898168498168499, "grad_norm": 1.2265625, "learning_rate": 8.263504026677273e-05, "loss": 0.1367, "step": 14876 }, { "epoch": 10.8989010989011, "grad_norm": 1.4765625, "learning_rate": 8.263243401727812e-05, "loss": 0.1299, "step": 14877 }, { "epoch": 10.8996336996337, "grad_norm": 0.9375, "learning_rate": 8.262982761382645e-05, "loss": 0.1166, "step": 14878 }, { "epoch": 10.9003663003663, "grad_norm": 1.25, "learning_rate": 8.262722105643025e-05, "loss": 0.1382, "step": 14879 }, { "epoch": 10.901098901098901, "grad_norm": 1.0859375, "learning_rate": 8.262461434510199e-05, "loss": 0.1145, "step": 14880 }, { "epoch": 10.901831501831502, "grad_norm": 0.7109375, "learning_rate": 8.262200747985412e-05, "loss": 0.1017, "step": 14881 }, { "epoch": 10.902564102564103, "grad_norm": 1.1015625, "learning_rate": 8.26194004606992e-05, "loss": 0.1397, "step": 14882 }, { "epoch": 10.903296703296704, "grad_norm": 1.921875, "learning_rate": 8.261679328764968e-05, "loss": 0.1473, "step": 14883 }, { "epoch": 10.904029304029304, "grad_norm": 1.0234375, "learning_rate": 8.261418596071806e-05, "loss": 0.1221, "step": 14884 }, { "epoch": 10.904761904761905, "grad_norm": 1.34375, "learning_rate": 8.261157847991684e-05, "loss": 0.1252, "step": 14885 }, { "epoch": 10.905494505494506, "grad_norm": 0.72265625, "learning_rate": 8.26089708452585e-05, "loss": 0.1084, "step": 14886 }, { "epoch": 10.906227106227107, "grad_norm": 1.6953125, "learning_rate": 8.260636305675553e-05, "loss": 0.1593, "step": 14887 }, { "epoch": 10.906959706959707, "grad_norm": 1.5390625, "learning_rate": 8.260375511442045e-05, "loss": 0.1093, "step": 14888 }, { "epoch": 10.907692307692308, "grad_norm": 1.0390625, "learning_rate": 8.260114701826574e-05, "loss": 0.1331, "step": 14889 }, { "epoch": 10.908424908424909, "grad_norm": 0.67578125, "learning_rate": 8.259853876830388e-05, "loss": 0.1015, "step": 14890 }, { "epoch": 10.90915750915751, "grad_norm": 0.6640625, "learning_rate": 8.25959303645474e-05, "loss": 0.1225, "step": 14891 }, { "epoch": 10.90989010989011, "grad_norm": 1.8125, "learning_rate": 8.259332180700879e-05, "loss": 0.1847, "step": 14892 }, { "epoch": 10.910622710622711, "grad_norm": 1.03125, "learning_rate": 8.259071309570052e-05, "loss": 0.1471, "step": 14893 }, { "epoch": 10.911355311355312, "grad_norm": 1.0859375, "learning_rate": 8.258810423063511e-05, "loss": 0.1203, "step": 14894 }, { "epoch": 10.912087912087912, "grad_norm": 1.3984375, "learning_rate": 8.258549521182506e-05, "loss": 0.1541, "step": 14895 }, { "epoch": 10.912820512820513, "grad_norm": 1.5390625, "learning_rate": 8.258288603928287e-05, "loss": 0.1337, "step": 14896 }, { "epoch": 10.913553113553114, "grad_norm": 0.671875, "learning_rate": 8.258027671302105e-05, "loss": 0.0943, "step": 14897 }, { "epoch": 10.914285714285715, "grad_norm": 0.6640625, "learning_rate": 8.257766723305209e-05, "loss": 0.1098, "step": 14898 }, { "epoch": 10.915018315018315, "grad_norm": 0.7109375, "learning_rate": 8.257505759938849e-05, "loss": 0.104, "step": 14899 }, { "epoch": 10.915750915750916, "grad_norm": 1.28125, "learning_rate": 8.257244781204276e-05, "loss": 0.1316, "step": 14900 }, { "epoch": 10.916483516483517, "grad_norm": 0.7734375, "learning_rate": 8.256983787102742e-05, "loss": 0.1194, "step": 14901 }, { "epoch": 10.917216117216118, "grad_norm": 0.7734375, "learning_rate": 8.256722777635494e-05, "loss": 0.1049, "step": 14902 }, { "epoch": 10.917948717948718, "grad_norm": 1.0078125, "learning_rate": 8.256461752803785e-05, "loss": 0.1237, "step": 14903 }, { "epoch": 10.918681318681319, "grad_norm": 1.3125, "learning_rate": 8.256200712608863e-05, "loss": 0.1508, "step": 14904 }, { "epoch": 10.91941391941392, "grad_norm": 0.8046875, "learning_rate": 8.255939657051984e-05, "loss": 0.1156, "step": 14905 }, { "epoch": 10.92014652014652, "grad_norm": 0.87109375, "learning_rate": 8.255678586134395e-05, "loss": 0.0883, "step": 14906 }, { "epoch": 10.920879120879121, "grad_norm": 1.8203125, "learning_rate": 8.255417499857346e-05, "loss": 0.1229, "step": 14907 }, { "epoch": 10.921611721611722, "grad_norm": 0.75390625, "learning_rate": 8.255156398222091e-05, "loss": 0.101, "step": 14908 }, { "epoch": 10.922344322344323, "grad_norm": 1.09375, "learning_rate": 8.25489528122988e-05, "loss": 0.0959, "step": 14909 }, { "epoch": 10.923076923076923, "grad_norm": 0.953125, "learning_rate": 8.254634148881962e-05, "loss": 0.1477, "step": 14910 }, { "epoch": 10.923809523809524, "grad_norm": 0.9296875, "learning_rate": 8.254373001179591e-05, "loss": 0.1661, "step": 14911 }, { "epoch": 10.924542124542125, "grad_norm": 0.9296875, "learning_rate": 8.254111838124016e-05, "loss": 0.1274, "step": 14912 }, { "epoch": 10.925274725274726, "grad_norm": 1.296875, "learning_rate": 8.25385065971649e-05, "loss": 0.159, "step": 14913 }, { "epoch": 10.926007326007326, "grad_norm": 1.0546875, "learning_rate": 8.253589465958265e-05, "loss": 0.1529, "step": 14914 }, { "epoch": 10.926739926739927, "grad_norm": 1.6953125, "learning_rate": 8.253328256850588e-05, "loss": 0.1372, "step": 14915 }, { "epoch": 10.927472527472528, "grad_norm": 0.828125, "learning_rate": 8.253067032394719e-05, "loss": 0.1478, "step": 14916 }, { "epoch": 10.928205128205128, "grad_norm": 1.1796875, "learning_rate": 8.252805792591898e-05, "loss": 0.1169, "step": 14917 }, { "epoch": 10.92893772893773, "grad_norm": 1.125, "learning_rate": 8.252544537443388e-05, "loss": 0.1314, "step": 14918 }, { "epoch": 10.92967032967033, "grad_norm": 0.8125, "learning_rate": 8.252283266950432e-05, "loss": 0.1366, "step": 14919 }, { "epoch": 10.93040293040293, "grad_norm": 1.2890625, "learning_rate": 8.252021981114287e-05, "loss": 0.1653, "step": 14920 }, { "epoch": 10.931135531135531, "grad_norm": 0.796875, "learning_rate": 8.251760679936203e-05, "loss": 0.12, "step": 14921 }, { "epoch": 10.931868131868132, "grad_norm": 1.046875, "learning_rate": 8.251499363417436e-05, "loss": 0.1052, "step": 14922 }, { "epoch": 10.932600732600733, "grad_norm": 1.1015625, "learning_rate": 8.25123803155923e-05, "loss": 0.1324, "step": 14923 }, { "epoch": 10.933333333333334, "grad_norm": 1.1796875, "learning_rate": 8.250976684362845e-05, "loss": 0.1443, "step": 14924 }, { "epoch": 10.934065934065934, "grad_norm": 0.734375, "learning_rate": 8.250715321829527e-05, "loss": 0.139, "step": 14925 }, { "epoch": 10.934798534798535, "grad_norm": 1.140625, "learning_rate": 8.250453943960534e-05, "loss": 0.1504, "step": 14926 }, { "epoch": 10.935531135531136, "grad_norm": 1.2890625, "learning_rate": 8.250192550757115e-05, "loss": 0.1499, "step": 14927 }, { "epoch": 10.936263736263736, "grad_norm": 1.0546875, "learning_rate": 8.249931142220523e-05, "loss": 0.1241, "step": 14928 }, { "epoch": 10.936996336996337, "grad_norm": 1.40625, "learning_rate": 8.249669718352008e-05, "loss": 0.1401, "step": 14929 }, { "epoch": 10.937728937728938, "grad_norm": 1.0859375, "learning_rate": 8.249408279152827e-05, "loss": 0.1216, "step": 14930 }, { "epoch": 10.938461538461539, "grad_norm": 1.0625, "learning_rate": 8.24914682462423e-05, "loss": 0.1371, "step": 14931 }, { "epoch": 10.93919413919414, "grad_norm": 0.8046875, "learning_rate": 8.24888535476747e-05, "loss": 0.1643, "step": 14932 }, { "epoch": 10.93992673992674, "grad_norm": 1.390625, "learning_rate": 8.248623869583802e-05, "loss": 0.1514, "step": 14933 }, { "epoch": 10.94065934065934, "grad_norm": 1.046875, "learning_rate": 8.248362369074476e-05, "loss": 0.1633, "step": 14934 }, { "epoch": 10.941391941391942, "grad_norm": 0.9765625, "learning_rate": 8.248100853240745e-05, "loss": 0.1339, "step": 14935 }, { "epoch": 10.942124542124542, "grad_norm": 0.90625, "learning_rate": 8.247839322083864e-05, "loss": 0.1397, "step": 14936 }, { "epoch": 10.942857142857143, "grad_norm": 0.97265625, "learning_rate": 8.247577775605085e-05, "loss": 0.1126, "step": 14937 }, { "epoch": 10.943589743589744, "grad_norm": 1.421875, "learning_rate": 8.24731621380566e-05, "loss": 0.1429, "step": 14938 }, { "epoch": 10.944322344322345, "grad_norm": 1.0234375, "learning_rate": 8.247054636686846e-05, "loss": 0.1268, "step": 14939 }, { "epoch": 10.945054945054945, "grad_norm": 1.0703125, "learning_rate": 8.24679304424989e-05, "loss": 0.1314, "step": 14940 }, { "epoch": 10.945787545787546, "grad_norm": 0.94140625, "learning_rate": 8.246531436496051e-05, "loss": 0.1651, "step": 14941 }, { "epoch": 10.946520146520147, "grad_norm": 1.015625, "learning_rate": 8.246269813426581e-05, "loss": 0.1193, "step": 14942 }, { "epoch": 10.947252747252747, "grad_norm": 1.3203125, "learning_rate": 8.246008175042732e-05, "loss": 0.1818, "step": 14943 }, { "epoch": 10.947985347985348, "grad_norm": 1.1953125, "learning_rate": 8.24574652134576e-05, "loss": 0.1139, "step": 14944 }, { "epoch": 10.948717948717949, "grad_norm": 2.125, "learning_rate": 8.245484852336917e-05, "loss": 0.1691, "step": 14945 }, { "epoch": 10.94945054945055, "grad_norm": 0.69140625, "learning_rate": 8.245223168017456e-05, "loss": 0.1156, "step": 14946 }, { "epoch": 10.95018315018315, "grad_norm": 0.921875, "learning_rate": 8.244961468388633e-05, "loss": 0.1186, "step": 14947 }, { "epoch": 10.950915750915751, "grad_norm": 0.58203125, "learning_rate": 8.2446997534517e-05, "loss": 0.1203, "step": 14948 }, { "epoch": 10.951648351648352, "grad_norm": 0.8828125, "learning_rate": 8.244438023207912e-05, "loss": 0.1335, "step": 14949 }, { "epoch": 10.952380952380953, "grad_norm": 1.0859375, "learning_rate": 8.244176277658524e-05, "loss": 0.1108, "step": 14950 }, { "epoch": 10.953113553113553, "grad_norm": 1.0703125, "learning_rate": 8.243914516804787e-05, "loss": 0.1174, "step": 14951 }, { "epoch": 10.953846153846154, "grad_norm": 0.828125, "learning_rate": 8.243652740647957e-05, "loss": 0.106, "step": 14952 }, { "epoch": 10.954578754578755, "grad_norm": 1.1875, "learning_rate": 8.243390949189289e-05, "loss": 0.1333, "step": 14953 }, { "epoch": 10.955311355311355, "grad_norm": 0.97265625, "learning_rate": 8.243129142430037e-05, "loss": 0.1288, "step": 14954 }, { "epoch": 10.956043956043956, "grad_norm": 1.4296875, "learning_rate": 8.242867320371455e-05, "loss": 0.1163, "step": 14955 }, { "epoch": 10.956776556776557, "grad_norm": 0.890625, "learning_rate": 8.242605483014798e-05, "loss": 0.175, "step": 14956 }, { "epoch": 10.957509157509158, "grad_norm": 1.21875, "learning_rate": 8.242343630361319e-05, "loss": 0.1421, "step": 14957 }, { "epoch": 10.958241758241758, "grad_norm": 0.625, "learning_rate": 8.242081762412273e-05, "loss": 0.098, "step": 14958 }, { "epoch": 10.95897435897436, "grad_norm": 0.578125, "learning_rate": 8.241819879168916e-05, "loss": 0.1321, "step": 14959 }, { "epoch": 10.95970695970696, "grad_norm": 0.80078125, "learning_rate": 8.241557980632504e-05, "loss": 0.1342, "step": 14960 }, { "epoch": 10.96043956043956, "grad_norm": 0.859375, "learning_rate": 8.241296066804289e-05, "loss": 0.1112, "step": 14961 }, { "epoch": 10.961172161172161, "grad_norm": 0.76953125, "learning_rate": 8.241034137685526e-05, "loss": 0.1059, "step": 14962 }, { "epoch": 10.961904761904762, "grad_norm": 1.0078125, "learning_rate": 8.240772193277472e-05, "loss": 0.1029, "step": 14963 }, { "epoch": 10.962637362637363, "grad_norm": 0.79296875, "learning_rate": 8.24051023358138e-05, "loss": 0.1046, "step": 14964 }, { "epoch": 10.963369963369964, "grad_norm": 1.6484375, "learning_rate": 8.240248258598506e-05, "loss": 0.1313, "step": 14965 }, { "epoch": 10.964102564102564, "grad_norm": 1.625, "learning_rate": 8.239986268330105e-05, "loss": 0.1845, "step": 14966 }, { "epoch": 10.964835164835165, "grad_norm": 1.5234375, "learning_rate": 8.239724262777433e-05, "loss": 0.1322, "step": 14967 }, { "epoch": 10.965567765567766, "grad_norm": 1.734375, "learning_rate": 8.239462241941745e-05, "loss": 0.1312, "step": 14968 }, { "epoch": 10.966300366300366, "grad_norm": 1.15625, "learning_rate": 8.239200205824297e-05, "loss": 0.1471, "step": 14969 }, { "epoch": 10.967032967032967, "grad_norm": 0.9140625, "learning_rate": 8.238938154426344e-05, "loss": 0.125, "step": 14970 }, { "epoch": 10.967765567765568, "grad_norm": 2.078125, "learning_rate": 8.238676087749142e-05, "loss": 0.1719, "step": 14971 }, { "epoch": 10.968498168498169, "grad_norm": 1.0390625, "learning_rate": 8.238414005793944e-05, "loss": 0.135, "step": 14972 }, { "epoch": 10.96923076923077, "grad_norm": 2.15625, "learning_rate": 8.238151908562008e-05, "loss": 0.1397, "step": 14973 }, { "epoch": 10.96996336996337, "grad_norm": 1.3203125, "learning_rate": 8.23788979605459e-05, "loss": 0.1364, "step": 14974 }, { "epoch": 10.97069597069597, "grad_norm": 1.4140625, "learning_rate": 8.237627668272947e-05, "loss": 0.1614, "step": 14975 }, { "epoch": 10.971428571428572, "grad_norm": 0.796875, "learning_rate": 8.237365525218333e-05, "loss": 0.1261, "step": 14976 }, { "epoch": 10.972161172161172, "grad_norm": 1.1171875, "learning_rate": 8.237103366892003e-05, "loss": 0.1167, "step": 14977 }, { "epoch": 10.972893772893773, "grad_norm": 1.984375, "learning_rate": 8.236841193295217e-05, "loss": 0.117, "step": 14978 }, { "epoch": 10.973626373626374, "grad_norm": 2.015625, "learning_rate": 8.236579004429226e-05, "loss": 0.1381, "step": 14979 }, { "epoch": 10.974358974358974, "grad_norm": 1.6640625, "learning_rate": 8.23631680029529e-05, "loss": 0.1227, "step": 14980 }, { "epoch": 10.975091575091575, "grad_norm": 1.1953125, "learning_rate": 8.236054580894663e-05, "loss": 0.1235, "step": 14981 }, { "epoch": 10.975824175824176, "grad_norm": 1.1171875, "learning_rate": 8.235792346228603e-05, "loss": 0.1647, "step": 14982 }, { "epoch": 10.976556776556777, "grad_norm": 0.6171875, "learning_rate": 8.235530096298367e-05, "loss": 0.1246, "step": 14983 }, { "epoch": 10.977289377289377, "grad_norm": 1.2109375, "learning_rate": 8.235267831105212e-05, "loss": 0.1299, "step": 14984 }, { "epoch": 10.978021978021978, "grad_norm": 1.390625, "learning_rate": 8.23500555065039e-05, "loss": 0.1365, "step": 14985 }, { "epoch": 10.978754578754579, "grad_norm": 1.2265625, "learning_rate": 8.234743254935162e-05, "loss": 0.1509, "step": 14986 }, { "epoch": 10.97948717948718, "grad_norm": 0.859375, "learning_rate": 8.234480943960784e-05, "loss": 0.1541, "step": 14987 }, { "epoch": 10.98021978021978, "grad_norm": 2.046875, "learning_rate": 8.234218617728511e-05, "loss": 0.1629, "step": 14988 }, { "epoch": 10.980952380952381, "grad_norm": 0.82421875, "learning_rate": 8.233956276239604e-05, "loss": 0.1579, "step": 14989 }, { "epoch": 10.981684981684982, "grad_norm": 0.70703125, "learning_rate": 8.233693919495315e-05, "loss": 0.104, "step": 14990 }, { "epoch": 10.982417582417582, "grad_norm": 0.98046875, "learning_rate": 8.233431547496905e-05, "loss": 0.1016, "step": 14991 }, { "epoch": 10.983150183150183, "grad_norm": 1.1328125, "learning_rate": 8.233169160245629e-05, "loss": 0.1005, "step": 14992 }, { "epoch": 10.983882783882784, "grad_norm": 1.25, "learning_rate": 8.232906757742742e-05, "loss": 0.145, "step": 14993 }, { "epoch": 10.984615384615385, "grad_norm": 1.15625, "learning_rate": 8.232644339989509e-05, "loss": 0.128, "step": 14994 }, { "epoch": 10.985347985347985, "grad_norm": 0.71875, "learning_rate": 8.232381906987177e-05, "loss": 0.1386, "step": 14995 }, { "epoch": 10.986080586080586, "grad_norm": 1.1484375, "learning_rate": 8.232119458737012e-05, "loss": 0.142, "step": 14996 }, { "epoch": 10.986813186813187, "grad_norm": 1.1953125, "learning_rate": 8.231856995240266e-05, "loss": 0.116, "step": 14997 }, { "epoch": 10.987545787545788, "grad_norm": 0.96875, "learning_rate": 8.231594516498199e-05, "loss": 0.1216, "step": 14998 }, { "epoch": 10.988278388278388, "grad_norm": 1.59375, "learning_rate": 8.231332022512068e-05, "loss": 0.1425, "step": 14999 }, { "epoch": 10.989010989010989, "grad_norm": 0.62890625, "learning_rate": 8.23106951328313e-05, "loss": 0.1458, "step": 15000 }, { "epoch": 10.98974358974359, "grad_norm": 0.6875, "learning_rate": 8.230806988812647e-05, "loss": 0.1255, "step": 15001 }, { "epoch": 10.99047619047619, "grad_norm": 0.98046875, "learning_rate": 8.230544449101872e-05, "loss": 0.1296, "step": 15002 }, { "epoch": 10.991208791208791, "grad_norm": 1.3359375, "learning_rate": 8.230281894152063e-05, "loss": 0.1355, "step": 15003 }, { "epoch": 10.991941391941392, "grad_norm": 1.203125, "learning_rate": 8.230019323964481e-05, "loss": 0.1385, "step": 15004 }, { "epoch": 10.992673992673993, "grad_norm": 1.5625, "learning_rate": 8.229756738540383e-05, "loss": 0.1641, "step": 15005 }, { "epoch": 10.993406593406593, "grad_norm": 0.98046875, "learning_rate": 8.229494137881025e-05, "loss": 0.1004, "step": 15006 }, { "epoch": 10.994139194139194, "grad_norm": 0.703125, "learning_rate": 8.229231521987668e-05, "loss": 0.1437, "step": 15007 }, { "epoch": 10.994871794871795, "grad_norm": 0.75390625, "learning_rate": 8.228968890861568e-05, "loss": 0.1259, "step": 15008 }, { "epoch": 10.995604395604396, "grad_norm": 1.1640625, "learning_rate": 8.228706244503985e-05, "loss": 0.1076, "step": 15009 }, { "epoch": 10.996336996336996, "grad_norm": 1.359375, "learning_rate": 8.228443582916178e-05, "loss": 0.1416, "step": 15010 }, { "epoch": 10.997069597069597, "grad_norm": 1.265625, "learning_rate": 8.228180906099403e-05, "loss": 0.162, "step": 15011 }, { "epoch": 10.997802197802198, "grad_norm": 0.8515625, "learning_rate": 8.227918214054919e-05, "loss": 0.1513, "step": 15012 }, { "epoch": 10.998534798534799, "grad_norm": 1.046875, "learning_rate": 8.227655506783987e-05, "loss": 0.1061, "step": 15013 }, { "epoch": 10.9992673992674, "grad_norm": 1.2109375, "learning_rate": 8.227392784287865e-05, "loss": 0.1134, "step": 15014 }, { "epoch": 11.0, "grad_norm": 0.9765625, "learning_rate": 8.22713004656781e-05, "loss": 0.1356, "step": 15015 }, { "epoch": 11.0007326007326, "grad_norm": 1.125, "learning_rate": 8.226867293625081e-05, "loss": 0.1082, "step": 15016 }, { "epoch": 11.001465201465201, "grad_norm": 0.9296875, "learning_rate": 8.226604525460939e-05, "loss": 0.1239, "step": 15017 }, { "epoch": 11.002197802197802, "grad_norm": 0.80078125, "learning_rate": 8.226341742076644e-05, "loss": 0.1072, "step": 15018 }, { "epoch": 11.002930402930403, "grad_norm": 0.7734375, "learning_rate": 8.226078943473449e-05, "loss": 0.1209, "step": 15019 }, { "epoch": 11.003663003663004, "grad_norm": 0.75, "learning_rate": 8.225816129652621e-05, "loss": 0.1232, "step": 15020 }, { "epoch": 11.004395604395604, "grad_norm": 0.8359375, "learning_rate": 8.225553300615412e-05, "loss": 0.1237, "step": 15021 }, { "epoch": 11.005128205128205, "grad_norm": 0.8125, "learning_rate": 8.225290456363088e-05, "loss": 0.1482, "step": 15022 }, { "epoch": 11.005860805860806, "grad_norm": 0.95703125, "learning_rate": 8.225027596896902e-05, "loss": 0.1189, "step": 15023 }, { "epoch": 11.006593406593407, "grad_norm": 0.8515625, "learning_rate": 8.224764722218116e-05, "loss": 0.1648, "step": 15024 }, { "epoch": 11.007326007326007, "grad_norm": 1.1484375, "learning_rate": 8.224501832327993e-05, "loss": 0.1241, "step": 15025 }, { "epoch": 11.008058608058608, "grad_norm": 0.8984375, "learning_rate": 8.224238927227788e-05, "loss": 0.1481, "step": 15026 }, { "epoch": 11.008791208791209, "grad_norm": 1.15625, "learning_rate": 8.223976006918764e-05, "loss": 0.1239, "step": 15027 }, { "epoch": 11.00952380952381, "grad_norm": 0.99609375, "learning_rate": 8.223713071402177e-05, "loss": 0.1076, "step": 15028 }, { "epoch": 11.01025641025641, "grad_norm": 0.78125, "learning_rate": 8.223450120679289e-05, "loss": 0.1292, "step": 15029 }, { "epoch": 11.010989010989011, "grad_norm": 1.1953125, "learning_rate": 8.223187154751361e-05, "loss": 0.1275, "step": 15030 }, { "epoch": 11.011721611721612, "grad_norm": 0.84765625, "learning_rate": 8.22292417361965e-05, "loss": 0.1595, "step": 15031 }, { "epoch": 11.012454212454212, "grad_norm": 0.6953125, "learning_rate": 8.222661177285417e-05, "loss": 0.1032, "step": 15032 }, { "epoch": 11.013186813186813, "grad_norm": 0.8828125, "learning_rate": 8.222398165749925e-05, "loss": 0.1289, "step": 15033 }, { "epoch": 11.013919413919414, "grad_norm": 1.0234375, "learning_rate": 8.22213513901443e-05, "loss": 0.1423, "step": 15034 }, { "epoch": 11.014652014652015, "grad_norm": 0.9140625, "learning_rate": 8.221872097080197e-05, "loss": 0.1256, "step": 15035 }, { "epoch": 11.015384615384615, "grad_norm": 1.234375, "learning_rate": 8.221609039948481e-05, "loss": 0.123, "step": 15036 }, { "epoch": 11.016117216117216, "grad_norm": 0.87890625, "learning_rate": 8.221345967620547e-05, "loss": 0.1163, "step": 15037 }, { "epoch": 11.016849816849817, "grad_norm": 1.1640625, "learning_rate": 8.221082880097651e-05, "loss": 0.1425, "step": 15038 }, { "epoch": 11.017582417582418, "grad_norm": 1.4453125, "learning_rate": 8.220819777381058e-05, "loss": 0.1269, "step": 15039 }, { "epoch": 11.018315018315018, "grad_norm": 1.1953125, "learning_rate": 8.220556659472027e-05, "loss": 0.1231, "step": 15040 }, { "epoch": 11.019047619047619, "grad_norm": 1.109375, "learning_rate": 8.220293526371817e-05, "loss": 0.1509, "step": 15041 }, { "epoch": 11.01978021978022, "grad_norm": 0.9609375, "learning_rate": 8.220030378081687e-05, "loss": 0.1101, "step": 15042 }, { "epoch": 11.02051282051282, "grad_norm": 1.0703125, "learning_rate": 8.219767214602906e-05, "loss": 0.1257, "step": 15043 }, { "epoch": 11.021245421245421, "grad_norm": 1.125, "learning_rate": 8.219504035936726e-05, "loss": 0.1224, "step": 15044 }, { "epoch": 11.021978021978022, "grad_norm": 1.8671875, "learning_rate": 8.219240842084413e-05, "loss": 0.1517, "step": 15045 }, { "epoch": 11.022710622710623, "grad_norm": 1.1640625, "learning_rate": 8.218977633047227e-05, "loss": 0.147, "step": 15046 }, { "epoch": 11.023443223443223, "grad_norm": 1.109375, "learning_rate": 8.218714408826429e-05, "loss": 0.1243, "step": 15047 }, { "epoch": 11.024175824175824, "grad_norm": 0.81640625, "learning_rate": 8.21845116942328e-05, "loss": 0.1168, "step": 15048 }, { "epoch": 11.024908424908425, "grad_norm": 1.578125, "learning_rate": 8.21818791483904e-05, "loss": 0.1187, "step": 15049 }, { "epoch": 11.025641025641026, "grad_norm": 0.77734375, "learning_rate": 8.217924645074974e-05, "loss": 0.0988, "step": 15050 }, { "epoch": 11.026373626373626, "grad_norm": 0.7421875, "learning_rate": 8.217661360132339e-05, "loss": 0.1113, "step": 15051 }, { "epoch": 11.027106227106227, "grad_norm": 1.0234375, "learning_rate": 8.217398060012398e-05, "loss": 0.1301, "step": 15052 }, { "epoch": 11.027838827838828, "grad_norm": 0.84375, "learning_rate": 8.217134744716416e-05, "loss": 0.1034, "step": 15053 }, { "epoch": 11.028571428571428, "grad_norm": 0.71875, "learning_rate": 8.21687141424565e-05, "loss": 0.1242, "step": 15054 }, { "epoch": 11.02930402930403, "grad_norm": 1.5625, "learning_rate": 8.216608068601363e-05, "loss": 0.1316, "step": 15055 }, { "epoch": 11.03003663003663, "grad_norm": 0.87109375, "learning_rate": 8.216344707784819e-05, "loss": 0.1047, "step": 15056 }, { "epoch": 11.03076923076923, "grad_norm": 1.2421875, "learning_rate": 8.216081331797277e-05, "loss": 0.1389, "step": 15057 }, { "epoch": 11.031501831501831, "grad_norm": 0.984375, "learning_rate": 8.21581794064e-05, "loss": 0.1474, "step": 15058 }, { "epoch": 11.032234432234432, "grad_norm": 0.9609375, "learning_rate": 8.21555453431425e-05, "loss": 0.1214, "step": 15059 }, { "epoch": 11.032967032967033, "grad_norm": 0.9921875, "learning_rate": 8.215291112821287e-05, "loss": 0.1244, "step": 15060 }, { "epoch": 11.033699633699634, "grad_norm": 1.1015625, "learning_rate": 8.215027676162377e-05, "loss": 0.1321, "step": 15061 }, { "epoch": 11.034432234432234, "grad_norm": 1.1328125, "learning_rate": 8.214764224338781e-05, "loss": 0.1294, "step": 15062 }, { "epoch": 11.035164835164835, "grad_norm": 1.328125, "learning_rate": 8.21450075735176e-05, "loss": 0.1219, "step": 15063 }, { "epoch": 11.035897435897436, "grad_norm": 1.015625, "learning_rate": 8.214237275202577e-05, "loss": 0.1384, "step": 15064 }, { "epoch": 11.036630036630036, "grad_norm": 0.90625, "learning_rate": 8.213973777892496e-05, "loss": 0.1365, "step": 15065 }, { "epoch": 11.037362637362637, "grad_norm": 0.98828125, "learning_rate": 8.213710265422775e-05, "loss": 0.1548, "step": 15066 }, { "epoch": 11.038095238095238, "grad_norm": 0.96484375, "learning_rate": 8.21344673779468e-05, "loss": 0.1359, "step": 15067 }, { "epoch": 11.038827838827839, "grad_norm": 1.5, "learning_rate": 8.213183195009476e-05, "loss": 0.1372, "step": 15068 }, { "epoch": 11.03956043956044, "grad_norm": 0.83984375, "learning_rate": 8.21291963706842e-05, "loss": 0.1161, "step": 15069 }, { "epoch": 11.04029304029304, "grad_norm": 1.3046875, "learning_rate": 8.212656063972779e-05, "loss": 0.1702, "step": 15070 }, { "epoch": 11.04102564102564, "grad_norm": 0.8203125, "learning_rate": 8.212392475723815e-05, "loss": 0.1277, "step": 15071 }, { "epoch": 11.041758241758242, "grad_norm": 0.9609375, "learning_rate": 8.21212887232279e-05, "loss": 0.1608, "step": 15072 }, { "epoch": 11.042490842490842, "grad_norm": 0.90625, "learning_rate": 8.211865253770967e-05, "loss": 0.1172, "step": 15073 }, { "epoch": 11.043223443223443, "grad_norm": 0.9296875, "learning_rate": 8.211601620069611e-05, "loss": 0.1238, "step": 15074 }, { "epoch": 11.043956043956044, "grad_norm": 1.234375, "learning_rate": 8.211337971219983e-05, "loss": 0.1294, "step": 15075 }, { "epoch": 11.044688644688645, "grad_norm": 0.75390625, "learning_rate": 8.211074307223346e-05, "loss": 0.0969, "step": 15076 }, { "epoch": 11.045421245421245, "grad_norm": 1.046875, "learning_rate": 8.210810628080965e-05, "loss": 0.1264, "step": 15077 }, { "epoch": 11.046153846153846, "grad_norm": 0.640625, "learning_rate": 8.210546933794102e-05, "loss": 0.1079, "step": 15078 }, { "epoch": 11.046886446886447, "grad_norm": 1.1484375, "learning_rate": 8.210283224364022e-05, "loss": 0.1371, "step": 15079 }, { "epoch": 11.047619047619047, "grad_norm": 0.68359375, "learning_rate": 8.210019499791989e-05, "loss": 0.1249, "step": 15080 }, { "epoch": 11.048351648351648, "grad_norm": 1.53125, "learning_rate": 8.209755760079262e-05, "loss": 0.1441, "step": 15081 }, { "epoch": 11.049084249084249, "grad_norm": 1.4921875, "learning_rate": 8.20949200522711e-05, "loss": 0.1101, "step": 15082 }, { "epoch": 11.04981684981685, "grad_norm": 0.828125, "learning_rate": 8.209228235236794e-05, "loss": 0.1383, "step": 15083 }, { "epoch": 11.05054945054945, "grad_norm": 1.046875, "learning_rate": 8.208964450109577e-05, "loss": 0.1533, "step": 15084 }, { "epoch": 11.051282051282051, "grad_norm": 0.83203125, "learning_rate": 8.208700649846728e-05, "loss": 0.1249, "step": 15085 }, { "epoch": 11.052014652014652, "grad_norm": 0.96484375, "learning_rate": 8.208436834449505e-05, "loss": 0.1262, "step": 15086 }, { "epoch": 11.052747252747253, "grad_norm": 0.90625, "learning_rate": 8.208173003919174e-05, "loss": 0.1483, "step": 15087 }, { "epoch": 11.053479853479853, "grad_norm": 1.078125, "learning_rate": 8.207909158257e-05, "loss": 0.1016, "step": 15088 }, { "epoch": 11.054212454212454, "grad_norm": 1.640625, "learning_rate": 8.207645297464246e-05, "loss": 0.1373, "step": 15089 }, { "epoch": 11.054945054945055, "grad_norm": 0.9453125, "learning_rate": 8.207381421542177e-05, "loss": 0.1483, "step": 15090 }, { "epoch": 11.055677655677655, "grad_norm": 1.0234375, "learning_rate": 8.207117530492058e-05, "loss": 0.1148, "step": 15091 }, { "epoch": 11.056410256410256, "grad_norm": 1.375, "learning_rate": 8.20685362431515e-05, "loss": 0.1495, "step": 15092 }, { "epoch": 11.057142857142857, "grad_norm": 1.1015625, "learning_rate": 8.206589703012724e-05, "loss": 0.1067, "step": 15093 }, { "epoch": 11.057875457875458, "grad_norm": 1.0625, "learning_rate": 8.206325766586037e-05, "loss": 0.1488, "step": 15094 }, { "epoch": 11.058608058608058, "grad_norm": 1.1171875, "learning_rate": 8.206061815036358e-05, "loss": 0.1341, "step": 15095 }, { "epoch": 11.05934065934066, "grad_norm": 0.91015625, "learning_rate": 8.205797848364952e-05, "loss": 0.1147, "step": 15096 }, { "epoch": 11.06007326007326, "grad_norm": 0.671875, "learning_rate": 8.20553386657308e-05, "loss": 0.1409, "step": 15097 }, { "epoch": 11.06080586080586, "grad_norm": 1.1484375, "learning_rate": 8.205269869662013e-05, "loss": 0.1385, "step": 15098 }, { "epoch": 11.061538461538461, "grad_norm": 1.1640625, "learning_rate": 8.205005857633008e-05, "loss": 0.1028, "step": 15099 }, { "epoch": 11.062271062271062, "grad_norm": 0.69921875, "learning_rate": 8.204741830487337e-05, "loss": 0.1522, "step": 15100 }, { "epoch": 11.063003663003663, "grad_norm": 1.1171875, "learning_rate": 8.204477788226261e-05, "loss": 0.1248, "step": 15101 }, { "epoch": 11.063736263736264, "grad_norm": 0.9453125, "learning_rate": 8.204213730851047e-05, "loss": 0.1026, "step": 15102 }, { "epoch": 11.064468864468864, "grad_norm": 1.0546875, "learning_rate": 8.20394965836296e-05, "loss": 0.1392, "step": 15103 }, { "epoch": 11.065201465201465, "grad_norm": 0.66796875, "learning_rate": 8.203685570763265e-05, "loss": 0.1177, "step": 15104 }, { "epoch": 11.065934065934066, "grad_norm": 1.328125, "learning_rate": 8.203421468053226e-05, "loss": 0.1514, "step": 15105 }, { "epoch": 11.066666666666666, "grad_norm": 1.265625, "learning_rate": 8.203157350234109e-05, "loss": 0.1659, "step": 15106 }, { "epoch": 11.067399267399267, "grad_norm": 1.3671875, "learning_rate": 8.202893217307183e-05, "loss": 0.156, "step": 15107 }, { "epoch": 11.068131868131868, "grad_norm": 1.2578125, "learning_rate": 8.202629069273707e-05, "loss": 0.1521, "step": 15108 }, { "epoch": 11.068864468864469, "grad_norm": 1.09375, "learning_rate": 8.202364906134953e-05, "loss": 0.1208, "step": 15109 }, { "epoch": 11.06959706959707, "grad_norm": 1.03125, "learning_rate": 8.202100727892183e-05, "loss": 0.1361, "step": 15110 }, { "epoch": 11.07032967032967, "grad_norm": 0.8359375, "learning_rate": 8.201836534546663e-05, "loss": 0.1364, "step": 15111 }, { "epoch": 11.07106227106227, "grad_norm": 1.1171875, "learning_rate": 8.201572326099659e-05, "loss": 0.1161, "step": 15112 }, { "epoch": 11.071794871794872, "grad_norm": 1.375, "learning_rate": 8.20130810255244e-05, "loss": 0.1387, "step": 15113 }, { "epoch": 11.072527472527472, "grad_norm": 1.0859375, "learning_rate": 8.201043863906266e-05, "loss": 0.1886, "step": 15114 }, { "epoch": 11.073260073260073, "grad_norm": 1.234375, "learning_rate": 8.200779610162409e-05, "loss": 0.1512, "step": 15115 }, { "epoch": 11.073992673992674, "grad_norm": 1.3828125, "learning_rate": 8.200515341322132e-05, "loss": 0.1478, "step": 15116 }, { "epoch": 11.074725274725274, "grad_norm": 1.2421875, "learning_rate": 8.2002510573867e-05, "loss": 0.1169, "step": 15117 }, { "epoch": 11.075457875457875, "grad_norm": 0.9609375, "learning_rate": 8.199986758357382e-05, "loss": 0.1114, "step": 15118 }, { "epoch": 11.076190476190476, "grad_norm": 1.2265625, "learning_rate": 8.199722444235445e-05, "loss": 0.1341, "step": 15119 }, { "epoch": 11.076923076923077, "grad_norm": 0.59765625, "learning_rate": 8.199458115022152e-05, "loss": 0.1159, "step": 15120 }, { "epoch": 11.077655677655677, "grad_norm": 0.8671875, "learning_rate": 8.199193770718772e-05, "loss": 0.1265, "step": 15121 }, { "epoch": 11.078388278388278, "grad_norm": 1.0078125, "learning_rate": 8.19892941132657e-05, "loss": 0.131, "step": 15122 }, { "epoch": 11.079120879120879, "grad_norm": 1.4765625, "learning_rate": 8.198665036846814e-05, "loss": 0.1778, "step": 15123 }, { "epoch": 11.07985347985348, "grad_norm": 0.73046875, "learning_rate": 8.19840064728077e-05, "loss": 0.1254, "step": 15124 }, { "epoch": 11.08058608058608, "grad_norm": 0.796875, "learning_rate": 8.198136242629706e-05, "loss": 0.1051, "step": 15125 }, { "epoch": 11.081318681318681, "grad_norm": 0.6875, "learning_rate": 8.197871822894887e-05, "loss": 0.1335, "step": 15126 }, { "epoch": 11.082051282051282, "grad_norm": 0.7109375, "learning_rate": 8.197607388077581e-05, "loss": 0.1194, "step": 15127 }, { "epoch": 11.082783882783882, "grad_norm": 0.60546875, "learning_rate": 8.197342938179054e-05, "loss": 0.1226, "step": 15128 }, { "epoch": 11.083516483516483, "grad_norm": 1.09375, "learning_rate": 8.197078473200574e-05, "loss": 0.1251, "step": 15129 }, { "epoch": 11.084249084249084, "grad_norm": 1.703125, "learning_rate": 8.196813993143409e-05, "loss": 0.1576, "step": 15130 }, { "epoch": 11.084981684981685, "grad_norm": 1.0703125, "learning_rate": 8.196549498008824e-05, "loss": 0.1372, "step": 15131 }, { "epoch": 11.085714285714285, "grad_norm": 1.0546875, "learning_rate": 8.196284987798088e-05, "loss": 0.1416, "step": 15132 }, { "epoch": 11.086446886446886, "grad_norm": 0.96484375, "learning_rate": 8.196020462512467e-05, "loss": 0.1466, "step": 15133 }, { "epoch": 11.087179487179487, "grad_norm": 1.0390625, "learning_rate": 8.195755922153228e-05, "loss": 0.112, "step": 15134 }, { "epoch": 11.087912087912088, "grad_norm": 0.875, "learning_rate": 8.195491366721642e-05, "loss": 0.1172, "step": 15135 }, { "epoch": 11.088644688644688, "grad_norm": 1.171875, "learning_rate": 8.195226796218974e-05, "loss": 0.146, "step": 15136 }, { "epoch": 11.089377289377289, "grad_norm": 1.3125, "learning_rate": 8.194962210646492e-05, "loss": 0.1237, "step": 15137 }, { "epoch": 11.09010989010989, "grad_norm": 1.1171875, "learning_rate": 8.194697610005463e-05, "loss": 0.1338, "step": 15138 }, { "epoch": 11.09084249084249, "grad_norm": 1.2734375, "learning_rate": 8.194432994297156e-05, "loss": 0.1363, "step": 15139 }, { "epoch": 11.091575091575091, "grad_norm": 0.91015625, "learning_rate": 8.194168363522839e-05, "loss": 0.127, "step": 15140 }, { "epoch": 11.092307692307692, "grad_norm": 1.453125, "learning_rate": 8.193903717683777e-05, "loss": 0.1299, "step": 15141 }, { "epoch": 11.093040293040293, "grad_norm": 1.09375, "learning_rate": 8.193639056781241e-05, "loss": 0.1481, "step": 15142 }, { "epoch": 11.093772893772893, "grad_norm": 1.46875, "learning_rate": 8.193374380816499e-05, "loss": 0.1351, "step": 15143 }, { "epoch": 11.094505494505494, "grad_norm": 1.078125, "learning_rate": 8.193109689790818e-05, "loss": 0.138, "step": 15144 }, { "epoch": 11.095238095238095, "grad_norm": 0.94140625, "learning_rate": 8.192844983705466e-05, "loss": 0.1379, "step": 15145 }, { "epoch": 11.095970695970696, "grad_norm": 1.0859375, "learning_rate": 8.192580262561712e-05, "loss": 0.1273, "step": 15146 }, { "epoch": 11.096703296703296, "grad_norm": 1.9921875, "learning_rate": 8.192315526360825e-05, "loss": 0.14, "step": 15147 }, { "epoch": 11.097435897435897, "grad_norm": 0.8046875, "learning_rate": 8.192050775104073e-05, "loss": 0.1169, "step": 15148 }, { "epoch": 11.098168498168498, "grad_norm": 2.15625, "learning_rate": 8.191786008792724e-05, "loss": 0.1426, "step": 15149 }, { "epoch": 11.098901098901099, "grad_norm": 1.4453125, "learning_rate": 8.191521227428046e-05, "loss": 0.1294, "step": 15150 }, { "epoch": 11.0996336996337, "grad_norm": 0.8359375, "learning_rate": 8.19125643101131e-05, "loss": 0.1119, "step": 15151 }, { "epoch": 11.1003663003663, "grad_norm": 1.0078125, "learning_rate": 8.190991619543783e-05, "loss": 0.1155, "step": 15152 }, { "epoch": 11.1010989010989, "grad_norm": 1.5390625, "learning_rate": 8.190726793026734e-05, "loss": 0.1431, "step": 15153 }, { "epoch": 11.101831501831501, "grad_norm": 1.3515625, "learning_rate": 8.190461951461428e-05, "loss": 0.1211, "step": 15154 }, { "epoch": 11.102564102564102, "grad_norm": 0.7578125, "learning_rate": 8.190197094849144e-05, "loss": 0.1354, "step": 15155 }, { "epoch": 11.103296703296703, "grad_norm": 0.7890625, "learning_rate": 8.189932223191141e-05, "loss": 0.1101, "step": 15156 }, { "epoch": 11.104029304029304, "grad_norm": 0.953125, "learning_rate": 8.189667336488695e-05, "loss": 0.1145, "step": 15157 }, { "epoch": 11.104761904761904, "grad_norm": 1.1484375, "learning_rate": 8.189402434743069e-05, "loss": 0.1216, "step": 15158 }, { "epoch": 11.105494505494505, "grad_norm": 1.0546875, "learning_rate": 8.189137517955536e-05, "loss": 0.1249, "step": 15159 }, { "epoch": 11.106227106227106, "grad_norm": 1.515625, "learning_rate": 8.188872586127365e-05, "loss": 0.1556, "step": 15160 }, { "epoch": 11.106959706959707, "grad_norm": 0.7734375, "learning_rate": 8.188607639259826e-05, "loss": 0.1229, "step": 15161 }, { "epoch": 11.107692307692307, "grad_norm": 0.91015625, "learning_rate": 8.188342677354186e-05, "loss": 0.1289, "step": 15162 }, { "epoch": 11.108424908424908, "grad_norm": 1.0625, "learning_rate": 8.188077700411717e-05, "loss": 0.1354, "step": 15163 }, { "epoch": 11.109157509157509, "grad_norm": 1.75, "learning_rate": 8.187812708433686e-05, "loss": 0.1147, "step": 15164 }, { "epoch": 11.10989010989011, "grad_norm": 1.09375, "learning_rate": 8.187547701421368e-05, "loss": 0.1264, "step": 15165 }, { "epoch": 11.11062271062271, "grad_norm": 1.0859375, "learning_rate": 8.187282679376026e-05, "loss": 0.1259, "step": 15166 }, { "epoch": 11.111355311355311, "grad_norm": 0.921875, "learning_rate": 8.187017642298932e-05, "loss": 0.1366, "step": 15167 }, { "epoch": 11.112087912087912, "grad_norm": 1.234375, "learning_rate": 8.186752590191358e-05, "loss": 0.1284, "step": 15168 }, { "epoch": 11.112820512820512, "grad_norm": 0.96875, "learning_rate": 8.186487523054573e-05, "loss": 0.1225, "step": 15169 }, { "epoch": 11.113553113553113, "grad_norm": 1.0390625, "learning_rate": 8.186222440889846e-05, "loss": 0.1438, "step": 15170 }, { "epoch": 11.114285714285714, "grad_norm": 0.98046875, "learning_rate": 8.185957343698448e-05, "loss": 0.1257, "step": 15171 }, { "epoch": 11.115018315018315, "grad_norm": 1.2734375, "learning_rate": 8.185692231481649e-05, "loss": 0.1417, "step": 15172 }, { "epoch": 11.115750915750915, "grad_norm": 1.0703125, "learning_rate": 8.185427104240717e-05, "loss": 0.1319, "step": 15173 }, { "epoch": 11.116483516483516, "grad_norm": 0.8125, "learning_rate": 8.185161961976928e-05, "loss": 0.1174, "step": 15174 }, { "epoch": 11.117216117216117, "grad_norm": 1.15625, "learning_rate": 8.184896804691545e-05, "loss": 0.1137, "step": 15175 }, { "epoch": 11.117948717948718, "grad_norm": 1.359375, "learning_rate": 8.184631632385847e-05, "loss": 0.1324, "step": 15176 }, { "epoch": 11.118681318681318, "grad_norm": 1.4140625, "learning_rate": 8.184366445061097e-05, "loss": 0.1368, "step": 15177 }, { "epoch": 11.119413919413919, "grad_norm": 1.234375, "learning_rate": 8.184101242718567e-05, "loss": 0.1163, "step": 15178 }, { "epoch": 11.12014652014652, "grad_norm": 1.015625, "learning_rate": 8.183836025359532e-05, "loss": 0.113, "step": 15179 }, { "epoch": 11.12087912087912, "grad_norm": 0.9609375, "learning_rate": 8.183570792985258e-05, "loss": 0.1368, "step": 15180 }, { "epoch": 11.121611721611721, "grad_norm": 1.3359375, "learning_rate": 8.183305545597019e-05, "loss": 0.121, "step": 15181 }, { "epoch": 11.122344322344322, "grad_norm": 1.2421875, "learning_rate": 8.183040283196084e-05, "loss": 0.0995, "step": 15182 }, { "epoch": 11.123076923076923, "grad_norm": 0.9375, "learning_rate": 8.182775005783724e-05, "loss": 0.1147, "step": 15183 }, { "epoch": 11.123809523809523, "grad_norm": 1.765625, "learning_rate": 8.18250971336121e-05, "loss": 0.1565, "step": 15184 }, { "epoch": 11.124542124542124, "grad_norm": 1.140625, "learning_rate": 8.182244405929814e-05, "loss": 0.1278, "step": 15185 }, { "epoch": 11.125274725274725, "grad_norm": 1.6796875, "learning_rate": 8.181979083490807e-05, "loss": 0.0782, "step": 15186 }, { "epoch": 11.126007326007326, "grad_norm": 1.1640625, "learning_rate": 8.181713746045462e-05, "loss": 0.1216, "step": 15187 }, { "epoch": 11.126739926739926, "grad_norm": 1.1171875, "learning_rate": 8.181448393595045e-05, "loss": 0.1645, "step": 15188 }, { "epoch": 11.127472527472527, "grad_norm": 1.0859375, "learning_rate": 8.181183026140832e-05, "loss": 0.1629, "step": 15189 }, { "epoch": 11.128205128205128, "grad_norm": 1.203125, "learning_rate": 8.180917643684094e-05, "loss": 0.1174, "step": 15190 }, { "epoch": 11.128937728937728, "grad_norm": 1.40625, "learning_rate": 8.180652246226103e-05, "loss": 0.1123, "step": 15191 }, { "epoch": 11.12967032967033, "grad_norm": 1.6015625, "learning_rate": 8.180386833768126e-05, "loss": 0.1267, "step": 15192 }, { "epoch": 11.13040293040293, "grad_norm": 1.3984375, "learning_rate": 8.18012140631144e-05, "loss": 0.1389, "step": 15193 }, { "epoch": 11.13113553113553, "grad_norm": 1.234375, "learning_rate": 8.179855963857316e-05, "loss": 0.1487, "step": 15194 }, { "epoch": 11.131868131868131, "grad_norm": 1.109375, "learning_rate": 8.179590506407024e-05, "loss": 0.1456, "step": 15195 }, { "epoch": 11.132600732600732, "grad_norm": 1.578125, "learning_rate": 8.179325033961835e-05, "loss": 0.1424, "step": 15196 }, { "epoch": 11.133333333333333, "grad_norm": 1.1796875, "learning_rate": 8.179059546523023e-05, "loss": 0.1634, "step": 15197 }, { "epoch": 11.134065934065934, "grad_norm": 1.5390625, "learning_rate": 8.17879404409186e-05, "loss": 0.1521, "step": 15198 }, { "epoch": 11.134798534798534, "grad_norm": 1.3359375, "learning_rate": 8.178528526669616e-05, "loss": 0.1142, "step": 15199 }, { "epoch": 11.135531135531135, "grad_norm": 0.8671875, "learning_rate": 8.178262994257567e-05, "loss": 0.1197, "step": 15200 }, { "epoch": 11.136263736263736, "grad_norm": 1.390625, "learning_rate": 8.17799744685698e-05, "loss": 0.1456, "step": 15201 }, { "epoch": 11.136996336996337, "grad_norm": 1.4765625, "learning_rate": 8.177731884469134e-05, "loss": 0.1019, "step": 15202 }, { "epoch": 11.137728937728937, "grad_norm": 1.09375, "learning_rate": 8.177466307095296e-05, "loss": 0.1231, "step": 15203 }, { "epoch": 11.138461538461538, "grad_norm": 1.0390625, "learning_rate": 8.177200714736742e-05, "loss": 0.1089, "step": 15204 }, { "epoch": 11.139194139194139, "grad_norm": 0.9375, "learning_rate": 8.17693510739474e-05, "loss": 0.125, "step": 15205 }, { "epoch": 11.13992673992674, "grad_norm": 1.34375, "learning_rate": 8.176669485070568e-05, "loss": 0.1296, "step": 15206 }, { "epoch": 11.14065934065934, "grad_norm": 0.84765625, "learning_rate": 8.176403847765494e-05, "loss": 0.1343, "step": 15207 }, { "epoch": 11.141391941391941, "grad_norm": 0.76953125, "learning_rate": 8.176138195480793e-05, "loss": 0.1258, "step": 15208 }, { "epoch": 11.142124542124542, "grad_norm": 0.87890625, "learning_rate": 8.175872528217739e-05, "loss": 0.0938, "step": 15209 }, { "epoch": 11.142857142857142, "grad_norm": 1.640625, "learning_rate": 8.175606845977602e-05, "loss": 0.1084, "step": 15210 }, { "epoch": 11.143589743589743, "grad_norm": 1.3984375, "learning_rate": 8.175341148761657e-05, "loss": 0.1591, "step": 15211 }, { "epoch": 11.144322344322344, "grad_norm": 1.4921875, "learning_rate": 8.175075436571178e-05, "loss": 0.1359, "step": 15212 }, { "epoch": 11.145054945054945, "grad_norm": 1.0703125, "learning_rate": 8.174809709407436e-05, "loss": 0.1228, "step": 15213 }, { "epoch": 11.145787545787545, "grad_norm": 1.4375, "learning_rate": 8.174543967271704e-05, "loss": 0.152, "step": 15214 }, { "epoch": 11.146520146520146, "grad_norm": 1.484375, "learning_rate": 8.174278210165257e-05, "loss": 0.1346, "step": 15215 }, { "epoch": 11.147252747252747, "grad_norm": 1.234375, "learning_rate": 8.174012438089367e-05, "loss": 0.1538, "step": 15216 }, { "epoch": 11.147985347985347, "grad_norm": 1.6328125, "learning_rate": 8.173746651045308e-05, "loss": 0.1247, "step": 15217 }, { "epoch": 11.148717948717948, "grad_norm": 1.203125, "learning_rate": 8.173480849034354e-05, "loss": 0.1105, "step": 15218 }, { "epoch": 11.149450549450549, "grad_norm": 0.984375, "learning_rate": 8.173215032057776e-05, "loss": 0.1205, "step": 15219 }, { "epoch": 11.15018315018315, "grad_norm": 0.87109375, "learning_rate": 8.17294920011685e-05, "loss": 0.1442, "step": 15220 }, { "epoch": 11.15091575091575, "grad_norm": 1.3515625, "learning_rate": 8.172683353212852e-05, "loss": 0.127, "step": 15221 }, { "epoch": 11.151648351648351, "grad_norm": 0.98046875, "learning_rate": 8.17241749134705e-05, "loss": 0.1408, "step": 15222 }, { "epoch": 11.152380952380952, "grad_norm": 1.1796875, "learning_rate": 8.17215161452072e-05, "loss": 0.1701, "step": 15223 }, { "epoch": 11.153113553113553, "grad_norm": 0.55078125, "learning_rate": 8.171885722735139e-05, "loss": 0.1197, "step": 15224 }, { "epoch": 11.153846153846153, "grad_norm": 0.84765625, "learning_rate": 8.171619815991579e-05, "loss": 0.1283, "step": 15225 }, { "epoch": 11.154578754578754, "grad_norm": 1.1875, "learning_rate": 8.171353894291312e-05, "loss": 0.1214, "step": 15226 }, { "epoch": 11.155311355311355, "grad_norm": 1.578125, "learning_rate": 8.171087957635614e-05, "loss": 0.1435, "step": 15227 }, { "epoch": 11.156043956043955, "grad_norm": 1.8046875, "learning_rate": 8.17082200602576e-05, "loss": 0.1214, "step": 15228 }, { "epoch": 11.156776556776556, "grad_norm": 1.0546875, "learning_rate": 8.170556039463023e-05, "loss": 0.1341, "step": 15229 }, { "epoch": 11.157509157509157, "grad_norm": 1.15625, "learning_rate": 8.170290057948677e-05, "loss": 0.1323, "step": 15230 }, { "epoch": 11.158241758241758, "grad_norm": 0.98828125, "learning_rate": 8.170024061483998e-05, "loss": 0.1392, "step": 15231 }, { "epoch": 11.158974358974358, "grad_norm": 1.1953125, "learning_rate": 8.169758050070258e-05, "loss": 0.1309, "step": 15232 }, { "epoch": 11.15970695970696, "grad_norm": 1.5390625, "learning_rate": 8.169492023708735e-05, "loss": 0.1229, "step": 15233 }, { "epoch": 11.16043956043956, "grad_norm": 1.1953125, "learning_rate": 8.1692259824007e-05, "loss": 0.1195, "step": 15234 }, { "epoch": 11.16117216117216, "grad_norm": 1.4296875, "learning_rate": 8.168959926147429e-05, "loss": 0.1256, "step": 15235 }, { "epoch": 11.161904761904761, "grad_norm": 1.234375, "learning_rate": 8.1686938549502e-05, "loss": 0.13, "step": 15236 }, { "epoch": 11.162637362637362, "grad_norm": 1.078125, "learning_rate": 8.16842776881028e-05, "loss": 0.157, "step": 15237 }, { "epoch": 11.163369963369963, "grad_norm": 1.265625, "learning_rate": 8.168161667728953e-05, "loss": 0.123, "step": 15238 }, { "epoch": 11.164102564102564, "grad_norm": 0.9375, "learning_rate": 8.167895551707489e-05, "loss": 0.1398, "step": 15239 }, { "epoch": 11.164835164835164, "grad_norm": 0.86328125, "learning_rate": 8.167629420747163e-05, "loss": 0.1156, "step": 15240 }, { "epoch": 11.165567765567765, "grad_norm": 1.21875, "learning_rate": 8.16736327484925e-05, "loss": 0.1583, "step": 15241 }, { "epoch": 11.166300366300366, "grad_norm": 1.1328125, "learning_rate": 8.16709711401503e-05, "loss": 0.134, "step": 15242 }, { "epoch": 11.167032967032966, "grad_norm": 0.86328125, "learning_rate": 8.166830938245771e-05, "loss": 0.108, "step": 15243 }, { "epoch": 11.167765567765567, "grad_norm": 1.9453125, "learning_rate": 8.166564747542753e-05, "loss": 0.1682, "step": 15244 }, { "epoch": 11.168498168498168, "grad_norm": 1.4921875, "learning_rate": 8.16629854190725e-05, "loss": 0.1323, "step": 15245 }, { "epoch": 11.169230769230769, "grad_norm": 0.9453125, "learning_rate": 8.166032321340538e-05, "loss": 0.1182, "step": 15246 }, { "epoch": 11.16996336996337, "grad_norm": 0.9609375, "learning_rate": 8.165766085843892e-05, "loss": 0.1064, "step": 15247 }, { "epoch": 11.17069597069597, "grad_norm": 0.8125, "learning_rate": 8.165499835418588e-05, "loss": 0.1072, "step": 15248 }, { "epoch": 11.17142857142857, "grad_norm": 1.0859375, "learning_rate": 8.165233570065901e-05, "loss": 0.1044, "step": 15249 }, { "epoch": 11.172161172161172, "grad_norm": 1.9375, "learning_rate": 8.164967289787107e-05, "loss": 0.1576, "step": 15250 }, { "epoch": 11.172893772893772, "grad_norm": 1.65625, "learning_rate": 8.164700994583485e-05, "loss": 0.1148, "step": 15251 }, { "epoch": 11.173626373626373, "grad_norm": 0.7734375, "learning_rate": 8.164434684456305e-05, "loss": 0.1245, "step": 15252 }, { "epoch": 11.174358974358974, "grad_norm": 1.1875, "learning_rate": 8.164168359406846e-05, "loss": 0.136, "step": 15253 }, { "epoch": 11.175091575091574, "grad_norm": 1.3671875, "learning_rate": 8.163902019436385e-05, "loss": 0.1488, "step": 15254 }, { "epoch": 11.175824175824175, "grad_norm": 1.390625, "learning_rate": 8.163635664546196e-05, "loss": 0.1288, "step": 15255 }, { "epoch": 11.176556776556776, "grad_norm": 0.85546875, "learning_rate": 8.163369294737558e-05, "loss": 0.1118, "step": 15256 }, { "epoch": 11.177289377289377, "grad_norm": 0.84375, "learning_rate": 8.163102910011746e-05, "loss": 0.1309, "step": 15257 }, { "epoch": 11.178021978021977, "grad_norm": 0.6796875, "learning_rate": 8.162836510370034e-05, "loss": 0.1257, "step": 15258 }, { "epoch": 11.178754578754578, "grad_norm": 1.1875, "learning_rate": 8.162570095813703e-05, "loss": 0.1509, "step": 15259 }, { "epoch": 11.179487179487179, "grad_norm": 1.0, "learning_rate": 8.162303666344026e-05, "loss": 0.1428, "step": 15260 }, { "epoch": 11.18021978021978, "grad_norm": 1.109375, "learning_rate": 8.162037221962278e-05, "loss": 0.1269, "step": 15261 }, { "epoch": 11.18095238095238, "grad_norm": 1.4296875, "learning_rate": 8.16177076266974e-05, "loss": 0.1366, "step": 15262 }, { "epoch": 11.181684981684981, "grad_norm": 1.265625, "learning_rate": 8.161504288467686e-05, "loss": 0.1216, "step": 15263 }, { "epoch": 11.182417582417582, "grad_norm": 0.98046875, "learning_rate": 8.161237799357395e-05, "loss": 0.141, "step": 15264 }, { "epoch": 11.183150183150182, "grad_norm": 1.4375, "learning_rate": 8.160971295340142e-05, "loss": 0.1605, "step": 15265 }, { "epoch": 11.183882783882783, "grad_norm": 0.85546875, "learning_rate": 8.160704776417203e-05, "loss": 0.1199, "step": 15266 }, { "epoch": 11.184615384615384, "grad_norm": 0.81640625, "learning_rate": 8.160438242589858e-05, "loss": 0.1246, "step": 15267 }, { "epoch": 11.185347985347985, "grad_norm": 1.0078125, "learning_rate": 8.160171693859381e-05, "loss": 0.1162, "step": 15268 }, { "epoch": 11.186080586080585, "grad_norm": 1.0078125, "learning_rate": 8.159905130227051e-05, "loss": 0.1285, "step": 15269 }, { "epoch": 11.186813186813186, "grad_norm": 0.75, "learning_rate": 8.159638551694145e-05, "loss": 0.1297, "step": 15270 }, { "epoch": 11.187545787545787, "grad_norm": 1.328125, "learning_rate": 8.159371958261938e-05, "loss": 0.1303, "step": 15271 }, { "epoch": 11.188278388278388, "grad_norm": 0.5625, "learning_rate": 8.15910534993171e-05, "loss": 0.118, "step": 15272 }, { "epoch": 11.189010989010988, "grad_norm": 0.88671875, "learning_rate": 8.158838726704739e-05, "loss": 0.1463, "step": 15273 }, { "epoch": 11.189743589743589, "grad_norm": 1.0546875, "learning_rate": 8.158572088582299e-05, "loss": 0.1371, "step": 15274 }, { "epoch": 11.19047619047619, "grad_norm": 0.83203125, "learning_rate": 8.15830543556567e-05, "loss": 0.1125, "step": 15275 }, { "epoch": 11.19120879120879, "grad_norm": 1.015625, "learning_rate": 8.158038767656132e-05, "loss": 0.1349, "step": 15276 }, { "epoch": 11.191941391941391, "grad_norm": 0.75390625, "learning_rate": 8.157772084854959e-05, "loss": 0.1179, "step": 15277 }, { "epoch": 11.192673992673992, "grad_norm": 1.03125, "learning_rate": 8.157505387163428e-05, "loss": 0.1258, "step": 15278 }, { "epoch": 11.193406593406593, "grad_norm": 1.453125, "learning_rate": 8.157238674582819e-05, "loss": 0.1258, "step": 15279 }, { "epoch": 11.194139194139193, "grad_norm": 0.921875, "learning_rate": 8.15697194711441e-05, "loss": 0.1216, "step": 15280 }, { "epoch": 11.194871794871794, "grad_norm": 1.109375, "learning_rate": 8.156705204759478e-05, "loss": 0.1467, "step": 15281 }, { "epoch": 11.195604395604395, "grad_norm": 1.015625, "learning_rate": 8.156438447519301e-05, "loss": 0.1087, "step": 15282 }, { "epoch": 11.196336996336996, "grad_norm": 0.96484375, "learning_rate": 8.15617167539516e-05, "loss": 0.0899, "step": 15283 }, { "epoch": 11.197069597069596, "grad_norm": 0.6171875, "learning_rate": 8.155904888388328e-05, "loss": 0.1261, "step": 15284 }, { "epoch": 11.197802197802197, "grad_norm": 1.625, "learning_rate": 8.155638086500088e-05, "loss": 0.138, "step": 15285 }, { "epoch": 11.198534798534798, "grad_norm": 1.96875, "learning_rate": 8.155371269731715e-05, "loss": 0.1488, "step": 15286 }, { "epoch": 11.199267399267399, "grad_norm": 1.5546875, "learning_rate": 8.15510443808449e-05, "loss": 0.1237, "step": 15287 }, { "epoch": 11.2, "grad_norm": 0.91796875, "learning_rate": 8.154837591559688e-05, "loss": 0.1213, "step": 15288 }, { "epoch": 11.2007326007326, "grad_norm": 1.0546875, "learning_rate": 8.154570730158593e-05, "loss": 0.1743, "step": 15289 }, { "epoch": 11.2014652014652, "grad_norm": 0.8671875, "learning_rate": 8.154303853882479e-05, "loss": 0.1389, "step": 15290 }, { "epoch": 11.202197802197801, "grad_norm": 1.375, "learning_rate": 8.154036962732626e-05, "loss": 0.167, "step": 15291 }, { "epoch": 11.202930402930402, "grad_norm": 1.96875, "learning_rate": 8.153770056710312e-05, "loss": 0.1185, "step": 15292 }, { "epoch": 11.203663003663003, "grad_norm": 1.1875, "learning_rate": 8.153503135816819e-05, "loss": 0.0976, "step": 15293 }, { "epoch": 11.204395604395604, "grad_norm": 0.78515625, "learning_rate": 8.153236200053424e-05, "loss": 0.1228, "step": 15294 }, { "epoch": 11.205128205128204, "grad_norm": 1.7578125, "learning_rate": 8.152969249421403e-05, "loss": 0.1539, "step": 15295 }, { "epoch": 11.205860805860805, "grad_norm": 1.21875, "learning_rate": 8.152702283922039e-05, "loss": 0.1271, "step": 15296 }, { "epoch": 11.206593406593406, "grad_norm": 1.8984375, "learning_rate": 8.152435303556609e-05, "loss": 0.1338, "step": 15297 }, { "epoch": 11.207326007326007, "grad_norm": 1.53125, "learning_rate": 8.152168308326395e-05, "loss": 0.1219, "step": 15298 }, { "epoch": 11.208058608058607, "grad_norm": 1.40625, "learning_rate": 8.151901298232671e-05, "loss": 0.1452, "step": 15299 }, { "epoch": 11.208791208791208, "grad_norm": 1.2265625, "learning_rate": 8.151634273276724e-05, "loss": 0.1244, "step": 15300 }, { "epoch": 11.209523809523809, "grad_norm": 1.0703125, "learning_rate": 8.151367233459825e-05, "loss": 0.1172, "step": 15301 }, { "epoch": 11.21025641025641, "grad_norm": 1.0859375, "learning_rate": 8.151100178783259e-05, "loss": 0.1108, "step": 15302 }, { "epoch": 11.21098901098901, "grad_norm": 0.9453125, "learning_rate": 8.150833109248304e-05, "loss": 0.1257, "step": 15303 }, { "epoch": 11.211721611721611, "grad_norm": 1.65625, "learning_rate": 8.150566024856239e-05, "loss": 0.1622, "step": 15304 }, { "epoch": 11.212454212454212, "grad_norm": 1.3828125, "learning_rate": 8.150298925608346e-05, "loss": 0.1503, "step": 15305 }, { "epoch": 11.213186813186812, "grad_norm": 1.0078125, "learning_rate": 8.150031811505902e-05, "loss": 0.1237, "step": 15306 }, { "epoch": 11.213919413919413, "grad_norm": 1.53125, "learning_rate": 8.149764682550189e-05, "loss": 0.1185, "step": 15307 }, { "epoch": 11.214652014652014, "grad_norm": 0.828125, "learning_rate": 8.149497538742483e-05, "loss": 0.1165, "step": 15308 }, { "epoch": 11.215384615384615, "grad_norm": 1.4921875, "learning_rate": 8.14923038008407e-05, "loss": 0.1283, "step": 15309 }, { "epoch": 11.216117216117215, "grad_norm": 1.140625, "learning_rate": 8.148963206576226e-05, "loss": 0.1487, "step": 15310 }, { "epoch": 11.216849816849816, "grad_norm": 1.390625, "learning_rate": 8.148696018220231e-05, "loss": 0.1732, "step": 15311 }, { "epoch": 11.217582417582417, "grad_norm": 2.109375, "learning_rate": 8.148428815017367e-05, "loss": 0.1234, "step": 15312 }, { "epoch": 11.218315018315018, "grad_norm": 1.640625, "learning_rate": 8.148161596968912e-05, "loss": 0.1171, "step": 15313 }, { "epoch": 11.219047619047618, "grad_norm": 1.6875, "learning_rate": 8.14789436407615e-05, "loss": 0.1421, "step": 15314 }, { "epoch": 11.219780219780219, "grad_norm": 1.0703125, "learning_rate": 8.147627116340356e-05, "loss": 0.1132, "step": 15315 }, { "epoch": 11.22051282051282, "grad_norm": 1.1796875, "learning_rate": 8.147359853762818e-05, "loss": 0.1601, "step": 15316 }, { "epoch": 11.22124542124542, "grad_norm": 1.6640625, "learning_rate": 8.147092576344807e-05, "loss": 0.1442, "step": 15317 }, { "epoch": 11.221978021978021, "grad_norm": 1.625, "learning_rate": 8.146825284087612e-05, "loss": 0.1248, "step": 15318 }, { "epoch": 11.222710622710622, "grad_norm": 1.671875, "learning_rate": 8.146557976992511e-05, "loss": 0.1237, "step": 15319 }, { "epoch": 11.223443223443223, "grad_norm": 1.65625, "learning_rate": 8.146290655060782e-05, "loss": 0.1478, "step": 15320 }, { "epoch": 11.224175824175823, "grad_norm": 1.75, "learning_rate": 8.146023318293708e-05, "loss": 0.1475, "step": 15321 }, { "epoch": 11.224908424908424, "grad_norm": 1.3125, "learning_rate": 8.145755966692571e-05, "loss": 0.1584, "step": 15322 }, { "epoch": 11.225641025641025, "grad_norm": 1.1796875, "learning_rate": 8.14548860025865e-05, "loss": 0.1411, "step": 15323 }, { "epoch": 11.226373626373626, "grad_norm": 0.8125, "learning_rate": 8.145221218993228e-05, "loss": 0.1507, "step": 15324 }, { "epoch": 11.227106227106226, "grad_norm": 0.77734375, "learning_rate": 8.144953822897585e-05, "loss": 0.1428, "step": 15325 }, { "epoch": 11.227838827838827, "grad_norm": 1.5546875, "learning_rate": 8.144686411973003e-05, "loss": 0.1195, "step": 15326 }, { "epoch": 11.228571428571428, "grad_norm": 0.578125, "learning_rate": 8.144418986220762e-05, "loss": 0.1054, "step": 15327 }, { "epoch": 11.229304029304028, "grad_norm": 1.15625, "learning_rate": 8.144151545642141e-05, "loss": 0.1062, "step": 15328 }, { "epoch": 11.23003663003663, "grad_norm": 1.1640625, "learning_rate": 8.143884090238428e-05, "loss": 0.1358, "step": 15329 }, { "epoch": 11.23076923076923, "grad_norm": 1.1328125, "learning_rate": 8.1436166200109e-05, "loss": 0.0926, "step": 15330 }, { "epoch": 11.23150183150183, "grad_norm": 1.7265625, "learning_rate": 8.143349134960837e-05, "loss": 0.169, "step": 15331 }, { "epoch": 11.232234432234431, "grad_norm": 1.8671875, "learning_rate": 8.143081635089526e-05, "loss": 0.1415, "step": 15332 }, { "epoch": 11.232967032967032, "grad_norm": 1.3359375, "learning_rate": 8.142814120398243e-05, "loss": 0.1417, "step": 15333 }, { "epoch": 11.233699633699633, "grad_norm": 1.4609375, "learning_rate": 8.142546590888272e-05, "loss": 0.1077, "step": 15334 }, { "epoch": 11.234432234432234, "grad_norm": 1.3984375, "learning_rate": 8.142279046560896e-05, "loss": 0.1275, "step": 15335 }, { "epoch": 11.235164835164834, "grad_norm": 0.69140625, "learning_rate": 8.142011487417396e-05, "loss": 0.1098, "step": 15336 }, { "epoch": 11.235897435897435, "grad_norm": 1.0234375, "learning_rate": 8.141743913459054e-05, "loss": 0.1315, "step": 15337 }, { "epoch": 11.236630036630036, "grad_norm": 1.1953125, "learning_rate": 8.141476324687152e-05, "loss": 0.1231, "step": 15338 }, { "epoch": 11.237362637362637, "grad_norm": 1.8984375, "learning_rate": 8.141208721102972e-05, "loss": 0.1437, "step": 15339 }, { "epoch": 11.238095238095237, "grad_norm": 1.359375, "learning_rate": 8.140941102707797e-05, "loss": 0.1483, "step": 15340 }, { "epoch": 11.238827838827838, "grad_norm": 1.4453125, "learning_rate": 8.140673469502906e-05, "loss": 0.1252, "step": 15341 }, { "epoch": 11.239560439560439, "grad_norm": 1.25, "learning_rate": 8.140405821489586e-05, "loss": 0.0989, "step": 15342 }, { "epoch": 11.24029304029304, "grad_norm": 1.109375, "learning_rate": 8.140138158669118e-05, "loss": 0.1472, "step": 15343 }, { "epoch": 11.24102564102564, "grad_norm": 1.140625, "learning_rate": 8.13987048104278e-05, "loss": 0.1133, "step": 15344 }, { "epoch": 11.241758241758241, "grad_norm": 1.3984375, "learning_rate": 8.13960278861186e-05, "loss": 0.147, "step": 15345 }, { "epoch": 11.242490842490842, "grad_norm": 1.40625, "learning_rate": 8.139335081377641e-05, "loss": 0.141, "step": 15346 }, { "epoch": 11.243223443223442, "grad_norm": 1.0703125, "learning_rate": 8.1390673593414e-05, "loss": 0.1327, "step": 15347 }, { "epoch": 11.243956043956043, "grad_norm": 1.8515625, "learning_rate": 8.138799622504427e-05, "loss": 0.1289, "step": 15348 }, { "epoch": 11.244688644688644, "grad_norm": 0.921875, "learning_rate": 8.138531870867999e-05, "loss": 0.146, "step": 15349 }, { "epoch": 11.245421245421245, "grad_norm": 0.62109375, "learning_rate": 8.138264104433399e-05, "loss": 0.1207, "step": 15350 }, { "epoch": 11.246153846153845, "grad_norm": 1.7109375, "learning_rate": 8.137996323201915e-05, "loss": 0.1506, "step": 15351 }, { "epoch": 11.246886446886446, "grad_norm": 1.2734375, "learning_rate": 8.137728527174824e-05, "loss": 0.1477, "step": 15352 }, { "epoch": 11.247619047619047, "grad_norm": 0.72265625, "learning_rate": 8.137460716353416e-05, "loss": 0.1211, "step": 15353 }, { "epoch": 11.248351648351647, "grad_norm": 0.6796875, "learning_rate": 8.137192890738968e-05, "loss": 0.123, "step": 15354 }, { "epoch": 11.249084249084248, "grad_norm": 1.3984375, "learning_rate": 8.136925050332765e-05, "loss": 0.1394, "step": 15355 }, { "epoch": 11.249816849816849, "grad_norm": 0.96484375, "learning_rate": 8.136657195136091e-05, "loss": 0.1318, "step": 15356 }, { "epoch": 11.25054945054945, "grad_norm": 0.86328125, "learning_rate": 8.136389325150229e-05, "loss": 0.1498, "step": 15357 }, { "epoch": 11.25128205128205, "grad_norm": 1.34375, "learning_rate": 8.136121440376461e-05, "loss": 0.141, "step": 15358 }, { "epoch": 11.252014652014651, "grad_norm": 1.1953125, "learning_rate": 8.135853540816073e-05, "loss": 0.1529, "step": 15359 }, { "epoch": 11.252747252747252, "grad_norm": 1.5390625, "learning_rate": 8.13558562647035e-05, "loss": 0.1039, "step": 15360 }, { "epoch": 11.253479853479853, "grad_norm": 1.421875, "learning_rate": 8.13531769734057e-05, "loss": 0.1406, "step": 15361 }, { "epoch": 11.254212454212453, "grad_norm": 1.234375, "learning_rate": 8.135049753428023e-05, "loss": 0.1299, "step": 15362 }, { "epoch": 11.254945054945054, "grad_norm": 0.8671875, "learning_rate": 8.134781794733988e-05, "loss": 0.1239, "step": 15363 }, { "epoch": 11.255677655677655, "grad_norm": 0.9296875, "learning_rate": 8.134513821259751e-05, "loss": 0.1374, "step": 15364 }, { "epoch": 11.256410256410255, "grad_norm": 1.375, "learning_rate": 8.134245833006597e-05, "loss": 0.1373, "step": 15365 }, { "epoch": 11.257142857142856, "grad_norm": 1.25, "learning_rate": 8.133977829975806e-05, "loss": 0.1614, "step": 15366 }, { "epoch": 11.257875457875459, "grad_norm": 1.9921875, "learning_rate": 8.133709812168668e-05, "loss": 0.1097, "step": 15367 }, { "epoch": 11.25860805860806, "grad_norm": 2.03125, "learning_rate": 8.133441779586463e-05, "loss": 0.1402, "step": 15368 }, { "epoch": 11.25934065934066, "grad_norm": 1.921875, "learning_rate": 8.133173732230476e-05, "loss": 0.1505, "step": 15369 }, { "epoch": 11.260073260073261, "grad_norm": 1.453125, "learning_rate": 8.13290567010199e-05, "loss": 0.1331, "step": 15370 }, { "epoch": 11.260805860805862, "grad_norm": 0.90625, "learning_rate": 8.132637593202291e-05, "loss": 0.099, "step": 15371 }, { "epoch": 11.261538461538462, "grad_norm": 1.0078125, "learning_rate": 8.132369501532666e-05, "loss": 0.1238, "step": 15372 }, { "epoch": 11.262271062271063, "grad_norm": 0.78515625, "learning_rate": 8.132101395094396e-05, "loss": 0.131, "step": 15373 }, { "epoch": 11.263003663003664, "grad_norm": 2.09375, "learning_rate": 8.131833273888765e-05, "loss": 0.1381, "step": 15374 }, { "epoch": 11.263736263736265, "grad_norm": 1.6796875, "learning_rate": 8.13156513791706e-05, "loss": 0.1317, "step": 15375 }, { "epoch": 11.264468864468865, "grad_norm": 1.9296875, "learning_rate": 8.131296987180564e-05, "loss": 0.1614, "step": 15376 }, { "epoch": 11.265201465201466, "grad_norm": 1.71875, "learning_rate": 8.131028821680563e-05, "loss": 0.1284, "step": 15377 }, { "epoch": 11.265934065934067, "grad_norm": 1.984375, "learning_rate": 8.130760641418342e-05, "loss": 0.1485, "step": 15378 }, { "epoch": 11.266666666666667, "grad_norm": 1.15625, "learning_rate": 8.130492446395186e-05, "loss": 0.1394, "step": 15379 }, { "epoch": 11.267399267399268, "grad_norm": 1.1953125, "learning_rate": 8.130224236612378e-05, "loss": 0.1193, "step": 15380 }, { "epoch": 11.268131868131869, "grad_norm": 0.8828125, "learning_rate": 8.129956012071204e-05, "loss": 0.1192, "step": 15381 }, { "epoch": 11.26886446886447, "grad_norm": 0.8125, "learning_rate": 8.129687772772952e-05, "loss": 0.1188, "step": 15382 }, { "epoch": 11.26959706959707, "grad_norm": 1.0078125, "learning_rate": 8.129419518718903e-05, "loss": 0.1314, "step": 15383 }, { "epoch": 11.270329670329671, "grad_norm": 1.4765625, "learning_rate": 8.129151249910345e-05, "loss": 0.1297, "step": 15384 }, { "epoch": 11.271062271062272, "grad_norm": 1.203125, "learning_rate": 8.128882966348562e-05, "loss": 0.1139, "step": 15385 }, { "epoch": 11.271794871794873, "grad_norm": 1.2890625, "learning_rate": 8.12861466803484e-05, "loss": 0.1438, "step": 15386 }, { "epoch": 11.272527472527473, "grad_norm": 1.3046875, "learning_rate": 8.128346354970466e-05, "loss": 0.168, "step": 15387 }, { "epoch": 11.273260073260074, "grad_norm": 1.4140625, "learning_rate": 8.128078027156721e-05, "loss": 0.1843, "step": 15388 }, { "epoch": 11.273992673992675, "grad_norm": 1.2734375, "learning_rate": 8.127809684594897e-05, "loss": 0.1292, "step": 15389 }, { "epoch": 11.274725274725276, "grad_norm": 1.046875, "learning_rate": 8.127541327286276e-05, "loss": 0.1354, "step": 15390 }, { "epoch": 11.275457875457876, "grad_norm": 1.4921875, "learning_rate": 8.127272955232142e-05, "loss": 0.1427, "step": 15391 }, { "epoch": 11.276190476190477, "grad_norm": 1.4296875, "learning_rate": 8.127004568433784e-05, "loss": 0.1474, "step": 15392 }, { "epoch": 11.276923076923078, "grad_norm": 1.0078125, "learning_rate": 8.126736166892486e-05, "loss": 0.1421, "step": 15393 }, { "epoch": 11.277655677655678, "grad_norm": 0.93359375, "learning_rate": 8.126467750609537e-05, "loss": 0.1392, "step": 15394 }, { "epoch": 11.27838827838828, "grad_norm": 0.98046875, "learning_rate": 8.126199319586222e-05, "loss": 0.1186, "step": 15395 }, { "epoch": 11.27912087912088, "grad_norm": 0.796875, "learning_rate": 8.125930873823822e-05, "loss": 0.1253, "step": 15396 }, { "epoch": 11.27985347985348, "grad_norm": 1.5546875, "learning_rate": 8.125662413323631e-05, "loss": 0.1691, "step": 15397 }, { "epoch": 11.280586080586081, "grad_norm": 1.828125, "learning_rate": 8.125393938086932e-05, "loss": 0.1524, "step": 15398 }, { "epoch": 11.281318681318682, "grad_norm": 0.91015625, "learning_rate": 8.125125448115009e-05, "loss": 0.1045, "step": 15399 }, { "epoch": 11.282051282051283, "grad_norm": 1.0625, "learning_rate": 8.124856943409152e-05, "loss": 0.1349, "step": 15400 }, { "epoch": 11.282783882783884, "grad_norm": 1.2109375, "learning_rate": 8.124588423970644e-05, "loss": 0.1621, "step": 15401 }, { "epoch": 11.283516483516484, "grad_norm": 1.2109375, "learning_rate": 8.124319889800776e-05, "loss": 0.1204, "step": 15402 }, { "epoch": 11.284249084249085, "grad_norm": 0.8359375, "learning_rate": 8.12405134090083e-05, "loss": 0.1362, "step": 15403 }, { "epoch": 11.284981684981686, "grad_norm": 1.0234375, "learning_rate": 8.123782777272096e-05, "loss": 0.1242, "step": 15404 }, { "epoch": 11.285714285714286, "grad_norm": 0.9140625, "learning_rate": 8.123514198915858e-05, "loss": 0.1469, "step": 15405 }, { "epoch": 11.286446886446887, "grad_norm": 1.1875, "learning_rate": 8.123245605833407e-05, "loss": 0.1107, "step": 15406 }, { "epoch": 11.287179487179488, "grad_norm": 0.84375, "learning_rate": 8.122976998026026e-05, "loss": 0.1206, "step": 15407 }, { "epoch": 11.287912087912089, "grad_norm": 1.9453125, "learning_rate": 8.122708375495005e-05, "loss": 0.1417, "step": 15408 }, { "epoch": 11.28864468864469, "grad_norm": 1.4921875, "learning_rate": 8.122439738241627e-05, "loss": 0.1304, "step": 15409 }, { "epoch": 11.28937728937729, "grad_norm": 1.3203125, "learning_rate": 8.122171086267183e-05, "loss": 0.1349, "step": 15410 }, { "epoch": 11.29010989010989, "grad_norm": 1.5859375, "learning_rate": 8.121902419572959e-05, "loss": 0.158, "step": 15411 }, { "epoch": 11.290842490842492, "grad_norm": 0.96875, "learning_rate": 8.121633738160241e-05, "loss": 0.1272, "step": 15412 }, { "epoch": 11.291575091575092, "grad_norm": 1.75, "learning_rate": 8.121365042030318e-05, "loss": 0.2185, "step": 15413 }, { "epoch": 11.292307692307693, "grad_norm": 0.8984375, "learning_rate": 8.121096331184477e-05, "loss": 0.1108, "step": 15414 }, { "epoch": 11.293040293040294, "grad_norm": 2.0, "learning_rate": 8.120827605624007e-05, "loss": 0.2009, "step": 15415 }, { "epoch": 11.293772893772895, "grad_norm": 1.1484375, "learning_rate": 8.120558865350191e-05, "loss": 0.1408, "step": 15416 }, { "epoch": 11.294505494505495, "grad_norm": 0.7734375, "learning_rate": 8.12029011036432e-05, "loss": 0.1013, "step": 15417 }, { "epoch": 11.295238095238096, "grad_norm": 1.5078125, "learning_rate": 8.120021340667682e-05, "loss": 0.1476, "step": 15418 }, { "epoch": 11.295970695970697, "grad_norm": 0.9609375, "learning_rate": 8.119752556261563e-05, "loss": 0.1372, "step": 15419 }, { "epoch": 11.296703296703297, "grad_norm": 0.921875, "learning_rate": 8.119483757147254e-05, "loss": 0.1497, "step": 15420 }, { "epoch": 11.297435897435898, "grad_norm": 0.62890625, "learning_rate": 8.119214943326038e-05, "loss": 0.1095, "step": 15421 }, { "epoch": 11.298168498168499, "grad_norm": 1.109375, "learning_rate": 8.118946114799205e-05, "loss": 0.1394, "step": 15422 }, { "epoch": 11.2989010989011, "grad_norm": 0.84375, "learning_rate": 8.118677271568047e-05, "loss": 0.1235, "step": 15423 }, { "epoch": 11.2996336996337, "grad_norm": 1.28125, "learning_rate": 8.118408413633848e-05, "loss": 0.115, "step": 15424 }, { "epoch": 11.300366300366301, "grad_norm": 1.609375, "learning_rate": 8.118139540997895e-05, "loss": 0.1614, "step": 15425 }, { "epoch": 11.301098901098902, "grad_norm": 0.64453125, "learning_rate": 8.11787065366148e-05, "loss": 0.0917, "step": 15426 }, { "epoch": 11.301831501831503, "grad_norm": 1.2578125, "learning_rate": 8.11760175162589e-05, "loss": 0.1333, "step": 15427 }, { "epoch": 11.302564102564103, "grad_norm": 0.63671875, "learning_rate": 8.117332834892412e-05, "loss": 0.1166, "step": 15428 }, { "epoch": 11.303296703296704, "grad_norm": 0.98046875, "learning_rate": 8.117063903462335e-05, "loss": 0.1429, "step": 15429 }, { "epoch": 11.304029304029305, "grad_norm": 1.171875, "learning_rate": 8.116794957336947e-05, "loss": 0.109, "step": 15430 }, { "epoch": 11.304761904761905, "grad_norm": 1.171875, "learning_rate": 8.116525996517542e-05, "loss": 0.1381, "step": 15431 }, { "epoch": 11.305494505494506, "grad_norm": 1.4609375, "learning_rate": 8.116257021005399e-05, "loss": 0.1503, "step": 15432 }, { "epoch": 11.306227106227107, "grad_norm": 1.34375, "learning_rate": 8.115988030801814e-05, "loss": 0.1137, "step": 15433 }, { "epoch": 11.306959706959708, "grad_norm": 1.3046875, "learning_rate": 8.115719025908076e-05, "loss": 0.1385, "step": 15434 }, { "epoch": 11.307692307692308, "grad_norm": 1.3828125, "learning_rate": 8.11545000632547e-05, "loss": 0.1412, "step": 15435 }, { "epoch": 11.308424908424909, "grad_norm": 1.7421875, "learning_rate": 8.115180972055287e-05, "loss": 0.1548, "step": 15436 }, { "epoch": 11.30915750915751, "grad_norm": 1.25, "learning_rate": 8.114911923098815e-05, "loss": 0.1062, "step": 15437 }, { "epoch": 11.30989010989011, "grad_norm": 0.9296875, "learning_rate": 8.114642859457343e-05, "loss": 0.1279, "step": 15438 }, { "epoch": 11.310622710622711, "grad_norm": 0.86328125, "learning_rate": 8.114373781132163e-05, "loss": 0.1279, "step": 15439 }, { "epoch": 11.311355311355312, "grad_norm": 0.88671875, "learning_rate": 8.11410468812456e-05, "loss": 0.0946, "step": 15440 }, { "epoch": 11.312087912087913, "grad_norm": 1.1015625, "learning_rate": 8.113835580435827e-05, "loss": 0.1582, "step": 15441 }, { "epoch": 11.312820512820513, "grad_norm": 0.79296875, "learning_rate": 8.113566458067253e-05, "loss": 0.1219, "step": 15442 }, { "epoch": 11.313553113553114, "grad_norm": 1.2109375, "learning_rate": 8.113297321020125e-05, "loss": 0.1087, "step": 15443 }, { "epoch": 11.314285714285715, "grad_norm": 1.3359375, "learning_rate": 8.113028169295735e-05, "loss": 0.1711, "step": 15444 }, { "epoch": 11.315018315018316, "grad_norm": 1.2890625, "learning_rate": 8.11275900289537e-05, "loss": 0.1313, "step": 15445 }, { "epoch": 11.315750915750916, "grad_norm": 0.76171875, "learning_rate": 8.112489821820323e-05, "loss": 0.118, "step": 15446 }, { "epoch": 11.316483516483517, "grad_norm": 0.8359375, "learning_rate": 8.112220626071882e-05, "loss": 0.1134, "step": 15447 }, { "epoch": 11.317216117216118, "grad_norm": 1.1875, "learning_rate": 8.111951415651334e-05, "loss": 0.1509, "step": 15448 }, { "epoch": 11.317948717948719, "grad_norm": 1.390625, "learning_rate": 8.111682190559973e-05, "loss": 0.1474, "step": 15449 }, { "epoch": 11.31868131868132, "grad_norm": 1.4140625, "learning_rate": 8.111412950799089e-05, "loss": 0.1315, "step": 15450 }, { "epoch": 11.31941391941392, "grad_norm": 0.84375, "learning_rate": 8.111143696369968e-05, "loss": 0.1299, "step": 15451 }, { "epoch": 11.32014652014652, "grad_norm": 0.8828125, "learning_rate": 8.110874427273904e-05, "loss": 0.1233, "step": 15452 }, { "epoch": 11.320879120879122, "grad_norm": 1.109375, "learning_rate": 8.110605143512186e-05, "loss": 0.1462, "step": 15453 }, { "epoch": 11.321611721611722, "grad_norm": 1.59375, "learning_rate": 8.110335845086105e-05, "loss": 0.129, "step": 15454 }, { "epoch": 11.322344322344323, "grad_norm": 1.59375, "learning_rate": 8.110066531996949e-05, "loss": 0.1548, "step": 15455 }, { "epoch": 11.323076923076924, "grad_norm": 1.0859375, "learning_rate": 8.10979720424601e-05, "loss": 0.1278, "step": 15456 }, { "epoch": 11.323809523809524, "grad_norm": 0.93359375, "learning_rate": 8.109527861834577e-05, "loss": 0.131, "step": 15457 }, { "epoch": 11.324542124542125, "grad_norm": 1.34375, "learning_rate": 8.109258504763942e-05, "loss": 0.1201, "step": 15458 }, { "epoch": 11.325274725274726, "grad_norm": 0.90234375, "learning_rate": 8.108989133035396e-05, "loss": 0.1214, "step": 15459 }, { "epoch": 11.326007326007327, "grad_norm": 0.84765625, "learning_rate": 8.108719746650229e-05, "loss": 0.0973, "step": 15460 }, { "epoch": 11.326739926739927, "grad_norm": 0.98046875, "learning_rate": 8.108450345609731e-05, "loss": 0.1344, "step": 15461 }, { "epoch": 11.327472527472528, "grad_norm": 0.84375, "learning_rate": 8.108180929915195e-05, "loss": 0.1172, "step": 15462 }, { "epoch": 11.328205128205129, "grad_norm": 1.1875, "learning_rate": 8.107911499567909e-05, "loss": 0.121, "step": 15463 }, { "epoch": 11.32893772893773, "grad_norm": 0.828125, "learning_rate": 8.107642054569165e-05, "loss": 0.1312, "step": 15464 }, { "epoch": 11.32967032967033, "grad_norm": 0.94140625, "learning_rate": 8.107372594920253e-05, "loss": 0.1394, "step": 15465 }, { "epoch": 11.330402930402931, "grad_norm": 1.1171875, "learning_rate": 8.107103120622465e-05, "loss": 0.1209, "step": 15466 }, { "epoch": 11.331135531135532, "grad_norm": 1.6015625, "learning_rate": 8.106833631677096e-05, "loss": 0.1201, "step": 15467 }, { "epoch": 11.331868131868132, "grad_norm": 0.9140625, "learning_rate": 8.10656412808543e-05, "loss": 0.1155, "step": 15468 }, { "epoch": 11.332600732600733, "grad_norm": 1.234375, "learning_rate": 8.106294609848763e-05, "loss": 0.0884, "step": 15469 }, { "epoch": 11.333333333333334, "grad_norm": 0.7265625, "learning_rate": 8.106025076968385e-05, "loss": 0.1166, "step": 15470 }, { "epoch": 11.334065934065935, "grad_norm": 1.21875, "learning_rate": 8.10575552944559e-05, "loss": 0.1565, "step": 15471 }, { "epoch": 11.334798534798535, "grad_norm": 1.515625, "learning_rate": 8.105485967281664e-05, "loss": 0.1533, "step": 15472 }, { "epoch": 11.335531135531136, "grad_norm": 0.8203125, "learning_rate": 8.105216390477903e-05, "loss": 0.1372, "step": 15473 }, { "epoch": 11.336263736263737, "grad_norm": 0.828125, "learning_rate": 8.104946799035597e-05, "loss": 0.1005, "step": 15474 }, { "epoch": 11.336996336996338, "grad_norm": 1.21875, "learning_rate": 8.104677192956039e-05, "loss": 0.1551, "step": 15475 }, { "epoch": 11.337728937728938, "grad_norm": 0.57421875, "learning_rate": 8.104407572240519e-05, "loss": 0.0985, "step": 15476 }, { "epoch": 11.338461538461539, "grad_norm": 0.6328125, "learning_rate": 8.104137936890329e-05, "loss": 0.0959, "step": 15477 }, { "epoch": 11.33919413919414, "grad_norm": 0.98828125, "learning_rate": 8.103868286906761e-05, "loss": 0.1514, "step": 15478 }, { "epoch": 11.33992673992674, "grad_norm": 0.76171875, "learning_rate": 8.103598622291109e-05, "loss": 0.113, "step": 15479 }, { "epoch": 11.340659340659341, "grad_norm": 0.8203125, "learning_rate": 8.103328943044663e-05, "loss": 0.1371, "step": 15480 }, { "epoch": 11.341391941391942, "grad_norm": 1.2421875, "learning_rate": 8.103059249168716e-05, "loss": 0.1333, "step": 15481 }, { "epoch": 11.342124542124543, "grad_norm": 1.296875, "learning_rate": 8.102789540664561e-05, "loss": 0.13, "step": 15482 }, { "epoch": 11.342857142857143, "grad_norm": 1.390625, "learning_rate": 8.102519817533487e-05, "loss": 0.1087, "step": 15483 }, { "epoch": 11.343589743589744, "grad_norm": 1.1796875, "learning_rate": 8.10225007977679e-05, "loss": 0.1488, "step": 15484 }, { "epoch": 11.344322344322345, "grad_norm": 1.0078125, "learning_rate": 8.101980327395761e-05, "loss": 0.153, "step": 15485 }, { "epoch": 11.345054945054946, "grad_norm": 1.8359375, "learning_rate": 8.101710560391691e-05, "loss": 0.1439, "step": 15486 }, { "epoch": 11.345787545787546, "grad_norm": 1.5390625, "learning_rate": 8.101440778765874e-05, "loss": 0.147, "step": 15487 }, { "epoch": 11.346520146520147, "grad_norm": 1.3828125, "learning_rate": 8.101170982519603e-05, "loss": 0.1105, "step": 15488 }, { "epoch": 11.347252747252748, "grad_norm": 1.015625, "learning_rate": 8.100901171654171e-05, "loss": 0.14, "step": 15489 }, { "epoch": 11.347985347985349, "grad_norm": 0.8828125, "learning_rate": 8.100631346170869e-05, "loss": 0.1344, "step": 15490 }, { "epoch": 11.34871794871795, "grad_norm": 0.98828125, "learning_rate": 8.10036150607099e-05, "loss": 0.1171, "step": 15491 }, { "epoch": 11.34945054945055, "grad_norm": 0.65234375, "learning_rate": 8.100091651355827e-05, "loss": 0.1118, "step": 15492 }, { "epoch": 11.35018315018315, "grad_norm": 1.5625, "learning_rate": 8.099821782026676e-05, "loss": 0.1341, "step": 15493 }, { "epoch": 11.350915750915751, "grad_norm": 1.4453125, "learning_rate": 8.099551898084828e-05, "loss": 0.1165, "step": 15494 }, { "epoch": 11.351648351648352, "grad_norm": 1.640625, "learning_rate": 8.099281999531573e-05, "loss": 0.1494, "step": 15495 }, { "epoch": 11.352380952380953, "grad_norm": 1.453125, "learning_rate": 8.09901208636821e-05, "loss": 0.1141, "step": 15496 }, { "epoch": 11.353113553113554, "grad_norm": 0.58203125, "learning_rate": 8.098742158596028e-05, "loss": 0.1034, "step": 15497 }, { "epoch": 11.353846153846154, "grad_norm": 1.65625, "learning_rate": 8.098472216216321e-05, "loss": 0.1441, "step": 15498 }, { "epoch": 11.354578754578755, "grad_norm": 1.453125, "learning_rate": 8.098202259230383e-05, "loss": 0.1179, "step": 15499 }, { "epoch": 11.355311355311356, "grad_norm": 1.3125, "learning_rate": 8.097932287639507e-05, "loss": 0.1282, "step": 15500 }, { "epoch": 11.356043956043957, "grad_norm": 1.2421875, "learning_rate": 8.097662301444988e-05, "loss": 0.126, "step": 15501 }, { "epoch": 11.356776556776557, "grad_norm": 1.3671875, "learning_rate": 8.097392300648117e-05, "loss": 0.1638, "step": 15502 }, { "epoch": 11.357509157509158, "grad_norm": 0.88671875, "learning_rate": 8.097122285250191e-05, "loss": 0.1, "step": 15503 }, { "epoch": 11.358241758241759, "grad_norm": 1.2734375, "learning_rate": 8.096852255252499e-05, "loss": 0.1341, "step": 15504 }, { "epoch": 11.35897435897436, "grad_norm": 1.1484375, "learning_rate": 8.096582210656339e-05, "loss": 0.1221, "step": 15505 }, { "epoch": 11.35970695970696, "grad_norm": 1.9140625, "learning_rate": 8.096312151463004e-05, "loss": 0.1358, "step": 15506 }, { "epoch": 11.360439560439561, "grad_norm": 1.65625, "learning_rate": 8.096042077673787e-05, "loss": 0.1079, "step": 15507 }, { "epoch": 11.361172161172162, "grad_norm": 1.4375, "learning_rate": 8.095771989289983e-05, "loss": 0.1366, "step": 15508 }, { "epoch": 11.361904761904762, "grad_norm": 1.1875, "learning_rate": 8.095501886312885e-05, "loss": 0.1077, "step": 15509 }, { "epoch": 11.362637362637363, "grad_norm": 0.87109375, "learning_rate": 8.095231768743789e-05, "loss": 0.1466, "step": 15510 }, { "epoch": 11.363369963369964, "grad_norm": 1.015625, "learning_rate": 8.094961636583985e-05, "loss": 0.1426, "step": 15511 }, { "epoch": 11.364102564102565, "grad_norm": 1.0390625, "learning_rate": 8.094691489834772e-05, "loss": 0.134, "step": 15512 }, { "epoch": 11.364835164835165, "grad_norm": 1.7109375, "learning_rate": 8.094421328497443e-05, "loss": 0.1262, "step": 15513 }, { "epoch": 11.365567765567766, "grad_norm": 1.6875, "learning_rate": 8.09415115257329e-05, "loss": 0.1111, "step": 15514 }, { "epoch": 11.366300366300367, "grad_norm": 0.80078125, "learning_rate": 8.093880962063613e-05, "loss": 0.1016, "step": 15515 }, { "epoch": 11.367032967032968, "grad_norm": 0.87890625, "learning_rate": 8.093610756969699e-05, "loss": 0.1214, "step": 15516 }, { "epoch": 11.367765567765568, "grad_norm": 0.86328125, "learning_rate": 8.093340537292851e-05, "loss": 0.1207, "step": 15517 }, { "epoch": 11.368498168498169, "grad_norm": 1.3984375, "learning_rate": 8.093070303034356e-05, "loss": 0.1502, "step": 15518 }, { "epoch": 11.36923076923077, "grad_norm": 1.6875, "learning_rate": 8.092800054195513e-05, "loss": 0.1583, "step": 15519 }, { "epoch": 11.36996336996337, "grad_norm": 1.375, "learning_rate": 8.092529790777617e-05, "loss": 0.1215, "step": 15520 }, { "epoch": 11.370695970695971, "grad_norm": 1.2890625, "learning_rate": 8.09225951278196e-05, "loss": 0.1238, "step": 15521 }, { "epoch": 11.371428571428572, "grad_norm": 1.4140625, "learning_rate": 8.091989220209842e-05, "loss": 0.1839, "step": 15522 }, { "epoch": 11.372161172161173, "grad_norm": 1.1953125, "learning_rate": 8.091718913062553e-05, "loss": 0.1291, "step": 15523 }, { "epoch": 11.372893772893773, "grad_norm": 0.796875, "learning_rate": 8.091448591341392e-05, "loss": 0.1217, "step": 15524 }, { "epoch": 11.373626373626374, "grad_norm": 0.77734375, "learning_rate": 8.09117825504765e-05, "loss": 0.1207, "step": 15525 }, { "epoch": 11.374358974358975, "grad_norm": 1.671875, "learning_rate": 8.090907904182625e-05, "loss": 0.1467, "step": 15526 }, { "epoch": 11.375091575091576, "grad_norm": 1.296875, "learning_rate": 8.090637538747612e-05, "loss": 0.1654, "step": 15527 }, { "epoch": 11.375824175824176, "grad_norm": 0.88671875, "learning_rate": 8.090367158743908e-05, "loss": 0.1126, "step": 15528 }, { "epoch": 11.376556776556777, "grad_norm": 0.99609375, "learning_rate": 8.090096764172806e-05, "loss": 0.1468, "step": 15529 }, { "epoch": 11.377289377289378, "grad_norm": 1.390625, "learning_rate": 8.089826355035601e-05, "loss": 0.1414, "step": 15530 }, { "epoch": 11.378021978021978, "grad_norm": 1.625, "learning_rate": 8.089555931333591e-05, "loss": 0.1566, "step": 15531 }, { "epoch": 11.37875457875458, "grad_norm": 1.109375, "learning_rate": 8.089285493068072e-05, "loss": 0.0982, "step": 15532 }, { "epoch": 11.37948717948718, "grad_norm": 1.515625, "learning_rate": 8.089015040240337e-05, "loss": 0.1437, "step": 15533 }, { "epoch": 11.38021978021978, "grad_norm": 1.125, "learning_rate": 8.088744572851683e-05, "loss": 0.1395, "step": 15534 }, { "epoch": 11.380952380952381, "grad_norm": 1.34375, "learning_rate": 8.088474090903406e-05, "loss": 0.1679, "step": 15535 }, { "epoch": 11.381684981684982, "grad_norm": 1.40625, "learning_rate": 8.088203594396805e-05, "loss": 0.149, "step": 15536 }, { "epoch": 11.382417582417583, "grad_norm": 1.2421875, "learning_rate": 8.08793308333317e-05, "loss": 0.1085, "step": 15537 }, { "epoch": 11.383150183150184, "grad_norm": 1.40625, "learning_rate": 8.087662557713803e-05, "loss": 0.1516, "step": 15538 }, { "epoch": 11.383882783882784, "grad_norm": 1.109375, "learning_rate": 8.087392017539996e-05, "loss": 0.1445, "step": 15539 }, { "epoch": 11.384615384615385, "grad_norm": 1.3671875, "learning_rate": 8.087121462813046e-05, "loss": 0.1506, "step": 15540 }, { "epoch": 11.385347985347986, "grad_norm": 0.65234375, "learning_rate": 8.08685089353425e-05, "loss": 0.1135, "step": 15541 }, { "epoch": 11.386080586080586, "grad_norm": 1.3828125, "learning_rate": 8.086580309704906e-05, "loss": 0.1513, "step": 15542 }, { "epoch": 11.386813186813187, "grad_norm": 1.15625, "learning_rate": 8.086309711326307e-05, "loss": 0.1225, "step": 15543 }, { "epoch": 11.387545787545788, "grad_norm": 0.69140625, "learning_rate": 8.086039098399753e-05, "loss": 0.1265, "step": 15544 }, { "epoch": 11.388278388278389, "grad_norm": 1.1796875, "learning_rate": 8.085768470926538e-05, "loss": 0.153, "step": 15545 }, { "epoch": 11.38901098901099, "grad_norm": 0.796875, "learning_rate": 8.085497828907961e-05, "loss": 0.1031, "step": 15546 }, { "epoch": 11.38974358974359, "grad_norm": 0.578125, "learning_rate": 8.085227172345316e-05, "loss": 0.1115, "step": 15547 }, { "epoch": 11.39047619047619, "grad_norm": 1.03125, "learning_rate": 8.084956501239903e-05, "loss": 0.1254, "step": 15548 }, { "epoch": 11.391208791208792, "grad_norm": 0.8671875, "learning_rate": 8.084685815593017e-05, "loss": 0.1412, "step": 15549 }, { "epoch": 11.391941391941392, "grad_norm": 0.77734375, "learning_rate": 8.084415115405952e-05, "loss": 0.1448, "step": 15550 }, { "epoch": 11.392673992673993, "grad_norm": 1.2578125, "learning_rate": 8.084144400680011e-05, "loss": 0.1535, "step": 15551 }, { "epoch": 11.393406593406594, "grad_norm": 1.0, "learning_rate": 8.083873671416485e-05, "loss": 0.1261, "step": 15552 }, { "epoch": 11.394139194139195, "grad_norm": 1.109375, "learning_rate": 8.083602927616677e-05, "loss": 0.1052, "step": 15553 }, { "epoch": 11.394871794871795, "grad_norm": 1.359375, "learning_rate": 8.083332169281882e-05, "loss": 0.1657, "step": 15554 }, { "epoch": 11.395604395604396, "grad_norm": 0.890625, "learning_rate": 8.083061396413397e-05, "loss": 0.1596, "step": 15555 }, { "epoch": 11.396336996336997, "grad_norm": 0.87109375, "learning_rate": 8.082790609012518e-05, "loss": 0.1287, "step": 15556 }, { "epoch": 11.397069597069597, "grad_norm": 0.79296875, "learning_rate": 8.082519807080542e-05, "loss": 0.1311, "step": 15557 }, { "epoch": 11.397802197802198, "grad_norm": 1.6953125, "learning_rate": 8.08224899061877e-05, "loss": 0.1544, "step": 15558 }, { "epoch": 11.398534798534799, "grad_norm": 1.265625, "learning_rate": 8.081978159628497e-05, "loss": 0.1091, "step": 15559 }, { "epoch": 11.3992673992674, "grad_norm": 0.9921875, "learning_rate": 8.081707314111022e-05, "loss": 0.1105, "step": 15560 }, { "epoch": 11.4, "grad_norm": 0.7734375, "learning_rate": 8.081436454067642e-05, "loss": 0.1234, "step": 15561 }, { "epoch": 11.400732600732601, "grad_norm": 0.80078125, "learning_rate": 8.081165579499655e-05, "loss": 0.1331, "step": 15562 }, { "epoch": 11.401465201465202, "grad_norm": 0.83203125, "learning_rate": 8.080894690408357e-05, "loss": 0.1229, "step": 15563 }, { "epoch": 11.402197802197803, "grad_norm": 0.93359375, "learning_rate": 8.080623786795049e-05, "loss": 0.1237, "step": 15564 }, { "epoch": 11.402930402930403, "grad_norm": 1.109375, "learning_rate": 8.080352868661026e-05, "loss": 0.1449, "step": 15565 }, { "epoch": 11.403663003663004, "grad_norm": 1.109375, "learning_rate": 8.080081936007588e-05, "loss": 0.1249, "step": 15566 }, { "epoch": 11.404395604395605, "grad_norm": 0.91015625, "learning_rate": 8.079810988836036e-05, "loss": 0.1143, "step": 15567 }, { "epoch": 11.405128205128205, "grad_norm": 1.78125, "learning_rate": 8.079540027147661e-05, "loss": 0.1709, "step": 15568 }, { "epoch": 11.405860805860806, "grad_norm": 1.3203125, "learning_rate": 8.079269050943767e-05, "loss": 0.1337, "step": 15569 }, { "epoch": 11.406593406593407, "grad_norm": 1.3046875, "learning_rate": 8.078998060225649e-05, "loss": 0.1331, "step": 15570 }, { "epoch": 11.407326007326008, "grad_norm": 1.34375, "learning_rate": 8.078727054994608e-05, "loss": 0.1466, "step": 15571 }, { "epoch": 11.408058608058608, "grad_norm": 0.84375, "learning_rate": 8.078456035251944e-05, "loss": 0.1246, "step": 15572 }, { "epoch": 11.40879120879121, "grad_norm": 0.7890625, "learning_rate": 8.078185000998948e-05, "loss": 0.1233, "step": 15573 }, { "epoch": 11.40952380952381, "grad_norm": 0.81640625, "learning_rate": 8.077913952236926e-05, "loss": 0.1203, "step": 15574 }, { "epoch": 11.41025641025641, "grad_norm": 1.546875, "learning_rate": 8.077642888967174e-05, "loss": 0.153, "step": 15575 }, { "epoch": 11.410989010989011, "grad_norm": 1.34375, "learning_rate": 8.077371811190992e-05, "loss": 0.1112, "step": 15576 }, { "epoch": 11.411721611721612, "grad_norm": 1.8984375, "learning_rate": 8.077100718909679e-05, "loss": 0.1286, "step": 15577 }, { "epoch": 11.412454212454213, "grad_norm": 1.0390625, "learning_rate": 8.076829612124529e-05, "loss": 0.1371, "step": 15578 }, { "epoch": 11.413186813186813, "grad_norm": 0.8515625, "learning_rate": 8.076558490836846e-05, "loss": 0.1253, "step": 15579 }, { "epoch": 11.413919413919414, "grad_norm": 1.1015625, "learning_rate": 8.076287355047929e-05, "loss": 0.1321, "step": 15580 }, { "epoch": 11.414652014652015, "grad_norm": 1.078125, "learning_rate": 8.076016204759075e-05, "loss": 0.1249, "step": 15581 }, { "epoch": 11.415384615384616, "grad_norm": 1.734375, "learning_rate": 8.075745039971584e-05, "loss": 0.1163, "step": 15582 }, { "epoch": 11.416117216117216, "grad_norm": 1.4765625, "learning_rate": 8.075473860686754e-05, "loss": 0.1383, "step": 15583 }, { "epoch": 11.416849816849817, "grad_norm": 0.80859375, "learning_rate": 8.075202666905888e-05, "loss": 0.126, "step": 15584 }, { "epoch": 11.417582417582418, "grad_norm": 0.75390625, "learning_rate": 8.074931458630282e-05, "loss": 0.1135, "step": 15585 }, { "epoch": 11.418315018315019, "grad_norm": 0.94140625, "learning_rate": 8.074660235861235e-05, "loss": 0.1336, "step": 15586 }, { "epoch": 11.41904761904762, "grad_norm": 0.8203125, "learning_rate": 8.07438899860005e-05, "loss": 0.1177, "step": 15587 }, { "epoch": 11.41978021978022, "grad_norm": 1.359375, "learning_rate": 8.074117746848024e-05, "loss": 0.152, "step": 15588 }, { "epoch": 11.42051282051282, "grad_norm": 1.46875, "learning_rate": 8.073846480606455e-05, "loss": 0.1379, "step": 15589 }, { "epoch": 11.421245421245422, "grad_norm": 1.2265625, "learning_rate": 8.073575199876647e-05, "loss": 0.1272, "step": 15590 }, { "epoch": 11.421978021978022, "grad_norm": 1.109375, "learning_rate": 8.073303904659896e-05, "loss": 0.1239, "step": 15591 }, { "epoch": 11.422710622710623, "grad_norm": 0.86328125, "learning_rate": 8.073032594957506e-05, "loss": 0.1394, "step": 15592 }, { "epoch": 11.423443223443224, "grad_norm": 1.0703125, "learning_rate": 8.072761270770773e-05, "loss": 0.1022, "step": 15593 }, { "epoch": 11.424175824175824, "grad_norm": 1.0078125, "learning_rate": 8.072489932100998e-05, "loss": 0.1023, "step": 15594 }, { "epoch": 11.424908424908425, "grad_norm": 1.28125, "learning_rate": 8.072218578949482e-05, "loss": 0.1263, "step": 15595 }, { "epoch": 11.425641025641026, "grad_norm": 1.140625, "learning_rate": 8.071947211317524e-05, "loss": 0.1238, "step": 15596 }, { "epoch": 11.426373626373627, "grad_norm": 0.77734375, "learning_rate": 8.071675829206426e-05, "loss": 0.1143, "step": 15597 }, { "epoch": 11.427106227106227, "grad_norm": 0.984375, "learning_rate": 8.071404432617486e-05, "loss": 0.1294, "step": 15598 }, { "epoch": 11.427838827838828, "grad_norm": 1.234375, "learning_rate": 8.071133021552007e-05, "loss": 0.1348, "step": 15599 }, { "epoch": 11.428571428571429, "grad_norm": 1.015625, "learning_rate": 8.070861596011287e-05, "loss": 0.1451, "step": 15600 }, { "epoch": 11.42930402930403, "grad_norm": 1.3828125, "learning_rate": 8.070590155996627e-05, "loss": 0.1181, "step": 15601 }, { "epoch": 11.43003663003663, "grad_norm": 0.75390625, "learning_rate": 8.070318701509329e-05, "loss": 0.1237, "step": 15602 }, { "epoch": 11.430769230769231, "grad_norm": 1.2421875, "learning_rate": 8.070047232550691e-05, "loss": 0.1532, "step": 15603 }, { "epoch": 11.431501831501832, "grad_norm": 0.984375, "learning_rate": 8.069775749122016e-05, "loss": 0.1054, "step": 15604 }, { "epoch": 11.432234432234432, "grad_norm": 0.83984375, "learning_rate": 8.069504251224606e-05, "loss": 0.1309, "step": 15605 }, { "epoch": 11.432967032967033, "grad_norm": 0.921875, "learning_rate": 8.069232738859759e-05, "loss": 0.1087, "step": 15606 }, { "epoch": 11.433699633699634, "grad_norm": 1.375, "learning_rate": 8.068961212028775e-05, "loss": 0.1663, "step": 15607 }, { "epoch": 11.434432234432235, "grad_norm": 0.90625, "learning_rate": 8.068689670732957e-05, "loss": 0.119, "step": 15608 }, { "epoch": 11.435164835164835, "grad_norm": 1.390625, "learning_rate": 8.068418114973607e-05, "loss": 0.1638, "step": 15609 }, { "epoch": 11.435897435897436, "grad_norm": 0.97265625, "learning_rate": 8.068146544752024e-05, "loss": 0.1359, "step": 15610 }, { "epoch": 11.436630036630037, "grad_norm": 1.0390625, "learning_rate": 8.067874960069512e-05, "loss": 0.1218, "step": 15611 }, { "epoch": 11.437362637362638, "grad_norm": 1.1796875, "learning_rate": 8.067603360927369e-05, "loss": 0.1304, "step": 15612 }, { "epoch": 11.438095238095238, "grad_norm": 0.94140625, "learning_rate": 8.067331747326897e-05, "loss": 0.1057, "step": 15613 }, { "epoch": 11.438827838827839, "grad_norm": 1.6015625, "learning_rate": 8.067060119269399e-05, "loss": 0.1344, "step": 15614 }, { "epoch": 11.43956043956044, "grad_norm": 1.0078125, "learning_rate": 8.066788476756175e-05, "loss": 0.1187, "step": 15615 }, { "epoch": 11.44029304029304, "grad_norm": 0.99609375, "learning_rate": 8.066516819788528e-05, "loss": 0.1513, "step": 15616 }, { "epoch": 11.441025641025641, "grad_norm": 1.1640625, "learning_rate": 8.066245148367757e-05, "loss": 0.1718, "step": 15617 }, { "epoch": 11.441758241758242, "grad_norm": 0.7421875, "learning_rate": 8.065973462495167e-05, "loss": 0.136, "step": 15618 }, { "epoch": 11.442490842490843, "grad_norm": 0.66015625, "learning_rate": 8.065701762172058e-05, "loss": 0.1066, "step": 15619 }, { "epoch": 11.443223443223443, "grad_norm": 1.1171875, "learning_rate": 8.065430047399732e-05, "loss": 0.134, "step": 15620 }, { "epoch": 11.443956043956044, "grad_norm": 0.97265625, "learning_rate": 8.065158318179489e-05, "loss": 0.1384, "step": 15621 }, { "epoch": 11.444688644688645, "grad_norm": 0.9921875, "learning_rate": 8.064886574512634e-05, "loss": 0.0931, "step": 15622 }, { "epoch": 11.445421245421246, "grad_norm": 1.84375, "learning_rate": 8.064614816400467e-05, "loss": 0.1331, "step": 15623 }, { "epoch": 11.446153846153846, "grad_norm": 0.9140625, "learning_rate": 8.064343043844291e-05, "loss": 0.0995, "step": 15624 }, { "epoch": 11.446886446886447, "grad_norm": 0.88671875, "learning_rate": 8.064071256845409e-05, "loss": 0.1458, "step": 15625 }, { "epoch": 11.447619047619048, "grad_norm": 1.3984375, "learning_rate": 8.063799455405121e-05, "loss": 0.0985, "step": 15626 }, { "epoch": 11.448351648351649, "grad_norm": 1.6328125, "learning_rate": 8.063527639524731e-05, "loss": 0.1315, "step": 15627 }, { "epoch": 11.44908424908425, "grad_norm": 0.98046875, "learning_rate": 8.063255809205541e-05, "loss": 0.1054, "step": 15628 }, { "epoch": 11.44981684981685, "grad_norm": 0.921875, "learning_rate": 8.062983964448852e-05, "loss": 0.1298, "step": 15629 }, { "epoch": 11.45054945054945, "grad_norm": 1.3046875, "learning_rate": 8.062712105255969e-05, "loss": 0.1237, "step": 15630 }, { "epoch": 11.451282051282051, "grad_norm": 1.7109375, "learning_rate": 8.062440231628191e-05, "loss": 0.1466, "step": 15631 }, { "epoch": 11.452014652014652, "grad_norm": 0.8828125, "learning_rate": 8.062168343566826e-05, "loss": 0.105, "step": 15632 }, { "epoch": 11.452747252747253, "grad_norm": 1.234375, "learning_rate": 8.061896441073173e-05, "loss": 0.1397, "step": 15633 }, { "epoch": 11.453479853479854, "grad_norm": 1.453125, "learning_rate": 8.061624524148534e-05, "loss": 0.1261, "step": 15634 }, { "epoch": 11.454212454212454, "grad_norm": 1.0, "learning_rate": 8.061352592794214e-05, "loss": 0.1584, "step": 15635 }, { "epoch": 11.454945054945055, "grad_norm": 1.1015625, "learning_rate": 8.061080647011515e-05, "loss": 0.1539, "step": 15636 }, { "epoch": 11.455677655677656, "grad_norm": 1.4921875, "learning_rate": 8.06080868680174e-05, "loss": 0.1341, "step": 15637 }, { "epoch": 11.456410256410257, "grad_norm": 2.03125, "learning_rate": 8.060536712166192e-05, "loss": 0.1142, "step": 15638 }, { "epoch": 11.457142857142857, "grad_norm": 1.1640625, "learning_rate": 8.060264723106177e-05, "loss": 0.133, "step": 15639 }, { "epoch": 11.457875457875458, "grad_norm": 1.46875, "learning_rate": 8.059992719622991e-05, "loss": 0.1298, "step": 15640 }, { "epoch": 11.458608058608059, "grad_norm": 1.640625, "learning_rate": 8.059720701717946e-05, "loss": 0.1682, "step": 15641 }, { "epoch": 11.45934065934066, "grad_norm": 1.0078125, "learning_rate": 8.05944866939234e-05, "loss": 0.109, "step": 15642 }, { "epoch": 11.46007326007326, "grad_norm": 1.1015625, "learning_rate": 8.059176622647475e-05, "loss": 0.1216, "step": 15643 }, { "epoch": 11.460805860805861, "grad_norm": 0.91796875, "learning_rate": 8.058904561484659e-05, "loss": 0.1199, "step": 15644 }, { "epoch": 11.461538461538462, "grad_norm": 0.96484375, "learning_rate": 8.058632485905194e-05, "loss": 0.2009, "step": 15645 }, { "epoch": 11.462271062271062, "grad_norm": 1.25, "learning_rate": 8.058360395910381e-05, "loss": 0.1206, "step": 15646 }, { "epoch": 11.463003663003663, "grad_norm": 1.4765625, "learning_rate": 8.058088291501527e-05, "loss": 0.1346, "step": 15647 }, { "epoch": 11.463736263736264, "grad_norm": 2.09375, "learning_rate": 8.057816172679935e-05, "loss": 0.1392, "step": 15648 }, { "epoch": 11.464468864468865, "grad_norm": 1.1484375, "learning_rate": 8.057544039446908e-05, "loss": 0.1368, "step": 15649 }, { "epoch": 11.465201465201465, "grad_norm": 1.1328125, "learning_rate": 8.05727189180375e-05, "loss": 0.12, "step": 15650 }, { "epoch": 11.465934065934066, "grad_norm": 1.015625, "learning_rate": 8.056999729751765e-05, "loss": 0.1246, "step": 15651 }, { "epoch": 11.466666666666667, "grad_norm": 1.171875, "learning_rate": 8.056727553292258e-05, "loss": 0.134, "step": 15652 }, { "epoch": 11.467399267399268, "grad_norm": 1.4296875, "learning_rate": 8.056455362426531e-05, "loss": 0.1469, "step": 15653 }, { "epoch": 11.468131868131868, "grad_norm": 0.99609375, "learning_rate": 8.05618315715589e-05, "loss": 0.1471, "step": 15654 }, { "epoch": 11.468864468864469, "grad_norm": 0.72265625, "learning_rate": 8.055910937481639e-05, "loss": 0.1202, "step": 15655 }, { "epoch": 11.46959706959707, "grad_norm": 1.0234375, "learning_rate": 8.055638703405081e-05, "loss": 0.1055, "step": 15656 }, { "epoch": 11.47032967032967, "grad_norm": 1.03125, "learning_rate": 8.055366454927523e-05, "loss": 0.1242, "step": 15657 }, { "epoch": 11.471062271062271, "grad_norm": 1.1015625, "learning_rate": 8.055094192050266e-05, "loss": 0.1108, "step": 15658 }, { "epoch": 11.471794871794872, "grad_norm": 0.9140625, "learning_rate": 8.054821914774616e-05, "loss": 0.1473, "step": 15659 }, { "epoch": 11.472527472527473, "grad_norm": 1.2890625, "learning_rate": 8.05454962310188e-05, "loss": 0.1604, "step": 15660 }, { "epoch": 11.473260073260073, "grad_norm": 0.8046875, "learning_rate": 8.054277317033357e-05, "loss": 0.1304, "step": 15661 }, { "epoch": 11.473992673992674, "grad_norm": 1.09375, "learning_rate": 8.054004996570358e-05, "loss": 0.1344, "step": 15662 }, { "epoch": 11.474725274725275, "grad_norm": 0.5703125, "learning_rate": 8.053732661714185e-05, "loss": 0.0851, "step": 15663 }, { "epoch": 11.475457875457876, "grad_norm": 1.203125, "learning_rate": 8.053460312466139e-05, "loss": 0.1369, "step": 15664 }, { "epoch": 11.476190476190476, "grad_norm": 0.87109375, "learning_rate": 8.053187948827533e-05, "loss": 0.1119, "step": 15665 }, { "epoch": 11.476923076923077, "grad_norm": 1.1796875, "learning_rate": 8.052915570799667e-05, "loss": 0.162, "step": 15666 }, { "epoch": 11.477655677655678, "grad_norm": 1.1796875, "learning_rate": 8.052643178383846e-05, "loss": 0.1335, "step": 15667 }, { "epoch": 11.478388278388278, "grad_norm": 1.1171875, "learning_rate": 8.052370771581374e-05, "loss": 0.1225, "step": 15668 }, { "epoch": 11.47912087912088, "grad_norm": 1.4453125, "learning_rate": 8.05209835039356e-05, "loss": 0.1143, "step": 15669 }, { "epoch": 11.47985347985348, "grad_norm": 1.265625, "learning_rate": 8.051825914821706e-05, "loss": 0.1396, "step": 15670 }, { "epoch": 11.48058608058608, "grad_norm": 1.0625, "learning_rate": 8.051553464867121e-05, "loss": 0.1332, "step": 15671 }, { "epoch": 11.481318681318681, "grad_norm": 1.03125, "learning_rate": 8.051281000531105e-05, "loss": 0.1229, "step": 15672 }, { "epoch": 11.482051282051282, "grad_norm": 0.75, "learning_rate": 8.051008521814969e-05, "loss": 0.1102, "step": 15673 }, { "epoch": 11.482783882783883, "grad_norm": 1.234375, "learning_rate": 8.050736028720015e-05, "loss": 0.1307, "step": 15674 }, { "epoch": 11.483516483516484, "grad_norm": 1.421875, "learning_rate": 8.050463521247549e-05, "loss": 0.105, "step": 15675 }, { "epoch": 11.484249084249084, "grad_norm": 1.0859375, "learning_rate": 8.05019099939888e-05, "loss": 0.1554, "step": 15676 }, { "epoch": 11.484981684981685, "grad_norm": 1.3203125, "learning_rate": 8.049918463175307e-05, "loss": 0.1305, "step": 15677 }, { "epoch": 11.485714285714286, "grad_norm": 1.5625, "learning_rate": 8.049645912578142e-05, "loss": 0.1371, "step": 15678 }, { "epoch": 11.486446886446886, "grad_norm": 1.3984375, "learning_rate": 8.049373347608689e-05, "loss": 0.1393, "step": 15679 }, { "epoch": 11.487179487179487, "grad_norm": 1.8671875, "learning_rate": 8.049100768268253e-05, "loss": 0.1233, "step": 15680 }, { "epoch": 11.487912087912088, "grad_norm": 1.171875, "learning_rate": 8.048828174558141e-05, "loss": 0.1084, "step": 15681 }, { "epoch": 11.488644688644689, "grad_norm": 1.4609375, "learning_rate": 8.048555566479658e-05, "loss": 0.1431, "step": 15682 }, { "epoch": 11.48937728937729, "grad_norm": 0.6171875, "learning_rate": 8.048282944034112e-05, "loss": 0.0993, "step": 15683 }, { "epoch": 11.49010989010989, "grad_norm": 0.828125, "learning_rate": 8.048010307222808e-05, "loss": 0.1181, "step": 15684 }, { "epoch": 11.49084249084249, "grad_norm": 1.3828125, "learning_rate": 8.047737656047052e-05, "loss": 0.1667, "step": 15685 }, { "epoch": 11.491575091575092, "grad_norm": 1.515625, "learning_rate": 8.04746499050815e-05, "loss": 0.1431, "step": 15686 }, { "epoch": 11.492307692307692, "grad_norm": 1.96875, "learning_rate": 8.04719231060741e-05, "loss": 0.1962, "step": 15687 }, { "epoch": 11.493040293040293, "grad_norm": 1.1953125, "learning_rate": 8.046919616346137e-05, "loss": 0.1354, "step": 15688 }, { "epoch": 11.493772893772894, "grad_norm": 1.453125, "learning_rate": 8.04664690772564e-05, "loss": 0.1284, "step": 15689 }, { "epoch": 11.494505494505495, "grad_norm": 0.88671875, "learning_rate": 8.046374184747224e-05, "loss": 0.1484, "step": 15690 }, { "epoch": 11.495238095238095, "grad_norm": 1.015625, "learning_rate": 8.046101447412195e-05, "loss": 0.1135, "step": 15691 }, { "epoch": 11.495970695970696, "grad_norm": 1.3359375, "learning_rate": 8.04582869572186e-05, "loss": 0.1469, "step": 15692 }, { "epoch": 11.496703296703297, "grad_norm": 0.74609375, "learning_rate": 8.045555929677525e-05, "loss": 0.1259, "step": 15693 }, { "epoch": 11.497435897435897, "grad_norm": 1.015625, "learning_rate": 8.045283149280501e-05, "loss": 0.0876, "step": 15694 }, { "epoch": 11.498168498168498, "grad_norm": 1.0390625, "learning_rate": 8.045010354532088e-05, "loss": 0.1325, "step": 15695 }, { "epoch": 11.498901098901099, "grad_norm": 0.71875, "learning_rate": 8.044737545433602e-05, "loss": 0.1398, "step": 15696 }, { "epoch": 11.4996336996337, "grad_norm": 0.65234375, "learning_rate": 8.044464721986343e-05, "loss": 0.1301, "step": 15697 }, { "epoch": 11.5003663003663, "grad_norm": 0.5703125, "learning_rate": 8.04419188419162e-05, "loss": 0.1212, "step": 15698 }, { "epoch": 11.501098901098901, "grad_norm": 1.0, "learning_rate": 8.043919032050741e-05, "loss": 0.1444, "step": 15699 }, { "epoch": 11.501831501831502, "grad_norm": 0.75, "learning_rate": 8.043646165565014e-05, "loss": 0.1238, "step": 15700 }, { "epoch": 11.502564102564103, "grad_norm": 1.4609375, "learning_rate": 8.043373284735746e-05, "loss": 0.1263, "step": 15701 }, { "epoch": 11.503296703296703, "grad_norm": 0.98046875, "learning_rate": 8.043100389564242e-05, "loss": 0.1392, "step": 15702 }, { "epoch": 11.504029304029304, "grad_norm": 1.625, "learning_rate": 8.042827480051814e-05, "loss": 0.1541, "step": 15703 }, { "epoch": 11.504761904761905, "grad_norm": 0.73828125, "learning_rate": 8.042554556199765e-05, "loss": 0.1119, "step": 15704 }, { "epoch": 11.505494505494505, "grad_norm": 1.046875, "learning_rate": 8.042281618009405e-05, "loss": 0.1214, "step": 15705 }, { "epoch": 11.506227106227106, "grad_norm": 0.90625, "learning_rate": 8.042008665482041e-05, "loss": 0.1112, "step": 15706 }, { "epoch": 11.506959706959707, "grad_norm": 1.1640625, "learning_rate": 8.041735698618982e-05, "loss": 0.1374, "step": 15707 }, { "epoch": 11.507692307692308, "grad_norm": 1.390625, "learning_rate": 8.041462717421535e-05, "loss": 0.1402, "step": 15708 }, { "epoch": 11.508424908424908, "grad_norm": 0.87109375, "learning_rate": 8.041189721891007e-05, "loss": 0.1146, "step": 15709 }, { "epoch": 11.50915750915751, "grad_norm": 0.98828125, "learning_rate": 8.04091671202871e-05, "loss": 0.119, "step": 15710 }, { "epoch": 11.50989010989011, "grad_norm": 1.15625, "learning_rate": 8.040643687835946e-05, "loss": 0.1316, "step": 15711 }, { "epoch": 11.51062271062271, "grad_norm": 1.1328125, "learning_rate": 8.040370649314027e-05, "loss": 0.1627, "step": 15712 }, { "epoch": 11.511355311355311, "grad_norm": 0.98828125, "learning_rate": 8.040097596464263e-05, "loss": 0.1255, "step": 15713 }, { "epoch": 11.512087912087912, "grad_norm": 1.453125, "learning_rate": 8.039824529287958e-05, "loss": 0.1788, "step": 15714 }, { "epoch": 11.512820512820513, "grad_norm": 1.34375, "learning_rate": 8.039551447786423e-05, "loss": 0.1283, "step": 15715 }, { "epoch": 11.513553113553114, "grad_norm": 1.1484375, "learning_rate": 8.039278351960964e-05, "loss": 0.132, "step": 15716 }, { "epoch": 11.514285714285714, "grad_norm": 0.9921875, "learning_rate": 8.039005241812893e-05, "loss": 0.1232, "step": 15717 }, { "epoch": 11.515018315018315, "grad_norm": 0.9921875, "learning_rate": 8.038732117343515e-05, "loss": 0.1481, "step": 15718 }, { "epoch": 11.515750915750916, "grad_norm": 1.1640625, "learning_rate": 8.03845897855414e-05, "loss": 0.1159, "step": 15719 }, { "epoch": 11.516483516483516, "grad_norm": 1.3046875, "learning_rate": 8.038185825446079e-05, "loss": 0.1263, "step": 15720 }, { "epoch": 11.517216117216117, "grad_norm": 1.734375, "learning_rate": 8.037912658020637e-05, "loss": 0.101, "step": 15721 }, { "epoch": 11.517948717948718, "grad_norm": 1.34375, "learning_rate": 8.037639476279127e-05, "loss": 0.1022, "step": 15722 }, { "epoch": 11.518681318681319, "grad_norm": 1.1484375, "learning_rate": 8.037366280222852e-05, "loss": 0.1092, "step": 15723 }, { "epoch": 11.51941391941392, "grad_norm": 1.203125, "learning_rate": 8.037093069853127e-05, "loss": 0.0957, "step": 15724 }, { "epoch": 11.52014652014652, "grad_norm": 0.98828125, "learning_rate": 8.036819845171257e-05, "loss": 0.1469, "step": 15725 }, { "epoch": 11.52087912087912, "grad_norm": 1.1953125, "learning_rate": 8.036546606178554e-05, "loss": 0.1654, "step": 15726 }, { "epoch": 11.521611721611722, "grad_norm": 1.4375, "learning_rate": 8.036273352876324e-05, "loss": 0.1234, "step": 15727 }, { "epoch": 11.522344322344322, "grad_norm": 0.88671875, "learning_rate": 8.03600008526588e-05, "loss": 0.1074, "step": 15728 }, { "epoch": 11.523076923076923, "grad_norm": 1.0, "learning_rate": 8.035726803348528e-05, "loss": 0.1263, "step": 15729 }, { "epoch": 11.523809523809524, "grad_norm": 1.21875, "learning_rate": 8.03545350712558e-05, "loss": 0.1403, "step": 15730 }, { "epoch": 11.524542124542124, "grad_norm": 1.09375, "learning_rate": 8.035180196598343e-05, "loss": 0.1045, "step": 15731 }, { "epoch": 11.525274725274725, "grad_norm": 1.359375, "learning_rate": 8.034906871768127e-05, "loss": 0.1618, "step": 15732 }, { "epoch": 11.526007326007326, "grad_norm": 1.4453125, "learning_rate": 8.034633532636245e-05, "loss": 0.1614, "step": 15733 }, { "epoch": 11.526739926739927, "grad_norm": 0.89453125, "learning_rate": 8.034360179204001e-05, "loss": 0.1117, "step": 15734 }, { "epoch": 11.527472527472527, "grad_norm": 1.0078125, "learning_rate": 8.034086811472711e-05, "loss": 0.1271, "step": 15735 }, { "epoch": 11.528205128205128, "grad_norm": 1.375, "learning_rate": 8.033813429443678e-05, "loss": 0.1194, "step": 15736 }, { "epoch": 11.528937728937729, "grad_norm": 1.34375, "learning_rate": 8.033540033118218e-05, "loss": 0.1203, "step": 15737 }, { "epoch": 11.52967032967033, "grad_norm": 0.9765625, "learning_rate": 8.033266622497637e-05, "loss": 0.1168, "step": 15738 }, { "epoch": 11.53040293040293, "grad_norm": 1.125, "learning_rate": 8.032993197583247e-05, "loss": 0.114, "step": 15739 }, { "epoch": 11.531135531135531, "grad_norm": 0.70703125, "learning_rate": 8.032719758376356e-05, "loss": 0.1315, "step": 15740 }, { "epoch": 11.531868131868132, "grad_norm": 1.28125, "learning_rate": 8.032446304878276e-05, "loss": 0.1299, "step": 15741 }, { "epoch": 11.532600732600732, "grad_norm": 0.8671875, "learning_rate": 8.032172837090319e-05, "loss": 0.1114, "step": 15742 }, { "epoch": 11.533333333333333, "grad_norm": 0.90625, "learning_rate": 8.031899355013789e-05, "loss": 0.1237, "step": 15743 }, { "epoch": 11.534065934065934, "grad_norm": 0.6640625, "learning_rate": 8.031625858650003e-05, "loss": 0.116, "step": 15744 }, { "epoch": 11.534798534798535, "grad_norm": 0.6640625, "learning_rate": 8.031352348000267e-05, "loss": 0.1165, "step": 15745 }, { "epoch": 11.535531135531135, "grad_norm": 1.0859375, "learning_rate": 8.031078823065895e-05, "loss": 0.1246, "step": 15746 }, { "epoch": 11.536263736263736, "grad_norm": 0.84765625, "learning_rate": 8.030805283848193e-05, "loss": 0.1079, "step": 15747 }, { "epoch": 11.536996336996337, "grad_norm": 0.98046875, "learning_rate": 8.030531730348478e-05, "loss": 0.0892, "step": 15748 }, { "epoch": 11.537728937728938, "grad_norm": 1.3203125, "learning_rate": 8.030258162568055e-05, "loss": 0.1013, "step": 15749 }, { "epoch": 11.538461538461538, "grad_norm": 1.171875, "learning_rate": 8.029984580508237e-05, "loss": 0.1286, "step": 15750 }, { "epoch": 11.539194139194139, "grad_norm": 1.25, "learning_rate": 8.029710984170334e-05, "loss": 0.1161, "step": 15751 }, { "epoch": 11.53992673992674, "grad_norm": 1.34375, "learning_rate": 8.029437373555658e-05, "loss": 0.1536, "step": 15752 }, { "epoch": 11.54065934065934, "grad_norm": 1.5703125, "learning_rate": 8.029163748665519e-05, "loss": 0.1689, "step": 15753 }, { "epoch": 11.541391941391941, "grad_norm": 0.953125, "learning_rate": 8.02889010950123e-05, "loss": 0.1539, "step": 15754 }, { "epoch": 11.542124542124542, "grad_norm": 1.7421875, "learning_rate": 8.028616456064099e-05, "loss": 0.114, "step": 15755 }, { "epoch": 11.542857142857143, "grad_norm": 1.6171875, "learning_rate": 8.02834278835544e-05, "loss": 0.1405, "step": 15756 }, { "epoch": 11.543589743589743, "grad_norm": 1.6640625, "learning_rate": 8.028069106376562e-05, "loss": 0.122, "step": 15757 }, { "epoch": 11.544322344322344, "grad_norm": 1.890625, "learning_rate": 8.027795410128777e-05, "loss": 0.1555, "step": 15758 }, { "epoch": 11.545054945054945, "grad_norm": 1.734375, "learning_rate": 8.027521699613398e-05, "loss": 0.144, "step": 15759 }, { "epoch": 11.545787545787546, "grad_norm": 1.015625, "learning_rate": 8.027247974831735e-05, "loss": 0.1048, "step": 15760 }, { "epoch": 11.546520146520146, "grad_norm": 1.6484375, "learning_rate": 8.026974235785099e-05, "loss": 0.1401, "step": 15761 }, { "epoch": 11.547252747252747, "grad_norm": 1.4453125, "learning_rate": 8.026700482474802e-05, "loss": 0.1222, "step": 15762 }, { "epoch": 11.547985347985348, "grad_norm": 1.6484375, "learning_rate": 8.026426714902155e-05, "loss": 0.1604, "step": 15763 }, { "epoch": 11.548717948717949, "grad_norm": 0.85546875, "learning_rate": 8.026152933068471e-05, "loss": 0.1542, "step": 15764 }, { "epoch": 11.54945054945055, "grad_norm": 1.3125, "learning_rate": 8.025879136975063e-05, "loss": 0.1373, "step": 15765 }, { "epoch": 11.55018315018315, "grad_norm": 1.4140625, "learning_rate": 8.025605326623239e-05, "loss": 0.154, "step": 15766 }, { "epoch": 11.55091575091575, "grad_norm": 1.0703125, "learning_rate": 8.025331502014314e-05, "loss": 0.1355, "step": 15767 }, { "epoch": 11.551648351648351, "grad_norm": 1.546875, "learning_rate": 8.025057663149598e-05, "loss": 0.1216, "step": 15768 }, { "epoch": 11.552380952380952, "grad_norm": 1.4609375, "learning_rate": 8.024783810030406e-05, "loss": 0.1302, "step": 15769 }, { "epoch": 11.553113553113553, "grad_norm": 0.90234375, "learning_rate": 8.024509942658048e-05, "loss": 0.1263, "step": 15770 }, { "epoch": 11.553846153846154, "grad_norm": 0.6796875, "learning_rate": 8.024236061033834e-05, "loss": 0.1126, "step": 15771 }, { "epoch": 11.554578754578754, "grad_norm": 0.87890625, "learning_rate": 8.023962165159082e-05, "loss": 0.121, "step": 15772 }, { "epoch": 11.555311355311355, "grad_norm": 1.296875, "learning_rate": 8.023688255035098e-05, "loss": 0.1336, "step": 15773 }, { "epoch": 11.556043956043956, "grad_norm": 1.609375, "learning_rate": 8.023414330663199e-05, "loss": 0.1679, "step": 15774 }, { "epoch": 11.556776556776557, "grad_norm": 1.53125, "learning_rate": 8.023140392044696e-05, "loss": 0.124, "step": 15775 }, { "epoch": 11.557509157509157, "grad_norm": 1.640625, "learning_rate": 8.022866439180899e-05, "loss": 0.1328, "step": 15776 }, { "epoch": 11.558241758241758, "grad_norm": 0.75, "learning_rate": 8.022592472073126e-05, "loss": 0.0963, "step": 15777 }, { "epoch": 11.558974358974359, "grad_norm": 1.6640625, "learning_rate": 8.022318490722685e-05, "loss": 0.1389, "step": 15778 }, { "epoch": 11.55970695970696, "grad_norm": 2.3125, "learning_rate": 8.022044495130891e-05, "loss": 0.1646, "step": 15779 }, { "epoch": 11.56043956043956, "grad_norm": 1.203125, "learning_rate": 8.021770485299056e-05, "loss": 0.1299, "step": 15780 }, { "epoch": 11.561172161172161, "grad_norm": 0.96875, "learning_rate": 8.021496461228492e-05, "loss": 0.1313, "step": 15781 }, { "epoch": 11.561904761904762, "grad_norm": 1.75, "learning_rate": 8.021222422920515e-05, "loss": 0.1243, "step": 15782 }, { "epoch": 11.562637362637362, "grad_norm": 1.046875, "learning_rate": 8.020948370376436e-05, "loss": 0.1228, "step": 15783 }, { "epoch": 11.563369963369963, "grad_norm": 1.0, "learning_rate": 8.020674303597565e-05, "loss": 0.1039, "step": 15784 }, { "epoch": 11.564102564102564, "grad_norm": 1.265625, "learning_rate": 8.020400222585221e-05, "loss": 0.1269, "step": 15785 }, { "epoch": 11.564835164835165, "grad_norm": 1.3359375, "learning_rate": 8.020126127340714e-05, "loss": 0.1529, "step": 15786 }, { "epoch": 11.565567765567765, "grad_norm": 1.1484375, "learning_rate": 8.019852017865356e-05, "loss": 0.1422, "step": 15787 }, { "epoch": 11.566300366300366, "grad_norm": 0.72265625, "learning_rate": 8.019577894160465e-05, "loss": 0.1295, "step": 15788 }, { "epoch": 11.567032967032967, "grad_norm": 1.0625, "learning_rate": 8.019303756227349e-05, "loss": 0.1538, "step": 15789 }, { "epoch": 11.567765567765568, "grad_norm": 0.99609375, "learning_rate": 8.019029604067325e-05, "loss": 0.1413, "step": 15790 }, { "epoch": 11.568498168498168, "grad_norm": 0.95703125, "learning_rate": 8.018755437681706e-05, "loss": 0.1331, "step": 15791 }, { "epoch": 11.569230769230769, "grad_norm": 1.390625, "learning_rate": 8.018481257071803e-05, "loss": 0.117, "step": 15792 }, { "epoch": 11.56996336996337, "grad_norm": 1.0625, "learning_rate": 8.018207062238934e-05, "loss": 0.1224, "step": 15793 }, { "epoch": 11.57069597069597, "grad_norm": 1.328125, "learning_rate": 8.017932853184408e-05, "loss": 0.1394, "step": 15794 }, { "epoch": 11.571428571428571, "grad_norm": 1.1796875, "learning_rate": 8.017658629909544e-05, "loss": 0.1153, "step": 15795 }, { "epoch": 11.572161172161172, "grad_norm": 1.6953125, "learning_rate": 8.017384392415652e-05, "loss": 0.1154, "step": 15796 }, { "epoch": 11.572893772893773, "grad_norm": 1.1328125, "learning_rate": 8.017110140704048e-05, "loss": 0.1227, "step": 15797 }, { "epoch": 11.573626373626373, "grad_norm": 1.0234375, "learning_rate": 8.016835874776046e-05, "loss": 0.1371, "step": 15798 }, { "epoch": 11.574358974358974, "grad_norm": 1.7734375, "learning_rate": 8.016561594632959e-05, "loss": 0.1743, "step": 15799 }, { "epoch": 11.575091575091575, "grad_norm": 0.84375, "learning_rate": 8.016287300276098e-05, "loss": 0.1354, "step": 15800 }, { "epoch": 11.575824175824176, "grad_norm": 1.0234375, "learning_rate": 8.016012991706786e-05, "loss": 0.11, "step": 15801 }, { "epoch": 11.576556776556776, "grad_norm": 1.078125, "learning_rate": 8.015738668926329e-05, "loss": 0.1285, "step": 15802 }, { "epoch": 11.577289377289377, "grad_norm": 1.0625, "learning_rate": 8.015464331936045e-05, "loss": 0.1187, "step": 15803 }, { "epoch": 11.578021978021978, "grad_norm": 1.3203125, "learning_rate": 8.015189980737249e-05, "loss": 0.1141, "step": 15804 }, { "epoch": 11.578754578754578, "grad_norm": 1.28125, "learning_rate": 8.014915615331252e-05, "loss": 0.1269, "step": 15805 }, { "epoch": 11.57948717948718, "grad_norm": 1.046875, "learning_rate": 8.014641235719373e-05, "loss": 0.0984, "step": 15806 }, { "epoch": 11.58021978021978, "grad_norm": 0.77734375, "learning_rate": 8.014366841902924e-05, "loss": 0.1096, "step": 15807 }, { "epoch": 11.58095238095238, "grad_norm": 1.28125, "learning_rate": 8.01409243388322e-05, "loss": 0.1359, "step": 15808 }, { "epoch": 11.581684981684981, "grad_norm": 1.0390625, "learning_rate": 8.013818011661577e-05, "loss": 0.1332, "step": 15809 }, { "epoch": 11.582417582417582, "grad_norm": 1.5390625, "learning_rate": 8.013543575239309e-05, "loss": 0.1308, "step": 15810 }, { "epoch": 11.583150183150183, "grad_norm": 1.15625, "learning_rate": 8.01326912461773e-05, "loss": 0.1377, "step": 15811 }, { "epoch": 11.583882783882784, "grad_norm": 1.1015625, "learning_rate": 8.012994659798155e-05, "loss": 0.1519, "step": 15812 }, { "epoch": 11.584615384615384, "grad_norm": 0.85546875, "learning_rate": 8.012720180781902e-05, "loss": 0.1141, "step": 15813 }, { "epoch": 11.585347985347985, "grad_norm": 1.53125, "learning_rate": 8.012445687570282e-05, "loss": 0.1656, "step": 15814 }, { "epoch": 11.586080586080586, "grad_norm": 1.8046875, "learning_rate": 8.012171180164614e-05, "loss": 0.1603, "step": 15815 }, { "epoch": 11.586813186813187, "grad_norm": 0.96484375, "learning_rate": 8.01189665856621e-05, "loss": 0.1245, "step": 15816 }, { "epoch": 11.587545787545787, "grad_norm": 0.9765625, "learning_rate": 8.011622122776388e-05, "loss": 0.1547, "step": 15817 }, { "epoch": 11.588278388278388, "grad_norm": 0.8125, "learning_rate": 8.011347572796461e-05, "loss": 0.1277, "step": 15818 }, { "epoch": 11.589010989010989, "grad_norm": 1.28125, "learning_rate": 8.011073008627748e-05, "loss": 0.1349, "step": 15819 }, { "epoch": 11.58974358974359, "grad_norm": 0.6640625, "learning_rate": 8.010798430271558e-05, "loss": 0.0918, "step": 15820 }, { "epoch": 11.59047619047619, "grad_norm": 1.171875, "learning_rate": 8.010523837729214e-05, "loss": 0.1392, "step": 15821 }, { "epoch": 11.59120879120879, "grad_norm": 1.0234375, "learning_rate": 8.010249231002028e-05, "loss": 0.1397, "step": 15822 }, { "epoch": 11.591941391941392, "grad_norm": 1.3046875, "learning_rate": 8.009974610091315e-05, "loss": 0.1453, "step": 15823 }, { "epoch": 11.592673992673992, "grad_norm": 0.9453125, "learning_rate": 8.009699974998393e-05, "loss": 0.1264, "step": 15824 }, { "epoch": 11.593406593406593, "grad_norm": 1.6875, "learning_rate": 8.009425325724576e-05, "loss": 0.1442, "step": 15825 }, { "epoch": 11.594139194139194, "grad_norm": 1.0625, "learning_rate": 8.009150662271182e-05, "loss": 0.1178, "step": 15826 }, { "epoch": 11.594871794871795, "grad_norm": 1.65625, "learning_rate": 8.008875984639526e-05, "loss": 0.1281, "step": 15827 }, { "epoch": 11.595604395604395, "grad_norm": 1.3828125, "learning_rate": 8.008601292830921e-05, "loss": 0.1339, "step": 15828 }, { "epoch": 11.596336996336996, "grad_norm": 0.98828125, "learning_rate": 8.008326586846689e-05, "loss": 0.1403, "step": 15829 }, { "epoch": 11.597069597069597, "grad_norm": 1.0859375, "learning_rate": 8.008051866688144e-05, "loss": 0.1351, "step": 15830 }, { "epoch": 11.597802197802197, "grad_norm": 1.1640625, "learning_rate": 8.007777132356597e-05, "loss": 0.1249, "step": 15831 }, { "epoch": 11.598534798534798, "grad_norm": 0.96875, "learning_rate": 8.007502383853371e-05, "loss": 0.131, "step": 15832 }, { "epoch": 11.599267399267399, "grad_norm": 1.375, "learning_rate": 8.007227621179781e-05, "loss": 0.1191, "step": 15833 }, { "epoch": 11.6, "grad_norm": 1.1484375, "learning_rate": 8.006952844337143e-05, "loss": 0.1418, "step": 15834 }, { "epoch": 11.6007326007326, "grad_norm": 1.1015625, "learning_rate": 8.006678053326771e-05, "loss": 0.1447, "step": 15835 }, { "epoch": 11.601465201465201, "grad_norm": 0.97265625, "learning_rate": 8.006403248149987e-05, "loss": 0.1358, "step": 15836 }, { "epoch": 11.602197802197802, "grad_norm": 1.2578125, "learning_rate": 8.006128428808101e-05, "loss": 0.1498, "step": 15837 }, { "epoch": 11.602930402930403, "grad_norm": 1.6015625, "learning_rate": 8.005853595302435e-05, "loss": 0.1367, "step": 15838 }, { "epoch": 11.603663003663003, "grad_norm": 0.9609375, "learning_rate": 8.005578747634305e-05, "loss": 0.1198, "step": 15839 }, { "epoch": 11.604395604395604, "grad_norm": 1.375, "learning_rate": 8.005303885805024e-05, "loss": 0.1339, "step": 15840 }, { "epoch": 11.605128205128205, "grad_norm": 1.328125, "learning_rate": 8.005029009815914e-05, "loss": 0.1837, "step": 15841 }, { "epoch": 11.605860805860805, "grad_norm": 1.4296875, "learning_rate": 8.00475411966829e-05, "loss": 0.1586, "step": 15842 }, { "epoch": 11.606593406593406, "grad_norm": 1.296875, "learning_rate": 8.00447921536347e-05, "loss": 0.1021, "step": 15843 }, { "epoch": 11.607326007326007, "grad_norm": 0.94921875, "learning_rate": 8.004204296902769e-05, "loss": 0.1228, "step": 15844 }, { "epoch": 11.608058608058608, "grad_norm": 1.390625, "learning_rate": 8.003929364287504e-05, "loss": 0.1477, "step": 15845 }, { "epoch": 11.608791208791208, "grad_norm": 0.99609375, "learning_rate": 8.003654417518995e-05, "loss": 0.1455, "step": 15846 }, { "epoch": 11.60952380952381, "grad_norm": 0.953125, "learning_rate": 8.003379456598558e-05, "loss": 0.1551, "step": 15847 }, { "epoch": 11.61025641025641, "grad_norm": 1.203125, "learning_rate": 8.003104481527511e-05, "loss": 0.157, "step": 15848 }, { "epoch": 11.61098901098901, "grad_norm": 0.93359375, "learning_rate": 8.002829492307171e-05, "loss": 0.1295, "step": 15849 }, { "epoch": 11.611721611721611, "grad_norm": 1.046875, "learning_rate": 8.002554488938855e-05, "loss": 0.143, "step": 15850 }, { "epoch": 11.612454212454212, "grad_norm": 0.98828125, "learning_rate": 8.002279471423881e-05, "loss": 0.1293, "step": 15851 }, { "epoch": 11.613186813186813, "grad_norm": 1.2421875, "learning_rate": 8.002004439763568e-05, "loss": 0.1098, "step": 15852 }, { "epoch": 11.613919413919414, "grad_norm": 0.79296875, "learning_rate": 8.001729393959231e-05, "loss": 0.1113, "step": 15853 }, { "epoch": 11.614652014652014, "grad_norm": 1.0859375, "learning_rate": 8.00145433401219e-05, "loss": 0.1439, "step": 15854 }, { "epoch": 11.615384615384615, "grad_norm": 1.6171875, "learning_rate": 8.001179259923763e-05, "loss": 0.1396, "step": 15855 }, { "epoch": 11.616117216117216, "grad_norm": 0.94921875, "learning_rate": 8.00090417169527e-05, "loss": 0.1061, "step": 15856 }, { "epoch": 11.616849816849816, "grad_norm": 0.81640625, "learning_rate": 8.000629069328022e-05, "loss": 0.1196, "step": 15857 }, { "epoch": 11.617582417582417, "grad_norm": 1.375, "learning_rate": 8.000353952823344e-05, "loss": 0.1438, "step": 15858 }, { "epoch": 11.618315018315018, "grad_norm": 1.3515625, "learning_rate": 8.000078822182549e-05, "loss": 0.1504, "step": 15859 }, { "epoch": 11.619047619047619, "grad_norm": 1.4140625, "learning_rate": 7.99980367740696e-05, "loss": 0.1467, "step": 15860 }, { "epoch": 11.61978021978022, "grad_norm": 1.1484375, "learning_rate": 7.999528518497893e-05, "loss": 0.1352, "step": 15861 }, { "epoch": 11.62051282051282, "grad_norm": 0.734375, "learning_rate": 7.999253345456666e-05, "loss": 0.1103, "step": 15862 }, { "epoch": 11.62124542124542, "grad_norm": 0.78515625, "learning_rate": 7.998978158284599e-05, "loss": 0.1271, "step": 15863 }, { "epoch": 11.621978021978022, "grad_norm": 1.5703125, "learning_rate": 7.998702956983008e-05, "loss": 0.1299, "step": 15864 }, { "epoch": 11.622710622710622, "grad_norm": 0.9140625, "learning_rate": 7.998427741553215e-05, "loss": 0.1246, "step": 15865 }, { "epoch": 11.623443223443223, "grad_norm": 1.2890625, "learning_rate": 7.998152511996537e-05, "loss": 0.1231, "step": 15866 }, { "epoch": 11.624175824175824, "grad_norm": 1.0859375, "learning_rate": 7.99787726831429e-05, "loss": 0.1166, "step": 15867 }, { "epoch": 11.624908424908424, "grad_norm": 0.5625, "learning_rate": 7.997602010507797e-05, "loss": 0.1363, "step": 15868 }, { "epoch": 11.625641025641025, "grad_norm": 0.8671875, "learning_rate": 7.997326738578376e-05, "loss": 0.1316, "step": 15869 }, { "epoch": 11.626373626373626, "grad_norm": 1.2265625, "learning_rate": 7.997051452527344e-05, "loss": 0.1229, "step": 15870 }, { "epoch": 11.627106227106227, "grad_norm": 1.3046875, "learning_rate": 7.996776152356021e-05, "loss": 0.1362, "step": 15871 }, { "epoch": 11.627838827838827, "grad_norm": 1.6171875, "learning_rate": 7.996500838065726e-05, "loss": 0.1076, "step": 15872 }, { "epoch": 11.628571428571428, "grad_norm": 1.0234375, "learning_rate": 7.996225509657779e-05, "loss": 0.1409, "step": 15873 }, { "epoch": 11.629304029304029, "grad_norm": 1.8125, "learning_rate": 7.995950167133496e-05, "loss": 0.1217, "step": 15874 }, { "epoch": 11.63003663003663, "grad_norm": 1.6328125, "learning_rate": 7.995674810494201e-05, "loss": 0.183, "step": 15875 }, { "epoch": 11.63076923076923, "grad_norm": 0.80078125, "learning_rate": 7.995399439741211e-05, "loss": 0.1257, "step": 15876 }, { "epoch": 11.631501831501831, "grad_norm": 1.734375, "learning_rate": 7.995124054875844e-05, "loss": 0.1402, "step": 15877 }, { "epoch": 11.632234432234432, "grad_norm": 2.21875, "learning_rate": 7.994848655899421e-05, "loss": 0.1529, "step": 15878 }, { "epoch": 11.632967032967032, "grad_norm": 1.140625, "learning_rate": 7.994573242813262e-05, "loss": 0.1279, "step": 15879 }, { "epoch": 11.633699633699633, "grad_norm": 0.96484375, "learning_rate": 7.994297815618687e-05, "loss": 0.1499, "step": 15880 }, { "epoch": 11.634432234432234, "grad_norm": 1.3359375, "learning_rate": 7.994022374317013e-05, "loss": 0.1356, "step": 15881 }, { "epoch": 11.635164835164835, "grad_norm": 1.296875, "learning_rate": 7.993746918909562e-05, "loss": 0.1632, "step": 15882 }, { "epoch": 11.635897435897435, "grad_norm": 0.859375, "learning_rate": 7.993471449397653e-05, "loss": 0.1266, "step": 15883 }, { "epoch": 11.636630036630036, "grad_norm": 1.1875, "learning_rate": 7.993195965782606e-05, "loss": 0.1158, "step": 15884 }, { "epoch": 11.637362637362637, "grad_norm": 0.84375, "learning_rate": 7.99292046806574e-05, "loss": 0.111, "step": 15885 }, { "epoch": 11.638095238095238, "grad_norm": 1.0390625, "learning_rate": 7.992644956248378e-05, "loss": 0.14, "step": 15886 }, { "epoch": 11.638827838827838, "grad_norm": 0.71875, "learning_rate": 7.992369430331837e-05, "loss": 0.1058, "step": 15887 }, { "epoch": 11.639560439560439, "grad_norm": 1.2578125, "learning_rate": 7.992093890317437e-05, "loss": 0.1147, "step": 15888 }, { "epoch": 11.64029304029304, "grad_norm": 0.80078125, "learning_rate": 7.991818336206501e-05, "loss": 0.1343, "step": 15889 }, { "epoch": 11.64102564102564, "grad_norm": 1.6328125, "learning_rate": 7.991542768000347e-05, "loss": 0.1671, "step": 15890 }, { "epoch": 11.641758241758241, "grad_norm": 1.0625, "learning_rate": 7.991267185700296e-05, "loss": 0.1151, "step": 15891 }, { "epoch": 11.642490842490842, "grad_norm": 1.3125, "learning_rate": 7.990991589307668e-05, "loss": 0.1352, "step": 15892 }, { "epoch": 11.643223443223443, "grad_norm": 1.1640625, "learning_rate": 7.990715978823784e-05, "loss": 0.1436, "step": 15893 }, { "epoch": 11.643956043956043, "grad_norm": 0.89453125, "learning_rate": 7.990440354249965e-05, "loss": 0.1264, "step": 15894 }, { "epoch": 11.644688644688644, "grad_norm": 0.8828125, "learning_rate": 7.990164715587531e-05, "loss": 0.1355, "step": 15895 }, { "epoch": 11.645421245421245, "grad_norm": 1.03125, "learning_rate": 7.989889062837801e-05, "loss": 0.1508, "step": 15896 }, { "epoch": 11.646153846153846, "grad_norm": 0.8984375, "learning_rate": 7.989613396002098e-05, "loss": 0.1405, "step": 15897 }, { "epoch": 11.646886446886446, "grad_norm": 1.6484375, "learning_rate": 7.989337715081743e-05, "loss": 0.1677, "step": 15898 }, { "epoch": 11.647619047619047, "grad_norm": 0.9375, "learning_rate": 7.989062020078057e-05, "loss": 0.1448, "step": 15899 }, { "epoch": 11.648351648351648, "grad_norm": 0.84375, "learning_rate": 7.988786310992358e-05, "loss": 0.116, "step": 15900 }, { "epoch": 11.649084249084249, "grad_norm": 1.21875, "learning_rate": 7.98851058782597e-05, "loss": 0.1466, "step": 15901 }, { "epoch": 11.64981684981685, "grad_norm": 0.87109375, "learning_rate": 7.988234850580213e-05, "loss": 0.1428, "step": 15902 }, { "epoch": 11.65054945054945, "grad_norm": 1.1015625, "learning_rate": 7.987959099256409e-05, "loss": 0.1708, "step": 15903 }, { "epoch": 11.65128205128205, "grad_norm": 1.5859375, "learning_rate": 7.987683333855877e-05, "loss": 0.1297, "step": 15904 }, { "epoch": 11.652014652014651, "grad_norm": 0.76171875, "learning_rate": 7.987407554379941e-05, "loss": 0.1154, "step": 15905 }, { "epoch": 11.652747252747252, "grad_norm": 1.25, "learning_rate": 7.98713176082992e-05, "loss": 0.1336, "step": 15906 }, { "epoch": 11.653479853479853, "grad_norm": 1.484375, "learning_rate": 7.986855953207137e-05, "loss": 0.1577, "step": 15907 }, { "epoch": 11.654212454212454, "grad_norm": 0.88671875, "learning_rate": 7.986580131512914e-05, "loss": 0.1316, "step": 15908 }, { "epoch": 11.654945054945054, "grad_norm": 1.2734375, "learning_rate": 7.98630429574857e-05, "loss": 0.1161, "step": 15909 }, { "epoch": 11.655677655677655, "grad_norm": 0.8671875, "learning_rate": 7.98602844591543e-05, "loss": 0.1195, "step": 15910 }, { "epoch": 11.656410256410256, "grad_norm": 1.328125, "learning_rate": 7.985752582014813e-05, "loss": 0.1229, "step": 15911 }, { "epoch": 11.657142857142857, "grad_norm": 1.078125, "learning_rate": 7.98547670404804e-05, "loss": 0.1222, "step": 15912 }, { "epoch": 11.657875457875457, "grad_norm": 0.8359375, "learning_rate": 7.985200812016437e-05, "loss": 0.1144, "step": 15913 }, { "epoch": 11.658608058608058, "grad_norm": 1.1640625, "learning_rate": 7.984924905921322e-05, "loss": 0.1333, "step": 15914 }, { "epoch": 11.659340659340659, "grad_norm": 1.15625, "learning_rate": 7.984648985764018e-05, "loss": 0.1156, "step": 15915 }, { "epoch": 11.66007326007326, "grad_norm": 1.15625, "learning_rate": 7.984373051545848e-05, "loss": 0.1323, "step": 15916 }, { "epoch": 11.66080586080586, "grad_norm": 0.90234375, "learning_rate": 7.984097103268134e-05, "loss": 0.1362, "step": 15917 }, { "epoch": 11.661538461538461, "grad_norm": 1.1953125, "learning_rate": 7.983821140932198e-05, "loss": 0.1019, "step": 15918 }, { "epoch": 11.662271062271062, "grad_norm": 1.0859375, "learning_rate": 7.98354516453936e-05, "loss": 0.1339, "step": 15919 }, { "epoch": 11.663003663003662, "grad_norm": 0.765625, "learning_rate": 7.983269174090944e-05, "loss": 0.1233, "step": 15920 }, { "epoch": 11.663736263736263, "grad_norm": 1.65625, "learning_rate": 7.982993169588273e-05, "loss": 0.1754, "step": 15921 }, { "epoch": 11.664468864468864, "grad_norm": 1.1171875, "learning_rate": 7.98271715103267e-05, "loss": 0.1298, "step": 15922 }, { "epoch": 11.665201465201465, "grad_norm": 0.953125, "learning_rate": 7.982441118425456e-05, "loss": 0.1398, "step": 15923 }, { "epoch": 11.665934065934065, "grad_norm": 1.1015625, "learning_rate": 7.982165071767952e-05, "loss": 0.1492, "step": 15924 }, { "epoch": 11.666666666666666, "grad_norm": 1.7421875, "learning_rate": 7.981889011061485e-05, "loss": 0.1373, "step": 15925 }, { "epoch": 11.667399267399267, "grad_norm": 1.6875, "learning_rate": 7.981612936307374e-05, "loss": 0.1658, "step": 15926 }, { "epoch": 11.668131868131868, "grad_norm": 1.046875, "learning_rate": 7.981336847506943e-05, "loss": 0.1291, "step": 15927 }, { "epoch": 11.668864468864468, "grad_norm": 0.921875, "learning_rate": 7.981060744661515e-05, "loss": 0.1223, "step": 15928 }, { "epoch": 11.669597069597069, "grad_norm": 0.796875, "learning_rate": 7.980784627772414e-05, "loss": 0.1515, "step": 15929 }, { "epoch": 11.67032967032967, "grad_norm": 1.46875, "learning_rate": 7.98050849684096e-05, "loss": 0.1228, "step": 15930 }, { "epoch": 11.67106227106227, "grad_norm": 1.0859375, "learning_rate": 7.980232351868479e-05, "loss": 0.1131, "step": 15931 }, { "epoch": 11.671794871794871, "grad_norm": 1.7265625, "learning_rate": 7.979956192856291e-05, "loss": 0.1123, "step": 15932 }, { "epoch": 11.672527472527472, "grad_norm": 1.59375, "learning_rate": 7.979680019805724e-05, "loss": 0.1178, "step": 15933 }, { "epoch": 11.673260073260073, "grad_norm": 2.0, "learning_rate": 7.979403832718098e-05, "loss": 0.1061, "step": 15934 }, { "epoch": 11.673992673992673, "grad_norm": 1.3671875, "learning_rate": 7.979127631594733e-05, "loss": 0.1095, "step": 15935 }, { "epoch": 11.674725274725274, "grad_norm": 1.2421875, "learning_rate": 7.978851416436958e-05, "loss": 0.1228, "step": 15936 }, { "epoch": 11.675457875457875, "grad_norm": 0.9609375, "learning_rate": 7.978575187246095e-05, "loss": 0.1177, "step": 15937 }, { "epoch": 11.676190476190476, "grad_norm": 0.765625, "learning_rate": 7.978298944023466e-05, "loss": 0.1445, "step": 15938 }, { "epoch": 11.676923076923076, "grad_norm": 1.3125, "learning_rate": 7.978022686770398e-05, "loss": 0.1368, "step": 15939 }, { "epoch": 11.677655677655677, "grad_norm": 1.3828125, "learning_rate": 7.977746415488208e-05, "loss": 0.1427, "step": 15940 }, { "epoch": 11.678388278388278, "grad_norm": 0.7890625, "learning_rate": 7.977470130178226e-05, "loss": 0.1236, "step": 15941 }, { "epoch": 11.679120879120878, "grad_norm": 0.9296875, "learning_rate": 7.977193830841773e-05, "loss": 0.0974, "step": 15942 }, { "epoch": 11.67985347985348, "grad_norm": 1.71875, "learning_rate": 7.976917517480175e-05, "loss": 0.1341, "step": 15943 }, { "epoch": 11.68058608058608, "grad_norm": 1.1171875, "learning_rate": 7.976641190094754e-05, "loss": 0.1396, "step": 15944 }, { "epoch": 11.68131868131868, "grad_norm": 1.6953125, "learning_rate": 7.976364848686832e-05, "loss": 0.1464, "step": 15945 }, { "epoch": 11.682051282051281, "grad_norm": 0.8359375, "learning_rate": 7.976088493257738e-05, "loss": 0.1284, "step": 15946 }, { "epoch": 11.682783882783882, "grad_norm": 1.1171875, "learning_rate": 7.975812123808792e-05, "loss": 0.1289, "step": 15947 }, { "epoch": 11.683516483516483, "grad_norm": 0.94921875, "learning_rate": 7.975535740341319e-05, "loss": 0.1316, "step": 15948 }, { "epoch": 11.684249084249084, "grad_norm": 0.8671875, "learning_rate": 7.975259342856646e-05, "loss": 0.1548, "step": 15949 }, { "epoch": 11.684981684981684, "grad_norm": 0.97265625, "learning_rate": 7.974982931356094e-05, "loss": 0.1227, "step": 15950 }, { "epoch": 11.685714285714285, "grad_norm": 1.15625, "learning_rate": 7.974706505840989e-05, "loss": 0.1259, "step": 15951 }, { "epoch": 11.686446886446886, "grad_norm": 1.3359375, "learning_rate": 7.974430066312654e-05, "loss": 0.1288, "step": 15952 }, { "epoch": 11.687179487179487, "grad_norm": 1.2421875, "learning_rate": 7.974153612772415e-05, "loss": 0.1673, "step": 15953 }, { "epoch": 11.687912087912087, "grad_norm": 1.0625, "learning_rate": 7.973877145221597e-05, "loss": 0.141, "step": 15954 }, { "epoch": 11.688644688644688, "grad_norm": 2.03125, "learning_rate": 7.973600663661522e-05, "loss": 0.1349, "step": 15955 }, { "epoch": 11.689377289377289, "grad_norm": 1.6640625, "learning_rate": 7.973324168093518e-05, "loss": 0.1326, "step": 15956 }, { "epoch": 11.69010989010989, "grad_norm": 1.0234375, "learning_rate": 7.973047658518908e-05, "loss": 0.15, "step": 15957 }, { "epoch": 11.69084249084249, "grad_norm": 0.85546875, "learning_rate": 7.972771134939015e-05, "loss": 0.1255, "step": 15958 }, { "epoch": 11.691575091575091, "grad_norm": 1.59375, "learning_rate": 7.972494597355167e-05, "loss": 0.1592, "step": 15959 }, { "epoch": 11.692307692307692, "grad_norm": 0.6171875, "learning_rate": 7.972218045768691e-05, "loss": 0.1286, "step": 15960 }, { "epoch": 11.693040293040292, "grad_norm": 1.1796875, "learning_rate": 7.971941480180906e-05, "loss": 0.162, "step": 15961 }, { "epoch": 11.693772893772893, "grad_norm": 0.69921875, "learning_rate": 7.971664900593142e-05, "loss": 0.1488, "step": 15962 }, { "epoch": 11.694505494505494, "grad_norm": 1.078125, "learning_rate": 7.97138830700672e-05, "loss": 0.0964, "step": 15963 }, { "epoch": 11.695238095238095, "grad_norm": 1.296875, "learning_rate": 7.971111699422969e-05, "loss": 0.1285, "step": 15964 }, { "epoch": 11.695970695970695, "grad_norm": 1.515625, "learning_rate": 7.970835077843213e-05, "loss": 0.1207, "step": 15965 }, { "epoch": 11.696703296703296, "grad_norm": 0.9296875, "learning_rate": 7.970558442268777e-05, "loss": 0.1305, "step": 15966 }, { "epoch": 11.697435897435897, "grad_norm": 1.2109375, "learning_rate": 7.970281792700987e-05, "loss": 0.133, "step": 15967 }, { "epoch": 11.698168498168497, "grad_norm": 0.83984375, "learning_rate": 7.97000512914117e-05, "loss": 0.1388, "step": 15968 }, { "epoch": 11.698901098901098, "grad_norm": 0.62109375, "learning_rate": 7.969728451590646e-05, "loss": 0.1312, "step": 15969 }, { "epoch": 11.699633699633699, "grad_norm": 0.921875, "learning_rate": 7.969451760050747e-05, "loss": 0.1395, "step": 15970 }, { "epoch": 11.7003663003663, "grad_norm": 1.75, "learning_rate": 7.969175054522794e-05, "loss": 0.1095, "step": 15971 }, { "epoch": 11.7010989010989, "grad_norm": 0.8515625, "learning_rate": 7.968898335008119e-05, "loss": 0.1012, "step": 15972 }, { "epoch": 11.701831501831501, "grad_norm": 1.390625, "learning_rate": 7.968621601508041e-05, "loss": 0.1896, "step": 15973 }, { "epoch": 11.702564102564102, "grad_norm": 1.1015625, "learning_rate": 7.968344854023889e-05, "loss": 0.1383, "step": 15974 }, { "epoch": 11.703296703296703, "grad_norm": 0.984375, "learning_rate": 7.96806809255699e-05, "loss": 0.1301, "step": 15975 }, { "epoch": 11.704029304029303, "grad_norm": 0.86328125, "learning_rate": 7.967791317108667e-05, "loss": 0.127, "step": 15976 }, { "epoch": 11.704761904761904, "grad_norm": 0.71875, "learning_rate": 7.967514527680248e-05, "loss": 0.1059, "step": 15977 }, { "epoch": 11.705494505494505, "grad_norm": 0.95703125, "learning_rate": 7.967237724273062e-05, "loss": 0.1376, "step": 15978 }, { "epoch": 11.706227106227105, "grad_norm": 1.140625, "learning_rate": 7.966960906888429e-05, "loss": 0.1179, "step": 15979 }, { "epoch": 11.706959706959706, "grad_norm": 0.88671875, "learning_rate": 7.966684075527682e-05, "loss": 0.1579, "step": 15980 }, { "epoch": 11.707692307692307, "grad_norm": 0.9140625, "learning_rate": 7.966407230192141e-05, "loss": 0.1326, "step": 15981 }, { "epoch": 11.708424908424908, "grad_norm": 1.1640625, "learning_rate": 7.966130370883137e-05, "loss": 0.1617, "step": 15982 }, { "epoch": 11.709157509157508, "grad_norm": 0.99609375, "learning_rate": 7.965853497601995e-05, "loss": 0.1303, "step": 15983 }, { "epoch": 11.70989010989011, "grad_norm": 1.65625, "learning_rate": 7.96557661035004e-05, "loss": 0.1425, "step": 15984 }, { "epoch": 11.71062271062271, "grad_norm": 1.8125, "learning_rate": 7.965299709128604e-05, "loss": 0.1312, "step": 15985 }, { "epoch": 11.71135531135531, "grad_norm": 1.1171875, "learning_rate": 7.965022793939007e-05, "loss": 0.1528, "step": 15986 }, { "epoch": 11.712087912087911, "grad_norm": 1.5, "learning_rate": 7.964745864782582e-05, "loss": 0.1291, "step": 15987 }, { "epoch": 11.712820512820512, "grad_norm": 1.5, "learning_rate": 7.964468921660649e-05, "loss": 0.1396, "step": 15988 }, { "epoch": 11.713553113553113, "grad_norm": 0.828125, "learning_rate": 7.96419196457454e-05, "loss": 0.1324, "step": 15989 }, { "epoch": 11.714285714285714, "grad_norm": 1.4765625, "learning_rate": 7.963914993525581e-05, "loss": 0.1497, "step": 15990 }, { "epoch": 11.715018315018314, "grad_norm": 1.671875, "learning_rate": 7.9636380085151e-05, "loss": 0.1506, "step": 15991 }, { "epoch": 11.715750915750915, "grad_norm": 1.3046875, "learning_rate": 7.96336100954442e-05, "loss": 0.1377, "step": 15992 }, { "epoch": 11.716483516483516, "grad_norm": 1.8984375, "learning_rate": 7.963083996614873e-05, "loss": 0.1373, "step": 15993 }, { "epoch": 11.717216117216116, "grad_norm": 1.203125, "learning_rate": 7.962806969727783e-05, "loss": 0.1252, "step": 15994 }, { "epoch": 11.717948717948717, "grad_norm": 1.4375, "learning_rate": 7.962529928884479e-05, "loss": 0.1342, "step": 15995 }, { "epoch": 11.718681318681318, "grad_norm": 1.0078125, "learning_rate": 7.962252874086288e-05, "loss": 0.1352, "step": 15996 }, { "epoch": 11.719413919413919, "grad_norm": 0.77734375, "learning_rate": 7.961975805334538e-05, "loss": 0.0995, "step": 15997 }, { "epoch": 11.72014652014652, "grad_norm": 0.66796875, "learning_rate": 7.961698722630556e-05, "loss": 0.1058, "step": 15998 }, { "epoch": 11.72087912087912, "grad_norm": 0.88671875, "learning_rate": 7.961421625975668e-05, "loss": 0.1379, "step": 15999 }, { "epoch": 11.72161172161172, "grad_norm": 0.80078125, "learning_rate": 7.961144515371205e-05, "loss": 0.1264, "step": 16000 }, { "epoch": 11.722344322344322, "grad_norm": 1.21875, "learning_rate": 7.960867390818493e-05, "loss": 0.1261, "step": 16001 }, { "epoch": 11.723076923076922, "grad_norm": 1.5234375, "learning_rate": 7.960590252318858e-05, "loss": 0.1453, "step": 16002 }, { "epoch": 11.723809523809523, "grad_norm": 1.125, "learning_rate": 7.960313099873632e-05, "loss": 0.1323, "step": 16003 }, { "epoch": 11.724542124542124, "grad_norm": 1.3515625, "learning_rate": 7.96003593348414e-05, "loss": 0.1476, "step": 16004 }, { "epoch": 11.725274725274724, "grad_norm": 0.66796875, "learning_rate": 7.95975875315171e-05, "loss": 0.1412, "step": 16005 }, { "epoch": 11.726007326007325, "grad_norm": 0.74609375, "learning_rate": 7.959481558877669e-05, "loss": 0.1121, "step": 16006 }, { "epoch": 11.726739926739926, "grad_norm": 0.9609375, "learning_rate": 7.959204350663348e-05, "loss": 0.1292, "step": 16007 }, { "epoch": 11.727472527472527, "grad_norm": 1.6953125, "learning_rate": 7.958927128510075e-05, "loss": 0.1201, "step": 16008 }, { "epoch": 11.728205128205127, "grad_norm": 1.1640625, "learning_rate": 7.958649892419176e-05, "loss": 0.121, "step": 16009 }, { "epoch": 11.728937728937728, "grad_norm": 1.21875, "learning_rate": 7.958372642391981e-05, "loss": 0.12, "step": 16010 }, { "epoch": 11.729670329670329, "grad_norm": 0.91015625, "learning_rate": 7.958095378429817e-05, "loss": 0.1388, "step": 16011 }, { "epoch": 11.73040293040293, "grad_norm": 0.859375, "learning_rate": 7.957818100534015e-05, "loss": 0.1057, "step": 16012 }, { "epoch": 11.73113553113553, "grad_norm": 0.78125, "learning_rate": 7.957540808705901e-05, "loss": 0.1115, "step": 16013 }, { "epoch": 11.731868131868131, "grad_norm": 1.28125, "learning_rate": 7.957263502946805e-05, "loss": 0.1141, "step": 16014 }, { "epoch": 11.732600732600732, "grad_norm": 0.8984375, "learning_rate": 7.956986183258054e-05, "loss": 0.1177, "step": 16015 }, { "epoch": 11.733333333333333, "grad_norm": 0.9453125, "learning_rate": 7.956708849640979e-05, "loss": 0.1443, "step": 16016 }, { "epoch": 11.734065934065933, "grad_norm": 0.97265625, "learning_rate": 7.956431502096908e-05, "loss": 0.1209, "step": 16017 }, { "epoch": 11.734798534798536, "grad_norm": 0.80859375, "learning_rate": 7.956154140627168e-05, "loss": 0.1189, "step": 16018 }, { "epoch": 11.735531135531136, "grad_norm": 0.6640625, "learning_rate": 7.95587676523309e-05, "loss": 0.1213, "step": 16019 }, { "epoch": 11.736263736263737, "grad_norm": 1.0234375, "learning_rate": 7.955599375916004e-05, "loss": 0.1158, "step": 16020 }, { "epoch": 11.736996336996338, "grad_norm": 0.59765625, "learning_rate": 7.955321972677236e-05, "loss": 0.0968, "step": 16021 }, { "epoch": 11.737728937728939, "grad_norm": 1.5546875, "learning_rate": 7.955044555518117e-05, "loss": 0.1342, "step": 16022 }, { "epoch": 11.73846153846154, "grad_norm": 0.8046875, "learning_rate": 7.954767124439975e-05, "loss": 0.134, "step": 16023 }, { "epoch": 11.73919413919414, "grad_norm": 1.0625, "learning_rate": 7.954489679444141e-05, "loss": 0.1161, "step": 16024 }, { "epoch": 11.73992673992674, "grad_norm": 0.71484375, "learning_rate": 7.954212220531945e-05, "loss": 0.143, "step": 16025 }, { "epoch": 11.740659340659342, "grad_norm": 0.84765625, "learning_rate": 7.953934747704713e-05, "loss": 0.1158, "step": 16026 }, { "epoch": 11.741391941391942, "grad_norm": 0.6015625, "learning_rate": 7.953657260963776e-05, "loss": 0.1175, "step": 16027 }, { "epoch": 11.742124542124543, "grad_norm": 0.75390625, "learning_rate": 7.953379760310463e-05, "loss": 0.132, "step": 16028 }, { "epoch": 11.742857142857144, "grad_norm": 0.86328125, "learning_rate": 7.953102245746107e-05, "loss": 0.1293, "step": 16029 }, { "epoch": 11.743589743589745, "grad_norm": 0.90234375, "learning_rate": 7.952824717272035e-05, "loss": 0.1072, "step": 16030 }, { "epoch": 11.744322344322345, "grad_norm": 2.3125, "learning_rate": 7.952547174889576e-05, "loss": 0.1692, "step": 16031 }, { "epoch": 11.745054945054946, "grad_norm": 0.9453125, "learning_rate": 7.95226961860006e-05, "loss": 0.159, "step": 16032 }, { "epoch": 11.745787545787547, "grad_norm": 0.8046875, "learning_rate": 7.951992048404819e-05, "loss": 0.1292, "step": 16033 }, { "epoch": 11.746520146520147, "grad_norm": 0.94921875, "learning_rate": 7.95171446430518e-05, "loss": 0.1261, "step": 16034 }, { "epoch": 11.747252747252748, "grad_norm": 1.09375, "learning_rate": 7.951436866302475e-05, "loss": 0.1271, "step": 16035 }, { "epoch": 11.747985347985349, "grad_norm": 0.9765625, "learning_rate": 7.951159254398032e-05, "loss": 0.1307, "step": 16036 }, { "epoch": 11.74871794871795, "grad_norm": 0.98828125, "learning_rate": 7.950881628593186e-05, "loss": 0.1223, "step": 16037 }, { "epoch": 11.74945054945055, "grad_norm": 1.2265625, "learning_rate": 7.950603988889263e-05, "loss": 0.1483, "step": 16038 }, { "epoch": 11.750183150183151, "grad_norm": 0.92578125, "learning_rate": 7.950326335287591e-05, "loss": 0.116, "step": 16039 }, { "epoch": 11.750915750915752, "grad_norm": 0.86328125, "learning_rate": 7.950048667789505e-05, "loss": 0.1595, "step": 16040 }, { "epoch": 11.751648351648353, "grad_norm": 1.2421875, "learning_rate": 7.949770986396335e-05, "loss": 0.1366, "step": 16041 }, { "epoch": 11.752380952380953, "grad_norm": 0.8828125, "learning_rate": 7.94949329110941e-05, "loss": 0.1031, "step": 16042 }, { "epoch": 11.753113553113554, "grad_norm": 0.9296875, "learning_rate": 7.94921558193006e-05, "loss": 0.1146, "step": 16043 }, { "epoch": 11.753846153846155, "grad_norm": 1.3359375, "learning_rate": 7.948937858859617e-05, "loss": 0.1802, "step": 16044 }, { "epoch": 11.754578754578755, "grad_norm": 1.015625, "learning_rate": 7.94866012189941e-05, "loss": 0.1399, "step": 16045 }, { "epoch": 11.755311355311356, "grad_norm": 1.046875, "learning_rate": 7.94838237105077e-05, "loss": 0.1086, "step": 16046 }, { "epoch": 11.756043956043957, "grad_norm": 1.203125, "learning_rate": 7.948104606315032e-05, "loss": 0.1216, "step": 16047 }, { "epoch": 11.756776556776558, "grad_norm": 1.171875, "learning_rate": 7.947826827693521e-05, "loss": 0.1377, "step": 16048 }, { "epoch": 11.757509157509158, "grad_norm": 1.296875, "learning_rate": 7.947549035187572e-05, "loss": 0.1427, "step": 16049 }, { "epoch": 11.758241758241759, "grad_norm": 1.2265625, "learning_rate": 7.947271228798513e-05, "loss": 0.1217, "step": 16050 }, { "epoch": 11.75897435897436, "grad_norm": 0.9453125, "learning_rate": 7.946993408527677e-05, "loss": 0.1475, "step": 16051 }, { "epoch": 11.75970695970696, "grad_norm": 0.81640625, "learning_rate": 7.946715574376394e-05, "loss": 0.1503, "step": 16052 }, { "epoch": 11.760439560439561, "grad_norm": 1.171875, "learning_rate": 7.946437726345996e-05, "loss": 0.1102, "step": 16053 }, { "epoch": 11.761172161172162, "grad_norm": 1.4921875, "learning_rate": 7.946159864437813e-05, "loss": 0.1341, "step": 16054 }, { "epoch": 11.761904761904763, "grad_norm": 0.78515625, "learning_rate": 7.945881988653179e-05, "loss": 0.1251, "step": 16055 }, { "epoch": 11.762637362637363, "grad_norm": 0.95703125, "learning_rate": 7.945604098993423e-05, "loss": 0.1468, "step": 16056 }, { "epoch": 11.763369963369964, "grad_norm": 1.6640625, "learning_rate": 7.945326195459877e-05, "loss": 0.1318, "step": 16057 }, { "epoch": 11.764102564102565, "grad_norm": 0.9140625, "learning_rate": 7.945048278053874e-05, "loss": 0.1228, "step": 16058 }, { "epoch": 11.764835164835166, "grad_norm": 1.0234375, "learning_rate": 7.944770346776744e-05, "loss": 0.1226, "step": 16059 }, { "epoch": 11.765567765567766, "grad_norm": 1.4609375, "learning_rate": 7.944492401629818e-05, "loss": 0.1545, "step": 16060 }, { "epoch": 11.766300366300367, "grad_norm": 1.1328125, "learning_rate": 7.944214442614428e-05, "loss": 0.1338, "step": 16061 }, { "epoch": 11.767032967032968, "grad_norm": 0.6484375, "learning_rate": 7.943936469731908e-05, "loss": 0.1143, "step": 16062 }, { "epoch": 11.767765567765569, "grad_norm": 1.359375, "learning_rate": 7.943658482983588e-05, "loss": 0.1078, "step": 16063 }, { "epoch": 11.76849816849817, "grad_norm": 0.890625, "learning_rate": 7.943380482370801e-05, "loss": 0.1264, "step": 16064 }, { "epoch": 11.76923076923077, "grad_norm": 0.7265625, "learning_rate": 7.943102467894876e-05, "loss": 0.1192, "step": 16065 }, { "epoch": 11.76996336996337, "grad_norm": 1.390625, "learning_rate": 7.942824439557149e-05, "loss": 0.1064, "step": 16066 }, { "epoch": 11.770695970695972, "grad_norm": 0.94140625, "learning_rate": 7.94254639735895e-05, "loss": 0.1007, "step": 16067 }, { "epoch": 11.771428571428572, "grad_norm": 0.984375, "learning_rate": 7.942268341301612e-05, "loss": 0.146, "step": 16068 }, { "epoch": 11.772161172161173, "grad_norm": 1.0859375, "learning_rate": 7.941990271386468e-05, "loss": 0.1042, "step": 16069 }, { "epoch": 11.772893772893774, "grad_norm": 1.4375, "learning_rate": 7.941712187614847e-05, "loss": 0.1713, "step": 16070 }, { "epoch": 11.773626373626374, "grad_norm": 1.0234375, "learning_rate": 7.941434089988084e-05, "loss": 0.111, "step": 16071 }, { "epoch": 11.774358974358975, "grad_norm": 1.234375, "learning_rate": 7.941155978507513e-05, "loss": 0.1376, "step": 16072 }, { "epoch": 11.775091575091576, "grad_norm": 1.4921875, "learning_rate": 7.940877853174463e-05, "loss": 0.1372, "step": 16073 }, { "epoch": 11.775824175824177, "grad_norm": 1.484375, "learning_rate": 7.94059971399027e-05, "loss": 0.1496, "step": 16074 }, { "epoch": 11.776556776556777, "grad_norm": 1.2578125, "learning_rate": 7.940321560956262e-05, "loss": 0.1389, "step": 16075 }, { "epoch": 11.777289377289378, "grad_norm": 1.1328125, "learning_rate": 7.940043394073776e-05, "loss": 0.1341, "step": 16076 }, { "epoch": 11.778021978021979, "grad_norm": 1.1171875, "learning_rate": 7.939765213344143e-05, "loss": 0.1195, "step": 16077 }, { "epoch": 11.77875457875458, "grad_norm": 1.625, "learning_rate": 7.939487018768697e-05, "loss": 0.1318, "step": 16078 }, { "epoch": 11.77948717948718, "grad_norm": 1.2890625, "learning_rate": 7.939208810348771e-05, "loss": 0.1252, "step": 16079 }, { "epoch": 11.780219780219781, "grad_norm": 1.140625, "learning_rate": 7.938930588085693e-05, "loss": 0.1117, "step": 16080 }, { "epoch": 11.780952380952382, "grad_norm": 1.3203125, "learning_rate": 7.938652351980806e-05, "loss": 0.1357, "step": 16081 }, { "epoch": 11.781684981684982, "grad_norm": 1.3359375, "learning_rate": 7.938374102035433e-05, "loss": 0.1433, "step": 16082 }, { "epoch": 11.782417582417583, "grad_norm": 0.77734375, "learning_rate": 7.938095838250912e-05, "loss": 0.129, "step": 16083 }, { "epoch": 11.783150183150184, "grad_norm": 0.9453125, "learning_rate": 7.937817560628576e-05, "loss": 0.1305, "step": 16084 }, { "epoch": 11.783882783882785, "grad_norm": 1.1171875, "learning_rate": 7.937539269169759e-05, "loss": 0.1488, "step": 16085 }, { "epoch": 11.784615384615385, "grad_norm": 0.90625, "learning_rate": 7.937260963875792e-05, "loss": 0.1292, "step": 16086 }, { "epoch": 11.785347985347986, "grad_norm": 0.87890625, "learning_rate": 7.936982644748013e-05, "loss": 0.1146, "step": 16087 }, { "epoch": 11.786080586080587, "grad_norm": 0.98046875, "learning_rate": 7.93670431178775e-05, "loss": 0.1155, "step": 16088 }, { "epoch": 11.786813186813188, "grad_norm": 0.7265625, "learning_rate": 7.936425964996339e-05, "loss": 0.1096, "step": 16089 }, { "epoch": 11.787545787545788, "grad_norm": 1.234375, "learning_rate": 7.936147604375113e-05, "loss": 0.1369, "step": 16090 }, { "epoch": 11.788278388278389, "grad_norm": 0.8125, "learning_rate": 7.935869229925408e-05, "loss": 0.136, "step": 16091 }, { "epoch": 11.78901098901099, "grad_norm": 0.8359375, "learning_rate": 7.935590841648556e-05, "loss": 0.1397, "step": 16092 }, { "epoch": 11.78974358974359, "grad_norm": 1.0859375, "learning_rate": 7.93531243954589e-05, "loss": 0.1187, "step": 16093 }, { "epoch": 11.790476190476191, "grad_norm": 0.9296875, "learning_rate": 7.935034023618745e-05, "loss": 0.1166, "step": 16094 }, { "epoch": 11.791208791208792, "grad_norm": 1.0625, "learning_rate": 7.934755593868456e-05, "loss": 0.1631, "step": 16095 }, { "epoch": 11.791941391941393, "grad_norm": 0.94140625, "learning_rate": 7.934477150296354e-05, "loss": 0.1363, "step": 16096 }, { "epoch": 11.792673992673993, "grad_norm": 1.1328125, "learning_rate": 7.934198692903777e-05, "loss": 0.1324, "step": 16097 }, { "epoch": 11.793406593406594, "grad_norm": 0.74609375, "learning_rate": 7.933920221692058e-05, "loss": 0.1169, "step": 16098 }, { "epoch": 11.794139194139195, "grad_norm": 0.80078125, "learning_rate": 7.933641736662528e-05, "loss": 0.1239, "step": 16099 }, { "epoch": 11.794871794871796, "grad_norm": 1.15625, "learning_rate": 7.933363237816526e-05, "loss": 0.1398, "step": 16100 }, { "epoch": 11.795604395604396, "grad_norm": 1.203125, "learning_rate": 7.933084725155381e-05, "loss": 0.1307, "step": 16101 }, { "epoch": 11.796336996336997, "grad_norm": 1.7890625, "learning_rate": 7.932806198680433e-05, "loss": 0.1437, "step": 16102 }, { "epoch": 11.797069597069598, "grad_norm": 0.8203125, "learning_rate": 7.932527658393015e-05, "loss": 0.1165, "step": 16103 }, { "epoch": 11.797802197802199, "grad_norm": 0.8359375, "learning_rate": 7.93224910429446e-05, "loss": 0.1625, "step": 16104 }, { "epoch": 11.7985347985348, "grad_norm": 0.8359375, "learning_rate": 7.931970536386102e-05, "loss": 0.1225, "step": 16105 }, { "epoch": 11.7992673992674, "grad_norm": 0.90234375, "learning_rate": 7.931691954669278e-05, "loss": 0.1173, "step": 16106 }, { "epoch": 11.8, "grad_norm": 1.546875, "learning_rate": 7.931413359145323e-05, "loss": 0.1284, "step": 16107 }, { "epoch": 11.800732600732601, "grad_norm": 0.84765625, "learning_rate": 7.931134749815568e-05, "loss": 0.137, "step": 16108 }, { "epoch": 11.801465201465202, "grad_norm": 1.0234375, "learning_rate": 7.930856126681352e-05, "loss": 0.1289, "step": 16109 }, { "epoch": 11.802197802197803, "grad_norm": 1.390625, "learning_rate": 7.930577489744008e-05, "loss": 0.1731, "step": 16110 }, { "epoch": 11.802930402930404, "grad_norm": 1.3125, "learning_rate": 7.930298839004873e-05, "loss": 0.1445, "step": 16111 }, { "epoch": 11.803663003663004, "grad_norm": 1.0546875, "learning_rate": 7.93002017446528e-05, "loss": 0.113, "step": 16112 }, { "epoch": 11.804395604395605, "grad_norm": 0.703125, "learning_rate": 7.929741496126565e-05, "loss": 0.108, "step": 16113 }, { "epoch": 11.805128205128206, "grad_norm": 1.5234375, "learning_rate": 7.929462803990063e-05, "loss": 0.1276, "step": 16114 }, { "epoch": 11.805860805860807, "grad_norm": 0.8671875, "learning_rate": 7.92918409805711e-05, "loss": 0.1164, "step": 16115 }, { "epoch": 11.806593406593407, "grad_norm": 1.46875, "learning_rate": 7.92890537832904e-05, "loss": 0.1601, "step": 16116 }, { "epoch": 11.807326007326008, "grad_norm": 0.90234375, "learning_rate": 7.928626644807188e-05, "loss": 0.1247, "step": 16117 }, { "epoch": 11.808058608058609, "grad_norm": 0.76953125, "learning_rate": 7.928347897492894e-05, "loss": 0.1413, "step": 16118 }, { "epoch": 11.80879120879121, "grad_norm": 1.7734375, "learning_rate": 7.92806913638749e-05, "loss": 0.157, "step": 16119 }, { "epoch": 11.80952380952381, "grad_norm": 1.171875, "learning_rate": 7.927790361492309e-05, "loss": 0.1297, "step": 16120 }, { "epoch": 11.810256410256411, "grad_norm": 1.3984375, "learning_rate": 7.92751157280869e-05, "loss": 0.1126, "step": 16121 }, { "epoch": 11.810989010989012, "grad_norm": 1.234375, "learning_rate": 7.927232770337969e-05, "loss": 0.1271, "step": 16122 }, { "epoch": 11.811721611721612, "grad_norm": 1.0078125, "learning_rate": 7.926953954081483e-05, "loss": 0.1284, "step": 16123 }, { "epoch": 11.812454212454213, "grad_norm": 1.0625, "learning_rate": 7.926675124040565e-05, "loss": 0.1479, "step": 16124 }, { "epoch": 11.813186813186814, "grad_norm": 0.8046875, "learning_rate": 7.926396280216553e-05, "loss": 0.1423, "step": 16125 }, { "epoch": 11.813919413919415, "grad_norm": 1.2421875, "learning_rate": 7.926117422610781e-05, "loss": 0.128, "step": 16126 }, { "epoch": 11.814652014652015, "grad_norm": 0.98828125, "learning_rate": 7.925838551224586e-05, "loss": 0.1352, "step": 16127 }, { "epoch": 11.815384615384616, "grad_norm": 1.0625, "learning_rate": 7.925559666059306e-05, "loss": 0.1174, "step": 16128 }, { "epoch": 11.816117216117217, "grad_norm": 1.3125, "learning_rate": 7.925280767116273e-05, "loss": 0.1344, "step": 16129 }, { "epoch": 11.816849816849818, "grad_norm": 1.1640625, "learning_rate": 7.925001854396827e-05, "loss": 0.154, "step": 16130 }, { "epoch": 11.817582417582418, "grad_norm": 1.015625, "learning_rate": 7.924722927902306e-05, "loss": 0.1229, "step": 16131 }, { "epoch": 11.818315018315019, "grad_norm": 0.98828125, "learning_rate": 7.924443987634042e-05, "loss": 0.1207, "step": 16132 }, { "epoch": 11.81904761904762, "grad_norm": 0.62109375, "learning_rate": 7.924165033593373e-05, "loss": 0.117, "step": 16133 }, { "epoch": 11.81978021978022, "grad_norm": 1.09375, "learning_rate": 7.923886065781637e-05, "loss": 0.155, "step": 16134 }, { "epoch": 11.820512820512821, "grad_norm": 0.81640625, "learning_rate": 7.923607084200167e-05, "loss": 0.1081, "step": 16135 }, { "epoch": 11.821245421245422, "grad_norm": 1.09375, "learning_rate": 7.923328088850303e-05, "loss": 0.1398, "step": 16136 }, { "epoch": 11.821978021978023, "grad_norm": 0.859375, "learning_rate": 7.92304907973338e-05, "loss": 0.095, "step": 16137 }, { "epoch": 11.822710622710623, "grad_norm": 0.765625, "learning_rate": 7.922770056850737e-05, "loss": 0.1475, "step": 16138 }, { "epoch": 11.823443223443224, "grad_norm": 1.125, "learning_rate": 7.922491020203711e-05, "loss": 0.1509, "step": 16139 }, { "epoch": 11.824175824175825, "grad_norm": 0.97265625, "learning_rate": 7.922211969793634e-05, "loss": 0.1393, "step": 16140 }, { "epoch": 11.824908424908426, "grad_norm": 0.57421875, "learning_rate": 7.92193290562185e-05, "loss": 0.103, "step": 16141 }, { "epoch": 11.825641025641026, "grad_norm": 1.2734375, "learning_rate": 7.921653827689691e-05, "loss": 0.1565, "step": 16142 }, { "epoch": 11.826373626373627, "grad_norm": 0.9609375, "learning_rate": 7.921374735998495e-05, "loss": 0.1092, "step": 16143 }, { "epoch": 11.827106227106228, "grad_norm": 0.8515625, "learning_rate": 7.921095630549602e-05, "loss": 0.1479, "step": 16144 }, { "epoch": 11.827838827838828, "grad_norm": 0.69921875, "learning_rate": 7.920816511344347e-05, "loss": 0.1289, "step": 16145 }, { "epoch": 11.82857142857143, "grad_norm": 0.98828125, "learning_rate": 7.920537378384067e-05, "loss": 0.1397, "step": 16146 }, { "epoch": 11.82930402930403, "grad_norm": 0.9140625, "learning_rate": 7.9202582316701e-05, "loss": 0.125, "step": 16147 }, { "epoch": 11.83003663003663, "grad_norm": 0.796875, "learning_rate": 7.919979071203784e-05, "loss": 0.1134, "step": 16148 }, { "epoch": 11.830769230769231, "grad_norm": 0.859375, "learning_rate": 7.919699896986454e-05, "loss": 0.1082, "step": 16149 }, { "epoch": 11.831501831501832, "grad_norm": 1.1171875, "learning_rate": 7.919420709019452e-05, "loss": 0.1406, "step": 16150 }, { "epoch": 11.832234432234433, "grad_norm": 0.62890625, "learning_rate": 7.919141507304113e-05, "loss": 0.1376, "step": 16151 }, { "epoch": 11.832967032967034, "grad_norm": 1.3671875, "learning_rate": 7.918862291841776e-05, "loss": 0.1485, "step": 16152 }, { "epoch": 11.833699633699634, "grad_norm": 1.0625, "learning_rate": 7.918583062633777e-05, "loss": 0.1617, "step": 16153 }, { "epoch": 11.834432234432235, "grad_norm": 1.4609375, "learning_rate": 7.918303819681454e-05, "loss": 0.1196, "step": 16154 }, { "epoch": 11.835164835164836, "grad_norm": 0.85546875, "learning_rate": 7.918024562986147e-05, "loss": 0.1474, "step": 16155 }, { "epoch": 11.835897435897436, "grad_norm": 1.3828125, "learning_rate": 7.917745292549193e-05, "loss": 0.1216, "step": 16156 }, { "epoch": 11.836630036630037, "grad_norm": 1.5234375, "learning_rate": 7.91746600837193e-05, "loss": 0.1607, "step": 16157 }, { "epoch": 11.837362637362638, "grad_norm": 1.2421875, "learning_rate": 7.917186710455696e-05, "loss": 0.1279, "step": 16158 }, { "epoch": 11.838095238095239, "grad_norm": 1.0625, "learning_rate": 7.916907398801827e-05, "loss": 0.1445, "step": 16159 }, { "epoch": 11.83882783882784, "grad_norm": 0.92578125, "learning_rate": 7.916628073411665e-05, "loss": 0.1145, "step": 16160 }, { "epoch": 11.83956043956044, "grad_norm": 1.015625, "learning_rate": 7.916348734286547e-05, "loss": 0.1158, "step": 16161 }, { "epoch": 11.84029304029304, "grad_norm": 0.6796875, "learning_rate": 7.91606938142781e-05, "loss": 0.1198, "step": 16162 }, { "epoch": 11.841025641025642, "grad_norm": 0.79296875, "learning_rate": 7.915790014836797e-05, "loss": 0.1249, "step": 16163 }, { "epoch": 11.841758241758242, "grad_norm": 1.0, "learning_rate": 7.91551063451484e-05, "loss": 0.1058, "step": 16164 }, { "epoch": 11.842490842490843, "grad_norm": 1.1953125, "learning_rate": 7.915231240463283e-05, "loss": 0.1116, "step": 16165 }, { "epoch": 11.843223443223444, "grad_norm": 0.890625, "learning_rate": 7.914951832683462e-05, "loss": 0.1351, "step": 16166 }, { "epoch": 11.843956043956045, "grad_norm": 1.078125, "learning_rate": 7.914672411176716e-05, "loss": 0.151, "step": 16167 }, { "epoch": 11.844688644688645, "grad_norm": 1.375, "learning_rate": 7.914392975944384e-05, "loss": 0.1198, "step": 16168 }, { "epoch": 11.845421245421246, "grad_norm": 0.796875, "learning_rate": 7.914113526987804e-05, "loss": 0.124, "step": 16169 }, { "epoch": 11.846153846153847, "grad_norm": 0.74609375, "learning_rate": 7.913834064308317e-05, "loss": 0.1276, "step": 16170 }, { "epoch": 11.846886446886447, "grad_norm": 1.46875, "learning_rate": 7.913554587907261e-05, "loss": 0.1344, "step": 16171 }, { "epoch": 11.847619047619048, "grad_norm": 1.59375, "learning_rate": 7.913275097785975e-05, "loss": 0.1655, "step": 16172 }, { "epoch": 11.848351648351649, "grad_norm": 1.5234375, "learning_rate": 7.912995593945798e-05, "loss": 0.1775, "step": 16173 }, { "epoch": 11.84908424908425, "grad_norm": 1.0703125, "learning_rate": 7.912716076388069e-05, "loss": 0.1564, "step": 16174 }, { "epoch": 11.84981684981685, "grad_norm": 0.98828125, "learning_rate": 7.912436545114127e-05, "loss": 0.1172, "step": 16175 }, { "epoch": 11.850549450549451, "grad_norm": 1.0546875, "learning_rate": 7.912157000125312e-05, "loss": 0.1247, "step": 16176 }, { "epoch": 11.851282051282052, "grad_norm": 1.0625, "learning_rate": 7.911877441422964e-05, "loss": 0.1101, "step": 16177 }, { "epoch": 11.852014652014653, "grad_norm": 1.0703125, "learning_rate": 7.91159786900842e-05, "loss": 0.109, "step": 16178 }, { "epoch": 11.852747252747253, "grad_norm": 0.86328125, "learning_rate": 7.911318282883023e-05, "loss": 0.1419, "step": 16179 }, { "epoch": 11.853479853479854, "grad_norm": 0.87890625, "learning_rate": 7.911038683048109e-05, "loss": 0.1298, "step": 16180 }, { "epoch": 11.854212454212455, "grad_norm": 0.93359375, "learning_rate": 7.910759069505023e-05, "loss": 0.128, "step": 16181 }, { "epoch": 11.854945054945055, "grad_norm": 1.4921875, "learning_rate": 7.910479442255097e-05, "loss": 0.1202, "step": 16182 }, { "epoch": 11.855677655677656, "grad_norm": 1.0546875, "learning_rate": 7.910199801299678e-05, "loss": 0.136, "step": 16183 }, { "epoch": 11.856410256410257, "grad_norm": 1.15625, "learning_rate": 7.909920146640101e-05, "loss": 0.153, "step": 16184 }, { "epoch": 11.857142857142858, "grad_norm": 1.4453125, "learning_rate": 7.909640478277708e-05, "loss": 0.1713, "step": 16185 }, { "epoch": 11.857875457875458, "grad_norm": 1.0234375, "learning_rate": 7.909360796213839e-05, "loss": 0.1228, "step": 16186 }, { "epoch": 11.858608058608059, "grad_norm": 0.92578125, "learning_rate": 7.909081100449834e-05, "loss": 0.1465, "step": 16187 }, { "epoch": 11.85934065934066, "grad_norm": 1.21875, "learning_rate": 7.908801390987032e-05, "loss": 0.1178, "step": 16188 }, { "epoch": 11.86007326007326, "grad_norm": 1.2734375, "learning_rate": 7.908521667826775e-05, "loss": 0.1317, "step": 16189 }, { "epoch": 11.860805860805861, "grad_norm": 1.09375, "learning_rate": 7.908241930970401e-05, "loss": 0.1371, "step": 16190 }, { "epoch": 11.861538461538462, "grad_norm": 1.109375, "learning_rate": 7.907962180419255e-05, "loss": 0.1242, "step": 16191 }, { "epoch": 11.862271062271063, "grad_norm": 0.984375, "learning_rate": 7.907682416174672e-05, "loss": 0.1497, "step": 16192 }, { "epoch": 11.863003663003663, "grad_norm": 0.5859375, "learning_rate": 7.907402638237993e-05, "loss": 0.1129, "step": 16193 }, { "epoch": 11.863736263736264, "grad_norm": 1.34375, "learning_rate": 7.90712284661056e-05, "loss": 0.1432, "step": 16194 }, { "epoch": 11.864468864468865, "grad_norm": 1.4140625, "learning_rate": 7.906843041293714e-05, "loss": 0.1719, "step": 16195 }, { "epoch": 11.865201465201466, "grad_norm": 1.03125, "learning_rate": 7.906563222288795e-05, "loss": 0.1468, "step": 16196 }, { "epoch": 11.865934065934066, "grad_norm": 1.3359375, "learning_rate": 7.906283389597146e-05, "loss": 0.1874, "step": 16197 }, { "epoch": 11.866666666666667, "grad_norm": 0.953125, "learning_rate": 7.906003543220104e-05, "loss": 0.1482, "step": 16198 }, { "epoch": 11.867399267399268, "grad_norm": 0.66015625, "learning_rate": 7.905723683159012e-05, "loss": 0.1061, "step": 16199 }, { "epoch": 11.868131868131869, "grad_norm": 1.0390625, "learning_rate": 7.90544380941521e-05, "loss": 0.0998, "step": 16200 }, { "epoch": 11.86886446886447, "grad_norm": 1.2109375, "learning_rate": 7.905163921990037e-05, "loss": 0.1226, "step": 16201 }, { "epoch": 11.86959706959707, "grad_norm": 0.7109375, "learning_rate": 7.904884020884841e-05, "loss": 0.118, "step": 16202 }, { "epoch": 11.87032967032967, "grad_norm": 1.0, "learning_rate": 7.904604106100956e-05, "loss": 0.1221, "step": 16203 }, { "epoch": 11.871062271062272, "grad_norm": 1.046875, "learning_rate": 7.904324177639725e-05, "loss": 0.113, "step": 16204 }, { "epoch": 11.871794871794872, "grad_norm": 1.140625, "learning_rate": 7.904044235502491e-05, "loss": 0.1191, "step": 16205 }, { "epoch": 11.872527472527473, "grad_norm": 1.171875, "learning_rate": 7.903764279690594e-05, "loss": 0.1211, "step": 16206 }, { "epoch": 11.873260073260074, "grad_norm": 1.34375, "learning_rate": 7.903484310205375e-05, "loss": 0.1247, "step": 16207 }, { "epoch": 11.873992673992674, "grad_norm": 1.8515625, "learning_rate": 7.903204327048175e-05, "loss": 0.1392, "step": 16208 }, { "epoch": 11.874725274725275, "grad_norm": 1.015625, "learning_rate": 7.902924330220338e-05, "loss": 0.0923, "step": 16209 }, { "epoch": 11.875457875457876, "grad_norm": 0.98828125, "learning_rate": 7.902644319723203e-05, "loss": 0.1078, "step": 16210 }, { "epoch": 11.876190476190477, "grad_norm": 1.0546875, "learning_rate": 7.902364295558113e-05, "loss": 0.1117, "step": 16211 }, { "epoch": 11.876923076923077, "grad_norm": 1.453125, "learning_rate": 7.90208425772641e-05, "loss": 0.1565, "step": 16212 }, { "epoch": 11.877655677655678, "grad_norm": 1.3046875, "learning_rate": 7.901804206229434e-05, "loss": 0.1168, "step": 16213 }, { "epoch": 11.878388278388279, "grad_norm": 1.21875, "learning_rate": 7.90152414106853e-05, "loss": 0.1414, "step": 16214 }, { "epoch": 11.87912087912088, "grad_norm": 1.0, "learning_rate": 7.901244062245035e-05, "loss": 0.1344, "step": 16215 }, { "epoch": 11.87985347985348, "grad_norm": 1.0078125, "learning_rate": 7.900963969760294e-05, "loss": 0.1398, "step": 16216 }, { "epoch": 11.880586080586081, "grad_norm": 1.84375, "learning_rate": 7.90068386361565e-05, "loss": 0.1568, "step": 16217 }, { "epoch": 11.881318681318682, "grad_norm": 0.78515625, "learning_rate": 7.900403743812444e-05, "loss": 0.1162, "step": 16218 }, { "epoch": 11.882051282051282, "grad_norm": 1.265625, "learning_rate": 7.900123610352016e-05, "loss": 0.1345, "step": 16219 }, { "epoch": 11.882783882783883, "grad_norm": 1.796875, "learning_rate": 7.899843463235713e-05, "loss": 0.1545, "step": 16220 }, { "epoch": 11.883516483516484, "grad_norm": 1.6953125, "learning_rate": 7.89956330246487e-05, "loss": 0.1306, "step": 16221 }, { "epoch": 11.884249084249085, "grad_norm": 1.0625, "learning_rate": 7.899283128040839e-05, "loss": 0.1209, "step": 16222 }, { "epoch": 11.884981684981685, "grad_norm": 0.87890625, "learning_rate": 7.899002939964953e-05, "loss": 0.1409, "step": 16223 }, { "epoch": 11.885714285714286, "grad_norm": 0.87109375, "learning_rate": 7.89872273823856e-05, "loss": 0.0993, "step": 16224 }, { "epoch": 11.886446886446887, "grad_norm": 0.61328125, "learning_rate": 7.898442522863002e-05, "loss": 0.0945, "step": 16225 }, { "epoch": 11.887179487179488, "grad_norm": 1.3359375, "learning_rate": 7.898162293839621e-05, "loss": 0.1444, "step": 16226 }, { "epoch": 11.887912087912088, "grad_norm": 0.9140625, "learning_rate": 7.897882051169758e-05, "loss": 0.1402, "step": 16227 }, { "epoch": 11.888644688644689, "grad_norm": 1.0390625, "learning_rate": 7.89760179485476e-05, "loss": 0.1296, "step": 16228 }, { "epoch": 11.88937728937729, "grad_norm": 0.9375, "learning_rate": 7.897321524895962e-05, "loss": 0.1386, "step": 16229 }, { "epoch": 11.89010989010989, "grad_norm": 1.5546875, "learning_rate": 7.897041241294717e-05, "loss": 0.1267, "step": 16230 }, { "epoch": 11.890842490842491, "grad_norm": 1.125, "learning_rate": 7.896760944052359e-05, "loss": 0.1278, "step": 16231 }, { "epoch": 11.891575091575092, "grad_norm": 0.72265625, "learning_rate": 7.896480633170237e-05, "loss": 0.1203, "step": 16232 }, { "epoch": 11.892307692307693, "grad_norm": 1.703125, "learning_rate": 7.896200308649691e-05, "loss": 0.1334, "step": 16233 }, { "epoch": 11.893040293040293, "grad_norm": 1.046875, "learning_rate": 7.895919970492066e-05, "loss": 0.1822, "step": 16234 }, { "epoch": 11.893772893772894, "grad_norm": 1.59375, "learning_rate": 7.895639618698703e-05, "loss": 0.1624, "step": 16235 }, { "epoch": 11.894505494505495, "grad_norm": 0.8515625, "learning_rate": 7.895359253270948e-05, "loss": 0.1489, "step": 16236 }, { "epoch": 11.895238095238096, "grad_norm": 1.7109375, "learning_rate": 7.895078874210141e-05, "loss": 0.1229, "step": 16237 }, { "epoch": 11.895970695970696, "grad_norm": 1.75, "learning_rate": 7.894798481517628e-05, "loss": 0.1728, "step": 16238 }, { "epoch": 11.896703296703297, "grad_norm": 1.765625, "learning_rate": 7.89451807519475e-05, "loss": 0.125, "step": 16239 }, { "epoch": 11.897435897435898, "grad_norm": 1.578125, "learning_rate": 7.894237655242854e-05, "loss": 0.129, "step": 16240 }, { "epoch": 11.898168498168499, "grad_norm": 1.265625, "learning_rate": 7.893957221663281e-05, "loss": 0.1432, "step": 16241 }, { "epoch": 11.8989010989011, "grad_norm": 0.91015625, "learning_rate": 7.893676774457377e-05, "loss": 0.1182, "step": 16242 }, { "epoch": 11.8996336996337, "grad_norm": 0.82421875, "learning_rate": 7.893396313626482e-05, "loss": 0.1089, "step": 16243 }, { "epoch": 11.9003663003663, "grad_norm": 0.96484375, "learning_rate": 7.893115839171941e-05, "loss": 0.1417, "step": 16244 }, { "epoch": 11.901098901098901, "grad_norm": 0.8125, "learning_rate": 7.892835351095101e-05, "loss": 0.1366, "step": 16245 }, { "epoch": 11.901831501831502, "grad_norm": 1.421875, "learning_rate": 7.892554849397303e-05, "loss": 0.1525, "step": 16246 }, { "epoch": 11.902564102564103, "grad_norm": 1.1484375, "learning_rate": 7.89227433407989e-05, "loss": 0.12, "step": 16247 }, { "epoch": 11.903296703296704, "grad_norm": 1.4296875, "learning_rate": 7.891993805144209e-05, "loss": 0.121, "step": 16248 }, { "epoch": 11.904029304029304, "grad_norm": 1.703125, "learning_rate": 7.891713262591602e-05, "loss": 0.1437, "step": 16249 }, { "epoch": 11.904761904761905, "grad_norm": 0.9765625, "learning_rate": 7.891432706423416e-05, "loss": 0.1191, "step": 16250 }, { "epoch": 11.905494505494506, "grad_norm": 1.2265625, "learning_rate": 7.891152136640991e-05, "loss": 0.153, "step": 16251 }, { "epoch": 11.906227106227107, "grad_norm": 1.1484375, "learning_rate": 7.890871553245673e-05, "loss": 0.1293, "step": 16252 }, { "epoch": 11.906959706959707, "grad_norm": 0.76171875, "learning_rate": 7.890590956238809e-05, "loss": 0.0941, "step": 16253 }, { "epoch": 11.907692307692308, "grad_norm": 2.0, "learning_rate": 7.89031034562174e-05, "loss": 0.1001, "step": 16254 }, { "epoch": 11.908424908424909, "grad_norm": 1.171875, "learning_rate": 7.890029721395811e-05, "loss": 0.1507, "step": 16255 }, { "epoch": 11.90915750915751, "grad_norm": 1.2890625, "learning_rate": 7.889749083562368e-05, "loss": 0.1278, "step": 16256 }, { "epoch": 11.90989010989011, "grad_norm": 1.3203125, "learning_rate": 7.889468432122754e-05, "loss": 0.1162, "step": 16257 }, { "epoch": 11.910622710622711, "grad_norm": 1.7890625, "learning_rate": 7.889187767078316e-05, "loss": 0.1329, "step": 16258 }, { "epoch": 11.911355311355312, "grad_norm": 1.0390625, "learning_rate": 7.888907088430397e-05, "loss": 0.1208, "step": 16259 }, { "epoch": 11.912087912087912, "grad_norm": 1.359375, "learning_rate": 7.888626396180341e-05, "loss": 0.1348, "step": 16260 }, { "epoch": 11.912820512820513, "grad_norm": 1.1875, "learning_rate": 7.888345690329497e-05, "loss": 0.1425, "step": 16261 }, { "epoch": 11.913553113553114, "grad_norm": 0.78515625, "learning_rate": 7.888064970879205e-05, "loss": 0.1236, "step": 16262 }, { "epoch": 11.914285714285715, "grad_norm": 1.3671875, "learning_rate": 7.887784237830811e-05, "loss": 0.1351, "step": 16263 }, { "epoch": 11.915018315018315, "grad_norm": 1.1328125, "learning_rate": 7.887503491185664e-05, "loss": 0.1406, "step": 16264 }, { "epoch": 11.915750915750916, "grad_norm": 1.59375, "learning_rate": 7.887222730945103e-05, "loss": 0.1592, "step": 16265 }, { "epoch": 11.916483516483517, "grad_norm": 0.92578125, "learning_rate": 7.886941957110478e-05, "loss": 0.1139, "step": 16266 }, { "epoch": 11.917216117216118, "grad_norm": 1.203125, "learning_rate": 7.886661169683134e-05, "loss": 0.1095, "step": 16267 }, { "epoch": 11.917948717948718, "grad_norm": 1.09375, "learning_rate": 7.886380368664412e-05, "loss": 0.1252, "step": 16268 }, { "epoch": 11.918681318681319, "grad_norm": 0.87890625, "learning_rate": 7.886099554055665e-05, "loss": 0.1149, "step": 16269 }, { "epoch": 11.91941391941392, "grad_norm": 0.7421875, "learning_rate": 7.88581872585823e-05, "loss": 0.1264, "step": 16270 }, { "epoch": 11.92014652014652, "grad_norm": 0.9765625, "learning_rate": 7.885537884073458e-05, "loss": 0.1114, "step": 16271 }, { "epoch": 11.920879120879121, "grad_norm": 1.53125, "learning_rate": 7.885257028702692e-05, "loss": 0.152, "step": 16272 }, { "epoch": 11.921611721611722, "grad_norm": 0.91015625, "learning_rate": 7.884976159747282e-05, "loss": 0.1246, "step": 16273 }, { "epoch": 11.922344322344323, "grad_norm": 0.87109375, "learning_rate": 7.88469527720857e-05, "loss": 0.1272, "step": 16274 }, { "epoch": 11.923076923076923, "grad_norm": 1.3828125, "learning_rate": 7.8844143810879e-05, "loss": 0.1458, "step": 16275 }, { "epoch": 11.923809523809524, "grad_norm": 0.8046875, "learning_rate": 7.88413347138662e-05, "loss": 0.1328, "step": 16276 }, { "epoch": 11.924542124542125, "grad_norm": 0.91015625, "learning_rate": 7.883852548106077e-05, "loss": 0.1102, "step": 16277 }, { "epoch": 11.925274725274726, "grad_norm": 0.80859375, "learning_rate": 7.883571611247616e-05, "loss": 0.12, "step": 16278 }, { "epoch": 11.926007326007326, "grad_norm": 0.87890625, "learning_rate": 7.883290660812586e-05, "loss": 0.1309, "step": 16279 }, { "epoch": 11.926739926739927, "grad_norm": 1.0234375, "learning_rate": 7.883009696802327e-05, "loss": 0.1543, "step": 16280 }, { "epoch": 11.927472527472528, "grad_norm": 1.1875, "learning_rate": 7.882728719218188e-05, "loss": 0.1616, "step": 16281 }, { "epoch": 11.928205128205128, "grad_norm": 1.3125, "learning_rate": 7.882447728061518e-05, "loss": 0.1414, "step": 16282 }, { "epoch": 11.92893772893773, "grad_norm": 0.9375, "learning_rate": 7.88216672333366e-05, "loss": 0.1283, "step": 16283 }, { "epoch": 11.92967032967033, "grad_norm": 0.921875, "learning_rate": 7.881885705035963e-05, "loss": 0.1511, "step": 16284 }, { "epoch": 11.93040293040293, "grad_norm": 0.7265625, "learning_rate": 7.881604673169771e-05, "loss": 0.1393, "step": 16285 }, { "epoch": 11.931135531135531, "grad_norm": 0.6953125, "learning_rate": 7.88132362773643e-05, "loss": 0.1312, "step": 16286 }, { "epoch": 11.931868131868132, "grad_norm": 1.1640625, "learning_rate": 7.88104256873729e-05, "loss": 0.1563, "step": 16287 }, { "epoch": 11.932600732600733, "grad_norm": 0.6328125, "learning_rate": 7.880761496173695e-05, "loss": 0.1136, "step": 16288 }, { "epoch": 11.933333333333334, "grad_norm": 1.296875, "learning_rate": 7.880480410046992e-05, "loss": 0.1468, "step": 16289 }, { "epoch": 11.934065934065934, "grad_norm": 0.80859375, "learning_rate": 7.88019931035853e-05, "loss": 0.1115, "step": 16290 }, { "epoch": 11.934798534798535, "grad_norm": 1.078125, "learning_rate": 7.879918197109652e-05, "loss": 0.1381, "step": 16291 }, { "epoch": 11.935531135531136, "grad_norm": 1.0546875, "learning_rate": 7.879637070301707e-05, "loss": 0.1342, "step": 16292 }, { "epoch": 11.936263736263736, "grad_norm": 0.66015625, "learning_rate": 7.879355929936044e-05, "loss": 0.124, "step": 16293 }, { "epoch": 11.936996336996337, "grad_norm": 1.0703125, "learning_rate": 7.879074776014004e-05, "loss": 0.1364, "step": 16294 }, { "epoch": 11.937728937728938, "grad_norm": 1.3828125, "learning_rate": 7.878793608536942e-05, "loss": 0.1296, "step": 16295 }, { "epoch": 11.938461538461539, "grad_norm": 1.5390625, "learning_rate": 7.8785124275062e-05, "loss": 0.1439, "step": 16296 }, { "epoch": 11.93919413919414, "grad_norm": 1.03125, "learning_rate": 7.878231232923125e-05, "loss": 0.1474, "step": 16297 }, { "epoch": 11.93992673992674, "grad_norm": 1.140625, "learning_rate": 7.877950024789066e-05, "loss": 0.1297, "step": 16298 }, { "epoch": 11.94065934065934, "grad_norm": 1.0234375, "learning_rate": 7.877668803105371e-05, "loss": 0.131, "step": 16299 }, { "epoch": 11.941391941391942, "grad_norm": 1.0390625, "learning_rate": 7.877387567873387e-05, "loss": 0.1337, "step": 16300 }, { "epoch": 11.942124542124542, "grad_norm": 1.0, "learning_rate": 7.877106319094459e-05, "loss": 0.1664, "step": 16301 }, { "epoch": 11.942857142857143, "grad_norm": 0.7109375, "learning_rate": 7.87682505676994e-05, "loss": 0.1049, "step": 16302 }, { "epoch": 11.943589743589744, "grad_norm": 0.93359375, "learning_rate": 7.876543780901171e-05, "loss": 0.1361, "step": 16303 }, { "epoch": 11.944322344322345, "grad_norm": 0.94140625, "learning_rate": 7.876262491489502e-05, "loss": 0.1237, "step": 16304 }, { "epoch": 11.945054945054945, "grad_norm": 0.8984375, "learning_rate": 7.875981188536285e-05, "loss": 0.1704, "step": 16305 }, { "epoch": 11.945787545787546, "grad_norm": 0.703125, "learning_rate": 7.875699872042863e-05, "loss": 0.1277, "step": 16306 }, { "epoch": 11.946520146520147, "grad_norm": 1.1328125, "learning_rate": 7.875418542010584e-05, "loss": 0.1345, "step": 16307 }, { "epoch": 11.947252747252747, "grad_norm": 0.94921875, "learning_rate": 7.875137198440799e-05, "loss": 0.1233, "step": 16308 }, { "epoch": 11.947985347985348, "grad_norm": 1.0234375, "learning_rate": 7.874855841334853e-05, "loss": 0.1537, "step": 16309 }, { "epoch": 11.948717948717949, "grad_norm": 1.484375, "learning_rate": 7.874574470694096e-05, "loss": 0.1532, "step": 16310 }, { "epoch": 11.94945054945055, "grad_norm": 1.265625, "learning_rate": 7.874293086519878e-05, "loss": 0.1288, "step": 16311 }, { "epoch": 11.95018315018315, "grad_norm": 1.0078125, "learning_rate": 7.874011688813541e-05, "loss": 0.1061, "step": 16312 }, { "epoch": 11.950915750915751, "grad_norm": 0.93359375, "learning_rate": 7.873730277576438e-05, "loss": 0.1421, "step": 16313 }, { "epoch": 11.951648351648352, "grad_norm": 1.359375, "learning_rate": 7.873448852809915e-05, "loss": 0.1548, "step": 16314 }, { "epoch": 11.952380952380953, "grad_norm": 1.0, "learning_rate": 7.873167414515324e-05, "loss": 0.1368, "step": 16315 }, { "epoch": 11.953113553113553, "grad_norm": 1.3671875, "learning_rate": 7.872885962694009e-05, "loss": 0.129, "step": 16316 }, { "epoch": 11.953846153846154, "grad_norm": 1.6171875, "learning_rate": 7.872604497347322e-05, "loss": 0.1274, "step": 16317 }, { "epoch": 11.954578754578755, "grad_norm": 1.265625, "learning_rate": 7.87232301847661e-05, "loss": 0.1428, "step": 16318 }, { "epoch": 11.955311355311355, "grad_norm": 1.1796875, "learning_rate": 7.872041526083222e-05, "loss": 0.1126, "step": 16319 }, { "epoch": 11.956043956043956, "grad_norm": 1.0859375, "learning_rate": 7.871760020168508e-05, "loss": 0.1289, "step": 16320 }, { "epoch": 11.956776556776557, "grad_norm": 1.09375, "learning_rate": 7.871478500733814e-05, "loss": 0.1321, "step": 16321 }, { "epoch": 11.957509157509158, "grad_norm": 1.28125, "learning_rate": 7.871196967780489e-05, "loss": 0.1339, "step": 16322 }, { "epoch": 11.958241758241758, "grad_norm": 1.6640625, "learning_rate": 7.870915421309884e-05, "loss": 0.1367, "step": 16323 }, { "epoch": 11.95897435897436, "grad_norm": 0.84765625, "learning_rate": 7.870633861323348e-05, "loss": 0.1317, "step": 16324 }, { "epoch": 11.95970695970696, "grad_norm": 0.78515625, "learning_rate": 7.870352287822228e-05, "loss": 0.1251, "step": 16325 }, { "epoch": 11.96043956043956, "grad_norm": 1.4609375, "learning_rate": 7.870070700807877e-05, "loss": 0.134, "step": 16326 }, { "epoch": 11.961172161172161, "grad_norm": 1.3125, "learning_rate": 7.869789100281639e-05, "loss": 0.1571, "step": 16327 }, { "epoch": 11.961904761904762, "grad_norm": 0.87109375, "learning_rate": 7.869507486244866e-05, "loss": 0.1305, "step": 16328 }, { "epoch": 11.962637362637363, "grad_norm": 1.5703125, "learning_rate": 7.869225858698907e-05, "loss": 0.1262, "step": 16329 }, { "epoch": 11.963369963369964, "grad_norm": 1.234375, "learning_rate": 7.868944217645113e-05, "loss": 0.1371, "step": 16330 }, { "epoch": 11.964102564102564, "grad_norm": 0.83203125, "learning_rate": 7.86866256308483e-05, "loss": 0.119, "step": 16331 }, { "epoch": 11.964835164835165, "grad_norm": 1.0078125, "learning_rate": 7.868380895019411e-05, "loss": 0.1133, "step": 16332 }, { "epoch": 11.965567765567766, "grad_norm": 1.359375, "learning_rate": 7.868099213450202e-05, "loss": 0.1537, "step": 16333 }, { "epoch": 11.966300366300366, "grad_norm": 1.0546875, "learning_rate": 7.867817518378556e-05, "loss": 0.1398, "step": 16334 }, { "epoch": 11.967032967032967, "grad_norm": 0.7890625, "learning_rate": 7.86753580980582e-05, "loss": 0.1058, "step": 16335 }, { "epoch": 11.967765567765568, "grad_norm": 1.84375, "learning_rate": 7.867254087733346e-05, "loss": 0.1609, "step": 16336 }, { "epoch": 11.968498168498169, "grad_norm": 1.0703125, "learning_rate": 7.866972352162483e-05, "loss": 0.1323, "step": 16337 }, { "epoch": 11.96923076923077, "grad_norm": 1.140625, "learning_rate": 7.86669060309458e-05, "loss": 0.1557, "step": 16338 }, { "epoch": 11.96996336996337, "grad_norm": 1.0, "learning_rate": 7.866408840530989e-05, "loss": 0.1231, "step": 16339 }, { "epoch": 11.97069597069597, "grad_norm": 1.4609375, "learning_rate": 7.866127064473057e-05, "loss": 0.1373, "step": 16340 }, { "epoch": 11.971428571428572, "grad_norm": 1.1328125, "learning_rate": 7.865845274922135e-05, "loss": 0.1008, "step": 16341 }, { "epoch": 11.972161172161172, "grad_norm": 1.0546875, "learning_rate": 7.865563471879576e-05, "loss": 0.155, "step": 16342 }, { "epoch": 11.972893772893773, "grad_norm": 0.90234375, "learning_rate": 7.865281655346728e-05, "loss": 0.1128, "step": 16343 }, { "epoch": 11.973626373626374, "grad_norm": 0.8203125, "learning_rate": 7.864999825324942e-05, "loss": 0.1138, "step": 16344 }, { "epoch": 11.974358974358974, "grad_norm": 0.74609375, "learning_rate": 7.864717981815565e-05, "loss": 0.1074, "step": 16345 }, { "epoch": 11.975091575091575, "grad_norm": 1.84375, "learning_rate": 7.864436124819953e-05, "loss": 0.1877, "step": 16346 }, { "epoch": 11.975824175824176, "grad_norm": 0.8671875, "learning_rate": 7.864154254339451e-05, "loss": 0.1223, "step": 16347 }, { "epoch": 11.976556776556777, "grad_norm": 0.81640625, "learning_rate": 7.863872370375414e-05, "loss": 0.1247, "step": 16348 }, { "epoch": 11.977289377289377, "grad_norm": 0.82421875, "learning_rate": 7.863590472929189e-05, "loss": 0.126, "step": 16349 }, { "epoch": 11.978021978021978, "grad_norm": 1.359375, "learning_rate": 7.86330856200213e-05, "loss": 0.1397, "step": 16350 }, { "epoch": 11.978754578754579, "grad_norm": 0.99609375, "learning_rate": 7.863026637595586e-05, "loss": 0.1417, "step": 16351 }, { "epoch": 11.97948717948718, "grad_norm": 1.0390625, "learning_rate": 7.862744699710908e-05, "loss": 0.1316, "step": 16352 }, { "epoch": 11.98021978021978, "grad_norm": 1.7890625, "learning_rate": 7.862462748349446e-05, "loss": 0.1524, "step": 16353 }, { "epoch": 11.980952380952381, "grad_norm": 1.0234375, "learning_rate": 7.862180783512551e-05, "loss": 0.1356, "step": 16354 }, { "epoch": 11.981684981684982, "grad_norm": 1.453125, "learning_rate": 7.861898805201576e-05, "loss": 0.1267, "step": 16355 }, { "epoch": 11.982417582417582, "grad_norm": 1.171875, "learning_rate": 7.861616813417872e-05, "loss": 0.1398, "step": 16356 }, { "epoch": 11.983150183150183, "grad_norm": 1.328125, "learning_rate": 7.861334808162787e-05, "loss": 0.1236, "step": 16357 }, { "epoch": 11.983882783882784, "grad_norm": 0.58984375, "learning_rate": 7.861052789437674e-05, "loss": 0.1261, "step": 16358 }, { "epoch": 11.984615384615385, "grad_norm": 1.03125, "learning_rate": 7.860770757243885e-05, "loss": 0.1601, "step": 16359 }, { "epoch": 11.985347985347985, "grad_norm": 1.421875, "learning_rate": 7.860488711582769e-05, "loss": 0.1086, "step": 16360 }, { "epoch": 11.986080586080586, "grad_norm": 0.75, "learning_rate": 7.860206652455682e-05, "loss": 0.1158, "step": 16361 }, { "epoch": 11.986813186813187, "grad_norm": 1.0546875, "learning_rate": 7.85992457986397e-05, "loss": 0.1277, "step": 16362 }, { "epoch": 11.987545787545788, "grad_norm": 0.8359375, "learning_rate": 7.859642493808989e-05, "loss": 0.1168, "step": 16363 }, { "epoch": 11.988278388278388, "grad_norm": 1.671875, "learning_rate": 7.859360394292086e-05, "loss": 0.1342, "step": 16364 }, { "epoch": 11.989010989010989, "grad_norm": 1.4375, "learning_rate": 7.859078281314615e-05, "loss": 0.1252, "step": 16365 }, { "epoch": 11.98974358974359, "grad_norm": 2.0, "learning_rate": 7.85879615487793e-05, "loss": 0.1438, "step": 16366 }, { "epoch": 11.99047619047619, "grad_norm": 1.015625, "learning_rate": 7.858514014983379e-05, "loss": 0.1278, "step": 16367 }, { "epoch": 11.991208791208791, "grad_norm": 0.625, "learning_rate": 7.858231861632316e-05, "loss": 0.1095, "step": 16368 }, { "epoch": 11.991941391941392, "grad_norm": 1.25, "learning_rate": 7.857949694826092e-05, "loss": 0.142, "step": 16369 }, { "epoch": 11.992673992673993, "grad_norm": 1.3828125, "learning_rate": 7.857667514566061e-05, "loss": 0.1344, "step": 16370 }, { "epoch": 11.993406593406593, "grad_norm": 1.1875, "learning_rate": 7.85738532085357e-05, "loss": 0.1261, "step": 16371 }, { "epoch": 11.994139194139194, "grad_norm": 1.3203125, "learning_rate": 7.857103113689977e-05, "loss": 0.1236, "step": 16372 }, { "epoch": 11.994871794871795, "grad_norm": 1.7578125, "learning_rate": 7.856820893076628e-05, "loss": 0.1355, "step": 16373 }, { "epoch": 11.995604395604396, "grad_norm": 1.265625, "learning_rate": 7.856538659014883e-05, "loss": 0.1328, "step": 16374 }, { "epoch": 11.996336996336996, "grad_norm": 0.9140625, "learning_rate": 7.856256411506088e-05, "loss": 0.1258, "step": 16375 }, { "epoch": 11.997069597069597, "grad_norm": 0.8125, "learning_rate": 7.855974150551597e-05, "loss": 0.1138, "step": 16376 }, { "epoch": 11.997802197802198, "grad_norm": 0.56640625, "learning_rate": 7.855691876152763e-05, "loss": 0.119, "step": 16377 }, { "epoch": 11.998534798534799, "grad_norm": 1.515625, "learning_rate": 7.855409588310938e-05, "loss": 0.1834, "step": 16378 }, { "epoch": 11.9992673992674, "grad_norm": 1.03125, "learning_rate": 7.855127287027476e-05, "loss": 0.1277, "step": 16379 }, { "epoch": 12.0, "grad_norm": 0.859375, "learning_rate": 7.854844972303726e-05, "loss": 0.1335, "step": 16380 }, { "epoch": 12.0007326007326, "grad_norm": 0.96484375, "learning_rate": 7.854562644141045e-05, "loss": 0.1286, "step": 16381 }, { "epoch": 12.001465201465201, "grad_norm": 1.359375, "learning_rate": 7.854280302540781e-05, "loss": 0.1377, "step": 16382 }, { "epoch": 12.002197802197802, "grad_norm": 0.98046875, "learning_rate": 7.853997947504293e-05, "loss": 0.1207, "step": 16383 }, { "epoch": 12.002930402930403, "grad_norm": 1.1875, "learning_rate": 7.853715579032928e-05, "loss": 0.1313, "step": 16384 }, { "epoch": 12.003663003663004, "grad_norm": 1.03125, "learning_rate": 7.853433197128041e-05, "loss": 0.1204, "step": 16385 }, { "epoch": 12.004395604395604, "grad_norm": 1.8046875, "learning_rate": 7.853150801790987e-05, "loss": 0.121, "step": 16386 }, { "epoch": 12.005128205128205, "grad_norm": 1.0703125, "learning_rate": 7.852868393023117e-05, "loss": 0.1429, "step": 16387 }, { "epoch": 12.005860805860806, "grad_norm": 1.171875, "learning_rate": 7.852585970825783e-05, "loss": 0.1229, "step": 16388 }, { "epoch": 12.006593406593407, "grad_norm": 1.0703125, "learning_rate": 7.852303535200341e-05, "loss": 0.1428, "step": 16389 }, { "epoch": 12.007326007326007, "grad_norm": 1.0546875, "learning_rate": 7.852021086148142e-05, "loss": 0.1045, "step": 16390 }, { "epoch": 12.008058608058608, "grad_norm": 1.3828125, "learning_rate": 7.851738623670541e-05, "loss": 0.1541, "step": 16391 }, { "epoch": 12.008791208791209, "grad_norm": 1.203125, "learning_rate": 7.85145614776889e-05, "loss": 0.1038, "step": 16392 }, { "epoch": 12.00952380952381, "grad_norm": 1.3984375, "learning_rate": 7.851173658444542e-05, "loss": 0.1488, "step": 16393 }, { "epoch": 12.01025641025641, "grad_norm": 1.5546875, "learning_rate": 7.850891155698854e-05, "loss": 0.1427, "step": 16394 }, { "epoch": 12.010989010989011, "grad_norm": 0.76953125, "learning_rate": 7.850608639533174e-05, "loss": 0.1432, "step": 16395 }, { "epoch": 12.011721611721612, "grad_norm": 0.92578125, "learning_rate": 7.85032610994886e-05, "loss": 0.0951, "step": 16396 }, { "epoch": 12.012454212454212, "grad_norm": 1.4609375, "learning_rate": 7.850043566947265e-05, "loss": 0.1448, "step": 16397 }, { "epoch": 12.013186813186813, "grad_norm": 2.03125, "learning_rate": 7.849761010529741e-05, "loss": 0.1581, "step": 16398 }, { "epoch": 12.013919413919414, "grad_norm": 1.0234375, "learning_rate": 7.849478440697642e-05, "loss": 0.1233, "step": 16399 }, { "epoch": 12.014652014652015, "grad_norm": 0.62890625, "learning_rate": 7.849195857452326e-05, "loss": 0.1239, "step": 16400 }, { "epoch": 12.015384615384615, "grad_norm": 0.78515625, "learning_rate": 7.848913260795141e-05, "loss": 0.1183, "step": 16401 }, { "epoch": 12.016117216117216, "grad_norm": 0.90625, "learning_rate": 7.848630650727446e-05, "loss": 0.1176, "step": 16402 }, { "epoch": 12.016849816849817, "grad_norm": 0.97265625, "learning_rate": 7.848348027250589e-05, "loss": 0.1446, "step": 16403 }, { "epoch": 12.017582417582418, "grad_norm": 0.98828125, "learning_rate": 7.848065390365932e-05, "loss": 0.1167, "step": 16404 }, { "epoch": 12.018315018315018, "grad_norm": 1.265625, "learning_rate": 7.847782740074822e-05, "loss": 0.1247, "step": 16405 }, { "epoch": 12.019047619047619, "grad_norm": 0.90625, "learning_rate": 7.847500076378618e-05, "loss": 0.1596, "step": 16406 }, { "epoch": 12.01978021978022, "grad_norm": 1.1640625, "learning_rate": 7.847217399278672e-05, "loss": 0.1577, "step": 16407 }, { "epoch": 12.02051282051282, "grad_norm": 0.734375, "learning_rate": 7.84693470877634e-05, "loss": 0.1164, "step": 16408 }, { "epoch": 12.021245421245421, "grad_norm": 0.8203125, "learning_rate": 7.846652004872976e-05, "loss": 0.1271, "step": 16409 }, { "epoch": 12.021978021978022, "grad_norm": 1.421875, "learning_rate": 7.846369287569934e-05, "loss": 0.1608, "step": 16410 }, { "epoch": 12.022710622710623, "grad_norm": 1.0546875, "learning_rate": 7.846086556868569e-05, "loss": 0.1236, "step": 16411 }, { "epoch": 12.023443223443223, "grad_norm": 0.66796875, "learning_rate": 7.845803812770234e-05, "loss": 0.1183, "step": 16412 }, { "epoch": 12.024175824175824, "grad_norm": 0.96484375, "learning_rate": 7.845521055276285e-05, "loss": 0.1293, "step": 16413 }, { "epoch": 12.024908424908425, "grad_norm": 1.390625, "learning_rate": 7.845238284388078e-05, "loss": 0.1333, "step": 16414 }, { "epoch": 12.025641025641026, "grad_norm": 1.421875, "learning_rate": 7.844955500106967e-05, "loss": 0.1045, "step": 16415 }, { "epoch": 12.026373626373626, "grad_norm": 0.953125, "learning_rate": 7.844672702434307e-05, "loss": 0.1326, "step": 16416 }, { "epoch": 12.027106227106227, "grad_norm": 0.97265625, "learning_rate": 7.844389891371451e-05, "loss": 0.1205, "step": 16417 }, { "epoch": 12.027838827838828, "grad_norm": 1.2578125, "learning_rate": 7.844107066919758e-05, "loss": 0.1178, "step": 16418 }, { "epoch": 12.028571428571428, "grad_norm": 0.984375, "learning_rate": 7.843824229080578e-05, "loss": 0.1229, "step": 16419 }, { "epoch": 12.02930402930403, "grad_norm": 0.8359375, "learning_rate": 7.843541377855271e-05, "loss": 0.1196, "step": 16420 }, { "epoch": 12.03003663003663, "grad_norm": 1.2890625, "learning_rate": 7.843258513245189e-05, "loss": 0.1283, "step": 16421 }, { "epoch": 12.03076923076923, "grad_norm": 0.734375, "learning_rate": 7.84297563525169e-05, "loss": 0.116, "step": 16422 }, { "epoch": 12.031501831501831, "grad_norm": 1.5234375, "learning_rate": 7.842692743876127e-05, "loss": 0.1308, "step": 16423 }, { "epoch": 12.032234432234432, "grad_norm": 0.60546875, "learning_rate": 7.842409839119856e-05, "loss": 0.1359, "step": 16424 }, { "epoch": 12.032967032967033, "grad_norm": 1.3671875, "learning_rate": 7.842126920984233e-05, "loss": 0.1443, "step": 16425 }, { "epoch": 12.033699633699634, "grad_norm": 0.953125, "learning_rate": 7.841843989470613e-05, "loss": 0.127, "step": 16426 }, { "epoch": 12.034432234432234, "grad_norm": 0.921875, "learning_rate": 7.841561044580353e-05, "loss": 0.1272, "step": 16427 }, { "epoch": 12.035164835164835, "grad_norm": 0.92578125, "learning_rate": 7.841278086314808e-05, "loss": 0.1633, "step": 16428 }, { "epoch": 12.035897435897436, "grad_norm": 1.0546875, "learning_rate": 7.84099511467533e-05, "loss": 0.1475, "step": 16429 }, { "epoch": 12.036630036630036, "grad_norm": 0.74609375, "learning_rate": 7.840712129663282e-05, "loss": 0.1499, "step": 16430 }, { "epoch": 12.037362637362637, "grad_norm": 0.74609375, "learning_rate": 7.840429131280015e-05, "loss": 0.1346, "step": 16431 }, { "epoch": 12.038095238095238, "grad_norm": 1.671875, "learning_rate": 7.840146119526883e-05, "loss": 0.1471, "step": 16432 }, { "epoch": 12.038827838827839, "grad_norm": 1.296875, "learning_rate": 7.839863094405249e-05, "loss": 0.1099, "step": 16433 }, { "epoch": 12.03956043956044, "grad_norm": 0.75390625, "learning_rate": 7.839580055916463e-05, "loss": 0.1106, "step": 16434 }, { "epoch": 12.04029304029304, "grad_norm": 1.109375, "learning_rate": 7.839297004061883e-05, "loss": 0.1298, "step": 16435 }, { "epoch": 12.04102564102564, "grad_norm": 0.8828125, "learning_rate": 7.839013938842867e-05, "loss": 0.1234, "step": 16436 }, { "epoch": 12.041758241758242, "grad_norm": 0.890625, "learning_rate": 7.83873086026077e-05, "loss": 0.1109, "step": 16437 }, { "epoch": 12.042490842490842, "grad_norm": 0.9453125, "learning_rate": 7.838447768316946e-05, "loss": 0.1518, "step": 16438 }, { "epoch": 12.043223443223443, "grad_norm": 1.0625, "learning_rate": 7.838164663012754e-05, "loss": 0.1399, "step": 16439 }, { "epoch": 12.043956043956044, "grad_norm": 1.203125, "learning_rate": 7.83788154434955e-05, "loss": 0.1444, "step": 16440 }, { "epoch": 12.044688644688645, "grad_norm": 0.8984375, "learning_rate": 7.83759841232869e-05, "loss": 0.1224, "step": 16441 }, { "epoch": 12.045421245421245, "grad_norm": 1.390625, "learning_rate": 7.83731526695153e-05, "loss": 0.1178, "step": 16442 }, { "epoch": 12.046153846153846, "grad_norm": 0.7578125, "learning_rate": 7.83703210821943e-05, "loss": 0.1302, "step": 16443 }, { "epoch": 12.046886446886447, "grad_norm": 1.484375, "learning_rate": 7.836748936133743e-05, "loss": 0.1306, "step": 16444 }, { "epoch": 12.047619047619047, "grad_norm": 1.3203125, "learning_rate": 7.836465750695826e-05, "loss": 0.1298, "step": 16445 }, { "epoch": 12.048351648351648, "grad_norm": 0.8125, "learning_rate": 7.836182551907038e-05, "loss": 0.1268, "step": 16446 }, { "epoch": 12.049084249084249, "grad_norm": 1.3359375, "learning_rate": 7.835899339768734e-05, "loss": 0.1516, "step": 16447 }, { "epoch": 12.04981684981685, "grad_norm": 0.90625, "learning_rate": 7.835616114282273e-05, "loss": 0.1063, "step": 16448 }, { "epoch": 12.05054945054945, "grad_norm": 1.046875, "learning_rate": 7.83533287544901e-05, "loss": 0.1435, "step": 16449 }, { "epoch": 12.051282051282051, "grad_norm": 1.109375, "learning_rate": 7.835049623270303e-05, "loss": 0.1417, "step": 16450 }, { "epoch": 12.052014652014652, "grad_norm": 1.078125, "learning_rate": 7.83476635774751e-05, "loss": 0.1474, "step": 16451 }, { "epoch": 12.052747252747253, "grad_norm": 1.125, "learning_rate": 7.834483078881984e-05, "loss": 0.138, "step": 16452 }, { "epoch": 12.053479853479853, "grad_norm": 0.859375, "learning_rate": 7.834199786675089e-05, "loss": 0.1253, "step": 16453 }, { "epoch": 12.054212454212454, "grad_norm": 0.828125, "learning_rate": 7.833916481128178e-05, "loss": 0.1371, "step": 16454 }, { "epoch": 12.054945054945055, "grad_norm": 0.98046875, "learning_rate": 7.833633162242607e-05, "loss": 0.1107, "step": 16455 }, { "epoch": 12.055677655677655, "grad_norm": 1.1484375, "learning_rate": 7.833349830019737e-05, "loss": 0.0973, "step": 16456 }, { "epoch": 12.056410256410256, "grad_norm": 0.828125, "learning_rate": 7.833066484460925e-05, "loss": 0.1184, "step": 16457 }, { "epoch": 12.057142857142857, "grad_norm": 1.1796875, "learning_rate": 7.832783125567527e-05, "loss": 0.1372, "step": 16458 }, { "epoch": 12.057875457875458, "grad_norm": 1.0859375, "learning_rate": 7.832499753340903e-05, "loss": 0.1303, "step": 16459 }, { "epoch": 12.058608058608058, "grad_norm": 0.94140625, "learning_rate": 7.832216367782406e-05, "loss": 0.1001, "step": 16460 }, { "epoch": 12.05934065934066, "grad_norm": 0.734375, "learning_rate": 7.8319329688934e-05, "loss": 0.119, "step": 16461 }, { "epoch": 12.06007326007326, "grad_norm": 1.453125, "learning_rate": 7.83164955667524e-05, "loss": 0.1473, "step": 16462 }, { "epoch": 12.06080586080586, "grad_norm": 0.71484375, "learning_rate": 7.831366131129283e-05, "loss": 0.1227, "step": 16463 }, { "epoch": 12.061538461538461, "grad_norm": 0.8046875, "learning_rate": 7.831082692256889e-05, "loss": 0.1251, "step": 16464 }, { "epoch": 12.062271062271062, "grad_norm": 0.68359375, "learning_rate": 7.830799240059412e-05, "loss": 0.1251, "step": 16465 }, { "epoch": 12.063003663003663, "grad_norm": 1.109375, "learning_rate": 7.830515774538214e-05, "loss": 0.1639, "step": 16466 }, { "epoch": 12.063736263736264, "grad_norm": 1.1953125, "learning_rate": 7.830232295694653e-05, "loss": 0.1401, "step": 16467 }, { "epoch": 12.064468864468864, "grad_norm": 1.3046875, "learning_rate": 7.829948803530086e-05, "loss": 0.12, "step": 16468 }, { "epoch": 12.065201465201465, "grad_norm": 1.2578125, "learning_rate": 7.829665298045872e-05, "loss": 0.1407, "step": 16469 }, { "epoch": 12.065934065934066, "grad_norm": 1.5703125, "learning_rate": 7.829381779243368e-05, "loss": 0.1444, "step": 16470 }, { "epoch": 12.066666666666666, "grad_norm": 1.0, "learning_rate": 7.829098247123934e-05, "loss": 0.146, "step": 16471 }, { "epoch": 12.067399267399267, "grad_norm": 1.2109375, "learning_rate": 7.82881470168893e-05, "loss": 0.1268, "step": 16472 }, { "epoch": 12.068131868131868, "grad_norm": 0.93359375, "learning_rate": 7.82853114293971e-05, "loss": 0.1077, "step": 16473 }, { "epoch": 12.068864468864469, "grad_norm": 1.3203125, "learning_rate": 7.828247570877636e-05, "loss": 0.1482, "step": 16474 }, { "epoch": 12.06959706959707, "grad_norm": 0.796875, "learning_rate": 7.827963985504066e-05, "loss": 0.1154, "step": 16475 }, { "epoch": 12.07032967032967, "grad_norm": 1.1015625, "learning_rate": 7.827680386820358e-05, "loss": 0.1647, "step": 16476 }, { "epoch": 12.07106227106227, "grad_norm": 1.0859375, "learning_rate": 7.827396774827872e-05, "loss": 0.1704, "step": 16477 }, { "epoch": 12.071794871794872, "grad_norm": 0.82421875, "learning_rate": 7.827113149527965e-05, "loss": 0.1281, "step": 16478 }, { "epoch": 12.072527472527472, "grad_norm": 1.2421875, "learning_rate": 7.826829510922e-05, "loss": 0.1178, "step": 16479 }, { "epoch": 12.073260073260073, "grad_norm": 1.0390625, "learning_rate": 7.826545859011331e-05, "loss": 0.1404, "step": 16480 }, { "epoch": 12.073992673992674, "grad_norm": 1.3984375, "learning_rate": 7.826262193797319e-05, "loss": 0.1494, "step": 16481 }, { "epoch": 12.074725274725274, "grad_norm": 1.0859375, "learning_rate": 7.825978515281323e-05, "loss": 0.1301, "step": 16482 }, { "epoch": 12.075457875457875, "grad_norm": 1.546875, "learning_rate": 7.825694823464707e-05, "loss": 0.1295, "step": 16483 }, { "epoch": 12.076190476190476, "grad_norm": 0.92578125, "learning_rate": 7.825411118348821e-05, "loss": 0.1345, "step": 16484 }, { "epoch": 12.076923076923077, "grad_norm": 1.1796875, "learning_rate": 7.825127399935032e-05, "loss": 0.1666, "step": 16485 }, { "epoch": 12.077655677655677, "grad_norm": 1.2109375, "learning_rate": 7.824843668224695e-05, "loss": 0.1502, "step": 16486 }, { "epoch": 12.078388278388278, "grad_norm": 1.375, "learning_rate": 7.824559923219172e-05, "loss": 0.1552, "step": 16487 }, { "epoch": 12.079120879120879, "grad_norm": 0.80859375, "learning_rate": 7.824276164919822e-05, "loss": 0.1139, "step": 16488 }, { "epoch": 12.07985347985348, "grad_norm": 0.83984375, "learning_rate": 7.823992393328003e-05, "loss": 0.1057, "step": 16489 }, { "epoch": 12.08058608058608, "grad_norm": 0.99609375, "learning_rate": 7.823708608445076e-05, "loss": 0.1115, "step": 16490 }, { "epoch": 12.081318681318681, "grad_norm": 1.1875, "learning_rate": 7.823424810272402e-05, "loss": 0.1411, "step": 16491 }, { "epoch": 12.082051282051282, "grad_norm": 0.6796875, "learning_rate": 7.823140998811337e-05, "loss": 0.1267, "step": 16492 }, { "epoch": 12.082783882783882, "grad_norm": 0.8671875, "learning_rate": 7.822857174063244e-05, "loss": 0.1463, "step": 16493 }, { "epoch": 12.083516483516483, "grad_norm": 1.203125, "learning_rate": 7.822573336029483e-05, "loss": 0.1374, "step": 16494 }, { "epoch": 12.084249084249084, "grad_norm": 1.03125, "learning_rate": 7.822289484711412e-05, "loss": 0.124, "step": 16495 }, { "epoch": 12.084981684981685, "grad_norm": 1.3046875, "learning_rate": 7.82200562011039e-05, "loss": 0.1904, "step": 16496 }, { "epoch": 12.085714285714285, "grad_norm": 1.5, "learning_rate": 7.821721742227783e-05, "loss": 0.141, "step": 16497 }, { "epoch": 12.086446886446886, "grad_norm": 0.87109375, "learning_rate": 7.821437851064945e-05, "loss": 0.1357, "step": 16498 }, { "epoch": 12.087179487179487, "grad_norm": 0.8125, "learning_rate": 7.82115394662324e-05, "loss": 0.1386, "step": 16499 }, { "epoch": 12.087912087912088, "grad_norm": 1.3203125, "learning_rate": 7.820870028904025e-05, "loss": 0.1324, "step": 16500 }, { "epoch": 12.088644688644688, "grad_norm": 1.5390625, "learning_rate": 7.820586097908662e-05, "loss": 0.1355, "step": 16501 }, { "epoch": 12.089377289377289, "grad_norm": 1.171875, "learning_rate": 7.820302153638513e-05, "loss": 0.1373, "step": 16502 }, { "epoch": 12.09010989010989, "grad_norm": 1.0859375, "learning_rate": 7.820018196094938e-05, "loss": 0.1555, "step": 16503 }, { "epoch": 12.09084249084249, "grad_norm": 1.0390625, "learning_rate": 7.819734225279294e-05, "loss": 0.1141, "step": 16504 }, { "epoch": 12.091575091575091, "grad_norm": 0.8125, "learning_rate": 7.819450241192943e-05, "loss": 0.129, "step": 16505 }, { "epoch": 12.092307692307692, "grad_norm": 1.1875, "learning_rate": 7.819166243837249e-05, "loss": 0.1051, "step": 16506 }, { "epoch": 12.093040293040293, "grad_norm": 1.28125, "learning_rate": 7.81888223321357e-05, "loss": 0.115, "step": 16507 }, { "epoch": 12.093772893772893, "grad_norm": 1.8984375, "learning_rate": 7.818598209323266e-05, "loss": 0.1876, "step": 16508 }, { "epoch": 12.094505494505494, "grad_norm": 1.078125, "learning_rate": 7.818314172167702e-05, "loss": 0.1153, "step": 16509 }, { "epoch": 12.095238095238095, "grad_norm": 1.40625, "learning_rate": 7.818030121748233e-05, "loss": 0.1411, "step": 16510 }, { "epoch": 12.095970695970696, "grad_norm": 0.76171875, "learning_rate": 7.817746058066225e-05, "loss": 0.1392, "step": 16511 }, { "epoch": 12.096703296703296, "grad_norm": 1.5703125, "learning_rate": 7.817461981123036e-05, "loss": 0.1103, "step": 16512 }, { "epoch": 12.097435897435897, "grad_norm": 1.0234375, "learning_rate": 7.81717789092003e-05, "loss": 0.1085, "step": 16513 }, { "epoch": 12.098168498168498, "grad_norm": 1.59375, "learning_rate": 7.816893787458565e-05, "loss": 0.1246, "step": 16514 }, { "epoch": 12.098901098901099, "grad_norm": 1.0859375, "learning_rate": 7.816609670740003e-05, "loss": 0.1338, "step": 16515 }, { "epoch": 12.0996336996337, "grad_norm": 1.640625, "learning_rate": 7.816325540765705e-05, "loss": 0.1332, "step": 16516 }, { "epoch": 12.1003663003663, "grad_norm": 1.703125, "learning_rate": 7.816041397537037e-05, "loss": 0.1442, "step": 16517 }, { "epoch": 12.1010989010989, "grad_norm": 1.453125, "learning_rate": 7.815757241055353e-05, "loss": 0.1622, "step": 16518 }, { "epoch": 12.101831501831501, "grad_norm": 1.1875, "learning_rate": 7.815473071322018e-05, "loss": 0.1199, "step": 16519 }, { "epoch": 12.102564102564102, "grad_norm": 1.4765625, "learning_rate": 7.815188888338395e-05, "loss": 0.113, "step": 16520 }, { "epoch": 12.103296703296703, "grad_norm": 0.78515625, "learning_rate": 7.814904692105844e-05, "loss": 0.1112, "step": 16521 }, { "epoch": 12.104029304029304, "grad_norm": 1.0625, "learning_rate": 7.814620482625728e-05, "loss": 0.1058, "step": 16522 }, { "epoch": 12.104761904761904, "grad_norm": 0.890625, "learning_rate": 7.814336259899406e-05, "loss": 0.1237, "step": 16523 }, { "epoch": 12.105494505494505, "grad_norm": 1.3515625, "learning_rate": 7.814052023928242e-05, "loss": 0.1421, "step": 16524 }, { "epoch": 12.106227106227106, "grad_norm": 1.1796875, "learning_rate": 7.813767774713597e-05, "loss": 0.1165, "step": 16525 }, { "epoch": 12.106959706959707, "grad_norm": 1.8125, "learning_rate": 7.813483512256836e-05, "loss": 0.1193, "step": 16526 }, { "epoch": 12.107692307692307, "grad_norm": 1.3828125, "learning_rate": 7.813199236559316e-05, "loss": 0.1272, "step": 16527 }, { "epoch": 12.108424908424908, "grad_norm": 0.80859375, "learning_rate": 7.812914947622401e-05, "loss": 0.1489, "step": 16528 }, { "epoch": 12.109157509157509, "grad_norm": 0.7578125, "learning_rate": 7.812630645447455e-05, "loss": 0.0989, "step": 16529 }, { "epoch": 12.10989010989011, "grad_norm": 1.15625, "learning_rate": 7.812346330035837e-05, "loss": 0.1172, "step": 16530 }, { "epoch": 12.11062271062271, "grad_norm": 1.0078125, "learning_rate": 7.812062001388914e-05, "loss": 0.124, "step": 16531 }, { "epoch": 12.111355311355311, "grad_norm": 0.65625, "learning_rate": 7.811777659508042e-05, "loss": 0.1294, "step": 16532 }, { "epoch": 12.112087912087912, "grad_norm": 1.4296875, "learning_rate": 7.811493304394588e-05, "loss": 0.1451, "step": 16533 }, { "epoch": 12.112820512820512, "grad_norm": 1.2109375, "learning_rate": 7.81120893604991e-05, "loss": 0.1376, "step": 16534 }, { "epoch": 12.113553113553113, "grad_norm": 1.328125, "learning_rate": 7.810924554475377e-05, "loss": 0.1585, "step": 16535 }, { "epoch": 12.114285714285714, "grad_norm": 0.8046875, "learning_rate": 7.810640159672347e-05, "loss": 0.1173, "step": 16536 }, { "epoch": 12.115018315018315, "grad_norm": 1.171875, "learning_rate": 7.810355751642184e-05, "loss": 0.1053, "step": 16537 }, { "epoch": 12.115750915750915, "grad_norm": 1.0546875, "learning_rate": 7.810071330386251e-05, "loss": 0.1017, "step": 16538 }, { "epoch": 12.116483516483516, "grad_norm": 1.546875, "learning_rate": 7.80978689590591e-05, "loss": 0.131, "step": 16539 }, { "epoch": 12.117216117216117, "grad_norm": 1.0703125, "learning_rate": 7.809502448202523e-05, "loss": 0.1418, "step": 16540 }, { "epoch": 12.117948717948718, "grad_norm": 0.5703125, "learning_rate": 7.809217987277454e-05, "loss": 0.1205, "step": 16541 }, { "epoch": 12.118681318681318, "grad_norm": 0.765625, "learning_rate": 7.808933513132067e-05, "loss": 0.1512, "step": 16542 }, { "epoch": 12.119413919413919, "grad_norm": 1.171875, "learning_rate": 7.808649025767724e-05, "loss": 0.1489, "step": 16543 }, { "epoch": 12.12014652014652, "grad_norm": 0.87890625, "learning_rate": 7.808364525185786e-05, "loss": 0.1304, "step": 16544 }, { "epoch": 12.12087912087912, "grad_norm": 1.015625, "learning_rate": 7.808080011387619e-05, "loss": 0.1113, "step": 16545 }, { "epoch": 12.121611721611721, "grad_norm": 1.0234375, "learning_rate": 7.807795484374585e-05, "loss": 0.1484, "step": 16546 }, { "epoch": 12.122344322344322, "grad_norm": 0.5703125, "learning_rate": 7.807510944148048e-05, "loss": 0.1106, "step": 16547 }, { "epoch": 12.123076923076923, "grad_norm": 0.84375, "learning_rate": 7.807226390709372e-05, "loss": 0.1342, "step": 16548 }, { "epoch": 12.123809523809523, "grad_norm": 0.796875, "learning_rate": 7.806941824059916e-05, "loss": 0.1389, "step": 16549 }, { "epoch": 12.124542124542124, "grad_norm": 1.109375, "learning_rate": 7.80665724420105e-05, "loss": 0.125, "step": 16550 }, { "epoch": 12.125274725274725, "grad_norm": 0.91796875, "learning_rate": 7.806372651134132e-05, "loss": 0.1059, "step": 16551 }, { "epoch": 12.126007326007326, "grad_norm": 1.171875, "learning_rate": 7.806088044860531e-05, "loss": 0.1571, "step": 16552 }, { "epoch": 12.126739926739926, "grad_norm": 1.0, "learning_rate": 7.805803425381604e-05, "loss": 0.1153, "step": 16553 }, { "epoch": 12.127472527472527, "grad_norm": 0.89453125, "learning_rate": 7.805518792698719e-05, "loss": 0.1413, "step": 16554 }, { "epoch": 12.128205128205128, "grad_norm": 0.9375, "learning_rate": 7.805234146813238e-05, "loss": 0.1047, "step": 16555 }, { "epoch": 12.128937728937728, "grad_norm": 1.2421875, "learning_rate": 7.804949487726528e-05, "loss": 0.1519, "step": 16556 }, { "epoch": 12.12967032967033, "grad_norm": 0.96484375, "learning_rate": 7.804664815439949e-05, "loss": 0.1002, "step": 16557 }, { "epoch": 12.13040293040293, "grad_norm": 0.84765625, "learning_rate": 7.804380129954867e-05, "loss": 0.1126, "step": 16558 }, { "epoch": 12.13113553113553, "grad_norm": 0.88671875, "learning_rate": 7.804095431272645e-05, "loss": 0.1112, "step": 16559 }, { "epoch": 12.131868131868131, "grad_norm": 0.91796875, "learning_rate": 7.803810719394647e-05, "loss": 0.1211, "step": 16560 }, { "epoch": 12.132600732600732, "grad_norm": 1.046875, "learning_rate": 7.80352599432224e-05, "loss": 0.1305, "step": 16561 }, { "epoch": 12.133333333333333, "grad_norm": 1.328125, "learning_rate": 7.803241256056785e-05, "loss": 0.136, "step": 16562 }, { "epoch": 12.134065934065934, "grad_norm": 0.99609375, "learning_rate": 7.802956504599649e-05, "loss": 0.121, "step": 16563 }, { "epoch": 12.134798534798534, "grad_norm": 1.0625, "learning_rate": 7.802671739952192e-05, "loss": 0.1224, "step": 16564 }, { "epoch": 12.135531135531135, "grad_norm": 1.1640625, "learning_rate": 7.802386962115781e-05, "loss": 0.1549, "step": 16565 }, { "epoch": 12.136263736263736, "grad_norm": 0.99609375, "learning_rate": 7.802102171091782e-05, "loss": 0.1346, "step": 16566 }, { "epoch": 12.136996336996337, "grad_norm": 0.984375, "learning_rate": 7.801817366881556e-05, "loss": 0.1346, "step": 16567 }, { "epoch": 12.137728937728937, "grad_norm": 1.5, "learning_rate": 7.801532549486474e-05, "loss": 0.1213, "step": 16568 }, { "epoch": 12.138461538461538, "grad_norm": 1.375, "learning_rate": 7.801247718907893e-05, "loss": 0.131, "step": 16569 }, { "epoch": 12.139194139194139, "grad_norm": 1.3828125, "learning_rate": 7.800962875147179e-05, "loss": 0.1324, "step": 16570 }, { "epoch": 12.13992673992674, "grad_norm": 0.8359375, "learning_rate": 7.800678018205703e-05, "loss": 0.1344, "step": 16571 }, { "epoch": 12.14065934065934, "grad_norm": 0.77734375, "learning_rate": 7.800393148084824e-05, "loss": 0.1287, "step": 16572 }, { "epoch": 12.141391941391941, "grad_norm": 1.671875, "learning_rate": 7.800108264785908e-05, "loss": 0.1499, "step": 16573 }, { "epoch": 12.142124542124542, "grad_norm": 1.3359375, "learning_rate": 7.799823368310322e-05, "loss": 0.1441, "step": 16574 }, { "epoch": 12.142857142857142, "grad_norm": 1.1640625, "learning_rate": 7.799538458659429e-05, "loss": 0.104, "step": 16575 }, { "epoch": 12.143589743589743, "grad_norm": 1.640625, "learning_rate": 7.799253535834595e-05, "loss": 0.1265, "step": 16576 }, { "epoch": 12.144322344322344, "grad_norm": 1.3984375, "learning_rate": 7.798968599837182e-05, "loss": 0.1327, "step": 16577 }, { "epoch": 12.145054945054945, "grad_norm": 1.296875, "learning_rate": 7.798683650668561e-05, "loss": 0.1058, "step": 16578 }, { "epoch": 12.145787545787545, "grad_norm": 1.2578125, "learning_rate": 7.798398688330093e-05, "loss": 0.1187, "step": 16579 }, { "epoch": 12.146520146520146, "grad_norm": 1.28125, "learning_rate": 7.798113712823144e-05, "loss": 0.1423, "step": 16580 }, { "epoch": 12.147252747252747, "grad_norm": 0.9921875, "learning_rate": 7.79782872414908e-05, "loss": 0.1271, "step": 16581 }, { "epoch": 12.147985347985347, "grad_norm": 1.6875, "learning_rate": 7.797543722309269e-05, "loss": 0.1513, "step": 16582 }, { "epoch": 12.148717948717948, "grad_norm": 0.85546875, "learning_rate": 7.797258707305071e-05, "loss": 0.1017, "step": 16583 }, { "epoch": 12.149450549450549, "grad_norm": 0.5625, "learning_rate": 7.796973679137858e-05, "loss": 0.1214, "step": 16584 }, { "epoch": 12.15018315018315, "grad_norm": 0.7578125, "learning_rate": 7.79668863780899e-05, "loss": 0.1429, "step": 16585 }, { "epoch": 12.15091575091575, "grad_norm": 0.59765625, "learning_rate": 7.796403583319836e-05, "loss": 0.1042, "step": 16586 }, { "epoch": 12.151648351648351, "grad_norm": 0.81640625, "learning_rate": 7.79611851567176e-05, "loss": 0.125, "step": 16587 }, { "epoch": 12.152380952380952, "grad_norm": 1.0546875, "learning_rate": 7.79583343486613e-05, "loss": 0.0998, "step": 16588 }, { "epoch": 12.153113553113553, "grad_norm": 1.0234375, "learning_rate": 7.79554834090431e-05, "loss": 0.1168, "step": 16589 }, { "epoch": 12.153846153846153, "grad_norm": 1.1171875, "learning_rate": 7.795263233787666e-05, "loss": 0.108, "step": 16590 }, { "epoch": 12.154578754578754, "grad_norm": 1.0, "learning_rate": 7.794978113517564e-05, "loss": 0.113, "step": 16591 }, { "epoch": 12.155311355311355, "grad_norm": 1.21875, "learning_rate": 7.794692980095372e-05, "loss": 0.1188, "step": 16592 }, { "epoch": 12.156043956043955, "grad_norm": 0.87890625, "learning_rate": 7.794407833522454e-05, "loss": 0.131, "step": 16593 }, { "epoch": 12.156776556776556, "grad_norm": 0.87890625, "learning_rate": 7.794122673800178e-05, "loss": 0.1346, "step": 16594 }, { "epoch": 12.157509157509157, "grad_norm": 0.91015625, "learning_rate": 7.79383750092991e-05, "loss": 0.1309, "step": 16595 }, { "epoch": 12.158241758241758, "grad_norm": 0.7265625, "learning_rate": 7.793552314913014e-05, "loss": 0.1034, "step": 16596 }, { "epoch": 12.158974358974358, "grad_norm": 1.609375, "learning_rate": 7.79326711575086e-05, "loss": 0.1331, "step": 16597 }, { "epoch": 12.15970695970696, "grad_norm": 1.4921875, "learning_rate": 7.792981903444811e-05, "loss": 0.1468, "step": 16598 }, { "epoch": 12.16043956043956, "grad_norm": 1.0390625, "learning_rate": 7.792696677996236e-05, "loss": 0.1484, "step": 16599 }, { "epoch": 12.16117216117216, "grad_norm": 0.90234375, "learning_rate": 7.792411439406502e-05, "loss": 0.1138, "step": 16600 }, { "epoch": 12.161904761904761, "grad_norm": 1.1875, "learning_rate": 7.792126187676973e-05, "loss": 0.1334, "step": 16601 }, { "epoch": 12.162637362637362, "grad_norm": 1.15625, "learning_rate": 7.791840922809018e-05, "loss": 0.1242, "step": 16602 }, { "epoch": 12.163369963369963, "grad_norm": 0.84375, "learning_rate": 7.791555644804005e-05, "loss": 0.1338, "step": 16603 }, { "epoch": 12.164102564102564, "grad_norm": 1.2109375, "learning_rate": 7.791270353663294e-05, "loss": 0.1353, "step": 16604 }, { "epoch": 12.164835164835164, "grad_norm": 0.83203125, "learning_rate": 7.79098504938826e-05, "loss": 0.0994, "step": 16605 }, { "epoch": 12.165567765567765, "grad_norm": 0.95703125, "learning_rate": 7.790699731980265e-05, "loss": 0.1392, "step": 16606 }, { "epoch": 12.166300366300366, "grad_norm": 1.9140625, "learning_rate": 7.790414401440681e-05, "loss": 0.1475, "step": 16607 }, { "epoch": 12.167032967032966, "grad_norm": 0.7265625, "learning_rate": 7.79012905777087e-05, "loss": 0.1084, "step": 16608 }, { "epoch": 12.167765567765567, "grad_norm": 0.59765625, "learning_rate": 7.789843700972201e-05, "loss": 0.0991, "step": 16609 }, { "epoch": 12.168498168498168, "grad_norm": 1.0625, "learning_rate": 7.789558331046042e-05, "loss": 0.1805, "step": 16610 }, { "epoch": 12.169230769230769, "grad_norm": 1.4921875, "learning_rate": 7.78927294799376e-05, "loss": 0.1189, "step": 16611 }, { "epoch": 12.16996336996337, "grad_norm": 1.390625, "learning_rate": 7.788987551816721e-05, "loss": 0.1168, "step": 16612 }, { "epoch": 12.17069597069597, "grad_norm": 0.8828125, "learning_rate": 7.788702142516295e-05, "loss": 0.1199, "step": 16613 }, { "epoch": 12.17142857142857, "grad_norm": 1.1953125, "learning_rate": 7.788416720093845e-05, "loss": 0.1414, "step": 16614 }, { "epoch": 12.172161172161172, "grad_norm": 1.15625, "learning_rate": 7.788131284550744e-05, "loss": 0.1618, "step": 16615 }, { "epoch": 12.172893772893772, "grad_norm": 1.5625, "learning_rate": 7.787845835888358e-05, "loss": 0.1245, "step": 16616 }, { "epoch": 12.173626373626373, "grad_norm": 1.390625, "learning_rate": 7.787560374108052e-05, "loss": 0.1149, "step": 16617 }, { "epoch": 12.174358974358974, "grad_norm": 1.875, "learning_rate": 7.787274899211196e-05, "loss": 0.1554, "step": 16618 }, { "epoch": 12.175091575091574, "grad_norm": 1.4765625, "learning_rate": 7.786989411199157e-05, "loss": 0.159, "step": 16619 }, { "epoch": 12.175824175824175, "grad_norm": 0.76171875, "learning_rate": 7.786703910073305e-05, "loss": 0.1283, "step": 16620 }, { "epoch": 12.176556776556776, "grad_norm": 1.1953125, "learning_rate": 7.786418395835004e-05, "loss": 0.1245, "step": 16621 }, { "epoch": 12.177289377289377, "grad_norm": 1.34375, "learning_rate": 7.786132868485625e-05, "loss": 0.1381, "step": 16622 }, { "epoch": 12.178021978021977, "grad_norm": 0.8046875, "learning_rate": 7.785847328026536e-05, "loss": 0.1095, "step": 16623 }, { "epoch": 12.178754578754578, "grad_norm": 1.3984375, "learning_rate": 7.785561774459103e-05, "loss": 0.1172, "step": 16624 }, { "epoch": 12.179487179487179, "grad_norm": 0.73046875, "learning_rate": 7.785276207784695e-05, "loss": 0.1088, "step": 16625 }, { "epoch": 12.18021978021978, "grad_norm": 1.1015625, "learning_rate": 7.784990628004682e-05, "loss": 0.1136, "step": 16626 }, { "epoch": 12.18095238095238, "grad_norm": 1.4375, "learning_rate": 7.784705035120431e-05, "loss": 0.1177, "step": 16627 }, { "epoch": 12.181684981684981, "grad_norm": 2.15625, "learning_rate": 7.784419429133311e-05, "loss": 0.1062, "step": 16628 }, { "epoch": 12.182417582417582, "grad_norm": 1.03125, "learning_rate": 7.784133810044689e-05, "loss": 0.1335, "step": 16629 }, { "epoch": 12.183150183150182, "grad_norm": 0.70703125, "learning_rate": 7.783848177855934e-05, "loss": 0.1128, "step": 16630 }, { "epoch": 12.183882783882783, "grad_norm": 1.15625, "learning_rate": 7.783562532568415e-05, "loss": 0.1728, "step": 16631 }, { "epoch": 12.184615384615384, "grad_norm": 0.81640625, "learning_rate": 7.783276874183501e-05, "loss": 0.106, "step": 16632 }, { "epoch": 12.185347985347985, "grad_norm": 0.91796875, "learning_rate": 7.782991202702561e-05, "loss": 0.1404, "step": 16633 }, { "epoch": 12.186080586080585, "grad_norm": 1.140625, "learning_rate": 7.782705518126963e-05, "loss": 0.1352, "step": 16634 }, { "epoch": 12.186813186813186, "grad_norm": 1.03125, "learning_rate": 7.782419820458074e-05, "loss": 0.1263, "step": 16635 }, { "epoch": 12.187545787545787, "grad_norm": 1.2109375, "learning_rate": 7.782134109697266e-05, "loss": 0.122, "step": 16636 }, { "epoch": 12.188278388278388, "grad_norm": 1.171875, "learning_rate": 7.781848385845904e-05, "loss": 0.1304, "step": 16637 }, { "epoch": 12.189010989010988, "grad_norm": 1.21875, "learning_rate": 7.781562648905363e-05, "loss": 0.1401, "step": 16638 }, { "epoch": 12.189743589743589, "grad_norm": 0.59765625, "learning_rate": 7.781276898877007e-05, "loss": 0.1106, "step": 16639 }, { "epoch": 12.19047619047619, "grad_norm": 1.328125, "learning_rate": 7.780991135762207e-05, "loss": 0.1352, "step": 16640 }, { "epoch": 12.19120879120879, "grad_norm": 1.4140625, "learning_rate": 7.780705359562332e-05, "loss": 0.1094, "step": 16641 }, { "epoch": 12.191941391941391, "grad_norm": 1.1875, "learning_rate": 7.780419570278753e-05, "loss": 0.1644, "step": 16642 }, { "epoch": 12.192673992673992, "grad_norm": 1.0625, "learning_rate": 7.780133767912835e-05, "loss": 0.1136, "step": 16643 }, { "epoch": 12.193406593406593, "grad_norm": 1.078125, "learning_rate": 7.779847952465951e-05, "loss": 0.1452, "step": 16644 }, { "epoch": 12.194139194139193, "grad_norm": 1.3359375, "learning_rate": 7.77956212393947e-05, "loss": 0.1234, "step": 16645 }, { "epoch": 12.194871794871794, "grad_norm": 1.078125, "learning_rate": 7.779276282334758e-05, "loss": 0.117, "step": 16646 }, { "epoch": 12.195604395604395, "grad_norm": 1.3828125, "learning_rate": 7.77899042765319e-05, "loss": 0.1262, "step": 16647 }, { "epoch": 12.196336996336996, "grad_norm": 1.2734375, "learning_rate": 7.778704559896131e-05, "loss": 0.1586, "step": 16648 }, { "epoch": 12.197069597069596, "grad_norm": 0.7734375, "learning_rate": 7.778418679064954e-05, "loss": 0.1111, "step": 16649 }, { "epoch": 12.197802197802197, "grad_norm": 1.078125, "learning_rate": 7.778132785161027e-05, "loss": 0.1121, "step": 16650 }, { "epoch": 12.198534798534798, "grad_norm": 0.98828125, "learning_rate": 7.77784687818572e-05, "loss": 0.1337, "step": 16651 }, { "epoch": 12.199267399267399, "grad_norm": 1.609375, "learning_rate": 7.777560958140405e-05, "loss": 0.1227, "step": 16652 }, { "epoch": 12.2, "grad_norm": 0.82421875, "learning_rate": 7.777275025026447e-05, "loss": 0.1315, "step": 16653 }, { "epoch": 12.2007326007326, "grad_norm": 1.25, "learning_rate": 7.776989078845223e-05, "loss": 0.1101, "step": 16654 }, { "epoch": 12.2014652014652, "grad_norm": 1.140625, "learning_rate": 7.776703119598097e-05, "loss": 0.146, "step": 16655 }, { "epoch": 12.202197802197801, "grad_norm": 1.2890625, "learning_rate": 7.77641714728644e-05, "loss": 0.1505, "step": 16656 }, { "epoch": 12.202930402930402, "grad_norm": 0.8046875, "learning_rate": 7.776131161911626e-05, "loss": 0.1062, "step": 16657 }, { "epoch": 12.203663003663003, "grad_norm": 1.5859375, "learning_rate": 7.77584516347502e-05, "loss": 0.1215, "step": 16658 }, { "epoch": 12.204395604395604, "grad_norm": 0.9453125, "learning_rate": 7.775559151977995e-05, "loss": 0.0925, "step": 16659 }, { "epoch": 12.205128205128204, "grad_norm": 1.578125, "learning_rate": 7.775273127421924e-05, "loss": 0.1149, "step": 16660 }, { "epoch": 12.205860805860805, "grad_norm": 1.9140625, "learning_rate": 7.774987089808174e-05, "loss": 0.1388, "step": 16661 }, { "epoch": 12.206593406593406, "grad_norm": 1.8671875, "learning_rate": 7.774701039138114e-05, "loss": 0.141, "step": 16662 }, { "epoch": 12.207326007326007, "grad_norm": 1.296875, "learning_rate": 7.774414975413118e-05, "loss": 0.1431, "step": 16663 }, { "epoch": 12.208058608058607, "grad_norm": 1.296875, "learning_rate": 7.774128898634557e-05, "loss": 0.1661, "step": 16664 }, { "epoch": 12.208791208791208, "grad_norm": 2.015625, "learning_rate": 7.773842808803798e-05, "loss": 0.137, "step": 16665 }, { "epoch": 12.209523809523809, "grad_norm": 1.1953125, "learning_rate": 7.773556705922216e-05, "loss": 0.1236, "step": 16666 }, { "epoch": 12.21025641025641, "grad_norm": 0.8046875, "learning_rate": 7.773270589991177e-05, "loss": 0.1183, "step": 16667 }, { "epoch": 12.21098901098901, "grad_norm": 1.1640625, "learning_rate": 7.772984461012057e-05, "loss": 0.1187, "step": 16668 }, { "epoch": 12.211721611721611, "grad_norm": 0.81640625, "learning_rate": 7.772698318986224e-05, "loss": 0.1121, "step": 16669 }, { "epoch": 12.212454212454212, "grad_norm": 1.4921875, "learning_rate": 7.772412163915051e-05, "loss": 0.1305, "step": 16670 }, { "epoch": 12.213186813186812, "grad_norm": 1.6796875, "learning_rate": 7.772125995799905e-05, "loss": 0.1371, "step": 16671 }, { "epoch": 12.213919413919413, "grad_norm": 1.078125, "learning_rate": 7.771839814642161e-05, "loss": 0.1415, "step": 16672 }, { "epoch": 12.214652014652014, "grad_norm": 0.88671875, "learning_rate": 7.77155362044319e-05, "loss": 0.1294, "step": 16673 }, { "epoch": 12.215384615384615, "grad_norm": 1.359375, "learning_rate": 7.77126741320436e-05, "loss": 0.1167, "step": 16674 }, { "epoch": 12.216117216117215, "grad_norm": 0.96484375, "learning_rate": 7.770981192927045e-05, "loss": 0.1014, "step": 16675 }, { "epoch": 12.216849816849816, "grad_norm": 0.51953125, "learning_rate": 7.770694959612616e-05, "loss": 0.1289, "step": 16676 }, { "epoch": 12.217582417582417, "grad_norm": 1.1796875, "learning_rate": 7.770408713262444e-05, "loss": 0.1333, "step": 16677 }, { "epoch": 12.218315018315018, "grad_norm": 1.0078125, "learning_rate": 7.770122453877902e-05, "loss": 0.1203, "step": 16678 }, { "epoch": 12.219047619047618, "grad_norm": 1.375, "learning_rate": 7.769836181460359e-05, "loss": 0.1318, "step": 16679 }, { "epoch": 12.219780219780219, "grad_norm": 1.3984375, "learning_rate": 7.76954989601119e-05, "loss": 0.1616, "step": 16680 }, { "epoch": 12.22051282051282, "grad_norm": 1.0703125, "learning_rate": 7.769263597531764e-05, "loss": 0.1206, "step": 16681 }, { "epoch": 12.22124542124542, "grad_norm": 1.09375, "learning_rate": 7.768977286023452e-05, "loss": 0.1101, "step": 16682 }, { "epoch": 12.221978021978021, "grad_norm": 1.0859375, "learning_rate": 7.768690961487627e-05, "loss": 0.1299, "step": 16683 }, { "epoch": 12.222710622710622, "grad_norm": 1.359375, "learning_rate": 7.768404623925663e-05, "loss": 0.1709, "step": 16684 }, { "epoch": 12.223443223443223, "grad_norm": 0.96484375, "learning_rate": 7.76811827333893e-05, "loss": 0.1232, "step": 16685 }, { "epoch": 12.224175824175823, "grad_norm": 1.140625, "learning_rate": 7.767831909728799e-05, "loss": 0.1353, "step": 16686 }, { "epoch": 12.224908424908424, "grad_norm": 0.9296875, "learning_rate": 7.767545533096643e-05, "loss": 0.1164, "step": 16687 }, { "epoch": 12.225641025641025, "grad_norm": 0.87109375, "learning_rate": 7.767259143443834e-05, "loss": 0.1105, "step": 16688 }, { "epoch": 12.226373626373626, "grad_norm": 0.90234375, "learning_rate": 7.766972740771744e-05, "loss": 0.1325, "step": 16689 }, { "epoch": 12.227106227106226, "grad_norm": 1.1328125, "learning_rate": 7.766686325081748e-05, "loss": 0.1434, "step": 16690 }, { "epoch": 12.227838827838827, "grad_norm": 1.109375, "learning_rate": 7.766399896375215e-05, "loss": 0.1418, "step": 16691 }, { "epoch": 12.228571428571428, "grad_norm": 0.84765625, "learning_rate": 7.766113454653518e-05, "loss": 0.1263, "step": 16692 }, { "epoch": 12.229304029304028, "grad_norm": 0.8671875, "learning_rate": 7.76582699991803e-05, "loss": 0.1126, "step": 16693 }, { "epoch": 12.23003663003663, "grad_norm": 1.09375, "learning_rate": 7.765540532170124e-05, "loss": 0.1486, "step": 16694 }, { "epoch": 12.23076923076923, "grad_norm": 1.2265625, "learning_rate": 7.76525405141117e-05, "loss": 0.1408, "step": 16695 }, { "epoch": 12.23150183150183, "grad_norm": 1.140625, "learning_rate": 7.764967557642544e-05, "loss": 0.1228, "step": 16696 }, { "epoch": 12.232234432234431, "grad_norm": 1.234375, "learning_rate": 7.764681050865617e-05, "loss": 0.1535, "step": 16697 }, { "epoch": 12.232967032967032, "grad_norm": 0.90234375, "learning_rate": 7.764394531081761e-05, "loss": 0.1213, "step": 16698 }, { "epoch": 12.233699633699633, "grad_norm": 1.25, "learning_rate": 7.764107998292351e-05, "loss": 0.1417, "step": 16699 }, { "epoch": 12.234432234432234, "grad_norm": 0.921875, "learning_rate": 7.763821452498757e-05, "loss": 0.1574, "step": 16700 }, { "epoch": 12.235164835164834, "grad_norm": 1.2421875, "learning_rate": 7.763534893702356e-05, "loss": 0.1153, "step": 16701 }, { "epoch": 12.235897435897435, "grad_norm": 0.94921875, "learning_rate": 7.763248321904516e-05, "loss": 0.1245, "step": 16702 }, { "epoch": 12.236630036630036, "grad_norm": 0.984375, "learning_rate": 7.762961737106615e-05, "loss": 0.1317, "step": 16703 }, { "epoch": 12.237362637362637, "grad_norm": 1.40625, "learning_rate": 7.762675139310021e-05, "loss": 0.1284, "step": 16704 }, { "epoch": 12.238095238095237, "grad_norm": 0.828125, "learning_rate": 7.762388528516111e-05, "loss": 0.1198, "step": 16705 }, { "epoch": 12.238827838827838, "grad_norm": 1.1953125, "learning_rate": 7.762101904726258e-05, "loss": 0.1197, "step": 16706 }, { "epoch": 12.239560439560439, "grad_norm": 1.7890625, "learning_rate": 7.761815267941834e-05, "loss": 0.1331, "step": 16707 }, { "epoch": 12.24029304029304, "grad_norm": 1.1796875, "learning_rate": 7.761528618164213e-05, "loss": 0.1268, "step": 16708 }, { "epoch": 12.24102564102564, "grad_norm": 0.8125, "learning_rate": 7.761241955394767e-05, "loss": 0.1156, "step": 16709 }, { "epoch": 12.241758241758241, "grad_norm": 1.8359375, "learning_rate": 7.76095527963487e-05, "loss": 0.1346, "step": 16710 }, { "epoch": 12.242490842490842, "grad_norm": 1.7578125, "learning_rate": 7.7606685908859e-05, "loss": 0.1472, "step": 16711 }, { "epoch": 12.243223443223442, "grad_norm": 0.546875, "learning_rate": 7.760381889149224e-05, "loss": 0.1185, "step": 16712 }, { "epoch": 12.243956043956043, "grad_norm": 0.90625, "learning_rate": 7.760095174426218e-05, "loss": 0.1067, "step": 16713 }, { "epoch": 12.244688644688644, "grad_norm": 1.0703125, "learning_rate": 7.75980844671826e-05, "loss": 0.1375, "step": 16714 }, { "epoch": 12.245421245421245, "grad_norm": 1.1015625, "learning_rate": 7.759521706026717e-05, "loss": 0.171, "step": 16715 }, { "epoch": 12.246153846153845, "grad_norm": 1.0625, "learning_rate": 7.759234952352967e-05, "loss": 0.1105, "step": 16716 }, { "epoch": 12.246886446886446, "grad_norm": 1.0, "learning_rate": 7.758948185698383e-05, "loss": 0.1643, "step": 16717 }, { "epoch": 12.247619047619047, "grad_norm": 1.3125, "learning_rate": 7.758661406064338e-05, "loss": 0.1571, "step": 16718 }, { "epoch": 12.248351648351647, "grad_norm": 1.125, "learning_rate": 7.758374613452209e-05, "loss": 0.1066, "step": 16719 }, { "epoch": 12.249084249084248, "grad_norm": 1.2734375, "learning_rate": 7.758087807863367e-05, "loss": 0.12, "step": 16720 }, { "epoch": 12.249816849816849, "grad_norm": 0.67578125, "learning_rate": 7.757800989299187e-05, "loss": 0.1369, "step": 16721 }, { "epoch": 12.25054945054945, "grad_norm": 1.4453125, "learning_rate": 7.757514157761045e-05, "loss": 0.1792, "step": 16722 }, { "epoch": 12.25128205128205, "grad_norm": 0.734375, "learning_rate": 7.757227313250311e-05, "loss": 0.1104, "step": 16723 }, { "epoch": 12.252014652014651, "grad_norm": 1.0546875, "learning_rate": 7.756940455768365e-05, "loss": 0.1409, "step": 16724 }, { "epoch": 12.252747252747252, "grad_norm": 0.94921875, "learning_rate": 7.756653585316578e-05, "loss": 0.1137, "step": 16725 }, { "epoch": 12.253479853479853, "grad_norm": 0.70703125, "learning_rate": 7.756366701896323e-05, "loss": 0.1475, "step": 16726 }, { "epoch": 12.254212454212453, "grad_norm": 0.73828125, "learning_rate": 7.75607980550898e-05, "loss": 0.1181, "step": 16727 }, { "epoch": 12.254945054945054, "grad_norm": 0.94140625, "learning_rate": 7.755792896155918e-05, "loss": 0.1241, "step": 16728 }, { "epoch": 12.255677655677655, "grad_norm": 1.03125, "learning_rate": 7.755505973838515e-05, "loss": 0.114, "step": 16729 }, { "epoch": 12.256410256410255, "grad_norm": 1.1328125, "learning_rate": 7.755219038558145e-05, "loss": 0.1325, "step": 16730 }, { "epoch": 12.257142857142856, "grad_norm": 0.482421875, "learning_rate": 7.754932090316182e-05, "loss": 0.112, "step": 16731 }, { "epoch": 12.257875457875459, "grad_norm": 0.78125, "learning_rate": 7.754645129114002e-05, "loss": 0.1061, "step": 16732 }, { "epoch": 12.25860805860806, "grad_norm": 1.0, "learning_rate": 7.754358154952978e-05, "loss": 0.129, "step": 16733 }, { "epoch": 12.25934065934066, "grad_norm": 0.55078125, "learning_rate": 7.754071167834486e-05, "loss": 0.1024, "step": 16734 }, { "epoch": 12.260073260073261, "grad_norm": 0.84765625, "learning_rate": 7.753784167759902e-05, "loss": 0.0916, "step": 16735 }, { "epoch": 12.260805860805862, "grad_norm": 0.66796875, "learning_rate": 7.753497154730602e-05, "loss": 0.116, "step": 16736 }, { "epoch": 12.261538461538462, "grad_norm": 0.90625, "learning_rate": 7.753210128747957e-05, "loss": 0.1302, "step": 16737 }, { "epoch": 12.262271062271063, "grad_norm": 1.4375, "learning_rate": 7.752923089813347e-05, "loss": 0.1317, "step": 16738 }, { "epoch": 12.263003663003664, "grad_norm": 1.0859375, "learning_rate": 7.752636037928143e-05, "loss": 0.1148, "step": 16739 }, { "epoch": 12.263736263736265, "grad_norm": 1.1171875, "learning_rate": 7.752348973093725e-05, "loss": 0.1185, "step": 16740 }, { "epoch": 12.264468864468865, "grad_norm": 1.4453125, "learning_rate": 7.752061895311465e-05, "loss": 0.1255, "step": 16741 }, { "epoch": 12.265201465201466, "grad_norm": 1.0625, "learning_rate": 7.75177480458274e-05, "loss": 0.1152, "step": 16742 }, { "epoch": 12.265934065934067, "grad_norm": 1.3046875, "learning_rate": 7.751487700908924e-05, "loss": 0.1479, "step": 16743 }, { "epoch": 12.266666666666667, "grad_norm": 1.015625, "learning_rate": 7.751200584291392e-05, "loss": 0.1665, "step": 16744 }, { "epoch": 12.267399267399268, "grad_norm": 1.0546875, "learning_rate": 7.750913454731524e-05, "loss": 0.1456, "step": 16745 }, { "epoch": 12.268131868131869, "grad_norm": 1.1484375, "learning_rate": 7.750626312230693e-05, "loss": 0.0996, "step": 16746 }, { "epoch": 12.26886446886447, "grad_norm": 0.73046875, "learning_rate": 7.750339156790273e-05, "loss": 0.1194, "step": 16747 }, { "epoch": 12.26959706959707, "grad_norm": 1.3671875, "learning_rate": 7.750051988411644e-05, "loss": 0.1302, "step": 16748 }, { "epoch": 12.270329670329671, "grad_norm": 0.9140625, "learning_rate": 7.749764807096177e-05, "loss": 0.1458, "step": 16749 }, { "epoch": 12.271062271062272, "grad_norm": 1.1796875, "learning_rate": 7.749477612845252e-05, "loss": 0.1323, "step": 16750 }, { "epoch": 12.271794871794873, "grad_norm": 1.0234375, "learning_rate": 7.749190405660244e-05, "loss": 0.1291, "step": 16751 }, { "epoch": 12.272527472527473, "grad_norm": 1.3515625, "learning_rate": 7.748903185542527e-05, "loss": 0.128, "step": 16752 }, { "epoch": 12.273260073260074, "grad_norm": 0.7890625, "learning_rate": 7.74861595249348e-05, "loss": 0.1337, "step": 16753 }, { "epoch": 12.273992673992675, "grad_norm": 1.2109375, "learning_rate": 7.748328706514477e-05, "loss": 0.1221, "step": 16754 }, { "epoch": 12.274725274725276, "grad_norm": 2.046875, "learning_rate": 7.748041447606897e-05, "loss": 0.1391, "step": 16755 }, { "epoch": 12.275457875457876, "grad_norm": 1.0, "learning_rate": 7.747754175772114e-05, "loss": 0.1225, "step": 16756 }, { "epoch": 12.276190476190477, "grad_norm": 0.8359375, "learning_rate": 7.747466891011504e-05, "loss": 0.1673, "step": 16757 }, { "epoch": 12.276923076923078, "grad_norm": 1.4921875, "learning_rate": 7.747179593326446e-05, "loss": 0.1149, "step": 16758 }, { "epoch": 12.277655677655678, "grad_norm": 1.453125, "learning_rate": 7.746892282718316e-05, "loss": 0.1176, "step": 16759 }, { "epoch": 12.27838827838828, "grad_norm": 1.609375, "learning_rate": 7.746604959188487e-05, "loss": 0.1186, "step": 16760 }, { "epoch": 12.27912087912088, "grad_norm": 1.0859375, "learning_rate": 7.746317622738341e-05, "loss": 0.1115, "step": 16761 }, { "epoch": 12.27985347985348, "grad_norm": 1.3046875, "learning_rate": 7.74603027336925e-05, "loss": 0.1334, "step": 16762 }, { "epoch": 12.280586080586081, "grad_norm": 0.90234375, "learning_rate": 7.745742911082595e-05, "loss": 0.1536, "step": 16763 }, { "epoch": 12.281318681318682, "grad_norm": 1.0, "learning_rate": 7.745455535879749e-05, "loss": 0.1431, "step": 16764 }, { "epoch": 12.282051282051283, "grad_norm": 1.0859375, "learning_rate": 7.74516814776209e-05, "loss": 0.1344, "step": 16765 }, { "epoch": 12.282783882783884, "grad_norm": 1.0234375, "learning_rate": 7.744880746730998e-05, "loss": 0.1309, "step": 16766 }, { "epoch": 12.283516483516484, "grad_norm": 1.2265625, "learning_rate": 7.744593332787847e-05, "loss": 0.1373, "step": 16767 }, { "epoch": 12.284249084249085, "grad_norm": 1.09375, "learning_rate": 7.744305905934014e-05, "loss": 0.1481, "step": 16768 }, { "epoch": 12.284981684981686, "grad_norm": 0.91796875, "learning_rate": 7.744018466170878e-05, "loss": 0.1332, "step": 16769 }, { "epoch": 12.285714285714286, "grad_norm": 1.3671875, "learning_rate": 7.743731013499814e-05, "loss": 0.1277, "step": 16770 }, { "epoch": 12.286446886446887, "grad_norm": 0.8515625, "learning_rate": 7.743443547922201e-05, "loss": 0.105, "step": 16771 }, { "epoch": 12.287179487179488, "grad_norm": 0.99609375, "learning_rate": 7.743156069439416e-05, "loss": 0.1113, "step": 16772 }, { "epoch": 12.287912087912089, "grad_norm": 1.9609375, "learning_rate": 7.742868578052836e-05, "loss": 0.1276, "step": 16773 }, { "epoch": 12.28864468864469, "grad_norm": 1.3046875, "learning_rate": 7.742581073763839e-05, "loss": 0.0979, "step": 16774 }, { "epoch": 12.28937728937729, "grad_norm": 1.015625, "learning_rate": 7.7422935565738e-05, "loss": 0.1492, "step": 16775 }, { "epoch": 12.29010989010989, "grad_norm": 1.125, "learning_rate": 7.742006026484101e-05, "loss": 0.1384, "step": 16776 }, { "epoch": 12.290842490842492, "grad_norm": 1.734375, "learning_rate": 7.741718483496117e-05, "loss": 0.1067, "step": 16777 }, { "epoch": 12.291575091575092, "grad_norm": 1.453125, "learning_rate": 7.741430927611226e-05, "loss": 0.1432, "step": 16778 }, { "epoch": 12.292307692307693, "grad_norm": 1.2578125, "learning_rate": 7.741143358830806e-05, "loss": 0.1102, "step": 16779 }, { "epoch": 12.293040293040294, "grad_norm": 0.91015625, "learning_rate": 7.740855777156235e-05, "loss": 0.1207, "step": 16780 }, { "epoch": 12.293772893772895, "grad_norm": 0.515625, "learning_rate": 7.74056818258889e-05, "loss": 0.1141, "step": 16781 }, { "epoch": 12.294505494505495, "grad_norm": 0.921875, "learning_rate": 7.74028057513015e-05, "loss": 0.1377, "step": 16782 }, { "epoch": 12.295238095238096, "grad_norm": 1.0546875, "learning_rate": 7.73999295478139e-05, "loss": 0.1277, "step": 16783 }, { "epoch": 12.295970695970697, "grad_norm": 1.09375, "learning_rate": 7.739705321543994e-05, "loss": 0.1199, "step": 16784 }, { "epoch": 12.296703296703297, "grad_norm": 1.0546875, "learning_rate": 7.739417675419337e-05, "loss": 0.1156, "step": 16785 }, { "epoch": 12.297435897435898, "grad_norm": 0.71484375, "learning_rate": 7.739130016408794e-05, "loss": 0.1075, "step": 16786 }, { "epoch": 12.298168498168499, "grad_norm": 0.97265625, "learning_rate": 7.73884234451375e-05, "loss": 0.1288, "step": 16787 }, { "epoch": 12.2989010989011, "grad_norm": 1.109375, "learning_rate": 7.738554659735576e-05, "loss": 0.1121, "step": 16788 }, { "epoch": 12.2996336996337, "grad_norm": 0.90234375, "learning_rate": 7.738266962075656e-05, "loss": 0.0985, "step": 16789 }, { "epoch": 12.300366300366301, "grad_norm": 0.83203125, "learning_rate": 7.737979251535367e-05, "loss": 0.1284, "step": 16790 }, { "epoch": 12.301098901098902, "grad_norm": 1.0234375, "learning_rate": 7.737691528116087e-05, "loss": 0.1173, "step": 16791 }, { "epoch": 12.301831501831503, "grad_norm": 1.1953125, "learning_rate": 7.737403791819194e-05, "loss": 0.1144, "step": 16792 }, { "epoch": 12.302564102564103, "grad_norm": 1.3125, "learning_rate": 7.737116042646068e-05, "loss": 0.1279, "step": 16793 }, { "epoch": 12.303296703296704, "grad_norm": 1.1171875, "learning_rate": 7.736828280598086e-05, "loss": 0.1557, "step": 16794 }, { "epoch": 12.304029304029305, "grad_norm": 1.2265625, "learning_rate": 7.736540505676627e-05, "loss": 0.14, "step": 16795 }, { "epoch": 12.304761904761905, "grad_norm": 2.171875, "learning_rate": 7.736252717883071e-05, "loss": 0.1021, "step": 16796 }, { "epoch": 12.305494505494506, "grad_norm": 1.28125, "learning_rate": 7.735964917218796e-05, "loss": 0.1208, "step": 16797 }, { "epoch": 12.306227106227107, "grad_norm": 1.1796875, "learning_rate": 7.735677103685183e-05, "loss": 0.1522, "step": 16798 }, { "epoch": 12.306959706959708, "grad_norm": 1.2265625, "learning_rate": 7.735389277283608e-05, "loss": 0.1171, "step": 16799 }, { "epoch": 12.307692307692308, "grad_norm": 1.3984375, "learning_rate": 7.735101438015451e-05, "loss": 0.1383, "step": 16800 }, { "epoch": 12.308424908424909, "grad_norm": 1.6328125, "learning_rate": 7.734813585882092e-05, "loss": 0.1437, "step": 16801 }, { "epoch": 12.30915750915751, "grad_norm": 1.71875, "learning_rate": 7.734525720884911e-05, "loss": 0.1053, "step": 16802 }, { "epoch": 12.30989010989011, "grad_norm": 1.1953125, "learning_rate": 7.734237843025285e-05, "loss": 0.141, "step": 16803 }, { "epoch": 12.310622710622711, "grad_norm": 0.859375, "learning_rate": 7.733949952304593e-05, "loss": 0.1068, "step": 16804 }, { "epoch": 12.311355311355312, "grad_norm": 1.2109375, "learning_rate": 7.733662048724219e-05, "loss": 0.1387, "step": 16805 }, { "epoch": 12.312087912087913, "grad_norm": 1.359375, "learning_rate": 7.733374132285536e-05, "loss": 0.1777, "step": 16806 }, { "epoch": 12.312820512820513, "grad_norm": 1.2578125, "learning_rate": 7.733086202989928e-05, "loss": 0.1557, "step": 16807 }, { "epoch": 12.313553113553114, "grad_norm": 1.0703125, "learning_rate": 7.732798260838772e-05, "loss": 0.1418, "step": 16808 }, { "epoch": 12.314285714285715, "grad_norm": 0.92578125, "learning_rate": 7.73251030583345e-05, "loss": 0.0994, "step": 16809 }, { "epoch": 12.315018315018316, "grad_norm": 1.7109375, "learning_rate": 7.732222337975342e-05, "loss": 0.1113, "step": 16810 }, { "epoch": 12.315750915750916, "grad_norm": 1.2109375, "learning_rate": 7.731934357265825e-05, "loss": 0.1303, "step": 16811 }, { "epoch": 12.316483516483517, "grad_norm": 1.5859375, "learning_rate": 7.731646363706277e-05, "loss": 0.132, "step": 16812 }, { "epoch": 12.317216117216118, "grad_norm": 1.2890625, "learning_rate": 7.731358357298086e-05, "loss": 0.1668, "step": 16813 }, { "epoch": 12.317948717948719, "grad_norm": 0.72265625, "learning_rate": 7.731070338042623e-05, "loss": 0.1002, "step": 16814 }, { "epoch": 12.31868131868132, "grad_norm": 1.3046875, "learning_rate": 7.730782305941274e-05, "loss": 0.1453, "step": 16815 }, { "epoch": 12.31941391941392, "grad_norm": 1.0, "learning_rate": 7.730494260995417e-05, "loss": 0.1311, "step": 16816 }, { "epoch": 12.32014652014652, "grad_norm": 1.4765625, "learning_rate": 7.73020620320643e-05, "loss": 0.1239, "step": 16817 }, { "epoch": 12.320879120879122, "grad_norm": 1.7734375, "learning_rate": 7.729918132575697e-05, "loss": 0.1475, "step": 16818 }, { "epoch": 12.321611721611722, "grad_norm": 1.140625, "learning_rate": 7.729630049104597e-05, "loss": 0.1471, "step": 16819 }, { "epoch": 12.322344322344323, "grad_norm": 0.6171875, "learning_rate": 7.729341952794509e-05, "loss": 0.1288, "step": 16820 }, { "epoch": 12.323076923076924, "grad_norm": 0.53515625, "learning_rate": 7.729053843646813e-05, "loss": 0.1046, "step": 16821 }, { "epoch": 12.323809523809524, "grad_norm": 1.2578125, "learning_rate": 7.728765721662892e-05, "loss": 0.1113, "step": 16822 }, { "epoch": 12.324542124542125, "grad_norm": 0.86328125, "learning_rate": 7.728477586844124e-05, "loss": 0.128, "step": 16823 }, { "epoch": 12.325274725274726, "grad_norm": 1.2578125, "learning_rate": 7.728189439191892e-05, "loss": 0.1168, "step": 16824 }, { "epoch": 12.326007326007327, "grad_norm": 1.375, "learning_rate": 7.727901278707574e-05, "loss": 0.1285, "step": 16825 }, { "epoch": 12.326739926739927, "grad_norm": 1.1484375, "learning_rate": 7.727613105392552e-05, "loss": 0.1624, "step": 16826 }, { "epoch": 12.327472527472528, "grad_norm": 0.97265625, "learning_rate": 7.727324919248206e-05, "loss": 0.1221, "step": 16827 }, { "epoch": 12.328205128205129, "grad_norm": 0.73046875, "learning_rate": 7.72703672027592e-05, "loss": 0.1131, "step": 16828 }, { "epoch": 12.32893772893773, "grad_norm": 1.4140625, "learning_rate": 7.72674850847707e-05, "loss": 0.1346, "step": 16829 }, { "epoch": 12.32967032967033, "grad_norm": 1.4375, "learning_rate": 7.726460283853038e-05, "loss": 0.124, "step": 16830 }, { "epoch": 12.330402930402931, "grad_norm": 1.4453125, "learning_rate": 7.726172046405208e-05, "loss": 0.1528, "step": 16831 }, { "epoch": 12.331135531135532, "grad_norm": 2.21875, "learning_rate": 7.725883796134961e-05, "loss": 0.1364, "step": 16832 }, { "epoch": 12.331868131868132, "grad_norm": 0.8671875, "learning_rate": 7.725595533043672e-05, "loss": 0.1017, "step": 16833 }, { "epoch": 12.332600732600733, "grad_norm": 1.6640625, "learning_rate": 7.72530725713273e-05, "loss": 0.1459, "step": 16834 }, { "epoch": 12.333333333333334, "grad_norm": 1.1796875, "learning_rate": 7.725018968403512e-05, "loss": 0.1082, "step": 16835 }, { "epoch": 12.334065934065935, "grad_norm": 1.5234375, "learning_rate": 7.724730666857399e-05, "loss": 0.1298, "step": 16836 }, { "epoch": 12.334798534798535, "grad_norm": 0.9453125, "learning_rate": 7.724442352495775e-05, "loss": 0.117, "step": 16837 }, { "epoch": 12.335531135531136, "grad_norm": 1.359375, "learning_rate": 7.724154025320017e-05, "loss": 0.1163, "step": 16838 }, { "epoch": 12.336263736263737, "grad_norm": 1.1484375, "learning_rate": 7.723865685331512e-05, "loss": 0.1289, "step": 16839 }, { "epoch": 12.336996336996338, "grad_norm": 1.2734375, "learning_rate": 7.723577332531637e-05, "loss": 0.1555, "step": 16840 }, { "epoch": 12.337728937728938, "grad_norm": 1.4296875, "learning_rate": 7.723288966921774e-05, "loss": 0.1183, "step": 16841 }, { "epoch": 12.338461538461539, "grad_norm": 2.09375, "learning_rate": 7.723000588503309e-05, "loss": 0.1352, "step": 16842 }, { "epoch": 12.33919413919414, "grad_norm": 2.40625, "learning_rate": 7.72271219727762e-05, "loss": 0.1487, "step": 16843 }, { "epoch": 12.33992673992674, "grad_norm": 1.1015625, "learning_rate": 7.722423793246087e-05, "loss": 0.1423, "step": 16844 }, { "epoch": 12.340659340659341, "grad_norm": 1.78125, "learning_rate": 7.722135376410096e-05, "loss": 0.1649, "step": 16845 }, { "epoch": 12.341391941391942, "grad_norm": 1.5, "learning_rate": 7.721846946771028e-05, "loss": 0.1367, "step": 16846 }, { "epoch": 12.342124542124543, "grad_norm": 0.8671875, "learning_rate": 7.721558504330265e-05, "loss": 0.1083, "step": 16847 }, { "epoch": 12.342857142857143, "grad_norm": 1.0234375, "learning_rate": 7.721270049089186e-05, "loss": 0.1285, "step": 16848 }, { "epoch": 12.343589743589744, "grad_norm": 1.3203125, "learning_rate": 7.720981581049176e-05, "loss": 0.1692, "step": 16849 }, { "epoch": 12.344322344322345, "grad_norm": 1.1875, "learning_rate": 7.720693100211617e-05, "loss": 0.1469, "step": 16850 }, { "epoch": 12.345054945054946, "grad_norm": 1.421875, "learning_rate": 7.720404606577889e-05, "loss": 0.1495, "step": 16851 }, { "epoch": 12.345787545787546, "grad_norm": 1.203125, "learning_rate": 7.720116100149378e-05, "loss": 0.1388, "step": 16852 }, { "epoch": 12.346520146520147, "grad_norm": 1.140625, "learning_rate": 7.719827580927462e-05, "loss": 0.1155, "step": 16853 }, { "epoch": 12.347252747252748, "grad_norm": 2.03125, "learning_rate": 7.719539048913527e-05, "loss": 0.15, "step": 16854 }, { "epoch": 12.347985347985349, "grad_norm": 1.5546875, "learning_rate": 7.719250504108953e-05, "loss": 0.1354, "step": 16855 }, { "epoch": 12.34871794871795, "grad_norm": 1.09375, "learning_rate": 7.718961946515125e-05, "loss": 0.1125, "step": 16856 }, { "epoch": 12.34945054945055, "grad_norm": 0.8828125, "learning_rate": 7.718673376133423e-05, "loss": 0.1143, "step": 16857 }, { "epoch": 12.35018315018315, "grad_norm": 1.734375, "learning_rate": 7.718384792965234e-05, "loss": 0.1326, "step": 16858 }, { "epoch": 12.350915750915751, "grad_norm": 1.53125, "learning_rate": 7.718096197011933e-05, "loss": 0.1235, "step": 16859 }, { "epoch": 12.351648351648352, "grad_norm": 1.53125, "learning_rate": 7.717807588274908e-05, "loss": 0.115, "step": 16860 }, { "epoch": 12.352380952380953, "grad_norm": 1.1875, "learning_rate": 7.717518966755543e-05, "loss": 0.1325, "step": 16861 }, { "epoch": 12.353113553113554, "grad_norm": 1.5625, "learning_rate": 7.717230332455219e-05, "loss": 0.118, "step": 16862 }, { "epoch": 12.353846153846154, "grad_norm": 1.140625, "learning_rate": 7.716941685375317e-05, "loss": 0.1105, "step": 16863 }, { "epoch": 12.354578754578755, "grad_norm": 1.09375, "learning_rate": 7.716653025517224e-05, "loss": 0.1363, "step": 16864 }, { "epoch": 12.355311355311356, "grad_norm": 1.0703125, "learning_rate": 7.71636435288232e-05, "loss": 0.1364, "step": 16865 }, { "epoch": 12.356043956043957, "grad_norm": 1.1328125, "learning_rate": 7.716075667471988e-05, "loss": 0.145, "step": 16866 }, { "epoch": 12.356776556776557, "grad_norm": 1.0, "learning_rate": 7.715786969287613e-05, "loss": 0.1206, "step": 16867 }, { "epoch": 12.357509157509158, "grad_norm": 1.2109375, "learning_rate": 7.715498258330579e-05, "loss": 0.1275, "step": 16868 }, { "epoch": 12.358241758241759, "grad_norm": 1.265625, "learning_rate": 7.715209534602266e-05, "loss": 0.1311, "step": 16869 }, { "epoch": 12.35897435897436, "grad_norm": 0.84375, "learning_rate": 7.714920798104058e-05, "loss": 0.1258, "step": 16870 }, { "epoch": 12.35970695970696, "grad_norm": 1.1640625, "learning_rate": 7.714632048837343e-05, "loss": 0.133, "step": 16871 }, { "epoch": 12.360439560439561, "grad_norm": 1.4921875, "learning_rate": 7.714343286803497e-05, "loss": 0.1538, "step": 16872 }, { "epoch": 12.361172161172162, "grad_norm": 1.046875, "learning_rate": 7.714054512003911e-05, "loss": 0.1413, "step": 16873 }, { "epoch": 12.361904761904762, "grad_norm": 1.4453125, "learning_rate": 7.713765724439964e-05, "loss": 0.1344, "step": 16874 }, { "epoch": 12.362637362637363, "grad_norm": 0.91796875, "learning_rate": 7.713476924113043e-05, "loss": 0.1328, "step": 16875 }, { "epoch": 12.363369963369964, "grad_norm": 1.234375, "learning_rate": 7.713188111024529e-05, "loss": 0.1245, "step": 16876 }, { "epoch": 12.364102564102565, "grad_norm": 0.921875, "learning_rate": 7.712899285175804e-05, "loss": 0.1152, "step": 16877 }, { "epoch": 12.364835164835165, "grad_norm": 0.76171875, "learning_rate": 7.712610446568256e-05, "loss": 0.1306, "step": 16878 }, { "epoch": 12.365567765567766, "grad_norm": 1.1328125, "learning_rate": 7.712321595203266e-05, "loss": 0.1403, "step": 16879 }, { "epoch": 12.366300366300367, "grad_norm": 1.3828125, "learning_rate": 7.712032731082221e-05, "loss": 0.1405, "step": 16880 }, { "epoch": 12.367032967032968, "grad_norm": 2.046875, "learning_rate": 7.711743854206504e-05, "loss": 0.152, "step": 16881 }, { "epoch": 12.367765567765568, "grad_norm": 1.0078125, "learning_rate": 7.711454964577497e-05, "loss": 0.1226, "step": 16882 }, { "epoch": 12.368498168498169, "grad_norm": 1.2265625, "learning_rate": 7.711166062196584e-05, "loss": 0.1633, "step": 16883 }, { "epoch": 12.36923076923077, "grad_norm": 1.390625, "learning_rate": 7.710877147065153e-05, "loss": 0.1397, "step": 16884 }, { "epoch": 12.36996336996337, "grad_norm": 1.2578125, "learning_rate": 7.710588219184586e-05, "loss": 0.1545, "step": 16885 }, { "epoch": 12.370695970695971, "grad_norm": 1.4609375, "learning_rate": 7.710299278556267e-05, "loss": 0.1495, "step": 16886 }, { "epoch": 12.371428571428572, "grad_norm": 1.0234375, "learning_rate": 7.710010325181579e-05, "loss": 0.118, "step": 16887 }, { "epoch": 12.372161172161173, "grad_norm": 1.109375, "learning_rate": 7.70972135906191e-05, "loss": 0.1201, "step": 16888 }, { "epoch": 12.372893772893773, "grad_norm": 0.75, "learning_rate": 7.709432380198644e-05, "loss": 0.1035, "step": 16889 }, { "epoch": 12.373626373626374, "grad_norm": 0.78125, "learning_rate": 7.709143388593164e-05, "loss": 0.1464, "step": 16890 }, { "epoch": 12.374358974358975, "grad_norm": 0.7421875, "learning_rate": 7.708854384246854e-05, "loss": 0.1314, "step": 16891 }, { "epoch": 12.375091575091576, "grad_norm": 1.375, "learning_rate": 7.708565367161099e-05, "loss": 0.1866, "step": 16892 }, { "epoch": 12.375824175824176, "grad_norm": 1.0625, "learning_rate": 7.708276337337287e-05, "loss": 0.1548, "step": 16893 }, { "epoch": 12.376556776556777, "grad_norm": 0.94921875, "learning_rate": 7.7079872947768e-05, "loss": 0.1356, "step": 16894 }, { "epoch": 12.377289377289378, "grad_norm": 1.0859375, "learning_rate": 7.707698239481021e-05, "loss": 0.1264, "step": 16895 }, { "epoch": 12.378021978021978, "grad_norm": 1.96875, "learning_rate": 7.70740917145134e-05, "loss": 0.164, "step": 16896 }, { "epoch": 12.37875457875458, "grad_norm": 1.0859375, "learning_rate": 7.707120090689138e-05, "loss": 0.1006, "step": 16897 }, { "epoch": 12.37948717948718, "grad_norm": 0.7421875, "learning_rate": 7.706830997195801e-05, "loss": 0.1374, "step": 16898 }, { "epoch": 12.38021978021978, "grad_norm": 1.25, "learning_rate": 7.706541890972715e-05, "loss": 0.1517, "step": 16899 }, { "epoch": 12.380952380952381, "grad_norm": 1.1796875, "learning_rate": 7.706252772021264e-05, "loss": 0.1565, "step": 16900 }, { "epoch": 12.381684981684982, "grad_norm": 1.203125, "learning_rate": 7.705963640342835e-05, "loss": 0.1431, "step": 16901 }, { "epoch": 12.382417582417583, "grad_norm": 0.82421875, "learning_rate": 7.705674495938812e-05, "loss": 0.1321, "step": 16902 }, { "epoch": 12.383150183150184, "grad_norm": 1.1640625, "learning_rate": 7.705385338810581e-05, "loss": 0.1394, "step": 16903 }, { "epoch": 12.383882783882784, "grad_norm": 1.265625, "learning_rate": 7.705096168959528e-05, "loss": 0.1264, "step": 16904 }, { "epoch": 12.384615384615385, "grad_norm": 1.3984375, "learning_rate": 7.704806986387034e-05, "loss": 0.14, "step": 16905 }, { "epoch": 12.385347985347986, "grad_norm": 1.4921875, "learning_rate": 7.704517791094492e-05, "loss": 0.1338, "step": 16906 }, { "epoch": 12.386080586080586, "grad_norm": 0.8359375, "learning_rate": 7.704228583083282e-05, "loss": 0.119, "step": 16907 }, { "epoch": 12.386813186813187, "grad_norm": 1.046875, "learning_rate": 7.703939362354793e-05, "loss": 0.1, "step": 16908 }, { "epoch": 12.387545787545788, "grad_norm": 0.9296875, "learning_rate": 7.703650128910407e-05, "loss": 0.1035, "step": 16909 }, { "epoch": 12.388278388278389, "grad_norm": 1.1171875, "learning_rate": 7.703360882751514e-05, "loss": 0.1318, "step": 16910 }, { "epoch": 12.38901098901099, "grad_norm": 1.28125, "learning_rate": 7.703071623879497e-05, "loss": 0.1393, "step": 16911 }, { "epoch": 12.38974358974359, "grad_norm": 1.453125, "learning_rate": 7.702782352295743e-05, "loss": 0.1179, "step": 16912 }, { "epoch": 12.39047619047619, "grad_norm": 1.59375, "learning_rate": 7.702493068001638e-05, "loss": 0.1319, "step": 16913 }, { "epoch": 12.391208791208792, "grad_norm": 0.83984375, "learning_rate": 7.70220377099857e-05, "loss": 0.1154, "step": 16914 }, { "epoch": 12.391941391941392, "grad_norm": 0.9921875, "learning_rate": 7.70191446128792e-05, "loss": 0.1236, "step": 16915 }, { "epoch": 12.392673992673993, "grad_norm": 1.4453125, "learning_rate": 7.701625138871079e-05, "loss": 0.1293, "step": 16916 }, { "epoch": 12.393406593406594, "grad_norm": 1.0078125, "learning_rate": 7.70133580374943e-05, "loss": 0.1491, "step": 16917 }, { "epoch": 12.394139194139195, "grad_norm": 2.015625, "learning_rate": 7.701046455924364e-05, "loss": 0.1772, "step": 16918 }, { "epoch": 12.394871794871795, "grad_norm": 1.09375, "learning_rate": 7.70075709539726e-05, "loss": 0.1312, "step": 16919 }, { "epoch": 12.395604395604396, "grad_norm": 1.34375, "learning_rate": 7.700467722169511e-05, "loss": 0.1041, "step": 16920 }, { "epoch": 12.396336996336997, "grad_norm": 0.80859375, "learning_rate": 7.700178336242501e-05, "loss": 0.1115, "step": 16921 }, { "epoch": 12.397069597069597, "grad_norm": 1.1015625, "learning_rate": 7.699888937617615e-05, "loss": 0.1119, "step": 16922 }, { "epoch": 12.397802197802198, "grad_norm": 0.65234375, "learning_rate": 7.699599526296243e-05, "loss": 0.1132, "step": 16923 }, { "epoch": 12.398534798534799, "grad_norm": 1.46875, "learning_rate": 7.699310102279768e-05, "loss": 0.1099, "step": 16924 }, { "epoch": 12.3992673992674, "grad_norm": 1.0078125, "learning_rate": 7.69902066556958e-05, "loss": 0.1111, "step": 16925 }, { "epoch": 12.4, "grad_norm": 1.3125, "learning_rate": 7.698731216167064e-05, "loss": 0.1437, "step": 16926 }, { "epoch": 12.400732600732601, "grad_norm": 1.359375, "learning_rate": 7.698441754073607e-05, "loss": 0.1603, "step": 16927 }, { "epoch": 12.401465201465202, "grad_norm": 1.4296875, "learning_rate": 7.698152279290599e-05, "loss": 0.1234, "step": 16928 }, { "epoch": 12.402197802197803, "grad_norm": 1.0703125, "learning_rate": 7.69786279181942e-05, "loss": 0.148, "step": 16929 }, { "epoch": 12.402930402930403, "grad_norm": 0.8359375, "learning_rate": 7.697573291661463e-05, "loss": 0.1403, "step": 16930 }, { "epoch": 12.403663003663004, "grad_norm": 0.91015625, "learning_rate": 7.697283778818112e-05, "loss": 0.1215, "step": 16931 }, { "epoch": 12.404395604395605, "grad_norm": 1.4375, "learning_rate": 7.696994253290757e-05, "loss": 0.11, "step": 16932 }, { "epoch": 12.405128205128205, "grad_norm": 1.0625, "learning_rate": 7.696704715080783e-05, "loss": 0.1355, "step": 16933 }, { "epoch": 12.405860805860806, "grad_norm": 1.359375, "learning_rate": 7.696415164189578e-05, "loss": 0.125, "step": 16934 }, { "epoch": 12.406593406593407, "grad_norm": 0.92578125, "learning_rate": 7.696125600618528e-05, "loss": 0.1158, "step": 16935 }, { "epoch": 12.407326007326008, "grad_norm": 1.9375, "learning_rate": 7.695836024369023e-05, "loss": 0.1267, "step": 16936 }, { "epoch": 12.408058608058608, "grad_norm": 1.890625, "learning_rate": 7.695546435442448e-05, "loss": 0.1621, "step": 16937 }, { "epoch": 12.40879120879121, "grad_norm": 1.0625, "learning_rate": 7.695256833840193e-05, "loss": 0.1088, "step": 16938 }, { "epoch": 12.40952380952381, "grad_norm": 1.4609375, "learning_rate": 7.694967219563642e-05, "loss": 0.131, "step": 16939 }, { "epoch": 12.41025641025641, "grad_norm": 1.34375, "learning_rate": 7.694677592614185e-05, "loss": 0.1245, "step": 16940 }, { "epoch": 12.410989010989011, "grad_norm": 0.984375, "learning_rate": 7.694387952993213e-05, "loss": 0.1334, "step": 16941 }, { "epoch": 12.411721611721612, "grad_norm": 0.9921875, "learning_rate": 7.694098300702106e-05, "loss": 0.1351, "step": 16942 }, { "epoch": 12.412454212454213, "grad_norm": 1.125, "learning_rate": 7.693808635742258e-05, "loss": 0.1082, "step": 16943 }, { "epoch": 12.413186813186813, "grad_norm": 1.0234375, "learning_rate": 7.693518958115055e-05, "loss": 0.1213, "step": 16944 }, { "epoch": 12.413919413919414, "grad_norm": 1.4296875, "learning_rate": 7.693229267821885e-05, "loss": 0.1376, "step": 16945 }, { "epoch": 12.414652014652015, "grad_norm": 1.3359375, "learning_rate": 7.692939564864135e-05, "loss": 0.1119, "step": 16946 }, { "epoch": 12.415384615384616, "grad_norm": 1.1328125, "learning_rate": 7.692649849243195e-05, "loss": 0.1519, "step": 16947 }, { "epoch": 12.416117216117216, "grad_norm": 1.2734375, "learning_rate": 7.69236012096045e-05, "loss": 0.118, "step": 16948 }, { "epoch": 12.416849816849817, "grad_norm": 1.5, "learning_rate": 7.692070380017293e-05, "loss": 0.145, "step": 16949 }, { "epoch": 12.417582417582418, "grad_norm": 1.0, "learning_rate": 7.691780626415106e-05, "loss": 0.1221, "step": 16950 }, { "epoch": 12.418315018315019, "grad_norm": 1.1328125, "learning_rate": 7.691490860155283e-05, "loss": 0.1245, "step": 16951 }, { "epoch": 12.41904761904762, "grad_norm": 1.578125, "learning_rate": 7.69120108123921e-05, "loss": 0.1362, "step": 16952 }, { "epoch": 12.41978021978022, "grad_norm": 0.97265625, "learning_rate": 7.690911289668276e-05, "loss": 0.1259, "step": 16953 }, { "epoch": 12.42051282051282, "grad_norm": 1.171875, "learning_rate": 7.69062148544387e-05, "loss": 0.1183, "step": 16954 }, { "epoch": 12.421245421245422, "grad_norm": 1.7734375, "learning_rate": 7.690331668567377e-05, "loss": 0.2137, "step": 16955 }, { "epoch": 12.421978021978022, "grad_norm": 1.1484375, "learning_rate": 7.69004183904019e-05, "loss": 0.11, "step": 16956 }, { "epoch": 12.422710622710623, "grad_norm": 0.73046875, "learning_rate": 7.689751996863693e-05, "loss": 0.1194, "step": 16957 }, { "epoch": 12.423443223443224, "grad_norm": 1.3671875, "learning_rate": 7.68946214203928e-05, "loss": 0.1394, "step": 16958 }, { "epoch": 12.424175824175824, "grad_norm": 0.91015625, "learning_rate": 7.689172274568339e-05, "loss": 0.1371, "step": 16959 }, { "epoch": 12.424908424908425, "grad_norm": 1.34375, "learning_rate": 7.688882394452255e-05, "loss": 0.1067, "step": 16960 }, { "epoch": 12.425641025641026, "grad_norm": 0.8046875, "learning_rate": 7.68859250169242e-05, "loss": 0.1167, "step": 16961 }, { "epoch": 12.426373626373627, "grad_norm": 1.1484375, "learning_rate": 7.688302596290221e-05, "loss": 0.1223, "step": 16962 }, { "epoch": 12.427106227106227, "grad_norm": 1.5, "learning_rate": 7.688012678247049e-05, "loss": 0.198, "step": 16963 }, { "epoch": 12.427838827838828, "grad_norm": 0.70703125, "learning_rate": 7.687722747564293e-05, "loss": 0.102, "step": 16964 }, { "epoch": 12.428571428571429, "grad_norm": 1.2890625, "learning_rate": 7.687432804243341e-05, "loss": 0.1433, "step": 16965 }, { "epoch": 12.42930402930403, "grad_norm": 0.7109375, "learning_rate": 7.687142848285581e-05, "loss": 0.1196, "step": 16966 }, { "epoch": 12.43003663003663, "grad_norm": 1.015625, "learning_rate": 7.686852879692406e-05, "loss": 0.129, "step": 16967 }, { "epoch": 12.430769230769231, "grad_norm": 0.69140625, "learning_rate": 7.686562898465202e-05, "loss": 0.1332, "step": 16968 }, { "epoch": 12.431501831501832, "grad_norm": 0.75390625, "learning_rate": 7.686272904605361e-05, "loss": 0.1084, "step": 16969 }, { "epoch": 12.432234432234432, "grad_norm": 0.81640625, "learning_rate": 7.685982898114269e-05, "loss": 0.1108, "step": 16970 }, { "epoch": 12.432967032967033, "grad_norm": 0.94140625, "learning_rate": 7.68569287899332e-05, "loss": 0.1312, "step": 16971 }, { "epoch": 12.433699633699634, "grad_norm": 1.1484375, "learning_rate": 7.6854028472439e-05, "loss": 0.1092, "step": 16972 }, { "epoch": 12.434432234432235, "grad_norm": 1.4765625, "learning_rate": 7.685112802867399e-05, "loss": 0.1397, "step": 16973 }, { "epoch": 12.435164835164835, "grad_norm": 1.3671875, "learning_rate": 7.684822745865207e-05, "loss": 0.1152, "step": 16974 }, { "epoch": 12.435897435897436, "grad_norm": 1.015625, "learning_rate": 7.684532676238715e-05, "loss": 0.1215, "step": 16975 }, { "epoch": 12.436630036630037, "grad_norm": 0.60546875, "learning_rate": 7.684242593989313e-05, "loss": 0.1105, "step": 16976 }, { "epoch": 12.437362637362638, "grad_norm": 1.21875, "learning_rate": 7.68395249911839e-05, "loss": 0.1358, "step": 16977 }, { "epoch": 12.438095238095238, "grad_norm": 0.69140625, "learning_rate": 7.683662391627335e-05, "loss": 0.11, "step": 16978 }, { "epoch": 12.438827838827839, "grad_norm": 0.78515625, "learning_rate": 7.68337227151754e-05, "loss": 0.1203, "step": 16979 }, { "epoch": 12.43956043956044, "grad_norm": 1.171875, "learning_rate": 7.683082138790394e-05, "loss": 0.1461, "step": 16980 }, { "epoch": 12.44029304029304, "grad_norm": 1.46875, "learning_rate": 7.682791993447285e-05, "loss": 0.1459, "step": 16981 }, { "epoch": 12.441025641025641, "grad_norm": 0.984375, "learning_rate": 7.682501835489608e-05, "loss": 0.1214, "step": 16982 }, { "epoch": 12.441758241758242, "grad_norm": 1.0, "learning_rate": 7.68221166491875e-05, "loss": 0.1372, "step": 16983 }, { "epoch": 12.442490842490843, "grad_norm": 0.71875, "learning_rate": 7.6819214817361e-05, "loss": 0.1095, "step": 16984 }, { "epoch": 12.443223443223443, "grad_norm": 1.109375, "learning_rate": 7.68163128594305e-05, "loss": 0.1283, "step": 16985 }, { "epoch": 12.443956043956044, "grad_norm": 0.8984375, "learning_rate": 7.681341077540993e-05, "loss": 0.1159, "step": 16986 }, { "epoch": 12.444688644688645, "grad_norm": 1.3046875, "learning_rate": 7.681050856531316e-05, "loss": 0.1765, "step": 16987 }, { "epoch": 12.445421245421246, "grad_norm": 0.79296875, "learning_rate": 7.68076062291541e-05, "loss": 0.1334, "step": 16988 }, { "epoch": 12.446153846153846, "grad_norm": 0.90234375, "learning_rate": 7.680470376694665e-05, "loss": 0.1194, "step": 16989 }, { "epoch": 12.446886446886447, "grad_norm": 0.96875, "learning_rate": 7.680180117870477e-05, "loss": 0.1232, "step": 16990 }, { "epoch": 12.447619047619048, "grad_norm": 0.98046875, "learning_rate": 7.67988984644423e-05, "loss": 0.1217, "step": 16991 }, { "epoch": 12.448351648351649, "grad_norm": 1.1015625, "learning_rate": 7.679599562417318e-05, "loss": 0.1282, "step": 16992 }, { "epoch": 12.44908424908425, "grad_norm": 1.515625, "learning_rate": 7.679309265791131e-05, "loss": 0.1219, "step": 16993 }, { "epoch": 12.44981684981685, "grad_norm": 1.234375, "learning_rate": 7.679018956567059e-05, "loss": 0.1255, "step": 16994 }, { "epoch": 12.45054945054945, "grad_norm": 1.4765625, "learning_rate": 7.678728634746495e-05, "loss": 0.1227, "step": 16995 }, { "epoch": 12.451282051282051, "grad_norm": 1.109375, "learning_rate": 7.67843830033083e-05, "loss": 0.1527, "step": 16996 }, { "epoch": 12.452014652014652, "grad_norm": 1.4453125, "learning_rate": 7.678147953321452e-05, "loss": 0.1381, "step": 16997 }, { "epoch": 12.452747252747253, "grad_norm": 0.9921875, "learning_rate": 7.677857593719757e-05, "loss": 0.107, "step": 16998 }, { "epoch": 12.453479853479854, "grad_norm": 1.03125, "learning_rate": 7.677567221527132e-05, "loss": 0.1526, "step": 16999 }, { "epoch": 12.454212454212454, "grad_norm": 1.4609375, "learning_rate": 7.677276836744969e-05, "loss": 0.1819, "step": 17000 }, { "epoch": 12.454945054945055, "grad_norm": 1.40625, "learning_rate": 7.676986439374664e-05, "loss": 0.1136, "step": 17001 }, { "epoch": 12.455677655677656, "grad_norm": 0.96875, "learning_rate": 7.676696029417602e-05, "loss": 0.1274, "step": 17002 }, { "epoch": 12.456410256410257, "grad_norm": 0.90234375, "learning_rate": 7.676405606875177e-05, "loss": 0.1311, "step": 17003 }, { "epoch": 12.457142857142857, "grad_norm": 1.75, "learning_rate": 7.676115171748781e-05, "loss": 0.176, "step": 17004 }, { "epoch": 12.457875457875458, "grad_norm": 0.99609375, "learning_rate": 7.675824724039804e-05, "loss": 0.1218, "step": 17005 }, { "epoch": 12.458608058608059, "grad_norm": 1.1796875, "learning_rate": 7.675534263749641e-05, "loss": 0.124, "step": 17006 }, { "epoch": 12.45934065934066, "grad_norm": 1.1484375, "learning_rate": 7.67524379087968e-05, "loss": 0.1418, "step": 17007 }, { "epoch": 12.46007326007326, "grad_norm": 1.46875, "learning_rate": 7.674953305431314e-05, "loss": 0.1097, "step": 17008 }, { "epoch": 12.460805860805861, "grad_norm": 1.5546875, "learning_rate": 7.674662807405937e-05, "loss": 0.1434, "step": 17009 }, { "epoch": 12.461538461538462, "grad_norm": 1.5625, "learning_rate": 7.674372296804935e-05, "loss": 0.1652, "step": 17010 }, { "epoch": 12.462271062271062, "grad_norm": 0.953125, "learning_rate": 7.674081773629705e-05, "loss": 0.1404, "step": 17011 }, { "epoch": 12.463003663003663, "grad_norm": 1.0234375, "learning_rate": 7.67379123788164e-05, "loss": 0.1585, "step": 17012 }, { "epoch": 12.463736263736264, "grad_norm": 0.7578125, "learning_rate": 7.673500689562127e-05, "loss": 0.1351, "step": 17013 }, { "epoch": 12.464468864468865, "grad_norm": 1.265625, "learning_rate": 7.673210128672564e-05, "loss": 0.136, "step": 17014 }, { "epoch": 12.465201465201465, "grad_norm": 1.3046875, "learning_rate": 7.67291955521434e-05, "loss": 0.1349, "step": 17015 }, { "epoch": 12.465934065934066, "grad_norm": 1.21875, "learning_rate": 7.672628969188845e-05, "loss": 0.1408, "step": 17016 }, { "epoch": 12.466666666666667, "grad_norm": 0.74609375, "learning_rate": 7.672338370597474e-05, "loss": 0.1102, "step": 17017 }, { "epoch": 12.467399267399268, "grad_norm": 1.2578125, "learning_rate": 7.67204775944162e-05, "loss": 0.127, "step": 17018 }, { "epoch": 12.468131868131868, "grad_norm": 0.69921875, "learning_rate": 7.671757135722675e-05, "loss": 0.1272, "step": 17019 }, { "epoch": 12.468864468864469, "grad_norm": 0.84765625, "learning_rate": 7.671466499442029e-05, "loss": 0.1245, "step": 17020 }, { "epoch": 12.46959706959707, "grad_norm": 1.421875, "learning_rate": 7.671175850601077e-05, "loss": 0.143, "step": 17021 }, { "epoch": 12.47032967032967, "grad_norm": 1.2890625, "learning_rate": 7.670885189201213e-05, "loss": 0.1403, "step": 17022 }, { "epoch": 12.471062271062271, "grad_norm": 1.46875, "learning_rate": 7.670594515243827e-05, "loss": 0.1093, "step": 17023 }, { "epoch": 12.471794871794872, "grad_norm": 1.1796875, "learning_rate": 7.67030382873031e-05, "loss": 0.1402, "step": 17024 }, { "epoch": 12.472527472527473, "grad_norm": 1.1171875, "learning_rate": 7.670013129662058e-05, "loss": 0.1297, "step": 17025 }, { "epoch": 12.473260073260073, "grad_norm": 0.83984375, "learning_rate": 7.669722418040464e-05, "loss": 0.1267, "step": 17026 }, { "epoch": 12.473992673992674, "grad_norm": 0.9296875, "learning_rate": 7.66943169386692e-05, "loss": 0.1291, "step": 17027 }, { "epoch": 12.474725274725275, "grad_norm": 1.1015625, "learning_rate": 7.669140957142818e-05, "loss": 0.1072, "step": 17028 }, { "epoch": 12.475457875457876, "grad_norm": 1.390625, "learning_rate": 7.668850207869554e-05, "loss": 0.1087, "step": 17029 }, { "epoch": 12.476190476190476, "grad_norm": 0.86328125, "learning_rate": 7.668559446048515e-05, "loss": 0.1338, "step": 17030 }, { "epoch": 12.476923076923077, "grad_norm": 1.421875, "learning_rate": 7.668268671681102e-05, "loss": 0.1357, "step": 17031 }, { "epoch": 12.477655677655678, "grad_norm": 1.015625, "learning_rate": 7.667977884768702e-05, "loss": 0.106, "step": 17032 }, { "epoch": 12.478388278388278, "grad_norm": 1.296875, "learning_rate": 7.667687085312711e-05, "loss": 0.1095, "step": 17033 }, { "epoch": 12.47912087912088, "grad_norm": 1.2734375, "learning_rate": 7.667396273314523e-05, "loss": 0.1457, "step": 17034 }, { "epoch": 12.47985347985348, "grad_norm": 1.21875, "learning_rate": 7.66710544877553e-05, "loss": 0.1183, "step": 17035 }, { "epoch": 12.48058608058608, "grad_norm": 0.7734375, "learning_rate": 7.666814611697123e-05, "loss": 0.1361, "step": 17036 }, { "epoch": 12.481318681318681, "grad_norm": 0.94140625, "learning_rate": 7.6665237620807e-05, "loss": 0.1272, "step": 17037 }, { "epoch": 12.482051282051282, "grad_norm": 1.4765625, "learning_rate": 7.666232899927653e-05, "loss": 0.1273, "step": 17038 }, { "epoch": 12.482783882783883, "grad_norm": 1.125, "learning_rate": 7.665942025239376e-05, "loss": 0.0912, "step": 17039 }, { "epoch": 12.483516483516484, "grad_norm": 1.3125, "learning_rate": 7.66565113801726e-05, "loss": 0.1699, "step": 17040 }, { "epoch": 12.484249084249084, "grad_norm": 1.125, "learning_rate": 7.665360238262701e-05, "loss": 0.1456, "step": 17041 }, { "epoch": 12.484981684981685, "grad_norm": 1.140625, "learning_rate": 7.665069325977094e-05, "loss": 0.1658, "step": 17042 }, { "epoch": 12.485714285714286, "grad_norm": 1.1953125, "learning_rate": 7.664778401161831e-05, "loss": 0.1515, "step": 17043 }, { "epoch": 12.486446886446886, "grad_norm": 0.57421875, "learning_rate": 7.664487463818307e-05, "loss": 0.1157, "step": 17044 }, { "epoch": 12.487179487179487, "grad_norm": 0.98828125, "learning_rate": 7.664196513947915e-05, "loss": 0.1166, "step": 17045 }, { "epoch": 12.487912087912088, "grad_norm": 0.9921875, "learning_rate": 7.663905551552047e-05, "loss": 0.1072, "step": 17046 }, { "epoch": 12.488644688644689, "grad_norm": 1.1171875, "learning_rate": 7.663614576632103e-05, "loss": 0.1114, "step": 17047 }, { "epoch": 12.48937728937729, "grad_norm": 1.171875, "learning_rate": 7.663323589189469e-05, "loss": 0.1516, "step": 17048 }, { "epoch": 12.49010989010989, "grad_norm": 0.8984375, "learning_rate": 7.663032589225547e-05, "loss": 0.1432, "step": 17049 }, { "epoch": 12.49084249084249, "grad_norm": 0.80859375, "learning_rate": 7.662741576741726e-05, "loss": 0.0948, "step": 17050 }, { "epoch": 12.491575091575092, "grad_norm": 0.94140625, "learning_rate": 7.662450551739405e-05, "loss": 0.1124, "step": 17051 }, { "epoch": 12.492307692307692, "grad_norm": 1.1640625, "learning_rate": 7.662159514219975e-05, "loss": 0.1291, "step": 17052 }, { "epoch": 12.493040293040293, "grad_norm": 1.328125, "learning_rate": 7.66186846418483e-05, "loss": 0.148, "step": 17053 }, { "epoch": 12.493772893772894, "grad_norm": 1.0390625, "learning_rate": 7.661577401635368e-05, "loss": 0.1072, "step": 17054 }, { "epoch": 12.494505494505495, "grad_norm": 0.921875, "learning_rate": 7.661286326572979e-05, "loss": 0.1269, "step": 17055 }, { "epoch": 12.495238095238095, "grad_norm": 1.25, "learning_rate": 7.660995238999061e-05, "loss": 0.1535, "step": 17056 }, { "epoch": 12.495970695970696, "grad_norm": 1.390625, "learning_rate": 7.660704138915008e-05, "loss": 0.1426, "step": 17057 }, { "epoch": 12.496703296703297, "grad_norm": 0.9296875, "learning_rate": 7.660413026322215e-05, "loss": 0.1265, "step": 17058 }, { "epoch": 12.497435897435897, "grad_norm": 1.0625, "learning_rate": 7.660121901222075e-05, "loss": 0.1107, "step": 17059 }, { "epoch": 12.498168498168498, "grad_norm": 1.515625, "learning_rate": 7.659830763615985e-05, "loss": 0.1401, "step": 17060 }, { "epoch": 12.498901098901099, "grad_norm": 0.7890625, "learning_rate": 7.659539613505338e-05, "loss": 0.1038, "step": 17061 }, { "epoch": 12.4996336996337, "grad_norm": 1.3359375, "learning_rate": 7.65924845089153e-05, "loss": 0.1088, "step": 17062 }, { "epoch": 12.5003663003663, "grad_norm": 1.171875, "learning_rate": 7.658957275775958e-05, "loss": 0.0969, "step": 17063 }, { "epoch": 12.501098901098901, "grad_norm": 0.93359375, "learning_rate": 7.658666088160013e-05, "loss": 0.1222, "step": 17064 }, { "epoch": 12.501831501831502, "grad_norm": 2.0625, "learning_rate": 7.658374888045092e-05, "loss": 0.123, "step": 17065 }, { "epoch": 12.502564102564103, "grad_norm": 1.171875, "learning_rate": 7.658083675432593e-05, "loss": 0.1396, "step": 17066 }, { "epoch": 12.503296703296703, "grad_norm": 0.7890625, "learning_rate": 7.657792450323907e-05, "loss": 0.1189, "step": 17067 }, { "epoch": 12.504029304029304, "grad_norm": 1.09375, "learning_rate": 7.657501212720434e-05, "loss": 0.1387, "step": 17068 }, { "epoch": 12.504761904761905, "grad_norm": 1.0078125, "learning_rate": 7.657209962623564e-05, "loss": 0.1477, "step": 17069 }, { "epoch": 12.505494505494505, "grad_norm": 1.6171875, "learning_rate": 7.656918700034695e-05, "loss": 0.1288, "step": 17070 }, { "epoch": 12.506227106227106, "grad_norm": 1.75, "learning_rate": 7.656627424955225e-05, "loss": 0.1283, "step": 17071 }, { "epoch": 12.506959706959707, "grad_norm": 1.296875, "learning_rate": 7.656336137386546e-05, "loss": 0.1226, "step": 17072 }, { "epoch": 12.507692307692308, "grad_norm": 1.09375, "learning_rate": 7.656044837330054e-05, "loss": 0.1265, "step": 17073 }, { "epoch": 12.508424908424908, "grad_norm": 1.4375, "learning_rate": 7.655753524787148e-05, "loss": 0.1547, "step": 17074 }, { "epoch": 12.50915750915751, "grad_norm": 0.8125, "learning_rate": 7.655462199759218e-05, "loss": 0.0994, "step": 17075 }, { "epoch": 12.50989010989011, "grad_norm": 0.8984375, "learning_rate": 7.655170862247668e-05, "loss": 0.1313, "step": 17076 }, { "epoch": 12.51062271062271, "grad_norm": 1.7890625, "learning_rate": 7.654879512253884e-05, "loss": 0.1376, "step": 17077 }, { "epoch": 12.511355311355311, "grad_norm": 2.203125, "learning_rate": 7.654588149779272e-05, "loss": 0.1178, "step": 17078 }, { "epoch": 12.512087912087912, "grad_norm": 1.9921875, "learning_rate": 7.654296774825222e-05, "loss": 0.1279, "step": 17079 }, { "epoch": 12.512820512820513, "grad_norm": 1.53125, "learning_rate": 7.654005387393129e-05, "loss": 0.1086, "step": 17080 }, { "epoch": 12.513553113553114, "grad_norm": 0.97265625, "learning_rate": 7.653713987484393e-05, "loss": 0.1256, "step": 17081 }, { "epoch": 12.514285714285714, "grad_norm": 1.15625, "learning_rate": 7.653422575100407e-05, "loss": 0.134, "step": 17082 }, { "epoch": 12.515018315018315, "grad_norm": 1.2265625, "learning_rate": 7.653131150242569e-05, "loss": 0.1348, "step": 17083 }, { "epoch": 12.515750915750916, "grad_norm": 1.6640625, "learning_rate": 7.652839712912277e-05, "loss": 0.1172, "step": 17084 }, { "epoch": 12.516483516483516, "grad_norm": 0.8828125, "learning_rate": 7.652548263110925e-05, "loss": 0.1146, "step": 17085 }, { "epoch": 12.517216117216117, "grad_norm": 1.8515625, "learning_rate": 7.652256800839909e-05, "loss": 0.1215, "step": 17086 }, { "epoch": 12.517948717948718, "grad_norm": 0.94921875, "learning_rate": 7.651965326100627e-05, "loss": 0.1221, "step": 17087 }, { "epoch": 12.518681318681319, "grad_norm": 1.3046875, "learning_rate": 7.651673838894474e-05, "loss": 0.1113, "step": 17088 }, { "epoch": 12.51941391941392, "grad_norm": 1.0078125, "learning_rate": 7.65138233922285e-05, "loss": 0.1397, "step": 17089 }, { "epoch": 12.52014652014652, "grad_norm": 0.9453125, "learning_rate": 7.651090827087143e-05, "loss": 0.1528, "step": 17090 }, { "epoch": 12.52087912087912, "grad_norm": 0.94140625, "learning_rate": 7.650799302488762e-05, "loss": 0.1244, "step": 17091 }, { "epoch": 12.521611721611722, "grad_norm": 1.9453125, "learning_rate": 7.650507765429097e-05, "loss": 0.1589, "step": 17092 }, { "epoch": 12.522344322344322, "grad_norm": 1.2578125, "learning_rate": 7.650216215909544e-05, "loss": 0.0884, "step": 17093 }, { "epoch": 12.523076923076923, "grad_norm": 1.6328125, "learning_rate": 7.649924653931505e-05, "loss": 0.1149, "step": 17094 }, { "epoch": 12.523809523809524, "grad_norm": 0.9765625, "learning_rate": 7.64963307949637e-05, "loss": 0.116, "step": 17095 }, { "epoch": 12.524542124542124, "grad_norm": 0.87890625, "learning_rate": 7.64934149260554e-05, "loss": 0.1289, "step": 17096 }, { "epoch": 12.525274725274725, "grad_norm": 1.0234375, "learning_rate": 7.649049893260412e-05, "loss": 0.1129, "step": 17097 }, { "epoch": 12.526007326007326, "grad_norm": 1.0625, "learning_rate": 7.648758281462382e-05, "loss": 0.112, "step": 17098 }, { "epoch": 12.526739926739927, "grad_norm": 1.9375, "learning_rate": 7.648466657212849e-05, "loss": 0.133, "step": 17099 }, { "epoch": 12.527472527472527, "grad_norm": 1.1171875, "learning_rate": 7.64817502051321e-05, "loss": 0.15, "step": 17100 }, { "epoch": 12.528205128205128, "grad_norm": 1.125, "learning_rate": 7.64788337136486e-05, "loss": 0.1272, "step": 17101 }, { "epoch": 12.528937728937729, "grad_norm": 0.828125, "learning_rate": 7.647591709769198e-05, "loss": 0.1063, "step": 17102 }, { "epoch": 12.52967032967033, "grad_norm": 1.4765625, "learning_rate": 7.64730003572762e-05, "loss": 0.1281, "step": 17103 }, { "epoch": 12.53040293040293, "grad_norm": 1.171875, "learning_rate": 7.647008349241528e-05, "loss": 0.1314, "step": 17104 }, { "epoch": 12.531135531135531, "grad_norm": 1.1796875, "learning_rate": 7.646716650312315e-05, "loss": 0.1531, "step": 17105 }, { "epoch": 12.531868131868132, "grad_norm": 0.69140625, "learning_rate": 7.646424938941379e-05, "loss": 0.0979, "step": 17106 }, { "epoch": 12.532600732600732, "grad_norm": 1.0546875, "learning_rate": 7.64613321513012e-05, "loss": 0.145, "step": 17107 }, { "epoch": 12.533333333333333, "grad_norm": 0.8515625, "learning_rate": 7.645841478879934e-05, "loss": 0.1119, "step": 17108 }, { "epoch": 12.534065934065934, "grad_norm": 0.88671875, "learning_rate": 7.64554973019222e-05, "loss": 0.137, "step": 17109 }, { "epoch": 12.534798534798535, "grad_norm": 1.375, "learning_rate": 7.645257969068375e-05, "loss": 0.1728, "step": 17110 }, { "epoch": 12.535531135531135, "grad_norm": 0.859375, "learning_rate": 7.644966195509797e-05, "loss": 0.0959, "step": 17111 }, { "epoch": 12.536263736263736, "grad_norm": 1.1171875, "learning_rate": 7.644674409517886e-05, "loss": 0.1571, "step": 17112 }, { "epoch": 12.536996336996337, "grad_norm": 1.0859375, "learning_rate": 7.644382611094034e-05, "loss": 0.1349, "step": 17113 }, { "epoch": 12.537728937728938, "grad_norm": 0.7109375, "learning_rate": 7.644090800239646e-05, "loss": 0.1204, "step": 17114 }, { "epoch": 12.538461538461538, "grad_norm": 1.5625, "learning_rate": 7.643798976956116e-05, "loss": 0.2034, "step": 17115 }, { "epoch": 12.539194139194139, "grad_norm": 0.8515625, "learning_rate": 7.643507141244846e-05, "loss": 0.1304, "step": 17116 }, { "epoch": 12.53992673992674, "grad_norm": 1.015625, "learning_rate": 7.64321529310723e-05, "loss": 0.1513, "step": 17117 }, { "epoch": 12.54065934065934, "grad_norm": 0.875, "learning_rate": 7.642923432544668e-05, "loss": 0.1201, "step": 17118 }, { "epoch": 12.541391941391941, "grad_norm": 1.1640625, "learning_rate": 7.64263155955856e-05, "loss": 0.1281, "step": 17119 }, { "epoch": 12.542124542124542, "grad_norm": 1.1796875, "learning_rate": 7.642339674150302e-05, "loss": 0.1471, "step": 17120 }, { "epoch": 12.542857142857143, "grad_norm": 0.97265625, "learning_rate": 7.642047776321295e-05, "loss": 0.121, "step": 17121 }, { "epoch": 12.543589743589743, "grad_norm": 1.1875, "learning_rate": 7.641755866072936e-05, "loss": 0.1412, "step": 17122 }, { "epoch": 12.544322344322344, "grad_norm": 1.6953125, "learning_rate": 7.641463943406622e-05, "loss": 0.187, "step": 17123 }, { "epoch": 12.545054945054945, "grad_norm": 1.0703125, "learning_rate": 7.641172008323754e-05, "loss": 0.125, "step": 17124 }, { "epoch": 12.545787545787546, "grad_norm": 0.69140625, "learning_rate": 7.640880060825732e-05, "loss": 0.1124, "step": 17125 }, { "epoch": 12.546520146520146, "grad_norm": 1.3359375, "learning_rate": 7.64058810091395e-05, "loss": 0.123, "step": 17126 }, { "epoch": 12.547252747252747, "grad_norm": 1.5234375, "learning_rate": 7.640296128589812e-05, "loss": 0.1177, "step": 17127 }, { "epoch": 12.547985347985348, "grad_norm": 1.6015625, "learning_rate": 7.640004143854716e-05, "loss": 0.1261, "step": 17128 }, { "epoch": 12.548717948717949, "grad_norm": 1.3203125, "learning_rate": 7.639712146710056e-05, "loss": 0.1749, "step": 17129 }, { "epoch": 12.54945054945055, "grad_norm": 1.125, "learning_rate": 7.63942013715724e-05, "loss": 0.1288, "step": 17130 }, { "epoch": 12.55018315018315, "grad_norm": 0.8203125, "learning_rate": 7.639128115197659e-05, "loss": 0.1292, "step": 17131 }, { "epoch": 12.55091575091575, "grad_norm": 1.7109375, "learning_rate": 7.638836080832716e-05, "loss": 0.2023, "step": 17132 }, { "epoch": 12.551648351648351, "grad_norm": 1.125, "learning_rate": 7.638544034063809e-05, "loss": 0.1428, "step": 17133 }, { "epoch": 12.552380952380952, "grad_norm": 0.86328125, "learning_rate": 7.638251974892336e-05, "loss": 0.1275, "step": 17134 }, { "epoch": 12.553113553113553, "grad_norm": 0.62890625, "learning_rate": 7.637959903319702e-05, "loss": 0.0994, "step": 17135 }, { "epoch": 12.553846153846154, "grad_norm": 1.2890625, "learning_rate": 7.637667819347302e-05, "loss": 0.1308, "step": 17136 }, { "epoch": 12.554578754578754, "grad_norm": 1.7578125, "learning_rate": 7.637375722976534e-05, "loss": 0.1433, "step": 17137 }, { "epoch": 12.555311355311355, "grad_norm": 0.98046875, "learning_rate": 7.637083614208801e-05, "loss": 0.1114, "step": 17138 }, { "epoch": 12.556043956043956, "grad_norm": 0.93359375, "learning_rate": 7.6367914930455e-05, "loss": 0.0938, "step": 17139 }, { "epoch": 12.556776556776557, "grad_norm": 1.015625, "learning_rate": 7.636499359488032e-05, "loss": 0.1162, "step": 17140 }, { "epoch": 12.557509157509157, "grad_norm": 1.6171875, "learning_rate": 7.636207213537796e-05, "loss": 0.1112, "step": 17141 }, { "epoch": 12.558241758241758, "grad_norm": 2.46875, "learning_rate": 7.635915055196193e-05, "loss": 0.1209, "step": 17142 }, { "epoch": 12.558974358974359, "grad_norm": 0.95703125, "learning_rate": 7.635622884464622e-05, "loss": 0.1267, "step": 17143 }, { "epoch": 12.55970695970696, "grad_norm": 1.34375, "learning_rate": 7.635330701344482e-05, "loss": 0.1292, "step": 17144 }, { "epoch": 12.56043956043956, "grad_norm": 1.4296875, "learning_rate": 7.635038505837176e-05, "loss": 0.1411, "step": 17145 }, { "epoch": 12.561172161172161, "grad_norm": 1.2890625, "learning_rate": 7.6347462979441e-05, "loss": 0.1248, "step": 17146 }, { "epoch": 12.561904761904762, "grad_norm": 1.2109375, "learning_rate": 7.634454077666657e-05, "loss": 0.1148, "step": 17147 }, { "epoch": 12.562637362637362, "grad_norm": 2.15625, "learning_rate": 7.634161845006246e-05, "loss": 0.1742, "step": 17148 }, { "epoch": 12.563369963369963, "grad_norm": 1.4140625, "learning_rate": 7.633869599964267e-05, "loss": 0.1563, "step": 17149 }, { "epoch": 12.564102564102564, "grad_norm": 1.5859375, "learning_rate": 7.633577342542122e-05, "loss": 0.1205, "step": 17150 }, { "epoch": 12.564835164835165, "grad_norm": 1.609375, "learning_rate": 7.63328507274121e-05, "loss": 0.13, "step": 17151 }, { "epoch": 12.565567765567765, "grad_norm": 0.95703125, "learning_rate": 7.632992790562927e-05, "loss": 0.1056, "step": 17152 }, { "epoch": 12.566300366300366, "grad_norm": 1.3203125, "learning_rate": 7.632700496008681e-05, "loss": 0.1477, "step": 17153 }, { "epoch": 12.567032967032967, "grad_norm": 0.84765625, "learning_rate": 7.632408189079869e-05, "loss": 0.105, "step": 17154 }, { "epoch": 12.567765567765568, "grad_norm": 0.82421875, "learning_rate": 7.63211586977789e-05, "loss": 0.1342, "step": 17155 }, { "epoch": 12.568498168498168, "grad_norm": 1.0, "learning_rate": 7.631823538104146e-05, "loss": 0.1343, "step": 17156 }, { "epoch": 12.569230769230769, "grad_norm": 0.84375, "learning_rate": 7.631531194060042e-05, "loss": 0.1115, "step": 17157 }, { "epoch": 12.56996336996337, "grad_norm": 1.234375, "learning_rate": 7.63123883764697e-05, "loss": 0.1256, "step": 17158 }, { "epoch": 12.57069597069597, "grad_norm": 0.859375, "learning_rate": 7.630946468866337e-05, "loss": 0.1238, "step": 17159 }, { "epoch": 12.571428571428571, "grad_norm": 0.97265625, "learning_rate": 7.630654087719542e-05, "loss": 0.1159, "step": 17160 }, { "epoch": 12.572161172161172, "grad_norm": 0.75, "learning_rate": 7.630361694207985e-05, "loss": 0.1115, "step": 17161 }, { "epoch": 12.572893772893773, "grad_norm": 1.0625, "learning_rate": 7.63006928833307e-05, "loss": 0.1238, "step": 17162 }, { "epoch": 12.573626373626373, "grad_norm": 1.125, "learning_rate": 7.629776870096197e-05, "loss": 0.1349, "step": 17163 }, { "epoch": 12.574358974358974, "grad_norm": 1.28125, "learning_rate": 7.629484439498765e-05, "loss": 0.125, "step": 17164 }, { "epoch": 12.575091575091575, "grad_norm": 1.1171875, "learning_rate": 7.629191996542175e-05, "loss": 0.1094, "step": 17165 }, { "epoch": 12.575824175824176, "grad_norm": 0.74609375, "learning_rate": 7.628899541227829e-05, "loss": 0.1207, "step": 17166 }, { "epoch": 12.576556776556776, "grad_norm": 1.515625, "learning_rate": 7.62860707355713e-05, "loss": 0.1589, "step": 17167 }, { "epoch": 12.577289377289377, "grad_norm": 1.1484375, "learning_rate": 7.628314593531476e-05, "loss": 0.1454, "step": 17168 }, { "epoch": 12.578021978021978, "grad_norm": 0.69921875, "learning_rate": 7.628022101152274e-05, "loss": 0.1078, "step": 17169 }, { "epoch": 12.578754578754578, "grad_norm": 0.8828125, "learning_rate": 7.627729596420921e-05, "loss": 0.1211, "step": 17170 }, { "epoch": 12.57948717948718, "grad_norm": 0.9140625, "learning_rate": 7.627437079338819e-05, "loss": 0.0897, "step": 17171 }, { "epoch": 12.58021978021978, "grad_norm": 1.7265625, "learning_rate": 7.627144549907368e-05, "loss": 0.1409, "step": 17172 }, { "epoch": 12.58095238095238, "grad_norm": 1.109375, "learning_rate": 7.626852008127973e-05, "loss": 0.1236, "step": 17173 }, { "epoch": 12.581684981684981, "grad_norm": 0.73828125, "learning_rate": 7.626559454002034e-05, "loss": 0.1253, "step": 17174 }, { "epoch": 12.582417582417582, "grad_norm": 0.99609375, "learning_rate": 7.626266887530953e-05, "loss": 0.1125, "step": 17175 }, { "epoch": 12.583150183150183, "grad_norm": 1.03125, "learning_rate": 7.62597430871613e-05, "loss": 0.1228, "step": 17176 }, { "epoch": 12.583882783882784, "grad_norm": 1.3671875, "learning_rate": 7.62568171755897e-05, "loss": 0.1463, "step": 17177 }, { "epoch": 12.584615384615384, "grad_norm": 1.1328125, "learning_rate": 7.625389114060872e-05, "loss": 0.1258, "step": 17178 }, { "epoch": 12.585347985347985, "grad_norm": 1.359375, "learning_rate": 7.62509649822324e-05, "loss": 0.1365, "step": 17179 }, { "epoch": 12.586080586080586, "grad_norm": 1.3984375, "learning_rate": 7.624803870047477e-05, "loss": 0.138, "step": 17180 }, { "epoch": 12.586813186813187, "grad_norm": 1.3671875, "learning_rate": 7.624511229534982e-05, "loss": 0.1317, "step": 17181 }, { "epoch": 12.587545787545787, "grad_norm": 0.8515625, "learning_rate": 7.624218576687159e-05, "loss": 0.1225, "step": 17182 }, { "epoch": 12.588278388278388, "grad_norm": 0.83984375, "learning_rate": 7.623925911505409e-05, "loss": 0.1309, "step": 17183 }, { "epoch": 12.589010989010989, "grad_norm": 1.3046875, "learning_rate": 7.623633233991136e-05, "loss": 0.1327, "step": 17184 }, { "epoch": 12.58974358974359, "grad_norm": 0.75390625, "learning_rate": 7.62334054414574e-05, "loss": 0.1236, "step": 17185 }, { "epoch": 12.59047619047619, "grad_norm": 1.5078125, "learning_rate": 7.623047841970627e-05, "loss": 0.1387, "step": 17186 }, { "epoch": 12.59120879120879, "grad_norm": 1.15625, "learning_rate": 7.622755127467194e-05, "loss": 0.1412, "step": 17187 }, { "epoch": 12.591941391941392, "grad_norm": 0.87890625, "learning_rate": 7.622462400636849e-05, "loss": 0.1303, "step": 17188 }, { "epoch": 12.592673992673992, "grad_norm": 1.5859375, "learning_rate": 7.62216966148099e-05, "loss": 0.1231, "step": 17189 }, { "epoch": 12.593406593406593, "grad_norm": 1.4921875, "learning_rate": 7.621876910001024e-05, "loss": 0.1319, "step": 17190 }, { "epoch": 12.594139194139194, "grad_norm": 0.75390625, "learning_rate": 7.62158414619835e-05, "loss": 0.1305, "step": 17191 }, { "epoch": 12.594871794871795, "grad_norm": 1.0546875, "learning_rate": 7.621291370074372e-05, "loss": 0.1378, "step": 17192 }, { "epoch": 12.595604395604395, "grad_norm": 0.95703125, "learning_rate": 7.620998581630493e-05, "loss": 0.1317, "step": 17193 }, { "epoch": 12.596336996336996, "grad_norm": 1.1171875, "learning_rate": 7.620705780868117e-05, "loss": 0.1359, "step": 17194 }, { "epoch": 12.597069597069597, "grad_norm": 0.75390625, "learning_rate": 7.620412967788645e-05, "loss": 0.1146, "step": 17195 }, { "epoch": 12.597802197802197, "grad_norm": 0.9375, "learning_rate": 7.620120142393481e-05, "loss": 0.1317, "step": 17196 }, { "epoch": 12.598534798534798, "grad_norm": 1.1953125, "learning_rate": 7.619827304684027e-05, "loss": 0.1464, "step": 17197 }, { "epoch": 12.599267399267399, "grad_norm": 1.0390625, "learning_rate": 7.619534454661688e-05, "loss": 0.1316, "step": 17198 }, { "epoch": 12.6, "grad_norm": 0.65625, "learning_rate": 7.619241592327864e-05, "loss": 0.1236, "step": 17199 }, { "epoch": 12.6007326007326, "grad_norm": 0.890625, "learning_rate": 7.61894871768396e-05, "loss": 0.1369, "step": 17200 }, { "epoch": 12.601465201465201, "grad_norm": 1.078125, "learning_rate": 7.618655830731381e-05, "loss": 0.1861, "step": 17201 }, { "epoch": 12.602197802197802, "grad_norm": 1.265625, "learning_rate": 7.618362931471527e-05, "loss": 0.1429, "step": 17202 }, { "epoch": 12.602930402930403, "grad_norm": 0.68359375, "learning_rate": 7.618070019905807e-05, "loss": 0.1343, "step": 17203 }, { "epoch": 12.603663003663003, "grad_norm": 1.640625, "learning_rate": 7.617777096035617e-05, "loss": 0.1246, "step": 17204 }, { "epoch": 12.604395604395604, "grad_norm": 1.0078125, "learning_rate": 7.617484159862364e-05, "loss": 0.1058, "step": 17205 }, { "epoch": 12.605128205128205, "grad_norm": 1.0625, "learning_rate": 7.617191211387453e-05, "loss": 0.166, "step": 17206 }, { "epoch": 12.605860805860805, "grad_norm": 0.80078125, "learning_rate": 7.616898250612284e-05, "loss": 0.0929, "step": 17207 }, { "epoch": 12.606593406593406, "grad_norm": 0.734375, "learning_rate": 7.616605277538264e-05, "loss": 0.105, "step": 17208 }, { "epoch": 12.607326007326007, "grad_norm": 1.328125, "learning_rate": 7.616312292166797e-05, "loss": 0.1474, "step": 17209 }, { "epoch": 12.608058608058608, "grad_norm": 1.3828125, "learning_rate": 7.616019294499283e-05, "loss": 0.1371, "step": 17210 }, { "epoch": 12.608791208791208, "grad_norm": 0.93359375, "learning_rate": 7.615726284537129e-05, "loss": 0.1292, "step": 17211 }, { "epoch": 12.60952380952381, "grad_norm": 0.7890625, "learning_rate": 7.615433262281741e-05, "loss": 0.1602, "step": 17212 }, { "epoch": 12.61025641025641, "grad_norm": 1.25, "learning_rate": 7.615140227734517e-05, "loss": 0.1419, "step": 17213 }, { "epoch": 12.61098901098901, "grad_norm": 1.3046875, "learning_rate": 7.614847180896865e-05, "loss": 0.1665, "step": 17214 }, { "epoch": 12.611721611721611, "grad_norm": 1.3515625, "learning_rate": 7.614554121770186e-05, "loss": 0.1435, "step": 17215 }, { "epoch": 12.612454212454212, "grad_norm": 1.2890625, "learning_rate": 7.614261050355887e-05, "loss": 0.1345, "step": 17216 }, { "epoch": 12.613186813186813, "grad_norm": 0.8671875, "learning_rate": 7.613967966655373e-05, "loss": 0.1127, "step": 17217 }, { "epoch": 12.613919413919414, "grad_norm": 0.984375, "learning_rate": 7.613674870670047e-05, "loss": 0.1215, "step": 17218 }, { "epoch": 12.614652014652014, "grad_norm": 0.8125, "learning_rate": 7.613381762401313e-05, "loss": 0.115, "step": 17219 }, { "epoch": 12.615384615384615, "grad_norm": 1.2265625, "learning_rate": 7.613088641850573e-05, "loss": 0.1244, "step": 17220 }, { "epoch": 12.616117216117216, "grad_norm": 1.6953125, "learning_rate": 7.612795509019236e-05, "loss": 0.1694, "step": 17221 }, { "epoch": 12.616849816849816, "grad_norm": 1.3828125, "learning_rate": 7.612502363908706e-05, "loss": 0.1411, "step": 17222 }, { "epoch": 12.617582417582417, "grad_norm": 1.375, "learning_rate": 7.612209206520384e-05, "loss": 0.1263, "step": 17223 }, { "epoch": 12.618315018315018, "grad_norm": 1.1328125, "learning_rate": 7.611916036855677e-05, "loss": 0.1156, "step": 17224 }, { "epoch": 12.619047619047619, "grad_norm": 0.828125, "learning_rate": 7.611622854915989e-05, "loss": 0.1294, "step": 17225 }, { "epoch": 12.61978021978022, "grad_norm": 1.2109375, "learning_rate": 7.611329660702724e-05, "loss": 0.1201, "step": 17226 }, { "epoch": 12.62051282051282, "grad_norm": 0.8359375, "learning_rate": 7.61103645421729e-05, "loss": 0.1171, "step": 17227 }, { "epoch": 12.62124542124542, "grad_norm": 1.0, "learning_rate": 7.610743235461088e-05, "loss": 0.113, "step": 17228 }, { "epoch": 12.621978021978022, "grad_norm": 1.1171875, "learning_rate": 7.610450004435525e-05, "loss": 0.1167, "step": 17229 }, { "epoch": 12.622710622710622, "grad_norm": 0.90625, "learning_rate": 7.610156761142005e-05, "loss": 0.127, "step": 17230 }, { "epoch": 12.623443223443223, "grad_norm": 1.390625, "learning_rate": 7.609863505581932e-05, "loss": 0.1428, "step": 17231 }, { "epoch": 12.624175824175824, "grad_norm": 1.25, "learning_rate": 7.609570237756714e-05, "loss": 0.1371, "step": 17232 }, { "epoch": 12.624908424908424, "grad_norm": 1.765625, "learning_rate": 7.609276957667756e-05, "loss": 0.1356, "step": 17233 }, { "epoch": 12.625641025641025, "grad_norm": 0.9765625, "learning_rate": 7.608983665316462e-05, "loss": 0.1045, "step": 17234 }, { "epoch": 12.626373626373626, "grad_norm": 0.82421875, "learning_rate": 7.608690360704236e-05, "loss": 0.126, "step": 17235 }, { "epoch": 12.627106227106227, "grad_norm": 1.5078125, "learning_rate": 7.608397043832484e-05, "loss": 0.1502, "step": 17236 }, { "epoch": 12.627838827838827, "grad_norm": 0.890625, "learning_rate": 7.608103714702613e-05, "loss": 0.1282, "step": 17237 }, { "epoch": 12.628571428571428, "grad_norm": 0.99609375, "learning_rate": 7.607810373316027e-05, "loss": 0.1389, "step": 17238 }, { "epoch": 12.629304029304029, "grad_norm": 0.765625, "learning_rate": 7.607517019674132e-05, "loss": 0.1367, "step": 17239 }, { "epoch": 12.63003663003663, "grad_norm": 1.75, "learning_rate": 7.607223653778334e-05, "loss": 0.1438, "step": 17240 }, { "epoch": 12.63076923076923, "grad_norm": 1.453125, "learning_rate": 7.606930275630037e-05, "loss": 0.1306, "step": 17241 }, { "epoch": 12.631501831501831, "grad_norm": 1.2421875, "learning_rate": 7.606636885230649e-05, "loss": 0.1178, "step": 17242 }, { "epoch": 12.632234432234432, "grad_norm": 1.2734375, "learning_rate": 7.606343482581573e-05, "loss": 0.1573, "step": 17243 }, { "epoch": 12.632967032967032, "grad_norm": 1.34375, "learning_rate": 7.606050067684215e-05, "loss": 0.1196, "step": 17244 }, { "epoch": 12.633699633699633, "grad_norm": 1.3359375, "learning_rate": 7.605756640539984e-05, "loss": 0.0951, "step": 17245 }, { "epoch": 12.634432234432234, "grad_norm": 1.1796875, "learning_rate": 7.605463201150283e-05, "loss": 0.1236, "step": 17246 }, { "epoch": 12.635164835164835, "grad_norm": 0.79296875, "learning_rate": 7.60516974951652e-05, "loss": 0.1076, "step": 17247 }, { "epoch": 12.635897435897435, "grad_norm": 1.15625, "learning_rate": 7.6048762856401e-05, "loss": 0.1339, "step": 17248 }, { "epoch": 12.636630036630036, "grad_norm": 1.0078125, "learning_rate": 7.604582809522428e-05, "loss": 0.1446, "step": 17249 }, { "epoch": 12.637362637362637, "grad_norm": 0.8515625, "learning_rate": 7.604289321164911e-05, "loss": 0.1186, "step": 17250 }, { "epoch": 12.638095238095238, "grad_norm": 2.03125, "learning_rate": 7.603995820568958e-05, "loss": 0.1628, "step": 17251 }, { "epoch": 12.638827838827838, "grad_norm": 1.0, "learning_rate": 7.603702307735968e-05, "loss": 0.1005, "step": 17252 }, { "epoch": 12.639560439560439, "grad_norm": 1.1875, "learning_rate": 7.603408782667356e-05, "loss": 0.1512, "step": 17253 }, { "epoch": 12.64029304029304, "grad_norm": 1.09375, "learning_rate": 7.60311524536452e-05, "loss": 0.1422, "step": 17254 }, { "epoch": 12.64102564102564, "grad_norm": 0.69140625, "learning_rate": 7.602821695828875e-05, "loss": 0.1318, "step": 17255 }, { "epoch": 12.641758241758241, "grad_norm": 1.21875, "learning_rate": 7.602528134061822e-05, "loss": 0.1161, "step": 17256 }, { "epoch": 12.642490842490842, "grad_norm": 0.7578125, "learning_rate": 7.602234560064765e-05, "loss": 0.1132, "step": 17257 }, { "epoch": 12.643223443223443, "grad_norm": 0.5546875, "learning_rate": 7.601940973839118e-05, "loss": 0.1012, "step": 17258 }, { "epoch": 12.643956043956043, "grad_norm": 1.0703125, "learning_rate": 7.601647375386287e-05, "loss": 0.131, "step": 17259 }, { "epoch": 12.644688644688644, "grad_norm": 1.0078125, "learning_rate": 7.601353764707671e-05, "loss": 0.1323, "step": 17260 }, { "epoch": 12.645421245421245, "grad_norm": 0.97265625, "learning_rate": 7.601060141804683e-05, "loss": 0.1377, "step": 17261 }, { "epoch": 12.646153846153846, "grad_norm": 0.97265625, "learning_rate": 7.600766506678728e-05, "loss": 0.1227, "step": 17262 }, { "epoch": 12.646886446886446, "grad_norm": 0.96875, "learning_rate": 7.600472859331215e-05, "loss": 0.1351, "step": 17263 }, { "epoch": 12.647619047619047, "grad_norm": 0.98828125, "learning_rate": 7.600179199763549e-05, "loss": 0.131, "step": 17264 }, { "epoch": 12.648351648351648, "grad_norm": 0.9921875, "learning_rate": 7.599885527977135e-05, "loss": 0.124, "step": 17265 }, { "epoch": 12.649084249084249, "grad_norm": 1.0, "learning_rate": 7.599591843973385e-05, "loss": 0.1232, "step": 17266 }, { "epoch": 12.64981684981685, "grad_norm": 0.76953125, "learning_rate": 7.599298147753702e-05, "loss": 0.114, "step": 17267 }, { "epoch": 12.65054945054945, "grad_norm": 1.2421875, "learning_rate": 7.599004439319495e-05, "loss": 0.1304, "step": 17268 }, { "epoch": 12.65128205128205, "grad_norm": 0.91796875, "learning_rate": 7.59871071867217e-05, "loss": 0.1189, "step": 17269 }, { "epoch": 12.652014652014651, "grad_norm": 0.78515625, "learning_rate": 7.598416985813135e-05, "loss": 0.1305, "step": 17270 }, { "epoch": 12.652747252747252, "grad_norm": 1.15625, "learning_rate": 7.5981232407438e-05, "loss": 0.1544, "step": 17271 }, { "epoch": 12.653479853479853, "grad_norm": 0.93359375, "learning_rate": 7.597829483465568e-05, "loss": 0.127, "step": 17272 }, { "epoch": 12.654212454212454, "grad_norm": 0.703125, "learning_rate": 7.597535713979849e-05, "loss": 0.1307, "step": 17273 }, { "epoch": 12.654945054945054, "grad_norm": 1.328125, "learning_rate": 7.597241932288052e-05, "loss": 0.1349, "step": 17274 }, { "epoch": 12.655677655677655, "grad_norm": 0.9375, "learning_rate": 7.596948138391579e-05, "loss": 0.1537, "step": 17275 }, { "epoch": 12.656410256410256, "grad_norm": 1.46875, "learning_rate": 7.596654332291845e-05, "loss": 0.1114, "step": 17276 }, { "epoch": 12.657142857142857, "grad_norm": 0.734375, "learning_rate": 7.596360513990253e-05, "loss": 0.1287, "step": 17277 }, { "epoch": 12.657875457875457, "grad_norm": 1.0078125, "learning_rate": 7.596066683488211e-05, "loss": 0.1174, "step": 17278 }, { "epoch": 12.658608058608058, "grad_norm": 1.203125, "learning_rate": 7.595772840787128e-05, "loss": 0.1516, "step": 17279 }, { "epoch": 12.659340659340659, "grad_norm": 1.859375, "learning_rate": 7.595478985888411e-05, "loss": 0.1529, "step": 17280 }, { "epoch": 12.66007326007326, "grad_norm": 1.4296875, "learning_rate": 7.59518511879347e-05, "loss": 0.1222, "step": 17281 }, { "epoch": 12.66080586080586, "grad_norm": 1.25, "learning_rate": 7.594891239503711e-05, "loss": 0.1154, "step": 17282 }, { "epoch": 12.661538461538461, "grad_norm": 1.015625, "learning_rate": 7.594597348020541e-05, "loss": 0.131, "step": 17283 }, { "epoch": 12.662271062271062, "grad_norm": 0.73046875, "learning_rate": 7.594303444345373e-05, "loss": 0.1061, "step": 17284 }, { "epoch": 12.663003663003662, "grad_norm": 1.1171875, "learning_rate": 7.59400952847961e-05, "loss": 0.1023, "step": 17285 }, { "epoch": 12.663736263736263, "grad_norm": 1.015625, "learning_rate": 7.593715600424662e-05, "loss": 0.1189, "step": 17286 }, { "epoch": 12.664468864468864, "grad_norm": 1.265625, "learning_rate": 7.593421660181939e-05, "loss": 0.1287, "step": 17287 }, { "epoch": 12.665201465201465, "grad_norm": 0.97265625, "learning_rate": 7.593127707752846e-05, "loss": 0.1315, "step": 17288 }, { "epoch": 12.665934065934065, "grad_norm": 1.6796875, "learning_rate": 7.592833743138795e-05, "loss": 0.1262, "step": 17289 }, { "epoch": 12.666666666666666, "grad_norm": 0.75390625, "learning_rate": 7.592539766341191e-05, "loss": 0.1138, "step": 17290 }, { "epoch": 12.667399267399267, "grad_norm": 1.109375, "learning_rate": 7.592245777361445e-05, "loss": 0.149, "step": 17291 }, { "epoch": 12.668131868131868, "grad_norm": 0.95703125, "learning_rate": 7.591951776200965e-05, "loss": 0.1496, "step": 17292 }, { "epoch": 12.668864468864468, "grad_norm": 0.796875, "learning_rate": 7.591657762861159e-05, "loss": 0.1239, "step": 17293 }, { "epoch": 12.669597069597069, "grad_norm": 1.09375, "learning_rate": 7.591363737343438e-05, "loss": 0.1522, "step": 17294 }, { "epoch": 12.67032967032967, "grad_norm": 0.859375, "learning_rate": 7.591069699649208e-05, "loss": 0.116, "step": 17295 }, { "epoch": 12.67106227106227, "grad_norm": 0.94140625, "learning_rate": 7.590775649779879e-05, "loss": 0.1186, "step": 17296 }, { "epoch": 12.671794871794871, "grad_norm": 1.296875, "learning_rate": 7.590481587736858e-05, "loss": 0.1042, "step": 17297 }, { "epoch": 12.672527472527472, "grad_norm": 1.421875, "learning_rate": 7.590187513521559e-05, "loss": 0.158, "step": 17298 }, { "epoch": 12.673260073260073, "grad_norm": 0.64453125, "learning_rate": 7.589893427135385e-05, "loss": 0.114, "step": 17299 }, { "epoch": 12.673992673992673, "grad_norm": 1.1796875, "learning_rate": 7.589599328579749e-05, "loss": 0.1303, "step": 17300 }, { "epoch": 12.674725274725274, "grad_norm": 0.94921875, "learning_rate": 7.58930521785606e-05, "loss": 0.0952, "step": 17301 }, { "epoch": 12.675457875457875, "grad_norm": 0.72265625, "learning_rate": 7.589011094965725e-05, "loss": 0.0943, "step": 17302 }, { "epoch": 12.676190476190476, "grad_norm": 1.0, "learning_rate": 7.588716959910155e-05, "loss": 0.1098, "step": 17303 }, { "epoch": 12.676923076923076, "grad_norm": 0.9921875, "learning_rate": 7.588422812690758e-05, "loss": 0.1728, "step": 17304 }, { "epoch": 12.677655677655677, "grad_norm": 1.0234375, "learning_rate": 7.588128653308944e-05, "loss": 0.1561, "step": 17305 }, { "epoch": 12.678388278388278, "grad_norm": 0.83203125, "learning_rate": 7.587834481766121e-05, "loss": 0.1184, "step": 17306 }, { "epoch": 12.679120879120878, "grad_norm": 1.1484375, "learning_rate": 7.5875402980637e-05, "loss": 0.1076, "step": 17307 }, { "epoch": 12.67985347985348, "grad_norm": 1.125, "learning_rate": 7.587246102203094e-05, "loss": 0.1303, "step": 17308 }, { "epoch": 12.68058608058608, "grad_norm": 0.81640625, "learning_rate": 7.586951894185707e-05, "loss": 0.1207, "step": 17309 }, { "epoch": 12.68131868131868, "grad_norm": 1.2265625, "learning_rate": 7.586657674012949e-05, "loss": 0.1693, "step": 17310 }, { "epoch": 12.682051282051281, "grad_norm": 1.3203125, "learning_rate": 7.586363441686232e-05, "loss": 0.1149, "step": 17311 }, { "epoch": 12.682783882783882, "grad_norm": 2.046875, "learning_rate": 7.586069197206966e-05, "loss": 0.1592, "step": 17312 }, { "epoch": 12.683516483516483, "grad_norm": 0.6640625, "learning_rate": 7.58577494057656e-05, "loss": 0.1458, "step": 17313 }, { "epoch": 12.684249084249084, "grad_norm": 1.546875, "learning_rate": 7.585480671796424e-05, "loss": 0.1408, "step": 17314 }, { "epoch": 12.684981684981684, "grad_norm": 1.6875, "learning_rate": 7.585186390867968e-05, "loss": 0.1277, "step": 17315 }, { "epoch": 12.685714285714285, "grad_norm": 1.46875, "learning_rate": 7.584892097792601e-05, "loss": 0.1494, "step": 17316 }, { "epoch": 12.686446886446886, "grad_norm": 1.0546875, "learning_rate": 7.584597792571734e-05, "loss": 0.13, "step": 17317 }, { "epoch": 12.687179487179487, "grad_norm": 1.046875, "learning_rate": 7.584303475206776e-05, "loss": 0.121, "step": 17318 }, { "epoch": 12.687912087912087, "grad_norm": 1.171875, "learning_rate": 7.584009145699139e-05, "loss": 0.1079, "step": 17319 }, { "epoch": 12.688644688644688, "grad_norm": 0.921875, "learning_rate": 7.583714804050231e-05, "loss": 0.1249, "step": 17320 }, { "epoch": 12.689377289377289, "grad_norm": 1.2109375, "learning_rate": 7.583420450261466e-05, "loss": 0.1361, "step": 17321 }, { "epoch": 12.69010989010989, "grad_norm": 0.890625, "learning_rate": 7.58312608433425e-05, "loss": 0.1596, "step": 17322 }, { "epoch": 12.69084249084249, "grad_norm": 1.0703125, "learning_rate": 7.582831706269995e-05, "loss": 0.1171, "step": 17323 }, { "epoch": 12.691575091575091, "grad_norm": 1.15625, "learning_rate": 7.582537316070115e-05, "loss": 0.125, "step": 17324 }, { "epoch": 12.692307692307692, "grad_norm": 1.28125, "learning_rate": 7.582242913736013e-05, "loss": 0.1362, "step": 17325 }, { "epoch": 12.693040293040292, "grad_norm": 1.0859375, "learning_rate": 7.58194849926911e-05, "loss": 0.1327, "step": 17326 }, { "epoch": 12.693772893772893, "grad_norm": 1.125, "learning_rate": 7.581654072670805e-05, "loss": 0.1657, "step": 17327 }, { "epoch": 12.694505494505494, "grad_norm": 0.8828125, "learning_rate": 7.581359633942515e-05, "loss": 0.1347, "step": 17328 }, { "epoch": 12.695238095238095, "grad_norm": 1.8046875, "learning_rate": 7.581065183085652e-05, "loss": 0.1502, "step": 17329 }, { "epoch": 12.695970695970695, "grad_norm": 0.828125, "learning_rate": 7.580770720101623e-05, "loss": 0.1285, "step": 17330 }, { "epoch": 12.696703296703296, "grad_norm": 1.3125, "learning_rate": 7.580476244991844e-05, "loss": 0.1393, "step": 17331 }, { "epoch": 12.697435897435897, "grad_norm": 0.96484375, "learning_rate": 7.580181757757719e-05, "loss": 0.1237, "step": 17332 }, { "epoch": 12.698168498168497, "grad_norm": 1.15625, "learning_rate": 7.579887258400662e-05, "loss": 0.1096, "step": 17333 }, { "epoch": 12.698901098901098, "grad_norm": 1.6796875, "learning_rate": 7.579592746922088e-05, "loss": 0.1094, "step": 17334 }, { "epoch": 12.699633699633699, "grad_norm": 1.1328125, "learning_rate": 7.579298223323402e-05, "loss": 0.1215, "step": 17335 }, { "epoch": 12.7003663003663, "grad_norm": 1.1875, "learning_rate": 7.57900368760602e-05, "loss": 0.1358, "step": 17336 }, { "epoch": 12.7010989010989, "grad_norm": 0.93359375, "learning_rate": 7.578709139771349e-05, "loss": 0.1212, "step": 17337 }, { "epoch": 12.701831501831501, "grad_norm": 1.2421875, "learning_rate": 7.578414579820804e-05, "loss": 0.117, "step": 17338 }, { "epoch": 12.702564102564102, "grad_norm": 0.97265625, "learning_rate": 7.578120007755794e-05, "loss": 0.1335, "step": 17339 }, { "epoch": 12.703296703296703, "grad_norm": 1.0, "learning_rate": 7.57782542357773e-05, "loss": 0.1291, "step": 17340 }, { "epoch": 12.704029304029303, "grad_norm": 1.1015625, "learning_rate": 7.577530827288027e-05, "loss": 0.1877, "step": 17341 }, { "epoch": 12.704761904761904, "grad_norm": 1.0859375, "learning_rate": 7.577236218888094e-05, "loss": 0.1199, "step": 17342 }, { "epoch": 12.705494505494505, "grad_norm": 1.1484375, "learning_rate": 7.576941598379342e-05, "loss": 0.1336, "step": 17343 }, { "epoch": 12.706227106227105, "grad_norm": 0.83203125, "learning_rate": 7.576646965763183e-05, "loss": 0.1144, "step": 17344 }, { "epoch": 12.706959706959706, "grad_norm": 1.4609375, "learning_rate": 7.576352321041029e-05, "loss": 0.1327, "step": 17345 }, { "epoch": 12.707692307692307, "grad_norm": 1.015625, "learning_rate": 7.57605766421429e-05, "loss": 0.1304, "step": 17346 }, { "epoch": 12.708424908424908, "grad_norm": 0.94140625, "learning_rate": 7.57576299528438e-05, "loss": 0.1274, "step": 17347 }, { "epoch": 12.709157509157508, "grad_norm": 1.0859375, "learning_rate": 7.575468314252712e-05, "loss": 0.1254, "step": 17348 }, { "epoch": 12.70989010989011, "grad_norm": 1.2734375, "learning_rate": 7.575173621120697e-05, "loss": 0.1507, "step": 17349 }, { "epoch": 12.71062271062271, "grad_norm": 0.89453125, "learning_rate": 7.574878915889743e-05, "loss": 0.1318, "step": 17350 }, { "epoch": 12.71135531135531, "grad_norm": 0.76171875, "learning_rate": 7.574584198561268e-05, "loss": 0.1299, "step": 17351 }, { "epoch": 12.712087912087911, "grad_norm": 1.2109375, "learning_rate": 7.57428946913668e-05, "loss": 0.1275, "step": 17352 }, { "epoch": 12.712820512820512, "grad_norm": 1.5, "learning_rate": 7.573994727617393e-05, "loss": 0.0928, "step": 17353 }, { "epoch": 12.713553113553113, "grad_norm": 1.296875, "learning_rate": 7.573699974004819e-05, "loss": 0.1265, "step": 17354 }, { "epoch": 12.714285714285714, "grad_norm": 2.125, "learning_rate": 7.57340520830037e-05, "loss": 0.1311, "step": 17355 }, { "epoch": 12.715018315018314, "grad_norm": 1.0859375, "learning_rate": 7.573110430505457e-05, "loss": 0.1275, "step": 17356 }, { "epoch": 12.715750915750915, "grad_norm": 0.921875, "learning_rate": 7.572815640621494e-05, "loss": 0.1317, "step": 17357 }, { "epoch": 12.716483516483516, "grad_norm": 1.359375, "learning_rate": 7.572520838649893e-05, "loss": 0.1341, "step": 17358 }, { "epoch": 12.717216117216116, "grad_norm": 0.96484375, "learning_rate": 7.572226024592067e-05, "loss": 0.1468, "step": 17359 }, { "epoch": 12.717948717948717, "grad_norm": 1.6015625, "learning_rate": 7.57193119844943e-05, "loss": 0.1478, "step": 17360 }, { "epoch": 12.718681318681318, "grad_norm": 1.03125, "learning_rate": 7.57163636022339e-05, "loss": 0.1319, "step": 17361 }, { "epoch": 12.719413919413919, "grad_norm": 1.0625, "learning_rate": 7.571341509915363e-05, "loss": 0.1035, "step": 17362 }, { "epoch": 12.72014652014652, "grad_norm": 1.4296875, "learning_rate": 7.571046647526761e-05, "loss": 0.1397, "step": 17363 }, { "epoch": 12.72087912087912, "grad_norm": 1.0625, "learning_rate": 7.570751773058998e-05, "loss": 0.121, "step": 17364 }, { "epoch": 12.72161172161172, "grad_norm": 1.4375, "learning_rate": 7.570456886513487e-05, "loss": 0.1025, "step": 17365 }, { "epoch": 12.722344322344322, "grad_norm": 0.99609375, "learning_rate": 7.570161987891636e-05, "loss": 0.128, "step": 17366 }, { "epoch": 12.723076923076922, "grad_norm": 0.83984375, "learning_rate": 7.569867077194864e-05, "loss": 0.1143, "step": 17367 }, { "epoch": 12.723809523809523, "grad_norm": 1.03125, "learning_rate": 7.56957215442458e-05, "loss": 0.1333, "step": 17368 }, { "epoch": 12.724542124542124, "grad_norm": 1.1015625, "learning_rate": 7.5692772195822e-05, "loss": 0.1079, "step": 17369 }, { "epoch": 12.725274725274724, "grad_norm": 1.390625, "learning_rate": 7.568982272669135e-05, "loss": 0.1452, "step": 17370 }, { "epoch": 12.726007326007325, "grad_norm": 1.828125, "learning_rate": 7.568687313686799e-05, "loss": 0.163, "step": 17371 }, { "epoch": 12.726739926739926, "grad_norm": 1.765625, "learning_rate": 7.568392342636607e-05, "loss": 0.156, "step": 17372 }, { "epoch": 12.727472527472527, "grad_norm": 1.09375, "learning_rate": 7.56809735951997e-05, "loss": 0.1262, "step": 17373 }, { "epoch": 12.728205128205127, "grad_norm": 1.96875, "learning_rate": 7.567802364338299e-05, "loss": 0.1309, "step": 17374 }, { "epoch": 12.728937728937728, "grad_norm": 2.53125, "learning_rate": 7.567507357093013e-05, "loss": 0.1574, "step": 17375 }, { "epoch": 12.729670329670329, "grad_norm": 1.359375, "learning_rate": 7.56721233778552e-05, "loss": 0.1503, "step": 17376 }, { "epoch": 12.73040293040293, "grad_norm": 1.3984375, "learning_rate": 7.566917306417242e-05, "loss": 0.151, "step": 17377 }, { "epoch": 12.73113553113553, "grad_norm": 0.98828125, "learning_rate": 7.566622262989583e-05, "loss": 0.1069, "step": 17378 }, { "epoch": 12.731868131868131, "grad_norm": 0.70703125, "learning_rate": 7.56632720750396e-05, "loss": 0.1244, "step": 17379 }, { "epoch": 12.732600732600732, "grad_norm": 0.6953125, "learning_rate": 7.566032139961789e-05, "loss": 0.1273, "step": 17380 }, { "epoch": 12.733333333333333, "grad_norm": 1.375, "learning_rate": 7.565737060364483e-05, "loss": 0.136, "step": 17381 }, { "epoch": 12.734065934065933, "grad_norm": 0.58203125, "learning_rate": 7.565441968713451e-05, "loss": 0.1022, "step": 17382 }, { "epoch": 12.734798534798536, "grad_norm": 1.0859375, "learning_rate": 7.565146865010116e-05, "loss": 0.1252, "step": 17383 }, { "epoch": 12.735531135531136, "grad_norm": 0.83984375, "learning_rate": 7.564851749255882e-05, "loss": 0.1012, "step": 17384 }, { "epoch": 12.736263736263737, "grad_norm": 0.97265625, "learning_rate": 7.564556621452169e-05, "loss": 0.1147, "step": 17385 }, { "epoch": 12.736996336996338, "grad_norm": 0.490234375, "learning_rate": 7.56426148160039e-05, "loss": 0.1091, "step": 17386 }, { "epoch": 12.737728937728939, "grad_norm": 0.84765625, "learning_rate": 7.563966329701959e-05, "loss": 0.1243, "step": 17387 }, { "epoch": 12.73846153846154, "grad_norm": 0.75, "learning_rate": 7.563671165758293e-05, "loss": 0.0915, "step": 17388 }, { "epoch": 12.73919413919414, "grad_norm": 1.328125, "learning_rate": 7.563375989770798e-05, "loss": 0.123, "step": 17389 }, { "epoch": 12.73992673992674, "grad_norm": 1.0, "learning_rate": 7.563080801740896e-05, "loss": 0.1376, "step": 17390 }, { "epoch": 12.740659340659342, "grad_norm": 0.609375, "learning_rate": 7.56278560167e-05, "loss": 0.1176, "step": 17391 }, { "epoch": 12.741391941391942, "grad_norm": 1.2890625, "learning_rate": 7.562490389559522e-05, "loss": 0.1288, "step": 17392 }, { "epoch": 12.742124542124543, "grad_norm": 0.8203125, "learning_rate": 7.562195165410877e-05, "loss": 0.1455, "step": 17393 }, { "epoch": 12.742857142857144, "grad_norm": 1.1171875, "learning_rate": 7.561899929225483e-05, "loss": 0.1083, "step": 17394 }, { "epoch": 12.743589743589745, "grad_norm": 1.015625, "learning_rate": 7.561604681004749e-05, "loss": 0.1394, "step": 17395 }, { "epoch": 12.744322344322345, "grad_norm": 0.98046875, "learning_rate": 7.561309420750094e-05, "loss": 0.1166, "step": 17396 }, { "epoch": 12.745054945054946, "grad_norm": 1.03125, "learning_rate": 7.561014148462931e-05, "loss": 0.1088, "step": 17397 }, { "epoch": 12.745787545787547, "grad_norm": 1.2734375, "learning_rate": 7.560718864144674e-05, "loss": 0.0997, "step": 17398 }, { "epoch": 12.746520146520147, "grad_norm": 1.515625, "learning_rate": 7.560423567796738e-05, "loss": 0.1151, "step": 17399 }, { "epoch": 12.747252747252748, "grad_norm": 1.78125, "learning_rate": 7.56012825942054e-05, "loss": 0.1471, "step": 17400 }, { "epoch": 12.747985347985349, "grad_norm": 1.1328125, "learning_rate": 7.559832939017494e-05, "loss": 0.1272, "step": 17401 }, { "epoch": 12.74871794871795, "grad_norm": 1.8828125, "learning_rate": 7.559537606589014e-05, "loss": 0.1515, "step": 17402 }, { "epoch": 12.74945054945055, "grad_norm": 1.0625, "learning_rate": 7.559242262136515e-05, "loss": 0.1209, "step": 17403 }, { "epoch": 12.750183150183151, "grad_norm": 1.0234375, "learning_rate": 7.558946905661413e-05, "loss": 0.1258, "step": 17404 }, { "epoch": 12.750915750915752, "grad_norm": 1.6953125, "learning_rate": 7.558651537165123e-05, "loss": 0.152, "step": 17405 }, { "epoch": 12.751648351648353, "grad_norm": 1.171875, "learning_rate": 7.55835615664906e-05, "loss": 0.113, "step": 17406 }, { "epoch": 12.752380952380953, "grad_norm": 1.0390625, "learning_rate": 7.558060764114639e-05, "loss": 0.165, "step": 17407 }, { "epoch": 12.753113553113554, "grad_norm": 0.8671875, "learning_rate": 7.557765359563275e-05, "loss": 0.1314, "step": 17408 }, { "epoch": 12.753846153846155, "grad_norm": 1.203125, "learning_rate": 7.557469942996382e-05, "loss": 0.1288, "step": 17409 }, { "epoch": 12.754578754578755, "grad_norm": 1.140625, "learning_rate": 7.557174514415382e-05, "loss": 0.1364, "step": 17410 }, { "epoch": 12.755311355311356, "grad_norm": 0.9453125, "learning_rate": 7.556879073821683e-05, "loss": 0.1419, "step": 17411 }, { "epoch": 12.756043956043957, "grad_norm": 1.25, "learning_rate": 7.556583621216704e-05, "loss": 0.1244, "step": 17412 }, { "epoch": 12.756776556776558, "grad_norm": 1.9453125, "learning_rate": 7.556288156601858e-05, "loss": 0.1434, "step": 17413 }, { "epoch": 12.757509157509158, "grad_norm": 1.4140625, "learning_rate": 7.555992679978567e-05, "loss": 0.1609, "step": 17414 }, { "epoch": 12.758241758241759, "grad_norm": 1.1796875, "learning_rate": 7.55569719134824e-05, "loss": 0.1101, "step": 17415 }, { "epoch": 12.75897435897436, "grad_norm": 1.484375, "learning_rate": 7.555401690712293e-05, "loss": 0.1491, "step": 17416 }, { "epoch": 12.75970695970696, "grad_norm": 0.6796875, "learning_rate": 7.555106178072147e-05, "loss": 0.121, "step": 17417 }, { "epoch": 12.760439560439561, "grad_norm": 0.921875, "learning_rate": 7.554810653429215e-05, "loss": 0.115, "step": 17418 }, { "epoch": 12.761172161172162, "grad_norm": 1.21875, "learning_rate": 7.554515116784912e-05, "loss": 0.1322, "step": 17419 }, { "epoch": 12.761904761904763, "grad_norm": 1.53125, "learning_rate": 7.554219568140655e-05, "loss": 0.1373, "step": 17420 }, { "epoch": 12.762637362637363, "grad_norm": 1.1640625, "learning_rate": 7.553924007497858e-05, "loss": 0.124, "step": 17421 }, { "epoch": 12.763369963369964, "grad_norm": 1.3515625, "learning_rate": 7.553628434857942e-05, "loss": 0.1368, "step": 17422 }, { "epoch": 12.764102564102565, "grad_norm": 1.3046875, "learning_rate": 7.55333285022232e-05, "loss": 0.1422, "step": 17423 }, { "epoch": 12.764835164835166, "grad_norm": 1.28125, "learning_rate": 7.553037253592407e-05, "loss": 0.1617, "step": 17424 }, { "epoch": 12.765567765567766, "grad_norm": 0.95703125, "learning_rate": 7.552741644969622e-05, "loss": 0.1339, "step": 17425 }, { "epoch": 12.766300366300367, "grad_norm": 0.9921875, "learning_rate": 7.55244602435538e-05, "loss": 0.103, "step": 17426 }, { "epoch": 12.767032967032968, "grad_norm": 0.953125, "learning_rate": 7.552150391751097e-05, "loss": 0.1182, "step": 17427 }, { "epoch": 12.767765567765569, "grad_norm": 1.1875, "learning_rate": 7.551854747158191e-05, "loss": 0.1723, "step": 17428 }, { "epoch": 12.76849816849817, "grad_norm": 0.77734375, "learning_rate": 7.551559090578077e-05, "loss": 0.1163, "step": 17429 }, { "epoch": 12.76923076923077, "grad_norm": 1.140625, "learning_rate": 7.551263422012172e-05, "loss": 0.1254, "step": 17430 }, { "epoch": 12.76996336996337, "grad_norm": 1.171875, "learning_rate": 7.550967741461893e-05, "loss": 0.1183, "step": 17431 }, { "epoch": 12.770695970695972, "grad_norm": 1.6328125, "learning_rate": 7.550672048928656e-05, "loss": 0.1253, "step": 17432 }, { "epoch": 12.771428571428572, "grad_norm": 0.6484375, "learning_rate": 7.55037634441388e-05, "loss": 0.1064, "step": 17433 }, { "epoch": 12.772161172161173, "grad_norm": 1.1171875, "learning_rate": 7.550080627918978e-05, "loss": 0.1249, "step": 17434 }, { "epoch": 12.772893772893774, "grad_norm": 1.2109375, "learning_rate": 7.54978489944537e-05, "loss": 0.1247, "step": 17435 }, { "epoch": 12.773626373626374, "grad_norm": 0.9140625, "learning_rate": 7.549489158994473e-05, "loss": 0.1341, "step": 17436 }, { "epoch": 12.774358974358975, "grad_norm": 1.109375, "learning_rate": 7.5491934065677e-05, "loss": 0.128, "step": 17437 }, { "epoch": 12.775091575091576, "grad_norm": 0.62890625, "learning_rate": 7.548897642166473e-05, "loss": 0.1189, "step": 17438 }, { "epoch": 12.775824175824177, "grad_norm": 0.74609375, "learning_rate": 7.548601865792207e-05, "loss": 0.1252, "step": 17439 }, { "epoch": 12.776556776556777, "grad_norm": 1.0234375, "learning_rate": 7.548306077446319e-05, "loss": 0.117, "step": 17440 }, { "epoch": 12.777289377289378, "grad_norm": 0.96875, "learning_rate": 7.548010277130228e-05, "loss": 0.1069, "step": 17441 }, { "epoch": 12.778021978021979, "grad_norm": 1.953125, "learning_rate": 7.547714464845346e-05, "loss": 0.1648, "step": 17442 }, { "epoch": 12.77875457875458, "grad_norm": 1.0625, "learning_rate": 7.547418640593098e-05, "loss": 0.1049, "step": 17443 }, { "epoch": 12.77948717948718, "grad_norm": 1.421875, "learning_rate": 7.547122804374895e-05, "loss": 0.1083, "step": 17444 }, { "epoch": 12.780219780219781, "grad_norm": 1.1953125, "learning_rate": 7.546826956192159e-05, "loss": 0.1366, "step": 17445 }, { "epoch": 12.780952380952382, "grad_norm": 0.7421875, "learning_rate": 7.546531096046304e-05, "loss": 0.1049, "step": 17446 }, { "epoch": 12.781684981684982, "grad_norm": 0.72265625, "learning_rate": 7.54623522393875e-05, "loss": 0.1193, "step": 17447 }, { "epoch": 12.782417582417583, "grad_norm": 0.89453125, "learning_rate": 7.54593933987091e-05, "loss": 0.1345, "step": 17448 }, { "epoch": 12.783150183150184, "grad_norm": 0.78515625, "learning_rate": 7.545643443844209e-05, "loss": 0.1368, "step": 17449 }, { "epoch": 12.783882783882785, "grad_norm": 0.984375, "learning_rate": 7.545347535860058e-05, "loss": 0.0997, "step": 17450 }, { "epoch": 12.784615384615385, "grad_norm": 0.7421875, "learning_rate": 7.54505161591988e-05, "loss": 0.1112, "step": 17451 }, { "epoch": 12.785347985347986, "grad_norm": 1.4921875, "learning_rate": 7.54475568402509e-05, "loss": 0.1254, "step": 17452 }, { "epoch": 12.786080586080587, "grad_norm": 1.0234375, "learning_rate": 7.544459740177108e-05, "loss": 0.1393, "step": 17453 }, { "epoch": 12.786813186813188, "grad_norm": 0.69921875, "learning_rate": 7.544163784377348e-05, "loss": 0.1151, "step": 17454 }, { "epoch": 12.787545787545788, "grad_norm": 1.3359375, "learning_rate": 7.543867816627232e-05, "loss": 0.105, "step": 17455 }, { "epoch": 12.788278388278389, "grad_norm": 1.4453125, "learning_rate": 7.543571836928177e-05, "loss": 0.1419, "step": 17456 }, { "epoch": 12.78901098901099, "grad_norm": 1.8359375, "learning_rate": 7.543275845281599e-05, "loss": 0.1273, "step": 17457 }, { "epoch": 12.78974358974359, "grad_norm": 1.234375, "learning_rate": 7.54297984168892e-05, "loss": 0.1466, "step": 17458 }, { "epoch": 12.790476190476191, "grad_norm": 1.8125, "learning_rate": 7.542683826151556e-05, "loss": 0.1657, "step": 17459 }, { "epoch": 12.791208791208792, "grad_norm": 1.4375, "learning_rate": 7.542387798670924e-05, "loss": 0.1135, "step": 17460 }, { "epoch": 12.791941391941393, "grad_norm": 0.58984375, "learning_rate": 7.542091759248444e-05, "loss": 0.1022, "step": 17461 }, { "epoch": 12.792673992673993, "grad_norm": 0.9296875, "learning_rate": 7.541795707885535e-05, "loss": 0.1546, "step": 17462 }, { "epoch": 12.793406593406594, "grad_norm": 0.87109375, "learning_rate": 7.541499644583616e-05, "loss": 0.131, "step": 17463 }, { "epoch": 12.794139194139195, "grad_norm": 0.7421875, "learning_rate": 7.541203569344102e-05, "loss": 0.1463, "step": 17464 }, { "epoch": 12.794871794871796, "grad_norm": 0.765625, "learning_rate": 7.540907482168415e-05, "loss": 0.0878, "step": 17465 }, { "epoch": 12.795604395604396, "grad_norm": 0.7421875, "learning_rate": 7.540611383057973e-05, "loss": 0.1279, "step": 17466 }, { "epoch": 12.796336996336997, "grad_norm": 1.1796875, "learning_rate": 7.540315272014195e-05, "loss": 0.1374, "step": 17467 }, { "epoch": 12.797069597069598, "grad_norm": 0.83984375, "learning_rate": 7.540019149038495e-05, "loss": 0.1364, "step": 17468 }, { "epoch": 12.797802197802199, "grad_norm": 0.86328125, "learning_rate": 7.539723014132299e-05, "loss": 0.126, "step": 17469 }, { "epoch": 12.7985347985348, "grad_norm": 1.359375, "learning_rate": 7.539426867297024e-05, "loss": 0.1572, "step": 17470 }, { "epoch": 12.7992673992674, "grad_norm": 0.69921875, "learning_rate": 7.539130708534085e-05, "loss": 0.1019, "step": 17471 }, { "epoch": 12.8, "grad_norm": 1.171875, "learning_rate": 7.538834537844904e-05, "loss": 0.1154, "step": 17472 }, { "epoch": 12.800732600732601, "grad_norm": 0.90625, "learning_rate": 7.5385383552309e-05, "loss": 0.119, "step": 17473 }, { "epoch": 12.801465201465202, "grad_norm": 1.0390625, "learning_rate": 7.538242160693492e-05, "loss": 0.1383, "step": 17474 }, { "epoch": 12.802197802197803, "grad_norm": 1.0546875, "learning_rate": 7.537945954234099e-05, "loss": 0.1347, "step": 17475 }, { "epoch": 12.802930402930404, "grad_norm": 1.171875, "learning_rate": 7.53764973585414e-05, "loss": 0.1198, "step": 17476 }, { "epoch": 12.803663003663004, "grad_norm": 1.171875, "learning_rate": 7.537353505555035e-05, "loss": 0.1257, "step": 17477 }, { "epoch": 12.804395604395605, "grad_norm": 1.7421875, "learning_rate": 7.537057263338202e-05, "loss": 0.1505, "step": 17478 }, { "epoch": 12.805128205128206, "grad_norm": 0.93359375, "learning_rate": 7.536761009205062e-05, "loss": 0.1312, "step": 17479 }, { "epoch": 12.805860805860807, "grad_norm": 0.9296875, "learning_rate": 7.536464743157034e-05, "loss": 0.1615, "step": 17480 }, { "epoch": 12.806593406593407, "grad_norm": 1.171875, "learning_rate": 7.536168465195535e-05, "loss": 0.1444, "step": 17481 }, { "epoch": 12.807326007326008, "grad_norm": 0.93359375, "learning_rate": 7.53587217532199e-05, "loss": 0.0994, "step": 17482 }, { "epoch": 12.808058608058609, "grad_norm": 1.2421875, "learning_rate": 7.535575873537814e-05, "loss": 0.1252, "step": 17483 }, { "epoch": 12.80879120879121, "grad_norm": 1.1875, "learning_rate": 7.535279559844427e-05, "loss": 0.1095, "step": 17484 }, { "epoch": 12.80952380952381, "grad_norm": 1.2109375, "learning_rate": 7.53498323424325e-05, "loss": 0.1037, "step": 17485 }, { "epoch": 12.810256410256411, "grad_norm": 0.96484375, "learning_rate": 7.534686896735702e-05, "loss": 0.1336, "step": 17486 }, { "epoch": 12.810989010989012, "grad_norm": 1.2890625, "learning_rate": 7.534390547323205e-05, "loss": 0.1371, "step": 17487 }, { "epoch": 12.811721611721612, "grad_norm": 1.390625, "learning_rate": 7.534094186007175e-05, "loss": 0.1003, "step": 17488 }, { "epoch": 12.812454212454213, "grad_norm": 1.0703125, "learning_rate": 7.533797812789036e-05, "loss": 0.1421, "step": 17489 }, { "epoch": 12.813186813186814, "grad_norm": 1.1328125, "learning_rate": 7.533501427670209e-05, "loss": 0.1123, "step": 17490 }, { "epoch": 12.813919413919415, "grad_norm": 1.6328125, "learning_rate": 7.533205030652108e-05, "loss": 0.1516, "step": 17491 }, { "epoch": 12.814652014652015, "grad_norm": 1.1328125, "learning_rate": 7.532908621736156e-05, "loss": 0.0996, "step": 17492 }, { "epoch": 12.815384615384616, "grad_norm": 1.1796875, "learning_rate": 7.532612200923776e-05, "loss": 0.1089, "step": 17493 }, { "epoch": 12.816117216117217, "grad_norm": 0.84375, "learning_rate": 7.532315768216383e-05, "loss": 0.1031, "step": 17494 }, { "epoch": 12.816849816849818, "grad_norm": 0.75390625, "learning_rate": 7.532019323615404e-05, "loss": 0.0991, "step": 17495 }, { "epoch": 12.817582417582418, "grad_norm": 1.0390625, "learning_rate": 7.531722867122253e-05, "loss": 0.1314, "step": 17496 }, { "epoch": 12.818315018315019, "grad_norm": 1.09375, "learning_rate": 7.531426398738355e-05, "loss": 0.114, "step": 17497 }, { "epoch": 12.81904761904762, "grad_norm": 1.203125, "learning_rate": 7.531129918465127e-05, "loss": 0.1356, "step": 17498 }, { "epoch": 12.81978021978022, "grad_norm": 0.77734375, "learning_rate": 7.53083342630399e-05, "loss": 0.111, "step": 17499 }, { "epoch": 12.820512820512821, "grad_norm": 1.0390625, "learning_rate": 7.530536922256367e-05, "loss": 0.1081, "step": 17500 }, { "epoch": 12.821245421245422, "grad_norm": 0.79296875, "learning_rate": 7.530240406323676e-05, "loss": 0.1261, "step": 17501 }, { "epoch": 12.821978021978023, "grad_norm": 0.70703125, "learning_rate": 7.52994387850734e-05, "loss": 0.1008, "step": 17502 }, { "epoch": 12.822710622710623, "grad_norm": 1.234375, "learning_rate": 7.529647338808781e-05, "loss": 0.1659, "step": 17503 }, { "epoch": 12.823443223443224, "grad_norm": 2.0625, "learning_rate": 7.529350787229415e-05, "loss": 0.1383, "step": 17504 }, { "epoch": 12.824175824175825, "grad_norm": 0.9296875, "learning_rate": 7.529054223770666e-05, "loss": 0.1224, "step": 17505 }, { "epoch": 12.824908424908426, "grad_norm": 0.92578125, "learning_rate": 7.528757648433955e-05, "loss": 0.1023, "step": 17506 }, { "epoch": 12.825641025641026, "grad_norm": 1.296875, "learning_rate": 7.528461061220702e-05, "loss": 0.1445, "step": 17507 }, { "epoch": 12.826373626373627, "grad_norm": 1.21875, "learning_rate": 7.528164462132329e-05, "loss": 0.1359, "step": 17508 }, { "epoch": 12.827106227106228, "grad_norm": 1.0234375, "learning_rate": 7.527867851170255e-05, "loss": 0.1158, "step": 17509 }, { "epoch": 12.827838827838828, "grad_norm": 0.57421875, "learning_rate": 7.527571228335903e-05, "loss": 0.1212, "step": 17510 }, { "epoch": 12.82857142857143, "grad_norm": 1.015625, "learning_rate": 7.527274593630695e-05, "loss": 0.1359, "step": 17511 }, { "epoch": 12.82930402930403, "grad_norm": 1.5078125, "learning_rate": 7.526977947056053e-05, "loss": 0.1558, "step": 17512 }, { "epoch": 12.83003663003663, "grad_norm": 1.265625, "learning_rate": 7.526681288613391e-05, "loss": 0.1213, "step": 17513 }, { "epoch": 12.830769230769231, "grad_norm": 1.6015625, "learning_rate": 7.52638461830414e-05, "loss": 0.1255, "step": 17514 }, { "epoch": 12.831501831501832, "grad_norm": 1.9375, "learning_rate": 7.526087936129717e-05, "loss": 0.1405, "step": 17515 }, { "epoch": 12.832234432234433, "grad_norm": 1.03125, "learning_rate": 7.525791242091543e-05, "loss": 0.1071, "step": 17516 }, { "epoch": 12.832967032967034, "grad_norm": 1.3359375, "learning_rate": 7.525494536191038e-05, "loss": 0.1248, "step": 17517 }, { "epoch": 12.833699633699634, "grad_norm": 1.2265625, "learning_rate": 7.525197818429631e-05, "loss": 0.1025, "step": 17518 }, { "epoch": 12.834432234432235, "grad_norm": 0.78125, "learning_rate": 7.524901088808737e-05, "loss": 0.1167, "step": 17519 }, { "epoch": 12.835164835164836, "grad_norm": 1.0234375, "learning_rate": 7.524604347329777e-05, "loss": 0.1101, "step": 17520 }, { "epoch": 12.835897435897436, "grad_norm": 1.296875, "learning_rate": 7.524307593994178e-05, "loss": 0.1292, "step": 17521 }, { "epoch": 12.836630036630037, "grad_norm": 1.2578125, "learning_rate": 7.524010828803357e-05, "loss": 0.1293, "step": 17522 }, { "epoch": 12.837362637362638, "grad_norm": 1.296875, "learning_rate": 7.52371405175874e-05, "loss": 0.1358, "step": 17523 }, { "epoch": 12.838095238095239, "grad_norm": 1.7265625, "learning_rate": 7.523417262861746e-05, "loss": 0.1312, "step": 17524 }, { "epoch": 12.83882783882784, "grad_norm": 2.0, "learning_rate": 7.523120462113797e-05, "loss": 0.136, "step": 17525 }, { "epoch": 12.83956043956044, "grad_norm": 1.3125, "learning_rate": 7.522823649516318e-05, "loss": 0.1277, "step": 17526 }, { "epoch": 12.84029304029304, "grad_norm": 0.65625, "learning_rate": 7.522526825070728e-05, "loss": 0.1138, "step": 17527 }, { "epoch": 12.841025641025642, "grad_norm": 1.4296875, "learning_rate": 7.522229988778449e-05, "loss": 0.1486, "step": 17528 }, { "epoch": 12.841758241758242, "grad_norm": 1.5234375, "learning_rate": 7.521933140640907e-05, "loss": 0.1284, "step": 17529 }, { "epoch": 12.842490842490843, "grad_norm": 1.6015625, "learning_rate": 7.52163628065952e-05, "loss": 0.1581, "step": 17530 }, { "epoch": 12.843223443223444, "grad_norm": 1.3671875, "learning_rate": 7.521339408835713e-05, "loss": 0.1294, "step": 17531 }, { "epoch": 12.843956043956045, "grad_norm": 1.03125, "learning_rate": 7.521042525170908e-05, "loss": 0.1387, "step": 17532 }, { "epoch": 12.844688644688645, "grad_norm": 1.8359375, "learning_rate": 7.520745629666527e-05, "loss": 0.1307, "step": 17533 }, { "epoch": 12.845421245421246, "grad_norm": 0.88671875, "learning_rate": 7.520448722323992e-05, "loss": 0.1176, "step": 17534 }, { "epoch": 12.846153846153847, "grad_norm": 1.0859375, "learning_rate": 7.520151803144725e-05, "loss": 0.1637, "step": 17535 }, { "epoch": 12.846886446886447, "grad_norm": 1.4921875, "learning_rate": 7.519854872130152e-05, "loss": 0.1479, "step": 17536 }, { "epoch": 12.847619047619048, "grad_norm": 0.984375, "learning_rate": 7.519557929281692e-05, "loss": 0.1177, "step": 17537 }, { "epoch": 12.848351648351649, "grad_norm": 1.6171875, "learning_rate": 7.519260974600771e-05, "loss": 0.1508, "step": 17538 }, { "epoch": 12.84908424908425, "grad_norm": 1.453125, "learning_rate": 7.518964008088807e-05, "loss": 0.1553, "step": 17539 }, { "epoch": 12.84981684981685, "grad_norm": 1.0, "learning_rate": 7.518667029747228e-05, "loss": 0.0981, "step": 17540 }, { "epoch": 12.850549450549451, "grad_norm": 1.0, "learning_rate": 7.518370039577454e-05, "loss": 0.1403, "step": 17541 }, { "epoch": 12.851282051282052, "grad_norm": 1.28125, "learning_rate": 7.518073037580912e-05, "loss": 0.1604, "step": 17542 }, { "epoch": 12.852014652014653, "grad_norm": 1.3671875, "learning_rate": 7.51777602375902e-05, "loss": 0.1282, "step": 17543 }, { "epoch": 12.852747252747253, "grad_norm": 1.59375, "learning_rate": 7.5174789981132e-05, "loss": 0.1564, "step": 17544 }, { "epoch": 12.853479853479854, "grad_norm": 1.1953125, "learning_rate": 7.517181960644882e-05, "loss": 0.149, "step": 17545 }, { "epoch": 12.854212454212455, "grad_norm": 0.703125, "learning_rate": 7.516884911355483e-05, "loss": 0.1467, "step": 17546 }, { "epoch": 12.854945054945055, "grad_norm": 0.90625, "learning_rate": 7.51658785024643e-05, "loss": 0.1286, "step": 17547 }, { "epoch": 12.855677655677656, "grad_norm": 1.0703125, "learning_rate": 7.516290777319144e-05, "loss": 0.1442, "step": 17548 }, { "epoch": 12.856410256410257, "grad_norm": 0.71484375, "learning_rate": 7.51599369257505e-05, "loss": 0.118, "step": 17549 }, { "epoch": 12.857142857142858, "grad_norm": 0.82421875, "learning_rate": 7.515696596015569e-05, "loss": 0.1364, "step": 17550 }, { "epoch": 12.857875457875458, "grad_norm": 0.765625, "learning_rate": 7.515399487642128e-05, "loss": 0.1147, "step": 17551 }, { "epoch": 12.858608058608059, "grad_norm": 0.8359375, "learning_rate": 7.515102367456147e-05, "loss": 0.143, "step": 17552 }, { "epoch": 12.85934065934066, "grad_norm": 1.40625, "learning_rate": 7.514805235459053e-05, "loss": 0.126, "step": 17553 }, { "epoch": 12.86007326007326, "grad_norm": 0.66796875, "learning_rate": 7.514508091652268e-05, "loss": 0.1204, "step": 17554 }, { "epoch": 12.860805860805861, "grad_norm": 0.83203125, "learning_rate": 7.514210936037215e-05, "loss": 0.1301, "step": 17555 }, { "epoch": 12.861538461538462, "grad_norm": 1.0625, "learning_rate": 7.513913768615317e-05, "loss": 0.1584, "step": 17556 }, { "epoch": 12.862271062271063, "grad_norm": 0.7890625, "learning_rate": 7.513616589388002e-05, "loss": 0.1033, "step": 17557 }, { "epoch": 12.863003663003663, "grad_norm": 1.0078125, "learning_rate": 7.513319398356691e-05, "loss": 0.1348, "step": 17558 }, { "epoch": 12.863736263736264, "grad_norm": 1.0859375, "learning_rate": 7.513022195522805e-05, "loss": 0.133, "step": 17559 }, { "epoch": 12.864468864468865, "grad_norm": 0.9375, "learning_rate": 7.512724980887774e-05, "loss": 0.1077, "step": 17560 }, { "epoch": 12.865201465201466, "grad_norm": 1.0, "learning_rate": 7.512427754453019e-05, "loss": 0.1056, "step": 17561 }, { "epoch": 12.865934065934066, "grad_norm": 1.53125, "learning_rate": 7.512130516219963e-05, "loss": 0.1555, "step": 17562 }, { "epoch": 12.866666666666667, "grad_norm": 0.859375, "learning_rate": 7.51183326619003e-05, "loss": 0.1264, "step": 17563 }, { "epoch": 12.867399267399268, "grad_norm": 1.3671875, "learning_rate": 7.51153600436465e-05, "loss": 0.1154, "step": 17564 }, { "epoch": 12.868131868131869, "grad_norm": 1.109375, "learning_rate": 7.511238730745241e-05, "loss": 0.1341, "step": 17565 }, { "epoch": 12.86886446886447, "grad_norm": 0.91015625, "learning_rate": 7.510941445333228e-05, "loss": 0.1354, "step": 17566 }, { "epoch": 12.86959706959707, "grad_norm": 0.95703125, "learning_rate": 7.510644148130037e-05, "loss": 0.1335, "step": 17567 }, { "epoch": 12.87032967032967, "grad_norm": 1.3125, "learning_rate": 7.510346839137094e-05, "loss": 0.1043, "step": 17568 }, { "epoch": 12.871062271062272, "grad_norm": 1.0859375, "learning_rate": 7.510049518355819e-05, "loss": 0.1305, "step": 17569 }, { "epoch": 12.871794871794872, "grad_norm": 1.390625, "learning_rate": 7.509752185787641e-05, "loss": 0.1111, "step": 17570 }, { "epoch": 12.872527472527473, "grad_norm": 1.140625, "learning_rate": 7.509454841433982e-05, "loss": 0.1344, "step": 17571 }, { "epoch": 12.873260073260074, "grad_norm": 1.1640625, "learning_rate": 7.509157485296268e-05, "loss": 0.1568, "step": 17572 }, { "epoch": 12.873992673992674, "grad_norm": 1.0859375, "learning_rate": 7.508860117375924e-05, "loss": 0.1211, "step": 17573 }, { "epoch": 12.874725274725275, "grad_norm": 1.71875, "learning_rate": 7.50856273767437e-05, "loss": 0.1333, "step": 17574 }, { "epoch": 12.875457875457876, "grad_norm": 1.3515625, "learning_rate": 7.50826534619304e-05, "loss": 0.1392, "step": 17575 }, { "epoch": 12.876190476190477, "grad_norm": 1.3515625, "learning_rate": 7.50796794293335e-05, "loss": 0.1053, "step": 17576 }, { "epoch": 12.876923076923077, "grad_norm": 1.25, "learning_rate": 7.50767052789673e-05, "loss": 0.1573, "step": 17577 }, { "epoch": 12.877655677655678, "grad_norm": 0.82421875, "learning_rate": 7.507373101084604e-05, "loss": 0.124, "step": 17578 }, { "epoch": 12.878388278388279, "grad_norm": 0.921875, "learning_rate": 7.507075662498396e-05, "loss": 0.1322, "step": 17579 }, { "epoch": 12.87912087912088, "grad_norm": 0.859375, "learning_rate": 7.506778212139532e-05, "loss": 0.1353, "step": 17580 }, { "epoch": 12.87985347985348, "grad_norm": 0.81640625, "learning_rate": 7.506480750009439e-05, "loss": 0.1164, "step": 17581 }, { "epoch": 12.880586080586081, "grad_norm": 1.2109375, "learning_rate": 7.506183276109538e-05, "loss": 0.1413, "step": 17582 }, { "epoch": 12.881318681318682, "grad_norm": 0.79296875, "learning_rate": 7.505885790441259e-05, "loss": 0.1197, "step": 17583 }, { "epoch": 12.882051282051282, "grad_norm": 1.28125, "learning_rate": 7.505588293006023e-05, "loss": 0.1471, "step": 17584 }, { "epoch": 12.882783882783883, "grad_norm": 1.4375, "learning_rate": 7.505290783805258e-05, "loss": 0.1262, "step": 17585 }, { "epoch": 12.883516483516484, "grad_norm": 0.9921875, "learning_rate": 7.504993262840389e-05, "loss": 0.1279, "step": 17586 }, { "epoch": 12.884249084249085, "grad_norm": 0.90625, "learning_rate": 7.504695730112841e-05, "loss": 0.1527, "step": 17587 }, { "epoch": 12.884981684981685, "grad_norm": 0.6015625, "learning_rate": 7.50439818562404e-05, "loss": 0.0988, "step": 17588 }, { "epoch": 12.885714285714286, "grad_norm": 1.3984375, "learning_rate": 7.504100629375411e-05, "loss": 0.1175, "step": 17589 }, { "epoch": 12.886446886446887, "grad_norm": 0.67578125, "learning_rate": 7.503803061368381e-05, "loss": 0.1288, "step": 17590 }, { "epoch": 12.887179487179488, "grad_norm": 0.78515625, "learning_rate": 7.503505481604374e-05, "loss": 0.1168, "step": 17591 }, { "epoch": 12.887912087912088, "grad_norm": 0.828125, "learning_rate": 7.503207890084819e-05, "loss": 0.1069, "step": 17592 }, { "epoch": 12.888644688644689, "grad_norm": 0.796875, "learning_rate": 7.502910286811137e-05, "loss": 0.1272, "step": 17593 }, { "epoch": 12.88937728937729, "grad_norm": 0.9765625, "learning_rate": 7.502612671784758e-05, "loss": 0.1095, "step": 17594 }, { "epoch": 12.89010989010989, "grad_norm": 1.1171875, "learning_rate": 7.502315045007105e-05, "loss": 0.1154, "step": 17595 }, { "epoch": 12.890842490842491, "grad_norm": 0.875, "learning_rate": 7.502017406479607e-05, "loss": 0.1381, "step": 17596 }, { "epoch": 12.891575091575092, "grad_norm": 0.75, "learning_rate": 7.50171975620369e-05, "loss": 0.1113, "step": 17597 }, { "epoch": 12.892307692307693, "grad_norm": 1.1796875, "learning_rate": 7.501422094180777e-05, "loss": 0.1185, "step": 17598 }, { "epoch": 12.893040293040293, "grad_norm": 0.828125, "learning_rate": 7.501124420412298e-05, "loss": 0.0982, "step": 17599 }, { "epoch": 12.893772893772894, "grad_norm": 0.95703125, "learning_rate": 7.500826734899674e-05, "loss": 0.1309, "step": 17600 }, { "epoch": 12.894505494505495, "grad_norm": 0.921875, "learning_rate": 7.500529037644337e-05, "loss": 0.1198, "step": 17601 }, { "epoch": 12.895238095238096, "grad_norm": 0.82421875, "learning_rate": 7.50023132864771e-05, "loss": 0.1057, "step": 17602 }, { "epoch": 12.895970695970696, "grad_norm": 0.91015625, "learning_rate": 7.499933607911219e-05, "loss": 0.1231, "step": 17603 }, { "epoch": 12.896703296703297, "grad_norm": 1.0, "learning_rate": 7.499635875436294e-05, "loss": 0.1295, "step": 17604 }, { "epoch": 12.897435897435898, "grad_norm": 1.171875, "learning_rate": 7.499338131224359e-05, "loss": 0.1489, "step": 17605 }, { "epoch": 12.898168498168499, "grad_norm": 0.99609375, "learning_rate": 7.499040375276841e-05, "loss": 0.1148, "step": 17606 }, { "epoch": 12.8989010989011, "grad_norm": 1.6796875, "learning_rate": 7.498742607595167e-05, "loss": 0.1113, "step": 17607 }, { "epoch": 12.8996336996337, "grad_norm": 1.2578125, "learning_rate": 7.498444828180762e-05, "loss": 0.1165, "step": 17608 }, { "epoch": 12.9003663003663, "grad_norm": 1.2890625, "learning_rate": 7.498147037035053e-05, "loss": 0.1354, "step": 17609 }, { "epoch": 12.901098901098901, "grad_norm": 0.85546875, "learning_rate": 7.497849234159471e-05, "loss": 0.1318, "step": 17610 }, { "epoch": 12.901831501831502, "grad_norm": 0.9921875, "learning_rate": 7.497551419555437e-05, "loss": 0.1384, "step": 17611 }, { "epoch": 12.902564102564103, "grad_norm": 1.9140625, "learning_rate": 7.497253593224383e-05, "loss": 0.1372, "step": 17612 }, { "epoch": 12.903296703296704, "grad_norm": 1.140625, "learning_rate": 7.496955755167733e-05, "loss": 0.1362, "step": 17613 }, { "epoch": 12.904029304029304, "grad_norm": 1.5078125, "learning_rate": 7.496657905386913e-05, "loss": 0.1325, "step": 17614 }, { "epoch": 12.904761904761905, "grad_norm": 1.171875, "learning_rate": 7.496360043883353e-05, "loss": 0.1632, "step": 17615 }, { "epoch": 12.905494505494506, "grad_norm": 1.3203125, "learning_rate": 7.496062170658479e-05, "loss": 0.1476, "step": 17616 }, { "epoch": 12.906227106227107, "grad_norm": 0.72265625, "learning_rate": 7.495764285713717e-05, "loss": 0.1052, "step": 17617 }, { "epoch": 12.906959706959707, "grad_norm": 0.73046875, "learning_rate": 7.495466389050498e-05, "loss": 0.1157, "step": 17618 }, { "epoch": 12.907692307692308, "grad_norm": 1.0, "learning_rate": 7.495168480670244e-05, "loss": 0.1118, "step": 17619 }, { "epoch": 12.908424908424909, "grad_norm": 1.109375, "learning_rate": 7.494870560574388e-05, "loss": 0.1448, "step": 17620 }, { "epoch": 12.90915750915751, "grad_norm": 1.2578125, "learning_rate": 7.494572628764352e-05, "loss": 0.1225, "step": 17621 }, { "epoch": 12.90989010989011, "grad_norm": 1.3515625, "learning_rate": 7.494274685241567e-05, "loss": 0.11, "step": 17622 }, { "epoch": 12.910622710622711, "grad_norm": 0.91015625, "learning_rate": 7.49397673000746e-05, "loss": 0.1237, "step": 17623 }, { "epoch": 12.911355311355312, "grad_norm": 0.94921875, "learning_rate": 7.493678763063457e-05, "loss": 0.1431, "step": 17624 }, { "epoch": 12.912087912087912, "grad_norm": 1.015625, "learning_rate": 7.493380784410988e-05, "loss": 0.1188, "step": 17625 }, { "epoch": 12.912820512820513, "grad_norm": 0.94921875, "learning_rate": 7.493082794051479e-05, "loss": 0.1184, "step": 17626 }, { "epoch": 12.913553113553114, "grad_norm": 1.0, "learning_rate": 7.492784791986358e-05, "loss": 0.138, "step": 17627 }, { "epoch": 12.914285714285715, "grad_norm": 0.93359375, "learning_rate": 7.492486778217053e-05, "loss": 0.138, "step": 17628 }, { "epoch": 12.915018315018315, "grad_norm": 1.6875, "learning_rate": 7.492188752744993e-05, "loss": 0.1392, "step": 17629 }, { "epoch": 12.915750915750916, "grad_norm": 2.140625, "learning_rate": 7.491890715571605e-05, "loss": 0.1916, "step": 17630 }, { "epoch": 12.916483516483517, "grad_norm": 1.265625, "learning_rate": 7.491592666698316e-05, "loss": 0.1523, "step": 17631 }, { "epoch": 12.917216117216118, "grad_norm": 1.40625, "learning_rate": 7.491294606126556e-05, "loss": 0.1478, "step": 17632 }, { "epoch": 12.917948717948718, "grad_norm": 1.0546875, "learning_rate": 7.490996533857753e-05, "loss": 0.123, "step": 17633 }, { "epoch": 12.918681318681319, "grad_norm": 0.8515625, "learning_rate": 7.490698449893331e-05, "loss": 0.1271, "step": 17634 }, { "epoch": 12.91941391941392, "grad_norm": 1.09375, "learning_rate": 7.490400354234724e-05, "loss": 0.1105, "step": 17635 }, { "epoch": 12.92014652014652, "grad_norm": 1.265625, "learning_rate": 7.490102246883358e-05, "loss": 0.13, "step": 17636 }, { "epoch": 12.920879120879121, "grad_norm": 1.0390625, "learning_rate": 7.48980412784066e-05, "loss": 0.1327, "step": 17637 }, { "epoch": 12.921611721611722, "grad_norm": 0.7265625, "learning_rate": 7.48950599710806e-05, "loss": 0.1277, "step": 17638 }, { "epoch": 12.922344322344323, "grad_norm": 1.2265625, "learning_rate": 7.489207854686984e-05, "loss": 0.1327, "step": 17639 }, { "epoch": 12.923076923076923, "grad_norm": 0.796875, "learning_rate": 7.488909700578865e-05, "loss": 0.1234, "step": 17640 }, { "epoch": 12.923809523809524, "grad_norm": 1.0625, "learning_rate": 7.488611534785127e-05, "loss": 0.1384, "step": 17641 }, { "epoch": 12.924542124542125, "grad_norm": 1.3203125, "learning_rate": 7.488313357307202e-05, "loss": 0.1494, "step": 17642 }, { "epoch": 12.925274725274726, "grad_norm": 0.6171875, "learning_rate": 7.488015168146517e-05, "loss": 0.1292, "step": 17643 }, { "epoch": 12.926007326007326, "grad_norm": 0.7109375, "learning_rate": 7.487716967304502e-05, "loss": 0.1075, "step": 17644 }, { "epoch": 12.926739926739927, "grad_norm": 1.0, "learning_rate": 7.487418754782584e-05, "loss": 0.1235, "step": 17645 }, { "epoch": 12.927472527472528, "grad_norm": 1.125, "learning_rate": 7.487120530582193e-05, "loss": 0.1006, "step": 17646 }, { "epoch": 12.928205128205128, "grad_norm": 1.0625, "learning_rate": 7.486822294704754e-05, "loss": 0.1181, "step": 17647 }, { "epoch": 12.92893772893773, "grad_norm": 1.484375, "learning_rate": 7.486524047151704e-05, "loss": 0.1211, "step": 17648 }, { "epoch": 12.92967032967033, "grad_norm": 0.95703125, "learning_rate": 7.486225787924465e-05, "loss": 0.1214, "step": 17649 }, { "epoch": 12.93040293040293, "grad_norm": 1.21875, "learning_rate": 7.485927517024469e-05, "loss": 0.1213, "step": 17650 }, { "epoch": 12.931135531135531, "grad_norm": 1.6640625, "learning_rate": 7.485629234453145e-05, "loss": 0.1257, "step": 17651 }, { "epoch": 12.931868131868132, "grad_norm": 1.2734375, "learning_rate": 7.485330940211921e-05, "loss": 0.1261, "step": 17652 }, { "epoch": 12.932600732600733, "grad_norm": 1.6171875, "learning_rate": 7.48503263430223e-05, "loss": 0.1514, "step": 17653 }, { "epoch": 12.933333333333334, "grad_norm": 1.3671875, "learning_rate": 7.484734316725494e-05, "loss": 0.1535, "step": 17654 }, { "epoch": 12.934065934065934, "grad_norm": 1.328125, "learning_rate": 7.484435987483148e-05, "loss": 0.1286, "step": 17655 }, { "epoch": 12.934798534798535, "grad_norm": 0.99609375, "learning_rate": 7.48413764657662e-05, "loss": 0.1281, "step": 17656 }, { "epoch": 12.935531135531136, "grad_norm": 1.1875, "learning_rate": 7.483839294007341e-05, "loss": 0.1336, "step": 17657 }, { "epoch": 12.936263736263736, "grad_norm": 1.1796875, "learning_rate": 7.483540929776737e-05, "loss": 0.1315, "step": 17658 }, { "epoch": 12.936996336996337, "grad_norm": 1.3359375, "learning_rate": 7.48324255388624e-05, "loss": 0.1648, "step": 17659 }, { "epoch": 12.937728937728938, "grad_norm": 1.3125, "learning_rate": 7.48294416633728e-05, "loss": 0.1305, "step": 17660 }, { "epoch": 12.938461538461539, "grad_norm": 1.0078125, "learning_rate": 7.482645767131285e-05, "loss": 0.1362, "step": 17661 }, { "epoch": 12.93919413919414, "grad_norm": 1.3359375, "learning_rate": 7.482347356269688e-05, "loss": 0.1323, "step": 17662 }, { "epoch": 12.93992673992674, "grad_norm": 1.3359375, "learning_rate": 7.482048933753912e-05, "loss": 0.1542, "step": 17663 }, { "epoch": 12.94065934065934, "grad_norm": 0.94140625, "learning_rate": 7.481750499585394e-05, "loss": 0.1273, "step": 17664 }, { "epoch": 12.941391941391942, "grad_norm": 1.4296875, "learning_rate": 7.48145205376556e-05, "loss": 0.149, "step": 17665 }, { "epoch": 12.942124542124542, "grad_norm": 1.234375, "learning_rate": 7.481153596295842e-05, "loss": 0.1619, "step": 17666 }, { "epoch": 12.942857142857143, "grad_norm": 1.4921875, "learning_rate": 7.480855127177669e-05, "loss": 0.1783, "step": 17667 }, { "epoch": 12.943589743589744, "grad_norm": 1.359375, "learning_rate": 7.48055664641247e-05, "loss": 0.1228, "step": 17668 }, { "epoch": 12.944322344322345, "grad_norm": 0.98046875, "learning_rate": 7.480258154001676e-05, "loss": 0.1494, "step": 17669 }, { "epoch": 12.945054945054945, "grad_norm": 1.25, "learning_rate": 7.47995964994672e-05, "loss": 0.1109, "step": 17670 }, { "epoch": 12.945787545787546, "grad_norm": 2.171875, "learning_rate": 7.479661134249027e-05, "loss": 0.1503, "step": 17671 }, { "epoch": 12.946520146520147, "grad_norm": 0.87890625, "learning_rate": 7.47936260691003e-05, "loss": 0.123, "step": 17672 }, { "epoch": 12.947252747252747, "grad_norm": 0.92578125, "learning_rate": 7.47906406793116e-05, "loss": 0.1101, "step": 17673 }, { "epoch": 12.947985347985348, "grad_norm": 1.0078125, "learning_rate": 7.478765517313847e-05, "loss": 0.1247, "step": 17674 }, { "epoch": 12.948717948717949, "grad_norm": 1.1953125, "learning_rate": 7.47846695505952e-05, "loss": 0.1329, "step": 17675 }, { "epoch": 12.94945054945055, "grad_norm": 1.390625, "learning_rate": 7.478168381169612e-05, "loss": 0.1366, "step": 17676 }, { "epoch": 12.95018315018315, "grad_norm": 2.171875, "learning_rate": 7.477869795645551e-05, "loss": 0.1213, "step": 17677 }, { "epoch": 12.950915750915751, "grad_norm": 1.0625, "learning_rate": 7.477571198488769e-05, "loss": 0.1536, "step": 17678 }, { "epoch": 12.951648351648352, "grad_norm": 1.0078125, "learning_rate": 7.477272589700698e-05, "loss": 0.1295, "step": 17679 }, { "epoch": 12.952380952380953, "grad_norm": 0.83984375, "learning_rate": 7.476973969282765e-05, "loss": 0.1428, "step": 17680 }, { "epoch": 12.953113553113553, "grad_norm": 0.59375, "learning_rate": 7.476675337236403e-05, "loss": 0.1191, "step": 17681 }, { "epoch": 12.953846153846154, "grad_norm": 0.91015625, "learning_rate": 7.476376693563044e-05, "loss": 0.1283, "step": 17682 }, { "epoch": 12.954578754578755, "grad_norm": 1.25, "learning_rate": 7.476078038264118e-05, "loss": 0.1531, "step": 17683 }, { "epoch": 12.955311355311355, "grad_norm": 0.859375, "learning_rate": 7.475779371341055e-05, "loss": 0.125, "step": 17684 }, { "epoch": 12.956043956043956, "grad_norm": 1.1796875, "learning_rate": 7.475480692795287e-05, "loss": 0.0859, "step": 17685 }, { "epoch": 12.956776556776557, "grad_norm": 0.9609375, "learning_rate": 7.475182002628245e-05, "loss": 0.109, "step": 17686 }, { "epoch": 12.957509157509158, "grad_norm": 0.81640625, "learning_rate": 7.47488330084136e-05, "loss": 0.1197, "step": 17687 }, { "epoch": 12.958241758241758, "grad_norm": 1.9375, "learning_rate": 7.474584587436062e-05, "loss": 0.1413, "step": 17688 }, { "epoch": 12.95897435897436, "grad_norm": 1.328125, "learning_rate": 7.474285862413784e-05, "loss": 0.1043, "step": 17689 }, { "epoch": 12.95970695970696, "grad_norm": 2.125, "learning_rate": 7.473987125775956e-05, "loss": 0.123, "step": 17690 }, { "epoch": 12.96043956043956, "grad_norm": 1.6796875, "learning_rate": 7.473688377524011e-05, "loss": 0.1434, "step": 17691 }, { "epoch": 12.961172161172161, "grad_norm": 0.76953125, "learning_rate": 7.473389617659378e-05, "loss": 0.089, "step": 17692 }, { "epoch": 12.961904761904762, "grad_norm": 1.7109375, "learning_rate": 7.473090846183492e-05, "loss": 0.1954, "step": 17693 }, { "epoch": 12.962637362637363, "grad_norm": 0.93359375, "learning_rate": 7.47279206309778e-05, "loss": 0.1444, "step": 17694 }, { "epoch": 12.963369963369964, "grad_norm": 1.46875, "learning_rate": 7.472493268403676e-05, "loss": 0.1398, "step": 17695 }, { "epoch": 12.964102564102564, "grad_norm": 1.625, "learning_rate": 7.472194462102613e-05, "loss": 0.1655, "step": 17696 }, { "epoch": 12.964835164835165, "grad_norm": 1.328125, "learning_rate": 7.471895644196019e-05, "loss": 0.1213, "step": 17697 }, { "epoch": 12.965567765567766, "grad_norm": 1.7578125, "learning_rate": 7.47159681468533e-05, "loss": 0.1259, "step": 17698 }, { "epoch": 12.966300366300366, "grad_norm": 2.03125, "learning_rate": 7.471297973571974e-05, "loss": 0.1284, "step": 17699 }, { "epoch": 12.967032967032967, "grad_norm": 1.359375, "learning_rate": 7.470999120857387e-05, "loss": 0.1244, "step": 17700 }, { "epoch": 12.967765567765568, "grad_norm": 1.21875, "learning_rate": 7.470700256542998e-05, "loss": 0.1272, "step": 17701 }, { "epoch": 12.968498168498169, "grad_norm": 1.2890625, "learning_rate": 7.470401380630237e-05, "loss": 0.1199, "step": 17702 }, { "epoch": 12.96923076923077, "grad_norm": 1.5703125, "learning_rate": 7.47010249312054e-05, "loss": 0.1237, "step": 17703 }, { "epoch": 12.96996336996337, "grad_norm": 1.5625, "learning_rate": 7.469803594015337e-05, "loss": 0.1413, "step": 17704 }, { "epoch": 12.97069597069597, "grad_norm": 2.171875, "learning_rate": 7.469504683316062e-05, "loss": 0.1588, "step": 17705 }, { "epoch": 12.971428571428572, "grad_norm": 1.5, "learning_rate": 7.469205761024144e-05, "loss": 0.1311, "step": 17706 }, { "epoch": 12.972161172161172, "grad_norm": 1.3671875, "learning_rate": 7.468906827141016e-05, "loss": 0.1478, "step": 17707 }, { "epoch": 12.972893772893773, "grad_norm": 1.1484375, "learning_rate": 7.468607881668114e-05, "loss": 0.1207, "step": 17708 }, { "epoch": 12.973626373626374, "grad_norm": 1.03125, "learning_rate": 7.468308924606867e-05, "loss": 0.1171, "step": 17709 }, { "epoch": 12.974358974358974, "grad_norm": 0.82421875, "learning_rate": 7.468009955958708e-05, "loss": 0.109, "step": 17710 }, { "epoch": 12.975091575091575, "grad_norm": 0.91015625, "learning_rate": 7.467710975725071e-05, "loss": 0.1173, "step": 17711 }, { "epoch": 12.975824175824176, "grad_norm": 0.72265625, "learning_rate": 7.467411983907383e-05, "loss": 0.1231, "step": 17712 }, { "epoch": 12.976556776556777, "grad_norm": 1.046875, "learning_rate": 7.467112980507084e-05, "loss": 0.1148, "step": 17713 }, { "epoch": 12.977289377289377, "grad_norm": 1.0546875, "learning_rate": 7.466813965525603e-05, "loss": 0.1431, "step": 17714 }, { "epoch": 12.978021978021978, "grad_norm": 1.2109375, "learning_rate": 7.466514938964371e-05, "loss": 0.1455, "step": 17715 }, { "epoch": 12.978754578754579, "grad_norm": 1.15625, "learning_rate": 7.466215900824823e-05, "loss": 0.1309, "step": 17716 }, { "epoch": 12.97948717948718, "grad_norm": 0.8515625, "learning_rate": 7.465916851108392e-05, "loss": 0.1378, "step": 17717 }, { "epoch": 12.98021978021978, "grad_norm": 1.125, "learning_rate": 7.46561778981651e-05, "loss": 0.108, "step": 17718 }, { "epoch": 12.980952380952381, "grad_norm": 0.91015625, "learning_rate": 7.465318716950612e-05, "loss": 0.1132, "step": 17719 }, { "epoch": 12.981684981684982, "grad_norm": 1.6328125, "learning_rate": 7.465019632512127e-05, "loss": 0.1429, "step": 17720 }, { "epoch": 12.982417582417582, "grad_norm": 1.0234375, "learning_rate": 7.464720536502492e-05, "loss": 0.1281, "step": 17721 }, { "epoch": 12.983150183150183, "grad_norm": 1.6875, "learning_rate": 7.464421428923138e-05, "loss": 0.1528, "step": 17722 }, { "epoch": 12.983882783882784, "grad_norm": 1.28125, "learning_rate": 7.464122309775496e-05, "loss": 0.1186, "step": 17723 }, { "epoch": 12.984615384615385, "grad_norm": 0.921875, "learning_rate": 7.463823179061003e-05, "loss": 0.1235, "step": 17724 }, { "epoch": 12.985347985347985, "grad_norm": 0.6875, "learning_rate": 7.46352403678109e-05, "loss": 0.1114, "step": 17725 }, { "epoch": 12.986080586080586, "grad_norm": 0.9609375, "learning_rate": 7.463224882937192e-05, "loss": 0.104, "step": 17726 }, { "epoch": 12.986813186813187, "grad_norm": 0.96875, "learning_rate": 7.462925717530741e-05, "loss": 0.1367, "step": 17727 }, { "epoch": 12.987545787545788, "grad_norm": 1.5078125, "learning_rate": 7.462626540563172e-05, "loss": 0.1663, "step": 17728 }, { "epoch": 12.988278388278388, "grad_norm": 1.578125, "learning_rate": 7.462327352035916e-05, "loss": 0.1562, "step": 17729 }, { "epoch": 12.989010989010989, "grad_norm": 0.9609375, "learning_rate": 7.462028151950407e-05, "loss": 0.1227, "step": 17730 }, { "epoch": 12.98974358974359, "grad_norm": 1.140625, "learning_rate": 7.46172894030808e-05, "loss": 0.1365, "step": 17731 }, { "epoch": 12.99047619047619, "grad_norm": 0.796875, "learning_rate": 7.461429717110367e-05, "loss": 0.1058, "step": 17732 }, { "epoch": 12.991208791208791, "grad_norm": 1.15625, "learning_rate": 7.461130482358703e-05, "loss": 0.1432, "step": 17733 }, { "epoch": 12.991941391941392, "grad_norm": 0.91796875, "learning_rate": 7.460831236054523e-05, "loss": 0.1152, "step": 17734 }, { "epoch": 12.992673992673993, "grad_norm": 1.5390625, "learning_rate": 7.460531978199257e-05, "loss": 0.1545, "step": 17735 }, { "epoch": 12.993406593406593, "grad_norm": 1.515625, "learning_rate": 7.460232708794343e-05, "loss": 0.1202, "step": 17736 }, { "epoch": 12.994139194139194, "grad_norm": 1.3828125, "learning_rate": 7.459933427841213e-05, "loss": 0.1696, "step": 17737 }, { "epoch": 12.994871794871795, "grad_norm": 0.98046875, "learning_rate": 7.459634135341299e-05, "loss": 0.1512, "step": 17738 }, { "epoch": 12.995604395604396, "grad_norm": 1.5546875, "learning_rate": 7.459334831296038e-05, "loss": 0.1523, "step": 17739 }, { "epoch": 12.996336996336996, "grad_norm": 1.4296875, "learning_rate": 7.459035515706862e-05, "loss": 0.1188, "step": 17740 }, { "epoch": 12.997069597069597, "grad_norm": 2.28125, "learning_rate": 7.458736188575206e-05, "loss": 0.1393, "step": 17741 }, { "epoch": 12.997802197802198, "grad_norm": 1.5234375, "learning_rate": 7.458436849902505e-05, "loss": 0.1327, "step": 17742 }, { "epoch": 12.998534798534799, "grad_norm": 1.421875, "learning_rate": 7.458137499690192e-05, "loss": 0.1392, "step": 17743 }, { "epoch": 12.9992673992674, "grad_norm": 0.89453125, "learning_rate": 7.4578381379397e-05, "loss": 0.1236, "step": 17744 }, { "epoch": 13.0, "grad_norm": 0.8203125, "learning_rate": 7.45753876465247e-05, "loss": 0.0979, "step": 17745 }, { "epoch": 13.0007326007326, "grad_norm": 1.2734375, "learning_rate": 7.457239379829927e-05, "loss": 0.1296, "step": 17746 }, { "epoch": 13.001465201465201, "grad_norm": 1.03125, "learning_rate": 7.456939983473511e-05, "loss": 0.1384, "step": 17747 }, { "epoch": 13.002197802197802, "grad_norm": 0.73828125, "learning_rate": 7.456640575584656e-05, "loss": 0.1182, "step": 17748 }, { "epoch": 13.002930402930403, "grad_norm": 1.0546875, "learning_rate": 7.456341156164796e-05, "loss": 0.1111, "step": 17749 }, { "epoch": 13.003663003663004, "grad_norm": 0.80859375, "learning_rate": 7.456041725215364e-05, "loss": 0.142, "step": 17750 }, { "epoch": 13.004395604395604, "grad_norm": 0.5859375, "learning_rate": 7.455742282737798e-05, "loss": 0.1275, "step": 17751 }, { "epoch": 13.005128205128205, "grad_norm": 0.88671875, "learning_rate": 7.45544282873353e-05, "loss": 0.107, "step": 17752 }, { "epoch": 13.005860805860806, "grad_norm": 0.96484375, "learning_rate": 7.455143363203996e-05, "loss": 0.1344, "step": 17753 }, { "epoch": 13.006593406593407, "grad_norm": 1.75, "learning_rate": 7.45484388615063e-05, "loss": 0.1121, "step": 17754 }, { "epoch": 13.007326007326007, "grad_norm": 1.03125, "learning_rate": 7.45454439757487e-05, "loss": 0.1132, "step": 17755 }, { "epoch": 13.008058608058608, "grad_norm": 1.328125, "learning_rate": 7.454244897478146e-05, "loss": 0.137, "step": 17756 }, { "epoch": 13.008791208791209, "grad_norm": 1.1640625, "learning_rate": 7.453945385861894e-05, "loss": 0.1461, "step": 17757 }, { "epoch": 13.00952380952381, "grad_norm": 0.765625, "learning_rate": 7.453645862727553e-05, "loss": 0.1015, "step": 17758 }, { "epoch": 13.01025641025641, "grad_norm": 1.1640625, "learning_rate": 7.453346328076554e-05, "loss": 0.1549, "step": 17759 }, { "epoch": 13.010989010989011, "grad_norm": 0.94921875, "learning_rate": 7.453046781910335e-05, "loss": 0.148, "step": 17760 }, { "epoch": 13.011721611721612, "grad_norm": 0.765625, "learning_rate": 7.452747224230331e-05, "loss": 0.1138, "step": 17761 }, { "epoch": 13.012454212454212, "grad_norm": 1.5703125, "learning_rate": 7.452447655037976e-05, "loss": 0.146, "step": 17762 }, { "epoch": 13.013186813186813, "grad_norm": 1.390625, "learning_rate": 7.452148074334704e-05, "loss": 0.1606, "step": 17763 }, { "epoch": 13.013919413919414, "grad_norm": 1.2578125, "learning_rate": 7.451848482121951e-05, "loss": 0.1466, "step": 17764 }, { "epoch": 13.014652014652015, "grad_norm": 0.546875, "learning_rate": 7.451548878401157e-05, "loss": 0.1161, "step": 17765 }, { "epoch": 13.015384615384615, "grad_norm": 1.0703125, "learning_rate": 7.451249263173753e-05, "loss": 0.127, "step": 17766 }, { "epoch": 13.016117216117216, "grad_norm": 1.328125, "learning_rate": 7.450949636441173e-05, "loss": 0.1445, "step": 17767 }, { "epoch": 13.016849816849817, "grad_norm": 1.1640625, "learning_rate": 7.450649998204857e-05, "loss": 0.1316, "step": 17768 }, { "epoch": 13.017582417582418, "grad_norm": 1.1484375, "learning_rate": 7.450350348466238e-05, "loss": 0.1339, "step": 17769 }, { "epoch": 13.018315018315018, "grad_norm": 1.0390625, "learning_rate": 7.450050687226753e-05, "loss": 0.1277, "step": 17770 }, { "epoch": 13.019047619047619, "grad_norm": 0.8671875, "learning_rate": 7.449751014487837e-05, "loss": 0.1173, "step": 17771 }, { "epoch": 13.01978021978022, "grad_norm": 0.796875, "learning_rate": 7.449451330250926e-05, "loss": 0.089, "step": 17772 }, { "epoch": 13.02051282051282, "grad_norm": 0.9453125, "learning_rate": 7.449151634517459e-05, "loss": 0.1304, "step": 17773 }, { "epoch": 13.021245421245421, "grad_norm": 0.60546875, "learning_rate": 7.448851927288865e-05, "loss": 0.0962, "step": 17774 }, { "epoch": 13.021978021978022, "grad_norm": 1.1953125, "learning_rate": 7.448552208566585e-05, "loss": 0.1594, "step": 17775 }, { "epoch": 13.022710622710623, "grad_norm": 1.03125, "learning_rate": 7.448252478352055e-05, "loss": 0.1302, "step": 17776 }, { "epoch": 13.023443223443223, "grad_norm": 1.2109375, "learning_rate": 7.447952736646708e-05, "loss": 0.1338, "step": 17777 }, { "epoch": 13.024175824175824, "grad_norm": 1.6015625, "learning_rate": 7.447652983451984e-05, "loss": 0.1437, "step": 17778 }, { "epoch": 13.024908424908425, "grad_norm": 0.80078125, "learning_rate": 7.447353218769317e-05, "loss": 0.1064, "step": 17779 }, { "epoch": 13.025641025641026, "grad_norm": 0.76171875, "learning_rate": 7.447053442600144e-05, "loss": 0.1151, "step": 17780 }, { "epoch": 13.026373626373626, "grad_norm": 1.0078125, "learning_rate": 7.446753654945901e-05, "loss": 0.1394, "step": 17781 }, { "epoch": 13.027106227106227, "grad_norm": 0.9140625, "learning_rate": 7.446453855808025e-05, "loss": 0.1547, "step": 17782 }, { "epoch": 13.027838827838828, "grad_norm": 1.046875, "learning_rate": 7.446154045187949e-05, "loss": 0.1255, "step": 17783 }, { "epoch": 13.028571428571428, "grad_norm": 0.8828125, "learning_rate": 7.445854223087115e-05, "loss": 0.1232, "step": 17784 }, { "epoch": 13.02930402930403, "grad_norm": 1.15625, "learning_rate": 7.445554389506956e-05, "loss": 0.1095, "step": 17785 }, { "epoch": 13.03003663003663, "grad_norm": 1.03125, "learning_rate": 7.44525454444891e-05, "loss": 0.1171, "step": 17786 }, { "epoch": 13.03076923076923, "grad_norm": 1.03125, "learning_rate": 7.444954687914413e-05, "loss": 0.1115, "step": 17787 }, { "epoch": 13.031501831501831, "grad_norm": 0.8984375, "learning_rate": 7.444654819904902e-05, "loss": 0.1016, "step": 17788 }, { "epoch": 13.032234432234432, "grad_norm": 1.453125, "learning_rate": 7.444354940421814e-05, "loss": 0.1354, "step": 17789 }, { "epoch": 13.032967032967033, "grad_norm": 1.1015625, "learning_rate": 7.444055049466586e-05, "loss": 0.1062, "step": 17790 }, { "epoch": 13.033699633699634, "grad_norm": 1.046875, "learning_rate": 7.443755147040652e-05, "loss": 0.141, "step": 17791 }, { "epoch": 13.034432234432234, "grad_norm": 0.71484375, "learning_rate": 7.443455233145454e-05, "loss": 0.1276, "step": 17792 }, { "epoch": 13.035164835164835, "grad_norm": 0.97265625, "learning_rate": 7.443155307782424e-05, "loss": 0.1246, "step": 17793 }, { "epoch": 13.035897435897436, "grad_norm": 0.9609375, "learning_rate": 7.442855370953002e-05, "loss": 0.1544, "step": 17794 }, { "epoch": 13.036630036630036, "grad_norm": 1.2734375, "learning_rate": 7.442555422658623e-05, "loss": 0.0982, "step": 17795 }, { "epoch": 13.037362637362637, "grad_norm": 1.921875, "learning_rate": 7.442255462900729e-05, "loss": 0.1559, "step": 17796 }, { "epoch": 13.038095238095238, "grad_norm": 0.859375, "learning_rate": 7.441955491680753e-05, "loss": 0.1036, "step": 17797 }, { "epoch": 13.038827838827839, "grad_norm": 1.3125, "learning_rate": 7.441655509000131e-05, "loss": 0.1714, "step": 17798 }, { "epoch": 13.03956043956044, "grad_norm": 1.234375, "learning_rate": 7.441355514860303e-05, "loss": 0.1205, "step": 17799 }, { "epoch": 13.04029304029304, "grad_norm": 0.6484375, "learning_rate": 7.441055509262706e-05, "loss": 0.1172, "step": 17800 }, { "epoch": 13.04102564102564, "grad_norm": 1.2578125, "learning_rate": 7.440755492208777e-05, "loss": 0.1518, "step": 17801 }, { "epoch": 13.041758241758242, "grad_norm": 1.0625, "learning_rate": 7.440455463699954e-05, "loss": 0.1309, "step": 17802 }, { "epoch": 13.042490842490842, "grad_norm": 1.53125, "learning_rate": 7.440155423737673e-05, "loss": 0.1057, "step": 17803 }, { "epoch": 13.043223443223443, "grad_norm": 0.96484375, "learning_rate": 7.439855372323373e-05, "loss": 0.1048, "step": 17804 }, { "epoch": 13.043956043956044, "grad_norm": 1.1015625, "learning_rate": 7.439555309458492e-05, "loss": 0.1237, "step": 17805 }, { "epoch": 13.044688644688645, "grad_norm": 1.140625, "learning_rate": 7.439255235144468e-05, "loss": 0.1471, "step": 17806 }, { "epoch": 13.045421245421245, "grad_norm": 0.76171875, "learning_rate": 7.438955149382735e-05, "loss": 0.1044, "step": 17807 }, { "epoch": 13.046153846153846, "grad_norm": 1.0546875, "learning_rate": 7.438655052174737e-05, "loss": 0.128, "step": 17808 }, { "epoch": 13.046886446886447, "grad_norm": 1.265625, "learning_rate": 7.438354943521905e-05, "loss": 0.1504, "step": 17809 }, { "epoch": 13.047619047619047, "grad_norm": 1.125, "learning_rate": 7.438054823425684e-05, "loss": 0.1067, "step": 17810 }, { "epoch": 13.048351648351648, "grad_norm": 0.99609375, "learning_rate": 7.437754691887505e-05, "loss": 0.1245, "step": 17811 }, { "epoch": 13.049084249084249, "grad_norm": 1.21875, "learning_rate": 7.437454548908812e-05, "loss": 0.1171, "step": 17812 }, { "epoch": 13.04981684981685, "grad_norm": 1.0390625, "learning_rate": 7.437154394491039e-05, "loss": 0.1202, "step": 17813 }, { "epoch": 13.05054945054945, "grad_norm": 0.8125, "learning_rate": 7.436854228635627e-05, "loss": 0.1099, "step": 17814 }, { "epoch": 13.051282051282051, "grad_norm": 0.91015625, "learning_rate": 7.436554051344013e-05, "loss": 0.1087, "step": 17815 }, { "epoch": 13.052014652014652, "grad_norm": 0.90234375, "learning_rate": 7.436253862617634e-05, "loss": 0.1231, "step": 17816 }, { "epoch": 13.052747252747253, "grad_norm": 1.40625, "learning_rate": 7.43595366245793e-05, "loss": 0.1022, "step": 17817 }, { "epoch": 13.053479853479853, "grad_norm": 0.98046875, "learning_rate": 7.43565345086634e-05, "loss": 0.1227, "step": 17818 }, { "epoch": 13.054212454212454, "grad_norm": 0.7734375, "learning_rate": 7.435353227844297e-05, "loss": 0.1206, "step": 17819 }, { "epoch": 13.054945054945055, "grad_norm": 1.46875, "learning_rate": 7.435052993393246e-05, "loss": 0.1617, "step": 17820 }, { "epoch": 13.055677655677655, "grad_norm": 0.90625, "learning_rate": 7.434752747514624e-05, "loss": 0.1076, "step": 17821 }, { "epoch": 13.056410256410256, "grad_norm": 1.21875, "learning_rate": 7.434452490209867e-05, "loss": 0.1452, "step": 17822 }, { "epoch": 13.057142857142857, "grad_norm": 0.90625, "learning_rate": 7.434152221480417e-05, "loss": 0.1311, "step": 17823 }, { "epoch": 13.057875457875458, "grad_norm": 1.609375, "learning_rate": 7.43385194132771e-05, "loss": 0.0995, "step": 17824 }, { "epoch": 13.058608058608058, "grad_norm": 0.9296875, "learning_rate": 7.433551649753186e-05, "loss": 0.1159, "step": 17825 }, { "epoch": 13.05934065934066, "grad_norm": 0.8671875, "learning_rate": 7.433251346758286e-05, "loss": 0.1037, "step": 17826 }, { "epoch": 13.06007326007326, "grad_norm": 1.28125, "learning_rate": 7.432951032344442e-05, "loss": 0.1592, "step": 17827 }, { "epoch": 13.06080586080586, "grad_norm": 1.2109375, "learning_rate": 7.4326507065131e-05, "loss": 0.1406, "step": 17828 }, { "epoch": 13.061538461538461, "grad_norm": 0.94140625, "learning_rate": 7.432350369265695e-05, "loss": 0.1361, "step": 17829 }, { "epoch": 13.062271062271062, "grad_norm": 1.0078125, "learning_rate": 7.432050020603669e-05, "loss": 0.1132, "step": 17830 }, { "epoch": 13.063003663003663, "grad_norm": 1.0234375, "learning_rate": 7.431749660528459e-05, "loss": 0.1627, "step": 17831 }, { "epoch": 13.063736263736264, "grad_norm": 0.984375, "learning_rate": 7.431449289041503e-05, "loss": 0.1283, "step": 17832 }, { "epoch": 13.064468864468864, "grad_norm": 1.125, "learning_rate": 7.431148906144243e-05, "loss": 0.1183, "step": 17833 }, { "epoch": 13.065201465201465, "grad_norm": 1.5234375, "learning_rate": 7.430848511838117e-05, "loss": 0.1633, "step": 17834 }, { "epoch": 13.065934065934066, "grad_norm": 1.5234375, "learning_rate": 7.430548106124564e-05, "loss": 0.1133, "step": 17835 }, { "epoch": 13.066666666666666, "grad_norm": 1.34375, "learning_rate": 7.430247689005022e-05, "loss": 0.1326, "step": 17836 }, { "epoch": 13.067399267399267, "grad_norm": 0.6484375, "learning_rate": 7.429947260480933e-05, "loss": 0.0982, "step": 17837 }, { "epoch": 13.068131868131868, "grad_norm": 0.98828125, "learning_rate": 7.429646820553738e-05, "loss": 0.1013, "step": 17838 }, { "epoch": 13.068864468864469, "grad_norm": 1.09375, "learning_rate": 7.42934636922487e-05, "loss": 0.1202, "step": 17839 }, { "epoch": 13.06959706959707, "grad_norm": 1.1015625, "learning_rate": 7.429045906495774e-05, "loss": 0.1076, "step": 17840 }, { "epoch": 13.07032967032967, "grad_norm": 0.96875, "learning_rate": 7.428745432367889e-05, "loss": 0.1361, "step": 17841 }, { "epoch": 13.07106227106227, "grad_norm": 1.2734375, "learning_rate": 7.428444946842652e-05, "loss": 0.1476, "step": 17842 }, { "epoch": 13.071794871794872, "grad_norm": 1.0390625, "learning_rate": 7.428144449921505e-05, "loss": 0.1318, "step": 17843 }, { "epoch": 13.072527472527472, "grad_norm": 1.609375, "learning_rate": 7.427843941605887e-05, "loss": 0.1871, "step": 17844 }, { "epoch": 13.073260073260073, "grad_norm": 0.875, "learning_rate": 7.42754342189724e-05, "loss": 0.1381, "step": 17845 }, { "epoch": 13.073992673992674, "grad_norm": 0.97265625, "learning_rate": 7.427242890797e-05, "loss": 0.1325, "step": 17846 }, { "epoch": 13.074725274725274, "grad_norm": 0.9765625, "learning_rate": 7.42694234830661e-05, "loss": 0.1231, "step": 17847 }, { "epoch": 13.075457875457875, "grad_norm": 1.1640625, "learning_rate": 7.426641794427506e-05, "loss": 0.1567, "step": 17848 }, { "epoch": 13.076190476190476, "grad_norm": 1.1484375, "learning_rate": 7.426341229161135e-05, "loss": 0.1236, "step": 17849 }, { "epoch": 13.076923076923077, "grad_norm": 1.25, "learning_rate": 7.426040652508932e-05, "loss": 0.1701, "step": 17850 }, { "epoch": 13.077655677655677, "grad_norm": 1.125, "learning_rate": 7.425740064472337e-05, "loss": 0.1556, "step": 17851 }, { "epoch": 13.078388278388278, "grad_norm": 0.609375, "learning_rate": 7.425439465052793e-05, "loss": 0.1008, "step": 17852 }, { "epoch": 13.079120879120879, "grad_norm": 0.75390625, "learning_rate": 7.425138854251736e-05, "loss": 0.1113, "step": 17853 }, { "epoch": 13.07985347985348, "grad_norm": 1.2265625, "learning_rate": 7.424838232070612e-05, "loss": 0.1344, "step": 17854 }, { "epoch": 13.08058608058608, "grad_norm": 0.86328125, "learning_rate": 7.424537598510857e-05, "loss": 0.1133, "step": 17855 }, { "epoch": 13.081318681318681, "grad_norm": 0.8046875, "learning_rate": 7.424236953573914e-05, "loss": 0.1168, "step": 17856 }, { "epoch": 13.082051282051282, "grad_norm": 0.9140625, "learning_rate": 7.423936297261222e-05, "loss": 0.1062, "step": 17857 }, { "epoch": 13.082783882783882, "grad_norm": 0.71875, "learning_rate": 7.42363562957422e-05, "loss": 0.1131, "step": 17858 }, { "epoch": 13.083516483516483, "grad_norm": 1.125, "learning_rate": 7.423334950514351e-05, "loss": 0.1227, "step": 17859 }, { "epoch": 13.084249084249084, "grad_norm": 1.0390625, "learning_rate": 7.423034260083057e-05, "loss": 0.1271, "step": 17860 }, { "epoch": 13.084981684981685, "grad_norm": 1.3046875, "learning_rate": 7.422733558281775e-05, "loss": 0.154, "step": 17861 }, { "epoch": 13.085714285714285, "grad_norm": 1.34375, "learning_rate": 7.422432845111948e-05, "loss": 0.1239, "step": 17862 }, { "epoch": 13.086446886446886, "grad_norm": 1.375, "learning_rate": 7.422132120575017e-05, "loss": 0.1571, "step": 17863 }, { "epoch": 13.087179487179487, "grad_norm": 1.0390625, "learning_rate": 7.421831384672423e-05, "loss": 0.1453, "step": 17864 }, { "epoch": 13.087912087912088, "grad_norm": 1.765625, "learning_rate": 7.421530637405606e-05, "loss": 0.1314, "step": 17865 }, { "epoch": 13.088644688644688, "grad_norm": 1.6015625, "learning_rate": 7.421229878776006e-05, "loss": 0.0944, "step": 17866 }, { "epoch": 13.089377289377289, "grad_norm": 1.34375, "learning_rate": 7.420929108785066e-05, "loss": 0.1133, "step": 17867 }, { "epoch": 13.09010989010989, "grad_norm": 1.21875, "learning_rate": 7.420628327434226e-05, "loss": 0.1203, "step": 17868 }, { "epoch": 13.09084249084249, "grad_norm": 1.8046875, "learning_rate": 7.420327534724928e-05, "loss": 0.149, "step": 17869 }, { "epoch": 13.091575091575091, "grad_norm": 0.81640625, "learning_rate": 7.420026730658613e-05, "loss": 0.1388, "step": 17870 }, { "epoch": 13.092307692307692, "grad_norm": 1.5078125, "learning_rate": 7.41972591523672e-05, "loss": 0.1401, "step": 17871 }, { "epoch": 13.093040293040293, "grad_norm": 1.1875, "learning_rate": 7.419425088460693e-05, "loss": 0.1508, "step": 17872 }, { "epoch": 13.093772893772893, "grad_norm": 1.3984375, "learning_rate": 7.419124250331973e-05, "loss": 0.1251, "step": 17873 }, { "epoch": 13.094505494505494, "grad_norm": 1.40625, "learning_rate": 7.418823400852e-05, "loss": 0.1044, "step": 17874 }, { "epoch": 13.095238095238095, "grad_norm": 0.921875, "learning_rate": 7.418522540022219e-05, "loss": 0.1274, "step": 17875 }, { "epoch": 13.095970695970696, "grad_norm": 1.375, "learning_rate": 7.418221667844067e-05, "loss": 0.1345, "step": 17876 }, { "epoch": 13.096703296703296, "grad_norm": 0.94921875, "learning_rate": 7.417920784318988e-05, "loss": 0.1151, "step": 17877 }, { "epoch": 13.097435897435897, "grad_norm": 1.2109375, "learning_rate": 7.417619889448423e-05, "loss": 0.1382, "step": 17878 }, { "epoch": 13.098168498168498, "grad_norm": 1.265625, "learning_rate": 7.417318983233813e-05, "loss": 0.1394, "step": 17879 }, { "epoch": 13.098901098901099, "grad_norm": 1.578125, "learning_rate": 7.417018065676603e-05, "loss": 0.1747, "step": 17880 }, { "epoch": 13.0996336996337, "grad_norm": 1.21875, "learning_rate": 7.41671713677823e-05, "loss": 0.1134, "step": 17881 }, { "epoch": 13.1003663003663, "grad_norm": 0.82421875, "learning_rate": 7.416416196540141e-05, "loss": 0.1209, "step": 17882 }, { "epoch": 13.1010989010989, "grad_norm": 1.6484375, "learning_rate": 7.416115244963772e-05, "loss": 0.1399, "step": 17883 }, { "epoch": 13.101831501831501, "grad_norm": 1.5546875, "learning_rate": 7.415814282050571e-05, "loss": 0.1087, "step": 17884 }, { "epoch": 13.102564102564102, "grad_norm": 1.25, "learning_rate": 7.415513307801974e-05, "loss": 0.1117, "step": 17885 }, { "epoch": 13.103296703296703, "grad_norm": 1.8125, "learning_rate": 7.415212322219427e-05, "loss": 0.1195, "step": 17886 }, { "epoch": 13.104029304029304, "grad_norm": 1.2578125, "learning_rate": 7.414911325304373e-05, "loss": 0.1448, "step": 17887 }, { "epoch": 13.104761904761904, "grad_norm": 1.03125, "learning_rate": 7.414610317058251e-05, "loss": 0.1497, "step": 17888 }, { "epoch": 13.105494505494505, "grad_norm": 1.1796875, "learning_rate": 7.414309297482505e-05, "loss": 0.122, "step": 17889 }, { "epoch": 13.106227106227106, "grad_norm": 1.2578125, "learning_rate": 7.414008266578577e-05, "loss": 0.1333, "step": 17890 }, { "epoch": 13.106959706959707, "grad_norm": 1.03125, "learning_rate": 7.413707224347913e-05, "loss": 0.1279, "step": 17891 }, { "epoch": 13.107692307692307, "grad_norm": 0.93359375, "learning_rate": 7.413406170791946e-05, "loss": 0.1229, "step": 17892 }, { "epoch": 13.108424908424908, "grad_norm": 1.3125, "learning_rate": 7.413105105912127e-05, "loss": 0.1244, "step": 17893 }, { "epoch": 13.109157509157509, "grad_norm": 1.34375, "learning_rate": 7.412804029709896e-05, "loss": 0.1676, "step": 17894 }, { "epoch": 13.10989010989011, "grad_norm": 1.453125, "learning_rate": 7.412502942186693e-05, "loss": 0.1166, "step": 17895 }, { "epoch": 13.11062271062271, "grad_norm": 1.7109375, "learning_rate": 7.412201843343965e-05, "loss": 0.1329, "step": 17896 }, { "epoch": 13.111355311355311, "grad_norm": 1.1953125, "learning_rate": 7.411900733183153e-05, "loss": 0.1033, "step": 17897 }, { "epoch": 13.112087912087912, "grad_norm": 1.5234375, "learning_rate": 7.411599611705697e-05, "loss": 0.1148, "step": 17898 }, { "epoch": 13.112820512820512, "grad_norm": 0.72265625, "learning_rate": 7.411298478913042e-05, "loss": 0.1169, "step": 17899 }, { "epoch": 13.113553113553113, "grad_norm": 0.5703125, "learning_rate": 7.410997334806632e-05, "loss": 0.1171, "step": 17900 }, { "epoch": 13.114285714285714, "grad_norm": 0.66796875, "learning_rate": 7.410696179387909e-05, "loss": 0.1417, "step": 17901 }, { "epoch": 13.115018315018315, "grad_norm": 0.84375, "learning_rate": 7.410395012658314e-05, "loss": 0.1166, "step": 17902 }, { "epoch": 13.115750915750915, "grad_norm": 1.1171875, "learning_rate": 7.410093834619294e-05, "loss": 0.1196, "step": 17903 }, { "epoch": 13.116483516483516, "grad_norm": 0.87890625, "learning_rate": 7.40979264527229e-05, "loss": 0.1197, "step": 17904 }, { "epoch": 13.117216117216117, "grad_norm": 0.73828125, "learning_rate": 7.409491444618742e-05, "loss": 0.1235, "step": 17905 }, { "epoch": 13.117948717948718, "grad_norm": 0.80859375, "learning_rate": 7.409190232660098e-05, "loss": 0.1232, "step": 17906 }, { "epoch": 13.118681318681318, "grad_norm": 1.0234375, "learning_rate": 7.408889009397797e-05, "loss": 0.1259, "step": 17907 }, { "epoch": 13.119413919413919, "grad_norm": 1.078125, "learning_rate": 7.408587774833287e-05, "loss": 0.1283, "step": 17908 }, { "epoch": 13.12014652014652, "grad_norm": 0.828125, "learning_rate": 7.408286528968008e-05, "loss": 0.1258, "step": 17909 }, { "epoch": 13.12087912087912, "grad_norm": 1.2890625, "learning_rate": 7.407985271803402e-05, "loss": 0.1445, "step": 17910 }, { "epoch": 13.121611721611721, "grad_norm": 1.2265625, "learning_rate": 7.407684003340915e-05, "loss": 0.1153, "step": 17911 }, { "epoch": 13.122344322344322, "grad_norm": 0.87109375, "learning_rate": 7.407382723581993e-05, "loss": 0.1027, "step": 17912 }, { "epoch": 13.123076923076923, "grad_norm": 0.92578125, "learning_rate": 7.407081432528074e-05, "loss": 0.1305, "step": 17913 }, { "epoch": 13.123809523809523, "grad_norm": 0.77734375, "learning_rate": 7.406780130180604e-05, "loss": 0.1123, "step": 17914 }, { "epoch": 13.124542124542124, "grad_norm": 1.3515625, "learning_rate": 7.406478816541027e-05, "loss": 0.1109, "step": 17915 }, { "epoch": 13.125274725274725, "grad_norm": 1.421875, "learning_rate": 7.406177491610788e-05, "loss": 0.164, "step": 17916 }, { "epoch": 13.126007326007326, "grad_norm": 1.296875, "learning_rate": 7.405876155391328e-05, "loss": 0.1533, "step": 17917 }, { "epoch": 13.126739926739926, "grad_norm": 1.2734375, "learning_rate": 7.405574807884092e-05, "loss": 0.1602, "step": 17918 }, { "epoch": 13.127472527472527, "grad_norm": 1.046875, "learning_rate": 7.405273449090522e-05, "loss": 0.1184, "step": 17919 }, { "epoch": 13.128205128205128, "grad_norm": 1.046875, "learning_rate": 7.404972079012066e-05, "loss": 0.1216, "step": 17920 }, { "epoch": 13.128937728937728, "grad_norm": 1.09375, "learning_rate": 7.404670697650165e-05, "loss": 0.1279, "step": 17921 }, { "epoch": 13.12967032967033, "grad_norm": 1.6953125, "learning_rate": 7.404369305006265e-05, "loss": 0.1662, "step": 17922 }, { "epoch": 13.13040293040293, "grad_norm": 1.4453125, "learning_rate": 7.404067901081806e-05, "loss": 0.1438, "step": 17923 }, { "epoch": 13.13113553113553, "grad_norm": 1.0390625, "learning_rate": 7.403766485878235e-05, "loss": 0.1027, "step": 17924 }, { "epoch": 13.131868131868131, "grad_norm": 0.859375, "learning_rate": 7.403465059396999e-05, "loss": 0.1175, "step": 17925 }, { "epoch": 13.132600732600732, "grad_norm": 1.40625, "learning_rate": 7.403163621639537e-05, "loss": 0.128, "step": 17926 }, { "epoch": 13.133333333333333, "grad_norm": 1.203125, "learning_rate": 7.402862172607296e-05, "loss": 0.1134, "step": 17927 }, { "epoch": 13.134065934065934, "grad_norm": 1.046875, "learning_rate": 7.40256071230172e-05, "loss": 0.1256, "step": 17928 }, { "epoch": 13.134798534798534, "grad_norm": 1.0703125, "learning_rate": 7.402259240724254e-05, "loss": 0.1478, "step": 17929 }, { "epoch": 13.135531135531135, "grad_norm": 1.140625, "learning_rate": 7.401957757876342e-05, "loss": 0.1251, "step": 17930 }, { "epoch": 13.136263736263736, "grad_norm": 1.65625, "learning_rate": 7.401656263759427e-05, "loss": 0.138, "step": 17931 }, { "epoch": 13.136996336996337, "grad_norm": 1.4921875, "learning_rate": 7.401354758374957e-05, "loss": 0.1223, "step": 17932 }, { "epoch": 13.137728937728937, "grad_norm": 0.97265625, "learning_rate": 7.401053241724372e-05, "loss": 0.1227, "step": 17933 }, { "epoch": 13.138461538461538, "grad_norm": 1.15625, "learning_rate": 7.40075171380912e-05, "loss": 0.105, "step": 17934 }, { "epoch": 13.139194139194139, "grad_norm": 0.79296875, "learning_rate": 7.400450174630645e-05, "loss": 0.1252, "step": 17935 }, { "epoch": 13.13992673992674, "grad_norm": 1.0703125, "learning_rate": 7.400148624190392e-05, "loss": 0.1156, "step": 17936 }, { "epoch": 13.14065934065934, "grad_norm": 0.96875, "learning_rate": 7.399847062489803e-05, "loss": 0.107, "step": 17937 }, { "epoch": 13.141391941391941, "grad_norm": 1.3203125, "learning_rate": 7.399545489530328e-05, "loss": 0.1333, "step": 17938 }, { "epoch": 13.142124542124542, "grad_norm": 0.73828125, "learning_rate": 7.399243905313409e-05, "loss": 0.1041, "step": 17939 }, { "epoch": 13.142857142857142, "grad_norm": 0.68359375, "learning_rate": 7.39894230984049e-05, "loss": 0.1311, "step": 17940 }, { "epoch": 13.143589743589743, "grad_norm": 1.4609375, "learning_rate": 7.398640703113017e-05, "loss": 0.1483, "step": 17941 }, { "epoch": 13.144322344322344, "grad_norm": 0.78515625, "learning_rate": 7.398339085132437e-05, "loss": 0.1088, "step": 17942 }, { "epoch": 13.145054945054945, "grad_norm": 1.3359375, "learning_rate": 7.398037455900196e-05, "loss": 0.1312, "step": 17943 }, { "epoch": 13.145787545787545, "grad_norm": 1.1640625, "learning_rate": 7.397735815417732e-05, "loss": 0.1333, "step": 17944 }, { "epoch": 13.146520146520146, "grad_norm": 0.953125, "learning_rate": 7.397434163686496e-05, "loss": 0.1554, "step": 17945 }, { "epoch": 13.147252747252747, "grad_norm": 0.953125, "learning_rate": 7.397132500707934e-05, "loss": 0.091, "step": 17946 }, { "epoch": 13.147985347985347, "grad_norm": 1.109375, "learning_rate": 7.396830826483489e-05, "loss": 0.1389, "step": 17947 }, { "epoch": 13.148717948717948, "grad_norm": 1.0859375, "learning_rate": 7.396529141014607e-05, "loss": 0.128, "step": 17948 }, { "epoch": 13.149450549450549, "grad_norm": 0.6875, "learning_rate": 7.396227444302733e-05, "loss": 0.1277, "step": 17949 }, { "epoch": 13.15018315018315, "grad_norm": 1.046875, "learning_rate": 7.395925736349313e-05, "loss": 0.1057, "step": 17950 }, { "epoch": 13.15091575091575, "grad_norm": 1.2578125, "learning_rate": 7.395624017155793e-05, "loss": 0.1271, "step": 17951 }, { "epoch": 13.151648351648351, "grad_norm": 0.94921875, "learning_rate": 7.395322286723617e-05, "loss": 0.1326, "step": 17952 }, { "epoch": 13.152380952380952, "grad_norm": 1.140625, "learning_rate": 7.395020545054234e-05, "loss": 0.1257, "step": 17953 }, { "epoch": 13.153113553113553, "grad_norm": 1.3515625, "learning_rate": 7.394718792149087e-05, "loss": 0.1348, "step": 17954 }, { "epoch": 13.153846153846153, "grad_norm": 1.1640625, "learning_rate": 7.394417028009624e-05, "loss": 0.1429, "step": 17955 }, { "epoch": 13.154578754578754, "grad_norm": 0.625, "learning_rate": 7.394115252637288e-05, "loss": 0.1064, "step": 17956 }, { "epoch": 13.155311355311355, "grad_norm": 1.1484375, "learning_rate": 7.393813466033527e-05, "loss": 0.1215, "step": 17957 }, { "epoch": 13.156043956043955, "grad_norm": 1.5703125, "learning_rate": 7.393511668199786e-05, "loss": 0.1312, "step": 17958 }, { "epoch": 13.156776556776556, "grad_norm": 1.5234375, "learning_rate": 7.39320985913751e-05, "loss": 0.1402, "step": 17959 }, { "epoch": 13.157509157509157, "grad_norm": 1.2109375, "learning_rate": 7.392908038848148e-05, "loss": 0.1528, "step": 17960 }, { "epoch": 13.158241758241758, "grad_norm": 1.25, "learning_rate": 7.392606207333143e-05, "loss": 0.1586, "step": 17961 }, { "epoch": 13.158974358974358, "grad_norm": 1.515625, "learning_rate": 7.392304364593943e-05, "loss": 0.1389, "step": 17962 }, { "epoch": 13.15970695970696, "grad_norm": 1.984375, "learning_rate": 7.392002510631995e-05, "loss": 0.1223, "step": 17963 }, { "epoch": 13.16043956043956, "grad_norm": 0.859375, "learning_rate": 7.391700645448742e-05, "loss": 0.1028, "step": 17964 }, { "epoch": 13.16117216117216, "grad_norm": 1.0859375, "learning_rate": 7.391398769045634e-05, "loss": 0.1292, "step": 17965 }, { "epoch": 13.161904761904761, "grad_norm": 1.265625, "learning_rate": 7.391096881424117e-05, "loss": 0.1595, "step": 17966 }, { "epoch": 13.162637362637362, "grad_norm": 0.89453125, "learning_rate": 7.390794982585635e-05, "loss": 0.1486, "step": 17967 }, { "epoch": 13.163369963369963, "grad_norm": 1.2890625, "learning_rate": 7.390493072531637e-05, "loss": 0.1332, "step": 17968 }, { "epoch": 13.164102564102564, "grad_norm": 1.9765625, "learning_rate": 7.390191151263566e-05, "loss": 0.1472, "step": 17969 }, { "epoch": 13.164835164835164, "grad_norm": 1.3515625, "learning_rate": 7.389889218782872e-05, "loss": 0.1382, "step": 17970 }, { "epoch": 13.165567765567765, "grad_norm": 1.4921875, "learning_rate": 7.389587275091002e-05, "loss": 0.1365, "step": 17971 }, { "epoch": 13.166300366300366, "grad_norm": 1.546875, "learning_rate": 7.389285320189399e-05, "loss": 0.123, "step": 17972 }, { "epoch": 13.167032967032966, "grad_norm": 1.265625, "learning_rate": 7.388983354079514e-05, "loss": 0.1553, "step": 17973 }, { "epoch": 13.167765567765567, "grad_norm": 0.83984375, "learning_rate": 7.38868137676279e-05, "loss": 0.091, "step": 17974 }, { "epoch": 13.168498168498168, "grad_norm": 0.59375, "learning_rate": 7.388379388240678e-05, "loss": 0.1081, "step": 17975 }, { "epoch": 13.169230769230769, "grad_norm": 0.91796875, "learning_rate": 7.388077388514621e-05, "loss": 0.1129, "step": 17976 }, { "epoch": 13.16996336996337, "grad_norm": 1.34375, "learning_rate": 7.38777537758607e-05, "loss": 0.1328, "step": 17977 }, { "epoch": 13.17069597069597, "grad_norm": 1.0546875, "learning_rate": 7.387473355456467e-05, "loss": 0.1208, "step": 17978 }, { "epoch": 13.17142857142857, "grad_norm": 0.79296875, "learning_rate": 7.387171322127263e-05, "loss": 0.1395, "step": 17979 }, { "epoch": 13.172161172161172, "grad_norm": 0.87109375, "learning_rate": 7.386869277599902e-05, "loss": 0.1326, "step": 17980 }, { "epoch": 13.172893772893772, "grad_norm": 1.328125, "learning_rate": 7.386567221875836e-05, "loss": 0.1313, "step": 17981 }, { "epoch": 13.173626373626373, "grad_norm": 1.890625, "learning_rate": 7.386265154956509e-05, "loss": 0.1421, "step": 17982 }, { "epoch": 13.174358974358974, "grad_norm": 2.734375, "learning_rate": 7.385963076843367e-05, "loss": 0.1384, "step": 17983 }, { "epoch": 13.175091575091574, "grad_norm": 1.734375, "learning_rate": 7.385660987537861e-05, "loss": 0.131, "step": 17984 }, { "epoch": 13.175824175824175, "grad_norm": 1.8828125, "learning_rate": 7.385358887041434e-05, "loss": 0.1225, "step": 17985 }, { "epoch": 13.176556776556776, "grad_norm": 1.2421875, "learning_rate": 7.385056775355538e-05, "loss": 0.107, "step": 17986 }, { "epoch": 13.177289377289377, "grad_norm": 0.7421875, "learning_rate": 7.384754652481618e-05, "loss": 0.1104, "step": 17987 }, { "epoch": 13.178021978021977, "grad_norm": 1.359375, "learning_rate": 7.384452518421122e-05, "loss": 0.1883, "step": 17988 }, { "epoch": 13.178754578754578, "grad_norm": 1.671875, "learning_rate": 7.384150373175498e-05, "loss": 0.149, "step": 17989 }, { "epoch": 13.179487179487179, "grad_norm": 1.5703125, "learning_rate": 7.383848216746193e-05, "loss": 0.1157, "step": 17990 }, { "epoch": 13.18021978021978, "grad_norm": 0.80859375, "learning_rate": 7.383546049134654e-05, "loss": 0.1318, "step": 17991 }, { "epoch": 13.18095238095238, "grad_norm": 1.4140625, "learning_rate": 7.383243870342333e-05, "loss": 0.1277, "step": 17992 }, { "epoch": 13.181684981684981, "grad_norm": 0.6015625, "learning_rate": 7.38294168037067e-05, "loss": 0.125, "step": 17993 }, { "epoch": 13.182417582417582, "grad_norm": 1.109375, "learning_rate": 7.382639479221122e-05, "loss": 0.1047, "step": 17994 }, { "epoch": 13.183150183150182, "grad_norm": 1.21875, "learning_rate": 7.382337266895131e-05, "loss": 0.1721, "step": 17995 }, { "epoch": 13.183882783882783, "grad_norm": 0.89453125, "learning_rate": 7.382035043394145e-05, "loss": 0.1219, "step": 17996 }, { "epoch": 13.184615384615384, "grad_norm": 1.4140625, "learning_rate": 7.381732808719615e-05, "loss": 0.1302, "step": 17997 }, { "epoch": 13.185347985347985, "grad_norm": 1.1484375, "learning_rate": 7.381430562872987e-05, "loss": 0.1141, "step": 17998 }, { "epoch": 13.186080586080585, "grad_norm": 0.93359375, "learning_rate": 7.381128305855709e-05, "loss": 0.1344, "step": 17999 }, { "epoch": 13.186813186813186, "grad_norm": 0.64453125, "learning_rate": 7.38082603766923e-05, "loss": 0.1538, "step": 18000 }, { "epoch": 13.187545787545787, "grad_norm": 1.09375, "learning_rate": 7.380523758315001e-05, "loss": 0.0982, "step": 18001 }, { "epoch": 13.188278388278388, "grad_norm": 1.3828125, "learning_rate": 7.380221467794466e-05, "loss": 0.1319, "step": 18002 }, { "epoch": 13.189010989010988, "grad_norm": 0.6796875, "learning_rate": 7.379919166109076e-05, "loss": 0.1212, "step": 18003 }, { "epoch": 13.189743589743589, "grad_norm": 1.53125, "learning_rate": 7.379616853260275e-05, "loss": 0.1354, "step": 18004 }, { "epoch": 13.19047619047619, "grad_norm": 0.94140625, "learning_rate": 7.379314529249519e-05, "loss": 0.1262, "step": 18005 }, { "epoch": 13.19120879120879, "grad_norm": 1.203125, "learning_rate": 7.379012194078249e-05, "loss": 0.1335, "step": 18006 }, { "epoch": 13.191941391941391, "grad_norm": 0.62890625, "learning_rate": 7.378709847747919e-05, "loss": 0.0896, "step": 18007 }, { "epoch": 13.192673992673992, "grad_norm": 0.8359375, "learning_rate": 7.378407490259976e-05, "loss": 0.1309, "step": 18008 }, { "epoch": 13.193406593406593, "grad_norm": 1.84375, "learning_rate": 7.378105121615868e-05, "loss": 0.1478, "step": 18009 }, { "epoch": 13.194139194139193, "grad_norm": 0.90234375, "learning_rate": 7.377802741817043e-05, "loss": 0.1142, "step": 18010 }, { "epoch": 13.194871794871794, "grad_norm": 1.109375, "learning_rate": 7.377500350864951e-05, "loss": 0.17, "step": 18011 }, { "epoch": 13.195604395604395, "grad_norm": 1.4140625, "learning_rate": 7.377197948761043e-05, "loss": 0.112, "step": 18012 }, { "epoch": 13.196336996336996, "grad_norm": 1.03125, "learning_rate": 7.376895535506763e-05, "loss": 0.1572, "step": 18013 }, { "epoch": 13.197069597069596, "grad_norm": 1.53125, "learning_rate": 7.376593111103564e-05, "loss": 0.1151, "step": 18014 }, { "epoch": 13.197802197802197, "grad_norm": 1.2734375, "learning_rate": 7.376290675552894e-05, "loss": 0.132, "step": 18015 }, { "epoch": 13.198534798534798, "grad_norm": 0.8046875, "learning_rate": 7.375988228856202e-05, "loss": 0.1088, "step": 18016 }, { "epoch": 13.199267399267399, "grad_norm": 1.75, "learning_rate": 7.375685771014933e-05, "loss": 0.1305, "step": 18017 }, { "epoch": 13.2, "grad_norm": 0.91796875, "learning_rate": 7.375383302030544e-05, "loss": 0.1122, "step": 18018 }, { "epoch": 13.2007326007326, "grad_norm": 1.515625, "learning_rate": 7.375080821904478e-05, "loss": 0.1277, "step": 18019 }, { "epoch": 13.2014652014652, "grad_norm": 0.7890625, "learning_rate": 7.374778330638189e-05, "loss": 0.0973, "step": 18020 }, { "epoch": 13.202197802197801, "grad_norm": 0.9609375, "learning_rate": 7.374475828233123e-05, "loss": 0.104, "step": 18021 }, { "epoch": 13.202930402930402, "grad_norm": 1.8515625, "learning_rate": 7.374173314690729e-05, "loss": 0.1273, "step": 18022 }, { "epoch": 13.203663003663003, "grad_norm": 0.875, "learning_rate": 7.373870790012458e-05, "loss": 0.1156, "step": 18023 }, { "epoch": 13.204395604395604, "grad_norm": 1.1171875, "learning_rate": 7.37356825419976e-05, "loss": 0.1271, "step": 18024 }, { "epoch": 13.205128205128204, "grad_norm": 1.59375, "learning_rate": 7.373265707254083e-05, "loss": 0.1564, "step": 18025 }, { "epoch": 13.205860805860805, "grad_norm": 1.171875, "learning_rate": 7.372963149176877e-05, "loss": 0.1209, "step": 18026 }, { "epoch": 13.206593406593406, "grad_norm": 0.9765625, "learning_rate": 7.372660579969591e-05, "loss": 0.1092, "step": 18027 }, { "epoch": 13.207326007326007, "grad_norm": 1.3359375, "learning_rate": 7.372357999633678e-05, "loss": 0.131, "step": 18028 }, { "epoch": 13.208058608058607, "grad_norm": 1.34375, "learning_rate": 7.372055408170586e-05, "loss": 0.1278, "step": 18029 }, { "epoch": 13.208791208791208, "grad_norm": 1.8046875, "learning_rate": 7.371752805581762e-05, "loss": 0.1485, "step": 18030 }, { "epoch": 13.209523809523809, "grad_norm": 1.625, "learning_rate": 7.371450191868659e-05, "loss": 0.1305, "step": 18031 }, { "epoch": 13.21025641025641, "grad_norm": 0.92578125, "learning_rate": 7.371147567032725e-05, "loss": 0.1098, "step": 18032 }, { "epoch": 13.21098901098901, "grad_norm": 1.2890625, "learning_rate": 7.370844931075414e-05, "loss": 0.1276, "step": 18033 }, { "epoch": 13.211721611721611, "grad_norm": 0.75, "learning_rate": 7.370542283998171e-05, "loss": 0.1051, "step": 18034 }, { "epoch": 13.212454212454212, "grad_norm": 1.0, "learning_rate": 7.370239625802447e-05, "loss": 0.1276, "step": 18035 }, { "epoch": 13.213186813186812, "grad_norm": 1.390625, "learning_rate": 7.369936956489695e-05, "loss": 0.1187, "step": 18036 }, { "epoch": 13.213919413919413, "grad_norm": 1.1875, "learning_rate": 7.369634276061364e-05, "loss": 0.135, "step": 18037 }, { "epoch": 13.214652014652014, "grad_norm": 1.171875, "learning_rate": 7.369331584518902e-05, "loss": 0.1755, "step": 18038 }, { "epoch": 13.215384615384615, "grad_norm": 1.1875, "learning_rate": 7.369028881863762e-05, "loss": 0.1284, "step": 18039 }, { "epoch": 13.216117216117215, "grad_norm": 1.4140625, "learning_rate": 7.368726168097393e-05, "loss": 0.1498, "step": 18040 }, { "epoch": 13.216849816849816, "grad_norm": 1.0625, "learning_rate": 7.368423443221246e-05, "loss": 0.0981, "step": 18041 }, { "epoch": 13.217582417582417, "grad_norm": 1.1015625, "learning_rate": 7.36812070723677e-05, "loss": 0.1377, "step": 18042 }, { "epoch": 13.218315018315018, "grad_norm": 1.4296875, "learning_rate": 7.367817960145417e-05, "loss": 0.13, "step": 18043 }, { "epoch": 13.219047619047618, "grad_norm": 1.53125, "learning_rate": 7.36751520194864e-05, "loss": 0.1454, "step": 18044 }, { "epoch": 13.219780219780219, "grad_norm": 1.015625, "learning_rate": 7.367212432647884e-05, "loss": 0.1138, "step": 18045 }, { "epoch": 13.22051282051282, "grad_norm": 1.9140625, "learning_rate": 7.366909652244603e-05, "loss": 0.1496, "step": 18046 }, { "epoch": 13.22124542124542, "grad_norm": 1.171875, "learning_rate": 7.366606860740248e-05, "loss": 0.1756, "step": 18047 }, { "epoch": 13.221978021978021, "grad_norm": 0.87109375, "learning_rate": 7.366304058136268e-05, "loss": 0.1444, "step": 18048 }, { "epoch": 13.222710622710622, "grad_norm": 1.5703125, "learning_rate": 7.366001244434117e-05, "loss": 0.1411, "step": 18049 }, { "epoch": 13.223443223443223, "grad_norm": 0.79296875, "learning_rate": 7.365698419635241e-05, "loss": 0.0993, "step": 18050 }, { "epoch": 13.224175824175823, "grad_norm": 1.2734375, "learning_rate": 7.365395583741096e-05, "loss": 0.1286, "step": 18051 }, { "epoch": 13.224908424908424, "grad_norm": 0.6953125, "learning_rate": 7.365092736753129e-05, "loss": 0.1139, "step": 18052 }, { "epoch": 13.225641025641025, "grad_norm": 0.8671875, "learning_rate": 7.364789878672793e-05, "loss": 0.1271, "step": 18053 }, { "epoch": 13.226373626373626, "grad_norm": 1.078125, "learning_rate": 7.36448700950154e-05, "loss": 0.1686, "step": 18054 }, { "epoch": 13.227106227106226, "grad_norm": 0.76171875, "learning_rate": 7.364184129240818e-05, "loss": 0.1315, "step": 18055 }, { "epoch": 13.227838827838827, "grad_norm": 1.125, "learning_rate": 7.363881237892082e-05, "loss": 0.1645, "step": 18056 }, { "epoch": 13.228571428571428, "grad_norm": 0.85546875, "learning_rate": 7.36357833545678e-05, "loss": 0.1195, "step": 18057 }, { "epoch": 13.229304029304028, "grad_norm": 1.0859375, "learning_rate": 7.363275421936364e-05, "loss": 0.1576, "step": 18058 }, { "epoch": 13.23003663003663, "grad_norm": 1.171875, "learning_rate": 7.362972497332289e-05, "loss": 0.1298, "step": 18059 }, { "epoch": 13.23076923076923, "grad_norm": 1.2421875, "learning_rate": 7.362669561646001e-05, "loss": 0.1233, "step": 18060 }, { "epoch": 13.23150183150183, "grad_norm": 0.76953125, "learning_rate": 7.362366614878955e-05, "loss": 0.1328, "step": 18061 }, { "epoch": 13.232234432234431, "grad_norm": 1.0390625, "learning_rate": 7.362063657032602e-05, "loss": 0.1401, "step": 18062 }, { "epoch": 13.232967032967032, "grad_norm": 0.7265625, "learning_rate": 7.361760688108392e-05, "loss": 0.0988, "step": 18063 }, { "epoch": 13.233699633699633, "grad_norm": 1.25, "learning_rate": 7.361457708107778e-05, "loss": 0.1577, "step": 18064 }, { "epoch": 13.234432234432234, "grad_norm": 1.0859375, "learning_rate": 7.361154717032211e-05, "loss": 0.1026, "step": 18065 }, { "epoch": 13.235164835164834, "grad_norm": 1.484375, "learning_rate": 7.360851714883143e-05, "loss": 0.1224, "step": 18066 }, { "epoch": 13.235897435897435, "grad_norm": 1.0546875, "learning_rate": 7.360548701662025e-05, "loss": 0.1328, "step": 18067 }, { "epoch": 13.236630036630036, "grad_norm": 1.546875, "learning_rate": 7.360245677370312e-05, "loss": 0.1258, "step": 18068 }, { "epoch": 13.237362637362637, "grad_norm": 1.6171875, "learning_rate": 7.35994264200945e-05, "loss": 0.1416, "step": 18069 }, { "epoch": 13.238095238095237, "grad_norm": 0.66796875, "learning_rate": 7.359639595580898e-05, "loss": 0.1158, "step": 18070 }, { "epoch": 13.238827838827838, "grad_norm": 0.625, "learning_rate": 7.359336538086102e-05, "loss": 0.1319, "step": 18071 }, { "epoch": 13.239560439560439, "grad_norm": 0.93359375, "learning_rate": 7.35903346952652e-05, "loss": 0.0991, "step": 18072 }, { "epoch": 13.24029304029304, "grad_norm": 1.015625, "learning_rate": 7.358730389903598e-05, "loss": 0.1282, "step": 18073 }, { "epoch": 13.24102564102564, "grad_norm": 1.5546875, "learning_rate": 7.35842729921879e-05, "loss": 0.1341, "step": 18074 }, { "epoch": 13.241758241758241, "grad_norm": 1.2421875, "learning_rate": 7.358124197473551e-05, "loss": 0.1447, "step": 18075 }, { "epoch": 13.242490842490842, "grad_norm": 0.7734375, "learning_rate": 7.35782108466933e-05, "loss": 0.1289, "step": 18076 }, { "epoch": 13.243223443223442, "grad_norm": 1.046875, "learning_rate": 7.357517960807582e-05, "loss": 0.1447, "step": 18077 }, { "epoch": 13.243956043956043, "grad_norm": 1.296875, "learning_rate": 7.357214825889756e-05, "loss": 0.1156, "step": 18078 }, { "epoch": 13.244688644688644, "grad_norm": 0.83203125, "learning_rate": 7.356911679917308e-05, "loss": 0.1048, "step": 18079 }, { "epoch": 13.245421245421245, "grad_norm": 0.9375, "learning_rate": 7.356608522891687e-05, "loss": 0.1133, "step": 18080 }, { "epoch": 13.246153846153845, "grad_norm": 1.265625, "learning_rate": 7.35630535481435e-05, "loss": 0.1228, "step": 18081 }, { "epoch": 13.246886446886446, "grad_norm": 1.828125, "learning_rate": 7.356002175686745e-05, "loss": 0.1501, "step": 18082 }, { "epoch": 13.247619047619047, "grad_norm": 1.59375, "learning_rate": 7.355698985510328e-05, "loss": 0.1207, "step": 18083 }, { "epoch": 13.248351648351647, "grad_norm": 0.97265625, "learning_rate": 7.355395784286548e-05, "loss": 0.1157, "step": 18084 }, { "epoch": 13.249084249084248, "grad_norm": 1.2578125, "learning_rate": 7.355092572016862e-05, "loss": 0.1457, "step": 18085 }, { "epoch": 13.249816849816849, "grad_norm": 1.1484375, "learning_rate": 7.35478934870272e-05, "loss": 0.0954, "step": 18086 }, { "epoch": 13.25054945054945, "grad_norm": 0.67578125, "learning_rate": 7.354486114345575e-05, "loss": 0.1131, "step": 18087 }, { "epoch": 13.25128205128205, "grad_norm": 0.7109375, "learning_rate": 7.354182868946882e-05, "loss": 0.1145, "step": 18088 }, { "epoch": 13.252014652014651, "grad_norm": 1.3671875, "learning_rate": 7.35387961250809e-05, "loss": 0.1237, "step": 18089 }, { "epoch": 13.252747252747252, "grad_norm": 0.9765625, "learning_rate": 7.353576345030657e-05, "loss": 0.1157, "step": 18090 }, { "epoch": 13.253479853479853, "grad_norm": 1.5703125, "learning_rate": 7.353273066516033e-05, "loss": 0.1519, "step": 18091 }, { "epoch": 13.254212454212453, "grad_norm": 0.8671875, "learning_rate": 7.35296977696567e-05, "loss": 0.1006, "step": 18092 }, { "epoch": 13.254945054945054, "grad_norm": 1.3046875, "learning_rate": 7.352666476381024e-05, "loss": 0.121, "step": 18093 }, { "epoch": 13.255677655677655, "grad_norm": 0.65234375, "learning_rate": 7.352363164763548e-05, "loss": 0.1298, "step": 18094 }, { "epoch": 13.256410256410255, "grad_norm": 0.98046875, "learning_rate": 7.352059842114693e-05, "loss": 0.1276, "step": 18095 }, { "epoch": 13.257142857142856, "grad_norm": 0.8046875, "learning_rate": 7.351756508435914e-05, "loss": 0.0976, "step": 18096 }, { "epoch": 13.257875457875459, "grad_norm": 0.81640625, "learning_rate": 7.351453163728664e-05, "loss": 0.122, "step": 18097 }, { "epoch": 13.25860805860806, "grad_norm": 0.84765625, "learning_rate": 7.351149807994397e-05, "loss": 0.1046, "step": 18098 }, { "epoch": 13.25934065934066, "grad_norm": 1.15625, "learning_rate": 7.350846441234564e-05, "loss": 0.1153, "step": 18099 }, { "epoch": 13.260073260073261, "grad_norm": 1.6015625, "learning_rate": 7.350543063450623e-05, "loss": 0.1631, "step": 18100 }, { "epoch": 13.260805860805862, "grad_norm": 0.84765625, "learning_rate": 7.350239674644023e-05, "loss": 0.1456, "step": 18101 }, { "epoch": 13.261538461538462, "grad_norm": 1.046875, "learning_rate": 7.349936274816219e-05, "loss": 0.1114, "step": 18102 }, { "epoch": 13.262271062271063, "grad_norm": 0.76953125, "learning_rate": 7.349632863968666e-05, "loss": 0.0977, "step": 18103 }, { "epoch": 13.263003663003664, "grad_norm": 0.87890625, "learning_rate": 7.349329442102818e-05, "loss": 0.1187, "step": 18104 }, { "epoch": 13.263736263736265, "grad_norm": 1.1796875, "learning_rate": 7.349026009220125e-05, "loss": 0.1143, "step": 18105 }, { "epoch": 13.264468864468865, "grad_norm": 0.70703125, "learning_rate": 7.348722565322048e-05, "loss": 0.1295, "step": 18106 }, { "epoch": 13.265201465201466, "grad_norm": 1.0234375, "learning_rate": 7.348419110410035e-05, "loss": 0.1134, "step": 18107 }, { "epoch": 13.265934065934067, "grad_norm": 1.421875, "learning_rate": 7.34811564448554e-05, "loss": 0.1298, "step": 18108 }, { "epoch": 13.266666666666667, "grad_norm": 1.1875, "learning_rate": 7.347812167550019e-05, "loss": 0.1074, "step": 18109 }, { "epoch": 13.267399267399268, "grad_norm": 0.765625, "learning_rate": 7.347508679604927e-05, "loss": 0.1329, "step": 18110 }, { "epoch": 13.268131868131869, "grad_norm": 0.98828125, "learning_rate": 7.347205180651716e-05, "loss": 0.1344, "step": 18111 }, { "epoch": 13.26886446886447, "grad_norm": 0.92578125, "learning_rate": 7.34690167069184e-05, "loss": 0.1238, "step": 18112 }, { "epoch": 13.26959706959707, "grad_norm": 0.5859375, "learning_rate": 7.346598149726755e-05, "loss": 0.1088, "step": 18113 }, { "epoch": 13.270329670329671, "grad_norm": 0.9296875, "learning_rate": 7.346294617757914e-05, "loss": 0.1381, "step": 18114 }, { "epoch": 13.271062271062272, "grad_norm": 0.703125, "learning_rate": 7.345991074786771e-05, "loss": 0.098, "step": 18115 }, { "epoch": 13.271794871794873, "grad_norm": 0.6640625, "learning_rate": 7.345687520814781e-05, "loss": 0.1162, "step": 18116 }, { "epoch": 13.272527472527473, "grad_norm": 0.953125, "learning_rate": 7.3453839558434e-05, "loss": 0.1294, "step": 18117 }, { "epoch": 13.273260073260074, "grad_norm": 0.8203125, "learning_rate": 7.345080379874078e-05, "loss": 0.1216, "step": 18118 }, { "epoch": 13.273992673992675, "grad_norm": 1.203125, "learning_rate": 7.344776792908275e-05, "loss": 0.1286, "step": 18119 }, { "epoch": 13.274725274725276, "grad_norm": 0.98046875, "learning_rate": 7.344473194947444e-05, "loss": 0.1439, "step": 18120 }, { "epoch": 13.275457875457876, "grad_norm": 0.62890625, "learning_rate": 7.344169585993038e-05, "loss": 0.1396, "step": 18121 }, { "epoch": 13.276190476190477, "grad_norm": 0.91015625, "learning_rate": 7.343865966046511e-05, "loss": 0.1, "step": 18122 }, { "epoch": 13.276923076923078, "grad_norm": 0.6171875, "learning_rate": 7.343562335109319e-05, "loss": 0.1241, "step": 18123 }, { "epoch": 13.277655677655678, "grad_norm": 1.0625, "learning_rate": 7.343258693182919e-05, "loss": 0.1341, "step": 18124 }, { "epoch": 13.27838827838828, "grad_norm": 1.125, "learning_rate": 7.342955040268762e-05, "loss": 0.1145, "step": 18125 }, { "epoch": 13.27912087912088, "grad_norm": 1.328125, "learning_rate": 7.342651376368304e-05, "loss": 0.1283, "step": 18126 }, { "epoch": 13.27985347985348, "grad_norm": 1.390625, "learning_rate": 7.342347701483004e-05, "loss": 0.1398, "step": 18127 }, { "epoch": 13.280586080586081, "grad_norm": 0.66015625, "learning_rate": 7.34204401561431e-05, "loss": 0.1272, "step": 18128 }, { "epoch": 13.281318681318682, "grad_norm": 0.55078125, "learning_rate": 7.341740318763682e-05, "loss": 0.1235, "step": 18129 }, { "epoch": 13.282051282051283, "grad_norm": 0.84765625, "learning_rate": 7.341436610932574e-05, "loss": 0.1, "step": 18130 }, { "epoch": 13.282783882783884, "grad_norm": 0.9453125, "learning_rate": 7.341132892122441e-05, "loss": 0.1213, "step": 18131 }, { "epoch": 13.283516483516484, "grad_norm": 0.91015625, "learning_rate": 7.340829162334738e-05, "loss": 0.125, "step": 18132 }, { "epoch": 13.284249084249085, "grad_norm": 1.0625, "learning_rate": 7.340525421570921e-05, "loss": 0.1362, "step": 18133 }, { "epoch": 13.284981684981686, "grad_norm": 1.109375, "learning_rate": 7.340221669832443e-05, "loss": 0.1291, "step": 18134 }, { "epoch": 13.285714285714286, "grad_norm": 1.4765625, "learning_rate": 7.339917907120763e-05, "loss": 0.1114, "step": 18135 }, { "epoch": 13.286446886446887, "grad_norm": 0.88671875, "learning_rate": 7.339614133437335e-05, "loss": 0.1242, "step": 18136 }, { "epoch": 13.287179487179488, "grad_norm": 0.89453125, "learning_rate": 7.339310348783612e-05, "loss": 0.1269, "step": 18137 }, { "epoch": 13.287912087912089, "grad_norm": 0.890625, "learning_rate": 7.339006553161055e-05, "loss": 0.1217, "step": 18138 }, { "epoch": 13.28864468864469, "grad_norm": 1.25, "learning_rate": 7.338702746571111e-05, "loss": 0.1757, "step": 18139 }, { "epoch": 13.28937728937729, "grad_norm": 0.83984375, "learning_rate": 7.338398929015244e-05, "loss": 0.1028, "step": 18140 }, { "epoch": 13.29010989010989, "grad_norm": 1.3125, "learning_rate": 7.338095100494907e-05, "loss": 0.1231, "step": 18141 }, { "epoch": 13.290842490842492, "grad_norm": 1.6484375, "learning_rate": 7.337791261011554e-05, "loss": 0.1492, "step": 18142 }, { "epoch": 13.291575091575092, "grad_norm": 1.53125, "learning_rate": 7.337487410566642e-05, "loss": 0.1241, "step": 18143 }, { "epoch": 13.292307692307693, "grad_norm": 1.8125, "learning_rate": 7.337183549161628e-05, "loss": 0.1331, "step": 18144 }, { "epoch": 13.293040293040294, "grad_norm": 1.1171875, "learning_rate": 7.336879676797966e-05, "loss": 0.15, "step": 18145 }, { "epoch": 13.293772893772895, "grad_norm": 1.0546875, "learning_rate": 7.336575793477112e-05, "loss": 0.1164, "step": 18146 }, { "epoch": 13.294505494505495, "grad_norm": 1.2578125, "learning_rate": 7.336271899200524e-05, "loss": 0.1201, "step": 18147 }, { "epoch": 13.295238095238096, "grad_norm": 1.15625, "learning_rate": 7.335967993969657e-05, "loss": 0.1487, "step": 18148 }, { "epoch": 13.295970695970697, "grad_norm": 1.5, "learning_rate": 7.335664077785965e-05, "loss": 0.1116, "step": 18149 }, { "epoch": 13.296703296703297, "grad_norm": 1.1328125, "learning_rate": 7.335360150650908e-05, "loss": 0.1235, "step": 18150 }, { "epoch": 13.297435897435898, "grad_norm": 1.7734375, "learning_rate": 7.335056212565941e-05, "loss": 0.1397, "step": 18151 }, { "epoch": 13.298168498168499, "grad_norm": 1.6171875, "learning_rate": 7.334752263532517e-05, "loss": 0.1345, "step": 18152 }, { "epoch": 13.2989010989011, "grad_norm": 1.453125, "learning_rate": 7.334448303552097e-05, "loss": 0.1108, "step": 18153 }, { "epoch": 13.2996336996337, "grad_norm": 1.125, "learning_rate": 7.334144332626134e-05, "loss": 0.1231, "step": 18154 }, { "epoch": 13.300366300366301, "grad_norm": 0.7109375, "learning_rate": 7.333840350756086e-05, "loss": 0.1035, "step": 18155 }, { "epoch": 13.301098901098902, "grad_norm": 1.140625, "learning_rate": 7.333536357943409e-05, "loss": 0.1497, "step": 18156 }, { "epoch": 13.301831501831503, "grad_norm": 1.9140625, "learning_rate": 7.333232354189558e-05, "loss": 0.1584, "step": 18157 }, { "epoch": 13.302564102564103, "grad_norm": 1.546875, "learning_rate": 7.332928339495995e-05, "loss": 0.1212, "step": 18158 }, { "epoch": 13.303296703296704, "grad_norm": 1.9765625, "learning_rate": 7.332624313864172e-05, "loss": 0.1534, "step": 18159 }, { "epoch": 13.304029304029305, "grad_norm": 1.234375, "learning_rate": 7.332320277295544e-05, "loss": 0.1219, "step": 18160 }, { "epoch": 13.304761904761905, "grad_norm": 1.15625, "learning_rate": 7.332016229791572e-05, "loss": 0.1162, "step": 18161 }, { "epoch": 13.305494505494506, "grad_norm": 1.2890625, "learning_rate": 7.331712171353709e-05, "loss": 0.1069, "step": 18162 }, { "epoch": 13.306227106227107, "grad_norm": 0.71875, "learning_rate": 7.331408101983418e-05, "loss": 0.1096, "step": 18163 }, { "epoch": 13.306959706959708, "grad_norm": 0.80078125, "learning_rate": 7.33110402168215e-05, "loss": 0.1084, "step": 18164 }, { "epoch": 13.307692307692308, "grad_norm": 2.140625, "learning_rate": 7.330799930451363e-05, "loss": 0.1274, "step": 18165 }, { "epoch": 13.308424908424909, "grad_norm": 1.484375, "learning_rate": 7.330495828292517e-05, "loss": 0.1437, "step": 18166 }, { "epoch": 13.30915750915751, "grad_norm": 0.98828125, "learning_rate": 7.330191715207065e-05, "loss": 0.1153, "step": 18167 }, { "epoch": 13.30989010989011, "grad_norm": 0.9921875, "learning_rate": 7.329887591196464e-05, "loss": 0.1216, "step": 18168 }, { "epoch": 13.310622710622711, "grad_norm": 1.0546875, "learning_rate": 7.329583456262175e-05, "loss": 0.1286, "step": 18169 }, { "epoch": 13.311355311355312, "grad_norm": 1.015625, "learning_rate": 7.329279310405654e-05, "loss": 0.1552, "step": 18170 }, { "epoch": 13.312087912087913, "grad_norm": 1.3203125, "learning_rate": 7.328975153628357e-05, "loss": 0.1319, "step": 18171 }, { "epoch": 13.312820512820513, "grad_norm": 1.03125, "learning_rate": 7.328670985931742e-05, "loss": 0.1258, "step": 18172 }, { "epoch": 13.313553113553114, "grad_norm": 1.296875, "learning_rate": 7.328366807317268e-05, "loss": 0.1415, "step": 18173 }, { "epoch": 13.314285714285715, "grad_norm": 0.90234375, "learning_rate": 7.328062617786388e-05, "loss": 0.1318, "step": 18174 }, { "epoch": 13.315018315018316, "grad_norm": 1.125, "learning_rate": 7.327758417340562e-05, "loss": 0.1305, "step": 18175 }, { "epoch": 13.315750915750916, "grad_norm": 0.85546875, "learning_rate": 7.32745420598125e-05, "loss": 0.1507, "step": 18176 }, { "epoch": 13.316483516483517, "grad_norm": 0.73046875, "learning_rate": 7.327149983709906e-05, "loss": 0.1097, "step": 18177 }, { "epoch": 13.317216117216118, "grad_norm": 0.9453125, "learning_rate": 7.326845750527988e-05, "loss": 0.131, "step": 18178 }, { "epoch": 13.317948717948719, "grad_norm": 1.6796875, "learning_rate": 7.326541506436956e-05, "loss": 0.1711, "step": 18179 }, { "epoch": 13.31868131868132, "grad_norm": 0.9375, "learning_rate": 7.326237251438267e-05, "loss": 0.1268, "step": 18180 }, { "epoch": 13.31941391941392, "grad_norm": 1.125, "learning_rate": 7.325932985533375e-05, "loss": 0.1204, "step": 18181 }, { "epoch": 13.32014652014652, "grad_norm": 1.484375, "learning_rate": 7.325628708723742e-05, "loss": 0.156, "step": 18182 }, { "epoch": 13.320879120879122, "grad_norm": 0.8984375, "learning_rate": 7.325324421010827e-05, "loss": 0.1, "step": 18183 }, { "epoch": 13.321611721611722, "grad_norm": 1.0390625, "learning_rate": 7.325020122396085e-05, "loss": 0.1226, "step": 18184 }, { "epoch": 13.322344322344323, "grad_norm": 1.328125, "learning_rate": 7.324715812880974e-05, "loss": 0.1314, "step": 18185 }, { "epoch": 13.323076923076924, "grad_norm": 0.82421875, "learning_rate": 7.324411492466953e-05, "loss": 0.1174, "step": 18186 }, { "epoch": 13.323809523809524, "grad_norm": 1.28125, "learning_rate": 7.32410716115548e-05, "loss": 0.1586, "step": 18187 }, { "epoch": 13.324542124542125, "grad_norm": 0.85546875, "learning_rate": 7.323802818948013e-05, "loss": 0.1117, "step": 18188 }, { "epoch": 13.325274725274726, "grad_norm": 0.9453125, "learning_rate": 7.32349846584601e-05, "loss": 0.1034, "step": 18189 }, { "epoch": 13.326007326007327, "grad_norm": 2.140625, "learning_rate": 7.32319410185093e-05, "loss": 0.124, "step": 18190 }, { "epoch": 13.326739926739927, "grad_norm": 0.90625, "learning_rate": 7.32288972696423e-05, "loss": 0.1168, "step": 18191 }, { "epoch": 13.327472527472528, "grad_norm": 1.265625, "learning_rate": 7.322585341187371e-05, "loss": 0.1293, "step": 18192 }, { "epoch": 13.328205128205129, "grad_norm": 1.1328125, "learning_rate": 7.322280944521807e-05, "loss": 0.1579, "step": 18193 }, { "epoch": 13.32893772893773, "grad_norm": 1.40625, "learning_rate": 7.321976536969e-05, "loss": 0.1735, "step": 18194 }, { "epoch": 13.32967032967033, "grad_norm": 1.421875, "learning_rate": 7.321672118530409e-05, "loss": 0.1221, "step": 18195 }, { "epoch": 13.330402930402931, "grad_norm": 1.4921875, "learning_rate": 7.32136768920749e-05, "loss": 0.133, "step": 18196 }, { "epoch": 13.331135531135532, "grad_norm": 0.7734375, "learning_rate": 7.321063249001701e-05, "loss": 0.1025, "step": 18197 }, { "epoch": 13.331868131868132, "grad_norm": 1.1015625, "learning_rate": 7.320758797914506e-05, "loss": 0.1254, "step": 18198 }, { "epoch": 13.332600732600733, "grad_norm": 1.0078125, "learning_rate": 7.320454335947358e-05, "loss": 0.1243, "step": 18199 }, { "epoch": 13.333333333333334, "grad_norm": 0.70703125, "learning_rate": 7.320149863101719e-05, "loss": 0.1293, "step": 18200 }, { "epoch": 13.334065934065935, "grad_norm": 1.328125, "learning_rate": 7.319845379379044e-05, "loss": 0.1221, "step": 18201 }, { "epoch": 13.334798534798535, "grad_norm": 1.0078125, "learning_rate": 7.319540884780799e-05, "loss": 0.1211, "step": 18202 }, { "epoch": 13.335531135531136, "grad_norm": 1.4453125, "learning_rate": 7.319236379308437e-05, "loss": 0.1232, "step": 18203 }, { "epoch": 13.336263736263737, "grad_norm": 1.3828125, "learning_rate": 7.318931862963416e-05, "loss": 0.1098, "step": 18204 }, { "epoch": 13.336996336996338, "grad_norm": 1.8984375, "learning_rate": 7.318627335747201e-05, "loss": 0.1349, "step": 18205 }, { "epoch": 13.337728937728938, "grad_norm": 1.203125, "learning_rate": 7.318322797661246e-05, "loss": 0.1231, "step": 18206 }, { "epoch": 13.338461538461539, "grad_norm": 1.1796875, "learning_rate": 7.318018248707012e-05, "loss": 0.1251, "step": 18207 }, { "epoch": 13.33919413919414, "grad_norm": 1.171875, "learning_rate": 7.317713688885958e-05, "loss": 0.151, "step": 18208 }, { "epoch": 13.33992673992674, "grad_norm": 1.90625, "learning_rate": 7.317409118199541e-05, "loss": 0.129, "step": 18209 }, { "epoch": 13.340659340659341, "grad_norm": 1.328125, "learning_rate": 7.317104536649227e-05, "loss": 0.0998, "step": 18210 }, { "epoch": 13.341391941391942, "grad_norm": 0.74609375, "learning_rate": 7.316799944236468e-05, "loss": 0.1118, "step": 18211 }, { "epoch": 13.342124542124543, "grad_norm": 1.75, "learning_rate": 7.316495340962728e-05, "loss": 0.1339, "step": 18212 }, { "epoch": 13.342857142857143, "grad_norm": 0.94921875, "learning_rate": 7.316190726829464e-05, "loss": 0.1123, "step": 18213 }, { "epoch": 13.343589743589744, "grad_norm": 1.96875, "learning_rate": 7.315886101838137e-05, "loss": 0.1424, "step": 18214 }, { "epoch": 13.344322344322345, "grad_norm": 1.1328125, "learning_rate": 7.315581465990204e-05, "loss": 0.1274, "step": 18215 }, { "epoch": 13.345054945054946, "grad_norm": 2.265625, "learning_rate": 7.315276819287127e-05, "loss": 0.1367, "step": 18216 }, { "epoch": 13.345787545787546, "grad_norm": 1.3203125, "learning_rate": 7.314972161730366e-05, "loss": 0.1278, "step": 18217 }, { "epoch": 13.346520146520147, "grad_norm": 1.0703125, "learning_rate": 7.31466749332138e-05, "loss": 0.1395, "step": 18218 }, { "epoch": 13.347252747252748, "grad_norm": 1.640625, "learning_rate": 7.314362814061628e-05, "loss": 0.1246, "step": 18219 }, { "epoch": 13.347985347985349, "grad_norm": 1.3984375, "learning_rate": 7.31405812395257e-05, "loss": 0.1293, "step": 18220 }, { "epoch": 13.34871794871795, "grad_norm": 1.1015625, "learning_rate": 7.313753422995667e-05, "loss": 0.1197, "step": 18221 }, { "epoch": 13.34945054945055, "grad_norm": 1.53125, "learning_rate": 7.313448711192378e-05, "loss": 0.1283, "step": 18222 }, { "epoch": 13.35018315018315, "grad_norm": 1.8515625, "learning_rate": 7.313143988544162e-05, "loss": 0.1428, "step": 18223 }, { "epoch": 13.350915750915751, "grad_norm": 1.953125, "learning_rate": 7.312839255052482e-05, "loss": 0.1403, "step": 18224 }, { "epoch": 13.351648351648352, "grad_norm": 0.79296875, "learning_rate": 7.312534510718796e-05, "loss": 0.1258, "step": 18225 }, { "epoch": 13.352380952380953, "grad_norm": 0.75390625, "learning_rate": 7.312229755544564e-05, "loss": 0.1196, "step": 18226 }, { "epoch": 13.353113553113554, "grad_norm": 0.9375, "learning_rate": 7.311924989531247e-05, "loss": 0.125, "step": 18227 }, { "epoch": 13.353846153846154, "grad_norm": 1.484375, "learning_rate": 7.311620212680304e-05, "loss": 0.1202, "step": 18228 }, { "epoch": 13.354578754578755, "grad_norm": 1.203125, "learning_rate": 7.311315424993198e-05, "loss": 0.1643, "step": 18229 }, { "epoch": 13.355311355311356, "grad_norm": 0.59375, "learning_rate": 7.311010626471385e-05, "loss": 0.1107, "step": 18230 }, { "epoch": 13.356043956043957, "grad_norm": 1.1796875, "learning_rate": 7.310705817116328e-05, "loss": 0.136, "step": 18231 }, { "epoch": 13.356776556776557, "grad_norm": 0.87890625, "learning_rate": 7.310400996929489e-05, "loss": 0.129, "step": 18232 }, { "epoch": 13.357509157509158, "grad_norm": 1.0703125, "learning_rate": 7.310096165912325e-05, "loss": 0.1072, "step": 18233 }, { "epoch": 13.358241758241759, "grad_norm": 1.140625, "learning_rate": 7.3097913240663e-05, "loss": 0.1084, "step": 18234 }, { "epoch": 13.35897435897436, "grad_norm": 1.4453125, "learning_rate": 7.309486471392872e-05, "loss": 0.147, "step": 18235 }, { "epoch": 13.35970695970696, "grad_norm": 0.87109375, "learning_rate": 7.309181607893504e-05, "loss": 0.1092, "step": 18236 }, { "epoch": 13.360439560439561, "grad_norm": 0.73828125, "learning_rate": 7.308876733569655e-05, "loss": 0.1104, "step": 18237 }, { "epoch": 13.361172161172162, "grad_norm": 1.171875, "learning_rate": 7.308571848422784e-05, "loss": 0.124, "step": 18238 }, { "epoch": 13.361904761904762, "grad_norm": 1.609375, "learning_rate": 7.308266952454356e-05, "loss": 0.1598, "step": 18239 }, { "epoch": 13.362637362637363, "grad_norm": 1.6953125, "learning_rate": 7.307962045665827e-05, "loss": 0.1654, "step": 18240 }, { "epoch": 13.363369963369964, "grad_norm": 1.2421875, "learning_rate": 7.307657128058663e-05, "loss": 0.1336, "step": 18241 }, { "epoch": 13.364102564102565, "grad_norm": 0.65234375, "learning_rate": 7.307352199634321e-05, "loss": 0.1117, "step": 18242 }, { "epoch": 13.364835164835165, "grad_norm": 1.125, "learning_rate": 7.307047260394264e-05, "loss": 0.1405, "step": 18243 }, { "epoch": 13.365567765567766, "grad_norm": 1.703125, "learning_rate": 7.306742310339954e-05, "loss": 0.1464, "step": 18244 }, { "epoch": 13.366300366300367, "grad_norm": 0.70703125, "learning_rate": 7.306437349472849e-05, "loss": 0.1278, "step": 18245 }, { "epoch": 13.367032967032968, "grad_norm": 1.375, "learning_rate": 7.306132377794412e-05, "loss": 0.1168, "step": 18246 }, { "epoch": 13.367765567765568, "grad_norm": 1.3359375, "learning_rate": 7.305827395306106e-05, "loss": 0.1257, "step": 18247 }, { "epoch": 13.368498168498169, "grad_norm": 1.1640625, "learning_rate": 7.30552240200939e-05, "loss": 0.1405, "step": 18248 }, { "epoch": 13.36923076923077, "grad_norm": 1.796875, "learning_rate": 7.305217397905725e-05, "loss": 0.1479, "step": 18249 }, { "epoch": 13.36996336996337, "grad_norm": 1.4921875, "learning_rate": 7.304912382996573e-05, "loss": 0.1707, "step": 18250 }, { "epoch": 13.370695970695971, "grad_norm": 0.96875, "learning_rate": 7.304607357283395e-05, "loss": 0.1273, "step": 18251 }, { "epoch": 13.371428571428572, "grad_norm": 1.453125, "learning_rate": 7.304302320767654e-05, "loss": 0.132, "step": 18252 }, { "epoch": 13.372161172161173, "grad_norm": 1.1328125, "learning_rate": 7.30399727345081e-05, "loss": 0.1329, "step": 18253 }, { "epoch": 13.372893772893773, "grad_norm": 1.453125, "learning_rate": 7.303692215334325e-05, "loss": 0.1631, "step": 18254 }, { "epoch": 13.373626373626374, "grad_norm": 1.5546875, "learning_rate": 7.303387146419661e-05, "loss": 0.1634, "step": 18255 }, { "epoch": 13.374358974358975, "grad_norm": 0.8203125, "learning_rate": 7.30308206670828e-05, "loss": 0.1167, "step": 18256 }, { "epoch": 13.375091575091576, "grad_norm": 1.5390625, "learning_rate": 7.302776976201642e-05, "loss": 0.126, "step": 18257 }, { "epoch": 13.375824175824176, "grad_norm": 0.97265625, "learning_rate": 7.302471874901211e-05, "loss": 0.1477, "step": 18258 }, { "epoch": 13.376556776556777, "grad_norm": 1.0859375, "learning_rate": 7.302166762808447e-05, "loss": 0.1076, "step": 18259 }, { "epoch": 13.377289377289378, "grad_norm": 1.015625, "learning_rate": 7.301861639924814e-05, "loss": 0.1002, "step": 18260 }, { "epoch": 13.378021978021978, "grad_norm": 1.0703125, "learning_rate": 7.30155650625177e-05, "loss": 0.1251, "step": 18261 }, { "epoch": 13.37875457875458, "grad_norm": 1.4296875, "learning_rate": 7.301251361790783e-05, "loss": 0.1467, "step": 18262 }, { "epoch": 13.37948717948718, "grad_norm": 0.9375, "learning_rate": 7.30094620654331e-05, "loss": 0.1347, "step": 18263 }, { "epoch": 13.38021978021978, "grad_norm": 1.03125, "learning_rate": 7.300641040510814e-05, "loss": 0.134, "step": 18264 }, { "epoch": 13.380952380952381, "grad_norm": 0.78125, "learning_rate": 7.30033586369476e-05, "loss": 0.1452, "step": 18265 }, { "epoch": 13.381684981684982, "grad_norm": 0.9140625, "learning_rate": 7.300030676096606e-05, "loss": 0.1242, "step": 18266 }, { "epoch": 13.382417582417583, "grad_norm": 0.80078125, "learning_rate": 7.299725477717817e-05, "loss": 0.1475, "step": 18267 }, { "epoch": 13.383150183150184, "grad_norm": 1.0703125, "learning_rate": 7.299420268559857e-05, "loss": 0.1311, "step": 18268 }, { "epoch": 13.383882783882784, "grad_norm": 0.90234375, "learning_rate": 7.299115048624183e-05, "loss": 0.1089, "step": 18269 }, { "epoch": 13.384615384615385, "grad_norm": 0.9921875, "learning_rate": 7.298809817912261e-05, "loss": 0.1264, "step": 18270 }, { "epoch": 13.385347985347986, "grad_norm": 0.93359375, "learning_rate": 7.298504576425554e-05, "loss": 0.1076, "step": 18271 }, { "epoch": 13.386080586080586, "grad_norm": 1.3984375, "learning_rate": 7.298199324165523e-05, "loss": 0.1517, "step": 18272 }, { "epoch": 13.386813186813187, "grad_norm": 0.86328125, "learning_rate": 7.297894061133632e-05, "loss": 0.124, "step": 18273 }, { "epoch": 13.387545787545788, "grad_norm": 1.203125, "learning_rate": 7.29758878733134e-05, "loss": 0.1138, "step": 18274 }, { "epoch": 13.388278388278389, "grad_norm": 1.71875, "learning_rate": 7.297283502760113e-05, "loss": 0.1392, "step": 18275 }, { "epoch": 13.38901098901099, "grad_norm": 0.91015625, "learning_rate": 7.296978207421416e-05, "loss": 0.1372, "step": 18276 }, { "epoch": 13.38974358974359, "grad_norm": 1.578125, "learning_rate": 7.296672901316706e-05, "loss": 0.1392, "step": 18277 }, { "epoch": 13.39047619047619, "grad_norm": 1.5, "learning_rate": 7.29636758444745e-05, "loss": 0.1273, "step": 18278 }, { "epoch": 13.391208791208792, "grad_norm": 1.6015625, "learning_rate": 7.296062256815109e-05, "loss": 0.144, "step": 18279 }, { "epoch": 13.391941391941392, "grad_norm": 1.0859375, "learning_rate": 7.295756918421145e-05, "loss": 0.1134, "step": 18280 }, { "epoch": 13.392673992673993, "grad_norm": 1.3984375, "learning_rate": 7.295451569267024e-05, "loss": 0.1535, "step": 18281 }, { "epoch": 13.393406593406594, "grad_norm": 1.0078125, "learning_rate": 7.295146209354206e-05, "loss": 0.1298, "step": 18282 }, { "epoch": 13.394139194139195, "grad_norm": 1.3359375, "learning_rate": 7.294840838684156e-05, "loss": 0.1388, "step": 18283 }, { "epoch": 13.394871794871795, "grad_norm": 1.21875, "learning_rate": 7.294535457258339e-05, "loss": 0.1229, "step": 18284 }, { "epoch": 13.395604395604396, "grad_norm": 0.88671875, "learning_rate": 7.294230065078212e-05, "loss": 0.1433, "step": 18285 }, { "epoch": 13.396336996336997, "grad_norm": 1.3046875, "learning_rate": 7.293924662145244e-05, "loss": 0.1226, "step": 18286 }, { "epoch": 13.397069597069597, "grad_norm": 0.78125, "learning_rate": 7.293619248460896e-05, "loss": 0.1266, "step": 18287 }, { "epoch": 13.397802197802198, "grad_norm": 1.140625, "learning_rate": 7.293313824026632e-05, "loss": 0.1258, "step": 18288 }, { "epoch": 13.398534798534799, "grad_norm": 1.0625, "learning_rate": 7.293008388843914e-05, "loss": 0.146, "step": 18289 }, { "epoch": 13.3992673992674, "grad_norm": 0.77734375, "learning_rate": 7.292702942914208e-05, "loss": 0.144, "step": 18290 }, { "epoch": 13.4, "grad_norm": 1.40625, "learning_rate": 7.292397486238975e-05, "loss": 0.1147, "step": 18291 }, { "epoch": 13.400732600732601, "grad_norm": 1.0546875, "learning_rate": 7.292092018819681e-05, "loss": 0.1216, "step": 18292 }, { "epoch": 13.401465201465202, "grad_norm": 1.6484375, "learning_rate": 7.291786540657787e-05, "loss": 0.1485, "step": 18293 }, { "epoch": 13.402197802197803, "grad_norm": 1.1796875, "learning_rate": 7.291481051754757e-05, "loss": 0.1481, "step": 18294 }, { "epoch": 13.402930402930403, "grad_norm": 1.484375, "learning_rate": 7.291175552112057e-05, "loss": 0.1365, "step": 18295 }, { "epoch": 13.403663003663004, "grad_norm": 1.625, "learning_rate": 7.290870041731147e-05, "loss": 0.1425, "step": 18296 }, { "epoch": 13.404395604395605, "grad_norm": 1.0234375, "learning_rate": 7.290564520613496e-05, "loss": 0.1198, "step": 18297 }, { "epoch": 13.405128205128205, "grad_norm": 1.234375, "learning_rate": 7.290258988760561e-05, "loss": 0.1041, "step": 18298 }, { "epoch": 13.405860805860806, "grad_norm": 0.9453125, "learning_rate": 7.289953446173811e-05, "loss": 0.153, "step": 18299 }, { "epoch": 13.406593406593407, "grad_norm": 1.1953125, "learning_rate": 7.28964789285471e-05, "loss": 0.1346, "step": 18300 }, { "epoch": 13.407326007326008, "grad_norm": 1.21875, "learning_rate": 7.28934232880472e-05, "loss": 0.1371, "step": 18301 }, { "epoch": 13.408058608058608, "grad_norm": 1.28125, "learning_rate": 7.289036754025308e-05, "loss": 0.14, "step": 18302 }, { "epoch": 13.40879120879121, "grad_norm": 2.171875, "learning_rate": 7.288731168517934e-05, "loss": 0.1374, "step": 18303 }, { "epoch": 13.40952380952381, "grad_norm": 1.515625, "learning_rate": 7.288425572284064e-05, "loss": 0.1698, "step": 18304 }, { "epoch": 13.41025641025641, "grad_norm": 1.078125, "learning_rate": 7.288119965325162e-05, "loss": 0.1433, "step": 18305 }, { "epoch": 13.410989010989011, "grad_norm": 1.1953125, "learning_rate": 7.287814347642694e-05, "loss": 0.1165, "step": 18306 }, { "epoch": 13.411721611721612, "grad_norm": 0.66015625, "learning_rate": 7.287508719238121e-05, "loss": 0.0992, "step": 18307 }, { "epoch": 13.412454212454213, "grad_norm": 1.1015625, "learning_rate": 7.287203080112909e-05, "loss": 0.1449, "step": 18308 }, { "epoch": 13.413186813186813, "grad_norm": 1.9453125, "learning_rate": 7.286897430268522e-05, "loss": 0.1704, "step": 18309 }, { "epoch": 13.413919413919414, "grad_norm": 1.875, "learning_rate": 7.286591769706428e-05, "loss": 0.161, "step": 18310 }, { "epoch": 13.414652014652015, "grad_norm": 1.1796875, "learning_rate": 7.286286098428087e-05, "loss": 0.1312, "step": 18311 }, { "epoch": 13.415384615384616, "grad_norm": 1.265625, "learning_rate": 7.285980416434964e-05, "loss": 0.1237, "step": 18312 }, { "epoch": 13.416117216117216, "grad_norm": 0.81640625, "learning_rate": 7.285674723728527e-05, "loss": 0.1314, "step": 18313 }, { "epoch": 13.416849816849817, "grad_norm": 1.03125, "learning_rate": 7.285369020310237e-05, "loss": 0.1076, "step": 18314 }, { "epoch": 13.417582417582418, "grad_norm": 1.0, "learning_rate": 7.285063306181562e-05, "loss": 0.1078, "step": 18315 }, { "epoch": 13.418315018315019, "grad_norm": 0.73046875, "learning_rate": 7.284757581343964e-05, "loss": 0.1364, "step": 18316 }, { "epoch": 13.41904761904762, "grad_norm": 1.1796875, "learning_rate": 7.284451845798909e-05, "loss": 0.1062, "step": 18317 }, { "epoch": 13.41978021978022, "grad_norm": 1.3125, "learning_rate": 7.284146099547861e-05, "loss": 0.1455, "step": 18318 }, { "epoch": 13.42051282051282, "grad_norm": 1.125, "learning_rate": 7.283840342592286e-05, "loss": 0.1281, "step": 18319 }, { "epoch": 13.421245421245422, "grad_norm": 0.921875, "learning_rate": 7.283534574933649e-05, "loss": 0.1123, "step": 18320 }, { "epoch": 13.421978021978022, "grad_norm": 0.73828125, "learning_rate": 7.283228796573413e-05, "loss": 0.1253, "step": 18321 }, { "epoch": 13.422710622710623, "grad_norm": 1.0, "learning_rate": 7.282923007513045e-05, "loss": 0.1374, "step": 18322 }, { "epoch": 13.423443223443224, "grad_norm": 0.80859375, "learning_rate": 7.282617207754011e-05, "loss": 0.1042, "step": 18323 }, { "epoch": 13.424175824175824, "grad_norm": 1.5078125, "learning_rate": 7.282311397297775e-05, "loss": 0.1554, "step": 18324 }, { "epoch": 13.424908424908425, "grad_norm": 0.88671875, "learning_rate": 7.282005576145802e-05, "loss": 0.1403, "step": 18325 }, { "epoch": 13.425641025641026, "grad_norm": 0.7890625, "learning_rate": 7.281699744299558e-05, "loss": 0.1308, "step": 18326 }, { "epoch": 13.426373626373627, "grad_norm": 1.140625, "learning_rate": 7.281393901760507e-05, "loss": 0.1228, "step": 18327 }, { "epoch": 13.427106227106227, "grad_norm": 1.2265625, "learning_rate": 7.281088048530115e-05, "loss": 0.1225, "step": 18328 }, { "epoch": 13.427838827838828, "grad_norm": 1.2265625, "learning_rate": 7.280782184609848e-05, "loss": 0.1359, "step": 18329 }, { "epoch": 13.428571428571429, "grad_norm": 1.0859375, "learning_rate": 7.280476310001172e-05, "loss": 0.1472, "step": 18330 }, { "epoch": 13.42930402930403, "grad_norm": 0.95703125, "learning_rate": 7.280170424705552e-05, "loss": 0.1151, "step": 18331 }, { "epoch": 13.43003663003663, "grad_norm": 1.6171875, "learning_rate": 7.279864528724453e-05, "loss": 0.1433, "step": 18332 }, { "epoch": 13.430769230769231, "grad_norm": 0.828125, "learning_rate": 7.279558622059341e-05, "loss": 0.1073, "step": 18333 }, { "epoch": 13.431501831501832, "grad_norm": 0.93359375, "learning_rate": 7.279252704711682e-05, "loss": 0.1292, "step": 18334 }, { "epoch": 13.432234432234432, "grad_norm": 1.9609375, "learning_rate": 7.27894677668294e-05, "loss": 0.1194, "step": 18335 }, { "epoch": 13.432967032967033, "grad_norm": 1.0, "learning_rate": 7.278640837974584e-05, "loss": 0.1131, "step": 18336 }, { "epoch": 13.433699633699634, "grad_norm": 1.3125, "learning_rate": 7.278334888588077e-05, "loss": 0.1746, "step": 18337 }, { "epoch": 13.434432234432235, "grad_norm": 1.1640625, "learning_rate": 7.278028928524888e-05, "loss": 0.1412, "step": 18338 }, { "epoch": 13.435164835164835, "grad_norm": 0.77734375, "learning_rate": 7.277722957786477e-05, "loss": 0.1066, "step": 18339 }, { "epoch": 13.435897435897436, "grad_norm": 1.34375, "learning_rate": 7.277416976374317e-05, "loss": 0.1298, "step": 18340 }, { "epoch": 13.436630036630037, "grad_norm": 0.7890625, "learning_rate": 7.277110984289872e-05, "loss": 0.1061, "step": 18341 }, { "epoch": 13.437362637362638, "grad_norm": 1.1640625, "learning_rate": 7.276804981534605e-05, "loss": 0.1372, "step": 18342 }, { "epoch": 13.438095238095238, "grad_norm": 1.09375, "learning_rate": 7.276498968109985e-05, "loss": 0.1231, "step": 18343 }, { "epoch": 13.438827838827839, "grad_norm": 0.74609375, "learning_rate": 7.276192944017478e-05, "loss": 0.1105, "step": 18344 }, { "epoch": 13.43956043956044, "grad_norm": 1.03125, "learning_rate": 7.275886909258549e-05, "loss": 0.1135, "step": 18345 }, { "epoch": 13.44029304029304, "grad_norm": 0.93359375, "learning_rate": 7.275580863834666e-05, "loss": 0.1254, "step": 18346 }, { "epoch": 13.441025641025641, "grad_norm": 1.3515625, "learning_rate": 7.275274807747293e-05, "loss": 0.125, "step": 18347 }, { "epoch": 13.441758241758242, "grad_norm": 1.0078125, "learning_rate": 7.274968740997899e-05, "loss": 0.1001, "step": 18348 }, { "epoch": 13.442490842490843, "grad_norm": 1.890625, "learning_rate": 7.27466266358795e-05, "loss": 0.1458, "step": 18349 }, { "epoch": 13.443223443223443, "grad_norm": 1.3671875, "learning_rate": 7.274356575518911e-05, "loss": 0.139, "step": 18350 }, { "epoch": 13.443956043956044, "grad_norm": 1.0703125, "learning_rate": 7.274050476792249e-05, "loss": 0.1339, "step": 18351 }, { "epoch": 13.444688644688645, "grad_norm": 1.296875, "learning_rate": 7.273744367409431e-05, "loss": 0.1283, "step": 18352 }, { "epoch": 13.445421245421246, "grad_norm": 1.0703125, "learning_rate": 7.273438247371925e-05, "loss": 0.1276, "step": 18353 }, { "epoch": 13.446153846153846, "grad_norm": 0.640625, "learning_rate": 7.273132116681195e-05, "loss": 0.1066, "step": 18354 }, { "epoch": 13.446886446886447, "grad_norm": 1.4765625, "learning_rate": 7.27282597533871e-05, "loss": 0.1544, "step": 18355 }, { "epoch": 13.447619047619048, "grad_norm": 1.2734375, "learning_rate": 7.272519823345937e-05, "loss": 0.1318, "step": 18356 }, { "epoch": 13.448351648351649, "grad_norm": 1.359375, "learning_rate": 7.272213660704341e-05, "loss": 0.1613, "step": 18357 }, { "epoch": 13.44908424908425, "grad_norm": 0.93359375, "learning_rate": 7.27190748741539e-05, "loss": 0.1276, "step": 18358 }, { "epoch": 13.44981684981685, "grad_norm": 0.67578125, "learning_rate": 7.271601303480551e-05, "loss": 0.1, "step": 18359 }, { "epoch": 13.45054945054945, "grad_norm": 0.64453125, "learning_rate": 7.271295108901289e-05, "loss": 0.1326, "step": 18360 }, { "epoch": 13.451282051282051, "grad_norm": 0.92578125, "learning_rate": 7.270988903679074e-05, "loss": 0.1051, "step": 18361 }, { "epoch": 13.452014652014652, "grad_norm": 1.078125, "learning_rate": 7.270682687815373e-05, "loss": 0.125, "step": 18362 }, { "epoch": 13.452747252747253, "grad_norm": 1.09375, "learning_rate": 7.270376461311651e-05, "loss": 0.1543, "step": 18363 }, { "epoch": 13.453479853479854, "grad_norm": 1.71875, "learning_rate": 7.270070224169377e-05, "loss": 0.1875, "step": 18364 }, { "epoch": 13.454212454212454, "grad_norm": 1.1875, "learning_rate": 7.269763976390018e-05, "loss": 0.1031, "step": 18365 }, { "epoch": 13.454945054945055, "grad_norm": 1.1015625, "learning_rate": 7.269457717975041e-05, "loss": 0.1041, "step": 18366 }, { "epoch": 13.455677655677656, "grad_norm": 1.140625, "learning_rate": 7.269151448925914e-05, "loss": 0.1187, "step": 18367 }, { "epoch": 13.456410256410257, "grad_norm": 0.8203125, "learning_rate": 7.268845169244102e-05, "loss": 0.1138, "step": 18368 }, { "epoch": 13.457142857142857, "grad_norm": 1.875, "learning_rate": 7.268538878931078e-05, "loss": 0.1346, "step": 18369 }, { "epoch": 13.457875457875458, "grad_norm": 1.2421875, "learning_rate": 7.268232577988304e-05, "loss": 0.1736, "step": 18370 }, { "epoch": 13.458608058608059, "grad_norm": 1.3125, "learning_rate": 7.267926266417248e-05, "loss": 0.1184, "step": 18371 }, { "epoch": 13.45934065934066, "grad_norm": 1.25, "learning_rate": 7.267619944219382e-05, "loss": 0.1354, "step": 18372 }, { "epoch": 13.46007326007326, "grad_norm": 1.328125, "learning_rate": 7.267313611396168e-05, "loss": 0.1309, "step": 18373 }, { "epoch": 13.460805860805861, "grad_norm": 1.0625, "learning_rate": 7.267007267949077e-05, "loss": 0.1246, "step": 18374 }, { "epoch": 13.461538461538462, "grad_norm": 1.203125, "learning_rate": 7.266700913879579e-05, "loss": 0.1371, "step": 18375 }, { "epoch": 13.462271062271062, "grad_norm": 0.984375, "learning_rate": 7.266394549189136e-05, "loss": 0.1271, "step": 18376 }, { "epoch": 13.463003663003663, "grad_norm": 1.2578125, "learning_rate": 7.266088173879221e-05, "loss": 0.0909, "step": 18377 }, { "epoch": 13.463736263736264, "grad_norm": 1.3046875, "learning_rate": 7.2657817879513e-05, "loss": 0.1232, "step": 18378 }, { "epoch": 13.464468864468865, "grad_norm": 1.328125, "learning_rate": 7.265475391406841e-05, "loss": 0.1293, "step": 18379 }, { "epoch": 13.465201465201465, "grad_norm": 1.640625, "learning_rate": 7.265168984247311e-05, "loss": 0.1695, "step": 18380 }, { "epoch": 13.465934065934066, "grad_norm": 1.8125, "learning_rate": 7.26486256647418e-05, "loss": 0.1267, "step": 18381 }, { "epoch": 13.466666666666667, "grad_norm": 0.8203125, "learning_rate": 7.264556138088916e-05, "loss": 0.1066, "step": 18382 }, { "epoch": 13.467399267399268, "grad_norm": 0.953125, "learning_rate": 7.264249699092987e-05, "loss": 0.1362, "step": 18383 }, { "epoch": 13.468131868131868, "grad_norm": 0.8359375, "learning_rate": 7.26394324948786e-05, "loss": 0.1107, "step": 18384 }, { "epoch": 13.468864468864469, "grad_norm": 1.265625, "learning_rate": 7.263636789275001e-05, "loss": 0.1422, "step": 18385 }, { "epoch": 13.46959706959707, "grad_norm": 0.84375, "learning_rate": 7.263330318455885e-05, "loss": 0.1359, "step": 18386 }, { "epoch": 13.47032967032967, "grad_norm": 1.0546875, "learning_rate": 7.263023837031975e-05, "loss": 0.1314, "step": 18387 }, { "epoch": 13.471062271062271, "grad_norm": 1.03125, "learning_rate": 7.262717345004744e-05, "loss": 0.1357, "step": 18388 }, { "epoch": 13.471794871794872, "grad_norm": 1.0078125, "learning_rate": 7.262410842375656e-05, "loss": 0.119, "step": 18389 }, { "epoch": 13.472527472527473, "grad_norm": 0.98046875, "learning_rate": 7.262104329146181e-05, "loss": 0.1099, "step": 18390 }, { "epoch": 13.473260073260073, "grad_norm": 1.2421875, "learning_rate": 7.261797805317789e-05, "loss": 0.1412, "step": 18391 }, { "epoch": 13.473992673992674, "grad_norm": 1.4296875, "learning_rate": 7.261491270891946e-05, "loss": 0.1836, "step": 18392 }, { "epoch": 13.474725274725275, "grad_norm": 1.6328125, "learning_rate": 7.261184725870125e-05, "loss": 0.1275, "step": 18393 }, { "epoch": 13.475457875457876, "grad_norm": 1.3203125, "learning_rate": 7.26087817025379e-05, "loss": 0.1364, "step": 18394 }, { "epoch": 13.476190476190476, "grad_norm": 1.2578125, "learning_rate": 7.260571604044413e-05, "loss": 0.1428, "step": 18395 }, { "epoch": 13.476923076923077, "grad_norm": 0.96484375, "learning_rate": 7.260265027243461e-05, "loss": 0.0965, "step": 18396 }, { "epoch": 13.477655677655678, "grad_norm": 0.86328125, "learning_rate": 7.259958439852404e-05, "loss": 0.1412, "step": 18397 }, { "epoch": 13.478388278388278, "grad_norm": 0.7109375, "learning_rate": 7.25965184187271e-05, "loss": 0.1006, "step": 18398 }, { "epoch": 13.47912087912088, "grad_norm": 1.28125, "learning_rate": 7.25934523330585e-05, "loss": 0.1654, "step": 18399 }, { "epoch": 13.47985347985348, "grad_norm": 1.1640625, "learning_rate": 7.25903861415329e-05, "loss": 0.1453, "step": 18400 }, { "epoch": 13.48058608058608, "grad_norm": 1.28125, "learning_rate": 7.258731984416503e-05, "loss": 0.1444, "step": 18401 }, { "epoch": 13.481318681318681, "grad_norm": 0.7421875, "learning_rate": 7.258425344096954e-05, "loss": 0.118, "step": 18402 }, { "epoch": 13.482051282051282, "grad_norm": 0.92578125, "learning_rate": 7.258118693196114e-05, "loss": 0.107, "step": 18403 }, { "epoch": 13.482783882783883, "grad_norm": 0.890625, "learning_rate": 7.257812031715454e-05, "loss": 0.1533, "step": 18404 }, { "epoch": 13.483516483516484, "grad_norm": 0.70703125, "learning_rate": 7.257505359656442e-05, "loss": 0.1039, "step": 18405 }, { "epoch": 13.484249084249084, "grad_norm": 1.046875, "learning_rate": 7.257198677020547e-05, "loss": 0.1297, "step": 18406 }, { "epoch": 13.484981684981685, "grad_norm": 1.3125, "learning_rate": 7.256891983809236e-05, "loss": 0.1247, "step": 18407 }, { "epoch": 13.485714285714286, "grad_norm": 1.203125, "learning_rate": 7.256585280023984e-05, "loss": 0.1085, "step": 18408 }, { "epoch": 13.486446886446886, "grad_norm": 0.87890625, "learning_rate": 7.256278565666257e-05, "loss": 0.1346, "step": 18409 }, { "epoch": 13.487179487179487, "grad_norm": 0.6015625, "learning_rate": 7.255971840737524e-05, "loss": 0.1065, "step": 18410 }, { "epoch": 13.487912087912088, "grad_norm": 1.1015625, "learning_rate": 7.255665105239257e-05, "loss": 0.1517, "step": 18411 }, { "epoch": 13.488644688644689, "grad_norm": 1.3046875, "learning_rate": 7.255358359172923e-05, "loss": 0.1173, "step": 18412 }, { "epoch": 13.48937728937729, "grad_norm": 1.015625, "learning_rate": 7.255051602539994e-05, "loss": 0.1334, "step": 18413 }, { "epoch": 13.49010989010989, "grad_norm": 0.98828125, "learning_rate": 7.25474483534194e-05, "loss": 0.12, "step": 18414 }, { "epoch": 13.49084249084249, "grad_norm": 1.1328125, "learning_rate": 7.254438057580228e-05, "loss": 0.1302, "step": 18415 }, { "epoch": 13.491575091575092, "grad_norm": 2.125, "learning_rate": 7.254131269256328e-05, "loss": 0.1489, "step": 18416 }, { "epoch": 13.492307692307692, "grad_norm": 1.453125, "learning_rate": 7.253824470371714e-05, "loss": 0.156, "step": 18417 }, { "epoch": 13.493040293040293, "grad_norm": 1.09375, "learning_rate": 7.253517660927853e-05, "loss": 0.1019, "step": 18418 }, { "epoch": 13.493772893772894, "grad_norm": 0.640625, "learning_rate": 7.253210840926216e-05, "loss": 0.1107, "step": 18419 }, { "epoch": 13.494505494505495, "grad_norm": 1.171875, "learning_rate": 7.252904010368273e-05, "loss": 0.1189, "step": 18420 }, { "epoch": 13.495238095238095, "grad_norm": 1.3828125, "learning_rate": 7.252597169255492e-05, "loss": 0.1369, "step": 18421 }, { "epoch": 13.495970695970696, "grad_norm": 0.83984375, "learning_rate": 7.252290317589347e-05, "loss": 0.1313, "step": 18422 }, { "epoch": 13.496703296703297, "grad_norm": 1.40625, "learning_rate": 7.251983455371304e-05, "loss": 0.1354, "step": 18423 }, { "epoch": 13.497435897435897, "grad_norm": 1.359375, "learning_rate": 7.251676582602836e-05, "loss": 0.135, "step": 18424 }, { "epoch": 13.498168498168498, "grad_norm": 0.9453125, "learning_rate": 7.251369699285413e-05, "loss": 0.1387, "step": 18425 }, { "epoch": 13.498901098901099, "grad_norm": 1.5703125, "learning_rate": 7.251062805420506e-05, "loss": 0.1443, "step": 18426 }, { "epoch": 13.4996336996337, "grad_norm": 0.796875, "learning_rate": 7.250755901009584e-05, "loss": 0.1366, "step": 18427 }, { "epoch": 13.5003663003663, "grad_norm": 0.95703125, "learning_rate": 7.250448986054116e-05, "loss": 0.1079, "step": 18428 }, { "epoch": 13.501098901098901, "grad_norm": 1.1953125, "learning_rate": 7.250142060555577e-05, "loss": 0.1269, "step": 18429 }, { "epoch": 13.501831501831502, "grad_norm": 1.1171875, "learning_rate": 7.249835124515435e-05, "loss": 0.1195, "step": 18430 }, { "epoch": 13.502564102564103, "grad_norm": 1.40625, "learning_rate": 7.24952817793516e-05, "loss": 0.1112, "step": 18431 }, { "epoch": 13.503296703296703, "grad_norm": 0.6015625, "learning_rate": 7.249221220816225e-05, "loss": 0.1195, "step": 18432 }, { "epoch": 13.504029304029304, "grad_norm": 1.046875, "learning_rate": 7.248914253160096e-05, "loss": 0.1181, "step": 18433 }, { "epoch": 13.504761904761905, "grad_norm": 1.265625, "learning_rate": 7.24860727496825e-05, "loss": 0.1362, "step": 18434 }, { "epoch": 13.505494505494505, "grad_norm": 0.921875, "learning_rate": 7.248300286242154e-05, "loss": 0.1179, "step": 18435 }, { "epoch": 13.506227106227106, "grad_norm": 1.40625, "learning_rate": 7.24799328698328e-05, "loss": 0.1182, "step": 18436 }, { "epoch": 13.506959706959707, "grad_norm": 1.25, "learning_rate": 7.247686277193095e-05, "loss": 0.1233, "step": 18437 }, { "epoch": 13.507692307692308, "grad_norm": 1.609375, "learning_rate": 7.247379256873077e-05, "loss": 0.1469, "step": 18438 }, { "epoch": 13.508424908424908, "grad_norm": 1.2421875, "learning_rate": 7.247072226024693e-05, "loss": 0.1374, "step": 18439 }, { "epoch": 13.50915750915751, "grad_norm": 1.1484375, "learning_rate": 7.246765184649415e-05, "loss": 0.1594, "step": 18440 }, { "epoch": 13.50989010989011, "grad_norm": 1.09375, "learning_rate": 7.246458132748714e-05, "loss": 0.1424, "step": 18441 }, { "epoch": 13.51062271062271, "grad_norm": 0.88671875, "learning_rate": 7.246151070324061e-05, "loss": 0.1066, "step": 18442 }, { "epoch": 13.511355311355311, "grad_norm": 1.140625, "learning_rate": 7.245843997376925e-05, "loss": 0.1179, "step": 18443 }, { "epoch": 13.512087912087912, "grad_norm": 1.234375, "learning_rate": 7.245536913908782e-05, "loss": 0.1217, "step": 18444 }, { "epoch": 13.512820512820513, "grad_norm": 0.7734375, "learning_rate": 7.245229819921101e-05, "loss": 0.1452, "step": 18445 }, { "epoch": 13.513553113553114, "grad_norm": 1.46875, "learning_rate": 7.244922715415353e-05, "loss": 0.1501, "step": 18446 }, { "epoch": 13.514285714285714, "grad_norm": 1.1875, "learning_rate": 7.24461560039301e-05, "loss": 0.1155, "step": 18447 }, { "epoch": 13.515018315018315, "grad_norm": 1.234375, "learning_rate": 7.244308474855543e-05, "loss": 0.1436, "step": 18448 }, { "epoch": 13.515750915750916, "grad_norm": 0.98828125, "learning_rate": 7.244001338804422e-05, "loss": 0.1321, "step": 18449 }, { "epoch": 13.516483516483516, "grad_norm": 0.96484375, "learning_rate": 7.243694192241121e-05, "loss": 0.1254, "step": 18450 }, { "epoch": 13.517216117216117, "grad_norm": 1.328125, "learning_rate": 7.243387035167112e-05, "loss": 0.1411, "step": 18451 }, { "epoch": 13.517948717948718, "grad_norm": 1.6875, "learning_rate": 7.243079867583866e-05, "loss": 0.1206, "step": 18452 }, { "epoch": 13.518681318681319, "grad_norm": 1.0703125, "learning_rate": 7.242772689492854e-05, "loss": 0.1586, "step": 18453 }, { "epoch": 13.51941391941392, "grad_norm": 1.984375, "learning_rate": 7.242465500895546e-05, "loss": 0.1398, "step": 18454 }, { "epoch": 13.52014652014652, "grad_norm": 1.8515625, "learning_rate": 7.24215830179342e-05, "loss": 0.1291, "step": 18455 }, { "epoch": 13.52087912087912, "grad_norm": 0.828125, "learning_rate": 7.24185109218794e-05, "loss": 0.119, "step": 18456 }, { "epoch": 13.521611721611722, "grad_norm": 1.4296875, "learning_rate": 7.241543872080585e-05, "loss": 0.1579, "step": 18457 }, { "epoch": 13.522344322344322, "grad_norm": 1.3671875, "learning_rate": 7.241236641472823e-05, "loss": 0.1184, "step": 18458 }, { "epoch": 13.523076923076923, "grad_norm": 1.640625, "learning_rate": 7.240929400366126e-05, "loss": 0.1146, "step": 18459 }, { "epoch": 13.523809523809524, "grad_norm": 1.1953125, "learning_rate": 7.240622148761969e-05, "loss": 0.1333, "step": 18460 }, { "epoch": 13.524542124542124, "grad_norm": 1.0703125, "learning_rate": 7.240314886661822e-05, "loss": 0.1271, "step": 18461 }, { "epoch": 13.525274725274725, "grad_norm": 1.1484375, "learning_rate": 7.240007614067155e-05, "loss": 0.1355, "step": 18462 }, { "epoch": 13.526007326007326, "grad_norm": 0.75, "learning_rate": 7.239700330979443e-05, "loss": 0.1133, "step": 18463 }, { "epoch": 13.526739926739927, "grad_norm": 1.5703125, "learning_rate": 7.239393037400158e-05, "loss": 0.1264, "step": 18464 }, { "epoch": 13.527472527472527, "grad_norm": 1.703125, "learning_rate": 7.239085733330774e-05, "loss": 0.1299, "step": 18465 }, { "epoch": 13.528205128205128, "grad_norm": 1.21875, "learning_rate": 7.238778418772761e-05, "loss": 0.124, "step": 18466 }, { "epoch": 13.528937728937729, "grad_norm": 1.5859375, "learning_rate": 7.23847109372759e-05, "loss": 0.1118, "step": 18467 }, { "epoch": 13.52967032967033, "grad_norm": 1.6484375, "learning_rate": 7.238163758196738e-05, "loss": 0.1123, "step": 18468 }, { "epoch": 13.53040293040293, "grad_norm": 1.65625, "learning_rate": 7.237856412181673e-05, "loss": 0.1211, "step": 18469 }, { "epoch": 13.531135531135531, "grad_norm": 1.1328125, "learning_rate": 7.23754905568387e-05, "loss": 0.1447, "step": 18470 }, { "epoch": 13.531868131868132, "grad_norm": 0.44921875, "learning_rate": 7.237241688704804e-05, "loss": 0.084, "step": 18471 }, { "epoch": 13.532600732600732, "grad_norm": 0.70703125, "learning_rate": 7.236934311245944e-05, "loss": 0.1299, "step": 18472 }, { "epoch": 13.533333333333333, "grad_norm": 1.28125, "learning_rate": 7.236626923308762e-05, "loss": 0.1308, "step": 18473 }, { "epoch": 13.534065934065934, "grad_norm": 1.4609375, "learning_rate": 7.236319524894735e-05, "loss": 0.1482, "step": 18474 }, { "epoch": 13.534798534798535, "grad_norm": 0.8984375, "learning_rate": 7.236012116005331e-05, "loss": 0.114, "step": 18475 }, { "epoch": 13.535531135531135, "grad_norm": 1.171875, "learning_rate": 7.235704696642027e-05, "loss": 0.1595, "step": 18476 }, { "epoch": 13.536263736263736, "grad_norm": 1.1875, "learning_rate": 7.235397266806292e-05, "loss": 0.1414, "step": 18477 }, { "epoch": 13.536996336996337, "grad_norm": 1.0078125, "learning_rate": 7.235089826499603e-05, "loss": 0.1204, "step": 18478 }, { "epoch": 13.537728937728938, "grad_norm": 0.69921875, "learning_rate": 7.234782375723432e-05, "loss": 0.1131, "step": 18479 }, { "epoch": 13.538461538461538, "grad_norm": 1.5, "learning_rate": 7.234474914479252e-05, "loss": 0.1487, "step": 18480 }, { "epoch": 13.539194139194139, "grad_norm": 0.921875, "learning_rate": 7.234167442768533e-05, "loss": 0.1144, "step": 18481 }, { "epoch": 13.53992673992674, "grad_norm": 1.0234375, "learning_rate": 7.233859960592753e-05, "loss": 0.1135, "step": 18482 }, { "epoch": 13.54065934065934, "grad_norm": 1.546875, "learning_rate": 7.23355246795338e-05, "loss": 0.1416, "step": 18483 }, { "epoch": 13.541391941391941, "grad_norm": 0.93359375, "learning_rate": 7.233244964851894e-05, "loss": 0.1097, "step": 18484 }, { "epoch": 13.542124542124542, "grad_norm": 1.3125, "learning_rate": 7.232937451289763e-05, "loss": 0.1384, "step": 18485 }, { "epoch": 13.542857142857143, "grad_norm": 0.7890625, "learning_rate": 7.232629927268462e-05, "loss": 0.1161, "step": 18486 }, { "epoch": 13.543589743589743, "grad_norm": 0.8828125, "learning_rate": 7.232322392789465e-05, "loss": 0.1318, "step": 18487 }, { "epoch": 13.544322344322344, "grad_norm": 1.4375, "learning_rate": 7.232014847854245e-05, "loss": 0.0943, "step": 18488 }, { "epoch": 13.545054945054945, "grad_norm": 0.98828125, "learning_rate": 7.231707292464274e-05, "loss": 0.1264, "step": 18489 }, { "epoch": 13.545787545787546, "grad_norm": 1.640625, "learning_rate": 7.231399726621028e-05, "loss": 0.1206, "step": 18490 }, { "epoch": 13.546520146520146, "grad_norm": 0.76171875, "learning_rate": 7.231092150325981e-05, "loss": 0.1262, "step": 18491 }, { "epoch": 13.547252747252747, "grad_norm": 0.6171875, "learning_rate": 7.230784563580603e-05, "loss": 0.1334, "step": 18492 }, { "epoch": 13.547985347985348, "grad_norm": 1.671875, "learning_rate": 7.230476966386373e-05, "loss": 0.1284, "step": 18493 }, { "epoch": 13.548717948717949, "grad_norm": 1.0625, "learning_rate": 7.230169358744762e-05, "loss": 0.1401, "step": 18494 }, { "epoch": 13.54945054945055, "grad_norm": 0.90625, "learning_rate": 7.229861740657242e-05, "loss": 0.1522, "step": 18495 }, { "epoch": 13.55018315018315, "grad_norm": 1.0390625, "learning_rate": 7.229554112125289e-05, "loss": 0.1327, "step": 18496 }, { "epoch": 13.55091575091575, "grad_norm": 1.2265625, "learning_rate": 7.22924647315038e-05, "loss": 0.1142, "step": 18497 }, { "epoch": 13.551648351648351, "grad_norm": 0.6875, "learning_rate": 7.228938823733982e-05, "loss": 0.1013, "step": 18498 }, { "epoch": 13.552380952380952, "grad_norm": 0.828125, "learning_rate": 7.228631163877574e-05, "loss": 0.1067, "step": 18499 }, { "epoch": 13.553113553113553, "grad_norm": 1.2265625, "learning_rate": 7.228323493582629e-05, "loss": 0.1375, "step": 18500 }, { "epoch": 13.553846153846154, "grad_norm": 1.5703125, "learning_rate": 7.22801581285062e-05, "loss": 0.159, "step": 18501 }, { "epoch": 13.554578754578754, "grad_norm": 1.34375, "learning_rate": 7.227708121683023e-05, "loss": 0.1065, "step": 18502 }, { "epoch": 13.555311355311355, "grad_norm": 1.0703125, "learning_rate": 7.227400420081311e-05, "loss": 0.1146, "step": 18503 }, { "epoch": 13.556043956043956, "grad_norm": 1.484375, "learning_rate": 7.22709270804696e-05, "loss": 0.1786, "step": 18504 }, { "epoch": 13.556776556776557, "grad_norm": 0.8984375, "learning_rate": 7.226784985581443e-05, "loss": 0.1298, "step": 18505 }, { "epoch": 13.557509157509157, "grad_norm": 1.265625, "learning_rate": 7.226477252686233e-05, "loss": 0.1438, "step": 18506 }, { "epoch": 13.558241758241758, "grad_norm": 0.890625, "learning_rate": 7.226169509362807e-05, "loss": 0.121, "step": 18507 }, { "epoch": 13.558974358974359, "grad_norm": 1.421875, "learning_rate": 7.225861755612639e-05, "loss": 0.1323, "step": 18508 }, { "epoch": 13.55970695970696, "grad_norm": 1.078125, "learning_rate": 7.225553991437203e-05, "loss": 0.1095, "step": 18509 }, { "epoch": 13.56043956043956, "grad_norm": 1.03125, "learning_rate": 7.225246216837974e-05, "loss": 0.1711, "step": 18510 }, { "epoch": 13.561172161172161, "grad_norm": 1.375, "learning_rate": 7.224938431816425e-05, "loss": 0.1487, "step": 18511 }, { "epoch": 13.561904761904762, "grad_norm": 0.875, "learning_rate": 7.224630636374032e-05, "loss": 0.1446, "step": 18512 }, { "epoch": 13.562637362637362, "grad_norm": 1.359375, "learning_rate": 7.22432283051227e-05, "loss": 0.1369, "step": 18513 }, { "epoch": 13.563369963369963, "grad_norm": 1.171875, "learning_rate": 7.224015014232614e-05, "loss": 0.1558, "step": 18514 }, { "epoch": 13.564102564102564, "grad_norm": 1.59375, "learning_rate": 7.223707187536538e-05, "loss": 0.1265, "step": 18515 }, { "epoch": 13.564835164835165, "grad_norm": 1.046875, "learning_rate": 7.223399350425517e-05, "loss": 0.1074, "step": 18516 }, { "epoch": 13.565567765567765, "grad_norm": 1.21875, "learning_rate": 7.223091502901028e-05, "loss": 0.121, "step": 18517 }, { "epoch": 13.566300366300366, "grad_norm": 0.82421875, "learning_rate": 7.222783644964543e-05, "loss": 0.1529, "step": 18518 }, { "epoch": 13.567032967032967, "grad_norm": 0.78515625, "learning_rate": 7.222475776617536e-05, "loss": 0.1225, "step": 18519 }, { "epoch": 13.567765567765568, "grad_norm": 1.2734375, "learning_rate": 7.222167897861488e-05, "loss": 0.1231, "step": 18520 }, { "epoch": 13.568498168498168, "grad_norm": 1.328125, "learning_rate": 7.22186000869787e-05, "loss": 0.1204, "step": 18521 }, { "epoch": 13.569230769230769, "grad_norm": 1.015625, "learning_rate": 7.221552109128157e-05, "loss": 0.1819, "step": 18522 }, { "epoch": 13.56996336996337, "grad_norm": 1.296875, "learning_rate": 7.221244199153824e-05, "loss": 0.1098, "step": 18523 }, { "epoch": 13.57069597069597, "grad_norm": 0.8046875, "learning_rate": 7.220936278776348e-05, "loss": 0.1114, "step": 18524 }, { "epoch": 13.571428571428571, "grad_norm": 1.4453125, "learning_rate": 7.220628347997203e-05, "loss": 0.1325, "step": 18525 }, { "epoch": 13.572161172161172, "grad_norm": 0.9765625, "learning_rate": 7.220320406817866e-05, "loss": 0.1178, "step": 18526 }, { "epoch": 13.572893772893773, "grad_norm": 1.359375, "learning_rate": 7.220012455239812e-05, "loss": 0.1344, "step": 18527 }, { "epoch": 13.573626373626373, "grad_norm": 1.34375, "learning_rate": 7.219704493264514e-05, "loss": 0.1347, "step": 18528 }, { "epoch": 13.574358974358974, "grad_norm": 1.484375, "learning_rate": 7.219396520893451e-05, "loss": 0.1301, "step": 18529 }, { "epoch": 13.575091575091575, "grad_norm": 1.3203125, "learning_rate": 7.219088538128097e-05, "loss": 0.1209, "step": 18530 }, { "epoch": 13.575824175824176, "grad_norm": 0.84375, "learning_rate": 7.218780544969928e-05, "loss": 0.1511, "step": 18531 }, { "epoch": 13.576556776556776, "grad_norm": 1.375, "learning_rate": 7.218472541420418e-05, "loss": 0.1581, "step": 18532 }, { "epoch": 13.577289377289377, "grad_norm": 1.8984375, "learning_rate": 7.218164527481046e-05, "loss": 0.2003, "step": 18533 }, { "epoch": 13.578021978021978, "grad_norm": 1.1328125, "learning_rate": 7.217856503153285e-05, "loss": 0.1484, "step": 18534 }, { "epoch": 13.578754578754578, "grad_norm": 0.79296875, "learning_rate": 7.217548468438612e-05, "loss": 0.1324, "step": 18535 }, { "epoch": 13.57948717948718, "grad_norm": 1.3046875, "learning_rate": 7.217240423338503e-05, "loss": 0.1112, "step": 18536 }, { "epoch": 13.58021978021978, "grad_norm": 0.9609375, "learning_rate": 7.216932367854431e-05, "loss": 0.126, "step": 18537 }, { "epoch": 13.58095238095238, "grad_norm": 1.34375, "learning_rate": 7.216624301987878e-05, "loss": 0.1128, "step": 18538 }, { "epoch": 13.581684981684981, "grad_norm": 1.578125, "learning_rate": 7.216316225740316e-05, "loss": 0.1841, "step": 18539 }, { "epoch": 13.582417582417582, "grad_norm": 1.1171875, "learning_rate": 7.216008139113221e-05, "loss": 0.1242, "step": 18540 }, { "epoch": 13.583150183150183, "grad_norm": 1.546875, "learning_rate": 7.215700042108071e-05, "loss": 0.1359, "step": 18541 }, { "epoch": 13.583882783882784, "grad_norm": 1.0625, "learning_rate": 7.21539193472634e-05, "loss": 0.1194, "step": 18542 }, { "epoch": 13.584615384615384, "grad_norm": 1.171875, "learning_rate": 7.215083816969507e-05, "loss": 0.1393, "step": 18543 }, { "epoch": 13.585347985347985, "grad_norm": 1.078125, "learning_rate": 7.214775688839045e-05, "loss": 0.0924, "step": 18544 }, { "epoch": 13.586080586080586, "grad_norm": 1.359375, "learning_rate": 7.214467550336432e-05, "loss": 0.131, "step": 18545 }, { "epoch": 13.586813186813187, "grad_norm": 1.2734375, "learning_rate": 7.214159401463147e-05, "loss": 0.1236, "step": 18546 }, { "epoch": 13.587545787545787, "grad_norm": 1.609375, "learning_rate": 7.213851242220662e-05, "loss": 0.1344, "step": 18547 }, { "epoch": 13.588278388278388, "grad_norm": 1.3671875, "learning_rate": 7.213543072610455e-05, "loss": 0.1141, "step": 18548 }, { "epoch": 13.589010989010989, "grad_norm": 0.77734375, "learning_rate": 7.213234892634003e-05, "loss": 0.1253, "step": 18549 }, { "epoch": 13.58974358974359, "grad_norm": 1.328125, "learning_rate": 7.212926702292783e-05, "loss": 0.1219, "step": 18550 }, { "epoch": 13.59047619047619, "grad_norm": 1.2109375, "learning_rate": 7.212618501588271e-05, "loss": 0.1092, "step": 18551 }, { "epoch": 13.59120879120879, "grad_norm": 1.0234375, "learning_rate": 7.212310290521943e-05, "loss": 0.1327, "step": 18552 }, { "epoch": 13.591941391941392, "grad_norm": 1.71875, "learning_rate": 7.212002069095278e-05, "loss": 0.1872, "step": 18553 }, { "epoch": 13.592673992673992, "grad_norm": 1.1640625, "learning_rate": 7.211693837309749e-05, "loss": 0.1191, "step": 18554 }, { "epoch": 13.593406593406593, "grad_norm": 0.92578125, "learning_rate": 7.211385595166837e-05, "loss": 0.1055, "step": 18555 }, { "epoch": 13.594139194139194, "grad_norm": 1.3359375, "learning_rate": 7.211077342668016e-05, "loss": 0.1364, "step": 18556 }, { "epoch": 13.594871794871795, "grad_norm": 1.1875, "learning_rate": 7.210769079814767e-05, "loss": 0.1406, "step": 18557 }, { "epoch": 13.595604395604395, "grad_norm": 1.0, "learning_rate": 7.210460806608561e-05, "loss": 0.1431, "step": 18558 }, { "epoch": 13.596336996336996, "grad_norm": 0.95703125, "learning_rate": 7.210152523050879e-05, "loss": 0.1311, "step": 18559 }, { "epoch": 13.597069597069597, "grad_norm": 1.2109375, "learning_rate": 7.209844229143198e-05, "loss": 0.133, "step": 18560 }, { "epoch": 13.597802197802197, "grad_norm": 0.828125, "learning_rate": 7.209535924886992e-05, "loss": 0.1156, "step": 18561 }, { "epoch": 13.598534798534798, "grad_norm": 1.09375, "learning_rate": 7.209227610283743e-05, "loss": 0.1584, "step": 18562 }, { "epoch": 13.599267399267399, "grad_norm": 0.96484375, "learning_rate": 7.208919285334924e-05, "loss": 0.133, "step": 18563 }, { "epoch": 13.6, "grad_norm": 1.8828125, "learning_rate": 7.208610950042015e-05, "loss": 0.139, "step": 18564 }, { "epoch": 13.6007326007326, "grad_norm": 0.7734375, "learning_rate": 7.20830260440649e-05, "loss": 0.1028, "step": 18565 }, { "epoch": 13.601465201465201, "grad_norm": 1.59375, "learning_rate": 7.207994248429831e-05, "loss": 0.1728, "step": 18566 }, { "epoch": 13.602197802197802, "grad_norm": 1.171875, "learning_rate": 7.207685882113513e-05, "loss": 0.1379, "step": 18567 }, { "epoch": 13.602930402930403, "grad_norm": 1.03125, "learning_rate": 7.207377505459011e-05, "loss": 0.1074, "step": 18568 }, { "epoch": 13.603663003663003, "grad_norm": 1.1875, "learning_rate": 7.20706911846781e-05, "loss": 0.1595, "step": 18569 }, { "epoch": 13.604395604395604, "grad_norm": 0.87109375, "learning_rate": 7.20676072114138e-05, "loss": 0.1061, "step": 18570 }, { "epoch": 13.605128205128205, "grad_norm": 0.81640625, "learning_rate": 7.2064523134812e-05, "loss": 0.1049, "step": 18571 }, { "epoch": 13.605860805860805, "grad_norm": 1.1328125, "learning_rate": 7.206143895488751e-05, "loss": 0.1054, "step": 18572 }, { "epoch": 13.606593406593406, "grad_norm": 1.1015625, "learning_rate": 7.205835467165509e-05, "loss": 0.1169, "step": 18573 }, { "epoch": 13.607326007326007, "grad_norm": 0.92578125, "learning_rate": 7.20552702851295e-05, "loss": 0.1315, "step": 18574 }, { "epoch": 13.608058608058608, "grad_norm": 0.6015625, "learning_rate": 7.205218579532556e-05, "loss": 0.1048, "step": 18575 }, { "epoch": 13.608791208791208, "grad_norm": 1.3203125, "learning_rate": 7.2049101202258e-05, "loss": 0.1463, "step": 18576 }, { "epoch": 13.60952380952381, "grad_norm": 0.56640625, "learning_rate": 7.204601650594161e-05, "loss": 0.1143, "step": 18577 }, { "epoch": 13.61025641025641, "grad_norm": 0.79296875, "learning_rate": 7.204293170639121e-05, "loss": 0.1527, "step": 18578 }, { "epoch": 13.61098901098901, "grad_norm": 0.9765625, "learning_rate": 7.203984680362152e-05, "loss": 0.1307, "step": 18579 }, { "epoch": 13.611721611721611, "grad_norm": 0.80859375, "learning_rate": 7.203676179764738e-05, "loss": 0.1126, "step": 18580 }, { "epoch": 13.612454212454212, "grad_norm": 1.0234375, "learning_rate": 7.203367668848355e-05, "loss": 0.1312, "step": 18581 }, { "epoch": 13.613186813186813, "grad_norm": 1.0078125, "learning_rate": 7.20305914761448e-05, "loss": 0.1438, "step": 18582 }, { "epoch": 13.613919413919414, "grad_norm": 0.84375, "learning_rate": 7.20275061606459e-05, "loss": 0.1113, "step": 18583 }, { "epoch": 13.614652014652014, "grad_norm": 0.6484375, "learning_rate": 7.202442074200166e-05, "loss": 0.116, "step": 18584 }, { "epoch": 13.615384615384615, "grad_norm": 1.234375, "learning_rate": 7.202133522022686e-05, "loss": 0.16, "step": 18585 }, { "epoch": 13.616117216117216, "grad_norm": 1.578125, "learning_rate": 7.201824959533629e-05, "loss": 0.128, "step": 18586 }, { "epoch": 13.616849816849816, "grad_norm": 1.40625, "learning_rate": 7.20151638673447e-05, "loss": 0.152, "step": 18587 }, { "epoch": 13.617582417582417, "grad_norm": 1.1484375, "learning_rate": 7.20120780362669e-05, "loss": 0.1336, "step": 18588 }, { "epoch": 13.618315018315018, "grad_norm": 1.6171875, "learning_rate": 7.200899210211768e-05, "loss": 0.1487, "step": 18589 }, { "epoch": 13.619047619047619, "grad_norm": 1.2109375, "learning_rate": 7.200590606491182e-05, "loss": 0.1737, "step": 18590 }, { "epoch": 13.61978021978022, "grad_norm": 1.4375, "learning_rate": 7.200281992466409e-05, "loss": 0.1475, "step": 18591 }, { "epoch": 13.62051282051282, "grad_norm": 0.8203125, "learning_rate": 7.19997336813893e-05, "loss": 0.1049, "step": 18592 }, { "epoch": 13.62124542124542, "grad_norm": 0.9140625, "learning_rate": 7.199664733510224e-05, "loss": 0.1019, "step": 18593 }, { "epoch": 13.621978021978022, "grad_norm": 0.79296875, "learning_rate": 7.199356088581766e-05, "loss": 0.1097, "step": 18594 }, { "epoch": 13.622710622710622, "grad_norm": 0.67578125, "learning_rate": 7.199047433355039e-05, "loss": 0.1045, "step": 18595 }, { "epoch": 13.623443223443223, "grad_norm": 1.1640625, "learning_rate": 7.19873876783152e-05, "loss": 0.111, "step": 18596 }, { "epoch": 13.624175824175824, "grad_norm": 1.2265625, "learning_rate": 7.198430092012686e-05, "loss": 0.123, "step": 18597 }, { "epoch": 13.624908424908424, "grad_norm": 1.1875, "learning_rate": 7.198121405900022e-05, "loss": 0.1403, "step": 18598 }, { "epoch": 13.625641025641025, "grad_norm": 1.3203125, "learning_rate": 7.197812709495002e-05, "loss": 0.1176, "step": 18599 }, { "epoch": 13.626373626373626, "grad_norm": 1.0, "learning_rate": 7.197504002799104e-05, "loss": 0.1404, "step": 18600 }, { "epoch": 13.627106227106227, "grad_norm": 0.9296875, "learning_rate": 7.197195285813813e-05, "loss": 0.1217, "step": 18601 }, { "epoch": 13.627838827838827, "grad_norm": 1.046875, "learning_rate": 7.196886558540602e-05, "loss": 0.11, "step": 18602 }, { "epoch": 13.628571428571428, "grad_norm": 1.2578125, "learning_rate": 7.196577820980953e-05, "loss": 0.1447, "step": 18603 }, { "epoch": 13.629304029304029, "grad_norm": 0.890625, "learning_rate": 7.196269073136345e-05, "loss": 0.107, "step": 18604 }, { "epoch": 13.63003663003663, "grad_norm": 1.0078125, "learning_rate": 7.195960315008258e-05, "loss": 0.1448, "step": 18605 }, { "epoch": 13.63076923076923, "grad_norm": 0.6484375, "learning_rate": 7.19565154659817e-05, "loss": 0.096, "step": 18606 }, { "epoch": 13.631501831501831, "grad_norm": 1.4296875, "learning_rate": 7.19534276790756e-05, "loss": 0.1405, "step": 18607 }, { "epoch": 13.632234432234432, "grad_norm": 1.484375, "learning_rate": 7.19503397893791e-05, "loss": 0.15, "step": 18608 }, { "epoch": 13.632967032967032, "grad_norm": 1.109375, "learning_rate": 7.1947251796907e-05, "loss": 0.1145, "step": 18609 }, { "epoch": 13.633699633699633, "grad_norm": 1.546875, "learning_rate": 7.194416370167404e-05, "loss": 0.1327, "step": 18610 }, { "epoch": 13.634432234432234, "grad_norm": 0.84375, "learning_rate": 7.194107550369507e-05, "loss": 0.1213, "step": 18611 }, { "epoch": 13.635164835164835, "grad_norm": 2.421875, "learning_rate": 7.193798720298487e-05, "loss": 0.1323, "step": 18612 }, { "epoch": 13.635897435897435, "grad_norm": 0.7421875, "learning_rate": 7.193489879955822e-05, "loss": 0.1236, "step": 18613 }, { "epoch": 13.636630036630036, "grad_norm": 1.2890625, "learning_rate": 7.193181029342996e-05, "loss": 0.1214, "step": 18614 }, { "epoch": 13.637362637362637, "grad_norm": 1.234375, "learning_rate": 7.192872168461484e-05, "loss": 0.1055, "step": 18615 }, { "epoch": 13.638095238095238, "grad_norm": 1.2734375, "learning_rate": 7.192563297312769e-05, "loss": 0.1299, "step": 18616 }, { "epoch": 13.638827838827838, "grad_norm": 1.2421875, "learning_rate": 7.192254415898329e-05, "loss": 0.1219, "step": 18617 }, { "epoch": 13.639560439560439, "grad_norm": 1.421875, "learning_rate": 7.191945524219643e-05, "loss": 0.1097, "step": 18618 }, { "epoch": 13.64029304029304, "grad_norm": 2.5, "learning_rate": 7.191636622278197e-05, "loss": 0.165, "step": 18619 }, { "epoch": 13.64102564102564, "grad_norm": 1.71875, "learning_rate": 7.191327710075464e-05, "loss": 0.14, "step": 18620 }, { "epoch": 13.641758241758241, "grad_norm": 1.7734375, "learning_rate": 7.191018787612928e-05, "loss": 0.1479, "step": 18621 }, { "epoch": 13.642490842490842, "grad_norm": 1.3125, "learning_rate": 7.19070985489207e-05, "loss": 0.1272, "step": 18622 }, { "epoch": 13.643223443223443, "grad_norm": 2.15625, "learning_rate": 7.190400911914365e-05, "loss": 0.1286, "step": 18623 }, { "epoch": 13.643956043956043, "grad_norm": 1.7265625, "learning_rate": 7.190091958681299e-05, "loss": 0.1132, "step": 18624 }, { "epoch": 13.644688644688644, "grad_norm": 1.46875, "learning_rate": 7.18978299519435e-05, "loss": 0.1585, "step": 18625 }, { "epoch": 13.645421245421245, "grad_norm": 1.4140625, "learning_rate": 7.189474021454995e-05, "loss": 0.1098, "step": 18626 }, { "epoch": 13.646153846153846, "grad_norm": 1.2734375, "learning_rate": 7.189165037464721e-05, "loss": 0.1183, "step": 18627 }, { "epoch": 13.646886446886446, "grad_norm": 1.7890625, "learning_rate": 7.188856043225003e-05, "loss": 0.1256, "step": 18628 }, { "epoch": 13.647619047619047, "grad_norm": 1.828125, "learning_rate": 7.188547038737324e-05, "loss": 0.123, "step": 18629 }, { "epoch": 13.648351648351648, "grad_norm": 1.34375, "learning_rate": 7.188238024003164e-05, "loss": 0.1376, "step": 18630 }, { "epoch": 13.649084249084249, "grad_norm": 0.67578125, "learning_rate": 7.187928999024004e-05, "loss": 0.1088, "step": 18631 }, { "epoch": 13.64981684981685, "grad_norm": 1.4375, "learning_rate": 7.187619963801325e-05, "loss": 0.1335, "step": 18632 }, { "epoch": 13.65054945054945, "grad_norm": 0.9140625, "learning_rate": 7.187310918336606e-05, "loss": 0.1193, "step": 18633 }, { "epoch": 13.65128205128205, "grad_norm": 0.8203125, "learning_rate": 7.18700186263133e-05, "loss": 0.1203, "step": 18634 }, { "epoch": 13.652014652014651, "grad_norm": 1.265625, "learning_rate": 7.186692796686976e-05, "loss": 0.1259, "step": 18635 }, { "epoch": 13.652747252747252, "grad_norm": 0.68359375, "learning_rate": 7.186383720505024e-05, "loss": 0.1119, "step": 18636 }, { "epoch": 13.653479853479853, "grad_norm": 0.8828125, "learning_rate": 7.186074634086959e-05, "loss": 0.1532, "step": 18637 }, { "epoch": 13.654212454212454, "grad_norm": 0.90625, "learning_rate": 7.185765537434258e-05, "loss": 0.1206, "step": 18638 }, { "epoch": 13.654945054945054, "grad_norm": 1.5078125, "learning_rate": 7.185456430548402e-05, "loss": 0.1381, "step": 18639 }, { "epoch": 13.655677655677655, "grad_norm": 1.1484375, "learning_rate": 7.185147313430875e-05, "loss": 0.134, "step": 18640 }, { "epoch": 13.656410256410256, "grad_norm": 2.078125, "learning_rate": 7.184838186083154e-05, "loss": 0.1437, "step": 18641 }, { "epoch": 13.657142857142857, "grad_norm": 1.421875, "learning_rate": 7.184529048506725e-05, "loss": 0.1212, "step": 18642 }, { "epoch": 13.657875457875457, "grad_norm": 1.296875, "learning_rate": 7.184219900703064e-05, "loss": 0.1474, "step": 18643 }, { "epoch": 13.658608058608058, "grad_norm": 1.53125, "learning_rate": 7.183910742673656e-05, "loss": 0.1496, "step": 18644 }, { "epoch": 13.659340659340659, "grad_norm": 0.94140625, "learning_rate": 7.183601574419982e-05, "loss": 0.1108, "step": 18645 }, { "epoch": 13.66007326007326, "grad_norm": 1.0625, "learning_rate": 7.18329239594352e-05, "loss": 0.1181, "step": 18646 }, { "epoch": 13.66080586080586, "grad_norm": 1.3828125, "learning_rate": 7.182983207245757e-05, "loss": 0.137, "step": 18647 }, { "epoch": 13.661538461538461, "grad_norm": 0.625, "learning_rate": 7.182674008328171e-05, "loss": 0.12, "step": 18648 }, { "epoch": 13.662271062271062, "grad_norm": 1.2734375, "learning_rate": 7.18236479919224e-05, "loss": 0.1157, "step": 18649 }, { "epoch": 13.663003663003662, "grad_norm": 0.94921875, "learning_rate": 7.182055579839453e-05, "loss": 0.1198, "step": 18650 }, { "epoch": 13.663736263736263, "grad_norm": 0.98046875, "learning_rate": 7.181746350271288e-05, "loss": 0.1255, "step": 18651 }, { "epoch": 13.664468864468864, "grad_norm": 0.8046875, "learning_rate": 7.181437110489225e-05, "loss": 0.1029, "step": 18652 }, { "epoch": 13.665201465201465, "grad_norm": 1.6640625, "learning_rate": 7.181127860494747e-05, "loss": 0.0968, "step": 18653 }, { "epoch": 13.665934065934065, "grad_norm": 1.578125, "learning_rate": 7.180818600289335e-05, "loss": 0.1359, "step": 18654 }, { "epoch": 13.666666666666666, "grad_norm": 0.89453125, "learning_rate": 7.180509329874473e-05, "loss": 0.1286, "step": 18655 }, { "epoch": 13.667399267399267, "grad_norm": 1.1171875, "learning_rate": 7.18020004925164e-05, "loss": 0.1225, "step": 18656 }, { "epoch": 13.668131868131868, "grad_norm": 1.4296875, "learning_rate": 7.17989075842232e-05, "loss": 0.1225, "step": 18657 }, { "epoch": 13.668864468864468, "grad_norm": 0.87109375, "learning_rate": 7.179581457387993e-05, "loss": 0.1288, "step": 18658 }, { "epoch": 13.669597069597069, "grad_norm": 1.0625, "learning_rate": 7.179272146150144e-05, "loss": 0.1386, "step": 18659 }, { "epoch": 13.67032967032967, "grad_norm": 1.234375, "learning_rate": 7.178962824710253e-05, "loss": 0.1201, "step": 18660 }, { "epoch": 13.67106227106227, "grad_norm": 0.9140625, "learning_rate": 7.178653493069804e-05, "loss": 0.1461, "step": 18661 }, { "epoch": 13.671794871794871, "grad_norm": 1.34375, "learning_rate": 7.178344151230274e-05, "loss": 0.1117, "step": 18662 }, { "epoch": 13.672527472527472, "grad_norm": 1.6015625, "learning_rate": 7.17803479919315e-05, "loss": 0.1633, "step": 18663 }, { "epoch": 13.673260073260073, "grad_norm": 1.3203125, "learning_rate": 7.177725436959913e-05, "loss": 0.1378, "step": 18664 }, { "epoch": 13.673992673992673, "grad_norm": 1.3515625, "learning_rate": 7.177416064532045e-05, "loss": 0.1597, "step": 18665 }, { "epoch": 13.674725274725274, "grad_norm": 1.1875, "learning_rate": 7.177106681911026e-05, "loss": 0.1429, "step": 18666 }, { "epoch": 13.675457875457875, "grad_norm": 1.1875, "learning_rate": 7.176797289098341e-05, "loss": 0.1837, "step": 18667 }, { "epoch": 13.676190476190476, "grad_norm": 0.93359375, "learning_rate": 7.176487886095474e-05, "loss": 0.1389, "step": 18668 }, { "epoch": 13.676923076923076, "grad_norm": 0.96875, "learning_rate": 7.176178472903904e-05, "loss": 0.1395, "step": 18669 }, { "epoch": 13.677655677655677, "grad_norm": 1.1875, "learning_rate": 7.175869049525115e-05, "loss": 0.1431, "step": 18670 }, { "epoch": 13.678388278388278, "grad_norm": 1.484375, "learning_rate": 7.175559615960591e-05, "loss": 0.1064, "step": 18671 }, { "epoch": 13.679120879120878, "grad_norm": 1.1171875, "learning_rate": 7.175250172211812e-05, "loss": 0.1683, "step": 18672 }, { "epoch": 13.67985347985348, "grad_norm": 0.734375, "learning_rate": 7.174940718280262e-05, "loss": 0.1067, "step": 18673 }, { "epoch": 13.68058608058608, "grad_norm": 0.79296875, "learning_rate": 7.174631254167424e-05, "loss": 0.1131, "step": 18674 }, { "epoch": 13.68131868131868, "grad_norm": 1.4296875, "learning_rate": 7.174321779874778e-05, "loss": 0.1656, "step": 18675 }, { "epoch": 13.682051282051281, "grad_norm": 1.2734375, "learning_rate": 7.174012295403812e-05, "loss": 0.1374, "step": 18676 }, { "epoch": 13.682783882783882, "grad_norm": 1.578125, "learning_rate": 7.173702800756005e-05, "loss": 0.1231, "step": 18677 }, { "epoch": 13.683516483516483, "grad_norm": 1.375, "learning_rate": 7.17339329593284e-05, "loss": 0.1435, "step": 18678 }, { "epoch": 13.684249084249084, "grad_norm": 0.6484375, "learning_rate": 7.173083780935802e-05, "loss": 0.139, "step": 18679 }, { "epoch": 13.684981684981684, "grad_norm": 1.03125, "learning_rate": 7.17277425576637e-05, "loss": 0.1167, "step": 18680 }, { "epoch": 13.685714285714285, "grad_norm": 1.03125, "learning_rate": 7.172464720426031e-05, "loss": 0.131, "step": 18681 }, { "epoch": 13.686446886446886, "grad_norm": 1.1875, "learning_rate": 7.172155174916266e-05, "loss": 0.1249, "step": 18682 }, { "epoch": 13.687179487179487, "grad_norm": 0.8125, "learning_rate": 7.171845619238559e-05, "loss": 0.1245, "step": 18683 }, { "epoch": 13.687912087912087, "grad_norm": 1.078125, "learning_rate": 7.171536053394394e-05, "loss": 0.1201, "step": 18684 }, { "epoch": 13.688644688644688, "grad_norm": 0.71484375, "learning_rate": 7.171226477385254e-05, "loss": 0.1196, "step": 18685 }, { "epoch": 13.689377289377289, "grad_norm": 1.2734375, "learning_rate": 7.17091689121262e-05, "loss": 0.1341, "step": 18686 }, { "epoch": 13.69010989010989, "grad_norm": 0.9453125, "learning_rate": 7.17060729487798e-05, "loss": 0.1271, "step": 18687 }, { "epoch": 13.69084249084249, "grad_norm": 0.765625, "learning_rate": 7.17029768838281e-05, "loss": 0.1209, "step": 18688 }, { "epoch": 13.691575091575091, "grad_norm": 0.92578125, "learning_rate": 7.169988071728601e-05, "loss": 0.1057, "step": 18689 }, { "epoch": 13.692307692307692, "grad_norm": 0.93359375, "learning_rate": 7.169678444916831e-05, "loss": 0.1295, "step": 18690 }, { "epoch": 13.693040293040292, "grad_norm": 1.2578125, "learning_rate": 7.169368807948988e-05, "loss": 0.1374, "step": 18691 }, { "epoch": 13.693772893772893, "grad_norm": 1.3828125, "learning_rate": 7.169059160826551e-05, "loss": 0.1119, "step": 18692 }, { "epoch": 13.694505494505494, "grad_norm": 1.265625, "learning_rate": 7.168749503551007e-05, "loss": 0.1271, "step": 18693 }, { "epoch": 13.695238095238095, "grad_norm": 1.015625, "learning_rate": 7.168439836123838e-05, "loss": 0.094, "step": 18694 }, { "epoch": 13.695970695970695, "grad_norm": 1.765625, "learning_rate": 7.16813015854653e-05, "loss": 0.1378, "step": 18695 }, { "epoch": 13.696703296703296, "grad_norm": 0.8359375, "learning_rate": 7.167820470820561e-05, "loss": 0.094, "step": 18696 }, { "epoch": 13.697435897435897, "grad_norm": 1.2109375, "learning_rate": 7.167510772947422e-05, "loss": 0.1521, "step": 18697 }, { "epoch": 13.698168498168497, "grad_norm": 1.2421875, "learning_rate": 7.167201064928595e-05, "loss": 0.1425, "step": 18698 }, { "epoch": 13.698901098901098, "grad_norm": 1.125, "learning_rate": 7.166891346765561e-05, "loss": 0.1387, "step": 18699 }, { "epoch": 13.699633699633699, "grad_norm": 0.640625, "learning_rate": 7.166581618459806e-05, "loss": 0.1273, "step": 18700 }, { "epoch": 13.7003663003663, "grad_norm": 1.0, "learning_rate": 7.166271880012812e-05, "loss": 0.1482, "step": 18701 }, { "epoch": 13.7010989010989, "grad_norm": 1.421875, "learning_rate": 7.165962131426067e-05, "loss": 0.1205, "step": 18702 }, { "epoch": 13.701831501831501, "grad_norm": 0.8125, "learning_rate": 7.165652372701051e-05, "loss": 0.1313, "step": 18703 }, { "epoch": 13.702564102564102, "grad_norm": 1.328125, "learning_rate": 7.165342603839251e-05, "loss": 0.1254, "step": 18704 }, { "epoch": 13.703296703296703, "grad_norm": 1.46875, "learning_rate": 7.16503282484215e-05, "loss": 0.1178, "step": 18705 }, { "epoch": 13.704029304029303, "grad_norm": 0.90234375, "learning_rate": 7.164723035711233e-05, "loss": 0.1341, "step": 18706 }, { "epoch": 13.704761904761904, "grad_norm": 1.2890625, "learning_rate": 7.164413236447983e-05, "loss": 0.1198, "step": 18707 }, { "epoch": 13.705494505494505, "grad_norm": 1.109375, "learning_rate": 7.164103427053885e-05, "loss": 0.1085, "step": 18708 }, { "epoch": 13.706227106227105, "grad_norm": 0.640625, "learning_rate": 7.163793607530423e-05, "loss": 0.1012, "step": 18709 }, { "epoch": 13.706959706959706, "grad_norm": 1.6484375, "learning_rate": 7.163483777879082e-05, "loss": 0.1412, "step": 18710 }, { "epoch": 13.707692307692307, "grad_norm": 1.3515625, "learning_rate": 7.163173938101348e-05, "loss": 0.1255, "step": 18711 }, { "epoch": 13.708424908424908, "grad_norm": 1.0078125, "learning_rate": 7.162864088198703e-05, "loss": 0.129, "step": 18712 }, { "epoch": 13.709157509157508, "grad_norm": 0.796875, "learning_rate": 7.162554228172634e-05, "loss": 0.1122, "step": 18713 }, { "epoch": 13.70989010989011, "grad_norm": 1.09375, "learning_rate": 7.162244358024621e-05, "loss": 0.1387, "step": 18714 }, { "epoch": 13.71062271062271, "grad_norm": 1.0234375, "learning_rate": 7.161934477756154e-05, "loss": 0.1256, "step": 18715 }, { "epoch": 13.71135531135531, "grad_norm": 1.203125, "learning_rate": 7.161624587368716e-05, "loss": 0.1338, "step": 18716 }, { "epoch": 13.712087912087911, "grad_norm": 1.109375, "learning_rate": 7.16131468686379e-05, "loss": 0.1263, "step": 18717 }, { "epoch": 13.712820512820512, "grad_norm": 0.7109375, "learning_rate": 7.161004776242863e-05, "loss": 0.1119, "step": 18718 }, { "epoch": 13.713553113553113, "grad_norm": 1.3046875, "learning_rate": 7.160694855507419e-05, "loss": 0.1271, "step": 18719 }, { "epoch": 13.714285714285714, "grad_norm": 1.109375, "learning_rate": 7.160384924658942e-05, "loss": 0.1293, "step": 18720 }, { "epoch": 13.715018315018314, "grad_norm": 1.75, "learning_rate": 7.160074983698919e-05, "loss": 0.1645, "step": 18721 }, { "epoch": 13.715750915750915, "grad_norm": 1.1171875, "learning_rate": 7.159765032628833e-05, "loss": 0.1175, "step": 18722 }, { "epoch": 13.716483516483516, "grad_norm": 1.546875, "learning_rate": 7.159455071450173e-05, "loss": 0.1849, "step": 18723 }, { "epoch": 13.717216117216116, "grad_norm": 0.66015625, "learning_rate": 7.159145100164419e-05, "loss": 0.1105, "step": 18724 }, { "epoch": 13.717948717948717, "grad_norm": 1.1328125, "learning_rate": 7.158835118773058e-05, "loss": 0.146, "step": 18725 }, { "epoch": 13.718681318681318, "grad_norm": 1.1796875, "learning_rate": 7.158525127277578e-05, "loss": 0.1208, "step": 18726 }, { "epoch": 13.719413919413919, "grad_norm": 0.90625, "learning_rate": 7.15821512567946e-05, "loss": 0.1108, "step": 18727 }, { "epoch": 13.72014652014652, "grad_norm": 0.9375, "learning_rate": 7.157905113980191e-05, "loss": 0.1634, "step": 18728 }, { "epoch": 13.72087912087912, "grad_norm": 1.125, "learning_rate": 7.157595092181258e-05, "loss": 0.1445, "step": 18729 }, { "epoch": 13.72161172161172, "grad_norm": 0.85546875, "learning_rate": 7.157285060284144e-05, "loss": 0.1186, "step": 18730 }, { "epoch": 13.722344322344322, "grad_norm": 0.91796875, "learning_rate": 7.156975018290338e-05, "loss": 0.1022, "step": 18731 }, { "epoch": 13.723076923076922, "grad_norm": 0.83203125, "learning_rate": 7.156664966201319e-05, "loss": 0.1294, "step": 18732 }, { "epoch": 13.723809523809523, "grad_norm": 1.078125, "learning_rate": 7.156354904018577e-05, "loss": 0.115, "step": 18733 }, { "epoch": 13.724542124542124, "grad_norm": 1.1875, "learning_rate": 7.1560448317436e-05, "loss": 0.1326, "step": 18734 }, { "epoch": 13.725274725274724, "grad_norm": 1.078125, "learning_rate": 7.15573474937787e-05, "loss": 0.1603, "step": 18735 }, { "epoch": 13.726007326007325, "grad_norm": 1.234375, "learning_rate": 7.155424656922874e-05, "loss": 0.1192, "step": 18736 }, { "epoch": 13.726739926739926, "grad_norm": 1.0546875, "learning_rate": 7.155114554380096e-05, "loss": 0.1259, "step": 18737 }, { "epoch": 13.727472527472527, "grad_norm": 0.76953125, "learning_rate": 7.154804441751022e-05, "loss": 0.1417, "step": 18738 }, { "epoch": 13.728205128205127, "grad_norm": 1.4765625, "learning_rate": 7.154494319037142e-05, "loss": 0.1293, "step": 18739 }, { "epoch": 13.728937728937728, "grad_norm": 0.875, "learning_rate": 7.154184186239936e-05, "loss": 0.1203, "step": 18740 }, { "epoch": 13.729670329670329, "grad_norm": 0.91015625, "learning_rate": 7.153874043360893e-05, "loss": 0.1163, "step": 18741 }, { "epoch": 13.73040293040293, "grad_norm": 0.734375, "learning_rate": 7.153563890401502e-05, "loss": 0.1251, "step": 18742 }, { "epoch": 13.73113553113553, "grad_norm": 0.765625, "learning_rate": 7.153253727363242e-05, "loss": 0.1264, "step": 18743 }, { "epoch": 13.731868131868131, "grad_norm": 1.3984375, "learning_rate": 7.152943554247607e-05, "loss": 0.1127, "step": 18744 }, { "epoch": 13.732600732600732, "grad_norm": 1.1796875, "learning_rate": 7.152633371056075e-05, "loss": 0.1263, "step": 18745 }, { "epoch": 13.733333333333333, "grad_norm": 1.390625, "learning_rate": 7.152323177790138e-05, "loss": 0.149, "step": 18746 }, { "epoch": 13.734065934065933, "grad_norm": 0.9140625, "learning_rate": 7.152012974451282e-05, "loss": 0.1449, "step": 18747 }, { "epoch": 13.734798534798536, "grad_norm": 0.90625, "learning_rate": 7.151702761040989e-05, "loss": 0.1283, "step": 18748 }, { "epoch": 13.735531135531136, "grad_norm": 0.93359375, "learning_rate": 7.15139253756075e-05, "loss": 0.1143, "step": 18749 }, { "epoch": 13.736263736263737, "grad_norm": 1.46875, "learning_rate": 7.151082304012048e-05, "loss": 0.124, "step": 18750 }, { "epoch": 13.736996336996338, "grad_norm": 1.3984375, "learning_rate": 7.150772060396373e-05, "loss": 0.1334, "step": 18751 }, { "epoch": 13.737728937728939, "grad_norm": 1.1796875, "learning_rate": 7.150461806715208e-05, "loss": 0.1177, "step": 18752 }, { "epoch": 13.73846153846154, "grad_norm": 1.3671875, "learning_rate": 7.15015154297004e-05, "loss": 0.12, "step": 18753 }, { "epoch": 13.73919413919414, "grad_norm": 1.1171875, "learning_rate": 7.149841269162359e-05, "loss": 0.1204, "step": 18754 }, { "epoch": 13.73992673992674, "grad_norm": 0.64453125, "learning_rate": 7.149530985293649e-05, "loss": 0.1275, "step": 18755 }, { "epoch": 13.740659340659342, "grad_norm": 0.84765625, "learning_rate": 7.149220691365395e-05, "loss": 0.1094, "step": 18756 }, { "epoch": 13.741391941391942, "grad_norm": 1.03125, "learning_rate": 7.148910387379087e-05, "loss": 0.1285, "step": 18757 }, { "epoch": 13.742124542124543, "grad_norm": 1.234375, "learning_rate": 7.148600073336207e-05, "loss": 0.1553, "step": 18758 }, { "epoch": 13.742857142857144, "grad_norm": 1.015625, "learning_rate": 7.148289749238249e-05, "loss": 0.1547, "step": 18759 }, { "epoch": 13.743589743589745, "grad_norm": 0.91015625, "learning_rate": 7.147979415086695e-05, "loss": 0.1272, "step": 18760 }, { "epoch": 13.744322344322345, "grad_norm": 1.0234375, "learning_rate": 7.147669070883032e-05, "loss": 0.1143, "step": 18761 }, { "epoch": 13.745054945054946, "grad_norm": 0.71484375, "learning_rate": 7.147358716628749e-05, "loss": 0.1274, "step": 18762 }, { "epoch": 13.745787545787547, "grad_norm": 0.99609375, "learning_rate": 7.147048352325332e-05, "loss": 0.1388, "step": 18763 }, { "epoch": 13.746520146520147, "grad_norm": 1.046875, "learning_rate": 7.146737977974266e-05, "loss": 0.1107, "step": 18764 }, { "epoch": 13.747252747252748, "grad_norm": 1.0625, "learning_rate": 7.146427593577041e-05, "loss": 0.154, "step": 18765 }, { "epoch": 13.747985347985349, "grad_norm": 1.28125, "learning_rate": 7.146117199135144e-05, "loss": 0.1274, "step": 18766 }, { "epoch": 13.74871794871795, "grad_norm": 0.9296875, "learning_rate": 7.145806794650062e-05, "loss": 0.1423, "step": 18767 }, { "epoch": 13.74945054945055, "grad_norm": 0.8671875, "learning_rate": 7.145496380123282e-05, "loss": 0.124, "step": 18768 }, { "epoch": 13.750183150183151, "grad_norm": 0.5390625, "learning_rate": 7.145185955556287e-05, "loss": 0.1143, "step": 18769 }, { "epoch": 13.750915750915752, "grad_norm": 0.8359375, "learning_rate": 7.144875520950571e-05, "loss": 0.1225, "step": 18770 }, { "epoch": 13.751648351648353, "grad_norm": 0.8984375, "learning_rate": 7.144565076307619e-05, "loss": 0.1203, "step": 18771 }, { "epoch": 13.752380952380953, "grad_norm": 1.1875, "learning_rate": 7.144254621628917e-05, "loss": 0.1385, "step": 18772 }, { "epoch": 13.753113553113554, "grad_norm": 0.7265625, "learning_rate": 7.143944156915954e-05, "loss": 0.1044, "step": 18773 }, { "epoch": 13.753846153846155, "grad_norm": 0.91796875, "learning_rate": 7.143633682170217e-05, "loss": 0.1403, "step": 18774 }, { "epoch": 13.754578754578755, "grad_norm": 0.81640625, "learning_rate": 7.143323197393194e-05, "loss": 0.1089, "step": 18775 }, { "epoch": 13.755311355311356, "grad_norm": 0.6796875, "learning_rate": 7.143012702586373e-05, "loss": 0.0974, "step": 18776 }, { "epoch": 13.756043956043957, "grad_norm": 1.0859375, "learning_rate": 7.14270219775124e-05, "loss": 0.1301, "step": 18777 }, { "epoch": 13.756776556776558, "grad_norm": 0.9765625, "learning_rate": 7.142391682889285e-05, "loss": 0.122, "step": 18778 }, { "epoch": 13.757509157509158, "grad_norm": 1.125, "learning_rate": 7.142081158001993e-05, "loss": 0.1627, "step": 18779 }, { "epoch": 13.758241758241759, "grad_norm": 1.7890625, "learning_rate": 7.141770623090855e-05, "loss": 0.124, "step": 18780 }, { "epoch": 13.75897435897436, "grad_norm": 1.28125, "learning_rate": 7.141460078157356e-05, "loss": 0.1314, "step": 18781 }, { "epoch": 13.75970695970696, "grad_norm": 0.91015625, "learning_rate": 7.141149523202986e-05, "loss": 0.1334, "step": 18782 }, { "epoch": 13.760439560439561, "grad_norm": 1.3828125, "learning_rate": 7.140838958229232e-05, "loss": 0.1182, "step": 18783 }, { "epoch": 13.761172161172162, "grad_norm": 1.2421875, "learning_rate": 7.140528383237581e-05, "loss": 0.1766, "step": 18784 }, { "epoch": 13.761904761904763, "grad_norm": 2.015625, "learning_rate": 7.140217798229522e-05, "loss": 0.1506, "step": 18785 }, { "epoch": 13.762637362637363, "grad_norm": 1.4296875, "learning_rate": 7.139907203206546e-05, "loss": 0.1077, "step": 18786 }, { "epoch": 13.763369963369964, "grad_norm": 1.0546875, "learning_rate": 7.139596598170137e-05, "loss": 0.111, "step": 18787 }, { "epoch": 13.764102564102565, "grad_norm": 1.3125, "learning_rate": 7.139285983121785e-05, "loss": 0.1425, "step": 18788 }, { "epoch": 13.764835164835166, "grad_norm": 2.03125, "learning_rate": 7.138975358062979e-05, "loss": 0.1237, "step": 18789 }, { "epoch": 13.765567765567766, "grad_norm": 1.1484375, "learning_rate": 7.138664722995204e-05, "loss": 0.1375, "step": 18790 }, { "epoch": 13.766300366300367, "grad_norm": 1.421875, "learning_rate": 7.138354077919951e-05, "loss": 0.1309, "step": 18791 }, { "epoch": 13.767032967032968, "grad_norm": 1.21875, "learning_rate": 7.138043422838708e-05, "loss": 0.133, "step": 18792 }, { "epoch": 13.767765567765569, "grad_norm": 0.9375, "learning_rate": 7.137732757752965e-05, "loss": 0.1151, "step": 18793 }, { "epoch": 13.76849816849817, "grad_norm": 1.0546875, "learning_rate": 7.137422082664209e-05, "loss": 0.1283, "step": 18794 }, { "epoch": 13.76923076923077, "grad_norm": 1.1640625, "learning_rate": 7.137111397573928e-05, "loss": 0.0989, "step": 18795 }, { "epoch": 13.76996336996337, "grad_norm": 0.8515625, "learning_rate": 7.136800702483611e-05, "loss": 0.1019, "step": 18796 }, { "epoch": 13.770695970695972, "grad_norm": 1.890625, "learning_rate": 7.136489997394746e-05, "loss": 0.1245, "step": 18797 }, { "epoch": 13.771428571428572, "grad_norm": 1.0234375, "learning_rate": 7.136179282308824e-05, "loss": 0.1128, "step": 18798 }, { "epoch": 13.772161172161173, "grad_norm": 1.0390625, "learning_rate": 7.135868557227331e-05, "loss": 0.1016, "step": 18799 }, { "epoch": 13.772893772893774, "grad_norm": 1.1796875, "learning_rate": 7.135557822151757e-05, "loss": 0.114, "step": 18800 }, { "epoch": 13.773626373626374, "grad_norm": 1.234375, "learning_rate": 7.13524707708359e-05, "loss": 0.1653, "step": 18801 }, { "epoch": 13.774358974358975, "grad_norm": 1.015625, "learning_rate": 7.134936322024323e-05, "loss": 0.1036, "step": 18802 }, { "epoch": 13.775091575091576, "grad_norm": 1.3203125, "learning_rate": 7.13462555697544e-05, "loss": 0.1441, "step": 18803 }, { "epoch": 13.775824175824177, "grad_norm": 1.3046875, "learning_rate": 7.134314781938431e-05, "loss": 0.153, "step": 18804 }, { "epoch": 13.776556776556777, "grad_norm": 2.5625, "learning_rate": 7.134003996914786e-05, "loss": 0.1563, "step": 18805 }, { "epoch": 13.777289377289378, "grad_norm": 1.0859375, "learning_rate": 7.133693201905993e-05, "loss": 0.1281, "step": 18806 }, { "epoch": 13.778021978021979, "grad_norm": 1.015625, "learning_rate": 7.133382396913542e-05, "loss": 0.1186, "step": 18807 }, { "epoch": 13.77875457875458, "grad_norm": 1.7734375, "learning_rate": 7.133071581938923e-05, "loss": 0.1409, "step": 18808 }, { "epoch": 13.77948717948718, "grad_norm": 0.890625, "learning_rate": 7.132760756983623e-05, "loss": 0.1188, "step": 18809 }, { "epoch": 13.780219780219781, "grad_norm": 1.046875, "learning_rate": 7.132449922049133e-05, "loss": 0.134, "step": 18810 }, { "epoch": 13.780952380952382, "grad_norm": 1.296875, "learning_rate": 7.132139077136942e-05, "loss": 0.1471, "step": 18811 }, { "epoch": 13.781684981684982, "grad_norm": 1.4296875, "learning_rate": 7.13182822224854e-05, "loss": 0.1689, "step": 18812 }, { "epoch": 13.782417582417583, "grad_norm": 2.421875, "learning_rate": 7.131517357385413e-05, "loss": 0.1446, "step": 18813 }, { "epoch": 13.783150183150184, "grad_norm": 1.3203125, "learning_rate": 7.131206482549055e-05, "loss": 0.1314, "step": 18814 }, { "epoch": 13.783882783882785, "grad_norm": 0.92578125, "learning_rate": 7.130895597740955e-05, "loss": 0.1246, "step": 18815 }, { "epoch": 13.784615384615385, "grad_norm": 1.265625, "learning_rate": 7.130584702962597e-05, "loss": 0.1487, "step": 18816 }, { "epoch": 13.785347985347986, "grad_norm": 0.8515625, "learning_rate": 7.130273798215479e-05, "loss": 0.1217, "step": 18817 }, { "epoch": 13.786080586080587, "grad_norm": 1.953125, "learning_rate": 7.129962883501083e-05, "loss": 0.1508, "step": 18818 }, { "epoch": 13.786813186813188, "grad_norm": 1.109375, "learning_rate": 7.129651958820904e-05, "loss": 0.1176, "step": 18819 }, { "epoch": 13.787545787545788, "grad_norm": 1.296875, "learning_rate": 7.129341024176429e-05, "loss": 0.1289, "step": 18820 }, { "epoch": 13.788278388278389, "grad_norm": 0.97265625, "learning_rate": 7.129030079569147e-05, "loss": 0.1151, "step": 18821 }, { "epoch": 13.78901098901099, "grad_norm": 1.1171875, "learning_rate": 7.128719125000552e-05, "loss": 0.1357, "step": 18822 }, { "epoch": 13.78974358974359, "grad_norm": 0.86328125, "learning_rate": 7.12840816047213e-05, "loss": 0.1403, "step": 18823 }, { "epoch": 13.790476190476191, "grad_norm": 0.95703125, "learning_rate": 7.128097185985372e-05, "loss": 0.1163, "step": 18824 }, { "epoch": 13.791208791208792, "grad_norm": 1.359375, "learning_rate": 7.12778620154177e-05, "loss": 0.1392, "step": 18825 }, { "epoch": 13.791941391941393, "grad_norm": 1.703125, "learning_rate": 7.12747520714281e-05, "loss": 0.1281, "step": 18826 }, { "epoch": 13.792673992673993, "grad_norm": 2.21875, "learning_rate": 7.127164202789985e-05, "loss": 0.1369, "step": 18827 }, { "epoch": 13.793406593406594, "grad_norm": 1.046875, "learning_rate": 7.126853188484786e-05, "loss": 0.1266, "step": 18828 }, { "epoch": 13.794139194139195, "grad_norm": 0.87890625, "learning_rate": 7.126542164228699e-05, "loss": 0.1384, "step": 18829 }, { "epoch": 13.794871794871796, "grad_norm": 0.94921875, "learning_rate": 7.126231130023218e-05, "loss": 0.159, "step": 18830 }, { "epoch": 13.795604395604396, "grad_norm": 0.87890625, "learning_rate": 7.125920085869833e-05, "loss": 0.1267, "step": 18831 }, { "epoch": 13.796336996336997, "grad_norm": 1.3828125, "learning_rate": 7.125609031770034e-05, "loss": 0.1287, "step": 18832 }, { "epoch": 13.797069597069598, "grad_norm": 0.6875, "learning_rate": 7.125297967725309e-05, "loss": 0.1304, "step": 18833 }, { "epoch": 13.797802197802199, "grad_norm": 1.4296875, "learning_rate": 7.124986893737149e-05, "loss": 0.1633, "step": 18834 }, { "epoch": 13.7985347985348, "grad_norm": 1.4296875, "learning_rate": 7.124675809807046e-05, "loss": 0.1399, "step": 18835 }, { "epoch": 13.7992673992674, "grad_norm": 1.265625, "learning_rate": 7.124364715936491e-05, "loss": 0.1378, "step": 18836 }, { "epoch": 13.8, "grad_norm": 0.7890625, "learning_rate": 7.124053612126974e-05, "loss": 0.1104, "step": 18837 }, { "epoch": 13.800732600732601, "grad_norm": 1.0546875, "learning_rate": 7.123742498379986e-05, "loss": 0.1155, "step": 18838 }, { "epoch": 13.801465201465202, "grad_norm": 0.83203125, "learning_rate": 7.123431374697015e-05, "loss": 0.1305, "step": 18839 }, { "epoch": 13.802197802197803, "grad_norm": 1.109375, "learning_rate": 7.123120241079555e-05, "loss": 0.1301, "step": 18840 }, { "epoch": 13.802930402930404, "grad_norm": 1.1484375, "learning_rate": 7.122809097529096e-05, "loss": 0.1077, "step": 18841 }, { "epoch": 13.803663003663004, "grad_norm": 1.015625, "learning_rate": 7.122497944047127e-05, "loss": 0.1074, "step": 18842 }, { "epoch": 13.804395604395605, "grad_norm": 1.0234375, "learning_rate": 7.122186780635141e-05, "loss": 0.1154, "step": 18843 }, { "epoch": 13.805128205128206, "grad_norm": 1.375, "learning_rate": 7.121875607294624e-05, "loss": 0.1302, "step": 18844 }, { "epoch": 13.805860805860807, "grad_norm": 1.09375, "learning_rate": 7.121564424027077e-05, "loss": 0.1369, "step": 18845 }, { "epoch": 13.806593406593407, "grad_norm": 1.6484375, "learning_rate": 7.12125323083398e-05, "loss": 0.1196, "step": 18846 }, { "epoch": 13.807326007326008, "grad_norm": 0.83984375, "learning_rate": 7.120942027716831e-05, "loss": 0.1203, "step": 18847 }, { "epoch": 13.808058608058609, "grad_norm": 0.765625, "learning_rate": 7.120630814677117e-05, "loss": 0.1202, "step": 18848 }, { "epoch": 13.80879120879121, "grad_norm": 0.9765625, "learning_rate": 7.120319591716333e-05, "loss": 0.1008, "step": 18849 }, { "epoch": 13.80952380952381, "grad_norm": 0.93359375, "learning_rate": 7.120008358835967e-05, "loss": 0.1031, "step": 18850 }, { "epoch": 13.810256410256411, "grad_norm": 0.65625, "learning_rate": 7.119697116037513e-05, "loss": 0.0969, "step": 18851 }, { "epoch": 13.810989010989012, "grad_norm": 0.63671875, "learning_rate": 7.11938586332246e-05, "loss": 0.1065, "step": 18852 }, { "epoch": 13.811721611721612, "grad_norm": 0.73828125, "learning_rate": 7.1190746006923e-05, "loss": 0.1191, "step": 18853 }, { "epoch": 13.812454212454213, "grad_norm": 1.734375, "learning_rate": 7.118763328148523e-05, "loss": 0.1382, "step": 18854 }, { "epoch": 13.813186813186814, "grad_norm": 1.0546875, "learning_rate": 7.118452045692623e-05, "loss": 0.1223, "step": 18855 }, { "epoch": 13.813919413919415, "grad_norm": 1.40625, "learning_rate": 7.11814075332609e-05, "loss": 0.1054, "step": 18856 }, { "epoch": 13.814652014652015, "grad_norm": 1.375, "learning_rate": 7.117829451050415e-05, "loss": 0.1217, "step": 18857 }, { "epoch": 13.815384615384616, "grad_norm": 0.91015625, "learning_rate": 7.117518138867092e-05, "loss": 0.1335, "step": 18858 }, { "epoch": 13.816117216117217, "grad_norm": 1.6640625, "learning_rate": 7.117206816777611e-05, "loss": 0.1253, "step": 18859 }, { "epoch": 13.816849816849818, "grad_norm": 0.77734375, "learning_rate": 7.116895484783461e-05, "loss": 0.1101, "step": 18860 }, { "epoch": 13.817582417582418, "grad_norm": 1.609375, "learning_rate": 7.116584142886137e-05, "loss": 0.1021, "step": 18861 }, { "epoch": 13.818315018315019, "grad_norm": 1.75, "learning_rate": 7.11627279108713e-05, "loss": 0.1118, "step": 18862 }, { "epoch": 13.81904761904762, "grad_norm": 1.1484375, "learning_rate": 7.115961429387931e-05, "loss": 0.1265, "step": 18863 }, { "epoch": 13.81978021978022, "grad_norm": 0.859375, "learning_rate": 7.115650057790036e-05, "loss": 0.1143, "step": 18864 }, { "epoch": 13.820512820512821, "grad_norm": 0.953125, "learning_rate": 7.115338676294932e-05, "loss": 0.1306, "step": 18865 }, { "epoch": 13.821245421245422, "grad_norm": 1.0234375, "learning_rate": 7.11502728490411e-05, "loss": 0.1002, "step": 18866 }, { "epoch": 13.821978021978023, "grad_norm": 1.53125, "learning_rate": 7.114715883619068e-05, "loss": 0.1352, "step": 18867 }, { "epoch": 13.822710622710623, "grad_norm": 1.4609375, "learning_rate": 7.114404472441292e-05, "loss": 0.1486, "step": 18868 }, { "epoch": 13.823443223443224, "grad_norm": 1.0703125, "learning_rate": 7.114093051372277e-05, "loss": 0.1371, "step": 18869 }, { "epoch": 13.824175824175825, "grad_norm": 0.8046875, "learning_rate": 7.113781620413514e-05, "loss": 0.1185, "step": 18870 }, { "epoch": 13.824908424908426, "grad_norm": 1.515625, "learning_rate": 7.113470179566497e-05, "loss": 0.1385, "step": 18871 }, { "epoch": 13.825641025641026, "grad_norm": 0.83203125, "learning_rate": 7.113158728832718e-05, "loss": 0.1094, "step": 18872 }, { "epoch": 13.826373626373627, "grad_norm": 1.046875, "learning_rate": 7.112847268213665e-05, "loss": 0.1246, "step": 18873 }, { "epoch": 13.827106227106228, "grad_norm": 1.0390625, "learning_rate": 7.112535797710835e-05, "loss": 0.1257, "step": 18874 }, { "epoch": 13.827838827838828, "grad_norm": 1.3046875, "learning_rate": 7.112224317325718e-05, "loss": 0.1704, "step": 18875 }, { "epoch": 13.82857142857143, "grad_norm": 0.8515625, "learning_rate": 7.111912827059808e-05, "loss": 0.1266, "step": 18876 }, { "epoch": 13.82930402930403, "grad_norm": 1.453125, "learning_rate": 7.111601326914598e-05, "loss": 0.1475, "step": 18877 }, { "epoch": 13.83003663003663, "grad_norm": 0.87109375, "learning_rate": 7.111289816891578e-05, "loss": 0.1374, "step": 18878 }, { "epoch": 13.830769230769231, "grad_norm": 1.2890625, "learning_rate": 7.110978296992242e-05, "loss": 0.1748, "step": 18879 }, { "epoch": 13.831501831501832, "grad_norm": 1.0703125, "learning_rate": 7.110666767218084e-05, "loss": 0.1304, "step": 18880 }, { "epoch": 13.832234432234433, "grad_norm": 2.015625, "learning_rate": 7.110355227570592e-05, "loss": 0.1215, "step": 18881 }, { "epoch": 13.832967032967034, "grad_norm": 1.125, "learning_rate": 7.110043678051263e-05, "loss": 0.0961, "step": 18882 }, { "epoch": 13.833699633699634, "grad_norm": 0.734375, "learning_rate": 7.109732118661589e-05, "loss": 0.1293, "step": 18883 }, { "epoch": 13.834432234432235, "grad_norm": 1.34375, "learning_rate": 7.109420549403063e-05, "loss": 0.1022, "step": 18884 }, { "epoch": 13.835164835164836, "grad_norm": 1.390625, "learning_rate": 7.109108970277177e-05, "loss": 0.1468, "step": 18885 }, { "epoch": 13.835897435897436, "grad_norm": 1.0546875, "learning_rate": 7.108797381285423e-05, "loss": 0.1237, "step": 18886 }, { "epoch": 13.836630036630037, "grad_norm": 1.3046875, "learning_rate": 7.108485782429294e-05, "loss": 0.1451, "step": 18887 }, { "epoch": 13.837362637362638, "grad_norm": 0.74609375, "learning_rate": 7.108174173710284e-05, "loss": 0.1322, "step": 18888 }, { "epoch": 13.838095238095239, "grad_norm": 0.765625, "learning_rate": 7.107862555129887e-05, "loss": 0.1111, "step": 18889 }, { "epoch": 13.83882783882784, "grad_norm": 0.80859375, "learning_rate": 7.107550926689595e-05, "loss": 0.0975, "step": 18890 }, { "epoch": 13.83956043956044, "grad_norm": 0.75, "learning_rate": 7.1072392883909e-05, "loss": 0.1268, "step": 18891 }, { "epoch": 13.84029304029304, "grad_norm": 0.99609375, "learning_rate": 7.106927640235298e-05, "loss": 0.0926, "step": 18892 }, { "epoch": 13.841025641025642, "grad_norm": 0.69140625, "learning_rate": 7.106615982224279e-05, "loss": 0.0867, "step": 18893 }, { "epoch": 13.841758241758242, "grad_norm": 0.890625, "learning_rate": 7.106304314359339e-05, "loss": 0.1059, "step": 18894 }, { "epoch": 13.842490842490843, "grad_norm": 0.7421875, "learning_rate": 7.105992636641969e-05, "loss": 0.1202, "step": 18895 }, { "epoch": 13.843223443223444, "grad_norm": 1.09375, "learning_rate": 7.105680949073663e-05, "loss": 0.1285, "step": 18896 }, { "epoch": 13.843956043956045, "grad_norm": 1.484375, "learning_rate": 7.105369251655915e-05, "loss": 0.1488, "step": 18897 }, { "epoch": 13.844688644688645, "grad_norm": 1.1328125, "learning_rate": 7.105057544390218e-05, "loss": 0.1697, "step": 18898 }, { "epoch": 13.845421245421246, "grad_norm": 1.2578125, "learning_rate": 7.104745827278067e-05, "loss": 0.144, "step": 18899 }, { "epoch": 13.846153846153847, "grad_norm": 0.7734375, "learning_rate": 7.104434100320952e-05, "loss": 0.1298, "step": 18900 }, { "epoch": 13.846886446886447, "grad_norm": 0.95703125, "learning_rate": 7.10412236352037e-05, "loss": 0.1188, "step": 18901 }, { "epoch": 13.847619047619048, "grad_norm": 1.5625, "learning_rate": 7.103810616877813e-05, "loss": 0.1433, "step": 18902 }, { "epoch": 13.848351648351649, "grad_norm": 1.578125, "learning_rate": 7.103498860394777e-05, "loss": 0.1046, "step": 18903 }, { "epoch": 13.84908424908425, "grad_norm": 0.78125, "learning_rate": 7.103187094072751e-05, "loss": 0.1382, "step": 18904 }, { "epoch": 13.84981684981685, "grad_norm": 0.7578125, "learning_rate": 7.102875317913233e-05, "loss": 0.1509, "step": 18905 }, { "epoch": 13.850549450549451, "grad_norm": 0.75390625, "learning_rate": 7.102563531917717e-05, "loss": 0.1323, "step": 18906 }, { "epoch": 13.851282051282052, "grad_norm": 1.1953125, "learning_rate": 7.102251736087693e-05, "loss": 0.142, "step": 18907 }, { "epoch": 13.852014652014653, "grad_norm": 1.0078125, "learning_rate": 7.101939930424659e-05, "loss": 0.1213, "step": 18908 }, { "epoch": 13.852747252747253, "grad_norm": 0.984375, "learning_rate": 7.101628114930105e-05, "loss": 0.112, "step": 18909 }, { "epoch": 13.853479853479854, "grad_norm": 1.1796875, "learning_rate": 7.10131628960553e-05, "loss": 0.1361, "step": 18910 }, { "epoch": 13.854212454212455, "grad_norm": 0.90234375, "learning_rate": 7.101004454452424e-05, "loss": 0.1337, "step": 18911 }, { "epoch": 13.854945054945055, "grad_norm": 1.2578125, "learning_rate": 7.10069260947228e-05, "loss": 0.1559, "step": 18912 }, { "epoch": 13.855677655677656, "grad_norm": 1.046875, "learning_rate": 7.100380754666597e-05, "loss": 0.1395, "step": 18913 }, { "epoch": 13.856410256410257, "grad_norm": 1.140625, "learning_rate": 7.100068890036865e-05, "loss": 0.127, "step": 18914 }, { "epoch": 13.857142857142858, "grad_norm": 1.2734375, "learning_rate": 7.099757015584582e-05, "loss": 0.1312, "step": 18915 }, { "epoch": 13.857875457875458, "grad_norm": 0.81640625, "learning_rate": 7.099445131311239e-05, "loss": 0.1236, "step": 18916 }, { "epoch": 13.858608058608059, "grad_norm": 0.91015625, "learning_rate": 7.099133237218332e-05, "loss": 0.1155, "step": 18917 }, { "epoch": 13.85934065934066, "grad_norm": 1.0078125, "learning_rate": 7.098821333307356e-05, "loss": 0.1313, "step": 18918 }, { "epoch": 13.86007326007326, "grad_norm": 0.9375, "learning_rate": 7.098509419579802e-05, "loss": 0.1539, "step": 18919 }, { "epoch": 13.860805860805861, "grad_norm": 1.2734375, "learning_rate": 7.098197496037169e-05, "loss": 0.1525, "step": 18920 }, { "epoch": 13.861538461538462, "grad_norm": 1.21875, "learning_rate": 7.097885562680949e-05, "loss": 0.1348, "step": 18921 }, { "epoch": 13.862271062271063, "grad_norm": 1.21875, "learning_rate": 7.097573619512636e-05, "loss": 0.1249, "step": 18922 }, { "epoch": 13.863003663003663, "grad_norm": 0.91796875, "learning_rate": 7.097261666533727e-05, "loss": 0.1168, "step": 18923 }, { "epoch": 13.863736263736264, "grad_norm": 1.3203125, "learning_rate": 7.096949703745715e-05, "loss": 0.142, "step": 18924 }, { "epoch": 13.864468864468865, "grad_norm": 0.9921875, "learning_rate": 7.096637731150093e-05, "loss": 0.1293, "step": 18925 }, { "epoch": 13.865201465201466, "grad_norm": 0.96484375, "learning_rate": 7.09632574874836e-05, "loss": 0.1313, "step": 18926 }, { "epoch": 13.865934065934066, "grad_norm": 1.03125, "learning_rate": 7.096013756542009e-05, "loss": 0.1384, "step": 18927 }, { "epoch": 13.866666666666667, "grad_norm": 0.7734375, "learning_rate": 7.095701754532532e-05, "loss": 0.127, "step": 18928 }, { "epoch": 13.867399267399268, "grad_norm": 1.4140625, "learning_rate": 7.095389742721428e-05, "loss": 0.1589, "step": 18929 }, { "epoch": 13.868131868131869, "grad_norm": 1.359375, "learning_rate": 7.095077721110189e-05, "loss": 0.129, "step": 18930 }, { "epoch": 13.86886446886447, "grad_norm": 1.1328125, "learning_rate": 7.094765689700313e-05, "loss": 0.1161, "step": 18931 }, { "epoch": 13.86959706959707, "grad_norm": 0.9375, "learning_rate": 7.094453648493294e-05, "loss": 0.1272, "step": 18932 }, { "epoch": 13.87032967032967, "grad_norm": 1.0, "learning_rate": 7.094141597490625e-05, "loss": 0.1389, "step": 18933 }, { "epoch": 13.871062271062272, "grad_norm": 0.92578125, "learning_rate": 7.093829536693802e-05, "loss": 0.121, "step": 18934 }, { "epoch": 13.871794871794872, "grad_norm": 1.0703125, "learning_rate": 7.093517466104321e-05, "loss": 0.108, "step": 18935 }, { "epoch": 13.872527472527473, "grad_norm": 0.86328125, "learning_rate": 7.093205385723676e-05, "loss": 0.1368, "step": 18936 }, { "epoch": 13.873260073260074, "grad_norm": 0.8515625, "learning_rate": 7.092893295553365e-05, "loss": 0.1132, "step": 18937 }, { "epoch": 13.873992673992674, "grad_norm": 0.7421875, "learning_rate": 7.092581195594882e-05, "loss": 0.1205, "step": 18938 }, { "epoch": 13.874725274725275, "grad_norm": 0.63671875, "learning_rate": 7.092269085849721e-05, "loss": 0.1239, "step": 18939 }, { "epoch": 13.875457875457876, "grad_norm": 1.09375, "learning_rate": 7.091956966319379e-05, "loss": 0.1064, "step": 18940 }, { "epoch": 13.876190476190477, "grad_norm": 1.15625, "learning_rate": 7.091644837005348e-05, "loss": 0.1165, "step": 18941 }, { "epoch": 13.876923076923077, "grad_norm": 0.9375, "learning_rate": 7.09133269790913e-05, "loss": 0.1305, "step": 18942 }, { "epoch": 13.877655677655678, "grad_norm": 0.796875, "learning_rate": 7.091020549032214e-05, "loss": 0.1122, "step": 18943 }, { "epoch": 13.878388278388279, "grad_norm": 1.1640625, "learning_rate": 7.090708390376102e-05, "loss": 0.1125, "step": 18944 }, { "epoch": 13.87912087912088, "grad_norm": 1.0625, "learning_rate": 7.090396221942284e-05, "loss": 0.1609, "step": 18945 }, { "epoch": 13.87985347985348, "grad_norm": 1.0703125, "learning_rate": 7.090084043732258e-05, "loss": 0.1514, "step": 18946 }, { "epoch": 13.880586080586081, "grad_norm": 1.84375, "learning_rate": 7.08977185574752e-05, "loss": 0.1445, "step": 18947 }, { "epoch": 13.881318681318682, "grad_norm": 0.75390625, "learning_rate": 7.089459657989565e-05, "loss": 0.1136, "step": 18948 }, { "epoch": 13.882051282051282, "grad_norm": 0.99609375, "learning_rate": 7.089147450459888e-05, "loss": 0.1186, "step": 18949 }, { "epoch": 13.882783882783883, "grad_norm": 0.66796875, "learning_rate": 7.088835233159988e-05, "loss": 0.1121, "step": 18950 }, { "epoch": 13.883516483516484, "grad_norm": 0.90234375, "learning_rate": 7.08852300609136e-05, "loss": 0.1307, "step": 18951 }, { "epoch": 13.884249084249085, "grad_norm": 0.90234375, "learning_rate": 7.088210769255498e-05, "loss": 0.1412, "step": 18952 }, { "epoch": 13.884981684981685, "grad_norm": 0.8515625, "learning_rate": 7.087898522653899e-05, "loss": 0.1074, "step": 18953 }, { "epoch": 13.885714285714286, "grad_norm": 1.5, "learning_rate": 7.087586266288057e-05, "loss": 0.1574, "step": 18954 }, { "epoch": 13.886446886446887, "grad_norm": 0.96875, "learning_rate": 7.087274000159475e-05, "loss": 0.1228, "step": 18955 }, { "epoch": 13.887179487179488, "grad_norm": 0.97265625, "learning_rate": 7.08696172426964e-05, "loss": 0.1412, "step": 18956 }, { "epoch": 13.887912087912088, "grad_norm": 0.78125, "learning_rate": 7.086649438620055e-05, "loss": 0.1094, "step": 18957 }, { "epoch": 13.888644688644689, "grad_norm": 1.484375, "learning_rate": 7.086337143212214e-05, "loss": 0.1208, "step": 18958 }, { "epoch": 13.88937728937729, "grad_norm": 1.0, "learning_rate": 7.086024838047611e-05, "loss": 0.1202, "step": 18959 }, { "epoch": 13.89010989010989, "grad_norm": 0.8828125, "learning_rate": 7.085712523127749e-05, "loss": 0.1259, "step": 18960 }, { "epoch": 13.890842490842491, "grad_norm": 1.03125, "learning_rate": 7.085400198454116e-05, "loss": 0.1346, "step": 18961 }, { "epoch": 13.891575091575092, "grad_norm": 0.79296875, "learning_rate": 7.085087864028214e-05, "loss": 0.1239, "step": 18962 }, { "epoch": 13.892307692307693, "grad_norm": 1.234375, "learning_rate": 7.084775519851537e-05, "loss": 0.1317, "step": 18963 }, { "epoch": 13.893040293040293, "grad_norm": 1.5625, "learning_rate": 7.084463165925583e-05, "loss": 0.1419, "step": 18964 }, { "epoch": 13.893772893772894, "grad_norm": 1.9921875, "learning_rate": 7.084150802251848e-05, "loss": 0.1522, "step": 18965 }, { "epoch": 13.894505494505495, "grad_norm": 0.74609375, "learning_rate": 7.083838428831832e-05, "loss": 0.1278, "step": 18966 }, { "epoch": 13.895238095238096, "grad_norm": 0.87109375, "learning_rate": 7.083526045667024e-05, "loss": 0.1356, "step": 18967 }, { "epoch": 13.895970695970696, "grad_norm": 1.2734375, "learning_rate": 7.083213652758927e-05, "loss": 0.1238, "step": 18968 }, { "epoch": 13.896703296703297, "grad_norm": 0.66796875, "learning_rate": 7.082901250109034e-05, "loss": 0.1322, "step": 18969 }, { "epoch": 13.897435897435898, "grad_norm": 0.921875, "learning_rate": 7.082588837718846e-05, "loss": 0.1296, "step": 18970 }, { "epoch": 13.898168498168499, "grad_norm": 1.109375, "learning_rate": 7.082276415589859e-05, "loss": 0.1241, "step": 18971 }, { "epoch": 13.8989010989011, "grad_norm": 1.2734375, "learning_rate": 7.081963983723566e-05, "loss": 0.1177, "step": 18972 }, { "epoch": 13.8996336996337, "grad_norm": 0.97265625, "learning_rate": 7.081651542121467e-05, "loss": 0.1137, "step": 18973 }, { "epoch": 13.9003663003663, "grad_norm": 0.76953125, "learning_rate": 7.081339090785059e-05, "loss": 0.1228, "step": 18974 }, { "epoch": 13.901098901098901, "grad_norm": 1.203125, "learning_rate": 7.081026629715838e-05, "loss": 0.1537, "step": 18975 }, { "epoch": 13.901831501831502, "grad_norm": 1.265625, "learning_rate": 7.0807141589153e-05, "loss": 0.1323, "step": 18976 }, { "epoch": 13.902564102564103, "grad_norm": 1.5703125, "learning_rate": 7.080401678384947e-05, "loss": 0.11, "step": 18977 }, { "epoch": 13.903296703296704, "grad_norm": 1.3359375, "learning_rate": 7.080089188126272e-05, "loss": 0.1507, "step": 18978 }, { "epoch": 13.904029304029304, "grad_norm": 0.8828125, "learning_rate": 7.079776688140774e-05, "loss": 0.1374, "step": 18979 }, { "epoch": 13.904761904761905, "grad_norm": 0.875, "learning_rate": 7.079464178429946e-05, "loss": 0.1136, "step": 18980 }, { "epoch": 13.905494505494506, "grad_norm": 0.83203125, "learning_rate": 7.079151658995292e-05, "loss": 0.1083, "step": 18981 }, { "epoch": 13.906227106227107, "grad_norm": 1.09375, "learning_rate": 7.078839129838306e-05, "loss": 0.1283, "step": 18982 }, { "epoch": 13.906959706959707, "grad_norm": 1.125, "learning_rate": 7.078526590960486e-05, "loss": 0.1213, "step": 18983 }, { "epoch": 13.907692307692308, "grad_norm": 1.34375, "learning_rate": 7.07821404236333e-05, "loss": 0.1267, "step": 18984 }, { "epoch": 13.908424908424909, "grad_norm": 1.0546875, "learning_rate": 7.077901484048332e-05, "loss": 0.1113, "step": 18985 }, { "epoch": 13.90915750915751, "grad_norm": 1.203125, "learning_rate": 7.077588916016994e-05, "loss": 0.1188, "step": 18986 }, { "epoch": 13.90989010989011, "grad_norm": 1.328125, "learning_rate": 7.077276338270811e-05, "loss": 0.1144, "step": 18987 }, { "epoch": 13.910622710622711, "grad_norm": 2.0, "learning_rate": 7.076963750811281e-05, "loss": 0.1162, "step": 18988 }, { "epoch": 13.911355311355312, "grad_norm": 0.80078125, "learning_rate": 7.076651153639903e-05, "loss": 0.1263, "step": 18989 }, { "epoch": 13.912087912087912, "grad_norm": 1.2265625, "learning_rate": 7.076338546758175e-05, "loss": 0.1343, "step": 18990 }, { "epoch": 13.912820512820513, "grad_norm": 0.5703125, "learning_rate": 7.076025930167592e-05, "loss": 0.1031, "step": 18991 }, { "epoch": 13.913553113553114, "grad_norm": 1.109375, "learning_rate": 7.075713303869656e-05, "loss": 0.1241, "step": 18992 }, { "epoch": 13.914285714285715, "grad_norm": 1.140625, "learning_rate": 7.075400667865861e-05, "loss": 0.1337, "step": 18993 }, { "epoch": 13.915018315018315, "grad_norm": 1.4765625, "learning_rate": 7.075088022157708e-05, "loss": 0.1391, "step": 18994 }, { "epoch": 13.915750915750916, "grad_norm": 1.046875, "learning_rate": 7.074775366746691e-05, "loss": 0.1265, "step": 18995 }, { "epoch": 13.916483516483517, "grad_norm": 1.515625, "learning_rate": 7.074462701634312e-05, "loss": 0.1035, "step": 18996 }, { "epoch": 13.917216117216118, "grad_norm": 1.4609375, "learning_rate": 7.074150026822068e-05, "loss": 0.1197, "step": 18997 }, { "epoch": 13.917948717948718, "grad_norm": 1.5390625, "learning_rate": 7.073837342311456e-05, "loss": 0.1158, "step": 18998 }, { "epoch": 13.918681318681319, "grad_norm": 1.0078125, "learning_rate": 7.073524648103975e-05, "loss": 0.1089, "step": 18999 }, { "epoch": 13.91941391941392, "grad_norm": 1.109375, "learning_rate": 7.073211944201123e-05, "loss": 0.1075, "step": 19000 }, { "epoch": 13.92014652014652, "grad_norm": 0.73828125, "learning_rate": 7.072899230604399e-05, "loss": 0.1135, "step": 19001 }, { "epoch": 13.920879120879121, "grad_norm": 1.125, "learning_rate": 7.0725865073153e-05, "loss": 0.125, "step": 19002 }, { "epoch": 13.921611721611722, "grad_norm": 1.1328125, "learning_rate": 7.072273774335326e-05, "loss": 0.1622, "step": 19003 }, { "epoch": 13.922344322344323, "grad_norm": 0.9375, "learning_rate": 7.071961031665973e-05, "loss": 0.1137, "step": 19004 }, { "epoch": 13.923076923076923, "grad_norm": 1.1015625, "learning_rate": 7.071648279308743e-05, "loss": 0.1222, "step": 19005 }, { "epoch": 13.923809523809524, "grad_norm": 0.78515625, "learning_rate": 7.07133551726513e-05, "loss": 0.0941, "step": 19006 }, { "epoch": 13.924542124542125, "grad_norm": 0.6875, "learning_rate": 7.071022745536636e-05, "loss": 0.131, "step": 19007 }, { "epoch": 13.925274725274726, "grad_norm": 0.734375, "learning_rate": 7.070709964124759e-05, "loss": 0.1367, "step": 19008 }, { "epoch": 13.926007326007326, "grad_norm": 1.3046875, "learning_rate": 7.070397173030996e-05, "loss": 0.1081, "step": 19009 }, { "epoch": 13.926739926739927, "grad_norm": 0.9453125, "learning_rate": 7.070084372256849e-05, "loss": 0.1394, "step": 19010 }, { "epoch": 13.927472527472528, "grad_norm": 1.2265625, "learning_rate": 7.069771561803813e-05, "loss": 0.1362, "step": 19011 }, { "epoch": 13.928205128205128, "grad_norm": 0.93359375, "learning_rate": 7.06945874167339e-05, "loss": 0.1142, "step": 19012 }, { "epoch": 13.92893772893773, "grad_norm": 1.125, "learning_rate": 7.069145911867075e-05, "loss": 0.1134, "step": 19013 }, { "epoch": 13.92967032967033, "grad_norm": 1.4140625, "learning_rate": 7.068833072386371e-05, "loss": 0.1649, "step": 19014 }, { "epoch": 13.93040293040293, "grad_norm": 0.80859375, "learning_rate": 7.068520223232775e-05, "loss": 0.1238, "step": 19015 }, { "epoch": 13.931135531135531, "grad_norm": 0.96484375, "learning_rate": 7.068207364407783e-05, "loss": 0.1205, "step": 19016 }, { "epoch": 13.931868131868132, "grad_norm": 0.62890625, "learning_rate": 7.067894495912902e-05, "loss": 0.113, "step": 19017 }, { "epoch": 13.932600732600733, "grad_norm": 1.0234375, "learning_rate": 7.067581617749622e-05, "loss": 0.1103, "step": 19018 }, { "epoch": 13.933333333333334, "grad_norm": 1.4921875, "learning_rate": 7.067268729919448e-05, "loss": 0.1686, "step": 19019 }, { "epoch": 13.934065934065934, "grad_norm": 1.15625, "learning_rate": 7.066955832423875e-05, "loss": 0.1229, "step": 19020 }, { "epoch": 13.934798534798535, "grad_norm": 1.1875, "learning_rate": 7.066642925264407e-05, "loss": 0.1154, "step": 19021 }, { "epoch": 13.935531135531136, "grad_norm": 1.4765625, "learning_rate": 7.06633000844254e-05, "loss": 0.1648, "step": 19022 }, { "epoch": 13.936263736263736, "grad_norm": 1.4921875, "learning_rate": 7.066017081959774e-05, "loss": 0.1581, "step": 19023 }, { "epoch": 13.936996336996337, "grad_norm": 1.2734375, "learning_rate": 7.065704145817608e-05, "loss": 0.119, "step": 19024 }, { "epoch": 13.937728937728938, "grad_norm": 0.65234375, "learning_rate": 7.065391200017543e-05, "loss": 0.12, "step": 19025 }, { "epoch": 13.938461538461539, "grad_norm": 1.390625, "learning_rate": 7.065078244561075e-05, "loss": 0.126, "step": 19026 }, { "epoch": 13.93919413919414, "grad_norm": 0.8125, "learning_rate": 7.064765279449708e-05, "loss": 0.1274, "step": 19027 }, { "epoch": 13.93992673992674, "grad_norm": 1.2265625, "learning_rate": 7.064452304684936e-05, "loss": 0.1483, "step": 19028 }, { "epoch": 13.94065934065934, "grad_norm": 0.80078125, "learning_rate": 7.064139320268264e-05, "loss": 0.1388, "step": 19029 }, { "epoch": 13.941391941391942, "grad_norm": 1.171875, "learning_rate": 7.063826326201188e-05, "loss": 0.1509, "step": 19030 }, { "epoch": 13.942124542124542, "grad_norm": 1.1171875, "learning_rate": 7.063513322485212e-05, "loss": 0.1368, "step": 19031 }, { "epoch": 13.942857142857143, "grad_norm": 1.65625, "learning_rate": 7.06320030912183e-05, "loss": 0.1388, "step": 19032 }, { "epoch": 13.943589743589744, "grad_norm": 1.1953125, "learning_rate": 7.062887286112545e-05, "loss": 0.0978, "step": 19033 }, { "epoch": 13.944322344322345, "grad_norm": 1.03125, "learning_rate": 7.062574253458855e-05, "loss": 0.1015, "step": 19034 }, { "epoch": 13.945054945054945, "grad_norm": 0.6640625, "learning_rate": 7.062261211162265e-05, "loss": 0.126, "step": 19035 }, { "epoch": 13.945787545787546, "grad_norm": 0.76953125, "learning_rate": 7.061948159224267e-05, "loss": 0.1049, "step": 19036 }, { "epoch": 13.946520146520147, "grad_norm": 0.95703125, "learning_rate": 7.061635097646365e-05, "loss": 0.126, "step": 19037 }, { "epoch": 13.947252747252747, "grad_norm": 0.84765625, "learning_rate": 7.061322026430061e-05, "loss": 0.1304, "step": 19038 }, { "epoch": 13.947985347985348, "grad_norm": 1.109375, "learning_rate": 7.06100894557685e-05, "loss": 0.1312, "step": 19039 }, { "epoch": 13.948717948717949, "grad_norm": 0.9296875, "learning_rate": 7.060695855088238e-05, "loss": 0.1223, "step": 19040 }, { "epoch": 13.94945054945055, "grad_norm": 1.4765625, "learning_rate": 7.06038275496572e-05, "loss": 0.1118, "step": 19041 }, { "epoch": 13.95018315018315, "grad_norm": 1.3671875, "learning_rate": 7.060069645210799e-05, "loss": 0.1246, "step": 19042 }, { "epoch": 13.950915750915751, "grad_norm": 1.0390625, "learning_rate": 7.059756525824976e-05, "loss": 0.1571, "step": 19043 }, { "epoch": 13.951648351648352, "grad_norm": 0.7421875, "learning_rate": 7.05944339680975e-05, "loss": 0.1106, "step": 19044 }, { "epoch": 13.952380952380953, "grad_norm": 0.80078125, "learning_rate": 7.059130258166617e-05, "loss": 0.1319, "step": 19045 }, { "epoch": 13.953113553113553, "grad_norm": 0.703125, "learning_rate": 7.058817109897084e-05, "loss": 0.1053, "step": 19046 }, { "epoch": 13.953846153846154, "grad_norm": 1.0078125, "learning_rate": 7.05850395200265e-05, "loss": 0.1295, "step": 19047 }, { "epoch": 13.954578754578755, "grad_norm": 0.58203125, "learning_rate": 7.058190784484813e-05, "loss": 0.1057, "step": 19048 }, { "epoch": 13.955311355311355, "grad_norm": 1.328125, "learning_rate": 7.057877607345074e-05, "loss": 0.1449, "step": 19049 }, { "epoch": 13.956043956043956, "grad_norm": 0.89453125, "learning_rate": 7.057564420584936e-05, "loss": 0.0871, "step": 19050 }, { "epoch": 13.956776556776557, "grad_norm": 1.6015625, "learning_rate": 7.057251224205895e-05, "loss": 0.1385, "step": 19051 }, { "epoch": 13.957509157509158, "grad_norm": 1.265625, "learning_rate": 7.056938018209455e-05, "loss": 0.1299, "step": 19052 }, { "epoch": 13.958241758241758, "grad_norm": 1.25, "learning_rate": 7.056624802597118e-05, "loss": 0.131, "step": 19053 }, { "epoch": 13.95897435897436, "grad_norm": 0.921875, "learning_rate": 7.056311577370381e-05, "loss": 0.1228, "step": 19054 }, { "epoch": 13.95970695970696, "grad_norm": 0.875, "learning_rate": 7.055998342530747e-05, "loss": 0.1314, "step": 19055 }, { "epoch": 13.96043956043956, "grad_norm": 1.5078125, "learning_rate": 7.055685098079719e-05, "loss": 0.1353, "step": 19056 }, { "epoch": 13.961172161172161, "grad_norm": 1.4609375, "learning_rate": 7.055371844018793e-05, "loss": 0.124, "step": 19057 }, { "epoch": 13.961904761904762, "grad_norm": 1.6640625, "learning_rate": 7.055058580349472e-05, "loss": 0.1299, "step": 19058 }, { "epoch": 13.962637362637363, "grad_norm": 1.25, "learning_rate": 7.054745307073258e-05, "loss": 0.1082, "step": 19059 }, { "epoch": 13.963369963369964, "grad_norm": 1.6328125, "learning_rate": 7.054432024191649e-05, "loss": 0.1462, "step": 19060 }, { "epoch": 13.964102564102564, "grad_norm": 0.67578125, "learning_rate": 7.054118731706152e-05, "loss": 0.1182, "step": 19061 }, { "epoch": 13.964835164835165, "grad_norm": 0.828125, "learning_rate": 7.053805429618262e-05, "loss": 0.1165, "step": 19062 }, { "epoch": 13.965567765567766, "grad_norm": 1.203125, "learning_rate": 7.053492117929482e-05, "loss": 0.1395, "step": 19063 }, { "epoch": 13.966300366300366, "grad_norm": 1.1640625, "learning_rate": 7.053178796641314e-05, "loss": 0.1468, "step": 19064 }, { "epoch": 13.967032967032967, "grad_norm": 0.96875, "learning_rate": 7.052865465755259e-05, "loss": 0.1195, "step": 19065 }, { "epoch": 13.967765567765568, "grad_norm": 1.046875, "learning_rate": 7.052552125272818e-05, "loss": 0.1343, "step": 19066 }, { "epoch": 13.968498168498169, "grad_norm": 0.8984375, "learning_rate": 7.052238775195492e-05, "loss": 0.1163, "step": 19067 }, { "epoch": 13.96923076923077, "grad_norm": 1.421875, "learning_rate": 7.051925415524783e-05, "loss": 0.1359, "step": 19068 }, { "epoch": 13.96996336996337, "grad_norm": 1.3671875, "learning_rate": 7.051612046262192e-05, "loss": 0.1208, "step": 19069 }, { "epoch": 13.97069597069597, "grad_norm": 0.85546875, "learning_rate": 7.05129866740922e-05, "loss": 0.1017, "step": 19070 }, { "epoch": 13.971428571428572, "grad_norm": 1.046875, "learning_rate": 7.05098527896737e-05, "loss": 0.1549, "step": 19071 }, { "epoch": 13.972161172161172, "grad_norm": 1.2109375, "learning_rate": 7.050671880938143e-05, "loss": 0.1203, "step": 19072 }, { "epoch": 13.972893772893773, "grad_norm": 0.75, "learning_rate": 7.05035847332304e-05, "loss": 0.1142, "step": 19073 }, { "epoch": 13.973626373626374, "grad_norm": 1.0546875, "learning_rate": 7.050045056123563e-05, "loss": 0.115, "step": 19074 }, { "epoch": 13.974358974358974, "grad_norm": 2.25, "learning_rate": 7.049731629341213e-05, "loss": 0.1212, "step": 19075 }, { "epoch": 13.975091575091575, "grad_norm": 1.1796875, "learning_rate": 7.049418192977492e-05, "loss": 0.1195, "step": 19076 }, { "epoch": 13.975824175824176, "grad_norm": 0.96484375, "learning_rate": 7.049104747033903e-05, "loss": 0.1447, "step": 19077 }, { "epoch": 13.976556776556777, "grad_norm": 0.98828125, "learning_rate": 7.048791291511945e-05, "loss": 0.1742, "step": 19078 }, { "epoch": 13.977289377289377, "grad_norm": 0.80078125, "learning_rate": 7.048477826413124e-05, "loss": 0.1357, "step": 19079 }, { "epoch": 13.978021978021978, "grad_norm": 1.1328125, "learning_rate": 7.048164351738939e-05, "loss": 0.1479, "step": 19080 }, { "epoch": 13.978754578754579, "grad_norm": 0.90625, "learning_rate": 7.047850867490891e-05, "loss": 0.1318, "step": 19081 }, { "epoch": 13.97948717948718, "grad_norm": 1.28125, "learning_rate": 7.047537373670487e-05, "loss": 0.1255, "step": 19082 }, { "epoch": 13.98021978021978, "grad_norm": 1.0625, "learning_rate": 7.047223870279223e-05, "loss": 0.1126, "step": 19083 }, { "epoch": 13.980952380952381, "grad_norm": 1.1875, "learning_rate": 7.046910357318604e-05, "loss": 0.1332, "step": 19084 }, { "epoch": 13.981684981684982, "grad_norm": 0.97265625, "learning_rate": 7.046596834790133e-05, "loss": 0.126, "step": 19085 }, { "epoch": 13.982417582417582, "grad_norm": 1.2265625, "learning_rate": 7.04628330269531e-05, "loss": 0.1553, "step": 19086 }, { "epoch": 13.983150183150183, "grad_norm": 0.70703125, "learning_rate": 7.04596976103564e-05, "loss": 0.1274, "step": 19087 }, { "epoch": 13.983882783882784, "grad_norm": 1.265625, "learning_rate": 7.045656209812623e-05, "loss": 0.1289, "step": 19088 }, { "epoch": 13.984615384615385, "grad_norm": 1.03125, "learning_rate": 7.045342649027762e-05, "loss": 0.1295, "step": 19089 }, { "epoch": 13.985347985347985, "grad_norm": 0.984375, "learning_rate": 7.045029078682559e-05, "loss": 0.1249, "step": 19090 }, { "epoch": 13.986080586080586, "grad_norm": 0.8046875, "learning_rate": 7.044715498778516e-05, "loss": 0.125, "step": 19091 }, { "epoch": 13.986813186813187, "grad_norm": 0.69921875, "learning_rate": 7.044401909317137e-05, "loss": 0.1296, "step": 19092 }, { "epoch": 13.987545787545788, "grad_norm": 1.3828125, "learning_rate": 7.044088310299924e-05, "loss": 0.1758, "step": 19093 }, { "epoch": 13.988278388278388, "grad_norm": 1.2890625, "learning_rate": 7.043774701728378e-05, "loss": 0.1298, "step": 19094 }, { "epoch": 13.989010989010989, "grad_norm": 0.83203125, "learning_rate": 7.043461083604004e-05, "loss": 0.1311, "step": 19095 }, { "epoch": 13.98974358974359, "grad_norm": 0.9296875, "learning_rate": 7.043147455928306e-05, "loss": 0.1428, "step": 19096 }, { "epoch": 13.99047619047619, "grad_norm": 1.0234375, "learning_rate": 7.04283381870278e-05, "loss": 0.0982, "step": 19097 }, { "epoch": 13.991208791208791, "grad_norm": 1.09375, "learning_rate": 7.042520171928935e-05, "loss": 0.1345, "step": 19098 }, { "epoch": 13.991941391941392, "grad_norm": 1.0390625, "learning_rate": 7.04220651560827e-05, "loss": 0.1621, "step": 19099 }, { "epoch": 13.992673992673993, "grad_norm": 0.86328125, "learning_rate": 7.041892849742292e-05, "loss": 0.1244, "step": 19100 }, { "epoch": 13.993406593406593, "grad_norm": 0.94921875, "learning_rate": 7.0415791743325e-05, "loss": 0.1625, "step": 19101 }, { "epoch": 13.994139194139194, "grad_norm": 1.625, "learning_rate": 7.0412654893804e-05, "loss": 0.1488, "step": 19102 }, { "epoch": 13.994871794871795, "grad_norm": 1.0625, "learning_rate": 7.040951794887492e-05, "loss": 0.1233, "step": 19103 }, { "epoch": 13.995604395604396, "grad_norm": 0.79296875, "learning_rate": 7.04063809085528e-05, "loss": 0.1073, "step": 19104 }, { "epoch": 13.996336996336996, "grad_norm": 0.77734375, "learning_rate": 7.040324377285268e-05, "loss": 0.108, "step": 19105 }, { "epoch": 13.997069597069597, "grad_norm": 1.109375, "learning_rate": 7.040010654178958e-05, "loss": 0.1422, "step": 19106 }, { "epoch": 13.997802197802198, "grad_norm": 0.85546875, "learning_rate": 7.039696921537854e-05, "loss": 0.1052, "step": 19107 }, { "epoch": 13.998534798534799, "grad_norm": 1.53125, "learning_rate": 7.03938317936346e-05, "loss": 0.1251, "step": 19108 }, { "epoch": 13.9992673992674, "grad_norm": 1.4921875, "learning_rate": 7.039069427657276e-05, "loss": 0.1186, "step": 19109 }, { "epoch": 14.0, "grad_norm": 1.0, "learning_rate": 7.038755666420809e-05, "loss": 0.1233, "step": 19110 }, { "epoch": 14.0007326007326, "grad_norm": 1.09375, "learning_rate": 7.038441895655561e-05, "loss": 0.1311, "step": 19111 }, { "epoch": 14.001465201465201, "grad_norm": 0.84765625, "learning_rate": 7.038128115363034e-05, "loss": 0.089, "step": 19112 }, { "epoch": 14.002197802197802, "grad_norm": 0.796875, "learning_rate": 7.037814325544732e-05, "loss": 0.1244, "step": 19113 }, { "epoch": 14.002930402930403, "grad_norm": 1.0546875, "learning_rate": 7.037500526202162e-05, "loss": 0.122, "step": 19114 }, { "epoch": 14.003663003663004, "grad_norm": 1.28125, "learning_rate": 7.037186717336821e-05, "loss": 0.1429, "step": 19115 }, { "epoch": 14.004395604395604, "grad_norm": 1.21875, "learning_rate": 7.036872898950219e-05, "loss": 0.1165, "step": 19116 }, { "epoch": 14.005128205128205, "grad_norm": 1.03125, "learning_rate": 7.036559071043854e-05, "loss": 0.144, "step": 19117 }, { "epoch": 14.005860805860806, "grad_norm": 0.5859375, "learning_rate": 7.036245233619233e-05, "loss": 0.1112, "step": 19118 }, { "epoch": 14.006593406593407, "grad_norm": 0.8359375, "learning_rate": 7.03593138667786e-05, "loss": 0.128, "step": 19119 }, { "epoch": 14.007326007326007, "grad_norm": 0.75, "learning_rate": 7.035617530221236e-05, "loss": 0.108, "step": 19120 }, { "epoch": 14.008058608058608, "grad_norm": 0.8203125, "learning_rate": 7.03530366425087e-05, "loss": 0.1184, "step": 19121 }, { "epoch": 14.008791208791209, "grad_norm": 0.9453125, "learning_rate": 7.03498978876826e-05, "loss": 0.1291, "step": 19122 }, { "epoch": 14.00952380952381, "grad_norm": 0.75390625, "learning_rate": 7.034675903774912e-05, "loss": 0.1285, "step": 19123 }, { "epoch": 14.01025641025641, "grad_norm": 1.03125, "learning_rate": 7.03436200927233e-05, "loss": 0.1154, "step": 19124 }, { "epoch": 14.010989010989011, "grad_norm": 1.1875, "learning_rate": 7.03404810526202e-05, "loss": 0.1363, "step": 19125 }, { "epoch": 14.011721611721612, "grad_norm": 0.84375, "learning_rate": 7.033734191745484e-05, "loss": 0.1451, "step": 19126 }, { "epoch": 14.012454212454212, "grad_norm": 0.77734375, "learning_rate": 7.033420268724227e-05, "loss": 0.1124, "step": 19127 }, { "epoch": 14.013186813186813, "grad_norm": 1.0390625, "learning_rate": 7.03310633619975e-05, "loss": 0.13, "step": 19128 }, { "epoch": 14.013919413919414, "grad_norm": 1.015625, "learning_rate": 7.032792394173561e-05, "loss": 0.1344, "step": 19129 }, { "epoch": 14.014652014652015, "grad_norm": 1.0703125, "learning_rate": 7.032478442647161e-05, "loss": 0.1215, "step": 19130 }, { "epoch": 14.015384615384615, "grad_norm": 0.83203125, "learning_rate": 7.032164481622057e-05, "loss": 0.1126, "step": 19131 }, { "epoch": 14.016117216117216, "grad_norm": 1.1328125, "learning_rate": 7.031850511099753e-05, "loss": 0.147, "step": 19132 }, { "epoch": 14.016849816849817, "grad_norm": 0.9921875, "learning_rate": 7.031536531081752e-05, "loss": 0.1213, "step": 19133 }, { "epoch": 14.017582417582418, "grad_norm": 0.85546875, "learning_rate": 7.031222541569559e-05, "loss": 0.1335, "step": 19134 }, { "epoch": 14.018315018315018, "grad_norm": 1.6953125, "learning_rate": 7.030908542564679e-05, "loss": 0.159, "step": 19135 }, { "epoch": 14.019047619047619, "grad_norm": 1.625, "learning_rate": 7.030594534068614e-05, "loss": 0.1425, "step": 19136 }, { "epoch": 14.01978021978022, "grad_norm": 1.265625, "learning_rate": 7.030280516082872e-05, "loss": 0.1574, "step": 19137 }, { "epoch": 14.02051282051282, "grad_norm": 0.87109375, "learning_rate": 7.029966488608956e-05, "loss": 0.1042, "step": 19138 }, { "epoch": 14.021245421245421, "grad_norm": 1.2265625, "learning_rate": 7.02965245164837e-05, "loss": 0.1142, "step": 19139 }, { "epoch": 14.021978021978022, "grad_norm": 0.96484375, "learning_rate": 7.02933840520262e-05, "loss": 0.1367, "step": 19140 }, { "epoch": 14.022710622710623, "grad_norm": 1.578125, "learning_rate": 7.02902434927321e-05, "loss": 0.139, "step": 19141 }, { "epoch": 14.023443223443223, "grad_norm": 1.1171875, "learning_rate": 7.028710283861645e-05, "loss": 0.1334, "step": 19142 }, { "epoch": 14.024175824175824, "grad_norm": 0.69140625, "learning_rate": 7.028396208969427e-05, "loss": 0.0953, "step": 19143 }, { "epoch": 14.024908424908425, "grad_norm": 2.0, "learning_rate": 7.028082124598064e-05, "loss": 0.1951, "step": 19144 }, { "epoch": 14.025641025641026, "grad_norm": 1.125, "learning_rate": 7.027768030749062e-05, "loss": 0.152, "step": 19145 }, { "epoch": 14.026373626373626, "grad_norm": 0.96875, "learning_rate": 7.027453927423922e-05, "loss": 0.1414, "step": 19146 }, { "epoch": 14.027106227106227, "grad_norm": 1.2890625, "learning_rate": 7.027139814624154e-05, "loss": 0.1198, "step": 19147 }, { "epoch": 14.027838827838828, "grad_norm": 1.78125, "learning_rate": 7.026825692351259e-05, "loss": 0.138, "step": 19148 }, { "epoch": 14.028571428571428, "grad_norm": 1.7109375, "learning_rate": 7.026511560606742e-05, "loss": 0.1165, "step": 19149 }, { "epoch": 14.02930402930403, "grad_norm": 1.5234375, "learning_rate": 7.026197419392109e-05, "loss": 0.1207, "step": 19150 }, { "epoch": 14.03003663003663, "grad_norm": 1.5234375, "learning_rate": 7.025883268708866e-05, "loss": 0.1473, "step": 19151 }, { "epoch": 14.03076923076923, "grad_norm": 0.921875, "learning_rate": 7.025569108558519e-05, "loss": 0.1252, "step": 19152 }, { "epoch": 14.031501831501831, "grad_norm": 0.70703125, "learning_rate": 7.025254938942571e-05, "loss": 0.1313, "step": 19153 }, { "epoch": 14.032234432234432, "grad_norm": 1.3828125, "learning_rate": 7.024940759862527e-05, "loss": 0.1135, "step": 19154 }, { "epoch": 14.032967032967033, "grad_norm": 1.59375, "learning_rate": 7.024626571319895e-05, "loss": 0.1203, "step": 19155 }, { "epoch": 14.033699633699634, "grad_norm": 1.4453125, "learning_rate": 7.024312373316178e-05, "loss": 0.1267, "step": 19156 }, { "epoch": 14.034432234432234, "grad_norm": 1.53125, "learning_rate": 7.023998165852881e-05, "loss": 0.1212, "step": 19157 }, { "epoch": 14.035164835164835, "grad_norm": 1.15625, "learning_rate": 7.023683948931512e-05, "loss": 0.1246, "step": 19158 }, { "epoch": 14.035897435897436, "grad_norm": 1.421875, "learning_rate": 7.023369722553576e-05, "loss": 0.1314, "step": 19159 }, { "epoch": 14.036630036630036, "grad_norm": 1.1171875, "learning_rate": 7.023055486720578e-05, "loss": 0.1409, "step": 19160 }, { "epoch": 14.037362637362637, "grad_norm": 0.62109375, "learning_rate": 7.022741241434023e-05, "loss": 0.1145, "step": 19161 }, { "epoch": 14.038095238095238, "grad_norm": 0.86328125, "learning_rate": 7.022426986695417e-05, "loss": 0.118, "step": 19162 }, { "epoch": 14.038827838827839, "grad_norm": 1.015625, "learning_rate": 7.022112722506266e-05, "loss": 0.1443, "step": 19163 }, { "epoch": 14.03956043956044, "grad_norm": 2.265625, "learning_rate": 7.021798448868073e-05, "loss": 0.119, "step": 19164 }, { "epoch": 14.04029304029304, "grad_norm": 1.3984375, "learning_rate": 7.02148416578235e-05, "loss": 0.0941, "step": 19165 }, { "epoch": 14.04102564102564, "grad_norm": 1.0, "learning_rate": 7.021169873250599e-05, "loss": 0.0809, "step": 19166 }, { "epoch": 14.041758241758242, "grad_norm": 0.9609375, "learning_rate": 7.020855571274324e-05, "loss": 0.1727, "step": 19167 }, { "epoch": 14.042490842490842, "grad_norm": 1.2265625, "learning_rate": 7.020541259855033e-05, "loss": 0.1106, "step": 19168 }, { "epoch": 14.043223443223443, "grad_norm": 1.0078125, "learning_rate": 7.020226938994234e-05, "loss": 0.1186, "step": 19169 }, { "epoch": 14.043956043956044, "grad_norm": 1.40625, "learning_rate": 7.019912608693429e-05, "loss": 0.1561, "step": 19170 }, { "epoch": 14.044688644688645, "grad_norm": 0.953125, "learning_rate": 7.019598268954126e-05, "loss": 0.1187, "step": 19171 }, { "epoch": 14.045421245421245, "grad_norm": 1.1015625, "learning_rate": 7.019283919777832e-05, "loss": 0.1385, "step": 19172 }, { "epoch": 14.046153846153846, "grad_norm": 0.9140625, "learning_rate": 7.018969561166052e-05, "loss": 0.1003, "step": 19173 }, { "epoch": 14.046886446886447, "grad_norm": 0.875, "learning_rate": 7.018655193120294e-05, "loss": 0.1353, "step": 19174 }, { "epoch": 14.047619047619047, "grad_norm": 0.69921875, "learning_rate": 7.01834081564206e-05, "loss": 0.0994, "step": 19175 }, { "epoch": 14.048351648351648, "grad_norm": 1.2890625, "learning_rate": 7.018026428732862e-05, "loss": 0.1069, "step": 19176 }, { "epoch": 14.049084249084249, "grad_norm": 0.91796875, "learning_rate": 7.017712032394199e-05, "loss": 0.1475, "step": 19177 }, { "epoch": 14.04981684981685, "grad_norm": 0.83984375, "learning_rate": 7.017397626627585e-05, "loss": 0.1099, "step": 19178 }, { "epoch": 14.05054945054945, "grad_norm": 0.8828125, "learning_rate": 7.017083211434523e-05, "loss": 0.1349, "step": 19179 }, { "epoch": 14.051282051282051, "grad_norm": 0.7734375, "learning_rate": 7.016768786816517e-05, "loss": 0.1144, "step": 19180 }, { "epoch": 14.052014652014652, "grad_norm": 0.890625, "learning_rate": 7.016454352775079e-05, "loss": 0.1153, "step": 19181 }, { "epoch": 14.052747252747253, "grad_norm": 1.078125, "learning_rate": 7.016139909311712e-05, "loss": 0.1176, "step": 19182 }, { "epoch": 14.053479853479853, "grad_norm": 1.359375, "learning_rate": 7.015825456427923e-05, "loss": 0.1515, "step": 19183 }, { "epoch": 14.054212454212454, "grad_norm": 1.7734375, "learning_rate": 7.015510994125218e-05, "loss": 0.1354, "step": 19184 }, { "epoch": 14.054945054945055, "grad_norm": 0.8515625, "learning_rate": 7.015196522405105e-05, "loss": 0.122, "step": 19185 }, { "epoch": 14.055677655677655, "grad_norm": 1.3046875, "learning_rate": 7.014882041269091e-05, "loss": 0.113, "step": 19186 }, { "epoch": 14.056410256410256, "grad_norm": 1.1015625, "learning_rate": 7.014567550718684e-05, "loss": 0.1294, "step": 19187 }, { "epoch": 14.057142857142857, "grad_norm": 1.3359375, "learning_rate": 7.014253050755385e-05, "loss": 0.1358, "step": 19188 }, { "epoch": 14.057875457875458, "grad_norm": 1.203125, "learning_rate": 7.013938541380708e-05, "loss": 0.1581, "step": 19189 }, { "epoch": 14.058608058608058, "grad_norm": 1.203125, "learning_rate": 7.013624022596155e-05, "loss": 0.1245, "step": 19190 }, { "epoch": 14.05934065934066, "grad_norm": 1.546875, "learning_rate": 7.013309494403235e-05, "loss": 0.1435, "step": 19191 }, { "epoch": 14.06007326007326, "grad_norm": 1.28125, "learning_rate": 7.012994956803455e-05, "loss": 0.1241, "step": 19192 }, { "epoch": 14.06080586080586, "grad_norm": 1.0078125, "learning_rate": 7.01268040979832e-05, "loss": 0.118, "step": 19193 }, { "epoch": 14.061538461538461, "grad_norm": 1.1015625, "learning_rate": 7.012365853389341e-05, "loss": 0.1167, "step": 19194 }, { "epoch": 14.062271062271062, "grad_norm": 1.4609375, "learning_rate": 7.012051287578022e-05, "loss": 0.1523, "step": 19195 }, { "epoch": 14.063003663003663, "grad_norm": 0.9296875, "learning_rate": 7.011736712365872e-05, "loss": 0.1369, "step": 19196 }, { "epoch": 14.063736263736264, "grad_norm": 0.98828125, "learning_rate": 7.011422127754396e-05, "loss": 0.1322, "step": 19197 }, { "epoch": 14.064468864468864, "grad_norm": 0.58984375, "learning_rate": 7.011107533745103e-05, "loss": 0.1169, "step": 19198 }, { "epoch": 14.065201465201465, "grad_norm": 0.9296875, "learning_rate": 7.010792930339503e-05, "loss": 0.1072, "step": 19199 }, { "epoch": 14.065934065934066, "grad_norm": 1.578125, "learning_rate": 7.010478317539096e-05, "loss": 0.1318, "step": 19200 }, { "epoch": 14.066666666666666, "grad_norm": 1.5078125, "learning_rate": 7.010163695345397e-05, "loss": 0.1072, "step": 19201 }, { "epoch": 14.067399267399267, "grad_norm": 1.125, "learning_rate": 7.00984906375991e-05, "loss": 0.1247, "step": 19202 }, { "epoch": 14.068131868131868, "grad_norm": 0.98828125, "learning_rate": 7.009534422784141e-05, "loss": 0.1098, "step": 19203 }, { "epoch": 14.068864468864469, "grad_norm": 0.578125, "learning_rate": 7.0092197724196e-05, "loss": 0.1107, "step": 19204 }, { "epoch": 14.06959706959707, "grad_norm": 0.98046875, "learning_rate": 7.008905112667795e-05, "loss": 0.1244, "step": 19205 }, { "epoch": 14.07032967032967, "grad_norm": 1.2421875, "learning_rate": 7.00859044353023e-05, "loss": 0.1296, "step": 19206 }, { "epoch": 14.07106227106227, "grad_norm": 1.4375, "learning_rate": 7.008275765008419e-05, "loss": 0.135, "step": 19207 }, { "epoch": 14.071794871794872, "grad_norm": 1.109375, "learning_rate": 7.007961077103862e-05, "loss": 0.1282, "step": 19208 }, { "epoch": 14.072527472527472, "grad_norm": 0.67578125, "learning_rate": 7.007646379818073e-05, "loss": 0.1214, "step": 19209 }, { "epoch": 14.073260073260073, "grad_norm": 1.3046875, "learning_rate": 7.007331673152557e-05, "loss": 0.1344, "step": 19210 }, { "epoch": 14.073992673992674, "grad_norm": 1.734375, "learning_rate": 7.007016957108822e-05, "loss": 0.1404, "step": 19211 }, { "epoch": 14.074725274725274, "grad_norm": 1.2421875, "learning_rate": 7.006702231688377e-05, "loss": 0.1264, "step": 19212 }, { "epoch": 14.075457875457875, "grad_norm": 0.640625, "learning_rate": 7.006387496892729e-05, "loss": 0.1147, "step": 19213 }, { "epoch": 14.076190476190476, "grad_norm": 0.77734375, "learning_rate": 7.006072752723385e-05, "loss": 0.1125, "step": 19214 }, { "epoch": 14.076923076923077, "grad_norm": 0.72265625, "learning_rate": 7.005757999181857e-05, "loss": 0.1196, "step": 19215 }, { "epoch": 14.077655677655677, "grad_norm": 1.34375, "learning_rate": 7.005443236269648e-05, "loss": 0.1228, "step": 19216 }, { "epoch": 14.078388278388278, "grad_norm": 0.6875, "learning_rate": 7.005128463988269e-05, "loss": 0.1096, "step": 19217 }, { "epoch": 14.079120879120879, "grad_norm": 1.0390625, "learning_rate": 7.00481368233923e-05, "loss": 0.1285, "step": 19218 }, { "epoch": 14.07985347985348, "grad_norm": 0.90234375, "learning_rate": 7.004498891324034e-05, "loss": 0.1278, "step": 19219 }, { "epoch": 14.08058608058608, "grad_norm": 1.046875, "learning_rate": 7.004184090944194e-05, "loss": 0.1211, "step": 19220 }, { "epoch": 14.081318681318681, "grad_norm": 0.765625, "learning_rate": 7.003869281201216e-05, "loss": 0.1531, "step": 19221 }, { "epoch": 14.082051282051282, "grad_norm": 1.140625, "learning_rate": 7.003554462096607e-05, "loss": 0.1235, "step": 19222 }, { "epoch": 14.082783882783882, "grad_norm": 0.98046875, "learning_rate": 7.00323963363188e-05, "loss": 0.1366, "step": 19223 }, { "epoch": 14.083516483516483, "grad_norm": 1.3359375, "learning_rate": 7.002924795808538e-05, "loss": 0.1449, "step": 19224 }, { "epoch": 14.084249084249084, "grad_norm": 0.859375, "learning_rate": 7.002609948628095e-05, "loss": 0.1368, "step": 19225 }, { "epoch": 14.084981684981685, "grad_norm": 1.1640625, "learning_rate": 7.002295092092057e-05, "loss": 0.1211, "step": 19226 }, { "epoch": 14.085714285714285, "grad_norm": 0.9765625, "learning_rate": 7.001980226201929e-05, "loss": 0.1174, "step": 19227 }, { "epoch": 14.086446886446886, "grad_norm": 1.0546875, "learning_rate": 7.001665350959225e-05, "loss": 0.0922, "step": 19228 }, { "epoch": 14.087179487179487, "grad_norm": 1.171875, "learning_rate": 7.001350466365452e-05, "loss": 0.1337, "step": 19229 }, { "epoch": 14.087912087912088, "grad_norm": 1.484375, "learning_rate": 7.001035572422117e-05, "loss": 0.1525, "step": 19230 }, { "epoch": 14.088644688644688, "grad_norm": 0.74609375, "learning_rate": 7.000720669130732e-05, "loss": 0.1113, "step": 19231 }, { "epoch": 14.089377289377289, "grad_norm": 1.2734375, "learning_rate": 7.000405756492803e-05, "loss": 0.1402, "step": 19232 }, { "epoch": 14.09010989010989, "grad_norm": 0.60546875, "learning_rate": 7.000090834509838e-05, "loss": 0.136, "step": 19233 }, { "epoch": 14.09084249084249, "grad_norm": 0.81640625, "learning_rate": 6.999775903183351e-05, "loss": 0.1025, "step": 19234 }, { "epoch": 14.091575091575091, "grad_norm": 1.078125, "learning_rate": 6.999460962514844e-05, "loss": 0.1234, "step": 19235 }, { "epoch": 14.092307692307692, "grad_norm": 1.3984375, "learning_rate": 6.999146012505832e-05, "loss": 0.1181, "step": 19236 }, { "epoch": 14.093040293040293, "grad_norm": 1.25, "learning_rate": 6.99883105315782e-05, "loss": 0.1413, "step": 19237 }, { "epoch": 14.093772893772893, "grad_norm": 1.09375, "learning_rate": 6.99851608447232e-05, "loss": 0.1809, "step": 19238 }, { "epoch": 14.094505494505494, "grad_norm": 0.8515625, "learning_rate": 6.998201106450841e-05, "loss": 0.1149, "step": 19239 }, { "epoch": 14.095238095238095, "grad_norm": 1.578125, "learning_rate": 6.997886119094888e-05, "loss": 0.134, "step": 19240 }, { "epoch": 14.095970695970696, "grad_norm": 0.8359375, "learning_rate": 6.997571122405974e-05, "loss": 0.1061, "step": 19241 }, { "epoch": 14.096703296703296, "grad_norm": 1.0078125, "learning_rate": 6.997256116385608e-05, "loss": 0.1246, "step": 19242 }, { "epoch": 14.097435897435897, "grad_norm": 1.25, "learning_rate": 6.9969411010353e-05, "loss": 0.1433, "step": 19243 }, { "epoch": 14.098168498168498, "grad_norm": 0.64453125, "learning_rate": 6.996626076356557e-05, "loss": 0.1151, "step": 19244 }, { "epoch": 14.098901098901099, "grad_norm": 1.5625, "learning_rate": 6.996311042350889e-05, "loss": 0.1602, "step": 19245 }, { "epoch": 14.0996336996337, "grad_norm": 1.109375, "learning_rate": 6.995995999019804e-05, "loss": 0.107, "step": 19246 }, { "epoch": 14.1003663003663, "grad_norm": 0.85546875, "learning_rate": 6.995680946364816e-05, "loss": 0.1365, "step": 19247 }, { "epoch": 14.1010989010989, "grad_norm": 0.8671875, "learning_rate": 6.99536588438743e-05, "loss": 0.1134, "step": 19248 }, { "epoch": 14.101831501831501, "grad_norm": 0.78125, "learning_rate": 6.995050813089159e-05, "loss": 0.1363, "step": 19249 }, { "epoch": 14.102564102564102, "grad_norm": 0.82421875, "learning_rate": 6.994735732471512e-05, "loss": 0.1051, "step": 19250 }, { "epoch": 14.103296703296703, "grad_norm": 1.3515625, "learning_rate": 6.994420642535995e-05, "loss": 0.1122, "step": 19251 }, { "epoch": 14.104029304029304, "grad_norm": 1.03125, "learning_rate": 6.994105543284122e-05, "loss": 0.1149, "step": 19252 }, { "epoch": 14.104761904761904, "grad_norm": 1.4921875, "learning_rate": 6.9937904347174e-05, "loss": 0.1343, "step": 19253 }, { "epoch": 14.105494505494505, "grad_norm": 1.1171875, "learning_rate": 6.993475316837341e-05, "loss": 0.1506, "step": 19254 }, { "epoch": 14.106227106227106, "grad_norm": 1.3515625, "learning_rate": 6.993160189645453e-05, "loss": 0.1213, "step": 19255 }, { "epoch": 14.106959706959707, "grad_norm": 0.80859375, "learning_rate": 6.992845053143247e-05, "loss": 0.1094, "step": 19256 }, { "epoch": 14.107692307692307, "grad_norm": 2.03125, "learning_rate": 6.992529907332233e-05, "loss": 0.147, "step": 19257 }, { "epoch": 14.108424908424908, "grad_norm": 1.3125, "learning_rate": 6.992214752213921e-05, "loss": 0.1291, "step": 19258 }, { "epoch": 14.109157509157509, "grad_norm": 1.1796875, "learning_rate": 6.991899587789819e-05, "loss": 0.1206, "step": 19259 }, { "epoch": 14.10989010989011, "grad_norm": 1.1171875, "learning_rate": 6.99158441406144e-05, "loss": 0.1068, "step": 19260 }, { "epoch": 14.11062271062271, "grad_norm": 1.09375, "learning_rate": 6.991269231030291e-05, "loss": 0.1026, "step": 19261 }, { "epoch": 14.111355311355311, "grad_norm": 1.1328125, "learning_rate": 6.990954038697886e-05, "loss": 0.1371, "step": 19262 }, { "epoch": 14.112087912087912, "grad_norm": 0.8125, "learning_rate": 6.990638837065731e-05, "loss": 0.1318, "step": 19263 }, { "epoch": 14.112820512820512, "grad_norm": 0.984375, "learning_rate": 6.990323626135339e-05, "loss": 0.1117, "step": 19264 }, { "epoch": 14.113553113553113, "grad_norm": 0.7578125, "learning_rate": 6.990008405908221e-05, "loss": 0.1513, "step": 19265 }, { "epoch": 14.114285714285714, "grad_norm": 0.7734375, "learning_rate": 6.989693176385883e-05, "loss": 0.1123, "step": 19266 }, { "epoch": 14.115018315018315, "grad_norm": 0.89453125, "learning_rate": 6.98937793756984e-05, "loss": 0.1234, "step": 19267 }, { "epoch": 14.115750915750915, "grad_norm": 1.5390625, "learning_rate": 6.989062689461601e-05, "loss": 0.118, "step": 19268 }, { "epoch": 14.116483516483516, "grad_norm": 0.83203125, "learning_rate": 6.988747432062674e-05, "loss": 0.1167, "step": 19269 }, { "epoch": 14.117216117216117, "grad_norm": 1.1015625, "learning_rate": 6.988432165374575e-05, "loss": 0.1348, "step": 19270 }, { "epoch": 14.117948717948718, "grad_norm": 1.046875, "learning_rate": 6.988116889398809e-05, "loss": 0.1256, "step": 19271 }, { "epoch": 14.118681318681318, "grad_norm": 0.7265625, "learning_rate": 6.987801604136887e-05, "loss": 0.1203, "step": 19272 }, { "epoch": 14.119413919413919, "grad_norm": 1.2265625, "learning_rate": 6.987486309590325e-05, "loss": 0.1269, "step": 19273 }, { "epoch": 14.12014652014652, "grad_norm": 1.1875, "learning_rate": 6.987171005760628e-05, "loss": 0.1166, "step": 19274 }, { "epoch": 14.12087912087912, "grad_norm": 1.2265625, "learning_rate": 6.986855692649308e-05, "loss": 0.1413, "step": 19275 }, { "epoch": 14.121611721611721, "grad_norm": 1.65625, "learning_rate": 6.986540370257878e-05, "loss": 0.1702, "step": 19276 }, { "epoch": 14.122344322344322, "grad_norm": 1.265625, "learning_rate": 6.986225038587847e-05, "loss": 0.1101, "step": 19277 }, { "epoch": 14.123076923076923, "grad_norm": 1.0234375, "learning_rate": 6.985909697640727e-05, "loss": 0.1045, "step": 19278 }, { "epoch": 14.123809523809523, "grad_norm": 1.421875, "learning_rate": 6.985594347418028e-05, "loss": 0.109, "step": 19279 }, { "epoch": 14.124542124542124, "grad_norm": 1.09375, "learning_rate": 6.98527898792126e-05, "loss": 0.1717, "step": 19280 }, { "epoch": 14.125274725274725, "grad_norm": 1.2578125, "learning_rate": 6.984963619151935e-05, "loss": 0.1177, "step": 19281 }, { "epoch": 14.126007326007326, "grad_norm": 1.2421875, "learning_rate": 6.984648241111566e-05, "loss": 0.1442, "step": 19282 }, { "epoch": 14.126739926739926, "grad_norm": 1.28125, "learning_rate": 6.98433285380166e-05, "loss": 0.1269, "step": 19283 }, { "epoch": 14.127472527472527, "grad_norm": 1.296875, "learning_rate": 6.984017457223732e-05, "loss": 0.1346, "step": 19284 }, { "epoch": 14.128205128205128, "grad_norm": 1.0390625, "learning_rate": 6.98370205137929e-05, "loss": 0.1375, "step": 19285 }, { "epoch": 14.128937728937728, "grad_norm": 0.73828125, "learning_rate": 6.983386636269848e-05, "loss": 0.1191, "step": 19286 }, { "epoch": 14.12967032967033, "grad_norm": 0.87109375, "learning_rate": 6.983071211896913e-05, "loss": 0.1121, "step": 19287 }, { "epoch": 14.13040293040293, "grad_norm": 0.76953125, "learning_rate": 6.982755778262003e-05, "loss": 0.0989, "step": 19288 }, { "epoch": 14.13113553113553, "grad_norm": 0.984375, "learning_rate": 6.982440335366624e-05, "loss": 0.1407, "step": 19289 }, { "epoch": 14.131868131868131, "grad_norm": 0.96875, "learning_rate": 6.982124883212288e-05, "loss": 0.1196, "step": 19290 }, { "epoch": 14.132600732600732, "grad_norm": 1.078125, "learning_rate": 6.981809421800509e-05, "loss": 0.1152, "step": 19291 }, { "epoch": 14.133333333333333, "grad_norm": 0.90625, "learning_rate": 6.981493951132795e-05, "loss": 0.1106, "step": 19292 }, { "epoch": 14.134065934065934, "grad_norm": 1.734375, "learning_rate": 6.981178471210663e-05, "loss": 0.1445, "step": 19293 }, { "epoch": 14.134798534798534, "grad_norm": 1.0859375, "learning_rate": 6.980862982035618e-05, "loss": 0.1644, "step": 19294 }, { "epoch": 14.135531135531135, "grad_norm": 1.4375, "learning_rate": 6.980547483609174e-05, "loss": 0.1114, "step": 19295 }, { "epoch": 14.136263736263736, "grad_norm": 1.03125, "learning_rate": 6.980231975932847e-05, "loss": 0.1056, "step": 19296 }, { "epoch": 14.136996336996337, "grad_norm": 1.421875, "learning_rate": 6.97991645900814e-05, "loss": 0.1336, "step": 19297 }, { "epoch": 14.137728937728937, "grad_norm": 1.1484375, "learning_rate": 6.979600932836572e-05, "loss": 0.1461, "step": 19298 }, { "epoch": 14.138461538461538, "grad_norm": 1.0703125, "learning_rate": 6.979285397419655e-05, "loss": 0.1386, "step": 19299 }, { "epoch": 14.139194139194139, "grad_norm": 1.015625, "learning_rate": 6.978969852758894e-05, "loss": 0.106, "step": 19300 }, { "epoch": 14.13992673992674, "grad_norm": 1.234375, "learning_rate": 6.978654298855807e-05, "loss": 0.144, "step": 19301 }, { "epoch": 14.14065934065934, "grad_norm": 1.78125, "learning_rate": 6.978338735711904e-05, "loss": 0.1323, "step": 19302 }, { "epoch": 14.141391941391941, "grad_norm": 1.0703125, "learning_rate": 6.978023163328699e-05, "loss": 0.1119, "step": 19303 }, { "epoch": 14.142124542124542, "grad_norm": 1.859375, "learning_rate": 6.977707581707702e-05, "loss": 0.1359, "step": 19304 }, { "epoch": 14.142857142857142, "grad_norm": 1.296875, "learning_rate": 6.977391990850421e-05, "loss": 0.1271, "step": 19305 }, { "epoch": 14.143589743589743, "grad_norm": 1.0859375, "learning_rate": 6.977076390758377e-05, "loss": 0.1098, "step": 19306 }, { "epoch": 14.144322344322344, "grad_norm": 0.71484375, "learning_rate": 6.976760781433074e-05, "loss": 0.136, "step": 19307 }, { "epoch": 14.145054945054945, "grad_norm": 1.15625, "learning_rate": 6.97644516287603e-05, "loss": 0.1221, "step": 19308 }, { "epoch": 14.145787545787545, "grad_norm": 1.09375, "learning_rate": 6.976129535088753e-05, "loss": 0.1139, "step": 19309 }, { "epoch": 14.146520146520146, "grad_norm": 1.515625, "learning_rate": 6.975813898072757e-05, "loss": 0.1558, "step": 19310 }, { "epoch": 14.147252747252747, "grad_norm": 1.03125, "learning_rate": 6.975498251829555e-05, "loss": 0.1313, "step": 19311 }, { "epoch": 14.147985347985347, "grad_norm": 1.3359375, "learning_rate": 6.97518259636066e-05, "loss": 0.1185, "step": 19312 }, { "epoch": 14.148717948717948, "grad_norm": 1.125, "learning_rate": 6.974866931667581e-05, "loss": 0.132, "step": 19313 }, { "epoch": 14.149450549450549, "grad_norm": 1.5859375, "learning_rate": 6.974551257751833e-05, "loss": 0.097, "step": 19314 }, { "epoch": 14.15018315018315, "grad_norm": 1.7421875, "learning_rate": 6.974235574614929e-05, "loss": 0.1331, "step": 19315 }, { "epoch": 14.15091575091575, "grad_norm": 1.0703125, "learning_rate": 6.973919882258382e-05, "loss": 0.1305, "step": 19316 }, { "epoch": 14.151648351648351, "grad_norm": 2.15625, "learning_rate": 6.973604180683702e-05, "loss": 0.1366, "step": 19317 }, { "epoch": 14.152380952380952, "grad_norm": 2.0625, "learning_rate": 6.973288469892403e-05, "loss": 0.1527, "step": 19318 }, { "epoch": 14.153113553113553, "grad_norm": 1.1484375, "learning_rate": 6.972972749885998e-05, "loss": 0.1355, "step": 19319 }, { "epoch": 14.153846153846153, "grad_norm": 1.65625, "learning_rate": 6.972657020665997e-05, "loss": 0.0987, "step": 19320 }, { "epoch": 14.154578754578754, "grad_norm": 1.34375, "learning_rate": 6.972341282233916e-05, "loss": 0.116, "step": 19321 }, { "epoch": 14.155311355311355, "grad_norm": 1.0390625, "learning_rate": 6.972025534591268e-05, "loss": 0.1396, "step": 19322 }, { "epoch": 14.156043956043955, "grad_norm": 1.84375, "learning_rate": 6.971709777739565e-05, "loss": 0.1103, "step": 19323 }, { "epoch": 14.156776556776556, "grad_norm": 1.3515625, "learning_rate": 6.971394011680317e-05, "loss": 0.1348, "step": 19324 }, { "epoch": 14.157509157509157, "grad_norm": 1.5, "learning_rate": 6.971078236415042e-05, "loss": 0.1575, "step": 19325 }, { "epoch": 14.158241758241758, "grad_norm": 1.7734375, "learning_rate": 6.97076245194525e-05, "loss": 0.142, "step": 19326 }, { "epoch": 14.158974358974358, "grad_norm": 1.671875, "learning_rate": 6.970446658272455e-05, "loss": 0.1535, "step": 19327 }, { "epoch": 14.15970695970696, "grad_norm": 1.0546875, "learning_rate": 6.970130855398168e-05, "loss": 0.1027, "step": 19328 }, { "epoch": 14.16043956043956, "grad_norm": 1.1640625, "learning_rate": 6.969815043323905e-05, "loss": 0.1121, "step": 19329 }, { "epoch": 14.16117216117216, "grad_norm": 0.921875, "learning_rate": 6.96949922205118e-05, "loss": 0.0993, "step": 19330 }, { "epoch": 14.161904761904761, "grad_norm": 1.578125, "learning_rate": 6.9691833915815e-05, "loss": 0.1425, "step": 19331 }, { "epoch": 14.162637362637362, "grad_norm": 1.4375, "learning_rate": 6.968867551916385e-05, "loss": 0.1341, "step": 19332 }, { "epoch": 14.163369963369963, "grad_norm": 1.5234375, "learning_rate": 6.968551703057343e-05, "loss": 0.144, "step": 19333 }, { "epoch": 14.164102564102564, "grad_norm": 0.96484375, "learning_rate": 6.968235845005893e-05, "loss": 0.1165, "step": 19334 }, { "epoch": 14.164835164835164, "grad_norm": 1.46875, "learning_rate": 6.967919977763544e-05, "loss": 0.138, "step": 19335 }, { "epoch": 14.165567765567765, "grad_norm": 1.5390625, "learning_rate": 6.967604101331811e-05, "loss": 0.1371, "step": 19336 }, { "epoch": 14.166300366300366, "grad_norm": 1.15625, "learning_rate": 6.967288215712206e-05, "loss": 0.109, "step": 19337 }, { "epoch": 14.167032967032966, "grad_norm": 1.0859375, "learning_rate": 6.966972320906246e-05, "loss": 0.1525, "step": 19338 }, { "epoch": 14.167765567765567, "grad_norm": 1.578125, "learning_rate": 6.96665641691544e-05, "loss": 0.1267, "step": 19339 }, { "epoch": 14.168498168498168, "grad_norm": 0.7421875, "learning_rate": 6.966340503741305e-05, "loss": 0.1213, "step": 19340 }, { "epoch": 14.169230769230769, "grad_norm": 1.1015625, "learning_rate": 6.966024581385353e-05, "loss": 0.1216, "step": 19341 }, { "epoch": 14.16996336996337, "grad_norm": 0.70703125, "learning_rate": 6.9657086498491e-05, "loss": 0.1039, "step": 19342 }, { "epoch": 14.17069597069597, "grad_norm": 0.87109375, "learning_rate": 6.965392709134056e-05, "loss": 0.1305, "step": 19343 }, { "epoch": 14.17142857142857, "grad_norm": 0.57421875, "learning_rate": 6.965076759241737e-05, "loss": 0.1141, "step": 19344 }, { "epoch": 14.172161172161172, "grad_norm": 0.6328125, "learning_rate": 6.964760800173658e-05, "loss": 0.1012, "step": 19345 }, { "epoch": 14.172893772893772, "grad_norm": 1.4921875, "learning_rate": 6.96444483193133e-05, "loss": 0.1446, "step": 19346 }, { "epoch": 14.173626373626373, "grad_norm": 0.87890625, "learning_rate": 6.964128854516269e-05, "loss": 0.1251, "step": 19347 }, { "epoch": 14.174358974358974, "grad_norm": 1.375, "learning_rate": 6.963812867929987e-05, "loss": 0.1513, "step": 19348 }, { "epoch": 14.175091575091574, "grad_norm": 1.3828125, "learning_rate": 6.963496872174002e-05, "loss": 0.1278, "step": 19349 }, { "epoch": 14.175824175824175, "grad_norm": 1.140625, "learning_rate": 6.963180867249824e-05, "loss": 0.1549, "step": 19350 }, { "epoch": 14.176556776556776, "grad_norm": 0.875, "learning_rate": 6.96286485315897e-05, "loss": 0.1517, "step": 19351 }, { "epoch": 14.177289377289377, "grad_norm": 0.70703125, "learning_rate": 6.962548829902948e-05, "loss": 0.138, "step": 19352 }, { "epoch": 14.178021978021977, "grad_norm": 1.1640625, "learning_rate": 6.96223279748328e-05, "loss": 0.1039, "step": 19353 }, { "epoch": 14.178754578754578, "grad_norm": 1.5234375, "learning_rate": 6.961916755901476e-05, "loss": 0.113, "step": 19354 }, { "epoch": 14.179487179487179, "grad_norm": 1.21875, "learning_rate": 6.961600705159052e-05, "loss": 0.0907, "step": 19355 }, { "epoch": 14.18021978021978, "grad_norm": 1.3359375, "learning_rate": 6.961284645257521e-05, "loss": 0.1199, "step": 19356 }, { "epoch": 14.18095238095238, "grad_norm": 1.6015625, "learning_rate": 6.960968576198397e-05, "loss": 0.1356, "step": 19357 }, { "epoch": 14.181684981684981, "grad_norm": 1.4921875, "learning_rate": 6.960652497983198e-05, "loss": 0.1178, "step": 19358 }, { "epoch": 14.182417582417582, "grad_norm": 1.7109375, "learning_rate": 6.960336410613433e-05, "loss": 0.1527, "step": 19359 }, { "epoch": 14.183150183150182, "grad_norm": 1.4921875, "learning_rate": 6.96002031409062e-05, "loss": 0.1559, "step": 19360 }, { "epoch": 14.183882783882783, "grad_norm": 1.53125, "learning_rate": 6.959704208416272e-05, "loss": 0.1053, "step": 19361 }, { "epoch": 14.184615384615384, "grad_norm": 0.984375, "learning_rate": 6.959388093591906e-05, "loss": 0.1406, "step": 19362 }, { "epoch": 14.185347985347985, "grad_norm": 1.96875, "learning_rate": 6.959071969619031e-05, "loss": 0.1446, "step": 19363 }, { "epoch": 14.186080586080585, "grad_norm": 0.8046875, "learning_rate": 6.958755836499169e-05, "loss": 0.1104, "step": 19364 }, { "epoch": 14.186813186813186, "grad_norm": 0.8984375, "learning_rate": 6.958439694233831e-05, "loss": 0.119, "step": 19365 }, { "epoch": 14.187545787545787, "grad_norm": 1.1875, "learning_rate": 6.958123542824532e-05, "loss": 0.1528, "step": 19366 }, { "epoch": 14.188278388278388, "grad_norm": 1.265625, "learning_rate": 6.957807382272784e-05, "loss": 0.1387, "step": 19367 }, { "epoch": 14.189010989010988, "grad_norm": 1.3359375, "learning_rate": 6.957491212580106e-05, "loss": 0.1298, "step": 19368 }, { "epoch": 14.189743589743589, "grad_norm": 1.40625, "learning_rate": 6.957175033748014e-05, "loss": 0.1339, "step": 19369 }, { "epoch": 14.19047619047619, "grad_norm": 0.99609375, "learning_rate": 6.956858845778017e-05, "loss": 0.1302, "step": 19370 }, { "epoch": 14.19120879120879, "grad_norm": 1.3125, "learning_rate": 6.956542648671635e-05, "loss": 0.1442, "step": 19371 }, { "epoch": 14.191941391941391, "grad_norm": 1.046875, "learning_rate": 6.956226442430382e-05, "loss": 0.1298, "step": 19372 }, { "epoch": 14.192673992673992, "grad_norm": 0.64453125, "learning_rate": 6.95591022705577e-05, "loss": 0.1253, "step": 19373 }, { "epoch": 14.193406593406593, "grad_norm": 1.4453125, "learning_rate": 6.955594002549318e-05, "loss": 0.1321, "step": 19374 }, { "epoch": 14.194139194139193, "grad_norm": 1.390625, "learning_rate": 6.955277768912538e-05, "loss": 0.1137, "step": 19375 }, { "epoch": 14.194871794871794, "grad_norm": 0.99609375, "learning_rate": 6.954961526146947e-05, "loss": 0.1383, "step": 19376 }, { "epoch": 14.195604395604395, "grad_norm": 0.96875, "learning_rate": 6.954645274254063e-05, "loss": 0.0944, "step": 19377 }, { "epoch": 14.196336996336996, "grad_norm": 1.03125, "learning_rate": 6.954329013235393e-05, "loss": 0.1048, "step": 19378 }, { "epoch": 14.197069597069596, "grad_norm": 0.69921875, "learning_rate": 6.954012743092462e-05, "loss": 0.114, "step": 19379 }, { "epoch": 14.197802197802197, "grad_norm": 0.9375, "learning_rate": 6.953696463826779e-05, "loss": 0.1305, "step": 19380 }, { "epoch": 14.198534798534798, "grad_norm": 1.671875, "learning_rate": 6.953380175439863e-05, "loss": 0.1289, "step": 19381 }, { "epoch": 14.199267399267399, "grad_norm": 1.203125, "learning_rate": 6.953063877933227e-05, "loss": 0.1046, "step": 19382 }, { "epoch": 14.2, "grad_norm": 1.6171875, "learning_rate": 6.952747571308386e-05, "loss": 0.1245, "step": 19383 }, { "epoch": 14.2007326007326, "grad_norm": 1.1640625, "learning_rate": 6.952431255566857e-05, "loss": 0.1329, "step": 19384 }, { "epoch": 14.2014652014652, "grad_norm": 0.6328125, "learning_rate": 6.952114930710157e-05, "loss": 0.1078, "step": 19385 }, { "epoch": 14.202197802197801, "grad_norm": 1.9375, "learning_rate": 6.951798596739798e-05, "loss": 0.1337, "step": 19386 }, { "epoch": 14.202930402930402, "grad_norm": 1.109375, "learning_rate": 6.951482253657299e-05, "loss": 0.1138, "step": 19387 }, { "epoch": 14.203663003663003, "grad_norm": 1.515625, "learning_rate": 6.951165901464173e-05, "loss": 0.1365, "step": 19388 }, { "epoch": 14.204395604395604, "grad_norm": 0.9453125, "learning_rate": 6.950849540161938e-05, "loss": 0.1247, "step": 19389 }, { "epoch": 14.205128205128204, "grad_norm": 0.96484375, "learning_rate": 6.950533169752109e-05, "loss": 0.1191, "step": 19390 }, { "epoch": 14.205860805860805, "grad_norm": 1.5234375, "learning_rate": 6.950216790236199e-05, "loss": 0.1405, "step": 19391 }, { "epoch": 14.206593406593406, "grad_norm": 0.90234375, "learning_rate": 6.949900401615731e-05, "loss": 0.1114, "step": 19392 }, { "epoch": 14.207326007326007, "grad_norm": 1.0390625, "learning_rate": 6.949584003892213e-05, "loss": 0.1444, "step": 19393 }, { "epoch": 14.208058608058607, "grad_norm": 0.8984375, "learning_rate": 6.949267597067167e-05, "loss": 0.1049, "step": 19394 }, { "epoch": 14.208791208791208, "grad_norm": 1.609375, "learning_rate": 6.948951181142106e-05, "loss": 0.1413, "step": 19395 }, { "epoch": 14.209523809523809, "grad_norm": 0.921875, "learning_rate": 6.948634756118544e-05, "loss": 0.1353, "step": 19396 }, { "epoch": 14.21025641025641, "grad_norm": 0.98828125, "learning_rate": 6.948318321998003e-05, "loss": 0.1288, "step": 19397 }, { "epoch": 14.21098901098901, "grad_norm": 0.88671875, "learning_rate": 6.948001878781994e-05, "loss": 0.1435, "step": 19398 }, { "epoch": 14.211721611721611, "grad_norm": 1.2421875, "learning_rate": 6.947685426472033e-05, "loss": 0.1446, "step": 19399 }, { "epoch": 14.212454212454212, "grad_norm": 0.91796875, "learning_rate": 6.94736896506964e-05, "loss": 0.1172, "step": 19400 }, { "epoch": 14.213186813186812, "grad_norm": 0.86328125, "learning_rate": 6.94705249457633e-05, "loss": 0.1293, "step": 19401 }, { "epoch": 14.213919413919413, "grad_norm": 1.203125, "learning_rate": 6.946736014993618e-05, "loss": 0.1463, "step": 19402 }, { "epoch": 14.214652014652014, "grad_norm": 0.67578125, "learning_rate": 6.946419526323022e-05, "loss": 0.1213, "step": 19403 }, { "epoch": 14.215384615384615, "grad_norm": 0.953125, "learning_rate": 6.946103028566057e-05, "loss": 0.122, "step": 19404 }, { "epoch": 14.216117216117215, "grad_norm": 1.015625, "learning_rate": 6.94578652172424e-05, "loss": 0.1571, "step": 19405 }, { "epoch": 14.216849816849816, "grad_norm": 1.1171875, "learning_rate": 6.945470005799087e-05, "loss": 0.1149, "step": 19406 }, { "epoch": 14.217582417582417, "grad_norm": 1.4375, "learning_rate": 6.945153480792116e-05, "loss": 0.1409, "step": 19407 }, { "epoch": 14.218315018315018, "grad_norm": 0.8203125, "learning_rate": 6.944836946704842e-05, "loss": 0.1338, "step": 19408 }, { "epoch": 14.219047619047618, "grad_norm": 0.94140625, "learning_rate": 6.944520403538781e-05, "loss": 0.1075, "step": 19409 }, { "epoch": 14.219780219780219, "grad_norm": 1.5859375, "learning_rate": 6.944203851295452e-05, "loss": 0.1483, "step": 19410 }, { "epoch": 14.22051282051282, "grad_norm": 1.421875, "learning_rate": 6.943887289976371e-05, "loss": 0.1343, "step": 19411 }, { "epoch": 14.22124542124542, "grad_norm": 1.1484375, "learning_rate": 6.943570719583053e-05, "loss": 0.1122, "step": 19412 }, { "epoch": 14.221978021978021, "grad_norm": 0.8515625, "learning_rate": 6.943254140117016e-05, "loss": 0.1411, "step": 19413 }, { "epoch": 14.222710622710622, "grad_norm": 1.0, "learning_rate": 6.942937551579777e-05, "loss": 0.1104, "step": 19414 }, { "epoch": 14.223443223443223, "grad_norm": 1.640625, "learning_rate": 6.942620953972853e-05, "loss": 0.1216, "step": 19415 }, { "epoch": 14.224175824175823, "grad_norm": 0.6953125, "learning_rate": 6.942304347297761e-05, "loss": 0.0931, "step": 19416 }, { "epoch": 14.224908424908424, "grad_norm": 1.125, "learning_rate": 6.941987731556017e-05, "loss": 0.125, "step": 19417 }, { "epoch": 14.225641025641025, "grad_norm": 1.0703125, "learning_rate": 6.94167110674914e-05, "loss": 0.1232, "step": 19418 }, { "epoch": 14.226373626373626, "grad_norm": 1.375, "learning_rate": 6.941354472878645e-05, "loss": 0.1108, "step": 19419 }, { "epoch": 14.227106227106226, "grad_norm": 1.25, "learning_rate": 6.94103782994605e-05, "loss": 0.1227, "step": 19420 }, { "epoch": 14.227838827838827, "grad_norm": 1.140625, "learning_rate": 6.940721177952872e-05, "loss": 0.1399, "step": 19421 }, { "epoch": 14.228571428571428, "grad_norm": 1.5546875, "learning_rate": 6.940404516900629e-05, "loss": 0.1242, "step": 19422 }, { "epoch": 14.229304029304028, "grad_norm": 1.2890625, "learning_rate": 6.940087846790837e-05, "loss": 0.1153, "step": 19423 }, { "epoch": 14.23003663003663, "grad_norm": 0.89453125, "learning_rate": 6.939771167625013e-05, "loss": 0.1011, "step": 19424 }, { "epoch": 14.23076923076923, "grad_norm": 1.484375, "learning_rate": 6.939454479404674e-05, "loss": 0.1401, "step": 19425 }, { "epoch": 14.23150183150183, "grad_norm": 1.4453125, "learning_rate": 6.939137782131338e-05, "loss": 0.1184, "step": 19426 }, { "epoch": 14.232234432234431, "grad_norm": 1.03125, "learning_rate": 6.938821075806524e-05, "loss": 0.1341, "step": 19427 }, { "epoch": 14.232967032967032, "grad_norm": 1.09375, "learning_rate": 6.938504360431749e-05, "loss": 0.1304, "step": 19428 }, { "epoch": 14.233699633699633, "grad_norm": 1.09375, "learning_rate": 6.938187636008528e-05, "loss": 0.1278, "step": 19429 }, { "epoch": 14.234432234432234, "grad_norm": 1.3203125, "learning_rate": 6.937870902538381e-05, "loss": 0.1088, "step": 19430 }, { "epoch": 14.235164835164834, "grad_norm": 1.3671875, "learning_rate": 6.937554160022825e-05, "loss": 0.1277, "step": 19431 }, { "epoch": 14.235897435897435, "grad_norm": 0.8359375, "learning_rate": 6.937237408463378e-05, "loss": 0.1047, "step": 19432 }, { "epoch": 14.236630036630036, "grad_norm": 0.68359375, "learning_rate": 6.936920647861556e-05, "loss": 0.1262, "step": 19433 }, { "epoch": 14.237362637362637, "grad_norm": 1.2265625, "learning_rate": 6.936603878218878e-05, "loss": 0.121, "step": 19434 }, { "epoch": 14.238095238095237, "grad_norm": 1.34375, "learning_rate": 6.936287099536861e-05, "loss": 0.1619, "step": 19435 }, { "epoch": 14.238827838827838, "grad_norm": 1.5078125, "learning_rate": 6.935970311817024e-05, "loss": 0.1475, "step": 19436 }, { "epoch": 14.239560439560439, "grad_norm": 1.53125, "learning_rate": 6.935653515060886e-05, "loss": 0.1401, "step": 19437 }, { "epoch": 14.24029304029304, "grad_norm": 2.03125, "learning_rate": 6.935336709269959e-05, "loss": 0.1634, "step": 19438 }, { "epoch": 14.24102564102564, "grad_norm": 1.0625, "learning_rate": 6.935019894445766e-05, "loss": 0.1235, "step": 19439 }, { "epoch": 14.241758241758241, "grad_norm": 1.53125, "learning_rate": 6.934703070589824e-05, "loss": 0.1223, "step": 19440 }, { "epoch": 14.242490842490842, "grad_norm": 1.4921875, "learning_rate": 6.934386237703652e-05, "loss": 0.1329, "step": 19441 }, { "epoch": 14.243223443223442, "grad_norm": 2.296875, "learning_rate": 6.934069395788768e-05, "loss": 0.1544, "step": 19442 }, { "epoch": 14.243956043956043, "grad_norm": 1.5390625, "learning_rate": 6.933752544846686e-05, "loss": 0.1201, "step": 19443 }, { "epoch": 14.244688644688644, "grad_norm": 0.75, "learning_rate": 6.933435684878928e-05, "loss": 0.092, "step": 19444 }, { "epoch": 14.245421245421245, "grad_norm": 0.85546875, "learning_rate": 6.933118815887013e-05, "loss": 0.1464, "step": 19445 }, { "epoch": 14.246153846153845, "grad_norm": 0.6796875, "learning_rate": 6.932801937872456e-05, "loss": 0.1157, "step": 19446 }, { "epoch": 14.246886446886446, "grad_norm": 0.80078125, "learning_rate": 6.932485050836778e-05, "loss": 0.1244, "step": 19447 }, { "epoch": 14.247619047619047, "grad_norm": 0.83203125, "learning_rate": 6.932168154781495e-05, "loss": 0.1032, "step": 19448 }, { "epoch": 14.248351648351647, "grad_norm": 0.8359375, "learning_rate": 6.931851249708127e-05, "loss": 0.1373, "step": 19449 }, { "epoch": 14.249084249084248, "grad_norm": 1.9375, "learning_rate": 6.931534335618193e-05, "loss": 0.1388, "step": 19450 }, { "epoch": 14.249816849816849, "grad_norm": 0.83984375, "learning_rate": 6.93121741251321e-05, "loss": 0.0992, "step": 19451 }, { "epoch": 14.25054945054945, "grad_norm": 0.84765625, "learning_rate": 6.930900480394694e-05, "loss": 0.1196, "step": 19452 }, { "epoch": 14.25128205128205, "grad_norm": 3.078125, "learning_rate": 6.930583539264169e-05, "loss": 0.1671, "step": 19453 }, { "epoch": 14.252014652014651, "grad_norm": 1.5546875, "learning_rate": 6.93026658912315e-05, "loss": 0.1309, "step": 19454 }, { "epoch": 14.252747252747252, "grad_norm": 0.56640625, "learning_rate": 6.929949629973157e-05, "loss": 0.109, "step": 19455 }, { "epoch": 14.253479853479853, "grad_norm": 0.90234375, "learning_rate": 6.929632661815708e-05, "loss": 0.1328, "step": 19456 }, { "epoch": 14.254212454212453, "grad_norm": 1.6328125, "learning_rate": 6.929315684652323e-05, "loss": 0.1418, "step": 19457 }, { "epoch": 14.254945054945054, "grad_norm": 1.8125, "learning_rate": 6.928998698484515e-05, "loss": 0.096, "step": 19458 }, { "epoch": 14.255677655677655, "grad_norm": 1.171875, "learning_rate": 6.928681703313813e-05, "loss": 0.1349, "step": 19459 }, { "epoch": 14.256410256410255, "grad_norm": 1.1484375, "learning_rate": 6.928364699141729e-05, "loss": 0.1183, "step": 19460 }, { "epoch": 14.257142857142856, "grad_norm": 1.171875, "learning_rate": 6.928047685969781e-05, "loss": 0.1129, "step": 19461 }, { "epoch": 14.257875457875459, "grad_norm": 0.60546875, "learning_rate": 6.927730663799492e-05, "loss": 0.1249, "step": 19462 }, { "epoch": 14.25860805860806, "grad_norm": 0.7421875, "learning_rate": 6.927413632632379e-05, "loss": 0.1372, "step": 19463 }, { "epoch": 14.25934065934066, "grad_norm": 1.2265625, "learning_rate": 6.927096592469958e-05, "loss": 0.1371, "step": 19464 }, { "epoch": 14.260073260073261, "grad_norm": 0.92578125, "learning_rate": 6.926779543313754e-05, "loss": 0.1362, "step": 19465 }, { "epoch": 14.260805860805862, "grad_norm": 1.0859375, "learning_rate": 6.926462485165282e-05, "loss": 0.1263, "step": 19466 }, { "epoch": 14.261538461538462, "grad_norm": 1.0859375, "learning_rate": 6.926145418026062e-05, "loss": 0.1371, "step": 19467 }, { "epoch": 14.262271062271063, "grad_norm": 0.80859375, "learning_rate": 6.925828341897615e-05, "loss": 0.1174, "step": 19468 }, { "epoch": 14.263003663003664, "grad_norm": 1.21875, "learning_rate": 6.925511256781457e-05, "loss": 0.1456, "step": 19469 }, { "epoch": 14.263736263736265, "grad_norm": 1.2578125, "learning_rate": 6.92519416267911e-05, "loss": 0.1245, "step": 19470 }, { "epoch": 14.264468864468865, "grad_norm": 1.2109375, "learning_rate": 6.924877059592091e-05, "loss": 0.1382, "step": 19471 }, { "epoch": 14.265201465201466, "grad_norm": 1.375, "learning_rate": 6.92455994752192e-05, "loss": 0.1355, "step": 19472 }, { "epoch": 14.265934065934067, "grad_norm": 0.78515625, "learning_rate": 6.92424282647012e-05, "loss": 0.1424, "step": 19473 }, { "epoch": 14.266666666666667, "grad_norm": 1.15625, "learning_rate": 6.923925696438204e-05, "loss": 0.1514, "step": 19474 }, { "epoch": 14.267399267399268, "grad_norm": 1.546875, "learning_rate": 6.923608557427697e-05, "loss": 0.1466, "step": 19475 }, { "epoch": 14.268131868131869, "grad_norm": 1.4375, "learning_rate": 6.923291409440114e-05, "loss": 0.1716, "step": 19476 }, { "epoch": 14.26886446886447, "grad_norm": 1.3046875, "learning_rate": 6.92297425247698e-05, "loss": 0.1562, "step": 19477 }, { "epoch": 14.26959706959707, "grad_norm": 1.6640625, "learning_rate": 6.922657086539809e-05, "loss": 0.1356, "step": 19478 }, { "epoch": 14.270329670329671, "grad_norm": 1.234375, "learning_rate": 6.922339911630123e-05, "loss": 0.1355, "step": 19479 }, { "epoch": 14.271062271062272, "grad_norm": 1.03125, "learning_rate": 6.922022727749444e-05, "loss": 0.1243, "step": 19480 }, { "epoch": 14.271794871794873, "grad_norm": 1.453125, "learning_rate": 6.921705534899288e-05, "loss": 0.147, "step": 19481 }, { "epoch": 14.272527472527473, "grad_norm": 0.84765625, "learning_rate": 6.921388333081177e-05, "loss": 0.1295, "step": 19482 }, { "epoch": 14.273260073260074, "grad_norm": 1.7734375, "learning_rate": 6.92107112229663e-05, "loss": 0.1554, "step": 19483 }, { "epoch": 14.273992673992675, "grad_norm": 1.828125, "learning_rate": 6.920753902547167e-05, "loss": 0.1464, "step": 19484 }, { "epoch": 14.274725274725276, "grad_norm": 1.640625, "learning_rate": 6.920436673834307e-05, "loss": 0.1136, "step": 19485 }, { "epoch": 14.275457875457876, "grad_norm": 1.0234375, "learning_rate": 6.920119436159574e-05, "loss": 0.0998, "step": 19486 }, { "epoch": 14.276190476190477, "grad_norm": 1.046875, "learning_rate": 6.919802189524481e-05, "loss": 0.1151, "step": 19487 }, { "epoch": 14.276923076923078, "grad_norm": 1.375, "learning_rate": 6.919484933930554e-05, "loss": 0.1344, "step": 19488 }, { "epoch": 14.277655677655678, "grad_norm": 1.3125, "learning_rate": 6.91916766937931e-05, "loss": 0.156, "step": 19489 }, { "epoch": 14.27838827838828, "grad_norm": 1.21875, "learning_rate": 6.918850395872272e-05, "loss": 0.139, "step": 19490 }, { "epoch": 14.27912087912088, "grad_norm": 1.921875, "learning_rate": 6.918533113410955e-05, "loss": 0.159, "step": 19491 }, { "epoch": 14.27985347985348, "grad_norm": 1.453125, "learning_rate": 6.918215821996885e-05, "loss": 0.1239, "step": 19492 }, { "epoch": 14.280586080586081, "grad_norm": 0.859375, "learning_rate": 6.917898521631578e-05, "loss": 0.1519, "step": 19493 }, { "epoch": 14.281318681318682, "grad_norm": 0.71875, "learning_rate": 6.917581212316558e-05, "loss": 0.1413, "step": 19494 }, { "epoch": 14.282051282051283, "grad_norm": 0.98046875, "learning_rate": 6.917263894053342e-05, "loss": 0.1669, "step": 19495 }, { "epoch": 14.282783882783884, "grad_norm": 0.93359375, "learning_rate": 6.916946566843453e-05, "loss": 0.1219, "step": 19496 }, { "epoch": 14.283516483516484, "grad_norm": 1.0859375, "learning_rate": 6.916629230688408e-05, "loss": 0.1328, "step": 19497 }, { "epoch": 14.284249084249085, "grad_norm": 0.96875, "learning_rate": 6.916311885589731e-05, "loss": 0.1373, "step": 19498 }, { "epoch": 14.284981684981686, "grad_norm": 0.8203125, "learning_rate": 6.915994531548941e-05, "loss": 0.1027, "step": 19499 }, { "epoch": 14.285714285714286, "grad_norm": 1.0390625, "learning_rate": 6.915677168567558e-05, "loss": 0.135, "step": 19500 }, { "epoch": 14.286446886446887, "grad_norm": 0.8828125, "learning_rate": 6.915359796647104e-05, "loss": 0.1247, "step": 19501 }, { "epoch": 14.287179487179488, "grad_norm": 0.9140625, "learning_rate": 6.915042415789098e-05, "loss": 0.1282, "step": 19502 }, { "epoch": 14.287912087912089, "grad_norm": 0.984375, "learning_rate": 6.914725025995064e-05, "loss": 0.133, "step": 19503 }, { "epoch": 14.28864468864469, "grad_norm": 0.765625, "learning_rate": 6.914407627266517e-05, "loss": 0.119, "step": 19504 }, { "epoch": 14.28937728937729, "grad_norm": 1.1328125, "learning_rate": 6.914090219604982e-05, "loss": 0.1226, "step": 19505 }, { "epoch": 14.29010989010989, "grad_norm": 1.0, "learning_rate": 6.913772803011979e-05, "loss": 0.1181, "step": 19506 }, { "epoch": 14.290842490842492, "grad_norm": 1.1328125, "learning_rate": 6.913455377489029e-05, "loss": 0.1466, "step": 19507 }, { "epoch": 14.291575091575092, "grad_norm": 0.61328125, "learning_rate": 6.913137943037652e-05, "loss": 0.1073, "step": 19508 }, { "epoch": 14.292307692307693, "grad_norm": 1.046875, "learning_rate": 6.912820499659371e-05, "loss": 0.1379, "step": 19509 }, { "epoch": 14.293040293040294, "grad_norm": 0.61328125, "learning_rate": 6.912503047355703e-05, "loss": 0.1205, "step": 19510 }, { "epoch": 14.293772893772895, "grad_norm": 0.94140625, "learning_rate": 6.912185586128173e-05, "loss": 0.1158, "step": 19511 }, { "epoch": 14.294505494505495, "grad_norm": 1.109375, "learning_rate": 6.911868115978302e-05, "loss": 0.124, "step": 19512 }, { "epoch": 14.295238095238096, "grad_norm": 1.0078125, "learning_rate": 6.911550636907607e-05, "loss": 0.1177, "step": 19513 }, { "epoch": 14.295970695970697, "grad_norm": 0.86328125, "learning_rate": 6.911233148917611e-05, "loss": 0.1211, "step": 19514 }, { "epoch": 14.296703296703297, "grad_norm": 0.7734375, "learning_rate": 6.91091565200984e-05, "loss": 0.1361, "step": 19515 }, { "epoch": 14.297435897435898, "grad_norm": 0.91015625, "learning_rate": 6.910598146185808e-05, "loss": 0.1316, "step": 19516 }, { "epoch": 14.298168498168499, "grad_norm": 0.71484375, "learning_rate": 6.910280631447041e-05, "loss": 0.105, "step": 19517 }, { "epoch": 14.2989010989011, "grad_norm": 1.1875, "learning_rate": 6.909963107795057e-05, "loss": 0.1645, "step": 19518 }, { "epoch": 14.2996336996337, "grad_norm": 1.03125, "learning_rate": 6.909645575231381e-05, "loss": 0.1176, "step": 19519 }, { "epoch": 14.300366300366301, "grad_norm": 1.1015625, "learning_rate": 6.909328033757532e-05, "loss": 0.1167, "step": 19520 }, { "epoch": 14.301098901098902, "grad_norm": 1.0625, "learning_rate": 6.909010483375031e-05, "loss": 0.146, "step": 19521 }, { "epoch": 14.301831501831503, "grad_norm": 1.203125, "learning_rate": 6.908692924085403e-05, "loss": 0.1239, "step": 19522 }, { "epoch": 14.302564102564103, "grad_norm": 0.9140625, "learning_rate": 6.908375355890166e-05, "loss": 0.1354, "step": 19523 }, { "epoch": 14.303296703296704, "grad_norm": 1.0078125, "learning_rate": 6.90805777879084e-05, "loss": 0.1215, "step": 19524 }, { "epoch": 14.304029304029305, "grad_norm": 1.359375, "learning_rate": 6.907740192788953e-05, "loss": 0.1351, "step": 19525 }, { "epoch": 14.304761904761905, "grad_norm": 1.1953125, "learning_rate": 6.907422597886021e-05, "loss": 0.135, "step": 19526 }, { "epoch": 14.305494505494506, "grad_norm": 1.1796875, "learning_rate": 6.907104994083568e-05, "loss": 0.1378, "step": 19527 }, { "epoch": 14.306227106227107, "grad_norm": 0.95703125, "learning_rate": 6.906787381383114e-05, "loss": 0.1421, "step": 19528 }, { "epoch": 14.306959706959708, "grad_norm": 1.265625, "learning_rate": 6.906469759786184e-05, "loss": 0.0963, "step": 19529 }, { "epoch": 14.307692307692308, "grad_norm": 0.7421875, "learning_rate": 6.906152129294297e-05, "loss": 0.12, "step": 19530 }, { "epoch": 14.308424908424909, "grad_norm": 1.0859375, "learning_rate": 6.905834489908975e-05, "loss": 0.1196, "step": 19531 }, { "epoch": 14.30915750915751, "grad_norm": 0.75, "learning_rate": 6.905516841631743e-05, "loss": 0.1162, "step": 19532 }, { "epoch": 14.30989010989011, "grad_norm": 0.87109375, "learning_rate": 6.905199184464118e-05, "loss": 0.1219, "step": 19533 }, { "epoch": 14.310622710622711, "grad_norm": 0.78515625, "learning_rate": 6.904881518407627e-05, "loss": 0.1349, "step": 19534 }, { "epoch": 14.311355311355312, "grad_norm": 0.97265625, "learning_rate": 6.90456384346379e-05, "loss": 0.1309, "step": 19535 }, { "epoch": 14.312087912087913, "grad_norm": 1.5390625, "learning_rate": 6.904246159634126e-05, "loss": 0.1413, "step": 19536 }, { "epoch": 14.312820512820513, "grad_norm": 1.296875, "learning_rate": 6.903928466920161e-05, "loss": 0.1157, "step": 19537 }, { "epoch": 14.313553113553114, "grad_norm": 0.7265625, "learning_rate": 6.903610765323417e-05, "loss": 0.1032, "step": 19538 }, { "epoch": 14.314285714285715, "grad_norm": 1.546875, "learning_rate": 6.903293054845414e-05, "loss": 0.1431, "step": 19539 }, { "epoch": 14.315018315018316, "grad_norm": 0.734375, "learning_rate": 6.902975335487678e-05, "loss": 0.0918, "step": 19540 }, { "epoch": 14.315750915750916, "grad_norm": 0.7890625, "learning_rate": 6.902657607251727e-05, "loss": 0.1032, "step": 19541 }, { "epoch": 14.316483516483517, "grad_norm": 0.75, "learning_rate": 6.902339870139085e-05, "loss": 0.1259, "step": 19542 }, { "epoch": 14.317216117216118, "grad_norm": 1.640625, "learning_rate": 6.902022124151277e-05, "loss": 0.1206, "step": 19543 }, { "epoch": 14.317948717948719, "grad_norm": 0.83984375, "learning_rate": 6.90170436928982e-05, "loss": 0.0974, "step": 19544 }, { "epoch": 14.31868131868132, "grad_norm": 0.9453125, "learning_rate": 6.90138660555624e-05, "loss": 0.1622, "step": 19545 }, { "epoch": 14.31941391941392, "grad_norm": 0.921875, "learning_rate": 6.901068832952062e-05, "loss": 0.1228, "step": 19546 }, { "epoch": 14.32014652014652, "grad_norm": 1.28125, "learning_rate": 6.900751051478803e-05, "loss": 0.1669, "step": 19547 }, { "epoch": 14.320879120879122, "grad_norm": 0.6953125, "learning_rate": 6.900433261137989e-05, "loss": 0.1008, "step": 19548 }, { "epoch": 14.321611721611722, "grad_norm": 1.09375, "learning_rate": 6.90011546193114e-05, "loss": 0.1414, "step": 19549 }, { "epoch": 14.322344322344323, "grad_norm": 0.8984375, "learning_rate": 6.899797653859782e-05, "loss": 0.1053, "step": 19550 }, { "epoch": 14.323076923076924, "grad_norm": 1.359375, "learning_rate": 6.899479836925437e-05, "loss": 0.1272, "step": 19551 }, { "epoch": 14.323809523809524, "grad_norm": 1.4609375, "learning_rate": 6.899162011129625e-05, "loss": 0.1345, "step": 19552 }, { "epoch": 14.324542124542125, "grad_norm": 1.109375, "learning_rate": 6.898844176473872e-05, "loss": 0.1339, "step": 19553 }, { "epoch": 14.325274725274726, "grad_norm": 1.671875, "learning_rate": 6.898526332959698e-05, "loss": 0.1104, "step": 19554 }, { "epoch": 14.326007326007327, "grad_norm": 1.6953125, "learning_rate": 6.89820848058863e-05, "loss": 0.2187, "step": 19555 }, { "epoch": 14.326739926739927, "grad_norm": 1.265625, "learning_rate": 6.897890619362187e-05, "loss": 0.1028, "step": 19556 }, { "epoch": 14.327472527472528, "grad_norm": 1.9609375, "learning_rate": 6.897572749281893e-05, "loss": 0.142, "step": 19557 }, { "epoch": 14.328205128205129, "grad_norm": 0.88671875, "learning_rate": 6.897254870349273e-05, "loss": 0.11, "step": 19558 }, { "epoch": 14.32893772893773, "grad_norm": 2.171875, "learning_rate": 6.896936982565848e-05, "loss": 0.1464, "step": 19559 }, { "epoch": 14.32967032967033, "grad_norm": 2.390625, "learning_rate": 6.89661908593314e-05, "loss": 0.1414, "step": 19560 }, { "epoch": 14.330402930402931, "grad_norm": 1.1171875, "learning_rate": 6.896301180452675e-05, "loss": 0.1102, "step": 19561 }, { "epoch": 14.331135531135532, "grad_norm": 1.8203125, "learning_rate": 6.895983266125975e-05, "loss": 0.1415, "step": 19562 }, { "epoch": 14.331868131868132, "grad_norm": 0.91015625, "learning_rate": 6.895665342954563e-05, "loss": 0.1219, "step": 19563 }, { "epoch": 14.332600732600733, "grad_norm": 1.4609375, "learning_rate": 6.895347410939962e-05, "loss": 0.1526, "step": 19564 }, { "epoch": 14.333333333333334, "grad_norm": 1.3515625, "learning_rate": 6.895029470083697e-05, "loss": 0.1149, "step": 19565 }, { "epoch": 14.334065934065935, "grad_norm": 0.859375, "learning_rate": 6.894711520387289e-05, "loss": 0.1362, "step": 19566 }, { "epoch": 14.334798534798535, "grad_norm": 1.6640625, "learning_rate": 6.894393561852262e-05, "loss": 0.149, "step": 19567 }, { "epoch": 14.335531135531136, "grad_norm": 1.3203125, "learning_rate": 6.894075594480141e-05, "loss": 0.1102, "step": 19568 }, { "epoch": 14.336263736263737, "grad_norm": 1.09375, "learning_rate": 6.893757618272449e-05, "loss": 0.1081, "step": 19569 }, { "epoch": 14.336996336996338, "grad_norm": 0.90625, "learning_rate": 6.893439633230709e-05, "loss": 0.1327, "step": 19570 }, { "epoch": 14.337728937728938, "grad_norm": 0.69140625, "learning_rate": 6.893121639356445e-05, "loss": 0.1189, "step": 19571 }, { "epoch": 14.338461538461539, "grad_norm": 0.61328125, "learning_rate": 6.892803636651177e-05, "loss": 0.119, "step": 19572 }, { "epoch": 14.33919413919414, "grad_norm": 1.265625, "learning_rate": 6.892485625116434e-05, "loss": 0.1347, "step": 19573 }, { "epoch": 14.33992673992674, "grad_norm": 1.1953125, "learning_rate": 6.892167604753738e-05, "loss": 0.1502, "step": 19574 }, { "epoch": 14.340659340659341, "grad_norm": 1.171875, "learning_rate": 6.891849575564612e-05, "loss": 0.1353, "step": 19575 }, { "epoch": 14.341391941391942, "grad_norm": 1.4453125, "learning_rate": 6.891531537550581e-05, "loss": 0.1096, "step": 19576 }, { "epoch": 14.342124542124543, "grad_norm": 0.89453125, "learning_rate": 6.891213490713167e-05, "loss": 0.1463, "step": 19577 }, { "epoch": 14.342857142857143, "grad_norm": 0.66796875, "learning_rate": 6.890895435053893e-05, "loss": 0.1165, "step": 19578 }, { "epoch": 14.343589743589744, "grad_norm": 0.87109375, "learning_rate": 6.890577370574287e-05, "loss": 0.1365, "step": 19579 }, { "epoch": 14.344322344322345, "grad_norm": 0.64453125, "learning_rate": 6.890259297275869e-05, "loss": 0.1207, "step": 19580 }, { "epoch": 14.345054945054946, "grad_norm": 0.75390625, "learning_rate": 6.889941215160164e-05, "loss": 0.1289, "step": 19581 }, { "epoch": 14.345787545787546, "grad_norm": 0.59375, "learning_rate": 6.889623124228698e-05, "loss": 0.1056, "step": 19582 }, { "epoch": 14.346520146520147, "grad_norm": 0.75390625, "learning_rate": 6.889305024482992e-05, "loss": 0.122, "step": 19583 }, { "epoch": 14.347252747252748, "grad_norm": 0.5390625, "learning_rate": 6.888986915924574e-05, "loss": 0.1139, "step": 19584 }, { "epoch": 14.347985347985349, "grad_norm": 0.89453125, "learning_rate": 6.888668798554966e-05, "loss": 0.1145, "step": 19585 }, { "epoch": 14.34871794871795, "grad_norm": 0.8359375, "learning_rate": 6.88835067237569e-05, "loss": 0.1246, "step": 19586 }, { "epoch": 14.34945054945055, "grad_norm": 0.9140625, "learning_rate": 6.888032537388275e-05, "loss": 0.1423, "step": 19587 }, { "epoch": 14.35018315018315, "grad_norm": 0.86328125, "learning_rate": 6.887714393594243e-05, "loss": 0.1186, "step": 19588 }, { "epoch": 14.350915750915751, "grad_norm": 0.94921875, "learning_rate": 6.887396240995115e-05, "loss": 0.1341, "step": 19589 }, { "epoch": 14.351648351648352, "grad_norm": 1.390625, "learning_rate": 6.887078079592422e-05, "loss": 0.1308, "step": 19590 }, { "epoch": 14.352380952380953, "grad_norm": 1.1328125, "learning_rate": 6.88675990938768e-05, "loss": 0.1503, "step": 19591 }, { "epoch": 14.353113553113554, "grad_norm": 1.0, "learning_rate": 6.886441730382424e-05, "loss": 0.1197, "step": 19592 }, { "epoch": 14.353846153846154, "grad_norm": 1.0390625, "learning_rate": 6.886123542578168e-05, "loss": 0.1026, "step": 19593 }, { "epoch": 14.354578754578755, "grad_norm": 0.5859375, "learning_rate": 6.885805345976444e-05, "loss": 0.1177, "step": 19594 }, { "epoch": 14.355311355311356, "grad_norm": 0.90234375, "learning_rate": 6.885487140578774e-05, "loss": 0.1083, "step": 19595 }, { "epoch": 14.356043956043957, "grad_norm": 1.359375, "learning_rate": 6.885168926386682e-05, "loss": 0.1473, "step": 19596 }, { "epoch": 14.356776556776557, "grad_norm": 1.2109375, "learning_rate": 6.884850703401693e-05, "loss": 0.1259, "step": 19597 }, { "epoch": 14.357509157509158, "grad_norm": 1.046875, "learning_rate": 6.884532471625333e-05, "loss": 0.1361, "step": 19598 }, { "epoch": 14.358241758241759, "grad_norm": 0.9921875, "learning_rate": 6.884214231059124e-05, "loss": 0.1497, "step": 19599 }, { "epoch": 14.35897435897436, "grad_norm": 1.0625, "learning_rate": 6.883895981704593e-05, "loss": 0.1252, "step": 19600 }, { "epoch": 14.35970695970696, "grad_norm": 0.80859375, "learning_rate": 6.883577723563265e-05, "loss": 0.1251, "step": 19601 }, { "epoch": 14.360439560439561, "grad_norm": 1.421875, "learning_rate": 6.883259456636663e-05, "loss": 0.1607, "step": 19602 }, { "epoch": 14.361172161172162, "grad_norm": 1.796875, "learning_rate": 6.882941180926315e-05, "loss": 0.1215, "step": 19603 }, { "epoch": 14.361904761904762, "grad_norm": 1.828125, "learning_rate": 6.882622896433742e-05, "loss": 0.1324, "step": 19604 }, { "epoch": 14.362637362637363, "grad_norm": 0.921875, "learning_rate": 6.882304603160472e-05, "loss": 0.1228, "step": 19605 }, { "epoch": 14.363369963369964, "grad_norm": 1.296875, "learning_rate": 6.88198630110803e-05, "loss": 0.1334, "step": 19606 }, { "epoch": 14.364102564102565, "grad_norm": 1.765625, "learning_rate": 6.881667990277938e-05, "loss": 0.1091, "step": 19607 }, { "epoch": 14.364835164835165, "grad_norm": 1.203125, "learning_rate": 6.881349670671727e-05, "loss": 0.1069, "step": 19608 }, { "epoch": 14.365567765567766, "grad_norm": 0.8203125, "learning_rate": 6.881031342290914e-05, "loss": 0.1211, "step": 19609 }, { "epoch": 14.366300366300367, "grad_norm": 0.734375, "learning_rate": 6.880713005137033e-05, "loss": 0.1184, "step": 19610 }, { "epoch": 14.367032967032968, "grad_norm": 0.96484375, "learning_rate": 6.880394659211604e-05, "loss": 0.1057, "step": 19611 }, { "epoch": 14.367765567765568, "grad_norm": 1.0234375, "learning_rate": 6.880076304516151e-05, "loss": 0.1203, "step": 19612 }, { "epoch": 14.368498168498169, "grad_norm": 0.765625, "learning_rate": 6.879757941052204e-05, "loss": 0.1212, "step": 19613 }, { "epoch": 14.36923076923077, "grad_norm": 1.3359375, "learning_rate": 6.879439568821286e-05, "loss": 0.108, "step": 19614 }, { "epoch": 14.36996336996337, "grad_norm": 1.2734375, "learning_rate": 6.879121187824921e-05, "loss": 0.1388, "step": 19615 }, { "epoch": 14.370695970695971, "grad_norm": 1.40625, "learning_rate": 6.878802798064636e-05, "loss": 0.1412, "step": 19616 }, { "epoch": 14.371428571428572, "grad_norm": 1.09375, "learning_rate": 6.878484399541956e-05, "loss": 0.1352, "step": 19617 }, { "epoch": 14.372161172161173, "grad_norm": 0.97265625, "learning_rate": 6.87816599225841e-05, "loss": 0.1095, "step": 19618 }, { "epoch": 14.372893772893773, "grad_norm": 1.59375, "learning_rate": 6.877847576215517e-05, "loss": 0.1517, "step": 19619 }, { "epoch": 14.373626373626374, "grad_norm": 2.40625, "learning_rate": 6.877529151414807e-05, "loss": 0.1382, "step": 19620 }, { "epoch": 14.374358974358975, "grad_norm": 1.796875, "learning_rate": 6.877210717857806e-05, "loss": 0.1494, "step": 19621 }, { "epoch": 14.375091575091576, "grad_norm": 1.0859375, "learning_rate": 6.876892275546038e-05, "loss": 0.12, "step": 19622 }, { "epoch": 14.375824175824176, "grad_norm": 1.609375, "learning_rate": 6.876573824481029e-05, "loss": 0.1577, "step": 19623 }, { "epoch": 14.376556776556777, "grad_norm": 1.25, "learning_rate": 6.876255364664307e-05, "loss": 0.1339, "step": 19624 }, { "epoch": 14.377289377289378, "grad_norm": 0.89453125, "learning_rate": 6.875936896097395e-05, "loss": 0.1316, "step": 19625 }, { "epoch": 14.378021978021978, "grad_norm": 1.3671875, "learning_rate": 6.87561841878182e-05, "loss": 0.13, "step": 19626 }, { "epoch": 14.37875457875458, "grad_norm": 0.91796875, "learning_rate": 6.875299932719107e-05, "loss": 0.1194, "step": 19627 }, { "epoch": 14.37948717948718, "grad_norm": 1.0078125, "learning_rate": 6.874981437910784e-05, "loss": 0.1314, "step": 19628 }, { "epoch": 14.38021978021978, "grad_norm": 1.3359375, "learning_rate": 6.874662934358375e-05, "loss": 0.1143, "step": 19629 }, { "epoch": 14.380952380952381, "grad_norm": 1.15625, "learning_rate": 6.874344422063406e-05, "loss": 0.1103, "step": 19630 }, { "epoch": 14.381684981684982, "grad_norm": 0.921875, "learning_rate": 6.874025901027405e-05, "loss": 0.1311, "step": 19631 }, { "epoch": 14.382417582417583, "grad_norm": 0.76953125, "learning_rate": 6.873707371251899e-05, "loss": 0.128, "step": 19632 }, { "epoch": 14.383150183150184, "grad_norm": 1.71875, "learning_rate": 6.87338883273841e-05, "loss": 0.1584, "step": 19633 }, { "epoch": 14.383882783882784, "grad_norm": 1.015625, "learning_rate": 6.873070285488467e-05, "loss": 0.1106, "step": 19634 }, { "epoch": 14.384615384615385, "grad_norm": 1.09375, "learning_rate": 6.872751729503594e-05, "loss": 0.1095, "step": 19635 }, { "epoch": 14.385347985347986, "grad_norm": 0.609375, "learning_rate": 6.872433164785322e-05, "loss": 0.1008, "step": 19636 }, { "epoch": 14.386080586080586, "grad_norm": 0.953125, "learning_rate": 6.872114591335174e-05, "loss": 0.1154, "step": 19637 }, { "epoch": 14.386813186813187, "grad_norm": 1.109375, "learning_rate": 6.871796009154676e-05, "loss": 0.1132, "step": 19638 }, { "epoch": 14.387545787545788, "grad_norm": 1.3671875, "learning_rate": 6.871477418245357e-05, "loss": 0.1153, "step": 19639 }, { "epoch": 14.388278388278389, "grad_norm": 0.890625, "learning_rate": 6.87115881860874e-05, "loss": 0.117, "step": 19640 }, { "epoch": 14.38901098901099, "grad_norm": 1.4765625, "learning_rate": 6.870840210246352e-05, "loss": 0.1053, "step": 19641 }, { "epoch": 14.38974358974359, "grad_norm": 1.5625, "learning_rate": 6.870521593159723e-05, "loss": 0.1303, "step": 19642 }, { "epoch": 14.39047619047619, "grad_norm": 1.1171875, "learning_rate": 6.870202967350376e-05, "loss": 0.1496, "step": 19643 }, { "epoch": 14.391208791208792, "grad_norm": 1.296875, "learning_rate": 6.869884332819841e-05, "loss": 0.1286, "step": 19644 }, { "epoch": 14.391941391941392, "grad_norm": 0.80078125, "learning_rate": 6.86956568956964e-05, "loss": 0.1186, "step": 19645 }, { "epoch": 14.392673992673993, "grad_norm": 1.3046875, "learning_rate": 6.869247037601304e-05, "loss": 0.1101, "step": 19646 }, { "epoch": 14.393406593406594, "grad_norm": 1.0546875, "learning_rate": 6.868928376916358e-05, "loss": 0.1198, "step": 19647 }, { "epoch": 14.394139194139195, "grad_norm": 1.9765625, "learning_rate": 6.868609707516328e-05, "loss": 0.1445, "step": 19648 }, { "epoch": 14.394871794871795, "grad_norm": 0.9765625, "learning_rate": 6.868291029402743e-05, "loss": 0.1401, "step": 19649 }, { "epoch": 14.395604395604396, "grad_norm": 1.34375, "learning_rate": 6.86797234257713e-05, "loss": 0.1072, "step": 19650 }, { "epoch": 14.396336996336997, "grad_norm": 1.3515625, "learning_rate": 6.867653647041013e-05, "loss": 0.1224, "step": 19651 }, { "epoch": 14.397069597069597, "grad_norm": 1.3828125, "learning_rate": 6.867334942795924e-05, "loss": 0.1572, "step": 19652 }, { "epoch": 14.397802197802198, "grad_norm": 1.1953125, "learning_rate": 6.867016229843383e-05, "loss": 0.1116, "step": 19653 }, { "epoch": 14.398534798534799, "grad_norm": 1.359375, "learning_rate": 6.866697508184922e-05, "loss": 0.1228, "step": 19654 }, { "epoch": 14.3992673992674, "grad_norm": 0.92578125, "learning_rate": 6.866378777822066e-05, "loss": 0.1251, "step": 19655 }, { "epoch": 14.4, "grad_norm": 0.86328125, "learning_rate": 6.866060038756343e-05, "loss": 0.0969, "step": 19656 }, { "epoch": 14.400732600732601, "grad_norm": 0.94921875, "learning_rate": 6.86574129098928e-05, "loss": 0.1349, "step": 19657 }, { "epoch": 14.401465201465202, "grad_norm": 1.1953125, "learning_rate": 6.865422534522406e-05, "loss": 0.1545, "step": 19658 }, { "epoch": 14.402197802197803, "grad_norm": 1.1640625, "learning_rate": 6.865103769357246e-05, "loss": 0.1247, "step": 19659 }, { "epoch": 14.402930402930403, "grad_norm": 1.328125, "learning_rate": 6.864784995495327e-05, "loss": 0.1461, "step": 19660 }, { "epoch": 14.403663003663004, "grad_norm": 0.5859375, "learning_rate": 6.864466212938178e-05, "loss": 0.1018, "step": 19661 }, { "epoch": 14.404395604395605, "grad_norm": 1.296875, "learning_rate": 6.864147421687326e-05, "loss": 0.1628, "step": 19662 }, { "epoch": 14.405128205128205, "grad_norm": 1.171875, "learning_rate": 6.863828621744298e-05, "loss": 0.1369, "step": 19663 }, { "epoch": 14.405860805860806, "grad_norm": 1.296875, "learning_rate": 6.863509813110621e-05, "loss": 0.0901, "step": 19664 }, { "epoch": 14.406593406593407, "grad_norm": 1.0546875, "learning_rate": 6.863190995787825e-05, "loss": 0.1153, "step": 19665 }, { "epoch": 14.407326007326008, "grad_norm": 1.1328125, "learning_rate": 6.862872169777433e-05, "loss": 0.0999, "step": 19666 }, { "epoch": 14.408058608058608, "grad_norm": 0.7734375, "learning_rate": 6.862553335080978e-05, "loss": 0.1098, "step": 19667 }, { "epoch": 14.40879120879121, "grad_norm": 0.87890625, "learning_rate": 6.862234491699985e-05, "loss": 0.1265, "step": 19668 }, { "epoch": 14.40952380952381, "grad_norm": 1.125, "learning_rate": 6.861915639635979e-05, "loss": 0.1329, "step": 19669 }, { "epoch": 14.41025641025641, "grad_norm": 1.3359375, "learning_rate": 6.86159677889049e-05, "loss": 0.1347, "step": 19670 }, { "epoch": 14.410989010989011, "grad_norm": 0.6796875, "learning_rate": 6.861277909465049e-05, "loss": 0.1084, "step": 19671 }, { "epoch": 14.411721611721612, "grad_norm": 1.5, "learning_rate": 6.860959031361179e-05, "loss": 0.1778, "step": 19672 }, { "epoch": 14.412454212454213, "grad_norm": 0.73046875, "learning_rate": 6.860640144580412e-05, "loss": 0.1179, "step": 19673 }, { "epoch": 14.413186813186813, "grad_norm": 1.1796875, "learning_rate": 6.860321249124271e-05, "loss": 0.1258, "step": 19674 }, { "epoch": 14.413919413919414, "grad_norm": 0.86328125, "learning_rate": 6.860002344994289e-05, "loss": 0.1274, "step": 19675 }, { "epoch": 14.414652014652015, "grad_norm": 0.74609375, "learning_rate": 6.85968343219199e-05, "loss": 0.1463, "step": 19676 }, { "epoch": 14.415384615384616, "grad_norm": 1.3828125, "learning_rate": 6.859364510718903e-05, "loss": 0.1043, "step": 19677 }, { "epoch": 14.416117216117216, "grad_norm": 2.046875, "learning_rate": 6.859045580576559e-05, "loss": 0.1632, "step": 19678 }, { "epoch": 14.416849816849817, "grad_norm": 0.7421875, "learning_rate": 6.858726641766482e-05, "loss": 0.1226, "step": 19679 }, { "epoch": 14.417582417582418, "grad_norm": 1.09375, "learning_rate": 6.858407694290202e-05, "loss": 0.124, "step": 19680 }, { "epoch": 14.418315018315019, "grad_norm": 0.60546875, "learning_rate": 6.858088738149249e-05, "loss": 0.1026, "step": 19681 }, { "epoch": 14.41904761904762, "grad_norm": 0.80078125, "learning_rate": 6.857769773345146e-05, "loss": 0.1032, "step": 19682 }, { "epoch": 14.41978021978022, "grad_norm": 1.0078125, "learning_rate": 6.857450799879426e-05, "loss": 0.1049, "step": 19683 }, { "epoch": 14.42051282051282, "grad_norm": 1.015625, "learning_rate": 6.857131817753616e-05, "loss": 0.1539, "step": 19684 }, { "epoch": 14.421245421245422, "grad_norm": 0.69921875, "learning_rate": 6.856812826969245e-05, "loss": 0.1471, "step": 19685 }, { "epoch": 14.421978021978022, "grad_norm": 1.1015625, "learning_rate": 6.856493827527838e-05, "loss": 0.117, "step": 19686 }, { "epoch": 14.422710622710623, "grad_norm": 1.6796875, "learning_rate": 6.856174819430928e-05, "loss": 0.1507, "step": 19687 }, { "epoch": 14.423443223443224, "grad_norm": 1.21875, "learning_rate": 6.855855802680043e-05, "loss": 0.1221, "step": 19688 }, { "epoch": 14.424175824175824, "grad_norm": 0.8203125, "learning_rate": 6.855536777276708e-05, "loss": 0.1132, "step": 19689 }, { "epoch": 14.424908424908425, "grad_norm": 0.71484375, "learning_rate": 6.855217743222454e-05, "loss": 0.1284, "step": 19690 }, { "epoch": 14.425641025641026, "grad_norm": 1.0546875, "learning_rate": 6.85489870051881e-05, "loss": 0.1379, "step": 19691 }, { "epoch": 14.426373626373627, "grad_norm": 0.9609375, "learning_rate": 6.854579649167303e-05, "loss": 0.1547, "step": 19692 }, { "epoch": 14.427106227106227, "grad_norm": 0.82421875, "learning_rate": 6.854260589169463e-05, "loss": 0.1144, "step": 19693 }, { "epoch": 14.427838827838828, "grad_norm": 0.73046875, "learning_rate": 6.853941520526817e-05, "loss": 0.1093, "step": 19694 }, { "epoch": 14.428571428571429, "grad_norm": 1.015625, "learning_rate": 6.853622443240896e-05, "loss": 0.1246, "step": 19695 }, { "epoch": 14.42930402930403, "grad_norm": 1.1328125, "learning_rate": 6.853303357313227e-05, "loss": 0.1097, "step": 19696 }, { "epoch": 14.43003663003663, "grad_norm": 1.0390625, "learning_rate": 6.852984262745342e-05, "loss": 0.1179, "step": 19697 }, { "epoch": 14.430769230769231, "grad_norm": 0.8515625, "learning_rate": 6.852665159538763e-05, "loss": 0.1151, "step": 19698 }, { "epoch": 14.431501831501832, "grad_norm": 0.78125, "learning_rate": 6.852346047695028e-05, "loss": 0.0967, "step": 19699 }, { "epoch": 14.432234432234432, "grad_norm": 1.4609375, "learning_rate": 6.852026927215659e-05, "loss": 0.1705, "step": 19700 }, { "epoch": 14.432967032967033, "grad_norm": 0.96875, "learning_rate": 6.851707798102188e-05, "loss": 0.1076, "step": 19701 }, { "epoch": 14.433699633699634, "grad_norm": 0.79296875, "learning_rate": 6.851388660356144e-05, "loss": 0.1235, "step": 19702 }, { "epoch": 14.434432234432235, "grad_norm": 1.3984375, "learning_rate": 6.851069513979054e-05, "loss": 0.1182, "step": 19703 }, { "epoch": 14.435164835164835, "grad_norm": 1.671875, "learning_rate": 6.850750358972451e-05, "loss": 0.1525, "step": 19704 }, { "epoch": 14.435897435897436, "grad_norm": 1.140625, "learning_rate": 6.850431195337861e-05, "loss": 0.098, "step": 19705 }, { "epoch": 14.436630036630037, "grad_norm": 0.90234375, "learning_rate": 6.850112023076813e-05, "loss": 0.1182, "step": 19706 }, { "epoch": 14.437362637362638, "grad_norm": 0.93359375, "learning_rate": 6.849792842190838e-05, "loss": 0.1363, "step": 19707 }, { "epoch": 14.438095238095238, "grad_norm": 1.1171875, "learning_rate": 6.849473652681463e-05, "loss": 0.1437, "step": 19708 }, { "epoch": 14.438827838827839, "grad_norm": 1.53125, "learning_rate": 6.849154454550222e-05, "loss": 0.1542, "step": 19709 }, { "epoch": 14.43956043956044, "grad_norm": 0.99609375, "learning_rate": 6.84883524779864e-05, "loss": 0.1236, "step": 19710 }, { "epoch": 14.44029304029304, "grad_norm": 0.875, "learning_rate": 6.848516032428249e-05, "loss": 0.112, "step": 19711 }, { "epoch": 14.441025641025641, "grad_norm": 1.1796875, "learning_rate": 6.848196808440575e-05, "loss": 0.1439, "step": 19712 }, { "epoch": 14.441758241758242, "grad_norm": 1.1640625, "learning_rate": 6.84787757583715e-05, "loss": 0.1253, "step": 19713 }, { "epoch": 14.442490842490843, "grad_norm": 0.94921875, "learning_rate": 6.847558334619506e-05, "loss": 0.1312, "step": 19714 }, { "epoch": 14.443223443223443, "grad_norm": 1.734375, "learning_rate": 6.847239084789168e-05, "loss": 0.1373, "step": 19715 }, { "epoch": 14.443956043956044, "grad_norm": 1.0625, "learning_rate": 6.846919826347669e-05, "loss": 0.1122, "step": 19716 }, { "epoch": 14.444688644688645, "grad_norm": 0.80859375, "learning_rate": 6.846600559296536e-05, "loss": 0.1179, "step": 19717 }, { "epoch": 14.445421245421246, "grad_norm": 0.796875, "learning_rate": 6.846281283637301e-05, "loss": 0.1242, "step": 19718 }, { "epoch": 14.446153846153846, "grad_norm": 1.328125, "learning_rate": 6.84596199937149e-05, "loss": 0.154, "step": 19719 }, { "epoch": 14.446886446886447, "grad_norm": 0.9453125, "learning_rate": 6.845642706500636e-05, "loss": 0.113, "step": 19720 }, { "epoch": 14.447619047619048, "grad_norm": 0.88671875, "learning_rate": 6.845323405026271e-05, "loss": 0.127, "step": 19721 }, { "epoch": 14.448351648351649, "grad_norm": 1.046875, "learning_rate": 6.84500409494992e-05, "loss": 0.1576, "step": 19722 }, { "epoch": 14.44908424908425, "grad_norm": 0.71875, "learning_rate": 6.844684776273115e-05, "loss": 0.1062, "step": 19723 }, { "epoch": 14.44981684981685, "grad_norm": 1.5625, "learning_rate": 6.844365448997388e-05, "loss": 0.1396, "step": 19724 }, { "epoch": 14.45054945054945, "grad_norm": 1.6171875, "learning_rate": 6.844046113124265e-05, "loss": 0.1638, "step": 19725 }, { "epoch": 14.451282051282051, "grad_norm": 1.046875, "learning_rate": 6.843726768655278e-05, "loss": 0.1134, "step": 19726 }, { "epoch": 14.452014652014652, "grad_norm": 0.60546875, "learning_rate": 6.84340741559196e-05, "loss": 0.1251, "step": 19727 }, { "epoch": 14.452747252747253, "grad_norm": 0.8203125, "learning_rate": 6.843088053935837e-05, "loss": 0.1033, "step": 19728 }, { "epoch": 14.453479853479854, "grad_norm": 0.6640625, "learning_rate": 6.842768683688439e-05, "loss": 0.1084, "step": 19729 }, { "epoch": 14.454212454212454, "grad_norm": 1.03125, "learning_rate": 6.842449304851301e-05, "loss": 0.1181, "step": 19730 }, { "epoch": 14.454945054945055, "grad_norm": 1.671875, "learning_rate": 6.842129917425945e-05, "loss": 0.1293, "step": 19731 }, { "epoch": 14.455677655677656, "grad_norm": 1.125, "learning_rate": 6.84181052141391e-05, "loss": 0.1419, "step": 19732 }, { "epoch": 14.456410256410257, "grad_norm": 1.6484375, "learning_rate": 6.841491116816723e-05, "loss": 0.1624, "step": 19733 }, { "epoch": 14.457142857142857, "grad_norm": 1.734375, "learning_rate": 6.841171703635911e-05, "loss": 0.1523, "step": 19734 }, { "epoch": 14.457875457875458, "grad_norm": 0.7421875, "learning_rate": 6.84085228187301e-05, "loss": 0.1076, "step": 19735 }, { "epoch": 14.458608058608059, "grad_norm": 1.421875, "learning_rate": 6.840532851529546e-05, "loss": 0.1359, "step": 19736 }, { "epoch": 14.45934065934066, "grad_norm": 1.5859375, "learning_rate": 6.84021341260705e-05, "loss": 0.1373, "step": 19737 }, { "epoch": 14.46007326007326, "grad_norm": 1.1875, "learning_rate": 6.839893965107057e-05, "loss": 0.1223, "step": 19738 }, { "epoch": 14.460805860805861, "grad_norm": 1.6875, "learning_rate": 6.839574509031094e-05, "loss": 0.1126, "step": 19739 }, { "epoch": 14.461538461538462, "grad_norm": 1.34375, "learning_rate": 6.839255044380691e-05, "loss": 0.1442, "step": 19740 }, { "epoch": 14.462271062271062, "grad_norm": 1.2265625, "learning_rate": 6.83893557115738e-05, "loss": 0.1411, "step": 19741 }, { "epoch": 14.463003663003663, "grad_norm": 1.328125, "learning_rate": 6.838616089362692e-05, "loss": 0.1362, "step": 19742 }, { "epoch": 14.463736263736264, "grad_norm": 1.2421875, "learning_rate": 6.838296598998158e-05, "loss": 0.1407, "step": 19743 }, { "epoch": 14.464468864468865, "grad_norm": 2.0625, "learning_rate": 6.837977100065306e-05, "loss": 0.1566, "step": 19744 }, { "epoch": 14.465201465201465, "grad_norm": 1.2734375, "learning_rate": 6.837657592565669e-05, "loss": 0.1441, "step": 19745 }, { "epoch": 14.465934065934066, "grad_norm": 1.875, "learning_rate": 6.837338076500777e-05, "loss": 0.1156, "step": 19746 }, { "epoch": 14.466666666666667, "grad_norm": 1.84375, "learning_rate": 6.837018551872163e-05, "loss": 0.1386, "step": 19747 }, { "epoch": 14.467399267399268, "grad_norm": 1.7890625, "learning_rate": 6.836699018681357e-05, "loss": 0.123, "step": 19748 }, { "epoch": 14.468131868131868, "grad_norm": 1.015625, "learning_rate": 6.836379476929887e-05, "loss": 0.1229, "step": 19749 }, { "epoch": 14.468864468864469, "grad_norm": 1.71875, "learning_rate": 6.836059926619288e-05, "loss": 0.1191, "step": 19750 }, { "epoch": 14.46959706959707, "grad_norm": 1.59375, "learning_rate": 6.83574036775109e-05, "loss": 0.1273, "step": 19751 }, { "epoch": 14.47032967032967, "grad_norm": 1.03125, "learning_rate": 6.835420800326823e-05, "loss": 0.1074, "step": 19752 }, { "epoch": 14.471062271062271, "grad_norm": 1.0859375, "learning_rate": 6.835101224348019e-05, "loss": 0.1657, "step": 19753 }, { "epoch": 14.471794871794872, "grad_norm": 1.2890625, "learning_rate": 6.83478163981621e-05, "loss": 0.0998, "step": 19754 }, { "epoch": 14.472527472527473, "grad_norm": 0.83984375, "learning_rate": 6.834462046732924e-05, "loss": 0.1174, "step": 19755 }, { "epoch": 14.473260073260073, "grad_norm": 0.83203125, "learning_rate": 6.834142445099696e-05, "loss": 0.1225, "step": 19756 }, { "epoch": 14.473992673992674, "grad_norm": 0.6953125, "learning_rate": 6.833822834918056e-05, "loss": 0.1226, "step": 19757 }, { "epoch": 14.474725274725275, "grad_norm": 1.140625, "learning_rate": 6.833503216189534e-05, "loss": 0.1139, "step": 19758 }, { "epoch": 14.475457875457876, "grad_norm": 1.171875, "learning_rate": 6.833183588915665e-05, "loss": 0.1261, "step": 19759 }, { "epoch": 14.476190476190476, "grad_norm": 0.94140625, "learning_rate": 6.832863953097977e-05, "loss": 0.1166, "step": 19760 }, { "epoch": 14.476923076923077, "grad_norm": 1.921875, "learning_rate": 6.832544308737999e-05, "loss": 0.1319, "step": 19761 }, { "epoch": 14.477655677655678, "grad_norm": 1.5234375, "learning_rate": 6.83222465583727e-05, "loss": 0.1239, "step": 19762 }, { "epoch": 14.478388278388278, "grad_norm": 1.0390625, "learning_rate": 6.831904994397317e-05, "loss": 0.0855, "step": 19763 }, { "epoch": 14.47912087912088, "grad_norm": 0.69140625, "learning_rate": 6.831585324419672e-05, "loss": 0.1075, "step": 19764 }, { "epoch": 14.47985347985348, "grad_norm": 0.8828125, "learning_rate": 6.831265645905867e-05, "loss": 0.1142, "step": 19765 }, { "epoch": 14.48058608058608, "grad_norm": 1.109375, "learning_rate": 6.830945958857434e-05, "loss": 0.1258, "step": 19766 }, { "epoch": 14.481318681318681, "grad_norm": 1.015625, "learning_rate": 6.830626263275904e-05, "loss": 0.1321, "step": 19767 }, { "epoch": 14.482051282051282, "grad_norm": 0.9375, "learning_rate": 6.830306559162808e-05, "loss": 0.1219, "step": 19768 }, { "epoch": 14.482783882783883, "grad_norm": 1.140625, "learning_rate": 6.82998684651968e-05, "loss": 0.1267, "step": 19769 }, { "epoch": 14.483516483516484, "grad_norm": 1.2578125, "learning_rate": 6.82966712534805e-05, "loss": 0.1396, "step": 19770 }, { "epoch": 14.484249084249084, "grad_norm": 0.72265625, "learning_rate": 6.829347395649449e-05, "loss": 0.1212, "step": 19771 }, { "epoch": 14.484981684981685, "grad_norm": 0.94140625, "learning_rate": 6.829027657425414e-05, "loss": 0.0896, "step": 19772 }, { "epoch": 14.485714285714286, "grad_norm": 1.0546875, "learning_rate": 6.828707910677471e-05, "loss": 0.149, "step": 19773 }, { "epoch": 14.486446886446886, "grad_norm": 1.109375, "learning_rate": 6.828388155407155e-05, "loss": 0.1426, "step": 19774 }, { "epoch": 14.487179487179487, "grad_norm": 1.171875, "learning_rate": 6.828068391615997e-05, "loss": 0.1208, "step": 19775 }, { "epoch": 14.487912087912088, "grad_norm": 0.671875, "learning_rate": 6.827748619305531e-05, "loss": 0.1186, "step": 19776 }, { "epoch": 14.488644688644689, "grad_norm": 0.72265625, "learning_rate": 6.827428838477288e-05, "loss": 0.1405, "step": 19777 }, { "epoch": 14.48937728937729, "grad_norm": 0.8203125, "learning_rate": 6.827109049132797e-05, "loss": 0.13, "step": 19778 }, { "epoch": 14.49010989010989, "grad_norm": 1.578125, "learning_rate": 6.826789251273596e-05, "loss": 0.139, "step": 19779 }, { "epoch": 14.49084249084249, "grad_norm": 1.578125, "learning_rate": 6.826469444901214e-05, "loss": 0.1509, "step": 19780 }, { "epoch": 14.491575091575092, "grad_norm": 1.0546875, "learning_rate": 6.826149630017182e-05, "loss": 0.1194, "step": 19781 }, { "epoch": 14.492307692307692, "grad_norm": 1.6015625, "learning_rate": 6.825829806623037e-05, "loss": 0.1399, "step": 19782 }, { "epoch": 14.493040293040293, "grad_norm": 0.6875, "learning_rate": 6.825509974720308e-05, "loss": 0.1007, "step": 19783 }, { "epoch": 14.493772893772894, "grad_norm": 0.921875, "learning_rate": 6.825190134310525e-05, "loss": 0.1312, "step": 19784 }, { "epoch": 14.494505494505495, "grad_norm": 1.0546875, "learning_rate": 6.824870285395225e-05, "loss": 0.1109, "step": 19785 }, { "epoch": 14.495238095238095, "grad_norm": 0.65625, "learning_rate": 6.82455042797594e-05, "loss": 0.1234, "step": 19786 }, { "epoch": 14.495970695970696, "grad_norm": 1.234375, "learning_rate": 6.824230562054199e-05, "loss": 0.1174, "step": 19787 }, { "epoch": 14.496703296703297, "grad_norm": 0.72265625, "learning_rate": 6.823910687631538e-05, "loss": 0.1065, "step": 19788 }, { "epoch": 14.497435897435897, "grad_norm": 1.2578125, "learning_rate": 6.823590804709489e-05, "loss": 0.1256, "step": 19789 }, { "epoch": 14.498168498168498, "grad_norm": 0.88671875, "learning_rate": 6.823270913289583e-05, "loss": 0.1357, "step": 19790 }, { "epoch": 14.498901098901099, "grad_norm": 1.125, "learning_rate": 6.822951013373357e-05, "loss": 0.103, "step": 19791 }, { "epoch": 14.4996336996337, "grad_norm": 1.3125, "learning_rate": 6.822631104962338e-05, "loss": 0.1594, "step": 19792 }, { "epoch": 14.5003663003663, "grad_norm": 0.98046875, "learning_rate": 6.822311188058062e-05, "loss": 0.1355, "step": 19793 }, { "epoch": 14.501098901098901, "grad_norm": 1.5078125, "learning_rate": 6.82199126266206e-05, "loss": 0.1205, "step": 19794 }, { "epoch": 14.501831501831502, "grad_norm": 1.03125, "learning_rate": 6.821671328775868e-05, "loss": 0.1221, "step": 19795 }, { "epoch": 14.502564102564103, "grad_norm": 1.5078125, "learning_rate": 6.821351386401017e-05, "loss": 0.1698, "step": 19796 }, { "epoch": 14.503296703296703, "grad_norm": 1.734375, "learning_rate": 6.821031435539038e-05, "loss": 0.1331, "step": 19797 }, { "epoch": 14.504029304029304, "grad_norm": 0.734375, "learning_rate": 6.820711476191469e-05, "loss": 0.1096, "step": 19798 }, { "epoch": 14.504761904761905, "grad_norm": 0.89453125, "learning_rate": 6.820391508359838e-05, "loss": 0.1172, "step": 19799 }, { "epoch": 14.505494505494505, "grad_norm": 1.21875, "learning_rate": 6.82007153204568e-05, "loss": 0.1394, "step": 19800 }, { "epoch": 14.506227106227106, "grad_norm": 1.3671875, "learning_rate": 6.81975154725053e-05, "loss": 0.1363, "step": 19801 }, { "epoch": 14.506959706959707, "grad_norm": 1.109375, "learning_rate": 6.819431553975916e-05, "loss": 0.1458, "step": 19802 }, { "epoch": 14.507692307692308, "grad_norm": 1.8515625, "learning_rate": 6.819111552223379e-05, "loss": 0.1408, "step": 19803 }, { "epoch": 14.508424908424908, "grad_norm": 0.60546875, "learning_rate": 6.818791541994445e-05, "loss": 0.101, "step": 19804 }, { "epoch": 14.50915750915751, "grad_norm": 1.3359375, "learning_rate": 6.818471523290651e-05, "loss": 0.1438, "step": 19805 }, { "epoch": 14.50989010989011, "grad_norm": 1.09375, "learning_rate": 6.818151496113529e-05, "loss": 0.1115, "step": 19806 }, { "epoch": 14.51062271062271, "grad_norm": 0.7109375, "learning_rate": 6.817831460464611e-05, "loss": 0.128, "step": 19807 }, { "epoch": 14.511355311355311, "grad_norm": 1.46875, "learning_rate": 6.817511416345434e-05, "loss": 0.1012, "step": 19808 }, { "epoch": 14.512087912087912, "grad_norm": 0.828125, "learning_rate": 6.81719136375753e-05, "loss": 0.1162, "step": 19809 }, { "epoch": 14.512820512820513, "grad_norm": 1.2421875, "learning_rate": 6.816871302702431e-05, "loss": 0.1194, "step": 19810 }, { "epoch": 14.513553113553114, "grad_norm": 1.21875, "learning_rate": 6.816551233181672e-05, "loss": 0.1026, "step": 19811 }, { "epoch": 14.514285714285714, "grad_norm": 0.8203125, "learning_rate": 6.816231155196787e-05, "loss": 0.1108, "step": 19812 }, { "epoch": 14.515018315018315, "grad_norm": 1.0546875, "learning_rate": 6.815911068749306e-05, "loss": 0.1348, "step": 19813 }, { "epoch": 14.515750915750916, "grad_norm": 0.77734375, "learning_rate": 6.815590973840768e-05, "loss": 0.1248, "step": 19814 }, { "epoch": 14.516483516483516, "grad_norm": 1.2734375, "learning_rate": 6.815270870472703e-05, "loss": 0.1261, "step": 19815 }, { "epoch": 14.517216117216117, "grad_norm": 1.25, "learning_rate": 6.814950758646647e-05, "loss": 0.1211, "step": 19816 }, { "epoch": 14.517948717948718, "grad_norm": 0.953125, "learning_rate": 6.814630638364133e-05, "loss": 0.0913, "step": 19817 }, { "epoch": 14.518681318681319, "grad_norm": 1.25, "learning_rate": 6.814310509626691e-05, "loss": 0.1149, "step": 19818 }, { "epoch": 14.51941391941392, "grad_norm": 0.7421875, "learning_rate": 6.81399037243586e-05, "loss": 0.1169, "step": 19819 }, { "epoch": 14.52014652014652, "grad_norm": 1.421875, "learning_rate": 6.813670226793173e-05, "loss": 0.144, "step": 19820 }, { "epoch": 14.52087912087912, "grad_norm": 1.5390625, "learning_rate": 6.813350072700162e-05, "loss": 0.1393, "step": 19821 }, { "epoch": 14.521611721611722, "grad_norm": 1.4296875, "learning_rate": 6.813029910158362e-05, "loss": 0.1479, "step": 19822 }, { "epoch": 14.522344322344322, "grad_norm": 1.3828125, "learning_rate": 6.812709739169307e-05, "loss": 0.1467, "step": 19823 }, { "epoch": 14.523076923076923, "grad_norm": 1.7890625, "learning_rate": 6.812389559734529e-05, "loss": 0.1686, "step": 19824 }, { "epoch": 14.523809523809524, "grad_norm": 1.0859375, "learning_rate": 6.812069371855567e-05, "loss": 0.1239, "step": 19825 }, { "epoch": 14.524542124542124, "grad_norm": 1.0, "learning_rate": 6.81174917553395e-05, "loss": 0.1346, "step": 19826 }, { "epoch": 14.525274725274725, "grad_norm": 0.9375, "learning_rate": 6.811428970771217e-05, "loss": 0.1195, "step": 19827 }, { "epoch": 14.526007326007326, "grad_norm": 0.83984375, "learning_rate": 6.811108757568895e-05, "loss": 0.1101, "step": 19828 }, { "epoch": 14.526739926739927, "grad_norm": 1.6015625, "learning_rate": 6.810788535928528e-05, "loss": 0.1103, "step": 19829 }, { "epoch": 14.527472527472527, "grad_norm": 1.265625, "learning_rate": 6.810468305851644e-05, "loss": 0.1085, "step": 19830 }, { "epoch": 14.528205128205128, "grad_norm": 1.2109375, "learning_rate": 6.810148067339776e-05, "loss": 0.1525, "step": 19831 }, { "epoch": 14.528937728937729, "grad_norm": 0.77734375, "learning_rate": 6.809827820394463e-05, "loss": 0.1419, "step": 19832 }, { "epoch": 14.52967032967033, "grad_norm": 0.98046875, "learning_rate": 6.809507565017236e-05, "loss": 0.1557, "step": 19833 }, { "epoch": 14.53040293040293, "grad_norm": 1.203125, "learning_rate": 6.809187301209631e-05, "loss": 0.153, "step": 19834 }, { "epoch": 14.531135531135531, "grad_norm": 1.5390625, "learning_rate": 6.808867028973183e-05, "loss": 0.1317, "step": 19835 }, { "epoch": 14.531868131868132, "grad_norm": 0.6484375, "learning_rate": 6.808546748309424e-05, "loss": 0.1216, "step": 19836 }, { "epoch": 14.532600732600732, "grad_norm": 1.0703125, "learning_rate": 6.80822645921989e-05, "loss": 0.1375, "step": 19837 }, { "epoch": 14.533333333333333, "grad_norm": 1.0234375, "learning_rate": 6.807906161706117e-05, "loss": 0.1552, "step": 19838 }, { "epoch": 14.534065934065934, "grad_norm": 1.1953125, "learning_rate": 6.807585855769639e-05, "loss": 0.1162, "step": 19839 }, { "epoch": 14.534798534798535, "grad_norm": 1.1484375, "learning_rate": 6.807265541411989e-05, "loss": 0.1316, "step": 19840 }, { "epoch": 14.535531135531135, "grad_norm": 1.1953125, "learning_rate": 6.806945218634704e-05, "loss": 0.1499, "step": 19841 }, { "epoch": 14.536263736263736, "grad_norm": 1.265625, "learning_rate": 6.806624887439318e-05, "loss": 0.1278, "step": 19842 }, { "epoch": 14.536996336996337, "grad_norm": 0.8046875, "learning_rate": 6.806304547827364e-05, "loss": 0.1216, "step": 19843 }, { "epoch": 14.537728937728938, "grad_norm": 1.3671875, "learning_rate": 6.805984199800378e-05, "loss": 0.1331, "step": 19844 }, { "epoch": 14.538461538461538, "grad_norm": 1.140625, "learning_rate": 6.805663843359897e-05, "loss": 0.1071, "step": 19845 }, { "epoch": 14.539194139194139, "grad_norm": 1.3203125, "learning_rate": 6.805343478507453e-05, "loss": 0.1242, "step": 19846 }, { "epoch": 14.53992673992674, "grad_norm": 0.92578125, "learning_rate": 6.805023105244584e-05, "loss": 0.1069, "step": 19847 }, { "epoch": 14.54065934065934, "grad_norm": 1.2109375, "learning_rate": 6.804702723572822e-05, "loss": 0.152, "step": 19848 }, { "epoch": 14.541391941391941, "grad_norm": 1.375, "learning_rate": 6.804382333493702e-05, "loss": 0.1327, "step": 19849 }, { "epoch": 14.542124542124542, "grad_norm": 1.5390625, "learning_rate": 6.80406193500876e-05, "loss": 0.1578, "step": 19850 }, { "epoch": 14.542857142857143, "grad_norm": 0.984375, "learning_rate": 6.803741528119534e-05, "loss": 0.1375, "step": 19851 }, { "epoch": 14.543589743589743, "grad_norm": 1.3359375, "learning_rate": 6.803421112827555e-05, "loss": 0.1258, "step": 19852 }, { "epoch": 14.544322344322344, "grad_norm": 0.8359375, "learning_rate": 6.80310068913436e-05, "loss": 0.1313, "step": 19853 }, { "epoch": 14.545054945054945, "grad_norm": 1.09375, "learning_rate": 6.802780257041485e-05, "loss": 0.1251, "step": 19854 }, { "epoch": 14.545787545787546, "grad_norm": 1.0234375, "learning_rate": 6.802459816550463e-05, "loss": 0.1114, "step": 19855 }, { "epoch": 14.546520146520146, "grad_norm": 1.28125, "learning_rate": 6.802139367662833e-05, "loss": 0.1149, "step": 19856 }, { "epoch": 14.547252747252747, "grad_norm": 1.0625, "learning_rate": 6.801818910380126e-05, "loss": 0.1379, "step": 19857 }, { "epoch": 14.547985347985348, "grad_norm": 1.1640625, "learning_rate": 6.80149844470388e-05, "loss": 0.1229, "step": 19858 }, { "epoch": 14.548717948717949, "grad_norm": 1.2578125, "learning_rate": 6.801177970635629e-05, "loss": 0.1311, "step": 19859 }, { "epoch": 14.54945054945055, "grad_norm": 1.9453125, "learning_rate": 6.800857488176912e-05, "loss": 0.1181, "step": 19860 }, { "epoch": 14.55018315018315, "grad_norm": 1.84375, "learning_rate": 6.800536997329261e-05, "loss": 0.1444, "step": 19861 }, { "epoch": 14.55091575091575, "grad_norm": 1.34375, "learning_rate": 6.80021649809421e-05, "loss": 0.1132, "step": 19862 }, { "epoch": 14.551648351648351, "grad_norm": 1.34375, "learning_rate": 6.799895990473299e-05, "loss": 0.1452, "step": 19863 }, { "epoch": 14.552380952380952, "grad_norm": 0.953125, "learning_rate": 6.799575474468063e-05, "loss": 0.0935, "step": 19864 }, { "epoch": 14.553113553113553, "grad_norm": 0.953125, "learning_rate": 6.799254950080035e-05, "loss": 0.1317, "step": 19865 }, { "epoch": 14.553846153846154, "grad_norm": 0.7890625, "learning_rate": 6.798934417310755e-05, "loss": 0.0858, "step": 19866 }, { "epoch": 14.554578754578754, "grad_norm": 0.8359375, "learning_rate": 6.798613876161752e-05, "loss": 0.126, "step": 19867 }, { "epoch": 14.555311355311355, "grad_norm": 1.6953125, "learning_rate": 6.798293326634569e-05, "loss": 0.1732, "step": 19868 }, { "epoch": 14.556043956043956, "grad_norm": 1.4921875, "learning_rate": 6.797972768730738e-05, "loss": 0.1411, "step": 19869 }, { "epoch": 14.556776556776557, "grad_norm": 1.1171875, "learning_rate": 6.797652202451795e-05, "loss": 0.1181, "step": 19870 }, { "epoch": 14.557509157509157, "grad_norm": 1.7578125, "learning_rate": 6.797331627799278e-05, "loss": 0.1356, "step": 19871 }, { "epoch": 14.558241758241758, "grad_norm": 0.96875, "learning_rate": 6.79701104477472e-05, "loss": 0.1107, "step": 19872 }, { "epoch": 14.558974358974359, "grad_norm": 1.796875, "learning_rate": 6.796690453379658e-05, "loss": 0.0926, "step": 19873 }, { "epoch": 14.55970695970696, "grad_norm": 1.078125, "learning_rate": 6.79636985361563e-05, "loss": 0.1239, "step": 19874 }, { "epoch": 14.56043956043956, "grad_norm": 1.6640625, "learning_rate": 6.796049245484171e-05, "loss": 0.1418, "step": 19875 }, { "epoch": 14.561172161172161, "grad_norm": 1.984375, "learning_rate": 6.795728628986815e-05, "loss": 0.158, "step": 19876 }, { "epoch": 14.561904761904762, "grad_norm": 1.4375, "learning_rate": 6.7954080041251e-05, "loss": 0.1052, "step": 19877 }, { "epoch": 14.562637362637362, "grad_norm": 1.125, "learning_rate": 6.795087370900566e-05, "loss": 0.1417, "step": 19878 }, { "epoch": 14.563369963369963, "grad_norm": 1.6484375, "learning_rate": 6.794766729314743e-05, "loss": 0.1241, "step": 19879 }, { "epoch": 14.564102564102564, "grad_norm": 1.0546875, "learning_rate": 6.79444607936917e-05, "loss": 0.1151, "step": 19880 }, { "epoch": 14.564835164835165, "grad_norm": 1.09375, "learning_rate": 6.794125421065382e-05, "loss": 0.1172, "step": 19881 }, { "epoch": 14.565567765567765, "grad_norm": 1.3203125, "learning_rate": 6.793804754404918e-05, "loss": 0.1655, "step": 19882 }, { "epoch": 14.566300366300366, "grad_norm": 1.671875, "learning_rate": 6.793484079389314e-05, "loss": 0.1367, "step": 19883 }, { "epoch": 14.567032967032967, "grad_norm": 1.1171875, "learning_rate": 6.793163396020104e-05, "loss": 0.1098, "step": 19884 }, { "epoch": 14.567765567765568, "grad_norm": 2.28125, "learning_rate": 6.792842704298826e-05, "loss": 0.1335, "step": 19885 }, { "epoch": 14.568498168498168, "grad_norm": 1.328125, "learning_rate": 6.792522004227017e-05, "loss": 0.1348, "step": 19886 }, { "epoch": 14.569230769230769, "grad_norm": 0.921875, "learning_rate": 6.792201295806212e-05, "loss": 0.1363, "step": 19887 }, { "epoch": 14.56996336996337, "grad_norm": 1.5625, "learning_rate": 6.791880579037949e-05, "loss": 0.1205, "step": 19888 }, { "epoch": 14.57069597069597, "grad_norm": 1.6796875, "learning_rate": 6.791559853923764e-05, "loss": 0.1275, "step": 19889 }, { "epoch": 14.571428571428571, "grad_norm": 1.2109375, "learning_rate": 6.791239120465193e-05, "loss": 0.1086, "step": 19890 }, { "epoch": 14.572161172161172, "grad_norm": 1.1484375, "learning_rate": 6.790918378663777e-05, "loss": 0.1167, "step": 19891 }, { "epoch": 14.572893772893773, "grad_norm": 0.625, "learning_rate": 6.790597628521047e-05, "loss": 0.1112, "step": 19892 }, { "epoch": 14.573626373626373, "grad_norm": 1.0546875, "learning_rate": 6.790276870038544e-05, "loss": 0.1112, "step": 19893 }, { "epoch": 14.574358974358974, "grad_norm": 1.1015625, "learning_rate": 6.789956103217801e-05, "loss": 0.1251, "step": 19894 }, { "epoch": 14.575091575091575, "grad_norm": 1.1640625, "learning_rate": 6.78963532806036e-05, "loss": 0.131, "step": 19895 }, { "epoch": 14.575824175824176, "grad_norm": 1.1875, "learning_rate": 6.789314544567754e-05, "loss": 0.1139, "step": 19896 }, { "epoch": 14.576556776556776, "grad_norm": 1.2265625, "learning_rate": 6.78899375274152e-05, "loss": 0.1494, "step": 19897 }, { "epoch": 14.577289377289377, "grad_norm": 1.171875, "learning_rate": 6.788672952583196e-05, "loss": 0.1411, "step": 19898 }, { "epoch": 14.578021978021978, "grad_norm": 0.85546875, "learning_rate": 6.788352144094321e-05, "loss": 0.101, "step": 19899 }, { "epoch": 14.578754578754578, "grad_norm": 0.84765625, "learning_rate": 6.788031327276429e-05, "loss": 0.1508, "step": 19900 }, { "epoch": 14.57948717948718, "grad_norm": 0.9453125, "learning_rate": 6.787710502131058e-05, "loss": 0.1118, "step": 19901 }, { "epoch": 14.58021978021978, "grad_norm": 0.953125, "learning_rate": 6.787389668659744e-05, "loss": 0.1169, "step": 19902 }, { "epoch": 14.58095238095238, "grad_norm": 1.140625, "learning_rate": 6.78706882686403e-05, "loss": 0.1416, "step": 19903 }, { "epoch": 14.581684981684981, "grad_norm": 0.86328125, "learning_rate": 6.786747976745446e-05, "loss": 0.1092, "step": 19904 }, { "epoch": 14.582417582417582, "grad_norm": 1.9140625, "learning_rate": 6.786427118305533e-05, "loss": 0.1331, "step": 19905 }, { "epoch": 14.583150183150183, "grad_norm": 1.78125, "learning_rate": 6.786106251545827e-05, "loss": 0.1562, "step": 19906 }, { "epoch": 14.583882783882784, "grad_norm": 1.3359375, "learning_rate": 6.785785376467868e-05, "loss": 0.1257, "step": 19907 }, { "epoch": 14.584615384615384, "grad_norm": 1.4453125, "learning_rate": 6.78546449307319e-05, "loss": 0.131, "step": 19908 }, { "epoch": 14.585347985347985, "grad_norm": 0.79296875, "learning_rate": 6.785143601363332e-05, "loss": 0.0875, "step": 19909 }, { "epoch": 14.586080586080586, "grad_norm": 0.63671875, "learning_rate": 6.784822701339832e-05, "loss": 0.1386, "step": 19910 }, { "epoch": 14.586813186813187, "grad_norm": 1.390625, "learning_rate": 6.784501793004224e-05, "loss": 0.1062, "step": 19911 }, { "epoch": 14.587545787545787, "grad_norm": 1.2265625, "learning_rate": 6.784180876358052e-05, "loss": 0.1016, "step": 19912 }, { "epoch": 14.588278388278388, "grad_norm": 0.8828125, "learning_rate": 6.783859951402849e-05, "loss": 0.1251, "step": 19913 }, { "epoch": 14.589010989010989, "grad_norm": 1.265625, "learning_rate": 6.783539018140154e-05, "loss": 0.1344, "step": 19914 }, { "epoch": 14.58974358974359, "grad_norm": 1.84375, "learning_rate": 6.783218076571504e-05, "loss": 0.1268, "step": 19915 }, { "epoch": 14.59047619047619, "grad_norm": 0.7265625, "learning_rate": 6.782897126698436e-05, "loss": 0.1291, "step": 19916 }, { "epoch": 14.59120879120879, "grad_norm": 0.6328125, "learning_rate": 6.782576168522491e-05, "loss": 0.127, "step": 19917 }, { "epoch": 14.591941391941392, "grad_norm": 1.34375, "learning_rate": 6.782255202045204e-05, "loss": 0.1501, "step": 19918 }, { "epoch": 14.592673992673992, "grad_norm": 1.25, "learning_rate": 6.781934227268114e-05, "loss": 0.113, "step": 19919 }, { "epoch": 14.593406593406593, "grad_norm": 1.0703125, "learning_rate": 6.781613244192759e-05, "loss": 0.1137, "step": 19920 }, { "epoch": 14.594139194139194, "grad_norm": 1.0, "learning_rate": 6.781292252820677e-05, "loss": 0.1346, "step": 19921 }, { "epoch": 14.594871794871795, "grad_norm": 1.1875, "learning_rate": 6.780971253153403e-05, "loss": 0.1116, "step": 19922 }, { "epoch": 14.595604395604395, "grad_norm": 0.91015625, "learning_rate": 6.78065024519248e-05, "loss": 0.1232, "step": 19923 }, { "epoch": 14.596336996336996, "grad_norm": 1.265625, "learning_rate": 6.780329228939443e-05, "loss": 0.1355, "step": 19924 }, { "epoch": 14.597069597069597, "grad_norm": 0.9375, "learning_rate": 6.780008204395829e-05, "loss": 0.1389, "step": 19925 }, { "epoch": 14.597802197802197, "grad_norm": 1.1484375, "learning_rate": 6.779687171563178e-05, "loss": 0.1579, "step": 19926 }, { "epoch": 14.598534798534798, "grad_norm": 0.96484375, "learning_rate": 6.77936613044303e-05, "loss": 0.1105, "step": 19927 }, { "epoch": 14.599267399267399, "grad_norm": 0.9375, "learning_rate": 6.77904508103692e-05, "loss": 0.1566, "step": 19928 }, { "epoch": 14.6, "grad_norm": 1.0390625, "learning_rate": 6.778724023346387e-05, "loss": 0.1423, "step": 19929 }, { "epoch": 14.6007326007326, "grad_norm": 1.3203125, "learning_rate": 6.77840295737297e-05, "loss": 0.1028, "step": 19930 }, { "epoch": 14.601465201465201, "grad_norm": 0.84375, "learning_rate": 6.778081883118206e-05, "loss": 0.1002, "step": 19931 }, { "epoch": 14.602197802197802, "grad_norm": 0.96875, "learning_rate": 6.777760800583636e-05, "loss": 0.1302, "step": 19932 }, { "epoch": 14.602930402930403, "grad_norm": 0.86328125, "learning_rate": 6.777439709770797e-05, "loss": 0.1468, "step": 19933 }, { "epoch": 14.603663003663003, "grad_norm": 0.7890625, "learning_rate": 6.777118610681227e-05, "loss": 0.1314, "step": 19934 }, { "epoch": 14.604395604395604, "grad_norm": 1.0078125, "learning_rate": 6.776797503316463e-05, "loss": 0.1072, "step": 19935 }, { "epoch": 14.605128205128205, "grad_norm": 1.234375, "learning_rate": 6.776476387678048e-05, "loss": 0.1347, "step": 19936 }, { "epoch": 14.605860805860805, "grad_norm": 1.2265625, "learning_rate": 6.776155263767517e-05, "loss": 0.1479, "step": 19937 }, { "epoch": 14.606593406593406, "grad_norm": 1.5390625, "learning_rate": 6.775834131586408e-05, "loss": 0.1494, "step": 19938 }, { "epoch": 14.607326007326007, "grad_norm": 0.765625, "learning_rate": 6.775512991136261e-05, "loss": 0.1251, "step": 19939 }, { "epoch": 14.608058608058608, "grad_norm": 0.94921875, "learning_rate": 6.775191842418617e-05, "loss": 0.1433, "step": 19940 }, { "epoch": 14.608791208791208, "grad_norm": 1.296875, "learning_rate": 6.77487068543501e-05, "loss": 0.103, "step": 19941 }, { "epoch": 14.60952380952381, "grad_norm": 0.90234375, "learning_rate": 6.774549520186984e-05, "loss": 0.1129, "step": 19942 }, { "epoch": 14.61025641025641, "grad_norm": 0.98828125, "learning_rate": 6.774228346676074e-05, "loss": 0.0929, "step": 19943 }, { "epoch": 14.61098901098901, "grad_norm": 0.7421875, "learning_rate": 6.77390716490382e-05, "loss": 0.1414, "step": 19944 }, { "epoch": 14.611721611721611, "grad_norm": 1.1171875, "learning_rate": 6.773585974871759e-05, "loss": 0.0971, "step": 19945 }, { "epoch": 14.612454212454212, "grad_norm": 1.1328125, "learning_rate": 6.773264776581434e-05, "loss": 0.1422, "step": 19946 }, { "epoch": 14.613186813186813, "grad_norm": 1.234375, "learning_rate": 6.772943570034382e-05, "loss": 0.1324, "step": 19947 }, { "epoch": 14.613919413919414, "grad_norm": 1.0078125, "learning_rate": 6.772622355232141e-05, "loss": 0.0989, "step": 19948 }, { "epoch": 14.614652014652014, "grad_norm": 1.2421875, "learning_rate": 6.772301132176251e-05, "loss": 0.1746, "step": 19949 }, { "epoch": 14.615384615384615, "grad_norm": 0.97265625, "learning_rate": 6.77197990086825e-05, "loss": 0.1252, "step": 19950 }, { "epoch": 14.616117216117216, "grad_norm": 0.88671875, "learning_rate": 6.771658661309679e-05, "loss": 0.101, "step": 19951 }, { "epoch": 14.616849816849816, "grad_norm": 1.21875, "learning_rate": 6.771337413502076e-05, "loss": 0.1272, "step": 19952 }, { "epoch": 14.617582417582417, "grad_norm": 0.96484375, "learning_rate": 6.77101615744698e-05, "loss": 0.1217, "step": 19953 }, { "epoch": 14.618315018315018, "grad_norm": 0.8828125, "learning_rate": 6.770694893145931e-05, "loss": 0.1379, "step": 19954 }, { "epoch": 14.619047619047619, "grad_norm": 1.65625, "learning_rate": 6.770373620600466e-05, "loss": 0.1286, "step": 19955 }, { "epoch": 14.61978021978022, "grad_norm": 0.7890625, "learning_rate": 6.770052339812129e-05, "loss": 0.1287, "step": 19956 }, { "epoch": 14.62051282051282, "grad_norm": 0.87109375, "learning_rate": 6.769731050782456e-05, "loss": 0.1221, "step": 19957 }, { "epoch": 14.62124542124542, "grad_norm": 0.96484375, "learning_rate": 6.769409753512985e-05, "loss": 0.1506, "step": 19958 }, { "epoch": 14.621978021978022, "grad_norm": 1.3125, "learning_rate": 6.769088448005259e-05, "loss": 0.1271, "step": 19959 }, { "epoch": 14.622710622710622, "grad_norm": 0.94140625, "learning_rate": 6.768767134260818e-05, "loss": 0.1119, "step": 19960 }, { "epoch": 14.623443223443223, "grad_norm": 1.796875, "learning_rate": 6.768445812281195e-05, "loss": 0.133, "step": 19961 }, { "epoch": 14.624175824175824, "grad_norm": 1.65625, "learning_rate": 6.768124482067937e-05, "loss": 0.1415, "step": 19962 }, { "epoch": 14.624908424908424, "grad_norm": 1.390625, "learning_rate": 6.76780314362258e-05, "loss": 0.1064, "step": 19963 }, { "epoch": 14.625641025641025, "grad_norm": 2.203125, "learning_rate": 6.767481796946664e-05, "loss": 0.1348, "step": 19964 }, { "epoch": 14.626373626373626, "grad_norm": 1.1875, "learning_rate": 6.767160442041729e-05, "loss": 0.131, "step": 19965 }, { "epoch": 14.627106227106227, "grad_norm": 0.9765625, "learning_rate": 6.766839078909313e-05, "loss": 0.1497, "step": 19966 }, { "epoch": 14.627838827838827, "grad_norm": 1.1484375, "learning_rate": 6.766517707550961e-05, "loss": 0.1063, "step": 19967 }, { "epoch": 14.628571428571428, "grad_norm": 1.078125, "learning_rate": 6.766196327968206e-05, "loss": 0.1398, "step": 19968 }, { "epoch": 14.629304029304029, "grad_norm": 0.86328125, "learning_rate": 6.765874940162591e-05, "loss": 0.1387, "step": 19969 }, { "epoch": 14.63003663003663, "grad_norm": 1.5546875, "learning_rate": 6.765553544135659e-05, "loss": 0.126, "step": 19970 }, { "epoch": 14.63076923076923, "grad_norm": 0.859375, "learning_rate": 6.765232139888944e-05, "loss": 0.1198, "step": 19971 }, { "epoch": 14.631501831501831, "grad_norm": 1.0546875, "learning_rate": 6.764910727423988e-05, "loss": 0.1323, "step": 19972 }, { "epoch": 14.632234432234432, "grad_norm": 0.734375, "learning_rate": 6.764589306742335e-05, "loss": 0.1381, "step": 19973 }, { "epoch": 14.632967032967032, "grad_norm": 0.9140625, "learning_rate": 6.764267877845521e-05, "loss": 0.1035, "step": 19974 }, { "epoch": 14.633699633699633, "grad_norm": 0.82421875, "learning_rate": 6.763946440735085e-05, "loss": 0.1067, "step": 19975 }, { "epoch": 14.634432234432234, "grad_norm": 1.03125, "learning_rate": 6.76362499541257e-05, "loss": 0.1198, "step": 19976 }, { "epoch": 14.635164835164835, "grad_norm": 0.85546875, "learning_rate": 6.763303541879516e-05, "loss": 0.1152, "step": 19977 }, { "epoch": 14.635897435897435, "grad_norm": 0.7421875, "learning_rate": 6.76298208013746e-05, "loss": 0.1129, "step": 19978 }, { "epoch": 14.636630036630036, "grad_norm": 1.6015625, "learning_rate": 6.762660610187944e-05, "loss": 0.1295, "step": 19979 }, { "epoch": 14.637362637362637, "grad_norm": 0.87890625, "learning_rate": 6.762339132032512e-05, "loss": 0.1015, "step": 19980 }, { "epoch": 14.638095238095238, "grad_norm": 0.890625, "learning_rate": 6.762017645672699e-05, "loss": 0.1207, "step": 19981 }, { "epoch": 14.638827838827838, "grad_norm": 1.671875, "learning_rate": 6.761696151110049e-05, "loss": 0.1419, "step": 19982 }, { "epoch": 14.639560439560439, "grad_norm": 1.1796875, "learning_rate": 6.761374648346099e-05, "loss": 0.1365, "step": 19983 }, { "epoch": 14.64029304029304, "grad_norm": 1.65625, "learning_rate": 6.761053137382391e-05, "loss": 0.1742, "step": 19984 }, { "epoch": 14.64102564102564, "grad_norm": 0.890625, "learning_rate": 6.760731618220468e-05, "loss": 0.131, "step": 19985 }, { "epoch": 14.641758241758241, "grad_norm": 1.2421875, "learning_rate": 6.760410090861867e-05, "loss": 0.1145, "step": 19986 }, { "epoch": 14.642490842490842, "grad_norm": 1.078125, "learning_rate": 6.760088555308131e-05, "loss": 0.1024, "step": 19987 }, { "epoch": 14.643223443223443, "grad_norm": 1.1953125, "learning_rate": 6.759767011560798e-05, "loss": 0.1231, "step": 19988 }, { "epoch": 14.643956043956043, "grad_norm": 1.5859375, "learning_rate": 6.759445459621408e-05, "loss": 0.1358, "step": 19989 }, { "epoch": 14.644688644688644, "grad_norm": 0.78515625, "learning_rate": 6.759123899491507e-05, "loss": 0.125, "step": 19990 }, { "epoch": 14.645421245421245, "grad_norm": 1.3515625, "learning_rate": 6.75880233117263e-05, "loss": 0.1481, "step": 19991 }, { "epoch": 14.646153846153846, "grad_norm": 1.125, "learning_rate": 6.758480754666322e-05, "loss": 0.1388, "step": 19992 }, { "epoch": 14.646886446886446, "grad_norm": 0.91015625, "learning_rate": 6.758159169974121e-05, "loss": 0.1282, "step": 19993 }, { "epoch": 14.647619047619047, "grad_norm": 1.03125, "learning_rate": 6.757837577097568e-05, "loss": 0.1036, "step": 19994 }, { "epoch": 14.648351648351648, "grad_norm": 1.4296875, "learning_rate": 6.757515976038205e-05, "loss": 0.1196, "step": 19995 }, { "epoch": 14.649084249084249, "grad_norm": 0.77734375, "learning_rate": 6.757194366797573e-05, "loss": 0.1032, "step": 19996 }, { "epoch": 14.64981684981685, "grad_norm": 1.4296875, "learning_rate": 6.756872749377212e-05, "loss": 0.1223, "step": 19997 }, { "epoch": 14.65054945054945, "grad_norm": 1.0859375, "learning_rate": 6.756551123778664e-05, "loss": 0.1239, "step": 19998 }, { "epoch": 14.65128205128205, "grad_norm": 1.328125, "learning_rate": 6.756229490003468e-05, "loss": 0.1236, "step": 19999 }, { "epoch": 14.652014652014651, "grad_norm": 1.109375, "learning_rate": 6.755907848053168e-05, "loss": 0.121, "step": 20000 }, { "epoch": 14.652747252747252, "grad_norm": 1.296875, "learning_rate": 6.755586197929302e-05, "loss": 0.1171, "step": 20001 }, { "epoch": 14.653479853479853, "grad_norm": 0.66796875, "learning_rate": 6.755264539633413e-05, "loss": 0.1086, "step": 20002 }, { "epoch": 14.654212454212454, "grad_norm": 0.96484375, "learning_rate": 6.754942873167043e-05, "loss": 0.1322, "step": 20003 }, { "epoch": 14.654945054945054, "grad_norm": 1.1015625, "learning_rate": 6.75462119853173e-05, "loss": 0.1029, "step": 20004 }, { "epoch": 14.655677655677655, "grad_norm": 0.86328125, "learning_rate": 6.754299515729016e-05, "loss": 0.1243, "step": 20005 }, { "epoch": 14.656410256410256, "grad_norm": 0.8125, "learning_rate": 6.753977824760448e-05, "loss": 0.1321, "step": 20006 }, { "epoch": 14.657142857142857, "grad_norm": 1.2109375, "learning_rate": 6.75365612562756e-05, "loss": 0.1362, "step": 20007 }, { "epoch": 14.657875457875457, "grad_norm": 1.5390625, "learning_rate": 6.753334418331896e-05, "loss": 0.1284, "step": 20008 }, { "epoch": 14.658608058608058, "grad_norm": 0.62109375, "learning_rate": 6.753012702875e-05, "loss": 0.1037, "step": 20009 }, { "epoch": 14.659340659340659, "grad_norm": 0.859375, "learning_rate": 6.752690979258409e-05, "loss": 0.1139, "step": 20010 }, { "epoch": 14.66007326007326, "grad_norm": 1.078125, "learning_rate": 6.752369247483667e-05, "loss": 0.1369, "step": 20011 }, { "epoch": 14.66080586080586, "grad_norm": 1.0546875, "learning_rate": 6.752047507552316e-05, "loss": 0.1318, "step": 20012 }, { "epoch": 14.661538461538461, "grad_norm": 0.828125, "learning_rate": 6.751725759465894e-05, "loss": 0.1108, "step": 20013 }, { "epoch": 14.662271062271062, "grad_norm": 1.5390625, "learning_rate": 6.751404003225948e-05, "loss": 0.1321, "step": 20014 }, { "epoch": 14.663003663003662, "grad_norm": 0.96875, "learning_rate": 6.751082238834016e-05, "loss": 0.1081, "step": 20015 }, { "epoch": 14.663736263736263, "grad_norm": 1.3828125, "learning_rate": 6.75076046629164e-05, "loss": 0.1534, "step": 20016 }, { "epoch": 14.664468864468864, "grad_norm": 0.98046875, "learning_rate": 6.750438685600363e-05, "loss": 0.1331, "step": 20017 }, { "epoch": 14.665201465201465, "grad_norm": 1.78125, "learning_rate": 6.750116896761724e-05, "loss": 0.1402, "step": 20018 }, { "epoch": 14.665934065934065, "grad_norm": 0.9765625, "learning_rate": 6.74979509977727e-05, "loss": 0.1033, "step": 20019 }, { "epoch": 14.666666666666666, "grad_norm": 1.75, "learning_rate": 6.749473294648538e-05, "loss": 0.1373, "step": 20020 }, { "epoch": 14.667399267399267, "grad_norm": 1.53125, "learning_rate": 6.749151481377072e-05, "loss": 0.1368, "step": 20021 }, { "epoch": 14.668131868131868, "grad_norm": 0.83984375, "learning_rate": 6.748829659964413e-05, "loss": 0.0943, "step": 20022 }, { "epoch": 14.668864468864468, "grad_norm": 0.80078125, "learning_rate": 6.748507830412104e-05, "loss": 0.1287, "step": 20023 }, { "epoch": 14.669597069597069, "grad_norm": 0.81640625, "learning_rate": 6.748185992721688e-05, "loss": 0.1185, "step": 20024 }, { "epoch": 14.67032967032967, "grad_norm": 0.7890625, "learning_rate": 6.747864146894703e-05, "loss": 0.1341, "step": 20025 }, { "epoch": 14.67106227106227, "grad_norm": 0.9375, "learning_rate": 6.747542292932694e-05, "loss": 0.0996, "step": 20026 }, { "epoch": 14.671794871794871, "grad_norm": 1.0625, "learning_rate": 6.747220430837204e-05, "loss": 0.1396, "step": 20027 }, { "epoch": 14.672527472527472, "grad_norm": 0.90625, "learning_rate": 6.74689856060977e-05, "loss": 0.1262, "step": 20028 }, { "epoch": 14.673260073260073, "grad_norm": 0.78125, "learning_rate": 6.746576682251941e-05, "loss": 0.153, "step": 20029 }, { "epoch": 14.673992673992673, "grad_norm": 1.25, "learning_rate": 6.746254795765255e-05, "loss": 0.1409, "step": 20030 }, { "epoch": 14.674725274725274, "grad_norm": 1.78125, "learning_rate": 6.745932901151255e-05, "loss": 0.147, "step": 20031 }, { "epoch": 14.675457875457875, "grad_norm": 1.0546875, "learning_rate": 6.745610998411485e-05, "loss": 0.144, "step": 20032 }, { "epoch": 14.676190476190476, "grad_norm": 1.2734375, "learning_rate": 6.745289087547488e-05, "loss": 0.1471, "step": 20033 }, { "epoch": 14.676923076923076, "grad_norm": 1.1953125, "learning_rate": 6.7449671685608e-05, "loss": 0.1327, "step": 20034 }, { "epoch": 14.677655677655677, "grad_norm": 1.1953125, "learning_rate": 6.744645241452971e-05, "loss": 0.1242, "step": 20035 }, { "epoch": 14.678388278388278, "grad_norm": 1.59375, "learning_rate": 6.744323306225538e-05, "loss": 0.1317, "step": 20036 }, { "epoch": 14.679120879120878, "grad_norm": 1.96875, "learning_rate": 6.744001362880049e-05, "loss": 0.1566, "step": 20037 }, { "epoch": 14.67985347985348, "grad_norm": 2.140625, "learning_rate": 6.743679411418042e-05, "loss": 0.1256, "step": 20038 }, { "epoch": 14.68058608058608, "grad_norm": 1.1328125, "learning_rate": 6.743357451841059e-05, "loss": 0.1235, "step": 20039 }, { "epoch": 14.68131868131868, "grad_norm": 1.4765625, "learning_rate": 6.743035484150648e-05, "loss": 0.1067, "step": 20040 }, { "epoch": 14.682051282051281, "grad_norm": 1.015625, "learning_rate": 6.742713508348345e-05, "loss": 0.1046, "step": 20041 }, { "epoch": 14.682783882783882, "grad_norm": 1.3203125, "learning_rate": 6.742391524435697e-05, "loss": 0.1512, "step": 20042 }, { "epoch": 14.683516483516483, "grad_norm": 1.171875, "learning_rate": 6.742069532414246e-05, "loss": 0.1374, "step": 20043 }, { "epoch": 14.684249084249084, "grad_norm": 1.1640625, "learning_rate": 6.741747532285534e-05, "loss": 0.1199, "step": 20044 }, { "epoch": 14.684981684981684, "grad_norm": 1.3125, "learning_rate": 6.741425524051105e-05, "loss": 0.1244, "step": 20045 }, { "epoch": 14.685714285714285, "grad_norm": 1.6875, "learning_rate": 6.741103507712501e-05, "loss": 0.1044, "step": 20046 }, { "epoch": 14.686446886446886, "grad_norm": 0.796875, "learning_rate": 6.740781483271263e-05, "loss": 0.1337, "step": 20047 }, { "epoch": 14.687179487179487, "grad_norm": 0.953125, "learning_rate": 6.740459450728939e-05, "loss": 0.1211, "step": 20048 }, { "epoch": 14.687912087912087, "grad_norm": 1.765625, "learning_rate": 6.740137410087066e-05, "loss": 0.1369, "step": 20049 }, { "epoch": 14.688644688644688, "grad_norm": 1.328125, "learning_rate": 6.73981536134719e-05, "loss": 0.1185, "step": 20050 }, { "epoch": 14.689377289377289, "grad_norm": 1.234375, "learning_rate": 6.739493304510857e-05, "loss": 0.1294, "step": 20051 }, { "epoch": 14.69010989010989, "grad_norm": 1.5546875, "learning_rate": 6.739171239579603e-05, "loss": 0.1142, "step": 20052 }, { "epoch": 14.69084249084249, "grad_norm": 1.609375, "learning_rate": 6.738849166554977e-05, "loss": 0.1664, "step": 20053 }, { "epoch": 14.691575091575091, "grad_norm": 1.1953125, "learning_rate": 6.73852708543852e-05, "loss": 0.1597, "step": 20054 }, { "epoch": 14.692307692307692, "grad_norm": 0.5625, "learning_rate": 6.738204996231776e-05, "loss": 0.1082, "step": 20055 }, { "epoch": 14.693040293040292, "grad_norm": 0.6171875, "learning_rate": 6.737882898936288e-05, "loss": 0.1233, "step": 20056 }, { "epoch": 14.693772893772893, "grad_norm": 1.1328125, "learning_rate": 6.737560793553596e-05, "loss": 0.1431, "step": 20057 }, { "epoch": 14.694505494505494, "grad_norm": 0.80078125, "learning_rate": 6.737238680085249e-05, "loss": 0.1187, "step": 20058 }, { "epoch": 14.695238095238095, "grad_norm": 0.8671875, "learning_rate": 6.736916558532789e-05, "loss": 0.1226, "step": 20059 }, { "epoch": 14.695970695970695, "grad_norm": 1.046875, "learning_rate": 6.736594428897754e-05, "loss": 0.1017, "step": 20060 }, { "epoch": 14.696703296703296, "grad_norm": 1.328125, "learning_rate": 6.736272291181695e-05, "loss": 0.1431, "step": 20061 }, { "epoch": 14.697435897435897, "grad_norm": 1.6640625, "learning_rate": 6.73595014538615e-05, "loss": 0.1311, "step": 20062 }, { "epoch": 14.698168498168497, "grad_norm": 0.9375, "learning_rate": 6.735627991512664e-05, "loss": 0.1285, "step": 20063 }, { "epoch": 14.698901098901098, "grad_norm": 1.1640625, "learning_rate": 6.735305829562782e-05, "loss": 0.1185, "step": 20064 }, { "epoch": 14.699633699633699, "grad_norm": 0.984375, "learning_rate": 6.734983659538045e-05, "loss": 0.1419, "step": 20065 }, { "epoch": 14.7003663003663, "grad_norm": 1.2265625, "learning_rate": 6.73466148144e-05, "loss": 0.1445, "step": 20066 }, { "epoch": 14.7010989010989, "grad_norm": 1.3046875, "learning_rate": 6.734339295270188e-05, "loss": 0.1235, "step": 20067 }, { "epoch": 14.701831501831501, "grad_norm": 0.74609375, "learning_rate": 6.734017101030153e-05, "loss": 0.1367, "step": 20068 }, { "epoch": 14.702564102564102, "grad_norm": 1.0625, "learning_rate": 6.73369489872144e-05, "loss": 0.1408, "step": 20069 }, { "epoch": 14.703296703296703, "grad_norm": 1.2890625, "learning_rate": 6.733372688345592e-05, "loss": 0.1035, "step": 20070 }, { "epoch": 14.704029304029303, "grad_norm": 1.3671875, "learning_rate": 6.733050469904153e-05, "loss": 0.1501, "step": 20071 }, { "epoch": 14.704761904761904, "grad_norm": 0.8515625, "learning_rate": 6.732728243398667e-05, "loss": 0.1503, "step": 20072 }, { "epoch": 14.705494505494505, "grad_norm": 1.2890625, "learning_rate": 6.732406008830676e-05, "loss": 0.1271, "step": 20073 }, { "epoch": 14.706227106227105, "grad_norm": 0.7890625, "learning_rate": 6.732083766201726e-05, "loss": 0.1137, "step": 20074 }, { "epoch": 14.706959706959706, "grad_norm": 1.3125, "learning_rate": 6.731761515513361e-05, "loss": 0.1099, "step": 20075 }, { "epoch": 14.707692307692307, "grad_norm": 0.9140625, "learning_rate": 6.731439256767125e-05, "loss": 0.1087, "step": 20076 }, { "epoch": 14.708424908424908, "grad_norm": 1.1875, "learning_rate": 6.731116989964561e-05, "loss": 0.122, "step": 20077 }, { "epoch": 14.709157509157508, "grad_norm": 1.171875, "learning_rate": 6.730794715107214e-05, "loss": 0.0999, "step": 20078 }, { "epoch": 14.70989010989011, "grad_norm": 0.71875, "learning_rate": 6.730472432196627e-05, "loss": 0.139, "step": 20079 }, { "epoch": 14.71062271062271, "grad_norm": 0.859375, "learning_rate": 6.730150141234346e-05, "loss": 0.1091, "step": 20080 }, { "epoch": 14.71135531135531, "grad_norm": 1.0234375, "learning_rate": 6.729827842221912e-05, "loss": 0.1242, "step": 20081 }, { "epoch": 14.712087912087911, "grad_norm": 1.0234375, "learning_rate": 6.729505535160873e-05, "loss": 0.1183, "step": 20082 }, { "epoch": 14.712820512820512, "grad_norm": 1.296875, "learning_rate": 6.729183220052772e-05, "loss": 0.1181, "step": 20083 }, { "epoch": 14.713553113553113, "grad_norm": 1.15625, "learning_rate": 6.728860896899152e-05, "loss": 0.118, "step": 20084 }, { "epoch": 14.714285714285714, "grad_norm": 0.734375, "learning_rate": 6.728538565701559e-05, "loss": 0.1195, "step": 20085 }, { "epoch": 14.715018315018314, "grad_norm": 1.40625, "learning_rate": 6.728216226461536e-05, "loss": 0.129, "step": 20086 }, { "epoch": 14.715750915750915, "grad_norm": 1.28125, "learning_rate": 6.727893879180629e-05, "loss": 0.1175, "step": 20087 }, { "epoch": 14.716483516483516, "grad_norm": 0.921875, "learning_rate": 6.727571523860381e-05, "loss": 0.1019, "step": 20088 }, { "epoch": 14.717216117216116, "grad_norm": 0.609375, "learning_rate": 6.727249160502337e-05, "loss": 0.1067, "step": 20089 }, { "epoch": 14.717948717948717, "grad_norm": 0.9296875, "learning_rate": 6.726926789108042e-05, "loss": 0.125, "step": 20090 }, { "epoch": 14.718681318681318, "grad_norm": 0.79296875, "learning_rate": 6.726604409679039e-05, "loss": 0.0864, "step": 20091 }, { "epoch": 14.719413919413919, "grad_norm": 1.40625, "learning_rate": 6.726282022216876e-05, "loss": 0.1461, "step": 20092 }, { "epoch": 14.72014652014652, "grad_norm": 1.2734375, "learning_rate": 6.725959626723094e-05, "loss": 0.1538, "step": 20093 }, { "epoch": 14.72087912087912, "grad_norm": 0.8203125, "learning_rate": 6.72563722319924e-05, "loss": 0.1071, "step": 20094 }, { "epoch": 14.72161172161172, "grad_norm": 1.4375, "learning_rate": 6.725314811646858e-05, "loss": 0.1393, "step": 20095 }, { "epoch": 14.722344322344322, "grad_norm": 1.25, "learning_rate": 6.724992392067492e-05, "loss": 0.1439, "step": 20096 }, { "epoch": 14.723076923076922, "grad_norm": 1.34375, "learning_rate": 6.724669964462689e-05, "loss": 0.1138, "step": 20097 }, { "epoch": 14.723809523809523, "grad_norm": 0.609375, "learning_rate": 6.724347528833991e-05, "loss": 0.1243, "step": 20098 }, { "epoch": 14.724542124542124, "grad_norm": 0.91015625, "learning_rate": 6.724025085182945e-05, "loss": 0.1067, "step": 20099 }, { "epoch": 14.725274725274724, "grad_norm": 1.1875, "learning_rate": 6.723702633511097e-05, "loss": 0.1202, "step": 20100 }, { "epoch": 14.726007326007325, "grad_norm": 0.828125, "learning_rate": 6.723380173819988e-05, "loss": 0.1314, "step": 20101 }, { "epoch": 14.726739926739926, "grad_norm": 1.0859375, "learning_rate": 6.723057706111164e-05, "loss": 0.1603, "step": 20102 }, { "epoch": 14.727472527472527, "grad_norm": 0.95703125, "learning_rate": 6.722735230386174e-05, "loss": 0.1673, "step": 20103 }, { "epoch": 14.728205128205127, "grad_norm": 1.09375, "learning_rate": 6.72241274664656e-05, "loss": 0.1332, "step": 20104 }, { "epoch": 14.728937728937728, "grad_norm": 0.98046875, "learning_rate": 6.722090254893866e-05, "loss": 0.1276, "step": 20105 }, { "epoch": 14.729670329670329, "grad_norm": 1.2265625, "learning_rate": 6.72176775512964e-05, "loss": 0.1202, "step": 20106 }, { "epoch": 14.73040293040293, "grad_norm": 1.1640625, "learning_rate": 6.721445247355424e-05, "loss": 0.1062, "step": 20107 }, { "epoch": 14.73113553113553, "grad_norm": 1.109375, "learning_rate": 6.721122731572767e-05, "loss": 0.1409, "step": 20108 }, { "epoch": 14.731868131868131, "grad_norm": 1.59375, "learning_rate": 6.720800207783213e-05, "loss": 0.1169, "step": 20109 }, { "epoch": 14.732600732600732, "grad_norm": 0.875, "learning_rate": 6.720477675988305e-05, "loss": 0.0995, "step": 20110 }, { "epoch": 14.733333333333333, "grad_norm": 0.97265625, "learning_rate": 6.720155136189591e-05, "loss": 0.113, "step": 20111 }, { "epoch": 14.734065934065933, "grad_norm": 0.87109375, "learning_rate": 6.719832588388615e-05, "loss": 0.1141, "step": 20112 }, { "epoch": 14.734798534798536, "grad_norm": 2.234375, "learning_rate": 6.719510032586924e-05, "loss": 0.1722, "step": 20113 }, { "epoch": 14.735531135531136, "grad_norm": 1.265625, "learning_rate": 6.71918746878606e-05, "loss": 0.129, "step": 20114 }, { "epoch": 14.736263736263737, "grad_norm": 1.3359375, "learning_rate": 6.718864896987572e-05, "loss": 0.1233, "step": 20115 }, { "epoch": 14.736996336996338, "grad_norm": 1.390625, "learning_rate": 6.718542317193006e-05, "loss": 0.1282, "step": 20116 }, { "epoch": 14.737728937728939, "grad_norm": 1.2421875, "learning_rate": 6.718219729403905e-05, "loss": 0.1196, "step": 20117 }, { "epoch": 14.73846153846154, "grad_norm": 1.4765625, "learning_rate": 6.717897133621815e-05, "loss": 0.148, "step": 20118 }, { "epoch": 14.73919413919414, "grad_norm": 0.91796875, "learning_rate": 6.717574529848283e-05, "loss": 0.115, "step": 20119 }, { "epoch": 14.73992673992674, "grad_norm": 1.890625, "learning_rate": 6.717251918084853e-05, "loss": 0.1436, "step": 20120 }, { "epoch": 14.740659340659342, "grad_norm": 0.8984375, "learning_rate": 6.716929298333073e-05, "loss": 0.1283, "step": 20121 }, { "epoch": 14.741391941391942, "grad_norm": 0.9140625, "learning_rate": 6.716606670594486e-05, "loss": 0.1143, "step": 20122 }, { "epoch": 14.742124542124543, "grad_norm": 1.125, "learning_rate": 6.716284034870641e-05, "loss": 0.1177, "step": 20123 }, { "epoch": 14.742857142857144, "grad_norm": 1.21875, "learning_rate": 6.715961391163081e-05, "loss": 0.124, "step": 20124 }, { "epoch": 14.743589743589745, "grad_norm": 1.0703125, "learning_rate": 6.715638739473354e-05, "loss": 0.1245, "step": 20125 }, { "epoch": 14.744322344322345, "grad_norm": 1.1953125, "learning_rate": 6.715316079803005e-05, "loss": 0.1338, "step": 20126 }, { "epoch": 14.745054945054946, "grad_norm": 1.3203125, "learning_rate": 6.714993412153578e-05, "loss": 0.1413, "step": 20127 }, { "epoch": 14.745787545787547, "grad_norm": 0.921875, "learning_rate": 6.714670736526624e-05, "loss": 0.121, "step": 20128 }, { "epoch": 14.746520146520147, "grad_norm": 0.734375, "learning_rate": 6.714348052923686e-05, "loss": 0.1391, "step": 20129 }, { "epoch": 14.747252747252748, "grad_norm": 0.8203125, "learning_rate": 6.714025361346307e-05, "loss": 0.1098, "step": 20130 }, { "epoch": 14.747985347985349, "grad_norm": 0.9375, "learning_rate": 6.713702661796039e-05, "loss": 0.1365, "step": 20131 }, { "epoch": 14.74871794871795, "grad_norm": 1.3203125, "learning_rate": 6.713379954274424e-05, "loss": 0.116, "step": 20132 }, { "epoch": 14.74945054945055, "grad_norm": 1.484375, "learning_rate": 6.713057238783009e-05, "loss": 0.1188, "step": 20133 }, { "epoch": 14.750183150183151, "grad_norm": 1.015625, "learning_rate": 6.712734515323342e-05, "loss": 0.1059, "step": 20134 }, { "epoch": 14.750915750915752, "grad_norm": 0.98046875, "learning_rate": 6.712411783896966e-05, "loss": 0.1011, "step": 20135 }, { "epoch": 14.751648351648353, "grad_norm": 1.0234375, "learning_rate": 6.712089044505433e-05, "loss": 0.1408, "step": 20136 }, { "epoch": 14.752380952380953, "grad_norm": 1.1640625, "learning_rate": 6.711766297150285e-05, "loss": 0.1196, "step": 20137 }, { "epoch": 14.753113553113554, "grad_norm": 1.21875, "learning_rate": 6.711443541833067e-05, "loss": 0.1403, "step": 20138 }, { "epoch": 14.753846153846155, "grad_norm": 1.234375, "learning_rate": 6.71112077855533e-05, "loss": 0.1401, "step": 20139 }, { "epoch": 14.754578754578755, "grad_norm": 2.8125, "learning_rate": 6.710798007318615e-05, "loss": 0.1649, "step": 20140 }, { "epoch": 14.755311355311356, "grad_norm": 0.9375, "learning_rate": 6.710475228124474e-05, "loss": 0.1191, "step": 20141 }, { "epoch": 14.756043956043957, "grad_norm": 0.76953125, "learning_rate": 6.71015244097445e-05, "loss": 0.1287, "step": 20142 }, { "epoch": 14.756776556776558, "grad_norm": 0.82421875, "learning_rate": 6.709829645870092e-05, "loss": 0.1348, "step": 20143 }, { "epoch": 14.757509157509158, "grad_norm": 0.9765625, "learning_rate": 6.709506842812944e-05, "loss": 0.1539, "step": 20144 }, { "epoch": 14.758241758241759, "grad_norm": 1.1328125, "learning_rate": 6.709184031804553e-05, "loss": 0.1182, "step": 20145 }, { "epoch": 14.75897435897436, "grad_norm": 1.0546875, "learning_rate": 6.708861212846468e-05, "loss": 0.1268, "step": 20146 }, { "epoch": 14.75970695970696, "grad_norm": 1.3359375, "learning_rate": 6.708538385940235e-05, "loss": 0.1786, "step": 20147 }, { "epoch": 14.760439560439561, "grad_norm": 2.0, "learning_rate": 6.708215551087397e-05, "loss": 0.1146, "step": 20148 }, { "epoch": 14.761172161172162, "grad_norm": 0.828125, "learning_rate": 6.707892708289508e-05, "loss": 0.1259, "step": 20149 }, { "epoch": 14.761904761904763, "grad_norm": 0.765625, "learning_rate": 6.707569857548109e-05, "loss": 0.1084, "step": 20150 }, { "epoch": 14.762637362637363, "grad_norm": 0.97265625, "learning_rate": 6.707246998864748e-05, "loss": 0.1465, "step": 20151 }, { "epoch": 14.763369963369964, "grad_norm": 1.109375, "learning_rate": 6.706924132240975e-05, "loss": 0.115, "step": 20152 }, { "epoch": 14.764102564102565, "grad_norm": 1.0078125, "learning_rate": 6.706601257678332e-05, "loss": 0.1518, "step": 20153 }, { "epoch": 14.764835164835166, "grad_norm": 0.8046875, "learning_rate": 6.70627837517837e-05, "loss": 0.1087, "step": 20154 }, { "epoch": 14.765567765567766, "grad_norm": 1.203125, "learning_rate": 6.705955484742636e-05, "loss": 0.1641, "step": 20155 }, { "epoch": 14.766300366300367, "grad_norm": 0.84765625, "learning_rate": 6.705632586372671e-05, "loss": 0.1307, "step": 20156 }, { "epoch": 14.767032967032968, "grad_norm": 0.88671875, "learning_rate": 6.70530968007003e-05, "loss": 0.1265, "step": 20157 }, { "epoch": 14.767765567765569, "grad_norm": 1.4140625, "learning_rate": 6.704986765836256e-05, "loss": 0.1109, "step": 20158 }, { "epoch": 14.76849816849817, "grad_norm": 0.85546875, "learning_rate": 6.704663843672897e-05, "loss": 0.1065, "step": 20159 }, { "epoch": 14.76923076923077, "grad_norm": 1.2265625, "learning_rate": 6.704340913581502e-05, "loss": 0.1614, "step": 20160 }, { "epoch": 14.76996336996337, "grad_norm": 1.3125, "learning_rate": 6.704017975563614e-05, "loss": 0.1136, "step": 20161 }, { "epoch": 14.770695970695972, "grad_norm": 1.125, "learning_rate": 6.703695029620784e-05, "loss": 0.1162, "step": 20162 }, { "epoch": 14.771428571428572, "grad_norm": 1.203125, "learning_rate": 6.70337207575456e-05, "loss": 0.1296, "step": 20163 }, { "epoch": 14.772161172161173, "grad_norm": 0.80859375, "learning_rate": 6.703049113966485e-05, "loss": 0.0908, "step": 20164 }, { "epoch": 14.772893772893774, "grad_norm": 0.93359375, "learning_rate": 6.702726144258112e-05, "loss": 0.0985, "step": 20165 }, { "epoch": 14.773626373626374, "grad_norm": 1.265625, "learning_rate": 6.702403166630982e-05, "loss": 0.1081, "step": 20166 }, { "epoch": 14.774358974358975, "grad_norm": 0.6015625, "learning_rate": 6.702080181086649e-05, "loss": 0.1049, "step": 20167 }, { "epoch": 14.775091575091576, "grad_norm": 0.6328125, "learning_rate": 6.701757187626656e-05, "loss": 0.1122, "step": 20168 }, { "epoch": 14.775824175824177, "grad_norm": 1.3046875, "learning_rate": 6.701434186252552e-05, "loss": 0.1087, "step": 20169 }, { "epoch": 14.776556776556777, "grad_norm": 1.1875, "learning_rate": 6.701111176965885e-05, "loss": 0.132, "step": 20170 }, { "epoch": 14.777289377289378, "grad_norm": 1.078125, "learning_rate": 6.700788159768203e-05, "loss": 0.1033, "step": 20171 }, { "epoch": 14.778021978021979, "grad_norm": 0.796875, "learning_rate": 6.70046513466105e-05, "loss": 0.1098, "step": 20172 }, { "epoch": 14.77875457875458, "grad_norm": 0.7890625, "learning_rate": 6.70014210164598e-05, "loss": 0.1011, "step": 20173 }, { "epoch": 14.77948717948718, "grad_norm": 1.140625, "learning_rate": 6.699819060724536e-05, "loss": 0.1355, "step": 20174 }, { "epoch": 14.780219780219781, "grad_norm": 1.9296875, "learning_rate": 6.699496011898268e-05, "loss": 0.1274, "step": 20175 }, { "epoch": 14.780952380952382, "grad_norm": 0.7578125, "learning_rate": 6.699172955168723e-05, "loss": 0.1009, "step": 20176 }, { "epoch": 14.781684981684982, "grad_norm": 1.1796875, "learning_rate": 6.69884989053745e-05, "loss": 0.1452, "step": 20177 }, { "epoch": 14.782417582417583, "grad_norm": 0.79296875, "learning_rate": 6.698526818005994e-05, "loss": 0.1266, "step": 20178 }, { "epoch": 14.783150183150184, "grad_norm": 1.5859375, "learning_rate": 6.698203737575905e-05, "loss": 0.1154, "step": 20179 }, { "epoch": 14.783882783882785, "grad_norm": 0.828125, "learning_rate": 6.697880649248732e-05, "loss": 0.1174, "step": 20180 }, { "epoch": 14.784615384615385, "grad_norm": 0.69921875, "learning_rate": 6.697557553026021e-05, "loss": 0.1271, "step": 20181 }, { "epoch": 14.785347985347986, "grad_norm": 0.875, "learning_rate": 6.697234448909323e-05, "loss": 0.1022, "step": 20182 }, { "epoch": 14.786080586080587, "grad_norm": 0.67578125, "learning_rate": 6.696911336900181e-05, "loss": 0.1261, "step": 20183 }, { "epoch": 14.786813186813188, "grad_norm": 1.1953125, "learning_rate": 6.696588217000147e-05, "loss": 0.1307, "step": 20184 }, { "epoch": 14.787545787545788, "grad_norm": 0.9921875, "learning_rate": 6.696265089210768e-05, "loss": 0.1225, "step": 20185 }, { "epoch": 14.788278388278389, "grad_norm": 1.0078125, "learning_rate": 6.695941953533594e-05, "loss": 0.1345, "step": 20186 }, { "epoch": 14.78901098901099, "grad_norm": 1.046875, "learning_rate": 6.695618809970171e-05, "loss": 0.1223, "step": 20187 }, { "epoch": 14.78974358974359, "grad_norm": 1.265625, "learning_rate": 6.695295658522048e-05, "loss": 0.1528, "step": 20188 }, { "epoch": 14.790476190476191, "grad_norm": 1.0625, "learning_rate": 6.694972499190773e-05, "loss": 0.1208, "step": 20189 }, { "epoch": 14.791208791208792, "grad_norm": 1.1953125, "learning_rate": 6.694649331977894e-05, "loss": 0.1485, "step": 20190 }, { "epoch": 14.791941391941393, "grad_norm": 1.234375, "learning_rate": 6.694326156884964e-05, "loss": 0.1348, "step": 20191 }, { "epoch": 14.792673992673993, "grad_norm": 0.94921875, "learning_rate": 6.694002973913522e-05, "loss": 0.107, "step": 20192 }, { "epoch": 14.793406593406594, "grad_norm": 0.76953125, "learning_rate": 6.693679783065126e-05, "loss": 0.1347, "step": 20193 }, { "epoch": 14.794139194139195, "grad_norm": 0.64453125, "learning_rate": 6.69335658434132e-05, "loss": 0.1225, "step": 20194 }, { "epoch": 14.794871794871796, "grad_norm": 1.125, "learning_rate": 6.693033377743653e-05, "loss": 0.137, "step": 20195 }, { "epoch": 14.795604395604396, "grad_norm": 1.765625, "learning_rate": 6.692710163273674e-05, "loss": 0.1925, "step": 20196 }, { "epoch": 14.796336996336997, "grad_norm": 0.875, "learning_rate": 6.69238694093293e-05, "loss": 0.1516, "step": 20197 }, { "epoch": 14.797069597069598, "grad_norm": 1.21875, "learning_rate": 6.69206371072297e-05, "loss": 0.1345, "step": 20198 }, { "epoch": 14.797802197802199, "grad_norm": 0.94921875, "learning_rate": 6.691740472645347e-05, "loss": 0.1054, "step": 20199 }, { "epoch": 14.7985347985348, "grad_norm": 0.8046875, "learning_rate": 6.691417226701603e-05, "loss": 0.1565, "step": 20200 }, { "epoch": 14.7992673992674, "grad_norm": 0.83984375, "learning_rate": 6.691093972893293e-05, "loss": 0.1273, "step": 20201 }, { "epoch": 14.8, "grad_norm": 1.390625, "learning_rate": 6.690770711221964e-05, "loss": 0.1291, "step": 20202 }, { "epoch": 14.800732600732601, "grad_norm": 1.453125, "learning_rate": 6.690447441689159e-05, "loss": 0.1662, "step": 20203 }, { "epoch": 14.801465201465202, "grad_norm": 1.0859375, "learning_rate": 6.690124164296436e-05, "loss": 0.1242, "step": 20204 }, { "epoch": 14.802197802197803, "grad_norm": 1.171875, "learning_rate": 6.689800879045339e-05, "loss": 0.1348, "step": 20205 }, { "epoch": 14.802930402930404, "grad_norm": 1.3671875, "learning_rate": 6.689477585937417e-05, "loss": 0.1237, "step": 20206 }, { "epoch": 14.803663003663004, "grad_norm": 0.8359375, "learning_rate": 6.68915428497422e-05, "loss": 0.1168, "step": 20207 }, { "epoch": 14.804395604395605, "grad_norm": 1.1796875, "learning_rate": 6.688830976157296e-05, "loss": 0.1258, "step": 20208 }, { "epoch": 14.805128205128206, "grad_norm": 0.6953125, "learning_rate": 6.688507659488196e-05, "loss": 0.1355, "step": 20209 }, { "epoch": 14.805860805860807, "grad_norm": 1.21875, "learning_rate": 6.688184334968466e-05, "loss": 0.1323, "step": 20210 }, { "epoch": 14.806593406593407, "grad_norm": 0.85546875, "learning_rate": 6.687861002599657e-05, "loss": 0.1056, "step": 20211 }, { "epoch": 14.807326007326008, "grad_norm": 1.375, "learning_rate": 6.687537662383322e-05, "loss": 0.1415, "step": 20212 }, { "epoch": 14.808058608058609, "grad_norm": 1.4375, "learning_rate": 6.687214314321003e-05, "loss": 0.1301, "step": 20213 }, { "epoch": 14.80879120879121, "grad_norm": 1.0546875, "learning_rate": 6.686890958414254e-05, "loss": 0.1193, "step": 20214 }, { "epoch": 14.80952380952381, "grad_norm": 0.71484375, "learning_rate": 6.686567594664624e-05, "loss": 0.1078, "step": 20215 }, { "epoch": 14.810256410256411, "grad_norm": 1.25, "learning_rate": 6.686244223073659e-05, "loss": 0.1107, "step": 20216 }, { "epoch": 14.810989010989012, "grad_norm": 1.046875, "learning_rate": 6.685920843642913e-05, "loss": 0.1277, "step": 20217 }, { "epoch": 14.811721611721612, "grad_norm": 1.015625, "learning_rate": 6.685597456373932e-05, "loss": 0.1234, "step": 20218 }, { "epoch": 14.812454212454213, "grad_norm": 0.87890625, "learning_rate": 6.685274061268269e-05, "loss": 0.1353, "step": 20219 }, { "epoch": 14.813186813186814, "grad_norm": 1.0234375, "learning_rate": 6.684950658327467e-05, "loss": 0.1243, "step": 20220 }, { "epoch": 14.813919413919415, "grad_norm": 0.91796875, "learning_rate": 6.684627247553083e-05, "loss": 0.1272, "step": 20221 }, { "epoch": 14.814652014652015, "grad_norm": 1.2734375, "learning_rate": 6.68430382894666e-05, "loss": 0.1216, "step": 20222 }, { "epoch": 14.815384615384616, "grad_norm": 1.6953125, "learning_rate": 6.683980402509753e-05, "loss": 0.1347, "step": 20223 }, { "epoch": 14.816117216117217, "grad_norm": 0.578125, "learning_rate": 6.68365696824391e-05, "loss": 0.1198, "step": 20224 }, { "epoch": 14.816849816849818, "grad_norm": 0.8203125, "learning_rate": 6.683333526150679e-05, "loss": 0.1117, "step": 20225 }, { "epoch": 14.817582417582418, "grad_norm": 1.8671875, "learning_rate": 6.683010076231612e-05, "loss": 0.1492, "step": 20226 }, { "epoch": 14.818315018315019, "grad_norm": 1.5546875, "learning_rate": 6.682686618488257e-05, "loss": 0.1485, "step": 20227 }, { "epoch": 14.81904761904762, "grad_norm": 0.953125, "learning_rate": 6.682363152922165e-05, "loss": 0.1413, "step": 20228 }, { "epoch": 14.81978021978022, "grad_norm": 1.171875, "learning_rate": 6.682039679534884e-05, "loss": 0.1332, "step": 20229 }, { "epoch": 14.820512820512821, "grad_norm": 0.78515625, "learning_rate": 6.681716198327966e-05, "loss": 0.1268, "step": 20230 }, { "epoch": 14.821245421245422, "grad_norm": 0.84765625, "learning_rate": 6.68139270930296e-05, "loss": 0.1166, "step": 20231 }, { "epoch": 14.821978021978023, "grad_norm": 0.8203125, "learning_rate": 6.681069212461418e-05, "loss": 0.1337, "step": 20232 }, { "epoch": 14.822710622710623, "grad_norm": 0.85546875, "learning_rate": 6.680745707804886e-05, "loss": 0.1013, "step": 20233 }, { "epoch": 14.823443223443224, "grad_norm": 0.98828125, "learning_rate": 6.680422195334914e-05, "loss": 0.1349, "step": 20234 }, { "epoch": 14.824175824175825, "grad_norm": 1.4453125, "learning_rate": 6.680098675053055e-05, "loss": 0.1412, "step": 20235 }, { "epoch": 14.824908424908426, "grad_norm": 1.078125, "learning_rate": 6.679775146960859e-05, "loss": 0.1156, "step": 20236 }, { "epoch": 14.825641025641026, "grad_norm": 1.1640625, "learning_rate": 6.679451611059875e-05, "loss": 0.1388, "step": 20237 }, { "epoch": 14.826373626373627, "grad_norm": 1.0546875, "learning_rate": 6.679128067351655e-05, "loss": 0.1316, "step": 20238 }, { "epoch": 14.827106227106228, "grad_norm": 0.8046875, "learning_rate": 6.678804515837746e-05, "loss": 0.1467, "step": 20239 }, { "epoch": 14.827838827838828, "grad_norm": 1.328125, "learning_rate": 6.6784809565197e-05, "loss": 0.1611, "step": 20240 }, { "epoch": 14.82857142857143, "grad_norm": 0.92578125, "learning_rate": 6.678157389399069e-05, "loss": 0.1155, "step": 20241 }, { "epoch": 14.82930402930403, "grad_norm": 0.8515625, "learning_rate": 6.677833814477397e-05, "loss": 0.125, "step": 20242 }, { "epoch": 14.83003663003663, "grad_norm": 1.046875, "learning_rate": 6.677510231756243e-05, "loss": 0.1311, "step": 20243 }, { "epoch": 14.830769230769231, "grad_norm": 1.03125, "learning_rate": 6.677186641237152e-05, "loss": 0.1332, "step": 20244 }, { "epoch": 14.831501831501832, "grad_norm": 1.3671875, "learning_rate": 6.676863042921675e-05, "loss": 0.1094, "step": 20245 }, { "epoch": 14.832234432234433, "grad_norm": 1.3046875, "learning_rate": 6.676539436811365e-05, "loss": 0.1319, "step": 20246 }, { "epoch": 14.832967032967034, "grad_norm": 1.125, "learning_rate": 6.676215822907768e-05, "loss": 0.1365, "step": 20247 }, { "epoch": 14.833699633699634, "grad_norm": 1.09375, "learning_rate": 6.675892201212438e-05, "loss": 0.1199, "step": 20248 }, { "epoch": 14.834432234432235, "grad_norm": 1.1171875, "learning_rate": 6.675568571726924e-05, "loss": 0.1064, "step": 20249 }, { "epoch": 14.835164835164836, "grad_norm": 1.265625, "learning_rate": 6.675244934452779e-05, "loss": 0.1262, "step": 20250 }, { "epoch": 14.835897435897436, "grad_norm": 1.5859375, "learning_rate": 6.674921289391552e-05, "loss": 0.1469, "step": 20251 }, { "epoch": 14.836630036630037, "grad_norm": 0.93359375, "learning_rate": 6.674597636544792e-05, "loss": 0.1417, "step": 20252 }, { "epoch": 14.837362637362638, "grad_norm": 0.9375, "learning_rate": 6.674273975914053e-05, "loss": 0.1256, "step": 20253 }, { "epoch": 14.838095238095239, "grad_norm": 1.1484375, "learning_rate": 6.673950307500884e-05, "loss": 0.135, "step": 20254 }, { "epoch": 14.83882783882784, "grad_norm": 1.109375, "learning_rate": 6.673626631306834e-05, "loss": 0.1164, "step": 20255 }, { "epoch": 14.83956043956044, "grad_norm": 1.2421875, "learning_rate": 6.673302947333458e-05, "loss": 0.1258, "step": 20256 }, { "epoch": 14.84029304029304, "grad_norm": 1.0546875, "learning_rate": 6.672979255582303e-05, "loss": 0.1407, "step": 20257 }, { "epoch": 14.841025641025642, "grad_norm": 1.421875, "learning_rate": 6.672655556054924e-05, "loss": 0.1555, "step": 20258 }, { "epoch": 14.841758241758242, "grad_norm": 0.69921875, "learning_rate": 6.672331848752869e-05, "loss": 0.107, "step": 20259 }, { "epoch": 14.842490842490843, "grad_norm": 1.046875, "learning_rate": 6.672008133677687e-05, "loss": 0.1296, "step": 20260 }, { "epoch": 14.843223443223444, "grad_norm": 1.0625, "learning_rate": 6.671684410830932e-05, "loss": 0.1301, "step": 20261 }, { "epoch": 14.843956043956045, "grad_norm": 1.125, "learning_rate": 6.671360680214158e-05, "loss": 0.1394, "step": 20262 }, { "epoch": 14.844688644688645, "grad_norm": 1.0234375, "learning_rate": 6.671036941828908e-05, "loss": 0.1577, "step": 20263 }, { "epoch": 14.845421245421246, "grad_norm": 0.73828125, "learning_rate": 6.670713195676742e-05, "loss": 0.1192, "step": 20264 }, { "epoch": 14.846153846153847, "grad_norm": 1.1875, "learning_rate": 6.670389441759206e-05, "loss": 0.1267, "step": 20265 }, { "epoch": 14.846886446886447, "grad_norm": 0.953125, "learning_rate": 6.670065680077851e-05, "loss": 0.1398, "step": 20266 }, { "epoch": 14.847619047619048, "grad_norm": 0.97265625, "learning_rate": 6.669741910634231e-05, "loss": 0.1392, "step": 20267 }, { "epoch": 14.848351648351649, "grad_norm": 0.78515625, "learning_rate": 6.669418133429894e-05, "loss": 0.1006, "step": 20268 }, { "epoch": 14.84908424908425, "grad_norm": 1.0234375, "learning_rate": 6.669094348466394e-05, "loss": 0.0937, "step": 20269 }, { "epoch": 14.84981684981685, "grad_norm": 0.91796875, "learning_rate": 6.668770555745282e-05, "loss": 0.1059, "step": 20270 }, { "epoch": 14.850549450549451, "grad_norm": 0.6640625, "learning_rate": 6.66844675526811e-05, "loss": 0.089, "step": 20271 }, { "epoch": 14.851282051282052, "grad_norm": 1.34375, "learning_rate": 6.668122947036425e-05, "loss": 0.1301, "step": 20272 }, { "epoch": 14.852014652014653, "grad_norm": 1.109375, "learning_rate": 6.667799131051784e-05, "loss": 0.0972, "step": 20273 }, { "epoch": 14.852747252747253, "grad_norm": 1.25, "learning_rate": 6.667475307315735e-05, "loss": 0.1122, "step": 20274 }, { "epoch": 14.853479853479854, "grad_norm": 1.4453125, "learning_rate": 6.667151475829832e-05, "loss": 0.1237, "step": 20275 }, { "epoch": 14.854212454212455, "grad_norm": 0.859375, "learning_rate": 6.666827636595624e-05, "loss": 0.0905, "step": 20276 }, { "epoch": 14.854945054945055, "grad_norm": 1.1875, "learning_rate": 6.666503789614667e-05, "loss": 0.1308, "step": 20277 }, { "epoch": 14.855677655677656, "grad_norm": 0.97265625, "learning_rate": 6.666179934888506e-05, "loss": 0.1204, "step": 20278 }, { "epoch": 14.856410256410257, "grad_norm": 1.09375, "learning_rate": 6.6658560724187e-05, "loss": 0.1201, "step": 20279 }, { "epoch": 14.857142857142858, "grad_norm": 0.86328125, "learning_rate": 6.665532202206794e-05, "loss": 0.1131, "step": 20280 }, { "epoch": 14.857875457875458, "grad_norm": 0.90234375, "learning_rate": 6.665208324254344e-05, "loss": 0.1144, "step": 20281 }, { "epoch": 14.858608058608059, "grad_norm": 0.72265625, "learning_rate": 6.6648844385629e-05, "loss": 0.1169, "step": 20282 }, { "epoch": 14.85934065934066, "grad_norm": 1.15625, "learning_rate": 6.664560545134017e-05, "loss": 0.1175, "step": 20283 }, { "epoch": 14.86007326007326, "grad_norm": 1.0546875, "learning_rate": 6.664236643969242e-05, "loss": 0.1304, "step": 20284 }, { "epoch": 14.860805860805861, "grad_norm": 1.7734375, "learning_rate": 6.66391273507013e-05, "loss": 0.1362, "step": 20285 }, { "epoch": 14.861538461538462, "grad_norm": 1.6484375, "learning_rate": 6.663588818438233e-05, "loss": 0.1282, "step": 20286 }, { "epoch": 14.862271062271063, "grad_norm": 0.921875, "learning_rate": 6.663264894075101e-05, "loss": 0.1022, "step": 20287 }, { "epoch": 14.863003663003663, "grad_norm": 0.9140625, "learning_rate": 6.662940961982287e-05, "loss": 0.1112, "step": 20288 }, { "epoch": 14.863736263736264, "grad_norm": 1.046875, "learning_rate": 6.662617022161344e-05, "loss": 0.1347, "step": 20289 }, { "epoch": 14.864468864468865, "grad_norm": 1.484375, "learning_rate": 6.662293074613823e-05, "loss": 0.1226, "step": 20290 }, { "epoch": 14.865201465201466, "grad_norm": 1.375, "learning_rate": 6.661969119341277e-05, "loss": 0.1095, "step": 20291 }, { "epoch": 14.865934065934066, "grad_norm": 0.9453125, "learning_rate": 6.661645156345259e-05, "loss": 0.1007, "step": 20292 }, { "epoch": 14.866666666666667, "grad_norm": 1.671875, "learning_rate": 6.66132118562732e-05, "loss": 0.1404, "step": 20293 }, { "epoch": 14.867399267399268, "grad_norm": 1.03125, "learning_rate": 6.66099720718901e-05, "loss": 0.1554, "step": 20294 }, { "epoch": 14.868131868131869, "grad_norm": 1.265625, "learning_rate": 6.660673221031886e-05, "loss": 0.1496, "step": 20295 }, { "epoch": 14.86886446886447, "grad_norm": 1.875, "learning_rate": 6.660349227157496e-05, "loss": 0.1406, "step": 20296 }, { "epoch": 14.86959706959707, "grad_norm": 1.796875, "learning_rate": 6.660025225567394e-05, "loss": 0.1442, "step": 20297 }, { "epoch": 14.87032967032967, "grad_norm": 0.96484375, "learning_rate": 6.659701216263135e-05, "loss": 0.1262, "step": 20298 }, { "epoch": 14.871062271062272, "grad_norm": 1.375, "learning_rate": 6.659377199246266e-05, "loss": 0.1258, "step": 20299 }, { "epoch": 14.871794871794872, "grad_norm": 0.80859375, "learning_rate": 6.659053174518343e-05, "loss": 0.1015, "step": 20300 }, { "epoch": 14.872527472527473, "grad_norm": 1.0625, "learning_rate": 6.65872914208092e-05, "loss": 0.1215, "step": 20301 }, { "epoch": 14.873260073260074, "grad_norm": 0.92578125, "learning_rate": 6.658405101935546e-05, "loss": 0.1223, "step": 20302 }, { "epoch": 14.873992673992674, "grad_norm": 0.90234375, "learning_rate": 6.658081054083777e-05, "loss": 0.1476, "step": 20303 }, { "epoch": 14.874725274725275, "grad_norm": 1.109375, "learning_rate": 6.657756998527161e-05, "loss": 0.1247, "step": 20304 }, { "epoch": 14.875457875457876, "grad_norm": 1.46875, "learning_rate": 6.657432935267255e-05, "loss": 0.1325, "step": 20305 }, { "epoch": 14.876190476190477, "grad_norm": 1.7578125, "learning_rate": 6.657108864305611e-05, "loss": 0.1216, "step": 20306 }, { "epoch": 14.876923076923077, "grad_norm": 0.921875, "learning_rate": 6.65678478564378e-05, "loss": 0.1466, "step": 20307 }, { "epoch": 14.877655677655678, "grad_norm": 0.8203125, "learning_rate": 6.656460699283315e-05, "loss": 0.1, "step": 20308 }, { "epoch": 14.878388278388279, "grad_norm": 1.03125, "learning_rate": 6.65613660522577e-05, "loss": 0.1341, "step": 20309 }, { "epoch": 14.87912087912088, "grad_norm": 1.0625, "learning_rate": 6.655812503472697e-05, "loss": 0.1323, "step": 20310 }, { "epoch": 14.87985347985348, "grad_norm": 0.7421875, "learning_rate": 6.65548839402565e-05, "loss": 0.1084, "step": 20311 }, { "epoch": 14.880586080586081, "grad_norm": 0.75390625, "learning_rate": 6.65516427688618e-05, "loss": 0.1217, "step": 20312 }, { "epoch": 14.881318681318682, "grad_norm": 0.76171875, "learning_rate": 6.65484015205584e-05, "loss": 0.1128, "step": 20313 }, { "epoch": 14.882051282051282, "grad_norm": 0.875, "learning_rate": 6.654516019536187e-05, "loss": 0.1105, "step": 20314 }, { "epoch": 14.882783882783883, "grad_norm": 1.484375, "learning_rate": 6.654191879328768e-05, "loss": 0.1524, "step": 20315 }, { "epoch": 14.883516483516484, "grad_norm": 1.125, "learning_rate": 6.653867731435142e-05, "loss": 0.127, "step": 20316 }, { "epoch": 14.884249084249085, "grad_norm": 1.15625, "learning_rate": 6.653543575856855e-05, "loss": 0.1498, "step": 20317 }, { "epoch": 14.884981684981685, "grad_norm": 0.8125, "learning_rate": 6.653219412595468e-05, "loss": 0.1259, "step": 20318 }, { "epoch": 14.885714285714286, "grad_norm": 1.1953125, "learning_rate": 6.652895241652531e-05, "loss": 0.1467, "step": 20319 }, { "epoch": 14.886446886446887, "grad_norm": 0.7265625, "learning_rate": 6.652571063029595e-05, "loss": 0.0968, "step": 20320 }, { "epoch": 14.887179487179488, "grad_norm": 1.8515625, "learning_rate": 6.652246876728214e-05, "loss": 0.1523, "step": 20321 }, { "epoch": 14.887912087912088, "grad_norm": 1.140625, "learning_rate": 6.651922682749944e-05, "loss": 0.1165, "step": 20322 }, { "epoch": 14.888644688644689, "grad_norm": 1.171875, "learning_rate": 6.651598481096336e-05, "loss": 0.1556, "step": 20323 }, { "epoch": 14.88937728937729, "grad_norm": 1.3203125, "learning_rate": 6.651274271768944e-05, "loss": 0.1135, "step": 20324 }, { "epoch": 14.89010989010989, "grad_norm": 1.1328125, "learning_rate": 6.650950054769322e-05, "loss": 0.1283, "step": 20325 }, { "epoch": 14.890842490842491, "grad_norm": 1.3984375, "learning_rate": 6.650625830099019e-05, "loss": 0.1355, "step": 20326 }, { "epoch": 14.891575091575092, "grad_norm": 1.03125, "learning_rate": 6.650301597759597e-05, "loss": 0.1217, "step": 20327 }, { "epoch": 14.892307692307693, "grad_norm": 0.796875, "learning_rate": 6.649977357752603e-05, "loss": 0.1203, "step": 20328 }, { "epoch": 14.893040293040293, "grad_norm": 1.5390625, "learning_rate": 6.64965311007959e-05, "loss": 0.1219, "step": 20329 }, { "epoch": 14.893772893772894, "grad_norm": 1.328125, "learning_rate": 6.649328854742116e-05, "loss": 0.1343, "step": 20330 }, { "epoch": 14.894505494505495, "grad_norm": 1.1015625, "learning_rate": 6.649004591741734e-05, "loss": 0.1366, "step": 20331 }, { "epoch": 14.895238095238096, "grad_norm": 1.34375, "learning_rate": 6.648680321079993e-05, "loss": 0.109, "step": 20332 }, { "epoch": 14.895970695970696, "grad_norm": 1.7578125, "learning_rate": 6.648356042758451e-05, "loss": 0.1438, "step": 20333 }, { "epoch": 14.896703296703297, "grad_norm": 1.796875, "learning_rate": 6.648031756778662e-05, "loss": 0.1393, "step": 20334 }, { "epoch": 14.897435897435898, "grad_norm": 1.0703125, "learning_rate": 6.647707463142176e-05, "loss": 0.1097, "step": 20335 }, { "epoch": 14.898168498168499, "grad_norm": 1.3984375, "learning_rate": 6.647383161850552e-05, "loss": 0.1553, "step": 20336 }, { "epoch": 14.8989010989011, "grad_norm": 0.83203125, "learning_rate": 6.647058852905335e-05, "loss": 0.1443, "step": 20337 }, { "epoch": 14.8996336996337, "grad_norm": 2.046875, "learning_rate": 6.64673453630809e-05, "loss": 0.1608, "step": 20338 }, { "epoch": 14.9003663003663, "grad_norm": 1.78125, "learning_rate": 6.646410212060363e-05, "loss": 0.1417, "step": 20339 }, { "epoch": 14.901098901098901, "grad_norm": 1.6328125, "learning_rate": 6.646085880163713e-05, "loss": 0.1654, "step": 20340 }, { "epoch": 14.901831501831502, "grad_norm": 0.921875, "learning_rate": 6.64576154061969e-05, "loss": 0.1112, "step": 20341 }, { "epoch": 14.902564102564103, "grad_norm": 1.0859375, "learning_rate": 6.645437193429852e-05, "loss": 0.1082, "step": 20342 }, { "epoch": 14.903296703296704, "grad_norm": 0.88671875, "learning_rate": 6.645112838595747e-05, "loss": 0.1129, "step": 20343 }, { "epoch": 14.904029304029304, "grad_norm": 1.296875, "learning_rate": 6.644788476118934e-05, "loss": 0.1119, "step": 20344 }, { "epoch": 14.904761904761905, "grad_norm": 1.6953125, "learning_rate": 6.644464106000967e-05, "loss": 0.1516, "step": 20345 }, { "epoch": 14.905494505494506, "grad_norm": 1.78125, "learning_rate": 6.644139728243398e-05, "loss": 0.1054, "step": 20346 }, { "epoch": 14.906227106227107, "grad_norm": 0.8359375, "learning_rate": 6.643815342847783e-05, "loss": 0.1212, "step": 20347 }, { "epoch": 14.906959706959707, "grad_norm": 1.4453125, "learning_rate": 6.643490949815675e-05, "loss": 0.1166, "step": 20348 }, { "epoch": 14.907692307692308, "grad_norm": 1.4453125, "learning_rate": 6.643166549148629e-05, "loss": 0.1324, "step": 20349 }, { "epoch": 14.908424908424909, "grad_norm": 1.484375, "learning_rate": 6.642842140848198e-05, "loss": 0.1692, "step": 20350 }, { "epoch": 14.90915750915751, "grad_norm": 0.84375, "learning_rate": 6.642517724915939e-05, "loss": 0.1136, "step": 20351 }, { "epoch": 14.90989010989011, "grad_norm": 1.0546875, "learning_rate": 6.642193301353403e-05, "loss": 0.11, "step": 20352 }, { "epoch": 14.910622710622711, "grad_norm": 0.8984375, "learning_rate": 6.641868870162148e-05, "loss": 0.1225, "step": 20353 }, { "epoch": 14.911355311355312, "grad_norm": 1.2890625, "learning_rate": 6.641544431343726e-05, "loss": 0.1297, "step": 20354 }, { "epoch": 14.912087912087912, "grad_norm": 1.0625, "learning_rate": 6.641219984899693e-05, "loss": 0.1228, "step": 20355 }, { "epoch": 14.912820512820513, "grad_norm": 0.78515625, "learning_rate": 6.640895530831599e-05, "loss": 0.1246, "step": 20356 }, { "epoch": 14.913553113553114, "grad_norm": 1.2109375, "learning_rate": 6.640571069141006e-05, "loss": 0.1274, "step": 20357 }, { "epoch": 14.914285714285715, "grad_norm": 1.4765625, "learning_rate": 6.640246599829465e-05, "loss": 0.0979, "step": 20358 }, { "epoch": 14.915018315018315, "grad_norm": 0.69921875, "learning_rate": 6.639922122898528e-05, "loss": 0.1253, "step": 20359 }, { "epoch": 14.915750915750916, "grad_norm": 0.90625, "learning_rate": 6.639597638349755e-05, "loss": 0.1254, "step": 20360 }, { "epoch": 14.916483516483517, "grad_norm": 0.98828125, "learning_rate": 6.639273146184695e-05, "loss": 0.1088, "step": 20361 }, { "epoch": 14.917216117216118, "grad_norm": 1.4296875, "learning_rate": 6.638948646404907e-05, "loss": 0.1406, "step": 20362 }, { "epoch": 14.917948717948718, "grad_norm": 0.9921875, "learning_rate": 6.638624139011945e-05, "loss": 0.1004, "step": 20363 }, { "epoch": 14.918681318681319, "grad_norm": 0.75390625, "learning_rate": 6.638299624007362e-05, "loss": 0.0972, "step": 20364 }, { "epoch": 14.91941391941392, "grad_norm": 0.78125, "learning_rate": 6.637975101392715e-05, "loss": 0.1171, "step": 20365 }, { "epoch": 14.92014652014652, "grad_norm": 0.86328125, "learning_rate": 6.637650571169559e-05, "loss": 0.1147, "step": 20366 }, { "epoch": 14.920879120879121, "grad_norm": 0.91015625, "learning_rate": 6.637326033339446e-05, "loss": 0.1427, "step": 20367 }, { "epoch": 14.921611721611722, "grad_norm": 0.59765625, "learning_rate": 6.637001487903933e-05, "loss": 0.1069, "step": 20368 }, { "epoch": 14.922344322344323, "grad_norm": 1.046875, "learning_rate": 6.636676934864575e-05, "loss": 0.1269, "step": 20369 }, { "epoch": 14.923076923076923, "grad_norm": 0.86328125, "learning_rate": 6.636352374222929e-05, "loss": 0.1138, "step": 20370 }, { "epoch": 14.923809523809524, "grad_norm": 1.046875, "learning_rate": 6.636027805980547e-05, "loss": 0.1181, "step": 20371 }, { "epoch": 14.924542124542125, "grad_norm": 0.9765625, "learning_rate": 6.635703230138985e-05, "loss": 0.121, "step": 20372 }, { "epoch": 14.925274725274726, "grad_norm": 0.96484375, "learning_rate": 6.635378646699796e-05, "loss": 0.1431, "step": 20373 }, { "epoch": 14.926007326007326, "grad_norm": 0.70703125, "learning_rate": 6.635054055664538e-05, "loss": 0.128, "step": 20374 }, { "epoch": 14.926739926739927, "grad_norm": 1.25, "learning_rate": 6.63472945703477e-05, "loss": 0.1053, "step": 20375 }, { "epoch": 14.927472527472528, "grad_norm": 1.1484375, "learning_rate": 6.634404850812039e-05, "loss": 0.1024, "step": 20376 }, { "epoch": 14.928205128205128, "grad_norm": 0.61328125, "learning_rate": 6.634080236997904e-05, "loss": 0.104, "step": 20377 }, { "epoch": 14.92893772893773, "grad_norm": 1.359375, "learning_rate": 6.633755615593921e-05, "loss": 0.1391, "step": 20378 }, { "epoch": 14.92967032967033, "grad_norm": 1.015625, "learning_rate": 6.633430986601648e-05, "loss": 0.1144, "step": 20379 }, { "epoch": 14.93040293040293, "grad_norm": 1.8359375, "learning_rate": 6.633106350022634e-05, "loss": 0.1448, "step": 20380 }, { "epoch": 14.931135531135531, "grad_norm": 1.046875, "learning_rate": 6.632781705858438e-05, "loss": 0.1005, "step": 20381 }, { "epoch": 14.931868131868132, "grad_norm": 0.86328125, "learning_rate": 6.632457054110615e-05, "loss": 0.1341, "step": 20382 }, { "epoch": 14.932600732600733, "grad_norm": 0.9765625, "learning_rate": 6.632132394780724e-05, "loss": 0.1197, "step": 20383 }, { "epoch": 14.933333333333334, "grad_norm": 1.578125, "learning_rate": 6.631807727870316e-05, "loss": 0.1301, "step": 20384 }, { "epoch": 14.934065934065934, "grad_norm": 0.9453125, "learning_rate": 6.631483053380945e-05, "loss": 0.126, "step": 20385 }, { "epoch": 14.934798534798535, "grad_norm": 1.0546875, "learning_rate": 6.631158371314173e-05, "loss": 0.1293, "step": 20386 }, { "epoch": 14.935531135531136, "grad_norm": 1.5859375, "learning_rate": 6.630833681671549e-05, "loss": 0.119, "step": 20387 }, { "epoch": 14.936263736263736, "grad_norm": 0.9765625, "learning_rate": 6.630508984454634e-05, "loss": 0.1293, "step": 20388 }, { "epoch": 14.936996336996337, "grad_norm": 1.125, "learning_rate": 6.630184279664981e-05, "loss": 0.1237, "step": 20389 }, { "epoch": 14.937728937728938, "grad_norm": 0.765625, "learning_rate": 6.629859567304146e-05, "loss": 0.1112, "step": 20390 }, { "epoch": 14.938461538461539, "grad_norm": 0.9921875, "learning_rate": 6.629534847373686e-05, "loss": 0.1087, "step": 20391 }, { "epoch": 14.93919413919414, "grad_norm": 1.578125, "learning_rate": 6.629210119875156e-05, "loss": 0.1205, "step": 20392 }, { "epoch": 14.93992673992674, "grad_norm": 0.9921875, "learning_rate": 6.628885384810112e-05, "loss": 0.1067, "step": 20393 }, { "epoch": 14.94065934065934, "grad_norm": 0.8125, "learning_rate": 6.62856064218011e-05, "loss": 0.1084, "step": 20394 }, { "epoch": 14.941391941391942, "grad_norm": 1.5390625, "learning_rate": 6.628235891986705e-05, "loss": 0.1296, "step": 20395 }, { "epoch": 14.942124542124542, "grad_norm": 1.328125, "learning_rate": 6.627911134231455e-05, "loss": 0.1154, "step": 20396 }, { "epoch": 14.942857142857143, "grad_norm": 0.99609375, "learning_rate": 6.627586368915913e-05, "loss": 0.1075, "step": 20397 }, { "epoch": 14.943589743589744, "grad_norm": 0.5703125, "learning_rate": 6.627261596041636e-05, "loss": 0.1158, "step": 20398 }, { "epoch": 14.944322344322345, "grad_norm": 1.078125, "learning_rate": 6.626936815610183e-05, "loss": 0.1123, "step": 20399 }, { "epoch": 14.945054945054945, "grad_norm": 0.921875, "learning_rate": 6.626612027623108e-05, "loss": 0.1366, "step": 20400 }, { "epoch": 14.945787545787546, "grad_norm": 1.4765625, "learning_rate": 6.626287232081967e-05, "loss": 0.1321, "step": 20401 }, { "epoch": 14.946520146520147, "grad_norm": 1.0, "learning_rate": 6.625962428988317e-05, "loss": 0.1217, "step": 20402 }, { "epoch": 14.947252747252747, "grad_norm": 1.140625, "learning_rate": 6.625637618343711e-05, "loss": 0.1108, "step": 20403 }, { "epoch": 14.947985347985348, "grad_norm": 0.97265625, "learning_rate": 6.625312800149711e-05, "loss": 0.1303, "step": 20404 }, { "epoch": 14.948717948717949, "grad_norm": 0.73828125, "learning_rate": 6.624987974407869e-05, "loss": 0.1135, "step": 20405 }, { "epoch": 14.94945054945055, "grad_norm": 1.8515625, "learning_rate": 6.624663141119742e-05, "loss": 0.1431, "step": 20406 }, { "epoch": 14.95018315018315, "grad_norm": 1.28125, "learning_rate": 6.624338300286888e-05, "loss": 0.1413, "step": 20407 }, { "epoch": 14.950915750915751, "grad_norm": 0.9453125, "learning_rate": 6.624013451910861e-05, "loss": 0.1447, "step": 20408 }, { "epoch": 14.951648351648352, "grad_norm": 1.125, "learning_rate": 6.62368859599322e-05, "loss": 0.1314, "step": 20409 }, { "epoch": 14.952380952380953, "grad_norm": 1.0703125, "learning_rate": 6.62336373253552e-05, "loss": 0.13, "step": 20410 }, { "epoch": 14.953113553113553, "grad_norm": 0.98828125, "learning_rate": 6.623038861539318e-05, "loss": 0.1226, "step": 20411 }, { "epoch": 14.953846153846154, "grad_norm": 0.7734375, "learning_rate": 6.62271398300617e-05, "loss": 0.1258, "step": 20412 }, { "epoch": 14.954578754578755, "grad_norm": 0.73046875, "learning_rate": 6.622389096937634e-05, "loss": 0.0998, "step": 20413 }, { "epoch": 14.955311355311355, "grad_norm": 1.2578125, "learning_rate": 6.622064203335264e-05, "loss": 0.1278, "step": 20414 }, { "epoch": 14.956043956043956, "grad_norm": 1.265625, "learning_rate": 6.62173930220062e-05, "loss": 0.1227, "step": 20415 }, { "epoch": 14.956776556776557, "grad_norm": 0.74609375, "learning_rate": 6.621414393535254e-05, "loss": 0.1063, "step": 20416 }, { "epoch": 14.957509157509158, "grad_norm": 1.46875, "learning_rate": 6.62108947734073e-05, "loss": 0.1381, "step": 20417 }, { "epoch": 14.958241758241758, "grad_norm": 1.578125, "learning_rate": 6.620764553618599e-05, "loss": 0.1073, "step": 20418 }, { "epoch": 14.95897435897436, "grad_norm": 2.0625, "learning_rate": 6.620439622370417e-05, "loss": 0.1095, "step": 20419 }, { "epoch": 14.95970695970696, "grad_norm": 1.015625, "learning_rate": 6.620114683597747e-05, "loss": 0.1064, "step": 20420 }, { "epoch": 14.96043956043956, "grad_norm": 1.8359375, "learning_rate": 6.61978973730214e-05, "loss": 0.1308, "step": 20421 }, { "epoch": 14.961172161172161, "grad_norm": 1.0390625, "learning_rate": 6.619464783485156e-05, "loss": 0.1143, "step": 20422 }, { "epoch": 14.961904761904762, "grad_norm": 1.6171875, "learning_rate": 6.619139822148352e-05, "loss": 0.1475, "step": 20423 }, { "epoch": 14.962637362637363, "grad_norm": 0.77734375, "learning_rate": 6.61881485329328e-05, "loss": 0.0908, "step": 20424 }, { "epoch": 14.963369963369964, "grad_norm": 0.93359375, "learning_rate": 6.618489876921505e-05, "loss": 0.0999, "step": 20425 }, { "epoch": 14.964102564102564, "grad_norm": 0.87109375, "learning_rate": 6.618164893034577e-05, "loss": 0.1069, "step": 20426 }, { "epoch": 14.964835164835165, "grad_norm": 0.828125, "learning_rate": 6.617839901634059e-05, "loss": 0.1203, "step": 20427 }, { "epoch": 14.965567765567766, "grad_norm": 0.83203125, "learning_rate": 6.617514902721503e-05, "loss": 0.1302, "step": 20428 }, { "epoch": 14.966300366300366, "grad_norm": 1.1015625, "learning_rate": 6.617189896298469e-05, "loss": 0.1101, "step": 20429 }, { "epoch": 14.967032967032967, "grad_norm": 0.89453125, "learning_rate": 6.616864882366516e-05, "loss": 0.1174, "step": 20430 }, { "epoch": 14.967765567765568, "grad_norm": 1.4609375, "learning_rate": 6.616539860927196e-05, "loss": 0.195, "step": 20431 }, { "epoch": 14.968498168498169, "grad_norm": 1.265625, "learning_rate": 6.616214831982071e-05, "loss": 0.1702, "step": 20432 }, { "epoch": 14.96923076923077, "grad_norm": 0.9609375, "learning_rate": 6.615889795532696e-05, "loss": 0.1233, "step": 20433 }, { "epoch": 14.96996336996337, "grad_norm": 0.7421875, "learning_rate": 6.615564751580628e-05, "loss": 0.1455, "step": 20434 }, { "epoch": 14.97069597069597, "grad_norm": 1.9375, "learning_rate": 6.615239700127428e-05, "loss": 0.1181, "step": 20435 }, { "epoch": 14.971428571428572, "grad_norm": 1.7734375, "learning_rate": 6.614914641174649e-05, "loss": 0.1145, "step": 20436 }, { "epoch": 14.972161172161172, "grad_norm": 1.25, "learning_rate": 6.61458957472385e-05, "loss": 0.1155, "step": 20437 }, { "epoch": 14.972893772893773, "grad_norm": 0.89453125, "learning_rate": 6.61426450077659e-05, "loss": 0.1052, "step": 20438 }, { "epoch": 14.973626373626374, "grad_norm": 1.4765625, "learning_rate": 6.613939419334424e-05, "loss": 0.1295, "step": 20439 }, { "epoch": 14.974358974358974, "grad_norm": 1.25, "learning_rate": 6.613614330398911e-05, "loss": 0.157, "step": 20440 }, { "epoch": 14.975091575091575, "grad_norm": 1.21875, "learning_rate": 6.613289233971609e-05, "loss": 0.1267, "step": 20441 }, { "epoch": 14.975824175824176, "grad_norm": 0.7265625, "learning_rate": 6.612964130054075e-05, "loss": 0.1366, "step": 20442 }, { "epoch": 14.976556776556777, "grad_norm": 1.3046875, "learning_rate": 6.612639018647866e-05, "loss": 0.1186, "step": 20443 }, { "epoch": 14.977289377289377, "grad_norm": 1.4140625, "learning_rate": 6.612313899754542e-05, "loss": 0.1269, "step": 20444 }, { "epoch": 14.978021978021978, "grad_norm": 1.1640625, "learning_rate": 6.611988773375656e-05, "loss": 0.1122, "step": 20445 }, { "epoch": 14.978754578754579, "grad_norm": 0.84375, "learning_rate": 6.611663639512773e-05, "loss": 0.0963, "step": 20446 }, { "epoch": 14.97948717948718, "grad_norm": 1.109375, "learning_rate": 6.611338498167445e-05, "loss": 0.1467, "step": 20447 }, { "epoch": 14.98021978021978, "grad_norm": 1.15625, "learning_rate": 6.611013349341232e-05, "loss": 0.133, "step": 20448 }, { "epoch": 14.980952380952381, "grad_norm": 0.9765625, "learning_rate": 6.610688193035692e-05, "loss": 0.1003, "step": 20449 }, { "epoch": 14.981684981684982, "grad_norm": 1.1953125, "learning_rate": 6.610363029252381e-05, "loss": 0.1386, "step": 20450 }, { "epoch": 14.982417582417582, "grad_norm": 1.1328125, "learning_rate": 6.61003785799286e-05, "loss": 0.1336, "step": 20451 }, { "epoch": 14.983150183150183, "grad_norm": 1.25, "learning_rate": 6.609712679258685e-05, "loss": 0.1416, "step": 20452 }, { "epoch": 14.983882783882784, "grad_norm": 1.390625, "learning_rate": 6.609387493051414e-05, "loss": 0.1134, "step": 20453 }, { "epoch": 14.984615384615385, "grad_norm": 0.96875, "learning_rate": 6.609062299372606e-05, "loss": 0.0916, "step": 20454 }, { "epoch": 14.985347985347985, "grad_norm": 1.2109375, "learning_rate": 6.608737098223818e-05, "loss": 0.1547, "step": 20455 }, { "epoch": 14.986080586080586, "grad_norm": 0.73828125, "learning_rate": 6.60841188960661e-05, "loss": 0.1098, "step": 20456 }, { "epoch": 14.986813186813187, "grad_norm": 1.0390625, "learning_rate": 6.60808667352254e-05, "loss": 0.1293, "step": 20457 }, { "epoch": 14.987545787545788, "grad_norm": 1.28125, "learning_rate": 6.607761449973163e-05, "loss": 0.1548, "step": 20458 }, { "epoch": 14.988278388278388, "grad_norm": 0.91796875, "learning_rate": 6.60743621896004e-05, "loss": 0.1432, "step": 20459 }, { "epoch": 14.989010989010989, "grad_norm": 0.75, "learning_rate": 6.60711098048473e-05, "loss": 0.1252, "step": 20460 }, { "epoch": 14.98974358974359, "grad_norm": 0.89453125, "learning_rate": 6.606785734548789e-05, "loss": 0.0997, "step": 20461 }, { "epoch": 14.99047619047619, "grad_norm": 1.1640625, "learning_rate": 6.606460481153778e-05, "loss": 0.154, "step": 20462 }, { "epoch": 14.991208791208791, "grad_norm": 0.6875, "learning_rate": 6.606135220301253e-05, "loss": 0.103, "step": 20463 }, { "epoch": 14.991941391941392, "grad_norm": 1.0859375, "learning_rate": 6.605809951992772e-05, "loss": 0.1498, "step": 20464 }, { "epoch": 14.992673992673993, "grad_norm": 1.46875, "learning_rate": 6.605484676229897e-05, "loss": 0.1278, "step": 20465 }, { "epoch": 14.993406593406593, "grad_norm": 0.89453125, "learning_rate": 6.605159393014182e-05, "loss": 0.1198, "step": 20466 }, { "epoch": 14.994139194139194, "grad_norm": 1.75, "learning_rate": 6.604834102347189e-05, "loss": 0.1511, "step": 20467 }, { "epoch": 14.994871794871795, "grad_norm": 1.2421875, "learning_rate": 6.604508804230477e-05, "loss": 0.1588, "step": 20468 }, { "epoch": 14.995604395604396, "grad_norm": 0.91015625, "learning_rate": 6.604183498665599e-05, "loss": 0.1534, "step": 20469 }, { "epoch": 14.996336996336996, "grad_norm": 1.03125, "learning_rate": 6.603858185654121e-05, "loss": 0.0947, "step": 20470 }, { "epoch": 14.997069597069597, "grad_norm": 0.8359375, "learning_rate": 6.603532865197595e-05, "loss": 0.1188, "step": 20471 }, { "epoch": 14.997802197802198, "grad_norm": 1.2109375, "learning_rate": 6.603207537297588e-05, "loss": 0.1173, "step": 20472 }, { "epoch": 14.998534798534799, "grad_norm": 1.421875, "learning_rate": 6.602882201955651e-05, "loss": 0.1307, "step": 20473 }, { "epoch": 14.9992673992674, "grad_norm": 1.0546875, "learning_rate": 6.602556859173345e-05, "loss": 0.1118, "step": 20474 }, { "epoch": 15.0, "grad_norm": 1.0703125, "learning_rate": 6.602231508952232e-05, "loss": 0.1396, "step": 20475 }, { "epoch": 15.0007326007326, "grad_norm": 2.125, "learning_rate": 6.601906151293866e-05, "loss": 0.1278, "step": 20476 }, { "epoch": 15.001465201465201, "grad_norm": 0.8828125, "learning_rate": 6.601580786199808e-05, "loss": 0.1173, "step": 20477 }, { "epoch": 15.002197802197802, "grad_norm": 1.6328125, "learning_rate": 6.601255413671617e-05, "loss": 0.134, "step": 20478 }, { "epoch": 15.002930402930403, "grad_norm": 0.92578125, "learning_rate": 6.600930033710855e-05, "loss": 0.127, "step": 20479 }, { "epoch": 15.003663003663004, "grad_norm": 1.921875, "learning_rate": 6.600604646319073e-05, "loss": 0.1239, "step": 20480 }, { "epoch": 15.004395604395604, "grad_norm": 1.8515625, "learning_rate": 6.600279251497839e-05, "loss": 0.1526, "step": 20481 }, { "epoch": 15.005128205128205, "grad_norm": 1.5546875, "learning_rate": 6.599953849248708e-05, "loss": 0.1211, "step": 20482 }, { "epoch": 15.005860805860806, "grad_norm": 0.98046875, "learning_rate": 6.599628439573238e-05, "loss": 0.1072, "step": 20483 }, { "epoch": 15.006593406593407, "grad_norm": 0.90234375, "learning_rate": 6.599303022472991e-05, "loss": 0.0972, "step": 20484 }, { "epoch": 15.007326007326007, "grad_norm": 0.6640625, "learning_rate": 6.598977597949524e-05, "loss": 0.1021, "step": 20485 }, { "epoch": 15.008058608058608, "grad_norm": 1.5625, "learning_rate": 6.598652166004395e-05, "loss": 0.1368, "step": 20486 }, { "epoch": 15.008791208791209, "grad_norm": 0.71484375, "learning_rate": 6.598326726639168e-05, "loss": 0.0939, "step": 20487 }, { "epoch": 15.00952380952381, "grad_norm": 0.9375, "learning_rate": 6.598001279855397e-05, "loss": 0.1219, "step": 20488 }, { "epoch": 15.01025641025641, "grad_norm": 1.328125, "learning_rate": 6.597675825654644e-05, "loss": 0.1433, "step": 20489 }, { "epoch": 15.010989010989011, "grad_norm": 1.3515625, "learning_rate": 6.597350364038469e-05, "loss": 0.1375, "step": 20490 }, { "epoch": 15.011721611721612, "grad_norm": 1.6328125, "learning_rate": 6.59702489500843e-05, "loss": 0.141, "step": 20491 }, { "epoch": 15.012454212454212, "grad_norm": 0.84375, "learning_rate": 6.596699418566088e-05, "loss": 0.1086, "step": 20492 }, { "epoch": 15.013186813186813, "grad_norm": 1.4140625, "learning_rate": 6.596373934712999e-05, "loss": 0.1131, "step": 20493 }, { "epoch": 15.013919413919414, "grad_norm": 1.2421875, "learning_rate": 6.596048443450725e-05, "loss": 0.1198, "step": 20494 }, { "epoch": 15.014652014652015, "grad_norm": 1.5703125, "learning_rate": 6.595722944780828e-05, "loss": 0.1143, "step": 20495 }, { "epoch": 15.015384615384615, "grad_norm": 0.8046875, "learning_rate": 6.595397438704864e-05, "loss": 0.1244, "step": 20496 }, { "epoch": 15.016117216117216, "grad_norm": 1.28125, "learning_rate": 6.595071925224391e-05, "loss": 0.1452, "step": 20497 }, { "epoch": 15.016849816849817, "grad_norm": 1.3203125, "learning_rate": 6.594746404340973e-05, "loss": 0.0989, "step": 20498 }, { "epoch": 15.017582417582418, "grad_norm": 0.8203125, "learning_rate": 6.594420876056168e-05, "loss": 0.1192, "step": 20499 }, { "epoch": 15.018315018315018, "grad_norm": 1.3125, "learning_rate": 6.594095340371536e-05, "loss": 0.1191, "step": 20500 }, { "epoch": 15.019047619047619, "grad_norm": 0.71875, "learning_rate": 6.593769797288637e-05, "loss": 0.129, "step": 20501 }, { "epoch": 15.01978021978022, "grad_norm": 1.5078125, "learning_rate": 6.593444246809028e-05, "loss": 0.1429, "step": 20502 }, { "epoch": 15.02051282051282, "grad_norm": 1.078125, "learning_rate": 6.593118688934272e-05, "loss": 0.1032, "step": 20503 }, { "epoch": 15.021245421245421, "grad_norm": 2.3125, "learning_rate": 6.592793123665929e-05, "loss": 0.167, "step": 20504 }, { "epoch": 15.021978021978022, "grad_norm": 0.8671875, "learning_rate": 6.592467551005554e-05, "loss": 0.1168, "step": 20505 }, { "epoch": 15.022710622710623, "grad_norm": 0.8828125, "learning_rate": 6.592141970954715e-05, "loss": 0.1185, "step": 20506 }, { "epoch": 15.023443223443223, "grad_norm": 1.40625, "learning_rate": 6.591816383514963e-05, "loss": 0.1039, "step": 20507 }, { "epoch": 15.024175824175824, "grad_norm": 1.7421875, "learning_rate": 6.591490788687867e-05, "loss": 0.1374, "step": 20508 }, { "epoch": 15.024908424908425, "grad_norm": 1.2109375, "learning_rate": 6.591165186474981e-05, "loss": 0.1428, "step": 20509 }, { "epoch": 15.025641025641026, "grad_norm": 1.0703125, "learning_rate": 6.590839576877868e-05, "loss": 0.1275, "step": 20510 }, { "epoch": 15.026373626373626, "grad_norm": 1.0390625, "learning_rate": 6.590513959898085e-05, "loss": 0.1251, "step": 20511 }, { "epoch": 15.027106227106227, "grad_norm": 0.8671875, "learning_rate": 6.590188335537191e-05, "loss": 0.1099, "step": 20512 }, { "epoch": 15.027838827838828, "grad_norm": 1.359375, "learning_rate": 6.589862703796754e-05, "loss": 0.1474, "step": 20513 }, { "epoch": 15.028571428571428, "grad_norm": 0.80078125, "learning_rate": 6.589537064678328e-05, "loss": 0.0906, "step": 20514 }, { "epoch": 15.02930402930403, "grad_norm": 0.69921875, "learning_rate": 6.589211418183474e-05, "loss": 0.0951, "step": 20515 }, { "epoch": 15.03003663003663, "grad_norm": 0.83984375, "learning_rate": 6.588885764313753e-05, "loss": 0.1258, "step": 20516 }, { "epoch": 15.03076923076923, "grad_norm": 0.76171875, "learning_rate": 6.588560103070724e-05, "loss": 0.1194, "step": 20517 }, { "epoch": 15.031501831501831, "grad_norm": 0.76953125, "learning_rate": 6.588234434455949e-05, "loss": 0.1183, "step": 20518 }, { "epoch": 15.032234432234432, "grad_norm": 1.171875, "learning_rate": 6.587908758470988e-05, "loss": 0.1477, "step": 20519 }, { "epoch": 15.032967032967033, "grad_norm": 0.78515625, "learning_rate": 6.587583075117402e-05, "loss": 0.1013, "step": 20520 }, { "epoch": 15.033699633699634, "grad_norm": 0.71875, "learning_rate": 6.58725738439675e-05, "loss": 0.1149, "step": 20521 }, { "epoch": 15.034432234432234, "grad_norm": 0.98046875, "learning_rate": 6.586931686310593e-05, "loss": 0.1369, "step": 20522 }, { "epoch": 15.035164835164835, "grad_norm": 0.77734375, "learning_rate": 6.586605980860492e-05, "loss": 0.1122, "step": 20523 }, { "epoch": 15.035897435897436, "grad_norm": 0.94140625, "learning_rate": 6.58628026804801e-05, "loss": 0.1365, "step": 20524 }, { "epoch": 15.036630036630036, "grad_norm": 0.78125, "learning_rate": 6.5859545478747e-05, "loss": 0.1093, "step": 20525 }, { "epoch": 15.037362637362637, "grad_norm": 1.359375, "learning_rate": 6.585628820342131e-05, "loss": 0.1412, "step": 20526 }, { "epoch": 15.038095238095238, "grad_norm": 1.171875, "learning_rate": 6.585303085451859e-05, "loss": 0.1396, "step": 20527 }, { "epoch": 15.038827838827839, "grad_norm": 0.63671875, "learning_rate": 6.584977343205446e-05, "loss": 0.0824, "step": 20528 }, { "epoch": 15.03956043956044, "grad_norm": 1.4765625, "learning_rate": 6.584651593604454e-05, "loss": 0.1181, "step": 20529 }, { "epoch": 15.04029304029304, "grad_norm": 1.0234375, "learning_rate": 6.58432583665044e-05, "loss": 0.1169, "step": 20530 }, { "epoch": 15.04102564102564, "grad_norm": 1.3046875, "learning_rate": 6.584000072344968e-05, "loss": 0.1157, "step": 20531 }, { "epoch": 15.041758241758242, "grad_norm": 1.5703125, "learning_rate": 6.583674300689598e-05, "loss": 0.1588, "step": 20532 }, { "epoch": 15.042490842490842, "grad_norm": 1.3984375, "learning_rate": 6.583348521685891e-05, "loss": 0.1304, "step": 20533 }, { "epoch": 15.043223443223443, "grad_norm": 0.80078125, "learning_rate": 6.583022735335409e-05, "loss": 0.1077, "step": 20534 }, { "epoch": 15.043956043956044, "grad_norm": 0.796875, "learning_rate": 6.582696941639712e-05, "loss": 0.0932, "step": 20535 }, { "epoch": 15.044688644688645, "grad_norm": 1.015625, "learning_rate": 6.58237114060036e-05, "loss": 0.1055, "step": 20536 }, { "epoch": 15.045421245421245, "grad_norm": 1.84375, "learning_rate": 6.582045332218914e-05, "loss": 0.1507, "step": 20537 }, { "epoch": 15.046153846153846, "grad_norm": 1.3046875, "learning_rate": 6.581719516496937e-05, "loss": 0.1123, "step": 20538 }, { "epoch": 15.046886446886447, "grad_norm": 1.3125, "learning_rate": 6.581393693435988e-05, "loss": 0.1318, "step": 20539 }, { "epoch": 15.047619047619047, "grad_norm": 1.1796875, "learning_rate": 6.58106786303763e-05, "loss": 0.1127, "step": 20540 }, { "epoch": 15.048351648351648, "grad_norm": 1.03125, "learning_rate": 6.580742025303422e-05, "loss": 0.1208, "step": 20541 }, { "epoch": 15.049084249084249, "grad_norm": 0.84765625, "learning_rate": 6.580416180234929e-05, "loss": 0.1113, "step": 20542 }, { "epoch": 15.04981684981685, "grad_norm": 0.97265625, "learning_rate": 6.580090327833707e-05, "loss": 0.1273, "step": 20543 }, { "epoch": 15.05054945054945, "grad_norm": 0.80078125, "learning_rate": 6.579764468101321e-05, "loss": 0.1214, "step": 20544 }, { "epoch": 15.051282051282051, "grad_norm": 0.953125, "learning_rate": 6.579438601039331e-05, "loss": 0.1172, "step": 20545 }, { "epoch": 15.052014652014652, "grad_norm": 1.78125, "learning_rate": 6.579112726649299e-05, "loss": 0.1591, "step": 20546 }, { "epoch": 15.052747252747253, "grad_norm": 1.671875, "learning_rate": 6.578786844932786e-05, "loss": 0.1278, "step": 20547 }, { "epoch": 15.053479853479853, "grad_norm": 1.1484375, "learning_rate": 6.578460955891354e-05, "loss": 0.1174, "step": 20548 }, { "epoch": 15.054212454212454, "grad_norm": 1.0625, "learning_rate": 6.578135059526561e-05, "loss": 0.1368, "step": 20549 }, { "epoch": 15.054945054945055, "grad_norm": 1.3125, "learning_rate": 6.577809155839974e-05, "loss": 0.1132, "step": 20550 }, { "epoch": 15.055677655677655, "grad_norm": 0.86328125, "learning_rate": 6.577483244833151e-05, "loss": 0.1128, "step": 20551 }, { "epoch": 15.056410256410256, "grad_norm": 1.2734375, "learning_rate": 6.577157326507655e-05, "loss": 0.1399, "step": 20552 }, { "epoch": 15.057142857142857, "grad_norm": 1.4609375, "learning_rate": 6.576831400865046e-05, "loss": 0.1328, "step": 20553 }, { "epoch": 15.057875457875458, "grad_norm": 1.5234375, "learning_rate": 6.576505467906887e-05, "loss": 0.1522, "step": 20554 }, { "epoch": 15.058608058608058, "grad_norm": 1.765625, "learning_rate": 6.576179527634738e-05, "loss": 0.1443, "step": 20555 }, { "epoch": 15.05934065934066, "grad_norm": 1.328125, "learning_rate": 6.575853580050164e-05, "loss": 0.1289, "step": 20556 }, { "epoch": 15.06007326007326, "grad_norm": 0.80078125, "learning_rate": 6.575527625154721e-05, "loss": 0.1339, "step": 20557 }, { "epoch": 15.06080586080586, "grad_norm": 1.3203125, "learning_rate": 6.575201662949978e-05, "loss": 0.1501, "step": 20558 }, { "epoch": 15.061538461538461, "grad_norm": 1.1015625, "learning_rate": 6.574875693437492e-05, "loss": 0.1382, "step": 20559 }, { "epoch": 15.062271062271062, "grad_norm": 1.0625, "learning_rate": 6.574549716618825e-05, "loss": 0.1454, "step": 20560 }, { "epoch": 15.063003663003663, "grad_norm": 1.0390625, "learning_rate": 6.574223732495542e-05, "loss": 0.1082, "step": 20561 }, { "epoch": 15.063736263736264, "grad_norm": 1.28125, "learning_rate": 6.5738977410692e-05, "loss": 0.1645, "step": 20562 }, { "epoch": 15.064468864468864, "grad_norm": 0.88671875, "learning_rate": 6.573571742341367e-05, "loss": 0.1261, "step": 20563 }, { "epoch": 15.065201465201465, "grad_norm": 0.921875, "learning_rate": 6.573245736313598e-05, "loss": 0.1247, "step": 20564 }, { "epoch": 15.065934065934066, "grad_norm": 1.1640625, "learning_rate": 6.572919722987461e-05, "loss": 0.1111, "step": 20565 }, { "epoch": 15.066666666666666, "grad_norm": 0.92578125, "learning_rate": 6.572593702364517e-05, "loss": 0.1239, "step": 20566 }, { "epoch": 15.067399267399267, "grad_norm": 0.87109375, "learning_rate": 6.572267674446324e-05, "loss": 0.1274, "step": 20567 }, { "epoch": 15.068131868131868, "grad_norm": 0.953125, "learning_rate": 6.571941639234447e-05, "loss": 0.1277, "step": 20568 }, { "epoch": 15.068864468864469, "grad_norm": 1.5859375, "learning_rate": 6.571615596730447e-05, "loss": 0.169, "step": 20569 }, { "epoch": 15.06959706959707, "grad_norm": 0.98046875, "learning_rate": 6.57128954693589e-05, "loss": 0.1215, "step": 20570 }, { "epoch": 15.07032967032967, "grad_norm": 1.0625, "learning_rate": 6.570963489852333e-05, "loss": 0.1477, "step": 20571 }, { "epoch": 15.07106227106227, "grad_norm": 0.94921875, "learning_rate": 6.570637425481342e-05, "loss": 0.1229, "step": 20572 }, { "epoch": 15.071794871794872, "grad_norm": 0.890625, "learning_rate": 6.570311353824478e-05, "loss": 0.1256, "step": 20573 }, { "epoch": 15.072527472527472, "grad_norm": 0.96875, "learning_rate": 6.569985274883304e-05, "loss": 0.1287, "step": 20574 }, { "epoch": 15.073260073260073, "grad_norm": 0.98828125, "learning_rate": 6.569659188659379e-05, "loss": 0.0917, "step": 20575 }, { "epoch": 15.073992673992674, "grad_norm": 0.84765625, "learning_rate": 6.56933309515427e-05, "loss": 0.1161, "step": 20576 }, { "epoch": 15.074725274725274, "grad_norm": 0.85546875, "learning_rate": 6.569006994369537e-05, "loss": 0.0927, "step": 20577 }, { "epoch": 15.075457875457875, "grad_norm": 0.9140625, "learning_rate": 6.568680886306742e-05, "loss": 0.0932, "step": 20578 }, { "epoch": 15.076190476190476, "grad_norm": 1.53125, "learning_rate": 6.56835477096745e-05, "loss": 0.1317, "step": 20579 }, { "epoch": 15.076923076923077, "grad_norm": 1.796875, "learning_rate": 6.56802864835322e-05, "loss": 0.1493, "step": 20580 }, { "epoch": 15.077655677655677, "grad_norm": 0.9609375, "learning_rate": 6.567702518465617e-05, "loss": 0.1399, "step": 20581 }, { "epoch": 15.078388278388278, "grad_norm": 1.0078125, "learning_rate": 6.567376381306201e-05, "loss": 0.1181, "step": 20582 }, { "epoch": 15.079120879120879, "grad_norm": 0.8984375, "learning_rate": 6.56705023687654e-05, "loss": 0.1252, "step": 20583 }, { "epoch": 15.07985347985348, "grad_norm": 1.0078125, "learning_rate": 6.566724085178193e-05, "loss": 0.1452, "step": 20584 }, { "epoch": 15.08058608058608, "grad_norm": 0.7890625, "learning_rate": 6.566397926212721e-05, "loss": 0.1179, "step": 20585 }, { "epoch": 15.081318681318681, "grad_norm": 0.83203125, "learning_rate": 6.566071759981691e-05, "loss": 0.0998, "step": 20586 }, { "epoch": 15.082051282051282, "grad_norm": 1.2578125, "learning_rate": 6.565745586486662e-05, "loss": 0.1235, "step": 20587 }, { "epoch": 15.082783882783882, "grad_norm": 1.4296875, "learning_rate": 6.565419405729198e-05, "loss": 0.1392, "step": 20588 }, { "epoch": 15.083516483516483, "grad_norm": 1.171875, "learning_rate": 6.565093217710864e-05, "loss": 0.1571, "step": 20589 }, { "epoch": 15.084249084249084, "grad_norm": 0.85546875, "learning_rate": 6.564767022433219e-05, "loss": 0.1315, "step": 20590 }, { "epoch": 15.084981684981685, "grad_norm": 1.0390625, "learning_rate": 6.564440819897829e-05, "loss": 0.134, "step": 20591 }, { "epoch": 15.085714285714285, "grad_norm": 1.2265625, "learning_rate": 6.564114610106255e-05, "loss": 0.1128, "step": 20592 }, { "epoch": 15.086446886446886, "grad_norm": 0.87890625, "learning_rate": 6.563788393060061e-05, "loss": 0.102, "step": 20593 }, { "epoch": 15.087179487179487, "grad_norm": 1.234375, "learning_rate": 6.563462168760812e-05, "loss": 0.1653, "step": 20594 }, { "epoch": 15.087912087912088, "grad_norm": 0.94921875, "learning_rate": 6.563135937210066e-05, "loss": 0.1125, "step": 20595 }, { "epoch": 15.088644688644688, "grad_norm": 1.03125, "learning_rate": 6.56280969840939e-05, "loss": 0.1385, "step": 20596 }, { "epoch": 15.089377289377289, "grad_norm": 0.8828125, "learning_rate": 6.562483452360347e-05, "loss": 0.133, "step": 20597 }, { "epoch": 15.09010989010989, "grad_norm": 1.15625, "learning_rate": 6.562157199064498e-05, "loss": 0.1132, "step": 20598 }, { "epoch": 15.09084249084249, "grad_norm": 1.734375, "learning_rate": 6.561830938523409e-05, "loss": 0.1397, "step": 20599 }, { "epoch": 15.091575091575091, "grad_norm": 1.6171875, "learning_rate": 6.561504670738641e-05, "loss": 0.1408, "step": 20600 }, { "epoch": 15.092307692307692, "grad_norm": 0.71484375, "learning_rate": 6.561178395711757e-05, "loss": 0.1275, "step": 20601 }, { "epoch": 15.093040293040293, "grad_norm": 2.015625, "learning_rate": 6.560852113444323e-05, "loss": 0.1361, "step": 20602 }, { "epoch": 15.093772893772893, "grad_norm": 2.671875, "learning_rate": 6.560525823937898e-05, "loss": 0.1542, "step": 20603 }, { "epoch": 15.094505494505494, "grad_norm": 1.9609375, "learning_rate": 6.56019952719405e-05, "loss": 0.1506, "step": 20604 }, { "epoch": 15.095238095238095, "grad_norm": 1.8515625, "learning_rate": 6.55987322321434e-05, "loss": 0.1389, "step": 20605 }, { "epoch": 15.095970695970696, "grad_norm": 0.79296875, "learning_rate": 6.55954691200033e-05, "loss": 0.1392, "step": 20606 }, { "epoch": 15.096703296703296, "grad_norm": 1.3515625, "learning_rate": 6.559220593553588e-05, "loss": 0.1153, "step": 20607 }, { "epoch": 15.097435897435897, "grad_norm": 1.0859375, "learning_rate": 6.558894267875672e-05, "loss": 0.1212, "step": 20608 }, { "epoch": 15.098168498168498, "grad_norm": 1.3203125, "learning_rate": 6.558567934968149e-05, "loss": 0.1247, "step": 20609 }, { "epoch": 15.098901098901099, "grad_norm": 2.015625, "learning_rate": 6.558241594832583e-05, "loss": 0.133, "step": 20610 }, { "epoch": 15.0996336996337, "grad_norm": 1.578125, "learning_rate": 6.557915247470535e-05, "loss": 0.1267, "step": 20611 }, { "epoch": 15.1003663003663, "grad_norm": 1.5546875, "learning_rate": 6.557588892883571e-05, "loss": 0.1148, "step": 20612 }, { "epoch": 15.1010989010989, "grad_norm": 1.0078125, "learning_rate": 6.557262531073254e-05, "loss": 0.1234, "step": 20613 }, { "epoch": 15.101831501831501, "grad_norm": 1.59375, "learning_rate": 6.556936162041147e-05, "loss": 0.1472, "step": 20614 }, { "epoch": 15.102564102564102, "grad_norm": 0.8671875, "learning_rate": 6.556609785788814e-05, "loss": 0.1201, "step": 20615 }, { "epoch": 15.103296703296703, "grad_norm": 0.9140625, "learning_rate": 6.556283402317818e-05, "loss": 0.1225, "step": 20616 }, { "epoch": 15.104029304029304, "grad_norm": 1.109375, "learning_rate": 6.555957011629726e-05, "loss": 0.1214, "step": 20617 }, { "epoch": 15.104761904761904, "grad_norm": 2.015625, "learning_rate": 6.555630613726098e-05, "loss": 0.1167, "step": 20618 }, { "epoch": 15.105494505494505, "grad_norm": 1.4609375, "learning_rate": 6.5553042086085e-05, "loss": 0.1263, "step": 20619 }, { "epoch": 15.106227106227106, "grad_norm": 1.4765625, "learning_rate": 6.554977796278494e-05, "loss": 0.1404, "step": 20620 }, { "epoch": 15.106959706959707, "grad_norm": 1.3984375, "learning_rate": 6.554651376737646e-05, "loss": 0.1069, "step": 20621 }, { "epoch": 15.107692307692307, "grad_norm": 1.3984375, "learning_rate": 6.554324949987519e-05, "loss": 0.1188, "step": 20622 }, { "epoch": 15.108424908424908, "grad_norm": 0.9140625, "learning_rate": 6.553998516029679e-05, "loss": 0.1, "step": 20623 }, { "epoch": 15.109157509157509, "grad_norm": 0.640625, "learning_rate": 6.553672074865687e-05, "loss": 0.1189, "step": 20624 }, { "epoch": 15.10989010989011, "grad_norm": 1.4609375, "learning_rate": 6.553345626497109e-05, "loss": 0.1166, "step": 20625 }, { "epoch": 15.11062271062271, "grad_norm": 1.2890625, "learning_rate": 6.55301917092551e-05, "loss": 0.1127, "step": 20626 }, { "epoch": 15.111355311355311, "grad_norm": 0.875, "learning_rate": 6.552692708152449e-05, "loss": 0.1379, "step": 20627 }, { "epoch": 15.112087912087912, "grad_norm": 1.9765625, "learning_rate": 6.552366238179498e-05, "loss": 0.1052, "step": 20628 }, { "epoch": 15.112820512820512, "grad_norm": 1.203125, "learning_rate": 6.552039761008214e-05, "loss": 0.1292, "step": 20629 }, { "epoch": 15.113553113553113, "grad_norm": 2.453125, "learning_rate": 6.551713276640166e-05, "loss": 0.1461, "step": 20630 }, { "epoch": 15.114285714285714, "grad_norm": 0.87890625, "learning_rate": 6.551386785076918e-05, "loss": 0.1154, "step": 20631 }, { "epoch": 15.115018315018315, "grad_norm": 1.2890625, "learning_rate": 6.55106028632003e-05, "loss": 0.1327, "step": 20632 }, { "epoch": 15.115750915750915, "grad_norm": 1.6328125, "learning_rate": 6.55073378037107e-05, "loss": 0.1173, "step": 20633 }, { "epoch": 15.116483516483516, "grad_norm": 1.046875, "learning_rate": 6.550407267231603e-05, "loss": 0.1004, "step": 20634 }, { "epoch": 15.117216117216117, "grad_norm": 0.91015625, "learning_rate": 6.550080746903193e-05, "loss": 0.128, "step": 20635 }, { "epoch": 15.117948717948718, "grad_norm": 1.5625, "learning_rate": 6.549754219387401e-05, "loss": 0.0954, "step": 20636 }, { "epoch": 15.118681318681318, "grad_norm": 1.7421875, "learning_rate": 6.549427684685796e-05, "loss": 0.1342, "step": 20637 }, { "epoch": 15.119413919413919, "grad_norm": 0.5390625, "learning_rate": 6.549101142799942e-05, "loss": 0.1068, "step": 20638 }, { "epoch": 15.12014652014652, "grad_norm": 1.3359375, "learning_rate": 6.548774593731401e-05, "loss": 0.1469, "step": 20639 }, { "epoch": 15.12087912087912, "grad_norm": 1.0625, "learning_rate": 6.548448037481737e-05, "loss": 0.1106, "step": 20640 }, { "epoch": 15.121611721611721, "grad_norm": 0.62109375, "learning_rate": 6.548121474052519e-05, "loss": 0.1164, "step": 20641 }, { "epoch": 15.122344322344322, "grad_norm": 1.0546875, "learning_rate": 6.547794903445308e-05, "loss": 0.088, "step": 20642 }, { "epoch": 15.123076923076923, "grad_norm": 1.734375, "learning_rate": 6.547468325661671e-05, "loss": 0.1383, "step": 20643 }, { "epoch": 15.123809523809523, "grad_norm": 0.9609375, "learning_rate": 6.547141740703172e-05, "loss": 0.1329, "step": 20644 }, { "epoch": 15.124542124542124, "grad_norm": 1.5390625, "learning_rate": 6.546815148571373e-05, "loss": 0.1128, "step": 20645 }, { "epoch": 15.125274725274725, "grad_norm": 1.171875, "learning_rate": 6.546488549267843e-05, "loss": 0.1245, "step": 20646 }, { "epoch": 15.126007326007326, "grad_norm": 0.89453125, "learning_rate": 6.546161942794145e-05, "loss": 0.1139, "step": 20647 }, { "epoch": 15.126739926739926, "grad_norm": 0.84765625, "learning_rate": 6.545835329151843e-05, "loss": 0.1274, "step": 20648 }, { "epoch": 15.127472527472527, "grad_norm": 1.09375, "learning_rate": 6.545508708342505e-05, "loss": 0.1238, "step": 20649 }, { "epoch": 15.128205128205128, "grad_norm": 1.1875, "learning_rate": 6.54518208036769e-05, "loss": 0.1352, "step": 20650 }, { "epoch": 15.128937728937728, "grad_norm": 1.171875, "learning_rate": 6.544855445228972e-05, "loss": 0.1389, "step": 20651 }, { "epoch": 15.12967032967033, "grad_norm": 0.94140625, "learning_rate": 6.544528802927908e-05, "loss": 0.097, "step": 20652 }, { "epoch": 15.13040293040293, "grad_norm": 0.609375, "learning_rate": 6.544202153466065e-05, "loss": 0.1153, "step": 20653 }, { "epoch": 15.13113553113553, "grad_norm": 1.2734375, "learning_rate": 6.543875496845011e-05, "loss": 0.1276, "step": 20654 }, { "epoch": 15.131868131868131, "grad_norm": 0.71875, "learning_rate": 6.543548833066308e-05, "loss": 0.113, "step": 20655 }, { "epoch": 15.132600732600732, "grad_norm": 1.0703125, "learning_rate": 6.543222162131523e-05, "loss": 0.123, "step": 20656 }, { "epoch": 15.133333333333333, "grad_norm": 1.3515625, "learning_rate": 6.54289548404222e-05, "loss": 0.1546, "step": 20657 }, { "epoch": 15.134065934065934, "grad_norm": 1.3125, "learning_rate": 6.542568798799964e-05, "loss": 0.152, "step": 20658 }, { "epoch": 15.134798534798534, "grad_norm": 0.99609375, "learning_rate": 6.54224210640632e-05, "loss": 0.117, "step": 20659 }, { "epoch": 15.135531135531135, "grad_norm": 1.625, "learning_rate": 6.541915406862856e-05, "loss": 0.1422, "step": 20660 }, { "epoch": 15.136263736263736, "grad_norm": 0.57421875, "learning_rate": 6.541588700171135e-05, "loss": 0.1201, "step": 20661 }, { "epoch": 15.136996336996337, "grad_norm": 0.97265625, "learning_rate": 6.541261986332723e-05, "loss": 0.1192, "step": 20662 }, { "epoch": 15.137728937728937, "grad_norm": 0.87109375, "learning_rate": 6.540935265349185e-05, "loss": 0.1207, "step": 20663 }, { "epoch": 15.138461538461538, "grad_norm": 1.546875, "learning_rate": 6.540608537222087e-05, "loss": 0.1256, "step": 20664 }, { "epoch": 15.139194139194139, "grad_norm": 1.1171875, "learning_rate": 6.540281801952996e-05, "loss": 0.1312, "step": 20665 }, { "epoch": 15.13992673992674, "grad_norm": 0.97265625, "learning_rate": 6.539955059543473e-05, "loss": 0.143, "step": 20666 }, { "epoch": 15.14065934065934, "grad_norm": 0.9921875, "learning_rate": 6.539628309995087e-05, "loss": 0.1243, "step": 20667 }, { "epoch": 15.141391941391941, "grad_norm": 0.8515625, "learning_rate": 6.539301553309403e-05, "loss": 0.1191, "step": 20668 }, { "epoch": 15.142124542124542, "grad_norm": 0.7734375, "learning_rate": 6.538974789487985e-05, "loss": 0.1247, "step": 20669 }, { "epoch": 15.142857142857142, "grad_norm": 0.73046875, "learning_rate": 6.538648018532403e-05, "loss": 0.1155, "step": 20670 }, { "epoch": 15.143589743589743, "grad_norm": 1.0859375, "learning_rate": 6.538321240444216e-05, "loss": 0.1244, "step": 20671 }, { "epoch": 15.144322344322344, "grad_norm": 1.53125, "learning_rate": 6.537994455224994e-05, "loss": 0.1411, "step": 20672 }, { "epoch": 15.145054945054945, "grad_norm": 1.5703125, "learning_rate": 6.537667662876303e-05, "loss": 0.1341, "step": 20673 }, { "epoch": 15.145787545787545, "grad_norm": 0.79296875, "learning_rate": 6.53734086339971e-05, "loss": 0.1271, "step": 20674 }, { "epoch": 15.146520146520146, "grad_norm": 1.34375, "learning_rate": 6.537014056796775e-05, "loss": 0.1236, "step": 20675 }, { "epoch": 15.147252747252747, "grad_norm": 1.0546875, "learning_rate": 6.536687243069069e-05, "loss": 0.123, "step": 20676 }, { "epoch": 15.147985347985347, "grad_norm": 1.09375, "learning_rate": 6.536360422218157e-05, "loss": 0.1401, "step": 20677 }, { "epoch": 15.148717948717948, "grad_norm": 0.90625, "learning_rate": 6.536033594245605e-05, "loss": 0.1209, "step": 20678 }, { "epoch": 15.149450549450549, "grad_norm": 0.609375, "learning_rate": 6.535706759152976e-05, "loss": 0.1021, "step": 20679 }, { "epoch": 15.15018315018315, "grad_norm": 0.94921875, "learning_rate": 6.535379916941841e-05, "loss": 0.1388, "step": 20680 }, { "epoch": 15.15091575091575, "grad_norm": 1.1328125, "learning_rate": 6.535053067613761e-05, "loss": 0.1262, "step": 20681 }, { "epoch": 15.151648351648351, "grad_norm": 1.015625, "learning_rate": 6.534726211170304e-05, "loss": 0.0862, "step": 20682 }, { "epoch": 15.152380952380952, "grad_norm": 0.77734375, "learning_rate": 6.534399347613037e-05, "loss": 0.1107, "step": 20683 }, { "epoch": 15.153113553113553, "grad_norm": 1.203125, "learning_rate": 6.534072476943526e-05, "loss": 0.1161, "step": 20684 }, { "epoch": 15.153846153846153, "grad_norm": 1.234375, "learning_rate": 6.533745599163334e-05, "loss": 0.1367, "step": 20685 }, { "epoch": 15.154578754578754, "grad_norm": 0.91015625, "learning_rate": 6.533418714274032e-05, "loss": 0.1363, "step": 20686 }, { "epoch": 15.155311355311355, "grad_norm": 0.703125, "learning_rate": 6.533091822277184e-05, "loss": 0.0999, "step": 20687 }, { "epoch": 15.156043956043955, "grad_norm": 1.109375, "learning_rate": 6.532764923174355e-05, "loss": 0.1506, "step": 20688 }, { "epoch": 15.156776556776556, "grad_norm": 0.88671875, "learning_rate": 6.532438016967113e-05, "loss": 0.1429, "step": 20689 }, { "epoch": 15.157509157509157, "grad_norm": 1.703125, "learning_rate": 6.532111103657026e-05, "loss": 0.1466, "step": 20690 }, { "epoch": 15.158241758241758, "grad_norm": 0.921875, "learning_rate": 6.531784183245657e-05, "loss": 0.121, "step": 20691 }, { "epoch": 15.158974358974358, "grad_norm": 1.0390625, "learning_rate": 6.531457255734572e-05, "loss": 0.1306, "step": 20692 }, { "epoch": 15.15970695970696, "grad_norm": 0.9921875, "learning_rate": 6.53113032112534e-05, "loss": 0.1222, "step": 20693 }, { "epoch": 15.16043956043956, "grad_norm": 1.0546875, "learning_rate": 6.530803379419527e-05, "loss": 0.1254, "step": 20694 }, { "epoch": 15.16117216117216, "grad_norm": 1.140625, "learning_rate": 6.530476430618697e-05, "loss": 0.1509, "step": 20695 }, { "epoch": 15.161904761904761, "grad_norm": 0.67578125, "learning_rate": 6.530149474724421e-05, "loss": 0.0991, "step": 20696 }, { "epoch": 15.162637362637362, "grad_norm": 0.703125, "learning_rate": 6.52982251173826e-05, "loss": 0.1228, "step": 20697 }, { "epoch": 15.163369963369963, "grad_norm": 1.3203125, "learning_rate": 6.529495541661785e-05, "loss": 0.1057, "step": 20698 }, { "epoch": 15.164102564102564, "grad_norm": 1.265625, "learning_rate": 6.529168564496562e-05, "loss": 0.1231, "step": 20699 }, { "epoch": 15.164835164835164, "grad_norm": 1.6171875, "learning_rate": 6.528841580244155e-05, "loss": 0.174, "step": 20700 }, { "epoch": 15.165567765567765, "grad_norm": 1.234375, "learning_rate": 6.528514588906135e-05, "loss": 0.1437, "step": 20701 }, { "epoch": 15.166300366300366, "grad_norm": 1.71875, "learning_rate": 6.528187590484064e-05, "loss": 0.155, "step": 20702 }, { "epoch": 15.167032967032966, "grad_norm": 0.81640625, "learning_rate": 6.527860584979512e-05, "loss": 0.1322, "step": 20703 }, { "epoch": 15.167765567765567, "grad_norm": 0.890625, "learning_rate": 6.527533572394044e-05, "loss": 0.1037, "step": 20704 }, { "epoch": 15.168498168498168, "grad_norm": 0.78125, "learning_rate": 6.527206552729229e-05, "loss": 0.12, "step": 20705 }, { "epoch": 15.169230769230769, "grad_norm": 1.4375, "learning_rate": 6.526879525986633e-05, "loss": 0.1201, "step": 20706 }, { "epoch": 15.16996336996337, "grad_norm": 0.7890625, "learning_rate": 6.52655249216782e-05, "loss": 0.1038, "step": 20707 }, { "epoch": 15.17069597069597, "grad_norm": 1.0546875, "learning_rate": 6.52622545127436e-05, "loss": 0.1029, "step": 20708 }, { "epoch": 15.17142857142857, "grad_norm": 1.375, "learning_rate": 6.525898403307821e-05, "loss": 0.1125, "step": 20709 }, { "epoch": 15.172161172161172, "grad_norm": 2.25, "learning_rate": 6.525571348269766e-05, "loss": 0.1268, "step": 20710 }, { "epoch": 15.172893772893772, "grad_norm": 1.2734375, "learning_rate": 6.525244286161765e-05, "loss": 0.1314, "step": 20711 }, { "epoch": 15.173626373626373, "grad_norm": 1.234375, "learning_rate": 6.524917216985384e-05, "loss": 0.1091, "step": 20712 }, { "epoch": 15.174358974358974, "grad_norm": 1.546875, "learning_rate": 6.52459014074219e-05, "loss": 0.1278, "step": 20713 }, { "epoch": 15.175091575091574, "grad_norm": 1.0625, "learning_rate": 6.524263057433752e-05, "loss": 0.1156, "step": 20714 }, { "epoch": 15.175824175824175, "grad_norm": 1.015625, "learning_rate": 6.523935967061635e-05, "loss": 0.126, "step": 20715 }, { "epoch": 15.176556776556776, "grad_norm": 0.8828125, "learning_rate": 6.523608869627408e-05, "loss": 0.1042, "step": 20716 }, { "epoch": 15.177289377289377, "grad_norm": 1.6796875, "learning_rate": 6.523281765132637e-05, "loss": 0.1551, "step": 20717 }, { "epoch": 15.178021978021977, "grad_norm": 1.3828125, "learning_rate": 6.522954653578888e-05, "loss": 0.1354, "step": 20718 }, { "epoch": 15.178754578754578, "grad_norm": 1.1484375, "learning_rate": 6.52262753496773e-05, "loss": 0.1265, "step": 20719 }, { "epoch": 15.179487179487179, "grad_norm": 0.8125, "learning_rate": 6.522300409300733e-05, "loss": 0.103, "step": 20720 }, { "epoch": 15.18021978021978, "grad_norm": 0.7578125, "learning_rate": 6.521973276579456e-05, "loss": 0.1233, "step": 20721 }, { "epoch": 15.18095238095238, "grad_norm": 1.2421875, "learning_rate": 6.521646136805476e-05, "loss": 0.1146, "step": 20722 }, { "epoch": 15.181684981684981, "grad_norm": 1.4296875, "learning_rate": 6.521318989980354e-05, "loss": 0.1326, "step": 20723 }, { "epoch": 15.182417582417582, "grad_norm": 1.3671875, "learning_rate": 6.520991836105661e-05, "loss": 0.1047, "step": 20724 }, { "epoch": 15.183150183150182, "grad_norm": 0.89453125, "learning_rate": 6.520664675182962e-05, "loss": 0.123, "step": 20725 }, { "epoch": 15.183882783882783, "grad_norm": 0.95703125, "learning_rate": 6.520337507213826e-05, "loss": 0.1412, "step": 20726 }, { "epoch": 15.184615384615384, "grad_norm": 2.03125, "learning_rate": 6.520010332199823e-05, "loss": 0.1393, "step": 20727 }, { "epoch": 15.185347985347985, "grad_norm": 1.109375, "learning_rate": 6.519683150142513e-05, "loss": 0.1207, "step": 20728 }, { "epoch": 15.186080586080585, "grad_norm": 1.390625, "learning_rate": 6.519355961043473e-05, "loss": 0.1531, "step": 20729 }, { "epoch": 15.186813186813186, "grad_norm": 0.80859375, "learning_rate": 6.519028764904265e-05, "loss": 0.1049, "step": 20730 }, { "epoch": 15.187545787545787, "grad_norm": 0.734375, "learning_rate": 6.518701561726457e-05, "loss": 0.0998, "step": 20731 }, { "epoch": 15.188278388278388, "grad_norm": 0.89453125, "learning_rate": 6.518374351511619e-05, "loss": 0.128, "step": 20732 }, { "epoch": 15.189010989010988, "grad_norm": 1.7421875, "learning_rate": 6.518047134261317e-05, "loss": 0.1231, "step": 20733 }, { "epoch": 15.189743589743589, "grad_norm": 2.375, "learning_rate": 6.517719909977119e-05, "loss": 0.1571, "step": 20734 }, { "epoch": 15.19047619047619, "grad_norm": 1.2890625, "learning_rate": 6.517392678660594e-05, "loss": 0.1334, "step": 20735 }, { "epoch": 15.19120879120879, "grad_norm": 1.03125, "learning_rate": 6.517065440313307e-05, "loss": 0.1223, "step": 20736 }, { "epoch": 15.191941391941391, "grad_norm": 1.5390625, "learning_rate": 6.516738194936827e-05, "loss": 0.1075, "step": 20737 }, { "epoch": 15.192673992673992, "grad_norm": 0.85546875, "learning_rate": 6.516410942532727e-05, "loss": 0.1164, "step": 20738 }, { "epoch": 15.193406593406593, "grad_norm": 0.89453125, "learning_rate": 6.516083683102568e-05, "loss": 0.1389, "step": 20739 }, { "epoch": 15.194139194139193, "grad_norm": 0.62109375, "learning_rate": 6.515756416647921e-05, "loss": 0.1206, "step": 20740 }, { "epoch": 15.194871794871794, "grad_norm": 1.3828125, "learning_rate": 6.515429143170354e-05, "loss": 0.1224, "step": 20741 }, { "epoch": 15.195604395604395, "grad_norm": 2.03125, "learning_rate": 6.515101862671435e-05, "loss": 0.1136, "step": 20742 }, { "epoch": 15.196336996336996, "grad_norm": 1.734375, "learning_rate": 6.514774575152731e-05, "loss": 0.1287, "step": 20743 }, { "epoch": 15.197069597069596, "grad_norm": 1.484375, "learning_rate": 6.514447280615813e-05, "loss": 0.1111, "step": 20744 }, { "epoch": 15.197802197802197, "grad_norm": 1.109375, "learning_rate": 6.514119979062246e-05, "loss": 0.1298, "step": 20745 }, { "epoch": 15.198534798534798, "grad_norm": 1.109375, "learning_rate": 6.513792670493601e-05, "loss": 0.1205, "step": 20746 }, { "epoch": 15.199267399267399, "grad_norm": 1.6875, "learning_rate": 6.513465354911443e-05, "loss": 0.1252, "step": 20747 }, { "epoch": 15.2, "grad_norm": 0.9375, "learning_rate": 6.513138032317342e-05, "loss": 0.111, "step": 20748 }, { "epoch": 15.2007326007326, "grad_norm": 1.109375, "learning_rate": 6.512810702712869e-05, "loss": 0.1482, "step": 20749 }, { "epoch": 15.2014652014652, "grad_norm": 1.1953125, "learning_rate": 6.512483366099588e-05, "loss": 0.1238, "step": 20750 }, { "epoch": 15.202197802197801, "grad_norm": 0.76953125, "learning_rate": 6.512156022479069e-05, "loss": 0.1319, "step": 20751 }, { "epoch": 15.202930402930402, "grad_norm": 1.3046875, "learning_rate": 6.51182867185288e-05, "loss": 0.1047, "step": 20752 }, { "epoch": 15.203663003663003, "grad_norm": 0.8828125, "learning_rate": 6.511501314222592e-05, "loss": 0.1115, "step": 20753 }, { "epoch": 15.204395604395604, "grad_norm": 0.79296875, "learning_rate": 6.511173949589768e-05, "loss": 0.1403, "step": 20754 }, { "epoch": 15.205128205128204, "grad_norm": 0.75, "learning_rate": 6.510846577955982e-05, "loss": 0.1022, "step": 20755 }, { "epoch": 15.205860805860805, "grad_norm": 1.4453125, "learning_rate": 6.5105191993228e-05, "loss": 0.1338, "step": 20756 }, { "epoch": 15.206593406593406, "grad_norm": 1.2890625, "learning_rate": 6.510191813691793e-05, "loss": 0.1449, "step": 20757 }, { "epoch": 15.207326007326007, "grad_norm": 1.84375, "learning_rate": 6.509864421064526e-05, "loss": 0.1568, "step": 20758 }, { "epoch": 15.208058608058607, "grad_norm": 0.921875, "learning_rate": 6.509537021442568e-05, "loss": 0.1014, "step": 20759 }, { "epoch": 15.208791208791208, "grad_norm": 0.9375, "learning_rate": 6.509209614827491e-05, "loss": 0.1205, "step": 20760 }, { "epoch": 15.209523809523809, "grad_norm": 1.25, "learning_rate": 6.508882201220861e-05, "loss": 0.1432, "step": 20761 }, { "epoch": 15.21025641025641, "grad_norm": 1.4921875, "learning_rate": 6.508554780624247e-05, "loss": 0.1241, "step": 20762 }, { "epoch": 15.21098901098901, "grad_norm": 0.9296875, "learning_rate": 6.50822735303922e-05, "loss": 0.1031, "step": 20763 }, { "epoch": 15.211721611721611, "grad_norm": 1.265625, "learning_rate": 6.507899918467347e-05, "loss": 0.1293, "step": 20764 }, { "epoch": 15.212454212454212, "grad_norm": 0.61328125, "learning_rate": 6.507572476910194e-05, "loss": 0.1259, "step": 20765 }, { "epoch": 15.213186813186812, "grad_norm": 1.1328125, "learning_rate": 6.507245028369334e-05, "loss": 0.1349, "step": 20766 }, { "epoch": 15.213919413919413, "grad_norm": 0.765625, "learning_rate": 6.506917572846335e-05, "loss": 0.1082, "step": 20767 }, { "epoch": 15.214652014652014, "grad_norm": 1.4453125, "learning_rate": 6.506590110342766e-05, "loss": 0.1142, "step": 20768 }, { "epoch": 15.215384615384615, "grad_norm": 0.93359375, "learning_rate": 6.506262640860197e-05, "loss": 0.1143, "step": 20769 }, { "epoch": 15.216117216117215, "grad_norm": 0.609375, "learning_rate": 6.505935164400192e-05, "loss": 0.1302, "step": 20770 }, { "epoch": 15.216849816849816, "grad_norm": 0.83984375, "learning_rate": 6.505607680964328e-05, "loss": 0.1404, "step": 20771 }, { "epoch": 15.217582417582417, "grad_norm": 0.9453125, "learning_rate": 6.505280190554167e-05, "loss": 0.1212, "step": 20772 }, { "epoch": 15.218315018315018, "grad_norm": 2.140625, "learning_rate": 6.504952693171282e-05, "loss": 0.1617, "step": 20773 }, { "epoch": 15.219047619047618, "grad_norm": 1.0078125, "learning_rate": 6.504625188817239e-05, "loss": 0.1275, "step": 20774 }, { "epoch": 15.219780219780219, "grad_norm": 1.0546875, "learning_rate": 6.50429767749361e-05, "loss": 0.1331, "step": 20775 }, { "epoch": 15.22051282051282, "grad_norm": 0.95703125, "learning_rate": 6.503970159201964e-05, "loss": 0.1718, "step": 20776 }, { "epoch": 15.22124542124542, "grad_norm": 0.71484375, "learning_rate": 6.50364263394387e-05, "loss": 0.1348, "step": 20777 }, { "epoch": 15.221978021978021, "grad_norm": 0.734375, "learning_rate": 6.503315101720896e-05, "loss": 0.1091, "step": 20778 }, { "epoch": 15.222710622710622, "grad_norm": 1.109375, "learning_rate": 6.502987562534613e-05, "loss": 0.1255, "step": 20779 }, { "epoch": 15.223443223443223, "grad_norm": 1.5703125, "learning_rate": 6.50266001638659e-05, "loss": 0.1478, "step": 20780 }, { "epoch": 15.224175824175823, "grad_norm": 1.4453125, "learning_rate": 6.502332463278396e-05, "loss": 0.1662, "step": 20781 }, { "epoch": 15.224908424908424, "grad_norm": 0.97265625, "learning_rate": 6.5020049032116e-05, "loss": 0.1104, "step": 20782 }, { "epoch": 15.225641025641025, "grad_norm": 0.90234375, "learning_rate": 6.50167733618777e-05, "loss": 0.1062, "step": 20783 }, { "epoch": 15.226373626373626, "grad_norm": 0.97265625, "learning_rate": 6.50134976220848e-05, "loss": 0.1342, "step": 20784 }, { "epoch": 15.227106227106226, "grad_norm": 0.8828125, "learning_rate": 6.501022181275297e-05, "loss": 0.1159, "step": 20785 }, { "epoch": 15.227838827838827, "grad_norm": 0.98046875, "learning_rate": 6.500694593389788e-05, "loss": 0.1238, "step": 20786 }, { "epoch": 15.228571428571428, "grad_norm": 0.9375, "learning_rate": 6.500366998553526e-05, "loss": 0.1093, "step": 20787 }, { "epoch": 15.229304029304028, "grad_norm": 1.203125, "learning_rate": 6.50003939676808e-05, "loss": 0.1122, "step": 20788 }, { "epoch": 15.23003663003663, "grad_norm": 1.4296875, "learning_rate": 6.499711788035019e-05, "loss": 0.1218, "step": 20789 }, { "epoch": 15.23076923076923, "grad_norm": 0.8984375, "learning_rate": 6.499384172355912e-05, "loss": 0.1176, "step": 20790 }, { "epoch": 15.23150183150183, "grad_norm": 0.984375, "learning_rate": 6.499056549732332e-05, "loss": 0.1653, "step": 20791 }, { "epoch": 15.232234432234431, "grad_norm": 1.2421875, "learning_rate": 6.498728920165844e-05, "loss": 0.1542, "step": 20792 }, { "epoch": 15.232967032967032, "grad_norm": 0.79296875, "learning_rate": 6.498401283658021e-05, "loss": 0.1281, "step": 20793 }, { "epoch": 15.233699633699633, "grad_norm": 0.7109375, "learning_rate": 6.498073640210433e-05, "loss": 0.1433, "step": 20794 }, { "epoch": 15.234432234432234, "grad_norm": 1.859375, "learning_rate": 6.497745989824649e-05, "loss": 0.1484, "step": 20795 }, { "epoch": 15.235164835164834, "grad_norm": 1.140625, "learning_rate": 6.497418332502236e-05, "loss": 0.1203, "step": 20796 }, { "epoch": 15.235897435897435, "grad_norm": 1.3984375, "learning_rate": 6.497090668244769e-05, "loss": 0.1086, "step": 20797 }, { "epoch": 15.236630036630036, "grad_norm": 1.0546875, "learning_rate": 6.496762997053815e-05, "loss": 0.1216, "step": 20798 }, { "epoch": 15.237362637362637, "grad_norm": 1.140625, "learning_rate": 6.496435318930943e-05, "loss": 0.1134, "step": 20799 }, { "epoch": 15.238095238095237, "grad_norm": 1.0, "learning_rate": 6.496107633877726e-05, "loss": 0.1217, "step": 20800 }, { "epoch": 15.238827838827838, "grad_norm": 0.703125, "learning_rate": 6.495779941895734e-05, "loss": 0.1171, "step": 20801 }, { "epoch": 15.239560439560439, "grad_norm": 0.6328125, "learning_rate": 6.495452242986533e-05, "loss": 0.0939, "step": 20802 }, { "epoch": 15.24029304029304, "grad_norm": 0.85546875, "learning_rate": 6.495124537151698e-05, "loss": 0.1313, "step": 20803 }, { "epoch": 15.24102564102564, "grad_norm": 1.109375, "learning_rate": 6.494796824392795e-05, "loss": 0.1103, "step": 20804 }, { "epoch": 15.241758241758241, "grad_norm": 1.25, "learning_rate": 6.494469104711398e-05, "loss": 0.1186, "step": 20805 }, { "epoch": 15.242490842490842, "grad_norm": 1.515625, "learning_rate": 6.494141378109073e-05, "loss": 0.1297, "step": 20806 }, { "epoch": 15.243223443223442, "grad_norm": 1.5078125, "learning_rate": 6.493813644587395e-05, "loss": 0.1176, "step": 20807 }, { "epoch": 15.243956043956043, "grad_norm": 0.9375, "learning_rate": 6.493485904147931e-05, "loss": 0.1381, "step": 20808 }, { "epoch": 15.244688644688644, "grad_norm": 0.85546875, "learning_rate": 6.49315815679225e-05, "loss": 0.1411, "step": 20809 }, { "epoch": 15.245421245421245, "grad_norm": 1.0078125, "learning_rate": 6.492830402521928e-05, "loss": 0.1295, "step": 20810 }, { "epoch": 15.246153846153845, "grad_norm": 1.109375, "learning_rate": 6.49250264133853e-05, "loss": 0.1241, "step": 20811 }, { "epoch": 15.246886446886446, "grad_norm": 1.421875, "learning_rate": 6.492174873243628e-05, "loss": 0.1524, "step": 20812 }, { "epoch": 15.247619047619047, "grad_norm": 1.4765625, "learning_rate": 6.491847098238793e-05, "loss": 0.1546, "step": 20813 }, { "epoch": 15.248351648351647, "grad_norm": 1.6953125, "learning_rate": 6.491519316325595e-05, "loss": 0.1311, "step": 20814 }, { "epoch": 15.249084249084248, "grad_norm": 1.015625, "learning_rate": 6.491191527505607e-05, "loss": 0.1245, "step": 20815 }, { "epoch": 15.249816849816849, "grad_norm": 0.94140625, "learning_rate": 6.490863731780394e-05, "loss": 0.139, "step": 20816 }, { "epoch": 15.25054945054945, "grad_norm": 1.2109375, "learning_rate": 6.490535929151532e-05, "loss": 0.1278, "step": 20817 }, { "epoch": 15.25128205128205, "grad_norm": 1.375, "learning_rate": 6.49020811962059e-05, "loss": 0.1374, "step": 20818 }, { "epoch": 15.252014652014651, "grad_norm": 0.8671875, "learning_rate": 6.489880303189135e-05, "loss": 0.1321, "step": 20819 }, { "epoch": 15.252747252747252, "grad_norm": 0.79296875, "learning_rate": 6.489552479858745e-05, "loss": 0.1071, "step": 20820 }, { "epoch": 15.253479853479853, "grad_norm": 1.0078125, "learning_rate": 6.489224649630984e-05, "loss": 0.1042, "step": 20821 }, { "epoch": 15.254212454212453, "grad_norm": 0.87109375, "learning_rate": 6.488896812507425e-05, "loss": 0.1082, "step": 20822 }, { "epoch": 15.254945054945054, "grad_norm": 0.88671875, "learning_rate": 6.48856896848964e-05, "loss": 0.1051, "step": 20823 }, { "epoch": 15.255677655677655, "grad_norm": 0.93359375, "learning_rate": 6.488241117579198e-05, "loss": 0.1326, "step": 20824 }, { "epoch": 15.256410256410255, "grad_norm": 0.8125, "learning_rate": 6.48791325977767e-05, "loss": 0.1176, "step": 20825 }, { "epoch": 15.257142857142856, "grad_norm": 1.546875, "learning_rate": 6.487585395086631e-05, "loss": 0.1224, "step": 20826 }, { "epoch": 15.257875457875459, "grad_norm": 0.859375, "learning_rate": 6.487257523507644e-05, "loss": 0.098, "step": 20827 }, { "epoch": 15.25860805860806, "grad_norm": 1.625, "learning_rate": 6.486929645042288e-05, "loss": 0.1248, "step": 20828 }, { "epoch": 15.25934065934066, "grad_norm": 1.1015625, "learning_rate": 6.486601759692127e-05, "loss": 0.1452, "step": 20829 }, { "epoch": 15.260073260073261, "grad_norm": 1.28125, "learning_rate": 6.486273867458738e-05, "loss": 0.1199, "step": 20830 }, { "epoch": 15.260805860805862, "grad_norm": 1.2265625, "learning_rate": 6.485945968343689e-05, "loss": 0.1214, "step": 20831 }, { "epoch": 15.261538461538462, "grad_norm": 1.2890625, "learning_rate": 6.485618062348549e-05, "loss": 0.11, "step": 20832 }, { "epoch": 15.262271062271063, "grad_norm": 1.6171875, "learning_rate": 6.485290149474895e-05, "loss": 0.1209, "step": 20833 }, { "epoch": 15.263003663003664, "grad_norm": 1.21875, "learning_rate": 6.484962229724295e-05, "loss": 0.1379, "step": 20834 }, { "epoch": 15.263736263736265, "grad_norm": 1.2578125, "learning_rate": 6.484634303098317e-05, "loss": 0.1214, "step": 20835 }, { "epoch": 15.264468864468865, "grad_norm": 0.76171875, "learning_rate": 6.484306369598536e-05, "loss": 0.1212, "step": 20836 }, { "epoch": 15.265201465201466, "grad_norm": 0.92578125, "learning_rate": 6.483978429226524e-05, "loss": 0.1326, "step": 20837 }, { "epoch": 15.265934065934067, "grad_norm": 0.8125, "learning_rate": 6.483650481983848e-05, "loss": 0.1181, "step": 20838 }, { "epoch": 15.266666666666667, "grad_norm": 1.109375, "learning_rate": 6.483322527872083e-05, "loss": 0.1285, "step": 20839 }, { "epoch": 15.267399267399268, "grad_norm": 0.92578125, "learning_rate": 6.482994566892798e-05, "loss": 0.1515, "step": 20840 }, { "epoch": 15.268131868131869, "grad_norm": 1.171875, "learning_rate": 6.482666599047569e-05, "loss": 0.1422, "step": 20841 }, { "epoch": 15.26886446886447, "grad_norm": 1.015625, "learning_rate": 6.482338624337962e-05, "loss": 0.1282, "step": 20842 }, { "epoch": 15.26959706959707, "grad_norm": 1.390625, "learning_rate": 6.482010642765549e-05, "loss": 0.1267, "step": 20843 }, { "epoch": 15.270329670329671, "grad_norm": 1.34375, "learning_rate": 6.481682654331906e-05, "loss": 0.1458, "step": 20844 }, { "epoch": 15.271062271062272, "grad_norm": 0.65234375, "learning_rate": 6.4813546590386e-05, "loss": 0.0985, "step": 20845 }, { "epoch": 15.271794871794873, "grad_norm": 1.3125, "learning_rate": 6.481026656887202e-05, "loss": 0.1339, "step": 20846 }, { "epoch": 15.272527472527473, "grad_norm": 1.0078125, "learning_rate": 6.480698647879289e-05, "loss": 0.1149, "step": 20847 }, { "epoch": 15.273260073260074, "grad_norm": 1.8125, "learning_rate": 6.480370632016426e-05, "loss": 0.1375, "step": 20848 }, { "epoch": 15.273992673992675, "grad_norm": 1.0546875, "learning_rate": 6.48004260930019e-05, "loss": 0.1208, "step": 20849 }, { "epoch": 15.274725274725276, "grad_norm": 0.7578125, "learning_rate": 6.47971457973215e-05, "loss": 0.1316, "step": 20850 }, { "epoch": 15.275457875457876, "grad_norm": 0.76171875, "learning_rate": 6.479386543313877e-05, "loss": 0.1195, "step": 20851 }, { "epoch": 15.276190476190477, "grad_norm": 1.3828125, "learning_rate": 6.479058500046943e-05, "loss": 0.1063, "step": 20852 }, { "epoch": 15.276923076923078, "grad_norm": 1.6328125, "learning_rate": 6.478730449932922e-05, "loss": 0.1242, "step": 20853 }, { "epoch": 15.277655677655678, "grad_norm": 0.828125, "learning_rate": 6.478402392973387e-05, "loss": 0.1467, "step": 20854 }, { "epoch": 15.27838827838828, "grad_norm": 1.375, "learning_rate": 6.478074329169904e-05, "loss": 0.1277, "step": 20855 }, { "epoch": 15.27912087912088, "grad_norm": 0.9140625, "learning_rate": 6.477746258524048e-05, "loss": 0.1195, "step": 20856 }, { "epoch": 15.27985347985348, "grad_norm": 1.265625, "learning_rate": 6.477418181037391e-05, "loss": 0.1266, "step": 20857 }, { "epoch": 15.280586080586081, "grad_norm": 1.40625, "learning_rate": 6.477090096711506e-05, "loss": 0.1277, "step": 20858 }, { "epoch": 15.281318681318682, "grad_norm": 0.953125, "learning_rate": 6.476762005547964e-05, "loss": 0.1106, "step": 20859 }, { "epoch": 15.282051282051283, "grad_norm": 1.515625, "learning_rate": 6.476433907548335e-05, "loss": 0.1422, "step": 20860 }, { "epoch": 15.282783882783884, "grad_norm": 1.859375, "learning_rate": 6.476105802714196e-05, "loss": 0.1193, "step": 20861 }, { "epoch": 15.283516483516484, "grad_norm": 0.9453125, "learning_rate": 6.475777691047114e-05, "loss": 0.1094, "step": 20862 }, { "epoch": 15.284249084249085, "grad_norm": 1.25, "learning_rate": 6.475449572548663e-05, "loss": 0.1017, "step": 20863 }, { "epoch": 15.284981684981686, "grad_norm": 1.4375, "learning_rate": 6.475121447220416e-05, "loss": 0.1567, "step": 20864 }, { "epoch": 15.285714285714286, "grad_norm": 0.75390625, "learning_rate": 6.474793315063942e-05, "loss": 0.1195, "step": 20865 }, { "epoch": 15.286446886446887, "grad_norm": 0.984375, "learning_rate": 6.474465176080817e-05, "loss": 0.1413, "step": 20866 }, { "epoch": 15.287179487179488, "grad_norm": 1.1875, "learning_rate": 6.474137030272613e-05, "loss": 0.1104, "step": 20867 }, { "epoch": 15.287912087912089, "grad_norm": 1.46875, "learning_rate": 6.473808877640898e-05, "loss": 0.1238, "step": 20868 }, { "epoch": 15.28864468864469, "grad_norm": 1.265625, "learning_rate": 6.47348071818725e-05, "loss": 0.1365, "step": 20869 }, { "epoch": 15.28937728937729, "grad_norm": 1.15625, "learning_rate": 6.473152551913238e-05, "loss": 0.1142, "step": 20870 }, { "epoch": 15.29010989010989, "grad_norm": 1.1953125, "learning_rate": 6.472824378820435e-05, "loss": 0.1058, "step": 20871 }, { "epoch": 15.290842490842492, "grad_norm": 0.75, "learning_rate": 6.472496198910412e-05, "loss": 0.1108, "step": 20872 }, { "epoch": 15.291575091575092, "grad_norm": 1.5078125, "learning_rate": 6.472168012184744e-05, "loss": 0.1138, "step": 20873 }, { "epoch": 15.292307692307693, "grad_norm": 0.734375, "learning_rate": 6.471839818645001e-05, "loss": 0.1166, "step": 20874 }, { "epoch": 15.293040293040294, "grad_norm": 0.8203125, "learning_rate": 6.47151161829276e-05, "loss": 0.1307, "step": 20875 }, { "epoch": 15.293772893772895, "grad_norm": 0.83203125, "learning_rate": 6.471183411129586e-05, "loss": 0.1023, "step": 20876 }, { "epoch": 15.294505494505495, "grad_norm": 1.234375, "learning_rate": 6.470855197157059e-05, "loss": 0.1343, "step": 20877 }, { "epoch": 15.295238095238096, "grad_norm": 1.34375, "learning_rate": 6.470526976376745e-05, "loss": 0.099, "step": 20878 }, { "epoch": 15.295970695970697, "grad_norm": 0.79296875, "learning_rate": 6.470198748790222e-05, "loss": 0.1242, "step": 20879 }, { "epoch": 15.296703296703297, "grad_norm": 0.7109375, "learning_rate": 6.46987051439906e-05, "loss": 0.1262, "step": 20880 }, { "epoch": 15.297435897435898, "grad_norm": 0.8671875, "learning_rate": 6.469542273204833e-05, "loss": 0.1212, "step": 20881 }, { "epoch": 15.298168498168499, "grad_norm": 0.91796875, "learning_rate": 6.469214025209113e-05, "loss": 0.1181, "step": 20882 }, { "epoch": 15.2989010989011, "grad_norm": 1.421875, "learning_rate": 6.468885770413471e-05, "loss": 0.1168, "step": 20883 }, { "epoch": 15.2996336996337, "grad_norm": 1.421875, "learning_rate": 6.468557508819483e-05, "loss": 0.1651, "step": 20884 }, { "epoch": 15.300366300366301, "grad_norm": 1.3046875, "learning_rate": 6.46822924042872e-05, "loss": 0.1273, "step": 20885 }, { "epoch": 15.301098901098902, "grad_norm": 1.15625, "learning_rate": 6.467900965242756e-05, "loss": 0.1236, "step": 20886 }, { "epoch": 15.301831501831503, "grad_norm": 0.64453125, "learning_rate": 6.467572683263161e-05, "loss": 0.1378, "step": 20887 }, { "epoch": 15.302564102564103, "grad_norm": 0.859375, "learning_rate": 6.467244394491511e-05, "loss": 0.1308, "step": 20888 }, { "epoch": 15.303296703296704, "grad_norm": 1.375, "learning_rate": 6.466916098929378e-05, "loss": 0.1709, "step": 20889 }, { "epoch": 15.304029304029305, "grad_norm": 0.734375, "learning_rate": 6.466587796578335e-05, "loss": 0.1461, "step": 20890 }, { "epoch": 15.304761904761905, "grad_norm": 1.3203125, "learning_rate": 6.466259487439955e-05, "loss": 0.1324, "step": 20891 }, { "epoch": 15.305494505494506, "grad_norm": 1.234375, "learning_rate": 6.465931171515809e-05, "loss": 0.1335, "step": 20892 }, { "epoch": 15.306227106227107, "grad_norm": 0.94921875, "learning_rate": 6.465602848807475e-05, "loss": 0.1065, "step": 20893 }, { "epoch": 15.306959706959708, "grad_norm": 1.296875, "learning_rate": 6.46527451931652e-05, "loss": 0.1134, "step": 20894 }, { "epoch": 15.307692307692308, "grad_norm": 1.125, "learning_rate": 6.464946183044521e-05, "loss": 0.1304, "step": 20895 }, { "epoch": 15.308424908424909, "grad_norm": 1.4296875, "learning_rate": 6.464617839993052e-05, "loss": 0.1236, "step": 20896 }, { "epoch": 15.30915750915751, "grad_norm": 1.46875, "learning_rate": 6.464289490163684e-05, "loss": 0.1086, "step": 20897 }, { "epoch": 15.30989010989011, "grad_norm": 1.3203125, "learning_rate": 6.46396113355799e-05, "loss": 0.1166, "step": 20898 }, { "epoch": 15.310622710622711, "grad_norm": 0.9453125, "learning_rate": 6.463632770177545e-05, "loss": 0.1305, "step": 20899 }, { "epoch": 15.311355311355312, "grad_norm": 1.0859375, "learning_rate": 6.463304400023919e-05, "loss": 0.1369, "step": 20900 }, { "epoch": 15.312087912087913, "grad_norm": 0.828125, "learning_rate": 6.46297602309869e-05, "loss": 0.0953, "step": 20901 }, { "epoch": 15.312820512820513, "grad_norm": 1.125, "learning_rate": 6.462647639403428e-05, "loss": 0.1194, "step": 20902 }, { "epoch": 15.313553113553114, "grad_norm": 1.0625, "learning_rate": 6.462319248939708e-05, "loss": 0.1417, "step": 20903 }, { "epoch": 15.314285714285715, "grad_norm": 0.875, "learning_rate": 6.461990851709101e-05, "loss": 0.1185, "step": 20904 }, { "epoch": 15.315018315018316, "grad_norm": 1.140625, "learning_rate": 6.461662447713184e-05, "loss": 0.1443, "step": 20905 }, { "epoch": 15.315750915750916, "grad_norm": 1.046875, "learning_rate": 6.461334036953529e-05, "loss": 0.1145, "step": 20906 }, { "epoch": 15.316483516483517, "grad_norm": 1.921875, "learning_rate": 6.461005619431711e-05, "loss": 0.1378, "step": 20907 }, { "epoch": 15.317216117216118, "grad_norm": 0.78125, "learning_rate": 6.460677195149297e-05, "loss": 0.1031, "step": 20908 }, { "epoch": 15.317948717948719, "grad_norm": 1.3515625, "learning_rate": 6.46034876410787e-05, "loss": 0.1498, "step": 20909 }, { "epoch": 15.31868131868132, "grad_norm": 0.69921875, "learning_rate": 6.460020326308998e-05, "loss": 0.1314, "step": 20910 }, { "epoch": 15.31941391941392, "grad_norm": 0.8828125, "learning_rate": 6.459691881754253e-05, "loss": 0.098, "step": 20911 }, { "epoch": 15.32014652014652, "grad_norm": 0.72265625, "learning_rate": 6.459363430445214e-05, "loss": 0.1122, "step": 20912 }, { "epoch": 15.320879120879122, "grad_norm": 0.9296875, "learning_rate": 6.459034972383452e-05, "loss": 0.1191, "step": 20913 }, { "epoch": 15.321611721611722, "grad_norm": 1.375, "learning_rate": 6.458706507570541e-05, "loss": 0.1412, "step": 20914 }, { "epoch": 15.322344322344323, "grad_norm": 0.69140625, "learning_rate": 6.458378036008054e-05, "loss": 0.1232, "step": 20915 }, { "epoch": 15.323076923076924, "grad_norm": 1.1484375, "learning_rate": 6.458049557697565e-05, "loss": 0.1252, "step": 20916 }, { "epoch": 15.323809523809524, "grad_norm": 0.87109375, "learning_rate": 6.457721072640649e-05, "loss": 0.1223, "step": 20917 }, { "epoch": 15.324542124542125, "grad_norm": 1.1796875, "learning_rate": 6.457392580838878e-05, "loss": 0.1407, "step": 20918 }, { "epoch": 15.325274725274726, "grad_norm": 1.265625, "learning_rate": 6.45706408229383e-05, "loss": 0.1305, "step": 20919 }, { "epoch": 15.326007326007327, "grad_norm": 1.09375, "learning_rate": 6.456735577007075e-05, "loss": 0.117, "step": 20920 }, { "epoch": 15.326739926739927, "grad_norm": 0.96875, "learning_rate": 6.456407064980187e-05, "loss": 0.1216, "step": 20921 }, { "epoch": 15.327472527472528, "grad_norm": 0.8984375, "learning_rate": 6.456078546214742e-05, "loss": 0.0974, "step": 20922 }, { "epoch": 15.328205128205129, "grad_norm": 1.2109375, "learning_rate": 6.455750020712313e-05, "loss": 0.1346, "step": 20923 }, { "epoch": 15.32893772893773, "grad_norm": 1.265625, "learning_rate": 6.455421488474473e-05, "loss": 0.1349, "step": 20924 }, { "epoch": 15.32967032967033, "grad_norm": 1.15625, "learning_rate": 6.4550929495028e-05, "loss": 0.1031, "step": 20925 }, { "epoch": 15.330402930402931, "grad_norm": 1.140625, "learning_rate": 6.454764403798862e-05, "loss": 0.1104, "step": 20926 }, { "epoch": 15.331135531135532, "grad_norm": 1.0390625, "learning_rate": 6.45443585136424e-05, "loss": 0.096, "step": 20927 }, { "epoch": 15.331868131868132, "grad_norm": 0.81640625, "learning_rate": 6.454107292200501e-05, "loss": 0.1044, "step": 20928 }, { "epoch": 15.332600732600733, "grad_norm": 1.4140625, "learning_rate": 6.453778726309224e-05, "loss": 0.1274, "step": 20929 }, { "epoch": 15.333333333333334, "grad_norm": 0.9453125, "learning_rate": 6.453450153691984e-05, "loss": 0.1475, "step": 20930 }, { "epoch": 15.334065934065935, "grad_norm": 1.03125, "learning_rate": 6.453121574350352e-05, "loss": 0.1316, "step": 20931 }, { "epoch": 15.334798534798535, "grad_norm": 1.25, "learning_rate": 6.452792988285905e-05, "loss": 0.1279, "step": 20932 }, { "epoch": 15.335531135531136, "grad_norm": 0.9609375, "learning_rate": 6.452464395500216e-05, "loss": 0.1207, "step": 20933 }, { "epoch": 15.336263736263737, "grad_norm": 0.80078125, "learning_rate": 6.45213579599486e-05, "loss": 0.1229, "step": 20934 }, { "epoch": 15.336996336996338, "grad_norm": 1.109375, "learning_rate": 6.451807189771411e-05, "loss": 0.1482, "step": 20935 }, { "epoch": 15.337728937728938, "grad_norm": 0.734375, "learning_rate": 6.451478576831443e-05, "loss": 0.11, "step": 20936 }, { "epoch": 15.338461538461539, "grad_norm": 1.328125, "learning_rate": 6.451149957176534e-05, "loss": 0.1428, "step": 20937 }, { "epoch": 15.33919413919414, "grad_norm": 1.0390625, "learning_rate": 6.450821330808253e-05, "loss": 0.1296, "step": 20938 }, { "epoch": 15.33992673992674, "grad_norm": 0.97265625, "learning_rate": 6.450492697728175e-05, "loss": 0.1083, "step": 20939 }, { "epoch": 15.340659340659341, "grad_norm": 0.7734375, "learning_rate": 6.450164057937879e-05, "loss": 0.1111, "step": 20940 }, { "epoch": 15.341391941391942, "grad_norm": 0.80078125, "learning_rate": 6.449835411438939e-05, "loss": 0.1333, "step": 20941 }, { "epoch": 15.342124542124543, "grad_norm": 1.6875, "learning_rate": 6.449506758232924e-05, "loss": 0.1372, "step": 20942 }, { "epoch": 15.342857142857143, "grad_norm": 1.3828125, "learning_rate": 6.449178098321415e-05, "loss": 0.1156, "step": 20943 }, { "epoch": 15.343589743589744, "grad_norm": 0.91796875, "learning_rate": 6.448849431705984e-05, "loss": 0.1115, "step": 20944 }, { "epoch": 15.344322344322345, "grad_norm": 0.71484375, "learning_rate": 6.448520758388208e-05, "loss": 0.1058, "step": 20945 }, { "epoch": 15.345054945054946, "grad_norm": 1.3671875, "learning_rate": 6.448192078369659e-05, "loss": 0.1459, "step": 20946 }, { "epoch": 15.345787545787546, "grad_norm": 1.0703125, "learning_rate": 6.447863391651912e-05, "loss": 0.1201, "step": 20947 }, { "epoch": 15.346520146520147, "grad_norm": 0.8515625, "learning_rate": 6.447534698236543e-05, "loss": 0.1072, "step": 20948 }, { "epoch": 15.347252747252748, "grad_norm": 0.90625, "learning_rate": 6.447205998125126e-05, "loss": 0.1395, "step": 20949 }, { "epoch": 15.347985347985349, "grad_norm": 1.140625, "learning_rate": 6.446877291319237e-05, "loss": 0.1365, "step": 20950 }, { "epoch": 15.34871794871795, "grad_norm": 0.6796875, "learning_rate": 6.44654857782045e-05, "loss": 0.1334, "step": 20951 }, { "epoch": 15.34945054945055, "grad_norm": 1.1171875, "learning_rate": 6.44621985763034e-05, "loss": 0.1258, "step": 20952 }, { "epoch": 15.35018315018315, "grad_norm": 0.9921875, "learning_rate": 6.445891130750483e-05, "loss": 0.1313, "step": 20953 }, { "epoch": 15.350915750915751, "grad_norm": 1.4140625, "learning_rate": 6.445562397182452e-05, "loss": 0.1453, "step": 20954 }, { "epoch": 15.351648351648352, "grad_norm": 1.2421875, "learning_rate": 6.445233656927825e-05, "loss": 0.1496, "step": 20955 }, { "epoch": 15.352380952380953, "grad_norm": 1.3046875, "learning_rate": 6.444904909988175e-05, "loss": 0.1391, "step": 20956 }, { "epoch": 15.353113553113554, "grad_norm": 0.7265625, "learning_rate": 6.444576156365078e-05, "loss": 0.0896, "step": 20957 }, { "epoch": 15.353846153846154, "grad_norm": 1.0859375, "learning_rate": 6.444247396060108e-05, "loss": 0.1435, "step": 20958 }, { "epoch": 15.354578754578755, "grad_norm": 0.80078125, "learning_rate": 6.443918629074842e-05, "loss": 0.1259, "step": 20959 }, { "epoch": 15.355311355311356, "grad_norm": 1.28125, "learning_rate": 6.443589855410852e-05, "loss": 0.1101, "step": 20960 }, { "epoch": 15.356043956043957, "grad_norm": 0.92578125, "learning_rate": 6.443261075069719e-05, "loss": 0.1274, "step": 20961 }, { "epoch": 15.356776556776557, "grad_norm": 1.34375, "learning_rate": 6.442932288053014e-05, "loss": 0.1577, "step": 20962 }, { "epoch": 15.357509157509158, "grad_norm": 1.609375, "learning_rate": 6.44260349436231e-05, "loss": 0.1405, "step": 20963 }, { "epoch": 15.358241758241759, "grad_norm": 1.2734375, "learning_rate": 6.44227469399919e-05, "loss": 0.1286, "step": 20964 }, { "epoch": 15.35897435897436, "grad_norm": 1.7890625, "learning_rate": 6.441945886965222e-05, "loss": 0.1795, "step": 20965 }, { "epoch": 15.35970695970696, "grad_norm": 0.9453125, "learning_rate": 6.441617073261984e-05, "loss": 0.1279, "step": 20966 }, { "epoch": 15.360439560439561, "grad_norm": 0.87109375, "learning_rate": 6.441288252891053e-05, "loss": 0.1185, "step": 20967 }, { "epoch": 15.361172161172162, "grad_norm": 1.125, "learning_rate": 6.440959425854001e-05, "loss": 0.1315, "step": 20968 }, { "epoch": 15.361904761904762, "grad_norm": 1.265625, "learning_rate": 6.44063059215241e-05, "loss": 0.1039, "step": 20969 }, { "epoch": 15.362637362637363, "grad_norm": 0.8984375, "learning_rate": 6.440301751787848e-05, "loss": 0.0922, "step": 20970 }, { "epoch": 15.363369963369964, "grad_norm": 1.203125, "learning_rate": 6.439972904761895e-05, "loss": 0.0999, "step": 20971 }, { "epoch": 15.364102564102565, "grad_norm": 1.265625, "learning_rate": 6.439644051076125e-05, "loss": 0.117, "step": 20972 }, { "epoch": 15.364835164835165, "grad_norm": 1.5078125, "learning_rate": 6.439315190732115e-05, "loss": 0.1179, "step": 20973 }, { "epoch": 15.365567765567766, "grad_norm": 0.76171875, "learning_rate": 6.438986323731439e-05, "loss": 0.1147, "step": 20974 }, { "epoch": 15.366300366300367, "grad_norm": 1.0625, "learning_rate": 6.438657450075674e-05, "loss": 0.1074, "step": 20975 }, { "epoch": 15.367032967032968, "grad_norm": 1.1171875, "learning_rate": 6.438328569766397e-05, "loss": 0.1688, "step": 20976 }, { "epoch": 15.367765567765568, "grad_norm": 0.84765625, "learning_rate": 6.437999682805179e-05, "loss": 0.1047, "step": 20977 }, { "epoch": 15.368498168498169, "grad_norm": 0.7109375, "learning_rate": 6.4376707891936e-05, "loss": 0.1056, "step": 20978 }, { "epoch": 15.36923076923077, "grad_norm": 0.953125, "learning_rate": 6.437341888933235e-05, "loss": 0.1349, "step": 20979 }, { "epoch": 15.36996336996337, "grad_norm": 0.8046875, "learning_rate": 6.437012982025659e-05, "loss": 0.1308, "step": 20980 }, { "epoch": 15.370695970695971, "grad_norm": 1.3125, "learning_rate": 6.436684068472448e-05, "loss": 0.1662, "step": 20981 }, { "epoch": 15.371428571428572, "grad_norm": 1.328125, "learning_rate": 6.43635514827518e-05, "loss": 0.1362, "step": 20982 }, { "epoch": 15.372161172161173, "grad_norm": 0.75, "learning_rate": 6.436026221435429e-05, "loss": 0.1206, "step": 20983 }, { "epoch": 15.372893772893773, "grad_norm": 1.03125, "learning_rate": 6.435697287954772e-05, "loss": 0.1569, "step": 20984 }, { "epoch": 15.373626373626374, "grad_norm": 0.64453125, "learning_rate": 6.435368347834784e-05, "loss": 0.107, "step": 20985 }, { "epoch": 15.374358974358975, "grad_norm": 0.8203125, "learning_rate": 6.43503940107704e-05, "loss": 0.1152, "step": 20986 }, { "epoch": 15.375091575091576, "grad_norm": 0.71875, "learning_rate": 6.434710447683118e-05, "loss": 0.1028, "step": 20987 }, { "epoch": 15.375824175824176, "grad_norm": 1.4609375, "learning_rate": 6.434381487654596e-05, "loss": 0.1576, "step": 20988 }, { "epoch": 15.376556776556777, "grad_norm": 1.1796875, "learning_rate": 6.434052520993045e-05, "loss": 0.1232, "step": 20989 }, { "epoch": 15.377289377289378, "grad_norm": 0.65625, "learning_rate": 6.433723547700045e-05, "loss": 0.122, "step": 20990 }, { "epoch": 15.378021978021978, "grad_norm": 0.64453125, "learning_rate": 6.433394567777171e-05, "loss": 0.1181, "step": 20991 }, { "epoch": 15.37875457875458, "grad_norm": 1.0703125, "learning_rate": 6.433065581226e-05, "loss": 0.1141, "step": 20992 }, { "epoch": 15.37948717948718, "grad_norm": 0.9296875, "learning_rate": 6.432736588048107e-05, "loss": 0.1379, "step": 20993 }, { "epoch": 15.38021978021978, "grad_norm": 1.53125, "learning_rate": 6.432407588245069e-05, "loss": 0.14, "step": 20994 }, { "epoch": 15.380952380952381, "grad_norm": 1.53125, "learning_rate": 6.432078581818463e-05, "loss": 0.1271, "step": 20995 }, { "epoch": 15.381684981684982, "grad_norm": 1.2109375, "learning_rate": 6.431749568769865e-05, "loss": 0.1417, "step": 20996 }, { "epoch": 15.382417582417583, "grad_norm": 1.5390625, "learning_rate": 6.431420549100852e-05, "loss": 0.1837, "step": 20997 }, { "epoch": 15.383150183150184, "grad_norm": 0.67578125, "learning_rate": 6.431091522812999e-05, "loss": 0.1031, "step": 20998 }, { "epoch": 15.383882783882784, "grad_norm": 0.8671875, "learning_rate": 6.430762489907882e-05, "loss": 0.1259, "step": 20999 }, { "epoch": 15.384615384615385, "grad_norm": 1.1015625, "learning_rate": 6.43043345038708e-05, "loss": 0.1324, "step": 21000 }, { "epoch": 15.385347985347986, "grad_norm": 0.98046875, "learning_rate": 6.430104404252168e-05, "loss": 0.113, "step": 21001 }, { "epoch": 15.386080586080586, "grad_norm": 0.81640625, "learning_rate": 6.429775351504722e-05, "loss": 0.1205, "step": 21002 }, { "epoch": 15.386813186813187, "grad_norm": 1.4375, "learning_rate": 6.429446292146321e-05, "loss": 0.1228, "step": 21003 }, { "epoch": 15.387545787545788, "grad_norm": 1.296875, "learning_rate": 6.429117226178537e-05, "loss": 0.1242, "step": 21004 }, { "epoch": 15.388278388278389, "grad_norm": 1.0078125, "learning_rate": 6.428788153602952e-05, "loss": 0.1164, "step": 21005 }, { "epoch": 15.38901098901099, "grad_norm": 1.234375, "learning_rate": 6.42845907442114e-05, "loss": 0.1352, "step": 21006 }, { "epoch": 15.38974358974359, "grad_norm": 0.69140625, "learning_rate": 6.428129988634677e-05, "loss": 0.1057, "step": 21007 }, { "epoch": 15.39047619047619, "grad_norm": 1.1171875, "learning_rate": 6.427800896245141e-05, "loss": 0.1224, "step": 21008 }, { "epoch": 15.391208791208792, "grad_norm": 1.25, "learning_rate": 6.427471797254108e-05, "loss": 0.1333, "step": 21009 }, { "epoch": 15.391941391941392, "grad_norm": 0.92578125, "learning_rate": 6.427142691663158e-05, "loss": 0.1108, "step": 21010 }, { "epoch": 15.392673992673993, "grad_norm": 1.203125, "learning_rate": 6.426813579473864e-05, "loss": 0.1115, "step": 21011 }, { "epoch": 15.393406593406594, "grad_norm": 0.71484375, "learning_rate": 6.426484460687804e-05, "loss": 0.1204, "step": 21012 }, { "epoch": 15.394139194139195, "grad_norm": 0.92578125, "learning_rate": 6.426155335306556e-05, "loss": 0.1172, "step": 21013 }, { "epoch": 15.394871794871795, "grad_norm": 1.09375, "learning_rate": 6.425826203331695e-05, "loss": 0.1479, "step": 21014 }, { "epoch": 15.395604395604396, "grad_norm": 0.90625, "learning_rate": 6.425497064764799e-05, "loss": 0.1197, "step": 21015 }, { "epoch": 15.396336996336997, "grad_norm": 0.9296875, "learning_rate": 6.425167919607446e-05, "loss": 0.1176, "step": 21016 }, { "epoch": 15.397069597069597, "grad_norm": 0.9453125, "learning_rate": 6.42483876786121e-05, "loss": 0.1166, "step": 21017 }, { "epoch": 15.397802197802198, "grad_norm": 0.57421875, "learning_rate": 6.424509609527672e-05, "loss": 0.1037, "step": 21018 }, { "epoch": 15.398534798534799, "grad_norm": 1.203125, "learning_rate": 6.424180444608407e-05, "loss": 0.146, "step": 21019 }, { "epoch": 15.3992673992674, "grad_norm": 0.94921875, "learning_rate": 6.42385127310499e-05, "loss": 0.1262, "step": 21020 }, { "epoch": 15.4, "grad_norm": 1.0546875, "learning_rate": 6.423522095019004e-05, "loss": 0.1603, "step": 21021 }, { "epoch": 15.400732600732601, "grad_norm": 1.125, "learning_rate": 6.423192910352021e-05, "loss": 0.1361, "step": 21022 }, { "epoch": 15.401465201465202, "grad_norm": 0.76953125, "learning_rate": 6.42286371910562e-05, "loss": 0.1123, "step": 21023 }, { "epoch": 15.402197802197803, "grad_norm": 1.140625, "learning_rate": 6.422534521281378e-05, "loss": 0.1091, "step": 21024 }, { "epoch": 15.402930402930403, "grad_norm": 0.9296875, "learning_rate": 6.422205316880872e-05, "loss": 0.0951, "step": 21025 }, { "epoch": 15.403663003663004, "grad_norm": 1.21875, "learning_rate": 6.421876105905681e-05, "loss": 0.1499, "step": 21026 }, { "epoch": 15.404395604395605, "grad_norm": 1.9375, "learning_rate": 6.421546888357382e-05, "loss": 0.1525, "step": 21027 }, { "epoch": 15.405128205128205, "grad_norm": 0.82421875, "learning_rate": 6.42121766423755e-05, "loss": 0.0951, "step": 21028 }, { "epoch": 15.405860805860806, "grad_norm": 1.078125, "learning_rate": 6.420888433547766e-05, "loss": 0.1016, "step": 21029 }, { "epoch": 15.406593406593407, "grad_norm": 1.015625, "learning_rate": 6.420559196289603e-05, "loss": 0.13, "step": 21030 }, { "epoch": 15.407326007326008, "grad_norm": 1.0625, "learning_rate": 6.420229952464642e-05, "loss": 0.1361, "step": 21031 }, { "epoch": 15.408058608058608, "grad_norm": 1.3359375, "learning_rate": 6.419900702074461e-05, "loss": 0.1337, "step": 21032 }, { "epoch": 15.40879120879121, "grad_norm": 1.046875, "learning_rate": 6.419571445120633e-05, "loss": 0.1263, "step": 21033 }, { "epoch": 15.40952380952381, "grad_norm": 1.1796875, "learning_rate": 6.419242181604742e-05, "loss": 0.1028, "step": 21034 }, { "epoch": 15.41025641025641, "grad_norm": 1.359375, "learning_rate": 6.41891291152836e-05, "loss": 0.1197, "step": 21035 }, { "epoch": 15.410989010989011, "grad_norm": 1.296875, "learning_rate": 6.418583634893067e-05, "loss": 0.119, "step": 21036 }, { "epoch": 15.411721611721612, "grad_norm": 0.79296875, "learning_rate": 6.418254351700441e-05, "loss": 0.1224, "step": 21037 }, { "epoch": 15.412454212454213, "grad_norm": 1.8984375, "learning_rate": 6.417925061952059e-05, "loss": 0.1244, "step": 21038 }, { "epoch": 15.413186813186813, "grad_norm": 0.9375, "learning_rate": 6.417595765649499e-05, "loss": 0.159, "step": 21039 }, { "epoch": 15.413919413919414, "grad_norm": 0.97265625, "learning_rate": 6.41726646279434e-05, "loss": 0.1251, "step": 21040 }, { "epoch": 15.414652014652015, "grad_norm": 1.234375, "learning_rate": 6.416937153388156e-05, "loss": 0.1281, "step": 21041 }, { "epoch": 15.415384615384616, "grad_norm": 0.74609375, "learning_rate": 6.41660783743253e-05, "loss": 0.1014, "step": 21042 }, { "epoch": 15.416117216117216, "grad_norm": 0.953125, "learning_rate": 6.416278514929035e-05, "loss": 0.1054, "step": 21043 }, { "epoch": 15.416849816849817, "grad_norm": 0.84375, "learning_rate": 6.415949185879253e-05, "loss": 0.1395, "step": 21044 }, { "epoch": 15.417582417582418, "grad_norm": 1.46875, "learning_rate": 6.415619850284758e-05, "loss": 0.1429, "step": 21045 }, { "epoch": 15.418315018315019, "grad_norm": 0.93359375, "learning_rate": 6.415290508147131e-05, "loss": 0.1151, "step": 21046 }, { "epoch": 15.41904761904762, "grad_norm": 0.80859375, "learning_rate": 6.41496115946795e-05, "loss": 0.136, "step": 21047 }, { "epoch": 15.41978021978022, "grad_norm": 0.796875, "learning_rate": 6.41463180424879e-05, "loss": 0.1161, "step": 21048 }, { "epoch": 15.42051282051282, "grad_norm": 1.2578125, "learning_rate": 6.414302442491233e-05, "loss": 0.1292, "step": 21049 }, { "epoch": 15.421245421245422, "grad_norm": 0.9921875, "learning_rate": 6.413973074196854e-05, "loss": 0.1343, "step": 21050 }, { "epoch": 15.421978021978022, "grad_norm": 0.69921875, "learning_rate": 6.413643699367232e-05, "loss": 0.1186, "step": 21051 }, { "epoch": 15.422710622710623, "grad_norm": 1.3515625, "learning_rate": 6.413314318003947e-05, "loss": 0.1282, "step": 21052 }, { "epoch": 15.423443223443224, "grad_norm": 1.0703125, "learning_rate": 6.412984930108574e-05, "loss": 0.1126, "step": 21053 }, { "epoch": 15.424175824175824, "grad_norm": 0.8984375, "learning_rate": 6.412655535682692e-05, "loss": 0.1231, "step": 21054 }, { "epoch": 15.424908424908425, "grad_norm": 1.1875, "learning_rate": 6.412326134727882e-05, "loss": 0.1035, "step": 21055 }, { "epoch": 15.425641025641026, "grad_norm": 1.234375, "learning_rate": 6.41199672724572e-05, "loss": 0.1328, "step": 21056 }, { "epoch": 15.426373626373627, "grad_norm": 1.1640625, "learning_rate": 6.411667313237782e-05, "loss": 0.1435, "step": 21057 }, { "epoch": 15.427106227106227, "grad_norm": 0.953125, "learning_rate": 6.411337892705652e-05, "loss": 0.1267, "step": 21058 }, { "epoch": 15.427838827838828, "grad_norm": 0.75390625, "learning_rate": 6.411008465650904e-05, "loss": 0.1036, "step": 21059 }, { "epoch": 15.428571428571429, "grad_norm": 1.53125, "learning_rate": 6.410679032075117e-05, "loss": 0.1374, "step": 21060 }, { "epoch": 15.42930402930403, "grad_norm": 1.140625, "learning_rate": 6.410349591979871e-05, "loss": 0.1286, "step": 21061 }, { "epoch": 15.43003663003663, "grad_norm": 1.1796875, "learning_rate": 6.410020145366744e-05, "loss": 0.1421, "step": 21062 }, { "epoch": 15.430769230769231, "grad_norm": 1.234375, "learning_rate": 6.409690692237313e-05, "loss": 0.1051, "step": 21063 }, { "epoch": 15.431501831501832, "grad_norm": 1.1640625, "learning_rate": 6.409361232593159e-05, "loss": 0.1122, "step": 21064 }, { "epoch": 15.432234432234432, "grad_norm": 1.984375, "learning_rate": 6.409031766435858e-05, "loss": 0.1102, "step": 21065 }, { "epoch": 15.432967032967033, "grad_norm": 1.25, "learning_rate": 6.40870229376699e-05, "loss": 0.1177, "step": 21066 }, { "epoch": 15.433699633699634, "grad_norm": 1.0390625, "learning_rate": 6.408372814588132e-05, "loss": 0.1068, "step": 21067 }, { "epoch": 15.434432234432235, "grad_norm": 0.7734375, "learning_rate": 6.408043328900866e-05, "loss": 0.1271, "step": 21068 }, { "epoch": 15.435164835164835, "grad_norm": 1.3515625, "learning_rate": 6.407713836706767e-05, "loss": 0.1177, "step": 21069 }, { "epoch": 15.435897435897436, "grad_norm": 1.234375, "learning_rate": 6.407384338007413e-05, "loss": 0.1373, "step": 21070 }, { "epoch": 15.436630036630037, "grad_norm": 1.140625, "learning_rate": 6.40705483280439e-05, "loss": 0.1311, "step": 21071 }, { "epoch": 15.437362637362638, "grad_norm": 1.0, "learning_rate": 6.40672532109927e-05, "loss": 0.129, "step": 21072 }, { "epoch": 15.438095238095238, "grad_norm": 1.4375, "learning_rate": 6.406395802893633e-05, "loss": 0.1506, "step": 21073 }, { "epoch": 15.438827838827839, "grad_norm": 1.1953125, "learning_rate": 6.406066278189057e-05, "loss": 0.152, "step": 21074 }, { "epoch": 15.43956043956044, "grad_norm": 0.6953125, "learning_rate": 6.405736746987125e-05, "loss": 0.1106, "step": 21075 }, { "epoch": 15.44029304029304, "grad_norm": 1.234375, "learning_rate": 6.405407209289412e-05, "loss": 0.1624, "step": 21076 }, { "epoch": 15.441025641025641, "grad_norm": 1.421875, "learning_rate": 6.405077665097499e-05, "loss": 0.1597, "step": 21077 }, { "epoch": 15.441758241758242, "grad_norm": 1.4375, "learning_rate": 6.404748114412963e-05, "loss": 0.1686, "step": 21078 }, { "epoch": 15.442490842490843, "grad_norm": 1.0078125, "learning_rate": 6.404418557237385e-05, "loss": 0.1237, "step": 21079 }, { "epoch": 15.443223443223443, "grad_norm": 1.0546875, "learning_rate": 6.404088993572342e-05, "loss": 0.1217, "step": 21080 }, { "epoch": 15.443956043956044, "grad_norm": 1.4140625, "learning_rate": 6.403759423419416e-05, "loss": 0.1394, "step": 21081 }, { "epoch": 15.444688644688645, "grad_norm": 0.890625, "learning_rate": 6.403429846780181e-05, "loss": 0.1089, "step": 21082 }, { "epoch": 15.445421245421246, "grad_norm": 1.125, "learning_rate": 6.40310026365622e-05, "loss": 0.1592, "step": 21083 }, { "epoch": 15.446153846153846, "grad_norm": 1.1171875, "learning_rate": 6.402770674049115e-05, "loss": 0.0998, "step": 21084 }, { "epoch": 15.446886446886447, "grad_norm": 1.203125, "learning_rate": 6.402441077960436e-05, "loss": 0.1116, "step": 21085 }, { "epoch": 15.447619047619048, "grad_norm": 1.375, "learning_rate": 6.402111475391772e-05, "loss": 0.0924, "step": 21086 }, { "epoch": 15.448351648351649, "grad_norm": 1.0625, "learning_rate": 6.401781866344698e-05, "loss": 0.1228, "step": 21087 }, { "epoch": 15.44908424908425, "grad_norm": 1.140625, "learning_rate": 6.401452250820793e-05, "loss": 0.1224, "step": 21088 }, { "epoch": 15.44981684981685, "grad_norm": 0.953125, "learning_rate": 6.401122628821635e-05, "loss": 0.1241, "step": 21089 }, { "epoch": 15.45054945054945, "grad_norm": 1.25, "learning_rate": 6.400793000348805e-05, "loss": 0.1454, "step": 21090 }, { "epoch": 15.451282051282051, "grad_norm": 1.5625, "learning_rate": 6.400463365403885e-05, "loss": 0.1471, "step": 21091 }, { "epoch": 15.452014652014652, "grad_norm": 0.85546875, "learning_rate": 6.40013372398845e-05, "loss": 0.1141, "step": 21092 }, { "epoch": 15.452747252747253, "grad_norm": 1.2734375, "learning_rate": 6.39980407610408e-05, "loss": 0.1418, "step": 21093 }, { "epoch": 15.453479853479854, "grad_norm": 1.1015625, "learning_rate": 6.399474421752357e-05, "loss": 0.107, "step": 21094 }, { "epoch": 15.454212454212454, "grad_norm": 0.87109375, "learning_rate": 6.399144760934858e-05, "loss": 0.1333, "step": 21095 }, { "epoch": 15.454945054945055, "grad_norm": 1.0234375, "learning_rate": 6.398815093653165e-05, "loss": 0.1287, "step": 21096 }, { "epoch": 15.455677655677656, "grad_norm": 1.2265625, "learning_rate": 6.398485419908854e-05, "loss": 0.094, "step": 21097 }, { "epoch": 15.456410256410257, "grad_norm": 1.53125, "learning_rate": 6.398155739703507e-05, "loss": 0.116, "step": 21098 }, { "epoch": 15.457142857142857, "grad_norm": 0.9609375, "learning_rate": 6.397826053038705e-05, "loss": 0.1292, "step": 21099 }, { "epoch": 15.457875457875458, "grad_norm": 1.203125, "learning_rate": 6.397496359916026e-05, "loss": 0.1586, "step": 21100 }, { "epoch": 15.458608058608059, "grad_norm": 1.140625, "learning_rate": 6.397166660337047e-05, "loss": 0.1348, "step": 21101 }, { "epoch": 15.45934065934066, "grad_norm": 1.4921875, "learning_rate": 6.396836954303354e-05, "loss": 0.1381, "step": 21102 }, { "epoch": 15.46007326007326, "grad_norm": 0.78125, "learning_rate": 6.396507241816519e-05, "loss": 0.1272, "step": 21103 }, { "epoch": 15.460805860805861, "grad_norm": 0.97265625, "learning_rate": 6.39617752287813e-05, "loss": 0.13, "step": 21104 }, { "epoch": 15.461538461538462, "grad_norm": 1.4921875, "learning_rate": 6.39584779748976e-05, "loss": 0.1151, "step": 21105 }, { "epoch": 15.462271062271062, "grad_norm": 0.87890625, "learning_rate": 6.39551806565299e-05, "loss": 0.1188, "step": 21106 }, { "epoch": 15.463003663003663, "grad_norm": 1.1875, "learning_rate": 6.395188327369404e-05, "loss": 0.1273, "step": 21107 }, { "epoch": 15.463736263736264, "grad_norm": 0.69140625, "learning_rate": 6.394858582640578e-05, "loss": 0.092, "step": 21108 }, { "epoch": 15.464468864468865, "grad_norm": 1.25, "learning_rate": 6.394528831468091e-05, "loss": 0.117, "step": 21109 }, { "epoch": 15.465201465201465, "grad_norm": 0.984375, "learning_rate": 6.394199073853528e-05, "loss": 0.1224, "step": 21110 }, { "epoch": 15.465934065934066, "grad_norm": 1.34375, "learning_rate": 6.393869309798464e-05, "loss": 0.1324, "step": 21111 }, { "epoch": 15.466666666666667, "grad_norm": 1.3984375, "learning_rate": 6.393539539304483e-05, "loss": 0.1359, "step": 21112 }, { "epoch": 15.467399267399268, "grad_norm": 1.46875, "learning_rate": 6.393209762373163e-05, "loss": 0.1655, "step": 21113 }, { "epoch": 15.468131868131868, "grad_norm": 1.0859375, "learning_rate": 6.392879979006083e-05, "loss": 0.1044, "step": 21114 }, { "epoch": 15.468864468864469, "grad_norm": 0.98046875, "learning_rate": 6.392550189204825e-05, "loss": 0.13, "step": 21115 }, { "epoch": 15.46959706959707, "grad_norm": 0.93359375, "learning_rate": 6.392220392970967e-05, "loss": 0.1138, "step": 21116 }, { "epoch": 15.47032967032967, "grad_norm": 0.90234375, "learning_rate": 6.39189059030609e-05, "loss": 0.1448, "step": 21117 }, { "epoch": 15.471062271062271, "grad_norm": 0.83203125, "learning_rate": 6.391560781211779e-05, "loss": 0.1103, "step": 21118 }, { "epoch": 15.471794871794872, "grad_norm": 1.0234375, "learning_rate": 6.391230965689604e-05, "loss": 0.1171, "step": 21119 }, { "epoch": 15.472527472527473, "grad_norm": 1.140625, "learning_rate": 6.390901143741155e-05, "loss": 0.1318, "step": 21120 }, { "epoch": 15.473260073260073, "grad_norm": 0.99609375, "learning_rate": 6.390571315368006e-05, "loss": 0.1139, "step": 21121 }, { "epoch": 15.473992673992674, "grad_norm": 1.3125, "learning_rate": 6.39024148057174e-05, "loss": 0.1117, "step": 21122 }, { "epoch": 15.474725274725275, "grad_norm": 1.3359375, "learning_rate": 6.38991163935394e-05, "loss": 0.145, "step": 21123 }, { "epoch": 15.475457875457876, "grad_norm": 1.2265625, "learning_rate": 6.38958179171618e-05, "loss": 0.1234, "step": 21124 }, { "epoch": 15.476190476190476, "grad_norm": 0.671875, "learning_rate": 6.389251937660047e-05, "loss": 0.0949, "step": 21125 }, { "epoch": 15.476923076923077, "grad_norm": 0.72265625, "learning_rate": 6.388922077187116e-05, "loss": 0.1211, "step": 21126 }, { "epoch": 15.477655677655678, "grad_norm": 0.81640625, "learning_rate": 6.388592210298971e-05, "loss": 0.1166, "step": 21127 }, { "epoch": 15.478388278388278, "grad_norm": 1.4921875, "learning_rate": 6.38826233699719e-05, "loss": 0.1742, "step": 21128 }, { "epoch": 15.47912087912088, "grad_norm": 0.96484375, "learning_rate": 6.387932457283355e-05, "loss": 0.1329, "step": 21129 }, { "epoch": 15.47985347985348, "grad_norm": 1.2578125, "learning_rate": 6.387602571159049e-05, "loss": 0.1044, "step": 21130 }, { "epoch": 15.48058608058608, "grad_norm": 1.2109375, "learning_rate": 6.387272678625847e-05, "loss": 0.1254, "step": 21131 }, { "epoch": 15.481318681318681, "grad_norm": 1.1328125, "learning_rate": 6.386942779685333e-05, "loss": 0.1123, "step": 21132 }, { "epoch": 15.482051282051282, "grad_norm": 1.7890625, "learning_rate": 6.38661287433909e-05, "loss": 0.1277, "step": 21133 }, { "epoch": 15.482783882783883, "grad_norm": 1.2890625, "learning_rate": 6.386282962588693e-05, "loss": 0.1277, "step": 21134 }, { "epoch": 15.483516483516484, "grad_norm": 1.3671875, "learning_rate": 6.385953044435725e-05, "loss": 0.1414, "step": 21135 }, { "epoch": 15.484249084249084, "grad_norm": 1.0234375, "learning_rate": 6.385623119881769e-05, "loss": 0.1244, "step": 21136 }, { "epoch": 15.484981684981685, "grad_norm": 1.96875, "learning_rate": 6.385293188928405e-05, "loss": 0.1398, "step": 21137 }, { "epoch": 15.485714285714286, "grad_norm": 1.3515625, "learning_rate": 6.384963251577212e-05, "loss": 0.1469, "step": 21138 }, { "epoch": 15.486446886446886, "grad_norm": 1.8359375, "learning_rate": 6.384633307829773e-05, "loss": 0.1163, "step": 21139 }, { "epoch": 15.487179487179487, "grad_norm": 1.2578125, "learning_rate": 6.384303357687665e-05, "loss": 0.0922, "step": 21140 }, { "epoch": 15.487912087912088, "grad_norm": 1.9765625, "learning_rate": 6.383973401152473e-05, "loss": 0.1453, "step": 21141 }, { "epoch": 15.488644688644689, "grad_norm": 2.171875, "learning_rate": 6.383643438225776e-05, "loss": 0.1127, "step": 21142 }, { "epoch": 15.48937728937729, "grad_norm": 1.5078125, "learning_rate": 6.383313468909158e-05, "loss": 0.1356, "step": 21143 }, { "epoch": 15.49010989010989, "grad_norm": 1.515625, "learning_rate": 6.382983493204195e-05, "loss": 0.1338, "step": 21144 }, { "epoch": 15.49084249084249, "grad_norm": 1.140625, "learning_rate": 6.382653511112469e-05, "loss": 0.1221, "step": 21145 }, { "epoch": 15.491575091575092, "grad_norm": 1.0703125, "learning_rate": 6.382323522635568e-05, "loss": 0.1349, "step": 21146 }, { "epoch": 15.492307692307692, "grad_norm": 0.9140625, "learning_rate": 6.381993527775063e-05, "loss": 0.1715, "step": 21147 }, { "epoch": 15.493040293040293, "grad_norm": 1.1796875, "learning_rate": 6.38166352653254e-05, "loss": 0.1129, "step": 21148 }, { "epoch": 15.493772893772894, "grad_norm": 1.1484375, "learning_rate": 6.381333518909582e-05, "loss": 0.1788, "step": 21149 }, { "epoch": 15.494505494505495, "grad_norm": 1.078125, "learning_rate": 6.381003504907765e-05, "loss": 0.1299, "step": 21150 }, { "epoch": 15.495238095238095, "grad_norm": 0.5859375, "learning_rate": 6.380673484528676e-05, "loss": 0.1241, "step": 21151 }, { "epoch": 15.495970695970696, "grad_norm": 1.765625, "learning_rate": 6.380343457773895e-05, "loss": 0.117, "step": 21152 }, { "epoch": 15.496703296703297, "grad_norm": 0.58984375, "learning_rate": 6.380013424644998e-05, "loss": 0.1238, "step": 21153 }, { "epoch": 15.497435897435897, "grad_norm": 1.6640625, "learning_rate": 6.37968338514357e-05, "loss": 0.1242, "step": 21154 }, { "epoch": 15.498168498168498, "grad_norm": 1.5, "learning_rate": 6.379353339271195e-05, "loss": 0.1408, "step": 21155 }, { "epoch": 15.498901098901099, "grad_norm": 1.6640625, "learning_rate": 6.379023287029452e-05, "loss": 0.1155, "step": 21156 }, { "epoch": 15.4996336996337, "grad_norm": 1.9609375, "learning_rate": 6.37869322841992e-05, "loss": 0.1343, "step": 21157 }, { "epoch": 15.5003663003663, "grad_norm": 1.09375, "learning_rate": 6.378363163444183e-05, "loss": 0.1061, "step": 21158 }, { "epoch": 15.501098901098901, "grad_norm": 0.88671875, "learning_rate": 6.378033092103823e-05, "loss": 0.1165, "step": 21159 }, { "epoch": 15.501831501831502, "grad_norm": 1.171875, "learning_rate": 6.377703014400419e-05, "loss": 0.1172, "step": 21160 }, { "epoch": 15.502564102564103, "grad_norm": 1.0078125, "learning_rate": 6.377372930335554e-05, "loss": 0.1219, "step": 21161 }, { "epoch": 15.503296703296703, "grad_norm": 1.8359375, "learning_rate": 6.377042839910813e-05, "loss": 0.1365, "step": 21162 }, { "epoch": 15.504029304029304, "grad_norm": 1.9765625, "learning_rate": 6.37671274312777e-05, "loss": 0.1691, "step": 21163 }, { "epoch": 15.504761904761905, "grad_norm": 1.0546875, "learning_rate": 6.376382639988012e-05, "loss": 0.1291, "step": 21164 }, { "epoch": 15.505494505494505, "grad_norm": 1.3125, "learning_rate": 6.376052530493121e-05, "loss": 0.1215, "step": 21165 }, { "epoch": 15.506227106227106, "grad_norm": 1.5, "learning_rate": 6.375722414644674e-05, "loss": 0.146, "step": 21166 }, { "epoch": 15.506959706959707, "grad_norm": 0.94921875, "learning_rate": 6.375392292444258e-05, "loss": 0.1238, "step": 21167 }, { "epoch": 15.507692307692308, "grad_norm": 1.0078125, "learning_rate": 6.375062163893452e-05, "loss": 0.1002, "step": 21168 }, { "epoch": 15.508424908424908, "grad_norm": 1.1171875, "learning_rate": 6.374732028993838e-05, "loss": 0.1176, "step": 21169 }, { "epoch": 15.50915750915751, "grad_norm": 1.1953125, "learning_rate": 6.374401887746999e-05, "loss": 0.1173, "step": 21170 }, { "epoch": 15.50989010989011, "grad_norm": 1.3671875, "learning_rate": 6.374071740154513e-05, "loss": 0.1258, "step": 21171 }, { "epoch": 15.51062271062271, "grad_norm": 1.03125, "learning_rate": 6.373741586217966e-05, "loss": 0.1534, "step": 21172 }, { "epoch": 15.511355311355311, "grad_norm": 0.9921875, "learning_rate": 6.37341142593894e-05, "loss": 0.1328, "step": 21173 }, { "epoch": 15.512087912087912, "grad_norm": 1.078125, "learning_rate": 6.373081259319014e-05, "loss": 0.1028, "step": 21174 }, { "epoch": 15.512820512820513, "grad_norm": 1.5, "learning_rate": 6.372751086359772e-05, "loss": 0.1431, "step": 21175 }, { "epoch": 15.513553113553114, "grad_norm": 1.453125, "learning_rate": 6.372420907062793e-05, "loss": 0.1078, "step": 21176 }, { "epoch": 15.514285714285714, "grad_norm": 1.2109375, "learning_rate": 6.372090721429664e-05, "loss": 0.1308, "step": 21177 }, { "epoch": 15.515018315018315, "grad_norm": 0.71875, "learning_rate": 6.371760529461966e-05, "loss": 0.1324, "step": 21178 }, { "epoch": 15.515750915750916, "grad_norm": 0.75390625, "learning_rate": 6.371430331161276e-05, "loss": 0.123, "step": 21179 }, { "epoch": 15.516483516483516, "grad_norm": 1.0234375, "learning_rate": 6.371100126529179e-05, "loss": 0.1254, "step": 21180 }, { "epoch": 15.517216117216117, "grad_norm": 1.0546875, "learning_rate": 6.37076991556726e-05, "loss": 0.1236, "step": 21181 }, { "epoch": 15.517948717948718, "grad_norm": 0.75390625, "learning_rate": 6.370439698277098e-05, "loss": 0.1538, "step": 21182 }, { "epoch": 15.518681318681319, "grad_norm": 1.171875, "learning_rate": 6.370109474660275e-05, "loss": 0.1187, "step": 21183 }, { "epoch": 15.51941391941392, "grad_norm": 1.1796875, "learning_rate": 6.369779244718376e-05, "loss": 0.1108, "step": 21184 }, { "epoch": 15.52014652014652, "grad_norm": 1.3671875, "learning_rate": 6.369449008452979e-05, "loss": 0.1322, "step": 21185 }, { "epoch": 15.52087912087912, "grad_norm": 1.75, "learning_rate": 6.36911876586567e-05, "loss": 0.1307, "step": 21186 }, { "epoch": 15.521611721611722, "grad_norm": 0.609375, "learning_rate": 6.36878851695803e-05, "loss": 0.0887, "step": 21187 }, { "epoch": 15.522344322344322, "grad_norm": 0.8671875, "learning_rate": 6.36845826173164e-05, "loss": 0.1432, "step": 21188 }, { "epoch": 15.523076923076923, "grad_norm": 0.7734375, "learning_rate": 6.368128000188086e-05, "loss": 0.107, "step": 21189 }, { "epoch": 15.523809523809524, "grad_norm": 1.3515625, "learning_rate": 6.367797732328946e-05, "loss": 0.1441, "step": 21190 }, { "epoch": 15.524542124542124, "grad_norm": 0.9296875, "learning_rate": 6.367467458155807e-05, "loss": 0.1374, "step": 21191 }, { "epoch": 15.525274725274725, "grad_norm": 1.4921875, "learning_rate": 6.367137177670246e-05, "loss": 0.1083, "step": 21192 }, { "epoch": 15.526007326007326, "grad_norm": 1.828125, "learning_rate": 6.366806890873851e-05, "loss": 0.1037, "step": 21193 }, { "epoch": 15.526739926739927, "grad_norm": 1.3359375, "learning_rate": 6.366476597768199e-05, "loss": 0.1398, "step": 21194 }, { "epoch": 15.527472527472527, "grad_norm": 1.09375, "learning_rate": 6.366146298354878e-05, "loss": 0.1289, "step": 21195 }, { "epoch": 15.528205128205128, "grad_norm": 1.6328125, "learning_rate": 6.365815992635468e-05, "loss": 0.1319, "step": 21196 }, { "epoch": 15.528937728937729, "grad_norm": 0.59765625, "learning_rate": 6.365485680611549e-05, "loss": 0.111, "step": 21197 }, { "epoch": 15.52967032967033, "grad_norm": 1.1640625, "learning_rate": 6.365155362284709e-05, "loss": 0.1477, "step": 21198 }, { "epoch": 15.53040293040293, "grad_norm": 1.2734375, "learning_rate": 6.364825037656525e-05, "loss": 0.0851, "step": 21199 }, { "epoch": 15.531135531135531, "grad_norm": 1.25, "learning_rate": 6.364494706728584e-05, "loss": 0.1356, "step": 21200 }, { "epoch": 15.531868131868132, "grad_norm": 0.92578125, "learning_rate": 6.364164369502469e-05, "loss": 0.1044, "step": 21201 }, { "epoch": 15.532600732600732, "grad_norm": 1.171875, "learning_rate": 6.36383402597976e-05, "loss": 0.1365, "step": 21202 }, { "epoch": 15.533333333333333, "grad_norm": 0.9765625, "learning_rate": 6.363503676162041e-05, "loss": 0.1399, "step": 21203 }, { "epoch": 15.534065934065934, "grad_norm": 1.1484375, "learning_rate": 6.363173320050895e-05, "loss": 0.1281, "step": 21204 }, { "epoch": 15.534798534798535, "grad_norm": 0.84375, "learning_rate": 6.362842957647904e-05, "loss": 0.1181, "step": 21205 }, { "epoch": 15.535531135531135, "grad_norm": 1.3515625, "learning_rate": 6.362512588954651e-05, "loss": 0.1077, "step": 21206 }, { "epoch": 15.536263736263736, "grad_norm": 1.0859375, "learning_rate": 6.362182213972718e-05, "loss": 0.0968, "step": 21207 }, { "epoch": 15.536996336996337, "grad_norm": 0.9921875, "learning_rate": 6.361851832703693e-05, "loss": 0.1421, "step": 21208 }, { "epoch": 15.537728937728938, "grad_norm": 1.53125, "learning_rate": 6.361521445149154e-05, "loss": 0.1062, "step": 21209 }, { "epoch": 15.538461538461538, "grad_norm": 0.71875, "learning_rate": 6.361191051310683e-05, "loss": 0.1253, "step": 21210 }, { "epoch": 15.539194139194139, "grad_norm": 1.078125, "learning_rate": 6.360860651189867e-05, "loss": 0.1167, "step": 21211 }, { "epoch": 15.53992673992674, "grad_norm": 0.7734375, "learning_rate": 6.360530244788286e-05, "loss": 0.1173, "step": 21212 }, { "epoch": 15.54065934065934, "grad_norm": 0.67578125, "learning_rate": 6.360199832107525e-05, "loss": 0.1033, "step": 21213 }, { "epoch": 15.541391941391941, "grad_norm": 0.71484375, "learning_rate": 6.359869413149166e-05, "loss": 0.1149, "step": 21214 }, { "epoch": 15.542124542124542, "grad_norm": 1.203125, "learning_rate": 6.359538987914793e-05, "loss": 0.1373, "step": 21215 }, { "epoch": 15.542857142857143, "grad_norm": 1.3984375, "learning_rate": 6.35920855640599e-05, "loss": 0.1321, "step": 21216 }, { "epoch": 15.543589743589743, "grad_norm": 1.21875, "learning_rate": 6.35887811862434e-05, "loss": 0.1262, "step": 21217 }, { "epoch": 15.544322344322344, "grad_norm": 1.7109375, "learning_rate": 6.358547674571422e-05, "loss": 0.1291, "step": 21218 }, { "epoch": 15.545054945054945, "grad_norm": 1.8515625, "learning_rate": 6.358217224248825e-05, "loss": 0.1109, "step": 21219 }, { "epoch": 15.545787545787546, "grad_norm": 1.015625, "learning_rate": 6.357886767658128e-05, "loss": 0.1199, "step": 21220 }, { "epoch": 15.546520146520146, "grad_norm": 1.6015625, "learning_rate": 6.357556304800918e-05, "loss": 0.1197, "step": 21221 }, { "epoch": 15.547252747252747, "grad_norm": 0.87109375, "learning_rate": 6.357225835678775e-05, "loss": 0.1187, "step": 21222 }, { "epoch": 15.547985347985348, "grad_norm": 1.1640625, "learning_rate": 6.356895360293285e-05, "loss": 0.1341, "step": 21223 }, { "epoch": 15.548717948717949, "grad_norm": 0.64453125, "learning_rate": 6.356564878646028e-05, "loss": 0.1051, "step": 21224 }, { "epoch": 15.54945054945055, "grad_norm": 1.046875, "learning_rate": 6.356234390738593e-05, "loss": 0.1442, "step": 21225 }, { "epoch": 15.55018315018315, "grad_norm": 1.9375, "learning_rate": 6.355903896572558e-05, "loss": 0.1437, "step": 21226 }, { "epoch": 15.55091575091575, "grad_norm": 1.1328125, "learning_rate": 6.35557339614951e-05, "loss": 0.1385, "step": 21227 }, { "epoch": 15.551648351648351, "grad_norm": 0.9140625, "learning_rate": 6.355242889471029e-05, "loss": 0.1442, "step": 21228 }, { "epoch": 15.552380952380952, "grad_norm": 1.3203125, "learning_rate": 6.354912376538703e-05, "loss": 0.1433, "step": 21229 }, { "epoch": 15.553113553113553, "grad_norm": 0.5625, "learning_rate": 6.354581857354113e-05, "loss": 0.0798, "step": 21230 }, { "epoch": 15.553846153846154, "grad_norm": 1.2578125, "learning_rate": 6.354251331918843e-05, "loss": 0.1766, "step": 21231 }, { "epoch": 15.554578754578754, "grad_norm": 1.2890625, "learning_rate": 6.353920800234478e-05, "loss": 0.1666, "step": 21232 }, { "epoch": 15.555311355311355, "grad_norm": 0.92578125, "learning_rate": 6.353590262302597e-05, "loss": 0.1163, "step": 21233 }, { "epoch": 15.556043956043956, "grad_norm": 0.91015625, "learning_rate": 6.353259718124791e-05, "loss": 0.1092, "step": 21234 }, { "epoch": 15.556776556776557, "grad_norm": 0.80859375, "learning_rate": 6.352929167702638e-05, "loss": 0.12, "step": 21235 }, { "epoch": 15.557509157509157, "grad_norm": 0.94140625, "learning_rate": 6.352598611037723e-05, "loss": 0.0971, "step": 21236 }, { "epoch": 15.558241758241758, "grad_norm": 0.9921875, "learning_rate": 6.352268048131631e-05, "loss": 0.1263, "step": 21237 }, { "epoch": 15.558974358974359, "grad_norm": 1.0078125, "learning_rate": 6.351937478985944e-05, "loss": 0.1276, "step": 21238 }, { "epoch": 15.55970695970696, "grad_norm": 1.0, "learning_rate": 6.35160690360225e-05, "loss": 0.1388, "step": 21239 }, { "epoch": 15.56043956043956, "grad_norm": 1.046875, "learning_rate": 6.351276321982128e-05, "loss": 0.1174, "step": 21240 }, { "epoch": 15.561172161172161, "grad_norm": 0.90234375, "learning_rate": 6.350945734127166e-05, "loss": 0.1306, "step": 21241 }, { "epoch": 15.561904761904762, "grad_norm": 1.359375, "learning_rate": 6.350615140038943e-05, "loss": 0.1699, "step": 21242 }, { "epoch": 15.562637362637362, "grad_norm": 0.91796875, "learning_rate": 6.350284539719047e-05, "loss": 0.1267, "step": 21243 }, { "epoch": 15.563369963369963, "grad_norm": 1.1328125, "learning_rate": 6.349953933169063e-05, "loss": 0.1426, "step": 21244 }, { "epoch": 15.564102564102564, "grad_norm": 1.0859375, "learning_rate": 6.34962332039057e-05, "loss": 0.1295, "step": 21245 }, { "epoch": 15.564835164835165, "grad_norm": 1.015625, "learning_rate": 6.349292701385157e-05, "loss": 0.1079, "step": 21246 }, { "epoch": 15.565567765567765, "grad_norm": 0.8984375, "learning_rate": 6.348962076154406e-05, "loss": 0.1177, "step": 21247 }, { "epoch": 15.566300366300366, "grad_norm": 1.2734375, "learning_rate": 6.3486314446999e-05, "loss": 0.1131, "step": 21248 }, { "epoch": 15.567032967032967, "grad_norm": 1.15625, "learning_rate": 6.348300807023224e-05, "loss": 0.1308, "step": 21249 }, { "epoch": 15.567765567765568, "grad_norm": 0.83984375, "learning_rate": 6.347970163125965e-05, "loss": 0.1044, "step": 21250 }, { "epoch": 15.568498168498168, "grad_norm": 1.015625, "learning_rate": 6.347639513009702e-05, "loss": 0.1229, "step": 21251 }, { "epoch": 15.569230769230769, "grad_norm": 1.3125, "learning_rate": 6.347308856676024e-05, "loss": 0.1297, "step": 21252 }, { "epoch": 15.56996336996337, "grad_norm": 1.9140625, "learning_rate": 6.346978194126514e-05, "loss": 0.1349, "step": 21253 }, { "epoch": 15.57069597069597, "grad_norm": 1.0234375, "learning_rate": 6.346647525362753e-05, "loss": 0.1212, "step": 21254 }, { "epoch": 15.571428571428571, "grad_norm": 1.1640625, "learning_rate": 6.346316850386331e-05, "loss": 0.1519, "step": 21255 }, { "epoch": 15.572161172161172, "grad_norm": 1.8046875, "learning_rate": 6.345986169198829e-05, "loss": 0.1741, "step": 21256 }, { "epoch": 15.572893772893773, "grad_norm": 1.4296875, "learning_rate": 6.34565548180183e-05, "loss": 0.1487, "step": 21257 }, { "epoch": 15.573626373626373, "grad_norm": 0.91015625, "learning_rate": 6.345324788196923e-05, "loss": 0.1163, "step": 21258 }, { "epoch": 15.574358974358974, "grad_norm": 0.890625, "learning_rate": 6.344994088385688e-05, "loss": 0.128, "step": 21259 }, { "epoch": 15.575091575091575, "grad_norm": 1.90625, "learning_rate": 6.344663382369712e-05, "loss": 0.1321, "step": 21260 }, { "epoch": 15.575824175824176, "grad_norm": 1.65625, "learning_rate": 6.344332670150577e-05, "loss": 0.1223, "step": 21261 }, { "epoch": 15.576556776556776, "grad_norm": 1.6484375, "learning_rate": 6.34400195172987e-05, "loss": 0.1097, "step": 21262 }, { "epoch": 15.577289377289377, "grad_norm": 1.0546875, "learning_rate": 6.343671227109174e-05, "loss": 0.1006, "step": 21263 }, { "epoch": 15.578021978021978, "grad_norm": 0.87109375, "learning_rate": 6.343340496290075e-05, "loss": 0.1363, "step": 21264 }, { "epoch": 15.578754578754578, "grad_norm": 1.3515625, "learning_rate": 6.34300975927416e-05, "loss": 0.149, "step": 21265 }, { "epoch": 15.57948717948718, "grad_norm": 1.078125, "learning_rate": 6.342679016063007e-05, "loss": 0.1051, "step": 21266 }, { "epoch": 15.58021978021978, "grad_norm": 0.90625, "learning_rate": 6.342348266658206e-05, "loss": 0.1038, "step": 21267 }, { "epoch": 15.58095238095238, "grad_norm": 1.0625, "learning_rate": 6.34201751106134e-05, "loss": 0.1315, "step": 21268 }, { "epoch": 15.581684981684981, "grad_norm": 1.6328125, "learning_rate": 6.341686749273995e-05, "loss": 0.1632, "step": 21269 }, { "epoch": 15.582417582417582, "grad_norm": 0.87890625, "learning_rate": 6.341355981297753e-05, "loss": 0.1278, "step": 21270 }, { "epoch": 15.583150183150183, "grad_norm": 1.15625, "learning_rate": 6.341025207134201e-05, "loss": 0.1499, "step": 21271 }, { "epoch": 15.583882783882784, "grad_norm": 1.0625, "learning_rate": 6.340694426784924e-05, "loss": 0.1213, "step": 21272 }, { "epoch": 15.584615384615384, "grad_norm": 2.15625, "learning_rate": 6.340363640251507e-05, "loss": 0.1415, "step": 21273 }, { "epoch": 15.585347985347985, "grad_norm": 1.265625, "learning_rate": 6.340032847535534e-05, "loss": 0.1476, "step": 21274 }, { "epoch": 15.586080586080586, "grad_norm": 0.953125, "learning_rate": 6.339702048638587e-05, "loss": 0.1206, "step": 21275 }, { "epoch": 15.586813186813187, "grad_norm": 1.1875, "learning_rate": 6.339371243562258e-05, "loss": 0.1654, "step": 21276 }, { "epoch": 15.587545787545787, "grad_norm": 0.73046875, "learning_rate": 6.339040432308126e-05, "loss": 0.1112, "step": 21277 }, { "epoch": 15.588278388278388, "grad_norm": 1.359375, "learning_rate": 6.338709614877779e-05, "loss": 0.1338, "step": 21278 }, { "epoch": 15.589010989010989, "grad_norm": 0.578125, "learning_rate": 6.338378791272801e-05, "loss": 0.1015, "step": 21279 }, { "epoch": 15.58974358974359, "grad_norm": 1.046875, "learning_rate": 6.338047961494777e-05, "loss": 0.1182, "step": 21280 }, { "epoch": 15.59047619047619, "grad_norm": 1.375, "learning_rate": 6.337717125545293e-05, "loss": 0.1139, "step": 21281 }, { "epoch": 15.59120879120879, "grad_norm": 1.1328125, "learning_rate": 6.337386283425934e-05, "loss": 0.1304, "step": 21282 }, { "epoch": 15.591941391941392, "grad_norm": 1.0078125, "learning_rate": 6.337055435138282e-05, "loss": 0.1158, "step": 21283 }, { "epoch": 15.592673992673992, "grad_norm": 0.63671875, "learning_rate": 6.336724580683927e-05, "loss": 0.117, "step": 21284 }, { "epoch": 15.593406593406593, "grad_norm": 0.85546875, "learning_rate": 6.336393720064453e-05, "loss": 0.1096, "step": 21285 }, { "epoch": 15.594139194139194, "grad_norm": 0.9140625, "learning_rate": 6.336062853281442e-05, "loss": 0.1169, "step": 21286 }, { "epoch": 15.594871794871795, "grad_norm": 1.125, "learning_rate": 6.335731980336484e-05, "loss": 0.1304, "step": 21287 }, { "epoch": 15.595604395604395, "grad_norm": 1.6328125, "learning_rate": 6.33540110123116e-05, "loss": 0.1472, "step": 21288 }, { "epoch": 15.596336996336996, "grad_norm": 1.921875, "learning_rate": 6.335070215967058e-05, "loss": 0.162, "step": 21289 }, { "epoch": 15.597069597069597, "grad_norm": 1.390625, "learning_rate": 6.334739324545762e-05, "loss": 0.1208, "step": 21290 }, { "epoch": 15.597802197802197, "grad_norm": 1.2578125, "learning_rate": 6.33440842696886e-05, "loss": 0.1513, "step": 21291 }, { "epoch": 15.598534798534798, "grad_norm": 1.6015625, "learning_rate": 6.334077523237935e-05, "loss": 0.153, "step": 21292 }, { "epoch": 15.599267399267399, "grad_norm": 0.69140625, "learning_rate": 6.333746613354574e-05, "loss": 0.1166, "step": 21293 }, { "epoch": 15.6, "grad_norm": 1.2890625, "learning_rate": 6.33341569732036e-05, "loss": 0.1687, "step": 21294 }, { "epoch": 15.6007326007326, "grad_norm": 1.0546875, "learning_rate": 6.333084775136881e-05, "loss": 0.1131, "step": 21295 }, { "epoch": 15.601465201465201, "grad_norm": 1.25, "learning_rate": 6.33275384680572e-05, "loss": 0.1452, "step": 21296 }, { "epoch": 15.602197802197802, "grad_norm": 1.5859375, "learning_rate": 6.332422912328467e-05, "loss": 0.1544, "step": 21297 }, { "epoch": 15.602930402930403, "grad_norm": 0.95703125, "learning_rate": 6.3320919717067e-05, "loss": 0.1207, "step": 21298 }, { "epoch": 15.603663003663003, "grad_norm": 0.72265625, "learning_rate": 6.331761024942013e-05, "loss": 0.1003, "step": 21299 }, { "epoch": 15.604395604395604, "grad_norm": 1.1640625, "learning_rate": 6.331430072035988e-05, "loss": 0.1176, "step": 21300 }, { "epoch": 15.605128205128205, "grad_norm": 0.8203125, "learning_rate": 6.331099112990211e-05, "loss": 0.0932, "step": 21301 }, { "epoch": 15.605860805860805, "grad_norm": 0.55078125, "learning_rate": 6.330768147806267e-05, "loss": 0.1021, "step": 21302 }, { "epoch": 15.606593406593406, "grad_norm": 1.2265625, "learning_rate": 6.330437176485743e-05, "loss": 0.1144, "step": 21303 }, { "epoch": 15.607326007326007, "grad_norm": 0.7265625, "learning_rate": 6.330106199030224e-05, "loss": 0.1097, "step": 21304 }, { "epoch": 15.608058608058608, "grad_norm": 1.59375, "learning_rate": 6.329775215441295e-05, "loss": 0.1387, "step": 21305 }, { "epoch": 15.608791208791208, "grad_norm": 0.6484375, "learning_rate": 6.329444225720542e-05, "loss": 0.0932, "step": 21306 }, { "epoch": 15.60952380952381, "grad_norm": 0.80078125, "learning_rate": 6.329113229869556e-05, "loss": 0.0964, "step": 21307 }, { "epoch": 15.61025641025641, "grad_norm": 1.140625, "learning_rate": 6.328782227889915e-05, "loss": 0.1285, "step": 21308 }, { "epoch": 15.61098901098901, "grad_norm": 1.265625, "learning_rate": 6.32845121978321e-05, "loss": 0.1301, "step": 21309 }, { "epoch": 15.611721611721611, "grad_norm": 1.0078125, "learning_rate": 6.328120205551024e-05, "loss": 0.1232, "step": 21310 }, { "epoch": 15.612454212454212, "grad_norm": 0.99609375, "learning_rate": 6.327789185194945e-05, "loss": 0.1041, "step": 21311 }, { "epoch": 15.613186813186813, "grad_norm": 1.4765625, "learning_rate": 6.32745815871656e-05, "loss": 0.131, "step": 21312 }, { "epoch": 15.613919413919414, "grad_norm": 0.78515625, "learning_rate": 6.327127126117452e-05, "loss": 0.1071, "step": 21313 }, { "epoch": 15.614652014652014, "grad_norm": 1.0390625, "learning_rate": 6.326796087399208e-05, "loss": 0.1023, "step": 21314 }, { "epoch": 15.615384615384615, "grad_norm": 1.578125, "learning_rate": 6.326465042563417e-05, "loss": 0.1177, "step": 21315 }, { "epoch": 15.616117216117216, "grad_norm": 1.53125, "learning_rate": 6.326133991611662e-05, "loss": 0.1525, "step": 21316 }, { "epoch": 15.616849816849816, "grad_norm": 0.83984375, "learning_rate": 6.325802934545529e-05, "loss": 0.1021, "step": 21317 }, { "epoch": 15.617582417582417, "grad_norm": 1.6328125, "learning_rate": 6.325471871366607e-05, "loss": 0.1434, "step": 21318 }, { "epoch": 15.618315018315018, "grad_norm": 1.4296875, "learning_rate": 6.32514080207648e-05, "loss": 0.1202, "step": 21319 }, { "epoch": 15.619047619047619, "grad_norm": 1.4140625, "learning_rate": 6.324809726676734e-05, "loss": 0.1336, "step": 21320 }, { "epoch": 15.61978021978022, "grad_norm": 1.3671875, "learning_rate": 6.324478645168959e-05, "loss": 0.115, "step": 21321 }, { "epoch": 15.62051282051282, "grad_norm": 1.1796875, "learning_rate": 6.324147557554735e-05, "loss": 0.1176, "step": 21322 }, { "epoch": 15.62124542124542, "grad_norm": 0.98828125, "learning_rate": 6.323816463835653e-05, "loss": 0.1243, "step": 21323 }, { "epoch": 15.621978021978022, "grad_norm": 0.671875, "learning_rate": 6.323485364013298e-05, "loss": 0.1022, "step": 21324 }, { "epoch": 15.622710622710622, "grad_norm": 1.0546875, "learning_rate": 6.323154258089256e-05, "loss": 0.1106, "step": 21325 }, { "epoch": 15.623443223443223, "grad_norm": 1.609375, "learning_rate": 6.322823146065115e-05, "loss": 0.124, "step": 21326 }, { "epoch": 15.624175824175824, "grad_norm": 1.734375, "learning_rate": 6.322492027942461e-05, "loss": 0.1386, "step": 21327 }, { "epoch": 15.624908424908424, "grad_norm": 1.5390625, "learning_rate": 6.32216090372288e-05, "loss": 0.1121, "step": 21328 }, { "epoch": 15.625641025641025, "grad_norm": 1.65625, "learning_rate": 6.321829773407957e-05, "loss": 0.1345, "step": 21329 }, { "epoch": 15.626373626373626, "grad_norm": 1.5546875, "learning_rate": 6.32149863699928e-05, "loss": 0.1358, "step": 21330 }, { "epoch": 15.627106227106227, "grad_norm": 1.546875, "learning_rate": 6.321167494498437e-05, "loss": 0.1377, "step": 21331 }, { "epoch": 15.627838827838827, "grad_norm": 1.1171875, "learning_rate": 6.320836345907011e-05, "loss": 0.1342, "step": 21332 }, { "epoch": 15.628571428571428, "grad_norm": 0.88671875, "learning_rate": 6.320505191226594e-05, "loss": 0.1294, "step": 21333 }, { "epoch": 15.629304029304029, "grad_norm": 1.65625, "learning_rate": 6.320174030458769e-05, "loss": 0.1119, "step": 21334 }, { "epoch": 15.63003663003663, "grad_norm": 1.265625, "learning_rate": 6.319842863605121e-05, "loss": 0.1163, "step": 21335 }, { "epoch": 15.63076923076923, "grad_norm": 1.0234375, "learning_rate": 6.319511690667242e-05, "loss": 0.135, "step": 21336 }, { "epoch": 15.631501831501831, "grad_norm": 1.1015625, "learning_rate": 6.319180511646713e-05, "loss": 0.1201, "step": 21337 }, { "epoch": 15.632234432234432, "grad_norm": 1.1796875, "learning_rate": 6.318849326545125e-05, "loss": 0.1372, "step": 21338 }, { "epoch": 15.632967032967032, "grad_norm": 1.09375, "learning_rate": 6.318518135364063e-05, "loss": 0.1103, "step": 21339 }, { "epoch": 15.633699633699633, "grad_norm": 1.296875, "learning_rate": 6.318186938105113e-05, "loss": 0.1466, "step": 21340 }, { "epoch": 15.634432234432234, "grad_norm": 0.99609375, "learning_rate": 6.317855734769863e-05, "loss": 0.0899, "step": 21341 }, { "epoch": 15.635164835164835, "grad_norm": 1.328125, "learning_rate": 6.317524525359902e-05, "loss": 0.1248, "step": 21342 }, { "epoch": 15.635897435897435, "grad_norm": 0.75, "learning_rate": 6.317193309876814e-05, "loss": 0.1157, "step": 21343 }, { "epoch": 15.636630036630036, "grad_norm": 1.6953125, "learning_rate": 6.316862088322186e-05, "loss": 0.1427, "step": 21344 }, { "epoch": 15.637362637362637, "grad_norm": 1.296875, "learning_rate": 6.316530860697607e-05, "loss": 0.13, "step": 21345 }, { "epoch": 15.638095238095238, "grad_norm": 1.8984375, "learning_rate": 6.316199627004663e-05, "loss": 0.1296, "step": 21346 }, { "epoch": 15.638827838827838, "grad_norm": 1.875, "learning_rate": 6.315868387244941e-05, "loss": 0.1601, "step": 21347 }, { "epoch": 15.639560439560439, "grad_norm": 1.1875, "learning_rate": 6.315537141420027e-05, "loss": 0.1445, "step": 21348 }, { "epoch": 15.64029304029304, "grad_norm": 1.3046875, "learning_rate": 6.315205889531509e-05, "loss": 0.1261, "step": 21349 }, { "epoch": 15.64102564102564, "grad_norm": 1.1796875, "learning_rate": 6.314874631580976e-05, "loss": 0.1089, "step": 21350 }, { "epoch": 15.641758241758241, "grad_norm": 1.546875, "learning_rate": 6.314543367570013e-05, "loss": 0.1375, "step": 21351 }, { "epoch": 15.642490842490842, "grad_norm": 1.484375, "learning_rate": 6.314212097500206e-05, "loss": 0.1338, "step": 21352 }, { "epoch": 15.643223443223443, "grad_norm": 0.83203125, "learning_rate": 6.313880821373143e-05, "loss": 0.1165, "step": 21353 }, { "epoch": 15.643956043956043, "grad_norm": 1.015625, "learning_rate": 6.313549539190415e-05, "loss": 0.1562, "step": 21354 }, { "epoch": 15.644688644688644, "grad_norm": 1.53125, "learning_rate": 6.313218250953605e-05, "loss": 0.1641, "step": 21355 }, { "epoch": 15.645421245421245, "grad_norm": 0.96484375, "learning_rate": 6.312886956664301e-05, "loss": 0.1016, "step": 21356 }, { "epoch": 15.646153846153846, "grad_norm": 1.3046875, "learning_rate": 6.312555656324093e-05, "loss": 0.1189, "step": 21357 }, { "epoch": 15.646886446886446, "grad_norm": 1.109375, "learning_rate": 6.312224349934564e-05, "loss": 0.139, "step": 21358 }, { "epoch": 15.647619047619047, "grad_norm": 1.171875, "learning_rate": 6.311893037497305e-05, "loss": 0.1109, "step": 21359 }, { "epoch": 15.648351648351648, "grad_norm": 0.71484375, "learning_rate": 6.311561719013904e-05, "loss": 0.1218, "step": 21360 }, { "epoch": 15.649084249084249, "grad_norm": 0.8828125, "learning_rate": 6.311230394485944e-05, "loss": 0.1293, "step": 21361 }, { "epoch": 15.64981684981685, "grad_norm": 1.09375, "learning_rate": 6.310899063915016e-05, "loss": 0.1363, "step": 21362 }, { "epoch": 15.65054945054945, "grad_norm": 1.3203125, "learning_rate": 6.310567727302706e-05, "loss": 0.1091, "step": 21363 }, { "epoch": 15.65128205128205, "grad_norm": 1.125, "learning_rate": 6.310236384650603e-05, "loss": 0.0841, "step": 21364 }, { "epoch": 15.652014652014651, "grad_norm": 1.03125, "learning_rate": 6.309905035960294e-05, "loss": 0.138, "step": 21365 }, { "epoch": 15.652747252747252, "grad_norm": 0.88671875, "learning_rate": 6.309573681233364e-05, "loss": 0.1082, "step": 21366 }, { "epoch": 15.653479853479853, "grad_norm": 0.8671875, "learning_rate": 6.309242320471406e-05, "loss": 0.1147, "step": 21367 }, { "epoch": 15.654212454212454, "grad_norm": 1.046875, "learning_rate": 6.308910953676005e-05, "loss": 0.1384, "step": 21368 }, { "epoch": 15.654945054945054, "grad_norm": 0.93359375, "learning_rate": 6.308579580848746e-05, "loss": 0.1261, "step": 21369 }, { "epoch": 15.655677655677655, "grad_norm": 1.0859375, "learning_rate": 6.308248201991219e-05, "loss": 0.1309, "step": 21370 }, { "epoch": 15.656410256410256, "grad_norm": 0.828125, "learning_rate": 6.307916817105013e-05, "loss": 0.1034, "step": 21371 }, { "epoch": 15.657142857142857, "grad_norm": 0.9453125, "learning_rate": 6.307585426191716e-05, "loss": 0.1282, "step": 21372 }, { "epoch": 15.657875457875457, "grad_norm": 1.5859375, "learning_rate": 6.307254029252912e-05, "loss": 0.1367, "step": 21373 }, { "epoch": 15.658608058608058, "grad_norm": 1.0234375, "learning_rate": 6.306922626290192e-05, "loss": 0.1401, "step": 21374 }, { "epoch": 15.659340659340659, "grad_norm": 1.5390625, "learning_rate": 6.306591217305144e-05, "loss": 0.1526, "step": 21375 }, { "epoch": 15.66007326007326, "grad_norm": 1.125, "learning_rate": 6.306259802299353e-05, "loss": 0.1335, "step": 21376 }, { "epoch": 15.66080586080586, "grad_norm": 1.125, "learning_rate": 6.30592838127441e-05, "loss": 0.1071, "step": 21377 }, { "epoch": 15.661538461538461, "grad_norm": 1.375, "learning_rate": 6.305596954231901e-05, "loss": 0.141, "step": 21378 }, { "epoch": 15.662271062271062, "grad_norm": 1.0703125, "learning_rate": 6.305265521173416e-05, "loss": 0.1056, "step": 21379 }, { "epoch": 15.663003663003662, "grad_norm": 1.640625, "learning_rate": 6.304934082100541e-05, "loss": 0.1128, "step": 21380 }, { "epoch": 15.663736263736263, "grad_norm": 0.87890625, "learning_rate": 6.304602637014866e-05, "loss": 0.1319, "step": 21381 }, { "epoch": 15.664468864468864, "grad_norm": 2.03125, "learning_rate": 6.304271185917977e-05, "loss": 0.1148, "step": 21382 }, { "epoch": 15.665201465201465, "grad_norm": 0.83984375, "learning_rate": 6.303939728811463e-05, "loss": 0.1169, "step": 21383 }, { "epoch": 15.665934065934065, "grad_norm": 1.5859375, "learning_rate": 6.303608265696913e-05, "loss": 0.1325, "step": 21384 }, { "epoch": 15.666666666666666, "grad_norm": 1.3359375, "learning_rate": 6.303276796575913e-05, "loss": 0.1094, "step": 21385 }, { "epoch": 15.667399267399267, "grad_norm": 1.578125, "learning_rate": 6.302945321450054e-05, "loss": 0.1162, "step": 21386 }, { "epoch": 15.668131868131868, "grad_norm": 0.56640625, "learning_rate": 6.302613840320921e-05, "loss": 0.1165, "step": 21387 }, { "epoch": 15.668864468864468, "grad_norm": 1.4296875, "learning_rate": 6.302282353190106e-05, "loss": 0.1401, "step": 21388 }, { "epoch": 15.669597069597069, "grad_norm": 1.09375, "learning_rate": 6.301950860059193e-05, "loss": 0.116, "step": 21389 }, { "epoch": 15.67032967032967, "grad_norm": 0.97265625, "learning_rate": 6.301619360929772e-05, "loss": 0.1381, "step": 21390 }, { "epoch": 15.67106227106227, "grad_norm": 0.71484375, "learning_rate": 6.301287855803433e-05, "loss": 0.1281, "step": 21391 }, { "epoch": 15.671794871794871, "grad_norm": 1.15625, "learning_rate": 6.300956344681762e-05, "loss": 0.1342, "step": 21392 }, { "epoch": 15.672527472527472, "grad_norm": 1.4375, "learning_rate": 6.300624827566351e-05, "loss": 0.136, "step": 21393 }, { "epoch": 15.673260073260073, "grad_norm": 1.5078125, "learning_rate": 6.300293304458784e-05, "loss": 0.1536, "step": 21394 }, { "epoch": 15.673992673992673, "grad_norm": 1.125, "learning_rate": 6.299961775360651e-05, "loss": 0.1192, "step": 21395 }, { "epoch": 15.674725274725274, "grad_norm": 1.0, "learning_rate": 6.299630240273541e-05, "loss": 0.1246, "step": 21396 }, { "epoch": 15.675457875457875, "grad_norm": 1.59375, "learning_rate": 6.299298699199044e-05, "loss": 0.1104, "step": 21397 }, { "epoch": 15.676190476190476, "grad_norm": 1.125, "learning_rate": 6.298967152138745e-05, "loss": 0.1034, "step": 21398 }, { "epoch": 15.676923076923076, "grad_norm": 0.94921875, "learning_rate": 6.298635599094236e-05, "loss": 0.1122, "step": 21399 }, { "epoch": 15.677655677655677, "grad_norm": 0.625, "learning_rate": 6.298304040067103e-05, "loss": 0.1225, "step": 21400 }, { "epoch": 15.678388278388278, "grad_norm": 1.375, "learning_rate": 6.297972475058936e-05, "loss": 0.122, "step": 21401 }, { "epoch": 15.679120879120878, "grad_norm": 1.2578125, "learning_rate": 6.297640904071321e-05, "loss": 0.1521, "step": 21402 }, { "epoch": 15.67985347985348, "grad_norm": 2.046875, "learning_rate": 6.297309327105851e-05, "loss": 0.1126, "step": 21403 }, { "epoch": 15.68058608058608, "grad_norm": 1.2109375, "learning_rate": 6.296977744164112e-05, "loss": 0.1382, "step": 21404 }, { "epoch": 15.68131868131868, "grad_norm": 1.1171875, "learning_rate": 6.296646155247694e-05, "loss": 0.1303, "step": 21405 }, { "epoch": 15.682051282051281, "grad_norm": 1.28125, "learning_rate": 6.296314560358183e-05, "loss": 0.1096, "step": 21406 }, { "epoch": 15.682783882783882, "grad_norm": 1.1328125, "learning_rate": 6.295982959497172e-05, "loss": 0.1295, "step": 21407 }, { "epoch": 15.683516483516483, "grad_norm": 1.265625, "learning_rate": 6.295651352666246e-05, "loss": 0.1251, "step": 21408 }, { "epoch": 15.684249084249084, "grad_norm": 1.34375, "learning_rate": 6.295319739866997e-05, "loss": 0.0981, "step": 21409 }, { "epoch": 15.684981684981684, "grad_norm": 0.96484375, "learning_rate": 6.294988121101011e-05, "loss": 0.1093, "step": 21410 }, { "epoch": 15.685714285714285, "grad_norm": 1.328125, "learning_rate": 6.294656496369879e-05, "loss": 0.1237, "step": 21411 }, { "epoch": 15.686446886446886, "grad_norm": 1.65625, "learning_rate": 6.29432486567519e-05, "loss": 0.1291, "step": 21412 }, { "epoch": 15.687179487179487, "grad_norm": 1.40625, "learning_rate": 6.293993229018529e-05, "loss": 0.1247, "step": 21413 }, { "epoch": 15.687912087912087, "grad_norm": 0.67578125, "learning_rate": 6.293661586401492e-05, "loss": 0.1177, "step": 21414 }, { "epoch": 15.688644688644688, "grad_norm": 1.15625, "learning_rate": 6.29332993782566e-05, "loss": 0.0977, "step": 21415 }, { "epoch": 15.689377289377289, "grad_norm": 1.4765625, "learning_rate": 6.29299828329263e-05, "loss": 0.1151, "step": 21416 }, { "epoch": 15.69010989010989, "grad_norm": 0.83203125, "learning_rate": 6.292666622803985e-05, "loss": 0.1242, "step": 21417 }, { "epoch": 15.69084249084249, "grad_norm": 1.015625, "learning_rate": 6.292334956361316e-05, "loss": 0.1486, "step": 21418 }, { "epoch": 15.691575091575091, "grad_norm": 1.078125, "learning_rate": 6.292003283966213e-05, "loss": 0.1096, "step": 21419 }, { "epoch": 15.692307692307692, "grad_norm": 1.171875, "learning_rate": 6.291671605620265e-05, "loss": 0.1264, "step": 21420 }, { "epoch": 15.693040293040292, "grad_norm": 1.390625, "learning_rate": 6.291339921325059e-05, "loss": 0.1292, "step": 21421 }, { "epoch": 15.693772893772893, "grad_norm": 1.4375, "learning_rate": 6.291008231082189e-05, "loss": 0.1182, "step": 21422 }, { "epoch": 15.694505494505494, "grad_norm": 1.5546875, "learning_rate": 6.290676534893238e-05, "loss": 0.1388, "step": 21423 }, { "epoch": 15.695238095238095, "grad_norm": 0.89453125, "learning_rate": 6.2903448327598e-05, "loss": 0.1226, "step": 21424 }, { "epoch": 15.695970695970695, "grad_norm": 1.1640625, "learning_rate": 6.290013124683463e-05, "loss": 0.1254, "step": 21425 }, { "epoch": 15.696703296703296, "grad_norm": 0.82421875, "learning_rate": 6.289681410665815e-05, "loss": 0.1532, "step": 21426 }, { "epoch": 15.697435897435897, "grad_norm": 1.2421875, "learning_rate": 6.289349690708449e-05, "loss": 0.098, "step": 21427 }, { "epoch": 15.698168498168497, "grad_norm": 0.9765625, "learning_rate": 6.289017964812949e-05, "loss": 0.1264, "step": 21428 }, { "epoch": 15.698901098901098, "grad_norm": 1.4140625, "learning_rate": 6.288686232980909e-05, "loss": 0.164, "step": 21429 }, { "epoch": 15.699633699633699, "grad_norm": 1.2890625, "learning_rate": 6.288354495213913e-05, "loss": 0.1507, "step": 21430 }, { "epoch": 15.7003663003663, "grad_norm": 0.74609375, "learning_rate": 6.288022751513556e-05, "loss": 0.117, "step": 21431 }, { "epoch": 15.7010989010989, "grad_norm": 1.046875, "learning_rate": 6.287691001881428e-05, "loss": 0.1399, "step": 21432 }, { "epoch": 15.701831501831501, "grad_norm": 1.046875, "learning_rate": 6.287359246319114e-05, "loss": 0.1075, "step": 21433 }, { "epoch": 15.702564102564102, "grad_norm": 0.78125, "learning_rate": 6.287027484828206e-05, "loss": 0.1073, "step": 21434 }, { "epoch": 15.703296703296703, "grad_norm": 1.09375, "learning_rate": 6.286695717410294e-05, "loss": 0.1392, "step": 21435 }, { "epoch": 15.704029304029303, "grad_norm": 1.28125, "learning_rate": 6.286363944066966e-05, "loss": 0.1301, "step": 21436 }, { "epoch": 15.704761904761904, "grad_norm": 0.8125, "learning_rate": 6.286032164799812e-05, "loss": 0.1303, "step": 21437 }, { "epoch": 15.705494505494505, "grad_norm": 0.875, "learning_rate": 6.285700379610423e-05, "loss": 0.137, "step": 21438 }, { "epoch": 15.706227106227105, "grad_norm": 1.4140625, "learning_rate": 6.285368588500388e-05, "loss": 0.1743, "step": 21439 }, { "epoch": 15.706959706959706, "grad_norm": 1.1953125, "learning_rate": 6.285036791471296e-05, "loss": 0.1385, "step": 21440 }, { "epoch": 15.707692307692307, "grad_norm": 1.8515625, "learning_rate": 6.284704988524737e-05, "loss": 0.1662, "step": 21441 }, { "epoch": 15.708424908424908, "grad_norm": 0.9375, "learning_rate": 6.284373179662302e-05, "loss": 0.1263, "step": 21442 }, { "epoch": 15.709157509157508, "grad_norm": 0.90234375, "learning_rate": 6.284041364885578e-05, "loss": 0.1366, "step": 21443 }, { "epoch": 15.70989010989011, "grad_norm": 0.6640625, "learning_rate": 6.283709544196158e-05, "loss": 0.1079, "step": 21444 }, { "epoch": 15.71062271062271, "grad_norm": 0.91796875, "learning_rate": 6.283377717595631e-05, "loss": 0.0866, "step": 21445 }, { "epoch": 15.71135531135531, "grad_norm": 1.1171875, "learning_rate": 6.283045885085586e-05, "loss": 0.1099, "step": 21446 }, { "epoch": 15.712087912087911, "grad_norm": 0.95703125, "learning_rate": 6.282714046667613e-05, "loss": 0.1311, "step": 21447 }, { "epoch": 15.712820512820512, "grad_norm": 1.265625, "learning_rate": 6.282382202343304e-05, "loss": 0.1126, "step": 21448 }, { "epoch": 15.713553113553113, "grad_norm": 0.91015625, "learning_rate": 6.282050352114246e-05, "loss": 0.0876, "step": 21449 }, { "epoch": 15.714285714285714, "grad_norm": 1.0859375, "learning_rate": 6.28171849598203e-05, "loss": 0.146, "step": 21450 }, { "epoch": 15.715018315018314, "grad_norm": 0.97265625, "learning_rate": 6.281386633948249e-05, "loss": 0.1463, "step": 21451 }, { "epoch": 15.715750915750915, "grad_norm": 0.87109375, "learning_rate": 6.281054766014487e-05, "loss": 0.1389, "step": 21452 }, { "epoch": 15.716483516483516, "grad_norm": 1.296875, "learning_rate": 6.280722892182339e-05, "loss": 0.1177, "step": 21453 }, { "epoch": 15.717216117216116, "grad_norm": 0.97265625, "learning_rate": 6.280391012453391e-05, "loss": 0.121, "step": 21454 }, { "epoch": 15.717948717948717, "grad_norm": 1.0234375, "learning_rate": 6.280059126829239e-05, "loss": 0.1073, "step": 21455 }, { "epoch": 15.718681318681318, "grad_norm": 1.2109375, "learning_rate": 6.279727235311469e-05, "loss": 0.1234, "step": 21456 }, { "epoch": 15.719413919413919, "grad_norm": 1.3828125, "learning_rate": 6.279395337901672e-05, "loss": 0.1349, "step": 21457 }, { "epoch": 15.72014652014652, "grad_norm": 0.81640625, "learning_rate": 6.279063434601438e-05, "loss": 0.1058, "step": 21458 }, { "epoch": 15.72087912087912, "grad_norm": 0.99609375, "learning_rate": 6.278731525412358e-05, "loss": 0.1388, "step": 21459 }, { "epoch": 15.72161172161172, "grad_norm": 1.296875, "learning_rate": 6.278399610336021e-05, "loss": 0.1258, "step": 21460 }, { "epoch": 15.722344322344322, "grad_norm": 1.71875, "learning_rate": 6.27806768937402e-05, "loss": 0.1224, "step": 21461 }, { "epoch": 15.723076923076922, "grad_norm": 0.97265625, "learning_rate": 6.277735762527942e-05, "loss": 0.1279, "step": 21462 }, { "epoch": 15.723809523809523, "grad_norm": 1.2890625, "learning_rate": 6.277403829799381e-05, "loss": 0.1321, "step": 21463 }, { "epoch": 15.724542124542124, "grad_norm": 1.0625, "learning_rate": 6.277071891189923e-05, "loss": 0.1408, "step": 21464 }, { "epoch": 15.725274725274724, "grad_norm": 0.67578125, "learning_rate": 6.276739946701162e-05, "loss": 0.1208, "step": 21465 }, { "epoch": 15.726007326007325, "grad_norm": 1.0546875, "learning_rate": 6.276407996334688e-05, "loss": 0.1269, "step": 21466 }, { "epoch": 15.726739926739926, "grad_norm": 0.984375, "learning_rate": 6.276076040092089e-05, "loss": 0.0966, "step": 21467 }, { "epoch": 15.727472527472527, "grad_norm": 1.375, "learning_rate": 6.275744077974957e-05, "loss": 0.1635, "step": 21468 }, { "epoch": 15.728205128205127, "grad_norm": 0.9140625, "learning_rate": 6.275412109984883e-05, "loss": 0.12, "step": 21469 }, { "epoch": 15.728937728937728, "grad_norm": 1.34375, "learning_rate": 6.275080136123458e-05, "loss": 0.1204, "step": 21470 }, { "epoch": 15.729670329670329, "grad_norm": 0.7421875, "learning_rate": 6.27474815639227e-05, "loss": 0.1003, "step": 21471 }, { "epoch": 15.73040293040293, "grad_norm": 0.828125, "learning_rate": 6.274416170792916e-05, "loss": 0.1258, "step": 21472 }, { "epoch": 15.73113553113553, "grad_norm": 1.0859375, "learning_rate": 6.274084179326977e-05, "loss": 0.1395, "step": 21473 }, { "epoch": 15.731868131868131, "grad_norm": 0.984375, "learning_rate": 6.273752181996054e-05, "loss": 0.147, "step": 21474 }, { "epoch": 15.732600732600732, "grad_norm": 0.953125, "learning_rate": 6.273420178801727e-05, "loss": 0.1474, "step": 21475 }, { "epoch": 15.733333333333333, "grad_norm": 1.0546875, "learning_rate": 6.273088169745596e-05, "loss": 0.1213, "step": 21476 }, { "epoch": 15.734065934065933, "grad_norm": 1.28125, "learning_rate": 6.272756154829248e-05, "loss": 0.1132, "step": 21477 }, { "epoch": 15.734798534798536, "grad_norm": 1.15625, "learning_rate": 6.272424134054273e-05, "loss": 0.1693, "step": 21478 }, { "epoch": 15.735531135531136, "grad_norm": 0.74609375, "learning_rate": 6.272092107422263e-05, "loss": 0.1257, "step": 21479 }, { "epoch": 15.736263736263737, "grad_norm": 0.7421875, "learning_rate": 6.271760074934808e-05, "loss": 0.118, "step": 21480 }, { "epoch": 15.736996336996338, "grad_norm": 0.7421875, "learning_rate": 6.271428036593499e-05, "loss": 0.123, "step": 21481 }, { "epoch": 15.737728937728939, "grad_norm": 1.4375, "learning_rate": 6.271095992399927e-05, "loss": 0.1106, "step": 21482 }, { "epoch": 15.73846153846154, "grad_norm": 0.80859375, "learning_rate": 6.270763942355685e-05, "loss": 0.0935, "step": 21483 }, { "epoch": 15.73919413919414, "grad_norm": 1.1953125, "learning_rate": 6.270431886462362e-05, "loss": 0.1244, "step": 21484 }, { "epoch": 15.73992673992674, "grad_norm": 0.92578125, "learning_rate": 6.270099824721549e-05, "loss": 0.1445, "step": 21485 }, { "epoch": 15.740659340659342, "grad_norm": 1.2890625, "learning_rate": 6.269767757134835e-05, "loss": 0.1161, "step": 21486 }, { "epoch": 15.741391941391942, "grad_norm": 1.453125, "learning_rate": 6.269435683703816e-05, "loss": 0.1434, "step": 21487 }, { "epoch": 15.742124542124543, "grad_norm": 1.296875, "learning_rate": 6.269103604430079e-05, "loss": 0.1152, "step": 21488 }, { "epoch": 15.742857142857144, "grad_norm": 1.3515625, "learning_rate": 6.268771519315217e-05, "loss": 0.1316, "step": 21489 }, { "epoch": 15.743589743589745, "grad_norm": 1.4921875, "learning_rate": 6.26843942836082e-05, "loss": 0.1389, "step": 21490 }, { "epoch": 15.744322344322345, "grad_norm": 1.3046875, "learning_rate": 6.26810733156848e-05, "loss": 0.1441, "step": 21491 }, { "epoch": 15.745054945054946, "grad_norm": 1.109375, "learning_rate": 6.267775228939789e-05, "loss": 0.1099, "step": 21492 }, { "epoch": 15.745787545787547, "grad_norm": 2.03125, "learning_rate": 6.267443120476335e-05, "loss": 0.128, "step": 21493 }, { "epoch": 15.746520146520147, "grad_norm": 1.3046875, "learning_rate": 6.267111006179712e-05, "loss": 0.1286, "step": 21494 }, { "epoch": 15.747252747252748, "grad_norm": 1.5, "learning_rate": 6.266778886051512e-05, "loss": 0.1446, "step": 21495 }, { "epoch": 15.747985347985349, "grad_norm": 0.80859375, "learning_rate": 6.266446760093323e-05, "loss": 0.1266, "step": 21496 }, { "epoch": 15.74871794871795, "grad_norm": 0.9140625, "learning_rate": 6.26611462830674e-05, "loss": 0.1105, "step": 21497 }, { "epoch": 15.74945054945055, "grad_norm": 1.4921875, "learning_rate": 6.265782490693351e-05, "loss": 0.1127, "step": 21498 }, { "epoch": 15.750183150183151, "grad_norm": 0.953125, "learning_rate": 6.26545034725475e-05, "loss": 0.0987, "step": 21499 }, { "epoch": 15.750915750915752, "grad_norm": 1.1015625, "learning_rate": 6.265118197992528e-05, "loss": 0.1136, "step": 21500 }, { "epoch": 15.751648351648353, "grad_norm": 1.015625, "learning_rate": 6.264786042908274e-05, "loss": 0.096, "step": 21501 }, { "epoch": 15.752380952380953, "grad_norm": 1.3828125, "learning_rate": 6.264453882003582e-05, "loss": 0.1299, "step": 21502 }, { "epoch": 15.753113553113554, "grad_norm": 1.0859375, "learning_rate": 6.264121715280043e-05, "loss": 0.1344, "step": 21503 }, { "epoch": 15.753846153846155, "grad_norm": 1.6484375, "learning_rate": 6.263789542739247e-05, "loss": 0.1216, "step": 21504 }, { "epoch": 15.754578754578755, "grad_norm": 1.421875, "learning_rate": 6.26345736438279e-05, "loss": 0.1173, "step": 21505 }, { "epoch": 15.755311355311356, "grad_norm": 0.83984375, "learning_rate": 6.263125180212259e-05, "loss": 0.1122, "step": 21506 }, { "epoch": 15.756043956043957, "grad_norm": 1.3359375, "learning_rate": 6.262792990229247e-05, "loss": 0.1007, "step": 21507 }, { "epoch": 15.756776556776558, "grad_norm": 1.15625, "learning_rate": 6.262460794435344e-05, "loss": 0.1383, "step": 21508 }, { "epoch": 15.757509157509158, "grad_norm": 1.203125, "learning_rate": 6.262128592832143e-05, "loss": 0.1086, "step": 21509 }, { "epoch": 15.758241758241759, "grad_norm": 1.6640625, "learning_rate": 6.261796385421239e-05, "loss": 0.1366, "step": 21510 }, { "epoch": 15.75897435897436, "grad_norm": 1.296875, "learning_rate": 6.261464172204218e-05, "loss": 0.1291, "step": 21511 }, { "epoch": 15.75970695970696, "grad_norm": 1.5625, "learning_rate": 6.261131953182677e-05, "loss": 0.1482, "step": 21512 }, { "epoch": 15.760439560439561, "grad_norm": 1.4453125, "learning_rate": 6.260799728358203e-05, "loss": 0.1554, "step": 21513 }, { "epoch": 15.761172161172162, "grad_norm": 2.0, "learning_rate": 6.260467497732391e-05, "loss": 0.138, "step": 21514 }, { "epoch": 15.761904761904763, "grad_norm": 0.74609375, "learning_rate": 6.260135261306832e-05, "loss": 0.1212, "step": 21515 }, { "epoch": 15.762637362637363, "grad_norm": 0.94921875, "learning_rate": 6.259803019083118e-05, "loss": 0.1358, "step": 21516 }, { "epoch": 15.763369963369964, "grad_norm": 1.1015625, "learning_rate": 6.259470771062839e-05, "loss": 0.1186, "step": 21517 }, { "epoch": 15.764102564102565, "grad_norm": 1.8359375, "learning_rate": 6.25913851724759e-05, "loss": 0.1264, "step": 21518 }, { "epoch": 15.764835164835166, "grad_norm": 0.8125, "learning_rate": 6.25880625763896e-05, "loss": 0.1272, "step": 21519 }, { "epoch": 15.765567765567766, "grad_norm": 1.25, "learning_rate": 6.258473992238544e-05, "loss": 0.119, "step": 21520 }, { "epoch": 15.766300366300367, "grad_norm": 1.40625, "learning_rate": 6.258141721047932e-05, "loss": 0.1173, "step": 21521 }, { "epoch": 15.767032967032968, "grad_norm": 1.640625, "learning_rate": 6.257809444068716e-05, "loss": 0.124, "step": 21522 }, { "epoch": 15.767765567765569, "grad_norm": 1.1015625, "learning_rate": 6.257477161302489e-05, "loss": 0.1414, "step": 21523 }, { "epoch": 15.76849816849817, "grad_norm": 0.87890625, "learning_rate": 6.257144872750843e-05, "loss": 0.1185, "step": 21524 }, { "epoch": 15.76923076923077, "grad_norm": 0.84765625, "learning_rate": 6.256812578415367e-05, "loss": 0.1203, "step": 21525 }, { "epoch": 15.76996336996337, "grad_norm": 1.0546875, "learning_rate": 6.256480278297659e-05, "loss": 0.1415, "step": 21526 }, { "epoch": 15.770695970695972, "grad_norm": 1.859375, "learning_rate": 6.256147972399306e-05, "loss": 0.1318, "step": 21527 }, { "epoch": 15.771428571428572, "grad_norm": 1.1953125, "learning_rate": 6.255815660721902e-05, "loss": 0.128, "step": 21528 }, { "epoch": 15.772161172161173, "grad_norm": 1.34375, "learning_rate": 6.255483343267041e-05, "loss": 0.1425, "step": 21529 }, { "epoch": 15.772893772893774, "grad_norm": 0.93359375, "learning_rate": 6.255151020036312e-05, "loss": 0.1533, "step": 21530 }, { "epoch": 15.773626373626374, "grad_norm": 1.0703125, "learning_rate": 6.25481869103131e-05, "loss": 0.1374, "step": 21531 }, { "epoch": 15.774358974358975, "grad_norm": 0.96875, "learning_rate": 6.254486356253625e-05, "loss": 0.1103, "step": 21532 }, { "epoch": 15.775091575091576, "grad_norm": 1.4453125, "learning_rate": 6.25415401570485e-05, "loss": 0.1251, "step": 21533 }, { "epoch": 15.775824175824177, "grad_norm": 1.6171875, "learning_rate": 6.253821669386579e-05, "loss": 0.1396, "step": 21534 }, { "epoch": 15.776556776556777, "grad_norm": 1.5703125, "learning_rate": 6.253489317300402e-05, "loss": 0.1362, "step": 21535 }, { "epoch": 15.777289377289378, "grad_norm": 0.56640625, "learning_rate": 6.253156959447914e-05, "loss": 0.0926, "step": 21536 }, { "epoch": 15.778021978021979, "grad_norm": 0.7109375, "learning_rate": 6.252824595830706e-05, "loss": 0.1088, "step": 21537 }, { "epoch": 15.77875457875458, "grad_norm": 1.140625, "learning_rate": 6.25249222645037e-05, "loss": 0.1168, "step": 21538 }, { "epoch": 15.77948717948718, "grad_norm": 0.7421875, "learning_rate": 6.252159851308499e-05, "loss": 0.1212, "step": 21539 }, { "epoch": 15.780219780219781, "grad_norm": 1.6953125, "learning_rate": 6.251827470406685e-05, "loss": 0.1201, "step": 21540 }, { "epoch": 15.780952380952382, "grad_norm": 1.4375, "learning_rate": 6.251495083746522e-05, "loss": 0.1204, "step": 21541 }, { "epoch": 15.781684981684982, "grad_norm": 1.6015625, "learning_rate": 6.251162691329601e-05, "loss": 0.1045, "step": 21542 }, { "epoch": 15.782417582417583, "grad_norm": 1.046875, "learning_rate": 6.250830293157516e-05, "loss": 0.12, "step": 21543 }, { "epoch": 15.783150183150184, "grad_norm": 1.2734375, "learning_rate": 6.250497889231859e-05, "loss": 0.1236, "step": 21544 }, { "epoch": 15.783882783882785, "grad_norm": 0.8515625, "learning_rate": 6.250165479554221e-05, "loss": 0.1172, "step": 21545 }, { "epoch": 15.784615384615385, "grad_norm": 1.4453125, "learning_rate": 6.249833064126197e-05, "loss": 0.1378, "step": 21546 }, { "epoch": 15.785347985347986, "grad_norm": 0.76953125, "learning_rate": 6.249500642949378e-05, "loss": 0.1206, "step": 21547 }, { "epoch": 15.786080586080587, "grad_norm": 0.890625, "learning_rate": 6.24916821602536e-05, "loss": 0.1186, "step": 21548 }, { "epoch": 15.786813186813188, "grad_norm": 1.109375, "learning_rate": 6.248835783355732e-05, "loss": 0.1049, "step": 21549 }, { "epoch": 15.787545787545788, "grad_norm": 1.0703125, "learning_rate": 6.248503344942087e-05, "loss": 0.1357, "step": 21550 }, { "epoch": 15.788278388278389, "grad_norm": 0.93359375, "learning_rate": 6.24817090078602e-05, "loss": 0.1114, "step": 21551 }, { "epoch": 15.78901098901099, "grad_norm": 1.078125, "learning_rate": 6.247838450889125e-05, "loss": 0.1321, "step": 21552 }, { "epoch": 15.78974358974359, "grad_norm": 0.76953125, "learning_rate": 6.24750599525299e-05, "loss": 0.1164, "step": 21553 }, { "epoch": 15.790476190476191, "grad_norm": 0.91796875, "learning_rate": 6.247173533879211e-05, "loss": 0.1221, "step": 21554 }, { "epoch": 15.791208791208792, "grad_norm": 0.859375, "learning_rate": 6.246841066769382e-05, "loss": 0.0895, "step": 21555 }, { "epoch": 15.791941391941393, "grad_norm": 1.0703125, "learning_rate": 6.246508593925092e-05, "loss": 0.1219, "step": 21556 }, { "epoch": 15.792673992673993, "grad_norm": 0.91015625, "learning_rate": 6.246176115347941e-05, "loss": 0.1295, "step": 21557 }, { "epoch": 15.793406593406594, "grad_norm": 1.4375, "learning_rate": 6.245843631039513e-05, "loss": 0.1333, "step": 21558 }, { "epoch": 15.794139194139195, "grad_norm": 1.015625, "learning_rate": 6.245511141001406e-05, "loss": 0.1308, "step": 21559 }, { "epoch": 15.794871794871796, "grad_norm": 1.1640625, "learning_rate": 6.245178645235215e-05, "loss": 0.108, "step": 21560 }, { "epoch": 15.795604395604396, "grad_norm": 1.1171875, "learning_rate": 6.24484614374253e-05, "loss": 0.1237, "step": 21561 }, { "epoch": 15.796336996336997, "grad_norm": 1.328125, "learning_rate": 6.244513636524944e-05, "loss": 0.1178, "step": 21562 }, { "epoch": 15.797069597069598, "grad_norm": 1.0, "learning_rate": 6.244181123584053e-05, "loss": 0.1331, "step": 21563 }, { "epoch": 15.797802197802199, "grad_norm": 1.78125, "learning_rate": 6.243848604921447e-05, "loss": 0.1501, "step": 21564 }, { "epoch": 15.7985347985348, "grad_norm": 1.0234375, "learning_rate": 6.243516080538719e-05, "loss": 0.111, "step": 21565 }, { "epoch": 15.7992673992674, "grad_norm": 0.82421875, "learning_rate": 6.243183550437466e-05, "loss": 0.1112, "step": 21566 }, { "epoch": 15.8, "grad_norm": 0.84765625, "learning_rate": 6.242851014619279e-05, "loss": 0.1149, "step": 21567 }, { "epoch": 15.800732600732601, "grad_norm": 1.46875, "learning_rate": 6.24251847308575e-05, "loss": 0.1302, "step": 21568 }, { "epoch": 15.801465201465202, "grad_norm": 1.1484375, "learning_rate": 6.242185925838473e-05, "loss": 0.1122, "step": 21569 }, { "epoch": 15.802197802197803, "grad_norm": 1.109375, "learning_rate": 6.241853372879042e-05, "loss": 0.1143, "step": 21570 }, { "epoch": 15.802930402930404, "grad_norm": 1.0859375, "learning_rate": 6.241520814209051e-05, "loss": 0.1406, "step": 21571 }, { "epoch": 15.803663003663004, "grad_norm": 1.015625, "learning_rate": 6.241188249830093e-05, "loss": 0.1275, "step": 21572 }, { "epoch": 15.804395604395605, "grad_norm": 1.4453125, "learning_rate": 6.240855679743759e-05, "loss": 0.1606, "step": 21573 }, { "epoch": 15.805128205128206, "grad_norm": 0.75390625, "learning_rate": 6.240523103951645e-05, "loss": 0.1435, "step": 21574 }, { "epoch": 15.805860805860807, "grad_norm": 0.62890625, "learning_rate": 6.240190522455345e-05, "loss": 0.1314, "step": 21575 }, { "epoch": 15.806593406593407, "grad_norm": 0.65625, "learning_rate": 6.239857935256453e-05, "loss": 0.1083, "step": 21576 }, { "epoch": 15.807326007326008, "grad_norm": 0.8828125, "learning_rate": 6.23952534235656e-05, "loss": 0.1246, "step": 21577 }, { "epoch": 15.808058608058609, "grad_norm": 1.1015625, "learning_rate": 6.23919274375726e-05, "loss": 0.1299, "step": 21578 }, { "epoch": 15.80879120879121, "grad_norm": 0.859375, "learning_rate": 6.238860139460146e-05, "loss": 0.1223, "step": 21579 }, { "epoch": 15.80952380952381, "grad_norm": 1.0234375, "learning_rate": 6.238527529466814e-05, "loss": 0.1006, "step": 21580 }, { "epoch": 15.810256410256411, "grad_norm": 0.7578125, "learning_rate": 6.238194913778857e-05, "loss": 0.1327, "step": 21581 }, { "epoch": 15.810989010989012, "grad_norm": 1.0390625, "learning_rate": 6.237862292397867e-05, "loss": 0.1131, "step": 21582 }, { "epoch": 15.811721611721612, "grad_norm": 1.453125, "learning_rate": 6.23752966532544e-05, "loss": 0.1126, "step": 21583 }, { "epoch": 15.812454212454213, "grad_norm": 1.5625, "learning_rate": 6.237197032563168e-05, "loss": 0.177, "step": 21584 }, { "epoch": 15.813186813186814, "grad_norm": 1.125, "learning_rate": 6.236864394112644e-05, "loss": 0.1396, "step": 21585 }, { "epoch": 15.813919413919415, "grad_norm": 0.9375, "learning_rate": 6.236531749975464e-05, "loss": 0.1358, "step": 21586 }, { "epoch": 15.814652014652015, "grad_norm": 1.3359375, "learning_rate": 6.23619910015322e-05, "loss": 0.1185, "step": 21587 }, { "epoch": 15.815384615384616, "grad_norm": 0.77734375, "learning_rate": 6.23586644464751e-05, "loss": 0.123, "step": 21588 }, { "epoch": 15.816117216117217, "grad_norm": 0.86328125, "learning_rate": 6.235533783459923e-05, "loss": 0.122, "step": 21589 }, { "epoch": 15.816849816849818, "grad_norm": 1.421875, "learning_rate": 6.235201116592053e-05, "loss": 0.1458, "step": 21590 }, { "epoch": 15.817582417582418, "grad_norm": 1.15625, "learning_rate": 6.234868444045496e-05, "loss": 0.1283, "step": 21591 }, { "epoch": 15.818315018315019, "grad_norm": 0.953125, "learning_rate": 6.234535765821844e-05, "loss": 0.1205, "step": 21592 }, { "epoch": 15.81904761904762, "grad_norm": 1.25, "learning_rate": 6.234203081922697e-05, "loss": 0.1393, "step": 21593 }, { "epoch": 15.81978021978022, "grad_norm": 1.2734375, "learning_rate": 6.23387039234964e-05, "loss": 0.1219, "step": 21594 }, { "epoch": 15.820512820512821, "grad_norm": 1.984375, "learning_rate": 6.233537697104273e-05, "loss": 0.1281, "step": 21595 }, { "epoch": 15.821245421245422, "grad_norm": 0.91796875, "learning_rate": 6.233204996188185e-05, "loss": 0.1279, "step": 21596 }, { "epoch": 15.821978021978023, "grad_norm": 0.99609375, "learning_rate": 6.232872289602978e-05, "loss": 0.1121, "step": 21597 }, { "epoch": 15.822710622710623, "grad_norm": 1.296875, "learning_rate": 6.232539577350239e-05, "loss": 0.1039, "step": 21598 }, { "epoch": 15.823443223443224, "grad_norm": 1.5390625, "learning_rate": 6.232206859431567e-05, "loss": 0.1462, "step": 21599 }, { "epoch": 15.824175824175825, "grad_norm": 1.15625, "learning_rate": 6.231874135848551e-05, "loss": 0.1308, "step": 21600 }, { "epoch": 15.824908424908426, "grad_norm": 1.0546875, "learning_rate": 6.231541406602789e-05, "loss": 0.1085, "step": 21601 }, { "epoch": 15.825641025641026, "grad_norm": 1.0, "learning_rate": 6.231208671695875e-05, "loss": 0.1158, "step": 21602 }, { "epoch": 15.826373626373627, "grad_norm": 1.03125, "learning_rate": 6.230875931129402e-05, "loss": 0.1256, "step": 21603 }, { "epoch": 15.827106227106228, "grad_norm": 2.03125, "learning_rate": 6.230543184904965e-05, "loss": 0.1408, "step": 21604 }, { "epoch": 15.827838827838828, "grad_norm": 1.234375, "learning_rate": 6.230210433024158e-05, "loss": 0.127, "step": 21605 }, { "epoch": 15.82857142857143, "grad_norm": 1.390625, "learning_rate": 6.229877675488576e-05, "loss": 0.1184, "step": 21606 }, { "epoch": 15.82930402930403, "grad_norm": 1.125, "learning_rate": 6.229544912299812e-05, "loss": 0.1456, "step": 21607 }, { "epoch": 15.83003663003663, "grad_norm": 1.4921875, "learning_rate": 6.229212143459459e-05, "loss": 0.1418, "step": 21608 }, { "epoch": 15.830769230769231, "grad_norm": 0.94140625, "learning_rate": 6.228879368969116e-05, "loss": 0.0922, "step": 21609 }, { "epoch": 15.831501831501832, "grad_norm": 1.921875, "learning_rate": 6.228546588830376e-05, "loss": 0.1339, "step": 21610 }, { "epoch": 15.832234432234433, "grad_norm": 1.5234375, "learning_rate": 6.22821380304483e-05, "loss": 0.1524, "step": 21611 }, { "epoch": 15.832967032967034, "grad_norm": 1.5546875, "learning_rate": 6.227881011614076e-05, "loss": 0.158, "step": 21612 }, { "epoch": 15.833699633699634, "grad_norm": 0.984375, "learning_rate": 6.227548214539706e-05, "loss": 0.1059, "step": 21613 }, { "epoch": 15.834432234432235, "grad_norm": 1.3203125, "learning_rate": 6.227215411823318e-05, "loss": 0.1369, "step": 21614 }, { "epoch": 15.835164835164836, "grad_norm": 1.2265625, "learning_rate": 6.226882603466503e-05, "loss": 0.0989, "step": 21615 }, { "epoch": 15.835897435897436, "grad_norm": 0.88671875, "learning_rate": 6.226549789470858e-05, "loss": 0.1258, "step": 21616 }, { "epoch": 15.836630036630037, "grad_norm": 0.828125, "learning_rate": 6.226216969837977e-05, "loss": 0.1279, "step": 21617 }, { "epoch": 15.837362637362638, "grad_norm": 1.015625, "learning_rate": 6.225884144569453e-05, "loss": 0.1293, "step": 21618 }, { "epoch": 15.838095238095239, "grad_norm": 1.2421875, "learning_rate": 6.225551313666884e-05, "loss": 0.1393, "step": 21619 }, { "epoch": 15.83882783882784, "grad_norm": 1.1328125, "learning_rate": 6.225218477131862e-05, "loss": 0.1309, "step": 21620 }, { "epoch": 15.83956043956044, "grad_norm": 0.734375, "learning_rate": 6.224885634965981e-05, "loss": 0.1593, "step": 21621 }, { "epoch": 15.84029304029304, "grad_norm": 0.83203125, "learning_rate": 6.224552787170838e-05, "loss": 0.1009, "step": 21622 }, { "epoch": 15.841025641025642, "grad_norm": 1.40625, "learning_rate": 6.224219933748028e-05, "loss": 0.1172, "step": 21623 }, { "epoch": 15.841758241758242, "grad_norm": 1.109375, "learning_rate": 6.223887074699143e-05, "loss": 0.1271, "step": 21624 }, { "epoch": 15.842490842490843, "grad_norm": 1.0078125, "learning_rate": 6.223554210025782e-05, "loss": 0.122, "step": 21625 }, { "epoch": 15.843223443223444, "grad_norm": 1.078125, "learning_rate": 6.223221339729535e-05, "loss": 0.1023, "step": 21626 }, { "epoch": 15.843956043956045, "grad_norm": 0.81640625, "learning_rate": 6.222888463812002e-05, "loss": 0.1245, "step": 21627 }, { "epoch": 15.844688644688645, "grad_norm": 0.81640625, "learning_rate": 6.222555582274775e-05, "loss": 0.1428, "step": 21628 }, { "epoch": 15.845421245421246, "grad_norm": 1.0078125, "learning_rate": 6.222222695119448e-05, "loss": 0.1232, "step": 21629 }, { "epoch": 15.846153846153847, "grad_norm": 0.81640625, "learning_rate": 6.221889802347618e-05, "loss": 0.1297, "step": 21630 }, { "epoch": 15.846886446886447, "grad_norm": 1.1171875, "learning_rate": 6.221556903960878e-05, "loss": 0.1132, "step": 21631 }, { "epoch": 15.847619047619048, "grad_norm": 1.6171875, "learning_rate": 6.221223999960826e-05, "loss": 0.1466, "step": 21632 }, { "epoch": 15.848351648351649, "grad_norm": 1.09375, "learning_rate": 6.220891090349056e-05, "loss": 0.1112, "step": 21633 }, { "epoch": 15.84908424908425, "grad_norm": 1.1171875, "learning_rate": 6.220558175127159e-05, "loss": 0.1373, "step": 21634 }, { "epoch": 15.84981684981685, "grad_norm": 1.5703125, "learning_rate": 6.220225254296734e-05, "loss": 0.1157, "step": 21635 }, { "epoch": 15.850549450549451, "grad_norm": 1.3515625, "learning_rate": 6.219892327859379e-05, "loss": 0.1146, "step": 21636 }, { "epoch": 15.851282051282052, "grad_norm": 0.7421875, "learning_rate": 6.219559395816683e-05, "loss": 0.1127, "step": 21637 }, { "epoch": 15.852014652014653, "grad_norm": 1.484375, "learning_rate": 6.219226458170246e-05, "loss": 0.111, "step": 21638 }, { "epoch": 15.852747252747253, "grad_norm": 1.0390625, "learning_rate": 6.21889351492166e-05, "loss": 0.1326, "step": 21639 }, { "epoch": 15.853479853479854, "grad_norm": 1.0703125, "learning_rate": 6.218560566072521e-05, "loss": 0.1463, "step": 21640 }, { "epoch": 15.854212454212455, "grad_norm": 0.87890625, "learning_rate": 6.218227611624426e-05, "loss": 0.1233, "step": 21641 }, { "epoch": 15.854945054945055, "grad_norm": 1.2890625, "learning_rate": 6.217894651578967e-05, "loss": 0.1174, "step": 21642 }, { "epoch": 15.855677655677656, "grad_norm": 0.8125, "learning_rate": 6.217561685937744e-05, "loss": 0.1414, "step": 21643 }, { "epoch": 15.856410256410257, "grad_norm": 0.84765625, "learning_rate": 6.217228714702347e-05, "loss": 0.1183, "step": 21644 }, { "epoch": 15.857142857142858, "grad_norm": 0.8828125, "learning_rate": 6.216895737874375e-05, "loss": 0.1337, "step": 21645 }, { "epoch": 15.857875457875458, "grad_norm": 1.1328125, "learning_rate": 6.216562755455425e-05, "loss": 0.1469, "step": 21646 }, { "epoch": 15.858608058608059, "grad_norm": 0.9375, "learning_rate": 6.216229767447087e-05, "loss": 0.1351, "step": 21647 }, { "epoch": 15.85934065934066, "grad_norm": 1.078125, "learning_rate": 6.215896773850959e-05, "loss": 0.1261, "step": 21648 }, { "epoch": 15.86007326007326, "grad_norm": 1.8671875, "learning_rate": 6.21556377466864e-05, "loss": 0.1373, "step": 21649 }, { "epoch": 15.860805860805861, "grad_norm": 1.484375, "learning_rate": 6.215230769901718e-05, "loss": 0.1235, "step": 21650 }, { "epoch": 15.861538461538462, "grad_norm": 0.94921875, "learning_rate": 6.214897759551795e-05, "loss": 0.1274, "step": 21651 }, { "epoch": 15.862271062271063, "grad_norm": 0.69921875, "learning_rate": 6.214564743620465e-05, "loss": 0.1096, "step": 21652 }, { "epoch": 15.863003663003663, "grad_norm": 1.0859375, "learning_rate": 6.214231722109323e-05, "loss": 0.109, "step": 21653 }, { "epoch": 15.863736263736264, "grad_norm": 0.8828125, "learning_rate": 6.213898695019963e-05, "loss": 0.1177, "step": 21654 }, { "epoch": 15.864468864468865, "grad_norm": 1.3046875, "learning_rate": 6.213565662353983e-05, "loss": 0.1382, "step": 21655 }, { "epoch": 15.865201465201466, "grad_norm": 0.94921875, "learning_rate": 6.21323262411298e-05, "loss": 0.1508, "step": 21656 }, { "epoch": 15.865934065934066, "grad_norm": 0.7578125, "learning_rate": 6.212899580298545e-05, "loss": 0.1223, "step": 21657 }, { "epoch": 15.866666666666667, "grad_norm": 0.99609375, "learning_rate": 6.212566530912277e-05, "loss": 0.112, "step": 21658 }, { "epoch": 15.867399267399268, "grad_norm": 0.91796875, "learning_rate": 6.212233475955772e-05, "loss": 0.1262, "step": 21659 }, { "epoch": 15.868131868131869, "grad_norm": 1.2109375, "learning_rate": 6.211900415430625e-05, "loss": 0.1287, "step": 21660 }, { "epoch": 15.86886446886447, "grad_norm": 0.734375, "learning_rate": 6.211567349338428e-05, "loss": 0.1108, "step": 21661 }, { "epoch": 15.86959706959707, "grad_norm": 1.203125, "learning_rate": 6.211234277680785e-05, "loss": 0.1109, "step": 21662 }, { "epoch": 15.87032967032967, "grad_norm": 1.6953125, "learning_rate": 6.210901200459285e-05, "loss": 0.1332, "step": 21663 }, { "epoch": 15.871062271062272, "grad_norm": 0.84765625, "learning_rate": 6.210568117675528e-05, "loss": 0.1201, "step": 21664 }, { "epoch": 15.871794871794872, "grad_norm": 1.296875, "learning_rate": 6.210235029331108e-05, "loss": 0.1158, "step": 21665 }, { "epoch": 15.872527472527473, "grad_norm": 1.2734375, "learning_rate": 6.209901935427619e-05, "loss": 0.1387, "step": 21666 }, { "epoch": 15.873260073260074, "grad_norm": 1.6484375, "learning_rate": 6.209568835966661e-05, "loss": 0.157, "step": 21667 }, { "epoch": 15.873992673992674, "grad_norm": 1.203125, "learning_rate": 6.209235730949827e-05, "loss": 0.1233, "step": 21668 }, { "epoch": 15.874725274725275, "grad_norm": 1.203125, "learning_rate": 6.208902620378715e-05, "loss": 0.125, "step": 21669 }, { "epoch": 15.875457875457876, "grad_norm": 0.91796875, "learning_rate": 6.20856950425492e-05, "loss": 0.1133, "step": 21670 }, { "epoch": 15.876190476190477, "grad_norm": 1.0546875, "learning_rate": 6.208236382580037e-05, "loss": 0.1136, "step": 21671 }, { "epoch": 15.876923076923077, "grad_norm": 0.984375, "learning_rate": 6.207903255355665e-05, "loss": 0.1225, "step": 21672 }, { "epoch": 15.877655677655678, "grad_norm": 0.76953125, "learning_rate": 6.207570122583397e-05, "loss": 0.1055, "step": 21673 }, { "epoch": 15.878388278388279, "grad_norm": 1.6484375, "learning_rate": 6.20723698426483e-05, "loss": 0.1322, "step": 21674 }, { "epoch": 15.87912087912088, "grad_norm": 1.21875, "learning_rate": 6.206903840401562e-05, "loss": 0.1287, "step": 21675 }, { "epoch": 15.87985347985348, "grad_norm": 0.9765625, "learning_rate": 6.206570690995188e-05, "loss": 0.1042, "step": 21676 }, { "epoch": 15.880586080586081, "grad_norm": 0.7578125, "learning_rate": 6.206237536047304e-05, "loss": 0.1185, "step": 21677 }, { "epoch": 15.881318681318682, "grad_norm": 1.421875, "learning_rate": 6.205904375559505e-05, "loss": 0.1259, "step": 21678 }, { "epoch": 15.882051282051282, "grad_norm": 1.3359375, "learning_rate": 6.205571209533391e-05, "loss": 0.157, "step": 21679 }, { "epoch": 15.882783882783883, "grad_norm": 0.63671875, "learning_rate": 6.205238037970554e-05, "loss": 0.1243, "step": 21680 }, { "epoch": 15.883516483516484, "grad_norm": 0.74609375, "learning_rate": 6.204904860872592e-05, "loss": 0.1473, "step": 21681 }, { "epoch": 15.884249084249085, "grad_norm": 0.89453125, "learning_rate": 6.204571678241105e-05, "loss": 0.1374, "step": 21682 }, { "epoch": 15.884981684981685, "grad_norm": 0.86328125, "learning_rate": 6.204238490077683e-05, "loss": 0.1242, "step": 21683 }, { "epoch": 15.885714285714286, "grad_norm": 1.3203125, "learning_rate": 6.203905296383927e-05, "loss": 0.1432, "step": 21684 }, { "epoch": 15.886446886446887, "grad_norm": 1.0625, "learning_rate": 6.20357209716143e-05, "loss": 0.1419, "step": 21685 }, { "epoch": 15.887179487179488, "grad_norm": 0.98828125, "learning_rate": 6.203238892411792e-05, "loss": 0.1251, "step": 21686 }, { "epoch": 15.887912087912088, "grad_norm": 1.0859375, "learning_rate": 6.202905682136607e-05, "loss": 0.1078, "step": 21687 }, { "epoch": 15.888644688644689, "grad_norm": 1.2109375, "learning_rate": 6.202572466337473e-05, "loss": 0.0941, "step": 21688 }, { "epoch": 15.88937728937729, "grad_norm": 0.9375, "learning_rate": 6.202239245015986e-05, "loss": 0.1137, "step": 21689 }, { "epoch": 15.89010989010989, "grad_norm": 1.46875, "learning_rate": 6.201906018173742e-05, "loss": 0.1381, "step": 21690 }, { "epoch": 15.890842490842491, "grad_norm": 1.234375, "learning_rate": 6.201572785812339e-05, "loss": 0.1391, "step": 21691 }, { "epoch": 15.891575091575092, "grad_norm": 1.0078125, "learning_rate": 6.201239547933373e-05, "loss": 0.1403, "step": 21692 }, { "epoch": 15.892307692307693, "grad_norm": 0.921875, "learning_rate": 6.200906304538439e-05, "loss": 0.1065, "step": 21693 }, { "epoch": 15.893040293040293, "grad_norm": 1.453125, "learning_rate": 6.200573055629135e-05, "loss": 0.145, "step": 21694 }, { "epoch": 15.893772893772894, "grad_norm": 0.6796875, "learning_rate": 6.20023980120706e-05, "loss": 0.127, "step": 21695 }, { "epoch": 15.894505494505495, "grad_norm": 1.8671875, "learning_rate": 6.199906541273807e-05, "loss": 0.1049, "step": 21696 }, { "epoch": 15.895238095238096, "grad_norm": 0.7109375, "learning_rate": 6.199573275830975e-05, "loss": 0.1168, "step": 21697 }, { "epoch": 15.895970695970696, "grad_norm": 0.921875, "learning_rate": 6.199240004880159e-05, "loss": 0.1323, "step": 21698 }, { "epoch": 15.896703296703297, "grad_norm": 0.85546875, "learning_rate": 6.198906728422956e-05, "loss": 0.1796, "step": 21699 }, { "epoch": 15.897435897435898, "grad_norm": 0.86328125, "learning_rate": 6.198573446460965e-05, "loss": 0.1351, "step": 21700 }, { "epoch": 15.898168498168499, "grad_norm": 0.82421875, "learning_rate": 6.198240158995781e-05, "loss": 0.1372, "step": 21701 }, { "epoch": 15.8989010989011, "grad_norm": 0.828125, "learning_rate": 6.197906866029002e-05, "loss": 0.1099, "step": 21702 }, { "epoch": 15.8996336996337, "grad_norm": 1.1640625, "learning_rate": 6.197573567562226e-05, "loss": 0.134, "step": 21703 }, { "epoch": 15.9003663003663, "grad_norm": 1.140625, "learning_rate": 6.197240263597046e-05, "loss": 0.1241, "step": 21704 }, { "epoch": 15.901098901098901, "grad_norm": 0.95703125, "learning_rate": 6.196906954135063e-05, "loss": 0.1097, "step": 21705 }, { "epoch": 15.901831501831502, "grad_norm": 1.2578125, "learning_rate": 6.196573639177871e-05, "loss": 0.1343, "step": 21706 }, { "epoch": 15.902564102564103, "grad_norm": 0.609375, "learning_rate": 6.196240318727067e-05, "loss": 0.1284, "step": 21707 }, { "epoch": 15.903296703296704, "grad_norm": 0.84375, "learning_rate": 6.195906992784252e-05, "loss": 0.1166, "step": 21708 }, { "epoch": 15.904029304029304, "grad_norm": 1.4296875, "learning_rate": 6.195573661351019e-05, "loss": 0.1437, "step": 21709 }, { "epoch": 15.904761904761905, "grad_norm": 0.84765625, "learning_rate": 6.195240324428966e-05, "loss": 0.1244, "step": 21710 }, { "epoch": 15.905494505494506, "grad_norm": 0.96875, "learning_rate": 6.194906982019692e-05, "loss": 0.1191, "step": 21711 }, { "epoch": 15.906227106227107, "grad_norm": 1.171875, "learning_rate": 6.194573634124794e-05, "loss": 0.1224, "step": 21712 }, { "epoch": 15.906959706959707, "grad_norm": 0.85546875, "learning_rate": 6.194240280745865e-05, "loss": 0.1318, "step": 21713 }, { "epoch": 15.907692307692308, "grad_norm": 0.94921875, "learning_rate": 6.193906921884506e-05, "loss": 0.1244, "step": 21714 }, { "epoch": 15.908424908424909, "grad_norm": 0.80859375, "learning_rate": 6.193573557542313e-05, "loss": 0.1148, "step": 21715 }, { "epoch": 15.90915750915751, "grad_norm": 0.50390625, "learning_rate": 6.193240187720884e-05, "loss": 0.0973, "step": 21716 }, { "epoch": 15.90989010989011, "grad_norm": 1.5234375, "learning_rate": 6.192906812421818e-05, "loss": 0.1062, "step": 21717 }, { "epoch": 15.910622710622711, "grad_norm": 0.68359375, "learning_rate": 6.192573431646709e-05, "loss": 0.1131, "step": 21718 }, { "epoch": 15.911355311355312, "grad_norm": 1.3671875, "learning_rate": 6.192240045397155e-05, "loss": 0.1292, "step": 21719 }, { "epoch": 15.912087912087912, "grad_norm": 1.421875, "learning_rate": 6.191906653674753e-05, "loss": 0.1328, "step": 21720 }, { "epoch": 15.912820512820513, "grad_norm": 0.99609375, "learning_rate": 6.191573256481104e-05, "loss": 0.1223, "step": 21721 }, { "epoch": 15.913553113553114, "grad_norm": 0.875, "learning_rate": 6.1912398538178e-05, "loss": 0.1137, "step": 21722 }, { "epoch": 15.914285714285715, "grad_norm": 1.25, "learning_rate": 6.190906445686444e-05, "loss": 0.1158, "step": 21723 }, { "epoch": 15.915018315018315, "grad_norm": 1.2734375, "learning_rate": 6.190573032088628e-05, "loss": 0.1662, "step": 21724 }, { "epoch": 15.915750915750916, "grad_norm": 1.125, "learning_rate": 6.190239613025954e-05, "loss": 0.1377, "step": 21725 }, { "epoch": 15.916483516483517, "grad_norm": 1.1015625, "learning_rate": 6.189906188500017e-05, "loss": 0.1414, "step": 21726 }, { "epoch": 15.917216117216118, "grad_norm": 1.0390625, "learning_rate": 6.189572758512415e-05, "loss": 0.1194, "step": 21727 }, { "epoch": 15.917948717948718, "grad_norm": 0.94140625, "learning_rate": 6.189239323064746e-05, "loss": 0.1229, "step": 21728 }, { "epoch": 15.918681318681319, "grad_norm": 0.83984375, "learning_rate": 6.188905882158609e-05, "loss": 0.0891, "step": 21729 }, { "epoch": 15.91941391941392, "grad_norm": 1.1640625, "learning_rate": 6.188572435795598e-05, "loss": 0.1341, "step": 21730 }, { "epoch": 15.92014652014652, "grad_norm": 0.734375, "learning_rate": 6.188238983977314e-05, "loss": 0.1112, "step": 21731 }, { "epoch": 15.920879120879121, "grad_norm": 1.2578125, "learning_rate": 6.187905526705353e-05, "loss": 0.1337, "step": 21732 }, { "epoch": 15.921611721611722, "grad_norm": 0.890625, "learning_rate": 6.187572063981312e-05, "loss": 0.1, "step": 21733 }, { "epoch": 15.922344322344323, "grad_norm": 0.99609375, "learning_rate": 6.187238595806791e-05, "loss": 0.1424, "step": 21734 }, { "epoch": 15.923076923076923, "grad_norm": 1.484375, "learning_rate": 6.186905122183388e-05, "loss": 0.1128, "step": 21735 }, { "epoch": 15.923809523809524, "grad_norm": 0.91796875, "learning_rate": 6.186571643112697e-05, "loss": 0.1215, "step": 21736 }, { "epoch": 15.924542124542125, "grad_norm": 1.1875, "learning_rate": 6.186238158596318e-05, "loss": 0.1269, "step": 21737 }, { "epoch": 15.925274725274726, "grad_norm": 1.0859375, "learning_rate": 6.18590466863585e-05, "loss": 0.1343, "step": 21738 }, { "epoch": 15.926007326007326, "grad_norm": 1.609375, "learning_rate": 6.185571173232889e-05, "loss": 0.1095, "step": 21739 }, { "epoch": 15.926739926739927, "grad_norm": 1.3203125, "learning_rate": 6.185237672389034e-05, "loss": 0.1354, "step": 21740 }, { "epoch": 15.927472527472528, "grad_norm": 1.515625, "learning_rate": 6.184904166105882e-05, "loss": 0.1049, "step": 21741 }, { "epoch": 15.928205128205128, "grad_norm": 1.8984375, "learning_rate": 6.184570654385034e-05, "loss": 0.1249, "step": 21742 }, { "epoch": 15.92893772893773, "grad_norm": 0.984375, "learning_rate": 6.184237137228082e-05, "loss": 0.0987, "step": 21743 }, { "epoch": 15.92967032967033, "grad_norm": 1.0, "learning_rate": 6.18390361463663e-05, "loss": 0.1282, "step": 21744 }, { "epoch": 15.93040293040293, "grad_norm": 1.46875, "learning_rate": 6.183570086612275e-05, "loss": 0.1564, "step": 21745 }, { "epoch": 15.931135531135531, "grad_norm": 1.03125, "learning_rate": 6.18323655315661e-05, "loss": 0.1265, "step": 21746 }, { "epoch": 15.931868131868132, "grad_norm": 1.328125, "learning_rate": 6.18290301427124e-05, "loss": 0.1115, "step": 21747 }, { "epoch": 15.932600732600733, "grad_norm": 1.171875, "learning_rate": 6.182569469957758e-05, "loss": 0.1182, "step": 21748 }, { "epoch": 15.933333333333334, "grad_norm": 1.8125, "learning_rate": 6.182235920217764e-05, "loss": 0.1213, "step": 21749 }, { "epoch": 15.934065934065934, "grad_norm": 1.421875, "learning_rate": 6.181902365052854e-05, "loss": 0.1274, "step": 21750 }, { "epoch": 15.934798534798535, "grad_norm": 0.80078125, "learning_rate": 6.181568804464631e-05, "loss": 0.1404, "step": 21751 }, { "epoch": 15.935531135531136, "grad_norm": 1.34375, "learning_rate": 6.181235238454689e-05, "loss": 0.1222, "step": 21752 }, { "epoch": 15.936263736263736, "grad_norm": 0.890625, "learning_rate": 6.180901667024628e-05, "loss": 0.1227, "step": 21753 }, { "epoch": 15.936996336996337, "grad_norm": 1.2734375, "learning_rate": 6.180568090176047e-05, "loss": 0.1097, "step": 21754 }, { "epoch": 15.937728937728938, "grad_norm": 1.0234375, "learning_rate": 6.180234507910542e-05, "loss": 0.1304, "step": 21755 }, { "epoch": 15.938461538461539, "grad_norm": 0.82421875, "learning_rate": 6.179900920229712e-05, "loss": 0.1325, "step": 21756 }, { "epoch": 15.93919413919414, "grad_norm": 0.890625, "learning_rate": 6.179567327135157e-05, "loss": 0.1238, "step": 21757 }, { "epoch": 15.93992673992674, "grad_norm": 0.8125, "learning_rate": 6.179233728628474e-05, "loss": 0.1144, "step": 21758 }, { "epoch": 15.94065934065934, "grad_norm": 0.87109375, "learning_rate": 6.17890012471126e-05, "loss": 0.1199, "step": 21759 }, { "epoch": 15.941391941391942, "grad_norm": 1.0859375, "learning_rate": 6.178566515385115e-05, "loss": 0.1501, "step": 21760 }, { "epoch": 15.942124542124542, "grad_norm": 0.71875, "learning_rate": 6.17823290065164e-05, "loss": 0.1043, "step": 21761 }, { "epoch": 15.942857142857143, "grad_norm": 1.0546875, "learning_rate": 6.177899280512428e-05, "loss": 0.138, "step": 21762 }, { "epoch": 15.943589743589744, "grad_norm": 1.2265625, "learning_rate": 6.177565654969082e-05, "loss": 0.1431, "step": 21763 }, { "epoch": 15.944322344322345, "grad_norm": 0.90625, "learning_rate": 6.177232024023197e-05, "loss": 0.1242, "step": 21764 }, { "epoch": 15.945054945054945, "grad_norm": 0.8125, "learning_rate": 6.176898387676375e-05, "loss": 0.1253, "step": 21765 }, { "epoch": 15.945787545787546, "grad_norm": 0.57421875, "learning_rate": 6.176564745930212e-05, "loss": 0.1223, "step": 21766 }, { "epoch": 15.946520146520147, "grad_norm": 0.70703125, "learning_rate": 6.176231098786307e-05, "loss": 0.1328, "step": 21767 }, { "epoch": 15.947252747252747, "grad_norm": 0.8125, "learning_rate": 6.175897446246261e-05, "loss": 0.115, "step": 21768 }, { "epoch": 15.947985347985348, "grad_norm": 0.734375, "learning_rate": 6.17556378831167e-05, "loss": 0.1047, "step": 21769 }, { "epoch": 15.948717948717949, "grad_norm": 0.76171875, "learning_rate": 6.175230124984133e-05, "loss": 0.1186, "step": 21770 }, { "epoch": 15.94945054945055, "grad_norm": 0.73828125, "learning_rate": 6.174896456265251e-05, "loss": 0.1004, "step": 21771 }, { "epoch": 15.95018315018315, "grad_norm": 0.82421875, "learning_rate": 6.174562782156616e-05, "loss": 0.1084, "step": 21772 }, { "epoch": 15.950915750915751, "grad_norm": 1.125, "learning_rate": 6.174229102659835e-05, "loss": 0.1276, "step": 21773 }, { "epoch": 15.951648351648352, "grad_norm": 0.60546875, "learning_rate": 6.173895417776504e-05, "loss": 0.1076, "step": 21774 }, { "epoch": 15.952380952380953, "grad_norm": 0.91796875, "learning_rate": 6.17356172750822e-05, "loss": 0.1325, "step": 21775 }, { "epoch": 15.953113553113553, "grad_norm": 0.51953125, "learning_rate": 6.173228031856582e-05, "loss": 0.1076, "step": 21776 }, { "epoch": 15.953846153846154, "grad_norm": 0.97265625, "learning_rate": 6.17289433082319e-05, "loss": 0.1202, "step": 21777 }, { "epoch": 15.954578754578755, "grad_norm": 1.234375, "learning_rate": 6.172560624409642e-05, "loss": 0.1286, "step": 21778 }, { "epoch": 15.955311355311355, "grad_norm": 1.3828125, "learning_rate": 6.17222691261754e-05, "loss": 0.1415, "step": 21779 }, { "epoch": 15.956043956043956, "grad_norm": 1.2265625, "learning_rate": 6.171893195448478e-05, "loss": 0.1625, "step": 21780 }, { "epoch": 15.956776556776557, "grad_norm": 1.1640625, "learning_rate": 6.171559472904059e-05, "loss": 0.1473, "step": 21781 }, { "epoch": 15.957509157509158, "grad_norm": 0.8984375, "learning_rate": 6.17122574498588e-05, "loss": 0.1421, "step": 21782 }, { "epoch": 15.958241758241758, "grad_norm": 1.109375, "learning_rate": 6.17089201169554e-05, "loss": 0.1287, "step": 21783 }, { "epoch": 15.95897435897436, "grad_norm": 0.734375, "learning_rate": 6.17055827303464e-05, "loss": 0.1159, "step": 21784 }, { "epoch": 15.95970695970696, "grad_norm": 0.74609375, "learning_rate": 6.170224529004775e-05, "loss": 0.1079, "step": 21785 }, { "epoch": 15.96043956043956, "grad_norm": 1.078125, "learning_rate": 6.169890779607548e-05, "loss": 0.1315, "step": 21786 }, { "epoch": 15.961172161172161, "grad_norm": 0.85546875, "learning_rate": 6.169557024844558e-05, "loss": 0.125, "step": 21787 }, { "epoch": 15.961904761904762, "grad_norm": 1.125, "learning_rate": 6.1692232647174e-05, "loss": 0.1187, "step": 21788 }, { "epoch": 15.962637362637363, "grad_norm": 0.80859375, "learning_rate": 6.168889499227676e-05, "loss": 0.1303, "step": 21789 }, { "epoch": 15.963369963369964, "grad_norm": 1.0859375, "learning_rate": 6.168555728376986e-05, "loss": 0.1302, "step": 21790 }, { "epoch": 15.964102564102564, "grad_norm": 1.4140625, "learning_rate": 6.16822195216693e-05, "loss": 0.1293, "step": 21791 }, { "epoch": 15.964835164835165, "grad_norm": 0.91796875, "learning_rate": 6.167888170599103e-05, "loss": 0.1278, "step": 21792 }, { "epoch": 15.965567765567766, "grad_norm": 1.0859375, "learning_rate": 6.167554383675109e-05, "loss": 0.105, "step": 21793 }, { "epoch": 15.966300366300366, "grad_norm": 0.84765625, "learning_rate": 6.167220591396545e-05, "loss": 0.1158, "step": 21794 }, { "epoch": 15.967032967032967, "grad_norm": 0.9296875, "learning_rate": 6.166886793765008e-05, "loss": 0.1221, "step": 21795 }, { "epoch": 15.967765567765568, "grad_norm": 0.8515625, "learning_rate": 6.1665529907821e-05, "loss": 0.1081, "step": 21796 }, { "epoch": 15.968498168498169, "grad_norm": 1.21875, "learning_rate": 6.166219182449422e-05, "loss": 0.1572, "step": 21797 }, { "epoch": 15.96923076923077, "grad_norm": 0.88671875, "learning_rate": 6.16588536876857e-05, "loss": 0.1024, "step": 21798 }, { "epoch": 15.96996336996337, "grad_norm": 1.6796875, "learning_rate": 6.165551549741146e-05, "loss": 0.1254, "step": 21799 }, { "epoch": 15.97069597069597, "grad_norm": 1.2734375, "learning_rate": 6.165217725368749e-05, "loss": 0.16, "step": 21800 }, { "epoch": 15.971428571428572, "grad_norm": 1.15625, "learning_rate": 6.164883895652974e-05, "loss": 0.1087, "step": 21801 }, { "epoch": 15.972161172161172, "grad_norm": 1.0703125, "learning_rate": 6.164550060595427e-05, "loss": 0.1339, "step": 21802 }, { "epoch": 15.972893772893773, "grad_norm": 1.1640625, "learning_rate": 6.164216220197703e-05, "loss": 0.1192, "step": 21803 }, { "epoch": 15.973626373626374, "grad_norm": 1.21875, "learning_rate": 6.163882374461406e-05, "loss": 0.1483, "step": 21804 }, { "epoch": 15.974358974358974, "grad_norm": 1.328125, "learning_rate": 6.163548523388133e-05, "loss": 0.1268, "step": 21805 }, { "epoch": 15.975091575091575, "grad_norm": 0.79296875, "learning_rate": 6.163214666979481e-05, "loss": 0.1176, "step": 21806 }, { "epoch": 15.975824175824176, "grad_norm": 1.171875, "learning_rate": 6.162880805237054e-05, "loss": 0.1542, "step": 21807 }, { "epoch": 15.976556776556777, "grad_norm": 0.9765625, "learning_rate": 6.162546938162448e-05, "loss": 0.1334, "step": 21808 }, { "epoch": 15.977289377289377, "grad_norm": 1.0234375, "learning_rate": 6.162213065757267e-05, "loss": 0.1507, "step": 21809 }, { "epoch": 15.978021978021978, "grad_norm": 1.53125, "learning_rate": 6.161879188023106e-05, "loss": 0.1583, "step": 21810 }, { "epoch": 15.978754578754579, "grad_norm": 1.4609375, "learning_rate": 6.161545304961568e-05, "loss": 0.1105, "step": 21811 }, { "epoch": 15.97948717948718, "grad_norm": 0.94921875, "learning_rate": 6.161211416574251e-05, "loss": 0.1299, "step": 21812 }, { "epoch": 15.98021978021978, "grad_norm": 1.7578125, "learning_rate": 6.160877522862755e-05, "loss": 0.1097, "step": 21813 }, { "epoch": 15.980952380952381, "grad_norm": 0.94921875, "learning_rate": 6.16054362382868e-05, "loss": 0.1153, "step": 21814 }, { "epoch": 15.981684981684982, "grad_norm": 1.0625, "learning_rate": 6.160209719473627e-05, "loss": 0.1075, "step": 21815 }, { "epoch": 15.982417582417582, "grad_norm": 0.93359375, "learning_rate": 6.159875809799191e-05, "loss": 0.0983, "step": 21816 }, { "epoch": 15.983150183150183, "grad_norm": 0.734375, "learning_rate": 6.159541894806981e-05, "loss": 0.1047, "step": 21817 }, { "epoch": 15.983882783882784, "grad_norm": 1.0078125, "learning_rate": 6.159207974498589e-05, "loss": 0.1148, "step": 21818 }, { "epoch": 15.984615384615385, "grad_norm": 0.7734375, "learning_rate": 6.158874048875618e-05, "loss": 0.1216, "step": 21819 }, { "epoch": 15.985347985347985, "grad_norm": 1.1875, "learning_rate": 6.158540117939667e-05, "loss": 0.1365, "step": 21820 }, { "epoch": 15.986080586080586, "grad_norm": 0.7890625, "learning_rate": 6.158206181692336e-05, "loss": 0.1099, "step": 21821 }, { "epoch": 15.986813186813187, "grad_norm": 0.94921875, "learning_rate": 6.157872240135225e-05, "loss": 0.1311, "step": 21822 }, { "epoch": 15.987545787545788, "grad_norm": 1.3203125, "learning_rate": 6.157538293269936e-05, "loss": 0.1473, "step": 21823 }, { "epoch": 15.988278388278388, "grad_norm": 0.71875, "learning_rate": 6.157204341098066e-05, "loss": 0.1081, "step": 21824 }, { "epoch": 15.989010989010989, "grad_norm": 0.91796875, "learning_rate": 6.156870383621218e-05, "loss": 0.1129, "step": 21825 }, { "epoch": 15.98974358974359, "grad_norm": 0.76953125, "learning_rate": 6.156536420840992e-05, "loss": 0.1037, "step": 21826 }, { "epoch": 15.99047619047619, "grad_norm": 1.34375, "learning_rate": 6.156202452758983e-05, "loss": 0.1393, "step": 21827 }, { "epoch": 15.991208791208791, "grad_norm": 1.234375, "learning_rate": 6.155868479376795e-05, "loss": 0.154, "step": 21828 }, { "epoch": 15.991941391941392, "grad_norm": 0.796875, "learning_rate": 6.15553450069603e-05, "loss": 0.121, "step": 21829 }, { "epoch": 15.992673992673993, "grad_norm": 0.83203125, "learning_rate": 6.155200516718285e-05, "loss": 0.129, "step": 21830 }, { "epoch": 15.993406593406593, "grad_norm": 1.203125, "learning_rate": 6.154866527445164e-05, "loss": 0.1471, "step": 21831 }, { "epoch": 15.994139194139194, "grad_norm": 1.28125, "learning_rate": 6.154532532878263e-05, "loss": 0.1137, "step": 21832 }, { "epoch": 15.994871794871795, "grad_norm": 1.0078125, "learning_rate": 6.154198533019184e-05, "loss": 0.1097, "step": 21833 }, { "epoch": 15.995604395604396, "grad_norm": 1.4375, "learning_rate": 6.153864527869526e-05, "loss": 0.0996, "step": 21834 }, { "epoch": 15.996336996336996, "grad_norm": 1.3984375, "learning_rate": 6.153530517430895e-05, "loss": 0.1475, "step": 21835 }, { "epoch": 15.997069597069597, "grad_norm": 0.828125, "learning_rate": 6.153196501704883e-05, "loss": 0.131, "step": 21836 }, { "epoch": 15.997802197802198, "grad_norm": 1.375, "learning_rate": 6.152862480693096e-05, "loss": 0.1348, "step": 21837 }, { "epoch": 15.998534798534799, "grad_norm": 0.73828125, "learning_rate": 6.152528454397133e-05, "loss": 0.1267, "step": 21838 }, { "epoch": 15.9992673992674, "grad_norm": 0.76953125, "learning_rate": 6.152194422818592e-05, "loss": 0.1252, "step": 21839 }, { "epoch": 16.0, "grad_norm": 0.94921875, "learning_rate": 6.151860385959078e-05, "loss": 0.133, "step": 21840 }, { "epoch": 16.0007326007326, "grad_norm": 0.93359375, "learning_rate": 6.151526343820187e-05, "loss": 0.1166, "step": 21841 }, { "epoch": 16.0014652014652, "grad_norm": 1.140625, "learning_rate": 6.151192296403524e-05, "loss": 0.1302, "step": 21842 }, { "epoch": 16.002197802197802, "grad_norm": 0.8828125, "learning_rate": 6.150858243710687e-05, "loss": 0.152, "step": 21843 }, { "epoch": 16.002930402930403, "grad_norm": 1.09375, "learning_rate": 6.150524185743276e-05, "loss": 0.1102, "step": 21844 }, { "epoch": 16.003663003663004, "grad_norm": 1.5546875, "learning_rate": 6.150190122502894e-05, "loss": 0.1444, "step": 21845 }, { "epoch": 16.004395604395604, "grad_norm": 0.640625, "learning_rate": 6.149856053991138e-05, "loss": 0.1165, "step": 21846 }, { "epoch": 16.005128205128205, "grad_norm": 1.1484375, "learning_rate": 6.14952198020961e-05, "loss": 0.1257, "step": 21847 }, { "epoch": 16.005860805860806, "grad_norm": 0.875, "learning_rate": 6.149187901159913e-05, "loss": 0.1479, "step": 21848 }, { "epoch": 16.006593406593407, "grad_norm": 1.0, "learning_rate": 6.148853816843646e-05, "loss": 0.119, "step": 21849 }, { "epoch": 16.007326007326007, "grad_norm": 1.0078125, "learning_rate": 6.148519727262407e-05, "loss": 0.1578, "step": 21850 }, { "epoch": 16.008058608058608, "grad_norm": 1.3984375, "learning_rate": 6.148185632417803e-05, "loss": 0.1301, "step": 21851 }, { "epoch": 16.00879120879121, "grad_norm": 0.9609375, "learning_rate": 6.147851532311428e-05, "loss": 0.1049, "step": 21852 }, { "epoch": 16.00952380952381, "grad_norm": 0.99609375, "learning_rate": 6.147517426944889e-05, "loss": 0.1201, "step": 21853 }, { "epoch": 16.01025641025641, "grad_norm": 1.015625, "learning_rate": 6.147183316319781e-05, "loss": 0.12, "step": 21854 }, { "epoch": 16.01098901098901, "grad_norm": 0.96875, "learning_rate": 6.146849200437707e-05, "loss": 0.1177, "step": 21855 }, { "epoch": 16.01172161172161, "grad_norm": 1.6796875, "learning_rate": 6.146515079300271e-05, "loss": 0.1368, "step": 21856 }, { "epoch": 16.012454212454212, "grad_norm": 1.2421875, "learning_rate": 6.146180952909071e-05, "loss": 0.1567, "step": 21857 }, { "epoch": 16.013186813186813, "grad_norm": 1.0390625, "learning_rate": 6.145846821265707e-05, "loss": 0.1045, "step": 21858 }, { "epoch": 16.013919413919414, "grad_norm": 1.515625, "learning_rate": 6.145512684371782e-05, "loss": 0.1238, "step": 21859 }, { "epoch": 16.014652014652015, "grad_norm": 1.7734375, "learning_rate": 6.145178542228894e-05, "loss": 0.129, "step": 21860 }, { "epoch": 16.015384615384615, "grad_norm": 1.0, "learning_rate": 6.144844394838649e-05, "loss": 0.137, "step": 21861 }, { "epoch": 16.016117216117216, "grad_norm": 1.0546875, "learning_rate": 6.144510242202645e-05, "loss": 0.1153, "step": 21862 }, { "epoch": 16.016849816849817, "grad_norm": 1.4296875, "learning_rate": 6.14417608432248e-05, "loss": 0.1305, "step": 21863 }, { "epoch": 16.017582417582418, "grad_norm": 0.91015625, "learning_rate": 6.143841921199763e-05, "loss": 0.1339, "step": 21864 }, { "epoch": 16.01831501831502, "grad_norm": 1.0078125, "learning_rate": 6.143507752836086e-05, "loss": 0.112, "step": 21865 }, { "epoch": 16.01904761904762, "grad_norm": 1.0, "learning_rate": 6.143173579233056e-05, "loss": 0.1323, "step": 21866 }, { "epoch": 16.01978021978022, "grad_norm": 1.5546875, "learning_rate": 6.14283940039227e-05, "loss": 0.1152, "step": 21867 }, { "epoch": 16.02051282051282, "grad_norm": 1.203125, "learning_rate": 6.142505216315336e-05, "loss": 0.1204, "step": 21868 }, { "epoch": 16.02124542124542, "grad_norm": 1.21875, "learning_rate": 6.14217102700385e-05, "loss": 0.111, "step": 21869 }, { "epoch": 16.021978021978022, "grad_norm": 1.3046875, "learning_rate": 6.141836832459412e-05, "loss": 0.1704, "step": 21870 }, { "epoch": 16.022710622710623, "grad_norm": 1.3046875, "learning_rate": 6.141502632683625e-05, "loss": 0.1238, "step": 21871 }, { "epoch": 16.023443223443223, "grad_norm": 1.0078125, "learning_rate": 6.141168427678092e-05, "loss": 0.1195, "step": 21872 }, { "epoch": 16.024175824175824, "grad_norm": 0.7578125, "learning_rate": 6.140834217444413e-05, "loss": 0.1168, "step": 21873 }, { "epoch": 16.024908424908425, "grad_norm": 1.078125, "learning_rate": 6.14050000198419e-05, "loss": 0.115, "step": 21874 }, { "epoch": 16.025641025641026, "grad_norm": 1.328125, "learning_rate": 6.140165781299023e-05, "loss": 0.1416, "step": 21875 }, { "epoch": 16.026373626373626, "grad_norm": 0.98046875, "learning_rate": 6.139831555390512e-05, "loss": 0.1193, "step": 21876 }, { "epoch": 16.027106227106227, "grad_norm": 0.8515625, "learning_rate": 6.139497324260261e-05, "loss": 0.1315, "step": 21877 }, { "epoch": 16.027838827838828, "grad_norm": 0.921875, "learning_rate": 6.139163087909871e-05, "loss": 0.1167, "step": 21878 }, { "epoch": 16.02857142857143, "grad_norm": 0.97265625, "learning_rate": 6.138828846340945e-05, "loss": 0.1335, "step": 21879 }, { "epoch": 16.02930402930403, "grad_norm": 1.3046875, "learning_rate": 6.13849459955508e-05, "loss": 0.1091, "step": 21880 }, { "epoch": 16.03003663003663, "grad_norm": 0.99609375, "learning_rate": 6.13816034755388e-05, "loss": 0.1123, "step": 21881 }, { "epoch": 16.03076923076923, "grad_norm": 0.96875, "learning_rate": 6.137826090338949e-05, "loss": 0.1367, "step": 21882 }, { "epoch": 16.03150183150183, "grad_norm": 1.28125, "learning_rate": 6.137491827911883e-05, "loss": 0.1475, "step": 21883 }, { "epoch": 16.032234432234432, "grad_norm": 0.91015625, "learning_rate": 6.137157560274288e-05, "loss": 0.1198, "step": 21884 }, { "epoch": 16.032967032967033, "grad_norm": 0.8359375, "learning_rate": 6.136823287427764e-05, "loss": 0.1241, "step": 21885 }, { "epoch": 16.033699633699634, "grad_norm": 1.046875, "learning_rate": 6.136489009373913e-05, "loss": 0.1029, "step": 21886 }, { "epoch": 16.034432234432234, "grad_norm": 1.5625, "learning_rate": 6.136154726114338e-05, "loss": 0.1234, "step": 21887 }, { "epoch": 16.035164835164835, "grad_norm": 1.296875, "learning_rate": 6.135820437650638e-05, "loss": 0.117, "step": 21888 }, { "epoch": 16.035897435897436, "grad_norm": 1.7265625, "learning_rate": 6.135486143984415e-05, "loss": 0.154, "step": 21889 }, { "epoch": 16.036630036630036, "grad_norm": 1.609375, "learning_rate": 6.135151845117274e-05, "loss": 0.1073, "step": 21890 }, { "epoch": 16.037362637362637, "grad_norm": 0.9453125, "learning_rate": 6.13481754105081e-05, "loss": 0.1077, "step": 21891 }, { "epoch": 16.038095238095238, "grad_norm": 1.2421875, "learning_rate": 6.134483231786634e-05, "loss": 0.155, "step": 21892 }, { "epoch": 16.03882783882784, "grad_norm": 1.09375, "learning_rate": 6.134148917326339e-05, "loss": 0.1223, "step": 21893 }, { "epoch": 16.03956043956044, "grad_norm": 1.6484375, "learning_rate": 6.133814597671531e-05, "loss": 0.1165, "step": 21894 }, { "epoch": 16.04029304029304, "grad_norm": 1.234375, "learning_rate": 6.133480272823812e-05, "loss": 0.1215, "step": 21895 }, { "epoch": 16.04102564102564, "grad_norm": 1.71875, "learning_rate": 6.133145942784784e-05, "loss": 0.1495, "step": 21896 }, { "epoch": 16.04175824175824, "grad_norm": 1.0625, "learning_rate": 6.132811607556049e-05, "loss": 0.1001, "step": 21897 }, { "epoch": 16.042490842490842, "grad_norm": 1.390625, "learning_rate": 6.132477267139209e-05, "loss": 0.1243, "step": 21898 }, { "epoch": 16.043223443223443, "grad_norm": 1.4609375, "learning_rate": 6.13214292153586e-05, "loss": 0.1132, "step": 21899 }, { "epoch": 16.043956043956044, "grad_norm": 1.4375, "learning_rate": 6.131808570747613e-05, "loss": 0.1318, "step": 21900 }, { "epoch": 16.044688644688645, "grad_norm": 1.4765625, "learning_rate": 6.131474214776068e-05, "loss": 0.1297, "step": 21901 }, { "epoch": 16.045421245421245, "grad_norm": 1.390625, "learning_rate": 6.13113985362282e-05, "loss": 0.142, "step": 21902 }, { "epoch": 16.046153846153846, "grad_norm": 1.015625, "learning_rate": 6.130805487289479e-05, "loss": 0.0908, "step": 21903 }, { "epoch": 16.046886446886447, "grad_norm": 1.34375, "learning_rate": 6.130471115777643e-05, "loss": 0.1379, "step": 21904 }, { "epoch": 16.047619047619047, "grad_norm": 1.3203125, "learning_rate": 6.130136739088916e-05, "loss": 0.1603, "step": 21905 }, { "epoch": 16.048351648351648, "grad_norm": 1.28125, "learning_rate": 6.129802357224899e-05, "loss": 0.1361, "step": 21906 }, { "epoch": 16.04908424908425, "grad_norm": 0.9296875, "learning_rate": 6.129467970187193e-05, "loss": 0.1162, "step": 21907 }, { "epoch": 16.04981684981685, "grad_norm": 1.375, "learning_rate": 6.129133577977403e-05, "loss": 0.144, "step": 21908 }, { "epoch": 16.05054945054945, "grad_norm": 1.5, "learning_rate": 6.12879918059713e-05, "loss": 0.1274, "step": 21909 }, { "epoch": 16.05128205128205, "grad_norm": 0.70703125, "learning_rate": 6.128464778047976e-05, "loss": 0.0991, "step": 21910 }, { "epoch": 16.052014652014652, "grad_norm": 0.8203125, "learning_rate": 6.128130370331542e-05, "loss": 0.1263, "step": 21911 }, { "epoch": 16.052747252747253, "grad_norm": 0.80859375, "learning_rate": 6.127795957449433e-05, "loss": 0.1122, "step": 21912 }, { "epoch": 16.053479853479853, "grad_norm": 1.0078125, "learning_rate": 6.127461539403248e-05, "loss": 0.1228, "step": 21913 }, { "epoch": 16.054212454212454, "grad_norm": 1.0703125, "learning_rate": 6.127127116194594e-05, "loss": 0.126, "step": 21914 }, { "epoch": 16.054945054945055, "grad_norm": 0.80859375, "learning_rate": 6.126792687825069e-05, "loss": 0.1159, "step": 21915 }, { "epoch": 16.055677655677655, "grad_norm": 1.015625, "learning_rate": 6.126458254296275e-05, "loss": 0.1156, "step": 21916 }, { "epoch": 16.056410256410256, "grad_norm": 0.96484375, "learning_rate": 6.126123815609816e-05, "loss": 0.1145, "step": 21917 }, { "epoch": 16.057142857142857, "grad_norm": 1.7734375, "learning_rate": 6.125789371767297e-05, "loss": 0.1354, "step": 21918 }, { "epoch": 16.057875457875458, "grad_norm": 1.6796875, "learning_rate": 6.125454922770317e-05, "loss": 0.1372, "step": 21919 }, { "epoch": 16.05860805860806, "grad_norm": 1.5546875, "learning_rate": 6.12512046862048e-05, "loss": 0.1385, "step": 21920 }, { "epoch": 16.05934065934066, "grad_norm": 0.921875, "learning_rate": 6.124786009319387e-05, "loss": 0.1356, "step": 21921 }, { "epoch": 16.06007326007326, "grad_norm": 1.25, "learning_rate": 6.124451544868642e-05, "loss": 0.1467, "step": 21922 }, { "epoch": 16.06080586080586, "grad_norm": 1.0859375, "learning_rate": 6.124117075269846e-05, "loss": 0.1366, "step": 21923 }, { "epoch": 16.06153846153846, "grad_norm": 1.1015625, "learning_rate": 6.123782600524604e-05, "loss": 0.148, "step": 21924 }, { "epoch": 16.062271062271062, "grad_norm": 1.2890625, "learning_rate": 6.123448120634515e-05, "loss": 0.1492, "step": 21925 }, { "epoch": 16.063003663003663, "grad_norm": 1.0703125, "learning_rate": 6.123113635601185e-05, "loss": 0.142, "step": 21926 }, { "epoch": 16.063736263736264, "grad_norm": 1.015625, "learning_rate": 6.122779145426216e-05, "loss": 0.1242, "step": 21927 }, { "epoch": 16.064468864468864, "grad_norm": 1.6328125, "learning_rate": 6.122444650111208e-05, "loss": 0.1196, "step": 21928 }, { "epoch": 16.065201465201465, "grad_norm": 1.3125, "learning_rate": 6.122110149657766e-05, "loss": 0.1387, "step": 21929 }, { "epoch": 16.065934065934066, "grad_norm": 0.98046875, "learning_rate": 6.121775644067493e-05, "loss": 0.1188, "step": 21930 }, { "epoch": 16.066666666666666, "grad_norm": 1.15625, "learning_rate": 6.121441133341991e-05, "loss": 0.1421, "step": 21931 }, { "epoch": 16.067399267399267, "grad_norm": 0.92578125, "learning_rate": 6.121106617482862e-05, "loss": 0.1003, "step": 21932 }, { "epoch": 16.068131868131868, "grad_norm": 1.171875, "learning_rate": 6.12077209649171e-05, "loss": 0.1058, "step": 21933 }, { "epoch": 16.06886446886447, "grad_norm": 1.1953125, "learning_rate": 6.120437570370139e-05, "loss": 0.1271, "step": 21934 }, { "epoch": 16.06959706959707, "grad_norm": 1.53125, "learning_rate": 6.120103039119749e-05, "loss": 0.1315, "step": 21935 }, { "epoch": 16.07032967032967, "grad_norm": 1.9375, "learning_rate": 6.119768502742144e-05, "loss": 0.1631, "step": 21936 }, { "epoch": 16.07106227106227, "grad_norm": 1.6796875, "learning_rate": 6.119433961238927e-05, "loss": 0.1246, "step": 21937 }, { "epoch": 16.07179487179487, "grad_norm": 1.3125, "learning_rate": 6.119099414611699e-05, "loss": 0.1193, "step": 21938 }, { "epoch": 16.072527472527472, "grad_norm": 1.15625, "learning_rate": 6.118764862862067e-05, "loss": 0.1301, "step": 21939 }, { "epoch": 16.073260073260073, "grad_norm": 1.0, "learning_rate": 6.118430305991633e-05, "loss": 0.1235, "step": 21940 }, { "epoch": 16.073992673992674, "grad_norm": 0.87890625, "learning_rate": 6.118095744001996e-05, "loss": 0.1152, "step": 21941 }, { "epoch": 16.074725274725274, "grad_norm": 0.77734375, "learning_rate": 6.117761176894762e-05, "loss": 0.0975, "step": 21942 }, { "epoch": 16.075457875457875, "grad_norm": 1.15625, "learning_rate": 6.117426604671535e-05, "loss": 0.1234, "step": 21943 }, { "epoch": 16.076190476190476, "grad_norm": 1.15625, "learning_rate": 6.117092027333915e-05, "loss": 0.126, "step": 21944 }, { "epoch": 16.076923076923077, "grad_norm": 1.453125, "learning_rate": 6.116757444883508e-05, "loss": 0.103, "step": 21945 }, { "epoch": 16.077655677655677, "grad_norm": 0.92578125, "learning_rate": 6.116422857321913e-05, "loss": 0.1688, "step": 21946 }, { "epoch": 16.078388278388278, "grad_norm": 1.78125, "learning_rate": 6.116088264650741e-05, "loss": 0.1356, "step": 21947 }, { "epoch": 16.07912087912088, "grad_norm": 1.4609375, "learning_rate": 6.115753666871588e-05, "loss": 0.1132, "step": 21948 }, { "epoch": 16.07985347985348, "grad_norm": 1.359375, "learning_rate": 6.115419063986058e-05, "loss": 0.1426, "step": 21949 }, { "epoch": 16.08058608058608, "grad_norm": 1.1875, "learning_rate": 6.115084455995757e-05, "loss": 0.1336, "step": 21950 }, { "epoch": 16.08131868131868, "grad_norm": 1.34375, "learning_rate": 6.114749842902287e-05, "loss": 0.1166, "step": 21951 }, { "epoch": 16.08205128205128, "grad_norm": 1.2421875, "learning_rate": 6.114415224707249e-05, "loss": 0.1428, "step": 21952 }, { "epoch": 16.082783882783882, "grad_norm": 1.1484375, "learning_rate": 6.114080601412251e-05, "loss": 0.1515, "step": 21953 }, { "epoch": 16.083516483516483, "grad_norm": 1.65625, "learning_rate": 6.11374597301889e-05, "loss": 0.1367, "step": 21954 }, { "epoch": 16.084249084249084, "grad_norm": 1.703125, "learning_rate": 6.113411339528776e-05, "loss": 0.1508, "step": 21955 }, { "epoch": 16.084981684981685, "grad_norm": 1.0859375, "learning_rate": 6.113076700943507e-05, "loss": 0.1232, "step": 21956 }, { "epoch": 16.085714285714285, "grad_norm": 1.4375, "learning_rate": 6.11274205726469e-05, "loss": 0.1056, "step": 21957 }, { "epoch": 16.086446886446886, "grad_norm": 1.0625, "learning_rate": 6.112407408493927e-05, "loss": 0.1324, "step": 21958 }, { "epoch": 16.087179487179487, "grad_norm": 1.0, "learning_rate": 6.11207275463282e-05, "loss": 0.1595, "step": 21959 }, { "epoch": 16.087912087912088, "grad_norm": 0.86328125, "learning_rate": 6.111738095682975e-05, "loss": 0.105, "step": 21960 }, { "epoch": 16.08864468864469, "grad_norm": 1.125, "learning_rate": 6.111403431645994e-05, "loss": 0.1097, "step": 21961 }, { "epoch": 16.08937728937729, "grad_norm": 0.75390625, "learning_rate": 6.11106876252348e-05, "loss": 0.1267, "step": 21962 }, { "epoch": 16.09010989010989, "grad_norm": 1.125, "learning_rate": 6.110734088317039e-05, "loss": 0.1261, "step": 21963 }, { "epoch": 16.09084249084249, "grad_norm": 1.28125, "learning_rate": 6.110399409028271e-05, "loss": 0.1078, "step": 21964 }, { "epoch": 16.09157509157509, "grad_norm": 0.66015625, "learning_rate": 6.110064724658781e-05, "loss": 0.1286, "step": 21965 }, { "epoch": 16.092307692307692, "grad_norm": 0.765625, "learning_rate": 6.109730035210174e-05, "loss": 0.1069, "step": 21966 }, { "epoch": 16.093040293040293, "grad_norm": 1.03125, "learning_rate": 6.109395340684052e-05, "loss": 0.1292, "step": 21967 }, { "epoch": 16.093772893772893, "grad_norm": 1.3828125, "learning_rate": 6.10906064108202e-05, "loss": 0.1304, "step": 21968 }, { "epoch": 16.094505494505494, "grad_norm": 1.234375, "learning_rate": 6.10872593640568e-05, "loss": 0.1453, "step": 21969 }, { "epoch": 16.095238095238095, "grad_norm": 1.3828125, "learning_rate": 6.108391226656635e-05, "loss": 0.1165, "step": 21970 }, { "epoch": 16.095970695970696, "grad_norm": 0.79296875, "learning_rate": 6.108056511836492e-05, "loss": 0.0973, "step": 21971 }, { "epoch": 16.096703296703296, "grad_norm": 0.88671875, "learning_rate": 6.107721791946855e-05, "loss": 0.1242, "step": 21972 }, { "epoch": 16.097435897435897, "grad_norm": 1.1171875, "learning_rate": 6.107387066989321e-05, "loss": 0.1233, "step": 21973 }, { "epoch": 16.098168498168498, "grad_norm": 1.09375, "learning_rate": 6.107052336965502e-05, "loss": 0.14, "step": 21974 }, { "epoch": 16.0989010989011, "grad_norm": 1.015625, "learning_rate": 6.106717601876998e-05, "loss": 0.1241, "step": 21975 }, { "epoch": 16.0996336996337, "grad_norm": 1.40625, "learning_rate": 6.106382861725411e-05, "loss": 0.135, "step": 21976 }, { "epoch": 16.1003663003663, "grad_norm": 0.953125, "learning_rate": 6.10604811651235e-05, "loss": 0.153, "step": 21977 }, { "epoch": 16.1010989010989, "grad_norm": 0.90234375, "learning_rate": 6.105713366239412e-05, "loss": 0.1241, "step": 21978 }, { "epoch": 16.1018315018315, "grad_norm": 0.81640625, "learning_rate": 6.105378610908207e-05, "loss": 0.1287, "step": 21979 }, { "epoch": 16.102564102564102, "grad_norm": 1.203125, "learning_rate": 6.105043850520336e-05, "loss": 0.0939, "step": 21980 }, { "epoch": 16.103296703296703, "grad_norm": 1.2734375, "learning_rate": 6.104709085077406e-05, "loss": 0.1189, "step": 21981 }, { "epoch": 16.104029304029304, "grad_norm": 0.76171875, "learning_rate": 6.104374314581015e-05, "loss": 0.114, "step": 21982 }, { "epoch": 16.104761904761904, "grad_norm": 1.0546875, "learning_rate": 6.104039539032771e-05, "loss": 0.15, "step": 21983 }, { "epoch": 16.105494505494505, "grad_norm": 1.1484375, "learning_rate": 6.10370475843428e-05, "loss": 0.1328, "step": 21984 }, { "epoch": 16.106227106227106, "grad_norm": 1.09375, "learning_rate": 6.1033699727871406e-05, "loss": 0.1173, "step": 21985 }, { "epoch": 16.106959706959707, "grad_norm": 0.7890625, "learning_rate": 6.103035182092962e-05, "loss": 0.111, "step": 21986 }, { "epoch": 16.107692307692307, "grad_norm": 0.92578125, "learning_rate": 6.102700386353346e-05, "loss": 0.1251, "step": 21987 }, { "epoch": 16.108424908424908, "grad_norm": 0.86328125, "learning_rate": 6.1023655855698955e-05, "loss": 0.1178, "step": 21988 }, { "epoch": 16.10915750915751, "grad_norm": 1.296875, "learning_rate": 6.1020307797442165e-05, "loss": 0.1329, "step": 21989 }, { "epoch": 16.10989010989011, "grad_norm": 0.984375, "learning_rate": 6.101695968877915e-05, "loss": 0.1173, "step": 21990 }, { "epoch": 16.11062271062271, "grad_norm": 1.0625, "learning_rate": 6.10136115297259e-05, "loss": 0.1015, "step": 21991 }, { "epoch": 16.11135531135531, "grad_norm": 0.84375, "learning_rate": 6.10102633202985e-05, "loss": 0.1014, "step": 21992 }, { "epoch": 16.11208791208791, "grad_norm": 0.9375, "learning_rate": 6.100691506051297e-05, "loss": 0.1262, "step": 21993 }, { "epoch": 16.112820512820512, "grad_norm": 1.2890625, "learning_rate": 6.100356675038536e-05, "loss": 0.1718, "step": 21994 }, { "epoch": 16.113553113553113, "grad_norm": 1.21875, "learning_rate": 6.100021838993171e-05, "loss": 0.1366, "step": 21995 }, { "epoch": 16.114285714285714, "grad_norm": 1.21875, "learning_rate": 6.099686997916807e-05, "loss": 0.1433, "step": 21996 }, { "epoch": 16.115018315018315, "grad_norm": 1.5546875, "learning_rate": 6.0993521518110486e-05, "loss": 0.1182, "step": 21997 }, { "epoch": 16.115750915750915, "grad_norm": 1.28125, "learning_rate": 6.099017300677499e-05, "loss": 0.1159, "step": 21998 }, { "epoch": 16.116483516483516, "grad_norm": 1.140625, "learning_rate": 6.0986824445177636e-05, "loss": 0.095, "step": 21999 }, { "epoch": 16.117216117216117, "grad_norm": 1.875, "learning_rate": 6.098347583333447e-05, "loss": 0.1388, "step": 22000 }, { "epoch": 16.117948717948718, "grad_norm": 0.7109375, "learning_rate": 6.098012717126153e-05, "loss": 0.1175, "step": 22001 }, { "epoch": 16.11868131868132, "grad_norm": 0.89453125, "learning_rate": 6.097677845897485e-05, "loss": 0.1304, "step": 22002 }, { "epoch": 16.11941391941392, "grad_norm": 0.89453125, "learning_rate": 6.0973429696490494e-05, "loss": 0.1397, "step": 22003 }, { "epoch": 16.12014652014652, "grad_norm": 1.0625, "learning_rate": 6.097008088382447e-05, "loss": 0.1194, "step": 22004 }, { "epoch": 16.12087912087912, "grad_norm": 1.328125, "learning_rate": 6.0966732020992886e-05, "loss": 0.1021, "step": 22005 }, { "epoch": 16.12161172161172, "grad_norm": 1.125, "learning_rate": 6.096338310801173e-05, "loss": 0.1064, "step": 22006 }, { "epoch": 16.122344322344322, "grad_norm": 1.875, "learning_rate": 6.0960034144897094e-05, "loss": 0.1317, "step": 22007 }, { "epoch": 16.123076923076923, "grad_norm": 0.91796875, "learning_rate": 6.095668513166498e-05, "loss": 0.1251, "step": 22008 }, { "epoch": 16.123809523809523, "grad_norm": 1.2421875, "learning_rate": 6.095333606833146e-05, "loss": 0.1516, "step": 22009 }, { "epoch": 16.124542124542124, "grad_norm": 1.34375, "learning_rate": 6.0949986954912574e-05, "loss": 0.1184, "step": 22010 }, { "epoch": 16.125274725274725, "grad_norm": 1.109375, "learning_rate": 6.0946637791424365e-05, "loss": 0.1212, "step": 22011 }, { "epoch": 16.126007326007326, "grad_norm": 0.953125, "learning_rate": 6.0943288577882906e-05, "loss": 0.1132, "step": 22012 }, { "epoch": 16.126739926739926, "grad_norm": 1.1796875, "learning_rate": 6.0939939314304215e-05, "loss": 0.1392, "step": 22013 }, { "epoch": 16.127472527472527, "grad_norm": 0.9375, "learning_rate": 6.093659000070433e-05, "loss": 0.1166, "step": 22014 }, { "epoch": 16.128205128205128, "grad_norm": 1.2109375, "learning_rate": 6.093324063709933e-05, "loss": 0.1391, "step": 22015 }, { "epoch": 16.12893772893773, "grad_norm": 0.88671875, "learning_rate": 6.0929891223505265e-05, "loss": 0.122, "step": 22016 }, { "epoch": 16.12967032967033, "grad_norm": 0.94140625, "learning_rate": 6.092654175993814e-05, "loss": 0.1123, "step": 22017 }, { "epoch": 16.13040293040293, "grad_norm": 2.0, "learning_rate": 6.092319224641404e-05, "loss": 0.1255, "step": 22018 }, { "epoch": 16.13113553113553, "grad_norm": 0.66015625, "learning_rate": 6.0919842682949005e-05, "loss": 0.1134, "step": 22019 }, { "epoch": 16.13186813186813, "grad_norm": 1.65625, "learning_rate": 6.091649306955908e-05, "loss": 0.1355, "step": 22020 }, { "epoch": 16.132600732600732, "grad_norm": 1.4921875, "learning_rate": 6.091314340626032e-05, "loss": 0.1447, "step": 22021 }, { "epoch": 16.133333333333333, "grad_norm": 0.80078125, "learning_rate": 6.090979369306876e-05, "loss": 0.1199, "step": 22022 }, { "epoch": 16.134065934065934, "grad_norm": 1.453125, "learning_rate": 6.090644393000049e-05, "loss": 0.1287, "step": 22023 }, { "epoch": 16.134798534798534, "grad_norm": 0.67578125, "learning_rate": 6.0903094117071514e-05, "loss": 0.1011, "step": 22024 }, { "epoch": 16.135531135531135, "grad_norm": 1.3046875, "learning_rate": 6.0899744254297894e-05, "loss": 0.1378, "step": 22025 }, { "epoch": 16.136263736263736, "grad_norm": 1.2265625, "learning_rate": 6.0896394341695705e-05, "loss": 0.1243, "step": 22026 }, { "epoch": 16.136996336996337, "grad_norm": 0.80078125, "learning_rate": 6.089304437928095e-05, "loss": 0.1258, "step": 22027 }, { "epoch": 16.137728937728937, "grad_norm": 0.89453125, "learning_rate": 6.088969436706973e-05, "loss": 0.1125, "step": 22028 }, { "epoch": 16.138461538461538, "grad_norm": 0.80859375, "learning_rate": 6.0886344305078066e-05, "loss": 0.1229, "step": 22029 }, { "epoch": 16.13919413919414, "grad_norm": 1.0234375, "learning_rate": 6.0882994193322006e-05, "loss": 0.1314, "step": 22030 }, { "epoch": 16.13992673992674, "grad_norm": 1.546875, "learning_rate": 6.0879644031817627e-05, "loss": 0.1304, "step": 22031 }, { "epoch": 16.14065934065934, "grad_norm": 0.8515625, "learning_rate": 6.087629382058097e-05, "loss": 0.1296, "step": 22032 }, { "epoch": 16.14139194139194, "grad_norm": 0.73046875, "learning_rate": 6.087294355962805e-05, "loss": 0.1094, "step": 22033 }, { "epoch": 16.14212454212454, "grad_norm": 0.8046875, "learning_rate": 6.0869593248974986e-05, "loss": 0.1227, "step": 22034 }, { "epoch": 16.142857142857142, "grad_norm": 1.1171875, "learning_rate": 6.086624288863778e-05, "loss": 0.1495, "step": 22035 }, { "epoch": 16.143589743589743, "grad_norm": 1.0078125, "learning_rate": 6.08628924786325e-05, "loss": 0.1161, "step": 22036 }, { "epoch": 16.144322344322344, "grad_norm": 0.66796875, "learning_rate": 6.0859542018975216e-05, "loss": 0.1046, "step": 22037 }, { "epoch": 16.145054945054945, "grad_norm": 1.7734375, "learning_rate": 6.085619150968196e-05, "loss": 0.143, "step": 22038 }, { "epoch": 16.145787545787545, "grad_norm": 1.6953125, "learning_rate": 6.08528409507688e-05, "loss": 0.1307, "step": 22039 }, { "epoch": 16.146520146520146, "grad_norm": 2.046875, "learning_rate": 6.084949034225176e-05, "loss": 0.1429, "step": 22040 }, { "epoch": 16.147252747252747, "grad_norm": 1.234375, "learning_rate": 6.084613968414692e-05, "loss": 0.1237, "step": 22041 }, { "epoch": 16.147985347985347, "grad_norm": 1.0859375, "learning_rate": 6.0842788976470336e-05, "loss": 0.1465, "step": 22042 }, { "epoch": 16.148717948717948, "grad_norm": 1.3203125, "learning_rate": 6.083943821923804e-05, "loss": 0.149, "step": 22043 }, { "epoch": 16.14945054945055, "grad_norm": 1.4453125, "learning_rate": 6.083608741246611e-05, "loss": 0.1266, "step": 22044 }, { "epoch": 16.15018315018315, "grad_norm": 1.09375, "learning_rate": 6.083273655617059e-05, "loss": 0.1101, "step": 22045 }, { "epoch": 16.15091575091575, "grad_norm": 0.67578125, "learning_rate": 6.0829385650367546e-05, "loss": 0.1144, "step": 22046 }, { "epoch": 16.15164835164835, "grad_norm": 0.78125, "learning_rate": 6.082603469507302e-05, "loss": 0.1016, "step": 22047 }, { "epoch": 16.152380952380952, "grad_norm": 1.234375, "learning_rate": 6.082268369030306e-05, "loss": 0.1284, "step": 22048 }, { "epoch": 16.153113553113553, "grad_norm": 1.125, "learning_rate": 6.0819332636073744e-05, "loss": 0.1216, "step": 22049 }, { "epoch": 16.153846153846153, "grad_norm": 1.28125, "learning_rate": 6.081598153240111e-05, "loss": 0.1187, "step": 22050 }, { "epoch": 16.154578754578754, "grad_norm": 1.0078125, "learning_rate": 6.081263037930124e-05, "loss": 0.1261, "step": 22051 }, { "epoch": 16.155311355311355, "grad_norm": 0.8046875, "learning_rate": 6.080927917679018e-05, "loss": 0.1127, "step": 22052 }, { "epoch": 16.156043956043955, "grad_norm": 1.453125, "learning_rate": 6.0805927924883956e-05, "loss": 0.1159, "step": 22053 }, { "epoch": 16.156776556776556, "grad_norm": 0.890625, "learning_rate": 6.080257662359865e-05, "loss": 0.1182, "step": 22054 }, { "epoch": 16.157509157509157, "grad_norm": 0.8828125, "learning_rate": 6.079922527295033e-05, "loss": 0.1382, "step": 22055 }, { "epoch": 16.158241758241758, "grad_norm": 0.9921875, "learning_rate": 6.079587387295502e-05, "loss": 0.0995, "step": 22056 }, { "epoch": 16.15897435897436, "grad_norm": 1.09375, "learning_rate": 6.0792522423628815e-05, "loss": 0.1251, "step": 22057 }, { "epoch": 16.15970695970696, "grad_norm": 0.5703125, "learning_rate": 6.0789170924987736e-05, "loss": 0.0962, "step": 22058 }, { "epoch": 16.16043956043956, "grad_norm": 0.75390625, "learning_rate": 6.078581937704788e-05, "loss": 0.1155, "step": 22059 }, { "epoch": 16.16117216117216, "grad_norm": 1.1796875, "learning_rate": 6.0782467779825293e-05, "loss": 0.1278, "step": 22060 }, { "epoch": 16.16190476190476, "grad_norm": 1.3046875, "learning_rate": 6.077911613333601e-05, "loss": 0.117, "step": 22061 }, { "epoch": 16.162637362637362, "grad_norm": 1.359375, "learning_rate": 6.0775764437596116e-05, "loss": 0.1359, "step": 22062 }, { "epoch": 16.163369963369963, "grad_norm": 1.109375, "learning_rate": 6.077241269262166e-05, "loss": 0.1354, "step": 22063 }, { "epoch": 16.164102564102564, "grad_norm": 0.8671875, "learning_rate": 6.076906089842871e-05, "loss": 0.1062, "step": 22064 }, { "epoch": 16.164835164835164, "grad_norm": 1.140625, "learning_rate": 6.076570905503331e-05, "loss": 0.1074, "step": 22065 }, { "epoch": 16.165567765567765, "grad_norm": 1.4921875, "learning_rate": 6.076235716245152e-05, "loss": 0.1594, "step": 22066 }, { "epoch": 16.166300366300366, "grad_norm": 0.859375, "learning_rate": 6.0759005220699414e-05, "loss": 0.13, "step": 22067 }, { "epoch": 16.167032967032966, "grad_norm": 0.8984375, "learning_rate": 6.075565322979306e-05, "loss": 0.108, "step": 22068 }, { "epoch": 16.167765567765567, "grad_norm": 0.9375, "learning_rate": 6.0752301189748486e-05, "loss": 0.1434, "step": 22069 }, { "epoch": 16.168498168498168, "grad_norm": 0.6953125, "learning_rate": 6.0748949100581786e-05, "loss": 0.1221, "step": 22070 }, { "epoch": 16.16923076923077, "grad_norm": 0.95703125, "learning_rate": 6.0745596962308994e-05, "loss": 0.1308, "step": 22071 }, { "epoch": 16.16996336996337, "grad_norm": 0.8046875, "learning_rate": 6.074224477494618e-05, "loss": 0.1231, "step": 22072 }, { "epoch": 16.17069597069597, "grad_norm": 1.5390625, "learning_rate": 6.073889253850943e-05, "loss": 0.1497, "step": 22073 }, { "epoch": 16.17142857142857, "grad_norm": 1.0859375, "learning_rate": 6.0735540253014756e-05, "loss": 0.1245, "step": 22074 }, { "epoch": 16.17216117216117, "grad_norm": 1.25, "learning_rate": 6.073218791847828e-05, "loss": 0.1625, "step": 22075 }, { "epoch": 16.172893772893772, "grad_norm": 1.3359375, "learning_rate": 6.072883553491601e-05, "loss": 0.1073, "step": 22076 }, { "epoch": 16.173626373626373, "grad_norm": 0.76171875, "learning_rate": 6.072548310234405e-05, "loss": 0.1244, "step": 22077 }, { "epoch": 16.174358974358974, "grad_norm": 1.1875, "learning_rate": 6.072213062077843e-05, "loss": 0.1022, "step": 22078 }, { "epoch": 16.175091575091574, "grad_norm": 1.0859375, "learning_rate": 6.071877809023523e-05, "loss": 0.1152, "step": 22079 }, { "epoch": 16.175824175824175, "grad_norm": 0.93359375, "learning_rate": 6.07154255107305e-05, "loss": 0.1408, "step": 22080 }, { "epoch": 16.176556776556776, "grad_norm": 0.828125, "learning_rate": 6.071207288228033e-05, "loss": 0.1356, "step": 22081 }, { "epoch": 16.177289377289377, "grad_norm": 0.9375, "learning_rate": 6.070872020490076e-05, "loss": 0.1242, "step": 22082 }, { "epoch": 16.178021978021977, "grad_norm": 0.99609375, "learning_rate": 6.0705367478607855e-05, "loss": 0.1169, "step": 22083 }, { "epoch": 16.178754578754578, "grad_norm": 1.109375, "learning_rate": 6.070201470341768e-05, "loss": 0.1288, "step": 22084 }, { "epoch": 16.17948717948718, "grad_norm": 1.84375, "learning_rate": 6.069866187934631e-05, "loss": 0.1498, "step": 22085 }, { "epoch": 16.18021978021978, "grad_norm": 1.21875, "learning_rate": 6.069530900640981e-05, "loss": 0.1452, "step": 22086 }, { "epoch": 16.18095238095238, "grad_norm": 1.6875, "learning_rate": 6.069195608462422e-05, "loss": 0.1247, "step": 22087 }, { "epoch": 16.18168498168498, "grad_norm": 1.8671875, "learning_rate": 6.0688603114005636e-05, "loss": 0.1219, "step": 22088 }, { "epoch": 16.18241758241758, "grad_norm": 0.96875, "learning_rate": 6.06852500945701e-05, "loss": 0.1648, "step": 22089 }, { "epoch": 16.183150183150182, "grad_norm": 0.60546875, "learning_rate": 6.06818970263337e-05, "loss": 0.124, "step": 22090 }, { "epoch": 16.183882783882783, "grad_norm": 1.484375, "learning_rate": 6.067854390931249e-05, "loss": 0.129, "step": 22091 }, { "epoch": 16.184615384615384, "grad_norm": 0.7109375, "learning_rate": 6.067519074352252e-05, "loss": 0.1187, "step": 22092 }, { "epoch": 16.185347985347985, "grad_norm": 0.80078125, "learning_rate": 6.0671837528979894e-05, "loss": 0.1169, "step": 22093 }, { "epoch": 16.186080586080585, "grad_norm": 0.89453125, "learning_rate": 6.066848426570064e-05, "loss": 0.1165, "step": 22094 }, { "epoch": 16.186813186813186, "grad_norm": 1.21875, "learning_rate": 6.0665130953700835e-05, "loss": 0.17, "step": 22095 }, { "epoch": 16.187545787545787, "grad_norm": 0.8828125, "learning_rate": 6.066177759299655e-05, "loss": 0.1477, "step": 22096 }, { "epoch": 16.188278388278388, "grad_norm": 1.03125, "learning_rate": 6.0658424183603865e-05, "loss": 0.1499, "step": 22097 }, { "epoch": 16.18901098901099, "grad_norm": 0.7890625, "learning_rate": 6.065507072553882e-05, "loss": 0.0855, "step": 22098 }, { "epoch": 16.18974358974359, "grad_norm": 0.76953125, "learning_rate": 6.0651717218817505e-05, "loss": 0.1247, "step": 22099 }, { "epoch": 16.19047619047619, "grad_norm": 1.25, "learning_rate": 6.064836366345598e-05, "loss": 0.1518, "step": 22100 }, { "epoch": 16.19120879120879, "grad_norm": 1.1875, "learning_rate": 6.064501005947032e-05, "loss": 0.1511, "step": 22101 }, { "epoch": 16.19194139194139, "grad_norm": 1.21875, "learning_rate": 6.064165640687657e-05, "loss": 0.1093, "step": 22102 }, { "epoch": 16.192673992673992, "grad_norm": 1.625, "learning_rate": 6.0638302705690833e-05, "loss": 0.1288, "step": 22103 }, { "epoch": 16.193406593406593, "grad_norm": 1.15625, "learning_rate": 6.063494895592916e-05, "loss": 0.1074, "step": 22104 }, { "epoch": 16.194139194139193, "grad_norm": 1.0234375, "learning_rate": 6.0631595157607614e-05, "loss": 0.111, "step": 22105 }, { "epoch": 16.194871794871794, "grad_norm": 1.171875, "learning_rate": 6.0628241310742265e-05, "loss": 0.1224, "step": 22106 }, { "epoch": 16.195604395604395, "grad_norm": 0.9296875, "learning_rate": 6.0624887415349186e-05, "loss": 0.1096, "step": 22107 }, { "epoch": 16.196336996336996, "grad_norm": 1.2265625, "learning_rate": 6.062153347144445e-05, "loss": 0.127, "step": 22108 }, { "epoch": 16.197069597069596, "grad_norm": 1.6796875, "learning_rate": 6.061817947904412e-05, "loss": 0.1381, "step": 22109 }, { "epoch": 16.197802197802197, "grad_norm": 1.5859375, "learning_rate": 6.061482543816428e-05, "loss": 0.1141, "step": 22110 }, { "epoch": 16.198534798534798, "grad_norm": 1.515625, "learning_rate": 6.061147134882098e-05, "loss": 0.1235, "step": 22111 }, { "epoch": 16.1992673992674, "grad_norm": 1.390625, "learning_rate": 6.060811721103031e-05, "loss": 0.1379, "step": 22112 }, { "epoch": 16.2, "grad_norm": 0.96484375, "learning_rate": 6.060476302480833e-05, "loss": 0.108, "step": 22113 }, { "epoch": 16.2007326007326, "grad_norm": 0.83203125, "learning_rate": 6.060140879017111e-05, "loss": 0.1063, "step": 22114 }, { "epoch": 16.2014652014652, "grad_norm": 0.703125, "learning_rate": 6.0598054507134725e-05, "loss": 0.1265, "step": 22115 }, { "epoch": 16.2021978021978, "grad_norm": 0.81640625, "learning_rate": 6.0594700175715255e-05, "loss": 0.1173, "step": 22116 }, { "epoch": 16.202930402930402, "grad_norm": 1.0546875, "learning_rate": 6.0591345795928766e-05, "loss": 0.1323, "step": 22117 }, { "epoch": 16.203663003663003, "grad_norm": 1.1953125, "learning_rate": 6.05879913677913e-05, "loss": 0.1211, "step": 22118 }, { "epoch": 16.204395604395604, "grad_norm": 1.5390625, "learning_rate": 6.058463689131898e-05, "loss": 0.126, "step": 22119 }, { "epoch": 16.205128205128204, "grad_norm": 1.21875, "learning_rate": 6.058128236652786e-05, "loss": 0.1148, "step": 22120 }, { "epoch": 16.205860805860805, "grad_norm": 0.73046875, "learning_rate": 6.057792779343398e-05, "loss": 0.1087, "step": 22121 }, { "epoch": 16.206593406593406, "grad_norm": 1.5234375, "learning_rate": 6.057457317205346e-05, "loss": 0.1079, "step": 22122 }, { "epoch": 16.207326007326007, "grad_norm": 1.0390625, "learning_rate": 6.0571218502402336e-05, "loss": 0.1447, "step": 22123 }, { "epoch": 16.208058608058607, "grad_norm": 1.5859375, "learning_rate": 6.056786378449671e-05, "loss": 0.1298, "step": 22124 }, { "epoch": 16.208791208791208, "grad_norm": 1.75, "learning_rate": 6.0564509018352646e-05, "loss": 0.1183, "step": 22125 }, { "epoch": 16.20952380952381, "grad_norm": 1.4296875, "learning_rate": 6.0561154203986194e-05, "loss": 0.1544, "step": 22126 }, { "epoch": 16.21025641025641, "grad_norm": 0.80859375, "learning_rate": 6.055779934141348e-05, "loss": 0.1114, "step": 22127 }, { "epoch": 16.21098901098901, "grad_norm": 1.8671875, "learning_rate": 6.055444443065053e-05, "loss": 0.1399, "step": 22128 }, { "epoch": 16.21172161172161, "grad_norm": 1.203125, "learning_rate": 6.055108947171344e-05, "loss": 0.102, "step": 22129 }, { "epoch": 16.21245421245421, "grad_norm": 1.2734375, "learning_rate": 6.0547734464618276e-05, "loss": 0.1119, "step": 22130 }, { "epoch": 16.213186813186812, "grad_norm": 1.390625, "learning_rate": 6.0544379409381125e-05, "loss": 0.1132, "step": 22131 }, { "epoch": 16.213919413919413, "grad_norm": 1.3046875, "learning_rate": 6.0541024306018056e-05, "loss": 0.1238, "step": 22132 }, { "epoch": 16.214652014652014, "grad_norm": 1.21875, "learning_rate": 6.053766915454514e-05, "loss": 0.1154, "step": 22133 }, { "epoch": 16.215384615384615, "grad_norm": 1.2265625, "learning_rate": 6.053431395497845e-05, "loss": 0.1528, "step": 22134 }, { "epoch": 16.216117216117215, "grad_norm": 1.2265625, "learning_rate": 6.053095870733407e-05, "loss": 0.0949, "step": 22135 }, { "epoch": 16.216849816849816, "grad_norm": 0.984375, "learning_rate": 6.052760341162809e-05, "loss": 0.1344, "step": 22136 }, { "epoch": 16.217582417582417, "grad_norm": 2.234375, "learning_rate": 6.0524248067876535e-05, "loss": 0.1222, "step": 22137 }, { "epoch": 16.218315018315018, "grad_norm": 1.2265625, "learning_rate": 6.052089267609554e-05, "loss": 0.1348, "step": 22138 }, { "epoch": 16.21904761904762, "grad_norm": 2.0, "learning_rate": 6.051753723630116e-05, "loss": 0.1185, "step": 22139 }, { "epoch": 16.21978021978022, "grad_norm": 1.234375, "learning_rate": 6.051418174850947e-05, "loss": 0.1525, "step": 22140 }, { "epoch": 16.22051282051282, "grad_norm": 1.0546875, "learning_rate": 6.051082621273654e-05, "loss": 0.1274, "step": 22141 }, { "epoch": 16.22124542124542, "grad_norm": 1.0, "learning_rate": 6.050747062899846e-05, "loss": 0.1392, "step": 22142 }, { "epoch": 16.22197802197802, "grad_norm": 1.5, "learning_rate": 6.050411499731131e-05, "loss": 0.1418, "step": 22143 }, { "epoch": 16.222710622710622, "grad_norm": 0.66796875, "learning_rate": 6.050075931769115e-05, "loss": 0.0909, "step": 22144 }, { "epoch": 16.223443223443223, "grad_norm": 0.984375, "learning_rate": 6.049740359015408e-05, "loss": 0.1406, "step": 22145 }, { "epoch": 16.224175824175823, "grad_norm": 0.8984375, "learning_rate": 6.049404781471617e-05, "loss": 0.1447, "step": 22146 }, { "epoch": 16.224908424908424, "grad_norm": 1.1328125, "learning_rate": 6.049069199139348e-05, "loss": 0.1235, "step": 22147 }, { "epoch": 16.225641025641025, "grad_norm": 1.2890625, "learning_rate": 6.04873361202021e-05, "loss": 0.1473, "step": 22148 }, { "epoch": 16.226373626373626, "grad_norm": 1.1171875, "learning_rate": 6.048398020115814e-05, "loss": 0.1516, "step": 22149 }, { "epoch": 16.227106227106226, "grad_norm": 1.2734375, "learning_rate": 6.048062423427763e-05, "loss": 0.1057, "step": 22150 }, { "epoch": 16.227838827838827, "grad_norm": 0.83203125, "learning_rate": 6.047726821957669e-05, "loss": 0.1303, "step": 22151 }, { "epoch": 16.228571428571428, "grad_norm": 0.9765625, "learning_rate": 6.047391215707137e-05, "loss": 0.1355, "step": 22152 }, { "epoch": 16.22930402930403, "grad_norm": 0.9296875, "learning_rate": 6.047055604677777e-05, "loss": 0.1316, "step": 22153 }, { "epoch": 16.23003663003663, "grad_norm": 1.171875, "learning_rate": 6.046719988871194e-05, "loss": 0.1139, "step": 22154 }, { "epoch": 16.23076923076923, "grad_norm": 0.95703125, "learning_rate": 6.0463843682890023e-05, "loss": 0.1427, "step": 22155 }, { "epoch": 16.23150183150183, "grad_norm": 1.6796875, "learning_rate": 6.046048742932804e-05, "loss": 0.1304, "step": 22156 }, { "epoch": 16.23223443223443, "grad_norm": 1.3046875, "learning_rate": 6.045713112804209e-05, "loss": 0.1061, "step": 22157 }, { "epoch": 16.232967032967032, "grad_norm": 1.1953125, "learning_rate": 6.045377477904827e-05, "loss": 0.1402, "step": 22158 }, { "epoch": 16.233699633699633, "grad_norm": 0.80078125, "learning_rate": 6.045041838236264e-05, "loss": 0.1066, "step": 22159 }, { "epoch": 16.234432234432234, "grad_norm": 0.91015625, "learning_rate": 6.044706193800128e-05, "loss": 0.1344, "step": 22160 }, { "epoch": 16.235164835164834, "grad_norm": 0.90234375, "learning_rate": 6.044370544598028e-05, "loss": 0.1193, "step": 22161 }, { "epoch": 16.235897435897435, "grad_norm": 0.65625, "learning_rate": 6.044034890631572e-05, "loss": 0.0973, "step": 22162 }, { "epoch": 16.236630036630036, "grad_norm": 0.9921875, "learning_rate": 6.04369923190237e-05, "loss": 0.1101, "step": 22163 }, { "epoch": 16.237362637362637, "grad_norm": 1.0078125, "learning_rate": 6.0433635684120294e-05, "loss": 0.1004, "step": 22164 }, { "epoch": 16.238095238095237, "grad_norm": 0.8203125, "learning_rate": 6.043027900162156e-05, "loss": 0.1079, "step": 22165 }, { "epoch": 16.238827838827838, "grad_norm": 1.125, "learning_rate": 6.0426922271543593e-05, "loss": 0.1297, "step": 22166 }, { "epoch": 16.23956043956044, "grad_norm": 1.2578125, "learning_rate": 6.042356549390251e-05, "loss": 0.1303, "step": 22167 }, { "epoch": 16.24029304029304, "grad_norm": 1.140625, "learning_rate": 6.042020866871435e-05, "loss": 0.1102, "step": 22168 }, { "epoch": 16.24102564102564, "grad_norm": 1.328125, "learning_rate": 6.041685179599522e-05, "loss": 0.1212, "step": 22169 }, { "epoch": 16.24175824175824, "grad_norm": 0.8828125, "learning_rate": 6.041349487576119e-05, "loss": 0.1075, "step": 22170 }, { "epoch": 16.24249084249084, "grad_norm": 0.97265625, "learning_rate": 6.041013790802835e-05, "loss": 0.1322, "step": 22171 }, { "epoch": 16.243223443223442, "grad_norm": 0.8203125, "learning_rate": 6.04067808928128e-05, "loss": 0.1147, "step": 22172 }, { "epoch": 16.243956043956043, "grad_norm": 0.94921875, "learning_rate": 6.040342383013059e-05, "loss": 0.1109, "step": 22173 }, { "epoch": 16.244688644688644, "grad_norm": 0.91015625, "learning_rate": 6.040006671999784e-05, "loss": 0.1078, "step": 22174 }, { "epoch": 16.245421245421245, "grad_norm": 1.0703125, "learning_rate": 6.039670956243062e-05, "loss": 0.1138, "step": 22175 }, { "epoch": 16.246153846153845, "grad_norm": 1.2578125, "learning_rate": 6.0393352357444996e-05, "loss": 0.1222, "step": 22176 }, { "epoch": 16.246886446886446, "grad_norm": 1.1484375, "learning_rate": 6.0389995105057106e-05, "loss": 0.1322, "step": 22177 }, { "epoch": 16.247619047619047, "grad_norm": 0.99609375, "learning_rate": 6.038663780528299e-05, "loss": 0.1333, "step": 22178 }, { "epoch": 16.248351648351647, "grad_norm": 1.0078125, "learning_rate": 6.0383280458138736e-05, "loss": 0.1208, "step": 22179 }, { "epoch": 16.249084249084248, "grad_norm": 1.171875, "learning_rate": 6.037992306364046e-05, "loss": 0.1388, "step": 22180 }, { "epoch": 16.24981684981685, "grad_norm": 0.953125, "learning_rate": 6.03765656218042e-05, "loss": 0.1284, "step": 22181 }, { "epoch": 16.25054945054945, "grad_norm": 1.25, "learning_rate": 6.037320813264611e-05, "loss": 0.0893, "step": 22182 }, { "epoch": 16.25128205128205, "grad_norm": 1.0625, "learning_rate": 6.036985059618221e-05, "loss": 0.1316, "step": 22183 }, { "epoch": 16.25201465201465, "grad_norm": 0.81640625, "learning_rate": 6.036649301242863e-05, "loss": 0.1142, "step": 22184 }, { "epoch": 16.252747252747252, "grad_norm": 1.359375, "learning_rate": 6.0363135381401434e-05, "loss": 0.1522, "step": 22185 }, { "epoch": 16.253479853479853, "grad_norm": 1.4609375, "learning_rate": 6.035977770311672e-05, "loss": 0.1172, "step": 22186 }, { "epoch": 16.254212454212453, "grad_norm": 0.8828125, "learning_rate": 6.035641997759057e-05, "loss": 0.1216, "step": 22187 }, { "epoch": 16.254945054945054, "grad_norm": 1.0234375, "learning_rate": 6.035306220483908e-05, "loss": 0.1123, "step": 22188 }, { "epoch": 16.255677655677655, "grad_norm": 1.0859375, "learning_rate": 6.0349704384878334e-05, "loss": 0.1436, "step": 22189 }, { "epoch": 16.256410256410255, "grad_norm": 1.09375, "learning_rate": 6.034634651772443e-05, "loss": 0.1117, "step": 22190 }, { "epoch": 16.257142857142856, "grad_norm": 1.0546875, "learning_rate": 6.0342988603393436e-05, "loss": 0.0895, "step": 22191 }, { "epoch": 16.257875457875457, "grad_norm": 1.3515625, "learning_rate": 6.033963064190146e-05, "loss": 0.1033, "step": 22192 }, { "epoch": 16.258608058608058, "grad_norm": 1.46875, "learning_rate": 6.033627263326459e-05, "loss": 0.1641, "step": 22193 }, { "epoch": 16.25934065934066, "grad_norm": 1.3671875, "learning_rate": 6.033291457749889e-05, "loss": 0.1464, "step": 22194 }, { "epoch": 16.26007326007326, "grad_norm": 0.87890625, "learning_rate": 6.0329556474620484e-05, "loss": 0.126, "step": 22195 }, { "epoch": 16.26080586080586, "grad_norm": 0.98828125, "learning_rate": 6.0326198324645434e-05, "loss": 0.1269, "step": 22196 }, { "epoch": 16.26153846153846, "grad_norm": 0.97265625, "learning_rate": 6.0322840127589854e-05, "loss": 0.099, "step": 22197 }, { "epoch": 16.26227106227106, "grad_norm": 1.109375, "learning_rate": 6.0319481883469816e-05, "loss": 0.1504, "step": 22198 }, { "epoch": 16.263003663003662, "grad_norm": 0.65234375, "learning_rate": 6.031612359230142e-05, "loss": 0.1272, "step": 22199 }, { "epoch": 16.263736263736263, "grad_norm": 1.3203125, "learning_rate": 6.0312765254100744e-05, "loss": 0.1194, "step": 22200 }, { "epoch": 16.264468864468864, "grad_norm": 0.7734375, "learning_rate": 6.03094068688839e-05, "loss": 0.1181, "step": 22201 }, { "epoch": 16.265201465201464, "grad_norm": 0.9296875, "learning_rate": 6.030604843666695e-05, "loss": 0.133, "step": 22202 }, { "epoch": 16.265934065934065, "grad_norm": 1.3671875, "learning_rate": 6.030268995746603e-05, "loss": 0.1357, "step": 22203 }, { "epoch": 16.266666666666666, "grad_norm": 1.328125, "learning_rate": 6.029933143129717e-05, "loss": 0.1115, "step": 22204 }, { "epoch": 16.267399267399266, "grad_norm": 1.609375, "learning_rate": 6.029597285817653e-05, "loss": 0.1238, "step": 22205 }, { "epoch": 16.268131868131867, "grad_norm": 0.78515625, "learning_rate": 6.029261423812016e-05, "loss": 0.1311, "step": 22206 }, { "epoch": 16.268864468864468, "grad_norm": 1.2734375, "learning_rate": 6.028925557114414e-05, "loss": 0.1327, "step": 22207 }, { "epoch": 16.26959706959707, "grad_norm": 0.99609375, "learning_rate": 6.028589685726461e-05, "loss": 0.1144, "step": 22208 }, { "epoch": 16.27032967032967, "grad_norm": 1.171875, "learning_rate": 6.028253809649761e-05, "loss": 0.1452, "step": 22209 }, { "epoch": 16.27106227106227, "grad_norm": 0.67578125, "learning_rate": 6.0279179288859275e-05, "loss": 0.1029, "step": 22210 }, { "epoch": 16.27179487179487, "grad_norm": 1.6171875, "learning_rate": 6.027582043436568e-05, "loss": 0.1416, "step": 22211 }, { "epoch": 16.27252747252747, "grad_norm": 0.64453125, "learning_rate": 6.027246153303292e-05, "loss": 0.1008, "step": 22212 }, { "epoch": 16.273260073260072, "grad_norm": 0.7109375, "learning_rate": 6.026910258487708e-05, "loss": 0.1068, "step": 22213 }, { "epoch": 16.273992673992673, "grad_norm": 1.1796875, "learning_rate": 6.026574358991427e-05, "loss": 0.1071, "step": 22214 }, { "epoch": 16.274725274725274, "grad_norm": 1.7109375, "learning_rate": 6.026238454816058e-05, "loss": 0.1481, "step": 22215 }, { "epoch": 16.275457875457874, "grad_norm": 1.59375, "learning_rate": 6.0259025459632104e-05, "loss": 0.1268, "step": 22216 }, { "epoch": 16.276190476190475, "grad_norm": 1.0546875, "learning_rate": 6.025566632434491e-05, "loss": 0.1204, "step": 22217 }, { "epoch": 16.276923076923076, "grad_norm": 0.7890625, "learning_rate": 6.025230714231513e-05, "loss": 0.1025, "step": 22218 }, { "epoch": 16.277655677655677, "grad_norm": 0.71484375, "learning_rate": 6.0248947913558875e-05, "loss": 0.1185, "step": 22219 }, { "epoch": 16.278388278388277, "grad_norm": 0.60546875, "learning_rate": 6.0245588638092164e-05, "loss": 0.1148, "step": 22220 }, { "epoch": 16.279120879120878, "grad_norm": 1.53125, "learning_rate": 6.024222931593116e-05, "loss": 0.1314, "step": 22221 }, { "epoch": 16.27985347985348, "grad_norm": 1.0390625, "learning_rate": 6.0238869947091925e-05, "loss": 0.129, "step": 22222 }, { "epoch": 16.28058608058608, "grad_norm": 1.1171875, "learning_rate": 6.023551053159059e-05, "loss": 0.1654, "step": 22223 }, { "epoch": 16.28131868131868, "grad_norm": 0.6484375, "learning_rate": 6.023215106944322e-05, "loss": 0.1096, "step": 22224 }, { "epoch": 16.28205128205128, "grad_norm": 0.85546875, "learning_rate": 6.022879156066591e-05, "loss": 0.1458, "step": 22225 }, { "epoch": 16.282783882783882, "grad_norm": 1.03125, "learning_rate": 6.022543200527476e-05, "loss": 0.1194, "step": 22226 }, { "epoch": 16.283516483516483, "grad_norm": 1.0, "learning_rate": 6.022207240328588e-05, "loss": 0.1425, "step": 22227 }, { "epoch": 16.284249084249083, "grad_norm": 0.640625, "learning_rate": 6.021871275471537e-05, "loss": 0.1326, "step": 22228 }, { "epoch": 16.284981684981684, "grad_norm": 1.5390625, "learning_rate": 6.0215353059579306e-05, "loss": 0.1243, "step": 22229 }, { "epoch": 16.285714285714285, "grad_norm": 1.640625, "learning_rate": 6.02119933178938e-05, "loss": 0.1293, "step": 22230 }, { "epoch": 16.286446886446885, "grad_norm": 1.15625, "learning_rate": 6.020863352967494e-05, "loss": 0.1231, "step": 22231 }, { "epoch": 16.287179487179486, "grad_norm": 1.359375, "learning_rate": 6.020527369493885e-05, "loss": 0.126, "step": 22232 }, { "epoch": 16.287912087912087, "grad_norm": 0.6328125, "learning_rate": 6.0201913813701583e-05, "loss": 0.1115, "step": 22233 }, { "epoch": 16.288644688644688, "grad_norm": 1.1640625, "learning_rate": 6.019855388597927e-05, "loss": 0.1087, "step": 22234 }, { "epoch": 16.28937728937729, "grad_norm": 0.79296875, "learning_rate": 6.0195193911788e-05, "loss": 0.1273, "step": 22235 }, { "epoch": 16.29010989010989, "grad_norm": 1.125, "learning_rate": 6.019183389114388e-05, "loss": 0.1208, "step": 22236 }, { "epoch": 16.29084249084249, "grad_norm": 0.8203125, "learning_rate": 6.018847382406301e-05, "loss": 0.1016, "step": 22237 }, { "epoch": 16.29157509157509, "grad_norm": 1.7109375, "learning_rate": 6.0185113710561455e-05, "loss": 0.1499, "step": 22238 }, { "epoch": 16.29230769230769, "grad_norm": 1.25, "learning_rate": 6.018175355065535e-05, "loss": 0.1073, "step": 22239 }, { "epoch": 16.293040293040292, "grad_norm": 1.3515625, "learning_rate": 6.0178393344360784e-05, "loss": 0.1455, "step": 22240 }, { "epoch": 16.293772893772893, "grad_norm": 1.0859375, "learning_rate": 6.017503309169387e-05, "loss": 0.1046, "step": 22241 }, { "epoch": 16.294505494505493, "grad_norm": 1.3984375, "learning_rate": 6.01716727926707e-05, "loss": 0.1304, "step": 22242 }, { "epoch": 16.295238095238094, "grad_norm": 0.62890625, "learning_rate": 6.016831244730736e-05, "loss": 0.106, "step": 22243 }, { "epoch": 16.295970695970695, "grad_norm": 0.96484375, "learning_rate": 6.0164952055619964e-05, "loss": 0.1351, "step": 22244 }, { "epoch": 16.296703296703296, "grad_norm": 1.2109375, "learning_rate": 6.0161591617624616e-05, "loss": 0.1432, "step": 22245 }, { "epoch": 16.297435897435896, "grad_norm": 1.0703125, "learning_rate": 6.015823113333739e-05, "loss": 0.1159, "step": 22246 }, { "epoch": 16.298168498168497, "grad_norm": 0.97265625, "learning_rate": 6.015487060277443e-05, "loss": 0.1217, "step": 22247 }, { "epoch": 16.298901098901098, "grad_norm": 0.77734375, "learning_rate": 6.0151510025951804e-05, "loss": 0.1271, "step": 22248 }, { "epoch": 16.2996336996337, "grad_norm": 1.125, "learning_rate": 6.0148149402885626e-05, "loss": 0.1222, "step": 22249 }, { "epoch": 16.3003663003663, "grad_norm": 1.2890625, "learning_rate": 6.014478873359201e-05, "loss": 0.1266, "step": 22250 }, { "epoch": 16.3010989010989, "grad_norm": 1.15625, "learning_rate": 6.014142801808702e-05, "loss": 0.1426, "step": 22251 }, { "epoch": 16.3018315018315, "grad_norm": 0.94921875, "learning_rate": 6.01380672563868e-05, "loss": 0.1294, "step": 22252 }, { "epoch": 16.3025641025641, "grad_norm": 1.671875, "learning_rate": 6.013470644850743e-05, "loss": 0.1858, "step": 22253 }, { "epoch": 16.303296703296702, "grad_norm": 1.40625, "learning_rate": 6.0131345594465017e-05, "loss": 0.1354, "step": 22254 }, { "epoch": 16.304029304029303, "grad_norm": 1.0, "learning_rate": 6.0127984694275685e-05, "loss": 0.1254, "step": 22255 }, { "epoch": 16.304761904761904, "grad_norm": 1.1015625, "learning_rate": 6.012462374795549e-05, "loss": 0.1378, "step": 22256 }, { "epoch": 16.305494505494504, "grad_norm": 1.78125, "learning_rate": 6.012126275552058e-05, "loss": 0.1334, "step": 22257 }, { "epoch": 16.306227106227105, "grad_norm": 0.7265625, "learning_rate": 6.011790171698705e-05, "loss": 0.1175, "step": 22258 }, { "epoch": 16.306959706959706, "grad_norm": 1.078125, "learning_rate": 6.011454063237098e-05, "loss": 0.1155, "step": 22259 }, { "epoch": 16.307692307692307, "grad_norm": 1.203125, "learning_rate": 6.01111795016885e-05, "loss": 0.1198, "step": 22260 }, { "epoch": 16.308424908424907, "grad_norm": 1.1015625, "learning_rate": 6.0107818324955676e-05, "loss": 0.085, "step": 22261 }, { "epoch": 16.309157509157508, "grad_norm": 1.4609375, "learning_rate": 6.010445710218867e-05, "loss": 0.1082, "step": 22262 }, { "epoch": 16.30989010989011, "grad_norm": 1.0, "learning_rate": 6.0101095833403556e-05, "loss": 0.1001, "step": 22263 }, { "epoch": 16.31062271062271, "grad_norm": 0.74609375, "learning_rate": 6.009773451861643e-05, "loss": 0.1151, "step": 22264 }, { "epoch": 16.31135531135531, "grad_norm": 0.9765625, "learning_rate": 6.009437315784341e-05, "loss": 0.1354, "step": 22265 }, { "epoch": 16.31208791208791, "grad_norm": 1.328125, "learning_rate": 6.009101175110059e-05, "loss": 0.1311, "step": 22266 }, { "epoch": 16.31282051282051, "grad_norm": 1.40625, "learning_rate": 6.0087650298404084e-05, "loss": 0.1411, "step": 22267 }, { "epoch": 16.313553113553112, "grad_norm": 0.83984375, "learning_rate": 6.008428879977003e-05, "loss": 0.1118, "step": 22268 }, { "epoch": 16.314285714285713, "grad_norm": 0.7578125, "learning_rate": 6.008092725521446e-05, "loss": 0.1033, "step": 22269 }, { "epoch": 16.315018315018314, "grad_norm": 0.76171875, "learning_rate": 6.0077565664753544e-05, "loss": 0.1198, "step": 22270 }, { "epoch": 16.315750915750915, "grad_norm": 1.3515625, "learning_rate": 6.007420402840337e-05, "loss": 0.1388, "step": 22271 }, { "epoch": 16.316483516483515, "grad_norm": 1.15625, "learning_rate": 6.0070842346180023e-05, "loss": 0.1254, "step": 22272 }, { "epoch": 16.317216117216116, "grad_norm": 1.0078125, "learning_rate": 6.0067480618099646e-05, "loss": 0.1536, "step": 22273 }, { "epoch": 16.317948717948717, "grad_norm": 1.3828125, "learning_rate": 6.006411884417832e-05, "loss": 0.1383, "step": 22274 }, { "epoch": 16.318681318681318, "grad_norm": 0.9296875, "learning_rate": 6.0060757024432166e-05, "loss": 0.1265, "step": 22275 }, { "epoch": 16.31941391941392, "grad_norm": 1.2421875, "learning_rate": 6.0057395158877284e-05, "loss": 0.0999, "step": 22276 }, { "epoch": 16.32014652014652, "grad_norm": 1.046875, "learning_rate": 6.005403324752978e-05, "loss": 0.113, "step": 22277 }, { "epoch": 16.32087912087912, "grad_norm": 1.3984375, "learning_rate": 6.0050671290405766e-05, "loss": 0.1225, "step": 22278 }, { "epoch": 16.32161172161172, "grad_norm": 1.7421875, "learning_rate": 6.0047309287521344e-05, "loss": 0.1395, "step": 22279 }, { "epoch": 16.32234432234432, "grad_norm": 1.3515625, "learning_rate": 6.0043947238892664e-05, "loss": 0.1236, "step": 22280 }, { "epoch": 16.323076923076922, "grad_norm": 0.640625, "learning_rate": 6.004058514453578e-05, "loss": 0.1239, "step": 22281 }, { "epoch": 16.323809523809523, "grad_norm": 0.58203125, "learning_rate": 6.00372230044668e-05, "loss": 0.0864, "step": 22282 }, { "epoch": 16.324542124542123, "grad_norm": 1.8203125, "learning_rate": 6.003386081870188e-05, "loss": 0.1585, "step": 22283 }, { "epoch": 16.325274725274724, "grad_norm": 1.8203125, "learning_rate": 6.00304985872571e-05, "loss": 0.1399, "step": 22284 }, { "epoch": 16.326007326007325, "grad_norm": 1.9765625, "learning_rate": 6.002713631014856e-05, "loss": 0.1562, "step": 22285 }, { "epoch": 16.326739926739926, "grad_norm": 0.96484375, "learning_rate": 6.00237739873924e-05, "loss": 0.1152, "step": 22286 }, { "epoch": 16.327472527472526, "grad_norm": 1.890625, "learning_rate": 6.0020411619004694e-05, "loss": 0.1509, "step": 22287 }, { "epoch": 16.328205128205127, "grad_norm": 1.2734375, "learning_rate": 6.0017049205001576e-05, "loss": 0.1512, "step": 22288 }, { "epoch": 16.328937728937728, "grad_norm": 1.1484375, "learning_rate": 6.0013686745399144e-05, "loss": 0.1092, "step": 22289 }, { "epoch": 16.32967032967033, "grad_norm": 2.1875, "learning_rate": 6.001032424021353e-05, "loss": 0.1474, "step": 22290 }, { "epoch": 16.33040293040293, "grad_norm": 1.3359375, "learning_rate": 6.0006961689460814e-05, "loss": 0.11, "step": 22291 }, { "epoch": 16.33113553113553, "grad_norm": 1.171875, "learning_rate": 6.000359909315714e-05, "loss": 0.1213, "step": 22292 }, { "epoch": 16.33186813186813, "grad_norm": 0.69140625, "learning_rate": 6.00002364513186e-05, "loss": 0.0924, "step": 22293 }, { "epoch": 16.33260073260073, "grad_norm": 1.6640625, "learning_rate": 5.99968737639613e-05, "loss": 0.123, "step": 22294 }, { "epoch": 16.333333333333332, "grad_norm": 0.625, "learning_rate": 5.999351103110137e-05, "loss": 0.1198, "step": 22295 }, { "epoch": 16.334065934065933, "grad_norm": 0.99609375, "learning_rate": 5.999014825275491e-05, "loss": 0.1223, "step": 22296 }, { "epoch": 16.334798534798534, "grad_norm": 1.203125, "learning_rate": 5.998678542893804e-05, "loss": 0.1174, "step": 22297 }, { "epoch": 16.335531135531134, "grad_norm": 0.63671875, "learning_rate": 5.9983422559666855e-05, "loss": 0.108, "step": 22298 }, { "epoch": 16.336263736263735, "grad_norm": 0.640625, "learning_rate": 5.998005964495749e-05, "loss": 0.1053, "step": 22299 }, { "epoch": 16.336996336996336, "grad_norm": 1.3515625, "learning_rate": 5.997669668482603e-05, "loss": 0.1042, "step": 22300 }, { "epoch": 16.337728937728937, "grad_norm": 0.84375, "learning_rate": 5.9973333679288624e-05, "loss": 0.1033, "step": 22301 }, { "epoch": 16.338461538461537, "grad_norm": 0.9375, "learning_rate": 5.996997062836135e-05, "loss": 0.1406, "step": 22302 }, { "epoch": 16.339194139194138, "grad_norm": 1.6953125, "learning_rate": 5.9966607532060346e-05, "loss": 0.1381, "step": 22303 }, { "epoch": 16.33992673992674, "grad_norm": 1.21875, "learning_rate": 5.9963244390401716e-05, "loss": 0.1408, "step": 22304 }, { "epoch": 16.34065934065934, "grad_norm": 0.76171875, "learning_rate": 5.995988120340158e-05, "loss": 0.1098, "step": 22305 }, { "epoch": 16.34139194139194, "grad_norm": 0.73828125, "learning_rate": 5.9956517971076044e-05, "loss": 0.1283, "step": 22306 }, { "epoch": 16.34212454212454, "grad_norm": 0.80859375, "learning_rate": 5.995315469344123e-05, "loss": 0.1153, "step": 22307 }, { "epoch": 16.34285714285714, "grad_norm": 0.84375, "learning_rate": 5.994979137051323e-05, "loss": 0.1235, "step": 22308 }, { "epoch": 16.343589743589742, "grad_norm": 1.2890625, "learning_rate": 5.994642800230821e-05, "loss": 0.0997, "step": 22309 }, { "epoch": 16.344322344322343, "grad_norm": 1.3125, "learning_rate": 5.9943064588842235e-05, "loss": 0.1101, "step": 22310 }, { "epoch": 16.345054945054944, "grad_norm": 0.81640625, "learning_rate": 5.9939701130131434e-05, "loss": 0.1296, "step": 22311 }, { "epoch": 16.345787545787545, "grad_norm": 1.0390625, "learning_rate": 5.993633762619194e-05, "loss": 0.1586, "step": 22312 }, { "epoch": 16.346520146520145, "grad_norm": 1.125, "learning_rate": 5.993297407703983e-05, "loss": 0.1216, "step": 22313 }, { "epoch": 16.347252747252746, "grad_norm": 1.15625, "learning_rate": 5.992961048269127e-05, "loss": 0.1258, "step": 22314 }, { "epoch": 16.347985347985347, "grad_norm": 0.9609375, "learning_rate": 5.992624684316233e-05, "loss": 0.1043, "step": 22315 }, { "epoch": 16.348717948717947, "grad_norm": 1.46875, "learning_rate": 5.992288315846917e-05, "loss": 0.12, "step": 22316 }, { "epoch": 16.349450549450548, "grad_norm": 1.3203125, "learning_rate": 5.9919519428627856e-05, "loss": 0.1187, "step": 22317 }, { "epoch": 16.35018315018315, "grad_norm": 0.890625, "learning_rate": 5.991615565365454e-05, "loss": 0.1301, "step": 22318 }, { "epoch": 16.35091575091575, "grad_norm": 1.296875, "learning_rate": 5.991279183356534e-05, "loss": 0.1135, "step": 22319 }, { "epoch": 16.35164835164835, "grad_norm": 0.859375, "learning_rate": 5.990942796837637e-05, "loss": 0.1487, "step": 22320 }, { "epoch": 16.35238095238095, "grad_norm": 1.125, "learning_rate": 5.990606405810373e-05, "loss": 0.1213, "step": 22321 }, { "epoch": 16.353113553113552, "grad_norm": 0.76953125, "learning_rate": 5.990270010276355e-05, "loss": 0.1055, "step": 22322 }, { "epoch": 16.353846153846153, "grad_norm": 0.8203125, "learning_rate": 5.9899336102371964e-05, "loss": 0.1179, "step": 22323 }, { "epoch": 16.354578754578753, "grad_norm": 1.1796875, "learning_rate": 5.989597205694505e-05, "loss": 0.1277, "step": 22324 }, { "epoch": 16.355311355311354, "grad_norm": 1.28125, "learning_rate": 5.989260796649896e-05, "loss": 0.1756, "step": 22325 }, { "epoch": 16.356043956043955, "grad_norm": 1.1875, "learning_rate": 5.98892438310498e-05, "loss": 0.1248, "step": 22326 }, { "epoch": 16.356776556776556, "grad_norm": 1.2578125, "learning_rate": 5.9885879650613696e-05, "loss": 0.1158, "step": 22327 }, { "epoch": 16.357509157509156, "grad_norm": 0.921875, "learning_rate": 5.9882515425206754e-05, "loss": 0.1428, "step": 22328 }, { "epoch": 16.358241758241757, "grad_norm": 0.66796875, "learning_rate": 5.987915115484509e-05, "loss": 0.1201, "step": 22329 }, { "epoch": 16.358974358974358, "grad_norm": 0.91015625, "learning_rate": 5.9875786839544846e-05, "loss": 0.1258, "step": 22330 }, { "epoch": 16.35970695970696, "grad_norm": 1.4296875, "learning_rate": 5.9872422479322135e-05, "loss": 0.1321, "step": 22331 }, { "epoch": 16.36043956043956, "grad_norm": 1.21875, "learning_rate": 5.9869058074193063e-05, "loss": 0.1386, "step": 22332 }, { "epoch": 16.36117216117216, "grad_norm": 1.28125, "learning_rate": 5.986569362417377e-05, "loss": 0.1184, "step": 22333 }, { "epoch": 16.36190476190476, "grad_norm": 1.65625, "learning_rate": 5.986232912928036e-05, "loss": 0.1379, "step": 22334 }, { "epoch": 16.36263736263736, "grad_norm": 1.109375, "learning_rate": 5.9858964589528953e-05, "loss": 0.1321, "step": 22335 }, { "epoch": 16.363369963369962, "grad_norm": 1.1328125, "learning_rate": 5.9855600004935685e-05, "loss": 0.1134, "step": 22336 }, { "epoch": 16.364102564102563, "grad_norm": 1.03125, "learning_rate": 5.985223537551664e-05, "loss": 0.126, "step": 22337 }, { "epoch": 16.364835164835164, "grad_norm": 0.8125, "learning_rate": 5.9848870701287986e-05, "loss": 0.1151, "step": 22338 }, { "epoch": 16.365567765567764, "grad_norm": 1.5078125, "learning_rate": 5.9845505982265805e-05, "loss": 0.1296, "step": 22339 }, { "epoch": 16.366300366300365, "grad_norm": 0.9765625, "learning_rate": 5.984214121846625e-05, "loss": 0.1296, "step": 22340 }, { "epoch": 16.367032967032966, "grad_norm": 1.0546875, "learning_rate": 5.983877640990544e-05, "loss": 0.1103, "step": 22341 }, { "epoch": 16.367765567765566, "grad_norm": 0.921875, "learning_rate": 5.9835411556599475e-05, "loss": 0.1357, "step": 22342 }, { "epoch": 16.368498168498167, "grad_norm": 1.0546875, "learning_rate": 5.983204665856449e-05, "loss": 0.151, "step": 22343 }, { "epoch": 16.369230769230768, "grad_norm": 1.484375, "learning_rate": 5.982868171581659e-05, "loss": 0.1413, "step": 22344 }, { "epoch": 16.36996336996337, "grad_norm": 1.4765625, "learning_rate": 5.982531672837194e-05, "loss": 0.1351, "step": 22345 }, { "epoch": 16.37069597069597, "grad_norm": 0.80078125, "learning_rate": 5.9821951696246626e-05, "loss": 0.0998, "step": 22346 }, { "epoch": 16.37142857142857, "grad_norm": 1.25, "learning_rate": 5.981858661945678e-05, "loss": 0.1445, "step": 22347 }, { "epoch": 16.37216117216117, "grad_norm": 1.625, "learning_rate": 5.981522149801854e-05, "loss": 0.1198, "step": 22348 }, { "epoch": 16.37289377289377, "grad_norm": 1.4609375, "learning_rate": 5.981185633194799e-05, "loss": 0.1013, "step": 22349 }, { "epoch": 16.373626373626372, "grad_norm": 2.328125, "learning_rate": 5.980849112126129e-05, "loss": 0.1573, "step": 22350 }, { "epoch": 16.374358974358973, "grad_norm": 0.55078125, "learning_rate": 5.9805125865974565e-05, "loss": 0.0962, "step": 22351 }, { "epoch": 16.375091575091574, "grad_norm": 1.328125, "learning_rate": 5.980176056610392e-05, "loss": 0.1215, "step": 22352 }, { "epoch": 16.375824175824174, "grad_norm": 1.2109375, "learning_rate": 5.979839522166548e-05, "loss": 0.1142, "step": 22353 }, { "epoch": 16.376556776556775, "grad_norm": 1.09375, "learning_rate": 5.979502983267538e-05, "loss": 0.1058, "step": 22354 }, { "epoch": 16.377289377289376, "grad_norm": 1.3515625, "learning_rate": 5.979166439914973e-05, "loss": 0.1213, "step": 22355 }, { "epoch": 16.378021978021977, "grad_norm": 1.3125, "learning_rate": 5.978829892110469e-05, "loss": 0.1352, "step": 22356 }, { "epoch": 16.378754578754577, "grad_norm": 0.91796875, "learning_rate": 5.978493339855633e-05, "loss": 0.1237, "step": 22357 }, { "epoch": 16.379487179487178, "grad_norm": 1.0, "learning_rate": 5.978156783152083e-05, "loss": 0.1023, "step": 22358 }, { "epoch": 16.38021978021978, "grad_norm": 0.63671875, "learning_rate": 5.97782022200143e-05, "loss": 0.1159, "step": 22359 }, { "epoch": 16.38095238095238, "grad_norm": 0.90234375, "learning_rate": 5.977483656405283e-05, "loss": 0.1229, "step": 22360 }, { "epoch": 16.38168498168498, "grad_norm": 0.77734375, "learning_rate": 5.97714708636526e-05, "loss": 0.1296, "step": 22361 }, { "epoch": 16.38241758241758, "grad_norm": 1.28125, "learning_rate": 5.9768105118829685e-05, "loss": 0.1253, "step": 22362 }, { "epoch": 16.383150183150182, "grad_norm": 1.4765625, "learning_rate": 5.976473932960025e-05, "loss": 0.1408, "step": 22363 }, { "epoch": 16.383882783882783, "grad_norm": 0.97265625, "learning_rate": 5.976137349598041e-05, "loss": 0.0958, "step": 22364 }, { "epoch": 16.384615384615383, "grad_norm": 0.8125, "learning_rate": 5.975800761798628e-05, "loss": 0.114, "step": 22365 }, { "epoch": 16.385347985347984, "grad_norm": 1.3125, "learning_rate": 5.9754641695634005e-05, "loss": 0.1414, "step": 22366 }, { "epoch": 16.386080586080585, "grad_norm": 1.65625, "learning_rate": 5.9751275728939685e-05, "loss": 0.1402, "step": 22367 }, { "epoch": 16.386813186813185, "grad_norm": 0.96875, "learning_rate": 5.9747909717919473e-05, "loss": 0.1183, "step": 22368 }, { "epoch": 16.387545787545786, "grad_norm": 1.0390625, "learning_rate": 5.9744543662589505e-05, "loss": 0.1179, "step": 22369 }, { "epoch": 16.388278388278387, "grad_norm": 1.390625, "learning_rate": 5.9741177562965866e-05, "loss": 0.113, "step": 22370 }, { "epoch": 16.389010989010988, "grad_norm": 0.765625, "learning_rate": 5.9737811419064746e-05, "loss": 0.1246, "step": 22371 }, { "epoch": 16.38974358974359, "grad_norm": 1.5234375, "learning_rate": 5.973444523090222e-05, "loss": 0.1168, "step": 22372 }, { "epoch": 16.39047619047619, "grad_norm": 1.421875, "learning_rate": 5.973107899849444e-05, "loss": 0.141, "step": 22373 }, { "epoch": 16.39120879120879, "grad_norm": 1.109375, "learning_rate": 5.972771272185753e-05, "loss": 0.1565, "step": 22374 }, { "epoch": 16.39194139194139, "grad_norm": 0.66015625, "learning_rate": 5.972434640100762e-05, "loss": 0.1262, "step": 22375 }, { "epoch": 16.39267399267399, "grad_norm": 1.46875, "learning_rate": 5.9720980035960836e-05, "loss": 0.1184, "step": 22376 }, { "epoch": 16.393406593406592, "grad_norm": 1.0234375, "learning_rate": 5.971761362673331e-05, "loss": 0.1274, "step": 22377 }, { "epoch": 16.394139194139193, "grad_norm": 1.2421875, "learning_rate": 5.9714247173341166e-05, "loss": 0.1467, "step": 22378 }, { "epoch": 16.394871794871793, "grad_norm": 1.9296875, "learning_rate": 5.971088067580054e-05, "loss": 0.1293, "step": 22379 }, { "epoch": 16.395604395604394, "grad_norm": 0.89453125, "learning_rate": 5.970751413412756e-05, "loss": 0.1244, "step": 22380 }, { "epoch": 16.396336996336995, "grad_norm": 1.3046875, "learning_rate": 5.9704147548338364e-05, "loss": 0.1325, "step": 22381 }, { "epoch": 16.397069597069596, "grad_norm": 1.1328125, "learning_rate": 5.9700780918449075e-05, "loss": 0.1134, "step": 22382 }, { "epoch": 16.397802197802196, "grad_norm": 0.671875, "learning_rate": 5.969741424447583e-05, "loss": 0.1184, "step": 22383 }, { "epoch": 16.398534798534797, "grad_norm": 0.71875, "learning_rate": 5.969404752643474e-05, "loss": 0.1129, "step": 22384 }, { "epoch": 16.399267399267398, "grad_norm": 1.0859375, "learning_rate": 5.9690680764341945e-05, "loss": 0.1092, "step": 22385 }, { "epoch": 16.4, "grad_norm": 1.6328125, "learning_rate": 5.96873139582136e-05, "loss": 0.1444, "step": 22386 }, { "epoch": 16.4007326007326, "grad_norm": 0.890625, "learning_rate": 5.968394710806582e-05, "loss": 0.1195, "step": 22387 }, { "epoch": 16.4014652014652, "grad_norm": 0.462890625, "learning_rate": 5.968058021391472e-05, "loss": 0.1052, "step": 22388 }, { "epoch": 16.4021978021978, "grad_norm": 1.0703125, "learning_rate": 5.9677213275776446e-05, "loss": 0.1322, "step": 22389 }, { "epoch": 16.4029304029304, "grad_norm": 0.8515625, "learning_rate": 5.967384629366713e-05, "loss": 0.1126, "step": 22390 }, { "epoch": 16.403663003663002, "grad_norm": 0.88671875, "learning_rate": 5.9670479267602916e-05, "loss": 0.1424, "step": 22391 }, { "epoch": 16.404395604395603, "grad_norm": 0.9453125, "learning_rate": 5.966711219759993e-05, "loss": 0.1089, "step": 22392 }, { "epoch": 16.405128205128204, "grad_norm": 0.875, "learning_rate": 5.9663745083674284e-05, "loss": 0.1615, "step": 22393 }, { "epoch": 16.405860805860804, "grad_norm": 0.96484375, "learning_rate": 5.9660377925842125e-05, "loss": 0.1085, "step": 22394 }, { "epoch": 16.406593406593405, "grad_norm": 0.90234375, "learning_rate": 5.96570107241196e-05, "loss": 0.1485, "step": 22395 }, { "epoch": 16.407326007326006, "grad_norm": 0.94921875, "learning_rate": 5.965364347852284e-05, "loss": 0.1242, "step": 22396 }, { "epoch": 16.408058608058607, "grad_norm": 0.828125, "learning_rate": 5.965027618906795e-05, "loss": 0.1354, "step": 22397 }, { "epoch": 16.408791208791207, "grad_norm": 1.03125, "learning_rate": 5.964690885577109e-05, "loss": 0.1587, "step": 22398 }, { "epoch": 16.409523809523808, "grad_norm": 0.9453125, "learning_rate": 5.964354147864838e-05, "loss": 0.1148, "step": 22399 }, { "epoch": 16.41025641025641, "grad_norm": 0.85546875, "learning_rate": 5.964017405771596e-05, "loss": 0.1459, "step": 22400 }, { "epoch": 16.41098901098901, "grad_norm": 1.40625, "learning_rate": 5.963680659298998e-05, "loss": 0.1441, "step": 22401 }, { "epoch": 16.41172161172161, "grad_norm": 1.15625, "learning_rate": 5.963343908448655e-05, "loss": 0.1071, "step": 22402 }, { "epoch": 16.41245421245421, "grad_norm": 1.296875, "learning_rate": 5.963007153222182e-05, "loss": 0.1261, "step": 22403 }, { "epoch": 16.41318681318681, "grad_norm": 1.5703125, "learning_rate": 5.9626703936211914e-05, "loss": 0.1498, "step": 22404 }, { "epoch": 16.413919413919412, "grad_norm": 1.296875, "learning_rate": 5.9623336296472984e-05, "loss": 0.176, "step": 22405 }, { "epoch": 16.414652014652013, "grad_norm": 1.40625, "learning_rate": 5.9619968613021136e-05, "loss": 0.1282, "step": 22406 }, { "epoch": 16.415384615384614, "grad_norm": 1.78125, "learning_rate": 5.961660088587253e-05, "loss": 0.1464, "step": 22407 }, { "epoch": 16.416117216117215, "grad_norm": 1.7109375, "learning_rate": 5.961323311504331e-05, "loss": 0.1355, "step": 22408 }, { "epoch": 16.416849816849815, "grad_norm": 1.375, "learning_rate": 5.9609865300549595e-05, "loss": 0.1386, "step": 22409 }, { "epoch": 16.417582417582416, "grad_norm": 1.7890625, "learning_rate": 5.9606497442407515e-05, "loss": 0.1171, "step": 22410 }, { "epoch": 16.418315018315017, "grad_norm": 0.90625, "learning_rate": 5.960312954063323e-05, "loss": 0.0881, "step": 22411 }, { "epoch": 16.419047619047618, "grad_norm": 1.1953125, "learning_rate": 5.9599761595242854e-05, "loss": 0.1554, "step": 22412 }, { "epoch": 16.41978021978022, "grad_norm": 0.76171875, "learning_rate": 5.9596393606252534e-05, "loss": 0.1258, "step": 22413 }, { "epoch": 16.42051282051282, "grad_norm": 1.0078125, "learning_rate": 5.959302557367842e-05, "loss": 0.0983, "step": 22414 }, { "epoch": 16.42124542124542, "grad_norm": 1.203125, "learning_rate": 5.958965749753661e-05, "loss": 0.1291, "step": 22415 }, { "epoch": 16.42197802197802, "grad_norm": 1.0625, "learning_rate": 5.9586289377843284e-05, "loss": 0.1081, "step": 22416 }, { "epoch": 16.42271062271062, "grad_norm": 0.94140625, "learning_rate": 5.958292121461455e-05, "loss": 0.1423, "step": 22417 }, { "epoch": 16.423443223443222, "grad_norm": 1.15625, "learning_rate": 5.957955300786657e-05, "loss": 0.1237, "step": 22418 }, { "epoch": 16.424175824175823, "grad_norm": 1.1015625, "learning_rate": 5.957618475761547e-05, "loss": 0.1182, "step": 22419 }, { "epoch": 16.424908424908423, "grad_norm": 0.96484375, "learning_rate": 5.957281646387738e-05, "loss": 0.1081, "step": 22420 }, { "epoch": 16.425641025641024, "grad_norm": 1.1015625, "learning_rate": 5.956944812666845e-05, "loss": 0.116, "step": 22421 }, { "epoch": 16.426373626373625, "grad_norm": 0.8125, "learning_rate": 5.9566079746004835e-05, "loss": 0.1292, "step": 22422 }, { "epoch": 16.427106227106226, "grad_norm": 1.09375, "learning_rate": 5.956271132190264e-05, "loss": 0.1334, "step": 22423 }, { "epoch": 16.427838827838826, "grad_norm": 0.66015625, "learning_rate": 5.955934285437802e-05, "loss": 0.1449, "step": 22424 }, { "epoch": 16.428571428571427, "grad_norm": 0.68359375, "learning_rate": 5.9555974343447106e-05, "loss": 0.1124, "step": 22425 }, { "epoch": 16.429304029304028, "grad_norm": 1.4609375, "learning_rate": 5.955260578912607e-05, "loss": 0.1246, "step": 22426 }, { "epoch": 16.43003663003663, "grad_norm": 0.703125, "learning_rate": 5.954923719143101e-05, "loss": 0.0964, "step": 22427 }, { "epoch": 16.43076923076923, "grad_norm": 0.78125, "learning_rate": 5.954586855037809e-05, "loss": 0.1091, "step": 22428 }, { "epoch": 16.43150183150183, "grad_norm": 0.82421875, "learning_rate": 5.954249986598344e-05, "loss": 0.1106, "step": 22429 }, { "epoch": 16.43223443223443, "grad_norm": 1.765625, "learning_rate": 5.9539131138263206e-05, "loss": 0.1242, "step": 22430 }, { "epoch": 16.43296703296703, "grad_norm": 0.953125, "learning_rate": 5.9535762367233536e-05, "loss": 0.1256, "step": 22431 }, { "epoch": 16.433699633699632, "grad_norm": 1.296875, "learning_rate": 5.9532393552910535e-05, "loss": 0.1501, "step": 22432 }, { "epoch": 16.434432234432233, "grad_norm": 0.8984375, "learning_rate": 5.9529024695310384e-05, "loss": 0.1176, "step": 22433 }, { "epoch": 16.435164835164834, "grad_norm": 1.40625, "learning_rate": 5.952565579444923e-05, "loss": 0.1231, "step": 22434 }, { "epoch": 16.435897435897434, "grad_norm": 1.296875, "learning_rate": 5.95222868503432e-05, "loss": 0.1705, "step": 22435 }, { "epoch": 16.436630036630035, "grad_norm": 1.6328125, "learning_rate": 5.9518917863008406e-05, "loss": 0.1531, "step": 22436 }, { "epoch": 16.437362637362636, "grad_norm": 1.046875, "learning_rate": 5.9515548832461034e-05, "loss": 0.1172, "step": 22437 }, { "epoch": 16.438095238095237, "grad_norm": 1.8984375, "learning_rate": 5.951217975871719e-05, "loss": 0.0932, "step": 22438 }, { "epoch": 16.438827838827837, "grad_norm": 1.2890625, "learning_rate": 5.950881064179306e-05, "loss": 0.1409, "step": 22439 }, { "epoch": 16.439560439560438, "grad_norm": 1.0859375, "learning_rate": 5.950544148170474e-05, "loss": 0.115, "step": 22440 }, { "epoch": 16.44029304029304, "grad_norm": 1.109375, "learning_rate": 5.950207227846839e-05, "loss": 0.133, "step": 22441 }, { "epoch": 16.44102564102564, "grad_norm": 0.84765625, "learning_rate": 5.9498703032100165e-05, "loss": 0.1131, "step": 22442 }, { "epoch": 16.44175824175824, "grad_norm": 1.0859375, "learning_rate": 5.94953337426162e-05, "loss": 0.1106, "step": 22443 }, { "epoch": 16.44249084249084, "grad_norm": 1.4296875, "learning_rate": 5.949196441003264e-05, "loss": 0.1689, "step": 22444 }, { "epoch": 16.44322344322344, "grad_norm": 1.2421875, "learning_rate": 5.948859503436562e-05, "loss": 0.13, "step": 22445 }, { "epoch": 16.443956043956042, "grad_norm": 1.6953125, "learning_rate": 5.948522561563129e-05, "loss": 0.1234, "step": 22446 }, { "epoch": 16.444688644688643, "grad_norm": 1.0703125, "learning_rate": 5.948185615384582e-05, "loss": 0.1322, "step": 22447 }, { "epoch": 16.445421245421244, "grad_norm": 1.2734375, "learning_rate": 5.947848664902531e-05, "loss": 0.1072, "step": 22448 }, { "epoch": 16.446153846153845, "grad_norm": 1.6640625, "learning_rate": 5.9475117101185914e-05, "loss": 0.1501, "step": 22449 }, { "epoch": 16.446886446886445, "grad_norm": 1.25, "learning_rate": 5.94717475103438e-05, "loss": 0.1027, "step": 22450 }, { "epoch": 16.447619047619046, "grad_norm": 1.1328125, "learning_rate": 5.946837787651508e-05, "loss": 0.1092, "step": 22451 }, { "epoch": 16.448351648351647, "grad_norm": 0.77734375, "learning_rate": 5.946500819971593e-05, "loss": 0.0987, "step": 22452 }, { "epoch": 16.449084249084247, "grad_norm": 1.2265625, "learning_rate": 5.946163847996249e-05, "loss": 0.1379, "step": 22453 }, { "epoch": 16.449816849816848, "grad_norm": 1.546875, "learning_rate": 5.945826871727089e-05, "loss": 0.1593, "step": 22454 }, { "epoch": 16.45054945054945, "grad_norm": 1.90625, "learning_rate": 5.945489891165727e-05, "loss": 0.1417, "step": 22455 }, { "epoch": 16.45128205128205, "grad_norm": 1.3515625, "learning_rate": 5.945152906313781e-05, "loss": 0.1318, "step": 22456 }, { "epoch": 16.45201465201465, "grad_norm": 1.375, "learning_rate": 5.9448159171728624e-05, "loss": 0.1086, "step": 22457 }, { "epoch": 16.45274725274725, "grad_norm": 1.203125, "learning_rate": 5.944478923744587e-05, "loss": 0.1439, "step": 22458 }, { "epoch": 16.453479853479852, "grad_norm": 1.5859375, "learning_rate": 5.944141926030571e-05, "loss": 0.1534, "step": 22459 }, { "epoch": 16.454212454212453, "grad_norm": 0.79296875, "learning_rate": 5.943804924032426e-05, "loss": 0.0945, "step": 22460 }, { "epoch": 16.454945054945053, "grad_norm": 0.98828125, "learning_rate": 5.9434679177517693e-05, "loss": 0.108, "step": 22461 }, { "epoch": 16.455677655677654, "grad_norm": 1.359375, "learning_rate": 5.943130907190213e-05, "loss": 0.1427, "step": 22462 }, { "epoch": 16.456410256410255, "grad_norm": 1.328125, "learning_rate": 5.9427938923493756e-05, "loss": 0.1171, "step": 22463 }, { "epoch": 16.457142857142856, "grad_norm": 1.1015625, "learning_rate": 5.942456873230866e-05, "loss": 0.1155, "step": 22464 }, { "epoch": 16.457875457875456, "grad_norm": 0.9375, "learning_rate": 5.9421198498363064e-05, "loss": 0.1283, "step": 22465 }, { "epoch": 16.458608058608057, "grad_norm": 1.015625, "learning_rate": 5.9417828221673065e-05, "loss": 0.1189, "step": 22466 }, { "epoch": 16.459340659340658, "grad_norm": 1.4296875, "learning_rate": 5.941445790225481e-05, "loss": 0.136, "step": 22467 }, { "epoch": 16.46007326007326, "grad_norm": 0.9140625, "learning_rate": 5.9411087540124466e-05, "loss": 0.1271, "step": 22468 }, { "epoch": 16.46080586080586, "grad_norm": 1.0, "learning_rate": 5.940771713529819e-05, "loss": 0.1262, "step": 22469 }, { "epoch": 16.46153846153846, "grad_norm": 1.5, "learning_rate": 5.9404346687792104e-05, "loss": 0.1382, "step": 22470 }, { "epoch": 16.46227106227106, "grad_norm": 1.203125, "learning_rate": 5.940097619762236e-05, "loss": 0.1085, "step": 22471 }, { "epoch": 16.46300366300366, "grad_norm": 1.078125, "learning_rate": 5.9397605664805134e-05, "loss": 0.1253, "step": 22472 }, { "epoch": 16.463736263736262, "grad_norm": 0.796875, "learning_rate": 5.939423508935655e-05, "loss": 0.1279, "step": 22473 }, { "epoch": 16.464468864468863, "grad_norm": 0.8359375, "learning_rate": 5.939086447129277e-05, "loss": 0.1485, "step": 22474 }, { "epoch": 16.465201465201464, "grad_norm": 0.6953125, "learning_rate": 5.938749381062993e-05, "loss": 0.1033, "step": 22475 }, { "epoch": 16.465934065934064, "grad_norm": 0.88671875, "learning_rate": 5.938412310738422e-05, "loss": 0.11, "step": 22476 }, { "epoch": 16.466666666666665, "grad_norm": 0.82421875, "learning_rate": 5.9380752361571736e-05, "loss": 0.1046, "step": 22477 }, { "epoch": 16.467399267399266, "grad_norm": 0.875, "learning_rate": 5.9377381573208656e-05, "loss": 0.094, "step": 22478 }, { "epoch": 16.468131868131866, "grad_norm": 1.03125, "learning_rate": 5.9374010742311125e-05, "loss": 0.126, "step": 22479 }, { "epoch": 16.468864468864467, "grad_norm": 0.828125, "learning_rate": 5.937063986889529e-05, "loss": 0.1068, "step": 22480 }, { "epoch": 16.469597069597068, "grad_norm": 1.046875, "learning_rate": 5.936726895297732e-05, "loss": 0.125, "step": 22481 }, { "epoch": 16.47032967032967, "grad_norm": 1.96875, "learning_rate": 5.9363897994573335e-05, "loss": 0.1287, "step": 22482 }, { "epoch": 16.47106227106227, "grad_norm": 0.828125, "learning_rate": 5.9360526993699514e-05, "loss": 0.1082, "step": 22483 }, { "epoch": 16.47179487179487, "grad_norm": 0.98046875, "learning_rate": 5.935715595037201e-05, "loss": 0.108, "step": 22484 }, { "epoch": 16.47252747252747, "grad_norm": 1.03125, "learning_rate": 5.935378486460694e-05, "loss": 0.1245, "step": 22485 }, { "epoch": 16.47326007326007, "grad_norm": 1.109375, "learning_rate": 5.9350413736420516e-05, "loss": 0.1261, "step": 22486 }, { "epoch": 16.473992673992672, "grad_norm": 0.79296875, "learning_rate": 5.934704256582884e-05, "loss": 0.1851, "step": 22487 }, { "epoch": 16.474725274725273, "grad_norm": 1.2265625, "learning_rate": 5.934367135284806e-05, "loss": 0.1682, "step": 22488 }, { "epoch": 16.475457875457874, "grad_norm": 1.1484375, "learning_rate": 5.934030009749437e-05, "loss": 0.1181, "step": 22489 }, { "epoch": 16.476190476190474, "grad_norm": 1.3828125, "learning_rate": 5.9336928799783885e-05, "loss": 0.142, "step": 22490 }, { "epoch": 16.476923076923075, "grad_norm": 0.97265625, "learning_rate": 5.933355745973278e-05, "loss": 0.1324, "step": 22491 }, { "epoch": 16.477655677655676, "grad_norm": 1.0546875, "learning_rate": 5.933018607735721e-05, "loss": 0.1191, "step": 22492 }, { "epoch": 16.478388278388277, "grad_norm": 0.85546875, "learning_rate": 5.93268146526733e-05, "loss": 0.1024, "step": 22493 }, { "epoch": 16.479120879120877, "grad_norm": 1.1328125, "learning_rate": 5.9323443185697235e-05, "loss": 0.1277, "step": 22494 }, { "epoch": 16.479853479853478, "grad_norm": 1.5859375, "learning_rate": 5.9320071676445156e-05, "loss": 0.114, "step": 22495 }, { "epoch": 16.48058608058608, "grad_norm": 1.4296875, "learning_rate": 5.931670012493322e-05, "loss": 0.0941, "step": 22496 }, { "epoch": 16.48131868131868, "grad_norm": 1.1015625, "learning_rate": 5.931332853117758e-05, "loss": 0.1293, "step": 22497 }, { "epoch": 16.48205128205128, "grad_norm": 0.8671875, "learning_rate": 5.930995689519437e-05, "loss": 0.1239, "step": 22498 }, { "epoch": 16.48278388278388, "grad_norm": 0.86328125, "learning_rate": 5.9306585216999797e-05, "loss": 0.1114, "step": 22499 }, { "epoch": 16.483516483516482, "grad_norm": 0.953125, "learning_rate": 5.930321349660997e-05, "loss": 0.095, "step": 22500 }, { "epoch": 16.484249084249083, "grad_norm": 0.8515625, "learning_rate": 5.929984173404105e-05, "loss": 0.1262, "step": 22501 }, { "epoch": 16.484981684981683, "grad_norm": 0.8359375, "learning_rate": 5.929646992930921e-05, "loss": 0.1192, "step": 22502 }, { "epoch": 16.485714285714284, "grad_norm": 1.1640625, "learning_rate": 5.9293098082430574e-05, "loss": 0.1182, "step": 22503 }, { "epoch": 16.486446886446885, "grad_norm": 1.125, "learning_rate": 5.928972619342135e-05, "loss": 0.1281, "step": 22504 }, { "epoch": 16.487179487179485, "grad_norm": 0.87109375, "learning_rate": 5.9286354262297646e-05, "loss": 0.1302, "step": 22505 }, { "epoch": 16.487912087912086, "grad_norm": 0.671875, "learning_rate": 5.928298228907563e-05, "loss": 0.1227, "step": 22506 }, { "epoch": 16.488644688644687, "grad_norm": 1.25, "learning_rate": 5.927961027377148e-05, "loss": 0.1051, "step": 22507 }, { "epoch": 16.489377289377288, "grad_norm": 1.8125, "learning_rate": 5.92762382164013e-05, "loss": 0.1286, "step": 22508 }, { "epoch": 16.49010989010989, "grad_norm": 1.4140625, "learning_rate": 5.927286611698131e-05, "loss": 0.1564, "step": 22509 }, { "epoch": 16.49084249084249, "grad_norm": 1.4140625, "learning_rate": 5.9269493975527635e-05, "loss": 0.1564, "step": 22510 }, { "epoch": 16.49157509157509, "grad_norm": 0.98046875, "learning_rate": 5.9266121792056425e-05, "loss": 0.1137, "step": 22511 }, { "epoch": 16.49230769230769, "grad_norm": 0.8125, "learning_rate": 5.926274956658387e-05, "loss": 0.0995, "step": 22512 }, { "epoch": 16.493040293040295, "grad_norm": 1.328125, "learning_rate": 5.925937729912608e-05, "loss": 0.149, "step": 22513 }, { "epoch": 16.493772893772892, "grad_norm": 1.28125, "learning_rate": 5.925600498969924e-05, "loss": 0.1161, "step": 22514 }, { "epoch": 16.494505494505496, "grad_norm": 0.71875, "learning_rate": 5.9252632638319526e-05, "loss": 0.1143, "step": 22515 }, { "epoch": 16.495238095238093, "grad_norm": 1.2109375, "learning_rate": 5.924926024500306e-05, "loss": 0.1092, "step": 22516 }, { "epoch": 16.495970695970698, "grad_norm": 0.8046875, "learning_rate": 5.924588780976602e-05, "loss": 0.1058, "step": 22517 }, { "epoch": 16.496703296703295, "grad_norm": 0.953125, "learning_rate": 5.924251533262457e-05, "loss": 0.1016, "step": 22518 }, { "epoch": 16.4974358974359, "grad_norm": 1.875, "learning_rate": 5.923914281359484e-05, "loss": 0.1409, "step": 22519 }, { "epoch": 16.498168498168496, "grad_norm": 1.9453125, "learning_rate": 5.9235770252693026e-05, "loss": 0.1469, "step": 22520 }, { "epoch": 16.4989010989011, "grad_norm": 1.0859375, "learning_rate": 5.9232397649935256e-05, "loss": 0.1305, "step": 22521 }, { "epoch": 16.499633699633698, "grad_norm": 1.2265625, "learning_rate": 5.9229025005337703e-05, "loss": 0.1531, "step": 22522 }, { "epoch": 16.500366300366302, "grad_norm": 0.8671875, "learning_rate": 5.9225652318916524e-05, "loss": 0.1289, "step": 22523 }, { "epoch": 16.5010989010989, "grad_norm": 1.140625, "learning_rate": 5.922227959068788e-05, "loss": 0.1261, "step": 22524 }, { "epoch": 16.501831501831504, "grad_norm": 1.46875, "learning_rate": 5.9218906820667955e-05, "loss": 0.1002, "step": 22525 }, { "epoch": 16.5025641025641, "grad_norm": 1.359375, "learning_rate": 5.921553400887286e-05, "loss": 0.1167, "step": 22526 }, { "epoch": 16.503296703296705, "grad_norm": 1.0859375, "learning_rate": 5.9212161155318794e-05, "loss": 0.1377, "step": 22527 }, { "epoch": 16.504029304029302, "grad_norm": 1.1953125, "learning_rate": 5.92087882600219e-05, "loss": 0.1555, "step": 22528 }, { "epoch": 16.504761904761907, "grad_norm": 0.75390625, "learning_rate": 5.920541532299832e-05, "loss": 0.1206, "step": 22529 }, { "epoch": 16.505494505494504, "grad_norm": 1.1953125, "learning_rate": 5.920204234426428e-05, "loss": 0.1152, "step": 22530 }, { "epoch": 16.506227106227108, "grad_norm": 1.0234375, "learning_rate": 5.919866932383586e-05, "loss": 0.1346, "step": 22531 }, { "epoch": 16.506959706959705, "grad_norm": 0.72265625, "learning_rate": 5.9195296261729286e-05, "loss": 0.1183, "step": 22532 }, { "epoch": 16.50769230769231, "grad_norm": 0.7265625, "learning_rate": 5.919192315796067e-05, "loss": 0.1038, "step": 22533 }, { "epoch": 16.508424908424907, "grad_norm": 0.9609375, "learning_rate": 5.918855001254622e-05, "loss": 0.0847, "step": 22534 }, { "epoch": 16.50915750915751, "grad_norm": 1.1640625, "learning_rate": 5.918517682550205e-05, "loss": 0.1335, "step": 22535 }, { "epoch": 16.509890109890108, "grad_norm": 1.0390625, "learning_rate": 5.918180359684436e-05, "loss": 0.1185, "step": 22536 }, { "epoch": 16.510622710622712, "grad_norm": 0.765625, "learning_rate": 5.9178430326589305e-05, "loss": 0.1105, "step": 22537 }, { "epoch": 16.51135531135531, "grad_norm": 0.8828125, "learning_rate": 5.9175057014753035e-05, "loss": 0.107, "step": 22538 }, { "epoch": 16.512087912087914, "grad_norm": 0.66796875, "learning_rate": 5.9171683661351726e-05, "loss": 0.117, "step": 22539 }, { "epoch": 16.51282051282051, "grad_norm": 1.359375, "learning_rate": 5.9168310266401526e-05, "loss": 0.147, "step": 22540 }, { "epoch": 16.513553113553115, "grad_norm": 0.87890625, "learning_rate": 5.9164936829918604e-05, "loss": 0.117, "step": 22541 }, { "epoch": 16.514285714285712, "grad_norm": 0.66796875, "learning_rate": 5.916156335191912e-05, "loss": 0.1201, "step": 22542 }, { "epoch": 16.515018315018317, "grad_norm": 0.97265625, "learning_rate": 5.9158189832419266e-05, "loss": 0.1376, "step": 22543 }, { "epoch": 16.515750915750917, "grad_norm": 0.91015625, "learning_rate": 5.915481627143518e-05, "loss": 0.1213, "step": 22544 }, { "epoch": 16.516483516483518, "grad_norm": 0.94140625, "learning_rate": 5.9151442668983e-05, "loss": 0.1097, "step": 22545 }, { "epoch": 16.51721611721612, "grad_norm": 1.609375, "learning_rate": 5.914806902507894e-05, "loss": 0.144, "step": 22546 }, { "epoch": 16.51794871794872, "grad_norm": 1.1484375, "learning_rate": 5.914469533973913e-05, "loss": 0.1225, "step": 22547 }, { "epoch": 16.51868131868132, "grad_norm": 0.84375, "learning_rate": 5.914132161297975e-05, "loss": 0.1239, "step": 22548 }, { "epoch": 16.51941391941392, "grad_norm": 0.8203125, "learning_rate": 5.9137947844816964e-05, "loss": 0.1308, "step": 22549 }, { "epoch": 16.520146520146522, "grad_norm": 1.140625, "learning_rate": 5.913457403526692e-05, "loss": 0.1211, "step": 22550 }, { "epoch": 16.520879120879123, "grad_norm": 1.328125, "learning_rate": 5.9131200184345825e-05, "loss": 0.1412, "step": 22551 }, { "epoch": 16.521611721611723, "grad_norm": 1.1484375, "learning_rate": 5.91278262920698e-05, "loss": 0.1103, "step": 22552 }, { "epoch": 16.522344322344324, "grad_norm": 0.79296875, "learning_rate": 5.912445235845503e-05, "loss": 0.1213, "step": 22553 }, { "epoch": 16.523076923076925, "grad_norm": 0.93359375, "learning_rate": 5.912107838351768e-05, "loss": 0.0979, "step": 22554 }, { "epoch": 16.523809523809526, "grad_norm": 1.1875, "learning_rate": 5.911770436727391e-05, "loss": 0.1334, "step": 22555 }, { "epoch": 16.524542124542126, "grad_norm": 0.78125, "learning_rate": 5.91143303097399e-05, "loss": 0.1129, "step": 22556 }, { "epoch": 16.525274725274727, "grad_norm": 0.79296875, "learning_rate": 5.91109562109318e-05, "loss": 0.121, "step": 22557 }, { "epoch": 16.526007326007328, "grad_norm": 0.88671875, "learning_rate": 5.9107582070865765e-05, "loss": 0.0931, "step": 22558 }, { "epoch": 16.52673992673993, "grad_norm": 0.82421875, "learning_rate": 5.9104207889557994e-05, "loss": 0.1185, "step": 22559 }, { "epoch": 16.52747252747253, "grad_norm": 0.77734375, "learning_rate": 5.9100833667024635e-05, "loss": 0.1013, "step": 22560 }, { "epoch": 16.52820512820513, "grad_norm": 1.125, "learning_rate": 5.9097459403281855e-05, "loss": 0.1124, "step": 22561 }, { "epoch": 16.52893772893773, "grad_norm": 1.3984375, "learning_rate": 5.909408509834583e-05, "loss": 0.1425, "step": 22562 }, { "epoch": 16.52967032967033, "grad_norm": 0.65625, "learning_rate": 5.9090710752232715e-05, "loss": 0.1176, "step": 22563 }, { "epoch": 16.530402930402932, "grad_norm": 0.88671875, "learning_rate": 5.90873363649587e-05, "loss": 0.1223, "step": 22564 }, { "epoch": 16.531135531135533, "grad_norm": 0.92578125, "learning_rate": 5.908396193653992e-05, "loss": 0.1216, "step": 22565 }, { "epoch": 16.531868131868134, "grad_norm": 0.81640625, "learning_rate": 5.908058746699256e-05, "loss": 0.1243, "step": 22566 }, { "epoch": 16.532600732600734, "grad_norm": 0.77734375, "learning_rate": 5.907721295633281e-05, "loss": 0.1129, "step": 22567 }, { "epoch": 16.533333333333335, "grad_norm": 2.0, "learning_rate": 5.90738384045768e-05, "loss": 0.148, "step": 22568 }, { "epoch": 16.534065934065936, "grad_norm": 1.3984375, "learning_rate": 5.9070463811740725e-05, "loss": 0.1478, "step": 22569 }, { "epoch": 16.534798534798536, "grad_norm": 1.21875, "learning_rate": 5.906708917784074e-05, "loss": 0.1169, "step": 22570 }, { "epoch": 16.535531135531137, "grad_norm": 0.90234375, "learning_rate": 5.906371450289301e-05, "loss": 0.1113, "step": 22571 }, { "epoch": 16.536263736263738, "grad_norm": 0.91796875, "learning_rate": 5.906033978691371e-05, "loss": 0.1257, "step": 22572 }, { "epoch": 16.53699633699634, "grad_norm": 0.74609375, "learning_rate": 5.905696502991903e-05, "loss": 0.1177, "step": 22573 }, { "epoch": 16.53772893772894, "grad_norm": 1.8125, "learning_rate": 5.9053590231925115e-05, "loss": 0.1212, "step": 22574 }, { "epoch": 16.53846153846154, "grad_norm": 1.125, "learning_rate": 5.905021539294814e-05, "loss": 0.1006, "step": 22575 }, { "epoch": 16.53919413919414, "grad_norm": 1.140625, "learning_rate": 5.9046840513004266e-05, "loss": 0.1167, "step": 22576 }, { "epoch": 16.53992673992674, "grad_norm": 1.2578125, "learning_rate": 5.904346559210969e-05, "loss": 0.111, "step": 22577 }, { "epoch": 16.540659340659342, "grad_norm": 1.7578125, "learning_rate": 5.904009063028057e-05, "loss": 0.123, "step": 22578 }, { "epoch": 16.541391941391943, "grad_norm": 1.2265625, "learning_rate": 5.903671562753306e-05, "loss": 0.1058, "step": 22579 }, { "epoch": 16.542124542124544, "grad_norm": 0.6796875, "learning_rate": 5.903334058388334e-05, "loss": 0.1112, "step": 22580 }, { "epoch": 16.542857142857144, "grad_norm": 0.91015625, "learning_rate": 5.902996549934757e-05, "loss": 0.1435, "step": 22581 }, { "epoch": 16.543589743589745, "grad_norm": 1.515625, "learning_rate": 5.902659037394197e-05, "loss": 0.1355, "step": 22582 }, { "epoch": 16.544322344322346, "grad_norm": 0.83203125, "learning_rate": 5.902321520768267e-05, "loss": 0.1242, "step": 22583 }, { "epoch": 16.545054945054947, "grad_norm": 1.0546875, "learning_rate": 5.901984000058584e-05, "loss": 0.1195, "step": 22584 }, { "epoch": 16.545787545787547, "grad_norm": 1.0234375, "learning_rate": 5.9016464752667644e-05, "loss": 0.1198, "step": 22585 }, { "epoch": 16.546520146520148, "grad_norm": 0.765625, "learning_rate": 5.901308946394428e-05, "loss": 0.1152, "step": 22586 }, { "epoch": 16.54725274725275, "grad_norm": 1.0390625, "learning_rate": 5.9009714134431905e-05, "loss": 0.1327, "step": 22587 }, { "epoch": 16.54798534798535, "grad_norm": 1.25, "learning_rate": 5.9006338764146716e-05, "loss": 0.1211, "step": 22588 }, { "epoch": 16.54871794871795, "grad_norm": 1.2734375, "learning_rate": 5.900296335310484e-05, "loss": 0.1128, "step": 22589 }, { "epoch": 16.54945054945055, "grad_norm": 1.1875, "learning_rate": 5.8999587901322495e-05, "loss": 0.1213, "step": 22590 }, { "epoch": 16.550183150183152, "grad_norm": 0.765625, "learning_rate": 5.899621240881583e-05, "loss": 0.1032, "step": 22591 }, { "epoch": 16.550915750915753, "grad_norm": 0.76171875, "learning_rate": 5.899283687560102e-05, "loss": 0.1123, "step": 22592 }, { "epoch": 16.551648351648353, "grad_norm": 0.81640625, "learning_rate": 5.898946130169424e-05, "loss": 0.1125, "step": 22593 }, { "epoch": 16.552380952380954, "grad_norm": 1.25, "learning_rate": 5.8986085687111656e-05, "loss": 0.1021, "step": 22594 }, { "epoch": 16.553113553113555, "grad_norm": 1.5, "learning_rate": 5.898271003186947e-05, "loss": 0.0912, "step": 22595 }, { "epoch": 16.553846153846155, "grad_norm": 1.46875, "learning_rate": 5.8979334335983824e-05, "loss": 0.1485, "step": 22596 }, { "epoch": 16.554578754578756, "grad_norm": 1.515625, "learning_rate": 5.8975958599470896e-05, "loss": 0.1357, "step": 22597 }, { "epoch": 16.555311355311357, "grad_norm": 0.83984375, "learning_rate": 5.8972582822346866e-05, "loss": 0.1204, "step": 22598 }, { "epoch": 16.556043956043958, "grad_norm": 0.67578125, "learning_rate": 5.896920700462793e-05, "loss": 0.1223, "step": 22599 }, { "epoch": 16.55677655677656, "grad_norm": 1.0625, "learning_rate": 5.896583114633023e-05, "loss": 0.1145, "step": 22600 }, { "epoch": 16.55750915750916, "grad_norm": 1.2109375, "learning_rate": 5.896245524746996e-05, "loss": 0.1, "step": 22601 }, { "epoch": 16.55824175824176, "grad_norm": 1.0625, "learning_rate": 5.895907930806327e-05, "loss": 0.1239, "step": 22602 }, { "epoch": 16.55897435897436, "grad_norm": 0.96484375, "learning_rate": 5.895570332812638e-05, "loss": 0.1199, "step": 22603 }, { "epoch": 16.55970695970696, "grad_norm": 1.1171875, "learning_rate": 5.8952327307675437e-05, "loss": 0.1154, "step": 22604 }, { "epoch": 16.560439560439562, "grad_norm": 0.93359375, "learning_rate": 5.89489512467266e-05, "loss": 0.0985, "step": 22605 }, { "epoch": 16.561172161172163, "grad_norm": 1.140625, "learning_rate": 5.8945575145296086e-05, "loss": 0.1382, "step": 22606 }, { "epoch": 16.561904761904763, "grad_norm": 1.140625, "learning_rate": 5.894219900340002e-05, "loss": 0.125, "step": 22607 }, { "epoch": 16.562637362637364, "grad_norm": 0.921875, "learning_rate": 5.893882282105464e-05, "loss": 0.0926, "step": 22608 }, { "epoch": 16.563369963369965, "grad_norm": 1.0859375, "learning_rate": 5.8935446598276065e-05, "loss": 0.1262, "step": 22609 }, { "epoch": 16.564102564102566, "grad_norm": 1.453125, "learning_rate": 5.8932070335080504e-05, "loss": 0.1394, "step": 22610 }, { "epoch": 16.564835164835166, "grad_norm": 1.109375, "learning_rate": 5.892869403148412e-05, "loss": 0.1151, "step": 22611 }, { "epoch": 16.565567765567767, "grad_norm": 1.1953125, "learning_rate": 5.892531768750311e-05, "loss": 0.1271, "step": 22612 }, { "epoch": 16.566300366300368, "grad_norm": 1.828125, "learning_rate": 5.892194130315363e-05, "loss": 0.1331, "step": 22613 }, { "epoch": 16.56703296703297, "grad_norm": 1.2421875, "learning_rate": 5.891856487845186e-05, "loss": 0.125, "step": 22614 }, { "epoch": 16.56776556776557, "grad_norm": 1.484375, "learning_rate": 5.891518841341399e-05, "loss": 0.1198, "step": 22615 }, { "epoch": 16.56849816849817, "grad_norm": 0.94140625, "learning_rate": 5.891181190805619e-05, "loss": 0.1174, "step": 22616 }, { "epoch": 16.56923076923077, "grad_norm": 1.65625, "learning_rate": 5.890843536239463e-05, "loss": 0.1389, "step": 22617 }, { "epoch": 16.56996336996337, "grad_norm": 1.2734375, "learning_rate": 5.89050587764455e-05, "loss": 0.1504, "step": 22618 }, { "epoch": 16.570695970695972, "grad_norm": 0.62109375, "learning_rate": 5.890168215022499e-05, "loss": 0.1021, "step": 22619 }, { "epoch": 16.571428571428573, "grad_norm": 0.7578125, "learning_rate": 5.8898305483749236e-05, "loss": 0.1221, "step": 22620 }, { "epoch": 16.572161172161174, "grad_norm": 1.4296875, "learning_rate": 5.8894928777034464e-05, "loss": 0.1192, "step": 22621 }, { "epoch": 16.572893772893774, "grad_norm": 0.73828125, "learning_rate": 5.889155203009683e-05, "loss": 0.1279, "step": 22622 }, { "epoch": 16.573626373626375, "grad_norm": 1.0078125, "learning_rate": 5.888817524295249e-05, "loss": 0.1357, "step": 22623 }, { "epoch": 16.574358974358976, "grad_norm": 0.9140625, "learning_rate": 5.888479841561766e-05, "loss": 0.1041, "step": 22624 }, { "epoch": 16.575091575091577, "grad_norm": 0.9453125, "learning_rate": 5.888142154810853e-05, "loss": 0.1194, "step": 22625 }, { "epoch": 16.575824175824177, "grad_norm": 1.3203125, "learning_rate": 5.8878044640441233e-05, "loss": 0.1333, "step": 22626 }, { "epoch": 16.576556776556778, "grad_norm": 1.2578125, "learning_rate": 5.887466769263199e-05, "loss": 0.128, "step": 22627 }, { "epoch": 16.57728937728938, "grad_norm": 0.90625, "learning_rate": 5.887129070469696e-05, "loss": 0.1288, "step": 22628 }, { "epoch": 16.57802197802198, "grad_norm": 0.890625, "learning_rate": 5.886791367665232e-05, "loss": 0.0958, "step": 22629 }, { "epoch": 16.57875457875458, "grad_norm": 1.015625, "learning_rate": 5.8864536608514274e-05, "loss": 0.1111, "step": 22630 }, { "epoch": 16.57948717948718, "grad_norm": 1.0859375, "learning_rate": 5.8861159500298966e-05, "loss": 0.1099, "step": 22631 }, { "epoch": 16.58021978021978, "grad_norm": 1.15625, "learning_rate": 5.8857782352022625e-05, "loss": 0.1399, "step": 22632 }, { "epoch": 16.580952380952382, "grad_norm": 1.375, "learning_rate": 5.885440516370138e-05, "loss": 0.1345, "step": 22633 }, { "epoch": 16.581684981684983, "grad_norm": 1.6171875, "learning_rate": 5.8851027935351457e-05, "loss": 0.0969, "step": 22634 }, { "epoch": 16.582417582417584, "grad_norm": 0.72265625, "learning_rate": 5.884765066698902e-05, "loss": 0.1259, "step": 22635 }, { "epoch": 16.583150183150185, "grad_norm": 0.96484375, "learning_rate": 5.8844273358630234e-05, "loss": 0.1386, "step": 22636 }, { "epoch": 16.583882783882785, "grad_norm": 2.125, "learning_rate": 5.884089601029129e-05, "loss": 0.1528, "step": 22637 }, { "epoch": 16.584615384615386, "grad_norm": 0.95703125, "learning_rate": 5.8837518621988385e-05, "loss": 0.1278, "step": 22638 }, { "epoch": 16.585347985347987, "grad_norm": 0.875, "learning_rate": 5.883414119373769e-05, "loss": 0.116, "step": 22639 }, { "epoch": 16.586080586080588, "grad_norm": 1.4921875, "learning_rate": 5.883076372555538e-05, "loss": 0.1242, "step": 22640 }, { "epoch": 16.58681318681319, "grad_norm": 0.84375, "learning_rate": 5.882738621745766e-05, "loss": 0.1076, "step": 22641 }, { "epoch": 16.58754578754579, "grad_norm": 1.1015625, "learning_rate": 5.8824008669460696e-05, "loss": 0.1167, "step": 22642 }, { "epoch": 16.58827838827839, "grad_norm": 1.34375, "learning_rate": 5.882063108158068e-05, "loss": 0.1164, "step": 22643 }, { "epoch": 16.58901098901099, "grad_norm": 1.125, "learning_rate": 5.881725345383376e-05, "loss": 0.123, "step": 22644 }, { "epoch": 16.58974358974359, "grad_norm": 0.89453125, "learning_rate": 5.881387578623618e-05, "loss": 0.1123, "step": 22645 }, { "epoch": 16.590476190476192, "grad_norm": 1.3203125, "learning_rate": 5.881049807880408e-05, "loss": 0.1112, "step": 22646 }, { "epoch": 16.591208791208793, "grad_norm": 0.8828125, "learning_rate": 5.8807120331553655e-05, "loss": 0.1074, "step": 22647 }, { "epoch": 16.591941391941393, "grad_norm": 0.640625, "learning_rate": 5.8803742544501106e-05, "loss": 0.1256, "step": 22648 }, { "epoch": 16.592673992673994, "grad_norm": 1.046875, "learning_rate": 5.880036471766257e-05, "loss": 0.1285, "step": 22649 }, { "epoch": 16.593406593406595, "grad_norm": 1.3515625, "learning_rate": 5.879698685105428e-05, "loss": 0.1384, "step": 22650 }, { "epoch": 16.594139194139196, "grad_norm": 1.53125, "learning_rate": 5.879360894469239e-05, "loss": 0.1473, "step": 22651 }, { "epoch": 16.594871794871796, "grad_norm": 1.2265625, "learning_rate": 5.879023099859311e-05, "loss": 0.1399, "step": 22652 }, { "epoch": 16.595604395604397, "grad_norm": 1.484375, "learning_rate": 5.8786853012772616e-05, "loss": 0.1177, "step": 22653 }, { "epoch": 16.596336996336998, "grad_norm": 0.9765625, "learning_rate": 5.878347498724707e-05, "loss": 0.1459, "step": 22654 }, { "epoch": 16.5970695970696, "grad_norm": 1.390625, "learning_rate": 5.878009692203269e-05, "loss": 0.1222, "step": 22655 }, { "epoch": 16.5978021978022, "grad_norm": 1.4609375, "learning_rate": 5.8776718817145646e-05, "loss": 0.1233, "step": 22656 }, { "epoch": 16.5985347985348, "grad_norm": 1.59375, "learning_rate": 5.8773340672602115e-05, "loss": 0.1416, "step": 22657 }, { "epoch": 16.5992673992674, "grad_norm": 0.9375, "learning_rate": 5.876996248841832e-05, "loss": 0.1149, "step": 22658 }, { "epoch": 16.6, "grad_norm": 1.25, "learning_rate": 5.876658426461039e-05, "loss": 0.1457, "step": 22659 }, { "epoch": 16.600732600732602, "grad_norm": 1.1015625, "learning_rate": 5.876320600119455e-05, "loss": 0.1338, "step": 22660 }, { "epoch": 16.601465201465203, "grad_norm": 0.70703125, "learning_rate": 5.8759827698186985e-05, "loss": 0.1111, "step": 22661 }, { "epoch": 16.602197802197804, "grad_norm": 1.328125, "learning_rate": 5.875644935560386e-05, "loss": 0.1357, "step": 22662 }, { "epoch": 16.602930402930404, "grad_norm": 1.2109375, "learning_rate": 5.875307097346138e-05, "loss": 0.1365, "step": 22663 }, { "epoch": 16.603663003663005, "grad_norm": 1.078125, "learning_rate": 5.8749692551775736e-05, "loss": 0.1211, "step": 22664 }, { "epoch": 16.604395604395606, "grad_norm": 1.1640625, "learning_rate": 5.8746314090563104e-05, "loss": 0.1413, "step": 22665 }, { "epoch": 16.605128205128207, "grad_norm": 0.81640625, "learning_rate": 5.874293558983968e-05, "loss": 0.0948, "step": 22666 }, { "epoch": 16.605860805860807, "grad_norm": 0.83203125, "learning_rate": 5.873955704962163e-05, "loss": 0.0872, "step": 22667 }, { "epoch": 16.606593406593408, "grad_norm": 0.8828125, "learning_rate": 5.873617846992516e-05, "loss": 0.0984, "step": 22668 }, { "epoch": 16.60732600732601, "grad_norm": 1.2890625, "learning_rate": 5.873279985076647e-05, "loss": 0.128, "step": 22669 }, { "epoch": 16.60805860805861, "grad_norm": 0.58984375, "learning_rate": 5.8729421192161714e-05, "loss": 0.0933, "step": 22670 }, { "epoch": 16.60879120879121, "grad_norm": 1.546875, "learning_rate": 5.872604249412712e-05, "loss": 0.1358, "step": 22671 }, { "epoch": 16.60952380952381, "grad_norm": 1.078125, "learning_rate": 5.872266375667885e-05, "loss": 0.0969, "step": 22672 }, { "epoch": 16.61025641025641, "grad_norm": 0.921875, "learning_rate": 5.871928497983309e-05, "loss": 0.1268, "step": 22673 }, { "epoch": 16.610989010989012, "grad_norm": 0.92578125, "learning_rate": 5.871590616360605e-05, "loss": 0.1317, "step": 22674 }, { "epoch": 16.611721611721613, "grad_norm": 1.8984375, "learning_rate": 5.8712527308013887e-05, "loss": 0.1293, "step": 22675 }, { "epoch": 16.612454212454214, "grad_norm": 1.515625, "learning_rate": 5.870914841307281e-05, "loss": 0.1332, "step": 22676 }, { "epoch": 16.613186813186815, "grad_norm": 1.4375, "learning_rate": 5.870576947879902e-05, "loss": 0.1163, "step": 22677 }, { "epoch": 16.613919413919415, "grad_norm": 1.1171875, "learning_rate": 5.870239050520869e-05, "loss": 0.1312, "step": 22678 }, { "epoch": 16.614652014652016, "grad_norm": 1.140625, "learning_rate": 5.869901149231804e-05, "loss": 0.0995, "step": 22679 }, { "epoch": 16.615384615384617, "grad_norm": 1.296875, "learning_rate": 5.86956324401432e-05, "loss": 0.1431, "step": 22680 }, { "epoch": 16.616117216117217, "grad_norm": 0.75, "learning_rate": 5.8692253348700424e-05, "loss": 0.1078, "step": 22681 }, { "epoch": 16.616849816849818, "grad_norm": 1.203125, "learning_rate": 5.868887421800585e-05, "loss": 0.1076, "step": 22682 }, { "epoch": 16.61758241758242, "grad_norm": 0.984375, "learning_rate": 5.868549504807569e-05, "loss": 0.1195, "step": 22683 }, { "epoch": 16.61831501831502, "grad_norm": 1.5, "learning_rate": 5.8682115838926143e-05, "loss": 0.1126, "step": 22684 }, { "epoch": 16.61904761904762, "grad_norm": 0.94140625, "learning_rate": 5.867873659057339e-05, "loss": 0.0867, "step": 22685 }, { "epoch": 16.61978021978022, "grad_norm": 1.0234375, "learning_rate": 5.8675357303033626e-05, "loss": 0.1156, "step": 22686 }, { "epoch": 16.620512820512822, "grad_norm": 0.8046875, "learning_rate": 5.867197797632305e-05, "loss": 0.1285, "step": 22687 }, { "epoch": 16.621245421245423, "grad_norm": 1.1171875, "learning_rate": 5.8668598610457835e-05, "loss": 0.1241, "step": 22688 }, { "epoch": 16.621978021978023, "grad_norm": 1.078125, "learning_rate": 5.866521920545419e-05, "loss": 0.0986, "step": 22689 }, { "epoch": 16.622710622710624, "grad_norm": 0.921875, "learning_rate": 5.86618397613283e-05, "loss": 0.1113, "step": 22690 }, { "epoch": 16.623443223443225, "grad_norm": 1.4765625, "learning_rate": 5.865846027809634e-05, "loss": 0.1389, "step": 22691 }, { "epoch": 16.624175824175826, "grad_norm": 1.046875, "learning_rate": 5.865508075577454e-05, "loss": 0.1298, "step": 22692 }, { "epoch": 16.624908424908426, "grad_norm": 0.9375, "learning_rate": 5.865170119437905e-05, "loss": 0.1108, "step": 22693 }, { "epoch": 16.625641025641027, "grad_norm": 1.25, "learning_rate": 5.86483215939261e-05, "loss": 0.1502, "step": 22694 }, { "epoch": 16.626373626373628, "grad_norm": 0.6640625, "learning_rate": 5.864494195443188e-05, "loss": 0.1069, "step": 22695 }, { "epoch": 16.62710622710623, "grad_norm": 1.3046875, "learning_rate": 5.864156227591253e-05, "loss": 0.169, "step": 22696 }, { "epoch": 16.62783882783883, "grad_norm": 1.046875, "learning_rate": 5.863818255838431e-05, "loss": 0.1107, "step": 22697 }, { "epoch": 16.62857142857143, "grad_norm": 1.2734375, "learning_rate": 5.8634802801863366e-05, "loss": 0.1378, "step": 22698 }, { "epoch": 16.62930402930403, "grad_norm": 1.203125, "learning_rate": 5.8631423006365926e-05, "loss": 0.163, "step": 22699 }, { "epoch": 16.63003663003663, "grad_norm": 1.0546875, "learning_rate": 5.862804317190816e-05, "loss": 0.1369, "step": 22700 }, { "epoch": 16.630769230769232, "grad_norm": 1.2890625, "learning_rate": 5.862466329850629e-05, "loss": 0.1449, "step": 22701 }, { "epoch": 16.631501831501833, "grad_norm": 1.484375, "learning_rate": 5.862128338617647e-05, "loss": 0.1396, "step": 22702 }, { "epoch": 16.632234432234434, "grad_norm": 1.078125, "learning_rate": 5.861790343493494e-05, "loss": 0.1108, "step": 22703 }, { "epoch": 16.632967032967034, "grad_norm": 1.3671875, "learning_rate": 5.861452344479784e-05, "loss": 0.1526, "step": 22704 }, { "epoch": 16.633699633699635, "grad_norm": 2.140625, "learning_rate": 5.861114341578141e-05, "loss": 0.1443, "step": 22705 }, { "epoch": 16.634432234432236, "grad_norm": 1.2890625, "learning_rate": 5.860776334790183e-05, "loss": 0.1135, "step": 22706 }, { "epoch": 16.635164835164836, "grad_norm": 1.3515625, "learning_rate": 5.860438324117529e-05, "loss": 0.1303, "step": 22707 }, { "epoch": 16.635897435897437, "grad_norm": 1.5390625, "learning_rate": 5.8601003095618004e-05, "loss": 0.145, "step": 22708 }, { "epoch": 16.636630036630038, "grad_norm": 0.80859375, "learning_rate": 5.859762291124613e-05, "loss": 0.1321, "step": 22709 }, { "epoch": 16.63736263736264, "grad_norm": 0.87890625, "learning_rate": 5.859424268807591e-05, "loss": 0.1409, "step": 22710 }, { "epoch": 16.63809523809524, "grad_norm": 0.6015625, "learning_rate": 5.859086242612348e-05, "loss": 0.1066, "step": 22711 }, { "epoch": 16.63882783882784, "grad_norm": 0.90625, "learning_rate": 5.85874821254051e-05, "loss": 0.1189, "step": 22712 }, { "epoch": 16.63956043956044, "grad_norm": 0.90234375, "learning_rate": 5.858410178593693e-05, "loss": 0.1012, "step": 22713 }, { "epoch": 16.64029304029304, "grad_norm": 1.0625, "learning_rate": 5.858072140773518e-05, "loss": 0.1441, "step": 22714 }, { "epoch": 16.641025641025642, "grad_norm": 1.6171875, "learning_rate": 5.857734099081603e-05, "loss": 0.1515, "step": 22715 }, { "epoch": 16.641758241758243, "grad_norm": 0.796875, "learning_rate": 5.857396053519572e-05, "loss": 0.1308, "step": 22716 }, { "epoch": 16.642490842490844, "grad_norm": 2.0, "learning_rate": 5.8570580040890376e-05, "loss": 0.1446, "step": 22717 }, { "epoch": 16.643223443223444, "grad_norm": 1.1640625, "learning_rate": 5.856719950791625e-05, "loss": 0.1236, "step": 22718 }, { "epoch": 16.643956043956045, "grad_norm": 0.94921875, "learning_rate": 5.856381893628952e-05, "loss": 0.1185, "step": 22719 }, { "epoch": 16.644688644688646, "grad_norm": 1.0703125, "learning_rate": 5.8560438326026376e-05, "loss": 0.1326, "step": 22720 }, { "epoch": 16.645421245421247, "grad_norm": 0.6875, "learning_rate": 5.8557057677143045e-05, "loss": 0.1169, "step": 22721 }, { "epoch": 16.646153846153847, "grad_norm": 1.578125, "learning_rate": 5.855367698965571e-05, "loss": 0.1544, "step": 22722 }, { "epoch": 16.646886446886448, "grad_norm": 0.84765625, "learning_rate": 5.855029626358054e-05, "loss": 0.1345, "step": 22723 }, { "epoch": 16.64761904761905, "grad_norm": 1.2578125, "learning_rate": 5.8546915498933774e-05, "loss": 0.1468, "step": 22724 }, { "epoch": 16.64835164835165, "grad_norm": 0.89453125, "learning_rate": 5.854353469573159e-05, "loss": 0.1578, "step": 22725 }, { "epoch": 16.64908424908425, "grad_norm": 1.109375, "learning_rate": 5.8540153853990175e-05, "loss": 0.1315, "step": 22726 }, { "epoch": 16.64981684981685, "grad_norm": 1.0546875, "learning_rate": 5.8536772973725774e-05, "loss": 0.127, "step": 22727 }, { "epoch": 16.650549450549452, "grad_norm": 0.8125, "learning_rate": 5.853339205495454e-05, "loss": 0.1139, "step": 22728 }, { "epoch": 16.651282051282053, "grad_norm": 1.375, "learning_rate": 5.853001109769268e-05, "loss": 0.1229, "step": 22729 }, { "epoch": 16.652014652014653, "grad_norm": 1.0234375, "learning_rate": 5.852663010195641e-05, "loss": 0.1442, "step": 22730 }, { "epoch": 16.652747252747254, "grad_norm": 1.4921875, "learning_rate": 5.852324906776192e-05, "loss": 0.1553, "step": 22731 }, { "epoch": 16.653479853479855, "grad_norm": 0.8125, "learning_rate": 5.85198679951254e-05, "loss": 0.1157, "step": 22732 }, { "epoch": 16.654212454212455, "grad_norm": 0.71875, "learning_rate": 5.851648688406308e-05, "loss": 0.1051, "step": 22733 }, { "epoch": 16.654945054945056, "grad_norm": 0.90625, "learning_rate": 5.851310573459115e-05, "loss": 0.169, "step": 22734 }, { "epoch": 16.655677655677657, "grad_norm": 1.84375, "learning_rate": 5.850972454672576e-05, "loss": 0.1214, "step": 22735 }, { "epoch": 16.656410256410258, "grad_norm": 1.0390625, "learning_rate": 5.850634332048317e-05, "loss": 0.1125, "step": 22736 }, { "epoch": 16.65714285714286, "grad_norm": 0.8125, "learning_rate": 5.850296205587955e-05, "loss": 0.1019, "step": 22737 }, { "epoch": 16.65787545787546, "grad_norm": 0.98828125, "learning_rate": 5.849958075293113e-05, "loss": 0.1057, "step": 22738 }, { "epoch": 16.65860805860806, "grad_norm": 1.03125, "learning_rate": 5.8496199411654084e-05, "loss": 0.1273, "step": 22739 }, { "epoch": 16.65934065934066, "grad_norm": 0.65234375, "learning_rate": 5.849281803206462e-05, "loss": 0.117, "step": 22740 }, { "epoch": 16.66007326007326, "grad_norm": 0.640625, "learning_rate": 5.8489436614178936e-05, "loss": 0.1261, "step": 22741 }, { "epoch": 16.660805860805862, "grad_norm": 1.0546875, "learning_rate": 5.848605515801325e-05, "loss": 0.1195, "step": 22742 }, { "epoch": 16.661538461538463, "grad_norm": 1.140625, "learning_rate": 5.8482673663583745e-05, "loss": 0.1263, "step": 22743 }, { "epoch": 16.662271062271063, "grad_norm": 0.74609375, "learning_rate": 5.847929213090664e-05, "loss": 0.1139, "step": 22744 }, { "epoch": 16.663003663003664, "grad_norm": 1.1328125, "learning_rate": 5.84759105599981e-05, "loss": 0.1125, "step": 22745 }, { "epoch": 16.663736263736265, "grad_norm": 0.96484375, "learning_rate": 5.8472528950874385e-05, "loss": 0.1168, "step": 22746 }, { "epoch": 16.664468864468866, "grad_norm": 1.0546875, "learning_rate": 5.846914730355165e-05, "loss": 0.1233, "step": 22747 }, { "epoch": 16.665201465201466, "grad_norm": 1.03125, "learning_rate": 5.846576561804612e-05, "loss": 0.1325, "step": 22748 }, { "epoch": 16.665934065934067, "grad_norm": 1.25, "learning_rate": 5.8462383894373996e-05, "loss": 0.1328, "step": 22749 }, { "epoch": 16.666666666666668, "grad_norm": 1.234375, "learning_rate": 5.8459002132551464e-05, "loss": 0.1118, "step": 22750 }, { "epoch": 16.66739926739927, "grad_norm": 0.92578125, "learning_rate": 5.845562033259476e-05, "loss": 0.1178, "step": 22751 }, { "epoch": 16.66813186813187, "grad_norm": 1.234375, "learning_rate": 5.845223849452004e-05, "loss": 0.128, "step": 22752 }, { "epoch": 16.66886446886447, "grad_norm": 1.296875, "learning_rate": 5.8448856618343554e-05, "loss": 0.112, "step": 22753 }, { "epoch": 16.66959706959707, "grad_norm": 0.62109375, "learning_rate": 5.8445474704081484e-05, "loss": 0.0853, "step": 22754 }, { "epoch": 16.67032967032967, "grad_norm": 0.7890625, "learning_rate": 5.844209275175005e-05, "loss": 0.1058, "step": 22755 }, { "epoch": 16.671062271062272, "grad_norm": 0.890625, "learning_rate": 5.843871076136541e-05, "loss": 0.1184, "step": 22756 }, { "epoch": 16.671794871794873, "grad_norm": 1.109375, "learning_rate": 5.8435328732943826e-05, "loss": 0.1356, "step": 22757 }, { "epoch": 16.672527472527474, "grad_norm": 0.7578125, "learning_rate": 5.843194666650146e-05, "loss": 0.0874, "step": 22758 }, { "epoch": 16.673260073260074, "grad_norm": 1.1640625, "learning_rate": 5.8428564562054546e-05, "loss": 0.1062, "step": 22759 }, { "epoch": 16.673992673992675, "grad_norm": 1.265625, "learning_rate": 5.842518241961927e-05, "loss": 0.1681, "step": 22760 }, { "epoch": 16.674725274725276, "grad_norm": 1.0390625, "learning_rate": 5.8421800239211845e-05, "loss": 0.117, "step": 22761 }, { "epoch": 16.675457875457877, "grad_norm": 1.828125, "learning_rate": 5.8418418020848475e-05, "loss": 0.163, "step": 22762 }, { "epoch": 16.676190476190477, "grad_norm": 0.78125, "learning_rate": 5.841503576454537e-05, "loss": 0.1204, "step": 22763 }, { "epoch": 16.676923076923078, "grad_norm": 0.96875, "learning_rate": 5.841165347031873e-05, "loss": 0.1127, "step": 22764 }, { "epoch": 16.67765567765568, "grad_norm": 0.86328125, "learning_rate": 5.8408271138184754e-05, "loss": 0.1244, "step": 22765 }, { "epoch": 16.67838827838828, "grad_norm": 0.98046875, "learning_rate": 5.8404888768159646e-05, "loss": 0.1326, "step": 22766 }, { "epoch": 16.67912087912088, "grad_norm": 1.0390625, "learning_rate": 5.8401506360259636e-05, "loss": 0.1533, "step": 22767 }, { "epoch": 16.67985347985348, "grad_norm": 1.65625, "learning_rate": 5.839812391450091e-05, "loss": 0.1586, "step": 22768 }, { "epoch": 16.68058608058608, "grad_norm": 0.8125, "learning_rate": 5.839474143089968e-05, "loss": 0.1389, "step": 22769 }, { "epoch": 16.681318681318682, "grad_norm": 1.109375, "learning_rate": 5.839135890947215e-05, "loss": 0.1169, "step": 22770 }, { "epoch": 16.682051282051283, "grad_norm": 1.0703125, "learning_rate": 5.838797635023453e-05, "loss": 0.1285, "step": 22771 }, { "epoch": 16.682783882783884, "grad_norm": 1.5234375, "learning_rate": 5.8384593753203046e-05, "loss": 0.1202, "step": 22772 }, { "epoch": 16.683516483516485, "grad_norm": 0.76953125, "learning_rate": 5.838121111839387e-05, "loss": 0.1076, "step": 22773 }, { "epoch": 16.684249084249085, "grad_norm": 0.70703125, "learning_rate": 5.8377828445823214e-05, "loss": 0.1274, "step": 22774 }, { "epoch": 16.684981684981686, "grad_norm": 0.59375, "learning_rate": 5.837444573550731e-05, "loss": 0.1074, "step": 22775 }, { "epoch": 16.685714285714287, "grad_norm": 1.0390625, "learning_rate": 5.837106298746233e-05, "loss": 0.1029, "step": 22776 }, { "epoch": 16.686446886446888, "grad_norm": 1.90625, "learning_rate": 5.8367680201704524e-05, "loss": 0.1384, "step": 22777 }, { "epoch": 16.68717948717949, "grad_norm": 1.2265625, "learning_rate": 5.836429737825007e-05, "loss": 0.1131, "step": 22778 }, { "epoch": 16.68791208791209, "grad_norm": 1.96875, "learning_rate": 5.836091451711517e-05, "loss": 0.1253, "step": 22779 }, { "epoch": 16.68864468864469, "grad_norm": 1.171875, "learning_rate": 5.835753161831608e-05, "loss": 0.122, "step": 22780 }, { "epoch": 16.68937728937729, "grad_norm": 1.125, "learning_rate": 5.835414868186897e-05, "loss": 0.1218, "step": 22781 }, { "epoch": 16.69010989010989, "grad_norm": 1.015625, "learning_rate": 5.835076570779004e-05, "loss": 0.1313, "step": 22782 }, { "epoch": 16.690842490842492, "grad_norm": 1.453125, "learning_rate": 5.834738269609553e-05, "loss": 0.1214, "step": 22783 }, { "epoch": 16.691575091575093, "grad_norm": 1.703125, "learning_rate": 5.8343999646801616e-05, "loss": 0.1727, "step": 22784 }, { "epoch": 16.692307692307693, "grad_norm": 1.1328125, "learning_rate": 5.834061655992454e-05, "loss": 0.1113, "step": 22785 }, { "epoch": 16.693040293040294, "grad_norm": 0.9609375, "learning_rate": 5.83372334354805e-05, "loss": 0.1059, "step": 22786 }, { "epoch": 16.693772893772895, "grad_norm": 0.8671875, "learning_rate": 5.833385027348569e-05, "loss": 0.1058, "step": 22787 }, { "epoch": 16.694505494505496, "grad_norm": 0.90234375, "learning_rate": 5.833046707395634e-05, "loss": 0.1112, "step": 22788 }, { "epoch": 16.695238095238096, "grad_norm": 1.03125, "learning_rate": 5.832708383690863e-05, "loss": 0.1531, "step": 22789 }, { "epoch": 16.695970695970697, "grad_norm": 0.890625, "learning_rate": 5.8323700562358826e-05, "loss": 0.1354, "step": 22790 }, { "epoch": 16.696703296703298, "grad_norm": 1.3203125, "learning_rate": 5.8320317250323075e-05, "loss": 0.1399, "step": 22791 }, { "epoch": 16.6974358974359, "grad_norm": 1.296875, "learning_rate": 5.831693390081763e-05, "loss": 0.1378, "step": 22792 }, { "epoch": 16.6981684981685, "grad_norm": 1.265625, "learning_rate": 5.8313550513858684e-05, "loss": 0.0955, "step": 22793 }, { "epoch": 16.6989010989011, "grad_norm": 1.3828125, "learning_rate": 5.831016708946247e-05, "loss": 0.1267, "step": 22794 }, { "epoch": 16.6996336996337, "grad_norm": 0.984375, "learning_rate": 5.830678362764515e-05, "loss": 0.1623, "step": 22795 }, { "epoch": 16.7003663003663, "grad_norm": 1.5625, "learning_rate": 5.830340012842299e-05, "loss": 0.1027, "step": 22796 }, { "epoch": 16.701098901098902, "grad_norm": 1.2109375, "learning_rate": 5.830001659181217e-05, "loss": 0.1147, "step": 22797 }, { "epoch": 16.701831501831503, "grad_norm": 1.0859375, "learning_rate": 5.8296633017828916e-05, "loss": 0.1085, "step": 22798 }, { "epoch": 16.702564102564104, "grad_norm": 1.234375, "learning_rate": 5.829324940648944e-05, "loss": 0.11, "step": 22799 }, { "epoch": 16.703296703296704, "grad_norm": 1.1875, "learning_rate": 5.828986575780992e-05, "loss": 0.1486, "step": 22800 }, { "epoch": 16.704029304029305, "grad_norm": 1.21875, "learning_rate": 5.8286482071806624e-05, "loss": 0.1318, "step": 22801 }, { "epoch": 16.704761904761906, "grad_norm": 1.1640625, "learning_rate": 5.828309834849573e-05, "loss": 0.1191, "step": 22802 }, { "epoch": 16.705494505494507, "grad_norm": 0.80859375, "learning_rate": 5.8279714587893444e-05, "loss": 0.1051, "step": 22803 }, { "epoch": 16.706227106227107, "grad_norm": 1.46875, "learning_rate": 5.8276330790016e-05, "loss": 0.1672, "step": 22804 }, { "epoch": 16.706959706959708, "grad_norm": 1.5703125, "learning_rate": 5.827294695487959e-05, "loss": 0.1519, "step": 22805 }, { "epoch": 16.70769230769231, "grad_norm": 1.5546875, "learning_rate": 5.8269563082500454e-05, "loss": 0.1399, "step": 22806 }, { "epoch": 16.70842490842491, "grad_norm": 1.4140625, "learning_rate": 5.82661791728948e-05, "loss": 0.1174, "step": 22807 }, { "epoch": 16.70915750915751, "grad_norm": 1.4453125, "learning_rate": 5.8262795226078806e-05, "loss": 0.1081, "step": 22808 }, { "epoch": 16.70989010989011, "grad_norm": 1.0390625, "learning_rate": 5.825941124206874e-05, "loss": 0.1314, "step": 22809 }, { "epoch": 16.71062271062271, "grad_norm": 1.0, "learning_rate": 5.825602722088076e-05, "loss": 0.1489, "step": 22810 }, { "epoch": 16.711355311355312, "grad_norm": 1.0390625, "learning_rate": 5.825264316253113e-05, "loss": 0.1214, "step": 22811 }, { "epoch": 16.712087912087913, "grad_norm": 0.96875, "learning_rate": 5.824925906703603e-05, "loss": 0.1453, "step": 22812 }, { "epoch": 16.712820512820514, "grad_norm": 1.15625, "learning_rate": 5.8245874934411693e-05, "loss": 0.1144, "step": 22813 }, { "epoch": 16.713553113553115, "grad_norm": 0.625, "learning_rate": 5.824249076467432e-05, "loss": 0.0976, "step": 22814 }, { "epoch": 16.714285714285715, "grad_norm": 1.078125, "learning_rate": 5.823910655784014e-05, "loss": 0.1389, "step": 22815 }, { "epoch": 16.715018315018316, "grad_norm": 0.91015625, "learning_rate": 5.823572231392535e-05, "loss": 0.1199, "step": 22816 }, { "epoch": 16.715750915750917, "grad_norm": 1.03125, "learning_rate": 5.823233803294618e-05, "loss": 0.1043, "step": 22817 }, { "epoch": 16.716483516483517, "grad_norm": 1.546875, "learning_rate": 5.822895371491882e-05, "loss": 0.1432, "step": 22818 }, { "epoch": 16.717216117216118, "grad_norm": 0.8984375, "learning_rate": 5.822556935985953e-05, "loss": 0.1179, "step": 22819 }, { "epoch": 16.71794871794872, "grad_norm": 1.1796875, "learning_rate": 5.822218496778451e-05, "loss": 0.1285, "step": 22820 }, { "epoch": 16.71868131868132, "grad_norm": 0.7578125, "learning_rate": 5.821880053870994e-05, "loss": 0.1028, "step": 22821 }, { "epoch": 16.71941391941392, "grad_norm": 0.90234375, "learning_rate": 5.8215416072652085e-05, "loss": 0.1421, "step": 22822 }, { "epoch": 16.72014652014652, "grad_norm": 0.9375, "learning_rate": 5.8212031569627125e-05, "loss": 0.1204, "step": 22823 }, { "epoch": 16.720879120879122, "grad_norm": 1.0546875, "learning_rate": 5.820864702965129e-05, "loss": 0.1217, "step": 22824 }, { "epoch": 16.721611721611723, "grad_norm": 0.62109375, "learning_rate": 5.8205262452740804e-05, "loss": 0.1165, "step": 22825 }, { "epoch": 16.722344322344323, "grad_norm": 1.21875, "learning_rate": 5.820187783891188e-05, "loss": 0.113, "step": 22826 }, { "epoch": 16.723076923076924, "grad_norm": 1.0703125, "learning_rate": 5.819849318818071e-05, "loss": 0.1257, "step": 22827 }, { "epoch": 16.723809523809525, "grad_norm": 1.40625, "learning_rate": 5.819510850056354e-05, "loss": 0.1378, "step": 22828 }, { "epoch": 16.724542124542126, "grad_norm": 1.0703125, "learning_rate": 5.81917237760766e-05, "loss": 0.1265, "step": 22829 }, { "epoch": 16.725274725274726, "grad_norm": 0.75, "learning_rate": 5.818833901473606e-05, "loss": 0.1178, "step": 22830 }, { "epoch": 16.726007326007327, "grad_norm": 1.625, "learning_rate": 5.8184954216558176e-05, "loss": 0.1135, "step": 22831 }, { "epoch": 16.726739926739928, "grad_norm": 0.94140625, "learning_rate": 5.818156938155915e-05, "loss": 0.1105, "step": 22832 }, { "epoch": 16.72747252747253, "grad_norm": 0.9765625, "learning_rate": 5.8178184509755216e-05, "loss": 0.13, "step": 22833 }, { "epoch": 16.72820512820513, "grad_norm": 1.2734375, "learning_rate": 5.8174799601162554e-05, "loss": 0.1203, "step": 22834 }, { "epoch": 16.72893772893773, "grad_norm": 0.82421875, "learning_rate": 5.8171414655797426e-05, "loss": 0.1341, "step": 22835 }, { "epoch": 16.72967032967033, "grad_norm": 1.1796875, "learning_rate": 5.816802967367603e-05, "loss": 0.1202, "step": 22836 }, { "epoch": 16.73040293040293, "grad_norm": 1.0546875, "learning_rate": 5.816464465481459e-05, "loss": 0.1362, "step": 22837 }, { "epoch": 16.731135531135532, "grad_norm": 0.79296875, "learning_rate": 5.816125959922932e-05, "loss": 0.0998, "step": 22838 }, { "epoch": 16.731868131868133, "grad_norm": 1.3671875, "learning_rate": 5.815787450693643e-05, "loss": 0.1417, "step": 22839 }, { "epoch": 16.732600732600734, "grad_norm": 1.078125, "learning_rate": 5.815448937795216e-05, "loss": 0.1251, "step": 22840 }, { "epoch": 16.733333333333334, "grad_norm": 1.015625, "learning_rate": 5.8151104212292705e-05, "loss": 0.1018, "step": 22841 }, { "epoch": 16.734065934065935, "grad_norm": 1.3203125, "learning_rate": 5.814771900997432e-05, "loss": 0.1508, "step": 22842 }, { "epoch": 16.734798534798536, "grad_norm": 1.171875, "learning_rate": 5.814433377101318e-05, "loss": 0.1547, "step": 22843 }, { "epoch": 16.735531135531136, "grad_norm": 1.0625, "learning_rate": 5.814094849542553e-05, "loss": 0.1284, "step": 22844 }, { "epoch": 16.736263736263737, "grad_norm": 1.921875, "learning_rate": 5.813756318322761e-05, "loss": 0.1485, "step": 22845 }, { "epoch": 16.736996336996338, "grad_norm": 0.9453125, "learning_rate": 5.81341778344356e-05, "loss": 0.1182, "step": 22846 }, { "epoch": 16.73772893772894, "grad_norm": 1.0078125, "learning_rate": 5.813079244906574e-05, "loss": 0.1084, "step": 22847 }, { "epoch": 16.73846153846154, "grad_norm": 1.28125, "learning_rate": 5.812740702713425e-05, "loss": 0.1308, "step": 22848 }, { "epoch": 16.73919413919414, "grad_norm": 1.203125, "learning_rate": 5.8124021568657336e-05, "loss": 0.104, "step": 22849 }, { "epoch": 16.73992673992674, "grad_norm": 1.2578125, "learning_rate": 5.8120636073651255e-05, "loss": 0.1379, "step": 22850 }, { "epoch": 16.74065934065934, "grad_norm": 0.8203125, "learning_rate": 5.811725054213219e-05, "loss": 0.1188, "step": 22851 }, { "epoch": 16.741391941391942, "grad_norm": 0.82421875, "learning_rate": 5.8113864974116384e-05, "loss": 0.1385, "step": 22852 }, { "epoch": 16.742124542124543, "grad_norm": 1.40625, "learning_rate": 5.811047936962005e-05, "loss": 0.1325, "step": 22853 }, { "epoch": 16.742857142857144, "grad_norm": 1.1015625, "learning_rate": 5.8107093728659414e-05, "loss": 0.1116, "step": 22854 }, { "epoch": 16.743589743589745, "grad_norm": 1.1796875, "learning_rate": 5.810370805125068e-05, "loss": 0.1015, "step": 22855 }, { "epoch": 16.744322344322345, "grad_norm": 1.3125, "learning_rate": 5.8100322337410096e-05, "loss": 0.1299, "step": 22856 }, { "epoch": 16.745054945054946, "grad_norm": 1.453125, "learning_rate": 5.809693658715386e-05, "loss": 0.1044, "step": 22857 }, { "epoch": 16.745787545787547, "grad_norm": 1.3671875, "learning_rate": 5.809355080049822e-05, "loss": 0.1213, "step": 22858 }, { "epoch": 16.746520146520147, "grad_norm": 0.90234375, "learning_rate": 5.80901649774594e-05, "loss": 0.1512, "step": 22859 }, { "epoch": 16.747252747252748, "grad_norm": 0.953125, "learning_rate": 5.808677911805358e-05, "loss": 0.108, "step": 22860 }, { "epoch": 16.74798534798535, "grad_norm": 0.84375, "learning_rate": 5.808339322229702e-05, "loss": 0.1184, "step": 22861 }, { "epoch": 16.74871794871795, "grad_norm": 0.84765625, "learning_rate": 5.808000729020593e-05, "loss": 0.0943, "step": 22862 }, { "epoch": 16.74945054945055, "grad_norm": 1.9921875, "learning_rate": 5.8076621321796554e-05, "loss": 0.1402, "step": 22863 }, { "epoch": 16.75018315018315, "grad_norm": 1.6875, "learning_rate": 5.80732353170851e-05, "loss": 0.1515, "step": 22864 }, { "epoch": 16.750915750915752, "grad_norm": 1.046875, "learning_rate": 5.806984927608775e-05, "loss": 0.1389, "step": 22865 }, { "epoch": 16.751648351648353, "grad_norm": 1.2265625, "learning_rate": 5.806646319882081e-05, "loss": 0.1205, "step": 22866 }, { "epoch": 16.752380952380953, "grad_norm": 0.953125, "learning_rate": 5.8063077085300446e-05, "loss": 0.128, "step": 22867 }, { "epoch": 16.753113553113554, "grad_norm": 0.71875, "learning_rate": 5.80596909355429e-05, "loss": 0.1307, "step": 22868 }, { "epoch": 16.753846153846155, "grad_norm": 0.6171875, "learning_rate": 5.805630474956437e-05, "loss": 0.1372, "step": 22869 }, { "epoch": 16.754578754578755, "grad_norm": 0.94921875, "learning_rate": 5.8052918527381114e-05, "loss": 0.1303, "step": 22870 }, { "epoch": 16.755311355311356, "grad_norm": 1.0703125, "learning_rate": 5.804953226900936e-05, "loss": 0.113, "step": 22871 }, { "epoch": 16.756043956043957, "grad_norm": 1.1640625, "learning_rate": 5.8046145974465317e-05, "loss": 0.1144, "step": 22872 }, { "epoch": 16.756776556776558, "grad_norm": 0.9609375, "learning_rate": 5.80427596437652e-05, "loss": 0.1, "step": 22873 }, { "epoch": 16.75750915750916, "grad_norm": 1.203125, "learning_rate": 5.803937327692524e-05, "loss": 0.1327, "step": 22874 }, { "epoch": 16.75824175824176, "grad_norm": 0.9296875, "learning_rate": 5.803598687396169e-05, "loss": 0.0954, "step": 22875 }, { "epoch": 16.75897435897436, "grad_norm": 1.078125, "learning_rate": 5.803260043489074e-05, "loss": 0.119, "step": 22876 }, { "epoch": 16.75970695970696, "grad_norm": 1.28125, "learning_rate": 5.802921395972865e-05, "loss": 0.123, "step": 22877 }, { "epoch": 16.76043956043956, "grad_norm": 0.62890625, "learning_rate": 5.802582744849159e-05, "loss": 0.099, "step": 22878 }, { "epoch": 16.761172161172162, "grad_norm": 0.8515625, "learning_rate": 5.802244090119584e-05, "loss": 0.1235, "step": 22879 }, { "epoch": 16.761904761904763, "grad_norm": 0.8515625, "learning_rate": 5.801905431785759e-05, "loss": 0.0804, "step": 22880 }, { "epoch": 16.762637362637363, "grad_norm": 1.2265625, "learning_rate": 5.801566769849309e-05, "loss": 0.1186, "step": 22881 }, { "epoch": 16.763369963369964, "grad_norm": 0.953125, "learning_rate": 5.8012281043118566e-05, "loss": 0.0912, "step": 22882 }, { "epoch": 16.764102564102565, "grad_norm": 1.2734375, "learning_rate": 5.8008894351750235e-05, "loss": 0.1373, "step": 22883 }, { "epoch": 16.764835164835166, "grad_norm": 1.140625, "learning_rate": 5.8005507624404335e-05, "loss": 0.1182, "step": 22884 }, { "epoch": 16.765567765567766, "grad_norm": 1.546875, "learning_rate": 5.800212086109706e-05, "loss": 0.1336, "step": 22885 }, { "epoch": 16.766300366300367, "grad_norm": 1.203125, "learning_rate": 5.799873406184467e-05, "loss": 0.1344, "step": 22886 }, { "epoch": 16.767032967032968, "grad_norm": 0.98828125, "learning_rate": 5.79953472266634e-05, "loss": 0.1483, "step": 22887 }, { "epoch": 16.76776556776557, "grad_norm": 1.6328125, "learning_rate": 5.799196035556944e-05, "loss": 0.1373, "step": 22888 }, { "epoch": 16.76849816849817, "grad_norm": 1.1015625, "learning_rate": 5.7988573448579056e-05, "loss": 0.1304, "step": 22889 }, { "epoch": 16.76923076923077, "grad_norm": 1.140625, "learning_rate": 5.7985186505708455e-05, "loss": 0.1253, "step": 22890 }, { "epoch": 16.76996336996337, "grad_norm": 0.76171875, "learning_rate": 5.7981799526973855e-05, "loss": 0.1074, "step": 22891 }, { "epoch": 16.77069597069597, "grad_norm": 0.83203125, "learning_rate": 5.7978412512391514e-05, "loss": 0.1142, "step": 22892 }, { "epoch": 16.771428571428572, "grad_norm": 1.125, "learning_rate": 5.7975025461977626e-05, "loss": 0.1268, "step": 22893 }, { "epoch": 16.772161172161173, "grad_norm": 1.015625, "learning_rate": 5.797163837574844e-05, "loss": 0.1372, "step": 22894 }, { "epoch": 16.772893772893774, "grad_norm": 0.7890625, "learning_rate": 5.796825125372019e-05, "loss": 0.1248, "step": 22895 }, { "epoch": 16.773626373626374, "grad_norm": 1.2734375, "learning_rate": 5.7964864095909086e-05, "loss": 0.1453, "step": 22896 }, { "epoch": 16.774358974358975, "grad_norm": 0.875, "learning_rate": 5.796147690233139e-05, "loss": 0.1211, "step": 22897 }, { "epoch": 16.775091575091576, "grad_norm": 0.76171875, "learning_rate": 5.795808967300329e-05, "loss": 0.1334, "step": 22898 }, { "epoch": 16.775824175824177, "grad_norm": 1.09375, "learning_rate": 5.7954702407941043e-05, "loss": 0.1044, "step": 22899 }, { "epoch": 16.776556776556777, "grad_norm": 1.4375, "learning_rate": 5.795131510716087e-05, "loss": 0.101, "step": 22900 }, { "epoch": 16.777289377289378, "grad_norm": 1.1640625, "learning_rate": 5.794792777067899e-05, "loss": 0.129, "step": 22901 }, { "epoch": 16.77802197802198, "grad_norm": 0.96484375, "learning_rate": 5.794454039851165e-05, "loss": 0.1371, "step": 22902 }, { "epoch": 16.77875457875458, "grad_norm": 0.76953125, "learning_rate": 5.7941152990675073e-05, "loss": 0.1147, "step": 22903 }, { "epoch": 16.77948717948718, "grad_norm": 1.1171875, "learning_rate": 5.793776554718548e-05, "loss": 0.1429, "step": 22904 }, { "epoch": 16.78021978021978, "grad_norm": 1.1875, "learning_rate": 5.7934378068059137e-05, "loss": 0.1148, "step": 22905 }, { "epoch": 16.78095238095238, "grad_norm": 1.1015625, "learning_rate": 5.793099055331223e-05, "loss": 0.111, "step": 22906 }, { "epoch": 16.781684981684982, "grad_norm": 1.1171875, "learning_rate": 5.7927603002960996e-05, "loss": 0.1484, "step": 22907 }, { "epoch": 16.782417582417583, "grad_norm": 1.1015625, "learning_rate": 5.792421541702169e-05, "loss": 0.1271, "step": 22908 }, { "epoch": 16.783150183150184, "grad_norm": 1.25, "learning_rate": 5.7920827795510525e-05, "loss": 0.1448, "step": 22909 }, { "epoch": 16.783882783882785, "grad_norm": 0.91015625, "learning_rate": 5.7917440138443746e-05, "loss": 0.1037, "step": 22910 }, { "epoch": 16.784615384615385, "grad_norm": 1.1015625, "learning_rate": 5.7914052445837585e-05, "loss": 0.0982, "step": 22911 }, { "epoch": 16.785347985347986, "grad_norm": 1.2265625, "learning_rate": 5.791066471770824e-05, "loss": 0.1359, "step": 22912 }, { "epoch": 16.786080586080587, "grad_norm": 0.76171875, "learning_rate": 5.790727695407199e-05, "loss": 0.1183, "step": 22913 }, { "epoch": 16.786813186813188, "grad_norm": 0.87890625, "learning_rate": 5.7903889154945036e-05, "loss": 0.1309, "step": 22914 }, { "epoch": 16.78754578754579, "grad_norm": 1.5859375, "learning_rate": 5.7900501320343614e-05, "loss": 0.1309, "step": 22915 }, { "epoch": 16.78827838827839, "grad_norm": 0.8046875, "learning_rate": 5.789711345028397e-05, "loss": 0.0904, "step": 22916 }, { "epoch": 16.78901098901099, "grad_norm": 1.3359375, "learning_rate": 5.7893725544782304e-05, "loss": 0.1229, "step": 22917 }, { "epoch": 16.78974358974359, "grad_norm": 1.6171875, "learning_rate": 5.78903376038549e-05, "loss": 0.149, "step": 22918 }, { "epoch": 16.79047619047619, "grad_norm": 1.1171875, "learning_rate": 5.788694962751795e-05, "loss": 0.1185, "step": 22919 }, { "epoch": 16.791208791208792, "grad_norm": 0.73828125, "learning_rate": 5.788356161578768e-05, "loss": 0.1302, "step": 22920 }, { "epoch": 16.791941391941393, "grad_norm": 0.91796875, "learning_rate": 5.788017356868037e-05, "loss": 0.1354, "step": 22921 }, { "epoch": 16.792673992673993, "grad_norm": 1.828125, "learning_rate": 5.7876785486212226e-05, "loss": 0.118, "step": 22922 }, { "epoch": 16.793406593406594, "grad_norm": 1.234375, "learning_rate": 5.7873397368399476e-05, "loss": 0.1193, "step": 22923 }, { "epoch": 16.794139194139195, "grad_norm": 0.8515625, "learning_rate": 5.787000921525837e-05, "loss": 0.1185, "step": 22924 }, { "epoch": 16.794871794871796, "grad_norm": 1.3515625, "learning_rate": 5.78666210268051e-05, "loss": 0.1584, "step": 22925 }, { "epoch": 16.795604395604396, "grad_norm": 0.8671875, "learning_rate": 5.786323280305596e-05, "loss": 0.1222, "step": 22926 }, { "epoch": 16.796336996336997, "grad_norm": 1.1328125, "learning_rate": 5.785984454402713e-05, "loss": 0.0879, "step": 22927 }, { "epoch": 16.797069597069598, "grad_norm": 1.40625, "learning_rate": 5.785645624973489e-05, "loss": 0.115, "step": 22928 }, { "epoch": 16.7978021978022, "grad_norm": 0.83984375, "learning_rate": 5.7853067920195445e-05, "loss": 0.1202, "step": 22929 }, { "epoch": 16.7985347985348, "grad_norm": 1.1171875, "learning_rate": 5.784967955542504e-05, "loss": 0.1503, "step": 22930 }, { "epoch": 16.7992673992674, "grad_norm": 1.53125, "learning_rate": 5.78462911554399e-05, "loss": 0.1338, "step": 22931 }, { "epoch": 16.8, "grad_norm": 0.80078125, "learning_rate": 5.784290272025627e-05, "loss": 0.1336, "step": 22932 }, { "epoch": 16.8007326007326, "grad_norm": 1.078125, "learning_rate": 5.783951424989038e-05, "loss": 0.136, "step": 22933 }, { "epoch": 16.801465201465202, "grad_norm": 1.234375, "learning_rate": 5.783612574435847e-05, "loss": 0.1305, "step": 22934 }, { "epoch": 16.802197802197803, "grad_norm": 1.4140625, "learning_rate": 5.783273720367678e-05, "loss": 0.1554, "step": 22935 }, { "epoch": 16.802930402930404, "grad_norm": 0.87109375, "learning_rate": 5.7829348627861536e-05, "loss": 0.1077, "step": 22936 }, { "epoch": 16.803663003663004, "grad_norm": 0.9453125, "learning_rate": 5.782596001692898e-05, "loss": 0.1081, "step": 22937 }, { "epoch": 16.804395604395605, "grad_norm": 1.1796875, "learning_rate": 5.782257137089534e-05, "loss": 0.1194, "step": 22938 }, { "epoch": 16.805128205128206, "grad_norm": 0.875, "learning_rate": 5.7819182689776864e-05, "loss": 0.101, "step": 22939 }, { "epoch": 16.805860805860807, "grad_norm": 0.455078125, "learning_rate": 5.781579397358976e-05, "loss": 0.0999, "step": 22940 }, { "epoch": 16.806593406593407, "grad_norm": 1.515625, "learning_rate": 5.7812405222350305e-05, "loss": 0.1347, "step": 22941 }, { "epoch": 16.807326007326008, "grad_norm": 1.078125, "learning_rate": 5.7809016436074705e-05, "loss": 0.1121, "step": 22942 }, { "epoch": 16.80805860805861, "grad_norm": 0.98828125, "learning_rate": 5.780562761477921e-05, "loss": 0.1159, "step": 22943 }, { "epoch": 16.80879120879121, "grad_norm": 0.9453125, "learning_rate": 5.780223875848007e-05, "loss": 0.1189, "step": 22944 }, { "epoch": 16.80952380952381, "grad_norm": 0.87109375, "learning_rate": 5.7798849867193484e-05, "loss": 0.1366, "step": 22945 }, { "epoch": 16.81025641025641, "grad_norm": 1.1640625, "learning_rate": 5.779546094093572e-05, "loss": 0.135, "step": 22946 }, { "epoch": 16.81098901098901, "grad_norm": 1.234375, "learning_rate": 5.7792071979723016e-05, "loss": 0.1204, "step": 22947 }, { "epoch": 16.811721611721612, "grad_norm": 1.0234375, "learning_rate": 5.7788682983571584e-05, "loss": 0.1199, "step": 22948 }, { "epoch": 16.812454212454213, "grad_norm": 0.82421875, "learning_rate": 5.77852939524977e-05, "loss": 0.091, "step": 22949 }, { "epoch": 16.813186813186814, "grad_norm": 1.3359375, "learning_rate": 5.778190488651758e-05, "loss": 0.1486, "step": 22950 }, { "epoch": 16.813919413919415, "grad_norm": 1.46875, "learning_rate": 5.7778515785647436e-05, "loss": 0.1344, "step": 22951 }, { "epoch": 16.814652014652015, "grad_norm": 1.3671875, "learning_rate": 5.7775126649903565e-05, "loss": 0.1614, "step": 22952 }, { "epoch": 16.815384615384616, "grad_norm": 0.9609375, "learning_rate": 5.7771737479302154e-05, "loss": 0.1283, "step": 22953 }, { "epoch": 16.816117216117217, "grad_norm": 1.125, "learning_rate": 5.7768348273859486e-05, "loss": 0.1327, "step": 22954 }, { "epoch": 16.816849816849818, "grad_norm": 1.484375, "learning_rate": 5.7764959033591756e-05, "loss": 0.1357, "step": 22955 }, { "epoch": 16.81758241758242, "grad_norm": 1.3359375, "learning_rate": 5.7761569758515214e-05, "loss": 0.1178, "step": 22956 }, { "epoch": 16.81831501831502, "grad_norm": 0.78125, "learning_rate": 5.775818044864613e-05, "loss": 0.1303, "step": 22957 }, { "epoch": 16.81904761904762, "grad_norm": 1.265625, "learning_rate": 5.7754791104000704e-05, "loss": 0.1105, "step": 22958 }, { "epoch": 16.81978021978022, "grad_norm": 0.80078125, "learning_rate": 5.775140172459519e-05, "loss": 0.1187, "step": 22959 }, { "epoch": 16.82051282051282, "grad_norm": 0.95703125, "learning_rate": 5.7748012310445845e-05, "loss": 0.1241, "step": 22960 }, { "epoch": 16.821245421245422, "grad_norm": 1.390625, "learning_rate": 5.7744622861568876e-05, "loss": 0.1241, "step": 22961 }, { "epoch": 16.821978021978023, "grad_norm": 1.328125, "learning_rate": 5.774123337798055e-05, "loss": 0.1161, "step": 22962 }, { "epoch": 16.822710622710623, "grad_norm": 0.94921875, "learning_rate": 5.7737843859697105e-05, "loss": 0.1238, "step": 22963 }, { "epoch": 16.823443223443224, "grad_norm": 1.4609375, "learning_rate": 5.773445430673476e-05, "loss": 0.1409, "step": 22964 }, { "epoch": 16.824175824175825, "grad_norm": 1.84375, "learning_rate": 5.7731064719109785e-05, "loss": 0.1358, "step": 22965 }, { "epoch": 16.824908424908426, "grad_norm": 1.0078125, "learning_rate": 5.7727675096838395e-05, "loss": 0.1397, "step": 22966 }, { "epoch": 16.825641025641026, "grad_norm": 1.0625, "learning_rate": 5.772428543993684e-05, "loss": 0.1361, "step": 22967 }, { "epoch": 16.826373626373627, "grad_norm": 0.87109375, "learning_rate": 5.772089574842137e-05, "loss": 0.1196, "step": 22968 }, { "epoch": 16.827106227106228, "grad_norm": 1.4296875, "learning_rate": 5.771750602230821e-05, "loss": 0.133, "step": 22969 }, { "epoch": 16.82783882783883, "grad_norm": 1.9921875, "learning_rate": 5.7714116261613614e-05, "loss": 0.1494, "step": 22970 }, { "epoch": 16.82857142857143, "grad_norm": 1.7734375, "learning_rate": 5.771072646635382e-05, "loss": 0.1221, "step": 22971 }, { "epoch": 16.82930402930403, "grad_norm": 1.5859375, "learning_rate": 5.770733663654505e-05, "loss": 0.1092, "step": 22972 }, { "epoch": 16.83003663003663, "grad_norm": 1.2734375, "learning_rate": 5.770394677220358e-05, "loss": 0.1512, "step": 22973 }, { "epoch": 16.83076923076923, "grad_norm": 1.15625, "learning_rate": 5.7700556873345634e-05, "loss": 0.1023, "step": 22974 }, { "epoch": 16.831501831501832, "grad_norm": 0.890625, "learning_rate": 5.769716693998747e-05, "loss": 0.0991, "step": 22975 }, { "epoch": 16.832234432234433, "grad_norm": 0.875, "learning_rate": 5.76937769721453e-05, "loss": 0.1215, "step": 22976 }, { "epoch": 16.832967032967034, "grad_norm": 0.88671875, "learning_rate": 5.7690386969835385e-05, "loss": 0.1173, "step": 22977 }, { "epoch": 16.833699633699634, "grad_norm": 1.3046875, "learning_rate": 5.7686996933073965e-05, "loss": 0.1201, "step": 22978 }, { "epoch": 16.834432234432235, "grad_norm": 0.9140625, "learning_rate": 5.7683606861877295e-05, "loss": 0.1344, "step": 22979 }, { "epoch": 16.835164835164836, "grad_norm": 0.67578125, "learning_rate": 5.76802167562616e-05, "loss": 0.1262, "step": 22980 }, { "epoch": 16.835897435897436, "grad_norm": 0.8515625, "learning_rate": 5.767682661624312e-05, "loss": 0.0984, "step": 22981 }, { "epoch": 16.836630036630037, "grad_norm": 0.98828125, "learning_rate": 5.7673436441838116e-05, "loss": 0.1225, "step": 22982 }, { "epoch": 16.837362637362638, "grad_norm": 1.0390625, "learning_rate": 5.7670046233062824e-05, "loss": 0.0959, "step": 22983 }, { "epoch": 16.83809523809524, "grad_norm": 0.9609375, "learning_rate": 5.766665598993348e-05, "loss": 0.0988, "step": 22984 }, { "epoch": 16.83882783882784, "grad_norm": 0.7890625, "learning_rate": 5.766326571246634e-05, "loss": 0.1039, "step": 22985 }, { "epoch": 16.83956043956044, "grad_norm": 1.21875, "learning_rate": 5.7659875400677645e-05, "loss": 0.1258, "step": 22986 }, { "epoch": 16.84029304029304, "grad_norm": 1.109375, "learning_rate": 5.765648505458362e-05, "loss": 0.1318, "step": 22987 }, { "epoch": 16.84102564102564, "grad_norm": 0.875, "learning_rate": 5.7653094674200545e-05, "loss": 0.1214, "step": 22988 }, { "epoch": 16.841758241758242, "grad_norm": 0.9140625, "learning_rate": 5.7649704259544646e-05, "loss": 0.1163, "step": 22989 }, { "epoch": 16.842490842490843, "grad_norm": 0.609375, "learning_rate": 5.764631381063216e-05, "loss": 0.1319, "step": 22990 }, { "epoch": 16.843223443223444, "grad_norm": 1.1015625, "learning_rate": 5.764292332747933e-05, "loss": 0.113, "step": 22991 }, { "epoch": 16.843956043956045, "grad_norm": 0.80859375, "learning_rate": 5.7639532810102416e-05, "loss": 0.1082, "step": 22992 }, { "epoch": 16.844688644688645, "grad_norm": 1.3984375, "learning_rate": 5.763614225851766e-05, "loss": 0.1397, "step": 22993 }, { "epoch": 16.845421245421246, "grad_norm": 0.89453125, "learning_rate": 5.76327516727413e-05, "loss": 0.0959, "step": 22994 }, { "epoch": 16.846153846153847, "grad_norm": 1.0078125, "learning_rate": 5.76293610527896e-05, "loss": 0.1505, "step": 22995 }, { "epoch": 16.846886446886447, "grad_norm": 0.8984375, "learning_rate": 5.762597039867876e-05, "loss": 0.1249, "step": 22996 }, { "epoch": 16.847619047619048, "grad_norm": 1.5234375, "learning_rate": 5.7622579710425075e-05, "loss": 0.1423, "step": 22997 }, { "epoch": 16.84835164835165, "grad_norm": 1.234375, "learning_rate": 5.761918898804477e-05, "loss": 0.1081, "step": 22998 }, { "epoch": 16.84908424908425, "grad_norm": 1.1953125, "learning_rate": 5.7615798231554094e-05, "loss": 0.0911, "step": 22999 }, { "epoch": 16.84981684981685, "grad_norm": 0.96484375, "learning_rate": 5.761240744096929e-05, "loss": 0.1357, "step": 23000 }, { "epoch": 16.85054945054945, "grad_norm": 1.7109375, "learning_rate": 5.760901661630662e-05, "loss": 0.1265, "step": 23001 }, { "epoch": 16.851282051282052, "grad_norm": 0.8671875, "learning_rate": 5.7605625757582315e-05, "loss": 0.1304, "step": 23002 }, { "epoch": 16.852014652014653, "grad_norm": 0.8984375, "learning_rate": 5.760223486481261e-05, "loss": 0.119, "step": 23003 }, { "epoch": 16.852747252747253, "grad_norm": 1.296875, "learning_rate": 5.759884393801378e-05, "loss": 0.1488, "step": 23004 }, { "epoch": 16.853479853479854, "grad_norm": 1.4296875, "learning_rate": 5.7595452977202046e-05, "loss": 0.135, "step": 23005 }, { "epoch": 16.854212454212455, "grad_norm": 1.3125, "learning_rate": 5.7592061982393685e-05, "loss": 0.1472, "step": 23006 }, { "epoch": 16.854945054945055, "grad_norm": 1.1015625, "learning_rate": 5.7588670953604916e-05, "loss": 0.143, "step": 23007 }, { "epoch": 16.855677655677656, "grad_norm": 1.015625, "learning_rate": 5.7585279890851996e-05, "loss": 0.118, "step": 23008 }, { "epoch": 16.856410256410257, "grad_norm": 1.2421875, "learning_rate": 5.758188879415116e-05, "loss": 0.1038, "step": 23009 }, { "epoch": 16.857142857142858, "grad_norm": 1.328125, "learning_rate": 5.7578497663518695e-05, "loss": 0.1483, "step": 23010 }, { "epoch": 16.85787545787546, "grad_norm": 1.0234375, "learning_rate": 5.757510649897081e-05, "loss": 0.108, "step": 23011 }, { "epoch": 16.85860805860806, "grad_norm": 1.8671875, "learning_rate": 5.757171530052378e-05, "loss": 0.1611, "step": 23012 }, { "epoch": 16.85934065934066, "grad_norm": 0.88671875, "learning_rate": 5.756832406819381e-05, "loss": 0.1184, "step": 23013 }, { "epoch": 16.86007326007326, "grad_norm": 1.3203125, "learning_rate": 5.756493280199722e-05, "loss": 0.131, "step": 23014 }, { "epoch": 16.86080586080586, "grad_norm": 1.03125, "learning_rate": 5.75615415019502e-05, "loss": 0.1234, "step": 23015 }, { "epoch": 16.861538461538462, "grad_norm": 0.81640625, "learning_rate": 5.755815016806901e-05, "loss": 0.1043, "step": 23016 }, { "epoch": 16.862271062271063, "grad_norm": 0.81640625, "learning_rate": 5.755475880036992e-05, "loss": 0.1298, "step": 23017 }, { "epoch": 16.863003663003663, "grad_norm": 0.9375, "learning_rate": 5.755136739886915e-05, "loss": 0.1167, "step": 23018 }, { "epoch": 16.863736263736264, "grad_norm": 0.79296875, "learning_rate": 5.754797596358297e-05, "loss": 0.1015, "step": 23019 }, { "epoch": 16.864468864468865, "grad_norm": 1.5390625, "learning_rate": 5.754458449452762e-05, "loss": 0.1142, "step": 23020 }, { "epoch": 16.865201465201466, "grad_norm": 0.9140625, "learning_rate": 5.754119299171936e-05, "loss": 0.1413, "step": 23021 }, { "epoch": 16.865934065934066, "grad_norm": 0.71484375, "learning_rate": 5.753780145517441e-05, "loss": 0.1075, "step": 23022 }, { "epoch": 16.866666666666667, "grad_norm": 0.8671875, "learning_rate": 5.7534409884909076e-05, "loss": 0.1075, "step": 23023 }, { "epoch": 16.867399267399268, "grad_norm": 1.046875, "learning_rate": 5.7531018280939556e-05, "loss": 0.1287, "step": 23024 }, { "epoch": 16.86813186813187, "grad_norm": 1.9375, "learning_rate": 5.752762664328213e-05, "loss": 0.1471, "step": 23025 }, { "epoch": 16.86886446886447, "grad_norm": 1.015625, "learning_rate": 5.752423497195303e-05, "loss": 0.1048, "step": 23026 }, { "epoch": 16.86959706959707, "grad_norm": 0.75390625, "learning_rate": 5.752084326696851e-05, "loss": 0.1295, "step": 23027 }, { "epoch": 16.87032967032967, "grad_norm": 1.0859375, "learning_rate": 5.751745152834484e-05, "loss": 0.1439, "step": 23028 }, { "epoch": 16.87106227106227, "grad_norm": 1.0234375, "learning_rate": 5.751405975609825e-05, "loss": 0.1431, "step": 23029 }, { "epoch": 16.871794871794872, "grad_norm": 0.76171875, "learning_rate": 5.751066795024499e-05, "loss": 0.1148, "step": 23030 }, { "epoch": 16.872527472527473, "grad_norm": 1.171875, "learning_rate": 5.750727611080132e-05, "loss": 0.1482, "step": 23031 }, { "epoch": 16.873260073260074, "grad_norm": 1.1796875, "learning_rate": 5.75038842377835e-05, "loss": 0.1092, "step": 23032 }, { "epoch": 16.873992673992674, "grad_norm": 1.296875, "learning_rate": 5.7500492331207766e-05, "loss": 0.1145, "step": 23033 }, { "epoch": 16.874725274725275, "grad_norm": 0.76953125, "learning_rate": 5.749710039109036e-05, "loss": 0.1153, "step": 23034 }, { "epoch": 16.875457875457876, "grad_norm": 1.3203125, "learning_rate": 5.749370841744757e-05, "loss": 0.1326, "step": 23035 }, { "epoch": 16.876190476190477, "grad_norm": 1.421875, "learning_rate": 5.749031641029562e-05, "loss": 0.1256, "step": 23036 }, { "epoch": 16.876923076923077, "grad_norm": 1.09375, "learning_rate": 5.748692436965076e-05, "loss": 0.1235, "step": 23037 }, { "epoch": 16.877655677655678, "grad_norm": 0.7578125, "learning_rate": 5.748353229552926e-05, "loss": 0.12, "step": 23038 }, { "epoch": 16.87838827838828, "grad_norm": 0.94140625, "learning_rate": 5.748014018794736e-05, "loss": 0.1271, "step": 23039 }, { "epoch": 16.87912087912088, "grad_norm": 1.140625, "learning_rate": 5.747674804692134e-05, "loss": 0.1543, "step": 23040 }, { "epoch": 16.87985347985348, "grad_norm": 1.1328125, "learning_rate": 5.747335587246742e-05, "loss": 0.1003, "step": 23041 }, { "epoch": 16.88058608058608, "grad_norm": 0.88671875, "learning_rate": 5.746996366460183e-05, "loss": 0.1045, "step": 23042 }, { "epoch": 16.88131868131868, "grad_norm": 1.3359375, "learning_rate": 5.7466571423340884e-05, "loss": 0.1453, "step": 23043 }, { "epoch": 16.882051282051282, "grad_norm": 1.796875, "learning_rate": 5.746317914870082e-05, "loss": 0.1352, "step": 23044 }, { "epoch": 16.882783882783883, "grad_norm": 1.2265625, "learning_rate": 5.745978684069785e-05, "loss": 0.1285, "step": 23045 }, { "epoch": 16.883516483516484, "grad_norm": 0.87890625, "learning_rate": 5.7456394499348275e-05, "loss": 0.1124, "step": 23046 }, { "epoch": 16.884249084249085, "grad_norm": 0.91015625, "learning_rate": 5.745300212466831e-05, "loss": 0.1265, "step": 23047 }, { "epoch": 16.884981684981685, "grad_norm": 1.1328125, "learning_rate": 5.744960971667425e-05, "loss": 0.1588, "step": 23048 }, { "epoch": 16.885714285714286, "grad_norm": 1.6796875, "learning_rate": 5.7446217275382316e-05, "loss": 0.1468, "step": 23049 }, { "epoch": 16.886446886446887, "grad_norm": 1.1796875, "learning_rate": 5.744282480080878e-05, "loss": 0.1116, "step": 23050 }, { "epoch": 16.887179487179488, "grad_norm": 1.0390625, "learning_rate": 5.74394322929699e-05, "loss": 0.1104, "step": 23051 }, { "epoch": 16.88791208791209, "grad_norm": 0.98046875, "learning_rate": 5.74360397518819e-05, "loss": 0.0851, "step": 23052 }, { "epoch": 16.88864468864469, "grad_norm": 1.296875, "learning_rate": 5.743264717756107e-05, "loss": 0.1441, "step": 23053 }, { "epoch": 16.88937728937729, "grad_norm": 0.6328125, "learning_rate": 5.742925457002366e-05, "loss": 0.1223, "step": 23054 }, { "epoch": 16.89010989010989, "grad_norm": 1.15625, "learning_rate": 5.7425861929285906e-05, "loss": 0.1272, "step": 23055 }, { "epoch": 16.89084249084249, "grad_norm": 0.88671875, "learning_rate": 5.7422469255364086e-05, "loss": 0.1073, "step": 23056 }, { "epoch": 16.891575091575092, "grad_norm": 1.578125, "learning_rate": 5.741907654827443e-05, "loss": 0.1773, "step": 23057 }, { "epoch": 16.892307692307693, "grad_norm": 1.1328125, "learning_rate": 5.7415683808033206e-05, "loss": 0.1209, "step": 23058 }, { "epoch": 16.893040293040293, "grad_norm": 1.390625, "learning_rate": 5.7412291034656686e-05, "loss": 0.1442, "step": 23059 }, { "epoch": 16.893772893772894, "grad_norm": 1.15625, "learning_rate": 5.740889822816109e-05, "loss": 0.1239, "step": 23060 }, { "epoch": 16.894505494505495, "grad_norm": 1.1640625, "learning_rate": 5.7405505388562695e-05, "loss": 0.1074, "step": 23061 }, { "epoch": 16.895238095238096, "grad_norm": 0.8984375, "learning_rate": 5.740211251587778e-05, "loss": 0.1154, "step": 23062 }, { "epoch": 16.895970695970696, "grad_norm": 0.8671875, "learning_rate": 5.739871961012255e-05, "loss": 0.1057, "step": 23063 }, { "epoch": 16.896703296703297, "grad_norm": 0.9140625, "learning_rate": 5.739532667131331e-05, "loss": 0.1, "step": 23064 }, { "epoch": 16.897435897435898, "grad_norm": 0.6953125, "learning_rate": 5.739193369946628e-05, "loss": 0.1151, "step": 23065 }, { "epoch": 16.8981684981685, "grad_norm": 1.1015625, "learning_rate": 5.738854069459776e-05, "loss": 0.1607, "step": 23066 }, { "epoch": 16.8989010989011, "grad_norm": 1.03125, "learning_rate": 5.738514765672396e-05, "loss": 0.1492, "step": 23067 }, { "epoch": 16.8996336996337, "grad_norm": 1.65625, "learning_rate": 5.738175458586115e-05, "loss": 0.1229, "step": 23068 }, { "epoch": 16.9003663003663, "grad_norm": 0.8671875, "learning_rate": 5.737836148202561e-05, "loss": 0.1423, "step": 23069 }, { "epoch": 16.9010989010989, "grad_norm": 1.7265625, "learning_rate": 5.737496834523359e-05, "loss": 0.1406, "step": 23070 }, { "epoch": 16.901831501831502, "grad_norm": 1.6953125, "learning_rate": 5.7371575175501327e-05, "loss": 0.1288, "step": 23071 }, { "epoch": 16.902564102564103, "grad_norm": 1.0859375, "learning_rate": 5.736818197284508e-05, "loss": 0.1346, "step": 23072 }, { "epoch": 16.903296703296704, "grad_norm": 1.3046875, "learning_rate": 5.736478873728113e-05, "loss": 0.1095, "step": 23073 }, { "epoch": 16.904029304029304, "grad_norm": 1.0703125, "learning_rate": 5.736139546882572e-05, "loss": 0.1309, "step": 23074 }, { "epoch": 16.904761904761905, "grad_norm": 1.1953125, "learning_rate": 5.7358002167495115e-05, "loss": 0.1292, "step": 23075 }, { "epoch": 16.905494505494506, "grad_norm": 1.703125, "learning_rate": 5.735460883330557e-05, "loss": 0.161, "step": 23076 }, { "epoch": 16.906227106227107, "grad_norm": 0.75390625, "learning_rate": 5.735121546627334e-05, "loss": 0.1089, "step": 23077 }, { "epoch": 16.906959706959707, "grad_norm": 1.3046875, "learning_rate": 5.734782206641469e-05, "loss": 0.1541, "step": 23078 }, { "epoch": 16.907692307692308, "grad_norm": 0.64453125, "learning_rate": 5.734442863374589e-05, "loss": 0.1227, "step": 23079 }, { "epoch": 16.90842490842491, "grad_norm": 1.5625, "learning_rate": 5.734103516828319e-05, "loss": 0.1298, "step": 23080 }, { "epoch": 16.90915750915751, "grad_norm": 0.9375, "learning_rate": 5.733764167004281e-05, "loss": 0.1399, "step": 23081 }, { "epoch": 16.90989010989011, "grad_norm": 0.91015625, "learning_rate": 5.733424813904107e-05, "loss": 0.13, "step": 23082 }, { "epoch": 16.91062271062271, "grad_norm": 0.55859375, "learning_rate": 5.7330854575294205e-05, "loss": 0.1076, "step": 23083 }, { "epoch": 16.91135531135531, "grad_norm": 0.8125, "learning_rate": 5.7327460978818447e-05, "loss": 0.0895, "step": 23084 }, { "epoch": 16.912087912087912, "grad_norm": 1.328125, "learning_rate": 5.732406734963012e-05, "loss": 0.1291, "step": 23085 }, { "epoch": 16.912820512820513, "grad_norm": 1.125, "learning_rate": 5.732067368774542e-05, "loss": 0.1078, "step": 23086 }, { "epoch": 16.913553113553114, "grad_norm": 1.671875, "learning_rate": 5.7317279993180625e-05, "loss": 0.164, "step": 23087 }, { "epoch": 16.914285714285715, "grad_norm": 1.2109375, "learning_rate": 5.7313886265952023e-05, "loss": 0.1204, "step": 23088 }, { "epoch": 16.915018315018315, "grad_norm": 1.1015625, "learning_rate": 5.731049250607585e-05, "loss": 0.144, "step": 23089 }, { "epoch": 16.915750915750916, "grad_norm": 0.7578125, "learning_rate": 5.730709871356837e-05, "loss": 0.1319, "step": 23090 }, { "epoch": 16.916483516483517, "grad_norm": 1.0078125, "learning_rate": 5.7303704888445854e-05, "loss": 0.1296, "step": 23091 }, { "epoch": 16.917216117216118, "grad_norm": 0.9375, "learning_rate": 5.730031103072455e-05, "loss": 0.1484, "step": 23092 }, { "epoch": 16.91794871794872, "grad_norm": 0.96875, "learning_rate": 5.7296917140420726e-05, "loss": 0.1329, "step": 23093 }, { "epoch": 16.91868131868132, "grad_norm": 0.7421875, "learning_rate": 5.729352321755063e-05, "loss": 0.1231, "step": 23094 }, { "epoch": 16.91941391941392, "grad_norm": 0.87890625, "learning_rate": 5.729012926213054e-05, "loss": 0.1285, "step": 23095 }, { "epoch": 16.92014652014652, "grad_norm": 1.015625, "learning_rate": 5.7286735274176715e-05, "loss": 0.1345, "step": 23096 }, { "epoch": 16.92087912087912, "grad_norm": 1.2421875, "learning_rate": 5.7283341253705415e-05, "loss": 0.1481, "step": 23097 }, { "epoch": 16.921611721611722, "grad_norm": 1.1328125, "learning_rate": 5.7279947200732885e-05, "loss": 0.1215, "step": 23098 }, { "epoch": 16.922344322344323, "grad_norm": 1.40625, "learning_rate": 5.727655311527541e-05, "loss": 0.1244, "step": 23099 }, { "epoch": 16.923076923076923, "grad_norm": 1.2421875, "learning_rate": 5.727315899734924e-05, "loss": 0.0987, "step": 23100 }, { "epoch": 16.923809523809524, "grad_norm": 0.99609375, "learning_rate": 5.7269764846970667e-05, "loss": 0.1142, "step": 23101 }, { "epoch": 16.924542124542125, "grad_norm": 0.91796875, "learning_rate": 5.726637066415589e-05, "loss": 0.1019, "step": 23102 }, { "epoch": 16.925274725274726, "grad_norm": 0.89453125, "learning_rate": 5.726297644892123e-05, "loss": 0.1069, "step": 23103 }, { "epoch": 16.926007326007326, "grad_norm": 1.3828125, "learning_rate": 5.7259582201282933e-05, "loss": 0.1266, "step": 23104 }, { "epoch": 16.926739926739927, "grad_norm": 0.98828125, "learning_rate": 5.7256187921257266e-05, "loss": 0.1174, "step": 23105 }, { "epoch": 16.927472527472528, "grad_norm": 0.6875, "learning_rate": 5.725279360886048e-05, "loss": 0.1319, "step": 23106 }, { "epoch": 16.92820512820513, "grad_norm": 0.87109375, "learning_rate": 5.7249399264108815e-05, "loss": 0.1166, "step": 23107 }, { "epoch": 16.92893772893773, "grad_norm": 1.3046875, "learning_rate": 5.72460048870186e-05, "loss": 0.1296, "step": 23108 }, { "epoch": 16.92967032967033, "grad_norm": 1.2890625, "learning_rate": 5.724261047760604e-05, "loss": 0.1294, "step": 23109 }, { "epoch": 16.93040293040293, "grad_norm": 1.1328125, "learning_rate": 5.723921603588742e-05, "loss": 0.1146, "step": 23110 }, { "epoch": 16.93113553113553, "grad_norm": 1.015625, "learning_rate": 5.7235821561878996e-05, "loss": 0.1391, "step": 23111 }, { "epoch": 16.931868131868132, "grad_norm": 0.74609375, "learning_rate": 5.723242705559706e-05, "loss": 0.1236, "step": 23112 }, { "epoch": 16.932600732600733, "grad_norm": 1.421875, "learning_rate": 5.722903251705786e-05, "loss": 0.1276, "step": 23113 }, { "epoch": 16.933333333333334, "grad_norm": 1.078125, "learning_rate": 5.722563794627764e-05, "loss": 0.1202, "step": 23114 }, { "epoch": 16.934065934065934, "grad_norm": 1.1171875, "learning_rate": 5.7222243343272685e-05, "loss": 0.102, "step": 23115 }, { "epoch": 16.934798534798535, "grad_norm": 1.5546875, "learning_rate": 5.7218848708059255e-05, "loss": 0.1183, "step": 23116 }, { "epoch": 16.935531135531136, "grad_norm": 0.78125, "learning_rate": 5.721545404065361e-05, "loss": 0.1067, "step": 23117 }, { "epoch": 16.936263736263736, "grad_norm": 0.95703125, "learning_rate": 5.7212059341072035e-05, "loss": 0.1022, "step": 23118 }, { "epoch": 16.936996336996337, "grad_norm": 0.9921875, "learning_rate": 5.720866460933078e-05, "loss": 0.0932, "step": 23119 }, { "epoch": 16.937728937728938, "grad_norm": 1.4296875, "learning_rate": 5.720526984544608e-05, "loss": 0.1301, "step": 23120 }, { "epoch": 16.93846153846154, "grad_norm": 1.2890625, "learning_rate": 5.720187504943426e-05, "loss": 0.1231, "step": 23121 }, { "epoch": 16.93919413919414, "grad_norm": 1.03125, "learning_rate": 5.719848022131156e-05, "loss": 0.1219, "step": 23122 }, { "epoch": 16.93992673992674, "grad_norm": 0.87890625, "learning_rate": 5.719508536109423e-05, "loss": 0.1197, "step": 23123 }, { "epoch": 16.94065934065934, "grad_norm": 0.98046875, "learning_rate": 5.719169046879853e-05, "loss": 0.1065, "step": 23124 }, { "epoch": 16.94139194139194, "grad_norm": 0.71484375, "learning_rate": 5.718829554444077e-05, "loss": 0.1094, "step": 23125 }, { "epoch": 16.942124542124542, "grad_norm": 1.3125, "learning_rate": 5.718490058803718e-05, "loss": 0.1339, "step": 23126 }, { "epoch": 16.942857142857143, "grad_norm": 1.453125, "learning_rate": 5.718150559960405e-05, "loss": 0.1287, "step": 23127 }, { "epoch": 16.943589743589744, "grad_norm": 0.83984375, "learning_rate": 5.717811057915762e-05, "loss": 0.109, "step": 23128 }, { "epoch": 16.944322344322345, "grad_norm": 1.640625, "learning_rate": 5.7174715526714185e-05, "loss": 0.1314, "step": 23129 }, { "epoch": 16.945054945054945, "grad_norm": 1.0703125, "learning_rate": 5.717132044228998e-05, "loss": 0.1224, "step": 23130 }, { "epoch": 16.945787545787546, "grad_norm": 1.5078125, "learning_rate": 5.7167925325901294e-05, "loss": 0.1255, "step": 23131 }, { "epoch": 16.946520146520147, "grad_norm": 1.21875, "learning_rate": 5.7164530177564404e-05, "loss": 0.1337, "step": 23132 }, { "epoch": 16.947252747252747, "grad_norm": 0.84375, "learning_rate": 5.716113499729554e-05, "loss": 0.1288, "step": 23133 }, { "epoch": 16.947985347985348, "grad_norm": 1.6796875, "learning_rate": 5.7157739785111e-05, "loss": 0.1382, "step": 23134 }, { "epoch": 16.94871794871795, "grad_norm": 0.63671875, "learning_rate": 5.715434454102704e-05, "loss": 0.1144, "step": 23135 }, { "epoch": 16.94945054945055, "grad_norm": 0.77734375, "learning_rate": 5.7150949265059936e-05, "loss": 0.0961, "step": 23136 }, { "epoch": 16.95018315018315, "grad_norm": 1.703125, "learning_rate": 5.714755395722594e-05, "loss": 0.1519, "step": 23137 }, { "epoch": 16.95091575091575, "grad_norm": 1.28125, "learning_rate": 5.714415861754133e-05, "loss": 0.1474, "step": 23138 }, { "epoch": 16.951648351648352, "grad_norm": 1.1484375, "learning_rate": 5.7140763246022385e-05, "loss": 0.1021, "step": 23139 }, { "epoch": 16.952380952380953, "grad_norm": 0.8515625, "learning_rate": 5.7137367842685374e-05, "loss": 0.1423, "step": 23140 }, { "epoch": 16.953113553113553, "grad_norm": 1.1953125, "learning_rate": 5.7133972407546544e-05, "loss": 0.1294, "step": 23141 }, { "epoch": 16.953846153846154, "grad_norm": 1.109375, "learning_rate": 5.713057694062217e-05, "loss": 0.1463, "step": 23142 }, { "epoch": 16.954578754578755, "grad_norm": 1.0234375, "learning_rate": 5.712718144192852e-05, "loss": 0.1333, "step": 23143 }, { "epoch": 16.955311355311355, "grad_norm": 1.6015625, "learning_rate": 5.7123785911481884e-05, "loss": 0.1095, "step": 23144 }, { "epoch": 16.956043956043956, "grad_norm": 1.0234375, "learning_rate": 5.7120390349298514e-05, "loss": 0.1141, "step": 23145 }, { "epoch": 16.956776556776557, "grad_norm": 1.3515625, "learning_rate": 5.711699475539467e-05, "loss": 0.1086, "step": 23146 }, { "epoch": 16.957509157509158, "grad_norm": 1.046875, "learning_rate": 5.7113599129786636e-05, "loss": 0.1285, "step": 23147 }, { "epoch": 16.95824175824176, "grad_norm": 1.0703125, "learning_rate": 5.711020347249069e-05, "loss": 0.1285, "step": 23148 }, { "epoch": 16.95897435897436, "grad_norm": 1.6875, "learning_rate": 5.710680778352307e-05, "loss": 0.1355, "step": 23149 }, { "epoch": 16.95970695970696, "grad_norm": 0.640625, "learning_rate": 5.7103412062900056e-05, "loss": 0.1062, "step": 23150 }, { "epoch": 16.96043956043956, "grad_norm": 1.328125, "learning_rate": 5.710001631063796e-05, "loss": 0.1302, "step": 23151 }, { "epoch": 16.96117216117216, "grad_norm": 1.484375, "learning_rate": 5.7096620526752984e-05, "loss": 0.1555, "step": 23152 }, { "epoch": 16.961904761904762, "grad_norm": 1.296875, "learning_rate": 5.709322471126146e-05, "loss": 0.1226, "step": 23153 }, { "epoch": 16.962637362637363, "grad_norm": 1.5, "learning_rate": 5.708982886417962e-05, "loss": 0.1339, "step": 23154 }, { "epoch": 16.963369963369964, "grad_norm": 1.25, "learning_rate": 5.708643298552375e-05, "loss": 0.0981, "step": 23155 }, { "epoch": 16.964102564102564, "grad_norm": 1.6171875, "learning_rate": 5.708303707531011e-05, "loss": 0.1304, "step": 23156 }, { "epoch": 16.964835164835165, "grad_norm": 0.84765625, "learning_rate": 5.707964113355499e-05, "loss": 0.0956, "step": 23157 }, { "epoch": 16.965567765567766, "grad_norm": 0.73046875, "learning_rate": 5.707624516027466e-05, "loss": 0.15, "step": 23158 }, { "epoch": 16.966300366300366, "grad_norm": 1.015625, "learning_rate": 5.707284915548537e-05, "loss": 0.1107, "step": 23159 }, { "epoch": 16.967032967032967, "grad_norm": 1.40625, "learning_rate": 5.7069453119203385e-05, "loss": 0.1276, "step": 23160 }, { "epoch": 16.967765567765568, "grad_norm": 0.90234375, "learning_rate": 5.706605705144501e-05, "loss": 0.1356, "step": 23161 }, { "epoch": 16.96849816849817, "grad_norm": 1.578125, "learning_rate": 5.70626609522265e-05, "loss": 0.1472, "step": 23162 }, { "epoch": 16.96923076923077, "grad_norm": 0.89453125, "learning_rate": 5.705926482156411e-05, "loss": 0.1256, "step": 23163 }, { "epoch": 16.96996336996337, "grad_norm": 1.3046875, "learning_rate": 5.705586865947415e-05, "loss": 0.1295, "step": 23164 }, { "epoch": 16.97069597069597, "grad_norm": 0.99609375, "learning_rate": 5.705247246597285e-05, "loss": 0.13, "step": 23165 }, { "epoch": 16.97142857142857, "grad_norm": 1.0390625, "learning_rate": 5.7049076241076524e-05, "loss": 0.116, "step": 23166 }, { "epoch": 16.972161172161172, "grad_norm": 0.83203125, "learning_rate": 5.704567998480142e-05, "loss": 0.1108, "step": 23167 }, { "epoch": 16.972893772893773, "grad_norm": 0.76953125, "learning_rate": 5.704228369716381e-05, "loss": 0.0974, "step": 23168 }, { "epoch": 16.973626373626374, "grad_norm": 0.69921875, "learning_rate": 5.703888737817996e-05, "loss": 0.1198, "step": 23169 }, { "epoch": 16.974358974358974, "grad_norm": 1.546875, "learning_rate": 5.7035491027866186e-05, "loss": 0.126, "step": 23170 }, { "epoch": 16.975091575091575, "grad_norm": 1.0546875, "learning_rate": 5.70320946462387e-05, "loss": 0.1226, "step": 23171 }, { "epoch": 16.975824175824176, "grad_norm": 0.97265625, "learning_rate": 5.7028698233313815e-05, "loss": 0.1383, "step": 23172 }, { "epoch": 16.976556776556777, "grad_norm": 1.5703125, "learning_rate": 5.7025301789107786e-05, "loss": 0.1085, "step": 23173 }, { "epoch": 16.977289377289377, "grad_norm": 0.921875, "learning_rate": 5.702190531363691e-05, "loss": 0.1397, "step": 23174 }, { "epoch": 16.978021978021978, "grad_norm": 1.15625, "learning_rate": 5.7018508806917424e-05, "loss": 0.1364, "step": 23175 }, { "epoch": 16.97875457875458, "grad_norm": 1.3046875, "learning_rate": 5.701511226896563e-05, "loss": 0.1668, "step": 23176 }, { "epoch": 16.97948717948718, "grad_norm": 1.5390625, "learning_rate": 5.7011715699797794e-05, "loss": 0.1526, "step": 23177 }, { "epoch": 16.98021978021978, "grad_norm": 1.3359375, "learning_rate": 5.700831909943021e-05, "loss": 0.1285, "step": 23178 }, { "epoch": 16.98095238095238, "grad_norm": 1.3359375, "learning_rate": 5.7004922467879106e-05, "loss": 0.1133, "step": 23179 }, { "epoch": 16.98168498168498, "grad_norm": 1.015625, "learning_rate": 5.7001525805160796e-05, "loss": 0.1476, "step": 23180 }, { "epoch": 16.982417582417582, "grad_norm": 1.0234375, "learning_rate": 5.699812911129154e-05, "loss": 0.1211, "step": 23181 }, { "epoch": 16.983150183150183, "grad_norm": 1.4765625, "learning_rate": 5.6994732386287606e-05, "loss": 0.1023, "step": 23182 }, { "epoch": 16.983882783882784, "grad_norm": 1.140625, "learning_rate": 5.69913356301653e-05, "loss": 0.14, "step": 23183 }, { "epoch": 16.984615384615385, "grad_norm": 1.328125, "learning_rate": 5.698793884294086e-05, "loss": 0.1362, "step": 23184 }, { "epoch": 16.985347985347985, "grad_norm": 1.078125, "learning_rate": 5.698454202463057e-05, "loss": 0.1051, "step": 23185 }, { "epoch": 16.986080586080586, "grad_norm": 0.82421875, "learning_rate": 5.6981145175250716e-05, "loss": 0.1137, "step": 23186 }, { "epoch": 16.986813186813187, "grad_norm": 1.109375, "learning_rate": 5.697774829481757e-05, "loss": 0.1359, "step": 23187 }, { "epoch": 16.987545787545788, "grad_norm": 0.85546875, "learning_rate": 5.69743513833474e-05, "loss": 0.126, "step": 23188 }, { "epoch": 16.98827838827839, "grad_norm": 0.67578125, "learning_rate": 5.6970954440856486e-05, "loss": 0.0924, "step": 23189 }, { "epoch": 16.98901098901099, "grad_norm": 1.375, "learning_rate": 5.696755746736109e-05, "loss": 0.146, "step": 23190 }, { "epoch": 16.98974358974359, "grad_norm": 1.21875, "learning_rate": 5.696416046287752e-05, "loss": 0.1134, "step": 23191 }, { "epoch": 16.99047619047619, "grad_norm": 0.98046875, "learning_rate": 5.6960763427422045e-05, "loss": 0.1562, "step": 23192 }, { "epoch": 16.99120879120879, "grad_norm": 1.53125, "learning_rate": 5.695736636101091e-05, "loss": 0.1184, "step": 23193 }, { "epoch": 16.991941391941392, "grad_norm": 1.2890625, "learning_rate": 5.695396926366044e-05, "loss": 0.1156, "step": 23194 }, { "epoch": 16.992673992673993, "grad_norm": 1.140625, "learning_rate": 5.695057213538686e-05, "loss": 0.1126, "step": 23195 }, { "epoch": 16.993406593406593, "grad_norm": 1.40625, "learning_rate": 5.6947174976206486e-05, "loss": 0.1318, "step": 23196 }, { "epoch": 16.994139194139194, "grad_norm": 0.90234375, "learning_rate": 5.694377778613558e-05, "loss": 0.1425, "step": 23197 }, { "epoch": 16.994871794871795, "grad_norm": 1.953125, "learning_rate": 5.694038056519041e-05, "loss": 0.1292, "step": 23198 }, { "epoch": 16.995604395604396, "grad_norm": 1.7578125, "learning_rate": 5.693698331338727e-05, "loss": 0.1053, "step": 23199 }, { "epoch": 16.996336996336996, "grad_norm": 0.875, "learning_rate": 5.693358603074244e-05, "loss": 0.126, "step": 23200 }, { "epoch": 16.997069597069597, "grad_norm": 1.0078125, "learning_rate": 5.693018871727216e-05, "loss": 0.0962, "step": 23201 }, { "epoch": 16.997802197802198, "grad_norm": 0.96875, "learning_rate": 5.692679137299274e-05, "loss": 0.1185, "step": 23202 }, { "epoch": 16.9985347985348, "grad_norm": 0.93359375, "learning_rate": 5.692339399792046e-05, "loss": 0.1201, "step": 23203 }, { "epoch": 16.9992673992674, "grad_norm": 1.75, "learning_rate": 5.691999659207159e-05, "loss": 0.1379, "step": 23204 }, { "epoch": 17.0, "grad_norm": 1.1953125, "learning_rate": 5.691659915546242e-05, "loss": 0.144, "step": 23205 }, { "epoch": 17.0007326007326, "grad_norm": 1.3359375, "learning_rate": 5.691320168810921e-05, "loss": 0.1548, "step": 23206 }, { "epoch": 17.0014652014652, "grad_norm": 1.3671875, "learning_rate": 5.690980419002825e-05, "loss": 0.1251, "step": 23207 }, { "epoch": 17.002197802197802, "grad_norm": 0.98046875, "learning_rate": 5.690640666123579e-05, "loss": 0.1347, "step": 23208 }, { "epoch": 17.002930402930403, "grad_norm": 1.4921875, "learning_rate": 5.690300910174815e-05, "loss": 0.1513, "step": 23209 }, { "epoch": 17.003663003663004, "grad_norm": 0.828125, "learning_rate": 5.68996115115816e-05, "loss": 0.1299, "step": 23210 }, { "epoch": 17.004395604395604, "grad_norm": 1.1640625, "learning_rate": 5.6896213890752405e-05, "loss": 0.1337, "step": 23211 }, { "epoch": 17.005128205128205, "grad_norm": 1.625, "learning_rate": 5.689281623927685e-05, "loss": 0.1186, "step": 23212 }, { "epoch": 17.005860805860806, "grad_norm": 1.7109375, "learning_rate": 5.688941855717121e-05, "loss": 0.1295, "step": 23213 }, { "epoch": 17.006593406593407, "grad_norm": 1.03125, "learning_rate": 5.688602084445176e-05, "loss": 0.1148, "step": 23214 }, { "epoch": 17.007326007326007, "grad_norm": 1.015625, "learning_rate": 5.688262310113479e-05, "loss": 0.1423, "step": 23215 }, { "epoch": 17.008058608058608, "grad_norm": 1.453125, "learning_rate": 5.687922532723659e-05, "loss": 0.1428, "step": 23216 }, { "epoch": 17.00879120879121, "grad_norm": 0.62890625, "learning_rate": 5.6875827522773416e-05, "loss": 0.0994, "step": 23217 }, { "epoch": 17.00952380952381, "grad_norm": 0.67578125, "learning_rate": 5.6872429687761575e-05, "loss": 0.125, "step": 23218 }, { "epoch": 17.01025641025641, "grad_norm": 1.21875, "learning_rate": 5.686903182221731e-05, "loss": 0.1222, "step": 23219 }, { "epoch": 17.01098901098901, "grad_norm": 0.83984375, "learning_rate": 5.6865633926156934e-05, "loss": 0.1269, "step": 23220 }, { "epoch": 17.01172161172161, "grad_norm": 1.15625, "learning_rate": 5.6862235999596715e-05, "loss": 0.1472, "step": 23221 }, { "epoch": 17.012454212454212, "grad_norm": 0.98046875, "learning_rate": 5.685883804255294e-05, "loss": 0.1473, "step": 23222 }, { "epoch": 17.013186813186813, "grad_norm": 1.2734375, "learning_rate": 5.685544005504189e-05, "loss": 0.1681, "step": 23223 }, { "epoch": 17.013919413919414, "grad_norm": 0.984375, "learning_rate": 5.685204203707982e-05, "loss": 0.1296, "step": 23224 }, { "epoch": 17.014652014652015, "grad_norm": 1.703125, "learning_rate": 5.684864398868305e-05, "loss": 0.1407, "step": 23225 }, { "epoch": 17.015384615384615, "grad_norm": 1.0859375, "learning_rate": 5.684524590986784e-05, "loss": 0.0931, "step": 23226 }, { "epoch": 17.016117216117216, "grad_norm": 1.3359375, "learning_rate": 5.6841847800650464e-05, "loss": 0.1686, "step": 23227 }, { "epoch": 17.016849816849817, "grad_norm": 0.94921875, "learning_rate": 5.68384496610472e-05, "loss": 0.1347, "step": 23228 }, { "epoch": 17.017582417582418, "grad_norm": 1.515625, "learning_rate": 5.6835051491074365e-05, "loss": 0.1475, "step": 23229 }, { "epoch": 17.01831501831502, "grad_norm": 0.875, "learning_rate": 5.683165329074821e-05, "loss": 0.1169, "step": 23230 }, { "epoch": 17.01904761904762, "grad_norm": 1.390625, "learning_rate": 5.682825506008503e-05, "loss": 0.1298, "step": 23231 }, { "epoch": 17.01978021978022, "grad_norm": 1.3671875, "learning_rate": 5.682485679910109e-05, "loss": 0.154, "step": 23232 }, { "epoch": 17.02051282051282, "grad_norm": 0.93359375, "learning_rate": 5.682145850781271e-05, "loss": 0.1042, "step": 23233 }, { "epoch": 17.02124542124542, "grad_norm": 1.0625, "learning_rate": 5.681806018623613e-05, "loss": 0.1289, "step": 23234 }, { "epoch": 17.021978021978022, "grad_norm": 0.80078125, "learning_rate": 5.681466183438765e-05, "loss": 0.0857, "step": 23235 }, { "epoch": 17.022710622710623, "grad_norm": 1.1640625, "learning_rate": 5.681126345228356e-05, "loss": 0.137, "step": 23236 }, { "epoch": 17.023443223443223, "grad_norm": 0.77734375, "learning_rate": 5.680786503994013e-05, "loss": 0.1304, "step": 23237 }, { "epoch": 17.024175824175824, "grad_norm": 1.1640625, "learning_rate": 5.6804466597373654e-05, "loss": 0.1473, "step": 23238 }, { "epoch": 17.024908424908425, "grad_norm": 1.2421875, "learning_rate": 5.6801068124600396e-05, "loss": 0.1158, "step": 23239 }, { "epoch": 17.025641025641026, "grad_norm": 1.0, "learning_rate": 5.679766962163666e-05, "loss": 0.1163, "step": 23240 }, { "epoch": 17.026373626373626, "grad_norm": 1.3046875, "learning_rate": 5.679427108849872e-05, "loss": 0.1302, "step": 23241 }, { "epoch": 17.027106227106227, "grad_norm": 0.9765625, "learning_rate": 5.6790872525202856e-05, "loss": 0.1244, "step": 23242 }, { "epoch": 17.027838827838828, "grad_norm": 0.9140625, "learning_rate": 5.6787473931765364e-05, "loss": 0.1303, "step": 23243 }, { "epoch": 17.02857142857143, "grad_norm": 1.3515625, "learning_rate": 5.6784075308202535e-05, "loss": 0.1378, "step": 23244 }, { "epoch": 17.02930402930403, "grad_norm": 1.109375, "learning_rate": 5.6780676654530606e-05, "loss": 0.1286, "step": 23245 }, { "epoch": 17.03003663003663, "grad_norm": 1.203125, "learning_rate": 5.677727797076592e-05, "loss": 0.1252, "step": 23246 }, { "epoch": 17.03076923076923, "grad_norm": 0.953125, "learning_rate": 5.677387925692472e-05, "loss": 0.1228, "step": 23247 }, { "epoch": 17.03150183150183, "grad_norm": 1.2265625, "learning_rate": 5.6770480513023306e-05, "loss": 0.1261, "step": 23248 }, { "epoch": 17.032234432234432, "grad_norm": 1.4296875, "learning_rate": 5.676708173907796e-05, "loss": 0.1766, "step": 23249 }, { "epoch": 17.032967032967033, "grad_norm": 1.578125, "learning_rate": 5.6763682935104964e-05, "loss": 0.1524, "step": 23250 }, { "epoch": 17.033699633699634, "grad_norm": 1.15625, "learning_rate": 5.676028410112063e-05, "loss": 0.1028, "step": 23251 }, { "epoch": 17.034432234432234, "grad_norm": 1.6171875, "learning_rate": 5.675688523714119e-05, "loss": 0.1463, "step": 23252 }, { "epoch": 17.035164835164835, "grad_norm": 0.85546875, "learning_rate": 5.675348634318298e-05, "loss": 0.1008, "step": 23253 }, { "epoch": 17.035897435897436, "grad_norm": 1.1953125, "learning_rate": 5.6750087419262256e-05, "loss": 0.1477, "step": 23254 }, { "epoch": 17.036630036630036, "grad_norm": 0.828125, "learning_rate": 5.6746688465395304e-05, "loss": 0.1391, "step": 23255 }, { "epoch": 17.037362637362637, "grad_norm": 1.890625, "learning_rate": 5.674328948159843e-05, "loss": 0.1189, "step": 23256 }, { "epoch": 17.038095238095238, "grad_norm": 1.71875, "learning_rate": 5.67398904678879e-05, "loss": 0.1761, "step": 23257 }, { "epoch": 17.03882783882784, "grad_norm": 1.4140625, "learning_rate": 5.673649142428001e-05, "loss": 0.1121, "step": 23258 }, { "epoch": 17.03956043956044, "grad_norm": 1.4609375, "learning_rate": 5.673309235079104e-05, "loss": 0.13, "step": 23259 }, { "epoch": 17.04029304029304, "grad_norm": 1.4375, "learning_rate": 5.67296932474373e-05, "loss": 0.1499, "step": 23260 }, { "epoch": 17.04102564102564, "grad_norm": 0.75, "learning_rate": 5.672629411423503e-05, "loss": 0.1213, "step": 23261 }, { "epoch": 17.04175824175824, "grad_norm": 1.09375, "learning_rate": 5.6722894951200546e-05, "loss": 0.1294, "step": 23262 }, { "epoch": 17.042490842490842, "grad_norm": 1.140625, "learning_rate": 5.671949575835011e-05, "loss": 0.1244, "step": 23263 }, { "epoch": 17.043223443223443, "grad_norm": 0.7890625, "learning_rate": 5.671609653570006e-05, "loss": 0.0993, "step": 23264 }, { "epoch": 17.043956043956044, "grad_norm": 1.484375, "learning_rate": 5.6712697283266623e-05, "loss": 0.139, "step": 23265 }, { "epoch": 17.044688644688645, "grad_norm": 1.203125, "learning_rate": 5.670929800106615e-05, "loss": 0.1026, "step": 23266 }, { "epoch": 17.045421245421245, "grad_norm": 1.25, "learning_rate": 5.670589868911486e-05, "loss": 0.133, "step": 23267 }, { "epoch": 17.046153846153846, "grad_norm": 1.25, "learning_rate": 5.6702499347429075e-05, "loss": 0.1283, "step": 23268 }, { "epoch": 17.046886446886447, "grad_norm": 0.61328125, "learning_rate": 5.6699099976025085e-05, "loss": 0.111, "step": 23269 }, { "epoch": 17.047619047619047, "grad_norm": 0.9375, "learning_rate": 5.6695700574919174e-05, "loss": 0.1035, "step": 23270 }, { "epoch": 17.048351648351648, "grad_norm": 0.98046875, "learning_rate": 5.6692301144127625e-05, "loss": 0.0949, "step": 23271 }, { "epoch": 17.04908424908425, "grad_norm": 1.0390625, "learning_rate": 5.668890168366673e-05, "loss": 0.1247, "step": 23272 }, { "epoch": 17.04981684981685, "grad_norm": 1.0078125, "learning_rate": 5.6685502193552785e-05, "loss": 0.1361, "step": 23273 }, { "epoch": 17.05054945054945, "grad_norm": 1.4296875, "learning_rate": 5.668210267380206e-05, "loss": 0.1491, "step": 23274 }, { "epoch": 17.05128205128205, "grad_norm": 0.96875, "learning_rate": 5.667870312443085e-05, "loss": 0.118, "step": 23275 }, { "epoch": 17.052014652014652, "grad_norm": 0.96484375, "learning_rate": 5.6675303545455436e-05, "loss": 0.1125, "step": 23276 }, { "epoch": 17.052747252747253, "grad_norm": 0.9921875, "learning_rate": 5.6671903936892135e-05, "loss": 0.118, "step": 23277 }, { "epoch": 17.053479853479853, "grad_norm": 1.1171875, "learning_rate": 5.66685042987572e-05, "loss": 0.1552, "step": 23278 }, { "epoch": 17.054212454212454, "grad_norm": 0.91015625, "learning_rate": 5.666510463106695e-05, "loss": 0.1344, "step": 23279 }, { "epoch": 17.054945054945055, "grad_norm": 1.3984375, "learning_rate": 5.666170493383765e-05, "loss": 0.1207, "step": 23280 }, { "epoch": 17.055677655677655, "grad_norm": 1.1875, "learning_rate": 5.6658305207085596e-05, "loss": 0.149, "step": 23281 }, { "epoch": 17.056410256410256, "grad_norm": 1.21875, "learning_rate": 5.665490545082709e-05, "loss": 0.1184, "step": 23282 }, { "epoch": 17.057142857142857, "grad_norm": 1.3515625, "learning_rate": 5.6651505665078396e-05, "loss": 0.1328, "step": 23283 }, { "epoch": 17.057875457875458, "grad_norm": 1.984375, "learning_rate": 5.6648105849855835e-05, "loss": 0.1323, "step": 23284 }, { "epoch": 17.05860805860806, "grad_norm": 1.3515625, "learning_rate": 5.6644706005175676e-05, "loss": 0.1442, "step": 23285 }, { "epoch": 17.05934065934066, "grad_norm": 0.8984375, "learning_rate": 5.664130613105422e-05, "loss": 0.087, "step": 23286 }, { "epoch": 17.06007326007326, "grad_norm": 1.0625, "learning_rate": 5.663790622750774e-05, "loss": 0.124, "step": 23287 }, { "epoch": 17.06080586080586, "grad_norm": 0.96484375, "learning_rate": 5.663450629455255e-05, "loss": 0.1244, "step": 23288 }, { "epoch": 17.06153846153846, "grad_norm": 1.5078125, "learning_rate": 5.6631106332204896e-05, "loss": 0.1275, "step": 23289 }, { "epoch": 17.062271062271062, "grad_norm": 1.3671875, "learning_rate": 5.6627706340481125e-05, "loss": 0.1539, "step": 23290 }, { "epoch": 17.063003663003663, "grad_norm": 1.4609375, "learning_rate": 5.662430631939749e-05, "loss": 0.1381, "step": 23291 }, { "epoch": 17.063736263736264, "grad_norm": 1.3828125, "learning_rate": 5.6620906268970306e-05, "loss": 0.1224, "step": 23292 }, { "epoch": 17.064468864468864, "grad_norm": 1.0390625, "learning_rate": 5.661750618921583e-05, "loss": 0.1382, "step": 23293 }, { "epoch": 17.065201465201465, "grad_norm": 1.2578125, "learning_rate": 5.6614106080150385e-05, "loss": 0.1211, "step": 23294 }, { "epoch": 17.065934065934066, "grad_norm": 1.1796875, "learning_rate": 5.661070594179026e-05, "loss": 0.1191, "step": 23295 }, { "epoch": 17.066666666666666, "grad_norm": 1.3203125, "learning_rate": 5.660730577415173e-05, "loss": 0.1022, "step": 23296 }, { "epoch": 17.067399267399267, "grad_norm": 1.21875, "learning_rate": 5.660390557725109e-05, "loss": 0.1115, "step": 23297 }, { "epoch": 17.068131868131868, "grad_norm": 1.2890625, "learning_rate": 5.660050535110464e-05, "loss": 0.133, "step": 23298 }, { "epoch": 17.06886446886447, "grad_norm": 1.4921875, "learning_rate": 5.659710509572866e-05, "loss": 0.1356, "step": 23299 }, { "epoch": 17.06959706959707, "grad_norm": 0.80078125, "learning_rate": 5.659370481113945e-05, "loss": 0.1388, "step": 23300 }, { "epoch": 17.07032967032967, "grad_norm": 1.5078125, "learning_rate": 5.659030449735331e-05, "loss": 0.1361, "step": 23301 }, { "epoch": 17.07106227106227, "grad_norm": 1.03125, "learning_rate": 5.65869041543865e-05, "loss": 0.1494, "step": 23302 }, { "epoch": 17.07179487179487, "grad_norm": 0.76171875, "learning_rate": 5.658350378225536e-05, "loss": 0.1086, "step": 23303 }, { "epoch": 17.072527472527472, "grad_norm": 0.73828125, "learning_rate": 5.658010338097613e-05, "loss": 0.1126, "step": 23304 }, { "epoch": 17.073260073260073, "grad_norm": 1.0625, "learning_rate": 5.657670295056513e-05, "loss": 0.126, "step": 23305 }, { "epoch": 17.073992673992674, "grad_norm": 1.1875, "learning_rate": 5.657330249103868e-05, "loss": 0.1491, "step": 23306 }, { "epoch": 17.074725274725274, "grad_norm": 0.7890625, "learning_rate": 5.656990200241301e-05, "loss": 0.131, "step": 23307 }, { "epoch": 17.075457875457875, "grad_norm": 0.84765625, "learning_rate": 5.656650148470447e-05, "loss": 0.1024, "step": 23308 }, { "epoch": 17.076190476190476, "grad_norm": 0.765625, "learning_rate": 5.656310093792933e-05, "loss": 0.1086, "step": 23309 }, { "epoch": 17.076923076923077, "grad_norm": 1.515625, "learning_rate": 5.6559700362103874e-05, "loss": 0.1348, "step": 23310 }, { "epoch": 17.077655677655677, "grad_norm": 1.015625, "learning_rate": 5.65562997572444e-05, "loss": 0.1366, "step": 23311 }, { "epoch": 17.078388278388278, "grad_norm": 1.1171875, "learning_rate": 5.655289912336722e-05, "loss": 0.1084, "step": 23312 }, { "epoch": 17.07912087912088, "grad_norm": 1.3671875, "learning_rate": 5.65494984604886e-05, "loss": 0.1198, "step": 23313 }, { "epoch": 17.07985347985348, "grad_norm": 1.359375, "learning_rate": 5.6546097768624865e-05, "loss": 0.115, "step": 23314 }, { "epoch": 17.08058608058608, "grad_norm": 1.4609375, "learning_rate": 5.654269704779227e-05, "loss": 0.1279, "step": 23315 }, { "epoch": 17.08131868131868, "grad_norm": 0.89453125, "learning_rate": 5.653929629800714e-05, "loss": 0.1287, "step": 23316 }, { "epoch": 17.08205128205128, "grad_norm": 1.5703125, "learning_rate": 5.6535895519285754e-05, "loss": 0.1108, "step": 23317 }, { "epoch": 17.082783882783882, "grad_norm": 0.625, "learning_rate": 5.65324947116444e-05, "loss": 0.1048, "step": 23318 }, { "epoch": 17.083516483516483, "grad_norm": 0.90234375, "learning_rate": 5.6529093875099406e-05, "loss": 0.1424, "step": 23319 }, { "epoch": 17.084249084249084, "grad_norm": 1.1328125, "learning_rate": 5.6525693009667025e-05, "loss": 0.1369, "step": 23320 }, { "epoch": 17.084981684981685, "grad_norm": 1.09375, "learning_rate": 5.6522292115363594e-05, "loss": 0.1272, "step": 23321 }, { "epoch": 17.085714285714285, "grad_norm": 0.94140625, "learning_rate": 5.651889119220537e-05, "loss": 0.1117, "step": 23322 }, { "epoch": 17.086446886446886, "grad_norm": 0.875, "learning_rate": 5.651549024020867e-05, "loss": 0.136, "step": 23323 }, { "epoch": 17.087179487179487, "grad_norm": 1.390625, "learning_rate": 5.651208925938978e-05, "loss": 0.1247, "step": 23324 }, { "epoch": 17.087912087912088, "grad_norm": 0.9765625, "learning_rate": 5.6508688249764995e-05, "loss": 0.1446, "step": 23325 }, { "epoch": 17.08864468864469, "grad_norm": 0.94140625, "learning_rate": 5.65052872113506e-05, "loss": 0.1144, "step": 23326 }, { "epoch": 17.08937728937729, "grad_norm": 1.109375, "learning_rate": 5.650188614416292e-05, "loss": 0.1444, "step": 23327 }, { "epoch": 17.09010989010989, "grad_norm": 0.8984375, "learning_rate": 5.649848504821822e-05, "loss": 0.1097, "step": 23328 }, { "epoch": 17.09084249084249, "grad_norm": 1.125, "learning_rate": 5.6495083923532823e-05, "loss": 0.1016, "step": 23329 }, { "epoch": 17.09157509157509, "grad_norm": 0.671875, "learning_rate": 5.649168277012299e-05, "loss": 0.1072, "step": 23330 }, { "epoch": 17.092307692307692, "grad_norm": 0.921875, "learning_rate": 5.648828158800504e-05, "loss": 0.1092, "step": 23331 }, { "epoch": 17.093040293040293, "grad_norm": 1.015625, "learning_rate": 5.6484880377195295e-05, "loss": 0.1169, "step": 23332 }, { "epoch": 17.093772893772893, "grad_norm": 0.99609375, "learning_rate": 5.6481479137709994e-05, "loss": 0.0955, "step": 23333 }, { "epoch": 17.094505494505494, "grad_norm": 1.0234375, "learning_rate": 5.647807786956547e-05, "loss": 0.1106, "step": 23334 }, { "epoch": 17.095238095238095, "grad_norm": 1.453125, "learning_rate": 5.647467657277803e-05, "loss": 0.1447, "step": 23335 }, { "epoch": 17.095970695970696, "grad_norm": 1.03125, "learning_rate": 5.647127524736393e-05, "loss": 0.1054, "step": 23336 }, { "epoch": 17.096703296703296, "grad_norm": 0.6796875, "learning_rate": 5.646787389333951e-05, "loss": 0.1098, "step": 23337 }, { "epoch": 17.097435897435897, "grad_norm": 0.84765625, "learning_rate": 5.646447251072105e-05, "loss": 0.1201, "step": 23338 }, { "epoch": 17.098168498168498, "grad_norm": 1.7109375, "learning_rate": 5.646107109952481e-05, "loss": 0.1335, "step": 23339 }, { "epoch": 17.0989010989011, "grad_norm": 1.3515625, "learning_rate": 5.645766965976715e-05, "loss": 0.1217, "step": 23340 }, { "epoch": 17.0996336996337, "grad_norm": 0.97265625, "learning_rate": 5.6454268191464335e-05, "loss": 0.0887, "step": 23341 }, { "epoch": 17.1003663003663, "grad_norm": 1.15625, "learning_rate": 5.645086669463265e-05, "loss": 0.1078, "step": 23342 }, { "epoch": 17.1010989010989, "grad_norm": 1.609375, "learning_rate": 5.6447465169288414e-05, "loss": 0.11, "step": 23343 }, { "epoch": 17.1018315018315, "grad_norm": 1.015625, "learning_rate": 5.644406361544794e-05, "loss": 0.1156, "step": 23344 }, { "epoch": 17.102564102564102, "grad_norm": 0.8203125, "learning_rate": 5.644066203312749e-05, "loss": 0.138, "step": 23345 }, { "epoch": 17.103296703296703, "grad_norm": 1.2265625, "learning_rate": 5.643726042234339e-05, "loss": 0.1268, "step": 23346 }, { "epoch": 17.104029304029304, "grad_norm": 1.3671875, "learning_rate": 5.6433858783111904e-05, "loss": 0.1149, "step": 23347 }, { "epoch": 17.104761904761904, "grad_norm": 1.015625, "learning_rate": 5.6430457115449375e-05, "loss": 0.121, "step": 23348 }, { "epoch": 17.105494505494505, "grad_norm": 1.1796875, "learning_rate": 5.642705541937207e-05, "loss": 0.1215, "step": 23349 }, { "epoch": 17.106227106227106, "grad_norm": 1.1875, "learning_rate": 5.642365369489631e-05, "loss": 0.1606, "step": 23350 }, { "epoch": 17.106959706959707, "grad_norm": 0.8984375, "learning_rate": 5.642025194203837e-05, "loss": 0.112, "step": 23351 }, { "epoch": 17.107692307692307, "grad_norm": 1.3828125, "learning_rate": 5.6416850160814547e-05, "loss": 0.1163, "step": 23352 }, { "epoch": 17.108424908424908, "grad_norm": 1.109375, "learning_rate": 5.6413448351241174e-05, "loss": 0.1118, "step": 23353 }, { "epoch": 17.10915750915751, "grad_norm": 1.3828125, "learning_rate": 5.6410046513334515e-05, "loss": 0.1465, "step": 23354 }, { "epoch": 17.10989010989011, "grad_norm": 1.1171875, "learning_rate": 5.640664464711088e-05, "loss": 0.1062, "step": 23355 }, { "epoch": 17.11062271062271, "grad_norm": 1.6328125, "learning_rate": 5.640324275258658e-05, "loss": 0.0907, "step": 23356 }, { "epoch": 17.11135531135531, "grad_norm": 1.0234375, "learning_rate": 5.6399840829777894e-05, "loss": 0.1138, "step": 23357 }, { "epoch": 17.11208791208791, "grad_norm": 1.1875, "learning_rate": 5.639643887870114e-05, "loss": 0.123, "step": 23358 }, { "epoch": 17.112820512820512, "grad_norm": 1.1640625, "learning_rate": 5.639303689937261e-05, "loss": 0.1037, "step": 23359 }, { "epoch": 17.113553113553113, "grad_norm": 1.2890625, "learning_rate": 5.6389634891808626e-05, "loss": 0.1012, "step": 23360 }, { "epoch": 17.114285714285714, "grad_norm": 2.171875, "learning_rate": 5.638623285602544e-05, "loss": 0.1419, "step": 23361 }, { "epoch": 17.115018315018315, "grad_norm": 0.77734375, "learning_rate": 5.63828307920394e-05, "loss": 0.1245, "step": 23362 }, { "epoch": 17.115750915750915, "grad_norm": 1.4296875, "learning_rate": 5.6379428699866774e-05, "loss": 0.1139, "step": 23363 }, { "epoch": 17.116483516483516, "grad_norm": 1.4921875, "learning_rate": 5.63760265795239e-05, "loss": 0.145, "step": 23364 }, { "epoch": 17.117216117216117, "grad_norm": 1.1484375, "learning_rate": 5.637262443102702e-05, "loss": 0.1106, "step": 23365 }, { "epoch": 17.117948717948718, "grad_norm": 0.82421875, "learning_rate": 5.6369222254392484e-05, "loss": 0.0935, "step": 23366 }, { "epoch": 17.11868131868132, "grad_norm": 1.0703125, "learning_rate": 5.6365820049636574e-05, "loss": 0.1289, "step": 23367 }, { "epoch": 17.11941391941392, "grad_norm": 0.96484375, "learning_rate": 5.63624178167756e-05, "loss": 0.1111, "step": 23368 }, { "epoch": 17.12014652014652, "grad_norm": 1.265625, "learning_rate": 5.635901555582585e-05, "loss": 0.1127, "step": 23369 }, { "epoch": 17.12087912087912, "grad_norm": 1.09375, "learning_rate": 5.635561326680363e-05, "loss": 0.1199, "step": 23370 }, { "epoch": 17.12161172161172, "grad_norm": 0.94140625, "learning_rate": 5.635221094972525e-05, "loss": 0.1336, "step": 23371 }, { "epoch": 17.122344322344322, "grad_norm": 1.0625, "learning_rate": 5.6348808604607015e-05, "loss": 0.1446, "step": 23372 }, { "epoch": 17.123076923076923, "grad_norm": 1.1171875, "learning_rate": 5.6345406231465204e-05, "loss": 0.1324, "step": 23373 }, { "epoch": 17.123809523809523, "grad_norm": 1.5703125, "learning_rate": 5.634200383031615e-05, "loss": 0.1373, "step": 23374 }, { "epoch": 17.124542124542124, "grad_norm": 0.94921875, "learning_rate": 5.6338601401176114e-05, "loss": 0.1223, "step": 23375 }, { "epoch": 17.125274725274725, "grad_norm": 0.6796875, "learning_rate": 5.633519894406145e-05, "loss": 0.1295, "step": 23376 }, { "epoch": 17.126007326007326, "grad_norm": 0.9765625, "learning_rate": 5.6331796458988415e-05, "loss": 0.1014, "step": 23377 }, { "epoch": 17.126739926739926, "grad_norm": 0.78515625, "learning_rate": 5.632839394597332e-05, "loss": 0.0947, "step": 23378 }, { "epoch": 17.127472527472527, "grad_norm": 2.453125, "learning_rate": 5.63249914050325e-05, "loss": 0.162, "step": 23379 }, { "epoch": 17.128205128205128, "grad_norm": 1.0625, "learning_rate": 5.632158883618222e-05, "loss": 0.1352, "step": 23380 }, { "epoch": 17.12893772893773, "grad_norm": 0.953125, "learning_rate": 5.631818623943881e-05, "loss": 0.1052, "step": 23381 }, { "epoch": 17.12967032967033, "grad_norm": 0.89453125, "learning_rate": 5.631478361481853e-05, "loss": 0.1174, "step": 23382 }, { "epoch": 17.13040293040293, "grad_norm": 0.921875, "learning_rate": 5.631138096233772e-05, "loss": 0.1258, "step": 23383 }, { "epoch": 17.13113553113553, "grad_norm": 1.59375, "learning_rate": 5.63079782820127e-05, "loss": 0.1408, "step": 23384 }, { "epoch": 17.13186813186813, "grad_norm": 1.5, "learning_rate": 5.630457557385973e-05, "loss": 0.1473, "step": 23385 }, { "epoch": 17.132600732600732, "grad_norm": 1.46875, "learning_rate": 5.630117283789514e-05, "loss": 0.0932, "step": 23386 }, { "epoch": 17.133333333333333, "grad_norm": 0.82421875, "learning_rate": 5.629777007413524e-05, "loss": 0.1264, "step": 23387 }, { "epoch": 17.134065934065934, "grad_norm": 1.8984375, "learning_rate": 5.62943672825963e-05, "loss": 0.1326, "step": 23388 }, { "epoch": 17.134798534798534, "grad_norm": 1.3515625, "learning_rate": 5.629096446329465e-05, "loss": 0.1527, "step": 23389 }, { "epoch": 17.135531135531135, "grad_norm": 1.1171875, "learning_rate": 5.6287561616246596e-05, "loss": 0.1267, "step": 23390 }, { "epoch": 17.136263736263736, "grad_norm": 1.03125, "learning_rate": 5.628415874146842e-05, "loss": 0.1503, "step": 23391 }, { "epoch": 17.136996336996337, "grad_norm": 1.453125, "learning_rate": 5.628075583897645e-05, "loss": 0.1037, "step": 23392 }, { "epoch": 17.137728937728937, "grad_norm": 0.8203125, "learning_rate": 5.6277352908786975e-05, "loss": 0.1059, "step": 23393 }, { "epoch": 17.138461538461538, "grad_norm": 1.0546875, "learning_rate": 5.627394995091632e-05, "loss": 0.1014, "step": 23394 }, { "epoch": 17.13919413919414, "grad_norm": 0.85546875, "learning_rate": 5.627054696538075e-05, "loss": 0.1194, "step": 23395 }, { "epoch": 17.13992673992674, "grad_norm": 0.93359375, "learning_rate": 5.626714395219661e-05, "loss": 0.0993, "step": 23396 }, { "epoch": 17.14065934065934, "grad_norm": 0.984375, "learning_rate": 5.6263740911380195e-05, "loss": 0.0989, "step": 23397 }, { "epoch": 17.14139194139194, "grad_norm": 1.2578125, "learning_rate": 5.6260337842947784e-05, "loss": 0.1194, "step": 23398 }, { "epoch": 17.14212454212454, "grad_norm": 1.40625, "learning_rate": 5.625693474691573e-05, "loss": 0.1065, "step": 23399 }, { "epoch": 17.142857142857142, "grad_norm": 0.9921875, "learning_rate": 5.6253531623300305e-05, "loss": 0.1053, "step": 23400 }, { "epoch": 17.143589743589743, "grad_norm": 1.09375, "learning_rate": 5.625012847211782e-05, "loss": 0.107, "step": 23401 }, { "epoch": 17.144322344322344, "grad_norm": 1.125, "learning_rate": 5.624672529338457e-05, "loss": 0.1358, "step": 23402 }, { "epoch": 17.145054945054945, "grad_norm": 1.7578125, "learning_rate": 5.624332208711689e-05, "loss": 0.1481, "step": 23403 }, { "epoch": 17.145787545787545, "grad_norm": 0.98046875, "learning_rate": 5.623991885333105e-05, "loss": 0.1373, "step": 23404 }, { "epoch": 17.146520146520146, "grad_norm": 1.4765625, "learning_rate": 5.623651559204339e-05, "loss": 0.1109, "step": 23405 }, { "epoch": 17.147252747252747, "grad_norm": 1.6953125, "learning_rate": 5.623311230327019e-05, "loss": 0.139, "step": 23406 }, { "epoch": 17.147985347985347, "grad_norm": 1.078125, "learning_rate": 5.6229708987027785e-05, "loss": 0.1335, "step": 23407 }, { "epoch": 17.148717948717948, "grad_norm": 1.0234375, "learning_rate": 5.622630564333244e-05, "loss": 0.1263, "step": 23408 }, { "epoch": 17.14945054945055, "grad_norm": 1.09375, "learning_rate": 5.6222902272200494e-05, "loss": 0.132, "step": 23409 }, { "epoch": 17.15018315018315, "grad_norm": 0.921875, "learning_rate": 5.621949887364825e-05, "loss": 0.1003, "step": 23410 }, { "epoch": 17.15091575091575, "grad_norm": 1.375, "learning_rate": 5.6216095447692e-05, "loss": 0.1296, "step": 23411 }, { "epoch": 17.15164835164835, "grad_norm": 1.1171875, "learning_rate": 5.6212691994348064e-05, "loss": 0.1059, "step": 23412 }, { "epoch": 17.152380952380952, "grad_norm": 1.03125, "learning_rate": 5.6209288513632754e-05, "loss": 0.1392, "step": 23413 }, { "epoch": 17.153113553113553, "grad_norm": 0.8515625, "learning_rate": 5.620588500556235e-05, "loss": 0.1276, "step": 23414 }, { "epoch": 17.153846153846153, "grad_norm": 0.96484375, "learning_rate": 5.6202481470153204e-05, "loss": 0.1133, "step": 23415 }, { "epoch": 17.154578754578754, "grad_norm": 0.8984375, "learning_rate": 5.6199077907421584e-05, "loss": 0.108, "step": 23416 }, { "epoch": 17.155311355311355, "grad_norm": 0.94921875, "learning_rate": 5.61956743173838e-05, "loss": 0.1259, "step": 23417 }, { "epoch": 17.156043956043955, "grad_norm": 1.5390625, "learning_rate": 5.6192270700056195e-05, "loss": 0.1277, "step": 23418 }, { "epoch": 17.156776556776556, "grad_norm": 1.46875, "learning_rate": 5.618886705545502e-05, "loss": 0.1229, "step": 23419 }, { "epoch": 17.157509157509157, "grad_norm": 0.81640625, "learning_rate": 5.618546338359663e-05, "loss": 0.114, "step": 23420 }, { "epoch": 17.158241758241758, "grad_norm": 1.421875, "learning_rate": 5.6182059684497326e-05, "loss": 0.161, "step": 23421 }, { "epoch": 17.15897435897436, "grad_norm": 1.4765625, "learning_rate": 5.617865595817339e-05, "loss": 0.1152, "step": 23422 }, { "epoch": 17.15970695970696, "grad_norm": 0.8203125, "learning_rate": 5.6175252204641174e-05, "loss": 0.1009, "step": 23423 }, { "epoch": 17.16043956043956, "grad_norm": 0.99609375, "learning_rate": 5.617184842391695e-05, "loss": 0.1188, "step": 23424 }, { "epoch": 17.16117216117216, "grad_norm": 1.1640625, "learning_rate": 5.616844461601703e-05, "loss": 0.0996, "step": 23425 }, { "epoch": 17.16190476190476, "grad_norm": 1.0859375, "learning_rate": 5.6165040780957755e-05, "loss": 0.0991, "step": 23426 }, { "epoch": 17.162637362637362, "grad_norm": 1.921875, "learning_rate": 5.616163691875539e-05, "loss": 0.1088, "step": 23427 }, { "epoch": 17.163369963369963, "grad_norm": 1.1640625, "learning_rate": 5.615823302942627e-05, "loss": 0.1085, "step": 23428 }, { "epoch": 17.164102564102564, "grad_norm": 1.109375, "learning_rate": 5.6154829112986706e-05, "loss": 0.1257, "step": 23429 }, { "epoch": 17.164835164835164, "grad_norm": 0.7734375, "learning_rate": 5.615142516945299e-05, "loss": 0.1169, "step": 23430 }, { "epoch": 17.165567765567765, "grad_norm": 1.0, "learning_rate": 5.614802119884144e-05, "loss": 0.1121, "step": 23431 }, { "epoch": 17.166300366300366, "grad_norm": 0.83203125, "learning_rate": 5.614461720116836e-05, "loss": 0.1102, "step": 23432 }, { "epoch": 17.167032967032966, "grad_norm": 1.921875, "learning_rate": 5.614121317645007e-05, "loss": 0.1291, "step": 23433 }, { "epoch": 17.167765567765567, "grad_norm": 1.2265625, "learning_rate": 5.6137809124702875e-05, "loss": 0.1147, "step": 23434 }, { "epoch": 17.168498168498168, "grad_norm": 1.09375, "learning_rate": 5.6134405045943083e-05, "loss": 0.1643, "step": 23435 }, { "epoch": 17.16923076923077, "grad_norm": 1.53125, "learning_rate": 5.613100094018703e-05, "loss": 0.1236, "step": 23436 }, { "epoch": 17.16996336996337, "grad_norm": 1.375, "learning_rate": 5.6127596807450966e-05, "loss": 0.1335, "step": 23437 }, { "epoch": 17.17069597069597, "grad_norm": 1.875, "learning_rate": 5.6124192647751274e-05, "loss": 0.0959, "step": 23438 }, { "epoch": 17.17142857142857, "grad_norm": 1.203125, "learning_rate": 5.612078846110421e-05, "loss": 0.129, "step": 23439 }, { "epoch": 17.17216117216117, "grad_norm": 1.4375, "learning_rate": 5.611738424752611e-05, "loss": 0.1284, "step": 23440 }, { "epoch": 17.172893772893772, "grad_norm": 1.1328125, "learning_rate": 5.611398000703327e-05, "loss": 0.1218, "step": 23441 }, { "epoch": 17.173626373626373, "grad_norm": 1.53125, "learning_rate": 5.611057573964201e-05, "loss": 0.17, "step": 23442 }, { "epoch": 17.174358974358974, "grad_norm": 0.79296875, "learning_rate": 5.610717144536865e-05, "loss": 0.1374, "step": 23443 }, { "epoch": 17.175091575091574, "grad_norm": 0.65625, "learning_rate": 5.610376712422949e-05, "loss": 0.1027, "step": 23444 }, { "epoch": 17.175824175824175, "grad_norm": 1.1796875, "learning_rate": 5.6100362776240825e-05, "loss": 0.1109, "step": 23445 }, { "epoch": 17.176556776556776, "grad_norm": 1.109375, "learning_rate": 5.6096958401419e-05, "loss": 0.1181, "step": 23446 }, { "epoch": 17.177289377289377, "grad_norm": 1.4453125, "learning_rate": 5.60935539997803e-05, "loss": 0.1292, "step": 23447 }, { "epoch": 17.178021978021977, "grad_norm": 0.94140625, "learning_rate": 5.609014957134106e-05, "loss": 0.1301, "step": 23448 }, { "epoch": 17.178754578754578, "grad_norm": 0.98046875, "learning_rate": 5.608674511611757e-05, "loss": 0.1124, "step": 23449 }, { "epoch": 17.17948717948718, "grad_norm": 1.1640625, "learning_rate": 5.6083340634126135e-05, "loss": 0.1163, "step": 23450 }, { "epoch": 17.18021978021978, "grad_norm": 1.328125, "learning_rate": 5.607993612538311e-05, "loss": 0.1336, "step": 23451 }, { "epoch": 17.18095238095238, "grad_norm": 1.3125, "learning_rate": 5.607653158990477e-05, "loss": 0.1197, "step": 23452 }, { "epoch": 17.18168498168498, "grad_norm": 0.7734375, "learning_rate": 5.6073127027707414e-05, "loss": 0.0957, "step": 23453 }, { "epoch": 17.18241758241758, "grad_norm": 1.8046875, "learning_rate": 5.6069722438807414e-05, "loss": 0.1624, "step": 23454 }, { "epoch": 17.183150183150182, "grad_norm": 0.8984375, "learning_rate": 5.6066317823221024e-05, "loss": 0.1305, "step": 23455 }, { "epoch": 17.183882783882783, "grad_norm": 0.80859375, "learning_rate": 5.6062913180964575e-05, "loss": 0.1098, "step": 23456 }, { "epoch": 17.184615384615384, "grad_norm": 0.8125, "learning_rate": 5.605950851205439e-05, "loss": 0.1201, "step": 23457 }, { "epoch": 17.185347985347985, "grad_norm": 1.0234375, "learning_rate": 5.6056103816506776e-05, "loss": 0.1091, "step": 23458 }, { "epoch": 17.186080586080585, "grad_norm": 1.515625, "learning_rate": 5.6052699094338034e-05, "loss": 0.1581, "step": 23459 }, { "epoch": 17.186813186813186, "grad_norm": 1.3984375, "learning_rate": 5.6049294345564495e-05, "loss": 0.1184, "step": 23460 }, { "epoch": 17.187545787545787, "grad_norm": 0.8359375, "learning_rate": 5.6045889570202456e-05, "loss": 0.0991, "step": 23461 }, { "epoch": 17.188278388278388, "grad_norm": 1.0, "learning_rate": 5.604248476826827e-05, "loss": 0.1452, "step": 23462 }, { "epoch": 17.18901098901099, "grad_norm": 1.5390625, "learning_rate": 5.6039079939778196e-05, "loss": 0.1171, "step": 23463 }, { "epoch": 17.18974358974359, "grad_norm": 0.8125, "learning_rate": 5.6035675084748575e-05, "loss": 0.147, "step": 23464 }, { "epoch": 17.19047619047619, "grad_norm": 1.5703125, "learning_rate": 5.603227020319572e-05, "loss": 0.1256, "step": 23465 }, { "epoch": 17.19120879120879, "grad_norm": 1.1796875, "learning_rate": 5.602886529513594e-05, "loss": 0.1227, "step": 23466 }, { "epoch": 17.19194139194139, "grad_norm": 0.75390625, "learning_rate": 5.602546036058555e-05, "loss": 0.0955, "step": 23467 }, { "epoch": 17.192673992673992, "grad_norm": 1.265625, "learning_rate": 5.602205539956087e-05, "loss": 0.1267, "step": 23468 }, { "epoch": 17.193406593406593, "grad_norm": 1.15625, "learning_rate": 5.60186504120782e-05, "loss": 0.1224, "step": 23469 }, { "epoch": 17.194139194139193, "grad_norm": 0.97265625, "learning_rate": 5.601524539815387e-05, "loss": 0.1274, "step": 23470 }, { "epoch": 17.194871794871794, "grad_norm": 1.7109375, "learning_rate": 5.6011840357804194e-05, "loss": 0.1462, "step": 23471 }, { "epoch": 17.195604395604395, "grad_norm": 0.8828125, "learning_rate": 5.600843529104547e-05, "loss": 0.1291, "step": 23472 }, { "epoch": 17.196336996336996, "grad_norm": 0.87109375, "learning_rate": 5.600503019789403e-05, "loss": 0.1228, "step": 23473 }, { "epoch": 17.197069597069596, "grad_norm": 0.87109375, "learning_rate": 5.6001625078366185e-05, "loss": 0.1178, "step": 23474 }, { "epoch": 17.197802197802197, "grad_norm": 1.265625, "learning_rate": 5.5998219932478243e-05, "loss": 0.1047, "step": 23475 }, { "epoch": 17.198534798534798, "grad_norm": 1.140625, "learning_rate": 5.599481476024654e-05, "loss": 0.1455, "step": 23476 }, { "epoch": 17.1992673992674, "grad_norm": 1.03125, "learning_rate": 5.599140956168736e-05, "loss": 0.1281, "step": 23477 }, { "epoch": 17.2, "grad_norm": 0.8203125, "learning_rate": 5.5988004336817045e-05, "loss": 0.1338, "step": 23478 }, { "epoch": 17.2007326007326, "grad_norm": 0.875, "learning_rate": 5.598459908565188e-05, "loss": 0.1308, "step": 23479 }, { "epoch": 17.2014652014652, "grad_norm": 0.91015625, "learning_rate": 5.598119380820822e-05, "loss": 0.1417, "step": 23480 }, { "epoch": 17.2021978021978, "grad_norm": 0.84765625, "learning_rate": 5.597778850450236e-05, "loss": 0.1057, "step": 23481 }, { "epoch": 17.202930402930402, "grad_norm": 0.984375, "learning_rate": 5.5974383174550604e-05, "loss": 0.1363, "step": 23482 }, { "epoch": 17.203663003663003, "grad_norm": 0.84765625, "learning_rate": 5.59709778183693e-05, "loss": 0.137, "step": 23483 }, { "epoch": 17.204395604395604, "grad_norm": 1.9609375, "learning_rate": 5.596757243597472e-05, "loss": 0.1253, "step": 23484 }, { "epoch": 17.205128205128204, "grad_norm": 1.03125, "learning_rate": 5.596416702738322e-05, "loss": 0.1278, "step": 23485 }, { "epoch": 17.205860805860805, "grad_norm": 1.1328125, "learning_rate": 5.59607615926111e-05, "loss": 0.1145, "step": 23486 }, { "epoch": 17.206593406593406, "grad_norm": 0.64453125, "learning_rate": 5.5957356131674674e-05, "loss": 0.1204, "step": 23487 }, { "epoch": 17.207326007326007, "grad_norm": 1.4140625, "learning_rate": 5.595395064459027e-05, "loss": 0.1439, "step": 23488 }, { "epoch": 17.208058608058607, "grad_norm": 0.98046875, "learning_rate": 5.595054513137421e-05, "loss": 0.1341, "step": 23489 }, { "epoch": 17.208791208791208, "grad_norm": 1.4140625, "learning_rate": 5.594713959204278e-05, "loss": 0.1375, "step": 23490 }, { "epoch": 17.20952380952381, "grad_norm": 1.265625, "learning_rate": 5.594373402661233e-05, "loss": 0.1434, "step": 23491 }, { "epoch": 17.21025641025641, "grad_norm": 1.671875, "learning_rate": 5.594032843509914e-05, "loss": 0.1054, "step": 23492 }, { "epoch": 17.21098901098901, "grad_norm": 0.828125, "learning_rate": 5.593692281751958e-05, "loss": 0.1044, "step": 23493 }, { "epoch": 17.21172161172161, "grad_norm": 0.921875, "learning_rate": 5.5933517173889924e-05, "loss": 0.1226, "step": 23494 }, { "epoch": 17.21245421245421, "grad_norm": 0.921875, "learning_rate": 5.593011150422649e-05, "loss": 0.1145, "step": 23495 }, { "epoch": 17.213186813186812, "grad_norm": 1.1484375, "learning_rate": 5.5926705808545635e-05, "loss": 0.109, "step": 23496 }, { "epoch": 17.213919413919413, "grad_norm": 1.4296875, "learning_rate": 5.592330008686363e-05, "loss": 0.1192, "step": 23497 }, { "epoch": 17.214652014652014, "grad_norm": 1.1796875, "learning_rate": 5.591989433919682e-05, "loss": 0.1184, "step": 23498 }, { "epoch": 17.215384615384615, "grad_norm": 1.5, "learning_rate": 5.591648856556152e-05, "loss": 0.1551, "step": 23499 }, { "epoch": 17.216117216117215, "grad_norm": 1.3046875, "learning_rate": 5.5913082765974045e-05, "loss": 0.1217, "step": 23500 }, { "epoch": 17.216849816849816, "grad_norm": 1.125, "learning_rate": 5.590967694045072e-05, "loss": 0.1339, "step": 23501 }, { "epoch": 17.217582417582417, "grad_norm": 1.0390625, "learning_rate": 5.590627108900785e-05, "loss": 0.1174, "step": 23502 }, { "epoch": 17.218315018315018, "grad_norm": 0.96484375, "learning_rate": 5.590286521166176e-05, "loss": 0.1501, "step": 23503 }, { "epoch": 17.21904761904762, "grad_norm": 0.9765625, "learning_rate": 5.589945930842877e-05, "loss": 0.1241, "step": 23504 }, { "epoch": 17.21978021978022, "grad_norm": 0.9375, "learning_rate": 5.5896053379325186e-05, "loss": 0.1317, "step": 23505 }, { "epoch": 17.22051282051282, "grad_norm": 1.6328125, "learning_rate": 5.589264742436736e-05, "loss": 0.1543, "step": 23506 }, { "epoch": 17.22124542124542, "grad_norm": 0.8671875, "learning_rate": 5.58892414435716e-05, "loss": 0.1058, "step": 23507 }, { "epoch": 17.22197802197802, "grad_norm": 1.125, "learning_rate": 5.588583543695418e-05, "loss": 0.1267, "step": 23508 }, { "epoch": 17.222710622710622, "grad_norm": 0.890625, "learning_rate": 5.5882429404531475e-05, "loss": 0.1222, "step": 23509 }, { "epoch": 17.223443223443223, "grad_norm": 1.5546875, "learning_rate": 5.587902334631977e-05, "loss": 0.1428, "step": 23510 }, { "epoch": 17.224175824175823, "grad_norm": 2.09375, "learning_rate": 5.58756172623354e-05, "loss": 0.1263, "step": 23511 }, { "epoch": 17.224908424908424, "grad_norm": 2.109375, "learning_rate": 5.58722111525947e-05, "loss": 0.1261, "step": 23512 }, { "epoch": 17.225641025641025, "grad_norm": 1.2890625, "learning_rate": 5.586880501711396e-05, "loss": 0.1362, "step": 23513 }, { "epoch": 17.226373626373626, "grad_norm": 1.15625, "learning_rate": 5.586539885590951e-05, "loss": 0.1453, "step": 23514 }, { "epoch": 17.227106227106226, "grad_norm": 1.1796875, "learning_rate": 5.586199266899769e-05, "loss": 0.1287, "step": 23515 }, { "epoch": 17.227838827838827, "grad_norm": 1.5234375, "learning_rate": 5.585858645639478e-05, "loss": 0.1229, "step": 23516 }, { "epoch": 17.228571428571428, "grad_norm": 1.546875, "learning_rate": 5.585518021811714e-05, "loss": 0.1131, "step": 23517 }, { "epoch": 17.22930402930403, "grad_norm": 1.2265625, "learning_rate": 5.585177395418106e-05, "loss": 0.1329, "step": 23518 }, { "epoch": 17.23003663003663, "grad_norm": 1.3671875, "learning_rate": 5.584836766460288e-05, "loss": 0.1307, "step": 23519 }, { "epoch": 17.23076923076923, "grad_norm": 1.6015625, "learning_rate": 5.584496134939893e-05, "loss": 0.11, "step": 23520 }, { "epoch": 17.23150183150183, "grad_norm": 0.99609375, "learning_rate": 5.5841555008585486e-05, "loss": 0.1167, "step": 23521 }, { "epoch": 17.23223443223443, "grad_norm": 1.3671875, "learning_rate": 5.5838148642178925e-05, "loss": 0.1332, "step": 23522 }, { "epoch": 17.232967032967032, "grad_norm": 0.88671875, "learning_rate": 5.5834742250195526e-05, "loss": 0.1304, "step": 23523 }, { "epoch": 17.233699633699633, "grad_norm": 1.390625, "learning_rate": 5.583133583265163e-05, "loss": 0.12, "step": 23524 }, { "epoch": 17.234432234432234, "grad_norm": 1.2265625, "learning_rate": 5.582792938956356e-05, "loss": 0.1055, "step": 23525 }, { "epoch": 17.235164835164834, "grad_norm": 0.9375, "learning_rate": 5.582452292094762e-05, "loss": 0.1088, "step": 23526 }, { "epoch": 17.235897435897435, "grad_norm": 0.80859375, "learning_rate": 5.582111642682015e-05, "loss": 0.1212, "step": 23527 }, { "epoch": 17.236630036630036, "grad_norm": 0.875, "learning_rate": 5.581770990719747e-05, "loss": 0.099, "step": 23528 }, { "epoch": 17.237362637362637, "grad_norm": 0.78515625, "learning_rate": 5.581430336209588e-05, "loss": 0.1254, "step": 23529 }, { "epoch": 17.238095238095237, "grad_norm": 0.7890625, "learning_rate": 5.581089679153173e-05, "loss": 0.1226, "step": 23530 }, { "epoch": 17.238827838827838, "grad_norm": 1.0234375, "learning_rate": 5.580749019552133e-05, "loss": 0.1281, "step": 23531 }, { "epoch": 17.23956043956044, "grad_norm": 1.3125, "learning_rate": 5.580408357408099e-05, "loss": 0.1317, "step": 23532 }, { "epoch": 17.24029304029304, "grad_norm": 1.6875, "learning_rate": 5.580067692722707e-05, "loss": 0.1226, "step": 23533 }, { "epoch": 17.24102564102564, "grad_norm": 1.1796875, "learning_rate": 5.579727025497584e-05, "loss": 0.1115, "step": 23534 }, { "epoch": 17.24175824175824, "grad_norm": 1.28125, "learning_rate": 5.579386355734365e-05, "loss": 0.1216, "step": 23535 }, { "epoch": 17.24249084249084, "grad_norm": 1.0703125, "learning_rate": 5.579045683434685e-05, "loss": 0.1229, "step": 23536 }, { "epoch": 17.243223443223442, "grad_norm": 1.71875, "learning_rate": 5.5787050086001705e-05, "loss": 0.1547, "step": 23537 }, { "epoch": 17.243956043956043, "grad_norm": 1.828125, "learning_rate": 5.5783643312324585e-05, "loss": 0.1191, "step": 23538 }, { "epoch": 17.244688644688644, "grad_norm": 1.4375, "learning_rate": 5.5780236513331785e-05, "loss": 0.1253, "step": 23539 }, { "epoch": 17.245421245421245, "grad_norm": 1.6015625, "learning_rate": 5.577682968903964e-05, "loss": 0.1541, "step": 23540 }, { "epoch": 17.246153846153845, "grad_norm": 1.609375, "learning_rate": 5.5773422839464475e-05, "loss": 0.1219, "step": 23541 }, { "epoch": 17.246886446886446, "grad_norm": 0.7265625, "learning_rate": 5.57700159646226e-05, "loss": 0.0952, "step": 23542 }, { "epoch": 17.247619047619047, "grad_norm": 0.9921875, "learning_rate": 5.576660906453037e-05, "loss": 0.1174, "step": 23543 }, { "epoch": 17.248351648351647, "grad_norm": 0.859375, "learning_rate": 5.576320213920407e-05, "loss": 0.1391, "step": 23544 }, { "epoch": 17.249084249084248, "grad_norm": 0.88671875, "learning_rate": 5.575979518866005e-05, "loss": 0.1194, "step": 23545 }, { "epoch": 17.24981684981685, "grad_norm": 1.421875, "learning_rate": 5.575638821291462e-05, "loss": 0.1345, "step": 23546 }, { "epoch": 17.25054945054945, "grad_norm": 0.71875, "learning_rate": 5.575298121198411e-05, "loss": 0.1099, "step": 23547 }, { "epoch": 17.25128205128205, "grad_norm": 1.6015625, "learning_rate": 5.574957418588484e-05, "loss": 0.1114, "step": 23548 }, { "epoch": 17.25201465201465, "grad_norm": 0.96484375, "learning_rate": 5.574616713463313e-05, "loss": 0.125, "step": 23549 }, { "epoch": 17.252747252747252, "grad_norm": 0.859375, "learning_rate": 5.5742760058245314e-05, "loss": 0.1589, "step": 23550 }, { "epoch": 17.253479853479853, "grad_norm": 1.046875, "learning_rate": 5.573935295673773e-05, "loss": 0.1197, "step": 23551 }, { "epoch": 17.254212454212453, "grad_norm": 0.86328125, "learning_rate": 5.573594583012667e-05, "loss": 0.1099, "step": 23552 }, { "epoch": 17.254945054945054, "grad_norm": 1.1640625, "learning_rate": 5.5732538678428484e-05, "loss": 0.1355, "step": 23553 }, { "epoch": 17.255677655677655, "grad_norm": 1.109375, "learning_rate": 5.5729131501659484e-05, "loss": 0.1277, "step": 23554 }, { "epoch": 17.256410256410255, "grad_norm": 0.8046875, "learning_rate": 5.572572429983599e-05, "loss": 0.1056, "step": 23555 }, { "epoch": 17.257142857142856, "grad_norm": 0.82421875, "learning_rate": 5.572231707297435e-05, "loss": 0.1077, "step": 23556 }, { "epoch": 17.257875457875457, "grad_norm": 1.03125, "learning_rate": 5.571890982109086e-05, "loss": 0.1212, "step": 23557 }, { "epoch": 17.258608058608058, "grad_norm": 1.3984375, "learning_rate": 5.571550254420187e-05, "loss": 0.1518, "step": 23558 }, { "epoch": 17.25934065934066, "grad_norm": 0.80859375, "learning_rate": 5.57120952423237e-05, "loss": 0.0972, "step": 23559 }, { "epoch": 17.26007326007326, "grad_norm": 1.2109375, "learning_rate": 5.5708687915472656e-05, "loss": 0.1212, "step": 23560 }, { "epoch": 17.26080586080586, "grad_norm": 1.0234375, "learning_rate": 5.570528056366508e-05, "loss": 0.1085, "step": 23561 }, { "epoch": 17.26153846153846, "grad_norm": 1.078125, "learning_rate": 5.57018731869173e-05, "loss": 0.147, "step": 23562 }, { "epoch": 17.26227106227106, "grad_norm": 1.0234375, "learning_rate": 5.569846578524562e-05, "loss": 0.124, "step": 23563 }, { "epoch": 17.263003663003662, "grad_norm": 0.8828125, "learning_rate": 5.569505835866642e-05, "loss": 0.1121, "step": 23564 }, { "epoch": 17.263736263736263, "grad_norm": 0.890625, "learning_rate": 5.569165090719597e-05, "loss": 0.113, "step": 23565 }, { "epoch": 17.264468864468864, "grad_norm": 1.015625, "learning_rate": 5.568824343085063e-05, "loss": 0.1305, "step": 23566 }, { "epoch": 17.265201465201464, "grad_norm": 1.3671875, "learning_rate": 5.5684835929646695e-05, "loss": 0.1182, "step": 23567 }, { "epoch": 17.265934065934065, "grad_norm": 1.09375, "learning_rate": 5.568142840360051e-05, "loss": 0.1112, "step": 23568 }, { "epoch": 17.266666666666666, "grad_norm": 0.84765625, "learning_rate": 5.5678020852728416e-05, "loss": 0.1075, "step": 23569 }, { "epoch": 17.267399267399266, "grad_norm": 0.9296875, "learning_rate": 5.56746132770467e-05, "loss": 0.1329, "step": 23570 }, { "epoch": 17.268131868131867, "grad_norm": 1.25, "learning_rate": 5.567120567657174e-05, "loss": 0.1207, "step": 23571 }, { "epoch": 17.268864468864468, "grad_norm": 0.76171875, "learning_rate": 5.566779805131983e-05, "loss": 0.1383, "step": 23572 }, { "epoch": 17.26959706959707, "grad_norm": 1.5390625, "learning_rate": 5.5664390401307285e-05, "loss": 0.1128, "step": 23573 }, { "epoch": 17.27032967032967, "grad_norm": 1.375, "learning_rate": 5.566098272655046e-05, "loss": 0.1262, "step": 23574 }, { "epoch": 17.27106227106227, "grad_norm": 1.46875, "learning_rate": 5.5657575027065676e-05, "loss": 0.1047, "step": 23575 }, { "epoch": 17.27179487179487, "grad_norm": 1.0703125, "learning_rate": 5.565416730286925e-05, "loss": 0.0898, "step": 23576 }, { "epoch": 17.27252747252747, "grad_norm": 0.95703125, "learning_rate": 5.5650759553977525e-05, "loss": 0.1153, "step": 23577 }, { "epoch": 17.273260073260072, "grad_norm": 1.6328125, "learning_rate": 5.564735178040681e-05, "loss": 0.1476, "step": 23578 }, { "epoch": 17.273992673992673, "grad_norm": 0.87890625, "learning_rate": 5.564394398217345e-05, "loss": 0.1208, "step": 23579 }, { "epoch": 17.274725274725274, "grad_norm": 1.0390625, "learning_rate": 5.564053615929378e-05, "loss": 0.1502, "step": 23580 }, { "epoch": 17.275457875457874, "grad_norm": 1.53125, "learning_rate": 5.563712831178408e-05, "loss": 0.1221, "step": 23581 }, { "epoch": 17.276190476190475, "grad_norm": 0.921875, "learning_rate": 5.563372043966074e-05, "loss": 0.1324, "step": 23582 }, { "epoch": 17.276923076923076, "grad_norm": 1.09375, "learning_rate": 5.5630312542940036e-05, "loss": 0.1228, "step": 23583 }, { "epoch": 17.277655677655677, "grad_norm": 0.6953125, "learning_rate": 5.5626904621638346e-05, "loss": 0.1337, "step": 23584 }, { "epoch": 17.278388278388277, "grad_norm": 1.0078125, "learning_rate": 5.562349667577197e-05, "loss": 0.1001, "step": 23585 }, { "epoch": 17.279120879120878, "grad_norm": 1.1328125, "learning_rate": 5.5620088705357234e-05, "loss": 0.1032, "step": 23586 }, { "epoch": 17.27985347985348, "grad_norm": 1.8984375, "learning_rate": 5.5616680710410454e-05, "loss": 0.1591, "step": 23587 }, { "epoch": 17.28058608058608, "grad_norm": 1.6640625, "learning_rate": 5.5613272690948006e-05, "loss": 0.1714, "step": 23588 }, { "epoch": 17.28131868131868, "grad_norm": 1.390625, "learning_rate": 5.5609864646986167e-05, "loss": 0.1303, "step": 23589 }, { "epoch": 17.28205128205128, "grad_norm": 1.21875, "learning_rate": 5.560645657854129e-05, "loss": 0.1661, "step": 23590 }, { "epoch": 17.282783882783882, "grad_norm": 1.7265625, "learning_rate": 5.560304848562972e-05, "loss": 0.1311, "step": 23591 }, { "epoch": 17.283516483516483, "grad_norm": 1.21875, "learning_rate": 5.559964036826776e-05, "loss": 0.1348, "step": 23592 }, { "epoch": 17.284249084249083, "grad_norm": 1.0390625, "learning_rate": 5.559623222647176e-05, "loss": 0.1497, "step": 23593 }, { "epoch": 17.284981684981684, "grad_norm": 0.91796875, "learning_rate": 5.559282406025802e-05, "loss": 0.1381, "step": 23594 }, { "epoch": 17.285714285714285, "grad_norm": 0.96484375, "learning_rate": 5.5589415869642894e-05, "loss": 0.1092, "step": 23595 }, { "epoch": 17.286446886446885, "grad_norm": 1.0078125, "learning_rate": 5.5586007654642714e-05, "loss": 0.1174, "step": 23596 }, { "epoch": 17.287179487179486, "grad_norm": 1.1796875, "learning_rate": 5.5582599415273795e-05, "loss": 0.1225, "step": 23597 }, { "epoch": 17.287912087912087, "grad_norm": 0.89453125, "learning_rate": 5.557919115155248e-05, "loss": 0.1174, "step": 23598 }, { "epoch": 17.288644688644688, "grad_norm": 1.0703125, "learning_rate": 5.5575782863495075e-05, "loss": 0.1182, "step": 23599 }, { "epoch": 17.28937728937729, "grad_norm": 1.4921875, "learning_rate": 5.557237455111795e-05, "loss": 0.149, "step": 23600 }, { "epoch": 17.29010989010989, "grad_norm": 0.83203125, "learning_rate": 5.556896621443739e-05, "loss": 0.1094, "step": 23601 }, { "epoch": 17.29084249084249, "grad_norm": 1.203125, "learning_rate": 5.556555785346977e-05, "loss": 0.1391, "step": 23602 }, { "epoch": 17.29157509157509, "grad_norm": 0.9609375, "learning_rate": 5.556214946823139e-05, "loss": 0.1292, "step": 23603 }, { "epoch": 17.29230769230769, "grad_norm": 1.0703125, "learning_rate": 5.55587410587386e-05, "loss": 0.1298, "step": 23604 }, { "epoch": 17.293040293040292, "grad_norm": 0.828125, "learning_rate": 5.55553326250077e-05, "loss": 0.1101, "step": 23605 }, { "epoch": 17.293772893772893, "grad_norm": 1.125, "learning_rate": 5.555192416705507e-05, "loss": 0.1159, "step": 23606 }, { "epoch": 17.294505494505493, "grad_norm": 1.6484375, "learning_rate": 5.5548515684897e-05, "loss": 0.1113, "step": 23607 }, { "epoch": 17.295238095238094, "grad_norm": 1.484375, "learning_rate": 5.5545107178549826e-05, "loss": 0.1394, "step": 23608 }, { "epoch": 17.295970695970695, "grad_norm": 0.890625, "learning_rate": 5.554169864802989e-05, "loss": 0.1198, "step": 23609 }, { "epoch": 17.296703296703296, "grad_norm": 1.453125, "learning_rate": 5.553829009335353e-05, "loss": 0.1354, "step": 23610 }, { "epoch": 17.297435897435896, "grad_norm": 1.1484375, "learning_rate": 5.5534881514537066e-05, "loss": 0.1351, "step": 23611 }, { "epoch": 17.298168498168497, "grad_norm": 0.88671875, "learning_rate": 5.5531472911596826e-05, "loss": 0.0907, "step": 23612 }, { "epoch": 17.298901098901098, "grad_norm": 1.359375, "learning_rate": 5.552806428454914e-05, "loss": 0.1334, "step": 23613 }, { "epoch": 17.2996336996337, "grad_norm": 1.0078125, "learning_rate": 5.552465563341036e-05, "loss": 0.1158, "step": 23614 }, { "epoch": 17.3003663003663, "grad_norm": 1.6328125, "learning_rate": 5.552124695819679e-05, "loss": 0.1183, "step": 23615 }, { "epoch": 17.3010989010989, "grad_norm": 0.56640625, "learning_rate": 5.5517838258924815e-05, "loss": 0.1119, "step": 23616 }, { "epoch": 17.3018315018315, "grad_norm": 0.984375, "learning_rate": 5.551442953561069e-05, "loss": 0.1317, "step": 23617 }, { "epoch": 17.3025641025641, "grad_norm": 0.69140625, "learning_rate": 5.5511020788270794e-05, "loss": 0.1205, "step": 23618 }, { "epoch": 17.303296703296702, "grad_norm": 1.046875, "learning_rate": 5.550761201692147e-05, "loss": 0.1227, "step": 23619 }, { "epoch": 17.304029304029303, "grad_norm": 1.0859375, "learning_rate": 5.550420322157902e-05, "loss": 0.1312, "step": 23620 }, { "epoch": 17.304761904761904, "grad_norm": 0.7890625, "learning_rate": 5.55007944022598e-05, "loss": 0.1098, "step": 23621 }, { "epoch": 17.305494505494504, "grad_norm": 0.8984375, "learning_rate": 5.549738555898012e-05, "loss": 0.1272, "step": 23622 }, { "epoch": 17.306227106227105, "grad_norm": 0.87890625, "learning_rate": 5.549397669175633e-05, "loss": 0.1501, "step": 23623 }, { "epoch": 17.306959706959706, "grad_norm": 0.7265625, "learning_rate": 5.5490567800604764e-05, "loss": 0.1233, "step": 23624 }, { "epoch": 17.307692307692307, "grad_norm": 0.68359375, "learning_rate": 5.5487158885541746e-05, "loss": 0.1312, "step": 23625 }, { "epoch": 17.308424908424907, "grad_norm": 0.65234375, "learning_rate": 5.54837499465836e-05, "loss": 0.112, "step": 23626 }, { "epoch": 17.309157509157508, "grad_norm": 1.609375, "learning_rate": 5.548034098374668e-05, "loss": 0.125, "step": 23627 }, { "epoch": 17.30989010989011, "grad_norm": 1.375, "learning_rate": 5.547693199704732e-05, "loss": 0.1365, "step": 23628 }, { "epoch": 17.31062271062271, "grad_norm": 1.3671875, "learning_rate": 5.547352298650184e-05, "loss": 0.1192, "step": 23629 }, { "epoch": 17.31135531135531, "grad_norm": 1.1328125, "learning_rate": 5.5470113952126566e-05, "loss": 0.1208, "step": 23630 }, { "epoch": 17.31208791208791, "grad_norm": 1.890625, "learning_rate": 5.546670489393786e-05, "loss": 0.1352, "step": 23631 }, { "epoch": 17.31282051282051, "grad_norm": 1.1640625, "learning_rate": 5.546329581195204e-05, "loss": 0.1144, "step": 23632 }, { "epoch": 17.313553113553112, "grad_norm": 0.93359375, "learning_rate": 5.545988670618544e-05, "loss": 0.1332, "step": 23633 }, { "epoch": 17.314285714285713, "grad_norm": 0.8359375, "learning_rate": 5.5456477576654396e-05, "loss": 0.1405, "step": 23634 }, { "epoch": 17.315018315018314, "grad_norm": 1.8828125, "learning_rate": 5.545306842337522e-05, "loss": 0.1321, "step": 23635 }, { "epoch": 17.315750915750915, "grad_norm": 1.3046875, "learning_rate": 5.5449659246364296e-05, "loss": 0.1494, "step": 23636 }, { "epoch": 17.316483516483515, "grad_norm": 1.390625, "learning_rate": 5.5446250045637926e-05, "loss": 0.1556, "step": 23637 }, { "epoch": 17.317216117216116, "grad_norm": 0.97265625, "learning_rate": 5.544284082121243e-05, "loss": 0.1247, "step": 23638 }, { "epoch": 17.317948717948717, "grad_norm": 1.1015625, "learning_rate": 5.543943157310416e-05, "loss": 0.1355, "step": 23639 }, { "epoch": 17.318681318681318, "grad_norm": 1.2265625, "learning_rate": 5.543602230132946e-05, "loss": 0.1849, "step": 23640 }, { "epoch": 17.31941391941392, "grad_norm": 0.83984375, "learning_rate": 5.5432613005904664e-05, "loss": 0.1282, "step": 23641 }, { "epoch": 17.32014652014652, "grad_norm": 0.64453125, "learning_rate": 5.54292036868461e-05, "loss": 0.1156, "step": 23642 }, { "epoch": 17.32087912087912, "grad_norm": 0.94921875, "learning_rate": 5.542579434417009e-05, "loss": 0.1053, "step": 23643 }, { "epoch": 17.32161172161172, "grad_norm": 1.15625, "learning_rate": 5.5422384977892994e-05, "loss": 0.1363, "step": 23644 }, { "epoch": 17.32234432234432, "grad_norm": 1.953125, "learning_rate": 5.541897558803114e-05, "loss": 0.1427, "step": 23645 }, { "epoch": 17.323076923076922, "grad_norm": 1.515625, "learning_rate": 5.541556617460084e-05, "loss": 0.1449, "step": 23646 }, { "epoch": 17.323809523809523, "grad_norm": 1.2578125, "learning_rate": 5.541215673761847e-05, "loss": 0.1211, "step": 23647 }, { "epoch": 17.324542124542123, "grad_norm": 1.546875, "learning_rate": 5.540874727710033e-05, "loss": 0.1157, "step": 23648 }, { "epoch": 17.325274725274724, "grad_norm": 1.203125, "learning_rate": 5.540533779306277e-05, "loss": 0.0989, "step": 23649 }, { "epoch": 17.326007326007325, "grad_norm": 1.5078125, "learning_rate": 5.540192828552214e-05, "loss": 0.1435, "step": 23650 }, { "epoch": 17.326739926739926, "grad_norm": 1.2421875, "learning_rate": 5.5398518754494744e-05, "loss": 0.1168, "step": 23651 }, { "epoch": 17.327472527472526, "grad_norm": 2.03125, "learning_rate": 5.539510919999694e-05, "loss": 0.1244, "step": 23652 }, { "epoch": 17.328205128205127, "grad_norm": 1.6328125, "learning_rate": 5.5391699622045056e-05, "loss": 0.1487, "step": 23653 }, { "epoch": 17.328937728937728, "grad_norm": 0.984375, "learning_rate": 5.538829002065545e-05, "loss": 0.1538, "step": 23654 }, { "epoch": 17.32967032967033, "grad_norm": 1.75, "learning_rate": 5.538488039584444e-05, "loss": 0.1166, "step": 23655 }, { "epoch": 17.33040293040293, "grad_norm": 1.125, "learning_rate": 5.538147074762836e-05, "loss": 0.1338, "step": 23656 }, { "epoch": 17.33113553113553, "grad_norm": 0.71484375, "learning_rate": 5.537806107602355e-05, "loss": 0.1207, "step": 23657 }, { "epoch": 17.33186813186813, "grad_norm": 1.6796875, "learning_rate": 5.537465138104635e-05, "loss": 0.1525, "step": 23658 }, { "epoch": 17.33260073260073, "grad_norm": 1.0625, "learning_rate": 5.537124166271309e-05, "loss": 0.126, "step": 23659 }, { "epoch": 17.333333333333332, "grad_norm": 0.8671875, "learning_rate": 5.536783192104013e-05, "loss": 0.1411, "step": 23660 }, { "epoch": 17.334065934065933, "grad_norm": 1.53125, "learning_rate": 5.536442215604376e-05, "loss": 0.1213, "step": 23661 }, { "epoch": 17.334798534798534, "grad_norm": 1.171875, "learning_rate": 5.536101236774037e-05, "loss": 0.1428, "step": 23662 }, { "epoch": 17.335531135531134, "grad_norm": 1.59375, "learning_rate": 5.5357602556146274e-05, "loss": 0.1348, "step": 23663 }, { "epoch": 17.336263736263735, "grad_norm": 0.90625, "learning_rate": 5.53541927212778e-05, "loss": 0.1144, "step": 23664 }, { "epoch": 17.336996336996336, "grad_norm": 1.1484375, "learning_rate": 5.535078286315128e-05, "loss": 0.1276, "step": 23665 }, { "epoch": 17.337728937728937, "grad_norm": 0.9921875, "learning_rate": 5.534737298178308e-05, "loss": 0.1186, "step": 23666 }, { "epoch": 17.338461538461537, "grad_norm": 1.6640625, "learning_rate": 5.534396307718953e-05, "loss": 0.1551, "step": 23667 }, { "epoch": 17.339194139194138, "grad_norm": 1.125, "learning_rate": 5.5340553149386974e-05, "loss": 0.1071, "step": 23668 }, { "epoch": 17.33992673992674, "grad_norm": 1.359375, "learning_rate": 5.533714319839172e-05, "loss": 0.1221, "step": 23669 }, { "epoch": 17.34065934065934, "grad_norm": 1.1328125, "learning_rate": 5.533373322422013e-05, "loss": 0.1232, "step": 23670 }, { "epoch": 17.34139194139194, "grad_norm": 1.0078125, "learning_rate": 5.533032322688856e-05, "loss": 0.1248, "step": 23671 }, { "epoch": 17.34212454212454, "grad_norm": 1.2109375, "learning_rate": 5.532691320641329e-05, "loss": 0.1064, "step": 23672 }, { "epoch": 17.34285714285714, "grad_norm": 1.0390625, "learning_rate": 5.532350316281071e-05, "loss": 0.1443, "step": 23673 }, { "epoch": 17.343589743589742, "grad_norm": 1.1484375, "learning_rate": 5.5320093096097135e-05, "loss": 0.0982, "step": 23674 }, { "epoch": 17.344322344322343, "grad_norm": 0.6328125, "learning_rate": 5.531668300628891e-05, "loss": 0.1311, "step": 23675 }, { "epoch": 17.345054945054944, "grad_norm": 1.046875, "learning_rate": 5.531327289340239e-05, "loss": 0.1166, "step": 23676 }, { "epoch": 17.345787545787545, "grad_norm": 0.828125, "learning_rate": 5.530986275745389e-05, "loss": 0.1361, "step": 23677 }, { "epoch": 17.346520146520145, "grad_norm": 0.93359375, "learning_rate": 5.530645259845977e-05, "loss": 0.1371, "step": 23678 }, { "epoch": 17.347252747252746, "grad_norm": 1.109375, "learning_rate": 5.5303042416436346e-05, "loss": 0.1351, "step": 23679 }, { "epoch": 17.347985347985347, "grad_norm": 1.1640625, "learning_rate": 5.5299632211399965e-05, "loss": 0.106, "step": 23680 }, { "epoch": 17.348717948717947, "grad_norm": 1.65625, "learning_rate": 5.529622198336698e-05, "loss": 0.1735, "step": 23681 }, { "epoch": 17.349450549450548, "grad_norm": 1.1484375, "learning_rate": 5.529281173235372e-05, "loss": 0.1244, "step": 23682 }, { "epoch": 17.35018315018315, "grad_norm": 1.390625, "learning_rate": 5.5289401458376523e-05, "loss": 0.1276, "step": 23683 }, { "epoch": 17.35091575091575, "grad_norm": 1.375, "learning_rate": 5.528599116145173e-05, "loss": 0.1336, "step": 23684 }, { "epoch": 17.35164835164835, "grad_norm": 0.640625, "learning_rate": 5.528258084159568e-05, "loss": 0.1093, "step": 23685 }, { "epoch": 17.35238095238095, "grad_norm": 0.8046875, "learning_rate": 5.5279170498824725e-05, "loss": 0.1206, "step": 23686 }, { "epoch": 17.353113553113552, "grad_norm": 1.53125, "learning_rate": 5.52757601331552e-05, "loss": 0.1068, "step": 23687 }, { "epoch": 17.353846153846153, "grad_norm": 0.796875, "learning_rate": 5.5272349744603425e-05, "loss": 0.1093, "step": 23688 }, { "epoch": 17.354578754578753, "grad_norm": 1.5703125, "learning_rate": 5.526893933318575e-05, "loss": 0.1522, "step": 23689 }, { "epoch": 17.355311355311354, "grad_norm": 0.84375, "learning_rate": 5.526552889891855e-05, "loss": 0.112, "step": 23690 }, { "epoch": 17.356043956043955, "grad_norm": 1.15625, "learning_rate": 5.52621184418181e-05, "loss": 0.1474, "step": 23691 }, { "epoch": 17.356776556776556, "grad_norm": 0.87890625, "learning_rate": 5.52587079619008e-05, "loss": 0.1053, "step": 23692 }, { "epoch": 17.357509157509156, "grad_norm": 0.7109375, "learning_rate": 5.525529745918296e-05, "loss": 0.1126, "step": 23693 }, { "epoch": 17.358241758241757, "grad_norm": 1.296875, "learning_rate": 5.525188693368095e-05, "loss": 0.1247, "step": 23694 }, { "epoch": 17.358974358974358, "grad_norm": 1.203125, "learning_rate": 5.524847638541106e-05, "loss": 0.1044, "step": 23695 }, { "epoch": 17.35970695970696, "grad_norm": 0.83984375, "learning_rate": 5.524506581438969e-05, "loss": 0.1086, "step": 23696 }, { "epoch": 17.36043956043956, "grad_norm": 0.91796875, "learning_rate": 5.524165522063312e-05, "loss": 0.1291, "step": 23697 }, { "epoch": 17.36117216117216, "grad_norm": 1.0859375, "learning_rate": 5.523824460415774e-05, "loss": 0.1114, "step": 23698 }, { "epoch": 17.36190476190476, "grad_norm": 0.8515625, "learning_rate": 5.5234833964979885e-05, "loss": 0.1247, "step": 23699 }, { "epoch": 17.36263736263736, "grad_norm": 1.375, "learning_rate": 5.523142330311586e-05, "loss": 0.1392, "step": 23700 }, { "epoch": 17.363369963369962, "grad_norm": 0.89453125, "learning_rate": 5.5228012618582056e-05, "loss": 0.1132, "step": 23701 }, { "epoch": 17.364102564102563, "grad_norm": 0.8359375, "learning_rate": 5.522460191139478e-05, "loss": 0.1435, "step": 23702 }, { "epoch": 17.364835164835164, "grad_norm": 1.03125, "learning_rate": 5.52211911815704e-05, "loss": 0.1315, "step": 23703 }, { "epoch": 17.365567765567764, "grad_norm": 1.25, "learning_rate": 5.521778042912521e-05, "loss": 0.1037, "step": 23704 }, { "epoch": 17.366300366300365, "grad_norm": 1.125, "learning_rate": 5.5214369654075616e-05, "loss": 0.1328, "step": 23705 }, { "epoch": 17.367032967032966, "grad_norm": 1.328125, "learning_rate": 5.521095885643791e-05, "loss": 0.1433, "step": 23706 }, { "epoch": 17.367765567765566, "grad_norm": 0.83203125, "learning_rate": 5.5207548036228474e-05, "loss": 0.1243, "step": 23707 }, { "epoch": 17.368498168498167, "grad_norm": 0.85546875, "learning_rate": 5.520413719346361e-05, "loss": 0.1185, "step": 23708 }, { "epoch": 17.369230769230768, "grad_norm": 1.25, "learning_rate": 5.520072632815969e-05, "loss": 0.1195, "step": 23709 }, { "epoch": 17.36996336996337, "grad_norm": 0.84375, "learning_rate": 5.5197315440333054e-05, "loss": 0.1093, "step": 23710 }, { "epoch": 17.37069597069597, "grad_norm": 0.74609375, "learning_rate": 5.5193904530000026e-05, "loss": 0.1124, "step": 23711 }, { "epoch": 17.37142857142857, "grad_norm": 1.375, "learning_rate": 5.519049359717697e-05, "loss": 0.134, "step": 23712 }, { "epoch": 17.37216117216117, "grad_norm": 0.90234375, "learning_rate": 5.518708264188021e-05, "loss": 0.1204, "step": 23713 }, { "epoch": 17.37289377289377, "grad_norm": 0.875, "learning_rate": 5.51836716641261e-05, "loss": 0.1372, "step": 23714 }, { "epoch": 17.373626373626372, "grad_norm": 1.0546875, "learning_rate": 5.518026066393098e-05, "loss": 0.1119, "step": 23715 }, { "epoch": 17.374358974358973, "grad_norm": 1.1328125, "learning_rate": 5.51768496413112e-05, "loss": 0.1438, "step": 23716 }, { "epoch": 17.375091575091574, "grad_norm": 1.0859375, "learning_rate": 5.51734385962831e-05, "loss": 0.1108, "step": 23717 }, { "epoch": 17.375824175824174, "grad_norm": 0.90625, "learning_rate": 5.5170027528863034e-05, "loss": 0.1557, "step": 23718 }, { "epoch": 17.376556776556775, "grad_norm": 2.515625, "learning_rate": 5.5166616439067306e-05, "loss": 0.1267, "step": 23719 }, { "epoch": 17.377289377289376, "grad_norm": 1.3203125, "learning_rate": 5.5163205326912306e-05, "loss": 0.0995, "step": 23720 }, { "epoch": 17.378021978021977, "grad_norm": 0.921875, "learning_rate": 5.515979419241434e-05, "loss": 0.1525, "step": 23721 }, { "epoch": 17.378754578754577, "grad_norm": 0.8203125, "learning_rate": 5.515638303558979e-05, "loss": 0.1349, "step": 23722 }, { "epoch": 17.379487179487178, "grad_norm": 1.28125, "learning_rate": 5.515297185645497e-05, "loss": 0.1557, "step": 23723 }, { "epoch": 17.38021978021978, "grad_norm": 1.34375, "learning_rate": 5.514956065502623e-05, "loss": 0.1175, "step": 23724 }, { "epoch": 17.38095238095238, "grad_norm": 0.92578125, "learning_rate": 5.514614943131993e-05, "loss": 0.1362, "step": 23725 }, { "epoch": 17.38168498168498, "grad_norm": 1.109375, "learning_rate": 5.5142738185352385e-05, "loss": 0.1239, "step": 23726 }, { "epoch": 17.38241758241758, "grad_norm": 1.09375, "learning_rate": 5.513932691713998e-05, "loss": 0.1302, "step": 23727 }, { "epoch": 17.383150183150182, "grad_norm": 1.75, "learning_rate": 5.513591562669902e-05, "loss": 0.1233, "step": 23728 }, { "epoch": 17.383882783882783, "grad_norm": 1.546875, "learning_rate": 5.513250431404586e-05, "loss": 0.1246, "step": 23729 }, { "epoch": 17.384615384615383, "grad_norm": 1.5078125, "learning_rate": 5.512909297919687e-05, "loss": 0.1518, "step": 23730 }, { "epoch": 17.385347985347984, "grad_norm": 1.703125, "learning_rate": 5.512568162216838e-05, "loss": 0.1402, "step": 23731 }, { "epoch": 17.386080586080585, "grad_norm": 1.3359375, "learning_rate": 5.512227024297672e-05, "loss": 0.1096, "step": 23732 }, { "epoch": 17.386813186813185, "grad_norm": 0.859375, "learning_rate": 5.511885884163823e-05, "loss": 0.1179, "step": 23733 }, { "epoch": 17.387545787545786, "grad_norm": 1.875, "learning_rate": 5.5115447418169296e-05, "loss": 0.128, "step": 23734 }, { "epoch": 17.388278388278387, "grad_norm": 0.85546875, "learning_rate": 5.511203597258624e-05, "loss": 0.1013, "step": 23735 }, { "epoch": 17.389010989010988, "grad_norm": 1.078125, "learning_rate": 5.5108624504905385e-05, "loss": 0.1351, "step": 23736 }, { "epoch": 17.38974358974359, "grad_norm": 1.0546875, "learning_rate": 5.51052130151431e-05, "loss": 0.1211, "step": 23737 }, { "epoch": 17.39047619047619, "grad_norm": 1.359375, "learning_rate": 5.510180150331575e-05, "loss": 0.1236, "step": 23738 }, { "epoch": 17.39120879120879, "grad_norm": 1.15625, "learning_rate": 5.5098389969439634e-05, "loss": 0.1017, "step": 23739 }, { "epoch": 17.39194139194139, "grad_norm": 0.59765625, "learning_rate": 5.509497841353113e-05, "loss": 0.1046, "step": 23740 }, { "epoch": 17.39267399267399, "grad_norm": 0.83984375, "learning_rate": 5.509156683560658e-05, "loss": 0.1001, "step": 23741 }, { "epoch": 17.393406593406592, "grad_norm": 0.8515625, "learning_rate": 5.508815523568231e-05, "loss": 0.1176, "step": 23742 }, { "epoch": 17.394139194139193, "grad_norm": 1.21875, "learning_rate": 5.5084743613774706e-05, "loss": 0.113, "step": 23743 }, { "epoch": 17.394871794871793, "grad_norm": 1.265625, "learning_rate": 5.5081331969900086e-05, "loss": 0.1416, "step": 23744 }, { "epoch": 17.395604395604394, "grad_norm": 0.82421875, "learning_rate": 5.5077920304074784e-05, "loss": 0.0973, "step": 23745 }, { "epoch": 17.396336996336995, "grad_norm": 0.9609375, "learning_rate": 5.507450861631519e-05, "loss": 0.1167, "step": 23746 }, { "epoch": 17.397069597069596, "grad_norm": 0.8046875, "learning_rate": 5.507109690663759e-05, "loss": 0.1241, "step": 23747 }, { "epoch": 17.397802197802196, "grad_norm": 0.98828125, "learning_rate": 5.506768517505839e-05, "loss": 0.111, "step": 23748 }, { "epoch": 17.398534798534797, "grad_norm": 1.625, "learning_rate": 5.5064273421593905e-05, "loss": 0.1629, "step": 23749 }, { "epoch": 17.399267399267398, "grad_norm": 1.171875, "learning_rate": 5.506086164626048e-05, "loss": 0.1029, "step": 23750 }, { "epoch": 17.4, "grad_norm": 0.7734375, "learning_rate": 5.505744984907447e-05, "loss": 0.0942, "step": 23751 }, { "epoch": 17.4007326007326, "grad_norm": 0.828125, "learning_rate": 5.505403803005222e-05, "loss": 0.1327, "step": 23752 }, { "epoch": 17.4014652014652, "grad_norm": 1.78125, "learning_rate": 5.50506261892101e-05, "loss": 0.1289, "step": 23753 }, { "epoch": 17.4021978021978, "grad_norm": 1.390625, "learning_rate": 5.5047214326564414e-05, "loss": 0.1285, "step": 23754 }, { "epoch": 17.4029304029304, "grad_norm": 0.85546875, "learning_rate": 5.504380244213152e-05, "loss": 0.1182, "step": 23755 }, { "epoch": 17.403663003663002, "grad_norm": 1.515625, "learning_rate": 5.50403905359278e-05, "loss": 0.1308, "step": 23756 }, { "epoch": 17.404395604395603, "grad_norm": 1.1484375, "learning_rate": 5.503697860796958e-05, "loss": 0.1389, "step": 23757 }, { "epoch": 17.405128205128204, "grad_norm": 1.0703125, "learning_rate": 5.5033566658273195e-05, "loss": 0.1173, "step": 23758 }, { "epoch": 17.405860805860804, "grad_norm": 1.15625, "learning_rate": 5.503015468685502e-05, "loss": 0.1309, "step": 23759 }, { "epoch": 17.406593406593405, "grad_norm": 0.890625, "learning_rate": 5.502674269373136e-05, "loss": 0.1128, "step": 23760 }, { "epoch": 17.407326007326006, "grad_norm": 0.84375, "learning_rate": 5.5023330678918606e-05, "loss": 0.0912, "step": 23761 }, { "epoch": 17.408058608058607, "grad_norm": 1.609375, "learning_rate": 5.50199186424331e-05, "loss": 0.1215, "step": 23762 }, { "epoch": 17.408791208791207, "grad_norm": 1.03125, "learning_rate": 5.501650658429116e-05, "loss": 0.1343, "step": 23763 }, { "epoch": 17.409523809523808, "grad_norm": 0.98046875, "learning_rate": 5.501309450450917e-05, "loss": 0.1172, "step": 23764 }, { "epoch": 17.41025641025641, "grad_norm": 0.7578125, "learning_rate": 5.500968240310345e-05, "loss": 0.1244, "step": 23765 }, { "epoch": 17.41098901098901, "grad_norm": 1.03125, "learning_rate": 5.5006270280090363e-05, "loss": 0.1151, "step": 23766 }, { "epoch": 17.41172161172161, "grad_norm": 1.3671875, "learning_rate": 5.5002858135486254e-05, "loss": 0.1321, "step": 23767 }, { "epoch": 17.41245421245421, "grad_norm": 1.859375, "learning_rate": 5.499944596930748e-05, "loss": 0.1139, "step": 23768 }, { "epoch": 17.41318681318681, "grad_norm": 1.2578125, "learning_rate": 5.499603378157039e-05, "loss": 0.1292, "step": 23769 }, { "epoch": 17.413919413919412, "grad_norm": 1.296875, "learning_rate": 5.4992621572291316e-05, "loss": 0.0968, "step": 23770 }, { "epoch": 17.414652014652013, "grad_norm": 1.1484375, "learning_rate": 5.4989209341486615e-05, "loss": 0.1309, "step": 23771 }, { "epoch": 17.415384615384614, "grad_norm": 0.921875, "learning_rate": 5.498579708917265e-05, "loss": 0.105, "step": 23772 }, { "epoch": 17.416117216117215, "grad_norm": 0.85546875, "learning_rate": 5.498238481536575e-05, "loss": 0.1126, "step": 23773 }, { "epoch": 17.416849816849815, "grad_norm": 0.6796875, "learning_rate": 5.4978972520082275e-05, "loss": 0.1024, "step": 23774 }, { "epoch": 17.417582417582416, "grad_norm": 1.015625, "learning_rate": 5.497556020333857e-05, "loss": 0.1386, "step": 23775 }, { "epoch": 17.418315018315017, "grad_norm": 0.8359375, "learning_rate": 5.497214786515099e-05, "loss": 0.0982, "step": 23776 }, { "epoch": 17.419047619047618, "grad_norm": 0.7734375, "learning_rate": 5.4968735505535884e-05, "loss": 0.1187, "step": 23777 }, { "epoch": 17.41978021978022, "grad_norm": 0.859375, "learning_rate": 5.496532312450959e-05, "loss": 0.1258, "step": 23778 }, { "epoch": 17.42051282051282, "grad_norm": 0.9765625, "learning_rate": 5.496191072208847e-05, "loss": 0.1275, "step": 23779 }, { "epoch": 17.42124542124542, "grad_norm": 1.4375, "learning_rate": 5.495849829828886e-05, "loss": 0.1112, "step": 23780 }, { "epoch": 17.42197802197802, "grad_norm": 0.66796875, "learning_rate": 5.495508585312714e-05, "loss": 0.0954, "step": 23781 }, { "epoch": 17.42271062271062, "grad_norm": 1.640625, "learning_rate": 5.495167338661963e-05, "loss": 0.1167, "step": 23782 }, { "epoch": 17.423443223443222, "grad_norm": 0.84765625, "learning_rate": 5.494826089878271e-05, "loss": 0.1097, "step": 23783 }, { "epoch": 17.424175824175823, "grad_norm": 1.0546875, "learning_rate": 5.494484838963268e-05, "loss": 0.1499, "step": 23784 }, { "epoch": 17.424908424908423, "grad_norm": 0.99609375, "learning_rate": 5.4941435859185964e-05, "loss": 0.1348, "step": 23785 }, { "epoch": 17.425641025641024, "grad_norm": 1.2578125, "learning_rate": 5.493802330745884e-05, "loss": 0.1119, "step": 23786 }, { "epoch": 17.426373626373625, "grad_norm": 0.9609375, "learning_rate": 5.49346107344677e-05, "loss": 0.1372, "step": 23787 }, { "epoch": 17.427106227106226, "grad_norm": 0.97265625, "learning_rate": 5.493119814022889e-05, "loss": 0.094, "step": 23788 }, { "epoch": 17.427838827838826, "grad_norm": 1.25, "learning_rate": 5.4927785524758736e-05, "loss": 0.1074, "step": 23789 }, { "epoch": 17.428571428571427, "grad_norm": 1.484375, "learning_rate": 5.4924372888073614e-05, "loss": 0.1443, "step": 23790 }, { "epoch": 17.429304029304028, "grad_norm": 1.015625, "learning_rate": 5.4920960230189866e-05, "loss": 0.1117, "step": 23791 }, { "epoch": 17.43003663003663, "grad_norm": 1.0078125, "learning_rate": 5.491754755112386e-05, "loss": 0.1289, "step": 23792 }, { "epoch": 17.43076923076923, "grad_norm": 1.265625, "learning_rate": 5.491413485089193e-05, "loss": 0.1594, "step": 23793 }, { "epoch": 17.43150183150183, "grad_norm": 1.6171875, "learning_rate": 5.491072212951042e-05, "loss": 0.1856, "step": 23794 }, { "epoch": 17.43223443223443, "grad_norm": 0.97265625, "learning_rate": 5.49073093869957e-05, "loss": 0.1406, "step": 23795 }, { "epoch": 17.43296703296703, "grad_norm": 0.82421875, "learning_rate": 5.4903896623364116e-05, "loss": 0.127, "step": 23796 }, { "epoch": 17.433699633699632, "grad_norm": 0.58203125, "learning_rate": 5.490048383863201e-05, "loss": 0.0978, "step": 23797 }, { "epoch": 17.434432234432233, "grad_norm": 1.0390625, "learning_rate": 5.489707103281576e-05, "loss": 0.1133, "step": 23798 }, { "epoch": 17.435164835164834, "grad_norm": 0.89453125, "learning_rate": 5.489365820593168e-05, "loss": 0.1198, "step": 23799 }, { "epoch": 17.435897435897434, "grad_norm": 0.87109375, "learning_rate": 5.489024535799615e-05, "loss": 0.1329, "step": 23800 }, { "epoch": 17.436630036630035, "grad_norm": 1.0859375, "learning_rate": 5.488683248902552e-05, "loss": 0.1088, "step": 23801 }, { "epoch": 17.437362637362636, "grad_norm": 1.3046875, "learning_rate": 5.488341959903611e-05, "loss": 0.137, "step": 23802 }, { "epoch": 17.438095238095237, "grad_norm": 1.03125, "learning_rate": 5.4880006688044314e-05, "loss": 0.1039, "step": 23803 }, { "epoch": 17.438827838827837, "grad_norm": 0.74609375, "learning_rate": 5.4876593756066464e-05, "loss": 0.0926, "step": 23804 }, { "epoch": 17.439560439560438, "grad_norm": 1.2265625, "learning_rate": 5.487318080311893e-05, "loss": 0.1011, "step": 23805 }, { "epoch": 17.44029304029304, "grad_norm": 1.234375, "learning_rate": 5.4869767829218046e-05, "loss": 0.1226, "step": 23806 }, { "epoch": 17.44102564102564, "grad_norm": 1.2890625, "learning_rate": 5.486635483438015e-05, "loss": 0.1464, "step": 23807 }, { "epoch": 17.44175824175824, "grad_norm": 0.82421875, "learning_rate": 5.486294181862164e-05, "loss": 0.1239, "step": 23808 }, { "epoch": 17.44249084249084, "grad_norm": 1.1171875, "learning_rate": 5.485952878195884e-05, "loss": 0.1241, "step": 23809 }, { "epoch": 17.44322344322344, "grad_norm": 1.28125, "learning_rate": 5.485611572440809e-05, "loss": 0.1372, "step": 23810 }, { "epoch": 17.443956043956042, "grad_norm": 0.92578125, "learning_rate": 5.485270264598578e-05, "loss": 0.1124, "step": 23811 }, { "epoch": 17.444688644688643, "grad_norm": 1.2734375, "learning_rate": 5.4849289546708214e-05, "loss": 0.1093, "step": 23812 }, { "epoch": 17.445421245421244, "grad_norm": 1.59375, "learning_rate": 5.4845876426591803e-05, "loss": 0.1334, "step": 23813 }, { "epoch": 17.446153846153845, "grad_norm": 1.1015625, "learning_rate": 5.484246328565286e-05, "loss": 0.1243, "step": 23814 }, { "epoch": 17.446886446886445, "grad_norm": 1.3671875, "learning_rate": 5.4839050123907745e-05, "loss": 0.1205, "step": 23815 }, { "epoch": 17.447619047619046, "grad_norm": 1.3125, "learning_rate": 5.4835636941372826e-05, "loss": 0.1262, "step": 23816 }, { "epoch": 17.448351648351647, "grad_norm": 1.5234375, "learning_rate": 5.483222373806444e-05, "loss": 0.1298, "step": 23817 }, { "epoch": 17.449084249084247, "grad_norm": 1.296875, "learning_rate": 5.482881051399895e-05, "loss": 0.1337, "step": 23818 }, { "epoch": 17.449816849816848, "grad_norm": 1.3515625, "learning_rate": 5.48253972691927e-05, "loss": 0.1236, "step": 23819 }, { "epoch": 17.45054945054945, "grad_norm": 0.92578125, "learning_rate": 5.482198400366206e-05, "loss": 0.106, "step": 23820 }, { "epoch": 17.45128205128205, "grad_norm": 1.0859375, "learning_rate": 5.4818570717423376e-05, "loss": 0.1037, "step": 23821 }, { "epoch": 17.45201465201465, "grad_norm": 1.0, "learning_rate": 5.481515741049301e-05, "loss": 0.1135, "step": 23822 }, { "epoch": 17.45274725274725, "grad_norm": 1.875, "learning_rate": 5.4811744082887306e-05, "loss": 0.1497, "step": 23823 }, { "epoch": 17.453479853479852, "grad_norm": 1.0703125, "learning_rate": 5.480833073462261e-05, "loss": 0.1191, "step": 23824 }, { "epoch": 17.454212454212453, "grad_norm": 0.9296875, "learning_rate": 5.480491736571529e-05, "loss": 0.1226, "step": 23825 }, { "epoch": 17.454945054945053, "grad_norm": 1.09375, "learning_rate": 5.48015039761817e-05, "loss": 0.1609, "step": 23826 }, { "epoch": 17.455677655677654, "grad_norm": 0.65234375, "learning_rate": 5.479809056603821e-05, "loss": 0.1157, "step": 23827 }, { "epoch": 17.456410256410255, "grad_norm": 1.3359375, "learning_rate": 5.479467713530112e-05, "loss": 0.1418, "step": 23828 }, { "epoch": 17.457142857142856, "grad_norm": 1.078125, "learning_rate": 5.479126368398685e-05, "loss": 0.1179, "step": 23829 }, { "epoch": 17.457875457875456, "grad_norm": 1.3984375, "learning_rate": 5.4787850212111715e-05, "loss": 0.1517, "step": 23830 }, { "epoch": 17.458608058608057, "grad_norm": 1.5390625, "learning_rate": 5.4784436719692076e-05, "loss": 0.114, "step": 23831 }, { "epoch": 17.459340659340658, "grad_norm": 0.9453125, "learning_rate": 5.478102320674431e-05, "loss": 0.1057, "step": 23832 }, { "epoch": 17.46007326007326, "grad_norm": 1.265625, "learning_rate": 5.4777609673284745e-05, "loss": 0.1298, "step": 23833 }, { "epoch": 17.46080586080586, "grad_norm": 1.1484375, "learning_rate": 5.4774196119329755e-05, "loss": 0.1189, "step": 23834 }, { "epoch": 17.46153846153846, "grad_norm": 0.88671875, "learning_rate": 5.47707825448957e-05, "loss": 0.108, "step": 23835 }, { "epoch": 17.46227106227106, "grad_norm": 0.94921875, "learning_rate": 5.47673689499989e-05, "loss": 0.1182, "step": 23836 }, { "epoch": 17.46300366300366, "grad_norm": 1.1953125, "learning_rate": 5.476395533465575e-05, "loss": 0.125, "step": 23837 }, { "epoch": 17.463736263736262, "grad_norm": 0.92578125, "learning_rate": 5.476054169888257e-05, "loss": 0.1096, "step": 23838 }, { "epoch": 17.464468864468863, "grad_norm": 0.76171875, "learning_rate": 5.475712804269576e-05, "loss": 0.0895, "step": 23839 }, { "epoch": 17.465201465201464, "grad_norm": 1.1015625, "learning_rate": 5.4753714366111644e-05, "loss": 0.1606, "step": 23840 }, { "epoch": 17.465934065934064, "grad_norm": 1.4375, "learning_rate": 5.475030066914658e-05, "loss": 0.1552, "step": 23841 }, { "epoch": 17.466666666666665, "grad_norm": 1.0078125, "learning_rate": 5.4746886951816945e-05, "loss": 0.1315, "step": 23842 }, { "epoch": 17.467399267399266, "grad_norm": 1.359375, "learning_rate": 5.4743473214139065e-05, "loss": 0.1178, "step": 23843 }, { "epoch": 17.468131868131866, "grad_norm": 1.6953125, "learning_rate": 5.474005945612931e-05, "loss": 0.1534, "step": 23844 }, { "epoch": 17.468864468864467, "grad_norm": 1.2109375, "learning_rate": 5.4736645677804055e-05, "loss": 0.1699, "step": 23845 }, { "epoch": 17.469597069597068, "grad_norm": 0.93359375, "learning_rate": 5.4733231879179614e-05, "loss": 0.113, "step": 23846 }, { "epoch": 17.47032967032967, "grad_norm": 1.453125, "learning_rate": 5.4729818060272386e-05, "loss": 0.1151, "step": 23847 }, { "epoch": 17.47106227106227, "grad_norm": 1.8046875, "learning_rate": 5.4726404221098716e-05, "loss": 0.1572, "step": 23848 }, { "epoch": 17.47179487179487, "grad_norm": 1.6171875, "learning_rate": 5.4722990361674954e-05, "loss": 0.1207, "step": 23849 }, { "epoch": 17.47252747252747, "grad_norm": 1.515625, "learning_rate": 5.471957648201745e-05, "loss": 0.1157, "step": 23850 }, { "epoch": 17.47326007326007, "grad_norm": 1.75, "learning_rate": 5.471616258214257e-05, "loss": 0.1511, "step": 23851 }, { "epoch": 17.473992673992672, "grad_norm": 2.28125, "learning_rate": 5.4712748662066695e-05, "loss": 0.1345, "step": 23852 }, { "epoch": 17.474725274725273, "grad_norm": 1.1484375, "learning_rate": 5.470933472180614e-05, "loss": 0.1019, "step": 23853 }, { "epoch": 17.475457875457874, "grad_norm": 1.203125, "learning_rate": 5.470592076137727e-05, "loss": 0.1415, "step": 23854 }, { "epoch": 17.476190476190474, "grad_norm": 1.2109375, "learning_rate": 5.470250678079647e-05, "loss": 0.1221, "step": 23855 }, { "epoch": 17.476923076923075, "grad_norm": 1.375, "learning_rate": 5.4699092780080066e-05, "loss": 0.104, "step": 23856 }, { "epoch": 17.477655677655676, "grad_norm": 1.28125, "learning_rate": 5.469567875924443e-05, "loss": 0.1443, "step": 23857 }, { "epoch": 17.478388278388277, "grad_norm": 2.65625, "learning_rate": 5.469226471830594e-05, "loss": 0.1177, "step": 23858 }, { "epoch": 17.479120879120877, "grad_norm": 1.171875, "learning_rate": 5.468885065728091e-05, "loss": 0.1204, "step": 23859 }, { "epoch": 17.479853479853478, "grad_norm": 1.15625, "learning_rate": 5.468543657618574e-05, "loss": 0.1385, "step": 23860 }, { "epoch": 17.48058608058608, "grad_norm": 1.75, "learning_rate": 5.4682022475036754e-05, "loss": 0.1334, "step": 23861 }, { "epoch": 17.48131868131868, "grad_norm": 0.8984375, "learning_rate": 5.4678608353850335e-05, "loss": 0.1269, "step": 23862 }, { "epoch": 17.48205128205128, "grad_norm": 0.80078125, "learning_rate": 5.467519421264284e-05, "loss": 0.1304, "step": 23863 }, { "epoch": 17.48278388278388, "grad_norm": 0.640625, "learning_rate": 5.46717800514306e-05, "loss": 0.1256, "step": 23864 }, { "epoch": 17.483516483516482, "grad_norm": 0.92578125, "learning_rate": 5.466836587023e-05, "loss": 0.1185, "step": 23865 }, { "epoch": 17.484249084249083, "grad_norm": 1.0234375, "learning_rate": 5.4664951669057405e-05, "loss": 0.1314, "step": 23866 }, { "epoch": 17.484981684981683, "grad_norm": 1.265625, "learning_rate": 5.4661537447929135e-05, "loss": 0.1353, "step": 23867 }, { "epoch": 17.485714285714284, "grad_norm": 1.796875, "learning_rate": 5.465812320686159e-05, "loss": 0.1369, "step": 23868 }, { "epoch": 17.486446886446885, "grad_norm": 1.0859375, "learning_rate": 5.46547089458711e-05, "loss": 0.1456, "step": 23869 }, { "epoch": 17.487179487179485, "grad_norm": 1.2421875, "learning_rate": 5.465129466497404e-05, "loss": 0.1356, "step": 23870 }, { "epoch": 17.487912087912086, "grad_norm": 1.265625, "learning_rate": 5.4647880364186757e-05, "loss": 0.1487, "step": 23871 }, { "epoch": 17.488644688644687, "grad_norm": 0.99609375, "learning_rate": 5.4644466043525624e-05, "loss": 0.1436, "step": 23872 }, { "epoch": 17.489377289377288, "grad_norm": 0.703125, "learning_rate": 5.464105170300701e-05, "loss": 0.1225, "step": 23873 }, { "epoch": 17.49010989010989, "grad_norm": 0.83203125, "learning_rate": 5.463763734264724e-05, "loss": 0.1089, "step": 23874 }, { "epoch": 17.49084249084249, "grad_norm": 1.3515625, "learning_rate": 5.463422296246269e-05, "loss": 0.1555, "step": 23875 }, { "epoch": 17.49157509157509, "grad_norm": 1.0234375, "learning_rate": 5.4630808562469736e-05, "loss": 0.1307, "step": 23876 }, { "epoch": 17.49230769230769, "grad_norm": 0.73046875, "learning_rate": 5.4627394142684694e-05, "loss": 0.1066, "step": 23877 }, { "epoch": 17.493040293040295, "grad_norm": 1.328125, "learning_rate": 5.4623979703123974e-05, "loss": 0.1132, "step": 23878 }, { "epoch": 17.493772893772892, "grad_norm": 1.46875, "learning_rate": 5.462056524380392e-05, "loss": 0.1288, "step": 23879 }, { "epoch": 17.494505494505496, "grad_norm": 1.171875, "learning_rate": 5.461715076474086e-05, "loss": 0.1304, "step": 23880 }, { "epoch": 17.495238095238093, "grad_norm": 0.73046875, "learning_rate": 5.46137362659512e-05, "loss": 0.1132, "step": 23881 }, { "epoch": 17.495970695970698, "grad_norm": 1.0078125, "learning_rate": 5.4610321747451274e-05, "loss": 0.1244, "step": 23882 }, { "epoch": 17.496703296703295, "grad_norm": 1.015625, "learning_rate": 5.460690720925743e-05, "loss": 0.1434, "step": 23883 }, { "epoch": 17.4974358974359, "grad_norm": 1.3359375, "learning_rate": 5.460349265138607e-05, "loss": 0.1653, "step": 23884 }, { "epoch": 17.498168498168496, "grad_norm": 0.9375, "learning_rate": 5.460007807385351e-05, "loss": 0.1128, "step": 23885 }, { "epoch": 17.4989010989011, "grad_norm": 0.55078125, "learning_rate": 5.4596663476676144e-05, "loss": 0.1081, "step": 23886 }, { "epoch": 17.499633699633698, "grad_norm": 0.96484375, "learning_rate": 5.459324885987033e-05, "loss": 0.1394, "step": 23887 }, { "epoch": 17.500366300366302, "grad_norm": 0.89453125, "learning_rate": 5.4589834223452384e-05, "loss": 0.1031, "step": 23888 }, { "epoch": 17.5010989010989, "grad_norm": 1.140625, "learning_rate": 5.458641956743873e-05, "loss": 0.1416, "step": 23889 }, { "epoch": 17.501831501831504, "grad_norm": 0.7421875, "learning_rate": 5.4583004891845666e-05, "loss": 0.1052, "step": 23890 }, { "epoch": 17.5025641025641, "grad_norm": 1.328125, "learning_rate": 5.457959019668962e-05, "loss": 0.1119, "step": 23891 }, { "epoch": 17.503296703296705, "grad_norm": 0.76171875, "learning_rate": 5.457617548198691e-05, "loss": 0.1162, "step": 23892 }, { "epoch": 17.504029304029302, "grad_norm": 0.609375, "learning_rate": 5.4572760747753894e-05, "loss": 0.112, "step": 23893 }, { "epoch": 17.504761904761907, "grad_norm": 1.0078125, "learning_rate": 5.456934599400696e-05, "loss": 0.1123, "step": 23894 }, { "epoch": 17.505494505494504, "grad_norm": 0.87890625, "learning_rate": 5.4565931220762424e-05, "loss": 0.1409, "step": 23895 }, { "epoch": 17.506227106227108, "grad_norm": 0.99609375, "learning_rate": 5.456251642803669e-05, "loss": 0.1279, "step": 23896 }, { "epoch": 17.506959706959705, "grad_norm": 0.7265625, "learning_rate": 5.4559101615846106e-05, "loss": 0.0927, "step": 23897 }, { "epoch": 17.50769230769231, "grad_norm": 1.953125, "learning_rate": 5.455568678420704e-05, "loss": 0.1153, "step": 23898 }, { "epoch": 17.508424908424907, "grad_norm": 1.796875, "learning_rate": 5.455227193313585e-05, "loss": 0.116, "step": 23899 }, { "epoch": 17.50915750915751, "grad_norm": 0.828125, "learning_rate": 5.454885706264889e-05, "loss": 0.1064, "step": 23900 }, { "epoch": 17.509890109890108, "grad_norm": 1.1484375, "learning_rate": 5.454544217276253e-05, "loss": 0.1214, "step": 23901 }, { "epoch": 17.510622710622712, "grad_norm": 1.15625, "learning_rate": 5.454202726349313e-05, "loss": 0.1607, "step": 23902 }, { "epoch": 17.51135531135531, "grad_norm": 1.3125, "learning_rate": 5.453861233485703e-05, "loss": 0.126, "step": 23903 }, { "epoch": 17.512087912087914, "grad_norm": 1.3203125, "learning_rate": 5.453519738687064e-05, "loss": 0.1201, "step": 23904 }, { "epoch": 17.51282051282051, "grad_norm": 1.2890625, "learning_rate": 5.4531782419550295e-05, "loss": 0.1554, "step": 23905 }, { "epoch": 17.513553113553115, "grad_norm": 0.76171875, "learning_rate": 5.452836743291233e-05, "loss": 0.1353, "step": 23906 }, { "epoch": 17.514285714285712, "grad_norm": 0.8359375, "learning_rate": 5.452495242697317e-05, "loss": 0.0932, "step": 23907 }, { "epoch": 17.515018315018317, "grad_norm": 0.87109375, "learning_rate": 5.45215374017491e-05, "loss": 0.1325, "step": 23908 }, { "epoch": 17.515750915750917, "grad_norm": 1.0859375, "learning_rate": 5.451812235725654e-05, "loss": 0.132, "step": 23909 }, { "epoch": 17.516483516483518, "grad_norm": 0.93359375, "learning_rate": 5.4514707293511845e-05, "loss": 0.1183, "step": 23910 }, { "epoch": 17.51721611721612, "grad_norm": 0.92578125, "learning_rate": 5.451129221053136e-05, "loss": 0.1172, "step": 23911 }, { "epoch": 17.51794871794872, "grad_norm": 2.265625, "learning_rate": 5.450787710833147e-05, "loss": 0.1594, "step": 23912 }, { "epoch": 17.51868131868132, "grad_norm": 0.9765625, "learning_rate": 5.450446198692852e-05, "loss": 0.1357, "step": 23913 }, { "epoch": 17.51941391941392, "grad_norm": 1.1015625, "learning_rate": 5.450104684633888e-05, "loss": 0.1613, "step": 23914 }, { "epoch": 17.520146520146522, "grad_norm": 1.390625, "learning_rate": 5.4497631686578896e-05, "loss": 0.1495, "step": 23915 }, { "epoch": 17.520879120879123, "grad_norm": 1.3203125, "learning_rate": 5.449421650766495e-05, "loss": 0.1476, "step": 23916 }, { "epoch": 17.521611721611723, "grad_norm": 0.87890625, "learning_rate": 5.4490801309613425e-05, "loss": 0.1437, "step": 23917 }, { "epoch": 17.522344322344324, "grad_norm": 1.734375, "learning_rate": 5.4487386092440654e-05, "loss": 0.1421, "step": 23918 }, { "epoch": 17.523076923076925, "grad_norm": 1.1328125, "learning_rate": 5.4483970856162993e-05, "loss": 0.1059, "step": 23919 }, { "epoch": 17.523809523809526, "grad_norm": 1.03125, "learning_rate": 5.4480555600796836e-05, "loss": 0.1518, "step": 23920 }, { "epoch": 17.524542124542126, "grad_norm": 1.359375, "learning_rate": 5.447714032635851e-05, "loss": 0.1411, "step": 23921 }, { "epoch": 17.525274725274727, "grad_norm": 1.296875, "learning_rate": 5.44737250328644e-05, "loss": 0.1272, "step": 23922 }, { "epoch": 17.526007326007328, "grad_norm": 1.421875, "learning_rate": 5.447030972033088e-05, "loss": 0.1565, "step": 23923 }, { "epoch": 17.52673992673993, "grad_norm": 1.0546875, "learning_rate": 5.4466894388774295e-05, "loss": 0.1621, "step": 23924 }, { "epoch": 17.52747252747253, "grad_norm": 0.80859375, "learning_rate": 5.446347903821103e-05, "loss": 0.0893, "step": 23925 }, { "epoch": 17.52820512820513, "grad_norm": 1.109375, "learning_rate": 5.446006366865743e-05, "loss": 0.127, "step": 23926 }, { "epoch": 17.52893772893773, "grad_norm": 1.1875, "learning_rate": 5.445664828012986e-05, "loss": 0.1038, "step": 23927 }, { "epoch": 17.52967032967033, "grad_norm": 1.2578125, "learning_rate": 5.44532328726447e-05, "loss": 0.1617, "step": 23928 }, { "epoch": 17.530402930402932, "grad_norm": 0.953125, "learning_rate": 5.4449817446218286e-05, "loss": 0.1022, "step": 23929 }, { "epoch": 17.531135531135533, "grad_norm": 1.0859375, "learning_rate": 5.4446402000867016e-05, "loss": 0.1419, "step": 23930 }, { "epoch": 17.531868131868134, "grad_norm": 1.578125, "learning_rate": 5.444298653660723e-05, "loss": 0.1529, "step": 23931 }, { "epoch": 17.532600732600734, "grad_norm": 1.03125, "learning_rate": 5.4439571053455294e-05, "loss": 0.1197, "step": 23932 }, { "epoch": 17.533333333333335, "grad_norm": 1.421875, "learning_rate": 5.443615555142759e-05, "loss": 0.1454, "step": 23933 }, { "epoch": 17.534065934065936, "grad_norm": 0.8203125, "learning_rate": 5.443274003054047e-05, "loss": 0.1224, "step": 23934 }, { "epoch": 17.534798534798536, "grad_norm": 1.21875, "learning_rate": 5.442932449081031e-05, "loss": 0.1146, "step": 23935 }, { "epoch": 17.535531135531137, "grad_norm": 0.98828125, "learning_rate": 5.442590893225345e-05, "loss": 0.1274, "step": 23936 }, { "epoch": 17.536263736263738, "grad_norm": 1.1171875, "learning_rate": 5.442249335488628e-05, "loss": 0.1451, "step": 23937 }, { "epoch": 17.53699633699634, "grad_norm": 0.98828125, "learning_rate": 5.441907775872517e-05, "loss": 0.094, "step": 23938 }, { "epoch": 17.53772893772894, "grad_norm": 1.0078125, "learning_rate": 5.441566214378646e-05, "loss": 0.1423, "step": 23939 }, { "epoch": 17.53846153846154, "grad_norm": 1.171875, "learning_rate": 5.441224651008652e-05, "loss": 0.1438, "step": 23940 }, { "epoch": 17.53919413919414, "grad_norm": 1.40625, "learning_rate": 5.4408830857641725e-05, "loss": 0.1128, "step": 23941 }, { "epoch": 17.53992673992674, "grad_norm": 1.5, "learning_rate": 5.440541518646844e-05, "loss": 0.1445, "step": 23942 }, { "epoch": 17.540659340659342, "grad_norm": 1.296875, "learning_rate": 5.440199949658304e-05, "loss": 0.1503, "step": 23943 }, { "epoch": 17.541391941391943, "grad_norm": 1.0625, "learning_rate": 5.439858378800188e-05, "loss": 0.1193, "step": 23944 }, { "epoch": 17.542124542124544, "grad_norm": 1.328125, "learning_rate": 5.43951680607413e-05, "loss": 0.1319, "step": 23945 }, { "epoch": 17.542857142857144, "grad_norm": 0.92578125, "learning_rate": 5.4391752314817723e-05, "loss": 0.1058, "step": 23946 }, { "epoch": 17.543589743589745, "grad_norm": 1.03125, "learning_rate": 5.4388336550247464e-05, "loss": 0.1245, "step": 23947 }, { "epoch": 17.544322344322346, "grad_norm": 0.90234375, "learning_rate": 5.438492076704689e-05, "loss": 0.1263, "step": 23948 }, { "epoch": 17.545054945054947, "grad_norm": 1.6796875, "learning_rate": 5.438150496523242e-05, "loss": 0.1442, "step": 23949 }, { "epoch": 17.545787545787547, "grad_norm": 1.1015625, "learning_rate": 5.4378089144820364e-05, "loss": 0.125, "step": 23950 }, { "epoch": 17.546520146520148, "grad_norm": 1.203125, "learning_rate": 5.437467330582712e-05, "loss": 0.1332, "step": 23951 }, { "epoch": 17.54725274725275, "grad_norm": 1.1328125, "learning_rate": 5.4371257448269055e-05, "loss": 0.1351, "step": 23952 }, { "epoch": 17.54798534798535, "grad_norm": 1.5078125, "learning_rate": 5.436784157216249e-05, "loss": 0.1273, "step": 23953 }, { "epoch": 17.54871794871795, "grad_norm": 0.8125, "learning_rate": 5.4364425677523866e-05, "loss": 0.0907, "step": 23954 }, { "epoch": 17.54945054945055, "grad_norm": 1.703125, "learning_rate": 5.4361009764369485e-05, "loss": 0.1159, "step": 23955 }, { "epoch": 17.550183150183152, "grad_norm": 0.89453125, "learning_rate": 5.435759383271575e-05, "loss": 0.103, "step": 23956 }, { "epoch": 17.550915750915753, "grad_norm": 1.3671875, "learning_rate": 5.435417788257903e-05, "loss": 0.1268, "step": 23957 }, { "epoch": 17.551648351648353, "grad_norm": 0.796875, "learning_rate": 5.435076191397564e-05, "loss": 0.1154, "step": 23958 }, { "epoch": 17.552380952380954, "grad_norm": 1.109375, "learning_rate": 5.434734592692201e-05, "loss": 0.1232, "step": 23959 }, { "epoch": 17.553113553113555, "grad_norm": 1.375, "learning_rate": 5.434392992143448e-05, "loss": 0.1377, "step": 23960 }, { "epoch": 17.553846153846155, "grad_norm": 1.125, "learning_rate": 5.434051389752941e-05, "loss": 0.1137, "step": 23961 }, { "epoch": 17.554578754578756, "grad_norm": 1.0546875, "learning_rate": 5.43370978552232e-05, "loss": 0.0933, "step": 23962 }, { "epoch": 17.555311355311357, "grad_norm": 1.8515625, "learning_rate": 5.4333681794532164e-05, "loss": 0.1549, "step": 23963 }, { "epoch": 17.556043956043958, "grad_norm": 0.80078125, "learning_rate": 5.433026571547272e-05, "loss": 0.0947, "step": 23964 }, { "epoch": 17.55677655677656, "grad_norm": 1.75, "learning_rate": 5.4326849618061216e-05, "loss": 0.141, "step": 23965 }, { "epoch": 17.55750915750916, "grad_norm": 0.8984375, "learning_rate": 5.432343350231402e-05, "loss": 0.1186, "step": 23966 }, { "epoch": 17.55824175824176, "grad_norm": 1.171875, "learning_rate": 5.43200173682475e-05, "loss": 0.1342, "step": 23967 }, { "epoch": 17.55897435897436, "grad_norm": 0.83203125, "learning_rate": 5.431660121587802e-05, "loss": 0.115, "step": 23968 }, { "epoch": 17.55970695970696, "grad_norm": 1.1015625, "learning_rate": 5.4313185045221936e-05, "loss": 0.1253, "step": 23969 }, { "epoch": 17.560439560439562, "grad_norm": 0.88671875, "learning_rate": 5.430976885629565e-05, "loss": 0.112, "step": 23970 }, { "epoch": 17.561172161172163, "grad_norm": 1.1796875, "learning_rate": 5.43063526491155e-05, "loss": 0.1433, "step": 23971 }, { "epoch": 17.561904761904763, "grad_norm": 1.203125, "learning_rate": 5.430293642369786e-05, "loss": 0.1265, "step": 23972 }, { "epoch": 17.562637362637364, "grad_norm": 1.234375, "learning_rate": 5.429952018005912e-05, "loss": 0.1136, "step": 23973 }, { "epoch": 17.563369963369965, "grad_norm": 1.0703125, "learning_rate": 5.4296103918215616e-05, "loss": 0.109, "step": 23974 }, { "epoch": 17.564102564102566, "grad_norm": 1.234375, "learning_rate": 5.429268763818375e-05, "loss": 0.1087, "step": 23975 }, { "epoch": 17.564835164835166, "grad_norm": 1.5703125, "learning_rate": 5.4289271339979844e-05, "loss": 0.1501, "step": 23976 }, { "epoch": 17.565567765567767, "grad_norm": 1.2265625, "learning_rate": 5.4285855023620325e-05, "loss": 0.1179, "step": 23977 }, { "epoch": 17.566300366300368, "grad_norm": 1.109375, "learning_rate": 5.4282438689121514e-05, "loss": 0.1274, "step": 23978 }, { "epoch": 17.56703296703297, "grad_norm": 1.1953125, "learning_rate": 5.4279022336499796e-05, "loss": 0.1154, "step": 23979 }, { "epoch": 17.56776556776557, "grad_norm": 1.453125, "learning_rate": 5.427560596577156e-05, "loss": 0.1212, "step": 23980 }, { "epoch": 17.56849816849817, "grad_norm": 1.4609375, "learning_rate": 5.427218957695314e-05, "loss": 0.1359, "step": 23981 }, { "epoch": 17.56923076923077, "grad_norm": 1.0703125, "learning_rate": 5.426877317006093e-05, "loss": 0.1481, "step": 23982 }, { "epoch": 17.56996336996337, "grad_norm": 0.8828125, "learning_rate": 5.4265356745111286e-05, "loss": 0.1052, "step": 23983 }, { "epoch": 17.570695970695972, "grad_norm": 0.78515625, "learning_rate": 5.4261940302120585e-05, "loss": 0.116, "step": 23984 }, { "epoch": 17.571428571428573, "grad_norm": 1.0, "learning_rate": 5.425852384110518e-05, "loss": 0.1166, "step": 23985 }, { "epoch": 17.572161172161174, "grad_norm": 1.125, "learning_rate": 5.425510736208147e-05, "loss": 0.1495, "step": 23986 }, { "epoch": 17.572893772893774, "grad_norm": 0.89453125, "learning_rate": 5.42516908650658e-05, "loss": 0.1407, "step": 23987 }, { "epoch": 17.573626373626375, "grad_norm": 1.046875, "learning_rate": 5.424827435007456e-05, "loss": 0.1229, "step": 23988 }, { "epoch": 17.574358974358976, "grad_norm": 0.97265625, "learning_rate": 5.424485781712409e-05, "loss": 0.1171, "step": 23989 }, { "epoch": 17.575091575091577, "grad_norm": 1.296875, "learning_rate": 5.4241441266230774e-05, "loss": 0.1434, "step": 23990 }, { "epoch": 17.575824175824177, "grad_norm": 1.09375, "learning_rate": 5.4238024697411005e-05, "loss": 0.1386, "step": 23991 }, { "epoch": 17.576556776556778, "grad_norm": 1.140625, "learning_rate": 5.42346081106811e-05, "loss": 0.1182, "step": 23992 }, { "epoch": 17.57728937728938, "grad_norm": 1.40625, "learning_rate": 5.423119150605751e-05, "loss": 0.1236, "step": 23993 }, { "epoch": 17.57802197802198, "grad_norm": 1.4453125, "learning_rate": 5.422777488355651e-05, "loss": 0.1357, "step": 23994 }, { "epoch": 17.57875457875458, "grad_norm": 0.8671875, "learning_rate": 5.422435824319454e-05, "loss": 0.1144, "step": 23995 }, { "epoch": 17.57948717948718, "grad_norm": 1.09375, "learning_rate": 5.422094158498795e-05, "loss": 0.1225, "step": 23996 }, { "epoch": 17.58021978021978, "grad_norm": 1.15625, "learning_rate": 5.42175249089531e-05, "loss": 0.1233, "step": 23997 }, { "epoch": 17.580952380952382, "grad_norm": 1.3515625, "learning_rate": 5.421410821510635e-05, "loss": 0.135, "step": 23998 }, { "epoch": 17.581684981684983, "grad_norm": 1.0546875, "learning_rate": 5.421069150346412e-05, "loss": 0.1497, "step": 23999 }, { "epoch": 17.582417582417584, "grad_norm": 1.609375, "learning_rate": 5.420727477404273e-05, "loss": 0.1249, "step": 24000 }, { "epoch": 17.583150183150185, "grad_norm": 1.2890625, "learning_rate": 5.420385802685858e-05, "loss": 0.1368, "step": 24001 }, { "epoch": 17.583882783882785, "grad_norm": 0.53515625, "learning_rate": 5.4200441261928014e-05, "loss": 0.1005, "step": 24002 }, { "epoch": 17.584615384615386, "grad_norm": 1.015625, "learning_rate": 5.4197024479267446e-05, "loss": 0.1098, "step": 24003 }, { "epoch": 17.585347985347987, "grad_norm": 1.4765625, "learning_rate": 5.419360767889321e-05, "loss": 0.1681, "step": 24004 }, { "epoch": 17.586080586080588, "grad_norm": 0.60546875, "learning_rate": 5.4190190860821676e-05, "loss": 0.1046, "step": 24005 }, { "epoch": 17.58681318681319, "grad_norm": 0.9921875, "learning_rate": 5.418677402506923e-05, "loss": 0.119, "step": 24006 }, { "epoch": 17.58754578754579, "grad_norm": 1.640625, "learning_rate": 5.418335717165224e-05, "loss": 0.1136, "step": 24007 }, { "epoch": 17.58827838827839, "grad_norm": 1.1328125, "learning_rate": 5.417994030058707e-05, "loss": 0.1388, "step": 24008 }, { "epoch": 17.58901098901099, "grad_norm": 1.421875, "learning_rate": 5.417652341189012e-05, "loss": 0.1017, "step": 24009 }, { "epoch": 17.58974358974359, "grad_norm": 1.1875, "learning_rate": 5.417310650557772e-05, "loss": 0.1139, "step": 24010 }, { "epoch": 17.590476190476192, "grad_norm": 0.8203125, "learning_rate": 5.416968958166626e-05, "loss": 0.1127, "step": 24011 }, { "epoch": 17.591208791208793, "grad_norm": 0.890625, "learning_rate": 5.416627264017212e-05, "loss": 0.1361, "step": 24012 }, { "epoch": 17.591941391941393, "grad_norm": 0.91015625, "learning_rate": 5.416285568111167e-05, "loss": 0.1174, "step": 24013 }, { "epoch": 17.592673992673994, "grad_norm": 0.7421875, "learning_rate": 5.415943870450126e-05, "loss": 0.1286, "step": 24014 }, { "epoch": 17.593406593406595, "grad_norm": 0.99609375, "learning_rate": 5.4156021710357275e-05, "loss": 0.1365, "step": 24015 }, { "epoch": 17.594139194139196, "grad_norm": 1.53125, "learning_rate": 5.415260469869612e-05, "loss": 0.15, "step": 24016 }, { "epoch": 17.594871794871796, "grad_norm": 1.109375, "learning_rate": 5.414918766953412e-05, "loss": 0.127, "step": 24017 }, { "epoch": 17.595604395604397, "grad_norm": 1.109375, "learning_rate": 5.414577062288766e-05, "loss": 0.1149, "step": 24018 }, { "epoch": 17.596336996336998, "grad_norm": 0.9921875, "learning_rate": 5.414235355877312e-05, "loss": 0.1508, "step": 24019 }, { "epoch": 17.5970695970696, "grad_norm": 1.1640625, "learning_rate": 5.413893647720686e-05, "loss": 0.1544, "step": 24020 }, { "epoch": 17.5978021978022, "grad_norm": 0.71875, "learning_rate": 5.413551937820528e-05, "loss": 0.0913, "step": 24021 }, { "epoch": 17.5985347985348, "grad_norm": 1.234375, "learning_rate": 5.413210226178471e-05, "loss": 0.1084, "step": 24022 }, { "epoch": 17.5992673992674, "grad_norm": 1.90625, "learning_rate": 5.412868512796155e-05, "loss": 0.1311, "step": 24023 }, { "epoch": 17.6, "grad_norm": 1.0078125, "learning_rate": 5.412526797675217e-05, "loss": 0.1179, "step": 24024 }, { "epoch": 17.600732600732602, "grad_norm": 1.1640625, "learning_rate": 5.412185080817296e-05, "loss": 0.1581, "step": 24025 }, { "epoch": 17.601465201465203, "grad_norm": 0.87890625, "learning_rate": 5.411843362224025e-05, "loss": 0.121, "step": 24026 }, { "epoch": 17.602197802197804, "grad_norm": 0.85546875, "learning_rate": 5.411501641897046e-05, "loss": 0.1089, "step": 24027 }, { "epoch": 17.602930402930404, "grad_norm": 1.265625, "learning_rate": 5.41115991983799e-05, "loss": 0.1118, "step": 24028 }, { "epoch": 17.603663003663005, "grad_norm": 1.328125, "learning_rate": 5.410818196048503e-05, "loss": 0.1113, "step": 24029 }, { "epoch": 17.604395604395606, "grad_norm": 0.671875, "learning_rate": 5.4104764705302155e-05, "loss": 0.1141, "step": 24030 }, { "epoch": 17.605128205128207, "grad_norm": 0.8125, "learning_rate": 5.410134743284766e-05, "loss": 0.1043, "step": 24031 }, { "epoch": 17.605860805860807, "grad_norm": 1.4921875, "learning_rate": 5.409793014313794e-05, "loss": 0.114, "step": 24032 }, { "epoch": 17.606593406593408, "grad_norm": 1.3671875, "learning_rate": 5.409451283618935e-05, "loss": 0.1557, "step": 24033 }, { "epoch": 17.60732600732601, "grad_norm": 1.390625, "learning_rate": 5.409109551201828e-05, "loss": 0.1474, "step": 24034 }, { "epoch": 17.60805860805861, "grad_norm": 0.83203125, "learning_rate": 5.4087678170641076e-05, "loss": 0.1295, "step": 24035 }, { "epoch": 17.60879120879121, "grad_norm": 1.4921875, "learning_rate": 5.4084260812074126e-05, "loss": 0.1283, "step": 24036 }, { "epoch": 17.60952380952381, "grad_norm": 1.4921875, "learning_rate": 5.408084343633381e-05, "loss": 0.121, "step": 24037 }, { "epoch": 17.61025641025641, "grad_norm": 0.75390625, "learning_rate": 5.4077426043436516e-05, "loss": 0.1164, "step": 24038 }, { "epoch": 17.610989010989012, "grad_norm": 1.3515625, "learning_rate": 5.407400863339858e-05, "loss": 0.1233, "step": 24039 }, { "epoch": 17.611721611721613, "grad_norm": 0.71875, "learning_rate": 5.40705912062364e-05, "loss": 0.0933, "step": 24040 }, { "epoch": 17.612454212454214, "grad_norm": 0.796875, "learning_rate": 5.406717376196634e-05, "loss": 0.0964, "step": 24041 }, { "epoch": 17.613186813186815, "grad_norm": 1.140625, "learning_rate": 5.40637563006048e-05, "loss": 0.1337, "step": 24042 }, { "epoch": 17.613919413919415, "grad_norm": 0.984375, "learning_rate": 5.406033882216811e-05, "loss": 0.1279, "step": 24043 }, { "epoch": 17.614652014652016, "grad_norm": 0.9140625, "learning_rate": 5.405692132667268e-05, "loss": 0.1069, "step": 24044 }, { "epoch": 17.615384615384617, "grad_norm": 0.9609375, "learning_rate": 5.4053503814134874e-05, "loss": 0.1393, "step": 24045 }, { "epoch": 17.616117216117217, "grad_norm": 1.34375, "learning_rate": 5.4050086284571054e-05, "loss": 0.1063, "step": 24046 }, { "epoch": 17.616849816849818, "grad_norm": 0.98046875, "learning_rate": 5.4046668737997606e-05, "loss": 0.1309, "step": 24047 }, { "epoch": 17.61758241758242, "grad_norm": 0.65625, "learning_rate": 5.404325117443092e-05, "loss": 0.1147, "step": 24048 }, { "epoch": 17.61831501831502, "grad_norm": 0.76953125, "learning_rate": 5.4039833593887347e-05, "loss": 0.1064, "step": 24049 }, { "epoch": 17.61904761904762, "grad_norm": 0.765625, "learning_rate": 5.403641599638326e-05, "loss": 0.115, "step": 24050 }, { "epoch": 17.61978021978022, "grad_norm": 0.8671875, "learning_rate": 5.4032998381935054e-05, "loss": 0.1169, "step": 24051 }, { "epoch": 17.620512820512822, "grad_norm": 1.3671875, "learning_rate": 5.402958075055907e-05, "loss": 0.1229, "step": 24052 }, { "epoch": 17.621245421245423, "grad_norm": 1.5078125, "learning_rate": 5.4026163102271734e-05, "loss": 0.1129, "step": 24053 }, { "epoch": 17.621978021978023, "grad_norm": 0.7265625, "learning_rate": 5.4022745437089374e-05, "loss": 0.1071, "step": 24054 }, { "epoch": 17.622710622710624, "grad_norm": 0.90234375, "learning_rate": 5.401932775502839e-05, "loss": 0.1154, "step": 24055 }, { "epoch": 17.623443223443225, "grad_norm": 1.2890625, "learning_rate": 5.4015910056105165e-05, "loss": 0.1461, "step": 24056 }, { "epoch": 17.624175824175826, "grad_norm": 1.5, "learning_rate": 5.401249234033605e-05, "loss": 0.144, "step": 24057 }, { "epoch": 17.624908424908426, "grad_norm": 0.97265625, "learning_rate": 5.400907460773744e-05, "loss": 0.1404, "step": 24058 }, { "epoch": 17.625641025641027, "grad_norm": 0.8125, "learning_rate": 5.400565685832568e-05, "loss": 0.1146, "step": 24059 }, { "epoch": 17.626373626373628, "grad_norm": 1.0234375, "learning_rate": 5.400223909211718e-05, "loss": 0.1098, "step": 24060 }, { "epoch": 17.62710622710623, "grad_norm": 1.1640625, "learning_rate": 5.399882130912831e-05, "loss": 0.1169, "step": 24061 }, { "epoch": 17.62783882783883, "grad_norm": 1.8828125, "learning_rate": 5.399540350937543e-05, "loss": 0.1457, "step": 24062 }, { "epoch": 17.62857142857143, "grad_norm": 1.6328125, "learning_rate": 5.399198569287492e-05, "loss": 0.1067, "step": 24063 }, { "epoch": 17.62930402930403, "grad_norm": 1.796875, "learning_rate": 5.398856785964317e-05, "loss": 0.1172, "step": 24064 }, { "epoch": 17.63003663003663, "grad_norm": 1.1171875, "learning_rate": 5.398515000969654e-05, "loss": 0.1261, "step": 24065 }, { "epoch": 17.630769230769232, "grad_norm": 1.1953125, "learning_rate": 5.398173214305141e-05, "loss": 0.1324, "step": 24066 }, { "epoch": 17.631501831501833, "grad_norm": 0.89453125, "learning_rate": 5.397831425972415e-05, "loss": 0.1183, "step": 24067 }, { "epoch": 17.632234432234434, "grad_norm": 1.4296875, "learning_rate": 5.3974896359731165e-05, "loss": 0.1632, "step": 24068 }, { "epoch": 17.632967032967034, "grad_norm": 1.5078125, "learning_rate": 5.3971478443088806e-05, "loss": 0.1215, "step": 24069 }, { "epoch": 17.633699633699635, "grad_norm": 0.99609375, "learning_rate": 5.3968060509813444e-05, "loss": 0.1366, "step": 24070 }, { "epoch": 17.634432234432236, "grad_norm": 0.98046875, "learning_rate": 5.3964642559921484e-05, "loss": 0.1391, "step": 24071 }, { "epoch": 17.635164835164836, "grad_norm": 1.4296875, "learning_rate": 5.3961224593429263e-05, "loss": 0.1076, "step": 24072 }, { "epoch": 17.635897435897437, "grad_norm": 0.7890625, "learning_rate": 5.39578066103532e-05, "loss": 0.1128, "step": 24073 }, { "epoch": 17.636630036630038, "grad_norm": 0.83203125, "learning_rate": 5.395438861070965e-05, "loss": 0.1207, "step": 24074 }, { "epoch": 17.63736263736264, "grad_norm": 0.8359375, "learning_rate": 5.3950970594514974e-05, "loss": 0.0884, "step": 24075 }, { "epoch": 17.63809523809524, "grad_norm": 0.8828125, "learning_rate": 5.394755256178555e-05, "loss": 0.1213, "step": 24076 }, { "epoch": 17.63882783882784, "grad_norm": 0.9765625, "learning_rate": 5.3944134512537804e-05, "loss": 0.1267, "step": 24077 }, { "epoch": 17.63956043956044, "grad_norm": 1.1171875, "learning_rate": 5.3940716446788054e-05, "loss": 0.1198, "step": 24078 }, { "epoch": 17.64029304029304, "grad_norm": 0.8046875, "learning_rate": 5.393729836455271e-05, "loss": 0.136, "step": 24079 }, { "epoch": 17.641025641025642, "grad_norm": 1.1015625, "learning_rate": 5.393388026584815e-05, "loss": 0.1036, "step": 24080 }, { "epoch": 17.641758241758243, "grad_norm": 1.421875, "learning_rate": 5.393046215069073e-05, "loss": 0.1437, "step": 24081 }, { "epoch": 17.642490842490844, "grad_norm": 1.8125, "learning_rate": 5.392704401909686e-05, "loss": 0.1759, "step": 24082 }, { "epoch": 17.643223443223444, "grad_norm": 1.3984375, "learning_rate": 5.392362587108287e-05, "loss": 0.1531, "step": 24083 }, { "epoch": 17.643956043956045, "grad_norm": 1.9375, "learning_rate": 5.392020770666518e-05, "loss": 0.1075, "step": 24084 }, { "epoch": 17.644688644688646, "grad_norm": 0.99609375, "learning_rate": 5.391678952586016e-05, "loss": 0.1473, "step": 24085 }, { "epoch": 17.645421245421247, "grad_norm": 1.203125, "learning_rate": 5.391337132868416e-05, "loss": 0.1266, "step": 24086 }, { "epoch": 17.646153846153847, "grad_norm": 1.3515625, "learning_rate": 5.390995311515359e-05, "loss": 0.1126, "step": 24087 }, { "epoch": 17.646886446886448, "grad_norm": 0.64453125, "learning_rate": 5.390653488528481e-05, "loss": 0.1084, "step": 24088 }, { "epoch": 17.64761904761905, "grad_norm": 1.046875, "learning_rate": 5.39031166390942e-05, "loss": 0.1313, "step": 24089 }, { "epoch": 17.64835164835165, "grad_norm": 1.359375, "learning_rate": 5.389969837659815e-05, "loss": 0.0945, "step": 24090 }, { "epoch": 17.64908424908425, "grad_norm": 1.03125, "learning_rate": 5.3896280097813027e-05, "loss": 0.114, "step": 24091 }, { "epoch": 17.64981684981685, "grad_norm": 0.94140625, "learning_rate": 5.3892861802755215e-05, "loss": 0.1121, "step": 24092 }, { "epoch": 17.650549450549452, "grad_norm": 1.15625, "learning_rate": 5.388944349144107e-05, "loss": 0.0961, "step": 24093 }, { "epoch": 17.651282051282053, "grad_norm": 1.0859375, "learning_rate": 5.388602516388701e-05, "loss": 0.1171, "step": 24094 }, { "epoch": 17.652014652014653, "grad_norm": 1.2265625, "learning_rate": 5.388260682010938e-05, "loss": 0.1319, "step": 24095 }, { "epoch": 17.652747252747254, "grad_norm": 1.0546875, "learning_rate": 5.387918846012457e-05, "loss": 0.1114, "step": 24096 }, { "epoch": 17.653479853479855, "grad_norm": 2.0, "learning_rate": 5.3875770083948976e-05, "loss": 0.1454, "step": 24097 }, { "epoch": 17.654212454212455, "grad_norm": 1.3125, "learning_rate": 5.3872351691598934e-05, "loss": 0.0914, "step": 24098 }, { "epoch": 17.654945054945056, "grad_norm": 1.1953125, "learning_rate": 5.386893328309086e-05, "loss": 0.0998, "step": 24099 }, { "epoch": 17.655677655677657, "grad_norm": 0.9375, "learning_rate": 5.3865514858441125e-05, "loss": 0.1171, "step": 24100 }, { "epoch": 17.656410256410258, "grad_norm": 1.453125, "learning_rate": 5.3862096417666095e-05, "loss": 0.146, "step": 24101 }, { "epoch": 17.65714285714286, "grad_norm": 0.70703125, "learning_rate": 5.385867796078216e-05, "loss": 0.1044, "step": 24102 }, { "epoch": 17.65787545787546, "grad_norm": 0.7734375, "learning_rate": 5.38552594878057e-05, "loss": 0.0926, "step": 24103 }, { "epoch": 17.65860805860806, "grad_norm": 0.65234375, "learning_rate": 5.3851840998753076e-05, "loss": 0.1347, "step": 24104 }, { "epoch": 17.65934065934066, "grad_norm": 1.3046875, "learning_rate": 5.38484224936407e-05, "loss": 0.1294, "step": 24105 }, { "epoch": 17.66007326007326, "grad_norm": 1.2265625, "learning_rate": 5.384500397248493e-05, "loss": 0.1166, "step": 24106 }, { "epoch": 17.660805860805862, "grad_norm": 0.88671875, "learning_rate": 5.384158543530213e-05, "loss": 0.1061, "step": 24107 }, { "epoch": 17.661538461538463, "grad_norm": 0.9453125, "learning_rate": 5.383816688210873e-05, "loss": 0.1178, "step": 24108 }, { "epoch": 17.662271062271063, "grad_norm": 0.7578125, "learning_rate": 5.383474831292105e-05, "loss": 0.1135, "step": 24109 }, { "epoch": 17.663003663003664, "grad_norm": 1.3203125, "learning_rate": 5.383132972775551e-05, "loss": 0.1317, "step": 24110 }, { "epoch": 17.663736263736265, "grad_norm": 1.1796875, "learning_rate": 5.3827911126628475e-05, "loss": 0.1213, "step": 24111 }, { "epoch": 17.664468864468866, "grad_norm": 0.99609375, "learning_rate": 5.382449250955631e-05, "loss": 0.1465, "step": 24112 }, { "epoch": 17.665201465201466, "grad_norm": 0.96484375, "learning_rate": 5.382107387655542e-05, "loss": 0.1298, "step": 24113 }, { "epoch": 17.665934065934067, "grad_norm": 0.69140625, "learning_rate": 5.381765522764217e-05, "loss": 0.0951, "step": 24114 }, { "epoch": 17.666666666666668, "grad_norm": 1.3203125, "learning_rate": 5.3814236562832945e-05, "loss": 0.1401, "step": 24115 }, { "epoch": 17.66739926739927, "grad_norm": 0.94921875, "learning_rate": 5.381081788214414e-05, "loss": 0.0951, "step": 24116 }, { "epoch": 17.66813186813187, "grad_norm": 0.78125, "learning_rate": 5.380739918559209e-05, "loss": 0.1236, "step": 24117 }, { "epoch": 17.66886446886447, "grad_norm": 0.8984375, "learning_rate": 5.3803980473193235e-05, "loss": 0.1063, "step": 24118 }, { "epoch": 17.66959706959707, "grad_norm": 1.6015625, "learning_rate": 5.3800561744963906e-05, "loss": 0.1234, "step": 24119 }, { "epoch": 17.67032967032967, "grad_norm": 0.99609375, "learning_rate": 5.379714300092051e-05, "loss": 0.1255, "step": 24120 }, { "epoch": 17.671062271062272, "grad_norm": 1.0703125, "learning_rate": 5.3793724241079426e-05, "loss": 0.1482, "step": 24121 }, { "epoch": 17.671794871794873, "grad_norm": 1.0703125, "learning_rate": 5.379030546545701e-05, "loss": 0.1376, "step": 24122 }, { "epoch": 17.672527472527474, "grad_norm": 0.9921875, "learning_rate": 5.378688667406968e-05, "loss": 0.1181, "step": 24123 }, { "epoch": 17.673260073260074, "grad_norm": 0.89453125, "learning_rate": 5.378346786693379e-05, "loss": 0.1125, "step": 24124 }, { "epoch": 17.673992673992675, "grad_norm": 1.234375, "learning_rate": 5.3780049044065714e-05, "loss": 0.114, "step": 24125 }, { "epoch": 17.674725274725276, "grad_norm": 1.4375, "learning_rate": 5.3776630205481856e-05, "loss": 0.1159, "step": 24126 }, { "epoch": 17.675457875457877, "grad_norm": 0.8828125, "learning_rate": 5.37732113511986e-05, "loss": 0.1144, "step": 24127 }, { "epoch": 17.676190476190477, "grad_norm": 1.0390625, "learning_rate": 5.3769792481232295e-05, "loss": 0.122, "step": 24128 }, { "epoch": 17.676923076923078, "grad_norm": 1.390625, "learning_rate": 5.376637359559935e-05, "loss": 0.139, "step": 24129 }, { "epoch": 17.67765567765568, "grad_norm": 1.25, "learning_rate": 5.376295469431613e-05, "loss": 0.1431, "step": 24130 }, { "epoch": 17.67838827838828, "grad_norm": 0.83203125, "learning_rate": 5.375953577739904e-05, "loss": 0.1083, "step": 24131 }, { "epoch": 17.67912087912088, "grad_norm": 1.390625, "learning_rate": 5.375611684486442e-05, "loss": 0.096, "step": 24132 }, { "epoch": 17.67985347985348, "grad_norm": 1.3984375, "learning_rate": 5.3752697896728685e-05, "loss": 0.1285, "step": 24133 }, { "epoch": 17.68058608058608, "grad_norm": 1.53125, "learning_rate": 5.374927893300823e-05, "loss": 0.1462, "step": 24134 }, { "epoch": 17.681318681318682, "grad_norm": 0.7734375, "learning_rate": 5.374585995371938e-05, "loss": 0.1335, "step": 24135 }, { "epoch": 17.682051282051283, "grad_norm": 0.66015625, "learning_rate": 5.374244095887857e-05, "loss": 0.1104, "step": 24136 }, { "epoch": 17.682783882783884, "grad_norm": 1.1875, "learning_rate": 5.3739021948502154e-05, "loss": 0.1121, "step": 24137 }, { "epoch": 17.683516483516485, "grad_norm": 1.4453125, "learning_rate": 5.373560292260651e-05, "loss": 0.116, "step": 24138 }, { "epoch": 17.684249084249085, "grad_norm": 0.92578125, "learning_rate": 5.3732183881208044e-05, "loss": 0.138, "step": 24139 }, { "epoch": 17.684981684981686, "grad_norm": 1.40625, "learning_rate": 5.372876482432312e-05, "loss": 0.116, "step": 24140 }, { "epoch": 17.685714285714287, "grad_norm": 0.8828125, "learning_rate": 5.372534575196812e-05, "loss": 0.1065, "step": 24141 }, { "epoch": 17.686446886446888, "grad_norm": 0.77734375, "learning_rate": 5.372192666415943e-05, "loss": 0.1046, "step": 24142 }, { "epoch": 17.68717948717949, "grad_norm": 1.1796875, "learning_rate": 5.3718507560913434e-05, "loss": 0.1053, "step": 24143 }, { "epoch": 17.68791208791209, "grad_norm": 2.5, "learning_rate": 5.371508844224651e-05, "loss": 0.1196, "step": 24144 }, { "epoch": 17.68864468864469, "grad_norm": 1.1875, "learning_rate": 5.371166930817504e-05, "loss": 0.1335, "step": 24145 }, { "epoch": 17.68937728937729, "grad_norm": 1.59375, "learning_rate": 5.3708250158715426e-05, "loss": 0.1177, "step": 24146 }, { "epoch": 17.69010989010989, "grad_norm": 1.578125, "learning_rate": 5.3704830993884015e-05, "loss": 0.1237, "step": 24147 }, { "epoch": 17.690842490842492, "grad_norm": 0.66015625, "learning_rate": 5.37014118136972e-05, "loss": 0.1088, "step": 24148 }, { "epoch": 17.691575091575093, "grad_norm": 0.92578125, "learning_rate": 5.369799261817139e-05, "loss": 0.1018, "step": 24149 }, { "epoch": 17.692307692307693, "grad_norm": 1.546875, "learning_rate": 5.369457340732294e-05, "loss": 0.1575, "step": 24150 }, { "epoch": 17.693040293040294, "grad_norm": 1.0625, "learning_rate": 5.369115418116824e-05, "loss": 0.1272, "step": 24151 }, { "epoch": 17.693772893772895, "grad_norm": 1.5703125, "learning_rate": 5.368773493972366e-05, "loss": 0.1324, "step": 24152 }, { "epoch": 17.694505494505496, "grad_norm": 0.890625, "learning_rate": 5.368431568300559e-05, "loss": 0.1148, "step": 24153 }, { "epoch": 17.695238095238096, "grad_norm": 1.7109375, "learning_rate": 5.368089641103045e-05, "loss": 0.1317, "step": 24154 }, { "epoch": 17.695970695970697, "grad_norm": 1.171875, "learning_rate": 5.3677477123814566e-05, "loss": 0.1238, "step": 24155 }, { "epoch": 17.696703296703298, "grad_norm": 1.78125, "learning_rate": 5.367405782137436e-05, "loss": 0.1515, "step": 24156 }, { "epoch": 17.6974358974359, "grad_norm": 1.234375, "learning_rate": 5.36706385037262e-05, "loss": 0.1363, "step": 24157 }, { "epoch": 17.6981684981685, "grad_norm": 2.578125, "learning_rate": 5.366721917088645e-05, "loss": 0.1435, "step": 24158 }, { "epoch": 17.6989010989011, "grad_norm": 1.0703125, "learning_rate": 5.366379982287154e-05, "loss": 0.1308, "step": 24159 }, { "epoch": 17.6996336996337, "grad_norm": 0.9609375, "learning_rate": 5.3660380459697815e-05, "loss": 0.1162, "step": 24160 }, { "epoch": 17.7003663003663, "grad_norm": 1.1328125, "learning_rate": 5.365696108138166e-05, "loss": 0.1071, "step": 24161 }, { "epoch": 17.701098901098902, "grad_norm": 1.4296875, "learning_rate": 5.36535416879395e-05, "loss": 0.1136, "step": 24162 }, { "epoch": 17.701831501831503, "grad_norm": 2.046875, "learning_rate": 5.365012227938767e-05, "loss": 0.1042, "step": 24163 }, { "epoch": 17.702564102564104, "grad_norm": 1.9921875, "learning_rate": 5.364670285574256e-05, "loss": 0.1338, "step": 24164 }, { "epoch": 17.703296703296704, "grad_norm": 1.0625, "learning_rate": 5.364328341702055e-05, "loss": 0.1568, "step": 24165 }, { "epoch": 17.704029304029305, "grad_norm": 1.578125, "learning_rate": 5.3639863963238055e-05, "loss": 0.1238, "step": 24166 }, { "epoch": 17.704761904761906, "grad_norm": 0.83203125, "learning_rate": 5.3636444494411445e-05, "loss": 0.1162, "step": 24167 }, { "epoch": 17.705494505494507, "grad_norm": 1.5390625, "learning_rate": 5.36330250105571e-05, "loss": 0.1484, "step": 24168 }, { "epoch": 17.706227106227107, "grad_norm": 1.453125, "learning_rate": 5.3629605511691397e-05, "loss": 0.1087, "step": 24169 }, { "epoch": 17.706959706959708, "grad_norm": 1.1640625, "learning_rate": 5.3626185997830735e-05, "loss": 0.1462, "step": 24170 }, { "epoch": 17.70769230769231, "grad_norm": 1.25, "learning_rate": 5.362276646899148e-05, "loss": 0.1052, "step": 24171 }, { "epoch": 17.70842490842491, "grad_norm": 0.6328125, "learning_rate": 5.361934692519004e-05, "loss": 0.1144, "step": 24172 }, { "epoch": 17.70915750915751, "grad_norm": 1.1640625, "learning_rate": 5.3615927366442774e-05, "loss": 0.1137, "step": 24173 }, { "epoch": 17.70989010989011, "grad_norm": 1.0390625, "learning_rate": 5.3612507792766075e-05, "loss": 0.1065, "step": 24174 }, { "epoch": 17.71062271062271, "grad_norm": 0.90625, "learning_rate": 5.360908820417634e-05, "loss": 0.1022, "step": 24175 }, { "epoch": 17.711355311355312, "grad_norm": 0.87109375, "learning_rate": 5.360566860068994e-05, "loss": 0.1058, "step": 24176 }, { "epoch": 17.712087912087913, "grad_norm": 1.140625, "learning_rate": 5.3602248982323256e-05, "loss": 0.1235, "step": 24177 }, { "epoch": 17.712820512820514, "grad_norm": 1.53125, "learning_rate": 5.359882934909266e-05, "loss": 0.1555, "step": 24178 }, { "epoch": 17.713553113553115, "grad_norm": 0.875, "learning_rate": 5.3595409701014574e-05, "loss": 0.0955, "step": 24179 }, { "epoch": 17.714285714285715, "grad_norm": 0.94921875, "learning_rate": 5.359199003810538e-05, "loss": 0.1405, "step": 24180 }, { "epoch": 17.715018315018316, "grad_norm": 1.375, "learning_rate": 5.358857036038143e-05, "loss": 0.1081, "step": 24181 }, { "epoch": 17.715750915750917, "grad_norm": 1.2734375, "learning_rate": 5.358515066785912e-05, "loss": 0.1488, "step": 24182 }, { "epoch": 17.716483516483517, "grad_norm": 1.03125, "learning_rate": 5.3581730960554855e-05, "loss": 0.1262, "step": 24183 }, { "epoch": 17.717216117216118, "grad_norm": 1.296875, "learning_rate": 5.357831123848499e-05, "loss": 0.1223, "step": 24184 }, { "epoch": 17.71794871794872, "grad_norm": 0.9765625, "learning_rate": 5.3574891501665945e-05, "loss": 0.1066, "step": 24185 }, { "epoch": 17.71868131868132, "grad_norm": 1.171875, "learning_rate": 5.3571471750114076e-05, "loss": 0.1128, "step": 24186 }, { "epoch": 17.71941391941392, "grad_norm": 1.09375, "learning_rate": 5.356805198384577e-05, "loss": 0.1343, "step": 24187 }, { "epoch": 17.72014652014652, "grad_norm": 2.125, "learning_rate": 5.356463220287743e-05, "loss": 0.1243, "step": 24188 }, { "epoch": 17.720879120879122, "grad_norm": 1.1484375, "learning_rate": 5.356121240722544e-05, "loss": 0.1481, "step": 24189 }, { "epoch": 17.721611721611723, "grad_norm": 0.6484375, "learning_rate": 5.355779259690614e-05, "loss": 0.1248, "step": 24190 }, { "epoch": 17.722344322344323, "grad_norm": 0.8046875, "learning_rate": 5.355437277193598e-05, "loss": 0.1497, "step": 24191 }, { "epoch": 17.723076923076924, "grad_norm": 0.90234375, "learning_rate": 5.35509529323313e-05, "loss": 0.1163, "step": 24192 }, { "epoch": 17.723809523809525, "grad_norm": 0.87890625, "learning_rate": 5.3547533078108525e-05, "loss": 0.1346, "step": 24193 }, { "epoch": 17.724542124542126, "grad_norm": 0.80859375, "learning_rate": 5.354411320928401e-05, "loss": 0.1226, "step": 24194 }, { "epoch": 17.725274725274726, "grad_norm": 1.328125, "learning_rate": 5.354069332587415e-05, "loss": 0.1235, "step": 24195 }, { "epoch": 17.726007326007327, "grad_norm": 1.25, "learning_rate": 5.353727342789533e-05, "loss": 0.1316, "step": 24196 }, { "epoch": 17.726739926739928, "grad_norm": 0.91015625, "learning_rate": 5.353385351536392e-05, "loss": 0.1128, "step": 24197 }, { "epoch": 17.72747252747253, "grad_norm": 1.0546875, "learning_rate": 5.353043358829636e-05, "loss": 0.1276, "step": 24198 }, { "epoch": 17.72820512820513, "grad_norm": 1.546875, "learning_rate": 5.352701364670897e-05, "loss": 0.1387, "step": 24199 }, { "epoch": 17.72893772893773, "grad_norm": 0.625, "learning_rate": 5.3523593690618154e-05, "loss": 0.1109, "step": 24200 }, { "epoch": 17.72967032967033, "grad_norm": 1.5859375, "learning_rate": 5.3520173720040336e-05, "loss": 0.1133, "step": 24201 }, { "epoch": 17.73040293040293, "grad_norm": 1.078125, "learning_rate": 5.3516753734991854e-05, "loss": 0.1432, "step": 24202 }, { "epoch": 17.731135531135532, "grad_norm": 1.1640625, "learning_rate": 5.351333373548912e-05, "loss": 0.0936, "step": 24203 }, { "epoch": 17.731868131868133, "grad_norm": 0.7109375, "learning_rate": 5.35099137215485e-05, "loss": 0.1272, "step": 24204 }, { "epoch": 17.732600732600734, "grad_norm": 1.078125, "learning_rate": 5.350649369318641e-05, "loss": 0.1109, "step": 24205 }, { "epoch": 17.733333333333334, "grad_norm": 1.09375, "learning_rate": 5.350307365041923e-05, "loss": 0.1213, "step": 24206 }, { "epoch": 17.734065934065935, "grad_norm": 0.87890625, "learning_rate": 5.3499653593263343e-05, "loss": 0.1082, "step": 24207 }, { "epoch": 17.734798534798536, "grad_norm": 1.09375, "learning_rate": 5.349623352173511e-05, "loss": 0.1445, "step": 24208 }, { "epoch": 17.735531135531136, "grad_norm": 1.390625, "learning_rate": 5.349281343585095e-05, "loss": 0.1276, "step": 24209 }, { "epoch": 17.736263736263737, "grad_norm": 1.1953125, "learning_rate": 5.3489393335627235e-05, "loss": 0.1348, "step": 24210 }, { "epoch": 17.736996336996338, "grad_norm": 0.9296875, "learning_rate": 5.348597322108035e-05, "loss": 0.1306, "step": 24211 }, { "epoch": 17.73772893772894, "grad_norm": 1.046875, "learning_rate": 5.348255309222671e-05, "loss": 0.1544, "step": 24212 }, { "epoch": 17.73846153846154, "grad_norm": 0.9453125, "learning_rate": 5.3479132949082645e-05, "loss": 0.0974, "step": 24213 }, { "epoch": 17.73919413919414, "grad_norm": 1.5078125, "learning_rate": 5.34757127916646e-05, "loss": 0.1372, "step": 24214 }, { "epoch": 17.73992673992674, "grad_norm": 1.203125, "learning_rate": 5.3472292619988914e-05, "loss": 0.1389, "step": 24215 }, { "epoch": 17.74065934065934, "grad_norm": 0.92578125, "learning_rate": 5.3468872434072024e-05, "loss": 0.1099, "step": 24216 }, { "epoch": 17.741391941391942, "grad_norm": 0.69140625, "learning_rate": 5.346545223393028e-05, "loss": 0.1135, "step": 24217 }, { "epoch": 17.742124542124543, "grad_norm": 0.9609375, "learning_rate": 5.346203201958008e-05, "loss": 0.1385, "step": 24218 }, { "epoch": 17.742857142857144, "grad_norm": 1.671875, "learning_rate": 5.345861179103783e-05, "loss": 0.1218, "step": 24219 }, { "epoch": 17.743589743589745, "grad_norm": 1.5625, "learning_rate": 5.3455191548319885e-05, "loss": 0.1174, "step": 24220 }, { "epoch": 17.744322344322345, "grad_norm": 1.71875, "learning_rate": 5.345177129144264e-05, "loss": 0.0972, "step": 24221 }, { "epoch": 17.745054945054946, "grad_norm": 0.890625, "learning_rate": 5.344835102042251e-05, "loss": 0.1029, "step": 24222 }, { "epoch": 17.745787545787547, "grad_norm": 0.9453125, "learning_rate": 5.3444930735275836e-05, "loss": 0.1295, "step": 24223 }, { "epoch": 17.746520146520147, "grad_norm": 0.84765625, "learning_rate": 5.3441510436019046e-05, "loss": 0.118, "step": 24224 }, { "epoch": 17.747252747252748, "grad_norm": 0.75, "learning_rate": 5.343809012266853e-05, "loss": 0.1461, "step": 24225 }, { "epoch": 17.74798534798535, "grad_norm": 0.87109375, "learning_rate": 5.3434669795240644e-05, "loss": 0.1106, "step": 24226 }, { "epoch": 17.74871794871795, "grad_norm": 1.078125, "learning_rate": 5.343124945375179e-05, "loss": 0.129, "step": 24227 }, { "epoch": 17.74945054945055, "grad_norm": 1.4375, "learning_rate": 5.342782909821835e-05, "loss": 0.138, "step": 24228 }, { "epoch": 17.75018315018315, "grad_norm": 0.953125, "learning_rate": 5.342440872865673e-05, "loss": 0.1212, "step": 24229 }, { "epoch": 17.750915750915752, "grad_norm": 1.4609375, "learning_rate": 5.34209883450833e-05, "loss": 0.1185, "step": 24230 }, { "epoch": 17.751648351648353, "grad_norm": 0.8828125, "learning_rate": 5.341756794751446e-05, "loss": 0.1177, "step": 24231 }, { "epoch": 17.752380952380953, "grad_norm": 0.87890625, "learning_rate": 5.34141475359666e-05, "loss": 0.1206, "step": 24232 }, { "epoch": 17.753113553113554, "grad_norm": 0.92578125, "learning_rate": 5.3410727110456106e-05, "loss": 0.0921, "step": 24233 }, { "epoch": 17.753846153846155, "grad_norm": 1.6640625, "learning_rate": 5.3407306670999344e-05, "loss": 0.1478, "step": 24234 }, { "epoch": 17.754578754578755, "grad_norm": 0.65625, "learning_rate": 5.340388621761274e-05, "loss": 0.1107, "step": 24235 }, { "epoch": 17.755311355311356, "grad_norm": 0.86328125, "learning_rate": 5.3400465750312656e-05, "loss": 0.1217, "step": 24236 }, { "epoch": 17.756043956043957, "grad_norm": 0.51953125, "learning_rate": 5.3397045269115486e-05, "loss": 0.0985, "step": 24237 }, { "epoch": 17.756776556776558, "grad_norm": 1.3671875, "learning_rate": 5.339362477403763e-05, "loss": 0.1802, "step": 24238 }, { "epoch": 17.75750915750916, "grad_norm": 1.5546875, "learning_rate": 5.3390204265095454e-05, "loss": 0.1334, "step": 24239 }, { "epoch": 17.75824175824176, "grad_norm": 0.79296875, "learning_rate": 5.338678374230538e-05, "loss": 0.1289, "step": 24240 }, { "epoch": 17.75897435897436, "grad_norm": 1.1328125, "learning_rate": 5.338336320568376e-05, "loss": 0.116, "step": 24241 }, { "epoch": 17.75970695970696, "grad_norm": 0.9375, "learning_rate": 5.337994265524702e-05, "loss": 0.1226, "step": 24242 }, { "epoch": 17.76043956043956, "grad_norm": 1.21875, "learning_rate": 5.337652209101151e-05, "loss": 0.1439, "step": 24243 }, { "epoch": 17.761172161172162, "grad_norm": 0.66015625, "learning_rate": 5.337310151299363e-05, "loss": 0.1268, "step": 24244 }, { "epoch": 17.761904761904763, "grad_norm": 1.1953125, "learning_rate": 5.3369680921209794e-05, "loss": 0.1308, "step": 24245 }, { "epoch": 17.762637362637363, "grad_norm": 1.1953125, "learning_rate": 5.3366260315676374e-05, "loss": 0.1067, "step": 24246 }, { "epoch": 17.763369963369964, "grad_norm": 0.7890625, "learning_rate": 5.336283969640975e-05, "loss": 0.1441, "step": 24247 }, { "epoch": 17.764102564102565, "grad_norm": 0.94921875, "learning_rate": 5.335941906342635e-05, "loss": 0.1252, "step": 24248 }, { "epoch": 17.764835164835166, "grad_norm": 1.109375, "learning_rate": 5.3355998416742504e-05, "loss": 0.1014, "step": 24249 }, { "epoch": 17.765567765567766, "grad_norm": 0.69140625, "learning_rate": 5.3352577756374643e-05, "loss": 0.1299, "step": 24250 }, { "epoch": 17.766300366300367, "grad_norm": 1.5, "learning_rate": 5.3349157082339154e-05, "loss": 0.1627, "step": 24251 }, { "epoch": 17.767032967032968, "grad_norm": 1.2578125, "learning_rate": 5.33457363946524e-05, "loss": 0.146, "step": 24252 }, { "epoch": 17.76776556776557, "grad_norm": 1.3046875, "learning_rate": 5.33423156933308e-05, "loss": 0.139, "step": 24253 }, { "epoch": 17.76849816849817, "grad_norm": 0.9453125, "learning_rate": 5.333889497839073e-05, "loss": 0.1158, "step": 24254 }, { "epoch": 17.76923076923077, "grad_norm": 1.765625, "learning_rate": 5.3335474249848583e-05, "loss": 0.1094, "step": 24255 }, { "epoch": 17.76996336996337, "grad_norm": 1.0, "learning_rate": 5.333205350772074e-05, "loss": 0.1383, "step": 24256 }, { "epoch": 17.77069597069597, "grad_norm": 1.40625, "learning_rate": 5.3328632752023596e-05, "loss": 0.1305, "step": 24257 }, { "epoch": 17.771428571428572, "grad_norm": 1.015625, "learning_rate": 5.332521198277356e-05, "loss": 0.1051, "step": 24258 }, { "epoch": 17.772161172161173, "grad_norm": 1.015625, "learning_rate": 5.3321791199987e-05, "loss": 0.1029, "step": 24259 }, { "epoch": 17.772893772893774, "grad_norm": 1.3984375, "learning_rate": 5.3318370403680316e-05, "loss": 0.129, "step": 24260 }, { "epoch": 17.773626373626374, "grad_norm": 1.0859375, "learning_rate": 5.331494959386989e-05, "loss": 0.1087, "step": 24261 }, { "epoch": 17.774358974358975, "grad_norm": 1.171875, "learning_rate": 5.331152877057212e-05, "loss": 0.1373, "step": 24262 }, { "epoch": 17.775091575091576, "grad_norm": 0.78125, "learning_rate": 5.3308107933803386e-05, "loss": 0.1246, "step": 24263 }, { "epoch": 17.775824175824177, "grad_norm": 1.0859375, "learning_rate": 5.33046870835801e-05, "loss": 0.1009, "step": 24264 }, { "epoch": 17.776556776556777, "grad_norm": 0.9140625, "learning_rate": 5.3301266219918624e-05, "loss": 0.1291, "step": 24265 }, { "epoch": 17.777289377289378, "grad_norm": 0.72265625, "learning_rate": 5.3297845342835365e-05, "loss": 0.1286, "step": 24266 }, { "epoch": 17.77802197802198, "grad_norm": 1.953125, "learning_rate": 5.329442445234669e-05, "loss": 0.1443, "step": 24267 }, { "epoch": 17.77875457875458, "grad_norm": 1.484375, "learning_rate": 5.329100354846903e-05, "loss": 0.1381, "step": 24268 }, { "epoch": 17.77948717948718, "grad_norm": 1.53125, "learning_rate": 5.3287582631218756e-05, "loss": 0.1458, "step": 24269 }, { "epoch": 17.78021978021978, "grad_norm": 1.078125, "learning_rate": 5.328416170061226e-05, "loss": 0.1235, "step": 24270 }, { "epoch": 17.78095238095238, "grad_norm": 1.4375, "learning_rate": 5.328074075666594e-05, "loss": 0.1358, "step": 24271 }, { "epoch": 17.781684981684982, "grad_norm": 1.1796875, "learning_rate": 5.3277319799396164e-05, "loss": 0.1339, "step": 24272 }, { "epoch": 17.782417582417583, "grad_norm": 0.9765625, "learning_rate": 5.3273898828819326e-05, "loss": 0.1163, "step": 24273 }, { "epoch": 17.783150183150184, "grad_norm": 1.2578125, "learning_rate": 5.327047784495185e-05, "loss": 0.1231, "step": 24274 }, { "epoch": 17.783882783882785, "grad_norm": 1.4140625, "learning_rate": 5.326705684781009e-05, "loss": 0.1702, "step": 24275 }, { "epoch": 17.784615384615385, "grad_norm": 1.546875, "learning_rate": 5.3263635837410475e-05, "loss": 0.155, "step": 24276 }, { "epoch": 17.785347985347986, "grad_norm": 1.6875, "learning_rate": 5.326021481376936e-05, "loss": 0.1331, "step": 24277 }, { "epoch": 17.786080586080587, "grad_norm": 0.81640625, "learning_rate": 5.3256793776903135e-05, "loss": 0.1006, "step": 24278 }, { "epoch": 17.786813186813188, "grad_norm": 1.2890625, "learning_rate": 5.325337272682823e-05, "loss": 0.137, "step": 24279 }, { "epoch": 17.78754578754579, "grad_norm": 1.4296875, "learning_rate": 5.3249951663560996e-05, "loss": 0.1199, "step": 24280 }, { "epoch": 17.78827838827839, "grad_norm": 0.546875, "learning_rate": 5.3246530587117846e-05, "loss": 0.0995, "step": 24281 }, { "epoch": 17.78901098901099, "grad_norm": 0.6328125, "learning_rate": 5.3243109497515166e-05, "loss": 0.0897, "step": 24282 }, { "epoch": 17.78974358974359, "grad_norm": 0.9765625, "learning_rate": 5.323968839476934e-05, "loss": 0.1046, "step": 24283 }, { "epoch": 17.79047619047619, "grad_norm": 0.9609375, "learning_rate": 5.323626727889678e-05, "loss": 0.1097, "step": 24284 }, { "epoch": 17.791208791208792, "grad_norm": 1.09375, "learning_rate": 5.323284614991387e-05, "loss": 0.1446, "step": 24285 }, { "epoch": 17.791941391941393, "grad_norm": 0.98046875, "learning_rate": 5.3229425007836977e-05, "loss": 0.1321, "step": 24286 }, { "epoch": 17.792673992673993, "grad_norm": 1.203125, "learning_rate": 5.3226003852682525e-05, "loss": 0.1444, "step": 24287 }, { "epoch": 17.793406593406594, "grad_norm": 1.203125, "learning_rate": 5.322258268446689e-05, "loss": 0.1265, "step": 24288 }, { "epoch": 17.794139194139195, "grad_norm": 0.97265625, "learning_rate": 5.3219161503206486e-05, "loss": 0.0895, "step": 24289 }, { "epoch": 17.794871794871796, "grad_norm": 0.81640625, "learning_rate": 5.3215740308917675e-05, "loss": 0.1412, "step": 24290 }, { "epoch": 17.795604395604396, "grad_norm": 1.0546875, "learning_rate": 5.3212319101616846e-05, "loss": 0.1338, "step": 24291 }, { "epoch": 17.796336996336997, "grad_norm": 1.2265625, "learning_rate": 5.320889788132042e-05, "loss": 0.1187, "step": 24292 }, { "epoch": 17.797069597069598, "grad_norm": 1.5625, "learning_rate": 5.320547664804476e-05, "loss": 0.1384, "step": 24293 }, { "epoch": 17.7978021978022, "grad_norm": 1.203125, "learning_rate": 5.320205540180628e-05, "loss": 0.132, "step": 24294 }, { "epoch": 17.7985347985348, "grad_norm": 1.0234375, "learning_rate": 5.319863414262137e-05, "loss": 0.1335, "step": 24295 }, { "epoch": 17.7992673992674, "grad_norm": 0.890625, "learning_rate": 5.3195212870506414e-05, "loss": 0.1233, "step": 24296 }, { "epoch": 17.8, "grad_norm": 1.1796875, "learning_rate": 5.319179158547782e-05, "loss": 0.0926, "step": 24297 }, { "epoch": 17.8007326007326, "grad_norm": 1.7109375, "learning_rate": 5.3188370287551956e-05, "loss": 0.1089, "step": 24298 }, { "epoch": 17.801465201465202, "grad_norm": 1.1171875, "learning_rate": 5.318494897674523e-05, "loss": 0.1424, "step": 24299 }, { "epoch": 17.802197802197803, "grad_norm": 0.80859375, "learning_rate": 5.318152765307405e-05, "loss": 0.1199, "step": 24300 }, { "epoch": 17.802930402930404, "grad_norm": 1.5078125, "learning_rate": 5.317810631655477e-05, "loss": 0.1503, "step": 24301 }, { "epoch": 17.803663003663004, "grad_norm": 0.7890625, "learning_rate": 5.3174684967203816e-05, "loss": 0.1343, "step": 24302 }, { "epoch": 17.804395604395605, "grad_norm": 1.234375, "learning_rate": 5.317126360503756e-05, "loss": 0.139, "step": 24303 }, { "epoch": 17.805128205128206, "grad_norm": 1.5546875, "learning_rate": 5.3167842230072395e-05, "loss": 0.1296, "step": 24304 }, { "epoch": 17.805860805860807, "grad_norm": 0.984375, "learning_rate": 5.316442084232475e-05, "loss": 0.1305, "step": 24305 }, { "epoch": 17.806593406593407, "grad_norm": 0.68359375, "learning_rate": 5.3160999441810954e-05, "loss": 0.1356, "step": 24306 }, { "epoch": 17.807326007326008, "grad_norm": 1.515625, "learning_rate": 5.315757802854745e-05, "loss": 0.1376, "step": 24307 }, { "epoch": 17.80805860805861, "grad_norm": 1.34375, "learning_rate": 5.315415660255062e-05, "loss": 0.1314, "step": 24308 }, { "epoch": 17.80879120879121, "grad_norm": 1.0859375, "learning_rate": 5.315073516383685e-05, "loss": 0.1294, "step": 24309 }, { "epoch": 17.80952380952381, "grad_norm": 0.79296875, "learning_rate": 5.314731371242255e-05, "loss": 0.1131, "step": 24310 }, { "epoch": 17.81025641025641, "grad_norm": 1.3125, "learning_rate": 5.314389224832409e-05, "loss": 0.1468, "step": 24311 }, { "epoch": 17.81098901098901, "grad_norm": 0.921875, "learning_rate": 5.314047077155787e-05, "loss": 0.1172, "step": 24312 }, { "epoch": 17.811721611721612, "grad_norm": 0.75, "learning_rate": 5.3137049282140316e-05, "loss": 0.1257, "step": 24313 }, { "epoch": 17.812454212454213, "grad_norm": 0.62890625, "learning_rate": 5.313362778008777e-05, "loss": 0.1042, "step": 24314 }, { "epoch": 17.813186813186814, "grad_norm": 0.91015625, "learning_rate": 5.313020626541665e-05, "loss": 0.1287, "step": 24315 }, { "epoch": 17.813919413919415, "grad_norm": 0.9296875, "learning_rate": 5.3126784738143356e-05, "loss": 0.1475, "step": 24316 }, { "epoch": 17.814652014652015, "grad_norm": 0.953125, "learning_rate": 5.312336319828426e-05, "loss": 0.1458, "step": 24317 }, { "epoch": 17.815384615384616, "grad_norm": 0.796875, "learning_rate": 5.311994164585578e-05, "loss": 0.1197, "step": 24318 }, { "epoch": 17.816117216117217, "grad_norm": 0.8984375, "learning_rate": 5.3116520080874286e-05, "loss": 0.1469, "step": 24319 }, { "epoch": 17.816849816849818, "grad_norm": 0.72265625, "learning_rate": 5.311309850335619e-05, "loss": 0.105, "step": 24320 }, { "epoch": 17.81758241758242, "grad_norm": 1.1171875, "learning_rate": 5.3109676913317895e-05, "loss": 0.1145, "step": 24321 }, { "epoch": 17.81831501831502, "grad_norm": 1.515625, "learning_rate": 5.3106255310775765e-05, "loss": 0.1204, "step": 24322 }, { "epoch": 17.81904761904762, "grad_norm": 1.2421875, "learning_rate": 5.3102833695746235e-05, "loss": 0.113, "step": 24323 }, { "epoch": 17.81978021978022, "grad_norm": 1.0234375, "learning_rate": 5.3099412068245656e-05, "loss": 0.1541, "step": 24324 }, { "epoch": 17.82051282051282, "grad_norm": 0.9765625, "learning_rate": 5.309599042829044e-05, "loss": 0.1051, "step": 24325 }, { "epoch": 17.821245421245422, "grad_norm": 1.421875, "learning_rate": 5.309256877589699e-05, "loss": 0.1262, "step": 24326 }, { "epoch": 17.821978021978023, "grad_norm": 0.8125, "learning_rate": 5.3089147111081684e-05, "loss": 0.139, "step": 24327 }, { "epoch": 17.822710622710623, "grad_norm": 1.359375, "learning_rate": 5.3085725433860935e-05, "loss": 0.1257, "step": 24328 }, { "epoch": 17.823443223443224, "grad_norm": 0.9765625, "learning_rate": 5.3082303744251116e-05, "loss": 0.1366, "step": 24329 }, { "epoch": 17.824175824175825, "grad_norm": 0.8125, "learning_rate": 5.3078882042268626e-05, "loss": 0.1313, "step": 24330 }, { "epoch": 17.824908424908426, "grad_norm": 1.40625, "learning_rate": 5.3075460327929884e-05, "loss": 0.1088, "step": 24331 }, { "epoch": 17.825641025641026, "grad_norm": 1.8046875, "learning_rate": 5.3072038601251254e-05, "loss": 0.1365, "step": 24332 }, { "epoch": 17.826373626373627, "grad_norm": 1.03125, "learning_rate": 5.3068616862249134e-05, "loss": 0.1482, "step": 24333 }, { "epoch": 17.827106227106228, "grad_norm": 1.015625, "learning_rate": 5.306519511093994e-05, "loss": 0.0989, "step": 24334 }, { "epoch": 17.82783882783883, "grad_norm": 1.0234375, "learning_rate": 5.306177334734005e-05, "loss": 0.1268, "step": 24335 }, { "epoch": 17.82857142857143, "grad_norm": 0.8671875, "learning_rate": 5.305835157146587e-05, "loss": 0.1318, "step": 24336 }, { "epoch": 17.82930402930403, "grad_norm": 0.82421875, "learning_rate": 5.3054929783333785e-05, "loss": 0.0886, "step": 24337 }, { "epoch": 17.83003663003663, "grad_norm": 1.1953125, "learning_rate": 5.305150798296019e-05, "loss": 0.1315, "step": 24338 }, { "epoch": 17.83076923076923, "grad_norm": 1.1953125, "learning_rate": 5.304808617036148e-05, "loss": 0.1421, "step": 24339 }, { "epoch": 17.831501831501832, "grad_norm": 1.859375, "learning_rate": 5.304466434555406e-05, "loss": 0.1375, "step": 24340 }, { "epoch": 17.832234432234433, "grad_norm": 1.2265625, "learning_rate": 5.304124250855431e-05, "loss": 0.1119, "step": 24341 }, { "epoch": 17.832967032967034, "grad_norm": 1.9765625, "learning_rate": 5.303782065937864e-05, "loss": 0.1225, "step": 24342 }, { "epoch": 17.833699633699634, "grad_norm": 1.421875, "learning_rate": 5.303439879804343e-05, "loss": 0.1282, "step": 24343 }, { "epoch": 17.834432234432235, "grad_norm": 1.2265625, "learning_rate": 5.303097692456509e-05, "loss": 0.1337, "step": 24344 }, { "epoch": 17.835164835164836, "grad_norm": 0.90234375, "learning_rate": 5.302755503896001e-05, "loss": 0.1561, "step": 24345 }, { "epoch": 17.835897435897436, "grad_norm": 0.99609375, "learning_rate": 5.302413314124457e-05, "loss": 0.1483, "step": 24346 }, { "epoch": 17.836630036630037, "grad_norm": 1.328125, "learning_rate": 5.3020711231435194e-05, "loss": 0.1484, "step": 24347 }, { "epoch": 17.837362637362638, "grad_norm": 0.91015625, "learning_rate": 5.301728930954826e-05, "loss": 0.1229, "step": 24348 }, { "epoch": 17.83809523809524, "grad_norm": 0.984375, "learning_rate": 5.3013867375600166e-05, "loss": 0.1044, "step": 24349 }, { "epoch": 17.83882783882784, "grad_norm": 1.875, "learning_rate": 5.301044542960731e-05, "loss": 0.1077, "step": 24350 }, { "epoch": 17.83956043956044, "grad_norm": 1.609375, "learning_rate": 5.300702347158608e-05, "loss": 0.1283, "step": 24351 }, { "epoch": 17.84029304029304, "grad_norm": 1.1015625, "learning_rate": 5.300360150155289e-05, "loss": 0.1212, "step": 24352 }, { "epoch": 17.84102564102564, "grad_norm": 0.98046875, "learning_rate": 5.300017951952412e-05, "loss": 0.149, "step": 24353 }, { "epoch": 17.841758241758242, "grad_norm": 2.359375, "learning_rate": 5.2996757525516164e-05, "loss": 0.1242, "step": 24354 }, { "epoch": 17.842490842490843, "grad_norm": 1.171875, "learning_rate": 5.299333551954543e-05, "loss": 0.1349, "step": 24355 }, { "epoch": 17.843223443223444, "grad_norm": 0.73046875, "learning_rate": 5.2989913501628274e-05, "loss": 0.1066, "step": 24356 }, { "epoch": 17.843956043956045, "grad_norm": 0.74609375, "learning_rate": 5.2986491471781163e-05, "loss": 0.1197, "step": 24357 }, { "epoch": 17.844688644688645, "grad_norm": 0.84375, "learning_rate": 5.2983069430020446e-05, "loss": 0.1068, "step": 24358 }, { "epoch": 17.845421245421246, "grad_norm": 2.234375, "learning_rate": 5.297964737636251e-05, "loss": 0.1536, "step": 24359 }, { "epoch": 17.846153846153847, "grad_norm": 2.359375, "learning_rate": 5.297622531082379e-05, "loss": 0.1175, "step": 24360 }, { "epoch": 17.846886446886447, "grad_norm": 1.734375, "learning_rate": 5.2972803233420654e-05, "loss": 0.1462, "step": 24361 }, { "epoch": 17.847619047619048, "grad_norm": 0.75, "learning_rate": 5.29693811441695e-05, "loss": 0.1198, "step": 24362 }, { "epoch": 17.84835164835165, "grad_norm": 1.109375, "learning_rate": 5.296595904308675e-05, "loss": 0.1151, "step": 24363 }, { "epoch": 17.84908424908425, "grad_norm": 0.93359375, "learning_rate": 5.2962536930188765e-05, "loss": 0.1056, "step": 24364 }, { "epoch": 17.84981684981685, "grad_norm": 1.0625, "learning_rate": 5.295911480549197e-05, "loss": 0.1399, "step": 24365 }, { "epoch": 17.85054945054945, "grad_norm": 1.203125, "learning_rate": 5.295569266901275e-05, "loss": 0.1381, "step": 24366 }, { "epoch": 17.851282051282052, "grad_norm": 1.578125, "learning_rate": 5.2952270520767487e-05, "loss": 0.1503, "step": 24367 }, { "epoch": 17.852014652014653, "grad_norm": 0.94921875, "learning_rate": 5.29488483607726e-05, "loss": 0.1072, "step": 24368 }, { "epoch": 17.852747252747253, "grad_norm": 1.0234375, "learning_rate": 5.294542618904445e-05, "loss": 0.127, "step": 24369 }, { "epoch": 17.853479853479854, "grad_norm": 1.1953125, "learning_rate": 5.29420040055995e-05, "loss": 0.1199, "step": 24370 }, { "epoch": 17.854212454212455, "grad_norm": 1.265625, "learning_rate": 5.293858181045408e-05, "loss": 0.1355, "step": 24371 }, { "epoch": 17.854945054945055, "grad_norm": 0.91015625, "learning_rate": 5.2935159603624615e-05, "loss": 0.0954, "step": 24372 }, { "epoch": 17.855677655677656, "grad_norm": 1.0703125, "learning_rate": 5.293173738512751e-05, "loss": 0.1323, "step": 24373 }, { "epoch": 17.856410256410257, "grad_norm": 1.0390625, "learning_rate": 5.2928315154979145e-05, "loss": 0.0961, "step": 24374 }, { "epoch": 17.857142857142858, "grad_norm": 1.3203125, "learning_rate": 5.2924892913195934e-05, "loss": 0.1297, "step": 24375 }, { "epoch": 17.85787545787546, "grad_norm": 1.3359375, "learning_rate": 5.292147065979427e-05, "loss": 0.1128, "step": 24376 }, { "epoch": 17.85860805860806, "grad_norm": 0.8828125, "learning_rate": 5.291804839479053e-05, "loss": 0.1274, "step": 24377 }, { "epoch": 17.85934065934066, "grad_norm": 0.98046875, "learning_rate": 5.291462611820114e-05, "loss": 0.1224, "step": 24378 }, { "epoch": 17.86007326007326, "grad_norm": 0.80859375, "learning_rate": 5.291120383004249e-05, "loss": 0.1226, "step": 24379 }, { "epoch": 17.86080586080586, "grad_norm": 0.97265625, "learning_rate": 5.290778153033094e-05, "loss": 0.0903, "step": 24380 }, { "epoch": 17.861538461538462, "grad_norm": 1.6171875, "learning_rate": 5.2904359219082946e-05, "loss": 0.1268, "step": 24381 }, { "epoch": 17.862271062271063, "grad_norm": 1.6171875, "learning_rate": 5.290093689631487e-05, "loss": 0.1396, "step": 24382 }, { "epoch": 17.863003663003663, "grad_norm": 1.515625, "learning_rate": 5.289751456204311e-05, "loss": 0.087, "step": 24383 }, { "epoch": 17.863736263736264, "grad_norm": 1.3515625, "learning_rate": 5.2894092216284066e-05, "loss": 0.1046, "step": 24384 }, { "epoch": 17.864468864468865, "grad_norm": 1.1796875, "learning_rate": 5.2890669859054154e-05, "loss": 0.1258, "step": 24385 }, { "epoch": 17.865201465201466, "grad_norm": 1.71875, "learning_rate": 5.288724749036976e-05, "loss": 0.1624, "step": 24386 }, { "epoch": 17.865934065934066, "grad_norm": 1.71875, "learning_rate": 5.2883825110247266e-05, "loss": 0.13, "step": 24387 }, { "epoch": 17.866666666666667, "grad_norm": 0.890625, "learning_rate": 5.2880402718703093e-05, "loss": 0.1187, "step": 24388 }, { "epoch": 17.867399267399268, "grad_norm": 1.2265625, "learning_rate": 5.287698031575363e-05, "loss": 0.1669, "step": 24389 }, { "epoch": 17.86813186813187, "grad_norm": 0.796875, "learning_rate": 5.287355790141527e-05, "loss": 0.1085, "step": 24390 }, { "epoch": 17.86886446886447, "grad_norm": 0.8359375, "learning_rate": 5.287013547570443e-05, "loss": 0.0861, "step": 24391 }, { "epoch": 17.86959706959707, "grad_norm": 1.328125, "learning_rate": 5.2866713038637485e-05, "loss": 0.12, "step": 24392 }, { "epoch": 17.87032967032967, "grad_norm": 1.015625, "learning_rate": 5.286329059023082e-05, "loss": 0.1282, "step": 24393 }, { "epoch": 17.87106227106227, "grad_norm": 1.421875, "learning_rate": 5.285986813050088e-05, "loss": 0.1704, "step": 24394 }, { "epoch": 17.871794871794872, "grad_norm": 1.3046875, "learning_rate": 5.285644565946401e-05, "loss": 0.1098, "step": 24395 }, { "epoch": 17.872527472527473, "grad_norm": 1.1484375, "learning_rate": 5.285302317713665e-05, "loss": 0.1399, "step": 24396 }, { "epoch": 17.873260073260074, "grad_norm": 0.7109375, "learning_rate": 5.284960068353518e-05, "loss": 0.1052, "step": 24397 }, { "epoch": 17.873992673992674, "grad_norm": 1.1171875, "learning_rate": 5.284617817867601e-05, "loss": 0.1298, "step": 24398 }, { "epoch": 17.874725274725275, "grad_norm": 1.3828125, "learning_rate": 5.284275566257553e-05, "loss": 0.0968, "step": 24399 }, { "epoch": 17.875457875457876, "grad_norm": 0.76171875, "learning_rate": 5.283933313525013e-05, "loss": 0.1367, "step": 24400 }, { "epoch": 17.876190476190477, "grad_norm": 1.40625, "learning_rate": 5.283591059671622e-05, "loss": 0.1257, "step": 24401 }, { "epoch": 17.876923076923077, "grad_norm": 0.76171875, "learning_rate": 5.28324880469902e-05, "loss": 0.1038, "step": 24402 }, { "epoch": 17.877655677655678, "grad_norm": 1.234375, "learning_rate": 5.282906548608847e-05, "loss": 0.1368, "step": 24403 }, { "epoch": 17.87838827838828, "grad_norm": 1.03125, "learning_rate": 5.282564291402742e-05, "loss": 0.1169, "step": 24404 }, { "epoch": 17.87912087912088, "grad_norm": 1.3828125, "learning_rate": 5.282222033082345e-05, "loss": 0.1204, "step": 24405 }, { "epoch": 17.87985347985348, "grad_norm": 0.79296875, "learning_rate": 5.281879773649294e-05, "loss": 0.1221, "step": 24406 }, { "epoch": 17.88058608058608, "grad_norm": 1.3671875, "learning_rate": 5.2815375131052335e-05, "loss": 0.1164, "step": 24407 }, { "epoch": 17.88131868131868, "grad_norm": 1.0078125, "learning_rate": 5.281195251451799e-05, "loss": 0.141, "step": 24408 }, { "epoch": 17.882051282051282, "grad_norm": 0.9921875, "learning_rate": 5.280852988690632e-05, "loss": 0.1027, "step": 24409 }, { "epoch": 17.882783882783883, "grad_norm": 0.76171875, "learning_rate": 5.280510724823375e-05, "loss": 0.1241, "step": 24410 }, { "epoch": 17.883516483516484, "grad_norm": 0.96484375, "learning_rate": 5.280168459851662e-05, "loss": 0.1519, "step": 24411 }, { "epoch": 17.884249084249085, "grad_norm": 0.98828125, "learning_rate": 5.27982619377714e-05, "loss": 0.122, "step": 24412 }, { "epoch": 17.884981684981685, "grad_norm": 1.4453125, "learning_rate": 5.2794839266014415e-05, "loss": 0.1404, "step": 24413 }, { "epoch": 17.885714285714286, "grad_norm": 1.421875, "learning_rate": 5.279141658326213e-05, "loss": 0.1086, "step": 24414 }, { "epoch": 17.886446886446887, "grad_norm": 1.2578125, "learning_rate": 5.278799388953092e-05, "loss": 0.1602, "step": 24415 }, { "epoch": 17.887179487179488, "grad_norm": 1.34375, "learning_rate": 5.278457118483716e-05, "loss": 0.1417, "step": 24416 }, { "epoch": 17.88791208791209, "grad_norm": 2.5, "learning_rate": 5.278114846919727e-05, "loss": 0.1517, "step": 24417 }, { "epoch": 17.88864468864469, "grad_norm": 0.765625, "learning_rate": 5.277772574262767e-05, "loss": 0.1025, "step": 24418 }, { "epoch": 17.88937728937729, "grad_norm": 1.7265625, "learning_rate": 5.277430300514472e-05, "loss": 0.1438, "step": 24419 }, { "epoch": 17.89010989010989, "grad_norm": 1.34375, "learning_rate": 5.277088025676484e-05, "loss": 0.1494, "step": 24420 }, { "epoch": 17.89084249084249, "grad_norm": 1.140625, "learning_rate": 5.276745749750444e-05, "loss": 0.1179, "step": 24421 }, { "epoch": 17.891575091575092, "grad_norm": 1.1953125, "learning_rate": 5.2764034727379885e-05, "loss": 0.1559, "step": 24422 }, { "epoch": 17.892307692307693, "grad_norm": 1.3828125, "learning_rate": 5.276061194640761e-05, "loss": 0.1322, "step": 24423 }, { "epoch": 17.893040293040293, "grad_norm": 0.8359375, "learning_rate": 5.275718915460399e-05, "loss": 0.1155, "step": 24424 }, { "epoch": 17.893772893772894, "grad_norm": 1.0703125, "learning_rate": 5.2753766351985446e-05, "loss": 0.1097, "step": 24425 }, { "epoch": 17.894505494505495, "grad_norm": 1.484375, "learning_rate": 5.2750343538568355e-05, "loss": 0.1234, "step": 24426 }, { "epoch": 17.895238095238096, "grad_norm": 0.80078125, "learning_rate": 5.274692071436915e-05, "loss": 0.1191, "step": 24427 }, { "epoch": 17.895970695970696, "grad_norm": 0.91796875, "learning_rate": 5.27434978794042e-05, "loss": 0.1227, "step": 24428 }, { "epoch": 17.896703296703297, "grad_norm": 0.796875, "learning_rate": 5.274007503368992e-05, "loss": 0.0981, "step": 24429 }, { "epoch": 17.897435897435898, "grad_norm": 1.3046875, "learning_rate": 5.273665217724271e-05, "loss": 0.1145, "step": 24430 }, { "epoch": 17.8981684981685, "grad_norm": 0.86328125, "learning_rate": 5.2733229310078946e-05, "loss": 0.0991, "step": 24431 }, { "epoch": 17.8989010989011, "grad_norm": 0.9765625, "learning_rate": 5.272980643221507e-05, "loss": 0.1411, "step": 24432 }, { "epoch": 17.8996336996337, "grad_norm": 1.3125, "learning_rate": 5.2726383543667444e-05, "loss": 0.1292, "step": 24433 }, { "epoch": 17.9003663003663, "grad_norm": 1.1796875, "learning_rate": 5.272296064445247e-05, "loss": 0.1032, "step": 24434 }, { "epoch": 17.9010989010989, "grad_norm": 1.3203125, "learning_rate": 5.2719537734586576e-05, "loss": 0.0939, "step": 24435 }, { "epoch": 17.901831501831502, "grad_norm": 1.03125, "learning_rate": 5.271611481408616e-05, "loss": 0.1194, "step": 24436 }, { "epoch": 17.902564102564103, "grad_norm": 1.1953125, "learning_rate": 5.271269188296759e-05, "loss": 0.1343, "step": 24437 }, { "epoch": 17.903296703296704, "grad_norm": 1.9765625, "learning_rate": 5.27092689412473e-05, "loss": 0.1269, "step": 24438 }, { "epoch": 17.904029304029304, "grad_norm": 1.6875, "learning_rate": 5.270584598894167e-05, "loss": 0.1406, "step": 24439 }, { "epoch": 17.904761904761905, "grad_norm": 0.734375, "learning_rate": 5.270242302606713e-05, "loss": 0.1205, "step": 24440 }, { "epoch": 17.905494505494506, "grad_norm": 0.75390625, "learning_rate": 5.269900005264003e-05, "loss": 0.1489, "step": 24441 }, { "epoch": 17.906227106227107, "grad_norm": 1.4140625, "learning_rate": 5.26955770686768e-05, "loss": 0.1292, "step": 24442 }, { "epoch": 17.906959706959707, "grad_norm": 1.1796875, "learning_rate": 5.2692154074193856e-05, "loss": 0.1167, "step": 24443 }, { "epoch": 17.907692307692308, "grad_norm": 1.0078125, "learning_rate": 5.2688731069207587e-05, "loss": 0.1339, "step": 24444 }, { "epoch": 17.90842490842491, "grad_norm": 1.421875, "learning_rate": 5.268530805373437e-05, "loss": 0.1237, "step": 24445 }, { "epoch": 17.90915750915751, "grad_norm": 1.375, "learning_rate": 5.268188502779064e-05, "loss": 0.1288, "step": 24446 }, { "epoch": 17.90989010989011, "grad_norm": 1.015625, "learning_rate": 5.2678461991392766e-05, "loss": 0.1218, "step": 24447 }, { "epoch": 17.91062271062271, "grad_norm": 0.92578125, "learning_rate": 5.267503894455717e-05, "loss": 0.1028, "step": 24448 }, { "epoch": 17.91135531135531, "grad_norm": 1.2109375, "learning_rate": 5.267161588730025e-05, "loss": 0.1627, "step": 24449 }, { "epoch": 17.912087912087912, "grad_norm": 1.4296875, "learning_rate": 5.266819281963841e-05, "loss": 0.1423, "step": 24450 }, { "epoch": 17.912820512820513, "grad_norm": 1.078125, "learning_rate": 5.266476974158804e-05, "loss": 0.1163, "step": 24451 }, { "epoch": 17.913553113553114, "grad_norm": 1.2109375, "learning_rate": 5.2661346653165546e-05, "loss": 0.1299, "step": 24452 }, { "epoch": 17.914285714285715, "grad_norm": 0.62109375, "learning_rate": 5.2657923554387346e-05, "loss": 0.1147, "step": 24453 }, { "epoch": 17.915018315018315, "grad_norm": 0.96875, "learning_rate": 5.265450044526981e-05, "loss": 0.102, "step": 24454 }, { "epoch": 17.915750915750916, "grad_norm": 1.0234375, "learning_rate": 5.265107732582937e-05, "loss": 0.1224, "step": 24455 }, { "epoch": 17.916483516483517, "grad_norm": 0.9453125, "learning_rate": 5.264765419608241e-05, "loss": 0.1415, "step": 24456 }, { "epoch": 17.917216117216118, "grad_norm": 1.0859375, "learning_rate": 5.264423105604532e-05, "loss": 0.1328, "step": 24457 }, { "epoch": 17.91794871794872, "grad_norm": 1.0546875, "learning_rate": 5.2640807905734524e-05, "loss": 0.1287, "step": 24458 }, { "epoch": 17.91868131868132, "grad_norm": 0.80078125, "learning_rate": 5.263738474516641e-05, "loss": 0.1023, "step": 24459 }, { "epoch": 17.91941391941392, "grad_norm": 1.3359375, "learning_rate": 5.263396157435738e-05, "loss": 0.1316, "step": 24460 }, { "epoch": 17.92014652014652, "grad_norm": 0.9453125, "learning_rate": 5.2630538393323836e-05, "loss": 0.1265, "step": 24461 }, { "epoch": 17.92087912087912, "grad_norm": 1.3359375, "learning_rate": 5.26271152020822e-05, "loss": 0.1489, "step": 24462 }, { "epoch": 17.921611721611722, "grad_norm": 0.71484375, "learning_rate": 5.262369200064886e-05, "loss": 0.1252, "step": 24463 }, { "epoch": 17.922344322344323, "grad_norm": 1.4296875, "learning_rate": 5.262026878904021e-05, "loss": 0.1358, "step": 24464 }, { "epoch": 17.923076923076923, "grad_norm": 1.28125, "learning_rate": 5.261684556727263e-05, "loss": 0.1359, "step": 24465 }, { "epoch": 17.923809523809524, "grad_norm": 1.546875, "learning_rate": 5.2613422335362584e-05, "loss": 0.1127, "step": 24466 }, { "epoch": 17.924542124542125, "grad_norm": 1.46875, "learning_rate": 5.2609999093326426e-05, "loss": 0.1477, "step": 24467 }, { "epoch": 17.925274725274726, "grad_norm": 1.2890625, "learning_rate": 5.260657584118056e-05, "loss": 0.1226, "step": 24468 }, { "epoch": 17.926007326007326, "grad_norm": 1.21875, "learning_rate": 5.260315257894141e-05, "loss": 0.1178, "step": 24469 }, { "epoch": 17.926739926739927, "grad_norm": 1.4921875, "learning_rate": 5.259972930662537e-05, "loss": 0.1154, "step": 24470 }, { "epoch": 17.927472527472528, "grad_norm": 1.0, "learning_rate": 5.259630602424882e-05, "loss": 0.1192, "step": 24471 }, { "epoch": 17.92820512820513, "grad_norm": 1.359375, "learning_rate": 5.2592882731828186e-05, "loss": 0.1312, "step": 24472 }, { "epoch": 17.92893772893773, "grad_norm": 1.3046875, "learning_rate": 5.258945942937986e-05, "loss": 0.1139, "step": 24473 }, { "epoch": 17.92967032967033, "grad_norm": 1.4140625, "learning_rate": 5.2586036116920247e-05, "loss": 0.1223, "step": 24474 }, { "epoch": 17.93040293040293, "grad_norm": 0.99609375, "learning_rate": 5.258261279446576e-05, "loss": 0.1268, "step": 24475 }, { "epoch": 17.93113553113553, "grad_norm": 0.9453125, "learning_rate": 5.257918946203277e-05, "loss": 0.1217, "step": 24476 }, { "epoch": 17.931868131868132, "grad_norm": 1.2578125, "learning_rate": 5.2575766119637725e-05, "loss": 0.1121, "step": 24477 }, { "epoch": 17.932600732600733, "grad_norm": 1.2578125, "learning_rate": 5.2572342767297e-05, "loss": 0.1206, "step": 24478 }, { "epoch": 17.933333333333334, "grad_norm": 2.53125, "learning_rate": 5.256891940502701e-05, "loss": 0.1575, "step": 24479 }, { "epoch": 17.934065934065934, "grad_norm": 1.7421875, "learning_rate": 5.256549603284413e-05, "loss": 0.1073, "step": 24480 }, { "epoch": 17.934798534798535, "grad_norm": 1.9375, "learning_rate": 5.2562072650764774e-05, "loss": 0.1427, "step": 24481 }, { "epoch": 17.935531135531136, "grad_norm": 1.7109375, "learning_rate": 5.255864925880537e-05, "loss": 0.1255, "step": 24482 }, { "epoch": 17.936263736263736, "grad_norm": 1.09375, "learning_rate": 5.2555225856982294e-05, "loss": 0.1146, "step": 24483 }, { "epoch": 17.936996336996337, "grad_norm": 0.78515625, "learning_rate": 5.2551802445311945e-05, "loss": 0.1248, "step": 24484 }, { "epoch": 17.937728937728938, "grad_norm": 0.83984375, "learning_rate": 5.2548379023810756e-05, "loss": 0.1216, "step": 24485 }, { "epoch": 17.93846153846154, "grad_norm": 1.078125, "learning_rate": 5.2544955592495104e-05, "loss": 0.1187, "step": 24486 }, { "epoch": 17.93919413919414, "grad_norm": 0.890625, "learning_rate": 5.254153215138139e-05, "loss": 0.128, "step": 24487 }, { "epoch": 17.93992673992674, "grad_norm": 1.046875, "learning_rate": 5.2538108700486035e-05, "loss": 0.1465, "step": 24488 }, { "epoch": 17.94065934065934, "grad_norm": 0.98828125, "learning_rate": 5.253468523982542e-05, "loss": 0.1096, "step": 24489 }, { "epoch": 17.94139194139194, "grad_norm": 1.3828125, "learning_rate": 5.2531261769415974e-05, "loss": 0.1508, "step": 24490 }, { "epoch": 17.942124542124542, "grad_norm": 1.8671875, "learning_rate": 5.252783828927408e-05, "loss": 0.1236, "step": 24491 }, { "epoch": 17.942857142857143, "grad_norm": 1.40625, "learning_rate": 5.2524414799416156e-05, "loss": 0.1144, "step": 24492 }, { "epoch": 17.943589743589744, "grad_norm": 1.34375, "learning_rate": 5.252099129985859e-05, "loss": 0.1219, "step": 24493 }, { "epoch": 17.944322344322345, "grad_norm": 1.3203125, "learning_rate": 5.2517567790617785e-05, "loss": 0.1136, "step": 24494 }, { "epoch": 17.945054945054945, "grad_norm": 1.6953125, "learning_rate": 5.251414427171015e-05, "loss": 0.1011, "step": 24495 }, { "epoch": 17.945787545787546, "grad_norm": 1.8203125, "learning_rate": 5.251072074315211e-05, "loss": 0.1313, "step": 24496 }, { "epoch": 17.946520146520147, "grad_norm": 1.1328125, "learning_rate": 5.2507297204960014e-05, "loss": 0.092, "step": 24497 }, { "epoch": 17.947252747252747, "grad_norm": 1.390625, "learning_rate": 5.250387365715033e-05, "loss": 0.1415, "step": 24498 }, { "epoch": 17.947985347985348, "grad_norm": 0.9140625, "learning_rate": 5.2500450099739396e-05, "loss": 0.1243, "step": 24499 }, { "epoch": 17.94871794871795, "grad_norm": 1.328125, "learning_rate": 5.249702653274366e-05, "loss": 0.1095, "step": 24500 }, { "epoch": 17.94945054945055, "grad_norm": 1.71875, "learning_rate": 5.249360295617953e-05, "loss": 0.1384, "step": 24501 }, { "epoch": 17.95018315018315, "grad_norm": 0.99609375, "learning_rate": 5.2490179370063386e-05, "loss": 0.1248, "step": 24502 }, { "epoch": 17.95091575091575, "grad_norm": 1.625, "learning_rate": 5.248675577441163e-05, "loss": 0.1076, "step": 24503 }, { "epoch": 17.951648351648352, "grad_norm": 1.125, "learning_rate": 5.248333216924069e-05, "loss": 0.1004, "step": 24504 }, { "epoch": 17.952380952380953, "grad_norm": 0.85546875, "learning_rate": 5.247990855456696e-05, "loss": 0.1107, "step": 24505 }, { "epoch": 17.953113553113553, "grad_norm": 0.9140625, "learning_rate": 5.247648493040682e-05, "loss": 0.1205, "step": 24506 }, { "epoch": 17.953846153846154, "grad_norm": 1.0, "learning_rate": 5.247306129677669e-05, "loss": 0.1136, "step": 24507 }, { "epoch": 17.954578754578755, "grad_norm": 1.046875, "learning_rate": 5.246963765369299e-05, "loss": 0.1072, "step": 24508 }, { "epoch": 17.955311355311355, "grad_norm": 1.0625, "learning_rate": 5.246621400117211e-05, "loss": 0.1557, "step": 24509 }, { "epoch": 17.956043956043956, "grad_norm": 1.21875, "learning_rate": 5.246279033923044e-05, "loss": 0.1442, "step": 24510 }, { "epoch": 17.956776556776557, "grad_norm": 1.3359375, "learning_rate": 5.24593666678844e-05, "loss": 0.1588, "step": 24511 }, { "epoch": 17.957509157509158, "grad_norm": 1.15625, "learning_rate": 5.24559429871504e-05, "loss": 0.1435, "step": 24512 }, { "epoch": 17.95824175824176, "grad_norm": 0.796875, "learning_rate": 5.2452519297044824e-05, "loss": 0.114, "step": 24513 }, { "epoch": 17.95897435897436, "grad_norm": 0.78515625, "learning_rate": 5.2449095597584094e-05, "loss": 0.1431, "step": 24514 }, { "epoch": 17.95970695970696, "grad_norm": 1.5390625, "learning_rate": 5.2445671888784596e-05, "loss": 0.1168, "step": 24515 }, { "epoch": 17.96043956043956, "grad_norm": 1.25, "learning_rate": 5.244224817066277e-05, "loss": 0.1163, "step": 24516 }, { "epoch": 17.96117216117216, "grad_norm": 1.125, "learning_rate": 5.2438824443234956e-05, "loss": 0.1483, "step": 24517 }, { "epoch": 17.961904761904762, "grad_norm": 0.8984375, "learning_rate": 5.243540070651763e-05, "loss": 0.1171, "step": 24518 }, { "epoch": 17.962637362637363, "grad_norm": 1.0625, "learning_rate": 5.243197696052716e-05, "loss": 0.1172, "step": 24519 }, { "epoch": 17.963369963369964, "grad_norm": 0.9609375, "learning_rate": 5.242855320527996e-05, "loss": 0.1199, "step": 24520 }, { "epoch": 17.964102564102564, "grad_norm": 1.09375, "learning_rate": 5.242512944079242e-05, "loss": 0.1461, "step": 24521 }, { "epoch": 17.964835164835165, "grad_norm": 2.171875, "learning_rate": 5.242170566708095e-05, "loss": 0.1287, "step": 24522 }, { "epoch": 17.965567765567766, "grad_norm": 1.125, "learning_rate": 5.241828188416196e-05, "loss": 0.1363, "step": 24523 }, { "epoch": 17.966300366300366, "grad_norm": 1.3984375, "learning_rate": 5.2414858092051845e-05, "loss": 0.1057, "step": 24524 }, { "epoch": 17.967032967032967, "grad_norm": 1.0, "learning_rate": 5.241143429076703e-05, "loss": 0.1196, "step": 24525 }, { "epoch": 17.967765567765568, "grad_norm": 1.2421875, "learning_rate": 5.2408010480323874e-05, "loss": 0.1338, "step": 24526 }, { "epoch": 17.96849816849817, "grad_norm": 1.390625, "learning_rate": 5.240458666073884e-05, "loss": 0.12, "step": 24527 }, { "epoch": 17.96923076923077, "grad_norm": 1.5234375, "learning_rate": 5.2401162832028306e-05, "loss": 0.1121, "step": 24528 }, { "epoch": 17.96996336996337, "grad_norm": 1.734375, "learning_rate": 5.239773899420868e-05, "loss": 0.1355, "step": 24529 }, { "epoch": 17.97069597069597, "grad_norm": 1.0859375, "learning_rate": 5.239431514729635e-05, "loss": 0.1603, "step": 24530 }, { "epoch": 17.97142857142857, "grad_norm": 0.70703125, "learning_rate": 5.2390891291307745e-05, "loss": 0.1004, "step": 24531 }, { "epoch": 17.972161172161172, "grad_norm": 0.703125, "learning_rate": 5.2387467426259264e-05, "loss": 0.0907, "step": 24532 }, { "epoch": 17.972893772893773, "grad_norm": 1.171875, "learning_rate": 5.238404355216728e-05, "loss": 0.1207, "step": 24533 }, { "epoch": 17.973626373626374, "grad_norm": 1.078125, "learning_rate": 5.238061966904826e-05, "loss": 0.1241, "step": 24534 }, { "epoch": 17.974358974358974, "grad_norm": 0.7578125, "learning_rate": 5.2377195776918565e-05, "loss": 0.1244, "step": 24535 }, { "epoch": 17.975091575091575, "grad_norm": 0.671875, "learning_rate": 5.23737718757946e-05, "loss": 0.1227, "step": 24536 }, { "epoch": 17.975824175824176, "grad_norm": 1.3671875, "learning_rate": 5.2370347965692754e-05, "loss": 0.1468, "step": 24537 }, { "epoch": 17.976556776556777, "grad_norm": 1.3984375, "learning_rate": 5.2366924046629496e-05, "loss": 0.1326, "step": 24538 }, { "epoch": 17.977289377289377, "grad_norm": 0.8125, "learning_rate": 5.236350011862118e-05, "loss": 0.1034, "step": 24539 }, { "epoch": 17.978021978021978, "grad_norm": 0.9375, "learning_rate": 5.2360076181684225e-05, "loss": 0.1079, "step": 24540 }, { "epoch": 17.97875457875458, "grad_norm": 0.9296875, "learning_rate": 5.2356652235835015e-05, "loss": 0.1349, "step": 24541 }, { "epoch": 17.97948717948718, "grad_norm": 0.88671875, "learning_rate": 5.235322828108999e-05, "loss": 0.1127, "step": 24542 }, { "epoch": 17.98021978021978, "grad_norm": 1.0390625, "learning_rate": 5.234980431746556e-05, "loss": 0.1138, "step": 24543 }, { "epoch": 17.98095238095238, "grad_norm": 1.03125, "learning_rate": 5.234638034497808e-05, "loss": 0.0973, "step": 24544 }, { "epoch": 17.98168498168498, "grad_norm": 1.3828125, "learning_rate": 5.2342956363644e-05, "loss": 0.118, "step": 24545 }, { "epoch": 17.982417582417582, "grad_norm": 1.078125, "learning_rate": 5.2339532373479707e-05, "loss": 0.1231, "step": 24546 }, { "epoch": 17.983150183150183, "grad_norm": 0.9453125, "learning_rate": 5.2336108374501615e-05, "loss": 0.096, "step": 24547 }, { "epoch": 17.983882783882784, "grad_norm": 1.375, "learning_rate": 5.2332684366726124e-05, "loss": 0.1154, "step": 24548 }, { "epoch": 17.984615384615385, "grad_norm": 0.9140625, "learning_rate": 5.232926035016963e-05, "loss": 0.1081, "step": 24549 }, { "epoch": 17.985347985347985, "grad_norm": 1.265625, "learning_rate": 5.232583632484855e-05, "loss": 0.1452, "step": 24550 }, { "epoch": 17.986080586080586, "grad_norm": 0.9921875, "learning_rate": 5.232241229077931e-05, "loss": 0.1021, "step": 24551 }, { "epoch": 17.986813186813187, "grad_norm": 1.2578125, "learning_rate": 5.231898824797827e-05, "loss": 0.1384, "step": 24552 }, { "epoch": 17.987545787545788, "grad_norm": 1.109375, "learning_rate": 5.231556419646188e-05, "loss": 0.1042, "step": 24553 }, { "epoch": 17.98827838827839, "grad_norm": 0.609375, "learning_rate": 5.23121401362465e-05, "loss": 0.132, "step": 24554 }, { "epoch": 17.98901098901099, "grad_norm": 0.91015625, "learning_rate": 5.2308716067348586e-05, "loss": 0.1029, "step": 24555 }, { "epoch": 17.98974358974359, "grad_norm": 0.8046875, "learning_rate": 5.230529198978451e-05, "loss": 0.132, "step": 24556 }, { "epoch": 17.99047619047619, "grad_norm": 0.6796875, "learning_rate": 5.230186790357069e-05, "loss": 0.1238, "step": 24557 }, { "epoch": 17.99120879120879, "grad_norm": 0.78125, "learning_rate": 5.229844380872353e-05, "loss": 0.1036, "step": 24558 }, { "epoch": 17.991941391941392, "grad_norm": 1.109375, "learning_rate": 5.229501970525943e-05, "loss": 0.1194, "step": 24559 }, { "epoch": 17.992673992673993, "grad_norm": 1.0390625, "learning_rate": 5.229159559319481e-05, "loss": 0.1161, "step": 24560 }, { "epoch": 17.993406593406593, "grad_norm": 1.0390625, "learning_rate": 5.228817147254607e-05, "loss": 0.1202, "step": 24561 }, { "epoch": 17.994139194139194, "grad_norm": 1.3046875, "learning_rate": 5.22847473433296e-05, "loss": 0.1087, "step": 24562 }, { "epoch": 17.994871794871795, "grad_norm": 0.82421875, "learning_rate": 5.228132320556182e-05, "loss": 0.108, "step": 24563 }, { "epoch": 17.995604395604396, "grad_norm": 1.140625, "learning_rate": 5.227789905925914e-05, "loss": 0.1137, "step": 24564 }, { "epoch": 17.996336996336996, "grad_norm": 0.87109375, "learning_rate": 5.2274474904437954e-05, "loss": 0.1148, "step": 24565 }, { "epoch": 17.997069597069597, "grad_norm": 1.515625, "learning_rate": 5.22710507411147e-05, "loss": 0.1345, "step": 24566 }, { "epoch": 17.997802197802198, "grad_norm": 0.953125, "learning_rate": 5.226762656930574e-05, "loss": 0.1183, "step": 24567 }, { "epoch": 17.9985347985348, "grad_norm": 0.9921875, "learning_rate": 5.2264202389027514e-05, "loss": 0.1245, "step": 24568 }, { "epoch": 17.9992673992674, "grad_norm": 1.0234375, "learning_rate": 5.22607782002964e-05, "loss": 0.127, "step": 24569 }, { "epoch": 18.0, "grad_norm": 0.984375, "learning_rate": 5.225735400312883e-05, "loss": 0.1181, "step": 24570 }, { "epoch": 18.0007326007326, "grad_norm": 0.7890625, "learning_rate": 5.22539297975412e-05, "loss": 0.1071, "step": 24571 }, { "epoch": 18.0014652014652, "grad_norm": 0.58984375, "learning_rate": 5.2250505583549916e-05, "loss": 0.1022, "step": 24572 }, { "epoch": 18.002197802197802, "grad_norm": 1.0390625, "learning_rate": 5.224708136117139e-05, "loss": 0.1273, "step": 24573 }, { "epoch": 18.002930402930403, "grad_norm": 0.5078125, "learning_rate": 5.2243657130422005e-05, "loss": 0.1072, "step": 24574 }, { "epoch": 18.003663003663004, "grad_norm": 0.76953125, "learning_rate": 5.22402328913182e-05, "loss": 0.1145, "step": 24575 }, { "epoch": 18.004395604395604, "grad_norm": 0.95703125, "learning_rate": 5.223680864387637e-05, "loss": 0.1191, "step": 24576 }, { "epoch": 18.005128205128205, "grad_norm": 0.78515625, "learning_rate": 5.2233384388112905e-05, "loss": 0.0952, "step": 24577 }, { "epoch": 18.005860805860806, "grad_norm": 0.7734375, "learning_rate": 5.222996012404424e-05, "loss": 0.1025, "step": 24578 }, { "epoch": 18.006593406593407, "grad_norm": 0.99609375, "learning_rate": 5.222653585168678e-05, "loss": 0.1139, "step": 24579 }, { "epoch": 18.007326007326007, "grad_norm": 0.90625, "learning_rate": 5.2223111571056896e-05, "loss": 0.1093, "step": 24580 }, { "epoch": 18.008058608058608, "grad_norm": 1.2421875, "learning_rate": 5.2219687282171024e-05, "loss": 0.1306, "step": 24581 }, { "epoch": 18.00879120879121, "grad_norm": 1.2734375, "learning_rate": 5.2216262985045575e-05, "loss": 0.1369, "step": 24582 }, { "epoch": 18.00952380952381, "grad_norm": 0.86328125, "learning_rate": 5.221283867969694e-05, "loss": 0.1241, "step": 24583 }, { "epoch": 18.01025641025641, "grad_norm": 0.8046875, "learning_rate": 5.220941436614154e-05, "loss": 0.1153, "step": 24584 }, { "epoch": 18.01098901098901, "grad_norm": 1.296875, "learning_rate": 5.2205990044395755e-05, "loss": 0.1341, "step": 24585 }, { "epoch": 18.01172161172161, "grad_norm": 0.72265625, "learning_rate": 5.220256571447602e-05, "loss": 0.1043, "step": 24586 }, { "epoch": 18.012454212454212, "grad_norm": 1.0859375, "learning_rate": 5.219914137639875e-05, "loss": 0.1015, "step": 24587 }, { "epoch": 18.013186813186813, "grad_norm": 1.21875, "learning_rate": 5.219571703018034e-05, "loss": 0.1285, "step": 24588 }, { "epoch": 18.013919413919414, "grad_norm": 1.265625, "learning_rate": 5.219229267583715e-05, "loss": 0.1582, "step": 24589 }, { "epoch": 18.014652014652015, "grad_norm": 1.09375, "learning_rate": 5.218886831338567e-05, "loss": 0.1246, "step": 24590 }, { "epoch": 18.015384615384615, "grad_norm": 1.4375, "learning_rate": 5.218544394284225e-05, "loss": 0.124, "step": 24591 }, { "epoch": 18.016117216117216, "grad_norm": 2.1875, "learning_rate": 5.2182019564223325e-05, "loss": 0.1785, "step": 24592 }, { "epoch": 18.016849816849817, "grad_norm": 0.77734375, "learning_rate": 5.217859517754528e-05, "loss": 0.1037, "step": 24593 }, { "epoch": 18.017582417582418, "grad_norm": 1.140625, "learning_rate": 5.217517078282454e-05, "loss": 0.1258, "step": 24594 }, { "epoch": 18.01831501831502, "grad_norm": 0.94921875, "learning_rate": 5.217174638007753e-05, "loss": 0.109, "step": 24595 }, { "epoch": 18.01904761904762, "grad_norm": 1.6875, "learning_rate": 5.216832196932059e-05, "loss": 0.1325, "step": 24596 }, { "epoch": 18.01978021978022, "grad_norm": 1.40625, "learning_rate": 5.2164897550570214e-05, "loss": 0.1234, "step": 24597 }, { "epoch": 18.02051282051282, "grad_norm": 1.78125, "learning_rate": 5.216147312384274e-05, "loss": 0.17, "step": 24598 }, { "epoch": 18.02124542124542, "grad_norm": 1.2578125, "learning_rate": 5.2158048689154614e-05, "loss": 0.1335, "step": 24599 }, { "epoch": 18.021978021978022, "grad_norm": 1.2265625, "learning_rate": 5.2154624246522234e-05, "loss": 0.1172, "step": 24600 }, { "epoch": 18.022710622710623, "grad_norm": 0.91796875, "learning_rate": 5.215119979596199e-05, "loss": 0.1243, "step": 24601 }, { "epoch": 18.023443223443223, "grad_norm": 1.296875, "learning_rate": 5.214777533749031e-05, "loss": 0.1363, "step": 24602 }, { "epoch": 18.024175824175824, "grad_norm": 1.15625, "learning_rate": 5.2144350871123606e-05, "loss": 0.1235, "step": 24603 }, { "epoch": 18.024908424908425, "grad_norm": 1.0703125, "learning_rate": 5.214092639687829e-05, "loss": 0.1295, "step": 24604 }, { "epoch": 18.025641025641026, "grad_norm": 1.1796875, "learning_rate": 5.213750191477074e-05, "loss": 0.1157, "step": 24605 }, { "epoch": 18.026373626373626, "grad_norm": 1.0078125, "learning_rate": 5.213407742481737e-05, "loss": 0.14, "step": 24606 }, { "epoch": 18.027106227106227, "grad_norm": 1.1953125, "learning_rate": 5.213065292703462e-05, "loss": 0.1335, "step": 24607 }, { "epoch": 18.027838827838828, "grad_norm": 0.7890625, "learning_rate": 5.2127228421438866e-05, "loss": 0.1034, "step": 24608 }, { "epoch": 18.02857142857143, "grad_norm": 0.9765625, "learning_rate": 5.212380390804653e-05, "loss": 0.1442, "step": 24609 }, { "epoch": 18.02930402930403, "grad_norm": 0.97265625, "learning_rate": 5.212037938687403e-05, "loss": 0.1424, "step": 24610 }, { "epoch": 18.03003663003663, "grad_norm": 0.84765625, "learning_rate": 5.211695485793773e-05, "loss": 0.1242, "step": 24611 }, { "epoch": 18.03076923076923, "grad_norm": 0.82421875, "learning_rate": 5.21135303212541e-05, "loss": 0.0874, "step": 24612 }, { "epoch": 18.03150183150183, "grad_norm": 1.453125, "learning_rate": 5.2110105776839504e-05, "loss": 0.136, "step": 24613 }, { "epoch": 18.032234432234432, "grad_norm": 1.0234375, "learning_rate": 5.210668122471034e-05, "loss": 0.1089, "step": 24614 }, { "epoch": 18.032967032967033, "grad_norm": 0.78515625, "learning_rate": 5.210325666488306e-05, "loss": 0.1194, "step": 24615 }, { "epoch": 18.033699633699634, "grad_norm": 1.375, "learning_rate": 5.209983209737405e-05, "loss": 0.1284, "step": 24616 }, { "epoch": 18.034432234432234, "grad_norm": 1.3359375, "learning_rate": 5.209640752219972e-05, "loss": 0.1214, "step": 24617 }, { "epoch": 18.035164835164835, "grad_norm": 1.0703125, "learning_rate": 5.209298293937648e-05, "loss": 0.1091, "step": 24618 }, { "epoch": 18.035897435897436, "grad_norm": 0.9453125, "learning_rate": 5.208955834892073e-05, "loss": 0.1, "step": 24619 }, { "epoch": 18.036630036630036, "grad_norm": 1.0546875, "learning_rate": 5.2086133750848886e-05, "loss": 0.1327, "step": 24620 }, { "epoch": 18.037362637362637, "grad_norm": 1.546875, "learning_rate": 5.2082709145177354e-05, "loss": 0.1332, "step": 24621 }, { "epoch": 18.038095238095238, "grad_norm": 1.078125, "learning_rate": 5.2079284531922546e-05, "loss": 0.145, "step": 24622 }, { "epoch": 18.03882783882784, "grad_norm": 1.359375, "learning_rate": 5.207585991110087e-05, "loss": 0.1231, "step": 24623 }, { "epoch": 18.03956043956044, "grad_norm": 1.3828125, "learning_rate": 5.207243528272872e-05, "loss": 0.1242, "step": 24624 }, { "epoch": 18.04029304029304, "grad_norm": 0.73046875, "learning_rate": 5.206901064682253e-05, "loss": 0.1082, "step": 24625 }, { "epoch": 18.04102564102564, "grad_norm": 0.828125, "learning_rate": 5.206558600339869e-05, "loss": 0.111, "step": 24626 }, { "epoch": 18.04175824175824, "grad_norm": 0.71875, "learning_rate": 5.20621613524736e-05, "loss": 0.1222, "step": 24627 }, { "epoch": 18.042490842490842, "grad_norm": 1.203125, "learning_rate": 5.205873669406369e-05, "loss": 0.1242, "step": 24628 }, { "epoch": 18.043223443223443, "grad_norm": 1.1953125, "learning_rate": 5.2055312028185356e-05, "loss": 0.1339, "step": 24629 }, { "epoch": 18.043956043956044, "grad_norm": 1.1484375, "learning_rate": 5.205188735485502e-05, "loss": 0.0929, "step": 24630 }, { "epoch": 18.044688644688645, "grad_norm": 1.21875, "learning_rate": 5.204846267408909e-05, "loss": 0.1039, "step": 24631 }, { "epoch": 18.045421245421245, "grad_norm": 1.2109375, "learning_rate": 5.204503798590395e-05, "loss": 0.1072, "step": 24632 }, { "epoch": 18.046153846153846, "grad_norm": 0.7578125, "learning_rate": 5.204161329031605e-05, "loss": 0.1121, "step": 24633 }, { "epoch": 18.046886446886447, "grad_norm": 1.7890625, "learning_rate": 5.2038188587341766e-05, "loss": 0.1209, "step": 24634 }, { "epoch": 18.047619047619047, "grad_norm": 0.8671875, "learning_rate": 5.203476387699751e-05, "loss": 0.1184, "step": 24635 }, { "epoch": 18.048351648351648, "grad_norm": 1.1328125, "learning_rate": 5.203133915929969e-05, "loss": 0.1274, "step": 24636 }, { "epoch": 18.04908424908425, "grad_norm": 0.89453125, "learning_rate": 5.202791443426472e-05, "loss": 0.1094, "step": 24637 }, { "epoch": 18.04981684981685, "grad_norm": 0.90625, "learning_rate": 5.202448970190902e-05, "loss": 0.1594, "step": 24638 }, { "epoch": 18.05054945054945, "grad_norm": 0.6640625, "learning_rate": 5.202106496224899e-05, "loss": 0.0922, "step": 24639 }, { "epoch": 18.05128205128205, "grad_norm": 0.60546875, "learning_rate": 5.2017640215301034e-05, "loss": 0.1087, "step": 24640 }, { "epoch": 18.052014652014652, "grad_norm": 1.2578125, "learning_rate": 5.201421546108156e-05, "loss": 0.1267, "step": 24641 }, { "epoch": 18.052747252747253, "grad_norm": 1.2421875, "learning_rate": 5.2010790699606984e-05, "loss": 0.1422, "step": 24642 }, { "epoch": 18.053479853479853, "grad_norm": 0.58203125, "learning_rate": 5.2007365930893726e-05, "loss": 0.1, "step": 24643 }, { "epoch": 18.054212454212454, "grad_norm": 0.91015625, "learning_rate": 5.2003941154958183e-05, "loss": 0.1016, "step": 24644 }, { "epoch": 18.054945054945055, "grad_norm": 1.0, "learning_rate": 5.200051637181674e-05, "loss": 0.1248, "step": 24645 }, { "epoch": 18.055677655677655, "grad_norm": 1.359375, "learning_rate": 5.1997091581485854e-05, "loss": 0.119, "step": 24646 }, { "epoch": 18.056410256410256, "grad_norm": 1.015625, "learning_rate": 5.199366678398192e-05, "loss": 0.1097, "step": 24647 }, { "epoch": 18.057142857142857, "grad_norm": 0.96875, "learning_rate": 5.199024197932131e-05, "loss": 0.1129, "step": 24648 }, { "epoch": 18.057875457875458, "grad_norm": 1.0859375, "learning_rate": 5.1986817167520486e-05, "loss": 0.1272, "step": 24649 }, { "epoch": 18.05860805860806, "grad_norm": 1.5390625, "learning_rate": 5.1983392348595805e-05, "loss": 0.1198, "step": 24650 }, { "epoch": 18.05934065934066, "grad_norm": 1.1875, "learning_rate": 5.197996752256373e-05, "loss": 0.1099, "step": 24651 }, { "epoch": 18.06007326007326, "grad_norm": 0.81640625, "learning_rate": 5.197654268944062e-05, "loss": 0.1137, "step": 24652 }, { "epoch": 18.06080586080586, "grad_norm": 0.80859375, "learning_rate": 5.1973117849242936e-05, "loss": 0.1247, "step": 24653 }, { "epoch": 18.06153846153846, "grad_norm": 0.953125, "learning_rate": 5.196969300198705e-05, "loss": 0.1118, "step": 24654 }, { "epoch": 18.062271062271062, "grad_norm": 1.1171875, "learning_rate": 5.1966268147689366e-05, "loss": 0.1311, "step": 24655 }, { "epoch": 18.063003663003663, "grad_norm": 0.85546875, "learning_rate": 5.1962843286366336e-05, "loss": 0.1125, "step": 24656 }, { "epoch": 18.063736263736264, "grad_norm": 1.1015625, "learning_rate": 5.195941841803433e-05, "loss": 0.1228, "step": 24657 }, { "epoch": 18.064468864468864, "grad_norm": 1.0390625, "learning_rate": 5.195599354270978e-05, "loss": 0.1122, "step": 24658 }, { "epoch": 18.065201465201465, "grad_norm": 1.46875, "learning_rate": 5.1952568660409085e-05, "loss": 0.1335, "step": 24659 }, { "epoch": 18.065934065934066, "grad_norm": 0.84765625, "learning_rate": 5.194914377114866e-05, "loss": 0.1327, "step": 24660 }, { "epoch": 18.066666666666666, "grad_norm": 1.2109375, "learning_rate": 5.1945718874944894e-05, "loss": 0.1115, "step": 24661 }, { "epoch": 18.067399267399267, "grad_norm": 2.1875, "learning_rate": 5.1942293971814236e-05, "loss": 0.1104, "step": 24662 }, { "epoch": 18.068131868131868, "grad_norm": 1.671875, "learning_rate": 5.193886906177304e-05, "loss": 0.1473, "step": 24663 }, { "epoch": 18.06886446886447, "grad_norm": 1.296875, "learning_rate": 5.193544414483779e-05, "loss": 0.1172, "step": 24664 }, { "epoch": 18.06959706959707, "grad_norm": 1.7109375, "learning_rate": 5.193201922102483e-05, "loss": 0.1108, "step": 24665 }, { "epoch": 18.07032967032967, "grad_norm": 1.671875, "learning_rate": 5.192859429035061e-05, "loss": 0.1328, "step": 24666 }, { "epoch": 18.07106227106227, "grad_norm": 1.125, "learning_rate": 5.192516935283152e-05, "loss": 0.1223, "step": 24667 }, { "epoch": 18.07179487179487, "grad_norm": 1.078125, "learning_rate": 5.1921744408483965e-05, "loss": 0.1145, "step": 24668 }, { "epoch": 18.072527472527472, "grad_norm": 1.1328125, "learning_rate": 5.1918319457324383e-05, "loss": 0.1166, "step": 24669 }, { "epoch": 18.073260073260073, "grad_norm": 1.0078125, "learning_rate": 5.191489449936916e-05, "loss": 0.1136, "step": 24670 }, { "epoch": 18.073992673992674, "grad_norm": 0.82421875, "learning_rate": 5.19114695346347e-05, "loss": 0.1291, "step": 24671 }, { "epoch": 18.074725274725274, "grad_norm": 1.0625, "learning_rate": 5.190804456313746e-05, "loss": 0.1254, "step": 24672 }, { "epoch": 18.075457875457875, "grad_norm": 1.1171875, "learning_rate": 5.19046195848938e-05, "loss": 0.14, "step": 24673 }, { "epoch": 18.076190476190476, "grad_norm": 1.203125, "learning_rate": 5.190119459992013e-05, "loss": 0.1134, "step": 24674 }, { "epoch": 18.076923076923077, "grad_norm": 1.328125, "learning_rate": 5.189776960823289e-05, "loss": 0.1113, "step": 24675 }, { "epoch": 18.077655677655677, "grad_norm": 0.8984375, "learning_rate": 5.189434460984848e-05, "loss": 0.104, "step": 24676 }, { "epoch": 18.078388278388278, "grad_norm": 1.1171875, "learning_rate": 5.189091960478331e-05, "loss": 0.1141, "step": 24677 }, { "epoch": 18.07912087912088, "grad_norm": 0.98046875, "learning_rate": 5.188749459305377e-05, "loss": 0.1601, "step": 24678 }, { "epoch": 18.07985347985348, "grad_norm": 1.265625, "learning_rate": 5.1884069574676295e-05, "loss": 0.1247, "step": 24679 }, { "epoch": 18.08058608058608, "grad_norm": 1.1171875, "learning_rate": 5.18806445496673e-05, "loss": 0.112, "step": 24680 }, { "epoch": 18.08131868131868, "grad_norm": 1.078125, "learning_rate": 5.1877219518043166e-05, "loss": 0.1385, "step": 24681 }, { "epoch": 18.08205128205128, "grad_norm": 0.83203125, "learning_rate": 5.187379447982033e-05, "loss": 0.1277, "step": 24682 }, { "epoch": 18.082783882783882, "grad_norm": 1.046875, "learning_rate": 5.187036943501521e-05, "loss": 0.1232, "step": 24683 }, { "epoch": 18.083516483516483, "grad_norm": 1.1171875, "learning_rate": 5.186694438364417e-05, "loss": 0.1239, "step": 24684 }, { "epoch": 18.084249084249084, "grad_norm": 1.2265625, "learning_rate": 5.1863519325723675e-05, "loss": 0.1093, "step": 24685 }, { "epoch": 18.084981684981685, "grad_norm": 1.0625, "learning_rate": 5.186009426127011e-05, "loss": 0.1181, "step": 24686 }, { "epoch": 18.085714285714285, "grad_norm": 1.828125, "learning_rate": 5.1856669190299866e-05, "loss": 0.1289, "step": 24687 }, { "epoch": 18.086446886446886, "grad_norm": 0.9453125, "learning_rate": 5.185324411282939e-05, "loss": 0.1197, "step": 24688 }, { "epoch": 18.087179487179487, "grad_norm": 1.09375, "learning_rate": 5.184981902887508e-05, "loss": 0.1313, "step": 24689 }, { "epoch": 18.087912087912088, "grad_norm": 1.3359375, "learning_rate": 5.184639393845333e-05, "loss": 0.1243, "step": 24690 }, { "epoch": 18.08864468864469, "grad_norm": 1.8046875, "learning_rate": 5.184296884158057e-05, "loss": 0.1671, "step": 24691 }, { "epoch": 18.08937728937729, "grad_norm": 1.328125, "learning_rate": 5.18395437382732e-05, "loss": 0.1291, "step": 24692 }, { "epoch": 18.09010989010989, "grad_norm": 1.0, "learning_rate": 5.1836118628547655e-05, "loss": 0.1475, "step": 24693 }, { "epoch": 18.09084249084249, "grad_norm": 1.1015625, "learning_rate": 5.18326935124203e-05, "loss": 0.1176, "step": 24694 }, { "epoch": 18.09157509157509, "grad_norm": 1.3203125, "learning_rate": 5.18292683899076e-05, "loss": 0.1292, "step": 24695 }, { "epoch": 18.092307692307692, "grad_norm": 1.265625, "learning_rate": 5.182584326102594e-05, "loss": 0.1355, "step": 24696 }, { "epoch": 18.093040293040293, "grad_norm": 1.171875, "learning_rate": 5.1822418125791714e-05, "loss": 0.1033, "step": 24697 }, { "epoch": 18.093772893772893, "grad_norm": 2.203125, "learning_rate": 5.181899298422136e-05, "loss": 0.1178, "step": 24698 }, { "epoch": 18.094505494505494, "grad_norm": 1.09375, "learning_rate": 5.181556783633127e-05, "loss": 0.1497, "step": 24699 }, { "epoch": 18.095238095238095, "grad_norm": 1.1640625, "learning_rate": 5.1812142682137865e-05, "loss": 0.0978, "step": 24700 }, { "epoch": 18.095970695970696, "grad_norm": 1.0390625, "learning_rate": 5.1808717521657544e-05, "loss": 0.1326, "step": 24701 }, { "epoch": 18.096703296703296, "grad_norm": 0.63671875, "learning_rate": 5.180529235490673e-05, "loss": 0.1053, "step": 24702 }, { "epoch": 18.097435897435897, "grad_norm": 1.4609375, "learning_rate": 5.180186718190184e-05, "loss": 0.1505, "step": 24703 }, { "epoch": 18.098168498168498, "grad_norm": 1.5390625, "learning_rate": 5.1798442002659274e-05, "loss": 0.1217, "step": 24704 }, { "epoch": 18.0989010989011, "grad_norm": 1.1484375, "learning_rate": 5.179501681719544e-05, "loss": 0.1131, "step": 24705 }, { "epoch": 18.0996336996337, "grad_norm": 1.140625, "learning_rate": 5.1791591625526767e-05, "loss": 0.1443, "step": 24706 }, { "epoch": 18.1003663003663, "grad_norm": 1.2421875, "learning_rate": 5.1788166427669646e-05, "loss": 0.1195, "step": 24707 }, { "epoch": 18.1010989010989, "grad_norm": 1.21875, "learning_rate": 5.17847412236405e-05, "loss": 0.1392, "step": 24708 }, { "epoch": 18.1018315018315, "grad_norm": 0.59375, "learning_rate": 5.178131601345575e-05, "loss": 0.1087, "step": 24709 }, { "epoch": 18.102564102564102, "grad_norm": 1.375, "learning_rate": 5.177789079713177e-05, "loss": 0.1503, "step": 24710 }, { "epoch": 18.103296703296703, "grad_norm": 1.2578125, "learning_rate": 5.1774465574685e-05, "loss": 0.1535, "step": 24711 }, { "epoch": 18.104029304029304, "grad_norm": 1.25, "learning_rate": 5.1771040346131875e-05, "loss": 0.1479, "step": 24712 }, { "epoch": 18.104761904761904, "grad_norm": 1.0078125, "learning_rate": 5.176761511148874e-05, "loss": 0.1246, "step": 24713 }, { "epoch": 18.105494505494505, "grad_norm": 1.40625, "learning_rate": 5.1764189870772074e-05, "loss": 0.1244, "step": 24714 }, { "epoch": 18.106227106227106, "grad_norm": 0.8359375, "learning_rate": 5.176076462399825e-05, "loss": 0.1059, "step": 24715 }, { "epoch": 18.106959706959707, "grad_norm": 1.0703125, "learning_rate": 5.1757339371183675e-05, "loss": 0.1009, "step": 24716 }, { "epoch": 18.107692307692307, "grad_norm": 0.6875, "learning_rate": 5.175391411234478e-05, "loss": 0.112, "step": 24717 }, { "epoch": 18.108424908424908, "grad_norm": 1.15625, "learning_rate": 5.175048884749798e-05, "loss": 0.1038, "step": 24718 }, { "epoch": 18.10915750915751, "grad_norm": 1.1171875, "learning_rate": 5.174706357665968e-05, "loss": 0.0874, "step": 24719 }, { "epoch": 18.10989010989011, "grad_norm": 0.94140625, "learning_rate": 5.1743638299846276e-05, "loss": 0.1115, "step": 24720 }, { "epoch": 18.11062271062271, "grad_norm": 0.91796875, "learning_rate": 5.1740213017074204e-05, "loss": 0.1268, "step": 24721 }, { "epoch": 18.11135531135531, "grad_norm": 0.93359375, "learning_rate": 5.173678772835986e-05, "loss": 0.1527, "step": 24722 }, { "epoch": 18.11208791208791, "grad_norm": 1.140625, "learning_rate": 5.1733362433719654e-05, "loss": 0.121, "step": 24723 }, { "epoch": 18.112820512820512, "grad_norm": 0.71875, "learning_rate": 5.172993713317003e-05, "loss": 0.1234, "step": 24724 }, { "epoch": 18.113553113553113, "grad_norm": 0.6796875, "learning_rate": 5.1726511826727355e-05, "loss": 0.0833, "step": 24725 }, { "epoch": 18.114285714285714, "grad_norm": 1.1796875, "learning_rate": 5.1723086514408036e-05, "loss": 0.1011, "step": 24726 }, { "epoch": 18.115018315018315, "grad_norm": 0.87890625, "learning_rate": 5.1719661196228544e-05, "loss": 0.121, "step": 24727 }, { "epoch": 18.115750915750915, "grad_norm": 0.7421875, "learning_rate": 5.1716235872205235e-05, "loss": 0.1177, "step": 24728 }, { "epoch": 18.116483516483516, "grad_norm": 0.79296875, "learning_rate": 5.171281054235455e-05, "loss": 0.1257, "step": 24729 }, { "epoch": 18.117216117216117, "grad_norm": 1.2109375, "learning_rate": 5.170938520669289e-05, "loss": 0.1294, "step": 24730 }, { "epoch": 18.117948717948718, "grad_norm": 1.3046875, "learning_rate": 5.170595986523665e-05, "loss": 0.1087, "step": 24731 }, { "epoch": 18.11868131868132, "grad_norm": 0.84765625, "learning_rate": 5.170253451800229e-05, "loss": 0.1039, "step": 24732 }, { "epoch": 18.11941391941392, "grad_norm": 1.28125, "learning_rate": 5.1699109165006164e-05, "loss": 0.1058, "step": 24733 }, { "epoch": 18.12014652014652, "grad_norm": 0.97265625, "learning_rate": 5.169568380626473e-05, "loss": 0.1439, "step": 24734 }, { "epoch": 18.12087912087912, "grad_norm": 1.234375, "learning_rate": 5.1692258441794386e-05, "loss": 0.1133, "step": 24735 }, { "epoch": 18.12161172161172, "grad_norm": 0.6171875, "learning_rate": 5.168883307161153e-05, "loss": 0.1117, "step": 24736 }, { "epoch": 18.122344322344322, "grad_norm": 1.0546875, "learning_rate": 5.168540769573259e-05, "loss": 0.0956, "step": 24737 }, { "epoch": 18.123076923076923, "grad_norm": 1.28125, "learning_rate": 5.168198231417397e-05, "loss": 0.1446, "step": 24738 }, { "epoch": 18.123809523809523, "grad_norm": 0.85546875, "learning_rate": 5.167855692695208e-05, "loss": 0.1155, "step": 24739 }, { "epoch": 18.124542124542124, "grad_norm": 1.515625, "learning_rate": 5.1675131534083344e-05, "loss": 0.119, "step": 24740 }, { "epoch": 18.125274725274725, "grad_norm": 1.4921875, "learning_rate": 5.167170613558416e-05, "loss": 0.1704, "step": 24741 }, { "epoch": 18.126007326007326, "grad_norm": 0.953125, "learning_rate": 5.1668280731470975e-05, "loss": 0.11, "step": 24742 }, { "epoch": 18.126739926739926, "grad_norm": 1.109375, "learning_rate": 5.166485532176014e-05, "loss": 0.1471, "step": 24743 }, { "epoch": 18.127472527472527, "grad_norm": 1.1328125, "learning_rate": 5.16614299064681e-05, "loss": 0.1498, "step": 24744 }, { "epoch": 18.128205128205128, "grad_norm": 1.3359375, "learning_rate": 5.165800448561129e-05, "loss": 0.1187, "step": 24745 }, { "epoch": 18.12893772893773, "grad_norm": 1.3046875, "learning_rate": 5.165457905920608e-05, "loss": 0.1129, "step": 24746 }, { "epoch": 18.12967032967033, "grad_norm": 1.1640625, "learning_rate": 5.1651153627268923e-05, "loss": 0.1346, "step": 24747 }, { "epoch": 18.13040293040293, "grad_norm": 0.984375, "learning_rate": 5.16477281898162e-05, "loss": 0.1379, "step": 24748 }, { "epoch": 18.13113553113553, "grad_norm": 0.85546875, "learning_rate": 5.164430274686434e-05, "loss": 0.1313, "step": 24749 }, { "epoch": 18.13186813186813, "grad_norm": 0.91015625, "learning_rate": 5.1640877298429744e-05, "loss": 0.1089, "step": 24750 }, { "epoch": 18.132600732600732, "grad_norm": 1.4375, "learning_rate": 5.163745184452884e-05, "loss": 0.1187, "step": 24751 }, { "epoch": 18.133333333333333, "grad_norm": 0.9296875, "learning_rate": 5.163402638517801e-05, "loss": 0.0931, "step": 24752 }, { "epoch": 18.134065934065934, "grad_norm": 0.78125, "learning_rate": 5.163060092039371e-05, "loss": 0.1352, "step": 24753 }, { "epoch": 18.134798534798534, "grad_norm": 0.515625, "learning_rate": 5.162717545019231e-05, "loss": 0.0988, "step": 24754 }, { "epoch": 18.135531135531135, "grad_norm": 0.92578125, "learning_rate": 5.1623749974590254e-05, "loss": 0.1291, "step": 24755 }, { "epoch": 18.136263736263736, "grad_norm": 0.63671875, "learning_rate": 5.1620324493603934e-05, "loss": 0.1142, "step": 24756 }, { "epoch": 18.136996336996337, "grad_norm": 0.9296875, "learning_rate": 5.1616899007249776e-05, "loss": 0.1121, "step": 24757 }, { "epoch": 18.137728937728937, "grad_norm": 1.265625, "learning_rate": 5.16134735155442e-05, "loss": 0.1188, "step": 24758 }, { "epoch": 18.138461538461538, "grad_norm": 1.0, "learning_rate": 5.16100480185036e-05, "loss": 0.1101, "step": 24759 }, { "epoch": 18.13919413919414, "grad_norm": 0.7734375, "learning_rate": 5.160662251614439e-05, "loss": 0.1128, "step": 24760 }, { "epoch": 18.13992673992674, "grad_norm": 1.375, "learning_rate": 5.1603197008482994e-05, "loss": 0.1429, "step": 24761 }, { "epoch": 18.14065934065934, "grad_norm": 0.7421875, "learning_rate": 5.159977149553581e-05, "loss": 0.1016, "step": 24762 }, { "epoch": 18.14139194139194, "grad_norm": 1.3671875, "learning_rate": 5.159634597731927e-05, "loss": 0.1159, "step": 24763 }, { "epoch": 18.14212454212454, "grad_norm": 1.0, "learning_rate": 5.159292045384978e-05, "loss": 0.1273, "step": 24764 }, { "epoch": 18.142857142857142, "grad_norm": 0.6328125, "learning_rate": 5.158949492514373e-05, "loss": 0.1099, "step": 24765 }, { "epoch": 18.143589743589743, "grad_norm": 0.8125, "learning_rate": 5.1586069391217575e-05, "loss": 0.1349, "step": 24766 }, { "epoch": 18.144322344322344, "grad_norm": 1.1015625, "learning_rate": 5.158264385208769e-05, "loss": 0.1167, "step": 24767 }, { "epoch": 18.145054945054945, "grad_norm": 1.734375, "learning_rate": 5.1579218307770504e-05, "loss": 0.1361, "step": 24768 }, { "epoch": 18.145787545787545, "grad_norm": 0.8125, "learning_rate": 5.1575792758282434e-05, "loss": 0.1072, "step": 24769 }, { "epoch": 18.146520146520146, "grad_norm": 0.85546875, "learning_rate": 5.157236720363987e-05, "loss": 0.1281, "step": 24770 }, { "epoch": 18.147252747252747, "grad_norm": 1.15625, "learning_rate": 5.156894164385927e-05, "loss": 0.1299, "step": 24771 }, { "epoch": 18.147985347985347, "grad_norm": 1.515625, "learning_rate": 5.156551607895701e-05, "loss": 0.112, "step": 24772 }, { "epoch": 18.148717948717948, "grad_norm": 0.5859375, "learning_rate": 5.156209050894951e-05, "loss": 0.1133, "step": 24773 }, { "epoch": 18.14945054945055, "grad_norm": 1.78125, "learning_rate": 5.155866493385319e-05, "loss": 0.1403, "step": 24774 }, { "epoch": 18.15018315018315, "grad_norm": 1.109375, "learning_rate": 5.1555239353684446e-05, "loss": 0.1172, "step": 24775 }, { "epoch": 18.15091575091575, "grad_norm": 1.015625, "learning_rate": 5.1551813768459724e-05, "loss": 0.1034, "step": 24776 }, { "epoch": 18.15164835164835, "grad_norm": 1.21875, "learning_rate": 5.1548388178195407e-05, "loss": 0.1356, "step": 24777 }, { "epoch": 18.152380952380952, "grad_norm": 1.6796875, "learning_rate": 5.1544962582907906e-05, "loss": 0.1225, "step": 24778 }, { "epoch": 18.153113553113553, "grad_norm": 1.375, "learning_rate": 5.154153698261367e-05, "loss": 0.1246, "step": 24779 }, { "epoch": 18.153846153846153, "grad_norm": 1.7734375, "learning_rate": 5.1538111377329065e-05, "loss": 0.141, "step": 24780 }, { "epoch": 18.154578754578754, "grad_norm": 0.98828125, "learning_rate": 5.1534685767070556e-05, "loss": 0.1436, "step": 24781 }, { "epoch": 18.155311355311355, "grad_norm": 1.0390625, "learning_rate": 5.153126015185449e-05, "loss": 0.1419, "step": 24782 }, { "epoch": 18.156043956043955, "grad_norm": 0.66015625, "learning_rate": 5.152783453169734e-05, "loss": 0.0982, "step": 24783 }, { "epoch": 18.156776556776556, "grad_norm": 1.25, "learning_rate": 5.1524408906615494e-05, "loss": 0.1162, "step": 24784 }, { "epoch": 18.157509157509157, "grad_norm": 1.0078125, "learning_rate": 5.152098327662538e-05, "loss": 0.1101, "step": 24785 }, { "epoch": 18.158241758241758, "grad_norm": 0.8828125, "learning_rate": 5.151755764174338e-05, "loss": 0.1314, "step": 24786 }, { "epoch": 18.15897435897436, "grad_norm": 1.3046875, "learning_rate": 5.151413200198595e-05, "loss": 0.1733, "step": 24787 }, { "epoch": 18.15970695970696, "grad_norm": 0.8203125, "learning_rate": 5.1510706357369466e-05, "loss": 0.0944, "step": 24788 }, { "epoch": 18.16043956043956, "grad_norm": 0.97265625, "learning_rate": 5.150728070791035e-05, "loss": 0.1256, "step": 24789 }, { "epoch": 18.16117216117216, "grad_norm": 1.40625, "learning_rate": 5.1503855053625036e-05, "loss": 0.1309, "step": 24790 }, { "epoch": 18.16190476190476, "grad_norm": 0.80859375, "learning_rate": 5.15004293945299e-05, "loss": 0.1253, "step": 24791 }, { "epoch": 18.162637362637362, "grad_norm": 0.9375, "learning_rate": 5.149700373064141e-05, "loss": 0.114, "step": 24792 }, { "epoch": 18.163369963369963, "grad_norm": 1.171875, "learning_rate": 5.1493578061975935e-05, "loss": 0.1068, "step": 24793 }, { "epoch": 18.164102564102564, "grad_norm": 0.9921875, "learning_rate": 5.1490152388549896e-05, "loss": 0.1284, "step": 24794 }, { "epoch": 18.164835164835164, "grad_norm": 0.87890625, "learning_rate": 5.1486726710379706e-05, "loss": 0.1233, "step": 24795 }, { "epoch": 18.165567765567765, "grad_norm": 1.4453125, "learning_rate": 5.148330102748179e-05, "loss": 0.1339, "step": 24796 }, { "epoch": 18.166300366300366, "grad_norm": 1.0546875, "learning_rate": 5.1479875339872566e-05, "loss": 0.1053, "step": 24797 }, { "epoch": 18.167032967032966, "grad_norm": 1.1796875, "learning_rate": 5.147644964756844e-05, "loss": 0.1126, "step": 24798 }, { "epoch": 18.167765567765567, "grad_norm": 1.078125, "learning_rate": 5.14730239505858e-05, "loss": 0.1402, "step": 24799 }, { "epoch": 18.168498168498168, "grad_norm": 1.171875, "learning_rate": 5.14695982489411e-05, "loss": 0.1308, "step": 24800 }, { "epoch": 18.16923076923077, "grad_norm": 1.25, "learning_rate": 5.146617254265074e-05, "loss": 0.1305, "step": 24801 }, { "epoch": 18.16996336996337, "grad_norm": 0.90625, "learning_rate": 5.146274683173111e-05, "loss": 0.1554, "step": 24802 }, { "epoch": 18.17069597069597, "grad_norm": 1.1953125, "learning_rate": 5.145932111619868e-05, "loss": 0.1185, "step": 24803 }, { "epoch": 18.17142857142857, "grad_norm": 1.4375, "learning_rate": 5.14558953960698e-05, "loss": 0.139, "step": 24804 }, { "epoch": 18.17216117216117, "grad_norm": 1.3359375, "learning_rate": 5.1452469671360914e-05, "loss": 0.1472, "step": 24805 }, { "epoch": 18.172893772893772, "grad_norm": 0.88671875, "learning_rate": 5.144904394208843e-05, "loss": 0.114, "step": 24806 }, { "epoch": 18.173626373626373, "grad_norm": 1.09375, "learning_rate": 5.1445618208268776e-05, "loss": 0.1658, "step": 24807 }, { "epoch": 18.174358974358974, "grad_norm": 1.671875, "learning_rate": 5.144219246991834e-05, "loss": 0.1245, "step": 24808 }, { "epoch": 18.175091575091574, "grad_norm": 1.234375, "learning_rate": 5.1438766727053556e-05, "loss": 0.1333, "step": 24809 }, { "epoch": 18.175824175824175, "grad_norm": 1.234375, "learning_rate": 5.143534097969084e-05, "loss": 0.1071, "step": 24810 }, { "epoch": 18.176556776556776, "grad_norm": 1.1015625, "learning_rate": 5.14319152278466e-05, "loss": 0.1293, "step": 24811 }, { "epoch": 18.177289377289377, "grad_norm": 0.640625, "learning_rate": 5.142848947153723e-05, "loss": 0.0989, "step": 24812 }, { "epoch": 18.178021978021977, "grad_norm": 1.0, "learning_rate": 5.142506371077919e-05, "loss": 0.0981, "step": 24813 }, { "epoch": 18.178754578754578, "grad_norm": 1.4453125, "learning_rate": 5.142163794558884e-05, "loss": 0.1214, "step": 24814 }, { "epoch": 18.17948717948718, "grad_norm": 0.99609375, "learning_rate": 5.1418212175982636e-05, "loss": 0.1278, "step": 24815 }, { "epoch": 18.18021978021978, "grad_norm": 0.58984375, "learning_rate": 5.141478640197698e-05, "loss": 0.114, "step": 24816 }, { "epoch": 18.18095238095238, "grad_norm": 1.46875, "learning_rate": 5.141136062358827e-05, "loss": 0.1127, "step": 24817 }, { "epoch": 18.18168498168498, "grad_norm": 2.09375, "learning_rate": 5.1407934840832936e-05, "loss": 0.127, "step": 24818 }, { "epoch": 18.18241758241758, "grad_norm": 0.95703125, "learning_rate": 5.140450905372738e-05, "loss": 0.097, "step": 24819 }, { "epoch": 18.183150183150182, "grad_norm": 1.359375, "learning_rate": 5.140108326228803e-05, "loss": 0.1203, "step": 24820 }, { "epoch": 18.183882783882783, "grad_norm": 1.5234375, "learning_rate": 5.1397657466531293e-05, "loss": 0.1522, "step": 24821 }, { "epoch": 18.184615384615384, "grad_norm": 0.51953125, "learning_rate": 5.139423166647358e-05, "loss": 0.1101, "step": 24822 }, { "epoch": 18.185347985347985, "grad_norm": 1.1015625, "learning_rate": 5.139080586213132e-05, "loss": 0.1193, "step": 24823 }, { "epoch": 18.186080586080585, "grad_norm": 0.65234375, "learning_rate": 5.138738005352093e-05, "loss": 0.1036, "step": 24824 }, { "epoch": 18.186813186813186, "grad_norm": 0.8515625, "learning_rate": 5.138395424065878e-05, "loss": 0.111, "step": 24825 }, { "epoch": 18.187545787545787, "grad_norm": 1.703125, "learning_rate": 5.1380528423561336e-05, "loss": 0.1399, "step": 24826 }, { "epoch": 18.188278388278388, "grad_norm": 0.671875, "learning_rate": 5.1377102602244984e-05, "loss": 0.1096, "step": 24827 }, { "epoch": 18.18901098901099, "grad_norm": 1.1015625, "learning_rate": 5.137367677672614e-05, "loss": 0.1327, "step": 24828 }, { "epoch": 18.18974358974359, "grad_norm": 1.1640625, "learning_rate": 5.137025094702124e-05, "loss": 0.1104, "step": 24829 }, { "epoch": 18.19047619047619, "grad_norm": 0.8359375, "learning_rate": 5.136682511314667e-05, "loss": 0.1386, "step": 24830 }, { "epoch": 18.19120879120879, "grad_norm": 0.73828125, "learning_rate": 5.1363399275118864e-05, "loss": 0.1054, "step": 24831 }, { "epoch": 18.19194139194139, "grad_norm": 1.0625, "learning_rate": 5.135997343295421e-05, "loss": 0.1305, "step": 24832 }, { "epoch": 18.192673992673992, "grad_norm": 1.09375, "learning_rate": 5.135654758666916e-05, "loss": 0.1196, "step": 24833 }, { "epoch": 18.193406593406593, "grad_norm": 0.98046875, "learning_rate": 5.1353121736280095e-05, "loss": 0.1093, "step": 24834 }, { "epoch": 18.194139194139193, "grad_norm": 1.5234375, "learning_rate": 5.134969588180345e-05, "loss": 0.1377, "step": 24835 }, { "epoch": 18.194871794871794, "grad_norm": 1.0078125, "learning_rate": 5.134627002325564e-05, "loss": 0.098, "step": 24836 }, { "epoch": 18.195604395604395, "grad_norm": 0.6953125, "learning_rate": 5.134284416065307e-05, "loss": 0.0904, "step": 24837 }, { "epoch": 18.196336996336996, "grad_norm": 1.015625, "learning_rate": 5.1339418294012144e-05, "loss": 0.1271, "step": 24838 }, { "epoch": 18.197069597069596, "grad_norm": 0.78515625, "learning_rate": 5.133599242334931e-05, "loss": 0.109, "step": 24839 }, { "epoch": 18.197802197802197, "grad_norm": 1.3359375, "learning_rate": 5.133256654868095e-05, "loss": 0.1441, "step": 24840 }, { "epoch": 18.198534798534798, "grad_norm": 1.0625, "learning_rate": 5.132914067002349e-05, "loss": 0.138, "step": 24841 }, { "epoch": 18.1992673992674, "grad_norm": 0.98828125, "learning_rate": 5.132571478739335e-05, "loss": 0.1562, "step": 24842 }, { "epoch": 18.2, "grad_norm": 0.63671875, "learning_rate": 5.1322288900806924e-05, "loss": 0.1093, "step": 24843 }, { "epoch": 18.2007326007326, "grad_norm": 0.98046875, "learning_rate": 5.131886301028067e-05, "loss": 0.1419, "step": 24844 }, { "epoch": 18.2014652014652, "grad_norm": 0.72265625, "learning_rate": 5.131543711583096e-05, "loss": 0.0955, "step": 24845 }, { "epoch": 18.2021978021978, "grad_norm": 0.984375, "learning_rate": 5.131201121747421e-05, "loss": 0.1109, "step": 24846 }, { "epoch": 18.202930402930402, "grad_norm": 1.0390625, "learning_rate": 5.1308585315226866e-05, "loss": 0.1148, "step": 24847 }, { "epoch": 18.203663003663003, "grad_norm": 0.58984375, "learning_rate": 5.1305159409105305e-05, "loss": 0.0932, "step": 24848 }, { "epoch": 18.204395604395604, "grad_norm": 1.265625, "learning_rate": 5.130173349912598e-05, "loss": 0.1022, "step": 24849 }, { "epoch": 18.205128205128204, "grad_norm": 0.93359375, "learning_rate": 5.1298307585305285e-05, "loss": 0.1162, "step": 24850 }, { "epoch": 18.205860805860805, "grad_norm": 0.7578125, "learning_rate": 5.129488166765962e-05, "loss": 0.1078, "step": 24851 }, { "epoch": 18.206593406593406, "grad_norm": 1.015625, "learning_rate": 5.1291455746205444e-05, "loss": 0.1291, "step": 24852 }, { "epoch": 18.207326007326007, "grad_norm": 1.078125, "learning_rate": 5.128802982095912e-05, "loss": 0.1138, "step": 24853 }, { "epoch": 18.208058608058607, "grad_norm": 1.1171875, "learning_rate": 5.1284603891937086e-05, "loss": 0.1382, "step": 24854 }, { "epoch": 18.208791208791208, "grad_norm": 1.453125, "learning_rate": 5.128117795915578e-05, "loss": 0.129, "step": 24855 }, { "epoch": 18.20952380952381, "grad_norm": 1.1875, "learning_rate": 5.1277752022631576e-05, "loss": 0.1663, "step": 24856 }, { "epoch": 18.21025641025641, "grad_norm": 0.90625, "learning_rate": 5.127432608238091e-05, "loss": 0.108, "step": 24857 }, { "epoch": 18.21098901098901, "grad_norm": 1.046875, "learning_rate": 5.12709001384202e-05, "loss": 0.1381, "step": 24858 }, { "epoch": 18.21172161172161, "grad_norm": 0.796875, "learning_rate": 5.126747419076583e-05, "loss": 0.1112, "step": 24859 }, { "epoch": 18.21245421245421, "grad_norm": 1.265625, "learning_rate": 5.126404823943426e-05, "loss": 0.1382, "step": 24860 }, { "epoch": 18.213186813186812, "grad_norm": 1.484375, "learning_rate": 5.126062228444187e-05, "loss": 0.1374, "step": 24861 }, { "epoch": 18.213919413919413, "grad_norm": 1.21875, "learning_rate": 5.12571963258051e-05, "loss": 0.1279, "step": 24862 }, { "epoch": 18.214652014652014, "grad_norm": 0.73828125, "learning_rate": 5.1253770363540366e-05, "loss": 0.1191, "step": 24863 }, { "epoch": 18.215384615384615, "grad_norm": 1.046875, "learning_rate": 5.125034439766405e-05, "loss": 0.1207, "step": 24864 }, { "epoch": 18.216117216117215, "grad_norm": 1.2265625, "learning_rate": 5.1246918428192605e-05, "loss": 0.1359, "step": 24865 }, { "epoch": 18.216849816849816, "grad_norm": 1.015625, "learning_rate": 5.124349245514242e-05, "loss": 0.1139, "step": 24866 }, { "epoch": 18.217582417582417, "grad_norm": 1.484375, "learning_rate": 5.1240066478529926e-05, "loss": 0.113, "step": 24867 }, { "epoch": 18.218315018315018, "grad_norm": 1.3125, "learning_rate": 5.123664049837151e-05, "loss": 0.1311, "step": 24868 }, { "epoch": 18.21904761904762, "grad_norm": 0.9375, "learning_rate": 5.123321451468361e-05, "loss": 0.1274, "step": 24869 }, { "epoch": 18.21978021978022, "grad_norm": 1.1015625, "learning_rate": 5.1229788527482655e-05, "loss": 0.1136, "step": 24870 }, { "epoch": 18.22051282051282, "grad_norm": 1.109375, "learning_rate": 5.1226362536785044e-05, "loss": 0.1167, "step": 24871 }, { "epoch": 18.22124542124542, "grad_norm": 0.91796875, "learning_rate": 5.122293654260718e-05, "loss": 0.1261, "step": 24872 }, { "epoch": 18.22197802197802, "grad_norm": 1.21875, "learning_rate": 5.121951054496549e-05, "loss": 0.1281, "step": 24873 }, { "epoch": 18.222710622710622, "grad_norm": 1.8359375, "learning_rate": 5.1216084543876395e-05, "loss": 0.1147, "step": 24874 }, { "epoch": 18.223443223443223, "grad_norm": 0.7265625, "learning_rate": 5.12126585393563e-05, "loss": 0.1143, "step": 24875 }, { "epoch": 18.224175824175823, "grad_norm": 0.8828125, "learning_rate": 5.120923253142164e-05, "loss": 0.1173, "step": 24876 }, { "epoch": 18.224908424908424, "grad_norm": 1.125, "learning_rate": 5.1205806520088794e-05, "loss": 0.096, "step": 24877 }, { "epoch": 18.225641025641025, "grad_norm": 0.87109375, "learning_rate": 5.120238050537421e-05, "loss": 0.1324, "step": 24878 }, { "epoch": 18.226373626373626, "grad_norm": 1.34375, "learning_rate": 5.119895448729428e-05, "loss": 0.1282, "step": 24879 }, { "epoch": 18.227106227106226, "grad_norm": 1.3203125, "learning_rate": 5.119552846586543e-05, "loss": 0.1132, "step": 24880 }, { "epoch": 18.227838827838827, "grad_norm": 1.2265625, "learning_rate": 5.119210244110408e-05, "loss": 0.1342, "step": 24881 }, { "epoch": 18.228571428571428, "grad_norm": 1.046875, "learning_rate": 5.1188676413026636e-05, "loss": 0.1228, "step": 24882 }, { "epoch": 18.22930402930403, "grad_norm": 1.3046875, "learning_rate": 5.118525038164953e-05, "loss": 0.1315, "step": 24883 }, { "epoch": 18.23003663003663, "grad_norm": 0.859375, "learning_rate": 5.118182434698915e-05, "loss": 0.1086, "step": 24884 }, { "epoch": 18.23076923076923, "grad_norm": 0.78515625, "learning_rate": 5.1178398309061925e-05, "loss": 0.1024, "step": 24885 }, { "epoch": 18.23150183150183, "grad_norm": 1.1484375, "learning_rate": 5.117497226788428e-05, "loss": 0.1527, "step": 24886 }, { "epoch": 18.23223443223443, "grad_norm": 1.15625, "learning_rate": 5.1171546223472596e-05, "loss": 0.142, "step": 24887 }, { "epoch": 18.232967032967032, "grad_norm": 1.0078125, "learning_rate": 5.1168120175843346e-05, "loss": 0.1194, "step": 24888 }, { "epoch": 18.233699633699633, "grad_norm": 1.0703125, "learning_rate": 5.1164694125012914e-05, "loss": 0.1081, "step": 24889 }, { "epoch": 18.234432234432234, "grad_norm": 1.0546875, "learning_rate": 5.116126807099769e-05, "loss": 0.0966, "step": 24890 }, { "epoch": 18.235164835164834, "grad_norm": 1.34375, "learning_rate": 5.1157842013814126e-05, "loss": 0.1098, "step": 24891 }, { "epoch": 18.235897435897435, "grad_norm": 1.421875, "learning_rate": 5.115441595347863e-05, "loss": 0.1128, "step": 24892 }, { "epoch": 18.236630036630036, "grad_norm": 1.046875, "learning_rate": 5.1150989890007606e-05, "loss": 0.1231, "step": 24893 }, { "epoch": 18.237362637362637, "grad_norm": 0.859375, "learning_rate": 5.114756382341749e-05, "loss": 0.1122, "step": 24894 }, { "epoch": 18.238095238095237, "grad_norm": 0.9765625, "learning_rate": 5.1144137753724663e-05, "loss": 0.1122, "step": 24895 }, { "epoch": 18.238827838827838, "grad_norm": 1.140625, "learning_rate": 5.114071168094556e-05, "loss": 0.1032, "step": 24896 }, { "epoch": 18.23956043956044, "grad_norm": 1.109375, "learning_rate": 5.113728560509661e-05, "loss": 0.1205, "step": 24897 }, { "epoch": 18.24029304029304, "grad_norm": 1.2578125, "learning_rate": 5.11338595261942e-05, "loss": 0.1105, "step": 24898 }, { "epoch": 18.24102564102564, "grad_norm": 1.4765625, "learning_rate": 5.113043344425478e-05, "loss": 0.137, "step": 24899 }, { "epoch": 18.24175824175824, "grad_norm": 1.3203125, "learning_rate": 5.112700735929472e-05, "loss": 0.1406, "step": 24900 }, { "epoch": 18.24249084249084, "grad_norm": 1.4296875, "learning_rate": 5.112358127133049e-05, "loss": 0.1132, "step": 24901 }, { "epoch": 18.243223443223442, "grad_norm": 1.03125, "learning_rate": 5.1120155180378474e-05, "loss": 0.0933, "step": 24902 }, { "epoch": 18.243956043956043, "grad_norm": 1.8984375, "learning_rate": 5.111672908645507e-05, "loss": 0.1396, "step": 24903 }, { "epoch": 18.244688644688644, "grad_norm": 1.7578125, "learning_rate": 5.111330298957674e-05, "loss": 0.1256, "step": 24904 }, { "epoch": 18.245421245421245, "grad_norm": 0.65625, "learning_rate": 5.110987688975984e-05, "loss": 0.1106, "step": 24905 }, { "epoch": 18.246153846153845, "grad_norm": 1.3359375, "learning_rate": 5.110645078702085e-05, "loss": 0.1114, "step": 24906 }, { "epoch": 18.246886446886446, "grad_norm": 1.875, "learning_rate": 5.110302468137617e-05, "loss": 0.1555, "step": 24907 }, { "epoch": 18.247619047619047, "grad_norm": 2.09375, "learning_rate": 5.1099598572842164e-05, "loss": 0.1466, "step": 24908 }, { "epoch": 18.248351648351647, "grad_norm": 1.703125, "learning_rate": 5.1096172461435297e-05, "loss": 0.0978, "step": 24909 }, { "epoch": 18.249084249084248, "grad_norm": 1.6484375, "learning_rate": 5.109274634717197e-05, "loss": 0.133, "step": 24910 }, { "epoch": 18.24981684981685, "grad_norm": 0.8125, "learning_rate": 5.10893202300686e-05, "loss": 0.1135, "step": 24911 }, { "epoch": 18.25054945054945, "grad_norm": 1.5390625, "learning_rate": 5.108589411014161e-05, "loss": 0.1556, "step": 24912 }, { "epoch": 18.25128205128205, "grad_norm": 1.359375, "learning_rate": 5.10824679874074e-05, "loss": 0.0975, "step": 24913 }, { "epoch": 18.25201465201465, "grad_norm": 0.9609375, "learning_rate": 5.1079041861882406e-05, "loss": 0.1137, "step": 24914 }, { "epoch": 18.252747252747252, "grad_norm": 1.3203125, "learning_rate": 5.1075615733583025e-05, "loss": 0.0893, "step": 24915 }, { "epoch": 18.253479853479853, "grad_norm": 2.125, "learning_rate": 5.107218960252568e-05, "loss": 0.1317, "step": 24916 }, { "epoch": 18.254212454212453, "grad_norm": 1.578125, "learning_rate": 5.1068763468726796e-05, "loss": 0.1349, "step": 24917 }, { "epoch": 18.254945054945054, "grad_norm": 0.7890625, "learning_rate": 5.1065337332202766e-05, "loss": 0.1256, "step": 24918 }, { "epoch": 18.255677655677655, "grad_norm": 1.296875, "learning_rate": 5.106191119297003e-05, "loss": 0.1327, "step": 24919 }, { "epoch": 18.256410256410255, "grad_norm": 1.2421875, "learning_rate": 5.1058485051045e-05, "loss": 0.1149, "step": 24920 }, { "epoch": 18.257142857142856, "grad_norm": 0.90625, "learning_rate": 5.1055058906444065e-05, "loss": 0.1421, "step": 24921 }, { "epoch": 18.257875457875457, "grad_norm": 0.8359375, "learning_rate": 5.105163275918366e-05, "loss": 0.1182, "step": 24922 }, { "epoch": 18.258608058608058, "grad_norm": 0.78515625, "learning_rate": 5.104820660928021e-05, "loss": 0.0848, "step": 24923 }, { "epoch": 18.25934065934066, "grad_norm": 0.9140625, "learning_rate": 5.104478045675012e-05, "loss": 0.1091, "step": 24924 }, { "epoch": 18.26007326007326, "grad_norm": 0.76953125, "learning_rate": 5.1041354301609814e-05, "loss": 0.122, "step": 24925 }, { "epoch": 18.26080586080586, "grad_norm": 1.078125, "learning_rate": 5.10379281438757e-05, "loss": 0.1164, "step": 24926 }, { "epoch": 18.26153846153846, "grad_norm": 1.3984375, "learning_rate": 5.10345019835642e-05, "loss": 0.1196, "step": 24927 }, { "epoch": 18.26227106227106, "grad_norm": 1.5546875, "learning_rate": 5.103107582069171e-05, "loss": 0.1407, "step": 24928 }, { "epoch": 18.263003663003662, "grad_norm": 1.5390625, "learning_rate": 5.1027649655274676e-05, "loss": 0.1352, "step": 24929 }, { "epoch": 18.263736263736263, "grad_norm": 1.21875, "learning_rate": 5.102422348732949e-05, "loss": 0.1417, "step": 24930 }, { "epoch": 18.264468864468864, "grad_norm": 1.1953125, "learning_rate": 5.102079731687258e-05, "loss": 0.1445, "step": 24931 }, { "epoch": 18.265201465201464, "grad_norm": 0.86328125, "learning_rate": 5.101737114392037e-05, "loss": 0.1191, "step": 24932 }, { "epoch": 18.265934065934065, "grad_norm": 1.03125, "learning_rate": 5.1013944968489256e-05, "loss": 0.1377, "step": 24933 }, { "epoch": 18.266666666666666, "grad_norm": 1.0078125, "learning_rate": 5.101051879059565e-05, "loss": 0.1282, "step": 24934 }, { "epoch": 18.267399267399266, "grad_norm": 1.1171875, "learning_rate": 5.100709261025599e-05, "loss": 0.1455, "step": 24935 }, { "epoch": 18.268131868131867, "grad_norm": 1.2734375, "learning_rate": 5.1003666427486695e-05, "loss": 0.1568, "step": 24936 }, { "epoch": 18.268864468864468, "grad_norm": 0.97265625, "learning_rate": 5.100024024230415e-05, "loss": 0.1157, "step": 24937 }, { "epoch": 18.26959706959707, "grad_norm": 1.203125, "learning_rate": 5.0996814054724804e-05, "loss": 0.1219, "step": 24938 }, { "epoch": 18.27032967032967, "grad_norm": 1.078125, "learning_rate": 5.099338786476505e-05, "loss": 0.1346, "step": 24939 }, { "epoch": 18.27106227106227, "grad_norm": 0.96484375, "learning_rate": 5.098996167244132e-05, "loss": 0.1296, "step": 24940 }, { "epoch": 18.27179487179487, "grad_norm": 1.3671875, "learning_rate": 5.0986535477770024e-05, "loss": 0.1133, "step": 24941 }, { "epoch": 18.27252747252747, "grad_norm": 1.0703125, "learning_rate": 5.098310928076757e-05, "loss": 0.1096, "step": 24942 }, { "epoch": 18.273260073260072, "grad_norm": 1.2421875, "learning_rate": 5.097968308145039e-05, "loss": 0.1143, "step": 24943 }, { "epoch": 18.273992673992673, "grad_norm": 1.1640625, "learning_rate": 5.097625687983487e-05, "loss": 0.1188, "step": 24944 }, { "epoch": 18.274725274725274, "grad_norm": 1.1640625, "learning_rate": 5.097283067593748e-05, "loss": 0.1331, "step": 24945 }, { "epoch": 18.275457875457874, "grad_norm": 1.0078125, "learning_rate": 5.0969404469774574e-05, "loss": 0.1372, "step": 24946 }, { "epoch": 18.276190476190475, "grad_norm": 0.8359375, "learning_rate": 5.09659782613626e-05, "loss": 0.1423, "step": 24947 }, { "epoch": 18.276923076923076, "grad_norm": 1.1015625, "learning_rate": 5.096255205071798e-05, "loss": 0.1171, "step": 24948 }, { "epoch": 18.277655677655677, "grad_norm": 1.3984375, "learning_rate": 5.095912583785713e-05, "loss": 0.1134, "step": 24949 }, { "epoch": 18.278388278388277, "grad_norm": 1.21875, "learning_rate": 5.095569962279644e-05, "loss": 0.1339, "step": 24950 }, { "epoch": 18.279120879120878, "grad_norm": 0.75, "learning_rate": 5.095227340555235e-05, "loss": 0.1037, "step": 24951 }, { "epoch": 18.27985347985348, "grad_norm": 1.3984375, "learning_rate": 5.094884718614127e-05, "loss": 0.1197, "step": 24952 }, { "epoch": 18.28058608058608, "grad_norm": 1.4453125, "learning_rate": 5.0945420964579616e-05, "loss": 0.1582, "step": 24953 }, { "epoch": 18.28131868131868, "grad_norm": 1.375, "learning_rate": 5.09419947408838e-05, "loss": 0.1181, "step": 24954 }, { "epoch": 18.28205128205128, "grad_norm": 1.1953125, "learning_rate": 5.093856851507026e-05, "loss": 0.1233, "step": 24955 }, { "epoch": 18.282783882783882, "grad_norm": 0.75, "learning_rate": 5.093514228715538e-05, "loss": 0.1238, "step": 24956 }, { "epoch": 18.283516483516483, "grad_norm": 1.03125, "learning_rate": 5.093171605715559e-05, "loss": 0.1185, "step": 24957 }, { "epoch": 18.284249084249083, "grad_norm": 1.0859375, "learning_rate": 5.092828982508731e-05, "loss": 0.1271, "step": 24958 }, { "epoch": 18.284981684981684, "grad_norm": 1.03125, "learning_rate": 5.0924863590966955e-05, "loss": 0.1229, "step": 24959 }, { "epoch": 18.285714285714285, "grad_norm": 1.4140625, "learning_rate": 5.092143735481093e-05, "loss": 0.1414, "step": 24960 }, { "epoch": 18.286446886446885, "grad_norm": 1.3203125, "learning_rate": 5.091801111663566e-05, "loss": 0.1192, "step": 24961 }, { "epoch": 18.287179487179486, "grad_norm": 0.7578125, "learning_rate": 5.091458487645756e-05, "loss": 0.0993, "step": 24962 }, { "epoch": 18.287912087912087, "grad_norm": 0.8984375, "learning_rate": 5.091115863429305e-05, "loss": 0.1075, "step": 24963 }, { "epoch": 18.288644688644688, "grad_norm": 1.859375, "learning_rate": 5.0907732390158556e-05, "loss": 0.1395, "step": 24964 }, { "epoch": 18.28937728937729, "grad_norm": 0.8203125, "learning_rate": 5.090430614407047e-05, "loss": 0.1065, "step": 24965 }, { "epoch": 18.29010989010989, "grad_norm": 1.3203125, "learning_rate": 5.090087989604523e-05, "loss": 0.1485, "step": 24966 }, { "epoch": 18.29084249084249, "grad_norm": 1.203125, "learning_rate": 5.0897453646099246e-05, "loss": 0.1571, "step": 24967 }, { "epoch": 18.29157509157509, "grad_norm": 1.484375, "learning_rate": 5.089402739424892e-05, "loss": 0.1399, "step": 24968 }, { "epoch": 18.29230769230769, "grad_norm": 1.328125, "learning_rate": 5.089060114051068e-05, "loss": 0.1366, "step": 24969 }, { "epoch": 18.293040293040292, "grad_norm": 1.8203125, "learning_rate": 5.088717488490094e-05, "loss": 0.1545, "step": 24970 }, { "epoch": 18.293772893772893, "grad_norm": 0.9921875, "learning_rate": 5.088374862743613e-05, "loss": 0.109, "step": 24971 }, { "epoch": 18.294505494505493, "grad_norm": 0.84765625, "learning_rate": 5.088032236813265e-05, "loss": 0.1593, "step": 24972 }, { "epoch": 18.295238095238094, "grad_norm": 1.109375, "learning_rate": 5.087689610700691e-05, "loss": 0.1222, "step": 24973 }, { "epoch": 18.295970695970695, "grad_norm": 1.2109375, "learning_rate": 5.087346984407534e-05, "loss": 0.1191, "step": 24974 }, { "epoch": 18.296703296703296, "grad_norm": 1.296875, "learning_rate": 5.087004357935435e-05, "loss": 0.1289, "step": 24975 }, { "epoch": 18.297435897435896, "grad_norm": 1.3671875, "learning_rate": 5.086661731286038e-05, "loss": 0.1078, "step": 24976 }, { "epoch": 18.298168498168497, "grad_norm": 1.3046875, "learning_rate": 5.086319104460982e-05, "loss": 0.139, "step": 24977 }, { "epoch": 18.298901098901098, "grad_norm": 0.8671875, "learning_rate": 5.085976477461907e-05, "loss": 0.1346, "step": 24978 }, { "epoch": 18.2996336996337, "grad_norm": 0.8125, "learning_rate": 5.085633850290459e-05, "loss": 0.1198, "step": 24979 }, { "epoch": 18.3003663003663, "grad_norm": 1.203125, "learning_rate": 5.0852912229482776e-05, "loss": 0.1218, "step": 24980 }, { "epoch": 18.3010989010989, "grad_norm": 1.015625, "learning_rate": 5.0849485954370034e-05, "loss": 0.1182, "step": 24981 }, { "epoch": 18.3018315018315, "grad_norm": 1.0703125, "learning_rate": 5.084605967758279e-05, "loss": 0.1225, "step": 24982 }, { "epoch": 18.3025641025641, "grad_norm": 0.8984375, "learning_rate": 5.084263339913745e-05, "loss": 0.1178, "step": 24983 }, { "epoch": 18.303296703296702, "grad_norm": 0.7265625, "learning_rate": 5.083920711905048e-05, "loss": 0.0997, "step": 24984 }, { "epoch": 18.304029304029303, "grad_norm": 1.046875, "learning_rate": 5.083578083733822e-05, "loss": 0.1111, "step": 24985 }, { "epoch": 18.304761904761904, "grad_norm": 0.86328125, "learning_rate": 5.083235455401714e-05, "loss": 0.1155, "step": 24986 }, { "epoch": 18.305494505494504, "grad_norm": 1.4921875, "learning_rate": 5.0828928269103635e-05, "loss": 0.149, "step": 24987 }, { "epoch": 18.306227106227105, "grad_norm": 1.390625, "learning_rate": 5.082550198261413e-05, "loss": 0.1406, "step": 24988 }, { "epoch": 18.306959706959706, "grad_norm": 0.90234375, "learning_rate": 5.082207569456502e-05, "loss": 0.1392, "step": 24989 }, { "epoch": 18.307692307692307, "grad_norm": 1.09375, "learning_rate": 5.081864940497276e-05, "loss": 0.1092, "step": 24990 }, { "epoch": 18.308424908424907, "grad_norm": 0.9765625, "learning_rate": 5.081522311385375e-05, "loss": 0.1371, "step": 24991 }, { "epoch": 18.309157509157508, "grad_norm": 0.83984375, "learning_rate": 5.081179682122439e-05, "loss": 0.1255, "step": 24992 }, { "epoch": 18.30989010989011, "grad_norm": 1.09375, "learning_rate": 5.0808370527101126e-05, "loss": 0.148, "step": 24993 }, { "epoch": 18.31062271062271, "grad_norm": 1.046875, "learning_rate": 5.080494423150033e-05, "loss": 0.1095, "step": 24994 }, { "epoch": 18.31135531135531, "grad_norm": 1.109375, "learning_rate": 5.080151793443847e-05, "loss": 0.1429, "step": 24995 }, { "epoch": 18.31208791208791, "grad_norm": 1.2265625, "learning_rate": 5.079809163593192e-05, "loss": 0.1399, "step": 24996 }, { "epoch": 18.31282051282051, "grad_norm": 0.55078125, "learning_rate": 5.079466533599714e-05, "loss": 0.1109, "step": 24997 }, { "epoch": 18.313553113553112, "grad_norm": 0.8359375, "learning_rate": 5.079123903465051e-05, "loss": 0.1213, "step": 24998 }, { "epoch": 18.314285714285713, "grad_norm": 0.79296875, "learning_rate": 5.0787812731908445e-05, "loss": 0.1068, "step": 24999 }, { "epoch": 18.315018315018314, "grad_norm": 1.0078125, "learning_rate": 5.078438642778738e-05, "loss": 0.137, "step": 25000 }, { "epoch": 18.315750915750915, "grad_norm": 1.53125, "learning_rate": 5.0780960122303735e-05, "loss": 0.1398, "step": 25001 }, { "epoch": 18.316483516483515, "grad_norm": 0.8203125, "learning_rate": 5.077753381547391e-05, "loss": 0.114, "step": 25002 }, { "epoch": 18.317216117216116, "grad_norm": 0.83984375, "learning_rate": 5.077410750731434e-05, "loss": 0.1433, "step": 25003 }, { "epoch": 18.317948717948717, "grad_norm": 1.3515625, "learning_rate": 5.077068119784141e-05, "loss": 0.1441, "step": 25004 }, { "epoch": 18.318681318681318, "grad_norm": 0.9296875, "learning_rate": 5.076725488707158e-05, "loss": 0.1426, "step": 25005 }, { "epoch": 18.31941391941392, "grad_norm": 0.9921875, "learning_rate": 5.0763828575021245e-05, "loss": 0.1165, "step": 25006 }, { "epoch": 18.32014652014652, "grad_norm": 0.87890625, "learning_rate": 5.0760402261706795e-05, "loss": 0.0834, "step": 25007 }, { "epoch": 18.32087912087912, "grad_norm": 0.87890625, "learning_rate": 5.075697594714469e-05, "loss": 0.136, "step": 25008 }, { "epoch": 18.32161172161172, "grad_norm": 1.71875, "learning_rate": 5.075354963135133e-05, "loss": 0.1253, "step": 25009 }, { "epoch": 18.32234432234432, "grad_norm": 0.66796875, "learning_rate": 5.075012331434313e-05, "loss": 0.0913, "step": 25010 }, { "epoch": 18.323076923076922, "grad_norm": 1.125, "learning_rate": 5.074669699613651e-05, "loss": 0.1121, "step": 25011 }, { "epoch": 18.323809523809523, "grad_norm": 1.4609375, "learning_rate": 5.0743270676747867e-05, "loss": 0.1561, "step": 25012 }, { "epoch": 18.324542124542123, "grad_norm": 1.0234375, "learning_rate": 5.073984435619364e-05, "loss": 0.1413, "step": 25013 }, { "epoch": 18.325274725274724, "grad_norm": 1.125, "learning_rate": 5.0736418034490253e-05, "loss": 0.135, "step": 25014 }, { "epoch": 18.326007326007325, "grad_norm": 1.0703125, "learning_rate": 5.0732991711654096e-05, "loss": 0.1356, "step": 25015 }, { "epoch": 18.326739926739926, "grad_norm": 0.921875, "learning_rate": 5.072956538770162e-05, "loss": 0.1325, "step": 25016 }, { "epoch": 18.327472527472526, "grad_norm": 0.859375, "learning_rate": 5.07261390626492e-05, "loss": 0.1426, "step": 25017 }, { "epoch": 18.328205128205127, "grad_norm": 1.1328125, "learning_rate": 5.0722712736513275e-05, "loss": 0.1091, "step": 25018 }, { "epoch": 18.328937728937728, "grad_norm": 1.1875, "learning_rate": 5.071928640931028e-05, "loss": 0.1355, "step": 25019 }, { "epoch": 18.32967032967033, "grad_norm": 1.2109375, "learning_rate": 5.0715860081056585e-05, "loss": 0.1083, "step": 25020 }, { "epoch": 18.33040293040293, "grad_norm": 0.72265625, "learning_rate": 5.071243375176866e-05, "loss": 0.1125, "step": 25021 }, { "epoch": 18.33113553113553, "grad_norm": 0.65234375, "learning_rate": 5.070900742146287e-05, "loss": 0.118, "step": 25022 }, { "epoch": 18.33186813186813, "grad_norm": 0.66015625, "learning_rate": 5.070558109015567e-05, "loss": 0.1214, "step": 25023 }, { "epoch": 18.33260073260073, "grad_norm": 0.90234375, "learning_rate": 5.070215475786346e-05, "loss": 0.1305, "step": 25024 }, { "epoch": 18.333333333333332, "grad_norm": 1.3125, "learning_rate": 5.069872842460266e-05, "loss": 0.1196, "step": 25025 }, { "epoch": 18.334065934065933, "grad_norm": 1.2265625, "learning_rate": 5.0695302090389686e-05, "loss": 0.1123, "step": 25026 }, { "epoch": 18.334798534798534, "grad_norm": 1.015625, "learning_rate": 5.069187575524097e-05, "loss": 0.1092, "step": 25027 }, { "epoch": 18.335531135531134, "grad_norm": 0.92578125, "learning_rate": 5.0688449419172877e-05, "loss": 0.1155, "step": 25028 }, { "epoch": 18.336263736263735, "grad_norm": 1.0078125, "learning_rate": 5.06850230822019e-05, "loss": 0.1126, "step": 25029 }, { "epoch": 18.336996336996336, "grad_norm": 1.171875, "learning_rate": 5.0681596744344394e-05, "loss": 0.1327, "step": 25030 }, { "epoch": 18.337728937728937, "grad_norm": 1.2421875, "learning_rate": 5.067817040561681e-05, "loss": 0.1187, "step": 25031 }, { "epoch": 18.338461538461537, "grad_norm": 0.6875, "learning_rate": 5.0674744066035554e-05, "loss": 0.1166, "step": 25032 }, { "epoch": 18.339194139194138, "grad_norm": 1.2421875, "learning_rate": 5.067131772561704e-05, "loss": 0.138, "step": 25033 }, { "epoch": 18.33992673992674, "grad_norm": 0.953125, "learning_rate": 5.066789138437769e-05, "loss": 0.1271, "step": 25034 }, { "epoch": 18.34065934065934, "grad_norm": 1.203125, "learning_rate": 5.066446504233391e-05, "loss": 0.0984, "step": 25035 }, { "epoch": 18.34139194139194, "grad_norm": 1.421875, "learning_rate": 5.0661038699502114e-05, "loss": 0.1218, "step": 25036 }, { "epoch": 18.34212454212454, "grad_norm": 0.9375, "learning_rate": 5.0657612355898755e-05, "loss": 0.1306, "step": 25037 }, { "epoch": 18.34285714285714, "grad_norm": 1.2421875, "learning_rate": 5.0654186011540196e-05, "loss": 0.1074, "step": 25038 }, { "epoch": 18.343589743589742, "grad_norm": 0.87109375, "learning_rate": 5.0650759666442895e-05, "loss": 0.1251, "step": 25039 }, { "epoch": 18.344322344322343, "grad_norm": 1.7265625, "learning_rate": 5.064733332062326e-05, "loss": 0.1185, "step": 25040 }, { "epoch": 18.345054945054944, "grad_norm": 1.2421875, "learning_rate": 5.064390697409769e-05, "loss": 0.1231, "step": 25041 }, { "epoch": 18.345787545787545, "grad_norm": 1.2734375, "learning_rate": 5.064048062688262e-05, "loss": 0.105, "step": 25042 }, { "epoch": 18.346520146520145, "grad_norm": 0.84765625, "learning_rate": 5.063705427899447e-05, "loss": 0.1397, "step": 25043 }, { "epoch": 18.347252747252746, "grad_norm": 1.1796875, "learning_rate": 5.0633627930449636e-05, "loss": 0.1069, "step": 25044 }, { "epoch": 18.347985347985347, "grad_norm": 0.95703125, "learning_rate": 5.063020158126456e-05, "loss": 0.1074, "step": 25045 }, { "epoch": 18.348717948717947, "grad_norm": 0.6484375, "learning_rate": 5.062677523145563e-05, "loss": 0.1006, "step": 25046 }, { "epoch": 18.349450549450548, "grad_norm": 1.2109375, "learning_rate": 5.062334888103929e-05, "loss": 0.1283, "step": 25047 }, { "epoch": 18.35018315018315, "grad_norm": 1.15625, "learning_rate": 5.061992253003194e-05, "loss": 0.1049, "step": 25048 }, { "epoch": 18.35091575091575, "grad_norm": 1.109375, "learning_rate": 5.061649617844999e-05, "loss": 0.1392, "step": 25049 }, { "epoch": 18.35164835164835, "grad_norm": 1.2578125, "learning_rate": 5.06130698263099e-05, "loss": 0.1217, "step": 25050 }, { "epoch": 18.35238095238095, "grad_norm": 0.88671875, "learning_rate": 5.060964347362802e-05, "loss": 0.1111, "step": 25051 }, { "epoch": 18.353113553113552, "grad_norm": 0.859375, "learning_rate": 5.060621712042081e-05, "loss": 0.1103, "step": 25052 }, { "epoch": 18.353846153846153, "grad_norm": 0.83984375, "learning_rate": 5.06027907667047e-05, "loss": 0.1351, "step": 25053 }, { "epoch": 18.354578754578753, "grad_norm": 0.69140625, "learning_rate": 5.0599364412496064e-05, "loss": 0.1335, "step": 25054 }, { "epoch": 18.355311355311354, "grad_norm": 1.65625, "learning_rate": 5.059593805781135e-05, "loss": 0.1384, "step": 25055 }, { "epoch": 18.356043956043955, "grad_norm": 1.7421875, "learning_rate": 5.0592511702666966e-05, "loss": 0.144, "step": 25056 }, { "epoch": 18.356776556776556, "grad_norm": 1.484375, "learning_rate": 5.0589085347079325e-05, "loss": 0.1226, "step": 25057 }, { "epoch": 18.357509157509156, "grad_norm": 0.65625, "learning_rate": 5.058565899106485e-05, "loss": 0.1109, "step": 25058 }, { "epoch": 18.358241758241757, "grad_norm": 0.98046875, "learning_rate": 5.058223263463995e-05, "loss": 0.1324, "step": 25059 }, { "epoch": 18.358974358974358, "grad_norm": 1.546875, "learning_rate": 5.0578806277821054e-05, "loss": 0.1442, "step": 25060 }, { "epoch": 18.35970695970696, "grad_norm": 0.82421875, "learning_rate": 5.057537992062455e-05, "loss": 0.1159, "step": 25061 }, { "epoch": 18.36043956043956, "grad_norm": 0.6953125, "learning_rate": 5.05719535630669e-05, "loss": 0.0811, "step": 25062 }, { "epoch": 18.36117216117216, "grad_norm": 1.09375, "learning_rate": 5.0568527205164495e-05, "loss": 0.104, "step": 25063 }, { "epoch": 18.36190476190476, "grad_norm": 2.390625, "learning_rate": 5.0565100846933744e-05, "loss": 0.1431, "step": 25064 }, { "epoch": 18.36263736263736, "grad_norm": 0.875, "learning_rate": 5.0561674488391076e-05, "loss": 0.0923, "step": 25065 }, { "epoch": 18.363369963369962, "grad_norm": 1.234375, "learning_rate": 5.055824812955291e-05, "loss": 0.1187, "step": 25066 }, { "epoch": 18.364102564102563, "grad_norm": 1.359375, "learning_rate": 5.0554821770435646e-05, "loss": 0.1326, "step": 25067 }, { "epoch": 18.364835164835164, "grad_norm": 0.921875, "learning_rate": 5.055139541105572e-05, "loss": 0.1297, "step": 25068 }, { "epoch": 18.365567765567764, "grad_norm": 1.421875, "learning_rate": 5.054796905142956e-05, "loss": 0.1234, "step": 25069 }, { "epoch": 18.366300366300365, "grad_norm": 1.0703125, "learning_rate": 5.054454269157354e-05, "loss": 0.1335, "step": 25070 }, { "epoch": 18.367032967032966, "grad_norm": 1.0625, "learning_rate": 5.054111633150411e-05, "loss": 0.1225, "step": 25071 }, { "epoch": 18.367765567765566, "grad_norm": 0.83203125, "learning_rate": 5.0537689971237677e-05, "loss": 0.1054, "step": 25072 }, { "epoch": 18.368498168498167, "grad_norm": 0.796875, "learning_rate": 5.053426361079067e-05, "loss": 0.0952, "step": 25073 }, { "epoch": 18.369230769230768, "grad_norm": 0.8984375, "learning_rate": 5.053083725017948e-05, "loss": 0.1081, "step": 25074 }, { "epoch": 18.36996336996337, "grad_norm": 1.125, "learning_rate": 5.0527410889420543e-05, "loss": 0.1074, "step": 25075 }, { "epoch": 18.37069597069597, "grad_norm": 0.98046875, "learning_rate": 5.052398452853025e-05, "loss": 0.1286, "step": 25076 }, { "epoch": 18.37142857142857, "grad_norm": 0.92578125, "learning_rate": 5.0520558167525076e-05, "loss": 0.1263, "step": 25077 }, { "epoch": 18.37216117216117, "grad_norm": 1.109375, "learning_rate": 5.051713180642138e-05, "loss": 0.1255, "step": 25078 }, { "epoch": 18.37289377289377, "grad_norm": 1.0234375, "learning_rate": 5.05137054452356e-05, "loss": 0.1069, "step": 25079 }, { "epoch": 18.373626373626372, "grad_norm": 0.80078125, "learning_rate": 5.0510279083984156e-05, "loss": 0.1046, "step": 25080 }, { "epoch": 18.374358974358973, "grad_norm": 1.2109375, "learning_rate": 5.0506852722683474e-05, "loss": 0.1316, "step": 25081 }, { "epoch": 18.375091575091574, "grad_norm": 0.91015625, "learning_rate": 5.050342636134994e-05, "loss": 0.1042, "step": 25082 }, { "epoch": 18.375824175824174, "grad_norm": 1.46875, "learning_rate": 5.05e-05, "loss": 0.1136, "step": 25083 }, { "epoch": 18.376556776556775, "grad_norm": 1.1171875, "learning_rate": 5.0496573638650055e-05, "loss": 0.1261, "step": 25084 }, { "epoch": 18.377289377289376, "grad_norm": 0.6953125, "learning_rate": 5.049314727731654e-05, "loss": 0.1136, "step": 25085 }, { "epoch": 18.378021978021977, "grad_norm": 1.3671875, "learning_rate": 5.048972091601584e-05, "loss": 0.1264, "step": 25086 }, { "epoch": 18.378754578754577, "grad_norm": 1.3828125, "learning_rate": 5.04862945547644e-05, "loss": 0.1304, "step": 25087 }, { "epoch": 18.379487179487178, "grad_norm": 1.2421875, "learning_rate": 5.048286819357862e-05, "loss": 0.163, "step": 25088 }, { "epoch": 18.38021978021978, "grad_norm": 0.86328125, "learning_rate": 5.0479441832474926e-05, "loss": 0.1302, "step": 25089 }, { "epoch": 18.38095238095238, "grad_norm": 1.4453125, "learning_rate": 5.0476015471469736e-05, "loss": 0.1212, "step": 25090 }, { "epoch": 18.38168498168498, "grad_norm": 0.7734375, "learning_rate": 5.047258911057947e-05, "loss": 0.1258, "step": 25091 }, { "epoch": 18.38241758241758, "grad_norm": 1.6171875, "learning_rate": 5.046916274982052e-05, "loss": 0.1037, "step": 25092 }, { "epoch": 18.383150183150182, "grad_norm": 0.8203125, "learning_rate": 5.046573638920935e-05, "loss": 0.1201, "step": 25093 }, { "epoch": 18.383882783882783, "grad_norm": 1.2265625, "learning_rate": 5.0462310028762325e-05, "loss": 0.125, "step": 25094 }, { "epoch": 18.384615384615383, "grad_norm": 2.296875, "learning_rate": 5.0458883668495904e-05, "loss": 0.1733, "step": 25095 }, { "epoch": 18.385347985347984, "grad_norm": 1.34375, "learning_rate": 5.045545730842648e-05, "loss": 0.1228, "step": 25096 }, { "epoch": 18.386080586080585, "grad_norm": 1.125, "learning_rate": 5.045203094857045e-05, "loss": 0.1162, "step": 25097 }, { "epoch": 18.386813186813185, "grad_norm": 1.28125, "learning_rate": 5.044860458894428e-05, "loss": 0.1255, "step": 25098 }, { "epoch": 18.387545787545786, "grad_norm": 2.28125, "learning_rate": 5.044517822956436e-05, "loss": 0.1471, "step": 25099 }, { "epoch": 18.388278388278387, "grad_norm": 1.9921875, "learning_rate": 5.044175187044711e-05, "loss": 0.1364, "step": 25100 }, { "epoch": 18.389010989010988, "grad_norm": 1.03125, "learning_rate": 5.043832551160893e-05, "loss": 0.1247, "step": 25101 }, { "epoch": 18.38974358974359, "grad_norm": 0.73046875, "learning_rate": 5.0434899153066265e-05, "loss": 0.1148, "step": 25102 }, { "epoch": 18.39047619047619, "grad_norm": 1.1484375, "learning_rate": 5.043147279483551e-05, "loss": 0.1409, "step": 25103 }, { "epoch": 18.39120879120879, "grad_norm": 1.484375, "learning_rate": 5.042804643693311e-05, "loss": 0.1459, "step": 25104 }, { "epoch": 18.39194139194139, "grad_norm": 0.88671875, "learning_rate": 5.042462007937544e-05, "loss": 0.1351, "step": 25105 }, { "epoch": 18.39267399267399, "grad_norm": 1.0234375, "learning_rate": 5.042119372217896e-05, "loss": 0.0989, "step": 25106 }, { "epoch": 18.393406593406592, "grad_norm": 1.125, "learning_rate": 5.0417767365360045e-05, "loss": 0.1362, "step": 25107 }, { "epoch": 18.394139194139193, "grad_norm": 1.53125, "learning_rate": 5.041434100893515e-05, "loss": 0.1278, "step": 25108 }, { "epoch": 18.394871794871793, "grad_norm": 1.1875, "learning_rate": 5.0410914652920684e-05, "loss": 0.1323, "step": 25109 }, { "epoch": 18.395604395604394, "grad_norm": 1.125, "learning_rate": 5.040748829733305e-05, "loss": 0.134, "step": 25110 }, { "epoch": 18.396336996336995, "grad_norm": 1.515625, "learning_rate": 5.040406194218866e-05, "loss": 0.1749, "step": 25111 }, { "epoch": 18.397069597069596, "grad_norm": 1.328125, "learning_rate": 5.040063558750393e-05, "loss": 0.1342, "step": 25112 }, { "epoch": 18.397802197802196, "grad_norm": 0.90234375, "learning_rate": 5.039720923329531e-05, "loss": 0.112, "step": 25113 }, { "epoch": 18.398534798534797, "grad_norm": 1.078125, "learning_rate": 5.03937828795792e-05, "loss": 0.1333, "step": 25114 }, { "epoch": 18.399267399267398, "grad_norm": 1.0078125, "learning_rate": 5.039035652637199e-05, "loss": 0.1265, "step": 25115 }, { "epoch": 18.4, "grad_norm": 1.078125, "learning_rate": 5.038693017369013e-05, "loss": 0.1239, "step": 25116 }, { "epoch": 18.4007326007326, "grad_norm": 0.54296875, "learning_rate": 5.0383503821550006e-05, "loss": 0.0852, "step": 25117 }, { "epoch": 18.4014652014652, "grad_norm": 1.4375, "learning_rate": 5.0380077469968065e-05, "loss": 0.1388, "step": 25118 }, { "epoch": 18.4021978021978, "grad_norm": 1.296875, "learning_rate": 5.0376651118960717e-05, "loss": 0.1384, "step": 25119 }, { "epoch": 18.4029304029304, "grad_norm": 1.0859375, "learning_rate": 5.037322476854438e-05, "loss": 0.1206, "step": 25120 }, { "epoch": 18.403663003663002, "grad_norm": 0.8125, "learning_rate": 5.036979841873546e-05, "loss": 0.1194, "step": 25121 }, { "epoch": 18.404395604395603, "grad_norm": 1.4609375, "learning_rate": 5.036637206955038e-05, "loss": 0.1478, "step": 25122 }, { "epoch": 18.405128205128204, "grad_norm": 1.2890625, "learning_rate": 5.036294572100554e-05, "loss": 0.1379, "step": 25123 }, { "epoch": 18.405860805860804, "grad_norm": 0.90625, "learning_rate": 5.035951937311738e-05, "loss": 0.1146, "step": 25124 }, { "epoch": 18.406593406593405, "grad_norm": 1.1328125, "learning_rate": 5.035609302590232e-05, "loss": 0.1042, "step": 25125 }, { "epoch": 18.407326007326006, "grad_norm": 0.8515625, "learning_rate": 5.035266667937676e-05, "loss": 0.1277, "step": 25126 }, { "epoch": 18.408058608058607, "grad_norm": 1.234375, "learning_rate": 5.0349240333557114e-05, "loss": 0.1295, "step": 25127 }, { "epoch": 18.408791208791207, "grad_norm": 0.90625, "learning_rate": 5.03458139884598e-05, "loss": 0.1095, "step": 25128 }, { "epoch": 18.409523809523808, "grad_norm": 0.8671875, "learning_rate": 5.034238764410125e-05, "loss": 0.0975, "step": 25129 }, { "epoch": 18.41025641025641, "grad_norm": 1.6328125, "learning_rate": 5.033896130049788e-05, "loss": 0.1448, "step": 25130 }, { "epoch": 18.41098901098901, "grad_norm": 0.58203125, "learning_rate": 5.0335534957666095e-05, "loss": 0.1027, "step": 25131 }, { "epoch": 18.41172161172161, "grad_norm": 0.9453125, "learning_rate": 5.0332108615622334e-05, "loss": 0.1345, "step": 25132 }, { "epoch": 18.41245421245421, "grad_norm": 0.8828125, "learning_rate": 5.0328682274382956e-05, "loss": 0.139, "step": 25133 }, { "epoch": 18.41318681318681, "grad_norm": 0.6640625, "learning_rate": 5.032525593396444e-05, "loss": 0.1028, "step": 25134 }, { "epoch": 18.413919413919412, "grad_norm": 0.921875, "learning_rate": 5.032182959438319e-05, "loss": 0.1423, "step": 25135 }, { "epoch": 18.414652014652013, "grad_norm": 0.7890625, "learning_rate": 5.031840325565561e-05, "loss": 0.1233, "step": 25136 }, { "epoch": 18.415384615384614, "grad_norm": 0.98046875, "learning_rate": 5.0314976917798114e-05, "loss": 0.1226, "step": 25137 }, { "epoch": 18.416117216117215, "grad_norm": 0.9140625, "learning_rate": 5.031155058082712e-05, "loss": 0.1155, "step": 25138 }, { "epoch": 18.416849816849815, "grad_norm": 0.88671875, "learning_rate": 5.0308124244759046e-05, "loss": 0.149, "step": 25139 }, { "epoch": 18.417582417582416, "grad_norm": 1.578125, "learning_rate": 5.030469790961032e-05, "loss": 0.1651, "step": 25140 }, { "epoch": 18.418315018315017, "grad_norm": 0.765625, "learning_rate": 5.030127157539735e-05, "loss": 0.1241, "step": 25141 }, { "epoch": 18.419047619047618, "grad_norm": 1.15625, "learning_rate": 5.029784524213655e-05, "loss": 0.1141, "step": 25142 }, { "epoch": 18.41978021978022, "grad_norm": 1.6796875, "learning_rate": 5.0294418909844343e-05, "loss": 0.1468, "step": 25143 }, { "epoch": 18.42051282051282, "grad_norm": 1.3046875, "learning_rate": 5.029099257853713e-05, "loss": 0.1567, "step": 25144 }, { "epoch": 18.42124542124542, "grad_norm": 1.421875, "learning_rate": 5.028756624823136e-05, "loss": 0.1248, "step": 25145 }, { "epoch": 18.42197802197802, "grad_norm": 0.640625, "learning_rate": 5.028413991894342e-05, "loss": 0.1367, "step": 25146 }, { "epoch": 18.42271062271062, "grad_norm": 1.2109375, "learning_rate": 5.0280713590689746e-05, "loss": 0.144, "step": 25147 }, { "epoch": 18.423443223443222, "grad_norm": 1.34375, "learning_rate": 5.027728726348674e-05, "loss": 0.1737, "step": 25148 }, { "epoch": 18.424175824175823, "grad_norm": 0.77734375, "learning_rate": 5.02738609373508e-05, "loss": 0.1175, "step": 25149 }, { "epoch": 18.424908424908423, "grad_norm": 0.91015625, "learning_rate": 5.027043461229839e-05, "loss": 0.1234, "step": 25150 }, { "epoch": 18.425641025641024, "grad_norm": 0.6953125, "learning_rate": 5.0267008288345906e-05, "loss": 0.1301, "step": 25151 }, { "epoch": 18.426373626373625, "grad_norm": 1.1171875, "learning_rate": 5.0263581965509755e-05, "loss": 0.1194, "step": 25152 }, { "epoch": 18.427106227106226, "grad_norm": 1.3359375, "learning_rate": 5.026015564380636e-05, "loss": 0.1069, "step": 25153 }, { "epoch": 18.427838827838826, "grad_norm": 0.828125, "learning_rate": 5.0256729323252135e-05, "loss": 0.1059, "step": 25154 }, { "epoch": 18.428571428571427, "grad_norm": 0.79296875, "learning_rate": 5.02533030038635e-05, "loss": 0.1425, "step": 25155 }, { "epoch": 18.429304029304028, "grad_norm": 1.140625, "learning_rate": 5.024987668565687e-05, "loss": 0.143, "step": 25156 }, { "epoch": 18.43003663003663, "grad_norm": 0.83984375, "learning_rate": 5.024645036864868e-05, "loss": 0.1081, "step": 25157 }, { "epoch": 18.43076923076923, "grad_norm": 0.8515625, "learning_rate": 5.024302405285532e-05, "loss": 0.1286, "step": 25158 }, { "epoch": 18.43150183150183, "grad_norm": 1.1328125, "learning_rate": 5.02395977382932e-05, "loss": 0.1291, "step": 25159 }, { "epoch": 18.43223443223443, "grad_norm": 1.203125, "learning_rate": 5.023617142497877e-05, "loss": 0.1349, "step": 25160 }, { "epoch": 18.43296703296703, "grad_norm": 1.265625, "learning_rate": 5.0232745112928426e-05, "loss": 0.1324, "step": 25161 }, { "epoch": 18.433699633699632, "grad_norm": 1.1328125, "learning_rate": 5.02293188021586e-05, "loss": 0.1074, "step": 25162 }, { "epoch": 18.434432234432233, "grad_norm": 1.3515625, "learning_rate": 5.022589249268569e-05, "loss": 0.1726, "step": 25163 }, { "epoch": 18.435164835164834, "grad_norm": 1.140625, "learning_rate": 5.022246618452608e-05, "loss": 0.1207, "step": 25164 }, { "epoch": 18.435897435897434, "grad_norm": 1.328125, "learning_rate": 5.021903987769627e-05, "loss": 0.1403, "step": 25165 }, { "epoch": 18.436630036630035, "grad_norm": 0.7578125, "learning_rate": 5.021561357221263e-05, "loss": 0.1079, "step": 25166 }, { "epoch": 18.437362637362636, "grad_norm": 1.1875, "learning_rate": 5.021218726809156e-05, "loss": 0.1436, "step": 25167 }, { "epoch": 18.438095238095237, "grad_norm": 0.9140625, "learning_rate": 5.020876096534951e-05, "loss": 0.1111, "step": 25168 }, { "epoch": 18.438827838827837, "grad_norm": 1.5703125, "learning_rate": 5.020533466400287e-05, "loss": 0.1449, "step": 25169 }, { "epoch": 18.439560439560438, "grad_norm": 1.8671875, "learning_rate": 5.0201908364068076e-05, "loss": 0.1322, "step": 25170 }, { "epoch": 18.44029304029304, "grad_norm": 1.234375, "learning_rate": 5.019848206556154e-05, "loss": 0.1101, "step": 25171 }, { "epoch": 18.44102564102564, "grad_norm": 1.125, "learning_rate": 5.019505576849968e-05, "loss": 0.0937, "step": 25172 }, { "epoch": 18.44175824175824, "grad_norm": 1.4453125, "learning_rate": 5.0191629472898896e-05, "loss": 0.1114, "step": 25173 }, { "epoch": 18.44249084249084, "grad_norm": 0.97265625, "learning_rate": 5.018820317877563e-05, "loss": 0.1144, "step": 25174 }, { "epoch": 18.44322344322344, "grad_norm": 0.8046875, "learning_rate": 5.0184776886146254e-05, "loss": 0.1048, "step": 25175 }, { "epoch": 18.443956043956042, "grad_norm": 0.91796875, "learning_rate": 5.0181350595027246e-05, "loss": 0.1355, "step": 25176 }, { "epoch": 18.444688644688643, "grad_norm": 1.796875, "learning_rate": 5.017792430543498e-05, "loss": 0.1488, "step": 25177 }, { "epoch": 18.445421245421244, "grad_norm": 1.59375, "learning_rate": 5.0174498017385887e-05, "loss": 0.119, "step": 25178 }, { "epoch": 18.446153846153845, "grad_norm": 0.9609375, "learning_rate": 5.017107173089638e-05, "loss": 0.1179, "step": 25179 }, { "epoch": 18.446886446886445, "grad_norm": 0.65234375, "learning_rate": 5.016764544598287e-05, "loss": 0.0917, "step": 25180 }, { "epoch": 18.447619047619046, "grad_norm": 1.0, "learning_rate": 5.016421916266178e-05, "loss": 0.1078, "step": 25181 }, { "epoch": 18.448351648351647, "grad_norm": 0.92578125, "learning_rate": 5.0160792880949536e-05, "loss": 0.1187, "step": 25182 }, { "epoch": 18.449084249084247, "grad_norm": 1.2421875, "learning_rate": 5.015736660086254e-05, "loss": 0.1391, "step": 25183 }, { "epoch": 18.449816849816848, "grad_norm": 1.109375, "learning_rate": 5.015394032241723e-05, "loss": 0.0924, "step": 25184 }, { "epoch": 18.45054945054945, "grad_norm": 0.6328125, "learning_rate": 5.015051404562997e-05, "loss": 0.1369, "step": 25185 }, { "epoch": 18.45128205128205, "grad_norm": 1.1875, "learning_rate": 5.014708777051723e-05, "loss": 0.1567, "step": 25186 }, { "epoch": 18.45201465201465, "grad_norm": 1.171875, "learning_rate": 5.014366149709542e-05, "loss": 0.1238, "step": 25187 }, { "epoch": 18.45274725274725, "grad_norm": 0.81640625, "learning_rate": 5.014023522538094e-05, "loss": 0.1335, "step": 25188 }, { "epoch": 18.453479853479852, "grad_norm": 1.0, "learning_rate": 5.013680895539021e-05, "loss": 0.1046, "step": 25189 }, { "epoch": 18.454212454212453, "grad_norm": 0.77734375, "learning_rate": 5.0133382687139616e-05, "loss": 0.1148, "step": 25190 }, { "epoch": 18.454945054945053, "grad_norm": 0.9609375, "learning_rate": 5.012995642064565e-05, "loss": 0.1413, "step": 25191 }, { "epoch": 18.455677655677654, "grad_norm": 1.0078125, "learning_rate": 5.012653015592467e-05, "loss": 0.1161, "step": 25192 }, { "epoch": 18.456410256410255, "grad_norm": 1.4140625, "learning_rate": 5.012310389299309e-05, "loss": 0.132, "step": 25193 }, { "epoch": 18.457142857142856, "grad_norm": 0.9140625, "learning_rate": 5.0119677631867366e-05, "loss": 0.1337, "step": 25194 }, { "epoch": 18.457875457875456, "grad_norm": 1.375, "learning_rate": 5.011625137256387e-05, "loss": 0.131, "step": 25195 }, { "epoch": 18.458608058608057, "grad_norm": 1.3359375, "learning_rate": 5.011282511509905e-05, "loss": 0.1113, "step": 25196 }, { "epoch": 18.459340659340658, "grad_norm": 1.2890625, "learning_rate": 5.010939885948932e-05, "loss": 0.1383, "step": 25197 }, { "epoch": 18.46007326007326, "grad_norm": 0.8984375, "learning_rate": 5.0105972605751094e-05, "loss": 0.1178, "step": 25198 }, { "epoch": 18.46080586080586, "grad_norm": 1.0703125, "learning_rate": 5.010254635390077e-05, "loss": 0.1426, "step": 25199 }, { "epoch": 18.46153846153846, "grad_norm": 1.3359375, "learning_rate": 5.0099120103954786e-05, "loss": 0.1107, "step": 25200 }, { "epoch": 18.46227106227106, "grad_norm": 1.484375, "learning_rate": 5.009569385592953e-05, "loss": 0.12, "step": 25201 }, { "epoch": 18.46300366300366, "grad_norm": 1.7578125, "learning_rate": 5.0092267609841446e-05, "loss": 0.1155, "step": 25202 }, { "epoch": 18.463736263736262, "grad_norm": 1.21875, "learning_rate": 5.008884136570695e-05, "loss": 0.1398, "step": 25203 }, { "epoch": 18.464468864468863, "grad_norm": 0.5859375, "learning_rate": 5.0085415123542447e-05, "loss": 0.0969, "step": 25204 }, { "epoch": 18.465201465201464, "grad_norm": 0.8671875, "learning_rate": 5.008198888336435e-05, "loss": 0.1238, "step": 25205 }, { "epoch": 18.465934065934064, "grad_norm": 1.21875, "learning_rate": 5.0078562645189077e-05, "loss": 0.1238, "step": 25206 }, { "epoch": 18.466666666666665, "grad_norm": 0.8125, "learning_rate": 5.007513640903305e-05, "loss": 0.1193, "step": 25207 }, { "epoch": 18.467399267399266, "grad_norm": 1.2421875, "learning_rate": 5.00717101749127e-05, "loss": 0.1422, "step": 25208 }, { "epoch": 18.468131868131866, "grad_norm": 1.3359375, "learning_rate": 5.006828394284442e-05, "loss": 0.114, "step": 25209 }, { "epoch": 18.468864468864467, "grad_norm": 1.046875, "learning_rate": 5.006485771284464e-05, "loss": 0.1515, "step": 25210 }, { "epoch": 18.469597069597068, "grad_norm": 1.0390625, "learning_rate": 5.006143148492974e-05, "loss": 0.0962, "step": 25211 }, { "epoch": 18.47032967032967, "grad_norm": 1.2265625, "learning_rate": 5.005800525911619e-05, "loss": 0.1002, "step": 25212 }, { "epoch": 18.47106227106227, "grad_norm": 0.7734375, "learning_rate": 5.005457903542039e-05, "loss": 0.1113, "step": 25213 }, { "epoch": 18.47179487179487, "grad_norm": 1.140625, "learning_rate": 5.0051152813858734e-05, "loss": 0.1227, "step": 25214 }, { "epoch": 18.47252747252747, "grad_norm": 1.671875, "learning_rate": 5.0047726594447664e-05, "loss": 0.129, "step": 25215 }, { "epoch": 18.47326007326007, "grad_norm": 1.375, "learning_rate": 5.0044300377203556e-05, "loss": 0.1273, "step": 25216 }, { "epoch": 18.473992673992672, "grad_norm": 1.1171875, "learning_rate": 5.004087416214288e-05, "loss": 0.1361, "step": 25217 }, { "epoch": 18.474725274725273, "grad_norm": 0.91796875, "learning_rate": 5.0037447949282026e-05, "loss": 0.123, "step": 25218 }, { "epoch": 18.475457875457874, "grad_norm": 1.1953125, "learning_rate": 5.0034021738637395e-05, "loss": 0.1224, "step": 25219 }, { "epoch": 18.476190476190474, "grad_norm": 1.8984375, "learning_rate": 5.0030595530225435e-05, "loss": 0.1312, "step": 25220 }, { "epoch": 18.476923076923075, "grad_norm": 0.8046875, "learning_rate": 5.0027169324062516e-05, "loss": 0.1195, "step": 25221 }, { "epoch": 18.477655677655676, "grad_norm": 1.2265625, "learning_rate": 5.002374312016512e-05, "loss": 0.1126, "step": 25222 }, { "epoch": 18.478388278388277, "grad_norm": 0.62109375, "learning_rate": 5.0020316918549616e-05, "loss": 0.1252, "step": 25223 }, { "epoch": 18.479120879120877, "grad_norm": 1.1171875, "learning_rate": 5.001689071923243e-05, "loss": 0.1117, "step": 25224 }, { "epoch": 18.479853479853478, "grad_norm": 1.0078125, "learning_rate": 5.001346452222999e-05, "loss": 0.0876, "step": 25225 }, { "epoch": 18.48058608058608, "grad_norm": 1.1015625, "learning_rate": 5.0010038327558694e-05, "loss": 0.1248, "step": 25226 }, { "epoch": 18.48131868131868, "grad_norm": 0.875, "learning_rate": 5.0006612135234945e-05, "loss": 0.109, "step": 25227 }, { "epoch": 18.48205128205128, "grad_norm": 0.73046875, "learning_rate": 5.00031859452752e-05, "loss": 0.1128, "step": 25228 }, { "epoch": 18.48278388278388, "grad_norm": 0.875, "learning_rate": 4.999975975769586e-05, "loss": 0.1197, "step": 25229 }, { "epoch": 18.483516483516482, "grad_norm": 0.8125, "learning_rate": 4.9996333572513314e-05, "loss": 0.1247, "step": 25230 }, { "epoch": 18.484249084249083, "grad_norm": 1.4921875, "learning_rate": 4.999290738974402e-05, "loss": 0.1471, "step": 25231 }, { "epoch": 18.484981684981683, "grad_norm": 1.765625, "learning_rate": 4.9989481209404354e-05, "loss": 0.1163, "step": 25232 }, { "epoch": 18.485714285714284, "grad_norm": 1.328125, "learning_rate": 4.998605503151076e-05, "loss": 0.1178, "step": 25233 }, { "epoch": 18.486446886446885, "grad_norm": 0.78125, "learning_rate": 4.998262885607965e-05, "loss": 0.1192, "step": 25234 }, { "epoch": 18.487179487179485, "grad_norm": 0.98828125, "learning_rate": 4.997920268312743e-05, "loss": 0.1268, "step": 25235 }, { "epoch": 18.487912087912086, "grad_norm": 0.91015625, "learning_rate": 4.9975776512670524e-05, "loss": 0.1662, "step": 25236 }, { "epoch": 18.488644688644687, "grad_norm": 0.96484375, "learning_rate": 4.9972350344725326e-05, "loss": 0.1192, "step": 25237 }, { "epoch": 18.489377289377288, "grad_norm": 1.0390625, "learning_rate": 4.996892417930829e-05, "loss": 0.1139, "step": 25238 }, { "epoch": 18.49010989010989, "grad_norm": 0.9140625, "learning_rate": 4.9965498016435816e-05, "loss": 0.1322, "step": 25239 }, { "epoch": 18.49084249084249, "grad_norm": 1.0625, "learning_rate": 4.996207185612432e-05, "loss": 0.141, "step": 25240 }, { "epoch": 18.49157509157509, "grad_norm": 1.0234375, "learning_rate": 4.995864569839021e-05, "loss": 0.143, "step": 25241 }, { "epoch": 18.49230769230769, "grad_norm": 1.7890625, "learning_rate": 4.9955219543249884e-05, "loss": 0.1677, "step": 25242 }, { "epoch": 18.493040293040295, "grad_norm": 0.85546875, "learning_rate": 4.99517933907198e-05, "loss": 0.127, "step": 25243 }, { "epoch": 18.493772893772892, "grad_norm": 1.078125, "learning_rate": 4.994836724081634e-05, "loss": 0.1239, "step": 25244 }, { "epoch": 18.494505494505496, "grad_norm": 1.21875, "learning_rate": 4.994494109355595e-05, "loss": 0.143, "step": 25245 }, { "epoch": 18.495238095238093, "grad_norm": 1.140625, "learning_rate": 4.9941514948955034e-05, "loss": 0.1285, "step": 25246 }, { "epoch": 18.495970695970698, "grad_norm": 0.7734375, "learning_rate": 4.993808880702997e-05, "loss": 0.1182, "step": 25247 }, { "epoch": 18.496703296703295, "grad_norm": 1.2109375, "learning_rate": 4.9934662667797236e-05, "loss": 0.1142, "step": 25248 }, { "epoch": 18.4974358974359, "grad_norm": 1.2578125, "learning_rate": 4.993123653127322e-05, "loss": 0.1504, "step": 25249 }, { "epoch": 18.498168498168496, "grad_norm": 0.80078125, "learning_rate": 4.992781039747433e-05, "loss": 0.105, "step": 25250 }, { "epoch": 18.4989010989011, "grad_norm": 0.91015625, "learning_rate": 4.9924384266417e-05, "loss": 0.1149, "step": 25251 }, { "epoch": 18.499633699633698, "grad_norm": 1.3046875, "learning_rate": 4.9920958138117616e-05, "loss": 0.1711, "step": 25252 }, { "epoch": 18.500366300366302, "grad_norm": 1.0234375, "learning_rate": 4.991753201259261e-05, "loss": 0.0973, "step": 25253 }, { "epoch": 18.5010989010989, "grad_norm": 0.7578125, "learning_rate": 4.9914105889858395e-05, "loss": 0.1247, "step": 25254 }, { "epoch": 18.501831501831504, "grad_norm": 1.4609375, "learning_rate": 4.991067976993141e-05, "loss": 0.1261, "step": 25255 }, { "epoch": 18.5025641025641, "grad_norm": 1.3203125, "learning_rate": 4.990725365282804e-05, "loss": 0.1239, "step": 25256 }, { "epoch": 18.503296703296705, "grad_norm": 1.109375, "learning_rate": 4.9903827538564726e-05, "loss": 0.1196, "step": 25257 }, { "epoch": 18.504029304029302, "grad_norm": 1.25, "learning_rate": 4.990040142715785e-05, "loss": 0.1338, "step": 25258 }, { "epoch": 18.504761904761907, "grad_norm": 0.97265625, "learning_rate": 4.989697531862385e-05, "loss": 0.1118, "step": 25259 }, { "epoch": 18.505494505494504, "grad_norm": 0.78515625, "learning_rate": 4.989354921297915e-05, "loss": 0.1363, "step": 25260 }, { "epoch": 18.506227106227108, "grad_norm": 1.3125, "learning_rate": 4.989012311024016e-05, "loss": 0.1668, "step": 25261 }, { "epoch": 18.506959706959705, "grad_norm": 0.87890625, "learning_rate": 4.9886697010423286e-05, "loss": 0.1209, "step": 25262 }, { "epoch": 18.50769230769231, "grad_norm": 1.1328125, "learning_rate": 4.9883270913544924e-05, "loss": 0.1099, "step": 25263 }, { "epoch": 18.508424908424907, "grad_norm": 0.9375, "learning_rate": 4.987984481962154e-05, "loss": 0.1425, "step": 25264 }, { "epoch": 18.50915750915751, "grad_norm": 0.94140625, "learning_rate": 4.9876418728669524e-05, "loss": 0.1331, "step": 25265 }, { "epoch": 18.509890109890108, "grad_norm": 1.1171875, "learning_rate": 4.987299264070529e-05, "loss": 0.1022, "step": 25266 }, { "epoch": 18.510622710622712, "grad_norm": 1.171875, "learning_rate": 4.986956655574525e-05, "loss": 0.1253, "step": 25267 }, { "epoch": 18.51135531135531, "grad_norm": 1.0625, "learning_rate": 4.98661404738058e-05, "loss": 0.1299, "step": 25268 }, { "epoch": 18.512087912087914, "grad_norm": 0.953125, "learning_rate": 4.986271439490341e-05, "loss": 0.1081, "step": 25269 }, { "epoch": 18.51282051282051, "grad_norm": 1.1640625, "learning_rate": 4.985928831905444e-05, "loss": 0.1233, "step": 25270 }, { "epoch": 18.513553113553115, "grad_norm": 1.5234375, "learning_rate": 4.985586224627535e-05, "loss": 0.1346, "step": 25271 }, { "epoch": 18.514285714285712, "grad_norm": 2.015625, "learning_rate": 4.985243617658254e-05, "loss": 0.1291, "step": 25272 }, { "epoch": 18.515018315018317, "grad_norm": 1.1796875, "learning_rate": 4.9849010109992396e-05, "loss": 0.1185, "step": 25273 }, { "epoch": 18.515750915750917, "grad_norm": 1.5859375, "learning_rate": 4.984558404652137e-05, "loss": 0.1372, "step": 25274 }, { "epoch": 18.516483516483518, "grad_norm": 1.1953125, "learning_rate": 4.9842157986185876e-05, "loss": 0.1162, "step": 25275 }, { "epoch": 18.51721611721612, "grad_norm": 0.80078125, "learning_rate": 4.983873192900232e-05, "loss": 0.0816, "step": 25276 }, { "epoch": 18.51794871794872, "grad_norm": 1.1015625, "learning_rate": 4.983530587498711e-05, "loss": 0.1099, "step": 25277 }, { "epoch": 18.51868131868132, "grad_norm": 0.78125, "learning_rate": 4.983187982415667e-05, "loss": 0.1062, "step": 25278 }, { "epoch": 18.51941391941392, "grad_norm": 0.91015625, "learning_rate": 4.982845377652739e-05, "loss": 0.1202, "step": 25279 }, { "epoch": 18.520146520146522, "grad_norm": 1.2109375, "learning_rate": 4.9825027732115735e-05, "loss": 0.1174, "step": 25280 }, { "epoch": 18.520879120879123, "grad_norm": 0.80078125, "learning_rate": 4.98216016909381e-05, "loss": 0.1293, "step": 25281 }, { "epoch": 18.521611721611723, "grad_norm": 0.9453125, "learning_rate": 4.9818175653010864e-05, "loss": 0.0939, "step": 25282 }, { "epoch": 18.522344322344324, "grad_norm": 1.4765625, "learning_rate": 4.98147496183505e-05, "loss": 0.1294, "step": 25283 }, { "epoch": 18.523076923076925, "grad_norm": 1.0625, "learning_rate": 4.981132358697337e-05, "loss": 0.1161, "step": 25284 }, { "epoch": 18.523809523809526, "grad_norm": 1.1171875, "learning_rate": 4.980789755889592e-05, "loss": 0.1366, "step": 25285 }, { "epoch": 18.524542124542126, "grad_norm": 1.7265625, "learning_rate": 4.980447153413458e-05, "loss": 0.1471, "step": 25286 }, { "epoch": 18.525274725274727, "grad_norm": 1.40625, "learning_rate": 4.980104551270574e-05, "loss": 0.1286, "step": 25287 }, { "epoch": 18.526007326007328, "grad_norm": 1.390625, "learning_rate": 4.979761949462582e-05, "loss": 0.1011, "step": 25288 }, { "epoch": 18.52673992673993, "grad_norm": 1.125, "learning_rate": 4.979419347991121e-05, "loss": 0.1179, "step": 25289 }, { "epoch": 18.52747252747253, "grad_norm": 0.98046875, "learning_rate": 4.979076746857837e-05, "loss": 0.1332, "step": 25290 }, { "epoch": 18.52820512820513, "grad_norm": 1.390625, "learning_rate": 4.9787341460643705e-05, "loss": 0.1458, "step": 25291 }, { "epoch": 18.52893772893773, "grad_norm": 1.375, "learning_rate": 4.978391545612362e-05, "loss": 0.1164, "step": 25292 }, { "epoch": 18.52967032967033, "grad_norm": 1.2109375, "learning_rate": 4.9780489455034514e-05, "loss": 0.1252, "step": 25293 }, { "epoch": 18.530402930402932, "grad_norm": 0.67578125, "learning_rate": 4.9777063457392825e-05, "loss": 0.0937, "step": 25294 }, { "epoch": 18.531135531135533, "grad_norm": 1.28125, "learning_rate": 4.977363746321497e-05, "loss": 0.0975, "step": 25295 }, { "epoch": 18.531868131868134, "grad_norm": 0.59765625, "learning_rate": 4.977021147251735e-05, "loss": 0.1042, "step": 25296 }, { "epoch": 18.532600732600734, "grad_norm": 1.1328125, "learning_rate": 4.976678548531639e-05, "loss": 0.1306, "step": 25297 }, { "epoch": 18.533333333333335, "grad_norm": 0.87109375, "learning_rate": 4.97633595016285e-05, "loss": 0.1008, "step": 25298 }, { "epoch": 18.534065934065936, "grad_norm": 1.0234375, "learning_rate": 4.975993352147009e-05, "loss": 0.1159, "step": 25299 }, { "epoch": 18.534798534798536, "grad_norm": 1.1171875, "learning_rate": 4.975650754485759e-05, "loss": 0.1452, "step": 25300 }, { "epoch": 18.535531135531137, "grad_norm": 1.4140625, "learning_rate": 4.975308157180741e-05, "loss": 0.1591, "step": 25301 }, { "epoch": 18.536263736263738, "grad_norm": 0.93359375, "learning_rate": 4.9749655602335957e-05, "loss": 0.1056, "step": 25302 }, { "epoch": 18.53699633699634, "grad_norm": 0.9140625, "learning_rate": 4.9746229636459656e-05, "loss": 0.1112, "step": 25303 }, { "epoch": 18.53772893772894, "grad_norm": 2.125, "learning_rate": 4.974280367419491e-05, "loss": 0.142, "step": 25304 }, { "epoch": 18.53846153846154, "grad_norm": 1.140625, "learning_rate": 4.9739377715558123e-05, "loss": 0.1541, "step": 25305 }, { "epoch": 18.53919413919414, "grad_norm": 0.8515625, "learning_rate": 4.973595176056575e-05, "loss": 0.1056, "step": 25306 }, { "epoch": 18.53992673992674, "grad_norm": 0.7734375, "learning_rate": 4.9732525809234184e-05, "loss": 0.1455, "step": 25307 }, { "epoch": 18.540659340659342, "grad_norm": 0.98828125, "learning_rate": 4.9729099861579825e-05, "loss": 0.1304, "step": 25308 }, { "epoch": 18.541391941391943, "grad_norm": 1.0, "learning_rate": 4.972567391761911e-05, "loss": 0.1124, "step": 25309 }, { "epoch": 18.542124542124544, "grad_norm": 1.140625, "learning_rate": 4.972224797736843e-05, "loss": 0.1236, "step": 25310 }, { "epoch": 18.542857142857144, "grad_norm": 1.0703125, "learning_rate": 4.971882204084423e-05, "loss": 0.1364, "step": 25311 }, { "epoch": 18.543589743589745, "grad_norm": 0.59765625, "learning_rate": 4.9715396108062916e-05, "loss": 0.1148, "step": 25312 }, { "epoch": 18.544322344322346, "grad_norm": 1.8125, "learning_rate": 4.97119701790409e-05, "loss": 0.1599, "step": 25313 }, { "epoch": 18.545054945054947, "grad_norm": 1.2890625, "learning_rate": 4.9708544253794585e-05, "loss": 0.1499, "step": 25314 }, { "epoch": 18.545787545787547, "grad_norm": 1.0859375, "learning_rate": 4.970511833234037e-05, "loss": 0.1318, "step": 25315 }, { "epoch": 18.546520146520148, "grad_norm": 1.1484375, "learning_rate": 4.9701692414694724e-05, "loss": 0.1506, "step": 25316 }, { "epoch": 18.54725274725275, "grad_norm": 1.015625, "learning_rate": 4.969826650087403e-05, "loss": 0.1004, "step": 25317 }, { "epoch": 18.54798534798535, "grad_norm": 0.7265625, "learning_rate": 4.9694840590894704e-05, "loss": 0.1516, "step": 25318 }, { "epoch": 18.54871794871795, "grad_norm": 0.87109375, "learning_rate": 4.969141468477315e-05, "loss": 0.1045, "step": 25319 }, { "epoch": 18.54945054945055, "grad_norm": 0.6328125, "learning_rate": 4.968798878252579e-05, "loss": 0.1127, "step": 25320 }, { "epoch": 18.550183150183152, "grad_norm": 1.0859375, "learning_rate": 4.968456288416906e-05, "loss": 0.1182, "step": 25321 }, { "epoch": 18.550915750915753, "grad_norm": 0.98046875, "learning_rate": 4.9681136989719347e-05, "loss": 0.1386, "step": 25322 }, { "epoch": 18.551648351648353, "grad_norm": 0.68359375, "learning_rate": 4.967771109919308e-05, "loss": 0.1202, "step": 25323 }, { "epoch": 18.552380952380954, "grad_norm": 1.40625, "learning_rate": 4.967428521260667e-05, "loss": 0.1309, "step": 25324 }, { "epoch": 18.553113553113555, "grad_norm": 1.2578125, "learning_rate": 4.9670859329976525e-05, "loss": 0.1409, "step": 25325 }, { "epoch": 18.553846153846155, "grad_norm": 1.1484375, "learning_rate": 4.966743345131906e-05, "loss": 0.1183, "step": 25326 }, { "epoch": 18.554578754578756, "grad_norm": 1.4765625, "learning_rate": 4.96640075766507e-05, "loss": 0.1321, "step": 25327 }, { "epoch": 18.555311355311357, "grad_norm": 0.8984375, "learning_rate": 4.966058170598786e-05, "loss": 0.1114, "step": 25328 }, { "epoch": 18.556043956043958, "grad_norm": 1.140625, "learning_rate": 4.965715583934695e-05, "loss": 0.1213, "step": 25329 }, { "epoch": 18.55677655677656, "grad_norm": 0.90625, "learning_rate": 4.965372997674438e-05, "loss": 0.1294, "step": 25330 }, { "epoch": 18.55750915750916, "grad_norm": 1.03125, "learning_rate": 4.9650304118196555e-05, "loss": 0.0871, "step": 25331 }, { "epoch": 18.55824175824176, "grad_norm": 0.9453125, "learning_rate": 4.9646878263719914e-05, "loss": 0.1318, "step": 25332 }, { "epoch": 18.55897435897436, "grad_norm": 1.375, "learning_rate": 4.9643452413330864e-05, "loss": 0.1241, "step": 25333 }, { "epoch": 18.55970695970696, "grad_norm": 1.078125, "learning_rate": 4.96400265670458e-05, "loss": 0.103, "step": 25334 }, { "epoch": 18.560439560439562, "grad_norm": 0.91015625, "learning_rate": 4.963660072488116e-05, "loss": 0.1203, "step": 25335 }, { "epoch": 18.561172161172163, "grad_norm": 1.203125, "learning_rate": 4.9633174886853335e-05, "loss": 0.1079, "step": 25336 }, { "epoch": 18.561904761904763, "grad_norm": 1.078125, "learning_rate": 4.962974905297877e-05, "loss": 0.1341, "step": 25337 }, { "epoch": 18.562637362637364, "grad_norm": 1.2890625, "learning_rate": 4.9626323223273866e-05, "loss": 0.129, "step": 25338 }, { "epoch": 18.563369963369965, "grad_norm": 0.80859375, "learning_rate": 4.962289739775504e-05, "loss": 0.1427, "step": 25339 }, { "epoch": 18.564102564102566, "grad_norm": 0.81640625, "learning_rate": 4.9619471576438686e-05, "loss": 0.1199, "step": 25340 }, { "epoch": 18.564835164835166, "grad_norm": 1.484375, "learning_rate": 4.9616045759341216e-05, "loss": 0.1129, "step": 25341 }, { "epoch": 18.565567765567767, "grad_norm": 1.515625, "learning_rate": 4.9612619946479086e-05, "loss": 0.1313, "step": 25342 }, { "epoch": 18.566300366300368, "grad_norm": 0.84375, "learning_rate": 4.960919413786869e-05, "loss": 0.1341, "step": 25343 }, { "epoch": 18.56703296703297, "grad_norm": 0.72265625, "learning_rate": 4.9605768333526424e-05, "loss": 0.1159, "step": 25344 }, { "epoch": 18.56776556776557, "grad_norm": 1.125, "learning_rate": 4.9602342533468715e-05, "loss": 0.1038, "step": 25345 }, { "epoch": 18.56849816849817, "grad_norm": 1.3671875, "learning_rate": 4.9598916737711976e-05, "loss": 0.143, "step": 25346 }, { "epoch": 18.56923076923077, "grad_norm": 1.234375, "learning_rate": 4.959549094627263e-05, "loss": 0.147, "step": 25347 }, { "epoch": 18.56996336996337, "grad_norm": 1.0625, "learning_rate": 4.959206515916708e-05, "loss": 0.1469, "step": 25348 }, { "epoch": 18.570695970695972, "grad_norm": 1.3359375, "learning_rate": 4.9588639376411747e-05, "loss": 0.135, "step": 25349 }, { "epoch": 18.571428571428573, "grad_norm": 0.96484375, "learning_rate": 4.958521359802305e-05, "loss": 0.114, "step": 25350 }, { "epoch": 18.572161172161174, "grad_norm": 0.9375, "learning_rate": 4.9581787824017386e-05, "loss": 0.1219, "step": 25351 }, { "epoch": 18.572893772893774, "grad_norm": 1.4296875, "learning_rate": 4.957836205441116e-05, "loss": 0.1351, "step": 25352 }, { "epoch": 18.573626373626375, "grad_norm": 0.8515625, "learning_rate": 4.957493628922083e-05, "loss": 0.1313, "step": 25353 }, { "epoch": 18.574358974358976, "grad_norm": 1.3671875, "learning_rate": 4.957151052846278e-05, "loss": 0.1182, "step": 25354 }, { "epoch": 18.575091575091577, "grad_norm": 1.46875, "learning_rate": 4.9568084772153426e-05, "loss": 0.1288, "step": 25355 }, { "epoch": 18.575824175824177, "grad_norm": 0.66015625, "learning_rate": 4.956465902030918e-05, "loss": 0.0947, "step": 25356 }, { "epoch": 18.576556776556778, "grad_norm": 0.796875, "learning_rate": 4.9561233272946446e-05, "loss": 0.1132, "step": 25357 }, { "epoch": 18.57728937728938, "grad_norm": 1.046875, "learning_rate": 4.9557807530081676e-05, "loss": 0.169, "step": 25358 }, { "epoch": 18.57802197802198, "grad_norm": 1.3515625, "learning_rate": 4.955438179173123e-05, "loss": 0.111, "step": 25359 }, { "epoch": 18.57875457875458, "grad_norm": 1.15625, "learning_rate": 4.955095605791158e-05, "loss": 0.1203, "step": 25360 }, { "epoch": 18.57948717948718, "grad_norm": 0.796875, "learning_rate": 4.954753032863911e-05, "loss": 0.1191, "step": 25361 }, { "epoch": 18.58021978021978, "grad_norm": 0.82421875, "learning_rate": 4.9544104603930205e-05, "loss": 0.1239, "step": 25362 }, { "epoch": 18.580952380952382, "grad_norm": 0.8828125, "learning_rate": 4.954067888380133e-05, "loss": 0.1112, "step": 25363 }, { "epoch": 18.581684981684983, "grad_norm": 1.1796875, "learning_rate": 4.953725316826889e-05, "loss": 0.1394, "step": 25364 }, { "epoch": 18.582417582417584, "grad_norm": 1.1171875, "learning_rate": 4.953382745734928e-05, "loss": 0.1276, "step": 25365 }, { "epoch": 18.583150183150185, "grad_norm": 1.15625, "learning_rate": 4.9530401751058916e-05, "loss": 0.1273, "step": 25366 }, { "epoch": 18.583882783882785, "grad_norm": 0.94921875, "learning_rate": 4.95269760494142e-05, "loss": 0.1341, "step": 25367 }, { "epoch": 18.584615384615386, "grad_norm": 1.171875, "learning_rate": 4.9523550352431576e-05, "loss": 0.1269, "step": 25368 }, { "epoch": 18.585347985347987, "grad_norm": 0.98828125, "learning_rate": 4.952012466012744e-05, "loss": 0.1291, "step": 25369 }, { "epoch": 18.586080586080588, "grad_norm": 0.94921875, "learning_rate": 4.951669897251822e-05, "loss": 0.1133, "step": 25370 }, { "epoch": 18.58681318681319, "grad_norm": 0.69921875, "learning_rate": 4.9513273289620296e-05, "loss": 0.1134, "step": 25371 }, { "epoch": 18.58754578754579, "grad_norm": 1.5, "learning_rate": 4.950984761145011e-05, "loss": 0.1131, "step": 25372 }, { "epoch": 18.58827838827839, "grad_norm": 1.3125, "learning_rate": 4.950642193802408e-05, "loss": 0.1503, "step": 25373 }, { "epoch": 18.58901098901099, "grad_norm": 1.0078125, "learning_rate": 4.95029962693586e-05, "loss": 0.1277, "step": 25374 }, { "epoch": 18.58974358974359, "grad_norm": 0.83203125, "learning_rate": 4.9499570605470095e-05, "loss": 0.1049, "step": 25375 }, { "epoch": 18.590476190476192, "grad_norm": 0.9921875, "learning_rate": 4.9496144946374986e-05, "loss": 0.1295, "step": 25376 }, { "epoch": 18.591208791208793, "grad_norm": 1.1328125, "learning_rate": 4.949271929208967e-05, "loss": 0.1377, "step": 25377 }, { "epoch": 18.591941391941393, "grad_norm": 0.6328125, "learning_rate": 4.948929364263055e-05, "loss": 0.1124, "step": 25378 }, { "epoch": 18.592673992673994, "grad_norm": 1.0625, "learning_rate": 4.9485867998014066e-05, "loss": 0.1062, "step": 25379 }, { "epoch": 18.593406593406595, "grad_norm": 0.7578125, "learning_rate": 4.948244235825663e-05, "loss": 0.1153, "step": 25380 }, { "epoch": 18.594139194139196, "grad_norm": 0.83984375, "learning_rate": 4.947901672337465e-05, "loss": 0.1548, "step": 25381 }, { "epoch": 18.594871794871796, "grad_norm": 1.2109375, "learning_rate": 4.947559109338453e-05, "loss": 0.1178, "step": 25382 }, { "epoch": 18.595604395604397, "grad_norm": 0.73828125, "learning_rate": 4.9472165468302664e-05, "loss": 0.1036, "step": 25383 }, { "epoch": 18.596336996336998, "grad_norm": 1.234375, "learning_rate": 4.946873984814552e-05, "loss": 0.146, "step": 25384 }, { "epoch": 18.5970695970696, "grad_norm": 1.1484375, "learning_rate": 4.946531423292947e-05, "loss": 0.1473, "step": 25385 }, { "epoch": 18.5978021978022, "grad_norm": 1.40625, "learning_rate": 4.9461888622670944e-05, "loss": 0.1096, "step": 25386 }, { "epoch": 18.5985347985348, "grad_norm": 1.609375, "learning_rate": 4.945846301738636e-05, "loss": 0.1152, "step": 25387 }, { "epoch": 18.5992673992674, "grad_norm": 1.03125, "learning_rate": 4.9455037417092096e-05, "loss": 0.0993, "step": 25388 }, { "epoch": 18.6, "grad_norm": 0.90234375, "learning_rate": 4.94516118218046e-05, "loss": 0.0896, "step": 25389 }, { "epoch": 18.600732600732602, "grad_norm": 0.8828125, "learning_rate": 4.944818623154029e-05, "loss": 0.1116, "step": 25390 }, { "epoch": 18.601465201465203, "grad_norm": 1.1171875, "learning_rate": 4.944476064631556e-05, "loss": 0.1237, "step": 25391 }, { "epoch": 18.602197802197804, "grad_norm": 0.66015625, "learning_rate": 4.9441335066146835e-05, "loss": 0.1164, "step": 25392 }, { "epoch": 18.602930402930404, "grad_norm": 1.28125, "learning_rate": 4.943790949105049e-05, "loss": 0.1327, "step": 25393 }, { "epoch": 18.603663003663005, "grad_norm": 0.73046875, "learning_rate": 4.9434483921042996e-05, "loss": 0.1, "step": 25394 }, { "epoch": 18.604395604395606, "grad_norm": 1.296875, "learning_rate": 4.943105835614074e-05, "loss": 0.1228, "step": 25395 }, { "epoch": 18.605128205128207, "grad_norm": 0.6796875, "learning_rate": 4.942763279636014e-05, "loss": 0.1097, "step": 25396 }, { "epoch": 18.605860805860807, "grad_norm": 1.1875, "learning_rate": 4.942420724171758e-05, "loss": 0.1257, "step": 25397 }, { "epoch": 18.606593406593408, "grad_norm": 1.0390625, "learning_rate": 4.94207816922295e-05, "loss": 0.1494, "step": 25398 }, { "epoch": 18.60732600732601, "grad_norm": 0.75, "learning_rate": 4.941735614791233e-05, "loss": 0.1023, "step": 25399 }, { "epoch": 18.60805860805861, "grad_norm": 0.91015625, "learning_rate": 4.941393060878244e-05, "loss": 0.0959, "step": 25400 }, { "epoch": 18.60879120879121, "grad_norm": 0.65234375, "learning_rate": 4.941050507485628e-05, "loss": 0.1055, "step": 25401 }, { "epoch": 18.60952380952381, "grad_norm": 0.90234375, "learning_rate": 4.940707954615025e-05, "loss": 0.1302, "step": 25402 }, { "epoch": 18.61025641025641, "grad_norm": 1.40625, "learning_rate": 4.9403654022680754e-05, "loss": 0.096, "step": 25403 }, { "epoch": 18.610989010989012, "grad_norm": 0.7109375, "learning_rate": 4.940022850446419e-05, "loss": 0.1069, "step": 25404 }, { "epoch": 18.611721611721613, "grad_norm": 1.40625, "learning_rate": 4.9396802991517015e-05, "loss": 0.1142, "step": 25405 }, { "epoch": 18.612454212454214, "grad_norm": 1.671875, "learning_rate": 4.939337748385562e-05, "loss": 0.1166, "step": 25406 }, { "epoch": 18.613186813186815, "grad_norm": 0.921875, "learning_rate": 4.9389951981496425e-05, "loss": 0.1093, "step": 25407 }, { "epoch": 18.613919413919415, "grad_norm": 1.34375, "learning_rate": 4.938652648445583e-05, "loss": 0.1582, "step": 25408 }, { "epoch": 18.614652014652016, "grad_norm": 1.390625, "learning_rate": 4.9383100992750226e-05, "loss": 0.1581, "step": 25409 }, { "epoch": 18.615384615384617, "grad_norm": 0.9765625, "learning_rate": 4.9379675506396075e-05, "loss": 0.1195, "step": 25410 }, { "epoch": 18.616117216117217, "grad_norm": 0.8203125, "learning_rate": 4.9376250025409755e-05, "loss": 0.1045, "step": 25411 }, { "epoch": 18.616849816849818, "grad_norm": 1.7421875, "learning_rate": 4.9372824549807705e-05, "loss": 0.1226, "step": 25412 }, { "epoch": 18.61758241758242, "grad_norm": 1.234375, "learning_rate": 4.936939907960632e-05, "loss": 0.1108, "step": 25413 }, { "epoch": 18.61831501831502, "grad_norm": 1.2265625, "learning_rate": 4.936597361482199e-05, "loss": 0.1355, "step": 25414 }, { "epoch": 18.61904761904762, "grad_norm": 0.7890625, "learning_rate": 4.9362548155471175e-05, "loss": 0.1317, "step": 25415 }, { "epoch": 18.61978021978022, "grad_norm": 0.79296875, "learning_rate": 4.9359122701570265e-05, "loss": 0.1223, "step": 25416 }, { "epoch": 18.620512820512822, "grad_norm": 0.7890625, "learning_rate": 4.935569725313568e-05, "loss": 0.1194, "step": 25417 }, { "epoch": 18.621245421245423, "grad_norm": 1.390625, "learning_rate": 4.935227181018382e-05, "loss": 0.1481, "step": 25418 }, { "epoch": 18.621978021978023, "grad_norm": 1.1953125, "learning_rate": 4.9348846372731085e-05, "loss": 0.1313, "step": 25419 }, { "epoch": 18.622710622710624, "grad_norm": 1.21875, "learning_rate": 4.9345420940793916e-05, "loss": 0.1328, "step": 25420 }, { "epoch": 18.623443223443225, "grad_norm": 0.8984375, "learning_rate": 4.9341995514388725e-05, "loss": 0.1023, "step": 25421 }, { "epoch": 18.624175824175826, "grad_norm": 1.421875, "learning_rate": 4.933857009353191e-05, "loss": 0.1221, "step": 25422 }, { "epoch": 18.624908424908426, "grad_norm": 1.046875, "learning_rate": 4.933514467823988e-05, "loss": 0.1303, "step": 25423 }, { "epoch": 18.625641025641027, "grad_norm": 1.3671875, "learning_rate": 4.933171926852905e-05, "loss": 0.1719, "step": 25424 }, { "epoch": 18.626373626373628, "grad_norm": 0.83984375, "learning_rate": 4.932829386441584e-05, "loss": 0.112, "step": 25425 }, { "epoch": 18.62710622710623, "grad_norm": 1.1171875, "learning_rate": 4.932486846591666e-05, "loss": 0.1436, "step": 25426 }, { "epoch": 18.62783882783883, "grad_norm": 0.8359375, "learning_rate": 4.932144307304792e-05, "loss": 0.1276, "step": 25427 }, { "epoch": 18.62857142857143, "grad_norm": 1.65625, "learning_rate": 4.931801768582605e-05, "loss": 0.1423, "step": 25428 }, { "epoch": 18.62930402930403, "grad_norm": 0.78515625, "learning_rate": 4.931459230426744e-05, "loss": 0.1152, "step": 25429 }, { "epoch": 18.63003663003663, "grad_norm": 0.75390625, "learning_rate": 4.931116692838847e-05, "loss": 0.1034, "step": 25430 }, { "epoch": 18.630769230769232, "grad_norm": 1.59375, "learning_rate": 4.930774155820563e-05, "loss": 0.1434, "step": 25431 }, { "epoch": 18.631501831501833, "grad_norm": 0.5625, "learning_rate": 4.930431619373528e-05, "loss": 0.1149, "step": 25432 }, { "epoch": 18.632234432234434, "grad_norm": 1.140625, "learning_rate": 4.9300890834993845e-05, "loss": 0.1037, "step": 25433 }, { "epoch": 18.632967032967034, "grad_norm": 1.40625, "learning_rate": 4.929746548199773e-05, "loss": 0.1357, "step": 25434 }, { "epoch": 18.633699633699635, "grad_norm": 1.1953125, "learning_rate": 4.929404013476335e-05, "loss": 0.1573, "step": 25435 }, { "epoch": 18.634432234432236, "grad_norm": 1.375, "learning_rate": 4.9290614793307135e-05, "loss": 0.167, "step": 25436 }, { "epoch": 18.635164835164836, "grad_norm": 1.1171875, "learning_rate": 4.928718945764546e-05, "loss": 0.0917, "step": 25437 }, { "epoch": 18.635897435897437, "grad_norm": 0.98046875, "learning_rate": 4.928376412779478e-05, "loss": 0.1072, "step": 25438 }, { "epoch": 18.636630036630038, "grad_norm": 1.3671875, "learning_rate": 4.928033880377148e-05, "loss": 0.1437, "step": 25439 }, { "epoch": 18.63736263736264, "grad_norm": 1.1015625, "learning_rate": 4.927691348559195e-05, "loss": 0.1275, "step": 25440 }, { "epoch": 18.63809523809524, "grad_norm": 1.296875, "learning_rate": 4.927348817327267e-05, "loss": 0.1055, "step": 25441 }, { "epoch": 18.63882783882784, "grad_norm": 1.5078125, "learning_rate": 4.927006286682999e-05, "loss": 0.1354, "step": 25442 }, { "epoch": 18.63956043956044, "grad_norm": 1.1796875, "learning_rate": 4.9266637566280355e-05, "loss": 0.1442, "step": 25443 }, { "epoch": 18.64029304029304, "grad_norm": 1.6796875, "learning_rate": 4.926321227164016e-05, "loss": 0.1406, "step": 25444 }, { "epoch": 18.641025641025642, "grad_norm": 1.0078125, "learning_rate": 4.92597869829258e-05, "loss": 0.0997, "step": 25445 }, { "epoch": 18.641758241758243, "grad_norm": 1.9765625, "learning_rate": 4.9256361700153726e-05, "loss": 0.139, "step": 25446 }, { "epoch": 18.642490842490844, "grad_norm": 2.125, "learning_rate": 4.9252936423340334e-05, "loss": 0.1619, "step": 25447 }, { "epoch": 18.643223443223444, "grad_norm": 1.328125, "learning_rate": 4.9249511152502035e-05, "loss": 0.1324, "step": 25448 }, { "epoch": 18.643956043956045, "grad_norm": 0.8359375, "learning_rate": 4.924608588765522e-05, "loss": 0.1062, "step": 25449 }, { "epoch": 18.644688644688646, "grad_norm": 1.0703125, "learning_rate": 4.924266062881633e-05, "loss": 0.1463, "step": 25450 }, { "epoch": 18.645421245421247, "grad_norm": 1.1875, "learning_rate": 4.923923537600176e-05, "loss": 0.1187, "step": 25451 }, { "epoch": 18.646153846153847, "grad_norm": 1.75, "learning_rate": 4.9235810129227935e-05, "loss": 0.136, "step": 25452 }, { "epoch": 18.646886446886448, "grad_norm": 1.6328125, "learning_rate": 4.9232384888511266e-05, "loss": 0.1458, "step": 25453 }, { "epoch": 18.64761904761905, "grad_norm": 1.0859375, "learning_rate": 4.922895965386815e-05, "loss": 0.1391, "step": 25454 }, { "epoch": 18.64835164835165, "grad_norm": 0.9765625, "learning_rate": 4.922553442531501e-05, "loss": 0.1032, "step": 25455 }, { "epoch": 18.64908424908425, "grad_norm": 1.2734375, "learning_rate": 4.922210920286823e-05, "loss": 0.1293, "step": 25456 }, { "epoch": 18.64981684981685, "grad_norm": 1.53125, "learning_rate": 4.9218683986544265e-05, "loss": 0.1467, "step": 25457 }, { "epoch": 18.650549450549452, "grad_norm": 0.6953125, "learning_rate": 4.9215258776359515e-05, "loss": 0.1019, "step": 25458 }, { "epoch": 18.651282051282053, "grad_norm": 1.2734375, "learning_rate": 4.921183357233037e-05, "loss": 0.1214, "step": 25459 }, { "epoch": 18.652014652014653, "grad_norm": 0.95703125, "learning_rate": 4.920840837447324e-05, "loss": 0.1284, "step": 25460 }, { "epoch": 18.652747252747254, "grad_norm": 0.99609375, "learning_rate": 4.920498318280456e-05, "loss": 0.1289, "step": 25461 }, { "epoch": 18.653479853479855, "grad_norm": 1.4609375, "learning_rate": 4.920155799734074e-05, "loss": 0.1528, "step": 25462 }, { "epoch": 18.654212454212455, "grad_norm": 1.3203125, "learning_rate": 4.9198132818098166e-05, "loss": 0.1097, "step": 25463 }, { "epoch": 18.654945054945056, "grad_norm": 1.1875, "learning_rate": 4.919470764509328e-05, "loss": 0.1088, "step": 25464 }, { "epoch": 18.655677655677657, "grad_norm": 0.8671875, "learning_rate": 4.919128247834247e-05, "loss": 0.1297, "step": 25465 }, { "epoch": 18.656410256410258, "grad_norm": 0.86328125, "learning_rate": 4.9187857317862144e-05, "loss": 0.1241, "step": 25466 }, { "epoch": 18.65714285714286, "grad_norm": 1.109375, "learning_rate": 4.918443216366874e-05, "loss": 0.0977, "step": 25467 }, { "epoch": 18.65787545787546, "grad_norm": 1.1640625, "learning_rate": 4.918100701577866e-05, "loss": 0.1108, "step": 25468 }, { "epoch": 18.65860805860806, "grad_norm": 1.171875, "learning_rate": 4.91775818742083e-05, "loss": 0.1002, "step": 25469 }, { "epoch": 18.65934065934066, "grad_norm": 1.09375, "learning_rate": 4.917415673897409e-05, "loss": 0.112, "step": 25470 }, { "epoch": 18.66007326007326, "grad_norm": 0.87109375, "learning_rate": 4.91707316100924e-05, "loss": 0.126, "step": 25471 }, { "epoch": 18.660805860805862, "grad_norm": 1.1484375, "learning_rate": 4.9167306487579685e-05, "loss": 0.1231, "step": 25472 }, { "epoch": 18.661538461538463, "grad_norm": 1.1171875, "learning_rate": 4.916388137145236e-05, "loss": 0.138, "step": 25473 }, { "epoch": 18.662271062271063, "grad_norm": 0.87890625, "learning_rate": 4.916045626172681e-05, "loss": 0.1097, "step": 25474 }, { "epoch": 18.663003663003664, "grad_norm": 0.9609375, "learning_rate": 4.915703115841944e-05, "loss": 0.1396, "step": 25475 }, { "epoch": 18.663736263736265, "grad_norm": 0.98046875, "learning_rate": 4.9153606061546674e-05, "loss": 0.0991, "step": 25476 }, { "epoch": 18.664468864468866, "grad_norm": 0.90234375, "learning_rate": 4.915018097112493e-05, "loss": 0.1101, "step": 25477 }, { "epoch": 18.665201465201466, "grad_norm": 1.1015625, "learning_rate": 4.914675588717061e-05, "loss": 0.1381, "step": 25478 }, { "epoch": 18.665934065934067, "grad_norm": 1.4140625, "learning_rate": 4.914333080970014e-05, "loss": 0.107, "step": 25479 }, { "epoch": 18.666666666666668, "grad_norm": 1.265625, "learning_rate": 4.913990573872992e-05, "loss": 0.1225, "step": 25480 }, { "epoch": 18.66739926739927, "grad_norm": 1.0703125, "learning_rate": 4.9136480674276354e-05, "loss": 0.1306, "step": 25481 }, { "epoch": 18.66813186813187, "grad_norm": 0.7421875, "learning_rate": 4.9133055616355826e-05, "loss": 0.0952, "step": 25482 }, { "epoch": 18.66886446886447, "grad_norm": 0.99609375, "learning_rate": 4.9129630564984806e-05, "loss": 0.1264, "step": 25483 }, { "epoch": 18.66959706959707, "grad_norm": 0.78125, "learning_rate": 4.9126205520179675e-05, "loss": 0.133, "step": 25484 }, { "epoch": 18.67032967032967, "grad_norm": 1.109375, "learning_rate": 4.912278048195685e-05, "loss": 0.1029, "step": 25485 }, { "epoch": 18.671062271062272, "grad_norm": 1.0703125, "learning_rate": 4.911935545033272e-05, "loss": 0.1131, "step": 25486 }, { "epoch": 18.671794871794873, "grad_norm": 1.1875, "learning_rate": 4.911593042532371e-05, "loss": 0.1255, "step": 25487 }, { "epoch": 18.672527472527474, "grad_norm": 0.94140625, "learning_rate": 4.911250540694624e-05, "loss": 0.1256, "step": 25488 }, { "epoch": 18.673260073260074, "grad_norm": 0.8046875, "learning_rate": 4.910908039521671e-05, "loss": 0.1102, "step": 25489 }, { "epoch": 18.673992673992675, "grad_norm": 1.4453125, "learning_rate": 4.910565539015154e-05, "loss": 0.154, "step": 25490 }, { "epoch": 18.674725274725276, "grad_norm": 0.796875, "learning_rate": 4.910223039176713e-05, "loss": 0.1089, "step": 25491 }, { "epoch": 18.675457875457877, "grad_norm": 1.1328125, "learning_rate": 4.909880540007987e-05, "loss": 0.088, "step": 25492 }, { "epoch": 18.676190476190477, "grad_norm": 0.8515625, "learning_rate": 4.909538041510622e-05, "loss": 0.1185, "step": 25493 }, { "epoch": 18.676923076923078, "grad_norm": 1.2265625, "learning_rate": 4.9091955436862556e-05, "loss": 0.1489, "step": 25494 }, { "epoch": 18.67765567765568, "grad_norm": 0.78125, "learning_rate": 4.90885304653653e-05, "loss": 0.1487, "step": 25495 }, { "epoch": 18.67838827838828, "grad_norm": 1.28125, "learning_rate": 4.908510550063086e-05, "loss": 0.1349, "step": 25496 }, { "epoch": 18.67912087912088, "grad_norm": 1.359375, "learning_rate": 4.9081680542675625e-05, "loss": 0.0935, "step": 25497 }, { "epoch": 18.67985347985348, "grad_norm": 1.125, "learning_rate": 4.907825559151604e-05, "loss": 0.1424, "step": 25498 }, { "epoch": 18.68058608058608, "grad_norm": 0.6953125, "learning_rate": 4.90748306471685e-05, "loss": 0.1121, "step": 25499 }, { "epoch": 18.681318681318682, "grad_norm": 1.5703125, "learning_rate": 4.907140570964941e-05, "loss": 0.1113, "step": 25500 }, { "epoch": 18.682051282051283, "grad_norm": 1.1796875, "learning_rate": 4.906798077897518e-05, "loss": 0.1544, "step": 25501 }, { "epoch": 18.682783882783884, "grad_norm": 1.1328125, "learning_rate": 4.906455585516222e-05, "loss": 0.1193, "step": 25502 }, { "epoch": 18.683516483516485, "grad_norm": 0.8984375, "learning_rate": 4.906113093822695e-05, "loss": 0.1181, "step": 25503 }, { "epoch": 18.684249084249085, "grad_norm": 0.97265625, "learning_rate": 4.9057706028185786e-05, "loss": 0.12, "step": 25504 }, { "epoch": 18.684981684981686, "grad_norm": 1.0, "learning_rate": 4.905428112505512e-05, "loss": 0.153, "step": 25505 }, { "epoch": 18.685714285714287, "grad_norm": 0.8203125, "learning_rate": 4.905085622885136e-05, "loss": 0.1154, "step": 25506 }, { "epoch": 18.686446886446888, "grad_norm": 0.9921875, "learning_rate": 4.9047431339590944e-05, "loss": 0.1197, "step": 25507 }, { "epoch": 18.68717948717949, "grad_norm": 1.0703125, "learning_rate": 4.904400645729023e-05, "loss": 0.1324, "step": 25508 }, { "epoch": 18.68791208791209, "grad_norm": 1.21875, "learning_rate": 4.9040581581965676e-05, "loss": 0.1429, "step": 25509 }, { "epoch": 18.68864468864469, "grad_norm": 1.1875, "learning_rate": 4.903715671363368e-05, "loss": 0.125, "step": 25510 }, { "epoch": 18.68937728937729, "grad_norm": 0.9453125, "learning_rate": 4.903373185231064e-05, "loss": 0.1063, "step": 25511 }, { "epoch": 18.69010989010989, "grad_norm": 1.15625, "learning_rate": 4.9030306998012965e-05, "loss": 0.1245, "step": 25512 }, { "epoch": 18.690842490842492, "grad_norm": 1.3203125, "learning_rate": 4.902688215075707e-05, "loss": 0.1457, "step": 25513 }, { "epoch": 18.691575091575093, "grad_norm": 0.9296875, "learning_rate": 4.902345731055938e-05, "loss": 0.1185, "step": 25514 }, { "epoch": 18.692307692307693, "grad_norm": 1.59375, "learning_rate": 4.902003247743628e-05, "loss": 0.1552, "step": 25515 }, { "epoch": 18.693040293040294, "grad_norm": 1.21875, "learning_rate": 4.90166076514042e-05, "loss": 0.1291, "step": 25516 }, { "epoch": 18.693772893772895, "grad_norm": 0.8671875, "learning_rate": 4.901318283247954e-05, "loss": 0.1136, "step": 25517 }, { "epoch": 18.694505494505496, "grad_norm": 0.734375, "learning_rate": 4.9009758020678695e-05, "loss": 0.1004, "step": 25518 }, { "epoch": 18.695238095238096, "grad_norm": 0.80859375, "learning_rate": 4.9006333216018097e-05, "loss": 0.1251, "step": 25519 }, { "epoch": 18.695970695970697, "grad_norm": 1.1484375, "learning_rate": 4.900290841851415e-05, "loss": 0.1232, "step": 25520 }, { "epoch": 18.696703296703298, "grad_norm": 1.0859375, "learning_rate": 4.899948362818327e-05, "loss": 0.1245, "step": 25521 }, { "epoch": 18.6974358974359, "grad_norm": 1.34375, "learning_rate": 4.8996058845041846e-05, "loss": 0.1541, "step": 25522 }, { "epoch": 18.6981684981685, "grad_norm": 1.7578125, "learning_rate": 4.899263406910628e-05, "loss": 0.114, "step": 25523 }, { "epoch": 18.6989010989011, "grad_norm": 1.1171875, "learning_rate": 4.898920930039302e-05, "loss": 0.1511, "step": 25524 }, { "epoch": 18.6996336996337, "grad_norm": 0.75390625, "learning_rate": 4.898578453891845e-05, "loss": 0.0983, "step": 25525 }, { "epoch": 18.7003663003663, "grad_norm": 1.0859375, "learning_rate": 4.898235978469898e-05, "loss": 0.1323, "step": 25526 }, { "epoch": 18.701098901098902, "grad_norm": 1.1953125, "learning_rate": 4.897893503775103e-05, "loss": 0.1349, "step": 25527 }, { "epoch": 18.701831501831503, "grad_norm": 0.828125, "learning_rate": 4.8975510298091e-05, "loss": 0.108, "step": 25528 }, { "epoch": 18.702564102564104, "grad_norm": 1.3046875, "learning_rate": 4.8972085565735275e-05, "loss": 0.1684, "step": 25529 }, { "epoch": 18.703296703296704, "grad_norm": 1.1640625, "learning_rate": 4.896866084070033e-05, "loss": 0.1272, "step": 25530 }, { "epoch": 18.704029304029305, "grad_norm": 1.234375, "learning_rate": 4.8965236123002515e-05, "loss": 0.1318, "step": 25531 }, { "epoch": 18.704761904761906, "grad_norm": 1.5625, "learning_rate": 4.8961811412658256e-05, "loss": 0.1238, "step": 25532 }, { "epoch": 18.705494505494507, "grad_norm": 0.94921875, "learning_rate": 4.895838670968397e-05, "loss": 0.1161, "step": 25533 }, { "epoch": 18.706227106227107, "grad_norm": 0.6171875, "learning_rate": 4.895496201409604e-05, "loss": 0.105, "step": 25534 }, { "epoch": 18.706959706959708, "grad_norm": 1.3046875, "learning_rate": 4.895153732591092e-05, "loss": 0.1149, "step": 25535 }, { "epoch": 18.70769230769231, "grad_norm": 1.234375, "learning_rate": 4.894811264514499e-05, "loss": 0.1084, "step": 25536 }, { "epoch": 18.70842490842491, "grad_norm": 0.66015625, "learning_rate": 4.894468797181465e-05, "loss": 0.1108, "step": 25537 }, { "epoch": 18.70915750915751, "grad_norm": 1.1953125, "learning_rate": 4.894126330593632e-05, "loss": 0.1119, "step": 25538 }, { "epoch": 18.70989010989011, "grad_norm": 0.796875, "learning_rate": 4.893783864752641e-05, "loss": 0.1224, "step": 25539 }, { "epoch": 18.71062271062271, "grad_norm": 1.296875, "learning_rate": 4.8934413996601326e-05, "loss": 0.1058, "step": 25540 }, { "epoch": 18.711355311355312, "grad_norm": 1.171875, "learning_rate": 4.893098935317749e-05, "loss": 0.1094, "step": 25541 }, { "epoch": 18.712087912087913, "grad_norm": 0.88671875, "learning_rate": 4.89275647172713e-05, "loss": 0.1205, "step": 25542 }, { "epoch": 18.712820512820514, "grad_norm": 0.953125, "learning_rate": 4.892414008889916e-05, "loss": 0.1213, "step": 25543 }, { "epoch": 18.713553113553115, "grad_norm": 0.94921875, "learning_rate": 4.892071546807746e-05, "loss": 0.1338, "step": 25544 }, { "epoch": 18.714285714285715, "grad_norm": 0.96875, "learning_rate": 4.891729085482265e-05, "loss": 0.1234, "step": 25545 }, { "epoch": 18.715018315018316, "grad_norm": 0.78515625, "learning_rate": 4.891386624915112e-05, "loss": 0.0975, "step": 25546 }, { "epoch": 18.715750915750917, "grad_norm": 0.87890625, "learning_rate": 4.8910441651079285e-05, "loss": 0.1118, "step": 25547 }, { "epoch": 18.716483516483517, "grad_norm": 1.8515625, "learning_rate": 4.890701706062354e-05, "loss": 0.1349, "step": 25548 }, { "epoch": 18.717216117216118, "grad_norm": 1.4765625, "learning_rate": 4.890359247780028e-05, "loss": 0.1428, "step": 25549 }, { "epoch": 18.71794871794872, "grad_norm": 0.85546875, "learning_rate": 4.890016790262596e-05, "loss": 0.0915, "step": 25550 }, { "epoch": 18.71868131868132, "grad_norm": 1.375, "learning_rate": 4.889674333511695e-05, "loss": 0.1265, "step": 25551 }, { "epoch": 18.71941391941392, "grad_norm": 1.7109375, "learning_rate": 4.889331877528967e-05, "loss": 0.1211, "step": 25552 }, { "epoch": 18.72014652014652, "grad_norm": 1.1953125, "learning_rate": 4.8889894223160525e-05, "loss": 0.1138, "step": 25553 }, { "epoch": 18.720879120879122, "grad_norm": 1.96875, "learning_rate": 4.8886469678745934e-05, "loss": 0.1194, "step": 25554 }, { "epoch": 18.721611721611723, "grad_norm": 0.7421875, "learning_rate": 4.888304514206227e-05, "loss": 0.1098, "step": 25555 }, { "epoch": 18.722344322344323, "grad_norm": 1.375, "learning_rate": 4.887962061312599e-05, "loss": 0.1556, "step": 25556 }, { "epoch": 18.723076923076924, "grad_norm": 1.40625, "learning_rate": 4.887619609195348e-05, "loss": 0.1309, "step": 25557 }, { "epoch": 18.723809523809525, "grad_norm": 0.91796875, "learning_rate": 4.887277157856115e-05, "loss": 0.1127, "step": 25558 }, { "epoch": 18.724542124542126, "grad_norm": 1.0078125, "learning_rate": 4.88693470729654e-05, "loss": 0.1241, "step": 25559 }, { "epoch": 18.725274725274726, "grad_norm": 1.421875, "learning_rate": 4.8865922575182625e-05, "loss": 0.1275, "step": 25560 }, { "epoch": 18.726007326007327, "grad_norm": 1.4765625, "learning_rate": 4.886249808522927e-05, "loss": 0.1007, "step": 25561 }, { "epoch": 18.726739926739928, "grad_norm": 0.75, "learning_rate": 4.8859073603121733e-05, "loss": 0.0967, "step": 25562 }, { "epoch": 18.72747252747253, "grad_norm": 1.296875, "learning_rate": 4.88556491288764e-05, "loss": 0.1119, "step": 25563 }, { "epoch": 18.72820512820513, "grad_norm": 1.2109375, "learning_rate": 4.88522246625097e-05, "loss": 0.1363, "step": 25564 }, { "epoch": 18.72893772893773, "grad_norm": 1.2578125, "learning_rate": 4.884880020403801e-05, "loss": 0.1208, "step": 25565 }, { "epoch": 18.72967032967033, "grad_norm": 1.296875, "learning_rate": 4.8845375753477775e-05, "loss": 0.1261, "step": 25566 }, { "epoch": 18.73040293040293, "grad_norm": 1.375, "learning_rate": 4.88419513108454e-05, "loss": 0.1571, "step": 25567 }, { "epoch": 18.731135531135532, "grad_norm": 0.69140625, "learning_rate": 4.8838526876157275e-05, "loss": 0.0957, "step": 25568 }, { "epoch": 18.731868131868133, "grad_norm": 1.4140625, "learning_rate": 4.8835102449429815e-05, "loss": 0.1155, "step": 25569 }, { "epoch": 18.732600732600734, "grad_norm": 1.2578125, "learning_rate": 4.88316780306794e-05, "loss": 0.1532, "step": 25570 }, { "epoch": 18.733333333333334, "grad_norm": 1.0546875, "learning_rate": 4.8828253619922495e-05, "loss": 0.1358, "step": 25571 }, { "epoch": 18.734065934065935, "grad_norm": 0.64453125, "learning_rate": 4.882482921717546e-05, "loss": 0.1033, "step": 25572 }, { "epoch": 18.734798534798536, "grad_norm": 0.796875, "learning_rate": 4.882140482245473e-05, "loss": 0.1165, "step": 25573 }, { "epoch": 18.735531135531136, "grad_norm": 0.65625, "learning_rate": 4.88179804357767e-05, "loss": 0.0982, "step": 25574 }, { "epoch": 18.736263736263737, "grad_norm": 1.34375, "learning_rate": 4.8814556057157754e-05, "loss": 0.1301, "step": 25575 }, { "epoch": 18.736996336996338, "grad_norm": 1.6953125, "learning_rate": 4.881113168661434e-05, "loss": 0.1255, "step": 25576 }, { "epoch": 18.73772893772894, "grad_norm": 1.234375, "learning_rate": 4.880770732416285e-05, "loss": 0.0934, "step": 25577 }, { "epoch": 18.73846153846154, "grad_norm": 0.9296875, "learning_rate": 4.8804282969819684e-05, "loss": 0.1132, "step": 25578 }, { "epoch": 18.73919413919414, "grad_norm": 0.95703125, "learning_rate": 4.880085862360126e-05, "loss": 0.108, "step": 25579 }, { "epoch": 18.73992673992674, "grad_norm": 0.65234375, "learning_rate": 4.879743428552399e-05, "loss": 0.1004, "step": 25580 }, { "epoch": 18.74065934065934, "grad_norm": 1.2265625, "learning_rate": 4.879400995560424e-05, "loss": 0.1051, "step": 25581 }, { "epoch": 18.741391941391942, "grad_norm": 1.15625, "learning_rate": 4.879058563385848e-05, "loss": 0.1248, "step": 25582 }, { "epoch": 18.742124542124543, "grad_norm": 0.9375, "learning_rate": 4.8787161320303076e-05, "loss": 0.1179, "step": 25583 }, { "epoch": 18.742857142857144, "grad_norm": 1.203125, "learning_rate": 4.878373701495445e-05, "loss": 0.1166, "step": 25584 }, { "epoch": 18.743589743589745, "grad_norm": 1.1953125, "learning_rate": 4.8780312717829005e-05, "loss": 0.1318, "step": 25585 }, { "epoch": 18.744322344322345, "grad_norm": 1.125, "learning_rate": 4.877688842894311e-05, "loss": 0.1291, "step": 25586 }, { "epoch": 18.745054945054946, "grad_norm": 1.453125, "learning_rate": 4.8773464148313244e-05, "loss": 0.1214, "step": 25587 }, { "epoch": 18.745787545787547, "grad_norm": 1.046875, "learning_rate": 4.877003987595578e-05, "loss": 0.0988, "step": 25588 }, { "epoch": 18.746520146520147, "grad_norm": 1.34375, "learning_rate": 4.876661561188711e-05, "loss": 0.1115, "step": 25589 }, { "epoch": 18.747252747252748, "grad_norm": 1.0859375, "learning_rate": 4.876319135612365e-05, "loss": 0.1518, "step": 25590 }, { "epoch": 18.74798534798535, "grad_norm": 0.91015625, "learning_rate": 4.875976710868181e-05, "loss": 0.1277, "step": 25591 }, { "epoch": 18.74871794871795, "grad_norm": 1.015625, "learning_rate": 4.8756342869578e-05, "loss": 0.0778, "step": 25592 }, { "epoch": 18.74945054945055, "grad_norm": 1.3046875, "learning_rate": 4.875291863882863e-05, "loss": 0.1112, "step": 25593 }, { "epoch": 18.75018315018315, "grad_norm": 0.90625, "learning_rate": 4.87494944164501e-05, "loss": 0.15, "step": 25594 }, { "epoch": 18.750915750915752, "grad_norm": 1.0234375, "learning_rate": 4.874607020245882e-05, "loss": 0.1106, "step": 25595 }, { "epoch": 18.751648351648353, "grad_norm": 0.828125, "learning_rate": 4.874264599687117e-05, "loss": 0.1174, "step": 25596 }, { "epoch": 18.752380952380953, "grad_norm": 1.28125, "learning_rate": 4.87392217997036e-05, "loss": 0.0987, "step": 25597 }, { "epoch": 18.753113553113554, "grad_norm": 1.21875, "learning_rate": 4.873579761097251e-05, "loss": 0.1502, "step": 25598 }, { "epoch": 18.753846153846155, "grad_norm": 1.2421875, "learning_rate": 4.873237343069427e-05, "loss": 0.0877, "step": 25599 }, { "epoch": 18.754578754578755, "grad_norm": 1.3515625, "learning_rate": 4.872894925888533e-05, "loss": 0.1089, "step": 25600 }, { "epoch": 18.755311355311356, "grad_norm": 1.15625, "learning_rate": 4.8725525095562035e-05, "loss": 0.14, "step": 25601 }, { "epoch": 18.756043956043957, "grad_norm": 1.265625, "learning_rate": 4.8722100940740865e-05, "loss": 0.138, "step": 25602 }, { "epoch": 18.756776556776558, "grad_norm": 1.109375, "learning_rate": 4.871867679443819e-05, "loss": 0.1176, "step": 25603 }, { "epoch": 18.75750915750916, "grad_norm": 0.859375, "learning_rate": 4.871525265667041e-05, "loss": 0.124, "step": 25604 }, { "epoch": 18.75824175824176, "grad_norm": 1.2109375, "learning_rate": 4.8711828527453955e-05, "loss": 0.1639, "step": 25605 }, { "epoch": 18.75897435897436, "grad_norm": 0.96875, "learning_rate": 4.870840440680521e-05, "loss": 0.1318, "step": 25606 }, { "epoch": 18.75970695970696, "grad_norm": 1.359375, "learning_rate": 4.8704980294740565e-05, "loss": 0.1245, "step": 25607 }, { "epoch": 18.76043956043956, "grad_norm": 1.3359375, "learning_rate": 4.870155619127647e-05, "loss": 0.1481, "step": 25608 }, { "epoch": 18.761172161172162, "grad_norm": 1.546875, "learning_rate": 4.869813209642932e-05, "loss": 0.1501, "step": 25609 }, { "epoch": 18.761904761904763, "grad_norm": 1.1953125, "learning_rate": 4.8694708010215504e-05, "loss": 0.1304, "step": 25610 }, { "epoch": 18.762637362637363, "grad_norm": 1.21875, "learning_rate": 4.869128393265143e-05, "loss": 0.1073, "step": 25611 }, { "epoch": 18.763369963369964, "grad_norm": 0.79296875, "learning_rate": 4.86878598637535e-05, "loss": 0.1059, "step": 25612 }, { "epoch": 18.764102564102565, "grad_norm": 1.4453125, "learning_rate": 4.8684435803538144e-05, "loss": 0.1194, "step": 25613 }, { "epoch": 18.764835164835166, "grad_norm": 1.3125, "learning_rate": 4.868101175202174e-05, "loss": 0.1367, "step": 25614 }, { "epoch": 18.765567765567766, "grad_norm": 0.9609375, "learning_rate": 4.867758770922071e-05, "loss": 0.0978, "step": 25615 }, { "epoch": 18.766300366300367, "grad_norm": 1.4296875, "learning_rate": 4.867416367515146e-05, "loss": 0.0928, "step": 25616 }, { "epoch": 18.767032967032968, "grad_norm": 1.3515625, "learning_rate": 4.867073964983038e-05, "loss": 0.1208, "step": 25617 }, { "epoch": 18.76776556776557, "grad_norm": 1.3125, "learning_rate": 4.866731563327389e-05, "loss": 0.1331, "step": 25618 }, { "epoch": 18.76849816849817, "grad_norm": 1.3984375, "learning_rate": 4.86638916254984e-05, "loss": 0.1052, "step": 25619 }, { "epoch": 18.76923076923077, "grad_norm": 0.88671875, "learning_rate": 4.866046762652031e-05, "loss": 0.1413, "step": 25620 }, { "epoch": 18.76996336996337, "grad_norm": 1.484375, "learning_rate": 4.865704363635602e-05, "loss": 0.1223, "step": 25621 }, { "epoch": 18.77069597069597, "grad_norm": 1.1328125, "learning_rate": 4.865361965502192e-05, "loss": 0.1469, "step": 25622 }, { "epoch": 18.771428571428572, "grad_norm": 1.125, "learning_rate": 4.8650195682534456e-05, "loss": 0.1056, "step": 25623 }, { "epoch": 18.772161172161173, "grad_norm": 1.3046875, "learning_rate": 4.864677171891001e-05, "loss": 0.1311, "step": 25624 }, { "epoch": 18.772893772893774, "grad_norm": 1.1640625, "learning_rate": 4.864334776416499e-05, "loss": 0.1465, "step": 25625 }, { "epoch": 18.773626373626374, "grad_norm": 1.1953125, "learning_rate": 4.8639923818315804e-05, "loss": 0.1356, "step": 25626 }, { "epoch": 18.774358974358975, "grad_norm": 1.65625, "learning_rate": 4.863649988137883e-05, "loss": 0.1329, "step": 25627 }, { "epoch": 18.775091575091576, "grad_norm": 1.03125, "learning_rate": 4.863307595337052e-05, "loss": 0.1346, "step": 25628 }, { "epoch": 18.775824175824177, "grad_norm": 0.953125, "learning_rate": 4.862965203430725e-05, "loss": 0.1511, "step": 25629 }, { "epoch": 18.776556776556777, "grad_norm": 1.1953125, "learning_rate": 4.8626228124205424e-05, "loss": 0.1205, "step": 25630 }, { "epoch": 18.777289377289378, "grad_norm": 1.4140625, "learning_rate": 4.8622804223081464e-05, "loss": 0.1487, "step": 25631 }, { "epoch": 18.77802197802198, "grad_norm": 0.84765625, "learning_rate": 4.861938033095176e-05, "loss": 0.1255, "step": 25632 }, { "epoch": 18.77875457875458, "grad_norm": 0.93359375, "learning_rate": 4.861595644783271e-05, "loss": 0.1379, "step": 25633 }, { "epoch": 18.77948717948718, "grad_norm": 0.79296875, "learning_rate": 4.861253257374075e-05, "loss": 0.1168, "step": 25634 }, { "epoch": 18.78021978021978, "grad_norm": 0.9609375, "learning_rate": 4.8609108708692264e-05, "loss": 0.1175, "step": 25635 }, { "epoch": 18.78095238095238, "grad_norm": 2.1875, "learning_rate": 4.860568485270366e-05, "loss": 0.099, "step": 25636 }, { "epoch": 18.781684981684982, "grad_norm": 1.296875, "learning_rate": 4.860226100579134e-05, "loss": 0.1255, "step": 25637 }, { "epoch": 18.782417582417583, "grad_norm": 1.28125, "learning_rate": 4.85988371679717e-05, "loss": 0.1059, "step": 25638 }, { "epoch": 18.783150183150184, "grad_norm": 1.625, "learning_rate": 4.8595413339261166e-05, "loss": 0.1152, "step": 25639 }, { "epoch": 18.783882783882785, "grad_norm": 0.640625, "learning_rate": 4.8591989519676134e-05, "loss": 0.1144, "step": 25640 }, { "epoch": 18.784615384615385, "grad_norm": 1.5390625, "learning_rate": 4.8588565709232994e-05, "loss": 0.1176, "step": 25641 }, { "epoch": 18.785347985347986, "grad_norm": 1.6015625, "learning_rate": 4.858514190794817e-05, "loss": 0.1183, "step": 25642 }, { "epoch": 18.786080586080587, "grad_norm": 0.67578125, "learning_rate": 4.8581718115838056e-05, "loss": 0.1078, "step": 25643 }, { "epoch": 18.786813186813188, "grad_norm": 1.4140625, "learning_rate": 4.857829433291906e-05, "loss": 0.1653, "step": 25644 }, { "epoch": 18.78754578754579, "grad_norm": 0.8125, "learning_rate": 4.85748705592076e-05, "loss": 0.1215, "step": 25645 }, { "epoch": 18.78827838827839, "grad_norm": 1.1328125, "learning_rate": 4.857144679472006e-05, "loss": 0.0997, "step": 25646 }, { "epoch": 18.78901098901099, "grad_norm": 1.046875, "learning_rate": 4.8568023039472856e-05, "loss": 0.1255, "step": 25647 }, { "epoch": 18.78974358974359, "grad_norm": 0.9453125, "learning_rate": 4.856459929348236e-05, "loss": 0.1087, "step": 25648 }, { "epoch": 18.79047619047619, "grad_norm": 1.5234375, "learning_rate": 4.856117555676503e-05, "loss": 0.1257, "step": 25649 }, { "epoch": 18.791208791208792, "grad_norm": 1.421875, "learning_rate": 4.855775182933725e-05, "loss": 0.1505, "step": 25650 }, { "epoch": 18.791941391941393, "grad_norm": 1.0234375, "learning_rate": 4.855432811121541e-05, "loss": 0.1176, "step": 25651 }, { "epoch": 18.792673992673993, "grad_norm": 1.0390625, "learning_rate": 4.855090440241593e-05, "loss": 0.1192, "step": 25652 }, { "epoch": 18.793406593406594, "grad_norm": 1.171875, "learning_rate": 4.854748070295518e-05, "loss": 0.1673, "step": 25653 }, { "epoch": 18.794139194139195, "grad_norm": 0.96484375, "learning_rate": 4.8544057012849616e-05, "loss": 0.1016, "step": 25654 }, { "epoch": 18.794871794871796, "grad_norm": 1.46875, "learning_rate": 4.854063333211561e-05, "loss": 0.1279, "step": 25655 }, { "epoch": 18.795604395604396, "grad_norm": 1.546875, "learning_rate": 4.8537209660769573e-05, "loss": 0.1171, "step": 25656 }, { "epoch": 18.796336996336997, "grad_norm": 1.65625, "learning_rate": 4.853378599882791e-05, "loss": 0.103, "step": 25657 }, { "epoch": 18.797069597069598, "grad_norm": 1.40625, "learning_rate": 4.853036234630704e-05, "loss": 0.1289, "step": 25658 }, { "epoch": 18.7978021978022, "grad_norm": 1.5703125, "learning_rate": 4.852693870322331e-05, "loss": 0.1186, "step": 25659 }, { "epoch": 18.7985347985348, "grad_norm": 0.6875, "learning_rate": 4.852351506959319e-05, "loss": 0.1005, "step": 25660 }, { "epoch": 18.7992673992674, "grad_norm": 1.6484375, "learning_rate": 4.8520091445433063e-05, "loss": 0.1361, "step": 25661 }, { "epoch": 18.8, "grad_norm": 1.5390625, "learning_rate": 4.851666783075932e-05, "loss": 0.134, "step": 25662 }, { "epoch": 18.8007326007326, "grad_norm": 1.140625, "learning_rate": 4.8513244225588376e-05, "loss": 0.1272, "step": 25663 }, { "epoch": 18.801465201465202, "grad_norm": 1.765625, "learning_rate": 4.8509820629936616e-05, "loss": 0.1655, "step": 25664 }, { "epoch": 18.802197802197803, "grad_norm": 1.40625, "learning_rate": 4.8506397043820476e-05, "loss": 0.155, "step": 25665 }, { "epoch": 18.802930402930404, "grad_norm": 0.73046875, "learning_rate": 4.850297346725634e-05, "loss": 0.1325, "step": 25666 }, { "epoch": 18.803663003663004, "grad_norm": 1.5, "learning_rate": 4.8499549900260606e-05, "loss": 0.1146, "step": 25667 }, { "epoch": 18.804395604395605, "grad_norm": 1.2265625, "learning_rate": 4.849612634284969e-05, "loss": 0.11, "step": 25668 }, { "epoch": 18.805128205128206, "grad_norm": 1.2109375, "learning_rate": 4.849270279503999e-05, "loss": 0.1114, "step": 25669 }, { "epoch": 18.805860805860807, "grad_norm": 1.0703125, "learning_rate": 4.848927925684791e-05, "loss": 0.1144, "step": 25670 }, { "epoch": 18.806593406593407, "grad_norm": 1.0078125, "learning_rate": 4.848585572828985e-05, "loss": 0.1131, "step": 25671 }, { "epoch": 18.807326007326008, "grad_norm": 1.1171875, "learning_rate": 4.848243220938223e-05, "loss": 0.1267, "step": 25672 }, { "epoch": 18.80805860805861, "grad_norm": 0.78125, "learning_rate": 4.847900870014143e-05, "loss": 0.1146, "step": 25673 }, { "epoch": 18.80879120879121, "grad_norm": 1.4765625, "learning_rate": 4.847558520058385e-05, "loss": 0.1105, "step": 25674 }, { "epoch": 18.80952380952381, "grad_norm": 1.078125, "learning_rate": 4.8472161710725925e-05, "loss": 0.1584, "step": 25675 }, { "epoch": 18.81025641025641, "grad_norm": 0.89453125, "learning_rate": 4.8468738230584035e-05, "loss": 0.1133, "step": 25676 }, { "epoch": 18.81098901098901, "grad_norm": 0.93359375, "learning_rate": 4.846531476017459e-05, "loss": 0.1093, "step": 25677 }, { "epoch": 18.811721611721612, "grad_norm": 1.328125, "learning_rate": 4.846189129951399e-05, "loss": 0.1449, "step": 25678 }, { "epoch": 18.812454212454213, "grad_norm": 0.66796875, "learning_rate": 4.845846784861861e-05, "loss": 0.1242, "step": 25679 }, { "epoch": 18.813186813186814, "grad_norm": 0.8046875, "learning_rate": 4.8455044407504905e-05, "loss": 0.1184, "step": 25680 }, { "epoch": 18.813919413919415, "grad_norm": 1.453125, "learning_rate": 4.8451620976189246e-05, "loss": 0.1236, "step": 25681 }, { "epoch": 18.814652014652015, "grad_norm": 1.015625, "learning_rate": 4.8448197554688063e-05, "loss": 0.1249, "step": 25682 }, { "epoch": 18.815384615384616, "grad_norm": 0.8125, "learning_rate": 4.844477414301772e-05, "loss": 0.1075, "step": 25683 }, { "epoch": 18.816117216117217, "grad_norm": 0.89453125, "learning_rate": 4.844135074119465e-05, "loss": 0.1101, "step": 25684 }, { "epoch": 18.816849816849818, "grad_norm": 1.1171875, "learning_rate": 4.843792734923522e-05, "loss": 0.1154, "step": 25685 }, { "epoch": 18.81758241758242, "grad_norm": 1.4765625, "learning_rate": 4.843450396715589e-05, "loss": 0.1087, "step": 25686 }, { "epoch": 18.81831501831502, "grad_norm": 1.5859375, "learning_rate": 4.8431080594973015e-05, "loss": 0.1457, "step": 25687 }, { "epoch": 18.81904761904762, "grad_norm": 0.94140625, "learning_rate": 4.8427657232703017e-05, "loss": 0.1225, "step": 25688 }, { "epoch": 18.81978021978022, "grad_norm": 1.5, "learning_rate": 4.842423388036229e-05, "loss": 0.1285, "step": 25689 }, { "epoch": 18.82051282051282, "grad_norm": 1.421875, "learning_rate": 4.842081053796722e-05, "loss": 0.1156, "step": 25690 }, { "epoch": 18.821245421245422, "grad_norm": 1.2265625, "learning_rate": 4.841738720553426e-05, "loss": 0.12, "step": 25691 }, { "epoch": 18.821978021978023, "grad_norm": 1.1640625, "learning_rate": 4.841396388307977e-05, "loss": 0.1135, "step": 25692 }, { "epoch": 18.822710622710623, "grad_norm": 1.2578125, "learning_rate": 4.841054057062015e-05, "loss": 0.1206, "step": 25693 }, { "epoch": 18.823443223443224, "grad_norm": 1.3125, "learning_rate": 4.8407117268171837e-05, "loss": 0.1424, "step": 25694 }, { "epoch": 18.824175824175825, "grad_norm": 1.484375, "learning_rate": 4.8403693975751204e-05, "loss": 0.1339, "step": 25695 }, { "epoch": 18.824908424908426, "grad_norm": 1.5390625, "learning_rate": 4.840027069337465e-05, "loss": 0.1332, "step": 25696 }, { "epoch": 18.825641025641026, "grad_norm": 0.9765625, "learning_rate": 4.839684742105861e-05, "loss": 0.1197, "step": 25697 }, { "epoch": 18.826373626373627, "grad_norm": 0.6640625, "learning_rate": 4.839342415881945e-05, "loss": 0.0943, "step": 25698 }, { "epoch": 18.827106227106228, "grad_norm": 1.0078125, "learning_rate": 4.83900009066736e-05, "loss": 0.147, "step": 25699 }, { "epoch": 18.82783882783883, "grad_norm": 1.53125, "learning_rate": 4.8386577664637425e-05, "loss": 0.1227, "step": 25700 }, { "epoch": 18.82857142857143, "grad_norm": 1.2890625, "learning_rate": 4.838315443272736e-05, "loss": 0.1552, "step": 25701 }, { "epoch": 18.82930402930403, "grad_norm": 0.86328125, "learning_rate": 4.837973121095981e-05, "loss": 0.112, "step": 25702 }, { "epoch": 18.83003663003663, "grad_norm": 1.515625, "learning_rate": 4.837630799935116e-05, "loss": 0.1208, "step": 25703 }, { "epoch": 18.83076923076923, "grad_norm": 1.1640625, "learning_rate": 4.837288479791782e-05, "loss": 0.095, "step": 25704 }, { "epoch": 18.831501831501832, "grad_norm": 1.28125, "learning_rate": 4.836946160667615e-05, "loss": 0.1552, "step": 25705 }, { "epoch": 18.832234432234433, "grad_norm": 1.0546875, "learning_rate": 4.836603842564263e-05, "loss": 0.1185, "step": 25706 }, { "epoch": 18.832967032967034, "grad_norm": 0.9921875, "learning_rate": 4.8362615254833594e-05, "loss": 0.0957, "step": 25707 }, { "epoch": 18.833699633699634, "grad_norm": 1.25, "learning_rate": 4.8359192094265484e-05, "loss": 0.1304, "step": 25708 }, { "epoch": 18.834432234432235, "grad_norm": 1.1875, "learning_rate": 4.83557689439547e-05, "loss": 0.1162, "step": 25709 }, { "epoch": 18.835164835164836, "grad_norm": 1.21875, "learning_rate": 4.8352345803917626e-05, "loss": 0.1429, "step": 25710 }, { "epoch": 18.835897435897436, "grad_norm": 0.81640625, "learning_rate": 4.834892267417064e-05, "loss": 0.1147, "step": 25711 }, { "epoch": 18.836630036630037, "grad_norm": 0.96875, "learning_rate": 4.8345499554730195e-05, "loss": 0.1076, "step": 25712 }, { "epoch": 18.837362637362638, "grad_norm": 0.9140625, "learning_rate": 4.834207644561266e-05, "loss": 0.123, "step": 25713 }, { "epoch": 18.83809523809524, "grad_norm": 1.0078125, "learning_rate": 4.833865334683446e-05, "loss": 0.1065, "step": 25714 }, { "epoch": 18.83882783882784, "grad_norm": 1.0078125, "learning_rate": 4.833523025841198e-05, "loss": 0.0929, "step": 25715 }, { "epoch": 18.83956043956044, "grad_norm": 0.8671875, "learning_rate": 4.83318071803616e-05, "loss": 0.1231, "step": 25716 }, { "epoch": 18.84029304029304, "grad_norm": 1.40625, "learning_rate": 4.832838411269976e-05, "loss": 0.1094, "step": 25717 }, { "epoch": 18.84102564102564, "grad_norm": 0.93359375, "learning_rate": 4.832496105544285e-05, "loss": 0.1034, "step": 25718 }, { "epoch": 18.841758241758242, "grad_norm": 0.73828125, "learning_rate": 4.832153800860725e-05, "loss": 0.1137, "step": 25719 }, { "epoch": 18.842490842490843, "grad_norm": 1.046875, "learning_rate": 4.8318114972209386e-05, "loss": 0.1738, "step": 25720 }, { "epoch": 18.843223443223444, "grad_norm": 0.703125, "learning_rate": 4.831469194626564e-05, "loss": 0.1237, "step": 25721 }, { "epoch": 18.843956043956045, "grad_norm": 1.328125, "learning_rate": 4.831126893079242e-05, "loss": 0.127, "step": 25722 }, { "epoch": 18.844688644688645, "grad_norm": 1.3828125, "learning_rate": 4.830784592580615e-05, "loss": 0.1259, "step": 25723 }, { "epoch": 18.845421245421246, "grad_norm": 0.85546875, "learning_rate": 4.83044229313232e-05, "loss": 0.1397, "step": 25724 }, { "epoch": 18.846153846153847, "grad_norm": 0.58203125, "learning_rate": 4.830099994735999e-05, "loss": 0.1071, "step": 25725 }, { "epoch": 18.846886446886447, "grad_norm": 1.15625, "learning_rate": 4.8297576973932887e-05, "loss": 0.127, "step": 25726 }, { "epoch": 18.847619047619048, "grad_norm": 1.0703125, "learning_rate": 4.829415401105833e-05, "loss": 0.1066, "step": 25727 }, { "epoch": 18.84835164835165, "grad_norm": 0.859375, "learning_rate": 4.8290731058752705e-05, "loss": 0.1165, "step": 25728 }, { "epoch": 18.84908424908425, "grad_norm": 1.015625, "learning_rate": 4.828730811703242e-05, "loss": 0.162, "step": 25729 }, { "epoch": 18.84981684981685, "grad_norm": 1.3984375, "learning_rate": 4.828388518591386e-05, "loss": 0.129, "step": 25730 }, { "epoch": 18.85054945054945, "grad_norm": 1.4609375, "learning_rate": 4.8280462265413426e-05, "loss": 0.149, "step": 25731 }, { "epoch": 18.851282051282052, "grad_norm": 0.9140625, "learning_rate": 4.827703935554753e-05, "loss": 0.135, "step": 25732 }, { "epoch": 18.852014652014653, "grad_norm": 0.76953125, "learning_rate": 4.827361645633257e-05, "loss": 0.1028, "step": 25733 }, { "epoch": 18.852747252747253, "grad_norm": 1.171875, "learning_rate": 4.827019356778495e-05, "loss": 0.1316, "step": 25734 }, { "epoch": 18.853479853479854, "grad_norm": 1.03125, "learning_rate": 4.826677068992107e-05, "loss": 0.1489, "step": 25735 }, { "epoch": 18.854212454212455, "grad_norm": 0.984375, "learning_rate": 4.826334782275732e-05, "loss": 0.1244, "step": 25736 }, { "epoch": 18.854945054945055, "grad_norm": 2.03125, "learning_rate": 4.825992496631008e-05, "loss": 0.1542, "step": 25737 }, { "epoch": 18.855677655677656, "grad_norm": 0.75, "learning_rate": 4.82565021205958e-05, "loss": 0.1121, "step": 25738 }, { "epoch": 18.856410256410257, "grad_norm": 0.7265625, "learning_rate": 4.8253079285630856e-05, "loss": 0.1082, "step": 25739 }, { "epoch": 18.857142857142858, "grad_norm": 0.921875, "learning_rate": 4.8249656461431647e-05, "loss": 0.1239, "step": 25740 }, { "epoch": 18.85787545787546, "grad_norm": 1.1640625, "learning_rate": 4.8246233648014576e-05, "loss": 0.1154, "step": 25741 }, { "epoch": 18.85860805860806, "grad_norm": 0.90234375, "learning_rate": 4.824281084539601e-05, "loss": 0.1183, "step": 25742 }, { "epoch": 18.85934065934066, "grad_norm": 0.8828125, "learning_rate": 4.82393880535924e-05, "loss": 0.1223, "step": 25743 }, { "epoch": 18.86007326007326, "grad_norm": 1.203125, "learning_rate": 4.823596527262013e-05, "loss": 0.1268, "step": 25744 }, { "epoch": 18.86080586080586, "grad_norm": 1.0859375, "learning_rate": 4.8232542502495585e-05, "loss": 0.1135, "step": 25745 }, { "epoch": 18.861538461538462, "grad_norm": 1.1796875, "learning_rate": 4.822911974323517e-05, "loss": 0.104, "step": 25746 }, { "epoch": 18.862271062271063, "grad_norm": 1.390625, "learning_rate": 4.8225696994855284e-05, "loss": 0.1347, "step": 25747 }, { "epoch": 18.863003663003663, "grad_norm": 0.90234375, "learning_rate": 4.822227425737234e-05, "loss": 0.1207, "step": 25748 }, { "epoch": 18.863736263736264, "grad_norm": 1.2109375, "learning_rate": 4.821885153080273e-05, "loss": 0.141, "step": 25749 }, { "epoch": 18.864468864468865, "grad_norm": 1.4375, "learning_rate": 4.821542881516285e-05, "loss": 0.1369, "step": 25750 }, { "epoch": 18.865201465201466, "grad_norm": 1.703125, "learning_rate": 4.8212006110469103e-05, "loss": 0.1527, "step": 25751 }, { "epoch": 18.865934065934066, "grad_norm": 1.2265625, "learning_rate": 4.820858341673787e-05, "loss": 0.1362, "step": 25752 }, { "epoch": 18.866666666666667, "grad_norm": 1.0390625, "learning_rate": 4.820516073398558e-05, "loss": 0.1289, "step": 25753 }, { "epoch": 18.867399267399268, "grad_norm": 0.96484375, "learning_rate": 4.820173806222862e-05, "loss": 0.1283, "step": 25754 }, { "epoch": 18.86813186813187, "grad_norm": 2.265625, "learning_rate": 4.819831540148339e-05, "loss": 0.1463, "step": 25755 }, { "epoch": 18.86886446886447, "grad_norm": 1.5546875, "learning_rate": 4.819489275176627e-05, "loss": 0.1064, "step": 25756 }, { "epoch": 18.86959706959707, "grad_norm": 0.8125, "learning_rate": 4.819147011309369e-05, "loss": 0.1051, "step": 25757 }, { "epoch": 18.87032967032967, "grad_norm": 1.15625, "learning_rate": 4.8188047485482025e-05, "loss": 0.1139, "step": 25758 }, { "epoch": 18.87106227106227, "grad_norm": 0.97265625, "learning_rate": 4.8184624868947673e-05, "loss": 0.1209, "step": 25759 }, { "epoch": 18.871794871794872, "grad_norm": 1.0546875, "learning_rate": 4.818120226350706e-05, "loss": 0.1129, "step": 25760 }, { "epoch": 18.872527472527473, "grad_norm": 0.95703125, "learning_rate": 4.817777966917658e-05, "loss": 0.1159, "step": 25761 }, { "epoch": 18.873260073260074, "grad_norm": 1.4765625, "learning_rate": 4.817435708597261e-05, "loss": 0.1406, "step": 25762 }, { "epoch": 18.873992673992674, "grad_norm": 0.8828125, "learning_rate": 4.8170934513911535e-05, "loss": 0.1168, "step": 25763 }, { "epoch": 18.874725274725275, "grad_norm": 0.96484375, "learning_rate": 4.8167511953009806e-05, "loss": 0.1009, "step": 25764 }, { "epoch": 18.875457875457876, "grad_norm": 1.1953125, "learning_rate": 4.816408940328378e-05, "loss": 0.1363, "step": 25765 }, { "epoch": 18.876190476190477, "grad_norm": 0.76953125, "learning_rate": 4.816066686474989e-05, "loss": 0.1182, "step": 25766 }, { "epoch": 18.876923076923077, "grad_norm": 1.40625, "learning_rate": 4.815724433742449e-05, "loss": 0.1405, "step": 25767 }, { "epoch": 18.877655677655678, "grad_norm": 0.86328125, "learning_rate": 4.8153821821324e-05, "loss": 0.1183, "step": 25768 }, { "epoch": 18.87838827838828, "grad_norm": 1.078125, "learning_rate": 4.815039931646483e-05, "loss": 0.1179, "step": 25769 }, { "epoch": 18.87912087912088, "grad_norm": 1.375, "learning_rate": 4.814697682286336e-05, "loss": 0.103, "step": 25770 }, { "epoch": 18.87985347985348, "grad_norm": 1.5390625, "learning_rate": 4.8143554340536e-05, "loss": 0.1405, "step": 25771 }, { "epoch": 18.88058608058608, "grad_norm": 1.5078125, "learning_rate": 4.814013186949915e-05, "loss": 0.1104, "step": 25772 }, { "epoch": 18.88131868131868, "grad_norm": 1.2421875, "learning_rate": 4.8136709409769184e-05, "loss": 0.1288, "step": 25773 }, { "epoch": 18.882051282051282, "grad_norm": 0.85546875, "learning_rate": 4.813328696136254e-05, "loss": 0.1374, "step": 25774 }, { "epoch": 18.882783882783883, "grad_norm": 0.8046875, "learning_rate": 4.812986452429559e-05, "loss": 0.1187, "step": 25775 }, { "epoch": 18.883516483516484, "grad_norm": 1.125, "learning_rate": 4.812644209858474e-05, "loss": 0.1124, "step": 25776 }, { "epoch": 18.884249084249085, "grad_norm": 1.5390625, "learning_rate": 4.812301968424639e-05, "loss": 0.148, "step": 25777 }, { "epoch": 18.884981684981685, "grad_norm": 0.72265625, "learning_rate": 4.811959728129691e-05, "loss": 0.1339, "step": 25778 }, { "epoch": 18.885714285714286, "grad_norm": 0.9453125, "learning_rate": 4.8116174889752736e-05, "loss": 0.1317, "step": 25779 }, { "epoch": 18.886446886446887, "grad_norm": 1.1953125, "learning_rate": 4.8112752509630254e-05, "loss": 0.1587, "step": 25780 }, { "epoch": 18.887179487179488, "grad_norm": 0.9921875, "learning_rate": 4.8109330140945855e-05, "loss": 0.1226, "step": 25781 }, { "epoch": 18.88791208791209, "grad_norm": 0.9296875, "learning_rate": 4.8105907783715936e-05, "loss": 0.1158, "step": 25782 }, { "epoch": 18.88864468864469, "grad_norm": 1.3125, "learning_rate": 4.810248543795691e-05, "loss": 0.1383, "step": 25783 }, { "epoch": 18.88937728937729, "grad_norm": 1.1484375, "learning_rate": 4.809906310368515e-05, "loss": 0.1362, "step": 25784 }, { "epoch": 18.89010989010989, "grad_norm": 1.0859375, "learning_rate": 4.809564078091707e-05, "loss": 0.1196, "step": 25785 }, { "epoch": 18.89084249084249, "grad_norm": 1.2578125, "learning_rate": 4.8092218469669065e-05, "loss": 0.1458, "step": 25786 }, { "epoch": 18.891575091575092, "grad_norm": 1.0, "learning_rate": 4.808879616995754e-05, "loss": 0.1268, "step": 25787 }, { "epoch": 18.892307692307693, "grad_norm": 0.859375, "learning_rate": 4.808537388179888e-05, "loss": 0.111, "step": 25788 }, { "epoch": 18.893040293040293, "grad_norm": 0.8515625, "learning_rate": 4.808195160520946e-05, "loss": 0.1175, "step": 25789 }, { "epoch": 18.893772893772894, "grad_norm": 1.59375, "learning_rate": 4.807852934020574e-05, "loss": 0.1362, "step": 25790 }, { "epoch": 18.894505494505495, "grad_norm": 1.421875, "learning_rate": 4.8075107086804074e-05, "loss": 0.137, "step": 25791 }, { "epoch": 18.895238095238096, "grad_norm": 1.171875, "learning_rate": 4.807168484502086e-05, "loss": 0.1569, "step": 25792 }, { "epoch": 18.895970695970696, "grad_norm": 1.09375, "learning_rate": 4.806826261487251e-05, "loss": 0.1012, "step": 25793 }, { "epoch": 18.896703296703297, "grad_norm": 0.6328125, "learning_rate": 4.806484039637539e-05, "loss": 0.131, "step": 25794 }, { "epoch": 18.897435897435898, "grad_norm": 1.046875, "learning_rate": 4.806141818954594e-05, "loss": 0.1252, "step": 25795 }, { "epoch": 18.8981684981685, "grad_norm": 0.99609375, "learning_rate": 4.805799599440052e-05, "loss": 0.1076, "step": 25796 }, { "epoch": 18.8989010989011, "grad_norm": 0.6796875, "learning_rate": 4.8054573810955556e-05, "loss": 0.1218, "step": 25797 }, { "epoch": 18.8996336996337, "grad_norm": 1.0703125, "learning_rate": 4.805115163922743e-05, "loss": 0.1344, "step": 25798 }, { "epoch": 18.9003663003663, "grad_norm": 1.28125, "learning_rate": 4.804772947923252e-05, "loss": 0.1242, "step": 25799 }, { "epoch": 18.9010989010989, "grad_norm": 0.734375, "learning_rate": 4.804430733098727e-05, "loss": 0.1014, "step": 25800 }, { "epoch": 18.901831501831502, "grad_norm": 1.3046875, "learning_rate": 4.804088519450804e-05, "loss": 0.1062, "step": 25801 }, { "epoch": 18.902564102564103, "grad_norm": 1.3046875, "learning_rate": 4.8037463069811243e-05, "loss": 0.1336, "step": 25802 }, { "epoch": 18.903296703296704, "grad_norm": 0.8828125, "learning_rate": 4.803404095691327e-05, "loss": 0.1266, "step": 25803 }, { "epoch": 18.904029304029304, "grad_norm": 0.69921875, "learning_rate": 4.8030618855830485e-05, "loss": 0.1061, "step": 25804 }, { "epoch": 18.904761904761905, "grad_norm": 1.0390625, "learning_rate": 4.802719676657935e-05, "loss": 0.1263, "step": 25805 }, { "epoch": 18.905494505494506, "grad_norm": 1.015625, "learning_rate": 4.802377468917622e-05, "loss": 0.1389, "step": 25806 }, { "epoch": 18.906227106227107, "grad_norm": 1.046875, "learning_rate": 4.80203526236375e-05, "loss": 0.1532, "step": 25807 }, { "epoch": 18.906959706959707, "grad_norm": 1.125, "learning_rate": 4.8016930569979576e-05, "loss": 0.1064, "step": 25808 }, { "epoch": 18.907692307692308, "grad_norm": 1.0546875, "learning_rate": 4.801350852821886e-05, "loss": 0.1099, "step": 25809 }, { "epoch": 18.90842490842491, "grad_norm": 0.83984375, "learning_rate": 4.801008649837173e-05, "loss": 0.1104, "step": 25810 }, { "epoch": 18.90915750915751, "grad_norm": 1.1171875, "learning_rate": 4.8006664480454595e-05, "loss": 0.1347, "step": 25811 }, { "epoch": 18.90989010989011, "grad_norm": 0.80859375, "learning_rate": 4.800324247448386e-05, "loss": 0.1112, "step": 25812 }, { "epoch": 18.91062271062271, "grad_norm": 0.88671875, "learning_rate": 4.7999820480475904e-05, "loss": 0.1064, "step": 25813 }, { "epoch": 18.91135531135531, "grad_norm": 0.93359375, "learning_rate": 4.799639849844714e-05, "loss": 0.1104, "step": 25814 }, { "epoch": 18.912087912087912, "grad_norm": 0.93359375, "learning_rate": 4.799297652841392e-05, "loss": 0.1003, "step": 25815 }, { "epoch": 18.912820512820513, "grad_norm": 1.484375, "learning_rate": 4.79895545703927e-05, "loss": 0.1383, "step": 25816 }, { "epoch": 18.913553113553114, "grad_norm": 1.3515625, "learning_rate": 4.798613262439984e-05, "loss": 0.1187, "step": 25817 }, { "epoch": 18.914285714285715, "grad_norm": 1.09375, "learning_rate": 4.798271069045176e-05, "loss": 0.1242, "step": 25818 }, { "epoch": 18.915018315018315, "grad_norm": 1.0546875, "learning_rate": 4.797928876856482e-05, "loss": 0.0989, "step": 25819 }, { "epoch": 18.915750915750916, "grad_norm": 1.234375, "learning_rate": 4.7975866858755425e-05, "loss": 0.111, "step": 25820 }, { "epoch": 18.916483516483517, "grad_norm": 0.90234375, "learning_rate": 4.7972444961040006e-05, "loss": 0.1074, "step": 25821 }, { "epoch": 18.917216117216118, "grad_norm": 0.8515625, "learning_rate": 4.7969023075434915e-05, "loss": 0.1294, "step": 25822 }, { "epoch": 18.91794871794872, "grad_norm": 1.40625, "learning_rate": 4.796560120195658e-05, "loss": 0.1317, "step": 25823 }, { "epoch": 18.91868131868132, "grad_norm": 0.87109375, "learning_rate": 4.796217934062138e-05, "loss": 0.1165, "step": 25824 }, { "epoch": 18.91941391941392, "grad_norm": 1.609375, "learning_rate": 4.795875749144569e-05, "loss": 0.1113, "step": 25825 }, { "epoch": 18.92014652014652, "grad_norm": 1.203125, "learning_rate": 4.7955335654445946e-05, "loss": 0.1319, "step": 25826 }, { "epoch": 18.92087912087912, "grad_norm": 1.9453125, "learning_rate": 4.795191382963853e-05, "loss": 0.1079, "step": 25827 }, { "epoch": 18.921611721611722, "grad_norm": 2.4375, "learning_rate": 4.7948492017039824e-05, "loss": 0.1186, "step": 25828 }, { "epoch": 18.922344322344323, "grad_norm": 1.1171875, "learning_rate": 4.794507021666624e-05, "loss": 0.1497, "step": 25829 }, { "epoch": 18.923076923076923, "grad_norm": 1.3828125, "learning_rate": 4.794164842853413e-05, "loss": 0.1492, "step": 25830 }, { "epoch": 18.923809523809524, "grad_norm": 1.609375, "learning_rate": 4.793822665265996e-05, "loss": 0.117, "step": 25831 }, { "epoch": 18.924542124542125, "grad_norm": 1.515625, "learning_rate": 4.7934804889060064e-05, "loss": 0.092, "step": 25832 }, { "epoch": 18.925274725274726, "grad_norm": 1.640625, "learning_rate": 4.793138313775088e-05, "loss": 0.1321, "step": 25833 }, { "epoch": 18.926007326007326, "grad_norm": 0.99609375, "learning_rate": 4.792796139874877e-05, "loss": 0.1257, "step": 25834 }, { "epoch": 18.926739926739927, "grad_norm": 1.421875, "learning_rate": 4.792453967207014e-05, "loss": 0.1375, "step": 25835 }, { "epoch": 18.927472527472528, "grad_norm": 0.828125, "learning_rate": 4.792111795773138e-05, "loss": 0.1109, "step": 25836 }, { "epoch": 18.92820512820513, "grad_norm": 0.72265625, "learning_rate": 4.791769625574889e-05, "loss": 0.114, "step": 25837 }, { "epoch": 18.92893772893773, "grad_norm": 0.65234375, "learning_rate": 4.791427456613908e-05, "loss": 0.1071, "step": 25838 }, { "epoch": 18.92967032967033, "grad_norm": 1.2109375, "learning_rate": 4.791085288891833e-05, "loss": 0.1145, "step": 25839 }, { "epoch": 18.93040293040293, "grad_norm": 1.3203125, "learning_rate": 4.790743122410303e-05, "loss": 0.1246, "step": 25840 }, { "epoch": 18.93113553113553, "grad_norm": 1.453125, "learning_rate": 4.790400957170956e-05, "loss": 0.114, "step": 25841 }, { "epoch": 18.931868131868132, "grad_norm": 1.3515625, "learning_rate": 4.7900587931754346e-05, "loss": 0.1668, "step": 25842 }, { "epoch": 18.932600732600733, "grad_norm": 1.140625, "learning_rate": 4.789716630425378e-05, "loss": 0.1297, "step": 25843 }, { "epoch": 18.933333333333334, "grad_norm": 0.6796875, "learning_rate": 4.7893744689224244e-05, "loss": 0.1204, "step": 25844 }, { "epoch": 18.934065934065934, "grad_norm": 0.94921875, "learning_rate": 4.789032308668211e-05, "loss": 0.1015, "step": 25845 }, { "epoch": 18.934798534798535, "grad_norm": 0.96875, "learning_rate": 4.7886901496643796e-05, "loss": 0.1221, "step": 25846 }, { "epoch": 18.935531135531136, "grad_norm": 1.0234375, "learning_rate": 4.7883479919125716e-05, "loss": 0.1231, "step": 25847 }, { "epoch": 18.936263736263736, "grad_norm": 1.375, "learning_rate": 4.788005835414423e-05, "loss": 0.1083, "step": 25848 }, { "epoch": 18.936996336996337, "grad_norm": 1.171875, "learning_rate": 4.787663680171575e-05, "loss": 0.1114, "step": 25849 }, { "epoch": 18.937728937728938, "grad_norm": 1.625, "learning_rate": 4.7873215261856667e-05, "loss": 0.133, "step": 25850 }, { "epoch": 18.93846153846154, "grad_norm": 0.67578125, "learning_rate": 4.786979373458335e-05, "loss": 0.1153, "step": 25851 }, { "epoch": 18.93919413919414, "grad_norm": 1.3515625, "learning_rate": 4.786637221991224e-05, "loss": 0.1074, "step": 25852 }, { "epoch": 18.93992673992674, "grad_norm": 1.296875, "learning_rate": 4.78629507178597e-05, "loss": 0.117, "step": 25853 }, { "epoch": 18.94065934065934, "grad_norm": 1.2109375, "learning_rate": 4.785952922844213e-05, "loss": 0.1658, "step": 25854 }, { "epoch": 18.94139194139194, "grad_norm": 0.8515625, "learning_rate": 4.785610775167592e-05, "loss": 0.0932, "step": 25855 }, { "epoch": 18.942124542124542, "grad_norm": 1.015625, "learning_rate": 4.7852686287577455e-05, "loss": 0.108, "step": 25856 }, { "epoch": 18.942857142857143, "grad_norm": 1.3203125, "learning_rate": 4.784926483616315e-05, "loss": 0.1317, "step": 25857 }, { "epoch": 18.943589743589744, "grad_norm": 1.2109375, "learning_rate": 4.784584339744939e-05, "loss": 0.1142, "step": 25858 }, { "epoch": 18.944322344322345, "grad_norm": 0.64453125, "learning_rate": 4.7842421971452565e-05, "loss": 0.1055, "step": 25859 }, { "epoch": 18.945054945054945, "grad_norm": 0.8125, "learning_rate": 4.7839000558189055e-05, "loss": 0.1139, "step": 25860 }, { "epoch": 18.945787545787546, "grad_norm": 1.1640625, "learning_rate": 4.783557915767528e-05, "loss": 0.1326, "step": 25861 }, { "epoch": 18.946520146520147, "grad_norm": 1.6328125, "learning_rate": 4.783215776992761e-05, "loss": 0.1411, "step": 25862 }, { "epoch": 18.947252747252747, "grad_norm": 1.0234375, "learning_rate": 4.7828736394962454e-05, "loss": 0.1211, "step": 25863 }, { "epoch": 18.947985347985348, "grad_norm": 0.92578125, "learning_rate": 4.78253150327962e-05, "loss": 0.128, "step": 25864 }, { "epoch": 18.94871794871795, "grad_norm": 1.28125, "learning_rate": 4.782189368344525e-05, "loss": 0.147, "step": 25865 }, { "epoch": 18.94945054945055, "grad_norm": 1.125, "learning_rate": 4.781847234692597e-05, "loss": 0.127, "step": 25866 }, { "epoch": 18.95018315018315, "grad_norm": 0.99609375, "learning_rate": 4.781505102325476e-05, "loss": 0.1189, "step": 25867 }, { "epoch": 18.95091575091575, "grad_norm": 1.5390625, "learning_rate": 4.781162971244804e-05, "loss": 0.1339, "step": 25868 }, { "epoch": 18.951648351648352, "grad_norm": 1.0703125, "learning_rate": 4.780820841452219e-05, "loss": 0.1388, "step": 25869 }, { "epoch": 18.952380952380953, "grad_norm": 1.1875, "learning_rate": 4.7804787129493595e-05, "loss": 0.1096, "step": 25870 }, { "epoch": 18.953113553113553, "grad_norm": 0.73046875, "learning_rate": 4.780136585737863e-05, "loss": 0.0911, "step": 25871 }, { "epoch": 18.953846153846154, "grad_norm": 0.91796875, "learning_rate": 4.7797944598193714e-05, "loss": 0.127, "step": 25872 }, { "epoch": 18.954578754578755, "grad_norm": 0.953125, "learning_rate": 4.779452335195525e-05, "loss": 0.1161, "step": 25873 }, { "epoch": 18.955311355311355, "grad_norm": 1.03125, "learning_rate": 4.77911021186796e-05, "loss": 0.1307, "step": 25874 }, { "epoch": 18.956043956043956, "grad_norm": 0.86328125, "learning_rate": 4.778768089838317e-05, "loss": 0.1342, "step": 25875 }, { "epoch": 18.956776556776557, "grad_norm": 0.9921875, "learning_rate": 4.7784259691082354e-05, "loss": 0.1508, "step": 25876 }, { "epoch": 18.957509157509158, "grad_norm": 0.74609375, "learning_rate": 4.778083849679352e-05, "loss": 0.1203, "step": 25877 }, { "epoch": 18.95824175824176, "grad_norm": 1.1875, "learning_rate": 4.777741731553311e-05, "loss": 0.1452, "step": 25878 }, { "epoch": 18.95897435897436, "grad_norm": 1.375, "learning_rate": 4.777399614731748e-05, "loss": 0.1291, "step": 25879 }, { "epoch": 18.95970695970696, "grad_norm": 0.74609375, "learning_rate": 4.777057499216304e-05, "loss": 0.1265, "step": 25880 }, { "epoch": 18.96043956043956, "grad_norm": 0.890625, "learning_rate": 4.776715385008615e-05, "loss": 0.1175, "step": 25881 }, { "epoch": 18.96117216117216, "grad_norm": 1.265625, "learning_rate": 4.776373272110322e-05, "loss": 0.1162, "step": 25882 }, { "epoch": 18.961904761904762, "grad_norm": 1.3203125, "learning_rate": 4.776031160523066e-05, "loss": 0.122, "step": 25883 }, { "epoch": 18.962637362637363, "grad_norm": 0.8828125, "learning_rate": 4.775689050248485e-05, "loss": 0.1325, "step": 25884 }, { "epoch": 18.963369963369964, "grad_norm": 1.109375, "learning_rate": 4.7753469412882176e-05, "loss": 0.1326, "step": 25885 }, { "epoch": 18.964102564102564, "grad_norm": 1.265625, "learning_rate": 4.775004833643901e-05, "loss": 0.1225, "step": 25886 }, { "epoch": 18.964835164835165, "grad_norm": 1.734375, "learning_rate": 4.774662727317179e-05, "loss": 0.1471, "step": 25887 }, { "epoch": 18.965567765567766, "grad_norm": 0.9375, "learning_rate": 4.774320622309686e-05, "loss": 0.1015, "step": 25888 }, { "epoch": 18.966300366300366, "grad_norm": 1.0625, "learning_rate": 4.7739785186230655e-05, "loss": 0.1068, "step": 25889 }, { "epoch": 18.967032967032967, "grad_norm": 1.734375, "learning_rate": 4.773636416258954e-05, "loss": 0.1373, "step": 25890 }, { "epoch": 18.967765567765568, "grad_norm": 1.2734375, "learning_rate": 4.7732943152189916e-05, "loss": 0.1281, "step": 25891 }, { "epoch": 18.96849816849817, "grad_norm": 1.1171875, "learning_rate": 4.7729522155048166e-05, "loss": 0.1079, "step": 25892 }, { "epoch": 18.96923076923077, "grad_norm": 0.62890625, "learning_rate": 4.772610117118066e-05, "loss": 0.1149, "step": 25893 }, { "epoch": 18.96996336996337, "grad_norm": 1.3125, "learning_rate": 4.772268020060385e-05, "loss": 0.1113, "step": 25894 }, { "epoch": 18.97069597069597, "grad_norm": 1.265625, "learning_rate": 4.771925924333408e-05, "loss": 0.1192, "step": 25895 }, { "epoch": 18.97142857142857, "grad_norm": 0.984375, "learning_rate": 4.771583829938776e-05, "loss": 0.171, "step": 25896 }, { "epoch": 18.972161172161172, "grad_norm": 1.296875, "learning_rate": 4.771241736878125e-05, "loss": 0.1102, "step": 25897 }, { "epoch": 18.972893772893773, "grad_norm": 1.078125, "learning_rate": 4.7708996451530966e-05, "loss": 0.1291, "step": 25898 }, { "epoch": 18.973626373626374, "grad_norm": 0.8828125, "learning_rate": 4.770557554765331e-05, "loss": 0.1144, "step": 25899 }, { "epoch": 18.974358974358974, "grad_norm": 1.3515625, "learning_rate": 4.770215465716466e-05, "loss": 0.1251, "step": 25900 }, { "epoch": 18.975091575091575, "grad_norm": 1.25, "learning_rate": 4.76987337800814e-05, "loss": 0.1287, "step": 25901 }, { "epoch": 18.975824175824176, "grad_norm": 0.9765625, "learning_rate": 4.7695312916419925e-05, "loss": 0.1066, "step": 25902 }, { "epoch": 18.976556776556777, "grad_norm": 0.75, "learning_rate": 4.7691892066196616e-05, "loss": 0.1205, "step": 25903 }, { "epoch": 18.977289377289377, "grad_norm": 0.9609375, "learning_rate": 4.768847122942788e-05, "loss": 0.1054, "step": 25904 }, { "epoch": 18.978021978021978, "grad_norm": 0.84375, "learning_rate": 4.7685050406130125e-05, "loss": 0.1157, "step": 25905 }, { "epoch": 18.97875457875458, "grad_norm": 1.3046875, "learning_rate": 4.76816295963197e-05, "loss": 0.1797, "step": 25906 }, { "epoch": 18.97948717948718, "grad_norm": 1.640625, "learning_rate": 4.767820880001301e-05, "loss": 0.1121, "step": 25907 }, { "epoch": 18.98021978021978, "grad_norm": 1.46875, "learning_rate": 4.7674788017226434e-05, "loss": 0.1126, "step": 25908 }, { "epoch": 18.98095238095238, "grad_norm": 0.75, "learning_rate": 4.767136724797639e-05, "loss": 0.0915, "step": 25909 }, { "epoch": 18.98168498168498, "grad_norm": 1.109375, "learning_rate": 4.766794649227927e-05, "loss": 0.1138, "step": 25910 }, { "epoch": 18.982417582417582, "grad_norm": 1.171875, "learning_rate": 4.7664525750151425e-05, "loss": 0.1349, "step": 25911 }, { "epoch": 18.983150183150183, "grad_norm": 1.0078125, "learning_rate": 4.7661105021609285e-05, "loss": 0.139, "step": 25912 }, { "epoch": 18.983882783882784, "grad_norm": 0.80078125, "learning_rate": 4.765768430666922e-05, "loss": 0.1179, "step": 25913 }, { "epoch": 18.984615384615385, "grad_norm": 1.5625, "learning_rate": 4.76542636053476e-05, "loss": 0.1286, "step": 25914 }, { "epoch": 18.985347985347985, "grad_norm": 0.89453125, "learning_rate": 4.765084291766086e-05, "loss": 0.1198, "step": 25915 }, { "epoch": 18.986080586080586, "grad_norm": 1.7578125, "learning_rate": 4.7647422243625365e-05, "loss": 0.1106, "step": 25916 }, { "epoch": 18.986813186813187, "grad_norm": 1.6015625, "learning_rate": 4.764400158325751e-05, "loss": 0.1645, "step": 25917 }, { "epoch": 18.987545787545788, "grad_norm": 1.6484375, "learning_rate": 4.764058093657368e-05, "loss": 0.1316, "step": 25918 }, { "epoch": 18.98827838827839, "grad_norm": 0.6875, "learning_rate": 4.763716030359024e-05, "loss": 0.0986, "step": 25919 }, { "epoch": 18.98901098901099, "grad_norm": 1.4296875, "learning_rate": 4.763373968432363e-05, "loss": 0.1317, "step": 25920 }, { "epoch": 18.98974358974359, "grad_norm": 1.25, "learning_rate": 4.7630319078790214e-05, "loss": 0.1177, "step": 25921 }, { "epoch": 18.99047619047619, "grad_norm": 1.4140625, "learning_rate": 4.762689848700638e-05, "loss": 0.1318, "step": 25922 }, { "epoch": 18.99120879120879, "grad_norm": 0.76171875, "learning_rate": 4.762347790898851e-05, "loss": 0.117, "step": 25923 }, { "epoch": 18.991941391941392, "grad_norm": 0.63671875, "learning_rate": 4.7620057344753e-05, "loss": 0.1034, "step": 25924 }, { "epoch": 18.992673992673993, "grad_norm": 1.296875, "learning_rate": 4.761663679431625e-05, "loss": 0.1129, "step": 25925 }, { "epoch": 18.993406593406593, "grad_norm": 1.0234375, "learning_rate": 4.761321625769463e-05, "loss": 0.1082, "step": 25926 }, { "epoch": 18.994139194139194, "grad_norm": 1.4140625, "learning_rate": 4.760979573490455e-05, "loss": 0.1164, "step": 25927 }, { "epoch": 18.994871794871795, "grad_norm": 1.8125, "learning_rate": 4.760637522596239e-05, "loss": 0.1306, "step": 25928 }, { "epoch": 18.995604395604396, "grad_norm": 1.40625, "learning_rate": 4.7602954730884516e-05, "loss": 0.1101, "step": 25929 }, { "epoch": 18.996336996336996, "grad_norm": 1.359375, "learning_rate": 4.759953424968735e-05, "loss": 0.1136, "step": 25930 }, { "epoch": 18.997069597069597, "grad_norm": 1.046875, "learning_rate": 4.759611378238727e-05, "loss": 0.1091, "step": 25931 }, { "epoch": 18.997802197802198, "grad_norm": 0.78515625, "learning_rate": 4.7592693329000665e-05, "loss": 0.11, "step": 25932 }, { "epoch": 18.9985347985348, "grad_norm": 1.203125, "learning_rate": 4.7589272889543924e-05, "loss": 0.1188, "step": 25933 }, { "epoch": 18.9992673992674, "grad_norm": 1.2578125, "learning_rate": 4.7585852464033403e-05, "loss": 0.1523, "step": 25934 }, { "epoch": 19.0, "grad_norm": 1.3828125, "learning_rate": 4.7582432052485544e-05, "loss": 0.138, "step": 25935 }, { "epoch": 19.0007326007326, "grad_norm": 1.5703125, "learning_rate": 4.757901165491671e-05, "loss": 0.1595, "step": 25936 }, { "epoch": 19.0014652014652, "grad_norm": 1.03125, "learning_rate": 4.757559127134328e-05, "loss": 0.1474, "step": 25937 }, { "epoch": 19.002197802197802, "grad_norm": 1.03125, "learning_rate": 4.757217090178166e-05, "loss": 0.1549, "step": 25938 }, { "epoch": 19.002930402930403, "grad_norm": 0.87109375, "learning_rate": 4.7568750546248235e-05, "loss": 0.1213, "step": 25939 }, { "epoch": 19.003663003663004, "grad_norm": 1.671875, "learning_rate": 4.7565330204759365e-05, "loss": 0.1199, "step": 25940 }, { "epoch": 19.004395604395604, "grad_norm": 1.15625, "learning_rate": 4.7561909877331486e-05, "loss": 0.1049, "step": 25941 }, { "epoch": 19.005128205128205, "grad_norm": 0.66796875, "learning_rate": 4.7558489563980956e-05, "loss": 0.1271, "step": 25942 }, { "epoch": 19.005860805860806, "grad_norm": 1.015625, "learning_rate": 4.755506926472417e-05, "loss": 0.1096, "step": 25943 }, { "epoch": 19.006593406593407, "grad_norm": 0.99609375, "learning_rate": 4.7551648979577515e-05, "loss": 0.1057, "step": 25944 }, { "epoch": 19.007326007326007, "grad_norm": 1.125, "learning_rate": 4.7548228708557355e-05, "loss": 0.1628, "step": 25945 }, { "epoch": 19.008058608058608, "grad_norm": 1.7890625, "learning_rate": 4.754480845168013e-05, "loss": 0.1308, "step": 25946 }, { "epoch": 19.00879120879121, "grad_norm": 0.99609375, "learning_rate": 4.754138820896219e-05, "loss": 0.1181, "step": 25947 }, { "epoch": 19.00952380952381, "grad_norm": 1.171875, "learning_rate": 4.753796798041993e-05, "loss": 0.1092, "step": 25948 }, { "epoch": 19.01025641025641, "grad_norm": 1.09375, "learning_rate": 4.753454776606973e-05, "loss": 0.1182, "step": 25949 }, { "epoch": 19.01098901098901, "grad_norm": 1.6640625, "learning_rate": 4.753112756592798e-05, "loss": 0.1394, "step": 25950 }, { "epoch": 19.01172161172161, "grad_norm": 1.34375, "learning_rate": 4.752770738001108e-05, "loss": 0.1099, "step": 25951 }, { "epoch": 19.012454212454212, "grad_norm": 1.4296875, "learning_rate": 4.752428720833541e-05, "loss": 0.1515, "step": 25952 }, { "epoch": 19.013186813186813, "grad_norm": 1.5859375, "learning_rate": 4.7520867050917364e-05, "loss": 0.1249, "step": 25953 }, { "epoch": 19.013919413919414, "grad_norm": 1.5859375, "learning_rate": 4.751744690777331e-05, "loss": 0.1178, "step": 25954 }, { "epoch": 19.014652014652015, "grad_norm": 0.9765625, "learning_rate": 4.7514026778919644e-05, "loss": 0.1068, "step": 25955 }, { "epoch": 19.015384615384615, "grad_norm": 0.828125, "learning_rate": 4.751060666437277e-05, "loss": 0.1311, "step": 25956 }, { "epoch": 19.016117216117216, "grad_norm": 0.81640625, "learning_rate": 4.750718656414906e-05, "loss": 0.1338, "step": 25957 }, { "epoch": 19.016849816849817, "grad_norm": 1.1171875, "learning_rate": 4.750376647826491e-05, "loss": 0.1123, "step": 25958 }, { "epoch": 19.017582417582418, "grad_norm": 0.96484375, "learning_rate": 4.7500346406736686e-05, "loss": 0.1146, "step": 25959 }, { "epoch": 19.01831501831502, "grad_norm": 0.97265625, "learning_rate": 4.749692634958078e-05, "loss": 0.1272, "step": 25960 }, { "epoch": 19.01904761904762, "grad_norm": 1.953125, "learning_rate": 4.749350630681359e-05, "loss": 0.1315, "step": 25961 }, { "epoch": 19.01978021978022, "grad_norm": 1.3125, "learning_rate": 4.74900862784515e-05, "loss": 0.1014, "step": 25962 }, { "epoch": 19.02051282051282, "grad_norm": 1.3828125, "learning_rate": 4.7486666264510894e-05, "loss": 0.1063, "step": 25963 }, { "epoch": 19.02124542124542, "grad_norm": 1.1171875, "learning_rate": 4.748324626500816e-05, "loss": 0.1302, "step": 25964 }, { "epoch": 19.021978021978022, "grad_norm": 1.1015625, "learning_rate": 4.747982627995969e-05, "loss": 0.1154, "step": 25965 }, { "epoch": 19.022710622710623, "grad_norm": 0.90234375, "learning_rate": 4.747640630938184e-05, "loss": 0.1085, "step": 25966 }, { "epoch": 19.023443223443223, "grad_norm": 1.0625, "learning_rate": 4.7472986353291045e-05, "loss": 0.1171, "step": 25967 }, { "epoch": 19.024175824175824, "grad_norm": 1.0390625, "learning_rate": 4.746956641170366e-05, "loss": 0.1216, "step": 25968 }, { "epoch": 19.024908424908425, "grad_norm": 1.65625, "learning_rate": 4.746614648463608e-05, "loss": 0.1369, "step": 25969 }, { "epoch": 19.025641025641026, "grad_norm": 1.2265625, "learning_rate": 4.7462726572104694e-05, "loss": 0.1163, "step": 25970 }, { "epoch": 19.026373626373626, "grad_norm": 0.78125, "learning_rate": 4.745930667412585e-05, "loss": 0.1312, "step": 25971 }, { "epoch": 19.027106227106227, "grad_norm": 1.1953125, "learning_rate": 4.7455886790716e-05, "loss": 0.1181, "step": 25972 }, { "epoch": 19.027838827838828, "grad_norm": 0.84375, "learning_rate": 4.7452466921891483e-05, "loss": 0.1034, "step": 25973 }, { "epoch": 19.02857142857143, "grad_norm": 0.86328125, "learning_rate": 4.744904706766871e-05, "loss": 0.1185, "step": 25974 }, { "epoch": 19.02930402930403, "grad_norm": 0.78515625, "learning_rate": 4.7445627228064033e-05, "loss": 0.1111, "step": 25975 }, { "epoch": 19.03003663003663, "grad_norm": 0.85546875, "learning_rate": 4.744220740309386e-05, "loss": 0.0936, "step": 25976 }, { "epoch": 19.03076923076923, "grad_norm": 1.5703125, "learning_rate": 4.743878759277458e-05, "loss": 0.116, "step": 25977 }, { "epoch": 19.03150183150183, "grad_norm": 1.0390625, "learning_rate": 4.743536779712258e-05, "loss": 0.1362, "step": 25978 }, { "epoch": 19.032234432234432, "grad_norm": 0.81640625, "learning_rate": 4.743194801615424e-05, "loss": 0.1106, "step": 25979 }, { "epoch": 19.032967032967033, "grad_norm": 1.0234375, "learning_rate": 4.742852824988595e-05, "loss": 0.1308, "step": 25980 }, { "epoch": 19.033699633699634, "grad_norm": 1.3359375, "learning_rate": 4.7425108498334057e-05, "loss": 0.1101, "step": 25981 }, { "epoch": 19.034432234432234, "grad_norm": 1.03125, "learning_rate": 4.7421688761515004e-05, "loss": 0.1431, "step": 25982 }, { "epoch": 19.035164835164835, "grad_norm": 1.7421875, "learning_rate": 4.7418269039445154e-05, "loss": 0.1599, "step": 25983 }, { "epoch": 19.035897435897436, "grad_norm": 0.76171875, "learning_rate": 4.741484933214089e-05, "loss": 0.1063, "step": 25984 }, { "epoch": 19.036630036630036, "grad_norm": 0.8359375, "learning_rate": 4.7411429639618584e-05, "loss": 0.1164, "step": 25985 }, { "epoch": 19.037362637362637, "grad_norm": 1.0078125, "learning_rate": 4.740800996189464e-05, "loss": 0.1246, "step": 25986 }, { "epoch": 19.038095238095238, "grad_norm": 1.3671875, "learning_rate": 4.740459029898542e-05, "loss": 0.1298, "step": 25987 }, { "epoch": 19.03882783882784, "grad_norm": 1.5, "learning_rate": 4.740117065090735e-05, "loss": 0.1439, "step": 25988 }, { "epoch": 19.03956043956044, "grad_norm": 1.3203125, "learning_rate": 4.7397751017676766e-05, "loss": 0.1151, "step": 25989 }, { "epoch": 19.04029304029304, "grad_norm": 0.84375, "learning_rate": 4.739433139931009e-05, "loss": 0.1264, "step": 25990 }, { "epoch": 19.04102564102564, "grad_norm": 1.0859375, "learning_rate": 4.739091179582369e-05, "loss": 0.1471, "step": 25991 }, { "epoch": 19.04175824175824, "grad_norm": 0.9375, "learning_rate": 4.738749220723393e-05, "loss": 0.1073, "step": 25992 }, { "epoch": 19.042490842490842, "grad_norm": 1.109375, "learning_rate": 4.7384072633557235e-05, "loss": 0.1385, "step": 25993 }, { "epoch": 19.043223443223443, "grad_norm": 1.34375, "learning_rate": 4.7380653074809974e-05, "loss": 0.1378, "step": 25994 }, { "epoch": 19.043956043956044, "grad_norm": 1.265625, "learning_rate": 4.7377233531008534e-05, "loss": 0.1362, "step": 25995 }, { "epoch": 19.044688644688645, "grad_norm": 1.296875, "learning_rate": 4.737381400216929e-05, "loss": 0.1265, "step": 25996 }, { "epoch": 19.045421245421245, "grad_norm": 0.9921875, "learning_rate": 4.73703944883086e-05, "loss": 0.1393, "step": 25997 }, { "epoch": 19.046153846153846, "grad_norm": 1.25, "learning_rate": 4.736697498944291e-05, "loss": 0.1188, "step": 25998 }, { "epoch": 19.046886446886447, "grad_norm": 1.1484375, "learning_rate": 4.7363555505588564e-05, "loss": 0.135, "step": 25999 }, { "epoch": 19.047619047619047, "grad_norm": 1.5234375, "learning_rate": 4.736013603676196e-05, "loss": 0.1679, "step": 26000 }, { "epoch": 19.048351648351648, "grad_norm": 0.828125, "learning_rate": 4.7356716582979456e-05, "loss": 0.1073, "step": 26001 }, { "epoch": 19.04908424908425, "grad_norm": 1.703125, "learning_rate": 4.735329714425746e-05, "loss": 0.1192, "step": 26002 }, { "epoch": 19.04981684981685, "grad_norm": 0.99609375, "learning_rate": 4.7349877720612354e-05, "loss": 0.1146, "step": 26003 }, { "epoch": 19.05054945054945, "grad_norm": 0.57421875, "learning_rate": 4.734645831206052e-05, "loss": 0.0997, "step": 26004 }, { "epoch": 19.05128205128205, "grad_norm": 0.96484375, "learning_rate": 4.734303891861834e-05, "loss": 0.1289, "step": 26005 }, { "epoch": 19.052014652014652, "grad_norm": 0.91796875, "learning_rate": 4.73396195403022e-05, "loss": 0.1297, "step": 26006 }, { "epoch": 19.052747252747253, "grad_norm": 0.765625, "learning_rate": 4.7336200177128456e-05, "loss": 0.1168, "step": 26007 }, { "epoch": 19.053479853479853, "grad_norm": 1.296875, "learning_rate": 4.7332780829113546e-05, "loss": 0.1362, "step": 26008 }, { "epoch": 19.054212454212454, "grad_norm": 1.171875, "learning_rate": 4.732936149627381e-05, "loss": 0.129, "step": 26009 }, { "epoch": 19.054945054945055, "grad_norm": 1.2890625, "learning_rate": 4.732594217862566e-05, "loss": 0.1142, "step": 26010 }, { "epoch": 19.055677655677655, "grad_norm": 1.5, "learning_rate": 4.732252287618545e-05, "loss": 0.123, "step": 26011 }, { "epoch": 19.056410256410256, "grad_norm": 1.015625, "learning_rate": 4.731910358896957e-05, "loss": 0.1225, "step": 26012 }, { "epoch": 19.057142857142857, "grad_norm": 1.1328125, "learning_rate": 4.7315684316994404e-05, "loss": 0.1282, "step": 26013 }, { "epoch": 19.057875457875458, "grad_norm": 0.96875, "learning_rate": 4.731226506027636e-05, "loss": 0.1089, "step": 26014 }, { "epoch": 19.05860805860806, "grad_norm": 0.859375, "learning_rate": 4.7308845818831785e-05, "loss": 0.0913, "step": 26015 }, { "epoch": 19.05934065934066, "grad_norm": 1.1484375, "learning_rate": 4.7305426592677085e-05, "loss": 0.1757, "step": 26016 }, { "epoch": 19.06007326007326, "grad_norm": 1.2578125, "learning_rate": 4.730200738182864e-05, "loss": 0.1081, "step": 26017 }, { "epoch": 19.06080586080586, "grad_norm": 1.1171875, "learning_rate": 4.729858818630279e-05, "loss": 0.1244, "step": 26018 }, { "epoch": 19.06153846153846, "grad_norm": 0.81640625, "learning_rate": 4.7295169006115994e-05, "loss": 0.0901, "step": 26019 }, { "epoch": 19.062271062271062, "grad_norm": 0.76171875, "learning_rate": 4.7291749841284596e-05, "loss": 0.1256, "step": 26020 }, { "epoch": 19.063003663003663, "grad_norm": 1.171875, "learning_rate": 4.728833069182497e-05, "loss": 0.0918, "step": 26021 }, { "epoch": 19.063736263736264, "grad_norm": 1.03125, "learning_rate": 4.728491155775351e-05, "loss": 0.1237, "step": 26022 }, { "epoch": 19.064468864468864, "grad_norm": 1.46875, "learning_rate": 4.7281492439086575e-05, "loss": 0.1113, "step": 26023 }, { "epoch": 19.065201465201465, "grad_norm": 1.21875, "learning_rate": 4.727807333584058e-05, "loss": 0.1739, "step": 26024 }, { "epoch": 19.065934065934066, "grad_norm": 1.3515625, "learning_rate": 4.7274654248031905e-05, "loss": 0.1453, "step": 26025 }, { "epoch": 19.066666666666666, "grad_norm": 1.109375, "learning_rate": 4.7271235175676895e-05, "loss": 0.1432, "step": 26026 }, { "epoch": 19.067399267399267, "grad_norm": 1.140625, "learning_rate": 4.726781611879197e-05, "loss": 0.1108, "step": 26027 }, { "epoch": 19.068131868131868, "grad_norm": 1.0, "learning_rate": 4.7264397077393495e-05, "loss": 0.1156, "step": 26028 }, { "epoch": 19.06886446886447, "grad_norm": 1.078125, "learning_rate": 4.7260978051497855e-05, "loss": 0.1103, "step": 26029 }, { "epoch": 19.06959706959707, "grad_norm": 0.61328125, "learning_rate": 4.7257559041121445e-05, "loss": 0.1279, "step": 26030 }, { "epoch": 19.07032967032967, "grad_norm": 1.6484375, "learning_rate": 4.7254140046280635e-05, "loss": 0.1279, "step": 26031 }, { "epoch": 19.07106227106227, "grad_norm": 1.4140625, "learning_rate": 4.72507210669918e-05, "loss": 0.1435, "step": 26032 }, { "epoch": 19.07179487179487, "grad_norm": 0.875, "learning_rate": 4.7247302103271303e-05, "loss": 0.1383, "step": 26033 }, { "epoch": 19.072527472527472, "grad_norm": 1.7265625, "learning_rate": 4.724388315513558e-05, "loss": 0.1285, "step": 26034 }, { "epoch": 19.073260073260073, "grad_norm": 0.953125, "learning_rate": 4.7240464222600974e-05, "loss": 0.1209, "step": 26035 }, { "epoch": 19.073992673992674, "grad_norm": 1.0234375, "learning_rate": 4.723704530568388e-05, "loss": 0.1061, "step": 26036 }, { "epoch": 19.074725274725274, "grad_norm": 1.125, "learning_rate": 4.7233626404400674e-05, "loss": 0.1218, "step": 26037 }, { "epoch": 19.075457875457875, "grad_norm": 1.1953125, "learning_rate": 4.7230207518767713e-05, "loss": 0.1217, "step": 26038 }, { "epoch": 19.076190476190476, "grad_norm": 0.6328125, "learning_rate": 4.7226788648801414e-05, "loss": 0.0948, "step": 26039 }, { "epoch": 19.076923076923077, "grad_norm": 0.96484375, "learning_rate": 4.7223369794518146e-05, "loss": 0.123, "step": 26040 }, { "epoch": 19.077655677655677, "grad_norm": 0.75390625, "learning_rate": 4.721995095593429e-05, "loss": 0.1141, "step": 26041 }, { "epoch": 19.078388278388278, "grad_norm": 1.484375, "learning_rate": 4.7216532133066225e-05, "loss": 0.1222, "step": 26042 }, { "epoch": 19.07912087912088, "grad_norm": 1.34375, "learning_rate": 4.721311332593034e-05, "loss": 0.1054, "step": 26043 }, { "epoch": 19.07985347985348, "grad_norm": 1.2109375, "learning_rate": 4.7209694534542984e-05, "loss": 0.1158, "step": 26044 }, { "epoch": 19.08058608058608, "grad_norm": 0.91015625, "learning_rate": 4.720627575892058e-05, "loss": 0.1175, "step": 26045 }, { "epoch": 19.08131868131868, "grad_norm": 1.3359375, "learning_rate": 4.7202856999079496e-05, "loss": 0.1472, "step": 26046 }, { "epoch": 19.08205128205128, "grad_norm": 0.83984375, "learning_rate": 4.71994382550361e-05, "loss": 0.1203, "step": 26047 }, { "epoch": 19.082783882783882, "grad_norm": 1.71875, "learning_rate": 4.719601952680679e-05, "loss": 0.1408, "step": 26048 }, { "epoch": 19.083516483516483, "grad_norm": 0.8203125, "learning_rate": 4.71926008144079e-05, "loss": 0.1285, "step": 26049 }, { "epoch": 19.084249084249084, "grad_norm": 2.0, "learning_rate": 4.7189182117855876e-05, "loss": 0.1347, "step": 26050 }, { "epoch": 19.084981684981685, "grad_norm": 1.09375, "learning_rate": 4.7185763437167064e-05, "loss": 0.1124, "step": 26051 }, { "epoch": 19.085714285714285, "grad_norm": 1.03125, "learning_rate": 4.718234477235784e-05, "loss": 0.1265, "step": 26052 }, { "epoch": 19.086446886446886, "grad_norm": 0.953125, "learning_rate": 4.71789261234446e-05, "loss": 0.1159, "step": 26053 }, { "epoch": 19.087179487179487, "grad_norm": 0.7421875, "learning_rate": 4.71755074904437e-05, "loss": 0.1029, "step": 26054 }, { "epoch": 19.087912087912088, "grad_norm": 0.77734375, "learning_rate": 4.717208887337155e-05, "loss": 0.095, "step": 26055 }, { "epoch": 19.08864468864469, "grad_norm": 1.421875, "learning_rate": 4.71686702722445e-05, "loss": 0.1561, "step": 26056 }, { "epoch": 19.08937728937729, "grad_norm": 0.7109375, "learning_rate": 4.716525168707897e-05, "loss": 0.1077, "step": 26057 }, { "epoch": 19.09010989010989, "grad_norm": 0.8046875, "learning_rate": 4.7161833117891296e-05, "loss": 0.1179, "step": 26058 }, { "epoch": 19.09084249084249, "grad_norm": 1.0078125, "learning_rate": 4.7158414564697856e-05, "loss": 0.1232, "step": 26059 }, { "epoch": 19.09157509157509, "grad_norm": 1.421875, "learning_rate": 4.715499602751508e-05, "loss": 0.1475, "step": 26060 }, { "epoch": 19.092307692307692, "grad_norm": 1.03125, "learning_rate": 4.7151577506359306e-05, "loss": 0.1299, "step": 26061 }, { "epoch": 19.093040293040293, "grad_norm": 1.0390625, "learning_rate": 4.714815900124693e-05, "loss": 0.1267, "step": 26062 }, { "epoch": 19.093772893772893, "grad_norm": 1.0234375, "learning_rate": 4.7144740512194324e-05, "loss": 0.1114, "step": 26063 }, { "epoch": 19.094505494505494, "grad_norm": 1.6328125, "learning_rate": 4.714132203921785e-05, "loss": 0.1178, "step": 26064 }, { "epoch": 19.095238095238095, "grad_norm": 1.125, "learning_rate": 4.7137903582333914e-05, "loss": 0.113, "step": 26065 }, { "epoch": 19.095970695970696, "grad_norm": 0.8671875, "learning_rate": 4.713448514155889e-05, "loss": 0.1229, "step": 26066 }, { "epoch": 19.096703296703296, "grad_norm": 0.640625, "learning_rate": 4.7131066716909144e-05, "loss": 0.0794, "step": 26067 }, { "epoch": 19.097435897435897, "grad_norm": 0.6875, "learning_rate": 4.712764830840108e-05, "loss": 0.129, "step": 26068 }, { "epoch": 19.098168498168498, "grad_norm": 1.21875, "learning_rate": 4.7124229916051046e-05, "loss": 0.1365, "step": 26069 }, { "epoch": 19.0989010989011, "grad_norm": 0.94921875, "learning_rate": 4.7120811539875424e-05, "loss": 0.1068, "step": 26070 }, { "epoch": 19.0996336996337, "grad_norm": 0.91015625, "learning_rate": 4.7117393179890626e-05, "loss": 0.1111, "step": 26071 }, { "epoch": 19.1003663003663, "grad_norm": 1.15625, "learning_rate": 4.7113974836113004e-05, "loss": 0.1161, "step": 26072 }, { "epoch": 19.1010989010989, "grad_norm": 1.234375, "learning_rate": 4.711055650855894e-05, "loss": 0.1603, "step": 26073 }, { "epoch": 19.1018315018315, "grad_norm": 0.65625, "learning_rate": 4.7107138197244814e-05, "loss": 0.1101, "step": 26074 }, { "epoch": 19.102564102564102, "grad_norm": 1.609375, "learning_rate": 4.7103719902186975e-05, "loss": 0.1442, "step": 26075 }, { "epoch": 19.103296703296703, "grad_norm": 1.265625, "learning_rate": 4.710030162340186e-05, "loss": 0.1084, "step": 26076 }, { "epoch": 19.104029304029304, "grad_norm": 0.86328125, "learning_rate": 4.709688336090581e-05, "loss": 0.1501, "step": 26077 }, { "epoch": 19.104761904761904, "grad_norm": 1.15625, "learning_rate": 4.709346511471519e-05, "loss": 0.121, "step": 26078 }, { "epoch": 19.105494505494505, "grad_norm": 0.69140625, "learning_rate": 4.709004688484642e-05, "loss": 0.1195, "step": 26079 }, { "epoch": 19.106227106227106, "grad_norm": 1.1875, "learning_rate": 4.708662867131585e-05, "loss": 0.128, "step": 26080 }, { "epoch": 19.106959706959707, "grad_norm": 0.99609375, "learning_rate": 4.708321047413985e-05, "loss": 0.14, "step": 26081 }, { "epoch": 19.107692307692307, "grad_norm": 1.640625, "learning_rate": 4.707979229333482e-05, "loss": 0.0898, "step": 26082 }, { "epoch": 19.108424908424908, "grad_norm": 0.8515625, "learning_rate": 4.707637412891713e-05, "loss": 0.118, "step": 26083 }, { "epoch": 19.10915750915751, "grad_norm": 1.2265625, "learning_rate": 4.707295598090316e-05, "loss": 0.0982, "step": 26084 }, { "epoch": 19.10989010989011, "grad_norm": 1.8359375, "learning_rate": 4.7069537849309265e-05, "loss": 0.0968, "step": 26085 }, { "epoch": 19.11062271062271, "grad_norm": 1.6171875, "learning_rate": 4.706611973415186e-05, "loss": 0.1179, "step": 26086 }, { "epoch": 19.11135531135531, "grad_norm": 0.486328125, "learning_rate": 4.706270163544729e-05, "loss": 0.101, "step": 26087 }, { "epoch": 19.11208791208791, "grad_norm": 1.125, "learning_rate": 4.7059283553211955e-05, "loss": 0.1359, "step": 26088 }, { "epoch": 19.112820512820512, "grad_norm": 1.3359375, "learning_rate": 4.7055865487462225e-05, "loss": 0.1508, "step": 26089 }, { "epoch": 19.113553113553113, "grad_norm": 0.8671875, "learning_rate": 4.705244743821445e-05, "loss": 0.1231, "step": 26090 }, { "epoch": 19.114285714285714, "grad_norm": 1.125, "learning_rate": 4.704902940548505e-05, "loss": 0.1276, "step": 26091 }, { "epoch": 19.115018315018315, "grad_norm": 1.4375, "learning_rate": 4.704561138929037e-05, "loss": 0.1239, "step": 26092 }, { "epoch": 19.115750915750915, "grad_norm": 1.421875, "learning_rate": 4.7042193389646813e-05, "loss": 0.1455, "step": 26093 }, { "epoch": 19.116483516483516, "grad_norm": 0.98828125, "learning_rate": 4.703877540657074e-05, "loss": 0.1097, "step": 26094 }, { "epoch": 19.117216117216117, "grad_norm": 1.109375, "learning_rate": 4.703535744007854e-05, "loss": 0.1337, "step": 26095 }, { "epoch": 19.117948717948718, "grad_norm": 1.171875, "learning_rate": 4.703193949018655e-05, "loss": 0.1182, "step": 26096 }, { "epoch": 19.11868131868132, "grad_norm": 1.0078125, "learning_rate": 4.7028521556911196e-05, "loss": 0.144, "step": 26097 }, { "epoch": 19.11941391941392, "grad_norm": 0.6953125, "learning_rate": 4.7025103640268843e-05, "loss": 0.1144, "step": 26098 }, { "epoch": 19.12014652014652, "grad_norm": 1.484375, "learning_rate": 4.702168574027585e-05, "loss": 0.1411, "step": 26099 }, { "epoch": 19.12087912087912, "grad_norm": 0.765625, "learning_rate": 4.701826785694861e-05, "loss": 0.1309, "step": 26100 }, { "epoch": 19.12161172161172, "grad_norm": 1.3203125, "learning_rate": 4.7014849990303464e-05, "loss": 0.1113, "step": 26101 }, { "epoch": 19.122344322344322, "grad_norm": 1.2734375, "learning_rate": 4.7011432140356845e-05, "loss": 0.1258, "step": 26102 }, { "epoch": 19.123076923076923, "grad_norm": 1.296875, "learning_rate": 4.70080143071251e-05, "loss": 0.1191, "step": 26103 }, { "epoch": 19.123809523809523, "grad_norm": 1.234375, "learning_rate": 4.7004596490624586e-05, "loss": 0.1344, "step": 26104 }, { "epoch": 19.124542124542124, "grad_norm": 0.86328125, "learning_rate": 4.700117869087171e-05, "loss": 0.1187, "step": 26105 }, { "epoch": 19.125274725274725, "grad_norm": 0.87890625, "learning_rate": 4.699776090788283e-05, "loss": 0.1341, "step": 26106 }, { "epoch": 19.126007326007326, "grad_norm": 1.1640625, "learning_rate": 4.699434314167433e-05, "loss": 0.1222, "step": 26107 }, { "epoch": 19.126739926739926, "grad_norm": 0.87109375, "learning_rate": 4.6990925392262584e-05, "loss": 0.0959, "step": 26108 }, { "epoch": 19.127472527472527, "grad_norm": 0.96484375, "learning_rate": 4.698750765966396e-05, "loss": 0.1031, "step": 26109 }, { "epoch": 19.128205128205128, "grad_norm": 1.1640625, "learning_rate": 4.698408994389486e-05, "loss": 0.1409, "step": 26110 }, { "epoch": 19.12893772893773, "grad_norm": 1.421875, "learning_rate": 4.6980672244971604e-05, "loss": 0.136, "step": 26111 }, { "epoch": 19.12967032967033, "grad_norm": 1.4140625, "learning_rate": 4.697725456291063e-05, "loss": 0.1019, "step": 26112 }, { "epoch": 19.13040293040293, "grad_norm": 0.9609375, "learning_rate": 4.697383689772828e-05, "loss": 0.125, "step": 26113 }, { "epoch": 19.13113553113553, "grad_norm": 1.53125, "learning_rate": 4.6970419249440936e-05, "loss": 0.1546, "step": 26114 }, { "epoch": 19.13186813186813, "grad_norm": 0.9765625, "learning_rate": 4.6967001618064975e-05, "loss": 0.1148, "step": 26115 }, { "epoch": 19.132600732600732, "grad_norm": 0.8828125, "learning_rate": 4.6963584003616765e-05, "loss": 0.1246, "step": 26116 }, { "epoch": 19.133333333333333, "grad_norm": 0.71484375, "learning_rate": 4.6960166406112676e-05, "loss": 0.1251, "step": 26117 }, { "epoch": 19.134065934065934, "grad_norm": 0.9765625, "learning_rate": 4.6956748825569094e-05, "loss": 0.1022, "step": 26118 }, { "epoch": 19.134798534798534, "grad_norm": 1.015625, "learning_rate": 4.69533312620024e-05, "loss": 0.1386, "step": 26119 }, { "epoch": 19.135531135531135, "grad_norm": 0.79296875, "learning_rate": 4.694991371542897e-05, "loss": 0.1076, "step": 26120 }, { "epoch": 19.136263736263736, "grad_norm": 1.4140625, "learning_rate": 4.694649618586515e-05, "loss": 0.1277, "step": 26121 }, { "epoch": 19.136996336996337, "grad_norm": 1.2265625, "learning_rate": 4.694307867332733e-05, "loss": 0.1157, "step": 26122 }, { "epoch": 19.137728937728937, "grad_norm": 1.3046875, "learning_rate": 4.69396611778319e-05, "loss": 0.0983, "step": 26123 }, { "epoch": 19.138461538461538, "grad_norm": 1.5234375, "learning_rate": 4.6936243699395224e-05, "loss": 0.1083, "step": 26124 }, { "epoch": 19.13919413919414, "grad_norm": 1.03125, "learning_rate": 4.693282623803367e-05, "loss": 0.1171, "step": 26125 }, { "epoch": 19.13992673992674, "grad_norm": 0.8046875, "learning_rate": 4.692940879376362e-05, "loss": 0.1065, "step": 26126 }, { "epoch": 19.14065934065934, "grad_norm": 1.046875, "learning_rate": 4.692599136660142e-05, "loss": 0.1283, "step": 26127 }, { "epoch": 19.14139194139194, "grad_norm": 1.234375, "learning_rate": 4.6922573956563506e-05, "loss": 0.1274, "step": 26128 }, { "epoch": 19.14212454212454, "grad_norm": 1.6640625, "learning_rate": 4.6919156563666195e-05, "loss": 0.159, "step": 26129 }, { "epoch": 19.142857142857142, "grad_norm": 0.75390625, "learning_rate": 4.6915739187925876e-05, "loss": 0.1256, "step": 26130 }, { "epoch": 19.143589743589743, "grad_norm": 1.015625, "learning_rate": 4.691232182935894e-05, "loss": 0.1162, "step": 26131 }, { "epoch": 19.144322344322344, "grad_norm": 0.91015625, "learning_rate": 4.6908904487981744e-05, "loss": 0.0977, "step": 26132 }, { "epoch": 19.145054945054945, "grad_norm": 0.55078125, "learning_rate": 4.6905487163810654e-05, "loss": 0.1161, "step": 26133 }, { "epoch": 19.145787545787545, "grad_norm": 1.15625, "learning_rate": 4.690206985686207e-05, "loss": 0.1032, "step": 26134 }, { "epoch": 19.146520146520146, "grad_norm": 0.8671875, "learning_rate": 4.689865256715235e-05, "loss": 0.1366, "step": 26135 }, { "epoch": 19.147252747252747, "grad_norm": 1.0390625, "learning_rate": 4.689523529469787e-05, "loss": 0.1289, "step": 26136 }, { "epoch": 19.147985347985347, "grad_norm": 0.828125, "learning_rate": 4.689181803951498e-05, "loss": 0.1082, "step": 26137 }, { "epoch": 19.148717948717948, "grad_norm": 1.5, "learning_rate": 4.6888400801620085e-05, "loss": 0.1336, "step": 26138 }, { "epoch": 19.14945054945055, "grad_norm": 1.1015625, "learning_rate": 4.6884983581029564e-05, "loss": 0.134, "step": 26139 }, { "epoch": 19.15018315018315, "grad_norm": 1.0859375, "learning_rate": 4.688156637775976e-05, "loss": 0.1231, "step": 26140 }, { "epoch": 19.15091575091575, "grad_norm": 0.91796875, "learning_rate": 4.6878149191827065e-05, "loss": 0.1089, "step": 26141 }, { "epoch": 19.15164835164835, "grad_norm": 0.8515625, "learning_rate": 4.687473202324784e-05, "loss": 0.13, "step": 26142 }, { "epoch": 19.152380952380952, "grad_norm": 1.203125, "learning_rate": 4.687131487203845e-05, "loss": 0.1747, "step": 26143 }, { "epoch": 19.153113553113553, "grad_norm": 0.62109375, "learning_rate": 4.68678977382153e-05, "loss": 0.1142, "step": 26144 }, { "epoch": 19.153846153846153, "grad_norm": 1.046875, "learning_rate": 4.686448062179474e-05, "loss": 0.1377, "step": 26145 }, { "epoch": 19.154578754578754, "grad_norm": 1.015625, "learning_rate": 4.686106352279315e-05, "loss": 0.1059, "step": 26146 }, { "epoch": 19.155311355311355, "grad_norm": 0.953125, "learning_rate": 4.6857646441226905e-05, "loss": 0.1276, "step": 26147 }, { "epoch": 19.156043956043955, "grad_norm": 1.0859375, "learning_rate": 4.6854229377112345e-05, "loss": 0.1256, "step": 26148 }, { "epoch": 19.156776556776556, "grad_norm": 0.96875, "learning_rate": 4.68508123304659e-05, "loss": 0.1095, "step": 26149 }, { "epoch": 19.157509157509157, "grad_norm": 0.89453125, "learning_rate": 4.68473953013039e-05, "loss": 0.0961, "step": 26150 }, { "epoch": 19.158241758241758, "grad_norm": 0.984375, "learning_rate": 4.684397828964273e-05, "loss": 0.1363, "step": 26151 }, { "epoch": 19.15897435897436, "grad_norm": 1.328125, "learning_rate": 4.684056129549876e-05, "loss": 0.1552, "step": 26152 }, { "epoch": 19.15970695970696, "grad_norm": 0.8203125, "learning_rate": 4.683714431888834e-05, "loss": 0.1032, "step": 26153 }, { "epoch": 19.16043956043956, "grad_norm": 0.96484375, "learning_rate": 4.683372735982789e-05, "loss": 0.1209, "step": 26154 }, { "epoch": 19.16117216117216, "grad_norm": 1.3828125, "learning_rate": 4.6830310418333755e-05, "loss": 0.1157, "step": 26155 }, { "epoch": 19.16190476190476, "grad_norm": 1.546875, "learning_rate": 4.682689349442229e-05, "loss": 0.1142, "step": 26156 }, { "epoch": 19.162637362637362, "grad_norm": 0.83203125, "learning_rate": 4.682347658810991e-05, "loss": 0.1221, "step": 26157 }, { "epoch": 19.163369963369963, "grad_norm": 1.171875, "learning_rate": 4.682005969941292e-05, "loss": 0.145, "step": 26158 }, { "epoch": 19.164102564102564, "grad_norm": 1.328125, "learning_rate": 4.681664282834777e-05, "loss": 0.1444, "step": 26159 }, { "epoch": 19.164835164835164, "grad_norm": 0.89453125, "learning_rate": 4.681322597493078e-05, "loss": 0.1093, "step": 26160 }, { "epoch": 19.165567765567765, "grad_norm": 1.40625, "learning_rate": 4.680980913917834e-05, "loss": 0.1085, "step": 26161 }, { "epoch": 19.166300366300366, "grad_norm": 1.3125, "learning_rate": 4.680639232110682e-05, "loss": 0.1046, "step": 26162 }, { "epoch": 19.167032967032966, "grad_norm": 0.73828125, "learning_rate": 4.680297552073256e-05, "loss": 0.0932, "step": 26163 }, { "epoch": 19.167765567765567, "grad_norm": 0.81640625, "learning_rate": 4.679955873807198e-05, "loss": 0.0978, "step": 26164 }, { "epoch": 19.168498168498168, "grad_norm": 0.8828125, "learning_rate": 4.679614197314143e-05, "loss": 0.1314, "step": 26165 }, { "epoch": 19.16923076923077, "grad_norm": 1.1640625, "learning_rate": 4.679272522595728e-05, "loss": 0.0872, "step": 26166 }, { "epoch": 19.16996336996337, "grad_norm": 0.6171875, "learning_rate": 4.6789308496535895e-05, "loss": 0.121, "step": 26167 }, { "epoch": 19.17069597069597, "grad_norm": 1.09375, "learning_rate": 4.678589178489365e-05, "loss": 0.1423, "step": 26168 }, { "epoch": 19.17142857142857, "grad_norm": 1.7265625, "learning_rate": 4.6782475091046914e-05, "loss": 0.1398, "step": 26169 }, { "epoch": 19.17216117216117, "grad_norm": 2.28125, "learning_rate": 4.677905841501206e-05, "loss": 0.1387, "step": 26170 }, { "epoch": 19.172893772893772, "grad_norm": 1.234375, "learning_rate": 4.6775641756805465e-05, "loss": 0.1395, "step": 26171 }, { "epoch": 19.173626373626373, "grad_norm": 0.9296875, "learning_rate": 4.6772225116443504e-05, "loss": 0.123, "step": 26172 }, { "epoch": 19.174358974358974, "grad_norm": 0.91015625, "learning_rate": 4.676880849394252e-05, "loss": 0.1619, "step": 26173 }, { "epoch": 19.175091575091574, "grad_norm": 1.3515625, "learning_rate": 4.6765391889318885e-05, "loss": 0.1535, "step": 26174 }, { "epoch": 19.175824175824175, "grad_norm": 0.953125, "learning_rate": 4.676197530258901e-05, "loss": 0.1451, "step": 26175 }, { "epoch": 19.176556776556776, "grad_norm": 1.8125, "learning_rate": 4.6758558733769235e-05, "loss": 0.1209, "step": 26176 }, { "epoch": 19.177289377289377, "grad_norm": 1.6953125, "learning_rate": 4.6755142182875935e-05, "loss": 0.1013, "step": 26177 }, { "epoch": 19.178021978021977, "grad_norm": 1.234375, "learning_rate": 4.675172564992548e-05, "loss": 0.1681, "step": 26178 }, { "epoch": 19.178754578754578, "grad_norm": 2.140625, "learning_rate": 4.6748309134934204e-05, "loss": 0.1252, "step": 26179 }, { "epoch": 19.17948717948718, "grad_norm": 1.921875, "learning_rate": 4.674489263791854e-05, "loss": 0.1242, "step": 26180 }, { "epoch": 19.18021978021978, "grad_norm": 1.5, "learning_rate": 4.674147615889483e-05, "loss": 0.1252, "step": 26181 }, { "epoch": 19.18095238095238, "grad_norm": 1.1796875, "learning_rate": 4.673805969787943e-05, "loss": 0.1311, "step": 26182 }, { "epoch": 19.18168498168498, "grad_norm": 1.078125, "learning_rate": 4.673464325488873e-05, "loss": 0.0948, "step": 26183 }, { "epoch": 19.18241758241758, "grad_norm": 1.2734375, "learning_rate": 4.6731226829939065e-05, "loss": 0.1518, "step": 26184 }, { "epoch": 19.183150183150182, "grad_norm": 1.1015625, "learning_rate": 4.672781042304686e-05, "loss": 0.1328, "step": 26185 }, { "epoch": 19.183882783882783, "grad_norm": 1.8984375, "learning_rate": 4.672439403422845e-05, "loss": 0.1368, "step": 26186 }, { "epoch": 19.184615384615384, "grad_norm": 1.0234375, "learning_rate": 4.6720977663500206e-05, "loss": 0.1322, "step": 26187 }, { "epoch": 19.185347985347985, "grad_norm": 1.03125, "learning_rate": 4.671756131087851e-05, "loss": 0.1375, "step": 26188 }, { "epoch": 19.186080586080585, "grad_norm": 1.15625, "learning_rate": 4.6714144976379704e-05, "loss": 0.1112, "step": 26189 }, { "epoch": 19.186813186813186, "grad_norm": 0.87890625, "learning_rate": 4.671072866002015e-05, "loss": 0.1312, "step": 26190 }, { "epoch": 19.187545787545787, "grad_norm": 0.9140625, "learning_rate": 4.670731236181627e-05, "loss": 0.1526, "step": 26191 }, { "epoch": 19.188278388278388, "grad_norm": 1.3046875, "learning_rate": 4.67038960817844e-05, "loss": 0.1304, "step": 26192 }, { "epoch": 19.18901098901099, "grad_norm": 0.9375, "learning_rate": 4.67004798199409e-05, "loss": 0.0939, "step": 26193 }, { "epoch": 19.18974358974359, "grad_norm": 1.0859375, "learning_rate": 4.669706357630215e-05, "loss": 0.1467, "step": 26194 }, { "epoch": 19.19047619047619, "grad_norm": 1.25, "learning_rate": 4.6693647350884515e-05, "loss": 0.1175, "step": 26195 }, { "epoch": 19.19120879120879, "grad_norm": 1.2421875, "learning_rate": 4.669023114370436e-05, "loss": 0.1069, "step": 26196 }, { "epoch": 19.19194139194139, "grad_norm": 0.9453125, "learning_rate": 4.668681495477807e-05, "loss": 0.1229, "step": 26197 }, { "epoch": 19.192673992673992, "grad_norm": 0.76171875, "learning_rate": 4.668339878412201e-05, "loss": 0.1145, "step": 26198 }, { "epoch": 19.193406593406593, "grad_norm": 1.796875, "learning_rate": 4.667998263175253e-05, "loss": 0.1718, "step": 26199 }, { "epoch": 19.194139194139193, "grad_norm": 1.1953125, "learning_rate": 4.6676566497685985e-05, "loss": 0.1295, "step": 26200 }, { "epoch": 19.194871794871794, "grad_norm": 1.453125, "learning_rate": 4.667315038193879e-05, "loss": 0.1542, "step": 26201 }, { "epoch": 19.195604395604395, "grad_norm": 1.6015625, "learning_rate": 4.666973428452728e-05, "loss": 0.1291, "step": 26202 }, { "epoch": 19.196336996336996, "grad_norm": 1.328125, "learning_rate": 4.666631820546784e-05, "loss": 0.1199, "step": 26203 }, { "epoch": 19.197069597069596, "grad_norm": 1.28125, "learning_rate": 4.666290214477683e-05, "loss": 0.1157, "step": 26204 }, { "epoch": 19.197802197802197, "grad_norm": 1.046875, "learning_rate": 4.665948610247059e-05, "loss": 0.0975, "step": 26205 }, { "epoch": 19.198534798534798, "grad_norm": 0.96875, "learning_rate": 4.665607007856553e-05, "loss": 0.1347, "step": 26206 }, { "epoch": 19.1992673992674, "grad_norm": 1.046875, "learning_rate": 4.6652654073077993e-05, "loss": 0.1346, "step": 26207 }, { "epoch": 19.2, "grad_norm": 1.234375, "learning_rate": 4.6649238086024366e-05, "loss": 0.1324, "step": 26208 }, { "epoch": 19.2007326007326, "grad_norm": 1.2734375, "learning_rate": 4.6645822117421006e-05, "loss": 0.1279, "step": 26209 }, { "epoch": 19.2014652014652, "grad_norm": 0.93359375, "learning_rate": 4.664240616728425e-05, "loss": 0.105, "step": 26210 }, { "epoch": 19.2021978021978, "grad_norm": 1.078125, "learning_rate": 4.6638990235630517e-05, "loss": 0.1262, "step": 26211 }, { "epoch": 19.202930402930402, "grad_norm": 1.4609375, "learning_rate": 4.663557432247615e-05, "loss": 0.114, "step": 26212 }, { "epoch": 19.203663003663003, "grad_norm": 0.87109375, "learning_rate": 4.663215842783751e-05, "loss": 0.1138, "step": 26213 }, { "epoch": 19.204395604395604, "grad_norm": 1.0, "learning_rate": 4.662874255173097e-05, "loss": 0.1001, "step": 26214 }, { "epoch": 19.205128205128204, "grad_norm": 0.63671875, "learning_rate": 4.6625326694172896e-05, "loss": 0.1063, "step": 26215 }, { "epoch": 19.205860805860805, "grad_norm": 1.046875, "learning_rate": 4.662191085517964e-05, "loss": 0.1382, "step": 26216 }, { "epoch": 19.206593406593406, "grad_norm": 1.2578125, "learning_rate": 4.661849503476759e-05, "loss": 0.125, "step": 26217 }, { "epoch": 19.207326007326007, "grad_norm": 1.2890625, "learning_rate": 4.661507923295311e-05, "loss": 0.1158, "step": 26218 }, { "epoch": 19.208058608058607, "grad_norm": 1.0390625, "learning_rate": 4.661166344975256e-05, "loss": 0.1144, "step": 26219 }, { "epoch": 19.208791208791208, "grad_norm": 0.82421875, "learning_rate": 4.6608247685182306e-05, "loss": 0.1089, "step": 26220 }, { "epoch": 19.20952380952381, "grad_norm": 0.80859375, "learning_rate": 4.660483193925871e-05, "loss": 0.1151, "step": 26221 }, { "epoch": 19.21025641025641, "grad_norm": 1.0, "learning_rate": 4.660141621199814e-05, "loss": 0.1222, "step": 26222 }, { "epoch": 19.21098901098901, "grad_norm": 0.9140625, "learning_rate": 4.659800050341697e-05, "loss": 0.1214, "step": 26223 }, { "epoch": 19.21172161172161, "grad_norm": 1.296875, "learning_rate": 4.6594584813531573e-05, "loss": 0.1289, "step": 26224 }, { "epoch": 19.21245421245421, "grad_norm": 0.8203125, "learning_rate": 4.659116914235829e-05, "loss": 0.1151, "step": 26225 }, { "epoch": 19.213186813186812, "grad_norm": 1.2890625, "learning_rate": 4.658775348991348e-05, "loss": 0.121, "step": 26226 }, { "epoch": 19.213919413919413, "grad_norm": 1.078125, "learning_rate": 4.658433785621355e-05, "loss": 0.1512, "step": 26227 }, { "epoch": 19.214652014652014, "grad_norm": 1.0625, "learning_rate": 4.658092224127485e-05, "loss": 0.1176, "step": 26228 }, { "epoch": 19.215384615384615, "grad_norm": 0.921875, "learning_rate": 4.657750664511373e-05, "loss": 0.1207, "step": 26229 }, { "epoch": 19.216117216117215, "grad_norm": 0.9921875, "learning_rate": 4.657409106774657e-05, "loss": 0.1226, "step": 26230 }, { "epoch": 19.216849816849816, "grad_norm": 1.28125, "learning_rate": 4.6570675509189705e-05, "loss": 0.1105, "step": 26231 }, { "epoch": 19.217582417582417, "grad_norm": 0.8046875, "learning_rate": 4.656725996945954e-05, "loss": 0.1124, "step": 26232 }, { "epoch": 19.218315018315018, "grad_norm": 0.875, "learning_rate": 4.6563844448572416e-05, "loss": 0.14, "step": 26233 }, { "epoch": 19.21904761904762, "grad_norm": 0.9921875, "learning_rate": 4.6560428946544715e-05, "loss": 0.1233, "step": 26234 }, { "epoch": 19.21978021978022, "grad_norm": 1.3359375, "learning_rate": 4.655701346339279e-05, "loss": 0.1271, "step": 26235 }, { "epoch": 19.22051282051282, "grad_norm": 1.4921875, "learning_rate": 4.655359799913299e-05, "loss": 0.1716, "step": 26236 }, { "epoch": 19.22124542124542, "grad_norm": 1.5234375, "learning_rate": 4.6550182553781716e-05, "loss": 0.1458, "step": 26237 }, { "epoch": 19.22197802197802, "grad_norm": 1.53125, "learning_rate": 4.654676712735532e-05, "loss": 0.1458, "step": 26238 }, { "epoch": 19.222710622710622, "grad_norm": 1.0703125, "learning_rate": 4.654335171987015e-05, "loss": 0.1345, "step": 26239 }, { "epoch": 19.223443223443223, "grad_norm": 1.046875, "learning_rate": 4.6539936331342596e-05, "loss": 0.1219, "step": 26240 }, { "epoch": 19.224175824175823, "grad_norm": 1.2734375, "learning_rate": 4.653652096178899e-05, "loss": 0.1349, "step": 26241 }, { "epoch": 19.224908424908424, "grad_norm": 0.6953125, "learning_rate": 4.653310561122571e-05, "loss": 0.0981, "step": 26242 }, { "epoch": 19.225641025641025, "grad_norm": 1.0078125, "learning_rate": 4.652969027966913e-05, "loss": 0.1346, "step": 26243 }, { "epoch": 19.226373626373626, "grad_norm": 0.88671875, "learning_rate": 4.652627496713562e-05, "loss": 0.1231, "step": 26244 }, { "epoch": 19.227106227106226, "grad_norm": 1.34375, "learning_rate": 4.652285967364151e-05, "loss": 0.1129, "step": 26245 }, { "epoch": 19.227838827838827, "grad_norm": 1.03125, "learning_rate": 4.651944439920319e-05, "loss": 0.1004, "step": 26246 }, { "epoch": 19.228571428571428, "grad_norm": 1.21875, "learning_rate": 4.651602914383702e-05, "loss": 0.1018, "step": 26247 }, { "epoch": 19.22930402930403, "grad_norm": 1.359375, "learning_rate": 4.651261390755936e-05, "loss": 0.1634, "step": 26248 }, { "epoch": 19.23003663003663, "grad_norm": 1.796875, "learning_rate": 4.6509198690386584e-05, "loss": 0.1419, "step": 26249 }, { "epoch": 19.23076923076923, "grad_norm": 0.8203125, "learning_rate": 4.6505783492335046e-05, "loss": 0.1264, "step": 26250 }, { "epoch": 19.23150183150183, "grad_norm": 1.0625, "learning_rate": 4.650236831342111e-05, "loss": 0.1161, "step": 26251 }, { "epoch": 19.23223443223443, "grad_norm": 0.90625, "learning_rate": 4.649895315366113e-05, "loss": 0.1092, "step": 26252 }, { "epoch": 19.232967032967032, "grad_norm": 1.34375, "learning_rate": 4.6495538013071487e-05, "loss": 0.1422, "step": 26253 }, { "epoch": 19.233699633699633, "grad_norm": 1.375, "learning_rate": 4.6492122891668536e-05, "loss": 0.1249, "step": 26254 }, { "epoch": 19.234432234432234, "grad_norm": 1.1484375, "learning_rate": 4.648870778946865e-05, "loss": 0.1371, "step": 26255 }, { "epoch": 19.235164835164834, "grad_norm": 0.8671875, "learning_rate": 4.648529270648817e-05, "loss": 0.1185, "step": 26256 }, { "epoch": 19.235897435897435, "grad_norm": 0.9140625, "learning_rate": 4.648187764274346e-05, "loss": 0.1211, "step": 26257 }, { "epoch": 19.236630036630036, "grad_norm": 0.90234375, "learning_rate": 4.647846259825091e-05, "loss": 0.1037, "step": 26258 }, { "epoch": 19.237362637362637, "grad_norm": 1.2734375, "learning_rate": 4.6475047573026855e-05, "loss": 0.1135, "step": 26259 }, { "epoch": 19.238095238095237, "grad_norm": 1.3203125, "learning_rate": 4.6471632567087676e-05, "loss": 0.1307, "step": 26260 }, { "epoch": 19.238827838827838, "grad_norm": 1.2578125, "learning_rate": 4.6468217580449734e-05, "loss": 0.1431, "step": 26261 }, { "epoch": 19.23956043956044, "grad_norm": 1.4296875, "learning_rate": 4.646480261312936e-05, "loss": 0.1328, "step": 26262 }, { "epoch": 19.24029304029304, "grad_norm": 2.515625, "learning_rate": 4.646138766514296e-05, "loss": 0.1296, "step": 26263 }, { "epoch": 19.24102564102564, "grad_norm": 1.296875, "learning_rate": 4.645797273650689e-05, "loss": 0.1408, "step": 26264 }, { "epoch": 19.24175824175824, "grad_norm": 1.109375, "learning_rate": 4.6454557827237485e-05, "loss": 0.1364, "step": 26265 }, { "epoch": 19.24249084249084, "grad_norm": 0.734375, "learning_rate": 4.645114293735113e-05, "loss": 0.13, "step": 26266 }, { "epoch": 19.243223443223442, "grad_norm": 1.046875, "learning_rate": 4.6447728066864173e-05, "loss": 0.133, "step": 26267 }, { "epoch": 19.243956043956043, "grad_norm": 1.03125, "learning_rate": 4.644431321579296e-05, "loss": 0.0935, "step": 26268 }, { "epoch": 19.244688644688644, "grad_norm": 1.2109375, "learning_rate": 4.6440898384153896e-05, "loss": 0.1233, "step": 26269 }, { "epoch": 19.245421245421245, "grad_norm": 1.0546875, "learning_rate": 4.6437483571963325e-05, "loss": 0.1024, "step": 26270 }, { "epoch": 19.246153846153845, "grad_norm": 1.03125, "learning_rate": 4.643406877923759e-05, "loss": 0.1241, "step": 26271 }, { "epoch": 19.246886446886446, "grad_norm": 1.484375, "learning_rate": 4.643065400599307e-05, "loss": 0.1401, "step": 26272 }, { "epoch": 19.247619047619047, "grad_norm": 0.92578125, "learning_rate": 4.6427239252246115e-05, "loss": 0.1231, "step": 26273 }, { "epoch": 19.248351648351647, "grad_norm": 1.46875, "learning_rate": 4.64238245180131e-05, "loss": 0.1361, "step": 26274 }, { "epoch": 19.249084249084248, "grad_norm": 0.92578125, "learning_rate": 4.6420409803310385e-05, "loss": 0.1209, "step": 26275 }, { "epoch": 19.24981684981685, "grad_norm": 1.1640625, "learning_rate": 4.641699510815433e-05, "loss": 0.113, "step": 26276 }, { "epoch": 19.25054945054945, "grad_norm": 1.0234375, "learning_rate": 4.641358043256129e-05, "loss": 0.1103, "step": 26277 }, { "epoch": 19.25128205128205, "grad_norm": 0.72265625, "learning_rate": 4.641016577654762e-05, "loss": 0.0986, "step": 26278 }, { "epoch": 19.25201465201465, "grad_norm": 1.578125, "learning_rate": 4.640675114012969e-05, "loss": 0.122, "step": 26279 }, { "epoch": 19.252747252747252, "grad_norm": 0.6484375, "learning_rate": 4.6403336523323865e-05, "loss": 0.1221, "step": 26280 }, { "epoch": 19.253479853479853, "grad_norm": 1.1953125, "learning_rate": 4.63999219261465e-05, "loss": 0.1222, "step": 26281 }, { "epoch": 19.254212454212453, "grad_norm": 1.03125, "learning_rate": 4.639650734861394e-05, "loss": 0.1377, "step": 26282 }, { "epoch": 19.254945054945054, "grad_norm": 0.89453125, "learning_rate": 4.6393092790742566e-05, "loss": 0.1034, "step": 26283 }, { "epoch": 19.255677655677655, "grad_norm": 1.015625, "learning_rate": 4.638967825254874e-05, "loss": 0.1221, "step": 26284 }, { "epoch": 19.256410256410255, "grad_norm": 1.1875, "learning_rate": 4.638626373404881e-05, "loss": 0.146, "step": 26285 }, { "epoch": 19.257142857142856, "grad_norm": 0.88671875, "learning_rate": 4.6382849235259145e-05, "loss": 0.1226, "step": 26286 }, { "epoch": 19.257875457875457, "grad_norm": 1.203125, "learning_rate": 4.63794347561961e-05, "loss": 0.1218, "step": 26287 }, { "epoch": 19.258608058608058, "grad_norm": 0.6484375, "learning_rate": 4.637602029687602e-05, "loss": 0.1011, "step": 26288 }, { "epoch": 19.25934065934066, "grad_norm": 1.15625, "learning_rate": 4.63726058573153e-05, "loss": 0.1303, "step": 26289 }, { "epoch": 19.26007326007326, "grad_norm": 0.92578125, "learning_rate": 4.636919143753029e-05, "loss": 0.1369, "step": 26290 }, { "epoch": 19.26080586080586, "grad_norm": 0.90625, "learning_rate": 4.636577703753732e-05, "loss": 0.1438, "step": 26291 }, { "epoch": 19.26153846153846, "grad_norm": 0.58984375, "learning_rate": 4.6362362657352785e-05, "loss": 0.0992, "step": 26292 }, { "epoch": 19.26227106227106, "grad_norm": 0.80859375, "learning_rate": 4.635894829699302e-05, "loss": 0.1206, "step": 26293 }, { "epoch": 19.263003663003662, "grad_norm": 1.4765625, "learning_rate": 4.635553395647437e-05, "loss": 0.1475, "step": 26294 }, { "epoch": 19.263736263736263, "grad_norm": 1.1328125, "learning_rate": 4.6352119635813245e-05, "loss": 0.1214, "step": 26295 }, { "epoch": 19.264468864468864, "grad_norm": 0.98828125, "learning_rate": 4.634870533502598e-05, "loss": 0.1289, "step": 26296 }, { "epoch": 19.265201465201464, "grad_norm": 0.6640625, "learning_rate": 4.6345291054128916e-05, "loss": 0.1129, "step": 26297 }, { "epoch": 19.265934065934065, "grad_norm": 1.3359375, "learning_rate": 4.6341876793138435e-05, "loss": 0.11, "step": 26298 }, { "epoch": 19.266666666666666, "grad_norm": 1.21875, "learning_rate": 4.633846255207087e-05, "loss": 0.1248, "step": 26299 }, { "epoch": 19.267399267399266, "grad_norm": 1.1171875, "learning_rate": 4.633504833094262e-05, "loss": 0.1312, "step": 26300 }, { "epoch": 19.268131868131867, "grad_norm": 1.203125, "learning_rate": 4.633163412977001e-05, "loss": 0.1181, "step": 26301 }, { "epoch": 19.268864468864468, "grad_norm": 1.0703125, "learning_rate": 4.6328219948569416e-05, "loss": 0.1325, "step": 26302 }, { "epoch": 19.26959706959707, "grad_norm": 0.80078125, "learning_rate": 4.632480578735719e-05, "loss": 0.1077, "step": 26303 }, { "epoch": 19.27032967032967, "grad_norm": 1.140625, "learning_rate": 4.632139164614967e-05, "loss": 0.1249, "step": 26304 }, { "epoch": 19.27106227106227, "grad_norm": 0.86328125, "learning_rate": 4.631797752496325e-05, "loss": 0.122, "step": 26305 }, { "epoch": 19.27179487179487, "grad_norm": 0.6171875, "learning_rate": 4.631456342381428e-05, "loss": 0.1111, "step": 26306 }, { "epoch": 19.27252747252747, "grad_norm": 1.296875, "learning_rate": 4.6311149342719104e-05, "loss": 0.13, "step": 26307 }, { "epoch": 19.273260073260072, "grad_norm": 0.97265625, "learning_rate": 4.6307735281694074e-05, "loss": 0.1225, "step": 26308 }, { "epoch": 19.273992673992673, "grad_norm": 1.5234375, "learning_rate": 4.630432124075556e-05, "loss": 0.1006, "step": 26309 }, { "epoch": 19.274725274725274, "grad_norm": 0.796875, "learning_rate": 4.630090721991994e-05, "loss": 0.1124, "step": 26310 }, { "epoch": 19.275457875457874, "grad_norm": 0.95703125, "learning_rate": 4.6297493219203544e-05, "loss": 0.1149, "step": 26311 }, { "epoch": 19.276190476190475, "grad_norm": 0.859375, "learning_rate": 4.629407923862274e-05, "loss": 0.1063, "step": 26312 }, { "epoch": 19.276923076923076, "grad_norm": 1.2109375, "learning_rate": 4.629066527819388e-05, "loss": 0.1214, "step": 26313 }, { "epoch": 19.277655677655677, "grad_norm": 0.98046875, "learning_rate": 4.6287251337933314e-05, "loss": 0.1261, "step": 26314 }, { "epoch": 19.278388278388277, "grad_norm": 0.9140625, "learning_rate": 4.6283837417857426e-05, "loss": 0.1047, "step": 26315 }, { "epoch": 19.279120879120878, "grad_norm": 1.4453125, "learning_rate": 4.628042351798255e-05, "loss": 0.141, "step": 26316 }, { "epoch": 19.27985347985348, "grad_norm": 0.69140625, "learning_rate": 4.627700963832506e-05, "loss": 0.1089, "step": 26317 }, { "epoch": 19.28058608058608, "grad_norm": 1.0234375, "learning_rate": 4.627359577890129e-05, "loss": 0.119, "step": 26318 }, { "epoch": 19.28131868131868, "grad_norm": 0.96484375, "learning_rate": 4.627018193972763e-05, "loss": 0.1076, "step": 26319 }, { "epoch": 19.28205128205128, "grad_norm": 0.85546875, "learning_rate": 4.626676812082039e-05, "loss": 0.1181, "step": 26320 }, { "epoch": 19.282783882783882, "grad_norm": 1.5078125, "learning_rate": 4.626335432219597e-05, "loss": 0.1356, "step": 26321 }, { "epoch": 19.283516483516483, "grad_norm": 1.2265625, "learning_rate": 4.6259940543870704e-05, "loss": 0.1126, "step": 26322 }, { "epoch": 19.284249084249083, "grad_norm": 0.78515625, "learning_rate": 4.625652678586095e-05, "loss": 0.109, "step": 26323 }, { "epoch": 19.284981684981684, "grad_norm": 1.21875, "learning_rate": 4.6253113048183085e-05, "loss": 0.1385, "step": 26324 }, { "epoch": 19.285714285714285, "grad_norm": 0.96484375, "learning_rate": 4.6249699330853416e-05, "loss": 0.136, "step": 26325 }, { "epoch": 19.286446886446885, "grad_norm": 1.71875, "learning_rate": 4.6246285633888365e-05, "loss": 0.1313, "step": 26326 }, { "epoch": 19.287179487179486, "grad_norm": 1.3671875, "learning_rate": 4.624287195730425e-05, "loss": 0.125, "step": 26327 }, { "epoch": 19.287912087912087, "grad_norm": 1.2890625, "learning_rate": 4.6239458301117436e-05, "loss": 0.1189, "step": 26328 }, { "epoch": 19.288644688644688, "grad_norm": 1.109375, "learning_rate": 4.623604466534428e-05, "loss": 0.1265, "step": 26329 }, { "epoch": 19.28937728937729, "grad_norm": 1.234375, "learning_rate": 4.623263105000111e-05, "loss": 0.1363, "step": 26330 }, { "epoch": 19.29010989010989, "grad_norm": 1.078125, "learning_rate": 4.6229217455104325e-05, "loss": 0.0949, "step": 26331 }, { "epoch": 19.29084249084249, "grad_norm": 0.80078125, "learning_rate": 4.6225803880670254e-05, "loss": 0.1273, "step": 26332 }, { "epoch": 19.29157509157509, "grad_norm": 1.7578125, "learning_rate": 4.622239032671527e-05, "loss": 0.136, "step": 26333 }, { "epoch": 19.29230769230769, "grad_norm": 0.8984375, "learning_rate": 4.62189767932557e-05, "loss": 0.0994, "step": 26334 }, { "epoch": 19.293040293040292, "grad_norm": 1.828125, "learning_rate": 4.621556328030792e-05, "loss": 0.12, "step": 26335 }, { "epoch": 19.293772893772893, "grad_norm": 1.1875, "learning_rate": 4.62121497878883e-05, "loss": 0.0843, "step": 26336 }, { "epoch": 19.294505494505493, "grad_norm": 1.2578125, "learning_rate": 4.620873631601316e-05, "loss": 0.1222, "step": 26337 }, { "epoch": 19.295238095238094, "grad_norm": 0.8125, "learning_rate": 4.6205322864698886e-05, "loss": 0.1098, "step": 26338 }, { "epoch": 19.295970695970695, "grad_norm": 0.78515625, "learning_rate": 4.620190943396182e-05, "loss": 0.124, "step": 26339 }, { "epoch": 19.296703296703296, "grad_norm": 1.515625, "learning_rate": 4.619849602381829e-05, "loss": 0.1339, "step": 26340 }, { "epoch": 19.297435897435896, "grad_norm": 0.82421875, "learning_rate": 4.619508263428472e-05, "loss": 0.1254, "step": 26341 }, { "epoch": 19.298168498168497, "grad_norm": 1.1796875, "learning_rate": 4.6191669265377396e-05, "loss": 0.1366, "step": 26342 }, { "epoch": 19.298901098901098, "grad_norm": 0.89453125, "learning_rate": 4.6188255917112716e-05, "loss": 0.1311, "step": 26343 }, { "epoch": 19.2996336996337, "grad_norm": 1.3515625, "learning_rate": 4.618484258950701e-05, "loss": 0.1373, "step": 26344 }, { "epoch": 19.3003663003663, "grad_norm": 1.1328125, "learning_rate": 4.618142928257664e-05, "loss": 0.1279, "step": 26345 }, { "epoch": 19.3010989010989, "grad_norm": 1.15625, "learning_rate": 4.617801599633794e-05, "loss": 0.0984, "step": 26346 }, { "epoch": 19.3018315018315, "grad_norm": 0.875, "learning_rate": 4.61746027308073e-05, "loss": 0.1029, "step": 26347 }, { "epoch": 19.3025641025641, "grad_norm": 1.2734375, "learning_rate": 4.617118948600106e-05, "loss": 0.1254, "step": 26348 }, { "epoch": 19.303296703296702, "grad_norm": 1.2578125, "learning_rate": 4.616777626193557e-05, "loss": 0.1513, "step": 26349 }, { "epoch": 19.304029304029303, "grad_norm": 1.234375, "learning_rate": 4.6164363058627196e-05, "loss": 0.1204, "step": 26350 }, { "epoch": 19.304761904761904, "grad_norm": 1.25, "learning_rate": 4.616094987609225e-05, "loss": 0.1236, "step": 26351 }, { "epoch": 19.305494505494504, "grad_norm": 1.03125, "learning_rate": 4.615753671434715e-05, "loss": 0.1385, "step": 26352 }, { "epoch": 19.306227106227105, "grad_norm": 1.671875, "learning_rate": 4.615412357340821e-05, "loss": 0.1065, "step": 26353 }, { "epoch": 19.306959706959706, "grad_norm": 0.7421875, "learning_rate": 4.6150710453291795e-05, "loss": 0.1226, "step": 26354 }, { "epoch": 19.307692307692307, "grad_norm": 1.3046875, "learning_rate": 4.614729735401425e-05, "loss": 0.1224, "step": 26355 }, { "epoch": 19.308424908424907, "grad_norm": 1.375, "learning_rate": 4.6143884275591914e-05, "loss": 0.1246, "step": 26356 }, { "epoch": 19.309157509157508, "grad_norm": 0.6875, "learning_rate": 4.614047121804118e-05, "loss": 0.1148, "step": 26357 }, { "epoch": 19.30989010989011, "grad_norm": 1.0625, "learning_rate": 4.613705818137838e-05, "loss": 0.1278, "step": 26358 }, { "epoch": 19.31062271062271, "grad_norm": 0.85546875, "learning_rate": 4.613364516561986e-05, "loss": 0.1244, "step": 26359 }, { "epoch": 19.31135531135531, "grad_norm": 0.953125, "learning_rate": 4.6130232170781976e-05, "loss": 0.1104, "step": 26360 }, { "epoch": 19.31208791208791, "grad_norm": 0.95703125, "learning_rate": 4.6126819196881085e-05, "loss": 0.1127, "step": 26361 }, { "epoch": 19.31282051282051, "grad_norm": 0.98828125, "learning_rate": 4.6123406243933545e-05, "loss": 0.0934, "step": 26362 }, { "epoch": 19.313553113553112, "grad_norm": 0.89453125, "learning_rate": 4.6119993311955695e-05, "loss": 0.1222, "step": 26363 }, { "epoch": 19.314285714285713, "grad_norm": 1.046875, "learning_rate": 4.61165804009639e-05, "loss": 0.0898, "step": 26364 }, { "epoch": 19.315018315018314, "grad_norm": 0.69921875, "learning_rate": 4.611316751097451e-05, "loss": 0.1136, "step": 26365 }, { "epoch": 19.315750915750915, "grad_norm": 0.984375, "learning_rate": 4.6109754642003856e-05, "loss": 0.126, "step": 26366 }, { "epoch": 19.316483516483515, "grad_norm": 1.0546875, "learning_rate": 4.6106341794068324e-05, "loss": 0.1171, "step": 26367 }, { "epoch": 19.317216117216116, "grad_norm": 0.90625, "learning_rate": 4.610292896718426e-05, "loss": 0.1373, "step": 26368 }, { "epoch": 19.317948717948717, "grad_norm": 0.84375, "learning_rate": 4.6099516161368e-05, "loss": 0.0962, "step": 26369 }, { "epoch": 19.318681318681318, "grad_norm": 0.9609375, "learning_rate": 4.60961033766359e-05, "loss": 0.1128, "step": 26370 }, { "epoch": 19.31941391941392, "grad_norm": 1.5078125, "learning_rate": 4.6092690613004316e-05, "loss": 0.1616, "step": 26371 }, { "epoch": 19.32014652014652, "grad_norm": 0.8203125, "learning_rate": 4.608927787048958e-05, "loss": 0.1267, "step": 26372 }, { "epoch": 19.32087912087912, "grad_norm": 1.359375, "learning_rate": 4.6085865149108085e-05, "loss": 0.1239, "step": 26373 }, { "epoch": 19.32161172161172, "grad_norm": 1.1953125, "learning_rate": 4.6082452448876145e-05, "loss": 0.1328, "step": 26374 }, { "epoch": 19.32234432234432, "grad_norm": 0.60546875, "learning_rate": 4.6079039769810136e-05, "loss": 0.1122, "step": 26375 }, { "epoch": 19.323076923076922, "grad_norm": 2.09375, "learning_rate": 4.607562711192641e-05, "loss": 0.1363, "step": 26376 }, { "epoch": 19.323809523809523, "grad_norm": 1.046875, "learning_rate": 4.607221447524127e-05, "loss": 0.1066, "step": 26377 }, { "epoch": 19.324542124542123, "grad_norm": 1.0, "learning_rate": 4.6068801859771135e-05, "loss": 0.1369, "step": 26378 }, { "epoch": 19.325274725274724, "grad_norm": 0.56640625, "learning_rate": 4.606538926553232e-05, "loss": 0.0999, "step": 26379 }, { "epoch": 19.326007326007325, "grad_norm": 0.95703125, "learning_rate": 4.606197669254118e-05, "loss": 0.1222, "step": 26380 }, { "epoch": 19.326739926739926, "grad_norm": 0.92578125, "learning_rate": 4.6058564140814065e-05, "loss": 0.0976, "step": 26381 }, { "epoch": 19.327472527472526, "grad_norm": 0.80859375, "learning_rate": 4.605515161036731e-05, "loss": 0.0998, "step": 26382 }, { "epoch": 19.328205128205127, "grad_norm": 0.82421875, "learning_rate": 4.60517391012173e-05, "loss": 0.1056, "step": 26383 }, { "epoch": 19.328937728937728, "grad_norm": 1.5078125, "learning_rate": 4.6048326613380374e-05, "loss": 0.1213, "step": 26384 }, { "epoch": 19.32967032967033, "grad_norm": 0.99609375, "learning_rate": 4.604491414687287e-05, "loss": 0.1182, "step": 26385 }, { "epoch": 19.33040293040293, "grad_norm": 1.6796875, "learning_rate": 4.6041501701711136e-05, "loss": 0.1213, "step": 26386 }, { "epoch": 19.33113553113553, "grad_norm": 1.0390625, "learning_rate": 4.603808927791154e-05, "loss": 0.1333, "step": 26387 }, { "epoch": 19.33186813186813, "grad_norm": 1.1328125, "learning_rate": 4.603467687549043e-05, "loss": 0.1147, "step": 26388 }, { "epoch": 19.33260073260073, "grad_norm": 0.95703125, "learning_rate": 4.603126449446413e-05, "loss": 0.1009, "step": 26389 }, { "epoch": 19.333333333333332, "grad_norm": 0.66015625, "learning_rate": 4.602785213484903e-05, "loss": 0.1263, "step": 26390 }, { "epoch": 19.334065934065933, "grad_norm": 0.6640625, "learning_rate": 4.602443979666145e-05, "loss": 0.1024, "step": 26391 }, { "epoch": 19.334798534798534, "grad_norm": 0.7890625, "learning_rate": 4.602102747991775e-05, "loss": 0.1002, "step": 26392 }, { "epoch": 19.335531135531134, "grad_norm": 0.82421875, "learning_rate": 4.601761518463426e-05, "loss": 0.1241, "step": 26393 }, { "epoch": 19.336263736263735, "grad_norm": 0.7734375, "learning_rate": 4.6014202910827365e-05, "loss": 0.1207, "step": 26394 }, { "epoch": 19.336996336996336, "grad_norm": 1.140625, "learning_rate": 4.6010790658513394e-05, "loss": 0.1357, "step": 26395 }, { "epoch": 19.337728937728937, "grad_norm": 1.2890625, "learning_rate": 4.60073784277087e-05, "loss": 0.1082, "step": 26396 }, { "epoch": 19.338461538461537, "grad_norm": 1.671875, "learning_rate": 4.6003966218429627e-05, "loss": 0.1216, "step": 26397 }, { "epoch": 19.339194139194138, "grad_norm": 1.0859375, "learning_rate": 4.600055403069252e-05, "loss": 0.1087, "step": 26398 }, { "epoch": 19.33992673992674, "grad_norm": 1.09375, "learning_rate": 4.5997141864513755e-05, "loss": 0.1132, "step": 26399 }, { "epoch": 19.34065934065934, "grad_norm": 0.58203125, "learning_rate": 4.5993729719909645e-05, "loss": 0.1009, "step": 26400 }, { "epoch": 19.34139194139194, "grad_norm": 1.15625, "learning_rate": 4.599031759689657e-05, "loss": 0.1444, "step": 26401 }, { "epoch": 19.34212454212454, "grad_norm": 1.4765625, "learning_rate": 4.5986905495490854e-05, "loss": 0.1276, "step": 26402 }, { "epoch": 19.34285714285714, "grad_norm": 1.3046875, "learning_rate": 4.598349341570884e-05, "loss": 0.1581, "step": 26403 }, { "epoch": 19.343589743589742, "grad_norm": 0.8046875, "learning_rate": 4.5980081357566914e-05, "loss": 0.1246, "step": 26404 }, { "epoch": 19.344322344322343, "grad_norm": 1.015625, "learning_rate": 4.5976669321081396e-05, "loss": 0.1329, "step": 26405 }, { "epoch": 19.345054945054944, "grad_norm": 1.3046875, "learning_rate": 4.5973257306268647e-05, "loss": 0.1342, "step": 26406 }, { "epoch": 19.345787545787545, "grad_norm": 1.0625, "learning_rate": 4.596984531314501e-05, "loss": 0.124, "step": 26407 }, { "epoch": 19.346520146520145, "grad_norm": 1.0859375, "learning_rate": 4.596643334172681e-05, "loss": 0.1253, "step": 26408 }, { "epoch": 19.347252747252746, "grad_norm": 0.94140625, "learning_rate": 4.5963021392030434e-05, "loss": 0.1064, "step": 26409 }, { "epoch": 19.347985347985347, "grad_norm": 0.91796875, "learning_rate": 4.5959609464072205e-05, "loss": 0.132, "step": 26410 }, { "epoch": 19.348717948717947, "grad_norm": 1.671875, "learning_rate": 4.595619755786849e-05, "loss": 0.1053, "step": 26411 }, { "epoch": 19.349450549450548, "grad_norm": 1.3359375, "learning_rate": 4.595278567343561e-05, "loss": 0.1345, "step": 26412 }, { "epoch": 19.35018315018315, "grad_norm": 0.76953125, "learning_rate": 4.594937381078992e-05, "loss": 0.1204, "step": 26413 }, { "epoch": 19.35091575091575, "grad_norm": 1.1796875, "learning_rate": 4.594596196994778e-05, "loss": 0.1347, "step": 26414 }, { "epoch": 19.35164835164835, "grad_norm": 0.92578125, "learning_rate": 4.594255015092554e-05, "loss": 0.1281, "step": 26415 }, { "epoch": 19.35238095238095, "grad_norm": 1.09375, "learning_rate": 4.593913835373954e-05, "loss": 0.109, "step": 26416 }, { "epoch": 19.353113553113552, "grad_norm": 0.875, "learning_rate": 4.593572657840612e-05, "loss": 0.1217, "step": 26417 }, { "epoch": 19.353846153846153, "grad_norm": 1.4375, "learning_rate": 4.5932314824941636e-05, "loss": 0.1207, "step": 26418 }, { "epoch": 19.354578754578753, "grad_norm": 1.0859375, "learning_rate": 4.592890309336241e-05, "loss": 0.1086, "step": 26419 }, { "epoch": 19.355311355311354, "grad_norm": 0.6484375, "learning_rate": 4.592549138368483e-05, "loss": 0.1138, "step": 26420 }, { "epoch": 19.356043956043955, "grad_norm": 0.5546875, "learning_rate": 4.5922079695925225e-05, "loss": 0.1174, "step": 26421 }, { "epoch": 19.356776556776556, "grad_norm": 1.0546875, "learning_rate": 4.591866803009994e-05, "loss": 0.0982, "step": 26422 }, { "epoch": 19.357509157509156, "grad_norm": 1.2109375, "learning_rate": 4.591525638622531e-05, "loss": 0.1208, "step": 26423 }, { "epoch": 19.358241758241757, "grad_norm": 1.25, "learning_rate": 4.591184476431769e-05, "loss": 0.1304, "step": 26424 }, { "epoch": 19.358974358974358, "grad_norm": 0.78125, "learning_rate": 4.5908433164393434e-05, "loss": 0.1479, "step": 26425 }, { "epoch": 19.35970695970696, "grad_norm": 0.796875, "learning_rate": 4.5905021586468875e-05, "loss": 0.1368, "step": 26426 }, { "epoch": 19.36043956043956, "grad_norm": 0.9140625, "learning_rate": 4.590161003056038e-05, "loss": 0.122, "step": 26427 }, { "epoch": 19.36117216117216, "grad_norm": 0.73046875, "learning_rate": 4.589819849668428e-05, "loss": 0.1298, "step": 26428 }, { "epoch": 19.36190476190476, "grad_norm": 1.171875, "learning_rate": 4.589478698485689e-05, "loss": 0.1647, "step": 26429 }, { "epoch": 19.36263736263736, "grad_norm": 1.28125, "learning_rate": 4.589137549509462e-05, "loss": 0.1148, "step": 26430 }, { "epoch": 19.363369963369962, "grad_norm": 0.9375, "learning_rate": 4.588796402741378e-05, "loss": 0.1075, "step": 26431 }, { "epoch": 19.364102564102563, "grad_norm": 0.5546875, "learning_rate": 4.588455258183072e-05, "loss": 0.1239, "step": 26432 }, { "epoch": 19.364835164835164, "grad_norm": 0.671875, "learning_rate": 4.588114115836178e-05, "loss": 0.1179, "step": 26433 }, { "epoch": 19.365567765567764, "grad_norm": 1.2890625, "learning_rate": 4.587772975702329e-05, "loss": 0.1355, "step": 26434 }, { "epoch": 19.366300366300365, "grad_norm": 1.3203125, "learning_rate": 4.5874318377831634e-05, "loss": 0.1339, "step": 26435 }, { "epoch": 19.367032967032966, "grad_norm": 1.3359375, "learning_rate": 4.587090702080314e-05, "loss": 0.1441, "step": 26436 }, { "epoch": 19.367765567765566, "grad_norm": 1.1015625, "learning_rate": 4.586749568595414e-05, "loss": 0.1184, "step": 26437 }, { "epoch": 19.368498168498167, "grad_norm": 0.86328125, "learning_rate": 4.586408437330099e-05, "loss": 0.1163, "step": 26438 }, { "epoch": 19.369230769230768, "grad_norm": 1.3671875, "learning_rate": 4.5860673082860035e-05, "loss": 0.137, "step": 26439 }, { "epoch": 19.36996336996337, "grad_norm": 1.0546875, "learning_rate": 4.585726181464761e-05, "loss": 0.1436, "step": 26440 }, { "epoch": 19.37069597069597, "grad_norm": 1.25, "learning_rate": 4.585385056868009e-05, "loss": 0.0973, "step": 26441 }, { "epoch": 19.37142857142857, "grad_norm": 1.0234375, "learning_rate": 4.585043934497378e-05, "loss": 0.162, "step": 26442 }, { "epoch": 19.37216117216117, "grad_norm": 0.6484375, "learning_rate": 4.5847028143545054e-05, "loss": 0.1054, "step": 26443 }, { "epoch": 19.37289377289377, "grad_norm": 0.78515625, "learning_rate": 4.584361696441024e-05, "loss": 0.1255, "step": 26444 }, { "epoch": 19.373626373626372, "grad_norm": 1.15625, "learning_rate": 4.584020580758566e-05, "loss": 0.1311, "step": 26445 }, { "epoch": 19.374358974358973, "grad_norm": 1.3125, "learning_rate": 4.5836794673087716e-05, "loss": 0.1277, "step": 26446 }, { "epoch": 19.375091575091574, "grad_norm": 1.2421875, "learning_rate": 4.5833383560932717e-05, "loss": 0.1331, "step": 26447 }, { "epoch": 19.375824175824174, "grad_norm": 1.46875, "learning_rate": 4.5829972471137e-05, "loss": 0.1321, "step": 26448 }, { "epoch": 19.376556776556775, "grad_norm": 0.94921875, "learning_rate": 4.582656140371691e-05, "loss": 0.0952, "step": 26449 }, { "epoch": 19.377289377289376, "grad_norm": 1.46875, "learning_rate": 4.58231503586888e-05, "loss": 0.1454, "step": 26450 }, { "epoch": 19.378021978021977, "grad_norm": 1.9453125, "learning_rate": 4.581973933606903e-05, "loss": 0.1564, "step": 26451 }, { "epoch": 19.378754578754577, "grad_norm": 1.2265625, "learning_rate": 4.58163283358739e-05, "loss": 0.1275, "step": 26452 }, { "epoch": 19.379487179487178, "grad_norm": 0.796875, "learning_rate": 4.58129173581198e-05, "loss": 0.1472, "step": 26453 }, { "epoch": 19.38021978021978, "grad_norm": 0.78515625, "learning_rate": 4.580950640282305e-05, "loss": 0.1115, "step": 26454 }, { "epoch": 19.38095238095238, "grad_norm": 0.9453125, "learning_rate": 4.5806095469999976e-05, "loss": 0.1343, "step": 26455 }, { "epoch": 19.38168498168498, "grad_norm": 1.234375, "learning_rate": 4.5802684559666954e-05, "loss": 0.1127, "step": 26456 }, { "epoch": 19.38241758241758, "grad_norm": 0.8203125, "learning_rate": 4.5799273671840316e-05, "loss": 0.1409, "step": 26457 }, { "epoch": 19.383150183150182, "grad_norm": 1.015625, "learning_rate": 4.57958628065364e-05, "loss": 0.128, "step": 26458 }, { "epoch": 19.383882783882783, "grad_norm": 1.125, "learning_rate": 4.579245196377155e-05, "loss": 0.1273, "step": 26459 }, { "epoch": 19.384615384615383, "grad_norm": 0.890625, "learning_rate": 4.5789041143562087e-05, "loss": 0.1053, "step": 26460 }, { "epoch": 19.385347985347984, "grad_norm": 1.265625, "learning_rate": 4.57856303459244e-05, "loss": 0.1253, "step": 26461 }, { "epoch": 19.386080586080585, "grad_norm": 1.4609375, "learning_rate": 4.5782219570874804e-05, "loss": 0.115, "step": 26462 }, { "epoch": 19.386813186813185, "grad_norm": 1.046875, "learning_rate": 4.5778808818429625e-05, "loss": 0.0955, "step": 26463 }, { "epoch": 19.387545787545786, "grad_norm": 0.87890625, "learning_rate": 4.577539808860524e-05, "loss": 0.1368, "step": 26464 }, { "epoch": 19.388278388278387, "grad_norm": 1.2734375, "learning_rate": 4.577198738141795e-05, "loss": 0.112, "step": 26465 }, { "epoch": 19.389010989010988, "grad_norm": 1.375, "learning_rate": 4.5768576696884135e-05, "loss": 0.1438, "step": 26466 }, { "epoch": 19.38974358974359, "grad_norm": 1.0390625, "learning_rate": 4.576516603502014e-05, "loss": 0.1203, "step": 26467 }, { "epoch": 19.39047619047619, "grad_norm": 0.90234375, "learning_rate": 4.576175539584227e-05, "loss": 0.1184, "step": 26468 }, { "epoch": 19.39120879120879, "grad_norm": 0.9765625, "learning_rate": 4.57583447793669e-05, "loss": 0.1227, "step": 26469 }, { "epoch": 19.39194139194139, "grad_norm": 0.765625, "learning_rate": 4.575493418561035e-05, "loss": 0.1192, "step": 26470 }, { "epoch": 19.39267399267399, "grad_norm": 1.0703125, "learning_rate": 4.575152361458894e-05, "loss": 0.1157, "step": 26471 }, { "epoch": 19.393406593406592, "grad_norm": 1.3203125, "learning_rate": 4.5748113066319074e-05, "loss": 0.1209, "step": 26472 }, { "epoch": 19.394139194139193, "grad_norm": 0.84375, "learning_rate": 4.574470254081705e-05, "loss": 0.1033, "step": 26473 }, { "epoch": 19.394871794871793, "grad_norm": 0.9609375, "learning_rate": 4.574129203809922e-05, "loss": 0.1138, "step": 26474 }, { "epoch": 19.395604395604394, "grad_norm": 0.9921875, "learning_rate": 4.573788155818191e-05, "loss": 0.109, "step": 26475 }, { "epoch": 19.396336996336995, "grad_norm": 1.125, "learning_rate": 4.573447110108147e-05, "loss": 0.1276, "step": 26476 }, { "epoch": 19.397069597069596, "grad_norm": 1.2890625, "learning_rate": 4.573106066681426e-05, "loss": 0.1475, "step": 26477 }, { "epoch": 19.397802197802196, "grad_norm": 1.0859375, "learning_rate": 4.5727650255396584e-05, "loss": 0.1362, "step": 26478 }, { "epoch": 19.398534798534797, "grad_norm": 0.73046875, "learning_rate": 4.572423986684482e-05, "loss": 0.1197, "step": 26479 }, { "epoch": 19.399267399267398, "grad_norm": 1.3984375, "learning_rate": 4.572082950117529e-05, "loss": 0.1156, "step": 26480 }, { "epoch": 19.4, "grad_norm": 0.94140625, "learning_rate": 4.571741915840431e-05, "loss": 0.1101, "step": 26481 }, { "epoch": 19.4007326007326, "grad_norm": 0.67578125, "learning_rate": 4.571400883854828e-05, "loss": 0.1159, "step": 26482 }, { "epoch": 19.4014652014652, "grad_norm": 1.2578125, "learning_rate": 4.571059854162349e-05, "loss": 0.1251, "step": 26483 }, { "epoch": 19.4021978021978, "grad_norm": 0.640625, "learning_rate": 4.57071882676463e-05, "loss": 0.1237, "step": 26484 }, { "epoch": 19.4029304029304, "grad_norm": 1.5, "learning_rate": 4.5703778016633045e-05, "loss": 0.1564, "step": 26485 }, { "epoch": 19.403663003663002, "grad_norm": 1.3125, "learning_rate": 4.570036778860004e-05, "loss": 0.1008, "step": 26486 }, { "epoch": 19.404395604395603, "grad_norm": 1.0546875, "learning_rate": 4.569695758356366e-05, "loss": 0.1005, "step": 26487 }, { "epoch": 19.405128205128204, "grad_norm": 1.1953125, "learning_rate": 4.569354740154025e-05, "loss": 0.1243, "step": 26488 }, { "epoch": 19.405860805860804, "grad_norm": 1.0859375, "learning_rate": 4.5690137242546115e-05, "loss": 0.1295, "step": 26489 }, { "epoch": 19.406593406593405, "grad_norm": 0.82421875, "learning_rate": 4.5686727106597624e-05, "loss": 0.1268, "step": 26490 }, { "epoch": 19.407326007326006, "grad_norm": 1.3515625, "learning_rate": 4.568331699371109e-05, "loss": 0.119, "step": 26491 }, { "epoch": 19.408058608058607, "grad_norm": 1.34375, "learning_rate": 4.567990690390286e-05, "loss": 0.1544, "step": 26492 }, { "epoch": 19.408791208791207, "grad_norm": 1.0625, "learning_rate": 4.5676496837189304e-05, "loss": 0.1468, "step": 26493 }, { "epoch": 19.409523809523808, "grad_norm": 1.1171875, "learning_rate": 4.567308679358673e-05, "loss": 0.1107, "step": 26494 }, { "epoch": 19.41025641025641, "grad_norm": 0.89453125, "learning_rate": 4.5669676773111476e-05, "loss": 0.1114, "step": 26495 }, { "epoch": 19.41098901098901, "grad_norm": 0.765625, "learning_rate": 4.5666266775779886e-05, "loss": 0.1429, "step": 26496 }, { "epoch": 19.41172161172161, "grad_norm": 0.85546875, "learning_rate": 4.566285680160828e-05, "loss": 0.1028, "step": 26497 }, { "epoch": 19.41245421245421, "grad_norm": 1.265625, "learning_rate": 4.565944685061304e-05, "loss": 0.1234, "step": 26498 }, { "epoch": 19.41318681318681, "grad_norm": 1.0703125, "learning_rate": 4.5656036922810476e-05, "loss": 0.1228, "step": 26499 }, { "epoch": 19.413919413919412, "grad_norm": 0.78515625, "learning_rate": 4.5652627018216925e-05, "loss": 0.1161, "step": 26500 }, { "epoch": 19.414652014652013, "grad_norm": 1.0390625, "learning_rate": 4.5649217136848726e-05, "loss": 0.1194, "step": 26501 }, { "epoch": 19.415384615384614, "grad_norm": 1.1171875, "learning_rate": 4.564580727872222e-05, "loss": 0.1413, "step": 26502 }, { "epoch": 19.416117216117215, "grad_norm": 1.296875, "learning_rate": 4.564239744385375e-05, "loss": 0.1369, "step": 26503 }, { "epoch": 19.416849816849815, "grad_norm": 1.1171875, "learning_rate": 4.563898763225964e-05, "loss": 0.1031, "step": 26504 }, { "epoch": 19.417582417582416, "grad_norm": 1.2265625, "learning_rate": 4.5635577843956243e-05, "loss": 0.1585, "step": 26505 }, { "epoch": 19.418315018315017, "grad_norm": 1.4296875, "learning_rate": 4.563216807895989e-05, "loss": 0.1288, "step": 26506 }, { "epoch": 19.419047619047618, "grad_norm": 0.953125, "learning_rate": 4.5628758337286905e-05, "loss": 0.116, "step": 26507 }, { "epoch": 19.41978021978022, "grad_norm": 1.203125, "learning_rate": 4.562534861895365e-05, "loss": 0.1219, "step": 26508 }, { "epoch": 19.42051282051282, "grad_norm": 0.91796875, "learning_rate": 4.562193892397646e-05, "loss": 0.1149, "step": 26509 }, { "epoch": 19.42124542124542, "grad_norm": 0.6484375, "learning_rate": 4.5618529252371654e-05, "loss": 0.0941, "step": 26510 }, { "epoch": 19.42197802197802, "grad_norm": 1.03125, "learning_rate": 4.5615119604155575e-05, "loss": 0.1039, "step": 26511 }, { "epoch": 19.42271062271062, "grad_norm": 1.0546875, "learning_rate": 4.561170997934455e-05, "loss": 0.1018, "step": 26512 }, { "epoch": 19.423443223443222, "grad_norm": 0.93359375, "learning_rate": 4.560830037795494e-05, "loss": 0.1613, "step": 26513 }, { "epoch": 19.424175824175823, "grad_norm": 1.4375, "learning_rate": 4.5604890800003076e-05, "loss": 0.1511, "step": 26514 }, { "epoch": 19.424908424908423, "grad_norm": 1.171875, "learning_rate": 4.5601481245505264e-05, "loss": 0.1605, "step": 26515 }, { "epoch": 19.425641025641024, "grad_norm": 0.81640625, "learning_rate": 4.559807171447789e-05, "loss": 0.1262, "step": 26516 }, { "epoch": 19.426373626373625, "grad_norm": 0.94140625, "learning_rate": 4.559466220693724e-05, "loss": 0.1105, "step": 26517 }, { "epoch": 19.427106227106226, "grad_norm": 0.7421875, "learning_rate": 4.559125272289968e-05, "loss": 0.1429, "step": 26518 }, { "epoch": 19.427838827838826, "grad_norm": 0.984375, "learning_rate": 4.558784326238155e-05, "loss": 0.145, "step": 26519 }, { "epoch": 19.428571428571427, "grad_norm": 0.9453125, "learning_rate": 4.558443382539918e-05, "loss": 0.1129, "step": 26520 }, { "epoch": 19.429304029304028, "grad_norm": 0.890625, "learning_rate": 4.558102441196889e-05, "loss": 0.1024, "step": 26521 }, { "epoch": 19.43003663003663, "grad_norm": 1.0625, "learning_rate": 4.557761502210703e-05, "loss": 0.1283, "step": 26522 }, { "epoch": 19.43076923076923, "grad_norm": 1.453125, "learning_rate": 4.557420565582992e-05, "loss": 0.1194, "step": 26523 }, { "epoch": 19.43150183150183, "grad_norm": 1.34375, "learning_rate": 4.557079631315392e-05, "loss": 0.1337, "step": 26524 }, { "epoch": 19.43223443223443, "grad_norm": 1.09375, "learning_rate": 4.556738699409535e-05, "loss": 0.1427, "step": 26525 }, { "epoch": 19.43296703296703, "grad_norm": 0.87890625, "learning_rate": 4.5563977698670556e-05, "loss": 0.111, "step": 26526 }, { "epoch": 19.433699633699632, "grad_norm": 0.7421875, "learning_rate": 4.556056842689585e-05, "loss": 0.1108, "step": 26527 }, { "epoch": 19.434432234432233, "grad_norm": 1.2109375, "learning_rate": 4.555715917878759e-05, "loss": 0.1021, "step": 26528 }, { "epoch": 19.435164835164834, "grad_norm": 0.87109375, "learning_rate": 4.555374995436209e-05, "loss": 0.1281, "step": 26529 }, { "epoch": 19.435897435897434, "grad_norm": 1.2265625, "learning_rate": 4.555034075363572e-05, "loss": 0.1253, "step": 26530 }, { "epoch": 19.436630036630035, "grad_norm": 0.98046875, "learning_rate": 4.554693157662479e-05, "loss": 0.1439, "step": 26531 }, { "epoch": 19.437362637362636, "grad_norm": 0.53515625, "learning_rate": 4.554352242334563e-05, "loss": 0.0905, "step": 26532 }, { "epoch": 19.438095238095237, "grad_norm": 1.5390625, "learning_rate": 4.5540113293814564e-05, "loss": 0.1608, "step": 26533 }, { "epoch": 19.438827838827837, "grad_norm": 1.125, "learning_rate": 4.553670418804797e-05, "loss": 0.1364, "step": 26534 }, { "epoch": 19.439560439560438, "grad_norm": 0.6484375, "learning_rate": 4.5533295106062145e-05, "loss": 0.0957, "step": 26535 }, { "epoch": 19.44029304029304, "grad_norm": 0.94921875, "learning_rate": 4.552988604787344e-05, "loss": 0.1134, "step": 26536 }, { "epoch": 19.44102564102564, "grad_norm": 0.69921875, "learning_rate": 4.5526477013498184e-05, "loss": 0.1217, "step": 26537 }, { "epoch": 19.44175824175824, "grad_norm": 1.40625, "learning_rate": 4.552306800295269e-05, "loss": 0.1276, "step": 26538 }, { "epoch": 19.44249084249084, "grad_norm": 0.86328125, "learning_rate": 4.551965901625333e-05, "loss": 0.118, "step": 26539 }, { "epoch": 19.44322344322344, "grad_norm": 0.78125, "learning_rate": 4.551625005341641e-05, "loss": 0.101, "step": 26540 }, { "epoch": 19.443956043956042, "grad_norm": 1.125, "learning_rate": 4.551284111445827e-05, "loss": 0.1378, "step": 26541 }, { "epoch": 19.444688644688643, "grad_norm": 0.98828125, "learning_rate": 4.550943219939525e-05, "loss": 0.1066, "step": 26542 }, { "epoch": 19.445421245421244, "grad_norm": 1.09375, "learning_rate": 4.5506023308243676e-05, "loss": 0.1077, "step": 26543 }, { "epoch": 19.446153846153845, "grad_norm": 0.77734375, "learning_rate": 4.550261444101988e-05, "loss": 0.1197, "step": 26544 }, { "epoch": 19.446886446886445, "grad_norm": 0.70703125, "learning_rate": 4.5499205597740214e-05, "loss": 0.1186, "step": 26545 }, { "epoch": 19.447619047619046, "grad_norm": 1.046875, "learning_rate": 4.549579677842099e-05, "loss": 0.1396, "step": 26546 }, { "epoch": 19.448351648351647, "grad_norm": 1.2890625, "learning_rate": 4.549238798307855e-05, "loss": 0.1045, "step": 26547 }, { "epoch": 19.449084249084247, "grad_norm": 1.1875, "learning_rate": 4.5488979211729214e-05, "loss": 0.1455, "step": 26548 }, { "epoch": 19.449816849816848, "grad_norm": 0.9609375, "learning_rate": 4.548557046438931e-05, "loss": 0.1354, "step": 26549 }, { "epoch": 19.45054945054945, "grad_norm": 1.0546875, "learning_rate": 4.548216174107521e-05, "loss": 0.1378, "step": 26550 }, { "epoch": 19.45128205128205, "grad_norm": 0.9140625, "learning_rate": 4.547875304180321e-05, "loss": 0.1125, "step": 26551 }, { "epoch": 19.45201465201465, "grad_norm": 1.09375, "learning_rate": 4.547534436658966e-05, "loss": 0.1092, "step": 26552 }, { "epoch": 19.45274725274725, "grad_norm": 1.25, "learning_rate": 4.547193571545087e-05, "loss": 0.1434, "step": 26553 }, { "epoch": 19.453479853479852, "grad_norm": 0.9765625, "learning_rate": 4.546852708840319e-05, "loss": 0.1338, "step": 26554 }, { "epoch": 19.454212454212453, "grad_norm": 0.96484375, "learning_rate": 4.546511848546295e-05, "loss": 0.1292, "step": 26555 }, { "epoch": 19.454945054945053, "grad_norm": 0.80078125, "learning_rate": 4.5461709906646486e-05, "loss": 0.1138, "step": 26556 }, { "epoch": 19.455677655677654, "grad_norm": 0.96484375, "learning_rate": 4.545830135197012e-05, "loss": 0.1196, "step": 26557 }, { "epoch": 19.456410256410255, "grad_norm": 0.76171875, "learning_rate": 4.545489282145019e-05, "loss": 0.095, "step": 26558 }, { "epoch": 19.457142857142856, "grad_norm": 1.0546875, "learning_rate": 4.545148431510301e-05, "loss": 0.125, "step": 26559 }, { "epoch": 19.457875457875456, "grad_norm": 0.81640625, "learning_rate": 4.544807583294494e-05, "loss": 0.1014, "step": 26560 }, { "epoch": 19.458608058608057, "grad_norm": 1.09375, "learning_rate": 4.54446673749923e-05, "loss": 0.1288, "step": 26561 }, { "epoch": 19.459340659340658, "grad_norm": 0.7265625, "learning_rate": 4.544125894126142e-05, "loss": 0.1018, "step": 26562 }, { "epoch": 19.46007326007326, "grad_norm": 0.984375, "learning_rate": 4.543785053176862e-05, "loss": 0.0966, "step": 26563 }, { "epoch": 19.46080586080586, "grad_norm": 1.21875, "learning_rate": 4.543444214653023e-05, "loss": 0.1451, "step": 26564 }, { "epoch": 19.46153846153846, "grad_norm": 1.90625, "learning_rate": 4.543103378556261e-05, "loss": 0.1373, "step": 26565 }, { "epoch": 19.46227106227106, "grad_norm": 0.80078125, "learning_rate": 4.542762544888207e-05, "loss": 0.1152, "step": 26566 }, { "epoch": 19.46300366300366, "grad_norm": 1.171875, "learning_rate": 4.542421713650493e-05, "loss": 0.1363, "step": 26567 }, { "epoch": 19.463736263736262, "grad_norm": 1.203125, "learning_rate": 4.5420808848447546e-05, "loss": 0.1256, "step": 26568 }, { "epoch": 19.464468864468863, "grad_norm": 0.70703125, "learning_rate": 4.541740058472621e-05, "loss": 0.1331, "step": 26569 }, { "epoch": 19.465201465201464, "grad_norm": 1.0390625, "learning_rate": 4.5413992345357295e-05, "loss": 0.0861, "step": 26570 }, { "epoch": 19.465934065934064, "grad_norm": 0.91015625, "learning_rate": 4.541058413035711e-05, "loss": 0.1058, "step": 26571 }, { "epoch": 19.466666666666665, "grad_norm": 1.203125, "learning_rate": 4.540717593974199e-05, "loss": 0.1611, "step": 26572 }, { "epoch": 19.467399267399266, "grad_norm": 1.09375, "learning_rate": 4.5403767773528265e-05, "loss": 0.1331, "step": 26573 }, { "epoch": 19.468131868131866, "grad_norm": 1.0859375, "learning_rate": 4.540035963173226e-05, "loss": 0.1412, "step": 26574 }, { "epoch": 19.468864468864467, "grad_norm": 1.8515625, "learning_rate": 4.539695151437029e-05, "loss": 0.1656, "step": 26575 }, { "epoch": 19.469597069597068, "grad_norm": 1.171875, "learning_rate": 4.5393543421458715e-05, "loss": 0.1474, "step": 26576 }, { "epoch": 19.47032967032967, "grad_norm": 1.2890625, "learning_rate": 4.539013535301385e-05, "loss": 0.1324, "step": 26577 }, { "epoch": 19.47106227106227, "grad_norm": 0.9140625, "learning_rate": 4.538672730905201e-05, "loss": 0.1117, "step": 26578 }, { "epoch": 19.47179487179487, "grad_norm": 0.86328125, "learning_rate": 4.538331928958956e-05, "loss": 0.1011, "step": 26579 }, { "epoch": 19.47252747252747, "grad_norm": 1.5546875, "learning_rate": 4.537991129464279e-05, "loss": 0.1365, "step": 26580 }, { "epoch": 19.47326007326007, "grad_norm": 0.9921875, "learning_rate": 4.537650332422804e-05, "loss": 0.0811, "step": 26581 }, { "epoch": 19.473992673992672, "grad_norm": 1.34375, "learning_rate": 4.537309537836166e-05, "loss": 0.1186, "step": 26582 }, { "epoch": 19.474725274725273, "grad_norm": 1.0078125, "learning_rate": 4.5369687457059966e-05, "loss": 0.1174, "step": 26583 }, { "epoch": 19.475457875457874, "grad_norm": 1.0625, "learning_rate": 4.536627956033928e-05, "loss": 0.1244, "step": 26584 }, { "epoch": 19.476190476190474, "grad_norm": 0.9921875, "learning_rate": 4.5362871688215916e-05, "loss": 0.1305, "step": 26585 }, { "epoch": 19.476923076923075, "grad_norm": 0.86328125, "learning_rate": 4.535946384070624e-05, "loss": 0.1184, "step": 26586 }, { "epoch": 19.477655677655676, "grad_norm": 1.3359375, "learning_rate": 4.5356056017826565e-05, "loss": 0.1371, "step": 26587 }, { "epoch": 19.478388278388277, "grad_norm": 0.96875, "learning_rate": 4.5352648219593205e-05, "loss": 0.1079, "step": 26588 }, { "epoch": 19.479120879120877, "grad_norm": 1.0390625, "learning_rate": 4.5349240446022504e-05, "loss": 0.1287, "step": 26589 }, { "epoch": 19.479853479853478, "grad_norm": 1.4140625, "learning_rate": 4.534583269713076e-05, "loss": 0.1003, "step": 26590 }, { "epoch": 19.48058608058608, "grad_norm": 1.1796875, "learning_rate": 4.5342424972934346e-05, "loss": 0.1365, "step": 26591 }, { "epoch": 19.48131868131868, "grad_norm": 1.21875, "learning_rate": 4.533901727344956e-05, "loss": 0.1066, "step": 26592 }, { "epoch": 19.48205128205128, "grad_norm": 0.69921875, "learning_rate": 4.533560959869273e-05, "loss": 0.1138, "step": 26593 }, { "epoch": 19.48278388278388, "grad_norm": 0.9375, "learning_rate": 4.53322019486802e-05, "loss": 0.1251, "step": 26594 }, { "epoch": 19.483516483516482, "grad_norm": 1.0, "learning_rate": 4.532879432342827e-05, "loss": 0.1612, "step": 26595 }, { "epoch": 19.484249084249083, "grad_norm": 1.28125, "learning_rate": 4.53253867229533e-05, "loss": 0.105, "step": 26596 }, { "epoch": 19.484981684981683, "grad_norm": 1.9375, "learning_rate": 4.5321979147271606e-05, "loss": 0.1447, "step": 26597 }, { "epoch": 19.485714285714284, "grad_norm": 1.234375, "learning_rate": 4.531857159639951e-05, "loss": 0.1267, "step": 26598 }, { "epoch": 19.486446886446885, "grad_norm": 0.88671875, "learning_rate": 4.531516407035333e-05, "loss": 0.1507, "step": 26599 }, { "epoch": 19.487179487179485, "grad_norm": 1.0078125, "learning_rate": 4.5311756569149405e-05, "loss": 0.1114, "step": 26600 }, { "epoch": 19.487912087912086, "grad_norm": 1.28125, "learning_rate": 4.530834909280403e-05, "loss": 0.1364, "step": 26601 }, { "epoch": 19.488644688644687, "grad_norm": 0.85546875, "learning_rate": 4.530494164133359e-05, "loss": 0.1629, "step": 26602 }, { "epoch": 19.489377289377288, "grad_norm": 1.2265625, "learning_rate": 4.5301534214754375e-05, "loss": 0.107, "step": 26603 }, { "epoch": 19.49010989010989, "grad_norm": 1.1015625, "learning_rate": 4.529812681308271e-05, "loss": 0.1277, "step": 26604 }, { "epoch": 19.49084249084249, "grad_norm": 0.7578125, "learning_rate": 4.529471943633494e-05, "loss": 0.0938, "step": 26605 }, { "epoch": 19.49157509157509, "grad_norm": 0.83984375, "learning_rate": 4.529131208452736e-05, "loss": 0.1096, "step": 26606 }, { "epoch": 19.49230769230769, "grad_norm": 0.7578125, "learning_rate": 4.528790475767632e-05, "loss": 0.1312, "step": 26607 }, { "epoch": 19.493040293040295, "grad_norm": 1.03125, "learning_rate": 4.528449745579814e-05, "loss": 0.1256, "step": 26608 }, { "epoch": 19.493772893772892, "grad_norm": 0.7734375, "learning_rate": 4.528109017890915e-05, "loss": 0.132, "step": 26609 }, { "epoch": 19.494505494505496, "grad_norm": 0.88671875, "learning_rate": 4.5277682927025674e-05, "loss": 0.1007, "step": 26610 }, { "epoch": 19.495238095238093, "grad_norm": 1.0859375, "learning_rate": 4.527427570016401e-05, "loss": 0.1241, "step": 26611 }, { "epoch": 19.495970695970698, "grad_norm": 0.66796875, "learning_rate": 4.527086849834053e-05, "loss": 0.1126, "step": 26612 }, { "epoch": 19.496703296703295, "grad_norm": 1.0703125, "learning_rate": 4.526746132157153e-05, "loss": 0.1195, "step": 26613 }, { "epoch": 19.4974358974359, "grad_norm": 0.7734375, "learning_rate": 4.5264054169873345e-05, "loss": 0.1222, "step": 26614 }, { "epoch": 19.498168498168496, "grad_norm": 1.0, "learning_rate": 4.5260647043262296e-05, "loss": 0.1192, "step": 26615 }, { "epoch": 19.4989010989011, "grad_norm": 1.0, "learning_rate": 4.525723994175468e-05, "loss": 0.1051, "step": 26616 }, { "epoch": 19.499633699633698, "grad_norm": 0.96484375, "learning_rate": 4.525383286536687e-05, "loss": 0.1198, "step": 26617 }, { "epoch": 19.500366300366302, "grad_norm": 0.875, "learning_rate": 4.5250425814115185e-05, "loss": 0.1238, "step": 26618 }, { "epoch": 19.5010989010989, "grad_norm": 1.109375, "learning_rate": 4.5247018788015904e-05, "loss": 0.1409, "step": 26619 }, { "epoch": 19.501831501831504, "grad_norm": 0.65625, "learning_rate": 4.52436117870854e-05, "loss": 0.1088, "step": 26620 }, { "epoch": 19.5025641025641, "grad_norm": 1.265625, "learning_rate": 4.5240204811339974e-05, "loss": 0.098, "step": 26621 }, { "epoch": 19.503296703296705, "grad_norm": 0.859375, "learning_rate": 4.523679786079593e-05, "loss": 0.1055, "step": 26622 }, { "epoch": 19.504029304029302, "grad_norm": 1.1328125, "learning_rate": 4.5233390935469644e-05, "loss": 0.1165, "step": 26623 }, { "epoch": 19.504761904761907, "grad_norm": 0.94921875, "learning_rate": 4.5229984035377404e-05, "loss": 0.1165, "step": 26624 }, { "epoch": 19.505494505494504, "grad_norm": 1.3125, "learning_rate": 4.522657716053554e-05, "loss": 0.1549, "step": 26625 }, { "epoch": 19.506227106227108, "grad_norm": 2.09375, "learning_rate": 4.5223170310960386e-05, "loss": 0.1688, "step": 26626 }, { "epoch": 19.506959706959705, "grad_norm": 1.1171875, "learning_rate": 4.5219763486668224e-05, "loss": 0.1532, "step": 26627 }, { "epoch": 19.50769230769231, "grad_norm": 0.91796875, "learning_rate": 4.5216356687675424e-05, "loss": 0.1294, "step": 26628 }, { "epoch": 19.508424908424907, "grad_norm": 0.87890625, "learning_rate": 4.521294991399831e-05, "loss": 0.1245, "step": 26629 }, { "epoch": 19.50915750915751, "grad_norm": 1.1484375, "learning_rate": 4.5209543165653175e-05, "loss": 0.1284, "step": 26630 }, { "epoch": 19.509890109890108, "grad_norm": 0.80859375, "learning_rate": 4.5206136442656355e-05, "loss": 0.1038, "step": 26631 }, { "epoch": 19.510622710622712, "grad_norm": 1.359375, "learning_rate": 4.520272974502417e-05, "loss": 0.1036, "step": 26632 }, { "epoch": 19.51135531135531, "grad_norm": 1.3125, "learning_rate": 4.5199323072772946e-05, "loss": 0.1244, "step": 26633 }, { "epoch": 19.512087912087914, "grad_norm": 0.890625, "learning_rate": 4.519591642591901e-05, "loss": 0.1199, "step": 26634 }, { "epoch": 19.51282051282051, "grad_norm": 0.65625, "learning_rate": 4.5192509804478686e-05, "loss": 0.1014, "step": 26635 }, { "epoch": 19.513553113553115, "grad_norm": 0.703125, "learning_rate": 4.5189103208468285e-05, "loss": 0.1137, "step": 26636 }, { "epoch": 19.514285714285712, "grad_norm": 1.0234375, "learning_rate": 4.518569663790412e-05, "loss": 0.1435, "step": 26637 }, { "epoch": 19.515018315018317, "grad_norm": 1.609375, "learning_rate": 4.5182290092802545e-05, "loss": 0.1489, "step": 26638 }, { "epoch": 19.515750915750917, "grad_norm": 1.390625, "learning_rate": 4.517888357317986e-05, "loss": 0.1683, "step": 26639 }, { "epoch": 19.516483516483518, "grad_norm": 0.91796875, "learning_rate": 4.517547707905239e-05, "loss": 0.1273, "step": 26640 }, { "epoch": 19.51721611721612, "grad_norm": 0.84765625, "learning_rate": 4.517207061043647e-05, "loss": 0.11, "step": 26641 }, { "epoch": 19.51794871794872, "grad_norm": 1.1953125, "learning_rate": 4.5168664167348376e-05, "loss": 0.1165, "step": 26642 }, { "epoch": 19.51868131868132, "grad_norm": 0.73828125, "learning_rate": 4.5165257749804476e-05, "loss": 0.1195, "step": 26643 }, { "epoch": 19.51941391941392, "grad_norm": 0.65234375, "learning_rate": 4.5161851357821084e-05, "loss": 0.1411, "step": 26644 }, { "epoch": 19.520146520146522, "grad_norm": 1.125, "learning_rate": 4.5158444991414516e-05, "loss": 0.1576, "step": 26645 }, { "epoch": 19.520879120879123, "grad_norm": 0.69921875, "learning_rate": 4.5155038650601095e-05, "loss": 0.1163, "step": 26646 }, { "epoch": 19.521611721611723, "grad_norm": 1.3125, "learning_rate": 4.515163233539713e-05, "loss": 0.1465, "step": 26647 }, { "epoch": 19.522344322344324, "grad_norm": 1.5, "learning_rate": 4.514822604581894e-05, "loss": 0.1452, "step": 26648 }, { "epoch": 19.523076923076925, "grad_norm": 1.59375, "learning_rate": 4.514481978188287e-05, "loss": 0.1328, "step": 26649 }, { "epoch": 19.523809523809526, "grad_norm": 0.75390625, "learning_rate": 4.514141354360523e-05, "loss": 0.1078, "step": 26650 }, { "epoch": 19.524542124542126, "grad_norm": 1.1328125, "learning_rate": 4.5138007331002335e-05, "loss": 0.1341, "step": 26651 }, { "epoch": 19.525274725274727, "grad_norm": 0.9921875, "learning_rate": 4.513460114409051e-05, "loss": 0.1138, "step": 26652 }, { "epoch": 19.526007326007328, "grad_norm": 0.7734375, "learning_rate": 4.513119498288605e-05, "loss": 0.1216, "step": 26653 }, { "epoch": 19.52673992673993, "grad_norm": 1.015625, "learning_rate": 4.5127788847405315e-05, "loss": 0.1349, "step": 26654 }, { "epoch": 19.52747252747253, "grad_norm": 1.2890625, "learning_rate": 4.512438273766461e-05, "loss": 0.1079, "step": 26655 }, { "epoch": 19.52820512820513, "grad_norm": 1.375, "learning_rate": 4.512097665368024e-05, "loss": 0.1316, "step": 26656 }, { "epoch": 19.52893772893773, "grad_norm": 1.4609375, "learning_rate": 4.5117570595468554e-05, "loss": 0.1583, "step": 26657 }, { "epoch": 19.52967032967033, "grad_norm": 1.7734375, "learning_rate": 4.511416456304584e-05, "loss": 0.1092, "step": 26658 }, { "epoch": 19.530402930402932, "grad_norm": 1.8515625, "learning_rate": 4.5110758556428425e-05, "loss": 0.1116, "step": 26659 }, { "epoch": 19.531135531135533, "grad_norm": 0.83984375, "learning_rate": 4.510735257563265e-05, "loss": 0.1076, "step": 26660 }, { "epoch": 19.531868131868134, "grad_norm": 0.8984375, "learning_rate": 4.5103946620674816e-05, "loss": 0.1144, "step": 26661 }, { "epoch": 19.532600732600734, "grad_norm": 1.4140625, "learning_rate": 4.510054069157125e-05, "loss": 0.1221, "step": 26662 }, { "epoch": 19.533333333333335, "grad_norm": 0.68359375, "learning_rate": 4.509713478833824e-05, "loss": 0.132, "step": 26663 }, { "epoch": 19.534065934065936, "grad_norm": 0.91796875, "learning_rate": 4.509372891099216e-05, "loss": 0.1069, "step": 26664 }, { "epoch": 19.534798534798536, "grad_norm": 1.109375, "learning_rate": 4.50903230595493e-05, "loss": 0.1275, "step": 26665 }, { "epoch": 19.535531135531137, "grad_norm": 0.92578125, "learning_rate": 4.5086917234025964e-05, "loss": 0.1361, "step": 26666 }, { "epoch": 19.536263736263738, "grad_norm": 0.80078125, "learning_rate": 4.5083511434438496e-05, "loss": 0.1162, "step": 26667 }, { "epoch": 19.53699633699634, "grad_norm": 0.91796875, "learning_rate": 4.508010566080318e-05, "loss": 0.1338, "step": 26668 }, { "epoch": 19.53772893772894, "grad_norm": 1.0546875, "learning_rate": 4.507669991313638e-05, "loss": 0.1308, "step": 26669 }, { "epoch": 19.53846153846154, "grad_norm": 1.203125, "learning_rate": 4.507329419145438e-05, "loss": 0.1663, "step": 26670 }, { "epoch": 19.53919413919414, "grad_norm": 0.984375, "learning_rate": 4.506988849577351e-05, "loss": 0.1101, "step": 26671 }, { "epoch": 19.53992673992674, "grad_norm": 1.2109375, "learning_rate": 4.50664828261101e-05, "loss": 0.0882, "step": 26672 }, { "epoch": 19.540659340659342, "grad_norm": 0.8125, "learning_rate": 4.5063077182480444e-05, "loss": 0.1115, "step": 26673 }, { "epoch": 19.541391941391943, "grad_norm": 0.82421875, "learning_rate": 4.5059671564900854e-05, "loss": 0.1121, "step": 26674 }, { "epoch": 19.542124542124544, "grad_norm": 0.70703125, "learning_rate": 4.5056265973387686e-05, "loss": 0.0995, "step": 26675 }, { "epoch": 19.542857142857144, "grad_norm": 1.203125, "learning_rate": 4.505286040795723e-05, "loss": 0.1312, "step": 26676 }, { "epoch": 19.543589743589745, "grad_norm": 0.76171875, "learning_rate": 4.504945486862582e-05, "loss": 0.134, "step": 26677 }, { "epoch": 19.544322344322346, "grad_norm": 1.3359375, "learning_rate": 4.504604935540975e-05, "loss": 0.1113, "step": 26678 }, { "epoch": 19.545054945054947, "grad_norm": 0.9921875, "learning_rate": 4.504264386832533e-05, "loss": 0.1283, "step": 26679 }, { "epoch": 19.545787545787547, "grad_norm": 1.171875, "learning_rate": 4.503923840738891e-05, "loss": 0.1446, "step": 26680 }, { "epoch": 19.546520146520148, "grad_norm": 0.96484375, "learning_rate": 4.50358329726168e-05, "loss": 0.1628, "step": 26681 }, { "epoch": 19.54725274725275, "grad_norm": 1.125, "learning_rate": 4.5032427564025294e-05, "loss": 0.1179, "step": 26682 }, { "epoch": 19.54798534798535, "grad_norm": 1.4375, "learning_rate": 4.502902218163073e-05, "loss": 0.1353, "step": 26683 }, { "epoch": 19.54871794871795, "grad_norm": 1.046875, "learning_rate": 4.502561682544941e-05, "loss": 0.1238, "step": 26684 }, { "epoch": 19.54945054945055, "grad_norm": 1.703125, "learning_rate": 4.502221149549765e-05, "loss": 0.1269, "step": 26685 }, { "epoch": 19.550183150183152, "grad_norm": 0.83203125, "learning_rate": 4.501880619179179e-05, "loss": 0.1075, "step": 26686 }, { "epoch": 19.550915750915753, "grad_norm": 1.046875, "learning_rate": 4.5015400914348136e-05, "loss": 0.112, "step": 26687 }, { "epoch": 19.551648351648353, "grad_norm": 1.1640625, "learning_rate": 4.501199566318299e-05, "loss": 0.0952, "step": 26688 }, { "epoch": 19.552380952380954, "grad_norm": 1.28125, "learning_rate": 4.500859043831265e-05, "loss": 0.1312, "step": 26689 }, { "epoch": 19.553113553113555, "grad_norm": 1.15625, "learning_rate": 4.5005185239753475e-05, "loss": 0.1337, "step": 26690 }, { "epoch": 19.553846153846155, "grad_norm": 0.67578125, "learning_rate": 4.5001780067521765e-05, "loss": 0.1042, "step": 26691 }, { "epoch": 19.554578754578756, "grad_norm": 1.03125, "learning_rate": 4.499837492163384e-05, "loss": 0.1211, "step": 26692 }, { "epoch": 19.555311355311357, "grad_norm": 1.6484375, "learning_rate": 4.499496980210598e-05, "loss": 0.115, "step": 26693 }, { "epoch": 19.556043956043958, "grad_norm": 0.52734375, "learning_rate": 4.4991564708954535e-05, "loss": 0.1214, "step": 26694 }, { "epoch": 19.55677655677656, "grad_norm": 0.8984375, "learning_rate": 4.498815964219583e-05, "loss": 0.0932, "step": 26695 }, { "epoch": 19.55750915750916, "grad_norm": 0.91015625, "learning_rate": 4.498475460184613e-05, "loss": 0.1189, "step": 26696 }, { "epoch": 19.55824175824176, "grad_norm": 0.98046875, "learning_rate": 4.498134958792181e-05, "loss": 0.122, "step": 26697 }, { "epoch": 19.55897435897436, "grad_norm": 1.1796875, "learning_rate": 4.4977944600439155e-05, "loss": 0.1309, "step": 26698 }, { "epoch": 19.55970695970696, "grad_norm": 0.9609375, "learning_rate": 4.497453963941447e-05, "loss": 0.1159, "step": 26699 }, { "epoch": 19.560439560439562, "grad_norm": 1.2421875, "learning_rate": 4.497113470486407e-05, "loss": 0.1239, "step": 26700 }, { "epoch": 19.561172161172163, "grad_norm": 0.87109375, "learning_rate": 4.49677297968043e-05, "loss": 0.1054, "step": 26701 }, { "epoch": 19.561904761904763, "grad_norm": 1.078125, "learning_rate": 4.496432491525144e-05, "loss": 0.1249, "step": 26702 }, { "epoch": 19.562637362637364, "grad_norm": 1.1953125, "learning_rate": 4.4960920060221827e-05, "loss": 0.1406, "step": 26703 }, { "epoch": 19.563369963369965, "grad_norm": 0.77734375, "learning_rate": 4.495751523173176e-05, "loss": 0.121, "step": 26704 }, { "epoch": 19.564102564102566, "grad_norm": 1.3359375, "learning_rate": 4.495411042979753e-05, "loss": 0.1074, "step": 26705 }, { "epoch": 19.564835164835166, "grad_norm": 1.15625, "learning_rate": 4.4950705654435514e-05, "loss": 0.1075, "step": 26706 }, { "epoch": 19.565567765567767, "grad_norm": 2.125, "learning_rate": 4.4947300905661975e-05, "loss": 0.1442, "step": 26707 }, { "epoch": 19.566300366300368, "grad_norm": 1.390625, "learning_rate": 4.494389618349324e-05, "loss": 0.1217, "step": 26708 }, { "epoch": 19.56703296703297, "grad_norm": 1.609375, "learning_rate": 4.494049148794563e-05, "loss": 0.1581, "step": 26709 }, { "epoch": 19.56776556776557, "grad_norm": 1.015625, "learning_rate": 4.493708681903542e-05, "loss": 0.1287, "step": 26710 }, { "epoch": 19.56849816849817, "grad_norm": 0.73046875, "learning_rate": 4.4933682176778985e-05, "loss": 0.1281, "step": 26711 }, { "epoch": 19.56923076923077, "grad_norm": 1.0390625, "learning_rate": 4.493027756119261e-05, "loss": 0.1263, "step": 26712 }, { "epoch": 19.56996336996337, "grad_norm": 0.87109375, "learning_rate": 4.492687297229259e-05, "loss": 0.093, "step": 26713 }, { "epoch": 19.570695970695972, "grad_norm": 1.96875, "learning_rate": 4.492346841009526e-05, "loss": 0.1574, "step": 26714 }, { "epoch": 19.571428571428573, "grad_norm": 1.015625, "learning_rate": 4.4920063874616896e-05, "loss": 0.1113, "step": 26715 }, { "epoch": 19.572161172161174, "grad_norm": 1.21875, "learning_rate": 4.491665936587386e-05, "loss": 0.1024, "step": 26716 }, { "epoch": 19.572893772893774, "grad_norm": 0.98828125, "learning_rate": 4.491325488388245e-05, "loss": 0.1227, "step": 26717 }, { "epoch": 19.573626373626375, "grad_norm": 1.2734375, "learning_rate": 4.4909850428658965e-05, "loss": 0.1269, "step": 26718 }, { "epoch": 19.574358974358976, "grad_norm": 0.94140625, "learning_rate": 4.490644600021971e-05, "loss": 0.1457, "step": 26719 }, { "epoch": 19.575091575091577, "grad_norm": 1.15625, "learning_rate": 4.4903041598581005e-05, "loss": 0.0856, "step": 26720 }, { "epoch": 19.575824175824177, "grad_norm": 1.1640625, "learning_rate": 4.4899637223759177e-05, "loss": 0.111, "step": 26721 }, { "epoch": 19.576556776556778, "grad_norm": 1.25, "learning_rate": 4.489623287577052e-05, "loss": 0.1278, "step": 26722 }, { "epoch": 19.57728937728938, "grad_norm": 0.76953125, "learning_rate": 4.489282855463136e-05, "loss": 0.1154, "step": 26723 }, { "epoch": 19.57802197802198, "grad_norm": 1.7109375, "learning_rate": 4.4889424260357995e-05, "loss": 0.1136, "step": 26724 }, { "epoch": 19.57875457875458, "grad_norm": 0.76171875, "learning_rate": 4.488601999296675e-05, "loss": 0.1036, "step": 26725 }, { "epoch": 19.57948717948718, "grad_norm": 0.6484375, "learning_rate": 4.488261575247389e-05, "loss": 0.1131, "step": 26726 }, { "epoch": 19.58021978021978, "grad_norm": 1.2578125, "learning_rate": 4.48792115388958e-05, "loss": 0.1196, "step": 26727 }, { "epoch": 19.580952380952382, "grad_norm": 1.5546875, "learning_rate": 4.487580735224874e-05, "loss": 0.1068, "step": 26728 }, { "epoch": 19.581684981684983, "grad_norm": 1.46875, "learning_rate": 4.4872403192549036e-05, "loss": 0.1137, "step": 26729 }, { "epoch": 19.582417582417584, "grad_norm": 2.015625, "learning_rate": 4.4868999059813e-05, "loss": 0.1641, "step": 26730 }, { "epoch": 19.583150183150185, "grad_norm": 1.015625, "learning_rate": 4.486559495405691e-05, "loss": 0.1179, "step": 26731 }, { "epoch": 19.583882783882785, "grad_norm": 0.82421875, "learning_rate": 4.4862190875297134e-05, "loss": 0.1325, "step": 26732 }, { "epoch": 19.584615384615386, "grad_norm": 1.0078125, "learning_rate": 4.485878682354995e-05, "loss": 0.1226, "step": 26733 }, { "epoch": 19.585347985347987, "grad_norm": 1.59375, "learning_rate": 4.485538279883165e-05, "loss": 0.1253, "step": 26734 }, { "epoch": 19.586080586080588, "grad_norm": 1.0390625, "learning_rate": 4.485197880115859e-05, "loss": 0.1414, "step": 26735 }, { "epoch": 19.58681318681319, "grad_norm": 0.7578125, "learning_rate": 4.484857483054702e-05, "loss": 0.1176, "step": 26736 }, { "epoch": 19.58754578754579, "grad_norm": 0.8203125, "learning_rate": 4.484517088701331e-05, "loss": 0.1181, "step": 26737 }, { "epoch": 19.58827838827839, "grad_norm": 0.7265625, "learning_rate": 4.484176697057375e-05, "loss": 0.1294, "step": 26738 }, { "epoch": 19.58901098901099, "grad_norm": 1.125, "learning_rate": 4.483836308124462e-05, "loss": 0.0839, "step": 26739 }, { "epoch": 19.58974358974359, "grad_norm": 0.8828125, "learning_rate": 4.4834959219042274e-05, "loss": 0.1347, "step": 26740 }, { "epoch": 19.590476190476192, "grad_norm": 0.9765625, "learning_rate": 4.483155538398296e-05, "loss": 0.1032, "step": 26741 }, { "epoch": 19.591208791208793, "grad_norm": 0.8046875, "learning_rate": 4.482815157608306e-05, "loss": 0.0856, "step": 26742 }, { "epoch": 19.591941391941393, "grad_norm": 1.046875, "learning_rate": 4.4824747795358835e-05, "loss": 0.1197, "step": 26743 }, { "epoch": 19.592673992673994, "grad_norm": 0.86328125, "learning_rate": 4.482134404182661e-05, "loss": 0.1286, "step": 26744 }, { "epoch": 19.593406593406595, "grad_norm": 1.0078125, "learning_rate": 4.481794031550268e-05, "loss": 0.1342, "step": 26745 }, { "epoch": 19.594139194139196, "grad_norm": 1.0390625, "learning_rate": 4.481453661640337e-05, "loss": 0.1312, "step": 26746 }, { "epoch": 19.594871794871796, "grad_norm": 0.95703125, "learning_rate": 4.481113294454499e-05, "loss": 0.0959, "step": 26747 }, { "epoch": 19.595604395604397, "grad_norm": 0.9921875, "learning_rate": 4.480772929994383e-05, "loss": 0.1258, "step": 26748 }, { "epoch": 19.596336996336998, "grad_norm": 0.9375, "learning_rate": 4.480432568261621e-05, "loss": 0.1197, "step": 26749 }, { "epoch": 19.5970695970696, "grad_norm": 0.84375, "learning_rate": 4.480092209257844e-05, "loss": 0.1265, "step": 26750 }, { "epoch": 19.5978021978022, "grad_norm": 0.84765625, "learning_rate": 4.4797518529846825e-05, "loss": 0.1318, "step": 26751 }, { "epoch": 19.5985347985348, "grad_norm": 1.1875, "learning_rate": 4.479411499443765e-05, "loss": 0.1443, "step": 26752 }, { "epoch": 19.5992673992674, "grad_norm": 1.078125, "learning_rate": 4.479071148636726e-05, "loss": 0.1516, "step": 26753 }, { "epoch": 19.6, "grad_norm": 0.89453125, "learning_rate": 4.4787308005651945e-05, "loss": 0.1285, "step": 26754 }, { "epoch": 19.600732600732602, "grad_norm": 0.828125, "learning_rate": 4.478390455230802e-05, "loss": 0.1302, "step": 26755 }, { "epoch": 19.601465201465203, "grad_norm": 1.0390625, "learning_rate": 4.4780501126351775e-05, "loss": 0.1099, "step": 26756 }, { "epoch": 19.602197802197804, "grad_norm": 0.89453125, "learning_rate": 4.477709772779951e-05, "loss": 0.1045, "step": 26757 }, { "epoch": 19.602930402930404, "grad_norm": 0.91796875, "learning_rate": 4.4773694356667574e-05, "loss": 0.1353, "step": 26758 }, { "epoch": 19.603663003663005, "grad_norm": 0.77734375, "learning_rate": 4.477029101297223e-05, "loss": 0.0982, "step": 26759 }, { "epoch": 19.604395604395606, "grad_norm": 0.86328125, "learning_rate": 4.476688769672982e-05, "loss": 0.0984, "step": 26760 }, { "epoch": 19.605128205128207, "grad_norm": 1.609375, "learning_rate": 4.476348440795662e-05, "loss": 0.1847, "step": 26761 }, { "epoch": 19.605860805860807, "grad_norm": 1.2421875, "learning_rate": 4.4760081146668945e-05, "loss": 0.1125, "step": 26762 }, { "epoch": 19.606593406593408, "grad_norm": 0.94140625, "learning_rate": 4.475667791288312e-05, "loss": 0.1111, "step": 26763 }, { "epoch": 19.60732600732601, "grad_norm": 0.64453125, "learning_rate": 4.4753274706615436e-05, "loss": 0.1029, "step": 26764 }, { "epoch": 19.60805860805861, "grad_norm": 0.67578125, "learning_rate": 4.4749871527882206e-05, "loss": 0.098, "step": 26765 }, { "epoch": 19.60879120879121, "grad_norm": 0.70703125, "learning_rate": 4.474646837669972e-05, "loss": 0.1083, "step": 26766 }, { "epoch": 19.60952380952381, "grad_norm": 1.0546875, "learning_rate": 4.4743065253084276e-05, "loss": 0.1186, "step": 26767 }, { "epoch": 19.61025641025641, "grad_norm": 1.1484375, "learning_rate": 4.473966215705221e-05, "loss": 0.1314, "step": 26768 }, { "epoch": 19.610989010989012, "grad_norm": 0.80859375, "learning_rate": 4.473625908861983e-05, "loss": 0.1305, "step": 26769 }, { "epoch": 19.611721611721613, "grad_norm": 1.2109375, "learning_rate": 4.473285604780341e-05, "loss": 0.1143, "step": 26770 }, { "epoch": 19.612454212454214, "grad_norm": 1.2109375, "learning_rate": 4.4729453034619264e-05, "loss": 0.0938, "step": 26771 }, { "epoch": 19.613186813186815, "grad_norm": 1.796875, "learning_rate": 4.47260500490837e-05, "loss": 0.1336, "step": 26772 }, { "epoch": 19.613919413919415, "grad_norm": 1.3046875, "learning_rate": 4.472264709121304e-05, "loss": 0.106, "step": 26773 }, { "epoch": 19.614652014652016, "grad_norm": 0.93359375, "learning_rate": 4.471924416102356e-05, "loss": 0.1198, "step": 26774 }, { "epoch": 19.615384615384617, "grad_norm": 1.4140625, "learning_rate": 4.4715841258531595e-05, "loss": 0.1348, "step": 26775 }, { "epoch": 19.616117216117217, "grad_norm": 1.03125, "learning_rate": 4.471243838375343e-05, "loss": 0.0869, "step": 26776 }, { "epoch": 19.616849816849818, "grad_norm": 1.359375, "learning_rate": 4.4709035536705375e-05, "loss": 0.1133, "step": 26777 }, { "epoch": 19.61758241758242, "grad_norm": 1.1484375, "learning_rate": 4.4705632717403705e-05, "loss": 0.1492, "step": 26778 }, { "epoch": 19.61831501831502, "grad_norm": 0.8203125, "learning_rate": 4.470222992586478e-05, "loss": 0.1135, "step": 26779 }, { "epoch": 19.61904761904762, "grad_norm": 1.328125, "learning_rate": 4.469882716210487e-05, "loss": 0.1312, "step": 26780 }, { "epoch": 19.61978021978022, "grad_norm": 0.70703125, "learning_rate": 4.469542442614028e-05, "loss": 0.087, "step": 26781 }, { "epoch": 19.620512820512822, "grad_norm": 1.0390625, "learning_rate": 4.4692021717987324e-05, "loss": 0.1423, "step": 26782 }, { "epoch": 19.621245421245423, "grad_norm": 0.875, "learning_rate": 4.468861903766228e-05, "loss": 0.0898, "step": 26783 }, { "epoch": 19.621978021978023, "grad_norm": 1.0859375, "learning_rate": 4.468521638518148e-05, "loss": 0.1217, "step": 26784 }, { "epoch": 19.622710622710624, "grad_norm": 1.4296875, "learning_rate": 4.4681813760561216e-05, "loss": 0.121, "step": 26785 }, { "epoch": 19.623443223443225, "grad_norm": 1.234375, "learning_rate": 4.46784111638178e-05, "loss": 0.1449, "step": 26786 }, { "epoch": 19.624175824175826, "grad_norm": 0.5546875, "learning_rate": 4.467500859496752e-05, "loss": 0.0952, "step": 26787 }, { "epoch": 19.624908424908426, "grad_norm": 1.15625, "learning_rate": 4.467160605402667e-05, "loss": 0.1209, "step": 26788 }, { "epoch": 19.625641025641027, "grad_norm": 0.7578125, "learning_rate": 4.4668203541011594e-05, "loss": 0.1157, "step": 26789 }, { "epoch": 19.626373626373628, "grad_norm": 0.94921875, "learning_rate": 4.4664801055938567e-05, "loss": 0.1532, "step": 26790 }, { "epoch": 19.62710622710623, "grad_norm": 0.94140625, "learning_rate": 4.4661398598823895e-05, "loss": 0.096, "step": 26791 }, { "epoch": 19.62783882783883, "grad_norm": 0.80859375, "learning_rate": 4.4657996169683875e-05, "loss": 0.1365, "step": 26792 }, { "epoch": 19.62857142857143, "grad_norm": 0.9921875, "learning_rate": 4.465459376853479e-05, "loss": 0.1151, "step": 26793 }, { "epoch": 19.62930402930403, "grad_norm": 1.375, "learning_rate": 4.465119139539299e-05, "loss": 0.1336, "step": 26794 }, { "epoch": 19.63003663003663, "grad_norm": 1.0703125, "learning_rate": 4.464778905027475e-05, "loss": 0.1142, "step": 26795 }, { "epoch": 19.630769230769232, "grad_norm": 1.5078125, "learning_rate": 4.4644386733196385e-05, "loss": 0.1039, "step": 26796 }, { "epoch": 19.631501831501833, "grad_norm": 1.203125, "learning_rate": 4.464098444417417e-05, "loss": 0.1586, "step": 26797 }, { "epoch": 19.632234432234434, "grad_norm": 1.125, "learning_rate": 4.4637582183224406e-05, "loss": 0.1254, "step": 26798 }, { "epoch": 19.632967032967034, "grad_norm": 1.5078125, "learning_rate": 4.4634179950363435e-05, "loss": 0.1559, "step": 26799 }, { "epoch": 19.633699633699635, "grad_norm": 0.99609375, "learning_rate": 4.4630777745607525e-05, "loss": 0.1151, "step": 26800 }, { "epoch": 19.634432234432236, "grad_norm": 0.9609375, "learning_rate": 4.462737556897299e-05, "loss": 0.1482, "step": 26801 }, { "epoch": 19.635164835164836, "grad_norm": 1.015625, "learning_rate": 4.462397342047613e-05, "loss": 0.1741, "step": 26802 }, { "epoch": 19.635897435897437, "grad_norm": 0.90234375, "learning_rate": 4.4620571300133235e-05, "loss": 0.1396, "step": 26803 }, { "epoch": 19.636630036630038, "grad_norm": 1.2421875, "learning_rate": 4.46171692079606e-05, "loss": 0.1345, "step": 26804 }, { "epoch": 19.63736263736264, "grad_norm": 1.2890625, "learning_rate": 4.461376714397455e-05, "loss": 0.1101, "step": 26805 }, { "epoch": 19.63809523809524, "grad_norm": 0.66796875, "learning_rate": 4.4610365108191396e-05, "loss": 0.1052, "step": 26806 }, { "epoch": 19.63882783882784, "grad_norm": 1.125, "learning_rate": 4.46069631006274e-05, "loss": 0.1207, "step": 26807 }, { "epoch": 19.63956043956044, "grad_norm": 0.9921875, "learning_rate": 4.460356112129887e-05, "loss": 0.126, "step": 26808 }, { "epoch": 19.64029304029304, "grad_norm": 1.375, "learning_rate": 4.4600159170222114e-05, "loss": 0.1239, "step": 26809 }, { "epoch": 19.641025641025642, "grad_norm": 1.375, "learning_rate": 4.459675724741344e-05, "loss": 0.1254, "step": 26810 }, { "epoch": 19.641758241758243, "grad_norm": 1.296875, "learning_rate": 4.459335535288913e-05, "loss": 0.1119, "step": 26811 }, { "epoch": 19.642490842490844, "grad_norm": 0.8515625, "learning_rate": 4.458995348666551e-05, "loss": 0.1277, "step": 26812 }, { "epoch": 19.643223443223444, "grad_norm": 0.8984375, "learning_rate": 4.4586551648758855e-05, "loss": 0.1124, "step": 26813 }, { "epoch": 19.643956043956045, "grad_norm": 1.1171875, "learning_rate": 4.458314983918545e-05, "loss": 0.1004, "step": 26814 }, { "epoch": 19.644688644688646, "grad_norm": 0.96484375, "learning_rate": 4.4579748057961645e-05, "loss": 0.1117, "step": 26815 }, { "epoch": 19.645421245421247, "grad_norm": 1.2578125, "learning_rate": 4.457634630510371e-05, "loss": 0.1064, "step": 26816 }, { "epoch": 19.646153846153847, "grad_norm": 1.0859375, "learning_rate": 4.457294458062794e-05, "loss": 0.1397, "step": 26817 }, { "epoch": 19.646886446886448, "grad_norm": 1.1796875, "learning_rate": 4.456954288455065e-05, "loss": 0.1258, "step": 26818 }, { "epoch": 19.64761904761905, "grad_norm": 1.3046875, "learning_rate": 4.4566141216888084e-05, "loss": 0.0972, "step": 26819 }, { "epoch": 19.64835164835165, "grad_norm": 1.640625, "learning_rate": 4.456273957765663e-05, "loss": 0.1528, "step": 26820 }, { "epoch": 19.64908424908425, "grad_norm": 1.4296875, "learning_rate": 4.455933796687252e-05, "loss": 0.1187, "step": 26821 }, { "epoch": 19.64981684981685, "grad_norm": 0.86328125, "learning_rate": 4.455593638455208e-05, "loss": 0.1309, "step": 26822 }, { "epoch": 19.650549450549452, "grad_norm": 1.1015625, "learning_rate": 4.455253483071158e-05, "loss": 0.1457, "step": 26823 }, { "epoch": 19.651282051282053, "grad_norm": 0.96484375, "learning_rate": 4.4549133305367365e-05, "loss": 0.1438, "step": 26824 }, { "epoch": 19.652014652014653, "grad_norm": 1.171875, "learning_rate": 4.4545731808535673e-05, "loss": 0.1421, "step": 26825 }, { "epoch": 19.652747252747254, "grad_norm": 0.98046875, "learning_rate": 4.454233034023286e-05, "loss": 0.1279, "step": 26826 }, { "epoch": 19.653479853479855, "grad_norm": 1.90625, "learning_rate": 4.4538928900475195e-05, "loss": 0.1232, "step": 26827 }, { "epoch": 19.654212454212455, "grad_norm": 0.69921875, "learning_rate": 4.453552748927898e-05, "loss": 0.1134, "step": 26828 }, { "epoch": 19.654945054945056, "grad_norm": 1.4609375, "learning_rate": 4.4532126106660514e-05, "loss": 0.1346, "step": 26829 }, { "epoch": 19.655677655677657, "grad_norm": 0.78515625, "learning_rate": 4.452872475263607e-05, "loss": 0.1299, "step": 26830 }, { "epoch": 19.656410256410258, "grad_norm": 1.265625, "learning_rate": 4.452532342722199e-05, "loss": 0.1391, "step": 26831 }, { "epoch": 19.65714285714286, "grad_norm": 0.7578125, "learning_rate": 4.4521922130434534e-05, "loss": 0.1395, "step": 26832 }, { "epoch": 19.65787545787546, "grad_norm": 1.078125, "learning_rate": 4.451852086229002e-05, "loss": 0.1314, "step": 26833 }, { "epoch": 19.65860805860806, "grad_norm": 1.4765625, "learning_rate": 4.451511962280473e-05, "loss": 0.1552, "step": 26834 }, { "epoch": 19.65934065934066, "grad_norm": 0.69921875, "learning_rate": 4.4511718411994954e-05, "loss": 0.1203, "step": 26835 }, { "epoch": 19.66007326007326, "grad_norm": 0.97265625, "learning_rate": 4.4508317229877026e-05, "loss": 0.1104, "step": 26836 }, { "epoch": 19.660805860805862, "grad_norm": 1.2890625, "learning_rate": 4.45049160764672e-05, "loss": 0.1242, "step": 26837 }, { "epoch": 19.661538461538463, "grad_norm": 0.9375, "learning_rate": 4.45015149517818e-05, "loss": 0.1217, "step": 26838 }, { "epoch": 19.662271062271063, "grad_norm": 1.0859375, "learning_rate": 4.4498113855837105e-05, "loss": 0.0874, "step": 26839 }, { "epoch": 19.663003663003664, "grad_norm": 1.0625, "learning_rate": 4.44947127886494e-05, "loss": 0.1198, "step": 26840 }, { "epoch": 19.663736263736265, "grad_norm": 0.91796875, "learning_rate": 4.449131175023502e-05, "loss": 0.1003, "step": 26841 }, { "epoch": 19.664468864468866, "grad_norm": 0.72265625, "learning_rate": 4.448791074061024e-05, "loss": 0.1043, "step": 26842 }, { "epoch": 19.665201465201466, "grad_norm": 1.484375, "learning_rate": 4.448450975979135e-05, "loss": 0.1068, "step": 26843 }, { "epoch": 19.665934065934067, "grad_norm": 1.3359375, "learning_rate": 4.4481108807794646e-05, "loss": 0.1265, "step": 26844 }, { "epoch": 19.666666666666668, "grad_norm": 1.078125, "learning_rate": 4.447770788463641e-05, "loss": 0.0889, "step": 26845 }, { "epoch": 19.66739926739927, "grad_norm": 0.94140625, "learning_rate": 4.447430699033296e-05, "loss": 0.1384, "step": 26846 }, { "epoch": 19.66813186813187, "grad_norm": 1.3125, "learning_rate": 4.44709061249006e-05, "loss": 0.1249, "step": 26847 }, { "epoch": 19.66886446886447, "grad_norm": 0.765625, "learning_rate": 4.4467505288355603e-05, "loss": 0.0998, "step": 26848 }, { "epoch": 19.66959706959707, "grad_norm": 1.4453125, "learning_rate": 4.446410448071426e-05, "loss": 0.1303, "step": 26849 }, { "epoch": 19.67032967032967, "grad_norm": 1.21875, "learning_rate": 4.446070370199288e-05, "loss": 0.1397, "step": 26850 }, { "epoch": 19.671062271062272, "grad_norm": 1.2578125, "learning_rate": 4.445730295220773e-05, "loss": 0.1387, "step": 26851 }, { "epoch": 19.671794871794873, "grad_norm": 1.03125, "learning_rate": 4.445390223137515e-05, "loss": 0.1094, "step": 26852 }, { "epoch": 19.672527472527474, "grad_norm": 2.28125, "learning_rate": 4.445050153951141e-05, "loss": 0.134, "step": 26853 }, { "epoch": 19.673260073260074, "grad_norm": 1.9453125, "learning_rate": 4.4447100876632795e-05, "loss": 0.1272, "step": 26854 }, { "epoch": 19.673992673992675, "grad_norm": 1.09375, "learning_rate": 4.4443700242755614e-05, "loss": 0.1268, "step": 26855 }, { "epoch": 19.674725274725276, "grad_norm": 1.703125, "learning_rate": 4.4440299637896134e-05, "loss": 0.1677, "step": 26856 }, { "epoch": 19.675457875457877, "grad_norm": 0.8828125, "learning_rate": 4.443689906207068e-05, "loss": 0.1327, "step": 26857 }, { "epoch": 19.676190476190477, "grad_norm": 0.6796875, "learning_rate": 4.443349851529554e-05, "loss": 0.1125, "step": 26858 }, { "epoch": 19.676923076923078, "grad_norm": 1.21875, "learning_rate": 4.4430097997587e-05, "loss": 0.1551, "step": 26859 }, { "epoch": 19.67765567765568, "grad_norm": 0.953125, "learning_rate": 4.442669750896134e-05, "loss": 0.1171, "step": 26860 }, { "epoch": 19.67838827838828, "grad_norm": 1.53125, "learning_rate": 4.4423297049434866e-05, "loss": 0.1372, "step": 26861 }, { "epoch": 19.67912087912088, "grad_norm": 0.7890625, "learning_rate": 4.441989661902388e-05, "loss": 0.1258, "step": 26862 }, { "epoch": 19.67985347985348, "grad_norm": 0.7734375, "learning_rate": 4.441649621774466e-05, "loss": 0.1028, "step": 26863 }, { "epoch": 19.68058608058608, "grad_norm": 1.1484375, "learning_rate": 4.441309584561351e-05, "loss": 0.143, "step": 26864 }, { "epoch": 19.681318681318682, "grad_norm": 1.015625, "learning_rate": 4.440969550264672e-05, "loss": 0.0973, "step": 26865 }, { "epoch": 19.682051282051283, "grad_norm": 0.90625, "learning_rate": 4.440629518886055e-05, "loss": 0.1247, "step": 26866 }, { "epoch": 19.682783882783884, "grad_norm": 0.91796875, "learning_rate": 4.440289490427134e-05, "loss": 0.1359, "step": 26867 }, { "epoch": 19.683516483516485, "grad_norm": 1.03125, "learning_rate": 4.439949464889537e-05, "loss": 0.1291, "step": 26868 }, { "epoch": 19.684249084249085, "grad_norm": 1.234375, "learning_rate": 4.439609442274892e-05, "loss": 0.1307, "step": 26869 }, { "epoch": 19.684981684981686, "grad_norm": 1.5078125, "learning_rate": 4.4392694225848294e-05, "loss": 0.144, "step": 26870 }, { "epoch": 19.685714285714287, "grad_norm": 1.3828125, "learning_rate": 4.4389294058209746e-05, "loss": 0.1637, "step": 26871 }, { "epoch": 19.686446886446888, "grad_norm": 1.2578125, "learning_rate": 4.438589391984961e-05, "loss": 0.1017, "step": 26872 }, { "epoch": 19.68717948717949, "grad_norm": 1.4921875, "learning_rate": 4.438249381078418e-05, "loss": 0.1223, "step": 26873 }, { "epoch": 19.68791208791209, "grad_norm": 0.74609375, "learning_rate": 4.437909373102971e-05, "loss": 0.0953, "step": 26874 }, { "epoch": 19.68864468864469, "grad_norm": 1.53125, "learning_rate": 4.4375693680602523e-05, "loss": 0.1481, "step": 26875 }, { "epoch": 19.68937728937729, "grad_norm": 0.8515625, "learning_rate": 4.437229365951889e-05, "loss": 0.1313, "step": 26876 }, { "epoch": 19.69010989010989, "grad_norm": 1.0859375, "learning_rate": 4.4368893667795106e-05, "loss": 0.1465, "step": 26877 }, { "epoch": 19.690842490842492, "grad_norm": 1.09375, "learning_rate": 4.4365493705447475e-05, "loss": 0.0994, "step": 26878 }, { "epoch": 19.691575091575093, "grad_norm": 0.83203125, "learning_rate": 4.4362093772492273e-05, "loss": 0.1058, "step": 26879 }, { "epoch": 19.692307692307693, "grad_norm": 0.85546875, "learning_rate": 4.43586938689458e-05, "loss": 0.1126, "step": 26880 }, { "epoch": 19.693040293040294, "grad_norm": 1.109375, "learning_rate": 4.4355293994824346e-05, "loss": 0.123, "step": 26881 }, { "epoch": 19.693772893772895, "grad_norm": 0.75390625, "learning_rate": 4.435189415014416e-05, "loss": 0.1069, "step": 26882 }, { "epoch": 19.694505494505496, "grad_norm": 1.296875, "learning_rate": 4.4348494334921606e-05, "loss": 0.1256, "step": 26883 }, { "epoch": 19.695238095238096, "grad_norm": 1.6796875, "learning_rate": 4.434509454917293e-05, "loss": 0.1126, "step": 26884 }, { "epoch": 19.695970695970697, "grad_norm": 1.0078125, "learning_rate": 4.4341694792914426e-05, "loss": 0.1082, "step": 26885 }, { "epoch": 19.696703296703298, "grad_norm": 0.83203125, "learning_rate": 4.433829506616237e-05, "loss": 0.1072, "step": 26886 }, { "epoch": 19.6974358974359, "grad_norm": 1.7265625, "learning_rate": 4.433489536893306e-05, "loss": 0.1276, "step": 26887 }, { "epoch": 19.6981684981685, "grad_norm": 0.8203125, "learning_rate": 4.433149570124281e-05, "loss": 0.106, "step": 26888 }, { "epoch": 19.6989010989011, "grad_norm": 0.9453125, "learning_rate": 4.4328096063107874e-05, "loss": 0.1034, "step": 26889 }, { "epoch": 19.6996336996337, "grad_norm": 1.140625, "learning_rate": 4.432469645454457e-05, "loss": 0.1224, "step": 26890 }, { "epoch": 19.7003663003663, "grad_norm": 1.171875, "learning_rate": 4.432129687556918e-05, "loss": 0.1093, "step": 26891 }, { "epoch": 19.701098901098902, "grad_norm": 0.76171875, "learning_rate": 4.431789732619795e-05, "loss": 0.114, "step": 26892 }, { "epoch": 19.701831501831503, "grad_norm": 0.7421875, "learning_rate": 4.4314497806447224e-05, "loss": 0.1029, "step": 26893 }, { "epoch": 19.702564102564104, "grad_norm": 1.390625, "learning_rate": 4.4311098316333275e-05, "loss": 0.1574, "step": 26894 }, { "epoch": 19.703296703296704, "grad_norm": 0.62109375, "learning_rate": 4.430769885587239e-05, "loss": 0.1278, "step": 26895 }, { "epoch": 19.704029304029305, "grad_norm": 0.91796875, "learning_rate": 4.430429942508084e-05, "loss": 0.156, "step": 26896 }, { "epoch": 19.704761904761906, "grad_norm": 0.8671875, "learning_rate": 4.430090002397491e-05, "loss": 0.1191, "step": 26897 }, { "epoch": 19.705494505494507, "grad_norm": 0.7578125, "learning_rate": 4.4297500652570934e-05, "loss": 0.1044, "step": 26898 }, { "epoch": 19.706227106227107, "grad_norm": 1.1640625, "learning_rate": 4.4294101310885164e-05, "loss": 0.1324, "step": 26899 }, { "epoch": 19.706959706959708, "grad_norm": 0.91796875, "learning_rate": 4.4290701998933875e-05, "loss": 0.1312, "step": 26900 }, { "epoch": 19.70769230769231, "grad_norm": 0.9765625, "learning_rate": 4.428730271673338e-05, "loss": 0.1215, "step": 26901 }, { "epoch": 19.70842490842491, "grad_norm": 1.171875, "learning_rate": 4.4283903464299964e-05, "loss": 0.1464, "step": 26902 }, { "epoch": 19.70915750915751, "grad_norm": 1.53125, "learning_rate": 4.428050424164988e-05, "loss": 0.1406, "step": 26903 }, { "epoch": 19.70989010989011, "grad_norm": 0.65234375, "learning_rate": 4.427710504879947e-05, "loss": 0.0955, "step": 26904 }, { "epoch": 19.71062271062271, "grad_norm": 1.421875, "learning_rate": 4.4273705885764995e-05, "loss": 0.0893, "step": 26905 }, { "epoch": 19.711355311355312, "grad_norm": 1.0, "learning_rate": 4.427030675256274e-05, "loss": 0.1341, "step": 26906 }, { "epoch": 19.712087912087913, "grad_norm": 0.7890625, "learning_rate": 4.426690764920897e-05, "loss": 0.1273, "step": 26907 }, { "epoch": 19.712820512820514, "grad_norm": 0.9921875, "learning_rate": 4.426350857571999e-05, "loss": 0.116, "step": 26908 }, { "epoch": 19.713553113553115, "grad_norm": 1.015625, "learning_rate": 4.42601095321121e-05, "loss": 0.1244, "step": 26909 }, { "epoch": 19.714285714285715, "grad_norm": 1.40625, "learning_rate": 4.425671051840158e-05, "loss": 0.1417, "step": 26910 }, { "epoch": 19.715018315018316, "grad_norm": 1.625, "learning_rate": 4.4253311534604704e-05, "loss": 0.1126, "step": 26911 }, { "epoch": 19.715750915750917, "grad_norm": 1.0703125, "learning_rate": 4.424991258073776e-05, "loss": 0.1062, "step": 26912 }, { "epoch": 19.716483516483517, "grad_norm": 1.0390625, "learning_rate": 4.424651365681703e-05, "loss": 0.128, "step": 26913 }, { "epoch": 19.717216117216118, "grad_norm": 1.15625, "learning_rate": 4.424311476285881e-05, "loss": 0.1478, "step": 26914 }, { "epoch": 19.71794871794872, "grad_norm": 1.1796875, "learning_rate": 4.4239715898879384e-05, "loss": 0.0996, "step": 26915 }, { "epoch": 19.71868131868132, "grad_norm": 0.7578125, "learning_rate": 4.423631706489504e-05, "loss": 0.1197, "step": 26916 }, { "epoch": 19.71941391941392, "grad_norm": 1.0, "learning_rate": 4.423291826092205e-05, "loss": 0.1382, "step": 26917 }, { "epoch": 19.72014652014652, "grad_norm": 1.3125, "learning_rate": 4.4229519486976696e-05, "loss": 0.1097, "step": 26918 }, { "epoch": 19.720879120879122, "grad_norm": 0.92578125, "learning_rate": 4.422612074307529e-05, "loss": 0.1205, "step": 26919 }, { "epoch": 19.721611721611723, "grad_norm": 1.1875, "learning_rate": 4.42227220292341e-05, "loss": 0.1229, "step": 26920 }, { "epoch": 19.722344322344323, "grad_norm": 1.1796875, "learning_rate": 4.421932334546941e-05, "loss": 0.1258, "step": 26921 }, { "epoch": 19.723076923076924, "grad_norm": 0.6875, "learning_rate": 4.42159246917975e-05, "loss": 0.1015, "step": 26922 }, { "epoch": 19.723809523809525, "grad_norm": 0.765625, "learning_rate": 4.421252606823463e-05, "loss": 0.1034, "step": 26923 }, { "epoch": 19.724542124542126, "grad_norm": 1.140625, "learning_rate": 4.4209127474797146e-05, "loss": 0.0967, "step": 26924 }, { "epoch": 19.725274725274726, "grad_norm": 1.234375, "learning_rate": 4.4205728911501295e-05, "loss": 0.1195, "step": 26925 }, { "epoch": 19.726007326007327, "grad_norm": 1.3125, "learning_rate": 4.420233037836335e-05, "loss": 0.1224, "step": 26926 }, { "epoch": 19.726739926739928, "grad_norm": 1.1953125, "learning_rate": 4.419893187539962e-05, "loss": 0.1262, "step": 26927 }, { "epoch": 19.72747252747253, "grad_norm": 1.6875, "learning_rate": 4.419553340262637e-05, "loss": 0.147, "step": 26928 }, { "epoch": 19.72820512820513, "grad_norm": 1.2265625, "learning_rate": 4.419213496005987e-05, "loss": 0.1096, "step": 26929 }, { "epoch": 19.72893772893773, "grad_norm": 0.921875, "learning_rate": 4.4188736547716445e-05, "loss": 0.1309, "step": 26930 }, { "epoch": 19.72967032967033, "grad_norm": 1.15625, "learning_rate": 4.418533816561236e-05, "loss": 0.1128, "step": 26931 }, { "epoch": 19.73040293040293, "grad_norm": 0.96875, "learning_rate": 4.418193981376388e-05, "loss": 0.1064, "step": 26932 }, { "epoch": 19.731135531135532, "grad_norm": 1.5234375, "learning_rate": 4.417854149218731e-05, "loss": 0.0885, "step": 26933 }, { "epoch": 19.731868131868133, "grad_norm": 1.4140625, "learning_rate": 4.41751432008989e-05, "loss": 0.1654, "step": 26934 }, { "epoch": 19.732600732600734, "grad_norm": 1.3125, "learning_rate": 4.417174493991498e-05, "loss": 0.1158, "step": 26935 }, { "epoch": 19.733333333333334, "grad_norm": 1.1875, "learning_rate": 4.41683467092518e-05, "loss": 0.1491, "step": 26936 }, { "epoch": 19.734065934065935, "grad_norm": 1.140625, "learning_rate": 4.416494850892566e-05, "loss": 0.1033, "step": 26937 }, { "epoch": 19.734798534798536, "grad_norm": 1.0703125, "learning_rate": 4.4161550338952815e-05, "loss": 0.15, "step": 26938 }, { "epoch": 19.735531135531136, "grad_norm": 1.2890625, "learning_rate": 4.415815219934956e-05, "loss": 0.1212, "step": 26939 }, { "epoch": 19.736263736263737, "grad_norm": 1.15625, "learning_rate": 4.415475409013218e-05, "loss": 0.1098, "step": 26940 }, { "epoch": 19.736996336996338, "grad_norm": 1.5390625, "learning_rate": 4.415135601131697e-05, "loss": 0.1485, "step": 26941 }, { "epoch": 19.73772893772894, "grad_norm": 1.2421875, "learning_rate": 4.4147957962920194e-05, "loss": 0.1407, "step": 26942 }, { "epoch": 19.73846153846154, "grad_norm": 1.4453125, "learning_rate": 4.4144559944958134e-05, "loss": 0.126, "step": 26943 }, { "epoch": 19.73919413919414, "grad_norm": 1.5078125, "learning_rate": 4.414116195744706e-05, "loss": 0.1231, "step": 26944 }, { "epoch": 19.73992673992674, "grad_norm": 1.1171875, "learning_rate": 4.413776400040329e-05, "loss": 0.1361, "step": 26945 }, { "epoch": 19.74065934065934, "grad_norm": 0.91015625, "learning_rate": 4.413436607384307e-05, "loss": 0.1409, "step": 26946 }, { "epoch": 19.741391941391942, "grad_norm": 0.94921875, "learning_rate": 4.41309681777827e-05, "loss": 0.1242, "step": 26947 }, { "epoch": 19.742124542124543, "grad_norm": 0.69140625, "learning_rate": 4.4127570312238454e-05, "loss": 0.1169, "step": 26948 }, { "epoch": 19.742857142857144, "grad_norm": 1.46875, "learning_rate": 4.4124172477226586e-05, "loss": 0.1062, "step": 26949 }, { "epoch": 19.743589743589745, "grad_norm": 1.1796875, "learning_rate": 4.4120774672763415e-05, "loss": 0.1124, "step": 26950 }, { "epoch": 19.744322344322345, "grad_norm": 1.296875, "learning_rate": 4.411737689886522e-05, "loss": 0.1247, "step": 26951 }, { "epoch": 19.745054945054946, "grad_norm": 1.3125, "learning_rate": 4.4113979155548243e-05, "loss": 0.111, "step": 26952 }, { "epoch": 19.745787545787547, "grad_norm": 1.453125, "learning_rate": 4.411058144282881e-05, "loss": 0.1548, "step": 26953 }, { "epoch": 19.746520146520147, "grad_norm": 0.8671875, "learning_rate": 4.410718376072318e-05, "loss": 0.1164, "step": 26954 }, { "epoch": 19.747252747252748, "grad_norm": 1.3359375, "learning_rate": 4.4103786109247604e-05, "loss": 0.1179, "step": 26955 }, { "epoch": 19.74798534798535, "grad_norm": 1.4140625, "learning_rate": 4.4100388488418404e-05, "loss": 0.1065, "step": 26956 }, { "epoch": 19.74871794871795, "grad_norm": 1.234375, "learning_rate": 4.409699089825185e-05, "loss": 0.1076, "step": 26957 }, { "epoch": 19.74945054945055, "grad_norm": 1.4921875, "learning_rate": 4.4093593338764215e-05, "loss": 0.1158, "step": 26958 }, { "epoch": 19.75018315018315, "grad_norm": 1.2734375, "learning_rate": 4.4090195809971786e-05, "loss": 0.1023, "step": 26959 }, { "epoch": 19.750915750915752, "grad_norm": 1.3203125, "learning_rate": 4.40867983118908e-05, "loss": 0.1292, "step": 26960 }, { "epoch": 19.751648351648353, "grad_norm": 1.109375, "learning_rate": 4.408340084453759e-05, "loss": 0.1265, "step": 26961 }, { "epoch": 19.752380952380953, "grad_norm": 0.94921875, "learning_rate": 4.408000340792842e-05, "loss": 0.1223, "step": 26962 }, { "epoch": 19.753113553113554, "grad_norm": 1.71875, "learning_rate": 4.4076606002079555e-05, "loss": 0.1483, "step": 26963 }, { "epoch": 19.753846153846155, "grad_norm": 1.3515625, "learning_rate": 4.407320862700727e-05, "loss": 0.1467, "step": 26964 }, { "epoch": 19.754578754578755, "grad_norm": 1.2578125, "learning_rate": 4.4069811282727854e-05, "loss": 0.1076, "step": 26965 }, { "epoch": 19.755311355311356, "grad_norm": 1.4453125, "learning_rate": 4.406641396925758e-05, "loss": 0.1166, "step": 26966 }, { "epoch": 19.756043956043957, "grad_norm": 1.3203125, "learning_rate": 4.406301668661274e-05, "loss": 0.1253, "step": 26967 }, { "epoch": 19.756776556776558, "grad_norm": 0.8359375, "learning_rate": 4.4059619434809604e-05, "loss": 0.1036, "step": 26968 }, { "epoch": 19.75750915750916, "grad_norm": 0.62109375, "learning_rate": 4.405622221386444e-05, "loss": 0.1202, "step": 26969 }, { "epoch": 19.75824175824176, "grad_norm": 0.62890625, "learning_rate": 4.4052825023793516e-05, "loss": 0.1113, "step": 26970 }, { "epoch": 19.75897435897436, "grad_norm": 1.296875, "learning_rate": 4.404942786461314e-05, "loss": 0.1215, "step": 26971 }, { "epoch": 19.75970695970696, "grad_norm": 1.375, "learning_rate": 4.404603073633957e-05, "loss": 0.1185, "step": 26972 }, { "epoch": 19.76043956043956, "grad_norm": 1.7578125, "learning_rate": 4.404263363898909e-05, "loss": 0.1442, "step": 26973 }, { "epoch": 19.761172161172162, "grad_norm": 1.125, "learning_rate": 4.403923657257798e-05, "loss": 0.0974, "step": 26974 }, { "epoch": 19.761904761904763, "grad_norm": 1.3046875, "learning_rate": 4.403583953712247e-05, "loss": 0.1509, "step": 26975 }, { "epoch": 19.762637362637363, "grad_norm": 1.125, "learning_rate": 4.403244253263891e-05, "loss": 0.127, "step": 26976 }, { "epoch": 19.763369963369964, "grad_norm": 1.4765625, "learning_rate": 4.4029045559143536e-05, "loss": 0.1504, "step": 26977 }, { "epoch": 19.764102564102565, "grad_norm": 0.9453125, "learning_rate": 4.402564861665262e-05, "loss": 0.1227, "step": 26978 }, { "epoch": 19.764835164835166, "grad_norm": 1.1640625, "learning_rate": 4.402225170518245e-05, "loss": 0.1161, "step": 26979 }, { "epoch": 19.765567765567766, "grad_norm": 0.76953125, "learning_rate": 4.4018854824749307e-05, "loss": 0.0907, "step": 26980 }, { "epoch": 19.766300366300367, "grad_norm": 1.2421875, "learning_rate": 4.401545797536943e-05, "loss": 0.1369, "step": 26981 }, { "epoch": 19.767032967032968, "grad_norm": 0.94140625, "learning_rate": 4.401206115705915e-05, "loss": 0.0933, "step": 26982 }, { "epoch": 19.76776556776557, "grad_norm": 1.15625, "learning_rate": 4.400866436983472e-05, "loss": 0.1137, "step": 26983 }, { "epoch": 19.76849816849817, "grad_norm": 1.390625, "learning_rate": 4.40052676137124e-05, "loss": 0.1152, "step": 26984 }, { "epoch": 19.76923076923077, "grad_norm": 0.94921875, "learning_rate": 4.400187088870848e-05, "loss": 0.1373, "step": 26985 }, { "epoch": 19.76996336996337, "grad_norm": 1.0859375, "learning_rate": 4.3998474194839206e-05, "loss": 0.1386, "step": 26986 }, { "epoch": 19.77069597069597, "grad_norm": 1.296875, "learning_rate": 4.3995077532120896e-05, "loss": 0.1678, "step": 26987 }, { "epoch": 19.771428571428572, "grad_norm": 0.8046875, "learning_rate": 4.3991680900569807e-05, "loss": 0.0957, "step": 26988 }, { "epoch": 19.772161172161173, "grad_norm": 0.8125, "learning_rate": 4.3988284300202215e-05, "loss": 0.1026, "step": 26989 }, { "epoch": 19.772893772893774, "grad_norm": 1.0, "learning_rate": 4.398488773103437e-05, "loss": 0.1053, "step": 26990 }, { "epoch": 19.773626373626374, "grad_norm": 1.234375, "learning_rate": 4.398149119308258e-05, "loss": 0.14, "step": 26991 }, { "epoch": 19.774358974358975, "grad_norm": 1.1875, "learning_rate": 4.39780946863631e-05, "loss": 0.1493, "step": 26992 }, { "epoch": 19.775091575091576, "grad_norm": 1.78125, "learning_rate": 4.3974698210892216e-05, "loss": 0.1225, "step": 26993 }, { "epoch": 19.775824175824177, "grad_norm": 1.2421875, "learning_rate": 4.39713017666862e-05, "loss": 0.1252, "step": 26994 }, { "epoch": 19.776556776556777, "grad_norm": 2.03125, "learning_rate": 4.3967905353761315e-05, "loss": 0.1289, "step": 26995 }, { "epoch": 19.777289377289378, "grad_norm": 0.96875, "learning_rate": 4.396450897213382e-05, "loss": 0.1316, "step": 26996 }, { "epoch": 19.77802197802198, "grad_norm": 1.125, "learning_rate": 4.3961112621820034e-05, "loss": 0.1246, "step": 26997 }, { "epoch": 19.77875457875458, "grad_norm": 0.94140625, "learning_rate": 4.3957716302836205e-05, "loss": 0.1174, "step": 26998 }, { "epoch": 19.77948717948718, "grad_norm": 1.9296875, "learning_rate": 4.39543200151986e-05, "loss": 0.1503, "step": 26999 }, { "epoch": 19.78021978021978, "grad_norm": 0.71875, "learning_rate": 4.39509237589235e-05, "loss": 0.1158, "step": 27000 }, { "epoch": 19.78095238095238, "grad_norm": 0.8359375, "learning_rate": 4.394752753402715e-05, "loss": 0.1331, "step": 27001 }, { "epoch": 19.781684981684982, "grad_norm": 0.703125, "learning_rate": 4.394413134052587e-05, "loss": 0.1249, "step": 27002 }, { "epoch": 19.782417582417583, "grad_norm": 1.21875, "learning_rate": 4.39407351784359e-05, "loss": 0.127, "step": 27003 }, { "epoch": 19.783150183150184, "grad_norm": 0.72265625, "learning_rate": 4.3937339047773516e-05, "loss": 0.121, "step": 27004 }, { "epoch": 19.783882783882785, "grad_norm": 1.015625, "learning_rate": 4.393394294855501e-05, "loss": 0.1573, "step": 27005 }, { "epoch": 19.784615384615385, "grad_norm": 1.421875, "learning_rate": 4.393054688079664e-05, "loss": 0.1208, "step": 27006 }, { "epoch": 19.785347985347986, "grad_norm": 1.0546875, "learning_rate": 4.392715084451465e-05, "loss": 0.1103, "step": 27007 }, { "epoch": 19.786080586080587, "grad_norm": 1.3984375, "learning_rate": 4.392375483972536e-05, "loss": 0.1208, "step": 27008 }, { "epoch": 19.786813186813188, "grad_norm": 0.9296875, "learning_rate": 4.392035886644502e-05, "loss": 0.1186, "step": 27009 }, { "epoch": 19.78754578754579, "grad_norm": 0.78515625, "learning_rate": 4.39169629246899e-05, "loss": 0.1328, "step": 27010 }, { "epoch": 19.78827838827839, "grad_norm": 1.015625, "learning_rate": 4.391356701447627e-05, "loss": 0.1333, "step": 27011 }, { "epoch": 19.78901098901099, "grad_norm": 1.40625, "learning_rate": 4.3910171135820384e-05, "loss": 0.1324, "step": 27012 }, { "epoch": 19.78974358974359, "grad_norm": 1.1328125, "learning_rate": 4.390677528873855e-05, "loss": 0.1398, "step": 27013 }, { "epoch": 19.79047619047619, "grad_norm": 0.86328125, "learning_rate": 4.3903379473247024e-05, "loss": 0.115, "step": 27014 }, { "epoch": 19.791208791208792, "grad_norm": 1.28125, "learning_rate": 4.3899983689362064e-05, "loss": 0.0999, "step": 27015 }, { "epoch": 19.791941391941393, "grad_norm": 0.9609375, "learning_rate": 4.389658793709995e-05, "loss": 0.1491, "step": 27016 }, { "epoch": 19.792673992673993, "grad_norm": 0.9453125, "learning_rate": 4.389319221647695e-05, "loss": 0.1356, "step": 27017 }, { "epoch": 19.793406593406594, "grad_norm": 0.5859375, "learning_rate": 4.388979652750933e-05, "loss": 0.1102, "step": 27018 }, { "epoch": 19.794139194139195, "grad_norm": 1.0390625, "learning_rate": 4.388640087021337e-05, "loss": 0.1281, "step": 27019 }, { "epoch": 19.794871794871796, "grad_norm": 1.140625, "learning_rate": 4.388300524460535e-05, "loss": 0.0929, "step": 27020 }, { "epoch": 19.795604395604396, "grad_norm": 0.8203125, "learning_rate": 4.387960965070151e-05, "loss": 0.1238, "step": 27021 }, { "epoch": 19.796336996336997, "grad_norm": 0.87890625, "learning_rate": 4.387621408851812e-05, "loss": 0.1335, "step": 27022 }, { "epoch": 19.797069597069598, "grad_norm": 0.859375, "learning_rate": 4.387281855807148e-05, "loss": 0.1355, "step": 27023 }, { "epoch": 19.7978021978022, "grad_norm": 1.0546875, "learning_rate": 4.386942305937784e-05, "loss": 0.1006, "step": 27024 }, { "epoch": 19.7985347985348, "grad_norm": 1.3046875, "learning_rate": 4.386602759245347e-05, "loss": 0.1341, "step": 27025 }, { "epoch": 19.7992673992674, "grad_norm": 1.0625, "learning_rate": 4.386263215731465e-05, "loss": 0.1213, "step": 27026 }, { "epoch": 19.8, "grad_norm": 1.203125, "learning_rate": 4.38592367539776e-05, "loss": 0.1338, "step": 27027 }, { "epoch": 19.8007326007326, "grad_norm": 0.89453125, "learning_rate": 4.3855841382458666e-05, "loss": 0.1107, "step": 27028 }, { "epoch": 19.801465201465202, "grad_norm": 1.1328125, "learning_rate": 4.385244604277407e-05, "loss": 0.1314, "step": 27029 }, { "epoch": 19.802197802197803, "grad_norm": 0.83203125, "learning_rate": 4.384905073494008e-05, "loss": 0.1165, "step": 27030 }, { "epoch": 19.802930402930404, "grad_norm": 0.68359375, "learning_rate": 4.384565545897297e-05, "loss": 0.1199, "step": 27031 }, { "epoch": 19.803663003663004, "grad_norm": 1.3671875, "learning_rate": 4.384226021488902e-05, "loss": 0.1475, "step": 27032 }, { "epoch": 19.804395604395605, "grad_norm": 0.82421875, "learning_rate": 4.383886500270446e-05, "loss": 0.1093, "step": 27033 }, { "epoch": 19.805128205128206, "grad_norm": 1.296875, "learning_rate": 4.383546982243562e-05, "loss": 0.1199, "step": 27034 }, { "epoch": 19.805860805860807, "grad_norm": 1.453125, "learning_rate": 4.3832074674098715e-05, "loss": 0.1088, "step": 27035 }, { "epoch": 19.806593406593407, "grad_norm": 0.96875, "learning_rate": 4.382867955771003e-05, "loss": 0.1241, "step": 27036 }, { "epoch": 19.807326007326008, "grad_norm": 1.1875, "learning_rate": 4.3825284473285844e-05, "loss": 0.1155, "step": 27037 }, { "epoch": 19.80805860805861, "grad_norm": 1.3203125, "learning_rate": 4.3821889420842375e-05, "loss": 0.0965, "step": 27038 }, { "epoch": 19.80879120879121, "grad_norm": 0.83984375, "learning_rate": 4.381849440039596e-05, "loss": 0.1121, "step": 27039 }, { "epoch": 19.80952380952381, "grad_norm": 1.5, "learning_rate": 4.3815099411962826e-05, "loss": 0.1418, "step": 27040 }, { "epoch": 19.81025641025641, "grad_norm": 0.98828125, "learning_rate": 4.381170445555923e-05, "loss": 0.1354, "step": 27041 }, { "epoch": 19.81098901098901, "grad_norm": 1.96875, "learning_rate": 4.380830953120147e-05, "loss": 0.1219, "step": 27042 }, { "epoch": 19.811721611721612, "grad_norm": 0.75, "learning_rate": 4.380491463890579e-05, "loss": 0.1501, "step": 27043 }, { "epoch": 19.812454212454213, "grad_norm": 1.53125, "learning_rate": 4.380151977868845e-05, "loss": 0.1035, "step": 27044 }, { "epoch": 19.813186813186814, "grad_norm": 1.234375, "learning_rate": 4.3798124950565746e-05, "loss": 0.1168, "step": 27045 }, { "epoch": 19.813919413919415, "grad_norm": 1.0546875, "learning_rate": 4.379473015455392e-05, "loss": 0.1311, "step": 27046 }, { "epoch": 19.814652014652015, "grad_norm": 1.21875, "learning_rate": 4.379133539066925e-05, "loss": 0.1226, "step": 27047 }, { "epoch": 19.815384615384616, "grad_norm": 0.97265625, "learning_rate": 4.3787940658927974e-05, "loss": 0.1291, "step": 27048 }, { "epoch": 19.816117216117217, "grad_norm": 0.9375, "learning_rate": 4.378454595934639e-05, "loss": 0.0985, "step": 27049 }, { "epoch": 19.816849816849818, "grad_norm": 1.3046875, "learning_rate": 4.378115129194076e-05, "loss": 0.1251, "step": 27050 }, { "epoch": 19.81758241758242, "grad_norm": 0.890625, "learning_rate": 4.377775665672733e-05, "loss": 0.0977, "step": 27051 }, { "epoch": 19.81831501831502, "grad_norm": 0.921875, "learning_rate": 4.377436205372238e-05, "loss": 0.1244, "step": 27052 }, { "epoch": 19.81904761904762, "grad_norm": 1.4921875, "learning_rate": 4.3770967482942165e-05, "loss": 0.0888, "step": 27053 }, { "epoch": 19.81978021978022, "grad_norm": 1.328125, "learning_rate": 4.376757294440294e-05, "loss": 0.1228, "step": 27054 }, { "epoch": 19.82051282051282, "grad_norm": 1.140625, "learning_rate": 4.3764178438121e-05, "loss": 0.1094, "step": 27055 }, { "epoch": 19.821245421245422, "grad_norm": 1.140625, "learning_rate": 4.376078396411258e-05, "loss": 0.127, "step": 27056 }, { "epoch": 19.821978021978023, "grad_norm": 0.97265625, "learning_rate": 4.3757389522393976e-05, "loss": 0.1214, "step": 27057 }, { "epoch": 19.822710622710623, "grad_norm": 0.84375, "learning_rate": 4.375399511298143e-05, "loss": 0.1265, "step": 27058 }, { "epoch": 19.823443223443224, "grad_norm": 0.77734375, "learning_rate": 4.375060073589118e-05, "loss": 0.1189, "step": 27059 }, { "epoch": 19.824175824175825, "grad_norm": 0.9765625, "learning_rate": 4.3747206391139544e-05, "loss": 0.1154, "step": 27060 }, { "epoch": 19.824908424908426, "grad_norm": 0.8671875, "learning_rate": 4.374381207874276e-05, "loss": 0.1142, "step": 27061 }, { "epoch": 19.825641025641026, "grad_norm": 1.046875, "learning_rate": 4.374041779871708e-05, "loss": 0.1162, "step": 27062 }, { "epoch": 19.826373626373627, "grad_norm": 1.078125, "learning_rate": 4.373702355107878e-05, "loss": 0.1362, "step": 27063 }, { "epoch": 19.827106227106228, "grad_norm": 1.171875, "learning_rate": 4.373362933584411e-05, "loss": 0.1569, "step": 27064 }, { "epoch": 19.82783882783883, "grad_norm": 1.71875, "learning_rate": 4.3730235153029356e-05, "loss": 0.1301, "step": 27065 }, { "epoch": 19.82857142857143, "grad_norm": 0.93359375, "learning_rate": 4.372684100265077e-05, "loss": 0.1374, "step": 27066 }, { "epoch": 19.82930402930403, "grad_norm": 0.91015625, "learning_rate": 4.37234468847246e-05, "loss": 0.0975, "step": 27067 }, { "epoch": 19.83003663003663, "grad_norm": 0.79296875, "learning_rate": 4.372005279926714e-05, "loss": 0.1131, "step": 27068 }, { "epoch": 19.83076923076923, "grad_norm": 0.94921875, "learning_rate": 4.371665874629461e-05, "loss": 0.1145, "step": 27069 }, { "epoch": 19.831501831501832, "grad_norm": 0.9375, "learning_rate": 4.37132647258233e-05, "loss": 0.1257, "step": 27070 }, { "epoch": 19.832234432234433, "grad_norm": 1.3125, "learning_rate": 4.370987073786947e-05, "loss": 0.142, "step": 27071 }, { "epoch": 19.832967032967034, "grad_norm": 1.3203125, "learning_rate": 4.370647678244939e-05, "loss": 0.1326, "step": 27072 }, { "epoch": 19.833699633699634, "grad_norm": 0.71875, "learning_rate": 4.3703082859579303e-05, "loss": 0.1139, "step": 27073 }, { "epoch": 19.834432234432235, "grad_norm": 0.82421875, "learning_rate": 4.369968896927546e-05, "loss": 0.1257, "step": 27074 }, { "epoch": 19.835164835164836, "grad_norm": 1.3359375, "learning_rate": 4.369629511155416e-05, "loss": 0.1171, "step": 27075 }, { "epoch": 19.835897435897436, "grad_norm": 1.359375, "learning_rate": 4.369290128643164e-05, "loss": 0.1267, "step": 27076 }, { "epoch": 19.836630036630037, "grad_norm": 1.2890625, "learning_rate": 4.368950749392417e-05, "loss": 0.1061, "step": 27077 }, { "epoch": 19.837362637362638, "grad_norm": 0.9296875, "learning_rate": 4.368611373404799e-05, "loss": 0.1215, "step": 27078 }, { "epoch": 19.83809523809524, "grad_norm": 0.6484375, "learning_rate": 4.3682720006819383e-05, "loss": 0.1285, "step": 27079 }, { "epoch": 19.83882783882784, "grad_norm": 1.8359375, "learning_rate": 4.36793263122546e-05, "loss": 0.1189, "step": 27080 }, { "epoch": 19.83956043956044, "grad_norm": 1.140625, "learning_rate": 4.36759326503699e-05, "loss": 0.1242, "step": 27081 }, { "epoch": 19.84029304029304, "grad_norm": 1.0625, "learning_rate": 4.3672539021181555e-05, "loss": 0.1511, "step": 27082 }, { "epoch": 19.84102564102564, "grad_norm": 1.0078125, "learning_rate": 4.3669145424705825e-05, "loss": 0.1237, "step": 27083 }, { "epoch": 19.841758241758242, "grad_norm": 0.765625, "learning_rate": 4.366575186095895e-05, "loss": 0.1288, "step": 27084 }, { "epoch": 19.842490842490843, "grad_norm": 0.7578125, "learning_rate": 4.3662358329957194e-05, "loss": 0.1206, "step": 27085 }, { "epoch": 19.843223443223444, "grad_norm": 1.171875, "learning_rate": 4.3658964831716835e-05, "loss": 0.1074, "step": 27086 }, { "epoch": 19.843956043956045, "grad_norm": 1.125, "learning_rate": 4.365557136625412e-05, "loss": 0.1174, "step": 27087 }, { "epoch": 19.844688644688645, "grad_norm": 1.140625, "learning_rate": 4.365217793358532e-05, "loss": 0.1481, "step": 27088 }, { "epoch": 19.845421245421246, "grad_norm": 0.89453125, "learning_rate": 4.364878453372667e-05, "loss": 0.1125, "step": 27089 }, { "epoch": 19.846153846153847, "grad_norm": 0.87109375, "learning_rate": 4.364539116669443e-05, "loss": 0.1053, "step": 27090 }, { "epoch": 19.846886446886447, "grad_norm": 1.03125, "learning_rate": 4.364199783250489e-05, "loss": 0.1328, "step": 27091 }, { "epoch": 19.847619047619048, "grad_norm": 0.84765625, "learning_rate": 4.363860453117429e-05, "loss": 0.1067, "step": 27092 }, { "epoch": 19.84835164835165, "grad_norm": 1.328125, "learning_rate": 4.3635211262718885e-05, "loss": 0.1398, "step": 27093 }, { "epoch": 19.84908424908425, "grad_norm": 1.25, "learning_rate": 4.3631818027154935e-05, "loss": 0.1102, "step": 27094 }, { "epoch": 19.84981684981685, "grad_norm": 1.2734375, "learning_rate": 4.3628424824498696e-05, "loss": 0.1494, "step": 27095 }, { "epoch": 19.85054945054945, "grad_norm": 1.015625, "learning_rate": 4.362503165476643e-05, "loss": 0.1501, "step": 27096 }, { "epoch": 19.851282051282052, "grad_norm": 0.84375, "learning_rate": 4.3621638517974396e-05, "loss": 0.1244, "step": 27097 }, { "epoch": 19.852014652014653, "grad_norm": 0.91796875, "learning_rate": 4.3618245414138855e-05, "loss": 0.1068, "step": 27098 }, { "epoch": 19.852747252747253, "grad_norm": 0.8046875, "learning_rate": 4.3614852343276064e-05, "loss": 0.1167, "step": 27099 }, { "epoch": 19.853479853479854, "grad_norm": 1.6484375, "learning_rate": 4.3611459305402244e-05, "loss": 0.1137, "step": 27100 }, { "epoch": 19.854212454212455, "grad_norm": 1.4609375, "learning_rate": 4.3608066300533714e-05, "loss": 0.1304, "step": 27101 }, { "epoch": 19.854945054945055, "grad_norm": 0.890625, "learning_rate": 4.36046733286867e-05, "loss": 0.1312, "step": 27102 }, { "epoch": 19.855677655677656, "grad_norm": 1.2109375, "learning_rate": 4.360128038987746e-05, "loss": 0.1397, "step": 27103 }, { "epoch": 19.856410256410257, "grad_norm": 1.1875, "learning_rate": 4.359788748412225e-05, "loss": 0.1302, "step": 27104 }, { "epoch": 19.857142857142858, "grad_norm": 1.0859375, "learning_rate": 4.3594494611437314e-05, "loss": 0.1186, "step": 27105 }, { "epoch": 19.85787545787546, "grad_norm": 0.60546875, "learning_rate": 4.359110177183892e-05, "loss": 0.108, "step": 27106 }, { "epoch": 19.85860805860806, "grad_norm": 0.56640625, "learning_rate": 4.358770896534333e-05, "loss": 0.1101, "step": 27107 }, { "epoch": 19.85934065934066, "grad_norm": 1.1640625, "learning_rate": 4.35843161919668e-05, "loss": 0.139, "step": 27108 }, { "epoch": 19.86007326007326, "grad_norm": 1.203125, "learning_rate": 4.358092345172559e-05, "loss": 0.1323, "step": 27109 }, { "epoch": 19.86080586080586, "grad_norm": 0.91796875, "learning_rate": 4.3577530744635944e-05, "loss": 0.1126, "step": 27110 }, { "epoch": 19.861538461538462, "grad_norm": 1.328125, "learning_rate": 4.3574138070714096e-05, "loss": 0.1022, "step": 27111 }, { "epoch": 19.862271062271063, "grad_norm": 1.203125, "learning_rate": 4.357074542997634e-05, "loss": 0.1289, "step": 27112 }, { "epoch": 19.863003663003663, "grad_norm": 0.89453125, "learning_rate": 4.356735282243893e-05, "loss": 0.0937, "step": 27113 }, { "epoch": 19.863736263736264, "grad_norm": 1.078125, "learning_rate": 4.3563960248118104e-05, "loss": 0.1358, "step": 27114 }, { "epoch": 19.864468864468865, "grad_norm": 1.2578125, "learning_rate": 4.356056770703013e-05, "loss": 0.1047, "step": 27115 }, { "epoch": 19.865201465201466, "grad_norm": 1.03125, "learning_rate": 4.3557175199191224e-05, "loss": 0.115, "step": 27116 }, { "epoch": 19.865934065934066, "grad_norm": 0.875, "learning_rate": 4.355378272461769e-05, "loss": 0.1309, "step": 27117 }, { "epoch": 19.866666666666667, "grad_norm": 1.5859375, "learning_rate": 4.355039028332577e-05, "loss": 0.1088, "step": 27118 }, { "epoch": 19.867399267399268, "grad_norm": 1.828125, "learning_rate": 4.3546997875331695e-05, "loss": 0.1375, "step": 27119 }, { "epoch": 19.86813186813187, "grad_norm": 1.4140625, "learning_rate": 4.3543605500651754e-05, "loss": 0.1277, "step": 27120 }, { "epoch": 19.86886446886447, "grad_norm": 0.7421875, "learning_rate": 4.3540213159302146e-05, "loss": 0.1338, "step": 27121 }, { "epoch": 19.86959706959707, "grad_norm": 0.94921875, "learning_rate": 4.35368208512992e-05, "loss": 0.1127, "step": 27122 }, { "epoch": 19.87032967032967, "grad_norm": 1.15625, "learning_rate": 4.353342857665912e-05, "loss": 0.1591, "step": 27123 }, { "epoch": 19.87106227106227, "grad_norm": 0.87890625, "learning_rate": 4.3530036335398176e-05, "loss": 0.12, "step": 27124 }, { "epoch": 19.871794871794872, "grad_norm": 0.89453125, "learning_rate": 4.352664412753262e-05, "loss": 0.0955, "step": 27125 }, { "epoch": 19.872527472527473, "grad_norm": 1.359375, "learning_rate": 4.352325195307867e-05, "loss": 0.1622, "step": 27126 }, { "epoch": 19.873260073260074, "grad_norm": 0.8984375, "learning_rate": 4.351985981205264e-05, "loss": 0.1316, "step": 27127 }, { "epoch": 19.873992673992674, "grad_norm": 1.4765625, "learning_rate": 4.351646770447075e-05, "loss": 0.1314, "step": 27128 }, { "epoch": 19.874725274725275, "grad_norm": 1.328125, "learning_rate": 4.3513075630349255e-05, "loss": 0.1326, "step": 27129 }, { "epoch": 19.875457875457876, "grad_norm": 1.140625, "learning_rate": 4.3509683589704393e-05, "loss": 0.1212, "step": 27130 }, { "epoch": 19.876190476190477, "grad_norm": 0.6875, "learning_rate": 4.350629158255245e-05, "loss": 0.1162, "step": 27131 }, { "epoch": 19.876923076923077, "grad_norm": 1.0, "learning_rate": 4.350289960890964e-05, "loss": 0.1385, "step": 27132 }, { "epoch": 19.877655677655678, "grad_norm": 0.8828125, "learning_rate": 4.349950766879224e-05, "loss": 0.1216, "step": 27133 }, { "epoch": 19.87838827838828, "grad_norm": 1.40625, "learning_rate": 4.3496115762216516e-05, "loss": 0.1144, "step": 27134 }, { "epoch": 19.87912087912088, "grad_norm": 1.2890625, "learning_rate": 4.349272388919869e-05, "loss": 0.1415, "step": 27135 }, { "epoch": 19.87985347985348, "grad_norm": 1.15625, "learning_rate": 4.3489332049755025e-05, "loss": 0.1282, "step": 27136 }, { "epoch": 19.88058608058608, "grad_norm": 1.2421875, "learning_rate": 4.3485940243901754e-05, "loss": 0.114, "step": 27137 }, { "epoch": 19.88131868131868, "grad_norm": 1.34375, "learning_rate": 4.348254847165517e-05, "loss": 0.1454, "step": 27138 }, { "epoch": 19.882051282051282, "grad_norm": 0.78515625, "learning_rate": 4.34791567330315e-05, "loss": 0.1144, "step": 27139 }, { "epoch": 19.882783882783883, "grad_norm": 1.3671875, "learning_rate": 4.347576502804699e-05, "loss": 0.1276, "step": 27140 }, { "epoch": 19.883516483516484, "grad_norm": 0.86328125, "learning_rate": 4.34723733567179e-05, "loss": 0.1368, "step": 27141 }, { "epoch": 19.884249084249085, "grad_norm": 0.88671875, "learning_rate": 4.346898171906044e-05, "loss": 0.1271, "step": 27142 }, { "epoch": 19.884981684981685, "grad_norm": 1.0078125, "learning_rate": 4.346559011509093e-05, "loss": 0.0961, "step": 27143 }, { "epoch": 19.885714285714286, "grad_norm": 1.3203125, "learning_rate": 4.346219854482559e-05, "loss": 0.1509, "step": 27144 }, { "epoch": 19.886446886446887, "grad_norm": 1.1796875, "learning_rate": 4.345880700828065e-05, "loss": 0.1094, "step": 27145 }, { "epoch": 19.887179487179488, "grad_norm": 0.97265625, "learning_rate": 4.345541550547239e-05, "loss": 0.1138, "step": 27146 }, { "epoch": 19.88791208791209, "grad_norm": 1.671875, "learning_rate": 4.345202403641703e-05, "loss": 0.1412, "step": 27147 }, { "epoch": 19.88864468864469, "grad_norm": 1.375, "learning_rate": 4.344863260113086e-05, "loss": 0.1594, "step": 27148 }, { "epoch": 19.88937728937729, "grad_norm": 1.1953125, "learning_rate": 4.34452411996301e-05, "loss": 0.1296, "step": 27149 }, { "epoch": 19.89010989010989, "grad_norm": 1.0078125, "learning_rate": 4.3441849831931e-05, "loss": 0.1302, "step": 27150 }, { "epoch": 19.89084249084249, "grad_norm": 1.125, "learning_rate": 4.343845849804982e-05, "loss": 0.1596, "step": 27151 }, { "epoch": 19.891575091575092, "grad_norm": 0.921875, "learning_rate": 4.343506719800279e-05, "loss": 0.0921, "step": 27152 }, { "epoch": 19.892307692307693, "grad_norm": 1.21875, "learning_rate": 4.343167593180618e-05, "loss": 0.1392, "step": 27153 }, { "epoch": 19.893040293040293, "grad_norm": 0.80859375, "learning_rate": 4.342828469947624e-05, "loss": 0.1095, "step": 27154 }, { "epoch": 19.893772893772894, "grad_norm": 1.2109375, "learning_rate": 4.342489350102921e-05, "loss": 0.1353, "step": 27155 }, { "epoch": 19.894505494505495, "grad_norm": 1.2421875, "learning_rate": 4.342150233648132e-05, "loss": 0.138, "step": 27156 }, { "epoch": 19.895238095238096, "grad_norm": 1.1640625, "learning_rate": 4.341811120584885e-05, "loss": 0.1155, "step": 27157 }, { "epoch": 19.895970695970696, "grad_norm": 0.984375, "learning_rate": 4.341472010914802e-05, "loss": 0.1208, "step": 27158 }, { "epoch": 19.896703296703297, "grad_norm": 1.2421875, "learning_rate": 4.34113290463951e-05, "loss": 0.1206, "step": 27159 }, { "epoch": 19.897435897435898, "grad_norm": 1.3984375, "learning_rate": 4.340793801760633e-05, "loss": 0.1321, "step": 27160 }, { "epoch": 19.8981684981685, "grad_norm": 1.0625, "learning_rate": 4.340454702279796e-05, "loss": 0.1204, "step": 27161 }, { "epoch": 19.8989010989011, "grad_norm": 0.84765625, "learning_rate": 4.3401156061986244e-05, "loss": 0.1316, "step": 27162 }, { "epoch": 19.8996336996337, "grad_norm": 1.3828125, "learning_rate": 4.339776513518739e-05, "loss": 0.1236, "step": 27163 }, { "epoch": 19.9003663003663, "grad_norm": 0.7265625, "learning_rate": 4.33943742424177e-05, "loss": 0.1251, "step": 27164 }, { "epoch": 19.9010989010989, "grad_norm": 1.15625, "learning_rate": 4.3390983383693394e-05, "loss": 0.1339, "step": 27165 }, { "epoch": 19.901831501831502, "grad_norm": 0.76171875, "learning_rate": 4.3387592559030716e-05, "loss": 0.1378, "step": 27166 }, { "epoch": 19.902564102564103, "grad_norm": 0.796875, "learning_rate": 4.338420176844592e-05, "loss": 0.1249, "step": 27167 }, { "epoch": 19.903296703296704, "grad_norm": 1.296875, "learning_rate": 4.338081101195523e-05, "loss": 0.1551, "step": 27168 }, { "epoch": 19.904029304029304, "grad_norm": 1.2265625, "learning_rate": 4.3377420289574934e-05, "loss": 0.1233, "step": 27169 }, { "epoch": 19.904761904761905, "grad_norm": 0.9296875, "learning_rate": 4.3374029601321254e-05, "loss": 0.1098, "step": 27170 }, { "epoch": 19.905494505494506, "grad_norm": 1.1171875, "learning_rate": 4.337063894721042e-05, "loss": 0.1277, "step": 27171 }, { "epoch": 19.906227106227107, "grad_norm": 1.09375, "learning_rate": 4.336724832725871e-05, "loss": 0.1322, "step": 27172 }, { "epoch": 19.906959706959707, "grad_norm": 1.40625, "learning_rate": 4.336385774148234e-05, "loss": 0.1537, "step": 27173 }, { "epoch": 19.907692307692308, "grad_norm": 1.4453125, "learning_rate": 4.336046718989759e-05, "loss": 0.1297, "step": 27174 }, { "epoch": 19.90842490842491, "grad_norm": 1.1875, "learning_rate": 4.335707667252068e-05, "loss": 0.1244, "step": 27175 }, { "epoch": 19.90915750915751, "grad_norm": 1.359375, "learning_rate": 4.335368618936786e-05, "loss": 0.16, "step": 27176 }, { "epoch": 19.90989010989011, "grad_norm": 0.89453125, "learning_rate": 4.335029574045538e-05, "loss": 0.1244, "step": 27177 }, { "epoch": 19.91062271062271, "grad_norm": 0.80078125, "learning_rate": 4.3346905325799457e-05, "loss": 0.1212, "step": 27178 }, { "epoch": 19.91135531135531, "grad_norm": 1.125, "learning_rate": 4.334351494541638e-05, "loss": 0.1501, "step": 27179 }, { "epoch": 19.912087912087912, "grad_norm": 0.765625, "learning_rate": 4.334012459932237e-05, "loss": 0.1025, "step": 27180 }, { "epoch": 19.912820512820513, "grad_norm": 1.2421875, "learning_rate": 4.333673428753368e-05, "loss": 0.1398, "step": 27181 }, { "epoch": 19.913553113553114, "grad_norm": 1.796875, "learning_rate": 4.3333344010066534e-05, "loss": 0.111, "step": 27182 }, { "epoch": 19.914285714285715, "grad_norm": 0.8203125, "learning_rate": 4.3329953766937205e-05, "loss": 0.1269, "step": 27183 }, { "epoch": 19.915018315018315, "grad_norm": 1.1796875, "learning_rate": 4.33265635581619e-05, "loss": 0.1121, "step": 27184 }, { "epoch": 19.915750915750916, "grad_norm": 1.109375, "learning_rate": 4.332317338375689e-05, "loss": 0.1085, "step": 27185 }, { "epoch": 19.916483516483517, "grad_norm": 1.1171875, "learning_rate": 4.331978324373842e-05, "loss": 0.1707, "step": 27186 }, { "epoch": 19.917216117216118, "grad_norm": 0.75, "learning_rate": 4.331639313812273e-05, "loss": 0.1226, "step": 27187 }, { "epoch": 19.91794871794872, "grad_norm": 0.7578125, "learning_rate": 4.331300306692605e-05, "loss": 0.1097, "step": 27188 }, { "epoch": 19.91868131868132, "grad_norm": 0.96484375, "learning_rate": 4.330961303016462e-05, "loss": 0.1168, "step": 27189 }, { "epoch": 19.91941391941392, "grad_norm": 0.85546875, "learning_rate": 4.330622302785471e-05, "loss": 0.1321, "step": 27190 }, { "epoch": 19.92014652014652, "grad_norm": 1.1484375, "learning_rate": 4.330283306001255e-05, "loss": 0.1182, "step": 27191 }, { "epoch": 19.92087912087912, "grad_norm": 1.296875, "learning_rate": 4.3299443126654375e-05, "loss": 0.1457, "step": 27192 }, { "epoch": 19.921611721611722, "grad_norm": 1.2421875, "learning_rate": 4.329605322779643e-05, "loss": 0.1029, "step": 27193 }, { "epoch": 19.922344322344323, "grad_norm": 0.7109375, "learning_rate": 4.3292663363454945e-05, "loss": 0.1251, "step": 27194 }, { "epoch": 19.923076923076923, "grad_norm": 0.95703125, "learning_rate": 4.32892735336462e-05, "loss": 0.1439, "step": 27195 }, { "epoch": 19.923809523809524, "grad_norm": 1.109375, "learning_rate": 4.3285883738386395e-05, "loss": 0.1437, "step": 27196 }, { "epoch": 19.924542124542125, "grad_norm": 0.8828125, "learning_rate": 4.328249397769181e-05, "loss": 0.1118, "step": 27197 }, { "epoch": 19.925274725274726, "grad_norm": 0.76953125, "learning_rate": 4.327910425157865e-05, "loss": 0.1142, "step": 27198 }, { "epoch": 19.926007326007326, "grad_norm": 1.5390625, "learning_rate": 4.327571456006316e-05, "loss": 0.0956, "step": 27199 }, { "epoch": 19.926739926739927, "grad_norm": 1.8203125, "learning_rate": 4.327232490316161e-05, "loss": 0.1477, "step": 27200 }, { "epoch": 19.927472527472528, "grad_norm": 0.8984375, "learning_rate": 4.3268935280890224e-05, "loss": 0.1106, "step": 27201 }, { "epoch": 19.92820512820513, "grad_norm": 0.75, "learning_rate": 4.326554569326525e-05, "loss": 0.1434, "step": 27202 }, { "epoch": 19.92893772893773, "grad_norm": 1.0703125, "learning_rate": 4.326215614030291e-05, "loss": 0.1355, "step": 27203 }, { "epoch": 19.92967032967033, "grad_norm": 0.73828125, "learning_rate": 4.3258766622019445e-05, "loss": 0.1081, "step": 27204 }, { "epoch": 19.93040293040293, "grad_norm": 1.4296875, "learning_rate": 4.325537713843113e-05, "loss": 0.1304, "step": 27205 }, { "epoch": 19.93113553113553, "grad_norm": 0.68359375, "learning_rate": 4.325198768955418e-05, "loss": 0.0963, "step": 27206 }, { "epoch": 19.931868131868132, "grad_norm": 0.64453125, "learning_rate": 4.324859827540482e-05, "loss": 0.1126, "step": 27207 }, { "epoch": 19.932600732600733, "grad_norm": 0.890625, "learning_rate": 4.324520889599931e-05, "loss": 0.1274, "step": 27208 }, { "epoch": 19.933333333333334, "grad_norm": 1.1953125, "learning_rate": 4.324181955135389e-05, "loss": 0.1421, "step": 27209 }, { "epoch": 19.934065934065934, "grad_norm": 0.89453125, "learning_rate": 4.3238430241484795e-05, "loss": 0.1248, "step": 27210 }, { "epoch": 19.934798534798535, "grad_norm": 1.0078125, "learning_rate": 4.323504096640826e-05, "loss": 0.1254, "step": 27211 }, { "epoch": 19.935531135531136, "grad_norm": 1.390625, "learning_rate": 4.3231651726140537e-05, "loss": 0.1081, "step": 27212 }, { "epoch": 19.936263736263736, "grad_norm": 1.2578125, "learning_rate": 4.3228262520697855e-05, "loss": 0.1255, "step": 27213 }, { "epoch": 19.936996336996337, "grad_norm": 1.0859375, "learning_rate": 4.322487335009646e-05, "loss": 0.1246, "step": 27214 }, { "epoch": 19.937728937728938, "grad_norm": 1.0703125, "learning_rate": 4.322148421435255e-05, "loss": 0.1367, "step": 27215 }, { "epoch": 19.93846153846154, "grad_norm": 0.53515625, "learning_rate": 4.321809511348244e-05, "loss": 0.1039, "step": 27216 }, { "epoch": 19.93919413919414, "grad_norm": 0.921875, "learning_rate": 4.3214706047502314e-05, "loss": 0.0905, "step": 27217 }, { "epoch": 19.93992673992674, "grad_norm": 0.546875, "learning_rate": 4.321131701642842e-05, "loss": 0.0986, "step": 27218 }, { "epoch": 19.94065934065934, "grad_norm": 0.75390625, "learning_rate": 4.320792802027701e-05, "loss": 0.1247, "step": 27219 }, { "epoch": 19.94139194139194, "grad_norm": 0.86328125, "learning_rate": 4.320453905906429e-05, "loss": 0.1182, "step": 27220 }, { "epoch": 19.942124542124542, "grad_norm": 0.53515625, "learning_rate": 4.3201150132806525e-05, "loss": 0.1093, "step": 27221 }, { "epoch": 19.942857142857143, "grad_norm": 1.609375, "learning_rate": 4.319776124151995e-05, "loss": 0.1391, "step": 27222 }, { "epoch": 19.943589743589744, "grad_norm": 0.8671875, "learning_rate": 4.319437238522079e-05, "loss": 0.126, "step": 27223 }, { "epoch": 19.944322344322345, "grad_norm": 1.1796875, "learning_rate": 4.319098356392531e-05, "loss": 0.1073, "step": 27224 }, { "epoch": 19.945054945054945, "grad_norm": 0.8984375, "learning_rate": 4.3187594777649704e-05, "loss": 0.0986, "step": 27225 }, { "epoch": 19.945787545787546, "grad_norm": 1.0234375, "learning_rate": 4.318420602641024e-05, "loss": 0.1083, "step": 27226 }, { "epoch": 19.946520146520147, "grad_norm": 1.3046875, "learning_rate": 4.318081731022316e-05, "loss": 0.1258, "step": 27227 }, { "epoch": 19.947252747252747, "grad_norm": 1.6953125, "learning_rate": 4.317742862910468e-05, "loss": 0.126, "step": 27228 }, { "epoch": 19.947985347985348, "grad_norm": 1.3984375, "learning_rate": 4.3174039983071043e-05, "loss": 0.1499, "step": 27229 }, { "epoch": 19.94871794871795, "grad_norm": 1.1171875, "learning_rate": 4.3170651372138466e-05, "loss": 0.1085, "step": 27230 }, { "epoch": 19.94945054945055, "grad_norm": 0.609375, "learning_rate": 4.3167262796323225e-05, "loss": 0.1115, "step": 27231 }, { "epoch": 19.95018315018315, "grad_norm": 0.65234375, "learning_rate": 4.3163874255641534e-05, "loss": 0.1137, "step": 27232 }, { "epoch": 19.95091575091575, "grad_norm": 1.1953125, "learning_rate": 4.316048575010963e-05, "loss": 0.1086, "step": 27233 }, { "epoch": 19.951648351648352, "grad_norm": 1.296875, "learning_rate": 4.315709727974374e-05, "loss": 0.0972, "step": 27234 }, { "epoch": 19.952380952380953, "grad_norm": 0.73046875, "learning_rate": 4.3153708844560114e-05, "loss": 0.1138, "step": 27235 }, { "epoch": 19.953113553113553, "grad_norm": 0.89453125, "learning_rate": 4.3150320444574974e-05, "loss": 0.1068, "step": 27236 }, { "epoch": 19.953846153846154, "grad_norm": 0.80859375, "learning_rate": 4.314693207980456e-05, "loss": 0.1046, "step": 27237 }, { "epoch": 19.954578754578755, "grad_norm": 1.03125, "learning_rate": 4.314354375026512e-05, "loss": 0.1358, "step": 27238 }, { "epoch": 19.955311355311355, "grad_norm": 0.8203125, "learning_rate": 4.314015545597287e-05, "loss": 0.1254, "step": 27239 }, { "epoch": 19.956043956043956, "grad_norm": 1.28125, "learning_rate": 4.3136767196944064e-05, "loss": 0.1258, "step": 27240 }, { "epoch": 19.956776556776557, "grad_norm": 1.1015625, "learning_rate": 4.3133378973194896e-05, "loss": 0.1209, "step": 27241 }, { "epoch": 19.957509157509158, "grad_norm": 0.79296875, "learning_rate": 4.3129990784741644e-05, "loss": 0.1129, "step": 27242 }, { "epoch": 19.95824175824176, "grad_norm": 0.87109375, "learning_rate": 4.312660263160053e-05, "loss": 0.0916, "step": 27243 }, { "epoch": 19.95897435897436, "grad_norm": 1.1640625, "learning_rate": 4.312321451378778e-05, "loss": 0.1315, "step": 27244 }, { "epoch": 19.95970695970696, "grad_norm": 1.171875, "learning_rate": 4.3119826431319624e-05, "loss": 0.1296, "step": 27245 }, { "epoch": 19.96043956043956, "grad_norm": 1.1484375, "learning_rate": 4.3116438384212305e-05, "loss": 0.1194, "step": 27246 }, { "epoch": 19.96117216117216, "grad_norm": 1.0078125, "learning_rate": 4.311305037248206e-05, "loss": 0.1112, "step": 27247 }, { "epoch": 19.961904761904762, "grad_norm": 1.015625, "learning_rate": 4.310966239614511e-05, "loss": 0.1205, "step": 27248 }, { "epoch": 19.962637362637363, "grad_norm": 0.94921875, "learning_rate": 4.31062744552177e-05, "loss": 0.1232, "step": 27249 }, { "epoch": 19.963369963369964, "grad_norm": 1.796875, "learning_rate": 4.310288654971605e-05, "loss": 0.1294, "step": 27250 }, { "epoch": 19.964102564102564, "grad_norm": 1.671875, "learning_rate": 4.309949867965639e-05, "loss": 0.1167, "step": 27251 }, { "epoch": 19.964835164835165, "grad_norm": 1.6875, "learning_rate": 4.309611084505498e-05, "loss": 0.1187, "step": 27252 }, { "epoch": 19.965567765567766, "grad_norm": 1.953125, "learning_rate": 4.309272304592802e-05, "loss": 0.1338, "step": 27253 }, { "epoch": 19.966300366300366, "grad_norm": 0.9921875, "learning_rate": 4.308933528229177e-05, "loss": 0.1413, "step": 27254 }, { "epoch": 19.967032967032967, "grad_norm": 0.68359375, "learning_rate": 4.308594755416245e-05, "loss": 0.1228, "step": 27255 }, { "epoch": 19.967765567765568, "grad_norm": 1.109375, "learning_rate": 4.308255986155628e-05, "loss": 0.1267, "step": 27256 }, { "epoch": 19.96849816849817, "grad_norm": 1.7734375, "learning_rate": 4.307917220448948e-05, "loss": 0.1786, "step": 27257 }, { "epoch": 19.96923076923077, "grad_norm": 1.390625, "learning_rate": 4.307578458297832e-05, "loss": 0.1189, "step": 27258 }, { "epoch": 19.96996336996337, "grad_norm": 0.8828125, "learning_rate": 4.307239699703902e-05, "loss": 0.1386, "step": 27259 }, { "epoch": 19.97069597069597, "grad_norm": 1.21875, "learning_rate": 4.3069009446687794e-05, "loss": 0.1297, "step": 27260 }, { "epoch": 19.97142857142857, "grad_norm": 1.03125, "learning_rate": 4.306562193194089e-05, "loss": 0.124, "step": 27261 }, { "epoch": 19.972161172161172, "grad_norm": 1.4375, "learning_rate": 4.306223445281451e-05, "loss": 0.0955, "step": 27262 }, { "epoch": 19.972893772893773, "grad_norm": 1.2265625, "learning_rate": 4.305884700932493e-05, "loss": 0.1202, "step": 27263 }, { "epoch": 19.973626373626374, "grad_norm": 1.078125, "learning_rate": 4.305545960148836e-05, "loss": 0.0964, "step": 27264 }, { "epoch": 19.974358974358974, "grad_norm": 0.85546875, "learning_rate": 4.305207222932102e-05, "loss": 0.1177, "step": 27265 }, { "epoch": 19.975091575091575, "grad_norm": 1.125, "learning_rate": 4.304868489283915e-05, "loss": 0.1538, "step": 27266 }, { "epoch": 19.975824175824176, "grad_norm": 0.9140625, "learning_rate": 4.304529759205896e-05, "loss": 0.1161, "step": 27267 }, { "epoch": 19.976556776556777, "grad_norm": 1.296875, "learning_rate": 4.304191032699671e-05, "loss": 0.1181, "step": 27268 }, { "epoch": 19.977289377289377, "grad_norm": 0.78515625, "learning_rate": 4.3038523097668626e-05, "loss": 0.1306, "step": 27269 }, { "epoch": 19.978021978021978, "grad_norm": 0.6015625, "learning_rate": 4.3035135904090916e-05, "loss": 0.1274, "step": 27270 }, { "epoch": 19.97875457875458, "grad_norm": 1.3125, "learning_rate": 4.303174874627982e-05, "loss": 0.1507, "step": 27271 }, { "epoch": 19.97948717948718, "grad_norm": 1.0234375, "learning_rate": 4.302836162425156e-05, "loss": 0.1063, "step": 27272 }, { "epoch": 19.98021978021978, "grad_norm": 0.90234375, "learning_rate": 4.302497453802238e-05, "loss": 0.1081, "step": 27273 }, { "epoch": 19.98095238095238, "grad_norm": 0.9609375, "learning_rate": 4.30215874876085e-05, "loss": 0.1073, "step": 27274 }, { "epoch": 19.98168498168498, "grad_norm": 1.4296875, "learning_rate": 4.301820047302615e-05, "loss": 0.1164, "step": 27275 }, { "epoch": 19.982417582417582, "grad_norm": 1.734375, "learning_rate": 4.301481349429157e-05, "loss": 0.134, "step": 27276 }, { "epoch": 19.983150183150183, "grad_norm": 1.484375, "learning_rate": 4.301142655142094e-05, "loss": 0.1692, "step": 27277 }, { "epoch": 19.983882783882784, "grad_norm": 1.234375, "learning_rate": 4.300803964443055e-05, "loss": 0.1051, "step": 27278 }, { "epoch": 19.984615384615385, "grad_norm": 1.984375, "learning_rate": 4.300465277333661e-05, "loss": 0.1234, "step": 27279 }, { "epoch": 19.985347985347985, "grad_norm": 1.0078125, "learning_rate": 4.300126593815533e-05, "loss": 0.109, "step": 27280 }, { "epoch": 19.986080586080586, "grad_norm": 1.15625, "learning_rate": 4.299787913890296e-05, "loss": 0.1513, "step": 27281 }, { "epoch": 19.986813186813187, "grad_norm": 0.71875, "learning_rate": 4.29944923755957e-05, "loss": 0.1165, "step": 27282 }, { "epoch": 19.987545787545788, "grad_norm": 0.9765625, "learning_rate": 4.299110564824977e-05, "loss": 0.1231, "step": 27283 }, { "epoch": 19.98827838827839, "grad_norm": 1.0703125, "learning_rate": 4.298771895688144e-05, "loss": 0.1044, "step": 27284 }, { "epoch": 19.98901098901099, "grad_norm": 0.75390625, "learning_rate": 4.298433230150692e-05, "loss": 0.1052, "step": 27285 }, { "epoch": 19.98974358974359, "grad_norm": 1.1328125, "learning_rate": 4.298094568214242e-05, "loss": 0.1192, "step": 27286 }, { "epoch": 19.99047619047619, "grad_norm": 1.15625, "learning_rate": 4.2977559098804185e-05, "loss": 0.1209, "step": 27287 }, { "epoch": 19.99120879120879, "grad_norm": 0.80078125, "learning_rate": 4.297417255150841e-05, "loss": 0.1222, "step": 27288 }, { "epoch": 19.991941391941392, "grad_norm": 0.8828125, "learning_rate": 4.297078604027137e-05, "loss": 0.1246, "step": 27289 }, { "epoch": 19.992673992673993, "grad_norm": 1.15625, "learning_rate": 4.296739956510927e-05, "loss": 0.1231, "step": 27290 }, { "epoch": 19.993406593406593, "grad_norm": 0.84765625, "learning_rate": 4.2964013126038326e-05, "loss": 0.1358, "step": 27291 }, { "epoch": 19.994139194139194, "grad_norm": 1.40625, "learning_rate": 4.2960626723074766e-05, "loss": 0.0967, "step": 27292 }, { "epoch": 19.994871794871795, "grad_norm": 0.57421875, "learning_rate": 4.29572403562348e-05, "loss": 0.116, "step": 27293 }, { "epoch": 19.995604395604396, "grad_norm": 1.046875, "learning_rate": 4.295385402553469e-05, "loss": 0.142, "step": 27294 }, { "epoch": 19.996336996336996, "grad_norm": 0.98828125, "learning_rate": 4.295046773099065e-05, "loss": 0.121, "step": 27295 }, { "epoch": 19.997069597069597, "grad_norm": 0.7421875, "learning_rate": 4.29470814726189e-05, "loss": 0.1149, "step": 27296 }, { "epoch": 19.997802197802198, "grad_norm": 1.25, "learning_rate": 4.294369525043564e-05, "loss": 0.1127, "step": 27297 }, { "epoch": 19.9985347985348, "grad_norm": 0.8515625, "learning_rate": 4.2940309064457124e-05, "loss": 0.1133, "step": 27298 }, { "epoch": 19.9992673992674, "grad_norm": 2.03125, "learning_rate": 4.2936922914699576e-05, "loss": 0.126, "step": 27299 }, { "epoch": 20.0, "grad_norm": 1.28125, "learning_rate": 4.29335368011792e-05, "loss": 0.1348, "step": 27300 }, { "epoch": 20.0007326007326, "grad_norm": 0.78515625, "learning_rate": 4.293015072391225e-05, "loss": 0.1276, "step": 27301 }, { "epoch": 20.0014652014652, "grad_norm": 0.96484375, "learning_rate": 4.292676468291493e-05, "loss": 0.1193, "step": 27302 }, { "epoch": 20.002197802197802, "grad_norm": 1.1796875, "learning_rate": 4.292337867820345e-05, "loss": 0.1128, "step": 27303 }, { "epoch": 20.002930402930403, "grad_norm": 0.640625, "learning_rate": 4.291999270979406e-05, "loss": 0.113, "step": 27304 }, { "epoch": 20.003663003663004, "grad_norm": 0.86328125, "learning_rate": 4.291660677770298e-05, "loss": 0.1151, "step": 27305 }, { "epoch": 20.004395604395604, "grad_norm": 0.8359375, "learning_rate": 4.291322088194643e-05, "loss": 0.1008, "step": 27306 }, { "epoch": 20.005128205128205, "grad_norm": 0.734375, "learning_rate": 4.290983502254063e-05, "loss": 0.1374, "step": 27307 }, { "epoch": 20.005860805860806, "grad_norm": 0.9609375, "learning_rate": 4.290644919950179e-05, "loss": 0.1107, "step": 27308 }, { "epoch": 20.006593406593407, "grad_norm": 1.1328125, "learning_rate": 4.290306341284613e-05, "loss": 0.1541, "step": 27309 }, { "epoch": 20.007326007326007, "grad_norm": 1.1328125, "learning_rate": 4.289967766258992e-05, "loss": 0.1251, "step": 27310 }, { "epoch": 20.008058608058608, "grad_norm": 0.87890625, "learning_rate": 4.2896291948749326e-05, "loss": 0.1113, "step": 27311 }, { "epoch": 20.00879120879121, "grad_norm": 0.79296875, "learning_rate": 4.289290627134061e-05, "loss": 0.1147, "step": 27312 }, { "epoch": 20.00952380952381, "grad_norm": 0.59375, "learning_rate": 4.288952063037997e-05, "loss": 0.0973, "step": 27313 }, { "epoch": 20.01025641025641, "grad_norm": 1.390625, "learning_rate": 4.288613502588362e-05, "loss": 0.1173, "step": 27314 }, { "epoch": 20.01098901098901, "grad_norm": 1.7421875, "learning_rate": 4.288274945786781e-05, "loss": 0.144, "step": 27315 }, { "epoch": 20.01172161172161, "grad_norm": 1.03125, "learning_rate": 4.287936392634876e-05, "loss": 0.1103, "step": 27316 }, { "epoch": 20.012454212454212, "grad_norm": 0.8359375, "learning_rate": 4.2875978431342666e-05, "loss": 0.1044, "step": 27317 }, { "epoch": 20.013186813186813, "grad_norm": 1.3515625, "learning_rate": 4.2872592972865775e-05, "loss": 0.1319, "step": 27318 }, { "epoch": 20.013919413919414, "grad_norm": 1.28125, "learning_rate": 4.286920755093426e-05, "loss": 0.1185, "step": 27319 }, { "epoch": 20.014652014652015, "grad_norm": 1.0859375, "learning_rate": 4.2865822165564406e-05, "loss": 0.1249, "step": 27320 }, { "epoch": 20.015384615384615, "grad_norm": 1.78125, "learning_rate": 4.286243681677241e-05, "loss": 0.1469, "step": 27321 }, { "epoch": 20.016117216117216, "grad_norm": 1.5390625, "learning_rate": 4.2859051504574474e-05, "loss": 0.1054, "step": 27322 }, { "epoch": 20.016849816849817, "grad_norm": 0.66015625, "learning_rate": 4.285566622898682e-05, "loss": 0.117, "step": 27323 }, { "epoch": 20.017582417582418, "grad_norm": 0.83984375, "learning_rate": 4.28522809900257e-05, "loss": 0.1134, "step": 27324 }, { "epoch": 20.01831501831502, "grad_norm": 1.1015625, "learning_rate": 4.28488957877073e-05, "loss": 0.1049, "step": 27325 }, { "epoch": 20.01904761904762, "grad_norm": 1.296875, "learning_rate": 4.2845510622047846e-05, "loss": 0.108, "step": 27326 }, { "epoch": 20.01978021978022, "grad_norm": 0.94921875, "learning_rate": 4.284212549306358e-05, "loss": 0.1304, "step": 27327 }, { "epoch": 20.02051282051282, "grad_norm": 0.83203125, "learning_rate": 4.283874040077071e-05, "loss": 0.1143, "step": 27328 }, { "epoch": 20.02124542124542, "grad_norm": 0.74609375, "learning_rate": 4.283535534518541e-05, "loss": 0.1044, "step": 27329 }, { "epoch": 20.021978021978022, "grad_norm": 1.4140625, "learning_rate": 4.283197032632397e-05, "loss": 0.1096, "step": 27330 }, { "epoch": 20.022710622710623, "grad_norm": 1.3359375, "learning_rate": 4.2828585344202576e-05, "loss": 0.1398, "step": 27331 }, { "epoch": 20.023443223443223, "grad_norm": 1.015625, "learning_rate": 4.282520039883745e-05, "loss": 0.1241, "step": 27332 }, { "epoch": 20.024175824175824, "grad_norm": 1.1484375, "learning_rate": 4.282181549024481e-05, "loss": 0.116, "step": 27333 }, { "epoch": 20.024908424908425, "grad_norm": 0.85546875, "learning_rate": 4.281843061844087e-05, "loss": 0.1189, "step": 27334 }, { "epoch": 20.025641025641026, "grad_norm": 1.078125, "learning_rate": 4.2815045783441826e-05, "loss": 0.1187, "step": 27335 }, { "epoch": 20.026373626373626, "grad_norm": 0.76953125, "learning_rate": 4.281166098526395e-05, "loss": 0.1399, "step": 27336 }, { "epoch": 20.027106227106227, "grad_norm": 1.109375, "learning_rate": 4.2808276223923415e-05, "loss": 0.1228, "step": 27337 }, { "epoch": 20.027838827838828, "grad_norm": 0.9296875, "learning_rate": 4.280489149943646e-05, "loss": 0.1352, "step": 27338 }, { "epoch": 20.02857142857143, "grad_norm": 1.109375, "learning_rate": 4.2801506811819304e-05, "loss": 0.1344, "step": 27339 }, { "epoch": 20.02930402930403, "grad_norm": 1.46875, "learning_rate": 4.2798122161088136e-05, "loss": 0.1175, "step": 27340 }, { "epoch": 20.03003663003663, "grad_norm": 0.97265625, "learning_rate": 4.27947375472592e-05, "loss": 0.1121, "step": 27341 }, { "epoch": 20.03076923076923, "grad_norm": 1.4296875, "learning_rate": 4.2791352970348714e-05, "loss": 0.1112, "step": 27342 }, { "epoch": 20.03150183150183, "grad_norm": 1.0234375, "learning_rate": 4.278796843037289e-05, "loss": 0.1044, "step": 27343 }, { "epoch": 20.032234432234432, "grad_norm": 0.90625, "learning_rate": 4.278458392734794e-05, "loss": 0.1155, "step": 27344 }, { "epoch": 20.032967032967033, "grad_norm": 1.4140625, "learning_rate": 4.278119946129006e-05, "loss": 0.166, "step": 27345 }, { "epoch": 20.033699633699634, "grad_norm": 1.8359375, "learning_rate": 4.2777815032215506e-05, "loss": 0.1454, "step": 27346 }, { "epoch": 20.034432234432234, "grad_norm": 0.984375, "learning_rate": 4.277443064014048e-05, "loss": 0.1265, "step": 27347 }, { "epoch": 20.035164835164835, "grad_norm": 0.75390625, "learning_rate": 4.277104628508118e-05, "loss": 0.1267, "step": 27348 }, { "epoch": 20.035897435897436, "grad_norm": 1.25, "learning_rate": 4.2767661967053844e-05, "loss": 0.1504, "step": 27349 }, { "epoch": 20.036630036630036, "grad_norm": 0.87109375, "learning_rate": 4.276427768607466e-05, "loss": 0.1067, "step": 27350 }, { "epoch": 20.037362637362637, "grad_norm": 0.91015625, "learning_rate": 4.2760893442159886e-05, "loss": 0.1131, "step": 27351 }, { "epoch": 20.038095238095238, "grad_norm": 1.09375, "learning_rate": 4.2757509235325685e-05, "loss": 0.1231, "step": 27352 }, { "epoch": 20.03882783882784, "grad_norm": 1.0078125, "learning_rate": 4.2754125065588315e-05, "loss": 0.1911, "step": 27353 }, { "epoch": 20.03956043956044, "grad_norm": 1.0859375, "learning_rate": 4.2750740932963986e-05, "loss": 0.1208, "step": 27354 }, { "epoch": 20.04029304029304, "grad_norm": 1.09375, "learning_rate": 4.274735683746887e-05, "loss": 0.1172, "step": 27355 }, { "epoch": 20.04102564102564, "grad_norm": 0.96875, "learning_rate": 4.274397277911924e-05, "loss": 0.1098, "step": 27356 }, { "epoch": 20.04175824175824, "grad_norm": 1.09375, "learning_rate": 4.274058875793127e-05, "loss": 0.1262, "step": 27357 }, { "epoch": 20.042490842490842, "grad_norm": 1.609375, "learning_rate": 4.27372047739212e-05, "loss": 0.1492, "step": 27358 }, { "epoch": 20.043223443223443, "grad_norm": 1.2265625, "learning_rate": 4.2733820827105225e-05, "loss": 0.1211, "step": 27359 }, { "epoch": 20.043956043956044, "grad_norm": 1.1875, "learning_rate": 4.273043691749955e-05, "loss": 0.1522, "step": 27360 }, { "epoch": 20.044688644688645, "grad_norm": 0.8984375, "learning_rate": 4.2727053045120406e-05, "loss": 0.1098, "step": 27361 }, { "epoch": 20.045421245421245, "grad_norm": 1.0390625, "learning_rate": 4.2723669209984015e-05, "loss": 0.134, "step": 27362 }, { "epoch": 20.046153846153846, "grad_norm": 0.765625, "learning_rate": 4.272028541210657e-05, "loss": 0.111, "step": 27363 }, { "epoch": 20.046886446886447, "grad_norm": 1.1328125, "learning_rate": 4.2716901651504296e-05, "loss": 0.1381, "step": 27364 }, { "epoch": 20.047619047619047, "grad_norm": 0.73828125, "learning_rate": 4.2713517928193405e-05, "loss": 0.1188, "step": 27365 }, { "epoch": 20.048351648351648, "grad_norm": 1.8359375, "learning_rate": 4.271013424219007e-05, "loss": 0.105, "step": 27366 }, { "epoch": 20.04908424908425, "grad_norm": 1.3359375, "learning_rate": 4.270675059351058e-05, "loss": 0.1091, "step": 27367 }, { "epoch": 20.04981684981685, "grad_norm": 1.046875, "learning_rate": 4.27033669821711e-05, "loss": 0.1393, "step": 27368 }, { "epoch": 20.05054945054945, "grad_norm": 0.91796875, "learning_rate": 4.269998340818784e-05, "loss": 0.1082, "step": 27369 }, { "epoch": 20.05128205128205, "grad_norm": 0.76171875, "learning_rate": 4.269659987157703e-05, "loss": 0.109, "step": 27370 }, { "epoch": 20.052014652014652, "grad_norm": 1.3515625, "learning_rate": 4.269321637235485e-05, "loss": 0.1462, "step": 27371 }, { "epoch": 20.052747252747253, "grad_norm": 1.0703125, "learning_rate": 4.2689832910537555e-05, "loss": 0.1003, "step": 27372 }, { "epoch": 20.053479853479853, "grad_norm": 1.1171875, "learning_rate": 4.2686449486141325e-05, "loss": 0.1336, "step": 27373 }, { "epoch": 20.054212454212454, "grad_norm": 1.046875, "learning_rate": 4.2683066099182386e-05, "loss": 0.1214, "step": 27374 }, { "epoch": 20.054945054945055, "grad_norm": 1.0859375, "learning_rate": 4.267968274967693e-05, "loss": 0.124, "step": 27375 }, { "epoch": 20.055677655677655, "grad_norm": 0.75, "learning_rate": 4.2676299437641196e-05, "loss": 0.1125, "step": 27376 }, { "epoch": 20.056410256410256, "grad_norm": 0.8359375, "learning_rate": 4.2672916163091355e-05, "loss": 0.133, "step": 27377 }, { "epoch": 20.057142857142857, "grad_norm": 1.1484375, "learning_rate": 4.2669532926043675e-05, "loss": 0.1055, "step": 27378 }, { "epoch": 20.057875457875458, "grad_norm": 0.91796875, "learning_rate": 4.266614972651432e-05, "loss": 0.1258, "step": 27379 }, { "epoch": 20.05860805860806, "grad_norm": 0.95703125, "learning_rate": 4.2662766564519515e-05, "loss": 0.1332, "step": 27380 }, { "epoch": 20.05934065934066, "grad_norm": 1.265625, "learning_rate": 4.265938344007546e-05, "loss": 0.1375, "step": 27381 }, { "epoch": 20.06007326007326, "grad_norm": 0.94921875, "learning_rate": 4.265600035319838e-05, "loss": 0.1382, "step": 27382 }, { "epoch": 20.06080586080586, "grad_norm": 0.99609375, "learning_rate": 4.2652617303904476e-05, "loss": 0.1275, "step": 27383 }, { "epoch": 20.06153846153846, "grad_norm": 1.046875, "learning_rate": 4.2649234292209966e-05, "loss": 0.1255, "step": 27384 }, { "epoch": 20.062271062271062, "grad_norm": 1.0859375, "learning_rate": 4.2645851318131044e-05, "loss": 0.126, "step": 27385 }, { "epoch": 20.063003663003663, "grad_norm": 1.328125, "learning_rate": 4.2642468381683925e-05, "loss": 0.1391, "step": 27386 }, { "epoch": 20.063736263736264, "grad_norm": 0.78515625, "learning_rate": 4.263908548288482e-05, "loss": 0.1382, "step": 27387 }, { "epoch": 20.064468864468864, "grad_norm": 0.7109375, "learning_rate": 4.263570262174994e-05, "loss": 0.1177, "step": 27388 }, { "epoch": 20.065201465201465, "grad_norm": 0.94921875, "learning_rate": 4.2632319798295485e-05, "loss": 0.1292, "step": 27389 }, { "epoch": 20.065934065934066, "grad_norm": 0.85546875, "learning_rate": 4.2628937012537676e-05, "loss": 0.1239, "step": 27390 }, { "epoch": 20.066666666666666, "grad_norm": 1.1484375, "learning_rate": 4.262555426449272e-05, "loss": 0.1188, "step": 27391 }, { "epoch": 20.067399267399267, "grad_norm": 0.671875, "learning_rate": 4.2622171554176795e-05, "loss": 0.1273, "step": 27392 }, { "epoch": 20.068131868131868, "grad_norm": 1.0625, "learning_rate": 4.261878888160614e-05, "loss": 0.1192, "step": 27393 }, { "epoch": 20.06886446886447, "grad_norm": 1.03125, "learning_rate": 4.2615406246796976e-05, "loss": 0.1655, "step": 27394 }, { "epoch": 20.06959706959707, "grad_norm": 1.046875, "learning_rate": 4.261202364976547e-05, "loss": 0.1297, "step": 27395 }, { "epoch": 20.07032967032967, "grad_norm": 1.0859375, "learning_rate": 4.260864109052786e-05, "loss": 0.1292, "step": 27396 }, { "epoch": 20.07106227106227, "grad_norm": 1.0, "learning_rate": 4.260525856910032e-05, "loss": 0.1469, "step": 27397 }, { "epoch": 20.07179487179487, "grad_norm": 1.0546875, "learning_rate": 4.260187608549909e-05, "loss": 0.1, "step": 27398 }, { "epoch": 20.072527472527472, "grad_norm": 1.4609375, "learning_rate": 4.259849363974037e-05, "loss": 0.0934, "step": 27399 }, { "epoch": 20.073260073260073, "grad_norm": 1.125, "learning_rate": 4.259511123184037e-05, "loss": 0.1022, "step": 27400 }, { "epoch": 20.073992673992674, "grad_norm": 1.15625, "learning_rate": 4.259172886181526e-05, "loss": 0.1515, "step": 27401 }, { "epoch": 20.074725274725274, "grad_norm": 1.859375, "learning_rate": 4.2588346529681285e-05, "loss": 0.126, "step": 27402 }, { "epoch": 20.075457875457875, "grad_norm": 0.7890625, "learning_rate": 4.2584964235454635e-05, "loss": 0.1097, "step": 27403 }, { "epoch": 20.076190476190476, "grad_norm": 1.6640625, "learning_rate": 4.258158197915153e-05, "loss": 0.1048, "step": 27404 }, { "epoch": 20.076923076923077, "grad_norm": 1.265625, "learning_rate": 4.257819976078817e-05, "loss": 0.1283, "step": 27405 }, { "epoch": 20.077655677655677, "grad_norm": 0.73828125, "learning_rate": 4.2574817580380746e-05, "loss": 0.1068, "step": 27406 }, { "epoch": 20.078388278388278, "grad_norm": 1.0078125, "learning_rate": 4.2571435437945456e-05, "loss": 0.1119, "step": 27407 }, { "epoch": 20.07912087912088, "grad_norm": 0.796875, "learning_rate": 4.256805333349854e-05, "loss": 0.1154, "step": 27408 }, { "epoch": 20.07985347985348, "grad_norm": 1.2421875, "learning_rate": 4.2564671267056197e-05, "loss": 0.1169, "step": 27409 }, { "epoch": 20.08058608058608, "grad_norm": 1.0078125, "learning_rate": 4.25612892386346e-05, "loss": 0.1222, "step": 27410 }, { "epoch": 20.08131868131868, "grad_norm": 0.92578125, "learning_rate": 4.255790724824998e-05, "loss": 0.1141, "step": 27411 }, { "epoch": 20.08205128205128, "grad_norm": 0.59375, "learning_rate": 4.255452529591853e-05, "loss": 0.1079, "step": 27412 }, { "epoch": 20.082783882783882, "grad_norm": 1.265625, "learning_rate": 4.2551143381656455e-05, "loss": 0.1137, "step": 27413 }, { "epoch": 20.083516483516483, "grad_norm": 0.94140625, "learning_rate": 4.254776150547997e-05, "loss": 0.1351, "step": 27414 }, { "epoch": 20.084249084249084, "grad_norm": 0.890625, "learning_rate": 4.254437966740525e-05, "loss": 0.1236, "step": 27415 }, { "epoch": 20.084981684981685, "grad_norm": 0.95703125, "learning_rate": 4.2540997867448545e-05, "loss": 0.1177, "step": 27416 }, { "epoch": 20.085714285714285, "grad_norm": 1.109375, "learning_rate": 4.2537616105626026e-05, "loss": 0.1639, "step": 27417 }, { "epoch": 20.086446886446886, "grad_norm": 1.125, "learning_rate": 4.253423438195388e-05, "loss": 0.1129, "step": 27418 }, { "epoch": 20.087179487179487, "grad_norm": 1.5625, "learning_rate": 4.253085269644835e-05, "loss": 0.1745, "step": 27419 }, { "epoch": 20.087912087912088, "grad_norm": 0.93359375, "learning_rate": 4.252747104912563e-05, "loss": 0.1217, "step": 27420 }, { "epoch": 20.08864468864469, "grad_norm": 2.078125, "learning_rate": 4.252408944000191e-05, "loss": 0.1441, "step": 27421 }, { "epoch": 20.08937728937729, "grad_norm": 0.71484375, "learning_rate": 4.252070786909339e-05, "loss": 0.1183, "step": 27422 }, { "epoch": 20.09010989010989, "grad_norm": 0.79296875, "learning_rate": 4.251732633641626e-05, "loss": 0.1008, "step": 27423 }, { "epoch": 20.09084249084249, "grad_norm": 0.86328125, "learning_rate": 4.251394484198676e-05, "loss": 0.1158, "step": 27424 }, { "epoch": 20.09157509157509, "grad_norm": 1.453125, "learning_rate": 4.251056338582107e-05, "loss": 0.1279, "step": 27425 }, { "epoch": 20.092307692307692, "grad_norm": 1.0703125, "learning_rate": 4.2507181967935394e-05, "loss": 0.1276, "step": 27426 }, { "epoch": 20.093040293040293, "grad_norm": 0.83203125, "learning_rate": 4.250380058834593e-05, "loss": 0.1153, "step": 27427 }, { "epoch": 20.093772893772893, "grad_norm": 1.3359375, "learning_rate": 4.250041924706888e-05, "loss": 0.0982, "step": 27428 }, { "epoch": 20.094505494505494, "grad_norm": 1.40625, "learning_rate": 4.249703794412045e-05, "loss": 0.1208, "step": 27429 }, { "epoch": 20.095238095238095, "grad_norm": 1.1953125, "learning_rate": 4.2493656679516836e-05, "loss": 0.1356, "step": 27430 }, { "epoch": 20.095970695970696, "grad_norm": 1.3828125, "learning_rate": 4.249027545327425e-05, "loss": 0.123, "step": 27431 }, { "epoch": 20.096703296703296, "grad_norm": 1.125, "learning_rate": 4.248689426540889e-05, "loss": 0.1167, "step": 27432 }, { "epoch": 20.097435897435897, "grad_norm": 0.75390625, "learning_rate": 4.248351311593692e-05, "loss": 0.1281, "step": 27433 }, { "epoch": 20.098168498168498, "grad_norm": 1.484375, "learning_rate": 4.248013200487459e-05, "loss": 0.0887, "step": 27434 }, { "epoch": 20.0989010989011, "grad_norm": 0.92578125, "learning_rate": 4.247675093223809e-05, "loss": 0.1201, "step": 27435 }, { "epoch": 20.0996336996337, "grad_norm": 1.015625, "learning_rate": 4.24733698980436e-05, "loss": 0.1423, "step": 27436 }, { "epoch": 20.1003663003663, "grad_norm": 1.0390625, "learning_rate": 4.2469988902307335e-05, "loss": 0.115, "step": 27437 }, { "epoch": 20.1010989010989, "grad_norm": 0.86328125, "learning_rate": 4.246660794504548e-05, "loss": 0.109, "step": 27438 }, { "epoch": 20.1018315018315, "grad_norm": 0.6953125, "learning_rate": 4.246322702627424e-05, "loss": 0.1017, "step": 27439 }, { "epoch": 20.102564102564102, "grad_norm": 0.77734375, "learning_rate": 4.245984614600983e-05, "loss": 0.0894, "step": 27440 }, { "epoch": 20.103296703296703, "grad_norm": 1.0859375, "learning_rate": 4.2456465304268424e-05, "loss": 0.1386, "step": 27441 }, { "epoch": 20.104029304029304, "grad_norm": 1.6484375, "learning_rate": 4.245308450106624e-05, "loss": 0.1532, "step": 27442 }, { "epoch": 20.104761904761904, "grad_norm": 1.0234375, "learning_rate": 4.244970373641948e-05, "loss": 0.1247, "step": 27443 }, { "epoch": 20.105494505494505, "grad_norm": 1.28125, "learning_rate": 4.24463230103443e-05, "loss": 0.0969, "step": 27444 }, { "epoch": 20.106227106227106, "grad_norm": 1.0078125, "learning_rate": 4.244294232285696e-05, "loss": 0.1039, "step": 27445 }, { "epoch": 20.106959706959707, "grad_norm": 1.21875, "learning_rate": 4.2439561673973626e-05, "loss": 0.1371, "step": 27446 }, { "epoch": 20.107692307692307, "grad_norm": 1.1640625, "learning_rate": 4.24361810637105e-05, "loss": 0.1306, "step": 27447 }, { "epoch": 20.108424908424908, "grad_norm": 1.0078125, "learning_rate": 4.2432800492083775e-05, "loss": 0.0999, "step": 27448 }, { "epoch": 20.10915750915751, "grad_norm": 0.921875, "learning_rate": 4.242941995910963e-05, "loss": 0.116, "step": 27449 }, { "epoch": 20.10989010989011, "grad_norm": 1.671875, "learning_rate": 4.2426039464804305e-05, "loss": 0.1637, "step": 27450 }, { "epoch": 20.11062271062271, "grad_norm": 0.72265625, "learning_rate": 4.2422659009183976e-05, "loss": 0.1085, "step": 27451 }, { "epoch": 20.11135531135531, "grad_norm": 0.98046875, "learning_rate": 4.241927859226483e-05, "loss": 0.1516, "step": 27452 }, { "epoch": 20.11208791208791, "grad_norm": 1.265625, "learning_rate": 4.241589821406308e-05, "loss": 0.1275, "step": 27453 }, { "epoch": 20.112820512820512, "grad_norm": 1.125, "learning_rate": 4.2412517874594905e-05, "loss": 0.0997, "step": 27454 }, { "epoch": 20.113553113553113, "grad_norm": 0.84765625, "learning_rate": 4.240913757387651e-05, "loss": 0.1341, "step": 27455 }, { "epoch": 20.114285714285714, "grad_norm": 1.015625, "learning_rate": 4.2405757311924113e-05, "loss": 0.1448, "step": 27456 }, { "epoch": 20.115018315018315, "grad_norm": 1.078125, "learning_rate": 4.2402377088753884e-05, "loss": 0.1164, "step": 27457 }, { "epoch": 20.115750915750915, "grad_norm": 1.2109375, "learning_rate": 4.239899690438202e-05, "loss": 0.1571, "step": 27458 }, { "epoch": 20.116483516483516, "grad_norm": 1.1640625, "learning_rate": 4.2395616758824706e-05, "loss": 0.1247, "step": 27459 }, { "epoch": 20.117216117216117, "grad_norm": 1.25, "learning_rate": 4.2392236652098176e-05, "loss": 0.1353, "step": 27460 }, { "epoch": 20.117948717948718, "grad_norm": 0.73046875, "learning_rate": 4.2388856584218596e-05, "loss": 0.1186, "step": 27461 }, { "epoch": 20.11868131868132, "grad_norm": 1.328125, "learning_rate": 4.238547655520217e-05, "loss": 0.1383, "step": 27462 }, { "epoch": 20.11941391941392, "grad_norm": 1.578125, "learning_rate": 4.238209656506509e-05, "loss": 0.1166, "step": 27463 }, { "epoch": 20.12014652014652, "grad_norm": 1.1640625, "learning_rate": 4.237871661382353e-05, "loss": 0.1211, "step": 27464 }, { "epoch": 20.12087912087912, "grad_norm": 1.1484375, "learning_rate": 4.237533670149372e-05, "loss": 0.1422, "step": 27465 }, { "epoch": 20.12161172161172, "grad_norm": 1.6015625, "learning_rate": 4.237195682809184e-05, "loss": 0.0948, "step": 27466 }, { "epoch": 20.122344322344322, "grad_norm": 1.484375, "learning_rate": 4.2368576993634076e-05, "loss": 0.1266, "step": 27467 }, { "epoch": 20.123076923076923, "grad_norm": 1.203125, "learning_rate": 4.2365197198136636e-05, "loss": 0.0921, "step": 27468 }, { "epoch": 20.123809523809523, "grad_norm": 1.109375, "learning_rate": 4.2361817441615716e-05, "loss": 0.1368, "step": 27469 }, { "epoch": 20.124542124542124, "grad_norm": 0.671875, "learning_rate": 4.235843772408747e-05, "loss": 0.1046, "step": 27470 }, { "epoch": 20.125274725274725, "grad_norm": 1.25, "learning_rate": 4.2355058045568145e-05, "loss": 0.1137, "step": 27471 }, { "epoch": 20.126007326007326, "grad_norm": 1.1875, "learning_rate": 4.2351678406073907e-05, "loss": 0.1161, "step": 27472 }, { "epoch": 20.126739926739926, "grad_norm": 0.77734375, "learning_rate": 4.2348298805620954e-05, "loss": 0.1051, "step": 27473 }, { "epoch": 20.127472527472527, "grad_norm": 0.68359375, "learning_rate": 4.2344919244225484e-05, "loss": 0.1071, "step": 27474 }, { "epoch": 20.128205128205128, "grad_norm": 1.0859375, "learning_rate": 4.2341539721903665e-05, "loss": 0.1394, "step": 27475 }, { "epoch": 20.12893772893773, "grad_norm": 1.3515625, "learning_rate": 4.233816023867172e-05, "loss": 0.1194, "step": 27476 }, { "epoch": 20.12967032967033, "grad_norm": 0.69921875, "learning_rate": 4.2334780794545824e-05, "loss": 0.0815, "step": 27477 }, { "epoch": 20.13040293040293, "grad_norm": 1.28125, "learning_rate": 4.233140138954217e-05, "loss": 0.1335, "step": 27478 }, { "epoch": 20.13113553113553, "grad_norm": 0.7265625, "learning_rate": 4.232802202367696e-05, "loss": 0.0877, "step": 27479 }, { "epoch": 20.13186813186813, "grad_norm": 0.9609375, "learning_rate": 4.2324642696966376e-05, "loss": 0.1248, "step": 27480 }, { "epoch": 20.132600732600732, "grad_norm": 1.0625, "learning_rate": 4.232126340942661e-05, "loss": 0.0992, "step": 27481 }, { "epoch": 20.133333333333333, "grad_norm": 1.6484375, "learning_rate": 4.231788416107386e-05, "loss": 0.1416, "step": 27482 }, { "epoch": 20.134065934065934, "grad_norm": 0.8828125, "learning_rate": 4.2314504951924324e-05, "loss": 0.1247, "step": 27483 }, { "epoch": 20.134798534798534, "grad_norm": 0.81640625, "learning_rate": 4.231112578199417e-05, "loss": 0.1449, "step": 27484 }, { "epoch": 20.135531135531135, "grad_norm": 0.765625, "learning_rate": 4.230774665129961e-05, "loss": 0.1284, "step": 27485 }, { "epoch": 20.136263736263736, "grad_norm": 1.03125, "learning_rate": 4.23043675598568e-05, "loss": 0.1301, "step": 27486 }, { "epoch": 20.136996336996337, "grad_norm": 1.0234375, "learning_rate": 4.230098850768198e-05, "loss": 0.1072, "step": 27487 }, { "epoch": 20.137728937728937, "grad_norm": 1.2578125, "learning_rate": 4.229760949479131e-05, "loss": 0.1261, "step": 27488 }, { "epoch": 20.138461538461538, "grad_norm": 1.2890625, "learning_rate": 4.229423052120099e-05, "loss": 0.1484, "step": 27489 }, { "epoch": 20.13919413919414, "grad_norm": 0.8203125, "learning_rate": 4.229085158692719e-05, "loss": 0.1194, "step": 27490 }, { "epoch": 20.13992673992674, "grad_norm": 0.6484375, "learning_rate": 4.2287472691986115e-05, "loss": 0.1071, "step": 27491 }, { "epoch": 20.14065934065934, "grad_norm": 1.28125, "learning_rate": 4.228409383639396e-05, "loss": 0.1323, "step": 27492 }, { "epoch": 20.14139194139194, "grad_norm": 1.2734375, "learning_rate": 4.228071502016692e-05, "loss": 0.116, "step": 27493 }, { "epoch": 20.14212454212454, "grad_norm": 1.2421875, "learning_rate": 4.227733624332117e-05, "loss": 0.1204, "step": 27494 }, { "epoch": 20.142857142857142, "grad_norm": 1.375, "learning_rate": 4.2273957505872895e-05, "loss": 0.1388, "step": 27495 }, { "epoch": 20.143589743589743, "grad_norm": 0.84765625, "learning_rate": 4.227057880783828e-05, "loss": 0.1083, "step": 27496 }, { "epoch": 20.144322344322344, "grad_norm": 0.8828125, "learning_rate": 4.226720014923354e-05, "loss": 0.1328, "step": 27497 }, { "epoch": 20.145054945054945, "grad_norm": 0.68359375, "learning_rate": 4.226382153007484e-05, "loss": 0.1043, "step": 27498 }, { "epoch": 20.145787545787545, "grad_norm": 1.3671875, "learning_rate": 4.226044295037839e-05, "loss": 0.1201, "step": 27499 }, { "epoch": 20.146520146520146, "grad_norm": 1.796875, "learning_rate": 4.2257064410160343e-05, "loss": 0.136, "step": 27500 }, { "epoch": 20.147252747252747, "grad_norm": 0.6328125, "learning_rate": 4.22536859094369e-05, "loss": 0.1302, "step": 27501 }, { "epoch": 20.147985347985347, "grad_norm": 1.1484375, "learning_rate": 4.2250307448224266e-05, "loss": 0.1304, "step": 27502 }, { "epoch": 20.148717948717948, "grad_norm": 1.4453125, "learning_rate": 4.224692902653862e-05, "loss": 0.1412, "step": 27503 }, { "epoch": 20.14945054945055, "grad_norm": 1.1796875, "learning_rate": 4.2243550644396144e-05, "loss": 0.1056, "step": 27504 }, { "epoch": 20.15018315018315, "grad_norm": 1.21875, "learning_rate": 4.224017230181304e-05, "loss": 0.1137, "step": 27505 }, { "epoch": 20.15091575091575, "grad_norm": 1.1015625, "learning_rate": 4.223679399880546e-05, "loss": 0.1217, "step": 27506 }, { "epoch": 20.15164835164835, "grad_norm": 0.87890625, "learning_rate": 4.2233415735389616e-05, "loss": 0.1332, "step": 27507 }, { "epoch": 20.152380952380952, "grad_norm": 0.984375, "learning_rate": 4.22300375115817e-05, "loss": 0.104, "step": 27508 }, { "epoch": 20.153113553113553, "grad_norm": 1.5078125, "learning_rate": 4.2226659327397887e-05, "loss": 0.1293, "step": 27509 }, { "epoch": 20.153846153846153, "grad_norm": 0.94140625, "learning_rate": 4.222328118285437e-05, "loss": 0.1102, "step": 27510 }, { "epoch": 20.154578754578754, "grad_norm": 1.796875, "learning_rate": 4.221990307796733e-05, "loss": 0.1137, "step": 27511 }, { "epoch": 20.155311355311355, "grad_norm": 0.83984375, "learning_rate": 4.2216525012752935e-05, "loss": 0.0974, "step": 27512 }, { "epoch": 20.156043956043955, "grad_norm": 2.109375, "learning_rate": 4.2213146987227407e-05, "loss": 0.1318, "step": 27513 }, { "epoch": 20.156776556776556, "grad_norm": 1.9609375, "learning_rate": 4.220976900140691e-05, "loss": 0.1819, "step": 27514 }, { "epoch": 20.157509157509157, "grad_norm": 1.34375, "learning_rate": 4.2206391055307624e-05, "loss": 0.1555, "step": 27515 }, { "epoch": 20.158241758241758, "grad_norm": 1.9375, "learning_rate": 4.220301314894574e-05, "loss": 0.1093, "step": 27516 }, { "epoch": 20.15897435897436, "grad_norm": 0.875, "learning_rate": 4.219963528233744e-05, "loss": 0.1296, "step": 27517 }, { "epoch": 20.15970695970696, "grad_norm": 1.140625, "learning_rate": 4.219625745549892e-05, "loss": 0.1107, "step": 27518 }, { "epoch": 20.16043956043956, "grad_norm": 1.015625, "learning_rate": 4.219287966844635e-05, "loss": 0.1096, "step": 27519 }, { "epoch": 20.16117216117216, "grad_norm": 0.68359375, "learning_rate": 4.2189501921195936e-05, "loss": 0.1223, "step": 27520 }, { "epoch": 20.16190476190476, "grad_norm": 0.6640625, "learning_rate": 4.2186124213763846e-05, "loss": 0.1165, "step": 27521 }, { "epoch": 20.162637362637362, "grad_norm": 0.9296875, "learning_rate": 4.218274654616623e-05, "loss": 0.13, "step": 27522 }, { "epoch": 20.163369963369963, "grad_norm": 1.359375, "learning_rate": 4.2179368918419336e-05, "loss": 0.116, "step": 27523 }, { "epoch": 20.164102564102564, "grad_norm": 1.875, "learning_rate": 4.217599133053931e-05, "loss": 0.1085, "step": 27524 }, { "epoch": 20.164835164835164, "grad_norm": 0.9765625, "learning_rate": 4.217261378254236e-05, "loss": 0.1117, "step": 27525 }, { "epoch": 20.165567765567765, "grad_norm": 1.2421875, "learning_rate": 4.216923627444463e-05, "loss": 0.1177, "step": 27526 }, { "epoch": 20.166300366300366, "grad_norm": 1.0234375, "learning_rate": 4.216585880626232e-05, "loss": 0.1274, "step": 27527 }, { "epoch": 20.167032967032966, "grad_norm": 1.1875, "learning_rate": 4.2162481378011624e-05, "loss": 0.1136, "step": 27528 }, { "epoch": 20.167765567765567, "grad_norm": 1.0703125, "learning_rate": 4.2159103989708724e-05, "loss": 0.1136, "step": 27529 }, { "epoch": 20.168498168498168, "grad_norm": 1.015625, "learning_rate": 4.215572664136979e-05, "loss": 0.111, "step": 27530 }, { "epoch": 20.16923076923077, "grad_norm": 1.09375, "learning_rate": 4.215234933301101e-05, "loss": 0.1322, "step": 27531 }, { "epoch": 20.16996336996337, "grad_norm": 1.0078125, "learning_rate": 4.214897206464856e-05, "loss": 0.1228, "step": 27532 }, { "epoch": 20.17069597069597, "grad_norm": 1.1328125, "learning_rate": 4.2145594836298616e-05, "loss": 0.1211, "step": 27533 }, { "epoch": 20.17142857142857, "grad_norm": 1.0859375, "learning_rate": 4.214221764797739e-05, "loss": 0.1416, "step": 27534 }, { "epoch": 20.17216117216117, "grad_norm": 1.3828125, "learning_rate": 4.213884049970103e-05, "loss": 0.1029, "step": 27535 }, { "epoch": 20.172893772893772, "grad_norm": 0.6796875, "learning_rate": 4.213546339148575e-05, "loss": 0.0857, "step": 27536 }, { "epoch": 20.173626373626373, "grad_norm": 0.99609375, "learning_rate": 4.213208632334769e-05, "loss": 0.1129, "step": 27537 }, { "epoch": 20.174358974358974, "grad_norm": 1.1484375, "learning_rate": 4.212870929530305e-05, "loss": 0.1286, "step": 27538 }, { "epoch": 20.175091575091574, "grad_norm": 1.3671875, "learning_rate": 4.2125332307368025e-05, "loss": 0.1145, "step": 27539 }, { "epoch": 20.175824175824175, "grad_norm": 1.078125, "learning_rate": 4.212195535955878e-05, "loss": 0.1285, "step": 27540 }, { "epoch": 20.176556776556776, "grad_norm": 1.2890625, "learning_rate": 4.211857845189149e-05, "loss": 0.1154, "step": 27541 }, { "epoch": 20.177289377289377, "grad_norm": 1.1328125, "learning_rate": 4.2115201584382344e-05, "loss": 0.1033, "step": 27542 }, { "epoch": 20.178021978021977, "grad_norm": 1.1875, "learning_rate": 4.211182475704752e-05, "loss": 0.1195, "step": 27543 }, { "epoch": 20.178754578754578, "grad_norm": 1.28125, "learning_rate": 4.2108447969903194e-05, "loss": 0.1517, "step": 27544 }, { "epoch": 20.17948717948718, "grad_norm": 0.75, "learning_rate": 4.210507122296556e-05, "loss": 0.0983, "step": 27545 }, { "epoch": 20.18021978021978, "grad_norm": 0.5546875, "learning_rate": 4.210169451625078e-05, "loss": 0.1097, "step": 27546 }, { "epoch": 20.18095238095238, "grad_norm": 1.1484375, "learning_rate": 4.209831784977504e-05, "loss": 0.1361, "step": 27547 }, { "epoch": 20.18168498168498, "grad_norm": 0.86328125, "learning_rate": 4.20949412235545e-05, "loss": 0.1021, "step": 27548 }, { "epoch": 20.18241758241758, "grad_norm": 1.328125, "learning_rate": 4.209156463760537e-05, "loss": 0.1264, "step": 27549 }, { "epoch": 20.183150183150182, "grad_norm": 1.4375, "learning_rate": 4.208818809194382e-05, "loss": 0.1561, "step": 27550 }, { "epoch": 20.183882783882783, "grad_norm": 1.234375, "learning_rate": 4.208481158658603e-05, "loss": 0.107, "step": 27551 }, { "epoch": 20.184615384615384, "grad_norm": 1.15625, "learning_rate": 4.208143512154816e-05, "loss": 0.1366, "step": 27552 }, { "epoch": 20.185347985347985, "grad_norm": 1.4140625, "learning_rate": 4.2078058696846374e-05, "loss": 0.1289, "step": 27553 }, { "epoch": 20.186080586080585, "grad_norm": 1.21875, "learning_rate": 4.2074682312496905e-05, "loss": 0.1151, "step": 27554 }, { "epoch": 20.186813186813186, "grad_norm": 0.91015625, "learning_rate": 4.207130596851589e-05, "loss": 0.109, "step": 27555 }, { "epoch": 20.187545787545787, "grad_norm": 1.078125, "learning_rate": 4.206792966491951e-05, "loss": 0.1447, "step": 27556 }, { "epoch": 20.188278388278388, "grad_norm": 0.9765625, "learning_rate": 4.206455340172395e-05, "loss": 0.1248, "step": 27557 }, { "epoch": 20.18901098901099, "grad_norm": 1.015625, "learning_rate": 4.2061177178945366e-05, "loss": 0.1449, "step": 27558 }, { "epoch": 20.18974358974359, "grad_norm": 1.4375, "learning_rate": 4.2057800996599974e-05, "loss": 0.1368, "step": 27559 }, { "epoch": 20.19047619047619, "grad_norm": 0.8125, "learning_rate": 4.2054424854703936e-05, "loss": 0.1362, "step": 27560 }, { "epoch": 20.19120879120879, "grad_norm": 1.1484375, "learning_rate": 4.205104875327341e-05, "loss": 0.0949, "step": 27561 }, { "epoch": 20.19194139194139, "grad_norm": 0.765625, "learning_rate": 4.20476726923246e-05, "loss": 0.1246, "step": 27562 }, { "epoch": 20.192673992673992, "grad_norm": 1.6640625, "learning_rate": 4.2044296671873644e-05, "loss": 0.1151, "step": 27563 }, { "epoch": 20.193406593406593, "grad_norm": 1.3828125, "learning_rate": 4.204092069193672e-05, "loss": 0.149, "step": 27564 }, { "epoch": 20.194139194139193, "grad_norm": 0.96875, "learning_rate": 4.203754475253006e-05, "loss": 0.1166, "step": 27565 }, { "epoch": 20.194871794871794, "grad_norm": 1.0078125, "learning_rate": 4.2034168853669786e-05, "loss": 0.1348, "step": 27566 }, { "epoch": 20.195604395604395, "grad_norm": 1.140625, "learning_rate": 4.203079299537208e-05, "loss": 0.1382, "step": 27567 }, { "epoch": 20.196336996336996, "grad_norm": 1.09375, "learning_rate": 4.2027417177653136e-05, "loss": 0.1261, "step": 27568 }, { "epoch": 20.197069597069596, "grad_norm": 0.96484375, "learning_rate": 4.202404140052912e-05, "loss": 0.1395, "step": 27569 }, { "epoch": 20.197802197802197, "grad_norm": 0.97265625, "learning_rate": 4.202066566401619e-05, "loss": 0.1137, "step": 27570 }, { "epoch": 20.198534798534798, "grad_norm": 1.046875, "learning_rate": 4.201728996813055e-05, "loss": 0.1221, "step": 27571 }, { "epoch": 20.1992673992674, "grad_norm": 0.82421875, "learning_rate": 4.201391431288835e-05, "loss": 0.1188, "step": 27572 }, { "epoch": 20.2, "grad_norm": 1.15625, "learning_rate": 4.2010538698305784e-05, "loss": 0.1412, "step": 27573 }, { "epoch": 20.2007326007326, "grad_norm": 1.265625, "learning_rate": 4.2007163124398984e-05, "loss": 0.127, "step": 27574 }, { "epoch": 20.2014652014652, "grad_norm": 1.0, "learning_rate": 4.200378759118418e-05, "loss": 0.1037, "step": 27575 }, { "epoch": 20.2021978021978, "grad_norm": 0.57421875, "learning_rate": 4.2000412098677514e-05, "loss": 0.1255, "step": 27576 }, { "epoch": 20.202930402930402, "grad_norm": 1.4375, "learning_rate": 4.199703664689517e-05, "loss": 0.119, "step": 27577 }, { "epoch": 20.203663003663003, "grad_norm": 0.9765625, "learning_rate": 4.199366123585331e-05, "loss": 0.0994, "step": 27578 }, { "epoch": 20.204395604395604, "grad_norm": 1.4609375, "learning_rate": 4.199028586556809e-05, "loss": 0.1042, "step": 27579 }, { "epoch": 20.205128205128204, "grad_norm": 1.1015625, "learning_rate": 4.198691053605573e-05, "loss": 0.1117, "step": 27580 }, { "epoch": 20.205860805860805, "grad_norm": 0.9609375, "learning_rate": 4.198353524733238e-05, "loss": 0.1212, "step": 27581 }, { "epoch": 20.206593406593406, "grad_norm": 0.875, "learning_rate": 4.1980159999414184e-05, "loss": 0.1124, "step": 27582 }, { "epoch": 20.207326007326007, "grad_norm": 1.2421875, "learning_rate": 4.197678479231736e-05, "loss": 0.1398, "step": 27583 }, { "epoch": 20.208058608058607, "grad_norm": 1.1328125, "learning_rate": 4.1973409626058035e-05, "loss": 0.1058, "step": 27584 }, { "epoch": 20.208791208791208, "grad_norm": 1.9609375, "learning_rate": 4.1970034500652416e-05, "loss": 0.1625, "step": 27585 }, { "epoch": 20.20952380952381, "grad_norm": 1.828125, "learning_rate": 4.196665941611667e-05, "loss": 0.1278, "step": 27586 }, { "epoch": 20.21025641025641, "grad_norm": 1.1484375, "learning_rate": 4.196328437246696e-05, "loss": 0.1085, "step": 27587 }, { "epoch": 20.21098901098901, "grad_norm": 1.40625, "learning_rate": 4.1959909369719455e-05, "loss": 0.1115, "step": 27588 }, { "epoch": 20.21172161172161, "grad_norm": 1.6328125, "learning_rate": 4.195653440789033e-05, "loss": 0.1102, "step": 27589 }, { "epoch": 20.21245421245421, "grad_norm": 0.890625, "learning_rate": 4.195315948699573e-05, "loss": 0.1249, "step": 27590 }, { "epoch": 20.213186813186812, "grad_norm": 1.5546875, "learning_rate": 4.194978460705187e-05, "loss": 0.1226, "step": 27591 }, { "epoch": 20.213919413919413, "grad_norm": 1.140625, "learning_rate": 4.19464097680749e-05, "loss": 0.1357, "step": 27592 }, { "epoch": 20.214652014652014, "grad_norm": 0.97265625, "learning_rate": 4.194303497008098e-05, "loss": 0.109, "step": 27593 }, { "epoch": 20.215384615384615, "grad_norm": 0.7890625, "learning_rate": 4.1939660213086295e-05, "loss": 0.0946, "step": 27594 }, { "epoch": 20.216117216117215, "grad_norm": 0.8671875, "learning_rate": 4.1936285497107e-05, "loss": 0.0995, "step": 27595 }, { "epoch": 20.216849816849816, "grad_norm": 0.84375, "learning_rate": 4.193291082215926e-05, "loss": 0.1072, "step": 27596 }, { "epoch": 20.217582417582417, "grad_norm": 0.81640625, "learning_rate": 4.192953618825929e-05, "loss": 0.1221, "step": 27597 }, { "epoch": 20.218315018315018, "grad_norm": 1.5546875, "learning_rate": 4.192616159542321e-05, "loss": 0.1525, "step": 27598 }, { "epoch": 20.21904761904762, "grad_norm": 0.98828125, "learning_rate": 4.1922787043667214e-05, "loss": 0.1179, "step": 27599 }, { "epoch": 20.21978021978022, "grad_norm": 1.0390625, "learning_rate": 4.191941253300743e-05, "loss": 0.1488, "step": 27600 }, { "epoch": 20.22051282051282, "grad_norm": 1.0078125, "learning_rate": 4.191603806346009e-05, "loss": 0.1609, "step": 27601 }, { "epoch": 20.22124542124542, "grad_norm": 0.953125, "learning_rate": 4.1912663635041324e-05, "loss": 0.1359, "step": 27602 }, { "epoch": 20.22197802197802, "grad_norm": 0.6796875, "learning_rate": 4.19092892477673e-05, "loss": 0.1014, "step": 27603 }, { "epoch": 20.222710622710622, "grad_norm": 0.765625, "learning_rate": 4.19059149016542e-05, "loss": 0.1254, "step": 27604 }, { "epoch": 20.223443223443223, "grad_norm": 1.234375, "learning_rate": 4.1902540596718154e-05, "loss": 0.145, "step": 27605 }, { "epoch": 20.224175824175823, "grad_norm": 1.078125, "learning_rate": 4.189916633297538e-05, "loss": 0.1391, "step": 27606 }, { "epoch": 20.224908424908424, "grad_norm": 1.0546875, "learning_rate": 4.1895792110442015e-05, "loss": 0.123, "step": 27607 }, { "epoch": 20.225641025641025, "grad_norm": 1.0859375, "learning_rate": 4.189241792913425e-05, "loss": 0.1592, "step": 27608 }, { "epoch": 20.226373626373626, "grad_norm": 1.8515625, "learning_rate": 4.1889043789068236e-05, "loss": 0.1129, "step": 27609 }, { "epoch": 20.227106227106226, "grad_norm": 1.0234375, "learning_rate": 4.188566969026011e-05, "loss": 0.1307, "step": 27610 }, { "epoch": 20.227838827838827, "grad_norm": 1.40625, "learning_rate": 4.18822956327261e-05, "loss": 0.1232, "step": 27611 }, { "epoch": 20.228571428571428, "grad_norm": 0.90625, "learning_rate": 4.187892161648233e-05, "loss": 0.1094, "step": 27612 }, { "epoch": 20.22930402930403, "grad_norm": 1.3984375, "learning_rate": 4.187554764154498e-05, "loss": 0.1109, "step": 27613 }, { "epoch": 20.23003663003663, "grad_norm": 0.7578125, "learning_rate": 4.187217370793021e-05, "loss": 0.1143, "step": 27614 }, { "epoch": 20.23076923076923, "grad_norm": 1.0, "learning_rate": 4.1868799815654204e-05, "loss": 0.1426, "step": 27615 }, { "epoch": 20.23150183150183, "grad_norm": 0.9296875, "learning_rate": 4.186542596473308e-05, "loss": 0.1091, "step": 27616 }, { "epoch": 20.23223443223443, "grad_norm": 1.015625, "learning_rate": 4.186205215518305e-05, "loss": 0.1238, "step": 27617 }, { "epoch": 20.232967032967032, "grad_norm": 0.9609375, "learning_rate": 4.185867838702027e-05, "loss": 0.1014, "step": 27618 }, { "epoch": 20.233699633699633, "grad_norm": 0.88671875, "learning_rate": 4.1855304660260885e-05, "loss": 0.1381, "step": 27619 }, { "epoch": 20.234432234432234, "grad_norm": 0.88671875, "learning_rate": 4.1851930974921085e-05, "loss": 0.1229, "step": 27620 }, { "epoch": 20.235164835164834, "grad_norm": 1.1484375, "learning_rate": 4.184855733101701e-05, "loss": 0.1159, "step": 27621 }, { "epoch": 20.235897435897435, "grad_norm": 2.0625, "learning_rate": 4.184518372856484e-05, "loss": 0.1541, "step": 27622 }, { "epoch": 20.236630036630036, "grad_norm": 0.8515625, "learning_rate": 4.184181016758075e-05, "loss": 0.0999, "step": 27623 }, { "epoch": 20.237362637362637, "grad_norm": 1.5, "learning_rate": 4.183843664808089e-05, "loss": 0.1239, "step": 27624 }, { "epoch": 20.238095238095237, "grad_norm": 1.9140625, "learning_rate": 4.183506317008142e-05, "loss": 0.1238, "step": 27625 }, { "epoch": 20.238827838827838, "grad_norm": 0.8671875, "learning_rate": 4.183168973359848e-05, "loss": 0.0996, "step": 27626 }, { "epoch": 20.23956043956044, "grad_norm": 0.90625, "learning_rate": 4.182831633864829e-05, "loss": 0.1156, "step": 27627 }, { "epoch": 20.24029304029304, "grad_norm": 1.1640625, "learning_rate": 4.182494298524698e-05, "loss": 0.152, "step": 27628 }, { "epoch": 20.24102564102564, "grad_norm": 1.109375, "learning_rate": 4.182156967341072e-05, "loss": 0.1394, "step": 27629 }, { "epoch": 20.24175824175824, "grad_norm": 1.0, "learning_rate": 4.181819640315565e-05, "loss": 0.1159, "step": 27630 }, { "epoch": 20.24249084249084, "grad_norm": 1.421875, "learning_rate": 4.181482317449795e-05, "loss": 0.1466, "step": 27631 }, { "epoch": 20.243223443223442, "grad_norm": 0.90234375, "learning_rate": 4.1811449987453796e-05, "loss": 0.1123, "step": 27632 }, { "epoch": 20.243956043956043, "grad_norm": 0.8515625, "learning_rate": 4.180807684203933e-05, "loss": 0.1049, "step": 27633 }, { "epoch": 20.244688644688644, "grad_norm": 1.09375, "learning_rate": 4.180470373827074e-05, "loss": 0.141, "step": 27634 }, { "epoch": 20.245421245421245, "grad_norm": 1.203125, "learning_rate": 4.180133067616415e-05, "loss": 0.1312, "step": 27635 }, { "epoch": 20.246153846153845, "grad_norm": 0.91796875, "learning_rate": 4.179795765573573e-05, "loss": 0.112, "step": 27636 }, { "epoch": 20.246886446886446, "grad_norm": 1.0, "learning_rate": 4.179458467700167e-05, "loss": 0.119, "step": 27637 }, { "epoch": 20.247619047619047, "grad_norm": 1.0703125, "learning_rate": 4.179121173997812e-05, "loss": 0.1383, "step": 27638 }, { "epoch": 20.248351648351647, "grad_norm": 1.1171875, "learning_rate": 4.178783884468123e-05, "loss": 0.1025, "step": 27639 }, { "epoch": 20.249084249084248, "grad_norm": 0.9921875, "learning_rate": 4.178446599112716e-05, "loss": 0.1318, "step": 27640 }, { "epoch": 20.24981684981685, "grad_norm": 1.171875, "learning_rate": 4.1781093179332074e-05, "loss": 0.1325, "step": 27641 }, { "epoch": 20.25054945054945, "grad_norm": 1.3203125, "learning_rate": 4.177772040931211e-05, "loss": 0.1448, "step": 27642 }, { "epoch": 20.25128205128205, "grad_norm": 1.09375, "learning_rate": 4.177434768108348e-05, "loss": 0.0941, "step": 27643 }, { "epoch": 20.25201465201465, "grad_norm": 0.94921875, "learning_rate": 4.177097499466231e-05, "loss": 0.1129, "step": 27644 }, { "epoch": 20.252747252747252, "grad_norm": 1.3203125, "learning_rate": 4.176760235006475e-05, "loss": 0.119, "step": 27645 }, { "epoch": 20.253479853479853, "grad_norm": 1.1328125, "learning_rate": 4.1764229747306996e-05, "loss": 0.1684, "step": 27646 }, { "epoch": 20.254212454212453, "grad_norm": 1.3671875, "learning_rate": 4.176085718640517e-05, "loss": 0.0975, "step": 27647 }, { "epoch": 20.254945054945054, "grad_norm": 1.84375, "learning_rate": 4.175748466737544e-05, "loss": 0.1653, "step": 27648 }, { "epoch": 20.255677655677655, "grad_norm": 1.015625, "learning_rate": 4.1754112190233985e-05, "loss": 0.102, "step": 27649 }, { "epoch": 20.256410256410255, "grad_norm": 0.921875, "learning_rate": 4.1750739754996956e-05, "loss": 0.1033, "step": 27650 }, { "epoch": 20.257142857142856, "grad_norm": 1.0546875, "learning_rate": 4.1747367361680496e-05, "loss": 0.1124, "step": 27651 }, { "epoch": 20.257875457875457, "grad_norm": 0.58984375, "learning_rate": 4.174399501030075e-05, "loss": 0.1057, "step": 27652 }, { "epoch": 20.258608058608058, "grad_norm": 1.5078125, "learning_rate": 4.174062270087393e-05, "loss": 0.1519, "step": 27653 }, { "epoch": 20.25934065934066, "grad_norm": 1.3359375, "learning_rate": 4.1737250433416155e-05, "loss": 0.1087, "step": 27654 }, { "epoch": 20.26007326007326, "grad_norm": 0.9921875, "learning_rate": 4.173387820794359e-05, "loss": 0.1073, "step": 27655 }, { "epoch": 20.26080586080586, "grad_norm": 0.96875, "learning_rate": 4.173050602447239e-05, "loss": 0.1209, "step": 27656 }, { "epoch": 20.26153846153846, "grad_norm": 1.484375, "learning_rate": 4.1727133883018694e-05, "loss": 0.1227, "step": 27657 }, { "epoch": 20.26227106227106, "grad_norm": 0.8984375, "learning_rate": 4.1723761783598706e-05, "loss": 0.1207, "step": 27658 }, { "epoch": 20.263003663003662, "grad_norm": 1.0859375, "learning_rate": 4.1720389726228544e-05, "loss": 0.1292, "step": 27659 }, { "epoch": 20.263736263736263, "grad_norm": 1.0078125, "learning_rate": 4.171701771092438e-05, "loss": 0.1199, "step": 27660 }, { "epoch": 20.264468864468864, "grad_norm": 1.3828125, "learning_rate": 4.1713645737702376e-05, "loss": 0.1139, "step": 27661 }, { "epoch": 20.265201465201464, "grad_norm": 1.3828125, "learning_rate": 4.171027380657865e-05, "loss": 0.1382, "step": 27662 }, { "epoch": 20.265934065934065, "grad_norm": 1.046875, "learning_rate": 4.170690191756942e-05, "loss": 0.1225, "step": 27663 }, { "epoch": 20.266666666666666, "grad_norm": 1.1953125, "learning_rate": 4.17035300706908e-05, "loss": 0.1138, "step": 27664 }, { "epoch": 20.267399267399266, "grad_norm": 0.96875, "learning_rate": 4.170015826595896e-05, "loss": 0.1038, "step": 27665 }, { "epoch": 20.268131868131867, "grad_norm": 0.85546875, "learning_rate": 4.169678650339005e-05, "loss": 0.1138, "step": 27666 }, { "epoch": 20.268864468864468, "grad_norm": 1.125, "learning_rate": 4.1693414783000226e-05, "loss": 0.1217, "step": 27667 }, { "epoch": 20.26959706959707, "grad_norm": 1.1640625, "learning_rate": 4.169004310480563e-05, "loss": 0.1384, "step": 27668 }, { "epoch": 20.27032967032967, "grad_norm": 0.62109375, "learning_rate": 4.168667146882244e-05, "loss": 0.1045, "step": 27669 }, { "epoch": 20.27106227106227, "grad_norm": 1.03125, "learning_rate": 4.16832998750668e-05, "loss": 0.1209, "step": 27670 }, { "epoch": 20.27179487179487, "grad_norm": 1.4453125, "learning_rate": 4.167992832355485e-05, "loss": 0.1591, "step": 27671 }, { "epoch": 20.27252747252747, "grad_norm": 1.2734375, "learning_rate": 4.167655681430278e-05, "loss": 0.1207, "step": 27672 }, { "epoch": 20.273260073260072, "grad_norm": 1.4765625, "learning_rate": 4.16731853473267e-05, "loss": 0.1088, "step": 27673 }, { "epoch": 20.273992673992673, "grad_norm": 0.640625, "learning_rate": 4.16698139226428e-05, "loss": 0.1385, "step": 27674 }, { "epoch": 20.274725274725274, "grad_norm": 1.1953125, "learning_rate": 4.166644254026723e-05, "loss": 0.1289, "step": 27675 }, { "epoch": 20.275457875457874, "grad_norm": 0.84765625, "learning_rate": 4.1663071200216124e-05, "loss": 0.1088, "step": 27676 }, { "epoch": 20.276190476190475, "grad_norm": 1.5546875, "learning_rate": 4.165969990250565e-05, "loss": 0.0961, "step": 27677 }, { "epoch": 20.276923076923076, "grad_norm": 0.77734375, "learning_rate": 4.165632864715194e-05, "loss": 0.1153, "step": 27678 }, { "epoch": 20.277655677655677, "grad_norm": 1.109375, "learning_rate": 4.165295743417118e-05, "loss": 0.1038, "step": 27679 }, { "epoch": 20.278388278388277, "grad_norm": 1.0078125, "learning_rate": 4.16495862635795e-05, "loss": 0.1078, "step": 27680 }, { "epoch": 20.279120879120878, "grad_norm": 0.87109375, "learning_rate": 4.1646215135393055e-05, "loss": 0.1165, "step": 27681 }, { "epoch": 20.27985347985348, "grad_norm": 0.7421875, "learning_rate": 4.1642844049628e-05, "loss": 0.1032, "step": 27682 }, { "epoch": 20.28058608058608, "grad_norm": 0.97265625, "learning_rate": 4.163947300630049e-05, "loss": 0.1075, "step": 27683 }, { "epoch": 20.28131868131868, "grad_norm": 0.734375, "learning_rate": 4.163610200542667e-05, "loss": 0.1237, "step": 27684 }, { "epoch": 20.28205128205128, "grad_norm": 0.84765625, "learning_rate": 4.163273104702269e-05, "loss": 0.1113, "step": 27685 }, { "epoch": 20.282783882783882, "grad_norm": 0.8125, "learning_rate": 4.162936013110472e-05, "loss": 0.0991, "step": 27686 }, { "epoch": 20.283516483516483, "grad_norm": 1.2578125, "learning_rate": 4.16259892576889e-05, "loss": 0.1205, "step": 27687 }, { "epoch": 20.284249084249083, "grad_norm": 1.1796875, "learning_rate": 4.1622618426791374e-05, "loss": 0.1496, "step": 27688 }, { "epoch": 20.284981684981684, "grad_norm": 0.9609375, "learning_rate": 4.161924763842827e-05, "loss": 0.1031, "step": 27689 }, { "epoch": 20.285714285714285, "grad_norm": 1.0390625, "learning_rate": 4.16158768926158e-05, "loss": 0.1193, "step": 27690 }, { "epoch": 20.286446886446885, "grad_norm": 0.828125, "learning_rate": 4.161250618937007e-05, "loss": 0.0933, "step": 27691 }, { "epoch": 20.287179487179486, "grad_norm": 1.7265625, "learning_rate": 4.160913552870724e-05, "loss": 0.1477, "step": 27692 }, { "epoch": 20.287912087912087, "grad_norm": 1.25, "learning_rate": 4.160576491064347e-05, "loss": 0.1223, "step": 27693 }, { "epoch": 20.288644688644688, "grad_norm": 1.8984375, "learning_rate": 4.1602394335194875e-05, "loss": 0.1177, "step": 27694 }, { "epoch": 20.28937728937729, "grad_norm": 0.9453125, "learning_rate": 4.1599023802377644e-05, "loss": 0.0968, "step": 27695 }, { "epoch": 20.29010989010989, "grad_norm": 1.3203125, "learning_rate": 4.159565331220792e-05, "loss": 0.1018, "step": 27696 }, { "epoch": 20.29084249084249, "grad_norm": 1.1796875, "learning_rate": 4.159228286470183e-05, "loss": 0.1203, "step": 27697 }, { "epoch": 20.29157509157509, "grad_norm": 1.1171875, "learning_rate": 4.158891245987554e-05, "loss": 0.1274, "step": 27698 }, { "epoch": 20.29230769230769, "grad_norm": 1.078125, "learning_rate": 4.158554209774519e-05, "loss": 0.1129, "step": 27699 }, { "epoch": 20.293040293040292, "grad_norm": 1.3671875, "learning_rate": 4.158217177832695e-05, "loss": 0.1548, "step": 27700 }, { "epoch": 20.293772893772893, "grad_norm": 1.1640625, "learning_rate": 4.157880150163695e-05, "loss": 0.1423, "step": 27701 }, { "epoch": 20.294505494505493, "grad_norm": 0.734375, "learning_rate": 4.157543126769134e-05, "loss": 0.1368, "step": 27702 }, { "epoch": 20.295238095238094, "grad_norm": 1.1171875, "learning_rate": 4.1572061076506274e-05, "loss": 0.1172, "step": 27703 }, { "epoch": 20.295970695970695, "grad_norm": 1.2734375, "learning_rate": 4.156869092809786e-05, "loss": 0.1317, "step": 27704 }, { "epoch": 20.296703296703296, "grad_norm": 1.0, "learning_rate": 4.1565320822482315e-05, "loss": 0.1141, "step": 27705 }, { "epoch": 20.297435897435896, "grad_norm": 0.94921875, "learning_rate": 4.1561950759675745e-05, "loss": 0.1152, "step": 27706 }, { "epoch": 20.298168498168497, "grad_norm": 0.921875, "learning_rate": 4.155858073969431e-05, "loss": 0.1297, "step": 27707 }, { "epoch": 20.298901098901098, "grad_norm": 0.71484375, "learning_rate": 4.155521076255413e-05, "loss": 0.1077, "step": 27708 }, { "epoch": 20.2996336996337, "grad_norm": 0.95703125, "learning_rate": 4.155184082827137e-05, "loss": 0.1296, "step": 27709 }, { "epoch": 20.3003663003663, "grad_norm": 1.6484375, "learning_rate": 4.15484709368622e-05, "loss": 0.1169, "step": 27710 }, { "epoch": 20.3010989010989, "grad_norm": 1.4609375, "learning_rate": 4.154510108834272e-05, "loss": 0.1625, "step": 27711 }, { "epoch": 20.3018315018315, "grad_norm": 1.3828125, "learning_rate": 4.154173128272913e-05, "loss": 0.1394, "step": 27712 }, { "epoch": 20.3025641025641, "grad_norm": 1.0390625, "learning_rate": 4.153836152003753e-05, "loss": 0.0998, "step": 27713 }, { "epoch": 20.303296703296702, "grad_norm": 0.78125, "learning_rate": 4.153499180028409e-05, "loss": 0.0923, "step": 27714 }, { "epoch": 20.304029304029303, "grad_norm": 0.71484375, "learning_rate": 4.153162212348492e-05, "loss": 0.1127, "step": 27715 }, { "epoch": 20.304761904761904, "grad_norm": 0.828125, "learning_rate": 4.152825248965622e-05, "loss": 0.1358, "step": 27716 }, { "epoch": 20.305494505494504, "grad_norm": 0.98828125, "learning_rate": 4.15248828988141e-05, "loss": 0.0969, "step": 27717 }, { "epoch": 20.306227106227105, "grad_norm": 1.1484375, "learning_rate": 4.152151335097472e-05, "loss": 0.1416, "step": 27718 }, { "epoch": 20.306959706959706, "grad_norm": 0.95703125, "learning_rate": 4.151814384615421e-05, "loss": 0.1287, "step": 27719 }, { "epoch": 20.307692307692307, "grad_norm": 1.125, "learning_rate": 4.1514774384368705e-05, "loss": 0.0969, "step": 27720 }, { "epoch": 20.308424908424907, "grad_norm": 0.9140625, "learning_rate": 4.151140496563439e-05, "loss": 0.1018, "step": 27721 }, { "epoch": 20.309157509157508, "grad_norm": 0.58984375, "learning_rate": 4.150803558996738e-05, "loss": 0.115, "step": 27722 }, { "epoch": 20.30989010989011, "grad_norm": 0.8515625, "learning_rate": 4.1504666257383805e-05, "loss": 0.138, "step": 27723 }, { "epoch": 20.31062271062271, "grad_norm": 1.53125, "learning_rate": 4.1501296967899844e-05, "loss": 0.1215, "step": 27724 }, { "epoch": 20.31135531135531, "grad_norm": 1.8125, "learning_rate": 4.149792772153161e-05, "loss": 0.1475, "step": 27725 }, { "epoch": 20.31208791208791, "grad_norm": 1.2578125, "learning_rate": 4.149455851829527e-05, "loss": 0.1218, "step": 27726 }, { "epoch": 20.31282051282051, "grad_norm": 1.0546875, "learning_rate": 4.1491189358206964e-05, "loss": 0.1369, "step": 27727 }, { "epoch": 20.313553113553112, "grad_norm": 0.9453125, "learning_rate": 4.148782024128282e-05, "loss": 0.155, "step": 27728 }, { "epoch": 20.314285714285713, "grad_norm": 1.4453125, "learning_rate": 4.148445116753899e-05, "loss": 0.1433, "step": 27729 }, { "epoch": 20.315018315018314, "grad_norm": 1.953125, "learning_rate": 4.1481082136991596e-05, "loss": 0.1525, "step": 27730 }, { "epoch": 20.315750915750915, "grad_norm": 0.77734375, "learning_rate": 4.147771314965682e-05, "loss": 0.1245, "step": 27731 }, { "epoch": 20.316483516483515, "grad_norm": 1.0546875, "learning_rate": 4.147434420555078e-05, "loss": 0.108, "step": 27732 }, { "epoch": 20.317216117216116, "grad_norm": 1.1953125, "learning_rate": 4.147097530468961e-05, "loss": 0.145, "step": 27733 }, { "epoch": 20.317948717948717, "grad_norm": 0.77734375, "learning_rate": 4.146760644708946e-05, "loss": 0.1407, "step": 27734 }, { "epoch": 20.318681318681318, "grad_norm": 1.6328125, "learning_rate": 4.146423763276648e-05, "loss": 0.1043, "step": 27735 }, { "epoch": 20.31941391941392, "grad_norm": 1.0390625, "learning_rate": 4.146086886173681e-05, "loss": 0.1049, "step": 27736 }, { "epoch": 20.32014652014652, "grad_norm": 1.4453125, "learning_rate": 4.1457500134016564e-05, "loss": 0.1592, "step": 27737 }, { "epoch": 20.32087912087912, "grad_norm": 1.2109375, "learning_rate": 4.1454131449621926e-05, "loss": 0.1494, "step": 27738 }, { "epoch": 20.32161172161172, "grad_norm": 0.921875, "learning_rate": 4.145076280856901e-05, "loss": 0.1228, "step": 27739 }, { "epoch": 20.32234432234432, "grad_norm": 0.6640625, "learning_rate": 4.144739421087396e-05, "loss": 0.1254, "step": 27740 }, { "epoch": 20.323076923076922, "grad_norm": 0.84375, "learning_rate": 4.144402565655289e-05, "loss": 0.1322, "step": 27741 }, { "epoch": 20.323809523809523, "grad_norm": 1.7734375, "learning_rate": 4.144065714562199e-05, "loss": 0.1231, "step": 27742 }, { "epoch": 20.324542124542123, "grad_norm": 1.0703125, "learning_rate": 4.143728867809738e-05, "loss": 0.1172, "step": 27743 }, { "epoch": 20.325274725274724, "grad_norm": 1.03125, "learning_rate": 4.143392025399519e-05, "loss": 0.1406, "step": 27744 }, { "epoch": 20.326007326007325, "grad_norm": 1.3203125, "learning_rate": 4.143055187333156e-05, "loss": 0.0947, "step": 27745 }, { "epoch": 20.326739926739926, "grad_norm": 1.1484375, "learning_rate": 4.142718353612263e-05, "loss": 0.16, "step": 27746 }, { "epoch": 20.327472527472526, "grad_norm": 1.0234375, "learning_rate": 4.142381524238455e-05, "loss": 0.1275, "step": 27747 }, { "epoch": 20.328205128205127, "grad_norm": 1.03125, "learning_rate": 4.142044699213344e-05, "loss": 0.116, "step": 27748 }, { "epoch": 20.328937728937728, "grad_norm": 0.56640625, "learning_rate": 4.141707878538546e-05, "loss": 0.1022, "step": 27749 }, { "epoch": 20.32967032967033, "grad_norm": 1.2578125, "learning_rate": 4.141371062215674e-05, "loss": 0.1366, "step": 27750 }, { "epoch": 20.33040293040293, "grad_norm": 0.80078125, "learning_rate": 4.141034250246339e-05, "loss": 0.1248, "step": 27751 }, { "epoch": 20.33113553113553, "grad_norm": 1.0078125, "learning_rate": 4.14069744263216e-05, "loss": 0.126, "step": 27752 }, { "epoch": 20.33186813186813, "grad_norm": 1.28125, "learning_rate": 4.1403606393747474e-05, "loss": 0.1479, "step": 27753 }, { "epoch": 20.33260073260073, "grad_norm": 1.2734375, "learning_rate": 4.1400238404757155e-05, "loss": 0.106, "step": 27754 }, { "epoch": 20.333333333333332, "grad_norm": 1.2421875, "learning_rate": 4.1396870459366785e-05, "loss": 0.1173, "step": 27755 }, { "epoch": 20.334065934065933, "grad_norm": 0.74609375, "learning_rate": 4.139350255759248e-05, "loss": 0.1111, "step": 27756 }, { "epoch": 20.334798534798534, "grad_norm": 1.59375, "learning_rate": 4.1390134699450414e-05, "loss": 0.1414, "step": 27757 }, { "epoch": 20.335531135531134, "grad_norm": 0.88671875, "learning_rate": 4.1386766884956695e-05, "loss": 0.14, "step": 27758 }, { "epoch": 20.336263736263735, "grad_norm": 0.98828125, "learning_rate": 4.1383399114127476e-05, "loss": 0.1094, "step": 27759 }, { "epoch": 20.336996336996336, "grad_norm": 1.1796875, "learning_rate": 4.1380031386978866e-05, "loss": 0.1282, "step": 27760 }, { "epoch": 20.337728937728937, "grad_norm": 1.234375, "learning_rate": 4.137666370352703e-05, "loss": 0.1228, "step": 27761 }, { "epoch": 20.338461538461537, "grad_norm": 1.3359375, "learning_rate": 4.1373296063788094e-05, "loss": 0.1424, "step": 27762 }, { "epoch": 20.339194139194138, "grad_norm": 1.328125, "learning_rate": 4.136992846777818e-05, "loss": 0.1055, "step": 27763 }, { "epoch": 20.33992673992674, "grad_norm": 1.1953125, "learning_rate": 4.136656091551346e-05, "loss": 0.1125, "step": 27764 }, { "epoch": 20.34065934065934, "grad_norm": 1.234375, "learning_rate": 4.1363193407010036e-05, "loss": 0.0932, "step": 27765 }, { "epoch": 20.34139194139194, "grad_norm": 1.078125, "learning_rate": 4.135982594228405e-05, "loss": 0.1105, "step": 27766 }, { "epoch": 20.34212454212454, "grad_norm": 1.1328125, "learning_rate": 4.1356458521351626e-05, "loss": 0.1146, "step": 27767 }, { "epoch": 20.34285714285714, "grad_norm": 1.4609375, "learning_rate": 4.1353091144228926e-05, "loss": 0.133, "step": 27768 }, { "epoch": 20.343589743589742, "grad_norm": 1.2890625, "learning_rate": 4.134972381093207e-05, "loss": 0.1423, "step": 27769 }, { "epoch": 20.344322344322343, "grad_norm": 1.2109375, "learning_rate": 4.134635652147719e-05, "loss": 0.102, "step": 27770 }, { "epoch": 20.345054945054944, "grad_norm": 1.7109375, "learning_rate": 4.1342989275880424e-05, "loss": 0.1503, "step": 27771 }, { "epoch": 20.345787545787545, "grad_norm": 1.640625, "learning_rate": 4.133962207415788e-05, "loss": 0.1085, "step": 27772 }, { "epoch": 20.346520146520145, "grad_norm": 1.234375, "learning_rate": 4.133625491632573e-05, "loss": 0.1427, "step": 27773 }, { "epoch": 20.347252747252746, "grad_norm": 1.1796875, "learning_rate": 4.133288780240009e-05, "loss": 0.1303, "step": 27774 }, { "epoch": 20.347985347985347, "grad_norm": 1.0546875, "learning_rate": 4.132952073239709e-05, "loss": 0.0931, "step": 27775 }, { "epoch": 20.348717948717947, "grad_norm": 1.015625, "learning_rate": 4.132615370633288e-05, "loss": 0.1497, "step": 27776 }, { "epoch": 20.349450549450548, "grad_norm": 1.6953125, "learning_rate": 4.1322786724223556e-05, "loss": 0.1227, "step": 27777 }, { "epoch": 20.35018315018315, "grad_norm": 1.2421875, "learning_rate": 4.131941978608529e-05, "loss": 0.1063, "step": 27778 }, { "epoch": 20.35091575091575, "grad_norm": 1.8046875, "learning_rate": 4.13160528919342e-05, "loss": 0.1378, "step": 27779 }, { "epoch": 20.35164835164835, "grad_norm": 1.3828125, "learning_rate": 4.131268604178642e-05, "loss": 0.1265, "step": 27780 }, { "epoch": 20.35238095238095, "grad_norm": 1.59375, "learning_rate": 4.1309319235658064e-05, "loss": 0.1204, "step": 27781 }, { "epoch": 20.353113553113552, "grad_norm": 1.109375, "learning_rate": 4.130595247356527e-05, "loss": 0.1133, "step": 27782 }, { "epoch": 20.353846153846153, "grad_norm": 0.734375, "learning_rate": 4.130258575552419e-05, "loss": 0.1183, "step": 27783 }, { "epoch": 20.354578754578753, "grad_norm": 0.75390625, "learning_rate": 4.129921908155094e-05, "loss": 0.1543, "step": 27784 }, { "epoch": 20.355311355311354, "grad_norm": 1.0, "learning_rate": 4.129585245166165e-05, "loss": 0.1095, "step": 27785 }, { "epoch": 20.356043956043955, "grad_norm": 1.3203125, "learning_rate": 4.129248586587245e-05, "loss": 0.1254, "step": 27786 }, { "epoch": 20.356776556776556, "grad_norm": 1.1484375, "learning_rate": 4.1289119324199465e-05, "loss": 0.1303, "step": 27787 }, { "epoch": 20.357509157509156, "grad_norm": 0.94921875, "learning_rate": 4.1285752826658836e-05, "loss": 0.1207, "step": 27788 }, { "epoch": 20.358241758241757, "grad_norm": 0.9296875, "learning_rate": 4.12823863732667e-05, "loss": 0.1267, "step": 27789 }, { "epoch": 20.358974358974358, "grad_norm": 1.15625, "learning_rate": 4.1279019964039186e-05, "loss": 0.1321, "step": 27790 }, { "epoch": 20.35970695970696, "grad_norm": 1.2109375, "learning_rate": 4.1275653598992406e-05, "loss": 0.1314, "step": 27791 }, { "epoch": 20.36043956043956, "grad_norm": 1.1015625, "learning_rate": 4.12722872781425e-05, "loss": 0.1327, "step": 27792 }, { "epoch": 20.36117216117216, "grad_norm": 1.265625, "learning_rate": 4.126892100150556e-05, "loss": 0.1255, "step": 27793 }, { "epoch": 20.36190476190476, "grad_norm": 1.1328125, "learning_rate": 4.1265554769097786e-05, "loss": 0.1394, "step": 27794 }, { "epoch": 20.36263736263736, "grad_norm": 1.21875, "learning_rate": 4.126218858093528e-05, "loss": 0.1147, "step": 27795 }, { "epoch": 20.363369963369962, "grad_norm": 1.015625, "learning_rate": 4.1258822437034136e-05, "loss": 0.1155, "step": 27796 }, { "epoch": 20.364102564102563, "grad_norm": 1.296875, "learning_rate": 4.125545633741052e-05, "loss": 0.0986, "step": 27797 }, { "epoch": 20.364835164835164, "grad_norm": 0.97265625, "learning_rate": 4.125209028208052e-05, "loss": 0.1239, "step": 27798 }, { "epoch": 20.365567765567764, "grad_norm": 1.1484375, "learning_rate": 4.1248724271060324e-05, "loss": 0.1024, "step": 27799 }, { "epoch": 20.366300366300365, "grad_norm": 1.125, "learning_rate": 4.124535830436601e-05, "loss": 0.1063, "step": 27800 }, { "epoch": 20.367032967032966, "grad_norm": 1.3046875, "learning_rate": 4.124199238201374e-05, "loss": 0.1268, "step": 27801 }, { "epoch": 20.367765567765566, "grad_norm": 1.5234375, "learning_rate": 4.1238626504019616e-05, "loss": 0.1149, "step": 27802 }, { "epoch": 20.368498168498167, "grad_norm": 1.015625, "learning_rate": 4.1235260670399756e-05, "loss": 0.1085, "step": 27803 }, { "epoch": 20.369230769230768, "grad_norm": 1.5390625, "learning_rate": 4.123189488117032e-05, "loss": 0.1368, "step": 27804 }, { "epoch": 20.36996336996337, "grad_norm": 1.109375, "learning_rate": 4.122852913634742e-05, "loss": 0.1322, "step": 27805 }, { "epoch": 20.37069597069597, "grad_norm": 1.25, "learning_rate": 4.1225163435947175e-05, "loss": 0.1432, "step": 27806 }, { "epoch": 20.37142857142857, "grad_norm": 1.2421875, "learning_rate": 4.1221797779985726e-05, "loss": 0.1233, "step": 27807 }, { "epoch": 20.37216117216117, "grad_norm": 1.109375, "learning_rate": 4.121843216847916e-05, "loss": 0.1039, "step": 27808 }, { "epoch": 20.37289377289377, "grad_norm": 1.2265625, "learning_rate": 4.1215066601443664e-05, "loss": 0.1176, "step": 27809 }, { "epoch": 20.373626373626372, "grad_norm": 0.97265625, "learning_rate": 4.121170107889533e-05, "loss": 0.1328, "step": 27810 }, { "epoch": 20.374358974358973, "grad_norm": 0.83984375, "learning_rate": 4.120833560085028e-05, "loss": 0.0942, "step": 27811 }, { "epoch": 20.375091575091574, "grad_norm": 0.9609375, "learning_rate": 4.120497016732464e-05, "loss": 0.0919, "step": 27812 }, { "epoch": 20.375824175824174, "grad_norm": 1.59375, "learning_rate": 4.120160477833453e-05, "loss": 0.1246, "step": 27813 }, { "epoch": 20.376556776556775, "grad_norm": 0.79296875, "learning_rate": 4.1198239433896085e-05, "loss": 0.1073, "step": 27814 }, { "epoch": 20.377289377289376, "grad_norm": 0.89453125, "learning_rate": 4.1194874134025444e-05, "loss": 0.1348, "step": 27815 }, { "epoch": 20.378021978021977, "grad_norm": 1.515625, "learning_rate": 4.1191508878738714e-05, "loss": 0.1313, "step": 27816 }, { "epoch": 20.378754578754577, "grad_norm": 0.86328125, "learning_rate": 4.118814366805202e-05, "loss": 0.1185, "step": 27817 }, { "epoch": 20.379487179487178, "grad_norm": 0.73828125, "learning_rate": 4.118477850198149e-05, "loss": 0.1021, "step": 27818 }, { "epoch": 20.38021978021978, "grad_norm": 0.8125, "learning_rate": 4.118141338054322e-05, "loss": 0.138, "step": 27819 }, { "epoch": 20.38095238095238, "grad_norm": 2.1875, "learning_rate": 4.117804830375338e-05, "loss": 0.1373, "step": 27820 }, { "epoch": 20.38168498168498, "grad_norm": 1.5078125, "learning_rate": 4.1174683271628075e-05, "loss": 0.1239, "step": 27821 }, { "epoch": 20.38241758241758, "grad_norm": 0.7265625, "learning_rate": 4.1171318284183416e-05, "loss": 0.1099, "step": 27822 }, { "epoch": 20.383150183150182, "grad_norm": 1.8984375, "learning_rate": 4.1167953341435526e-05, "loss": 0.1601, "step": 27823 }, { "epoch": 20.383882783882783, "grad_norm": 1.4921875, "learning_rate": 4.1164588443400534e-05, "loss": 0.1268, "step": 27824 }, { "epoch": 20.384615384615383, "grad_norm": 0.765625, "learning_rate": 4.116122359009458e-05, "loss": 0.1197, "step": 27825 }, { "epoch": 20.385347985347984, "grad_norm": 0.8515625, "learning_rate": 4.1157858781533746e-05, "loss": 0.1011, "step": 27826 }, { "epoch": 20.386080586080585, "grad_norm": 1.140625, "learning_rate": 4.11544940177342e-05, "loss": 0.1291, "step": 27827 }, { "epoch": 20.386813186813185, "grad_norm": 0.8515625, "learning_rate": 4.1151129298712036e-05, "loss": 0.1172, "step": 27828 }, { "epoch": 20.387545787545786, "grad_norm": 0.953125, "learning_rate": 4.114776462448336e-05, "loss": 0.1046, "step": 27829 }, { "epoch": 20.388278388278387, "grad_norm": 0.94140625, "learning_rate": 4.114439999506434e-05, "loss": 0.1329, "step": 27830 }, { "epoch": 20.389010989010988, "grad_norm": 0.8828125, "learning_rate": 4.114103541047106e-05, "loss": 0.1044, "step": 27831 }, { "epoch": 20.38974358974359, "grad_norm": 0.98828125, "learning_rate": 4.113767087071966e-05, "loss": 0.1215, "step": 27832 }, { "epoch": 20.39047619047619, "grad_norm": 0.6171875, "learning_rate": 4.1134306375826255e-05, "loss": 0.1147, "step": 27833 }, { "epoch": 20.39120879120879, "grad_norm": 0.6796875, "learning_rate": 4.113094192580693e-05, "loss": 0.1314, "step": 27834 }, { "epoch": 20.39194139194139, "grad_norm": 1.15625, "learning_rate": 4.112757752067787e-05, "loss": 0.122, "step": 27835 }, { "epoch": 20.39267399267399, "grad_norm": 1.46875, "learning_rate": 4.1124213160455156e-05, "loss": 0.1195, "step": 27836 }, { "epoch": 20.393406593406592, "grad_norm": 1.0859375, "learning_rate": 4.1120848845154925e-05, "loss": 0.1358, "step": 27837 }, { "epoch": 20.394139194139193, "grad_norm": 0.8203125, "learning_rate": 4.111748457479326e-05, "loss": 0.1143, "step": 27838 }, { "epoch": 20.394871794871793, "grad_norm": 1.1875, "learning_rate": 4.111412034938632e-05, "loss": 0.1336, "step": 27839 }, { "epoch": 20.395604395604394, "grad_norm": 1.0546875, "learning_rate": 4.111075616895021e-05, "loss": 0.1133, "step": 27840 }, { "epoch": 20.396336996336995, "grad_norm": 0.87109375, "learning_rate": 4.1107392033501044e-05, "loss": 0.1203, "step": 27841 }, { "epoch": 20.397069597069596, "grad_norm": 1.5390625, "learning_rate": 4.110402794305496e-05, "loss": 0.1583, "step": 27842 }, { "epoch": 20.397802197802196, "grad_norm": 0.79296875, "learning_rate": 4.110066389762806e-05, "loss": 0.0902, "step": 27843 }, { "epoch": 20.398534798534797, "grad_norm": 0.953125, "learning_rate": 4.109729989723646e-05, "loss": 0.1205, "step": 27844 }, { "epoch": 20.399267399267398, "grad_norm": 0.70703125, "learning_rate": 4.109393594189627e-05, "loss": 0.121, "step": 27845 }, { "epoch": 20.4, "grad_norm": 1.3984375, "learning_rate": 4.109057203162364e-05, "loss": 0.1149, "step": 27846 }, { "epoch": 20.4007326007326, "grad_norm": 1.5859375, "learning_rate": 4.108720816643467e-05, "loss": 0.1059, "step": 27847 }, { "epoch": 20.4014652014652, "grad_norm": 1.34375, "learning_rate": 4.1083844346345465e-05, "loss": 0.1346, "step": 27848 }, { "epoch": 20.4021978021978, "grad_norm": 1.578125, "learning_rate": 4.108048057137215e-05, "loss": 0.1112, "step": 27849 }, { "epoch": 20.4029304029304, "grad_norm": 1.2890625, "learning_rate": 4.107711684153085e-05, "loss": 0.1195, "step": 27850 }, { "epoch": 20.403663003663002, "grad_norm": 0.79296875, "learning_rate": 4.107375315683768e-05, "loss": 0.1213, "step": 27851 }, { "epoch": 20.404395604395603, "grad_norm": 1.078125, "learning_rate": 4.1070389517308746e-05, "loss": 0.1085, "step": 27852 }, { "epoch": 20.405128205128204, "grad_norm": 1.671875, "learning_rate": 4.106702592296018e-05, "loss": 0.1397, "step": 27853 }, { "epoch": 20.405860805860804, "grad_norm": 0.98046875, "learning_rate": 4.1063662373808095e-05, "loss": 0.1332, "step": 27854 }, { "epoch": 20.406593406593405, "grad_norm": 0.90234375, "learning_rate": 4.106029886986857e-05, "loss": 0.1199, "step": 27855 }, { "epoch": 20.407326007326006, "grad_norm": 1.15625, "learning_rate": 4.1056935411157774e-05, "loss": 0.123, "step": 27856 }, { "epoch": 20.408058608058607, "grad_norm": 1.6171875, "learning_rate": 4.10535719976918e-05, "loss": 0.1207, "step": 27857 }, { "epoch": 20.408791208791207, "grad_norm": 1.2734375, "learning_rate": 4.105020862948677e-05, "loss": 0.1216, "step": 27858 }, { "epoch": 20.409523809523808, "grad_norm": 1.3671875, "learning_rate": 4.104684530655879e-05, "loss": 0.1004, "step": 27859 }, { "epoch": 20.41025641025641, "grad_norm": 1.3359375, "learning_rate": 4.104348202892396e-05, "loss": 0.1234, "step": 27860 }, { "epoch": 20.41098901098901, "grad_norm": 1.0703125, "learning_rate": 4.104011879659843e-05, "loss": 0.1425, "step": 27861 }, { "epoch": 20.41172161172161, "grad_norm": 0.9375, "learning_rate": 4.10367556095983e-05, "loss": 0.1283, "step": 27862 }, { "epoch": 20.41245421245421, "grad_norm": 1.28125, "learning_rate": 4.1033392467939656e-05, "loss": 0.0991, "step": 27863 }, { "epoch": 20.41318681318681, "grad_norm": 0.74609375, "learning_rate": 4.1030029371638665e-05, "loss": 0.0927, "step": 27864 }, { "epoch": 20.413919413919412, "grad_norm": 0.921875, "learning_rate": 4.1026666320711385e-05, "loss": 0.1133, "step": 27865 }, { "epoch": 20.414652014652013, "grad_norm": 1.171875, "learning_rate": 4.102330331517397e-05, "loss": 0.0935, "step": 27866 }, { "epoch": 20.415384615384614, "grad_norm": 0.7578125, "learning_rate": 4.1019940355042526e-05, "loss": 0.1093, "step": 27867 }, { "epoch": 20.416117216117215, "grad_norm": 1.3671875, "learning_rate": 4.101657744033316e-05, "loss": 0.1573, "step": 27868 }, { "epoch": 20.416849816849815, "grad_norm": 1.1953125, "learning_rate": 4.1013214571061984e-05, "loss": 0.1151, "step": 27869 }, { "epoch": 20.417582417582416, "grad_norm": 0.69140625, "learning_rate": 4.100985174724511e-05, "loss": 0.1108, "step": 27870 }, { "epoch": 20.418315018315017, "grad_norm": 2.0, "learning_rate": 4.1006488968898635e-05, "loss": 0.1116, "step": 27871 }, { "epoch": 20.419047619047618, "grad_norm": 0.8984375, "learning_rate": 4.1003126236038706e-05, "loss": 0.1306, "step": 27872 }, { "epoch": 20.41978021978022, "grad_norm": 0.84375, "learning_rate": 4.099976354868141e-05, "loss": 0.101, "step": 27873 }, { "epoch": 20.42051282051282, "grad_norm": 1.0234375, "learning_rate": 4.099640090684287e-05, "loss": 0.126, "step": 27874 }, { "epoch": 20.42124542124542, "grad_norm": 1.2265625, "learning_rate": 4.099303831053919e-05, "loss": 0.1313, "step": 27875 }, { "epoch": 20.42197802197802, "grad_norm": 1.1171875, "learning_rate": 4.098967575978648e-05, "loss": 0.1237, "step": 27876 }, { "epoch": 20.42271062271062, "grad_norm": 1.3671875, "learning_rate": 4.0986313254600865e-05, "loss": 0.1034, "step": 27877 }, { "epoch": 20.423443223443222, "grad_norm": 0.94921875, "learning_rate": 4.098295079499843e-05, "loss": 0.1402, "step": 27878 }, { "epoch": 20.424175824175823, "grad_norm": 0.8984375, "learning_rate": 4.097958838099532e-05, "loss": 0.1391, "step": 27879 }, { "epoch": 20.424908424908423, "grad_norm": 0.61328125, "learning_rate": 4.0976226012607625e-05, "loss": 0.1061, "step": 27880 }, { "epoch": 20.425641025641024, "grad_norm": 1.0078125, "learning_rate": 4.0972863689851436e-05, "loss": 0.1529, "step": 27881 }, { "epoch": 20.426373626373625, "grad_norm": 1.359375, "learning_rate": 4.096950141274291e-05, "loss": 0.141, "step": 27882 }, { "epoch": 20.427106227106226, "grad_norm": 1.0390625, "learning_rate": 4.096613918129813e-05, "loss": 0.1389, "step": 27883 }, { "epoch": 20.427838827838826, "grad_norm": 1.2890625, "learning_rate": 4.0962776995533206e-05, "loss": 0.1193, "step": 27884 }, { "epoch": 20.428571428571427, "grad_norm": 1.4609375, "learning_rate": 4.095941485546425e-05, "loss": 0.1134, "step": 27885 }, { "epoch": 20.429304029304028, "grad_norm": 0.94921875, "learning_rate": 4.0956052761107344e-05, "loss": 0.1365, "step": 27886 }, { "epoch": 20.43003663003663, "grad_norm": 1.09375, "learning_rate": 4.0952690712478644e-05, "loss": 0.1157, "step": 27887 }, { "epoch": 20.43076923076923, "grad_norm": 1.234375, "learning_rate": 4.0949328709594236e-05, "loss": 0.142, "step": 27888 }, { "epoch": 20.43150183150183, "grad_norm": 1.0859375, "learning_rate": 4.094596675247022e-05, "loss": 0.1112, "step": 27889 }, { "epoch": 20.43223443223443, "grad_norm": 1.3125, "learning_rate": 4.0942604841122725e-05, "loss": 0.1281, "step": 27890 }, { "epoch": 20.43296703296703, "grad_norm": 1.9609375, "learning_rate": 4.093924297556784e-05, "loss": 0.1297, "step": 27891 }, { "epoch": 20.433699633699632, "grad_norm": 1.203125, "learning_rate": 4.093588115582168e-05, "loss": 0.1273, "step": 27892 }, { "epoch": 20.434432234432233, "grad_norm": 1.7265625, "learning_rate": 4.0932519381900356e-05, "loss": 0.1119, "step": 27893 }, { "epoch": 20.435164835164834, "grad_norm": 1.6875, "learning_rate": 4.0929157653819985e-05, "loss": 0.1329, "step": 27894 }, { "epoch": 20.435897435897434, "grad_norm": 1.0078125, "learning_rate": 4.092579597159665e-05, "loss": 0.1398, "step": 27895 }, { "epoch": 20.436630036630035, "grad_norm": 1.1328125, "learning_rate": 4.092243433524647e-05, "loss": 0.1206, "step": 27896 }, { "epoch": 20.437362637362636, "grad_norm": 0.76171875, "learning_rate": 4.091907274478554e-05, "loss": 0.1386, "step": 27897 }, { "epoch": 20.438095238095237, "grad_norm": 0.5703125, "learning_rate": 4.091571120022999e-05, "loss": 0.1094, "step": 27898 }, { "epoch": 20.438827838827837, "grad_norm": 0.76953125, "learning_rate": 4.0912349701595925e-05, "loss": 0.1088, "step": 27899 }, { "epoch": 20.439560439560438, "grad_norm": 0.83203125, "learning_rate": 4.090898824889943e-05, "loss": 0.1024, "step": 27900 }, { "epoch": 20.44029304029304, "grad_norm": 0.93359375, "learning_rate": 4.0905626842156614e-05, "loss": 0.0958, "step": 27901 }, { "epoch": 20.44102564102564, "grad_norm": 0.80078125, "learning_rate": 4.090226548138359e-05, "loss": 0.1238, "step": 27902 }, { "epoch": 20.44175824175824, "grad_norm": 1.3515625, "learning_rate": 4.0898904166596466e-05, "loss": 0.1308, "step": 27903 }, { "epoch": 20.44249084249084, "grad_norm": 1.125, "learning_rate": 4.0895542897811336e-05, "loss": 0.1493, "step": 27904 }, { "epoch": 20.44322344322344, "grad_norm": 1.0703125, "learning_rate": 4.0892181675044326e-05, "loss": 0.1323, "step": 27905 }, { "epoch": 20.443956043956042, "grad_norm": 0.73828125, "learning_rate": 4.088882049831153e-05, "loss": 0.1119, "step": 27906 }, { "epoch": 20.444688644688643, "grad_norm": 1.09375, "learning_rate": 4.088545936762903e-05, "loss": 0.1096, "step": 27907 }, { "epoch": 20.445421245421244, "grad_norm": 1.296875, "learning_rate": 4.0882098283012966e-05, "loss": 0.1207, "step": 27908 }, { "epoch": 20.446153846153845, "grad_norm": 0.89453125, "learning_rate": 4.0878737244479425e-05, "loss": 0.1238, "step": 27909 }, { "epoch": 20.446886446886445, "grad_norm": 0.83984375, "learning_rate": 4.087537625204452e-05, "loss": 0.1347, "step": 27910 }, { "epoch": 20.447619047619046, "grad_norm": 1.1640625, "learning_rate": 4.0872015305724344e-05, "loss": 0.1626, "step": 27911 }, { "epoch": 20.448351648351647, "grad_norm": 0.6875, "learning_rate": 4.086865440553498e-05, "loss": 0.1285, "step": 27912 }, { "epoch": 20.449084249084247, "grad_norm": 1.5234375, "learning_rate": 4.086529355149258e-05, "loss": 0.094, "step": 27913 }, { "epoch": 20.449816849816848, "grad_norm": 0.91015625, "learning_rate": 4.086193274361321e-05, "loss": 0.1006, "step": 27914 }, { "epoch": 20.45054945054945, "grad_norm": 1.1328125, "learning_rate": 4.0858571981912986e-05, "loss": 0.1332, "step": 27915 }, { "epoch": 20.45128205128205, "grad_norm": 0.9921875, "learning_rate": 4.085521126640801e-05, "loss": 0.1394, "step": 27916 }, { "epoch": 20.45201465201465, "grad_norm": 1.25, "learning_rate": 4.0851850597114396e-05, "loss": 0.1315, "step": 27917 }, { "epoch": 20.45274725274725, "grad_norm": 1.1875, "learning_rate": 4.0848489974048204e-05, "loss": 0.1312, "step": 27918 }, { "epoch": 20.453479853479852, "grad_norm": 1.6640625, "learning_rate": 4.084512939722558e-05, "loss": 0.1601, "step": 27919 }, { "epoch": 20.454212454212453, "grad_norm": 0.9921875, "learning_rate": 4.084176886666262e-05, "loss": 0.0864, "step": 27920 }, { "epoch": 20.454945054945053, "grad_norm": 0.796875, "learning_rate": 4.0838408382375406e-05, "loss": 0.0998, "step": 27921 }, { "epoch": 20.455677655677654, "grad_norm": 1.4765625, "learning_rate": 4.083504794438006e-05, "loss": 0.1652, "step": 27922 }, { "epoch": 20.456410256410255, "grad_norm": 1.1484375, "learning_rate": 4.083168755269265e-05, "loss": 0.1253, "step": 27923 }, { "epoch": 20.457142857142856, "grad_norm": 0.77734375, "learning_rate": 4.0828327207329316e-05, "loss": 0.1065, "step": 27924 }, { "epoch": 20.457875457875456, "grad_norm": 0.80078125, "learning_rate": 4.082496690830614e-05, "loss": 0.1458, "step": 27925 }, { "epoch": 20.458608058608057, "grad_norm": 1.9140625, "learning_rate": 4.0821606655639224e-05, "loss": 0.1753, "step": 27926 }, { "epoch": 20.459340659340658, "grad_norm": 0.71875, "learning_rate": 4.0818246449344664e-05, "loss": 0.1158, "step": 27927 }, { "epoch": 20.46007326007326, "grad_norm": 1.34375, "learning_rate": 4.0814886289438554e-05, "loss": 0.0964, "step": 27928 }, { "epoch": 20.46080586080586, "grad_norm": 0.796875, "learning_rate": 4.081152617593701e-05, "loss": 0.0944, "step": 27929 }, { "epoch": 20.46153846153846, "grad_norm": 1.0625, "learning_rate": 4.0808166108856135e-05, "loss": 0.1513, "step": 27930 }, { "epoch": 20.46227106227106, "grad_norm": 0.90234375, "learning_rate": 4.080480608821201e-05, "loss": 0.1037, "step": 27931 }, { "epoch": 20.46300366300366, "grad_norm": 0.9609375, "learning_rate": 4.080144611402075e-05, "loss": 0.1362, "step": 27932 }, { "epoch": 20.463736263736262, "grad_norm": 1.109375, "learning_rate": 4.0798086186298425e-05, "loss": 0.131, "step": 27933 }, { "epoch": 20.464468864468863, "grad_norm": 1.0390625, "learning_rate": 4.079472630506117e-05, "loss": 0.1099, "step": 27934 }, { "epoch": 20.465201465201464, "grad_norm": 0.57421875, "learning_rate": 4.0791366470325063e-05, "loss": 0.1026, "step": 27935 }, { "epoch": 20.465934065934064, "grad_norm": 1.453125, "learning_rate": 4.078800668210622e-05, "loss": 0.1118, "step": 27936 }, { "epoch": 20.466666666666665, "grad_norm": 1.109375, "learning_rate": 4.078464694042071e-05, "loss": 0.1323, "step": 27937 }, { "epoch": 20.467399267399266, "grad_norm": 0.96875, "learning_rate": 4.078128724528464e-05, "loss": 0.1178, "step": 27938 }, { "epoch": 20.468131868131866, "grad_norm": 0.890625, "learning_rate": 4.077792759671412e-05, "loss": 0.1231, "step": 27939 }, { "epoch": 20.468864468864467, "grad_norm": 1.09375, "learning_rate": 4.077456799472525e-05, "loss": 0.1208, "step": 27940 }, { "epoch": 20.469597069597068, "grad_norm": 1.6484375, "learning_rate": 4.0771208439334105e-05, "loss": 0.1345, "step": 27941 }, { "epoch": 20.47032967032967, "grad_norm": 0.9296875, "learning_rate": 4.0767848930556804e-05, "loss": 0.1372, "step": 27942 }, { "epoch": 20.47106227106227, "grad_norm": 0.92578125, "learning_rate": 4.076448946840943e-05, "loss": 0.1343, "step": 27943 }, { "epoch": 20.47179487179487, "grad_norm": 1.6953125, "learning_rate": 4.0761130052908064e-05, "loss": 0.1345, "step": 27944 }, { "epoch": 20.47252747252747, "grad_norm": 1.265625, "learning_rate": 4.0757770684068846e-05, "loss": 0.1276, "step": 27945 }, { "epoch": 20.47326007326007, "grad_norm": 1.265625, "learning_rate": 4.0754411361907845e-05, "loss": 0.1145, "step": 27946 }, { "epoch": 20.473992673992672, "grad_norm": 0.8828125, "learning_rate": 4.075105208644116e-05, "loss": 0.1164, "step": 27947 }, { "epoch": 20.474725274725273, "grad_norm": 1.109375, "learning_rate": 4.0747692857684875e-05, "loss": 0.1276, "step": 27948 }, { "epoch": 20.475457875457874, "grad_norm": 1.5234375, "learning_rate": 4.074433367565508e-05, "loss": 0.1201, "step": 27949 }, { "epoch": 20.476190476190474, "grad_norm": 1.109375, "learning_rate": 4.074097454036792e-05, "loss": 0.1181, "step": 27950 }, { "epoch": 20.476923076923075, "grad_norm": 0.87890625, "learning_rate": 4.073761545183944e-05, "loss": 0.1532, "step": 27951 }, { "epoch": 20.477655677655676, "grad_norm": 1.0, "learning_rate": 4.073425641008575e-05, "loss": 0.1288, "step": 27952 }, { "epoch": 20.478388278388277, "grad_norm": 1.5234375, "learning_rate": 4.073089741512293e-05, "loss": 0.1328, "step": 27953 }, { "epoch": 20.479120879120877, "grad_norm": 1.3125, "learning_rate": 4.072753846696709e-05, "loss": 0.1382, "step": 27954 }, { "epoch": 20.479853479853478, "grad_norm": 1.375, "learning_rate": 4.0724179565634326e-05, "loss": 0.1046, "step": 27955 }, { "epoch": 20.48058608058608, "grad_norm": 1.1484375, "learning_rate": 4.072082071114073e-05, "loss": 0.1304, "step": 27956 }, { "epoch": 20.48131868131868, "grad_norm": 1.5625, "learning_rate": 4.07174619035024e-05, "loss": 0.1134, "step": 27957 }, { "epoch": 20.48205128205128, "grad_norm": 1.046875, "learning_rate": 4.0714103142735425e-05, "loss": 0.126, "step": 27958 }, { "epoch": 20.48278388278388, "grad_norm": 1.21875, "learning_rate": 4.0710744428855854e-05, "loss": 0.1285, "step": 27959 }, { "epoch": 20.483516483516482, "grad_norm": 1.0859375, "learning_rate": 4.070738576187986e-05, "loss": 0.1343, "step": 27960 }, { "epoch": 20.484249084249083, "grad_norm": 0.984375, "learning_rate": 4.0704027141823486e-05, "loss": 0.1195, "step": 27961 }, { "epoch": 20.484981684981683, "grad_norm": 1.171875, "learning_rate": 4.0700668568702834e-05, "loss": 0.1263, "step": 27962 }, { "epoch": 20.485714285714284, "grad_norm": 0.81640625, "learning_rate": 4.0697310042534005e-05, "loss": 0.1482, "step": 27963 }, { "epoch": 20.486446886446885, "grad_norm": 1.3984375, "learning_rate": 4.069395156333306e-05, "loss": 0.1276, "step": 27964 }, { "epoch": 20.487179487179485, "grad_norm": 0.96484375, "learning_rate": 4.069059313111612e-05, "loss": 0.1112, "step": 27965 }, { "epoch": 20.487912087912086, "grad_norm": 0.70703125, "learning_rate": 4.068723474589927e-05, "loss": 0.1078, "step": 27966 }, { "epoch": 20.488644688644687, "grad_norm": 0.8125, "learning_rate": 4.06838764076986e-05, "loss": 0.0904, "step": 27967 }, { "epoch": 20.489377289377288, "grad_norm": 1.2109375, "learning_rate": 4.0680518116530206e-05, "loss": 0.0962, "step": 27968 }, { "epoch": 20.49010989010989, "grad_norm": 0.8828125, "learning_rate": 4.067715987241017e-05, "loss": 0.1, "step": 27969 }, { "epoch": 20.49084249084249, "grad_norm": 1.09375, "learning_rate": 4.067380167535457e-05, "loss": 0.1136, "step": 27970 }, { "epoch": 20.49157509157509, "grad_norm": 1.390625, "learning_rate": 4.0670443525379525e-05, "loss": 0.1047, "step": 27971 }, { "epoch": 20.49230769230769, "grad_norm": 1.15625, "learning_rate": 4.066708542250112e-05, "loss": 0.1035, "step": 27972 }, { "epoch": 20.493040293040295, "grad_norm": 1.5078125, "learning_rate": 4.066372736673544e-05, "loss": 0.1257, "step": 27973 }, { "epoch": 20.493772893772892, "grad_norm": 1.0, "learning_rate": 4.0660369358098564e-05, "loss": 0.0951, "step": 27974 }, { "epoch": 20.494505494505496, "grad_norm": 0.78515625, "learning_rate": 4.0657011396606566e-05, "loss": 0.1247, "step": 27975 }, { "epoch": 20.495238095238093, "grad_norm": 0.7734375, "learning_rate": 4.065365348227558e-05, "loss": 0.1093, "step": 27976 }, { "epoch": 20.495970695970698, "grad_norm": 0.84375, "learning_rate": 4.065029561512168e-05, "loss": 0.1182, "step": 27977 }, { "epoch": 20.496703296703295, "grad_norm": 0.99609375, "learning_rate": 4.064693779516093e-05, "loss": 0.1064, "step": 27978 }, { "epoch": 20.4974358974359, "grad_norm": 0.94140625, "learning_rate": 4.0643580022409446e-05, "loss": 0.1338, "step": 27979 }, { "epoch": 20.498168498168496, "grad_norm": 1.109375, "learning_rate": 4.06402222968833e-05, "loss": 0.1755, "step": 27980 }, { "epoch": 20.4989010989011, "grad_norm": 1.578125, "learning_rate": 4.0636864618598575e-05, "loss": 0.1219, "step": 27981 }, { "epoch": 20.499633699633698, "grad_norm": 1.5625, "learning_rate": 4.063350698757139e-05, "loss": 0.1261, "step": 27982 }, { "epoch": 20.500366300366302, "grad_norm": 1.0, "learning_rate": 4.0630149403817804e-05, "loss": 0.1413, "step": 27983 }, { "epoch": 20.5010989010989, "grad_norm": 0.90625, "learning_rate": 4.062679186735392e-05, "loss": 0.1011, "step": 27984 }, { "epoch": 20.501831501831504, "grad_norm": 1.359375, "learning_rate": 4.062343437819579e-05, "loss": 0.1308, "step": 27985 }, { "epoch": 20.5025641025641, "grad_norm": 1.4375, "learning_rate": 4.062007693635955e-05, "loss": 0.122, "step": 27986 }, { "epoch": 20.503296703296705, "grad_norm": 1.8203125, "learning_rate": 4.061671954186127e-05, "loss": 0.1401, "step": 27987 }, { "epoch": 20.504029304029302, "grad_norm": 0.625, "learning_rate": 4.061336219471703e-05, "loss": 0.114, "step": 27988 }, { "epoch": 20.504761904761907, "grad_norm": 1.09375, "learning_rate": 4.0610004894942916e-05, "loss": 0.1133, "step": 27989 }, { "epoch": 20.505494505494504, "grad_norm": 0.74609375, "learning_rate": 4.060664764255499e-05, "loss": 0.1051, "step": 27990 }, { "epoch": 20.506227106227108, "grad_norm": 0.984375, "learning_rate": 4.060329043756939e-05, "loss": 0.1034, "step": 27991 }, { "epoch": 20.506959706959705, "grad_norm": 0.87109375, "learning_rate": 4.059993328000217e-05, "loss": 0.1307, "step": 27992 }, { "epoch": 20.50769230769231, "grad_norm": 1.0078125, "learning_rate": 4.059657616986941e-05, "loss": 0.0947, "step": 27993 }, { "epoch": 20.508424908424907, "grad_norm": 1.1015625, "learning_rate": 4.0593219107187223e-05, "loss": 0.0904, "step": 27994 }, { "epoch": 20.50915750915751, "grad_norm": 1.2265625, "learning_rate": 4.058986209197167e-05, "loss": 0.1414, "step": 27995 }, { "epoch": 20.509890109890108, "grad_norm": 1.296875, "learning_rate": 4.058650512423882e-05, "loss": 0.1312, "step": 27996 }, { "epoch": 20.510622710622712, "grad_norm": 1.65625, "learning_rate": 4.0583148204004794e-05, "loss": 0.1098, "step": 27997 }, { "epoch": 20.51135531135531, "grad_norm": 1.6484375, "learning_rate": 4.0579791331285674e-05, "loss": 0.1357, "step": 27998 }, { "epoch": 20.512087912087914, "grad_norm": 0.90625, "learning_rate": 4.057643450609752e-05, "loss": 0.1256, "step": 27999 }, { "epoch": 20.51282051282051, "grad_norm": 0.94921875, "learning_rate": 4.057307772845642e-05, "loss": 0.1218, "step": 28000 }, { "epoch": 20.513553113553115, "grad_norm": 1.03125, "learning_rate": 4.056972099837844e-05, "loss": 0.1109, "step": 28001 }, { "epoch": 20.514285714285712, "grad_norm": 1.0546875, "learning_rate": 4.056636431587973e-05, "loss": 0.1068, "step": 28002 }, { "epoch": 20.515018315018317, "grad_norm": 1.1875, "learning_rate": 4.056300768097631e-05, "loss": 0.1169, "step": 28003 }, { "epoch": 20.515750915750917, "grad_norm": 0.69921875, "learning_rate": 4.055965109368428e-05, "loss": 0.1317, "step": 28004 }, { "epoch": 20.516483516483518, "grad_norm": 1.0625, "learning_rate": 4.0556294554019734e-05, "loss": 0.1204, "step": 28005 }, { "epoch": 20.51721611721612, "grad_norm": 1.3046875, "learning_rate": 4.055293806199874e-05, "loss": 0.0877, "step": 28006 }, { "epoch": 20.51794871794872, "grad_norm": 0.71875, "learning_rate": 4.054958161763738e-05, "loss": 0.0997, "step": 28007 }, { "epoch": 20.51868131868132, "grad_norm": 0.97265625, "learning_rate": 4.054622522095175e-05, "loss": 0.1138, "step": 28008 }, { "epoch": 20.51941391941392, "grad_norm": 1.53125, "learning_rate": 4.0542868871957915e-05, "loss": 0.1176, "step": 28009 }, { "epoch": 20.520146520146522, "grad_norm": 0.62109375, "learning_rate": 4.0539512570671975e-05, "loss": 0.1064, "step": 28010 }, { "epoch": 20.520879120879123, "grad_norm": 1.0546875, "learning_rate": 4.053615631710998e-05, "loss": 0.1142, "step": 28011 }, { "epoch": 20.521611721611723, "grad_norm": 1.2578125, "learning_rate": 4.0532800111288046e-05, "loss": 0.1599, "step": 28012 }, { "epoch": 20.522344322344324, "grad_norm": 1.140625, "learning_rate": 4.0529443953222245e-05, "loss": 0.1397, "step": 28013 }, { "epoch": 20.523076923076925, "grad_norm": 1.3125, "learning_rate": 4.052608784292865e-05, "loss": 0.1543, "step": 28014 }, { "epoch": 20.523809523809526, "grad_norm": 0.63671875, "learning_rate": 4.052273178042334e-05, "loss": 0.0913, "step": 28015 }, { "epoch": 20.524542124542126, "grad_norm": 0.97265625, "learning_rate": 4.0519375765722374e-05, "loss": 0.1233, "step": 28016 }, { "epoch": 20.525274725274727, "grad_norm": 0.8828125, "learning_rate": 4.051601979884188e-05, "loss": 0.0937, "step": 28017 }, { "epoch": 20.526007326007328, "grad_norm": 0.796875, "learning_rate": 4.0512663879797904e-05, "loss": 0.1253, "step": 28018 }, { "epoch": 20.52673992673993, "grad_norm": 1.46875, "learning_rate": 4.0509308008606536e-05, "loss": 0.13, "step": 28019 }, { "epoch": 20.52747252747253, "grad_norm": 1.0625, "learning_rate": 4.050595218528385e-05, "loss": 0.1399, "step": 28020 }, { "epoch": 20.52820512820513, "grad_norm": 0.69921875, "learning_rate": 4.050259640984594e-05, "loss": 0.0999, "step": 28021 }, { "epoch": 20.52893772893773, "grad_norm": 0.6875, "learning_rate": 4.049924068230885e-05, "loss": 0.1159, "step": 28022 }, { "epoch": 20.52967032967033, "grad_norm": 1.4765625, "learning_rate": 4.049588500268869e-05, "loss": 0.1415, "step": 28023 }, { "epoch": 20.530402930402932, "grad_norm": 1.6640625, "learning_rate": 4.049252937100155e-05, "loss": 0.1551, "step": 28024 }, { "epoch": 20.531135531135533, "grad_norm": 1.7578125, "learning_rate": 4.0489173787263475e-05, "loss": 0.1697, "step": 28025 }, { "epoch": 20.531868131868134, "grad_norm": 0.921875, "learning_rate": 4.048581825149055e-05, "loss": 0.1165, "step": 28026 }, { "epoch": 20.532600732600734, "grad_norm": 0.84375, "learning_rate": 4.0482462763698844e-05, "loss": 0.1347, "step": 28027 }, { "epoch": 20.533333333333335, "grad_norm": 0.73828125, "learning_rate": 4.0479107323904467e-05, "loss": 0.1057, "step": 28028 }, { "epoch": 20.534065934065936, "grad_norm": 0.640625, "learning_rate": 4.0475751932123474e-05, "loss": 0.1043, "step": 28029 }, { "epoch": 20.534798534798536, "grad_norm": 1.0390625, "learning_rate": 4.047239658837193e-05, "loss": 0.1211, "step": 28030 }, { "epoch": 20.535531135531137, "grad_norm": 1.203125, "learning_rate": 4.0469041292665943e-05, "loss": 0.1179, "step": 28031 }, { "epoch": 20.536263736263738, "grad_norm": 1.6328125, "learning_rate": 4.046568604502156e-05, "loss": 0.1466, "step": 28032 }, { "epoch": 20.53699633699634, "grad_norm": 1.015625, "learning_rate": 4.046233084545487e-05, "loss": 0.1355, "step": 28033 }, { "epoch": 20.53772893772894, "grad_norm": 0.93359375, "learning_rate": 4.045897569398196e-05, "loss": 0.1135, "step": 28034 }, { "epoch": 20.53846153846154, "grad_norm": 1.140625, "learning_rate": 4.045562059061889e-05, "loss": 0.1323, "step": 28035 }, { "epoch": 20.53919413919414, "grad_norm": 0.94921875, "learning_rate": 4.045226553538174e-05, "loss": 0.1248, "step": 28036 }, { "epoch": 20.53992673992674, "grad_norm": 1.5703125, "learning_rate": 4.044891052828657e-05, "loss": 0.1433, "step": 28037 }, { "epoch": 20.540659340659342, "grad_norm": 1.140625, "learning_rate": 4.044555556934948e-05, "loss": 0.1342, "step": 28038 }, { "epoch": 20.541391941391943, "grad_norm": 0.83203125, "learning_rate": 4.044220065858654e-05, "loss": 0.116, "step": 28039 }, { "epoch": 20.542124542124544, "grad_norm": 1.0078125, "learning_rate": 4.0438845796013815e-05, "loss": 0.1082, "step": 28040 }, { "epoch": 20.542857142857144, "grad_norm": 1.0, "learning_rate": 4.043549098164738e-05, "loss": 0.1024, "step": 28041 }, { "epoch": 20.543589743589745, "grad_norm": 0.70703125, "learning_rate": 4.0432136215503295e-05, "loss": 0.1128, "step": 28042 }, { "epoch": 20.544322344322346, "grad_norm": 1.1875, "learning_rate": 4.0428781497597666e-05, "loss": 0.1167, "step": 28043 }, { "epoch": 20.545054945054947, "grad_norm": 0.73046875, "learning_rate": 4.0425426827946556e-05, "loss": 0.1221, "step": 28044 }, { "epoch": 20.545787545787547, "grad_norm": 0.6796875, "learning_rate": 4.042207220656603e-05, "loss": 0.1266, "step": 28045 }, { "epoch": 20.546520146520148, "grad_norm": 1.328125, "learning_rate": 4.041871763347217e-05, "loss": 0.1426, "step": 28046 }, { "epoch": 20.54725274725275, "grad_norm": 0.67578125, "learning_rate": 4.0415363108681034e-05, "loss": 0.1189, "step": 28047 }, { "epoch": 20.54798534798535, "grad_norm": 0.91015625, "learning_rate": 4.041200863220869e-05, "loss": 0.1107, "step": 28048 }, { "epoch": 20.54871794871795, "grad_norm": 0.90625, "learning_rate": 4.040865420407125e-05, "loss": 0.135, "step": 28049 }, { "epoch": 20.54945054945055, "grad_norm": 0.6796875, "learning_rate": 4.0405299824284754e-05, "loss": 0.1036, "step": 28050 }, { "epoch": 20.550183150183152, "grad_norm": 0.83984375, "learning_rate": 4.0401945492865284e-05, "loss": 0.1278, "step": 28051 }, { "epoch": 20.550915750915753, "grad_norm": 1.15625, "learning_rate": 4.0398591209828904e-05, "loss": 0.1275, "step": 28052 }, { "epoch": 20.551648351648353, "grad_norm": 1.25, "learning_rate": 4.0395236975191676e-05, "loss": 0.1472, "step": 28053 }, { "epoch": 20.552380952380954, "grad_norm": 1.1484375, "learning_rate": 4.0391882788969705e-05, "loss": 0.119, "step": 28054 }, { "epoch": 20.553113553113555, "grad_norm": 1.171875, "learning_rate": 4.038852865117903e-05, "loss": 0.1362, "step": 28055 }, { "epoch": 20.553846153846155, "grad_norm": 0.72265625, "learning_rate": 4.038517456183573e-05, "loss": 0.1201, "step": 28056 }, { "epoch": 20.554578754578756, "grad_norm": 1.171875, "learning_rate": 4.0381820520955886e-05, "loss": 0.1437, "step": 28057 }, { "epoch": 20.555311355311357, "grad_norm": 0.81640625, "learning_rate": 4.037846652855556e-05, "loss": 0.1348, "step": 28058 }, { "epoch": 20.556043956043958, "grad_norm": 0.9140625, "learning_rate": 4.037511258465082e-05, "loss": 0.0941, "step": 28059 }, { "epoch": 20.55677655677656, "grad_norm": 1.078125, "learning_rate": 4.037175868925775e-05, "loss": 0.1075, "step": 28060 }, { "epoch": 20.55750915750916, "grad_norm": 0.8046875, "learning_rate": 4.036840484239241e-05, "loss": 0.1029, "step": 28061 }, { "epoch": 20.55824175824176, "grad_norm": 1.3515625, "learning_rate": 4.036505104407087e-05, "loss": 0.1186, "step": 28062 }, { "epoch": 20.55897435897436, "grad_norm": 0.75, "learning_rate": 4.036169729430917e-05, "loss": 0.1073, "step": 28063 }, { "epoch": 20.55970695970696, "grad_norm": 1.2421875, "learning_rate": 4.0358343593123424e-05, "loss": 0.1068, "step": 28064 }, { "epoch": 20.560439560439562, "grad_norm": 1.5078125, "learning_rate": 4.035498994052969e-05, "loss": 0.1284, "step": 28065 }, { "epoch": 20.561172161172163, "grad_norm": 1.1796875, "learning_rate": 4.035163633654403e-05, "loss": 0.1353, "step": 28066 }, { "epoch": 20.561904761904763, "grad_norm": 0.66015625, "learning_rate": 4.034828278118251e-05, "loss": 0.1092, "step": 28067 }, { "epoch": 20.562637362637364, "grad_norm": 0.78515625, "learning_rate": 4.034492927446118e-05, "loss": 0.1494, "step": 28068 }, { "epoch": 20.563369963369965, "grad_norm": 0.98046875, "learning_rate": 4.034157581639615e-05, "loss": 0.0936, "step": 28069 }, { "epoch": 20.564102564102566, "grad_norm": 0.984375, "learning_rate": 4.033822240700345e-05, "loss": 0.1091, "step": 28070 }, { "epoch": 20.564835164835166, "grad_norm": 1.296875, "learning_rate": 4.0334869046299174e-05, "loss": 0.1316, "step": 28071 }, { "epoch": 20.565567765567767, "grad_norm": 1.4296875, "learning_rate": 4.033151573429939e-05, "loss": 0.1449, "step": 28072 }, { "epoch": 20.566300366300368, "grad_norm": 1.6328125, "learning_rate": 4.0328162471020135e-05, "loss": 0.1033, "step": 28073 }, { "epoch": 20.56703296703297, "grad_norm": 1.3671875, "learning_rate": 4.032480925647748e-05, "loss": 0.1114, "step": 28074 }, { "epoch": 20.56776556776557, "grad_norm": 0.73046875, "learning_rate": 4.032145609068752e-05, "loss": 0.1142, "step": 28075 }, { "epoch": 20.56849816849817, "grad_norm": 0.9609375, "learning_rate": 4.0318102973666306e-05, "loss": 0.1331, "step": 28076 }, { "epoch": 20.56923076923077, "grad_norm": 1.2265625, "learning_rate": 4.03147499054299e-05, "loss": 0.1153, "step": 28077 }, { "epoch": 20.56996336996337, "grad_norm": 0.94921875, "learning_rate": 4.031139688599439e-05, "loss": 0.1509, "step": 28078 }, { "epoch": 20.570695970695972, "grad_norm": 1.3828125, "learning_rate": 4.030804391537578e-05, "loss": 0.1353, "step": 28079 }, { "epoch": 20.571428571428573, "grad_norm": 1.046875, "learning_rate": 4.030469099359021e-05, "loss": 0.1517, "step": 28080 }, { "epoch": 20.572161172161174, "grad_norm": 1.140625, "learning_rate": 4.03013381206537e-05, "loss": 0.1359, "step": 28081 }, { "epoch": 20.572893772893774, "grad_norm": 0.8125, "learning_rate": 4.029798529658233e-05, "loss": 0.1346, "step": 28082 }, { "epoch": 20.573626373626375, "grad_norm": 1.5703125, "learning_rate": 4.029463252139216e-05, "loss": 0.1394, "step": 28083 }, { "epoch": 20.574358974358976, "grad_norm": 1.0625, "learning_rate": 4.0291279795099264e-05, "loss": 0.1006, "step": 28084 }, { "epoch": 20.575091575091577, "grad_norm": 0.96875, "learning_rate": 4.0287927117719675e-05, "loss": 0.1339, "step": 28085 }, { "epoch": 20.575824175824177, "grad_norm": 0.83984375, "learning_rate": 4.0284574489269506e-05, "loss": 0.1222, "step": 28086 }, { "epoch": 20.576556776556778, "grad_norm": 0.79296875, "learning_rate": 4.028122190976479e-05, "loss": 0.1137, "step": 28087 }, { "epoch": 20.57728937728938, "grad_norm": 0.7265625, "learning_rate": 4.0277869379221595e-05, "loss": 0.1193, "step": 28088 }, { "epoch": 20.57802197802198, "grad_norm": 0.83203125, "learning_rate": 4.027451689765596e-05, "loss": 0.1191, "step": 28089 }, { "epoch": 20.57875457875458, "grad_norm": 0.859375, "learning_rate": 4.027116446508399e-05, "loss": 0.1361, "step": 28090 }, { "epoch": 20.57948717948718, "grad_norm": 0.79296875, "learning_rate": 4.026781208152173e-05, "loss": 0.1204, "step": 28091 }, { "epoch": 20.58021978021978, "grad_norm": 0.8125, "learning_rate": 4.0264459746985246e-05, "loss": 0.1208, "step": 28092 }, { "epoch": 20.580952380952382, "grad_norm": 1.0078125, "learning_rate": 4.0261107461490586e-05, "loss": 0.1269, "step": 28093 }, { "epoch": 20.581684981684983, "grad_norm": 1.1015625, "learning_rate": 4.025775522505382e-05, "loss": 0.156, "step": 28094 }, { "epoch": 20.582417582417584, "grad_norm": 0.66796875, "learning_rate": 4.025440303769102e-05, "loss": 0.1003, "step": 28095 }, { "epoch": 20.583150183150185, "grad_norm": 1.0234375, "learning_rate": 4.025105089941822e-05, "loss": 0.1287, "step": 28096 }, { "epoch": 20.583882783882785, "grad_norm": 0.76171875, "learning_rate": 4.024769881025152e-05, "loss": 0.1258, "step": 28097 }, { "epoch": 20.584615384615386, "grad_norm": 0.97265625, "learning_rate": 4.024434677020696e-05, "loss": 0.1191, "step": 28098 }, { "epoch": 20.585347985347987, "grad_norm": 0.90625, "learning_rate": 4.02409947793006e-05, "loss": 0.108, "step": 28099 }, { "epoch": 20.586080586080588, "grad_norm": 1.015625, "learning_rate": 4.023764283754848e-05, "loss": 0.0944, "step": 28100 }, { "epoch": 20.58681318681319, "grad_norm": 1.046875, "learning_rate": 4.02342909449667e-05, "loss": 0.0918, "step": 28101 }, { "epoch": 20.58754578754579, "grad_norm": 1.0625, "learning_rate": 4.023093910157131e-05, "loss": 0.1053, "step": 28102 }, { "epoch": 20.58827838827839, "grad_norm": 1.140625, "learning_rate": 4.0227587307378355e-05, "loss": 0.1139, "step": 28103 }, { "epoch": 20.58901098901099, "grad_norm": 0.76171875, "learning_rate": 4.0224235562403906e-05, "loss": 0.1248, "step": 28104 }, { "epoch": 20.58974358974359, "grad_norm": 1.40625, "learning_rate": 4.022088386666399e-05, "loss": 0.1379, "step": 28105 }, { "epoch": 20.590476190476192, "grad_norm": 0.9609375, "learning_rate": 4.021753222017472e-05, "loss": 0.1132, "step": 28106 }, { "epoch": 20.591208791208793, "grad_norm": 0.703125, "learning_rate": 4.021418062295213e-05, "loss": 0.1199, "step": 28107 }, { "epoch": 20.591941391941393, "grad_norm": 1.0390625, "learning_rate": 4.0210829075012266e-05, "loss": 0.1369, "step": 28108 }, { "epoch": 20.592673992673994, "grad_norm": 0.80078125, "learning_rate": 4.020747757637121e-05, "loss": 0.1323, "step": 28109 }, { "epoch": 20.593406593406595, "grad_norm": 0.79296875, "learning_rate": 4.0204126127044976e-05, "loss": 0.1037, "step": 28110 }, { "epoch": 20.594139194139196, "grad_norm": 0.98046875, "learning_rate": 4.020077472704969e-05, "loss": 0.1434, "step": 28111 }, { "epoch": 20.594871794871796, "grad_norm": 1.765625, "learning_rate": 4.019742337640136e-05, "loss": 0.1371, "step": 28112 }, { "epoch": 20.595604395604397, "grad_norm": 1.140625, "learning_rate": 4.0194072075116066e-05, "loss": 0.1245, "step": 28113 }, { "epoch": 20.596336996336998, "grad_norm": 0.921875, "learning_rate": 4.019072082320986e-05, "loss": 0.1106, "step": 28114 }, { "epoch": 20.5970695970696, "grad_norm": 0.9296875, "learning_rate": 4.018736962069875e-05, "loss": 0.1018, "step": 28115 }, { "epoch": 20.5978021978022, "grad_norm": 1.078125, "learning_rate": 4.0184018467598876e-05, "loss": 0.1187, "step": 28116 }, { "epoch": 20.5985347985348, "grad_norm": 0.93359375, "learning_rate": 4.018066736392626e-05, "loss": 0.1321, "step": 28117 }, { "epoch": 20.5992673992674, "grad_norm": 1.3515625, "learning_rate": 4.017731630969695e-05, "loss": 0.0986, "step": 28118 }, { "epoch": 20.6, "grad_norm": 1.125, "learning_rate": 4.017396530492699e-05, "loss": 0.1256, "step": 28119 }, { "epoch": 20.600732600732602, "grad_norm": 1.0703125, "learning_rate": 4.017061434963247e-05, "loss": 0.1112, "step": 28120 }, { "epoch": 20.601465201465203, "grad_norm": 1.421875, "learning_rate": 4.016726344382941e-05, "loss": 0.1388, "step": 28121 }, { "epoch": 20.602197802197804, "grad_norm": 1.046875, "learning_rate": 4.016391258753389e-05, "loss": 0.1096, "step": 28122 }, { "epoch": 20.602930402930404, "grad_norm": 0.796875, "learning_rate": 4.016056178076196e-05, "loss": 0.1237, "step": 28123 }, { "epoch": 20.603663003663005, "grad_norm": 0.78125, "learning_rate": 4.015721102352969e-05, "loss": 0.106, "step": 28124 }, { "epoch": 20.604395604395606, "grad_norm": 0.8515625, "learning_rate": 4.0153860315853096e-05, "loss": 0.1192, "step": 28125 }, { "epoch": 20.605128205128207, "grad_norm": 0.87109375, "learning_rate": 4.0150509657748256e-05, "loss": 0.1128, "step": 28126 }, { "epoch": 20.605860805860807, "grad_norm": 1.234375, "learning_rate": 4.014715904923122e-05, "loss": 0.1078, "step": 28127 }, { "epoch": 20.606593406593408, "grad_norm": 0.98046875, "learning_rate": 4.0143808490318054e-05, "loss": 0.1112, "step": 28128 }, { "epoch": 20.60732600732601, "grad_norm": 0.8203125, "learning_rate": 4.01404579810248e-05, "loss": 0.1027, "step": 28129 }, { "epoch": 20.60805860805861, "grad_norm": 0.87109375, "learning_rate": 4.0137107521367513e-05, "loss": 0.1349, "step": 28130 }, { "epoch": 20.60879120879121, "grad_norm": 0.83203125, "learning_rate": 4.013375711136223e-05, "loss": 0.1031, "step": 28131 }, { "epoch": 20.60952380952381, "grad_norm": 1.34375, "learning_rate": 4.013040675102503e-05, "loss": 0.1286, "step": 28132 }, { "epoch": 20.61025641025641, "grad_norm": 1.1015625, "learning_rate": 4.012705644037196e-05, "loss": 0.1145, "step": 28133 }, { "epoch": 20.610989010989012, "grad_norm": 1.0390625, "learning_rate": 4.0123706179419054e-05, "loss": 0.1289, "step": 28134 }, { "epoch": 20.611721611721613, "grad_norm": 1.2265625, "learning_rate": 4.0120355968182396e-05, "loss": 0.1296, "step": 28135 }, { "epoch": 20.612454212454214, "grad_norm": 0.70703125, "learning_rate": 4.0117005806677996e-05, "loss": 0.1259, "step": 28136 }, { "epoch": 20.613186813186815, "grad_norm": 1.5390625, "learning_rate": 4.011365569492195e-05, "loss": 0.1252, "step": 28137 }, { "epoch": 20.613919413919415, "grad_norm": 0.5390625, "learning_rate": 4.011030563293028e-05, "loss": 0.0957, "step": 28138 }, { "epoch": 20.614652014652016, "grad_norm": 0.6953125, "learning_rate": 4.0106955620719063e-05, "loss": 0.1106, "step": 28139 }, { "epoch": 20.615384615384617, "grad_norm": 1.90625, "learning_rate": 4.010360565830433e-05, "loss": 0.1114, "step": 28140 }, { "epoch": 20.616117216117217, "grad_norm": 1.1328125, "learning_rate": 4.01002557457021e-05, "loss": 0.1245, "step": 28141 }, { "epoch": 20.616849816849818, "grad_norm": 1.4765625, "learning_rate": 4.0096905882928495e-05, "loss": 0.1235, "step": 28142 }, { "epoch": 20.61758241758242, "grad_norm": 1.96875, "learning_rate": 4.009355606999952e-05, "loss": 0.1151, "step": 28143 }, { "epoch": 20.61831501831502, "grad_norm": 1.3515625, "learning_rate": 4.009020630693124e-05, "loss": 0.114, "step": 28144 }, { "epoch": 20.61904761904762, "grad_norm": 1.1171875, "learning_rate": 4.008685659373969e-05, "loss": 0.0961, "step": 28145 }, { "epoch": 20.61978021978022, "grad_norm": 0.89453125, "learning_rate": 4.008350693044093e-05, "loss": 0.1324, "step": 28146 }, { "epoch": 20.620512820512822, "grad_norm": 1.1015625, "learning_rate": 4.0080157317051004e-05, "loss": 0.1103, "step": 28147 }, { "epoch": 20.621245421245423, "grad_norm": 1.0703125, "learning_rate": 4.007680775358596e-05, "loss": 0.0939, "step": 28148 }, { "epoch": 20.621978021978023, "grad_norm": 1.2265625, "learning_rate": 4.007345824006187e-05, "loss": 0.1284, "step": 28149 }, { "epoch": 20.622710622710624, "grad_norm": 0.80078125, "learning_rate": 4.007010877649476e-05, "loss": 0.1453, "step": 28150 }, { "epoch": 20.623443223443225, "grad_norm": 0.75390625, "learning_rate": 4.006675936290068e-05, "loss": 0.1099, "step": 28151 }, { "epoch": 20.624175824175826, "grad_norm": 1.2734375, "learning_rate": 4.006340999929566e-05, "loss": 0.1126, "step": 28152 }, { "epoch": 20.624908424908426, "grad_norm": 0.796875, "learning_rate": 4.00600606856958e-05, "loss": 0.1039, "step": 28153 }, { "epoch": 20.625641025641027, "grad_norm": 0.9296875, "learning_rate": 4.005671142211711e-05, "loss": 0.111, "step": 28154 }, { "epoch": 20.626373626373628, "grad_norm": 1.0859375, "learning_rate": 4.005336220857564e-05, "loss": 0.1084, "step": 28155 }, { "epoch": 20.62710622710623, "grad_norm": 0.92578125, "learning_rate": 4.005001304508744e-05, "loss": 0.133, "step": 28156 }, { "epoch": 20.62783882783883, "grad_norm": 1.140625, "learning_rate": 4.004666393166855e-05, "loss": 0.1298, "step": 28157 }, { "epoch": 20.62857142857143, "grad_norm": 1.0625, "learning_rate": 4.004331486833503e-05, "loss": 0.1081, "step": 28158 }, { "epoch": 20.62930402930403, "grad_norm": 1.046875, "learning_rate": 4.0039965855102915e-05, "loss": 0.1367, "step": 28159 }, { "epoch": 20.63003663003663, "grad_norm": 1.0703125, "learning_rate": 4.003661689198827e-05, "loss": 0.1135, "step": 28160 }, { "epoch": 20.630769230769232, "grad_norm": 0.875, "learning_rate": 4.0033267979007136e-05, "loss": 0.0781, "step": 28161 }, { "epoch": 20.631501831501833, "grad_norm": 0.875, "learning_rate": 4.0029919116175515e-05, "loss": 0.1231, "step": 28162 }, { "epoch": 20.632234432234434, "grad_norm": 0.66015625, "learning_rate": 4.002657030350952e-05, "loss": 0.1179, "step": 28163 }, { "epoch": 20.632967032967034, "grad_norm": 1.890625, "learning_rate": 4.002322154102517e-05, "loss": 0.1631, "step": 28164 }, { "epoch": 20.633699633699635, "grad_norm": 0.78515625, "learning_rate": 4.0019872828738495e-05, "loss": 0.1073, "step": 28165 }, { "epoch": 20.634432234432236, "grad_norm": 1.1328125, "learning_rate": 4.001652416666555e-05, "loss": 0.134, "step": 28166 }, { "epoch": 20.635164835164836, "grad_norm": 1.796875, "learning_rate": 4.001317555482236e-05, "loss": 0.1387, "step": 28167 }, { "epoch": 20.635897435897437, "grad_norm": 1.4765625, "learning_rate": 4.0009826993225e-05, "loss": 0.1393, "step": 28168 }, { "epoch": 20.636630036630038, "grad_norm": 1.1796875, "learning_rate": 4.0006478481889516e-05, "loss": 0.1186, "step": 28169 }, { "epoch": 20.63736263736264, "grad_norm": 1.09375, "learning_rate": 4.000313002083194e-05, "loss": 0.1326, "step": 28170 }, { "epoch": 20.63809523809524, "grad_norm": 0.87109375, "learning_rate": 3.99997816100683e-05, "loss": 0.1205, "step": 28171 }, { "epoch": 20.63882783882784, "grad_norm": 0.8203125, "learning_rate": 3.999643324961466e-05, "loss": 0.1103, "step": 28172 }, { "epoch": 20.63956043956044, "grad_norm": 1.296875, "learning_rate": 3.9993084939487045e-05, "loss": 0.1223, "step": 28173 }, { "epoch": 20.64029304029304, "grad_norm": 1.046875, "learning_rate": 3.9989736679701513e-05, "loss": 0.1213, "step": 28174 }, { "epoch": 20.641025641025642, "grad_norm": 1.1484375, "learning_rate": 3.998638847027411e-05, "loss": 0.111, "step": 28175 }, { "epoch": 20.641758241758243, "grad_norm": 1.3046875, "learning_rate": 3.998304031122087e-05, "loss": 0.105, "step": 28176 }, { "epoch": 20.642490842490844, "grad_norm": 1.21875, "learning_rate": 3.9979692202557844e-05, "loss": 0.1184, "step": 28177 }, { "epoch": 20.643223443223444, "grad_norm": 1.0703125, "learning_rate": 3.997634414430104e-05, "loss": 0.1196, "step": 28178 }, { "epoch": 20.643956043956045, "grad_norm": 1.515625, "learning_rate": 3.997299613646655e-05, "loss": 0.1517, "step": 28179 }, { "epoch": 20.644688644688646, "grad_norm": 0.9765625, "learning_rate": 3.996964817907039e-05, "loss": 0.1235, "step": 28180 }, { "epoch": 20.645421245421247, "grad_norm": 0.68359375, "learning_rate": 3.99663002721286e-05, "loss": 0.112, "step": 28181 }, { "epoch": 20.646153846153847, "grad_norm": 1.296875, "learning_rate": 3.9962952415657226e-05, "loss": 0.121, "step": 28182 }, { "epoch": 20.646886446886448, "grad_norm": 1.171875, "learning_rate": 3.9959604609672286e-05, "loss": 0.12, "step": 28183 }, { "epoch": 20.64761904761905, "grad_norm": 0.98828125, "learning_rate": 3.995625685418986e-05, "loss": 0.1164, "step": 28184 }, { "epoch": 20.64835164835165, "grad_norm": 0.6328125, "learning_rate": 3.995290914922596e-05, "loss": 0.1276, "step": 28185 }, { "epoch": 20.64908424908425, "grad_norm": 0.87890625, "learning_rate": 3.994956149479664e-05, "loss": 0.0963, "step": 28186 }, { "epoch": 20.64981684981685, "grad_norm": 1.3359375, "learning_rate": 3.994621389091794e-05, "loss": 0.1463, "step": 28187 }, { "epoch": 20.650549450549452, "grad_norm": 0.84765625, "learning_rate": 3.994286633760587e-05, "loss": 0.0936, "step": 28188 }, { "epoch": 20.651282051282053, "grad_norm": 0.93359375, "learning_rate": 3.993951883487651e-05, "loss": 0.1472, "step": 28189 }, { "epoch": 20.652014652014653, "grad_norm": 1.0078125, "learning_rate": 3.993617138274589e-05, "loss": 0.1319, "step": 28190 }, { "epoch": 20.652747252747254, "grad_norm": 0.85546875, "learning_rate": 3.993282398123004e-05, "loss": 0.1088, "step": 28191 }, { "epoch": 20.653479853479855, "grad_norm": 1.4609375, "learning_rate": 3.992947663034499e-05, "loss": 0.1143, "step": 28192 }, { "epoch": 20.654212454212455, "grad_norm": 1.265625, "learning_rate": 3.9926129330106774e-05, "loss": 0.1419, "step": 28193 }, { "epoch": 20.654945054945056, "grad_norm": 0.9375, "learning_rate": 3.992278208053147e-05, "loss": 0.1201, "step": 28194 }, { "epoch": 20.655677655677657, "grad_norm": 0.8046875, "learning_rate": 3.991943488163508e-05, "loss": 0.1222, "step": 28195 }, { "epoch": 20.656410256410258, "grad_norm": 0.8828125, "learning_rate": 3.991608773343365e-05, "loss": 0.1401, "step": 28196 }, { "epoch": 20.65714285714286, "grad_norm": 1.1875, "learning_rate": 3.991274063594321e-05, "loss": 0.1242, "step": 28197 }, { "epoch": 20.65787545787546, "grad_norm": 0.953125, "learning_rate": 3.990939358917982e-05, "loss": 0.1051, "step": 28198 }, { "epoch": 20.65860805860806, "grad_norm": 1.0, "learning_rate": 3.990604659315949e-05, "loss": 0.1183, "step": 28199 }, { "epoch": 20.65934065934066, "grad_norm": 1.5390625, "learning_rate": 3.990269964789827e-05, "loss": 0.1211, "step": 28200 }, { "epoch": 20.66007326007326, "grad_norm": 1.234375, "learning_rate": 3.989935275341219e-05, "loss": 0.1373, "step": 28201 }, { "epoch": 20.660805860805862, "grad_norm": 1.5234375, "learning_rate": 3.989600590971731e-05, "loss": 0.1234, "step": 28202 }, { "epoch": 20.661538461538463, "grad_norm": 1.1015625, "learning_rate": 3.989265911682965e-05, "loss": 0.0914, "step": 28203 }, { "epoch": 20.662271062271063, "grad_norm": 0.875, "learning_rate": 3.98893123747652e-05, "loss": 0.1114, "step": 28204 }, { "epoch": 20.663003663003664, "grad_norm": 0.64453125, "learning_rate": 3.988596568354007e-05, "loss": 0.0984, "step": 28205 }, { "epoch": 20.663736263736265, "grad_norm": 0.8515625, "learning_rate": 3.988261904317027e-05, "loss": 0.1281, "step": 28206 }, { "epoch": 20.664468864468866, "grad_norm": 1.0, "learning_rate": 3.9879272453671815e-05, "loss": 0.1315, "step": 28207 }, { "epoch": 20.665201465201466, "grad_norm": 0.953125, "learning_rate": 3.987592591506075e-05, "loss": 0.1188, "step": 28208 }, { "epoch": 20.665934065934067, "grad_norm": 0.8828125, "learning_rate": 3.987257942735311e-05, "loss": 0.1241, "step": 28209 }, { "epoch": 20.666666666666668, "grad_norm": 0.5703125, "learning_rate": 3.986923299056494e-05, "loss": 0.1152, "step": 28210 }, { "epoch": 20.66739926739927, "grad_norm": 1.390625, "learning_rate": 3.986588660471225e-05, "loss": 0.1234, "step": 28211 }, { "epoch": 20.66813186813187, "grad_norm": 0.828125, "learning_rate": 3.9862540269811106e-05, "loss": 0.0946, "step": 28212 }, { "epoch": 20.66886446886447, "grad_norm": 1.1796875, "learning_rate": 3.985919398587752e-05, "loss": 0.1343, "step": 28213 }, { "epoch": 20.66959706959707, "grad_norm": 0.86328125, "learning_rate": 3.9855847752927506e-05, "loss": 0.1248, "step": 28214 }, { "epoch": 20.67032967032967, "grad_norm": 0.92578125, "learning_rate": 3.9852501570977145e-05, "loss": 0.1327, "step": 28215 }, { "epoch": 20.671062271062272, "grad_norm": 1.0078125, "learning_rate": 3.984915544004244e-05, "loss": 0.1157, "step": 28216 }, { "epoch": 20.671794871794873, "grad_norm": 0.921875, "learning_rate": 3.984580936013944e-05, "loss": 0.1267, "step": 28217 }, { "epoch": 20.672527472527474, "grad_norm": 1.046875, "learning_rate": 3.984246333128415e-05, "loss": 0.1226, "step": 28218 }, { "epoch": 20.673260073260074, "grad_norm": 1.15625, "learning_rate": 3.9839117353492595e-05, "loss": 0.1406, "step": 28219 }, { "epoch": 20.673992673992675, "grad_norm": 1.0546875, "learning_rate": 3.9835771426780854e-05, "loss": 0.1314, "step": 28220 }, { "epoch": 20.674725274725276, "grad_norm": 1.125, "learning_rate": 3.983242555116493e-05, "loss": 0.1138, "step": 28221 }, { "epoch": 20.675457875457877, "grad_norm": 1.4609375, "learning_rate": 3.9829079726660875e-05, "loss": 0.1273, "step": 28222 }, { "epoch": 20.676190476190477, "grad_norm": 0.98828125, "learning_rate": 3.982573395328467e-05, "loss": 0.1407, "step": 28223 }, { "epoch": 20.676923076923078, "grad_norm": 1.0703125, "learning_rate": 3.9822388231052394e-05, "loss": 0.1011, "step": 28224 }, { "epoch": 20.67765567765568, "grad_norm": 1.140625, "learning_rate": 3.981904255998005e-05, "loss": 0.1278, "step": 28225 }, { "epoch": 20.67838827838828, "grad_norm": 1.15625, "learning_rate": 3.9815696940083685e-05, "loss": 0.1465, "step": 28226 }, { "epoch": 20.67912087912088, "grad_norm": 1.078125, "learning_rate": 3.981235137137933e-05, "loss": 0.1091, "step": 28227 }, { "epoch": 20.67985347985348, "grad_norm": 1.2265625, "learning_rate": 3.9809005853883016e-05, "loss": 0.1725, "step": 28228 }, { "epoch": 20.68058608058608, "grad_norm": 1.6015625, "learning_rate": 3.980566038761076e-05, "loss": 0.126, "step": 28229 }, { "epoch": 20.681318681318682, "grad_norm": 1.4453125, "learning_rate": 3.9802314972578566e-05, "loss": 0.1326, "step": 28230 }, { "epoch": 20.682051282051283, "grad_norm": 1.265625, "learning_rate": 3.979896960880252e-05, "loss": 0.1186, "step": 28231 }, { "epoch": 20.682783882783884, "grad_norm": 0.83203125, "learning_rate": 3.979562429629863e-05, "loss": 0.1224, "step": 28232 }, { "epoch": 20.683516483516485, "grad_norm": 1.203125, "learning_rate": 3.979227903508291e-05, "loss": 0.1182, "step": 28233 }, { "epoch": 20.684249084249085, "grad_norm": 1.6328125, "learning_rate": 3.9788933825171384e-05, "loss": 0.1324, "step": 28234 }, { "epoch": 20.684981684981686, "grad_norm": 0.91796875, "learning_rate": 3.97855886665801e-05, "loss": 0.1202, "step": 28235 }, { "epoch": 20.685714285714287, "grad_norm": 0.8046875, "learning_rate": 3.9782243559325085e-05, "loss": 0.1017, "step": 28236 }, { "epoch": 20.686446886446888, "grad_norm": 1.34375, "learning_rate": 3.9778898503422344e-05, "loss": 0.1561, "step": 28237 }, { "epoch": 20.68717948717949, "grad_norm": 1.7578125, "learning_rate": 3.9775553498887936e-05, "loss": 0.1455, "step": 28238 }, { "epoch": 20.68791208791209, "grad_norm": 0.91015625, "learning_rate": 3.977220854573787e-05, "loss": 0.1496, "step": 28239 }, { "epoch": 20.68864468864469, "grad_norm": 1.296875, "learning_rate": 3.9768863643988155e-05, "loss": 0.1279, "step": 28240 }, { "epoch": 20.68937728937729, "grad_norm": 1.3046875, "learning_rate": 3.9765518793654855e-05, "loss": 0.1103, "step": 28241 }, { "epoch": 20.69010989010989, "grad_norm": 0.734375, "learning_rate": 3.976217399475398e-05, "loss": 0.1178, "step": 28242 }, { "epoch": 20.690842490842492, "grad_norm": 1.1328125, "learning_rate": 3.975882924730156e-05, "loss": 0.1532, "step": 28243 }, { "epoch": 20.691575091575093, "grad_norm": 1.1328125, "learning_rate": 3.9755484551313605e-05, "loss": 0.1217, "step": 28244 }, { "epoch": 20.692307692307693, "grad_norm": 1.125, "learning_rate": 3.975213990680613e-05, "loss": 0.1141, "step": 28245 }, { "epoch": 20.693040293040294, "grad_norm": 0.8203125, "learning_rate": 3.9748795313795215e-05, "loss": 0.1195, "step": 28246 }, { "epoch": 20.693772893772895, "grad_norm": 1.1953125, "learning_rate": 3.9745450772296836e-05, "loss": 0.1053, "step": 28247 }, { "epoch": 20.694505494505496, "grad_norm": 1.265625, "learning_rate": 3.974210628232705e-05, "loss": 0.1369, "step": 28248 }, { "epoch": 20.695238095238096, "grad_norm": 0.90625, "learning_rate": 3.973876184390184e-05, "loss": 0.1385, "step": 28249 }, { "epoch": 20.695970695970697, "grad_norm": 0.80078125, "learning_rate": 3.9735417457037266e-05, "loss": 0.0995, "step": 28250 }, { "epoch": 20.696703296703298, "grad_norm": 1.34375, "learning_rate": 3.973207312174932e-05, "loss": 0.1588, "step": 28251 }, { "epoch": 20.6974358974359, "grad_norm": 0.82421875, "learning_rate": 3.9728728838054074e-05, "loss": 0.116, "step": 28252 }, { "epoch": 20.6981684981685, "grad_norm": 1.390625, "learning_rate": 3.972538460596752e-05, "loss": 0.1278, "step": 28253 }, { "epoch": 20.6989010989011, "grad_norm": 1.015625, "learning_rate": 3.9722040425505686e-05, "loss": 0.1271, "step": 28254 }, { "epoch": 20.6996336996337, "grad_norm": 0.9296875, "learning_rate": 3.971869629668459e-05, "loss": 0.13, "step": 28255 }, { "epoch": 20.7003663003663, "grad_norm": 0.7578125, "learning_rate": 3.971535221952025e-05, "loss": 0.0924, "step": 28256 }, { "epoch": 20.701098901098902, "grad_norm": 1.0078125, "learning_rate": 3.971200819402871e-05, "loss": 0.1238, "step": 28257 }, { "epoch": 20.701831501831503, "grad_norm": 1.09375, "learning_rate": 3.9708664220225974e-05, "loss": 0.1288, "step": 28258 }, { "epoch": 20.702564102564104, "grad_norm": 0.91796875, "learning_rate": 3.970532029812809e-05, "loss": 0.106, "step": 28259 }, { "epoch": 20.703296703296704, "grad_norm": 1.2734375, "learning_rate": 3.970197642775103e-05, "loss": 0.1347, "step": 28260 }, { "epoch": 20.704029304029305, "grad_norm": 1.921875, "learning_rate": 3.969863260911085e-05, "loss": 0.1848, "step": 28261 }, { "epoch": 20.704761904761906, "grad_norm": 0.76171875, "learning_rate": 3.969528884222358e-05, "loss": 0.1234, "step": 28262 }, { "epoch": 20.705494505494507, "grad_norm": 0.9609375, "learning_rate": 3.9691945127105224e-05, "loss": 0.1194, "step": 28263 }, { "epoch": 20.706227106227107, "grad_norm": 0.7265625, "learning_rate": 3.968860146377181e-05, "loss": 0.1044, "step": 28264 }, { "epoch": 20.706959706959708, "grad_norm": 1.25, "learning_rate": 3.9685257852239354e-05, "loss": 0.0993, "step": 28265 }, { "epoch": 20.70769230769231, "grad_norm": 1.7421875, "learning_rate": 3.968191429252387e-05, "loss": 0.1294, "step": 28266 }, { "epoch": 20.70842490842491, "grad_norm": 1.5625, "learning_rate": 3.967857078464138e-05, "loss": 0.1219, "step": 28267 }, { "epoch": 20.70915750915751, "grad_norm": 0.71484375, "learning_rate": 3.967522732860794e-05, "loss": 0.1087, "step": 28268 }, { "epoch": 20.70989010989011, "grad_norm": 1.28125, "learning_rate": 3.967188392443952e-05, "loss": 0.1421, "step": 28269 }, { "epoch": 20.71062271062271, "grad_norm": 1.2109375, "learning_rate": 3.9668540572152166e-05, "loss": 0.139, "step": 28270 }, { "epoch": 20.711355311355312, "grad_norm": 0.734375, "learning_rate": 3.966519727176187e-05, "loss": 0.1008, "step": 28271 }, { "epoch": 20.712087912087913, "grad_norm": 0.8125, "learning_rate": 3.9661854023284686e-05, "loss": 0.122, "step": 28272 }, { "epoch": 20.712820512820514, "grad_norm": 1.078125, "learning_rate": 3.965851082673662e-05, "loss": 0.1386, "step": 28273 }, { "epoch": 20.713553113553115, "grad_norm": 1.28125, "learning_rate": 3.965516768213368e-05, "loss": 0.0955, "step": 28274 }, { "epoch": 20.714285714285715, "grad_norm": 1.1484375, "learning_rate": 3.9651824589491906e-05, "loss": 0.1315, "step": 28275 }, { "epoch": 20.715018315018316, "grad_norm": 1.125, "learning_rate": 3.964848154882729e-05, "loss": 0.1159, "step": 28276 }, { "epoch": 20.715750915750917, "grad_norm": 1.6953125, "learning_rate": 3.9645138560155846e-05, "loss": 0.1208, "step": 28277 }, { "epoch": 20.716483516483517, "grad_norm": 0.84375, "learning_rate": 3.9641795623493624e-05, "loss": 0.1144, "step": 28278 }, { "epoch": 20.717216117216118, "grad_norm": 1.1484375, "learning_rate": 3.963845273885663e-05, "loss": 0.1198, "step": 28279 }, { "epoch": 20.71794871794872, "grad_norm": 1.0390625, "learning_rate": 3.963510990626087e-05, "loss": 0.1293, "step": 28280 }, { "epoch": 20.71868131868132, "grad_norm": 1.265625, "learning_rate": 3.963176712572237e-05, "loss": 0.1408, "step": 28281 }, { "epoch": 20.71941391941392, "grad_norm": 0.81640625, "learning_rate": 3.962842439725712e-05, "loss": 0.1104, "step": 28282 }, { "epoch": 20.72014652014652, "grad_norm": 0.92578125, "learning_rate": 3.9625081720881175e-05, "loss": 0.102, "step": 28283 }, { "epoch": 20.720879120879122, "grad_norm": 1.125, "learning_rate": 3.962173909661054e-05, "loss": 0.1442, "step": 28284 }, { "epoch": 20.721611721611723, "grad_norm": 0.94140625, "learning_rate": 3.9618396524461214e-05, "loss": 0.1054, "step": 28285 }, { "epoch": 20.722344322344323, "grad_norm": 1.328125, "learning_rate": 3.9615054004449214e-05, "loss": 0.1064, "step": 28286 }, { "epoch": 20.723076923076924, "grad_norm": 0.76171875, "learning_rate": 3.961171153659057e-05, "loss": 0.1327, "step": 28287 }, { "epoch": 20.723809523809525, "grad_norm": 1.03125, "learning_rate": 3.9608369120901295e-05, "loss": 0.1053, "step": 28288 }, { "epoch": 20.724542124542126, "grad_norm": 1.203125, "learning_rate": 3.960502675739739e-05, "loss": 0.1222, "step": 28289 }, { "epoch": 20.725274725274726, "grad_norm": 1.2109375, "learning_rate": 3.9601684446094896e-05, "loss": 0.1198, "step": 28290 }, { "epoch": 20.726007326007327, "grad_norm": 1.140625, "learning_rate": 3.9598342187009796e-05, "loss": 0.1352, "step": 28291 }, { "epoch": 20.726739926739928, "grad_norm": 0.73046875, "learning_rate": 3.959499998015811e-05, "loss": 0.1128, "step": 28292 }, { "epoch": 20.72747252747253, "grad_norm": 0.875, "learning_rate": 3.959165782555587e-05, "loss": 0.1173, "step": 28293 }, { "epoch": 20.72820512820513, "grad_norm": 1.1796875, "learning_rate": 3.9588315723219075e-05, "loss": 0.1247, "step": 28294 }, { "epoch": 20.72893772893773, "grad_norm": 0.9453125, "learning_rate": 3.9584973673163755e-05, "loss": 0.1083, "step": 28295 }, { "epoch": 20.72967032967033, "grad_norm": 0.73046875, "learning_rate": 3.95816316754059e-05, "loss": 0.099, "step": 28296 }, { "epoch": 20.73040293040293, "grad_norm": 0.890625, "learning_rate": 3.957828972996152e-05, "loss": 0.1208, "step": 28297 }, { "epoch": 20.731135531135532, "grad_norm": 1.2890625, "learning_rate": 3.9574947836846646e-05, "loss": 0.1348, "step": 28298 }, { "epoch": 20.731868131868133, "grad_norm": 1.0, "learning_rate": 3.957160599607729e-05, "loss": 0.1237, "step": 28299 }, { "epoch": 20.732600732600734, "grad_norm": 1.328125, "learning_rate": 3.9568264207669446e-05, "loss": 0.1111, "step": 28300 }, { "epoch": 20.733333333333334, "grad_norm": 1.453125, "learning_rate": 3.956492247163915e-05, "loss": 0.1077, "step": 28301 }, { "epoch": 20.734065934065935, "grad_norm": 0.859375, "learning_rate": 3.9561580788002404e-05, "loss": 0.1273, "step": 28302 }, { "epoch": 20.734798534798536, "grad_norm": 1.2734375, "learning_rate": 3.955823915677519e-05, "loss": 0.1192, "step": 28303 }, { "epoch": 20.735531135531136, "grad_norm": 1.203125, "learning_rate": 3.955489757797356e-05, "loss": 0.1225, "step": 28304 }, { "epoch": 20.736263736263737, "grad_norm": 1.375, "learning_rate": 3.955155605161352e-05, "loss": 0.1106, "step": 28305 }, { "epoch": 20.736996336996338, "grad_norm": 1.1640625, "learning_rate": 3.954821457771106e-05, "loss": 0.1673, "step": 28306 }, { "epoch": 20.73772893772894, "grad_norm": 0.828125, "learning_rate": 3.95448731562822e-05, "loss": 0.1123, "step": 28307 }, { "epoch": 20.73846153846154, "grad_norm": 1.640625, "learning_rate": 3.9541531787342934e-05, "loss": 0.1211, "step": 28308 }, { "epoch": 20.73919413919414, "grad_norm": 0.88671875, "learning_rate": 3.95381904709093e-05, "loss": 0.1129, "step": 28309 }, { "epoch": 20.73992673992674, "grad_norm": 1.265625, "learning_rate": 3.95348492069973e-05, "loss": 0.1354, "step": 28310 }, { "epoch": 20.74065934065934, "grad_norm": 1.46875, "learning_rate": 3.9531507995622935e-05, "loss": 0.1052, "step": 28311 }, { "epoch": 20.741391941391942, "grad_norm": 1.1796875, "learning_rate": 3.9528166836802205e-05, "loss": 0.1603, "step": 28312 }, { "epoch": 20.742124542124543, "grad_norm": 1.0859375, "learning_rate": 3.952482573055113e-05, "loss": 0.1208, "step": 28313 }, { "epoch": 20.742857142857144, "grad_norm": 1.0078125, "learning_rate": 3.952148467688572e-05, "loss": 0.1326, "step": 28314 }, { "epoch": 20.743589743589745, "grad_norm": 1.015625, "learning_rate": 3.951814367582198e-05, "loss": 0.109, "step": 28315 }, { "epoch": 20.744322344322345, "grad_norm": 1.2890625, "learning_rate": 3.9514802727375925e-05, "loss": 0.1083, "step": 28316 }, { "epoch": 20.745054945054946, "grad_norm": 0.88671875, "learning_rate": 3.951146183156357e-05, "loss": 0.0927, "step": 28317 }, { "epoch": 20.745787545787547, "grad_norm": 0.87890625, "learning_rate": 3.950812098840087e-05, "loss": 0.0863, "step": 28318 }, { "epoch": 20.746520146520147, "grad_norm": 1.6484375, "learning_rate": 3.9504780197903895e-05, "loss": 0.1251, "step": 28319 }, { "epoch": 20.747252747252748, "grad_norm": 1.3984375, "learning_rate": 3.950143946008863e-05, "loss": 0.1136, "step": 28320 }, { "epoch": 20.74798534798535, "grad_norm": 1.4609375, "learning_rate": 3.9498098774971086e-05, "loss": 0.1184, "step": 28321 }, { "epoch": 20.74871794871795, "grad_norm": 1.1640625, "learning_rate": 3.949475814256726e-05, "loss": 0.1315, "step": 28322 }, { "epoch": 20.74945054945055, "grad_norm": 1.5234375, "learning_rate": 3.9491417562893135e-05, "loss": 0.1363, "step": 28323 }, { "epoch": 20.75018315018315, "grad_norm": 0.9765625, "learning_rate": 3.948807703596476e-05, "loss": 0.1307, "step": 28324 }, { "epoch": 20.750915750915752, "grad_norm": 1.2578125, "learning_rate": 3.948473656179812e-05, "loss": 0.1893, "step": 28325 }, { "epoch": 20.751648351648353, "grad_norm": 0.76953125, "learning_rate": 3.948139614040923e-05, "loss": 0.107, "step": 28326 }, { "epoch": 20.752380952380953, "grad_norm": 0.91796875, "learning_rate": 3.9478055771814085e-05, "loss": 0.1165, "step": 28327 }, { "epoch": 20.753113553113554, "grad_norm": 0.6328125, "learning_rate": 3.94747154560287e-05, "loss": 0.1158, "step": 28328 }, { "epoch": 20.753846153846155, "grad_norm": 0.625, "learning_rate": 3.947137519306905e-05, "loss": 0.0939, "step": 28329 }, { "epoch": 20.754578754578755, "grad_norm": 1.375, "learning_rate": 3.946803498295118e-05, "loss": 0.1342, "step": 28330 }, { "epoch": 20.755311355311356, "grad_norm": 0.9921875, "learning_rate": 3.946469482569107e-05, "loss": 0.1296, "step": 28331 }, { "epoch": 20.756043956043957, "grad_norm": 0.80078125, "learning_rate": 3.946135472130474e-05, "loss": 0.0817, "step": 28332 }, { "epoch": 20.756776556776558, "grad_norm": 1.7734375, "learning_rate": 3.945801466980819e-05, "loss": 0.1311, "step": 28333 }, { "epoch": 20.75750915750916, "grad_norm": 0.89453125, "learning_rate": 3.945467467121737e-05, "loss": 0.1256, "step": 28334 }, { "epoch": 20.75824175824176, "grad_norm": 1.03125, "learning_rate": 3.945133472554837e-05, "loss": 0.1199, "step": 28335 }, { "epoch": 20.75897435897436, "grad_norm": 1.2890625, "learning_rate": 3.9447994832817155e-05, "loss": 0.1203, "step": 28336 }, { "epoch": 20.75970695970696, "grad_norm": 1.03125, "learning_rate": 3.9444654993039716e-05, "loss": 0.1216, "step": 28337 }, { "epoch": 20.76043956043956, "grad_norm": 0.78515625, "learning_rate": 3.944131520623206e-05, "loss": 0.1346, "step": 28338 }, { "epoch": 20.761172161172162, "grad_norm": 1.765625, "learning_rate": 3.943797547241018e-05, "loss": 0.1553, "step": 28339 }, { "epoch": 20.761904761904763, "grad_norm": 0.765625, "learning_rate": 3.943463579159011e-05, "loss": 0.1184, "step": 28340 }, { "epoch": 20.762637362637363, "grad_norm": 0.9140625, "learning_rate": 3.9431296163787826e-05, "loss": 0.1193, "step": 28341 }, { "epoch": 20.763369963369964, "grad_norm": 0.984375, "learning_rate": 3.942795658901935e-05, "loss": 0.1082, "step": 28342 }, { "epoch": 20.764102564102565, "grad_norm": 0.828125, "learning_rate": 3.942461706730066e-05, "loss": 0.1142, "step": 28343 }, { "epoch": 20.764835164835166, "grad_norm": 0.83203125, "learning_rate": 3.9421277598647744e-05, "loss": 0.1138, "step": 28344 }, { "epoch": 20.765567765567766, "grad_norm": 1.015625, "learning_rate": 3.941793818307665e-05, "loss": 0.1263, "step": 28345 }, { "epoch": 20.766300366300367, "grad_norm": 0.66015625, "learning_rate": 3.9414598820603346e-05, "loss": 0.1217, "step": 28346 }, { "epoch": 20.767032967032968, "grad_norm": 0.9140625, "learning_rate": 3.941125951124384e-05, "loss": 0.1401, "step": 28347 }, { "epoch": 20.76776556776557, "grad_norm": 1.28125, "learning_rate": 3.940792025501413e-05, "loss": 0.1182, "step": 28348 }, { "epoch": 20.76849816849817, "grad_norm": 0.84765625, "learning_rate": 3.9404581051930215e-05, "loss": 0.1438, "step": 28349 }, { "epoch": 20.76923076923077, "grad_norm": 1.1171875, "learning_rate": 3.940124190200808e-05, "loss": 0.1227, "step": 28350 }, { "epoch": 20.76996336996337, "grad_norm": 0.8828125, "learning_rate": 3.939790280526376e-05, "loss": 0.0956, "step": 28351 }, { "epoch": 20.77069597069597, "grad_norm": 1.0, "learning_rate": 3.939456376171321e-05, "loss": 0.1207, "step": 28352 }, { "epoch": 20.771428571428572, "grad_norm": 0.99609375, "learning_rate": 3.939122477137246e-05, "loss": 0.11, "step": 28353 }, { "epoch": 20.772161172161173, "grad_norm": 0.99609375, "learning_rate": 3.938788583425752e-05, "loss": 0.1153, "step": 28354 }, { "epoch": 20.772893772893774, "grad_norm": 0.79296875, "learning_rate": 3.9384546950384324e-05, "loss": 0.1063, "step": 28355 }, { "epoch": 20.773626373626374, "grad_norm": 1.1640625, "learning_rate": 3.9381208119768944e-05, "loss": 0.122, "step": 28356 }, { "epoch": 20.774358974358975, "grad_norm": 1.1015625, "learning_rate": 3.937786934242734e-05, "loss": 0.1136, "step": 28357 }, { "epoch": 20.775091575091576, "grad_norm": 0.9453125, "learning_rate": 3.937453061837552e-05, "loss": 0.1092, "step": 28358 }, { "epoch": 20.775824175824177, "grad_norm": 0.83984375, "learning_rate": 3.9371191947629476e-05, "loss": 0.1189, "step": 28359 }, { "epoch": 20.776556776556777, "grad_norm": 1.40625, "learning_rate": 3.936785333020519e-05, "loss": 0.1405, "step": 28360 }, { "epoch": 20.777289377289378, "grad_norm": 1.140625, "learning_rate": 3.936451476611868e-05, "loss": 0.1173, "step": 28361 }, { "epoch": 20.77802197802198, "grad_norm": 1.6953125, "learning_rate": 3.936117625538594e-05, "loss": 0.1286, "step": 28362 }, { "epoch": 20.77875457875458, "grad_norm": 1.171875, "learning_rate": 3.935783779802297e-05, "loss": 0.1748, "step": 28363 }, { "epoch": 20.77948717948718, "grad_norm": 1.1328125, "learning_rate": 3.935449939404573e-05, "loss": 0.1208, "step": 28364 }, { "epoch": 20.78021978021978, "grad_norm": 1.359375, "learning_rate": 3.9351161043470256e-05, "loss": 0.1338, "step": 28365 }, { "epoch": 20.78095238095238, "grad_norm": 0.828125, "learning_rate": 3.9347822746312534e-05, "loss": 0.1184, "step": 28366 }, { "epoch": 20.781684981684982, "grad_norm": 1.1171875, "learning_rate": 3.934448450258855e-05, "loss": 0.1208, "step": 28367 }, { "epoch": 20.782417582417583, "grad_norm": 0.8671875, "learning_rate": 3.934114631231431e-05, "loss": 0.0916, "step": 28368 }, { "epoch": 20.783150183150184, "grad_norm": 1.0859375, "learning_rate": 3.93378081755058e-05, "loss": 0.1245, "step": 28369 }, { "epoch": 20.783882783882785, "grad_norm": 1.109375, "learning_rate": 3.9334470092178996e-05, "loss": 0.1312, "step": 28370 }, { "epoch": 20.784615384615385, "grad_norm": 0.921875, "learning_rate": 3.933113206234993e-05, "loss": 0.1391, "step": 28371 }, { "epoch": 20.785347985347986, "grad_norm": 1.2890625, "learning_rate": 3.932779408603458e-05, "loss": 0.1219, "step": 28372 }, { "epoch": 20.786080586080587, "grad_norm": 1.03125, "learning_rate": 3.932445616324893e-05, "loss": 0.1108, "step": 28373 }, { "epoch": 20.786813186813188, "grad_norm": 0.78515625, "learning_rate": 3.932111829400898e-05, "loss": 0.1067, "step": 28374 }, { "epoch": 20.78754578754579, "grad_norm": 0.859375, "learning_rate": 3.931778047833072e-05, "loss": 0.1317, "step": 28375 }, { "epoch": 20.78827838827839, "grad_norm": 1.046875, "learning_rate": 3.931444271623014e-05, "loss": 0.1149, "step": 28376 }, { "epoch": 20.78901098901099, "grad_norm": 0.98046875, "learning_rate": 3.9311105007723244e-05, "loss": 0.1231, "step": 28377 }, { "epoch": 20.78974358974359, "grad_norm": 1.25, "learning_rate": 3.930776735282601e-05, "loss": 0.1085, "step": 28378 }, { "epoch": 20.79047619047619, "grad_norm": 0.828125, "learning_rate": 3.930442975155445e-05, "loss": 0.1082, "step": 28379 }, { "epoch": 20.791208791208792, "grad_norm": 1.5, "learning_rate": 3.930109220392454e-05, "loss": 0.1418, "step": 28380 }, { "epoch": 20.791941391941393, "grad_norm": 1.2421875, "learning_rate": 3.929775470995224e-05, "loss": 0.0955, "step": 28381 }, { "epoch": 20.792673992673993, "grad_norm": 0.79296875, "learning_rate": 3.929441726965361e-05, "loss": 0.1264, "step": 28382 }, { "epoch": 20.793406593406594, "grad_norm": 0.69921875, "learning_rate": 3.9291079883044604e-05, "loss": 0.0863, "step": 28383 }, { "epoch": 20.794139194139195, "grad_norm": 1.421875, "learning_rate": 3.928774255014121e-05, "loss": 0.1191, "step": 28384 }, { "epoch": 20.794871794871796, "grad_norm": 0.6640625, "learning_rate": 3.928440527095942e-05, "loss": 0.1108, "step": 28385 }, { "epoch": 20.795604395604396, "grad_norm": 1.71875, "learning_rate": 3.928106804551522e-05, "loss": 0.1226, "step": 28386 }, { "epoch": 20.796336996336997, "grad_norm": 0.56640625, "learning_rate": 3.9277730873824607e-05, "loss": 0.1228, "step": 28387 }, { "epoch": 20.797069597069598, "grad_norm": 0.8515625, "learning_rate": 3.9274393755903585e-05, "loss": 0.1011, "step": 28388 }, { "epoch": 20.7978021978022, "grad_norm": 1.0859375, "learning_rate": 3.927105669176811e-05, "loss": 0.1315, "step": 28389 }, { "epoch": 20.7985347985348, "grad_norm": 1.1640625, "learning_rate": 3.92677196814342e-05, "loss": 0.1027, "step": 28390 }, { "epoch": 20.7992673992674, "grad_norm": 0.9453125, "learning_rate": 3.926438272491782e-05, "loss": 0.1168, "step": 28391 }, { "epoch": 20.8, "grad_norm": 1.203125, "learning_rate": 3.926104582223498e-05, "loss": 0.1224, "step": 28392 }, { "epoch": 20.8007326007326, "grad_norm": 0.88671875, "learning_rate": 3.9257708973401654e-05, "loss": 0.1391, "step": 28393 }, { "epoch": 20.801465201465202, "grad_norm": 1.578125, "learning_rate": 3.925437217843384e-05, "loss": 0.1714, "step": 28394 }, { "epoch": 20.802197802197803, "grad_norm": 1.75, "learning_rate": 3.9251035437347536e-05, "loss": 0.1287, "step": 28395 }, { "epoch": 20.802930402930404, "grad_norm": 0.91015625, "learning_rate": 3.924769875015867e-05, "loss": 0.1012, "step": 28396 }, { "epoch": 20.803663003663004, "grad_norm": 1.3984375, "learning_rate": 3.924436211688331e-05, "loss": 0.1297, "step": 28397 }, { "epoch": 20.804395604395605, "grad_norm": 2.203125, "learning_rate": 3.92410255375374e-05, "loss": 0.1297, "step": 28398 }, { "epoch": 20.805128205128206, "grad_norm": 0.84765625, "learning_rate": 3.923768901213693e-05, "loss": 0.106, "step": 28399 }, { "epoch": 20.805860805860807, "grad_norm": 1.3359375, "learning_rate": 3.923435254069789e-05, "loss": 0.1548, "step": 28400 }, { "epoch": 20.806593406593407, "grad_norm": 1.203125, "learning_rate": 3.923101612323626e-05, "loss": 0.1272, "step": 28401 }, { "epoch": 20.807326007326008, "grad_norm": 1.0546875, "learning_rate": 3.922767975976803e-05, "loss": 0.1302, "step": 28402 }, { "epoch": 20.80805860805861, "grad_norm": 1.140625, "learning_rate": 3.92243434503092e-05, "loss": 0.1283, "step": 28403 }, { "epoch": 20.80879120879121, "grad_norm": 0.81640625, "learning_rate": 3.922100719487573e-05, "loss": 0.1281, "step": 28404 }, { "epoch": 20.80952380952381, "grad_norm": 1.1328125, "learning_rate": 3.9217670993483616e-05, "loss": 0.143, "step": 28405 }, { "epoch": 20.81025641025641, "grad_norm": 0.7734375, "learning_rate": 3.9214334846148856e-05, "loss": 0.111, "step": 28406 }, { "epoch": 20.81098901098901, "grad_norm": 0.68359375, "learning_rate": 3.92109987528874e-05, "loss": 0.1243, "step": 28407 }, { "epoch": 20.811721611721612, "grad_norm": 1.1953125, "learning_rate": 3.920766271371528e-05, "loss": 0.1375, "step": 28408 }, { "epoch": 20.812454212454213, "grad_norm": 0.96875, "learning_rate": 3.920432672864844e-05, "loss": 0.1124, "step": 28409 }, { "epoch": 20.813186813186814, "grad_norm": 1.0234375, "learning_rate": 3.92009907977029e-05, "loss": 0.1203, "step": 28410 }, { "epoch": 20.813919413919415, "grad_norm": 1.015625, "learning_rate": 3.91976549208946e-05, "loss": 0.1062, "step": 28411 }, { "epoch": 20.814652014652015, "grad_norm": 1.515625, "learning_rate": 3.919431909823954e-05, "loss": 0.1332, "step": 28412 }, { "epoch": 20.815384615384616, "grad_norm": 0.75390625, "learning_rate": 3.919098332975372e-05, "loss": 0.1338, "step": 28413 }, { "epoch": 20.816117216117217, "grad_norm": 1.5234375, "learning_rate": 3.918764761545312e-05, "loss": 0.1096, "step": 28414 }, { "epoch": 20.816849816849818, "grad_norm": 1.109375, "learning_rate": 3.9184311955353696e-05, "loss": 0.1513, "step": 28415 }, { "epoch": 20.81758241758242, "grad_norm": 0.7421875, "learning_rate": 3.9180976349471466e-05, "loss": 0.1018, "step": 28416 }, { "epoch": 20.81831501831502, "grad_norm": 1.4296875, "learning_rate": 3.9177640797822385e-05, "loss": 0.0942, "step": 28417 }, { "epoch": 20.81904761904762, "grad_norm": 1.28125, "learning_rate": 3.917430530042243e-05, "loss": 0.106, "step": 28418 }, { "epoch": 20.81978021978022, "grad_norm": 0.984375, "learning_rate": 3.917096985728762e-05, "loss": 0.1292, "step": 28419 }, { "epoch": 20.82051282051282, "grad_norm": 0.77734375, "learning_rate": 3.9167634468433906e-05, "loss": 0.1181, "step": 28420 }, { "epoch": 20.821245421245422, "grad_norm": 1.265625, "learning_rate": 3.916429913387728e-05, "loss": 0.1469, "step": 28421 }, { "epoch": 20.821978021978023, "grad_norm": 1.484375, "learning_rate": 3.916096385363369e-05, "loss": 0.1109, "step": 28422 }, { "epoch": 20.822710622710623, "grad_norm": 1.015625, "learning_rate": 3.915762862771917e-05, "loss": 0.1233, "step": 28423 }, { "epoch": 20.823443223443224, "grad_norm": 1.2578125, "learning_rate": 3.915429345614968e-05, "loss": 0.1263, "step": 28424 }, { "epoch": 20.824175824175825, "grad_norm": 1.1015625, "learning_rate": 3.9150958338941186e-05, "loss": 0.1298, "step": 28425 }, { "epoch": 20.824908424908426, "grad_norm": 0.5703125, "learning_rate": 3.914762327610968e-05, "loss": 0.1134, "step": 28426 }, { "epoch": 20.825641025641026, "grad_norm": 2.265625, "learning_rate": 3.914428826767113e-05, "loss": 0.1496, "step": 28427 }, { "epoch": 20.826373626373627, "grad_norm": 1.328125, "learning_rate": 3.914095331364152e-05, "loss": 0.1417, "step": 28428 }, { "epoch": 20.827106227106228, "grad_norm": 1.53125, "learning_rate": 3.913761841403684e-05, "loss": 0.131, "step": 28429 }, { "epoch": 20.82783882783883, "grad_norm": 0.9140625, "learning_rate": 3.913428356887305e-05, "loss": 0.1113, "step": 28430 }, { "epoch": 20.82857142857143, "grad_norm": 0.87890625, "learning_rate": 3.9130948778166146e-05, "loss": 0.1119, "step": 28431 }, { "epoch": 20.82930402930403, "grad_norm": 0.96875, "learning_rate": 3.91276140419321e-05, "loss": 0.1503, "step": 28432 }, { "epoch": 20.83003663003663, "grad_norm": 1.3828125, "learning_rate": 3.912427936018688e-05, "loss": 0.1352, "step": 28433 }, { "epoch": 20.83076923076923, "grad_norm": 1.046875, "learning_rate": 3.912094473294648e-05, "loss": 0.1158, "step": 28434 }, { "epoch": 20.831501831501832, "grad_norm": 0.9765625, "learning_rate": 3.9117610160226874e-05, "loss": 0.1274, "step": 28435 }, { "epoch": 20.832234432234433, "grad_norm": 0.6484375, "learning_rate": 3.911427564204403e-05, "loss": 0.0999, "step": 28436 }, { "epoch": 20.832967032967034, "grad_norm": 1.1484375, "learning_rate": 3.9110941178413935e-05, "loss": 0.1905, "step": 28437 }, { "epoch": 20.833699633699634, "grad_norm": 1.078125, "learning_rate": 3.910760676935253e-05, "loss": 0.1046, "step": 28438 }, { "epoch": 20.834432234432235, "grad_norm": 1.1875, "learning_rate": 3.910427241487585e-05, "loss": 0.1523, "step": 28439 }, { "epoch": 20.835164835164836, "grad_norm": 0.921875, "learning_rate": 3.910093811499985e-05, "loss": 0.1218, "step": 28440 }, { "epoch": 20.835897435897436, "grad_norm": 0.9765625, "learning_rate": 3.909760386974047e-05, "loss": 0.1165, "step": 28441 }, { "epoch": 20.836630036630037, "grad_norm": 1.234375, "learning_rate": 3.9094269679113725e-05, "loss": 0.0965, "step": 28442 }, { "epoch": 20.837362637362638, "grad_norm": 1.1015625, "learning_rate": 3.9090935543135576e-05, "loss": 0.1034, "step": 28443 }, { "epoch": 20.83809523809524, "grad_norm": 1.4765625, "learning_rate": 3.908760146182199e-05, "loss": 0.1144, "step": 28444 }, { "epoch": 20.83882783882784, "grad_norm": 0.96484375, "learning_rate": 3.908426743518898e-05, "loss": 0.1305, "step": 28445 }, { "epoch": 20.83956043956044, "grad_norm": 0.96484375, "learning_rate": 3.908093346325248e-05, "loss": 0.1306, "step": 28446 }, { "epoch": 20.84029304029304, "grad_norm": 1.171875, "learning_rate": 3.907759954602847e-05, "loss": 0.1192, "step": 28447 }, { "epoch": 20.84102564102564, "grad_norm": 1.5078125, "learning_rate": 3.9074265683532926e-05, "loss": 0.1287, "step": 28448 }, { "epoch": 20.841758241758242, "grad_norm": 0.8359375, "learning_rate": 3.907093187578184e-05, "loss": 0.1408, "step": 28449 }, { "epoch": 20.842490842490843, "grad_norm": 1.203125, "learning_rate": 3.906759812279116e-05, "loss": 0.1284, "step": 28450 }, { "epoch": 20.843223443223444, "grad_norm": 1.140625, "learning_rate": 3.9064264424576884e-05, "loss": 0.1075, "step": 28451 }, { "epoch": 20.843956043956045, "grad_norm": 0.96875, "learning_rate": 3.906093078115497e-05, "loss": 0.1187, "step": 28452 }, { "epoch": 20.844688644688645, "grad_norm": 0.67578125, "learning_rate": 3.905759719254137e-05, "loss": 0.1288, "step": 28453 }, { "epoch": 20.845421245421246, "grad_norm": 0.97265625, "learning_rate": 3.9054263658752094e-05, "loss": 0.1292, "step": 28454 }, { "epoch": 20.846153846153847, "grad_norm": 1.21875, "learning_rate": 3.905093017980308e-05, "loss": 0.155, "step": 28455 }, { "epoch": 20.846886446886447, "grad_norm": 1.03125, "learning_rate": 3.9047596755710345e-05, "loss": 0.1117, "step": 28456 }, { "epoch": 20.847619047619048, "grad_norm": 0.703125, "learning_rate": 3.904426338648983e-05, "loss": 0.0993, "step": 28457 }, { "epoch": 20.84835164835165, "grad_norm": 1.3671875, "learning_rate": 3.904093007215751e-05, "loss": 0.1069, "step": 28458 }, { "epoch": 20.84908424908425, "grad_norm": 1.0078125, "learning_rate": 3.9037596812729335e-05, "loss": 0.1071, "step": 28459 }, { "epoch": 20.84981684981685, "grad_norm": 1.609375, "learning_rate": 3.903426360822131e-05, "loss": 0.1727, "step": 28460 }, { "epoch": 20.85054945054945, "grad_norm": 0.91015625, "learning_rate": 3.9030930458649396e-05, "loss": 0.1098, "step": 28461 }, { "epoch": 20.851282051282052, "grad_norm": 1.0703125, "learning_rate": 3.9027597364029564e-05, "loss": 0.151, "step": 28462 }, { "epoch": 20.852014652014653, "grad_norm": 1.6484375, "learning_rate": 3.902426432437777e-05, "loss": 0.1206, "step": 28463 }, { "epoch": 20.852747252747253, "grad_norm": 0.859375, "learning_rate": 3.902093133970998e-05, "loss": 0.1068, "step": 28464 }, { "epoch": 20.853479853479854, "grad_norm": 0.890625, "learning_rate": 3.90175984100422e-05, "loss": 0.1407, "step": 28465 }, { "epoch": 20.854212454212455, "grad_norm": 0.625, "learning_rate": 3.9014265535390364e-05, "loss": 0.0951, "step": 28466 }, { "epoch": 20.854945054945055, "grad_norm": 1.0859375, "learning_rate": 3.901093271577045e-05, "loss": 0.0914, "step": 28467 }, { "epoch": 20.855677655677656, "grad_norm": 1.3125, "learning_rate": 3.900759995119843e-05, "loss": 0.1, "step": 28468 }, { "epoch": 20.856410256410257, "grad_norm": 0.5625, "learning_rate": 3.900426724169027e-05, "loss": 0.1067, "step": 28469 }, { "epoch": 20.857142857142858, "grad_norm": 0.94921875, "learning_rate": 3.9000934587261936e-05, "loss": 0.1059, "step": 28470 }, { "epoch": 20.85787545787546, "grad_norm": 0.80859375, "learning_rate": 3.899760198792941e-05, "loss": 0.1349, "step": 28471 }, { "epoch": 20.85860805860806, "grad_norm": 1.0, "learning_rate": 3.899426944370865e-05, "loss": 0.13, "step": 28472 }, { "epoch": 20.85934065934066, "grad_norm": 0.97265625, "learning_rate": 3.899093695461562e-05, "loss": 0.1123, "step": 28473 }, { "epoch": 20.86007326007326, "grad_norm": 1.1171875, "learning_rate": 3.898760452066628e-05, "loss": 0.1258, "step": 28474 }, { "epoch": 20.86080586080586, "grad_norm": 0.8984375, "learning_rate": 3.898427214187662e-05, "loss": 0.1018, "step": 28475 }, { "epoch": 20.861538461538462, "grad_norm": 1.2109375, "learning_rate": 3.8980939818262584e-05, "loss": 0.1073, "step": 28476 }, { "epoch": 20.862271062271063, "grad_norm": 1.6875, "learning_rate": 3.897760754984016e-05, "loss": 0.1259, "step": 28477 }, { "epoch": 20.863003663003663, "grad_norm": 1.4296875, "learning_rate": 3.897427533662529e-05, "loss": 0.1349, "step": 28478 }, { "epoch": 20.863736263736264, "grad_norm": 0.953125, "learning_rate": 3.897094317863394e-05, "loss": 0.1498, "step": 28479 }, { "epoch": 20.864468864468865, "grad_norm": 1.1640625, "learning_rate": 3.896761107588209e-05, "loss": 0.1239, "step": 28480 }, { "epoch": 20.865201465201466, "grad_norm": 1.3203125, "learning_rate": 3.89642790283857e-05, "loss": 0.1289, "step": 28481 }, { "epoch": 20.865934065934066, "grad_norm": 1.859375, "learning_rate": 3.896094703616074e-05, "loss": 0.1149, "step": 28482 }, { "epoch": 20.866666666666667, "grad_norm": 1.40625, "learning_rate": 3.895761509922319e-05, "loss": 0.1403, "step": 28483 }, { "epoch": 20.867399267399268, "grad_norm": 0.6796875, "learning_rate": 3.8954283217588975e-05, "loss": 0.1083, "step": 28484 }, { "epoch": 20.86813186813187, "grad_norm": 1.2890625, "learning_rate": 3.895095139127407e-05, "loss": 0.1191, "step": 28485 }, { "epoch": 20.86886446886447, "grad_norm": 1.390625, "learning_rate": 3.894761962029447e-05, "loss": 0.1235, "step": 28486 }, { "epoch": 20.86959706959707, "grad_norm": 0.83984375, "learning_rate": 3.89442879046661e-05, "loss": 0.1104, "step": 28487 }, { "epoch": 20.87032967032967, "grad_norm": 0.875, "learning_rate": 3.8940956244404965e-05, "loss": 0.1224, "step": 28488 }, { "epoch": 20.87106227106227, "grad_norm": 1.0703125, "learning_rate": 3.893762463952699e-05, "loss": 0.1421, "step": 28489 }, { "epoch": 20.871794871794872, "grad_norm": 1.1015625, "learning_rate": 3.8934293090048125e-05, "loss": 0.1334, "step": 28490 }, { "epoch": 20.872527472527473, "grad_norm": 1.015625, "learning_rate": 3.8930961595984386e-05, "loss": 0.1331, "step": 28491 }, { "epoch": 20.873260073260074, "grad_norm": 0.79296875, "learning_rate": 3.892763015735171e-05, "loss": 0.1112, "step": 28492 }, { "epoch": 20.873992673992674, "grad_norm": 1.2265625, "learning_rate": 3.892429877416605e-05, "loss": 0.1335, "step": 28493 }, { "epoch": 20.874725274725275, "grad_norm": 1.65625, "learning_rate": 3.892096744644337e-05, "loss": 0.1072, "step": 28494 }, { "epoch": 20.875457875457876, "grad_norm": 1.25, "learning_rate": 3.891763617419964e-05, "loss": 0.1117, "step": 28495 }, { "epoch": 20.876190476190477, "grad_norm": 1.0, "learning_rate": 3.8914304957450806e-05, "loss": 0.1391, "step": 28496 }, { "epoch": 20.876923076923077, "grad_norm": 0.80859375, "learning_rate": 3.8910973796212854e-05, "loss": 0.096, "step": 28497 }, { "epoch": 20.877655677655678, "grad_norm": 1.4140625, "learning_rate": 3.890764269050174e-05, "loss": 0.1159, "step": 28498 }, { "epoch": 20.87838827838828, "grad_norm": 1.2109375, "learning_rate": 3.890431164033341e-05, "loss": 0.1022, "step": 28499 }, { "epoch": 20.87912087912088, "grad_norm": 1.5234375, "learning_rate": 3.890098064572381e-05, "loss": 0.1458, "step": 28500 }, { "epoch": 20.87985347985348, "grad_norm": 1.0859375, "learning_rate": 3.889764970668893e-05, "loss": 0.1531, "step": 28501 }, { "epoch": 20.88058608058608, "grad_norm": 0.68359375, "learning_rate": 3.889431882324473e-05, "loss": 0.1168, "step": 28502 }, { "epoch": 20.88131868131868, "grad_norm": 1.84375, "learning_rate": 3.8890987995407155e-05, "loss": 0.128, "step": 28503 }, { "epoch": 20.882051282051282, "grad_norm": 1.1640625, "learning_rate": 3.888765722319217e-05, "loss": 0.1519, "step": 28504 }, { "epoch": 20.882783882783883, "grad_norm": 1.0546875, "learning_rate": 3.888432650661572e-05, "loss": 0.1203, "step": 28505 }, { "epoch": 20.883516483516484, "grad_norm": 0.8515625, "learning_rate": 3.888099584569378e-05, "loss": 0.1029, "step": 28506 }, { "epoch": 20.884249084249085, "grad_norm": 1.0234375, "learning_rate": 3.887766524044229e-05, "loss": 0.128, "step": 28507 }, { "epoch": 20.884981684981685, "grad_norm": 0.875, "learning_rate": 3.887433469087724e-05, "loss": 0.1038, "step": 28508 }, { "epoch": 20.885714285714286, "grad_norm": 0.62890625, "learning_rate": 3.887100419701456e-05, "loss": 0.1064, "step": 28509 }, { "epoch": 20.886446886446887, "grad_norm": 1.6015625, "learning_rate": 3.8867673758870234e-05, "loss": 0.1263, "step": 28510 }, { "epoch": 20.887179487179488, "grad_norm": 1.359375, "learning_rate": 3.886434337646016e-05, "loss": 0.1196, "step": 28511 }, { "epoch": 20.88791208791209, "grad_norm": 0.87109375, "learning_rate": 3.886101304980037e-05, "loss": 0.1207, "step": 28512 }, { "epoch": 20.88864468864469, "grad_norm": 1.015625, "learning_rate": 3.885768277890679e-05, "loss": 0.1214, "step": 28513 }, { "epoch": 20.88937728937729, "grad_norm": 1.1953125, "learning_rate": 3.885435256379536e-05, "loss": 0.1539, "step": 28514 }, { "epoch": 20.89010989010989, "grad_norm": 0.90625, "learning_rate": 3.885102240448207e-05, "loss": 0.1401, "step": 28515 }, { "epoch": 20.89084249084249, "grad_norm": 1.0078125, "learning_rate": 3.8847692300982826e-05, "loss": 0.1197, "step": 28516 }, { "epoch": 20.891575091575092, "grad_norm": 0.8515625, "learning_rate": 3.884436225331363e-05, "loss": 0.1188, "step": 28517 }, { "epoch": 20.892307692307693, "grad_norm": 0.8203125, "learning_rate": 3.8841032261490425e-05, "loss": 0.1297, "step": 28518 }, { "epoch": 20.893040293040293, "grad_norm": 0.9921875, "learning_rate": 3.883770232552915e-05, "loss": 0.13, "step": 28519 }, { "epoch": 20.893772893772894, "grad_norm": 1.0625, "learning_rate": 3.8834372445445776e-05, "loss": 0.1091, "step": 28520 }, { "epoch": 20.894505494505495, "grad_norm": 1.125, "learning_rate": 3.883104262125624e-05, "loss": 0.1251, "step": 28521 }, { "epoch": 20.895238095238096, "grad_norm": 1.28125, "learning_rate": 3.882771285297653e-05, "loss": 0.1818, "step": 28522 }, { "epoch": 20.895970695970696, "grad_norm": 1.34375, "learning_rate": 3.882438314062258e-05, "loss": 0.1166, "step": 28523 }, { "epoch": 20.896703296703297, "grad_norm": 1.734375, "learning_rate": 3.882105348421034e-05, "loss": 0.1206, "step": 28524 }, { "epoch": 20.897435897435898, "grad_norm": 1.1015625, "learning_rate": 3.8817723883755765e-05, "loss": 0.1164, "step": 28525 }, { "epoch": 20.8981684981685, "grad_norm": 1.125, "learning_rate": 3.881439433927479e-05, "loss": 0.1082, "step": 28526 }, { "epoch": 20.8989010989011, "grad_norm": 0.92578125, "learning_rate": 3.8811064850783406e-05, "loss": 0.1397, "step": 28527 }, { "epoch": 20.8996336996337, "grad_norm": 0.97265625, "learning_rate": 3.880773541829756e-05, "loss": 0.1387, "step": 28528 }, { "epoch": 20.9003663003663, "grad_norm": 1.421875, "learning_rate": 3.880440604183318e-05, "loss": 0.1114, "step": 28529 }, { "epoch": 20.9010989010989, "grad_norm": 0.7734375, "learning_rate": 3.880107672140623e-05, "loss": 0.1232, "step": 28530 }, { "epoch": 20.901831501831502, "grad_norm": 1.1953125, "learning_rate": 3.879774745703266e-05, "loss": 0.1557, "step": 28531 }, { "epoch": 20.902564102564103, "grad_norm": 0.87109375, "learning_rate": 3.879441824872842e-05, "loss": 0.1011, "step": 28532 }, { "epoch": 20.903296703296704, "grad_norm": 0.81640625, "learning_rate": 3.879108909650946e-05, "loss": 0.1472, "step": 28533 }, { "epoch": 20.904029304029304, "grad_norm": 0.76171875, "learning_rate": 3.878776000039175e-05, "loss": 0.1243, "step": 28534 }, { "epoch": 20.904761904761905, "grad_norm": 1.0625, "learning_rate": 3.878443096039123e-05, "loss": 0.1404, "step": 28535 }, { "epoch": 20.905494505494506, "grad_norm": 0.9609375, "learning_rate": 3.8781101976523844e-05, "loss": 0.1064, "step": 28536 }, { "epoch": 20.906227106227107, "grad_norm": 0.76953125, "learning_rate": 3.877777304880552e-05, "loss": 0.1114, "step": 28537 }, { "epoch": 20.906959706959707, "grad_norm": 1.1796875, "learning_rate": 3.877444417725226e-05, "loss": 0.1295, "step": 28538 }, { "epoch": 20.907692307692308, "grad_norm": 1.453125, "learning_rate": 3.8771115361879987e-05, "loss": 0.1517, "step": 28539 }, { "epoch": 20.90842490842491, "grad_norm": 0.96484375, "learning_rate": 3.8767786602704654e-05, "loss": 0.1367, "step": 28540 }, { "epoch": 20.90915750915751, "grad_norm": 1.3125, "learning_rate": 3.87644578997422e-05, "loss": 0.1495, "step": 28541 }, { "epoch": 20.90989010989011, "grad_norm": 0.80859375, "learning_rate": 3.876112925300856e-05, "loss": 0.1034, "step": 28542 }, { "epoch": 20.91062271062271, "grad_norm": 1.234375, "learning_rate": 3.8757800662519725e-05, "loss": 0.1418, "step": 28543 }, { "epoch": 20.91135531135531, "grad_norm": 1.109375, "learning_rate": 3.8754472128291624e-05, "loss": 0.1279, "step": 28544 }, { "epoch": 20.912087912087912, "grad_norm": 1.09375, "learning_rate": 3.8751143650340194e-05, "loss": 0.1037, "step": 28545 }, { "epoch": 20.912820512820513, "grad_norm": 1.828125, "learning_rate": 3.87478152286814e-05, "loss": 0.1136, "step": 28546 }, { "epoch": 20.913553113553114, "grad_norm": 1.2265625, "learning_rate": 3.874448686333116e-05, "loss": 0.1197, "step": 28547 }, { "epoch": 20.914285714285715, "grad_norm": 1.0234375, "learning_rate": 3.874115855430547e-05, "loss": 0.1143, "step": 28548 }, { "epoch": 20.915018315018315, "grad_norm": 0.78515625, "learning_rate": 3.8737830301620236e-05, "loss": 0.097, "step": 28549 }, { "epoch": 20.915750915750916, "grad_norm": 1.1640625, "learning_rate": 3.8734502105291425e-05, "loss": 0.1127, "step": 28550 }, { "epoch": 20.916483516483517, "grad_norm": 1.2421875, "learning_rate": 3.873117396533498e-05, "loss": 0.1116, "step": 28551 }, { "epoch": 20.917216117216118, "grad_norm": 0.90625, "learning_rate": 3.872784588176684e-05, "loss": 0.1086, "step": 28552 }, { "epoch": 20.91794871794872, "grad_norm": 1.1875, "learning_rate": 3.872451785460294e-05, "loss": 0.1715, "step": 28553 }, { "epoch": 20.91868131868132, "grad_norm": 1.0390625, "learning_rate": 3.872118988385926e-05, "loss": 0.124, "step": 28554 }, { "epoch": 20.91941391941392, "grad_norm": 1.03125, "learning_rate": 3.871786196955172e-05, "loss": 0.1058, "step": 28555 }, { "epoch": 20.92014652014652, "grad_norm": 0.57421875, "learning_rate": 3.871453411169626e-05, "loss": 0.117, "step": 28556 }, { "epoch": 20.92087912087912, "grad_norm": 1.0859375, "learning_rate": 3.871120631030885e-05, "loss": 0.1226, "step": 28557 }, { "epoch": 20.921611721611722, "grad_norm": 1.1484375, "learning_rate": 3.870787856540541e-05, "loss": 0.1099, "step": 28558 }, { "epoch": 20.922344322344323, "grad_norm": 1.015625, "learning_rate": 3.870455087700189e-05, "loss": 0.1173, "step": 28559 }, { "epoch": 20.923076923076923, "grad_norm": 0.5625, "learning_rate": 3.870122324511426e-05, "loss": 0.105, "step": 28560 }, { "epoch": 20.923809523809524, "grad_norm": 1.0625, "learning_rate": 3.869789566975844e-05, "loss": 0.1176, "step": 28561 }, { "epoch": 20.924542124542125, "grad_norm": 1.328125, "learning_rate": 3.869456815095037e-05, "loss": 0.1274, "step": 28562 }, { "epoch": 20.925274725274726, "grad_norm": 1.1953125, "learning_rate": 3.8691240688705985e-05, "loss": 0.0972, "step": 28563 }, { "epoch": 20.926007326007326, "grad_norm": 1.015625, "learning_rate": 3.868791328304126e-05, "loss": 0.1363, "step": 28564 }, { "epoch": 20.926739926739927, "grad_norm": 1.2265625, "learning_rate": 3.8684585933972115e-05, "loss": 0.1392, "step": 28565 }, { "epoch": 20.927472527472528, "grad_norm": 1.140625, "learning_rate": 3.8681258641514514e-05, "loss": 0.136, "step": 28566 }, { "epoch": 20.92820512820513, "grad_norm": 1.1015625, "learning_rate": 3.867793140568437e-05, "loss": 0.1152, "step": 28567 }, { "epoch": 20.92893772893773, "grad_norm": 0.87890625, "learning_rate": 3.8674604226497615e-05, "loss": 0.1187, "step": 28568 }, { "epoch": 20.92967032967033, "grad_norm": 1.5078125, "learning_rate": 3.867127710397023e-05, "loss": 0.102, "step": 28569 }, { "epoch": 20.93040293040293, "grad_norm": 0.796875, "learning_rate": 3.866795003811815e-05, "loss": 0.1341, "step": 28570 }, { "epoch": 20.93113553113553, "grad_norm": 0.97265625, "learning_rate": 3.866462302895729e-05, "loss": 0.1344, "step": 28571 }, { "epoch": 20.931868131868132, "grad_norm": 1.2890625, "learning_rate": 3.866129607650362e-05, "loss": 0.1089, "step": 28572 }, { "epoch": 20.932600732600733, "grad_norm": 0.71484375, "learning_rate": 3.865796918077304e-05, "loss": 0.1531, "step": 28573 }, { "epoch": 20.933333333333334, "grad_norm": 0.97265625, "learning_rate": 3.865464234178155e-05, "loss": 0.1269, "step": 28574 }, { "epoch": 20.934065934065934, "grad_norm": 1.0078125, "learning_rate": 3.865131555954505e-05, "loss": 0.1264, "step": 28575 }, { "epoch": 20.934798534798535, "grad_norm": 1.296875, "learning_rate": 3.864798883407949e-05, "loss": 0.1427, "step": 28576 }, { "epoch": 20.935531135531136, "grad_norm": 0.8125, "learning_rate": 3.86446621654008e-05, "loss": 0.1213, "step": 28577 }, { "epoch": 20.936263736263736, "grad_norm": 0.83984375, "learning_rate": 3.8641335553524926e-05, "loss": 0.1253, "step": 28578 }, { "epoch": 20.936996336996337, "grad_norm": 1.234375, "learning_rate": 3.863800899846779e-05, "loss": 0.1213, "step": 28579 }, { "epoch": 20.937728937728938, "grad_norm": 1.265625, "learning_rate": 3.863468250024536e-05, "loss": 0.1239, "step": 28580 }, { "epoch": 20.93846153846154, "grad_norm": 0.69921875, "learning_rate": 3.863135605887357e-05, "loss": 0.1334, "step": 28581 }, { "epoch": 20.93919413919414, "grad_norm": 0.80859375, "learning_rate": 3.862802967436835e-05, "loss": 0.1135, "step": 28582 }, { "epoch": 20.93992673992674, "grad_norm": 1.3671875, "learning_rate": 3.862470334674562e-05, "loss": 0.1492, "step": 28583 }, { "epoch": 20.94065934065934, "grad_norm": 1.2890625, "learning_rate": 3.862137707602134e-05, "loss": 0.1085, "step": 28584 }, { "epoch": 20.94139194139194, "grad_norm": 1.0859375, "learning_rate": 3.8618050862211444e-05, "loss": 0.1274, "step": 28585 }, { "epoch": 20.942124542124542, "grad_norm": 1.296875, "learning_rate": 3.8614724705331864e-05, "loss": 0.1387, "step": 28586 }, { "epoch": 20.942857142857143, "grad_norm": 0.69140625, "learning_rate": 3.861139860539855e-05, "loss": 0.1163, "step": 28587 }, { "epoch": 20.943589743589744, "grad_norm": 1.4296875, "learning_rate": 3.860807256242743e-05, "loss": 0.1496, "step": 28588 }, { "epoch": 20.944322344322345, "grad_norm": 0.76171875, "learning_rate": 3.8604746576434414e-05, "loss": 0.1263, "step": 28589 }, { "epoch": 20.945054945054945, "grad_norm": 1.171875, "learning_rate": 3.860142064743548e-05, "loss": 0.1212, "step": 28590 }, { "epoch": 20.945787545787546, "grad_norm": 1.2421875, "learning_rate": 3.859809477544655e-05, "loss": 0.1024, "step": 28591 }, { "epoch": 20.946520146520147, "grad_norm": 1.0625, "learning_rate": 3.859476896048355e-05, "loss": 0.1456, "step": 28592 }, { "epoch": 20.947252747252747, "grad_norm": 1.3359375, "learning_rate": 3.859144320256243e-05, "loss": 0.117, "step": 28593 }, { "epoch": 20.947985347985348, "grad_norm": 1.1875, "learning_rate": 3.858811750169908e-05, "loss": 0.1386, "step": 28594 }, { "epoch": 20.94871794871795, "grad_norm": 0.75, "learning_rate": 3.85847918579095e-05, "loss": 0.1039, "step": 28595 }, { "epoch": 20.94945054945055, "grad_norm": 1.2890625, "learning_rate": 3.858146627120958e-05, "loss": 0.1293, "step": 28596 }, { "epoch": 20.95018315018315, "grad_norm": 1.046875, "learning_rate": 3.8578140741615284e-05, "loss": 0.1362, "step": 28597 }, { "epoch": 20.95091575091575, "grad_norm": 1.4375, "learning_rate": 3.8574815269142527e-05, "loss": 0.1437, "step": 28598 }, { "epoch": 20.951648351648352, "grad_norm": 0.921875, "learning_rate": 3.8571489853807224e-05, "loss": 0.1241, "step": 28599 }, { "epoch": 20.952380952380953, "grad_norm": 1.609375, "learning_rate": 3.8568164495625347e-05, "loss": 0.1243, "step": 28600 }, { "epoch": 20.953113553113553, "grad_norm": 1.3203125, "learning_rate": 3.856483919461281e-05, "loss": 0.1419, "step": 28601 }, { "epoch": 20.953846153846154, "grad_norm": 1.296875, "learning_rate": 3.856151395078555e-05, "loss": 0.1013, "step": 28602 }, { "epoch": 20.954578754578755, "grad_norm": 1.1953125, "learning_rate": 3.8558188764159494e-05, "loss": 0.1068, "step": 28603 }, { "epoch": 20.955311355311355, "grad_norm": 1.4375, "learning_rate": 3.855486363475058e-05, "loss": 0.1429, "step": 28604 }, { "epoch": 20.956043956043956, "grad_norm": 1.4296875, "learning_rate": 3.85515385625747e-05, "loss": 0.1313, "step": 28605 }, { "epoch": 20.956776556776557, "grad_norm": 1.3203125, "learning_rate": 3.854821354764785e-05, "loss": 0.1087, "step": 28606 }, { "epoch": 20.957509157509158, "grad_norm": 1.046875, "learning_rate": 3.854488858998594e-05, "loss": 0.1311, "step": 28607 }, { "epoch": 20.95824175824176, "grad_norm": 1.34375, "learning_rate": 3.8541563689604883e-05, "loss": 0.1089, "step": 28608 }, { "epoch": 20.95897435897436, "grad_norm": 1.0625, "learning_rate": 3.853823884652062e-05, "loss": 0.1344, "step": 28609 }, { "epoch": 20.95970695970696, "grad_norm": 1.2109375, "learning_rate": 3.8534914060749075e-05, "loss": 0.1242, "step": 28610 }, { "epoch": 20.96043956043956, "grad_norm": 1.453125, "learning_rate": 3.853158933230619e-05, "loss": 0.1254, "step": 28611 }, { "epoch": 20.96117216117216, "grad_norm": 1.046875, "learning_rate": 3.8528264661207895e-05, "loss": 0.1052, "step": 28612 }, { "epoch": 20.961904761904762, "grad_norm": 1.046875, "learning_rate": 3.8524940047470115e-05, "loss": 0.1735, "step": 28613 }, { "epoch": 20.962637362637363, "grad_norm": 0.96875, "learning_rate": 3.852161549110878e-05, "loss": 0.1117, "step": 28614 }, { "epoch": 20.963369963369964, "grad_norm": 1.0859375, "learning_rate": 3.8518290992139794e-05, "loss": 0.1263, "step": 28615 }, { "epoch": 20.964102564102564, "grad_norm": 0.8984375, "learning_rate": 3.851496655057913e-05, "loss": 0.1304, "step": 28616 }, { "epoch": 20.964835164835165, "grad_norm": 1.1796875, "learning_rate": 3.851164216644269e-05, "loss": 0.1344, "step": 28617 }, { "epoch": 20.965567765567766, "grad_norm": 1.2265625, "learning_rate": 3.850831783974643e-05, "loss": 0.1342, "step": 28618 }, { "epoch": 20.966300366300366, "grad_norm": 1.0234375, "learning_rate": 3.8504993570506233e-05, "loss": 0.1149, "step": 28619 }, { "epoch": 20.967032967032967, "grad_norm": 0.8515625, "learning_rate": 3.850166935873804e-05, "loss": 0.0942, "step": 28620 }, { "epoch": 20.967765567765568, "grad_norm": 0.68359375, "learning_rate": 3.84983452044578e-05, "loss": 0.1314, "step": 28621 }, { "epoch": 20.96849816849817, "grad_norm": 1.5234375, "learning_rate": 3.849502110768143e-05, "loss": 0.1155, "step": 28622 }, { "epoch": 20.96923076923077, "grad_norm": 1.3359375, "learning_rate": 3.8491697068424854e-05, "loss": 0.1009, "step": 28623 }, { "epoch": 20.96996336996337, "grad_norm": 0.88671875, "learning_rate": 3.8488373086704e-05, "loss": 0.1106, "step": 28624 }, { "epoch": 20.97069597069597, "grad_norm": 1.453125, "learning_rate": 3.8485049162534776e-05, "loss": 0.16, "step": 28625 }, { "epoch": 20.97142857142857, "grad_norm": 1.234375, "learning_rate": 3.848172529593315e-05, "loss": 0.1271, "step": 28626 }, { "epoch": 20.972161172161172, "grad_norm": 1.4609375, "learning_rate": 3.8478401486915016e-05, "loss": 0.1487, "step": 28627 }, { "epoch": 20.972893772893773, "grad_norm": 0.62890625, "learning_rate": 3.847507773549632e-05, "loss": 0.0961, "step": 28628 }, { "epoch": 20.973626373626374, "grad_norm": 0.85546875, "learning_rate": 3.847175404169296e-05, "loss": 0.127, "step": 28629 }, { "epoch": 20.974358974358974, "grad_norm": 1.5703125, "learning_rate": 3.846843040552088e-05, "loss": 0.1608, "step": 28630 }, { "epoch": 20.975091575091575, "grad_norm": 0.73828125, "learning_rate": 3.846510682699598e-05, "loss": 0.1059, "step": 28631 }, { "epoch": 20.975824175824176, "grad_norm": 0.90625, "learning_rate": 3.8461783306134226e-05, "loss": 0.1068, "step": 28632 }, { "epoch": 20.976556776556777, "grad_norm": 1.234375, "learning_rate": 3.845845984295151e-05, "loss": 0.1271, "step": 28633 }, { "epoch": 20.977289377289377, "grad_norm": 1.375, "learning_rate": 3.8455136437463775e-05, "loss": 0.112, "step": 28634 }, { "epoch": 20.978021978021978, "grad_norm": 1.9296875, "learning_rate": 3.845181308968693e-05, "loss": 0.1234, "step": 28635 }, { "epoch": 20.97875457875458, "grad_norm": 1.4453125, "learning_rate": 3.8448489799636886e-05, "loss": 0.1646, "step": 28636 }, { "epoch": 20.97948717948718, "grad_norm": 0.9453125, "learning_rate": 3.8445166567329606e-05, "loss": 0.1174, "step": 28637 }, { "epoch": 20.98021978021978, "grad_norm": 1.28125, "learning_rate": 3.844184339278099e-05, "loss": 0.1003, "step": 28638 }, { "epoch": 20.98095238095238, "grad_norm": 1.5703125, "learning_rate": 3.843852027600696e-05, "loss": 0.1229, "step": 28639 }, { "epoch": 20.98168498168498, "grad_norm": 1.3515625, "learning_rate": 3.8435197217023434e-05, "loss": 0.1119, "step": 28640 }, { "epoch": 20.982417582417582, "grad_norm": 1.3359375, "learning_rate": 3.8431874215846325e-05, "loss": 0.1236, "step": 28641 }, { "epoch": 20.983150183150183, "grad_norm": 0.80078125, "learning_rate": 3.842855127249159e-05, "loss": 0.0927, "step": 28642 }, { "epoch": 20.983882783882784, "grad_norm": 0.828125, "learning_rate": 3.842522838697512e-05, "loss": 0.1271, "step": 28643 }, { "epoch": 20.984615384615385, "grad_norm": 0.9453125, "learning_rate": 3.8421905559312856e-05, "loss": 0.1188, "step": 28644 }, { "epoch": 20.985347985347985, "grad_norm": 0.73046875, "learning_rate": 3.8418582789520693e-05, "loss": 0.1215, "step": 28645 }, { "epoch": 20.986080586080586, "grad_norm": 0.83984375, "learning_rate": 3.841526007761456e-05, "loss": 0.0967, "step": 28646 }, { "epoch": 20.986813186813187, "grad_norm": 1.21875, "learning_rate": 3.84119374236104e-05, "loss": 0.1298, "step": 28647 }, { "epoch": 20.987545787545788, "grad_norm": 0.92578125, "learning_rate": 3.84086148275241e-05, "loss": 0.1311, "step": 28648 }, { "epoch": 20.98827838827839, "grad_norm": 0.93359375, "learning_rate": 3.8405292289371616e-05, "loss": 0.1043, "step": 28649 }, { "epoch": 20.98901098901099, "grad_norm": 1.125, "learning_rate": 3.8401969809168844e-05, "loss": 0.1275, "step": 28650 }, { "epoch": 20.98974358974359, "grad_norm": 1.125, "learning_rate": 3.839864738693168e-05, "loss": 0.1338, "step": 28651 }, { "epoch": 20.99047619047619, "grad_norm": 1.0234375, "learning_rate": 3.839532502267609e-05, "loss": 0.1046, "step": 28652 }, { "epoch": 20.99120879120879, "grad_norm": 1.25, "learning_rate": 3.839200271641797e-05, "loss": 0.0943, "step": 28653 }, { "epoch": 20.991941391941392, "grad_norm": 0.9765625, "learning_rate": 3.838868046817325e-05, "loss": 0.1274, "step": 28654 }, { "epoch": 20.992673992673993, "grad_norm": 0.9375, "learning_rate": 3.838535827795783e-05, "loss": 0.1109, "step": 28655 }, { "epoch": 20.993406593406593, "grad_norm": 1.1640625, "learning_rate": 3.8382036145787636e-05, "loss": 0.1104, "step": 28656 }, { "epoch": 20.994139194139194, "grad_norm": 0.890625, "learning_rate": 3.837871407167856e-05, "loss": 0.1054, "step": 28657 }, { "epoch": 20.994871794871795, "grad_norm": 1.484375, "learning_rate": 3.837539205564658e-05, "loss": 0.1563, "step": 28658 }, { "epoch": 20.995604395604396, "grad_norm": 1.3828125, "learning_rate": 3.8372070097707564e-05, "loss": 0.1217, "step": 28659 }, { "epoch": 20.996336996336996, "grad_norm": 0.875, "learning_rate": 3.836874819787743e-05, "loss": 0.1136, "step": 28660 }, { "epoch": 20.997069597069597, "grad_norm": 1.0234375, "learning_rate": 3.8365426356172125e-05, "loss": 0.0898, "step": 28661 }, { "epoch": 20.997802197802198, "grad_norm": 1.09375, "learning_rate": 3.8362104572607514e-05, "loss": 0.1146, "step": 28662 }, { "epoch": 20.9985347985348, "grad_norm": 1.1484375, "learning_rate": 3.835878284719957e-05, "loss": 0.109, "step": 28663 }, { "epoch": 20.9992673992674, "grad_norm": 0.80078125, "learning_rate": 3.8355461179964187e-05, "loss": 0.12, "step": 28664 }, { "epoch": 21.0, "grad_norm": 1.15625, "learning_rate": 3.835213957091727e-05, "loss": 0.1261, "step": 28665 }, { "epoch": 21.0007326007326, "grad_norm": 1.5234375, "learning_rate": 3.834881802007475e-05, "loss": 0.1039, "step": 28666 }, { "epoch": 21.0014652014652, "grad_norm": 1.125, "learning_rate": 3.834549652745251e-05, "loss": 0.1308, "step": 28667 }, { "epoch": 21.002197802197802, "grad_norm": 1.015625, "learning_rate": 3.83421750930665e-05, "loss": 0.125, "step": 28668 }, { "epoch": 21.002930402930403, "grad_norm": 0.84765625, "learning_rate": 3.8338853716932615e-05, "loss": 0.1129, "step": 28669 }, { "epoch": 21.003663003663004, "grad_norm": 0.98046875, "learning_rate": 3.833553239906678e-05, "loss": 0.1186, "step": 28670 }, { "epoch": 21.004395604395604, "grad_norm": 0.96875, "learning_rate": 3.8332211139484895e-05, "loss": 0.12, "step": 28671 }, { "epoch": 21.005128205128205, "grad_norm": 0.71484375, "learning_rate": 3.8328889938202886e-05, "loss": 0.1141, "step": 28672 }, { "epoch": 21.005860805860806, "grad_norm": 1.0703125, "learning_rate": 3.832556879523666e-05, "loss": 0.1325, "step": 28673 }, { "epoch": 21.006593406593407, "grad_norm": 0.890625, "learning_rate": 3.832224771060213e-05, "loss": 0.1171, "step": 28674 }, { "epoch": 21.007326007326007, "grad_norm": 0.75390625, "learning_rate": 3.831892668431521e-05, "loss": 0.1125, "step": 28675 }, { "epoch": 21.008058608058608, "grad_norm": 1.90625, "learning_rate": 3.831560571639181e-05, "loss": 0.1085, "step": 28676 }, { "epoch": 21.00879120879121, "grad_norm": 0.8984375, "learning_rate": 3.831228480684783e-05, "loss": 0.1145, "step": 28677 }, { "epoch": 21.00952380952381, "grad_norm": 1.0625, "learning_rate": 3.830896395569921e-05, "loss": 0.1081, "step": 28678 }, { "epoch": 21.01025641025641, "grad_norm": 1.0546875, "learning_rate": 3.8305643162961853e-05, "loss": 0.1195, "step": 28679 }, { "epoch": 21.01098901098901, "grad_norm": 0.9921875, "learning_rate": 3.8302322428651656e-05, "loss": 0.1163, "step": 28680 }, { "epoch": 21.01172161172161, "grad_norm": 1.0078125, "learning_rate": 3.829900175278454e-05, "loss": 0.1058, "step": 28681 }, { "epoch": 21.012454212454212, "grad_norm": 1.2890625, "learning_rate": 3.829568113537641e-05, "loss": 0.1418, "step": 28682 }, { "epoch": 21.013186813186813, "grad_norm": 1.5234375, "learning_rate": 3.829236057644316e-05, "loss": 0.1301, "step": 28683 }, { "epoch": 21.013919413919414, "grad_norm": 0.93359375, "learning_rate": 3.828904007600073e-05, "loss": 0.1199, "step": 28684 }, { "epoch": 21.014652014652015, "grad_norm": 0.7578125, "learning_rate": 3.828571963406503e-05, "loss": 0.1018, "step": 28685 }, { "epoch": 21.015384615384615, "grad_norm": 0.97265625, "learning_rate": 3.828239925065194e-05, "loss": 0.116, "step": 28686 }, { "epoch": 21.016117216117216, "grad_norm": 1.34375, "learning_rate": 3.8279078925777396e-05, "loss": 0.1395, "step": 28687 }, { "epoch": 21.016849816849817, "grad_norm": 0.67578125, "learning_rate": 3.827575865945728e-05, "loss": 0.0928, "step": 28688 }, { "epoch": 21.017582417582418, "grad_norm": 1.3828125, "learning_rate": 3.827243845170753e-05, "loss": 0.1431, "step": 28689 }, { "epoch": 21.01831501831502, "grad_norm": 1.1171875, "learning_rate": 3.8269118302544046e-05, "loss": 0.1172, "step": 28690 }, { "epoch": 21.01904761904762, "grad_norm": 1.2265625, "learning_rate": 3.826579821198273e-05, "loss": 0.1325, "step": 28691 }, { "epoch": 21.01978021978022, "grad_norm": 0.94140625, "learning_rate": 3.82624781800395e-05, "loss": 0.1214, "step": 28692 }, { "epoch": 21.02051282051282, "grad_norm": 0.99609375, "learning_rate": 3.8259158206730215e-05, "loss": 0.1077, "step": 28693 }, { "epoch": 21.02124542124542, "grad_norm": 0.91796875, "learning_rate": 3.825583829207086e-05, "loss": 0.0939, "step": 28694 }, { "epoch": 21.021978021978022, "grad_norm": 1.1328125, "learning_rate": 3.82525184360773e-05, "loss": 0.1114, "step": 28695 }, { "epoch": 21.022710622710623, "grad_norm": 1.2578125, "learning_rate": 3.8249198638765436e-05, "loss": 0.1105, "step": 28696 }, { "epoch": 21.023443223443223, "grad_norm": 0.94921875, "learning_rate": 3.8245878900151175e-05, "loss": 0.1096, "step": 28697 }, { "epoch": 21.024175824175824, "grad_norm": 0.9453125, "learning_rate": 3.824255922025043e-05, "loss": 0.1069, "step": 28698 }, { "epoch": 21.024908424908425, "grad_norm": 1.1640625, "learning_rate": 3.823923959907913e-05, "loss": 0.1328, "step": 28699 }, { "epoch": 21.025641025641026, "grad_norm": 0.58203125, "learning_rate": 3.823592003665314e-05, "loss": 0.106, "step": 28700 }, { "epoch": 21.026373626373626, "grad_norm": 1.125, "learning_rate": 3.823260053298839e-05, "loss": 0.1078, "step": 28701 }, { "epoch": 21.027106227106227, "grad_norm": 1.390625, "learning_rate": 3.822928108810078e-05, "loss": 0.121, "step": 28702 }, { "epoch": 21.027838827838828, "grad_norm": 0.546875, "learning_rate": 3.82259617020062e-05, "loss": 0.1087, "step": 28703 }, { "epoch": 21.02857142857143, "grad_norm": 1.1640625, "learning_rate": 3.8222642374720576e-05, "loss": 0.1677, "step": 28704 }, { "epoch": 21.02930402930403, "grad_norm": 0.99609375, "learning_rate": 3.821932310625981e-05, "loss": 0.1355, "step": 28705 }, { "epoch": 21.03003663003663, "grad_norm": 1.5703125, "learning_rate": 3.821600389663979e-05, "loss": 0.1343, "step": 28706 }, { "epoch": 21.03076923076923, "grad_norm": 2.015625, "learning_rate": 3.8212684745876434e-05, "loss": 0.112, "step": 28707 }, { "epoch": 21.03150183150183, "grad_norm": 1.3828125, "learning_rate": 3.820936565398564e-05, "loss": 0.1464, "step": 28708 }, { "epoch": 21.032234432234432, "grad_norm": 1.3203125, "learning_rate": 3.820604662098329e-05, "loss": 0.1115, "step": 28709 }, { "epoch": 21.032967032967033, "grad_norm": 0.81640625, "learning_rate": 3.820272764688533e-05, "loss": 0.1355, "step": 28710 }, { "epoch": 21.033699633699634, "grad_norm": 1.046875, "learning_rate": 3.8199408731707617e-05, "loss": 0.1302, "step": 28711 }, { "epoch": 21.034432234432234, "grad_norm": 1.03125, "learning_rate": 3.81960898754661e-05, "loss": 0.1044, "step": 28712 }, { "epoch": 21.035164835164835, "grad_norm": 1.8671875, "learning_rate": 3.819277107817664e-05, "loss": 0.1165, "step": 28713 }, { "epoch": 21.035897435897436, "grad_norm": 0.8671875, "learning_rate": 3.818945233985514e-05, "loss": 0.1166, "step": 28714 }, { "epoch": 21.036630036630036, "grad_norm": 1.3984375, "learning_rate": 3.818613366051754e-05, "loss": 0.1221, "step": 28715 }, { "epoch": 21.037362637362637, "grad_norm": 0.69921875, "learning_rate": 3.818281504017971e-05, "loss": 0.0934, "step": 28716 }, { "epoch": 21.038095238095238, "grad_norm": 1.5, "learning_rate": 3.817949647885756e-05, "loss": 0.1443, "step": 28717 }, { "epoch": 21.03882783882784, "grad_norm": 0.74609375, "learning_rate": 3.817617797656698e-05, "loss": 0.0991, "step": 28718 }, { "epoch": 21.03956043956044, "grad_norm": 1.2890625, "learning_rate": 3.817285953332386e-05, "loss": 0.128, "step": 28719 }, { "epoch": 21.04029304029304, "grad_norm": 1.1015625, "learning_rate": 3.8169541149144145e-05, "loss": 0.1249, "step": 28720 }, { "epoch": 21.04102564102564, "grad_norm": 0.98828125, "learning_rate": 3.81662228240437e-05, "loss": 0.1153, "step": 28721 }, { "epoch": 21.04175824175824, "grad_norm": 1.140625, "learning_rate": 3.816290455803843e-05, "loss": 0.101, "step": 28722 }, { "epoch": 21.042490842490842, "grad_norm": 0.69140625, "learning_rate": 3.8159586351144225e-05, "loss": 0.1069, "step": 28723 }, { "epoch": 21.043223443223443, "grad_norm": 1.21875, "learning_rate": 3.815626820337699e-05, "loss": 0.149, "step": 28724 }, { "epoch": 21.043956043956044, "grad_norm": 0.95703125, "learning_rate": 3.8152950114752644e-05, "loss": 0.1378, "step": 28725 }, { "epoch": 21.044688644688645, "grad_norm": 1.296875, "learning_rate": 3.814963208528705e-05, "loss": 0.1399, "step": 28726 }, { "epoch": 21.045421245421245, "grad_norm": 1.390625, "learning_rate": 3.814631411499614e-05, "loss": 0.1203, "step": 28727 }, { "epoch": 21.046153846153846, "grad_norm": 0.9296875, "learning_rate": 3.8142996203895785e-05, "loss": 0.1229, "step": 28728 }, { "epoch": 21.046886446886447, "grad_norm": 0.8984375, "learning_rate": 3.813967835200188e-05, "loss": 0.1105, "step": 28729 }, { "epoch": 21.047619047619047, "grad_norm": 1.0546875, "learning_rate": 3.8136360559330346e-05, "loss": 0.1394, "step": 28730 }, { "epoch": 21.048351648351648, "grad_norm": 0.9375, "learning_rate": 3.8133042825897074e-05, "loss": 0.1101, "step": 28731 }, { "epoch": 21.04908424908425, "grad_norm": 1.25, "learning_rate": 3.812972515171795e-05, "loss": 0.1058, "step": 28732 }, { "epoch": 21.04981684981685, "grad_norm": 1.1171875, "learning_rate": 3.812640753680888e-05, "loss": 0.1264, "step": 28733 }, { "epoch": 21.05054945054945, "grad_norm": 0.734375, "learning_rate": 3.812308998118575e-05, "loss": 0.0975, "step": 28734 }, { "epoch": 21.05128205128205, "grad_norm": 0.81640625, "learning_rate": 3.811977248486444e-05, "loss": 0.1324, "step": 28735 }, { "epoch": 21.052014652014652, "grad_norm": 1.515625, "learning_rate": 3.8116455047860875e-05, "loss": 0.1305, "step": 28736 }, { "epoch": 21.052747252747253, "grad_norm": 1.125, "learning_rate": 3.8113137670190934e-05, "loss": 0.117, "step": 28737 }, { "epoch": 21.053479853479853, "grad_norm": 1.0703125, "learning_rate": 3.810982035187053e-05, "loss": 0.1354, "step": 28738 }, { "epoch": 21.054212454212454, "grad_norm": 1.1640625, "learning_rate": 3.8106503092915543e-05, "loss": 0.1171, "step": 28739 }, { "epoch": 21.054945054945055, "grad_norm": 0.73828125, "learning_rate": 3.810318589334185e-05, "loss": 0.1087, "step": 28740 }, { "epoch": 21.055677655677655, "grad_norm": 1.2265625, "learning_rate": 3.8099868753165376e-05, "loss": 0.1062, "step": 28741 }, { "epoch": 21.056410256410256, "grad_norm": 1.09375, "learning_rate": 3.8096551672402004e-05, "loss": 0.1182, "step": 28742 }, { "epoch": 21.057142857142857, "grad_norm": 1.5625, "learning_rate": 3.809323465106763e-05, "loss": 0.1399, "step": 28743 }, { "epoch": 21.057875457875458, "grad_norm": 0.921875, "learning_rate": 3.8089917689178135e-05, "loss": 0.0884, "step": 28744 }, { "epoch": 21.05860805860806, "grad_norm": 0.84375, "learning_rate": 3.8086600786749405e-05, "loss": 0.1283, "step": 28745 }, { "epoch": 21.05934065934066, "grad_norm": 0.92578125, "learning_rate": 3.808328394379736e-05, "loss": 0.1292, "step": 28746 }, { "epoch": 21.06007326007326, "grad_norm": 1.046875, "learning_rate": 3.807996716033788e-05, "loss": 0.1098, "step": 28747 }, { "epoch": 21.06080586080586, "grad_norm": 1.0234375, "learning_rate": 3.8076650436386866e-05, "loss": 0.1337, "step": 28748 }, { "epoch": 21.06153846153846, "grad_norm": 1.1796875, "learning_rate": 3.807333377196017e-05, "loss": 0.1241, "step": 28749 }, { "epoch": 21.062271062271062, "grad_norm": 1.1875, "learning_rate": 3.807001716707372e-05, "loss": 0.1063, "step": 28750 }, { "epoch": 21.063003663003663, "grad_norm": 0.7734375, "learning_rate": 3.80667006217434e-05, "loss": 0.1008, "step": 28751 }, { "epoch": 21.063736263736264, "grad_norm": 0.91796875, "learning_rate": 3.80633841359851e-05, "loss": 0.1187, "step": 28752 }, { "epoch": 21.064468864468864, "grad_norm": 0.85546875, "learning_rate": 3.8060067709814716e-05, "loss": 0.1247, "step": 28753 }, { "epoch": 21.065201465201465, "grad_norm": 1.2578125, "learning_rate": 3.805675134324813e-05, "loss": 0.135, "step": 28754 }, { "epoch": 21.065934065934066, "grad_norm": 1.4765625, "learning_rate": 3.8053435036301216e-05, "loss": 0.1655, "step": 28755 }, { "epoch": 21.066666666666666, "grad_norm": 1.375, "learning_rate": 3.80501187889899e-05, "loss": 0.1157, "step": 28756 }, { "epoch": 21.067399267399267, "grad_norm": 1.125, "learning_rate": 3.804680260133004e-05, "loss": 0.1122, "step": 28757 }, { "epoch": 21.068131868131868, "grad_norm": 1.109375, "learning_rate": 3.8043486473337546e-05, "loss": 0.1287, "step": 28758 }, { "epoch": 21.06886446886447, "grad_norm": 0.91796875, "learning_rate": 3.8040170405028306e-05, "loss": 0.1228, "step": 28759 }, { "epoch": 21.06959706959707, "grad_norm": 0.87890625, "learning_rate": 3.803685439641818e-05, "loss": 0.1339, "step": 28760 }, { "epoch": 21.07032967032967, "grad_norm": 0.7734375, "learning_rate": 3.803353844752307e-05, "loss": 0.1266, "step": 28761 }, { "epoch": 21.07106227106227, "grad_norm": 0.9765625, "learning_rate": 3.8030222558358895e-05, "loss": 0.1447, "step": 28762 }, { "epoch": 21.07179487179487, "grad_norm": 0.9453125, "learning_rate": 3.80269067289415e-05, "loss": 0.1328, "step": 28763 }, { "epoch": 21.072527472527472, "grad_norm": 0.9921875, "learning_rate": 3.80235909592868e-05, "loss": 0.1072, "step": 28764 }, { "epoch": 21.073260073260073, "grad_norm": 1.4609375, "learning_rate": 3.802027524941067e-05, "loss": 0.1302, "step": 28765 }, { "epoch": 21.073992673992674, "grad_norm": 1.140625, "learning_rate": 3.801695959932898e-05, "loss": 0.1223, "step": 28766 }, { "epoch": 21.074725274725274, "grad_norm": 0.96875, "learning_rate": 3.8013644009057654e-05, "loss": 0.0829, "step": 28767 }, { "epoch": 21.075457875457875, "grad_norm": 1.3203125, "learning_rate": 3.801032847861255e-05, "loss": 0.0947, "step": 28768 }, { "epoch": 21.076190476190476, "grad_norm": 1.328125, "learning_rate": 3.800701300800958e-05, "loss": 0.1262, "step": 28769 }, { "epoch": 21.076923076923077, "grad_norm": 0.86328125, "learning_rate": 3.8003697597264603e-05, "loss": 0.1126, "step": 28770 }, { "epoch": 21.077655677655677, "grad_norm": 1.03125, "learning_rate": 3.800038224639349e-05, "loss": 0.0954, "step": 28771 }, { "epoch": 21.078388278388278, "grad_norm": 0.8359375, "learning_rate": 3.799706695541218e-05, "loss": 0.1218, "step": 28772 }, { "epoch": 21.07912087912088, "grad_norm": 1.1328125, "learning_rate": 3.799375172433651e-05, "loss": 0.1302, "step": 28773 }, { "epoch": 21.07985347985348, "grad_norm": 1.3359375, "learning_rate": 3.799043655318239e-05, "loss": 0.1522, "step": 28774 }, { "epoch": 21.08058608058608, "grad_norm": 1.390625, "learning_rate": 3.798712144196568e-05, "loss": 0.1257, "step": 28775 }, { "epoch": 21.08131868131868, "grad_norm": 1.078125, "learning_rate": 3.7983806390702285e-05, "loss": 0.1373, "step": 28776 }, { "epoch": 21.08205128205128, "grad_norm": 0.96875, "learning_rate": 3.798049139940809e-05, "loss": 0.0978, "step": 28777 }, { "epoch": 21.082783882783882, "grad_norm": 1.8984375, "learning_rate": 3.797717646809896e-05, "loss": 0.1463, "step": 28778 }, { "epoch": 21.083516483516483, "grad_norm": 1.1796875, "learning_rate": 3.7973861596790804e-05, "loss": 0.1549, "step": 28779 }, { "epoch": 21.084249084249084, "grad_norm": 0.96875, "learning_rate": 3.7970546785499484e-05, "loss": 0.1332, "step": 28780 }, { "epoch": 21.084981684981685, "grad_norm": 0.79296875, "learning_rate": 3.796723203424089e-05, "loss": 0.1015, "step": 28781 }, { "epoch": 21.085714285714285, "grad_norm": 0.92578125, "learning_rate": 3.796391734303088e-05, "loss": 0.1451, "step": 28782 }, { "epoch": 21.086446886446886, "grad_norm": 0.7578125, "learning_rate": 3.7960602711885374e-05, "loss": 0.0967, "step": 28783 }, { "epoch": 21.087179487179487, "grad_norm": 0.91015625, "learning_rate": 3.795728814082024e-05, "loss": 0.114, "step": 28784 }, { "epoch": 21.087912087912088, "grad_norm": 0.96484375, "learning_rate": 3.795397362985136e-05, "loss": 0.1509, "step": 28785 }, { "epoch": 21.08864468864469, "grad_norm": 0.875, "learning_rate": 3.79506591789946e-05, "loss": 0.1262, "step": 28786 }, { "epoch": 21.08937728937729, "grad_norm": 1.140625, "learning_rate": 3.7947344788265844e-05, "loss": 0.1357, "step": 28787 }, { "epoch": 21.09010989010989, "grad_norm": 1.03125, "learning_rate": 3.7944030457680995e-05, "loss": 0.139, "step": 28788 }, { "epoch": 21.09084249084249, "grad_norm": 1.828125, "learning_rate": 3.794071618725591e-05, "loss": 0.1302, "step": 28789 }, { "epoch": 21.09157509157509, "grad_norm": 1.375, "learning_rate": 3.7937401977006483e-05, "loss": 0.1426, "step": 28790 }, { "epoch": 21.092307692307692, "grad_norm": 1.1015625, "learning_rate": 3.793408782694858e-05, "loss": 0.113, "step": 28791 }, { "epoch": 21.093040293040293, "grad_norm": 1.171875, "learning_rate": 3.793077373709808e-05, "loss": 0.1377, "step": 28792 }, { "epoch": 21.093772893772893, "grad_norm": 1.1953125, "learning_rate": 3.7927459707470886e-05, "loss": 0.1405, "step": 28793 }, { "epoch": 21.094505494505494, "grad_norm": 1.203125, "learning_rate": 3.792414573808286e-05, "loss": 0.113, "step": 28794 }, { "epoch": 21.095238095238095, "grad_norm": 0.9453125, "learning_rate": 3.7920831828949874e-05, "loss": 0.1018, "step": 28795 }, { "epoch": 21.095970695970696, "grad_norm": 0.640625, "learning_rate": 3.791751798008782e-05, "loss": 0.1134, "step": 28796 }, { "epoch": 21.096703296703296, "grad_norm": 1.0390625, "learning_rate": 3.7914204191512545e-05, "loss": 0.1152, "step": 28797 }, { "epoch": 21.097435897435897, "grad_norm": 1.1640625, "learning_rate": 3.791089046323997e-05, "loss": 0.1505, "step": 28798 }, { "epoch": 21.098168498168498, "grad_norm": 0.91015625, "learning_rate": 3.7907576795285946e-05, "loss": 0.1, "step": 28799 }, { "epoch": 21.0989010989011, "grad_norm": 0.65234375, "learning_rate": 3.7904263187666364e-05, "loss": 0.1282, "step": 28800 }, { "epoch": 21.0996336996337, "grad_norm": 1.171875, "learning_rate": 3.7900949640397073e-05, "loss": 0.0843, "step": 28801 }, { "epoch": 21.1003663003663, "grad_norm": 1.265625, "learning_rate": 3.7897636153493984e-05, "loss": 0.1268, "step": 28802 }, { "epoch": 21.1010989010989, "grad_norm": 0.9765625, "learning_rate": 3.789432272697294e-05, "loss": 0.139, "step": 28803 }, { "epoch": 21.1018315018315, "grad_norm": 1.0234375, "learning_rate": 3.789100936084986e-05, "loss": 0.1048, "step": 28804 }, { "epoch": 21.102564102564102, "grad_norm": 1.3671875, "learning_rate": 3.788769605514057e-05, "loss": 0.116, "step": 28805 }, { "epoch": 21.103296703296703, "grad_norm": 0.98046875, "learning_rate": 3.788438280986099e-05, "loss": 0.133, "step": 28806 }, { "epoch": 21.104029304029304, "grad_norm": 0.78515625, "learning_rate": 3.788106962502696e-05, "loss": 0.1201, "step": 28807 }, { "epoch": 21.104761904761904, "grad_norm": 1.421875, "learning_rate": 3.787775650065436e-05, "loss": 0.1356, "step": 28808 }, { "epoch": 21.105494505494505, "grad_norm": 1.203125, "learning_rate": 3.787444343675909e-05, "loss": 0.1231, "step": 28809 }, { "epoch": 21.106227106227106, "grad_norm": 0.65625, "learning_rate": 3.7871130433357e-05, "loss": 0.111, "step": 28810 }, { "epoch": 21.106959706959707, "grad_norm": 0.6953125, "learning_rate": 3.786781749046397e-05, "loss": 0.1412, "step": 28811 }, { "epoch": 21.107692307692307, "grad_norm": 0.54296875, "learning_rate": 3.786450460809586e-05, "loss": 0.1144, "step": 28812 }, { "epoch": 21.108424908424908, "grad_norm": 1.0390625, "learning_rate": 3.786119178626857e-05, "loss": 0.1038, "step": 28813 }, { "epoch": 21.10915750915751, "grad_norm": 1.0078125, "learning_rate": 3.785787902499796e-05, "loss": 0.108, "step": 28814 }, { "epoch": 21.10989010989011, "grad_norm": 1.5390625, "learning_rate": 3.785456632429989e-05, "loss": 0.1217, "step": 28815 }, { "epoch": 21.11062271062271, "grad_norm": 0.70703125, "learning_rate": 3.7851253684190266e-05, "loss": 0.0953, "step": 28816 }, { "epoch": 21.11135531135531, "grad_norm": 1.0703125, "learning_rate": 3.7847941104684917e-05, "loss": 0.1006, "step": 28817 }, { "epoch": 21.11208791208791, "grad_norm": 0.91796875, "learning_rate": 3.784462858579973e-05, "loss": 0.1247, "step": 28818 }, { "epoch": 21.112820512820512, "grad_norm": 0.53125, "learning_rate": 3.78413161275506e-05, "loss": 0.122, "step": 28819 }, { "epoch": 21.113553113553113, "grad_norm": 0.921875, "learning_rate": 3.783800372995338e-05, "loss": 0.1326, "step": 28820 }, { "epoch": 21.114285714285714, "grad_norm": 1.3984375, "learning_rate": 3.783469139302394e-05, "loss": 0.1104, "step": 28821 }, { "epoch": 21.115018315018315, "grad_norm": 1.234375, "learning_rate": 3.783137911677815e-05, "loss": 0.1403, "step": 28822 }, { "epoch": 21.115750915750915, "grad_norm": 0.9375, "learning_rate": 3.7828066901231866e-05, "loss": 0.1057, "step": 28823 }, { "epoch": 21.116483516483516, "grad_norm": 0.984375, "learning_rate": 3.7824754746400986e-05, "loss": 0.1191, "step": 28824 }, { "epoch": 21.117216117216117, "grad_norm": 0.8828125, "learning_rate": 3.782144265230137e-05, "loss": 0.1306, "step": 28825 }, { "epoch": 21.117948717948718, "grad_norm": 1.6328125, "learning_rate": 3.7818130618948874e-05, "loss": 0.1276, "step": 28826 }, { "epoch": 21.11868131868132, "grad_norm": 1.2265625, "learning_rate": 3.7814818646359395e-05, "loss": 0.1, "step": 28827 }, { "epoch": 21.11941391941392, "grad_norm": 0.9609375, "learning_rate": 3.7811506734548766e-05, "loss": 0.1194, "step": 28828 }, { "epoch": 21.12014652014652, "grad_norm": 0.84765625, "learning_rate": 3.780819488353288e-05, "loss": 0.1035, "step": 28829 }, { "epoch": 21.12087912087912, "grad_norm": 1.4453125, "learning_rate": 3.78048830933276e-05, "loss": 0.1203, "step": 28830 }, { "epoch": 21.12161172161172, "grad_norm": 1.2265625, "learning_rate": 3.7801571363948795e-05, "loss": 0.1464, "step": 28831 }, { "epoch": 21.122344322344322, "grad_norm": 1.1796875, "learning_rate": 3.779825969541234e-05, "loss": 0.1489, "step": 28832 }, { "epoch": 21.123076923076923, "grad_norm": 1.34375, "learning_rate": 3.7794948087734085e-05, "loss": 0.1177, "step": 28833 }, { "epoch": 21.123809523809523, "grad_norm": 1.0390625, "learning_rate": 3.779163654092988e-05, "loss": 0.1507, "step": 28834 }, { "epoch": 21.124542124542124, "grad_norm": 0.8046875, "learning_rate": 3.778832505501564e-05, "loss": 0.124, "step": 28835 }, { "epoch": 21.125274725274725, "grad_norm": 1.234375, "learning_rate": 3.778501363000721e-05, "loss": 0.106, "step": 28836 }, { "epoch": 21.126007326007326, "grad_norm": 1.5390625, "learning_rate": 3.7781702265920454e-05, "loss": 0.1386, "step": 28837 }, { "epoch": 21.126739926739926, "grad_norm": 1.625, "learning_rate": 3.777839096277122e-05, "loss": 0.0836, "step": 28838 }, { "epoch": 21.127472527472527, "grad_norm": 1.359375, "learning_rate": 3.77750797205754e-05, "loss": 0.1273, "step": 28839 }, { "epoch": 21.128205128205128, "grad_norm": 1.09375, "learning_rate": 3.7771768539348855e-05, "loss": 0.1403, "step": 28840 }, { "epoch": 21.12893772893773, "grad_norm": 0.6953125, "learning_rate": 3.776845741910744e-05, "loss": 0.1137, "step": 28841 }, { "epoch": 21.12967032967033, "grad_norm": 1.171875, "learning_rate": 3.776514635986704e-05, "loss": 0.1224, "step": 28842 }, { "epoch": 21.13040293040293, "grad_norm": 0.796875, "learning_rate": 3.776183536164349e-05, "loss": 0.119, "step": 28843 }, { "epoch": 21.13113553113553, "grad_norm": 1.0703125, "learning_rate": 3.775852442445265e-05, "loss": 0.1363, "step": 28844 }, { "epoch": 21.13186813186813, "grad_norm": 1.5625, "learning_rate": 3.775521354831043e-05, "loss": 0.1137, "step": 28845 }, { "epoch": 21.132600732600732, "grad_norm": 1.4140625, "learning_rate": 3.775190273323266e-05, "loss": 0.1026, "step": 28846 }, { "epoch": 21.133333333333333, "grad_norm": 1.09375, "learning_rate": 3.7748591979235215e-05, "loss": 0.1191, "step": 28847 }, { "epoch": 21.134065934065934, "grad_norm": 1.3125, "learning_rate": 3.774528128633395e-05, "loss": 0.1425, "step": 28848 }, { "epoch": 21.134798534798534, "grad_norm": 1.078125, "learning_rate": 3.774197065454471e-05, "loss": 0.1297, "step": 28849 }, { "epoch": 21.135531135531135, "grad_norm": 1.1796875, "learning_rate": 3.773866008388339e-05, "loss": 0.1183, "step": 28850 }, { "epoch": 21.136263736263736, "grad_norm": 0.890625, "learning_rate": 3.773534957436585e-05, "loss": 0.1102, "step": 28851 }, { "epoch": 21.136996336996337, "grad_norm": 1.359375, "learning_rate": 3.7732039126007925e-05, "loss": 0.1404, "step": 28852 }, { "epoch": 21.137728937728937, "grad_norm": 1.1953125, "learning_rate": 3.77287287388255e-05, "loss": 0.1439, "step": 28853 }, { "epoch": 21.138461538461538, "grad_norm": 1.1328125, "learning_rate": 3.772541841283442e-05, "loss": 0.1348, "step": 28854 }, { "epoch": 21.13919413919414, "grad_norm": 1.09375, "learning_rate": 3.772210814805055e-05, "loss": 0.1053, "step": 28855 }, { "epoch": 21.13992673992674, "grad_norm": 0.6953125, "learning_rate": 3.771879794448977e-05, "loss": 0.0984, "step": 28856 }, { "epoch": 21.14065934065934, "grad_norm": 0.93359375, "learning_rate": 3.771548780216792e-05, "loss": 0.1364, "step": 28857 }, { "epoch": 21.14139194139194, "grad_norm": 0.95703125, "learning_rate": 3.771217772110087e-05, "loss": 0.1238, "step": 28858 }, { "epoch": 21.14212454212454, "grad_norm": 0.84375, "learning_rate": 3.7708867701304477e-05, "loss": 0.1198, "step": 28859 }, { "epoch": 21.142857142857142, "grad_norm": 0.82421875, "learning_rate": 3.770555774279457e-05, "loss": 0.1231, "step": 28860 }, { "epoch": 21.143589743589743, "grad_norm": 0.96484375, "learning_rate": 3.770224784558706e-05, "loss": 0.1141, "step": 28861 }, { "epoch": 21.144322344322344, "grad_norm": 0.90625, "learning_rate": 3.7698938009697785e-05, "loss": 0.1154, "step": 28862 }, { "epoch": 21.145054945054945, "grad_norm": 1.3125, "learning_rate": 3.7695628235142596e-05, "loss": 0.1125, "step": 28863 }, { "epoch": 21.145787545787545, "grad_norm": 0.953125, "learning_rate": 3.769231852193734e-05, "loss": 0.1018, "step": 28864 }, { "epoch": 21.146520146520146, "grad_norm": 0.61328125, "learning_rate": 3.76890088700979e-05, "loss": 0.0887, "step": 28865 }, { "epoch": 21.147252747252747, "grad_norm": 1.8671875, "learning_rate": 3.768569927964013e-05, "loss": 0.128, "step": 28866 }, { "epoch": 21.147985347985347, "grad_norm": 0.8671875, "learning_rate": 3.768238975057987e-05, "loss": 0.106, "step": 28867 }, { "epoch": 21.148717948717948, "grad_norm": 1.0390625, "learning_rate": 3.767908028293301e-05, "loss": 0.128, "step": 28868 }, { "epoch": 21.14945054945055, "grad_norm": 1.3671875, "learning_rate": 3.767577087671537e-05, "loss": 0.1509, "step": 28869 }, { "epoch": 21.15018315018315, "grad_norm": 1.3125, "learning_rate": 3.7672461531942805e-05, "loss": 0.1279, "step": 28870 }, { "epoch": 21.15091575091575, "grad_norm": 0.74609375, "learning_rate": 3.766915224863121e-05, "loss": 0.1031, "step": 28871 }, { "epoch": 21.15164835164835, "grad_norm": 0.9921875, "learning_rate": 3.7665843026796415e-05, "loss": 0.1285, "step": 28872 }, { "epoch": 21.152380952380952, "grad_norm": 1.1171875, "learning_rate": 3.7662533866454285e-05, "loss": 0.1241, "step": 28873 }, { "epoch": 21.153113553113553, "grad_norm": 1.21875, "learning_rate": 3.765922476762067e-05, "loss": 0.1089, "step": 28874 }, { "epoch": 21.153846153846153, "grad_norm": 0.671875, "learning_rate": 3.7655915730311395e-05, "loss": 0.1113, "step": 28875 }, { "epoch": 21.154578754578754, "grad_norm": 0.9296875, "learning_rate": 3.765260675454237e-05, "loss": 0.095, "step": 28876 }, { "epoch": 21.155311355311355, "grad_norm": 1.0, "learning_rate": 3.764929784032943e-05, "loss": 0.1227, "step": 28877 }, { "epoch": 21.156043956043955, "grad_norm": 1.3359375, "learning_rate": 3.7645988987688405e-05, "loss": 0.1279, "step": 28878 }, { "epoch": 21.156776556776556, "grad_norm": 1.140625, "learning_rate": 3.7642680196635185e-05, "loss": 0.143, "step": 28879 }, { "epoch": 21.157509157509157, "grad_norm": 0.640625, "learning_rate": 3.763937146718559e-05, "loss": 0.1129, "step": 28880 }, { "epoch": 21.158241758241758, "grad_norm": 0.65234375, "learning_rate": 3.763606279935548e-05, "loss": 0.1038, "step": 28881 }, { "epoch": 21.15897435897436, "grad_norm": 1.0703125, "learning_rate": 3.763275419316074e-05, "loss": 0.1233, "step": 28882 }, { "epoch": 21.15970695970696, "grad_norm": 1.3359375, "learning_rate": 3.762944564861719e-05, "loss": 0.1316, "step": 28883 }, { "epoch": 21.16043956043956, "grad_norm": 0.9765625, "learning_rate": 3.7626137165740695e-05, "loss": 0.1255, "step": 28884 }, { "epoch": 21.16117216117216, "grad_norm": 1.0546875, "learning_rate": 3.7622828744547095e-05, "loss": 0.1092, "step": 28885 }, { "epoch": 21.16190476190476, "grad_norm": 1.6328125, "learning_rate": 3.7619520385052236e-05, "loss": 0.1175, "step": 28886 }, { "epoch": 21.162637362637362, "grad_norm": 0.9375, "learning_rate": 3.7616212087272e-05, "loss": 0.1211, "step": 28887 }, { "epoch": 21.163369963369963, "grad_norm": 1.3359375, "learning_rate": 3.761290385122223e-05, "loss": 0.1401, "step": 28888 }, { "epoch": 21.164102564102564, "grad_norm": 1.4609375, "learning_rate": 3.760959567691876e-05, "loss": 0.1226, "step": 28889 }, { "epoch": 21.164835164835164, "grad_norm": 1.5, "learning_rate": 3.760628756437744e-05, "loss": 0.119, "step": 28890 }, { "epoch": 21.165567765567765, "grad_norm": 1.734375, "learning_rate": 3.760297951361413e-05, "loss": 0.1057, "step": 28891 }, { "epoch": 21.166300366300366, "grad_norm": 0.7734375, "learning_rate": 3.759967152464468e-05, "loss": 0.1367, "step": 28892 }, { "epoch": 21.167032967032966, "grad_norm": 1.28125, "learning_rate": 3.759636359748495e-05, "loss": 0.1424, "step": 28893 }, { "epoch": 21.167765567765567, "grad_norm": 1.359375, "learning_rate": 3.759305573215077e-05, "loss": 0.126, "step": 28894 }, { "epoch": 21.168498168498168, "grad_norm": 1.828125, "learning_rate": 3.7589747928658e-05, "loss": 0.1162, "step": 28895 }, { "epoch": 21.16923076923077, "grad_norm": 1.0703125, "learning_rate": 3.758644018702247e-05, "loss": 0.1385, "step": 28896 }, { "epoch": 21.16996336996337, "grad_norm": 0.921875, "learning_rate": 3.758313250726006e-05, "loss": 0.1202, "step": 28897 }, { "epoch": 21.17069597069597, "grad_norm": 1.125, "learning_rate": 3.757982488938661e-05, "loss": 0.1339, "step": 28898 }, { "epoch": 21.17142857142857, "grad_norm": 1.0859375, "learning_rate": 3.757651733341795e-05, "loss": 0.1212, "step": 28899 }, { "epoch": 21.17216117216117, "grad_norm": 0.75, "learning_rate": 3.757320983936995e-05, "loss": 0.0962, "step": 28900 }, { "epoch": 21.172893772893772, "grad_norm": 0.95703125, "learning_rate": 3.7569902407258415e-05, "loss": 0.1134, "step": 28901 }, { "epoch": 21.173626373626373, "grad_norm": 0.98046875, "learning_rate": 3.756659503709925e-05, "loss": 0.1136, "step": 28902 }, { "epoch": 21.174358974358974, "grad_norm": 1.078125, "learning_rate": 3.756328772890827e-05, "loss": 0.1145, "step": 28903 }, { "epoch": 21.175091575091574, "grad_norm": 1.046875, "learning_rate": 3.755998048270132e-05, "loss": 0.1133, "step": 28904 }, { "epoch": 21.175824175824175, "grad_norm": 0.88671875, "learning_rate": 3.7556673298494247e-05, "loss": 0.148, "step": 28905 }, { "epoch": 21.176556776556776, "grad_norm": 0.67578125, "learning_rate": 3.7553366176302914e-05, "loss": 0.0896, "step": 28906 }, { "epoch": 21.177289377289377, "grad_norm": 0.9296875, "learning_rate": 3.755005911614313e-05, "loss": 0.1274, "step": 28907 }, { "epoch": 21.178021978021977, "grad_norm": 1.3984375, "learning_rate": 3.7546752118030795e-05, "loss": 0.1544, "step": 28908 }, { "epoch": 21.178754578754578, "grad_norm": 0.9609375, "learning_rate": 3.754344518198171e-05, "loss": 0.1065, "step": 28909 }, { "epoch": 21.17948717948718, "grad_norm": 1.0390625, "learning_rate": 3.7540138308011734e-05, "loss": 0.0959, "step": 28910 }, { "epoch": 21.18021978021978, "grad_norm": 1.1953125, "learning_rate": 3.753683149613671e-05, "loss": 0.1217, "step": 28911 }, { "epoch": 21.18095238095238, "grad_norm": 0.8359375, "learning_rate": 3.7533524746372454e-05, "loss": 0.103, "step": 28912 }, { "epoch": 21.18168498168498, "grad_norm": 0.98046875, "learning_rate": 3.7530218058734875e-05, "loss": 0.125, "step": 28913 }, { "epoch": 21.18241758241758, "grad_norm": 0.8515625, "learning_rate": 3.752691143323977e-05, "loss": 0.1402, "step": 28914 }, { "epoch": 21.183150183150182, "grad_norm": 0.6484375, "learning_rate": 3.752360486990299e-05, "loss": 0.1002, "step": 28915 }, { "epoch": 21.183882783882783, "grad_norm": 1.0703125, "learning_rate": 3.752029836874037e-05, "loss": 0.1217, "step": 28916 }, { "epoch": 21.184615384615384, "grad_norm": 1.109375, "learning_rate": 3.751699192976776e-05, "loss": 0.1274, "step": 28917 }, { "epoch": 21.185347985347985, "grad_norm": 0.92578125, "learning_rate": 3.751368555300101e-05, "loss": 0.1241, "step": 28918 }, { "epoch": 21.186080586080585, "grad_norm": 1.125, "learning_rate": 3.7510379238455965e-05, "loss": 0.1068, "step": 28919 }, { "epoch": 21.186813186813186, "grad_norm": 0.9375, "learning_rate": 3.750707298614845e-05, "loss": 0.1032, "step": 28920 }, { "epoch": 21.187545787545787, "grad_norm": 1.078125, "learning_rate": 3.7503766796094316e-05, "loss": 0.1454, "step": 28921 }, { "epoch": 21.188278388278388, "grad_norm": 1.359375, "learning_rate": 3.7500460668309383e-05, "loss": 0.1159, "step": 28922 }, { "epoch": 21.18901098901099, "grad_norm": 2.3125, "learning_rate": 3.749715460280953e-05, "loss": 0.1286, "step": 28923 }, { "epoch": 21.18974358974359, "grad_norm": 1.03125, "learning_rate": 3.749384859961058e-05, "loss": 0.1423, "step": 28924 }, { "epoch": 21.19047619047619, "grad_norm": 0.86328125, "learning_rate": 3.7490542658728366e-05, "loss": 0.0995, "step": 28925 }, { "epoch": 21.19120879120879, "grad_norm": 1.2109375, "learning_rate": 3.748723678017874e-05, "loss": 0.1213, "step": 28926 }, { "epoch": 21.19194139194139, "grad_norm": 0.8359375, "learning_rate": 3.7483930963977505e-05, "loss": 0.1064, "step": 28927 }, { "epoch": 21.192673992673992, "grad_norm": 0.9453125, "learning_rate": 3.7480625210140556e-05, "loss": 0.1285, "step": 28928 }, { "epoch": 21.193406593406593, "grad_norm": 1.0078125, "learning_rate": 3.747731951868371e-05, "loss": 0.1416, "step": 28929 }, { "epoch": 21.194139194139193, "grad_norm": 0.73828125, "learning_rate": 3.7474013889622776e-05, "loss": 0.1339, "step": 28930 }, { "epoch": 21.194871794871794, "grad_norm": 1.0078125, "learning_rate": 3.7470708322973645e-05, "loss": 0.1451, "step": 28931 }, { "epoch": 21.195604395604395, "grad_norm": 0.734375, "learning_rate": 3.746740281875211e-05, "loss": 0.1194, "step": 28932 }, { "epoch": 21.196336996336996, "grad_norm": 1.28125, "learning_rate": 3.7464097376974023e-05, "loss": 0.1117, "step": 28933 }, { "epoch": 21.197069597069596, "grad_norm": 0.83203125, "learning_rate": 3.746079199765524e-05, "loss": 0.0853, "step": 28934 }, { "epoch": 21.197802197802197, "grad_norm": 0.96875, "learning_rate": 3.745748668081158e-05, "loss": 0.1112, "step": 28935 }, { "epoch": 21.198534798534798, "grad_norm": 1.03125, "learning_rate": 3.745418142645889e-05, "loss": 0.1182, "step": 28936 }, { "epoch": 21.1992673992674, "grad_norm": 1.1484375, "learning_rate": 3.745087623461299e-05, "loss": 0.1196, "step": 28937 }, { "epoch": 21.2, "grad_norm": 1.1328125, "learning_rate": 3.744757110528971e-05, "loss": 0.1296, "step": 28938 }, { "epoch": 21.2007326007326, "grad_norm": 0.6328125, "learning_rate": 3.744426603850492e-05, "loss": 0.1147, "step": 28939 }, { "epoch": 21.2014652014652, "grad_norm": 1.375, "learning_rate": 3.744096103427443e-05, "loss": 0.1229, "step": 28940 }, { "epoch": 21.2021978021978, "grad_norm": 1.234375, "learning_rate": 3.743765609261409e-05, "loss": 0.1103, "step": 28941 }, { "epoch": 21.202930402930402, "grad_norm": 1.734375, "learning_rate": 3.7434351213539725e-05, "loss": 0.1186, "step": 28942 }, { "epoch": 21.203663003663003, "grad_norm": 0.65234375, "learning_rate": 3.743104639706717e-05, "loss": 0.1078, "step": 28943 }, { "epoch": 21.204395604395604, "grad_norm": 1.2421875, "learning_rate": 3.7427741643212256e-05, "loss": 0.1187, "step": 28944 }, { "epoch": 21.205128205128204, "grad_norm": 0.7578125, "learning_rate": 3.742443695199084e-05, "loss": 0.1169, "step": 28945 }, { "epoch": 21.205860805860805, "grad_norm": 1.0078125, "learning_rate": 3.7421132323418735e-05, "loss": 0.1265, "step": 28946 }, { "epoch": 21.206593406593406, "grad_norm": 1.8125, "learning_rate": 3.741782775751177e-05, "loss": 0.0973, "step": 28947 }, { "epoch": 21.207326007326007, "grad_norm": 1.171875, "learning_rate": 3.741452325428578e-05, "loss": 0.1175, "step": 28948 }, { "epoch": 21.208058608058607, "grad_norm": 0.9453125, "learning_rate": 3.7411218813756615e-05, "loss": 0.1168, "step": 28949 }, { "epoch": 21.208791208791208, "grad_norm": 0.83984375, "learning_rate": 3.7407914435940105e-05, "loss": 0.1022, "step": 28950 }, { "epoch": 21.20952380952381, "grad_norm": 0.78125, "learning_rate": 3.740461012085207e-05, "loss": 0.1194, "step": 28951 }, { "epoch": 21.21025641025641, "grad_norm": 1.0078125, "learning_rate": 3.740130586850835e-05, "loss": 0.1282, "step": 28952 }, { "epoch": 21.21098901098901, "grad_norm": 1.4296875, "learning_rate": 3.739800167892476e-05, "loss": 0.1079, "step": 28953 }, { "epoch": 21.21172161172161, "grad_norm": 0.84375, "learning_rate": 3.7394697552117144e-05, "loss": 0.101, "step": 28954 }, { "epoch": 21.21245421245421, "grad_norm": 1.1640625, "learning_rate": 3.739139348810136e-05, "loss": 0.1176, "step": 28955 }, { "epoch": 21.213186813186812, "grad_norm": 1.1171875, "learning_rate": 3.7388089486893185e-05, "loss": 0.1343, "step": 28956 }, { "epoch": 21.213919413919413, "grad_norm": 0.96875, "learning_rate": 3.738478554850849e-05, "loss": 0.1257, "step": 28957 }, { "epoch": 21.214652014652014, "grad_norm": 0.9140625, "learning_rate": 3.738148167296308e-05, "loss": 0.0955, "step": 28958 }, { "epoch": 21.215384615384615, "grad_norm": 0.9140625, "learning_rate": 3.737817786027281e-05, "loss": 0.1289, "step": 28959 }, { "epoch": 21.216117216117215, "grad_norm": 1.21875, "learning_rate": 3.73748741104535e-05, "loss": 0.1242, "step": 28960 }, { "epoch": 21.216849816849816, "grad_norm": 0.87109375, "learning_rate": 3.737157042352098e-05, "loss": 0.1361, "step": 28961 }, { "epoch": 21.217582417582417, "grad_norm": 1.1015625, "learning_rate": 3.736826679949108e-05, "loss": 0.1356, "step": 28962 }, { "epoch": 21.218315018315018, "grad_norm": 1.1484375, "learning_rate": 3.736496323837961e-05, "loss": 0.1274, "step": 28963 }, { "epoch": 21.21904761904762, "grad_norm": 0.97265625, "learning_rate": 3.736165974020241e-05, "loss": 0.1136, "step": 28964 }, { "epoch": 21.21978021978022, "grad_norm": 0.796875, "learning_rate": 3.7358356304975315e-05, "loss": 0.0929, "step": 28965 }, { "epoch": 21.22051282051282, "grad_norm": 1.4140625, "learning_rate": 3.7355052932714165e-05, "loss": 0.1404, "step": 28966 }, { "epoch": 21.22124542124542, "grad_norm": 1.1796875, "learning_rate": 3.735174962343475e-05, "loss": 0.1329, "step": 28967 }, { "epoch": 21.22197802197802, "grad_norm": 1.25, "learning_rate": 3.7348446377152935e-05, "loss": 0.1135, "step": 28968 }, { "epoch": 21.222710622710622, "grad_norm": 1.21875, "learning_rate": 3.7345143193884514e-05, "loss": 0.1339, "step": 28969 }, { "epoch": 21.223443223443223, "grad_norm": 0.8125, "learning_rate": 3.734184007364534e-05, "loss": 0.0835, "step": 28970 }, { "epoch": 21.224175824175823, "grad_norm": 0.71484375, "learning_rate": 3.733853701645123e-05, "loss": 0.0982, "step": 28971 }, { "epoch": 21.224908424908424, "grad_norm": 1.2734375, "learning_rate": 3.733523402231801e-05, "loss": 0.1114, "step": 28972 }, { "epoch": 21.225641025641025, "grad_norm": 1.171875, "learning_rate": 3.733193109126152e-05, "loss": 0.1209, "step": 28973 }, { "epoch": 21.226373626373626, "grad_norm": 0.90625, "learning_rate": 3.732862822329753e-05, "loss": 0.1404, "step": 28974 }, { "epoch": 21.227106227106226, "grad_norm": 1.2109375, "learning_rate": 3.732532541844194e-05, "loss": 0.1321, "step": 28975 }, { "epoch": 21.227838827838827, "grad_norm": 2.03125, "learning_rate": 3.732202267671054e-05, "loss": 0.1463, "step": 28976 }, { "epoch": 21.228571428571428, "grad_norm": 1.0703125, "learning_rate": 3.7318719998119154e-05, "loss": 0.1074, "step": 28977 }, { "epoch": 21.22930402930403, "grad_norm": 0.984375, "learning_rate": 3.731541738268361e-05, "loss": 0.1284, "step": 28978 }, { "epoch": 21.23003663003663, "grad_norm": 1.59375, "learning_rate": 3.731211483041971e-05, "loss": 0.1451, "step": 28979 }, { "epoch": 21.23076923076923, "grad_norm": 1.015625, "learning_rate": 3.7308812341343304e-05, "loss": 0.1213, "step": 28980 }, { "epoch": 21.23150183150183, "grad_norm": 1.03125, "learning_rate": 3.730550991547022e-05, "loss": 0.1415, "step": 28981 }, { "epoch": 21.23223443223443, "grad_norm": 1.6484375, "learning_rate": 3.730220755281626e-05, "loss": 0.1586, "step": 28982 }, { "epoch": 21.232967032967032, "grad_norm": 1.1953125, "learning_rate": 3.729890525339726e-05, "loss": 0.1252, "step": 28983 }, { "epoch": 21.233699633699633, "grad_norm": 0.76953125, "learning_rate": 3.729560301722905e-05, "loss": 0.121, "step": 28984 }, { "epoch": 21.234432234432234, "grad_norm": 1.046875, "learning_rate": 3.7292300844327404e-05, "loss": 0.1335, "step": 28985 }, { "epoch": 21.235164835164834, "grad_norm": 0.859375, "learning_rate": 3.728899873470821e-05, "loss": 0.1747, "step": 28986 }, { "epoch": 21.235897435897435, "grad_norm": 0.8046875, "learning_rate": 3.7285696688387264e-05, "loss": 0.1449, "step": 28987 }, { "epoch": 21.236630036630036, "grad_norm": 1.0, "learning_rate": 3.728239470538037e-05, "loss": 0.1063, "step": 28988 }, { "epoch": 21.237362637362637, "grad_norm": 0.953125, "learning_rate": 3.7279092785703376e-05, "loss": 0.0985, "step": 28989 }, { "epoch": 21.238095238095237, "grad_norm": 1.2109375, "learning_rate": 3.727579092937206e-05, "loss": 0.1364, "step": 28990 }, { "epoch": 21.238827838827838, "grad_norm": 0.890625, "learning_rate": 3.727248913640229e-05, "loss": 0.1069, "step": 28991 }, { "epoch": 21.23956043956044, "grad_norm": 0.81640625, "learning_rate": 3.726918740680988e-05, "loss": 0.1259, "step": 28992 }, { "epoch": 21.24029304029304, "grad_norm": 0.67578125, "learning_rate": 3.726588574061062e-05, "loss": 0.089, "step": 28993 }, { "epoch": 21.24102564102564, "grad_norm": 0.89453125, "learning_rate": 3.7262584137820345e-05, "loss": 0.1198, "step": 28994 }, { "epoch": 21.24175824175824, "grad_norm": 0.875, "learning_rate": 3.7259282598454874e-05, "loss": 0.1235, "step": 28995 }, { "epoch": 21.24249084249084, "grad_norm": 0.796875, "learning_rate": 3.725598112253002e-05, "loss": 0.1135, "step": 28996 }, { "epoch": 21.243223443223442, "grad_norm": 0.796875, "learning_rate": 3.7252679710061634e-05, "loss": 0.0955, "step": 28997 }, { "epoch": 21.243956043956043, "grad_norm": 1.6484375, "learning_rate": 3.7249378361065496e-05, "loss": 0.1089, "step": 28998 }, { "epoch": 21.244688644688644, "grad_norm": 1.0, "learning_rate": 3.724607707555743e-05, "loss": 0.0992, "step": 28999 }, { "epoch": 21.245421245421245, "grad_norm": 1.5546875, "learning_rate": 3.724277585355326e-05, "loss": 0.134, "step": 29000 }, { "epoch": 21.246153846153845, "grad_norm": 0.81640625, "learning_rate": 3.72394746950688e-05, "loss": 0.1101, "step": 29001 }, { "epoch": 21.246886446886446, "grad_norm": 1.4921875, "learning_rate": 3.723617360011989e-05, "loss": 0.1401, "step": 29002 }, { "epoch": 21.247619047619047, "grad_norm": 0.71875, "learning_rate": 3.723287256872231e-05, "loss": 0.1363, "step": 29003 }, { "epoch": 21.248351648351647, "grad_norm": 0.84375, "learning_rate": 3.72295716008919e-05, "loss": 0.1136, "step": 29004 }, { "epoch": 21.249084249084248, "grad_norm": 1.3046875, "learning_rate": 3.7226270696644454e-05, "loss": 0.1401, "step": 29005 }, { "epoch": 21.24981684981685, "grad_norm": 1.0859375, "learning_rate": 3.722296985599582e-05, "loss": 0.1346, "step": 29006 }, { "epoch": 21.25054945054945, "grad_norm": 0.72265625, "learning_rate": 3.721966907896178e-05, "loss": 0.1034, "step": 29007 }, { "epoch": 21.25128205128205, "grad_norm": 1.265625, "learning_rate": 3.721636836555818e-05, "loss": 0.1153, "step": 29008 }, { "epoch": 21.25201465201465, "grad_norm": 1.2265625, "learning_rate": 3.721306771580081e-05, "loss": 0.1324, "step": 29009 }, { "epoch": 21.252747252747252, "grad_norm": 1.2578125, "learning_rate": 3.720976712970551e-05, "loss": 0.1135, "step": 29010 }, { "epoch": 21.253479853479853, "grad_norm": 1.0234375, "learning_rate": 3.720646660728805e-05, "loss": 0.1161, "step": 29011 }, { "epoch": 21.254212454212453, "grad_norm": 1.015625, "learning_rate": 3.72031661485643e-05, "loss": 0.1408, "step": 29012 }, { "epoch": 21.254945054945054, "grad_norm": 1.8359375, "learning_rate": 3.7199865753550034e-05, "loss": 0.1323, "step": 29013 }, { "epoch": 21.255677655677655, "grad_norm": 1.1015625, "learning_rate": 3.719656542226109e-05, "loss": 0.1354, "step": 29014 }, { "epoch": 21.256410256410255, "grad_norm": 1.328125, "learning_rate": 3.7193265154713264e-05, "loss": 0.0909, "step": 29015 }, { "epoch": 21.257142857142856, "grad_norm": 0.83984375, "learning_rate": 3.718996495092234e-05, "loss": 0.1291, "step": 29016 }, { "epoch": 21.257875457875457, "grad_norm": 0.73046875, "learning_rate": 3.71866648109042e-05, "loss": 0.1112, "step": 29017 }, { "epoch": 21.258608058608058, "grad_norm": 1.2578125, "learning_rate": 3.718336473467461e-05, "loss": 0.1065, "step": 29018 }, { "epoch": 21.25934065934066, "grad_norm": 1.1171875, "learning_rate": 3.718006472224938e-05, "loss": 0.1283, "step": 29019 }, { "epoch": 21.26007326007326, "grad_norm": 1.1875, "learning_rate": 3.7176764773644354e-05, "loss": 0.1173, "step": 29020 }, { "epoch": 21.26080586080586, "grad_norm": 1.1640625, "learning_rate": 3.7173464888875305e-05, "loss": 0.1247, "step": 29021 }, { "epoch": 21.26153846153846, "grad_norm": 1.0546875, "learning_rate": 3.717016506795806e-05, "loss": 0.1124, "step": 29022 }, { "epoch": 21.26227106227106, "grad_norm": 1.5390625, "learning_rate": 3.716686531090845e-05, "loss": 0.1339, "step": 29023 }, { "epoch": 21.263003663003662, "grad_norm": 0.859375, "learning_rate": 3.716356561774225e-05, "loss": 0.0953, "step": 29024 }, { "epoch": 21.263736263736263, "grad_norm": 1.234375, "learning_rate": 3.716026598847529e-05, "loss": 0.1096, "step": 29025 }, { "epoch": 21.264468864468864, "grad_norm": 0.8203125, "learning_rate": 3.715696642312335e-05, "loss": 0.1172, "step": 29026 }, { "epoch": 21.265201465201464, "grad_norm": 1.4609375, "learning_rate": 3.715366692170229e-05, "loss": 0.1166, "step": 29027 }, { "epoch": 21.265934065934065, "grad_norm": 1.46875, "learning_rate": 3.71503674842279e-05, "loss": 0.1003, "step": 29028 }, { "epoch": 21.266666666666666, "grad_norm": 0.7265625, "learning_rate": 3.714706811071596e-05, "loss": 0.1009, "step": 29029 }, { "epoch": 21.267399267399266, "grad_norm": 0.890625, "learning_rate": 3.714376880118232e-05, "loss": 0.1226, "step": 29030 }, { "epoch": 21.268131868131867, "grad_norm": 0.9140625, "learning_rate": 3.714046955564275e-05, "loss": 0.1409, "step": 29031 }, { "epoch": 21.268864468864468, "grad_norm": 1.34375, "learning_rate": 3.713717037411309e-05, "loss": 0.1276, "step": 29032 }, { "epoch": 21.26959706959707, "grad_norm": 1.53125, "learning_rate": 3.713387125660912e-05, "loss": 0.1229, "step": 29033 }, { "epoch": 21.27032967032967, "grad_norm": 0.875, "learning_rate": 3.7130572203146666e-05, "loss": 0.1132, "step": 29034 }, { "epoch": 21.27106227106227, "grad_norm": 1.203125, "learning_rate": 3.712727321374154e-05, "loss": 0.1061, "step": 29035 }, { "epoch": 21.27179487179487, "grad_norm": 1.4765625, "learning_rate": 3.7123974288409536e-05, "loss": 0.1289, "step": 29036 }, { "epoch": 21.27252747252747, "grad_norm": 1.3828125, "learning_rate": 3.712067542716644e-05, "loss": 0.1465, "step": 29037 }, { "epoch": 21.273260073260072, "grad_norm": 0.90625, "learning_rate": 3.71173766300281e-05, "loss": 0.1035, "step": 29038 }, { "epoch": 21.273992673992673, "grad_norm": 1.3359375, "learning_rate": 3.7114077897010305e-05, "loss": 0.1403, "step": 29039 }, { "epoch": 21.274725274725274, "grad_norm": 1.203125, "learning_rate": 3.7110779228128856e-05, "loss": 0.1267, "step": 29040 }, { "epoch": 21.275457875457874, "grad_norm": 0.91015625, "learning_rate": 3.710748062339956e-05, "loss": 0.1063, "step": 29041 }, { "epoch": 21.276190476190475, "grad_norm": 1.03125, "learning_rate": 3.7104182082838196e-05, "loss": 0.115, "step": 29042 }, { "epoch": 21.276923076923076, "grad_norm": 1.3671875, "learning_rate": 3.710088360646062e-05, "loss": 0.1426, "step": 29043 }, { "epoch": 21.277655677655677, "grad_norm": 1.34375, "learning_rate": 3.70975851942826e-05, "loss": 0.1004, "step": 29044 }, { "epoch": 21.278388278388277, "grad_norm": 1.125, "learning_rate": 3.7094286846319945e-05, "loss": 0.1176, "step": 29045 }, { "epoch": 21.279120879120878, "grad_norm": 0.91015625, "learning_rate": 3.709098856258847e-05, "loss": 0.1012, "step": 29046 }, { "epoch": 21.27985347985348, "grad_norm": 0.8515625, "learning_rate": 3.708769034310396e-05, "loss": 0.0916, "step": 29047 }, { "epoch": 21.28058608058608, "grad_norm": 0.96484375, "learning_rate": 3.708439218788224e-05, "loss": 0.1271, "step": 29048 }, { "epoch": 21.28131868131868, "grad_norm": 1.21875, "learning_rate": 3.70810940969391e-05, "loss": 0.1245, "step": 29049 }, { "epoch": 21.28205128205128, "grad_norm": 0.72265625, "learning_rate": 3.7077796070290346e-05, "loss": 0.1093, "step": 29050 }, { "epoch": 21.282783882783882, "grad_norm": 1.5859375, "learning_rate": 3.7074498107951774e-05, "loss": 0.1246, "step": 29051 }, { "epoch": 21.283516483516483, "grad_norm": 1.2421875, "learning_rate": 3.7071200209939174e-05, "loss": 0.1028, "step": 29052 }, { "epoch": 21.284249084249083, "grad_norm": 1.421875, "learning_rate": 3.7067902376268376e-05, "loss": 0.1358, "step": 29053 }, { "epoch": 21.284981684981684, "grad_norm": 0.953125, "learning_rate": 3.7064604606955175e-05, "loss": 0.121, "step": 29054 }, { "epoch": 21.285714285714285, "grad_norm": 0.765625, "learning_rate": 3.706130690201536e-05, "loss": 0.1173, "step": 29055 }, { "epoch": 21.286446886446885, "grad_norm": 0.9453125, "learning_rate": 3.705800926146472e-05, "loss": 0.132, "step": 29056 }, { "epoch": 21.287179487179486, "grad_norm": 0.9765625, "learning_rate": 3.705471168531908e-05, "loss": 0.1303, "step": 29057 }, { "epoch": 21.287912087912087, "grad_norm": 1.3203125, "learning_rate": 3.705141417359423e-05, "loss": 0.1256, "step": 29058 }, { "epoch": 21.288644688644688, "grad_norm": 0.890625, "learning_rate": 3.7048116726305964e-05, "loss": 0.134, "step": 29059 }, { "epoch": 21.28937728937729, "grad_norm": 1.015625, "learning_rate": 3.7044819343470105e-05, "loss": 0.1343, "step": 29060 }, { "epoch": 21.29010989010989, "grad_norm": 0.92578125, "learning_rate": 3.7041522025102426e-05, "loss": 0.1382, "step": 29061 }, { "epoch": 21.29084249084249, "grad_norm": 0.7890625, "learning_rate": 3.7038224771218734e-05, "loss": 0.1071, "step": 29062 }, { "epoch": 21.29157509157509, "grad_norm": 1.1484375, "learning_rate": 3.70349275818348e-05, "loss": 0.1382, "step": 29063 }, { "epoch": 21.29230769230769, "grad_norm": 1.0859375, "learning_rate": 3.7031630456966484e-05, "loss": 0.1114, "step": 29064 }, { "epoch": 21.293040293040292, "grad_norm": 0.78125, "learning_rate": 3.7028333396629537e-05, "loss": 0.1142, "step": 29065 }, { "epoch": 21.293772893772893, "grad_norm": 0.7578125, "learning_rate": 3.7025036400839766e-05, "loss": 0.1026, "step": 29066 }, { "epoch": 21.294505494505493, "grad_norm": 0.796875, "learning_rate": 3.702173946961297e-05, "loss": 0.0898, "step": 29067 }, { "epoch": 21.295238095238094, "grad_norm": 0.9140625, "learning_rate": 3.701844260296492e-05, "loss": 0.1254, "step": 29068 }, { "epoch": 21.295970695970695, "grad_norm": 0.8125, "learning_rate": 3.701514580091147e-05, "loss": 0.1172, "step": 29069 }, { "epoch": 21.296703296703296, "grad_norm": 1.3203125, "learning_rate": 3.701184906346837e-05, "loss": 0.1346, "step": 29070 }, { "epoch": 21.297435897435896, "grad_norm": 1.46875, "learning_rate": 3.700855239065144e-05, "loss": 0.1358, "step": 29071 }, { "epoch": 21.298168498168497, "grad_norm": 1.3984375, "learning_rate": 3.700525578247645e-05, "loss": 0.1454, "step": 29072 }, { "epoch": 21.298901098901098, "grad_norm": 1.375, "learning_rate": 3.70019592389592e-05, "loss": 0.1103, "step": 29073 }, { "epoch": 21.2996336996337, "grad_norm": 1.640625, "learning_rate": 3.699866276011551e-05, "loss": 0.1459, "step": 29074 }, { "epoch": 21.3003663003663, "grad_norm": 0.8359375, "learning_rate": 3.699536634596117e-05, "loss": 0.1015, "step": 29075 }, { "epoch": 21.3010989010989, "grad_norm": 0.84765625, "learning_rate": 3.699206999651195e-05, "loss": 0.0937, "step": 29076 }, { "epoch": 21.3018315018315, "grad_norm": 0.99609375, "learning_rate": 3.698877371178367e-05, "loss": 0.1361, "step": 29077 }, { "epoch": 21.3025641025641, "grad_norm": 0.87109375, "learning_rate": 3.6985477491792077e-05, "loss": 0.1201, "step": 29078 }, { "epoch": 21.303296703296702, "grad_norm": 0.80859375, "learning_rate": 3.698218133655303e-05, "loss": 0.1162, "step": 29079 }, { "epoch": 21.304029304029303, "grad_norm": 1.21875, "learning_rate": 3.697888524608228e-05, "loss": 0.1402, "step": 29080 }, { "epoch": 21.304761904761904, "grad_norm": 1.234375, "learning_rate": 3.697558922039564e-05, "loss": 0.1174, "step": 29081 }, { "epoch": 21.305494505494504, "grad_norm": 1.7734375, "learning_rate": 3.697229325950887e-05, "loss": 0.1409, "step": 29082 }, { "epoch": 21.306227106227105, "grad_norm": 1.15625, "learning_rate": 3.696899736343779e-05, "loss": 0.133, "step": 29083 }, { "epoch": 21.306959706959706, "grad_norm": 1.8828125, "learning_rate": 3.69657015321982e-05, "loss": 0.1525, "step": 29084 }, { "epoch": 21.307692307692307, "grad_norm": 0.65625, "learning_rate": 3.696240576580586e-05, "loss": 0.1086, "step": 29085 }, { "epoch": 21.308424908424907, "grad_norm": 1.0390625, "learning_rate": 3.6959110064276584e-05, "loss": 0.1154, "step": 29086 }, { "epoch": 21.309157509157508, "grad_norm": 1.1015625, "learning_rate": 3.6955814427626163e-05, "loss": 0.1371, "step": 29087 }, { "epoch": 21.30989010989011, "grad_norm": 0.9375, "learning_rate": 3.6952518855870386e-05, "loss": 0.1206, "step": 29088 }, { "epoch": 21.31062271062271, "grad_norm": 0.66796875, "learning_rate": 3.694922334902501e-05, "loss": 0.0998, "step": 29089 }, { "epoch": 21.31135531135531, "grad_norm": 1.015625, "learning_rate": 3.6945927907105874e-05, "loss": 0.1109, "step": 29090 }, { "epoch": 21.31208791208791, "grad_norm": 0.87890625, "learning_rate": 3.694263253012876e-05, "loss": 0.1255, "step": 29091 }, { "epoch": 21.31282051282051, "grad_norm": 1.1015625, "learning_rate": 3.6939337218109434e-05, "loss": 0.1309, "step": 29092 }, { "epoch": 21.313553113553112, "grad_norm": 0.8125, "learning_rate": 3.6936041971063696e-05, "loss": 0.1162, "step": 29093 }, { "epoch": 21.314285714285713, "grad_norm": 0.91796875, "learning_rate": 3.693274678900731e-05, "loss": 0.093, "step": 29094 }, { "epoch": 21.315018315018314, "grad_norm": 0.984375, "learning_rate": 3.692945167195611e-05, "loss": 0.1328, "step": 29095 }, { "epoch": 21.315750915750915, "grad_norm": 1.3203125, "learning_rate": 3.692615661992586e-05, "loss": 0.1413, "step": 29096 }, { "epoch": 21.316483516483515, "grad_norm": 0.71484375, "learning_rate": 3.6922861632932344e-05, "loss": 0.108, "step": 29097 }, { "epoch": 21.317216117216116, "grad_norm": 1.0546875, "learning_rate": 3.691956671099137e-05, "loss": 0.1083, "step": 29098 }, { "epoch": 21.317948717948717, "grad_norm": 1.109375, "learning_rate": 3.691627185411868e-05, "loss": 0.1167, "step": 29099 }, { "epoch": 21.318681318681318, "grad_norm": 0.84375, "learning_rate": 3.691297706233012e-05, "loss": 0.1145, "step": 29100 }, { "epoch": 21.31941391941392, "grad_norm": 1.5234375, "learning_rate": 3.690968233564144e-05, "loss": 0.143, "step": 29101 }, { "epoch": 21.32014652014652, "grad_norm": 0.9375, "learning_rate": 3.6906387674068434e-05, "loss": 0.1261, "step": 29102 }, { "epoch": 21.32087912087912, "grad_norm": 1.046875, "learning_rate": 3.690309307762689e-05, "loss": 0.1645, "step": 29103 }, { "epoch": 21.32161172161172, "grad_norm": 1.046875, "learning_rate": 3.689979854633257e-05, "loss": 0.1457, "step": 29104 }, { "epoch": 21.32234432234432, "grad_norm": 1.6171875, "learning_rate": 3.689650408020129e-05, "loss": 0.1714, "step": 29105 }, { "epoch": 21.323076923076922, "grad_norm": 1.265625, "learning_rate": 3.689320967924884e-05, "loss": 0.1208, "step": 29106 }, { "epoch": 21.323809523809523, "grad_norm": 1.171875, "learning_rate": 3.6889915343490984e-05, "loss": 0.1253, "step": 29107 }, { "epoch": 21.324542124542123, "grad_norm": 1.1875, "learning_rate": 3.6886621072943493e-05, "loss": 0.1318, "step": 29108 }, { "epoch": 21.325274725274724, "grad_norm": 1.15625, "learning_rate": 3.6883326867622174e-05, "loss": 0.1268, "step": 29109 }, { "epoch": 21.326007326007325, "grad_norm": 1.625, "learning_rate": 3.688003272754282e-05, "loss": 0.1116, "step": 29110 }, { "epoch": 21.326739926739926, "grad_norm": 0.9375, "learning_rate": 3.6876738652721186e-05, "loss": 0.13, "step": 29111 }, { "epoch": 21.327472527472526, "grad_norm": 1.0078125, "learning_rate": 3.687344464317309e-05, "loss": 0.105, "step": 29112 }, { "epoch": 21.328205128205127, "grad_norm": 1.1171875, "learning_rate": 3.687015069891428e-05, "loss": 0.1259, "step": 29113 }, { "epoch": 21.328937728937728, "grad_norm": 1.515625, "learning_rate": 3.6866856819960565e-05, "loss": 0.1366, "step": 29114 }, { "epoch": 21.32967032967033, "grad_norm": 1.3984375, "learning_rate": 3.6863563006327684e-05, "loss": 0.1329, "step": 29115 }, { "epoch": 21.33040293040293, "grad_norm": 0.89453125, "learning_rate": 3.686026925803147e-05, "loss": 0.1173, "step": 29116 }, { "epoch": 21.33113553113553, "grad_norm": 0.6640625, "learning_rate": 3.6856975575087695e-05, "loss": 0.1005, "step": 29117 }, { "epoch": 21.33186813186813, "grad_norm": 1.0390625, "learning_rate": 3.6853681957512116e-05, "loss": 0.122, "step": 29118 }, { "epoch": 21.33260073260073, "grad_norm": 1.203125, "learning_rate": 3.6850388405320535e-05, "loss": 0.0928, "step": 29119 }, { "epoch": 21.333333333333332, "grad_norm": 1.0234375, "learning_rate": 3.684709491852869e-05, "loss": 0.1272, "step": 29120 }, { "epoch": 21.334065934065933, "grad_norm": 0.7265625, "learning_rate": 3.684380149715243e-05, "loss": 0.1207, "step": 29121 }, { "epoch": 21.334798534798534, "grad_norm": 1.1171875, "learning_rate": 3.684050814120749e-05, "loss": 0.1073, "step": 29122 }, { "epoch": 21.335531135531134, "grad_norm": 1.015625, "learning_rate": 3.683721485070966e-05, "loss": 0.1113, "step": 29123 }, { "epoch": 21.336263736263735, "grad_norm": 0.7421875, "learning_rate": 3.683392162567473e-05, "loss": 0.1002, "step": 29124 }, { "epoch": 21.336996336996336, "grad_norm": 1.09375, "learning_rate": 3.683062846611843e-05, "loss": 0.1294, "step": 29125 }, { "epoch": 21.337728937728937, "grad_norm": 1.1328125, "learning_rate": 3.6827335372056615e-05, "loss": 0.1435, "step": 29126 }, { "epoch": 21.338461538461537, "grad_norm": 0.76171875, "learning_rate": 3.6824042343505016e-05, "loss": 0.1137, "step": 29127 }, { "epoch": 21.339194139194138, "grad_norm": 0.734375, "learning_rate": 3.6820749380479426e-05, "loss": 0.1116, "step": 29128 }, { "epoch": 21.33992673992674, "grad_norm": 1.203125, "learning_rate": 3.6817456482995606e-05, "loss": 0.1236, "step": 29129 }, { "epoch": 21.34065934065934, "grad_norm": 0.875, "learning_rate": 3.681416365106933e-05, "loss": 0.1005, "step": 29130 }, { "epoch": 21.34139194139194, "grad_norm": 0.78515625, "learning_rate": 3.681087088471641e-05, "loss": 0.1007, "step": 29131 }, { "epoch": 21.34212454212454, "grad_norm": 1.2421875, "learning_rate": 3.68075781839526e-05, "loss": 0.1478, "step": 29132 }, { "epoch": 21.34285714285714, "grad_norm": 1.5078125, "learning_rate": 3.680428554879368e-05, "loss": 0.1578, "step": 29133 }, { "epoch": 21.343589743589742, "grad_norm": 0.94921875, "learning_rate": 3.680099297925541e-05, "loss": 0.1034, "step": 29134 }, { "epoch": 21.344322344322343, "grad_norm": 1.953125, "learning_rate": 3.679770047535358e-05, "loss": 0.1229, "step": 29135 }, { "epoch": 21.345054945054944, "grad_norm": 1.3515625, "learning_rate": 3.6794408037103977e-05, "loss": 0.1314, "step": 29136 }, { "epoch": 21.345787545787545, "grad_norm": 1.46875, "learning_rate": 3.679111566452235e-05, "loss": 0.1006, "step": 29137 }, { "epoch": 21.346520146520145, "grad_norm": 1.5234375, "learning_rate": 3.67878233576245e-05, "loss": 0.1216, "step": 29138 }, { "epoch": 21.347252747252746, "grad_norm": 1.0390625, "learning_rate": 3.6784531116426195e-05, "loss": 0.1121, "step": 29139 }, { "epoch": 21.347985347985347, "grad_norm": 1.3125, "learning_rate": 3.67812389409432e-05, "loss": 0.1626, "step": 29140 }, { "epoch": 21.348717948717947, "grad_norm": 1.6171875, "learning_rate": 3.6777946831191274e-05, "loss": 0.084, "step": 29141 }, { "epoch": 21.349450549450548, "grad_norm": 0.66796875, "learning_rate": 3.677465478718623e-05, "loss": 0.1315, "step": 29142 }, { "epoch": 21.35018315018315, "grad_norm": 1.484375, "learning_rate": 3.6771362808943815e-05, "loss": 0.1085, "step": 29143 }, { "epoch": 21.35091575091575, "grad_norm": 0.73046875, "learning_rate": 3.676807089647982e-05, "loss": 0.1044, "step": 29144 }, { "epoch": 21.35164835164835, "grad_norm": 1.1953125, "learning_rate": 3.676477904980999e-05, "loss": 0.1165, "step": 29145 }, { "epoch": 21.35238095238095, "grad_norm": 1.109375, "learning_rate": 3.676148726895009e-05, "loss": 0.1325, "step": 29146 }, { "epoch": 21.353113553113552, "grad_norm": 0.93359375, "learning_rate": 3.675819555391595e-05, "loss": 0.0974, "step": 29147 }, { "epoch": 21.353846153846153, "grad_norm": 0.90234375, "learning_rate": 3.6754903904723295e-05, "loss": 0.1045, "step": 29148 }, { "epoch": 21.354578754578753, "grad_norm": 1.4375, "learning_rate": 3.675161232138792e-05, "loss": 0.1411, "step": 29149 }, { "epoch": 21.355311355311354, "grad_norm": 1.65625, "learning_rate": 3.674832080392557e-05, "loss": 0.1741, "step": 29150 }, { "epoch": 21.356043956043955, "grad_norm": 1.9921875, "learning_rate": 3.674502935235202e-05, "loss": 0.1426, "step": 29151 }, { "epoch": 21.356776556776556, "grad_norm": 0.97265625, "learning_rate": 3.674173796668306e-05, "loss": 0.1139, "step": 29152 }, { "epoch": 21.357509157509156, "grad_norm": 1.21875, "learning_rate": 3.673844664693446e-05, "loss": 0.1259, "step": 29153 }, { "epoch": 21.358241758241757, "grad_norm": 0.60546875, "learning_rate": 3.673515539312197e-05, "loss": 0.0941, "step": 29154 }, { "epoch": 21.358974358974358, "grad_norm": 1.1015625, "learning_rate": 3.6731864205261373e-05, "loss": 0.121, "step": 29155 }, { "epoch": 21.35970695970696, "grad_norm": 1.03125, "learning_rate": 3.6728573083368424e-05, "loss": 0.1249, "step": 29156 }, { "epoch": 21.36043956043956, "grad_norm": 1.359375, "learning_rate": 3.672528202745891e-05, "loss": 0.1182, "step": 29157 }, { "epoch": 21.36117216117216, "grad_norm": 1.3203125, "learning_rate": 3.672199103754859e-05, "loss": 0.1056, "step": 29158 }, { "epoch": 21.36190476190476, "grad_norm": 1.015625, "learning_rate": 3.6718700113653246e-05, "loss": 0.1317, "step": 29159 }, { "epoch": 21.36263736263736, "grad_norm": 1.0546875, "learning_rate": 3.671540925578862e-05, "loss": 0.0995, "step": 29160 }, { "epoch": 21.363369963369962, "grad_norm": 1.0703125, "learning_rate": 3.671211846397049e-05, "loss": 0.1541, "step": 29161 }, { "epoch": 21.364102564102563, "grad_norm": 1.34375, "learning_rate": 3.670882773821464e-05, "loss": 0.1091, "step": 29162 }, { "epoch": 21.364835164835164, "grad_norm": 1.0625, "learning_rate": 3.670553707853681e-05, "loss": 0.102, "step": 29163 }, { "epoch": 21.365567765567764, "grad_norm": 1.0625, "learning_rate": 3.67022464849528e-05, "loss": 0.1243, "step": 29164 }, { "epoch": 21.366300366300365, "grad_norm": 0.73828125, "learning_rate": 3.669895595747834e-05, "loss": 0.0997, "step": 29165 }, { "epoch": 21.367032967032966, "grad_norm": 1.1875, "learning_rate": 3.669566549612922e-05, "loss": 0.1005, "step": 29166 }, { "epoch": 21.367765567765566, "grad_norm": 0.75390625, "learning_rate": 3.6692375100921177e-05, "loss": 0.1193, "step": 29167 }, { "epoch": 21.368498168498167, "grad_norm": 1.3125, "learning_rate": 3.6689084771870026e-05, "loss": 0.0978, "step": 29168 }, { "epoch": 21.369230769230768, "grad_norm": 0.9765625, "learning_rate": 3.66857945089915e-05, "loss": 0.113, "step": 29169 }, { "epoch": 21.36996336996337, "grad_norm": 2.0, "learning_rate": 3.6682504312301364e-05, "loss": 0.1433, "step": 29170 }, { "epoch": 21.37069597069597, "grad_norm": 1.515625, "learning_rate": 3.667921418181538e-05, "loss": 0.1141, "step": 29171 }, { "epoch": 21.37142857142857, "grad_norm": 0.953125, "learning_rate": 3.6675924117549306e-05, "loss": 0.1202, "step": 29172 }, { "epoch": 21.37216117216117, "grad_norm": 1.1484375, "learning_rate": 3.667263411951894e-05, "loss": 0.1466, "step": 29173 }, { "epoch": 21.37289377289377, "grad_norm": 0.80859375, "learning_rate": 3.6669344187740006e-05, "loss": 0.1252, "step": 29174 }, { "epoch": 21.373626373626372, "grad_norm": 1.0546875, "learning_rate": 3.66660543222283e-05, "loss": 0.1405, "step": 29175 }, { "epoch": 21.374358974358973, "grad_norm": 1.2578125, "learning_rate": 3.6662764522999564e-05, "loss": 0.1159, "step": 29176 }, { "epoch": 21.375091575091574, "grad_norm": 1.0390625, "learning_rate": 3.665947479006955e-05, "loss": 0.1026, "step": 29177 }, { "epoch": 21.375824175824174, "grad_norm": 0.8125, "learning_rate": 3.6656185123454065e-05, "loss": 0.1239, "step": 29178 }, { "epoch": 21.376556776556775, "grad_norm": 0.7734375, "learning_rate": 3.6652895523168825e-05, "loss": 0.1278, "step": 29179 }, { "epoch": 21.377289377289376, "grad_norm": 1.0703125, "learning_rate": 3.6649605989229616e-05, "loss": 0.1371, "step": 29180 }, { "epoch": 21.378021978021977, "grad_norm": 0.9296875, "learning_rate": 3.664631652165219e-05, "loss": 0.1152, "step": 29181 }, { "epoch": 21.378754578754577, "grad_norm": 1.1953125, "learning_rate": 3.664302712045228e-05, "loss": 0.1258, "step": 29182 }, { "epoch": 21.379487179487178, "grad_norm": 1.078125, "learning_rate": 3.6639737785645714e-05, "loss": 0.1323, "step": 29183 }, { "epoch": 21.38021978021978, "grad_norm": 0.98828125, "learning_rate": 3.66364485172482e-05, "loss": 0.1272, "step": 29184 }, { "epoch": 21.38095238095238, "grad_norm": 1.078125, "learning_rate": 3.6633159315275524e-05, "loss": 0.1135, "step": 29185 }, { "epoch": 21.38168498168498, "grad_norm": 1.2734375, "learning_rate": 3.662987017974342e-05, "loss": 0.1064, "step": 29186 }, { "epoch": 21.38241758241758, "grad_norm": 1.1875, "learning_rate": 3.6626581110667654e-05, "loss": 0.107, "step": 29187 }, { "epoch": 21.383150183150182, "grad_norm": 0.6796875, "learning_rate": 3.6623292108064e-05, "loss": 0.0929, "step": 29188 }, { "epoch": 21.383882783882783, "grad_norm": 0.82421875, "learning_rate": 3.662000317194822e-05, "loss": 0.1315, "step": 29189 }, { "epoch": 21.384615384615383, "grad_norm": 1.390625, "learning_rate": 3.661671430233606e-05, "loss": 0.1299, "step": 29190 }, { "epoch": 21.385347985347984, "grad_norm": 1.546875, "learning_rate": 3.661342549924327e-05, "loss": 0.1684, "step": 29191 }, { "epoch": 21.386080586080585, "grad_norm": 0.953125, "learning_rate": 3.6610136762685625e-05, "loss": 0.1241, "step": 29192 }, { "epoch": 21.386813186813185, "grad_norm": 1.375, "learning_rate": 3.660684809267885e-05, "loss": 0.1392, "step": 29193 }, { "epoch": 21.387545787545786, "grad_norm": 1.2265625, "learning_rate": 3.6603559489238754e-05, "loss": 0.142, "step": 29194 }, { "epoch": 21.388278388278387, "grad_norm": 0.83984375, "learning_rate": 3.660027095238106e-05, "loss": 0.0978, "step": 29195 }, { "epoch": 21.389010989010988, "grad_norm": 1.1640625, "learning_rate": 3.659698248212154e-05, "loss": 0.114, "step": 29196 }, { "epoch": 21.38974358974359, "grad_norm": 1.1796875, "learning_rate": 3.659369407847592e-05, "loss": 0.1102, "step": 29197 }, { "epoch": 21.39047619047619, "grad_norm": 1.25, "learning_rate": 3.659040574145998e-05, "loss": 0.1319, "step": 29198 }, { "epoch": 21.39120879120879, "grad_norm": 1.21875, "learning_rate": 3.658711747108949e-05, "loss": 0.1395, "step": 29199 }, { "epoch": 21.39194139194139, "grad_norm": 0.6484375, "learning_rate": 3.658382926738017e-05, "loss": 0.0966, "step": 29200 }, { "epoch": 21.39267399267399, "grad_norm": 1.4453125, "learning_rate": 3.6580541130347805e-05, "loss": 0.1075, "step": 29201 }, { "epoch": 21.393406593406592, "grad_norm": 1.2265625, "learning_rate": 3.6577253060008135e-05, "loss": 0.1183, "step": 29202 }, { "epoch": 21.394139194139193, "grad_norm": 1.15625, "learning_rate": 3.657396505637689e-05, "loss": 0.1125, "step": 29203 }, { "epoch": 21.394871794871793, "grad_norm": 1.1796875, "learning_rate": 3.657067711946987e-05, "loss": 0.1379, "step": 29204 }, { "epoch": 21.395604395604394, "grad_norm": 1.046875, "learning_rate": 3.6567389249302825e-05, "loss": 0.1416, "step": 29205 }, { "epoch": 21.396336996336995, "grad_norm": 1.3515625, "learning_rate": 3.656410144589148e-05, "loss": 0.1385, "step": 29206 }, { "epoch": 21.397069597069596, "grad_norm": 0.765625, "learning_rate": 3.6560813709251606e-05, "loss": 0.1149, "step": 29207 }, { "epoch": 21.397802197802196, "grad_norm": 1.1796875, "learning_rate": 3.6557526039398926e-05, "loss": 0.1276, "step": 29208 }, { "epoch": 21.398534798534797, "grad_norm": 1.0078125, "learning_rate": 3.655423843634923e-05, "loss": 0.1316, "step": 29209 }, { "epoch": 21.399267399267398, "grad_norm": 0.72265625, "learning_rate": 3.655095090011826e-05, "loss": 0.1108, "step": 29210 }, { "epoch": 21.4, "grad_norm": 1.0078125, "learning_rate": 3.654766343072177e-05, "loss": 0.1242, "step": 29211 }, { "epoch": 21.4007326007326, "grad_norm": 1.328125, "learning_rate": 3.654437602817548e-05, "loss": 0.1581, "step": 29212 }, { "epoch": 21.4014652014652, "grad_norm": 1.140625, "learning_rate": 3.6541088692495197e-05, "loss": 0.1343, "step": 29213 }, { "epoch": 21.4021978021978, "grad_norm": 1.09375, "learning_rate": 3.6537801423696604e-05, "loss": 0.1268, "step": 29214 }, { "epoch": 21.4029304029304, "grad_norm": 2.328125, "learning_rate": 3.6534514221795515e-05, "loss": 0.1343, "step": 29215 }, { "epoch": 21.403663003663002, "grad_norm": 1.515625, "learning_rate": 3.653122708680765e-05, "loss": 0.1116, "step": 29216 }, { "epoch": 21.404395604395603, "grad_norm": 0.9765625, "learning_rate": 3.652794001874876e-05, "loss": 0.1257, "step": 29217 }, { "epoch": 21.405128205128204, "grad_norm": 0.94921875, "learning_rate": 3.65246530176346e-05, "loss": 0.1168, "step": 29218 }, { "epoch": 21.405860805860804, "grad_norm": 0.62109375, "learning_rate": 3.652136608348089e-05, "loss": 0.0972, "step": 29219 }, { "epoch": 21.406593406593405, "grad_norm": 1.0234375, "learning_rate": 3.6518079216303424e-05, "loss": 0.139, "step": 29220 }, { "epoch": 21.407326007326006, "grad_norm": 1.265625, "learning_rate": 3.6514792416117935e-05, "loss": 0.1406, "step": 29221 }, { "epoch": 21.408058608058607, "grad_norm": 0.71484375, "learning_rate": 3.651150568294017e-05, "loss": 0.1156, "step": 29222 }, { "epoch": 21.408791208791207, "grad_norm": 1.0, "learning_rate": 3.650821901678585e-05, "loss": 0.1238, "step": 29223 }, { "epoch": 21.409523809523808, "grad_norm": 1.6640625, "learning_rate": 3.650493241767076e-05, "loss": 0.1224, "step": 29224 }, { "epoch": 21.41025641025641, "grad_norm": 1.46875, "learning_rate": 3.6501645885610635e-05, "loss": 0.1295, "step": 29225 }, { "epoch": 21.41098901098901, "grad_norm": 0.953125, "learning_rate": 3.649835942062122e-05, "loss": 0.1129, "step": 29226 }, { "epoch": 21.41172161172161, "grad_norm": 1.34375, "learning_rate": 3.649507302271826e-05, "loss": 0.121, "step": 29227 }, { "epoch": 21.41245421245421, "grad_norm": 1.3671875, "learning_rate": 3.649178669191751e-05, "loss": 0.1494, "step": 29228 }, { "epoch": 21.41318681318681, "grad_norm": 1.3671875, "learning_rate": 3.648850042823468e-05, "loss": 0.1027, "step": 29229 }, { "epoch": 21.413919413919412, "grad_norm": 0.921875, "learning_rate": 3.648521423168557e-05, "loss": 0.1095, "step": 29230 }, { "epoch": 21.414652014652013, "grad_norm": 0.703125, "learning_rate": 3.64819281022859e-05, "loss": 0.1188, "step": 29231 }, { "epoch": 21.415384615384614, "grad_norm": 1.2421875, "learning_rate": 3.647864204005141e-05, "loss": 0.092, "step": 29232 }, { "epoch": 21.416117216117215, "grad_norm": 1.234375, "learning_rate": 3.647535604499785e-05, "loss": 0.147, "step": 29233 }, { "epoch": 21.416849816849815, "grad_norm": 1.015625, "learning_rate": 3.647207011714095e-05, "loss": 0.1178, "step": 29234 }, { "epoch": 21.417582417582416, "grad_norm": 1.046875, "learning_rate": 3.646878425649648e-05, "loss": 0.1044, "step": 29235 }, { "epoch": 21.418315018315017, "grad_norm": 1.0390625, "learning_rate": 3.646549846308017e-05, "loss": 0.1183, "step": 29236 }, { "epoch": 21.419047619047618, "grad_norm": 1.03125, "learning_rate": 3.646221273690777e-05, "loss": 0.1072, "step": 29237 }, { "epoch": 21.41978021978022, "grad_norm": 1.0, "learning_rate": 3.6458927077995e-05, "loss": 0.1484, "step": 29238 }, { "epoch": 21.42051282051282, "grad_norm": 0.73046875, "learning_rate": 3.645564148635764e-05, "loss": 0.1129, "step": 29239 }, { "epoch": 21.42124542124542, "grad_norm": 0.9296875, "learning_rate": 3.645235596201138e-05, "loss": 0.1116, "step": 29240 }, { "epoch": 21.42197802197802, "grad_norm": 0.9140625, "learning_rate": 3.644907050497202e-05, "loss": 0.1309, "step": 29241 }, { "epoch": 21.42271062271062, "grad_norm": 1.0859375, "learning_rate": 3.644578511525527e-05, "loss": 0.1162, "step": 29242 }, { "epoch": 21.423443223443222, "grad_norm": 0.609375, "learning_rate": 3.644249979287689e-05, "loss": 0.1154, "step": 29243 }, { "epoch": 21.424175824175823, "grad_norm": 1.53125, "learning_rate": 3.6439214537852606e-05, "loss": 0.1298, "step": 29244 }, { "epoch": 21.424908424908423, "grad_norm": 1.171875, "learning_rate": 3.6435929350198136e-05, "loss": 0.1076, "step": 29245 }, { "epoch": 21.425641025641024, "grad_norm": 0.94921875, "learning_rate": 3.6432644229929256e-05, "loss": 0.1127, "step": 29246 }, { "epoch": 21.426373626373625, "grad_norm": 1.8125, "learning_rate": 3.642935917706171e-05, "loss": 0.1164, "step": 29247 }, { "epoch": 21.427106227106226, "grad_norm": 1.015625, "learning_rate": 3.642607419161122e-05, "loss": 0.1176, "step": 29248 }, { "epoch": 21.427838827838826, "grad_norm": 0.953125, "learning_rate": 3.6422789273593515e-05, "loss": 0.1386, "step": 29249 }, { "epoch": 21.428571428571427, "grad_norm": 1.359375, "learning_rate": 3.6419504423024344e-05, "loss": 0.1127, "step": 29250 }, { "epoch": 21.429304029304028, "grad_norm": 1.1328125, "learning_rate": 3.641621963991948e-05, "loss": 0.1356, "step": 29251 }, { "epoch": 21.43003663003663, "grad_norm": 1.5, "learning_rate": 3.64129349242946e-05, "loss": 0.1246, "step": 29252 }, { "epoch": 21.43076923076923, "grad_norm": 0.97265625, "learning_rate": 3.640965027616549e-05, "loss": 0.1382, "step": 29253 }, { "epoch": 21.43150183150183, "grad_norm": 1.0078125, "learning_rate": 3.640636569554787e-05, "loss": 0.1167, "step": 29254 }, { "epoch": 21.43223443223443, "grad_norm": 1.265625, "learning_rate": 3.640308118245746e-05, "loss": 0.1704, "step": 29255 }, { "epoch": 21.43296703296703, "grad_norm": 1.1953125, "learning_rate": 3.639979673691004e-05, "loss": 0.1261, "step": 29256 }, { "epoch": 21.433699633699632, "grad_norm": 1.3828125, "learning_rate": 3.6396512358921317e-05, "loss": 0.1128, "step": 29257 }, { "epoch": 21.434432234432233, "grad_norm": 0.9609375, "learning_rate": 3.6393228048507035e-05, "loss": 0.1306, "step": 29258 }, { "epoch": 21.435164835164834, "grad_norm": 0.84375, "learning_rate": 3.638994380568293e-05, "loss": 0.1369, "step": 29259 }, { "epoch": 21.435897435897434, "grad_norm": 1.03125, "learning_rate": 3.638665963046471e-05, "loss": 0.1227, "step": 29260 }, { "epoch": 21.436630036630035, "grad_norm": 1.171875, "learning_rate": 3.638337552286816e-05, "loss": 0.139, "step": 29261 }, { "epoch": 21.437362637362636, "grad_norm": 0.8671875, "learning_rate": 3.638009148290899e-05, "loss": 0.1189, "step": 29262 }, { "epoch": 21.438095238095237, "grad_norm": 1.03125, "learning_rate": 3.6376807510602926e-05, "loss": 0.1315, "step": 29263 }, { "epoch": 21.438827838827837, "grad_norm": 1.6328125, "learning_rate": 3.6373523605965734e-05, "loss": 0.1493, "step": 29264 }, { "epoch": 21.439560439560438, "grad_norm": 1.6171875, "learning_rate": 3.637023976901312e-05, "loss": 0.1201, "step": 29265 }, { "epoch": 21.44029304029304, "grad_norm": 0.88671875, "learning_rate": 3.636695599976081e-05, "loss": 0.0901, "step": 29266 }, { "epoch": 21.44102564102564, "grad_norm": 1.1484375, "learning_rate": 3.636367229822457e-05, "loss": 0.0918, "step": 29267 }, { "epoch": 21.44175824175824, "grad_norm": 0.8125, "learning_rate": 3.636038866442012e-05, "loss": 0.1006, "step": 29268 }, { "epoch": 21.44249084249084, "grad_norm": 0.9921875, "learning_rate": 3.635710509836318e-05, "loss": 0.1214, "step": 29269 }, { "epoch": 21.44322344322344, "grad_norm": 1.0625, "learning_rate": 3.63538216000695e-05, "loss": 0.1231, "step": 29270 }, { "epoch": 21.443956043956042, "grad_norm": 0.8046875, "learning_rate": 3.635053816955478e-05, "loss": 0.1092, "step": 29271 }, { "epoch": 21.444688644688643, "grad_norm": 0.6015625, "learning_rate": 3.63472548068348e-05, "loss": 0.1048, "step": 29272 }, { "epoch": 21.445421245421244, "grad_norm": 0.78515625, "learning_rate": 3.634397151192527e-05, "loss": 0.131, "step": 29273 }, { "epoch": 21.446153846153845, "grad_norm": 1.0546875, "learning_rate": 3.634068828484192e-05, "loss": 0.1137, "step": 29274 }, { "epoch": 21.446886446886445, "grad_norm": 0.90234375, "learning_rate": 3.633740512560047e-05, "loss": 0.1041, "step": 29275 }, { "epoch": 21.447619047619046, "grad_norm": 0.9296875, "learning_rate": 3.6334122034216663e-05, "loss": 0.1014, "step": 29276 }, { "epoch": 21.448351648351647, "grad_norm": 1.078125, "learning_rate": 3.6330839010706233e-05, "loss": 0.1094, "step": 29277 }, { "epoch": 21.449084249084247, "grad_norm": 0.9140625, "learning_rate": 3.63275560550849e-05, "loss": 0.1096, "step": 29278 }, { "epoch": 21.449816849816848, "grad_norm": 1.6328125, "learning_rate": 3.63242731673684e-05, "loss": 0.1087, "step": 29279 }, { "epoch": 21.45054945054945, "grad_norm": 0.8984375, "learning_rate": 3.632099034757246e-05, "loss": 0.1167, "step": 29280 }, { "epoch": 21.45128205128205, "grad_norm": 1.0390625, "learning_rate": 3.6317707595712795e-05, "loss": 0.1024, "step": 29281 }, { "epoch": 21.45201465201465, "grad_norm": 1.109375, "learning_rate": 3.6314424911805175e-05, "loss": 0.1129, "step": 29282 }, { "epoch": 21.45274725274725, "grad_norm": 1.7265625, "learning_rate": 3.631114229586529e-05, "loss": 0.1086, "step": 29283 }, { "epoch": 21.453479853479852, "grad_norm": 0.91015625, "learning_rate": 3.630785974790889e-05, "loss": 0.1196, "step": 29284 }, { "epoch": 21.454212454212453, "grad_norm": 0.7421875, "learning_rate": 3.630457726795169e-05, "loss": 0.1212, "step": 29285 }, { "epoch": 21.454945054945053, "grad_norm": 1.0625, "learning_rate": 3.63012948560094e-05, "loss": 0.1196, "step": 29286 }, { "epoch": 21.455677655677654, "grad_norm": 1.0625, "learning_rate": 3.629801251209779e-05, "loss": 0.0996, "step": 29287 }, { "epoch": 21.456410256410255, "grad_norm": 0.6875, "learning_rate": 3.629473023623256e-05, "loss": 0.0883, "step": 29288 }, { "epoch": 21.457142857142856, "grad_norm": 1.3203125, "learning_rate": 3.629144802842943e-05, "loss": 0.1085, "step": 29289 }, { "epoch": 21.457875457875456, "grad_norm": 1.03125, "learning_rate": 3.628816588870414e-05, "loss": 0.1086, "step": 29290 }, { "epoch": 21.458608058608057, "grad_norm": 1.1171875, "learning_rate": 3.628488381707243e-05, "loss": 0.1131, "step": 29291 }, { "epoch": 21.459340659340658, "grad_norm": 1.3671875, "learning_rate": 3.6281601813549977e-05, "loss": 0.1141, "step": 29292 }, { "epoch": 21.46007326007326, "grad_norm": 1.046875, "learning_rate": 3.627831987815256e-05, "loss": 0.1198, "step": 29293 }, { "epoch": 21.46080586080586, "grad_norm": 0.78125, "learning_rate": 3.6275038010895884e-05, "loss": 0.1325, "step": 29294 }, { "epoch": 21.46153846153846, "grad_norm": 0.94140625, "learning_rate": 3.627175621179567e-05, "loss": 0.1061, "step": 29295 }, { "epoch": 21.46227106227106, "grad_norm": 1.125, "learning_rate": 3.6268474480867634e-05, "loss": 0.1183, "step": 29296 }, { "epoch": 21.46300366300366, "grad_norm": 0.91015625, "learning_rate": 3.6265192818127505e-05, "loss": 0.1308, "step": 29297 }, { "epoch": 21.463736263736262, "grad_norm": 0.94140625, "learning_rate": 3.626191122359101e-05, "loss": 0.0932, "step": 29298 }, { "epoch": 21.464468864468863, "grad_norm": 1.46875, "learning_rate": 3.625862969727389e-05, "loss": 0.1443, "step": 29299 }, { "epoch": 21.465201465201464, "grad_norm": 1.0859375, "learning_rate": 3.625534823919184e-05, "loss": 0.125, "step": 29300 }, { "epoch": 21.465934065934064, "grad_norm": 1.65625, "learning_rate": 3.6252066849360585e-05, "loss": 0.141, "step": 29301 }, { "epoch": 21.466666666666665, "grad_norm": 0.71484375, "learning_rate": 3.624878552779586e-05, "loss": 0.0914, "step": 29302 }, { "epoch": 21.467399267399266, "grad_norm": 1.046875, "learning_rate": 3.624550427451338e-05, "loss": 0.108, "step": 29303 }, { "epoch": 21.468131868131866, "grad_norm": 1.1328125, "learning_rate": 3.6242223089528875e-05, "loss": 0.1399, "step": 29304 }, { "epoch": 21.468864468864467, "grad_norm": 1.1640625, "learning_rate": 3.6238941972858053e-05, "loss": 0.1179, "step": 29305 }, { "epoch": 21.469597069597068, "grad_norm": 0.8828125, "learning_rate": 3.6235660924516655e-05, "loss": 0.1084, "step": 29306 }, { "epoch": 21.47032967032967, "grad_norm": 1.703125, "learning_rate": 3.623237994452036e-05, "loss": 0.1323, "step": 29307 }, { "epoch": 21.47106227106227, "grad_norm": 0.984375, "learning_rate": 3.622909903288494e-05, "loss": 0.126, "step": 29308 }, { "epoch": 21.47179487179487, "grad_norm": 1.5234375, "learning_rate": 3.622581818962609e-05, "loss": 0.1084, "step": 29309 }, { "epoch": 21.47252747252747, "grad_norm": 1.265625, "learning_rate": 3.622253741475954e-05, "loss": 0.1146, "step": 29310 }, { "epoch": 21.47326007326007, "grad_norm": 0.80078125, "learning_rate": 3.6219256708300985e-05, "loss": 0.1258, "step": 29311 }, { "epoch": 21.473992673992672, "grad_norm": 1.4453125, "learning_rate": 3.621597607026615e-05, "loss": 0.1462, "step": 29312 }, { "epoch": 21.474725274725273, "grad_norm": 1.1640625, "learning_rate": 3.621269550067078e-05, "loss": 0.1198, "step": 29313 }, { "epoch": 21.475457875457874, "grad_norm": 0.6328125, "learning_rate": 3.6209414999530575e-05, "loss": 0.1016, "step": 29314 }, { "epoch": 21.476190476190474, "grad_norm": 0.9921875, "learning_rate": 3.620613456686124e-05, "loss": 0.1095, "step": 29315 }, { "epoch": 21.476923076923075, "grad_norm": 0.8125, "learning_rate": 3.6202854202678524e-05, "loss": 0.1201, "step": 29316 }, { "epoch": 21.477655677655676, "grad_norm": 0.92578125, "learning_rate": 3.619957390699812e-05, "loss": 0.1172, "step": 29317 }, { "epoch": 21.478388278388277, "grad_norm": 1.6328125, "learning_rate": 3.619629367983574e-05, "loss": 0.1354, "step": 29318 }, { "epoch": 21.479120879120877, "grad_norm": 1.4140625, "learning_rate": 3.619301352120713e-05, "loss": 0.1463, "step": 29319 }, { "epoch": 21.479853479853478, "grad_norm": 0.75390625, "learning_rate": 3.618973343112798e-05, "loss": 0.1218, "step": 29320 }, { "epoch": 21.48058608058608, "grad_norm": 0.7421875, "learning_rate": 3.618645340961403e-05, "loss": 0.127, "step": 29321 }, { "epoch": 21.48131868131868, "grad_norm": 1.2421875, "learning_rate": 3.618317345668096e-05, "loss": 0.1173, "step": 29322 }, { "epoch": 21.48205128205128, "grad_norm": 1.1015625, "learning_rate": 3.61798935723445e-05, "loss": 0.1241, "step": 29323 }, { "epoch": 21.48278388278388, "grad_norm": 1.0078125, "learning_rate": 3.617661375662039e-05, "loss": 0.1174, "step": 29324 }, { "epoch": 21.483516483516482, "grad_norm": 1.046875, "learning_rate": 3.617333400952433e-05, "loss": 0.1186, "step": 29325 }, { "epoch": 21.484249084249083, "grad_norm": 0.77734375, "learning_rate": 3.6170054331072025e-05, "loss": 0.1025, "step": 29326 }, { "epoch": 21.484981684981683, "grad_norm": 1.15625, "learning_rate": 3.6166774721279186e-05, "loss": 0.1423, "step": 29327 }, { "epoch": 21.485714285714284, "grad_norm": 0.83984375, "learning_rate": 3.6163495180161536e-05, "loss": 0.1393, "step": 29328 }, { "epoch": 21.486446886446885, "grad_norm": 1.203125, "learning_rate": 3.6160215707734775e-05, "loss": 0.1404, "step": 29329 }, { "epoch": 21.487179487179485, "grad_norm": 0.83203125, "learning_rate": 3.6156936304014644e-05, "loss": 0.1074, "step": 29330 }, { "epoch": 21.487912087912086, "grad_norm": 0.71484375, "learning_rate": 3.6153656969016844e-05, "loss": 0.1289, "step": 29331 }, { "epoch": 21.488644688644687, "grad_norm": 0.83984375, "learning_rate": 3.615037770275708e-05, "loss": 0.0877, "step": 29332 }, { "epoch": 21.489377289377288, "grad_norm": 1.2578125, "learning_rate": 3.6147098505251046e-05, "loss": 0.1202, "step": 29333 }, { "epoch": 21.49010989010989, "grad_norm": 0.9296875, "learning_rate": 3.614381937651449e-05, "loss": 0.1024, "step": 29334 }, { "epoch": 21.49084249084249, "grad_norm": 1.6015625, "learning_rate": 3.614054031656313e-05, "loss": 0.1516, "step": 29335 }, { "epoch": 21.49157509157509, "grad_norm": 1.6640625, "learning_rate": 3.6137261325412634e-05, "loss": 0.1216, "step": 29336 }, { "epoch": 21.49230769230769, "grad_norm": 0.99609375, "learning_rate": 3.6133982403078735e-05, "loss": 0.1145, "step": 29337 }, { "epoch": 21.493040293040295, "grad_norm": 1.2578125, "learning_rate": 3.613070354957713e-05, "loss": 0.1263, "step": 29338 }, { "epoch": 21.493772893772892, "grad_norm": 0.8671875, "learning_rate": 3.612742476492356e-05, "loss": 0.1027, "step": 29339 }, { "epoch": 21.494505494505496, "grad_norm": 0.87890625, "learning_rate": 3.612414604913372e-05, "loss": 0.12, "step": 29340 }, { "epoch": 21.495238095238093, "grad_norm": 1.265625, "learning_rate": 3.61208674022233e-05, "loss": 0.1226, "step": 29341 }, { "epoch": 21.495970695970698, "grad_norm": 0.640625, "learning_rate": 3.6117588824208034e-05, "loss": 0.1263, "step": 29342 }, { "epoch": 21.496703296703295, "grad_norm": 1.03125, "learning_rate": 3.611431031510362e-05, "loss": 0.1019, "step": 29343 }, { "epoch": 21.4974358974359, "grad_norm": 1.546875, "learning_rate": 3.611103187492575e-05, "loss": 0.1491, "step": 29344 }, { "epoch": 21.498168498168496, "grad_norm": 0.859375, "learning_rate": 3.610775350369017e-05, "loss": 0.1204, "step": 29345 }, { "epoch": 21.4989010989011, "grad_norm": 1.203125, "learning_rate": 3.610447520141257e-05, "loss": 0.1288, "step": 29346 }, { "epoch": 21.499633699633698, "grad_norm": 0.71484375, "learning_rate": 3.6101196968108645e-05, "loss": 0.0951, "step": 29347 }, { "epoch": 21.500366300366302, "grad_norm": 0.6484375, "learning_rate": 3.6097918803794126e-05, "loss": 0.1101, "step": 29348 }, { "epoch": 21.5010989010989, "grad_norm": 1.265625, "learning_rate": 3.609464070848468e-05, "loss": 0.1457, "step": 29349 }, { "epoch": 21.501831501831504, "grad_norm": 0.93359375, "learning_rate": 3.6091362682196054e-05, "loss": 0.0918, "step": 29350 }, { "epoch": 21.5025641025641, "grad_norm": 1.1171875, "learning_rate": 3.608808472494395e-05, "loss": 0.1383, "step": 29351 }, { "epoch": 21.503296703296705, "grad_norm": 1.15625, "learning_rate": 3.608480683674406e-05, "loss": 0.1217, "step": 29352 }, { "epoch": 21.504029304029302, "grad_norm": 1.3828125, "learning_rate": 3.608152901761208e-05, "loss": 0.1535, "step": 29353 }, { "epoch": 21.504761904761907, "grad_norm": 0.91015625, "learning_rate": 3.607825126756373e-05, "loss": 0.1088, "step": 29354 }, { "epoch": 21.505494505494504, "grad_norm": 1.15625, "learning_rate": 3.607497358661471e-05, "loss": 0.1341, "step": 29355 }, { "epoch": 21.506227106227108, "grad_norm": 0.66015625, "learning_rate": 3.607169597478073e-05, "loss": 0.1051, "step": 29356 }, { "epoch": 21.506959706959705, "grad_norm": 0.953125, "learning_rate": 3.606841843207751e-05, "loss": 0.1157, "step": 29357 }, { "epoch": 21.50769230769231, "grad_norm": 1.15625, "learning_rate": 3.6065140958520715e-05, "loss": 0.1452, "step": 29358 }, { "epoch": 21.508424908424907, "grad_norm": 0.69921875, "learning_rate": 3.606186355412606e-05, "loss": 0.11, "step": 29359 }, { "epoch": 21.50915750915751, "grad_norm": 1.375, "learning_rate": 3.605858621890928e-05, "loss": 0.129, "step": 29360 }, { "epoch": 21.509890109890108, "grad_norm": 0.96484375, "learning_rate": 3.605530895288604e-05, "loss": 0.1341, "step": 29361 }, { "epoch": 21.510622710622712, "grad_norm": 1.0625, "learning_rate": 3.605203175607207e-05, "loss": 0.1441, "step": 29362 }, { "epoch": 21.51135531135531, "grad_norm": 1.0234375, "learning_rate": 3.604875462848305e-05, "loss": 0.1129, "step": 29363 }, { "epoch": 21.512087912087914, "grad_norm": 1.1875, "learning_rate": 3.6045477570134674e-05, "loss": 0.1114, "step": 29364 }, { "epoch": 21.51282051282051, "grad_norm": 1.1171875, "learning_rate": 3.604220058104268e-05, "loss": 0.1182, "step": 29365 }, { "epoch": 21.513553113553115, "grad_norm": 1.921875, "learning_rate": 3.603892366122275e-05, "loss": 0.1282, "step": 29366 }, { "epoch": 21.514285714285712, "grad_norm": 1.4296875, "learning_rate": 3.603564681069057e-05, "loss": 0.1214, "step": 29367 }, { "epoch": 21.515018315018317, "grad_norm": 1.171875, "learning_rate": 3.603237002946187e-05, "loss": 0.1074, "step": 29368 }, { "epoch": 21.515750915750917, "grad_norm": 0.828125, "learning_rate": 3.602909331755234e-05, "loss": 0.1132, "step": 29369 }, { "epoch": 21.516483516483518, "grad_norm": 1.359375, "learning_rate": 3.602581667497764e-05, "loss": 0.1195, "step": 29370 }, { "epoch": 21.51721611721612, "grad_norm": 1.3359375, "learning_rate": 3.602254010175353e-05, "loss": 0.116, "step": 29371 }, { "epoch": 21.51794871794872, "grad_norm": 0.99609375, "learning_rate": 3.601926359789568e-05, "loss": 0.1143, "step": 29372 }, { "epoch": 21.51868131868132, "grad_norm": 0.625, "learning_rate": 3.6015987163419794e-05, "loss": 0.1232, "step": 29373 }, { "epoch": 21.51941391941392, "grad_norm": 0.79296875, "learning_rate": 3.601271079834157e-05, "loss": 0.1048, "step": 29374 }, { "epoch": 21.520146520146522, "grad_norm": 1.6484375, "learning_rate": 3.6009434502676684e-05, "loss": 0.1377, "step": 29375 }, { "epoch": 21.520879120879123, "grad_norm": 1.2421875, "learning_rate": 3.600615827644088e-05, "loss": 0.1167, "step": 29376 }, { "epoch": 21.521611721611723, "grad_norm": 1.2109375, "learning_rate": 3.6002882119649826e-05, "loss": 0.1331, "step": 29377 }, { "epoch": 21.522344322344324, "grad_norm": 0.88671875, "learning_rate": 3.599960603231921e-05, "loss": 0.1055, "step": 29378 }, { "epoch": 21.523076923076925, "grad_norm": 1.0078125, "learning_rate": 3.599633001446475e-05, "loss": 0.1418, "step": 29379 }, { "epoch": 21.523809523809526, "grad_norm": 1.65625, "learning_rate": 3.5993054066102126e-05, "loss": 0.138, "step": 29380 }, { "epoch": 21.524542124542126, "grad_norm": 0.671875, "learning_rate": 3.598977818724705e-05, "loss": 0.1065, "step": 29381 }, { "epoch": 21.525274725274727, "grad_norm": 1.5546875, "learning_rate": 3.5986502377915205e-05, "loss": 0.1373, "step": 29382 }, { "epoch": 21.526007326007328, "grad_norm": 1.3046875, "learning_rate": 3.59832266381223e-05, "loss": 0.1139, "step": 29383 }, { "epoch": 21.52673992673993, "grad_norm": 1.40625, "learning_rate": 3.5979950967884025e-05, "loss": 0.1294, "step": 29384 }, { "epoch": 21.52747252747253, "grad_norm": 2.0, "learning_rate": 3.597667536721605e-05, "loss": 0.1355, "step": 29385 }, { "epoch": 21.52820512820513, "grad_norm": 0.84375, "learning_rate": 3.597339983613411e-05, "loss": 0.1095, "step": 29386 }, { "epoch": 21.52893772893773, "grad_norm": 1.5625, "learning_rate": 3.597012437465388e-05, "loss": 0.1353, "step": 29387 }, { "epoch": 21.52967032967033, "grad_norm": 0.6171875, "learning_rate": 3.5966848982791045e-05, "loss": 0.1115, "step": 29388 }, { "epoch": 21.530402930402932, "grad_norm": 0.8984375, "learning_rate": 3.596357366056132e-05, "loss": 0.1234, "step": 29389 }, { "epoch": 21.531135531135533, "grad_norm": 1.21875, "learning_rate": 3.596029840798036e-05, "loss": 0.1615, "step": 29390 }, { "epoch": 21.531868131868134, "grad_norm": 1.234375, "learning_rate": 3.59570232250639e-05, "loss": 0.1242, "step": 29391 }, { "epoch": 21.532600732600734, "grad_norm": 1.140625, "learning_rate": 3.5953748111827626e-05, "loss": 0.1523, "step": 29392 }, { "epoch": 21.533333333333335, "grad_norm": 1.3984375, "learning_rate": 3.595047306828721e-05, "loss": 0.13, "step": 29393 }, { "epoch": 21.534065934065936, "grad_norm": 0.84375, "learning_rate": 3.594719809445835e-05, "loss": 0.0914, "step": 29394 }, { "epoch": 21.534798534798536, "grad_norm": 0.8828125, "learning_rate": 3.5943923190356746e-05, "loss": 0.1415, "step": 29395 }, { "epoch": 21.535531135531137, "grad_norm": 1.0390625, "learning_rate": 3.594064835599807e-05, "loss": 0.1277, "step": 29396 }, { "epoch": 21.536263736263738, "grad_norm": 1.8828125, "learning_rate": 3.593737359139805e-05, "loss": 0.178, "step": 29397 }, { "epoch": 21.53699633699634, "grad_norm": 0.90234375, "learning_rate": 3.593409889657235e-05, "loss": 0.0898, "step": 29398 }, { "epoch": 21.53772893772894, "grad_norm": 1.3828125, "learning_rate": 3.5930824271536656e-05, "loss": 0.1007, "step": 29399 }, { "epoch": 21.53846153846154, "grad_norm": 1.234375, "learning_rate": 3.592754971630668e-05, "loss": 0.133, "step": 29400 }, { "epoch": 21.53919413919414, "grad_norm": 1.1640625, "learning_rate": 3.592427523089806e-05, "loss": 0.1156, "step": 29401 }, { "epoch": 21.53992673992674, "grad_norm": 0.7890625, "learning_rate": 3.592100081532656e-05, "loss": 0.1203, "step": 29402 }, { "epoch": 21.540659340659342, "grad_norm": 1.125, "learning_rate": 3.591772646960782e-05, "loss": 0.12, "step": 29403 }, { "epoch": 21.541391941391943, "grad_norm": 1.0390625, "learning_rate": 3.591445219375753e-05, "loss": 0.1364, "step": 29404 }, { "epoch": 21.542124542124544, "grad_norm": 0.97265625, "learning_rate": 3.5911177987791405e-05, "loss": 0.1068, "step": 29405 }, { "epoch": 21.542857142857144, "grad_norm": 0.90625, "learning_rate": 3.5907903851725093e-05, "loss": 0.1326, "step": 29406 }, { "epoch": 21.543589743589745, "grad_norm": 1.1953125, "learning_rate": 3.590462978557431e-05, "loss": 0.1321, "step": 29407 }, { "epoch": 21.544322344322346, "grad_norm": 0.97265625, "learning_rate": 3.590135578935475e-05, "loss": 0.1229, "step": 29408 }, { "epoch": 21.545054945054947, "grad_norm": 0.8515625, "learning_rate": 3.5898081863082096e-05, "loss": 0.1167, "step": 29409 }, { "epoch": 21.545787545787547, "grad_norm": 0.66796875, "learning_rate": 3.589480800677201e-05, "loss": 0.1123, "step": 29410 }, { "epoch": 21.546520146520148, "grad_norm": 0.7109375, "learning_rate": 3.5891534220440175e-05, "loss": 0.0916, "step": 29411 }, { "epoch": 21.54725274725275, "grad_norm": 0.86328125, "learning_rate": 3.5888260504102325e-05, "loss": 0.0956, "step": 29412 }, { "epoch": 21.54798534798535, "grad_norm": 0.60546875, "learning_rate": 3.5884986857774105e-05, "loss": 0.1255, "step": 29413 }, { "epoch": 21.54871794871795, "grad_norm": 1.6171875, "learning_rate": 3.588171328147121e-05, "loss": 0.1086, "step": 29414 }, { "epoch": 21.54945054945055, "grad_norm": 0.82421875, "learning_rate": 3.5878439775209336e-05, "loss": 0.1094, "step": 29415 }, { "epoch": 21.550183150183152, "grad_norm": 0.9140625, "learning_rate": 3.5875166339004146e-05, "loss": 0.1359, "step": 29416 }, { "epoch": 21.550915750915753, "grad_norm": 0.81640625, "learning_rate": 3.587189297287133e-05, "loss": 0.1122, "step": 29417 }, { "epoch": 21.551648351648353, "grad_norm": 0.578125, "learning_rate": 3.5868619676826585e-05, "loss": 0.1078, "step": 29418 }, { "epoch": 21.552380952380954, "grad_norm": 1.0078125, "learning_rate": 3.5865346450885575e-05, "loss": 0.1179, "step": 29419 }, { "epoch": 21.553113553113555, "grad_norm": 0.66796875, "learning_rate": 3.586207329506401e-05, "loss": 0.1168, "step": 29420 }, { "epoch": 21.553846153846155, "grad_norm": 1.4453125, "learning_rate": 3.5858800209377555e-05, "loss": 0.1259, "step": 29421 }, { "epoch": 21.554578754578756, "grad_norm": 0.74609375, "learning_rate": 3.5855527193841874e-05, "loss": 0.1226, "step": 29422 }, { "epoch": 21.555311355311357, "grad_norm": 1.359375, "learning_rate": 3.5852254248472687e-05, "loss": 0.1369, "step": 29423 }, { "epoch": 21.556043956043958, "grad_norm": 0.83984375, "learning_rate": 3.584898137328566e-05, "loss": 0.1281, "step": 29424 }, { "epoch": 21.55677655677656, "grad_norm": 1.0234375, "learning_rate": 3.5845708568296485e-05, "loss": 0.1316, "step": 29425 }, { "epoch": 21.55750915750916, "grad_norm": 0.9609375, "learning_rate": 3.584243583352081e-05, "loss": 0.105, "step": 29426 }, { "epoch": 21.55824175824176, "grad_norm": 1.1875, "learning_rate": 3.583916316897433e-05, "loss": 0.1189, "step": 29427 }, { "epoch": 21.55897435897436, "grad_norm": 0.98828125, "learning_rate": 3.583589057467275e-05, "loss": 0.1429, "step": 29428 }, { "epoch": 21.55970695970696, "grad_norm": 0.625, "learning_rate": 3.5832618050631736e-05, "loss": 0.1178, "step": 29429 }, { "epoch": 21.560439560439562, "grad_norm": 1.875, "learning_rate": 3.5829345596866946e-05, "loss": 0.1627, "step": 29430 }, { "epoch": 21.561172161172163, "grad_norm": 0.73046875, "learning_rate": 3.582607321339409e-05, "loss": 0.1236, "step": 29431 }, { "epoch": 21.561904761904763, "grad_norm": 1.1328125, "learning_rate": 3.5822800900228824e-05, "loss": 0.1126, "step": 29432 }, { "epoch": 21.562637362637364, "grad_norm": 0.65625, "learning_rate": 3.581952865738684e-05, "loss": 0.1362, "step": 29433 }, { "epoch": 21.563369963369965, "grad_norm": 1.0234375, "learning_rate": 3.581625648488382e-05, "loss": 0.1177, "step": 29434 }, { "epoch": 21.564102564102566, "grad_norm": 1.0078125, "learning_rate": 3.581298438273543e-05, "loss": 0.1129, "step": 29435 }, { "epoch": 21.564835164835166, "grad_norm": 0.64453125, "learning_rate": 3.5809712350957365e-05, "loss": 0.1028, "step": 29436 }, { "epoch": 21.565567765567767, "grad_norm": 1.296875, "learning_rate": 3.5806440389565265e-05, "loss": 0.1313, "step": 29437 }, { "epoch": 21.566300366300368, "grad_norm": 0.9453125, "learning_rate": 3.580316849857485e-05, "loss": 0.127, "step": 29438 }, { "epoch": 21.56703296703297, "grad_norm": 1.40625, "learning_rate": 3.5799896678001795e-05, "loss": 0.1288, "step": 29439 }, { "epoch": 21.56776556776557, "grad_norm": 0.9921875, "learning_rate": 3.5796624927861745e-05, "loss": 0.0991, "step": 29440 }, { "epoch": 21.56849816849817, "grad_norm": 1.40625, "learning_rate": 3.5793353248170394e-05, "loss": 0.1565, "step": 29441 }, { "epoch": 21.56923076923077, "grad_norm": 1.09375, "learning_rate": 3.579008163894341e-05, "loss": 0.1054, "step": 29442 }, { "epoch": 21.56996336996337, "grad_norm": 0.78125, "learning_rate": 3.578681010019647e-05, "loss": 0.1208, "step": 29443 }, { "epoch": 21.570695970695972, "grad_norm": 1.0859375, "learning_rate": 3.5783538631945245e-05, "loss": 0.1098, "step": 29444 }, { "epoch": 21.571428571428573, "grad_norm": 0.97265625, "learning_rate": 3.578026723420544e-05, "loss": 0.1311, "step": 29445 }, { "epoch": 21.572161172161174, "grad_norm": 0.9453125, "learning_rate": 3.577699590699271e-05, "loss": 0.1105, "step": 29446 }, { "epoch": 21.572893772893774, "grad_norm": 1.1328125, "learning_rate": 3.577372465032272e-05, "loss": 0.1076, "step": 29447 }, { "epoch": 21.573626373626375, "grad_norm": 1.4140625, "learning_rate": 3.577045346421113e-05, "loss": 0.1658, "step": 29448 }, { "epoch": 21.574358974358976, "grad_norm": 1.015625, "learning_rate": 3.576718234867365e-05, "loss": 0.1199, "step": 29449 }, { "epoch": 21.575091575091577, "grad_norm": 1.296875, "learning_rate": 3.5763911303725925e-05, "loss": 0.1261, "step": 29450 }, { "epoch": 21.575824175824177, "grad_norm": 2.015625, "learning_rate": 3.576064032938366e-05, "loss": 0.1205, "step": 29451 }, { "epoch": 21.576556776556778, "grad_norm": 1.296875, "learning_rate": 3.5757369425662495e-05, "loss": 0.1556, "step": 29452 }, { "epoch": 21.57728937728938, "grad_norm": 0.859375, "learning_rate": 3.5754098592578095e-05, "loss": 0.144, "step": 29453 }, { "epoch": 21.57802197802198, "grad_norm": 1.0625, "learning_rate": 3.5750827830146164e-05, "loss": 0.1215, "step": 29454 }, { "epoch": 21.57875457875458, "grad_norm": 1.2109375, "learning_rate": 3.5747557138382373e-05, "loss": 0.1078, "step": 29455 }, { "epoch": 21.57948717948718, "grad_norm": 1.125, "learning_rate": 3.574428651730236e-05, "loss": 0.1202, "step": 29456 }, { "epoch": 21.58021978021978, "grad_norm": 0.91796875, "learning_rate": 3.574101596692182e-05, "loss": 0.1138, "step": 29457 }, { "epoch": 21.580952380952382, "grad_norm": 0.734375, "learning_rate": 3.5737745487256406e-05, "loss": 0.1089, "step": 29458 }, { "epoch": 21.581684981684983, "grad_norm": 1.25, "learning_rate": 3.57344750783218e-05, "loss": 0.1227, "step": 29459 }, { "epoch": 21.582417582417584, "grad_norm": 1.203125, "learning_rate": 3.573120474013369e-05, "loss": 0.1319, "step": 29460 }, { "epoch": 21.583150183150185, "grad_norm": 1.21875, "learning_rate": 3.572793447270772e-05, "loss": 0.1255, "step": 29461 }, { "epoch": 21.583882783882785, "grad_norm": 0.8359375, "learning_rate": 3.572466427605957e-05, "loss": 0.0893, "step": 29462 }, { "epoch": 21.584615384615386, "grad_norm": 1.078125, "learning_rate": 3.572139415020488e-05, "loss": 0.1334, "step": 29463 }, { "epoch": 21.585347985347987, "grad_norm": 1.40625, "learning_rate": 3.5718124095159366e-05, "loss": 0.1424, "step": 29464 }, { "epoch": 21.586080586080588, "grad_norm": 0.9140625, "learning_rate": 3.571485411093867e-05, "loss": 0.1088, "step": 29465 }, { "epoch": 21.58681318681319, "grad_norm": 1.046875, "learning_rate": 3.5711584197558456e-05, "loss": 0.1105, "step": 29466 }, { "epoch": 21.58754578754579, "grad_norm": 1.4375, "learning_rate": 3.57083143550344e-05, "loss": 0.1287, "step": 29467 }, { "epoch": 21.58827838827839, "grad_norm": 0.9375, "learning_rate": 3.570504458338216e-05, "loss": 0.1257, "step": 29468 }, { "epoch": 21.58901098901099, "grad_norm": 0.97265625, "learning_rate": 3.57017748826174e-05, "loss": 0.1347, "step": 29469 }, { "epoch": 21.58974358974359, "grad_norm": 1.953125, "learning_rate": 3.56985052527558e-05, "loss": 0.1193, "step": 29470 }, { "epoch": 21.590476190476192, "grad_norm": 0.83984375, "learning_rate": 3.569523569381304e-05, "loss": 0.1357, "step": 29471 }, { "epoch": 21.591208791208793, "grad_norm": 1.0390625, "learning_rate": 3.569196620580475e-05, "loss": 0.1238, "step": 29472 }, { "epoch": 21.591941391941393, "grad_norm": 1.4140625, "learning_rate": 3.568869678874662e-05, "loss": 0.1564, "step": 29473 }, { "epoch": 21.592673992673994, "grad_norm": 0.94921875, "learning_rate": 3.568542744265428e-05, "loss": 0.1378, "step": 29474 }, { "epoch": 21.593406593406595, "grad_norm": 1.2265625, "learning_rate": 3.568215816754345e-05, "loss": 0.1064, "step": 29475 }, { "epoch": 21.594139194139196, "grad_norm": 1.0390625, "learning_rate": 3.567888896342976e-05, "loss": 0.1233, "step": 29476 }, { "epoch": 21.594871794871796, "grad_norm": 1.0078125, "learning_rate": 3.5675619830328874e-05, "loss": 0.1291, "step": 29477 }, { "epoch": 21.595604395604397, "grad_norm": 0.98828125, "learning_rate": 3.5672350768256455e-05, "loss": 0.1247, "step": 29478 }, { "epoch": 21.596336996336998, "grad_norm": 1.46875, "learning_rate": 3.5669081777228164e-05, "loss": 0.1103, "step": 29479 }, { "epoch": 21.5970695970696, "grad_norm": 0.69140625, "learning_rate": 3.5665812857259684e-05, "loss": 0.1016, "step": 29480 }, { "epoch": 21.5978021978022, "grad_norm": 0.8125, "learning_rate": 3.566254400836667e-05, "loss": 0.1187, "step": 29481 }, { "epoch": 21.5985347985348, "grad_norm": 0.6015625, "learning_rate": 3.5659275230564765e-05, "loss": 0.1107, "step": 29482 }, { "epoch": 21.5992673992674, "grad_norm": 0.9453125, "learning_rate": 3.565600652386965e-05, "loss": 0.1257, "step": 29483 }, { "epoch": 21.6, "grad_norm": 0.8984375, "learning_rate": 3.565273788829697e-05, "loss": 0.1192, "step": 29484 }, { "epoch": 21.600732600732602, "grad_norm": 0.7421875, "learning_rate": 3.5649469323862404e-05, "loss": 0.0905, "step": 29485 }, { "epoch": 21.601465201465203, "grad_norm": 0.8984375, "learning_rate": 3.5646200830581613e-05, "loss": 0.1075, "step": 29486 }, { "epoch": 21.602197802197804, "grad_norm": 0.765625, "learning_rate": 3.564293240847025e-05, "loss": 0.1041, "step": 29487 }, { "epoch": 21.602930402930404, "grad_norm": 1.3828125, "learning_rate": 3.5639664057543975e-05, "loss": 0.1047, "step": 29488 }, { "epoch": 21.603663003663005, "grad_norm": 0.92578125, "learning_rate": 3.5636395777818425e-05, "loss": 0.1122, "step": 29489 }, { "epoch": 21.604395604395606, "grad_norm": 0.8203125, "learning_rate": 3.5633127569309306e-05, "loss": 0.0992, "step": 29490 }, { "epoch": 21.605128205128207, "grad_norm": 0.75390625, "learning_rate": 3.5629859432032255e-05, "loss": 0.1283, "step": 29491 }, { "epoch": 21.605860805860807, "grad_norm": 1.3046875, "learning_rate": 3.562659136600293e-05, "loss": 0.1507, "step": 29492 }, { "epoch": 21.606593406593408, "grad_norm": 0.82421875, "learning_rate": 3.562332337123697e-05, "loss": 0.1239, "step": 29493 }, { "epoch": 21.60732600732601, "grad_norm": 0.92578125, "learning_rate": 3.5620055447750064e-05, "loss": 0.095, "step": 29494 }, { "epoch": 21.60805860805861, "grad_norm": 1.171875, "learning_rate": 3.561678759555785e-05, "loss": 0.1213, "step": 29495 }, { "epoch": 21.60879120879121, "grad_norm": 0.96875, "learning_rate": 3.561351981467599e-05, "loss": 0.121, "step": 29496 }, { "epoch": 21.60952380952381, "grad_norm": 1.0234375, "learning_rate": 3.561025210512016e-05, "loss": 0.1254, "step": 29497 }, { "epoch": 21.61025641025641, "grad_norm": 1.5859375, "learning_rate": 3.560698446690599e-05, "loss": 0.1233, "step": 29498 }, { "epoch": 21.610989010989012, "grad_norm": 1.1328125, "learning_rate": 3.560371690004915e-05, "loss": 0.1281, "step": 29499 }, { "epoch": 21.611721611721613, "grad_norm": 1.09375, "learning_rate": 3.560044940456528e-05, "loss": 0.1189, "step": 29500 }, { "epoch": 21.612454212454214, "grad_norm": 0.89453125, "learning_rate": 3.5597181980470066e-05, "loss": 0.1477, "step": 29501 }, { "epoch": 21.613186813186815, "grad_norm": 1.015625, "learning_rate": 3.5593914627779136e-05, "loss": 0.1147, "step": 29502 }, { "epoch": 21.613919413919415, "grad_norm": 0.90234375, "learning_rate": 3.559064734650816e-05, "loss": 0.1421, "step": 29503 }, { "epoch": 21.614652014652016, "grad_norm": 1.296875, "learning_rate": 3.558738013667278e-05, "loss": 0.0941, "step": 29504 }, { "epoch": 21.615384615384617, "grad_norm": 1.2734375, "learning_rate": 3.558411299828865e-05, "loss": 0.1657, "step": 29505 }, { "epoch": 21.616117216117217, "grad_norm": 2.234375, "learning_rate": 3.558084593137145e-05, "loss": 0.1162, "step": 29506 }, { "epoch": 21.616849816849818, "grad_norm": 0.8671875, "learning_rate": 3.55775789359368e-05, "loss": 0.1105, "step": 29507 }, { "epoch": 21.61758241758242, "grad_norm": 1.0234375, "learning_rate": 3.557431201200038e-05, "loss": 0.1241, "step": 29508 }, { "epoch": 21.61831501831502, "grad_norm": 1.03125, "learning_rate": 3.557104515957783e-05, "loss": 0.1332, "step": 29509 }, { "epoch": 21.61904761904762, "grad_norm": 1.3203125, "learning_rate": 3.5567778378684776e-05, "loss": 0.1322, "step": 29510 }, { "epoch": 21.61978021978022, "grad_norm": 1.2265625, "learning_rate": 3.5564511669336926e-05, "loss": 0.1232, "step": 29511 }, { "epoch": 21.620512820512822, "grad_norm": 1.6171875, "learning_rate": 3.55612450315499e-05, "loss": 0.1391, "step": 29512 }, { "epoch": 21.621245421245423, "grad_norm": 1.0078125, "learning_rate": 3.5557978465339356e-05, "loss": 0.1126, "step": 29513 }, { "epoch": 21.621978021978023, "grad_norm": 1.9765625, "learning_rate": 3.5554711970720944e-05, "loss": 0.1411, "step": 29514 }, { "epoch": 21.622710622710624, "grad_norm": 1.6328125, "learning_rate": 3.555144554771029e-05, "loss": 0.1228, "step": 29515 }, { "epoch": 21.623443223443225, "grad_norm": 1.1171875, "learning_rate": 3.554817919632308e-05, "loss": 0.1063, "step": 29516 }, { "epoch": 21.624175824175826, "grad_norm": 1.0078125, "learning_rate": 3.554491291657497e-05, "loss": 0.1274, "step": 29517 }, { "epoch": 21.624908424908426, "grad_norm": 1.546875, "learning_rate": 3.554164670848158e-05, "loss": 0.1197, "step": 29518 }, { "epoch": 21.625641025641027, "grad_norm": 1.2578125, "learning_rate": 3.553838057205856e-05, "loss": 0.1412, "step": 29519 }, { "epoch": 21.626373626373628, "grad_norm": 1.03125, "learning_rate": 3.5535114507321584e-05, "loss": 0.0911, "step": 29520 }, { "epoch": 21.62710622710623, "grad_norm": 1.65625, "learning_rate": 3.5531848514286275e-05, "loss": 0.1484, "step": 29521 }, { "epoch": 21.62783882783883, "grad_norm": 1.15625, "learning_rate": 3.5528582592968296e-05, "loss": 0.1501, "step": 29522 }, { "epoch": 21.62857142857143, "grad_norm": 1.5703125, "learning_rate": 3.55253167433833e-05, "loss": 0.1012, "step": 29523 }, { "epoch": 21.62930402930403, "grad_norm": 0.9921875, "learning_rate": 3.5522050965546924e-05, "loss": 0.0993, "step": 29524 }, { "epoch": 21.63003663003663, "grad_norm": 0.875, "learning_rate": 3.551878525947483e-05, "loss": 0.1035, "step": 29525 }, { "epoch": 21.630769230769232, "grad_norm": 1.28125, "learning_rate": 3.5515519625182626e-05, "loss": 0.1391, "step": 29526 }, { "epoch": 21.631501831501833, "grad_norm": 0.82421875, "learning_rate": 3.551225406268601e-05, "loss": 0.0961, "step": 29527 }, { "epoch": 21.632234432234434, "grad_norm": 1.09375, "learning_rate": 3.5508988572000606e-05, "loss": 0.1263, "step": 29528 }, { "epoch": 21.632967032967034, "grad_norm": 0.90234375, "learning_rate": 3.550572315314205e-05, "loss": 0.1194, "step": 29529 }, { "epoch": 21.633699633699635, "grad_norm": 0.62890625, "learning_rate": 3.5502457806126e-05, "loss": 0.0993, "step": 29530 }, { "epoch": 21.634432234432236, "grad_norm": 0.91796875, "learning_rate": 3.5499192530968084e-05, "loss": 0.0976, "step": 29531 }, { "epoch": 21.635164835164836, "grad_norm": 1.2578125, "learning_rate": 3.549592732768397e-05, "loss": 0.1417, "step": 29532 }, { "epoch": 21.635897435897437, "grad_norm": 0.76953125, "learning_rate": 3.5492662196289304e-05, "loss": 0.0891, "step": 29533 }, { "epoch": 21.636630036630038, "grad_norm": 0.8125, "learning_rate": 3.548939713679971e-05, "loss": 0.1112, "step": 29534 }, { "epoch": 21.63736263736264, "grad_norm": 1.5390625, "learning_rate": 3.548613214923085e-05, "loss": 0.1163, "step": 29535 }, { "epoch": 21.63809523809524, "grad_norm": 1.078125, "learning_rate": 3.548286723359834e-05, "loss": 0.1467, "step": 29536 }, { "epoch": 21.63882783882784, "grad_norm": 1.3046875, "learning_rate": 3.547960238991786e-05, "loss": 0.1249, "step": 29537 }, { "epoch": 21.63956043956044, "grad_norm": 1.140625, "learning_rate": 3.5476337618205044e-05, "loss": 0.1417, "step": 29538 }, { "epoch": 21.64029304029304, "grad_norm": 0.84375, "learning_rate": 3.547307291847551e-05, "loss": 0.1297, "step": 29539 }, { "epoch": 21.641025641025642, "grad_norm": 1.1484375, "learning_rate": 3.546980829074493e-05, "loss": 0.1139, "step": 29540 }, { "epoch": 21.641758241758243, "grad_norm": 1.2421875, "learning_rate": 3.5466543735028914e-05, "loss": 0.117, "step": 29541 }, { "epoch": 21.642490842490844, "grad_norm": 1.21875, "learning_rate": 3.546327925134313e-05, "loss": 0.1166, "step": 29542 }, { "epoch": 21.643223443223444, "grad_norm": 1.4140625, "learning_rate": 3.546001483970322e-05, "loss": 0.133, "step": 29543 }, { "epoch": 21.643956043956045, "grad_norm": 1.2265625, "learning_rate": 3.5456750500124816e-05, "loss": 0.1453, "step": 29544 }, { "epoch": 21.644688644688646, "grad_norm": 0.89453125, "learning_rate": 3.545348623262354e-05, "loss": 0.1107, "step": 29545 }, { "epoch": 21.645421245421247, "grad_norm": 1.2421875, "learning_rate": 3.545022203721507e-05, "loss": 0.141, "step": 29546 }, { "epoch": 21.646153846153847, "grad_norm": 0.90625, "learning_rate": 3.544695791391502e-05, "loss": 0.0952, "step": 29547 }, { "epoch": 21.646886446886448, "grad_norm": 0.61328125, "learning_rate": 3.5443693862739034e-05, "loss": 0.1138, "step": 29548 }, { "epoch": 21.64761904761905, "grad_norm": 0.640625, "learning_rate": 3.544042988370276e-05, "loss": 0.0889, "step": 29549 }, { "epoch": 21.64835164835165, "grad_norm": 0.87109375, "learning_rate": 3.543716597682183e-05, "loss": 0.1146, "step": 29550 }, { "epoch": 21.64908424908425, "grad_norm": 1.203125, "learning_rate": 3.543390214211188e-05, "loss": 0.1391, "step": 29551 }, { "epoch": 21.64981684981685, "grad_norm": 0.65234375, "learning_rate": 3.5430638379588535e-05, "loss": 0.1138, "step": 29552 }, { "epoch": 21.650549450549452, "grad_norm": 1.2421875, "learning_rate": 3.5427374689267465e-05, "loss": 0.1432, "step": 29553 }, { "epoch": 21.651282051282053, "grad_norm": 0.69140625, "learning_rate": 3.54241110711643e-05, "loss": 0.1198, "step": 29554 }, { "epoch": 21.652014652014653, "grad_norm": 0.90234375, "learning_rate": 3.5420847525294665e-05, "loss": 0.129, "step": 29555 }, { "epoch": 21.652747252747254, "grad_norm": 0.765625, "learning_rate": 3.541758405167419e-05, "loss": 0.0966, "step": 29556 }, { "epoch": 21.653479853479855, "grad_norm": 0.94921875, "learning_rate": 3.54143206503185e-05, "loss": 0.1147, "step": 29557 }, { "epoch": 21.654212454212455, "grad_norm": 1.296875, "learning_rate": 3.541105732124328e-05, "loss": 0.1278, "step": 29558 }, { "epoch": 21.654945054945056, "grad_norm": 1.5703125, "learning_rate": 3.5407794064464134e-05, "loss": 0.1513, "step": 29559 }, { "epoch": 21.655677655677657, "grad_norm": 1.4453125, "learning_rate": 3.54045308799967e-05, "loss": 0.1269, "step": 29560 }, { "epoch": 21.656410256410258, "grad_norm": 1.09375, "learning_rate": 3.540126776785662e-05, "loss": 0.1179, "step": 29561 }, { "epoch": 21.65714285714286, "grad_norm": 1.4609375, "learning_rate": 3.5398004728059506e-05, "loss": 0.1584, "step": 29562 }, { "epoch": 21.65787545787546, "grad_norm": 1.65625, "learning_rate": 3.539474176062102e-05, "loss": 0.1135, "step": 29563 }, { "epoch": 21.65860805860806, "grad_norm": 0.87890625, "learning_rate": 3.539147886555679e-05, "loss": 0.1216, "step": 29564 }, { "epoch": 21.65934065934066, "grad_norm": 1.03125, "learning_rate": 3.5388216042882444e-05, "loss": 0.1379, "step": 29565 }, { "epoch": 21.66007326007326, "grad_norm": 1.1640625, "learning_rate": 3.5384953292613616e-05, "loss": 0.1144, "step": 29566 }, { "epoch": 21.660805860805862, "grad_norm": 1.7265625, "learning_rate": 3.5381690614765915e-05, "loss": 0.1277, "step": 29567 }, { "epoch": 21.661538461538463, "grad_norm": 1.3515625, "learning_rate": 3.537842800935502e-05, "loss": 0.1297, "step": 29568 }, { "epoch": 21.662271062271063, "grad_norm": 0.6484375, "learning_rate": 3.537516547639654e-05, "loss": 0.1065, "step": 29569 }, { "epoch": 21.663003663003664, "grad_norm": 0.578125, "learning_rate": 3.537190301590611e-05, "loss": 0.1111, "step": 29570 }, { "epoch": 21.663736263736265, "grad_norm": 1.1640625, "learning_rate": 3.5368640627899345e-05, "loss": 0.1143, "step": 29571 }, { "epoch": 21.664468864468866, "grad_norm": 1.0, "learning_rate": 3.536537831239191e-05, "loss": 0.1088, "step": 29572 }, { "epoch": 21.665201465201466, "grad_norm": 1.25, "learning_rate": 3.536211606939939e-05, "loss": 0.1272, "step": 29573 }, { "epoch": 21.665934065934067, "grad_norm": 1.453125, "learning_rate": 3.535885389893745e-05, "loss": 0.114, "step": 29574 }, { "epoch": 21.666666666666668, "grad_norm": 1.5546875, "learning_rate": 3.535559180102172e-05, "loss": 0.1265, "step": 29575 }, { "epoch": 21.66739926739927, "grad_norm": 1.3125, "learning_rate": 3.535232977566782e-05, "loss": 0.1191, "step": 29576 }, { "epoch": 21.66813186813187, "grad_norm": 1.703125, "learning_rate": 3.534906782289138e-05, "loss": 0.1312, "step": 29577 }, { "epoch": 21.66886446886447, "grad_norm": 1.359375, "learning_rate": 3.534580594270802e-05, "loss": 0.107, "step": 29578 }, { "epoch": 21.66959706959707, "grad_norm": 0.89453125, "learning_rate": 3.5342544135133396e-05, "loss": 0.1235, "step": 29579 }, { "epoch": 21.67032967032967, "grad_norm": 1.328125, "learning_rate": 3.533928240018311e-05, "loss": 0.1493, "step": 29580 }, { "epoch": 21.671062271062272, "grad_norm": 0.9375, "learning_rate": 3.53360207378728e-05, "loss": 0.0972, "step": 29581 }, { "epoch": 21.671794871794873, "grad_norm": 0.98046875, "learning_rate": 3.533275914821809e-05, "loss": 0.131, "step": 29582 }, { "epoch": 21.672527472527474, "grad_norm": 2.171875, "learning_rate": 3.53294976312346e-05, "loss": 0.1773, "step": 29583 }, { "epoch": 21.673260073260074, "grad_norm": 1.640625, "learning_rate": 3.532623618693798e-05, "loss": 0.1214, "step": 29584 }, { "epoch": 21.673992673992675, "grad_norm": 1.2109375, "learning_rate": 3.5322974815343835e-05, "loss": 0.1215, "step": 29585 }, { "epoch": 21.674725274725276, "grad_norm": 1.0703125, "learning_rate": 3.531971351646782e-05, "loss": 0.1034, "step": 29586 }, { "epoch": 21.675457875457877, "grad_norm": 0.9765625, "learning_rate": 3.531645229032553e-05, "loss": 0.1395, "step": 29587 }, { "epoch": 21.676190476190477, "grad_norm": 1.5546875, "learning_rate": 3.5313191136932574e-05, "loss": 0.0992, "step": 29588 }, { "epoch": 21.676923076923078, "grad_norm": 2.203125, "learning_rate": 3.5309930056304646e-05, "loss": 0.1391, "step": 29589 }, { "epoch": 21.67765567765568, "grad_norm": 1.3046875, "learning_rate": 3.530666904845731e-05, "loss": 0.1323, "step": 29590 }, { "epoch": 21.67838827838828, "grad_norm": 1.3671875, "learning_rate": 3.530340811340622e-05, "loss": 0.1134, "step": 29591 }, { "epoch": 21.67912087912088, "grad_norm": 0.8515625, "learning_rate": 3.530014725116699e-05, "loss": 0.1204, "step": 29592 }, { "epoch": 21.67985347985348, "grad_norm": 1.8046875, "learning_rate": 3.5296886461755226e-05, "loss": 0.1688, "step": 29593 }, { "epoch": 21.68058608058608, "grad_norm": 1.296875, "learning_rate": 3.5293625745186585e-05, "loss": 0.1404, "step": 29594 }, { "epoch": 21.681318681318682, "grad_norm": 0.8046875, "learning_rate": 3.5290365101476676e-05, "loss": 0.131, "step": 29595 }, { "epoch": 21.682051282051283, "grad_norm": 2.328125, "learning_rate": 3.5287104530641124e-05, "loss": 0.1458, "step": 29596 }, { "epoch": 21.682783882783884, "grad_norm": 1.1015625, "learning_rate": 3.528384403269553e-05, "loss": 0.1216, "step": 29597 }, { "epoch": 21.683516483516485, "grad_norm": 1.65625, "learning_rate": 3.528058360765556e-05, "loss": 0.1435, "step": 29598 }, { "epoch": 21.684249084249085, "grad_norm": 1.0625, "learning_rate": 3.527732325553678e-05, "loss": 0.1232, "step": 29599 }, { "epoch": 21.684981684981686, "grad_norm": 0.86328125, "learning_rate": 3.527406297635485e-05, "loss": 0.1337, "step": 29600 }, { "epoch": 21.685714285714287, "grad_norm": 1.5859375, "learning_rate": 3.52708027701254e-05, "loss": 0.1178, "step": 29601 }, { "epoch": 21.686446886446888, "grad_norm": 1.2578125, "learning_rate": 3.5267542636864024e-05, "loss": 0.1379, "step": 29602 }, { "epoch": 21.68717948717949, "grad_norm": 1.03125, "learning_rate": 3.5264282576586364e-05, "loss": 0.1209, "step": 29603 }, { "epoch": 21.68791208791209, "grad_norm": 0.69921875, "learning_rate": 3.5261022589307994e-05, "loss": 0.1357, "step": 29604 }, { "epoch": 21.68864468864469, "grad_norm": 1.140625, "learning_rate": 3.525776267504459e-05, "loss": 0.1385, "step": 29605 }, { "epoch": 21.68937728937729, "grad_norm": 1.0546875, "learning_rate": 3.525450283381175e-05, "loss": 0.1277, "step": 29606 }, { "epoch": 21.69010989010989, "grad_norm": 1.0859375, "learning_rate": 3.5251243065625094e-05, "loss": 0.1021, "step": 29607 }, { "epoch": 21.690842490842492, "grad_norm": 0.6796875, "learning_rate": 3.524798337050023e-05, "loss": 0.1173, "step": 29608 }, { "epoch": 21.691575091575093, "grad_norm": 1.3671875, "learning_rate": 3.524472374845279e-05, "loss": 0.1165, "step": 29609 }, { "epoch": 21.692307692307693, "grad_norm": 1.2265625, "learning_rate": 3.524146419949838e-05, "loss": 0.1398, "step": 29610 }, { "epoch": 21.693040293040294, "grad_norm": 1.0859375, "learning_rate": 3.5238204723652626e-05, "loss": 0.1188, "step": 29611 }, { "epoch": 21.693772893772895, "grad_norm": 1.0390625, "learning_rate": 3.523494532093115e-05, "loss": 0.1276, "step": 29612 }, { "epoch": 21.694505494505496, "grad_norm": 0.9140625, "learning_rate": 3.523168599134956e-05, "loss": 0.1098, "step": 29613 }, { "epoch": 21.695238095238096, "grad_norm": 1.65625, "learning_rate": 3.522842673492346e-05, "loss": 0.1271, "step": 29614 }, { "epoch": 21.695970695970697, "grad_norm": 0.84375, "learning_rate": 3.522516755166849e-05, "loss": 0.1356, "step": 29615 }, { "epoch": 21.696703296703298, "grad_norm": 1.25, "learning_rate": 3.522190844160027e-05, "loss": 0.113, "step": 29616 }, { "epoch": 21.6974358974359, "grad_norm": 1.0546875, "learning_rate": 3.521864940473439e-05, "loss": 0.1235, "step": 29617 }, { "epoch": 21.6981684981685, "grad_norm": 1.015625, "learning_rate": 3.5215390441086486e-05, "loss": 0.1281, "step": 29618 }, { "epoch": 21.6989010989011, "grad_norm": 1.109375, "learning_rate": 3.5212131550672144e-05, "loss": 0.1218, "step": 29619 }, { "epoch": 21.6996336996337, "grad_norm": 1.140625, "learning_rate": 3.5208872733507016e-05, "loss": 0.1035, "step": 29620 }, { "epoch": 21.7003663003663, "grad_norm": 1.8828125, "learning_rate": 3.52056139896067e-05, "loss": 0.1008, "step": 29621 }, { "epoch": 21.701098901098902, "grad_norm": 1.2109375, "learning_rate": 3.5202355318986805e-05, "loss": 0.1277, "step": 29622 }, { "epoch": 21.701831501831503, "grad_norm": 0.99609375, "learning_rate": 3.519909672166294e-05, "loss": 0.1262, "step": 29623 }, { "epoch": 21.702564102564104, "grad_norm": 1.265625, "learning_rate": 3.519583819765074e-05, "loss": 0.1212, "step": 29624 }, { "epoch": 21.703296703296704, "grad_norm": 1.2578125, "learning_rate": 3.519257974696577e-05, "loss": 0.1199, "step": 29625 }, { "epoch": 21.704029304029305, "grad_norm": 1.1953125, "learning_rate": 3.51893213696237e-05, "loss": 0.112, "step": 29626 }, { "epoch": 21.704761904761906, "grad_norm": 0.921875, "learning_rate": 3.5186063065640124e-05, "loss": 0.1067, "step": 29627 }, { "epoch": 21.705494505494507, "grad_norm": 1.234375, "learning_rate": 3.5182804835030655e-05, "loss": 0.1154, "step": 29628 }, { "epoch": 21.706227106227107, "grad_norm": 1.5625, "learning_rate": 3.517954667781088e-05, "loss": 0.1432, "step": 29629 }, { "epoch": 21.706959706959708, "grad_norm": 0.91015625, "learning_rate": 3.517628859399641e-05, "loss": 0.1163, "step": 29630 }, { "epoch": 21.70769230769231, "grad_norm": 0.6015625, "learning_rate": 3.517303058360289e-05, "loss": 0.096, "step": 29631 }, { "epoch": 21.70842490842491, "grad_norm": 0.76171875, "learning_rate": 3.516977264664592e-05, "loss": 0.118, "step": 29632 }, { "epoch": 21.70915750915751, "grad_norm": 0.9140625, "learning_rate": 3.51665147831411e-05, "loss": 0.1241, "step": 29633 }, { "epoch": 21.70989010989011, "grad_norm": 0.95703125, "learning_rate": 3.516325699310402e-05, "loss": 0.1217, "step": 29634 }, { "epoch": 21.71062271062271, "grad_norm": 0.60546875, "learning_rate": 3.515999927655032e-05, "loss": 0.1111, "step": 29635 }, { "epoch": 21.711355311355312, "grad_norm": 1.015625, "learning_rate": 3.515674163349561e-05, "loss": 0.1373, "step": 29636 }, { "epoch": 21.712087912087913, "grad_norm": 1.3046875, "learning_rate": 3.515348406395548e-05, "loss": 0.1288, "step": 29637 }, { "epoch": 21.712820512820514, "grad_norm": 1.0, "learning_rate": 3.515022656794555e-05, "loss": 0.1191, "step": 29638 }, { "epoch": 21.713553113553115, "grad_norm": 0.91015625, "learning_rate": 3.514696914548142e-05, "loss": 0.1221, "step": 29639 }, { "epoch": 21.714285714285715, "grad_norm": 0.98828125, "learning_rate": 3.514371179657869e-05, "loss": 0.1433, "step": 29640 }, { "epoch": 21.715018315018316, "grad_norm": 1.265625, "learning_rate": 3.5140454521253e-05, "loss": 0.1077, "step": 29641 }, { "epoch": 21.715750915750917, "grad_norm": 1.03125, "learning_rate": 3.513719731951993e-05, "loss": 0.1304, "step": 29642 }, { "epoch": 21.716483516483517, "grad_norm": 1.3046875, "learning_rate": 3.513394019139508e-05, "loss": 0.1446, "step": 29643 }, { "epoch": 21.717216117216118, "grad_norm": 0.875, "learning_rate": 3.5130683136894076e-05, "loss": 0.093, "step": 29644 }, { "epoch": 21.71794871794872, "grad_norm": 1.265625, "learning_rate": 3.512742615603252e-05, "loss": 0.122, "step": 29645 }, { "epoch": 21.71868131868132, "grad_norm": 0.99609375, "learning_rate": 3.512416924882599e-05, "loss": 0.1084, "step": 29646 }, { "epoch": 21.71941391941392, "grad_norm": 1.140625, "learning_rate": 3.512091241529013e-05, "loss": 0.127, "step": 29647 }, { "epoch": 21.72014652014652, "grad_norm": 1.0, "learning_rate": 3.5117655655440526e-05, "loss": 0.117, "step": 29648 }, { "epoch": 21.720879120879122, "grad_norm": 0.671875, "learning_rate": 3.511439896929277e-05, "loss": 0.1255, "step": 29649 }, { "epoch": 21.721611721611723, "grad_norm": 0.76171875, "learning_rate": 3.5111142356862494e-05, "loss": 0.1439, "step": 29650 }, { "epoch": 21.722344322344323, "grad_norm": 1.3125, "learning_rate": 3.510788581816527e-05, "loss": 0.1319, "step": 29651 }, { "epoch": 21.723076923076924, "grad_norm": 1.125, "learning_rate": 3.5104629353216725e-05, "loss": 0.129, "step": 29652 }, { "epoch": 21.723809523809525, "grad_norm": 1.3515625, "learning_rate": 3.510137296203247e-05, "loss": 0.1478, "step": 29653 }, { "epoch": 21.724542124542126, "grad_norm": 1.0546875, "learning_rate": 3.509811664462808e-05, "loss": 0.1138, "step": 29654 }, { "epoch": 21.725274725274726, "grad_norm": 1.34375, "learning_rate": 3.509486040101919e-05, "loss": 0.1468, "step": 29655 }, { "epoch": 21.726007326007327, "grad_norm": 1.3046875, "learning_rate": 3.509160423122134e-05, "loss": 0.1148, "step": 29656 }, { "epoch": 21.726739926739928, "grad_norm": 0.89453125, "learning_rate": 3.50883481352502e-05, "loss": 0.1106, "step": 29657 }, { "epoch": 21.72747252747253, "grad_norm": 1.390625, "learning_rate": 3.508509211312134e-05, "loss": 0.1232, "step": 29658 }, { "epoch": 21.72820512820513, "grad_norm": 0.68359375, "learning_rate": 3.5081836164850364e-05, "loss": 0.1081, "step": 29659 }, { "epoch": 21.72893772893773, "grad_norm": 1.2734375, "learning_rate": 3.507858029045287e-05, "loss": 0.1251, "step": 29660 }, { "epoch": 21.72967032967033, "grad_norm": 1.046875, "learning_rate": 3.507532448994445e-05, "loss": 0.1807, "step": 29661 }, { "epoch": 21.73040293040293, "grad_norm": 0.97265625, "learning_rate": 3.507206876334073e-05, "loss": 0.1217, "step": 29662 }, { "epoch": 21.731135531135532, "grad_norm": 0.7421875, "learning_rate": 3.5068813110657286e-05, "loss": 0.1396, "step": 29663 }, { "epoch": 21.731868131868133, "grad_norm": 1.1640625, "learning_rate": 3.506555753190973e-05, "loss": 0.1049, "step": 29664 }, { "epoch": 21.732600732600734, "grad_norm": 1.4765625, "learning_rate": 3.506230202711365e-05, "loss": 0.1476, "step": 29665 }, { "epoch": 21.733333333333334, "grad_norm": 0.98046875, "learning_rate": 3.505904659628464e-05, "loss": 0.1166, "step": 29666 }, { "epoch": 21.734065934065935, "grad_norm": 1.375, "learning_rate": 3.505579123943832e-05, "loss": 0.1023, "step": 29667 }, { "epoch": 21.734798534798536, "grad_norm": 1.1328125, "learning_rate": 3.505253595659027e-05, "loss": 0.1373, "step": 29668 }, { "epoch": 21.735531135531136, "grad_norm": 1.4453125, "learning_rate": 3.50492807477561e-05, "loss": 0.1231, "step": 29669 }, { "epoch": 21.736263736263737, "grad_norm": 1.203125, "learning_rate": 3.5046025612951385e-05, "loss": 0.1386, "step": 29670 }, { "epoch": 21.736996336996338, "grad_norm": 1.1953125, "learning_rate": 3.504277055219175e-05, "loss": 0.1369, "step": 29671 }, { "epoch": 21.73772893772894, "grad_norm": 1.2578125, "learning_rate": 3.5039515565492745e-05, "loss": 0.1012, "step": 29672 }, { "epoch": 21.73846153846154, "grad_norm": 1.328125, "learning_rate": 3.5036260652870023e-05, "loss": 0.1072, "step": 29673 }, { "epoch": 21.73919413919414, "grad_norm": 1.109375, "learning_rate": 3.5033005814339144e-05, "loss": 0.0951, "step": 29674 }, { "epoch": 21.73992673992674, "grad_norm": 0.8515625, "learning_rate": 3.502975104991571e-05, "loss": 0.0949, "step": 29675 }, { "epoch": 21.74065934065934, "grad_norm": 0.70703125, "learning_rate": 3.5026496359615336e-05, "loss": 0.1292, "step": 29676 }, { "epoch": 21.741391941391942, "grad_norm": 0.625, "learning_rate": 3.5023241743453556e-05, "loss": 0.1115, "step": 29677 }, { "epoch": 21.742124542124543, "grad_norm": 1.6015625, "learning_rate": 3.501998720144604e-05, "loss": 0.1513, "step": 29678 }, { "epoch": 21.742857142857144, "grad_norm": 1.0234375, "learning_rate": 3.5016732733608345e-05, "loss": 0.1217, "step": 29679 }, { "epoch": 21.743589743589745, "grad_norm": 1.078125, "learning_rate": 3.501347833995606e-05, "loss": 0.0947, "step": 29680 }, { "epoch": 21.744322344322345, "grad_norm": 1.2421875, "learning_rate": 3.5010224020504795e-05, "loss": 0.1148, "step": 29681 }, { "epoch": 21.745054945054946, "grad_norm": 0.890625, "learning_rate": 3.500696977527011e-05, "loss": 0.1277, "step": 29682 }, { "epoch": 21.745787545787547, "grad_norm": 0.91796875, "learning_rate": 3.5003715604267625e-05, "loss": 0.11, "step": 29683 }, { "epoch": 21.746520146520147, "grad_norm": 0.7265625, "learning_rate": 3.5000461507512934e-05, "loss": 0.1217, "step": 29684 }, { "epoch": 21.747252747252748, "grad_norm": 0.7109375, "learning_rate": 3.4997207485021626e-05, "loss": 0.0937, "step": 29685 }, { "epoch": 21.74798534798535, "grad_norm": 0.5390625, "learning_rate": 3.499395353680927e-05, "loss": 0.1161, "step": 29686 }, { "epoch": 21.74871794871795, "grad_norm": 0.81640625, "learning_rate": 3.4990699662891475e-05, "loss": 0.1282, "step": 29687 }, { "epoch": 21.74945054945055, "grad_norm": 0.73828125, "learning_rate": 3.4987445863283836e-05, "loss": 0.1173, "step": 29688 }, { "epoch": 21.75018315018315, "grad_norm": 1.1796875, "learning_rate": 3.498419213800192e-05, "loss": 0.1376, "step": 29689 }, { "epoch": 21.750915750915752, "grad_norm": 1.078125, "learning_rate": 3.498093848706136e-05, "loss": 0.1272, "step": 29690 }, { "epoch": 21.751648351648353, "grad_norm": 0.85546875, "learning_rate": 3.4977684910477706e-05, "loss": 0.0925, "step": 29691 }, { "epoch": 21.752380952380953, "grad_norm": 1.25, "learning_rate": 3.497443140826654e-05, "loss": 0.126, "step": 29692 }, { "epoch": 21.753113553113554, "grad_norm": 0.7578125, "learning_rate": 3.4971177980443495e-05, "loss": 0.1122, "step": 29693 }, { "epoch": 21.753846153846155, "grad_norm": 0.87890625, "learning_rate": 3.4967924627024135e-05, "loss": 0.1301, "step": 29694 }, { "epoch": 21.754578754578755, "grad_norm": 1.109375, "learning_rate": 3.496467134802404e-05, "loss": 0.111, "step": 29695 }, { "epoch": 21.755311355311356, "grad_norm": 1.625, "learning_rate": 3.4961418143458816e-05, "loss": 0.1272, "step": 29696 }, { "epoch": 21.756043956043957, "grad_norm": 1.4296875, "learning_rate": 3.4958165013344026e-05, "loss": 0.1117, "step": 29697 }, { "epoch": 21.756776556776558, "grad_norm": 1.2578125, "learning_rate": 3.495491195769525e-05, "loss": 0.174, "step": 29698 }, { "epoch": 21.75750915750916, "grad_norm": 0.890625, "learning_rate": 3.4951658976528115e-05, "loss": 0.1147, "step": 29699 }, { "epoch": 21.75824175824176, "grad_norm": 0.96484375, "learning_rate": 3.4948406069858184e-05, "loss": 0.1012, "step": 29700 }, { "epoch": 21.75897435897436, "grad_norm": 1.28125, "learning_rate": 3.494515323770106e-05, "loss": 0.1413, "step": 29701 }, { "epoch": 21.75970695970696, "grad_norm": 2.0, "learning_rate": 3.494190048007229e-05, "loss": 0.1452, "step": 29702 }, { "epoch": 21.76043956043956, "grad_norm": 1.5234375, "learning_rate": 3.4938647796987476e-05, "loss": 0.1358, "step": 29703 }, { "epoch": 21.761172161172162, "grad_norm": 1.046875, "learning_rate": 3.493539518846224e-05, "loss": 0.1207, "step": 29704 }, { "epoch": 21.761904761904763, "grad_norm": 0.85546875, "learning_rate": 3.4932142654512115e-05, "loss": 0.104, "step": 29705 }, { "epoch": 21.762637362637363, "grad_norm": 0.6796875, "learning_rate": 3.4928890195152714e-05, "loss": 0.124, "step": 29706 }, { "epoch": 21.763369963369964, "grad_norm": 1.078125, "learning_rate": 3.492563781039961e-05, "loss": 0.0872, "step": 29707 }, { "epoch": 21.764102564102565, "grad_norm": 0.80078125, "learning_rate": 3.4922385500268374e-05, "loss": 0.1158, "step": 29708 }, { "epoch": 21.764835164835166, "grad_norm": 0.79296875, "learning_rate": 3.491913326477462e-05, "loss": 0.1047, "step": 29709 }, { "epoch": 21.765567765567766, "grad_norm": 1.2109375, "learning_rate": 3.491588110393391e-05, "loss": 0.1108, "step": 29710 }, { "epoch": 21.766300366300367, "grad_norm": 0.9453125, "learning_rate": 3.491262901776183e-05, "loss": 0.1075, "step": 29711 }, { "epoch": 21.767032967032968, "grad_norm": 1.1875, "learning_rate": 3.490937700627395e-05, "loss": 0.1337, "step": 29712 }, { "epoch": 21.76776556776557, "grad_norm": 0.828125, "learning_rate": 3.490612506948586e-05, "loss": 0.1064, "step": 29713 }, { "epoch": 21.76849816849817, "grad_norm": 1.203125, "learning_rate": 3.490287320741317e-05, "loss": 0.1133, "step": 29714 }, { "epoch": 21.76923076923077, "grad_norm": 1.0, "learning_rate": 3.489962142007141e-05, "loss": 0.1225, "step": 29715 }, { "epoch": 21.76996336996337, "grad_norm": 1.0859375, "learning_rate": 3.48963697074762e-05, "loss": 0.1214, "step": 29716 }, { "epoch": 21.77069597069597, "grad_norm": 1.03125, "learning_rate": 3.4893118069643105e-05, "loss": 0.1234, "step": 29717 }, { "epoch": 21.771428571428572, "grad_norm": 1.34375, "learning_rate": 3.488986650658768e-05, "loss": 0.1425, "step": 29718 }, { "epoch": 21.772161172161173, "grad_norm": 1.625, "learning_rate": 3.488661501832555e-05, "loss": 0.1266, "step": 29719 }, { "epoch": 21.772893772893774, "grad_norm": 1.375, "learning_rate": 3.488336360487228e-05, "loss": 0.1471, "step": 29720 }, { "epoch": 21.773626373626374, "grad_norm": 1.171875, "learning_rate": 3.488011226624344e-05, "loss": 0.1379, "step": 29721 }, { "epoch": 21.774358974358975, "grad_norm": 1.296875, "learning_rate": 3.487686100245461e-05, "loss": 0.1347, "step": 29722 }, { "epoch": 21.775091575091576, "grad_norm": 1.421875, "learning_rate": 3.487360981352135e-05, "loss": 0.1517, "step": 29723 }, { "epoch": 21.775824175824177, "grad_norm": 1.6328125, "learning_rate": 3.487035869945926e-05, "loss": 0.1452, "step": 29724 }, { "epoch": 21.776556776556777, "grad_norm": 1.09375, "learning_rate": 3.4867107660283923e-05, "loss": 0.1253, "step": 29725 }, { "epoch": 21.777289377289378, "grad_norm": 1.40625, "learning_rate": 3.486385669601089e-05, "loss": 0.1517, "step": 29726 }, { "epoch": 21.77802197802198, "grad_norm": 1.40625, "learning_rate": 3.4860605806655764e-05, "loss": 0.1209, "step": 29727 }, { "epoch": 21.77875457875458, "grad_norm": 1.1484375, "learning_rate": 3.485735499223412e-05, "loss": 0.1399, "step": 29728 }, { "epoch": 21.77948717948718, "grad_norm": 1.1796875, "learning_rate": 3.48541042527615e-05, "loss": 0.1376, "step": 29729 }, { "epoch": 21.78021978021978, "grad_norm": 1.0, "learning_rate": 3.485085358825351e-05, "loss": 0.1336, "step": 29730 }, { "epoch": 21.78095238095238, "grad_norm": 1.21875, "learning_rate": 3.4847602998725734e-05, "loss": 0.1265, "step": 29731 }, { "epoch": 21.781684981684982, "grad_norm": 0.8984375, "learning_rate": 3.484435248419372e-05, "loss": 0.1315, "step": 29732 }, { "epoch": 21.782417582417583, "grad_norm": 1.3125, "learning_rate": 3.484110204467305e-05, "loss": 0.1367, "step": 29733 }, { "epoch": 21.783150183150184, "grad_norm": 1.09375, "learning_rate": 3.483785168017929e-05, "loss": 0.1301, "step": 29734 }, { "epoch": 21.783882783882785, "grad_norm": 1.0703125, "learning_rate": 3.483460139072804e-05, "loss": 0.093, "step": 29735 }, { "epoch": 21.784615384615385, "grad_norm": 0.7109375, "learning_rate": 3.4831351176334864e-05, "loss": 0.1288, "step": 29736 }, { "epoch": 21.785347985347986, "grad_norm": 0.875, "learning_rate": 3.482810103701532e-05, "loss": 0.106, "step": 29737 }, { "epoch": 21.786080586080587, "grad_norm": 0.828125, "learning_rate": 3.482485097278498e-05, "loss": 0.1449, "step": 29738 }, { "epoch": 21.786813186813188, "grad_norm": 1.0859375, "learning_rate": 3.482160098365943e-05, "loss": 0.1302, "step": 29739 }, { "epoch": 21.78754578754579, "grad_norm": 1.0859375, "learning_rate": 3.481835106965423e-05, "loss": 0.1336, "step": 29740 }, { "epoch": 21.78827838827839, "grad_norm": 0.98828125, "learning_rate": 3.481510123078497e-05, "loss": 0.1163, "step": 29741 }, { "epoch": 21.78901098901099, "grad_norm": 1.28125, "learning_rate": 3.4811851467067205e-05, "loss": 0.1134, "step": 29742 }, { "epoch": 21.78974358974359, "grad_norm": 1.9453125, "learning_rate": 3.480860177851652e-05, "loss": 0.1346, "step": 29743 }, { "epoch": 21.79047619047619, "grad_norm": 0.7578125, "learning_rate": 3.480535216514844e-05, "loss": 0.1002, "step": 29744 }, { "epoch": 21.791208791208792, "grad_norm": 0.91015625, "learning_rate": 3.48021026269786e-05, "loss": 0.1403, "step": 29745 }, { "epoch": 21.791941391941393, "grad_norm": 0.79296875, "learning_rate": 3.4798853164022544e-05, "loss": 0.1233, "step": 29746 }, { "epoch": 21.792673992673993, "grad_norm": 0.765625, "learning_rate": 3.4795603776295834e-05, "loss": 0.1165, "step": 29747 }, { "epoch": 21.793406593406594, "grad_norm": 1.1640625, "learning_rate": 3.4792354463814036e-05, "loss": 0.1379, "step": 29748 }, { "epoch": 21.794139194139195, "grad_norm": 0.8046875, "learning_rate": 3.4789105226592713e-05, "loss": 0.1028, "step": 29749 }, { "epoch": 21.794871794871796, "grad_norm": 0.8984375, "learning_rate": 3.4785856064647447e-05, "loss": 0.1208, "step": 29750 }, { "epoch": 21.795604395604396, "grad_norm": 0.94921875, "learning_rate": 3.4782606977993813e-05, "loss": 0.1015, "step": 29751 }, { "epoch": 21.796336996336997, "grad_norm": 0.8046875, "learning_rate": 3.4779357966647366e-05, "loss": 0.1329, "step": 29752 }, { "epoch": 21.797069597069598, "grad_norm": 1.0390625, "learning_rate": 3.4776109030623675e-05, "loss": 0.1289, "step": 29753 }, { "epoch": 21.7978021978022, "grad_norm": 1.3046875, "learning_rate": 3.4772860169938314e-05, "loss": 0.1471, "step": 29754 }, { "epoch": 21.7985347985348, "grad_norm": 1.2109375, "learning_rate": 3.4769611384606826e-05, "loss": 0.116, "step": 29755 }, { "epoch": 21.7992673992674, "grad_norm": 1.0390625, "learning_rate": 3.476636267464481e-05, "loss": 0.1257, "step": 29756 }, { "epoch": 21.8, "grad_norm": 0.7890625, "learning_rate": 3.4763114040067814e-05, "loss": 0.088, "step": 29757 }, { "epoch": 21.8007326007326, "grad_norm": 1.390625, "learning_rate": 3.4759865480891406e-05, "loss": 0.1372, "step": 29758 }, { "epoch": 21.801465201465202, "grad_norm": 0.6875, "learning_rate": 3.475661699713114e-05, "loss": 0.1234, "step": 29759 }, { "epoch": 21.802197802197803, "grad_norm": 0.98828125, "learning_rate": 3.4753368588802584e-05, "loss": 0.1491, "step": 29760 }, { "epoch": 21.802930402930404, "grad_norm": 1.0625, "learning_rate": 3.475012025592133e-05, "loss": 0.1308, "step": 29761 }, { "epoch": 21.803663003663004, "grad_norm": 1.0078125, "learning_rate": 3.4746871998502914e-05, "loss": 0.1308, "step": 29762 }, { "epoch": 21.804395604395605, "grad_norm": 0.79296875, "learning_rate": 3.4743623816562895e-05, "loss": 0.0946, "step": 29763 }, { "epoch": 21.805128205128206, "grad_norm": 1.359375, "learning_rate": 3.474037571011685e-05, "loss": 0.1086, "step": 29764 }, { "epoch": 21.805860805860807, "grad_norm": 0.94140625, "learning_rate": 3.4737127679180335e-05, "loss": 0.0969, "step": 29765 }, { "epoch": 21.806593406593407, "grad_norm": 1.6484375, "learning_rate": 3.473387972376892e-05, "loss": 0.1181, "step": 29766 }, { "epoch": 21.807326007326008, "grad_norm": 1.0390625, "learning_rate": 3.4730631843898174e-05, "loss": 0.1505, "step": 29767 }, { "epoch": 21.80805860805861, "grad_norm": 1.078125, "learning_rate": 3.472738403958364e-05, "loss": 0.1311, "step": 29768 }, { "epoch": 21.80879120879121, "grad_norm": 0.83203125, "learning_rate": 3.472413631084089e-05, "loss": 0.1212, "step": 29769 }, { "epoch": 21.80952380952381, "grad_norm": 0.98828125, "learning_rate": 3.472088865768546e-05, "loss": 0.0933, "step": 29770 }, { "epoch": 21.81025641025641, "grad_norm": 1.1640625, "learning_rate": 3.4717641080132956e-05, "loss": 0.1133, "step": 29771 }, { "epoch": 21.81098901098901, "grad_norm": 0.98046875, "learning_rate": 3.4714393578198915e-05, "loss": 0.106, "step": 29772 }, { "epoch": 21.811721611721612, "grad_norm": 1.2890625, "learning_rate": 3.4711146151898886e-05, "loss": 0.1228, "step": 29773 }, { "epoch": 21.812454212454213, "grad_norm": 0.7265625, "learning_rate": 3.470789880124845e-05, "loss": 0.0976, "step": 29774 }, { "epoch": 21.813186813186814, "grad_norm": 1.5390625, "learning_rate": 3.470465152626314e-05, "loss": 0.1396, "step": 29775 }, { "epoch": 21.813919413919415, "grad_norm": 1.0859375, "learning_rate": 3.4701404326958537e-05, "loss": 0.1185, "step": 29776 }, { "epoch": 21.814652014652015, "grad_norm": 0.84765625, "learning_rate": 3.4698157203350196e-05, "loss": 0.1312, "step": 29777 }, { "epoch": 21.815384615384616, "grad_norm": 0.84765625, "learning_rate": 3.4694910155453666e-05, "loss": 0.1098, "step": 29778 }, { "epoch": 21.816117216117217, "grad_norm": 1.4765625, "learning_rate": 3.469166318328451e-05, "loss": 0.116, "step": 29779 }, { "epoch": 21.816849816849818, "grad_norm": 1.5625, "learning_rate": 3.46884162868583e-05, "loss": 0.1582, "step": 29780 }, { "epoch": 21.81758241758242, "grad_norm": 0.84375, "learning_rate": 3.468516946619054e-05, "loss": 0.1362, "step": 29781 }, { "epoch": 21.81831501831502, "grad_norm": 1.5390625, "learning_rate": 3.468192272129685e-05, "loss": 0.1234, "step": 29782 }, { "epoch": 21.81904761904762, "grad_norm": 1.640625, "learning_rate": 3.467867605219278e-05, "loss": 0.1705, "step": 29783 }, { "epoch": 21.81978021978022, "grad_norm": 1.296875, "learning_rate": 3.467542945889385e-05, "loss": 0.1588, "step": 29784 }, { "epoch": 21.82051282051282, "grad_norm": 1.28125, "learning_rate": 3.4672182941415634e-05, "loss": 0.1329, "step": 29785 }, { "epoch": 21.821245421245422, "grad_norm": 1.6796875, "learning_rate": 3.466893649977367e-05, "loss": 0.1335, "step": 29786 }, { "epoch": 21.821978021978023, "grad_norm": 0.98046875, "learning_rate": 3.466569013398354e-05, "loss": 0.1207, "step": 29787 }, { "epoch": 21.822710622710623, "grad_norm": 1.25, "learning_rate": 3.4662443844060796e-05, "loss": 0.1208, "step": 29788 }, { "epoch": 21.823443223443224, "grad_norm": 1.1171875, "learning_rate": 3.4659197630020965e-05, "loss": 0.1284, "step": 29789 }, { "epoch": 21.824175824175825, "grad_norm": 1.5234375, "learning_rate": 3.465595149187963e-05, "loss": 0.1633, "step": 29790 }, { "epoch": 21.824908424908426, "grad_norm": 0.90625, "learning_rate": 3.465270542965233e-05, "loss": 0.1192, "step": 29791 }, { "epoch": 21.825641025641026, "grad_norm": 1.3203125, "learning_rate": 3.4649459443354604e-05, "loss": 0.1862, "step": 29792 }, { "epoch": 21.826373626373627, "grad_norm": 1.1171875, "learning_rate": 3.4646213533002044e-05, "loss": 0.1176, "step": 29793 }, { "epoch": 21.827106227106228, "grad_norm": 1.25, "learning_rate": 3.464296769861017e-05, "loss": 0.1162, "step": 29794 }, { "epoch": 21.82783882783883, "grad_norm": 0.72265625, "learning_rate": 3.463972194019456e-05, "loss": 0.1151, "step": 29795 }, { "epoch": 21.82857142857143, "grad_norm": 1.515625, "learning_rate": 3.4636476257770714e-05, "loss": 0.0971, "step": 29796 }, { "epoch": 21.82930402930403, "grad_norm": 1.4140625, "learning_rate": 3.463323065135424e-05, "loss": 0.1424, "step": 29797 }, { "epoch": 21.83003663003663, "grad_norm": 0.66796875, "learning_rate": 3.462998512096067e-05, "loss": 0.1076, "step": 29798 }, { "epoch": 21.83076923076923, "grad_norm": 0.9609375, "learning_rate": 3.4626739666605556e-05, "loss": 0.1172, "step": 29799 }, { "epoch": 21.831501831501832, "grad_norm": 1.09375, "learning_rate": 3.462349428830444e-05, "loss": 0.1276, "step": 29800 }, { "epoch": 21.832234432234433, "grad_norm": 1.390625, "learning_rate": 3.462024898607286e-05, "loss": 0.1482, "step": 29801 }, { "epoch": 21.832967032967034, "grad_norm": 1.109375, "learning_rate": 3.4617003759926375e-05, "loss": 0.1379, "step": 29802 }, { "epoch": 21.833699633699634, "grad_norm": 0.83984375, "learning_rate": 3.461375860988056e-05, "loss": 0.1078, "step": 29803 }, { "epoch": 21.834432234432235, "grad_norm": 0.83984375, "learning_rate": 3.461051353595093e-05, "loss": 0.1282, "step": 29804 }, { "epoch": 21.835164835164836, "grad_norm": 0.765625, "learning_rate": 3.4607268538153054e-05, "loss": 0.0995, "step": 29805 }, { "epoch": 21.835897435897436, "grad_norm": 1.3203125, "learning_rate": 3.4604023616502474e-05, "loss": 0.1584, "step": 29806 }, { "epoch": 21.836630036630037, "grad_norm": 0.96484375, "learning_rate": 3.460077877101472e-05, "loss": 0.12, "step": 29807 }, { "epoch": 21.837362637362638, "grad_norm": 0.95703125, "learning_rate": 3.4597534001705364e-05, "loss": 0.139, "step": 29808 }, { "epoch": 21.83809523809524, "grad_norm": 0.8671875, "learning_rate": 3.4594289308589944e-05, "loss": 0.1245, "step": 29809 }, { "epoch": 21.83882783882784, "grad_norm": 1.3203125, "learning_rate": 3.459104469168401e-05, "loss": 0.1598, "step": 29810 }, { "epoch": 21.83956043956044, "grad_norm": 1.203125, "learning_rate": 3.4587800151003106e-05, "loss": 0.1339, "step": 29811 }, { "epoch": 21.84029304029304, "grad_norm": 0.81640625, "learning_rate": 3.458455568656275e-05, "loss": 0.1223, "step": 29812 }, { "epoch": 21.84102564102564, "grad_norm": 0.87109375, "learning_rate": 3.458131129837854e-05, "loss": 0.1299, "step": 29813 }, { "epoch": 21.841758241758242, "grad_norm": 0.81640625, "learning_rate": 3.457806698646598e-05, "loss": 0.1182, "step": 29814 }, { "epoch": 21.842490842490843, "grad_norm": 0.99609375, "learning_rate": 3.4574822750840626e-05, "loss": 0.1245, "step": 29815 }, { "epoch": 21.843223443223444, "grad_norm": 0.94921875, "learning_rate": 3.457157859151803e-05, "loss": 0.1272, "step": 29816 }, { "epoch": 21.843956043956045, "grad_norm": 0.80078125, "learning_rate": 3.456833450851372e-05, "loss": 0.1288, "step": 29817 }, { "epoch": 21.844688644688645, "grad_norm": 1.03125, "learning_rate": 3.4565090501843255e-05, "loss": 0.1302, "step": 29818 }, { "epoch": 21.845421245421246, "grad_norm": 1.375, "learning_rate": 3.456184657152218e-05, "loss": 0.1152, "step": 29819 }, { "epoch": 21.846153846153847, "grad_norm": 1.703125, "learning_rate": 3.455860271756604e-05, "loss": 0.1582, "step": 29820 }, { "epoch": 21.846886446886447, "grad_norm": 0.8046875, "learning_rate": 3.455535893999035e-05, "loss": 0.1017, "step": 29821 }, { "epoch": 21.847619047619048, "grad_norm": 1.0625, "learning_rate": 3.4552115238810665e-05, "loss": 0.1004, "step": 29822 }, { "epoch": 21.84835164835165, "grad_norm": 0.7421875, "learning_rate": 3.454887161404253e-05, "loss": 0.1128, "step": 29823 }, { "epoch": 21.84908424908425, "grad_norm": 0.7578125, "learning_rate": 3.4545628065701506e-05, "loss": 0.1075, "step": 29824 }, { "epoch": 21.84981684981685, "grad_norm": 1.0859375, "learning_rate": 3.454238459380311e-05, "loss": 0.1012, "step": 29825 }, { "epoch": 21.85054945054945, "grad_norm": 0.75, "learning_rate": 3.453914119836289e-05, "loss": 0.0972, "step": 29826 }, { "epoch": 21.851282051282052, "grad_norm": 1.03125, "learning_rate": 3.453589787939637e-05, "loss": 0.1413, "step": 29827 }, { "epoch": 21.852014652014653, "grad_norm": 1.1171875, "learning_rate": 3.453265463691911e-05, "loss": 0.1426, "step": 29828 }, { "epoch": 21.852747252747253, "grad_norm": 0.73828125, "learning_rate": 3.4529411470946644e-05, "loss": 0.1082, "step": 29829 }, { "epoch": 21.853479853479854, "grad_norm": 1.484375, "learning_rate": 3.452616838149451e-05, "loss": 0.1398, "step": 29830 }, { "epoch": 21.854212454212455, "grad_norm": 0.77734375, "learning_rate": 3.452292536857826e-05, "loss": 0.0868, "step": 29831 }, { "epoch": 21.854945054945055, "grad_norm": 1.125, "learning_rate": 3.4519682432213406e-05, "loss": 0.1283, "step": 29832 }, { "epoch": 21.855677655677656, "grad_norm": 0.98046875, "learning_rate": 3.451643957241549e-05, "loss": 0.1517, "step": 29833 }, { "epoch": 21.856410256410257, "grad_norm": 1.0078125, "learning_rate": 3.451319678920007e-05, "loss": 0.1123, "step": 29834 }, { "epoch": 21.857142857142858, "grad_norm": 1.109375, "learning_rate": 3.450995408258267e-05, "loss": 0.1134, "step": 29835 }, { "epoch": 21.85787545787546, "grad_norm": 1.078125, "learning_rate": 3.450671145257885e-05, "loss": 0.1149, "step": 29836 }, { "epoch": 21.85860805860806, "grad_norm": 1.21875, "learning_rate": 3.450346889920411e-05, "loss": 0.1329, "step": 29837 }, { "epoch": 21.85934065934066, "grad_norm": 1.0546875, "learning_rate": 3.450022642247398e-05, "loss": 0.1389, "step": 29838 }, { "epoch": 21.86007326007326, "grad_norm": 1.4765625, "learning_rate": 3.449698402240404e-05, "loss": 0.1255, "step": 29839 }, { "epoch": 21.86080586080586, "grad_norm": 0.90234375, "learning_rate": 3.449374169900981e-05, "loss": 0.1161, "step": 29840 }, { "epoch": 21.861538461538462, "grad_norm": 0.75390625, "learning_rate": 3.44904994523068e-05, "loss": 0.1228, "step": 29841 }, { "epoch": 21.862271062271063, "grad_norm": 1.1875, "learning_rate": 3.448725728231058e-05, "loss": 0.1161, "step": 29842 }, { "epoch": 21.863003663003663, "grad_norm": 1.125, "learning_rate": 3.448401518903665e-05, "loss": 0.1177, "step": 29843 }, { "epoch": 21.863736263736264, "grad_norm": 0.67578125, "learning_rate": 3.448077317250056e-05, "loss": 0.133, "step": 29844 }, { "epoch": 21.864468864468865, "grad_norm": 1.5, "learning_rate": 3.4477531232717865e-05, "loss": 0.1411, "step": 29845 }, { "epoch": 21.865201465201466, "grad_norm": 0.89453125, "learning_rate": 3.4474289369704063e-05, "loss": 0.1071, "step": 29846 }, { "epoch": 21.865934065934066, "grad_norm": 0.9140625, "learning_rate": 3.447104758347471e-05, "loss": 0.1391, "step": 29847 }, { "epoch": 21.866666666666667, "grad_norm": 1.4296875, "learning_rate": 3.446780587404533e-05, "loss": 0.136, "step": 29848 }, { "epoch": 21.867399267399268, "grad_norm": 0.75390625, "learning_rate": 3.446456424143144e-05, "loss": 0.1053, "step": 29849 }, { "epoch": 21.86813186813187, "grad_norm": 1.5859375, "learning_rate": 3.44613226856486e-05, "loss": 0.1271, "step": 29850 }, { "epoch": 21.86886446886447, "grad_norm": 0.98828125, "learning_rate": 3.445808120671233e-05, "loss": 0.1102, "step": 29851 }, { "epoch": 21.86959706959707, "grad_norm": 1.109375, "learning_rate": 3.445483980463816e-05, "loss": 0.1068, "step": 29852 }, { "epoch": 21.87032967032967, "grad_norm": 1.140625, "learning_rate": 3.44515984794416e-05, "loss": 0.1216, "step": 29853 }, { "epoch": 21.87106227106227, "grad_norm": 0.7890625, "learning_rate": 3.444835723113821e-05, "loss": 0.1177, "step": 29854 }, { "epoch": 21.871794871794872, "grad_norm": 0.703125, "learning_rate": 3.4445116059743514e-05, "loss": 0.1182, "step": 29855 }, { "epoch": 21.872527472527473, "grad_norm": 1.671875, "learning_rate": 3.444187496527304e-05, "loss": 0.1118, "step": 29856 }, { "epoch": 21.873260073260074, "grad_norm": 1.8828125, "learning_rate": 3.4438633947742315e-05, "loss": 0.1305, "step": 29857 }, { "epoch": 21.873992673992674, "grad_norm": 1.7421875, "learning_rate": 3.443539300716687e-05, "loss": 0.1104, "step": 29858 }, { "epoch": 21.874725274725275, "grad_norm": 0.99609375, "learning_rate": 3.443215214356221e-05, "loss": 0.1426, "step": 29859 }, { "epoch": 21.875457875457876, "grad_norm": 1.9609375, "learning_rate": 3.44289113569439e-05, "loss": 0.159, "step": 29860 }, { "epoch": 21.876190476190477, "grad_norm": 0.75, "learning_rate": 3.442567064732746e-05, "loss": 0.1003, "step": 29861 }, { "epoch": 21.876923076923077, "grad_norm": 1.21875, "learning_rate": 3.4422430014728405e-05, "loss": 0.1176, "step": 29862 }, { "epoch": 21.877655677655678, "grad_norm": 1.171875, "learning_rate": 3.4419189459162263e-05, "loss": 0.1716, "step": 29863 }, { "epoch": 21.87838827838828, "grad_norm": 0.9296875, "learning_rate": 3.441594898064454e-05, "loss": 0.1078, "step": 29864 }, { "epoch": 21.87912087912088, "grad_norm": 0.94921875, "learning_rate": 3.441270857919081e-05, "loss": 0.1355, "step": 29865 }, { "epoch": 21.87985347985348, "grad_norm": 1.03125, "learning_rate": 3.4409468254816576e-05, "loss": 0.1162, "step": 29866 }, { "epoch": 21.88058608058608, "grad_norm": 1.6875, "learning_rate": 3.440622800753735e-05, "loss": 0.1516, "step": 29867 }, { "epoch": 21.88131868131868, "grad_norm": 1.6484375, "learning_rate": 3.4402987837368674e-05, "loss": 0.1272, "step": 29868 }, { "epoch": 21.882051282051282, "grad_norm": 1.03125, "learning_rate": 3.4399747744326066e-05, "loss": 0.128, "step": 29869 }, { "epoch": 21.882783882783883, "grad_norm": 1.0625, "learning_rate": 3.4396507728425046e-05, "loss": 0.1102, "step": 29870 }, { "epoch": 21.883516483516484, "grad_norm": 0.72265625, "learning_rate": 3.439326778968116e-05, "loss": 0.1236, "step": 29871 }, { "epoch": 21.884249084249085, "grad_norm": 0.97265625, "learning_rate": 3.439002792810991e-05, "loss": 0.1173, "step": 29872 }, { "epoch": 21.884981684981685, "grad_norm": 0.734375, "learning_rate": 3.438678814372682e-05, "loss": 0.1162, "step": 29873 }, { "epoch": 21.885714285714286, "grad_norm": 1.1484375, "learning_rate": 3.438354843654743e-05, "loss": 0.1168, "step": 29874 }, { "epoch": 21.886446886446887, "grad_norm": 1.3671875, "learning_rate": 3.438030880658723e-05, "loss": 0.1158, "step": 29875 }, { "epoch": 21.887179487179488, "grad_norm": 0.86328125, "learning_rate": 3.437706925386177e-05, "loss": 0.1191, "step": 29876 }, { "epoch": 21.88791208791209, "grad_norm": 0.66015625, "learning_rate": 3.437382977838658e-05, "loss": 0.0876, "step": 29877 }, { "epoch": 21.88864468864469, "grad_norm": 0.953125, "learning_rate": 3.437059038017715e-05, "loss": 0.1131, "step": 29878 }, { "epoch": 21.88937728937729, "grad_norm": 1.2109375, "learning_rate": 3.436735105924902e-05, "loss": 0.1274, "step": 29879 }, { "epoch": 21.89010989010989, "grad_norm": 0.86328125, "learning_rate": 3.436411181561769e-05, "loss": 0.0924, "step": 29880 }, { "epoch": 21.89084249084249, "grad_norm": 0.81640625, "learning_rate": 3.436087264929871e-05, "loss": 0.0891, "step": 29881 }, { "epoch": 21.891575091575092, "grad_norm": 1.375, "learning_rate": 3.435763356030759e-05, "loss": 0.1375, "step": 29882 }, { "epoch": 21.892307692307693, "grad_norm": 1.6484375, "learning_rate": 3.435439454865985e-05, "loss": 0.1358, "step": 29883 }, { "epoch": 21.893040293040293, "grad_norm": 0.8203125, "learning_rate": 3.435115561437101e-05, "loss": 0.1271, "step": 29884 }, { "epoch": 21.893772893772894, "grad_norm": 1.0546875, "learning_rate": 3.434791675745656e-05, "loss": 0.1203, "step": 29885 }, { "epoch": 21.894505494505495, "grad_norm": 0.78125, "learning_rate": 3.4344677977932064e-05, "loss": 0.116, "step": 29886 }, { "epoch": 21.895238095238096, "grad_norm": 0.765625, "learning_rate": 3.434143927581302e-05, "loss": 0.1124, "step": 29887 }, { "epoch": 21.895970695970696, "grad_norm": 1.1015625, "learning_rate": 3.433820065111495e-05, "loss": 0.1355, "step": 29888 }, { "epoch": 21.896703296703297, "grad_norm": 0.77734375, "learning_rate": 3.433496210385336e-05, "loss": 0.1174, "step": 29889 }, { "epoch": 21.897435897435898, "grad_norm": 1.0078125, "learning_rate": 3.433172363404375e-05, "loss": 0.0915, "step": 29890 }, { "epoch": 21.8981684981685, "grad_norm": 0.5859375, "learning_rate": 3.4328485241701686e-05, "loss": 0.0842, "step": 29891 }, { "epoch": 21.8989010989011, "grad_norm": 0.8984375, "learning_rate": 3.4325246926842655e-05, "loss": 0.1276, "step": 29892 }, { "epoch": 21.8996336996337, "grad_norm": 1.375, "learning_rate": 3.432200868948217e-05, "loss": 0.1728, "step": 29893 }, { "epoch": 21.9003663003663, "grad_norm": 1.640625, "learning_rate": 3.4318770529635755e-05, "loss": 0.0991, "step": 29894 }, { "epoch": 21.9010989010989, "grad_norm": 1.1640625, "learning_rate": 3.431553244731892e-05, "loss": 0.1302, "step": 29895 }, { "epoch": 21.901831501831502, "grad_norm": 1.671875, "learning_rate": 3.431229444254718e-05, "loss": 0.1266, "step": 29896 }, { "epoch": 21.902564102564103, "grad_norm": 0.91796875, "learning_rate": 3.430905651533606e-05, "loss": 0.1204, "step": 29897 }, { "epoch": 21.903296703296704, "grad_norm": 1.3203125, "learning_rate": 3.4305818665701064e-05, "loss": 0.1031, "step": 29898 }, { "epoch": 21.904029304029304, "grad_norm": 1.0078125, "learning_rate": 3.430258089365771e-05, "loss": 0.1384, "step": 29899 }, { "epoch": 21.904761904761905, "grad_norm": 1.015625, "learning_rate": 3.429934319922151e-05, "loss": 0.1164, "step": 29900 }, { "epoch": 21.905494505494506, "grad_norm": 1.140625, "learning_rate": 3.429610558240795e-05, "loss": 0.1507, "step": 29901 }, { "epoch": 21.906227106227107, "grad_norm": 1.0859375, "learning_rate": 3.429286804323259e-05, "loss": 0.1138, "step": 29902 }, { "epoch": 21.906959706959707, "grad_norm": 0.671875, "learning_rate": 3.4289630581710915e-05, "loss": 0.1153, "step": 29903 }, { "epoch": 21.907692307692308, "grad_norm": 1.0546875, "learning_rate": 3.428639319785845e-05, "loss": 0.1215, "step": 29904 }, { "epoch": 21.90842490842491, "grad_norm": 0.796875, "learning_rate": 3.428315589169068e-05, "loss": 0.118, "step": 29905 }, { "epoch": 21.90915750915751, "grad_norm": 0.80859375, "learning_rate": 3.427991866322314e-05, "loss": 0.0999, "step": 29906 }, { "epoch": 21.90989010989011, "grad_norm": 1.0390625, "learning_rate": 3.427668151247132e-05, "loss": 0.1299, "step": 29907 }, { "epoch": 21.91062271062271, "grad_norm": 1.359375, "learning_rate": 3.427344443945077e-05, "loss": 0.1019, "step": 29908 }, { "epoch": 21.91135531135531, "grad_norm": 1.1953125, "learning_rate": 3.427020744417697e-05, "loss": 0.1298, "step": 29909 }, { "epoch": 21.912087912087912, "grad_norm": 0.71875, "learning_rate": 3.4266970526665436e-05, "loss": 0.1026, "step": 29910 }, { "epoch": 21.912820512820513, "grad_norm": 0.5625, "learning_rate": 3.426373368693165e-05, "loss": 0.13, "step": 29911 }, { "epoch": 21.913553113553114, "grad_norm": 1.2109375, "learning_rate": 3.426049692499118e-05, "loss": 0.1307, "step": 29912 }, { "epoch": 21.914285714285715, "grad_norm": 1.015625, "learning_rate": 3.425726024085948e-05, "loss": 0.1162, "step": 29913 }, { "epoch": 21.915018315018315, "grad_norm": 1.5, "learning_rate": 3.4254023634552094e-05, "loss": 0.1291, "step": 29914 }, { "epoch": 21.915750915750916, "grad_norm": 1.5859375, "learning_rate": 3.4250787106084505e-05, "loss": 0.1068, "step": 29915 }, { "epoch": 21.916483516483517, "grad_norm": 0.66796875, "learning_rate": 3.4247550655472215e-05, "loss": 0.0966, "step": 29916 }, { "epoch": 21.917216117216118, "grad_norm": 0.8359375, "learning_rate": 3.424431428273076e-05, "loss": 0.1221, "step": 29917 }, { "epoch": 21.91794871794872, "grad_norm": 0.671875, "learning_rate": 3.4241077987875645e-05, "loss": 0.1146, "step": 29918 }, { "epoch": 21.91868131868132, "grad_norm": 1.15625, "learning_rate": 3.423784177092233e-05, "loss": 0.1468, "step": 29919 }, { "epoch": 21.91941391941392, "grad_norm": 0.96484375, "learning_rate": 3.423460563188638e-05, "loss": 0.1108, "step": 29920 }, { "epoch": 21.92014652014652, "grad_norm": 0.7890625, "learning_rate": 3.423136957078325e-05, "loss": 0.1016, "step": 29921 }, { "epoch": 21.92087912087912, "grad_norm": 0.95703125, "learning_rate": 3.4228133587628486e-05, "loss": 0.1121, "step": 29922 }, { "epoch": 21.921611721611722, "grad_norm": 0.9375, "learning_rate": 3.4224897682437576e-05, "loss": 0.1298, "step": 29923 }, { "epoch": 21.922344322344323, "grad_norm": 1.1953125, "learning_rate": 3.422166185522603e-05, "loss": 0.1512, "step": 29924 }, { "epoch": 21.923076923076923, "grad_norm": 0.95703125, "learning_rate": 3.421842610600935e-05, "loss": 0.1078, "step": 29925 }, { "epoch": 21.923809523809524, "grad_norm": 1.0859375, "learning_rate": 3.421519043480302e-05, "loss": 0.1232, "step": 29926 }, { "epoch": 21.924542124542125, "grad_norm": 0.8984375, "learning_rate": 3.421195484162255e-05, "loss": 0.1289, "step": 29927 }, { "epoch": 21.925274725274726, "grad_norm": 1.5546875, "learning_rate": 3.420871932648347e-05, "loss": 0.1415, "step": 29928 }, { "epoch": 21.926007326007326, "grad_norm": 0.8671875, "learning_rate": 3.4205483889401254e-05, "loss": 0.1358, "step": 29929 }, { "epoch": 21.926739926739927, "grad_norm": 0.82421875, "learning_rate": 3.4202248530391405e-05, "loss": 0.1329, "step": 29930 }, { "epoch": 21.927472527472528, "grad_norm": 0.87109375, "learning_rate": 3.4199013249469456e-05, "loss": 0.1029, "step": 29931 }, { "epoch": 21.92820512820513, "grad_norm": 0.70703125, "learning_rate": 3.419577804665087e-05, "loss": 0.0922, "step": 29932 }, { "epoch": 21.92893772893773, "grad_norm": 1.4140625, "learning_rate": 3.419254292195116e-05, "loss": 0.1152, "step": 29933 }, { "epoch": 21.92967032967033, "grad_norm": 1.2265625, "learning_rate": 3.4189307875385844e-05, "loss": 0.1462, "step": 29934 }, { "epoch": 21.93040293040293, "grad_norm": 1.0, "learning_rate": 3.4186072906970406e-05, "loss": 0.0895, "step": 29935 }, { "epoch": 21.93113553113553, "grad_norm": 1.0078125, "learning_rate": 3.418283801672035e-05, "loss": 0.1035, "step": 29936 }, { "epoch": 21.931868131868132, "grad_norm": 0.8125, "learning_rate": 3.417960320465116e-05, "loss": 0.1298, "step": 29937 }, { "epoch": 21.932600732600733, "grad_norm": 1.15625, "learning_rate": 3.4176368470778364e-05, "loss": 0.1086, "step": 29938 }, { "epoch": 21.933333333333334, "grad_norm": 0.86328125, "learning_rate": 3.4173133815117434e-05, "loss": 0.1049, "step": 29939 }, { "epoch": 21.934065934065934, "grad_norm": 1.4921875, "learning_rate": 3.416989923768389e-05, "loss": 0.1416, "step": 29940 }, { "epoch": 21.934798534798535, "grad_norm": 1.3828125, "learning_rate": 3.4166664738493225e-05, "loss": 0.136, "step": 29941 }, { "epoch": 21.935531135531136, "grad_norm": 0.9296875, "learning_rate": 3.41634303175609e-05, "loss": 0.1005, "step": 29942 }, { "epoch": 21.936263736263736, "grad_norm": 0.95703125, "learning_rate": 3.416019597490247e-05, "loss": 0.119, "step": 29943 }, { "epoch": 21.936996336996337, "grad_norm": 0.7421875, "learning_rate": 3.415696171053341e-05, "loss": 0.1135, "step": 29944 }, { "epoch": 21.937728937728938, "grad_norm": 1.0859375, "learning_rate": 3.415372752446919e-05, "loss": 0.1176, "step": 29945 }, { "epoch": 21.93846153846154, "grad_norm": 0.76171875, "learning_rate": 3.415049341672534e-05, "loss": 0.1134, "step": 29946 }, { "epoch": 21.93919413919414, "grad_norm": 1.4296875, "learning_rate": 3.414725938731733e-05, "loss": 0.1308, "step": 29947 }, { "epoch": 21.93992673992674, "grad_norm": 1.015625, "learning_rate": 3.414402543626068e-05, "loss": 0.1173, "step": 29948 }, { "epoch": 21.94065934065934, "grad_norm": 1.2109375, "learning_rate": 3.4140791563570883e-05, "loss": 0.1177, "step": 29949 }, { "epoch": 21.94139194139194, "grad_norm": 0.875, "learning_rate": 3.413755776926341e-05, "loss": 0.1214, "step": 29950 }, { "epoch": 21.942124542124542, "grad_norm": 1.2734375, "learning_rate": 3.413432405335378e-05, "loss": 0.0957, "step": 29951 }, { "epoch": 21.942857142857143, "grad_norm": 1.015625, "learning_rate": 3.4131090415857475e-05, "loss": 0.1235, "step": 29952 }, { "epoch": 21.943589743589744, "grad_norm": 0.92578125, "learning_rate": 3.412785685678997e-05, "loss": 0.1447, "step": 29953 }, { "epoch": 21.944322344322345, "grad_norm": 0.90625, "learning_rate": 3.4124623376166795e-05, "loss": 0.1144, "step": 29954 }, { "epoch": 21.945054945054945, "grad_norm": 0.9765625, "learning_rate": 3.412138997400343e-05, "loss": 0.1589, "step": 29955 }, { "epoch": 21.945787545787546, "grad_norm": 0.9296875, "learning_rate": 3.411815665031534e-05, "loss": 0.1306, "step": 29956 }, { "epoch": 21.946520146520147, "grad_norm": 1.5234375, "learning_rate": 3.411492340511806e-05, "loss": 0.1267, "step": 29957 }, { "epoch": 21.947252747252747, "grad_norm": 0.88671875, "learning_rate": 3.411169023842705e-05, "loss": 0.1298, "step": 29958 }, { "epoch": 21.947985347985348, "grad_norm": 1.0078125, "learning_rate": 3.410845715025781e-05, "loss": 0.105, "step": 29959 }, { "epoch": 21.94871794871795, "grad_norm": 1.1171875, "learning_rate": 3.410522414062584e-05, "loss": 0.1344, "step": 29960 }, { "epoch": 21.94945054945055, "grad_norm": 1.2578125, "learning_rate": 3.410199120954663e-05, "loss": 0.0991, "step": 29961 }, { "epoch": 21.95018315018315, "grad_norm": 1.109375, "learning_rate": 3.409875835703566e-05, "loss": 0.1106, "step": 29962 }, { "epoch": 21.95091575091575, "grad_norm": 1.515625, "learning_rate": 3.40955255831084e-05, "loss": 0.1479, "step": 29963 }, { "epoch": 21.951648351648352, "grad_norm": 1.296875, "learning_rate": 3.409229288778039e-05, "loss": 0.118, "step": 29964 }, { "epoch": 21.952380952380953, "grad_norm": 1.3984375, "learning_rate": 3.408906027106708e-05, "loss": 0.1318, "step": 29965 }, { "epoch": 21.953113553113553, "grad_norm": 0.6640625, "learning_rate": 3.408582773298397e-05, "loss": 0.0936, "step": 29966 }, { "epoch": 21.953846153846154, "grad_norm": 1.4296875, "learning_rate": 3.408259527354656e-05, "loss": 0.1215, "step": 29967 }, { "epoch": 21.954578754578755, "grad_norm": 1.0078125, "learning_rate": 3.4079362892770294e-05, "loss": 0.1318, "step": 29968 }, { "epoch": 21.955311355311355, "grad_norm": 1.5703125, "learning_rate": 3.407613059067072e-05, "loss": 0.11, "step": 29969 }, { "epoch": 21.956043956043956, "grad_norm": 1.25, "learning_rate": 3.407289836726328e-05, "loss": 0.1293, "step": 29970 }, { "epoch": 21.956776556776557, "grad_norm": 0.80078125, "learning_rate": 3.406966622256349e-05, "loss": 0.1106, "step": 29971 }, { "epoch": 21.957509157509158, "grad_norm": 1.109375, "learning_rate": 3.4066434156586816e-05, "loss": 0.1311, "step": 29972 }, { "epoch": 21.95824175824176, "grad_norm": 0.77734375, "learning_rate": 3.4063202169348735e-05, "loss": 0.1208, "step": 29973 }, { "epoch": 21.95897435897436, "grad_norm": 1.0703125, "learning_rate": 3.405997026086477e-05, "loss": 0.1206, "step": 29974 }, { "epoch": 21.95970695970696, "grad_norm": 1.2421875, "learning_rate": 3.405673843115039e-05, "loss": 0.1145, "step": 29975 }, { "epoch": 21.96043956043956, "grad_norm": 1.2109375, "learning_rate": 3.405350668022106e-05, "loss": 0.1222, "step": 29976 }, { "epoch": 21.96117216117216, "grad_norm": 1.265625, "learning_rate": 3.405027500809229e-05, "loss": 0.141, "step": 29977 }, { "epoch": 21.961904761904762, "grad_norm": 0.84765625, "learning_rate": 3.4047043414779544e-05, "loss": 0.1216, "step": 29978 }, { "epoch": 21.962637362637363, "grad_norm": 0.86328125, "learning_rate": 3.40438119002983e-05, "loss": 0.1096, "step": 29979 }, { "epoch": 21.963369963369964, "grad_norm": 0.88671875, "learning_rate": 3.404058046466407e-05, "loss": 0.1159, "step": 29980 }, { "epoch": 21.964102564102564, "grad_norm": 1.4140625, "learning_rate": 3.403734910789233e-05, "loss": 0.1324, "step": 29981 }, { "epoch": 21.964835164835165, "grad_norm": 1.4765625, "learning_rate": 3.403411782999854e-05, "loss": 0.1337, "step": 29982 }, { "epoch": 21.965567765567766, "grad_norm": 1.53125, "learning_rate": 3.403088663099821e-05, "loss": 0.1102, "step": 29983 }, { "epoch": 21.966300366300366, "grad_norm": 1.28125, "learning_rate": 3.40276555109068e-05, "loss": 0.1308, "step": 29984 }, { "epoch": 21.967032967032967, "grad_norm": 0.71875, "learning_rate": 3.402442446973979e-05, "loss": 0.1329, "step": 29985 }, { "epoch": 21.967765567765568, "grad_norm": 0.828125, "learning_rate": 3.402119350751269e-05, "loss": 0.1251, "step": 29986 }, { "epoch": 21.96849816849817, "grad_norm": 0.90625, "learning_rate": 3.4017962624240956e-05, "loss": 0.1005, "step": 29987 }, { "epoch": 21.96923076923077, "grad_norm": 1.1171875, "learning_rate": 3.401473181994008e-05, "loss": 0.1084, "step": 29988 }, { "epoch": 21.96996336996337, "grad_norm": 1.640625, "learning_rate": 3.4011501094625514e-05, "loss": 0.1585, "step": 29989 }, { "epoch": 21.97069597069597, "grad_norm": 1.1640625, "learning_rate": 3.4008270448312776e-05, "loss": 0.1232, "step": 29990 }, { "epoch": 21.97142857142857, "grad_norm": 0.953125, "learning_rate": 3.400503988101732e-05, "loss": 0.1153, "step": 29991 }, { "epoch": 21.972161172161172, "grad_norm": 0.84765625, "learning_rate": 3.4001809392754644e-05, "loss": 0.109, "step": 29992 }, { "epoch": 21.972893772893773, "grad_norm": 1.5625, "learning_rate": 3.399857898354022e-05, "loss": 0.1625, "step": 29993 }, { "epoch": 21.973626373626374, "grad_norm": 1.578125, "learning_rate": 3.399534865338949e-05, "loss": 0.0936, "step": 29994 }, { "epoch": 21.974358974358974, "grad_norm": 0.9453125, "learning_rate": 3.399211840231798e-05, "loss": 0.1202, "step": 29995 }, { "epoch": 21.975091575091575, "grad_norm": 0.7265625, "learning_rate": 3.398888823034115e-05, "loss": 0.1282, "step": 29996 }, { "epoch": 21.975824175824176, "grad_norm": 1.265625, "learning_rate": 3.39856581374745e-05, "loss": 0.1527, "step": 29997 }, { "epoch": 21.976556776556777, "grad_norm": 2.171875, "learning_rate": 3.398242812373345e-05, "loss": 0.1293, "step": 29998 }, { "epoch": 21.977289377289377, "grad_norm": 1.1328125, "learning_rate": 3.3979198189133515e-05, "loss": 0.1262, "step": 29999 }, { "epoch": 21.978021978021978, "grad_norm": 0.70703125, "learning_rate": 3.3975968333690175e-05, "loss": 0.1231, "step": 30000 }, { "epoch": 21.97875457875458, "grad_norm": 1.140625, "learning_rate": 3.3972738557418896e-05, "loss": 0.1335, "step": 30001 }, { "epoch": 21.97948717948718, "grad_norm": 1.28125, "learning_rate": 3.3969508860335155e-05, "loss": 0.1486, "step": 30002 }, { "epoch": 21.98021978021978, "grad_norm": 1.421875, "learning_rate": 3.396627924245442e-05, "loss": 0.1254, "step": 30003 }, { "epoch": 21.98095238095238, "grad_norm": 1.3671875, "learning_rate": 3.396304970379217e-05, "loss": 0.1501, "step": 30004 }, { "epoch": 21.98168498168498, "grad_norm": 1.75, "learning_rate": 3.395982024436386e-05, "loss": 0.1407, "step": 30005 }, { "epoch": 21.982417582417582, "grad_norm": 1.3046875, "learning_rate": 3.3956590864185e-05, "loss": 0.1343, "step": 30006 }, { "epoch": 21.983150183150183, "grad_norm": 1.4453125, "learning_rate": 3.395336156327104e-05, "loss": 0.1263, "step": 30007 }, { "epoch": 21.983882783882784, "grad_norm": 0.8203125, "learning_rate": 3.395013234163745e-05, "loss": 0.118, "step": 30008 }, { "epoch": 21.984615384615385, "grad_norm": 0.7109375, "learning_rate": 3.394690319929972e-05, "loss": 0.0961, "step": 30009 }, { "epoch": 21.985347985347985, "grad_norm": 1.2734375, "learning_rate": 3.394367413627329e-05, "loss": 0.1276, "step": 30010 }, { "epoch": 21.986080586080586, "grad_norm": 1.5078125, "learning_rate": 3.394044515257367e-05, "loss": 0.1292, "step": 30011 }, { "epoch": 21.986813186813187, "grad_norm": 1.2265625, "learning_rate": 3.393721624821631e-05, "loss": 0.1007, "step": 30012 }, { "epoch": 21.987545787545788, "grad_norm": 1.1796875, "learning_rate": 3.393398742321669e-05, "loss": 0.1395, "step": 30013 }, { "epoch": 21.98827838827839, "grad_norm": 0.9453125, "learning_rate": 3.3930758677590276e-05, "loss": 0.124, "step": 30014 }, { "epoch": 21.98901098901099, "grad_norm": 0.71484375, "learning_rate": 3.392753001135252e-05, "loss": 0.1032, "step": 30015 }, { "epoch": 21.98974358974359, "grad_norm": 0.7421875, "learning_rate": 3.392430142451891e-05, "loss": 0.1076, "step": 30016 }, { "epoch": 21.99047619047619, "grad_norm": 1.2734375, "learning_rate": 3.3921072917104926e-05, "loss": 0.1183, "step": 30017 }, { "epoch": 21.99120879120879, "grad_norm": 1.125, "learning_rate": 3.391784448912603e-05, "loss": 0.1351, "step": 30018 }, { "epoch": 21.991941391941392, "grad_norm": 1.0625, "learning_rate": 3.391461614059768e-05, "loss": 0.1344, "step": 30019 }, { "epoch": 21.992673992673993, "grad_norm": 1.046875, "learning_rate": 3.3911387871535327e-05, "loss": 0.1119, "step": 30020 }, { "epoch": 21.993406593406593, "grad_norm": 1.15625, "learning_rate": 3.3908159681954475e-05, "loss": 0.1167, "step": 30021 }, { "epoch": 21.994139194139194, "grad_norm": 0.98828125, "learning_rate": 3.390493157187057e-05, "loss": 0.1361, "step": 30022 }, { "epoch": 21.994871794871795, "grad_norm": 1.734375, "learning_rate": 3.3901703541299104e-05, "loss": 0.1522, "step": 30023 }, { "epoch": 21.995604395604396, "grad_norm": 1.421875, "learning_rate": 3.389847559025552e-05, "loss": 0.1217, "step": 30024 }, { "epoch": 21.996336996336996, "grad_norm": 1.1015625, "learning_rate": 3.3895247718755265e-05, "loss": 0.1286, "step": 30025 }, { "epoch": 21.997069597069597, "grad_norm": 0.8359375, "learning_rate": 3.389201992681385e-05, "loss": 0.1505, "step": 30026 }, { "epoch": 21.997802197802198, "grad_norm": 0.97265625, "learning_rate": 3.388879221444672e-05, "loss": 0.1186, "step": 30027 }, { "epoch": 21.9985347985348, "grad_norm": 1.015625, "learning_rate": 3.3885564581669344e-05, "loss": 0.1181, "step": 30028 }, { "epoch": 21.9992673992674, "grad_norm": 0.80859375, "learning_rate": 3.388233702849718e-05, "loss": 0.1155, "step": 30029 }, { "epoch": 22.0, "grad_norm": 1.2109375, "learning_rate": 3.3879109554945694e-05, "loss": 0.1255, "step": 30030 }, { "epoch": 22.0007326007326, "grad_norm": 1.0390625, "learning_rate": 3.387588216103033e-05, "loss": 0.1038, "step": 30031 }, { "epoch": 22.0014652014652, "grad_norm": 0.7890625, "learning_rate": 3.387265484676659e-05, "loss": 0.1148, "step": 30032 }, { "epoch": 22.002197802197802, "grad_norm": 0.87109375, "learning_rate": 3.386942761216993e-05, "loss": 0.1141, "step": 30033 }, { "epoch": 22.002930402930403, "grad_norm": 0.80078125, "learning_rate": 3.386620045725578e-05, "loss": 0.1091, "step": 30034 }, { "epoch": 22.003663003663004, "grad_norm": 1.28125, "learning_rate": 3.386297338203964e-05, "loss": 0.1206, "step": 30035 }, { "epoch": 22.004395604395604, "grad_norm": 1.3125, "learning_rate": 3.385974638653693e-05, "loss": 0.1315, "step": 30036 }, { "epoch": 22.005128205128205, "grad_norm": 1.2890625, "learning_rate": 3.385651947076316e-05, "loss": 0.1548, "step": 30037 }, { "epoch": 22.005860805860806, "grad_norm": 1.109375, "learning_rate": 3.385329263473377e-05, "loss": 0.1359, "step": 30038 }, { "epoch": 22.006593406593407, "grad_norm": 1.1328125, "learning_rate": 3.385006587846422e-05, "loss": 0.0955, "step": 30039 }, { "epoch": 22.007326007326007, "grad_norm": 1.078125, "learning_rate": 3.3846839201969974e-05, "loss": 0.1236, "step": 30040 }, { "epoch": 22.008058608058608, "grad_norm": 1.1484375, "learning_rate": 3.384361260526646e-05, "loss": 0.107, "step": 30041 }, { "epoch": 22.00879120879121, "grad_norm": 0.9375, "learning_rate": 3.384038608836919e-05, "loss": 0.1251, "step": 30042 }, { "epoch": 22.00952380952381, "grad_norm": 0.73828125, "learning_rate": 3.3837159651293604e-05, "loss": 0.0996, "step": 30043 }, { "epoch": 22.01025641025641, "grad_norm": 0.9453125, "learning_rate": 3.383393329405515e-05, "loss": 0.1029, "step": 30044 }, { "epoch": 22.01098901098901, "grad_norm": 1.1328125, "learning_rate": 3.3830707016669283e-05, "loss": 0.1313, "step": 30045 }, { "epoch": 22.01172161172161, "grad_norm": 1.0, "learning_rate": 3.382748081915147e-05, "loss": 0.1257, "step": 30046 }, { "epoch": 22.012454212454212, "grad_norm": 1.53125, "learning_rate": 3.382425470151718e-05, "loss": 0.1052, "step": 30047 }, { "epoch": 22.013186813186813, "grad_norm": 2.109375, "learning_rate": 3.382102866378186e-05, "loss": 0.111, "step": 30048 }, { "epoch": 22.013919413919414, "grad_norm": 1.390625, "learning_rate": 3.3817802705960964e-05, "loss": 0.1525, "step": 30049 }, { "epoch": 22.014652014652015, "grad_norm": 1.2109375, "learning_rate": 3.381457682806996e-05, "loss": 0.1114, "step": 30050 }, { "epoch": 22.015384615384615, "grad_norm": 0.9765625, "learning_rate": 3.3811351030124265e-05, "loss": 0.1199, "step": 30051 }, { "epoch": 22.016117216117216, "grad_norm": 1.03125, "learning_rate": 3.38081253121394e-05, "loss": 0.0993, "step": 30052 }, { "epoch": 22.016849816849817, "grad_norm": 0.921875, "learning_rate": 3.380489967413078e-05, "loss": 0.1207, "step": 30053 }, { "epoch": 22.017582417582418, "grad_norm": 0.84765625, "learning_rate": 3.3801674116113866e-05, "loss": 0.1249, "step": 30054 }, { "epoch": 22.01831501831502, "grad_norm": 1.0390625, "learning_rate": 3.379844863810411e-05, "loss": 0.108, "step": 30055 }, { "epoch": 22.01904761904762, "grad_norm": 1.40625, "learning_rate": 3.3795223240116975e-05, "loss": 0.1147, "step": 30056 }, { "epoch": 22.01978021978022, "grad_norm": 1.171875, "learning_rate": 3.379199792216788e-05, "loss": 0.1134, "step": 30057 }, { "epoch": 22.02051282051282, "grad_norm": 0.96875, "learning_rate": 3.378877268427234e-05, "loss": 0.1121, "step": 30058 }, { "epoch": 22.02124542124542, "grad_norm": 1.03125, "learning_rate": 3.378554752644577e-05, "loss": 0.1168, "step": 30059 }, { "epoch": 22.021978021978022, "grad_norm": 0.69921875, "learning_rate": 3.3782322448703625e-05, "loss": 0.1008, "step": 30060 }, { "epoch": 22.022710622710623, "grad_norm": 1.171875, "learning_rate": 3.3779097451061357e-05, "loss": 0.1174, "step": 30061 }, { "epoch": 22.023443223443223, "grad_norm": 1.3671875, "learning_rate": 3.377587253353441e-05, "loss": 0.1232, "step": 30062 }, { "epoch": 22.024175824175824, "grad_norm": 1.625, "learning_rate": 3.377264769613827e-05, "loss": 0.121, "step": 30063 }, { "epoch": 22.024908424908425, "grad_norm": 1.3203125, "learning_rate": 3.376942293888836e-05, "loss": 0.1498, "step": 30064 }, { "epoch": 22.025641025641026, "grad_norm": 0.9921875, "learning_rate": 3.376619826180015e-05, "loss": 0.1061, "step": 30065 }, { "epoch": 22.026373626373626, "grad_norm": 1.046875, "learning_rate": 3.376297366488907e-05, "loss": 0.1002, "step": 30066 }, { "epoch": 22.027106227106227, "grad_norm": 2.234375, "learning_rate": 3.375974914817055e-05, "loss": 0.1403, "step": 30067 }, { "epoch": 22.027838827838828, "grad_norm": 1.03125, "learning_rate": 3.3756524711660094e-05, "loss": 0.1146, "step": 30068 }, { "epoch": 22.02857142857143, "grad_norm": 0.8515625, "learning_rate": 3.375330035537312e-05, "loss": 0.1236, "step": 30069 }, { "epoch": 22.02930402930403, "grad_norm": 1.046875, "learning_rate": 3.375007607932509e-05, "loss": 0.137, "step": 30070 }, { "epoch": 22.03003663003663, "grad_norm": 1.7734375, "learning_rate": 3.374685188353143e-05, "loss": 0.0961, "step": 30071 }, { "epoch": 22.03076923076923, "grad_norm": 1.03125, "learning_rate": 3.3743627768007606e-05, "loss": 0.1106, "step": 30072 }, { "epoch": 22.03150183150183, "grad_norm": 1.203125, "learning_rate": 3.3740403732769065e-05, "loss": 0.106, "step": 30073 }, { "epoch": 22.032234432234432, "grad_norm": 0.73828125, "learning_rate": 3.3737179777831245e-05, "loss": 0.119, "step": 30074 }, { "epoch": 22.032967032967033, "grad_norm": 1.28125, "learning_rate": 3.3733955903209605e-05, "loss": 0.107, "step": 30075 }, { "epoch": 22.033699633699634, "grad_norm": 1.046875, "learning_rate": 3.37307321089196e-05, "loss": 0.1048, "step": 30076 }, { "epoch": 22.034432234432234, "grad_norm": 0.6796875, "learning_rate": 3.372750839497665e-05, "loss": 0.119, "step": 30077 }, { "epoch": 22.035164835164835, "grad_norm": 1.2109375, "learning_rate": 3.37242847613962e-05, "loss": 0.0972, "step": 30078 }, { "epoch": 22.035897435897436, "grad_norm": 1.1484375, "learning_rate": 3.3721061208193724e-05, "loss": 0.117, "step": 30079 }, { "epoch": 22.036630036630036, "grad_norm": 0.73828125, "learning_rate": 3.371783773538465e-05, "loss": 0.1118, "step": 30080 }, { "epoch": 22.037362637362637, "grad_norm": 0.94921875, "learning_rate": 3.371461434298443e-05, "loss": 0.1159, "step": 30081 }, { "epoch": 22.038095238095238, "grad_norm": 0.8359375, "learning_rate": 3.37113910310085e-05, "loss": 0.1151, "step": 30082 }, { "epoch": 22.03882783882784, "grad_norm": 1.734375, "learning_rate": 3.370816779947229e-05, "loss": 0.1335, "step": 30083 }, { "epoch": 22.03956043956044, "grad_norm": 0.84765625, "learning_rate": 3.370494464839128e-05, "loss": 0.1108, "step": 30084 }, { "epoch": 22.04029304029304, "grad_norm": 0.84375, "learning_rate": 3.3701721577780886e-05, "loss": 0.1264, "step": 30085 }, { "epoch": 22.04102564102564, "grad_norm": 1.109375, "learning_rate": 3.3698498587656555e-05, "loss": 0.1234, "step": 30086 }, { "epoch": 22.04175824175824, "grad_norm": 1.109375, "learning_rate": 3.369527567803375e-05, "loss": 0.1161, "step": 30087 }, { "epoch": 22.042490842490842, "grad_norm": 0.7265625, "learning_rate": 3.3692052848927866e-05, "loss": 0.1191, "step": 30088 }, { "epoch": 22.043223443223443, "grad_norm": 1.453125, "learning_rate": 3.3688830100354394e-05, "loss": 0.1177, "step": 30089 }, { "epoch": 22.043956043956044, "grad_norm": 1.4140625, "learning_rate": 3.368560743232876e-05, "loss": 0.1412, "step": 30090 }, { "epoch": 22.044688644688645, "grad_norm": 1.0234375, "learning_rate": 3.36823848448664e-05, "loss": 0.1074, "step": 30091 }, { "epoch": 22.045421245421245, "grad_norm": 1.0859375, "learning_rate": 3.367916233798275e-05, "loss": 0.1109, "step": 30092 }, { "epoch": 22.046153846153846, "grad_norm": 0.9453125, "learning_rate": 3.367593991169324e-05, "loss": 0.1019, "step": 30093 }, { "epoch": 22.046886446886447, "grad_norm": 1.6015625, "learning_rate": 3.367271756601335e-05, "loss": 0.1583, "step": 30094 }, { "epoch": 22.047619047619047, "grad_norm": 1.53125, "learning_rate": 3.3669495300958483e-05, "loss": 0.1318, "step": 30095 }, { "epoch": 22.048351648351648, "grad_norm": 0.75390625, "learning_rate": 3.3666273116544096e-05, "loss": 0.1053, "step": 30096 }, { "epoch": 22.04908424908425, "grad_norm": 1.0703125, "learning_rate": 3.366305101278561e-05, "loss": 0.1125, "step": 30097 }, { "epoch": 22.04981684981685, "grad_norm": 0.8671875, "learning_rate": 3.365982898969847e-05, "loss": 0.112, "step": 30098 }, { "epoch": 22.05054945054945, "grad_norm": 1.0859375, "learning_rate": 3.365660704729813e-05, "loss": 0.1181, "step": 30099 }, { "epoch": 22.05128205128205, "grad_norm": 1.1796875, "learning_rate": 3.36533851856e-05, "loss": 0.1216, "step": 30100 }, { "epoch": 22.052014652014652, "grad_norm": 1.1015625, "learning_rate": 3.365016340461955e-05, "loss": 0.1359, "step": 30101 }, { "epoch": 22.052747252747253, "grad_norm": 1.578125, "learning_rate": 3.3646941704372206e-05, "loss": 0.1567, "step": 30102 }, { "epoch": 22.053479853479853, "grad_norm": 1.140625, "learning_rate": 3.364372008487338e-05, "loss": 0.1263, "step": 30103 }, { "epoch": 22.054212454212454, "grad_norm": 0.70703125, "learning_rate": 3.364049854613851e-05, "loss": 0.1218, "step": 30104 }, { "epoch": 22.054945054945055, "grad_norm": 1.1875, "learning_rate": 3.363727708818307e-05, "loss": 0.1247, "step": 30105 }, { "epoch": 22.055677655677655, "grad_norm": 0.75, "learning_rate": 3.363405571102246e-05, "loss": 0.1075, "step": 30106 }, { "epoch": 22.056410256410256, "grad_norm": 0.72265625, "learning_rate": 3.363083441467214e-05, "loss": 0.1001, "step": 30107 }, { "epoch": 22.057142857142857, "grad_norm": 1.4375, "learning_rate": 3.362761319914752e-05, "loss": 0.1292, "step": 30108 }, { "epoch": 22.057875457875458, "grad_norm": 1.5859375, "learning_rate": 3.3624392064464026e-05, "loss": 0.1096, "step": 30109 }, { "epoch": 22.05860805860806, "grad_norm": 1.1953125, "learning_rate": 3.362117101063714e-05, "loss": 0.12, "step": 30110 }, { "epoch": 22.05934065934066, "grad_norm": 1.1953125, "learning_rate": 3.3617950037682245e-05, "loss": 0.1224, "step": 30111 }, { "epoch": 22.06007326007326, "grad_norm": 0.9765625, "learning_rate": 3.361472914561481e-05, "loss": 0.1423, "step": 30112 }, { "epoch": 22.06080586080586, "grad_norm": 0.671875, "learning_rate": 3.3611508334450245e-05, "loss": 0.099, "step": 30113 }, { "epoch": 22.06153846153846, "grad_norm": 1.1484375, "learning_rate": 3.360828760420397e-05, "loss": 0.145, "step": 30114 }, { "epoch": 22.062271062271062, "grad_norm": 1.1953125, "learning_rate": 3.360506695489145e-05, "loss": 0.1482, "step": 30115 }, { "epoch": 22.063003663003663, "grad_norm": 1.3984375, "learning_rate": 3.3601846386528096e-05, "loss": 0.1278, "step": 30116 }, { "epoch": 22.063736263736264, "grad_norm": 0.8828125, "learning_rate": 3.359862589912935e-05, "loss": 0.0978, "step": 30117 }, { "epoch": 22.064468864468864, "grad_norm": 1.1953125, "learning_rate": 3.359540549271064e-05, "loss": 0.1248, "step": 30118 }, { "epoch": 22.065201465201465, "grad_norm": 0.515625, "learning_rate": 3.359218516728737e-05, "loss": 0.1147, "step": 30119 }, { "epoch": 22.065934065934066, "grad_norm": 1.1328125, "learning_rate": 3.358896492287501e-05, "loss": 0.0986, "step": 30120 }, { "epoch": 22.066666666666666, "grad_norm": 0.8203125, "learning_rate": 3.358574475948897e-05, "loss": 0.1339, "step": 30121 }, { "epoch": 22.067399267399267, "grad_norm": 1.0234375, "learning_rate": 3.358252467714468e-05, "loss": 0.1322, "step": 30122 }, { "epoch": 22.068131868131868, "grad_norm": 0.8046875, "learning_rate": 3.357930467585755e-05, "loss": 0.1193, "step": 30123 }, { "epoch": 22.06886446886447, "grad_norm": 0.64453125, "learning_rate": 3.3576084755643034e-05, "loss": 0.1298, "step": 30124 }, { "epoch": 22.06959706959707, "grad_norm": 0.9140625, "learning_rate": 3.3572864916516564e-05, "loss": 0.1386, "step": 30125 }, { "epoch": 22.07032967032967, "grad_norm": 0.84765625, "learning_rate": 3.356964515849354e-05, "loss": 0.1344, "step": 30126 }, { "epoch": 22.07106227106227, "grad_norm": 1.2265625, "learning_rate": 3.356642548158942e-05, "loss": 0.1169, "step": 30127 }, { "epoch": 22.07179487179487, "grad_norm": 1.203125, "learning_rate": 3.356320588581961e-05, "loss": 0.1313, "step": 30128 }, { "epoch": 22.072527472527472, "grad_norm": 1.5859375, "learning_rate": 3.355998637119954e-05, "loss": 0.1267, "step": 30129 }, { "epoch": 22.073260073260073, "grad_norm": 0.671875, "learning_rate": 3.355676693774462e-05, "loss": 0.1682, "step": 30130 }, { "epoch": 22.073992673992674, "grad_norm": 0.73828125, "learning_rate": 3.35535475854703e-05, "loss": 0.1094, "step": 30131 }, { "epoch": 22.074725274725274, "grad_norm": 0.89453125, "learning_rate": 3.355032831439201e-05, "loss": 0.1181, "step": 30132 }, { "epoch": 22.075457875457875, "grad_norm": 1.1640625, "learning_rate": 3.3547109124525154e-05, "loss": 0.1432, "step": 30133 }, { "epoch": 22.076190476190476, "grad_norm": 1.1484375, "learning_rate": 3.354389001588517e-05, "loss": 0.1349, "step": 30134 }, { "epoch": 22.076923076923077, "grad_norm": 0.9453125, "learning_rate": 3.354067098848745e-05, "loss": 0.1236, "step": 30135 }, { "epoch": 22.077655677655677, "grad_norm": 1.5859375, "learning_rate": 3.3537452042347464e-05, "loss": 0.1736, "step": 30136 }, { "epoch": 22.078388278388278, "grad_norm": 0.7109375, "learning_rate": 3.35342331774806e-05, "loss": 0.0981, "step": 30137 }, { "epoch": 22.07912087912088, "grad_norm": 0.93359375, "learning_rate": 3.353101439390231e-05, "loss": 0.1286, "step": 30138 }, { "epoch": 22.07985347985348, "grad_norm": 0.94921875, "learning_rate": 3.3527795691628e-05, "loss": 0.1018, "step": 30139 }, { "epoch": 22.08058608058608, "grad_norm": 1.421875, "learning_rate": 3.352457707067307e-05, "loss": 0.1233, "step": 30140 }, { "epoch": 22.08131868131868, "grad_norm": 1.0859375, "learning_rate": 3.352135853105299e-05, "loss": 0.1345, "step": 30141 }, { "epoch": 22.08205128205128, "grad_norm": 1.0, "learning_rate": 3.351814007278315e-05, "loss": 0.1094, "step": 30142 }, { "epoch": 22.082783882783882, "grad_norm": 1.0078125, "learning_rate": 3.351492169587897e-05, "loss": 0.1206, "step": 30143 }, { "epoch": 22.083516483516483, "grad_norm": 1.5234375, "learning_rate": 3.3511703400355884e-05, "loss": 0.1702, "step": 30144 }, { "epoch": 22.084249084249084, "grad_norm": 1.0703125, "learning_rate": 3.3508485186229286e-05, "loss": 0.1148, "step": 30145 }, { "epoch": 22.084981684981685, "grad_norm": 1.765625, "learning_rate": 3.350526705351462e-05, "loss": 0.113, "step": 30146 }, { "epoch": 22.085714285714285, "grad_norm": 0.89453125, "learning_rate": 3.350204900222731e-05, "loss": 0.1153, "step": 30147 }, { "epoch": 22.086446886446886, "grad_norm": 0.94140625, "learning_rate": 3.3498831032382764e-05, "loss": 0.1451, "step": 30148 }, { "epoch": 22.087179487179487, "grad_norm": 1.078125, "learning_rate": 3.349561314399639e-05, "loss": 0.1172, "step": 30149 }, { "epoch": 22.087912087912088, "grad_norm": 0.9375, "learning_rate": 3.34923953370836e-05, "loss": 0.1292, "step": 30150 }, { "epoch": 22.08864468864469, "grad_norm": 1.2421875, "learning_rate": 3.3489177611659854e-05, "loss": 0.1125, "step": 30151 }, { "epoch": 22.08937728937729, "grad_norm": 1.2578125, "learning_rate": 3.348595996774053e-05, "loss": 0.1286, "step": 30152 }, { "epoch": 22.09010989010989, "grad_norm": 1.546875, "learning_rate": 3.348274240534107e-05, "loss": 0.1273, "step": 30153 }, { "epoch": 22.09084249084249, "grad_norm": 0.63671875, "learning_rate": 3.347952492447687e-05, "loss": 0.097, "step": 30154 }, { "epoch": 22.09157509157509, "grad_norm": 0.88671875, "learning_rate": 3.347630752516335e-05, "loss": 0.1276, "step": 30155 }, { "epoch": 22.092307692307692, "grad_norm": 2.0, "learning_rate": 3.347309020741592e-05, "loss": 0.1084, "step": 30156 }, { "epoch": 22.093040293040293, "grad_norm": 1.59375, "learning_rate": 3.3469872971250024e-05, "loss": 0.1797, "step": 30157 }, { "epoch": 22.093772893772893, "grad_norm": 1.09375, "learning_rate": 3.3466655816681046e-05, "loss": 0.1395, "step": 30158 }, { "epoch": 22.094505494505494, "grad_norm": 1.1875, "learning_rate": 3.346343874372442e-05, "loss": 0.0949, "step": 30159 }, { "epoch": 22.095238095238095, "grad_norm": 1.4921875, "learning_rate": 3.346022175239554e-05, "loss": 0.1127, "step": 30160 }, { "epoch": 22.095970695970696, "grad_norm": 1.4296875, "learning_rate": 3.345700484270983e-05, "loss": 0.1239, "step": 30161 }, { "epoch": 22.096703296703296, "grad_norm": 0.74609375, "learning_rate": 3.3453788014682715e-05, "loss": 0.1239, "step": 30162 }, { "epoch": 22.097435897435897, "grad_norm": 1.5, "learning_rate": 3.345057126832959e-05, "loss": 0.1358, "step": 30163 }, { "epoch": 22.098168498168498, "grad_norm": 1.0546875, "learning_rate": 3.344735460366588e-05, "loss": 0.1574, "step": 30164 }, { "epoch": 22.0989010989011, "grad_norm": 1.0390625, "learning_rate": 3.3444138020707e-05, "loss": 0.1153, "step": 30165 }, { "epoch": 22.0996336996337, "grad_norm": 0.6640625, "learning_rate": 3.344092151946833e-05, "loss": 0.1228, "step": 30166 }, { "epoch": 22.1003663003663, "grad_norm": 1.0859375, "learning_rate": 3.3437705099965325e-05, "loss": 0.1224, "step": 30167 }, { "epoch": 22.1010989010989, "grad_norm": 1.4453125, "learning_rate": 3.343448876221337e-05, "loss": 0.1091, "step": 30168 }, { "epoch": 22.1018315018315, "grad_norm": 1.078125, "learning_rate": 3.343127250622789e-05, "loss": 0.11, "step": 30169 }, { "epoch": 22.102564102564102, "grad_norm": 1.203125, "learning_rate": 3.342805633202429e-05, "loss": 0.1003, "step": 30170 }, { "epoch": 22.103296703296703, "grad_norm": 0.87890625, "learning_rate": 3.3424840239617945e-05, "loss": 0.117, "step": 30171 }, { "epoch": 22.104029304029304, "grad_norm": 1.1796875, "learning_rate": 3.342162422902432e-05, "loss": 0.1427, "step": 30172 }, { "epoch": 22.104761904761904, "grad_norm": 1.2265625, "learning_rate": 3.34184083002588e-05, "loss": 0.1367, "step": 30173 }, { "epoch": 22.105494505494505, "grad_norm": 1.0546875, "learning_rate": 3.341519245333679e-05, "loss": 0.1166, "step": 30174 }, { "epoch": 22.106227106227106, "grad_norm": 0.8828125, "learning_rate": 3.3411976688273696e-05, "loss": 0.1135, "step": 30175 }, { "epoch": 22.106959706959707, "grad_norm": 1.15625, "learning_rate": 3.340876100508494e-05, "loss": 0.1243, "step": 30176 }, { "epoch": 22.107692307692307, "grad_norm": 1.3046875, "learning_rate": 3.340554540378591e-05, "loss": 0.1345, "step": 30177 }, { "epoch": 22.108424908424908, "grad_norm": 0.90234375, "learning_rate": 3.340232988439204e-05, "loss": 0.0911, "step": 30178 }, { "epoch": 22.10915750915751, "grad_norm": 1.1875, "learning_rate": 3.339911444691871e-05, "loss": 0.1096, "step": 30179 }, { "epoch": 22.10989010989011, "grad_norm": 1.609375, "learning_rate": 3.3395899091381344e-05, "loss": 0.1637, "step": 30180 }, { "epoch": 22.11062271062271, "grad_norm": 0.85546875, "learning_rate": 3.3392683817795345e-05, "loss": 0.1072, "step": 30181 }, { "epoch": 22.11135531135531, "grad_norm": 1.0, "learning_rate": 3.338946862617608e-05, "loss": 0.1207, "step": 30182 }, { "epoch": 22.11208791208791, "grad_norm": 1.5703125, "learning_rate": 3.3386253516539015e-05, "loss": 0.135, "step": 30183 }, { "epoch": 22.112820512820512, "grad_norm": 1.1875, "learning_rate": 3.338303848889953e-05, "loss": 0.0993, "step": 30184 }, { "epoch": 22.113553113553113, "grad_norm": 1.171875, "learning_rate": 3.337982354327303e-05, "loss": 0.0978, "step": 30185 }, { "epoch": 22.114285714285714, "grad_norm": 1.1796875, "learning_rate": 3.3376608679674894e-05, "loss": 0.133, "step": 30186 }, { "epoch": 22.115018315018315, "grad_norm": 1.21875, "learning_rate": 3.3373393898120556e-05, "loss": 0.1184, "step": 30187 }, { "epoch": 22.115750915750915, "grad_norm": 0.98046875, "learning_rate": 3.3370179198625415e-05, "loss": 0.1425, "step": 30188 }, { "epoch": 22.116483516483516, "grad_norm": 1.03125, "learning_rate": 3.3366964581204866e-05, "loss": 0.1252, "step": 30189 }, { "epoch": 22.117216117216117, "grad_norm": 1.078125, "learning_rate": 3.336375004587431e-05, "loss": 0.1345, "step": 30190 }, { "epoch": 22.117948717948718, "grad_norm": 1.40625, "learning_rate": 3.336053559264917e-05, "loss": 0.153, "step": 30191 }, { "epoch": 22.11868131868132, "grad_norm": 0.87890625, "learning_rate": 3.33573212215448e-05, "loss": 0.1305, "step": 30192 }, { "epoch": 22.11941391941392, "grad_norm": 1.03125, "learning_rate": 3.335410693257666e-05, "loss": 0.1286, "step": 30193 }, { "epoch": 22.12014652014652, "grad_norm": 0.98828125, "learning_rate": 3.3350892725760114e-05, "loss": 0.1272, "step": 30194 }, { "epoch": 22.12087912087912, "grad_norm": 0.6640625, "learning_rate": 3.334767860111057e-05, "loss": 0.1284, "step": 30195 }, { "epoch": 22.12161172161172, "grad_norm": 1.0, "learning_rate": 3.334446455864343e-05, "loss": 0.1205, "step": 30196 }, { "epoch": 22.122344322344322, "grad_norm": 1.421875, "learning_rate": 3.334125059837408e-05, "loss": 0.1163, "step": 30197 }, { "epoch": 22.123076923076923, "grad_norm": 0.9765625, "learning_rate": 3.333803672031794e-05, "loss": 0.1176, "step": 30198 }, { "epoch": 22.123809523809523, "grad_norm": 0.6328125, "learning_rate": 3.333482292449041e-05, "loss": 0.1133, "step": 30199 }, { "epoch": 22.124542124542124, "grad_norm": 1.484375, "learning_rate": 3.3331609210906876e-05, "loss": 0.1255, "step": 30200 }, { "epoch": 22.125274725274725, "grad_norm": 1.5234375, "learning_rate": 3.3328395579582725e-05, "loss": 0.1114, "step": 30201 }, { "epoch": 22.126007326007326, "grad_norm": 1.5546875, "learning_rate": 3.332518203053337e-05, "loss": 0.1443, "step": 30202 }, { "epoch": 22.126739926739926, "grad_norm": 1.171875, "learning_rate": 3.33219685637742e-05, "loss": 0.1509, "step": 30203 }, { "epoch": 22.127472527472527, "grad_norm": 1.5, "learning_rate": 3.3318755179320636e-05, "loss": 0.1431, "step": 30204 }, { "epoch": 22.128205128205128, "grad_norm": 1.0703125, "learning_rate": 3.331554187718805e-05, "loss": 0.1512, "step": 30205 }, { "epoch": 22.12893772893773, "grad_norm": 1.359375, "learning_rate": 3.331232865739185e-05, "loss": 0.0966, "step": 30206 }, { "epoch": 22.12967032967033, "grad_norm": 1.0078125, "learning_rate": 3.330911551994742e-05, "loss": 0.1067, "step": 30207 }, { "epoch": 22.13040293040293, "grad_norm": 1.0390625, "learning_rate": 3.330590246487014e-05, "loss": 0.1561, "step": 30208 }, { "epoch": 22.13113553113553, "grad_norm": 1.0859375, "learning_rate": 3.3302689492175445e-05, "loss": 0.1081, "step": 30209 }, { "epoch": 22.13186813186813, "grad_norm": 1.3515625, "learning_rate": 3.329947660187872e-05, "loss": 0.1344, "step": 30210 }, { "epoch": 22.132600732600732, "grad_norm": 0.5703125, "learning_rate": 3.329626379399535e-05, "loss": 0.1058, "step": 30211 }, { "epoch": 22.133333333333333, "grad_norm": 0.765625, "learning_rate": 3.329305106854071e-05, "loss": 0.1229, "step": 30212 }, { "epoch": 22.134065934065934, "grad_norm": 0.7265625, "learning_rate": 3.3289838425530215e-05, "loss": 0.1121, "step": 30213 }, { "epoch": 22.134798534798534, "grad_norm": 1.015625, "learning_rate": 3.328662586497926e-05, "loss": 0.0874, "step": 30214 }, { "epoch": 22.135531135531135, "grad_norm": 0.8359375, "learning_rate": 3.328341338690322e-05, "loss": 0.1275, "step": 30215 }, { "epoch": 22.136263736263736, "grad_norm": 0.671875, "learning_rate": 3.328020099131751e-05, "loss": 0.1213, "step": 30216 }, { "epoch": 22.136996336996337, "grad_norm": 0.9375, "learning_rate": 3.327698867823751e-05, "loss": 0.1168, "step": 30217 }, { "epoch": 22.137728937728937, "grad_norm": 1.265625, "learning_rate": 3.32737764476786e-05, "loss": 0.186, "step": 30218 }, { "epoch": 22.138461538461538, "grad_norm": 0.80859375, "learning_rate": 3.3270564299656195e-05, "loss": 0.1059, "step": 30219 }, { "epoch": 22.13919413919414, "grad_norm": 1.6328125, "learning_rate": 3.3267352234185665e-05, "loss": 0.1315, "step": 30220 }, { "epoch": 22.13992673992674, "grad_norm": 1.015625, "learning_rate": 3.326414025128241e-05, "loss": 0.1116, "step": 30221 }, { "epoch": 22.14065934065934, "grad_norm": 1.234375, "learning_rate": 3.326092835096182e-05, "loss": 0.1238, "step": 30222 }, { "epoch": 22.14139194139194, "grad_norm": 1.0625, "learning_rate": 3.325771653323927e-05, "loss": 0.096, "step": 30223 }, { "epoch": 22.14212454212454, "grad_norm": 1.296875, "learning_rate": 3.325450479813017e-05, "loss": 0.1632, "step": 30224 }, { "epoch": 22.142857142857142, "grad_norm": 1.6171875, "learning_rate": 3.32512931456499e-05, "loss": 0.1325, "step": 30225 }, { "epoch": 22.143589743589743, "grad_norm": 0.68359375, "learning_rate": 3.3248081575813844e-05, "loss": 0.119, "step": 30226 }, { "epoch": 22.144322344322344, "grad_norm": 0.88671875, "learning_rate": 3.32448700886374e-05, "loss": 0.1383, "step": 30227 }, { "epoch": 22.145054945054945, "grad_norm": 1.7265625, "learning_rate": 3.324165868413594e-05, "loss": 0.1131, "step": 30228 }, { "epoch": 22.145787545787545, "grad_norm": 1.0546875, "learning_rate": 3.323844736232485e-05, "loss": 0.1308, "step": 30229 }, { "epoch": 22.146520146520146, "grad_norm": 0.96484375, "learning_rate": 3.323523612321954e-05, "loss": 0.1019, "step": 30230 }, { "epoch": 22.147252747252747, "grad_norm": 1.3046875, "learning_rate": 3.323202496683537e-05, "loss": 0.1165, "step": 30231 }, { "epoch": 22.147985347985347, "grad_norm": 0.9296875, "learning_rate": 3.322881389318775e-05, "loss": 0.1202, "step": 30232 }, { "epoch": 22.148717948717948, "grad_norm": 0.83984375, "learning_rate": 3.322560290229205e-05, "loss": 0.1, "step": 30233 }, { "epoch": 22.14945054945055, "grad_norm": 1.0625, "learning_rate": 3.3222391994163645e-05, "loss": 0.1293, "step": 30234 }, { "epoch": 22.15018315018315, "grad_norm": 0.77734375, "learning_rate": 3.321918116881794e-05, "loss": 0.1222, "step": 30235 }, { "epoch": 22.15091575091575, "grad_norm": 1.078125, "learning_rate": 3.321597042627032e-05, "loss": 0.1402, "step": 30236 }, { "epoch": 22.15164835164835, "grad_norm": 0.95703125, "learning_rate": 3.3212759766536156e-05, "loss": 0.122, "step": 30237 }, { "epoch": 22.152380952380952, "grad_norm": 0.92578125, "learning_rate": 3.3209549189630824e-05, "loss": 0.1042, "step": 30238 }, { "epoch": 22.153113553113553, "grad_norm": 0.8671875, "learning_rate": 3.320633869556971e-05, "loss": 0.1267, "step": 30239 }, { "epoch": 22.153846153846153, "grad_norm": 1.2109375, "learning_rate": 3.320312828436822e-05, "loss": 0.1128, "step": 30240 }, { "epoch": 22.154578754578754, "grad_norm": 0.85546875, "learning_rate": 3.319991795604172e-05, "loss": 0.107, "step": 30241 }, { "epoch": 22.155311355311355, "grad_norm": 0.91796875, "learning_rate": 3.31967077106056e-05, "loss": 0.1321, "step": 30242 }, { "epoch": 22.156043956043955, "grad_norm": 1.21875, "learning_rate": 3.319349754807523e-05, "loss": 0.1051, "step": 30243 }, { "epoch": 22.156776556776556, "grad_norm": 1.0078125, "learning_rate": 3.319028746846597e-05, "loss": 0.1183, "step": 30244 }, { "epoch": 22.157509157509157, "grad_norm": 0.97265625, "learning_rate": 3.318707747179325e-05, "loss": 0.106, "step": 30245 }, { "epoch": 22.158241758241758, "grad_norm": 0.91015625, "learning_rate": 3.3183867558072417e-05, "loss": 0.1379, "step": 30246 }, { "epoch": 22.15897435897436, "grad_norm": 0.91015625, "learning_rate": 3.318065772731886e-05, "loss": 0.1229, "step": 30247 }, { "epoch": 22.15970695970696, "grad_norm": 0.9921875, "learning_rate": 3.317744797954797e-05, "loss": 0.1049, "step": 30248 }, { "epoch": 22.16043956043956, "grad_norm": 0.95703125, "learning_rate": 3.317423831477509e-05, "loss": 0.1382, "step": 30249 }, { "epoch": 22.16117216117216, "grad_norm": 1.15625, "learning_rate": 3.317102873301564e-05, "loss": 0.0918, "step": 30250 }, { "epoch": 22.16190476190476, "grad_norm": 1.0546875, "learning_rate": 3.316781923428497e-05, "loss": 0.1254, "step": 30251 }, { "epoch": 22.162637362637362, "grad_norm": 0.65625, "learning_rate": 3.316460981859847e-05, "loss": 0.1074, "step": 30252 }, { "epoch": 22.163369963369963, "grad_norm": 0.69921875, "learning_rate": 3.316140048597153e-05, "loss": 0.1126, "step": 30253 }, { "epoch": 22.164102564102564, "grad_norm": 1.328125, "learning_rate": 3.315819123641949e-05, "loss": 0.1407, "step": 30254 }, { "epoch": 22.164835164835164, "grad_norm": 0.8984375, "learning_rate": 3.3154982069957746e-05, "loss": 0.1366, "step": 30255 }, { "epoch": 22.165567765567765, "grad_norm": 0.86328125, "learning_rate": 3.31517729866017e-05, "loss": 0.1357, "step": 30256 }, { "epoch": 22.166300366300366, "grad_norm": 1.0703125, "learning_rate": 3.31485639863667e-05, "loss": 0.1109, "step": 30257 }, { "epoch": 22.167032967032966, "grad_norm": 1.0859375, "learning_rate": 3.314535506926812e-05, "loss": 0.1064, "step": 30258 }, { "epoch": 22.167765567765567, "grad_norm": 1.0859375, "learning_rate": 3.3142146235321346e-05, "loss": 0.1206, "step": 30259 }, { "epoch": 22.168498168498168, "grad_norm": 1.4921875, "learning_rate": 3.3138937484541725e-05, "loss": 0.1194, "step": 30260 }, { "epoch": 22.16923076923077, "grad_norm": 0.6484375, "learning_rate": 3.3135728816944675e-05, "loss": 0.1123, "step": 30261 }, { "epoch": 22.16996336996337, "grad_norm": 1.09375, "learning_rate": 3.313252023254555e-05, "loss": 0.1045, "step": 30262 }, { "epoch": 22.17069597069597, "grad_norm": 1.0625, "learning_rate": 3.312931173135973e-05, "loss": 0.1293, "step": 30263 }, { "epoch": 22.17142857142857, "grad_norm": 1.8046875, "learning_rate": 3.312610331340256e-05, "loss": 0.1366, "step": 30264 }, { "epoch": 22.17216117216117, "grad_norm": 0.796875, "learning_rate": 3.312289497868943e-05, "loss": 0.125, "step": 30265 }, { "epoch": 22.172893772893772, "grad_norm": 1.328125, "learning_rate": 3.3119686727235725e-05, "loss": 0.1288, "step": 30266 }, { "epoch": 22.173626373626373, "grad_norm": 0.97265625, "learning_rate": 3.31164785590568e-05, "loss": 0.0984, "step": 30267 }, { "epoch": 22.174358974358974, "grad_norm": 0.7109375, "learning_rate": 3.3113270474168046e-05, "loss": 0.1165, "step": 30268 }, { "epoch": 22.175091575091574, "grad_norm": 1.40625, "learning_rate": 3.311006247258481e-05, "loss": 0.1157, "step": 30269 }, { "epoch": 22.175824175824175, "grad_norm": 1.0625, "learning_rate": 3.310685455432247e-05, "loss": 0.0953, "step": 30270 }, { "epoch": 22.176556776556776, "grad_norm": 0.86328125, "learning_rate": 3.310364671939641e-05, "loss": 0.1105, "step": 30271 }, { "epoch": 22.177289377289377, "grad_norm": 1.6328125, "learning_rate": 3.310043896782198e-05, "loss": 0.1259, "step": 30272 }, { "epoch": 22.178021978021977, "grad_norm": 1.296875, "learning_rate": 3.309723129961458e-05, "loss": 0.1414, "step": 30273 }, { "epoch": 22.178754578754578, "grad_norm": 1.84375, "learning_rate": 3.3094023714789544e-05, "loss": 0.1472, "step": 30274 }, { "epoch": 22.17948717948718, "grad_norm": 1.328125, "learning_rate": 3.309081621336224e-05, "loss": 0.1269, "step": 30275 }, { "epoch": 22.18021978021978, "grad_norm": 1.7265625, "learning_rate": 3.308760879534806e-05, "loss": 0.1301, "step": 30276 }, { "epoch": 22.18095238095238, "grad_norm": 1.125, "learning_rate": 3.308440146076237e-05, "loss": 0.1297, "step": 30277 }, { "epoch": 22.18168498168498, "grad_norm": 1.1171875, "learning_rate": 3.308119420962052e-05, "loss": 0.1229, "step": 30278 }, { "epoch": 22.18241758241758, "grad_norm": 1.40625, "learning_rate": 3.3077987041937904e-05, "loss": 0.1333, "step": 30279 }, { "epoch": 22.183150183150182, "grad_norm": 1.078125, "learning_rate": 3.307477995772986e-05, "loss": 0.1399, "step": 30280 }, { "epoch": 22.183882783882783, "grad_norm": 0.921875, "learning_rate": 3.3071572957011746e-05, "loss": 0.1218, "step": 30281 }, { "epoch": 22.184615384615384, "grad_norm": 0.90625, "learning_rate": 3.306836603979898e-05, "loss": 0.1206, "step": 30282 }, { "epoch": 22.185347985347985, "grad_norm": 1.140625, "learning_rate": 3.3065159206106885e-05, "loss": 0.1014, "step": 30283 }, { "epoch": 22.186080586080585, "grad_norm": 1.515625, "learning_rate": 3.3061952455950834e-05, "loss": 0.1398, "step": 30284 }, { "epoch": 22.186813186813186, "grad_norm": 0.99609375, "learning_rate": 3.305874578934619e-05, "loss": 0.0832, "step": 30285 }, { "epoch": 22.187545787545787, "grad_norm": 1.3046875, "learning_rate": 3.305553920630831e-05, "loss": 0.1091, "step": 30286 }, { "epoch": 22.188278388278388, "grad_norm": 0.8984375, "learning_rate": 3.305233270685258e-05, "loss": 0.1163, "step": 30287 }, { "epoch": 22.18901098901099, "grad_norm": 1.0703125, "learning_rate": 3.304912629099436e-05, "loss": 0.1056, "step": 30288 }, { "epoch": 22.18974358974359, "grad_norm": 1.4609375, "learning_rate": 3.3045919958749004e-05, "loss": 0.1278, "step": 30289 }, { "epoch": 22.19047619047619, "grad_norm": 0.8984375, "learning_rate": 3.3042713710131854e-05, "loss": 0.1131, "step": 30290 }, { "epoch": 22.19120879120879, "grad_norm": 1.40625, "learning_rate": 3.30395075451583e-05, "loss": 0.1559, "step": 30291 }, { "epoch": 22.19194139194139, "grad_norm": 1.1484375, "learning_rate": 3.30363014638437e-05, "loss": 0.1549, "step": 30292 }, { "epoch": 22.192673992673992, "grad_norm": 1.2734375, "learning_rate": 3.3033095466203424e-05, "loss": 0.1101, "step": 30293 }, { "epoch": 22.193406593406593, "grad_norm": 1.0625, "learning_rate": 3.302988955225282e-05, "loss": 0.1153, "step": 30294 }, { "epoch": 22.194139194139193, "grad_norm": 0.66796875, "learning_rate": 3.3026683722007246e-05, "loss": 0.1105, "step": 30295 }, { "epoch": 22.194871794871794, "grad_norm": 1.390625, "learning_rate": 3.302347797548205e-05, "loss": 0.1283, "step": 30296 }, { "epoch": 22.195604395604395, "grad_norm": 1.078125, "learning_rate": 3.302027231269263e-05, "loss": 0.1002, "step": 30297 }, { "epoch": 22.196336996336996, "grad_norm": 0.99609375, "learning_rate": 3.301706673365432e-05, "loss": 0.1199, "step": 30298 }, { "epoch": 22.197069597069596, "grad_norm": 1.359375, "learning_rate": 3.301386123838248e-05, "loss": 0.1185, "step": 30299 }, { "epoch": 22.197802197802197, "grad_norm": 1.390625, "learning_rate": 3.301065582689248e-05, "loss": 0.132, "step": 30300 }, { "epoch": 22.198534798534798, "grad_norm": 1.359375, "learning_rate": 3.300745049919964e-05, "loss": 0.1324, "step": 30301 }, { "epoch": 22.1992673992674, "grad_norm": 0.83984375, "learning_rate": 3.3004245255319374e-05, "loss": 0.1351, "step": 30302 }, { "epoch": 22.2, "grad_norm": 1.015625, "learning_rate": 3.300104009526701e-05, "loss": 0.1167, "step": 30303 }, { "epoch": 22.2007326007326, "grad_norm": 1.015625, "learning_rate": 3.2997835019057896e-05, "loss": 0.1251, "step": 30304 }, { "epoch": 22.2014652014652, "grad_norm": 1.3359375, "learning_rate": 3.2994630026707416e-05, "loss": 0.1716, "step": 30305 }, { "epoch": 22.2021978021978, "grad_norm": 1.328125, "learning_rate": 3.2991425118230904e-05, "loss": 0.1298, "step": 30306 }, { "epoch": 22.202930402930402, "grad_norm": 1.625, "learning_rate": 3.29882202936437e-05, "loss": 0.138, "step": 30307 }, { "epoch": 22.203663003663003, "grad_norm": 0.86328125, "learning_rate": 3.2985015552961205e-05, "loss": 0.1249, "step": 30308 }, { "epoch": 22.204395604395604, "grad_norm": 0.66015625, "learning_rate": 3.2981810896198755e-05, "loss": 0.11, "step": 30309 }, { "epoch": 22.205128205128204, "grad_norm": 1.59375, "learning_rate": 3.297860632337169e-05, "loss": 0.1231, "step": 30310 }, { "epoch": 22.205860805860805, "grad_norm": 0.75, "learning_rate": 3.297540183449538e-05, "loss": 0.12, "step": 30311 }, { "epoch": 22.206593406593406, "grad_norm": 1.1640625, "learning_rate": 3.2972197429585154e-05, "loss": 0.1445, "step": 30312 }, { "epoch": 22.207326007326007, "grad_norm": 1.2421875, "learning_rate": 3.29689931086564e-05, "loss": 0.1141, "step": 30313 }, { "epoch": 22.208058608058607, "grad_norm": 1.0703125, "learning_rate": 3.296578887172446e-05, "loss": 0.1197, "step": 30314 }, { "epoch": 22.208791208791208, "grad_norm": 0.91015625, "learning_rate": 3.296258471880468e-05, "loss": 0.1139, "step": 30315 }, { "epoch": 22.20952380952381, "grad_norm": 1.0859375, "learning_rate": 3.2959380649912395e-05, "loss": 0.1231, "step": 30316 }, { "epoch": 22.21025641025641, "grad_norm": 1.0546875, "learning_rate": 3.2956176665062985e-05, "loss": 0.1286, "step": 30317 }, { "epoch": 22.21098901098901, "grad_norm": 0.94921875, "learning_rate": 3.295297276427179e-05, "loss": 0.1275, "step": 30318 }, { "epoch": 22.21172161172161, "grad_norm": 1.390625, "learning_rate": 3.294976894755417e-05, "loss": 0.1575, "step": 30319 }, { "epoch": 22.21245421245421, "grad_norm": 0.96484375, "learning_rate": 3.2946565214925474e-05, "loss": 0.12, "step": 30320 }, { "epoch": 22.213186813186812, "grad_norm": 0.6640625, "learning_rate": 3.2943361566401046e-05, "loss": 0.1119, "step": 30321 }, { "epoch": 22.213919413919413, "grad_norm": 0.6796875, "learning_rate": 3.2940158001996205e-05, "loss": 0.1096, "step": 30322 }, { "epoch": 22.214652014652014, "grad_norm": 1.453125, "learning_rate": 3.293695452172637e-05, "loss": 0.1338, "step": 30323 }, { "epoch": 22.215384615384615, "grad_norm": 1.5, "learning_rate": 3.293375112560684e-05, "loss": 0.1327, "step": 30324 }, { "epoch": 22.216117216117215, "grad_norm": 0.80859375, "learning_rate": 3.293054781365297e-05, "loss": 0.1126, "step": 30325 }, { "epoch": 22.216849816849816, "grad_norm": 0.890625, "learning_rate": 3.292734458588012e-05, "loss": 0.1308, "step": 30326 }, { "epoch": 22.217582417582417, "grad_norm": 1.1484375, "learning_rate": 3.292414144230362e-05, "loss": 0.1069, "step": 30327 }, { "epoch": 22.218315018315018, "grad_norm": 1.1796875, "learning_rate": 3.2920938382938834e-05, "loss": 0.1375, "step": 30328 }, { "epoch": 22.21904761904762, "grad_norm": 1.1640625, "learning_rate": 3.2917735407801106e-05, "loss": 0.0977, "step": 30329 }, { "epoch": 22.21978021978022, "grad_norm": 1.078125, "learning_rate": 3.2914532516905765e-05, "loss": 0.1357, "step": 30330 }, { "epoch": 22.22051282051282, "grad_norm": 1.1015625, "learning_rate": 3.29113297102682e-05, "loss": 0.116, "step": 30331 }, { "epoch": 22.22124542124542, "grad_norm": 1.296875, "learning_rate": 3.2908126987903704e-05, "loss": 0.1247, "step": 30332 }, { "epoch": 22.22197802197802, "grad_norm": 1.1015625, "learning_rate": 3.2904924349827646e-05, "loss": 0.1449, "step": 30333 }, { "epoch": 22.222710622710622, "grad_norm": 0.8515625, "learning_rate": 3.290172179605538e-05, "loss": 0.0915, "step": 30334 }, { "epoch": 22.223443223443223, "grad_norm": 1.25, "learning_rate": 3.289851932660225e-05, "loss": 0.1171, "step": 30335 }, { "epoch": 22.224175824175823, "grad_norm": 0.6015625, "learning_rate": 3.2895316941483586e-05, "loss": 0.1204, "step": 30336 }, { "epoch": 22.224908424908424, "grad_norm": 1.0390625, "learning_rate": 3.289211464071474e-05, "loss": 0.1191, "step": 30337 }, { "epoch": 22.225641025641025, "grad_norm": 1.0703125, "learning_rate": 3.2888912424311034e-05, "loss": 0.1356, "step": 30338 }, { "epoch": 22.226373626373626, "grad_norm": 1.4921875, "learning_rate": 3.288571029228785e-05, "loss": 0.1091, "step": 30339 }, { "epoch": 22.227106227106226, "grad_norm": 0.8984375, "learning_rate": 3.2882508244660504e-05, "loss": 0.1293, "step": 30340 }, { "epoch": 22.227838827838827, "grad_norm": 1.0703125, "learning_rate": 3.287930628144434e-05, "loss": 0.1126, "step": 30341 }, { "epoch": 22.228571428571428, "grad_norm": 1.0234375, "learning_rate": 3.2876104402654716e-05, "loss": 0.1396, "step": 30342 }, { "epoch": 22.22930402930403, "grad_norm": 1.0703125, "learning_rate": 3.287290260830695e-05, "loss": 0.0856, "step": 30343 }, { "epoch": 22.23003663003663, "grad_norm": 0.90625, "learning_rate": 3.28697008984164e-05, "loss": 0.1329, "step": 30344 }, { "epoch": 22.23076923076923, "grad_norm": 1.4140625, "learning_rate": 3.2866499272998406e-05, "loss": 0.1229, "step": 30345 }, { "epoch": 22.23150183150183, "grad_norm": 1.4140625, "learning_rate": 3.2863297732068295e-05, "loss": 0.1396, "step": 30346 }, { "epoch": 22.23223443223443, "grad_norm": 0.66015625, "learning_rate": 3.286009627564142e-05, "loss": 0.0981, "step": 30347 }, { "epoch": 22.232967032967032, "grad_norm": 1.40625, "learning_rate": 3.285689490373309e-05, "loss": 0.1497, "step": 30348 }, { "epoch": 22.233699633699633, "grad_norm": 0.91796875, "learning_rate": 3.28536936163587e-05, "loss": 0.1035, "step": 30349 }, { "epoch": 22.234432234432234, "grad_norm": 1.0078125, "learning_rate": 3.2850492413533545e-05, "loss": 0.1103, "step": 30350 }, { "epoch": 22.235164835164834, "grad_norm": 1.0703125, "learning_rate": 3.2847291295272984e-05, "loss": 0.1148, "step": 30351 }, { "epoch": 22.235897435897435, "grad_norm": 0.80078125, "learning_rate": 3.2844090261592335e-05, "loss": 0.0984, "step": 30352 }, { "epoch": 22.236630036630036, "grad_norm": 0.6875, "learning_rate": 3.2840889312506933e-05, "loss": 0.1244, "step": 30353 }, { "epoch": 22.237362637362637, "grad_norm": 0.96875, "learning_rate": 3.283768844803215e-05, "loss": 0.1352, "step": 30354 }, { "epoch": 22.238095238095237, "grad_norm": 1.390625, "learning_rate": 3.283448766818329e-05, "loss": 0.1134, "step": 30355 }, { "epoch": 22.238827838827838, "grad_norm": 1.25, "learning_rate": 3.2831286972975695e-05, "loss": 0.1093, "step": 30356 }, { "epoch": 22.23956043956044, "grad_norm": 0.734375, "learning_rate": 3.282808636242472e-05, "loss": 0.1104, "step": 30357 }, { "epoch": 22.24029304029304, "grad_norm": 0.9296875, "learning_rate": 3.2824885836545673e-05, "loss": 0.1184, "step": 30358 }, { "epoch": 22.24102564102564, "grad_norm": 0.98046875, "learning_rate": 3.2821685395353885e-05, "loss": 0.121, "step": 30359 }, { "epoch": 22.24175824175824, "grad_norm": 1.5390625, "learning_rate": 3.2818485038864735e-05, "loss": 0.1487, "step": 30360 }, { "epoch": 22.24249084249084, "grad_norm": 0.859375, "learning_rate": 3.2815284767093515e-05, "loss": 0.1117, "step": 30361 }, { "epoch": 22.243223443223442, "grad_norm": 0.734375, "learning_rate": 3.2812084580055576e-05, "loss": 0.1109, "step": 30362 }, { "epoch": 22.243956043956043, "grad_norm": 1.5625, "learning_rate": 3.2808884477766244e-05, "loss": 0.1219, "step": 30363 }, { "epoch": 22.244688644688644, "grad_norm": 1.1171875, "learning_rate": 3.280568446024084e-05, "loss": 0.1184, "step": 30364 }, { "epoch": 22.245421245421245, "grad_norm": 1.0703125, "learning_rate": 3.280248452749472e-05, "loss": 0.1397, "step": 30365 }, { "epoch": 22.246153846153845, "grad_norm": 0.96484375, "learning_rate": 3.279928467954321e-05, "loss": 0.1182, "step": 30366 }, { "epoch": 22.246886446886446, "grad_norm": 1.0859375, "learning_rate": 3.279608491640163e-05, "loss": 0.1332, "step": 30367 }, { "epoch": 22.247619047619047, "grad_norm": 0.73828125, "learning_rate": 3.279288523808533e-05, "loss": 0.1287, "step": 30368 }, { "epoch": 22.248351648351647, "grad_norm": 0.89453125, "learning_rate": 3.278968564460962e-05, "loss": 0.135, "step": 30369 }, { "epoch": 22.249084249084248, "grad_norm": 0.921875, "learning_rate": 3.278648613598985e-05, "loss": 0.1258, "step": 30370 }, { "epoch": 22.24981684981685, "grad_norm": 1.0078125, "learning_rate": 3.278328671224133e-05, "loss": 0.1185, "step": 30371 }, { "epoch": 22.25054945054945, "grad_norm": 0.94140625, "learning_rate": 3.278008737337941e-05, "loss": 0.1144, "step": 30372 }, { "epoch": 22.25128205128205, "grad_norm": 0.9375, "learning_rate": 3.277688811941941e-05, "loss": 0.1064, "step": 30373 }, { "epoch": 22.25201465201465, "grad_norm": 0.76953125, "learning_rate": 3.2773688950376627e-05, "loss": 0.1069, "step": 30374 }, { "epoch": 22.252747252747252, "grad_norm": 1.328125, "learning_rate": 3.277048986626645e-05, "loss": 0.1341, "step": 30375 }, { "epoch": 22.253479853479853, "grad_norm": 1.5703125, "learning_rate": 3.276729086710417e-05, "loss": 0.1108, "step": 30376 }, { "epoch": 22.254212454212453, "grad_norm": 1.171875, "learning_rate": 3.276409195290512e-05, "loss": 0.16, "step": 30377 }, { "epoch": 22.254945054945054, "grad_norm": 1.0546875, "learning_rate": 3.276089312368463e-05, "loss": 0.1112, "step": 30378 }, { "epoch": 22.255677655677655, "grad_norm": 1.078125, "learning_rate": 3.275769437945801e-05, "loss": 0.1282, "step": 30379 }, { "epoch": 22.256410256410255, "grad_norm": 0.68359375, "learning_rate": 3.275449572024062e-05, "loss": 0.1255, "step": 30380 }, { "epoch": 22.257142857142856, "grad_norm": 1.3203125, "learning_rate": 3.275129714604776e-05, "loss": 0.1218, "step": 30381 }, { "epoch": 22.257875457875457, "grad_norm": 0.83984375, "learning_rate": 3.274809865689475e-05, "loss": 0.1201, "step": 30382 }, { "epoch": 22.258608058608058, "grad_norm": 0.95703125, "learning_rate": 3.274490025279695e-05, "loss": 0.1298, "step": 30383 }, { "epoch": 22.25934065934066, "grad_norm": 0.71484375, "learning_rate": 3.2741701933769644e-05, "loss": 0.1146, "step": 30384 }, { "epoch": 22.26007326007326, "grad_norm": 0.8515625, "learning_rate": 3.273850369982817e-05, "loss": 0.1255, "step": 30385 }, { "epoch": 22.26080586080586, "grad_norm": 1.2265625, "learning_rate": 3.273530555098787e-05, "loss": 0.1417, "step": 30386 }, { "epoch": 22.26153846153846, "grad_norm": 0.73828125, "learning_rate": 3.273210748726405e-05, "loss": 0.128, "step": 30387 }, { "epoch": 22.26227106227106, "grad_norm": 0.70703125, "learning_rate": 3.272890950867203e-05, "loss": 0.1311, "step": 30388 }, { "epoch": 22.263003663003662, "grad_norm": 0.93359375, "learning_rate": 3.272571161522715e-05, "loss": 0.1259, "step": 30389 }, { "epoch": 22.263736263736263, "grad_norm": 1.203125, "learning_rate": 3.272251380694469e-05, "loss": 0.1369, "step": 30390 }, { "epoch": 22.264468864468864, "grad_norm": 1.109375, "learning_rate": 3.271931608384003e-05, "loss": 0.135, "step": 30391 }, { "epoch": 22.265201465201464, "grad_norm": 1.21875, "learning_rate": 3.271611844592847e-05, "loss": 0.126, "step": 30392 }, { "epoch": 22.265934065934065, "grad_norm": 1.1171875, "learning_rate": 3.2712920893225304e-05, "loss": 0.1064, "step": 30393 }, { "epoch": 22.266666666666666, "grad_norm": 1.1875, "learning_rate": 3.2709723425745884e-05, "loss": 0.1192, "step": 30394 }, { "epoch": 22.267399267399266, "grad_norm": 1.1171875, "learning_rate": 3.270652604350551e-05, "loss": 0.1155, "step": 30395 }, { "epoch": 22.268131868131867, "grad_norm": 1.03125, "learning_rate": 3.270332874651951e-05, "loss": 0.124, "step": 30396 }, { "epoch": 22.268864468864468, "grad_norm": 0.96484375, "learning_rate": 3.270013153480322e-05, "loss": 0.1502, "step": 30397 }, { "epoch": 22.26959706959707, "grad_norm": 0.70703125, "learning_rate": 3.269693440837194e-05, "loss": 0.1052, "step": 30398 }, { "epoch": 22.27032967032967, "grad_norm": 0.75, "learning_rate": 3.269373736724099e-05, "loss": 0.1188, "step": 30399 }, { "epoch": 22.27106227106227, "grad_norm": 0.9921875, "learning_rate": 3.2690540411425665e-05, "loss": 0.102, "step": 30400 }, { "epoch": 22.27179487179487, "grad_norm": 1.234375, "learning_rate": 3.2687343540941334e-05, "loss": 0.1133, "step": 30401 }, { "epoch": 22.27252747252747, "grad_norm": 0.96484375, "learning_rate": 3.2684146755803284e-05, "loss": 0.1439, "step": 30402 }, { "epoch": 22.273260073260072, "grad_norm": 1.0078125, "learning_rate": 3.268095005602684e-05, "loss": 0.1062, "step": 30403 }, { "epoch": 22.273992673992673, "grad_norm": 1.1953125, "learning_rate": 3.267775344162731e-05, "loss": 0.1081, "step": 30404 }, { "epoch": 22.274725274725274, "grad_norm": 1.0234375, "learning_rate": 3.2674556912619996e-05, "loss": 0.1189, "step": 30405 }, { "epoch": 22.275457875457874, "grad_norm": 0.7109375, "learning_rate": 3.267136046902025e-05, "loss": 0.1249, "step": 30406 }, { "epoch": 22.276190476190475, "grad_norm": 1.03125, "learning_rate": 3.266816411084336e-05, "loss": 0.1256, "step": 30407 }, { "epoch": 22.276923076923076, "grad_norm": 0.6796875, "learning_rate": 3.266496783810466e-05, "loss": 0.1132, "step": 30408 }, { "epoch": 22.277655677655677, "grad_norm": 1.3046875, "learning_rate": 3.266177165081946e-05, "loss": 0.1292, "step": 30409 }, { "epoch": 22.278388278388277, "grad_norm": 0.78515625, "learning_rate": 3.265857554900306e-05, "loss": 0.1501, "step": 30410 }, { "epoch": 22.279120879120878, "grad_norm": 0.76171875, "learning_rate": 3.2655379532670765e-05, "loss": 0.1116, "step": 30411 }, { "epoch": 22.27985347985348, "grad_norm": 1.40625, "learning_rate": 3.265218360183792e-05, "loss": 0.1351, "step": 30412 }, { "epoch": 22.28058608058608, "grad_norm": 1.1953125, "learning_rate": 3.264898775651983e-05, "loss": 0.1275, "step": 30413 }, { "epoch": 22.28131868131868, "grad_norm": 1.0546875, "learning_rate": 3.264579199673179e-05, "loss": 0.1346, "step": 30414 }, { "epoch": 22.28205128205128, "grad_norm": 0.8203125, "learning_rate": 3.264259632248913e-05, "loss": 0.104, "step": 30415 }, { "epoch": 22.282783882783882, "grad_norm": 1.2890625, "learning_rate": 3.263940073380713e-05, "loss": 0.1399, "step": 30416 }, { "epoch": 22.283516483516483, "grad_norm": 0.86328125, "learning_rate": 3.263620523070114e-05, "loss": 0.1357, "step": 30417 }, { "epoch": 22.284249084249083, "grad_norm": 0.98828125, "learning_rate": 3.263300981318646e-05, "loss": 0.1225, "step": 30418 }, { "epoch": 22.284981684981684, "grad_norm": 1.0625, "learning_rate": 3.262981448127839e-05, "loss": 0.0909, "step": 30419 }, { "epoch": 22.285714285714285, "grad_norm": 0.890625, "learning_rate": 3.262661923499224e-05, "loss": 0.1119, "step": 30420 }, { "epoch": 22.286446886446885, "grad_norm": 0.9140625, "learning_rate": 3.2623424074343325e-05, "loss": 0.1184, "step": 30421 }, { "epoch": 22.287179487179486, "grad_norm": 1.0234375, "learning_rate": 3.2620228999346954e-05, "loss": 0.1088, "step": 30422 }, { "epoch": 22.287912087912087, "grad_norm": 1.2890625, "learning_rate": 3.2617034010018444e-05, "loss": 0.1135, "step": 30423 }, { "epoch": 22.288644688644688, "grad_norm": 1.2421875, "learning_rate": 3.26138391063731e-05, "loss": 0.1308, "step": 30424 }, { "epoch": 22.28937728937729, "grad_norm": 0.828125, "learning_rate": 3.2610644288426214e-05, "loss": 0.0828, "step": 30425 }, { "epoch": 22.29010989010989, "grad_norm": 1.1875, "learning_rate": 3.2607449556193095e-05, "loss": 0.1188, "step": 30426 }, { "epoch": 22.29084249084249, "grad_norm": 0.6328125, "learning_rate": 3.260425490968907e-05, "loss": 0.116, "step": 30427 }, { "epoch": 22.29157509157509, "grad_norm": 0.80859375, "learning_rate": 3.260106034892943e-05, "loss": 0.1222, "step": 30428 }, { "epoch": 22.29230769230769, "grad_norm": 1.0234375, "learning_rate": 3.2597865873929495e-05, "loss": 0.1302, "step": 30429 }, { "epoch": 22.293040293040292, "grad_norm": 1.6953125, "learning_rate": 3.259467148470456e-05, "loss": 0.1247, "step": 30430 }, { "epoch": 22.293772893772893, "grad_norm": 1.703125, "learning_rate": 3.259147718126992e-05, "loss": 0.1213, "step": 30431 }, { "epoch": 22.294505494505493, "grad_norm": 1.296875, "learning_rate": 3.2588282963640895e-05, "loss": 0.1253, "step": 30432 }, { "epoch": 22.295238095238094, "grad_norm": 1.140625, "learning_rate": 3.2585088831832786e-05, "loss": 0.1464, "step": 30433 }, { "epoch": 22.295970695970695, "grad_norm": 1.0625, "learning_rate": 3.258189478586091e-05, "loss": 0.1325, "step": 30434 }, { "epoch": 22.296703296703296, "grad_norm": 1.359375, "learning_rate": 3.257870082574055e-05, "loss": 0.1416, "step": 30435 }, { "epoch": 22.297435897435896, "grad_norm": 1.3125, "learning_rate": 3.257550695148703e-05, "loss": 0.1014, "step": 30436 }, { "epoch": 22.298168498168497, "grad_norm": 1.1015625, "learning_rate": 3.257231316311561e-05, "loss": 0.1077, "step": 30437 }, { "epoch": 22.298901098901098, "grad_norm": 0.98828125, "learning_rate": 3.256911946064165e-05, "loss": 0.1027, "step": 30438 }, { "epoch": 22.2996336996337, "grad_norm": 1.140625, "learning_rate": 3.2565925844080415e-05, "loss": 0.1049, "step": 30439 }, { "epoch": 22.3003663003663, "grad_norm": 0.734375, "learning_rate": 3.2562732313447226e-05, "loss": 0.1082, "step": 30440 }, { "epoch": 22.3010989010989, "grad_norm": 0.93359375, "learning_rate": 3.2559538868757367e-05, "loss": 0.1137, "step": 30441 }, { "epoch": 22.3018315018315, "grad_norm": 1.3671875, "learning_rate": 3.255634551002613e-05, "loss": 0.1314, "step": 30442 }, { "epoch": 22.3025641025641, "grad_norm": 0.99609375, "learning_rate": 3.255315223726885e-05, "loss": 0.1117, "step": 30443 }, { "epoch": 22.303296703296702, "grad_norm": 1.2890625, "learning_rate": 3.2549959050500815e-05, "loss": 0.1418, "step": 30444 }, { "epoch": 22.304029304029303, "grad_norm": 1.1953125, "learning_rate": 3.254676594973731e-05, "loss": 0.1003, "step": 30445 }, { "epoch": 22.304761904761904, "grad_norm": 1.1796875, "learning_rate": 3.254357293499364e-05, "loss": 0.1145, "step": 30446 }, { "epoch": 22.305494505494504, "grad_norm": 1.078125, "learning_rate": 3.2540380006285116e-05, "loss": 0.1091, "step": 30447 }, { "epoch": 22.306227106227105, "grad_norm": 1.671875, "learning_rate": 3.2537187163627015e-05, "loss": 0.1258, "step": 30448 }, { "epoch": 22.306959706959706, "grad_norm": 1.4375, "learning_rate": 3.253399440703466e-05, "loss": 0.1265, "step": 30449 }, { "epoch": 22.307692307692307, "grad_norm": 0.83984375, "learning_rate": 3.253080173652333e-05, "loss": 0.1035, "step": 30450 }, { "epoch": 22.308424908424907, "grad_norm": 0.70703125, "learning_rate": 3.2527609152108335e-05, "loss": 0.0996, "step": 30451 }, { "epoch": 22.309157509157508, "grad_norm": 0.796875, "learning_rate": 3.2524416653804946e-05, "loss": 0.1382, "step": 30452 }, { "epoch": 22.30989010989011, "grad_norm": 1.046875, "learning_rate": 3.252122424162848e-05, "loss": 0.1025, "step": 30453 }, { "epoch": 22.31062271062271, "grad_norm": 1.4453125, "learning_rate": 3.2518031915594255e-05, "loss": 0.0948, "step": 30454 }, { "epoch": 22.31135531135531, "grad_norm": 1.0078125, "learning_rate": 3.2514839675717534e-05, "loss": 0.1095, "step": 30455 }, { "epoch": 22.31208791208791, "grad_norm": 0.921875, "learning_rate": 3.25116475220136e-05, "loss": 0.1556, "step": 30456 }, { "epoch": 22.31282051282051, "grad_norm": 1.25, "learning_rate": 3.250845545449778e-05, "loss": 0.1153, "step": 30457 }, { "epoch": 22.313553113553112, "grad_norm": 0.8671875, "learning_rate": 3.250526347318536e-05, "loss": 0.1028, "step": 30458 }, { "epoch": 22.314285714285713, "grad_norm": 1.296875, "learning_rate": 3.250207157809163e-05, "loss": 0.1493, "step": 30459 }, { "epoch": 22.315018315018314, "grad_norm": 0.98828125, "learning_rate": 3.2498879769231884e-05, "loss": 0.128, "step": 30460 }, { "epoch": 22.315750915750915, "grad_norm": 1.078125, "learning_rate": 3.2495688046621413e-05, "loss": 0.1101, "step": 30461 }, { "epoch": 22.316483516483515, "grad_norm": 1.109375, "learning_rate": 3.2492496410275516e-05, "loss": 0.1399, "step": 30462 }, { "epoch": 22.317216117216116, "grad_norm": 0.94921875, "learning_rate": 3.248930486020946e-05, "loss": 0.119, "step": 30463 }, { "epoch": 22.317948717948717, "grad_norm": 0.71484375, "learning_rate": 3.2486113396438575e-05, "loss": 0.0973, "step": 30464 }, { "epoch": 22.318681318681318, "grad_norm": 0.96875, "learning_rate": 3.2482922018978126e-05, "loss": 0.1078, "step": 30465 }, { "epoch": 22.31941391941392, "grad_norm": 1.015625, "learning_rate": 3.2479730727843427e-05, "loss": 0.1432, "step": 30466 }, { "epoch": 22.32014652014652, "grad_norm": 1.4609375, "learning_rate": 3.247653952304975e-05, "loss": 0.1208, "step": 30467 }, { "epoch": 22.32087912087912, "grad_norm": 1.0703125, "learning_rate": 3.2473348404612356e-05, "loss": 0.1182, "step": 30468 }, { "epoch": 22.32161172161172, "grad_norm": 1.015625, "learning_rate": 3.24701573725466e-05, "loss": 0.1183, "step": 30469 }, { "epoch": 22.32234432234432, "grad_norm": 1.1484375, "learning_rate": 3.246696642686774e-05, "loss": 0.1034, "step": 30470 }, { "epoch": 22.323076923076922, "grad_norm": 1.25, "learning_rate": 3.246377556759105e-05, "loss": 0.1452, "step": 30471 }, { "epoch": 22.323809523809523, "grad_norm": 0.63671875, "learning_rate": 3.246058479473184e-05, "loss": 0.1208, "step": 30472 }, { "epoch": 22.324542124542123, "grad_norm": 0.8125, "learning_rate": 3.2457394108305375e-05, "loss": 0.102, "step": 30473 }, { "epoch": 22.325274725274724, "grad_norm": 1.15625, "learning_rate": 3.2454203508326976e-05, "loss": 0.1155, "step": 30474 }, { "epoch": 22.326007326007325, "grad_norm": 1.0546875, "learning_rate": 3.245101299481191e-05, "loss": 0.1309, "step": 30475 }, { "epoch": 22.326739926739926, "grad_norm": 1.6875, "learning_rate": 3.244782256777547e-05, "loss": 0.135, "step": 30476 }, { "epoch": 22.327472527472526, "grad_norm": 0.73046875, "learning_rate": 3.244463222723293e-05, "loss": 0.1355, "step": 30477 }, { "epoch": 22.328205128205127, "grad_norm": 0.8828125, "learning_rate": 3.244144197319957e-05, "loss": 0.1179, "step": 30478 }, { "epoch": 22.328937728937728, "grad_norm": 0.70703125, "learning_rate": 3.243825180569071e-05, "loss": 0.1078, "step": 30479 }, { "epoch": 22.32967032967033, "grad_norm": 1.65625, "learning_rate": 3.243506172472162e-05, "loss": 0.1499, "step": 30480 }, { "epoch": 22.33040293040293, "grad_norm": 1.0703125, "learning_rate": 3.243187173030757e-05, "loss": 0.1082, "step": 30481 }, { "epoch": 22.33113553113553, "grad_norm": 1.640625, "learning_rate": 3.242868182246385e-05, "loss": 0.1485, "step": 30482 }, { "epoch": 22.33186813186813, "grad_norm": 1.34375, "learning_rate": 3.242549200120574e-05, "loss": 0.1376, "step": 30483 }, { "epoch": 22.33260073260073, "grad_norm": 1.1796875, "learning_rate": 3.242230226654855e-05, "loss": 0.1174, "step": 30484 }, { "epoch": 22.333333333333332, "grad_norm": 0.66796875, "learning_rate": 3.241911261850753e-05, "loss": 0.0932, "step": 30485 }, { "epoch": 22.334065934065933, "grad_norm": 1.34375, "learning_rate": 3.241592305709799e-05, "loss": 0.1498, "step": 30486 }, { "epoch": 22.334798534798534, "grad_norm": 0.9921875, "learning_rate": 3.24127335823352e-05, "loss": 0.1281, "step": 30487 }, { "epoch": 22.335531135531134, "grad_norm": 1.46875, "learning_rate": 3.2409544194234435e-05, "loss": 0.1168, "step": 30488 }, { "epoch": 22.336263736263735, "grad_norm": 1.4453125, "learning_rate": 3.240635489281097e-05, "loss": 0.1242, "step": 30489 }, { "epoch": 22.336996336996336, "grad_norm": 1.25, "learning_rate": 3.240316567808011e-05, "loss": 0.1085, "step": 30490 }, { "epoch": 22.337728937728937, "grad_norm": 0.87890625, "learning_rate": 3.2399976550057134e-05, "loss": 0.1086, "step": 30491 }, { "epoch": 22.338461538461537, "grad_norm": 1.4609375, "learning_rate": 3.23967875087573e-05, "loss": 0.1257, "step": 30492 }, { "epoch": 22.339194139194138, "grad_norm": 0.88671875, "learning_rate": 3.2393598554195906e-05, "loss": 0.1098, "step": 30493 }, { "epoch": 22.33992673992674, "grad_norm": 1.0, "learning_rate": 3.2390409686388214e-05, "loss": 0.121, "step": 30494 }, { "epoch": 22.34065934065934, "grad_norm": 1.2734375, "learning_rate": 3.238722090534953e-05, "loss": 0.1047, "step": 30495 }, { "epoch": 22.34139194139194, "grad_norm": 0.71875, "learning_rate": 3.2384032211095106e-05, "loss": 0.1209, "step": 30496 }, { "epoch": 22.34212454212454, "grad_norm": 0.703125, "learning_rate": 3.238084360364022e-05, "loss": 0.1054, "step": 30497 }, { "epoch": 22.34285714285714, "grad_norm": 1.171875, "learning_rate": 3.2377655083000186e-05, "loss": 0.1136, "step": 30498 }, { "epoch": 22.343589743589742, "grad_norm": 0.87109375, "learning_rate": 3.237446664919023e-05, "loss": 0.1204, "step": 30499 }, { "epoch": 22.344322344322343, "grad_norm": 1.078125, "learning_rate": 3.2371278302225665e-05, "loss": 0.142, "step": 30500 }, { "epoch": 22.345054945054944, "grad_norm": 0.77734375, "learning_rate": 3.236809004212177e-05, "loss": 0.1317, "step": 30501 }, { "epoch": 22.345787545787545, "grad_norm": 0.95703125, "learning_rate": 3.23649018688938e-05, "loss": 0.1316, "step": 30502 }, { "epoch": 22.346520146520145, "grad_norm": 1.09375, "learning_rate": 3.236171378255704e-05, "loss": 0.1245, "step": 30503 }, { "epoch": 22.347252747252746, "grad_norm": 0.9765625, "learning_rate": 3.2358525783126744e-05, "loss": 0.1043, "step": 30504 }, { "epoch": 22.347985347985347, "grad_norm": 0.96484375, "learning_rate": 3.235533787061823e-05, "loss": 0.1158, "step": 30505 }, { "epoch": 22.348717948717947, "grad_norm": 1.0078125, "learning_rate": 3.2352150045046745e-05, "loss": 0.1211, "step": 30506 }, { "epoch": 22.349450549450548, "grad_norm": 0.68359375, "learning_rate": 3.2348962306427566e-05, "loss": 0.1152, "step": 30507 }, { "epoch": 22.35018315018315, "grad_norm": 1.3984375, "learning_rate": 3.234577465477596e-05, "loss": 0.1123, "step": 30508 }, { "epoch": 22.35091575091575, "grad_norm": 0.93359375, "learning_rate": 3.234258709010721e-05, "loss": 0.1258, "step": 30509 }, { "epoch": 22.35164835164835, "grad_norm": 0.63671875, "learning_rate": 3.233939961243659e-05, "loss": 0.1203, "step": 30510 }, { "epoch": 22.35238095238095, "grad_norm": 0.8671875, "learning_rate": 3.233621222177936e-05, "loss": 0.1007, "step": 30511 }, { "epoch": 22.353113553113552, "grad_norm": 1.3671875, "learning_rate": 3.23330249181508e-05, "loss": 0.1138, "step": 30512 }, { "epoch": 22.353846153846153, "grad_norm": 1.0546875, "learning_rate": 3.23298377015662e-05, "loss": 0.1198, "step": 30513 }, { "epoch": 22.354578754578753, "grad_norm": 1.125, "learning_rate": 3.23266505720408e-05, "loss": 0.1176, "step": 30514 }, { "epoch": 22.355311355311354, "grad_norm": 1.0390625, "learning_rate": 3.232346352958986e-05, "loss": 0.1239, "step": 30515 }, { "epoch": 22.356043956043955, "grad_norm": 1.1484375, "learning_rate": 3.2320276574228704e-05, "loss": 0.0948, "step": 30516 }, { "epoch": 22.356776556776556, "grad_norm": 0.69140625, "learning_rate": 3.2317089705972565e-05, "loss": 0.1038, "step": 30517 }, { "epoch": 22.357509157509156, "grad_norm": 0.8203125, "learning_rate": 3.231390292483672e-05, "loss": 0.104, "step": 30518 }, { "epoch": 22.358241758241757, "grad_norm": 1.296875, "learning_rate": 3.231071623083644e-05, "loss": 0.1266, "step": 30519 }, { "epoch": 22.358974358974358, "grad_norm": 1.0, "learning_rate": 3.230752962398697e-05, "loss": 0.1313, "step": 30520 }, { "epoch": 22.35970695970696, "grad_norm": 1.078125, "learning_rate": 3.2304343104303616e-05, "loss": 0.1297, "step": 30521 }, { "epoch": 22.36043956043956, "grad_norm": 1.203125, "learning_rate": 3.2301156671801605e-05, "loss": 0.1279, "step": 30522 }, { "epoch": 22.36117216117216, "grad_norm": 1.203125, "learning_rate": 3.229797032649625e-05, "loss": 0.1787, "step": 30523 }, { "epoch": 22.36190476190476, "grad_norm": 1.109375, "learning_rate": 3.229478406840279e-05, "loss": 0.1216, "step": 30524 }, { "epoch": 22.36263736263736, "grad_norm": 1.6796875, "learning_rate": 3.229159789753647e-05, "loss": 0.1423, "step": 30525 }, { "epoch": 22.363369963369962, "grad_norm": 0.8046875, "learning_rate": 3.2288411813912615e-05, "loss": 0.1055, "step": 30526 }, { "epoch": 22.364102564102563, "grad_norm": 1.03125, "learning_rate": 3.228522581754645e-05, "loss": 0.1036, "step": 30527 }, { "epoch": 22.364835164835164, "grad_norm": 0.875, "learning_rate": 3.228203990845325e-05, "loss": 0.1278, "step": 30528 }, { "epoch": 22.365567765567764, "grad_norm": 1.5625, "learning_rate": 3.227885408664828e-05, "loss": 0.1312, "step": 30529 }, { "epoch": 22.366300366300365, "grad_norm": 1.0703125, "learning_rate": 3.227566835214678e-05, "loss": 0.158, "step": 30530 }, { "epoch": 22.367032967032966, "grad_norm": 2.203125, "learning_rate": 3.227248270496405e-05, "loss": 0.152, "step": 30531 }, { "epoch": 22.367765567765566, "grad_norm": 0.828125, "learning_rate": 3.226929714511534e-05, "loss": 0.1144, "step": 30532 }, { "epoch": 22.368498168498167, "grad_norm": 0.84375, "learning_rate": 3.226611167261592e-05, "loss": 0.1086, "step": 30533 }, { "epoch": 22.369230769230768, "grad_norm": 1.390625, "learning_rate": 3.226292628748103e-05, "loss": 0.1086, "step": 30534 }, { "epoch": 22.36996336996337, "grad_norm": 0.78515625, "learning_rate": 3.2259740989725956e-05, "loss": 0.1229, "step": 30535 }, { "epoch": 22.37069597069597, "grad_norm": 0.87109375, "learning_rate": 3.225655577936594e-05, "loss": 0.1111, "step": 30536 }, { "epoch": 22.37142857142857, "grad_norm": 0.96875, "learning_rate": 3.225337065641626e-05, "loss": 0.1278, "step": 30537 }, { "epoch": 22.37216117216117, "grad_norm": 0.91796875, "learning_rate": 3.225018562089217e-05, "loss": 0.1302, "step": 30538 }, { "epoch": 22.37289377289377, "grad_norm": 0.8671875, "learning_rate": 3.224700067280894e-05, "loss": 0.1044, "step": 30539 }, { "epoch": 22.373626373626372, "grad_norm": 1.046875, "learning_rate": 3.224381581218182e-05, "loss": 0.1319, "step": 30540 }, { "epoch": 22.374358974358973, "grad_norm": 0.796875, "learning_rate": 3.2240631039026056e-05, "loss": 0.0981, "step": 30541 }, { "epoch": 22.375091575091574, "grad_norm": 1.203125, "learning_rate": 3.2237446353356934e-05, "loss": 0.1166, "step": 30542 }, { "epoch": 22.375824175824174, "grad_norm": 1.2421875, "learning_rate": 3.223426175518972e-05, "loss": 0.0956, "step": 30543 }, { "epoch": 22.376556776556775, "grad_norm": 1.359375, "learning_rate": 3.223107724453963e-05, "loss": 0.1322, "step": 30544 }, { "epoch": 22.377289377289376, "grad_norm": 0.69140625, "learning_rate": 3.222789282142196e-05, "loss": 0.1239, "step": 30545 }, { "epoch": 22.378021978021977, "grad_norm": 1.21875, "learning_rate": 3.222470848585194e-05, "loss": 0.1473, "step": 30546 }, { "epoch": 22.378754578754577, "grad_norm": 0.61328125, "learning_rate": 3.222152423784485e-05, "loss": 0.1183, "step": 30547 }, { "epoch": 22.379487179487178, "grad_norm": 1.1875, "learning_rate": 3.2218340077415924e-05, "loss": 0.1143, "step": 30548 }, { "epoch": 22.38021978021978, "grad_norm": 1.1796875, "learning_rate": 3.2215156004580445e-05, "loss": 0.1327, "step": 30549 }, { "epoch": 22.38095238095238, "grad_norm": 0.7890625, "learning_rate": 3.2211972019353665e-05, "loss": 0.1124, "step": 30550 }, { "epoch": 22.38168498168498, "grad_norm": 1.046875, "learning_rate": 3.2208788121750804e-05, "loss": 0.1294, "step": 30551 }, { "epoch": 22.38241758241758, "grad_norm": 0.82421875, "learning_rate": 3.220560431178716e-05, "loss": 0.1459, "step": 30552 }, { "epoch": 22.383150183150182, "grad_norm": 1.0546875, "learning_rate": 3.220242058947798e-05, "loss": 0.1255, "step": 30553 }, { "epoch": 22.383882783882783, "grad_norm": 1.46875, "learning_rate": 3.219923695483851e-05, "loss": 0.1165, "step": 30554 }, { "epoch": 22.384615384615383, "grad_norm": 0.75390625, "learning_rate": 3.219605340788399e-05, "loss": 0.1118, "step": 30555 }, { "epoch": 22.385347985347984, "grad_norm": 1.5390625, "learning_rate": 3.219286994862968e-05, "loss": 0.1263, "step": 30556 }, { "epoch": 22.386080586080585, "grad_norm": 0.8828125, "learning_rate": 3.2189686577090855e-05, "loss": 0.1255, "step": 30557 }, { "epoch": 22.386813186813185, "grad_norm": 1.234375, "learning_rate": 3.2186503293282755e-05, "loss": 0.1371, "step": 30558 }, { "epoch": 22.387545787545786, "grad_norm": 1.7265625, "learning_rate": 3.218332009722063e-05, "loss": 0.141, "step": 30559 }, { "epoch": 22.388278388278387, "grad_norm": 0.86328125, "learning_rate": 3.218013698891971e-05, "loss": 0.1117, "step": 30560 }, { "epoch": 22.389010989010988, "grad_norm": 1.0703125, "learning_rate": 3.217695396839529e-05, "loss": 0.1133, "step": 30561 }, { "epoch": 22.38974358974359, "grad_norm": 0.89453125, "learning_rate": 3.217377103566259e-05, "loss": 0.124, "step": 30562 }, { "epoch": 22.39047619047619, "grad_norm": 0.94140625, "learning_rate": 3.217058819073686e-05, "loss": 0.1225, "step": 30563 }, { "epoch": 22.39120879120879, "grad_norm": 1.109375, "learning_rate": 3.2167405433633374e-05, "loss": 0.1305, "step": 30564 }, { "epoch": 22.39194139194139, "grad_norm": 0.98046875, "learning_rate": 3.216422276436737e-05, "loss": 0.107, "step": 30565 }, { "epoch": 22.39267399267399, "grad_norm": 1.0234375, "learning_rate": 3.216104018295409e-05, "loss": 0.1117, "step": 30566 }, { "epoch": 22.393406593406592, "grad_norm": 0.8671875, "learning_rate": 3.2157857689408764e-05, "loss": 0.1275, "step": 30567 }, { "epoch": 22.394139194139193, "grad_norm": 1.0703125, "learning_rate": 3.215467528374669e-05, "loss": 0.1123, "step": 30568 }, { "epoch": 22.394871794871793, "grad_norm": 1.0078125, "learning_rate": 3.215149296598309e-05, "loss": 0.1442, "step": 30569 }, { "epoch": 22.395604395604394, "grad_norm": 0.90234375, "learning_rate": 3.21483107361332e-05, "loss": 0.1419, "step": 30570 }, { "epoch": 22.396336996336995, "grad_norm": 1.25, "learning_rate": 3.214512859421227e-05, "loss": 0.1388, "step": 30571 }, { "epoch": 22.397069597069596, "grad_norm": 0.8125, "learning_rate": 3.2141946540235564e-05, "loss": 0.0884, "step": 30572 }, { "epoch": 22.397802197802196, "grad_norm": 1.1171875, "learning_rate": 3.213876457421832e-05, "loss": 0.1085, "step": 30573 }, { "epoch": 22.398534798534797, "grad_norm": 0.70703125, "learning_rate": 3.213558269617578e-05, "loss": 0.1111, "step": 30574 }, { "epoch": 22.399267399267398, "grad_norm": 1.234375, "learning_rate": 3.213240090612319e-05, "loss": 0.1242, "step": 30575 }, { "epoch": 22.4, "grad_norm": 1.2265625, "learning_rate": 3.212921920407581e-05, "loss": 0.1126, "step": 30576 }, { "epoch": 22.4007326007326, "grad_norm": 1.0078125, "learning_rate": 3.212603759004884e-05, "loss": 0.1307, "step": 30577 }, { "epoch": 22.4014652014652, "grad_norm": 0.6171875, "learning_rate": 3.212285606405758e-05, "loss": 0.1081, "step": 30578 }, { "epoch": 22.4021978021978, "grad_norm": 1.25, "learning_rate": 3.211967462611726e-05, "loss": 0.1198, "step": 30579 }, { "epoch": 22.4029304029304, "grad_norm": 0.83984375, "learning_rate": 3.2116493276243096e-05, "loss": 0.1471, "step": 30580 }, { "epoch": 22.403663003663002, "grad_norm": 1.25, "learning_rate": 3.2113312014450364e-05, "loss": 0.1278, "step": 30581 }, { "epoch": 22.404395604395603, "grad_norm": 1.0859375, "learning_rate": 3.2110130840754256e-05, "loss": 0.1079, "step": 30582 }, { "epoch": 22.405128205128204, "grad_norm": 1.375, "learning_rate": 3.210694975517007e-05, "loss": 0.119, "step": 30583 }, { "epoch": 22.405860805860804, "grad_norm": 1.0, "learning_rate": 3.210376875771303e-05, "loss": 0.1112, "step": 30584 }, { "epoch": 22.406593406593405, "grad_norm": 1.1796875, "learning_rate": 3.210058784839838e-05, "loss": 0.1665, "step": 30585 }, { "epoch": 22.407326007326006, "grad_norm": 0.828125, "learning_rate": 3.209740702724133e-05, "loss": 0.1072, "step": 30586 }, { "epoch": 22.408058608058607, "grad_norm": 0.81640625, "learning_rate": 3.2094226294257165e-05, "loss": 0.1155, "step": 30587 }, { "epoch": 22.408791208791207, "grad_norm": 0.828125, "learning_rate": 3.2091045649461074e-05, "loss": 0.1102, "step": 30588 }, { "epoch": 22.409523809523808, "grad_norm": 1.1171875, "learning_rate": 3.208786509286836e-05, "loss": 0.107, "step": 30589 }, { "epoch": 22.41025641025641, "grad_norm": 1.2265625, "learning_rate": 3.2084684624494215e-05, "loss": 0.0951, "step": 30590 }, { "epoch": 22.41098901098901, "grad_norm": 1.40625, "learning_rate": 3.208150424435389e-05, "loss": 0.1184, "step": 30591 }, { "epoch": 22.41172161172161, "grad_norm": 0.8984375, "learning_rate": 3.2078323952462634e-05, "loss": 0.1287, "step": 30592 }, { "epoch": 22.41245421245421, "grad_norm": 1.109375, "learning_rate": 3.207514374883565e-05, "loss": 0.111, "step": 30593 }, { "epoch": 22.41318681318681, "grad_norm": 1.0546875, "learning_rate": 3.207196363348823e-05, "loss": 0.1244, "step": 30594 }, { "epoch": 22.413919413919412, "grad_norm": 0.91796875, "learning_rate": 3.206878360643558e-05, "loss": 0.1327, "step": 30595 }, { "epoch": 22.414652014652013, "grad_norm": 1.1796875, "learning_rate": 3.206560366769292e-05, "loss": 0.1499, "step": 30596 }, { "epoch": 22.415384615384614, "grad_norm": 0.640625, "learning_rate": 3.206242381727552e-05, "loss": 0.0939, "step": 30597 }, { "epoch": 22.416117216117215, "grad_norm": 1.2734375, "learning_rate": 3.205924405519858e-05, "loss": 0.1628, "step": 30598 }, { "epoch": 22.416849816849815, "grad_norm": 0.84765625, "learning_rate": 3.2056064381477376e-05, "loss": 0.1404, "step": 30599 }, { "epoch": 22.417582417582416, "grad_norm": 1.0625, "learning_rate": 3.205288479612711e-05, "loss": 0.0948, "step": 30600 }, { "epoch": 22.418315018315017, "grad_norm": 0.80859375, "learning_rate": 3.204970529916304e-05, "loss": 0.109, "step": 30601 }, { "epoch": 22.419047619047618, "grad_norm": 0.82421875, "learning_rate": 3.2046525890600386e-05, "loss": 0.1225, "step": 30602 }, { "epoch": 22.41978021978022, "grad_norm": 1.140625, "learning_rate": 3.204334657045437e-05, "loss": 0.1319, "step": 30603 }, { "epoch": 22.42051282051282, "grad_norm": 1.0625, "learning_rate": 3.2040167338740254e-05, "loss": 0.1054, "step": 30604 }, { "epoch": 22.42124542124542, "grad_norm": 1.375, "learning_rate": 3.203698819547326e-05, "loss": 0.1554, "step": 30605 }, { "epoch": 22.42197802197802, "grad_norm": 1.1484375, "learning_rate": 3.2033809140668614e-05, "loss": 0.137, "step": 30606 }, { "epoch": 22.42271062271062, "grad_norm": 1.234375, "learning_rate": 3.203063017434155e-05, "loss": 0.1224, "step": 30607 }, { "epoch": 22.423443223443222, "grad_norm": 0.8203125, "learning_rate": 3.2027451296507275e-05, "loss": 0.1278, "step": 30608 }, { "epoch": 22.424175824175823, "grad_norm": 0.91796875, "learning_rate": 3.202427250718108e-05, "loss": 0.1135, "step": 30609 }, { "epoch": 22.424908424908423, "grad_norm": 1.2890625, "learning_rate": 3.202109380637814e-05, "loss": 0.1234, "step": 30610 }, { "epoch": 22.425641025641024, "grad_norm": 0.8203125, "learning_rate": 3.2017915194113726e-05, "loss": 0.1191, "step": 30611 }, { "epoch": 22.426373626373625, "grad_norm": 1.5546875, "learning_rate": 3.2014736670403026e-05, "loss": 0.0872, "step": 30612 }, { "epoch": 22.427106227106226, "grad_norm": 1.140625, "learning_rate": 3.20115582352613e-05, "loss": 0.1151, "step": 30613 }, { "epoch": 22.427838827838826, "grad_norm": 1.21875, "learning_rate": 3.200837988870375e-05, "loss": 0.1313, "step": 30614 }, { "epoch": 22.428571428571427, "grad_norm": 1.4921875, "learning_rate": 3.2005201630745645e-05, "loss": 0.1054, "step": 30615 }, { "epoch": 22.429304029304028, "grad_norm": 1.203125, "learning_rate": 3.200202346140218e-05, "loss": 0.1282, "step": 30616 }, { "epoch": 22.43003663003663, "grad_norm": 1.0390625, "learning_rate": 3.199884538068862e-05, "loss": 0.1128, "step": 30617 }, { "epoch": 22.43076923076923, "grad_norm": 0.890625, "learning_rate": 3.199566738862014e-05, "loss": 0.0973, "step": 30618 }, { "epoch": 22.43150183150183, "grad_norm": 0.9375, "learning_rate": 3.1992489485211974e-05, "loss": 0.1413, "step": 30619 }, { "epoch": 22.43223443223443, "grad_norm": 1.046875, "learning_rate": 3.198931167047939e-05, "loss": 0.1125, "step": 30620 }, { "epoch": 22.43296703296703, "grad_norm": 0.7734375, "learning_rate": 3.1986133944437595e-05, "loss": 0.107, "step": 30621 }, { "epoch": 22.433699633699632, "grad_norm": 1.28125, "learning_rate": 3.1982956307101805e-05, "loss": 0.1172, "step": 30622 }, { "epoch": 22.434432234432233, "grad_norm": 1.2265625, "learning_rate": 3.197977875848725e-05, "loss": 0.1523, "step": 30623 }, { "epoch": 22.435164835164834, "grad_norm": 1.0234375, "learning_rate": 3.1976601298609144e-05, "loss": 0.1023, "step": 30624 }, { "epoch": 22.435897435897434, "grad_norm": 1.5078125, "learning_rate": 3.197342392748273e-05, "loss": 0.1016, "step": 30625 }, { "epoch": 22.436630036630035, "grad_norm": 0.8515625, "learning_rate": 3.1970246645123224e-05, "loss": 0.1086, "step": 30626 }, { "epoch": 22.437362637362636, "grad_norm": 0.79296875, "learning_rate": 3.1967069451545854e-05, "loss": 0.1091, "step": 30627 }, { "epoch": 22.438095238095237, "grad_norm": 0.76953125, "learning_rate": 3.196389234676584e-05, "loss": 0.1181, "step": 30628 }, { "epoch": 22.438827838827837, "grad_norm": 0.97265625, "learning_rate": 3.196071533079838e-05, "loss": 0.1217, "step": 30629 }, { "epoch": 22.439560439560438, "grad_norm": 0.5546875, "learning_rate": 3.1957538403658746e-05, "loss": 0.0992, "step": 30630 }, { "epoch": 22.44029304029304, "grad_norm": 0.6484375, "learning_rate": 3.195436156536213e-05, "loss": 0.1082, "step": 30631 }, { "epoch": 22.44102564102564, "grad_norm": 0.5625, "learning_rate": 3.195118481592375e-05, "loss": 0.1094, "step": 30632 }, { "epoch": 22.44175824175824, "grad_norm": 0.734375, "learning_rate": 3.194800815535883e-05, "loss": 0.1372, "step": 30633 }, { "epoch": 22.44249084249084, "grad_norm": 0.66796875, "learning_rate": 3.194483158368258e-05, "loss": 0.1291, "step": 30634 }, { "epoch": 22.44322344322344, "grad_norm": 0.95703125, "learning_rate": 3.194165510091024e-05, "loss": 0.1422, "step": 30635 }, { "epoch": 22.443956043956042, "grad_norm": 0.828125, "learning_rate": 3.193847870705705e-05, "loss": 0.103, "step": 30636 }, { "epoch": 22.444688644688643, "grad_norm": 1.109375, "learning_rate": 3.193530240213817e-05, "loss": 0.1253, "step": 30637 }, { "epoch": 22.445421245421244, "grad_norm": 0.921875, "learning_rate": 3.1932126186168867e-05, "loss": 0.1009, "step": 30638 }, { "epoch": 22.446153846153845, "grad_norm": 0.921875, "learning_rate": 3.1928950059164345e-05, "loss": 0.1009, "step": 30639 }, { "epoch": 22.446886446886445, "grad_norm": 0.703125, "learning_rate": 3.1925774021139795e-05, "loss": 0.1376, "step": 30640 }, { "epoch": 22.447619047619046, "grad_norm": 0.74609375, "learning_rate": 3.192259807211048e-05, "loss": 0.1123, "step": 30641 }, { "epoch": 22.448351648351647, "grad_norm": 1.4921875, "learning_rate": 3.191942221209161e-05, "loss": 0.1127, "step": 30642 }, { "epoch": 22.449084249084247, "grad_norm": 0.9921875, "learning_rate": 3.191624644109837e-05, "loss": 0.1412, "step": 30643 }, { "epoch": 22.449816849816848, "grad_norm": 1.1015625, "learning_rate": 3.1913070759146e-05, "loss": 0.1239, "step": 30644 }, { "epoch": 22.45054945054945, "grad_norm": 0.95703125, "learning_rate": 3.190989516624969e-05, "loss": 0.1105, "step": 30645 }, { "epoch": 22.45128205128205, "grad_norm": 0.78515625, "learning_rate": 3.190671966242469e-05, "loss": 0.1129, "step": 30646 }, { "epoch": 22.45201465201465, "grad_norm": 0.92578125, "learning_rate": 3.190354424768621e-05, "loss": 0.1124, "step": 30647 }, { "epoch": 22.45274725274725, "grad_norm": 1.4140625, "learning_rate": 3.190036892204944e-05, "loss": 0.1208, "step": 30648 }, { "epoch": 22.453479853479852, "grad_norm": 0.69140625, "learning_rate": 3.1897193685529607e-05, "loss": 0.1495, "step": 30649 }, { "epoch": 22.454212454212453, "grad_norm": 1.0, "learning_rate": 3.189401853814193e-05, "loss": 0.0998, "step": 30650 }, { "epoch": 22.454945054945053, "grad_norm": 1.1171875, "learning_rate": 3.189084347990162e-05, "loss": 0.1145, "step": 30651 }, { "epoch": 22.455677655677654, "grad_norm": 0.91015625, "learning_rate": 3.188766851082388e-05, "loss": 0.1273, "step": 30652 }, { "epoch": 22.456410256410255, "grad_norm": 1.2421875, "learning_rate": 3.1884493630923946e-05, "loss": 0.1468, "step": 30653 }, { "epoch": 22.457142857142856, "grad_norm": 0.84765625, "learning_rate": 3.1881318840217014e-05, "loss": 0.1048, "step": 30654 }, { "epoch": 22.457875457875456, "grad_norm": 1.2890625, "learning_rate": 3.187814413871826e-05, "loss": 0.153, "step": 30655 }, { "epoch": 22.458608058608057, "grad_norm": 1.2734375, "learning_rate": 3.187496952644297e-05, "loss": 0.166, "step": 30656 }, { "epoch": 22.459340659340658, "grad_norm": 0.94921875, "learning_rate": 3.187179500340631e-05, "loss": 0.1125, "step": 30657 }, { "epoch": 22.46007326007326, "grad_norm": 0.89453125, "learning_rate": 3.186862056962349e-05, "loss": 0.1273, "step": 30658 }, { "epoch": 22.46080586080586, "grad_norm": 0.96875, "learning_rate": 3.1865446225109727e-05, "loss": 0.1259, "step": 30659 }, { "epoch": 22.46153846153846, "grad_norm": 0.9375, "learning_rate": 3.186227196988021e-05, "loss": 0.091, "step": 30660 }, { "epoch": 22.46227106227106, "grad_norm": 1.171875, "learning_rate": 3.185909780395019e-05, "loss": 0.1191, "step": 30661 }, { "epoch": 22.46300366300366, "grad_norm": 0.87109375, "learning_rate": 3.185592372733485e-05, "loss": 0.1325, "step": 30662 }, { "epoch": 22.463736263736262, "grad_norm": 1.0, "learning_rate": 3.185274974004939e-05, "loss": 0.0983, "step": 30663 }, { "epoch": 22.464468864468863, "grad_norm": 0.9765625, "learning_rate": 3.184957584210902e-05, "loss": 0.129, "step": 30664 }, { "epoch": 22.465201465201464, "grad_norm": 0.99609375, "learning_rate": 3.184640203352897e-05, "loss": 0.1137, "step": 30665 }, { "epoch": 22.465934065934064, "grad_norm": 1.765625, "learning_rate": 3.1843228314324416e-05, "loss": 0.1259, "step": 30666 }, { "epoch": 22.466666666666665, "grad_norm": 0.796875, "learning_rate": 3.1840054684510596e-05, "loss": 0.1142, "step": 30667 }, { "epoch": 22.467399267399266, "grad_norm": 0.68359375, "learning_rate": 3.18368811441027e-05, "loss": 0.1134, "step": 30668 }, { "epoch": 22.468131868131866, "grad_norm": 1.125, "learning_rate": 3.183370769311593e-05, "loss": 0.1492, "step": 30669 }, { "epoch": 22.468864468864467, "grad_norm": 0.9765625, "learning_rate": 3.183053433156549e-05, "loss": 0.1116, "step": 30670 }, { "epoch": 22.469597069597068, "grad_norm": 1.1328125, "learning_rate": 3.1827361059466585e-05, "loss": 0.1347, "step": 30671 }, { "epoch": 22.47032967032967, "grad_norm": 0.91796875, "learning_rate": 3.182418787683443e-05, "loss": 0.1202, "step": 30672 }, { "epoch": 22.47106227106227, "grad_norm": 0.68359375, "learning_rate": 3.1821014783684225e-05, "loss": 0.1114, "step": 30673 }, { "epoch": 22.47179487179487, "grad_norm": 1.1171875, "learning_rate": 3.181784178003117e-05, "loss": 0.1112, "step": 30674 }, { "epoch": 22.47252747252747, "grad_norm": 1.5078125, "learning_rate": 3.181466886589045e-05, "loss": 0.1325, "step": 30675 }, { "epoch": 22.47326007326007, "grad_norm": 0.94921875, "learning_rate": 3.18114960412773e-05, "loss": 0.1045, "step": 30676 }, { "epoch": 22.473992673992672, "grad_norm": 0.921875, "learning_rate": 3.1808323306206904e-05, "loss": 0.1202, "step": 30677 }, { "epoch": 22.474725274725273, "grad_norm": 0.6640625, "learning_rate": 3.180515066069447e-05, "loss": 0.134, "step": 30678 }, { "epoch": 22.475457875457874, "grad_norm": 1.5703125, "learning_rate": 3.18019781047552e-05, "loss": 0.1226, "step": 30679 }, { "epoch": 22.476190476190474, "grad_norm": 0.86328125, "learning_rate": 3.1798805638404285e-05, "loss": 0.1227, "step": 30680 }, { "epoch": 22.476923076923075, "grad_norm": 1.140625, "learning_rate": 3.1795633261656915e-05, "loss": 0.1379, "step": 30681 }, { "epoch": 22.477655677655676, "grad_norm": 0.734375, "learning_rate": 3.1792460974528334e-05, "loss": 0.1156, "step": 30682 }, { "epoch": 22.478388278388277, "grad_norm": 0.9296875, "learning_rate": 3.17892887770337e-05, "loss": 0.1293, "step": 30683 }, { "epoch": 22.479120879120877, "grad_norm": 0.6328125, "learning_rate": 3.178611666918824e-05, "loss": 0.0901, "step": 30684 }, { "epoch": 22.479853479853478, "grad_norm": 0.64453125, "learning_rate": 3.178294465100713e-05, "loss": 0.1147, "step": 30685 }, { "epoch": 22.48058608058608, "grad_norm": 1.1015625, "learning_rate": 3.1779772722505565e-05, "loss": 0.1384, "step": 30686 }, { "epoch": 22.48131868131868, "grad_norm": 0.51171875, "learning_rate": 3.1776600883698764e-05, "loss": 0.1074, "step": 30687 }, { "epoch": 22.48205128205128, "grad_norm": 1.046875, "learning_rate": 3.177342913460192e-05, "loss": 0.1347, "step": 30688 }, { "epoch": 22.48278388278388, "grad_norm": 1.34375, "learning_rate": 3.177025747523022e-05, "loss": 0.1424, "step": 30689 }, { "epoch": 22.483516483516482, "grad_norm": 1.421875, "learning_rate": 3.176708590559886e-05, "loss": 0.1249, "step": 30690 }, { "epoch": 22.484249084249083, "grad_norm": 0.76953125, "learning_rate": 3.176391442572305e-05, "loss": 0.1317, "step": 30691 }, { "epoch": 22.484981684981683, "grad_norm": 1.015625, "learning_rate": 3.176074303561796e-05, "loss": 0.1369, "step": 30692 }, { "epoch": 22.485714285714284, "grad_norm": 0.76953125, "learning_rate": 3.1757571735298815e-05, "loss": 0.1041, "step": 30693 }, { "epoch": 22.486446886446885, "grad_norm": 1.078125, "learning_rate": 3.1754400524780796e-05, "loss": 0.1218, "step": 30694 }, { "epoch": 22.487179487179485, "grad_norm": 0.76953125, "learning_rate": 3.175122940407911e-05, "loss": 0.1234, "step": 30695 }, { "epoch": 22.487912087912086, "grad_norm": 0.79296875, "learning_rate": 3.1748058373208924e-05, "loss": 0.1224, "step": 30696 }, { "epoch": 22.488644688644687, "grad_norm": 1.5703125, "learning_rate": 3.1744887432185435e-05, "loss": 0.1331, "step": 30697 }, { "epoch": 22.489377289377288, "grad_norm": 0.84375, "learning_rate": 3.174171658102386e-05, "loss": 0.1317, "step": 30698 }, { "epoch": 22.49010989010989, "grad_norm": 1.21875, "learning_rate": 3.173854581973938e-05, "loss": 0.1136, "step": 30699 }, { "epoch": 22.49084249084249, "grad_norm": 0.921875, "learning_rate": 3.1735375148347195e-05, "loss": 0.1056, "step": 30700 }, { "epoch": 22.49157509157509, "grad_norm": 1.2421875, "learning_rate": 3.173220456686248e-05, "loss": 0.1359, "step": 30701 }, { "epoch": 22.49230769230769, "grad_norm": 1.3984375, "learning_rate": 3.1729034075300416e-05, "loss": 0.1182, "step": 30702 }, { "epoch": 22.493040293040295, "grad_norm": 0.8203125, "learning_rate": 3.172586367367623e-05, "loss": 0.1177, "step": 30703 }, { "epoch": 22.493772893772892, "grad_norm": 1.1640625, "learning_rate": 3.17226933620051e-05, "loss": 0.1078, "step": 30704 }, { "epoch": 22.494505494505496, "grad_norm": 0.80078125, "learning_rate": 3.17195231403022e-05, "loss": 0.1069, "step": 30705 }, { "epoch": 22.495238095238093, "grad_norm": 1.1328125, "learning_rate": 3.171635300858274e-05, "loss": 0.1229, "step": 30706 }, { "epoch": 22.495970695970698, "grad_norm": 0.8125, "learning_rate": 3.171318296686187e-05, "loss": 0.1278, "step": 30707 }, { "epoch": 22.496703296703295, "grad_norm": 1.0, "learning_rate": 3.171001301515483e-05, "loss": 0.1021, "step": 30708 }, { "epoch": 22.4974358974359, "grad_norm": 1.1875, "learning_rate": 3.17068431534768e-05, "loss": 0.125, "step": 30709 }, { "epoch": 22.498168498168496, "grad_norm": 1.2265625, "learning_rate": 3.170367338184293e-05, "loss": 0.1259, "step": 30710 }, { "epoch": 22.4989010989011, "grad_norm": 1.140625, "learning_rate": 3.170050370026845e-05, "loss": 0.1278, "step": 30711 }, { "epoch": 22.499633699633698, "grad_norm": 0.98828125, "learning_rate": 3.1697334108768516e-05, "loss": 0.1637, "step": 30712 }, { "epoch": 22.500366300366302, "grad_norm": 0.8515625, "learning_rate": 3.169416460735832e-05, "loss": 0.1338, "step": 30713 }, { "epoch": 22.5010989010989, "grad_norm": 0.73046875, "learning_rate": 3.169099519605307e-05, "loss": 0.1085, "step": 30714 }, { "epoch": 22.501831501831504, "grad_norm": 1.09375, "learning_rate": 3.168782587486793e-05, "loss": 0.1379, "step": 30715 }, { "epoch": 22.5025641025641, "grad_norm": 1.859375, "learning_rate": 3.1684656643818086e-05, "loss": 0.1544, "step": 30716 }, { "epoch": 22.503296703296705, "grad_norm": 0.94921875, "learning_rate": 3.168148750291875e-05, "loss": 0.1172, "step": 30717 }, { "epoch": 22.504029304029302, "grad_norm": 1.53125, "learning_rate": 3.167831845218505e-05, "loss": 0.1051, "step": 30718 }, { "epoch": 22.504761904761907, "grad_norm": 1.4140625, "learning_rate": 3.1675149491632226e-05, "loss": 0.1269, "step": 30719 }, { "epoch": 22.505494505494504, "grad_norm": 0.953125, "learning_rate": 3.167198062127544e-05, "loss": 0.1158, "step": 30720 }, { "epoch": 22.506227106227108, "grad_norm": 1.046875, "learning_rate": 3.166881184112989e-05, "loss": 0.1621, "step": 30721 }, { "epoch": 22.506959706959705, "grad_norm": 1.0078125, "learning_rate": 3.166564315121073e-05, "loss": 0.1293, "step": 30722 }, { "epoch": 22.50769230769231, "grad_norm": 1.15625, "learning_rate": 3.1662474551533136e-05, "loss": 0.1384, "step": 30723 }, { "epoch": 22.508424908424907, "grad_norm": 1.078125, "learning_rate": 3.1659306042112344e-05, "loss": 0.1427, "step": 30724 }, { "epoch": 22.50915750915751, "grad_norm": 1.1328125, "learning_rate": 3.165613762296349e-05, "loss": 0.1413, "step": 30725 }, { "epoch": 22.509890109890108, "grad_norm": 1.1484375, "learning_rate": 3.1652969294101764e-05, "loss": 0.1307, "step": 30726 }, { "epoch": 22.510622710622712, "grad_norm": 1.1484375, "learning_rate": 3.164980105554235e-05, "loss": 0.1359, "step": 30727 }, { "epoch": 22.51135531135531, "grad_norm": 1.3515625, "learning_rate": 3.164663290730042e-05, "loss": 0.1179, "step": 30728 }, { "epoch": 22.512087912087914, "grad_norm": 1.1015625, "learning_rate": 3.1643464849391166e-05, "loss": 0.1115, "step": 30729 }, { "epoch": 22.51282051282051, "grad_norm": 0.5703125, "learning_rate": 3.1640296881829764e-05, "loss": 0.1011, "step": 30730 }, { "epoch": 22.513553113553115, "grad_norm": 1.140625, "learning_rate": 3.1637129004631395e-05, "loss": 0.1327, "step": 30731 }, { "epoch": 22.514285714285712, "grad_norm": 0.8515625, "learning_rate": 3.163396121781123e-05, "loss": 0.1288, "step": 30732 }, { "epoch": 22.515018315018317, "grad_norm": 1.2109375, "learning_rate": 3.163079352138444e-05, "loss": 0.1059, "step": 30733 }, { "epoch": 22.515750915750917, "grad_norm": 0.86328125, "learning_rate": 3.162762591536623e-05, "loss": 0.1299, "step": 30734 }, { "epoch": 22.516483516483518, "grad_norm": 1.3046875, "learning_rate": 3.162445839977175e-05, "loss": 0.1484, "step": 30735 }, { "epoch": 22.51721611721612, "grad_norm": 0.74609375, "learning_rate": 3.1621290974616195e-05, "loss": 0.1439, "step": 30736 }, { "epoch": 22.51794871794872, "grad_norm": 1.296875, "learning_rate": 3.161812363991473e-05, "loss": 0.1398, "step": 30737 }, { "epoch": 22.51868131868132, "grad_norm": 1.328125, "learning_rate": 3.1614956395682525e-05, "loss": 0.1025, "step": 30738 }, { "epoch": 22.51941391941392, "grad_norm": 0.65625, "learning_rate": 3.1611789241934766e-05, "loss": 0.1079, "step": 30739 }, { "epoch": 22.520146520146522, "grad_norm": 1.0703125, "learning_rate": 3.160862217868662e-05, "loss": 0.1329, "step": 30740 }, { "epoch": 22.520879120879123, "grad_norm": 1.1328125, "learning_rate": 3.1605455205953274e-05, "loss": 0.1158, "step": 30741 }, { "epoch": 22.521611721611723, "grad_norm": 1.5625, "learning_rate": 3.16022883237499e-05, "loss": 0.0881, "step": 30742 }, { "epoch": 22.522344322344324, "grad_norm": 0.9609375, "learning_rate": 3.159912153209166e-05, "loss": 0.1197, "step": 30743 }, { "epoch": 22.523076923076925, "grad_norm": 1.0625, "learning_rate": 3.1595954830993716e-05, "loss": 0.1274, "step": 30744 }, { "epoch": 22.523809523809526, "grad_norm": 1.1484375, "learning_rate": 3.159278822047128e-05, "loss": 0.1457, "step": 30745 }, { "epoch": 22.524542124542126, "grad_norm": 0.9296875, "learning_rate": 3.1589621700539504e-05, "loss": 0.1393, "step": 30746 }, { "epoch": 22.525274725274727, "grad_norm": 0.8828125, "learning_rate": 3.158645527121356e-05, "loss": 0.1267, "step": 30747 }, { "epoch": 22.526007326007328, "grad_norm": 0.92578125, "learning_rate": 3.1583288932508614e-05, "loss": 0.1268, "step": 30748 }, { "epoch": 22.52673992673993, "grad_norm": 0.8671875, "learning_rate": 3.158012268443982e-05, "loss": 0.0933, "step": 30749 }, { "epoch": 22.52747252747253, "grad_norm": 1.3125, "learning_rate": 3.1576956527022394e-05, "loss": 0.102, "step": 30750 }, { "epoch": 22.52820512820513, "grad_norm": 0.80859375, "learning_rate": 3.1573790460271475e-05, "loss": 0.1185, "step": 30751 }, { "epoch": 22.52893772893773, "grad_norm": 1.1640625, "learning_rate": 3.157062448420224e-05, "loss": 0.1121, "step": 30752 }, { "epoch": 22.52967032967033, "grad_norm": 0.6640625, "learning_rate": 3.1567458598829855e-05, "loss": 0.1131, "step": 30753 }, { "epoch": 22.530402930402932, "grad_norm": 1.21875, "learning_rate": 3.1564292804169485e-05, "loss": 0.1289, "step": 30754 }, { "epoch": 22.531135531135533, "grad_norm": 0.953125, "learning_rate": 3.1561127100236306e-05, "loss": 0.1248, "step": 30755 }, { "epoch": 22.531868131868134, "grad_norm": 0.8359375, "learning_rate": 3.155796148704549e-05, "loss": 0.1081, "step": 30756 }, { "epoch": 22.532600732600734, "grad_norm": 0.62890625, "learning_rate": 3.1554795964612204e-05, "loss": 0.1222, "step": 30757 }, { "epoch": 22.533333333333335, "grad_norm": 1.046875, "learning_rate": 3.155163053295161e-05, "loss": 0.1204, "step": 30758 }, { "epoch": 22.534065934065936, "grad_norm": 1.546875, "learning_rate": 3.1548465192078845e-05, "loss": 0.1231, "step": 30759 }, { "epoch": 22.534798534798536, "grad_norm": 0.89453125, "learning_rate": 3.154529994200914e-05, "loss": 0.1112, "step": 30760 }, { "epoch": 22.535531135531137, "grad_norm": 1.0546875, "learning_rate": 3.154213478275762e-05, "loss": 0.131, "step": 30761 }, { "epoch": 22.536263736263738, "grad_norm": 1.2265625, "learning_rate": 3.153896971433944e-05, "loss": 0.1551, "step": 30762 }, { "epoch": 22.53699633699634, "grad_norm": 2.03125, "learning_rate": 3.15358047367698e-05, "loss": 0.1407, "step": 30763 }, { "epoch": 22.53772893772894, "grad_norm": 1.1796875, "learning_rate": 3.1532639850063825e-05, "loss": 0.118, "step": 30764 }, { "epoch": 22.53846153846154, "grad_norm": 0.83984375, "learning_rate": 3.152947505423671e-05, "loss": 0.1087, "step": 30765 }, { "epoch": 22.53919413919414, "grad_norm": 0.953125, "learning_rate": 3.1526310349303605e-05, "loss": 0.1433, "step": 30766 }, { "epoch": 22.53992673992674, "grad_norm": 1.7734375, "learning_rate": 3.1523145735279675e-05, "loss": 0.1028, "step": 30767 }, { "epoch": 22.540659340659342, "grad_norm": 1.3046875, "learning_rate": 3.151998121218009e-05, "loss": 0.1544, "step": 30768 }, { "epoch": 22.541391941391943, "grad_norm": 1.3046875, "learning_rate": 3.151681678002e-05, "loss": 0.178, "step": 30769 }, { "epoch": 22.542124542124544, "grad_norm": 1.2265625, "learning_rate": 3.1513652438814564e-05, "loss": 0.1573, "step": 30770 }, { "epoch": 22.542857142857144, "grad_norm": 1.515625, "learning_rate": 3.151048818857896e-05, "loss": 0.1449, "step": 30771 }, { "epoch": 22.543589743589745, "grad_norm": 1.5625, "learning_rate": 3.150732402932835e-05, "loss": 0.1171, "step": 30772 }, { "epoch": 22.544322344322346, "grad_norm": 0.93359375, "learning_rate": 3.150415996107788e-05, "loss": 0.1267, "step": 30773 }, { "epoch": 22.545054945054947, "grad_norm": 1.171875, "learning_rate": 3.1500995983842716e-05, "loss": 0.1165, "step": 30774 }, { "epoch": 22.545787545787547, "grad_norm": 1.1796875, "learning_rate": 3.1497832097638e-05, "loss": 0.1224, "step": 30775 }, { "epoch": 22.546520146520148, "grad_norm": 1.328125, "learning_rate": 3.149466830247892e-05, "loss": 0.1176, "step": 30776 }, { "epoch": 22.54725274725275, "grad_norm": 0.75390625, "learning_rate": 3.149150459838064e-05, "loss": 0.0861, "step": 30777 }, { "epoch": 22.54798534798535, "grad_norm": 1.6015625, "learning_rate": 3.148834098535828e-05, "loss": 0.1238, "step": 30778 }, { "epoch": 22.54871794871795, "grad_norm": 0.83984375, "learning_rate": 3.148517746342703e-05, "loss": 0.1118, "step": 30779 }, { "epoch": 22.54945054945055, "grad_norm": 0.91015625, "learning_rate": 3.148201403260203e-05, "loss": 0.1314, "step": 30780 }, { "epoch": 22.550183150183152, "grad_norm": 1.1875, "learning_rate": 3.147885069289844e-05, "loss": 0.1293, "step": 30781 }, { "epoch": 22.550915750915753, "grad_norm": 1.4765625, "learning_rate": 3.147568744433144e-05, "loss": 0.1056, "step": 30782 }, { "epoch": 22.551648351648353, "grad_norm": 1.7109375, "learning_rate": 3.147252428691615e-05, "loss": 0.1219, "step": 30783 }, { "epoch": 22.552380952380954, "grad_norm": 1.28125, "learning_rate": 3.146936122066776e-05, "loss": 0.1083, "step": 30784 }, { "epoch": 22.553113553113555, "grad_norm": 1.0, "learning_rate": 3.146619824560137e-05, "loss": 0.1093, "step": 30785 }, { "epoch": 22.553846153846155, "grad_norm": 1.578125, "learning_rate": 3.1463035361732206e-05, "loss": 0.1466, "step": 30786 }, { "epoch": 22.554578754578756, "grad_norm": 0.8515625, "learning_rate": 3.145987256907539e-05, "loss": 0.1363, "step": 30787 }, { "epoch": 22.555311355311357, "grad_norm": 0.9453125, "learning_rate": 3.145670986764607e-05, "loss": 0.1165, "step": 30788 }, { "epoch": 22.556043956043958, "grad_norm": 0.94140625, "learning_rate": 3.1453547257459406e-05, "loss": 0.1218, "step": 30789 }, { "epoch": 22.55677655677656, "grad_norm": 0.859375, "learning_rate": 3.1450384738530536e-05, "loss": 0.1344, "step": 30790 }, { "epoch": 22.55750915750916, "grad_norm": 1.1484375, "learning_rate": 3.144722231087463e-05, "loss": 0.1505, "step": 30791 }, { "epoch": 22.55824175824176, "grad_norm": 1.1875, "learning_rate": 3.144405997450685e-05, "loss": 0.1451, "step": 30792 }, { "epoch": 22.55897435897436, "grad_norm": 1.03125, "learning_rate": 3.1440897729442315e-05, "loss": 0.1121, "step": 30793 }, { "epoch": 22.55970695970696, "grad_norm": 1.2578125, "learning_rate": 3.143773557569622e-05, "loss": 0.1376, "step": 30794 }, { "epoch": 22.560439560439562, "grad_norm": 0.9921875, "learning_rate": 3.143457351328367e-05, "loss": 0.1, "step": 30795 }, { "epoch": 22.561172161172163, "grad_norm": 1.0859375, "learning_rate": 3.143141154221983e-05, "loss": 0.138, "step": 30796 }, { "epoch": 22.561904761904763, "grad_norm": 1.1328125, "learning_rate": 3.142824966251988e-05, "loss": 0.1179, "step": 30797 }, { "epoch": 22.562637362637364, "grad_norm": 1.015625, "learning_rate": 3.1425087874198937e-05, "loss": 0.1063, "step": 30798 }, { "epoch": 22.563369963369965, "grad_norm": 0.69921875, "learning_rate": 3.1421926177272174e-05, "loss": 0.1129, "step": 30799 }, { "epoch": 22.564102564102566, "grad_norm": 1.0078125, "learning_rate": 3.141876457175471e-05, "loss": 0.1059, "step": 30800 }, { "epoch": 22.564835164835166, "grad_norm": 1.390625, "learning_rate": 3.14156030576617e-05, "loss": 0.1358, "step": 30801 }, { "epoch": 22.565567765567767, "grad_norm": 0.921875, "learning_rate": 3.141244163500832e-05, "loss": 0.127, "step": 30802 }, { "epoch": 22.566300366300368, "grad_norm": 0.7265625, "learning_rate": 3.140928030380969e-05, "loss": 0.124, "step": 30803 }, { "epoch": 22.56703296703297, "grad_norm": 1.3046875, "learning_rate": 3.1406119064080966e-05, "loss": 0.1284, "step": 30804 }, { "epoch": 22.56776556776557, "grad_norm": 1.21875, "learning_rate": 3.14029579158373e-05, "loss": 0.0971, "step": 30805 }, { "epoch": 22.56849816849817, "grad_norm": 0.7578125, "learning_rate": 3.139979685909381e-05, "loss": 0.1296, "step": 30806 }, { "epoch": 22.56923076923077, "grad_norm": 0.9453125, "learning_rate": 3.139663589386567e-05, "loss": 0.1261, "step": 30807 }, { "epoch": 22.56996336996337, "grad_norm": 0.80859375, "learning_rate": 3.1393475020168046e-05, "loss": 0.119, "step": 30808 }, { "epoch": 22.570695970695972, "grad_norm": 1.015625, "learning_rate": 3.1390314238016036e-05, "loss": 0.0812, "step": 30809 }, { "epoch": 22.571428571428573, "grad_norm": 0.8046875, "learning_rate": 3.1387153547424805e-05, "loss": 0.1182, "step": 30810 }, { "epoch": 22.572161172161174, "grad_norm": 1.1015625, "learning_rate": 3.138399294840948e-05, "loss": 0.1099, "step": 30811 }, { "epoch": 22.572893772893774, "grad_norm": 1.09375, "learning_rate": 3.1380832440985244e-05, "loss": 0.1112, "step": 30812 }, { "epoch": 22.573626373626375, "grad_norm": 1.296875, "learning_rate": 3.137767202516721e-05, "loss": 0.1147, "step": 30813 }, { "epoch": 22.574358974358976, "grad_norm": 0.921875, "learning_rate": 3.137451170097053e-05, "loss": 0.1151, "step": 30814 }, { "epoch": 22.575091575091577, "grad_norm": 0.74609375, "learning_rate": 3.137135146841034e-05, "loss": 0.1432, "step": 30815 }, { "epoch": 22.575824175824177, "grad_norm": 2.109375, "learning_rate": 3.136819132750177e-05, "loss": 0.17, "step": 30816 }, { "epoch": 22.576556776556778, "grad_norm": 0.6328125, "learning_rate": 3.136503127825999e-05, "loss": 0.1046, "step": 30817 }, { "epoch": 22.57728937728938, "grad_norm": 1.3671875, "learning_rate": 3.1361871320700134e-05, "loss": 0.1069, "step": 30818 }, { "epoch": 22.57802197802198, "grad_norm": 0.87890625, "learning_rate": 3.135871145483731e-05, "loss": 0.1157, "step": 30819 }, { "epoch": 22.57875457875458, "grad_norm": 1.609375, "learning_rate": 3.135555168068671e-05, "loss": 0.0976, "step": 30820 }, { "epoch": 22.57948717948718, "grad_norm": 1.625, "learning_rate": 3.135239199826343e-05, "loss": 0.1339, "step": 30821 }, { "epoch": 22.58021978021978, "grad_norm": 1.109375, "learning_rate": 3.134923240758262e-05, "loss": 0.1302, "step": 30822 }, { "epoch": 22.580952380952382, "grad_norm": 1.2109375, "learning_rate": 3.1346072908659444e-05, "loss": 0.1206, "step": 30823 }, { "epoch": 22.581684981684983, "grad_norm": 0.546875, "learning_rate": 3.134291350150902e-05, "loss": 0.1056, "step": 30824 }, { "epoch": 22.582417582417584, "grad_norm": 0.9140625, "learning_rate": 3.133975418614648e-05, "loss": 0.1137, "step": 30825 }, { "epoch": 22.583150183150185, "grad_norm": 0.7421875, "learning_rate": 3.133659496258697e-05, "loss": 0.1055, "step": 30826 }, { "epoch": 22.583882783882785, "grad_norm": 1.09375, "learning_rate": 3.133343583084561e-05, "loss": 0.1291, "step": 30827 }, { "epoch": 22.584615384615386, "grad_norm": 1.515625, "learning_rate": 3.133027679093756e-05, "loss": 0.1045, "step": 30828 }, { "epoch": 22.585347985347987, "grad_norm": 1.984375, "learning_rate": 3.132711784287795e-05, "loss": 0.1325, "step": 30829 }, { "epoch": 22.586080586080588, "grad_norm": 0.73046875, "learning_rate": 3.132395898668191e-05, "loss": 0.1451, "step": 30830 }, { "epoch": 22.58681318681319, "grad_norm": 1.109375, "learning_rate": 3.132080022236458e-05, "loss": 0.1095, "step": 30831 }, { "epoch": 22.58754578754579, "grad_norm": 0.8515625, "learning_rate": 3.1317641549941086e-05, "loss": 0.1208, "step": 30832 }, { "epoch": 22.58827838827839, "grad_norm": 1.4375, "learning_rate": 3.131448296942657e-05, "loss": 0.1604, "step": 30833 }, { "epoch": 22.58901098901099, "grad_norm": 1.4296875, "learning_rate": 3.1311324480836174e-05, "loss": 0.1181, "step": 30834 }, { "epoch": 22.58974358974359, "grad_norm": 1.0390625, "learning_rate": 3.130816608418501e-05, "loss": 0.1141, "step": 30835 }, { "epoch": 22.590476190476192, "grad_norm": 1.1171875, "learning_rate": 3.130500777948824e-05, "loss": 0.1113, "step": 30836 }, { "epoch": 22.591208791208793, "grad_norm": 1.1015625, "learning_rate": 3.1301849566760946e-05, "loss": 0.1083, "step": 30837 }, { "epoch": 22.591941391941393, "grad_norm": 0.9140625, "learning_rate": 3.129869144601832e-05, "loss": 0.113, "step": 30838 }, { "epoch": 22.592673992673994, "grad_norm": 0.76171875, "learning_rate": 3.129553341727546e-05, "loss": 0.1172, "step": 30839 }, { "epoch": 22.593406593406595, "grad_norm": 1.0859375, "learning_rate": 3.129237548054752e-05, "loss": 0.1182, "step": 30840 }, { "epoch": 22.594139194139196, "grad_norm": 0.6796875, "learning_rate": 3.1289217635849593e-05, "loss": 0.1292, "step": 30841 }, { "epoch": 22.594871794871796, "grad_norm": 0.87890625, "learning_rate": 3.1286059883196836e-05, "loss": 0.1287, "step": 30842 }, { "epoch": 22.595604395604397, "grad_norm": 0.9375, "learning_rate": 3.128290222260437e-05, "loss": 0.1146, "step": 30843 }, { "epoch": 22.596336996336998, "grad_norm": 0.97265625, "learning_rate": 3.127974465408733e-05, "loss": 0.1372, "step": 30844 }, { "epoch": 22.5970695970696, "grad_norm": 0.64453125, "learning_rate": 3.127658717766084e-05, "loss": 0.1357, "step": 30845 }, { "epoch": 22.5978021978022, "grad_norm": 1.484375, "learning_rate": 3.127342979334004e-05, "loss": 0.1198, "step": 30846 }, { "epoch": 22.5985347985348, "grad_norm": 0.9375, "learning_rate": 3.127027250114005e-05, "loss": 0.1176, "step": 30847 }, { "epoch": 22.5992673992674, "grad_norm": 0.9765625, "learning_rate": 3.126711530107598e-05, "loss": 0.1539, "step": 30848 }, { "epoch": 22.6, "grad_norm": 0.78515625, "learning_rate": 3.126395819316299e-05, "loss": 0.1027, "step": 30849 }, { "epoch": 22.600732600732602, "grad_norm": 0.765625, "learning_rate": 3.12608011774162e-05, "loss": 0.0902, "step": 30850 }, { "epoch": 22.601465201465203, "grad_norm": 1.3125, "learning_rate": 3.1257644253850717e-05, "loss": 0.1076, "step": 30851 }, { "epoch": 22.602197802197804, "grad_norm": 1.09375, "learning_rate": 3.125448742248168e-05, "loss": 0.1327, "step": 30852 }, { "epoch": 22.602930402930404, "grad_norm": 1.0859375, "learning_rate": 3.12513306833242e-05, "loss": 0.1049, "step": 30853 }, { "epoch": 22.603663003663005, "grad_norm": 0.76953125, "learning_rate": 3.124817403639342e-05, "loss": 0.1355, "step": 30854 }, { "epoch": 22.604395604395606, "grad_norm": 1.0390625, "learning_rate": 3.1245017481704455e-05, "loss": 0.1065, "step": 30855 }, { "epoch": 22.605128205128207, "grad_norm": 0.87890625, "learning_rate": 3.124186101927244e-05, "loss": 0.1198, "step": 30856 }, { "epoch": 22.605860805860807, "grad_norm": 1.3828125, "learning_rate": 3.1238704649112485e-05, "loss": 0.1062, "step": 30857 }, { "epoch": 22.606593406593408, "grad_norm": 0.8984375, "learning_rate": 3.123554837123973e-05, "loss": 0.1235, "step": 30858 }, { "epoch": 22.60732600732601, "grad_norm": 0.87109375, "learning_rate": 3.123239218566927e-05, "loss": 0.1199, "step": 30859 }, { "epoch": 22.60805860805861, "grad_norm": 1.546875, "learning_rate": 3.1229236092416256e-05, "loss": 0.1332, "step": 30860 }, { "epoch": 22.60879120879121, "grad_norm": 1.1796875, "learning_rate": 3.122608009149579e-05, "loss": 0.1036, "step": 30861 }, { "epoch": 22.60952380952381, "grad_norm": 1.109375, "learning_rate": 3.122292418292302e-05, "loss": 0.1281, "step": 30862 }, { "epoch": 22.61025641025641, "grad_norm": 1.1875, "learning_rate": 3.1219768366713016e-05, "loss": 0.1167, "step": 30863 }, { "epoch": 22.610989010989012, "grad_norm": 0.9765625, "learning_rate": 3.121661264288095e-05, "loss": 0.1317, "step": 30864 }, { "epoch": 22.611721611721613, "grad_norm": 0.76171875, "learning_rate": 3.1213457011441935e-05, "loss": 0.1178, "step": 30865 }, { "epoch": 22.612454212454214, "grad_norm": 1.015625, "learning_rate": 3.121030147241107e-05, "loss": 0.1596, "step": 30866 }, { "epoch": 22.613186813186815, "grad_norm": 1.6796875, "learning_rate": 3.1207146025803484e-05, "loss": 0.1321, "step": 30867 }, { "epoch": 22.613919413919415, "grad_norm": 1.296875, "learning_rate": 3.120399067163428e-05, "loss": 0.1334, "step": 30868 }, { "epoch": 22.614652014652016, "grad_norm": 0.83984375, "learning_rate": 3.1200835409918597e-05, "loss": 0.1177, "step": 30869 }, { "epoch": 22.615384615384617, "grad_norm": 1.078125, "learning_rate": 3.1197680240671544e-05, "loss": 0.1078, "step": 30870 }, { "epoch": 22.616117216117217, "grad_norm": 1.34375, "learning_rate": 3.1194525163908264e-05, "loss": 0.1129, "step": 30871 }, { "epoch": 22.616849816849818, "grad_norm": 0.83203125, "learning_rate": 3.1191370179643835e-05, "loss": 0.0959, "step": 30872 }, { "epoch": 22.61758241758242, "grad_norm": 1.2578125, "learning_rate": 3.1188215287893395e-05, "loss": 0.1431, "step": 30873 }, { "epoch": 22.61831501831502, "grad_norm": 1.1640625, "learning_rate": 3.118506048867204e-05, "loss": 0.1401, "step": 30874 }, { "epoch": 22.61904761904762, "grad_norm": 0.91796875, "learning_rate": 3.118190578199492e-05, "loss": 0.1037, "step": 30875 }, { "epoch": 22.61978021978022, "grad_norm": 1.4375, "learning_rate": 3.1178751167877125e-05, "loss": 0.1243, "step": 30876 }, { "epoch": 22.620512820512822, "grad_norm": 1.078125, "learning_rate": 3.1175596646333785e-05, "loss": 0.1238, "step": 30877 }, { "epoch": 22.621245421245423, "grad_norm": 0.67578125, "learning_rate": 3.1172442217379996e-05, "loss": 0.111, "step": 30878 }, { "epoch": 22.621978021978023, "grad_norm": 1.0859375, "learning_rate": 3.116928788103087e-05, "loss": 0.1283, "step": 30879 }, { "epoch": 22.622710622710624, "grad_norm": 1.1171875, "learning_rate": 3.116613363730154e-05, "loss": 0.1292, "step": 30880 }, { "epoch": 22.623443223443225, "grad_norm": 0.75390625, "learning_rate": 3.116297948620712e-05, "loss": 0.1289, "step": 30881 }, { "epoch": 22.624175824175826, "grad_norm": 1.6484375, "learning_rate": 3.11598254277627e-05, "loss": 0.1577, "step": 30882 }, { "epoch": 22.624908424908426, "grad_norm": 1.703125, "learning_rate": 3.115667146198342e-05, "loss": 0.1245, "step": 30883 }, { "epoch": 22.625641025641027, "grad_norm": 0.84375, "learning_rate": 3.115351758888435e-05, "loss": 0.1042, "step": 30884 }, { "epoch": 22.626373626373628, "grad_norm": 1.1484375, "learning_rate": 3.115036380848064e-05, "loss": 0.1272, "step": 30885 }, { "epoch": 22.62710622710623, "grad_norm": 1.0859375, "learning_rate": 3.1147210120787406e-05, "loss": 0.1281, "step": 30886 }, { "epoch": 22.62783882783883, "grad_norm": 1.453125, "learning_rate": 3.1144056525819736e-05, "loss": 0.1347, "step": 30887 }, { "epoch": 22.62857142857143, "grad_norm": 1.0078125, "learning_rate": 3.114090302359275e-05, "loss": 0.1192, "step": 30888 }, { "epoch": 22.62930402930403, "grad_norm": 1.2890625, "learning_rate": 3.1137749614121525e-05, "loss": 0.1239, "step": 30889 }, { "epoch": 22.63003663003663, "grad_norm": 1.3671875, "learning_rate": 3.1134596297421214e-05, "loss": 0.1089, "step": 30890 }, { "epoch": 22.630769230769232, "grad_norm": 1.109375, "learning_rate": 3.1131443073506924e-05, "loss": 0.143, "step": 30891 }, { "epoch": 22.631501831501833, "grad_norm": 1.2734375, "learning_rate": 3.112828994239374e-05, "loss": 0.0943, "step": 30892 }, { "epoch": 22.632234432234434, "grad_norm": 2.390625, "learning_rate": 3.112513690409677e-05, "loss": 0.12, "step": 30893 }, { "epoch": 22.632967032967034, "grad_norm": 1.796875, "learning_rate": 3.1121983958631135e-05, "loss": 0.1298, "step": 30894 }, { "epoch": 22.633699633699635, "grad_norm": 1.2734375, "learning_rate": 3.111883110601193e-05, "loss": 0.1002, "step": 30895 }, { "epoch": 22.634432234432236, "grad_norm": 1.5234375, "learning_rate": 3.1115678346254266e-05, "loss": 0.1274, "step": 30896 }, { "epoch": 22.635164835164836, "grad_norm": 0.890625, "learning_rate": 3.111252567937326e-05, "loss": 0.1331, "step": 30897 }, { "epoch": 22.635897435897437, "grad_norm": 1.3984375, "learning_rate": 3.1109373105384014e-05, "loss": 0.123, "step": 30898 }, { "epoch": 22.636630036630038, "grad_norm": 1.2890625, "learning_rate": 3.110622062430162e-05, "loss": 0.1192, "step": 30899 }, { "epoch": 22.63736263736264, "grad_norm": 1.2734375, "learning_rate": 3.1103068236141174e-05, "loss": 0.1238, "step": 30900 }, { "epoch": 22.63809523809524, "grad_norm": 1.265625, "learning_rate": 3.109991594091781e-05, "loss": 0.0966, "step": 30901 }, { "epoch": 22.63882783882784, "grad_norm": 1.015625, "learning_rate": 3.109676373864662e-05, "loss": 0.1119, "step": 30902 }, { "epoch": 22.63956043956044, "grad_norm": 1.0546875, "learning_rate": 3.109361162934271e-05, "loss": 0.1379, "step": 30903 }, { "epoch": 22.64029304029304, "grad_norm": 0.95703125, "learning_rate": 3.1090459613021174e-05, "loss": 0.1093, "step": 30904 }, { "epoch": 22.641025641025642, "grad_norm": 1.484375, "learning_rate": 3.1087307689697094e-05, "loss": 0.1208, "step": 30905 }, { "epoch": 22.641758241758243, "grad_norm": 1.140625, "learning_rate": 3.1084155859385615e-05, "loss": 0.0992, "step": 30906 }, { "epoch": 22.642490842490844, "grad_norm": 1.1171875, "learning_rate": 3.1081004122101826e-05, "loss": 0.1309, "step": 30907 }, { "epoch": 22.643223443223444, "grad_norm": 0.9609375, "learning_rate": 3.107785247786081e-05, "loss": 0.1524, "step": 30908 }, { "epoch": 22.643956043956045, "grad_norm": 0.71875, "learning_rate": 3.107470092667768e-05, "loss": 0.1314, "step": 30909 }, { "epoch": 22.644688644688646, "grad_norm": 1.1484375, "learning_rate": 3.1071549468567516e-05, "loss": 0.1058, "step": 30910 }, { "epoch": 22.645421245421247, "grad_norm": 0.78125, "learning_rate": 3.1068398103545465e-05, "loss": 0.1409, "step": 30911 }, { "epoch": 22.646153846153847, "grad_norm": 1.21875, "learning_rate": 3.10652468316266e-05, "loss": 0.1205, "step": 30912 }, { "epoch": 22.646886446886448, "grad_norm": 1.21875, "learning_rate": 3.106209565282601e-05, "loss": 0.1122, "step": 30913 }, { "epoch": 22.64761904761905, "grad_norm": 1.2890625, "learning_rate": 3.1058944567158796e-05, "loss": 0.1117, "step": 30914 }, { "epoch": 22.64835164835165, "grad_norm": 0.9921875, "learning_rate": 3.105579357464005e-05, "loss": 0.1036, "step": 30915 }, { "epoch": 22.64908424908425, "grad_norm": 0.7734375, "learning_rate": 3.10526426752849e-05, "loss": 0.12, "step": 30916 }, { "epoch": 22.64981684981685, "grad_norm": 1.0390625, "learning_rate": 3.104949186910842e-05, "loss": 0.147, "step": 30917 }, { "epoch": 22.650549450549452, "grad_norm": 0.921875, "learning_rate": 3.104634115612571e-05, "loss": 0.1147, "step": 30918 }, { "epoch": 22.651282051282053, "grad_norm": 1.0625, "learning_rate": 3.1043190536351854e-05, "loss": 0.1051, "step": 30919 }, { "epoch": 22.652014652014653, "grad_norm": 0.99609375, "learning_rate": 3.104004000980197e-05, "loss": 0.1174, "step": 30920 }, { "epoch": 22.652747252747254, "grad_norm": 1.0703125, "learning_rate": 3.1036889576491136e-05, "loss": 0.1271, "step": 30921 }, { "epoch": 22.653479853479855, "grad_norm": 1.1171875, "learning_rate": 3.103373923643445e-05, "loss": 0.1173, "step": 30922 }, { "epoch": 22.654212454212455, "grad_norm": 0.94921875, "learning_rate": 3.103058898964702e-05, "loss": 0.1479, "step": 30923 }, { "epoch": 22.654945054945056, "grad_norm": 0.69921875, "learning_rate": 3.102743883614393e-05, "loss": 0.1037, "step": 30924 }, { "epoch": 22.655677655677657, "grad_norm": 0.875, "learning_rate": 3.102428877594027e-05, "loss": 0.0996, "step": 30925 }, { "epoch": 22.656410256410258, "grad_norm": 0.91796875, "learning_rate": 3.102113880905112e-05, "loss": 0.114, "step": 30926 }, { "epoch": 22.65714285714286, "grad_norm": 1.0078125, "learning_rate": 3.101798893549161e-05, "loss": 0.1037, "step": 30927 }, { "epoch": 22.65787545787546, "grad_norm": 1.140625, "learning_rate": 3.1014839155276806e-05, "loss": 0.1362, "step": 30928 }, { "epoch": 22.65860805860806, "grad_norm": 1.2890625, "learning_rate": 3.10116894684218e-05, "loss": 0.1121, "step": 30929 }, { "epoch": 22.65934065934066, "grad_norm": 1.140625, "learning_rate": 3.100853987494169e-05, "loss": 0.1382, "step": 30930 }, { "epoch": 22.66007326007326, "grad_norm": 0.8359375, "learning_rate": 3.100539037485155e-05, "loss": 0.086, "step": 30931 }, { "epoch": 22.660805860805862, "grad_norm": 0.66015625, "learning_rate": 3.100224096816651e-05, "loss": 0.1109, "step": 30932 }, { "epoch": 22.661538461538463, "grad_norm": 1.4765625, "learning_rate": 3.099909165490162e-05, "loss": 0.1197, "step": 30933 }, { "epoch": 22.662271062271063, "grad_norm": 1.3515625, "learning_rate": 3.0995942435071976e-05, "loss": 0.1208, "step": 30934 }, { "epoch": 22.663003663003664, "grad_norm": 1.390625, "learning_rate": 3.09927933086927e-05, "loss": 0.107, "step": 30935 }, { "epoch": 22.663736263736265, "grad_norm": 1.3359375, "learning_rate": 3.0989644275778816e-05, "loss": 0.156, "step": 30936 }, { "epoch": 22.664468864468866, "grad_norm": 0.71875, "learning_rate": 3.0986495336345486e-05, "loss": 0.1072, "step": 30937 }, { "epoch": 22.665201465201466, "grad_norm": 1.140625, "learning_rate": 3.098334649040775e-05, "loss": 0.126, "step": 30938 }, { "epoch": 22.665934065934067, "grad_norm": 1.09375, "learning_rate": 3.098019773798071e-05, "loss": 0.1122, "step": 30939 }, { "epoch": 22.666666666666668, "grad_norm": 0.671875, "learning_rate": 3.097704907907946e-05, "loss": 0.1352, "step": 30940 }, { "epoch": 22.66739926739927, "grad_norm": 0.86328125, "learning_rate": 3.097390051371907e-05, "loss": 0.1278, "step": 30941 }, { "epoch": 22.66813186813187, "grad_norm": 1.9375, "learning_rate": 3.097075204191461e-05, "loss": 0.1271, "step": 30942 }, { "epoch": 22.66886446886447, "grad_norm": 1.84375, "learning_rate": 3.0967603663681214e-05, "loss": 0.1372, "step": 30943 }, { "epoch": 22.66959706959707, "grad_norm": 0.859375, "learning_rate": 3.096445537903393e-05, "loss": 0.1522, "step": 30944 }, { "epoch": 22.67032967032967, "grad_norm": 1.40625, "learning_rate": 3.096130718798785e-05, "loss": 0.1627, "step": 30945 }, { "epoch": 22.671062271062272, "grad_norm": 1.1171875, "learning_rate": 3.0958159090558074e-05, "loss": 0.1031, "step": 30946 }, { "epoch": 22.671794871794873, "grad_norm": 1.171875, "learning_rate": 3.095501108675966e-05, "loss": 0.1075, "step": 30947 }, { "epoch": 22.672527472527474, "grad_norm": 1.1328125, "learning_rate": 3.0951863176607714e-05, "loss": 0.1045, "step": 30948 }, { "epoch": 22.673260073260074, "grad_norm": 0.8125, "learning_rate": 3.094871536011731e-05, "loss": 0.0975, "step": 30949 }, { "epoch": 22.673992673992675, "grad_norm": 1.1796875, "learning_rate": 3.094556763730353e-05, "loss": 0.1401, "step": 30950 }, { "epoch": 22.674725274725276, "grad_norm": 1.265625, "learning_rate": 3.094242000818145e-05, "loss": 0.1296, "step": 30951 }, { "epoch": 22.675457875457877, "grad_norm": 1.4453125, "learning_rate": 3.093927247276615e-05, "loss": 0.1313, "step": 30952 }, { "epoch": 22.676190476190477, "grad_norm": 1.734375, "learning_rate": 3.0936125031072725e-05, "loss": 0.145, "step": 30953 }, { "epoch": 22.676923076923078, "grad_norm": 1.4140625, "learning_rate": 3.093297768311624e-05, "loss": 0.1351, "step": 30954 }, { "epoch": 22.67765567765568, "grad_norm": 1.0859375, "learning_rate": 3.09298304289118e-05, "loss": 0.1185, "step": 30955 }, { "epoch": 22.67838827838828, "grad_norm": 1.171875, "learning_rate": 3.092668326847446e-05, "loss": 0.1163, "step": 30956 }, { "epoch": 22.67912087912088, "grad_norm": 0.98046875, "learning_rate": 3.092353620181928e-05, "loss": 0.1325, "step": 30957 }, { "epoch": 22.67985347985348, "grad_norm": 1.015625, "learning_rate": 3.092038922896139e-05, "loss": 0.1305, "step": 30958 }, { "epoch": 22.68058608058608, "grad_norm": 0.65625, "learning_rate": 3.0917242349915834e-05, "loss": 0.1225, "step": 30959 }, { "epoch": 22.681318681318682, "grad_norm": 1.4375, "learning_rate": 3.09140955646977e-05, "loss": 0.1548, "step": 30960 }, { "epoch": 22.682051282051283, "grad_norm": 0.91796875, "learning_rate": 3.091094887332207e-05, "loss": 0.1131, "step": 30961 }, { "epoch": 22.682783882783884, "grad_norm": 0.73828125, "learning_rate": 3.0907802275804005e-05, "loss": 0.0976, "step": 30962 }, { "epoch": 22.683516483516485, "grad_norm": 1.4765625, "learning_rate": 3.0904655772158595e-05, "loss": 0.1328, "step": 30963 }, { "epoch": 22.684249084249085, "grad_norm": 0.85546875, "learning_rate": 3.090150936240092e-05, "loss": 0.1253, "step": 30964 }, { "epoch": 22.684981684981686, "grad_norm": 0.765625, "learning_rate": 3.089836304654604e-05, "loss": 0.1085, "step": 30965 }, { "epoch": 22.685714285714287, "grad_norm": 1.265625, "learning_rate": 3.089521682460905e-05, "loss": 0.1449, "step": 30966 }, { "epoch": 22.686446886446888, "grad_norm": 1.953125, "learning_rate": 3.0892070696605005e-05, "loss": 0.1134, "step": 30967 }, { "epoch": 22.68717948717949, "grad_norm": 1.015625, "learning_rate": 3.088892466254896e-05, "loss": 0.1177, "step": 30968 }, { "epoch": 22.68791208791209, "grad_norm": 0.984375, "learning_rate": 3.0885778722456046e-05, "loss": 0.1138, "step": 30969 }, { "epoch": 22.68864468864469, "grad_norm": 1.8984375, "learning_rate": 3.08826328763413e-05, "loss": 0.1697, "step": 30970 }, { "epoch": 22.68937728937729, "grad_norm": 0.890625, "learning_rate": 3.0879487124219786e-05, "loss": 0.1456, "step": 30971 }, { "epoch": 22.69010989010989, "grad_norm": 1.5234375, "learning_rate": 3.0876341466106605e-05, "loss": 0.1331, "step": 30972 }, { "epoch": 22.690842490842492, "grad_norm": 1.0546875, "learning_rate": 3.08731959020168e-05, "loss": 0.1206, "step": 30973 }, { "epoch": 22.691575091575093, "grad_norm": 1.3984375, "learning_rate": 3.087005043196546e-05, "loss": 0.1761, "step": 30974 }, { "epoch": 22.692307692307693, "grad_norm": 1.328125, "learning_rate": 3.086690505596766e-05, "loss": 0.1013, "step": 30975 }, { "epoch": 22.693040293040294, "grad_norm": 0.73828125, "learning_rate": 3.086375977403847e-05, "loss": 0.1309, "step": 30976 }, { "epoch": 22.693772893772895, "grad_norm": 0.90625, "learning_rate": 3.086061458619295e-05, "loss": 0.0961, "step": 30977 }, { "epoch": 22.694505494505496, "grad_norm": 1.0625, "learning_rate": 3.085746949244615e-05, "loss": 0.1514, "step": 30978 }, { "epoch": 22.695238095238096, "grad_norm": 1.5390625, "learning_rate": 3.085432449281318e-05, "loss": 0.1256, "step": 30979 }, { "epoch": 22.695970695970697, "grad_norm": 0.80078125, "learning_rate": 3.0851179587309094e-05, "loss": 0.1124, "step": 30980 }, { "epoch": 22.696703296703298, "grad_norm": 0.87890625, "learning_rate": 3.084803477594895e-05, "loss": 0.1281, "step": 30981 }, { "epoch": 22.6974358974359, "grad_norm": 1.0859375, "learning_rate": 3.084489005874783e-05, "loss": 0.1502, "step": 30982 }, { "epoch": 22.6981684981685, "grad_norm": 0.6015625, "learning_rate": 3.084174543572078e-05, "loss": 0.1208, "step": 30983 }, { "epoch": 22.6989010989011, "grad_norm": 1.3125, "learning_rate": 3.0838600906882886e-05, "loss": 0.1025, "step": 30984 }, { "epoch": 22.6996336996337, "grad_norm": 1.3203125, "learning_rate": 3.0835456472249206e-05, "loss": 0.1121, "step": 30985 }, { "epoch": 22.7003663003663, "grad_norm": 1.109375, "learning_rate": 3.083231213183483e-05, "loss": 0.1133, "step": 30986 }, { "epoch": 22.701098901098902, "grad_norm": 0.94140625, "learning_rate": 3.08291678856548e-05, "loss": 0.0863, "step": 30987 }, { "epoch": 22.701831501831503, "grad_norm": 1.53125, "learning_rate": 3.082602373372415e-05, "loss": 0.1518, "step": 30988 }, { "epoch": 22.702564102564104, "grad_norm": 1.3203125, "learning_rate": 3.0822879676058e-05, "loss": 0.1202, "step": 30989 }, { "epoch": 22.703296703296704, "grad_norm": 1.421875, "learning_rate": 3.0819735712671406e-05, "loss": 0.0935, "step": 30990 }, { "epoch": 22.704029304029305, "grad_norm": 0.9609375, "learning_rate": 3.081659184357941e-05, "loss": 0.1037, "step": 30991 }, { "epoch": 22.704761904761906, "grad_norm": 1.4140625, "learning_rate": 3.081344806879709e-05, "loss": 0.1334, "step": 30992 }, { "epoch": 22.705494505494507, "grad_norm": 0.875, "learning_rate": 3.08103043883395e-05, "loss": 0.1387, "step": 30993 }, { "epoch": 22.706227106227107, "grad_norm": 1.5078125, "learning_rate": 3.080716080222169e-05, "loss": 0.1348, "step": 30994 }, { "epoch": 22.706959706959708, "grad_norm": 1.2578125, "learning_rate": 3.080401731045874e-05, "loss": 0.1062, "step": 30995 }, { "epoch": 22.70769230769231, "grad_norm": 0.90625, "learning_rate": 3.080087391306573e-05, "loss": 0.12, "step": 30996 }, { "epoch": 22.70842490842491, "grad_norm": 1.4453125, "learning_rate": 3.079773061005768e-05, "loss": 0.1491, "step": 30997 }, { "epoch": 22.70915750915751, "grad_norm": 0.77734375, "learning_rate": 3.079458740144968e-05, "loss": 0.1301, "step": 30998 }, { "epoch": 22.70989010989011, "grad_norm": 0.8359375, "learning_rate": 3.0791444287256775e-05, "loss": 0.115, "step": 30999 }, { "epoch": 22.71062271062271, "grad_norm": 1.21875, "learning_rate": 3.078830126749403e-05, "loss": 0.1231, "step": 31000 }, { "epoch": 22.711355311355312, "grad_norm": 0.83984375, "learning_rate": 3.078515834217651e-05, "loss": 0.1441, "step": 31001 }, { "epoch": 22.712087912087913, "grad_norm": 1.1796875, "learning_rate": 3.078201551131927e-05, "loss": 0.1079, "step": 31002 }, { "epoch": 22.712820512820514, "grad_norm": 1.078125, "learning_rate": 3.0778872774937366e-05, "loss": 0.1364, "step": 31003 }, { "epoch": 22.713553113553115, "grad_norm": 1.1328125, "learning_rate": 3.0775730133045836e-05, "loss": 0.1482, "step": 31004 }, { "epoch": 22.714285714285715, "grad_norm": 1.5703125, "learning_rate": 3.077258758565978e-05, "loss": 0.1214, "step": 31005 }, { "epoch": 22.715018315018316, "grad_norm": 1.5703125, "learning_rate": 3.076944513279424e-05, "loss": 0.1025, "step": 31006 }, { "epoch": 22.715750915750917, "grad_norm": 0.75390625, "learning_rate": 3.0766302774464256e-05, "loss": 0.1154, "step": 31007 }, { "epoch": 22.716483516483517, "grad_norm": 0.9375, "learning_rate": 3.076316051068488e-05, "loss": 0.1191, "step": 31008 }, { "epoch": 22.717216117216118, "grad_norm": 0.9140625, "learning_rate": 3.0760018341471195e-05, "loss": 0.0986, "step": 31009 }, { "epoch": 22.71794871794872, "grad_norm": 0.921875, "learning_rate": 3.075687626683824e-05, "loss": 0.1003, "step": 31010 }, { "epoch": 22.71868131868132, "grad_norm": 1.21875, "learning_rate": 3.075373428680107e-05, "loss": 0.1043, "step": 31011 }, { "epoch": 22.71941391941392, "grad_norm": 1.2109375, "learning_rate": 3.075059240137474e-05, "loss": 0.1152, "step": 31012 }, { "epoch": 22.72014652014652, "grad_norm": 0.8203125, "learning_rate": 3.0747450610574315e-05, "loss": 0.1281, "step": 31013 }, { "epoch": 22.720879120879122, "grad_norm": 1.4765625, "learning_rate": 3.074430891441482e-05, "loss": 0.1299, "step": 31014 }, { "epoch": 22.721611721611723, "grad_norm": 0.88671875, "learning_rate": 3.074116731291134e-05, "loss": 0.1225, "step": 31015 }, { "epoch": 22.722344322344323, "grad_norm": 0.87109375, "learning_rate": 3.0738025806078916e-05, "loss": 0.1104, "step": 31016 }, { "epoch": 22.723076923076924, "grad_norm": 0.8359375, "learning_rate": 3.07348843939326e-05, "loss": 0.1283, "step": 31017 }, { "epoch": 22.723809523809525, "grad_norm": 1.8046875, "learning_rate": 3.073174307648744e-05, "loss": 0.1749, "step": 31018 }, { "epoch": 22.724542124542126, "grad_norm": 1.5390625, "learning_rate": 3.072860185375849e-05, "loss": 0.1208, "step": 31019 }, { "epoch": 22.725274725274726, "grad_norm": 1.109375, "learning_rate": 3.0725460725760767e-05, "loss": 0.1235, "step": 31020 }, { "epoch": 22.726007326007327, "grad_norm": 0.7734375, "learning_rate": 3.0722319692509385e-05, "loss": 0.0958, "step": 31021 }, { "epoch": 22.726739926739928, "grad_norm": 1.2890625, "learning_rate": 3.071917875401937e-05, "loss": 0.1519, "step": 31022 }, { "epoch": 22.72747252747253, "grad_norm": 1.484375, "learning_rate": 3.071603791030574e-05, "loss": 0.1354, "step": 31023 }, { "epoch": 22.72820512820513, "grad_norm": 1.4296875, "learning_rate": 3.071289716138359e-05, "loss": 0.1085, "step": 31024 }, { "epoch": 22.72893772893773, "grad_norm": 1.46875, "learning_rate": 3.070975650726791e-05, "loss": 0.161, "step": 31025 }, { "epoch": 22.72967032967033, "grad_norm": 1.1484375, "learning_rate": 3.0706615947973805e-05, "loss": 0.1396, "step": 31026 }, { "epoch": 22.73040293040293, "grad_norm": 0.734375, "learning_rate": 3.070347548351631e-05, "loss": 0.1096, "step": 31027 }, { "epoch": 22.731135531135532, "grad_norm": 1.53125, "learning_rate": 3.070033511391045e-05, "loss": 0.1378, "step": 31028 }, { "epoch": 22.731868131868133, "grad_norm": 0.90625, "learning_rate": 3.069719483917129e-05, "loss": 0.1256, "step": 31029 }, { "epoch": 22.732600732600734, "grad_norm": 0.96875, "learning_rate": 3.069405465931386e-05, "loss": 0.1205, "step": 31030 }, { "epoch": 22.733333333333334, "grad_norm": 1.078125, "learning_rate": 3.069091457435323e-05, "loss": 0.1117, "step": 31031 }, { "epoch": 22.734065934065935, "grad_norm": 1.671875, "learning_rate": 3.0687774584304425e-05, "loss": 0.1287, "step": 31032 }, { "epoch": 22.734798534798536, "grad_norm": 0.75, "learning_rate": 3.0684634689182503e-05, "loss": 0.1028, "step": 31033 }, { "epoch": 22.735531135531136, "grad_norm": 1.7578125, "learning_rate": 3.068149488900248e-05, "loss": 0.1166, "step": 31034 }, { "epoch": 22.736263736263737, "grad_norm": 1.203125, "learning_rate": 3.067835518377943e-05, "loss": 0.1172, "step": 31035 }, { "epoch": 22.736996336996338, "grad_norm": 0.87109375, "learning_rate": 3.0675215573528396e-05, "loss": 0.1201, "step": 31036 }, { "epoch": 22.73772893772894, "grad_norm": 1.375, "learning_rate": 3.067207605826441e-05, "loss": 0.1167, "step": 31037 }, { "epoch": 22.73846153846154, "grad_norm": 1.046875, "learning_rate": 3.0668936638002515e-05, "loss": 0.1354, "step": 31038 }, { "epoch": 22.73919413919414, "grad_norm": 0.9296875, "learning_rate": 3.066579731275777e-05, "loss": 0.1178, "step": 31039 }, { "epoch": 22.73992673992674, "grad_norm": 0.7890625, "learning_rate": 3.066265808254516e-05, "loss": 0.1204, "step": 31040 }, { "epoch": 22.74065934065934, "grad_norm": 0.9921875, "learning_rate": 3.06595189473798e-05, "loss": 0.1194, "step": 31041 }, { "epoch": 22.741391941391942, "grad_norm": 1.1953125, "learning_rate": 3.065637990727669e-05, "loss": 0.1213, "step": 31042 }, { "epoch": 22.742124542124543, "grad_norm": 0.8359375, "learning_rate": 3.065324096225089e-05, "loss": 0.1131, "step": 31043 }, { "epoch": 22.742857142857144, "grad_norm": 0.87109375, "learning_rate": 3.065010211231742e-05, "loss": 0.136, "step": 31044 }, { "epoch": 22.743589743589745, "grad_norm": 0.796875, "learning_rate": 3.0646963357491325e-05, "loss": 0.1163, "step": 31045 }, { "epoch": 22.744322344322345, "grad_norm": 1.2578125, "learning_rate": 3.064382469778763e-05, "loss": 0.1167, "step": 31046 }, { "epoch": 22.745054945054946, "grad_norm": 1.3359375, "learning_rate": 3.064068613322141e-05, "loss": 0.113, "step": 31047 }, { "epoch": 22.745787545787547, "grad_norm": 1.3828125, "learning_rate": 3.063754766380768e-05, "loss": 0.1461, "step": 31048 }, { "epoch": 22.746520146520147, "grad_norm": 0.95703125, "learning_rate": 3.0634409289561476e-05, "loss": 0.1167, "step": 31049 }, { "epoch": 22.747252747252748, "grad_norm": 1.0390625, "learning_rate": 3.063127101049784e-05, "loss": 0.114, "step": 31050 }, { "epoch": 22.74798534798535, "grad_norm": 1.0234375, "learning_rate": 3.0628132826631786e-05, "loss": 0.1228, "step": 31051 }, { "epoch": 22.74871794871795, "grad_norm": 1.046875, "learning_rate": 3.062499473797839e-05, "loss": 0.1134, "step": 31052 }, { "epoch": 22.74945054945055, "grad_norm": 1.0625, "learning_rate": 3.0621856744552686e-05, "loss": 0.1203, "step": 31053 }, { "epoch": 22.75018315018315, "grad_norm": 1.3125, "learning_rate": 3.061871884636968e-05, "loss": 0.1067, "step": 31054 }, { "epoch": 22.750915750915752, "grad_norm": 1.8203125, "learning_rate": 3.0615581043444415e-05, "loss": 0.1073, "step": 31055 }, { "epoch": 22.751648351648353, "grad_norm": 0.82421875, "learning_rate": 3.061244333579191e-05, "loss": 0.1019, "step": 31056 }, { "epoch": 22.752380952380953, "grad_norm": 1.546875, "learning_rate": 3.0609305723427244e-05, "loss": 0.1436, "step": 31057 }, { "epoch": 22.753113553113554, "grad_norm": 0.875, "learning_rate": 3.060616820636542e-05, "loss": 0.1312, "step": 31058 }, { "epoch": 22.753846153846155, "grad_norm": 1.4765625, "learning_rate": 3.060303078462148e-05, "loss": 0.1543, "step": 31059 }, { "epoch": 22.754578754578755, "grad_norm": 1.0625, "learning_rate": 3.059989345821043e-05, "loss": 0.119, "step": 31060 }, { "epoch": 22.755311355311356, "grad_norm": 0.8828125, "learning_rate": 3.059675622714733e-05, "loss": 0.1006, "step": 31061 }, { "epoch": 22.756043956043957, "grad_norm": 0.83203125, "learning_rate": 3.059361909144722e-05, "loss": 0.126, "step": 31062 }, { "epoch": 22.756776556776558, "grad_norm": 0.9453125, "learning_rate": 3.0590482051125095e-05, "loss": 0.1176, "step": 31063 }, { "epoch": 22.75750915750916, "grad_norm": 1.078125, "learning_rate": 3.058734510619602e-05, "loss": 0.1162, "step": 31064 }, { "epoch": 22.75824175824176, "grad_norm": 1.015625, "learning_rate": 3.058420825667502e-05, "loss": 0.1226, "step": 31065 }, { "epoch": 22.75897435897436, "grad_norm": 0.96875, "learning_rate": 3.0581071502577084e-05, "loss": 0.1184, "step": 31066 }, { "epoch": 22.75970695970696, "grad_norm": 0.70703125, "learning_rate": 3.057793484391729e-05, "loss": 0.0967, "step": 31067 }, { "epoch": 22.76043956043956, "grad_norm": 1.8046875, "learning_rate": 3.057479828071067e-05, "loss": 0.1674, "step": 31068 }, { "epoch": 22.761172161172162, "grad_norm": 1.296875, "learning_rate": 3.057166181297221e-05, "loss": 0.1457, "step": 31069 }, { "epoch": 22.761904761904763, "grad_norm": 1.21875, "learning_rate": 3.056852544071698e-05, "loss": 0.1415, "step": 31070 }, { "epoch": 22.762637362637363, "grad_norm": 1.171875, "learning_rate": 3.056538916395997e-05, "loss": 0.1425, "step": 31071 }, { "epoch": 22.763369963369964, "grad_norm": 1.203125, "learning_rate": 3.056225298271621e-05, "loss": 0.1267, "step": 31072 }, { "epoch": 22.764102564102565, "grad_norm": 0.64453125, "learning_rate": 3.055911689700077e-05, "loss": 0.1362, "step": 31073 }, { "epoch": 22.764835164835166, "grad_norm": 0.734375, "learning_rate": 3.055598090682864e-05, "loss": 0.1144, "step": 31074 }, { "epoch": 22.765567765567766, "grad_norm": 1.765625, "learning_rate": 3.0552845012214845e-05, "loss": 0.1271, "step": 31075 }, { "epoch": 22.766300366300367, "grad_norm": 0.95703125, "learning_rate": 3.054970921317443e-05, "loss": 0.1182, "step": 31076 }, { "epoch": 22.767032967032968, "grad_norm": 1.21875, "learning_rate": 3.054657350972239e-05, "loss": 0.1267, "step": 31077 }, { "epoch": 22.76776556776557, "grad_norm": 1.34375, "learning_rate": 3.0543437901873774e-05, "loss": 0.1108, "step": 31078 }, { "epoch": 22.76849816849817, "grad_norm": 1.140625, "learning_rate": 3.054030238964361e-05, "loss": 0.1172, "step": 31079 }, { "epoch": 22.76923076923077, "grad_norm": 1.6640625, "learning_rate": 3.0537166973046896e-05, "loss": 0.1005, "step": 31080 }, { "epoch": 22.76996336996337, "grad_norm": 0.89453125, "learning_rate": 3.053403165209868e-05, "loss": 0.1403, "step": 31081 }, { "epoch": 22.77069597069597, "grad_norm": 1.0859375, "learning_rate": 3.053089642681395e-05, "loss": 0.1453, "step": 31082 }, { "epoch": 22.771428571428572, "grad_norm": 1.109375, "learning_rate": 3.052776129720777e-05, "loss": 0.1531, "step": 31083 }, { "epoch": 22.772161172161173, "grad_norm": 0.70703125, "learning_rate": 3.052462626329515e-05, "loss": 0.0967, "step": 31084 }, { "epoch": 22.772893772893774, "grad_norm": 0.7421875, "learning_rate": 3.0521491325091094e-05, "loss": 0.1182, "step": 31085 }, { "epoch": 22.773626373626374, "grad_norm": 1.1953125, "learning_rate": 3.051835648261063e-05, "loss": 0.1215, "step": 31086 }, { "epoch": 22.774358974358975, "grad_norm": 1.1953125, "learning_rate": 3.0515221735868766e-05, "loss": 0.1483, "step": 31087 }, { "epoch": 22.775091575091576, "grad_norm": 1.0234375, "learning_rate": 3.051208708488055e-05, "loss": 0.1133, "step": 31088 }, { "epoch": 22.775824175824177, "grad_norm": 1.21875, "learning_rate": 3.0508952529660982e-05, "loss": 0.135, "step": 31089 }, { "epoch": 22.776556776556777, "grad_norm": 1.046875, "learning_rate": 3.050581807022509e-05, "loss": 0.1126, "step": 31090 }, { "epoch": 22.777289377289378, "grad_norm": 0.83203125, "learning_rate": 3.0502683706587894e-05, "loss": 0.1284, "step": 31091 }, { "epoch": 22.77802197802198, "grad_norm": 1.21875, "learning_rate": 3.0499549438764373e-05, "loss": 0.1253, "step": 31092 }, { "epoch": 22.77875457875458, "grad_norm": 0.99609375, "learning_rate": 3.0496415266769607e-05, "loss": 0.1299, "step": 31093 }, { "epoch": 22.77948717948718, "grad_norm": 0.9921875, "learning_rate": 3.0493281190618578e-05, "loss": 0.1197, "step": 31094 }, { "epoch": 22.78021978021978, "grad_norm": 1.1640625, "learning_rate": 3.0490147210326307e-05, "loss": 0.1146, "step": 31095 }, { "epoch": 22.78095238095238, "grad_norm": 0.70703125, "learning_rate": 3.048701332590781e-05, "loss": 0.113, "step": 31096 }, { "epoch": 22.781684981684982, "grad_norm": 0.60546875, "learning_rate": 3.0483879537378102e-05, "loss": 0.1227, "step": 31097 }, { "epoch": 22.782417582417583, "grad_norm": 0.859375, "learning_rate": 3.048074584475218e-05, "loss": 0.128, "step": 31098 }, { "epoch": 22.783150183150184, "grad_norm": 0.921875, "learning_rate": 3.0477612248045088e-05, "loss": 0.1371, "step": 31099 }, { "epoch": 22.783882783882785, "grad_norm": 1.3515625, "learning_rate": 3.047447874727183e-05, "loss": 0.1187, "step": 31100 }, { "epoch": 22.784615384615385, "grad_norm": 1.3984375, "learning_rate": 3.0471345342447422e-05, "loss": 0.1481, "step": 31101 }, { "epoch": 22.785347985347986, "grad_norm": 1.0546875, "learning_rate": 3.0468212033586873e-05, "loss": 0.1639, "step": 31102 }, { "epoch": 22.786080586080587, "grad_norm": 1.25, "learning_rate": 3.0465078820705185e-05, "loss": 0.1195, "step": 31103 }, { "epoch": 22.786813186813188, "grad_norm": 1.4296875, "learning_rate": 3.046194570381739e-05, "loss": 0.1256, "step": 31104 }, { "epoch": 22.78754578754579, "grad_norm": 1.2890625, "learning_rate": 3.045881268293849e-05, "loss": 0.1213, "step": 31105 }, { "epoch": 22.78827838827839, "grad_norm": 1.0078125, "learning_rate": 3.0455679758083507e-05, "loss": 0.1275, "step": 31106 }, { "epoch": 22.78901098901099, "grad_norm": 0.93359375, "learning_rate": 3.045254692926744e-05, "loss": 0.1555, "step": 31107 }, { "epoch": 22.78974358974359, "grad_norm": 0.8984375, "learning_rate": 3.0449414196505276e-05, "loss": 0.1258, "step": 31108 }, { "epoch": 22.79047619047619, "grad_norm": 0.98046875, "learning_rate": 3.044628155981208e-05, "loss": 0.1057, "step": 31109 }, { "epoch": 22.791208791208792, "grad_norm": 1.21875, "learning_rate": 3.0443149019202827e-05, "loss": 0.1044, "step": 31110 }, { "epoch": 22.791941391941393, "grad_norm": 0.875, "learning_rate": 3.0440016574692533e-05, "loss": 0.106, "step": 31111 }, { "epoch": 22.792673992673993, "grad_norm": 1.2421875, "learning_rate": 3.043688422629619e-05, "loss": 0.1589, "step": 31112 }, { "epoch": 22.793406593406594, "grad_norm": 0.875, "learning_rate": 3.0433751974028823e-05, "loss": 0.1364, "step": 31113 }, { "epoch": 22.794139194139195, "grad_norm": 1.15625, "learning_rate": 3.0430619817905454e-05, "loss": 0.117, "step": 31114 }, { "epoch": 22.794871794871796, "grad_norm": 1.0390625, "learning_rate": 3.0427487757941054e-05, "loss": 0.1321, "step": 31115 }, { "epoch": 22.795604395604396, "grad_norm": 0.96484375, "learning_rate": 3.0424355794150668e-05, "loss": 0.1123, "step": 31116 }, { "epoch": 22.796336996336997, "grad_norm": 0.83984375, "learning_rate": 3.0421223926549275e-05, "loss": 0.1245, "step": 31117 }, { "epoch": 22.797069597069598, "grad_norm": 1.46875, "learning_rate": 3.041809215515188e-05, "loss": 0.154, "step": 31118 }, { "epoch": 22.7978021978022, "grad_norm": 1.4140625, "learning_rate": 3.0414960479973513e-05, "loss": 0.1611, "step": 31119 }, { "epoch": 22.7985347985348, "grad_norm": 1.0234375, "learning_rate": 3.041182890102916e-05, "loss": 0.1099, "step": 31120 }, { "epoch": 22.7992673992674, "grad_norm": 1.484375, "learning_rate": 3.0408697418333832e-05, "loss": 0.1498, "step": 31121 }, { "epoch": 22.8, "grad_norm": 1.4296875, "learning_rate": 3.0405566031902536e-05, "loss": 0.1344, "step": 31122 }, { "epoch": 22.8007326007326, "grad_norm": 1.0859375, "learning_rate": 3.0402434741750247e-05, "loss": 0.1145, "step": 31123 }, { "epoch": 22.801465201465202, "grad_norm": 1.578125, "learning_rate": 3.0399303547892e-05, "loss": 0.1569, "step": 31124 }, { "epoch": 22.802197802197803, "grad_norm": 1.5859375, "learning_rate": 3.0396172450342795e-05, "loss": 0.1035, "step": 31125 }, { "epoch": 22.802930402930404, "grad_norm": 1.5390625, "learning_rate": 3.039304144911762e-05, "loss": 0.1013, "step": 31126 }, { "epoch": 22.803663003663004, "grad_norm": 1.109375, "learning_rate": 3.0389910544231492e-05, "loss": 0.1205, "step": 31127 }, { "epoch": 22.804395604395605, "grad_norm": 0.80078125, "learning_rate": 3.0386779735699412e-05, "loss": 0.1068, "step": 31128 }, { "epoch": 22.805128205128206, "grad_norm": 0.93359375, "learning_rate": 3.038364902353634e-05, "loss": 0.1107, "step": 31129 }, { "epoch": 22.805860805860807, "grad_norm": 1.046875, "learning_rate": 3.0380518407757342e-05, "loss": 0.1501, "step": 31130 }, { "epoch": 22.806593406593407, "grad_norm": 0.94140625, "learning_rate": 3.0377387888377374e-05, "loss": 0.1222, "step": 31131 }, { "epoch": 22.807326007326008, "grad_norm": 1.265625, "learning_rate": 3.0374257465411456e-05, "loss": 0.1133, "step": 31132 }, { "epoch": 22.80805860805861, "grad_norm": 1.609375, "learning_rate": 3.0371127138874574e-05, "loss": 0.123, "step": 31133 }, { "epoch": 22.80879120879121, "grad_norm": 0.98828125, "learning_rate": 3.036799690878171e-05, "loss": 0.1314, "step": 31134 }, { "epoch": 22.80952380952381, "grad_norm": 1.5390625, "learning_rate": 3.03648667751479e-05, "loss": 0.1572, "step": 31135 }, { "epoch": 22.81025641025641, "grad_norm": 0.82421875, "learning_rate": 3.0361736737988117e-05, "loss": 0.1183, "step": 31136 }, { "epoch": 22.81098901098901, "grad_norm": 0.70703125, "learning_rate": 3.035860679731737e-05, "loss": 0.115, "step": 31137 }, { "epoch": 22.811721611721612, "grad_norm": 0.63671875, "learning_rate": 3.035547695315064e-05, "loss": 0.1142, "step": 31138 }, { "epoch": 22.812454212454213, "grad_norm": 1.421875, "learning_rate": 3.0352347205502935e-05, "loss": 0.1295, "step": 31139 }, { "epoch": 22.813186813186814, "grad_norm": 1.3046875, "learning_rate": 3.0349217554389255e-05, "loss": 0.0986, "step": 31140 }, { "epoch": 22.813919413919415, "grad_norm": 0.859375, "learning_rate": 3.034608799982458e-05, "loss": 0.1251, "step": 31141 }, { "epoch": 22.814652014652015, "grad_norm": 1.5546875, "learning_rate": 3.034295854182393e-05, "loss": 0.1498, "step": 31142 }, { "epoch": 22.815384615384616, "grad_norm": 0.921875, "learning_rate": 3.0339829180402273e-05, "loss": 0.1228, "step": 31143 }, { "epoch": 22.816117216117217, "grad_norm": 1.4140625, "learning_rate": 3.0336699915574617e-05, "loss": 0.1524, "step": 31144 }, { "epoch": 22.816849816849818, "grad_norm": 0.8671875, "learning_rate": 3.0333570747355938e-05, "loss": 0.0965, "step": 31145 }, { "epoch": 22.81758241758242, "grad_norm": 1.3203125, "learning_rate": 3.033044167576125e-05, "loss": 0.1024, "step": 31146 }, { "epoch": 22.81831501831502, "grad_norm": 1.234375, "learning_rate": 3.032731270080554e-05, "loss": 0.1664, "step": 31147 }, { "epoch": 22.81904761904762, "grad_norm": 0.640625, "learning_rate": 3.03241838225038e-05, "loss": 0.1215, "step": 31148 }, { "epoch": 22.81978021978022, "grad_norm": 0.85546875, "learning_rate": 3.0321055040871004e-05, "loss": 0.1314, "step": 31149 }, { "epoch": 22.82051282051282, "grad_norm": 1.703125, "learning_rate": 3.031792635592216e-05, "loss": 0.1337, "step": 31150 }, { "epoch": 22.821245421245422, "grad_norm": 1.1875, "learning_rate": 3.0314797767672264e-05, "loss": 0.1339, "step": 31151 }, { "epoch": 22.821978021978023, "grad_norm": 1.0390625, "learning_rate": 3.03116692761363e-05, "loss": 0.1257, "step": 31152 }, { "epoch": 22.822710622710623, "grad_norm": 1.203125, "learning_rate": 3.030854088132925e-05, "loss": 0.1416, "step": 31153 }, { "epoch": 22.823443223443224, "grad_norm": 0.921875, "learning_rate": 3.0305412583266117e-05, "loss": 0.1272, "step": 31154 }, { "epoch": 22.824175824175825, "grad_norm": 1.140625, "learning_rate": 3.030228438196186e-05, "loss": 0.1372, "step": 31155 }, { "epoch": 22.824908424908426, "grad_norm": 1.265625, "learning_rate": 3.0299156277431513e-05, "loss": 0.1153, "step": 31156 }, { "epoch": 22.825641025641026, "grad_norm": 1.359375, "learning_rate": 3.0296028269690035e-05, "loss": 0.1095, "step": 31157 }, { "epoch": 22.826373626373627, "grad_norm": 1.03125, "learning_rate": 3.0292900358752425e-05, "loss": 0.159, "step": 31158 }, { "epoch": 22.827106227106228, "grad_norm": 0.93359375, "learning_rate": 3.0289772544633653e-05, "loss": 0.1074, "step": 31159 }, { "epoch": 22.82783882783883, "grad_norm": 1.0078125, "learning_rate": 3.0286644827348697e-05, "loss": 0.1041, "step": 31160 }, { "epoch": 22.82857142857143, "grad_norm": 1.0078125, "learning_rate": 3.0283517206912587e-05, "loss": 0.0992, "step": 31161 }, { "epoch": 22.82930402930403, "grad_norm": 0.73828125, "learning_rate": 3.028038968334028e-05, "loss": 0.1319, "step": 31162 }, { "epoch": 22.83003663003663, "grad_norm": 1.1953125, "learning_rate": 3.027726225664676e-05, "loss": 0.1072, "step": 31163 }, { "epoch": 22.83076923076923, "grad_norm": 1.0546875, "learning_rate": 3.0274134926847014e-05, "loss": 0.1126, "step": 31164 }, { "epoch": 22.831501831501832, "grad_norm": 1.2109375, "learning_rate": 3.0271007693956023e-05, "loss": 0.1056, "step": 31165 }, { "epoch": 22.832234432234433, "grad_norm": 0.9765625, "learning_rate": 3.026788055798877e-05, "loss": 0.1086, "step": 31166 }, { "epoch": 22.832967032967034, "grad_norm": 1.1328125, "learning_rate": 3.0264753518960258e-05, "loss": 0.1613, "step": 31167 }, { "epoch": 22.833699633699634, "grad_norm": 1.328125, "learning_rate": 3.0261626576885456e-05, "loss": 0.1581, "step": 31168 }, { "epoch": 22.834432234432235, "grad_norm": 0.4765625, "learning_rate": 3.0258499731779338e-05, "loss": 0.0894, "step": 31169 }, { "epoch": 22.835164835164836, "grad_norm": 1.609375, "learning_rate": 3.02553729836569e-05, "loss": 0.1344, "step": 31170 }, { "epoch": 22.835897435897436, "grad_norm": 1.234375, "learning_rate": 3.0252246332533098e-05, "loss": 0.126, "step": 31171 }, { "epoch": 22.836630036630037, "grad_norm": 0.81640625, "learning_rate": 3.024911977842294e-05, "loss": 0.1345, "step": 31172 }, { "epoch": 22.837362637362638, "grad_norm": 0.8125, "learning_rate": 3.0245993321341402e-05, "loss": 0.1249, "step": 31173 }, { "epoch": 22.83809523809524, "grad_norm": 0.85546875, "learning_rate": 3.0242866961303457e-05, "loss": 0.1042, "step": 31174 }, { "epoch": 22.83882783882784, "grad_norm": 1.109375, "learning_rate": 3.0239740698324083e-05, "loss": 0.112, "step": 31175 }, { "epoch": 22.83956043956044, "grad_norm": 0.953125, "learning_rate": 3.0236614532418256e-05, "loss": 0.1266, "step": 31176 }, { "epoch": 22.84029304029304, "grad_norm": 0.90234375, "learning_rate": 3.0233488463600975e-05, "loss": 0.1181, "step": 31177 }, { "epoch": 22.84102564102564, "grad_norm": 1.4296875, "learning_rate": 3.023036249188719e-05, "loss": 0.1327, "step": 31178 }, { "epoch": 22.841758241758242, "grad_norm": 0.9765625, "learning_rate": 3.0227236617291906e-05, "loss": 0.1197, "step": 31179 }, { "epoch": 22.842490842490843, "grad_norm": 1.1171875, "learning_rate": 3.0224110839830082e-05, "loss": 0.1286, "step": 31180 }, { "epoch": 22.843223443223444, "grad_norm": 0.94921875, "learning_rate": 3.022098515951669e-05, "loss": 0.1019, "step": 31181 }, { "epoch": 22.843956043956045, "grad_norm": 1.0234375, "learning_rate": 3.0217859576366725e-05, "loss": 0.1228, "step": 31182 }, { "epoch": 22.844688644688645, "grad_norm": 1.0390625, "learning_rate": 3.0214734090395156e-05, "loss": 0.1055, "step": 31183 }, { "epoch": 22.845421245421246, "grad_norm": 0.703125, "learning_rate": 3.021160870161695e-05, "loss": 0.0741, "step": 31184 }, { "epoch": 22.846153846153847, "grad_norm": 1.234375, "learning_rate": 3.020848341004709e-05, "loss": 0.1643, "step": 31185 }, { "epoch": 22.846886446886447, "grad_norm": 0.6796875, "learning_rate": 3.0205358215700537e-05, "loss": 0.1062, "step": 31186 }, { "epoch": 22.847619047619048, "grad_norm": 0.6875, "learning_rate": 3.0202233118592284e-05, "loss": 0.0849, "step": 31187 }, { "epoch": 22.84835164835165, "grad_norm": 1.21875, "learning_rate": 3.01991081187373e-05, "loss": 0.145, "step": 31188 }, { "epoch": 22.84908424908425, "grad_norm": 0.68359375, "learning_rate": 3.0195983216150537e-05, "loss": 0.1134, "step": 31189 }, { "epoch": 22.84981684981685, "grad_norm": 0.8828125, "learning_rate": 3.0192858410846997e-05, "loss": 0.1163, "step": 31190 }, { "epoch": 22.85054945054945, "grad_norm": 1.3125, "learning_rate": 3.018973370284163e-05, "loss": 0.151, "step": 31191 }, { "epoch": 22.851282051282052, "grad_norm": 1.0703125, "learning_rate": 3.018660909214942e-05, "loss": 0.1398, "step": 31192 }, { "epoch": 22.852014652014653, "grad_norm": 1.3203125, "learning_rate": 3.0183484578785337e-05, "loss": 0.1353, "step": 31193 }, { "epoch": 22.852747252747253, "grad_norm": 0.85546875, "learning_rate": 3.0180360162764354e-05, "loss": 0.1159, "step": 31194 }, { "epoch": 22.853479853479854, "grad_norm": 1.7578125, "learning_rate": 3.0177235844101443e-05, "loss": 0.1374, "step": 31195 }, { "epoch": 22.854212454212455, "grad_norm": 1.0546875, "learning_rate": 3.0174111622811545e-05, "loss": 0.1193, "step": 31196 }, { "epoch": 22.854945054945055, "grad_norm": 0.78515625, "learning_rate": 3.0170987498909647e-05, "loss": 0.126, "step": 31197 }, { "epoch": 22.855677655677656, "grad_norm": 0.6953125, "learning_rate": 3.0167863472410744e-05, "loss": 0.1152, "step": 31198 }, { "epoch": 22.856410256410257, "grad_norm": 1.8125, "learning_rate": 3.0164739543329775e-05, "loss": 0.1281, "step": 31199 }, { "epoch": 22.857142857142858, "grad_norm": 1.3125, "learning_rate": 3.0161615711681718e-05, "loss": 0.104, "step": 31200 }, { "epoch": 22.85787545787546, "grad_norm": 1.390625, "learning_rate": 3.0158491977481518e-05, "loss": 0.115, "step": 31201 }, { "epoch": 22.85860805860806, "grad_norm": 0.890625, "learning_rate": 3.0155368340744165e-05, "loss": 0.1286, "step": 31202 }, { "epoch": 22.85934065934066, "grad_norm": 0.80078125, "learning_rate": 3.0152244801484634e-05, "loss": 0.0984, "step": 31203 }, { "epoch": 22.86007326007326, "grad_norm": 1.3046875, "learning_rate": 3.014912135971787e-05, "loss": 0.1012, "step": 31204 }, { "epoch": 22.86080586080586, "grad_norm": 1.4765625, "learning_rate": 3.014599801545886e-05, "loss": 0.1024, "step": 31205 }, { "epoch": 22.861538461538462, "grad_norm": 0.828125, "learning_rate": 3.014287476872254e-05, "loss": 0.118, "step": 31206 }, { "epoch": 22.862271062271063, "grad_norm": 1.0703125, "learning_rate": 3.0139751619523877e-05, "loss": 0.1405, "step": 31207 }, { "epoch": 22.863003663003663, "grad_norm": 1.1171875, "learning_rate": 3.013662856787788e-05, "loss": 0.1103, "step": 31208 }, { "epoch": 22.863736263736264, "grad_norm": 1.5234375, "learning_rate": 3.0133505613799463e-05, "loss": 0.1417, "step": 31209 }, { "epoch": 22.864468864468865, "grad_norm": 1.2578125, "learning_rate": 3.0130382757303616e-05, "loss": 0.1014, "step": 31210 }, { "epoch": 22.865201465201466, "grad_norm": 1.5234375, "learning_rate": 3.0127259998405284e-05, "loss": 0.1224, "step": 31211 }, { "epoch": 22.865934065934066, "grad_norm": 1.3984375, "learning_rate": 3.0124137337119428e-05, "loss": 0.1452, "step": 31212 }, { "epoch": 22.866666666666667, "grad_norm": 1.375, "learning_rate": 3.012101477346103e-05, "loss": 0.1448, "step": 31213 }, { "epoch": 22.867399267399268, "grad_norm": 1.03125, "learning_rate": 3.0117892307445045e-05, "loss": 0.1452, "step": 31214 }, { "epoch": 22.86813186813187, "grad_norm": 1.0859375, "learning_rate": 3.011476993908641e-05, "loss": 0.1194, "step": 31215 }, { "epoch": 22.86886446886447, "grad_norm": 1.9921875, "learning_rate": 3.011164766840013e-05, "loss": 0.1811, "step": 31216 }, { "epoch": 22.86959706959707, "grad_norm": 1.8359375, "learning_rate": 3.0108525495401114e-05, "loss": 0.1764, "step": 31217 }, { "epoch": 22.87032967032967, "grad_norm": 1.5625, "learning_rate": 3.0105403420104357e-05, "loss": 0.1212, "step": 31218 }, { "epoch": 22.87106227106227, "grad_norm": 0.8046875, "learning_rate": 3.010228144252481e-05, "loss": 0.1348, "step": 31219 }, { "epoch": 22.871794871794872, "grad_norm": 1.1640625, "learning_rate": 3.0099159562677437e-05, "loss": 0.1056, "step": 31220 }, { "epoch": 22.872527472527473, "grad_norm": 1.296875, "learning_rate": 3.009603778057718e-05, "loss": 0.1318, "step": 31221 }, { "epoch": 22.873260073260074, "grad_norm": 1.1328125, "learning_rate": 3.009291609623901e-05, "loss": 0.1104, "step": 31222 }, { "epoch": 22.873992673992674, "grad_norm": 0.55859375, "learning_rate": 3.0089794509677856e-05, "loss": 0.0963, "step": 31223 }, { "epoch": 22.874725274725275, "grad_norm": 1.171875, "learning_rate": 3.008667302090871e-05, "loss": 0.1159, "step": 31224 }, { "epoch": 22.875457875457876, "grad_norm": 0.72265625, "learning_rate": 3.008355162994652e-05, "loss": 0.1046, "step": 31225 }, { "epoch": 22.876190476190477, "grad_norm": 1.0703125, "learning_rate": 3.008043033680625e-05, "loss": 0.1261, "step": 31226 }, { "epoch": 22.876923076923077, "grad_norm": 0.9765625, "learning_rate": 3.0077309141502812e-05, "loss": 0.1263, "step": 31227 }, { "epoch": 22.877655677655678, "grad_norm": 0.84375, "learning_rate": 3.0074188044051193e-05, "loss": 0.1086, "step": 31228 }, { "epoch": 22.87838827838828, "grad_norm": 0.99609375, "learning_rate": 3.007106704446636e-05, "loss": 0.1036, "step": 31229 }, { "epoch": 22.87912087912088, "grad_norm": 1.21875, "learning_rate": 3.0067946142763236e-05, "loss": 0.1262, "step": 31230 }, { "epoch": 22.87985347985348, "grad_norm": 1.8359375, "learning_rate": 3.0064825338956804e-05, "loss": 0.1408, "step": 31231 }, { "epoch": 22.88058608058608, "grad_norm": 1.59375, "learning_rate": 3.006170463306201e-05, "loss": 0.145, "step": 31232 }, { "epoch": 22.88131868131868, "grad_norm": 0.796875, "learning_rate": 3.0058584025093768e-05, "loss": 0.1142, "step": 31233 }, { "epoch": 22.882051282051282, "grad_norm": 1.1484375, "learning_rate": 3.0055463515067077e-05, "loss": 0.1235, "step": 31234 }, { "epoch": 22.882783882783883, "grad_norm": 0.7890625, "learning_rate": 3.0052343102996877e-05, "loss": 0.1076, "step": 31235 }, { "epoch": 22.883516483516484, "grad_norm": 0.828125, "learning_rate": 3.004922278889812e-05, "loss": 0.1269, "step": 31236 }, { "epoch": 22.884249084249085, "grad_norm": 1.203125, "learning_rate": 3.0046102572785738e-05, "loss": 0.1401, "step": 31237 }, { "epoch": 22.884981684981685, "grad_norm": 1.2109375, "learning_rate": 3.0042982454674683e-05, "loss": 0.1215, "step": 31238 }, { "epoch": 22.885714285714286, "grad_norm": 1.578125, "learning_rate": 3.003986243457993e-05, "loss": 0.1576, "step": 31239 }, { "epoch": 22.886446886446887, "grad_norm": 0.95703125, "learning_rate": 3.0036742512516412e-05, "loss": 0.0982, "step": 31240 }, { "epoch": 22.887179487179488, "grad_norm": 1.078125, "learning_rate": 3.003362268849907e-05, "loss": 0.1196, "step": 31241 }, { "epoch": 22.88791208791209, "grad_norm": 1.8828125, "learning_rate": 3.0030502962542873e-05, "loss": 0.1254, "step": 31242 }, { "epoch": 22.88864468864469, "grad_norm": 1.203125, "learning_rate": 3.002738333466274e-05, "loss": 0.117, "step": 31243 }, { "epoch": 22.88937728937729, "grad_norm": 0.8515625, "learning_rate": 3.0024263804873643e-05, "loss": 0.103, "step": 31244 }, { "epoch": 22.89010989010989, "grad_norm": 1.1875, "learning_rate": 3.002114437319052e-05, "loss": 0.0907, "step": 31245 }, { "epoch": 22.89084249084249, "grad_norm": 1.0859375, "learning_rate": 3.0018025039628323e-05, "loss": 0.1067, "step": 31246 }, { "epoch": 22.891575091575092, "grad_norm": 1.1015625, "learning_rate": 3.0014905804201988e-05, "loss": 0.1062, "step": 31247 }, { "epoch": 22.892307692307693, "grad_norm": 1.03125, "learning_rate": 3.0011786666926466e-05, "loss": 0.1097, "step": 31248 }, { "epoch": 22.893040293040293, "grad_norm": 1.4140625, "learning_rate": 3.0008667627816683e-05, "loss": 0.1301, "step": 31249 }, { "epoch": 22.893772893772894, "grad_norm": 1.953125, "learning_rate": 3.000554868688762e-05, "loss": 0.1331, "step": 31250 }, { "epoch": 22.894505494505495, "grad_norm": 0.859375, "learning_rate": 3.000242984415419e-05, "loss": 0.0923, "step": 31251 }, { "epoch": 22.895238095238096, "grad_norm": 1.28125, "learning_rate": 2.9999311099631362e-05, "loss": 0.1348, "step": 31252 }, { "epoch": 22.895970695970696, "grad_norm": 1.265625, "learning_rate": 2.999619245333405e-05, "loss": 0.0922, "step": 31253 }, { "epoch": 22.896703296703297, "grad_norm": 1.296875, "learning_rate": 2.99930739052772e-05, "loss": 0.1123, "step": 31254 }, { "epoch": 22.897435897435898, "grad_norm": 1.21875, "learning_rate": 2.9989955455475778e-05, "loss": 0.1314, "step": 31255 }, { "epoch": 22.8981684981685, "grad_norm": 0.91015625, "learning_rate": 2.998683710394472e-05, "loss": 0.1004, "step": 31256 }, { "epoch": 22.8989010989011, "grad_norm": 1.7578125, "learning_rate": 2.998371885069896e-05, "loss": 0.1356, "step": 31257 }, { "epoch": 22.8996336996337, "grad_norm": 1.0625, "learning_rate": 2.9980600695753434e-05, "loss": 0.1167, "step": 31258 }, { "epoch": 22.9003663003663, "grad_norm": 1.0703125, "learning_rate": 2.9977482639123065e-05, "loss": 0.1094, "step": 31259 }, { "epoch": 22.9010989010989, "grad_norm": 0.73046875, "learning_rate": 2.997436468082284e-05, "loss": 0.0856, "step": 31260 }, { "epoch": 22.901831501831502, "grad_norm": 0.97265625, "learning_rate": 2.997124682086767e-05, "loss": 0.103, "step": 31261 }, { "epoch": 22.902564102564103, "grad_norm": 1.3125, "learning_rate": 2.9968129059272497e-05, "loss": 0.1326, "step": 31262 }, { "epoch": 22.903296703296704, "grad_norm": 1.28125, "learning_rate": 2.9965011396052256e-05, "loss": 0.1221, "step": 31263 }, { "epoch": 22.904029304029304, "grad_norm": 0.765625, "learning_rate": 2.9961893831221867e-05, "loss": 0.139, "step": 31264 }, { "epoch": 22.904761904761905, "grad_norm": 0.77734375, "learning_rate": 2.995877636479631e-05, "loss": 0.0949, "step": 31265 }, { "epoch": 22.905494505494506, "grad_norm": 0.7421875, "learning_rate": 2.9955658996790494e-05, "loss": 0.1169, "step": 31266 }, { "epoch": 22.906227106227107, "grad_norm": 0.81640625, "learning_rate": 2.9952541727219348e-05, "loss": 0.1216, "step": 31267 }, { "epoch": 22.906959706959707, "grad_norm": 1.2734375, "learning_rate": 2.994942455609783e-05, "loss": 0.1226, "step": 31268 }, { "epoch": 22.907692307692308, "grad_norm": 0.8515625, "learning_rate": 2.994630748344086e-05, "loss": 0.1061, "step": 31269 }, { "epoch": 22.90842490842491, "grad_norm": 1.1953125, "learning_rate": 2.9943190509263375e-05, "loss": 0.1237, "step": 31270 }, { "epoch": 22.90915750915751, "grad_norm": 1.3515625, "learning_rate": 2.9940073633580324e-05, "loss": 0.1461, "step": 31271 }, { "epoch": 22.90989010989011, "grad_norm": 0.7109375, "learning_rate": 2.993695685640663e-05, "loss": 0.1064, "step": 31272 }, { "epoch": 22.91062271062271, "grad_norm": 1.3125, "learning_rate": 2.993384017775722e-05, "loss": 0.1574, "step": 31273 }, { "epoch": 22.91135531135531, "grad_norm": 0.703125, "learning_rate": 2.993072359764704e-05, "loss": 0.099, "step": 31274 }, { "epoch": 22.912087912087912, "grad_norm": 0.82421875, "learning_rate": 2.9927607116090995e-05, "loss": 0.0996, "step": 31275 }, { "epoch": 22.912820512820513, "grad_norm": 1.1875, "learning_rate": 2.9924490733104062e-05, "loss": 0.1281, "step": 31276 }, { "epoch": 22.913553113553114, "grad_norm": 0.82421875, "learning_rate": 2.9921374448701138e-05, "loss": 0.1139, "step": 31277 }, { "epoch": 22.914285714285715, "grad_norm": 0.9453125, "learning_rate": 2.9918258262897176e-05, "loss": 0.1242, "step": 31278 }, { "epoch": 22.915018315018315, "grad_norm": 0.8046875, "learning_rate": 2.9915142175707077e-05, "loss": 0.1197, "step": 31279 }, { "epoch": 22.915750915750916, "grad_norm": 1.28125, "learning_rate": 2.9912026187145786e-05, "loss": 0.1305, "step": 31280 }, { "epoch": 22.916483516483517, "grad_norm": 0.6953125, "learning_rate": 2.9908910297228244e-05, "loss": 0.1298, "step": 31281 }, { "epoch": 22.917216117216118, "grad_norm": 1.3984375, "learning_rate": 2.9905794505969376e-05, "loss": 0.1351, "step": 31282 }, { "epoch": 22.91794871794872, "grad_norm": 0.83203125, "learning_rate": 2.9902678813384117e-05, "loss": 0.1313, "step": 31283 }, { "epoch": 22.91868131868132, "grad_norm": 1.015625, "learning_rate": 2.9899563219487375e-05, "loss": 0.1401, "step": 31284 }, { "epoch": 22.91941391941392, "grad_norm": 1.0546875, "learning_rate": 2.989644772429409e-05, "loss": 0.1106, "step": 31285 }, { "epoch": 22.92014652014652, "grad_norm": 0.91015625, "learning_rate": 2.989333232781918e-05, "loss": 0.1029, "step": 31286 }, { "epoch": 22.92087912087912, "grad_norm": 1.390625, "learning_rate": 2.9890217030077578e-05, "loss": 0.1339, "step": 31287 }, { "epoch": 22.921611721611722, "grad_norm": 0.78515625, "learning_rate": 2.9887101831084235e-05, "loss": 0.1318, "step": 31288 }, { "epoch": 22.922344322344323, "grad_norm": 1.203125, "learning_rate": 2.9883986730854036e-05, "loss": 0.1148, "step": 31289 }, { "epoch": 22.923076923076923, "grad_norm": 0.89453125, "learning_rate": 2.9880871729401922e-05, "loss": 0.1354, "step": 31290 }, { "epoch": 22.923809523809524, "grad_norm": 1.1171875, "learning_rate": 2.987775682674282e-05, "loss": 0.1406, "step": 31291 }, { "epoch": 22.924542124542125, "grad_norm": 1.1875, "learning_rate": 2.9874642022891674e-05, "loss": 0.1373, "step": 31292 }, { "epoch": 22.925274725274726, "grad_norm": 1.2578125, "learning_rate": 2.9871527317863368e-05, "loss": 0.1182, "step": 31293 }, { "epoch": 22.926007326007326, "grad_norm": 0.8515625, "learning_rate": 2.9868412711672867e-05, "loss": 0.1095, "step": 31294 }, { "epoch": 22.926739926739927, "grad_norm": 0.9609375, "learning_rate": 2.9865298204335035e-05, "loss": 0.1146, "step": 31295 }, { "epoch": 22.927472527472528, "grad_norm": 0.9140625, "learning_rate": 2.9862183795864873e-05, "loss": 0.1174, "step": 31296 }, { "epoch": 22.92820512820513, "grad_norm": 1.234375, "learning_rate": 2.9859069486277243e-05, "loss": 0.1008, "step": 31297 }, { "epoch": 22.92893772893773, "grad_norm": 1.6484375, "learning_rate": 2.985595527558711e-05, "loss": 0.1153, "step": 31298 }, { "epoch": 22.92967032967033, "grad_norm": 1.3359375, "learning_rate": 2.9852841163809344e-05, "loss": 0.1287, "step": 31299 }, { "epoch": 22.93040293040293, "grad_norm": 1.4609375, "learning_rate": 2.98497271509589e-05, "loss": 0.1621, "step": 31300 }, { "epoch": 22.93113553113553, "grad_norm": 1.515625, "learning_rate": 2.9846613237050697e-05, "loss": 0.1247, "step": 31301 }, { "epoch": 22.931868131868132, "grad_norm": 1.9609375, "learning_rate": 2.9843499422099645e-05, "loss": 0.1409, "step": 31302 }, { "epoch": 22.932600732600733, "grad_norm": 1.171875, "learning_rate": 2.9840385706120684e-05, "loss": 0.0905, "step": 31303 }, { "epoch": 22.933333333333334, "grad_norm": 1.6171875, "learning_rate": 2.9837272089128703e-05, "loss": 0.1457, "step": 31304 }, { "epoch": 22.934065934065934, "grad_norm": 1.203125, "learning_rate": 2.9834158571138648e-05, "loss": 0.1274, "step": 31305 }, { "epoch": 22.934798534798535, "grad_norm": 1.3515625, "learning_rate": 2.9831045152165392e-05, "loss": 0.1195, "step": 31306 }, { "epoch": 22.935531135531136, "grad_norm": 2.25, "learning_rate": 2.9827931832223926e-05, "loss": 0.1327, "step": 31307 }, { "epoch": 22.936263736263736, "grad_norm": 0.859375, "learning_rate": 2.9824818611329098e-05, "loss": 0.0996, "step": 31308 }, { "epoch": 22.936996336996337, "grad_norm": 1.4140625, "learning_rate": 2.982170548949587e-05, "loss": 0.1008, "step": 31309 }, { "epoch": 22.937728937728938, "grad_norm": 0.76171875, "learning_rate": 2.981859246673912e-05, "loss": 0.094, "step": 31310 }, { "epoch": 22.93846153846154, "grad_norm": 1.25, "learning_rate": 2.9815479543073787e-05, "loss": 0.1024, "step": 31311 }, { "epoch": 22.93919413919414, "grad_norm": 1.4609375, "learning_rate": 2.981236671851478e-05, "loss": 0.1318, "step": 31312 }, { "epoch": 22.93992673992674, "grad_norm": 1.640625, "learning_rate": 2.980925399307702e-05, "loss": 0.1087, "step": 31313 }, { "epoch": 22.94065934065934, "grad_norm": 1.1796875, "learning_rate": 2.9806141366775425e-05, "loss": 0.1195, "step": 31314 }, { "epoch": 22.94139194139194, "grad_norm": 1.6640625, "learning_rate": 2.980302883962488e-05, "loss": 0.1328, "step": 31315 }, { "epoch": 22.942124542124542, "grad_norm": 1.1640625, "learning_rate": 2.9799916411640326e-05, "loss": 0.1566, "step": 31316 }, { "epoch": 22.942857142857143, "grad_norm": 1.0234375, "learning_rate": 2.9796804082836665e-05, "loss": 0.1166, "step": 31317 }, { "epoch": 22.943589743589744, "grad_norm": 1.4375, "learning_rate": 2.979369185322884e-05, "loss": 0.112, "step": 31318 }, { "epoch": 22.944322344322345, "grad_norm": 1.1171875, "learning_rate": 2.9790579722831708e-05, "loss": 0.1228, "step": 31319 }, { "epoch": 22.945054945054945, "grad_norm": 1.46875, "learning_rate": 2.9787467691660225e-05, "loss": 0.1443, "step": 31320 }, { "epoch": 22.945787545787546, "grad_norm": 1.140625, "learning_rate": 2.9784355759729244e-05, "loss": 0.1247, "step": 31321 }, { "epoch": 22.946520146520147, "grad_norm": 0.7265625, "learning_rate": 2.9781243927053752e-05, "loss": 0.1057, "step": 31322 }, { "epoch": 22.947252747252747, "grad_norm": 0.6953125, "learning_rate": 2.9778132193648618e-05, "loss": 0.0883, "step": 31323 }, { "epoch": 22.947985347985348, "grad_norm": 0.7421875, "learning_rate": 2.977502055952876e-05, "loss": 0.0734, "step": 31324 }, { "epoch": 22.94871794871795, "grad_norm": 1.3359375, "learning_rate": 2.977190902470906e-05, "loss": 0.1247, "step": 31325 }, { "epoch": 22.94945054945055, "grad_norm": 1.3828125, "learning_rate": 2.9768797589204457e-05, "loss": 0.1243, "step": 31326 }, { "epoch": 22.95018315018315, "grad_norm": 0.90234375, "learning_rate": 2.9765686253029846e-05, "loss": 0.0886, "step": 31327 }, { "epoch": 22.95091575091575, "grad_norm": 0.59375, "learning_rate": 2.976257501620014e-05, "loss": 0.0938, "step": 31328 }, { "epoch": 22.951648351648352, "grad_norm": 0.99609375, "learning_rate": 2.975946387873027e-05, "loss": 0.0988, "step": 31329 }, { "epoch": 22.952380952380953, "grad_norm": 1.1015625, "learning_rate": 2.9756352840635095e-05, "loss": 0.1299, "step": 31330 }, { "epoch": 22.953113553113553, "grad_norm": 1.046875, "learning_rate": 2.9753241901929556e-05, "loss": 0.1155, "step": 31331 }, { "epoch": 22.953846153846154, "grad_norm": 0.66015625, "learning_rate": 2.975013106262851e-05, "loss": 0.1054, "step": 31332 }, { "epoch": 22.954578754578755, "grad_norm": 1.2734375, "learning_rate": 2.9747020322746942e-05, "loss": 0.1118, "step": 31333 }, { "epoch": 22.955311355311355, "grad_norm": 1.0390625, "learning_rate": 2.9743909682299685e-05, "loss": 0.1125, "step": 31334 }, { "epoch": 22.956043956043956, "grad_norm": 0.6328125, "learning_rate": 2.9740799141301696e-05, "loss": 0.1048, "step": 31335 }, { "epoch": 22.956776556776557, "grad_norm": 0.97265625, "learning_rate": 2.9737688699767828e-05, "loss": 0.1304, "step": 31336 }, { "epoch": 22.957509157509158, "grad_norm": 1.1875, "learning_rate": 2.9734578357713016e-05, "loss": 0.1245, "step": 31337 }, { "epoch": 22.95824175824176, "grad_norm": 0.890625, "learning_rate": 2.9731468115152155e-05, "loss": 0.1249, "step": 31338 }, { "epoch": 22.95897435897436, "grad_norm": 0.67578125, "learning_rate": 2.972835797210014e-05, "loss": 0.1247, "step": 31339 }, { "epoch": 22.95970695970696, "grad_norm": 1.0, "learning_rate": 2.972524792857191e-05, "loss": 0.1437, "step": 31340 }, { "epoch": 22.96043956043956, "grad_norm": 0.6875, "learning_rate": 2.972213798458231e-05, "loss": 0.1112, "step": 31341 }, { "epoch": 22.96117216117216, "grad_norm": 0.55078125, "learning_rate": 2.971902814014627e-05, "loss": 0.0941, "step": 31342 }, { "epoch": 22.961904761904762, "grad_norm": 1.1015625, "learning_rate": 2.9715918395278696e-05, "loss": 0.1361, "step": 31343 }, { "epoch": 22.962637362637363, "grad_norm": 1.078125, "learning_rate": 2.971280874999449e-05, "loss": 0.0967, "step": 31344 }, { "epoch": 22.963369963369964, "grad_norm": 1.4296875, "learning_rate": 2.9709699204308523e-05, "loss": 0.127, "step": 31345 }, { "epoch": 22.964102564102564, "grad_norm": 1.046875, "learning_rate": 2.9706589758235737e-05, "loss": 0.1129, "step": 31346 }, { "epoch": 22.964835164835165, "grad_norm": 1.265625, "learning_rate": 2.9703480411790962e-05, "loss": 0.111, "step": 31347 }, { "epoch": 22.965567765567766, "grad_norm": 1.3984375, "learning_rate": 2.9700371164989182e-05, "loss": 0.1337, "step": 31348 }, { "epoch": 22.966300366300366, "grad_norm": 0.72265625, "learning_rate": 2.969726201784523e-05, "loss": 0.1246, "step": 31349 }, { "epoch": 22.967032967032967, "grad_norm": 0.8359375, "learning_rate": 2.969415297037402e-05, "loss": 0.1124, "step": 31350 }, { "epoch": 22.967765567765568, "grad_norm": 1.0078125, "learning_rate": 2.9691044022590482e-05, "loss": 0.1441, "step": 31351 }, { "epoch": 22.96849816849817, "grad_norm": 1.1484375, "learning_rate": 2.968793517450945e-05, "loss": 0.1012, "step": 31352 }, { "epoch": 22.96923076923077, "grad_norm": 0.7734375, "learning_rate": 2.968482642614587e-05, "loss": 0.1101, "step": 31353 }, { "epoch": 22.96996336996337, "grad_norm": 1.0859375, "learning_rate": 2.9681717777514613e-05, "loss": 0.1395, "step": 31354 }, { "epoch": 22.97069597069597, "grad_norm": 1.7265625, "learning_rate": 2.9678609228630593e-05, "loss": 0.1376, "step": 31355 }, { "epoch": 22.97142857142857, "grad_norm": 0.99609375, "learning_rate": 2.9675500779508674e-05, "loss": 0.1538, "step": 31356 }, { "epoch": 22.972161172161172, "grad_norm": 1.2265625, "learning_rate": 2.9672392430163794e-05, "loss": 0.1294, "step": 31357 }, { "epoch": 22.972893772893773, "grad_norm": 1.2265625, "learning_rate": 2.9669284180610772e-05, "loss": 0.1089, "step": 31358 }, { "epoch": 22.973626373626374, "grad_norm": 0.87109375, "learning_rate": 2.9666176030864595e-05, "loss": 0.1328, "step": 31359 }, { "epoch": 22.974358974358974, "grad_norm": 0.765625, "learning_rate": 2.9663067980940085e-05, "loss": 0.094, "step": 31360 }, { "epoch": 22.975091575091575, "grad_norm": 1.140625, "learning_rate": 2.9659960030852173e-05, "loss": 0.1293, "step": 31361 }, { "epoch": 22.975824175824176, "grad_norm": 1.6015625, "learning_rate": 2.9656852180615715e-05, "loss": 0.1173, "step": 31362 }, { "epoch": 22.976556776556777, "grad_norm": 1.1484375, "learning_rate": 2.965374443024563e-05, "loss": 0.1267, "step": 31363 }, { "epoch": 22.977289377289377, "grad_norm": 0.671875, "learning_rate": 2.9650636779756785e-05, "loss": 0.0923, "step": 31364 }, { "epoch": 22.978021978021978, "grad_norm": 0.87109375, "learning_rate": 2.9647529229164083e-05, "loss": 0.1406, "step": 31365 }, { "epoch": 22.97875457875458, "grad_norm": 1.3671875, "learning_rate": 2.9644421778482447e-05, "loss": 0.1684, "step": 31366 }, { "epoch": 22.97948717948718, "grad_norm": 0.953125, "learning_rate": 2.96413144277267e-05, "loss": 0.1034, "step": 31367 }, { "epoch": 22.98021978021978, "grad_norm": 0.9921875, "learning_rate": 2.9638207176911775e-05, "loss": 0.1243, "step": 31368 }, { "epoch": 22.98095238095238, "grad_norm": 0.82421875, "learning_rate": 2.963510002605254e-05, "loss": 0.1388, "step": 31369 }, { "epoch": 22.98168498168498, "grad_norm": 0.7578125, "learning_rate": 2.9631992975163914e-05, "loss": 0.1003, "step": 31370 }, { "epoch": 22.982417582417582, "grad_norm": 0.96875, "learning_rate": 2.9628886024260737e-05, "loss": 0.1095, "step": 31371 }, { "epoch": 22.983150183150183, "grad_norm": 0.8515625, "learning_rate": 2.9625779173357937e-05, "loss": 0.1199, "step": 31372 }, { "epoch": 22.983882783882784, "grad_norm": 1.328125, "learning_rate": 2.9622672422470364e-05, "loss": 0.1325, "step": 31373 }, { "epoch": 22.984615384615385, "grad_norm": 0.6328125, "learning_rate": 2.961956577161292e-05, "loss": 0.1232, "step": 31374 }, { "epoch": 22.985347985347985, "grad_norm": 0.95703125, "learning_rate": 2.961645922080049e-05, "loss": 0.1157, "step": 31375 }, { "epoch": 22.986080586080586, "grad_norm": 0.85546875, "learning_rate": 2.9613352770047968e-05, "loss": 0.1073, "step": 31376 }, { "epoch": 22.986813186813187, "grad_norm": 0.72265625, "learning_rate": 2.9610246419370236e-05, "loss": 0.1141, "step": 31377 }, { "epoch": 22.987545787545788, "grad_norm": 1.3125, "learning_rate": 2.960714016878216e-05, "loss": 0.1149, "step": 31378 }, { "epoch": 22.98827838827839, "grad_norm": 0.8828125, "learning_rate": 2.9604034018298632e-05, "loss": 0.1035, "step": 31379 }, { "epoch": 22.98901098901099, "grad_norm": 0.76953125, "learning_rate": 2.9600927967934544e-05, "loss": 0.1044, "step": 31380 }, { "epoch": 22.98974358974359, "grad_norm": 1.125, "learning_rate": 2.9597822017704784e-05, "loss": 0.1217, "step": 31381 }, { "epoch": 22.99047619047619, "grad_norm": 0.9609375, "learning_rate": 2.9594716167624202e-05, "loss": 0.1074, "step": 31382 }, { "epoch": 22.99120879120879, "grad_norm": 1.1328125, "learning_rate": 2.9591610417707715e-05, "loss": 0.1389, "step": 31383 }, { "epoch": 22.991941391941392, "grad_norm": 0.91796875, "learning_rate": 2.9588504767970144e-05, "loss": 0.1055, "step": 31384 }, { "epoch": 22.992673992673993, "grad_norm": 0.90234375, "learning_rate": 2.9585399218426453e-05, "loss": 0.1046, "step": 31385 }, { "epoch": 22.993406593406593, "grad_norm": 0.734375, "learning_rate": 2.9582293769091463e-05, "loss": 0.1166, "step": 31386 }, { "epoch": 22.994139194139194, "grad_norm": 0.68359375, "learning_rate": 2.957918841998009e-05, "loss": 0.1077, "step": 31387 }, { "epoch": 22.994871794871795, "grad_norm": 0.94921875, "learning_rate": 2.957608317110717e-05, "loss": 0.1108, "step": 31388 }, { "epoch": 22.995604395604396, "grad_norm": 0.97265625, "learning_rate": 2.9572978022487606e-05, "loss": 0.1313, "step": 31389 }, { "epoch": 22.996336996336996, "grad_norm": 0.91015625, "learning_rate": 2.956987297413628e-05, "loss": 0.1111, "step": 31390 }, { "epoch": 22.997069597069597, "grad_norm": 0.578125, "learning_rate": 2.9566768026068058e-05, "loss": 0.0971, "step": 31391 }, { "epoch": 22.997802197802198, "grad_norm": 0.8671875, "learning_rate": 2.9563663178297842e-05, "loss": 0.0995, "step": 31392 }, { "epoch": 22.9985347985348, "grad_norm": 1.03125, "learning_rate": 2.9560558430840467e-05, "loss": 0.1072, "step": 31393 }, { "epoch": 22.9992673992674, "grad_norm": 0.78515625, "learning_rate": 2.9557453783710835e-05, "loss": 0.1018, "step": 31394 }, { "epoch": 23.0, "grad_norm": 1.109375, "learning_rate": 2.9554349236923816e-05, "loss": 0.1161, "step": 31395 }, { "epoch": 23.0007326007326, "grad_norm": 0.87109375, "learning_rate": 2.9551244790494304e-05, "loss": 0.1216, "step": 31396 }, { "epoch": 23.0014652014652, "grad_norm": 0.83984375, "learning_rate": 2.954814044443714e-05, "loss": 0.1272, "step": 31397 }, { "epoch": 23.002197802197802, "grad_norm": 1.515625, "learning_rate": 2.954503619876723e-05, "loss": 0.1311, "step": 31398 }, { "epoch": 23.002930402930403, "grad_norm": 1.015625, "learning_rate": 2.9541932053499404e-05, "loss": 0.1217, "step": 31399 }, { "epoch": 23.003663003663004, "grad_norm": 1.1484375, "learning_rate": 2.9538828008648568e-05, "loss": 0.1199, "step": 31400 }, { "epoch": 23.004395604395604, "grad_norm": 0.96875, "learning_rate": 2.9535724064229585e-05, "loss": 0.1479, "step": 31401 }, { "epoch": 23.005128205128205, "grad_norm": 1.3515625, "learning_rate": 2.9532620220257335e-05, "loss": 0.1422, "step": 31402 }, { "epoch": 23.005860805860806, "grad_norm": 1.3984375, "learning_rate": 2.9529516476746704e-05, "loss": 0.115, "step": 31403 }, { "epoch": 23.006593406593407, "grad_norm": 1.484375, "learning_rate": 2.952641283371252e-05, "loss": 0.1211, "step": 31404 }, { "epoch": 23.007326007326007, "grad_norm": 1.359375, "learning_rate": 2.9523309291169683e-05, "loss": 0.1286, "step": 31405 }, { "epoch": 23.008058608058608, "grad_norm": 0.890625, "learning_rate": 2.9520205849133053e-05, "loss": 0.1072, "step": 31406 }, { "epoch": 23.00879120879121, "grad_norm": 1.03125, "learning_rate": 2.9517102507617522e-05, "loss": 0.1255, "step": 31407 }, { "epoch": 23.00952380952381, "grad_norm": 1.046875, "learning_rate": 2.9513999266637925e-05, "loss": 0.1338, "step": 31408 }, { "epoch": 23.01025641025641, "grad_norm": 0.921875, "learning_rate": 2.9510896126209164e-05, "loss": 0.1153, "step": 31409 }, { "epoch": 23.01098901098901, "grad_norm": 1.515625, "learning_rate": 2.9507793086346057e-05, "loss": 0.1061, "step": 31410 }, { "epoch": 23.01172161172161, "grad_norm": 1.3046875, "learning_rate": 2.9504690147063536e-05, "loss": 0.1213, "step": 31411 }, { "epoch": 23.012454212454212, "grad_norm": 1.25, "learning_rate": 2.9501587308376422e-05, "loss": 0.1126, "step": 31412 }, { "epoch": 23.013186813186813, "grad_norm": 1.171875, "learning_rate": 2.949848457029961e-05, "loss": 0.1152, "step": 31413 }, { "epoch": 23.013919413919414, "grad_norm": 1.6171875, "learning_rate": 2.949538193284794e-05, "loss": 0.1254, "step": 31414 }, { "epoch": 23.014652014652015, "grad_norm": 0.70703125, "learning_rate": 2.9492279396036287e-05, "loss": 0.1085, "step": 31415 }, { "epoch": 23.015384615384615, "grad_norm": 1.0234375, "learning_rate": 2.948917695987952e-05, "loss": 0.1274, "step": 31416 }, { "epoch": 23.016117216117216, "grad_norm": 1.515625, "learning_rate": 2.9486074624392507e-05, "loss": 0.1232, "step": 31417 }, { "epoch": 23.016849816849817, "grad_norm": 0.58203125, "learning_rate": 2.9482972389590123e-05, "loss": 0.1166, "step": 31418 }, { "epoch": 23.017582417582418, "grad_norm": 0.74609375, "learning_rate": 2.94798702554872e-05, "loss": 0.1378, "step": 31419 }, { "epoch": 23.01831501831502, "grad_norm": 0.984375, "learning_rate": 2.9476768222098616e-05, "loss": 0.1143, "step": 31420 }, { "epoch": 23.01904761904762, "grad_norm": 1.15625, "learning_rate": 2.947366628943925e-05, "loss": 0.1053, "step": 31421 }, { "epoch": 23.01978021978022, "grad_norm": 1.3515625, "learning_rate": 2.9470564457523962e-05, "loss": 0.1308, "step": 31422 }, { "epoch": 23.02051282051282, "grad_norm": 1.5390625, "learning_rate": 2.946746272636758e-05, "loss": 0.1398, "step": 31423 }, { "epoch": 23.02124542124542, "grad_norm": 0.82421875, "learning_rate": 2.946436109598501e-05, "loss": 0.1106, "step": 31424 }, { "epoch": 23.021978021978022, "grad_norm": 1.4765625, "learning_rate": 2.9461259566391074e-05, "loss": 0.1089, "step": 31425 }, { "epoch": 23.022710622710623, "grad_norm": 1.03125, "learning_rate": 2.945815813760065e-05, "loss": 0.1053, "step": 31426 }, { "epoch": 23.023443223443223, "grad_norm": 1.2109375, "learning_rate": 2.94550568096286e-05, "loss": 0.1242, "step": 31427 }, { "epoch": 23.024175824175824, "grad_norm": 2.140625, "learning_rate": 2.945195558248978e-05, "loss": 0.1498, "step": 31428 }, { "epoch": 23.024908424908425, "grad_norm": 0.87890625, "learning_rate": 2.9448854456199066e-05, "loss": 0.1092, "step": 31429 }, { "epoch": 23.025641025641026, "grad_norm": 0.80078125, "learning_rate": 2.944575343077129e-05, "loss": 0.1016, "step": 31430 }, { "epoch": 23.026373626373626, "grad_norm": 1.2734375, "learning_rate": 2.9442652506221313e-05, "loss": 0.101, "step": 31431 }, { "epoch": 23.027106227106227, "grad_norm": 1.8515625, "learning_rate": 2.9439551682564e-05, "loss": 0.1528, "step": 31432 }, { "epoch": 23.027838827838828, "grad_norm": 1.828125, "learning_rate": 2.9436450959814227e-05, "loss": 0.1133, "step": 31433 }, { "epoch": 23.02857142857143, "grad_norm": 1.0078125, "learning_rate": 2.9433350337986815e-05, "loss": 0.096, "step": 31434 }, { "epoch": 23.02930402930403, "grad_norm": 2.25, "learning_rate": 2.9430249817096667e-05, "loss": 0.1456, "step": 31435 }, { "epoch": 23.03003663003663, "grad_norm": 1.1640625, "learning_rate": 2.942714939715856e-05, "loss": 0.1242, "step": 31436 }, { "epoch": 23.03076923076923, "grad_norm": 0.58203125, "learning_rate": 2.942404907818744e-05, "loss": 0.0802, "step": 31437 }, { "epoch": 23.03150183150183, "grad_norm": 0.8046875, "learning_rate": 2.9420948860198095e-05, "loss": 0.118, "step": 31438 }, { "epoch": 23.032234432234432, "grad_norm": 1.0234375, "learning_rate": 2.9417848743205423e-05, "loss": 0.1146, "step": 31439 }, { "epoch": 23.032967032967033, "grad_norm": 1.6875, "learning_rate": 2.9414748727224246e-05, "loss": 0.115, "step": 31440 }, { "epoch": 23.033699633699634, "grad_norm": 0.86328125, "learning_rate": 2.9411648812269427e-05, "loss": 0.1003, "step": 31441 }, { "epoch": 23.034432234432234, "grad_norm": 1.234375, "learning_rate": 2.9408548998355816e-05, "loss": 0.1184, "step": 31442 }, { "epoch": 23.035164835164835, "grad_norm": 1.171875, "learning_rate": 2.940544928549828e-05, "loss": 0.1461, "step": 31443 }, { "epoch": 23.035897435897436, "grad_norm": 0.8671875, "learning_rate": 2.9402349673711672e-05, "loss": 0.1302, "step": 31444 }, { "epoch": 23.036630036630036, "grad_norm": 1.015625, "learning_rate": 2.9399250163010818e-05, "loss": 0.1216, "step": 31445 }, { "epoch": 23.037362637362637, "grad_norm": 0.95703125, "learning_rate": 2.9396150753410583e-05, "loss": 0.1075, "step": 31446 }, { "epoch": 23.038095238095238, "grad_norm": 0.859375, "learning_rate": 2.939305144492582e-05, "loss": 0.1219, "step": 31447 }, { "epoch": 23.03882783882784, "grad_norm": 0.796875, "learning_rate": 2.9389952237571393e-05, "loss": 0.1362, "step": 31448 }, { "epoch": 23.03956043956044, "grad_norm": 1.203125, "learning_rate": 2.9386853131362118e-05, "loss": 0.1118, "step": 31449 }, { "epoch": 23.04029304029304, "grad_norm": 0.8515625, "learning_rate": 2.9383754126312872e-05, "loss": 0.1215, "step": 31450 }, { "epoch": 23.04102564102564, "grad_norm": 0.92578125, "learning_rate": 2.938065522243848e-05, "loss": 0.1045, "step": 31451 }, { "epoch": 23.04175824175824, "grad_norm": 1.2109375, "learning_rate": 2.9377556419753805e-05, "loss": 0.1301, "step": 31452 }, { "epoch": 23.042490842490842, "grad_norm": 0.94921875, "learning_rate": 2.9374457718273684e-05, "loss": 0.1016, "step": 31453 }, { "epoch": 23.043223443223443, "grad_norm": 0.7578125, "learning_rate": 2.937135911801297e-05, "loss": 0.0819, "step": 31454 }, { "epoch": 23.043956043956044, "grad_norm": 1.515625, "learning_rate": 2.9368260618986536e-05, "loss": 0.1478, "step": 31455 }, { "epoch": 23.044688644688645, "grad_norm": 1.1484375, "learning_rate": 2.936516222120918e-05, "loss": 0.1424, "step": 31456 }, { "epoch": 23.045421245421245, "grad_norm": 1.203125, "learning_rate": 2.9362063924695772e-05, "loss": 0.1101, "step": 31457 }, { "epoch": 23.046153846153846, "grad_norm": 0.78515625, "learning_rate": 2.935896572946115e-05, "loss": 0.0885, "step": 31458 }, { "epoch": 23.046886446886447, "grad_norm": 0.98828125, "learning_rate": 2.935586763552019e-05, "loss": 0.1131, "step": 31459 }, { "epoch": 23.047619047619047, "grad_norm": 0.94140625, "learning_rate": 2.9352769642887683e-05, "loss": 0.1093, "step": 31460 }, { "epoch": 23.048351648351648, "grad_norm": 1.0859375, "learning_rate": 2.9349671751578516e-05, "loss": 0.1037, "step": 31461 }, { "epoch": 23.04908424908425, "grad_norm": 0.94921875, "learning_rate": 2.9346573961607486e-05, "loss": 0.098, "step": 31462 }, { "epoch": 23.04981684981685, "grad_norm": 1.0859375, "learning_rate": 2.9343476272989495e-05, "loss": 0.1243, "step": 31463 }, { "epoch": 23.05054945054945, "grad_norm": 0.97265625, "learning_rate": 2.9340378685739345e-05, "loss": 0.1274, "step": 31464 }, { "epoch": 23.05128205128205, "grad_norm": 1.1171875, "learning_rate": 2.933728119987188e-05, "loss": 0.1102, "step": 31465 }, { "epoch": 23.052014652014652, "grad_norm": 1.125, "learning_rate": 2.9334183815401967e-05, "loss": 0.1506, "step": 31466 }, { "epoch": 23.052747252747253, "grad_norm": 0.88671875, "learning_rate": 2.9331086532344408e-05, "loss": 0.1124, "step": 31467 }, { "epoch": 23.053479853479853, "grad_norm": 1.078125, "learning_rate": 2.9327989350714064e-05, "loss": 0.1485, "step": 31468 }, { "epoch": 23.054212454212454, "grad_norm": 1.5625, "learning_rate": 2.932489227052577e-05, "loss": 0.118, "step": 31469 }, { "epoch": 23.054945054945055, "grad_norm": 1.3359375, "learning_rate": 2.9321795291794398e-05, "loss": 0.1358, "step": 31470 }, { "epoch": 23.055677655677655, "grad_norm": 1.1640625, "learning_rate": 2.9318698414534728e-05, "loss": 0.141, "step": 31471 }, { "epoch": 23.056410256410256, "grad_norm": 0.8828125, "learning_rate": 2.931560163876163e-05, "loss": 0.1197, "step": 31472 }, { "epoch": 23.057142857142857, "grad_norm": 0.79296875, "learning_rate": 2.9312504964489934e-05, "loss": 0.097, "step": 31473 }, { "epoch": 23.057875457875458, "grad_norm": 1.2265625, "learning_rate": 2.930940839173451e-05, "loss": 0.1052, "step": 31474 }, { "epoch": 23.05860805860806, "grad_norm": 1.4921875, "learning_rate": 2.9306311920510143e-05, "loss": 0.1431, "step": 31475 }, { "epoch": 23.05934065934066, "grad_norm": 1.6328125, "learning_rate": 2.9303215550831714e-05, "loss": 0.1214, "step": 31476 }, { "epoch": 23.06007326007326, "grad_norm": 0.6796875, "learning_rate": 2.9300119282714017e-05, "loss": 0.0975, "step": 31477 }, { "epoch": 23.06080586080586, "grad_norm": 1.0390625, "learning_rate": 2.9297023116171907e-05, "loss": 0.1052, "step": 31478 }, { "epoch": 23.06153846153846, "grad_norm": 0.8515625, "learning_rate": 2.9293927051220223e-05, "loss": 0.0877, "step": 31479 }, { "epoch": 23.062271062271062, "grad_norm": 1.21875, "learning_rate": 2.9290831087873792e-05, "loss": 0.1289, "step": 31480 }, { "epoch": 23.063003663003663, "grad_norm": 1.484375, "learning_rate": 2.9287735226147477e-05, "loss": 0.1429, "step": 31481 }, { "epoch": 23.063736263736264, "grad_norm": 1.2578125, "learning_rate": 2.9284639466056058e-05, "loss": 0.111, "step": 31482 }, { "epoch": 23.064468864468864, "grad_norm": 0.94140625, "learning_rate": 2.9281543807614404e-05, "loss": 0.1097, "step": 31483 }, { "epoch": 23.065201465201465, "grad_norm": 1.1328125, "learning_rate": 2.9278448250837337e-05, "loss": 0.1233, "step": 31484 }, { "epoch": 23.065934065934066, "grad_norm": 0.85546875, "learning_rate": 2.9275352795739708e-05, "loss": 0.1238, "step": 31485 }, { "epoch": 23.066666666666666, "grad_norm": 1.125, "learning_rate": 2.9272257442336316e-05, "loss": 0.1347, "step": 31486 }, { "epoch": 23.067399267399267, "grad_norm": 1.234375, "learning_rate": 2.9269162190642017e-05, "loss": 0.1291, "step": 31487 }, { "epoch": 23.068131868131868, "grad_norm": 1.6328125, "learning_rate": 2.9266067040671602e-05, "loss": 0.1361, "step": 31488 }, { "epoch": 23.06886446886447, "grad_norm": 1.0234375, "learning_rate": 2.9262971992439974e-05, "loss": 0.1124, "step": 31489 }, { "epoch": 23.06959706959707, "grad_norm": 0.87109375, "learning_rate": 2.9259877045961893e-05, "loss": 0.1015, "step": 31490 }, { "epoch": 23.07032967032967, "grad_norm": 1.4296875, "learning_rate": 2.9256782201252208e-05, "loss": 0.1348, "step": 31491 }, { "epoch": 23.07106227106227, "grad_norm": 0.8671875, "learning_rate": 2.925368745832578e-05, "loss": 0.0992, "step": 31492 }, { "epoch": 23.07179487179487, "grad_norm": 0.99609375, "learning_rate": 2.9250592817197386e-05, "loss": 0.1019, "step": 31493 }, { "epoch": 23.072527472527472, "grad_norm": 2.125, "learning_rate": 2.9247498277881886e-05, "loss": 0.1329, "step": 31494 }, { "epoch": 23.073260073260073, "grad_norm": 1.0546875, "learning_rate": 2.924440384039409e-05, "loss": 0.149, "step": 31495 }, { "epoch": 23.073992673992674, "grad_norm": 0.73046875, "learning_rate": 2.9241309504748853e-05, "loss": 0.1095, "step": 31496 }, { "epoch": 23.074725274725274, "grad_norm": 1.0078125, "learning_rate": 2.9238215270960963e-05, "loss": 0.1518, "step": 31497 }, { "epoch": 23.075457875457875, "grad_norm": 1.15625, "learning_rate": 2.9235121139045268e-05, "loss": 0.1198, "step": 31498 }, { "epoch": 23.076190476190476, "grad_norm": 0.72265625, "learning_rate": 2.9232027109016584e-05, "loss": 0.1188, "step": 31499 }, { "epoch": 23.076923076923077, "grad_norm": 1.84375, "learning_rate": 2.9228933180889762e-05, "loss": 0.1522, "step": 31500 }, { "epoch": 23.077655677655677, "grad_norm": 0.9296875, "learning_rate": 2.922583935467958e-05, "loss": 0.1228, "step": 31501 }, { "epoch": 23.078388278388278, "grad_norm": 1.3359375, "learning_rate": 2.9222745630400904e-05, "loss": 0.1179, "step": 31502 }, { "epoch": 23.07912087912088, "grad_norm": 1.359375, "learning_rate": 2.9219652008068527e-05, "loss": 0.1207, "step": 31503 }, { "epoch": 23.07985347985348, "grad_norm": 1.09375, "learning_rate": 2.9216558487697277e-05, "loss": 0.1082, "step": 31504 }, { "epoch": 23.08058608058608, "grad_norm": 0.89453125, "learning_rate": 2.921346506930198e-05, "loss": 0.1221, "step": 31505 }, { "epoch": 23.08131868131868, "grad_norm": 1.2734375, "learning_rate": 2.9210371752897464e-05, "loss": 0.141, "step": 31506 }, { "epoch": 23.08205128205128, "grad_norm": 1.265625, "learning_rate": 2.920727853849856e-05, "loss": 0.1241, "step": 31507 }, { "epoch": 23.082783882783882, "grad_norm": 0.87109375, "learning_rate": 2.9204185426120063e-05, "loss": 0.1309, "step": 31508 }, { "epoch": 23.083516483516483, "grad_norm": 1.359375, "learning_rate": 2.9201092415776804e-05, "loss": 0.1381, "step": 31509 }, { "epoch": 23.084249084249084, "grad_norm": 1.2109375, "learning_rate": 2.9197999507483596e-05, "loss": 0.1137, "step": 31510 }, { "epoch": 23.084981684981685, "grad_norm": 1.125, "learning_rate": 2.919490670125529e-05, "loss": 0.114, "step": 31511 }, { "epoch": 23.085714285714285, "grad_norm": 0.890625, "learning_rate": 2.9191813997106666e-05, "loss": 0.1273, "step": 31512 }, { "epoch": 23.086446886446886, "grad_norm": 0.70703125, "learning_rate": 2.9188721395052566e-05, "loss": 0.103, "step": 31513 }, { "epoch": 23.087179487179487, "grad_norm": 1.484375, "learning_rate": 2.9185628895107763e-05, "loss": 0.0952, "step": 31514 }, { "epoch": 23.087912087912088, "grad_norm": 0.9140625, "learning_rate": 2.918253649728715e-05, "loss": 0.1141, "step": 31515 }, { "epoch": 23.08864468864469, "grad_norm": 1.0625, "learning_rate": 2.9179444201605477e-05, "loss": 0.1043, "step": 31516 }, { "epoch": 23.08937728937729, "grad_norm": 0.7890625, "learning_rate": 2.917635200807759e-05, "loss": 0.1113, "step": 31517 }, { "epoch": 23.09010989010989, "grad_norm": 1.2578125, "learning_rate": 2.9173259916718326e-05, "loss": 0.1156, "step": 31518 }, { "epoch": 23.09084249084249, "grad_norm": 1.2109375, "learning_rate": 2.917016792754244e-05, "loss": 0.0954, "step": 31519 }, { "epoch": 23.09157509157509, "grad_norm": 1.1640625, "learning_rate": 2.9167076040564796e-05, "loss": 0.1552, "step": 31520 }, { "epoch": 23.092307692307692, "grad_norm": 0.8671875, "learning_rate": 2.9163984255800186e-05, "loss": 0.105, "step": 31521 }, { "epoch": 23.093040293040293, "grad_norm": 1.3671875, "learning_rate": 2.916089257326346e-05, "loss": 0.1673, "step": 31522 }, { "epoch": 23.093772893772893, "grad_norm": 0.84375, "learning_rate": 2.915780099296937e-05, "loss": 0.1223, "step": 31523 }, { "epoch": 23.094505494505494, "grad_norm": 0.95703125, "learning_rate": 2.915470951493277e-05, "loss": 0.0928, "step": 31524 }, { "epoch": 23.095238095238095, "grad_norm": 1.4453125, "learning_rate": 2.915161813916846e-05, "loss": 0.1865, "step": 31525 }, { "epoch": 23.095970695970696, "grad_norm": 0.546875, "learning_rate": 2.914852686569128e-05, "loss": 0.1108, "step": 31526 }, { "epoch": 23.096703296703296, "grad_norm": 1.2890625, "learning_rate": 2.9145435694515994e-05, "loss": 0.1155, "step": 31527 }, { "epoch": 23.097435897435897, "grad_norm": 0.98828125, "learning_rate": 2.914234462565746e-05, "loss": 0.1328, "step": 31528 }, { "epoch": 23.098168498168498, "grad_norm": 1.8515625, "learning_rate": 2.913925365913044e-05, "loss": 0.1178, "step": 31529 }, { "epoch": 23.0989010989011, "grad_norm": 1.203125, "learning_rate": 2.913616279494977e-05, "loss": 0.142, "step": 31530 }, { "epoch": 23.0996336996337, "grad_norm": 1.3046875, "learning_rate": 2.9133072033130248e-05, "loss": 0.1603, "step": 31531 }, { "epoch": 23.1003663003663, "grad_norm": 1.171875, "learning_rate": 2.9129981373686703e-05, "loss": 0.1272, "step": 31532 }, { "epoch": 23.1010989010989, "grad_norm": 0.86328125, "learning_rate": 2.912689081663395e-05, "loss": 0.1065, "step": 31533 }, { "epoch": 23.1018315018315, "grad_norm": 0.90234375, "learning_rate": 2.9123800361986757e-05, "loss": 0.1229, "step": 31534 }, { "epoch": 23.102564102564102, "grad_norm": 1.3203125, "learning_rate": 2.9120710009759962e-05, "loss": 0.133, "step": 31535 }, { "epoch": 23.103296703296703, "grad_norm": 0.96875, "learning_rate": 2.9117619759968358e-05, "loss": 0.1223, "step": 31536 }, { "epoch": 23.104029304029304, "grad_norm": 1.234375, "learning_rate": 2.9114529612626777e-05, "loss": 0.114, "step": 31537 }, { "epoch": 23.104761904761904, "grad_norm": 0.796875, "learning_rate": 2.9111439567749983e-05, "loss": 0.0913, "step": 31538 }, { "epoch": 23.105494505494505, "grad_norm": 0.76171875, "learning_rate": 2.9108349625352825e-05, "loss": 0.0977, "step": 31539 }, { "epoch": 23.106227106227106, "grad_norm": 1.0625, "learning_rate": 2.9105259785450045e-05, "loss": 0.0842, "step": 31540 }, { "epoch": 23.106959706959707, "grad_norm": 1.7734375, "learning_rate": 2.9102170048056538e-05, "loss": 0.1226, "step": 31541 }, { "epoch": 23.107692307692307, "grad_norm": 0.72265625, "learning_rate": 2.9099080413187025e-05, "loss": 0.1121, "step": 31542 }, { "epoch": 23.108424908424908, "grad_norm": 1.375, "learning_rate": 2.9095990880856355e-05, "loss": 0.1405, "step": 31543 }, { "epoch": 23.10915750915751, "grad_norm": 1.1015625, "learning_rate": 2.909290145107933e-05, "loss": 0.1079, "step": 31544 }, { "epoch": 23.10989010989011, "grad_norm": 1.2265625, "learning_rate": 2.908981212387073e-05, "loss": 0.1174, "step": 31545 }, { "epoch": 23.11062271062271, "grad_norm": 0.9453125, "learning_rate": 2.9086722899245357e-05, "loss": 0.1223, "step": 31546 }, { "epoch": 23.11135531135531, "grad_norm": 0.87109375, "learning_rate": 2.9083633777218033e-05, "loss": 0.1215, "step": 31547 }, { "epoch": 23.11208791208791, "grad_norm": 1.078125, "learning_rate": 2.9080544757803567e-05, "loss": 0.1145, "step": 31548 }, { "epoch": 23.112820512820512, "grad_norm": 1.2578125, "learning_rate": 2.9077455841016727e-05, "loss": 0.1263, "step": 31549 }, { "epoch": 23.113553113553113, "grad_norm": 1.09375, "learning_rate": 2.9074367026872324e-05, "loss": 0.1338, "step": 31550 }, { "epoch": 23.114285714285714, "grad_norm": 1.71875, "learning_rate": 2.9071278315385163e-05, "loss": 0.1412, "step": 31551 }, { "epoch": 23.115018315018315, "grad_norm": 1.5, "learning_rate": 2.906818970657006e-05, "loss": 0.1045, "step": 31552 }, { "epoch": 23.115750915750915, "grad_norm": 1.265625, "learning_rate": 2.9065101200441785e-05, "loss": 0.1149, "step": 31553 }, { "epoch": 23.116483516483516, "grad_norm": 1.875, "learning_rate": 2.9062012797015154e-05, "loss": 0.1413, "step": 31554 }, { "epoch": 23.117216117216117, "grad_norm": 0.99609375, "learning_rate": 2.9058924496304945e-05, "loss": 0.1318, "step": 31555 }, { "epoch": 23.117948717948718, "grad_norm": 0.96875, "learning_rate": 2.9055836298325965e-05, "loss": 0.1165, "step": 31556 }, { "epoch": 23.11868131868132, "grad_norm": 1.2890625, "learning_rate": 2.905274820309301e-05, "loss": 0.1176, "step": 31557 }, { "epoch": 23.11941391941392, "grad_norm": 0.94921875, "learning_rate": 2.904966021062089e-05, "loss": 0.117, "step": 31558 }, { "epoch": 23.12014652014652, "grad_norm": 0.81640625, "learning_rate": 2.90465723209244e-05, "loss": 0.1072, "step": 31559 }, { "epoch": 23.12087912087912, "grad_norm": 1.546875, "learning_rate": 2.9043484534018312e-05, "loss": 0.1501, "step": 31560 }, { "epoch": 23.12161172161172, "grad_norm": 0.72265625, "learning_rate": 2.904039684991743e-05, "loss": 0.1522, "step": 31561 }, { "epoch": 23.122344322344322, "grad_norm": 1.609375, "learning_rate": 2.903730926863655e-05, "loss": 0.1421, "step": 31562 }, { "epoch": 23.123076923076923, "grad_norm": 0.92578125, "learning_rate": 2.903422179019048e-05, "loss": 0.0923, "step": 31563 }, { "epoch": 23.123809523809523, "grad_norm": 1.234375, "learning_rate": 2.903113441459399e-05, "loss": 0.1338, "step": 31564 }, { "epoch": 23.124542124542124, "grad_norm": 0.99609375, "learning_rate": 2.9028047141861897e-05, "loss": 0.1432, "step": 31565 }, { "epoch": 23.125274725274725, "grad_norm": 0.9609375, "learning_rate": 2.902495997200894e-05, "loss": 0.1319, "step": 31566 }, { "epoch": 23.126007326007326, "grad_norm": 1.3984375, "learning_rate": 2.9021872905049995e-05, "loss": 0.1316, "step": 31567 }, { "epoch": 23.126739926739926, "grad_norm": 0.9609375, "learning_rate": 2.9018785940999784e-05, "loss": 0.1197, "step": 31568 }, { "epoch": 23.127472527472527, "grad_norm": 1.2578125, "learning_rate": 2.901569907987312e-05, "loss": 0.127, "step": 31569 }, { "epoch": 23.128205128205128, "grad_norm": 1.640625, "learning_rate": 2.9012612321684824e-05, "loss": 0.1326, "step": 31570 }, { "epoch": 23.12893772893773, "grad_norm": 0.7734375, "learning_rate": 2.900952566644963e-05, "loss": 0.139, "step": 31571 }, { "epoch": 23.12967032967033, "grad_norm": 1.1484375, "learning_rate": 2.900643911418235e-05, "loss": 0.136, "step": 31572 }, { "epoch": 23.13040293040293, "grad_norm": 0.6484375, "learning_rate": 2.9003352664897776e-05, "loss": 0.1101, "step": 31573 }, { "epoch": 23.13113553113553, "grad_norm": 1.015625, "learning_rate": 2.9000266318610712e-05, "loss": 0.1079, "step": 31574 }, { "epoch": 23.13186813186813, "grad_norm": 1.5625, "learning_rate": 2.8997180075335923e-05, "loss": 0.1224, "step": 31575 }, { "epoch": 23.132600732600732, "grad_norm": 0.8203125, "learning_rate": 2.8994093935088214e-05, "loss": 0.131, "step": 31576 }, { "epoch": 23.133333333333333, "grad_norm": 1.0234375, "learning_rate": 2.899100789788232e-05, "loss": 0.1145, "step": 31577 }, { "epoch": 23.134065934065934, "grad_norm": 0.95703125, "learning_rate": 2.898792196373311e-05, "loss": 0.1245, "step": 31578 }, { "epoch": 23.134798534798534, "grad_norm": 0.734375, "learning_rate": 2.8984836132655313e-05, "loss": 0.1096, "step": 31579 }, { "epoch": 23.135531135531135, "grad_norm": 0.90234375, "learning_rate": 2.898175040466372e-05, "loss": 0.1142, "step": 31580 }, { "epoch": 23.136263736263736, "grad_norm": 0.6953125, "learning_rate": 2.897866477977315e-05, "loss": 0.1094, "step": 31581 }, { "epoch": 23.136996336996337, "grad_norm": 1.4453125, "learning_rate": 2.8975579257998348e-05, "loss": 0.1312, "step": 31582 }, { "epoch": 23.137728937728937, "grad_norm": 0.94140625, "learning_rate": 2.8972493839354104e-05, "loss": 0.1315, "step": 31583 }, { "epoch": 23.138461538461538, "grad_norm": 1.1875, "learning_rate": 2.8969408523855213e-05, "loss": 0.1106, "step": 31584 }, { "epoch": 23.13919413919414, "grad_norm": 1.078125, "learning_rate": 2.896632331151647e-05, "loss": 0.1424, "step": 31585 }, { "epoch": 23.13992673992674, "grad_norm": 1.6171875, "learning_rate": 2.8963238202352622e-05, "loss": 0.0958, "step": 31586 }, { "epoch": 23.14065934065934, "grad_norm": 0.7578125, "learning_rate": 2.8960153196378466e-05, "loss": 0.1066, "step": 31587 }, { "epoch": 23.14139194139194, "grad_norm": 0.81640625, "learning_rate": 2.8957068293608796e-05, "loss": 0.116, "step": 31588 }, { "epoch": 23.14212454212454, "grad_norm": 0.87890625, "learning_rate": 2.8953983494058402e-05, "loss": 0.1284, "step": 31589 }, { "epoch": 23.142857142857142, "grad_norm": 1.7734375, "learning_rate": 2.8950898797742025e-05, "loss": 0.1732, "step": 31590 }, { "epoch": 23.143589743589743, "grad_norm": 1.453125, "learning_rate": 2.894781420467448e-05, "loss": 0.127, "step": 31591 }, { "epoch": 23.144322344322344, "grad_norm": 1.0234375, "learning_rate": 2.8944729714870495e-05, "loss": 0.1401, "step": 31592 }, { "epoch": 23.145054945054945, "grad_norm": 1.0859375, "learning_rate": 2.894164532834493e-05, "loss": 0.1555, "step": 31593 }, { "epoch": 23.145787545787545, "grad_norm": 1.9609375, "learning_rate": 2.8938561045112495e-05, "loss": 0.1852, "step": 31594 }, { "epoch": 23.146520146520146, "grad_norm": 0.69140625, "learning_rate": 2.8935476865188e-05, "loss": 0.121, "step": 31595 }, { "epoch": 23.147252747252747, "grad_norm": 0.91015625, "learning_rate": 2.8932392788586226e-05, "loss": 0.1259, "step": 31596 }, { "epoch": 23.147985347985347, "grad_norm": 1.6796875, "learning_rate": 2.8929308815321897e-05, "loss": 0.1648, "step": 31597 }, { "epoch": 23.148717948717948, "grad_norm": 1.1015625, "learning_rate": 2.8926224945409874e-05, "loss": 0.1334, "step": 31598 }, { "epoch": 23.14945054945055, "grad_norm": 1.171875, "learning_rate": 2.8923141178864878e-05, "loss": 0.1325, "step": 31599 }, { "epoch": 23.15018315018315, "grad_norm": 1.6328125, "learning_rate": 2.892005751570171e-05, "loss": 0.1518, "step": 31600 }, { "epoch": 23.15091575091575, "grad_norm": 1.28125, "learning_rate": 2.8916973955935107e-05, "loss": 0.1237, "step": 31601 }, { "epoch": 23.15164835164835, "grad_norm": 0.69140625, "learning_rate": 2.8913890499579887e-05, "loss": 0.1329, "step": 31602 }, { "epoch": 23.152380952380952, "grad_norm": 1.3046875, "learning_rate": 2.8910807146650765e-05, "loss": 0.1403, "step": 31603 }, { "epoch": 23.153113553113553, "grad_norm": 2.15625, "learning_rate": 2.890772389716259e-05, "loss": 0.1302, "step": 31604 }, { "epoch": 23.153846153846153, "grad_norm": 0.78515625, "learning_rate": 2.8904640751130085e-05, "loss": 0.1418, "step": 31605 }, { "epoch": 23.154578754578754, "grad_norm": 1.1484375, "learning_rate": 2.8901557708568035e-05, "loss": 0.1285, "step": 31606 }, { "epoch": 23.155311355311355, "grad_norm": 1.5078125, "learning_rate": 2.8898474769491228e-05, "loss": 0.1107, "step": 31607 }, { "epoch": 23.156043956043955, "grad_norm": 1.078125, "learning_rate": 2.8895391933914407e-05, "loss": 0.1206, "step": 31608 }, { "epoch": 23.156776556776556, "grad_norm": 1.125, "learning_rate": 2.8892309201852348e-05, "loss": 0.1392, "step": 31609 }, { "epoch": 23.157509157509157, "grad_norm": 0.80859375, "learning_rate": 2.8889226573319826e-05, "loss": 0.1115, "step": 31610 }, { "epoch": 23.158241758241758, "grad_norm": 1.1171875, "learning_rate": 2.8886144048331647e-05, "loss": 0.1347, "step": 31611 }, { "epoch": 23.15897435897436, "grad_norm": 1.15625, "learning_rate": 2.8883061626902513e-05, "loss": 0.0845, "step": 31612 }, { "epoch": 23.15970695970696, "grad_norm": 1.03125, "learning_rate": 2.887997930904724e-05, "loss": 0.0944, "step": 31613 }, { "epoch": 23.16043956043956, "grad_norm": 0.85546875, "learning_rate": 2.8876897094780574e-05, "loss": 0.1157, "step": 31614 }, { "epoch": 23.16117216117216, "grad_norm": 0.765625, "learning_rate": 2.8873814984117308e-05, "loss": 0.1043, "step": 31615 }, { "epoch": 23.16190476190476, "grad_norm": 1.6171875, "learning_rate": 2.8870732977072192e-05, "loss": 0.1216, "step": 31616 }, { "epoch": 23.162637362637362, "grad_norm": 1.640625, "learning_rate": 2.8867651073660002e-05, "loss": 0.1261, "step": 31617 }, { "epoch": 23.163369963369963, "grad_norm": 0.9921875, "learning_rate": 2.886456927389546e-05, "loss": 0.1424, "step": 31618 }, { "epoch": 23.164102564102564, "grad_norm": 0.984375, "learning_rate": 2.8861487577793407e-05, "loss": 0.1213, "step": 31619 }, { "epoch": 23.164835164835164, "grad_norm": 0.8984375, "learning_rate": 2.8858405985368547e-05, "loss": 0.1017, "step": 31620 }, { "epoch": 23.165567765567765, "grad_norm": 1.390625, "learning_rate": 2.885532449663567e-05, "loss": 0.1281, "step": 31621 }, { "epoch": 23.166300366300366, "grad_norm": 1.078125, "learning_rate": 2.8852243111609563e-05, "loss": 0.1156, "step": 31622 }, { "epoch": 23.167032967032966, "grad_norm": 0.8671875, "learning_rate": 2.8849161830304926e-05, "loss": 0.1079, "step": 31623 }, { "epoch": 23.167765567765567, "grad_norm": 0.92578125, "learning_rate": 2.88460806527366e-05, "loss": 0.1161, "step": 31624 }, { "epoch": 23.168498168498168, "grad_norm": 0.9609375, "learning_rate": 2.884299957891929e-05, "loss": 0.1364, "step": 31625 }, { "epoch": 23.16923076923077, "grad_norm": 1.328125, "learning_rate": 2.88399186088678e-05, "loss": 0.1165, "step": 31626 }, { "epoch": 23.16996336996337, "grad_norm": 0.859375, "learning_rate": 2.883683774259685e-05, "loss": 0.1113, "step": 31627 }, { "epoch": 23.17069597069597, "grad_norm": 0.91796875, "learning_rate": 2.8833756980121245e-05, "loss": 0.0935, "step": 31628 }, { "epoch": 23.17142857142857, "grad_norm": 0.859375, "learning_rate": 2.883067632145568e-05, "loss": 0.1053, "step": 31629 }, { "epoch": 23.17216117216117, "grad_norm": 0.6640625, "learning_rate": 2.8827595766615e-05, "loss": 0.0916, "step": 31630 }, { "epoch": 23.172893772893772, "grad_norm": 0.91796875, "learning_rate": 2.8824515315613898e-05, "loss": 0.1054, "step": 31631 }, { "epoch": 23.173626373626373, "grad_norm": 0.91015625, "learning_rate": 2.882143496846716e-05, "loss": 0.1132, "step": 31632 }, { "epoch": 23.174358974358974, "grad_norm": 1.078125, "learning_rate": 2.8818354725189568e-05, "loss": 0.1231, "step": 31633 }, { "epoch": 23.175091575091574, "grad_norm": 1.90625, "learning_rate": 2.881527458579583e-05, "loss": 0.1391, "step": 31634 }, { "epoch": 23.175824175824175, "grad_norm": 1.3125, "learning_rate": 2.8812194550300736e-05, "loss": 0.1285, "step": 31635 }, { "epoch": 23.176556776556776, "grad_norm": 1.4921875, "learning_rate": 2.880911461871904e-05, "loss": 0.1214, "step": 31636 }, { "epoch": 23.177289377289377, "grad_norm": 1.0078125, "learning_rate": 2.880603479106551e-05, "loss": 0.1218, "step": 31637 }, { "epoch": 23.178021978021977, "grad_norm": 1.1484375, "learning_rate": 2.880295506735487e-05, "loss": 0.1176, "step": 31638 }, { "epoch": 23.178754578754578, "grad_norm": 0.95703125, "learning_rate": 2.8799875447601893e-05, "loss": 0.1421, "step": 31639 }, { "epoch": 23.17948717948718, "grad_norm": 1.2734375, "learning_rate": 2.879679593182134e-05, "loss": 0.1398, "step": 31640 }, { "epoch": 23.18021978021978, "grad_norm": 0.8671875, "learning_rate": 2.879371652002798e-05, "loss": 0.113, "step": 31641 }, { "epoch": 23.18095238095238, "grad_norm": 1.8359375, "learning_rate": 2.879063721223654e-05, "loss": 0.1274, "step": 31642 }, { "epoch": 23.18168498168498, "grad_norm": 0.96484375, "learning_rate": 2.8787558008461792e-05, "loss": 0.1098, "step": 31643 }, { "epoch": 23.18241758241758, "grad_norm": 1.140625, "learning_rate": 2.8784478908718442e-05, "loss": 0.1353, "step": 31644 }, { "epoch": 23.183150183150182, "grad_norm": 0.89453125, "learning_rate": 2.878139991302133e-05, "loss": 0.1223, "step": 31645 }, { "epoch": 23.183882783882783, "grad_norm": 0.828125, "learning_rate": 2.8778321021385128e-05, "loss": 0.0975, "step": 31646 }, { "epoch": 23.184615384615384, "grad_norm": 1.3984375, "learning_rate": 2.8775242233824628e-05, "loss": 0.1735, "step": 31647 }, { "epoch": 23.185347985347985, "grad_norm": 1.4609375, "learning_rate": 2.8772163550354593e-05, "loss": 0.1348, "step": 31648 }, { "epoch": 23.186080586080585, "grad_norm": 0.9140625, "learning_rate": 2.8769084970989722e-05, "loss": 0.101, "step": 31649 }, { "epoch": 23.186813186813186, "grad_norm": 0.85546875, "learning_rate": 2.8766006495744835e-05, "loss": 0.1158, "step": 31650 }, { "epoch": 23.187545787545787, "grad_norm": 0.95703125, "learning_rate": 2.8762928124634625e-05, "loss": 0.1115, "step": 31651 }, { "epoch": 23.188278388278388, "grad_norm": 1.0390625, "learning_rate": 2.8759849857673876e-05, "loss": 0.1139, "step": 31652 }, { "epoch": 23.18901098901099, "grad_norm": 0.7265625, "learning_rate": 2.875677169487731e-05, "loss": 0.0937, "step": 31653 }, { "epoch": 23.18974358974359, "grad_norm": 0.703125, "learning_rate": 2.8753693636259704e-05, "loss": 0.0974, "step": 31654 }, { "epoch": 23.19047619047619, "grad_norm": 1.0703125, "learning_rate": 2.875061568183576e-05, "loss": 0.1303, "step": 31655 }, { "epoch": 23.19120879120879, "grad_norm": 0.78515625, "learning_rate": 2.8747537831620285e-05, "loss": 0.1057, "step": 31656 }, { "epoch": 23.19194139194139, "grad_norm": 1.3046875, "learning_rate": 2.8744460085627983e-05, "loss": 0.1371, "step": 31657 }, { "epoch": 23.192673992673992, "grad_norm": 0.99609375, "learning_rate": 2.8741382443873616e-05, "loss": 0.0986, "step": 31658 }, { "epoch": 23.193406593406593, "grad_norm": 1.3515625, "learning_rate": 2.8738304906371942e-05, "loss": 0.1435, "step": 31659 }, { "epoch": 23.194139194139193, "grad_norm": 1.0859375, "learning_rate": 2.8735227473137676e-05, "loss": 0.1258, "step": 31660 }, { "epoch": 23.194871794871794, "grad_norm": 1.40625, "learning_rate": 2.8732150144185586e-05, "loss": 0.1108, "step": 31661 }, { "epoch": 23.195604395604395, "grad_norm": 0.67578125, "learning_rate": 2.8729072919530408e-05, "loss": 0.0863, "step": 31662 }, { "epoch": 23.196336996336996, "grad_norm": 0.86328125, "learning_rate": 2.8725995799186904e-05, "loss": 0.1187, "step": 31663 }, { "epoch": 23.197069597069596, "grad_norm": 1.40625, "learning_rate": 2.872291878316979e-05, "loss": 0.1209, "step": 31664 }, { "epoch": 23.197802197802197, "grad_norm": 0.796875, "learning_rate": 2.8719841871493813e-05, "loss": 0.1389, "step": 31665 }, { "epoch": 23.198534798534798, "grad_norm": 0.90625, "learning_rate": 2.8716765064173727e-05, "loss": 0.1028, "step": 31666 }, { "epoch": 23.1992673992674, "grad_norm": 1.484375, "learning_rate": 2.8713688361224288e-05, "loss": 0.1354, "step": 31667 }, { "epoch": 23.2, "grad_norm": 1.6484375, "learning_rate": 2.87106117626602e-05, "loss": 0.142, "step": 31668 }, { "epoch": 23.2007326007326, "grad_norm": 1.3125, "learning_rate": 2.870753526849625e-05, "loss": 0.1272, "step": 31669 }, { "epoch": 23.2014652014652, "grad_norm": 0.93359375, "learning_rate": 2.8704458878747102e-05, "loss": 0.0909, "step": 31670 }, { "epoch": 23.2021978021978, "grad_norm": 1.3125, "learning_rate": 2.8701382593427593e-05, "loss": 0.1443, "step": 31671 }, { "epoch": 23.202930402930402, "grad_norm": 1.1484375, "learning_rate": 2.86983064125524e-05, "loss": 0.108, "step": 31672 }, { "epoch": 23.203663003663003, "grad_norm": 0.796875, "learning_rate": 2.8695230336136275e-05, "loss": 0.1018, "step": 31673 }, { "epoch": 23.204395604395604, "grad_norm": 1.6875, "learning_rate": 2.8692154364193973e-05, "loss": 0.1106, "step": 31674 }, { "epoch": 23.205128205128204, "grad_norm": 1.1875, "learning_rate": 2.868907849674019e-05, "loss": 0.1261, "step": 31675 }, { "epoch": 23.205860805860805, "grad_norm": 1.5546875, "learning_rate": 2.868600273378972e-05, "loss": 0.1178, "step": 31676 }, { "epoch": 23.206593406593406, "grad_norm": 0.77734375, "learning_rate": 2.8682927075357262e-05, "loss": 0.1289, "step": 31677 }, { "epoch": 23.207326007326007, "grad_norm": 1.203125, "learning_rate": 2.8679851521457578e-05, "loss": 0.1263, "step": 31678 }, { "epoch": 23.208058608058607, "grad_norm": 0.77734375, "learning_rate": 2.8676776072105364e-05, "loss": 0.1167, "step": 31679 }, { "epoch": 23.208791208791208, "grad_norm": 1.2421875, "learning_rate": 2.867370072731541e-05, "loss": 0.124, "step": 31680 }, { "epoch": 23.20952380952381, "grad_norm": 1.4296875, "learning_rate": 2.8670625487102378e-05, "loss": 0.1496, "step": 31681 }, { "epoch": 23.21025641025641, "grad_norm": 0.98046875, "learning_rate": 2.866755035148108e-05, "loss": 0.1334, "step": 31682 }, { "epoch": 23.21098901098901, "grad_norm": 0.9765625, "learning_rate": 2.86644753204662e-05, "loss": 0.1312, "step": 31683 }, { "epoch": 23.21172161172161, "grad_norm": 0.96484375, "learning_rate": 2.866140039407248e-05, "loss": 0.1073, "step": 31684 }, { "epoch": 23.21245421245421, "grad_norm": 1.421875, "learning_rate": 2.8658325572314692e-05, "loss": 0.1354, "step": 31685 }, { "epoch": 23.213186813186812, "grad_norm": 0.765625, "learning_rate": 2.865525085520751e-05, "loss": 0.1444, "step": 31686 }, { "epoch": 23.213919413919413, "grad_norm": 1.2109375, "learning_rate": 2.8652176242765688e-05, "loss": 0.0887, "step": 31687 }, { "epoch": 23.214652014652014, "grad_norm": 0.796875, "learning_rate": 2.864910173500397e-05, "loss": 0.1301, "step": 31688 }, { "epoch": 23.215384615384615, "grad_norm": 1.140625, "learning_rate": 2.864602733193709e-05, "loss": 0.1246, "step": 31689 }, { "epoch": 23.216117216117215, "grad_norm": 1.2578125, "learning_rate": 2.8642953033579752e-05, "loss": 0.0932, "step": 31690 }, { "epoch": 23.216849816849816, "grad_norm": 1.078125, "learning_rate": 2.86398788399467e-05, "loss": 0.1378, "step": 31691 }, { "epoch": 23.217582417582417, "grad_norm": 0.6875, "learning_rate": 2.8636804751052663e-05, "loss": 0.1126, "step": 31692 }, { "epoch": 23.218315018315018, "grad_norm": 1.0859375, "learning_rate": 2.8633730766912398e-05, "loss": 0.1146, "step": 31693 }, { "epoch": 23.21904761904762, "grad_norm": 0.9375, "learning_rate": 2.863065688754058e-05, "loss": 0.1096, "step": 31694 }, { "epoch": 23.21978021978022, "grad_norm": 1.1328125, "learning_rate": 2.8627583112951966e-05, "loss": 0.129, "step": 31695 }, { "epoch": 23.22051282051282, "grad_norm": 0.9609375, "learning_rate": 2.8624509443161283e-05, "loss": 0.1482, "step": 31696 }, { "epoch": 23.22124542124542, "grad_norm": 0.98828125, "learning_rate": 2.8621435878183278e-05, "loss": 0.1306, "step": 31697 }, { "epoch": 23.22197802197802, "grad_norm": 0.890625, "learning_rate": 2.8618362418032636e-05, "loss": 0.1088, "step": 31698 }, { "epoch": 23.222710622710622, "grad_norm": 1.3046875, "learning_rate": 2.8615289062724098e-05, "loss": 0.1575, "step": 31699 }, { "epoch": 23.223443223443223, "grad_norm": 1.1484375, "learning_rate": 2.861221581227242e-05, "loss": 0.1054, "step": 31700 }, { "epoch": 23.224175824175823, "grad_norm": 0.98046875, "learning_rate": 2.8609142666692262e-05, "loss": 0.1167, "step": 31701 }, { "epoch": 23.224908424908424, "grad_norm": 0.984375, "learning_rate": 2.860606962599842e-05, "loss": 0.1225, "step": 31702 }, { "epoch": 23.225641025641025, "grad_norm": 0.80859375, "learning_rate": 2.8602996690205576e-05, "loss": 0.1232, "step": 31703 }, { "epoch": 23.226373626373626, "grad_norm": 0.93359375, "learning_rate": 2.8599923859328474e-05, "loss": 0.1137, "step": 31704 }, { "epoch": 23.227106227106226, "grad_norm": 0.90625, "learning_rate": 2.8596851133381807e-05, "loss": 0.1308, "step": 31705 }, { "epoch": 23.227838827838827, "grad_norm": 1.1484375, "learning_rate": 2.8593778512380344e-05, "loss": 0.1592, "step": 31706 }, { "epoch": 23.228571428571428, "grad_norm": 1.078125, "learning_rate": 2.8590705996338727e-05, "loss": 0.1428, "step": 31707 }, { "epoch": 23.22930402930403, "grad_norm": 0.8359375, "learning_rate": 2.8587633585271785e-05, "loss": 0.1169, "step": 31708 }, { "epoch": 23.23003663003663, "grad_norm": 1.296875, "learning_rate": 2.8584561279194156e-05, "loss": 0.1299, "step": 31709 }, { "epoch": 23.23076923076923, "grad_norm": 0.953125, "learning_rate": 2.858148907812059e-05, "loss": 0.113, "step": 31710 }, { "epoch": 23.23150183150183, "grad_norm": 1.25, "learning_rate": 2.857841698206582e-05, "loss": 0.1331, "step": 31711 }, { "epoch": 23.23223443223443, "grad_norm": 0.96484375, "learning_rate": 2.8575344991044518e-05, "loss": 0.1291, "step": 31712 }, { "epoch": 23.232967032967032, "grad_norm": 0.96484375, "learning_rate": 2.8572273105071473e-05, "loss": 0.1413, "step": 31713 }, { "epoch": 23.233699633699633, "grad_norm": 0.7890625, "learning_rate": 2.856920132416135e-05, "loss": 0.1169, "step": 31714 }, { "epoch": 23.234432234432234, "grad_norm": 1.1171875, "learning_rate": 2.8566129648328888e-05, "loss": 0.1511, "step": 31715 }, { "epoch": 23.235164835164834, "grad_norm": 1.6171875, "learning_rate": 2.8563058077588793e-05, "loss": 0.14, "step": 31716 }, { "epoch": 23.235897435897435, "grad_norm": 0.69140625, "learning_rate": 2.8559986611955784e-05, "loss": 0.1313, "step": 31717 }, { "epoch": 23.236630036630036, "grad_norm": 0.77734375, "learning_rate": 2.855691525144459e-05, "loss": 0.1174, "step": 31718 }, { "epoch": 23.237362637362637, "grad_norm": 0.96484375, "learning_rate": 2.8553843996069924e-05, "loss": 0.1144, "step": 31719 }, { "epoch": 23.238095238095237, "grad_norm": 0.66015625, "learning_rate": 2.855077284584648e-05, "loss": 0.1212, "step": 31720 }, { "epoch": 23.238827838827838, "grad_norm": 1.078125, "learning_rate": 2.8547701800788994e-05, "loss": 0.1122, "step": 31721 }, { "epoch": 23.23956043956044, "grad_norm": 1.2578125, "learning_rate": 2.8544630860912175e-05, "loss": 0.1354, "step": 31722 }, { "epoch": 23.24029304029304, "grad_norm": 1.6171875, "learning_rate": 2.8541560026230752e-05, "loss": 0.1048, "step": 31723 }, { "epoch": 23.24102564102564, "grad_norm": 0.9296875, "learning_rate": 2.8538489296759407e-05, "loss": 0.1213, "step": 31724 }, { "epoch": 23.24175824175824, "grad_norm": 0.8828125, "learning_rate": 2.853541867251287e-05, "loss": 0.1144, "step": 31725 }, { "epoch": 23.24249084249084, "grad_norm": 2.15625, "learning_rate": 2.8532348153505866e-05, "loss": 0.1505, "step": 31726 }, { "epoch": 23.243223443223442, "grad_norm": 1.109375, "learning_rate": 2.852927773975306e-05, "loss": 0.1405, "step": 31727 }, { "epoch": 23.243956043956043, "grad_norm": 1.0703125, "learning_rate": 2.8526207431269235e-05, "loss": 0.1176, "step": 31728 }, { "epoch": 23.244688644688644, "grad_norm": 0.984375, "learning_rate": 2.8523137228069048e-05, "loss": 0.121, "step": 31729 }, { "epoch": 23.245421245421245, "grad_norm": 1.3515625, "learning_rate": 2.8520067130167237e-05, "loss": 0.1269, "step": 31730 }, { "epoch": 23.246153846153845, "grad_norm": 0.8515625, "learning_rate": 2.8516997137578483e-05, "loss": 0.1173, "step": 31731 }, { "epoch": 23.246886446886446, "grad_norm": 1.109375, "learning_rate": 2.8513927250317533e-05, "loss": 0.1097, "step": 31732 }, { "epoch": 23.247619047619047, "grad_norm": 1.25, "learning_rate": 2.851085746839904e-05, "loss": 0.1355, "step": 31733 }, { "epoch": 23.248351648351647, "grad_norm": 1.0546875, "learning_rate": 2.8507787791837783e-05, "loss": 0.1242, "step": 31734 }, { "epoch": 23.249084249084248, "grad_norm": 1.5, "learning_rate": 2.8504718220648414e-05, "loss": 0.1155, "step": 31735 }, { "epoch": 23.24981684981685, "grad_norm": 0.875, "learning_rate": 2.8501648754845655e-05, "loss": 0.1322, "step": 31736 }, { "epoch": 23.25054945054945, "grad_norm": 1.5859375, "learning_rate": 2.8498579394444247e-05, "loss": 0.1293, "step": 31737 }, { "epoch": 23.25128205128205, "grad_norm": 1.21875, "learning_rate": 2.8495510139458824e-05, "loss": 0.118, "step": 31738 }, { "epoch": 23.25201465201465, "grad_norm": 0.76953125, "learning_rate": 2.8492440989904174e-05, "loss": 0.1183, "step": 31739 }, { "epoch": 23.252747252747252, "grad_norm": 0.765625, "learning_rate": 2.8489371945794945e-05, "loss": 0.1189, "step": 31740 }, { "epoch": 23.253479853479853, "grad_norm": 1.1484375, "learning_rate": 2.8486303007145882e-05, "loss": 0.1248, "step": 31741 }, { "epoch": 23.254212454212453, "grad_norm": 0.8359375, "learning_rate": 2.8483234173971645e-05, "loss": 0.0983, "step": 31742 }, { "epoch": 23.254945054945054, "grad_norm": 1.1796875, "learning_rate": 2.8480165446286965e-05, "loss": 0.1201, "step": 31743 }, { "epoch": 23.255677655677655, "grad_norm": 1.0234375, "learning_rate": 2.847709682410654e-05, "loss": 0.0889, "step": 31744 }, { "epoch": 23.256410256410255, "grad_norm": 1.0234375, "learning_rate": 2.8474028307445095e-05, "loss": 0.1007, "step": 31745 }, { "epoch": 23.257142857142856, "grad_norm": 0.9375, "learning_rate": 2.8470959896317285e-05, "loss": 0.12, "step": 31746 }, { "epoch": 23.257875457875457, "grad_norm": 1.0234375, "learning_rate": 2.846789159073785e-05, "loss": 0.1301, "step": 31747 }, { "epoch": 23.258608058608058, "grad_norm": 0.99609375, "learning_rate": 2.8464823390721464e-05, "loss": 0.1024, "step": 31748 }, { "epoch": 23.25934065934066, "grad_norm": 1.2421875, "learning_rate": 2.846175529628287e-05, "loss": 0.1192, "step": 31749 }, { "epoch": 23.26007326007326, "grad_norm": 1.296875, "learning_rate": 2.8458687307436727e-05, "loss": 0.1265, "step": 31750 }, { "epoch": 23.26080586080586, "grad_norm": 1.4375, "learning_rate": 2.8455619424197737e-05, "loss": 0.1377, "step": 31751 }, { "epoch": 23.26153846153846, "grad_norm": 1.4765625, "learning_rate": 2.8452551646580637e-05, "loss": 0.0933, "step": 31752 }, { "epoch": 23.26227106227106, "grad_norm": 1.109375, "learning_rate": 2.8449483974600056e-05, "loss": 0.1186, "step": 31753 }, { "epoch": 23.263003663003662, "grad_norm": 1.0390625, "learning_rate": 2.844641640827078e-05, "loss": 0.1087, "step": 31754 }, { "epoch": 23.263736263736263, "grad_norm": 1.125, "learning_rate": 2.844334894760744e-05, "loss": 0.0944, "step": 31755 }, { "epoch": 23.264468864468864, "grad_norm": 1.3828125, "learning_rate": 2.844028159262478e-05, "loss": 0.1251, "step": 31756 }, { "epoch": 23.265201465201464, "grad_norm": 1.1875, "learning_rate": 2.8437214343337454e-05, "loss": 0.1649, "step": 31757 }, { "epoch": 23.265934065934065, "grad_norm": 0.99609375, "learning_rate": 2.8434147199760193e-05, "loss": 0.1065, "step": 31758 }, { "epoch": 23.266666666666666, "grad_norm": 0.71875, "learning_rate": 2.8431080161907632e-05, "loss": 0.1048, "step": 31759 }, { "epoch": 23.267399267399266, "grad_norm": 1.0859375, "learning_rate": 2.842801322979456e-05, "loss": 0.1261, "step": 31760 }, { "epoch": 23.268131868131867, "grad_norm": 0.71484375, "learning_rate": 2.842494640343559e-05, "loss": 0.1099, "step": 31761 }, { "epoch": 23.268864468864468, "grad_norm": 0.96875, "learning_rate": 2.842187968284546e-05, "loss": 0.1374, "step": 31762 }, { "epoch": 23.26959706959707, "grad_norm": 1.0625, "learning_rate": 2.841881306803887e-05, "loss": 0.1001, "step": 31763 }, { "epoch": 23.27032967032967, "grad_norm": 1.0078125, "learning_rate": 2.8415746559030455e-05, "loss": 0.1266, "step": 31764 }, { "epoch": 23.27106227106227, "grad_norm": 0.62890625, "learning_rate": 2.8412680155834986e-05, "loss": 0.133, "step": 31765 }, { "epoch": 23.27179487179487, "grad_norm": 1.328125, "learning_rate": 2.8409613858467098e-05, "loss": 0.1155, "step": 31766 }, { "epoch": 23.27252747252747, "grad_norm": 1.0078125, "learning_rate": 2.8406547666941523e-05, "loss": 0.1413, "step": 31767 }, { "epoch": 23.273260073260072, "grad_norm": 1.484375, "learning_rate": 2.8403481581272912e-05, "loss": 0.1307, "step": 31768 }, { "epoch": 23.273992673992673, "grad_norm": 1.3359375, "learning_rate": 2.8400415601475973e-05, "loss": 0.1049, "step": 31769 }, { "epoch": 23.274725274725274, "grad_norm": 0.7421875, "learning_rate": 2.839734972756539e-05, "loss": 0.1119, "step": 31770 }, { "epoch": 23.275457875457874, "grad_norm": 1.2421875, "learning_rate": 2.8394283959555894e-05, "loss": 0.1217, "step": 31771 }, { "epoch": 23.276190476190475, "grad_norm": 1.4765625, "learning_rate": 2.8391218297462114e-05, "loss": 0.1501, "step": 31772 }, { "epoch": 23.276923076923076, "grad_norm": 1.3828125, "learning_rate": 2.8388152741298762e-05, "loss": 0.1316, "step": 31773 }, { "epoch": 23.277655677655677, "grad_norm": 0.96484375, "learning_rate": 2.8385087291080536e-05, "loss": 0.0977, "step": 31774 }, { "epoch": 23.278388278388277, "grad_norm": 0.9375, "learning_rate": 2.838202194682213e-05, "loss": 0.1268, "step": 31775 }, { "epoch": 23.279120879120878, "grad_norm": 1.3984375, "learning_rate": 2.8378956708538202e-05, "loss": 0.1814, "step": 31776 }, { "epoch": 23.27985347985348, "grad_norm": 1.4921875, "learning_rate": 2.8375891576243445e-05, "loss": 0.143, "step": 31777 }, { "epoch": 23.28058608058608, "grad_norm": 1.1015625, "learning_rate": 2.8372826549952585e-05, "loss": 0.1091, "step": 31778 }, { "epoch": 23.28131868131868, "grad_norm": 1.375, "learning_rate": 2.8369761629680235e-05, "loss": 0.155, "step": 31779 }, { "epoch": 23.28205128205128, "grad_norm": 1.203125, "learning_rate": 2.8366696815441163e-05, "loss": 0.1371, "step": 31780 }, { "epoch": 23.282783882783882, "grad_norm": 1.4375, "learning_rate": 2.836363210724999e-05, "loss": 0.1097, "step": 31781 }, { "epoch": 23.283516483516483, "grad_norm": 1.09375, "learning_rate": 2.8360567505121444e-05, "loss": 0.1539, "step": 31782 }, { "epoch": 23.284249084249083, "grad_norm": 0.92578125, "learning_rate": 2.8357503009070163e-05, "loss": 0.1112, "step": 31783 }, { "epoch": 23.284981684981684, "grad_norm": 1.15625, "learning_rate": 2.8354438619110874e-05, "loss": 0.1096, "step": 31784 }, { "epoch": 23.285714285714285, "grad_norm": 1.1953125, "learning_rate": 2.8351374335258202e-05, "loss": 0.1391, "step": 31785 }, { "epoch": 23.286446886446885, "grad_norm": 1.1640625, "learning_rate": 2.834831015752691e-05, "loss": 0.0974, "step": 31786 }, { "epoch": 23.287179487179486, "grad_norm": 1.015625, "learning_rate": 2.8345246085931608e-05, "loss": 0.1098, "step": 31787 }, { "epoch": 23.287912087912087, "grad_norm": 0.8203125, "learning_rate": 2.834218212048701e-05, "loss": 0.1175, "step": 31788 }, { "epoch": 23.288644688644688, "grad_norm": 1.0859375, "learning_rate": 2.8339118261207814e-05, "loss": 0.1012, "step": 31789 }, { "epoch": 23.28937728937729, "grad_norm": 1.3203125, "learning_rate": 2.8336054508108633e-05, "loss": 0.1213, "step": 31790 }, { "epoch": 23.29010989010989, "grad_norm": 1.3125, "learning_rate": 2.8332990861204233e-05, "loss": 0.1379, "step": 31791 }, { "epoch": 23.29084249084249, "grad_norm": 1.0390625, "learning_rate": 2.8329927320509232e-05, "loss": 0.1102, "step": 31792 }, { "epoch": 23.29157509157509, "grad_norm": 0.9140625, "learning_rate": 2.832686388603834e-05, "loss": 0.1136, "step": 31793 }, { "epoch": 23.29230769230769, "grad_norm": 0.89453125, "learning_rate": 2.8323800557806214e-05, "loss": 0.1425, "step": 31794 }, { "epoch": 23.293040293040292, "grad_norm": 1.0859375, "learning_rate": 2.8320737335827523e-05, "loss": 0.1093, "step": 31795 }, { "epoch": 23.293772893772893, "grad_norm": 0.96484375, "learning_rate": 2.8317674220116975e-05, "loss": 0.0951, "step": 31796 }, { "epoch": 23.294505494505493, "grad_norm": 1.5625, "learning_rate": 2.8314611210689246e-05, "loss": 0.1164, "step": 31797 }, { "epoch": 23.295238095238094, "grad_norm": 1.0390625, "learning_rate": 2.831154830755897e-05, "loss": 0.1205, "step": 31798 }, { "epoch": 23.295970695970695, "grad_norm": 0.95703125, "learning_rate": 2.830848551074086e-05, "loss": 0.0982, "step": 31799 }, { "epoch": 23.296703296703296, "grad_norm": 1.0, "learning_rate": 2.8305422820249584e-05, "loss": 0.1376, "step": 31800 }, { "epoch": 23.297435897435896, "grad_norm": 0.7578125, "learning_rate": 2.830236023609983e-05, "loss": 0.1167, "step": 31801 }, { "epoch": 23.298168498168497, "grad_norm": 0.78125, "learning_rate": 2.8299297758306233e-05, "loss": 0.1222, "step": 31802 }, { "epoch": 23.298901098901098, "grad_norm": 1.28125, "learning_rate": 2.8296235386883492e-05, "loss": 0.1126, "step": 31803 }, { "epoch": 23.2996336996337, "grad_norm": 2.03125, "learning_rate": 2.829317312184629e-05, "loss": 0.1252, "step": 31804 }, { "epoch": 23.3003663003663, "grad_norm": 1.015625, "learning_rate": 2.8290110963209265e-05, "loss": 0.1263, "step": 31805 }, { "epoch": 23.3010989010989, "grad_norm": 1.71875, "learning_rate": 2.8287048910987113e-05, "loss": 0.107, "step": 31806 }, { "epoch": 23.3018315018315, "grad_norm": 1.171875, "learning_rate": 2.8283986965194505e-05, "loss": 0.1441, "step": 31807 }, { "epoch": 23.3025641025641, "grad_norm": 1.140625, "learning_rate": 2.8280925125846125e-05, "loss": 0.1202, "step": 31808 }, { "epoch": 23.303296703296702, "grad_norm": 0.6640625, "learning_rate": 2.8277863392956604e-05, "loss": 0.1053, "step": 31809 }, { "epoch": 23.304029304029303, "grad_norm": 1.1875, "learning_rate": 2.8274801766540665e-05, "loss": 0.0912, "step": 31810 }, { "epoch": 23.304761904761904, "grad_norm": 0.84375, "learning_rate": 2.8271740246612892e-05, "loss": 0.1338, "step": 31811 }, { "epoch": 23.305494505494504, "grad_norm": 1.2421875, "learning_rate": 2.8268678833188057e-05, "loss": 0.1315, "step": 31812 }, { "epoch": 23.306227106227105, "grad_norm": 0.90234375, "learning_rate": 2.8265617526280763e-05, "loss": 0.1095, "step": 31813 }, { "epoch": 23.306959706959706, "grad_norm": 1.2265625, "learning_rate": 2.8262556325905687e-05, "loss": 0.1265, "step": 31814 }, { "epoch": 23.307692307692307, "grad_norm": 1.265625, "learning_rate": 2.8259495232077536e-05, "loss": 0.1237, "step": 31815 }, { "epoch": 23.308424908424907, "grad_norm": 1.1015625, "learning_rate": 2.825643424481089e-05, "loss": 0.1262, "step": 31816 }, { "epoch": 23.309157509157508, "grad_norm": 1.1015625, "learning_rate": 2.8253373364120517e-05, "loss": 0.1144, "step": 31817 }, { "epoch": 23.30989010989011, "grad_norm": 0.75, "learning_rate": 2.8250312590021012e-05, "loss": 0.1189, "step": 31818 }, { "epoch": 23.31062271062271, "grad_norm": 1.1640625, "learning_rate": 2.8247251922527085e-05, "loss": 0.1063, "step": 31819 }, { "epoch": 23.31135531135531, "grad_norm": 1.0546875, "learning_rate": 2.824419136165336e-05, "loss": 0.1068, "step": 31820 }, { "epoch": 23.31208791208791, "grad_norm": 0.9375, "learning_rate": 2.824113090741452e-05, "loss": 0.1202, "step": 31821 }, { "epoch": 23.31282051282051, "grad_norm": 1.0625, "learning_rate": 2.823807055982523e-05, "loss": 0.1192, "step": 31822 }, { "epoch": 23.313553113553112, "grad_norm": 0.7890625, "learning_rate": 2.8235010318900163e-05, "loss": 0.104, "step": 31823 }, { "epoch": 23.314285714285713, "grad_norm": 1.0625, "learning_rate": 2.8231950184653965e-05, "loss": 0.117, "step": 31824 }, { "epoch": 23.315018315018314, "grad_norm": 0.86328125, "learning_rate": 2.8228890157101294e-05, "loss": 0.1131, "step": 31825 }, { "epoch": 23.315750915750915, "grad_norm": 1.109375, "learning_rate": 2.8225830236256824e-05, "loss": 0.1453, "step": 31826 }, { "epoch": 23.316483516483515, "grad_norm": 0.578125, "learning_rate": 2.8222770422135215e-05, "loss": 0.1102, "step": 31827 }, { "epoch": 23.317216117216116, "grad_norm": 0.85546875, "learning_rate": 2.8219710714751145e-05, "loss": 0.1068, "step": 31828 }, { "epoch": 23.317948717948717, "grad_norm": 0.83984375, "learning_rate": 2.8216651114119234e-05, "loss": 0.0848, "step": 31829 }, { "epoch": 23.318681318681318, "grad_norm": 0.67578125, "learning_rate": 2.8213591620254186e-05, "loss": 0.1156, "step": 31830 }, { "epoch": 23.31941391941392, "grad_norm": 1.0390625, "learning_rate": 2.8210532233170618e-05, "loss": 0.1273, "step": 31831 }, { "epoch": 23.32014652014652, "grad_norm": 0.98046875, "learning_rate": 2.8207472952883203e-05, "loss": 0.1244, "step": 31832 }, { "epoch": 23.32087912087912, "grad_norm": 0.91015625, "learning_rate": 2.8204413779406595e-05, "loss": 0.1088, "step": 31833 }, { "epoch": 23.32161172161172, "grad_norm": 0.859375, "learning_rate": 2.820135471275549e-05, "loss": 0.1154, "step": 31834 }, { "epoch": 23.32234432234432, "grad_norm": 1.5078125, "learning_rate": 2.8198295752944497e-05, "loss": 0.1209, "step": 31835 }, { "epoch": 23.323076923076922, "grad_norm": 0.8828125, "learning_rate": 2.819523689998828e-05, "loss": 0.1332, "step": 31836 }, { "epoch": 23.323809523809523, "grad_norm": 1.28125, "learning_rate": 2.819217815390151e-05, "loss": 0.1133, "step": 31837 }, { "epoch": 23.324542124542123, "grad_norm": 0.9375, "learning_rate": 2.8189119514698865e-05, "loss": 0.1205, "step": 31838 }, { "epoch": 23.325274725274724, "grad_norm": 1.234375, "learning_rate": 2.8186060982394946e-05, "loss": 0.1347, "step": 31839 }, { "epoch": 23.326007326007325, "grad_norm": 0.91015625, "learning_rate": 2.8183002557004434e-05, "loss": 0.105, "step": 31840 }, { "epoch": 23.326739926739926, "grad_norm": 1.453125, "learning_rate": 2.8179944238542006e-05, "loss": 0.1071, "step": 31841 }, { "epoch": 23.327472527472526, "grad_norm": 1.3515625, "learning_rate": 2.8176886027022255e-05, "loss": 0.1271, "step": 31842 }, { "epoch": 23.328205128205127, "grad_norm": 0.75390625, "learning_rate": 2.81738279224599e-05, "loss": 0.1175, "step": 31843 }, { "epoch": 23.328937728937728, "grad_norm": 1.21875, "learning_rate": 2.8170769924869556e-05, "loss": 0.1282, "step": 31844 }, { "epoch": 23.32967032967033, "grad_norm": 1.1484375, "learning_rate": 2.8167712034265892e-05, "loss": 0.1582, "step": 31845 }, { "epoch": 23.33040293040293, "grad_norm": 0.80859375, "learning_rate": 2.816465425066354e-05, "loss": 0.1027, "step": 31846 }, { "epoch": 23.33113553113553, "grad_norm": 1.1796875, "learning_rate": 2.816159657407716e-05, "loss": 0.1276, "step": 31847 }, { "epoch": 23.33186813186813, "grad_norm": 1.0390625, "learning_rate": 2.8158539004521406e-05, "loss": 0.1292, "step": 31848 }, { "epoch": 23.33260073260073, "grad_norm": 0.90625, "learning_rate": 2.8155481542010938e-05, "loss": 0.1297, "step": 31849 }, { "epoch": 23.333333333333332, "grad_norm": 1.015625, "learning_rate": 2.8152424186560383e-05, "loss": 0.1108, "step": 31850 }, { "epoch": 23.334065934065933, "grad_norm": 1.109375, "learning_rate": 2.814936693818439e-05, "loss": 0.1328, "step": 31851 }, { "epoch": 23.334798534798534, "grad_norm": 1.203125, "learning_rate": 2.8146309796897623e-05, "loss": 0.1012, "step": 31852 }, { "epoch": 23.335531135531134, "grad_norm": 1.171875, "learning_rate": 2.814325276271472e-05, "loss": 0.127, "step": 31853 }, { "epoch": 23.336263736263735, "grad_norm": 0.796875, "learning_rate": 2.8140195835650358e-05, "loss": 0.1106, "step": 31854 }, { "epoch": 23.336996336996336, "grad_norm": 1.1171875, "learning_rate": 2.8137139015719138e-05, "loss": 0.1247, "step": 31855 }, { "epoch": 23.337728937728937, "grad_norm": 1.2578125, "learning_rate": 2.813408230293574e-05, "loss": 0.1571, "step": 31856 }, { "epoch": 23.338461538461537, "grad_norm": 0.9453125, "learning_rate": 2.813102569731478e-05, "loss": 0.1466, "step": 31857 }, { "epoch": 23.339194139194138, "grad_norm": 0.5859375, "learning_rate": 2.8127969198870923e-05, "loss": 0.1275, "step": 31858 }, { "epoch": 23.33992673992674, "grad_norm": 1.34375, "learning_rate": 2.8124912807618802e-05, "loss": 0.1197, "step": 31859 }, { "epoch": 23.34065934065934, "grad_norm": 1.328125, "learning_rate": 2.8121856523573097e-05, "loss": 0.1245, "step": 31860 }, { "epoch": 23.34139194139194, "grad_norm": 0.9453125, "learning_rate": 2.8118800346748396e-05, "loss": 0.1431, "step": 31861 }, { "epoch": 23.34212454212454, "grad_norm": 0.921875, "learning_rate": 2.811574427715937e-05, "loss": 0.1194, "step": 31862 }, { "epoch": 23.34285714285714, "grad_norm": 0.94921875, "learning_rate": 2.8112688314820672e-05, "loss": 0.1625, "step": 31863 }, { "epoch": 23.343589743589742, "grad_norm": 0.765625, "learning_rate": 2.8109632459746943e-05, "loss": 0.1125, "step": 31864 }, { "epoch": 23.344322344322343, "grad_norm": 1.1171875, "learning_rate": 2.81065767119528e-05, "loss": 0.109, "step": 31865 }, { "epoch": 23.345054945054944, "grad_norm": 1.046875, "learning_rate": 2.8103521071452894e-05, "loss": 0.1219, "step": 31866 }, { "epoch": 23.345787545787545, "grad_norm": 2.0625, "learning_rate": 2.8100465538261893e-05, "loss": 0.1375, "step": 31867 }, { "epoch": 23.346520146520145, "grad_norm": 1.3671875, "learning_rate": 2.8097410112394374e-05, "loss": 0.1366, "step": 31868 }, { "epoch": 23.347252747252746, "grad_norm": 1.0546875, "learning_rate": 2.809435479386507e-05, "loss": 0.0927, "step": 31869 }, { "epoch": 23.347985347985347, "grad_norm": 0.953125, "learning_rate": 2.809129958268853e-05, "loss": 0.1185, "step": 31870 }, { "epoch": 23.348717948717947, "grad_norm": 0.88671875, "learning_rate": 2.8088244478879455e-05, "loss": 0.0853, "step": 31871 }, { "epoch": 23.349450549450548, "grad_norm": 1.21875, "learning_rate": 2.808518948245244e-05, "loss": 0.1217, "step": 31872 }, { "epoch": 23.35018315018315, "grad_norm": 0.9453125, "learning_rate": 2.8082134593422148e-05, "loss": 0.1151, "step": 31873 }, { "epoch": 23.35091575091575, "grad_norm": 0.7421875, "learning_rate": 2.80790798118032e-05, "loss": 0.0948, "step": 31874 }, { "epoch": 23.35164835164835, "grad_norm": 1.109375, "learning_rate": 2.8076025137610255e-05, "loss": 0.0959, "step": 31875 }, { "epoch": 23.35238095238095, "grad_norm": 1.0, "learning_rate": 2.8072970570857928e-05, "loss": 0.1147, "step": 31876 }, { "epoch": 23.353113553113552, "grad_norm": 0.99609375, "learning_rate": 2.806991611156085e-05, "loss": 0.0997, "step": 31877 }, { "epoch": 23.353846153846153, "grad_norm": 0.9609375, "learning_rate": 2.8066861759733677e-05, "loss": 0.1275, "step": 31878 }, { "epoch": 23.354578754578753, "grad_norm": 1.3828125, "learning_rate": 2.8063807515391027e-05, "loss": 0.1153, "step": 31879 }, { "epoch": 23.355311355311354, "grad_norm": 0.89453125, "learning_rate": 2.806075337854756e-05, "loss": 0.1308, "step": 31880 }, { "epoch": 23.356043956043955, "grad_norm": 1.1484375, "learning_rate": 2.8057699349217875e-05, "loss": 0.1286, "step": 31881 }, { "epoch": 23.356776556776556, "grad_norm": 1.296875, "learning_rate": 2.8054645427416644e-05, "loss": 0.1345, "step": 31882 }, { "epoch": 23.357509157509156, "grad_norm": 1.109375, "learning_rate": 2.805159161315844e-05, "loss": 0.1169, "step": 31883 }, { "epoch": 23.358241758241757, "grad_norm": 0.953125, "learning_rate": 2.804853790645794e-05, "loss": 0.1369, "step": 31884 }, { "epoch": 23.358974358974358, "grad_norm": 0.82421875, "learning_rate": 2.8045484307329768e-05, "loss": 0.1218, "step": 31885 }, { "epoch": 23.35970695970696, "grad_norm": 0.765625, "learning_rate": 2.8042430815788568e-05, "loss": 0.1314, "step": 31886 }, { "epoch": 23.36043956043956, "grad_norm": 1.140625, "learning_rate": 2.8039377431848928e-05, "loss": 0.1203, "step": 31887 }, { "epoch": 23.36117216117216, "grad_norm": 1.3046875, "learning_rate": 2.803632415552552e-05, "loss": 0.1423, "step": 31888 }, { "epoch": 23.36190476190476, "grad_norm": 1.0703125, "learning_rate": 2.803327098683294e-05, "loss": 0.1351, "step": 31889 }, { "epoch": 23.36263736263736, "grad_norm": 1.421875, "learning_rate": 2.8030217925785857e-05, "loss": 0.1343, "step": 31890 }, { "epoch": 23.363369963369962, "grad_norm": 0.87890625, "learning_rate": 2.802716497239886e-05, "loss": 0.1366, "step": 31891 }, { "epoch": 23.364102564102563, "grad_norm": 0.99609375, "learning_rate": 2.8024112126686596e-05, "loss": 0.1174, "step": 31892 }, { "epoch": 23.364835164835164, "grad_norm": 1.703125, "learning_rate": 2.80210593886637e-05, "loss": 0.1574, "step": 31893 }, { "epoch": 23.365567765567764, "grad_norm": 1.609375, "learning_rate": 2.801800675834476e-05, "loss": 0.0993, "step": 31894 }, { "epoch": 23.366300366300365, "grad_norm": 0.97265625, "learning_rate": 2.8014954235744468e-05, "loss": 0.1238, "step": 31895 }, { "epoch": 23.367032967032966, "grad_norm": 1.1171875, "learning_rate": 2.8011901820877384e-05, "loss": 0.1277, "step": 31896 }, { "epoch": 23.367765567765566, "grad_norm": 0.58203125, "learning_rate": 2.800884951375819e-05, "loss": 0.1175, "step": 31897 }, { "epoch": 23.368498168498167, "grad_norm": 0.921875, "learning_rate": 2.800579731440145e-05, "loss": 0.1235, "step": 31898 }, { "epoch": 23.369230769230768, "grad_norm": 1.09375, "learning_rate": 2.800274522282183e-05, "loss": 0.1204, "step": 31899 }, { "epoch": 23.36996336996337, "grad_norm": 0.7890625, "learning_rate": 2.7999693239033938e-05, "loss": 0.1288, "step": 31900 }, { "epoch": 23.37069597069597, "grad_norm": 1.234375, "learning_rate": 2.7996641363052424e-05, "loss": 0.1261, "step": 31901 }, { "epoch": 23.37142857142857, "grad_norm": 0.78515625, "learning_rate": 2.7993589594891868e-05, "loss": 0.1149, "step": 31902 }, { "epoch": 23.37216117216117, "grad_norm": 0.953125, "learning_rate": 2.7990537934566908e-05, "loss": 0.1333, "step": 31903 }, { "epoch": 23.37289377289377, "grad_norm": 1.3359375, "learning_rate": 2.798748638209217e-05, "loss": 0.1053, "step": 31904 }, { "epoch": 23.373626373626372, "grad_norm": 1.25, "learning_rate": 2.798443493748228e-05, "loss": 0.1322, "step": 31905 }, { "epoch": 23.374358974358973, "grad_norm": 1.21875, "learning_rate": 2.7981383600751877e-05, "loss": 0.1204, "step": 31906 }, { "epoch": 23.375091575091574, "grad_norm": 1.4375, "learning_rate": 2.7978332371915532e-05, "loss": 0.131, "step": 31907 }, { "epoch": 23.375824175824174, "grad_norm": 1.7734375, "learning_rate": 2.797528125098791e-05, "loss": 0.1256, "step": 31908 }, { "epoch": 23.376556776556775, "grad_norm": 1.1796875, "learning_rate": 2.797223023798359e-05, "loss": 0.1482, "step": 31909 }, { "epoch": 23.377289377289376, "grad_norm": 1.2109375, "learning_rate": 2.7969179332917212e-05, "loss": 0.113, "step": 31910 }, { "epoch": 23.378021978021977, "grad_norm": 1.0, "learning_rate": 2.796612853580339e-05, "loss": 0.1164, "step": 31911 }, { "epoch": 23.378754578754577, "grad_norm": 0.6796875, "learning_rate": 2.7963077846656767e-05, "loss": 0.1218, "step": 31912 }, { "epoch": 23.379487179487178, "grad_norm": 0.6796875, "learning_rate": 2.7960027265491917e-05, "loss": 0.1179, "step": 31913 }, { "epoch": 23.38021978021978, "grad_norm": 1.0625, "learning_rate": 2.795697679232347e-05, "loss": 0.1182, "step": 31914 }, { "epoch": 23.38095238095238, "grad_norm": 0.99609375, "learning_rate": 2.7953926427166055e-05, "loss": 0.1192, "step": 31915 }, { "epoch": 23.38168498168498, "grad_norm": 0.828125, "learning_rate": 2.795087617003429e-05, "loss": 0.1074, "step": 31916 }, { "epoch": 23.38241758241758, "grad_norm": 1.3828125, "learning_rate": 2.794782602094277e-05, "loss": 0.1209, "step": 31917 }, { "epoch": 23.383150183150182, "grad_norm": 1.40625, "learning_rate": 2.7944775979906113e-05, "loss": 0.1412, "step": 31918 }, { "epoch": 23.383882783882783, "grad_norm": 1.390625, "learning_rate": 2.7941726046938952e-05, "loss": 0.1046, "step": 31919 }, { "epoch": 23.384615384615383, "grad_norm": 0.765625, "learning_rate": 2.7938676222055865e-05, "loss": 0.1171, "step": 31920 }, { "epoch": 23.385347985347984, "grad_norm": 1.0390625, "learning_rate": 2.7935626505271516e-05, "loss": 0.1027, "step": 31921 }, { "epoch": 23.386080586080585, "grad_norm": 1.1796875, "learning_rate": 2.7932576896600467e-05, "loss": 0.1045, "step": 31922 }, { "epoch": 23.386813186813185, "grad_norm": 1.1484375, "learning_rate": 2.792952739605737e-05, "loss": 0.1276, "step": 31923 }, { "epoch": 23.387545787545786, "grad_norm": 1.6875, "learning_rate": 2.7926478003656798e-05, "loss": 0.1198, "step": 31924 }, { "epoch": 23.388278388278387, "grad_norm": 1.0234375, "learning_rate": 2.7923428719413388e-05, "loss": 0.1119, "step": 31925 }, { "epoch": 23.389010989010988, "grad_norm": 0.6640625, "learning_rate": 2.792037954334173e-05, "loss": 0.1261, "step": 31926 }, { "epoch": 23.38974358974359, "grad_norm": 1.1875, "learning_rate": 2.7917330475456473e-05, "loss": 0.1247, "step": 31927 }, { "epoch": 23.39047619047619, "grad_norm": 0.98046875, "learning_rate": 2.791428151577217e-05, "loss": 0.1143, "step": 31928 }, { "epoch": 23.39120879120879, "grad_norm": 1.25, "learning_rate": 2.791123266430347e-05, "loss": 0.1664, "step": 31929 }, { "epoch": 23.39194139194139, "grad_norm": 1.2421875, "learning_rate": 2.7908183921064967e-05, "loss": 0.1045, "step": 31930 }, { "epoch": 23.39267399267399, "grad_norm": 0.921875, "learning_rate": 2.7905135286071267e-05, "loss": 0.1378, "step": 31931 }, { "epoch": 23.393406593406592, "grad_norm": 0.90625, "learning_rate": 2.790208675933701e-05, "loss": 0.1, "step": 31932 }, { "epoch": 23.394139194139193, "grad_norm": 0.95703125, "learning_rate": 2.7899038340876743e-05, "loss": 0.1302, "step": 31933 }, { "epoch": 23.394871794871793, "grad_norm": 0.84375, "learning_rate": 2.789599003070513e-05, "loss": 0.1187, "step": 31934 }, { "epoch": 23.395604395604394, "grad_norm": 1.078125, "learning_rate": 2.789294182883673e-05, "loss": 0.1457, "step": 31935 }, { "epoch": 23.396336996336995, "grad_norm": 1.7578125, "learning_rate": 2.788989373528616e-05, "loss": 0.1409, "step": 31936 }, { "epoch": 23.397069597069596, "grad_norm": 0.9453125, "learning_rate": 2.7886845750068036e-05, "loss": 0.1494, "step": 31937 }, { "epoch": 23.397802197802196, "grad_norm": 1.1875, "learning_rate": 2.7883797873196976e-05, "loss": 0.098, "step": 31938 }, { "epoch": 23.398534798534797, "grad_norm": 1.6484375, "learning_rate": 2.788075010468755e-05, "loss": 0.1164, "step": 31939 }, { "epoch": 23.399267399267398, "grad_norm": 1.4453125, "learning_rate": 2.7877702444554364e-05, "loss": 0.1227, "step": 31940 }, { "epoch": 23.4, "grad_norm": 0.88671875, "learning_rate": 2.7874654892812042e-05, "loss": 0.1133, "step": 31941 }, { "epoch": 23.4007326007326, "grad_norm": 1.46875, "learning_rate": 2.787160744947518e-05, "loss": 0.1093, "step": 31942 }, { "epoch": 23.4014652014652, "grad_norm": 0.82421875, "learning_rate": 2.7868560114558384e-05, "loss": 0.1243, "step": 31943 }, { "epoch": 23.4021978021978, "grad_norm": 0.9140625, "learning_rate": 2.786551288807623e-05, "loss": 0.135, "step": 31944 }, { "epoch": 23.4029304029304, "grad_norm": 1.0, "learning_rate": 2.786246577004335e-05, "loss": 0.1107, "step": 31945 }, { "epoch": 23.403663003663002, "grad_norm": 1.6328125, "learning_rate": 2.7859418760474297e-05, "loss": 0.1385, "step": 31946 }, { "epoch": 23.404395604395603, "grad_norm": 0.86328125, "learning_rate": 2.7856371859383735e-05, "loss": 0.1488, "step": 31947 }, { "epoch": 23.405128205128204, "grad_norm": 1.640625, "learning_rate": 2.7853325066786208e-05, "loss": 0.1216, "step": 31948 }, { "epoch": 23.405860805860804, "grad_norm": 0.6484375, "learning_rate": 2.785027838269636e-05, "loss": 0.0922, "step": 31949 }, { "epoch": 23.406593406593405, "grad_norm": 1.328125, "learning_rate": 2.784723180712874e-05, "loss": 0.1082, "step": 31950 }, { "epoch": 23.407326007326006, "grad_norm": 1.125, "learning_rate": 2.784418534009797e-05, "loss": 0.1575, "step": 31951 }, { "epoch": 23.408058608058607, "grad_norm": 1.5390625, "learning_rate": 2.7841138981618647e-05, "loss": 0.1449, "step": 31952 }, { "epoch": 23.408791208791207, "grad_norm": 1.3984375, "learning_rate": 2.7838092731705385e-05, "loss": 0.1169, "step": 31953 }, { "epoch": 23.409523809523808, "grad_norm": 0.8515625, "learning_rate": 2.7835046590372738e-05, "loss": 0.0963, "step": 31954 }, { "epoch": 23.41025641025641, "grad_norm": 0.76953125, "learning_rate": 2.7832000557635325e-05, "loss": 0.1103, "step": 31955 }, { "epoch": 23.41098901098901, "grad_norm": 1.421875, "learning_rate": 2.782895463350773e-05, "loss": 0.146, "step": 31956 }, { "epoch": 23.41172161172161, "grad_norm": 1.375, "learning_rate": 2.7825908818004572e-05, "loss": 0.1159, "step": 31957 }, { "epoch": 23.41245421245421, "grad_norm": 0.52734375, "learning_rate": 2.7822863111140436e-05, "loss": 0.1103, "step": 31958 }, { "epoch": 23.41318681318681, "grad_norm": 1.296875, "learning_rate": 2.7819817512929896e-05, "loss": 0.1481, "step": 31959 }, { "epoch": 23.413919413919412, "grad_norm": 0.75, "learning_rate": 2.781677202338756e-05, "loss": 0.1317, "step": 31960 }, { "epoch": 23.414652014652013, "grad_norm": 1.4921875, "learning_rate": 2.7813726642528008e-05, "loss": 0.1365, "step": 31961 }, { "epoch": 23.415384615384614, "grad_norm": 1.03125, "learning_rate": 2.7810681370365843e-05, "loss": 0.1109, "step": 31962 }, { "epoch": 23.416117216117215, "grad_norm": 1.2265625, "learning_rate": 2.780763620691565e-05, "loss": 0.0902, "step": 31963 }, { "epoch": 23.416849816849815, "grad_norm": 0.96484375, "learning_rate": 2.7804591152192034e-05, "loss": 0.1128, "step": 31964 }, { "epoch": 23.417582417582416, "grad_norm": 1.546875, "learning_rate": 2.7801546206209555e-05, "loss": 0.126, "step": 31965 }, { "epoch": 23.418315018315017, "grad_norm": 0.93359375, "learning_rate": 2.779850136898282e-05, "loss": 0.1098, "step": 31966 }, { "epoch": 23.419047619047618, "grad_norm": 0.9921875, "learning_rate": 2.7795456640526418e-05, "loss": 0.091, "step": 31967 }, { "epoch": 23.41978021978022, "grad_norm": 1.0390625, "learning_rate": 2.7792412020854942e-05, "loss": 0.121, "step": 31968 }, { "epoch": 23.42051282051282, "grad_norm": 1.09375, "learning_rate": 2.7789367509982984e-05, "loss": 0.128, "step": 31969 }, { "epoch": 23.42124542124542, "grad_norm": 1.140625, "learning_rate": 2.778632310792511e-05, "loss": 0.146, "step": 31970 }, { "epoch": 23.42197802197802, "grad_norm": 1.125, "learning_rate": 2.7783278814695935e-05, "loss": 0.1348, "step": 31971 }, { "epoch": 23.42271062271062, "grad_norm": 0.89453125, "learning_rate": 2.7780234630309988e-05, "loss": 0.1134, "step": 31972 }, { "epoch": 23.423443223443222, "grad_norm": 1.3046875, "learning_rate": 2.7777190554781933e-05, "loss": 0.1388, "step": 31973 }, { "epoch": 23.424175824175823, "grad_norm": 0.82421875, "learning_rate": 2.7774146588126303e-05, "loss": 0.1372, "step": 31974 }, { "epoch": 23.424908424908423, "grad_norm": 0.7734375, "learning_rate": 2.777110273035771e-05, "loss": 0.1168, "step": 31975 }, { "epoch": 23.425641025641024, "grad_norm": 0.8203125, "learning_rate": 2.7768058981490714e-05, "loss": 0.1025, "step": 31976 }, { "epoch": 23.426373626373625, "grad_norm": 1.015625, "learning_rate": 2.7765015341539907e-05, "loss": 0.1547, "step": 31977 }, { "epoch": 23.427106227106226, "grad_norm": 1.015625, "learning_rate": 2.7761971810519878e-05, "loss": 0.1042, "step": 31978 }, { "epoch": 23.427838827838826, "grad_norm": 1.09375, "learning_rate": 2.7758928388445216e-05, "loss": 0.1116, "step": 31979 }, { "epoch": 23.428571428571427, "grad_norm": 1.1171875, "learning_rate": 2.7755885075330484e-05, "loss": 0.1416, "step": 31980 }, { "epoch": 23.429304029304028, "grad_norm": 1.1875, "learning_rate": 2.775284187119026e-05, "loss": 0.1015, "step": 31981 }, { "epoch": 23.43003663003663, "grad_norm": 1.8203125, "learning_rate": 2.7749798776039153e-05, "loss": 0.0942, "step": 31982 }, { "epoch": 23.43076923076923, "grad_norm": 1.4453125, "learning_rate": 2.7746755789891722e-05, "loss": 0.122, "step": 31983 }, { "epoch": 23.43150183150183, "grad_norm": 0.87890625, "learning_rate": 2.774371291276257e-05, "loss": 0.0995, "step": 31984 }, { "epoch": 23.43223443223443, "grad_norm": 0.96875, "learning_rate": 2.7740670144666243e-05, "loss": 0.1238, "step": 31985 }, { "epoch": 23.43296703296703, "grad_norm": 0.6875, "learning_rate": 2.7737627485617362e-05, "loss": 0.1248, "step": 31986 }, { "epoch": 23.433699633699632, "grad_norm": 1.0625, "learning_rate": 2.7734584935630452e-05, "loss": 0.1145, "step": 31987 }, { "epoch": 23.434432234432233, "grad_norm": 1.0390625, "learning_rate": 2.773154249472012e-05, "loss": 0.1074, "step": 31988 }, { "epoch": 23.435164835164834, "grad_norm": 1.40625, "learning_rate": 2.772850016290095e-05, "loss": 0.0932, "step": 31989 }, { "epoch": 23.435897435897434, "grad_norm": 0.9921875, "learning_rate": 2.772545794018752e-05, "loss": 0.1014, "step": 31990 }, { "epoch": 23.436630036630035, "grad_norm": 1.796875, "learning_rate": 2.7722415826594383e-05, "loss": 0.1316, "step": 31991 }, { "epoch": 23.437362637362636, "grad_norm": 1.4609375, "learning_rate": 2.771937382213613e-05, "loss": 0.1182, "step": 31992 }, { "epoch": 23.438095238095237, "grad_norm": 0.92578125, "learning_rate": 2.771633192682734e-05, "loss": 0.1048, "step": 31993 }, { "epoch": 23.438827838827837, "grad_norm": 1.3515625, "learning_rate": 2.7713290140682574e-05, "loss": 0.1334, "step": 31994 }, { "epoch": 23.439560439560438, "grad_norm": 1.8515625, "learning_rate": 2.7710248463716436e-05, "loss": 0.0981, "step": 31995 }, { "epoch": 23.44029304029304, "grad_norm": 1.0390625, "learning_rate": 2.770720689594346e-05, "loss": 0.1274, "step": 31996 }, { "epoch": 23.44102564102564, "grad_norm": 1.1640625, "learning_rate": 2.770416543737826e-05, "loss": 0.1183, "step": 31997 }, { "epoch": 23.44175824175824, "grad_norm": 1.0546875, "learning_rate": 2.7701124088035353e-05, "loss": 0.1558, "step": 31998 }, { "epoch": 23.44249084249084, "grad_norm": 0.7890625, "learning_rate": 2.7698082847929378e-05, "loss": 0.1195, "step": 31999 }, { "epoch": 23.44322344322344, "grad_norm": 1.1328125, "learning_rate": 2.7695041717074854e-05, "loss": 0.1202, "step": 32000 }, { "epoch": 23.443956043956042, "grad_norm": 1.0078125, "learning_rate": 2.769200069548639e-05, "loss": 0.1034, "step": 32001 }, { "epoch": 23.444688644688643, "grad_norm": 1.0078125, "learning_rate": 2.7688959783178525e-05, "loss": 0.1227, "step": 32002 }, { "epoch": 23.445421245421244, "grad_norm": 0.89453125, "learning_rate": 2.768591898016583e-05, "loss": 0.1057, "step": 32003 }, { "epoch": 23.446153846153845, "grad_norm": 1.0390625, "learning_rate": 2.7682878286462892e-05, "loss": 0.0947, "step": 32004 }, { "epoch": 23.446886446886445, "grad_norm": 1.203125, "learning_rate": 2.76798377020843e-05, "loss": 0.1326, "step": 32005 }, { "epoch": 23.447619047619046, "grad_norm": 1.21875, "learning_rate": 2.7676797227044577e-05, "loss": 0.1109, "step": 32006 }, { "epoch": 23.448351648351647, "grad_norm": 0.93359375, "learning_rate": 2.7673756861358303e-05, "loss": 0.1147, "step": 32007 }, { "epoch": 23.449084249084247, "grad_norm": 0.75, "learning_rate": 2.7670716605040076e-05, "loss": 0.1127, "step": 32008 }, { "epoch": 23.449816849816848, "grad_norm": 1.5, "learning_rate": 2.7667676458104404e-05, "loss": 0.1046, "step": 32009 }, { "epoch": 23.45054945054945, "grad_norm": 1.03125, "learning_rate": 2.766463642056592e-05, "loss": 0.1178, "step": 32010 }, { "epoch": 23.45128205128205, "grad_norm": 0.9453125, "learning_rate": 2.766159649243915e-05, "loss": 0.1211, "step": 32011 }, { "epoch": 23.45201465201465, "grad_norm": 0.87109375, "learning_rate": 2.7658556673738683e-05, "loss": 0.117, "step": 32012 }, { "epoch": 23.45274725274725, "grad_norm": 0.9453125, "learning_rate": 2.765551696447905e-05, "loss": 0.1041, "step": 32013 }, { "epoch": 23.453479853479852, "grad_norm": 0.75, "learning_rate": 2.765247736467484e-05, "loss": 0.1303, "step": 32014 }, { "epoch": 23.454212454212453, "grad_norm": 1.0703125, "learning_rate": 2.7649437874340606e-05, "loss": 0.1039, "step": 32015 }, { "epoch": 23.454945054945053, "grad_norm": 1.2578125, "learning_rate": 2.7646398493490934e-05, "loss": 0.134, "step": 32016 }, { "epoch": 23.455677655677654, "grad_norm": 1.2265625, "learning_rate": 2.7643359222140348e-05, "loss": 0.097, "step": 32017 }, { "epoch": 23.456410256410255, "grad_norm": 1.1015625, "learning_rate": 2.764032006030344e-05, "loss": 0.1272, "step": 32018 }, { "epoch": 23.457142857142856, "grad_norm": 0.95703125, "learning_rate": 2.7637281007994763e-05, "loss": 0.1133, "step": 32019 }, { "epoch": 23.457875457875456, "grad_norm": 0.80078125, "learning_rate": 2.7634242065228864e-05, "loss": 0.1106, "step": 32020 }, { "epoch": 23.458608058608057, "grad_norm": 1.140625, "learning_rate": 2.763120323202035e-05, "loss": 0.1136, "step": 32021 }, { "epoch": 23.459340659340658, "grad_norm": 1.390625, "learning_rate": 2.7628164508383726e-05, "loss": 0.1632, "step": 32022 }, { "epoch": 23.46007326007326, "grad_norm": 1.0703125, "learning_rate": 2.7625125894333594e-05, "loss": 0.1111, "step": 32023 }, { "epoch": 23.46080586080586, "grad_norm": 0.96875, "learning_rate": 2.7622087389884453e-05, "loss": 0.1216, "step": 32024 }, { "epoch": 23.46153846153846, "grad_norm": 1.984375, "learning_rate": 2.761904899505094e-05, "loss": 0.1254, "step": 32025 }, { "epoch": 23.46227106227106, "grad_norm": 1.1953125, "learning_rate": 2.7616010709847557e-05, "loss": 0.1231, "step": 32026 }, { "epoch": 23.46300366300366, "grad_norm": 0.83984375, "learning_rate": 2.76129725342889e-05, "loss": 0.1303, "step": 32027 }, { "epoch": 23.463736263736262, "grad_norm": 1.125, "learning_rate": 2.7609934468389485e-05, "loss": 0.0956, "step": 32028 }, { "epoch": 23.464468864468863, "grad_norm": 0.890625, "learning_rate": 2.7606896512163886e-05, "loss": 0.1042, "step": 32029 }, { "epoch": 23.465201465201464, "grad_norm": 0.8046875, "learning_rate": 2.760385866562666e-05, "loss": 0.1151, "step": 32030 }, { "epoch": 23.465934065934064, "grad_norm": 1.0625, "learning_rate": 2.7600820928792386e-05, "loss": 0.104, "step": 32031 }, { "epoch": 23.466666666666665, "grad_norm": 1.328125, "learning_rate": 2.7597783301675574e-05, "loss": 0.1413, "step": 32032 }, { "epoch": 23.467399267399266, "grad_norm": 1.1640625, "learning_rate": 2.7594745784290803e-05, "loss": 0.1176, "step": 32033 }, { "epoch": 23.468131868131866, "grad_norm": 0.9765625, "learning_rate": 2.7591708376652642e-05, "loss": 0.1033, "step": 32034 }, { "epoch": 23.468864468864467, "grad_norm": 0.96484375, "learning_rate": 2.7588671078775585e-05, "loss": 0.1171, "step": 32035 }, { "epoch": 23.469597069597068, "grad_norm": 0.9140625, "learning_rate": 2.7585633890674273e-05, "loss": 0.1043, "step": 32036 }, { "epoch": 23.47032967032967, "grad_norm": 0.99609375, "learning_rate": 2.758259681236319e-05, "loss": 0.104, "step": 32037 }, { "epoch": 23.47106227106227, "grad_norm": 1.1953125, "learning_rate": 2.7579559843856923e-05, "loss": 0.1343, "step": 32038 }, { "epoch": 23.47179487179487, "grad_norm": 1.21875, "learning_rate": 2.757652298516999e-05, "loss": 0.1423, "step": 32039 }, { "epoch": 23.47252747252747, "grad_norm": 1.3203125, "learning_rate": 2.7573486236316966e-05, "loss": 0.1267, "step": 32040 }, { "epoch": 23.47326007326007, "grad_norm": 0.69921875, "learning_rate": 2.7570449597312388e-05, "loss": 0.1408, "step": 32041 }, { "epoch": 23.473992673992672, "grad_norm": 0.86328125, "learning_rate": 2.7567413068170833e-05, "loss": 0.1153, "step": 32042 }, { "epoch": 23.474725274725273, "grad_norm": 0.68359375, "learning_rate": 2.756437664890682e-05, "loss": 0.1105, "step": 32043 }, { "epoch": 23.475457875457874, "grad_norm": 0.9296875, "learning_rate": 2.75613403395349e-05, "loss": 0.1088, "step": 32044 }, { "epoch": 23.476190476190474, "grad_norm": 0.9140625, "learning_rate": 2.7558304140069635e-05, "loss": 0.1229, "step": 32045 }, { "epoch": 23.476923076923075, "grad_norm": 0.73828125, "learning_rate": 2.7555268050525562e-05, "loss": 0.1155, "step": 32046 }, { "epoch": 23.477655677655676, "grad_norm": 0.578125, "learning_rate": 2.7552232070917255e-05, "loss": 0.1082, "step": 32047 }, { "epoch": 23.478388278388277, "grad_norm": 0.87109375, "learning_rate": 2.7549196201259215e-05, "loss": 0.1118, "step": 32048 }, { "epoch": 23.479120879120877, "grad_norm": 1.1796875, "learning_rate": 2.7546160441566025e-05, "loss": 0.1414, "step": 32049 }, { "epoch": 23.479853479853478, "grad_norm": 0.66796875, "learning_rate": 2.7543124791852186e-05, "loss": 0.128, "step": 32050 }, { "epoch": 23.48058608058608, "grad_norm": 0.80859375, "learning_rate": 2.7540089252132307e-05, "loss": 0.1054, "step": 32051 }, { "epoch": 23.48131868131868, "grad_norm": 1.1953125, "learning_rate": 2.7537053822420876e-05, "loss": 0.1296, "step": 32052 }, { "epoch": 23.48205128205128, "grad_norm": 0.90234375, "learning_rate": 2.753401850273248e-05, "loss": 0.1263, "step": 32053 }, { "epoch": 23.48278388278388, "grad_norm": 1.078125, "learning_rate": 2.753098329308162e-05, "loss": 0.1486, "step": 32054 }, { "epoch": 23.483516483516482, "grad_norm": 0.875, "learning_rate": 2.752794819348286e-05, "loss": 0.1205, "step": 32055 }, { "epoch": 23.484249084249083, "grad_norm": 1.0234375, "learning_rate": 2.752491320395074e-05, "loss": 0.1467, "step": 32056 }, { "epoch": 23.484981684981683, "grad_norm": 1.609375, "learning_rate": 2.752187832449982e-05, "loss": 0.1021, "step": 32057 }, { "epoch": 23.485714285714284, "grad_norm": 1.03125, "learning_rate": 2.7518843555144613e-05, "loss": 0.1334, "step": 32058 }, { "epoch": 23.486446886446885, "grad_norm": 0.953125, "learning_rate": 2.751580889589967e-05, "loss": 0.1149, "step": 32059 }, { "epoch": 23.487179487179485, "grad_norm": 1.296875, "learning_rate": 2.7512774346779545e-05, "loss": 0.1417, "step": 32060 }, { "epoch": 23.487912087912086, "grad_norm": 1.15625, "learning_rate": 2.7509739907798727e-05, "loss": 0.1273, "step": 32061 }, { "epoch": 23.488644688644687, "grad_norm": 1.578125, "learning_rate": 2.7506705578971837e-05, "loss": 0.1075, "step": 32062 }, { "epoch": 23.489377289377288, "grad_norm": 2.171875, "learning_rate": 2.7503671360313338e-05, "loss": 0.1596, "step": 32063 }, { "epoch": 23.49010989010989, "grad_norm": 0.9453125, "learning_rate": 2.750063725183783e-05, "loss": 0.1274, "step": 32064 }, { "epoch": 23.49084249084249, "grad_norm": 1.6875, "learning_rate": 2.7497603253559795e-05, "loss": 0.1283, "step": 32065 }, { "epoch": 23.49157509157509, "grad_norm": 0.859375, "learning_rate": 2.7494569365493793e-05, "loss": 0.0913, "step": 32066 }, { "epoch": 23.49230769230769, "grad_norm": 1.1796875, "learning_rate": 2.749153558765436e-05, "loss": 0.1142, "step": 32067 }, { "epoch": 23.493040293040295, "grad_norm": 0.8359375, "learning_rate": 2.748850192005606e-05, "loss": 0.0959, "step": 32068 }, { "epoch": 23.493772893772892, "grad_norm": 2.03125, "learning_rate": 2.7485468362713374e-05, "loss": 0.1555, "step": 32069 }, { "epoch": 23.494505494505496, "grad_norm": 0.71875, "learning_rate": 2.7482434915640864e-05, "loss": 0.1015, "step": 32070 }, { "epoch": 23.495238095238093, "grad_norm": 1.1015625, "learning_rate": 2.747940157885307e-05, "loss": 0.1236, "step": 32071 }, { "epoch": 23.495970695970698, "grad_norm": 1.1484375, "learning_rate": 2.7476368352364516e-05, "loss": 0.1498, "step": 32072 }, { "epoch": 23.496703296703295, "grad_norm": 0.68359375, "learning_rate": 2.7473335236189763e-05, "loss": 0.1248, "step": 32073 }, { "epoch": 23.4974358974359, "grad_norm": 1.078125, "learning_rate": 2.7470302230343297e-05, "loss": 0.1353, "step": 32074 }, { "epoch": 23.498168498168496, "grad_norm": 1.125, "learning_rate": 2.7467269334839695e-05, "loss": 0.1405, "step": 32075 }, { "epoch": 23.4989010989011, "grad_norm": 1.203125, "learning_rate": 2.7464236549693424e-05, "loss": 0.1348, "step": 32076 }, { "epoch": 23.499633699633698, "grad_norm": 0.7578125, "learning_rate": 2.74612038749191e-05, "loss": 0.1212, "step": 32077 }, { "epoch": 23.500366300366302, "grad_norm": 1.1484375, "learning_rate": 2.7458171310531187e-05, "loss": 0.1259, "step": 32078 }, { "epoch": 23.5010989010989, "grad_norm": 0.9453125, "learning_rate": 2.7455138856544265e-05, "loss": 0.1279, "step": 32079 }, { "epoch": 23.501831501831504, "grad_norm": 1.140625, "learning_rate": 2.745210651297282e-05, "loss": 0.1183, "step": 32080 }, { "epoch": 23.5025641025641, "grad_norm": 0.7109375, "learning_rate": 2.7449074279831395e-05, "loss": 0.1222, "step": 32081 }, { "epoch": 23.503296703296705, "grad_norm": 0.734375, "learning_rate": 2.744604215713452e-05, "loss": 0.1262, "step": 32082 }, { "epoch": 23.504029304029302, "grad_norm": 0.96484375, "learning_rate": 2.744301014489673e-05, "loss": 0.114, "step": 32083 }, { "epoch": 23.504761904761907, "grad_norm": 1.3203125, "learning_rate": 2.7439978243132562e-05, "loss": 0.1388, "step": 32084 }, { "epoch": 23.505494505494504, "grad_norm": 1.15625, "learning_rate": 2.7436946451856516e-05, "loss": 0.1343, "step": 32085 }, { "epoch": 23.506227106227108, "grad_norm": 1.328125, "learning_rate": 2.7433914771083142e-05, "loss": 0.1288, "step": 32086 }, { "epoch": 23.506959706959705, "grad_norm": 1.0078125, "learning_rate": 2.7430883200826916e-05, "loss": 0.1205, "step": 32087 }, { "epoch": 23.50769230769231, "grad_norm": 0.796875, "learning_rate": 2.7427851741102446e-05, "loss": 0.1281, "step": 32088 }, { "epoch": 23.508424908424907, "grad_norm": 0.828125, "learning_rate": 2.7424820391924195e-05, "loss": 0.0974, "step": 32089 }, { "epoch": 23.50915750915751, "grad_norm": 1.234375, "learning_rate": 2.7421789153306715e-05, "loss": 0.1124, "step": 32090 }, { "epoch": 23.509890109890108, "grad_norm": 0.84765625, "learning_rate": 2.7418758025264514e-05, "loss": 0.1177, "step": 32091 }, { "epoch": 23.510622710622712, "grad_norm": 0.90234375, "learning_rate": 2.7415727007812096e-05, "loss": 0.1331, "step": 32092 }, { "epoch": 23.51135531135531, "grad_norm": 1.03125, "learning_rate": 2.7412696100964032e-05, "loss": 0.1159, "step": 32093 }, { "epoch": 23.512087912087914, "grad_norm": 1.1015625, "learning_rate": 2.7409665304734828e-05, "loss": 0.1122, "step": 32094 }, { "epoch": 23.51282051282051, "grad_norm": 2.09375, "learning_rate": 2.7406634619138976e-05, "loss": 0.1067, "step": 32095 }, { "epoch": 23.513553113553115, "grad_norm": 1.3828125, "learning_rate": 2.7403604044191027e-05, "loss": 0.1317, "step": 32096 }, { "epoch": 23.514285714285712, "grad_norm": 0.8125, "learning_rate": 2.740057357990549e-05, "loss": 0.1193, "step": 32097 }, { "epoch": 23.515018315018317, "grad_norm": 0.76953125, "learning_rate": 2.7397543226296884e-05, "loss": 0.1303, "step": 32098 }, { "epoch": 23.515750915750917, "grad_norm": 1.171875, "learning_rate": 2.7394512983379762e-05, "loss": 0.133, "step": 32099 }, { "epoch": 23.516483516483518, "grad_norm": 0.859375, "learning_rate": 2.7391482851168583e-05, "loss": 0.1218, "step": 32100 }, { "epoch": 23.51721611721612, "grad_norm": 0.94140625, "learning_rate": 2.738845282967792e-05, "loss": 0.1116, "step": 32101 }, { "epoch": 23.51794871794872, "grad_norm": 1.140625, "learning_rate": 2.7385422918922227e-05, "loss": 0.1486, "step": 32102 }, { "epoch": 23.51868131868132, "grad_norm": 0.85546875, "learning_rate": 2.7382393118916095e-05, "loss": 0.1048, "step": 32103 }, { "epoch": 23.51941391941392, "grad_norm": 1.265625, "learning_rate": 2.7379363429674e-05, "loss": 0.1365, "step": 32104 }, { "epoch": 23.520146520146522, "grad_norm": 0.671875, "learning_rate": 2.7376333851210467e-05, "loss": 0.0998, "step": 32105 }, { "epoch": 23.520879120879123, "grad_norm": 0.91015625, "learning_rate": 2.7373304383540003e-05, "loss": 0.1071, "step": 32106 }, { "epoch": 23.521611721611723, "grad_norm": 1.359375, "learning_rate": 2.737027502667713e-05, "loss": 0.1422, "step": 32107 }, { "epoch": 23.522344322344324, "grad_norm": 1.1484375, "learning_rate": 2.7367245780636353e-05, "loss": 0.1033, "step": 32108 }, { "epoch": 23.523076923076925, "grad_norm": 1.1796875, "learning_rate": 2.7364216645432206e-05, "loss": 0.1083, "step": 32109 }, { "epoch": 23.523809523809526, "grad_norm": 0.984375, "learning_rate": 2.7361187621079203e-05, "loss": 0.1281, "step": 32110 }, { "epoch": 23.524542124542126, "grad_norm": 0.9375, "learning_rate": 2.735815870759183e-05, "loss": 0.1155, "step": 32111 }, { "epoch": 23.525274725274727, "grad_norm": 0.6484375, "learning_rate": 2.7355129904984632e-05, "loss": 0.1111, "step": 32112 }, { "epoch": 23.526007326007328, "grad_norm": 0.99609375, "learning_rate": 2.735210121327207e-05, "loss": 0.1225, "step": 32113 }, { "epoch": 23.52673992673993, "grad_norm": 1.1484375, "learning_rate": 2.7349072632468726e-05, "loss": 0.1373, "step": 32114 }, { "epoch": 23.52747252747253, "grad_norm": 1.125, "learning_rate": 2.7346044162589058e-05, "loss": 0.1073, "step": 32115 }, { "epoch": 23.52820512820513, "grad_norm": 0.77734375, "learning_rate": 2.7343015803647615e-05, "loss": 0.1094, "step": 32116 }, { "epoch": 23.52893772893773, "grad_norm": 1.4453125, "learning_rate": 2.733998755565886e-05, "loss": 0.1738, "step": 32117 }, { "epoch": 23.52967032967033, "grad_norm": 1.0703125, "learning_rate": 2.733695941863733e-05, "loss": 0.1245, "step": 32118 }, { "epoch": 23.530402930402932, "grad_norm": 1.2109375, "learning_rate": 2.733393139259753e-05, "loss": 0.1459, "step": 32119 }, { "epoch": 23.531135531135533, "grad_norm": 1.0234375, "learning_rate": 2.733090347755399e-05, "loss": 0.1046, "step": 32120 }, { "epoch": 23.531868131868134, "grad_norm": 1.9453125, "learning_rate": 2.732787567352117e-05, "loss": 0.1169, "step": 32121 }, { "epoch": 23.532600732600734, "grad_norm": 0.79296875, "learning_rate": 2.732484798051362e-05, "loss": 0.096, "step": 32122 }, { "epoch": 23.533333333333335, "grad_norm": 0.828125, "learning_rate": 2.732182039854582e-05, "loss": 0.0982, "step": 32123 }, { "epoch": 23.534065934065936, "grad_norm": 0.84765625, "learning_rate": 2.731879292763229e-05, "loss": 0.0861, "step": 32124 }, { "epoch": 23.534798534798536, "grad_norm": 1.0078125, "learning_rate": 2.731576556778756e-05, "loss": 0.1371, "step": 32125 }, { "epoch": 23.535531135531137, "grad_norm": 1.3203125, "learning_rate": 2.7312738319026084e-05, "loss": 0.1306, "step": 32126 }, { "epoch": 23.536263736263738, "grad_norm": 1.125, "learning_rate": 2.7309711181362406e-05, "loss": 0.1299, "step": 32127 }, { "epoch": 23.53699633699634, "grad_norm": 0.76953125, "learning_rate": 2.730668415481098e-05, "loss": 0.1089, "step": 32128 }, { "epoch": 23.53772893772894, "grad_norm": 1.0390625, "learning_rate": 2.7303657239386378e-05, "loss": 0.1156, "step": 32129 }, { "epoch": 23.53846153846154, "grad_norm": 0.85546875, "learning_rate": 2.7300630435103054e-05, "loss": 0.1096, "step": 32130 }, { "epoch": 23.53919413919414, "grad_norm": 1.359375, "learning_rate": 2.729760374197554e-05, "loss": 0.1254, "step": 32131 }, { "epoch": 23.53992673992674, "grad_norm": 1.3515625, "learning_rate": 2.7294577160018308e-05, "loss": 0.1413, "step": 32132 }, { "epoch": 23.540659340659342, "grad_norm": 0.97265625, "learning_rate": 2.729155068924587e-05, "loss": 0.118, "step": 32133 }, { "epoch": 23.541391941391943, "grad_norm": 0.73828125, "learning_rate": 2.728852432967274e-05, "loss": 0.1219, "step": 32134 }, { "epoch": 23.542124542124544, "grad_norm": 1.7265625, "learning_rate": 2.728549808131341e-05, "loss": 0.112, "step": 32135 }, { "epoch": 23.542857142857144, "grad_norm": 1.2890625, "learning_rate": 2.728247194418239e-05, "loss": 0.1283, "step": 32136 }, { "epoch": 23.543589743589745, "grad_norm": 0.65234375, "learning_rate": 2.7279445918294156e-05, "loss": 0.1124, "step": 32137 }, { "epoch": 23.544322344322346, "grad_norm": 0.7734375, "learning_rate": 2.7276420003663235e-05, "loss": 0.1133, "step": 32138 }, { "epoch": 23.545054945054947, "grad_norm": 1.1875, "learning_rate": 2.727339420030407e-05, "loss": 0.115, "step": 32139 }, { "epoch": 23.545787545787547, "grad_norm": 1.5625, "learning_rate": 2.7270368508231242e-05, "loss": 0.1209, "step": 32140 }, { "epoch": 23.546520146520148, "grad_norm": 1.1953125, "learning_rate": 2.7267342927459177e-05, "loss": 0.1054, "step": 32141 }, { "epoch": 23.54725274725275, "grad_norm": 1.0546875, "learning_rate": 2.726431745800243e-05, "loss": 0.1309, "step": 32142 }, { "epoch": 23.54798534798535, "grad_norm": 1.0703125, "learning_rate": 2.7261292099875434e-05, "loss": 0.1155, "step": 32143 }, { "epoch": 23.54871794871795, "grad_norm": 1.0625, "learning_rate": 2.7258266853092722e-05, "loss": 0.1051, "step": 32144 }, { "epoch": 23.54945054945055, "grad_norm": 1.28125, "learning_rate": 2.725524171766879e-05, "loss": 0.1218, "step": 32145 }, { "epoch": 23.550183150183152, "grad_norm": 1.328125, "learning_rate": 2.7252216693618137e-05, "loss": 0.1101, "step": 32146 }, { "epoch": 23.550915750915753, "grad_norm": 0.80859375, "learning_rate": 2.7249191780955218e-05, "loss": 0.111, "step": 32147 }, { "epoch": 23.551648351648353, "grad_norm": 0.87890625, "learning_rate": 2.7246166979694577e-05, "loss": 0.107, "step": 32148 }, { "epoch": 23.552380952380954, "grad_norm": 0.84375, "learning_rate": 2.7243142289850665e-05, "loss": 0.1291, "step": 32149 }, { "epoch": 23.553113553113555, "grad_norm": 0.97265625, "learning_rate": 2.7240117711437996e-05, "loss": 0.1434, "step": 32150 }, { "epoch": 23.553846153846155, "grad_norm": 1.125, "learning_rate": 2.7237093244471078e-05, "loss": 0.101, "step": 32151 }, { "epoch": 23.554578754578756, "grad_norm": 1.3359375, "learning_rate": 2.723406888896437e-05, "loss": 0.1465, "step": 32152 }, { "epoch": 23.555311355311357, "grad_norm": 0.6484375, "learning_rate": 2.7231044644932385e-05, "loss": 0.1376, "step": 32153 }, { "epoch": 23.556043956043958, "grad_norm": 0.86328125, "learning_rate": 2.7228020512389575e-05, "loss": 0.1017, "step": 32154 }, { "epoch": 23.55677655677656, "grad_norm": 0.83984375, "learning_rate": 2.722499649135049e-05, "loss": 0.1135, "step": 32155 }, { "epoch": 23.55750915750916, "grad_norm": 1.109375, "learning_rate": 2.722197258182958e-05, "loss": 0.126, "step": 32156 }, { "epoch": 23.55824175824176, "grad_norm": 1.046875, "learning_rate": 2.7218948783841348e-05, "loss": 0.1228, "step": 32157 }, { "epoch": 23.55897435897436, "grad_norm": 1.046875, "learning_rate": 2.7215925097400257e-05, "loss": 0.0944, "step": 32158 }, { "epoch": 23.55970695970696, "grad_norm": 0.98828125, "learning_rate": 2.7212901522520816e-05, "loss": 0.1054, "step": 32159 }, { "epoch": 23.560439560439562, "grad_norm": 0.90234375, "learning_rate": 2.72098780592175e-05, "loss": 0.1023, "step": 32160 }, { "epoch": 23.561172161172163, "grad_norm": 1.1484375, "learning_rate": 2.720685470750482e-05, "loss": 0.134, "step": 32161 }, { "epoch": 23.561904761904763, "grad_norm": 1.546875, "learning_rate": 2.7203831467397256e-05, "loss": 0.1542, "step": 32162 }, { "epoch": 23.562637362637364, "grad_norm": 1.6015625, "learning_rate": 2.7200808338909267e-05, "loss": 0.1738, "step": 32163 }, { "epoch": 23.563369963369965, "grad_norm": 1.453125, "learning_rate": 2.719778532205537e-05, "loss": 0.1303, "step": 32164 }, { "epoch": 23.564102564102566, "grad_norm": 0.64453125, "learning_rate": 2.719476241684999e-05, "loss": 0.0983, "step": 32165 }, { "epoch": 23.564835164835166, "grad_norm": 0.92578125, "learning_rate": 2.7191739623307693e-05, "loss": 0.1282, "step": 32166 }, { "epoch": 23.565567765567767, "grad_norm": 1.1015625, "learning_rate": 2.718871694144291e-05, "loss": 0.1174, "step": 32167 }, { "epoch": 23.566300366300368, "grad_norm": 1.171875, "learning_rate": 2.7185694371270153e-05, "loss": 0.1025, "step": 32168 }, { "epoch": 23.56703296703297, "grad_norm": 0.83984375, "learning_rate": 2.7182671912803877e-05, "loss": 0.1377, "step": 32169 }, { "epoch": 23.56776556776557, "grad_norm": 1.2890625, "learning_rate": 2.7179649566058563e-05, "loss": 0.1469, "step": 32170 }, { "epoch": 23.56849816849817, "grad_norm": 0.95703125, "learning_rate": 2.7176627331048716e-05, "loss": 0.1262, "step": 32171 }, { "epoch": 23.56923076923077, "grad_norm": 1.109375, "learning_rate": 2.717360520778881e-05, "loss": 0.1221, "step": 32172 }, { "epoch": 23.56996336996337, "grad_norm": 0.76953125, "learning_rate": 2.7170583196293303e-05, "loss": 0.127, "step": 32173 }, { "epoch": 23.570695970695972, "grad_norm": 1.4765625, "learning_rate": 2.7167561296576695e-05, "loss": 0.1347, "step": 32174 }, { "epoch": 23.571428571428573, "grad_norm": 1.203125, "learning_rate": 2.7164539508653453e-05, "loss": 0.1191, "step": 32175 }, { "epoch": 23.572161172161174, "grad_norm": 0.8828125, "learning_rate": 2.716151783253807e-05, "loss": 0.1378, "step": 32176 }, { "epoch": 23.572893772893774, "grad_norm": 1.109375, "learning_rate": 2.7158496268245038e-05, "loss": 0.1195, "step": 32177 }, { "epoch": 23.573626373626375, "grad_norm": 0.96484375, "learning_rate": 2.715547481578878e-05, "loss": 0.127, "step": 32178 }, { "epoch": 23.574358974358976, "grad_norm": 0.828125, "learning_rate": 2.7152453475183836e-05, "loss": 0.1178, "step": 32179 }, { "epoch": 23.575091575091577, "grad_norm": 1.3046875, "learning_rate": 2.714943224644461e-05, "loss": 0.1115, "step": 32180 }, { "epoch": 23.575824175824177, "grad_norm": 0.90625, "learning_rate": 2.7146411129585658e-05, "loss": 0.1149, "step": 32181 }, { "epoch": 23.576556776556778, "grad_norm": 1.515625, "learning_rate": 2.7143390124621397e-05, "loss": 0.1407, "step": 32182 }, { "epoch": 23.57728937728938, "grad_norm": 0.87109375, "learning_rate": 2.7140369231566343e-05, "loss": 0.1386, "step": 32183 }, { "epoch": 23.57802197802198, "grad_norm": 0.91796875, "learning_rate": 2.7137348450434937e-05, "loss": 0.1171, "step": 32184 }, { "epoch": 23.57875457875458, "grad_norm": 1.109375, "learning_rate": 2.7134327781241648e-05, "loss": 0.1369, "step": 32185 }, { "epoch": 23.57948717948718, "grad_norm": 0.61328125, "learning_rate": 2.7131307224000975e-05, "loss": 0.1026, "step": 32186 }, { "epoch": 23.58021978021978, "grad_norm": 1.46875, "learning_rate": 2.712828677872738e-05, "loss": 0.1425, "step": 32187 }, { "epoch": 23.580952380952382, "grad_norm": 1.0703125, "learning_rate": 2.7125266445435348e-05, "loss": 0.1148, "step": 32188 }, { "epoch": 23.581684981684983, "grad_norm": 0.8828125, "learning_rate": 2.7122246224139326e-05, "loss": 0.113, "step": 32189 }, { "epoch": 23.582417582417584, "grad_norm": 1.9453125, "learning_rate": 2.7119226114853813e-05, "loss": 0.1597, "step": 32190 }, { "epoch": 23.583150183150185, "grad_norm": 1.125, "learning_rate": 2.7116206117593224e-05, "loss": 0.1115, "step": 32191 }, { "epoch": 23.583882783882785, "grad_norm": 0.7265625, "learning_rate": 2.7113186232372113e-05, "loss": 0.0895, "step": 32192 }, { "epoch": 23.584615384615386, "grad_norm": 1.59375, "learning_rate": 2.7110166459204872e-05, "loss": 0.1106, "step": 32193 }, { "epoch": 23.585347985347987, "grad_norm": 0.8515625, "learning_rate": 2.710714679810603e-05, "loss": 0.1148, "step": 32194 }, { "epoch": 23.586080586080588, "grad_norm": 1.09375, "learning_rate": 2.7104127249090005e-05, "loss": 0.1496, "step": 32195 }, { "epoch": 23.58681318681319, "grad_norm": 1.8671875, "learning_rate": 2.710110781217129e-05, "loss": 0.1063, "step": 32196 }, { "epoch": 23.58754578754579, "grad_norm": 1.15625, "learning_rate": 2.7098088487364348e-05, "loss": 0.1322, "step": 32197 }, { "epoch": 23.58827838827839, "grad_norm": 1.3984375, "learning_rate": 2.7095069274683643e-05, "loss": 0.125, "step": 32198 }, { "epoch": 23.58901098901099, "grad_norm": 1.3515625, "learning_rate": 2.709205017414367e-05, "loss": 0.1305, "step": 32199 }, { "epoch": 23.58974358974359, "grad_norm": 0.9765625, "learning_rate": 2.708903118575884e-05, "loss": 0.1074, "step": 32200 }, { "epoch": 23.590476190476192, "grad_norm": 0.8515625, "learning_rate": 2.708601230954365e-05, "loss": 0.1101, "step": 32201 }, { "epoch": 23.591208791208793, "grad_norm": 1.78125, "learning_rate": 2.7082993545512568e-05, "loss": 0.1327, "step": 32202 }, { "epoch": 23.591941391941393, "grad_norm": 1.0078125, "learning_rate": 2.7079974893680065e-05, "loss": 0.1431, "step": 32203 }, { "epoch": 23.592673992673994, "grad_norm": 1.2734375, "learning_rate": 2.7076956354060573e-05, "loss": 0.1448, "step": 32204 }, { "epoch": 23.593406593406595, "grad_norm": 1.21875, "learning_rate": 2.7073937926668593e-05, "loss": 0.1245, "step": 32205 }, { "epoch": 23.594139194139196, "grad_norm": 0.63671875, "learning_rate": 2.707091961151853e-05, "loss": 0.1006, "step": 32206 }, { "epoch": 23.594871794871796, "grad_norm": 1.5859375, "learning_rate": 2.7067901408624914e-05, "loss": 0.1206, "step": 32207 }, { "epoch": 23.595604395604397, "grad_norm": 1.2890625, "learning_rate": 2.706488331800216e-05, "loss": 0.1385, "step": 32208 }, { "epoch": 23.596336996336998, "grad_norm": 0.890625, "learning_rate": 2.706186533966476e-05, "loss": 0.0968, "step": 32209 }, { "epoch": 23.5970695970696, "grad_norm": 0.8515625, "learning_rate": 2.7058847473627137e-05, "loss": 0.1252, "step": 32210 }, { "epoch": 23.5978021978022, "grad_norm": 0.84375, "learning_rate": 2.7055829719903773e-05, "loss": 0.0838, "step": 32211 }, { "epoch": 23.5985347985348, "grad_norm": 1.3984375, "learning_rate": 2.7052812078509128e-05, "loss": 0.1188, "step": 32212 }, { "epoch": 23.5992673992674, "grad_norm": 1.03125, "learning_rate": 2.7049794549457652e-05, "loss": 0.1193, "step": 32213 }, { "epoch": 23.6, "grad_norm": 1.03125, "learning_rate": 2.704677713276383e-05, "loss": 0.1238, "step": 32214 }, { "epoch": 23.600732600732602, "grad_norm": 1.0, "learning_rate": 2.7043759828442076e-05, "loss": 0.124, "step": 32215 }, { "epoch": 23.601465201465203, "grad_norm": 0.80859375, "learning_rate": 2.7040742636506898e-05, "loss": 0.1033, "step": 32216 }, { "epoch": 23.602197802197804, "grad_norm": 0.76953125, "learning_rate": 2.7037725556972676e-05, "loss": 0.0978, "step": 32217 }, { "epoch": 23.602930402930404, "grad_norm": 0.7890625, "learning_rate": 2.7034708589853957e-05, "loss": 0.1356, "step": 32218 }, { "epoch": 23.603663003663005, "grad_norm": 1.53125, "learning_rate": 2.7031691735165122e-05, "loss": 0.1386, "step": 32219 }, { "epoch": 23.604395604395606, "grad_norm": 1.5625, "learning_rate": 2.7028674992920688e-05, "loss": 0.107, "step": 32220 }, { "epoch": 23.605128205128207, "grad_norm": 0.6328125, "learning_rate": 2.702565836313505e-05, "loss": 0.122, "step": 32221 }, { "epoch": 23.605860805860807, "grad_norm": 0.6640625, "learning_rate": 2.7022641845822693e-05, "loss": 0.1075, "step": 32222 }, { "epoch": 23.606593406593408, "grad_norm": 0.85546875, "learning_rate": 2.701962544099807e-05, "loss": 0.1082, "step": 32223 }, { "epoch": 23.60732600732601, "grad_norm": 0.90625, "learning_rate": 2.701660914867562e-05, "loss": 0.12, "step": 32224 }, { "epoch": 23.60805860805861, "grad_norm": 0.9140625, "learning_rate": 2.7013592968869827e-05, "loss": 0.1793, "step": 32225 }, { "epoch": 23.60879120879121, "grad_norm": 0.8515625, "learning_rate": 2.701057690159511e-05, "loss": 0.0942, "step": 32226 }, { "epoch": 23.60952380952381, "grad_norm": 0.984375, "learning_rate": 2.7007560946865917e-05, "loss": 0.1037, "step": 32227 }, { "epoch": 23.61025641025641, "grad_norm": 0.9140625, "learning_rate": 2.700454510469672e-05, "loss": 0.1111, "step": 32228 }, { "epoch": 23.610989010989012, "grad_norm": 1.21875, "learning_rate": 2.700152937510197e-05, "loss": 0.1388, "step": 32229 }, { "epoch": 23.611721611721613, "grad_norm": 0.95703125, "learning_rate": 2.69985137580961e-05, "loss": 0.1404, "step": 32230 }, { "epoch": 23.612454212454214, "grad_norm": 0.90625, "learning_rate": 2.6995498253693578e-05, "loss": 0.0961, "step": 32231 }, { "epoch": 23.613186813186815, "grad_norm": 0.65625, "learning_rate": 2.69924828619088e-05, "loss": 0.1208, "step": 32232 }, { "epoch": 23.613919413919415, "grad_norm": 1.078125, "learning_rate": 2.6989467582756295e-05, "loss": 0.1261, "step": 32233 }, { "epoch": 23.614652014652016, "grad_norm": 1.109375, "learning_rate": 2.698645241625045e-05, "loss": 0.1284, "step": 32234 }, { "epoch": 23.615384615384617, "grad_norm": 0.79296875, "learning_rate": 2.698343736240575e-05, "loss": 0.1165, "step": 32235 }, { "epoch": 23.616117216117217, "grad_norm": 1.3671875, "learning_rate": 2.69804224212366e-05, "loss": 0.1184, "step": 32236 }, { "epoch": 23.616849816849818, "grad_norm": 0.96875, "learning_rate": 2.697740759275747e-05, "loss": 0.1114, "step": 32237 }, { "epoch": 23.61758241758242, "grad_norm": 1.125, "learning_rate": 2.6974392876982794e-05, "loss": 0.1177, "step": 32238 }, { "epoch": 23.61831501831502, "grad_norm": 1.359375, "learning_rate": 2.6971378273927034e-05, "loss": 0.1384, "step": 32239 }, { "epoch": 23.61904761904762, "grad_norm": 0.98828125, "learning_rate": 2.6968363783604643e-05, "loss": 0.1125, "step": 32240 }, { "epoch": 23.61978021978022, "grad_norm": 0.98046875, "learning_rate": 2.6965349406030025e-05, "loss": 0.1354, "step": 32241 }, { "epoch": 23.620512820512822, "grad_norm": 1.0234375, "learning_rate": 2.6962335141217665e-05, "loss": 0.1504, "step": 32242 }, { "epoch": 23.621245421245423, "grad_norm": 1.0078125, "learning_rate": 2.695932098918194e-05, "loss": 0.1539, "step": 32243 }, { "epoch": 23.621978021978023, "grad_norm": 0.8515625, "learning_rate": 2.6956306949937382e-05, "loss": 0.1157, "step": 32244 }, { "epoch": 23.622710622710624, "grad_norm": 1.21875, "learning_rate": 2.695329302349836e-05, "loss": 0.112, "step": 32245 }, { "epoch": 23.623443223443225, "grad_norm": 1.171875, "learning_rate": 2.695027920987937e-05, "loss": 0.1235, "step": 32246 }, { "epoch": 23.624175824175826, "grad_norm": 1.859375, "learning_rate": 2.69472655090948e-05, "loss": 0.1678, "step": 32247 }, { "epoch": 23.624908424908426, "grad_norm": 1.3046875, "learning_rate": 2.694425192115911e-05, "loss": 0.1245, "step": 32248 }, { "epoch": 23.625641025641027, "grad_norm": 1.1015625, "learning_rate": 2.6941238446086737e-05, "loss": 0.1301, "step": 32249 }, { "epoch": 23.626373626373628, "grad_norm": 1.578125, "learning_rate": 2.6938225083892133e-05, "loss": 0.1363, "step": 32250 }, { "epoch": 23.62710622710623, "grad_norm": 0.875, "learning_rate": 2.693521183458974e-05, "loss": 0.1345, "step": 32251 }, { "epoch": 23.62783882783883, "grad_norm": 1.4921875, "learning_rate": 2.6932198698193965e-05, "loss": 0.1322, "step": 32252 }, { "epoch": 23.62857142857143, "grad_norm": 1.296875, "learning_rate": 2.692918567471927e-05, "loss": 0.1275, "step": 32253 }, { "epoch": 23.62930402930403, "grad_norm": 1.8046875, "learning_rate": 2.692617276418008e-05, "loss": 0.1202, "step": 32254 }, { "epoch": 23.63003663003663, "grad_norm": 0.67578125, "learning_rate": 2.692315996659085e-05, "loss": 0.1108, "step": 32255 }, { "epoch": 23.630769230769232, "grad_norm": 1.265625, "learning_rate": 2.6920147281965984e-05, "loss": 0.1268, "step": 32256 }, { "epoch": 23.631501831501833, "grad_norm": 1.1171875, "learning_rate": 2.6917134710319952e-05, "loss": 0.1283, "step": 32257 }, { "epoch": 23.632234432234434, "grad_norm": 1.0859375, "learning_rate": 2.691412225166713e-05, "loss": 0.1228, "step": 32258 }, { "epoch": 23.632967032967034, "grad_norm": 1.234375, "learning_rate": 2.691110990602204e-05, "loss": 0.114, "step": 32259 }, { "epoch": 23.633699633699635, "grad_norm": 0.98046875, "learning_rate": 2.6908097673399036e-05, "loss": 0.1266, "step": 32260 }, { "epoch": 23.634432234432236, "grad_norm": 1.109375, "learning_rate": 2.6905085553812597e-05, "loss": 0.0991, "step": 32261 }, { "epoch": 23.635164835164836, "grad_norm": 1.484375, "learning_rate": 2.6902073547277122e-05, "loss": 0.1054, "step": 32262 }, { "epoch": 23.635897435897437, "grad_norm": 1.4921875, "learning_rate": 2.6899061653807068e-05, "loss": 0.1216, "step": 32263 }, { "epoch": 23.636630036630038, "grad_norm": 1.15625, "learning_rate": 2.689604987341685e-05, "loss": 0.1108, "step": 32264 }, { "epoch": 23.63736263736264, "grad_norm": 1.2265625, "learning_rate": 2.6893038206120906e-05, "loss": 0.1188, "step": 32265 }, { "epoch": 23.63809523809524, "grad_norm": 0.9921875, "learning_rate": 2.6890026651933686e-05, "loss": 0.131, "step": 32266 }, { "epoch": 23.63882783882784, "grad_norm": 2.234375, "learning_rate": 2.688701521086958e-05, "loss": 0.146, "step": 32267 }, { "epoch": 23.63956043956044, "grad_norm": 0.73828125, "learning_rate": 2.688400388294306e-05, "loss": 0.0997, "step": 32268 }, { "epoch": 23.64029304029304, "grad_norm": 0.74609375, "learning_rate": 2.6880992668168487e-05, "loss": 0.1091, "step": 32269 }, { "epoch": 23.641025641025642, "grad_norm": 1.5234375, "learning_rate": 2.6877981566560363e-05, "loss": 0.1281, "step": 32270 }, { "epoch": 23.641758241758243, "grad_norm": 0.828125, "learning_rate": 2.6874970578133076e-05, "loss": 0.1145, "step": 32271 }, { "epoch": 23.642490842490844, "grad_norm": 0.8203125, "learning_rate": 2.687195970290108e-05, "loss": 0.1113, "step": 32272 }, { "epoch": 23.643223443223444, "grad_norm": 0.9140625, "learning_rate": 2.686894894087875e-05, "loss": 0.1341, "step": 32273 }, { "epoch": 23.643956043956045, "grad_norm": 0.9140625, "learning_rate": 2.686593829208055e-05, "loss": 0.1224, "step": 32274 }, { "epoch": 23.644688644688646, "grad_norm": 0.984375, "learning_rate": 2.68629277565209e-05, "loss": 0.1487, "step": 32275 }, { "epoch": 23.645421245421247, "grad_norm": 0.90625, "learning_rate": 2.6859917334214224e-05, "loss": 0.1048, "step": 32276 }, { "epoch": 23.646153846153847, "grad_norm": 1.3984375, "learning_rate": 2.685690702517496e-05, "loss": 0.1228, "step": 32277 }, { "epoch": 23.646886446886448, "grad_norm": 0.96484375, "learning_rate": 2.68538968294175e-05, "loss": 0.1386, "step": 32278 }, { "epoch": 23.64761904761905, "grad_norm": 1.265625, "learning_rate": 2.6850886746956278e-05, "loss": 0.1392, "step": 32279 }, { "epoch": 23.64835164835165, "grad_norm": 0.8046875, "learning_rate": 2.684787677780572e-05, "loss": 0.1596, "step": 32280 }, { "epoch": 23.64908424908425, "grad_norm": 1.0859375, "learning_rate": 2.6844866921980277e-05, "loss": 0.1227, "step": 32281 }, { "epoch": 23.64981684981685, "grad_norm": 1.625, "learning_rate": 2.684185717949432e-05, "loss": 0.1376, "step": 32282 }, { "epoch": 23.650549450549452, "grad_norm": 0.8671875, "learning_rate": 2.6838847550362307e-05, "loss": 0.1315, "step": 32283 }, { "epoch": 23.651282051282053, "grad_norm": 0.87109375, "learning_rate": 2.6835838034598608e-05, "loss": 0.0942, "step": 32284 }, { "epoch": 23.652014652014653, "grad_norm": 2.3125, "learning_rate": 2.683282863221771e-05, "loss": 0.1809, "step": 32285 }, { "epoch": 23.652747252747254, "grad_norm": 1.4375, "learning_rate": 2.682981934323398e-05, "loss": 0.1321, "step": 32286 }, { "epoch": 23.653479853479855, "grad_norm": 0.97265625, "learning_rate": 2.6826810167661887e-05, "loss": 0.1388, "step": 32287 }, { "epoch": 23.654212454212455, "grad_norm": 1.203125, "learning_rate": 2.682380110551579e-05, "loss": 0.1063, "step": 32288 }, { "epoch": 23.654945054945056, "grad_norm": 1.1015625, "learning_rate": 2.6820792156810133e-05, "loss": 0.1095, "step": 32289 }, { "epoch": 23.655677655677657, "grad_norm": 0.91015625, "learning_rate": 2.6817783321559337e-05, "loss": 0.1015, "step": 32290 }, { "epoch": 23.656410256410258, "grad_norm": 1.3984375, "learning_rate": 2.681477459977782e-05, "loss": 0.1036, "step": 32291 }, { "epoch": 23.65714285714286, "grad_norm": 1.6015625, "learning_rate": 2.6811765991479997e-05, "loss": 0.1229, "step": 32292 }, { "epoch": 23.65787545787546, "grad_norm": 0.95703125, "learning_rate": 2.6808757496680275e-05, "loss": 0.1364, "step": 32293 }, { "epoch": 23.65860805860806, "grad_norm": 0.52734375, "learning_rate": 2.680574911539309e-05, "loss": 0.1119, "step": 32294 }, { "epoch": 23.65934065934066, "grad_norm": 0.78125, "learning_rate": 2.6802740847632795e-05, "loss": 0.1102, "step": 32295 }, { "epoch": 23.66007326007326, "grad_norm": 0.9296875, "learning_rate": 2.679973269341389e-05, "loss": 0.128, "step": 32296 }, { "epoch": 23.660805860805862, "grad_norm": 0.9375, "learning_rate": 2.679672465275073e-05, "loss": 0.0969, "step": 32297 }, { "epoch": 23.661538461538463, "grad_norm": 1.6015625, "learning_rate": 2.679371672565776e-05, "loss": 0.1342, "step": 32298 }, { "epoch": 23.662271062271063, "grad_norm": 0.9296875, "learning_rate": 2.6790708912149353e-05, "loss": 0.1106, "step": 32299 }, { "epoch": 23.663003663003664, "grad_norm": 1.0390625, "learning_rate": 2.6787701212239945e-05, "loss": 0.1358, "step": 32300 }, { "epoch": 23.663736263736265, "grad_norm": 0.98046875, "learning_rate": 2.6784693625943947e-05, "loss": 0.1165, "step": 32301 }, { "epoch": 23.664468864468866, "grad_norm": 1.6171875, "learning_rate": 2.6781686153275765e-05, "loss": 0.1205, "step": 32302 }, { "epoch": 23.665201465201466, "grad_norm": 1.609375, "learning_rate": 2.6778678794249833e-05, "loss": 0.1194, "step": 32303 }, { "epoch": 23.665934065934067, "grad_norm": 1.28125, "learning_rate": 2.6775671548880517e-05, "loss": 0.1165, "step": 32304 }, { "epoch": 23.666666666666668, "grad_norm": 1.046875, "learning_rate": 2.6772664417182246e-05, "loss": 0.1423, "step": 32305 }, { "epoch": 23.66739926739927, "grad_norm": 1.3125, "learning_rate": 2.6769657399169424e-05, "loss": 0.123, "step": 32306 }, { "epoch": 23.66813186813187, "grad_norm": 0.609375, "learning_rate": 2.6766650494856488e-05, "loss": 0.1276, "step": 32307 }, { "epoch": 23.66886446886447, "grad_norm": 0.8828125, "learning_rate": 2.6763643704257805e-05, "loss": 0.1161, "step": 32308 }, { "epoch": 23.66959706959707, "grad_norm": 0.98828125, "learning_rate": 2.6760637027387815e-05, "loss": 0.1225, "step": 32309 }, { "epoch": 23.67032967032967, "grad_norm": 1.125, "learning_rate": 2.6757630464260864e-05, "loss": 0.1039, "step": 32310 }, { "epoch": 23.671062271062272, "grad_norm": 1.3203125, "learning_rate": 2.6754624014891437e-05, "loss": 0.1434, "step": 32311 }, { "epoch": 23.671794871794873, "grad_norm": 0.6640625, "learning_rate": 2.6751617679293887e-05, "loss": 0.1151, "step": 32312 }, { "epoch": 23.672527472527474, "grad_norm": 1.5859375, "learning_rate": 2.674861145748263e-05, "loss": 0.0967, "step": 32313 }, { "epoch": 23.673260073260074, "grad_norm": 1.1796875, "learning_rate": 2.6745605349472093e-05, "loss": 0.1156, "step": 32314 }, { "epoch": 23.673992673992675, "grad_norm": 1.03125, "learning_rate": 2.6742599355276637e-05, "loss": 0.1051, "step": 32315 }, { "epoch": 23.674725274725276, "grad_norm": 0.8359375, "learning_rate": 2.673959347491069e-05, "loss": 0.137, "step": 32316 }, { "epoch": 23.675457875457877, "grad_norm": 0.76953125, "learning_rate": 2.6736587708388648e-05, "loss": 0.1313, "step": 32317 }, { "epoch": 23.676190476190477, "grad_norm": 0.66796875, "learning_rate": 2.6733582055724937e-05, "loss": 0.1433, "step": 32318 }, { "epoch": 23.676923076923078, "grad_norm": 0.87890625, "learning_rate": 2.673057651693392e-05, "loss": 0.1266, "step": 32319 }, { "epoch": 23.67765567765568, "grad_norm": 1.2421875, "learning_rate": 2.6727571092030023e-05, "loss": 0.1007, "step": 32320 }, { "epoch": 23.67838827838828, "grad_norm": 1.75, "learning_rate": 2.672456578102761e-05, "loss": 0.1356, "step": 32321 }, { "epoch": 23.67912087912088, "grad_norm": 1.609375, "learning_rate": 2.672156058394114e-05, "loss": 0.1102, "step": 32322 }, { "epoch": 23.67985347985348, "grad_norm": 0.82421875, "learning_rate": 2.671855550078496e-05, "loss": 0.093, "step": 32323 }, { "epoch": 23.68058608058608, "grad_norm": 0.90625, "learning_rate": 2.6715550531573502e-05, "loss": 0.1128, "step": 32324 }, { "epoch": 23.681318681318682, "grad_norm": 0.79296875, "learning_rate": 2.6712545676321133e-05, "loss": 0.118, "step": 32325 }, { "epoch": 23.682051282051283, "grad_norm": 0.859375, "learning_rate": 2.670954093504227e-05, "loss": 0.1104, "step": 32326 }, { "epoch": 23.682783882783884, "grad_norm": 1.0859375, "learning_rate": 2.6706536307751306e-05, "loss": 0.1513, "step": 32327 }, { "epoch": 23.683516483516485, "grad_norm": 0.8671875, "learning_rate": 2.6703531794462638e-05, "loss": 0.1354, "step": 32328 }, { "epoch": 23.684249084249085, "grad_norm": 0.98828125, "learning_rate": 2.670052739519068e-05, "loss": 0.137, "step": 32329 }, { "epoch": 23.684981684981686, "grad_norm": 1.1484375, "learning_rate": 2.6697523109949785e-05, "loss": 0.1072, "step": 32330 }, { "epoch": 23.685714285714287, "grad_norm": 0.828125, "learning_rate": 2.6694518938754377e-05, "loss": 0.1347, "step": 32331 }, { "epoch": 23.686446886446888, "grad_norm": 0.84765625, "learning_rate": 2.6691514881618836e-05, "loss": 0.135, "step": 32332 }, { "epoch": 23.68717948717949, "grad_norm": 0.9765625, "learning_rate": 2.6688510938557586e-05, "loss": 0.1032, "step": 32333 }, { "epoch": 23.68791208791209, "grad_norm": 1.4453125, "learning_rate": 2.668550710958498e-05, "loss": 0.1491, "step": 32334 }, { "epoch": 23.68864468864469, "grad_norm": 1.3515625, "learning_rate": 2.668250339471544e-05, "loss": 0.1407, "step": 32335 }, { "epoch": 23.68937728937729, "grad_norm": 1.0703125, "learning_rate": 2.6679499793963312e-05, "loss": 0.1352, "step": 32336 }, { "epoch": 23.69010989010989, "grad_norm": 0.7578125, "learning_rate": 2.6676496307343056e-05, "loss": 0.1298, "step": 32337 }, { "epoch": 23.690842490842492, "grad_norm": 1.7578125, "learning_rate": 2.667349293486901e-05, "loss": 0.1346, "step": 32338 }, { "epoch": 23.691575091575093, "grad_norm": 0.828125, "learning_rate": 2.667048967655558e-05, "loss": 0.117, "step": 32339 }, { "epoch": 23.692307692307693, "grad_norm": 1.609375, "learning_rate": 2.6667486532417175e-05, "loss": 0.1286, "step": 32340 }, { "epoch": 23.693040293040294, "grad_norm": 0.875, "learning_rate": 2.6664483502468146e-05, "loss": 0.1095, "step": 32341 }, { "epoch": 23.693772893772895, "grad_norm": 1.0546875, "learning_rate": 2.6661480586722908e-05, "loss": 0.1508, "step": 32342 }, { "epoch": 23.694505494505496, "grad_norm": 0.8046875, "learning_rate": 2.6658477785195835e-05, "loss": 0.1169, "step": 32343 }, { "epoch": 23.695238095238096, "grad_norm": 0.77734375, "learning_rate": 2.665547509790134e-05, "loss": 0.134, "step": 32344 }, { "epoch": 23.695970695970697, "grad_norm": 2.125, "learning_rate": 2.6652472524853777e-05, "loss": 0.1333, "step": 32345 }, { "epoch": 23.696703296703298, "grad_norm": 1.2109375, "learning_rate": 2.664947006606756e-05, "loss": 0.1158, "step": 32346 }, { "epoch": 23.6974358974359, "grad_norm": 1.3828125, "learning_rate": 2.6646467721557027e-05, "loss": 0.1545, "step": 32347 }, { "epoch": 23.6981684981685, "grad_norm": 1.2734375, "learning_rate": 2.6643465491336637e-05, "loss": 0.1194, "step": 32348 }, { "epoch": 23.6989010989011, "grad_norm": 0.921875, "learning_rate": 2.6640463375420715e-05, "loss": 0.1132, "step": 32349 }, { "epoch": 23.6996336996337, "grad_norm": 0.76171875, "learning_rate": 2.6637461373823687e-05, "loss": 0.1239, "step": 32350 }, { "epoch": 23.7003663003663, "grad_norm": 1.0078125, "learning_rate": 2.6634459486559894e-05, "loss": 0.1147, "step": 32351 }, { "epoch": 23.701098901098902, "grad_norm": 0.97265625, "learning_rate": 2.6631457713643743e-05, "loss": 0.1012, "step": 32352 }, { "epoch": 23.701831501831503, "grad_norm": 0.71875, "learning_rate": 2.6628456055089607e-05, "loss": 0.1038, "step": 32353 }, { "epoch": 23.702564102564104, "grad_norm": 0.703125, "learning_rate": 2.6625454510911875e-05, "loss": 0.1096, "step": 32354 }, { "epoch": 23.703296703296704, "grad_norm": 0.87109375, "learning_rate": 2.6622453081124953e-05, "loss": 0.0956, "step": 32355 }, { "epoch": 23.704029304029305, "grad_norm": 0.71875, "learning_rate": 2.6619451765743176e-05, "loss": 0.1069, "step": 32356 }, { "epoch": 23.704761904761906, "grad_norm": 1.1875, "learning_rate": 2.6616450564780944e-05, "loss": 0.1503, "step": 32357 }, { "epoch": 23.705494505494507, "grad_norm": 0.8671875, "learning_rate": 2.661344947825264e-05, "loss": 0.1456, "step": 32358 }, { "epoch": 23.706227106227107, "grad_norm": 0.921875, "learning_rate": 2.661044850617266e-05, "loss": 0.1183, "step": 32359 }, { "epoch": 23.706959706959708, "grad_norm": 0.9453125, "learning_rate": 2.660744764855534e-05, "loss": 0.1136, "step": 32360 }, { "epoch": 23.70769230769231, "grad_norm": 1.2421875, "learning_rate": 2.6604446905415105e-05, "loss": 0.1131, "step": 32361 }, { "epoch": 23.70842490842491, "grad_norm": 1.09375, "learning_rate": 2.660144627676627e-05, "loss": 0.1437, "step": 32362 }, { "epoch": 23.70915750915751, "grad_norm": 1.0625, "learning_rate": 2.659844576262328e-05, "loss": 0.0836, "step": 32363 }, { "epoch": 23.70989010989011, "grad_norm": 0.83203125, "learning_rate": 2.6595445363000482e-05, "loss": 0.116, "step": 32364 }, { "epoch": 23.71062271062271, "grad_norm": 0.96875, "learning_rate": 2.659244507791224e-05, "loss": 0.1037, "step": 32365 }, { "epoch": 23.711355311355312, "grad_norm": 1.21875, "learning_rate": 2.6589444907372962e-05, "loss": 0.1365, "step": 32366 }, { "epoch": 23.712087912087913, "grad_norm": 1.140625, "learning_rate": 2.6586444851396987e-05, "loss": 0.1435, "step": 32367 }, { "epoch": 23.712820512820514, "grad_norm": 1.203125, "learning_rate": 2.6583444909998707e-05, "loss": 0.1442, "step": 32368 }, { "epoch": 23.713553113553115, "grad_norm": 1.1953125, "learning_rate": 2.658044508319249e-05, "loss": 0.1148, "step": 32369 }, { "epoch": 23.714285714285715, "grad_norm": 1.25, "learning_rate": 2.6577445370992727e-05, "loss": 0.1128, "step": 32370 }, { "epoch": 23.715018315018316, "grad_norm": 1.2109375, "learning_rate": 2.6574445773413764e-05, "loss": 0.1276, "step": 32371 }, { "epoch": 23.715750915750917, "grad_norm": 1.03125, "learning_rate": 2.6571446290470004e-05, "loss": 0.1109, "step": 32372 }, { "epoch": 23.716483516483517, "grad_norm": 0.703125, "learning_rate": 2.6568446922175765e-05, "loss": 0.1114, "step": 32373 }, { "epoch": 23.717216117216118, "grad_norm": 0.9765625, "learning_rate": 2.656544766854549e-05, "loss": 0.1159, "step": 32374 }, { "epoch": 23.71794871794872, "grad_norm": 0.83203125, "learning_rate": 2.6562448529593486e-05, "loss": 0.1165, "step": 32375 }, { "epoch": 23.71868131868132, "grad_norm": 1.1328125, "learning_rate": 2.6559449505334177e-05, "loss": 0.1302, "step": 32376 }, { "epoch": 23.71941391941392, "grad_norm": 1.1796875, "learning_rate": 2.655645059578188e-05, "loss": 0.0972, "step": 32377 }, { "epoch": 23.72014652014652, "grad_norm": 0.7734375, "learning_rate": 2.655345180095099e-05, "loss": 0.1126, "step": 32378 }, { "epoch": 23.720879120879122, "grad_norm": 0.96484375, "learning_rate": 2.6550453120855878e-05, "loss": 0.1053, "step": 32379 }, { "epoch": 23.721611721611723, "grad_norm": 0.94921875, "learning_rate": 2.6547454555510902e-05, "loss": 0.1309, "step": 32380 }, { "epoch": 23.722344322344323, "grad_norm": 1.34375, "learning_rate": 2.6544456104930448e-05, "loss": 0.1137, "step": 32381 }, { "epoch": 23.723076923076924, "grad_norm": 1.4375, "learning_rate": 2.654145776912886e-05, "loss": 0.1353, "step": 32382 }, { "epoch": 23.723809523809525, "grad_norm": 1.546875, "learning_rate": 2.653845954812051e-05, "loss": 0.1859, "step": 32383 }, { "epoch": 23.724542124542126, "grad_norm": 1.0234375, "learning_rate": 2.653546144191977e-05, "loss": 0.1129, "step": 32384 }, { "epoch": 23.725274725274726, "grad_norm": 1.0625, "learning_rate": 2.653246345054101e-05, "loss": 0.1254, "step": 32385 }, { "epoch": 23.726007326007327, "grad_norm": 1.15625, "learning_rate": 2.6529465573998573e-05, "loss": 0.1162, "step": 32386 }, { "epoch": 23.726739926739928, "grad_norm": 1.1328125, "learning_rate": 2.652646781230685e-05, "loss": 0.1188, "step": 32387 }, { "epoch": 23.72747252747253, "grad_norm": 1.921875, "learning_rate": 2.6523470165480157e-05, "loss": 0.1401, "step": 32388 }, { "epoch": 23.72820512820513, "grad_norm": 1.359375, "learning_rate": 2.652047263353292e-05, "loss": 0.1169, "step": 32389 }, { "epoch": 23.72893772893773, "grad_norm": 1.5078125, "learning_rate": 2.651747521647946e-05, "loss": 0.132, "step": 32390 }, { "epoch": 23.72967032967033, "grad_norm": 1.0859375, "learning_rate": 2.6514477914334152e-05, "loss": 0.1431, "step": 32391 }, { "epoch": 23.73040293040293, "grad_norm": 1.1015625, "learning_rate": 2.6511480727111365e-05, "loss": 0.1258, "step": 32392 }, { "epoch": 23.731135531135532, "grad_norm": 1.6484375, "learning_rate": 2.6508483654825433e-05, "loss": 0.1102, "step": 32393 }, { "epoch": 23.731868131868133, "grad_norm": 1.2265625, "learning_rate": 2.6505486697490728e-05, "loss": 0.1663, "step": 32394 }, { "epoch": 23.732600732600734, "grad_norm": 0.84375, "learning_rate": 2.650248985512162e-05, "loss": 0.1447, "step": 32395 }, { "epoch": 23.733333333333334, "grad_norm": 1.0625, "learning_rate": 2.6499493127732477e-05, "loss": 0.1291, "step": 32396 }, { "epoch": 23.734065934065935, "grad_norm": 0.84375, "learning_rate": 2.6496496515337622e-05, "loss": 0.1367, "step": 32397 }, { "epoch": 23.734798534798536, "grad_norm": 1.2109375, "learning_rate": 2.6493500017951457e-05, "loss": 0.1211, "step": 32398 }, { "epoch": 23.735531135531136, "grad_norm": 1.0625, "learning_rate": 2.6490503635588267e-05, "loss": 0.1369, "step": 32399 }, { "epoch": 23.736263736263737, "grad_norm": 0.7890625, "learning_rate": 2.6487507368262498e-05, "loss": 0.1204, "step": 32400 }, { "epoch": 23.736996336996338, "grad_norm": 1.171875, "learning_rate": 2.648451121598844e-05, "loss": 0.112, "step": 32401 }, { "epoch": 23.73772893772894, "grad_norm": 1.453125, "learning_rate": 2.6481515178780498e-05, "loss": 0.1281, "step": 32402 }, { "epoch": 23.73846153846154, "grad_norm": 1.1875, "learning_rate": 2.6478519256652978e-05, "loss": 0.0952, "step": 32403 }, { "epoch": 23.73919413919414, "grad_norm": 1.1015625, "learning_rate": 2.647552344962026e-05, "loss": 0.1339, "step": 32404 }, { "epoch": 23.73992673992674, "grad_norm": 1.1015625, "learning_rate": 2.6472527757696692e-05, "loss": 0.1484, "step": 32405 }, { "epoch": 23.74065934065934, "grad_norm": 1.0234375, "learning_rate": 2.6469532180896638e-05, "loss": 0.1214, "step": 32406 }, { "epoch": 23.741391941391942, "grad_norm": 0.93359375, "learning_rate": 2.6466536719234463e-05, "loss": 0.0969, "step": 32407 }, { "epoch": 23.742124542124543, "grad_norm": 1.28125, "learning_rate": 2.6463541372724472e-05, "loss": 0.1407, "step": 32408 }, { "epoch": 23.742857142857144, "grad_norm": 1.3046875, "learning_rate": 2.6460546141381055e-05, "loss": 0.1189, "step": 32409 }, { "epoch": 23.743589743589745, "grad_norm": 1.59375, "learning_rate": 2.645755102521855e-05, "loss": 0.1394, "step": 32410 }, { "epoch": 23.744322344322345, "grad_norm": 0.72265625, "learning_rate": 2.645455602425133e-05, "loss": 0.1065, "step": 32411 }, { "epoch": 23.745054945054946, "grad_norm": 0.8359375, "learning_rate": 2.64515611384937e-05, "loss": 0.1077, "step": 32412 }, { "epoch": 23.745787545787547, "grad_norm": 1.9140625, "learning_rate": 2.644856636796007e-05, "loss": 0.1545, "step": 32413 }, { "epoch": 23.746520146520147, "grad_norm": 0.72265625, "learning_rate": 2.6445571712664703e-05, "loss": 0.1158, "step": 32414 }, { "epoch": 23.747252747252748, "grad_norm": 0.93359375, "learning_rate": 2.6442577172622042e-05, "loss": 0.1441, "step": 32415 }, { "epoch": 23.74798534798535, "grad_norm": 1.21875, "learning_rate": 2.6439582747846366e-05, "loss": 0.1271, "step": 32416 }, { "epoch": 23.74871794871795, "grad_norm": 0.97265625, "learning_rate": 2.643658843835205e-05, "loss": 0.1223, "step": 32417 }, { "epoch": 23.74945054945055, "grad_norm": 1.5078125, "learning_rate": 2.643359424415346e-05, "loss": 0.1136, "step": 32418 }, { "epoch": 23.75018315018315, "grad_norm": 0.91015625, "learning_rate": 2.6430600165264897e-05, "loss": 0.0937, "step": 32419 }, { "epoch": 23.750915750915752, "grad_norm": 0.7890625, "learning_rate": 2.6427606201700734e-05, "loss": 0.117, "step": 32420 }, { "epoch": 23.751648351648353, "grad_norm": 2.078125, "learning_rate": 2.6424612353475313e-05, "loss": 0.1146, "step": 32421 }, { "epoch": 23.752380952380953, "grad_norm": 0.9296875, "learning_rate": 2.6421618620602996e-05, "loss": 0.1233, "step": 32422 }, { "epoch": 23.753113553113554, "grad_norm": 1.4375, "learning_rate": 2.6418625003098084e-05, "loss": 0.1353, "step": 32423 }, { "epoch": 23.753846153846155, "grad_norm": 1.484375, "learning_rate": 2.6415631500974973e-05, "loss": 0.1565, "step": 32424 }, { "epoch": 23.754578754578755, "grad_norm": 1.0625, "learning_rate": 2.6412638114247935e-05, "loss": 0.1552, "step": 32425 }, { "epoch": 23.755311355311356, "grad_norm": 1.5078125, "learning_rate": 2.6409644842931393e-05, "loss": 0.1205, "step": 32426 }, { "epoch": 23.756043956043957, "grad_norm": 1.5, "learning_rate": 2.640665168703963e-05, "loss": 0.1586, "step": 32427 }, { "epoch": 23.756776556776558, "grad_norm": 1.03125, "learning_rate": 2.640365864658701e-05, "loss": 0.1133, "step": 32428 }, { "epoch": 23.75750915750916, "grad_norm": 1.0234375, "learning_rate": 2.640066572158789e-05, "loss": 0.1051, "step": 32429 }, { "epoch": 23.75824175824176, "grad_norm": 0.95703125, "learning_rate": 2.639767291205658e-05, "loss": 0.1321, "step": 32430 }, { "epoch": 23.75897435897436, "grad_norm": 0.75390625, "learning_rate": 2.6394680218007424e-05, "loss": 0.0909, "step": 32431 }, { "epoch": 23.75970695970696, "grad_norm": 0.984375, "learning_rate": 2.6391687639454772e-05, "loss": 0.1162, "step": 32432 }, { "epoch": 23.76043956043956, "grad_norm": 1.4375, "learning_rate": 2.6388695176412975e-05, "loss": 0.1576, "step": 32433 }, { "epoch": 23.761172161172162, "grad_norm": 0.921875, "learning_rate": 2.638570282889633e-05, "loss": 0.1111, "step": 32434 }, { "epoch": 23.761904761904763, "grad_norm": 0.703125, "learning_rate": 2.6382710596919213e-05, "loss": 0.1049, "step": 32435 }, { "epoch": 23.762637362637363, "grad_norm": 0.8125, "learning_rate": 2.637971848049594e-05, "loss": 0.0952, "step": 32436 }, { "epoch": 23.763369963369964, "grad_norm": 0.6015625, "learning_rate": 2.637672647964087e-05, "loss": 0.1114, "step": 32437 }, { "epoch": 23.764102564102565, "grad_norm": 1.0859375, "learning_rate": 2.6373734594368306e-05, "loss": 0.1256, "step": 32438 }, { "epoch": 23.764835164835166, "grad_norm": 0.75390625, "learning_rate": 2.6370742824692612e-05, "loss": 0.1309, "step": 32439 }, { "epoch": 23.765567765567766, "grad_norm": 0.6640625, "learning_rate": 2.6367751170628097e-05, "loss": 0.1026, "step": 32440 }, { "epoch": 23.766300366300367, "grad_norm": 0.734375, "learning_rate": 2.636475963218911e-05, "loss": 0.0844, "step": 32441 }, { "epoch": 23.767032967032968, "grad_norm": 0.84375, "learning_rate": 2.636176820938998e-05, "loss": 0.1098, "step": 32442 }, { "epoch": 23.76776556776557, "grad_norm": 0.828125, "learning_rate": 2.635877690224505e-05, "loss": 0.1271, "step": 32443 }, { "epoch": 23.76849816849817, "grad_norm": 0.9765625, "learning_rate": 2.635578571076866e-05, "loss": 0.1114, "step": 32444 }, { "epoch": 23.76923076923077, "grad_norm": 0.8125, "learning_rate": 2.6352794634975104e-05, "loss": 0.1312, "step": 32445 }, { "epoch": 23.76996336996337, "grad_norm": 0.9140625, "learning_rate": 2.6349803674878737e-05, "loss": 0.112, "step": 32446 }, { "epoch": 23.77069597069597, "grad_norm": 1.515625, "learning_rate": 2.6346812830493883e-05, "loss": 0.1256, "step": 32447 }, { "epoch": 23.771428571428572, "grad_norm": 0.67578125, "learning_rate": 2.634382210183491e-05, "loss": 0.1036, "step": 32448 }, { "epoch": 23.772161172161173, "grad_norm": 1.453125, "learning_rate": 2.6340831488916086e-05, "loss": 0.1378, "step": 32449 }, { "epoch": 23.772893772893774, "grad_norm": 0.94140625, "learning_rate": 2.633784099175179e-05, "loss": 0.0873, "step": 32450 }, { "epoch": 23.773626373626374, "grad_norm": 1.2734375, "learning_rate": 2.6334850610356293e-05, "loss": 0.1436, "step": 32451 }, { "epoch": 23.774358974358975, "grad_norm": 1.4296875, "learning_rate": 2.6331860344743997e-05, "loss": 0.1307, "step": 32452 }, { "epoch": 23.775091575091576, "grad_norm": 1.2265625, "learning_rate": 2.6328870194929172e-05, "loss": 0.1285, "step": 32453 }, { "epoch": 23.775824175824177, "grad_norm": 1.078125, "learning_rate": 2.6325880160926163e-05, "loss": 0.0978, "step": 32454 }, { "epoch": 23.776556776556777, "grad_norm": 1.078125, "learning_rate": 2.6322890242749327e-05, "loss": 0.147, "step": 32455 }, { "epoch": 23.777289377289378, "grad_norm": 1.2890625, "learning_rate": 2.6319900440412933e-05, "loss": 0.1265, "step": 32456 }, { "epoch": 23.77802197802198, "grad_norm": 1.0234375, "learning_rate": 2.6316910753931334e-05, "loss": 0.1065, "step": 32457 }, { "epoch": 23.77875457875458, "grad_norm": 0.87109375, "learning_rate": 2.6313921183318858e-05, "loss": 0.1273, "step": 32458 }, { "epoch": 23.77948717948718, "grad_norm": 1.5546875, "learning_rate": 2.6310931728589837e-05, "loss": 0.1471, "step": 32459 }, { "epoch": 23.78021978021978, "grad_norm": 0.98046875, "learning_rate": 2.6307942389758573e-05, "loss": 0.1033, "step": 32460 }, { "epoch": 23.78095238095238, "grad_norm": 1.25, "learning_rate": 2.630495316683939e-05, "loss": 0.1116, "step": 32461 }, { "epoch": 23.781684981684982, "grad_norm": 0.9296875, "learning_rate": 2.630196405984663e-05, "loss": 0.1104, "step": 32462 }, { "epoch": 23.782417582417583, "grad_norm": 0.75, "learning_rate": 2.629897506879462e-05, "loss": 0.1344, "step": 32463 }, { "epoch": 23.783150183150184, "grad_norm": 2.203125, "learning_rate": 2.6295986193697643e-05, "loss": 0.1376, "step": 32464 }, { "epoch": 23.783882783882785, "grad_norm": 0.9609375, "learning_rate": 2.6292997434570056e-05, "loss": 0.1372, "step": 32465 }, { "epoch": 23.784615384615385, "grad_norm": 1.6953125, "learning_rate": 2.629000879142615e-05, "loss": 0.1221, "step": 32466 }, { "epoch": 23.785347985347986, "grad_norm": 1.0078125, "learning_rate": 2.628702026428026e-05, "loss": 0.1398, "step": 32467 }, { "epoch": 23.786080586080587, "grad_norm": 0.8046875, "learning_rate": 2.6284031853146708e-05, "loss": 0.1007, "step": 32468 }, { "epoch": 23.786813186813188, "grad_norm": 0.65625, "learning_rate": 2.6281043558039802e-05, "loss": 0.1152, "step": 32469 }, { "epoch": 23.78754578754579, "grad_norm": 0.96875, "learning_rate": 2.6278055378973893e-05, "loss": 0.1223, "step": 32470 }, { "epoch": 23.78827838827839, "grad_norm": 1.5703125, "learning_rate": 2.6275067315963254e-05, "loss": 0.1419, "step": 32471 }, { "epoch": 23.78901098901099, "grad_norm": 0.8515625, "learning_rate": 2.6272079369022214e-05, "loss": 0.1221, "step": 32472 }, { "epoch": 23.78974358974359, "grad_norm": 0.6796875, "learning_rate": 2.6269091538165098e-05, "loss": 0.1037, "step": 32473 }, { "epoch": 23.79047619047619, "grad_norm": 1.3671875, "learning_rate": 2.626610382340623e-05, "loss": 0.1146, "step": 32474 }, { "epoch": 23.791208791208792, "grad_norm": 0.8671875, "learning_rate": 2.6263116224759904e-05, "loss": 0.1087, "step": 32475 }, { "epoch": 23.791941391941393, "grad_norm": 0.734375, "learning_rate": 2.6260128742240463e-05, "loss": 0.1139, "step": 32476 }, { "epoch": 23.792673992673993, "grad_norm": 1.0859375, "learning_rate": 2.6257141375862165e-05, "loss": 0.1184, "step": 32477 }, { "epoch": 23.793406593406594, "grad_norm": 0.6171875, "learning_rate": 2.6254154125639395e-05, "loss": 0.132, "step": 32478 }, { "epoch": 23.794139194139195, "grad_norm": 1.1875, "learning_rate": 2.625116699158642e-05, "loss": 0.1386, "step": 32479 }, { "epoch": 23.794871794871796, "grad_norm": 0.94921875, "learning_rate": 2.6248179973717556e-05, "loss": 0.1048, "step": 32480 }, { "epoch": 23.795604395604396, "grad_norm": 0.87109375, "learning_rate": 2.6245193072047147e-05, "loss": 0.1242, "step": 32481 }, { "epoch": 23.796336996336997, "grad_norm": 1.1328125, "learning_rate": 2.6242206286589465e-05, "loss": 0.1086, "step": 32482 }, { "epoch": 23.797069597069598, "grad_norm": 0.75390625, "learning_rate": 2.6239219617358825e-05, "loss": 0.1248, "step": 32483 }, { "epoch": 23.7978021978022, "grad_norm": 1.171875, "learning_rate": 2.623623306436956e-05, "loss": 0.1294, "step": 32484 }, { "epoch": 23.7985347985348, "grad_norm": 0.7578125, "learning_rate": 2.6233246627635978e-05, "loss": 0.1141, "step": 32485 }, { "epoch": 23.7992673992674, "grad_norm": 0.490234375, "learning_rate": 2.623026030717236e-05, "loss": 0.1101, "step": 32486 }, { "epoch": 23.8, "grad_norm": 0.87109375, "learning_rate": 2.6227274102993038e-05, "loss": 0.1301, "step": 32487 }, { "epoch": 23.8007326007326, "grad_norm": 1.1328125, "learning_rate": 2.6224288015112304e-05, "loss": 0.1441, "step": 32488 }, { "epoch": 23.801465201465202, "grad_norm": 1.1953125, "learning_rate": 2.6221302043544498e-05, "loss": 0.0963, "step": 32489 }, { "epoch": 23.802197802197803, "grad_norm": 1.3515625, "learning_rate": 2.6218316188303894e-05, "loss": 0.1508, "step": 32490 }, { "epoch": 23.802930402930404, "grad_norm": 0.98046875, "learning_rate": 2.6215330449404818e-05, "loss": 0.1272, "step": 32491 }, { "epoch": 23.803663003663004, "grad_norm": 1.0859375, "learning_rate": 2.6212344826861552e-05, "loss": 0.0964, "step": 32492 }, { "epoch": 23.804395604395605, "grad_norm": 1.0703125, "learning_rate": 2.620935932068841e-05, "loss": 0.1413, "step": 32493 }, { "epoch": 23.805128205128206, "grad_norm": 1.734375, "learning_rate": 2.6206373930899702e-05, "loss": 0.1398, "step": 32494 }, { "epoch": 23.805860805860807, "grad_norm": 0.8359375, "learning_rate": 2.6203388657509736e-05, "loss": 0.1106, "step": 32495 }, { "epoch": 23.806593406593407, "grad_norm": 1.046875, "learning_rate": 2.6200403500532828e-05, "loss": 0.1278, "step": 32496 }, { "epoch": 23.807326007326008, "grad_norm": 1.1875, "learning_rate": 2.619741845998324e-05, "loss": 0.1289, "step": 32497 }, { "epoch": 23.80805860805861, "grad_norm": 0.75, "learning_rate": 2.6194433535875307e-05, "loss": 0.1308, "step": 32498 }, { "epoch": 23.80879120879121, "grad_norm": 1.515625, "learning_rate": 2.6191448728223318e-05, "loss": 0.1295, "step": 32499 }, { "epoch": 23.80952380952381, "grad_norm": 1.0078125, "learning_rate": 2.6188464037041594e-05, "loss": 0.1334, "step": 32500 }, { "epoch": 23.81025641025641, "grad_norm": 0.875, "learning_rate": 2.6185479462344404e-05, "loss": 0.1068, "step": 32501 }, { "epoch": 23.81098901098901, "grad_norm": 0.81640625, "learning_rate": 2.6182495004146086e-05, "loss": 0.0779, "step": 32502 }, { "epoch": 23.811721611721612, "grad_norm": 1.1640625, "learning_rate": 2.617951066246087e-05, "loss": 0.1486, "step": 32503 }, { "epoch": 23.812454212454213, "grad_norm": 1.0078125, "learning_rate": 2.617652643730315e-05, "loss": 0.1191, "step": 32504 }, { "epoch": 23.813186813186814, "grad_norm": 1.2421875, "learning_rate": 2.617354232868715e-05, "loss": 0.1207, "step": 32505 }, { "epoch": 23.813919413919415, "grad_norm": 1.140625, "learning_rate": 2.6170558336627205e-05, "loss": 0.1338, "step": 32506 }, { "epoch": 23.814652014652015, "grad_norm": 0.88671875, "learning_rate": 2.616757446113761e-05, "loss": 0.1351, "step": 32507 }, { "epoch": 23.815384615384616, "grad_norm": 0.87890625, "learning_rate": 2.6164590702232644e-05, "loss": 0.1121, "step": 32508 }, { "epoch": 23.816117216117217, "grad_norm": 1.140625, "learning_rate": 2.6161607059926608e-05, "loss": 0.1178, "step": 32509 }, { "epoch": 23.816849816849818, "grad_norm": 1.3828125, "learning_rate": 2.6158623534233794e-05, "loss": 0.1423, "step": 32510 }, { "epoch": 23.81758241758242, "grad_norm": 1.8046875, "learning_rate": 2.6155640125168536e-05, "loss": 0.1278, "step": 32511 }, { "epoch": 23.81831501831502, "grad_norm": 1.2265625, "learning_rate": 2.615265683274507e-05, "loss": 0.1426, "step": 32512 }, { "epoch": 23.81904761904762, "grad_norm": 1.203125, "learning_rate": 2.614967365697773e-05, "loss": 0.1005, "step": 32513 }, { "epoch": 23.81978021978022, "grad_norm": 1.34375, "learning_rate": 2.6146690597880785e-05, "loss": 0.097, "step": 32514 }, { "epoch": 23.82051282051282, "grad_norm": 1.5390625, "learning_rate": 2.6143707655468563e-05, "loss": 0.1588, "step": 32515 }, { "epoch": 23.821245421245422, "grad_norm": 1.1328125, "learning_rate": 2.614072482975532e-05, "loss": 0.1185, "step": 32516 }, { "epoch": 23.821978021978023, "grad_norm": 1.7265625, "learning_rate": 2.6137742120755372e-05, "loss": 0.099, "step": 32517 }, { "epoch": 23.822710622710623, "grad_norm": 0.84375, "learning_rate": 2.6134759528482982e-05, "loss": 0.097, "step": 32518 }, { "epoch": 23.823443223443224, "grad_norm": 1.046875, "learning_rate": 2.613177705295246e-05, "loss": 0.1332, "step": 32519 }, { "epoch": 23.824175824175825, "grad_norm": 1.25, "learning_rate": 2.6128794694178093e-05, "loss": 0.1565, "step": 32520 }, { "epoch": 23.824908424908426, "grad_norm": 0.7890625, "learning_rate": 2.6125812452174176e-05, "loss": 0.1222, "step": 32521 }, { "epoch": 23.825641025641026, "grad_norm": 0.984375, "learning_rate": 2.6122830326955004e-05, "loss": 0.1203, "step": 32522 }, { "epoch": 23.826373626373627, "grad_norm": 0.92578125, "learning_rate": 2.611984831853484e-05, "loss": 0.1129, "step": 32523 }, { "epoch": 23.827106227106228, "grad_norm": 1.2109375, "learning_rate": 2.6116866426927978e-05, "loss": 0.106, "step": 32524 }, { "epoch": 23.82783882783883, "grad_norm": 1.578125, "learning_rate": 2.6113884652148722e-05, "loss": 0.1263, "step": 32525 }, { "epoch": 23.82857142857143, "grad_norm": 0.98046875, "learning_rate": 2.6110902994211366e-05, "loss": 0.1109, "step": 32526 }, { "epoch": 23.82930402930403, "grad_norm": 0.80859375, "learning_rate": 2.6107921453130164e-05, "loss": 0.1251, "step": 32527 }, { "epoch": 23.83003663003663, "grad_norm": 1.5234375, "learning_rate": 2.6104940028919434e-05, "loss": 0.1428, "step": 32528 }, { "epoch": 23.83076923076923, "grad_norm": 1.03125, "learning_rate": 2.6101958721593412e-05, "loss": 0.0901, "step": 32529 }, { "epoch": 23.831501831501832, "grad_norm": 1.1171875, "learning_rate": 2.6098977531166446e-05, "loss": 0.1263, "step": 32530 }, { "epoch": 23.832234432234433, "grad_norm": 1.6015625, "learning_rate": 2.609599645765277e-05, "loss": 0.1195, "step": 32531 }, { "epoch": 23.832967032967034, "grad_norm": 0.8671875, "learning_rate": 2.6093015501066694e-05, "loss": 0.108, "step": 32532 }, { "epoch": 23.833699633699634, "grad_norm": 1.0625, "learning_rate": 2.6090034661422508e-05, "loss": 0.1333, "step": 32533 }, { "epoch": 23.834432234432235, "grad_norm": 1.4140625, "learning_rate": 2.6087053938734456e-05, "loss": 0.1246, "step": 32534 }, { "epoch": 23.835164835164836, "grad_norm": 1.1640625, "learning_rate": 2.6084073333016844e-05, "loss": 0.1348, "step": 32535 }, { "epoch": 23.835897435897436, "grad_norm": 0.69140625, "learning_rate": 2.6081092844283954e-05, "loss": 0.1246, "step": 32536 }, { "epoch": 23.836630036630037, "grad_norm": 1.5390625, "learning_rate": 2.6078112472550086e-05, "loss": 0.1145, "step": 32537 }, { "epoch": 23.837362637362638, "grad_norm": 0.67578125, "learning_rate": 2.607513221782948e-05, "loss": 0.1115, "step": 32538 }, { "epoch": 23.83809523809524, "grad_norm": 1.1640625, "learning_rate": 2.6072152080136425e-05, "loss": 0.1614, "step": 32539 }, { "epoch": 23.83882783882784, "grad_norm": 1.09375, "learning_rate": 2.606917205948522e-05, "loss": 0.112, "step": 32540 }, { "epoch": 23.83956043956044, "grad_norm": 1.4296875, "learning_rate": 2.606619215589014e-05, "loss": 0.1272, "step": 32541 }, { "epoch": 23.84029304029304, "grad_norm": 1.6171875, "learning_rate": 2.6063212369365435e-05, "loss": 0.1117, "step": 32542 }, { "epoch": 23.84102564102564, "grad_norm": 0.94921875, "learning_rate": 2.606023269992543e-05, "loss": 0.1267, "step": 32543 }, { "epoch": 23.841758241758242, "grad_norm": 1.28125, "learning_rate": 2.6057253147584348e-05, "loss": 0.13, "step": 32544 }, { "epoch": 23.842490842490843, "grad_norm": 1.0078125, "learning_rate": 2.6054273712356496e-05, "loss": 0.1161, "step": 32545 }, { "epoch": 23.843223443223444, "grad_norm": 1.125, "learning_rate": 2.6051294394256137e-05, "loss": 0.1138, "step": 32546 }, { "epoch": 23.843956043956045, "grad_norm": 1.0546875, "learning_rate": 2.6048315193297557e-05, "loss": 0.1226, "step": 32547 }, { "epoch": 23.844688644688645, "grad_norm": 0.66015625, "learning_rate": 2.6045336109495046e-05, "loss": 0.1063, "step": 32548 }, { "epoch": 23.845421245421246, "grad_norm": 0.875, "learning_rate": 2.6042357142862832e-05, "loss": 0.134, "step": 32549 }, { "epoch": 23.846153846153847, "grad_norm": 1.0859375, "learning_rate": 2.6039378293415218e-05, "loss": 0.1483, "step": 32550 }, { "epoch": 23.846886446886447, "grad_norm": 1.171875, "learning_rate": 2.6036399561166474e-05, "loss": 0.139, "step": 32551 }, { "epoch": 23.847619047619048, "grad_norm": 0.69921875, "learning_rate": 2.603342094613088e-05, "loss": 0.1252, "step": 32552 }, { "epoch": 23.84835164835165, "grad_norm": 1.3046875, "learning_rate": 2.603044244832269e-05, "loss": 0.1582, "step": 32553 }, { "epoch": 23.84908424908425, "grad_norm": 0.9765625, "learning_rate": 2.60274640677562e-05, "loss": 0.1039, "step": 32554 }, { "epoch": 23.84981684981685, "grad_norm": 0.8828125, "learning_rate": 2.6024485804445632e-05, "loss": 0.1195, "step": 32555 }, { "epoch": 23.85054945054945, "grad_norm": 1.234375, "learning_rate": 2.6021507658405307e-05, "loss": 0.1303, "step": 32556 }, { "epoch": 23.851282051282052, "grad_norm": 0.9375, "learning_rate": 2.6018529629649467e-05, "loss": 0.1173, "step": 32557 }, { "epoch": 23.852014652014653, "grad_norm": 1.2421875, "learning_rate": 2.6015551718192393e-05, "loss": 0.1215, "step": 32558 }, { "epoch": 23.852747252747253, "grad_norm": 0.83984375, "learning_rate": 2.6012573924048356e-05, "loss": 0.112, "step": 32559 }, { "epoch": 23.853479853479854, "grad_norm": 1.09375, "learning_rate": 2.600959624723159e-05, "loss": 0.1228, "step": 32560 }, { "epoch": 23.854212454212455, "grad_norm": 1.578125, "learning_rate": 2.6006618687756416e-05, "loss": 0.1244, "step": 32561 }, { "epoch": 23.854945054945055, "grad_norm": 0.765625, "learning_rate": 2.6003641245637063e-05, "loss": 0.1408, "step": 32562 }, { "epoch": 23.855677655677656, "grad_norm": 0.95703125, "learning_rate": 2.600066392088782e-05, "loss": 0.0983, "step": 32563 }, { "epoch": 23.856410256410257, "grad_norm": 1.328125, "learning_rate": 2.5997686713522918e-05, "loss": 0.1162, "step": 32564 }, { "epoch": 23.857142857142858, "grad_norm": 0.9921875, "learning_rate": 2.599470962355664e-05, "loss": 0.1438, "step": 32565 }, { "epoch": 23.85787545787546, "grad_norm": 1.328125, "learning_rate": 2.599173265100326e-05, "loss": 0.118, "step": 32566 }, { "epoch": 23.85860805860806, "grad_norm": 0.921875, "learning_rate": 2.5988755795877044e-05, "loss": 0.1515, "step": 32567 }, { "epoch": 23.85934065934066, "grad_norm": 0.84765625, "learning_rate": 2.5985779058192238e-05, "loss": 0.109, "step": 32568 }, { "epoch": 23.86007326007326, "grad_norm": 1.6953125, "learning_rate": 2.5982802437963105e-05, "loss": 0.1394, "step": 32569 }, { "epoch": 23.86080586080586, "grad_norm": 1.390625, "learning_rate": 2.5979825935203934e-05, "loss": 0.1268, "step": 32570 }, { "epoch": 23.861538461538462, "grad_norm": 1.703125, "learning_rate": 2.597684954992895e-05, "loss": 0.1506, "step": 32571 }, { "epoch": 23.862271062271063, "grad_norm": 0.7109375, "learning_rate": 2.597387328215243e-05, "loss": 0.092, "step": 32572 }, { "epoch": 23.863003663003663, "grad_norm": 0.81640625, "learning_rate": 2.597089713188863e-05, "loss": 0.117, "step": 32573 }, { "epoch": 23.863736263736264, "grad_norm": 1.015625, "learning_rate": 2.5967921099151837e-05, "loss": 0.1405, "step": 32574 }, { "epoch": 23.864468864468865, "grad_norm": 0.7578125, "learning_rate": 2.596494518395627e-05, "loss": 0.118, "step": 32575 }, { "epoch": 23.865201465201466, "grad_norm": 1.2265625, "learning_rate": 2.5961969386316205e-05, "loss": 0.1809, "step": 32576 }, { "epoch": 23.865934065934066, "grad_norm": 0.6953125, "learning_rate": 2.5958993706245894e-05, "loss": 0.1222, "step": 32577 }, { "epoch": 23.866666666666667, "grad_norm": 1.703125, "learning_rate": 2.595601814375962e-05, "loss": 0.128, "step": 32578 }, { "epoch": 23.867399267399268, "grad_norm": 0.921875, "learning_rate": 2.595304269887161e-05, "loss": 0.1266, "step": 32579 }, { "epoch": 23.86813186813187, "grad_norm": 1.0703125, "learning_rate": 2.5950067371596143e-05, "loss": 0.1189, "step": 32580 }, { "epoch": 23.86886446886447, "grad_norm": 0.98828125, "learning_rate": 2.5947092161947423e-05, "loss": 0.1156, "step": 32581 }, { "epoch": 23.86959706959707, "grad_norm": 1.0546875, "learning_rate": 2.594411706993979e-05, "loss": 0.1187, "step": 32582 }, { "epoch": 23.87032967032967, "grad_norm": 1.0859375, "learning_rate": 2.5941142095587424e-05, "loss": 0.1387, "step": 32583 }, { "epoch": 23.87106227106227, "grad_norm": 1.3046875, "learning_rate": 2.5938167238904614e-05, "loss": 0.1333, "step": 32584 }, { "epoch": 23.871794871794872, "grad_norm": 1.0859375, "learning_rate": 2.5935192499905632e-05, "loss": 0.1269, "step": 32585 }, { "epoch": 23.872527472527473, "grad_norm": 1.390625, "learning_rate": 2.593221787860466e-05, "loss": 0.1222, "step": 32586 }, { "epoch": 23.873260073260074, "grad_norm": 1.3359375, "learning_rate": 2.5929243375016038e-05, "loss": 0.1257, "step": 32587 }, { "epoch": 23.873992673992674, "grad_norm": 0.87109375, "learning_rate": 2.5926268989153958e-05, "loss": 0.13, "step": 32588 }, { "epoch": 23.874725274725275, "grad_norm": 1.046875, "learning_rate": 2.592329472103271e-05, "loss": 0.12, "step": 32589 }, { "epoch": 23.875457875457876, "grad_norm": 0.7734375, "learning_rate": 2.592032057066651e-05, "loss": 0.1114, "step": 32590 }, { "epoch": 23.876190476190477, "grad_norm": 1.671875, "learning_rate": 2.591734653806962e-05, "loss": 0.152, "step": 32591 }, { "epoch": 23.876923076923077, "grad_norm": 1.8125, "learning_rate": 2.591437262325629e-05, "loss": 0.1201, "step": 32592 }, { "epoch": 23.877655677655678, "grad_norm": 1.6640625, "learning_rate": 2.591139882624079e-05, "loss": 0.1385, "step": 32593 }, { "epoch": 23.87838827838828, "grad_norm": 1.6875, "learning_rate": 2.5908425147037336e-05, "loss": 0.1619, "step": 32594 }, { "epoch": 23.87912087912088, "grad_norm": 1.234375, "learning_rate": 2.5905451585660183e-05, "loss": 0.1342, "step": 32595 }, { "epoch": 23.87985347985348, "grad_norm": 0.61328125, "learning_rate": 2.5902478142123615e-05, "loss": 0.1055, "step": 32596 }, { "epoch": 23.88058608058608, "grad_norm": 1.078125, "learning_rate": 2.5899504816441816e-05, "loss": 0.1503, "step": 32597 }, { "epoch": 23.88131868131868, "grad_norm": 0.80078125, "learning_rate": 2.5896531608629083e-05, "loss": 0.1176, "step": 32598 }, { "epoch": 23.882051282051282, "grad_norm": 0.828125, "learning_rate": 2.589355851869963e-05, "loss": 0.1127, "step": 32599 }, { "epoch": 23.882783882783883, "grad_norm": 0.8828125, "learning_rate": 2.589058554666774e-05, "loss": 0.1325, "step": 32600 }, { "epoch": 23.883516483516484, "grad_norm": 0.984375, "learning_rate": 2.5887612692547614e-05, "loss": 0.1308, "step": 32601 }, { "epoch": 23.884249084249085, "grad_norm": 1.5390625, "learning_rate": 2.5884639956353513e-05, "loss": 0.1278, "step": 32602 }, { "epoch": 23.884981684981685, "grad_norm": 0.9921875, "learning_rate": 2.5881667338099685e-05, "loss": 0.1014, "step": 32603 }, { "epoch": 23.885714285714286, "grad_norm": 1.2734375, "learning_rate": 2.5878694837800387e-05, "loss": 0.1186, "step": 32604 }, { "epoch": 23.886446886446887, "grad_norm": 1.1953125, "learning_rate": 2.587572245546983e-05, "loss": 0.1276, "step": 32605 }, { "epoch": 23.887179487179488, "grad_norm": 1.046875, "learning_rate": 2.5872750191122285e-05, "loss": 0.1256, "step": 32606 }, { "epoch": 23.88791208791209, "grad_norm": 1.484375, "learning_rate": 2.5869778044771943e-05, "loss": 0.1288, "step": 32607 }, { "epoch": 23.88864468864469, "grad_norm": 0.9453125, "learning_rate": 2.586680601643312e-05, "loss": 0.1101, "step": 32608 }, { "epoch": 23.88937728937729, "grad_norm": 1.1796875, "learning_rate": 2.586383410611999e-05, "loss": 0.1069, "step": 32609 }, { "epoch": 23.89010989010989, "grad_norm": 1.4140625, "learning_rate": 2.5860862313846823e-05, "loss": 0.1388, "step": 32610 }, { "epoch": 23.89084249084249, "grad_norm": 0.8984375, "learning_rate": 2.5857890639627875e-05, "loss": 0.1085, "step": 32611 }, { "epoch": 23.891575091575092, "grad_norm": 1.3125, "learning_rate": 2.5854919083477314e-05, "loss": 0.1321, "step": 32612 }, { "epoch": 23.892307692307693, "grad_norm": 1.2890625, "learning_rate": 2.585194764540948e-05, "loss": 0.1147, "step": 32613 }, { "epoch": 23.893040293040293, "grad_norm": 0.8359375, "learning_rate": 2.5848976325438522e-05, "loss": 0.1171, "step": 32614 }, { "epoch": 23.893772893772894, "grad_norm": 1.3046875, "learning_rate": 2.5846005123578742e-05, "loss": 0.1289, "step": 32615 }, { "epoch": 23.894505494505495, "grad_norm": 1.0, "learning_rate": 2.5843034039844317e-05, "loss": 0.1456, "step": 32616 }, { "epoch": 23.895238095238096, "grad_norm": 1.40625, "learning_rate": 2.5840063074249515e-05, "loss": 0.1147, "step": 32617 }, { "epoch": 23.895970695970696, "grad_norm": 1.046875, "learning_rate": 2.5837092226808564e-05, "loss": 0.1262, "step": 32618 }, { "epoch": 23.896703296703297, "grad_norm": 1.1171875, "learning_rate": 2.583412149753572e-05, "loss": 0.1206, "step": 32619 }, { "epoch": 23.897435897435898, "grad_norm": 1.0703125, "learning_rate": 2.5831150886445178e-05, "loss": 0.1224, "step": 32620 }, { "epoch": 23.8981684981685, "grad_norm": 1.1640625, "learning_rate": 2.582818039355119e-05, "loss": 0.1181, "step": 32621 }, { "epoch": 23.8989010989011, "grad_norm": 1.6328125, "learning_rate": 2.5825210018868007e-05, "loss": 0.1506, "step": 32622 }, { "epoch": 23.8996336996337, "grad_norm": 0.890625, "learning_rate": 2.5822239762409827e-05, "loss": 0.1349, "step": 32623 }, { "epoch": 23.9003663003663, "grad_norm": 1.2578125, "learning_rate": 2.5819269624190895e-05, "loss": 0.127, "step": 32624 }, { "epoch": 23.9010989010989, "grad_norm": 1.46875, "learning_rate": 2.581629960422544e-05, "loss": 0.1339, "step": 32625 }, { "epoch": 23.901831501831502, "grad_norm": 1.0234375, "learning_rate": 2.581332970252773e-05, "loss": 0.0908, "step": 32626 }, { "epoch": 23.902564102564103, "grad_norm": 1.984375, "learning_rate": 2.5810359919111925e-05, "loss": 0.1534, "step": 32627 }, { "epoch": 23.903296703296704, "grad_norm": 1.09375, "learning_rate": 2.5807390253992308e-05, "loss": 0.1052, "step": 32628 }, { "epoch": 23.904029304029304, "grad_norm": 1.328125, "learning_rate": 2.580442070718308e-05, "loss": 0.1486, "step": 32629 }, { "epoch": 23.904761904761905, "grad_norm": 0.74609375, "learning_rate": 2.5801451278698498e-05, "loss": 0.1216, "step": 32630 }, { "epoch": 23.905494505494506, "grad_norm": 0.6796875, "learning_rate": 2.5798481968552758e-05, "loss": 0.136, "step": 32631 }, { "epoch": 23.906227106227107, "grad_norm": 1.1328125, "learning_rate": 2.5795512776760112e-05, "loss": 0.1104, "step": 32632 }, { "epoch": 23.906959706959707, "grad_norm": 1.171875, "learning_rate": 2.5792543703334738e-05, "loss": 0.0982, "step": 32633 }, { "epoch": 23.907692307692308, "grad_norm": 1.34375, "learning_rate": 2.578957474829094e-05, "loss": 0.1332, "step": 32634 }, { "epoch": 23.90842490842491, "grad_norm": 1.1953125, "learning_rate": 2.578660591164288e-05, "loss": 0.1303, "step": 32635 }, { "epoch": 23.90915750915751, "grad_norm": 0.90234375, "learning_rate": 2.5783637193404804e-05, "loss": 0.1552, "step": 32636 }, { "epoch": 23.90989010989011, "grad_norm": 0.9296875, "learning_rate": 2.578066859359095e-05, "loss": 0.1145, "step": 32637 }, { "epoch": 23.91062271062271, "grad_norm": 1.9375, "learning_rate": 2.5777700112215493e-05, "loss": 0.1277, "step": 32638 }, { "epoch": 23.91135531135531, "grad_norm": 1.0546875, "learning_rate": 2.577473174929273e-05, "loss": 0.1263, "step": 32639 }, { "epoch": 23.912087912087912, "grad_norm": 1.6171875, "learning_rate": 2.5771763504836822e-05, "loss": 0.1345, "step": 32640 }, { "epoch": 23.912820512820513, "grad_norm": 0.953125, "learning_rate": 2.5768795378862038e-05, "loss": 0.1326, "step": 32641 }, { "epoch": 23.913553113553114, "grad_norm": 1.5, "learning_rate": 2.576582737138255e-05, "loss": 0.1241, "step": 32642 }, { "epoch": 23.914285714285715, "grad_norm": 1.4296875, "learning_rate": 2.5762859482412604e-05, "loss": 0.1488, "step": 32643 }, { "epoch": 23.915018315018315, "grad_norm": 1.8203125, "learning_rate": 2.5759891711966418e-05, "loss": 0.1263, "step": 32644 }, { "epoch": 23.915750915750916, "grad_norm": 1.1171875, "learning_rate": 2.5756924060058234e-05, "loss": 0.1237, "step": 32645 }, { "epoch": 23.916483516483517, "grad_norm": 1.09375, "learning_rate": 2.5753956526702228e-05, "loss": 0.1034, "step": 32646 }, { "epoch": 23.917216117216118, "grad_norm": 0.953125, "learning_rate": 2.5750989111912644e-05, "loss": 0.1261, "step": 32647 }, { "epoch": 23.91794871794872, "grad_norm": 1.9765625, "learning_rate": 2.574802181570371e-05, "loss": 0.1335, "step": 32648 }, { "epoch": 23.91868131868132, "grad_norm": 1.1171875, "learning_rate": 2.5745054638089617e-05, "loss": 0.144, "step": 32649 }, { "epoch": 23.91941391941392, "grad_norm": 1.5, "learning_rate": 2.5742087579084584e-05, "loss": 0.1151, "step": 32650 }, { "epoch": 23.92014652014652, "grad_norm": 1.3203125, "learning_rate": 2.573912063870284e-05, "loss": 0.1497, "step": 32651 }, { "epoch": 23.92087912087912, "grad_norm": 1.1953125, "learning_rate": 2.573615381695862e-05, "loss": 0.1403, "step": 32652 }, { "epoch": 23.921611721611722, "grad_norm": 0.8125, "learning_rate": 2.573318711386609e-05, "loss": 0.1189, "step": 32653 }, { "epoch": 23.922344322344323, "grad_norm": 0.87109375, "learning_rate": 2.5730220529439497e-05, "loss": 0.1202, "step": 32654 }, { "epoch": 23.923076923076923, "grad_norm": 1.0859375, "learning_rate": 2.572725406369305e-05, "loss": 0.1223, "step": 32655 }, { "epoch": 23.923809523809524, "grad_norm": 0.796875, "learning_rate": 2.572428771664098e-05, "loss": 0.1403, "step": 32656 }, { "epoch": 23.924542124542125, "grad_norm": 1.3203125, "learning_rate": 2.5721321488297463e-05, "loss": 0.116, "step": 32657 }, { "epoch": 23.925274725274726, "grad_norm": 0.71875, "learning_rate": 2.5718355378676746e-05, "loss": 0.11, "step": 32658 }, { "epoch": 23.926007326007326, "grad_norm": 1.3359375, "learning_rate": 2.5715389387792975e-05, "loss": 0.1036, "step": 32659 }, { "epoch": 23.926739926739927, "grad_norm": 0.87890625, "learning_rate": 2.5712423515660462e-05, "loss": 0.1402, "step": 32660 }, { "epoch": 23.927472527472528, "grad_norm": 1.140625, "learning_rate": 2.5709457762293343e-05, "loss": 0.1248, "step": 32661 }, { "epoch": 23.92820512820513, "grad_norm": 0.71484375, "learning_rate": 2.5706492127705843e-05, "loss": 0.1409, "step": 32662 }, { "epoch": 23.92893772893773, "grad_norm": 1.2109375, "learning_rate": 2.570352661191221e-05, "loss": 0.1225, "step": 32663 }, { "epoch": 23.92967032967033, "grad_norm": 0.94140625, "learning_rate": 2.5700561214926573e-05, "loss": 0.113, "step": 32664 }, { "epoch": 23.93040293040293, "grad_norm": 0.91796875, "learning_rate": 2.569759593676323e-05, "loss": 0.1128, "step": 32665 }, { "epoch": 23.93113553113553, "grad_norm": 1.6484375, "learning_rate": 2.569463077743633e-05, "loss": 0.1375, "step": 32666 }, { "epoch": 23.931868131868132, "grad_norm": 1.3359375, "learning_rate": 2.569166573696011e-05, "loss": 0.1077, "step": 32667 }, { "epoch": 23.932600732600733, "grad_norm": 0.97265625, "learning_rate": 2.568870081534875e-05, "loss": 0.1191, "step": 32668 }, { "epoch": 23.933333333333334, "grad_norm": 1.5546875, "learning_rate": 2.568573601261649e-05, "loss": 0.1502, "step": 32669 }, { "epoch": 23.934065934065934, "grad_norm": 1.3515625, "learning_rate": 2.5682771328777466e-05, "loss": 0.1627, "step": 32670 }, { "epoch": 23.934798534798535, "grad_norm": 0.7109375, "learning_rate": 2.567980676384598e-05, "loss": 0.1124, "step": 32671 }, { "epoch": 23.935531135531136, "grad_norm": 1.109375, "learning_rate": 2.5676842317836163e-05, "loss": 0.1064, "step": 32672 }, { "epoch": 23.936263736263736, "grad_norm": 1.3125, "learning_rate": 2.5673877990762247e-05, "loss": 0.1114, "step": 32673 }, { "epoch": 23.936996336996337, "grad_norm": 1.171875, "learning_rate": 2.567091378263845e-05, "loss": 0.1589, "step": 32674 }, { "epoch": 23.937728937728938, "grad_norm": 1.3828125, "learning_rate": 2.566794969347892e-05, "loss": 0.108, "step": 32675 }, { "epoch": 23.93846153846154, "grad_norm": 0.9296875, "learning_rate": 2.566498572329793e-05, "loss": 0.1466, "step": 32676 }, { "epoch": 23.93919413919414, "grad_norm": 0.984375, "learning_rate": 2.5662021872109625e-05, "loss": 0.0992, "step": 32677 }, { "epoch": 23.93992673992674, "grad_norm": 0.83203125, "learning_rate": 2.565905813992825e-05, "loss": 0.1259, "step": 32678 }, { "epoch": 23.94065934065934, "grad_norm": 0.54296875, "learning_rate": 2.565609452676796e-05, "loss": 0.0959, "step": 32679 }, { "epoch": 23.94139194139194, "grad_norm": 1.2265625, "learning_rate": 2.5653131032642974e-05, "loss": 0.1147, "step": 32680 }, { "epoch": 23.942124542124542, "grad_norm": 1.5859375, "learning_rate": 2.5650167657567496e-05, "loss": 0.1135, "step": 32681 }, { "epoch": 23.942857142857143, "grad_norm": 1.3984375, "learning_rate": 2.5647204401555747e-05, "loss": 0.1295, "step": 32682 }, { "epoch": 23.943589743589744, "grad_norm": 0.88671875, "learning_rate": 2.5644241264621878e-05, "loss": 0.0952, "step": 32683 }, { "epoch": 23.944322344322345, "grad_norm": 0.6171875, "learning_rate": 2.564127824678011e-05, "loss": 0.0776, "step": 32684 }, { "epoch": 23.945054945054945, "grad_norm": 0.86328125, "learning_rate": 2.5638315348044638e-05, "loss": 0.1085, "step": 32685 }, { "epoch": 23.945787545787546, "grad_norm": 1.140625, "learning_rate": 2.5635352568429676e-05, "loss": 0.117, "step": 32686 }, { "epoch": 23.946520146520147, "grad_norm": 1.28125, "learning_rate": 2.563238990794939e-05, "loss": 0.1159, "step": 32687 }, { "epoch": 23.947252747252747, "grad_norm": 0.70703125, "learning_rate": 2.5629427366617977e-05, "loss": 0.1114, "step": 32688 }, { "epoch": 23.947985347985348, "grad_norm": 1.0859375, "learning_rate": 2.562646494444967e-05, "loss": 0.1273, "step": 32689 }, { "epoch": 23.94871794871795, "grad_norm": 0.7890625, "learning_rate": 2.562350264145859e-05, "loss": 0.1055, "step": 32690 }, { "epoch": 23.94945054945055, "grad_norm": 0.8984375, "learning_rate": 2.5620540457659016e-05, "loss": 0.1001, "step": 32691 }, { "epoch": 23.95018315018315, "grad_norm": 1.8828125, "learning_rate": 2.5617578393065083e-05, "loss": 0.1407, "step": 32692 }, { "epoch": 23.95091575091575, "grad_norm": 1.9296875, "learning_rate": 2.5614616447691008e-05, "loss": 0.1166, "step": 32693 }, { "epoch": 23.951648351648352, "grad_norm": 0.921875, "learning_rate": 2.5611654621550972e-05, "loss": 0.098, "step": 32694 }, { "epoch": 23.952380952380953, "grad_norm": 0.734375, "learning_rate": 2.5608692914659183e-05, "loss": 0.0972, "step": 32695 }, { "epoch": 23.953113553113553, "grad_norm": 0.74609375, "learning_rate": 2.5605731327029774e-05, "loss": 0.0972, "step": 32696 }, { "epoch": 23.953846153846154, "grad_norm": 0.91796875, "learning_rate": 2.560276985867702e-05, "loss": 0.0938, "step": 32697 }, { "epoch": 23.954578754578755, "grad_norm": 1.28125, "learning_rate": 2.5599808509615052e-05, "loss": 0.1205, "step": 32698 }, { "epoch": 23.955311355311355, "grad_norm": 0.67578125, "learning_rate": 2.559684727985807e-05, "loss": 0.0969, "step": 32699 }, { "epoch": 23.956043956043956, "grad_norm": 0.7265625, "learning_rate": 2.559388616942029e-05, "loss": 0.098, "step": 32700 }, { "epoch": 23.956776556776557, "grad_norm": 0.77734375, "learning_rate": 2.559092517831584e-05, "loss": 0.1033, "step": 32701 }, { "epoch": 23.957509157509158, "grad_norm": 1.1171875, "learning_rate": 2.5587964306558982e-05, "loss": 0.1313, "step": 32702 }, { "epoch": 23.95824175824176, "grad_norm": 1.25, "learning_rate": 2.558500355416385e-05, "loss": 0.1315, "step": 32703 }, { "epoch": 23.95897435897436, "grad_norm": 1.6484375, "learning_rate": 2.5582042921144662e-05, "loss": 0.1044, "step": 32704 }, { "epoch": 23.95970695970696, "grad_norm": 0.9921875, "learning_rate": 2.5579082407515565e-05, "loss": 0.1049, "step": 32705 }, { "epoch": 23.96043956043956, "grad_norm": 1.078125, "learning_rate": 2.5576122013290764e-05, "loss": 0.1288, "step": 32706 }, { "epoch": 23.96117216117216, "grad_norm": 0.81640625, "learning_rate": 2.5573161738484446e-05, "loss": 0.1213, "step": 32707 }, { "epoch": 23.961904761904762, "grad_norm": 1.0234375, "learning_rate": 2.5570201583110814e-05, "loss": 0.1214, "step": 32708 }, { "epoch": 23.962637362637363, "grad_norm": 1.515625, "learning_rate": 2.5567241547184013e-05, "loss": 0.1102, "step": 32709 }, { "epoch": 23.963369963369964, "grad_norm": 1.3671875, "learning_rate": 2.556428163071824e-05, "loss": 0.1076, "step": 32710 }, { "epoch": 23.964102564102564, "grad_norm": 1.734375, "learning_rate": 2.5561321833727675e-05, "loss": 0.1033, "step": 32711 }, { "epoch": 23.964835164835165, "grad_norm": 1.0234375, "learning_rate": 2.5558362156226528e-05, "loss": 0.1185, "step": 32712 }, { "epoch": 23.965567765567766, "grad_norm": 1.078125, "learning_rate": 2.5555402598228933e-05, "loss": 0.1188, "step": 32713 }, { "epoch": 23.966300366300366, "grad_norm": 1.328125, "learning_rate": 2.55524431597491e-05, "loss": 0.1157, "step": 32714 }, { "epoch": 23.967032967032967, "grad_norm": 1.296875, "learning_rate": 2.5549483840801214e-05, "loss": 0.132, "step": 32715 }, { "epoch": 23.967765567765568, "grad_norm": 0.95703125, "learning_rate": 2.5546524641399405e-05, "loss": 0.0888, "step": 32716 }, { "epoch": 23.96849816849817, "grad_norm": 0.9375, "learning_rate": 2.554356556155793e-05, "loss": 0.0988, "step": 32717 }, { "epoch": 23.96923076923077, "grad_norm": 1.171875, "learning_rate": 2.5540606601290905e-05, "loss": 0.144, "step": 32718 }, { "epoch": 23.96996336996337, "grad_norm": 1.1796875, "learning_rate": 2.5537647760612543e-05, "loss": 0.1312, "step": 32719 }, { "epoch": 23.97069597069597, "grad_norm": 1.2421875, "learning_rate": 2.5534689039536985e-05, "loss": 0.1179, "step": 32720 }, { "epoch": 23.97142857142857, "grad_norm": 1.09375, "learning_rate": 2.553173043807845e-05, "loss": 0.1177, "step": 32721 }, { "epoch": 23.972161172161172, "grad_norm": 0.9453125, "learning_rate": 2.5528771956251052e-05, "loss": 0.1322, "step": 32722 }, { "epoch": 23.972893772893773, "grad_norm": 1.265625, "learning_rate": 2.5525813594069042e-05, "loss": 0.1421, "step": 32723 }, { "epoch": 23.973626373626374, "grad_norm": 0.9296875, "learning_rate": 2.5522855351546543e-05, "loss": 0.1283, "step": 32724 }, { "epoch": 23.974358974358974, "grad_norm": 1.7421875, "learning_rate": 2.5519897228697738e-05, "loss": 0.1291, "step": 32725 }, { "epoch": 23.975091575091575, "grad_norm": 1.4140625, "learning_rate": 2.5516939225536823e-05, "loss": 0.1275, "step": 32726 }, { "epoch": 23.975824175824176, "grad_norm": 0.953125, "learning_rate": 2.551398134207792e-05, "loss": 0.103, "step": 32727 }, { "epoch": 23.976556776556777, "grad_norm": 1.2890625, "learning_rate": 2.5511023578335275e-05, "loss": 0.1155, "step": 32728 }, { "epoch": 23.977289377289377, "grad_norm": 0.75390625, "learning_rate": 2.5508065934322994e-05, "loss": 0.1139, "step": 32729 }, { "epoch": 23.978021978021978, "grad_norm": 1.8203125, "learning_rate": 2.55051084100553e-05, "loss": 0.1124, "step": 32730 }, { "epoch": 23.97875457875458, "grad_norm": 0.9921875, "learning_rate": 2.5502151005546315e-05, "loss": 0.1419, "step": 32731 }, { "epoch": 23.97948717948718, "grad_norm": 1.0703125, "learning_rate": 2.5499193720810227e-05, "loss": 0.1181, "step": 32732 }, { "epoch": 23.98021978021978, "grad_norm": 0.75, "learning_rate": 2.5496236555861208e-05, "loss": 0.1277, "step": 32733 }, { "epoch": 23.98095238095238, "grad_norm": 1.171875, "learning_rate": 2.549327951071346e-05, "loss": 0.124, "step": 32734 }, { "epoch": 23.98168498168498, "grad_norm": 0.66796875, "learning_rate": 2.5490322585381088e-05, "loss": 0.0993, "step": 32735 }, { "epoch": 23.982417582417582, "grad_norm": 1.15625, "learning_rate": 2.548736577987829e-05, "loss": 0.1052, "step": 32736 }, { "epoch": 23.983150183150183, "grad_norm": 2.359375, "learning_rate": 2.5484409094219243e-05, "loss": 0.1643, "step": 32737 }, { "epoch": 23.983882783882784, "grad_norm": 1.7734375, "learning_rate": 2.5481452528418115e-05, "loss": 0.132, "step": 32738 }, { "epoch": 23.984615384615385, "grad_norm": 1.0859375, "learning_rate": 2.5478496082489045e-05, "loss": 0.1102, "step": 32739 }, { "epoch": 23.985347985347985, "grad_norm": 1.0546875, "learning_rate": 2.547553975644621e-05, "loss": 0.1292, "step": 32740 }, { "epoch": 23.986080586080586, "grad_norm": 1.40625, "learning_rate": 2.54725835503038e-05, "loss": 0.1268, "step": 32741 }, { "epoch": 23.986813186813187, "grad_norm": 1.125, "learning_rate": 2.5469627464075923e-05, "loss": 0.1346, "step": 32742 }, { "epoch": 23.987545787545788, "grad_norm": 0.81640625, "learning_rate": 2.5466671497776813e-05, "loss": 0.115, "step": 32743 }, { "epoch": 23.98827838827839, "grad_norm": 1.0546875, "learning_rate": 2.5463715651420584e-05, "loss": 0.1254, "step": 32744 }, { "epoch": 23.98901098901099, "grad_norm": 1.03125, "learning_rate": 2.5460759925021422e-05, "loss": 0.1347, "step": 32745 }, { "epoch": 23.98974358974359, "grad_norm": 1.0859375, "learning_rate": 2.5457804318593465e-05, "loss": 0.1118, "step": 32746 }, { "epoch": 23.99047619047619, "grad_norm": 0.7578125, "learning_rate": 2.5454848832150914e-05, "loss": 0.1011, "step": 32747 }, { "epoch": 23.99120879120879, "grad_norm": 0.98046875, "learning_rate": 2.5451893465707856e-05, "loss": 0.1144, "step": 32748 }, { "epoch": 23.991941391941392, "grad_norm": 1.609375, "learning_rate": 2.5448938219278544e-05, "loss": 0.114, "step": 32749 }, { "epoch": 23.992673992673993, "grad_norm": 0.9609375, "learning_rate": 2.5445983092877068e-05, "loss": 0.124, "step": 32750 }, { "epoch": 23.993406593406593, "grad_norm": 1.5390625, "learning_rate": 2.5443028086517612e-05, "loss": 0.127, "step": 32751 }, { "epoch": 23.994139194139194, "grad_norm": 1.21875, "learning_rate": 2.5440073200214353e-05, "loss": 0.1443, "step": 32752 }, { "epoch": 23.994871794871795, "grad_norm": 1.21875, "learning_rate": 2.54371184339814e-05, "loss": 0.1277, "step": 32753 }, { "epoch": 23.995604395604396, "grad_norm": 0.90234375, "learning_rate": 2.543416378783297e-05, "loss": 0.1272, "step": 32754 }, { "epoch": 23.996336996336996, "grad_norm": 0.74609375, "learning_rate": 2.5431209261783172e-05, "loss": 0.1343, "step": 32755 }, { "epoch": 23.997069597069597, "grad_norm": 1.15625, "learning_rate": 2.5428254855846202e-05, "loss": 0.1224, "step": 32756 }, { "epoch": 23.997802197802198, "grad_norm": 1.1171875, "learning_rate": 2.5425300570036177e-05, "loss": 0.128, "step": 32757 }, { "epoch": 23.9985347985348, "grad_norm": 1.171875, "learning_rate": 2.5422346404367266e-05, "loss": 0.1178, "step": 32758 }, { "epoch": 23.9992673992674, "grad_norm": 1.2265625, "learning_rate": 2.5419392358853617e-05, "loss": 0.1675, "step": 32759 }, { "epoch": 24.0, "grad_norm": 0.984375, "learning_rate": 2.541643843350942e-05, "loss": 0.1116, "step": 32760 }, { "epoch": 24.0007326007326, "grad_norm": 1.03125, "learning_rate": 2.5413484628348782e-05, "loss": 0.1134, "step": 32761 }, { "epoch": 24.0014652014652, "grad_norm": 1.0859375, "learning_rate": 2.5410530943385875e-05, "loss": 0.1272, "step": 32762 }, { "epoch": 24.002197802197802, "grad_norm": 0.84765625, "learning_rate": 2.5407577378634845e-05, "loss": 0.1103, "step": 32763 }, { "epoch": 24.002930402930403, "grad_norm": 0.828125, "learning_rate": 2.5404623934109872e-05, "loss": 0.1149, "step": 32764 }, { "epoch": 24.003663003663004, "grad_norm": 1.0, "learning_rate": 2.540167060982507e-05, "loss": 0.1586, "step": 32765 }, { "epoch": 24.004395604395604, "grad_norm": 0.66796875, "learning_rate": 2.539871740579459e-05, "loss": 0.1121, "step": 32766 }, { "epoch": 24.005128205128205, "grad_norm": 0.86328125, "learning_rate": 2.539576432203263e-05, "loss": 0.1028, "step": 32767 }, { "epoch": 24.005860805860806, "grad_norm": 1.0625, "learning_rate": 2.539281135855326e-05, "loss": 0.1238, "step": 32768 }, { "epoch": 24.006593406593407, "grad_norm": 1.6171875, "learning_rate": 2.5389858515370702e-05, "loss": 0.1239, "step": 32769 }, { "epoch": 24.007326007326007, "grad_norm": 1.15625, "learning_rate": 2.5386905792499065e-05, "loss": 0.1368, "step": 32770 }, { "epoch": 24.008058608058608, "grad_norm": 1.3828125, "learning_rate": 2.5383953189952527e-05, "loss": 0.1217, "step": 32771 }, { "epoch": 24.00879120879121, "grad_norm": 0.9921875, "learning_rate": 2.538100070774519e-05, "loss": 0.1356, "step": 32772 }, { "epoch": 24.00952380952381, "grad_norm": 1.46875, "learning_rate": 2.537804834589125e-05, "loss": 0.1031, "step": 32773 }, { "epoch": 24.01025641025641, "grad_norm": 1.703125, "learning_rate": 2.537509610440478e-05, "loss": 0.1261, "step": 32774 }, { "epoch": 24.01098901098901, "grad_norm": 1.1484375, "learning_rate": 2.5372143983300018e-05, "loss": 0.1599, "step": 32775 }, { "epoch": 24.01172161172161, "grad_norm": 1.125, "learning_rate": 2.536919198259104e-05, "loss": 0.1447, "step": 32776 }, { "epoch": 24.012454212454212, "grad_norm": 1.2421875, "learning_rate": 2.5366240102292016e-05, "loss": 0.1169, "step": 32777 }, { "epoch": 24.013186813186813, "grad_norm": 0.91015625, "learning_rate": 2.5363288342417103e-05, "loss": 0.1096, "step": 32778 }, { "epoch": 24.013919413919414, "grad_norm": 0.65625, "learning_rate": 2.5360336702980392e-05, "loss": 0.1046, "step": 32779 }, { "epoch": 24.014652014652015, "grad_norm": 1.4140625, "learning_rate": 2.5357385183996103e-05, "loss": 0.1439, "step": 32780 }, { "epoch": 24.015384615384615, "grad_norm": 1.609375, "learning_rate": 2.5354433785478303e-05, "loss": 0.1419, "step": 32781 }, { "epoch": 24.016117216117216, "grad_norm": 0.6875, "learning_rate": 2.5351482507441193e-05, "loss": 0.1083, "step": 32782 }, { "epoch": 24.016849816849817, "grad_norm": 0.9765625, "learning_rate": 2.534853134989887e-05, "loss": 0.1062, "step": 32783 }, { "epoch": 24.017582417582418, "grad_norm": 0.76953125, "learning_rate": 2.5345580312865485e-05, "loss": 0.1228, "step": 32784 }, { "epoch": 24.01831501831502, "grad_norm": 1.0859375, "learning_rate": 2.534262939635519e-05, "loss": 0.1025, "step": 32785 }, { "epoch": 24.01904761904762, "grad_norm": 1.0703125, "learning_rate": 2.5339678600382127e-05, "loss": 0.1072, "step": 32786 }, { "epoch": 24.01978021978022, "grad_norm": 0.7109375, "learning_rate": 2.5336727924960403e-05, "loss": 0.1278, "step": 32787 }, { "epoch": 24.02051282051282, "grad_norm": 1.3828125, "learning_rate": 2.5333777370104177e-05, "loss": 0.1442, "step": 32788 }, { "epoch": 24.02124542124542, "grad_norm": 1.0625, "learning_rate": 2.533082693582759e-05, "loss": 0.1561, "step": 32789 }, { "epoch": 24.021978021978022, "grad_norm": 0.6640625, "learning_rate": 2.5327876622144785e-05, "loss": 0.1061, "step": 32790 }, { "epoch": 24.022710622710623, "grad_norm": 1.0, "learning_rate": 2.532492642906988e-05, "loss": 0.1181, "step": 32791 }, { "epoch": 24.023443223443223, "grad_norm": 1.90625, "learning_rate": 2.5321976356617005e-05, "loss": 0.132, "step": 32792 }, { "epoch": 24.024175824175824, "grad_norm": 1.34375, "learning_rate": 2.5319026404800332e-05, "loss": 0.094, "step": 32793 }, { "epoch": 24.024908424908425, "grad_norm": 0.69921875, "learning_rate": 2.5316076573633935e-05, "loss": 0.1105, "step": 32794 }, { "epoch": 24.025641025641026, "grad_norm": 0.94140625, "learning_rate": 2.5313126863132012e-05, "loss": 0.1066, "step": 32795 }, { "epoch": 24.026373626373626, "grad_norm": 1.28125, "learning_rate": 2.531017727330865e-05, "loss": 0.1465, "step": 32796 }, { "epoch": 24.027106227106227, "grad_norm": 0.984375, "learning_rate": 2.5307227804178018e-05, "loss": 0.1226, "step": 32797 }, { "epoch": 24.027838827838828, "grad_norm": 0.73046875, "learning_rate": 2.530427845575421e-05, "loss": 0.1221, "step": 32798 }, { "epoch": 24.02857142857143, "grad_norm": 0.86328125, "learning_rate": 2.530132922805138e-05, "loss": 0.109, "step": 32799 }, { "epoch": 24.02930402930403, "grad_norm": 0.953125, "learning_rate": 2.529838012108364e-05, "loss": 0.1142, "step": 32800 }, { "epoch": 24.03003663003663, "grad_norm": 0.93359375, "learning_rate": 2.5295431134865165e-05, "loss": 0.1103, "step": 32801 }, { "epoch": 24.03076923076923, "grad_norm": 1.3984375, "learning_rate": 2.5292482269410033e-05, "loss": 0.1113, "step": 32802 }, { "epoch": 24.03150183150183, "grad_norm": 0.984375, "learning_rate": 2.528953352473239e-05, "loss": 0.1157, "step": 32803 }, { "epoch": 24.032234432234432, "grad_norm": 1.21875, "learning_rate": 2.528658490084639e-05, "loss": 0.1021, "step": 32804 }, { "epoch": 24.032967032967033, "grad_norm": 1.015625, "learning_rate": 2.5283636397766098e-05, "loss": 0.1083, "step": 32805 }, { "epoch": 24.033699633699634, "grad_norm": 0.84765625, "learning_rate": 2.528068801550572e-05, "loss": 0.1026, "step": 32806 }, { "epoch": 24.034432234432234, "grad_norm": 1.0859375, "learning_rate": 2.5277739754079328e-05, "loss": 0.1276, "step": 32807 }, { "epoch": 24.035164835164835, "grad_norm": 0.765625, "learning_rate": 2.527479161350108e-05, "loss": 0.1039, "step": 32808 }, { "epoch": 24.035897435897436, "grad_norm": 1.0234375, "learning_rate": 2.527184359378507e-05, "loss": 0.1167, "step": 32809 }, { "epoch": 24.036630036630036, "grad_norm": 1.4296875, "learning_rate": 2.5268895694945438e-05, "loss": 0.1075, "step": 32810 }, { "epoch": 24.037362637362637, "grad_norm": 1.0234375, "learning_rate": 2.5265947916996314e-05, "loss": 0.1199, "step": 32811 }, { "epoch": 24.038095238095238, "grad_norm": 0.6875, "learning_rate": 2.526300025995183e-05, "loss": 0.1179, "step": 32812 }, { "epoch": 24.03882783882784, "grad_norm": 0.9765625, "learning_rate": 2.5260052723826077e-05, "loss": 0.1095, "step": 32813 }, { "epoch": 24.03956043956044, "grad_norm": 0.96875, "learning_rate": 2.52571053086332e-05, "loss": 0.136, "step": 32814 }, { "epoch": 24.04029304029304, "grad_norm": 0.87890625, "learning_rate": 2.5254158014387324e-05, "loss": 0.1235, "step": 32815 }, { "epoch": 24.04102564102564, "grad_norm": 1.3359375, "learning_rate": 2.525121084110256e-05, "loss": 0.1177, "step": 32816 }, { "epoch": 24.04175824175824, "grad_norm": 0.9453125, "learning_rate": 2.5248263788793048e-05, "loss": 0.0919, "step": 32817 }, { "epoch": 24.042490842490842, "grad_norm": 0.9921875, "learning_rate": 2.5245316857472883e-05, "loss": 0.1118, "step": 32818 }, { "epoch": 24.043223443223443, "grad_norm": 1.5390625, "learning_rate": 2.5242370047156207e-05, "loss": 0.1164, "step": 32819 }, { "epoch": 24.043956043956044, "grad_norm": 1.1015625, "learning_rate": 2.523942335785709e-05, "loss": 0.1342, "step": 32820 }, { "epoch": 24.044688644688645, "grad_norm": 0.828125, "learning_rate": 2.523647678958973e-05, "loss": 0.1262, "step": 32821 }, { "epoch": 24.045421245421245, "grad_norm": 1.09375, "learning_rate": 2.5233530342368183e-05, "loss": 0.1164, "step": 32822 }, { "epoch": 24.046153846153846, "grad_norm": 1.5859375, "learning_rate": 2.5230584016206605e-05, "loss": 0.1283, "step": 32823 }, { "epoch": 24.046886446886447, "grad_norm": 1.25, "learning_rate": 2.5227637811119082e-05, "loss": 0.1199, "step": 32824 }, { "epoch": 24.047619047619047, "grad_norm": 1.2578125, "learning_rate": 2.522469172711973e-05, "loss": 0.1058, "step": 32825 }, { "epoch": 24.048351648351648, "grad_norm": 0.671875, "learning_rate": 2.5221745764222686e-05, "loss": 0.1117, "step": 32826 }, { "epoch": 24.04908424908425, "grad_norm": 0.875, "learning_rate": 2.5218799922442075e-05, "loss": 0.102, "step": 32827 }, { "epoch": 24.04981684981685, "grad_norm": 1.390625, "learning_rate": 2.521585420179197e-05, "loss": 0.1194, "step": 32828 }, { "epoch": 24.05054945054945, "grad_norm": 0.8359375, "learning_rate": 2.5212908602286506e-05, "loss": 0.116, "step": 32829 }, { "epoch": 24.05128205128205, "grad_norm": 1.5859375, "learning_rate": 2.5209963123939822e-05, "loss": 0.1262, "step": 32830 }, { "epoch": 24.052014652014652, "grad_norm": 1.1171875, "learning_rate": 2.520701776676597e-05, "loss": 0.144, "step": 32831 }, { "epoch": 24.052747252747253, "grad_norm": 1.703125, "learning_rate": 2.5204072530779137e-05, "loss": 0.1486, "step": 32832 }, { "epoch": 24.053479853479853, "grad_norm": 0.671875, "learning_rate": 2.520112741599337e-05, "loss": 0.116, "step": 32833 }, { "epoch": 24.054212454212454, "grad_norm": 0.515625, "learning_rate": 2.5198182422422835e-05, "loss": 0.092, "step": 32834 }, { "epoch": 24.054945054945055, "grad_norm": 1.65625, "learning_rate": 2.519523755008159e-05, "loss": 0.152, "step": 32835 }, { "epoch": 24.055677655677655, "grad_norm": 0.87890625, "learning_rate": 2.5192292798983767e-05, "loss": 0.1178, "step": 32836 }, { "epoch": 24.056410256410256, "grad_norm": 1.1015625, "learning_rate": 2.5189348169143485e-05, "loss": 0.1056, "step": 32837 }, { "epoch": 24.057142857142857, "grad_norm": 1.3203125, "learning_rate": 2.5186403660574858e-05, "loss": 0.1232, "step": 32838 }, { "epoch": 24.057875457875458, "grad_norm": 0.91015625, "learning_rate": 2.5183459273291967e-05, "loss": 0.1101, "step": 32839 }, { "epoch": 24.05860805860806, "grad_norm": 1.0625, "learning_rate": 2.5180515007308924e-05, "loss": 0.1188, "step": 32840 }, { "epoch": 24.05934065934066, "grad_norm": 1.0, "learning_rate": 2.5177570862639853e-05, "loss": 0.1149, "step": 32841 }, { "epoch": 24.06007326007326, "grad_norm": 0.98828125, "learning_rate": 2.5174626839298848e-05, "loss": 0.1506, "step": 32842 }, { "epoch": 24.06080586080586, "grad_norm": 1.0078125, "learning_rate": 2.5171682937300045e-05, "loss": 0.1376, "step": 32843 }, { "epoch": 24.06153846153846, "grad_norm": 0.86328125, "learning_rate": 2.5168739156657502e-05, "loss": 0.1193, "step": 32844 }, { "epoch": 24.062271062271062, "grad_norm": 1.40625, "learning_rate": 2.5165795497385358e-05, "loss": 0.1688, "step": 32845 }, { "epoch": 24.063003663003663, "grad_norm": 1.46875, "learning_rate": 2.5162851959497674e-05, "loss": 0.1514, "step": 32846 }, { "epoch": 24.063736263736264, "grad_norm": 1.1640625, "learning_rate": 2.515990854300862e-05, "loss": 0.1208, "step": 32847 }, { "epoch": 24.064468864468864, "grad_norm": 1.2265625, "learning_rate": 2.5156965247932247e-05, "loss": 0.1224, "step": 32848 }, { "epoch": 24.065201465201465, "grad_norm": 1.171875, "learning_rate": 2.5154022074282686e-05, "loss": 0.1231, "step": 32849 }, { "epoch": 24.065934065934066, "grad_norm": 1.3125, "learning_rate": 2.515107902207401e-05, "loss": 0.1226, "step": 32850 }, { "epoch": 24.066666666666666, "grad_norm": 1.078125, "learning_rate": 2.5148136091320336e-05, "loss": 0.1082, "step": 32851 }, { "epoch": 24.067399267399267, "grad_norm": 0.99609375, "learning_rate": 2.5145193282035763e-05, "loss": 0.1204, "step": 32852 }, { "epoch": 24.068131868131868, "grad_norm": 1.171875, "learning_rate": 2.5142250594234414e-05, "loss": 0.1241, "step": 32853 }, { "epoch": 24.06886446886447, "grad_norm": 0.90625, "learning_rate": 2.5139308027930342e-05, "loss": 0.1128, "step": 32854 }, { "epoch": 24.06959706959707, "grad_norm": 0.6953125, "learning_rate": 2.5136365583137677e-05, "loss": 0.097, "step": 32855 }, { "epoch": 24.07032967032967, "grad_norm": 1.4921875, "learning_rate": 2.513342325987053e-05, "loss": 0.1356, "step": 32856 }, { "epoch": 24.07106227106227, "grad_norm": 1.3359375, "learning_rate": 2.5130481058142934e-05, "loss": 0.1425, "step": 32857 }, { "epoch": 24.07179487179487, "grad_norm": 0.625, "learning_rate": 2.5127538977969072e-05, "loss": 0.1088, "step": 32858 }, { "epoch": 24.072527472527472, "grad_norm": 1.2265625, "learning_rate": 2.5124597019362982e-05, "loss": 0.1228, "step": 32859 }, { "epoch": 24.073260073260073, "grad_norm": 1.0078125, "learning_rate": 2.5121655182338805e-05, "loss": 0.1364, "step": 32860 }, { "epoch": 24.073992673992674, "grad_norm": 0.8359375, "learning_rate": 2.511871346691058e-05, "loss": 0.1084, "step": 32861 }, { "epoch": 24.074725274725274, "grad_norm": 1.2109375, "learning_rate": 2.5115771873092436e-05, "loss": 0.1151, "step": 32862 }, { "epoch": 24.075457875457875, "grad_norm": 1.53125, "learning_rate": 2.511283040089846e-05, "loss": 0.1547, "step": 32863 }, { "epoch": 24.076190476190476, "grad_norm": 1.0234375, "learning_rate": 2.5109889050342768e-05, "loss": 0.1103, "step": 32864 }, { "epoch": 24.076923076923077, "grad_norm": 0.8984375, "learning_rate": 2.5106947821439413e-05, "loss": 0.1221, "step": 32865 }, { "epoch": 24.077655677655677, "grad_norm": 1.625, "learning_rate": 2.510400671420251e-05, "loss": 0.1263, "step": 32866 }, { "epoch": 24.078388278388278, "grad_norm": 1.1171875, "learning_rate": 2.510106572864614e-05, "loss": 0.1377, "step": 32867 }, { "epoch": 24.07912087912088, "grad_norm": 1.265625, "learning_rate": 2.509812486478441e-05, "loss": 0.1146, "step": 32868 }, { "epoch": 24.07985347985348, "grad_norm": 0.57421875, "learning_rate": 2.5095184122631417e-05, "loss": 0.1084, "step": 32869 }, { "epoch": 24.08058608058608, "grad_norm": 1.078125, "learning_rate": 2.5092243502201217e-05, "loss": 0.1129, "step": 32870 }, { "epoch": 24.08131868131868, "grad_norm": 1.1953125, "learning_rate": 2.508930300350794e-05, "loss": 0.1064, "step": 32871 }, { "epoch": 24.08205128205128, "grad_norm": 0.90625, "learning_rate": 2.508636262656564e-05, "loss": 0.1102, "step": 32872 }, { "epoch": 24.082783882783882, "grad_norm": 0.7734375, "learning_rate": 2.5083422371388414e-05, "loss": 0.0937, "step": 32873 }, { "epoch": 24.083516483516483, "grad_norm": 1.5703125, "learning_rate": 2.5080482237990355e-05, "loss": 0.1224, "step": 32874 }, { "epoch": 24.084249084249084, "grad_norm": 0.86328125, "learning_rate": 2.507754222638557e-05, "loss": 0.0999, "step": 32875 }, { "epoch": 24.084981684981685, "grad_norm": 1.046875, "learning_rate": 2.5074602336588106e-05, "loss": 0.143, "step": 32876 }, { "epoch": 24.085714285714285, "grad_norm": 0.8984375, "learning_rate": 2.507166256861207e-05, "loss": 0.1129, "step": 32877 }, { "epoch": 24.086446886446886, "grad_norm": 1.0078125, "learning_rate": 2.5068722922471544e-05, "loss": 0.1106, "step": 32878 }, { "epoch": 24.087179487179487, "grad_norm": 0.87109375, "learning_rate": 2.5065783398180637e-05, "loss": 0.0927, "step": 32879 }, { "epoch": 24.087912087912088, "grad_norm": 0.68359375, "learning_rate": 2.5062843995753394e-05, "loss": 0.1023, "step": 32880 }, { "epoch": 24.08864468864469, "grad_norm": 1.2890625, "learning_rate": 2.5059904715203912e-05, "loss": 0.1261, "step": 32881 }, { "epoch": 24.08937728937729, "grad_norm": 1.1875, "learning_rate": 2.5056965556546296e-05, "loss": 0.1045, "step": 32882 }, { "epoch": 24.09010989010989, "grad_norm": 0.6015625, "learning_rate": 2.505402651979457e-05, "loss": 0.109, "step": 32883 }, { "epoch": 24.09084249084249, "grad_norm": 0.7421875, "learning_rate": 2.5051087604962904e-05, "loss": 0.1103, "step": 32884 }, { "epoch": 24.09157509157509, "grad_norm": 0.75, "learning_rate": 2.5048148812065302e-05, "loss": 0.1079, "step": 32885 }, { "epoch": 24.092307692307692, "grad_norm": 1.234375, "learning_rate": 2.5045210141115905e-05, "loss": 0.1243, "step": 32886 }, { "epoch": 24.093040293040293, "grad_norm": 1.5390625, "learning_rate": 2.5042271592128736e-05, "loss": 0.129, "step": 32887 }, { "epoch": 24.093772893772893, "grad_norm": 1.296875, "learning_rate": 2.5039333165117902e-05, "loss": 0.1754, "step": 32888 }, { "epoch": 24.094505494505494, "grad_norm": 1.484375, "learning_rate": 2.5036394860097485e-05, "loss": 0.1117, "step": 32889 }, { "epoch": 24.095238095238095, "grad_norm": 1.0859375, "learning_rate": 2.503345667708157e-05, "loss": 0.1059, "step": 32890 }, { "epoch": 24.095970695970696, "grad_norm": 0.65234375, "learning_rate": 2.503051861608421e-05, "loss": 0.1297, "step": 32891 }, { "epoch": 24.096703296703296, "grad_norm": 0.80859375, "learning_rate": 2.5027580677119493e-05, "loss": 0.1041, "step": 32892 }, { "epoch": 24.097435897435897, "grad_norm": 0.953125, "learning_rate": 2.5024642860201503e-05, "loss": 0.1067, "step": 32893 }, { "epoch": 24.098168498168498, "grad_norm": 1.34375, "learning_rate": 2.5021705165344316e-05, "loss": 0.1391, "step": 32894 }, { "epoch": 24.0989010989011, "grad_norm": 0.70703125, "learning_rate": 2.5018767592562015e-05, "loss": 0.1073, "step": 32895 }, { "epoch": 24.0996336996337, "grad_norm": 0.96484375, "learning_rate": 2.501583014186865e-05, "loss": 0.1419, "step": 32896 }, { "epoch": 24.1003663003663, "grad_norm": 1.3359375, "learning_rate": 2.5012892813278322e-05, "loss": 0.1334, "step": 32897 }, { "epoch": 24.1010989010989, "grad_norm": 1.4453125, "learning_rate": 2.5009955606805076e-05, "loss": 0.132, "step": 32898 }, { "epoch": 24.1018315018315, "grad_norm": 1.0859375, "learning_rate": 2.5007018522463e-05, "loss": 0.11, "step": 32899 }, { "epoch": 24.102564102564102, "grad_norm": 0.9296875, "learning_rate": 2.5004081560266167e-05, "loss": 0.1451, "step": 32900 }, { "epoch": 24.103296703296703, "grad_norm": 0.99609375, "learning_rate": 2.5001144720228663e-05, "loss": 0.1117, "step": 32901 }, { "epoch": 24.104029304029304, "grad_norm": 0.74609375, "learning_rate": 2.4998208002364536e-05, "loss": 0.1108, "step": 32902 }, { "epoch": 24.104761904761904, "grad_norm": 1.1796875, "learning_rate": 2.4995271406687865e-05, "loss": 0.1324, "step": 32903 }, { "epoch": 24.105494505494505, "grad_norm": 1.1484375, "learning_rate": 2.4992334933212713e-05, "loss": 0.1257, "step": 32904 }, { "epoch": 24.106227106227106, "grad_norm": 0.5859375, "learning_rate": 2.4989398581953183e-05, "loss": 0.1065, "step": 32905 }, { "epoch": 24.106959706959707, "grad_norm": 1.375, "learning_rate": 2.4986462352923303e-05, "loss": 0.1339, "step": 32906 }, { "epoch": 24.107692307692307, "grad_norm": 1.4296875, "learning_rate": 2.498352624613715e-05, "loss": 0.1145, "step": 32907 }, { "epoch": 24.108424908424908, "grad_norm": 0.83984375, "learning_rate": 2.4980590261608815e-05, "loss": 0.1269, "step": 32908 }, { "epoch": 24.10915750915751, "grad_norm": 0.828125, "learning_rate": 2.4977654399352325e-05, "loss": 0.1429, "step": 32909 }, { "epoch": 24.10989010989011, "grad_norm": 0.62890625, "learning_rate": 2.49747186593818e-05, "loss": 0.1082, "step": 32910 }, { "epoch": 24.11062271062271, "grad_norm": 0.69140625, "learning_rate": 2.4971783041711264e-05, "loss": 0.1216, "step": 32911 }, { "epoch": 24.11135531135531, "grad_norm": 1.265625, "learning_rate": 2.496884754635481e-05, "loss": 0.1335, "step": 32912 }, { "epoch": 24.11208791208791, "grad_norm": 1.28125, "learning_rate": 2.496591217332647e-05, "loss": 0.1059, "step": 32913 }, { "epoch": 24.112820512820512, "grad_norm": 1.3203125, "learning_rate": 2.4962976922640327e-05, "loss": 0.1582, "step": 32914 }, { "epoch": 24.113553113553113, "grad_norm": 1.0546875, "learning_rate": 2.4960041794310442e-05, "loss": 0.1149, "step": 32915 }, { "epoch": 24.114285714285714, "grad_norm": 1.296875, "learning_rate": 2.4957106788350903e-05, "loss": 0.0992, "step": 32916 }, { "epoch": 24.115018315018315, "grad_norm": 1.0078125, "learning_rate": 2.4954171904775735e-05, "loss": 0.1261, "step": 32917 }, { "epoch": 24.115750915750915, "grad_norm": 1.1171875, "learning_rate": 2.4951237143599014e-05, "loss": 0.1065, "step": 32918 }, { "epoch": 24.116483516483516, "grad_norm": 1.9609375, "learning_rate": 2.4948302504834806e-05, "loss": 0.128, "step": 32919 }, { "epoch": 24.117216117216117, "grad_norm": 0.88671875, "learning_rate": 2.494536798849716e-05, "loss": 0.1135, "step": 32920 }, { "epoch": 24.117948717948718, "grad_norm": 0.6953125, "learning_rate": 2.4942433594600172e-05, "loss": 0.1225, "step": 32921 }, { "epoch": 24.11868131868132, "grad_norm": 0.6796875, "learning_rate": 2.493949932315785e-05, "loss": 0.1201, "step": 32922 }, { "epoch": 24.11941391941392, "grad_norm": 1.125, "learning_rate": 2.49365651741843e-05, "loss": 0.1167, "step": 32923 }, { "epoch": 24.12014652014652, "grad_norm": 0.75, "learning_rate": 2.4933631147693533e-05, "loss": 0.1194, "step": 32924 }, { "epoch": 24.12087912087912, "grad_norm": 0.73046875, "learning_rate": 2.493069724369964e-05, "loss": 0.1206, "step": 32925 }, { "epoch": 24.12161172161172, "grad_norm": 1.5703125, "learning_rate": 2.4927763462216668e-05, "loss": 0.1274, "step": 32926 }, { "epoch": 24.122344322344322, "grad_norm": 0.8828125, "learning_rate": 2.4924829803258693e-05, "loss": 0.1182, "step": 32927 }, { "epoch": 24.123076923076923, "grad_norm": 0.73828125, "learning_rate": 2.492189626683974e-05, "loss": 0.106, "step": 32928 }, { "epoch": 24.123809523809523, "grad_norm": 0.8046875, "learning_rate": 2.4918962852973873e-05, "loss": 0.1169, "step": 32929 }, { "epoch": 24.124542124542124, "grad_norm": 0.89453125, "learning_rate": 2.491602956167516e-05, "loss": 0.1541, "step": 32930 }, { "epoch": 24.125274725274725, "grad_norm": 0.8046875, "learning_rate": 2.491309639295764e-05, "loss": 0.0788, "step": 32931 }, { "epoch": 24.126007326007326, "grad_norm": 0.95703125, "learning_rate": 2.49101633468354e-05, "loss": 0.1403, "step": 32932 }, { "epoch": 24.126739926739926, "grad_norm": 1.1484375, "learning_rate": 2.4907230423322445e-05, "loss": 0.1324, "step": 32933 }, { "epoch": 24.127472527472527, "grad_norm": 0.9453125, "learning_rate": 2.490429762243287e-05, "loss": 0.1145, "step": 32934 }, { "epoch": 24.128205128205128, "grad_norm": 1.09375, "learning_rate": 2.4901364944180667e-05, "loss": 0.1508, "step": 32935 }, { "epoch": 24.12893772893773, "grad_norm": 0.84375, "learning_rate": 2.4898432388579972e-05, "loss": 0.1323, "step": 32936 }, { "epoch": 24.12967032967033, "grad_norm": 1.1015625, "learning_rate": 2.4895499955644773e-05, "loss": 0.1224, "step": 32937 }, { "epoch": 24.13040293040293, "grad_norm": 1.1875, "learning_rate": 2.489256764538915e-05, "loss": 0.1014, "step": 32938 }, { "epoch": 24.13113553113553, "grad_norm": 1.109375, "learning_rate": 2.488963545782713e-05, "loss": 0.1174, "step": 32939 }, { "epoch": 24.13186813186813, "grad_norm": 0.984375, "learning_rate": 2.4886703392972772e-05, "loss": 0.1336, "step": 32940 }, { "epoch": 24.132600732600732, "grad_norm": 1.296875, "learning_rate": 2.4883771450840128e-05, "loss": 0.1436, "step": 32941 }, { "epoch": 24.133333333333333, "grad_norm": 0.91015625, "learning_rate": 2.488083963144326e-05, "loss": 0.1025, "step": 32942 }, { "epoch": 24.134065934065934, "grad_norm": 0.99609375, "learning_rate": 2.4877907934796183e-05, "loss": 0.1, "step": 32943 }, { "epoch": 24.134798534798534, "grad_norm": 0.94140625, "learning_rate": 2.487497636091296e-05, "loss": 0.122, "step": 32944 }, { "epoch": 24.135531135531135, "grad_norm": 1.0859375, "learning_rate": 2.4872044909807635e-05, "loss": 0.109, "step": 32945 }, { "epoch": 24.136263736263736, "grad_norm": 0.953125, "learning_rate": 2.4869113581494262e-05, "loss": 0.135, "step": 32946 }, { "epoch": 24.136996336996337, "grad_norm": 1.71875, "learning_rate": 2.4866182375986887e-05, "loss": 0.1311, "step": 32947 }, { "epoch": 24.137728937728937, "grad_norm": 0.83203125, "learning_rate": 2.486325129329954e-05, "loss": 0.1164, "step": 32948 }, { "epoch": 24.138461538461538, "grad_norm": 0.6796875, "learning_rate": 2.4860320333446284e-05, "loss": 0.1122, "step": 32949 }, { "epoch": 24.13919413919414, "grad_norm": 1.4140625, "learning_rate": 2.485738949644114e-05, "loss": 0.112, "step": 32950 }, { "epoch": 24.13992673992674, "grad_norm": 1.3125, "learning_rate": 2.4854458782298152e-05, "loss": 0.1332, "step": 32951 }, { "epoch": 24.14065934065934, "grad_norm": 0.76171875, "learning_rate": 2.4851528191031375e-05, "loss": 0.1094, "step": 32952 }, { "epoch": 24.14139194139194, "grad_norm": 0.98046875, "learning_rate": 2.484859772265486e-05, "loss": 0.1105, "step": 32953 }, { "epoch": 24.14212454212454, "grad_norm": 1.140625, "learning_rate": 2.4845667377182617e-05, "loss": 0.1132, "step": 32954 }, { "epoch": 24.142857142857142, "grad_norm": 1.4453125, "learning_rate": 2.484273715462871e-05, "loss": 0.1326, "step": 32955 }, { "epoch": 24.143589743589743, "grad_norm": 1.453125, "learning_rate": 2.4839807055007168e-05, "loss": 0.1175, "step": 32956 }, { "epoch": 24.144322344322344, "grad_norm": 1.3984375, "learning_rate": 2.4836877078332035e-05, "loss": 0.1392, "step": 32957 }, { "epoch": 24.145054945054945, "grad_norm": 1.1171875, "learning_rate": 2.4833947224617364e-05, "loss": 0.1497, "step": 32958 }, { "epoch": 24.145787545787545, "grad_norm": 1.3359375, "learning_rate": 2.4831017493877163e-05, "loss": 0.1199, "step": 32959 }, { "epoch": 24.146520146520146, "grad_norm": 0.55078125, "learning_rate": 2.4828087886125498e-05, "loss": 0.1125, "step": 32960 }, { "epoch": 24.147252747252747, "grad_norm": 1.1171875, "learning_rate": 2.4825158401376356e-05, "loss": 0.1261, "step": 32961 }, { "epoch": 24.147985347985347, "grad_norm": 1.171875, "learning_rate": 2.4822229039643846e-05, "loss": 0.124, "step": 32962 }, { "epoch": 24.148717948717948, "grad_norm": 0.94140625, "learning_rate": 2.4819299800941946e-05, "loss": 0.1361, "step": 32963 }, { "epoch": 24.14945054945055, "grad_norm": 1.078125, "learning_rate": 2.4816370685284735e-05, "loss": 0.1539, "step": 32964 }, { "epoch": 24.15018315018315, "grad_norm": 0.8125, "learning_rate": 2.48134416926862e-05, "loss": 0.1293, "step": 32965 }, { "epoch": 24.15091575091575, "grad_norm": 0.69921875, "learning_rate": 2.48105128231604e-05, "loss": 0.1105, "step": 32966 }, { "epoch": 24.15164835164835, "grad_norm": 1.171875, "learning_rate": 2.4807584076721367e-05, "loss": 0.1074, "step": 32967 }, { "epoch": 24.152380952380952, "grad_norm": 0.8828125, "learning_rate": 2.4804655453383154e-05, "loss": 0.1098, "step": 32968 }, { "epoch": 24.153113553113553, "grad_norm": 1.5625, "learning_rate": 2.4801726953159746e-05, "loss": 0.1124, "step": 32969 }, { "epoch": 24.153846153846153, "grad_norm": 0.921875, "learning_rate": 2.47987985760652e-05, "loss": 0.1135, "step": 32970 }, { "epoch": 24.154578754578754, "grad_norm": 0.97265625, "learning_rate": 2.4795870322113554e-05, "loss": 0.1398, "step": 32971 }, { "epoch": 24.155311355311355, "grad_norm": 1.4921875, "learning_rate": 2.4792942191318833e-05, "loss": 0.1106, "step": 32972 }, { "epoch": 24.156043956043955, "grad_norm": 0.73046875, "learning_rate": 2.4790014183695073e-05, "loss": 0.1216, "step": 32973 }, { "epoch": 24.156776556776556, "grad_norm": 0.91796875, "learning_rate": 2.4787086299256283e-05, "loss": 0.1082, "step": 32974 }, { "epoch": 24.157509157509157, "grad_norm": 0.9765625, "learning_rate": 2.4784158538016525e-05, "loss": 0.1189, "step": 32975 }, { "epoch": 24.158241758241758, "grad_norm": 1.0625, "learning_rate": 2.4781230899989784e-05, "loss": 0.1113, "step": 32976 }, { "epoch": 24.15897435897436, "grad_norm": 0.9921875, "learning_rate": 2.4778303385190108e-05, "loss": 0.1478, "step": 32977 }, { "epoch": 24.15970695970696, "grad_norm": 1.1328125, "learning_rate": 2.4775375993631523e-05, "loss": 0.1124, "step": 32978 }, { "epoch": 24.16043956043956, "grad_norm": 1.1953125, "learning_rate": 2.477244872532808e-05, "loss": 0.159, "step": 32979 }, { "epoch": 24.16117216117216, "grad_norm": 1.078125, "learning_rate": 2.476952158029376e-05, "loss": 0.124, "step": 32980 }, { "epoch": 24.16190476190476, "grad_norm": 0.8984375, "learning_rate": 2.4766594558542604e-05, "loss": 0.0906, "step": 32981 }, { "epoch": 24.162637362637362, "grad_norm": 0.93359375, "learning_rate": 2.476366766008865e-05, "loss": 0.1287, "step": 32982 }, { "epoch": 24.163369963369963, "grad_norm": 0.93359375, "learning_rate": 2.4760740884945906e-05, "loss": 0.14, "step": 32983 }, { "epoch": 24.164102564102564, "grad_norm": 0.71875, "learning_rate": 2.4757814233128424e-05, "loss": 0.1124, "step": 32984 }, { "epoch": 24.164835164835164, "grad_norm": 0.64453125, "learning_rate": 2.4754887704650185e-05, "loss": 0.0975, "step": 32985 }, { "epoch": 24.165567765567765, "grad_norm": 0.71484375, "learning_rate": 2.4751961299525246e-05, "loss": 0.1058, "step": 32986 }, { "epoch": 24.166300366300366, "grad_norm": 0.8046875, "learning_rate": 2.474903501776758e-05, "loss": 0.1197, "step": 32987 }, { "epoch": 24.167032967032966, "grad_norm": 0.60546875, "learning_rate": 2.4746108859391282e-05, "loss": 0.1279, "step": 32988 }, { "epoch": 24.167765567765567, "grad_norm": 1.140625, "learning_rate": 2.4743182824410305e-05, "loss": 0.1241, "step": 32989 }, { "epoch": 24.168498168498168, "grad_norm": 0.890625, "learning_rate": 2.4740256912838717e-05, "loss": 0.0953, "step": 32990 }, { "epoch": 24.16923076923077, "grad_norm": 1.15625, "learning_rate": 2.473733112469049e-05, "loss": 0.1308, "step": 32991 }, { "epoch": 24.16996336996337, "grad_norm": 0.84375, "learning_rate": 2.4734405459979672e-05, "loss": 0.1395, "step": 32992 }, { "epoch": 24.17069597069597, "grad_norm": 0.87109375, "learning_rate": 2.4731479918720274e-05, "loss": 0.0872, "step": 32993 }, { "epoch": 24.17142857142857, "grad_norm": 0.66015625, "learning_rate": 2.4728554500926333e-05, "loss": 0.1159, "step": 32994 }, { "epoch": 24.17216117216117, "grad_norm": 1.1875, "learning_rate": 2.472562920661183e-05, "loss": 0.1332, "step": 32995 }, { "epoch": 24.172893772893772, "grad_norm": 0.77734375, "learning_rate": 2.4722704035790802e-05, "loss": 0.1062, "step": 32996 }, { "epoch": 24.173626373626373, "grad_norm": 1.0390625, "learning_rate": 2.4719778988477258e-05, "loss": 0.1384, "step": 32997 }, { "epoch": 24.174358974358974, "grad_norm": 1.25, "learning_rate": 2.4716854064685214e-05, "loss": 0.1259, "step": 32998 }, { "epoch": 24.175091575091574, "grad_norm": 0.62890625, "learning_rate": 2.471392926442871e-05, "loss": 0.1006, "step": 32999 }, { "epoch": 24.175824175824175, "grad_norm": 1.0, "learning_rate": 2.471100458772171e-05, "loss": 0.1137, "step": 33000 }, { "epoch": 24.176556776556776, "grad_norm": 1.2734375, "learning_rate": 2.4708080034578277e-05, "loss": 0.1271, "step": 33001 }, { "epoch": 24.177289377289377, "grad_norm": 0.984375, "learning_rate": 2.470515560501238e-05, "loss": 0.1249, "step": 33002 }, { "epoch": 24.178021978021977, "grad_norm": 0.765625, "learning_rate": 2.4702231299038047e-05, "loss": 0.1113, "step": 33003 }, { "epoch": 24.178754578754578, "grad_norm": 1.2734375, "learning_rate": 2.4699307116669297e-05, "loss": 0.1518, "step": 33004 }, { "epoch": 24.17948717948718, "grad_norm": 0.86328125, "learning_rate": 2.4696383057920153e-05, "loss": 0.1264, "step": 33005 }, { "epoch": 24.18021978021978, "grad_norm": 1.0078125, "learning_rate": 2.469345912280459e-05, "loss": 0.1387, "step": 33006 }, { "epoch": 24.18095238095238, "grad_norm": 0.97265625, "learning_rate": 2.469053531133664e-05, "loss": 0.1177, "step": 33007 }, { "epoch": 24.18168498168498, "grad_norm": 0.91796875, "learning_rate": 2.46876116235303e-05, "loss": 0.1118, "step": 33008 }, { "epoch": 24.18241758241758, "grad_norm": 0.7421875, "learning_rate": 2.4684688059399596e-05, "loss": 0.1147, "step": 33009 }, { "epoch": 24.183150183150182, "grad_norm": 1.171875, "learning_rate": 2.4681764618958538e-05, "loss": 0.116, "step": 33010 }, { "epoch": 24.183882783882783, "grad_norm": 0.95703125, "learning_rate": 2.4678841302221105e-05, "loss": 0.1172, "step": 33011 }, { "epoch": 24.184615384615384, "grad_norm": 0.7734375, "learning_rate": 2.467591810920134e-05, "loss": 0.102, "step": 33012 }, { "epoch": 24.185347985347985, "grad_norm": 0.7109375, "learning_rate": 2.4672995039913184e-05, "loss": 0.1127, "step": 33013 }, { "epoch": 24.186080586080585, "grad_norm": 0.62109375, "learning_rate": 2.4670072094370735e-05, "loss": 0.0978, "step": 33014 }, { "epoch": 24.186813186813186, "grad_norm": 0.87109375, "learning_rate": 2.4667149272587925e-05, "loss": 0.1024, "step": 33015 }, { "epoch": 24.187545787545787, "grad_norm": 1.0234375, "learning_rate": 2.4664226574578804e-05, "loss": 0.1237, "step": 33016 }, { "epoch": 24.188278388278388, "grad_norm": 1.0546875, "learning_rate": 2.4661304000357334e-05, "loss": 0.1452, "step": 33017 }, { "epoch": 24.18901098901099, "grad_norm": 0.92578125, "learning_rate": 2.4658381549937543e-05, "loss": 0.0877, "step": 33018 }, { "epoch": 24.18974358974359, "grad_norm": 1.140625, "learning_rate": 2.4655459223333422e-05, "loss": 0.1301, "step": 33019 }, { "epoch": 24.19047619047619, "grad_norm": 1.1953125, "learning_rate": 2.4652537020559004e-05, "loss": 0.1426, "step": 33020 }, { "epoch": 24.19120879120879, "grad_norm": 0.82421875, "learning_rate": 2.464961494162825e-05, "loss": 0.1287, "step": 33021 }, { "epoch": 24.19194139194139, "grad_norm": 0.7578125, "learning_rate": 2.464669298655517e-05, "loss": 0.1133, "step": 33022 }, { "epoch": 24.192673992673992, "grad_norm": 1.0390625, "learning_rate": 2.4643771155353778e-05, "loss": 0.1365, "step": 33023 }, { "epoch": 24.193406593406593, "grad_norm": 0.93359375, "learning_rate": 2.4640849448038064e-05, "loss": 0.1259, "step": 33024 }, { "epoch": 24.194139194139193, "grad_norm": 0.9609375, "learning_rate": 2.4637927864622044e-05, "loss": 0.1053, "step": 33025 }, { "epoch": 24.194871794871794, "grad_norm": 0.62109375, "learning_rate": 2.463500640511969e-05, "loss": 0.0929, "step": 33026 }, { "epoch": 24.195604395604395, "grad_norm": 0.81640625, "learning_rate": 2.4632085069545023e-05, "loss": 0.126, "step": 33027 }, { "epoch": 24.196336996336996, "grad_norm": 1.078125, "learning_rate": 2.4629163857912015e-05, "loss": 0.1083, "step": 33028 }, { "epoch": 24.197069597069596, "grad_norm": 1.46875, "learning_rate": 2.4626242770234672e-05, "loss": 0.1208, "step": 33029 }, { "epoch": 24.197802197802197, "grad_norm": 1.1328125, "learning_rate": 2.4623321806526994e-05, "loss": 0.1145, "step": 33030 }, { "epoch": 24.198534798534798, "grad_norm": 1.0703125, "learning_rate": 2.4620400966802995e-05, "loss": 0.1282, "step": 33031 }, { "epoch": 24.1992673992674, "grad_norm": 1.3671875, "learning_rate": 2.4617480251076637e-05, "loss": 0.0947, "step": 33032 }, { "epoch": 24.2, "grad_norm": 1.390625, "learning_rate": 2.461455965936192e-05, "loss": 0.0903, "step": 33033 }, { "epoch": 24.2007326007326, "grad_norm": 1.0625, "learning_rate": 2.4611639191672846e-05, "loss": 0.1233, "step": 33034 }, { "epoch": 24.2014652014652, "grad_norm": 0.88671875, "learning_rate": 2.460871884802341e-05, "loss": 0.1211, "step": 33035 }, { "epoch": 24.2021978021978, "grad_norm": 0.72265625, "learning_rate": 2.4605798628427623e-05, "loss": 0.1184, "step": 33036 }, { "epoch": 24.202930402930402, "grad_norm": 1.1484375, "learning_rate": 2.4602878532899433e-05, "loss": 0.1366, "step": 33037 }, { "epoch": 24.203663003663003, "grad_norm": 1.3828125, "learning_rate": 2.4599958561452875e-05, "loss": 0.1023, "step": 33038 }, { "epoch": 24.204395604395604, "grad_norm": 1.4140625, "learning_rate": 2.459703871410188e-05, "loss": 0.113, "step": 33039 }, { "epoch": 24.205128205128204, "grad_norm": 1.0234375, "learning_rate": 2.4594118990860512e-05, "loss": 0.1226, "step": 33040 }, { "epoch": 24.205860805860805, "grad_norm": 1.2109375, "learning_rate": 2.4591199391742702e-05, "loss": 0.1211, "step": 33041 }, { "epoch": 24.206593406593406, "grad_norm": 1.625, "learning_rate": 2.4588279916762484e-05, "loss": 0.1327, "step": 33042 }, { "epoch": 24.207326007326007, "grad_norm": 1.0625, "learning_rate": 2.4585360565933806e-05, "loss": 0.1219, "step": 33043 }, { "epoch": 24.208058608058607, "grad_norm": 1.5625, "learning_rate": 2.4582441339270667e-05, "loss": 0.1147, "step": 33044 }, { "epoch": 24.208791208791208, "grad_norm": 0.7734375, "learning_rate": 2.457952223678707e-05, "loss": 0.1139, "step": 33045 }, { "epoch": 24.20952380952381, "grad_norm": 0.8828125, "learning_rate": 2.457660325849698e-05, "loss": 0.1562, "step": 33046 }, { "epoch": 24.21025641025641, "grad_norm": 0.96484375, "learning_rate": 2.4573684404414415e-05, "loss": 0.1027, "step": 33047 }, { "epoch": 24.21098901098901, "grad_norm": 1.28125, "learning_rate": 2.4570765674553322e-05, "loss": 0.146, "step": 33048 }, { "epoch": 24.21172161172161, "grad_norm": 0.93359375, "learning_rate": 2.4567847068927703e-05, "loss": 0.1348, "step": 33049 }, { "epoch": 24.21245421245421, "grad_norm": 0.95703125, "learning_rate": 2.4564928587551545e-05, "loss": 0.1106, "step": 33050 }, { "epoch": 24.213186813186812, "grad_norm": 0.80859375, "learning_rate": 2.456201023043884e-05, "loss": 0.1153, "step": 33051 }, { "epoch": 24.213919413919413, "grad_norm": 0.59765625, "learning_rate": 2.4559091997603545e-05, "loss": 0.1008, "step": 33052 }, { "epoch": 24.214652014652014, "grad_norm": 1.1015625, "learning_rate": 2.4556173889059673e-05, "loss": 0.1104, "step": 33053 }, { "epoch": 24.215384615384615, "grad_norm": 0.79296875, "learning_rate": 2.455325590482117e-05, "loss": 0.1459, "step": 33054 }, { "epoch": 24.216117216117215, "grad_norm": 0.9140625, "learning_rate": 2.4550338044902035e-05, "loss": 0.1007, "step": 33055 }, { "epoch": 24.216849816849816, "grad_norm": 1.3046875, "learning_rate": 2.4547420309316252e-05, "loss": 0.1353, "step": 33056 }, { "epoch": 24.217582417582417, "grad_norm": 0.8828125, "learning_rate": 2.4544502698077813e-05, "loss": 0.1226, "step": 33057 }, { "epoch": 24.218315018315018, "grad_norm": 0.96484375, "learning_rate": 2.4541585211200667e-05, "loss": 0.1305, "step": 33058 }, { "epoch": 24.21904761904762, "grad_norm": 1.21875, "learning_rate": 2.4538667848698803e-05, "loss": 0.1442, "step": 33059 }, { "epoch": 24.21978021978022, "grad_norm": 1.5703125, "learning_rate": 2.4535750610586207e-05, "loss": 0.1167, "step": 33060 }, { "epoch": 24.22051282051282, "grad_norm": 1.2890625, "learning_rate": 2.4532833496876852e-05, "loss": 0.1098, "step": 33061 }, { "epoch": 24.22124542124542, "grad_norm": 1.109375, "learning_rate": 2.4529916507584735e-05, "loss": 0.1059, "step": 33062 }, { "epoch": 24.22197802197802, "grad_norm": 0.76953125, "learning_rate": 2.45269996427238e-05, "loss": 0.1207, "step": 33063 }, { "epoch": 24.222710622710622, "grad_norm": 0.765625, "learning_rate": 2.452408290230805e-05, "loss": 0.131, "step": 33064 }, { "epoch": 24.223443223443223, "grad_norm": 1.296875, "learning_rate": 2.4521166286351413e-05, "loss": 0.0993, "step": 33065 }, { "epoch": 24.224175824175823, "grad_norm": 0.94921875, "learning_rate": 2.4518249794867924e-05, "loss": 0.1248, "step": 33066 }, { "epoch": 24.224908424908424, "grad_norm": 0.6171875, "learning_rate": 2.4515333427871516e-05, "loss": 0.0911, "step": 33067 }, { "epoch": 24.225641025641025, "grad_norm": 1.3671875, "learning_rate": 2.45124171853762e-05, "loss": 0.0962, "step": 33068 }, { "epoch": 24.226373626373626, "grad_norm": 0.9140625, "learning_rate": 2.4509501067395903e-05, "loss": 0.0998, "step": 33069 }, { "epoch": 24.227106227106226, "grad_norm": 1.0546875, "learning_rate": 2.450658507394462e-05, "loss": 0.1403, "step": 33070 }, { "epoch": 24.227838827838827, "grad_norm": 1.0, "learning_rate": 2.4503669205036315e-05, "loss": 0.1161, "step": 33071 }, { "epoch": 24.228571428571428, "grad_norm": 0.9765625, "learning_rate": 2.4500753460684966e-05, "loss": 0.1087, "step": 33072 }, { "epoch": 24.22930402930403, "grad_norm": 0.77734375, "learning_rate": 2.4497837840904565e-05, "loss": 0.1331, "step": 33073 }, { "epoch": 24.23003663003663, "grad_norm": 1.3671875, "learning_rate": 2.449492234570904e-05, "loss": 0.1243, "step": 33074 }, { "epoch": 24.23076923076923, "grad_norm": 1.3046875, "learning_rate": 2.449200697511237e-05, "loss": 0.1382, "step": 33075 }, { "epoch": 24.23150183150183, "grad_norm": 0.90625, "learning_rate": 2.448909172912854e-05, "loss": 0.1028, "step": 33076 }, { "epoch": 24.23223443223443, "grad_norm": 1.1171875, "learning_rate": 2.448617660777153e-05, "loss": 0.1317, "step": 33077 }, { "epoch": 24.232967032967032, "grad_norm": 1.0859375, "learning_rate": 2.4483261611055263e-05, "loss": 0.1326, "step": 33078 }, { "epoch": 24.233699633699633, "grad_norm": 1.0625, "learning_rate": 2.448034673899376e-05, "loss": 0.1331, "step": 33079 }, { "epoch": 24.234432234432234, "grad_norm": 0.91796875, "learning_rate": 2.447743199160093e-05, "loss": 0.1308, "step": 33080 }, { "epoch": 24.235164835164834, "grad_norm": 1.2578125, "learning_rate": 2.4474517368890767e-05, "loss": 0.1176, "step": 33081 }, { "epoch": 24.235897435897435, "grad_norm": 0.92578125, "learning_rate": 2.4471602870877237e-05, "loss": 0.0876, "step": 33082 }, { "epoch": 24.236630036630036, "grad_norm": 1.4921875, "learning_rate": 2.4468688497574318e-05, "loss": 0.1182, "step": 33083 }, { "epoch": 24.237362637362637, "grad_norm": 0.890625, "learning_rate": 2.446577424899594e-05, "loss": 0.1196, "step": 33084 }, { "epoch": 24.238095238095237, "grad_norm": 1.390625, "learning_rate": 2.446286012515609e-05, "loss": 0.1201, "step": 33085 }, { "epoch": 24.238827838827838, "grad_norm": 1.828125, "learning_rate": 2.4459946126068716e-05, "loss": 0.1372, "step": 33086 }, { "epoch": 24.23956043956044, "grad_norm": 0.984375, "learning_rate": 2.445703225174779e-05, "loss": 0.13, "step": 33087 }, { "epoch": 24.24029304029304, "grad_norm": 1.1484375, "learning_rate": 2.4454118502207294e-05, "loss": 0.1192, "step": 33088 }, { "epoch": 24.24102564102564, "grad_norm": 1.0234375, "learning_rate": 2.4451204877461147e-05, "loss": 0.1141, "step": 33089 }, { "epoch": 24.24175824175824, "grad_norm": 1.0546875, "learning_rate": 2.444829137752335e-05, "loss": 0.1131, "step": 33090 }, { "epoch": 24.24249084249084, "grad_norm": 1.21875, "learning_rate": 2.4445378002407797e-05, "loss": 0.1189, "step": 33091 }, { "epoch": 24.243223443223442, "grad_norm": 1.328125, "learning_rate": 2.4442464752128537e-05, "loss": 0.1361, "step": 33092 }, { "epoch": 24.243956043956043, "grad_norm": 1.3203125, "learning_rate": 2.443955162669946e-05, "loss": 0.1476, "step": 33093 }, { "epoch": 24.244688644688644, "grad_norm": 1.296875, "learning_rate": 2.4436638626134557e-05, "loss": 0.1508, "step": 33094 }, { "epoch": 24.245421245421245, "grad_norm": 0.80859375, "learning_rate": 2.4433725750447764e-05, "loss": 0.112, "step": 33095 }, { "epoch": 24.246153846153845, "grad_norm": 0.78125, "learning_rate": 2.443081299965305e-05, "loss": 0.1309, "step": 33096 }, { "epoch": 24.246886446886446, "grad_norm": 1.0234375, "learning_rate": 2.4427900373764364e-05, "loss": 0.142, "step": 33097 }, { "epoch": 24.247619047619047, "grad_norm": 1.1171875, "learning_rate": 2.442498787279567e-05, "loss": 0.1183, "step": 33098 }, { "epoch": 24.248351648351647, "grad_norm": 1.0625, "learning_rate": 2.4422075496760933e-05, "loss": 0.0951, "step": 33099 }, { "epoch": 24.249084249084248, "grad_norm": 0.8203125, "learning_rate": 2.4419163245674077e-05, "loss": 0.1141, "step": 33100 }, { "epoch": 24.24981684981685, "grad_norm": 1.3046875, "learning_rate": 2.4416251119549093e-05, "loss": 0.1386, "step": 33101 }, { "epoch": 24.25054945054945, "grad_norm": 1.03125, "learning_rate": 2.4413339118399874e-05, "loss": 0.1378, "step": 33102 }, { "epoch": 24.25128205128205, "grad_norm": 1.1015625, "learning_rate": 2.4410427242240445e-05, "loss": 0.1287, "step": 33103 }, { "epoch": 24.25201465201465, "grad_norm": 1.3046875, "learning_rate": 2.440751549108471e-05, "loss": 0.1167, "step": 33104 }, { "epoch": 24.252747252747252, "grad_norm": 0.84765625, "learning_rate": 2.4404603864946642e-05, "loss": 0.1435, "step": 33105 }, { "epoch": 24.253479853479853, "grad_norm": 0.796875, "learning_rate": 2.4401692363840175e-05, "loss": 0.1041, "step": 33106 }, { "epoch": 24.254212454212453, "grad_norm": 0.97265625, "learning_rate": 2.4398780987779265e-05, "loss": 0.0904, "step": 33107 }, { "epoch": 24.254945054945054, "grad_norm": 1.671875, "learning_rate": 2.439586973677786e-05, "loss": 0.1203, "step": 33108 }, { "epoch": 24.255677655677655, "grad_norm": 0.66796875, "learning_rate": 2.439295861084994e-05, "loss": 0.1182, "step": 33109 }, { "epoch": 24.256410256410255, "grad_norm": 0.94140625, "learning_rate": 2.43900476100094e-05, "loss": 0.1167, "step": 33110 }, { "epoch": 24.257142857142856, "grad_norm": 1.3828125, "learning_rate": 2.4387136734270217e-05, "loss": 0.1071, "step": 33111 }, { "epoch": 24.257875457875457, "grad_norm": 1.015625, "learning_rate": 2.438422598364633e-05, "loss": 0.1337, "step": 33112 }, { "epoch": 24.258608058608058, "grad_norm": 0.96484375, "learning_rate": 2.438131535815169e-05, "loss": 0.0974, "step": 33113 }, { "epoch": 24.25934065934066, "grad_norm": 0.9765625, "learning_rate": 2.4378404857800262e-05, "loss": 0.1318, "step": 33114 }, { "epoch": 24.26007326007326, "grad_norm": 1.25, "learning_rate": 2.4375494482605954e-05, "loss": 0.1212, "step": 33115 }, { "epoch": 24.26080586080586, "grad_norm": 1.1875, "learning_rate": 2.437258423258275e-05, "loss": 0.1277, "step": 33116 }, { "epoch": 24.26153846153846, "grad_norm": 1.6328125, "learning_rate": 2.4369674107744526e-05, "loss": 0.1019, "step": 33117 }, { "epoch": 24.26227106227106, "grad_norm": 0.9765625, "learning_rate": 2.4366764108105318e-05, "loss": 0.134, "step": 33118 }, { "epoch": 24.263003663003662, "grad_norm": 0.72265625, "learning_rate": 2.4363854233678995e-05, "loss": 0.1159, "step": 33119 }, { "epoch": 24.263736263736263, "grad_norm": 1.0546875, "learning_rate": 2.436094448447955e-05, "loss": 0.1097, "step": 33120 }, { "epoch": 24.264468864468864, "grad_norm": 0.86328125, "learning_rate": 2.4358034860520886e-05, "loss": 0.1335, "step": 33121 }, { "epoch": 24.265201465201464, "grad_norm": 1.5546875, "learning_rate": 2.4355125361816952e-05, "loss": 0.1351, "step": 33122 }, { "epoch": 24.265934065934065, "grad_norm": 1.46875, "learning_rate": 2.4352215988381697e-05, "loss": 0.1058, "step": 33123 }, { "epoch": 24.266666666666666, "grad_norm": 1.15625, "learning_rate": 2.4349306740229056e-05, "loss": 0.1247, "step": 33124 }, { "epoch": 24.267399267399266, "grad_norm": 0.88671875, "learning_rate": 2.434639761737299e-05, "loss": 0.1457, "step": 33125 }, { "epoch": 24.268131868131867, "grad_norm": 1.5625, "learning_rate": 2.434348861982741e-05, "loss": 0.1335, "step": 33126 }, { "epoch": 24.268864468864468, "grad_norm": 0.953125, "learning_rate": 2.434057974760627e-05, "loss": 0.1088, "step": 33127 }, { "epoch": 24.26959706959707, "grad_norm": 1.125, "learning_rate": 2.4337671000723468e-05, "loss": 0.1172, "step": 33128 }, { "epoch": 24.27032967032967, "grad_norm": 0.83203125, "learning_rate": 2.4334762379193004e-05, "loss": 0.1007, "step": 33129 }, { "epoch": 24.27106227106227, "grad_norm": 0.734375, "learning_rate": 2.4331853883028772e-05, "loss": 0.1188, "step": 33130 }, { "epoch": 24.27179487179487, "grad_norm": 1.1953125, "learning_rate": 2.432894551224474e-05, "loss": 0.104, "step": 33131 }, { "epoch": 24.27252747252747, "grad_norm": 1.265625, "learning_rate": 2.4326037266854794e-05, "loss": 0.1172, "step": 33132 }, { "epoch": 24.273260073260072, "grad_norm": 1.5234375, "learning_rate": 2.4323129146872898e-05, "loss": 0.1011, "step": 33133 }, { "epoch": 24.273992673992673, "grad_norm": 0.85546875, "learning_rate": 2.4320221152312986e-05, "loss": 0.1231, "step": 33134 }, { "epoch": 24.274725274725274, "grad_norm": 1.0234375, "learning_rate": 2.4317313283189007e-05, "loss": 0.1357, "step": 33135 }, { "epoch": 24.275457875457874, "grad_norm": 0.83984375, "learning_rate": 2.4314405539514853e-05, "loss": 0.1316, "step": 33136 }, { "epoch": 24.276190476190475, "grad_norm": 0.546875, "learning_rate": 2.4311497921304483e-05, "loss": 0.1151, "step": 33137 }, { "epoch": 24.276923076923076, "grad_norm": 0.94921875, "learning_rate": 2.4308590428571816e-05, "loss": 0.1166, "step": 33138 }, { "epoch": 24.277655677655677, "grad_norm": 1.6015625, "learning_rate": 2.4305683061330796e-05, "loss": 0.1298, "step": 33139 }, { "epoch": 24.278388278388277, "grad_norm": 1.6875, "learning_rate": 2.430277581959537e-05, "loss": 0.1166, "step": 33140 }, { "epoch": 24.279120879120878, "grad_norm": 0.8984375, "learning_rate": 2.429986870337942e-05, "loss": 0.1544, "step": 33141 }, { "epoch": 24.27985347985348, "grad_norm": 0.76953125, "learning_rate": 2.4296961712696918e-05, "loss": 0.1113, "step": 33142 }, { "epoch": 24.28058608058608, "grad_norm": 0.75390625, "learning_rate": 2.4294054847561746e-05, "loss": 0.1264, "step": 33143 }, { "epoch": 24.28131868131868, "grad_norm": 1.1953125, "learning_rate": 2.4291148107987888e-05, "loss": 0.1174, "step": 33144 }, { "epoch": 24.28205128205128, "grad_norm": 1.078125, "learning_rate": 2.428824149398923e-05, "loss": 0.1448, "step": 33145 }, { "epoch": 24.282783882783882, "grad_norm": 1.453125, "learning_rate": 2.4285335005579726e-05, "loss": 0.123, "step": 33146 }, { "epoch": 24.283516483516483, "grad_norm": 1.015625, "learning_rate": 2.428242864277327e-05, "loss": 0.1106, "step": 33147 }, { "epoch": 24.284249084249083, "grad_norm": 1.078125, "learning_rate": 2.427952240558381e-05, "loss": 0.149, "step": 33148 }, { "epoch": 24.284981684981684, "grad_norm": 0.9296875, "learning_rate": 2.4276616294025256e-05, "loss": 0.1053, "step": 33149 }, { "epoch": 24.285714285714285, "grad_norm": 1.1171875, "learning_rate": 2.427371030811155e-05, "loss": 0.1149, "step": 33150 }, { "epoch": 24.286446886446885, "grad_norm": 0.9140625, "learning_rate": 2.4270804447856625e-05, "loss": 0.1308, "step": 33151 }, { "epoch": 24.287179487179486, "grad_norm": 1.6953125, "learning_rate": 2.4267898713274366e-05, "loss": 0.1419, "step": 33152 }, { "epoch": 24.287912087912087, "grad_norm": 1.4921875, "learning_rate": 2.4264993104378738e-05, "loss": 0.1005, "step": 33153 }, { "epoch": 24.288644688644688, "grad_norm": 1.171875, "learning_rate": 2.42620876211836e-05, "loss": 0.1489, "step": 33154 }, { "epoch": 24.28937728937729, "grad_norm": 0.88671875, "learning_rate": 2.425918226370295e-05, "loss": 0.1242, "step": 33155 }, { "epoch": 24.29010989010989, "grad_norm": 0.76171875, "learning_rate": 2.4256277031950656e-05, "loss": 0.1094, "step": 33156 }, { "epoch": 24.29084249084249, "grad_norm": 0.91796875, "learning_rate": 2.425337192594067e-05, "loss": 0.1209, "step": 33157 }, { "epoch": 24.29157509157509, "grad_norm": 0.72265625, "learning_rate": 2.425046694568688e-05, "loss": 0.1018, "step": 33158 }, { "epoch": 24.29230769230769, "grad_norm": 0.98828125, "learning_rate": 2.4247562091203216e-05, "loss": 0.1073, "step": 33159 }, { "epoch": 24.293040293040292, "grad_norm": 1.46875, "learning_rate": 2.42446573625036e-05, "loss": 0.1465, "step": 33160 }, { "epoch": 24.293772893772893, "grad_norm": 1.296875, "learning_rate": 2.4241752759601953e-05, "loss": 0.1481, "step": 33161 }, { "epoch": 24.294505494505493, "grad_norm": 1.203125, "learning_rate": 2.4238848282512207e-05, "loss": 0.112, "step": 33162 }, { "epoch": 24.295238095238094, "grad_norm": 1.1484375, "learning_rate": 2.4235943931248236e-05, "loss": 0.116, "step": 33163 }, { "epoch": 24.295970695970695, "grad_norm": 1.3203125, "learning_rate": 2.4233039705823986e-05, "loss": 0.1121, "step": 33164 }, { "epoch": 24.296703296703296, "grad_norm": 0.9140625, "learning_rate": 2.4230135606253365e-05, "loss": 0.1313, "step": 33165 }, { "epoch": 24.297435897435896, "grad_norm": 1.3359375, "learning_rate": 2.4227231632550302e-05, "loss": 0.1295, "step": 33166 }, { "epoch": 24.298168498168497, "grad_norm": 1.09375, "learning_rate": 2.4224327784728685e-05, "loss": 0.1141, "step": 33167 }, { "epoch": 24.298901098901098, "grad_norm": 0.859375, "learning_rate": 2.4221424062802455e-05, "loss": 0.1209, "step": 33168 }, { "epoch": 24.2996336996337, "grad_norm": 0.84375, "learning_rate": 2.4218520466785468e-05, "loss": 0.1181, "step": 33169 }, { "epoch": 24.3003663003663, "grad_norm": 1.53125, "learning_rate": 2.421561699669172e-05, "loss": 0.1209, "step": 33170 }, { "epoch": 24.3010989010989, "grad_norm": 1.140625, "learning_rate": 2.4212713652535058e-05, "loss": 0.1155, "step": 33171 }, { "epoch": 24.3018315018315, "grad_norm": 0.86328125, "learning_rate": 2.420981043432943e-05, "loss": 0.1217, "step": 33172 }, { "epoch": 24.3025641025641, "grad_norm": 0.7265625, "learning_rate": 2.4206907342088716e-05, "loss": 0.124, "step": 33173 }, { "epoch": 24.303296703296702, "grad_norm": 1.109375, "learning_rate": 2.420400437582684e-05, "loss": 0.1333, "step": 33174 }, { "epoch": 24.304029304029303, "grad_norm": 0.8671875, "learning_rate": 2.420110153555771e-05, "loss": 0.1161, "step": 33175 }, { "epoch": 24.304761904761904, "grad_norm": 1.0234375, "learning_rate": 2.4198198821295244e-05, "loss": 0.1033, "step": 33176 }, { "epoch": 24.305494505494504, "grad_norm": 0.73828125, "learning_rate": 2.419529623305335e-05, "loss": 0.0995, "step": 33177 }, { "epoch": 24.306227106227105, "grad_norm": 1.0625, "learning_rate": 2.419239377084591e-05, "loss": 0.1334, "step": 33178 }, { "epoch": 24.306959706959706, "grad_norm": 1.28125, "learning_rate": 2.418949143468687e-05, "loss": 0.1437, "step": 33179 }, { "epoch": 24.307692307692307, "grad_norm": 1.3046875, "learning_rate": 2.4186589224590078e-05, "loss": 0.1338, "step": 33180 }, { "epoch": 24.308424908424907, "grad_norm": 0.86328125, "learning_rate": 2.4183687140569507e-05, "loss": 0.1117, "step": 33181 }, { "epoch": 24.309157509157508, "grad_norm": 0.640625, "learning_rate": 2.418078518263901e-05, "loss": 0.1347, "step": 33182 }, { "epoch": 24.30989010989011, "grad_norm": 0.8515625, "learning_rate": 2.4177883350812537e-05, "loss": 0.134, "step": 33183 }, { "epoch": 24.31062271062271, "grad_norm": 0.8046875, "learning_rate": 2.4174981645103945e-05, "loss": 0.1001, "step": 33184 }, { "epoch": 24.31135531135531, "grad_norm": 1.0234375, "learning_rate": 2.4172080065527154e-05, "loss": 0.1199, "step": 33185 }, { "epoch": 24.31208791208791, "grad_norm": 1.21875, "learning_rate": 2.4169178612096073e-05, "loss": 0.1349, "step": 33186 }, { "epoch": 24.31282051282051, "grad_norm": 0.9921875, "learning_rate": 2.41662772848246e-05, "loss": 0.1001, "step": 33187 }, { "epoch": 24.313553113553112, "grad_norm": 1.359375, "learning_rate": 2.4163376083726656e-05, "loss": 0.1237, "step": 33188 }, { "epoch": 24.314285714285713, "grad_norm": 1.125, "learning_rate": 2.416047500881611e-05, "loss": 0.125, "step": 33189 }, { "epoch": 24.315018315018314, "grad_norm": 1.7265625, "learning_rate": 2.415757406010687e-05, "loss": 0.1972, "step": 33190 }, { "epoch": 24.315750915750915, "grad_norm": 1.0703125, "learning_rate": 2.4154673237612837e-05, "loss": 0.1125, "step": 33191 }, { "epoch": 24.316483516483515, "grad_norm": 2.046875, "learning_rate": 2.4151772541347937e-05, "loss": 0.155, "step": 33192 }, { "epoch": 24.317216117216116, "grad_norm": 0.9375, "learning_rate": 2.414887197132602e-05, "loss": 0.1201, "step": 33193 }, { "epoch": 24.317948717948717, "grad_norm": 1.78125, "learning_rate": 2.414597152756103e-05, "loss": 0.1387, "step": 33194 }, { "epoch": 24.318681318681318, "grad_norm": 0.91015625, "learning_rate": 2.4143071210066807e-05, "loss": 0.1093, "step": 33195 }, { "epoch": 24.31941391941392, "grad_norm": 1.2109375, "learning_rate": 2.414017101885732e-05, "loss": 0.1363, "step": 33196 }, { "epoch": 24.32014652014652, "grad_norm": 1.03125, "learning_rate": 2.4137270953946403e-05, "loss": 0.1245, "step": 33197 }, { "epoch": 24.32087912087912, "grad_norm": 0.68359375, "learning_rate": 2.4134371015348e-05, "loss": 0.0985, "step": 33198 }, { "epoch": 24.32161172161172, "grad_norm": 0.87109375, "learning_rate": 2.4131471203075958e-05, "loss": 0.1072, "step": 33199 }, { "epoch": 24.32234432234432, "grad_norm": 0.75390625, "learning_rate": 2.4128571517144194e-05, "loss": 0.0937, "step": 33200 }, { "epoch": 24.323076923076922, "grad_norm": 0.8046875, "learning_rate": 2.4125671957566602e-05, "loss": 0.1238, "step": 33201 }, { "epoch": 24.323809523809523, "grad_norm": 1.171875, "learning_rate": 2.4122772524357073e-05, "loss": 0.1575, "step": 33202 }, { "epoch": 24.324542124542123, "grad_norm": 1.0234375, "learning_rate": 2.4119873217529515e-05, "loss": 0.1192, "step": 33203 }, { "epoch": 24.325274725274724, "grad_norm": 1.046875, "learning_rate": 2.4116974037097795e-05, "loss": 0.1196, "step": 33204 }, { "epoch": 24.326007326007325, "grad_norm": 1.09375, "learning_rate": 2.4114074983075823e-05, "loss": 0.1304, "step": 33205 }, { "epoch": 24.326739926739926, "grad_norm": 1.265625, "learning_rate": 2.4111176055477448e-05, "loss": 0.1245, "step": 33206 }, { "epoch": 24.327472527472526, "grad_norm": 0.70703125, "learning_rate": 2.4108277254316626e-05, "loss": 0.1203, "step": 33207 }, { "epoch": 24.328205128205127, "grad_norm": 1.5078125, "learning_rate": 2.4105378579607195e-05, "loss": 0.1574, "step": 33208 }, { "epoch": 24.328937728937728, "grad_norm": 1.046875, "learning_rate": 2.4102480031363076e-05, "loss": 0.1252, "step": 33209 }, { "epoch": 24.32967032967033, "grad_norm": 0.92578125, "learning_rate": 2.4099581609598127e-05, "loss": 0.0987, "step": 33210 }, { "epoch": 24.33040293040293, "grad_norm": 0.75, "learning_rate": 2.4096683314326242e-05, "loss": 0.1104, "step": 33211 }, { "epoch": 24.33113553113553, "grad_norm": 0.77734375, "learning_rate": 2.409378514556132e-05, "loss": 0.1132, "step": 33212 }, { "epoch": 24.33186813186813, "grad_norm": 1.6015625, "learning_rate": 2.409088710331724e-05, "loss": 0.1072, "step": 33213 }, { "epoch": 24.33260073260073, "grad_norm": 1.3203125, "learning_rate": 2.4087989187607908e-05, "loss": 0.1243, "step": 33214 }, { "epoch": 24.333333333333332, "grad_norm": 1.1796875, "learning_rate": 2.4085091398447174e-05, "loss": 0.1196, "step": 33215 }, { "epoch": 24.334065934065933, "grad_norm": 0.86328125, "learning_rate": 2.4082193735848938e-05, "loss": 0.1033, "step": 33216 }, { "epoch": 24.334798534798534, "grad_norm": 1.125, "learning_rate": 2.407929619982708e-05, "loss": 0.1386, "step": 33217 }, { "epoch": 24.335531135531134, "grad_norm": 1.9375, "learning_rate": 2.407639879039551e-05, "loss": 0.1322, "step": 33218 }, { "epoch": 24.336263736263735, "grad_norm": 1.2578125, "learning_rate": 2.4073501507568067e-05, "loss": 0.1159, "step": 33219 }, { "epoch": 24.336996336996336, "grad_norm": 1.078125, "learning_rate": 2.4070604351358672e-05, "loss": 0.0953, "step": 33220 }, { "epoch": 24.337728937728937, "grad_norm": 1.1328125, "learning_rate": 2.4067707321781155e-05, "loss": 0.0908, "step": 33221 }, { "epoch": 24.338461538461537, "grad_norm": 1.03125, "learning_rate": 2.4064810418849465e-05, "loss": 0.1102, "step": 33222 }, { "epoch": 24.339194139194138, "grad_norm": 1.1484375, "learning_rate": 2.4061913642577425e-05, "loss": 0.1115, "step": 33223 }, { "epoch": 24.33992673992674, "grad_norm": 1.0625, "learning_rate": 2.4059016992978952e-05, "loss": 0.1036, "step": 33224 }, { "epoch": 24.34065934065934, "grad_norm": 1.4921875, "learning_rate": 2.4056120470067896e-05, "loss": 0.161, "step": 33225 }, { "epoch": 24.34139194139194, "grad_norm": 1.0390625, "learning_rate": 2.4053224073858147e-05, "loss": 0.1209, "step": 33226 }, { "epoch": 24.34212454212454, "grad_norm": 0.76171875, "learning_rate": 2.4050327804363577e-05, "loss": 0.0942, "step": 33227 }, { "epoch": 24.34285714285714, "grad_norm": 1.0859375, "learning_rate": 2.404743166159808e-05, "loss": 0.1222, "step": 33228 }, { "epoch": 24.343589743589742, "grad_norm": 1.5546875, "learning_rate": 2.4044535645575528e-05, "loss": 0.1268, "step": 33229 }, { "epoch": 24.344322344322343, "grad_norm": 0.94140625, "learning_rate": 2.404163975630978e-05, "loss": 0.1102, "step": 33230 }, { "epoch": 24.345054945054944, "grad_norm": 0.75, "learning_rate": 2.4038743993814743e-05, "loss": 0.1138, "step": 33231 }, { "epoch": 24.345787545787545, "grad_norm": 1.1875, "learning_rate": 2.4035848358104228e-05, "loss": 0.1245, "step": 33232 }, { "epoch": 24.346520146520145, "grad_norm": 0.7265625, "learning_rate": 2.4032952849192187e-05, "loss": 0.1119, "step": 33233 }, { "epoch": 24.347252747252746, "grad_norm": 1.0546875, "learning_rate": 2.403005746709244e-05, "loss": 0.1, "step": 33234 }, { "epoch": 24.347985347985347, "grad_norm": 0.81640625, "learning_rate": 2.4027162211818897e-05, "loss": 0.1465, "step": 33235 }, { "epoch": 24.348717948717947, "grad_norm": 0.890625, "learning_rate": 2.4024267083385392e-05, "loss": 0.1231, "step": 33236 }, { "epoch": 24.349450549450548, "grad_norm": 0.96875, "learning_rate": 2.4021372081805812e-05, "loss": 0.1289, "step": 33237 }, { "epoch": 24.35018315018315, "grad_norm": 1.46875, "learning_rate": 2.4018477207094034e-05, "loss": 0.128, "step": 33238 }, { "epoch": 24.35091575091575, "grad_norm": 1.5078125, "learning_rate": 2.4015582459263925e-05, "loss": 0.1089, "step": 33239 }, { "epoch": 24.35164835164835, "grad_norm": 1.109375, "learning_rate": 2.401268783832937e-05, "loss": 0.1452, "step": 33240 }, { "epoch": 24.35238095238095, "grad_norm": 0.76953125, "learning_rate": 2.4009793344304205e-05, "loss": 0.1145, "step": 33241 }, { "epoch": 24.353113553113552, "grad_norm": 1.09375, "learning_rate": 2.4006898977202318e-05, "loss": 0.109, "step": 33242 }, { "epoch": 24.353846153846153, "grad_norm": 1.609375, "learning_rate": 2.4004004737037578e-05, "loss": 0.1061, "step": 33243 }, { "epoch": 24.354578754578753, "grad_norm": 1.578125, "learning_rate": 2.400111062382386e-05, "loss": 0.1414, "step": 33244 }, { "epoch": 24.355311355311354, "grad_norm": 0.9375, "learning_rate": 2.3998216637575012e-05, "loss": 0.1103, "step": 33245 }, { "epoch": 24.356043956043955, "grad_norm": 1.3515625, "learning_rate": 2.399532277830491e-05, "loss": 0.129, "step": 33246 }, { "epoch": 24.356776556776556, "grad_norm": 0.9609375, "learning_rate": 2.3992429046027394e-05, "loss": 0.1451, "step": 33247 }, { "epoch": 24.357509157509156, "grad_norm": 1.578125, "learning_rate": 2.398953544075639e-05, "loss": 0.1202, "step": 33248 }, { "epoch": 24.358241758241757, "grad_norm": 0.9296875, "learning_rate": 2.39866419625057e-05, "loss": 0.1102, "step": 33249 }, { "epoch": 24.358974358974358, "grad_norm": 1.5546875, "learning_rate": 2.398374861128923e-05, "loss": 0.1119, "step": 33250 }, { "epoch": 24.35970695970696, "grad_norm": 1.1015625, "learning_rate": 2.3980855387120813e-05, "loss": 0.1286, "step": 33251 }, { "epoch": 24.36043956043956, "grad_norm": 1.234375, "learning_rate": 2.397796229001432e-05, "loss": 0.1708, "step": 33252 }, { "epoch": 24.36117216117216, "grad_norm": 1.078125, "learning_rate": 2.3975069319983615e-05, "loss": 0.1169, "step": 33253 }, { "epoch": 24.36190476190476, "grad_norm": 0.82421875, "learning_rate": 2.3972176477042564e-05, "loss": 0.1099, "step": 33254 }, { "epoch": 24.36263736263736, "grad_norm": 0.74609375, "learning_rate": 2.396928376120504e-05, "loss": 0.0963, "step": 33255 }, { "epoch": 24.363369963369962, "grad_norm": 0.734375, "learning_rate": 2.396639117248487e-05, "loss": 0.1026, "step": 33256 }, { "epoch": 24.364102564102563, "grad_norm": 0.92578125, "learning_rate": 2.3963498710895948e-05, "loss": 0.1374, "step": 33257 }, { "epoch": 24.364835164835164, "grad_norm": 1.6796875, "learning_rate": 2.3960606376452076e-05, "loss": 0.1525, "step": 33258 }, { "epoch": 24.365567765567764, "grad_norm": 1.328125, "learning_rate": 2.395771416916719e-05, "loss": 0.1245, "step": 33259 }, { "epoch": 24.366300366300365, "grad_norm": 1.0078125, "learning_rate": 2.395482208905509e-05, "loss": 0.1311, "step": 33260 }, { "epoch": 24.367032967032966, "grad_norm": 0.65625, "learning_rate": 2.3951930136129672e-05, "loss": 0.0923, "step": 33261 }, { "epoch": 24.367765567765566, "grad_norm": 1.625, "learning_rate": 2.3949038310404752e-05, "loss": 0.1112, "step": 33262 }, { "epoch": 24.368498168498167, "grad_norm": 0.9140625, "learning_rate": 2.3946146611894202e-05, "loss": 0.1224, "step": 33263 }, { "epoch": 24.369230769230768, "grad_norm": 1.1640625, "learning_rate": 2.3943255040611888e-05, "loss": 0.1135, "step": 33264 }, { "epoch": 24.36996336996337, "grad_norm": 0.91796875, "learning_rate": 2.3940363596571648e-05, "loss": 0.1428, "step": 33265 }, { "epoch": 24.37069597069597, "grad_norm": 1.109375, "learning_rate": 2.393747227978737e-05, "loss": 0.1194, "step": 33266 }, { "epoch": 24.37142857142857, "grad_norm": 0.890625, "learning_rate": 2.3934581090272863e-05, "loss": 0.1176, "step": 33267 }, { "epoch": 24.37216117216117, "grad_norm": 1.015625, "learning_rate": 2.3931690028041994e-05, "loss": 0.1656, "step": 33268 }, { "epoch": 24.37289377289377, "grad_norm": 0.7109375, "learning_rate": 2.3928799093108627e-05, "loss": 0.095, "step": 33269 }, { "epoch": 24.373626373626372, "grad_norm": 1.546875, "learning_rate": 2.392590828548662e-05, "loss": 0.1334, "step": 33270 }, { "epoch": 24.374358974358973, "grad_norm": 1.140625, "learning_rate": 2.3923017605189792e-05, "loss": 0.1458, "step": 33271 }, { "epoch": 24.375091575091574, "grad_norm": 1.0234375, "learning_rate": 2.3920127052232035e-05, "loss": 0.1065, "step": 33272 }, { "epoch": 24.375824175824174, "grad_norm": 1.9140625, "learning_rate": 2.391723662662713e-05, "loss": 0.1425, "step": 33273 }, { "epoch": 24.376556776556775, "grad_norm": 0.984375, "learning_rate": 2.3914346328389014e-05, "loss": 0.1204, "step": 33274 }, { "epoch": 24.377289377289376, "grad_norm": 1.0390625, "learning_rate": 2.391145615753147e-05, "loss": 0.1356, "step": 33275 }, { "epoch": 24.378021978021977, "grad_norm": 1.2890625, "learning_rate": 2.390856611406837e-05, "loss": 0.126, "step": 33276 }, { "epoch": 24.378754578754577, "grad_norm": 0.98828125, "learning_rate": 2.3905676198013577e-05, "loss": 0.1309, "step": 33277 }, { "epoch": 24.379487179487178, "grad_norm": 1.2578125, "learning_rate": 2.39027864093809e-05, "loss": 0.1261, "step": 33278 }, { "epoch": 24.38021978021978, "grad_norm": 1.1484375, "learning_rate": 2.3899896748184203e-05, "loss": 0.1275, "step": 33279 }, { "epoch": 24.38095238095238, "grad_norm": 0.81640625, "learning_rate": 2.389700721443734e-05, "loss": 0.1222, "step": 33280 }, { "epoch": 24.38168498168498, "grad_norm": 0.83203125, "learning_rate": 2.389411780815416e-05, "loss": 0.1009, "step": 33281 }, { "epoch": 24.38241758241758, "grad_norm": 0.65625, "learning_rate": 2.3891228529348483e-05, "loss": 0.083, "step": 33282 }, { "epoch": 24.383150183150182, "grad_norm": 0.83984375, "learning_rate": 2.388833937803418e-05, "loss": 0.1232, "step": 33283 }, { "epoch": 24.383882783882783, "grad_norm": 0.93359375, "learning_rate": 2.388545035422504e-05, "loss": 0.1337, "step": 33284 }, { "epoch": 24.384615384615383, "grad_norm": 1.0703125, "learning_rate": 2.3882561457934987e-05, "loss": 0.118, "step": 33285 }, { "epoch": 24.385347985347984, "grad_norm": 0.93359375, "learning_rate": 2.3879672689177798e-05, "loss": 0.1109, "step": 33286 }, { "epoch": 24.386080586080585, "grad_norm": 1.546875, "learning_rate": 2.3876784047967355e-05, "loss": 0.1204, "step": 33287 }, { "epoch": 24.386813186813185, "grad_norm": 1.625, "learning_rate": 2.3873895534317458e-05, "loss": 0.1036, "step": 33288 }, { "epoch": 24.387545787545786, "grad_norm": 0.96875, "learning_rate": 2.387100714824198e-05, "loss": 0.1074, "step": 33289 }, { "epoch": 24.388278388278387, "grad_norm": 1.0234375, "learning_rate": 2.3868118889754733e-05, "loss": 0.1252, "step": 33290 }, { "epoch": 24.389010989010988, "grad_norm": 0.7109375, "learning_rate": 2.386523075886958e-05, "loss": 0.0854, "step": 33291 }, { "epoch": 24.38974358974359, "grad_norm": 1.46875, "learning_rate": 2.3862342755600367e-05, "loss": 0.1558, "step": 33292 }, { "epoch": 24.39047619047619, "grad_norm": 1.1328125, "learning_rate": 2.3859454879960897e-05, "loss": 0.1238, "step": 33293 }, { "epoch": 24.39120879120879, "grad_norm": 0.671875, "learning_rate": 2.385656713196502e-05, "loss": 0.1076, "step": 33294 }, { "epoch": 24.39194139194139, "grad_norm": 1.234375, "learning_rate": 2.385367951162658e-05, "loss": 0.1266, "step": 33295 }, { "epoch": 24.39267399267399, "grad_norm": 1.0078125, "learning_rate": 2.3850792018959427e-05, "loss": 0.096, "step": 33296 }, { "epoch": 24.393406593406592, "grad_norm": 1.0078125, "learning_rate": 2.384790465397736e-05, "loss": 0.1096, "step": 33297 }, { "epoch": 24.394139194139193, "grad_norm": 0.93359375, "learning_rate": 2.384501741669425e-05, "loss": 0.1185, "step": 33298 }, { "epoch": 24.394871794871793, "grad_norm": 1.265625, "learning_rate": 2.384213030712387e-05, "loss": 0.1259, "step": 33299 }, { "epoch": 24.395604395604394, "grad_norm": 1.53125, "learning_rate": 2.3839243325280135e-05, "loss": 0.1071, "step": 33300 }, { "epoch": 24.396336996336995, "grad_norm": 1.0, "learning_rate": 2.383635647117682e-05, "loss": 0.1192, "step": 33301 }, { "epoch": 24.397069597069596, "grad_norm": 1.125, "learning_rate": 2.3833469744827773e-05, "loss": 0.1533, "step": 33302 }, { "epoch": 24.397802197802196, "grad_norm": 0.9296875, "learning_rate": 2.3830583146246847e-05, "loss": 0.1203, "step": 33303 }, { "epoch": 24.398534798534797, "grad_norm": 1.0703125, "learning_rate": 2.3827696675447828e-05, "loss": 0.1319, "step": 33304 }, { "epoch": 24.399267399267398, "grad_norm": 0.8828125, "learning_rate": 2.382481033244458e-05, "loss": 0.1257, "step": 33305 }, { "epoch": 24.4, "grad_norm": 0.90625, "learning_rate": 2.3821924117250914e-05, "loss": 0.1294, "step": 33306 }, { "epoch": 24.4007326007326, "grad_norm": 1.2578125, "learning_rate": 2.3819038029880684e-05, "loss": 0.1217, "step": 33307 }, { "epoch": 24.4014652014652, "grad_norm": 1.015625, "learning_rate": 2.3816152070347685e-05, "loss": 0.1113, "step": 33308 }, { "epoch": 24.4021978021978, "grad_norm": 0.74609375, "learning_rate": 2.3813266238665785e-05, "loss": 0.1272, "step": 33309 }, { "epoch": 24.4029304029304, "grad_norm": 0.87109375, "learning_rate": 2.3810380534848746e-05, "loss": 0.0954, "step": 33310 }, { "epoch": 24.403663003663002, "grad_norm": 1.0625, "learning_rate": 2.3807494958910474e-05, "loss": 0.1194, "step": 33311 }, { "epoch": 24.404395604395603, "grad_norm": 0.80078125, "learning_rate": 2.380460951086474e-05, "loss": 0.1157, "step": 33312 }, { "epoch": 24.405128205128204, "grad_norm": 1.3671875, "learning_rate": 2.3801724190725404e-05, "loss": 0.1661, "step": 33313 }, { "epoch": 24.405860805860804, "grad_norm": 1.1953125, "learning_rate": 2.3798838998506247e-05, "loss": 0.1479, "step": 33314 }, { "epoch": 24.406593406593405, "grad_norm": 1.078125, "learning_rate": 2.379595393422112e-05, "loss": 0.1197, "step": 33315 }, { "epoch": 24.407326007326006, "grad_norm": 0.7109375, "learning_rate": 2.3793068997883847e-05, "loss": 0.1097, "step": 33316 }, { "epoch": 24.408058608058607, "grad_norm": 0.94921875, "learning_rate": 2.379018418950825e-05, "loss": 0.1241, "step": 33317 }, { "epoch": 24.408791208791207, "grad_norm": 1.171875, "learning_rate": 2.378729950910816e-05, "loss": 0.1275, "step": 33318 }, { "epoch": 24.409523809523808, "grad_norm": 0.6328125, "learning_rate": 2.3784414956697375e-05, "loss": 0.1202, "step": 33319 }, { "epoch": 24.41025641025641, "grad_norm": 1.328125, "learning_rate": 2.3781530532289725e-05, "loss": 0.1143, "step": 33320 }, { "epoch": 24.41098901098901, "grad_norm": 1.15625, "learning_rate": 2.377864623589903e-05, "loss": 0.1376, "step": 33321 }, { "epoch": 24.41172161172161, "grad_norm": 1.2578125, "learning_rate": 2.3775762067539137e-05, "loss": 0.147, "step": 33322 }, { "epoch": 24.41245421245421, "grad_norm": 1.0703125, "learning_rate": 2.377287802722382e-05, "loss": 0.1297, "step": 33323 }, { "epoch": 24.41318681318681, "grad_norm": 0.75390625, "learning_rate": 2.376999411496694e-05, "loss": 0.1014, "step": 33324 }, { "epoch": 24.413919413919412, "grad_norm": 1.0390625, "learning_rate": 2.3767110330782255e-05, "loss": 0.1523, "step": 33325 }, { "epoch": 24.414652014652013, "grad_norm": 0.94921875, "learning_rate": 2.3764226674683653e-05, "loss": 0.1271, "step": 33326 }, { "epoch": 24.415384615384614, "grad_norm": 1.4765625, "learning_rate": 2.3761343146684907e-05, "loss": 0.1286, "step": 33327 }, { "epoch": 24.416117216117215, "grad_norm": 0.81640625, "learning_rate": 2.3758459746799836e-05, "loss": 0.1041, "step": 33328 }, { "epoch": 24.416849816849815, "grad_norm": 1.015625, "learning_rate": 2.3755576475042282e-05, "loss": 0.122, "step": 33329 }, { "epoch": 24.417582417582416, "grad_norm": 1.3046875, "learning_rate": 2.3752693331426026e-05, "loss": 0.1539, "step": 33330 }, { "epoch": 24.418315018315017, "grad_norm": 1.609375, "learning_rate": 2.3749810315964897e-05, "loss": 0.1329, "step": 33331 }, { "epoch": 24.419047619047618, "grad_norm": 0.91796875, "learning_rate": 2.3746927428672707e-05, "loss": 0.127, "step": 33332 }, { "epoch": 24.41978021978022, "grad_norm": 1.171875, "learning_rate": 2.3744044669563284e-05, "loss": 0.1467, "step": 33333 }, { "epoch": 24.42051282051282, "grad_norm": 0.875, "learning_rate": 2.3741162038650413e-05, "loss": 0.1045, "step": 33334 }, { "epoch": 24.42124542124542, "grad_norm": 1.09375, "learning_rate": 2.373827953594794e-05, "loss": 0.1171, "step": 33335 }, { "epoch": 24.42197802197802, "grad_norm": 1.28125, "learning_rate": 2.373539716146961e-05, "loss": 0.1192, "step": 33336 }, { "epoch": 24.42271062271062, "grad_norm": 1.1640625, "learning_rate": 2.3732514915229323e-05, "loss": 0.1218, "step": 33337 }, { "epoch": 24.423443223443222, "grad_norm": 1.3125, "learning_rate": 2.372963279724082e-05, "loss": 0.1155, "step": 33338 }, { "epoch": 24.424175824175823, "grad_norm": 1.09375, "learning_rate": 2.372675080751796e-05, "loss": 0.1194, "step": 33339 }, { "epoch": 24.424908424908423, "grad_norm": 1.1796875, "learning_rate": 2.3723868946074505e-05, "loss": 0.1375, "step": 33340 }, { "epoch": 24.425641025641024, "grad_norm": 1.1484375, "learning_rate": 2.3720987212924283e-05, "loss": 0.1197, "step": 33341 }, { "epoch": 24.426373626373625, "grad_norm": 1.3046875, "learning_rate": 2.3718105608081097e-05, "loss": 0.1259, "step": 33342 }, { "epoch": 24.427106227106226, "grad_norm": 1.2890625, "learning_rate": 2.3715224131558767e-05, "loss": 0.1178, "step": 33343 }, { "epoch": 24.427838827838826, "grad_norm": 1.34375, "learning_rate": 2.37123427833711e-05, "loss": 0.1017, "step": 33344 }, { "epoch": 24.428571428571427, "grad_norm": 1.0234375, "learning_rate": 2.370946156353188e-05, "loss": 0.0978, "step": 33345 }, { "epoch": 24.429304029304028, "grad_norm": 0.77734375, "learning_rate": 2.3706580472054927e-05, "loss": 0.1151, "step": 33346 }, { "epoch": 24.43003663003663, "grad_norm": 0.58203125, "learning_rate": 2.370369950895404e-05, "loss": 0.1018, "step": 33347 }, { "epoch": 24.43076923076923, "grad_norm": 1.1171875, "learning_rate": 2.3700818674243045e-05, "loss": 0.1152, "step": 33348 }, { "epoch": 24.43150183150183, "grad_norm": 1.1953125, "learning_rate": 2.3697937967935703e-05, "loss": 0.1243, "step": 33349 }, { "epoch": 24.43223443223443, "grad_norm": 0.69140625, "learning_rate": 2.3695057390045864e-05, "loss": 0.0958, "step": 33350 }, { "epoch": 24.43296703296703, "grad_norm": 0.765625, "learning_rate": 2.369217694058726e-05, "loss": 0.1392, "step": 33351 }, { "epoch": 24.433699633699632, "grad_norm": 1.0, "learning_rate": 2.3689296619573782e-05, "loss": 0.1068, "step": 33352 }, { "epoch": 24.434432234432233, "grad_norm": 0.96875, "learning_rate": 2.368641642701916e-05, "loss": 0.1037, "step": 33353 }, { "epoch": 24.435164835164834, "grad_norm": 0.97265625, "learning_rate": 2.3683536362937222e-05, "loss": 0.1109, "step": 33354 }, { "epoch": 24.435897435897434, "grad_norm": 0.890625, "learning_rate": 2.368065642734178e-05, "loss": 0.1091, "step": 33355 }, { "epoch": 24.436630036630035, "grad_norm": 1.8984375, "learning_rate": 2.36777766202466e-05, "loss": 0.2082, "step": 33356 }, { "epoch": 24.437362637362636, "grad_norm": 1.28125, "learning_rate": 2.36748969416655e-05, "loss": 0.1363, "step": 33357 }, { "epoch": 24.438095238095237, "grad_norm": 1.0546875, "learning_rate": 2.3672017391612276e-05, "loss": 0.131, "step": 33358 }, { "epoch": 24.438827838827837, "grad_norm": 0.89453125, "learning_rate": 2.366913797010074e-05, "loss": 0.1079, "step": 33359 }, { "epoch": 24.439560439560438, "grad_norm": 0.8984375, "learning_rate": 2.366625867714465e-05, "loss": 0.1271, "step": 33360 }, { "epoch": 24.44029304029304, "grad_norm": 1.234375, "learning_rate": 2.3663379512757846e-05, "loss": 0.1009, "step": 33361 }, { "epoch": 24.44102564102564, "grad_norm": 0.9140625, "learning_rate": 2.366050047695406e-05, "loss": 0.1125, "step": 33362 }, { "epoch": 24.44175824175824, "grad_norm": 1.0546875, "learning_rate": 2.365762156974717e-05, "loss": 0.1232, "step": 33363 }, { "epoch": 24.44249084249084, "grad_norm": 0.79296875, "learning_rate": 2.3654742791150905e-05, "loss": 0.1241, "step": 33364 }, { "epoch": 24.44322344322344, "grad_norm": 0.9921875, "learning_rate": 2.36518641411791e-05, "loss": 0.1289, "step": 33365 }, { "epoch": 24.443956043956042, "grad_norm": 0.69921875, "learning_rate": 2.3648985619845503e-05, "loss": 0.1028, "step": 33366 }, { "epoch": 24.444688644688643, "grad_norm": 0.7109375, "learning_rate": 2.3646107227163936e-05, "loss": 0.0938, "step": 33367 }, { "epoch": 24.445421245421244, "grad_norm": 1.1796875, "learning_rate": 2.3643228963148185e-05, "loss": 0.1202, "step": 33368 }, { "epoch": 24.446153846153845, "grad_norm": 2.1875, "learning_rate": 2.364035082781204e-05, "loss": 0.1715, "step": 33369 }, { "epoch": 24.446886446886445, "grad_norm": 1.1484375, "learning_rate": 2.363747282116931e-05, "loss": 0.1571, "step": 33370 }, { "epoch": 24.447619047619046, "grad_norm": 1.4453125, "learning_rate": 2.3634594943233748e-05, "loss": 0.1211, "step": 33371 }, { "epoch": 24.448351648351647, "grad_norm": 1.0234375, "learning_rate": 2.3631717194019164e-05, "loss": 0.1342, "step": 33372 }, { "epoch": 24.449084249084247, "grad_norm": 1.3046875, "learning_rate": 2.3628839573539334e-05, "loss": 0.1169, "step": 33373 }, { "epoch": 24.449816849816848, "grad_norm": 1.21875, "learning_rate": 2.362596208180808e-05, "loss": 0.0949, "step": 33374 }, { "epoch": 24.45054945054945, "grad_norm": 1.0078125, "learning_rate": 2.3623084718839143e-05, "loss": 0.1195, "step": 33375 }, { "epoch": 24.45128205128205, "grad_norm": 0.7890625, "learning_rate": 2.362020748464635e-05, "loss": 0.1063, "step": 33376 }, { "epoch": 24.45201465201465, "grad_norm": 0.64453125, "learning_rate": 2.361733037924343e-05, "loss": 0.1094, "step": 33377 }, { "epoch": 24.45274725274725, "grad_norm": 0.8984375, "learning_rate": 2.3614453402644244e-05, "loss": 0.1037, "step": 33378 }, { "epoch": 24.453479853479852, "grad_norm": 0.97265625, "learning_rate": 2.3611576554862516e-05, "loss": 0.139, "step": 33379 }, { "epoch": 24.454212454212453, "grad_norm": 1.1484375, "learning_rate": 2.3608699835912056e-05, "loss": 0.1396, "step": 33380 }, { "epoch": 24.454945054945053, "grad_norm": 0.83203125, "learning_rate": 2.3605823245806654e-05, "loss": 0.1236, "step": 33381 }, { "epoch": 24.455677655677654, "grad_norm": 1.0234375, "learning_rate": 2.3602946784560066e-05, "loss": 0.1458, "step": 33382 }, { "epoch": 24.456410256410255, "grad_norm": 0.63671875, "learning_rate": 2.3600070452186092e-05, "loss": 0.1081, "step": 33383 }, { "epoch": 24.457142857142856, "grad_norm": 1.46875, "learning_rate": 2.359719424869851e-05, "loss": 0.1276, "step": 33384 }, { "epoch": 24.457875457875456, "grad_norm": 1.21875, "learning_rate": 2.359431817411112e-05, "loss": 0.0975, "step": 33385 }, { "epoch": 24.458608058608057, "grad_norm": 1.0, "learning_rate": 2.3591442228437665e-05, "loss": 0.1034, "step": 33386 }, { "epoch": 24.459340659340658, "grad_norm": 0.84765625, "learning_rate": 2.358856641169197e-05, "loss": 0.1294, "step": 33387 }, { "epoch": 24.46007326007326, "grad_norm": 0.85546875, "learning_rate": 2.358569072388774e-05, "loss": 0.1152, "step": 33388 }, { "epoch": 24.46080586080586, "grad_norm": 1.109375, "learning_rate": 2.3582815165038845e-05, "loss": 0.1062, "step": 33389 }, { "epoch": 24.46153846153846, "grad_norm": 1.0078125, "learning_rate": 2.3579939735158998e-05, "loss": 0.1291, "step": 33390 }, { "epoch": 24.46227106227106, "grad_norm": 1.671875, "learning_rate": 2.3577064434262015e-05, "loss": 0.1311, "step": 33391 }, { "epoch": 24.46300366300366, "grad_norm": 1.015625, "learning_rate": 2.3574189262361634e-05, "loss": 0.1397, "step": 33392 }, { "epoch": 24.463736263736262, "grad_norm": 1.484375, "learning_rate": 2.3571314219471655e-05, "loss": 0.1378, "step": 33393 }, { "epoch": 24.464468864468863, "grad_norm": 0.9375, "learning_rate": 2.3568439305605848e-05, "loss": 0.1166, "step": 33394 }, { "epoch": 24.465201465201464, "grad_norm": 1.7578125, "learning_rate": 2.356556452077799e-05, "loss": 0.1178, "step": 33395 }, { "epoch": 24.465934065934064, "grad_norm": 0.890625, "learning_rate": 2.356268986500187e-05, "loss": 0.1037, "step": 33396 }, { "epoch": 24.466666666666665, "grad_norm": 1.0703125, "learning_rate": 2.3559815338291232e-05, "loss": 0.105, "step": 33397 }, { "epoch": 24.467399267399266, "grad_norm": 1.4375, "learning_rate": 2.3556940940659864e-05, "loss": 0.1229, "step": 33398 }, { "epoch": 24.468131868131866, "grad_norm": 0.92578125, "learning_rate": 2.355406667212153e-05, "loss": 0.1299, "step": 33399 }, { "epoch": 24.468864468864467, "grad_norm": 1.171875, "learning_rate": 2.355119253269003e-05, "loss": 0.1001, "step": 33400 }, { "epoch": 24.469597069597068, "grad_norm": 1.1953125, "learning_rate": 2.3548318522379095e-05, "loss": 0.1312, "step": 33401 }, { "epoch": 24.47032967032967, "grad_norm": 0.83203125, "learning_rate": 2.354544464120253e-05, "loss": 0.1259, "step": 33402 }, { "epoch": 24.47106227106227, "grad_norm": 1.03125, "learning_rate": 2.3542570889174058e-05, "loss": 0.1368, "step": 33403 }, { "epoch": 24.47179487179487, "grad_norm": 1.453125, "learning_rate": 2.353969726630751e-05, "loss": 0.1289, "step": 33404 }, { "epoch": 24.47252747252747, "grad_norm": 1.0625, "learning_rate": 2.3536823772616604e-05, "loss": 0.0989, "step": 33405 }, { "epoch": 24.47326007326007, "grad_norm": 0.90234375, "learning_rate": 2.3533950408115126e-05, "loss": 0.1332, "step": 33406 }, { "epoch": 24.473992673992672, "grad_norm": 1.1484375, "learning_rate": 2.353107717281687e-05, "loss": 0.1309, "step": 33407 }, { "epoch": 24.474725274725273, "grad_norm": 1.3671875, "learning_rate": 2.352820406673555e-05, "loss": 0.1519, "step": 33408 }, { "epoch": 24.475457875457874, "grad_norm": 1.21875, "learning_rate": 2.3525331089884963e-05, "loss": 0.104, "step": 33409 }, { "epoch": 24.476190476190474, "grad_norm": 1.359375, "learning_rate": 2.3522458242278866e-05, "loss": 0.1141, "step": 33410 }, { "epoch": 24.476923076923075, "grad_norm": 0.84375, "learning_rate": 2.3519585523931047e-05, "loss": 0.1384, "step": 33411 }, { "epoch": 24.477655677655676, "grad_norm": 0.9375, "learning_rate": 2.351671293485524e-05, "loss": 0.1142, "step": 33412 }, { "epoch": 24.478388278388277, "grad_norm": 1.203125, "learning_rate": 2.351384047506523e-05, "loss": 0.1043, "step": 33413 }, { "epoch": 24.479120879120877, "grad_norm": 1.4375, "learning_rate": 2.3510968144574733e-05, "loss": 0.1303, "step": 33414 }, { "epoch": 24.479853479853478, "grad_norm": 0.86328125, "learning_rate": 2.350809594339758e-05, "loss": 0.1282, "step": 33415 }, { "epoch": 24.48058608058608, "grad_norm": 1.046875, "learning_rate": 2.350522387154749e-05, "loss": 0.1381, "step": 33416 }, { "epoch": 24.48131868131868, "grad_norm": 1.1328125, "learning_rate": 2.350235192903823e-05, "loss": 0.1424, "step": 33417 }, { "epoch": 24.48205128205128, "grad_norm": 1.265625, "learning_rate": 2.3499480115883583e-05, "loss": 0.1362, "step": 33418 }, { "epoch": 24.48278388278388, "grad_norm": 1.015625, "learning_rate": 2.3496608432097276e-05, "loss": 0.1275, "step": 33419 }, { "epoch": 24.483516483516482, "grad_norm": 1.0390625, "learning_rate": 2.349373687769308e-05, "loss": 0.1608, "step": 33420 }, { "epoch": 24.484249084249083, "grad_norm": 0.94921875, "learning_rate": 2.3490865452684754e-05, "loss": 0.1229, "step": 33421 }, { "epoch": 24.484981684981683, "grad_norm": 0.83203125, "learning_rate": 2.348799415708608e-05, "loss": 0.0933, "step": 33422 }, { "epoch": 24.485714285714284, "grad_norm": 0.8359375, "learning_rate": 2.3485122990910773e-05, "loss": 0.1001, "step": 33423 }, { "epoch": 24.486446886446885, "grad_norm": 1.8203125, "learning_rate": 2.348225195417261e-05, "loss": 0.1731, "step": 33424 }, { "epoch": 24.487179487179485, "grad_norm": 1.0859375, "learning_rate": 2.3479381046885353e-05, "loss": 0.1309, "step": 33425 }, { "epoch": 24.487912087912086, "grad_norm": 1.3203125, "learning_rate": 2.347651026906276e-05, "loss": 0.1316, "step": 33426 }, { "epoch": 24.488644688644687, "grad_norm": 0.84375, "learning_rate": 2.3473639620718566e-05, "loss": 0.1304, "step": 33427 }, { "epoch": 24.489377289377288, "grad_norm": 1.5859375, "learning_rate": 2.347076910186655e-05, "loss": 0.1319, "step": 33428 }, { "epoch": 24.49010989010989, "grad_norm": 1.0, "learning_rate": 2.3467898712520426e-05, "loss": 0.109, "step": 33429 }, { "epoch": 24.49084249084249, "grad_norm": 1.15625, "learning_rate": 2.3465028452694e-05, "loss": 0.1069, "step": 33430 }, { "epoch": 24.49157509157509, "grad_norm": 1.171875, "learning_rate": 2.3462158322400983e-05, "loss": 0.116, "step": 33431 }, { "epoch": 24.49230769230769, "grad_norm": 0.69921875, "learning_rate": 2.345928832165514e-05, "loss": 0.1096, "step": 33432 }, { "epoch": 24.493040293040295, "grad_norm": 1.765625, "learning_rate": 2.345641845047024e-05, "loss": 0.1174, "step": 33433 }, { "epoch": 24.493772893772892, "grad_norm": 0.859375, "learning_rate": 2.3453548708860003e-05, "loss": 0.1309, "step": 33434 }, { "epoch": 24.494505494505496, "grad_norm": 0.78515625, "learning_rate": 2.3450679096838192e-05, "loss": 0.0956, "step": 33435 }, { "epoch": 24.495238095238093, "grad_norm": 1.03125, "learning_rate": 2.344780961441856e-05, "loss": 0.1129, "step": 33436 }, { "epoch": 24.495970695970698, "grad_norm": 1.015625, "learning_rate": 2.3444940261614866e-05, "loss": 0.1279, "step": 33437 }, { "epoch": 24.496703296703295, "grad_norm": 1.5625, "learning_rate": 2.3442071038440826e-05, "loss": 0.1314, "step": 33438 }, { "epoch": 24.4974358974359, "grad_norm": 0.88671875, "learning_rate": 2.3439201944910227e-05, "loss": 0.1186, "step": 33439 }, { "epoch": 24.498168498168496, "grad_norm": 1.234375, "learning_rate": 2.3436332981036757e-05, "loss": 0.1152, "step": 33440 }, { "epoch": 24.4989010989011, "grad_norm": 0.71875, "learning_rate": 2.3433464146834242e-05, "loss": 0.1203, "step": 33441 }, { "epoch": 24.499633699633698, "grad_norm": 0.72265625, "learning_rate": 2.3430595442316365e-05, "loss": 0.1142, "step": 33442 }, { "epoch": 24.500366300366302, "grad_norm": 0.828125, "learning_rate": 2.3427726867496884e-05, "loss": 0.1085, "step": 33443 }, { "epoch": 24.5010989010989, "grad_norm": 0.7421875, "learning_rate": 2.3424858422389576e-05, "loss": 0.1177, "step": 33444 }, { "epoch": 24.501831501831504, "grad_norm": 1.46875, "learning_rate": 2.3421990107008142e-05, "loss": 0.1297, "step": 33445 }, { "epoch": 24.5025641025641, "grad_norm": 0.828125, "learning_rate": 2.3419121921366338e-05, "loss": 0.1084, "step": 33446 }, { "epoch": 24.503296703296705, "grad_norm": 1.3984375, "learning_rate": 2.3416253865477913e-05, "loss": 0.1259, "step": 33447 }, { "epoch": 24.504029304029302, "grad_norm": 0.93359375, "learning_rate": 2.3413385939356626e-05, "loss": 0.1375, "step": 33448 }, { "epoch": 24.504761904761907, "grad_norm": 0.75, "learning_rate": 2.3410518143016182e-05, "loss": 0.1034, "step": 33449 }, { "epoch": 24.505494505494504, "grad_norm": 0.90234375, "learning_rate": 2.3407650476470337e-05, "loss": 0.1146, "step": 33450 }, { "epoch": 24.506227106227108, "grad_norm": 0.921875, "learning_rate": 2.3404782939732833e-05, "loss": 0.1134, "step": 33451 }, { "epoch": 24.506959706959705, "grad_norm": 1.2265625, "learning_rate": 2.3401915532817424e-05, "loss": 0.1025, "step": 33452 }, { "epoch": 24.50769230769231, "grad_norm": 1.0390625, "learning_rate": 2.339904825573782e-05, "loss": 0.1324, "step": 33453 }, { "epoch": 24.508424908424907, "grad_norm": 1.1875, "learning_rate": 2.3396181108507783e-05, "loss": 0.1739, "step": 33454 }, { "epoch": 24.50915750915751, "grad_norm": 1.2109375, "learning_rate": 2.339331409114101e-05, "loss": 0.1181, "step": 33455 }, { "epoch": 24.509890109890108, "grad_norm": 1.125, "learning_rate": 2.3390447203651306e-05, "loss": 0.0989, "step": 33456 }, { "epoch": 24.510622710622712, "grad_norm": 1.171875, "learning_rate": 2.3387580446052342e-05, "loss": 0.1013, "step": 33457 }, { "epoch": 24.51135531135531, "grad_norm": 0.58203125, "learning_rate": 2.3384713818357886e-05, "loss": 0.1025, "step": 33458 }, { "epoch": 24.512087912087914, "grad_norm": 1.0, "learning_rate": 2.3381847320581688e-05, "loss": 0.1182, "step": 33459 }, { "epoch": 24.51282051282051, "grad_norm": 1.4296875, "learning_rate": 2.337898095273744e-05, "loss": 0.1351, "step": 33460 }, { "epoch": 24.513553113553115, "grad_norm": 1.53125, "learning_rate": 2.3376114714838897e-05, "loss": 0.1046, "step": 33461 }, { "epoch": 24.514285714285712, "grad_norm": 1.1953125, "learning_rate": 2.3373248606899794e-05, "loss": 0.1418, "step": 33462 }, { "epoch": 24.515018315018317, "grad_norm": 0.94140625, "learning_rate": 2.3370382628933883e-05, "loss": 0.1144, "step": 33463 }, { "epoch": 24.515750915750917, "grad_norm": 0.9921875, "learning_rate": 2.3367516780954853e-05, "loss": 0.1292, "step": 33464 }, { "epoch": 24.516483516483518, "grad_norm": 0.79296875, "learning_rate": 2.3364651062976475e-05, "loss": 0.1237, "step": 33465 }, { "epoch": 24.51721611721612, "grad_norm": 1.3515625, "learning_rate": 2.3361785475012424e-05, "loss": 0.1501, "step": 33466 }, { "epoch": 24.51794871794872, "grad_norm": 1.1328125, "learning_rate": 2.335892001707651e-05, "loss": 0.1261, "step": 33467 }, { "epoch": 24.51868131868132, "grad_norm": 0.875, "learning_rate": 2.3356054689182397e-05, "loss": 0.1147, "step": 33468 }, { "epoch": 24.51941391941392, "grad_norm": 1.4453125, "learning_rate": 2.335318949134384e-05, "loss": 0.1244, "step": 33469 }, { "epoch": 24.520146520146522, "grad_norm": 0.70703125, "learning_rate": 2.335032442357458e-05, "loss": 0.1454, "step": 33470 }, { "epoch": 24.520879120879123, "grad_norm": 1.1796875, "learning_rate": 2.3347459485888308e-05, "loss": 0.1114, "step": 33471 }, { "epoch": 24.521611721611723, "grad_norm": 0.7734375, "learning_rate": 2.3344594678298774e-05, "loss": 0.1135, "step": 33472 }, { "epoch": 24.522344322344324, "grad_norm": 1.3828125, "learning_rate": 2.33417300008197e-05, "loss": 0.1188, "step": 33473 }, { "epoch": 24.523076923076925, "grad_norm": 0.77734375, "learning_rate": 2.333886545346483e-05, "loss": 0.1228, "step": 33474 }, { "epoch": 24.523809523809526, "grad_norm": 0.93359375, "learning_rate": 2.3336001036247858e-05, "loss": 0.1, "step": 33475 }, { "epoch": 24.524542124542126, "grad_norm": 0.8671875, "learning_rate": 2.3333136749182526e-05, "loss": 0.1151, "step": 33476 }, { "epoch": 24.525274725274727, "grad_norm": 0.671875, "learning_rate": 2.3330272592282546e-05, "loss": 0.1011, "step": 33477 }, { "epoch": 24.526007326007328, "grad_norm": 1.0625, "learning_rate": 2.332740856556167e-05, "loss": 0.1335, "step": 33478 }, { "epoch": 24.52673992673993, "grad_norm": 0.9765625, "learning_rate": 2.332454466903358e-05, "loss": 0.139, "step": 33479 }, { "epoch": 24.52747252747253, "grad_norm": 1.1796875, "learning_rate": 2.3321680902712037e-05, "loss": 0.1435, "step": 33480 }, { "epoch": 24.52820512820513, "grad_norm": 1.5703125, "learning_rate": 2.3318817266610706e-05, "loss": 0.1349, "step": 33481 }, { "epoch": 24.52893772893773, "grad_norm": 0.8828125, "learning_rate": 2.3315953760743387e-05, "loss": 0.1145, "step": 33482 }, { "epoch": 24.52967032967033, "grad_norm": 1.3046875, "learning_rate": 2.331309038512373e-05, "loss": 0.1012, "step": 33483 }, { "epoch": 24.530402930402932, "grad_norm": 1.2734375, "learning_rate": 2.3310227139765488e-05, "loss": 0.1792, "step": 33484 }, { "epoch": 24.531135531135533, "grad_norm": 1.3671875, "learning_rate": 2.330736402468239e-05, "loss": 0.1218, "step": 33485 }, { "epoch": 24.531868131868134, "grad_norm": 0.8046875, "learning_rate": 2.330450103988812e-05, "loss": 0.1267, "step": 33486 }, { "epoch": 24.532600732600734, "grad_norm": 1.6015625, "learning_rate": 2.330163818539641e-05, "loss": 0.1228, "step": 33487 }, { "epoch": 24.533333333333335, "grad_norm": 1.0078125, "learning_rate": 2.3298775461220983e-05, "loss": 0.1236, "step": 33488 }, { "epoch": 24.534065934065936, "grad_norm": 0.984375, "learning_rate": 2.3295912867375573e-05, "loss": 0.1007, "step": 33489 }, { "epoch": 24.534798534798536, "grad_norm": 1.3203125, "learning_rate": 2.329305040387385e-05, "loss": 0.1308, "step": 33490 }, { "epoch": 24.535531135531137, "grad_norm": 0.9765625, "learning_rate": 2.3290188070729575e-05, "loss": 0.1354, "step": 33491 }, { "epoch": 24.536263736263738, "grad_norm": 1.046875, "learning_rate": 2.3287325867956407e-05, "loss": 0.1364, "step": 33492 }, { "epoch": 24.53699633699634, "grad_norm": 1.5, "learning_rate": 2.3284463795568133e-05, "loss": 0.1456, "step": 33493 }, { "epoch": 24.53772893772894, "grad_norm": 1.3359375, "learning_rate": 2.3281601853578405e-05, "loss": 0.1087, "step": 33494 }, { "epoch": 24.53846153846154, "grad_norm": 0.8046875, "learning_rate": 2.3278740042000956e-05, "loss": 0.1032, "step": 33495 }, { "epoch": 24.53919413919414, "grad_norm": 1.5390625, "learning_rate": 2.3275878360849522e-05, "loss": 0.1352, "step": 33496 }, { "epoch": 24.53992673992674, "grad_norm": 1.125, "learning_rate": 2.327301681013777e-05, "loss": 0.1399, "step": 33497 }, { "epoch": 24.540659340659342, "grad_norm": 1.3046875, "learning_rate": 2.3270155389879433e-05, "loss": 0.1251, "step": 33498 }, { "epoch": 24.541391941391943, "grad_norm": 0.9765625, "learning_rate": 2.3267294100088225e-05, "loss": 0.1282, "step": 33499 }, { "epoch": 24.542124542124544, "grad_norm": 0.71484375, "learning_rate": 2.3264432940777863e-05, "loss": 0.1261, "step": 33500 }, { "epoch": 24.542857142857144, "grad_norm": 1.0703125, "learning_rate": 2.3261571911962033e-05, "loss": 0.1032, "step": 33501 }, { "epoch": 24.543589743589745, "grad_norm": 1.234375, "learning_rate": 2.3258711013654444e-05, "loss": 0.1073, "step": 33502 }, { "epoch": 24.544322344322346, "grad_norm": 1.3125, "learning_rate": 2.3255850245868815e-05, "loss": 0.139, "step": 33503 }, { "epoch": 24.545054945054947, "grad_norm": 0.65625, "learning_rate": 2.3252989608618877e-05, "loss": 0.1061, "step": 33504 }, { "epoch": 24.545787545787547, "grad_norm": 0.82421875, "learning_rate": 2.325012910191829e-05, "loss": 0.0991, "step": 33505 }, { "epoch": 24.546520146520148, "grad_norm": 0.99609375, "learning_rate": 2.3247268725780793e-05, "loss": 0.1097, "step": 33506 }, { "epoch": 24.54725274725275, "grad_norm": 1.015625, "learning_rate": 2.3244408480220043e-05, "loss": 0.1044, "step": 33507 }, { "epoch": 24.54798534798535, "grad_norm": 0.8359375, "learning_rate": 2.3241548365249818e-05, "loss": 0.1286, "step": 33508 }, { "epoch": 24.54871794871795, "grad_norm": 1.375, "learning_rate": 2.3238688380883766e-05, "loss": 0.147, "step": 33509 }, { "epoch": 24.54945054945055, "grad_norm": 1.109375, "learning_rate": 2.3235828527135605e-05, "loss": 0.1138, "step": 33510 }, { "epoch": 24.550183150183152, "grad_norm": 0.6875, "learning_rate": 2.3232968804019057e-05, "loss": 0.1163, "step": 33511 }, { "epoch": 24.550915750915753, "grad_norm": 0.8671875, "learning_rate": 2.323010921154779e-05, "loss": 0.1199, "step": 33512 }, { "epoch": 24.551648351648353, "grad_norm": 1.28125, "learning_rate": 2.3227249749735527e-05, "loss": 0.1267, "step": 33513 }, { "epoch": 24.552380952380954, "grad_norm": 0.953125, "learning_rate": 2.3224390418595954e-05, "loss": 0.1252, "step": 33514 }, { "epoch": 24.553113553113555, "grad_norm": 0.68359375, "learning_rate": 2.3221531218142805e-05, "loss": 0.1199, "step": 33515 }, { "epoch": 24.553846153846155, "grad_norm": 0.71875, "learning_rate": 2.3218672148389732e-05, "loss": 0.1188, "step": 33516 }, { "epoch": 24.554578754578756, "grad_norm": 0.91015625, "learning_rate": 2.3215813209350482e-05, "loss": 0.1093, "step": 33517 }, { "epoch": 24.555311355311357, "grad_norm": 1.7578125, "learning_rate": 2.321295440103868e-05, "loss": 0.1326, "step": 33518 }, { "epoch": 24.556043956043958, "grad_norm": 0.69921875, "learning_rate": 2.321009572346812e-05, "loss": 0.0835, "step": 33519 }, { "epoch": 24.55677655677656, "grad_norm": 0.984375, "learning_rate": 2.3207237176652426e-05, "loss": 0.1284, "step": 33520 }, { "epoch": 24.55750915750916, "grad_norm": 1.5546875, "learning_rate": 2.3204378760605318e-05, "loss": 0.1241, "step": 33521 }, { "epoch": 24.55824175824176, "grad_norm": 0.7109375, "learning_rate": 2.3201520475340507e-05, "loss": 0.1306, "step": 33522 }, { "epoch": 24.55897435897436, "grad_norm": 1.15625, "learning_rate": 2.3198662320871644e-05, "loss": 0.1389, "step": 33523 }, { "epoch": 24.55970695970696, "grad_norm": 0.57421875, "learning_rate": 2.3195804297212487e-05, "loss": 0.0908, "step": 33524 }, { "epoch": 24.560439560439562, "grad_norm": 0.9609375, "learning_rate": 2.3192946404376675e-05, "loss": 0.1038, "step": 33525 }, { "epoch": 24.561172161172163, "grad_norm": 1.2734375, "learning_rate": 2.319008864237794e-05, "loss": 0.1149, "step": 33526 }, { "epoch": 24.561904761904763, "grad_norm": 0.796875, "learning_rate": 2.3187231011229938e-05, "loss": 0.1222, "step": 33527 }, { "epoch": 24.562637362637364, "grad_norm": 1.5, "learning_rate": 2.3184373510946368e-05, "loss": 0.1155, "step": 33528 }, { "epoch": 24.563369963369965, "grad_norm": 1.1171875, "learning_rate": 2.3181516141540942e-05, "loss": 0.1194, "step": 33529 }, { "epoch": 24.564102564102566, "grad_norm": 1.0, "learning_rate": 2.317865890302736e-05, "loss": 0.1394, "step": 33530 }, { "epoch": 24.564835164835166, "grad_norm": 1.3515625, "learning_rate": 2.317580179541927e-05, "loss": 0.1172, "step": 33531 }, { "epoch": 24.565567765567767, "grad_norm": 1.140625, "learning_rate": 2.3172944818730385e-05, "loss": 0.0923, "step": 33532 }, { "epoch": 24.566300366300368, "grad_norm": 0.97265625, "learning_rate": 2.3170087972974415e-05, "loss": 0.1041, "step": 33533 }, { "epoch": 24.56703296703297, "grad_norm": 0.84765625, "learning_rate": 2.3167231258165e-05, "loss": 0.1073, "step": 33534 }, { "epoch": 24.56776556776557, "grad_norm": 0.6953125, "learning_rate": 2.3164374674315852e-05, "loss": 0.1073, "step": 33535 }, { "epoch": 24.56849816849817, "grad_norm": 1.3125, "learning_rate": 2.316151822144066e-05, "loss": 0.135, "step": 33536 }, { "epoch": 24.56923076923077, "grad_norm": 1.0390625, "learning_rate": 2.3158661899553133e-05, "loss": 0.1238, "step": 33537 }, { "epoch": 24.56996336996337, "grad_norm": 1.0546875, "learning_rate": 2.315580570866691e-05, "loss": 0.186, "step": 33538 }, { "epoch": 24.570695970695972, "grad_norm": 1.40625, "learning_rate": 2.3152949648795698e-05, "loss": 0.1423, "step": 33539 }, { "epoch": 24.571428571428573, "grad_norm": 1.0703125, "learning_rate": 2.3150093719953178e-05, "loss": 0.1344, "step": 33540 }, { "epoch": 24.572161172161174, "grad_norm": 0.87109375, "learning_rate": 2.314723792215306e-05, "loss": 0.1187, "step": 33541 }, { "epoch": 24.572893772893774, "grad_norm": 0.84375, "learning_rate": 2.3144382255408988e-05, "loss": 0.1389, "step": 33542 }, { "epoch": 24.573626373626375, "grad_norm": 0.7734375, "learning_rate": 2.3141526719734676e-05, "loss": 0.1193, "step": 33543 }, { "epoch": 24.574358974358976, "grad_norm": 1.0234375, "learning_rate": 2.3138671315143754e-05, "loss": 0.136, "step": 33544 }, { "epoch": 24.575091575091577, "grad_norm": 0.8984375, "learning_rate": 2.3135816041649976e-05, "loss": 0.1378, "step": 33545 }, { "epoch": 24.575824175824177, "grad_norm": 1.328125, "learning_rate": 2.3132960899266973e-05, "loss": 0.1239, "step": 33546 }, { "epoch": 24.576556776556778, "grad_norm": 1.0703125, "learning_rate": 2.313010588800843e-05, "loss": 0.1177, "step": 33547 }, { "epoch": 24.57728937728938, "grad_norm": 1.40625, "learning_rate": 2.3127251007888066e-05, "loss": 0.1377, "step": 33548 }, { "epoch": 24.57802197802198, "grad_norm": 0.79296875, "learning_rate": 2.3124396258919477e-05, "loss": 0.0839, "step": 33549 }, { "epoch": 24.57875457875458, "grad_norm": 1.1328125, "learning_rate": 2.3121541641116437e-05, "loss": 0.1138, "step": 33550 }, { "epoch": 24.57948717948718, "grad_norm": 1.5859375, "learning_rate": 2.311868715449256e-05, "loss": 0.128, "step": 33551 }, { "epoch": 24.58021978021978, "grad_norm": 1.53125, "learning_rate": 2.3115832799061555e-05, "loss": 0.1199, "step": 33552 }, { "epoch": 24.580952380952382, "grad_norm": 1.671875, "learning_rate": 2.311297857483707e-05, "loss": 0.1044, "step": 33553 }, { "epoch": 24.581684981684983, "grad_norm": 1.8984375, "learning_rate": 2.3110124481832798e-05, "loss": 0.1462, "step": 33554 }, { "epoch": 24.582417582417584, "grad_norm": 0.66796875, "learning_rate": 2.3107270520062406e-05, "loss": 0.0944, "step": 33555 }, { "epoch": 24.583150183150185, "grad_norm": 1.046875, "learning_rate": 2.3104416689539597e-05, "loss": 0.1362, "step": 33556 }, { "epoch": 24.583882783882785, "grad_norm": 1.0390625, "learning_rate": 2.3101562990278005e-05, "loss": 0.1244, "step": 33557 }, { "epoch": 24.584615384615386, "grad_norm": 0.8359375, "learning_rate": 2.309870942229131e-05, "loss": 0.1042, "step": 33558 }, { "epoch": 24.585347985347987, "grad_norm": 0.90234375, "learning_rate": 2.3095855985593216e-05, "loss": 0.1288, "step": 33559 }, { "epoch": 24.586080586080588, "grad_norm": 0.73046875, "learning_rate": 2.309300268019735e-05, "loss": 0.1182, "step": 33560 }, { "epoch": 24.58681318681319, "grad_norm": 1.0625, "learning_rate": 2.3090149506117414e-05, "loss": 0.1086, "step": 33561 }, { "epoch": 24.58754578754579, "grad_norm": 1.1328125, "learning_rate": 2.3087296463367064e-05, "loss": 0.1314, "step": 33562 }, { "epoch": 24.58827838827839, "grad_norm": 0.9609375, "learning_rate": 2.3084443551959994e-05, "loss": 0.1184, "step": 33563 }, { "epoch": 24.58901098901099, "grad_norm": 1.203125, "learning_rate": 2.3081590771909834e-05, "loss": 0.151, "step": 33564 }, { "epoch": 24.58974358974359, "grad_norm": 1.203125, "learning_rate": 2.3078738123230277e-05, "loss": 0.0973, "step": 33565 }, { "epoch": 24.590476190476192, "grad_norm": 0.875, "learning_rate": 2.3075885605934986e-05, "loss": 0.1127, "step": 33566 }, { "epoch": 24.591208791208793, "grad_norm": 1.1640625, "learning_rate": 2.307303322003765e-05, "loss": 0.135, "step": 33567 }, { "epoch": 24.591941391941393, "grad_norm": 0.98828125, "learning_rate": 2.3070180965551898e-05, "loss": 0.115, "step": 33568 }, { "epoch": 24.592673992673994, "grad_norm": 0.828125, "learning_rate": 2.3067328842491426e-05, "loss": 0.1332, "step": 33569 }, { "epoch": 24.593406593406595, "grad_norm": 1.015625, "learning_rate": 2.306447685086986e-05, "loss": 0.1006, "step": 33570 }, { "epoch": 24.594139194139196, "grad_norm": 1.015625, "learning_rate": 2.306162499070092e-05, "loss": 0.0983, "step": 33571 }, { "epoch": 24.594871794871796, "grad_norm": 1.4921875, "learning_rate": 2.3058773261998227e-05, "loss": 0.1443, "step": 33572 }, { "epoch": 24.595604395604397, "grad_norm": 1.4609375, "learning_rate": 2.305592166477546e-05, "loss": 0.1141, "step": 33573 }, { "epoch": 24.596336996336998, "grad_norm": 1.1875, "learning_rate": 2.3053070199046304e-05, "loss": 0.1092, "step": 33574 }, { "epoch": 24.5970695970696, "grad_norm": 1.078125, "learning_rate": 2.3050218864824353e-05, "loss": 0.1434, "step": 33575 }, { "epoch": 24.5978021978022, "grad_norm": 1.03125, "learning_rate": 2.3047367662123356e-05, "loss": 0.139, "step": 33576 }, { "epoch": 24.5985347985348, "grad_norm": 0.92578125, "learning_rate": 2.3044516590956916e-05, "loss": 0.0935, "step": 33577 }, { "epoch": 24.5992673992674, "grad_norm": 1.828125, "learning_rate": 2.304166565133873e-05, "loss": 0.1536, "step": 33578 }, { "epoch": 24.6, "grad_norm": 1.09375, "learning_rate": 2.303881484328241e-05, "loss": 0.1195, "step": 33579 }, { "epoch": 24.600732600732602, "grad_norm": 0.828125, "learning_rate": 2.303596416680165e-05, "loss": 0.1028, "step": 33580 }, { "epoch": 24.601465201465203, "grad_norm": 1.5234375, "learning_rate": 2.3033113621910103e-05, "loss": 0.1389, "step": 33581 }, { "epoch": 24.602197802197804, "grad_norm": 0.79296875, "learning_rate": 2.303026320862144e-05, "loss": 0.1281, "step": 33582 }, { "epoch": 24.602930402930404, "grad_norm": 0.83984375, "learning_rate": 2.3027412926949288e-05, "loss": 0.0967, "step": 33583 }, { "epoch": 24.603663003663005, "grad_norm": 1.0234375, "learning_rate": 2.3024562776907322e-05, "loss": 0.1158, "step": 33584 }, { "epoch": 24.604395604395606, "grad_norm": 1.46875, "learning_rate": 2.3021712758509205e-05, "loss": 0.151, "step": 33585 }, { "epoch": 24.605128205128207, "grad_norm": 0.8671875, "learning_rate": 2.3018862871768573e-05, "loss": 0.0952, "step": 33586 }, { "epoch": 24.605860805860807, "grad_norm": 0.75390625, "learning_rate": 2.301601311669909e-05, "loss": 0.144, "step": 33587 }, { "epoch": 24.606593406593408, "grad_norm": 1.09375, "learning_rate": 2.30131634933144e-05, "loss": 0.1156, "step": 33588 }, { "epoch": 24.60732600732601, "grad_norm": 1.09375, "learning_rate": 2.3010314001628198e-05, "loss": 0.1223, "step": 33589 }, { "epoch": 24.60805860805861, "grad_norm": 1.1015625, "learning_rate": 2.300746464165408e-05, "loss": 0.1159, "step": 33590 }, { "epoch": 24.60879120879121, "grad_norm": 1.65625, "learning_rate": 2.3004615413405727e-05, "loss": 0.1306, "step": 33591 }, { "epoch": 24.60952380952381, "grad_norm": 0.98046875, "learning_rate": 2.300176631689679e-05, "loss": 0.0922, "step": 33592 }, { "epoch": 24.61025641025641, "grad_norm": 1.109375, "learning_rate": 2.2998917352140928e-05, "loss": 0.1376, "step": 33593 }, { "epoch": 24.610989010989012, "grad_norm": 1.4453125, "learning_rate": 2.2996068519151766e-05, "loss": 0.1277, "step": 33594 }, { "epoch": 24.611721611721613, "grad_norm": 1.1015625, "learning_rate": 2.2993219817942994e-05, "loss": 0.1446, "step": 33595 }, { "epoch": 24.612454212454214, "grad_norm": 0.80859375, "learning_rate": 2.299037124852819e-05, "loss": 0.1359, "step": 33596 }, { "epoch": 24.613186813186815, "grad_norm": 1.3203125, "learning_rate": 2.298752281092109e-05, "loss": 0.1334, "step": 33597 }, { "epoch": 24.613919413919415, "grad_norm": 0.66015625, "learning_rate": 2.298467450513528e-05, "loss": 0.096, "step": 33598 }, { "epoch": 24.614652014652016, "grad_norm": 1.4453125, "learning_rate": 2.298182633118443e-05, "loss": 0.1335, "step": 33599 }, { "epoch": 24.615384615384617, "grad_norm": 0.73046875, "learning_rate": 2.29789782890822e-05, "loss": 0.115, "step": 33600 }, { "epoch": 24.616117216117217, "grad_norm": 1.453125, "learning_rate": 2.2976130378842178e-05, "loss": 0.1374, "step": 33601 }, { "epoch": 24.616849816849818, "grad_norm": 1.25, "learning_rate": 2.2973282600478095e-05, "loss": 0.1333, "step": 33602 }, { "epoch": 24.61758241758242, "grad_norm": 1.3203125, "learning_rate": 2.297043495400353e-05, "loss": 0.1143, "step": 33603 }, { "epoch": 24.61831501831502, "grad_norm": 0.88671875, "learning_rate": 2.2967587439432163e-05, "loss": 0.1221, "step": 33604 }, { "epoch": 24.61904761904762, "grad_norm": 1.609375, "learning_rate": 2.296474005677761e-05, "loss": 0.1546, "step": 33605 }, { "epoch": 24.61978021978022, "grad_norm": 0.8046875, "learning_rate": 2.2961892806053526e-05, "loss": 0.1044, "step": 33606 }, { "epoch": 24.620512820512822, "grad_norm": 1.015625, "learning_rate": 2.2959045687273556e-05, "loss": 0.1072, "step": 33607 }, { "epoch": 24.621245421245423, "grad_norm": 1.03125, "learning_rate": 2.2956198700451355e-05, "loss": 0.1272, "step": 33608 }, { "epoch": 24.621978021978023, "grad_norm": 1.1484375, "learning_rate": 2.295335184560053e-05, "loss": 0.1323, "step": 33609 }, { "epoch": 24.622710622710624, "grad_norm": 0.90234375, "learning_rate": 2.2950505122734732e-05, "loss": 0.0948, "step": 33610 }, { "epoch": 24.623443223443225, "grad_norm": 1.1484375, "learning_rate": 2.2947658531867636e-05, "loss": 0.1265, "step": 33611 }, { "epoch": 24.624175824175826, "grad_norm": 0.74609375, "learning_rate": 2.2944812073012828e-05, "loss": 0.1028, "step": 33612 }, { "epoch": 24.624908424908426, "grad_norm": 0.98828125, "learning_rate": 2.2941965746183974e-05, "loss": 0.1227, "step": 33613 }, { "epoch": 24.625641025641027, "grad_norm": 0.8984375, "learning_rate": 2.2939119551394712e-05, "loss": 0.1145, "step": 33614 }, { "epoch": 24.626373626373628, "grad_norm": 1.0703125, "learning_rate": 2.293627348865869e-05, "loss": 0.1214, "step": 33615 }, { "epoch": 24.62710622710623, "grad_norm": 0.78125, "learning_rate": 2.2933427557989514e-05, "loss": 0.0886, "step": 33616 }, { "epoch": 24.62783882783883, "grad_norm": 1.0, "learning_rate": 2.2930581759400835e-05, "loss": 0.1288, "step": 33617 }, { "epoch": 24.62857142857143, "grad_norm": 1.40625, "learning_rate": 2.292773609290629e-05, "loss": 0.1111, "step": 33618 }, { "epoch": 24.62930402930403, "grad_norm": 0.9375, "learning_rate": 2.292489055851953e-05, "loss": 0.1549, "step": 33619 }, { "epoch": 24.63003663003663, "grad_norm": 0.85546875, "learning_rate": 2.292204515625415e-05, "loss": 0.1038, "step": 33620 }, { "epoch": 24.630769230769232, "grad_norm": 0.89453125, "learning_rate": 2.2919199886123826e-05, "loss": 0.1387, "step": 33621 }, { "epoch": 24.631501831501833, "grad_norm": 1.046875, "learning_rate": 2.291635474814214e-05, "loss": 0.1233, "step": 33622 }, { "epoch": 24.632234432234434, "grad_norm": 1.28125, "learning_rate": 2.291350974232278e-05, "loss": 0.1166, "step": 33623 }, { "epoch": 24.632967032967034, "grad_norm": 0.9140625, "learning_rate": 2.291066486867934e-05, "loss": 0.1105, "step": 33624 }, { "epoch": 24.633699633699635, "grad_norm": 1.015625, "learning_rate": 2.290782012722546e-05, "loss": 0.1085, "step": 33625 }, { "epoch": 24.634432234432236, "grad_norm": 1.2109375, "learning_rate": 2.290497551797479e-05, "loss": 0.1277, "step": 33626 }, { "epoch": 24.635164835164836, "grad_norm": 0.92578125, "learning_rate": 2.2902131040940897e-05, "loss": 0.1239, "step": 33627 }, { "epoch": 24.635897435897437, "grad_norm": 1.078125, "learning_rate": 2.28992866961375e-05, "loss": 0.1267, "step": 33628 }, { "epoch": 24.636630036630038, "grad_norm": 1.4609375, "learning_rate": 2.2896442483578155e-05, "loss": 0.114, "step": 33629 }, { "epoch": 24.63736263736264, "grad_norm": 1.0859375, "learning_rate": 2.2893598403276542e-05, "loss": 0.1226, "step": 33630 }, { "epoch": 24.63809523809524, "grad_norm": 0.85546875, "learning_rate": 2.2890754455246235e-05, "loss": 0.1394, "step": 33631 }, { "epoch": 24.63882783882784, "grad_norm": 1.15625, "learning_rate": 2.2887910639500895e-05, "loss": 0.1585, "step": 33632 }, { "epoch": 24.63956043956044, "grad_norm": 1.2734375, "learning_rate": 2.2885066956054137e-05, "loss": 0.1081, "step": 33633 }, { "epoch": 24.64029304029304, "grad_norm": 1.3984375, "learning_rate": 2.2882223404919603e-05, "loss": 0.1267, "step": 33634 }, { "epoch": 24.641025641025642, "grad_norm": 1.234375, "learning_rate": 2.2879379986110888e-05, "loss": 0.1243, "step": 33635 }, { "epoch": 24.641758241758243, "grad_norm": 1.4609375, "learning_rate": 2.2876536699641634e-05, "loss": 0.0932, "step": 33636 }, { "epoch": 24.642490842490844, "grad_norm": 0.7890625, "learning_rate": 2.2873693545525473e-05, "loss": 0.122, "step": 33637 }, { "epoch": 24.643223443223444, "grad_norm": 1.7421875, "learning_rate": 2.2870850523776e-05, "loss": 0.1247, "step": 33638 }, { "epoch": 24.643956043956045, "grad_norm": 1.7578125, "learning_rate": 2.2868007634406855e-05, "loss": 0.1641, "step": 33639 }, { "epoch": 24.644688644688646, "grad_norm": 0.79296875, "learning_rate": 2.2865164877431653e-05, "loss": 0.1325, "step": 33640 }, { "epoch": 24.645421245421247, "grad_norm": 1.7890625, "learning_rate": 2.2862322252864034e-05, "loss": 0.1368, "step": 33641 }, { "epoch": 24.646153846153847, "grad_norm": 0.890625, "learning_rate": 2.285947976071759e-05, "loss": 0.0876, "step": 33642 }, { "epoch": 24.646886446886448, "grad_norm": 1.46875, "learning_rate": 2.2856637401005948e-05, "loss": 0.1112, "step": 33643 }, { "epoch": 24.64761904761905, "grad_norm": 1.1171875, "learning_rate": 2.285379517374273e-05, "loss": 0.1305, "step": 33644 }, { "epoch": 24.64835164835165, "grad_norm": 0.6875, "learning_rate": 2.2850953078941573e-05, "loss": 0.1047, "step": 33645 }, { "epoch": 24.64908424908425, "grad_norm": 1.578125, "learning_rate": 2.284811111661606e-05, "loss": 0.1335, "step": 33646 }, { "epoch": 24.64981684981685, "grad_norm": 0.74609375, "learning_rate": 2.2845269286779824e-05, "loss": 0.1041, "step": 33647 }, { "epoch": 24.650549450549452, "grad_norm": 1.390625, "learning_rate": 2.284242758944648e-05, "loss": 0.1619, "step": 33648 }, { "epoch": 24.651282051282053, "grad_norm": 1.6875, "learning_rate": 2.2839586024629667e-05, "loss": 0.1108, "step": 33649 }, { "epoch": 24.652014652014653, "grad_norm": 1.3046875, "learning_rate": 2.2836744592342955e-05, "loss": 0.1396, "step": 33650 }, { "epoch": 24.652747252747254, "grad_norm": 0.77734375, "learning_rate": 2.2833903292599983e-05, "loss": 0.1074, "step": 33651 }, { "epoch": 24.653479853479855, "grad_norm": 0.921875, "learning_rate": 2.2831062125414383e-05, "loss": 0.1156, "step": 33652 }, { "epoch": 24.654212454212455, "grad_norm": 2.015625, "learning_rate": 2.282822109079971e-05, "loss": 0.1061, "step": 33653 }, { "epoch": 24.654945054945056, "grad_norm": 1.34375, "learning_rate": 2.2825380188769646e-05, "loss": 0.1207, "step": 33654 }, { "epoch": 24.655677655677657, "grad_norm": 1.015625, "learning_rate": 2.2822539419337756e-05, "loss": 0.1074, "step": 33655 }, { "epoch": 24.656410256410258, "grad_norm": 1.328125, "learning_rate": 2.2819698782517682e-05, "loss": 0.152, "step": 33656 }, { "epoch": 24.65714285714286, "grad_norm": 1.296875, "learning_rate": 2.2816858278323e-05, "loss": 0.1147, "step": 33657 }, { "epoch": 24.65787545787546, "grad_norm": 1.0078125, "learning_rate": 2.2814017906767336e-05, "loss": 0.1394, "step": 33658 }, { "epoch": 24.65860805860806, "grad_norm": 0.88671875, "learning_rate": 2.2811177667864306e-05, "loss": 0.1235, "step": 33659 }, { "epoch": 24.65934065934066, "grad_norm": 0.890625, "learning_rate": 2.280833756162753e-05, "loss": 0.0906, "step": 33660 }, { "epoch": 24.66007326007326, "grad_norm": 1.484375, "learning_rate": 2.2805497588070573e-05, "loss": 0.1449, "step": 33661 }, { "epoch": 24.660805860805862, "grad_norm": 1.0859375, "learning_rate": 2.280265774720708e-05, "loss": 0.1201, "step": 33662 }, { "epoch": 24.661538461538463, "grad_norm": 0.90625, "learning_rate": 2.2799818039050662e-05, "loss": 0.1272, "step": 33663 }, { "epoch": 24.662271062271063, "grad_norm": 0.796875, "learning_rate": 2.279697846361487e-05, "loss": 0.1291, "step": 33664 }, { "epoch": 24.663003663003664, "grad_norm": 0.8359375, "learning_rate": 2.2794139020913385e-05, "loss": 0.104, "step": 33665 }, { "epoch": 24.663736263736265, "grad_norm": 1.1796875, "learning_rate": 2.2791299710959758e-05, "loss": 0.129, "step": 33666 }, { "epoch": 24.664468864468866, "grad_norm": 0.75, "learning_rate": 2.2788460533767625e-05, "loss": 0.1331, "step": 33667 }, { "epoch": 24.665201465201466, "grad_norm": 0.89453125, "learning_rate": 2.2785621489350563e-05, "loss": 0.1188, "step": 33668 }, { "epoch": 24.665934065934067, "grad_norm": 0.91015625, "learning_rate": 2.278278257772218e-05, "loss": 0.1032, "step": 33669 }, { "epoch": 24.666666666666668, "grad_norm": 0.80859375, "learning_rate": 2.2779943798896088e-05, "loss": 0.1231, "step": 33670 }, { "epoch": 24.66739926739927, "grad_norm": 0.72265625, "learning_rate": 2.27771051528859e-05, "loss": 0.0949, "step": 33671 }, { "epoch": 24.66813186813187, "grad_norm": 0.77734375, "learning_rate": 2.277426663970519e-05, "loss": 0.1297, "step": 33672 }, { "epoch": 24.66886446886447, "grad_norm": 1.328125, "learning_rate": 2.277142825936757e-05, "loss": 0.1363, "step": 33673 }, { "epoch": 24.66959706959707, "grad_norm": 1.0625, "learning_rate": 2.2768590011886634e-05, "loss": 0.1214, "step": 33674 }, { "epoch": 24.67032967032967, "grad_norm": 0.7578125, "learning_rate": 2.2765751897276012e-05, "loss": 0.124, "step": 33675 }, { "epoch": 24.671062271062272, "grad_norm": 0.98828125, "learning_rate": 2.276291391554925e-05, "loss": 0.1356, "step": 33676 }, { "epoch": 24.671794871794873, "grad_norm": 1.59375, "learning_rate": 2.2760076066719982e-05, "loss": 0.1467, "step": 33677 }, { "epoch": 24.672527472527474, "grad_norm": 1.21875, "learning_rate": 2.2757238350801806e-05, "loss": 0.138, "step": 33678 }, { "epoch": 24.673260073260074, "grad_norm": 0.75390625, "learning_rate": 2.275440076780828e-05, "loss": 0.1192, "step": 33679 }, { "epoch": 24.673992673992675, "grad_norm": 1.4453125, "learning_rate": 2.275156331775306e-05, "loss": 0.1407, "step": 33680 }, { "epoch": 24.674725274725276, "grad_norm": 0.71484375, "learning_rate": 2.274872600064969e-05, "loss": 0.0925, "step": 33681 }, { "epoch": 24.675457875457877, "grad_norm": 1.3046875, "learning_rate": 2.2745888816511805e-05, "loss": 0.137, "step": 33682 }, { "epoch": 24.676190476190477, "grad_norm": 0.85546875, "learning_rate": 2.2743051765352955e-05, "loss": 0.1044, "step": 33683 }, { "epoch": 24.676923076923078, "grad_norm": 1.0546875, "learning_rate": 2.274021484718676e-05, "loss": 0.1103, "step": 33684 }, { "epoch": 24.67765567765568, "grad_norm": 0.78515625, "learning_rate": 2.273737806202681e-05, "loss": 0.1135, "step": 33685 }, { "epoch": 24.67838827838828, "grad_norm": 0.91015625, "learning_rate": 2.2734541409886713e-05, "loss": 0.1269, "step": 33686 }, { "epoch": 24.67912087912088, "grad_norm": 0.8046875, "learning_rate": 2.2731704890780026e-05, "loss": 0.1333, "step": 33687 }, { "epoch": 24.67985347985348, "grad_norm": 1.0625, "learning_rate": 2.2728868504720354e-05, "loss": 0.1265, "step": 33688 }, { "epoch": 24.68058608058608, "grad_norm": 0.9609375, "learning_rate": 2.272603225172131e-05, "loss": 0.1271, "step": 33689 }, { "epoch": 24.681318681318682, "grad_norm": 1.5859375, "learning_rate": 2.272319613179642e-05, "loss": 0.1279, "step": 33690 }, { "epoch": 24.682051282051283, "grad_norm": 1.09375, "learning_rate": 2.272036014495936e-05, "loss": 0.1085, "step": 33691 }, { "epoch": 24.682783882783884, "grad_norm": 0.76953125, "learning_rate": 2.2717524291223652e-05, "loss": 0.1448, "step": 33692 }, { "epoch": 24.683516483516485, "grad_norm": 1.234375, "learning_rate": 2.2714688570602923e-05, "loss": 0.1283, "step": 33693 }, { "epoch": 24.684249084249085, "grad_norm": 0.921875, "learning_rate": 2.2711852983110722e-05, "loss": 0.1042, "step": 33694 }, { "epoch": 24.684981684981686, "grad_norm": 1.328125, "learning_rate": 2.2709017528760665e-05, "loss": 0.1229, "step": 33695 }, { "epoch": 24.685714285714287, "grad_norm": 1.34375, "learning_rate": 2.2706182207566322e-05, "loss": 0.1286, "step": 33696 }, { "epoch": 24.686446886446888, "grad_norm": 1.0078125, "learning_rate": 2.2703347019541303e-05, "loss": 0.1144, "step": 33697 }, { "epoch": 24.68717948717949, "grad_norm": 0.76953125, "learning_rate": 2.2700511964699153e-05, "loss": 0.1327, "step": 33698 }, { "epoch": 24.68791208791209, "grad_norm": 1.125, "learning_rate": 2.2697677043053477e-05, "loss": 0.1196, "step": 33699 }, { "epoch": 24.68864468864469, "grad_norm": 1.4453125, "learning_rate": 2.269484225461786e-05, "loss": 0.1281, "step": 33700 }, { "epoch": 24.68937728937729, "grad_norm": 1.2578125, "learning_rate": 2.26920075994059e-05, "loss": 0.106, "step": 33701 }, { "epoch": 24.69010989010989, "grad_norm": 0.8125, "learning_rate": 2.268917307743114e-05, "loss": 0.1304, "step": 33702 }, { "epoch": 24.690842490842492, "grad_norm": 0.9609375, "learning_rate": 2.2686338688707186e-05, "loss": 0.1193, "step": 33703 }, { "epoch": 24.691575091575093, "grad_norm": 1.0625, "learning_rate": 2.2683504433247627e-05, "loss": 0.112, "step": 33704 }, { "epoch": 24.692307692307693, "grad_norm": 0.90625, "learning_rate": 2.2680670311065995e-05, "loss": 0.1298, "step": 33705 }, { "epoch": 24.693040293040294, "grad_norm": 1.125, "learning_rate": 2.2677836322175934e-05, "loss": 0.1098, "step": 33706 }, { "epoch": 24.693772893772895, "grad_norm": 0.9453125, "learning_rate": 2.267500246659098e-05, "loss": 0.1349, "step": 33707 }, { "epoch": 24.694505494505496, "grad_norm": 1.2109375, "learning_rate": 2.2672168744324746e-05, "loss": 0.1321, "step": 33708 }, { "epoch": 24.695238095238096, "grad_norm": 1.125, "learning_rate": 2.2669335155390768e-05, "loss": 0.1114, "step": 33709 }, { "epoch": 24.695970695970697, "grad_norm": 0.875, "learning_rate": 2.2666501699802636e-05, "loss": 0.1401, "step": 33710 }, { "epoch": 24.696703296703298, "grad_norm": 0.9609375, "learning_rate": 2.2663668377573935e-05, "loss": 0.1094, "step": 33711 }, { "epoch": 24.6974358974359, "grad_norm": 0.94140625, "learning_rate": 2.266083518871825e-05, "loss": 0.1117, "step": 33712 }, { "epoch": 24.6981684981685, "grad_norm": 0.97265625, "learning_rate": 2.265800213324913e-05, "loss": 0.1164, "step": 33713 }, { "epoch": 24.6989010989011, "grad_norm": 1.328125, "learning_rate": 2.265516921118016e-05, "loss": 0.1278, "step": 33714 }, { "epoch": 24.6996336996337, "grad_norm": 0.984375, "learning_rate": 2.2652336422524932e-05, "loss": 0.1251, "step": 33715 }, { "epoch": 24.7003663003663, "grad_norm": 0.859375, "learning_rate": 2.2649503767296968e-05, "loss": 0.0955, "step": 33716 }, { "epoch": 24.701098901098902, "grad_norm": 0.6796875, "learning_rate": 2.2646671245509906e-05, "loss": 0.1264, "step": 33717 }, { "epoch": 24.701831501831503, "grad_norm": 0.95703125, "learning_rate": 2.2643838857177274e-05, "loss": 0.0982, "step": 33718 }, { "epoch": 24.702564102564104, "grad_norm": 1.0, "learning_rate": 2.2641006602312673e-05, "loss": 0.1058, "step": 33719 }, { "epoch": 24.703296703296704, "grad_norm": 1.3671875, "learning_rate": 2.263817448092963e-05, "loss": 0.1238, "step": 33720 }, { "epoch": 24.704029304029305, "grad_norm": 1.015625, "learning_rate": 2.2635342493041742e-05, "loss": 0.0864, "step": 33721 }, { "epoch": 24.704761904761906, "grad_norm": 0.90625, "learning_rate": 2.263251063866258e-05, "loss": 0.1282, "step": 33722 }, { "epoch": 24.705494505494507, "grad_norm": 1.734375, "learning_rate": 2.2629678917805722e-05, "loss": 0.1301, "step": 33723 }, { "epoch": 24.706227106227107, "grad_norm": 0.62890625, "learning_rate": 2.2626847330484704e-05, "loss": 0.103, "step": 33724 }, { "epoch": 24.706959706959708, "grad_norm": 0.70703125, "learning_rate": 2.262401587671311e-05, "loss": 0.1026, "step": 33725 }, { "epoch": 24.70769230769231, "grad_norm": 0.88671875, "learning_rate": 2.2621184556504504e-05, "loss": 0.144, "step": 33726 }, { "epoch": 24.70842490842491, "grad_norm": 0.68359375, "learning_rate": 2.261835336987248e-05, "loss": 0.1231, "step": 33727 }, { "epoch": 24.70915750915751, "grad_norm": 1.2109375, "learning_rate": 2.2615522316830557e-05, "loss": 0.1069, "step": 33728 }, { "epoch": 24.70989010989011, "grad_norm": 0.73046875, "learning_rate": 2.2612691397392317e-05, "loss": 0.1228, "step": 33729 }, { "epoch": 24.71062271062271, "grad_norm": 0.94140625, "learning_rate": 2.260986061157135e-05, "loss": 0.1268, "step": 33730 }, { "epoch": 24.711355311355312, "grad_norm": 0.98046875, "learning_rate": 2.2607029959381156e-05, "loss": 0.1204, "step": 33731 }, { "epoch": 24.712087912087913, "grad_norm": 1.3828125, "learning_rate": 2.2604199440835377e-05, "loss": 0.1067, "step": 33732 }, { "epoch": 24.712820512820514, "grad_norm": 0.7734375, "learning_rate": 2.2601369055947518e-05, "loss": 0.1241, "step": 33733 }, { "epoch": 24.713553113553115, "grad_norm": 1.46875, "learning_rate": 2.2598538804731174e-05, "loss": 0.1207, "step": 33734 }, { "epoch": 24.714285714285715, "grad_norm": 0.95703125, "learning_rate": 2.2595708687199876e-05, "loss": 0.1087, "step": 33735 }, { "epoch": 24.715018315018316, "grad_norm": 0.69921875, "learning_rate": 2.2592878703367216e-05, "loss": 0.097, "step": 33736 }, { "epoch": 24.715750915750917, "grad_norm": 0.83203125, "learning_rate": 2.2590048853246696e-05, "loss": 0.1145, "step": 33737 }, { "epoch": 24.716483516483517, "grad_norm": 0.87109375, "learning_rate": 2.2587219136851954e-05, "loss": 0.0996, "step": 33738 }, { "epoch": 24.717216117216118, "grad_norm": 1.09375, "learning_rate": 2.2584389554196483e-05, "loss": 0.1385, "step": 33739 }, { "epoch": 24.71794871794872, "grad_norm": 0.90234375, "learning_rate": 2.2581560105293878e-05, "loss": 0.1282, "step": 33740 }, { "epoch": 24.71868131868132, "grad_norm": 1.234375, "learning_rate": 2.257873079015769e-05, "loss": 0.127, "step": 33741 }, { "epoch": 24.71941391941392, "grad_norm": 0.58203125, "learning_rate": 2.257590160880144e-05, "loss": 0.1033, "step": 33742 }, { "epoch": 24.72014652014652, "grad_norm": 1.109375, "learning_rate": 2.2573072561238748e-05, "loss": 0.1006, "step": 33743 }, { "epoch": 24.720879120879122, "grad_norm": 1.171875, "learning_rate": 2.2570243647483105e-05, "loss": 0.1161, "step": 33744 }, { "epoch": 24.721611721611723, "grad_norm": 1.609375, "learning_rate": 2.256741486754812e-05, "loss": 0.1164, "step": 33745 }, { "epoch": 24.722344322344323, "grad_norm": 0.96875, "learning_rate": 2.2564586221447307e-05, "loss": 0.1048, "step": 33746 }, { "epoch": 24.723076923076924, "grad_norm": 0.765625, "learning_rate": 2.256175770919422e-05, "loss": 0.1052, "step": 33747 }, { "epoch": 24.723809523809525, "grad_norm": 0.93359375, "learning_rate": 2.255892933080243e-05, "loss": 0.1224, "step": 33748 }, { "epoch": 24.724542124542126, "grad_norm": 1.2109375, "learning_rate": 2.25561010862855e-05, "loss": 0.1181, "step": 33749 }, { "epoch": 24.725274725274726, "grad_norm": 1.4140625, "learning_rate": 2.2553272975656947e-05, "loss": 0.1211, "step": 33750 }, { "epoch": 24.726007326007327, "grad_norm": 1.2890625, "learning_rate": 2.2550444998930333e-05, "loss": 0.1174, "step": 33751 }, { "epoch": 24.726739926739928, "grad_norm": 0.66796875, "learning_rate": 2.2547617156119218e-05, "loss": 0.1152, "step": 33752 }, { "epoch": 24.72747252747253, "grad_norm": 0.98828125, "learning_rate": 2.2544789447237158e-05, "loss": 0.1265, "step": 33753 }, { "epoch": 24.72820512820513, "grad_norm": 1.78125, "learning_rate": 2.2541961872297672e-05, "loss": 0.1439, "step": 33754 }, { "epoch": 24.72893772893773, "grad_norm": 1.0390625, "learning_rate": 2.2539134431314325e-05, "loss": 0.1047, "step": 33755 }, { "epoch": 24.72967032967033, "grad_norm": 1.015625, "learning_rate": 2.2536307124300685e-05, "loss": 0.1297, "step": 33756 }, { "epoch": 24.73040293040293, "grad_norm": 1.453125, "learning_rate": 2.253347995127024e-05, "loss": 0.1123, "step": 33757 }, { "epoch": 24.731135531135532, "grad_norm": 1.0078125, "learning_rate": 2.25306529122366e-05, "loss": 0.1085, "step": 33758 }, { "epoch": 24.731868131868133, "grad_norm": 0.56640625, "learning_rate": 2.2527826007213278e-05, "loss": 0.1112, "step": 33759 }, { "epoch": 24.732600732600734, "grad_norm": 1.203125, "learning_rate": 2.2524999236213836e-05, "loss": 0.0933, "step": 33760 }, { "epoch": 24.733333333333334, "grad_norm": 1.140625, "learning_rate": 2.252217259925179e-05, "loss": 0.1448, "step": 33761 }, { "epoch": 24.734065934065935, "grad_norm": 0.71484375, "learning_rate": 2.2519346096340703e-05, "loss": 0.1063, "step": 33762 }, { "epoch": 24.734798534798536, "grad_norm": 1.375, "learning_rate": 2.2516519727494107e-05, "loss": 0.0998, "step": 33763 }, { "epoch": 24.735531135531136, "grad_norm": 0.85546875, "learning_rate": 2.2513693492725576e-05, "loss": 0.1173, "step": 33764 }, { "epoch": 24.736263736263737, "grad_norm": 0.98828125, "learning_rate": 2.2510867392048605e-05, "loss": 0.1008, "step": 33765 }, { "epoch": 24.736996336996338, "grad_norm": 0.8125, "learning_rate": 2.250804142547676e-05, "loss": 0.1179, "step": 33766 }, { "epoch": 24.73772893772894, "grad_norm": 1.28125, "learning_rate": 2.250521559302359e-05, "loss": 0.1359, "step": 33767 }, { "epoch": 24.73846153846154, "grad_norm": 1.3203125, "learning_rate": 2.250238989470259e-05, "loss": 0.1348, "step": 33768 }, { "epoch": 24.73919413919414, "grad_norm": 0.90234375, "learning_rate": 2.2499564330527366e-05, "loss": 0.1003, "step": 33769 }, { "epoch": 24.73992673992674, "grad_norm": 1.6640625, "learning_rate": 2.24967389005114e-05, "loss": 0.11, "step": 33770 }, { "epoch": 24.74065934065934, "grad_norm": 1.3515625, "learning_rate": 2.249391360466828e-05, "loss": 0.1465, "step": 33771 }, { "epoch": 24.741391941391942, "grad_norm": 1.75, "learning_rate": 2.249108844301149e-05, "loss": 0.1106, "step": 33772 }, { "epoch": 24.742124542124543, "grad_norm": 1.3203125, "learning_rate": 2.2488263415554586e-05, "loss": 0.1411, "step": 33773 }, { "epoch": 24.742857142857144, "grad_norm": 0.921875, "learning_rate": 2.248543852231111e-05, "loss": 0.1191, "step": 33774 }, { "epoch": 24.743589743589745, "grad_norm": 0.64453125, "learning_rate": 2.2482613763294612e-05, "loss": 0.0766, "step": 33775 }, { "epoch": 24.744322344322345, "grad_norm": 1.140625, "learning_rate": 2.2479789138518592e-05, "loss": 0.1065, "step": 33776 }, { "epoch": 24.745054945054946, "grad_norm": 1.046875, "learning_rate": 2.2476964647996596e-05, "loss": 0.1286, "step": 33777 }, { "epoch": 24.745787545787547, "grad_norm": 1.359375, "learning_rate": 2.247414029174217e-05, "loss": 0.1535, "step": 33778 }, { "epoch": 24.746520146520147, "grad_norm": 1.4296875, "learning_rate": 2.2471316069768833e-05, "loss": 0.1162, "step": 33779 }, { "epoch": 24.747252747252748, "grad_norm": 1.2578125, "learning_rate": 2.246849198209014e-05, "loss": 0.1241, "step": 33780 }, { "epoch": 24.74798534798535, "grad_norm": 1.03125, "learning_rate": 2.2465668028719587e-05, "loss": 0.1272, "step": 33781 }, { "epoch": 24.74871794871795, "grad_norm": 1.078125, "learning_rate": 2.246284420967074e-05, "loss": 0.1151, "step": 33782 }, { "epoch": 24.74945054945055, "grad_norm": 0.8125, "learning_rate": 2.2460020524957075e-05, "loss": 0.1054, "step": 33783 }, { "epoch": 24.75018315018315, "grad_norm": 1.2109375, "learning_rate": 2.2457196974592187e-05, "loss": 0.1143, "step": 33784 }, { "epoch": 24.750915750915752, "grad_norm": 1.0703125, "learning_rate": 2.2454373558589564e-05, "loss": 0.1223, "step": 33785 }, { "epoch": 24.751648351648353, "grad_norm": 0.6328125, "learning_rate": 2.245155027696276e-05, "loss": 0.0912, "step": 33786 }, { "epoch": 24.752380952380953, "grad_norm": 0.7734375, "learning_rate": 2.2448727129725266e-05, "loss": 0.1203, "step": 33787 }, { "epoch": 24.753113553113554, "grad_norm": 1.0703125, "learning_rate": 2.244590411689063e-05, "loss": 0.1209, "step": 33788 }, { "epoch": 24.753846153846155, "grad_norm": 1.0546875, "learning_rate": 2.2443081238472375e-05, "loss": 0.0972, "step": 33789 }, { "epoch": 24.754578754578755, "grad_norm": 0.93359375, "learning_rate": 2.244025849448405e-05, "loss": 0.1154, "step": 33790 }, { "epoch": 24.755311355311356, "grad_norm": 1.09375, "learning_rate": 2.2437435884939137e-05, "loss": 0.1073, "step": 33791 }, { "epoch": 24.756043956043957, "grad_norm": 1.1640625, "learning_rate": 2.2434613409851182e-05, "loss": 0.0999, "step": 33792 }, { "epoch": 24.756776556776558, "grad_norm": 0.9375, "learning_rate": 2.243179106923373e-05, "loss": 0.0997, "step": 33793 }, { "epoch": 24.75750915750916, "grad_norm": 1.34375, "learning_rate": 2.2428968863100238e-05, "loss": 0.1642, "step": 33794 }, { "epoch": 24.75824175824176, "grad_norm": 1.0390625, "learning_rate": 2.242614679146431e-05, "loss": 0.1283, "step": 33795 }, { "epoch": 24.75897435897436, "grad_norm": 1.1171875, "learning_rate": 2.242332485433941e-05, "loss": 0.1259, "step": 33796 }, { "epoch": 24.75970695970696, "grad_norm": 1.0078125, "learning_rate": 2.2420503051739098e-05, "loss": 0.1214, "step": 33797 }, { "epoch": 24.76043956043956, "grad_norm": 0.75390625, "learning_rate": 2.2417681383676853e-05, "loss": 0.1298, "step": 33798 }, { "epoch": 24.761172161172162, "grad_norm": 1.1953125, "learning_rate": 2.2414859850166216e-05, "loss": 0.12, "step": 33799 }, { "epoch": 24.761904761904763, "grad_norm": 0.96875, "learning_rate": 2.2412038451220708e-05, "loss": 0.1505, "step": 33800 }, { "epoch": 24.762637362637363, "grad_norm": 1.0390625, "learning_rate": 2.2409217186853857e-05, "loss": 0.1415, "step": 33801 }, { "epoch": 24.763369963369964, "grad_norm": 0.83984375, "learning_rate": 2.2406396057079157e-05, "loss": 0.1093, "step": 33802 }, { "epoch": 24.764102564102565, "grad_norm": 1.2421875, "learning_rate": 2.240357506191013e-05, "loss": 0.1497, "step": 33803 }, { "epoch": 24.764835164835166, "grad_norm": 0.74609375, "learning_rate": 2.24007542013603e-05, "loss": 0.1047, "step": 33804 }, { "epoch": 24.765567765567766, "grad_norm": 1.046875, "learning_rate": 2.239793347544318e-05, "loss": 0.1189, "step": 33805 }, { "epoch": 24.766300366300367, "grad_norm": 1.015625, "learning_rate": 2.2395112884172308e-05, "loss": 0.121, "step": 33806 }, { "epoch": 24.767032967032968, "grad_norm": 1.0859375, "learning_rate": 2.2392292427561158e-05, "loss": 0.1072, "step": 33807 }, { "epoch": 24.76776556776557, "grad_norm": 1.0, "learning_rate": 2.2389472105623277e-05, "loss": 0.1272, "step": 33808 }, { "epoch": 24.76849816849817, "grad_norm": 0.7890625, "learning_rate": 2.2386651918372132e-05, "loss": 0.1137, "step": 33809 }, { "epoch": 24.76923076923077, "grad_norm": 0.578125, "learning_rate": 2.2383831865821292e-05, "loss": 0.1083, "step": 33810 }, { "epoch": 24.76996336996337, "grad_norm": 1.1796875, "learning_rate": 2.2381011947984236e-05, "loss": 0.1351, "step": 33811 }, { "epoch": 24.77069597069597, "grad_norm": 0.94921875, "learning_rate": 2.2378192164874496e-05, "loss": 0.1312, "step": 33812 }, { "epoch": 24.771428571428572, "grad_norm": 0.69921875, "learning_rate": 2.2375372516505556e-05, "loss": 0.0993, "step": 33813 }, { "epoch": 24.772161172161173, "grad_norm": 0.94921875, "learning_rate": 2.237255300289093e-05, "loss": 0.1105, "step": 33814 }, { "epoch": 24.772893772893774, "grad_norm": 1.1171875, "learning_rate": 2.2369733624044145e-05, "loss": 0.0939, "step": 33815 }, { "epoch": 24.773626373626374, "grad_norm": 0.75390625, "learning_rate": 2.236691437997871e-05, "loss": 0.1213, "step": 33816 }, { "epoch": 24.774358974358975, "grad_norm": 0.9921875, "learning_rate": 2.2364095270708113e-05, "loss": 0.1156, "step": 33817 }, { "epoch": 24.775091575091576, "grad_norm": 1.125, "learning_rate": 2.236127629624587e-05, "loss": 0.126, "step": 33818 }, { "epoch": 24.775824175824177, "grad_norm": 1.0234375, "learning_rate": 2.2358457456605513e-05, "loss": 0.1224, "step": 33819 }, { "epoch": 24.776556776556777, "grad_norm": 0.84375, "learning_rate": 2.2355638751800478e-05, "loss": 0.1284, "step": 33820 }, { "epoch": 24.777289377289378, "grad_norm": 1.609375, "learning_rate": 2.235282018184436e-05, "loss": 0.1398, "step": 33821 }, { "epoch": 24.77802197802198, "grad_norm": 1.25, "learning_rate": 2.23500017467506e-05, "loss": 0.1315, "step": 33822 }, { "epoch": 24.77875457875458, "grad_norm": 1.328125, "learning_rate": 2.234718344653274e-05, "loss": 0.1087, "step": 33823 }, { "epoch": 24.77948717948718, "grad_norm": 1.203125, "learning_rate": 2.234436528120425e-05, "loss": 0.0875, "step": 33824 }, { "epoch": 24.78021978021978, "grad_norm": 0.94140625, "learning_rate": 2.234154725077865e-05, "loss": 0.1158, "step": 33825 }, { "epoch": 24.78095238095238, "grad_norm": 0.98046875, "learning_rate": 2.233872935526944e-05, "loss": 0.1101, "step": 33826 }, { "epoch": 24.781684981684982, "grad_norm": 0.83203125, "learning_rate": 2.233591159469014e-05, "loss": 0.1056, "step": 33827 }, { "epoch": 24.782417582417583, "grad_norm": 1.453125, "learning_rate": 2.2333093969054215e-05, "loss": 0.1362, "step": 33828 }, { "epoch": 24.783150183150184, "grad_norm": 0.6328125, "learning_rate": 2.2330276478375182e-05, "loss": 0.1145, "step": 33829 }, { "epoch": 24.783882783882785, "grad_norm": 1.4609375, "learning_rate": 2.232745912266654e-05, "loss": 0.1535, "step": 33830 }, { "epoch": 24.784615384615385, "grad_norm": 1.265625, "learning_rate": 2.2324641901941796e-05, "loss": 0.1399, "step": 33831 }, { "epoch": 24.785347985347986, "grad_norm": 1.3359375, "learning_rate": 2.2321824816214452e-05, "loss": 0.1435, "step": 33832 }, { "epoch": 24.786080586080587, "grad_norm": 1.2109375, "learning_rate": 2.2319007865497988e-05, "loss": 0.1185, "step": 33833 }, { "epoch": 24.786813186813188, "grad_norm": 0.80859375, "learning_rate": 2.2316191049805917e-05, "loss": 0.1183, "step": 33834 }, { "epoch": 24.78754578754579, "grad_norm": 1.1484375, "learning_rate": 2.2313374369151694e-05, "loss": 0.1344, "step": 33835 }, { "epoch": 24.78827838827839, "grad_norm": 1.5234375, "learning_rate": 2.2310557823548886e-05, "loss": 0.1424, "step": 33836 }, { "epoch": 24.78901098901099, "grad_norm": 0.75390625, "learning_rate": 2.2307741413010933e-05, "loss": 0.1212, "step": 33837 }, { "epoch": 24.78974358974359, "grad_norm": 0.87890625, "learning_rate": 2.2304925137551356e-05, "loss": 0.1325, "step": 33838 }, { "epoch": 24.79047619047619, "grad_norm": 1.796875, "learning_rate": 2.2302108997183628e-05, "loss": 0.1337, "step": 33839 }, { "epoch": 24.791208791208792, "grad_norm": 1.0078125, "learning_rate": 2.2299292991921246e-05, "loss": 0.111, "step": 33840 }, { "epoch": 24.791941391941393, "grad_norm": 1.0234375, "learning_rate": 2.2296477121777716e-05, "loss": 0.161, "step": 33841 }, { "epoch": 24.792673992673993, "grad_norm": 0.84765625, "learning_rate": 2.229366138676654e-05, "loss": 0.1271, "step": 33842 }, { "epoch": 24.793406593406594, "grad_norm": 0.7109375, "learning_rate": 2.2290845786901167e-05, "loss": 0.1216, "step": 33843 }, { "epoch": 24.794139194139195, "grad_norm": 1.4921875, "learning_rate": 2.2288030322195114e-05, "loss": 0.0991, "step": 33844 }, { "epoch": 24.794871794871796, "grad_norm": 0.91015625, "learning_rate": 2.2285214992661892e-05, "loss": 0.116, "step": 33845 }, { "epoch": 24.795604395604396, "grad_norm": 1.15625, "learning_rate": 2.228239979831493e-05, "loss": 0.1303, "step": 33846 }, { "epoch": 24.796336996336997, "grad_norm": 1.1640625, "learning_rate": 2.227958473916779e-05, "loss": 0.1486, "step": 33847 }, { "epoch": 24.797069597069598, "grad_norm": 0.80859375, "learning_rate": 2.2276769815233902e-05, "loss": 0.1142, "step": 33848 }, { "epoch": 24.7978021978022, "grad_norm": 2.09375, "learning_rate": 2.2273955026526792e-05, "loss": 0.1341, "step": 33849 }, { "epoch": 24.7985347985348, "grad_norm": 2.125, "learning_rate": 2.2271140373059914e-05, "loss": 0.1561, "step": 33850 }, { "epoch": 24.7992673992674, "grad_norm": 1.25, "learning_rate": 2.2268325854846775e-05, "loss": 0.1214, "step": 33851 }, { "epoch": 24.8, "grad_norm": 1.4375, "learning_rate": 2.2265511471900848e-05, "loss": 0.1499, "step": 33852 }, { "epoch": 24.8007326007326, "grad_norm": 0.92578125, "learning_rate": 2.2262697224235643e-05, "loss": 0.1051, "step": 33853 }, { "epoch": 24.801465201465202, "grad_norm": 1.6484375, "learning_rate": 2.2259883111864606e-05, "loss": 0.113, "step": 33854 }, { "epoch": 24.802197802197803, "grad_norm": 1.34375, "learning_rate": 2.2257069134801244e-05, "loss": 0.1246, "step": 33855 }, { "epoch": 24.802930402930404, "grad_norm": 1.140625, "learning_rate": 2.2254255293059036e-05, "loss": 0.1115, "step": 33856 }, { "epoch": 24.803663003663004, "grad_norm": 1.1796875, "learning_rate": 2.2251441586651456e-05, "loss": 0.1288, "step": 33857 }, { "epoch": 24.804395604395605, "grad_norm": 0.921875, "learning_rate": 2.2248628015592022e-05, "loss": 0.1393, "step": 33858 }, { "epoch": 24.805128205128206, "grad_norm": 1.9296875, "learning_rate": 2.224581457989416e-05, "loss": 0.1346, "step": 33859 }, { "epoch": 24.805860805860807, "grad_norm": 1.203125, "learning_rate": 2.2243001279571393e-05, "loss": 0.1222, "step": 33860 }, { "epoch": 24.806593406593407, "grad_norm": 0.84375, "learning_rate": 2.224018811463715e-05, "loss": 0.0957, "step": 33861 }, { "epoch": 24.807326007326008, "grad_norm": 0.875, "learning_rate": 2.2237375085104975e-05, "loss": 0.1069, "step": 33862 }, { "epoch": 24.80805860805861, "grad_norm": 0.91015625, "learning_rate": 2.22345621909883e-05, "loss": 0.1228, "step": 33863 }, { "epoch": 24.80879120879121, "grad_norm": 0.97265625, "learning_rate": 2.2231749432300632e-05, "loss": 0.118, "step": 33864 }, { "epoch": 24.80952380952381, "grad_norm": 0.88671875, "learning_rate": 2.222893680905541e-05, "loss": 0.0872, "step": 33865 }, { "epoch": 24.81025641025641, "grad_norm": 0.83984375, "learning_rate": 2.2226124321266142e-05, "loss": 0.129, "step": 33866 }, { "epoch": 24.81098901098901, "grad_norm": 0.75390625, "learning_rate": 2.222331196894629e-05, "loss": 0.099, "step": 33867 }, { "epoch": 24.811721611721612, "grad_norm": 1.015625, "learning_rate": 2.222049975210935e-05, "loss": 0.1037, "step": 33868 }, { "epoch": 24.812454212454213, "grad_norm": 1.015625, "learning_rate": 2.2217687670768762e-05, "loss": 0.1279, "step": 33869 }, { "epoch": 24.813186813186814, "grad_norm": 0.7734375, "learning_rate": 2.2214875724938018e-05, "loss": 0.1143, "step": 33870 }, { "epoch": 24.813919413919415, "grad_norm": 0.81640625, "learning_rate": 2.2212063914630604e-05, "loss": 0.1342, "step": 33871 }, { "epoch": 24.814652014652015, "grad_norm": 0.8125, "learning_rate": 2.2209252239859947e-05, "loss": 0.136, "step": 33872 }, { "epoch": 24.815384615384616, "grad_norm": 1.3125, "learning_rate": 2.220644070063958e-05, "loss": 0.1392, "step": 33873 }, { "epoch": 24.816117216117217, "grad_norm": 1.0078125, "learning_rate": 2.2203629296982932e-05, "loss": 0.1073, "step": 33874 }, { "epoch": 24.816849816849818, "grad_norm": 1.0, "learning_rate": 2.22008180289035e-05, "loss": 0.1304, "step": 33875 }, { "epoch": 24.81758241758242, "grad_norm": 0.94921875, "learning_rate": 2.219800689641472e-05, "loss": 0.1198, "step": 33876 }, { "epoch": 24.81831501831502, "grad_norm": 0.953125, "learning_rate": 2.2195195899530084e-05, "loss": 0.1392, "step": 33877 }, { "epoch": 24.81904761904762, "grad_norm": 1.0625, "learning_rate": 2.2192385038263053e-05, "loss": 0.1132, "step": 33878 }, { "epoch": 24.81978021978022, "grad_norm": 1.1953125, "learning_rate": 2.218957431262712e-05, "loss": 0.1082, "step": 33879 }, { "epoch": 24.82051282051282, "grad_norm": 1.078125, "learning_rate": 2.2186763722635708e-05, "loss": 0.1399, "step": 33880 }, { "epoch": 24.821245421245422, "grad_norm": 1.875, "learning_rate": 2.2183953268302305e-05, "loss": 0.1585, "step": 33881 }, { "epoch": 24.821978021978023, "grad_norm": 1.21875, "learning_rate": 2.2181142949640375e-05, "loss": 0.1334, "step": 33882 }, { "epoch": 24.822710622710623, "grad_norm": 1.6484375, "learning_rate": 2.217833276666339e-05, "loss": 0.148, "step": 33883 }, { "epoch": 24.823443223443224, "grad_norm": 1.6796875, "learning_rate": 2.217552271938483e-05, "loss": 0.1538, "step": 33884 }, { "epoch": 24.824175824175825, "grad_norm": 1.6953125, "learning_rate": 2.217271280781812e-05, "loss": 0.1291, "step": 33885 }, { "epoch": 24.824908424908426, "grad_norm": 1.1953125, "learning_rate": 2.2169903031976755e-05, "loss": 0.153, "step": 33886 }, { "epoch": 24.825641025641026, "grad_norm": 1.3125, "learning_rate": 2.2167093391874156e-05, "loss": 0.1199, "step": 33887 }, { "epoch": 24.826373626373627, "grad_norm": 1.09375, "learning_rate": 2.2164283887523844e-05, "loss": 0.1308, "step": 33888 }, { "epoch": 24.827106227106228, "grad_norm": 1.1328125, "learning_rate": 2.2161474518939235e-05, "loss": 0.1453, "step": 33889 }, { "epoch": 24.82783882783883, "grad_norm": 0.78515625, "learning_rate": 2.215866528613382e-05, "loss": 0.1131, "step": 33890 }, { "epoch": 24.82857142857143, "grad_norm": 1.046875, "learning_rate": 2.2155856189121026e-05, "loss": 0.1176, "step": 33891 }, { "epoch": 24.82930402930403, "grad_norm": 1.375, "learning_rate": 2.215304722791433e-05, "loss": 0.1149, "step": 33892 }, { "epoch": 24.83003663003663, "grad_norm": 0.7890625, "learning_rate": 2.2150238402527194e-05, "loss": 0.1196, "step": 33893 }, { "epoch": 24.83076923076923, "grad_norm": 1.0546875, "learning_rate": 2.2147429712973063e-05, "loss": 0.1345, "step": 33894 }, { "epoch": 24.831501831501832, "grad_norm": 0.9140625, "learning_rate": 2.214462115926543e-05, "loss": 0.1107, "step": 33895 }, { "epoch": 24.832234432234433, "grad_norm": 0.90625, "learning_rate": 2.2141812741417708e-05, "loss": 0.1141, "step": 33896 }, { "epoch": 24.832967032967034, "grad_norm": 0.9296875, "learning_rate": 2.2139004459443383e-05, "loss": 0.101, "step": 33897 }, { "epoch": 24.833699633699634, "grad_norm": 1.03125, "learning_rate": 2.2136196313355867e-05, "loss": 0.1377, "step": 33898 }, { "epoch": 24.834432234432235, "grad_norm": 1.171875, "learning_rate": 2.213338830316868e-05, "loss": 0.1604, "step": 33899 }, { "epoch": 24.835164835164836, "grad_norm": 1.203125, "learning_rate": 2.2130580428895224e-05, "loss": 0.1289, "step": 33900 }, { "epoch": 24.835897435897436, "grad_norm": 1.390625, "learning_rate": 2.212777269054899e-05, "loss": 0.153, "step": 33901 }, { "epoch": 24.836630036630037, "grad_norm": 1.0625, "learning_rate": 2.212496508814339e-05, "loss": 0.1136, "step": 33902 }, { "epoch": 24.837362637362638, "grad_norm": 1.03125, "learning_rate": 2.2122157621691896e-05, "loss": 0.1134, "step": 33903 }, { "epoch": 24.83809523809524, "grad_norm": 0.95703125, "learning_rate": 2.2119350291207966e-05, "loss": 0.112, "step": 33904 }, { "epoch": 24.83882783882784, "grad_norm": 0.7109375, "learning_rate": 2.211654309670506e-05, "loss": 0.1218, "step": 33905 }, { "epoch": 24.83956043956044, "grad_norm": 0.64453125, "learning_rate": 2.2113736038196595e-05, "loss": 0.1274, "step": 33906 }, { "epoch": 24.84029304029304, "grad_norm": 2.140625, "learning_rate": 2.211092911569604e-05, "loss": 0.1408, "step": 33907 }, { "epoch": 24.84102564102564, "grad_norm": 1.1484375, "learning_rate": 2.2108122329216847e-05, "loss": 0.1458, "step": 33908 }, { "epoch": 24.841758241758242, "grad_norm": 0.97265625, "learning_rate": 2.2105315678772452e-05, "loss": 0.1144, "step": 33909 }, { "epoch": 24.842490842490843, "grad_norm": 1.09375, "learning_rate": 2.2102509164376337e-05, "loss": 0.0978, "step": 33910 }, { "epoch": 24.843223443223444, "grad_norm": 1.296875, "learning_rate": 2.20997027860419e-05, "loss": 0.1237, "step": 33911 }, { "epoch": 24.843956043956045, "grad_norm": 0.9375, "learning_rate": 2.209689654378263e-05, "loss": 0.1319, "step": 33912 }, { "epoch": 24.844688644688645, "grad_norm": 0.78125, "learning_rate": 2.2094090437611913e-05, "loss": 0.1166, "step": 33913 }, { "epoch": 24.845421245421246, "grad_norm": 1.2890625, "learning_rate": 2.2091284467543277e-05, "loss": 0.1001, "step": 33914 }, { "epoch": 24.846153846153847, "grad_norm": 0.79296875, "learning_rate": 2.2088478633590098e-05, "loss": 0.1155, "step": 33915 }, { "epoch": 24.846886446886447, "grad_norm": 0.796875, "learning_rate": 2.2085672935765874e-05, "loss": 0.1229, "step": 33916 }, { "epoch": 24.847619047619048, "grad_norm": 1.4296875, "learning_rate": 2.2082867374083988e-05, "loss": 0.1488, "step": 33917 }, { "epoch": 24.84835164835165, "grad_norm": 1.328125, "learning_rate": 2.2080061948557915e-05, "loss": 0.0994, "step": 33918 }, { "epoch": 24.84908424908425, "grad_norm": 0.82421875, "learning_rate": 2.2077256659201103e-05, "loss": 0.1242, "step": 33919 }, { "epoch": 24.84981684981685, "grad_norm": 0.80078125, "learning_rate": 2.207445150602698e-05, "loss": 0.106, "step": 33920 }, { "epoch": 24.85054945054945, "grad_norm": 0.57421875, "learning_rate": 2.2071646489049007e-05, "loss": 0.0907, "step": 33921 }, { "epoch": 24.851282051282052, "grad_norm": 0.875, "learning_rate": 2.206884160828059e-05, "loss": 0.1254, "step": 33922 }, { "epoch": 24.852014652014653, "grad_norm": 0.68359375, "learning_rate": 2.206603686373521e-05, "loss": 0.1012, "step": 33923 }, { "epoch": 24.852747252747253, "grad_norm": 0.7890625, "learning_rate": 2.206323225542624e-05, "loss": 0.1163, "step": 33924 }, { "epoch": 24.853479853479854, "grad_norm": 1.0078125, "learning_rate": 2.20604277833672e-05, "loss": 0.1121, "step": 33925 }, { "epoch": 24.854212454212455, "grad_norm": 0.93359375, "learning_rate": 2.2057623447571464e-05, "loss": 0.1038, "step": 33926 }, { "epoch": 24.854945054945055, "grad_norm": 1.0703125, "learning_rate": 2.205481924805251e-05, "loss": 0.1159, "step": 33927 }, { "epoch": 24.855677655677656, "grad_norm": 1.234375, "learning_rate": 2.2052015184823737e-05, "loss": 0.1099, "step": 33928 }, { "epoch": 24.856410256410257, "grad_norm": 1.1015625, "learning_rate": 2.2049211257898604e-05, "loss": 0.1269, "step": 33929 }, { "epoch": 24.857142857142858, "grad_norm": 1.1796875, "learning_rate": 2.2046407467290532e-05, "loss": 0.1113, "step": 33930 }, { "epoch": 24.85787545787546, "grad_norm": 1.4765625, "learning_rate": 2.2043603813012982e-05, "loss": 0.1444, "step": 33931 }, { "epoch": 24.85860805860806, "grad_norm": 1.0, "learning_rate": 2.2040800295079354e-05, "loss": 0.1103, "step": 33932 }, { "epoch": 24.85934065934066, "grad_norm": 1.0703125, "learning_rate": 2.203799691350309e-05, "loss": 0.1188, "step": 33933 }, { "epoch": 24.86007326007326, "grad_norm": 0.96484375, "learning_rate": 2.203519366829763e-05, "loss": 0.106, "step": 33934 }, { "epoch": 24.86080586080586, "grad_norm": 1.9453125, "learning_rate": 2.2032390559476397e-05, "loss": 0.1357, "step": 33935 }, { "epoch": 24.861538461538462, "grad_norm": 0.85546875, "learning_rate": 2.2029587587052848e-05, "loss": 0.132, "step": 33936 }, { "epoch": 24.862271062271063, "grad_norm": 0.80859375, "learning_rate": 2.202678475104037e-05, "loss": 0.0985, "step": 33937 }, { "epoch": 24.863003663003663, "grad_norm": 0.8515625, "learning_rate": 2.2023982051452432e-05, "loss": 0.1049, "step": 33938 }, { "epoch": 24.863736263736264, "grad_norm": 0.87890625, "learning_rate": 2.202117948830241e-05, "loss": 0.0889, "step": 33939 }, { "epoch": 24.864468864468865, "grad_norm": 0.6171875, "learning_rate": 2.20183770616038e-05, "loss": 0.088, "step": 33940 }, { "epoch": 24.865201465201466, "grad_norm": 1.1796875, "learning_rate": 2.201557477136998e-05, "loss": 0.1334, "step": 33941 }, { "epoch": 24.865934065934066, "grad_norm": 1.4375, "learning_rate": 2.201277261761441e-05, "loss": 0.1287, "step": 33942 }, { "epoch": 24.866666666666667, "grad_norm": 0.71484375, "learning_rate": 2.200997060035048e-05, "loss": 0.127, "step": 33943 }, { "epoch": 24.867399267399268, "grad_norm": 1.6015625, "learning_rate": 2.200716871959163e-05, "loss": 0.1374, "step": 33944 }, { "epoch": 24.86813186813187, "grad_norm": 1.2890625, "learning_rate": 2.200436697535129e-05, "loss": 0.1062, "step": 33945 }, { "epoch": 24.86886446886447, "grad_norm": 1.265625, "learning_rate": 2.200156536764288e-05, "loss": 0.1054, "step": 33946 }, { "epoch": 24.86959706959707, "grad_norm": 0.91796875, "learning_rate": 2.199876389647985e-05, "loss": 0.1213, "step": 33947 }, { "epoch": 24.87032967032967, "grad_norm": 0.96875, "learning_rate": 2.1995962561875575e-05, "loss": 0.1103, "step": 33948 }, { "epoch": 24.87106227106227, "grad_norm": 0.90625, "learning_rate": 2.199316136384352e-05, "loss": 0.1074, "step": 33949 }, { "epoch": 24.871794871794872, "grad_norm": 1.34375, "learning_rate": 2.1990360302397053e-05, "loss": 0.1159, "step": 33950 }, { "epoch": 24.872527472527473, "grad_norm": 0.9140625, "learning_rate": 2.1987559377549666e-05, "loss": 0.1223, "step": 33951 }, { "epoch": 24.873260073260074, "grad_norm": 0.73046875, "learning_rate": 2.198475858931472e-05, "loss": 0.1141, "step": 33952 }, { "epoch": 24.873992673992674, "grad_norm": 1.0546875, "learning_rate": 2.1981957937705676e-05, "loss": 0.1122, "step": 33953 }, { "epoch": 24.874725274725275, "grad_norm": 1.046875, "learning_rate": 2.197915742273592e-05, "loss": 0.1185, "step": 33954 }, { "epoch": 24.875457875457876, "grad_norm": 0.6484375, "learning_rate": 2.1976357044418876e-05, "loss": 0.114, "step": 33955 }, { "epoch": 24.876190476190477, "grad_norm": 1.1796875, "learning_rate": 2.197355680276797e-05, "loss": 0.1571, "step": 33956 }, { "epoch": 24.876923076923077, "grad_norm": 1.03125, "learning_rate": 2.1970756697796638e-05, "loss": 0.1323, "step": 33957 }, { "epoch": 24.877655677655678, "grad_norm": 0.875, "learning_rate": 2.196795672951826e-05, "loss": 0.1005, "step": 33958 }, { "epoch": 24.87838827838828, "grad_norm": 1.34375, "learning_rate": 2.1965156897946265e-05, "loss": 0.1616, "step": 33959 }, { "epoch": 24.87912087912088, "grad_norm": 1.046875, "learning_rate": 2.1962357203094076e-05, "loss": 0.1077, "step": 33960 }, { "epoch": 24.87985347985348, "grad_norm": 0.9921875, "learning_rate": 2.1959557644975096e-05, "loss": 0.1148, "step": 33961 }, { "epoch": 24.88058608058608, "grad_norm": 0.890625, "learning_rate": 2.195675822360276e-05, "loss": 0.1168, "step": 33962 }, { "epoch": 24.88131868131868, "grad_norm": 1.046875, "learning_rate": 2.1953958938990454e-05, "loss": 0.12, "step": 33963 }, { "epoch": 24.882051282051282, "grad_norm": 1.546875, "learning_rate": 2.195115979115162e-05, "loss": 0.1331, "step": 33964 }, { "epoch": 24.882783882783883, "grad_norm": 1.1015625, "learning_rate": 2.194836078009963e-05, "loss": 0.1204, "step": 33965 }, { "epoch": 24.883516483516484, "grad_norm": 1.1171875, "learning_rate": 2.194556190584792e-05, "loss": 0.1055, "step": 33966 }, { "epoch": 24.884249084249085, "grad_norm": 1.234375, "learning_rate": 2.1942763168409894e-05, "loss": 0.1084, "step": 33967 }, { "epoch": 24.884981684981685, "grad_norm": 1.171875, "learning_rate": 2.1939964567798983e-05, "loss": 0.1211, "step": 33968 }, { "epoch": 24.885714285714286, "grad_norm": 0.86328125, "learning_rate": 2.1937166104028552e-05, "loss": 0.1336, "step": 33969 }, { "epoch": 24.886446886446887, "grad_norm": 1.3828125, "learning_rate": 2.1934367777112045e-05, "loss": 0.1497, "step": 33970 }, { "epoch": 24.887179487179488, "grad_norm": 0.7265625, "learning_rate": 2.193156958706286e-05, "loss": 0.1211, "step": 33971 }, { "epoch": 24.88791208791209, "grad_norm": 0.55859375, "learning_rate": 2.1928771533894394e-05, "loss": 0.0792, "step": 33972 }, { "epoch": 24.88864468864469, "grad_norm": 0.7734375, "learning_rate": 2.1925973617620087e-05, "loss": 0.1371, "step": 33973 }, { "epoch": 24.88937728937729, "grad_norm": 0.66796875, "learning_rate": 2.1923175838253307e-05, "loss": 0.1062, "step": 33974 }, { "epoch": 24.89010989010989, "grad_norm": 1.125, "learning_rate": 2.1920378195807482e-05, "loss": 0.15, "step": 33975 }, { "epoch": 24.89084249084249, "grad_norm": 1.15625, "learning_rate": 2.191758069029598e-05, "loss": 0.1113, "step": 33976 }, { "epoch": 24.891575091575092, "grad_norm": 0.84375, "learning_rate": 2.191478332173226e-05, "loss": 0.1518, "step": 33977 }, { "epoch": 24.892307692307693, "grad_norm": 1.0625, "learning_rate": 2.1911986090129684e-05, "loss": 0.1475, "step": 33978 }, { "epoch": 24.893040293040293, "grad_norm": 1.15625, "learning_rate": 2.1909188995501683e-05, "loss": 0.0966, "step": 33979 }, { "epoch": 24.893772893772894, "grad_norm": 1.171875, "learning_rate": 2.190639203786162e-05, "loss": 0.1209, "step": 33980 }, { "epoch": 24.894505494505495, "grad_norm": 1.046875, "learning_rate": 2.190359521722293e-05, "loss": 0.1336, "step": 33981 }, { "epoch": 24.895238095238096, "grad_norm": 1.171875, "learning_rate": 2.1900798533599e-05, "loss": 0.1172, "step": 33982 }, { "epoch": 24.895970695970696, "grad_norm": 0.91015625, "learning_rate": 2.1898001987003243e-05, "loss": 0.1322, "step": 33983 }, { "epoch": 24.896703296703297, "grad_norm": 1.46875, "learning_rate": 2.189520557744904e-05, "loss": 0.159, "step": 33984 }, { "epoch": 24.897435897435898, "grad_norm": 0.86328125, "learning_rate": 2.189240930494979e-05, "loss": 0.0989, "step": 33985 }, { "epoch": 24.8981684981685, "grad_norm": 1.3359375, "learning_rate": 2.18896131695189e-05, "loss": 0.1359, "step": 33986 }, { "epoch": 24.8989010989011, "grad_norm": 1.140625, "learning_rate": 2.1886817171169768e-05, "loss": 0.1412, "step": 33987 }, { "epoch": 24.8996336996337, "grad_norm": 1.5078125, "learning_rate": 2.188402130991581e-05, "loss": 0.1353, "step": 33988 }, { "epoch": 24.9003663003663, "grad_norm": 1.3046875, "learning_rate": 2.1881225585770373e-05, "loss": 0.1217, "step": 33989 }, { "epoch": 24.9010989010989, "grad_norm": 1.4609375, "learning_rate": 2.18784299987469e-05, "loss": 0.1327, "step": 33990 }, { "epoch": 24.901831501831502, "grad_norm": 0.7890625, "learning_rate": 2.1875634548858746e-05, "loss": 0.1188, "step": 33991 }, { "epoch": 24.902564102564103, "grad_norm": 1.234375, "learning_rate": 2.1872839236119326e-05, "loss": 0.1695, "step": 33992 }, { "epoch": 24.903296703296704, "grad_norm": 1.0390625, "learning_rate": 2.187004406054203e-05, "loss": 0.1464, "step": 33993 }, { "epoch": 24.904029304029304, "grad_norm": 1.0390625, "learning_rate": 2.186724902214027e-05, "loss": 0.1342, "step": 33994 }, { "epoch": 24.904761904761905, "grad_norm": 1.28125, "learning_rate": 2.1864454120927405e-05, "loss": 0.1724, "step": 33995 }, { "epoch": 24.905494505494506, "grad_norm": 0.8671875, "learning_rate": 2.1861659356916837e-05, "loss": 0.1062, "step": 33996 }, { "epoch": 24.906227106227107, "grad_norm": 1.546875, "learning_rate": 2.185886473012196e-05, "loss": 0.1578, "step": 33997 }, { "epoch": 24.906959706959707, "grad_norm": 1.1796875, "learning_rate": 2.1856070240556165e-05, "loss": 0.1487, "step": 33998 }, { "epoch": 24.907692307692308, "grad_norm": 0.828125, "learning_rate": 2.1853275888232858e-05, "loss": 0.1145, "step": 33999 }, { "epoch": 24.90842490842491, "grad_norm": 1.5234375, "learning_rate": 2.1850481673165393e-05, "loss": 0.1436, "step": 34000 }, { "epoch": 24.90915750915751, "grad_norm": 1.84375, "learning_rate": 2.184768759536719e-05, "loss": 0.131, "step": 34001 }, { "epoch": 24.90989010989011, "grad_norm": 1.3984375, "learning_rate": 2.1844893654851586e-05, "loss": 0.1376, "step": 34002 }, { "epoch": 24.91062271062271, "grad_norm": 1.0390625, "learning_rate": 2.184209985163204e-05, "loss": 0.1212, "step": 34003 }, { "epoch": 24.91135531135531, "grad_norm": 1.5390625, "learning_rate": 2.183930618572189e-05, "loss": 0.1619, "step": 34004 }, { "epoch": 24.912087912087912, "grad_norm": 1.0859375, "learning_rate": 2.1836512657134547e-05, "loss": 0.1386, "step": 34005 }, { "epoch": 24.912820512820513, "grad_norm": 0.80078125, "learning_rate": 2.1833719265883364e-05, "loss": 0.1113, "step": 34006 }, { "epoch": 24.913553113553114, "grad_norm": 0.8984375, "learning_rate": 2.183092601198174e-05, "loss": 0.132, "step": 34007 }, { "epoch": 24.914285714285715, "grad_norm": 1.0, "learning_rate": 2.1828132895443058e-05, "loss": 0.1424, "step": 34008 }, { "epoch": 24.915018315018315, "grad_norm": 0.78515625, "learning_rate": 2.1825339916280705e-05, "loss": 0.1272, "step": 34009 }, { "epoch": 24.915750915750916, "grad_norm": 1.796875, "learning_rate": 2.1822547074508086e-05, "loss": 0.1127, "step": 34010 }, { "epoch": 24.916483516483517, "grad_norm": 1.046875, "learning_rate": 2.1819754370138536e-05, "loss": 0.1113, "step": 34011 }, { "epoch": 24.917216117216118, "grad_norm": 0.84375, "learning_rate": 2.181696180318546e-05, "loss": 0.1122, "step": 34012 }, { "epoch": 24.91794871794872, "grad_norm": 1.6328125, "learning_rate": 2.181416937366224e-05, "loss": 0.1325, "step": 34013 }, { "epoch": 24.91868131868132, "grad_norm": 1.3203125, "learning_rate": 2.1811377081582263e-05, "loss": 0.1168, "step": 34014 }, { "epoch": 24.91941391941392, "grad_norm": 0.9765625, "learning_rate": 2.1808584926958874e-05, "loss": 0.1331, "step": 34015 }, { "epoch": 24.92014652014652, "grad_norm": 0.98828125, "learning_rate": 2.18057929098055e-05, "loss": 0.13, "step": 34016 }, { "epoch": 24.92087912087912, "grad_norm": 0.74609375, "learning_rate": 2.1803001030135468e-05, "loss": 0.1355, "step": 34017 }, { "epoch": 24.921611721611722, "grad_norm": 1.6796875, "learning_rate": 2.1800209287962182e-05, "loss": 0.1724, "step": 34018 }, { "epoch": 24.922344322344323, "grad_norm": 0.82421875, "learning_rate": 2.1797417683299014e-05, "loss": 0.1003, "step": 34019 }, { "epoch": 24.923076923076923, "grad_norm": 1.0, "learning_rate": 2.1794626216159356e-05, "loss": 0.1219, "step": 34020 }, { "epoch": 24.923809523809524, "grad_norm": 1.2109375, "learning_rate": 2.179183488655655e-05, "loss": 0.1108, "step": 34021 }, { "epoch": 24.924542124542125, "grad_norm": 1.2734375, "learning_rate": 2.1789043694503993e-05, "loss": 0.1287, "step": 34022 }, { "epoch": 24.925274725274726, "grad_norm": 0.82421875, "learning_rate": 2.178625264001505e-05, "loss": 0.1291, "step": 34023 }, { "epoch": 24.926007326007326, "grad_norm": 0.875, "learning_rate": 2.1783461723103095e-05, "loss": 0.1376, "step": 34024 }, { "epoch": 24.926739926739927, "grad_norm": 0.70703125, "learning_rate": 2.1780670943781522e-05, "loss": 0.1105, "step": 34025 }, { "epoch": 24.927472527472528, "grad_norm": 0.7890625, "learning_rate": 2.177788030206366e-05, "loss": 0.0985, "step": 34026 }, { "epoch": 24.92820512820513, "grad_norm": 0.7421875, "learning_rate": 2.1775089797962923e-05, "loss": 0.1311, "step": 34027 }, { "epoch": 24.92893772893773, "grad_norm": 0.875, "learning_rate": 2.1772299431492626e-05, "loss": 0.1433, "step": 34028 }, { "epoch": 24.92967032967033, "grad_norm": 1.171875, "learning_rate": 2.1769509202666206e-05, "loss": 0.1223, "step": 34029 }, { "epoch": 24.93040293040293, "grad_norm": 1.0625, "learning_rate": 2.1766719111496985e-05, "loss": 0.1166, "step": 34030 }, { "epoch": 24.93113553113553, "grad_norm": 0.84375, "learning_rate": 2.176392915799836e-05, "loss": 0.1093, "step": 34031 }, { "epoch": 24.931868131868132, "grad_norm": 1.0703125, "learning_rate": 2.1761139342183666e-05, "loss": 0.1217, "step": 34032 }, { "epoch": 24.932600732600733, "grad_norm": 0.765625, "learning_rate": 2.175834966406629e-05, "loss": 0.118, "step": 34033 }, { "epoch": 24.933333333333334, "grad_norm": 1.3203125, "learning_rate": 2.1755560123659596e-05, "loss": 0.123, "step": 34034 }, { "epoch": 24.934065934065934, "grad_norm": 0.98046875, "learning_rate": 2.1752770720976946e-05, "loss": 0.1287, "step": 34035 }, { "epoch": 24.934798534798535, "grad_norm": 1.3515625, "learning_rate": 2.1749981456031725e-05, "loss": 0.1129, "step": 34036 }, { "epoch": 24.935531135531136, "grad_norm": 1.0859375, "learning_rate": 2.1747192328837268e-05, "loss": 0.1217, "step": 34037 }, { "epoch": 24.936263736263736, "grad_norm": 0.94921875, "learning_rate": 2.1744403339406952e-05, "loss": 0.1289, "step": 34038 }, { "epoch": 24.936996336996337, "grad_norm": 0.9453125, "learning_rate": 2.1741614487754135e-05, "loss": 0.1181, "step": 34039 }, { "epoch": 24.937728937728938, "grad_norm": 1.1328125, "learning_rate": 2.173882577389221e-05, "loss": 0.1282, "step": 34040 }, { "epoch": 24.93846153846154, "grad_norm": 1.1640625, "learning_rate": 2.1736037197834487e-05, "loss": 0.1318, "step": 34041 }, { "epoch": 24.93919413919414, "grad_norm": 0.953125, "learning_rate": 2.173324875959437e-05, "loss": 0.0956, "step": 34042 }, { "epoch": 24.93992673992674, "grad_norm": 0.80078125, "learning_rate": 2.1730460459185186e-05, "loss": 0.1247, "step": 34043 }, { "epoch": 24.94065934065934, "grad_norm": 1.015625, "learning_rate": 2.1727672296620308e-05, "loss": 0.1375, "step": 34044 }, { "epoch": 24.94139194139194, "grad_norm": 1.140625, "learning_rate": 2.1724884271913103e-05, "loss": 0.1039, "step": 34045 }, { "epoch": 24.942124542124542, "grad_norm": 0.8828125, "learning_rate": 2.172209638507694e-05, "loss": 0.1243, "step": 34046 }, { "epoch": 24.942857142857143, "grad_norm": 1.0390625, "learning_rate": 2.1719308636125137e-05, "loss": 0.1322, "step": 34047 }, { "epoch": 24.943589743589744, "grad_norm": 0.859375, "learning_rate": 2.1716521025071083e-05, "loss": 0.1199, "step": 34048 }, { "epoch": 24.944322344322345, "grad_norm": 1.0, "learning_rate": 2.171373355192812e-05, "loss": 0.1325, "step": 34049 }, { "epoch": 24.945054945054945, "grad_norm": 0.8828125, "learning_rate": 2.171094621670961e-05, "loss": 0.1161, "step": 34050 }, { "epoch": 24.945787545787546, "grad_norm": 0.76953125, "learning_rate": 2.1708159019428924e-05, "loss": 0.1174, "step": 34051 }, { "epoch": 24.946520146520147, "grad_norm": 0.83203125, "learning_rate": 2.1705371960099385e-05, "loss": 0.1224, "step": 34052 }, { "epoch": 24.947252747252747, "grad_norm": 0.828125, "learning_rate": 2.1702585038734376e-05, "loss": 0.1147, "step": 34053 }, { "epoch": 24.947985347985348, "grad_norm": 0.58984375, "learning_rate": 2.1699798255347202e-05, "loss": 0.1031, "step": 34054 }, { "epoch": 24.94871794871795, "grad_norm": 0.9140625, "learning_rate": 2.1697011609951287e-05, "loss": 0.1403, "step": 34055 }, { "epoch": 24.94945054945055, "grad_norm": 1.0703125, "learning_rate": 2.169422510255992e-05, "loss": 0.1202, "step": 34056 }, { "epoch": 24.95018315018315, "grad_norm": 1.0546875, "learning_rate": 2.16914387331865e-05, "loss": 0.1096, "step": 34057 }, { "epoch": 24.95091575091575, "grad_norm": 0.84375, "learning_rate": 2.1688652501844338e-05, "loss": 0.131, "step": 34058 }, { "epoch": 24.951648351648352, "grad_norm": 0.81640625, "learning_rate": 2.1685866408546795e-05, "loss": 0.1332, "step": 34059 }, { "epoch": 24.952380952380953, "grad_norm": 1.0, "learning_rate": 2.168308045330723e-05, "loss": 0.1274, "step": 34060 }, { "epoch": 24.953113553113553, "grad_norm": 1.40625, "learning_rate": 2.1680294636138986e-05, "loss": 0.1426, "step": 34061 }, { "epoch": 24.953846153846154, "grad_norm": 0.88671875, "learning_rate": 2.1677508957055425e-05, "loss": 0.1291, "step": 34062 }, { "epoch": 24.954578754578755, "grad_norm": 1.2109375, "learning_rate": 2.1674723416069865e-05, "loss": 0.1563, "step": 34063 }, { "epoch": 24.955311355311355, "grad_norm": 0.66796875, "learning_rate": 2.167193801319567e-05, "loss": 0.1404, "step": 34064 }, { "epoch": 24.956043956043956, "grad_norm": 1.1875, "learning_rate": 2.166915274844618e-05, "loss": 0.126, "step": 34065 }, { "epoch": 24.956776556776557, "grad_norm": 0.80078125, "learning_rate": 2.1666367621834764e-05, "loss": 0.106, "step": 34066 }, { "epoch": 24.957509157509158, "grad_norm": 0.99609375, "learning_rate": 2.166358263337473e-05, "loss": 0.131, "step": 34067 }, { "epoch": 24.95824175824176, "grad_norm": 1.109375, "learning_rate": 2.1660797783079452e-05, "loss": 0.0931, "step": 34068 }, { "epoch": 24.95897435897436, "grad_norm": 0.89453125, "learning_rate": 2.1658013070962242e-05, "loss": 0.1202, "step": 34069 }, { "epoch": 24.95970695970696, "grad_norm": 0.91796875, "learning_rate": 2.165522849703646e-05, "loss": 0.1244, "step": 34070 }, { "epoch": 24.96043956043956, "grad_norm": 1.3359375, "learning_rate": 2.1652444061315448e-05, "loss": 0.1457, "step": 34071 }, { "epoch": 24.96117216117216, "grad_norm": 0.74609375, "learning_rate": 2.1649659763812565e-05, "loss": 0.1081, "step": 34072 }, { "epoch": 24.961904761904762, "grad_norm": 0.8046875, "learning_rate": 2.1646875604541112e-05, "loss": 0.1167, "step": 34073 }, { "epoch": 24.962637362637363, "grad_norm": 0.71484375, "learning_rate": 2.1644091583514452e-05, "loss": 0.112, "step": 34074 }, { "epoch": 24.963369963369964, "grad_norm": 1.2578125, "learning_rate": 2.1641307700745924e-05, "loss": 0.1214, "step": 34075 }, { "epoch": 24.964102564102564, "grad_norm": 1.75, "learning_rate": 2.163852395624886e-05, "loss": 0.1276, "step": 34076 }, { "epoch": 24.964835164835165, "grad_norm": 1.0, "learning_rate": 2.1635740350036626e-05, "loss": 0.1263, "step": 34077 }, { "epoch": 24.965567765567766, "grad_norm": 0.95703125, "learning_rate": 2.163295688212251e-05, "loss": 0.0997, "step": 34078 }, { "epoch": 24.966300366300366, "grad_norm": 0.84765625, "learning_rate": 2.1630173552519898e-05, "loss": 0.1055, "step": 34079 }, { "epoch": 24.967032967032967, "grad_norm": 1.1796875, "learning_rate": 2.1627390361242064e-05, "loss": 0.1209, "step": 34080 }, { "epoch": 24.967765567765568, "grad_norm": 1.1484375, "learning_rate": 2.1624607308302417e-05, "loss": 0.1171, "step": 34081 }, { "epoch": 24.96849816849817, "grad_norm": 1.1640625, "learning_rate": 2.1621824393714242e-05, "loss": 0.0896, "step": 34082 }, { "epoch": 24.96923076923077, "grad_norm": 0.8125, "learning_rate": 2.1619041617490898e-05, "loss": 0.1011, "step": 34083 }, { "epoch": 24.96996336996337, "grad_norm": 1.46875, "learning_rate": 2.1616258979645693e-05, "loss": 0.1184, "step": 34084 }, { "epoch": 24.97069597069597, "grad_norm": 0.94140625, "learning_rate": 2.1613476480191967e-05, "loss": 0.1007, "step": 34085 }, { "epoch": 24.97142857142857, "grad_norm": 1.1328125, "learning_rate": 2.161069411914306e-05, "loss": 0.13, "step": 34086 }, { "epoch": 24.972161172161172, "grad_norm": 1.2265625, "learning_rate": 2.1607911896512306e-05, "loss": 0.1018, "step": 34087 }, { "epoch": 24.972893772893773, "grad_norm": 0.9765625, "learning_rate": 2.1605129812313043e-05, "loss": 0.1112, "step": 34088 }, { "epoch": 24.973626373626374, "grad_norm": 1.390625, "learning_rate": 2.1602347866558578e-05, "loss": 0.1187, "step": 34089 }, { "epoch": 24.974358974358974, "grad_norm": 0.8828125, "learning_rate": 2.1599566059262246e-05, "loss": 0.1026, "step": 34090 }, { "epoch": 24.975091575091575, "grad_norm": 0.703125, "learning_rate": 2.159678439043738e-05, "loss": 0.0995, "step": 34091 }, { "epoch": 24.975824175824176, "grad_norm": 0.80859375, "learning_rate": 2.1594002860097325e-05, "loss": 0.1053, "step": 34092 }, { "epoch": 24.976556776556777, "grad_norm": 1.296875, "learning_rate": 2.159122146825538e-05, "loss": 0.0996, "step": 34093 }, { "epoch": 24.977289377289377, "grad_norm": 1.1015625, "learning_rate": 2.1588440214924894e-05, "loss": 0.1206, "step": 34094 }, { "epoch": 24.978021978021978, "grad_norm": 1.203125, "learning_rate": 2.1585659100119166e-05, "loss": 0.1242, "step": 34095 }, { "epoch": 24.97875457875458, "grad_norm": 1.0, "learning_rate": 2.1582878123851537e-05, "loss": 0.107, "step": 34096 }, { "epoch": 24.97948717948718, "grad_norm": 1.2265625, "learning_rate": 2.1580097286135335e-05, "loss": 0.1111, "step": 34097 }, { "epoch": 24.98021978021978, "grad_norm": 0.77734375, "learning_rate": 2.1577316586983893e-05, "loss": 0.1, "step": 34098 }, { "epoch": 24.98095238095238, "grad_norm": 0.76171875, "learning_rate": 2.1574536026410508e-05, "loss": 0.1135, "step": 34099 }, { "epoch": 24.98168498168498, "grad_norm": 1.53125, "learning_rate": 2.1571755604428518e-05, "loss": 0.1115, "step": 34100 }, { "epoch": 24.982417582417582, "grad_norm": 0.89453125, "learning_rate": 2.1568975321051238e-05, "loss": 0.1262, "step": 34101 }, { "epoch": 24.983150183150183, "grad_norm": 0.640625, "learning_rate": 2.1566195176292e-05, "loss": 0.0906, "step": 34102 }, { "epoch": 24.983882783882784, "grad_norm": 1.0625, "learning_rate": 2.1563415170164134e-05, "loss": 0.1095, "step": 34103 }, { "epoch": 24.984615384615385, "grad_norm": 0.9140625, "learning_rate": 2.156063530268093e-05, "loss": 0.1442, "step": 34104 }, { "epoch": 24.985347985347985, "grad_norm": 0.984375, "learning_rate": 2.1557855573855737e-05, "loss": 0.1176, "step": 34105 }, { "epoch": 24.986080586080586, "grad_norm": 1.1796875, "learning_rate": 2.1555075983701826e-05, "loss": 0.1064, "step": 34106 }, { "epoch": 24.986813186813187, "grad_norm": 1.1015625, "learning_rate": 2.1552296532232586e-05, "loss": 0.1041, "step": 34107 }, { "epoch": 24.987545787545788, "grad_norm": 1.5, "learning_rate": 2.1549517219461274e-05, "loss": 0.1574, "step": 34108 }, { "epoch": 24.98827838827839, "grad_norm": 0.83203125, "learning_rate": 2.1546738045401243e-05, "loss": 0.1055, "step": 34109 }, { "epoch": 24.98901098901099, "grad_norm": 1.234375, "learning_rate": 2.1543959010065784e-05, "loss": 0.1306, "step": 34110 }, { "epoch": 24.98974358974359, "grad_norm": 1.1875, "learning_rate": 2.1541180113468217e-05, "loss": 0.1239, "step": 34111 }, { "epoch": 24.99047619047619, "grad_norm": 1.71875, "learning_rate": 2.1538401355621866e-05, "loss": 0.1107, "step": 34112 }, { "epoch": 24.99120879120879, "grad_norm": 1.21875, "learning_rate": 2.153562273654004e-05, "loss": 0.136, "step": 34113 }, { "epoch": 24.991941391941392, "grad_norm": 1.6015625, "learning_rate": 2.1532844256236078e-05, "loss": 0.1341, "step": 34114 }, { "epoch": 24.992673992673993, "grad_norm": 1.1953125, "learning_rate": 2.1530065914723246e-05, "loss": 0.1475, "step": 34115 }, { "epoch": 24.993406593406593, "grad_norm": 0.81640625, "learning_rate": 2.152728771201488e-05, "loss": 0.0937, "step": 34116 }, { "epoch": 24.994139194139194, "grad_norm": 0.6796875, "learning_rate": 2.1524509648124287e-05, "loss": 0.1043, "step": 34117 }, { "epoch": 24.994871794871795, "grad_norm": 0.8671875, "learning_rate": 2.15217317230648e-05, "loss": 0.1088, "step": 34118 }, { "epoch": 24.995604395604396, "grad_norm": 0.84765625, "learning_rate": 2.151895393684969e-05, "loss": 0.1238, "step": 34119 }, { "epoch": 24.996336996336996, "grad_norm": 1.5546875, "learning_rate": 2.1516176289492304e-05, "loss": 0.1093, "step": 34120 }, { "epoch": 24.997069597069597, "grad_norm": 1.125, "learning_rate": 2.151339878100591e-05, "loss": 0.1179, "step": 34121 }, { "epoch": 24.997802197802198, "grad_norm": 2.09375, "learning_rate": 2.151062141140385e-05, "loss": 0.1183, "step": 34122 }, { "epoch": 24.9985347985348, "grad_norm": 1.125, "learning_rate": 2.150784418069941e-05, "loss": 0.1289, "step": 34123 }, { "epoch": 24.9992673992674, "grad_norm": 0.92578125, "learning_rate": 2.1505067088905908e-05, "loss": 0.1089, "step": 34124 }, { "epoch": 25.0, "grad_norm": 0.9140625, "learning_rate": 2.1502290136036664e-05, "loss": 0.0866, "step": 34125 }, { "epoch": 25.0007326007326, "grad_norm": 0.80078125, "learning_rate": 2.149951332210495e-05, "loss": 0.1, "step": 34126 }, { "epoch": 25.0014652014652, "grad_norm": 1.03125, "learning_rate": 2.1496736647124094e-05, "loss": 0.1403, "step": 34127 }, { "epoch": 25.002197802197802, "grad_norm": 0.953125, "learning_rate": 2.1493960111107384e-05, "loss": 0.115, "step": 34128 }, { "epoch": 25.002930402930403, "grad_norm": 1.203125, "learning_rate": 2.1491183714068157e-05, "loss": 0.1152, "step": 34129 }, { "epoch": 25.003663003663004, "grad_norm": 1.03125, "learning_rate": 2.1488407456019677e-05, "loss": 0.1263, "step": 34130 }, { "epoch": 25.004395604395604, "grad_norm": 1.2109375, "learning_rate": 2.148563133697527e-05, "loss": 0.1097, "step": 34131 }, { "epoch": 25.005128205128205, "grad_norm": 1.296875, "learning_rate": 2.14828553569482e-05, "loss": 0.0991, "step": 34132 }, { "epoch": 25.005860805860806, "grad_norm": 0.94140625, "learning_rate": 2.1480079515951828e-05, "loss": 0.096, "step": 34133 }, { "epoch": 25.006593406593407, "grad_norm": 1.078125, "learning_rate": 2.1477303813999404e-05, "loss": 0.0933, "step": 34134 }, { "epoch": 25.007326007326007, "grad_norm": 1.6796875, "learning_rate": 2.1474528251104265e-05, "loss": 0.1162, "step": 34135 }, { "epoch": 25.008058608058608, "grad_norm": 0.6640625, "learning_rate": 2.1471752827279663e-05, "loss": 0.1069, "step": 34136 }, { "epoch": 25.00879120879121, "grad_norm": 0.5703125, "learning_rate": 2.1468977542538936e-05, "loss": 0.0948, "step": 34137 }, { "epoch": 25.00952380952381, "grad_norm": 1.0625, "learning_rate": 2.1466202396895356e-05, "loss": 0.1244, "step": 34138 }, { "epoch": 25.01025641025641, "grad_norm": 1.046875, "learning_rate": 2.146342739036224e-05, "loss": 0.1347, "step": 34139 }, { "epoch": 25.01098901098901, "grad_norm": 1.2734375, "learning_rate": 2.146065252295289e-05, "loss": 0.1075, "step": 34140 }, { "epoch": 25.01172161172161, "grad_norm": 1.4921875, "learning_rate": 2.145787779468057e-05, "loss": 0.1637, "step": 34141 }, { "epoch": 25.012454212454212, "grad_norm": 1.0546875, "learning_rate": 2.1455103205558588e-05, "loss": 0.0993, "step": 34142 }, { "epoch": 25.013186813186813, "grad_norm": 1.484375, "learning_rate": 2.145232875560024e-05, "loss": 0.1175, "step": 34143 }, { "epoch": 25.013919413919414, "grad_norm": 0.6953125, "learning_rate": 2.1449554444818847e-05, "loss": 0.1235, "step": 34144 }, { "epoch": 25.014652014652015, "grad_norm": 1.1640625, "learning_rate": 2.1446780273227653e-05, "loss": 0.1155, "step": 34145 }, { "epoch": 25.015384615384615, "grad_norm": 0.87109375, "learning_rate": 2.1444006240839988e-05, "loss": 0.1653, "step": 34146 }, { "epoch": 25.016117216117216, "grad_norm": 1.140625, "learning_rate": 2.1441232347669113e-05, "loss": 0.1652, "step": 34147 }, { "epoch": 25.016849816849817, "grad_norm": 0.8359375, "learning_rate": 2.143845859372833e-05, "loss": 0.1315, "step": 34148 }, { "epoch": 25.017582417582418, "grad_norm": 1.4453125, "learning_rate": 2.1435684979030935e-05, "loss": 0.1051, "step": 34149 }, { "epoch": 25.01831501831502, "grad_norm": 1.5390625, "learning_rate": 2.1432911503590214e-05, "loss": 0.1311, "step": 34150 }, { "epoch": 25.01904761904762, "grad_norm": 0.7578125, "learning_rate": 2.1430138167419475e-05, "loss": 0.1114, "step": 34151 }, { "epoch": 25.01978021978022, "grad_norm": 1.0390625, "learning_rate": 2.1427364970531966e-05, "loss": 0.104, "step": 34152 }, { "epoch": 25.02051282051282, "grad_norm": 0.9765625, "learning_rate": 2.1424591912940998e-05, "loss": 0.1377, "step": 34153 }, { "epoch": 25.02124542124542, "grad_norm": 1.0078125, "learning_rate": 2.142181899465985e-05, "loss": 0.1144, "step": 34154 }, { "epoch": 25.021978021978022, "grad_norm": 1.6640625, "learning_rate": 2.1419046215701834e-05, "loss": 0.124, "step": 34155 }, { "epoch": 25.022710622710623, "grad_norm": 1.1015625, "learning_rate": 2.14162735760802e-05, "loss": 0.1209, "step": 34156 }, { "epoch": 25.023443223443223, "grad_norm": 1.0, "learning_rate": 2.141350107580826e-05, "loss": 0.1236, "step": 34157 }, { "epoch": 25.024175824175824, "grad_norm": 1.0859375, "learning_rate": 2.1410728714899248e-05, "loss": 0.1048, "step": 34158 }, { "epoch": 25.024908424908425, "grad_norm": 0.58203125, "learning_rate": 2.1407956493366526e-05, "loss": 0.1027, "step": 34159 }, { "epoch": 25.025641025641026, "grad_norm": 1.03125, "learning_rate": 2.140518441122331e-05, "loss": 0.0974, "step": 34160 }, { "epoch": 25.026373626373626, "grad_norm": 1.9140625, "learning_rate": 2.140241246848293e-05, "loss": 0.1742, "step": 34161 }, { "epoch": 25.027106227106227, "grad_norm": 1.421875, "learning_rate": 2.1399640665158623e-05, "loss": 0.1109, "step": 34162 }, { "epoch": 25.027838827838828, "grad_norm": 0.91015625, "learning_rate": 2.1396869001263696e-05, "loss": 0.1143, "step": 34163 }, { "epoch": 25.02857142857143, "grad_norm": 1.421875, "learning_rate": 2.1394097476811415e-05, "loss": 0.1389, "step": 34164 }, { "epoch": 25.02930402930403, "grad_norm": 0.9296875, "learning_rate": 2.1391326091815072e-05, "loss": 0.1124, "step": 34165 }, { "epoch": 25.03003663003663, "grad_norm": 0.91796875, "learning_rate": 2.138855484628796e-05, "loss": 0.1032, "step": 34166 }, { "epoch": 25.03076923076923, "grad_norm": 1.2890625, "learning_rate": 2.1385783740243317e-05, "loss": 0.1149, "step": 34167 }, { "epoch": 25.03150183150183, "grad_norm": 1.546875, "learning_rate": 2.1383012773694468e-05, "loss": 0.1243, "step": 34168 }, { "epoch": 25.032234432234432, "grad_norm": 1.828125, "learning_rate": 2.138024194665462e-05, "loss": 0.1443, "step": 34169 }, { "epoch": 25.032967032967033, "grad_norm": 0.84765625, "learning_rate": 2.1377471259137127e-05, "loss": 0.1026, "step": 34170 }, { "epoch": 25.033699633699634, "grad_norm": 0.671875, "learning_rate": 2.1374700711155212e-05, "loss": 0.1096, "step": 34171 }, { "epoch": 25.034432234432234, "grad_norm": 1.390625, "learning_rate": 2.137193030272219e-05, "loss": 0.1193, "step": 34172 }, { "epoch": 25.035164835164835, "grad_norm": 0.96484375, "learning_rate": 2.136916003385129e-05, "loss": 0.1185, "step": 34173 }, { "epoch": 25.035897435897436, "grad_norm": 1.5, "learning_rate": 2.136638990455581e-05, "loss": 0.14, "step": 34174 }, { "epoch": 25.036630036630036, "grad_norm": 0.87890625, "learning_rate": 2.136361991484902e-05, "loss": 0.1145, "step": 34175 }, { "epoch": 25.037362637362637, "grad_norm": 1.3203125, "learning_rate": 2.1360850064744194e-05, "loss": 0.1432, "step": 34176 }, { "epoch": 25.038095238095238, "grad_norm": 1.1484375, "learning_rate": 2.1358080354254613e-05, "loss": 0.1139, "step": 34177 }, { "epoch": 25.03882783882784, "grad_norm": 1.1796875, "learning_rate": 2.135531078339352e-05, "loss": 0.1108, "step": 34178 }, { "epoch": 25.03956043956044, "grad_norm": 0.94921875, "learning_rate": 2.13525413521742e-05, "loss": 0.1228, "step": 34179 }, { "epoch": 25.04029304029304, "grad_norm": 0.85546875, "learning_rate": 2.1349772060609924e-05, "loss": 0.0987, "step": 34180 }, { "epoch": 25.04102564102564, "grad_norm": 1.3984375, "learning_rate": 2.1347002908713978e-05, "loss": 0.1329, "step": 34181 }, { "epoch": 25.04175824175824, "grad_norm": 0.671875, "learning_rate": 2.1344233896499595e-05, "loss": 0.1045, "step": 34182 }, { "epoch": 25.042490842490842, "grad_norm": 1.3671875, "learning_rate": 2.1341465023980072e-05, "loss": 0.1067, "step": 34183 }, { "epoch": 25.043223443223443, "grad_norm": 0.99609375, "learning_rate": 2.133869629116863e-05, "loss": 0.1376, "step": 34184 }, { "epoch": 25.043956043956044, "grad_norm": 0.9140625, "learning_rate": 2.13359276980786e-05, "loss": 0.1133, "step": 34185 }, { "epoch": 25.044688644688645, "grad_norm": 1.0234375, "learning_rate": 2.13331592447232e-05, "loss": 0.1237, "step": 34186 }, { "epoch": 25.045421245421245, "grad_norm": 0.8671875, "learning_rate": 2.1330390931115724e-05, "loss": 0.1371, "step": 34187 }, { "epoch": 25.046153846153846, "grad_norm": 1.0546875, "learning_rate": 2.1327622757269406e-05, "loss": 0.1177, "step": 34188 }, { "epoch": 25.046886446886447, "grad_norm": 0.6171875, "learning_rate": 2.1324854723197517e-05, "loss": 0.1164, "step": 34189 }, { "epoch": 25.047619047619047, "grad_norm": 1.4296875, "learning_rate": 2.1322086828913332e-05, "loss": 0.0994, "step": 34190 }, { "epoch": 25.048351648351648, "grad_norm": 1.2890625, "learning_rate": 2.131931907443011e-05, "loss": 0.1269, "step": 34191 }, { "epoch": 25.04908424908425, "grad_norm": 1.375, "learning_rate": 2.1316551459761126e-05, "loss": 0.1306, "step": 34192 }, { "epoch": 25.04981684981685, "grad_norm": 1.3125, "learning_rate": 2.1313783984919604e-05, "loss": 0.1366, "step": 34193 }, { "epoch": 25.05054945054945, "grad_norm": 0.7109375, "learning_rate": 2.131101664991884e-05, "loss": 0.125, "step": 34194 }, { "epoch": 25.05128205128205, "grad_norm": 1.0625, "learning_rate": 2.1308249454772045e-05, "loss": 0.1096, "step": 34195 }, { "epoch": 25.052014652014652, "grad_norm": 0.72265625, "learning_rate": 2.1305482399492548e-05, "loss": 0.1129, "step": 34196 }, { "epoch": 25.052747252747253, "grad_norm": 1.0859375, "learning_rate": 2.1302715484093546e-05, "loss": 0.1118, "step": 34197 }, { "epoch": 25.053479853479853, "grad_norm": 0.90625, "learning_rate": 2.1299948708588336e-05, "loss": 0.1137, "step": 34198 }, { "epoch": 25.054212454212454, "grad_norm": 1.3828125, "learning_rate": 2.1297182072990144e-05, "loss": 0.1323, "step": 34199 }, { "epoch": 25.054945054945055, "grad_norm": 1.3046875, "learning_rate": 2.129441557731224e-05, "loss": 0.1305, "step": 34200 }, { "epoch": 25.055677655677655, "grad_norm": 1.390625, "learning_rate": 2.1291649221567873e-05, "loss": 0.1284, "step": 34201 }, { "epoch": 25.056410256410256, "grad_norm": 1.0234375, "learning_rate": 2.128888300577031e-05, "loss": 0.1312, "step": 34202 }, { "epoch": 25.057142857142857, "grad_norm": 1.390625, "learning_rate": 2.128611692993281e-05, "loss": 0.1367, "step": 34203 }, { "epoch": 25.057875457875458, "grad_norm": 0.984375, "learning_rate": 2.1283350994068597e-05, "loss": 0.1364, "step": 34204 }, { "epoch": 25.05860805860806, "grad_norm": 1.0234375, "learning_rate": 2.1280585198190944e-05, "loss": 0.1083, "step": 34205 }, { "epoch": 25.05934065934066, "grad_norm": 0.70703125, "learning_rate": 2.1277819542313095e-05, "loss": 0.1218, "step": 34206 }, { "epoch": 25.06007326007326, "grad_norm": 0.76953125, "learning_rate": 2.1275054026448322e-05, "loss": 0.1113, "step": 34207 }, { "epoch": 25.06080586080586, "grad_norm": 1.1328125, "learning_rate": 2.127228865060985e-05, "loss": 0.1328, "step": 34208 }, { "epoch": 25.06153846153846, "grad_norm": 1.125, "learning_rate": 2.1269523414810953e-05, "loss": 0.1005, "step": 34209 }, { "epoch": 25.062271062271062, "grad_norm": 0.72265625, "learning_rate": 2.126675831906483e-05, "loss": 0.1142, "step": 34210 }, { "epoch": 25.063003663003663, "grad_norm": 0.88671875, "learning_rate": 2.126399336338479e-05, "loss": 0.0936, "step": 34211 }, { "epoch": 25.063736263736264, "grad_norm": 1.296875, "learning_rate": 2.126122854778405e-05, "loss": 0.1499, "step": 34212 }, { "epoch": 25.064468864468864, "grad_norm": 0.99609375, "learning_rate": 2.1258463872275873e-05, "loss": 0.1386, "step": 34213 }, { "epoch": 25.065201465201465, "grad_norm": 1.859375, "learning_rate": 2.125569933687348e-05, "loss": 0.1375, "step": 34214 }, { "epoch": 25.065934065934066, "grad_norm": 1.046875, "learning_rate": 2.125293494159013e-05, "loss": 0.0991, "step": 34215 }, { "epoch": 25.066666666666666, "grad_norm": 0.89453125, "learning_rate": 2.1250170686439068e-05, "loss": 0.1344, "step": 34216 }, { "epoch": 25.067399267399267, "grad_norm": 1.125, "learning_rate": 2.124740657143354e-05, "loss": 0.1115, "step": 34217 }, { "epoch": 25.068131868131868, "grad_norm": 1.4296875, "learning_rate": 2.1244642596586817e-05, "loss": 0.1308, "step": 34218 }, { "epoch": 25.06886446886447, "grad_norm": 1.171875, "learning_rate": 2.124187876191209e-05, "loss": 0.1185, "step": 34219 }, { "epoch": 25.06959706959707, "grad_norm": 1.203125, "learning_rate": 2.1239115067422643e-05, "loss": 0.1097, "step": 34220 }, { "epoch": 25.07032967032967, "grad_norm": 0.63671875, "learning_rate": 2.1236351513131673e-05, "loss": 0.1129, "step": 34221 }, { "epoch": 25.07106227106227, "grad_norm": 0.9140625, "learning_rate": 2.123358809905248e-05, "loss": 0.1322, "step": 34222 }, { "epoch": 25.07179487179487, "grad_norm": 1.4765625, "learning_rate": 2.123082482519826e-05, "loss": 0.1158, "step": 34223 }, { "epoch": 25.072527472527472, "grad_norm": 0.9921875, "learning_rate": 2.122806169158228e-05, "loss": 0.1264, "step": 34224 }, { "epoch": 25.073260073260073, "grad_norm": 1.0234375, "learning_rate": 2.1225298698217745e-05, "loss": 0.1189, "step": 34225 }, { "epoch": 25.073992673992674, "grad_norm": 1.28125, "learning_rate": 2.1222535845117922e-05, "loss": 0.1033, "step": 34226 }, { "epoch": 25.074725274725274, "grad_norm": 1.3515625, "learning_rate": 2.1219773132296036e-05, "loss": 0.1429, "step": 34227 }, { "epoch": 25.075457875457875, "grad_norm": 1.203125, "learning_rate": 2.121701055976533e-05, "loss": 0.1218, "step": 34228 }, { "epoch": 25.076190476190476, "grad_norm": 0.66015625, "learning_rate": 2.121424812753906e-05, "loss": 0.1139, "step": 34229 }, { "epoch": 25.076923076923077, "grad_norm": 1.328125, "learning_rate": 2.1211485835630424e-05, "loss": 0.1192, "step": 34230 }, { "epoch": 25.077655677655677, "grad_norm": 1.1875, "learning_rate": 2.120872368405267e-05, "loss": 0.1367, "step": 34231 }, { "epoch": 25.078388278388278, "grad_norm": 1.0, "learning_rate": 2.120596167281904e-05, "loss": 0.1013, "step": 34232 }, { "epoch": 25.07912087912088, "grad_norm": 0.984375, "learning_rate": 2.1203199801942776e-05, "loss": 0.1178, "step": 34233 }, { "epoch": 25.07985347985348, "grad_norm": 1.3125, "learning_rate": 2.1200438071437087e-05, "loss": 0.1009, "step": 34234 }, { "epoch": 25.08058608058608, "grad_norm": 0.828125, "learning_rate": 2.119767648131523e-05, "loss": 0.1365, "step": 34235 }, { "epoch": 25.08131868131868, "grad_norm": 0.7578125, "learning_rate": 2.1194915031590396e-05, "loss": 0.0767, "step": 34236 }, { "epoch": 25.08205128205128, "grad_norm": 0.95703125, "learning_rate": 2.1192153722275875e-05, "loss": 0.1178, "step": 34237 }, { "epoch": 25.082783882783882, "grad_norm": 1.65625, "learning_rate": 2.1189392553384848e-05, "loss": 0.111, "step": 34238 }, { "epoch": 25.083516483516483, "grad_norm": 1.515625, "learning_rate": 2.1186631524930586e-05, "loss": 0.1263, "step": 34239 }, { "epoch": 25.084249084249084, "grad_norm": 0.99609375, "learning_rate": 2.1183870636926276e-05, "loss": 0.1352, "step": 34240 }, { "epoch": 25.084981684981685, "grad_norm": 1.1015625, "learning_rate": 2.1181109889385165e-05, "loss": 0.1387, "step": 34241 }, { "epoch": 25.085714285714285, "grad_norm": 0.9765625, "learning_rate": 2.117834928232048e-05, "loss": 0.1099, "step": 34242 }, { "epoch": 25.086446886446886, "grad_norm": 1.59375, "learning_rate": 2.117558881574545e-05, "loss": 0.1435, "step": 34243 }, { "epoch": 25.087179487179487, "grad_norm": 1.125, "learning_rate": 2.117282848967332e-05, "loss": 0.1017, "step": 34244 }, { "epoch": 25.087912087912088, "grad_norm": 1.25, "learning_rate": 2.1170068304117278e-05, "loss": 0.1262, "step": 34245 }, { "epoch": 25.08864468864469, "grad_norm": 1.0625, "learning_rate": 2.1167308259090585e-05, "loss": 0.1439, "step": 34246 }, { "epoch": 25.08937728937729, "grad_norm": 0.78515625, "learning_rate": 2.1164548354606405e-05, "loss": 0.144, "step": 34247 }, { "epoch": 25.09010989010989, "grad_norm": 1.03125, "learning_rate": 2.1161788590678048e-05, "loss": 0.0946, "step": 34248 }, { "epoch": 25.09084249084249, "grad_norm": 1.1484375, "learning_rate": 2.1159028967318668e-05, "loss": 0.1169, "step": 34249 }, { "epoch": 25.09157509157509, "grad_norm": 0.953125, "learning_rate": 2.1156269484541536e-05, "loss": 0.1008, "step": 34250 }, { "epoch": 25.092307692307692, "grad_norm": 0.64453125, "learning_rate": 2.1153510142359832e-05, "loss": 0.0972, "step": 34251 }, { "epoch": 25.093040293040293, "grad_norm": 0.8359375, "learning_rate": 2.1150750940786797e-05, "loss": 0.1298, "step": 34252 }, { "epoch": 25.093772893772893, "grad_norm": 0.9453125, "learning_rate": 2.114799187983564e-05, "loss": 0.1094, "step": 34253 }, { "epoch": 25.094505494505494, "grad_norm": 0.828125, "learning_rate": 2.1145232959519597e-05, "loss": 0.108, "step": 34254 }, { "epoch": 25.095238095238095, "grad_norm": 1.234375, "learning_rate": 2.1142474179851897e-05, "loss": 0.1439, "step": 34255 }, { "epoch": 25.095970695970696, "grad_norm": 1.234375, "learning_rate": 2.113971554084572e-05, "loss": 0.1222, "step": 34256 }, { "epoch": 25.096703296703296, "grad_norm": 0.76171875, "learning_rate": 2.1136957042514305e-05, "loss": 0.118, "step": 34257 }, { "epoch": 25.097435897435897, "grad_norm": 0.828125, "learning_rate": 2.1134198684870868e-05, "loss": 0.1228, "step": 34258 }, { "epoch": 25.098168498168498, "grad_norm": 0.953125, "learning_rate": 2.1131440467928642e-05, "loss": 0.1242, "step": 34259 }, { "epoch": 25.0989010989011, "grad_norm": 1.140625, "learning_rate": 2.112868239170081e-05, "loss": 0.1216, "step": 34260 }, { "epoch": 25.0996336996337, "grad_norm": 0.96484375, "learning_rate": 2.112592445620062e-05, "loss": 0.1119, "step": 34261 }, { "epoch": 25.1003663003663, "grad_norm": 1.203125, "learning_rate": 2.1123166661441234e-05, "loss": 0.1328, "step": 34262 }, { "epoch": 25.1010989010989, "grad_norm": 1.2109375, "learning_rate": 2.1120409007435933e-05, "loss": 0.1271, "step": 34263 }, { "epoch": 25.1018315018315, "grad_norm": 1.015625, "learning_rate": 2.1117651494197882e-05, "loss": 0.1472, "step": 34264 }, { "epoch": 25.102564102564102, "grad_norm": 1.6796875, "learning_rate": 2.1114894121740305e-05, "loss": 0.123, "step": 34265 }, { "epoch": 25.103296703296703, "grad_norm": 0.9765625, "learning_rate": 2.1112136890076433e-05, "loss": 0.1244, "step": 34266 }, { "epoch": 25.104029304029304, "grad_norm": 1.2421875, "learning_rate": 2.1109379799219446e-05, "loss": 0.1522, "step": 34267 }, { "epoch": 25.104761904761904, "grad_norm": 1.3125, "learning_rate": 2.1106622849182574e-05, "loss": 0.1308, "step": 34268 }, { "epoch": 25.105494505494505, "grad_norm": 1.5546875, "learning_rate": 2.1103866039979017e-05, "loss": 0.1651, "step": 34269 }, { "epoch": 25.106227106227106, "grad_norm": 1.5625, "learning_rate": 2.1101109371622002e-05, "loss": 0.1525, "step": 34270 }, { "epoch": 25.106959706959707, "grad_norm": 0.6328125, "learning_rate": 2.109835284412471e-05, "loss": 0.0951, "step": 34271 }, { "epoch": 25.107692307692307, "grad_norm": 0.81640625, "learning_rate": 2.109559645750038e-05, "loss": 0.1364, "step": 34272 }, { "epoch": 25.108424908424908, "grad_norm": 0.7578125, "learning_rate": 2.109284021176216e-05, "loss": 0.117, "step": 34273 }, { "epoch": 25.10915750915751, "grad_norm": 1.953125, "learning_rate": 2.1090084106923333e-05, "loss": 0.1245, "step": 34274 }, { "epoch": 25.10989010989011, "grad_norm": 0.82421875, "learning_rate": 2.1087328142997055e-05, "loss": 0.0994, "step": 34275 }, { "epoch": 25.11062271062271, "grad_norm": 0.9140625, "learning_rate": 2.1084572319996557e-05, "loss": 0.121, "step": 34276 }, { "epoch": 25.11135531135531, "grad_norm": 1.0390625, "learning_rate": 2.1081816637935013e-05, "loss": 0.124, "step": 34277 }, { "epoch": 25.11208791208791, "grad_norm": 0.7734375, "learning_rate": 2.107906109682564e-05, "loss": 0.1219, "step": 34278 }, { "epoch": 25.112820512820512, "grad_norm": 1.1484375, "learning_rate": 2.1076305696681642e-05, "loss": 0.0945, "step": 34279 }, { "epoch": 25.113553113553113, "grad_norm": 1.21875, "learning_rate": 2.1073550437516225e-05, "loss": 0.1364, "step": 34280 }, { "epoch": 25.114285714285714, "grad_norm": 0.8125, "learning_rate": 2.1070795319342607e-05, "loss": 0.0986, "step": 34281 }, { "epoch": 25.115018315018315, "grad_norm": 1.265625, "learning_rate": 2.106804034217395e-05, "loss": 0.0947, "step": 34282 }, { "epoch": 25.115750915750915, "grad_norm": 1.03125, "learning_rate": 2.106528550602348e-05, "loss": 0.1402, "step": 34283 }, { "epoch": 25.116483516483516, "grad_norm": 1.1171875, "learning_rate": 2.1062530810904382e-05, "loss": 0.1558, "step": 34284 }, { "epoch": 25.117216117216117, "grad_norm": 1.046875, "learning_rate": 2.1059776256829882e-05, "loss": 0.1339, "step": 34285 }, { "epoch": 25.117948717948718, "grad_norm": 1.3984375, "learning_rate": 2.1057021843813144e-05, "loss": 0.1237, "step": 34286 }, { "epoch": 25.11868131868132, "grad_norm": 0.9296875, "learning_rate": 2.1054267571867395e-05, "loss": 0.1145, "step": 34287 }, { "epoch": 25.11941391941392, "grad_norm": 1.0546875, "learning_rate": 2.1051513441005782e-05, "loss": 0.1321, "step": 34288 }, { "epoch": 25.12014652014652, "grad_norm": 0.88671875, "learning_rate": 2.1048759451241568e-05, "loss": 0.0975, "step": 34289 }, { "epoch": 25.12087912087912, "grad_norm": 1.3828125, "learning_rate": 2.10460056025879e-05, "loss": 0.1249, "step": 34290 }, { "epoch": 25.12161172161172, "grad_norm": 1.109375, "learning_rate": 2.1043251895057996e-05, "loss": 0.1303, "step": 34291 }, { "epoch": 25.122344322344322, "grad_norm": 0.99609375, "learning_rate": 2.1040498328665047e-05, "loss": 0.1029, "step": 34292 }, { "epoch": 25.123076923076923, "grad_norm": 1.1953125, "learning_rate": 2.1037744903422228e-05, "loss": 0.1226, "step": 34293 }, { "epoch": 25.123809523809523, "grad_norm": 1.3828125, "learning_rate": 2.103499161934275e-05, "loss": 0.1166, "step": 34294 }, { "epoch": 25.124542124542124, "grad_norm": 1.1171875, "learning_rate": 2.1032238476439796e-05, "loss": 0.1093, "step": 34295 }, { "epoch": 25.125274725274725, "grad_norm": 1.9609375, "learning_rate": 2.102948547472658e-05, "loss": 0.1398, "step": 34296 }, { "epoch": 25.126007326007326, "grad_norm": 0.6171875, "learning_rate": 2.1026732614216256e-05, "loss": 0.1018, "step": 34297 }, { "epoch": 25.126739926739926, "grad_norm": 0.90234375, "learning_rate": 2.1023979894922042e-05, "loss": 0.1217, "step": 34298 }, { "epoch": 25.127472527472527, "grad_norm": 0.8359375, "learning_rate": 2.102122731685709e-05, "loss": 0.1007, "step": 34299 }, { "epoch": 25.128205128205128, "grad_norm": 1.0703125, "learning_rate": 2.101847488003465e-05, "loss": 0.1169, "step": 34300 }, { "epoch": 25.12893772893773, "grad_norm": 0.79296875, "learning_rate": 2.1015722584467854e-05, "loss": 0.0991, "step": 34301 }, { "epoch": 25.12967032967033, "grad_norm": 1.0390625, "learning_rate": 2.1012970430169933e-05, "loss": 0.1331, "step": 34302 }, { "epoch": 25.13040293040293, "grad_norm": 0.7109375, "learning_rate": 2.1010218417154023e-05, "loss": 0.1142, "step": 34303 }, { "epoch": 25.13113553113553, "grad_norm": 0.97265625, "learning_rate": 2.1007466545433346e-05, "loss": 0.1159, "step": 34304 }, { "epoch": 25.13186813186813, "grad_norm": 1.1328125, "learning_rate": 2.1004714815021076e-05, "loss": 0.1547, "step": 34305 }, { "epoch": 25.132600732600732, "grad_norm": 1.2265625, "learning_rate": 2.1001963225930398e-05, "loss": 0.1184, "step": 34306 }, { "epoch": 25.133333333333333, "grad_norm": 0.8828125, "learning_rate": 2.0999211778174522e-05, "loss": 0.1033, "step": 34307 }, { "epoch": 25.134065934065934, "grad_norm": 0.828125, "learning_rate": 2.0996460471766583e-05, "loss": 0.1053, "step": 34308 }, { "epoch": 25.134798534798534, "grad_norm": 1.234375, "learning_rate": 2.0993709306719785e-05, "loss": 0.1779, "step": 34309 }, { "epoch": 25.135531135531135, "grad_norm": 1.375, "learning_rate": 2.0990958283047316e-05, "loss": 0.1362, "step": 34310 }, { "epoch": 25.136263736263736, "grad_norm": 0.86328125, "learning_rate": 2.0988207400762376e-05, "loss": 0.115, "step": 34311 }, { "epoch": 25.136996336996337, "grad_norm": 0.65234375, "learning_rate": 2.0985456659878094e-05, "loss": 0.0951, "step": 34312 }, { "epoch": 25.137728937728937, "grad_norm": 0.92578125, "learning_rate": 2.0982706060407704e-05, "loss": 0.132, "step": 34313 }, { "epoch": 25.138461538461538, "grad_norm": 0.625, "learning_rate": 2.097995560236432e-05, "loss": 0.0792, "step": 34314 }, { "epoch": 25.13919413919414, "grad_norm": 1.0546875, "learning_rate": 2.0977205285761198e-05, "loss": 0.1439, "step": 34315 }, { "epoch": 25.13992673992674, "grad_norm": 1.4765625, "learning_rate": 2.0974455110611455e-05, "loss": 0.1577, "step": 34316 }, { "epoch": 25.14065934065934, "grad_norm": 1.046875, "learning_rate": 2.0971705076928293e-05, "loss": 0.1225, "step": 34317 }, { "epoch": 25.14139194139194, "grad_norm": 0.5859375, "learning_rate": 2.0968955184724905e-05, "loss": 0.0925, "step": 34318 }, { "epoch": 25.14212454212454, "grad_norm": 1.046875, "learning_rate": 2.0966205434014423e-05, "loss": 0.1333, "step": 34319 }, { "epoch": 25.142857142857142, "grad_norm": 0.75, "learning_rate": 2.0963455824810052e-05, "loss": 0.096, "step": 34320 }, { "epoch": 25.143589743589743, "grad_norm": 1.4921875, "learning_rate": 2.0960706357124958e-05, "loss": 0.1146, "step": 34321 }, { "epoch": 25.144322344322344, "grad_norm": 1.0703125, "learning_rate": 2.0957957030972333e-05, "loss": 0.1179, "step": 34322 }, { "epoch": 25.145054945054945, "grad_norm": 1.015625, "learning_rate": 2.0955207846365317e-05, "loss": 0.1175, "step": 34323 }, { "epoch": 25.145787545787545, "grad_norm": 1.171875, "learning_rate": 2.0952458803317118e-05, "loss": 0.1373, "step": 34324 }, { "epoch": 25.146520146520146, "grad_norm": 1.1015625, "learning_rate": 2.094970990184086e-05, "loss": 0.1267, "step": 34325 }, { "epoch": 25.147252747252747, "grad_norm": 1.2734375, "learning_rate": 2.0946961141949767e-05, "loss": 0.1182, "step": 34326 }, { "epoch": 25.147985347985347, "grad_norm": 1.0390625, "learning_rate": 2.0944212523656972e-05, "loss": 0.1463, "step": 34327 }, { "epoch": 25.148717948717948, "grad_norm": 1.359375, "learning_rate": 2.0941464046975676e-05, "loss": 0.1172, "step": 34328 }, { "epoch": 25.14945054945055, "grad_norm": 0.8046875, "learning_rate": 2.093871571191901e-05, "loss": 0.1014, "step": 34329 }, { "epoch": 25.15018315018315, "grad_norm": 1.4296875, "learning_rate": 2.0935967518500157e-05, "loss": 0.1183, "step": 34330 }, { "epoch": 25.15091575091575, "grad_norm": 1.1171875, "learning_rate": 2.0933219466732295e-05, "loss": 0.1066, "step": 34331 }, { "epoch": 25.15164835164835, "grad_norm": 1.328125, "learning_rate": 2.0930471556628585e-05, "loss": 0.1306, "step": 34332 }, { "epoch": 25.152380952380952, "grad_norm": 0.671875, "learning_rate": 2.0927723788202205e-05, "loss": 0.1362, "step": 34333 }, { "epoch": 25.153113553113553, "grad_norm": 0.71484375, "learning_rate": 2.09249761614663e-05, "loss": 0.1213, "step": 34334 }, { "epoch": 25.153846153846153, "grad_norm": 1.0859375, "learning_rate": 2.0922228676434035e-05, "loss": 0.1084, "step": 34335 }, { "epoch": 25.154578754578754, "grad_norm": 1.1796875, "learning_rate": 2.0919481333118583e-05, "loss": 0.1478, "step": 34336 }, { "epoch": 25.155311355311355, "grad_norm": 0.98828125, "learning_rate": 2.0916734131533126e-05, "loss": 0.1108, "step": 34337 }, { "epoch": 25.156043956043955, "grad_norm": 0.90234375, "learning_rate": 2.0913987071690786e-05, "loss": 0.1347, "step": 34338 }, { "epoch": 25.156776556776556, "grad_norm": 1.5, "learning_rate": 2.091124015360477e-05, "loss": 0.1563, "step": 34339 }, { "epoch": 25.157509157509157, "grad_norm": 0.94921875, "learning_rate": 2.0908493377288176e-05, "loss": 0.1234, "step": 34340 }, { "epoch": 25.158241758241758, "grad_norm": 0.953125, "learning_rate": 2.090574674275424e-05, "loss": 0.1201, "step": 34341 }, { "epoch": 25.15897435897436, "grad_norm": 0.83203125, "learning_rate": 2.0903000250016072e-05, "loss": 0.0952, "step": 34342 }, { "epoch": 25.15970695970696, "grad_norm": 1.0625, "learning_rate": 2.0900253899086847e-05, "loss": 0.1241, "step": 34343 }, { "epoch": 25.16043956043956, "grad_norm": 1.2578125, "learning_rate": 2.0897507689979734e-05, "loss": 0.1376, "step": 34344 }, { "epoch": 25.16117216117216, "grad_norm": 1.1015625, "learning_rate": 2.0894761622707867e-05, "loss": 0.0947, "step": 34345 }, { "epoch": 25.16190476190476, "grad_norm": 0.8984375, "learning_rate": 2.0892015697284414e-05, "loss": 0.1363, "step": 34346 }, { "epoch": 25.162637362637362, "grad_norm": 0.890625, "learning_rate": 2.088926991372253e-05, "loss": 0.1225, "step": 34347 }, { "epoch": 25.163369963369963, "grad_norm": 0.91015625, "learning_rate": 2.08865242720354e-05, "loss": 0.1178, "step": 34348 }, { "epoch": 25.164102564102564, "grad_norm": 0.890625, "learning_rate": 2.0883778772236134e-05, "loss": 0.1531, "step": 34349 }, { "epoch": 25.164835164835164, "grad_norm": 0.9765625, "learning_rate": 2.0881033414337922e-05, "loss": 0.1434, "step": 34350 }, { "epoch": 25.165567765567765, "grad_norm": 1.046875, "learning_rate": 2.087828819835386e-05, "loss": 0.1301, "step": 34351 }, { "epoch": 25.166300366300366, "grad_norm": 1.140625, "learning_rate": 2.087554312429719e-05, "loss": 0.1166, "step": 34352 }, { "epoch": 25.167032967032966, "grad_norm": 1.3203125, "learning_rate": 2.0872798192180994e-05, "loss": 0.1268, "step": 34353 }, { "epoch": 25.167765567765567, "grad_norm": 1.703125, "learning_rate": 2.087005340201847e-05, "loss": 0.1533, "step": 34354 }, { "epoch": 25.168498168498168, "grad_norm": 1.8828125, "learning_rate": 2.0867308753822722e-05, "loss": 0.1287, "step": 34355 }, { "epoch": 25.16923076923077, "grad_norm": 1.0390625, "learning_rate": 2.0864564247606926e-05, "loss": 0.1537, "step": 34356 }, { "epoch": 25.16996336996337, "grad_norm": 1.4609375, "learning_rate": 2.086181988338424e-05, "loss": 0.1073, "step": 34357 }, { "epoch": 25.17069597069597, "grad_norm": 0.87890625, "learning_rate": 2.0859075661167803e-05, "loss": 0.1193, "step": 34358 }, { "epoch": 25.17142857142857, "grad_norm": 0.77734375, "learning_rate": 2.0856331580970775e-05, "loss": 0.1357, "step": 34359 }, { "epoch": 25.17216117216117, "grad_norm": 1.71875, "learning_rate": 2.0853587642806284e-05, "loss": 0.132, "step": 34360 }, { "epoch": 25.172893772893772, "grad_norm": 1.0234375, "learning_rate": 2.0850843846687478e-05, "loss": 0.1336, "step": 34361 }, { "epoch": 25.173626373626373, "grad_norm": 0.75, "learning_rate": 2.0848100192627525e-05, "loss": 0.1168, "step": 34362 }, { "epoch": 25.174358974358974, "grad_norm": 0.7109375, "learning_rate": 2.084535668063956e-05, "loss": 0.1042, "step": 34363 }, { "epoch": 25.175091575091574, "grad_norm": 0.75390625, "learning_rate": 2.084261331073672e-05, "loss": 0.1282, "step": 34364 }, { "epoch": 25.175824175824175, "grad_norm": 1.4921875, "learning_rate": 2.0839870082932174e-05, "loss": 0.1143, "step": 34365 }, { "epoch": 25.176556776556776, "grad_norm": 0.78125, "learning_rate": 2.0837126997239003e-05, "loss": 0.1198, "step": 34366 }, { "epoch": 25.177289377289377, "grad_norm": 0.9453125, "learning_rate": 2.0834384053670435e-05, "loss": 0.1394, "step": 34367 }, { "epoch": 25.178021978021977, "grad_norm": 1.078125, "learning_rate": 2.0831641252239552e-05, "loss": 0.1168, "step": 34368 }, { "epoch": 25.178754578754578, "grad_norm": 1.2890625, "learning_rate": 2.0828898592959524e-05, "loss": 0.11, "step": 34369 }, { "epoch": 25.17948717948718, "grad_norm": 1.34375, "learning_rate": 2.082615607584349e-05, "loss": 0.0999, "step": 34370 }, { "epoch": 25.18021978021978, "grad_norm": 1.796875, "learning_rate": 2.082341370090457e-05, "loss": 0.1618, "step": 34371 }, { "epoch": 25.18095238095238, "grad_norm": 1.265625, "learning_rate": 2.0820671468155916e-05, "loss": 0.146, "step": 34372 }, { "epoch": 25.18168498168498, "grad_norm": 0.96484375, "learning_rate": 2.0817929377610665e-05, "loss": 0.1297, "step": 34373 }, { "epoch": 25.18241758241758, "grad_norm": 1.2421875, "learning_rate": 2.0815187429281984e-05, "loss": 0.1325, "step": 34374 }, { "epoch": 25.183150183150182, "grad_norm": 1.125, "learning_rate": 2.0812445623182963e-05, "loss": 0.1469, "step": 34375 }, { "epoch": 25.183882783882783, "grad_norm": 0.97265625, "learning_rate": 2.0809703959326775e-05, "loss": 0.1248, "step": 34376 }, { "epoch": 25.184615384615384, "grad_norm": 1.265625, "learning_rate": 2.080696243772651e-05, "loss": 0.103, "step": 34377 }, { "epoch": 25.185347985347985, "grad_norm": 1.0859375, "learning_rate": 2.0804221058395377e-05, "loss": 0.11, "step": 34378 }, { "epoch": 25.186080586080585, "grad_norm": 0.953125, "learning_rate": 2.0801479821346443e-05, "loss": 0.1381, "step": 34379 }, { "epoch": 25.186813186813186, "grad_norm": 1.0546875, "learning_rate": 2.079873872659287e-05, "loss": 0.1427, "step": 34380 }, { "epoch": 25.187545787545787, "grad_norm": 1.015625, "learning_rate": 2.079599777414781e-05, "loss": 0.1049, "step": 34381 }, { "epoch": 25.188278388278388, "grad_norm": 1.15625, "learning_rate": 2.0793256964024358e-05, "loss": 0.107, "step": 34382 }, { "epoch": 25.18901098901099, "grad_norm": 1.375, "learning_rate": 2.079051629623566e-05, "loss": 0.1523, "step": 34383 }, { "epoch": 25.18974358974359, "grad_norm": 0.92578125, "learning_rate": 2.0787775770794852e-05, "loss": 0.1233, "step": 34384 }, { "epoch": 25.19047619047619, "grad_norm": 0.6328125, "learning_rate": 2.078503538771509e-05, "loss": 0.0975, "step": 34385 }, { "epoch": 25.19120879120879, "grad_norm": 0.87890625, "learning_rate": 2.078229514700945e-05, "loss": 0.127, "step": 34386 }, { "epoch": 25.19194139194139, "grad_norm": 1.5, "learning_rate": 2.07795550486911e-05, "loss": 0.1126, "step": 34387 }, { "epoch": 25.192673992673992, "grad_norm": 0.80859375, "learning_rate": 2.077681509277315e-05, "loss": 0.1197, "step": 34388 }, { "epoch": 25.193406593406593, "grad_norm": 1.2421875, "learning_rate": 2.0774075279268755e-05, "loss": 0.1271, "step": 34389 }, { "epoch": 25.194139194139193, "grad_norm": 1.1953125, "learning_rate": 2.077133560819101e-05, "loss": 0.1226, "step": 34390 }, { "epoch": 25.194871794871794, "grad_norm": 0.765625, "learning_rate": 2.0768596079553072e-05, "loss": 0.1, "step": 34391 }, { "epoch": 25.195604395604395, "grad_norm": 0.78125, "learning_rate": 2.0765856693368015e-05, "loss": 0.1154, "step": 34392 }, { "epoch": 25.196336996336996, "grad_norm": 1.1796875, "learning_rate": 2.076311744964903e-05, "loss": 0.1082, "step": 34393 }, { "epoch": 25.197069597069596, "grad_norm": 1.1796875, "learning_rate": 2.0760378348409197e-05, "loss": 0.1133, "step": 34394 }, { "epoch": 25.197802197802197, "grad_norm": 1.140625, "learning_rate": 2.075763938966166e-05, "loss": 0.1127, "step": 34395 }, { "epoch": 25.198534798534798, "grad_norm": 1.3046875, "learning_rate": 2.0754900573419543e-05, "loss": 0.1486, "step": 34396 }, { "epoch": 25.1992673992674, "grad_norm": 1.0234375, "learning_rate": 2.0752161899695955e-05, "loss": 0.1412, "step": 34397 }, { "epoch": 25.2, "grad_norm": 0.75390625, "learning_rate": 2.074942336850402e-05, "loss": 0.1061, "step": 34398 }, { "epoch": 25.2007326007326, "grad_norm": 0.81640625, "learning_rate": 2.0746684979856863e-05, "loss": 0.1009, "step": 34399 }, { "epoch": 25.2014652014652, "grad_norm": 0.8046875, "learning_rate": 2.0743946733767626e-05, "loss": 0.1073, "step": 34400 }, { "epoch": 25.2021978021978, "grad_norm": 1.0078125, "learning_rate": 2.074120863024939e-05, "loss": 0.1311, "step": 34401 }, { "epoch": 25.202930402930402, "grad_norm": 0.72265625, "learning_rate": 2.073847066931531e-05, "loss": 0.1294, "step": 34402 }, { "epoch": 25.203663003663003, "grad_norm": 1.0078125, "learning_rate": 2.073573285097845e-05, "loss": 0.1335, "step": 34403 }, { "epoch": 25.204395604395604, "grad_norm": 0.90625, "learning_rate": 2.0732995175252004e-05, "loss": 0.1052, "step": 34404 }, { "epoch": 25.205128205128204, "grad_norm": 1.171875, "learning_rate": 2.0730257642149026e-05, "loss": 0.1216, "step": 34405 }, { "epoch": 25.205860805860805, "grad_norm": 0.984375, "learning_rate": 2.0727520251682663e-05, "loss": 0.1203, "step": 34406 }, { "epoch": 25.206593406593406, "grad_norm": 1.2890625, "learning_rate": 2.072478300386604e-05, "loss": 0.1164, "step": 34407 }, { "epoch": 25.207326007326007, "grad_norm": 0.87890625, "learning_rate": 2.0722045898712232e-05, "loss": 0.1014, "step": 34408 }, { "epoch": 25.208058608058607, "grad_norm": 1.4296875, "learning_rate": 2.071930893623439e-05, "loss": 0.1107, "step": 34409 }, { "epoch": 25.208791208791208, "grad_norm": 1.09375, "learning_rate": 2.0716572116445605e-05, "loss": 0.1537, "step": 34410 }, { "epoch": 25.20952380952381, "grad_norm": 0.9765625, "learning_rate": 2.0713835439359023e-05, "loss": 0.0902, "step": 34411 }, { "epoch": 25.21025641025641, "grad_norm": 1.09375, "learning_rate": 2.071109890498771e-05, "loss": 0.11, "step": 34412 }, { "epoch": 25.21098901098901, "grad_norm": 1.1171875, "learning_rate": 2.070836251334481e-05, "loss": 0.1387, "step": 34413 }, { "epoch": 25.21172161172161, "grad_norm": 1.15625, "learning_rate": 2.070562626444342e-05, "loss": 0.1361, "step": 34414 }, { "epoch": 25.21245421245421, "grad_norm": 1.6015625, "learning_rate": 2.0702890158296676e-05, "loss": 0.1058, "step": 34415 }, { "epoch": 25.213186813186812, "grad_norm": 0.90625, "learning_rate": 2.070015419491765e-05, "loss": 0.1534, "step": 34416 }, { "epoch": 25.213919413919413, "grad_norm": 0.95703125, "learning_rate": 2.0697418374319477e-05, "loss": 0.1241, "step": 34417 }, { "epoch": 25.214652014652014, "grad_norm": 1.625, "learning_rate": 2.0694682696515224e-05, "loss": 0.0943, "step": 34418 }, { "epoch": 25.215384615384615, "grad_norm": 1.2578125, "learning_rate": 2.0691947161518073e-05, "loss": 0.1112, "step": 34419 }, { "epoch": 25.216117216117215, "grad_norm": 1.484375, "learning_rate": 2.0689211769341064e-05, "loss": 0.1059, "step": 34420 }, { "epoch": 25.216849816849816, "grad_norm": 1.125, "learning_rate": 2.068647651999733e-05, "loss": 0.1523, "step": 34421 }, { "epoch": 25.217582417582417, "grad_norm": 1.046875, "learning_rate": 2.0683741413499996e-05, "loss": 0.12, "step": 34422 }, { "epoch": 25.218315018315018, "grad_norm": 0.953125, "learning_rate": 2.068100644986212e-05, "loss": 0.1319, "step": 34423 }, { "epoch": 25.21904761904762, "grad_norm": 0.74609375, "learning_rate": 2.0678271629096837e-05, "loss": 0.1115, "step": 34424 }, { "epoch": 25.21978021978022, "grad_norm": 1.2578125, "learning_rate": 2.067553695121724e-05, "loss": 0.1198, "step": 34425 }, { "epoch": 25.22051282051282, "grad_norm": 1.203125, "learning_rate": 2.0672802416236454e-05, "loss": 0.1125, "step": 34426 }, { "epoch": 25.22124542124542, "grad_norm": 0.7734375, "learning_rate": 2.0670068024167547e-05, "loss": 0.1206, "step": 34427 }, { "epoch": 25.22197802197802, "grad_norm": 0.96484375, "learning_rate": 2.0667333775023652e-05, "loss": 0.136, "step": 34428 }, { "epoch": 25.222710622710622, "grad_norm": 0.5546875, "learning_rate": 2.0664599668817824e-05, "loss": 0.1071, "step": 34429 }, { "epoch": 25.223443223443223, "grad_norm": 1.2265625, "learning_rate": 2.066186570556323e-05, "loss": 0.1504, "step": 34430 }, { "epoch": 25.224175824175823, "grad_norm": 1.140625, "learning_rate": 2.0659131885272905e-05, "loss": 0.159, "step": 34431 }, { "epoch": 25.224908424908424, "grad_norm": 0.96484375, "learning_rate": 2.065639820795998e-05, "loss": 0.0976, "step": 34432 }, { "epoch": 25.225641025641025, "grad_norm": 1.0703125, "learning_rate": 2.0653664673637575e-05, "loss": 0.1232, "step": 34433 }, { "epoch": 25.226373626373626, "grad_norm": 0.765625, "learning_rate": 2.065093128231873e-05, "loss": 0.1021, "step": 34434 }, { "epoch": 25.227106227106226, "grad_norm": 0.81640625, "learning_rate": 2.0648198034016573e-05, "loss": 0.1181, "step": 34435 }, { "epoch": 25.227838827838827, "grad_norm": 1.125, "learning_rate": 2.0645464928744205e-05, "loss": 0.116, "step": 34436 }, { "epoch": 25.228571428571428, "grad_norm": 1.1953125, "learning_rate": 2.064273196651473e-05, "loss": 0.1245, "step": 34437 }, { "epoch": 25.22930402930403, "grad_norm": 1.1328125, "learning_rate": 2.0639999147341215e-05, "loss": 0.1156, "step": 34438 }, { "epoch": 25.23003663003663, "grad_norm": 1.015625, "learning_rate": 2.0637266471236758e-05, "loss": 0.146, "step": 34439 }, { "epoch": 25.23076923076923, "grad_norm": 1.109375, "learning_rate": 2.0634533938214466e-05, "loss": 0.1052, "step": 34440 }, { "epoch": 25.23150183150183, "grad_norm": 1.3046875, "learning_rate": 2.0631801548287443e-05, "loss": 0.1148, "step": 34441 }, { "epoch": 25.23223443223443, "grad_norm": 1.4609375, "learning_rate": 2.0629069301468747e-05, "loss": 0.1212, "step": 34442 }, { "epoch": 25.232967032967032, "grad_norm": 0.80078125, "learning_rate": 2.06263371977715e-05, "loss": 0.0943, "step": 34443 }, { "epoch": 25.233699633699633, "grad_norm": 1.171875, "learning_rate": 2.0623605237208745e-05, "loss": 0.1475, "step": 34444 }, { "epoch": 25.234432234432234, "grad_norm": 0.9609375, "learning_rate": 2.0620873419793642e-05, "loss": 0.1301, "step": 34445 }, { "epoch": 25.235164835164834, "grad_norm": 0.9140625, "learning_rate": 2.061814174553922e-05, "loss": 0.1419, "step": 34446 }, { "epoch": 25.235897435897435, "grad_norm": 1.0, "learning_rate": 2.06154102144586e-05, "loss": 0.121, "step": 34447 }, { "epoch": 25.236630036630036, "grad_norm": 1.15625, "learning_rate": 2.0612678826564874e-05, "loss": 0.1212, "step": 34448 }, { "epoch": 25.237362637362637, "grad_norm": 0.91796875, "learning_rate": 2.0609947581871095e-05, "loss": 0.1368, "step": 34449 }, { "epoch": 25.238095238095237, "grad_norm": 1.0234375, "learning_rate": 2.060721648039037e-05, "loss": 0.1397, "step": 34450 }, { "epoch": 25.238827838827838, "grad_norm": 1.0390625, "learning_rate": 2.0604485522135786e-05, "loss": 0.1476, "step": 34451 }, { "epoch": 25.23956043956044, "grad_norm": 0.99609375, "learning_rate": 2.0601754707120442e-05, "loss": 0.1281, "step": 34452 }, { "epoch": 25.24029304029304, "grad_norm": 0.8125, "learning_rate": 2.0599024035357384e-05, "loss": 0.1218, "step": 34453 }, { "epoch": 25.24102564102564, "grad_norm": 1.34375, "learning_rate": 2.059629350685974e-05, "loss": 0.117, "step": 34454 }, { "epoch": 25.24175824175824, "grad_norm": 0.7109375, "learning_rate": 2.0593563121640536e-05, "loss": 0.0935, "step": 34455 }, { "epoch": 25.24249084249084, "grad_norm": 1.171875, "learning_rate": 2.0590832879712922e-05, "loss": 0.1148, "step": 34456 }, { "epoch": 25.243223443223442, "grad_norm": 1.078125, "learning_rate": 2.0588102781089934e-05, "loss": 0.1298, "step": 34457 }, { "epoch": 25.243956043956043, "grad_norm": 2.109375, "learning_rate": 2.0585372825784656e-05, "loss": 0.1155, "step": 34458 }, { "epoch": 25.244688644688644, "grad_norm": 1.265625, "learning_rate": 2.05826430138102e-05, "loss": 0.0877, "step": 34459 }, { "epoch": 25.245421245421245, "grad_norm": 0.9453125, "learning_rate": 2.0579913345179605e-05, "loss": 0.1042, "step": 34460 }, { "epoch": 25.246153846153845, "grad_norm": 1.1328125, "learning_rate": 2.0577183819905965e-05, "loss": 0.1196, "step": 34461 }, { "epoch": 25.246886446886446, "grad_norm": 0.90625, "learning_rate": 2.0574454438002362e-05, "loss": 0.1323, "step": 34462 }, { "epoch": 25.247619047619047, "grad_norm": 1.6640625, "learning_rate": 2.057172519948189e-05, "loss": 0.1271, "step": 34463 }, { "epoch": 25.248351648351647, "grad_norm": 0.9296875, "learning_rate": 2.0568996104357588e-05, "loss": 0.1113, "step": 34464 }, { "epoch": 25.249084249084248, "grad_norm": 1.109375, "learning_rate": 2.056626715264255e-05, "loss": 0.0899, "step": 34465 }, { "epoch": 25.24981684981685, "grad_norm": 1.359375, "learning_rate": 2.0563538344349855e-05, "loss": 0.1005, "step": 34466 }, { "epoch": 25.25054945054945, "grad_norm": 1.21875, "learning_rate": 2.0560809679492594e-05, "loss": 0.1225, "step": 34467 }, { "epoch": 25.25128205128205, "grad_norm": 1.7265625, "learning_rate": 2.055808115808381e-05, "loss": 0.1524, "step": 34468 }, { "epoch": 25.25201465201465, "grad_norm": 0.87890625, "learning_rate": 2.0555352780136598e-05, "loss": 0.1322, "step": 34469 }, { "epoch": 25.252747252747252, "grad_norm": 0.76953125, "learning_rate": 2.055262454566398e-05, "loss": 0.1182, "step": 34470 }, { "epoch": 25.253479853479853, "grad_norm": 0.81640625, "learning_rate": 2.0549896454679114e-05, "loss": 0.0928, "step": 34471 }, { "epoch": 25.254212454212453, "grad_norm": 0.6953125, "learning_rate": 2.054716850719501e-05, "loss": 0.1055, "step": 34472 }, { "epoch": 25.254945054945054, "grad_norm": 1.1171875, "learning_rate": 2.054444070322475e-05, "loss": 0.118, "step": 34473 }, { "epoch": 25.255677655677655, "grad_norm": 0.78515625, "learning_rate": 2.0541713042781424e-05, "loss": 0.1173, "step": 34474 }, { "epoch": 25.256410256410255, "grad_norm": 1.625, "learning_rate": 2.0538985525878067e-05, "loss": 0.1346, "step": 34475 }, { "epoch": 25.257142857142856, "grad_norm": 1.34375, "learning_rate": 2.0536258152527778e-05, "loss": 0.1196, "step": 34476 }, { "epoch": 25.257875457875457, "grad_norm": 0.796875, "learning_rate": 2.0533530922743596e-05, "loss": 0.1125, "step": 34477 }, { "epoch": 25.258608058608058, "grad_norm": 1.328125, "learning_rate": 2.053080383653863e-05, "loss": 0.1505, "step": 34478 }, { "epoch": 25.25934065934066, "grad_norm": 1.109375, "learning_rate": 2.052807689392591e-05, "loss": 0.112, "step": 34479 }, { "epoch": 25.26007326007326, "grad_norm": 0.828125, "learning_rate": 2.052535009491852e-05, "loss": 0.1123, "step": 34480 }, { "epoch": 25.26080586080586, "grad_norm": 1.046875, "learning_rate": 2.052262343952949e-05, "loss": 0.1017, "step": 34481 }, { "epoch": 25.26153846153846, "grad_norm": 0.75, "learning_rate": 2.0519896927771944e-05, "loss": 0.0999, "step": 34482 }, { "epoch": 25.26227106227106, "grad_norm": 1.1953125, "learning_rate": 2.0517170559658892e-05, "loss": 0.12, "step": 34483 }, { "epoch": 25.263003663003662, "grad_norm": 0.8046875, "learning_rate": 2.0514444335203424e-05, "loss": 0.1075, "step": 34484 }, { "epoch": 25.263736263736263, "grad_norm": 1.546875, "learning_rate": 2.0511718254418612e-05, "loss": 0.109, "step": 34485 }, { "epoch": 25.264468864468864, "grad_norm": 1.7265625, "learning_rate": 2.0508992317317486e-05, "loss": 0.1499, "step": 34486 }, { "epoch": 25.265201465201464, "grad_norm": 1.453125, "learning_rate": 2.050626652391312e-05, "loss": 0.15, "step": 34487 }, { "epoch": 25.265934065934065, "grad_norm": 0.6328125, "learning_rate": 2.0503540874218582e-05, "loss": 0.1102, "step": 34488 }, { "epoch": 25.266666666666666, "grad_norm": 0.94921875, "learning_rate": 2.0500815368246943e-05, "loss": 0.1245, "step": 34489 }, { "epoch": 25.267399267399266, "grad_norm": 1.265625, "learning_rate": 2.0498090006011226e-05, "loss": 0.1451, "step": 34490 }, { "epoch": 25.268131868131867, "grad_norm": 1.6796875, "learning_rate": 2.049536478752451e-05, "loss": 0.1144, "step": 34491 }, { "epoch": 25.268864468864468, "grad_norm": 1.2890625, "learning_rate": 2.0492639712799853e-05, "loss": 0.1167, "step": 34492 }, { "epoch": 25.26959706959707, "grad_norm": 1.09375, "learning_rate": 2.0489914781850326e-05, "loss": 0.1234, "step": 34493 }, { "epoch": 25.27032967032967, "grad_norm": 1.484375, "learning_rate": 2.0487189994688953e-05, "loss": 0.1179, "step": 34494 }, { "epoch": 25.27106227106227, "grad_norm": 0.828125, "learning_rate": 2.0484465351328803e-05, "loss": 0.1171, "step": 34495 }, { "epoch": 25.27179487179487, "grad_norm": 1.3515625, "learning_rate": 2.0481740851782935e-05, "loss": 0.1121, "step": 34496 }, { "epoch": 25.27252747252747, "grad_norm": 1.0703125, "learning_rate": 2.0479016496064414e-05, "loss": 0.1333, "step": 34497 }, { "epoch": 25.273260073260072, "grad_norm": 0.5, "learning_rate": 2.0476292284186266e-05, "loss": 0.0915, "step": 34498 }, { "epoch": 25.273992673992673, "grad_norm": 0.859375, "learning_rate": 2.0473568216161557e-05, "loss": 0.1192, "step": 34499 }, { "epoch": 25.274725274725274, "grad_norm": 0.78515625, "learning_rate": 2.0470844292003358e-05, "loss": 0.1316, "step": 34500 }, { "epoch": 25.275457875457874, "grad_norm": 1.2109375, "learning_rate": 2.0468120511724685e-05, "loss": 0.1112, "step": 34501 }, { "epoch": 25.276190476190475, "grad_norm": 1.4921875, "learning_rate": 2.0465396875338608e-05, "loss": 0.1249, "step": 34502 }, { "epoch": 25.276923076923076, "grad_norm": 1.671875, "learning_rate": 2.0462673382858163e-05, "loss": 0.1114, "step": 34503 }, { "epoch": 25.277655677655677, "grad_norm": 1.1328125, "learning_rate": 2.045995003429644e-05, "loss": 0.1456, "step": 34504 }, { "epoch": 25.278388278388277, "grad_norm": 1.2421875, "learning_rate": 2.0457226829666436e-05, "loss": 0.1311, "step": 34505 }, { "epoch": 25.279120879120878, "grad_norm": 1.28125, "learning_rate": 2.0454503768981237e-05, "loss": 0.1208, "step": 34506 }, { "epoch": 25.27985347985348, "grad_norm": 1.1953125, "learning_rate": 2.045178085225384e-05, "loss": 0.1115, "step": 34507 }, { "epoch": 25.28058608058608, "grad_norm": 0.61328125, "learning_rate": 2.044905807949736e-05, "loss": 0.115, "step": 34508 }, { "epoch": 25.28131868131868, "grad_norm": 0.94140625, "learning_rate": 2.0446335450724795e-05, "loss": 0.1076, "step": 34509 }, { "epoch": 25.28205128205128, "grad_norm": 0.8515625, "learning_rate": 2.0443612965949195e-05, "loss": 0.1019, "step": 34510 }, { "epoch": 25.282783882783882, "grad_norm": 1.1953125, "learning_rate": 2.0440890625183633e-05, "loss": 0.135, "step": 34511 }, { "epoch": 25.283516483516483, "grad_norm": 1.0546875, "learning_rate": 2.0438168428441102e-05, "loss": 0.1097, "step": 34512 }, { "epoch": 25.284249084249083, "grad_norm": 1.140625, "learning_rate": 2.0435446375734705e-05, "loss": 0.1381, "step": 34513 }, { "epoch": 25.284981684981684, "grad_norm": 1.2109375, "learning_rate": 2.0432724467077434e-05, "loss": 0.1148, "step": 34514 }, { "epoch": 25.285714285714285, "grad_norm": 1.5859375, "learning_rate": 2.0430002702482368e-05, "loss": 0.1263, "step": 34515 }, { "epoch": 25.286446886446885, "grad_norm": 0.96484375, "learning_rate": 2.0427281081962514e-05, "loss": 0.1105, "step": 34516 }, { "epoch": 25.287179487179486, "grad_norm": 0.73828125, "learning_rate": 2.0424559605530934e-05, "loss": 0.1115, "step": 34517 }, { "epoch": 25.287912087912087, "grad_norm": 0.76171875, "learning_rate": 2.0421838273200654e-05, "loss": 0.1376, "step": 34518 }, { "epoch": 25.288644688644688, "grad_norm": 0.890625, "learning_rate": 2.041911708498474e-05, "loss": 0.1089, "step": 34519 }, { "epoch": 25.28937728937729, "grad_norm": 1.046875, "learning_rate": 2.0416396040896196e-05, "loss": 0.1288, "step": 34520 }, { "epoch": 25.29010989010989, "grad_norm": 1.4609375, "learning_rate": 2.0413675140948077e-05, "loss": 0.1069, "step": 34521 }, { "epoch": 25.29084249084249, "grad_norm": 1.0546875, "learning_rate": 2.0410954385153412e-05, "loss": 0.123, "step": 34522 }, { "epoch": 25.29157509157509, "grad_norm": 1.1953125, "learning_rate": 2.0408233773525262e-05, "loss": 0.1178, "step": 34523 }, { "epoch": 25.29230769230769, "grad_norm": 0.94140625, "learning_rate": 2.0405513306076625e-05, "loss": 0.1123, "step": 34524 }, { "epoch": 25.293040293040292, "grad_norm": 0.65625, "learning_rate": 2.040279298282055e-05, "loss": 0.1156, "step": 34525 }, { "epoch": 25.293772893772893, "grad_norm": 1.2265625, "learning_rate": 2.0400072803770094e-05, "loss": 0.1029, "step": 34526 }, { "epoch": 25.294505494505493, "grad_norm": 1.0546875, "learning_rate": 2.0397352768938255e-05, "loss": 0.1267, "step": 34527 }, { "epoch": 25.295238095238094, "grad_norm": 1.0, "learning_rate": 2.039463287833808e-05, "loss": 0.1223, "step": 34528 }, { "epoch": 25.295970695970695, "grad_norm": 1.2109375, "learning_rate": 2.03919131319826e-05, "loss": 0.1254, "step": 34529 }, { "epoch": 25.296703296703296, "grad_norm": 1.171875, "learning_rate": 2.038919352988486e-05, "loss": 0.1081, "step": 34530 }, { "epoch": 25.297435897435896, "grad_norm": 1.34375, "learning_rate": 2.0386474072057873e-05, "loss": 0.106, "step": 34531 }, { "epoch": 25.298168498168497, "grad_norm": 0.89453125, "learning_rate": 2.0383754758514688e-05, "loss": 0.1268, "step": 34532 }, { "epoch": 25.298901098901098, "grad_norm": 1.421875, "learning_rate": 2.038103558926828e-05, "loss": 0.1244, "step": 34533 }, { "epoch": 25.2996336996337, "grad_norm": 0.7421875, "learning_rate": 2.037831656433176e-05, "loss": 0.1164, "step": 34534 }, { "epoch": 25.3003663003663, "grad_norm": 1.296875, "learning_rate": 2.037559768371809e-05, "loss": 0.1133, "step": 34535 }, { "epoch": 25.3010989010989, "grad_norm": 1.359375, "learning_rate": 2.037287894744032e-05, "loss": 0.1272, "step": 34536 }, { "epoch": 25.3018315018315, "grad_norm": 1.703125, "learning_rate": 2.0370160355511506e-05, "loss": 0.1393, "step": 34537 }, { "epoch": 25.3025641025641, "grad_norm": 1.3828125, "learning_rate": 2.0367441907944597e-05, "loss": 0.1522, "step": 34538 }, { "epoch": 25.303296703296702, "grad_norm": 1.2109375, "learning_rate": 2.036472360475271e-05, "loss": 0.1048, "step": 34539 }, { "epoch": 25.304029304029303, "grad_norm": 0.7734375, "learning_rate": 2.0362005445948802e-05, "loss": 0.1081, "step": 34540 }, { "epoch": 25.304761904761904, "grad_norm": 1.203125, "learning_rate": 2.035928743154593e-05, "loss": 0.1393, "step": 34541 }, { "epoch": 25.305494505494504, "grad_norm": 1.578125, "learning_rate": 2.0356569561557097e-05, "loss": 0.1082, "step": 34542 }, { "epoch": 25.306227106227105, "grad_norm": 1.140625, "learning_rate": 2.035385183599534e-05, "loss": 0.1109, "step": 34543 }, { "epoch": 25.306959706959706, "grad_norm": 0.9140625, "learning_rate": 2.035113425487367e-05, "loss": 0.0857, "step": 34544 }, { "epoch": 25.307692307692307, "grad_norm": 0.87109375, "learning_rate": 2.0348416818205126e-05, "loss": 0.1243, "step": 34545 }, { "epoch": 25.308424908424907, "grad_norm": 1.109375, "learning_rate": 2.0345699526002708e-05, "loss": 0.1478, "step": 34546 }, { "epoch": 25.309157509157508, "grad_norm": 0.7578125, "learning_rate": 2.0342982378279435e-05, "loss": 0.1205, "step": 34547 }, { "epoch": 25.30989010989011, "grad_norm": 0.73828125, "learning_rate": 2.034026537504833e-05, "loss": 0.1279, "step": 34548 }, { "epoch": 25.31062271062271, "grad_norm": 0.9921875, "learning_rate": 2.0337548516322442e-05, "loss": 0.1109, "step": 34549 }, { "epoch": 25.31135531135531, "grad_norm": 1.203125, "learning_rate": 2.033483180211473e-05, "loss": 0.1525, "step": 34550 }, { "epoch": 25.31208791208791, "grad_norm": 0.828125, "learning_rate": 2.0332115232438257e-05, "loss": 0.1121, "step": 34551 }, { "epoch": 25.31282051282051, "grad_norm": 1.296875, "learning_rate": 2.0329398807306028e-05, "loss": 0.1578, "step": 34552 }, { "epoch": 25.313553113553112, "grad_norm": 0.7109375, "learning_rate": 2.0326682526731044e-05, "loss": 0.1207, "step": 34553 }, { "epoch": 25.314285714285713, "grad_norm": 1.390625, "learning_rate": 2.0323966390726325e-05, "loss": 0.1226, "step": 34554 }, { "epoch": 25.315018315018314, "grad_norm": 0.93359375, "learning_rate": 2.0321250399304892e-05, "loss": 0.1005, "step": 34555 }, { "epoch": 25.315750915750915, "grad_norm": 1.0546875, "learning_rate": 2.0318534552479773e-05, "loss": 0.1202, "step": 34556 }, { "epoch": 25.316483516483515, "grad_norm": 0.8984375, "learning_rate": 2.031581885026394e-05, "loss": 0.1218, "step": 34557 }, { "epoch": 25.317216117216116, "grad_norm": 1.046875, "learning_rate": 2.0313103292670455e-05, "loss": 0.1094, "step": 34558 }, { "epoch": 25.317948717948717, "grad_norm": 0.87890625, "learning_rate": 2.0310387879712257e-05, "loss": 0.103, "step": 34559 }, { "epoch": 25.318681318681318, "grad_norm": 0.9921875, "learning_rate": 2.030767261140244e-05, "loss": 0.1098, "step": 34560 }, { "epoch": 25.31941391941392, "grad_norm": 0.65625, "learning_rate": 2.030495748775396e-05, "loss": 0.1142, "step": 34561 }, { "epoch": 25.32014652014652, "grad_norm": 1.0234375, "learning_rate": 2.030224250877984e-05, "loss": 0.1194, "step": 34562 }, { "epoch": 25.32087912087912, "grad_norm": 1.4765625, "learning_rate": 2.0299527674493103e-05, "loss": 0.1418, "step": 34563 }, { "epoch": 25.32161172161172, "grad_norm": 0.984375, "learning_rate": 2.029681298490671e-05, "loss": 0.1251, "step": 34564 }, { "epoch": 25.32234432234432, "grad_norm": 1.1328125, "learning_rate": 2.029409844003374e-05, "loss": 0.1346, "step": 34565 }, { "epoch": 25.323076923076922, "grad_norm": 1.1953125, "learning_rate": 2.0291384039887137e-05, "loss": 0.1058, "step": 34566 }, { "epoch": 25.323809523809523, "grad_norm": 0.91796875, "learning_rate": 2.0288669784479953e-05, "loss": 0.1262, "step": 34567 }, { "epoch": 25.324542124542123, "grad_norm": 1.4140625, "learning_rate": 2.028595567382515e-05, "loss": 0.1193, "step": 34568 }, { "epoch": 25.325274725274724, "grad_norm": 1.609375, "learning_rate": 2.0283241707935753e-05, "loss": 0.1268, "step": 34569 }, { "epoch": 25.326007326007325, "grad_norm": 0.8125, "learning_rate": 2.028052788682476e-05, "loss": 0.1115, "step": 34570 }, { "epoch": 25.326739926739926, "grad_norm": 1.09375, "learning_rate": 2.0277814210505196e-05, "loss": 0.1082, "step": 34571 }, { "epoch": 25.327472527472526, "grad_norm": 1.1171875, "learning_rate": 2.0275100678990035e-05, "loss": 0.1086, "step": 34572 }, { "epoch": 25.328205128205127, "grad_norm": 1.1484375, "learning_rate": 2.027238729229228e-05, "loss": 0.1204, "step": 34573 }, { "epoch": 25.328937728937728, "grad_norm": 1.5390625, "learning_rate": 2.026967405042495e-05, "loss": 0.1176, "step": 34574 }, { "epoch": 25.32967032967033, "grad_norm": 0.89453125, "learning_rate": 2.0266960953401046e-05, "loss": 0.1405, "step": 34575 }, { "epoch": 25.33040293040293, "grad_norm": 0.96875, "learning_rate": 2.0264248001233544e-05, "loss": 0.1197, "step": 34576 }, { "epoch": 25.33113553113553, "grad_norm": 1.375, "learning_rate": 2.0261535193935445e-05, "loss": 0.1359, "step": 34577 }, { "epoch": 25.33186813186813, "grad_norm": 0.78515625, "learning_rate": 2.0258822531519788e-05, "loss": 0.0972, "step": 34578 }, { "epoch": 25.33260073260073, "grad_norm": 1.6796875, "learning_rate": 2.0256110013999515e-05, "loss": 0.1388, "step": 34579 }, { "epoch": 25.333333333333332, "grad_norm": 1.0546875, "learning_rate": 2.0253397641387652e-05, "loss": 0.0994, "step": 34580 }, { "epoch": 25.334065934065933, "grad_norm": 0.88671875, "learning_rate": 2.025068541369719e-05, "loss": 0.111, "step": 34581 }, { "epoch": 25.334798534798534, "grad_norm": 0.96484375, "learning_rate": 2.024797333094114e-05, "loss": 0.1071, "step": 34582 }, { "epoch": 25.335531135531134, "grad_norm": 1.7421875, "learning_rate": 2.0245261393132457e-05, "loss": 0.1258, "step": 34583 }, { "epoch": 25.336263736263735, "grad_norm": 0.66796875, "learning_rate": 2.0242549600284185e-05, "loss": 0.1278, "step": 34584 }, { "epoch": 25.336996336996336, "grad_norm": 0.8828125, "learning_rate": 2.0239837952409253e-05, "loss": 0.123, "step": 34585 }, { "epoch": 25.337728937728937, "grad_norm": 1.0234375, "learning_rate": 2.0237126449520728e-05, "loss": 0.1105, "step": 34586 }, { "epoch": 25.338461538461537, "grad_norm": 1.5546875, "learning_rate": 2.023441509163154e-05, "loss": 0.1674, "step": 34587 }, { "epoch": 25.339194139194138, "grad_norm": 0.92578125, "learning_rate": 2.0231703878754714e-05, "loss": 0.1167, "step": 34588 }, { "epoch": 25.33992673992674, "grad_norm": 1.421875, "learning_rate": 2.0228992810903245e-05, "loss": 0.1093, "step": 34589 }, { "epoch": 25.34065934065934, "grad_norm": 1.2265625, "learning_rate": 2.0226281888090072e-05, "loss": 0.1897, "step": 34590 }, { "epoch": 25.34139194139194, "grad_norm": 1.359375, "learning_rate": 2.0223571110328262e-05, "loss": 0.1388, "step": 34591 }, { "epoch": 25.34212454212454, "grad_norm": 1.21875, "learning_rate": 2.022086047763074e-05, "loss": 0.1217, "step": 34592 }, { "epoch": 25.34285714285714, "grad_norm": 1.0546875, "learning_rate": 2.0218149990010533e-05, "loss": 0.1281, "step": 34593 }, { "epoch": 25.343589743589742, "grad_norm": 0.7890625, "learning_rate": 2.021543964748059e-05, "loss": 0.0866, "step": 34594 }, { "epoch": 25.344322344322343, "grad_norm": 0.9375, "learning_rate": 2.021272945005392e-05, "loss": 0.1035, "step": 34595 }, { "epoch": 25.345054945054944, "grad_norm": 0.6875, "learning_rate": 2.0210019397743507e-05, "loss": 0.0949, "step": 34596 }, { "epoch": 25.345787545787545, "grad_norm": 0.8125, "learning_rate": 2.0207309490562354e-05, "loss": 0.1318, "step": 34597 }, { "epoch": 25.346520146520145, "grad_norm": 0.8203125, "learning_rate": 2.0204599728523406e-05, "loss": 0.1425, "step": 34598 }, { "epoch": 25.347252747252746, "grad_norm": 1.1484375, "learning_rate": 2.020189011163966e-05, "loss": 0.1206, "step": 34599 }, { "epoch": 25.347985347985347, "grad_norm": 1.203125, "learning_rate": 2.0199180639924132e-05, "loss": 0.11, "step": 34600 }, { "epoch": 25.348717948717947, "grad_norm": 1.828125, "learning_rate": 2.0196471313389752e-05, "loss": 0.1363, "step": 34601 }, { "epoch": 25.349450549450548, "grad_norm": 1.859375, "learning_rate": 2.019376213204953e-05, "loss": 0.1421, "step": 34602 }, { "epoch": 25.35018315018315, "grad_norm": 0.93359375, "learning_rate": 2.019105309591644e-05, "loss": 0.1229, "step": 34603 }, { "epoch": 25.35091575091575, "grad_norm": 1.234375, "learning_rate": 2.0188344205003483e-05, "loss": 0.1275, "step": 34604 }, { "epoch": 25.35164835164835, "grad_norm": 0.92578125, "learning_rate": 2.01856354593236e-05, "loss": 0.1189, "step": 34605 }, { "epoch": 25.35238095238095, "grad_norm": 0.90625, "learning_rate": 2.0182926858889793e-05, "loss": 0.1147, "step": 34606 }, { "epoch": 25.353113553113552, "grad_norm": 1.2890625, "learning_rate": 2.0180218403715035e-05, "loss": 0.1104, "step": 34607 }, { "epoch": 25.353846153846153, "grad_norm": 1.03125, "learning_rate": 2.017751009381232e-05, "loss": 0.1188, "step": 34608 }, { "epoch": 25.354578754578753, "grad_norm": 0.88671875, "learning_rate": 2.017480192919459e-05, "loss": 0.1275, "step": 34609 }, { "epoch": 25.355311355311354, "grad_norm": 0.75390625, "learning_rate": 2.0172093909874844e-05, "loss": 0.1223, "step": 34610 }, { "epoch": 25.356043956043955, "grad_norm": 0.91796875, "learning_rate": 2.0169386035866047e-05, "loss": 0.1046, "step": 34611 }, { "epoch": 25.356776556776556, "grad_norm": 0.46484375, "learning_rate": 2.0166678307181195e-05, "loss": 0.1042, "step": 34612 }, { "epoch": 25.357509157509156, "grad_norm": 1.2109375, "learning_rate": 2.0163970723833228e-05, "loss": 0.1175, "step": 34613 }, { "epoch": 25.358241758241757, "grad_norm": 0.80078125, "learning_rate": 2.016126328583514e-05, "loss": 0.1099, "step": 34614 }, { "epoch": 25.358974358974358, "grad_norm": 1.015625, "learning_rate": 2.015855599319991e-05, "loss": 0.1563, "step": 34615 }, { "epoch": 25.35970695970696, "grad_norm": 0.5546875, "learning_rate": 2.0155848845940477e-05, "loss": 0.1172, "step": 34616 }, { "epoch": 25.36043956043956, "grad_norm": 1.1640625, "learning_rate": 2.0153141844069857e-05, "loss": 0.102, "step": 34617 }, { "epoch": 25.36117216117216, "grad_norm": 0.8515625, "learning_rate": 2.0150434987600984e-05, "loss": 0.1328, "step": 34618 }, { "epoch": 25.36190476190476, "grad_norm": 0.9140625, "learning_rate": 2.0147728276546852e-05, "loss": 0.1489, "step": 34619 }, { "epoch": 25.36263736263736, "grad_norm": 1.078125, "learning_rate": 2.0145021710920403e-05, "loss": 0.1063, "step": 34620 }, { "epoch": 25.363369963369962, "grad_norm": 1.5703125, "learning_rate": 2.0142315290734616e-05, "loss": 0.131, "step": 34621 }, { "epoch": 25.364102564102563, "grad_norm": 1.234375, "learning_rate": 2.0139609016002476e-05, "loss": 0.1448, "step": 34622 }, { "epoch": 25.364835164835164, "grad_norm": 0.94140625, "learning_rate": 2.0136902886736938e-05, "loss": 0.1301, "step": 34623 }, { "epoch": 25.365567765567764, "grad_norm": 1.2421875, "learning_rate": 2.0134196902950956e-05, "loss": 0.1317, "step": 34624 }, { "epoch": 25.366300366300365, "grad_norm": 1.25, "learning_rate": 2.0131491064657505e-05, "loss": 0.1222, "step": 34625 }, { "epoch": 25.367032967032966, "grad_norm": 0.88671875, "learning_rate": 2.0128785371869566e-05, "loss": 0.1404, "step": 34626 }, { "epoch": 25.367765567765566, "grad_norm": 0.66796875, "learning_rate": 2.0126079824600048e-05, "loss": 0.1105, "step": 34627 }, { "epoch": 25.368498168498167, "grad_norm": 1.09375, "learning_rate": 2.0123374422861995e-05, "loss": 0.1213, "step": 34628 }, { "epoch": 25.369230769230768, "grad_norm": 1.3515625, "learning_rate": 2.0120669166668302e-05, "loss": 0.1345, "step": 34629 }, { "epoch": 25.36996336996337, "grad_norm": 0.953125, "learning_rate": 2.0117964056031976e-05, "loss": 0.1029, "step": 34630 }, { "epoch": 25.37069597069597, "grad_norm": 1.015625, "learning_rate": 2.0115259090965943e-05, "loss": 0.1234, "step": 34631 }, { "epoch": 25.37142857142857, "grad_norm": 0.8515625, "learning_rate": 2.011255427148318e-05, "loss": 0.1261, "step": 34632 }, { "epoch": 25.37216117216117, "grad_norm": 0.671875, "learning_rate": 2.0109849597596638e-05, "loss": 0.121, "step": 34633 }, { "epoch": 25.37289377289377, "grad_norm": 0.75, "learning_rate": 2.0107145069319305e-05, "loss": 0.1172, "step": 34634 }, { "epoch": 25.373626373626372, "grad_norm": 1.0390625, "learning_rate": 2.0104440686664097e-05, "loss": 0.1036, "step": 34635 }, { "epoch": 25.374358974358973, "grad_norm": 0.74609375, "learning_rate": 2.0101736449643997e-05, "loss": 0.0949, "step": 34636 }, { "epoch": 25.375091575091574, "grad_norm": 1.359375, "learning_rate": 2.009903235827195e-05, "loss": 0.1055, "step": 34637 }, { "epoch": 25.375824175824174, "grad_norm": 0.71875, "learning_rate": 2.0096328412560945e-05, "loss": 0.1029, "step": 34638 }, { "epoch": 25.376556776556775, "grad_norm": 1.2265625, "learning_rate": 2.0093624612523886e-05, "loss": 0.1189, "step": 34639 }, { "epoch": 25.377289377289376, "grad_norm": 0.9296875, "learning_rate": 2.0090920958173755e-05, "loss": 0.1545, "step": 34640 }, { "epoch": 25.378021978021977, "grad_norm": 0.9921875, "learning_rate": 2.0088217449523524e-05, "loss": 0.1214, "step": 34641 }, { "epoch": 25.378754578754577, "grad_norm": 1.4375, "learning_rate": 2.0085514086586094e-05, "loss": 0.1346, "step": 34642 }, { "epoch": 25.379487179487178, "grad_norm": 1.125, "learning_rate": 2.008281086937448e-05, "loss": 0.1107, "step": 34643 }, { "epoch": 25.38021978021978, "grad_norm": 0.8359375, "learning_rate": 2.008010779790159e-05, "loss": 0.1199, "step": 34644 }, { "epoch": 25.38095238095238, "grad_norm": 0.88671875, "learning_rate": 2.0077404872180406e-05, "loss": 0.1047, "step": 34645 }, { "epoch": 25.38168498168498, "grad_norm": 1.1875, "learning_rate": 2.0074702092223844e-05, "loss": 0.1313, "step": 34646 }, { "epoch": 25.38241758241758, "grad_norm": 1.75, "learning_rate": 2.0071999458044875e-05, "loss": 0.1197, "step": 34647 }, { "epoch": 25.383150183150182, "grad_norm": 1.0390625, "learning_rate": 2.006929696965644e-05, "loss": 0.1392, "step": 34648 }, { "epoch": 25.383882783882783, "grad_norm": 0.96875, "learning_rate": 2.006659462707152e-05, "loss": 0.1148, "step": 34649 }, { "epoch": 25.384615384615383, "grad_norm": 0.69140625, "learning_rate": 2.006389243030301e-05, "loss": 0.1226, "step": 34650 }, { "epoch": 25.385347985347984, "grad_norm": 1.140625, "learning_rate": 2.0061190379363885e-05, "loss": 0.127, "step": 34651 }, { "epoch": 25.386080586080585, "grad_norm": 1.5546875, "learning_rate": 2.0058488474267105e-05, "loss": 0.1415, "step": 34652 }, { "epoch": 25.386813186813185, "grad_norm": 0.98828125, "learning_rate": 2.005578671502557e-05, "loss": 0.108, "step": 34653 }, { "epoch": 25.387545787545786, "grad_norm": 0.72265625, "learning_rate": 2.0053085101652287e-05, "loss": 0.1151, "step": 34654 }, { "epoch": 25.388278388278387, "grad_norm": 1.7265625, "learning_rate": 2.005038363416015e-05, "loss": 0.1054, "step": 34655 }, { "epoch": 25.389010989010988, "grad_norm": 1.2265625, "learning_rate": 2.0047682312562142e-05, "loss": 0.1216, "step": 34656 }, { "epoch": 25.38974358974359, "grad_norm": 0.83984375, "learning_rate": 2.0044981136871166e-05, "loss": 0.1319, "step": 34657 }, { "epoch": 25.39047619047619, "grad_norm": 0.88671875, "learning_rate": 2.0042280107100182e-05, "loss": 0.1032, "step": 34658 }, { "epoch": 25.39120879120879, "grad_norm": 1.0390625, "learning_rate": 2.0039579223262133e-05, "loss": 0.1504, "step": 34659 }, { "epoch": 25.39194139194139, "grad_norm": 1.03125, "learning_rate": 2.0036878485369978e-05, "loss": 0.1136, "step": 34660 }, { "epoch": 25.39267399267399, "grad_norm": 0.7890625, "learning_rate": 2.0034177893436616e-05, "loss": 0.1199, "step": 34661 }, { "epoch": 25.393406593406592, "grad_norm": 0.66796875, "learning_rate": 2.003147744747502e-05, "loss": 0.107, "step": 34662 }, { "epoch": 25.394139194139193, "grad_norm": 0.61328125, "learning_rate": 2.002877714749811e-05, "loss": 0.1195, "step": 34663 }, { "epoch": 25.394871794871793, "grad_norm": 1.1484375, "learning_rate": 2.0026076993518844e-05, "loss": 0.1238, "step": 34664 }, { "epoch": 25.395604395604394, "grad_norm": 0.9921875, "learning_rate": 2.002337698555014e-05, "loss": 0.0957, "step": 34665 }, { "epoch": 25.396336996336995, "grad_norm": 0.796875, "learning_rate": 2.0020677123604933e-05, "loss": 0.1301, "step": 34666 }, { "epoch": 25.397069597069596, "grad_norm": 1.75, "learning_rate": 2.0017977407696193e-05, "loss": 0.1319, "step": 34667 }, { "epoch": 25.397802197802196, "grad_norm": 0.83984375, "learning_rate": 2.0015277837836794e-05, "loss": 0.122, "step": 34668 }, { "epoch": 25.398534798534797, "grad_norm": 0.9453125, "learning_rate": 2.0012578414039735e-05, "loss": 0.1207, "step": 34669 }, { "epoch": 25.399267399267398, "grad_norm": 0.75390625, "learning_rate": 2.0009879136317903e-05, "loss": 0.1133, "step": 34670 }, { "epoch": 25.4, "grad_norm": 1.0859375, "learning_rate": 2.0007180004684274e-05, "loss": 0.1274, "step": 34671 }, { "epoch": 25.4007326007326, "grad_norm": 1.3515625, "learning_rate": 2.0004481019151733e-05, "loss": 0.1353, "step": 34672 }, { "epoch": 25.4014652014652, "grad_norm": 0.96484375, "learning_rate": 2.000178217973324e-05, "loss": 0.102, "step": 34673 }, { "epoch": 25.4021978021978, "grad_norm": 1.3203125, "learning_rate": 1.9999083486441714e-05, "loss": 0.1128, "step": 34674 }, { "epoch": 25.4029304029304, "grad_norm": 1.421875, "learning_rate": 1.9996384939290116e-05, "loss": 0.1201, "step": 34675 }, { "epoch": 25.403663003663002, "grad_norm": 1.1328125, "learning_rate": 1.9993686538291328e-05, "loss": 0.1282, "step": 34676 }, { "epoch": 25.404395604395603, "grad_norm": 1.0078125, "learning_rate": 1.9990988283458305e-05, "loss": 0.108, "step": 34677 }, { "epoch": 25.405128205128204, "grad_norm": 0.74609375, "learning_rate": 1.998829017480399e-05, "loss": 0.1137, "step": 34678 }, { "epoch": 25.405860805860804, "grad_norm": 1.90625, "learning_rate": 1.9985592212341255e-05, "loss": 0.1208, "step": 34679 }, { "epoch": 25.406593406593405, "grad_norm": 0.76953125, "learning_rate": 1.9982894396083108e-05, "loss": 0.1152, "step": 34680 }, { "epoch": 25.407326007326006, "grad_norm": 1.2734375, "learning_rate": 1.9980196726042407e-05, "loss": 0.1649, "step": 34681 }, { "epoch": 25.408058608058607, "grad_norm": 0.93359375, "learning_rate": 1.9977499202232125e-05, "loss": 0.1115, "step": 34682 }, { "epoch": 25.408791208791207, "grad_norm": 1.1015625, "learning_rate": 1.997480182466514e-05, "loss": 0.1295, "step": 34683 }, { "epoch": 25.409523809523808, "grad_norm": 0.91015625, "learning_rate": 1.997210459335441e-05, "loss": 0.1065, "step": 34684 }, { "epoch": 25.41025641025641, "grad_norm": 0.83203125, "learning_rate": 1.9969407508312842e-05, "loss": 0.1135, "step": 34685 }, { "epoch": 25.41098901098901, "grad_norm": 1.0, "learning_rate": 1.996671056955338e-05, "loss": 0.1259, "step": 34686 }, { "epoch": 25.41172161172161, "grad_norm": 1.046875, "learning_rate": 1.996401377708892e-05, "loss": 0.1181, "step": 34687 }, { "epoch": 25.41245421245421, "grad_norm": 1.3203125, "learning_rate": 1.9961317130932393e-05, "loss": 0.1385, "step": 34688 }, { "epoch": 25.41318681318681, "grad_norm": 0.8984375, "learning_rate": 1.9958620631096714e-05, "loss": 0.1372, "step": 34689 }, { "epoch": 25.413919413919412, "grad_norm": 0.890625, "learning_rate": 1.9955924277594834e-05, "loss": 0.1427, "step": 34690 }, { "epoch": 25.414652014652013, "grad_norm": 0.984375, "learning_rate": 1.9953228070439626e-05, "loss": 0.1182, "step": 34691 }, { "epoch": 25.415384615384614, "grad_norm": 0.90625, "learning_rate": 1.9950532009644035e-05, "loss": 0.1285, "step": 34692 }, { "epoch": 25.416117216117215, "grad_norm": 1.203125, "learning_rate": 1.994783609522099e-05, "loss": 0.1287, "step": 34693 }, { "epoch": 25.416849816849815, "grad_norm": 0.828125, "learning_rate": 1.9945140327183355e-05, "loss": 0.1048, "step": 34694 }, { "epoch": 25.417582417582416, "grad_norm": 0.80078125, "learning_rate": 1.994244470554412e-05, "loss": 0.1448, "step": 34695 }, { "epoch": 25.418315018315017, "grad_norm": 0.8203125, "learning_rate": 1.9939749230316145e-05, "loss": 0.0973, "step": 34696 }, { "epoch": 25.419047619047618, "grad_norm": 1.2578125, "learning_rate": 1.9937053901512383e-05, "loss": 0.1223, "step": 34697 }, { "epoch": 25.41978021978022, "grad_norm": 1.1953125, "learning_rate": 1.993435871914571e-05, "loss": 0.1356, "step": 34698 }, { "epoch": 25.42051282051282, "grad_norm": 1.375, "learning_rate": 1.9931663683229057e-05, "loss": 0.1325, "step": 34699 }, { "epoch": 25.42124542124542, "grad_norm": 1.0390625, "learning_rate": 1.992896879377534e-05, "loss": 0.138, "step": 34700 }, { "epoch": 25.42197802197802, "grad_norm": 0.67578125, "learning_rate": 1.9926274050797486e-05, "loss": 0.1128, "step": 34701 }, { "epoch": 25.42271062271062, "grad_norm": 0.890625, "learning_rate": 1.9923579454308374e-05, "loss": 0.1021, "step": 34702 }, { "epoch": 25.423443223443222, "grad_norm": 1.109375, "learning_rate": 1.992088500432093e-05, "loss": 0.1173, "step": 34703 }, { "epoch": 25.424175824175823, "grad_norm": 1.203125, "learning_rate": 1.991819070084808e-05, "loss": 0.1233, "step": 34704 }, { "epoch": 25.424908424908423, "grad_norm": 0.8203125, "learning_rate": 1.9915496543902683e-05, "loss": 0.1, "step": 34705 }, { "epoch": 25.425641025641024, "grad_norm": 0.953125, "learning_rate": 1.9912802533497717e-05, "loss": 0.1185, "step": 34706 }, { "epoch": 25.426373626373625, "grad_norm": 0.90625, "learning_rate": 1.991010866964604e-05, "loss": 0.1383, "step": 34707 }, { "epoch": 25.427106227106226, "grad_norm": 1.3359375, "learning_rate": 1.990741495236059e-05, "loss": 0.117, "step": 34708 }, { "epoch": 25.427838827838826, "grad_norm": 0.9375, "learning_rate": 1.9904721381654242e-05, "loss": 0.131, "step": 34709 }, { "epoch": 25.428571428571427, "grad_norm": 0.9453125, "learning_rate": 1.990202795753992e-05, "loss": 0.107, "step": 34710 }, { "epoch": 25.429304029304028, "grad_norm": 1.0625, "learning_rate": 1.9899334680030523e-05, "loss": 0.1112, "step": 34711 }, { "epoch": 25.43003663003663, "grad_norm": 1.1953125, "learning_rate": 1.9896641549138975e-05, "loss": 0.1056, "step": 34712 }, { "epoch": 25.43076923076923, "grad_norm": 1.0546875, "learning_rate": 1.9893948564878145e-05, "loss": 0.1278, "step": 34713 }, { "epoch": 25.43150183150183, "grad_norm": 0.93359375, "learning_rate": 1.989125572726096e-05, "loss": 0.1204, "step": 34714 }, { "epoch": 25.43223443223443, "grad_norm": 1.1328125, "learning_rate": 1.9888563036300315e-05, "loss": 0.1609, "step": 34715 }, { "epoch": 25.43296703296703, "grad_norm": 1.125, "learning_rate": 1.988587049200913e-05, "loss": 0.1184, "step": 34716 }, { "epoch": 25.433699633699632, "grad_norm": 0.84765625, "learning_rate": 1.9883178094400276e-05, "loss": 0.1167, "step": 34717 }, { "epoch": 25.434432234432233, "grad_norm": 1.015625, "learning_rate": 1.9880485843486664e-05, "loss": 0.1144, "step": 34718 }, { "epoch": 25.435164835164834, "grad_norm": 1.0546875, "learning_rate": 1.9877793739281207e-05, "loss": 0.1581, "step": 34719 }, { "epoch": 25.435897435897434, "grad_norm": 0.9140625, "learning_rate": 1.987510178179677e-05, "loss": 0.1327, "step": 34720 }, { "epoch": 25.436630036630035, "grad_norm": 1.0625, "learning_rate": 1.9872409971046303e-05, "loss": 0.1161, "step": 34721 }, { "epoch": 25.437362637362636, "grad_norm": 1.0078125, "learning_rate": 1.9869718307042653e-05, "loss": 0.1236, "step": 34722 }, { "epoch": 25.438095238095237, "grad_norm": 1.1484375, "learning_rate": 1.9867026789798757e-05, "loss": 0.1345, "step": 34723 }, { "epoch": 25.438827838827837, "grad_norm": 0.984375, "learning_rate": 1.986433541932748e-05, "loss": 0.0981, "step": 34724 }, { "epoch": 25.439560439560438, "grad_norm": 1.25, "learning_rate": 1.9861644195641727e-05, "loss": 0.1355, "step": 34725 }, { "epoch": 25.44029304029304, "grad_norm": 0.8828125, "learning_rate": 1.985895311875439e-05, "loss": 0.1212, "step": 34726 }, { "epoch": 25.44102564102564, "grad_norm": 0.9296875, "learning_rate": 1.985626218867839e-05, "loss": 0.1035, "step": 34727 }, { "epoch": 25.44175824175824, "grad_norm": 0.8515625, "learning_rate": 1.9853571405426577e-05, "loss": 0.1344, "step": 34728 }, { "epoch": 25.44249084249084, "grad_norm": 1.109375, "learning_rate": 1.9850880769011864e-05, "loss": 0.1217, "step": 34729 }, { "epoch": 25.44322344322344, "grad_norm": 0.8984375, "learning_rate": 1.9848190279447164e-05, "loss": 0.1249, "step": 34730 }, { "epoch": 25.443956043956042, "grad_norm": 0.94921875, "learning_rate": 1.9845499936745306e-05, "loss": 0.1077, "step": 34731 }, { "epoch": 25.444688644688643, "grad_norm": 0.9296875, "learning_rate": 1.984280974091926e-05, "loss": 0.13, "step": 34732 }, { "epoch": 25.445421245421244, "grad_norm": 0.78125, "learning_rate": 1.9840119691981857e-05, "loss": 0.118, "step": 34733 }, { "epoch": 25.446153846153845, "grad_norm": 0.984375, "learning_rate": 1.9837429789946024e-05, "loss": 0.1358, "step": 34734 }, { "epoch": 25.446886446886445, "grad_norm": 1.0546875, "learning_rate": 1.983474003482461e-05, "loss": 0.1133, "step": 34735 }, { "epoch": 25.447619047619046, "grad_norm": 1.640625, "learning_rate": 1.983205042663053e-05, "loss": 0.1272, "step": 34736 }, { "epoch": 25.448351648351647, "grad_norm": 0.9375, "learning_rate": 1.9829360965376654e-05, "loss": 0.1204, "step": 34737 }, { "epoch": 25.449084249084247, "grad_norm": 1.4140625, "learning_rate": 1.9826671651075905e-05, "loss": 0.1201, "step": 34738 }, { "epoch": 25.449816849816848, "grad_norm": 0.9609375, "learning_rate": 1.9823982483741122e-05, "loss": 0.129, "step": 34739 }, { "epoch": 25.45054945054945, "grad_norm": 1.15625, "learning_rate": 1.9821293463385207e-05, "loss": 0.1345, "step": 34740 }, { "epoch": 25.45128205128205, "grad_norm": 1.1328125, "learning_rate": 1.9818604590021045e-05, "loss": 0.1176, "step": 34741 }, { "epoch": 25.45201465201465, "grad_norm": 0.94921875, "learning_rate": 1.9815915863661528e-05, "loss": 0.1225, "step": 34742 }, { "epoch": 25.45274725274725, "grad_norm": 0.81640625, "learning_rate": 1.9813227284319544e-05, "loss": 0.104, "step": 34743 }, { "epoch": 25.453479853479852, "grad_norm": 1.03125, "learning_rate": 1.981053885200794e-05, "loss": 0.1264, "step": 34744 }, { "epoch": 25.454212454212453, "grad_norm": 0.71484375, "learning_rate": 1.9807850566739642e-05, "loss": 0.1229, "step": 34745 }, { "epoch": 25.454945054945053, "grad_norm": 0.53125, "learning_rate": 1.9805162428527467e-05, "loss": 0.0958, "step": 34746 }, { "epoch": 25.455677655677654, "grad_norm": 1.3359375, "learning_rate": 1.980247443738438e-05, "loss": 0.116, "step": 34747 }, { "epoch": 25.456410256410255, "grad_norm": 1.0234375, "learning_rate": 1.979978659332319e-05, "loss": 0.1485, "step": 34748 }, { "epoch": 25.457142857142856, "grad_norm": 0.546875, "learning_rate": 1.979709889635682e-05, "loss": 0.0919, "step": 34749 }, { "epoch": 25.457875457875456, "grad_norm": 1.1171875, "learning_rate": 1.9794411346498114e-05, "loss": 0.1032, "step": 34750 }, { "epoch": 25.458608058608057, "grad_norm": 0.8125, "learning_rate": 1.9791723943759954e-05, "loss": 0.1056, "step": 34751 }, { "epoch": 25.459340659340658, "grad_norm": 0.7265625, "learning_rate": 1.9789036688155233e-05, "loss": 0.1218, "step": 34752 }, { "epoch": 25.46007326007326, "grad_norm": 0.74609375, "learning_rate": 1.9786349579696835e-05, "loss": 0.1109, "step": 34753 }, { "epoch": 25.46080586080586, "grad_norm": 1.1015625, "learning_rate": 1.97836626183976e-05, "loss": 0.1196, "step": 34754 }, { "epoch": 25.46153846153846, "grad_norm": 1.0, "learning_rate": 1.9780975804270425e-05, "loss": 0.1335, "step": 34755 }, { "epoch": 25.46227106227106, "grad_norm": 0.84375, "learning_rate": 1.9778289137328195e-05, "loss": 0.1326, "step": 34756 }, { "epoch": 25.46300366300366, "grad_norm": 1.1640625, "learning_rate": 1.9775602617583726e-05, "loss": 0.1303, "step": 34757 }, { "epoch": 25.463736263736262, "grad_norm": 0.97265625, "learning_rate": 1.977291624504997e-05, "loss": 0.1173, "step": 34758 }, { "epoch": 25.464468864468863, "grad_norm": 1.5703125, "learning_rate": 1.9770230019739745e-05, "loss": 0.1099, "step": 34759 }, { "epoch": 25.465201465201464, "grad_norm": 1.171875, "learning_rate": 1.9767543941665953e-05, "loss": 0.1479, "step": 34760 }, { "epoch": 25.465934065934064, "grad_norm": 1.7421875, "learning_rate": 1.976485801084143e-05, "loss": 0.1201, "step": 34761 }, { "epoch": 25.466666666666665, "grad_norm": 0.78515625, "learning_rate": 1.9762172227279054e-05, "loss": 0.1087, "step": 34762 }, { "epoch": 25.467399267399266, "grad_norm": 1.34375, "learning_rate": 1.975948659099171e-05, "loss": 0.1037, "step": 34763 }, { "epoch": 25.468131868131866, "grad_norm": 1.140625, "learning_rate": 1.9756801101992267e-05, "loss": 0.1254, "step": 34764 }, { "epoch": 25.468864468864467, "grad_norm": 0.85546875, "learning_rate": 1.975411576029357e-05, "loss": 0.1283, "step": 34765 }, { "epoch": 25.469597069597068, "grad_norm": 1.0859375, "learning_rate": 1.97514305659085e-05, "loss": 0.1625, "step": 34766 }, { "epoch": 25.47032967032967, "grad_norm": 1.0, "learning_rate": 1.974874551884992e-05, "loss": 0.1419, "step": 34767 }, { "epoch": 25.47106227106227, "grad_norm": 1.0546875, "learning_rate": 1.974606061913069e-05, "loss": 0.1025, "step": 34768 }, { "epoch": 25.47179487179487, "grad_norm": 0.97265625, "learning_rate": 1.97433758667637e-05, "loss": 0.1226, "step": 34769 }, { "epoch": 25.47252747252747, "grad_norm": 1.0390625, "learning_rate": 1.9740691261761777e-05, "loss": 0.1198, "step": 34770 }, { "epoch": 25.47326007326007, "grad_norm": 1.109375, "learning_rate": 1.9738006804137813e-05, "loss": 0.0978, "step": 34771 }, { "epoch": 25.473992673992672, "grad_norm": 1.3359375, "learning_rate": 1.9735322493904624e-05, "loss": 0.1606, "step": 34772 }, { "epoch": 25.474725274725273, "grad_norm": 1.2578125, "learning_rate": 1.973263833107514e-05, "loss": 0.0971, "step": 34773 }, { "epoch": 25.475457875457874, "grad_norm": 1.2109375, "learning_rate": 1.9729954315662162e-05, "loss": 0.1276, "step": 34774 }, { "epoch": 25.476190476190474, "grad_norm": 1.03125, "learning_rate": 1.9727270447678598e-05, "loss": 0.1144, "step": 34775 }, { "epoch": 25.476923076923075, "grad_norm": 1.46875, "learning_rate": 1.972458672713727e-05, "loss": 0.1352, "step": 34776 }, { "epoch": 25.477655677655676, "grad_norm": 0.66015625, "learning_rate": 1.972190315405104e-05, "loss": 0.1295, "step": 34777 }, { "epoch": 25.478388278388277, "grad_norm": 1.0390625, "learning_rate": 1.971921972843278e-05, "loss": 0.1138, "step": 34778 }, { "epoch": 25.479120879120877, "grad_norm": 1.625, "learning_rate": 1.971653645029536e-05, "loss": 0.1202, "step": 34779 }, { "epoch": 25.479853479853478, "grad_norm": 1.296875, "learning_rate": 1.9713853319651606e-05, "loss": 0.1422, "step": 34780 }, { "epoch": 25.48058608058608, "grad_norm": 1.109375, "learning_rate": 1.9711170336514386e-05, "loss": 0.1331, "step": 34781 }, { "epoch": 25.48131868131868, "grad_norm": 1.21875, "learning_rate": 1.9708487500896573e-05, "loss": 0.118, "step": 34782 }, { "epoch": 25.48205128205128, "grad_norm": 1.375, "learning_rate": 1.9705804812810968e-05, "loss": 0.1244, "step": 34783 }, { "epoch": 25.48278388278388, "grad_norm": 1.4609375, "learning_rate": 1.9703122272270496e-05, "loss": 0.1107, "step": 34784 }, { "epoch": 25.483516483516482, "grad_norm": 1.3671875, "learning_rate": 1.9700439879287957e-05, "loss": 0.1185, "step": 34785 }, { "epoch": 25.484249084249083, "grad_norm": 0.796875, "learning_rate": 1.9697757633876237e-05, "loss": 0.1433, "step": 34786 }, { "epoch": 25.484981684981683, "grad_norm": 0.65625, "learning_rate": 1.969507553604816e-05, "loss": 0.1212, "step": 34787 }, { "epoch": 25.485714285714284, "grad_norm": 1.078125, "learning_rate": 1.9692393585816584e-05, "loss": 0.1411, "step": 34788 }, { "epoch": 25.486446886446885, "grad_norm": 0.61328125, "learning_rate": 1.9689711783194374e-05, "loss": 0.1079, "step": 34789 }, { "epoch": 25.487179487179485, "grad_norm": 0.80078125, "learning_rate": 1.9687030128194376e-05, "loss": 0.1216, "step": 34790 }, { "epoch": 25.487912087912086, "grad_norm": 0.9453125, "learning_rate": 1.9684348620829416e-05, "loss": 0.1407, "step": 34791 }, { "epoch": 25.488644688644687, "grad_norm": 1.09375, "learning_rate": 1.9681667261112365e-05, "loss": 0.1397, "step": 34792 }, { "epoch": 25.489377289377288, "grad_norm": 0.99609375, "learning_rate": 1.967898604905605e-05, "loss": 0.1078, "step": 34793 }, { "epoch": 25.49010989010989, "grad_norm": 0.828125, "learning_rate": 1.9676304984673342e-05, "loss": 0.1049, "step": 34794 }, { "epoch": 25.49084249084249, "grad_norm": 0.95703125, "learning_rate": 1.967362406797709e-05, "loss": 0.1408, "step": 34795 }, { "epoch": 25.49157509157509, "grad_norm": 0.9921875, "learning_rate": 1.967094329898011e-05, "loss": 0.1641, "step": 34796 }, { "epoch": 25.49230769230769, "grad_norm": 1.078125, "learning_rate": 1.9668262677695268e-05, "loss": 0.1045, "step": 34797 }, { "epoch": 25.493040293040295, "grad_norm": 0.98828125, "learning_rate": 1.966558220413537e-05, "loss": 0.1376, "step": 34798 }, { "epoch": 25.493772893772892, "grad_norm": 1.265625, "learning_rate": 1.966290187831333e-05, "loss": 0.1319, "step": 34799 }, { "epoch": 25.494505494505496, "grad_norm": 1.046875, "learning_rate": 1.9660221700241934e-05, "loss": 0.1292, "step": 34800 }, { "epoch": 25.495238095238093, "grad_norm": 0.88671875, "learning_rate": 1.965754166993405e-05, "loss": 0.1073, "step": 34801 }, { "epoch": 25.495970695970698, "grad_norm": 1.40625, "learning_rate": 1.9654861787402496e-05, "loss": 0.1265, "step": 34802 }, { "epoch": 25.496703296703295, "grad_norm": 1.1171875, "learning_rate": 1.9652182052660125e-05, "loss": 0.1182, "step": 34803 }, { "epoch": 25.4974358974359, "grad_norm": 1.1640625, "learning_rate": 1.964950246571978e-05, "loss": 0.1303, "step": 34804 }, { "epoch": 25.498168498168496, "grad_norm": 1.0546875, "learning_rate": 1.9646823026594306e-05, "loss": 0.1, "step": 34805 }, { "epoch": 25.4989010989011, "grad_norm": 0.8046875, "learning_rate": 1.9644143735296513e-05, "loss": 0.1113, "step": 34806 }, { "epoch": 25.499633699633698, "grad_norm": 0.98046875, "learning_rate": 1.964146459183926e-05, "loss": 0.1335, "step": 34807 }, { "epoch": 25.500366300366302, "grad_norm": 0.80078125, "learning_rate": 1.9638785596235404e-05, "loss": 0.1053, "step": 34808 }, { "epoch": 25.5010989010989, "grad_norm": 0.890625, "learning_rate": 1.963610674849772e-05, "loss": 0.1151, "step": 34809 }, { "epoch": 25.501831501831504, "grad_norm": 0.7890625, "learning_rate": 1.963342804863911e-05, "loss": 0.1031, "step": 34810 }, { "epoch": 25.5025641025641, "grad_norm": 0.98046875, "learning_rate": 1.9630749496672364e-05, "loss": 0.124, "step": 34811 }, { "epoch": 25.503296703296705, "grad_norm": 1.03125, "learning_rate": 1.9628071092610347e-05, "loss": 0.1299, "step": 34812 }, { "epoch": 25.504029304029302, "grad_norm": 0.6953125, "learning_rate": 1.9625392836465857e-05, "loss": 0.111, "step": 34813 }, { "epoch": 25.504761904761907, "grad_norm": 1.171875, "learning_rate": 1.9622714728251755e-05, "loss": 0.1167, "step": 34814 }, { "epoch": 25.505494505494504, "grad_norm": 0.78125, "learning_rate": 1.9620036767980853e-05, "loss": 0.1042, "step": 34815 }, { "epoch": 25.506227106227108, "grad_norm": 0.8515625, "learning_rate": 1.9617358955666012e-05, "loss": 0.1026, "step": 34816 }, { "epoch": 25.506959706959705, "grad_norm": 1.4765625, "learning_rate": 1.9614681291320026e-05, "loss": 0.1212, "step": 34817 }, { "epoch": 25.50769230769231, "grad_norm": 1.1328125, "learning_rate": 1.9612003774955742e-05, "loss": 0.1542, "step": 34818 }, { "epoch": 25.508424908424907, "grad_norm": 0.73046875, "learning_rate": 1.9609326406585986e-05, "loss": 0.1224, "step": 34819 }, { "epoch": 25.50915750915751, "grad_norm": 1.5625, "learning_rate": 1.9606649186223592e-05, "loss": 0.1264, "step": 34820 }, { "epoch": 25.509890109890108, "grad_norm": 0.82421875, "learning_rate": 1.9603972113881395e-05, "loss": 0.1249, "step": 34821 }, { "epoch": 25.510622710622712, "grad_norm": 1.015625, "learning_rate": 1.9601295189572195e-05, "loss": 0.1319, "step": 34822 }, { "epoch": 25.51135531135531, "grad_norm": 1.3125, "learning_rate": 1.9598618413308854e-05, "loss": 0.0926, "step": 34823 }, { "epoch": 25.512087912087914, "grad_norm": 0.76171875, "learning_rate": 1.9595941785104138e-05, "loss": 0.0949, "step": 34824 }, { "epoch": 25.51282051282051, "grad_norm": 1.2734375, "learning_rate": 1.9593265304970938e-05, "loss": 0.1291, "step": 34825 }, { "epoch": 25.513553113553115, "grad_norm": 1.421875, "learning_rate": 1.9590588972922042e-05, "loss": 0.1315, "step": 34826 }, { "epoch": 25.514285714285712, "grad_norm": 1.1875, "learning_rate": 1.9587912788970294e-05, "loss": 0.0886, "step": 34827 }, { "epoch": 25.515018315018317, "grad_norm": 1.1328125, "learning_rate": 1.9585236753128495e-05, "loss": 0.1169, "step": 34828 }, { "epoch": 25.515750915750917, "grad_norm": 0.58984375, "learning_rate": 1.9582560865409464e-05, "loss": 0.1074, "step": 34829 }, { "epoch": 25.516483516483518, "grad_norm": 0.99609375, "learning_rate": 1.957988512582604e-05, "loss": 0.1318, "step": 34830 }, { "epoch": 25.51721611721612, "grad_norm": 0.78125, "learning_rate": 1.9577209534391054e-05, "loss": 0.0986, "step": 34831 }, { "epoch": 25.51794871794872, "grad_norm": 0.76953125, "learning_rate": 1.957453409111729e-05, "loss": 0.1037, "step": 34832 }, { "epoch": 25.51868131868132, "grad_norm": 1.1796875, "learning_rate": 1.9571858796017582e-05, "loss": 0.1488, "step": 34833 }, { "epoch": 25.51941391941392, "grad_norm": 0.86328125, "learning_rate": 1.956918364910477e-05, "loss": 0.1007, "step": 34834 }, { "epoch": 25.520146520146522, "grad_norm": 0.63671875, "learning_rate": 1.9566508650391618e-05, "loss": 0.1214, "step": 34835 }, { "epoch": 25.520879120879123, "grad_norm": 1.2578125, "learning_rate": 1.9563833799891014e-05, "loss": 0.105, "step": 34836 }, { "epoch": 25.521611721611723, "grad_norm": 1.1484375, "learning_rate": 1.956115909761573e-05, "loss": 0.1285, "step": 34837 }, { "epoch": 25.522344322344324, "grad_norm": 1.03125, "learning_rate": 1.9558484543578588e-05, "loss": 0.1118, "step": 34838 }, { "epoch": 25.523076923076925, "grad_norm": 1.9921875, "learning_rate": 1.95558101377924e-05, "loss": 0.1327, "step": 34839 }, { "epoch": 25.523809523809526, "grad_norm": 0.921875, "learning_rate": 1.955313588026998e-05, "loss": 0.1238, "step": 34840 }, { "epoch": 25.524542124542126, "grad_norm": 0.9375, "learning_rate": 1.9550461771024146e-05, "loss": 0.101, "step": 34841 }, { "epoch": 25.525274725274727, "grad_norm": 1.421875, "learning_rate": 1.954778781006773e-05, "loss": 0.1535, "step": 34842 }, { "epoch": 25.526007326007328, "grad_norm": 1.03125, "learning_rate": 1.9545113997413502e-05, "loss": 0.0858, "step": 34843 }, { "epoch": 25.52673992673993, "grad_norm": 1.2421875, "learning_rate": 1.9542440333074294e-05, "loss": 0.0971, "step": 34844 }, { "epoch": 25.52747252747253, "grad_norm": 1.4296875, "learning_rate": 1.9539766817062915e-05, "loss": 0.139, "step": 34845 }, { "epoch": 25.52820512820513, "grad_norm": 1.015625, "learning_rate": 1.9537093449392182e-05, "loss": 0.1069, "step": 34846 }, { "epoch": 25.52893772893773, "grad_norm": 0.66796875, "learning_rate": 1.9534420230074916e-05, "loss": 0.1145, "step": 34847 }, { "epoch": 25.52967032967033, "grad_norm": 2.21875, "learning_rate": 1.953174715912389e-05, "loss": 0.1288, "step": 34848 }, { "epoch": 25.530402930402932, "grad_norm": 1.3125, "learning_rate": 1.9529074236551943e-05, "loss": 0.1458, "step": 34849 }, { "epoch": 25.531135531135533, "grad_norm": 1.1015625, "learning_rate": 1.952640146237184e-05, "loss": 0.1093, "step": 34850 }, { "epoch": 25.531868131868134, "grad_norm": 0.8671875, "learning_rate": 1.952372883659645e-05, "loss": 0.1325, "step": 34851 }, { "epoch": 25.532600732600734, "grad_norm": 1.03125, "learning_rate": 1.952105635923852e-05, "loss": 0.1287, "step": 34852 }, { "epoch": 25.533333333333335, "grad_norm": 1.328125, "learning_rate": 1.9518384030310897e-05, "loss": 0.1234, "step": 34853 }, { "epoch": 25.534065934065936, "grad_norm": 0.94921875, "learning_rate": 1.951571184982635e-05, "loss": 0.1311, "step": 34854 }, { "epoch": 25.534798534798536, "grad_norm": 1.1796875, "learning_rate": 1.95130398177977e-05, "loss": 0.118, "step": 34855 }, { "epoch": 25.535531135531137, "grad_norm": 1.125, "learning_rate": 1.9510367934237756e-05, "loss": 0.1402, "step": 34856 }, { "epoch": 25.536263736263738, "grad_norm": 1.015625, "learning_rate": 1.9507696199159305e-05, "loss": 0.1155, "step": 34857 }, { "epoch": 25.53699633699634, "grad_norm": 0.83203125, "learning_rate": 1.9505024612575172e-05, "loss": 0.0886, "step": 34858 }, { "epoch": 25.53772893772894, "grad_norm": 1.171875, "learning_rate": 1.9502353174498128e-05, "loss": 0.1283, "step": 34859 }, { "epoch": 25.53846153846154, "grad_norm": 1.1171875, "learning_rate": 1.9499681884941003e-05, "loss": 0.1138, "step": 34860 }, { "epoch": 25.53919413919414, "grad_norm": 1.0703125, "learning_rate": 1.9497010743916543e-05, "loss": 0.1123, "step": 34861 }, { "epoch": 25.53992673992674, "grad_norm": 1.109375, "learning_rate": 1.9494339751437613e-05, "loss": 0.1406, "step": 34862 }, { "epoch": 25.540659340659342, "grad_norm": 0.65625, "learning_rate": 1.9491668907516965e-05, "loss": 0.0985, "step": 34863 }, { "epoch": 25.541391941391943, "grad_norm": 1.09375, "learning_rate": 1.948899821216743e-05, "loss": 0.1192, "step": 34864 }, { "epoch": 25.542124542124544, "grad_norm": 0.9140625, "learning_rate": 1.948632766540176e-05, "loss": 0.1089, "step": 34865 }, { "epoch": 25.542857142857144, "grad_norm": 0.93359375, "learning_rate": 1.9483657267232783e-05, "loss": 0.1131, "step": 34866 }, { "epoch": 25.543589743589745, "grad_norm": 0.9140625, "learning_rate": 1.9480987017673285e-05, "loss": 0.1279, "step": 34867 }, { "epoch": 25.544322344322346, "grad_norm": 0.81640625, "learning_rate": 1.9478316916736078e-05, "loss": 0.1062, "step": 34868 }, { "epoch": 25.545054945054947, "grad_norm": 1.5078125, "learning_rate": 1.947564696443392e-05, "loss": 0.1284, "step": 34869 }, { "epoch": 25.545787545787547, "grad_norm": 1.15625, "learning_rate": 1.9472977160779625e-05, "loss": 0.1135, "step": 34870 }, { "epoch": 25.546520146520148, "grad_norm": 0.83203125, "learning_rate": 1.9470307505785973e-05, "loss": 0.1214, "step": 34871 }, { "epoch": 25.54725274725275, "grad_norm": 1.0390625, "learning_rate": 1.9467637999465773e-05, "loss": 0.1372, "step": 34872 }, { "epoch": 25.54798534798535, "grad_norm": 0.9140625, "learning_rate": 1.946496864183182e-05, "loss": 0.1005, "step": 34873 }, { "epoch": 25.54871794871795, "grad_norm": 1.1875, "learning_rate": 1.9462299432896875e-05, "loss": 0.116, "step": 34874 }, { "epoch": 25.54945054945055, "grad_norm": 0.93359375, "learning_rate": 1.945963037267376e-05, "loss": 0.1108, "step": 34875 }, { "epoch": 25.550183150183152, "grad_norm": 1.5625, "learning_rate": 1.9456961461175213e-05, "loss": 0.1305, "step": 34876 }, { "epoch": 25.550915750915753, "grad_norm": 1.0078125, "learning_rate": 1.9454292698414087e-05, "loss": 0.1501, "step": 34877 }, { "epoch": 25.551648351648353, "grad_norm": 0.94140625, "learning_rate": 1.9451624084403117e-05, "loss": 0.1313, "step": 34878 }, { "epoch": 25.552380952380954, "grad_norm": 1.5, "learning_rate": 1.9448955619155124e-05, "loss": 0.1301, "step": 34879 }, { "epoch": 25.553113553113555, "grad_norm": 0.82421875, "learning_rate": 1.9446287302682865e-05, "loss": 0.1347, "step": 34880 }, { "epoch": 25.553846153846155, "grad_norm": 1.203125, "learning_rate": 1.9443619134999135e-05, "loss": 0.0939, "step": 34881 }, { "epoch": 25.554578754578756, "grad_norm": 1.015625, "learning_rate": 1.9440951116116726e-05, "loss": 0.1318, "step": 34882 }, { "epoch": 25.555311355311357, "grad_norm": 1.78125, "learning_rate": 1.9438283246048414e-05, "loss": 0.1284, "step": 34883 }, { "epoch": 25.556043956043958, "grad_norm": 1.1796875, "learning_rate": 1.9435615524807e-05, "loss": 0.1101, "step": 34884 }, { "epoch": 25.55677655677656, "grad_norm": 0.75390625, "learning_rate": 1.943294795240523e-05, "loss": 0.1013, "step": 34885 }, { "epoch": 25.55750915750916, "grad_norm": 1.125, "learning_rate": 1.9430280528855925e-05, "loss": 0.0811, "step": 34886 }, { "epoch": 25.55824175824176, "grad_norm": 0.828125, "learning_rate": 1.9427613254171808e-05, "loss": 0.0906, "step": 34887 }, { "epoch": 25.55897435897436, "grad_norm": 1.3984375, "learning_rate": 1.9424946128365732e-05, "loss": 0.1043, "step": 34888 }, { "epoch": 25.55970695970696, "grad_norm": 0.91796875, "learning_rate": 1.9422279151450427e-05, "loss": 0.1151, "step": 34889 }, { "epoch": 25.560439560439562, "grad_norm": 1.5390625, "learning_rate": 1.9419612323438703e-05, "loss": 0.1251, "step": 34890 }, { "epoch": 25.561172161172163, "grad_norm": 1.0703125, "learning_rate": 1.9416945644343296e-05, "loss": 0.08, "step": 34891 }, { "epoch": 25.561904761904763, "grad_norm": 1.1796875, "learning_rate": 1.9414279114177013e-05, "loss": 0.1253, "step": 34892 }, { "epoch": 25.562637362637364, "grad_norm": 0.7109375, "learning_rate": 1.9411612732952617e-05, "loss": 0.1064, "step": 34893 }, { "epoch": 25.563369963369965, "grad_norm": 1.125, "learning_rate": 1.940894650068291e-05, "loss": 0.1222, "step": 34894 }, { "epoch": 25.564102564102566, "grad_norm": 0.87890625, "learning_rate": 1.940628041738063e-05, "loss": 0.1042, "step": 34895 }, { "epoch": 25.564835164835166, "grad_norm": 0.80859375, "learning_rate": 1.9403614483058567e-05, "loss": 0.1079, "step": 34896 }, { "epoch": 25.565567765567767, "grad_norm": 0.92578125, "learning_rate": 1.9400948697729496e-05, "loss": 0.1057, "step": 34897 }, { "epoch": 25.566300366300368, "grad_norm": 1.265625, "learning_rate": 1.939828306140619e-05, "loss": 0.1156, "step": 34898 }, { "epoch": 25.56703296703297, "grad_norm": 1.34375, "learning_rate": 1.9395617574101436e-05, "loss": 0.1058, "step": 34899 }, { "epoch": 25.56776556776557, "grad_norm": 1.40625, "learning_rate": 1.9392952235827973e-05, "loss": 0.1418, "step": 34900 }, { "epoch": 25.56849816849817, "grad_norm": 0.53125, "learning_rate": 1.93902870465986e-05, "loss": 0.1062, "step": 34901 }, { "epoch": 25.56923076923077, "grad_norm": 0.90234375, "learning_rate": 1.9387622006426043e-05, "loss": 0.111, "step": 34902 }, { "epoch": 25.56996336996337, "grad_norm": 1.4765625, "learning_rate": 1.9384957115323137e-05, "loss": 0.1583, "step": 34903 }, { "epoch": 25.570695970695972, "grad_norm": 0.8984375, "learning_rate": 1.9382292373302603e-05, "loss": 0.147, "step": 34904 }, { "epoch": 25.571428571428573, "grad_norm": 1.0390625, "learning_rate": 1.9379627780377235e-05, "loss": 0.1306, "step": 34905 }, { "epoch": 25.572161172161174, "grad_norm": 0.765625, "learning_rate": 1.9376963336559766e-05, "loss": 0.1252, "step": 34906 }, { "epoch": 25.572893772893774, "grad_norm": 0.88671875, "learning_rate": 1.937429904186299e-05, "loss": 0.1288, "step": 34907 }, { "epoch": 25.573626373626375, "grad_norm": 0.73828125, "learning_rate": 1.937163489629966e-05, "loss": 0.113, "step": 34908 }, { "epoch": 25.574358974358976, "grad_norm": 1.8046875, "learning_rate": 1.9368970899882546e-05, "loss": 0.1106, "step": 34909 }, { "epoch": 25.575091575091577, "grad_norm": 1.46875, "learning_rate": 1.936630705262443e-05, "loss": 0.1298, "step": 34910 }, { "epoch": 25.575824175824177, "grad_norm": 1.0859375, "learning_rate": 1.9363643354538037e-05, "loss": 0.1182, "step": 34911 }, { "epoch": 25.576556776556778, "grad_norm": 1.4609375, "learning_rate": 1.9360979805636173e-05, "loss": 0.1197, "step": 34912 }, { "epoch": 25.57728937728938, "grad_norm": 1.7734375, "learning_rate": 1.9358316405931542e-05, "loss": 0.1559, "step": 34913 }, { "epoch": 25.57802197802198, "grad_norm": 1.5390625, "learning_rate": 1.9355653155436973e-05, "loss": 0.1217, "step": 34914 }, { "epoch": 25.57875457875458, "grad_norm": 1.0859375, "learning_rate": 1.9352990054165174e-05, "loss": 0.106, "step": 34915 }, { "epoch": 25.57948717948718, "grad_norm": 0.97265625, "learning_rate": 1.935032710212895e-05, "loss": 0.1099, "step": 34916 }, { "epoch": 25.58021978021978, "grad_norm": 0.72265625, "learning_rate": 1.934766429934101e-05, "loss": 0.1249, "step": 34917 }, { "epoch": 25.580952380952382, "grad_norm": 1.1328125, "learning_rate": 1.9345001645814136e-05, "loss": 0.1211, "step": 34918 }, { "epoch": 25.581684981684983, "grad_norm": 0.984375, "learning_rate": 1.9342339141561092e-05, "loss": 0.1257, "step": 34919 }, { "epoch": 25.582417582417584, "grad_norm": 1.0078125, "learning_rate": 1.9339676786594644e-05, "loss": 0.1534, "step": 34920 }, { "epoch": 25.583150183150185, "grad_norm": 0.828125, "learning_rate": 1.9337014580927514e-05, "loss": 0.1298, "step": 34921 }, { "epoch": 25.583882783882785, "grad_norm": 0.77734375, "learning_rate": 1.9334352524572476e-05, "loss": 0.1131, "step": 34922 }, { "epoch": 25.584615384615386, "grad_norm": 1.0234375, "learning_rate": 1.9331690617542295e-05, "loss": 0.1416, "step": 34923 }, { "epoch": 25.585347985347987, "grad_norm": 1.40625, "learning_rate": 1.932902885984971e-05, "loss": 0.1305, "step": 34924 }, { "epoch": 25.586080586080588, "grad_norm": 1.09375, "learning_rate": 1.93263672515075e-05, "loss": 0.1275, "step": 34925 }, { "epoch": 25.58681318681319, "grad_norm": 1.1640625, "learning_rate": 1.932370579252838e-05, "loss": 0.1079, "step": 34926 }, { "epoch": 25.58754578754579, "grad_norm": 0.97265625, "learning_rate": 1.9321044482925136e-05, "loss": 0.1248, "step": 34927 }, { "epoch": 25.58827838827839, "grad_norm": 0.92578125, "learning_rate": 1.931838332271047e-05, "loss": 0.1153, "step": 34928 }, { "epoch": 25.58901098901099, "grad_norm": 1.2734375, "learning_rate": 1.9315722311897198e-05, "loss": 0.1124, "step": 34929 }, { "epoch": 25.58974358974359, "grad_norm": 1.203125, "learning_rate": 1.931306145049802e-05, "loss": 0.1146, "step": 34930 }, { "epoch": 25.590476190476192, "grad_norm": 1.1875, "learning_rate": 1.9310400738525727e-05, "loss": 0.1192, "step": 34931 }, { "epoch": 25.591208791208793, "grad_norm": 1.3515625, "learning_rate": 1.9307740175993024e-05, "loss": 0.1374, "step": 34932 }, { "epoch": 25.591941391941393, "grad_norm": 0.85546875, "learning_rate": 1.9305079762912667e-05, "loss": 0.1202, "step": 34933 }, { "epoch": 25.592673992673994, "grad_norm": 0.91796875, "learning_rate": 1.9302419499297423e-05, "loss": 0.0975, "step": 34934 }, { "epoch": 25.593406593406595, "grad_norm": 1.1953125, "learning_rate": 1.9299759385160025e-05, "loss": 0.0964, "step": 34935 }, { "epoch": 25.594139194139196, "grad_norm": 0.6640625, "learning_rate": 1.9297099420513246e-05, "loss": 0.1166, "step": 34936 }, { "epoch": 25.594871794871796, "grad_norm": 0.8515625, "learning_rate": 1.929443960536978e-05, "loss": 0.0961, "step": 34937 }, { "epoch": 25.595604395604397, "grad_norm": 1.46875, "learning_rate": 1.929177993974242e-05, "loss": 0.1661, "step": 34938 }, { "epoch": 25.596336996336998, "grad_norm": 1.0546875, "learning_rate": 1.9289120423643853e-05, "loss": 0.1391, "step": 34939 }, { "epoch": 25.5970695970696, "grad_norm": 1.0703125, "learning_rate": 1.9286461057086892e-05, "loss": 0.1283, "step": 34940 }, { "epoch": 25.5978021978022, "grad_norm": 0.796875, "learning_rate": 1.9283801840084224e-05, "loss": 0.0997, "step": 34941 }, { "epoch": 25.5985347985348, "grad_norm": 1.453125, "learning_rate": 1.9281142772648627e-05, "loss": 0.1468, "step": 34942 }, { "epoch": 25.5992673992674, "grad_norm": 0.79296875, "learning_rate": 1.9278483854792804e-05, "loss": 0.1067, "step": 34943 }, { "epoch": 25.6, "grad_norm": 1.1875, "learning_rate": 1.9275825086529516e-05, "loss": 0.1539, "step": 34944 }, { "epoch": 25.600732600732602, "grad_norm": 1.0859375, "learning_rate": 1.9273166467871497e-05, "loss": 0.1265, "step": 34945 }, { "epoch": 25.601465201465203, "grad_norm": 1.015625, "learning_rate": 1.927050799883151e-05, "loss": 0.1083, "step": 34946 }, { "epoch": 25.602197802197804, "grad_norm": 1.2265625, "learning_rate": 1.926784967942225e-05, "loss": 0.1182, "step": 34947 }, { "epoch": 25.602930402930404, "grad_norm": 0.703125, "learning_rate": 1.9265191509656477e-05, "loss": 0.1175, "step": 34948 }, { "epoch": 25.603663003663005, "grad_norm": 1.1953125, "learning_rate": 1.9262533489546928e-05, "loss": 0.132, "step": 34949 }, { "epoch": 25.604395604395606, "grad_norm": 0.8125, "learning_rate": 1.9259875619106332e-05, "loss": 0.1058, "step": 34950 }, { "epoch": 25.605128205128207, "grad_norm": 0.94140625, "learning_rate": 1.9257217898347446e-05, "loss": 0.1213, "step": 34951 }, { "epoch": 25.605860805860807, "grad_norm": 0.96484375, "learning_rate": 1.9254560327282967e-05, "loss": 0.1302, "step": 34952 }, { "epoch": 25.606593406593408, "grad_norm": 0.94140625, "learning_rate": 1.9251902905925667e-05, "loss": 0.1207, "step": 34953 }, { "epoch": 25.60732600732601, "grad_norm": 1.0546875, "learning_rate": 1.9249245634288217e-05, "loss": 0.1029, "step": 34954 }, { "epoch": 25.60805860805861, "grad_norm": 0.93359375, "learning_rate": 1.9246588512383434e-05, "loss": 0.1222, "step": 34955 }, { "epoch": 25.60879120879121, "grad_norm": 1.3203125, "learning_rate": 1.9243931540223982e-05, "loss": 0.1321, "step": 34956 }, { "epoch": 25.60952380952381, "grad_norm": 1.796875, "learning_rate": 1.924127471782263e-05, "loss": 0.1505, "step": 34957 }, { "epoch": 25.61025641025641, "grad_norm": 0.75, "learning_rate": 1.9238618045192083e-05, "loss": 0.1059, "step": 34958 }, { "epoch": 25.610989010989012, "grad_norm": 1.0234375, "learning_rate": 1.9235961522345076e-05, "loss": 0.1355, "step": 34959 }, { "epoch": 25.611721611721613, "grad_norm": 0.90625, "learning_rate": 1.9233305149294336e-05, "loss": 0.1205, "step": 34960 }, { "epoch": 25.612454212454214, "grad_norm": 0.95703125, "learning_rate": 1.9230648926052596e-05, "loss": 0.1137, "step": 34961 }, { "epoch": 25.613186813186815, "grad_norm": 1.5546875, "learning_rate": 1.9227992852632602e-05, "loss": 0.1478, "step": 34962 }, { "epoch": 25.613919413919415, "grad_norm": 1.15625, "learning_rate": 1.9225336929047043e-05, "loss": 0.1477, "step": 34963 }, { "epoch": 25.614652014652016, "grad_norm": 0.98046875, "learning_rate": 1.922268115530868e-05, "loss": 0.1004, "step": 34964 }, { "epoch": 25.615384615384617, "grad_norm": 1.0703125, "learning_rate": 1.9220025531430186e-05, "loss": 0.1089, "step": 34965 }, { "epoch": 25.616117216117217, "grad_norm": 1.1171875, "learning_rate": 1.921737005742435e-05, "loss": 0.1207, "step": 34966 }, { "epoch": 25.616849816849818, "grad_norm": 1.0859375, "learning_rate": 1.9214714733303845e-05, "loss": 0.1316, "step": 34967 }, { "epoch": 25.61758241758242, "grad_norm": 0.83984375, "learning_rate": 1.921205955908143e-05, "loss": 0.1322, "step": 34968 }, { "epoch": 25.61831501831502, "grad_norm": 0.6953125, "learning_rate": 1.9209404534769793e-05, "loss": 0.1006, "step": 34969 }, { "epoch": 25.61904761904762, "grad_norm": 1.078125, "learning_rate": 1.920674966038167e-05, "loss": 0.1322, "step": 34970 }, { "epoch": 25.61978021978022, "grad_norm": 1.1015625, "learning_rate": 1.9204094935929782e-05, "loss": 0.1371, "step": 34971 }, { "epoch": 25.620512820512822, "grad_norm": 0.72265625, "learning_rate": 1.9201440361426863e-05, "loss": 0.1273, "step": 34972 }, { "epoch": 25.621245421245423, "grad_norm": 0.828125, "learning_rate": 1.9198785936885608e-05, "loss": 0.1262, "step": 34973 }, { "epoch": 25.621978021978023, "grad_norm": 0.61328125, "learning_rate": 1.9196131662318737e-05, "loss": 0.1142, "step": 34974 }, { "epoch": 25.622710622710624, "grad_norm": 1.0390625, "learning_rate": 1.9193477537738984e-05, "loss": 0.1442, "step": 34975 }, { "epoch": 25.623443223443225, "grad_norm": 0.88671875, "learning_rate": 1.919082356315905e-05, "loss": 0.118, "step": 34976 }, { "epoch": 25.624175824175826, "grad_norm": 1.2109375, "learning_rate": 1.918816973859168e-05, "loss": 0.1323, "step": 34977 }, { "epoch": 25.624908424908426, "grad_norm": 1.34375, "learning_rate": 1.918551606404955e-05, "loss": 0.1385, "step": 34978 }, { "epoch": 25.625641025641027, "grad_norm": 0.8828125, "learning_rate": 1.918286253954541e-05, "loss": 0.1177, "step": 34979 }, { "epoch": 25.626373626373628, "grad_norm": 0.76171875, "learning_rate": 1.9180209165091916e-05, "loss": 0.1247, "step": 34980 }, { "epoch": 25.62710622710623, "grad_norm": 1.5, "learning_rate": 1.9177555940701863e-05, "loss": 0.1612, "step": 34981 }, { "epoch": 25.62783882783883, "grad_norm": 0.8359375, "learning_rate": 1.9174902866387905e-05, "loss": 0.1029, "step": 34982 }, { "epoch": 25.62857142857143, "grad_norm": 0.8359375, "learning_rate": 1.9172249942162784e-05, "loss": 0.1129, "step": 34983 }, { "epoch": 25.62930402930403, "grad_norm": 1.3046875, "learning_rate": 1.9169597168039186e-05, "loss": 0.1334, "step": 34984 }, { "epoch": 25.63003663003663, "grad_norm": 1.15625, "learning_rate": 1.9166944544029826e-05, "loss": 0.1195, "step": 34985 }, { "epoch": 25.630769230769232, "grad_norm": 1.140625, "learning_rate": 1.916429207014742e-05, "loss": 0.1471, "step": 34986 }, { "epoch": 25.631501831501833, "grad_norm": 1.109375, "learning_rate": 1.9161639746404684e-05, "loss": 0.1177, "step": 34987 }, { "epoch": 25.632234432234434, "grad_norm": 0.84765625, "learning_rate": 1.9158987572814333e-05, "loss": 0.0989, "step": 34988 }, { "epoch": 25.632967032967034, "grad_norm": 1.3359375, "learning_rate": 1.9156335549389046e-05, "loss": 0.1217, "step": 34989 }, { "epoch": 25.633699633699635, "grad_norm": 0.8671875, "learning_rate": 1.9153683676141566e-05, "loss": 0.1053, "step": 34990 }, { "epoch": 25.634432234432236, "grad_norm": 1.40625, "learning_rate": 1.9151031953084533e-05, "loss": 0.1577, "step": 34991 }, { "epoch": 25.635164835164836, "grad_norm": 0.84375, "learning_rate": 1.9148380380230742e-05, "loss": 0.1234, "step": 34992 }, { "epoch": 25.635897435897437, "grad_norm": 1.140625, "learning_rate": 1.914572895759283e-05, "loss": 0.1386, "step": 34993 }, { "epoch": 25.636630036630038, "grad_norm": 1.3671875, "learning_rate": 1.914307768518354e-05, "loss": 0.1076, "step": 34994 }, { "epoch": 25.63736263736264, "grad_norm": 1.2734375, "learning_rate": 1.914042656301554e-05, "loss": 0.1284, "step": 34995 }, { "epoch": 25.63809523809524, "grad_norm": 1.1015625, "learning_rate": 1.9137775591101556e-05, "loss": 0.1253, "step": 34996 }, { "epoch": 25.63882783882784, "grad_norm": 1.5390625, "learning_rate": 1.9135124769454282e-05, "loss": 0.1159, "step": 34997 }, { "epoch": 25.63956043956044, "grad_norm": 1.171875, "learning_rate": 1.913247409808642e-05, "loss": 0.121, "step": 34998 }, { "epoch": 25.64029304029304, "grad_norm": 1.2421875, "learning_rate": 1.9129823577010693e-05, "loss": 0.118, "step": 34999 }, { "epoch": 25.641025641025642, "grad_norm": 0.6484375, "learning_rate": 1.912717320623976e-05, "loss": 0.107, "step": 35000 }, { "epoch": 25.641758241758243, "grad_norm": 0.69140625, "learning_rate": 1.912452298578634e-05, "loss": 0.1109, "step": 35001 }, { "epoch": 25.642490842490844, "grad_norm": 0.90625, "learning_rate": 1.9121872915663132e-05, "loss": 0.1331, "step": 35002 }, { "epoch": 25.643223443223444, "grad_norm": 0.60546875, "learning_rate": 1.9119222995882845e-05, "loss": 0.1266, "step": 35003 }, { "epoch": 25.643956043956045, "grad_norm": 1.0234375, "learning_rate": 1.9116573226458143e-05, "loss": 0.1213, "step": 35004 }, { "epoch": 25.644688644688646, "grad_norm": 1.1953125, "learning_rate": 1.9113923607401766e-05, "loss": 0.1348, "step": 35005 }, { "epoch": 25.645421245421247, "grad_norm": 0.6796875, "learning_rate": 1.9111274138726345e-05, "loss": 0.0955, "step": 35006 }, { "epoch": 25.646153846153847, "grad_norm": 1.703125, "learning_rate": 1.910862482044465e-05, "loss": 0.1569, "step": 35007 }, { "epoch": 25.646886446886448, "grad_norm": 1.25, "learning_rate": 1.9105975652569322e-05, "loss": 0.1419, "step": 35008 }, { "epoch": 25.64761904761905, "grad_norm": 1.1640625, "learning_rate": 1.9103326635113085e-05, "loss": 0.1418, "step": 35009 }, { "epoch": 25.64835164835165, "grad_norm": 0.93359375, "learning_rate": 1.9100677768088598e-05, "loss": 0.1477, "step": 35010 }, { "epoch": 25.64908424908425, "grad_norm": 1.4453125, "learning_rate": 1.909802905150858e-05, "loss": 0.148, "step": 35011 }, { "epoch": 25.64981684981685, "grad_norm": 0.921875, "learning_rate": 1.9095380485385707e-05, "loss": 0.1043, "step": 35012 }, { "epoch": 25.650549450549452, "grad_norm": 0.5859375, "learning_rate": 1.9092732069732677e-05, "loss": 0.0881, "step": 35013 }, { "epoch": 25.651282051282053, "grad_norm": 1.03125, "learning_rate": 1.9090083804562192e-05, "loss": 0.1297, "step": 35014 }, { "epoch": 25.652014652014653, "grad_norm": 0.8046875, "learning_rate": 1.9087435689886914e-05, "loss": 0.1299, "step": 35015 }, { "epoch": 25.652747252747254, "grad_norm": 0.71875, "learning_rate": 1.908478772571956e-05, "loss": 0.1145, "step": 35016 }, { "epoch": 25.653479853479855, "grad_norm": 0.62890625, "learning_rate": 1.9082139912072763e-05, "loss": 0.116, "step": 35017 }, { "epoch": 25.654212454212455, "grad_norm": 1.1640625, "learning_rate": 1.9079492248959284e-05, "loss": 0.1255, "step": 35018 }, { "epoch": 25.654945054945056, "grad_norm": 0.91796875, "learning_rate": 1.907684473639176e-05, "loss": 0.0894, "step": 35019 }, { "epoch": 25.655677655677657, "grad_norm": 0.9609375, "learning_rate": 1.9074197374382897e-05, "loss": 0.1051, "step": 35020 }, { "epoch": 25.656410256410258, "grad_norm": 0.9453125, "learning_rate": 1.9071550162945356e-05, "loss": 0.1053, "step": 35021 }, { "epoch": 25.65714285714286, "grad_norm": 0.921875, "learning_rate": 1.9068903102091842e-05, "loss": 0.126, "step": 35022 }, { "epoch": 25.65787545787546, "grad_norm": 0.85546875, "learning_rate": 1.9066256191835026e-05, "loss": 0.1474, "step": 35023 }, { "epoch": 25.65860805860806, "grad_norm": 1.109375, "learning_rate": 1.9063609432187597e-05, "loss": 0.1301, "step": 35024 }, { "epoch": 25.65934065934066, "grad_norm": 1.46875, "learning_rate": 1.906096282316225e-05, "loss": 0.113, "step": 35025 }, { "epoch": 25.66007326007326, "grad_norm": 0.8828125, "learning_rate": 1.9058316364771636e-05, "loss": 0.1055, "step": 35026 }, { "epoch": 25.660805860805862, "grad_norm": 1.1796875, "learning_rate": 1.905567005702845e-05, "loss": 0.1031, "step": 35027 }, { "epoch": 25.661538461538463, "grad_norm": 1.5234375, "learning_rate": 1.9053023899945373e-05, "loss": 0.1447, "step": 35028 }, { "epoch": 25.662271062271063, "grad_norm": 0.9609375, "learning_rate": 1.9050377893535094e-05, "loss": 0.1232, "step": 35029 }, { "epoch": 25.663003663003664, "grad_norm": 0.953125, "learning_rate": 1.9047732037810265e-05, "loss": 0.1264, "step": 35030 }, { "epoch": 25.663736263736265, "grad_norm": 1.234375, "learning_rate": 1.9045086332783592e-05, "loss": 0.1396, "step": 35031 }, { "epoch": 25.664468864468866, "grad_norm": 1.265625, "learning_rate": 1.904244077846772e-05, "loss": 0.1273, "step": 35032 }, { "epoch": 25.665201465201466, "grad_norm": 1.15625, "learning_rate": 1.9039795374875342e-05, "loss": 0.1152, "step": 35033 }, { "epoch": 25.665934065934067, "grad_norm": 0.7734375, "learning_rate": 1.9037150122019133e-05, "loss": 0.1486, "step": 35034 }, { "epoch": 25.666666666666668, "grad_norm": 0.8671875, "learning_rate": 1.903450501991178e-05, "loss": 0.103, "step": 35035 }, { "epoch": 25.66739926739927, "grad_norm": 0.83203125, "learning_rate": 1.903186006856593e-05, "loss": 0.1318, "step": 35036 }, { "epoch": 25.66813186813187, "grad_norm": 0.80078125, "learning_rate": 1.9029215267994267e-05, "loss": 0.1211, "step": 35037 }, { "epoch": 25.66886446886447, "grad_norm": 1.0546875, "learning_rate": 1.9026570618209463e-05, "loss": 0.1476, "step": 35038 }, { "epoch": 25.66959706959707, "grad_norm": 0.96875, "learning_rate": 1.902392611922419e-05, "loss": 0.1123, "step": 35039 }, { "epoch": 25.67032967032967, "grad_norm": 0.99609375, "learning_rate": 1.9021281771051143e-05, "loss": 0.1067, "step": 35040 }, { "epoch": 25.671062271062272, "grad_norm": 0.97265625, "learning_rate": 1.901863757370295e-05, "loss": 0.1199, "step": 35041 }, { "epoch": 25.671794871794873, "grad_norm": 1.5, "learning_rate": 1.901599352719231e-05, "loss": 0.1394, "step": 35042 }, { "epoch": 25.672527472527474, "grad_norm": 1.328125, "learning_rate": 1.9013349631531855e-05, "loss": 0.1619, "step": 35043 }, { "epoch": 25.673260073260074, "grad_norm": 1.046875, "learning_rate": 1.9010705886734308e-05, "loss": 0.1308, "step": 35044 }, { "epoch": 25.673992673992675, "grad_norm": 1.21875, "learning_rate": 1.9008062292812288e-05, "loss": 0.1403, "step": 35045 }, { "epoch": 25.674725274725276, "grad_norm": 1.171875, "learning_rate": 1.90054188497785e-05, "loss": 0.0992, "step": 35046 }, { "epoch": 25.675457875457877, "grad_norm": 0.80859375, "learning_rate": 1.900277555764557e-05, "loss": 0.1135, "step": 35047 }, { "epoch": 25.676190476190477, "grad_norm": 1.203125, "learning_rate": 1.900013241642618e-05, "loss": 0.1226, "step": 35048 }, { "epoch": 25.676923076923078, "grad_norm": 2.453125, "learning_rate": 1.8997489426133e-05, "loss": 0.1454, "step": 35049 }, { "epoch": 25.67765567765568, "grad_norm": 0.671875, "learning_rate": 1.899484658677869e-05, "loss": 0.116, "step": 35050 }, { "epoch": 25.67838827838828, "grad_norm": 0.75390625, "learning_rate": 1.8992203898375926e-05, "loss": 0.0938, "step": 35051 }, { "epoch": 25.67912087912088, "grad_norm": 1.046875, "learning_rate": 1.8989561360937346e-05, "loss": 0.1224, "step": 35052 }, { "epoch": 25.67985347985348, "grad_norm": 0.703125, "learning_rate": 1.898691897447562e-05, "loss": 0.1211, "step": 35053 }, { "epoch": 25.68058608058608, "grad_norm": 0.8125, "learning_rate": 1.8984276739003407e-05, "loss": 0.111, "step": 35054 }, { "epoch": 25.681318681318682, "grad_norm": 0.7421875, "learning_rate": 1.8981634654533388e-05, "loss": 0.1288, "step": 35055 }, { "epoch": 25.682051282051283, "grad_norm": 1.4296875, "learning_rate": 1.8978992721078183e-05, "loss": 0.1244, "step": 35056 }, { "epoch": 25.682783882783884, "grad_norm": 1.2734375, "learning_rate": 1.8976350938650494e-05, "loss": 0.1864, "step": 35057 }, { "epoch": 25.683516483516485, "grad_norm": 0.890625, "learning_rate": 1.8973709307262935e-05, "loss": 0.1294, "step": 35058 }, { "epoch": 25.684249084249085, "grad_norm": 0.60546875, "learning_rate": 1.897106782692819e-05, "loss": 0.1023, "step": 35059 }, { "epoch": 25.684981684981686, "grad_norm": 0.93359375, "learning_rate": 1.8968426497658905e-05, "loss": 0.1503, "step": 35060 }, { "epoch": 25.685714285714287, "grad_norm": 1.5078125, "learning_rate": 1.896578531946776e-05, "loss": 0.1078, "step": 35061 }, { "epoch": 25.686446886446888, "grad_norm": 1.171875, "learning_rate": 1.896314429236737e-05, "loss": 0.1344, "step": 35062 }, { "epoch": 25.68717948717949, "grad_norm": 0.94140625, "learning_rate": 1.896050341637041e-05, "loss": 0.1117, "step": 35063 }, { "epoch": 25.68791208791209, "grad_norm": 1.09375, "learning_rate": 1.895786269148953e-05, "loss": 0.1136, "step": 35064 }, { "epoch": 25.68864468864469, "grad_norm": 1.390625, "learning_rate": 1.895522211773738e-05, "loss": 0.1271, "step": 35065 }, { "epoch": 25.68937728937729, "grad_norm": 1.3671875, "learning_rate": 1.8952581695126638e-05, "loss": 0.1618, "step": 35066 }, { "epoch": 25.69010989010989, "grad_norm": 1.7109375, "learning_rate": 1.8949941423669924e-05, "loss": 0.1109, "step": 35067 }, { "epoch": 25.690842490842492, "grad_norm": 1.6640625, "learning_rate": 1.8947301303379904e-05, "loss": 0.1528, "step": 35068 }, { "epoch": 25.691575091575093, "grad_norm": 1.1171875, "learning_rate": 1.8944661334269194e-05, "loss": 0.1179, "step": 35069 }, { "epoch": 25.692307692307693, "grad_norm": 1.3359375, "learning_rate": 1.89420215163505e-05, "loss": 0.1425, "step": 35070 }, { "epoch": 25.693040293040294, "grad_norm": 1.1484375, "learning_rate": 1.8939381849636427e-05, "loss": 0.1177, "step": 35071 }, { "epoch": 25.693772893772895, "grad_norm": 0.6875, "learning_rate": 1.8936742334139654e-05, "loss": 0.1246, "step": 35072 }, { "epoch": 25.694505494505496, "grad_norm": 0.875, "learning_rate": 1.8934102969872787e-05, "loss": 0.111, "step": 35073 }, { "epoch": 25.695238095238096, "grad_norm": 0.8125, "learning_rate": 1.89314637568485e-05, "loss": 0.1211, "step": 35074 }, { "epoch": 25.695970695970697, "grad_norm": 1.2265625, "learning_rate": 1.892882469507943e-05, "loss": 0.1321, "step": 35075 }, { "epoch": 25.696703296703298, "grad_norm": 0.69140625, "learning_rate": 1.892618578457823e-05, "loss": 0.1299, "step": 35076 }, { "epoch": 25.6974358974359, "grad_norm": 1.4609375, "learning_rate": 1.8923547025357552e-05, "loss": 0.1066, "step": 35077 }, { "epoch": 25.6981684981685, "grad_norm": 1.40625, "learning_rate": 1.8920908417430012e-05, "loss": 0.1013, "step": 35078 }, { "epoch": 25.6989010989011, "grad_norm": 1.3828125, "learning_rate": 1.8918269960808266e-05, "loss": 0.1311, "step": 35079 }, { "epoch": 25.6996336996337, "grad_norm": 0.79296875, "learning_rate": 1.8915631655504958e-05, "loss": 0.1124, "step": 35080 }, { "epoch": 25.7003663003663, "grad_norm": 1.1796875, "learning_rate": 1.8912993501532735e-05, "loss": 0.099, "step": 35081 }, { "epoch": 25.701098901098902, "grad_norm": 0.8359375, "learning_rate": 1.8910355498904223e-05, "loss": 0.1046, "step": 35082 }, { "epoch": 25.701831501831503, "grad_norm": 0.90625, "learning_rate": 1.890771764763208e-05, "loss": 0.1072, "step": 35083 }, { "epoch": 25.702564102564104, "grad_norm": 1.125, "learning_rate": 1.8905079947728914e-05, "loss": 0.1399, "step": 35084 }, { "epoch": 25.703296703296704, "grad_norm": 0.97265625, "learning_rate": 1.8902442399207386e-05, "loss": 0.137, "step": 35085 }, { "epoch": 25.704029304029305, "grad_norm": 1.1328125, "learning_rate": 1.8899805002080125e-05, "loss": 0.1155, "step": 35086 }, { "epoch": 25.704761904761906, "grad_norm": 1.1015625, "learning_rate": 1.8897167756359797e-05, "loss": 0.1158, "step": 35087 }, { "epoch": 25.705494505494507, "grad_norm": 1.03125, "learning_rate": 1.889453066205899e-05, "loss": 0.1323, "step": 35088 }, { "epoch": 25.706227106227107, "grad_norm": 0.96875, "learning_rate": 1.8891893719190358e-05, "loss": 0.1132, "step": 35089 }, { "epoch": 25.706959706959708, "grad_norm": 0.83984375, "learning_rate": 1.8889256927766546e-05, "loss": 0.1453, "step": 35090 }, { "epoch": 25.70769230769231, "grad_norm": 0.9921875, "learning_rate": 1.888662028780018e-05, "loss": 0.0995, "step": 35091 }, { "epoch": 25.70842490842491, "grad_norm": 0.8203125, "learning_rate": 1.888398379930391e-05, "loss": 0.1383, "step": 35092 }, { "epoch": 25.70915750915751, "grad_norm": 2.109375, "learning_rate": 1.8881347462290335e-05, "loss": 0.1586, "step": 35093 }, { "epoch": 25.70989010989011, "grad_norm": 0.65234375, "learning_rate": 1.8878711276772125e-05, "loss": 0.11, "step": 35094 }, { "epoch": 25.71062271062271, "grad_norm": 0.92578125, "learning_rate": 1.887607524276185e-05, "loss": 0.1182, "step": 35095 }, { "epoch": 25.711355311355312, "grad_norm": 0.82421875, "learning_rate": 1.8873439360272218e-05, "loss": 0.1039, "step": 35096 }, { "epoch": 25.712087912087913, "grad_norm": 0.77734375, "learning_rate": 1.88708036293158e-05, "loss": 0.1217, "step": 35097 }, { "epoch": 25.712820512820514, "grad_norm": 1.2421875, "learning_rate": 1.886816804990527e-05, "loss": 0.1426, "step": 35098 }, { "epoch": 25.713553113553115, "grad_norm": 0.69140625, "learning_rate": 1.8865532622053206e-05, "loss": 0.0941, "step": 35099 }, { "epoch": 25.714285714285715, "grad_norm": 0.8203125, "learning_rate": 1.8862897345772263e-05, "loss": 0.1042, "step": 35100 }, { "epoch": 25.715018315018316, "grad_norm": 1.453125, "learning_rate": 1.886026222107506e-05, "loss": 0.138, "step": 35101 }, { "epoch": 25.715750915750917, "grad_norm": 0.96875, "learning_rate": 1.8857627247974236e-05, "loss": 0.11, "step": 35102 }, { "epoch": 25.716483516483517, "grad_norm": 1.2265625, "learning_rate": 1.8854992426482417e-05, "loss": 0.1241, "step": 35103 }, { "epoch": 25.717216117216118, "grad_norm": 0.91796875, "learning_rate": 1.8852357756612206e-05, "loss": 0.1233, "step": 35104 }, { "epoch": 25.71794871794872, "grad_norm": 0.73046875, "learning_rate": 1.8849723238376233e-05, "loss": 0.1072, "step": 35105 }, { "epoch": 25.71868131868132, "grad_norm": 1.1640625, "learning_rate": 1.8847088871787128e-05, "loss": 0.1174, "step": 35106 }, { "epoch": 25.71941391941392, "grad_norm": 1.3984375, "learning_rate": 1.884445465685753e-05, "loss": 0.127, "step": 35107 }, { "epoch": 25.72014652014652, "grad_norm": 1.5546875, "learning_rate": 1.8841820593600023e-05, "loss": 0.1178, "step": 35108 }, { "epoch": 25.720879120879122, "grad_norm": 1.0625, "learning_rate": 1.883918668202726e-05, "loss": 0.1386, "step": 35109 }, { "epoch": 25.721611721611723, "grad_norm": 1.0625, "learning_rate": 1.883655292215183e-05, "loss": 0.1414, "step": 35110 }, { "epoch": 25.722344322344323, "grad_norm": 0.82421875, "learning_rate": 1.8833919313986377e-05, "loss": 0.1168, "step": 35111 }, { "epoch": 25.723076923076924, "grad_norm": 1.53125, "learning_rate": 1.883128585754351e-05, "loss": 0.1111, "step": 35112 }, { "epoch": 25.723809523809525, "grad_norm": 0.6875, "learning_rate": 1.8828652552835843e-05, "loss": 0.1243, "step": 35113 }, { "epoch": 25.724542124542126, "grad_norm": 0.953125, "learning_rate": 1.8826019399876022e-05, "loss": 0.1122, "step": 35114 }, { "epoch": 25.725274725274726, "grad_norm": 1.1328125, "learning_rate": 1.882338639867662e-05, "loss": 0.1354, "step": 35115 }, { "epoch": 25.726007326007327, "grad_norm": 0.8046875, "learning_rate": 1.8820753549250274e-05, "loss": 0.1136, "step": 35116 }, { "epoch": 25.726739926739928, "grad_norm": 0.61328125, "learning_rate": 1.88181208516096e-05, "loss": 0.119, "step": 35117 }, { "epoch": 25.72747252747253, "grad_norm": 1.3203125, "learning_rate": 1.8815488305767212e-05, "loss": 0.1459, "step": 35118 }, { "epoch": 25.72820512820513, "grad_norm": 0.84765625, "learning_rate": 1.8812855911735718e-05, "loss": 0.111, "step": 35119 }, { "epoch": 25.72893772893773, "grad_norm": 1.09375, "learning_rate": 1.8810223669527745e-05, "loss": 0.1191, "step": 35120 }, { "epoch": 25.72967032967033, "grad_norm": 0.9609375, "learning_rate": 1.8807591579155862e-05, "loss": 0.1242, "step": 35121 }, { "epoch": 25.73040293040293, "grad_norm": 1.015625, "learning_rate": 1.8804959640632748e-05, "loss": 0.1178, "step": 35122 }, { "epoch": 25.731135531135532, "grad_norm": 0.99609375, "learning_rate": 1.8802327853970955e-05, "loss": 0.1485, "step": 35123 }, { "epoch": 25.731868131868133, "grad_norm": 0.83203125, "learning_rate": 1.8799696219183132e-05, "loss": 0.1314, "step": 35124 }, { "epoch": 25.732600732600734, "grad_norm": 1.5, "learning_rate": 1.879706473628186e-05, "loss": 0.1175, "step": 35125 }, { "epoch": 25.733333333333334, "grad_norm": 0.75390625, "learning_rate": 1.879443340527975e-05, "loss": 0.0928, "step": 35126 }, { "epoch": 25.734065934065935, "grad_norm": 0.984375, "learning_rate": 1.8791802226189422e-05, "loss": 0.0844, "step": 35127 }, { "epoch": 25.734798534798536, "grad_norm": 0.6484375, "learning_rate": 1.878917119902348e-05, "loss": 0.11, "step": 35128 }, { "epoch": 25.735531135531136, "grad_norm": 1.0546875, "learning_rate": 1.8786540323794546e-05, "loss": 0.1048, "step": 35129 }, { "epoch": 25.736263736263737, "grad_norm": 1.6328125, "learning_rate": 1.8783909600515188e-05, "loss": 0.1408, "step": 35130 }, { "epoch": 25.736996336996338, "grad_norm": 1.2265625, "learning_rate": 1.8781279029198034e-05, "loss": 0.1061, "step": 35131 }, { "epoch": 25.73772893772894, "grad_norm": 0.91015625, "learning_rate": 1.877864860985568e-05, "loss": 0.1145, "step": 35132 }, { "epoch": 25.73846153846154, "grad_norm": 1.015625, "learning_rate": 1.8776018342500758e-05, "loss": 0.1531, "step": 35133 }, { "epoch": 25.73919413919414, "grad_norm": 1.1875, "learning_rate": 1.8773388227145823e-05, "loss": 0.1362, "step": 35134 }, { "epoch": 25.73992673992674, "grad_norm": 0.83984375, "learning_rate": 1.877075826380352e-05, "loss": 0.1424, "step": 35135 }, { "epoch": 25.74065934065934, "grad_norm": 1.5, "learning_rate": 1.8768128452486414e-05, "loss": 0.1345, "step": 35136 }, { "epoch": 25.741391941391942, "grad_norm": 1.265625, "learning_rate": 1.876549879320712e-05, "loss": 0.1246, "step": 35137 }, { "epoch": 25.742124542124543, "grad_norm": 1.1640625, "learning_rate": 1.8762869285978236e-05, "loss": 0.1062, "step": 35138 }, { "epoch": 25.742857142857144, "grad_norm": 0.91796875, "learning_rate": 1.876023993081237e-05, "loss": 0.1031, "step": 35139 }, { "epoch": 25.743589743589745, "grad_norm": 0.96484375, "learning_rate": 1.875761072772213e-05, "loss": 0.1079, "step": 35140 }, { "epoch": 25.744322344322345, "grad_norm": 0.75, "learning_rate": 1.875498167672008e-05, "loss": 0.116, "step": 35141 }, { "epoch": 25.745054945054946, "grad_norm": 1.5078125, "learning_rate": 1.8752352777818828e-05, "loss": 0.1644, "step": 35142 }, { "epoch": 25.745787545787547, "grad_norm": 1.046875, "learning_rate": 1.874972403103098e-05, "loss": 0.1184, "step": 35143 }, { "epoch": 25.746520146520147, "grad_norm": 1.0859375, "learning_rate": 1.874709543636915e-05, "loss": 0.0893, "step": 35144 }, { "epoch": 25.747252747252748, "grad_norm": 1.375, "learning_rate": 1.8744466993845882e-05, "loss": 0.1414, "step": 35145 }, { "epoch": 25.74798534798535, "grad_norm": 0.75390625, "learning_rate": 1.874183870347382e-05, "loss": 0.1215, "step": 35146 }, { "epoch": 25.74871794871795, "grad_norm": 1.046875, "learning_rate": 1.87392105652655e-05, "loss": 0.125, "step": 35147 }, { "epoch": 25.74945054945055, "grad_norm": 1.203125, "learning_rate": 1.8736582579233584e-05, "loss": 0.1402, "step": 35148 }, { "epoch": 25.75018315018315, "grad_norm": 0.90234375, "learning_rate": 1.873395474539061e-05, "loss": 0.1339, "step": 35149 }, { "epoch": 25.750915750915752, "grad_norm": 0.73828125, "learning_rate": 1.8731327063749202e-05, "loss": 0.092, "step": 35150 }, { "epoch": 25.751648351648353, "grad_norm": 0.984375, "learning_rate": 1.872869953432192e-05, "loss": 0.1058, "step": 35151 }, { "epoch": 25.752380952380953, "grad_norm": 0.92578125, "learning_rate": 1.8726072157121373e-05, "loss": 0.1026, "step": 35152 }, { "epoch": 25.753113553113554, "grad_norm": 1.6328125, "learning_rate": 1.8723444932160133e-05, "loss": 0.1473, "step": 35153 }, { "epoch": 25.753846153846155, "grad_norm": 1.296875, "learning_rate": 1.8720817859450808e-05, "loss": 0.106, "step": 35154 }, { "epoch": 25.754578754578755, "grad_norm": 0.9296875, "learning_rate": 1.8718190939005996e-05, "loss": 0.1023, "step": 35155 }, { "epoch": 25.755311355311356, "grad_norm": 0.8125, "learning_rate": 1.871556417083824e-05, "loss": 0.1112, "step": 35156 }, { "epoch": 25.756043956043957, "grad_norm": 1.0234375, "learning_rate": 1.8712937554960154e-05, "loss": 0.1218, "step": 35157 }, { "epoch": 25.756776556776558, "grad_norm": 1.328125, "learning_rate": 1.871031109138432e-05, "loss": 0.1152, "step": 35158 }, { "epoch": 25.75750915750916, "grad_norm": 0.77734375, "learning_rate": 1.870768478012334e-05, "loss": 0.1334, "step": 35159 }, { "epoch": 25.75824175824176, "grad_norm": 0.7265625, "learning_rate": 1.870505862118976e-05, "loss": 0.1106, "step": 35160 }, { "epoch": 25.75897435897436, "grad_norm": 0.703125, "learning_rate": 1.87024326145962e-05, "loss": 0.1353, "step": 35161 }, { "epoch": 25.75970695970696, "grad_norm": 1.75, "learning_rate": 1.86998067603552e-05, "loss": 0.1023, "step": 35162 }, { "epoch": 25.76043956043956, "grad_norm": 1.0703125, "learning_rate": 1.8697181058479377e-05, "loss": 0.1108, "step": 35163 }, { "epoch": 25.761172161172162, "grad_norm": 1.2734375, "learning_rate": 1.869455550898129e-05, "loss": 0.148, "step": 35164 }, { "epoch": 25.761904761904763, "grad_norm": 1.03125, "learning_rate": 1.8691930111873532e-05, "loss": 0.1458, "step": 35165 }, { "epoch": 25.762637362637363, "grad_norm": 0.66015625, "learning_rate": 1.8689304867168698e-05, "loss": 0.0941, "step": 35166 }, { "epoch": 25.763369963369964, "grad_norm": 0.77734375, "learning_rate": 1.8686679774879328e-05, "loss": 0.131, "step": 35167 }, { "epoch": 25.764102564102565, "grad_norm": 0.84765625, "learning_rate": 1.8684054835018012e-05, "loss": 0.1244, "step": 35168 }, { "epoch": 25.764835164835166, "grad_norm": 1.203125, "learning_rate": 1.8681430047597342e-05, "loss": 0.1406, "step": 35169 }, { "epoch": 25.765567765567766, "grad_norm": 0.78515625, "learning_rate": 1.8678805412629907e-05, "loss": 0.1458, "step": 35170 }, { "epoch": 25.766300366300367, "grad_norm": 1.4375, "learning_rate": 1.8676180930128236e-05, "loss": 0.1367, "step": 35171 }, { "epoch": 25.767032967032968, "grad_norm": 1.0546875, "learning_rate": 1.8673556600104954e-05, "loss": 0.1261, "step": 35172 }, { "epoch": 25.76776556776557, "grad_norm": 1.1640625, "learning_rate": 1.8670932422572566e-05, "loss": 0.1312, "step": 35173 }, { "epoch": 25.76849816849817, "grad_norm": 0.8203125, "learning_rate": 1.8668308397543734e-05, "loss": 0.1129, "step": 35174 }, { "epoch": 25.76923076923077, "grad_norm": 0.79296875, "learning_rate": 1.866568452503096e-05, "loss": 0.1255, "step": 35175 }, { "epoch": 25.76996336996337, "grad_norm": 0.83203125, "learning_rate": 1.8663060805046864e-05, "loss": 0.109, "step": 35176 }, { "epoch": 25.77069597069597, "grad_norm": 1.1171875, "learning_rate": 1.8660437237603973e-05, "loss": 0.1042, "step": 35177 }, { "epoch": 25.771428571428572, "grad_norm": 1.0703125, "learning_rate": 1.8657813822714887e-05, "loss": 0.1207, "step": 35178 }, { "epoch": 25.772161172161173, "grad_norm": 1.15625, "learning_rate": 1.8655190560392166e-05, "loss": 0.1059, "step": 35179 }, { "epoch": 25.772893772893774, "grad_norm": 1.3046875, "learning_rate": 1.8652567450648377e-05, "loss": 0.1435, "step": 35180 }, { "epoch": 25.773626373626374, "grad_norm": 1.3046875, "learning_rate": 1.864994449349611e-05, "loss": 0.1179, "step": 35181 }, { "epoch": 25.774358974358975, "grad_norm": 1.25, "learning_rate": 1.86473216889479e-05, "loss": 0.1041, "step": 35182 }, { "epoch": 25.775091575091576, "grad_norm": 0.87109375, "learning_rate": 1.8644699037016328e-05, "loss": 0.0983, "step": 35183 }, { "epoch": 25.775824175824177, "grad_norm": 1.1875, "learning_rate": 1.8642076537713964e-05, "loss": 0.1186, "step": 35184 }, { "epoch": 25.776556776556777, "grad_norm": 0.9453125, "learning_rate": 1.863945419105338e-05, "loss": 0.0939, "step": 35185 }, { "epoch": 25.777289377289378, "grad_norm": 1.4609375, "learning_rate": 1.8636831997047116e-05, "loss": 0.1493, "step": 35186 }, { "epoch": 25.77802197802198, "grad_norm": 0.8125, "learning_rate": 1.8634209955707766e-05, "loss": 0.1094, "step": 35187 }, { "epoch": 25.77875457875458, "grad_norm": 0.75, "learning_rate": 1.8631588067047865e-05, "loss": 0.1229, "step": 35188 }, { "epoch": 25.77948717948718, "grad_norm": 0.84765625, "learning_rate": 1.862896633107998e-05, "loss": 0.1158, "step": 35189 }, { "epoch": 25.78021978021978, "grad_norm": 1.546875, "learning_rate": 1.8626344747816683e-05, "loss": 0.1265, "step": 35190 }, { "epoch": 25.78095238095238, "grad_norm": 1.3359375, "learning_rate": 1.862372331727053e-05, "loss": 0.113, "step": 35191 }, { "epoch": 25.781684981684982, "grad_norm": 1.1640625, "learning_rate": 1.8621102039454106e-05, "loss": 0.1326, "step": 35192 }, { "epoch": 25.782417582417583, "grad_norm": 0.8203125, "learning_rate": 1.8618480914379924e-05, "loss": 0.1326, "step": 35193 }, { "epoch": 25.783150183150184, "grad_norm": 1.0703125, "learning_rate": 1.8615859942060564e-05, "loss": 0.1275, "step": 35194 }, { "epoch": 25.783882783882785, "grad_norm": 0.7890625, "learning_rate": 1.861323912250859e-05, "loss": 0.1289, "step": 35195 }, { "epoch": 25.784615384615385, "grad_norm": 1.0859375, "learning_rate": 1.8610618455736574e-05, "loss": 0.1373, "step": 35196 }, { "epoch": 25.785347985347986, "grad_norm": 1.140625, "learning_rate": 1.8607997941757035e-05, "loss": 0.123, "step": 35197 }, { "epoch": 25.786080586080587, "grad_norm": 0.7578125, "learning_rate": 1.8605377580582562e-05, "loss": 0.1248, "step": 35198 }, { "epoch": 25.786813186813188, "grad_norm": 0.859375, "learning_rate": 1.8602757372225664e-05, "loss": 0.1238, "step": 35199 }, { "epoch": 25.78754578754579, "grad_norm": 0.99609375, "learning_rate": 1.860013731669896e-05, "loss": 0.1245, "step": 35200 }, { "epoch": 25.78827838827839, "grad_norm": 1.125, "learning_rate": 1.8597517414014952e-05, "loss": 0.1166, "step": 35201 }, { "epoch": 25.78901098901099, "grad_norm": 0.78125, "learning_rate": 1.8594897664186227e-05, "loss": 0.1258, "step": 35202 }, { "epoch": 25.78974358974359, "grad_norm": 0.8203125, "learning_rate": 1.8592278067225307e-05, "loss": 0.1324, "step": 35203 }, { "epoch": 25.79047619047619, "grad_norm": 1.3828125, "learning_rate": 1.8589658623144752e-05, "loss": 0.1264, "step": 35204 }, { "epoch": 25.791208791208792, "grad_norm": 2.109375, "learning_rate": 1.8587039331957125e-05, "loss": 0.1142, "step": 35205 }, { "epoch": 25.791941391941393, "grad_norm": 0.796875, "learning_rate": 1.8584420193674963e-05, "loss": 0.1398, "step": 35206 }, { "epoch": 25.792673992673993, "grad_norm": 0.69921875, "learning_rate": 1.858180120831084e-05, "loss": 0.1055, "step": 35207 }, { "epoch": 25.793406593406594, "grad_norm": 0.7890625, "learning_rate": 1.8579182375877275e-05, "loss": 0.1295, "step": 35208 }, { "epoch": 25.794139194139195, "grad_norm": 1.03125, "learning_rate": 1.8576563696386816e-05, "loss": 0.1162, "step": 35209 }, { "epoch": 25.794871794871796, "grad_norm": 1.1328125, "learning_rate": 1.857394516985203e-05, "loss": 0.1129, "step": 35210 }, { "epoch": 25.795604395604396, "grad_norm": 1.46875, "learning_rate": 1.8571326796285463e-05, "loss": 0.113, "step": 35211 }, { "epoch": 25.796336996336997, "grad_norm": 1.03125, "learning_rate": 1.8568708575699637e-05, "loss": 0.1078, "step": 35212 }, { "epoch": 25.797069597069598, "grad_norm": 1.1015625, "learning_rate": 1.8566090508107124e-05, "loss": 0.1409, "step": 35213 }, { "epoch": 25.7978021978022, "grad_norm": 1.453125, "learning_rate": 1.856347259352044e-05, "loss": 0.1035, "step": 35214 }, { "epoch": 25.7985347985348, "grad_norm": 1.4296875, "learning_rate": 1.8560854831952144e-05, "loss": 0.1057, "step": 35215 }, { "epoch": 25.7992673992674, "grad_norm": 0.640625, "learning_rate": 1.8558237223414775e-05, "loss": 0.1135, "step": 35216 }, { "epoch": 25.8, "grad_norm": 1.2109375, "learning_rate": 1.8555619767920884e-05, "loss": 0.1027, "step": 35217 }, { "epoch": 25.8007326007326, "grad_norm": 0.8671875, "learning_rate": 1.8553002465483015e-05, "loss": 0.125, "step": 35218 }, { "epoch": 25.801465201465202, "grad_norm": 1.09375, "learning_rate": 1.8550385316113682e-05, "loss": 0.0826, "step": 35219 }, { "epoch": 25.802197802197803, "grad_norm": 0.765625, "learning_rate": 1.8547768319825446e-05, "loss": 0.1054, "step": 35220 }, { "epoch": 25.802930402930404, "grad_norm": 0.6171875, "learning_rate": 1.8545151476630836e-05, "loss": 0.0993, "step": 35221 }, { "epoch": 25.803663003663004, "grad_norm": 0.7734375, "learning_rate": 1.854253478654242e-05, "loss": 0.0905, "step": 35222 }, { "epoch": 25.804395604395605, "grad_norm": 0.84375, "learning_rate": 1.8539918249572682e-05, "loss": 0.1022, "step": 35223 }, { "epoch": 25.805128205128206, "grad_norm": 1.1875, "learning_rate": 1.8537301865734217e-05, "loss": 0.1427, "step": 35224 }, { "epoch": 25.805860805860807, "grad_norm": 1.1328125, "learning_rate": 1.8534685635039482e-05, "loss": 0.1406, "step": 35225 }, { "epoch": 25.806593406593407, "grad_norm": 1.0234375, "learning_rate": 1.8532069557501108e-05, "loss": 0.1056, "step": 35226 }, { "epoch": 25.807326007326008, "grad_norm": 1.390625, "learning_rate": 1.8529453633131562e-05, "loss": 0.1022, "step": 35227 }, { "epoch": 25.80805860805861, "grad_norm": 1.125, "learning_rate": 1.8526837861943402e-05, "loss": 0.1013, "step": 35228 }, { "epoch": 25.80879120879121, "grad_norm": 0.90234375, "learning_rate": 1.852422224394917e-05, "loss": 0.1303, "step": 35229 }, { "epoch": 25.80952380952381, "grad_norm": 1.5, "learning_rate": 1.8521606779161375e-05, "loss": 0.1199, "step": 35230 }, { "epoch": 25.81025641025641, "grad_norm": 1.5078125, "learning_rate": 1.8518991467592557e-05, "loss": 0.1283, "step": 35231 }, { "epoch": 25.81098901098901, "grad_norm": 1.1015625, "learning_rate": 1.851637630925525e-05, "loss": 0.117, "step": 35232 }, { "epoch": 25.811721611721612, "grad_norm": 1.734375, "learning_rate": 1.8513761304162002e-05, "loss": 0.1144, "step": 35233 }, { "epoch": 25.812454212454213, "grad_norm": 0.765625, "learning_rate": 1.85111464523253e-05, "loss": 0.1011, "step": 35234 }, { "epoch": 25.813186813186814, "grad_norm": 0.83984375, "learning_rate": 1.85085317537577e-05, "loss": 0.1239, "step": 35235 }, { "epoch": 25.813919413919415, "grad_norm": 0.93359375, "learning_rate": 1.8505917208471734e-05, "loss": 0.1198, "step": 35236 }, { "epoch": 25.814652014652015, "grad_norm": 1.25, "learning_rate": 1.8503302816479932e-05, "loss": 0.1101, "step": 35237 }, { "epoch": 25.815384615384616, "grad_norm": 0.96484375, "learning_rate": 1.8500688577794795e-05, "loss": 0.0929, "step": 35238 }, { "epoch": 25.816117216117217, "grad_norm": 1.234375, "learning_rate": 1.8498074492428876e-05, "loss": 0.1034, "step": 35239 }, { "epoch": 25.816849816849818, "grad_norm": 0.88671875, "learning_rate": 1.8495460560394677e-05, "loss": 0.1261, "step": 35240 }, { "epoch": 25.81758241758242, "grad_norm": 1.078125, "learning_rate": 1.849284678170473e-05, "loss": 0.0919, "step": 35241 }, { "epoch": 25.81831501831502, "grad_norm": 1.421875, "learning_rate": 1.8490233156371557e-05, "loss": 0.1151, "step": 35242 }, { "epoch": 25.81904761904762, "grad_norm": 0.8828125, "learning_rate": 1.848761968440769e-05, "loss": 0.1349, "step": 35243 }, { "epoch": 25.81978021978022, "grad_norm": 0.89453125, "learning_rate": 1.8485006365825662e-05, "loss": 0.1643, "step": 35244 }, { "epoch": 25.82051282051282, "grad_norm": 0.6953125, "learning_rate": 1.8482393200637967e-05, "loss": 0.1189, "step": 35245 }, { "epoch": 25.821245421245422, "grad_norm": 1.1171875, "learning_rate": 1.8479780188857133e-05, "loss": 0.122, "step": 35246 }, { "epoch": 25.821978021978023, "grad_norm": 1.0625, "learning_rate": 1.8477167330495677e-05, "loss": 0.1104, "step": 35247 }, { "epoch": 25.822710622710623, "grad_norm": 1.203125, "learning_rate": 1.847455462556615e-05, "loss": 0.1589, "step": 35248 }, { "epoch": 25.823443223443224, "grad_norm": 1.453125, "learning_rate": 1.8471942074081024e-05, "loss": 0.156, "step": 35249 }, { "epoch": 25.824175824175825, "grad_norm": 0.94140625, "learning_rate": 1.8469329676052854e-05, "loss": 0.1011, "step": 35250 }, { "epoch": 25.824908424908426, "grad_norm": 1.1875, "learning_rate": 1.846671743149411e-05, "loss": 0.1364, "step": 35251 }, { "epoch": 25.825641025641026, "grad_norm": 0.828125, "learning_rate": 1.8464105340417373e-05, "loss": 0.133, "step": 35252 }, { "epoch": 25.826373626373627, "grad_norm": 1.2578125, "learning_rate": 1.8461493402835104e-05, "loss": 0.1523, "step": 35253 }, { "epoch": 25.827106227106228, "grad_norm": 1.0859375, "learning_rate": 1.8458881618759836e-05, "loss": 0.1026, "step": 35254 }, { "epoch": 25.82783882783883, "grad_norm": 1.6328125, "learning_rate": 1.8456269988204107e-05, "loss": 0.1313, "step": 35255 }, { "epoch": 25.82857142857143, "grad_norm": 1.15625, "learning_rate": 1.845365851118039e-05, "loss": 0.1389, "step": 35256 }, { "epoch": 25.82930402930403, "grad_norm": 1.578125, "learning_rate": 1.8451047187701205e-05, "loss": 0.1766, "step": 35257 }, { "epoch": 25.83003663003663, "grad_norm": 0.91796875, "learning_rate": 1.844843601777909e-05, "loss": 0.1076, "step": 35258 }, { "epoch": 25.83076923076923, "grad_norm": 0.76953125, "learning_rate": 1.8445825001426546e-05, "loss": 0.1273, "step": 35259 }, { "epoch": 25.831501831501832, "grad_norm": 0.8984375, "learning_rate": 1.8443214138656063e-05, "loss": 0.1179, "step": 35260 }, { "epoch": 25.832234432234433, "grad_norm": 1.25, "learning_rate": 1.844060342948018e-05, "loss": 0.1591, "step": 35261 }, { "epoch": 25.832967032967034, "grad_norm": 1.1328125, "learning_rate": 1.8437992873911363e-05, "loss": 0.1047, "step": 35262 }, { "epoch": 25.833699633699634, "grad_norm": 1.03125, "learning_rate": 1.843538247196217e-05, "loss": 0.1264, "step": 35263 }, { "epoch": 25.834432234432235, "grad_norm": 1.203125, "learning_rate": 1.8432772223645078e-05, "loss": 0.1169, "step": 35264 }, { "epoch": 25.835164835164836, "grad_norm": 1.0390625, "learning_rate": 1.8430162128972612e-05, "loss": 0.0906, "step": 35265 }, { "epoch": 25.835897435897436, "grad_norm": 1.375, "learning_rate": 1.8427552187957253e-05, "loss": 0.1305, "step": 35266 }, { "epoch": 25.836630036630037, "grad_norm": 0.98828125, "learning_rate": 1.842494240061152e-05, "loss": 0.1208, "step": 35267 }, { "epoch": 25.837362637362638, "grad_norm": 1.7265625, "learning_rate": 1.842233276694792e-05, "loss": 0.1233, "step": 35268 }, { "epoch": 25.83809523809524, "grad_norm": 1.0625, "learning_rate": 1.8419723286978956e-05, "loss": 0.1332, "step": 35269 }, { "epoch": 25.83882783882784, "grad_norm": 0.82421875, "learning_rate": 1.8417113960717138e-05, "loss": 0.1077, "step": 35270 }, { "epoch": 25.83956043956044, "grad_norm": 1.0625, "learning_rate": 1.841450478817495e-05, "loss": 0.1414, "step": 35271 }, { "epoch": 25.84029304029304, "grad_norm": 0.57421875, "learning_rate": 1.84118957693649e-05, "loss": 0.1234, "step": 35272 }, { "epoch": 25.84102564102564, "grad_norm": 1.0078125, "learning_rate": 1.8409286904299488e-05, "loss": 0.1154, "step": 35273 }, { "epoch": 25.841758241758242, "grad_norm": 0.7734375, "learning_rate": 1.8406678192991234e-05, "loss": 0.1022, "step": 35274 }, { "epoch": 25.842490842490843, "grad_norm": 1.125, "learning_rate": 1.840406963545261e-05, "loss": 0.127, "step": 35275 }, { "epoch": 25.843223443223444, "grad_norm": 1.0625, "learning_rate": 1.8401461231696135e-05, "loss": 0.125, "step": 35276 }, { "epoch": 25.843956043956045, "grad_norm": 0.8125, "learning_rate": 1.8398852981734265e-05, "loss": 0.1336, "step": 35277 }, { "epoch": 25.844688644688645, "grad_norm": 1.1953125, "learning_rate": 1.8396244885579562e-05, "loss": 0.1526, "step": 35278 }, { "epoch": 25.845421245421246, "grad_norm": 1.296875, "learning_rate": 1.8393636943244473e-05, "loss": 0.1394, "step": 35279 }, { "epoch": 25.846153846153847, "grad_norm": 1.15625, "learning_rate": 1.8391029154741508e-05, "loss": 0.1382, "step": 35280 }, { "epoch": 25.846886446886447, "grad_norm": 0.96875, "learning_rate": 1.838842152008318e-05, "loss": 0.135, "step": 35281 }, { "epoch": 25.847619047619048, "grad_norm": 0.9609375, "learning_rate": 1.8385814039281947e-05, "loss": 0.1487, "step": 35282 }, { "epoch": 25.84835164835165, "grad_norm": 1.7265625, "learning_rate": 1.8383206712350325e-05, "loss": 0.1184, "step": 35283 }, { "epoch": 25.84908424908425, "grad_norm": 0.97265625, "learning_rate": 1.8380599539300804e-05, "loss": 0.1445, "step": 35284 }, { "epoch": 25.84981684981685, "grad_norm": 1.8125, "learning_rate": 1.8377992520145885e-05, "loss": 0.1136, "step": 35285 }, { "epoch": 25.85054945054945, "grad_norm": 1.8984375, "learning_rate": 1.837538565489804e-05, "loss": 0.1491, "step": 35286 }, { "epoch": 25.851282051282052, "grad_norm": 1.28125, "learning_rate": 1.837277894356978e-05, "loss": 0.1134, "step": 35287 }, { "epoch": 25.852014652014653, "grad_norm": 1.109375, "learning_rate": 1.837017238617355e-05, "loss": 0.1119, "step": 35288 }, { "epoch": 25.852747252747253, "grad_norm": 1.6328125, "learning_rate": 1.8367565982721904e-05, "loss": 0.1505, "step": 35289 }, { "epoch": 25.853479853479854, "grad_norm": 1.2109375, "learning_rate": 1.8364959733227283e-05, "loss": 0.1344, "step": 35290 }, { "epoch": 25.854212454212455, "grad_norm": 1.6015625, "learning_rate": 1.8362353637702202e-05, "loss": 0.1099, "step": 35291 }, { "epoch": 25.854945054945055, "grad_norm": 0.80078125, "learning_rate": 1.8359747696159125e-05, "loss": 0.1292, "step": 35292 }, { "epoch": 25.855677655677656, "grad_norm": 0.9140625, "learning_rate": 1.8357141908610545e-05, "loss": 0.1137, "step": 35293 }, { "epoch": 25.856410256410257, "grad_norm": 1.0, "learning_rate": 1.835453627506895e-05, "loss": 0.1183, "step": 35294 }, { "epoch": 25.857142857142858, "grad_norm": 1.0859375, "learning_rate": 1.835193079554682e-05, "loss": 0.1205, "step": 35295 }, { "epoch": 25.85787545787546, "grad_norm": 1.6796875, "learning_rate": 1.834932547005666e-05, "loss": 0.1544, "step": 35296 }, { "epoch": 25.85860805860806, "grad_norm": 1.3125, "learning_rate": 1.8346720298610925e-05, "loss": 0.099, "step": 35297 }, { "epoch": 25.85934065934066, "grad_norm": 1.21875, "learning_rate": 1.834411528122211e-05, "loss": 0.1404, "step": 35298 }, { "epoch": 25.86007326007326, "grad_norm": 0.91796875, "learning_rate": 1.834151041790269e-05, "loss": 0.093, "step": 35299 }, { "epoch": 25.86080586080586, "grad_norm": 0.6875, "learning_rate": 1.8338905708665167e-05, "loss": 0.1066, "step": 35300 }, { "epoch": 25.861538461538462, "grad_norm": 0.9453125, "learning_rate": 1.8336301153521986e-05, "loss": 0.1458, "step": 35301 }, { "epoch": 25.862271062271063, "grad_norm": 1.28125, "learning_rate": 1.8333696752485665e-05, "loss": 0.1315, "step": 35302 }, { "epoch": 25.863003663003663, "grad_norm": 1.1484375, "learning_rate": 1.833109250556863e-05, "loss": 0.1037, "step": 35303 }, { "epoch": 25.863736263736264, "grad_norm": 0.8125, "learning_rate": 1.832848841278342e-05, "loss": 0.1312, "step": 35304 }, { "epoch": 25.864468864468865, "grad_norm": 0.71875, "learning_rate": 1.832588447414247e-05, "loss": 0.1003, "step": 35305 }, { "epoch": 25.865201465201466, "grad_norm": 1.5390625, "learning_rate": 1.8323280689658274e-05, "loss": 0.1101, "step": 35306 }, { "epoch": 25.865934065934066, "grad_norm": 0.921875, "learning_rate": 1.8320677059343314e-05, "loss": 0.1299, "step": 35307 }, { "epoch": 25.866666666666667, "grad_norm": 1.21875, "learning_rate": 1.8318073583210034e-05, "loss": 0.1348, "step": 35308 }, { "epoch": 25.867399267399268, "grad_norm": 0.9375, "learning_rate": 1.8315470261270933e-05, "loss": 0.1063, "step": 35309 }, { "epoch": 25.86813186813187, "grad_norm": 1.2890625, "learning_rate": 1.8312867093538484e-05, "loss": 0.1145, "step": 35310 }, { "epoch": 25.86886446886447, "grad_norm": 0.98828125, "learning_rate": 1.8310264080025164e-05, "loss": 0.1074, "step": 35311 }, { "epoch": 25.86959706959707, "grad_norm": 1.3125, "learning_rate": 1.8307661220743418e-05, "loss": 0.1381, "step": 35312 }, { "epoch": 25.87032967032967, "grad_norm": 1.0546875, "learning_rate": 1.8305058515705758e-05, "loss": 0.1123, "step": 35313 }, { "epoch": 25.87106227106227, "grad_norm": 0.97265625, "learning_rate": 1.8302455964924595e-05, "loss": 0.1149, "step": 35314 }, { "epoch": 25.871794871794872, "grad_norm": 0.73828125, "learning_rate": 1.8299853568412466e-05, "loss": 0.0895, "step": 35315 }, { "epoch": 25.872527472527473, "grad_norm": 1.2265625, "learning_rate": 1.8297251326181798e-05, "loss": 0.1093, "step": 35316 }, { "epoch": 25.873260073260074, "grad_norm": 1.484375, "learning_rate": 1.8294649238245086e-05, "loss": 0.1329, "step": 35317 }, { "epoch": 25.873992673992674, "grad_norm": 0.9296875, "learning_rate": 1.829204730461476e-05, "loss": 0.1064, "step": 35318 }, { "epoch": 25.874725274725275, "grad_norm": 0.8125, "learning_rate": 1.8289445525303313e-05, "loss": 0.1211, "step": 35319 }, { "epoch": 25.875457875457876, "grad_norm": 1.1171875, "learning_rate": 1.828684390032321e-05, "loss": 0.1177, "step": 35320 }, { "epoch": 25.876190476190477, "grad_norm": 1.359375, "learning_rate": 1.8284242429686907e-05, "loss": 0.1317, "step": 35321 }, { "epoch": 25.876923076923077, "grad_norm": 0.98828125, "learning_rate": 1.8281641113406893e-05, "loss": 0.1155, "step": 35322 }, { "epoch": 25.877655677655678, "grad_norm": 0.6171875, "learning_rate": 1.8279039951495595e-05, "loss": 0.1093, "step": 35323 }, { "epoch": 25.87838827838828, "grad_norm": 0.98046875, "learning_rate": 1.8276438943965494e-05, "loss": 0.1449, "step": 35324 }, { "epoch": 25.87912087912088, "grad_norm": 1.0234375, "learning_rate": 1.8273838090829056e-05, "loss": 0.1256, "step": 35325 }, { "epoch": 25.87985347985348, "grad_norm": 1.0546875, "learning_rate": 1.827123739209875e-05, "loss": 0.1048, "step": 35326 }, { "epoch": 25.88058608058608, "grad_norm": 0.61328125, "learning_rate": 1.8268636847787015e-05, "loss": 0.0951, "step": 35327 }, { "epoch": 25.88131868131868, "grad_norm": 0.79296875, "learning_rate": 1.8266036457906336e-05, "loss": 0.1316, "step": 35328 }, { "epoch": 25.882051282051282, "grad_norm": 1.046875, "learning_rate": 1.8263436222469125e-05, "loss": 0.117, "step": 35329 }, { "epoch": 25.882783882783883, "grad_norm": 0.859375, "learning_rate": 1.8260836141487904e-05, "loss": 0.1287, "step": 35330 }, { "epoch": 25.883516483516484, "grad_norm": 1.0078125, "learning_rate": 1.8258236214975084e-05, "loss": 0.1059, "step": 35331 }, { "epoch": 25.884249084249085, "grad_norm": 0.84375, "learning_rate": 1.825563644294314e-05, "loss": 0.1383, "step": 35332 }, { "epoch": 25.884981684981685, "grad_norm": 1.234375, "learning_rate": 1.825303682540455e-05, "loss": 0.102, "step": 35333 }, { "epoch": 25.885714285714286, "grad_norm": 1.53125, "learning_rate": 1.8250437362371718e-05, "loss": 0.1092, "step": 35334 }, { "epoch": 25.886446886446887, "grad_norm": 1.2265625, "learning_rate": 1.8247838053857132e-05, "loss": 0.1423, "step": 35335 }, { "epoch": 25.887179487179488, "grad_norm": 1.3359375, "learning_rate": 1.8245238899873244e-05, "loss": 0.1187, "step": 35336 }, { "epoch": 25.88791208791209, "grad_norm": 1.21875, "learning_rate": 1.8242639900432527e-05, "loss": 0.1145, "step": 35337 }, { "epoch": 25.88864468864469, "grad_norm": 1.5390625, "learning_rate": 1.8240041055547384e-05, "loss": 0.1304, "step": 35338 }, { "epoch": 25.88937728937729, "grad_norm": 0.8671875, "learning_rate": 1.8237442365230318e-05, "loss": 0.1089, "step": 35339 }, { "epoch": 25.89010989010989, "grad_norm": 0.98046875, "learning_rate": 1.8234843829493722e-05, "loss": 0.1006, "step": 35340 }, { "epoch": 25.89084249084249, "grad_norm": 0.65234375, "learning_rate": 1.8232245448350112e-05, "loss": 0.0953, "step": 35341 }, { "epoch": 25.891575091575092, "grad_norm": 0.73046875, "learning_rate": 1.8229647221811892e-05, "loss": 0.1202, "step": 35342 }, { "epoch": 25.892307692307693, "grad_norm": 1.2265625, "learning_rate": 1.8227049149891524e-05, "loss": 0.1072, "step": 35343 }, { "epoch": 25.893040293040293, "grad_norm": 0.984375, "learning_rate": 1.822445123260148e-05, "loss": 0.1046, "step": 35344 }, { "epoch": 25.893772893772894, "grad_norm": 1.1640625, "learning_rate": 1.822185346995416e-05, "loss": 0.115, "step": 35345 }, { "epoch": 25.894505494505495, "grad_norm": 0.75, "learning_rate": 1.8219255861962042e-05, "loss": 0.1198, "step": 35346 }, { "epoch": 25.895238095238096, "grad_norm": 1.5546875, "learning_rate": 1.8216658408637567e-05, "loss": 0.1345, "step": 35347 }, { "epoch": 25.895970695970696, "grad_norm": 0.734375, "learning_rate": 1.8214061109993194e-05, "loss": 0.1253, "step": 35348 }, { "epoch": 25.896703296703297, "grad_norm": 0.498046875, "learning_rate": 1.8211463966041336e-05, "loss": 0.0946, "step": 35349 }, { "epoch": 25.897435897435898, "grad_norm": 0.7109375, "learning_rate": 1.8208866976794456e-05, "loss": 0.1219, "step": 35350 }, { "epoch": 25.8981684981685, "grad_norm": 1.09375, "learning_rate": 1.8206270142264998e-05, "loss": 0.094, "step": 35351 }, { "epoch": 25.8989010989011, "grad_norm": 1.0234375, "learning_rate": 1.820367346246541e-05, "loss": 0.1397, "step": 35352 }, { "epoch": 25.8996336996337, "grad_norm": 0.77734375, "learning_rate": 1.8201076937408115e-05, "loss": 0.0958, "step": 35353 }, { "epoch": 25.9003663003663, "grad_norm": 1.0390625, "learning_rate": 1.819848056710557e-05, "loss": 0.1033, "step": 35354 }, { "epoch": 25.9010989010989, "grad_norm": 0.9609375, "learning_rate": 1.8195884351570182e-05, "loss": 0.1177, "step": 35355 }, { "epoch": 25.901831501831502, "grad_norm": 0.77734375, "learning_rate": 1.8193288290814452e-05, "loss": 0.09, "step": 35356 }, { "epoch": 25.902564102564103, "grad_norm": 0.765625, "learning_rate": 1.8190692384850768e-05, "loss": 0.1031, "step": 35357 }, { "epoch": 25.903296703296704, "grad_norm": 1.09375, "learning_rate": 1.8188096633691577e-05, "loss": 0.1434, "step": 35358 }, { "epoch": 25.904029304029304, "grad_norm": 0.9609375, "learning_rate": 1.8185501037349342e-05, "loss": 0.1125, "step": 35359 }, { "epoch": 25.904761904761905, "grad_norm": 1.1015625, "learning_rate": 1.818290559583645e-05, "loss": 0.1167, "step": 35360 }, { "epoch": 25.905494505494506, "grad_norm": 1.4375, "learning_rate": 1.818031030916539e-05, "loss": 0.1431, "step": 35361 }, { "epoch": 25.906227106227107, "grad_norm": 1.1640625, "learning_rate": 1.8177715177348565e-05, "loss": 0.1068, "step": 35362 }, { "epoch": 25.906959706959707, "grad_norm": 1.1015625, "learning_rate": 1.817512020039843e-05, "loss": 0.1148, "step": 35363 }, { "epoch": 25.907692307692308, "grad_norm": 0.625, "learning_rate": 1.817252537832739e-05, "loss": 0.1028, "step": 35364 }, { "epoch": 25.90842490842491, "grad_norm": 0.90234375, "learning_rate": 1.816993071114791e-05, "loss": 0.111, "step": 35365 }, { "epoch": 25.90915750915751, "grad_norm": 0.91796875, "learning_rate": 1.8167336198872375e-05, "loss": 0.1183, "step": 35366 }, { "epoch": 25.90989010989011, "grad_norm": 0.8515625, "learning_rate": 1.8164741841513275e-05, "loss": 0.1699, "step": 35367 }, { "epoch": 25.91062271062271, "grad_norm": 0.93359375, "learning_rate": 1.8162147639082996e-05, "loss": 0.1259, "step": 35368 }, { "epoch": 25.91135531135531, "grad_norm": 1.4453125, "learning_rate": 1.8159553591593987e-05, "loss": 0.1547, "step": 35369 }, { "epoch": 25.912087912087912, "grad_norm": 1.1953125, "learning_rate": 1.815695969905869e-05, "loss": 0.0878, "step": 35370 }, { "epoch": 25.912820512820513, "grad_norm": 0.80078125, "learning_rate": 1.8154365961489495e-05, "loss": 0.1196, "step": 35371 }, { "epoch": 25.913553113553114, "grad_norm": 0.94921875, "learning_rate": 1.8151772378898855e-05, "loss": 0.1178, "step": 35372 }, { "epoch": 25.914285714285715, "grad_norm": 2.21875, "learning_rate": 1.814917895129919e-05, "loss": 0.13, "step": 35373 }, { "epoch": 25.915018315018315, "grad_norm": 0.97265625, "learning_rate": 1.8146585678702946e-05, "loss": 0.1342, "step": 35374 }, { "epoch": 25.915750915750916, "grad_norm": 0.94140625, "learning_rate": 1.8143992561122518e-05, "loss": 0.1018, "step": 35375 }, { "epoch": 25.916483516483517, "grad_norm": 1.7109375, "learning_rate": 1.814139959857034e-05, "loss": 0.1155, "step": 35376 }, { "epoch": 25.917216117216118, "grad_norm": 0.97265625, "learning_rate": 1.8138806791058838e-05, "loss": 0.1152, "step": 35377 }, { "epoch": 25.91794871794872, "grad_norm": 0.70703125, "learning_rate": 1.8136214138600458e-05, "loss": 0.1154, "step": 35378 }, { "epoch": 25.91868131868132, "grad_norm": 1.0703125, "learning_rate": 1.8133621641207576e-05, "loss": 0.1585, "step": 35379 }, { "epoch": 25.91941391941392, "grad_norm": 0.52734375, "learning_rate": 1.8131029298892657e-05, "loss": 0.1201, "step": 35380 }, { "epoch": 25.92014652014652, "grad_norm": 0.75, "learning_rate": 1.812843711166807e-05, "loss": 0.1227, "step": 35381 }, { "epoch": 25.92087912087912, "grad_norm": 0.8515625, "learning_rate": 1.8125845079546306e-05, "loss": 0.0969, "step": 35382 }, { "epoch": 25.921611721611722, "grad_norm": 0.91796875, "learning_rate": 1.8123253202539724e-05, "loss": 0.1263, "step": 35383 }, { "epoch": 25.922344322344323, "grad_norm": 1.1171875, "learning_rate": 1.8120661480660766e-05, "loss": 0.1205, "step": 35384 }, { "epoch": 25.923076923076923, "grad_norm": 1.390625, "learning_rate": 1.8118069913921866e-05, "loss": 0.1258, "step": 35385 }, { "epoch": 25.923809523809524, "grad_norm": 1.3046875, "learning_rate": 1.811547850233539e-05, "loss": 0.1633, "step": 35386 }, { "epoch": 25.924542124542125, "grad_norm": 0.97265625, "learning_rate": 1.811288724591382e-05, "loss": 0.1647, "step": 35387 }, { "epoch": 25.925274725274726, "grad_norm": 0.95703125, "learning_rate": 1.811029614466952e-05, "loss": 0.1142, "step": 35388 }, { "epoch": 25.926007326007326, "grad_norm": 0.87890625, "learning_rate": 1.8107705198614945e-05, "loss": 0.1354, "step": 35389 }, { "epoch": 25.926739926739927, "grad_norm": 1.3046875, "learning_rate": 1.810511440776247e-05, "loss": 0.1415, "step": 35390 }, { "epoch": 25.927472527472528, "grad_norm": 0.87890625, "learning_rate": 1.810252377212454e-05, "loss": 0.1158, "step": 35391 }, { "epoch": 25.92820512820513, "grad_norm": 1.328125, "learning_rate": 1.8099933291713522e-05, "loss": 0.1065, "step": 35392 }, { "epoch": 25.92893772893773, "grad_norm": 0.578125, "learning_rate": 1.80973429665419e-05, "loss": 0.1196, "step": 35393 }, { "epoch": 25.92967032967033, "grad_norm": 0.66796875, "learning_rate": 1.8094752796622016e-05, "loss": 0.1227, "step": 35394 }, { "epoch": 25.93040293040293, "grad_norm": 1.3125, "learning_rate": 1.809216278196632e-05, "loss": 0.1389, "step": 35395 }, { "epoch": 25.93113553113553, "grad_norm": 1.0703125, "learning_rate": 1.8089572922587223e-05, "loss": 0.1334, "step": 35396 }, { "epoch": 25.931868131868132, "grad_norm": 2.0, "learning_rate": 1.8086983218497104e-05, "loss": 0.1314, "step": 35397 }, { "epoch": 25.932600732600733, "grad_norm": 0.734375, "learning_rate": 1.8084393669708392e-05, "loss": 0.0982, "step": 35398 }, { "epoch": 25.933333333333334, "grad_norm": 0.6484375, "learning_rate": 1.8081804276233488e-05, "loss": 0.1106, "step": 35399 }, { "epoch": 25.934065934065934, "grad_norm": 1.3671875, "learning_rate": 1.807921503808482e-05, "loss": 0.1235, "step": 35400 }, { "epoch": 25.934798534798535, "grad_norm": 1.0390625, "learning_rate": 1.8076625955274763e-05, "loss": 0.1147, "step": 35401 }, { "epoch": 25.935531135531136, "grad_norm": 1.0625, "learning_rate": 1.8074037027815733e-05, "loss": 0.1408, "step": 35402 }, { "epoch": 25.936263736263736, "grad_norm": 0.79296875, "learning_rate": 1.8071448255720134e-05, "loss": 0.1222, "step": 35403 }, { "epoch": 25.936996336996337, "grad_norm": 1.109375, "learning_rate": 1.806885963900039e-05, "loss": 0.1121, "step": 35404 }, { "epoch": 25.937728937728938, "grad_norm": 1.15625, "learning_rate": 1.8066271177668868e-05, "loss": 0.1139, "step": 35405 }, { "epoch": 25.93846153846154, "grad_norm": 1.0859375, "learning_rate": 1.8063682871738007e-05, "loss": 0.14, "step": 35406 }, { "epoch": 25.93919413919414, "grad_norm": 1.1796875, "learning_rate": 1.806109472122015e-05, "loss": 0.1189, "step": 35407 }, { "epoch": 25.93992673992674, "grad_norm": 1.171875, "learning_rate": 1.805850672612778e-05, "loss": 0.105, "step": 35408 }, { "epoch": 25.94065934065934, "grad_norm": 0.6953125, "learning_rate": 1.805591888647323e-05, "loss": 0.1153, "step": 35409 }, { "epoch": 25.94139194139194, "grad_norm": 0.796875, "learning_rate": 1.8053331202268927e-05, "loss": 0.1274, "step": 35410 }, { "epoch": 25.942124542124542, "grad_norm": 0.7578125, "learning_rate": 1.8050743673527285e-05, "loss": 0.1224, "step": 35411 }, { "epoch": 25.942857142857143, "grad_norm": 0.8984375, "learning_rate": 1.8048156300260644e-05, "loss": 0.1124, "step": 35412 }, { "epoch": 25.943589743589744, "grad_norm": 0.92578125, "learning_rate": 1.8045569082481474e-05, "loss": 0.1089, "step": 35413 }, { "epoch": 25.944322344322345, "grad_norm": 0.99609375, "learning_rate": 1.8042982020202116e-05, "loss": 0.1244, "step": 35414 }, { "epoch": 25.945054945054945, "grad_norm": 0.94140625, "learning_rate": 1.804039511343501e-05, "loss": 0.1294, "step": 35415 }, { "epoch": 25.945787545787546, "grad_norm": 0.796875, "learning_rate": 1.803780836219251e-05, "loss": 0.1266, "step": 35416 }, { "epoch": 25.946520146520147, "grad_norm": 1.3203125, "learning_rate": 1.8035221766487036e-05, "loss": 0.1404, "step": 35417 }, { "epoch": 25.947252747252747, "grad_norm": 0.74609375, "learning_rate": 1.8032635326330942e-05, "loss": 0.0997, "step": 35418 }, { "epoch": 25.947985347985348, "grad_norm": 0.8984375, "learning_rate": 1.803004904173669e-05, "loss": 0.1232, "step": 35419 }, { "epoch": 25.94871794871795, "grad_norm": 1.6015625, "learning_rate": 1.8027462912716604e-05, "loss": 0.1451, "step": 35420 }, { "epoch": 25.94945054945055, "grad_norm": 1.1484375, "learning_rate": 1.8024876939283114e-05, "loss": 0.1084, "step": 35421 }, { "epoch": 25.95018315018315, "grad_norm": 0.99609375, "learning_rate": 1.8022291121448608e-05, "loss": 0.1137, "step": 35422 }, { "epoch": 25.95091575091575, "grad_norm": 1.3203125, "learning_rate": 1.8019705459225455e-05, "loss": 0.149, "step": 35423 }, { "epoch": 25.951648351648352, "grad_norm": 1.09375, "learning_rate": 1.801711995262606e-05, "loss": 0.1048, "step": 35424 }, { "epoch": 25.952380952380953, "grad_norm": 0.73828125, "learning_rate": 1.80145346016628e-05, "loss": 0.1432, "step": 35425 }, { "epoch": 25.953113553113553, "grad_norm": 0.58984375, "learning_rate": 1.8011949406348085e-05, "loss": 0.1157, "step": 35426 }, { "epoch": 25.953846153846154, "grad_norm": 1.125, "learning_rate": 1.8009364366694274e-05, "loss": 0.1275, "step": 35427 }, { "epoch": 25.954578754578755, "grad_norm": 1.578125, "learning_rate": 1.800677948271376e-05, "loss": 0.1016, "step": 35428 }, { "epoch": 25.955311355311355, "grad_norm": 0.83984375, "learning_rate": 1.8004194754418934e-05, "loss": 0.1189, "step": 35429 }, { "epoch": 25.956043956043956, "grad_norm": 0.765625, "learning_rate": 1.80016101818222e-05, "loss": 0.1031, "step": 35430 }, { "epoch": 25.956776556776557, "grad_norm": 0.984375, "learning_rate": 1.7999025764935895e-05, "loss": 0.118, "step": 35431 }, { "epoch": 25.957509157509158, "grad_norm": 1.546875, "learning_rate": 1.7996441503772447e-05, "loss": 0.149, "step": 35432 }, { "epoch": 25.95824175824176, "grad_norm": 1.203125, "learning_rate": 1.7993857398344186e-05, "loss": 0.1145, "step": 35433 }, { "epoch": 25.95897435897436, "grad_norm": 0.54296875, "learning_rate": 1.7991273448663555e-05, "loss": 0.082, "step": 35434 }, { "epoch": 25.95970695970696, "grad_norm": 1.265625, "learning_rate": 1.7988689654742895e-05, "loss": 0.1443, "step": 35435 }, { "epoch": 25.96043956043956, "grad_norm": 1.203125, "learning_rate": 1.7986106016594585e-05, "loss": 0.1354, "step": 35436 }, { "epoch": 25.96117216117216, "grad_norm": 0.74609375, "learning_rate": 1.798352253423104e-05, "loss": 0.1335, "step": 35437 }, { "epoch": 25.961904761904762, "grad_norm": 0.8203125, "learning_rate": 1.798093920766458e-05, "loss": 0.1002, "step": 35438 }, { "epoch": 25.962637362637363, "grad_norm": 0.96484375, "learning_rate": 1.797835603690764e-05, "loss": 0.1084, "step": 35439 }, { "epoch": 25.963369963369964, "grad_norm": 0.65234375, "learning_rate": 1.7975773021972568e-05, "loss": 0.1091, "step": 35440 }, { "epoch": 25.964102564102564, "grad_norm": 0.92578125, "learning_rate": 1.7973190162871753e-05, "loss": 0.1279, "step": 35441 }, { "epoch": 25.964835164835165, "grad_norm": 1.0078125, "learning_rate": 1.7970607459617546e-05, "loss": 0.1248, "step": 35442 }, { "epoch": 25.965567765567766, "grad_norm": 1.15625, "learning_rate": 1.796802491222236e-05, "loss": 0.1378, "step": 35443 }, { "epoch": 25.966300366300366, "grad_norm": 0.94921875, "learning_rate": 1.7965442520698506e-05, "loss": 0.1385, "step": 35444 }, { "epoch": 25.967032967032967, "grad_norm": 1.2421875, "learning_rate": 1.7962860285058433e-05, "loss": 0.1446, "step": 35445 }, { "epoch": 25.967765567765568, "grad_norm": 1.125, "learning_rate": 1.796027820531446e-05, "loss": 0.1355, "step": 35446 }, { "epoch": 25.96849816849817, "grad_norm": 1.2578125, "learning_rate": 1.7957696281478974e-05, "loss": 0.1039, "step": 35447 }, { "epoch": 25.96923076923077, "grad_norm": 1.9765625, "learning_rate": 1.795511451356437e-05, "loss": 0.1344, "step": 35448 }, { "epoch": 25.96996336996337, "grad_norm": 1.046875, "learning_rate": 1.7952532901582976e-05, "loss": 0.1594, "step": 35449 }, { "epoch": 25.97069597069597, "grad_norm": 0.828125, "learning_rate": 1.7949951445547186e-05, "loss": 0.1122, "step": 35450 }, { "epoch": 25.97142857142857, "grad_norm": 1.03125, "learning_rate": 1.7947370145469355e-05, "loss": 0.1195, "step": 35451 }, { "epoch": 25.972161172161172, "grad_norm": 0.609375, "learning_rate": 1.7944789001361888e-05, "loss": 0.1038, "step": 35452 }, { "epoch": 25.972893772893773, "grad_norm": 0.90234375, "learning_rate": 1.7942208013237096e-05, "loss": 0.1204, "step": 35453 }, { "epoch": 25.973626373626374, "grad_norm": 1.2265625, "learning_rate": 1.793962718110738e-05, "loss": 0.1219, "step": 35454 }, { "epoch": 25.974358974358974, "grad_norm": 0.859375, "learning_rate": 1.79370465049851e-05, "loss": 0.1568, "step": 35455 }, { "epoch": 25.975091575091575, "grad_norm": 0.9921875, "learning_rate": 1.793446598488263e-05, "loss": 0.1278, "step": 35456 }, { "epoch": 25.975824175824176, "grad_norm": 0.84375, "learning_rate": 1.7931885620812313e-05, "loss": 0.1266, "step": 35457 }, { "epoch": 25.976556776556777, "grad_norm": 1.0859375, "learning_rate": 1.7929305412786526e-05, "loss": 0.1096, "step": 35458 }, { "epoch": 25.977289377289377, "grad_norm": 1.2734375, "learning_rate": 1.7926725360817622e-05, "loss": 0.1089, "step": 35459 }, { "epoch": 25.978021978021978, "grad_norm": 1.1796875, "learning_rate": 1.7924145464917994e-05, "loss": 0.0857, "step": 35460 }, { "epoch": 25.97875457875458, "grad_norm": 1.078125, "learning_rate": 1.7921565725099958e-05, "loss": 0.1014, "step": 35461 }, { "epoch": 25.97948717948718, "grad_norm": 0.94921875, "learning_rate": 1.7918986141375898e-05, "loss": 0.1213, "step": 35462 }, { "epoch": 25.98021978021978, "grad_norm": 0.98828125, "learning_rate": 1.791640671375818e-05, "loss": 0.1078, "step": 35463 }, { "epoch": 25.98095238095238, "grad_norm": 1.21875, "learning_rate": 1.7913827442259146e-05, "loss": 0.1193, "step": 35464 }, { "epoch": 25.98168498168498, "grad_norm": 1.109375, "learning_rate": 1.791124832689116e-05, "loss": 0.1289, "step": 35465 }, { "epoch": 25.982417582417582, "grad_norm": 0.99609375, "learning_rate": 1.790866936766658e-05, "loss": 0.1167, "step": 35466 }, { "epoch": 25.983150183150183, "grad_norm": 0.984375, "learning_rate": 1.7906090564597784e-05, "loss": 0.1183, "step": 35467 }, { "epoch": 25.983882783882784, "grad_norm": 1.15625, "learning_rate": 1.7903511917697087e-05, "loss": 0.1262, "step": 35468 }, { "epoch": 25.984615384615385, "grad_norm": 1.3359375, "learning_rate": 1.790093342697688e-05, "loss": 0.1294, "step": 35469 }, { "epoch": 25.985347985347985, "grad_norm": 1.109375, "learning_rate": 1.789835509244947e-05, "loss": 0.1193, "step": 35470 }, { "epoch": 25.986080586080586, "grad_norm": 1.1484375, "learning_rate": 1.789577691412728e-05, "loss": 0.1268, "step": 35471 }, { "epoch": 25.986813186813187, "grad_norm": 1.140625, "learning_rate": 1.7893198892022612e-05, "loss": 0.1104, "step": 35472 }, { "epoch": 25.987545787545788, "grad_norm": 1.171875, "learning_rate": 1.7890621026147824e-05, "loss": 0.1195, "step": 35473 }, { "epoch": 25.98827838827839, "grad_norm": 1.15625, "learning_rate": 1.78880433165153e-05, "loss": 0.1387, "step": 35474 }, { "epoch": 25.98901098901099, "grad_norm": 1.0546875, "learning_rate": 1.7885465763137333e-05, "loss": 0.1161, "step": 35475 }, { "epoch": 25.98974358974359, "grad_norm": 1.0078125, "learning_rate": 1.7882888366026333e-05, "loss": 0.1167, "step": 35476 }, { "epoch": 25.99047619047619, "grad_norm": 1.15625, "learning_rate": 1.7880311125194606e-05, "loss": 0.119, "step": 35477 }, { "epoch": 25.99120879120879, "grad_norm": 0.98828125, "learning_rate": 1.7877734040654535e-05, "loss": 0.1197, "step": 35478 }, { "epoch": 25.991941391941392, "grad_norm": 0.63671875, "learning_rate": 1.787515711241844e-05, "loss": 0.1029, "step": 35479 }, { "epoch": 25.992673992673993, "grad_norm": 1.109375, "learning_rate": 1.7872580340498667e-05, "loss": 0.1093, "step": 35480 }, { "epoch": 25.993406593406593, "grad_norm": 1.125, "learning_rate": 1.7870003724907576e-05, "loss": 0.1053, "step": 35481 }, { "epoch": 25.994139194139194, "grad_norm": 0.7890625, "learning_rate": 1.786742726565753e-05, "loss": 0.1066, "step": 35482 }, { "epoch": 25.994871794871795, "grad_norm": 1.0625, "learning_rate": 1.7864850962760835e-05, "loss": 0.1069, "step": 35483 }, { "epoch": 25.995604395604396, "grad_norm": 0.7109375, "learning_rate": 1.7862274816229848e-05, "loss": 0.1115, "step": 35484 }, { "epoch": 25.996336996336996, "grad_norm": 0.875, "learning_rate": 1.7859698826076914e-05, "loss": 0.1142, "step": 35485 }, { "epoch": 25.997069597069597, "grad_norm": 0.78515625, "learning_rate": 1.7857122992314404e-05, "loss": 0.1283, "step": 35486 }, { "epoch": 25.997802197802198, "grad_norm": 1.15625, "learning_rate": 1.7854547314954614e-05, "loss": 0.1368, "step": 35487 }, { "epoch": 25.9985347985348, "grad_norm": 0.6796875, "learning_rate": 1.7851971794009903e-05, "loss": 0.0951, "step": 35488 }, { "epoch": 25.9992673992674, "grad_norm": 0.97265625, "learning_rate": 1.7849396429492624e-05, "loss": 0.1203, "step": 35489 }, { "epoch": 26.0, "grad_norm": 0.80078125, "learning_rate": 1.7846821221415098e-05, "loss": 0.1064, "step": 35490 }, { "epoch": 26.0007326007326, "grad_norm": 1.3984375, "learning_rate": 1.7844246169789662e-05, "loss": 0.1491, "step": 35491 }, { "epoch": 26.0014652014652, "grad_norm": 1.0390625, "learning_rate": 1.784167127462867e-05, "loss": 0.1261, "step": 35492 }, { "epoch": 26.002197802197802, "grad_norm": 0.859375, "learning_rate": 1.7839096535944464e-05, "loss": 0.1282, "step": 35493 }, { "epoch": 26.002930402930403, "grad_norm": 1.1171875, "learning_rate": 1.7836521953749356e-05, "loss": 0.1087, "step": 35494 }, { "epoch": 26.003663003663004, "grad_norm": 0.9140625, "learning_rate": 1.783394752805571e-05, "loss": 0.1145, "step": 35495 }, { "epoch": 26.004395604395604, "grad_norm": 0.9453125, "learning_rate": 1.783137325887581e-05, "loss": 0.1132, "step": 35496 }, { "epoch": 26.005128205128205, "grad_norm": 1.25, "learning_rate": 1.7828799146222056e-05, "loss": 0.1176, "step": 35497 }, { "epoch": 26.005860805860806, "grad_norm": 1.296875, "learning_rate": 1.7826225190106737e-05, "loss": 0.1187, "step": 35498 }, { "epoch": 26.006593406593407, "grad_norm": 1.4375, "learning_rate": 1.78236513905422e-05, "loss": 0.1245, "step": 35499 }, { "epoch": 26.007326007326007, "grad_norm": 0.97265625, "learning_rate": 1.782107774754079e-05, "loss": 0.137, "step": 35500 }, { "epoch": 26.008058608058608, "grad_norm": 0.9921875, "learning_rate": 1.781850426111479e-05, "loss": 0.1361, "step": 35501 }, { "epoch": 26.00879120879121, "grad_norm": 0.98828125, "learning_rate": 1.7815930931276603e-05, "loss": 0.1178, "step": 35502 }, { "epoch": 26.00952380952381, "grad_norm": 0.96484375, "learning_rate": 1.78133577580385e-05, "loss": 0.1291, "step": 35503 }, { "epoch": 26.01025641025641, "grad_norm": 0.58984375, "learning_rate": 1.7810784741412847e-05, "loss": 0.0986, "step": 35504 }, { "epoch": 26.01098901098901, "grad_norm": 0.953125, "learning_rate": 1.7808211881411945e-05, "loss": 0.1414, "step": 35505 }, { "epoch": 26.01172161172161, "grad_norm": 0.95703125, "learning_rate": 1.780563917804812e-05, "loss": 0.1248, "step": 35506 }, { "epoch": 26.012454212454212, "grad_norm": 1.34375, "learning_rate": 1.780306663133372e-05, "loss": 0.1149, "step": 35507 }, { "epoch": 26.013186813186813, "grad_norm": 0.87890625, "learning_rate": 1.7800494241281072e-05, "loss": 0.1304, "step": 35508 }, { "epoch": 26.013919413919414, "grad_norm": 1.3203125, "learning_rate": 1.779792200790248e-05, "loss": 0.1273, "step": 35509 }, { "epoch": 26.014652014652015, "grad_norm": 1.078125, "learning_rate": 1.7795349931210275e-05, "loss": 0.1342, "step": 35510 }, { "epoch": 26.015384615384615, "grad_norm": 1.0703125, "learning_rate": 1.779277801121679e-05, "loss": 0.1096, "step": 35511 }, { "epoch": 26.016117216117216, "grad_norm": 0.70703125, "learning_rate": 1.779020624793435e-05, "loss": 0.0985, "step": 35512 }, { "epoch": 26.016849816849817, "grad_norm": 1.2421875, "learning_rate": 1.7787634641375264e-05, "loss": 0.1012, "step": 35513 }, { "epoch": 26.017582417582418, "grad_norm": 1.0546875, "learning_rate": 1.7785063191551855e-05, "loss": 0.133, "step": 35514 }, { "epoch": 26.01831501831502, "grad_norm": 0.84765625, "learning_rate": 1.778249189847646e-05, "loss": 0.1408, "step": 35515 }, { "epoch": 26.01904761904762, "grad_norm": 0.859375, "learning_rate": 1.777992076216137e-05, "loss": 0.1084, "step": 35516 }, { "epoch": 26.01978021978022, "grad_norm": 0.82421875, "learning_rate": 1.7777349782618928e-05, "loss": 0.1084, "step": 35517 }, { "epoch": 26.02051282051282, "grad_norm": 1.078125, "learning_rate": 1.777477895986144e-05, "loss": 0.0976, "step": 35518 }, { "epoch": 26.02124542124542, "grad_norm": 1.1171875, "learning_rate": 1.7772208293901253e-05, "loss": 0.125, "step": 35519 }, { "epoch": 26.021978021978022, "grad_norm": 1.0390625, "learning_rate": 1.7769637784750633e-05, "loss": 0.1171, "step": 35520 }, { "epoch": 26.022710622710623, "grad_norm": 0.96484375, "learning_rate": 1.7767067432421945e-05, "loss": 0.1083, "step": 35521 }, { "epoch": 26.023443223443223, "grad_norm": 1.078125, "learning_rate": 1.7764497236927446e-05, "loss": 0.1466, "step": 35522 }, { "epoch": 26.024175824175824, "grad_norm": 0.7578125, "learning_rate": 1.7761927198279528e-05, "loss": 0.1125, "step": 35523 }, { "epoch": 26.024908424908425, "grad_norm": 0.57421875, "learning_rate": 1.775935731649044e-05, "loss": 0.0938, "step": 35524 }, { "epoch": 26.025641025641026, "grad_norm": 1.4140625, "learning_rate": 1.775678759157252e-05, "loss": 0.1132, "step": 35525 }, { "epoch": 26.026373626373626, "grad_norm": 0.921875, "learning_rate": 1.7754218023538096e-05, "loss": 0.1026, "step": 35526 }, { "epoch": 26.027106227106227, "grad_norm": 1.046875, "learning_rate": 1.7751648612399433e-05, "loss": 0.1044, "step": 35527 }, { "epoch": 26.027838827838828, "grad_norm": 0.77734375, "learning_rate": 1.7749079358168903e-05, "loss": 0.1159, "step": 35528 }, { "epoch": 26.02857142857143, "grad_norm": 1.3828125, "learning_rate": 1.774651026085877e-05, "loss": 0.1418, "step": 35529 }, { "epoch": 26.02930402930403, "grad_norm": 1.0703125, "learning_rate": 1.7743941320481374e-05, "loss": 0.1475, "step": 35530 }, { "epoch": 26.03003663003663, "grad_norm": 1.4453125, "learning_rate": 1.774137253704899e-05, "loss": 0.1505, "step": 35531 }, { "epoch": 26.03076923076923, "grad_norm": 1.2890625, "learning_rate": 1.773880391057395e-05, "loss": 0.1235, "step": 35532 }, { "epoch": 26.03150183150183, "grad_norm": 1.21875, "learning_rate": 1.7736235441068552e-05, "loss": 0.115, "step": 35533 }, { "epoch": 26.032234432234432, "grad_norm": 1.75, "learning_rate": 1.7733667128545123e-05, "loss": 0.1468, "step": 35534 }, { "epoch": 26.032967032967033, "grad_norm": 1.1015625, "learning_rate": 1.7731098973015933e-05, "loss": 0.1273, "step": 35535 }, { "epoch": 26.033699633699634, "grad_norm": 1.21875, "learning_rate": 1.7728530974493303e-05, "loss": 0.1364, "step": 35536 }, { "epoch": 26.034432234432234, "grad_norm": 1.3671875, "learning_rate": 1.772596313298954e-05, "loss": 0.1385, "step": 35537 }, { "epoch": 26.035164835164835, "grad_norm": 0.88671875, "learning_rate": 1.7723395448516966e-05, "loss": 0.1245, "step": 35538 }, { "epoch": 26.035897435897436, "grad_norm": 0.953125, "learning_rate": 1.7720827921087847e-05, "loss": 0.1095, "step": 35539 }, { "epoch": 26.036630036630036, "grad_norm": 0.9453125, "learning_rate": 1.7718260550714503e-05, "loss": 0.1307, "step": 35540 }, { "epoch": 26.037362637362637, "grad_norm": 0.88671875, "learning_rate": 1.771569333740925e-05, "loss": 0.1141, "step": 35541 }, { "epoch": 26.038095238095238, "grad_norm": 0.86328125, "learning_rate": 1.771312628118436e-05, "loss": 0.1124, "step": 35542 }, { "epoch": 26.03882783882784, "grad_norm": 1.6953125, "learning_rate": 1.7710559382052138e-05, "loss": 0.1181, "step": 35543 }, { "epoch": 26.03956043956044, "grad_norm": 1.546875, "learning_rate": 1.7707992640024894e-05, "loss": 0.1293, "step": 35544 }, { "epoch": 26.04029304029304, "grad_norm": 1.203125, "learning_rate": 1.7705426055114936e-05, "loss": 0.1416, "step": 35545 }, { "epoch": 26.04102564102564, "grad_norm": 0.94140625, "learning_rate": 1.7702859627334533e-05, "loss": 0.1469, "step": 35546 }, { "epoch": 26.04175824175824, "grad_norm": 0.70703125, "learning_rate": 1.7700293356696014e-05, "loss": 0.1009, "step": 35547 }, { "epoch": 26.042490842490842, "grad_norm": 1.046875, "learning_rate": 1.769772724321162e-05, "loss": 0.1234, "step": 35548 }, { "epoch": 26.043223443223443, "grad_norm": 0.73046875, "learning_rate": 1.7695161286893718e-05, "loss": 0.1164, "step": 35549 }, { "epoch": 26.043956043956044, "grad_norm": 1.6953125, "learning_rate": 1.7692595487754555e-05, "loss": 0.1038, "step": 35550 }, { "epoch": 26.044688644688645, "grad_norm": 1.609375, "learning_rate": 1.769002984580643e-05, "loss": 0.1184, "step": 35551 }, { "epoch": 26.045421245421245, "grad_norm": 1.2109375, "learning_rate": 1.7687464361061663e-05, "loss": 0.1428, "step": 35552 }, { "epoch": 26.046153846153846, "grad_norm": 0.6015625, "learning_rate": 1.7684899033532494e-05, "loss": 0.1104, "step": 35553 }, { "epoch": 26.046886446886447, "grad_norm": 0.80078125, "learning_rate": 1.768233386323128e-05, "loss": 0.1051, "step": 35554 }, { "epoch": 26.047619047619047, "grad_norm": 0.80078125, "learning_rate": 1.7679768850170254e-05, "loss": 0.1427, "step": 35555 }, { "epoch": 26.048351648351648, "grad_norm": 1.671875, "learning_rate": 1.7677203994361757e-05, "loss": 0.1021, "step": 35556 }, { "epoch": 26.04908424908425, "grad_norm": 1.0078125, "learning_rate": 1.7674639295818026e-05, "loss": 0.1179, "step": 35557 }, { "epoch": 26.04981684981685, "grad_norm": 0.75, "learning_rate": 1.7672074754551378e-05, "loss": 0.1119, "step": 35558 }, { "epoch": 26.05054945054945, "grad_norm": 0.9453125, "learning_rate": 1.766951037057409e-05, "loss": 0.1185, "step": 35559 }, { "epoch": 26.05128205128205, "grad_norm": 0.71484375, "learning_rate": 1.7666946143898478e-05, "loss": 0.1233, "step": 35560 }, { "epoch": 26.052014652014652, "grad_norm": 1.421875, "learning_rate": 1.7664382074536793e-05, "loss": 0.0933, "step": 35561 }, { "epoch": 26.052747252747253, "grad_norm": 0.984375, "learning_rate": 1.7661818162501323e-05, "loss": 0.1099, "step": 35562 }, { "epoch": 26.053479853479853, "grad_norm": 1.3984375, "learning_rate": 1.7659254407804368e-05, "loss": 0.1123, "step": 35563 }, { "epoch": 26.054212454212454, "grad_norm": 0.60546875, "learning_rate": 1.765669081045822e-05, "loss": 0.1149, "step": 35564 }, { "epoch": 26.054945054945055, "grad_norm": 1.671875, "learning_rate": 1.7654127370475136e-05, "loss": 0.1034, "step": 35565 }, { "epoch": 26.055677655677655, "grad_norm": 0.7421875, "learning_rate": 1.7651564087867407e-05, "loss": 0.1239, "step": 35566 }, { "epoch": 26.056410256410256, "grad_norm": 1.125, "learning_rate": 1.7649000962647334e-05, "loss": 0.1593, "step": 35567 }, { "epoch": 26.057142857142857, "grad_norm": 0.8046875, "learning_rate": 1.764643799482716e-05, "loss": 0.1455, "step": 35568 }, { "epoch": 26.057875457875458, "grad_norm": 1.0078125, "learning_rate": 1.7643875184419195e-05, "loss": 0.1042, "step": 35569 }, { "epoch": 26.05860805860806, "grad_norm": 1.2109375, "learning_rate": 1.7641312531435708e-05, "loss": 0.1501, "step": 35570 }, { "epoch": 26.05934065934066, "grad_norm": 1.2734375, "learning_rate": 1.7638750035888997e-05, "loss": 0.121, "step": 35571 }, { "epoch": 26.06007326007326, "grad_norm": 0.765625, "learning_rate": 1.7636187697791305e-05, "loss": 0.1073, "step": 35572 }, { "epoch": 26.06080586080586, "grad_norm": 0.9921875, "learning_rate": 1.7633625517154938e-05, "loss": 0.0984, "step": 35573 }, { "epoch": 26.06153846153846, "grad_norm": 1.1953125, "learning_rate": 1.7631063493992132e-05, "loss": 0.1086, "step": 35574 }, { "epoch": 26.062271062271062, "grad_norm": 1.4140625, "learning_rate": 1.7628501628315222e-05, "loss": 0.1104, "step": 35575 }, { "epoch": 26.063003663003663, "grad_norm": 1.2734375, "learning_rate": 1.7625939920136438e-05, "loss": 0.1028, "step": 35576 }, { "epoch": 26.063736263736264, "grad_norm": 1.0234375, "learning_rate": 1.7623378369468068e-05, "loss": 0.1467, "step": 35577 }, { "epoch": 26.064468864468864, "grad_norm": 1.265625, "learning_rate": 1.76208169763224e-05, "loss": 0.1352, "step": 35578 }, { "epoch": 26.065201465201465, "grad_norm": 1.234375, "learning_rate": 1.7618255740711655e-05, "loss": 0.1238, "step": 35579 }, { "epoch": 26.065934065934066, "grad_norm": 0.8046875, "learning_rate": 1.761569466264818e-05, "loss": 0.1152, "step": 35580 }, { "epoch": 26.066666666666666, "grad_norm": 1.7890625, "learning_rate": 1.7613133742144183e-05, "loss": 0.1347, "step": 35581 }, { "epoch": 26.067399267399267, "grad_norm": 0.79296875, "learning_rate": 1.7610572979211985e-05, "loss": 0.1011, "step": 35582 }, { "epoch": 26.068131868131868, "grad_norm": 1.6640625, "learning_rate": 1.7608012373863806e-05, "loss": 0.1451, "step": 35583 }, { "epoch": 26.06886446886447, "grad_norm": 1.0078125, "learning_rate": 1.7605451926111935e-05, "loss": 0.1366, "step": 35584 }, { "epoch": 26.06959706959707, "grad_norm": 0.9140625, "learning_rate": 1.7602891635968643e-05, "loss": 0.1257, "step": 35585 }, { "epoch": 26.07032967032967, "grad_norm": 1.015625, "learning_rate": 1.760033150344622e-05, "loss": 0.1448, "step": 35586 }, { "epoch": 26.07106227106227, "grad_norm": 0.6484375, "learning_rate": 1.7597771528556887e-05, "loss": 0.1104, "step": 35587 }, { "epoch": 26.07179487179487, "grad_norm": 1.65625, "learning_rate": 1.7595211711312933e-05, "loss": 0.1584, "step": 35588 }, { "epoch": 26.072527472527472, "grad_norm": 1.671875, "learning_rate": 1.7592652051726622e-05, "loss": 0.1075, "step": 35589 }, { "epoch": 26.073260073260073, "grad_norm": 1.2578125, "learning_rate": 1.759009254981022e-05, "loss": 0.1118, "step": 35590 }, { "epoch": 26.073992673992674, "grad_norm": 1.1796875, "learning_rate": 1.7587533205576e-05, "loss": 0.1111, "step": 35591 }, { "epoch": 26.074725274725274, "grad_norm": 1.234375, "learning_rate": 1.7584974019036195e-05, "loss": 0.1495, "step": 35592 }, { "epoch": 26.075457875457875, "grad_norm": 1.1484375, "learning_rate": 1.758241499020311e-05, "loss": 0.0992, "step": 35593 }, { "epoch": 26.076190476190476, "grad_norm": 1.2890625, "learning_rate": 1.7579856119088953e-05, "loss": 0.1425, "step": 35594 }, { "epoch": 26.076923076923077, "grad_norm": 1.6484375, "learning_rate": 1.7577297405706024e-05, "loss": 0.1305, "step": 35595 }, { "epoch": 26.077655677655677, "grad_norm": 1.1796875, "learning_rate": 1.7574738850066562e-05, "loss": 0.1266, "step": 35596 }, { "epoch": 26.078388278388278, "grad_norm": 0.796875, "learning_rate": 1.7572180452182853e-05, "loss": 0.1243, "step": 35597 }, { "epoch": 26.07912087912088, "grad_norm": 0.703125, "learning_rate": 1.7569622212067125e-05, "loss": 0.1148, "step": 35598 }, { "epoch": 26.07985347985348, "grad_norm": 0.92578125, "learning_rate": 1.7567064129731646e-05, "loss": 0.1177, "step": 35599 }, { "epoch": 26.08058608058608, "grad_norm": 1.859375, "learning_rate": 1.756450620518867e-05, "loss": 0.1216, "step": 35600 }, { "epoch": 26.08131868131868, "grad_norm": 0.8046875, "learning_rate": 1.7561948438450474e-05, "loss": 0.1194, "step": 35601 }, { "epoch": 26.08205128205128, "grad_norm": 0.7109375, "learning_rate": 1.755939082952928e-05, "loss": 0.1302, "step": 35602 }, { "epoch": 26.082783882783882, "grad_norm": 1.40625, "learning_rate": 1.755683337843736e-05, "loss": 0.0977, "step": 35603 }, { "epoch": 26.083516483516483, "grad_norm": 1.1640625, "learning_rate": 1.7554276085186984e-05, "loss": 0.1076, "step": 35604 }, { "epoch": 26.084249084249084, "grad_norm": 1.375, "learning_rate": 1.755171894979035e-05, "loss": 0.1056, "step": 35605 }, { "epoch": 26.084981684981685, "grad_norm": 1.0703125, "learning_rate": 1.7549161972259777e-05, "loss": 0.132, "step": 35606 }, { "epoch": 26.085714285714285, "grad_norm": 1.03125, "learning_rate": 1.7546605152607476e-05, "loss": 0.1036, "step": 35607 }, { "epoch": 26.086446886446886, "grad_norm": 1.3671875, "learning_rate": 1.7544048490845725e-05, "loss": 0.1645, "step": 35608 }, { "epoch": 26.087179487179487, "grad_norm": 0.65625, "learning_rate": 1.7541491986986734e-05, "loss": 0.1093, "step": 35609 }, { "epoch": 26.087912087912088, "grad_norm": 0.91015625, "learning_rate": 1.7538935641042776e-05, "loss": 0.1039, "step": 35610 }, { "epoch": 26.08864468864469, "grad_norm": 1.453125, "learning_rate": 1.75363794530261e-05, "loss": 0.1466, "step": 35611 }, { "epoch": 26.08937728937729, "grad_norm": 0.828125, "learning_rate": 1.7533823422948966e-05, "loss": 0.134, "step": 35612 }, { "epoch": 26.09010989010989, "grad_norm": 0.67578125, "learning_rate": 1.7531267550823587e-05, "loss": 0.1123, "step": 35613 }, { "epoch": 26.09084249084249, "grad_norm": 0.94140625, "learning_rate": 1.7528711836662232e-05, "loss": 0.1142, "step": 35614 }, { "epoch": 26.09157509157509, "grad_norm": 0.91015625, "learning_rate": 1.752615628047714e-05, "loss": 0.0993, "step": 35615 }, { "epoch": 26.092307692307692, "grad_norm": 1.046875, "learning_rate": 1.7523600882280563e-05, "loss": 0.0963, "step": 35616 }, { "epoch": 26.093040293040293, "grad_norm": 0.75, "learning_rate": 1.7521045642084754e-05, "loss": 0.1133, "step": 35617 }, { "epoch": 26.093772893772893, "grad_norm": 1.3125, "learning_rate": 1.751849055990192e-05, "loss": 0.1329, "step": 35618 }, { "epoch": 26.094505494505494, "grad_norm": 1.234375, "learning_rate": 1.7515935635744345e-05, "loss": 0.1374, "step": 35619 }, { "epoch": 26.095238095238095, "grad_norm": 1.1953125, "learning_rate": 1.7513380869624237e-05, "loss": 0.1072, "step": 35620 }, { "epoch": 26.095970695970696, "grad_norm": 1.0, "learning_rate": 1.7510826261553843e-05, "loss": 0.1328, "step": 35621 }, { "epoch": 26.096703296703296, "grad_norm": 0.73046875, "learning_rate": 1.7508271811545418e-05, "loss": 0.088, "step": 35622 }, { "epoch": 26.097435897435897, "grad_norm": 1.03125, "learning_rate": 1.7505717519611203e-05, "loss": 0.1118, "step": 35623 }, { "epoch": 26.098168498168498, "grad_norm": 1.0703125, "learning_rate": 1.7503163385763413e-05, "loss": 0.1303, "step": 35624 }, { "epoch": 26.0989010989011, "grad_norm": 1.1640625, "learning_rate": 1.7500609410014295e-05, "loss": 0.1105, "step": 35625 }, { "epoch": 26.0996336996337, "grad_norm": 0.98046875, "learning_rate": 1.7498055592376095e-05, "loss": 0.1088, "step": 35626 }, { "epoch": 26.1003663003663, "grad_norm": 1.1171875, "learning_rate": 1.749550193286106e-05, "loss": 0.138, "step": 35627 }, { "epoch": 26.1010989010989, "grad_norm": 1.2421875, "learning_rate": 1.7492948431481393e-05, "loss": 0.0984, "step": 35628 }, { "epoch": 26.1018315018315, "grad_norm": 1.0625, "learning_rate": 1.7490395088249348e-05, "loss": 0.1299, "step": 35629 }, { "epoch": 26.102564102564102, "grad_norm": 0.9296875, "learning_rate": 1.7487841903177175e-05, "loss": 0.1094, "step": 35630 }, { "epoch": 26.103296703296703, "grad_norm": 0.87890625, "learning_rate": 1.7485288876277052e-05, "loss": 0.1438, "step": 35631 }, { "epoch": 26.104029304029304, "grad_norm": 0.70703125, "learning_rate": 1.7482736007561276e-05, "loss": 0.1074, "step": 35632 }, { "epoch": 26.104761904761904, "grad_norm": 1.0390625, "learning_rate": 1.748018329704204e-05, "loss": 0.1298, "step": 35633 }, { "epoch": 26.105494505494505, "grad_norm": 0.96484375, "learning_rate": 1.7477630744731603e-05, "loss": 0.1188, "step": 35634 }, { "epoch": 26.106227106227106, "grad_norm": 1.0546875, "learning_rate": 1.747507835064216e-05, "loss": 0.1353, "step": 35635 }, { "epoch": 26.106959706959707, "grad_norm": 0.90234375, "learning_rate": 1.7472526114785955e-05, "loss": 0.12, "step": 35636 }, { "epoch": 26.107692307692307, "grad_norm": 1.171875, "learning_rate": 1.7469974037175225e-05, "loss": 0.1657, "step": 35637 }, { "epoch": 26.108424908424908, "grad_norm": 1.015625, "learning_rate": 1.7467422117822207e-05, "loss": 0.1141, "step": 35638 }, { "epoch": 26.10915750915751, "grad_norm": 1.21875, "learning_rate": 1.7464870356739094e-05, "loss": 0.1305, "step": 35639 }, { "epoch": 26.10989010989011, "grad_norm": 1.234375, "learning_rate": 1.7462318753938134e-05, "loss": 0.1469, "step": 35640 }, { "epoch": 26.11062271062271, "grad_norm": 1.0078125, "learning_rate": 1.745976730943155e-05, "loss": 0.1137, "step": 35641 }, { "epoch": 26.11135531135531, "grad_norm": 1.2109375, "learning_rate": 1.7457216023231566e-05, "loss": 0.1296, "step": 35642 }, { "epoch": 26.11208791208791, "grad_norm": 0.8515625, "learning_rate": 1.7454664895350426e-05, "loss": 0.1149, "step": 35643 }, { "epoch": 26.112820512820512, "grad_norm": 0.7890625, "learning_rate": 1.7452113925800314e-05, "loss": 0.0926, "step": 35644 }, { "epoch": 26.113553113553113, "grad_norm": 0.984375, "learning_rate": 1.7449563114593494e-05, "loss": 0.1039, "step": 35645 }, { "epoch": 26.114285714285714, "grad_norm": 1.0859375, "learning_rate": 1.744701246174215e-05, "loss": 0.1367, "step": 35646 }, { "epoch": 26.115018315018315, "grad_norm": 1.171875, "learning_rate": 1.744446196725852e-05, "loss": 0.1256, "step": 35647 }, { "epoch": 26.115750915750915, "grad_norm": 0.98828125, "learning_rate": 1.7441911631154823e-05, "loss": 0.1105, "step": 35648 }, { "epoch": 26.116483516483516, "grad_norm": 0.640625, "learning_rate": 1.7439361453443294e-05, "loss": 0.1041, "step": 35649 }, { "epoch": 26.117216117216117, "grad_norm": 0.70703125, "learning_rate": 1.7436811434136124e-05, "loss": 0.1259, "step": 35650 }, { "epoch": 26.117948717948718, "grad_norm": 1.0390625, "learning_rate": 1.7434261573245543e-05, "loss": 0.1084, "step": 35651 }, { "epoch": 26.11868131868132, "grad_norm": 0.8203125, "learning_rate": 1.7431711870783773e-05, "loss": 0.1146, "step": 35652 }, { "epoch": 26.11941391941392, "grad_norm": 1.21875, "learning_rate": 1.742916232676304e-05, "loss": 0.127, "step": 35653 }, { "epoch": 26.12014652014652, "grad_norm": 1.125, "learning_rate": 1.7426612941195526e-05, "loss": 0.1143, "step": 35654 }, { "epoch": 26.12087912087912, "grad_norm": 1.1953125, "learning_rate": 1.7424063714093474e-05, "loss": 0.0958, "step": 35655 }, { "epoch": 26.12161172161172, "grad_norm": 0.9609375, "learning_rate": 1.7421514645469102e-05, "loss": 0.0979, "step": 35656 }, { "epoch": 26.122344322344322, "grad_norm": 0.72265625, "learning_rate": 1.7418965735334585e-05, "loss": 0.1072, "step": 35657 }, { "epoch": 26.123076923076923, "grad_norm": 0.796875, "learning_rate": 1.741641698370219e-05, "loss": 0.1096, "step": 35658 }, { "epoch": 26.123809523809523, "grad_norm": 0.83203125, "learning_rate": 1.741386839058409e-05, "loss": 0.0931, "step": 35659 }, { "epoch": 26.124542124542124, "grad_norm": 0.875, "learning_rate": 1.7411319955992524e-05, "loss": 0.1027, "step": 35660 }, { "epoch": 26.125274725274725, "grad_norm": 1.5625, "learning_rate": 1.740877167993967e-05, "loss": 0.1201, "step": 35661 }, { "epoch": 26.126007326007326, "grad_norm": 0.95703125, "learning_rate": 1.7406223562437753e-05, "loss": 0.1413, "step": 35662 }, { "epoch": 26.126739926739926, "grad_norm": 1.3046875, "learning_rate": 1.740367560349898e-05, "loss": 0.1144, "step": 35663 }, { "epoch": 26.127472527472527, "grad_norm": 1.1328125, "learning_rate": 1.7401127803135586e-05, "loss": 0.0886, "step": 35664 }, { "epoch": 26.128205128205128, "grad_norm": 0.75390625, "learning_rate": 1.7398580161359727e-05, "loss": 0.1183, "step": 35665 }, { "epoch": 26.12893772893773, "grad_norm": 1.2578125, "learning_rate": 1.7396032678183646e-05, "loss": 0.1066, "step": 35666 }, { "epoch": 26.12967032967033, "grad_norm": 0.78515625, "learning_rate": 1.7393485353619536e-05, "loss": 0.132, "step": 35667 }, { "epoch": 26.13040293040293, "grad_norm": 0.9921875, "learning_rate": 1.7390938187679608e-05, "loss": 0.1325, "step": 35668 }, { "epoch": 26.13113553113553, "grad_norm": 1.09375, "learning_rate": 1.7388391180376083e-05, "loss": 0.1207, "step": 35669 }, { "epoch": 26.13186813186813, "grad_norm": 1.078125, "learning_rate": 1.7385844331721117e-05, "loss": 0.151, "step": 35670 }, { "epoch": 26.132600732600732, "grad_norm": 1.15625, "learning_rate": 1.738329764172697e-05, "loss": 0.1201, "step": 35671 }, { "epoch": 26.133333333333333, "grad_norm": 1.0234375, "learning_rate": 1.7380751110405792e-05, "loss": 0.1329, "step": 35672 }, { "epoch": 26.134065934065934, "grad_norm": 0.65234375, "learning_rate": 1.7378204737769817e-05, "loss": 0.1305, "step": 35673 }, { "epoch": 26.134798534798534, "grad_norm": 1.265625, "learning_rate": 1.7375658523831228e-05, "loss": 0.104, "step": 35674 }, { "epoch": 26.135531135531135, "grad_norm": 1.2734375, "learning_rate": 1.7373112468602253e-05, "loss": 0.1145, "step": 35675 }, { "epoch": 26.136263736263736, "grad_norm": 0.609375, "learning_rate": 1.737056657209505e-05, "loss": 0.1133, "step": 35676 }, { "epoch": 26.136996336996337, "grad_norm": 0.70703125, "learning_rate": 1.7368020834321844e-05, "loss": 0.1145, "step": 35677 }, { "epoch": 26.137728937728937, "grad_norm": 1.484375, "learning_rate": 1.7365475255294826e-05, "loss": 0.1193, "step": 35678 }, { "epoch": 26.138461538461538, "grad_norm": 0.75390625, "learning_rate": 1.7362929835026207e-05, "loss": 0.111, "step": 35679 }, { "epoch": 26.13919413919414, "grad_norm": 1.0703125, "learning_rate": 1.7360384573528157e-05, "loss": 0.1084, "step": 35680 }, { "epoch": 26.13992673992674, "grad_norm": 1.515625, "learning_rate": 1.7357839470812874e-05, "loss": 0.1281, "step": 35681 }, { "epoch": 26.14065934065934, "grad_norm": 1.5546875, "learning_rate": 1.735529452689259e-05, "loss": 0.1291, "step": 35682 }, { "epoch": 26.14139194139194, "grad_norm": 1.2265625, "learning_rate": 1.7352749741779435e-05, "loss": 0.1281, "step": 35683 }, { "epoch": 26.14212454212454, "grad_norm": 1.1171875, "learning_rate": 1.7350205115485667e-05, "loss": 0.1168, "step": 35684 }, { "epoch": 26.142857142857142, "grad_norm": 1.1640625, "learning_rate": 1.7347660648023428e-05, "loss": 0.1375, "step": 35685 }, { "epoch": 26.143589743589743, "grad_norm": 1.421875, "learning_rate": 1.7345116339404958e-05, "loss": 0.1342, "step": 35686 }, { "epoch": 26.144322344322344, "grad_norm": 1.4375, "learning_rate": 1.7342572189642392e-05, "loss": 0.1211, "step": 35687 }, { "epoch": 26.145054945054945, "grad_norm": 1.0546875, "learning_rate": 1.7340028198747956e-05, "loss": 0.1344, "step": 35688 }, { "epoch": 26.145787545787545, "grad_norm": 0.89453125, "learning_rate": 1.733748436673383e-05, "loss": 0.0971, "step": 35689 }, { "epoch": 26.146520146520146, "grad_norm": 0.98828125, "learning_rate": 1.7334940693612216e-05, "loss": 0.1264, "step": 35690 }, { "epoch": 26.147252747252747, "grad_norm": 0.68359375, "learning_rate": 1.733239717939527e-05, "loss": 0.0989, "step": 35691 }, { "epoch": 26.147985347985347, "grad_norm": 1.015625, "learning_rate": 1.73298538240952e-05, "loss": 0.1395, "step": 35692 }, { "epoch": 26.148717948717948, "grad_norm": 1.328125, "learning_rate": 1.7327310627724204e-05, "loss": 0.1344, "step": 35693 }, { "epoch": 26.14945054945055, "grad_norm": 0.81640625, "learning_rate": 1.7324767590294422e-05, "loss": 0.1134, "step": 35694 }, { "epoch": 26.15018315018315, "grad_norm": 1.625, "learning_rate": 1.7322224711818104e-05, "loss": 0.148, "step": 35695 }, { "epoch": 26.15091575091575, "grad_norm": 0.90625, "learning_rate": 1.7319681992307383e-05, "loss": 0.1136, "step": 35696 }, { "epoch": 26.15164835164835, "grad_norm": 0.9921875, "learning_rate": 1.731713943177447e-05, "loss": 0.136, "step": 35697 }, { "epoch": 26.152380952380952, "grad_norm": 1.640625, "learning_rate": 1.731459703023152e-05, "loss": 0.1306, "step": 35698 }, { "epoch": 26.153113553113553, "grad_norm": 1.2109375, "learning_rate": 1.731205478769073e-05, "loss": 0.1331, "step": 35699 }, { "epoch": 26.153846153846153, "grad_norm": 0.52734375, "learning_rate": 1.7309512704164283e-05, "loss": 0.123, "step": 35700 }, { "epoch": 26.154578754578754, "grad_norm": 1.2734375, "learning_rate": 1.7306970779664368e-05, "loss": 0.1362, "step": 35701 }, { "epoch": 26.155311355311355, "grad_norm": 1.5234375, "learning_rate": 1.730442901420314e-05, "loss": 0.1321, "step": 35702 }, { "epoch": 26.156043956043955, "grad_norm": 1.0390625, "learning_rate": 1.730188740779279e-05, "loss": 0.1079, "step": 35703 }, { "epoch": 26.156776556776556, "grad_norm": 1.0234375, "learning_rate": 1.7299345960445487e-05, "loss": 0.1219, "step": 35704 }, { "epoch": 26.157509157509157, "grad_norm": 0.79296875, "learning_rate": 1.7296804672173425e-05, "loss": 0.1253, "step": 35705 }, { "epoch": 26.158241758241758, "grad_norm": 0.98828125, "learning_rate": 1.7294263542988787e-05, "loss": 0.1254, "step": 35706 }, { "epoch": 26.15897435897436, "grad_norm": 1.1953125, "learning_rate": 1.7291722572903717e-05, "loss": 0.1248, "step": 35707 }, { "epoch": 26.15970695970696, "grad_norm": 0.671875, "learning_rate": 1.728918176193042e-05, "loss": 0.1104, "step": 35708 }, { "epoch": 26.16043956043956, "grad_norm": 0.83203125, "learning_rate": 1.728664111008103e-05, "loss": 0.1214, "step": 35709 }, { "epoch": 26.16117216117216, "grad_norm": 1.4375, "learning_rate": 1.728410061736777e-05, "loss": 0.108, "step": 35710 }, { "epoch": 26.16190476190476, "grad_norm": 1.234375, "learning_rate": 1.728156028380277e-05, "loss": 0.137, "step": 35711 }, { "epoch": 26.162637362637362, "grad_norm": 0.94140625, "learning_rate": 1.7279020109398242e-05, "loss": 0.1144, "step": 35712 }, { "epoch": 26.163369963369963, "grad_norm": 1.34375, "learning_rate": 1.7276480094166318e-05, "loss": 0.1098, "step": 35713 }, { "epoch": 26.164102564102564, "grad_norm": 1.15625, "learning_rate": 1.7273940238119186e-05, "loss": 0.1487, "step": 35714 }, { "epoch": 26.164835164835164, "grad_norm": 1.265625, "learning_rate": 1.7271400541269007e-05, "loss": 0.1335, "step": 35715 }, { "epoch": 26.165567765567765, "grad_norm": 0.97265625, "learning_rate": 1.7268861003627978e-05, "loss": 0.1108, "step": 35716 }, { "epoch": 26.166300366300366, "grad_norm": 1.1484375, "learning_rate": 1.7266321625208227e-05, "loss": 0.1122, "step": 35717 }, { "epoch": 26.167032967032966, "grad_norm": 1.0625, "learning_rate": 1.7263782406021936e-05, "loss": 0.1409, "step": 35718 }, { "epoch": 26.167765567765567, "grad_norm": 1.34375, "learning_rate": 1.7261243346081292e-05, "loss": 0.1384, "step": 35719 }, { "epoch": 26.168498168498168, "grad_norm": 1.09375, "learning_rate": 1.7258704445398413e-05, "loss": 0.1329, "step": 35720 }, { "epoch": 26.16923076923077, "grad_norm": 0.69921875, "learning_rate": 1.7256165703985522e-05, "loss": 0.116, "step": 35721 }, { "epoch": 26.16996336996337, "grad_norm": 1.2421875, "learning_rate": 1.7253627121854747e-05, "loss": 0.1296, "step": 35722 }, { "epoch": 26.17069597069597, "grad_norm": 0.9609375, "learning_rate": 1.725108869901826e-05, "loss": 0.1336, "step": 35723 }, { "epoch": 26.17142857142857, "grad_norm": 1.015625, "learning_rate": 1.724855043548822e-05, "loss": 0.1074, "step": 35724 }, { "epoch": 26.17216117216117, "grad_norm": 1.0703125, "learning_rate": 1.7246012331276785e-05, "loss": 0.1191, "step": 35725 }, { "epoch": 26.172893772893772, "grad_norm": 1.1796875, "learning_rate": 1.7243474386396116e-05, "loss": 0.1206, "step": 35726 }, { "epoch": 26.173626373626373, "grad_norm": 0.9921875, "learning_rate": 1.7240936600858405e-05, "loss": 0.1359, "step": 35727 }, { "epoch": 26.174358974358974, "grad_norm": 1.2421875, "learning_rate": 1.7238398974675763e-05, "loss": 0.0949, "step": 35728 }, { "epoch": 26.175091575091574, "grad_norm": 0.7890625, "learning_rate": 1.7235861507860368e-05, "loss": 0.1034, "step": 35729 }, { "epoch": 26.175824175824175, "grad_norm": 0.95703125, "learning_rate": 1.7233324200424386e-05, "loss": 0.1311, "step": 35730 }, { "epoch": 26.176556776556776, "grad_norm": 0.765625, "learning_rate": 1.7230787052379964e-05, "loss": 0.1299, "step": 35731 }, { "epoch": 26.177289377289377, "grad_norm": 0.796875, "learning_rate": 1.7228250063739286e-05, "loss": 0.0979, "step": 35732 }, { "epoch": 26.178021978021977, "grad_norm": 0.81640625, "learning_rate": 1.722571323451446e-05, "loss": 0.1067, "step": 35733 }, { "epoch": 26.178754578754578, "grad_norm": 1.3359375, "learning_rate": 1.7223176564717688e-05, "loss": 0.1432, "step": 35734 }, { "epoch": 26.17948717948718, "grad_norm": 0.69921875, "learning_rate": 1.722064005436106e-05, "loss": 0.1194, "step": 35735 }, { "epoch": 26.18021978021978, "grad_norm": 0.97265625, "learning_rate": 1.721810370345681e-05, "loss": 0.1243, "step": 35736 }, { "epoch": 26.18095238095238, "grad_norm": 0.8515625, "learning_rate": 1.7215567512017032e-05, "loss": 0.1224, "step": 35737 }, { "epoch": 26.18168498168498, "grad_norm": 1.09375, "learning_rate": 1.7213031480053915e-05, "loss": 0.1495, "step": 35738 }, { "epoch": 26.18241758241758, "grad_norm": 1.1484375, "learning_rate": 1.7210495607579576e-05, "loss": 0.1233, "step": 35739 }, { "epoch": 26.183150183150182, "grad_norm": 0.96484375, "learning_rate": 1.7207959894606176e-05, "loss": 0.1417, "step": 35740 }, { "epoch": 26.183882783882783, "grad_norm": 1.1875, "learning_rate": 1.720542434114587e-05, "loss": 0.096, "step": 35741 }, { "epoch": 26.184615384615384, "grad_norm": 1.078125, "learning_rate": 1.7202888947210827e-05, "loss": 0.1181, "step": 35742 }, { "epoch": 26.185347985347985, "grad_norm": 1.28125, "learning_rate": 1.7200353712813154e-05, "loss": 0.1189, "step": 35743 }, { "epoch": 26.186080586080585, "grad_norm": 1.5859375, "learning_rate": 1.719781863796502e-05, "loss": 0.1633, "step": 35744 }, { "epoch": 26.186813186813186, "grad_norm": 1.2265625, "learning_rate": 1.719528372267858e-05, "loss": 0.1176, "step": 35745 }, { "epoch": 26.187545787545787, "grad_norm": 0.9609375, "learning_rate": 1.7192748966965945e-05, "loss": 0.1313, "step": 35746 }, { "epoch": 26.188278388278388, "grad_norm": 1.1875, "learning_rate": 1.7190214370839314e-05, "loss": 0.1237, "step": 35747 }, { "epoch": 26.18901098901099, "grad_norm": 1.125, "learning_rate": 1.718767993431078e-05, "loss": 0.1153, "step": 35748 }, { "epoch": 26.18974358974359, "grad_norm": 1.7265625, "learning_rate": 1.7185145657392525e-05, "loss": 0.1383, "step": 35749 }, { "epoch": 26.19047619047619, "grad_norm": 1.0078125, "learning_rate": 1.7182611540096654e-05, "loss": 0.1015, "step": 35750 }, { "epoch": 26.19120879120879, "grad_norm": 0.7734375, "learning_rate": 1.7180077582435334e-05, "loss": 0.1181, "step": 35751 }, { "epoch": 26.19194139194139, "grad_norm": 0.78515625, "learning_rate": 1.7177543784420702e-05, "loss": 0.1092, "step": 35752 }, { "epoch": 26.192673992673992, "grad_norm": 1.7734375, "learning_rate": 1.7175010146064907e-05, "loss": 0.1624, "step": 35753 }, { "epoch": 26.193406593406593, "grad_norm": 1.2734375, "learning_rate": 1.7172476667380068e-05, "loss": 0.1337, "step": 35754 }, { "epoch": 26.194139194139193, "grad_norm": 1.1171875, "learning_rate": 1.716994334837833e-05, "loss": 0.111, "step": 35755 }, { "epoch": 26.194871794871794, "grad_norm": 1.296875, "learning_rate": 1.716741018907184e-05, "loss": 0.1099, "step": 35756 }, { "epoch": 26.195604395604395, "grad_norm": 1.171875, "learning_rate": 1.716487718947272e-05, "loss": 0.1136, "step": 35757 }, { "epoch": 26.196336996336996, "grad_norm": 1.015625, "learning_rate": 1.7162344349593144e-05, "loss": 0.1162, "step": 35758 }, { "epoch": 26.197069597069596, "grad_norm": 1.25, "learning_rate": 1.7159811669445197e-05, "loss": 0.1268, "step": 35759 }, { "epoch": 26.197802197802197, "grad_norm": 0.99609375, "learning_rate": 1.715727914904106e-05, "loss": 0.1446, "step": 35760 }, { "epoch": 26.198534798534798, "grad_norm": 0.8671875, "learning_rate": 1.7154746788392805e-05, "loss": 0.1174, "step": 35761 }, { "epoch": 26.1992673992674, "grad_norm": 0.890625, "learning_rate": 1.715221458751264e-05, "loss": 0.1331, "step": 35762 }, { "epoch": 26.2, "grad_norm": 1.8046875, "learning_rate": 1.714968254641265e-05, "loss": 0.1438, "step": 35763 }, { "epoch": 26.2007326007326, "grad_norm": 1.125, "learning_rate": 1.7147150665104985e-05, "loss": 0.1403, "step": 35764 }, { "epoch": 26.2014652014652, "grad_norm": 1.0, "learning_rate": 1.714461894360176e-05, "loss": 0.1039, "step": 35765 }, { "epoch": 26.2021978021978, "grad_norm": 1.265625, "learning_rate": 1.714208738191512e-05, "loss": 0.1271, "step": 35766 }, { "epoch": 26.202930402930402, "grad_norm": 0.91015625, "learning_rate": 1.7139555980057186e-05, "loss": 0.1594, "step": 35767 }, { "epoch": 26.203663003663003, "grad_norm": 1.1875, "learning_rate": 1.71370247380401e-05, "loss": 0.1253, "step": 35768 }, { "epoch": 26.204395604395604, "grad_norm": 0.78515625, "learning_rate": 1.7134493655875976e-05, "loss": 0.1173, "step": 35769 }, { "epoch": 26.205128205128204, "grad_norm": 1.140625, "learning_rate": 1.7131962733576937e-05, "loss": 0.1263, "step": 35770 }, { "epoch": 26.205860805860805, "grad_norm": 1.1796875, "learning_rate": 1.712943197115514e-05, "loss": 0.1398, "step": 35771 }, { "epoch": 26.206593406593406, "grad_norm": 0.625, "learning_rate": 1.7126901368622657e-05, "loss": 0.1065, "step": 35772 }, { "epoch": 26.207326007326007, "grad_norm": 1.1171875, "learning_rate": 1.7124370925991674e-05, "loss": 0.1107, "step": 35773 }, { "epoch": 26.208058608058607, "grad_norm": 0.796875, "learning_rate": 1.7121840643274275e-05, "loss": 0.1145, "step": 35774 }, { "epoch": 26.208791208791208, "grad_norm": 0.98046875, "learning_rate": 1.7119310520482608e-05, "loss": 0.1223, "step": 35775 }, { "epoch": 26.20952380952381, "grad_norm": 1.0234375, "learning_rate": 1.7116780557628766e-05, "loss": 0.1007, "step": 35776 }, { "epoch": 26.21025641025641, "grad_norm": 0.57421875, "learning_rate": 1.7114250754724885e-05, "loss": 0.1118, "step": 35777 }, { "epoch": 26.21098901098901, "grad_norm": 1.5234375, "learning_rate": 1.7111721111783096e-05, "loss": 0.1171, "step": 35778 }, { "epoch": 26.21172161172161, "grad_norm": 1.0234375, "learning_rate": 1.7109191628815526e-05, "loss": 0.1065, "step": 35779 }, { "epoch": 26.21245421245421, "grad_norm": 1.265625, "learning_rate": 1.7106662305834262e-05, "loss": 0.1148, "step": 35780 }, { "epoch": 26.213186813186812, "grad_norm": 1.0, "learning_rate": 1.710413314285144e-05, "loss": 0.126, "step": 35781 }, { "epoch": 26.213919413919413, "grad_norm": 0.921875, "learning_rate": 1.7101604139879184e-05, "loss": 0.1425, "step": 35782 }, { "epoch": 26.214652014652014, "grad_norm": 0.734375, "learning_rate": 1.7099075296929597e-05, "loss": 0.0955, "step": 35783 }, { "epoch": 26.215384615384615, "grad_norm": 1.171875, "learning_rate": 1.7096546614014828e-05, "loss": 0.0972, "step": 35784 }, { "epoch": 26.216117216117215, "grad_norm": 0.9296875, "learning_rate": 1.7094018091146954e-05, "loss": 0.1153, "step": 35785 }, { "epoch": 26.216849816849816, "grad_norm": 0.80859375, "learning_rate": 1.709148972833812e-05, "loss": 0.1185, "step": 35786 }, { "epoch": 26.217582417582417, "grad_norm": 1.3359375, "learning_rate": 1.7088961525600395e-05, "loss": 0.1304, "step": 35787 }, { "epoch": 26.218315018315018, "grad_norm": 1.0390625, "learning_rate": 1.7086433482945952e-05, "loss": 0.1318, "step": 35788 }, { "epoch": 26.21904761904762, "grad_norm": 1.2421875, "learning_rate": 1.708390560038686e-05, "loss": 0.1005, "step": 35789 }, { "epoch": 26.21978021978022, "grad_norm": 0.75, "learning_rate": 1.708137787793527e-05, "loss": 0.1227, "step": 35790 }, { "epoch": 26.22051282051282, "grad_norm": 0.7578125, "learning_rate": 1.707885031560325e-05, "loss": 0.1187, "step": 35791 }, { "epoch": 26.22124542124542, "grad_norm": 1.203125, "learning_rate": 1.7076322913402923e-05, "loss": 0.1417, "step": 35792 }, { "epoch": 26.22197802197802, "grad_norm": 0.71484375, "learning_rate": 1.707379567134641e-05, "loss": 0.1173, "step": 35793 }, { "epoch": 26.222710622710622, "grad_norm": 0.578125, "learning_rate": 1.707126858944583e-05, "loss": 0.102, "step": 35794 }, { "epoch": 26.223443223443223, "grad_norm": 1.3203125, "learning_rate": 1.706874166771326e-05, "loss": 0.1039, "step": 35795 }, { "epoch": 26.224175824175823, "grad_norm": 0.9921875, "learning_rate": 1.7066214906160825e-05, "loss": 0.136, "step": 35796 }, { "epoch": 26.224908424908424, "grad_norm": 0.6640625, "learning_rate": 1.706368830480065e-05, "loss": 0.1141, "step": 35797 }, { "epoch": 26.225641025641025, "grad_norm": 1.0625, "learning_rate": 1.706116186364478e-05, "loss": 0.1185, "step": 35798 }, { "epoch": 26.226373626373626, "grad_norm": 0.90234375, "learning_rate": 1.705863558270539e-05, "loss": 0.1068, "step": 35799 }, { "epoch": 26.227106227106226, "grad_norm": 0.8046875, "learning_rate": 1.705610946199455e-05, "loss": 0.1088, "step": 35800 }, { "epoch": 26.227838827838827, "grad_norm": 0.7890625, "learning_rate": 1.7053583501524385e-05, "loss": 0.0989, "step": 35801 }, { "epoch": 26.228571428571428, "grad_norm": 1.421875, "learning_rate": 1.705105770130696e-05, "loss": 0.1203, "step": 35802 }, { "epoch": 26.22930402930403, "grad_norm": 1.1171875, "learning_rate": 1.7048532061354397e-05, "loss": 0.106, "step": 35803 }, { "epoch": 26.23003663003663, "grad_norm": 1.0390625, "learning_rate": 1.7046006581678794e-05, "loss": 0.0939, "step": 35804 }, { "epoch": 26.23076923076923, "grad_norm": 1.21875, "learning_rate": 1.704348126229228e-05, "loss": 0.143, "step": 35805 }, { "epoch": 26.23150183150183, "grad_norm": 1.0390625, "learning_rate": 1.704095610320691e-05, "loss": 0.1331, "step": 35806 }, { "epoch": 26.23223443223443, "grad_norm": 1.3046875, "learning_rate": 1.7038431104434803e-05, "loss": 0.1174, "step": 35807 }, { "epoch": 26.232967032967032, "grad_norm": 1.3828125, "learning_rate": 1.703590626598806e-05, "loss": 0.1083, "step": 35808 }, { "epoch": 26.233699633699633, "grad_norm": 1.890625, "learning_rate": 1.703338158787878e-05, "loss": 0.1325, "step": 35809 }, { "epoch": 26.234432234432234, "grad_norm": 1.1953125, "learning_rate": 1.703085707011906e-05, "loss": 0.118, "step": 35810 }, { "epoch": 26.235164835164834, "grad_norm": 1.1640625, "learning_rate": 1.702833271272098e-05, "loss": 0.1328, "step": 35811 }, { "epoch": 26.235897435897435, "grad_norm": 1.0546875, "learning_rate": 1.7025808515696664e-05, "loss": 0.1181, "step": 35812 }, { "epoch": 26.236630036630036, "grad_norm": 0.99609375, "learning_rate": 1.702328447905816e-05, "loss": 0.1185, "step": 35813 }, { "epoch": 26.237362637362637, "grad_norm": 0.89453125, "learning_rate": 1.7020760602817616e-05, "loss": 0.1249, "step": 35814 }, { "epoch": 26.238095238095237, "grad_norm": 0.859375, "learning_rate": 1.701823688698708e-05, "loss": 0.0872, "step": 35815 }, { "epoch": 26.238827838827838, "grad_norm": 1.5, "learning_rate": 1.7015713331578683e-05, "loss": 0.1182, "step": 35816 }, { "epoch": 26.23956043956044, "grad_norm": 1.375, "learning_rate": 1.7013189936604477e-05, "loss": 0.1449, "step": 35817 }, { "epoch": 26.24029304029304, "grad_norm": 1.3046875, "learning_rate": 1.7010666702076574e-05, "loss": 0.1355, "step": 35818 }, { "epoch": 26.24102564102564, "grad_norm": 1.3125, "learning_rate": 1.7008143628007058e-05, "loss": 0.1019, "step": 35819 }, { "epoch": 26.24175824175824, "grad_norm": 0.8359375, "learning_rate": 1.700562071440804e-05, "loss": 0.1242, "step": 35820 }, { "epoch": 26.24249084249084, "grad_norm": 0.7734375, "learning_rate": 1.7003097961291568e-05, "loss": 0.0919, "step": 35821 }, { "epoch": 26.243223443223442, "grad_norm": 1.0, "learning_rate": 1.700057536866975e-05, "loss": 0.1425, "step": 35822 }, { "epoch": 26.243956043956043, "grad_norm": 1.171875, "learning_rate": 1.699805293655469e-05, "loss": 0.1412, "step": 35823 }, { "epoch": 26.244688644688644, "grad_norm": 1.0625, "learning_rate": 1.6995530664958427e-05, "loss": 0.1196, "step": 35824 }, { "epoch": 26.245421245421245, "grad_norm": 1.140625, "learning_rate": 1.6993008553893103e-05, "loss": 0.1279, "step": 35825 }, { "epoch": 26.246153846153845, "grad_norm": 1.3984375, "learning_rate": 1.6990486603370765e-05, "loss": 0.1618, "step": 35826 }, { "epoch": 26.246886446886446, "grad_norm": 0.8984375, "learning_rate": 1.698796481340352e-05, "loss": 0.1121, "step": 35827 }, { "epoch": 26.247619047619047, "grad_norm": 1.3125, "learning_rate": 1.6985443184003416e-05, "loss": 0.1085, "step": 35828 }, { "epoch": 26.248351648351647, "grad_norm": 0.91796875, "learning_rate": 1.6982921715182562e-05, "loss": 0.1234, "step": 35829 }, { "epoch": 26.249084249084248, "grad_norm": 0.7109375, "learning_rate": 1.6980400406953023e-05, "loss": 0.1039, "step": 35830 }, { "epoch": 26.24981684981685, "grad_norm": 0.94140625, "learning_rate": 1.6977879259326917e-05, "loss": 0.1349, "step": 35831 }, { "epoch": 26.25054945054945, "grad_norm": 0.93359375, "learning_rate": 1.6975358272316273e-05, "loss": 0.1255, "step": 35832 }, { "epoch": 26.25128205128205, "grad_norm": 1.0390625, "learning_rate": 1.69728374459332e-05, "loss": 0.1411, "step": 35833 }, { "epoch": 26.25201465201465, "grad_norm": 1.1171875, "learning_rate": 1.697031678018976e-05, "loss": 0.1329, "step": 35834 }, { "epoch": 26.252747252747252, "grad_norm": 1.1171875, "learning_rate": 1.6967796275098045e-05, "loss": 0.1165, "step": 35835 }, { "epoch": 26.253479853479853, "grad_norm": 0.54296875, "learning_rate": 1.696527593067014e-05, "loss": 0.102, "step": 35836 }, { "epoch": 26.254212454212453, "grad_norm": 1.0234375, "learning_rate": 1.696275574691809e-05, "loss": 0.1215, "step": 35837 }, { "epoch": 26.254945054945054, "grad_norm": 1.4140625, "learning_rate": 1.6960235723854004e-05, "loss": 0.1136, "step": 35838 }, { "epoch": 26.255677655677655, "grad_norm": 1.0703125, "learning_rate": 1.695771586148991e-05, "loss": 0.1375, "step": 35839 }, { "epoch": 26.256410256410255, "grad_norm": 1.75, "learning_rate": 1.695519615983794e-05, "loss": 0.1567, "step": 35840 }, { "epoch": 26.257142857142856, "grad_norm": 1.28125, "learning_rate": 1.6952676618910114e-05, "loss": 0.14, "step": 35841 }, { "epoch": 26.257875457875457, "grad_norm": 1.1171875, "learning_rate": 1.695015723871855e-05, "loss": 0.117, "step": 35842 }, { "epoch": 26.258608058608058, "grad_norm": 1.046875, "learning_rate": 1.6947638019275282e-05, "loss": 0.1073, "step": 35843 }, { "epoch": 26.25934065934066, "grad_norm": 1.0859375, "learning_rate": 1.6945118960592395e-05, "loss": 0.1149, "step": 35844 }, { "epoch": 26.26007326007326, "grad_norm": 1.1328125, "learning_rate": 1.694260006268196e-05, "loss": 0.1197, "step": 35845 }, { "epoch": 26.26080586080586, "grad_norm": 1.1171875, "learning_rate": 1.694008132555604e-05, "loss": 0.1313, "step": 35846 }, { "epoch": 26.26153846153846, "grad_norm": 1.515625, "learning_rate": 1.6937562749226725e-05, "loss": 0.1135, "step": 35847 }, { "epoch": 26.26227106227106, "grad_norm": 1.1015625, "learning_rate": 1.6935044333706048e-05, "loss": 0.1478, "step": 35848 }, { "epoch": 26.263003663003662, "grad_norm": 1.375, "learning_rate": 1.6932526079006106e-05, "loss": 0.1254, "step": 35849 }, { "epoch": 26.263736263736263, "grad_norm": 1.9296875, "learning_rate": 1.693000798513892e-05, "loss": 0.1409, "step": 35850 }, { "epoch": 26.264468864468864, "grad_norm": 0.7578125, "learning_rate": 1.692749005211662e-05, "loss": 0.1233, "step": 35851 }, { "epoch": 26.265201465201464, "grad_norm": 0.83203125, "learning_rate": 1.692497227995122e-05, "loss": 0.1038, "step": 35852 }, { "epoch": 26.265934065934065, "grad_norm": 0.98046875, "learning_rate": 1.6922454668654812e-05, "loss": 0.1173, "step": 35853 }, { "epoch": 26.266666666666666, "grad_norm": 0.81640625, "learning_rate": 1.6919937218239436e-05, "loss": 0.1246, "step": 35854 }, { "epoch": 26.267399267399266, "grad_norm": 1.0546875, "learning_rate": 1.6917419928717166e-05, "loss": 0.1255, "step": 35855 }, { "epoch": 26.268131868131867, "grad_norm": 1.1328125, "learning_rate": 1.6914902800100056e-05, "loss": 0.1255, "step": 35856 }, { "epoch": 26.268864468864468, "grad_norm": 1.046875, "learning_rate": 1.691238583240019e-05, "loss": 0.1075, "step": 35857 }, { "epoch": 26.26959706959707, "grad_norm": 1.359375, "learning_rate": 1.690986902562959e-05, "loss": 0.1342, "step": 35858 }, { "epoch": 26.27032967032967, "grad_norm": 1.203125, "learning_rate": 1.690735237980034e-05, "loss": 0.107, "step": 35859 }, { "epoch": 26.27106227106227, "grad_norm": 0.76171875, "learning_rate": 1.6904835894924492e-05, "loss": 0.1162, "step": 35860 }, { "epoch": 26.27179487179487, "grad_norm": 1.09375, "learning_rate": 1.6902319571014103e-05, "loss": 0.1144, "step": 35861 }, { "epoch": 26.27252747252747, "grad_norm": 0.90625, "learning_rate": 1.689980340808124e-05, "loss": 0.1128, "step": 35862 }, { "epoch": 26.273260073260072, "grad_norm": 1.109375, "learning_rate": 1.6897287406137942e-05, "loss": 0.1314, "step": 35863 }, { "epoch": 26.273992673992673, "grad_norm": 0.828125, "learning_rate": 1.6894771565196283e-05, "loss": 0.1016, "step": 35864 }, { "epoch": 26.274725274725274, "grad_norm": 1.0546875, "learning_rate": 1.689225588526827e-05, "loss": 0.1409, "step": 35865 }, { "epoch": 26.275457875457874, "grad_norm": 0.765625, "learning_rate": 1.6889740366366034e-05, "loss": 0.1187, "step": 35866 }, { "epoch": 26.276190476190475, "grad_norm": 1.2421875, "learning_rate": 1.688722500850156e-05, "loss": 0.1299, "step": 35867 }, { "epoch": 26.276923076923076, "grad_norm": 1.140625, "learning_rate": 1.6884709811686946e-05, "loss": 0.0902, "step": 35868 }, { "epoch": 26.277655677655677, "grad_norm": 0.71875, "learning_rate": 1.68821947759342e-05, "loss": 0.0917, "step": 35869 }, { "epoch": 26.278388278388277, "grad_norm": 0.68359375, "learning_rate": 1.6879679901255402e-05, "loss": 0.0997, "step": 35870 }, { "epoch": 26.279120879120878, "grad_norm": 1.21875, "learning_rate": 1.687716518766259e-05, "loss": 0.1206, "step": 35871 }, { "epoch": 26.27985347985348, "grad_norm": 1.171875, "learning_rate": 1.687465063516782e-05, "loss": 0.1254, "step": 35872 }, { "epoch": 26.28058608058608, "grad_norm": 0.87109375, "learning_rate": 1.6872136243783156e-05, "loss": 0.1259, "step": 35873 }, { "epoch": 26.28131868131868, "grad_norm": 1.0234375, "learning_rate": 1.6869622013520606e-05, "loss": 0.1279, "step": 35874 }, { "epoch": 26.28205128205128, "grad_norm": 1.5, "learning_rate": 1.6867107944392255e-05, "loss": 0.1215, "step": 35875 }, { "epoch": 26.282783882783882, "grad_norm": 0.9296875, "learning_rate": 1.68645940364101e-05, "loss": 0.1246, "step": 35876 }, { "epoch": 26.283516483516483, "grad_norm": 0.6328125, "learning_rate": 1.6862080289586244e-05, "loss": 0.1163, "step": 35877 }, { "epoch": 26.284249084249083, "grad_norm": 0.81640625, "learning_rate": 1.685956670393269e-05, "loss": 0.1356, "step": 35878 }, { "epoch": 26.284981684981684, "grad_norm": 0.80078125, "learning_rate": 1.6857053279461516e-05, "loss": 0.11, "step": 35879 }, { "epoch": 26.285714285714285, "grad_norm": 1.2578125, "learning_rate": 1.6854540016184725e-05, "loss": 0.0978, "step": 35880 }, { "epoch": 26.286446886446885, "grad_norm": 1.4140625, "learning_rate": 1.6852026914114374e-05, "loss": 0.1199, "step": 35881 }, { "epoch": 26.287179487179486, "grad_norm": 1.1640625, "learning_rate": 1.684951397326251e-05, "loss": 0.1278, "step": 35882 }, { "epoch": 26.287912087912087, "grad_norm": 1.09375, "learning_rate": 1.684700119364119e-05, "loss": 0.1407, "step": 35883 }, { "epoch": 26.288644688644688, "grad_norm": 1.0859375, "learning_rate": 1.6844488575262414e-05, "loss": 0.1216, "step": 35884 }, { "epoch": 26.28937728937729, "grad_norm": 1.1875, "learning_rate": 1.6841976118138247e-05, "loss": 0.0983, "step": 35885 }, { "epoch": 26.29010989010989, "grad_norm": 0.78125, "learning_rate": 1.683946382228072e-05, "loss": 0.0983, "step": 35886 }, { "epoch": 26.29084249084249, "grad_norm": 1.296875, "learning_rate": 1.683695168770187e-05, "loss": 0.1255, "step": 35887 }, { "epoch": 26.29157509157509, "grad_norm": 1.2734375, "learning_rate": 1.683443971441375e-05, "loss": 0.1208, "step": 35888 }, { "epoch": 26.29230769230769, "grad_norm": 1.3125, "learning_rate": 1.683192790242837e-05, "loss": 0.1109, "step": 35889 }, { "epoch": 26.293040293040292, "grad_norm": 1.4609375, "learning_rate": 1.6829416251757788e-05, "loss": 0.1248, "step": 35890 }, { "epoch": 26.293772893772893, "grad_norm": 1.0625, "learning_rate": 1.6826904762413997e-05, "loss": 0.1204, "step": 35891 }, { "epoch": 26.294505494505493, "grad_norm": 0.9140625, "learning_rate": 1.6824393434409092e-05, "loss": 0.1029, "step": 35892 }, { "epoch": 26.295238095238094, "grad_norm": 1.609375, "learning_rate": 1.6821882267755053e-05, "loss": 0.1217, "step": 35893 }, { "epoch": 26.295970695970695, "grad_norm": 0.90234375, "learning_rate": 1.6819371262463947e-05, "loss": 0.1303, "step": 35894 }, { "epoch": 26.296703296703296, "grad_norm": 0.82421875, "learning_rate": 1.681686041854778e-05, "loss": 0.1198, "step": 35895 }, { "epoch": 26.297435897435896, "grad_norm": 1.28125, "learning_rate": 1.681434973601859e-05, "loss": 0.1155, "step": 35896 }, { "epoch": 26.298168498168497, "grad_norm": 1.296875, "learning_rate": 1.6811839214888407e-05, "loss": 0.1073, "step": 35897 }, { "epoch": 26.298901098901098, "grad_norm": 1.109375, "learning_rate": 1.6809328855169263e-05, "loss": 0.1259, "step": 35898 }, { "epoch": 26.2996336996337, "grad_norm": 0.72265625, "learning_rate": 1.6806818656873196e-05, "loss": 0.1182, "step": 35899 }, { "epoch": 26.3003663003663, "grad_norm": 1.109375, "learning_rate": 1.6804308620012213e-05, "loss": 0.1142, "step": 35900 }, { "epoch": 26.3010989010989, "grad_norm": 1.125, "learning_rate": 1.6801798744598353e-05, "loss": 0.0995, "step": 35901 }, { "epoch": 26.3018315018315, "grad_norm": 1.0625, "learning_rate": 1.6799289030643616e-05, "loss": 0.1005, "step": 35902 }, { "epoch": 26.3025641025641, "grad_norm": 1.0078125, "learning_rate": 1.679677947816008e-05, "loss": 0.118, "step": 35903 }, { "epoch": 26.303296703296702, "grad_norm": 1.4765625, "learning_rate": 1.6794270087159718e-05, "loss": 0.143, "step": 35904 }, { "epoch": 26.304029304029303, "grad_norm": 1.0390625, "learning_rate": 1.6791760857654596e-05, "loss": 0.1029, "step": 35905 }, { "epoch": 26.304761904761904, "grad_norm": 0.8984375, "learning_rate": 1.678925178965669e-05, "loss": 0.114, "step": 35906 }, { "epoch": 26.305494505494504, "grad_norm": 1.078125, "learning_rate": 1.678674288317804e-05, "loss": 0.1099, "step": 35907 }, { "epoch": 26.306227106227105, "grad_norm": 1.1015625, "learning_rate": 1.6784234138230685e-05, "loss": 0.126, "step": 35908 }, { "epoch": 26.306959706959706, "grad_norm": 0.96875, "learning_rate": 1.6781725554826642e-05, "loss": 0.1293, "step": 35909 }, { "epoch": 26.307692307692307, "grad_norm": 0.98828125, "learning_rate": 1.6779217132977905e-05, "loss": 0.1196, "step": 35910 }, { "epoch": 26.308424908424907, "grad_norm": 1.53125, "learning_rate": 1.677670887269651e-05, "loss": 0.1198, "step": 35911 }, { "epoch": 26.309157509157508, "grad_norm": 0.90234375, "learning_rate": 1.6774200773994468e-05, "loss": 0.11, "step": 35912 }, { "epoch": 26.30989010989011, "grad_norm": 1.75, "learning_rate": 1.677169283688381e-05, "loss": 0.1226, "step": 35913 }, { "epoch": 26.31062271062271, "grad_norm": 0.84765625, "learning_rate": 1.6769185061376557e-05, "loss": 0.1213, "step": 35914 }, { "epoch": 26.31135531135531, "grad_norm": 0.73046875, "learning_rate": 1.676667744748469e-05, "loss": 0.1061, "step": 35915 }, { "epoch": 26.31208791208791, "grad_norm": 1.140625, "learning_rate": 1.676416999522027e-05, "loss": 0.0959, "step": 35916 }, { "epoch": 26.31282051282051, "grad_norm": 0.80859375, "learning_rate": 1.676166270459525e-05, "loss": 0.1311, "step": 35917 }, { "epoch": 26.313553113553112, "grad_norm": 1.0078125, "learning_rate": 1.675915557562171e-05, "loss": 0.1083, "step": 35918 }, { "epoch": 26.314285714285713, "grad_norm": 0.890625, "learning_rate": 1.6756648608311618e-05, "loss": 0.1339, "step": 35919 }, { "epoch": 26.315018315018314, "grad_norm": 1.109375, "learning_rate": 1.6754141802677018e-05, "loss": 0.1322, "step": 35920 }, { "epoch": 26.315750915750915, "grad_norm": 0.90625, "learning_rate": 1.6751635158729885e-05, "loss": 0.1281, "step": 35921 }, { "epoch": 26.316483516483515, "grad_norm": 1.3671875, "learning_rate": 1.674912867648224e-05, "loss": 0.1203, "step": 35922 }, { "epoch": 26.317216117216116, "grad_norm": 1.1875, "learning_rate": 1.674662235594611e-05, "loss": 0.1171, "step": 35923 }, { "epoch": 26.317948717948717, "grad_norm": 1.0703125, "learning_rate": 1.6744116197133483e-05, "loss": 0.1472, "step": 35924 }, { "epoch": 26.318681318681318, "grad_norm": 1.0625, "learning_rate": 1.6741610200056402e-05, "loss": 0.11, "step": 35925 }, { "epoch": 26.31941391941392, "grad_norm": 0.78515625, "learning_rate": 1.6739104364726825e-05, "loss": 0.1098, "step": 35926 }, { "epoch": 26.32014652014652, "grad_norm": 0.76171875, "learning_rate": 1.67365986911568e-05, "loss": 0.1109, "step": 35927 }, { "epoch": 26.32087912087912, "grad_norm": 1.0546875, "learning_rate": 1.6734093179358282e-05, "loss": 0.1115, "step": 35928 }, { "epoch": 26.32161172161172, "grad_norm": 1.1328125, "learning_rate": 1.6731587829343344e-05, "loss": 0.1384, "step": 35929 }, { "epoch": 26.32234432234432, "grad_norm": 1.0078125, "learning_rate": 1.672908264112394e-05, "loss": 0.1368, "step": 35930 }, { "epoch": 26.323076923076922, "grad_norm": 0.890625, "learning_rate": 1.6726577614712095e-05, "loss": 0.1306, "step": 35931 }, { "epoch": 26.323809523809523, "grad_norm": 1.09375, "learning_rate": 1.6724072750119793e-05, "loss": 0.1269, "step": 35932 }, { "epoch": 26.324542124542123, "grad_norm": 1.234375, "learning_rate": 1.6721568047359044e-05, "loss": 0.1226, "step": 35933 }, { "epoch": 26.325274725274724, "grad_norm": 0.8984375, "learning_rate": 1.6719063506441852e-05, "loss": 0.1197, "step": 35934 }, { "epoch": 26.326007326007325, "grad_norm": 1.0703125, "learning_rate": 1.6716559127380233e-05, "loss": 0.1061, "step": 35935 }, { "epoch": 26.326739926739926, "grad_norm": 1.4140625, "learning_rate": 1.6714054910186152e-05, "loss": 0.1073, "step": 35936 }, { "epoch": 26.327472527472526, "grad_norm": 1.0625, "learning_rate": 1.6711550854871628e-05, "loss": 0.1265, "step": 35937 }, { "epoch": 26.328205128205127, "grad_norm": 1.203125, "learning_rate": 1.670904696144865e-05, "loss": 0.1563, "step": 35938 }, { "epoch": 26.328937728937728, "grad_norm": 0.82421875, "learning_rate": 1.670654322992923e-05, "loss": 0.1233, "step": 35939 }, { "epoch": 26.32967032967033, "grad_norm": 0.78515625, "learning_rate": 1.670403966032537e-05, "loss": 0.1051, "step": 35940 }, { "epoch": 26.33040293040293, "grad_norm": 0.8203125, "learning_rate": 1.6701536252649028e-05, "loss": 0.1234, "step": 35941 }, { "epoch": 26.33113553113553, "grad_norm": 1.1484375, "learning_rate": 1.669903300691224e-05, "loss": 0.1142, "step": 35942 }, { "epoch": 26.33186813186813, "grad_norm": 1.2578125, "learning_rate": 1.6696529923126946e-05, "loss": 0.113, "step": 35943 }, { "epoch": 26.33260073260073, "grad_norm": 0.90234375, "learning_rate": 1.669402700130521e-05, "loss": 0.1051, "step": 35944 }, { "epoch": 26.333333333333332, "grad_norm": 1.234375, "learning_rate": 1.6691524241458976e-05, "loss": 0.139, "step": 35945 }, { "epoch": 26.334065934065933, "grad_norm": 1.2734375, "learning_rate": 1.6689021643600255e-05, "loss": 0.1212, "step": 35946 }, { "epoch": 26.334798534798534, "grad_norm": 1.140625, "learning_rate": 1.6686519207741023e-05, "loss": 0.1349, "step": 35947 }, { "epoch": 26.335531135531134, "grad_norm": 1.171875, "learning_rate": 1.6684016933893268e-05, "loss": 0.1293, "step": 35948 }, { "epoch": 26.336263736263735, "grad_norm": 1.5, "learning_rate": 1.6681514822068997e-05, "loss": 0.1194, "step": 35949 }, { "epoch": 26.336996336996336, "grad_norm": 1.140625, "learning_rate": 1.6679012872280187e-05, "loss": 0.1231, "step": 35950 }, { "epoch": 26.337728937728937, "grad_norm": 0.85546875, "learning_rate": 1.6676511084538847e-05, "loss": 0.1045, "step": 35951 }, { "epoch": 26.338461538461537, "grad_norm": 1.0703125, "learning_rate": 1.667400945885693e-05, "loss": 0.124, "step": 35952 }, { "epoch": 26.339194139194138, "grad_norm": 1.015625, "learning_rate": 1.667150799524645e-05, "loss": 0.0968, "step": 35953 }, { "epoch": 26.33992673992674, "grad_norm": 1.140625, "learning_rate": 1.6669006693719352e-05, "loss": 0.1067, "step": 35954 }, { "epoch": 26.34065934065934, "grad_norm": 1.1484375, "learning_rate": 1.6666505554287682e-05, "loss": 0.1321, "step": 35955 }, { "epoch": 26.34139194139194, "grad_norm": 1.203125, "learning_rate": 1.6664004576963373e-05, "loss": 0.1041, "step": 35956 }, { "epoch": 26.34212454212454, "grad_norm": 0.98828125, "learning_rate": 1.6661503761758442e-05, "loss": 0.1228, "step": 35957 }, { "epoch": 26.34285714285714, "grad_norm": 0.93359375, "learning_rate": 1.6659003108684836e-05, "loss": 0.1294, "step": 35958 }, { "epoch": 26.343589743589742, "grad_norm": 1.1875, "learning_rate": 1.665650261775456e-05, "loss": 0.1087, "step": 35959 }, { "epoch": 26.344322344322343, "grad_norm": 1.1015625, "learning_rate": 1.665400228897959e-05, "loss": 0.1055, "step": 35960 }, { "epoch": 26.345054945054944, "grad_norm": 1.1484375, "learning_rate": 1.66515021223719e-05, "loss": 0.1134, "step": 35961 }, { "epoch": 26.345787545787545, "grad_norm": 0.9609375, "learning_rate": 1.6649002117943495e-05, "loss": 0.1155, "step": 35962 }, { "epoch": 26.346520146520145, "grad_norm": 1.15625, "learning_rate": 1.664650227570632e-05, "loss": 0.1346, "step": 35963 }, { "epoch": 26.347252747252746, "grad_norm": 1.1015625, "learning_rate": 1.6644002595672363e-05, "loss": 0.1, "step": 35964 }, { "epoch": 26.347985347985347, "grad_norm": 0.93359375, "learning_rate": 1.6641503077853605e-05, "loss": 0.1069, "step": 35965 }, { "epoch": 26.348717948717947, "grad_norm": 1.0703125, "learning_rate": 1.6639003722262036e-05, "loss": 0.1279, "step": 35966 }, { "epoch": 26.349450549450548, "grad_norm": 0.94140625, "learning_rate": 1.66365045289096e-05, "loss": 0.1174, "step": 35967 }, { "epoch": 26.35018315018315, "grad_norm": 0.83984375, "learning_rate": 1.6634005497808304e-05, "loss": 0.1168, "step": 35968 }, { "epoch": 26.35091575091575, "grad_norm": 0.921875, "learning_rate": 1.6631506628970073e-05, "loss": 0.1145, "step": 35969 }, { "epoch": 26.35164835164835, "grad_norm": 1.3125, "learning_rate": 1.6629007922406942e-05, "loss": 0.0965, "step": 35970 }, { "epoch": 26.35238095238095, "grad_norm": 1.265625, "learning_rate": 1.6626509378130838e-05, "loss": 0.1398, "step": 35971 }, { "epoch": 26.353113553113552, "grad_norm": 1.078125, "learning_rate": 1.6624010996153768e-05, "loss": 0.1135, "step": 35972 }, { "epoch": 26.353846153846153, "grad_norm": 0.828125, "learning_rate": 1.6621512776487662e-05, "loss": 0.1102, "step": 35973 }, { "epoch": 26.354578754578753, "grad_norm": 0.92578125, "learning_rate": 1.661901471914451e-05, "loss": 0.091, "step": 35974 }, { "epoch": 26.355311355311354, "grad_norm": 1.0546875, "learning_rate": 1.6616516824136284e-05, "loss": 0.1626, "step": 35975 }, { "epoch": 26.356043956043955, "grad_norm": 1.015625, "learning_rate": 1.6614019091474943e-05, "loss": 0.1258, "step": 35976 }, { "epoch": 26.356776556776556, "grad_norm": 0.98828125, "learning_rate": 1.6611521521172477e-05, "loss": 0.1085, "step": 35977 }, { "epoch": 26.357509157509156, "grad_norm": 0.703125, "learning_rate": 1.660902411324083e-05, "loss": 0.1222, "step": 35978 }, { "epoch": 26.358241758241757, "grad_norm": 1.578125, "learning_rate": 1.6606526867691978e-05, "loss": 0.1269, "step": 35979 }, { "epoch": 26.358974358974358, "grad_norm": 1.0625, "learning_rate": 1.6604029784537855e-05, "loss": 0.1106, "step": 35980 }, { "epoch": 26.35970695970696, "grad_norm": 1.265625, "learning_rate": 1.6601532863790486e-05, "loss": 0.1231, "step": 35981 }, { "epoch": 26.36043956043956, "grad_norm": 0.6875, "learning_rate": 1.6599036105461785e-05, "loss": 0.1306, "step": 35982 }, { "epoch": 26.36117216117216, "grad_norm": 1.6640625, "learning_rate": 1.6596539509563745e-05, "loss": 0.1396, "step": 35983 }, { "epoch": 26.36190476190476, "grad_norm": 0.87890625, "learning_rate": 1.6594043076108297e-05, "loss": 0.1109, "step": 35984 }, { "epoch": 26.36263736263736, "grad_norm": 0.68359375, "learning_rate": 1.6591546805107423e-05, "loss": 0.1211, "step": 35985 }, { "epoch": 26.363369963369962, "grad_norm": 1.0234375, "learning_rate": 1.658905069657308e-05, "loss": 0.1523, "step": 35986 }, { "epoch": 26.364102564102563, "grad_norm": 1.2421875, "learning_rate": 1.6586554750517228e-05, "loss": 0.0996, "step": 35987 }, { "epoch": 26.364835164835164, "grad_norm": 1.40625, "learning_rate": 1.6584058966951838e-05, "loss": 0.1254, "step": 35988 }, { "epoch": 26.365567765567764, "grad_norm": 1.1328125, "learning_rate": 1.6581563345888838e-05, "loss": 0.127, "step": 35989 }, { "epoch": 26.366300366300365, "grad_norm": 0.98828125, "learning_rate": 1.6579067887340204e-05, "loss": 0.1, "step": 35990 }, { "epoch": 26.367032967032966, "grad_norm": 1.1796875, "learning_rate": 1.6576572591317895e-05, "loss": 0.1473, "step": 35991 }, { "epoch": 26.367765567765566, "grad_norm": 0.7265625, "learning_rate": 1.657407745783388e-05, "loss": 0.1068, "step": 35992 }, { "epoch": 26.368498168498167, "grad_norm": 1.859375, "learning_rate": 1.657158248690007e-05, "loss": 0.1379, "step": 35993 }, { "epoch": 26.369230769230768, "grad_norm": 1.1015625, "learning_rate": 1.656908767852847e-05, "loss": 0.1045, "step": 35994 }, { "epoch": 26.36996336996337, "grad_norm": 1.40625, "learning_rate": 1.656659303273098e-05, "loss": 0.1399, "step": 35995 }, { "epoch": 26.37069597069597, "grad_norm": 1.78125, "learning_rate": 1.656409854951961e-05, "loss": 0.138, "step": 35996 }, { "epoch": 26.37142857142857, "grad_norm": 0.8203125, "learning_rate": 1.6561604228906268e-05, "loss": 0.1358, "step": 35997 }, { "epoch": 26.37216117216117, "grad_norm": 0.671875, "learning_rate": 1.655911007090294e-05, "loss": 0.1133, "step": 35998 }, { "epoch": 26.37289377289377, "grad_norm": 1.0546875, "learning_rate": 1.6556616075521543e-05, "loss": 0.1485, "step": 35999 }, { "epoch": 26.373626373626372, "grad_norm": 1.6640625, "learning_rate": 1.655412224277404e-05, "loss": 0.1599, "step": 36000 }, { "epoch": 26.374358974358973, "grad_norm": 1.3671875, "learning_rate": 1.655162857267238e-05, "loss": 0.1373, "step": 36001 }, { "epoch": 26.375091575091574, "grad_norm": 1.359375, "learning_rate": 1.654913506522851e-05, "loss": 0.1076, "step": 36002 }, { "epoch": 26.375824175824174, "grad_norm": 0.71484375, "learning_rate": 1.65466417204544e-05, "loss": 0.1174, "step": 36003 }, { "epoch": 26.376556776556775, "grad_norm": 1.3125, "learning_rate": 1.6544148538361955e-05, "loss": 0.1564, "step": 36004 }, { "epoch": 26.377289377289376, "grad_norm": 0.8125, "learning_rate": 1.654165551896316e-05, "loss": 0.0971, "step": 36005 }, { "epoch": 26.378021978021977, "grad_norm": 0.99609375, "learning_rate": 1.653916266226991e-05, "loss": 0.1448, "step": 36006 }, { "epoch": 26.378754578754577, "grad_norm": 0.953125, "learning_rate": 1.653666996829421e-05, "loss": 0.1232, "step": 36007 }, { "epoch": 26.379487179487178, "grad_norm": 1.25, "learning_rate": 1.653417743704796e-05, "loss": 0.1379, "step": 36008 }, { "epoch": 26.38021978021978, "grad_norm": 0.85546875, "learning_rate": 1.653168506854313e-05, "loss": 0.1147, "step": 36009 }, { "epoch": 26.38095238095238, "grad_norm": 0.8359375, "learning_rate": 1.6529192862791635e-05, "loss": 0.0998, "step": 36010 }, { "epoch": 26.38168498168498, "grad_norm": 0.85546875, "learning_rate": 1.6526700819805424e-05, "loss": 0.0891, "step": 36011 }, { "epoch": 26.38241758241758, "grad_norm": 1.1015625, "learning_rate": 1.6524208939596445e-05, "loss": 0.1112, "step": 36012 }, { "epoch": 26.383150183150182, "grad_norm": 0.99609375, "learning_rate": 1.652171722217663e-05, "loss": 0.116, "step": 36013 }, { "epoch": 26.383882783882783, "grad_norm": 1.0078125, "learning_rate": 1.6519225667557946e-05, "loss": 0.1388, "step": 36014 }, { "epoch": 26.384615384615383, "grad_norm": 1.3125, "learning_rate": 1.6516734275752288e-05, "loss": 0.1253, "step": 36015 }, { "epoch": 26.385347985347984, "grad_norm": 0.84375, "learning_rate": 1.6514243046771614e-05, "loss": 0.1384, "step": 36016 }, { "epoch": 26.386080586080585, "grad_norm": 0.69140625, "learning_rate": 1.6511751980627854e-05, "loss": 0.0894, "step": 36017 }, { "epoch": 26.386813186813185, "grad_norm": 1.4921875, "learning_rate": 1.6509261077332964e-05, "loss": 0.1307, "step": 36018 }, { "epoch": 26.387545787545786, "grad_norm": 1.0, "learning_rate": 1.650677033689885e-05, "loss": 0.124, "step": 36019 }, { "epoch": 26.388278388278387, "grad_norm": 1.484375, "learning_rate": 1.650427975933747e-05, "loss": 0.1193, "step": 36020 }, { "epoch": 26.389010989010988, "grad_norm": 1.046875, "learning_rate": 1.650178934466072e-05, "loss": 0.1321, "step": 36021 }, { "epoch": 26.38974358974359, "grad_norm": 1.359375, "learning_rate": 1.649929909288058e-05, "loss": 0.1213, "step": 36022 }, { "epoch": 26.39047619047619, "grad_norm": 1.6328125, "learning_rate": 1.6496809004008955e-05, "loss": 0.1157, "step": 36023 }, { "epoch": 26.39120879120879, "grad_norm": 0.984375, "learning_rate": 1.6494319078057794e-05, "loss": 0.1306, "step": 36024 }, { "epoch": 26.39194139194139, "grad_norm": 1.828125, "learning_rate": 1.6491829315038996e-05, "loss": 0.1584, "step": 36025 }, { "epoch": 26.39267399267399, "grad_norm": 1.5078125, "learning_rate": 1.6489339714964507e-05, "loss": 0.1321, "step": 36026 }, { "epoch": 26.393406593406592, "grad_norm": 1.0546875, "learning_rate": 1.6486850277846254e-05, "loss": 0.1223, "step": 36027 }, { "epoch": 26.394139194139193, "grad_norm": 0.94140625, "learning_rate": 1.648436100369617e-05, "loss": 0.1271, "step": 36028 }, { "epoch": 26.394871794871793, "grad_norm": 1.4296875, "learning_rate": 1.6481871892526192e-05, "loss": 0.1044, "step": 36029 }, { "epoch": 26.395604395604394, "grad_norm": 0.6875, "learning_rate": 1.6479382944348225e-05, "loss": 0.1022, "step": 36030 }, { "epoch": 26.396336996336995, "grad_norm": 0.796875, "learning_rate": 1.6476894159174216e-05, "loss": 0.1355, "step": 36031 }, { "epoch": 26.397069597069596, "grad_norm": 1.1015625, "learning_rate": 1.647440553701604e-05, "loss": 0.1371, "step": 36032 }, { "epoch": 26.397802197802196, "grad_norm": 1.046875, "learning_rate": 1.647191707788569e-05, "loss": 0.0841, "step": 36033 }, { "epoch": 26.398534798534797, "grad_norm": 1.40625, "learning_rate": 1.646942878179504e-05, "loss": 0.1325, "step": 36034 }, { "epoch": 26.399267399267398, "grad_norm": 0.69921875, "learning_rate": 1.6466940648756045e-05, "loss": 0.1138, "step": 36035 }, { "epoch": 26.4, "grad_norm": 0.96484375, "learning_rate": 1.6464452678780593e-05, "loss": 0.0926, "step": 36036 }, { "epoch": 26.4007326007326, "grad_norm": 1.2109375, "learning_rate": 1.6461964871880626e-05, "loss": 0.1456, "step": 36037 }, { "epoch": 26.4014652014652, "grad_norm": 0.59765625, "learning_rate": 1.645947722806805e-05, "loss": 0.1069, "step": 36038 }, { "epoch": 26.4021978021978, "grad_norm": 0.8203125, "learning_rate": 1.6456989747354803e-05, "loss": 0.1154, "step": 36039 }, { "epoch": 26.4029304029304, "grad_norm": 1.109375, "learning_rate": 1.6454502429752803e-05, "loss": 0.1184, "step": 36040 }, { "epoch": 26.403663003663002, "grad_norm": 0.7421875, "learning_rate": 1.6452015275273946e-05, "loss": 0.1258, "step": 36041 }, { "epoch": 26.404395604395603, "grad_norm": 1.4765625, "learning_rate": 1.6449528283930156e-05, "loss": 0.0988, "step": 36042 }, { "epoch": 26.405128205128204, "grad_norm": 1.0078125, "learning_rate": 1.6447041455733357e-05, "loss": 0.1212, "step": 36043 }, { "epoch": 26.405860805860804, "grad_norm": 1.59375, "learning_rate": 1.644455479069548e-05, "loss": 0.15, "step": 36044 }, { "epoch": 26.406593406593405, "grad_norm": 0.70703125, "learning_rate": 1.6442068288828405e-05, "loss": 0.104, "step": 36045 }, { "epoch": 26.407326007326006, "grad_norm": 0.7734375, "learning_rate": 1.6439581950144073e-05, "loss": 0.0977, "step": 36046 }, { "epoch": 26.408058608058607, "grad_norm": 0.9296875, "learning_rate": 1.6437095774654357e-05, "loss": 0.1281, "step": 36047 }, { "epoch": 26.408791208791207, "grad_norm": 1.15625, "learning_rate": 1.6434609762371226e-05, "loss": 0.109, "step": 36048 }, { "epoch": 26.409523809523808, "grad_norm": 1.0234375, "learning_rate": 1.6432123913306548e-05, "loss": 0.126, "step": 36049 }, { "epoch": 26.41025641025641, "grad_norm": 0.97265625, "learning_rate": 1.6429638227472263e-05, "loss": 0.1148, "step": 36050 }, { "epoch": 26.41098901098901, "grad_norm": 1.0234375, "learning_rate": 1.6427152704880252e-05, "loss": 0.1062, "step": 36051 }, { "epoch": 26.41172161172161, "grad_norm": 0.66015625, "learning_rate": 1.6424667345542437e-05, "loss": 0.1006, "step": 36052 }, { "epoch": 26.41245421245421, "grad_norm": 1.4609375, "learning_rate": 1.6422182149470725e-05, "loss": 0.1128, "step": 36053 }, { "epoch": 26.41318681318681, "grad_norm": 1.234375, "learning_rate": 1.641969711667703e-05, "loss": 0.0974, "step": 36054 }, { "epoch": 26.413919413919412, "grad_norm": 1.84375, "learning_rate": 1.6417212247173266e-05, "loss": 0.1542, "step": 36055 }, { "epoch": 26.414652014652013, "grad_norm": 1.3984375, "learning_rate": 1.641472754097131e-05, "loss": 0.1476, "step": 36056 }, { "epoch": 26.415384615384614, "grad_norm": 0.7578125, "learning_rate": 1.6412242998083106e-05, "loss": 0.1038, "step": 36057 }, { "epoch": 26.416117216117215, "grad_norm": 1.640625, "learning_rate": 1.64097586185205e-05, "loss": 0.122, "step": 36058 }, { "epoch": 26.416849816849815, "grad_norm": 0.97265625, "learning_rate": 1.640727440229546e-05, "loss": 0.0992, "step": 36059 }, { "epoch": 26.417582417582416, "grad_norm": 1.0703125, "learning_rate": 1.6404790349419846e-05, "loss": 0.1207, "step": 36060 }, { "epoch": 26.418315018315017, "grad_norm": 1.125, "learning_rate": 1.6402306459905595e-05, "loss": 0.0966, "step": 36061 }, { "epoch": 26.419047619047618, "grad_norm": 1.203125, "learning_rate": 1.6399822733764563e-05, "loss": 0.1115, "step": 36062 }, { "epoch": 26.41978021978022, "grad_norm": 1.0078125, "learning_rate": 1.639733917100868e-05, "loss": 0.1263, "step": 36063 }, { "epoch": 26.42051282051282, "grad_norm": 1.3359375, "learning_rate": 1.6394855771649842e-05, "loss": 0.147, "step": 36064 }, { "epoch": 26.42124542124542, "grad_norm": 0.96484375, "learning_rate": 1.6392372535699946e-05, "loss": 0.1294, "step": 36065 }, { "epoch": 26.42197802197802, "grad_norm": 1.4921875, "learning_rate": 1.63898894631709e-05, "loss": 0.1431, "step": 36066 }, { "epoch": 26.42271062271062, "grad_norm": 0.8671875, "learning_rate": 1.638740655407458e-05, "loss": 0.1197, "step": 36067 }, { "epoch": 26.423443223443222, "grad_norm": 1.0703125, "learning_rate": 1.6384923808422894e-05, "loss": 0.1205, "step": 36068 }, { "epoch": 26.424175824175823, "grad_norm": 0.96484375, "learning_rate": 1.6382441226227734e-05, "loss": 0.1177, "step": 36069 }, { "epoch": 26.424908424908423, "grad_norm": 1.078125, "learning_rate": 1.6379958807501016e-05, "loss": 0.1034, "step": 36070 }, { "epoch": 26.425641025641024, "grad_norm": 0.7734375, "learning_rate": 1.6377476552254597e-05, "loss": 0.0926, "step": 36071 }, { "epoch": 26.426373626373625, "grad_norm": 1.4296875, "learning_rate": 1.637499446050041e-05, "loss": 0.1068, "step": 36072 }, { "epoch": 26.427106227106226, "grad_norm": 1.4609375, "learning_rate": 1.6372512532250295e-05, "loss": 0.1436, "step": 36073 }, { "epoch": 26.427838827838826, "grad_norm": 0.63671875, "learning_rate": 1.6370030767516208e-05, "loss": 0.1079, "step": 36074 }, { "epoch": 26.428571428571427, "grad_norm": 0.60546875, "learning_rate": 1.636754916630999e-05, "loss": 0.0962, "step": 36075 }, { "epoch": 26.429304029304028, "grad_norm": 1.328125, "learning_rate": 1.6365067728643545e-05, "loss": 0.109, "step": 36076 }, { "epoch": 26.43003663003663, "grad_norm": 1.484375, "learning_rate": 1.6362586454528786e-05, "loss": 0.1256, "step": 36077 }, { "epoch": 26.43076923076923, "grad_norm": 0.87109375, "learning_rate": 1.636010534397757e-05, "loss": 0.1175, "step": 36078 }, { "epoch": 26.43150183150183, "grad_norm": 1.15625, "learning_rate": 1.635762439700179e-05, "loss": 0.1344, "step": 36079 }, { "epoch": 26.43223443223443, "grad_norm": 0.7734375, "learning_rate": 1.6355143613613347e-05, "loss": 0.1242, "step": 36080 }, { "epoch": 26.43296703296703, "grad_norm": 1.2890625, "learning_rate": 1.635266299382413e-05, "loss": 0.1279, "step": 36081 }, { "epoch": 26.433699633699632, "grad_norm": 1.140625, "learning_rate": 1.6350182537646e-05, "loss": 0.1107, "step": 36082 }, { "epoch": 26.434432234432233, "grad_norm": 0.83203125, "learning_rate": 1.6347702245090872e-05, "loss": 0.1322, "step": 36083 }, { "epoch": 26.435164835164834, "grad_norm": 1.3515625, "learning_rate": 1.634522211617058e-05, "loss": 0.1282, "step": 36084 }, { "epoch": 26.435897435897434, "grad_norm": 0.984375, "learning_rate": 1.6342742150897077e-05, "loss": 0.1232, "step": 36085 }, { "epoch": 26.436630036630035, "grad_norm": 1.546875, "learning_rate": 1.6340262349282194e-05, "loss": 0.1711, "step": 36086 }, { "epoch": 26.437362637362636, "grad_norm": 0.89453125, "learning_rate": 1.633778271133784e-05, "loss": 0.111, "step": 36087 }, { "epoch": 26.438095238095237, "grad_norm": 1.1796875, "learning_rate": 1.633530323707587e-05, "loss": 0.1527, "step": 36088 }, { "epoch": 26.438827838827837, "grad_norm": 1.2734375, "learning_rate": 1.6332823926508175e-05, "loss": 0.1321, "step": 36089 }, { "epoch": 26.439560439560438, "grad_norm": 0.8046875, "learning_rate": 1.633034477964664e-05, "loss": 0.1203, "step": 36090 }, { "epoch": 26.44029304029304, "grad_norm": 1.1484375, "learning_rate": 1.632786579650314e-05, "loss": 0.11, "step": 36091 }, { "epoch": 26.44102564102564, "grad_norm": 1.1796875, "learning_rate": 1.6325386977089567e-05, "loss": 0.1487, "step": 36092 }, { "epoch": 26.44175824175824, "grad_norm": 0.88671875, "learning_rate": 1.6322908321417773e-05, "loss": 0.0973, "step": 36093 }, { "epoch": 26.44249084249084, "grad_norm": 1.1875, "learning_rate": 1.632042982949964e-05, "loss": 0.1356, "step": 36094 }, { "epoch": 26.44322344322344, "grad_norm": 0.98046875, "learning_rate": 1.6317951501347046e-05, "loss": 0.1517, "step": 36095 }, { "epoch": 26.443956043956042, "grad_norm": 1.109375, "learning_rate": 1.6315473336971888e-05, "loss": 0.1046, "step": 36096 }, { "epoch": 26.444688644688643, "grad_norm": 0.76171875, "learning_rate": 1.6312995336386003e-05, "loss": 0.1113, "step": 36097 }, { "epoch": 26.445421245421244, "grad_norm": 0.7890625, "learning_rate": 1.6310517499601295e-05, "loss": 0.0918, "step": 36098 }, { "epoch": 26.446153846153845, "grad_norm": 1.1953125, "learning_rate": 1.630803982662959e-05, "loss": 0.1121, "step": 36099 }, { "epoch": 26.446886446886445, "grad_norm": 1.1953125, "learning_rate": 1.630556231748282e-05, "loss": 0.1239, "step": 36100 }, { "epoch": 26.447619047619046, "grad_norm": 1.265625, "learning_rate": 1.6303084972172815e-05, "loss": 0.1363, "step": 36101 }, { "epoch": 26.448351648351647, "grad_norm": 1.1796875, "learning_rate": 1.6300607790711453e-05, "loss": 0.124, "step": 36102 }, { "epoch": 26.449084249084247, "grad_norm": 1.0546875, "learning_rate": 1.629813077311062e-05, "loss": 0.1384, "step": 36103 }, { "epoch": 26.449816849816848, "grad_norm": 1.078125, "learning_rate": 1.6295653919382157e-05, "loss": 0.1022, "step": 36104 }, { "epoch": 26.45054945054945, "grad_norm": 1.203125, "learning_rate": 1.6293177229537945e-05, "loss": 0.1074, "step": 36105 }, { "epoch": 26.45128205128205, "grad_norm": 1.671875, "learning_rate": 1.629070070358985e-05, "loss": 0.1435, "step": 36106 }, { "epoch": 26.45201465201465, "grad_norm": 0.703125, "learning_rate": 1.6288224341549753e-05, "loss": 0.1124, "step": 36107 }, { "epoch": 26.45274725274725, "grad_norm": 1.2578125, "learning_rate": 1.6285748143429492e-05, "loss": 0.1526, "step": 36108 }, { "epoch": 26.453479853479852, "grad_norm": 1.296875, "learning_rate": 1.6283272109240955e-05, "loss": 0.1103, "step": 36109 }, { "epoch": 26.454212454212453, "grad_norm": 1.2890625, "learning_rate": 1.6280796238995964e-05, "loss": 0.1282, "step": 36110 }, { "epoch": 26.454945054945053, "grad_norm": 1.140625, "learning_rate": 1.627832053270644e-05, "loss": 0.0974, "step": 36111 }, { "epoch": 26.455677655677654, "grad_norm": 0.91796875, "learning_rate": 1.6275844990384205e-05, "loss": 0.1062, "step": 36112 }, { "epoch": 26.456410256410255, "grad_norm": 0.86328125, "learning_rate": 1.6273369612041143e-05, "loss": 0.1245, "step": 36113 }, { "epoch": 26.457142857142856, "grad_norm": 0.875, "learning_rate": 1.6270894397689095e-05, "loss": 0.1061, "step": 36114 }, { "epoch": 26.457875457875456, "grad_norm": 0.8359375, "learning_rate": 1.626841934733992e-05, "loss": 0.0935, "step": 36115 }, { "epoch": 26.458608058608057, "grad_norm": 0.76953125, "learning_rate": 1.6265944461005495e-05, "loss": 0.1228, "step": 36116 }, { "epoch": 26.459340659340658, "grad_norm": 1.5703125, "learning_rate": 1.626346973869766e-05, "loss": 0.1281, "step": 36117 }, { "epoch": 26.46007326007326, "grad_norm": 1.609375, "learning_rate": 1.62609951804283e-05, "loss": 0.1435, "step": 36118 }, { "epoch": 26.46080586080586, "grad_norm": 1.234375, "learning_rate": 1.625852078620923e-05, "loss": 0.1221, "step": 36119 }, { "epoch": 26.46153846153846, "grad_norm": 0.7265625, "learning_rate": 1.6256046556052333e-05, "loss": 0.1094, "step": 36120 }, { "epoch": 26.46227106227106, "grad_norm": 0.6171875, "learning_rate": 1.625357248996946e-05, "loss": 0.1165, "step": 36121 }, { "epoch": 26.46300366300366, "grad_norm": 0.91015625, "learning_rate": 1.6251098587972476e-05, "loss": 0.1374, "step": 36122 }, { "epoch": 26.463736263736262, "grad_norm": 0.90625, "learning_rate": 1.624862485007321e-05, "loss": 0.1226, "step": 36123 }, { "epoch": 26.464468864468863, "grad_norm": 1.4453125, "learning_rate": 1.624615127628354e-05, "loss": 0.1507, "step": 36124 }, { "epoch": 26.465201465201464, "grad_norm": 0.78515625, "learning_rate": 1.6243677866615287e-05, "loss": 0.1086, "step": 36125 }, { "epoch": 26.465934065934064, "grad_norm": 1.34375, "learning_rate": 1.6241204621080323e-05, "loss": 0.1197, "step": 36126 }, { "epoch": 26.466666666666665, "grad_norm": 0.984375, "learning_rate": 1.6238731539690487e-05, "loss": 0.1168, "step": 36127 }, { "epoch": 26.467399267399266, "grad_norm": 0.859375, "learning_rate": 1.6236258622457646e-05, "loss": 0.1201, "step": 36128 }, { "epoch": 26.468131868131866, "grad_norm": 0.95703125, "learning_rate": 1.6233785869393644e-05, "loss": 0.1134, "step": 36129 }, { "epoch": 26.468864468864467, "grad_norm": 0.84375, "learning_rate": 1.623131328051031e-05, "loss": 0.1221, "step": 36130 }, { "epoch": 26.469597069597068, "grad_norm": 0.8671875, "learning_rate": 1.6228840855819504e-05, "loss": 0.1494, "step": 36131 }, { "epoch": 26.47032967032967, "grad_norm": 0.94140625, "learning_rate": 1.622636859533308e-05, "loss": 0.1297, "step": 36132 }, { "epoch": 26.47106227106227, "grad_norm": 1.078125, "learning_rate": 1.6223896499062876e-05, "loss": 0.1293, "step": 36133 }, { "epoch": 26.47179487179487, "grad_norm": 1.2109375, "learning_rate": 1.6221424567020727e-05, "loss": 0.142, "step": 36134 }, { "epoch": 26.47252747252747, "grad_norm": 1.2890625, "learning_rate": 1.62189527992185e-05, "loss": 0.1149, "step": 36135 }, { "epoch": 26.47326007326007, "grad_norm": 1.2734375, "learning_rate": 1.6216481195668e-05, "loss": 0.1153, "step": 36136 }, { "epoch": 26.473992673992672, "grad_norm": 0.62890625, "learning_rate": 1.6214009756381115e-05, "loss": 0.1207, "step": 36137 }, { "epoch": 26.474725274725273, "grad_norm": 1.8984375, "learning_rate": 1.6211538481369654e-05, "loss": 0.1637, "step": 36138 }, { "epoch": 26.475457875457874, "grad_norm": 0.95703125, "learning_rate": 1.6209067370645484e-05, "loss": 0.1169, "step": 36139 }, { "epoch": 26.476190476190474, "grad_norm": 1.21875, "learning_rate": 1.620659642422041e-05, "loss": 0.126, "step": 36140 }, { "epoch": 26.476923076923075, "grad_norm": 1.4765625, "learning_rate": 1.6204125642106297e-05, "loss": 0.1615, "step": 36141 }, { "epoch": 26.477655677655676, "grad_norm": 0.8203125, "learning_rate": 1.6201655024314973e-05, "loss": 0.1031, "step": 36142 }, { "epoch": 26.478388278388277, "grad_norm": 1.1015625, "learning_rate": 1.6199184570858277e-05, "loss": 0.1194, "step": 36143 }, { "epoch": 26.479120879120877, "grad_norm": 0.79296875, "learning_rate": 1.6196714281748063e-05, "loss": 0.0982, "step": 36144 }, { "epoch": 26.479853479853478, "grad_norm": 0.97265625, "learning_rate": 1.6194244156996142e-05, "loss": 0.1449, "step": 36145 }, { "epoch": 26.48058608058608, "grad_norm": 0.88671875, "learning_rate": 1.6191774196614354e-05, "loss": 0.126, "step": 36146 }, { "epoch": 26.48131868131868, "grad_norm": 0.89453125, "learning_rate": 1.6189304400614538e-05, "loss": 0.0919, "step": 36147 }, { "epoch": 26.48205128205128, "grad_norm": 0.875, "learning_rate": 1.6186834769008547e-05, "loss": 0.1181, "step": 36148 }, { "epoch": 26.48278388278388, "grad_norm": 1.0234375, "learning_rate": 1.618436530180818e-05, "loss": 0.1087, "step": 36149 }, { "epoch": 26.483516483516482, "grad_norm": 1.0, "learning_rate": 1.6181895999025296e-05, "loss": 0.1345, "step": 36150 }, { "epoch": 26.484249084249083, "grad_norm": 0.9453125, "learning_rate": 1.61794268606717e-05, "loss": 0.1219, "step": 36151 }, { "epoch": 26.484981684981683, "grad_norm": 0.65625, "learning_rate": 1.6176957886759236e-05, "loss": 0.1002, "step": 36152 }, { "epoch": 26.485714285714284, "grad_norm": 0.83984375, "learning_rate": 1.6174489077299738e-05, "loss": 0.1058, "step": 36153 }, { "epoch": 26.486446886446885, "grad_norm": 0.76171875, "learning_rate": 1.617202043230503e-05, "loss": 0.1354, "step": 36154 }, { "epoch": 26.487179487179485, "grad_norm": 1.875, "learning_rate": 1.6169551951786953e-05, "loss": 0.1259, "step": 36155 }, { "epoch": 26.487912087912086, "grad_norm": 1.2734375, "learning_rate": 1.616708363575731e-05, "loss": 0.1131, "step": 36156 }, { "epoch": 26.488644688644687, "grad_norm": 1.0234375, "learning_rate": 1.616461548422794e-05, "loss": 0.1078, "step": 36157 }, { "epoch": 26.489377289377288, "grad_norm": 1.3203125, "learning_rate": 1.6162147497210665e-05, "loss": 0.1127, "step": 36158 }, { "epoch": 26.49010989010989, "grad_norm": 0.703125, "learning_rate": 1.615967967471733e-05, "loss": 0.1139, "step": 36159 }, { "epoch": 26.49084249084249, "grad_norm": 0.68359375, "learning_rate": 1.615721201675973e-05, "loss": 0.11, "step": 36160 }, { "epoch": 26.49157509157509, "grad_norm": 1.25, "learning_rate": 1.6154744523349715e-05, "loss": 0.1596, "step": 36161 }, { "epoch": 26.49230769230769, "grad_norm": 0.7734375, "learning_rate": 1.6152277194499065e-05, "loss": 0.1208, "step": 36162 }, { "epoch": 26.493040293040295, "grad_norm": 1.1953125, "learning_rate": 1.614981003021966e-05, "loss": 0.1188, "step": 36163 }, { "epoch": 26.493772893772892, "grad_norm": 0.78515625, "learning_rate": 1.6147343030523277e-05, "loss": 0.1036, "step": 36164 }, { "epoch": 26.494505494505496, "grad_norm": 0.93359375, "learning_rate": 1.614487619542177e-05, "loss": 0.107, "step": 36165 }, { "epoch": 26.495238095238093, "grad_norm": 1.359375, "learning_rate": 1.6142409524926915e-05, "loss": 0.1257, "step": 36166 }, { "epoch": 26.495970695970698, "grad_norm": 0.72265625, "learning_rate": 1.6139943019050556e-05, "loss": 0.1085, "step": 36167 }, { "epoch": 26.496703296703295, "grad_norm": 1.4453125, "learning_rate": 1.6137476677804516e-05, "loss": 0.1199, "step": 36168 }, { "epoch": 26.4974358974359, "grad_norm": 0.9296875, "learning_rate": 1.61350105012006e-05, "loss": 0.1047, "step": 36169 }, { "epoch": 26.498168498168496, "grad_norm": 1.296875, "learning_rate": 1.6132544489250654e-05, "loss": 0.0977, "step": 36170 }, { "epoch": 26.4989010989011, "grad_norm": 1.0703125, "learning_rate": 1.6130078641966447e-05, "loss": 0.1191, "step": 36171 }, { "epoch": 26.499633699633698, "grad_norm": 1.171875, "learning_rate": 1.6127612959359817e-05, "loss": 0.1177, "step": 36172 }, { "epoch": 26.500366300366302, "grad_norm": 0.74609375, "learning_rate": 1.612514744144258e-05, "loss": 0.1344, "step": 36173 }, { "epoch": 26.5010989010989, "grad_norm": 1.828125, "learning_rate": 1.6122682088226553e-05, "loss": 0.113, "step": 36174 }, { "epoch": 26.501831501831504, "grad_norm": 1.234375, "learning_rate": 1.6120216899723532e-05, "loss": 0.1147, "step": 36175 }, { "epoch": 26.5025641025641, "grad_norm": 0.8828125, "learning_rate": 1.611775187594536e-05, "loss": 0.1412, "step": 36176 }, { "epoch": 26.503296703296705, "grad_norm": 1.1796875, "learning_rate": 1.6115287016903802e-05, "loss": 0.111, "step": 36177 }, { "epoch": 26.504029304029302, "grad_norm": 1.375, "learning_rate": 1.611282232261069e-05, "loss": 0.1166, "step": 36178 }, { "epoch": 26.504761904761907, "grad_norm": 1.0625, "learning_rate": 1.611035779307784e-05, "loss": 0.158, "step": 36179 }, { "epoch": 26.505494505494504, "grad_norm": 0.875, "learning_rate": 1.610789342831705e-05, "loss": 0.1073, "step": 36180 }, { "epoch": 26.506227106227108, "grad_norm": 0.9609375, "learning_rate": 1.6105429228340156e-05, "loss": 0.0948, "step": 36181 }, { "epoch": 26.506959706959705, "grad_norm": 0.75, "learning_rate": 1.6102965193158924e-05, "loss": 0.1264, "step": 36182 }, { "epoch": 26.50769230769231, "grad_norm": 1.0625, "learning_rate": 1.6100501322785173e-05, "loss": 0.0966, "step": 36183 }, { "epoch": 26.508424908424907, "grad_norm": 1.0078125, "learning_rate": 1.6098037617230712e-05, "loss": 0.1193, "step": 36184 }, { "epoch": 26.50915750915751, "grad_norm": 1.1484375, "learning_rate": 1.6095574076507368e-05, "loss": 0.1228, "step": 36185 }, { "epoch": 26.509890109890108, "grad_norm": 0.8515625, "learning_rate": 1.6093110700626905e-05, "loss": 0.1116, "step": 36186 }, { "epoch": 26.510622710622712, "grad_norm": 1.0859375, "learning_rate": 1.6090647489601157e-05, "loss": 0.1416, "step": 36187 }, { "epoch": 26.51135531135531, "grad_norm": 1.1640625, "learning_rate": 1.608818444344188e-05, "loss": 0.0938, "step": 36188 }, { "epoch": 26.512087912087914, "grad_norm": 0.83984375, "learning_rate": 1.6085721562160946e-05, "loss": 0.1331, "step": 36189 }, { "epoch": 26.51282051282051, "grad_norm": 0.921875, "learning_rate": 1.608325884577009e-05, "loss": 0.133, "step": 36190 }, { "epoch": 26.513553113553115, "grad_norm": 1.21875, "learning_rate": 1.608079629428115e-05, "loss": 0.132, "step": 36191 }, { "epoch": 26.514285714285712, "grad_norm": 1.265625, "learning_rate": 1.607833390770593e-05, "loss": 0.1239, "step": 36192 }, { "epoch": 26.515018315018317, "grad_norm": 0.98046875, "learning_rate": 1.6075871686056193e-05, "loss": 0.1182, "step": 36193 }, { "epoch": 26.515750915750917, "grad_norm": 1.015625, "learning_rate": 1.6073409629343754e-05, "loss": 0.1188, "step": 36194 }, { "epoch": 26.516483516483518, "grad_norm": 0.66015625, "learning_rate": 1.6070947737580418e-05, "loss": 0.1117, "step": 36195 }, { "epoch": 26.51721611721612, "grad_norm": 1.25, "learning_rate": 1.6068486010777978e-05, "loss": 0.121, "step": 36196 }, { "epoch": 26.51794871794872, "grad_norm": 0.7734375, "learning_rate": 1.6066024448948215e-05, "loss": 0.1134, "step": 36197 }, { "epoch": 26.51868131868132, "grad_norm": 0.76171875, "learning_rate": 1.6063563052102937e-05, "loss": 0.0934, "step": 36198 }, { "epoch": 26.51941391941392, "grad_norm": 1.125, "learning_rate": 1.6061101820253927e-05, "loss": 0.1041, "step": 36199 }, { "epoch": 26.520146520146522, "grad_norm": 0.94140625, "learning_rate": 1.6058640753413e-05, "loss": 0.114, "step": 36200 }, { "epoch": 26.520879120879123, "grad_norm": 0.71484375, "learning_rate": 1.605617985159192e-05, "loss": 0.1133, "step": 36201 }, { "epoch": 26.521611721611723, "grad_norm": 0.84765625, "learning_rate": 1.6053719114802503e-05, "loss": 0.1242, "step": 36202 }, { "epoch": 26.522344322344324, "grad_norm": 1.0859375, "learning_rate": 1.6051258543056507e-05, "loss": 0.0988, "step": 36203 }, { "epoch": 26.523076923076925, "grad_norm": 2.171875, "learning_rate": 1.604879813636574e-05, "loss": 0.136, "step": 36204 }, { "epoch": 26.523809523809526, "grad_norm": 1.109375, "learning_rate": 1.6046337894741997e-05, "loss": 0.1345, "step": 36205 }, { "epoch": 26.524542124542126, "grad_norm": 0.58984375, "learning_rate": 1.6043877818197057e-05, "loss": 0.1054, "step": 36206 }, { "epoch": 26.525274725274727, "grad_norm": 1.65625, "learning_rate": 1.6041417906742717e-05, "loss": 0.1443, "step": 36207 }, { "epoch": 26.526007326007328, "grad_norm": 1.859375, "learning_rate": 1.603895816039075e-05, "loss": 0.1766, "step": 36208 }, { "epoch": 26.52673992673993, "grad_norm": 0.9921875, "learning_rate": 1.6036498579152946e-05, "loss": 0.1211, "step": 36209 }, { "epoch": 26.52747252747253, "grad_norm": 1.328125, "learning_rate": 1.603403916304109e-05, "loss": 0.1267, "step": 36210 }, { "epoch": 26.52820512820513, "grad_norm": 0.91015625, "learning_rate": 1.6031579912066978e-05, "loss": 0.1076, "step": 36211 }, { "epoch": 26.52893772893773, "grad_norm": 0.83203125, "learning_rate": 1.602912082624237e-05, "loss": 0.1448, "step": 36212 }, { "epoch": 26.52967032967033, "grad_norm": 1.921875, "learning_rate": 1.6026661905579082e-05, "loss": 0.1075, "step": 36213 }, { "epoch": 26.530402930402932, "grad_norm": 1.0703125, "learning_rate": 1.6024203150088834e-05, "loss": 0.1025, "step": 36214 }, { "epoch": 26.531135531135533, "grad_norm": 1.1328125, "learning_rate": 1.6021744559783478e-05, "loss": 0.1123, "step": 36215 }, { "epoch": 26.531868131868134, "grad_norm": 1.640625, "learning_rate": 1.6019286134674745e-05, "loss": 0.1511, "step": 36216 }, { "epoch": 26.532600732600734, "grad_norm": 0.82421875, "learning_rate": 1.601682787477444e-05, "loss": 0.1234, "step": 36217 }, { "epoch": 26.533333333333335, "grad_norm": 1.3046875, "learning_rate": 1.6014369780094345e-05, "loss": 0.1434, "step": 36218 }, { "epoch": 26.534065934065936, "grad_norm": 1.03125, "learning_rate": 1.6011911850646202e-05, "loss": 0.1165, "step": 36219 }, { "epoch": 26.534798534798536, "grad_norm": 1.0390625, "learning_rate": 1.600945408644182e-05, "loss": 0.1206, "step": 36220 }, { "epoch": 26.535531135531137, "grad_norm": 0.83984375, "learning_rate": 1.600699648749296e-05, "loss": 0.1195, "step": 36221 }, { "epoch": 26.536263736263738, "grad_norm": 0.7265625, "learning_rate": 1.6004539053811423e-05, "loss": 0.0944, "step": 36222 }, { "epoch": 26.53699633699634, "grad_norm": 1.3125, "learning_rate": 1.6002081785408944e-05, "loss": 0.1378, "step": 36223 }, { "epoch": 26.53772893772894, "grad_norm": 1.140625, "learning_rate": 1.5999624682297322e-05, "loss": 0.1098, "step": 36224 }, { "epoch": 26.53846153846154, "grad_norm": 0.9375, "learning_rate": 1.5997167744488323e-05, "loss": 0.1121, "step": 36225 }, { "epoch": 26.53919413919414, "grad_norm": 0.8046875, "learning_rate": 1.5994710971993734e-05, "loss": 0.0969, "step": 36226 }, { "epoch": 26.53992673992674, "grad_norm": 0.7578125, "learning_rate": 1.5992254364825297e-05, "loss": 0.0979, "step": 36227 }, { "epoch": 26.540659340659342, "grad_norm": 1.4453125, "learning_rate": 1.5989797922994807e-05, "loss": 0.153, "step": 36228 }, { "epoch": 26.541391941391943, "grad_norm": 1.375, "learning_rate": 1.5987341646514015e-05, "loss": 0.128, "step": 36229 }, { "epoch": 26.542124542124544, "grad_norm": 0.7109375, "learning_rate": 1.59848855353947e-05, "loss": 0.11, "step": 36230 }, { "epoch": 26.542857142857144, "grad_norm": 0.78125, "learning_rate": 1.5982429589648624e-05, "loss": 0.1075, "step": 36231 }, { "epoch": 26.543589743589745, "grad_norm": 1.8671875, "learning_rate": 1.597997380928756e-05, "loss": 0.1624, "step": 36232 }, { "epoch": 26.544322344322346, "grad_norm": 0.83984375, "learning_rate": 1.5977518194323293e-05, "loss": 0.1147, "step": 36233 }, { "epoch": 26.545054945054947, "grad_norm": 1.0546875, "learning_rate": 1.597506274476755e-05, "loss": 0.1536, "step": 36234 }, { "epoch": 26.545787545787547, "grad_norm": 1.1171875, "learning_rate": 1.5972607460632115e-05, "loss": 0.1234, "step": 36235 }, { "epoch": 26.546520146520148, "grad_norm": 1.109375, "learning_rate": 1.5970152341928753e-05, "loss": 0.1389, "step": 36236 }, { "epoch": 26.54725274725275, "grad_norm": 1.3515625, "learning_rate": 1.596769738866924e-05, "loss": 0.1371, "step": 36237 }, { "epoch": 26.54798534798535, "grad_norm": 0.76953125, "learning_rate": 1.5965242600865314e-05, "loss": 0.1113, "step": 36238 }, { "epoch": 26.54871794871795, "grad_norm": 1.4140625, "learning_rate": 1.5962787978528765e-05, "loss": 0.118, "step": 36239 }, { "epoch": 26.54945054945055, "grad_norm": 0.953125, "learning_rate": 1.5960333521671303e-05, "loss": 0.1115, "step": 36240 }, { "epoch": 26.550183150183152, "grad_norm": 1.1171875, "learning_rate": 1.595787923030475e-05, "loss": 0.1269, "step": 36241 }, { "epoch": 26.550915750915753, "grad_norm": 0.96484375, "learning_rate": 1.5955425104440824e-05, "loss": 0.0963, "step": 36242 }, { "epoch": 26.551648351648353, "grad_norm": 1.3984375, "learning_rate": 1.5952971144091297e-05, "loss": 0.1083, "step": 36243 }, { "epoch": 26.552380952380954, "grad_norm": 1.1328125, "learning_rate": 1.5950517349267936e-05, "loss": 0.1251, "step": 36244 }, { "epoch": 26.553113553113555, "grad_norm": 0.75, "learning_rate": 1.5948063719982477e-05, "loss": 0.0978, "step": 36245 }, { "epoch": 26.553846153846155, "grad_norm": 0.80078125, "learning_rate": 1.594561025624669e-05, "loss": 0.1223, "step": 36246 }, { "epoch": 26.554578754578756, "grad_norm": 0.77734375, "learning_rate": 1.594315695807233e-05, "loss": 0.0963, "step": 36247 }, { "epoch": 26.555311355311357, "grad_norm": 0.796875, "learning_rate": 1.594070382547116e-05, "loss": 0.1014, "step": 36248 }, { "epoch": 26.556043956043958, "grad_norm": 0.73828125, "learning_rate": 1.5938250858454905e-05, "loss": 0.1129, "step": 36249 }, { "epoch": 26.55677655677656, "grad_norm": 0.9140625, "learning_rate": 1.5935798057035338e-05, "loss": 0.1314, "step": 36250 }, { "epoch": 26.55750915750916, "grad_norm": 0.9140625, "learning_rate": 1.593334542122421e-05, "loss": 0.1249, "step": 36251 }, { "epoch": 26.55824175824176, "grad_norm": 0.83984375, "learning_rate": 1.5930892951033287e-05, "loss": 0.1235, "step": 36252 }, { "epoch": 26.55897435897436, "grad_norm": 1.171875, "learning_rate": 1.592844064647429e-05, "loss": 0.1166, "step": 36253 }, { "epoch": 26.55970695970696, "grad_norm": 0.8671875, "learning_rate": 1.5925988507558998e-05, "loss": 0.1223, "step": 36254 }, { "epoch": 26.560439560439562, "grad_norm": 0.890625, "learning_rate": 1.5923536534299126e-05, "loss": 0.1071, "step": 36255 }, { "epoch": 26.561172161172163, "grad_norm": 1.046875, "learning_rate": 1.5921084726706444e-05, "loss": 0.1248, "step": 36256 }, { "epoch": 26.561904761904763, "grad_norm": 1.2421875, "learning_rate": 1.59186330847927e-05, "loss": 0.1507, "step": 36257 }, { "epoch": 26.562637362637364, "grad_norm": 1.0234375, "learning_rate": 1.5916181608569632e-05, "loss": 0.0895, "step": 36258 }, { "epoch": 26.563369963369965, "grad_norm": 1.3359375, "learning_rate": 1.591373029804901e-05, "loss": 0.1396, "step": 36259 }, { "epoch": 26.564102564102566, "grad_norm": 0.9140625, "learning_rate": 1.591127915324254e-05, "loss": 0.121, "step": 36260 }, { "epoch": 26.564835164835166, "grad_norm": 1.703125, "learning_rate": 1.590882817416199e-05, "loss": 0.1348, "step": 36261 }, { "epoch": 26.565567765567767, "grad_norm": 0.74609375, "learning_rate": 1.59063773608191e-05, "loss": 0.1334, "step": 36262 }, { "epoch": 26.566300366300368, "grad_norm": 1.390625, "learning_rate": 1.5903926713225628e-05, "loss": 0.1315, "step": 36263 }, { "epoch": 26.56703296703297, "grad_norm": 1.1015625, "learning_rate": 1.590147623139328e-05, "loss": 0.1503, "step": 36264 }, { "epoch": 26.56776556776557, "grad_norm": 1.3515625, "learning_rate": 1.589902591533384e-05, "loss": 0.1244, "step": 36265 }, { "epoch": 26.56849816849817, "grad_norm": 1.171875, "learning_rate": 1.589657576505899e-05, "loss": 0.0981, "step": 36266 }, { "epoch": 26.56923076923077, "grad_norm": 1.7890625, "learning_rate": 1.5894125780580535e-05, "loss": 0.147, "step": 36267 }, { "epoch": 26.56996336996337, "grad_norm": 1.296875, "learning_rate": 1.5891675961910172e-05, "loss": 0.1271, "step": 36268 }, { "epoch": 26.570695970695972, "grad_norm": 1.0, "learning_rate": 1.588922630905965e-05, "loss": 0.1349, "step": 36269 }, { "epoch": 26.571428571428573, "grad_norm": 1.3828125, "learning_rate": 1.5886776822040718e-05, "loss": 0.1662, "step": 36270 }, { "epoch": 26.572161172161174, "grad_norm": 0.95703125, "learning_rate": 1.588432750086509e-05, "loss": 0.1516, "step": 36271 }, { "epoch": 26.572893772893774, "grad_norm": 1.015625, "learning_rate": 1.588187834554451e-05, "loss": 0.1209, "step": 36272 }, { "epoch": 26.573626373626375, "grad_norm": 1.171875, "learning_rate": 1.5879429356090715e-05, "loss": 0.1368, "step": 36273 }, { "epoch": 26.574358974358976, "grad_norm": 0.75, "learning_rate": 1.5876980532515458e-05, "loss": 0.1345, "step": 36274 }, { "epoch": 26.575091575091577, "grad_norm": 0.90625, "learning_rate": 1.587453187483044e-05, "loss": 0.1178, "step": 36275 }, { "epoch": 26.575824175824177, "grad_norm": 1.0234375, "learning_rate": 1.58720833830474e-05, "loss": 0.1111, "step": 36276 }, { "epoch": 26.576556776556778, "grad_norm": 1.0703125, "learning_rate": 1.586963505717808e-05, "loss": 0.1525, "step": 36277 }, { "epoch": 26.57728937728938, "grad_norm": 1.6484375, "learning_rate": 1.586718689723422e-05, "loss": 0.1082, "step": 36278 }, { "epoch": 26.57802197802198, "grad_norm": 1.0859375, "learning_rate": 1.5864738903227523e-05, "loss": 0.1416, "step": 36279 }, { "epoch": 26.57875457875458, "grad_norm": 0.91015625, "learning_rate": 1.5862291075169747e-05, "loss": 0.1107, "step": 36280 }, { "epoch": 26.57948717948718, "grad_norm": 0.82421875, "learning_rate": 1.585984341307259e-05, "loss": 0.1268, "step": 36281 }, { "epoch": 26.58021978021978, "grad_norm": 0.80078125, "learning_rate": 1.5857395916947793e-05, "loss": 0.1171, "step": 36282 }, { "epoch": 26.580952380952382, "grad_norm": 0.6484375, "learning_rate": 1.5854948586807086e-05, "loss": 0.1164, "step": 36283 }, { "epoch": 26.581684981684983, "grad_norm": 1.0234375, "learning_rate": 1.5852501422662187e-05, "loss": 0.1303, "step": 36284 }, { "epoch": 26.582417582417584, "grad_norm": 1.046875, "learning_rate": 1.5850054424524853e-05, "loss": 0.1135, "step": 36285 }, { "epoch": 26.583150183150185, "grad_norm": 1.2109375, "learning_rate": 1.5847607592406766e-05, "loss": 0.1385, "step": 36286 }, { "epoch": 26.583882783882785, "grad_norm": 1.4296875, "learning_rate": 1.5845160926319665e-05, "loss": 0.1242, "step": 36287 }, { "epoch": 26.584615384615386, "grad_norm": 0.734375, "learning_rate": 1.5842714426275275e-05, "loss": 0.1092, "step": 36288 }, { "epoch": 26.585347985347987, "grad_norm": 0.85546875, "learning_rate": 1.584026809228533e-05, "loss": 0.1313, "step": 36289 }, { "epoch": 26.586080586080588, "grad_norm": 0.98046875, "learning_rate": 1.5837821924361526e-05, "loss": 0.1208, "step": 36290 }, { "epoch": 26.58681318681319, "grad_norm": 1.8671875, "learning_rate": 1.5835375922515613e-05, "loss": 0.1202, "step": 36291 }, { "epoch": 26.58754578754579, "grad_norm": 1.5859375, "learning_rate": 1.5832930086759258e-05, "loss": 0.1347, "step": 36292 }, { "epoch": 26.58827838827839, "grad_norm": 0.796875, "learning_rate": 1.5830484417104248e-05, "loss": 0.0914, "step": 36293 }, { "epoch": 26.58901098901099, "grad_norm": 0.91796875, "learning_rate": 1.5828038913562258e-05, "loss": 0.123, "step": 36294 }, { "epoch": 26.58974358974359, "grad_norm": 0.6640625, "learning_rate": 1.582559357614501e-05, "loss": 0.087, "step": 36295 }, { "epoch": 26.590476190476192, "grad_norm": 1.2890625, "learning_rate": 1.582314840486424e-05, "loss": 0.1237, "step": 36296 }, { "epoch": 26.591208791208793, "grad_norm": 1.5234375, "learning_rate": 1.5820703399731637e-05, "loss": 0.1346, "step": 36297 }, { "epoch": 26.591941391941393, "grad_norm": 0.70703125, "learning_rate": 1.5818258560758934e-05, "loss": 0.0853, "step": 36298 }, { "epoch": 26.592673992673994, "grad_norm": 0.9296875, "learning_rate": 1.5815813887957833e-05, "loss": 0.0969, "step": 36299 }, { "epoch": 26.593406593406595, "grad_norm": 1.203125, "learning_rate": 1.5813369381340066e-05, "loss": 0.1173, "step": 36300 }, { "epoch": 26.594139194139196, "grad_norm": 0.75390625, "learning_rate": 1.5810925040917326e-05, "loss": 0.1238, "step": 36301 }, { "epoch": 26.594871794871796, "grad_norm": 0.921875, "learning_rate": 1.5808480866701325e-05, "loss": 0.1184, "step": 36302 }, { "epoch": 26.595604395604397, "grad_norm": 1.3828125, "learning_rate": 1.5806036858703787e-05, "loss": 0.1422, "step": 36303 }, { "epoch": 26.596336996336998, "grad_norm": 1.046875, "learning_rate": 1.5803593016936424e-05, "loss": 0.1065, "step": 36304 }, { "epoch": 26.5970695970696, "grad_norm": 1.0859375, "learning_rate": 1.5801149341410928e-05, "loss": 0.1227, "step": 36305 }, { "epoch": 26.5978021978022, "grad_norm": 0.72265625, "learning_rate": 1.5798705832139013e-05, "loss": 0.1291, "step": 36306 }, { "epoch": 26.5985347985348, "grad_norm": 0.87890625, "learning_rate": 1.5796262489132405e-05, "loss": 0.1102, "step": 36307 }, { "epoch": 26.5992673992674, "grad_norm": 0.83984375, "learning_rate": 1.5793819312402783e-05, "loss": 0.1099, "step": 36308 }, { "epoch": 26.6, "grad_norm": 1.1953125, "learning_rate": 1.5791376301961862e-05, "loss": 0.1268, "step": 36309 }, { "epoch": 26.600732600732602, "grad_norm": 0.75, "learning_rate": 1.5788933457821353e-05, "loss": 0.1156, "step": 36310 }, { "epoch": 26.601465201465203, "grad_norm": 0.94921875, "learning_rate": 1.5786490779992974e-05, "loss": 0.1074, "step": 36311 }, { "epoch": 26.602197802197804, "grad_norm": 0.6484375, "learning_rate": 1.57840482684884e-05, "loss": 0.101, "step": 36312 }, { "epoch": 26.602930402930404, "grad_norm": 0.796875, "learning_rate": 1.578160592331935e-05, "loss": 0.105, "step": 36313 }, { "epoch": 26.603663003663005, "grad_norm": 1.25, "learning_rate": 1.5779163744497517e-05, "loss": 0.1358, "step": 36314 }, { "epoch": 26.604395604395606, "grad_norm": 0.86328125, "learning_rate": 1.5776721732034623e-05, "loss": 0.1119, "step": 36315 }, { "epoch": 26.605128205128207, "grad_norm": 0.9921875, "learning_rate": 1.5774279885942346e-05, "loss": 0.1258, "step": 36316 }, { "epoch": 26.605860805860807, "grad_norm": 1.2265625, "learning_rate": 1.57718382062324e-05, "loss": 0.1092, "step": 36317 }, { "epoch": 26.606593406593408, "grad_norm": 1.25, "learning_rate": 1.5769396692916455e-05, "loss": 0.1224, "step": 36318 }, { "epoch": 26.60732600732601, "grad_norm": 1.5078125, "learning_rate": 1.576695534600626e-05, "loss": 0.1055, "step": 36319 }, { "epoch": 26.60805860805861, "grad_norm": 1.3515625, "learning_rate": 1.5764514165513466e-05, "loss": 0.1164, "step": 36320 }, { "epoch": 26.60879120879121, "grad_norm": 1.0, "learning_rate": 1.576207315144979e-05, "loss": 0.1164, "step": 36321 }, { "epoch": 26.60952380952381, "grad_norm": 0.78515625, "learning_rate": 1.5759632303826942e-05, "loss": 0.121, "step": 36322 }, { "epoch": 26.61025641025641, "grad_norm": 1.1796875, "learning_rate": 1.575719162265657e-05, "loss": 0.1207, "step": 36323 }, { "epoch": 26.610989010989012, "grad_norm": 0.89453125, "learning_rate": 1.575475110795043e-05, "loss": 0.1533, "step": 36324 }, { "epoch": 26.611721611721613, "grad_norm": 0.78125, "learning_rate": 1.575231075972016e-05, "loss": 0.1104, "step": 36325 }, { "epoch": 26.612454212454214, "grad_norm": 1.078125, "learning_rate": 1.57498705779775e-05, "loss": 0.1554, "step": 36326 }, { "epoch": 26.613186813186815, "grad_norm": 0.9375, "learning_rate": 1.5747430562734094e-05, "loss": 0.152, "step": 36327 }, { "epoch": 26.613919413919415, "grad_norm": 0.984375, "learning_rate": 1.5744990714001678e-05, "loss": 0.12, "step": 36328 }, { "epoch": 26.614652014652016, "grad_norm": 1.1171875, "learning_rate": 1.5742551031791887e-05, "loss": 0.1224, "step": 36329 }, { "epoch": 26.615384615384617, "grad_norm": 0.7734375, "learning_rate": 1.574011151611648e-05, "loss": 0.1365, "step": 36330 }, { "epoch": 26.616117216117217, "grad_norm": 1.296875, "learning_rate": 1.5737672166987087e-05, "loss": 0.1349, "step": 36331 }, { "epoch": 26.616849816849818, "grad_norm": 1.6875, "learning_rate": 1.5735232984415426e-05, "loss": 0.1314, "step": 36332 }, { "epoch": 26.61758241758242, "grad_norm": 1.2734375, "learning_rate": 1.5732793968413186e-05, "loss": 0.1151, "step": 36333 }, { "epoch": 26.61831501831502, "grad_norm": 1.1640625, "learning_rate": 1.5730355118992033e-05, "loss": 0.1215, "step": 36334 }, { "epoch": 26.61904761904762, "grad_norm": 0.6953125, "learning_rate": 1.572791643616366e-05, "loss": 0.1124, "step": 36335 }, { "epoch": 26.61978021978022, "grad_norm": 1.015625, "learning_rate": 1.5725477919939754e-05, "loss": 0.1141, "step": 36336 }, { "epoch": 26.620512820512822, "grad_norm": 1.1171875, "learning_rate": 1.5723039570332015e-05, "loss": 0.1508, "step": 36337 }, { "epoch": 26.621245421245423, "grad_norm": 0.76171875, "learning_rate": 1.5720601387352093e-05, "loss": 0.1198, "step": 36338 }, { "epoch": 26.621978021978023, "grad_norm": 0.59375, "learning_rate": 1.5718163371011686e-05, "loss": 0.0898, "step": 36339 }, { "epoch": 26.622710622710624, "grad_norm": 0.7734375, "learning_rate": 1.571572552132248e-05, "loss": 0.115, "step": 36340 }, { "epoch": 26.623443223443225, "grad_norm": 0.890625, "learning_rate": 1.5713287838296167e-05, "loss": 0.1417, "step": 36341 }, { "epoch": 26.624175824175826, "grad_norm": 0.90625, "learning_rate": 1.5710850321944395e-05, "loss": 0.119, "step": 36342 }, { "epoch": 26.624908424908426, "grad_norm": 0.640625, "learning_rate": 1.5708412972278877e-05, "loss": 0.1083, "step": 36343 }, { "epoch": 26.625641025641027, "grad_norm": 1.078125, "learning_rate": 1.5705975789311243e-05, "loss": 0.152, "step": 36344 }, { "epoch": 26.626373626373628, "grad_norm": 1.6796875, "learning_rate": 1.5703538773053232e-05, "loss": 0.1069, "step": 36345 }, { "epoch": 26.62710622710623, "grad_norm": 1.0078125, "learning_rate": 1.570110192351647e-05, "loss": 0.1213, "step": 36346 }, { "epoch": 26.62783882783883, "grad_norm": 0.8515625, "learning_rate": 1.5698665240712656e-05, "loss": 0.1168, "step": 36347 }, { "epoch": 26.62857142857143, "grad_norm": 1.640625, "learning_rate": 1.5696228724653472e-05, "loss": 0.1311, "step": 36348 }, { "epoch": 26.62930402930403, "grad_norm": 1.25, "learning_rate": 1.569379237535056e-05, "loss": 0.117, "step": 36349 }, { "epoch": 26.63003663003663, "grad_norm": 0.97265625, "learning_rate": 1.569135619281564e-05, "loss": 0.1516, "step": 36350 }, { "epoch": 26.630769230769232, "grad_norm": 1.765625, "learning_rate": 1.568892017706034e-05, "loss": 0.1244, "step": 36351 }, { "epoch": 26.631501831501833, "grad_norm": 1.1015625, "learning_rate": 1.5686484328096375e-05, "loss": 0.126, "step": 36352 }, { "epoch": 26.632234432234434, "grad_norm": 1.109375, "learning_rate": 1.568404864593537e-05, "loss": 0.1065, "step": 36353 }, { "epoch": 26.632967032967034, "grad_norm": 1.109375, "learning_rate": 1.5681613130589036e-05, "loss": 0.1622, "step": 36354 }, { "epoch": 26.633699633699635, "grad_norm": 0.92578125, "learning_rate": 1.567917778206899e-05, "loss": 0.1077, "step": 36355 }, { "epoch": 26.634432234432236, "grad_norm": 0.63671875, "learning_rate": 1.567674260038697e-05, "loss": 0.1221, "step": 36356 }, { "epoch": 26.635164835164836, "grad_norm": 1.4453125, "learning_rate": 1.5674307585554587e-05, "loss": 0.1314, "step": 36357 }, { "epoch": 26.635897435897437, "grad_norm": 0.98828125, "learning_rate": 1.567187273758353e-05, "loss": 0.0951, "step": 36358 }, { "epoch": 26.636630036630038, "grad_norm": 0.92578125, "learning_rate": 1.566943805648548e-05, "loss": 0.1046, "step": 36359 }, { "epoch": 26.63736263736264, "grad_norm": 1.0234375, "learning_rate": 1.5667003542272074e-05, "loss": 0.0907, "step": 36360 }, { "epoch": 26.63809523809524, "grad_norm": 0.73046875, "learning_rate": 1.5664569194954988e-05, "loss": 0.1275, "step": 36361 }, { "epoch": 26.63882783882784, "grad_norm": 0.7734375, "learning_rate": 1.5662135014545883e-05, "loss": 0.1232, "step": 36362 }, { "epoch": 26.63956043956044, "grad_norm": 1.203125, "learning_rate": 1.5659701001056444e-05, "loss": 0.1279, "step": 36363 }, { "epoch": 26.64029304029304, "grad_norm": 1.0859375, "learning_rate": 1.5657267154498298e-05, "loss": 0.1476, "step": 36364 }, { "epoch": 26.641025641025642, "grad_norm": 1.2421875, "learning_rate": 1.5654833474883128e-05, "loss": 0.1229, "step": 36365 }, { "epoch": 26.641758241758243, "grad_norm": 1.1640625, "learning_rate": 1.5652399962222584e-05, "loss": 0.1254, "step": 36366 }, { "epoch": 26.642490842490844, "grad_norm": 1.2578125, "learning_rate": 1.5649966616528348e-05, "loss": 0.1588, "step": 36367 }, { "epoch": 26.643223443223444, "grad_norm": 1.3125, "learning_rate": 1.564753343781205e-05, "loss": 0.164, "step": 36368 }, { "epoch": 26.643956043956045, "grad_norm": 1.5859375, "learning_rate": 1.5645100426085372e-05, "loss": 0.1287, "step": 36369 }, { "epoch": 26.644688644688646, "grad_norm": 0.8515625, "learning_rate": 1.564266758135993e-05, "loss": 0.1243, "step": 36370 }, { "epoch": 26.645421245421247, "grad_norm": 1.65625, "learning_rate": 1.5640234903647443e-05, "loss": 0.1235, "step": 36371 }, { "epoch": 26.646153846153847, "grad_norm": 1.6171875, "learning_rate": 1.563780239295952e-05, "loss": 0.1488, "step": 36372 }, { "epoch": 26.646886446886448, "grad_norm": 0.8671875, "learning_rate": 1.5635370049307828e-05, "loss": 0.1253, "step": 36373 }, { "epoch": 26.64761904761905, "grad_norm": 0.828125, "learning_rate": 1.5632937872704038e-05, "loss": 0.1186, "step": 36374 }, { "epoch": 26.64835164835165, "grad_norm": 0.921875, "learning_rate": 1.5630505863159756e-05, "loss": 0.1314, "step": 36375 }, { "epoch": 26.64908424908425, "grad_norm": 0.6953125, "learning_rate": 1.5628074020686702e-05, "loss": 0.1118, "step": 36376 }, { "epoch": 26.64981684981685, "grad_norm": 1.3671875, "learning_rate": 1.5625642345296477e-05, "loss": 0.1183, "step": 36377 }, { "epoch": 26.650549450549452, "grad_norm": 0.9296875, "learning_rate": 1.5623210837000764e-05, "loss": 0.1171, "step": 36378 }, { "epoch": 26.651282051282053, "grad_norm": 0.8125, "learning_rate": 1.5620779495811183e-05, "loss": 0.133, "step": 36379 }, { "epoch": 26.652014652014653, "grad_norm": 1.6171875, "learning_rate": 1.561834832173941e-05, "loss": 0.146, "step": 36380 }, { "epoch": 26.652747252747254, "grad_norm": 1.3984375, "learning_rate": 1.561591731479705e-05, "loss": 0.1214, "step": 36381 }, { "epoch": 26.653479853479855, "grad_norm": 0.765625, "learning_rate": 1.5613486474995816e-05, "loss": 0.095, "step": 36382 }, { "epoch": 26.654212454212455, "grad_norm": 0.88671875, "learning_rate": 1.56110558023473e-05, "loss": 0.1492, "step": 36383 }, { "epoch": 26.654945054945056, "grad_norm": 1.1875, "learning_rate": 1.5608625296863178e-05, "loss": 0.126, "step": 36384 }, { "epoch": 26.655677655677657, "grad_norm": 0.7890625, "learning_rate": 1.5606194958555094e-05, "loss": 0.152, "step": 36385 }, { "epoch": 26.656410256410258, "grad_norm": 1.1484375, "learning_rate": 1.5603764787434672e-05, "loss": 0.1309, "step": 36386 }, { "epoch": 26.65714285714286, "grad_norm": 0.921875, "learning_rate": 1.5601334783513566e-05, "loss": 0.1304, "step": 36387 }, { "epoch": 26.65787545787546, "grad_norm": 0.6328125, "learning_rate": 1.5598904946803425e-05, "loss": 0.1124, "step": 36388 }, { "epoch": 26.65860805860806, "grad_norm": 0.67578125, "learning_rate": 1.5596475277315904e-05, "loss": 0.1095, "step": 36389 }, { "epoch": 26.65934065934066, "grad_norm": 1.0625, "learning_rate": 1.5594045775062607e-05, "loss": 0.1044, "step": 36390 }, { "epoch": 26.66007326007326, "grad_norm": 0.6640625, "learning_rate": 1.55916164400552e-05, "loss": 0.1204, "step": 36391 }, { "epoch": 26.660805860805862, "grad_norm": 0.96484375, "learning_rate": 1.5589187272305318e-05, "loss": 0.108, "step": 36392 }, { "epoch": 26.661538461538463, "grad_norm": 1.1953125, "learning_rate": 1.5586758271824616e-05, "loss": 0.0951, "step": 36393 }, { "epoch": 26.662271062271063, "grad_norm": 0.84375, "learning_rate": 1.5584329438624697e-05, "loss": 0.1077, "step": 36394 }, { "epoch": 26.663003663003664, "grad_norm": 1.4140625, "learning_rate": 1.5581900772717235e-05, "loss": 0.1213, "step": 36395 }, { "epoch": 26.663736263736265, "grad_norm": 0.67578125, "learning_rate": 1.5579472274113823e-05, "loss": 0.1291, "step": 36396 }, { "epoch": 26.664468864468866, "grad_norm": 0.96875, "learning_rate": 1.5577043942826147e-05, "loss": 0.1326, "step": 36397 }, { "epoch": 26.665201465201466, "grad_norm": 0.8984375, "learning_rate": 1.5574615778865805e-05, "loss": 0.1029, "step": 36398 }, { "epoch": 26.665934065934067, "grad_norm": 1.5078125, "learning_rate": 1.5572187782244445e-05, "loss": 0.1174, "step": 36399 }, { "epoch": 26.666666666666668, "grad_norm": 1.453125, "learning_rate": 1.556975995297372e-05, "loss": 0.1092, "step": 36400 }, { "epoch": 26.66739926739927, "grad_norm": 1.4921875, "learning_rate": 1.55673322910652e-05, "loss": 0.1481, "step": 36401 }, { "epoch": 26.66813186813187, "grad_norm": 0.90625, "learning_rate": 1.5564904796530598e-05, "loss": 0.1241, "step": 36402 }, { "epoch": 26.66886446886447, "grad_norm": 0.875, "learning_rate": 1.556247746938149e-05, "loss": 0.1092, "step": 36403 }, { "epoch": 26.66959706959707, "grad_norm": 0.921875, "learning_rate": 1.5560050309629533e-05, "loss": 0.1164, "step": 36404 }, { "epoch": 26.67032967032967, "grad_norm": 1.375, "learning_rate": 1.5557623317286326e-05, "loss": 0.121, "step": 36405 }, { "epoch": 26.671062271062272, "grad_norm": 0.9296875, "learning_rate": 1.5555196492363537e-05, "loss": 0.1356, "step": 36406 }, { "epoch": 26.671794871794873, "grad_norm": 0.90234375, "learning_rate": 1.5552769834872742e-05, "loss": 0.1276, "step": 36407 }, { "epoch": 26.672527472527474, "grad_norm": 0.85546875, "learning_rate": 1.5550343344825632e-05, "loss": 0.1104, "step": 36408 }, { "epoch": 26.673260073260074, "grad_norm": 0.95703125, "learning_rate": 1.5547917022233782e-05, "loss": 0.0804, "step": 36409 }, { "epoch": 26.673992673992675, "grad_norm": 1.1328125, "learning_rate": 1.5545490867108843e-05, "loss": 0.1153, "step": 36410 }, { "epoch": 26.674725274725276, "grad_norm": 1.015625, "learning_rate": 1.554306487946244e-05, "loss": 0.1063, "step": 36411 }, { "epoch": 26.675457875457877, "grad_norm": 1.015625, "learning_rate": 1.554063905930617e-05, "loss": 0.1152, "step": 36412 }, { "epoch": 26.676190476190477, "grad_norm": 1.15625, "learning_rate": 1.553821340665168e-05, "loss": 0.1325, "step": 36413 }, { "epoch": 26.676923076923078, "grad_norm": 0.99609375, "learning_rate": 1.5535787921510585e-05, "loss": 0.1185, "step": 36414 }, { "epoch": 26.67765567765568, "grad_norm": 1.5078125, "learning_rate": 1.5533362603894527e-05, "loss": 0.1458, "step": 36415 }, { "epoch": 26.67838827838828, "grad_norm": 0.90625, "learning_rate": 1.5530937453815088e-05, "loss": 0.1293, "step": 36416 }, { "epoch": 26.67912087912088, "grad_norm": 0.77734375, "learning_rate": 1.5528512471283912e-05, "loss": 0.103, "step": 36417 }, { "epoch": 26.67985347985348, "grad_norm": 1.0390625, "learning_rate": 1.552608765631261e-05, "loss": 0.0995, "step": 36418 }, { "epoch": 26.68058608058608, "grad_norm": 0.75, "learning_rate": 1.5523663008912817e-05, "loss": 0.1242, "step": 36419 }, { "epoch": 26.681318681318682, "grad_norm": 0.87890625, "learning_rate": 1.5521238529096126e-05, "loss": 0.1445, "step": 36420 }, { "epoch": 26.682051282051283, "grad_norm": 1.1328125, "learning_rate": 1.5518814216874176e-05, "loss": 0.1347, "step": 36421 }, { "epoch": 26.682783882783884, "grad_norm": 1.78125, "learning_rate": 1.5516390072258536e-05, "loss": 0.1351, "step": 36422 }, { "epoch": 26.683516483516485, "grad_norm": 1.2890625, "learning_rate": 1.5513966095260894e-05, "loss": 0.163, "step": 36423 }, { "epoch": 26.684249084249085, "grad_norm": 1.03125, "learning_rate": 1.5511542285892807e-05, "loss": 0.1177, "step": 36424 }, { "epoch": 26.684981684981686, "grad_norm": 1.2109375, "learning_rate": 1.5509118644165903e-05, "loss": 0.1098, "step": 36425 }, { "epoch": 26.685714285714287, "grad_norm": 1.3984375, "learning_rate": 1.5506695170091817e-05, "loss": 0.1386, "step": 36426 }, { "epoch": 26.686446886446888, "grad_norm": 1.1015625, "learning_rate": 1.5504271863682116e-05, "loss": 0.1113, "step": 36427 }, { "epoch": 26.68717948717949, "grad_norm": 0.75390625, "learning_rate": 1.550184872494846e-05, "loss": 0.1238, "step": 36428 }, { "epoch": 26.68791208791209, "grad_norm": 1.2734375, "learning_rate": 1.5499425753902425e-05, "loss": 0.1104, "step": 36429 }, { "epoch": 26.68864468864469, "grad_norm": 0.94921875, "learning_rate": 1.5497002950555643e-05, "loss": 0.1353, "step": 36430 }, { "epoch": 26.68937728937729, "grad_norm": 0.890625, "learning_rate": 1.5494580314919695e-05, "loss": 0.121, "step": 36431 }, { "epoch": 26.69010989010989, "grad_norm": 0.9453125, "learning_rate": 1.5492157847006225e-05, "loss": 0.1346, "step": 36432 }, { "epoch": 26.690842490842492, "grad_norm": 1.203125, "learning_rate": 1.5489735546826782e-05, "loss": 0.1373, "step": 36433 }, { "epoch": 26.691575091575093, "grad_norm": 0.67578125, "learning_rate": 1.5487313414393047e-05, "loss": 0.1227, "step": 36434 }, { "epoch": 26.692307692307693, "grad_norm": 0.98046875, "learning_rate": 1.5484891449716568e-05, "loss": 0.1182, "step": 36435 }, { "epoch": 26.693040293040294, "grad_norm": 0.76953125, "learning_rate": 1.548246965280897e-05, "loss": 0.1079, "step": 36436 }, { "epoch": 26.693772893772895, "grad_norm": 0.92578125, "learning_rate": 1.5480048023681868e-05, "loss": 0.1402, "step": 36437 }, { "epoch": 26.694505494505496, "grad_norm": 0.75, "learning_rate": 1.5477626562346824e-05, "loss": 0.109, "step": 36438 }, { "epoch": 26.695238095238096, "grad_norm": 1.515625, "learning_rate": 1.547520526881549e-05, "loss": 0.1353, "step": 36439 }, { "epoch": 26.695970695970697, "grad_norm": 0.85546875, "learning_rate": 1.5472784143099445e-05, "loss": 0.1208, "step": 36440 }, { "epoch": 26.696703296703298, "grad_norm": 0.6875, "learning_rate": 1.54703631852103e-05, "loss": 0.1444, "step": 36441 }, { "epoch": 26.6974358974359, "grad_norm": 0.85546875, "learning_rate": 1.5467942395159623e-05, "loss": 0.1005, "step": 36442 }, { "epoch": 26.6981684981685, "grad_norm": 1.8125, "learning_rate": 1.5465521772959042e-05, "loss": 0.1504, "step": 36443 }, { "epoch": 26.6989010989011, "grad_norm": 0.87890625, "learning_rate": 1.546310131862014e-05, "loss": 0.1106, "step": 36444 }, { "epoch": 26.6996336996337, "grad_norm": 1.1015625, "learning_rate": 1.546068103215454e-05, "loss": 0.1039, "step": 36445 }, { "epoch": 26.7003663003663, "grad_norm": 0.80859375, "learning_rate": 1.5458260913573804e-05, "loss": 0.1141, "step": 36446 }, { "epoch": 26.701098901098902, "grad_norm": 0.91015625, "learning_rate": 1.545584096288955e-05, "loss": 0.1173, "step": 36447 }, { "epoch": 26.701831501831503, "grad_norm": 1.0546875, "learning_rate": 1.5453421180113358e-05, "loss": 0.1174, "step": 36448 }, { "epoch": 26.702564102564104, "grad_norm": 1.2734375, "learning_rate": 1.5451001565256847e-05, "loss": 0.1387, "step": 36449 }, { "epoch": 26.703296703296704, "grad_norm": 0.98828125, "learning_rate": 1.544858211833158e-05, "loss": 0.1005, "step": 36450 }, { "epoch": 26.704029304029305, "grad_norm": 1.078125, "learning_rate": 1.5446162839349163e-05, "loss": 0.1318, "step": 36451 }, { "epoch": 26.704761904761906, "grad_norm": 1.0390625, "learning_rate": 1.5443743728321202e-05, "loss": 0.1204, "step": 36452 }, { "epoch": 26.705494505494507, "grad_norm": 0.7890625, "learning_rate": 1.5441324785259244e-05, "loss": 0.104, "step": 36453 }, { "epoch": 26.706227106227107, "grad_norm": 0.9609375, "learning_rate": 1.5438906010174934e-05, "loss": 0.1124, "step": 36454 }, { "epoch": 26.706959706959708, "grad_norm": 0.96875, "learning_rate": 1.5436487403079823e-05, "loss": 0.1263, "step": 36455 }, { "epoch": 26.70769230769231, "grad_norm": 0.8359375, "learning_rate": 1.5434068963985528e-05, "loss": 0.1015, "step": 36456 }, { "epoch": 26.70842490842491, "grad_norm": 1.2265625, "learning_rate": 1.54316506929036e-05, "loss": 0.0915, "step": 36457 }, { "epoch": 26.70915750915751, "grad_norm": 1.171875, "learning_rate": 1.5429232589845662e-05, "loss": 0.1585, "step": 36458 }, { "epoch": 26.70989010989011, "grad_norm": 1.1796875, "learning_rate": 1.5426814654823256e-05, "loss": 0.136, "step": 36459 }, { "epoch": 26.71062271062271, "grad_norm": 1.5390625, "learning_rate": 1.5424396887848026e-05, "loss": 0.1318, "step": 36460 }, { "epoch": 26.711355311355312, "grad_norm": 1.6796875, "learning_rate": 1.542197928893151e-05, "loss": 0.1487, "step": 36461 }, { "epoch": 26.712087912087913, "grad_norm": 1.390625, "learning_rate": 1.5419561858085305e-05, "loss": 0.1215, "step": 36462 }, { "epoch": 26.712820512820514, "grad_norm": 1.703125, "learning_rate": 1.541714459532101e-05, "loss": 0.13, "step": 36463 }, { "epoch": 26.713553113553115, "grad_norm": 1.4140625, "learning_rate": 1.5414727500650157e-05, "loss": 0.1116, "step": 36464 }, { "epoch": 26.714285714285715, "grad_norm": 0.81640625, "learning_rate": 1.54123105740844e-05, "loss": 0.1228, "step": 36465 }, { "epoch": 26.715018315018316, "grad_norm": 0.9921875, "learning_rate": 1.5409893815635263e-05, "loss": 0.1152, "step": 36466 }, { "epoch": 26.715750915750917, "grad_norm": 0.9375, "learning_rate": 1.540747722531436e-05, "loss": 0.0952, "step": 36467 }, { "epoch": 26.716483516483517, "grad_norm": 1.1328125, "learning_rate": 1.5405060803133235e-05, "loss": 0.1116, "step": 36468 }, { "epoch": 26.717216117216118, "grad_norm": 1.0546875, "learning_rate": 1.540264454910349e-05, "loss": 0.1041, "step": 36469 }, { "epoch": 26.71794871794872, "grad_norm": 1.1328125, "learning_rate": 1.5400228463236697e-05, "loss": 0.1137, "step": 36470 }, { "epoch": 26.71868131868132, "grad_norm": 1.015625, "learning_rate": 1.5397812545544448e-05, "loss": 0.1215, "step": 36471 }, { "epoch": 26.71941391941392, "grad_norm": 0.85546875, "learning_rate": 1.5395396796038285e-05, "loss": 0.1185, "step": 36472 }, { "epoch": 26.72014652014652, "grad_norm": 1.2109375, "learning_rate": 1.5392981214729796e-05, "loss": 0.0957, "step": 36473 }, { "epoch": 26.720879120879122, "grad_norm": 1.203125, "learning_rate": 1.5390565801630564e-05, "loss": 0.1346, "step": 36474 }, { "epoch": 26.721611721611723, "grad_norm": 0.76171875, "learning_rate": 1.5388150556752167e-05, "loss": 0.139, "step": 36475 }, { "epoch": 26.722344322344323, "grad_norm": 0.6953125, "learning_rate": 1.538573548010616e-05, "loss": 0.0981, "step": 36476 }, { "epoch": 26.723076923076924, "grad_norm": 0.6015625, "learning_rate": 1.5383320571704112e-05, "loss": 0.1146, "step": 36477 }, { "epoch": 26.723809523809525, "grad_norm": 0.765625, "learning_rate": 1.5380905831557622e-05, "loss": 0.1165, "step": 36478 }, { "epoch": 26.724542124542126, "grad_norm": 1.3125, "learning_rate": 1.537849125967822e-05, "loss": 0.1355, "step": 36479 }, { "epoch": 26.725274725274726, "grad_norm": 1.015625, "learning_rate": 1.5376076856077512e-05, "loss": 0.1144, "step": 36480 }, { "epoch": 26.726007326007327, "grad_norm": 1.7734375, "learning_rate": 1.537366262076704e-05, "loss": 0.1224, "step": 36481 }, { "epoch": 26.726739926739928, "grad_norm": 1.109375, "learning_rate": 1.5371248553758398e-05, "loss": 0.1134, "step": 36482 }, { "epoch": 26.72747252747253, "grad_norm": 0.9453125, "learning_rate": 1.5368834655063125e-05, "loss": 0.1158, "step": 36483 }, { "epoch": 26.72820512820513, "grad_norm": 1.03125, "learning_rate": 1.5366420924692808e-05, "loss": 0.1137, "step": 36484 }, { "epoch": 26.72893772893773, "grad_norm": 1.0703125, "learning_rate": 1.5364007362658976e-05, "loss": 0.116, "step": 36485 }, { "epoch": 26.72967032967033, "grad_norm": 1.046875, "learning_rate": 1.5361593968973247e-05, "loss": 0.1124, "step": 36486 }, { "epoch": 26.73040293040293, "grad_norm": 1.3125, "learning_rate": 1.5359180743647134e-05, "loss": 0.1346, "step": 36487 }, { "epoch": 26.731135531135532, "grad_norm": 1.453125, "learning_rate": 1.5356767686692234e-05, "loss": 0.1606, "step": 36488 }, { "epoch": 26.731868131868133, "grad_norm": 1.3515625, "learning_rate": 1.5354354798120106e-05, "loss": 0.1559, "step": 36489 }, { "epoch": 26.732600732600734, "grad_norm": 0.875, "learning_rate": 1.535194207794227e-05, "loss": 0.102, "step": 36490 }, { "epoch": 26.733333333333334, "grad_norm": 1.328125, "learning_rate": 1.5349529526170347e-05, "loss": 0.1287, "step": 36491 }, { "epoch": 26.734065934065935, "grad_norm": 0.6640625, "learning_rate": 1.5347117142815858e-05, "loss": 0.1366, "step": 36492 }, { "epoch": 26.734798534798536, "grad_norm": 1.9453125, "learning_rate": 1.5344704927890387e-05, "loss": 0.1227, "step": 36493 }, { "epoch": 26.735531135531136, "grad_norm": 0.875, "learning_rate": 1.5342292881405455e-05, "loss": 0.1195, "step": 36494 }, { "epoch": 26.736263736263737, "grad_norm": 1.1171875, "learning_rate": 1.5339881003372638e-05, "loss": 0.127, "step": 36495 }, { "epoch": 26.736996336996338, "grad_norm": 0.9609375, "learning_rate": 1.5337469293803497e-05, "loss": 0.116, "step": 36496 }, { "epoch": 26.73772893772894, "grad_norm": 0.9453125, "learning_rate": 1.53350577527096e-05, "loss": 0.0999, "step": 36497 }, { "epoch": 26.73846153846154, "grad_norm": 1.03125, "learning_rate": 1.5332646380102466e-05, "loss": 0.1449, "step": 36498 }, { "epoch": 26.73919413919414, "grad_norm": 1.4453125, "learning_rate": 1.533023517599367e-05, "loss": 0.1337, "step": 36499 }, { "epoch": 26.73992673992674, "grad_norm": 1.046875, "learning_rate": 1.532782414039476e-05, "loss": 0.1231, "step": 36500 }, { "epoch": 26.74065934065934, "grad_norm": 1.0078125, "learning_rate": 1.532541327331731e-05, "loss": 0.1449, "step": 36501 }, { "epoch": 26.741391941391942, "grad_norm": 1.2109375, "learning_rate": 1.5323002574772835e-05, "loss": 0.1168, "step": 36502 }, { "epoch": 26.742124542124543, "grad_norm": 0.58984375, "learning_rate": 1.5320592044772906e-05, "loss": 0.0928, "step": 36503 }, { "epoch": 26.742857142857144, "grad_norm": 1.09375, "learning_rate": 1.531818168332908e-05, "loss": 0.1274, "step": 36504 }, { "epoch": 26.743589743589745, "grad_norm": 0.921875, "learning_rate": 1.5315771490452863e-05, "loss": 0.1169, "step": 36505 }, { "epoch": 26.744322344322345, "grad_norm": 0.7890625, "learning_rate": 1.531336146615587e-05, "loss": 0.1415, "step": 36506 }, { "epoch": 26.745054945054946, "grad_norm": 1.1015625, "learning_rate": 1.5310951610449597e-05, "loss": 0.1469, "step": 36507 }, { "epoch": 26.745787545787547, "grad_norm": 1.1171875, "learning_rate": 1.5308541923345625e-05, "loss": 0.144, "step": 36508 }, { "epoch": 26.746520146520147, "grad_norm": 0.80859375, "learning_rate": 1.5306132404855457e-05, "loss": 0.0968, "step": 36509 }, { "epoch": 26.747252747252748, "grad_norm": 0.828125, "learning_rate": 1.5303723054990683e-05, "loss": 0.0974, "step": 36510 }, { "epoch": 26.74798534798535, "grad_norm": 0.6484375, "learning_rate": 1.5301313873762797e-05, "loss": 0.1077, "step": 36511 }, { "epoch": 26.74871794871795, "grad_norm": 1.0859375, "learning_rate": 1.5298904861183396e-05, "loss": 0.1099, "step": 36512 }, { "epoch": 26.74945054945055, "grad_norm": 0.83203125, "learning_rate": 1.529649601726399e-05, "loss": 0.1203, "step": 36513 }, { "epoch": 26.75018315018315, "grad_norm": 0.93359375, "learning_rate": 1.5294087342016127e-05, "loss": 0.1135, "step": 36514 }, { "epoch": 26.750915750915752, "grad_norm": 0.63671875, "learning_rate": 1.5291678835451355e-05, "loss": 0.1136, "step": 36515 }, { "epoch": 26.751648351648353, "grad_norm": 0.859375, "learning_rate": 1.5289270497581188e-05, "loss": 0.1055, "step": 36516 }, { "epoch": 26.752380952380953, "grad_norm": 0.99609375, "learning_rate": 1.5286862328417212e-05, "loss": 0.1478, "step": 36517 }, { "epoch": 26.753113553113554, "grad_norm": 0.79296875, "learning_rate": 1.528445432797092e-05, "loss": 0.1254, "step": 36518 }, { "epoch": 26.753846153846155, "grad_norm": 1.265625, "learning_rate": 1.5282046496253886e-05, "loss": 0.1042, "step": 36519 }, { "epoch": 26.754578754578755, "grad_norm": 0.8515625, "learning_rate": 1.527963883327761e-05, "loss": 0.1408, "step": 36520 }, { "epoch": 26.755311355311356, "grad_norm": 1.4453125, "learning_rate": 1.5277231339053653e-05, "loss": 0.166, "step": 36521 }, { "epoch": 26.756043956043957, "grad_norm": 1.390625, "learning_rate": 1.5274824013593538e-05, "loss": 0.1276, "step": 36522 }, { "epoch": 26.756776556776558, "grad_norm": 0.69921875, "learning_rate": 1.527241685690882e-05, "loss": 0.1264, "step": 36523 }, { "epoch": 26.75750915750916, "grad_norm": 0.68359375, "learning_rate": 1.5270009869011005e-05, "loss": 0.0926, "step": 36524 }, { "epoch": 26.75824175824176, "grad_norm": 1.21875, "learning_rate": 1.5267603049911634e-05, "loss": 0.1174, "step": 36525 }, { "epoch": 26.75897435897436, "grad_norm": 0.79296875, "learning_rate": 1.5265196399622247e-05, "loss": 0.1102, "step": 36526 }, { "epoch": 26.75970695970696, "grad_norm": 0.8828125, "learning_rate": 1.526278991815439e-05, "loss": 0.1006, "step": 36527 }, { "epoch": 26.76043956043956, "grad_norm": 0.80859375, "learning_rate": 1.5260383605519548e-05, "loss": 0.1172, "step": 36528 }, { "epoch": 26.761172161172162, "grad_norm": 0.91796875, "learning_rate": 1.5257977461729287e-05, "loss": 0.1145, "step": 36529 }, { "epoch": 26.761904761904763, "grad_norm": 0.9140625, "learning_rate": 1.5255571486795131e-05, "loss": 0.0998, "step": 36530 }, { "epoch": 26.762637362637363, "grad_norm": 0.984375, "learning_rate": 1.5253165680728573e-05, "loss": 0.1334, "step": 36531 }, { "epoch": 26.763369963369964, "grad_norm": 1.3671875, "learning_rate": 1.5250760043541209e-05, "loss": 0.1407, "step": 36532 }, { "epoch": 26.764102564102565, "grad_norm": 1.015625, "learning_rate": 1.5248354575244503e-05, "loss": 0.1119, "step": 36533 }, { "epoch": 26.764835164835166, "grad_norm": 0.85546875, "learning_rate": 1.5245949275850018e-05, "loss": 0.1032, "step": 36534 }, { "epoch": 26.765567765567766, "grad_norm": 1.09375, "learning_rate": 1.5243544145369249e-05, "loss": 0.1551, "step": 36535 }, { "epoch": 26.766300366300367, "grad_norm": 0.80859375, "learning_rate": 1.524113918381375e-05, "loss": 0.1391, "step": 36536 }, { "epoch": 26.767032967032968, "grad_norm": 0.9453125, "learning_rate": 1.5238734391194994e-05, "loss": 0.1064, "step": 36537 }, { "epoch": 26.76776556776557, "grad_norm": 0.81640625, "learning_rate": 1.5236329767524566e-05, "loss": 0.1013, "step": 36538 }, { "epoch": 26.76849816849817, "grad_norm": 1.2265625, "learning_rate": 1.5233925312813946e-05, "loss": 0.1217, "step": 36539 }, { "epoch": 26.76923076923077, "grad_norm": 1.2421875, "learning_rate": 1.5231521027074668e-05, "loss": 0.1019, "step": 36540 }, { "epoch": 26.76996336996337, "grad_norm": 0.99609375, "learning_rate": 1.5229116910318267e-05, "loss": 0.1109, "step": 36541 }, { "epoch": 26.77069597069597, "grad_norm": 0.5703125, "learning_rate": 1.5226712962556214e-05, "loss": 0.1021, "step": 36542 }, { "epoch": 26.771428571428572, "grad_norm": 0.86328125, "learning_rate": 1.5224309183800087e-05, "loss": 0.1298, "step": 36543 }, { "epoch": 26.772161172161173, "grad_norm": 1.0546875, "learning_rate": 1.5221905574061363e-05, "loss": 0.1242, "step": 36544 }, { "epoch": 26.772893772893774, "grad_norm": 0.796875, "learning_rate": 1.521950213335158e-05, "loss": 0.1116, "step": 36545 }, { "epoch": 26.773626373626374, "grad_norm": 1.2578125, "learning_rate": 1.5217098861682233e-05, "loss": 0.1095, "step": 36546 }, { "epoch": 26.774358974358975, "grad_norm": 0.9296875, "learning_rate": 1.5214695759064849e-05, "loss": 0.1383, "step": 36547 }, { "epoch": 26.775091575091576, "grad_norm": 0.63671875, "learning_rate": 1.5212292825510941e-05, "loss": 0.1061, "step": 36548 }, { "epoch": 26.775824175824177, "grad_norm": 0.93359375, "learning_rate": 1.5209890061032033e-05, "loss": 0.1177, "step": 36549 }, { "epoch": 26.776556776556777, "grad_norm": 1.109375, "learning_rate": 1.5207487465639616e-05, "loss": 0.1691, "step": 36550 }, { "epoch": 26.777289377289378, "grad_norm": 1.0625, "learning_rate": 1.5205085039345212e-05, "loss": 0.1101, "step": 36551 }, { "epoch": 26.77802197802198, "grad_norm": 1.1484375, "learning_rate": 1.520268278216033e-05, "loss": 0.1103, "step": 36552 }, { "epoch": 26.77875457875458, "grad_norm": 0.8828125, "learning_rate": 1.52002806940965e-05, "loss": 0.1142, "step": 36553 }, { "epoch": 26.77948717948718, "grad_norm": 0.85546875, "learning_rate": 1.5197878775165197e-05, "loss": 0.114, "step": 36554 }, { "epoch": 26.78021978021978, "grad_norm": 1.328125, "learning_rate": 1.5195477025377943e-05, "loss": 0.1083, "step": 36555 }, { "epoch": 26.78095238095238, "grad_norm": 1.390625, "learning_rate": 1.5193075444746265e-05, "loss": 0.1283, "step": 36556 }, { "epoch": 26.781684981684982, "grad_norm": 1.078125, "learning_rate": 1.5190674033281643e-05, "loss": 0.1274, "step": 36557 }, { "epoch": 26.782417582417583, "grad_norm": 1.0, "learning_rate": 1.5188272790995584e-05, "loss": 0.0977, "step": 36558 }, { "epoch": 26.783150183150184, "grad_norm": 0.9296875, "learning_rate": 1.5185871717899609e-05, "loss": 0.1476, "step": 36559 }, { "epoch": 26.783882783882785, "grad_norm": 0.88671875, "learning_rate": 1.518347081400523e-05, "loss": 0.1177, "step": 36560 }, { "epoch": 26.784615384615385, "grad_norm": 1.328125, "learning_rate": 1.518107007932392e-05, "loss": 0.1123, "step": 36561 }, { "epoch": 26.785347985347986, "grad_norm": 1.3046875, "learning_rate": 1.5178669513867194e-05, "loss": 0.1262, "step": 36562 }, { "epoch": 26.786080586080587, "grad_norm": 1.125, "learning_rate": 1.5176269117646564e-05, "loss": 0.1213, "step": 36563 }, { "epoch": 26.786813186813188, "grad_norm": 0.9453125, "learning_rate": 1.5173868890673534e-05, "loss": 0.1173, "step": 36564 }, { "epoch": 26.78754578754579, "grad_norm": 1.53125, "learning_rate": 1.5171468832959582e-05, "loss": 0.1367, "step": 36565 }, { "epoch": 26.78827838827839, "grad_norm": 0.9921875, "learning_rate": 1.516906894451622e-05, "loss": 0.1107, "step": 36566 }, { "epoch": 26.78901098901099, "grad_norm": 1.375, "learning_rate": 1.5166669225354966e-05, "loss": 0.1094, "step": 36567 }, { "epoch": 26.78974358974359, "grad_norm": 1.0234375, "learning_rate": 1.5164269675487264e-05, "loss": 0.1533, "step": 36568 }, { "epoch": 26.79047619047619, "grad_norm": 0.88671875, "learning_rate": 1.5161870294924676e-05, "loss": 0.106, "step": 36569 }, { "epoch": 26.791208791208792, "grad_norm": 1.09375, "learning_rate": 1.5159471083678658e-05, "loss": 0.1136, "step": 36570 }, { "epoch": 26.791941391941393, "grad_norm": 1.0703125, "learning_rate": 1.5157072041760728e-05, "loss": 0.1151, "step": 36571 }, { "epoch": 26.792673992673993, "grad_norm": 1.09375, "learning_rate": 1.5154673169182348e-05, "loss": 0.1382, "step": 36572 }, { "epoch": 26.793406593406594, "grad_norm": 1.328125, "learning_rate": 1.5152274465955041e-05, "loss": 0.1197, "step": 36573 }, { "epoch": 26.794139194139195, "grad_norm": 0.74609375, "learning_rate": 1.5149875932090282e-05, "loss": 0.0992, "step": 36574 }, { "epoch": 26.794871794871796, "grad_norm": 1.0703125, "learning_rate": 1.5147477567599594e-05, "loss": 0.1038, "step": 36575 }, { "epoch": 26.795604395604396, "grad_norm": 0.96484375, "learning_rate": 1.5145079372494429e-05, "loss": 0.101, "step": 36576 }, { "epoch": 26.796336996336997, "grad_norm": 0.85546875, "learning_rate": 1.5142681346786298e-05, "loss": 0.124, "step": 36577 }, { "epoch": 26.797069597069598, "grad_norm": 1.109375, "learning_rate": 1.5140283490486681e-05, "loss": 0.1156, "step": 36578 }, { "epoch": 26.7978021978022, "grad_norm": 0.77734375, "learning_rate": 1.5137885803607078e-05, "loss": 0.116, "step": 36579 }, { "epoch": 26.7985347985348, "grad_norm": 0.88671875, "learning_rate": 1.5135488286158983e-05, "loss": 0.1309, "step": 36580 }, { "epoch": 26.7992673992674, "grad_norm": 2.0625, "learning_rate": 1.5133090938153863e-05, "loss": 0.1441, "step": 36581 }, { "epoch": 26.8, "grad_norm": 1.1171875, "learning_rate": 1.513069375960323e-05, "loss": 0.095, "step": 36582 }, { "epoch": 26.8007326007326, "grad_norm": 0.6796875, "learning_rate": 1.512829675051854e-05, "loss": 0.1006, "step": 36583 }, { "epoch": 26.801465201465202, "grad_norm": 1.1953125, "learning_rate": 1.5125899910911287e-05, "loss": 0.1308, "step": 36584 }, { "epoch": 26.802197802197803, "grad_norm": 0.67578125, "learning_rate": 1.5123503240792966e-05, "loss": 0.1019, "step": 36585 }, { "epoch": 26.802930402930404, "grad_norm": 0.8984375, "learning_rate": 1.5121106740175065e-05, "loss": 0.1266, "step": 36586 }, { "epoch": 26.803663003663004, "grad_norm": 1.4765625, "learning_rate": 1.511871040906904e-05, "loss": 0.1202, "step": 36587 }, { "epoch": 26.804395604395605, "grad_norm": 1.1875, "learning_rate": 1.5116314247486391e-05, "loss": 0.1191, "step": 36588 }, { "epoch": 26.805128205128206, "grad_norm": 0.84375, "learning_rate": 1.51139182554386e-05, "loss": 0.1182, "step": 36589 }, { "epoch": 26.805860805860807, "grad_norm": 0.5703125, "learning_rate": 1.5111522432937153e-05, "loss": 0.0958, "step": 36590 }, { "epoch": 26.806593406593407, "grad_norm": 1.34375, "learning_rate": 1.5109126779993502e-05, "loss": 0.1259, "step": 36591 }, { "epoch": 26.807326007326008, "grad_norm": 1.765625, "learning_rate": 1.5106731296619148e-05, "loss": 0.1291, "step": 36592 }, { "epoch": 26.80805860805861, "grad_norm": 0.8515625, "learning_rate": 1.510433598282557e-05, "loss": 0.1191, "step": 36593 }, { "epoch": 26.80879120879121, "grad_norm": 1.5625, "learning_rate": 1.5101940838624217e-05, "loss": 0.1425, "step": 36594 }, { "epoch": 26.80952380952381, "grad_norm": 0.96875, "learning_rate": 1.509954586402661e-05, "loss": 0.1127, "step": 36595 }, { "epoch": 26.81025641025641, "grad_norm": 0.69140625, "learning_rate": 1.5097151059044193e-05, "loss": 0.1417, "step": 36596 }, { "epoch": 26.81098901098901, "grad_norm": 0.8515625, "learning_rate": 1.5094756423688455e-05, "loss": 0.1176, "step": 36597 }, { "epoch": 26.811721611721612, "grad_norm": 1.125, "learning_rate": 1.509236195797085e-05, "loss": 0.1295, "step": 36598 }, { "epoch": 26.812454212454213, "grad_norm": 1.21875, "learning_rate": 1.5089967661902857e-05, "loss": 0.1074, "step": 36599 }, { "epoch": 26.813186813186814, "grad_norm": 0.73828125, "learning_rate": 1.5087573535495961e-05, "loss": 0.1102, "step": 36600 }, { "epoch": 26.813919413919415, "grad_norm": 0.7265625, "learning_rate": 1.5085179578761634e-05, "loss": 0.1286, "step": 36601 }, { "epoch": 26.814652014652015, "grad_norm": 1.0390625, "learning_rate": 1.508278579171133e-05, "loss": 0.1333, "step": 36602 }, { "epoch": 26.815384615384616, "grad_norm": 0.8671875, "learning_rate": 1.5080392174356516e-05, "loss": 0.1076, "step": 36603 }, { "epoch": 26.816117216117217, "grad_norm": 2.0625, "learning_rate": 1.5077998726708674e-05, "loss": 0.1295, "step": 36604 }, { "epoch": 26.816849816849818, "grad_norm": 0.6328125, "learning_rate": 1.5075605448779276e-05, "loss": 0.1164, "step": 36605 }, { "epoch": 26.81758241758242, "grad_norm": 0.67578125, "learning_rate": 1.507321234057979e-05, "loss": 0.1092, "step": 36606 }, { "epoch": 26.81831501831502, "grad_norm": 0.88671875, "learning_rate": 1.5070819402121655e-05, "loss": 0.1377, "step": 36607 }, { "epoch": 26.81904761904762, "grad_norm": 0.953125, "learning_rate": 1.5068426633416375e-05, "loss": 0.1123, "step": 36608 }, { "epoch": 26.81978021978022, "grad_norm": 1.34375, "learning_rate": 1.5066034034475378e-05, "loss": 0.0989, "step": 36609 }, { "epoch": 26.82051282051282, "grad_norm": 1.0, "learning_rate": 1.5063641605310144e-05, "loss": 0.1161, "step": 36610 }, { "epoch": 26.821245421245422, "grad_norm": 0.54296875, "learning_rate": 1.5061249345932135e-05, "loss": 0.0971, "step": 36611 }, { "epoch": 26.821978021978023, "grad_norm": 1.390625, "learning_rate": 1.5058857256352832e-05, "loss": 0.1327, "step": 36612 }, { "epoch": 26.822710622710623, "grad_norm": 1.0546875, "learning_rate": 1.5056465336583664e-05, "loss": 0.1359, "step": 36613 }, { "epoch": 26.823443223443224, "grad_norm": 1.0234375, "learning_rate": 1.5054073586636103e-05, "loss": 0.1564, "step": 36614 }, { "epoch": 26.824175824175825, "grad_norm": 1.3203125, "learning_rate": 1.5051682006521611e-05, "loss": 0.1706, "step": 36615 }, { "epoch": 26.824908424908426, "grad_norm": 1.3671875, "learning_rate": 1.5049290596251666e-05, "loss": 0.1127, "step": 36616 }, { "epoch": 26.825641025641026, "grad_norm": 0.98046875, "learning_rate": 1.504689935583769e-05, "loss": 0.1372, "step": 36617 }, { "epoch": 26.826373626373627, "grad_norm": 0.92578125, "learning_rate": 1.5044508285291162e-05, "loss": 0.1252, "step": 36618 }, { "epoch": 26.827106227106228, "grad_norm": 1.0, "learning_rate": 1.5042117384623544e-05, "loss": 0.1265, "step": 36619 }, { "epoch": 26.82783882783883, "grad_norm": 0.93359375, "learning_rate": 1.5039726653846247e-05, "loss": 0.1253, "step": 36620 }, { "epoch": 26.82857142857143, "grad_norm": 1.796875, "learning_rate": 1.5037336092970803e-05, "loss": 0.13, "step": 36621 }, { "epoch": 26.82930402930403, "grad_norm": 1.0078125, "learning_rate": 1.5034945702008601e-05, "loss": 0.1245, "step": 36622 }, { "epoch": 26.83003663003663, "grad_norm": 1.015625, "learning_rate": 1.5032555480971133e-05, "loss": 0.1083, "step": 36623 }, { "epoch": 26.83076923076923, "grad_norm": 1.203125, "learning_rate": 1.5030165429869819e-05, "loss": 0.0995, "step": 36624 }, { "epoch": 26.831501831501832, "grad_norm": 0.8046875, "learning_rate": 1.5027775548716122e-05, "loss": 0.1043, "step": 36625 }, { "epoch": 26.832234432234433, "grad_norm": 1.03125, "learning_rate": 1.5025385837521502e-05, "loss": 0.1238, "step": 36626 }, { "epoch": 26.832967032967034, "grad_norm": 1.265625, "learning_rate": 1.5022996296297423e-05, "loss": 0.1466, "step": 36627 }, { "epoch": 26.833699633699634, "grad_norm": 1.3046875, "learning_rate": 1.5020606925055295e-05, "loss": 0.1181, "step": 36628 }, { "epoch": 26.834432234432235, "grad_norm": 0.75, "learning_rate": 1.5018217723806588e-05, "loss": 0.1124, "step": 36629 }, { "epoch": 26.835164835164836, "grad_norm": 1.15625, "learning_rate": 1.5015828692562746e-05, "loss": 0.1209, "step": 36630 }, { "epoch": 26.835897435897436, "grad_norm": 1.265625, "learning_rate": 1.5013439831335219e-05, "loss": 0.112, "step": 36631 }, { "epoch": 26.836630036630037, "grad_norm": 1.1796875, "learning_rate": 1.5011051140135462e-05, "loss": 0.1133, "step": 36632 }, { "epoch": 26.837362637362638, "grad_norm": 0.9921875, "learning_rate": 1.5008662618974903e-05, "loss": 0.107, "step": 36633 }, { "epoch": 26.83809523809524, "grad_norm": 1.4296875, "learning_rate": 1.5006274267864997e-05, "loss": 0.1209, "step": 36634 }, { "epoch": 26.83882783882784, "grad_norm": 0.80859375, "learning_rate": 1.500388608681718e-05, "loss": 0.0972, "step": 36635 }, { "epoch": 26.83956043956044, "grad_norm": 1.09375, "learning_rate": 1.5001498075842888e-05, "loss": 0.1485, "step": 36636 }, { "epoch": 26.84029304029304, "grad_norm": 0.7578125, "learning_rate": 1.4999110234953572e-05, "loss": 0.1143, "step": 36637 }, { "epoch": 26.84102564102564, "grad_norm": 1.1796875, "learning_rate": 1.4996722564160696e-05, "loss": 0.13, "step": 36638 }, { "epoch": 26.841758241758242, "grad_norm": 0.828125, "learning_rate": 1.4994335063475653e-05, "loss": 0.1203, "step": 36639 }, { "epoch": 26.842490842490843, "grad_norm": 1.3359375, "learning_rate": 1.4991947732909908e-05, "loss": 0.1161, "step": 36640 }, { "epoch": 26.843223443223444, "grad_norm": 0.96875, "learning_rate": 1.49895605724749e-05, "loss": 0.0985, "step": 36641 }, { "epoch": 26.843956043956045, "grad_norm": 0.875, "learning_rate": 1.4987173582182081e-05, "loss": 0.1238, "step": 36642 }, { "epoch": 26.844688644688645, "grad_norm": 0.95703125, "learning_rate": 1.4984786762042857e-05, "loss": 0.1316, "step": 36643 }, { "epoch": 26.845421245421246, "grad_norm": 1.078125, "learning_rate": 1.4982400112068674e-05, "loss": 0.1231, "step": 36644 }, { "epoch": 26.846153846153847, "grad_norm": 0.9140625, "learning_rate": 1.4980013632270996e-05, "loss": 0.1152, "step": 36645 }, { "epoch": 26.846886446886447, "grad_norm": 1.0625, "learning_rate": 1.4977627322661195e-05, "loss": 0.1116, "step": 36646 }, { "epoch": 26.847619047619048, "grad_norm": 0.71875, "learning_rate": 1.4975241183250771e-05, "loss": 0.1133, "step": 36647 }, { "epoch": 26.84835164835165, "grad_norm": 1.28125, "learning_rate": 1.4972855214051118e-05, "loss": 0.1132, "step": 36648 }, { "epoch": 26.84908424908425, "grad_norm": 1.46875, "learning_rate": 1.4970469415073693e-05, "loss": 0.1554, "step": 36649 }, { "epoch": 26.84981684981685, "grad_norm": 1.671875, "learning_rate": 1.4968083786329897e-05, "loss": 0.1007, "step": 36650 }, { "epoch": 26.85054945054945, "grad_norm": 1.09375, "learning_rate": 1.4965698327831176e-05, "loss": 0.1113, "step": 36651 }, { "epoch": 26.851282051282052, "grad_norm": 1.0078125, "learning_rate": 1.4963313039588961e-05, "loss": 0.1362, "step": 36652 }, { "epoch": 26.852014652014653, "grad_norm": 1.03125, "learning_rate": 1.496092792161469e-05, "loss": 0.1022, "step": 36653 }, { "epoch": 26.852747252747253, "grad_norm": 1.0859375, "learning_rate": 1.4958542973919768e-05, "loss": 0.1271, "step": 36654 }, { "epoch": 26.853479853479854, "grad_norm": 0.8359375, "learning_rate": 1.4956158196515634e-05, "loss": 0.1117, "step": 36655 }, { "epoch": 26.854212454212455, "grad_norm": 0.890625, "learning_rate": 1.4953773589413714e-05, "loss": 0.1082, "step": 36656 }, { "epoch": 26.854945054945055, "grad_norm": 1.0546875, "learning_rate": 1.4951389152625436e-05, "loss": 0.1111, "step": 36657 }, { "epoch": 26.855677655677656, "grad_norm": 1.1015625, "learning_rate": 1.4949004886162235e-05, "loss": 0.1451, "step": 36658 }, { "epoch": 26.856410256410257, "grad_norm": 0.89453125, "learning_rate": 1.4946620790035504e-05, "loss": 0.1239, "step": 36659 }, { "epoch": 26.857142857142858, "grad_norm": 0.921875, "learning_rate": 1.4944236864256703e-05, "loss": 0.1611, "step": 36660 }, { "epoch": 26.85787545787546, "grad_norm": 1.1796875, "learning_rate": 1.4941853108837223e-05, "loss": 0.1027, "step": 36661 }, { "epoch": 26.85860805860806, "grad_norm": 1.6796875, "learning_rate": 1.4939469523788493e-05, "loss": 0.1434, "step": 36662 }, { "epoch": 26.85934065934066, "grad_norm": 2.15625, "learning_rate": 1.4937086109121941e-05, "loss": 0.1439, "step": 36663 }, { "epoch": 26.86007326007326, "grad_norm": 1.109375, "learning_rate": 1.4934702864848998e-05, "loss": 0.1046, "step": 36664 }, { "epoch": 26.86080586080586, "grad_norm": 1.2890625, "learning_rate": 1.4932319790981053e-05, "loss": 0.1125, "step": 36665 }, { "epoch": 26.861538461538462, "grad_norm": 1.125, "learning_rate": 1.4929936887529541e-05, "loss": 0.1111, "step": 36666 }, { "epoch": 26.862271062271063, "grad_norm": 0.6953125, "learning_rate": 1.4927554154505882e-05, "loss": 0.1357, "step": 36667 }, { "epoch": 26.863003663003663, "grad_norm": 0.875, "learning_rate": 1.4925171591921499e-05, "loss": 0.1262, "step": 36668 }, { "epoch": 26.863736263736264, "grad_norm": 1.0, "learning_rate": 1.492278919978778e-05, "loss": 0.1726, "step": 36669 }, { "epoch": 26.864468864468865, "grad_norm": 1.0234375, "learning_rate": 1.4920406978116161e-05, "loss": 0.115, "step": 36670 }, { "epoch": 26.865201465201466, "grad_norm": 1.6171875, "learning_rate": 1.4918024926918067e-05, "loss": 0.1342, "step": 36671 }, { "epoch": 26.865934065934066, "grad_norm": 1.1328125, "learning_rate": 1.4915643046204869e-05, "loss": 0.126, "step": 36672 }, { "epoch": 26.866666666666667, "grad_norm": 1.03125, "learning_rate": 1.491326133598803e-05, "loss": 0.134, "step": 36673 }, { "epoch": 26.867399267399268, "grad_norm": 0.80078125, "learning_rate": 1.4910879796278925e-05, "loss": 0.115, "step": 36674 }, { "epoch": 26.86813186813187, "grad_norm": 1.3359375, "learning_rate": 1.4908498427088999e-05, "loss": 0.1104, "step": 36675 }, { "epoch": 26.86886446886447, "grad_norm": 1.3984375, "learning_rate": 1.4906117228429623e-05, "loss": 0.1411, "step": 36676 }, { "epoch": 26.86959706959707, "grad_norm": 1.5546875, "learning_rate": 1.490373620031222e-05, "loss": 0.1473, "step": 36677 }, { "epoch": 26.87032967032967, "grad_norm": 1.5390625, "learning_rate": 1.490135534274821e-05, "loss": 0.1442, "step": 36678 }, { "epoch": 26.87106227106227, "grad_norm": 1.1328125, "learning_rate": 1.4898974655748999e-05, "loss": 0.1053, "step": 36679 }, { "epoch": 26.871794871794872, "grad_norm": 1.09375, "learning_rate": 1.4896594139325975e-05, "loss": 0.1487, "step": 36680 }, { "epoch": 26.872527472527473, "grad_norm": 0.71875, "learning_rate": 1.4894213793490564e-05, "loss": 0.1336, "step": 36681 }, { "epoch": 26.873260073260074, "grad_norm": 0.89453125, "learning_rate": 1.4891833618254155e-05, "loss": 0.1544, "step": 36682 }, { "epoch": 26.873992673992674, "grad_norm": 1.3125, "learning_rate": 1.4889453613628165e-05, "loss": 0.1224, "step": 36683 }, { "epoch": 26.874725274725275, "grad_norm": 1.15625, "learning_rate": 1.4887073779624003e-05, "loss": 0.1175, "step": 36684 }, { "epoch": 26.875457875457876, "grad_norm": 0.62890625, "learning_rate": 1.4884694116253051e-05, "loss": 0.1042, "step": 36685 }, { "epoch": 26.876190476190477, "grad_norm": 1.328125, "learning_rate": 1.4882314623526736e-05, "loss": 0.1328, "step": 36686 }, { "epoch": 26.876923076923077, "grad_norm": 0.828125, "learning_rate": 1.4879935301456424e-05, "loss": 0.1066, "step": 36687 }, { "epoch": 26.877655677655678, "grad_norm": 1.453125, "learning_rate": 1.4877556150053543e-05, "loss": 0.1034, "step": 36688 }, { "epoch": 26.87838827838828, "grad_norm": 0.78125, "learning_rate": 1.4875177169329479e-05, "loss": 0.1219, "step": 36689 }, { "epoch": 26.87912087912088, "grad_norm": 1.1328125, "learning_rate": 1.4872798359295652e-05, "loss": 0.1328, "step": 36690 }, { "epoch": 26.87985347985348, "grad_norm": 1.03125, "learning_rate": 1.4870419719963427e-05, "loss": 0.1393, "step": 36691 }, { "epoch": 26.88058608058608, "grad_norm": 0.8203125, "learning_rate": 1.4868041251344223e-05, "loss": 0.1082, "step": 36692 }, { "epoch": 26.88131868131868, "grad_norm": 1.0859375, "learning_rate": 1.4865662953449425e-05, "loss": 0.1019, "step": 36693 }, { "epoch": 26.882051282051282, "grad_norm": 1.1015625, "learning_rate": 1.4863284826290439e-05, "loss": 0.1329, "step": 36694 }, { "epoch": 26.882783882783883, "grad_norm": 2.0, "learning_rate": 1.486090686987867e-05, "loss": 0.1357, "step": 36695 }, { "epoch": 26.883516483516484, "grad_norm": 0.86328125, "learning_rate": 1.4858529084225476e-05, "loss": 0.1004, "step": 36696 }, { "epoch": 26.884249084249085, "grad_norm": 0.765625, "learning_rate": 1.4856151469342284e-05, "loss": 0.1064, "step": 36697 }, { "epoch": 26.884981684981685, "grad_norm": 1.2265625, "learning_rate": 1.4853774025240444e-05, "loss": 0.1313, "step": 36698 }, { "epoch": 26.885714285714286, "grad_norm": 1.2421875, "learning_rate": 1.4851396751931402e-05, "loss": 0.1316, "step": 36699 }, { "epoch": 26.886446886446887, "grad_norm": 0.80078125, "learning_rate": 1.4849019649426505e-05, "loss": 0.1083, "step": 36700 }, { "epoch": 26.887179487179488, "grad_norm": 1.234375, "learning_rate": 1.484664271773718e-05, "loss": 0.1224, "step": 36701 }, { "epoch": 26.88791208791209, "grad_norm": 1.0625, "learning_rate": 1.4844265956874769e-05, "loss": 0.1126, "step": 36702 }, { "epoch": 26.88864468864469, "grad_norm": 0.59375, "learning_rate": 1.4841889366850695e-05, "loss": 0.1088, "step": 36703 }, { "epoch": 26.88937728937729, "grad_norm": 0.97265625, "learning_rate": 1.4839512947676328e-05, "loss": 0.1118, "step": 36704 }, { "epoch": 26.89010989010989, "grad_norm": 1.09375, "learning_rate": 1.4837136699363077e-05, "loss": 0.1065, "step": 36705 }, { "epoch": 26.89084249084249, "grad_norm": 1.1640625, "learning_rate": 1.48347606219223e-05, "loss": 0.1275, "step": 36706 }, { "epoch": 26.891575091575092, "grad_norm": 0.640625, "learning_rate": 1.4832384715365394e-05, "loss": 0.1119, "step": 36707 }, { "epoch": 26.892307692307693, "grad_norm": 1.0390625, "learning_rate": 1.4830008979703739e-05, "loss": 0.1437, "step": 36708 }, { "epoch": 26.893040293040293, "grad_norm": 0.98828125, "learning_rate": 1.4827633414948715e-05, "loss": 0.1027, "step": 36709 }, { "epoch": 26.893772893772894, "grad_norm": 0.8125, "learning_rate": 1.4825258021111734e-05, "loss": 0.1144, "step": 36710 }, { "epoch": 26.894505494505495, "grad_norm": 0.75, "learning_rate": 1.482288279820413e-05, "loss": 0.1362, "step": 36711 }, { "epoch": 26.895238095238096, "grad_norm": 0.8828125, "learning_rate": 1.4820507746237328e-05, "loss": 0.1258, "step": 36712 }, { "epoch": 26.895970695970696, "grad_norm": 0.9375, "learning_rate": 1.4818132865222667e-05, "loss": 0.1036, "step": 36713 }, { "epoch": 26.896703296703297, "grad_norm": 0.76953125, "learning_rate": 1.4815758155171547e-05, "loss": 0.0962, "step": 36714 }, { "epoch": 26.897435897435898, "grad_norm": 0.64453125, "learning_rate": 1.4813383616095345e-05, "loss": 0.086, "step": 36715 }, { "epoch": 26.8981684981685, "grad_norm": 1.1953125, "learning_rate": 1.481100924800545e-05, "loss": 0.1206, "step": 36716 }, { "epoch": 26.8989010989011, "grad_norm": 1.09375, "learning_rate": 1.4808635050913214e-05, "loss": 0.1509, "step": 36717 }, { "epoch": 26.8996336996337, "grad_norm": 0.65625, "learning_rate": 1.4806261024830021e-05, "loss": 0.1159, "step": 36718 }, { "epoch": 26.9003663003663, "grad_norm": 1.171875, "learning_rate": 1.480388716976725e-05, "loss": 0.1199, "step": 36719 }, { "epoch": 26.9010989010989, "grad_norm": 1.0078125, "learning_rate": 1.4801513485736274e-05, "loss": 0.1029, "step": 36720 }, { "epoch": 26.901831501831502, "grad_norm": 0.90234375, "learning_rate": 1.4799139972748483e-05, "loss": 0.1413, "step": 36721 }, { "epoch": 26.902564102564103, "grad_norm": 0.8828125, "learning_rate": 1.4796766630815211e-05, "loss": 0.1235, "step": 36722 }, { "epoch": 26.903296703296704, "grad_norm": 1.0, "learning_rate": 1.479439345994787e-05, "loss": 0.1242, "step": 36723 }, { "epoch": 26.904029304029304, "grad_norm": 0.90625, "learning_rate": 1.4792020460157782e-05, "loss": 0.1138, "step": 36724 }, { "epoch": 26.904761904761905, "grad_norm": 1.015625, "learning_rate": 1.4789647631456377e-05, "loss": 0.1104, "step": 36725 }, { "epoch": 26.905494505494506, "grad_norm": 0.7734375, "learning_rate": 1.4787274973854976e-05, "loss": 0.1374, "step": 36726 }, { "epoch": 26.906227106227107, "grad_norm": 0.89453125, "learning_rate": 1.4784902487364977e-05, "loss": 0.1341, "step": 36727 }, { "epoch": 26.906959706959707, "grad_norm": 1.1015625, "learning_rate": 1.4782530171997722e-05, "loss": 0.1441, "step": 36728 }, { "epoch": 26.907692307692308, "grad_norm": 0.94140625, "learning_rate": 1.4780158027764593e-05, "loss": 0.1298, "step": 36729 }, { "epoch": 26.90842490842491, "grad_norm": 1.078125, "learning_rate": 1.4777786054676945e-05, "loss": 0.1405, "step": 36730 }, { "epoch": 26.90915750915751, "grad_norm": 1.234375, "learning_rate": 1.4775414252746176e-05, "loss": 0.0831, "step": 36731 }, { "epoch": 26.90989010989011, "grad_norm": 1.1953125, "learning_rate": 1.4773042621983602e-05, "loss": 0.1187, "step": 36732 }, { "epoch": 26.91062271062271, "grad_norm": 0.75390625, "learning_rate": 1.4770671162400615e-05, "loss": 0.1159, "step": 36733 }, { "epoch": 26.91135531135531, "grad_norm": 1.5, "learning_rate": 1.4768299874008568e-05, "loss": 0.1385, "step": 36734 }, { "epoch": 26.912087912087912, "grad_norm": 0.9765625, "learning_rate": 1.4765928756818826e-05, "loss": 0.1283, "step": 36735 }, { "epoch": 26.912820512820513, "grad_norm": 0.69140625, "learning_rate": 1.4763557810842768e-05, "loss": 0.1108, "step": 36736 }, { "epoch": 26.913553113553114, "grad_norm": 0.69140625, "learning_rate": 1.476118703609172e-05, "loss": 0.1087, "step": 36737 }, { "epoch": 26.914285714285715, "grad_norm": 0.734375, "learning_rate": 1.4758816432577069e-05, "loss": 0.1136, "step": 36738 }, { "epoch": 26.915018315018315, "grad_norm": 1.3125, "learning_rate": 1.4756446000310149e-05, "loss": 0.1169, "step": 36739 }, { "epoch": 26.915750915750916, "grad_norm": 1.1640625, "learning_rate": 1.4754075739302333e-05, "loss": 0.1171, "step": 36740 }, { "epoch": 26.916483516483517, "grad_norm": 0.91796875, "learning_rate": 1.475170564956497e-05, "loss": 0.111, "step": 36741 }, { "epoch": 26.917216117216118, "grad_norm": 0.9140625, "learning_rate": 1.474933573110944e-05, "loss": 0.1225, "step": 36742 }, { "epoch": 26.91794871794872, "grad_norm": 0.96484375, "learning_rate": 1.4746965983947059e-05, "loss": 0.1431, "step": 36743 }, { "epoch": 26.91868131868132, "grad_norm": 1.2890625, "learning_rate": 1.4744596408089204e-05, "loss": 0.0817, "step": 36744 }, { "epoch": 26.91941391941392, "grad_norm": 1.296875, "learning_rate": 1.4742227003547223e-05, "loss": 0.1111, "step": 36745 }, { "epoch": 26.92014652014652, "grad_norm": 1.0, "learning_rate": 1.4739857770332474e-05, "loss": 0.129, "step": 36746 }, { "epoch": 26.92087912087912, "grad_norm": 1.609375, "learning_rate": 1.4737488708456315e-05, "loss": 0.0996, "step": 36747 }, { "epoch": 26.921611721611722, "grad_norm": 1.4453125, "learning_rate": 1.4735119817930079e-05, "loss": 0.1816, "step": 36748 }, { "epoch": 26.922344322344323, "grad_norm": 1.1015625, "learning_rate": 1.4732751098765138e-05, "loss": 0.1534, "step": 36749 }, { "epoch": 26.923076923076923, "grad_norm": 1.2578125, "learning_rate": 1.4730382550972799e-05, "loss": 0.1256, "step": 36750 }, { "epoch": 26.923809523809524, "grad_norm": 1.359375, "learning_rate": 1.472801417456447e-05, "loss": 0.1196, "step": 36751 }, { "epoch": 26.924542124542125, "grad_norm": 1.1171875, "learning_rate": 1.4725645969551447e-05, "loss": 0.1263, "step": 36752 }, { "epoch": 26.925274725274726, "grad_norm": 0.62109375, "learning_rate": 1.472327793594512e-05, "loss": 0.1046, "step": 36753 }, { "epoch": 26.926007326007326, "grad_norm": 1.140625, "learning_rate": 1.4720910073756794e-05, "loss": 0.1402, "step": 36754 }, { "epoch": 26.926739926739927, "grad_norm": 1.1875, "learning_rate": 1.4718542382997836e-05, "loss": 0.1088, "step": 36755 }, { "epoch": 26.927472527472528, "grad_norm": 1.34375, "learning_rate": 1.4716174863679588e-05, "loss": 0.1403, "step": 36756 }, { "epoch": 26.92820512820513, "grad_norm": 1.1796875, "learning_rate": 1.4713807515813404e-05, "loss": 0.0944, "step": 36757 }, { "epoch": 26.92893772893773, "grad_norm": 1.9765625, "learning_rate": 1.4711440339410607e-05, "loss": 0.1206, "step": 36758 }, { "epoch": 26.92967032967033, "grad_norm": 1.34375, "learning_rate": 1.4709073334482542e-05, "loss": 0.1178, "step": 36759 }, { "epoch": 26.93040293040293, "grad_norm": 0.79296875, "learning_rate": 1.4706706501040577e-05, "loss": 0.1138, "step": 36760 }, { "epoch": 26.93113553113553, "grad_norm": 1.0, "learning_rate": 1.4704339839095999e-05, "loss": 0.1365, "step": 36761 }, { "epoch": 26.931868131868132, "grad_norm": 0.94921875, "learning_rate": 1.4701973348660205e-05, "loss": 0.1288, "step": 36762 }, { "epoch": 26.932600732600733, "grad_norm": 0.85546875, "learning_rate": 1.4699607029744502e-05, "loss": 0.1081, "step": 36763 }, { "epoch": 26.933333333333334, "grad_norm": 0.8203125, "learning_rate": 1.4697240882360243e-05, "loss": 0.1276, "step": 36764 }, { "epoch": 26.934065934065934, "grad_norm": 0.76953125, "learning_rate": 1.4694874906518744e-05, "loss": 0.1197, "step": 36765 }, { "epoch": 26.934798534798535, "grad_norm": 1.1328125, "learning_rate": 1.469250910223135e-05, "loss": 0.1232, "step": 36766 }, { "epoch": 26.935531135531136, "grad_norm": 0.9140625, "learning_rate": 1.4690143469509405e-05, "loss": 0.1043, "step": 36767 }, { "epoch": 26.936263736263736, "grad_norm": 0.76171875, "learning_rate": 1.4687778008364245e-05, "loss": 0.1302, "step": 36768 }, { "epoch": 26.936996336996337, "grad_norm": 0.7578125, "learning_rate": 1.4685412718807187e-05, "loss": 0.1343, "step": 36769 }, { "epoch": 26.937728937728938, "grad_norm": 1.0078125, "learning_rate": 1.4683047600849579e-05, "loss": 0.1059, "step": 36770 }, { "epoch": 26.93846153846154, "grad_norm": 1.109375, "learning_rate": 1.4680682654502742e-05, "loss": 0.1301, "step": 36771 }, { "epoch": 26.93919413919414, "grad_norm": 1.2578125, "learning_rate": 1.4678317879778014e-05, "loss": 0.1544, "step": 36772 }, { "epoch": 26.93992673992674, "grad_norm": 0.81640625, "learning_rate": 1.467595327668674e-05, "loss": 0.1052, "step": 36773 }, { "epoch": 26.94065934065934, "grad_norm": 1.2109375, "learning_rate": 1.4673588845240216e-05, "loss": 0.0937, "step": 36774 }, { "epoch": 26.94139194139194, "grad_norm": 1.3515625, "learning_rate": 1.4671224585449808e-05, "loss": 0.1258, "step": 36775 }, { "epoch": 26.942124542124542, "grad_norm": 0.81640625, "learning_rate": 1.4668860497326794e-05, "loss": 0.0935, "step": 36776 }, { "epoch": 26.942857142857143, "grad_norm": 0.625, "learning_rate": 1.4666496580882561e-05, "loss": 0.1008, "step": 36777 }, { "epoch": 26.943589743589744, "grad_norm": 1.1640625, "learning_rate": 1.466413283612839e-05, "loss": 0.1195, "step": 36778 }, { "epoch": 26.944322344322345, "grad_norm": 1.1171875, "learning_rate": 1.4661769263075643e-05, "loss": 0.1338, "step": 36779 }, { "epoch": 26.945054945054945, "grad_norm": 1.1015625, "learning_rate": 1.4659405861735606e-05, "loss": 0.124, "step": 36780 }, { "epoch": 26.945787545787546, "grad_norm": 1.2890625, "learning_rate": 1.465704263211962e-05, "loss": 0.1456, "step": 36781 }, { "epoch": 26.946520146520147, "grad_norm": 1.3515625, "learning_rate": 1.4654679574239008e-05, "loss": 0.127, "step": 36782 }, { "epoch": 26.947252747252747, "grad_norm": 0.82421875, "learning_rate": 1.4652316688105113e-05, "loss": 0.1262, "step": 36783 }, { "epoch": 26.947985347985348, "grad_norm": 0.82421875, "learning_rate": 1.4649953973729218e-05, "loss": 0.0876, "step": 36784 }, { "epoch": 26.94871794871795, "grad_norm": 0.640625, "learning_rate": 1.4647591431122662e-05, "loss": 0.1154, "step": 36785 }, { "epoch": 26.94945054945055, "grad_norm": 0.7421875, "learning_rate": 1.4645229060296783e-05, "loss": 0.0967, "step": 36786 }, { "epoch": 26.95018315018315, "grad_norm": 1.171875, "learning_rate": 1.4642866861262855e-05, "loss": 0.1267, "step": 36787 }, { "epoch": 26.95091575091575, "grad_norm": 0.796875, "learning_rate": 1.4640504834032248e-05, "loss": 0.1121, "step": 36788 }, { "epoch": 26.951648351648352, "grad_norm": 1.4375, "learning_rate": 1.4638142978616237e-05, "loss": 0.1229, "step": 36789 }, { "epoch": 26.952380952380953, "grad_norm": 1.109375, "learning_rate": 1.4635781295026174e-05, "loss": 0.1313, "step": 36790 }, { "epoch": 26.953113553113553, "grad_norm": 0.91015625, "learning_rate": 1.4633419783273342e-05, "loss": 0.1327, "step": 36791 }, { "epoch": 26.953846153846154, "grad_norm": 0.859375, "learning_rate": 1.4631058443369069e-05, "loss": 0.114, "step": 36792 }, { "epoch": 26.954578754578755, "grad_norm": 0.92578125, "learning_rate": 1.4628697275324667e-05, "loss": 0.1236, "step": 36793 }, { "epoch": 26.955311355311355, "grad_norm": 1.0546875, "learning_rate": 1.462633627915147e-05, "loss": 0.1128, "step": 36794 }, { "epoch": 26.956043956043956, "grad_norm": 0.87890625, "learning_rate": 1.4623975454860761e-05, "loss": 0.1246, "step": 36795 }, { "epoch": 26.956776556776557, "grad_norm": 1.1015625, "learning_rate": 1.4621614802463856e-05, "loss": 0.1455, "step": 36796 }, { "epoch": 26.957509157509158, "grad_norm": 1.0703125, "learning_rate": 1.4619254321972075e-05, "loss": 0.1, "step": 36797 }, { "epoch": 26.95824175824176, "grad_norm": 0.9921875, "learning_rate": 1.4616894013396732e-05, "loss": 0.1219, "step": 36798 }, { "epoch": 26.95897435897436, "grad_norm": 1.1953125, "learning_rate": 1.4614533876749134e-05, "loss": 0.1185, "step": 36799 }, { "epoch": 26.95970695970696, "grad_norm": 1.0546875, "learning_rate": 1.4612173912040578e-05, "loss": 0.1294, "step": 36800 }, { "epoch": 26.96043956043956, "grad_norm": 0.84375, "learning_rate": 1.4609814119282388e-05, "loss": 0.1124, "step": 36801 }, { "epoch": 26.96117216117216, "grad_norm": 0.7578125, "learning_rate": 1.4607454498485831e-05, "loss": 0.1091, "step": 36802 }, { "epoch": 26.961904761904762, "grad_norm": 1.3984375, "learning_rate": 1.4605095049662277e-05, "loss": 0.1098, "step": 36803 }, { "epoch": 26.962637362637363, "grad_norm": 0.625, "learning_rate": 1.4602735772822975e-05, "loss": 0.1049, "step": 36804 }, { "epoch": 26.963369963369964, "grad_norm": 0.921875, "learning_rate": 1.4600376667979269e-05, "loss": 0.1542, "step": 36805 }, { "epoch": 26.964102564102564, "grad_norm": 0.71484375, "learning_rate": 1.4598017735142424e-05, "loss": 0.1015, "step": 36806 }, { "epoch": 26.964835164835165, "grad_norm": 0.88671875, "learning_rate": 1.4595658974323768e-05, "loss": 0.1139, "step": 36807 }, { "epoch": 26.965567765567766, "grad_norm": 1.46875, "learning_rate": 1.459330038553459e-05, "loss": 0.1485, "step": 36808 }, { "epoch": 26.966300366300366, "grad_norm": 1.4140625, "learning_rate": 1.4590941968786202e-05, "loss": 0.1074, "step": 36809 }, { "epoch": 26.967032967032967, "grad_norm": 1.265625, "learning_rate": 1.4588583724089911e-05, "loss": 0.1248, "step": 36810 }, { "epoch": 26.967765567765568, "grad_norm": 1.2265625, "learning_rate": 1.4586225651456991e-05, "loss": 0.1296, "step": 36811 }, { "epoch": 26.96849816849817, "grad_norm": 1.4609375, "learning_rate": 1.458386775089877e-05, "loss": 0.1173, "step": 36812 }, { "epoch": 26.96923076923077, "grad_norm": 1.0703125, "learning_rate": 1.45815100224265e-05, "loss": 0.1268, "step": 36813 }, { "epoch": 26.96996336996337, "grad_norm": 0.68359375, "learning_rate": 1.4579152466051537e-05, "loss": 0.1166, "step": 36814 }, { "epoch": 26.97069597069597, "grad_norm": 1.0859375, "learning_rate": 1.4576795081785128e-05, "loss": 0.1558, "step": 36815 }, { "epoch": 26.97142857142857, "grad_norm": 1.2421875, "learning_rate": 1.4574437869638602e-05, "loss": 0.126, "step": 36816 }, { "epoch": 26.972161172161172, "grad_norm": 1.0703125, "learning_rate": 1.4572080829623223e-05, "loss": 0.1141, "step": 36817 }, { "epoch": 26.972893772893773, "grad_norm": 1.5546875, "learning_rate": 1.45697239617503e-05, "loss": 0.115, "step": 36818 }, { "epoch": 26.973626373626374, "grad_norm": 1.2421875, "learning_rate": 1.4567367266031126e-05, "loss": 0.1338, "step": 36819 }, { "epoch": 26.974358974358974, "grad_norm": 1.0546875, "learning_rate": 1.4565010742477e-05, "loss": 0.1257, "step": 36820 }, { "epoch": 26.975091575091575, "grad_norm": 0.9765625, "learning_rate": 1.4562654391099196e-05, "loss": 0.1438, "step": 36821 }, { "epoch": 26.975824175824176, "grad_norm": 0.875, "learning_rate": 1.4560298211909007e-05, "loss": 0.1109, "step": 36822 }, { "epoch": 26.976556776556777, "grad_norm": 1.015625, "learning_rate": 1.4557942204917732e-05, "loss": 0.1064, "step": 36823 }, { "epoch": 26.977289377289377, "grad_norm": 1.171875, "learning_rate": 1.4555586370136656e-05, "loss": 0.0952, "step": 36824 }, { "epoch": 26.978021978021978, "grad_norm": 1.140625, "learning_rate": 1.4553230707577079e-05, "loss": 0.1299, "step": 36825 }, { "epoch": 26.97875457875458, "grad_norm": 1.5078125, "learning_rate": 1.455087521725026e-05, "loss": 0.1321, "step": 36826 }, { "epoch": 26.97948717948718, "grad_norm": 0.65234375, "learning_rate": 1.454851989916751e-05, "loss": 0.1092, "step": 36827 }, { "epoch": 26.98021978021978, "grad_norm": 0.875, "learning_rate": 1.4546164753340075e-05, "loss": 0.1218, "step": 36828 }, { "epoch": 26.98095238095238, "grad_norm": 0.984375, "learning_rate": 1.45438097797793e-05, "loss": 0.14, "step": 36829 }, { "epoch": 26.98168498168498, "grad_norm": 0.87109375, "learning_rate": 1.4541454978496417e-05, "loss": 0.1276, "step": 36830 }, { "epoch": 26.982417582417582, "grad_norm": 1.1796875, "learning_rate": 1.4539100349502746e-05, "loss": 0.119, "step": 36831 }, { "epoch": 26.983150183150183, "grad_norm": 0.8828125, "learning_rate": 1.453674589280953e-05, "loss": 0.129, "step": 36832 }, { "epoch": 26.983882783882784, "grad_norm": 0.7890625, "learning_rate": 1.4534391608428077e-05, "loss": 0.1234, "step": 36833 }, { "epoch": 26.984615384615385, "grad_norm": 1.8125, "learning_rate": 1.4532037496369661e-05, "loss": 0.1212, "step": 36834 }, { "epoch": 26.985347985347985, "grad_norm": 0.84765625, "learning_rate": 1.4529683556645554e-05, "loss": 0.1038, "step": 36835 }, { "epoch": 26.986080586080586, "grad_norm": 1.1796875, "learning_rate": 1.452732978926706e-05, "loss": 0.1274, "step": 36836 }, { "epoch": 26.986813186813187, "grad_norm": 1.1484375, "learning_rate": 1.4524976194245426e-05, "loss": 0.1222, "step": 36837 }, { "epoch": 26.987545787545788, "grad_norm": 0.89453125, "learning_rate": 1.4522622771591952e-05, "loss": 0.1175, "step": 36838 }, { "epoch": 26.98827838827839, "grad_norm": 0.78125, "learning_rate": 1.4520269521317875e-05, "loss": 0.132, "step": 36839 }, { "epoch": 26.98901098901099, "grad_norm": 0.92578125, "learning_rate": 1.4517916443434529e-05, "loss": 0.1052, "step": 36840 }, { "epoch": 26.98974358974359, "grad_norm": 0.72265625, "learning_rate": 1.451556353795314e-05, "loss": 0.1011, "step": 36841 }, { "epoch": 26.99047619047619, "grad_norm": 1.4609375, "learning_rate": 1.4513210804885014e-05, "loss": 0.1752, "step": 36842 }, { "epoch": 26.99120879120879, "grad_norm": 0.79296875, "learning_rate": 1.4510858244241397e-05, "loss": 0.1444, "step": 36843 }, { "epoch": 26.991941391941392, "grad_norm": 1.2421875, "learning_rate": 1.4508505856033569e-05, "loss": 0.1163, "step": 36844 }, { "epoch": 26.992673992673993, "grad_norm": 1.3515625, "learning_rate": 1.4506153640272808e-05, "loss": 0.1337, "step": 36845 }, { "epoch": 26.993406593406593, "grad_norm": 1.40625, "learning_rate": 1.450380159697039e-05, "loss": 0.1648, "step": 36846 }, { "epoch": 26.994139194139194, "grad_norm": 0.70703125, "learning_rate": 1.4501449726137565e-05, "loss": 0.1161, "step": 36847 }, { "epoch": 26.994871794871795, "grad_norm": 2.4375, "learning_rate": 1.4499098027785607e-05, "loss": 0.1303, "step": 36848 }, { "epoch": 26.995604395604396, "grad_norm": 1.15625, "learning_rate": 1.4496746501925792e-05, "loss": 0.1184, "step": 36849 }, { "epoch": 26.996336996336996, "grad_norm": 1.4296875, "learning_rate": 1.4494395148569387e-05, "loss": 0.153, "step": 36850 }, { "epoch": 26.997069597069597, "grad_norm": 0.9921875, "learning_rate": 1.449204396772767e-05, "loss": 0.1204, "step": 36851 }, { "epoch": 26.997802197802198, "grad_norm": 1.171875, "learning_rate": 1.4489692959411869e-05, "loss": 0.1103, "step": 36852 }, { "epoch": 26.9985347985348, "grad_norm": 1.8125, "learning_rate": 1.448734212363329e-05, "loss": 0.1269, "step": 36853 }, { "epoch": 26.9992673992674, "grad_norm": 0.52734375, "learning_rate": 1.4484991460403149e-05, "loss": 0.1307, "step": 36854 }, { "epoch": 27.0, "grad_norm": 1.109375, "learning_rate": 1.4482640969732763e-05, "loss": 0.1046, "step": 36855 }, { "epoch": 27.0007326007326, "grad_norm": 1.6328125, "learning_rate": 1.4480290651633355e-05, "loss": 0.1105, "step": 36856 }, { "epoch": 27.0014652014652, "grad_norm": 1.265625, "learning_rate": 1.4477940506116211e-05, "loss": 0.1259, "step": 36857 }, { "epoch": 27.002197802197802, "grad_norm": 0.9140625, "learning_rate": 1.4475590533192571e-05, "loss": 0.0997, "step": 36858 }, { "epoch": 27.002930402930403, "grad_norm": 1.0703125, "learning_rate": 1.4473240732873694e-05, "loss": 0.1002, "step": 36859 }, { "epoch": 27.003663003663004, "grad_norm": 1.109375, "learning_rate": 1.4470891105170856e-05, "loss": 0.1606, "step": 36860 }, { "epoch": 27.004395604395604, "grad_norm": 1.15625, "learning_rate": 1.4468541650095296e-05, "loss": 0.1143, "step": 36861 }, { "epoch": 27.005128205128205, "grad_norm": 0.73828125, "learning_rate": 1.4466192367658305e-05, "loss": 0.0998, "step": 36862 }, { "epoch": 27.005860805860806, "grad_norm": 1.0703125, "learning_rate": 1.4463843257871098e-05, "loss": 0.1469, "step": 36863 }, { "epoch": 27.006593406593407, "grad_norm": 0.87109375, "learning_rate": 1.4461494320744964e-05, "loss": 0.1177, "step": 36864 }, { "epoch": 27.007326007326007, "grad_norm": 0.625, "learning_rate": 1.4459145556291107e-05, "loss": 0.119, "step": 36865 }, { "epoch": 27.008058608058608, "grad_norm": 1.140625, "learning_rate": 1.445679696452085e-05, "loss": 0.1464, "step": 36866 }, { "epoch": 27.00879120879121, "grad_norm": 0.83203125, "learning_rate": 1.4454448545445396e-05, "loss": 0.0961, "step": 36867 }, { "epoch": 27.00952380952381, "grad_norm": 0.88671875, "learning_rate": 1.4452100299076024e-05, "loss": 0.1111, "step": 36868 }, { "epoch": 27.01025641025641, "grad_norm": 1.109375, "learning_rate": 1.4449752225423957e-05, "loss": 0.0923, "step": 36869 }, { "epoch": 27.01098901098901, "grad_norm": 1.1484375, "learning_rate": 1.4447404324500463e-05, "loss": 0.1341, "step": 36870 }, { "epoch": 27.01172161172161, "grad_norm": 1.515625, "learning_rate": 1.4445056596316795e-05, "loss": 0.1652, "step": 36871 }, { "epoch": 27.012454212454212, "grad_norm": 0.66796875, "learning_rate": 1.4442709040884206e-05, "loss": 0.1269, "step": 36872 }, { "epoch": 27.013186813186813, "grad_norm": 0.86328125, "learning_rate": 1.4440361658213921e-05, "loss": 0.139, "step": 36873 }, { "epoch": 27.013919413919414, "grad_norm": 0.7109375, "learning_rate": 1.4438014448317203e-05, "loss": 0.0904, "step": 36874 }, { "epoch": 27.014652014652015, "grad_norm": 0.8828125, "learning_rate": 1.4435667411205291e-05, "loss": 0.1199, "step": 36875 }, { "epoch": 27.015384615384615, "grad_norm": 0.98828125, "learning_rate": 1.4433320546889446e-05, "loss": 0.1252, "step": 36876 }, { "epoch": 27.016117216117216, "grad_norm": 1.2890625, "learning_rate": 1.4430973855380908e-05, "loss": 0.1238, "step": 36877 }, { "epoch": 27.016849816849817, "grad_norm": 0.6015625, "learning_rate": 1.4428627336690904e-05, "loss": 0.1086, "step": 36878 }, { "epoch": 27.017582417582418, "grad_norm": 1.1953125, "learning_rate": 1.4426280990830701e-05, "loss": 0.1305, "step": 36879 }, { "epoch": 27.01831501831502, "grad_norm": 0.7734375, "learning_rate": 1.4423934817811499e-05, "loss": 0.0978, "step": 36880 }, { "epoch": 27.01904761904762, "grad_norm": 1.0390625, "learning_rate": 1.4421588817644593e-05, "loss": 0.1174, "step": 36881 }, { "epoch": 27.01978021978022, "grad_norm": 1.0078125, "learning_rate": 1.4419242990341192e-05, "loss": 0.0992, "step": 36882 }, { "epoch": 27.02051282051282, "grad_norm": 0.60546875, "learning_rate": 1.4416897335912553e-05, "loss": 0.0925, "step": 36883 }, { "epoch": 27.02124542124542, "grad_norm": 1.6484375, "learning_rate": 1.4414551854369888e-05, "loss": 0.1549, "step": 36884 }, { "epoch": 27.021978021978022, "grad_norm": 0.97265625, "learning_rate": 1.4412206545724451e-05, "loss": 0.105, "step": 36885 }, { "epoch": 27.022710622710623, "grad_norm": 0.9140625, "learning_rate": 1.4409861409987488e-05, "loss": 0.1149, "step": 36886 }, { "epoch": 27.023443223443223, "grad_norm": 1.0390625, "learning_rate": 1.4407516447170217e-05, "loss": 0.1687, "step": 36887 }, { "epoch": 27.024175824175824, "grad_norm": 1.015625, "learning_rate": 1.44051716572839e-05, "loss": 0.1149, "step": 36888 }, { "epoch": 27.024908424908425, "grad_norm": 1.2421875, "learning_rate": 1.4402827040339742e-05, "loss": 0.1327, "step": 36889 }, { "epoch": 27.025641025641026, "grad_norm": 1.53125, "learning_rate": 1.4400482596349004e-05, "loss": 0.1618, "step": 36890 }, { "epoch": 27.026373626373626, "grad_norm": 0.7265625, "learning_rate": 1.4398138325322874e-05, "loss": 0.1111, "step": 36891 }, { "epoch": 27.027106227106227, "grad_norm": 0.6640625, "learning_rate": 1.439579422727264e-05, "loss": 0.1243, "step": 36892 }, { "epoch": 27.027838827838828, "grad_norm": 0.77734375, "learning_rate": 1.4393450302209504e-05, "loss": 0.1008, "step": 36893 }, { "epoch": 27.02857142857143, "grad_norm": 0.77734375, "learning_rate": 1.439110655014471e-05, "loss": 0.0925, "step": 36894 }, { "epoch": 27.02930402930403, "grad_norm": 1.2421875, "learning_rate": 1.4388762971089462e-05, "loss": 0.1199, "step": 36895 }, { "epoch": 27.03003663003663, "grad_norm": 0.82421875, "learning_rate": 1.4386419565055004e-05, "loss": 0.1286, "step": 36896 }, { "epoch": 27.03076923076923, "grad_norm": 0.796875, "learning_rate": 1.4384076332052568e-05, "loss": 0.1377, "step": 36897 }, { "epoch": 27.03150183150183, "grad_norm": 0.60546875, "learning_rate": 1.4381733272093395e-05, "loss": 0.1099, "step": 36898 }, { "epoch": 27.032234432234432, "grad_norm": 1.0625, "learning_rate": 1.4379390385188674e-05, "loss": 0.1157, "step": 36899 }, { "epoch": 27.032967032967033, "grad_norm": 1.09375, "learning_rate": 1.4377047671349652e-05, "loss": 0.1503, "step": 36900 }, { "epoch": 27.033699633699634, "grad_norm": 0.9921875, "learning_rate": 1.437470513058756e-05, "loss": 0.1219, "step": 36901 }, { "epoch": 27.034432234432234, "grad_norm": 1.03125, "learning_rate": 1.4372362762913608e-05, "loss": 0.1087, "step": 36902 }, { "epoch": 27.035164835164835, "grad_norm": 0.9921875, "learning_rate": 1.4370020568339046e-05, "loss": 0.1371, "step": 36903 }, { "epoch": 27.035897435897436, "grad_norm": 1.109375, "learning_rate": 1.4367678546875057e-05, "loss": 0.109, "step": 36904 }, { "epoch": 27.036630036630036, "grad_norm": 1.703125, "learning_rate": 1.43653366985329e-05, "loss": 0.141, "step": 36905 }, { "epoch": 27.037362637362637, "grad_norm": 1.515625, "learning_rate": 1.4362995023323744e-05, "loss": 0.1043, "step": 36906 }, { "epoch": 27.038095238095238, "grad_norm": 0.75, "learning_rate": 1.4360653521258871e-05, "loss": 0.1358, "step": 36907 }, { "epoch": 27.03882783882784, "grad_norm": 1.40625, "learning_rate": 1.4358312192349457e-05, "loss": 0.094, "step": 36908 }, { "epoch": 27.03956043956044, "grad_norm": 0.90625, "learning_rate": 1.4355971036606746e-05, "loss": 0.1255, "step": 36909 }, { "epoch": 27.04029304029304, "grad_norm": 1.1875, "learning_rate": 1.4353630054041937e-05, "loss": 0.1438, "step": 36910 }, { "epoch": 27.04102564102564, "grad_norm": 1.3203125, "learning_rate": 1.4351289244666247e-05, "loss": 0.1191, "step": 36911 }, { "epoch": 27.04175824175824, "grad_norm": 0.8046875, "learning_rate": 1.4348948608490896e-05, "loss": 0.1246, "step": 36912 }, { "epoch": 27.042490842490842, "grad_norm": 1.015625, "learning_rate": 1.4346608145527102e-05, "loss": 0.1079, "step": 36913 }, { "epoch": 27.043223443223443, "grad_norm": 0.953125, "learning_rate": 1.434426785578609e-05, "loss": 0.1334, "step": 36914 }, { "epoch": 27.043956043956044, "grad_norm": 1.1328125, "learning_rate": 1.4341927739279046e-05, "loss": 0.1124, "step": 36915 }, { "epoch": 27.044688644688645, "grad_norm": 1.2890625, "learning_rate": 1.4339587796017208e-05, "loss": 0.1425, "step": 36916 }, { "epoch": 27.045421245421245, "grad_norm": 0.796875, "learning_rate": 1.4337248026011748e-05, "loss": 0.1161, "step": 36917 }, { "epoch": 27.046153846153846, "grad_norm": 0.94921875, "learning_rate": 1.4334908429273935e-05, "loss": 0.1216, "step": 36918 }, { "epoch": 27.046886446886447, "grad_norm": 2.109375, "learning_rate": 1.4332569005814936e-05, "loss": 0.1435, "step": 36919 }, { "epoch": 27.047619047619047, "grad_norm": 0.58203125, "learning_rate": 1.4330229755645983e-05, "loss": 0.1115, "step": 36920 }, { "epoch": 27.048351648351648, "grad_norm": 0.78515625, "learning_rate": 1.4327890678778255e-05, "loss": 0.1148, "step": 36921 }, { "epoch": 27.04908424908425, "grad_norm": 1.0, "learning_rate": 1.4325551775222978e-05, "loss": 0.1671, "step": 36922 }, { "epoch": 27.04981684981685, "grad_norm": 1.0859375, "learning_rate": 1.4323213044991365e-05, "loss": 0.1314, "step": 36923 }, { "epoch": 27.05054945054945, "grad_norm": 1.421875, "learning_rate": 1.432087448809461e-05, "loss": 0.1258, "step": 36924 }, { "epoch": 27.05128205128205, "grad_norm": 1.03125, "learning_rate": 1.4318536104543934e-05, "loss": 0.1326, "step": 36925 }, { "epoch": 27.052014652014652, "grad_norm": 1.46875, "learning_rate": 1.4316197894350517e-05, "loss": 0.1189, "step": 36926 }, { "epoch": 27.052747252747253, "grad_norm": 1.15625, "learning_rate": 1.4313859857525574e-05, "loss": 0.1256, "step": 36927 }, { "epoch": 27.053479853479853, "grad_norm": 1.046875, "learning_rate": 1.4311521994080307e-05, "loss": 0.117, "step": 36928 }, { "epoch": 27.054212454212454, "grad_norm": 1.765625, "learning_rate": 1.4309184304025935e-05, "loss": 0.1088, "step": 36929 }, { "epoch": 27.054945054945055, "grad_norm": 0.66796875, "learning_rate": 1.4306846787373624e-05, "loss": 0.1142, "step": 36930 }, { "epoch": 27.055677655677655, "grad_norm": 1.1328125, "learning_rate": 1.4304509444134612e-05, "loss": 0.1333, "step": 36931 }, { "epoch": 27.056410256410256, "grad_norm": 1.25, "learning_rate": 1.4302172274320041e-05, "loss": 0.1022, "step": 36932 }, { "epoch": 27.057142857142857, "grad_norm": 1.1796875, "learning_rate": 1.4299835277941182e-05, "loss": 0.1118, "step": 36933 }, { "epoch": 27.057875457875458, "grad_norm": 0.94921875, "learning_rate": 1.4297498455009175e-05, "loss": 0.1329, "step": 36934 }, { "epoch": 27.05860805860806, "grad_norm": 1.0390625, "learning_rate": 1.4295161805535253e-05, "loss": 0.1344, "step": 36935 }, { "epoch": 27.05934065934066, "grad_norm": 0.9140625, "learning_rate": 1.4292825329530582e-05, "loss": 0.092, "step": 36936 }, { "epoch": 27.06007326007326, "grad_norm": 0.87109375, "learning_rate": 1.4290489027006373e-05, "loss": 0.105, "step": 36937 }, { "epoch": 27.06080586080586, "grad_norm": 0.62890625, "learning_rate": 1.4288152897973814e-05, "loss": 0.1014, "step": 36938 }, { "epoch": 27.06153846153846, "grad_norm": 0.96484375, "learning_rate": 1.4285816942444106e-05, "loss": 0.1078, "step": 36939 }, { "epoch": 27.062271062271062, "grad_norm": 1.421875, "learning_rate": 1.4283481160428453e-05, "loss": 0.1179, "step": 36940 }, { "epoch": 27.063003663003663, "grad_norm": 1.078125, "learning_rate": 1.4281145551938011e-05, "loss": 0.1346, "step": 36941 }, { "epoch": 27.063736263736264, "grad_norm": 0.85546875, "learning_rate": 1.4278810116984004e-05, "loss": 0.1086, "step": 36942 }, { "epoch": 27.064468864468864, "grad_norm": 0.984375, "learning_rate": 1.4276474855577585e-05, "loss": 0.1382, "step": 36943 }, { "epoch": 27.065201465201465, "grad_norm": 0.88671875, "learning_rate": 1.427413976772999e-05, "loss": 0.0982, "step": 36944 }, { "epoch": 27.065934065934066, "grad_norm": 1.1015625, "learning_rate": 1.4271804853452378e-05, "loss": 0.1274, "step": 36945 }, { "epoch": 27.066666666666666, "grad_norm": 1.2109375, "learning_rate": 1.4269470112755951e-05, "loss": 0.096, "step": 36946 }, { "epoch": 27.067399267399267, "grad_norm": 1.2421875, "learning_rate": 1.4267135545651869e-05, "loss": 0.1285, "step": 36947 }, { "epoch": 27.068131868131868, "grad_norm": 0.796875, "learning_rate": 1.4264801152151338e-05, "loss": 0.0969, "step": 36948 }, { "epoch": 27.06886446886447, "grad_norm": 1.1171875, "learning_rate": 1.4262466932265539e-05, "loss": 0.1315, "step": 36949 }, { "epoch": 27.06959706959707, "grad_norm": 1.1796875, "learning_rate": 1.4260132886005658e-05, "loss": 0.1397, "step": 36950 }, { "epoch": 27.07032967032967, "grad_norm": 1.1015625, "learning_rate": 1.4257799013382892e-05, "loss": 0.1293, "step": 36951 }, { "epoch": 27.07106227106227, "grad_norm": 0.69140625, "learning_rate": 1.425546531440839e-05, "loss": 0.1122, "step": 36952 }, { "epoch": 27.07179487179487, "grad_norm": 1.2890625, "learning_rate": 1.4253131789093358e-05, "loss": 0.1173, "step": 36953 }, { "epoch": 27.072527472527472, "grad_norm": 1.3515625, "learning_rate": 1.4250798437448967e-05, "loss": 0.1079, "step": 36954 }, { "epoch": 27.073260073260073, "grad_norm": 0.95703125, "learning_rate": 1.4248465259486412e-05, "loss": 0.1264, "step": 36955 }, { "epoch": 27.073992673992674, "grad_norm": 0.82421875, "learning_rate": 1.4246132255216851e-05, "loss": 0.105, "step": 36956 }, { "epoch": 27.074725274725274, "grad_norm": 1.2109375, "learning_rate": 1.4243799424651486e-05, "loss": 0.1291, "step": 36957 }, { "epoch": 27.075457875457875, "grad_norm": 0.78125, "learning_rate": 1.4241466767801451e-05, "loss": 0.1393, "step": 36958 }, { "epoch": 27.076190476190476, "grad_norm": 0.76953125, "learning_rate": 1.4239134284677982e-05, "loss": 0.1183, "step": 36959 }, { "epoch": 27.076923076923077, "grad_norm": 1.4140625, "learning_rate": 1.4236801975292208e-05, "loss": 0.1295, "step": 36960 }, { "epoch": 27.077655677655677, "grad_norm": 1.375, "learning_rate": 1.4234469839655332e-05, "loss": 0.1014, "step": 36961 }, { "epoch": 27.078388278388278, "grad_norm": 0.80859375, "learning_rate": 1.4232137877778507e-05, "loss": 0.1243, "step": 36962 }, { "epoch": 27.07912087912088, "grad_norm": 0.89453125, "learning_rate": 1.4229806089672908e-05, "loss": 0.1395, "step": 36963 }, { "epoch": 27.07985347985348, "grad_norm": 1.203125, "learning_rate": 1.4227474475349719e-05, "loss": 0.1397, "step": 36964 }, { "epoch": 27.08058608058608, "grad_norm": 1.0625, "learning_rate": 1.422514303482011e-05, "loss": 0.1171, "step": 36965 }, { "epoch": 27.08131868131868, "grad_norm": 1.1953125, "learning_rate": 1.4222811768095262e-05, "loss": 0.1228, "step": 36966 }, { "epoch": 27.08205128205128, "grad_norm": 0.6875, "learning_rate": 1.4220480675186316e-05, "loss": 0.0982, "step": 36967 }, { "epoch": 27.082783882783882, "grad_norm": 1.2890625, "learning_rate": 1.4218149756104471e-05, "loss": 0.1194, "step": 36968 }, { "epoch": 27.083516483516483, "grad_norm": 0.93359375, "learning_rate": 1.4215819010860854e-05, "loss": 0.1042, "step": 36969 }, { "epoch": 27.084249084249084, "grad_norm": 1.046875, "learning_rate": 1.4213488439466683e-05, "loss": 0.1759, "step": 36970 }, { "epoch": 27.084981684981685, "grad_norm": 0.890625, "learning_rate": 1.4211158041933093e-05, "loss": 0.1247, "step": 36971 }, { "epoch": 27.085714285714285, "grad_norm": 1.0390625, "learning_rate": 1.4208827818271267e-05, "loss": 0.1122, "step": 36972 }, { "epoch": 27.086446886446886, "grad_norm": 1.2890625, "learning_rate": 1.4206497768492353e-05, "loss": 0.1298, "step": 36973 }, { "epoch": 27.087179487179487, "grad_norm": 0.984375, "learning_rate": 1.4204167892607514e-05, "loss": 0.1108, "step": 36974 }, { "epoch": 27.087912087912088, "grad_norm": 1.171875, "learning_rate": 1.4201838190627925e-05, "loss": 0.1175, "step": 36975 }, { "epoch": 27.08864468864469, "grad_norm": 0.90234375, "learning_rate": 1.419950866256475e-05, "loss": 0.1203, "step": 36976 }, { "epoch": 27.08937728937729, "grad_norm": 1.234375, "learning_rate": 1.4197179308429156e-05, "loss": 0.1112, "step": 36977 }, { "epoch": 27.09010989010989, "grad_norm": 1.109375, "learning_rate": 1.4194850128232276e-05, "loss": 0.147, "step": 36978 }, { "epoch": 27.09084249084249, "grad_norm": 0.75, "learning_rate": 1.4192521121985292e-05, "loss": 0.1209, "step": 36979 }, { "epoch": 27.09157509157509, "grad_norm": 0.71875, "learning_rate": 1.4190192289699356e-05, "loss": 0.1029, "step": 36980 }, { "epoch": 27.092307692307692, "grad_norm": 1.0, "learning_rate": 1.4187863631385647e-05, "loss": 0.1294, "step": 36981 }, { "epoch": 27.093040293040293, "grad_norm": 0.9375, "learning_rate": 1.4185535147055281e-05, "loss": 0.1223, "step": 36982 }, { "epoch": 27.093772893772893, "grad_norm": 0.69140625, "learning_rate": 1.4183206836719459e-05, "loss": 0.0948, "step": 36983 }, { "epoch": 27.094505494505494, "grad_norm": 1.078125, "learning_rate": 1.4180878700389282e-05, "loss": 0.1224, "step": 36984 }, { "epoch": 27.095238095238095, "grad_norm": 0.86328125, "learning_rate": 1.4178550738075969e-05, "loss": 0.1084, "step": 36985 }, { "epoch": 27.095970695970696, "grad_norm": 1.03125, "learning_rate": 1.4176222949790632e-05, "loss": 0.1392, "step": 36986 }, { "epoch": 27.096703296703296, "grad_norm": 1.0078125, "learning_rate": 1.4173895335544446e-05, "loss": 0.1197, "step": 36987 }, { "epoch": 27.097435897435897, "grad_norm": 0.9765625, "learning_rate": 1.4171567895348536e-05, "loss": 0.1161, "step": 36988 }, { "epoch": 27.098168498168498, "grad_norm": 0.8359375, "learning_rate": 1.4169240629214075e-05, "loss": 0.0979, "step": 36989 }, { "epoch": 27.0989010989011, "grad_norm": 0.6171875, "learning_rate": 1.4166913537152205e-05, "loss": 0.1178, "step": 36990 }, { "epoch": 27.0996336996337, "grad_norm": 1.484375, "learning_rate": 1.4164586619174078e-05, "loss": 0.1122, "step": 36991 }, { "epoch": 27.1003663003663, "grad_norm": 0.8671875, "learning_rate": 1.4162259875290862e-05, "loss": 0.1105, "step": 36992 }, { "epoch": 27.1010989010989, "grad_norm": 0.875, "learning_rate": 1.4159933305513673e-05, "loss": 0.1584, "step": 36993 }, { "epoch": 27.1018315018315, "grad_norm": 1.5, "learning_rate": 1.4157606909853687e-05, "loss": 0.1134, "step": 36994 }, { "epoch": 27.102564102564102, "grad_norm": 0.78515625, "learning_rate": 1.4155280688322006e-05, "loss": 0.1171, "step": 36995 }, { "epoch": 27.103296703296703, "grad_norm": 1.015625, "learning_rate": 1.4152954640929842e-05, "loss": 0.123, "step": 36996 }, { "epoch": 27.104029304029304, "grad_norm": 0.5546875, "learning_rate": 1.4150628767688284e-05, "loss": 0.1083, "step": 36997 }, { "epoch": 27.104761904761904, "grad_norm": 1.0, "learning_rate": 1.4148303068608513e-05, "loss": 0.1318, "step": 36998 }, { "epoch": 27.105494505494505, "grad_norm": 0.78125, "learning_rate": 1.4145977543701642e-05, "loss": 0.0941, "step": 36999 }, { "epoch": 27.106227106227106, "grad_norm": 1.90625, "learning_rate": 1.414365219297882e-05, "loss": 0.124, "step": 37000 }, { "epoch": 27.106959706959707, "grad_norm": 1.2421875, "learning_rate": 1.4141327016451206e-05, "loss": 0.1259, "step": 37001 }, { "epoch": 27.107692307692307, "grad_norm": 0.99609375, "learning_rate": 1.4139002014129924e-05, "loss": 0.112, "step": 37002 }, { "epoch": 27.108424908424908, "grad_norm": 1.0703125, "learning_rate": 1.4136677186026132e-05, "loss": 0.1309, "step": 37003 }, { "epoch": 27.10915750915751, "grad_norm": 1.5390625, "learning_rate": 1.413435253215095e-05, "loss": 0.1122, "step": 37004 }, { "epoch": 27.10989010989011, "grad_norm": 1.1484375, "learning_rate": 1.4132028052515514e-05, "loss": 0.1289, "step": 37005 }, { "epoch": 27.11062271062271, "grad_norm": 1.203125, "learning_rate": 1.4129703747130976e-05, "loss": 0.1122, "step": 37006 }, { "epoch": 27.11135531135531, "grad_norm": 1.5859375, "learning_rate": 1.4127379616008479e-05, "loss": 0.1102, "step": 37007 }, { "epoch": 27.11208791208791, "grad_norm": 0.9765625, "learning_rate": 1.4125055659159134e-05, "loss": 0.1103, "step": 37008 }, { "epoch": 27.112820512820512, "grad_norm": 1.2421875, "learning_rate": 1.4122731876594105e-05, "loss": 0.1262, "step": 37009 }, { "epoch": 27.113553113553113, "grad_norm": 1.1640625, "learning_rate": 1.4120408268324476e-05, "loss": 0.1137, "step": 37010 }, { "epoch": 27.114285714285714, "grad_norm": 0.88671875, "learning_rate": 1.4118084834361442e-05, "loss": 0.1299, "step": 37011 }, { "epoch": 27.115018315018315, "grad_norm": 1.125, "learning_rate": 1.4115761574716095e-05, "loss": 0.1338, "step": 37012 }, { "epoch": 27.115750915750915, "grad_norm": 0.98046875, "learning_rate": 1.4113438489399598e-05, "loss": 0.092, "step": 37013 }, { "epoch": 27.116483516483516, "grad_norm": 0.94140625, "learning_rate": 1.4111115578423042e-05, "loss": 0.1394, "step": 37014 }, { "epoch": 27.117216117216117, "grad_norm": 0.88671875, "learning_rate": 1.4108792841797576e-05, "loss": 0.1174, "step": 37015 }, { "epoch": 27.117948717948718, "grad_norm": 1.4921875, "learning_rate": 1.4106470279534328e-05, "loss": 0.1369, "step": 37016 }, { "epoch": 27.11868131868132, "grad_norm": 1.21875, "learning_rate": 1.410414789164443e-05, "loss": 0.1234, "step": 37017 }, { "epoch": 27.11941391941392, "grad_norm": 0.84375, "learning_rate": 1.4101825678139024e-05, "loss": 0.1352, "step": 37018 }, { "epoch": 27.12014652014652, "grad_norm": 1.3203125, "learning_rate": 1.4099503639029207e-05, "loss": 0.1299, "step": 37019 }, { "epoch": 27.12087912087912, "grad_norm": 0.99609375, "learning_rate": 1.4097181774326128e-05, "loss": 0.118, "step": 37020 }, { "epoch": 27.12161172161172, "grad_norm": 1.2421875, "learning_rate": 1.4094860084040876e-05, "loss": 0.125, "step": 37021 }, { "epoch": 27.122344322344322, "grad_norm": 1.1171875, "learning_rate": 1.4092538568184627e-05, "loss": 0.1705, "step": 37022 }, { "epoch": 27.123076923076923, "grad_norm": 1.65625, "learning_rate": 1.4090217226768463e-05, "loss": 0.1245, "step": 37023 }, { "epoch": 27.123809523809523, "grad_norm": 1.0390625, "learning_rate": 1.4087896059803538e-05, "loss": 0.1157, "step": 37024 }, { "epoch": 27.124542124542124, "grad_norm": 0.82421875, "learning_rate": 1.4085575067300941e-05, "loss": 0.1003, "step": 37025 }, { "epoch": 27.125274725274725, "grad_norm": 1.421875, "learning_rate": 1.4083254249271807e-05, "loss": 0.1395, "step": 37026 }, { "epoch": 27.126007326007326, "grad_norm": 0.76171875, "learning_rate": 1.4080933605727253e-05, "loss": 0.0966, "step": 37027 }, { "epoch": 27.126739926739926, "grad_norm": 0.6328125, "learning_rate": 1.4078613136678407e-05, "loss": 0.1087, "step": 37028 }, { "epoch": 27.127472527472527, "grad_norm": 0.8984375, "learning_rate": 1.4076292842136396e-05, "loss": 0.1122, "step": 37029 }, { "epoch": 27.128205128205128, "grad_norm": 1.2578125, "learning_rate": 1.4073972722112302e-05, "loss": 0.1167, "step": 37030 }, { "epoch": 27.12893772893773, "grad_norm": 0.8125, "learning_rate": 1.407165277661727e-05, "loss": 0.0928, "step": 37031 }, { "epoch": 27.12967032967033, "grad_norm": 0.703125, "learning_rate": 1.4069333005662405e-05, "loss": 0.1005, "step": 37032 }, { "epoch": 27.13040293040293, "grad_norm": 1.4140625, "learning_rate": 1.4067013409258844e-05, "loss": 0.1204, "step": 37033 }, { "epoch": 27.13113553113553, "grad_norm": 1.046875, "learning_rate": 1.4064693987417662e-05, "loss": 0.1189, "step": 37034 }, { "epoch": 27.13186813186813, "grad_norm": 2.296875, "learning_rate": 1.4062374740150008e-05, "loss": 0.1206, "step": 37035 }, { "epoch": 27.132600732600732, "grad_norm": 1.1015625, "learning_rate": 1.4060055667466946e-05, "loss": 0.1287, "step": 37036 }, { "epoch": 27.133333333333333, "grad_norm": 0.7890625, "learning_rate": 1.4057736769379656e-05, "loss": 0.1193, "step": 37037 }, { "epoch": 27.134065934065934, "grad_norm": 0.89453125, "learning_rate": 1.4055418045899194e-05, "loss": 0.0943, "step": 37038 }, { "epoch": 27.134798534798534, "grad_norm": 0.85546875, "learning_rate": 1.4053099497036687e-05, "loss": 0.1253, "step": 37039 }, { "epoch": 27.135531135531135, "grad_norm": 0.87890625, "learning_rate": 1.4050781122803261e-05, "loss": 0.1118, "step": 37040 }, { "epoch": 27.136263736263736, "grad_norm": 0.90234375, "learning_rate": 1.4048462923209995e-05, "loss": 0.116, "step": 37041 }, { "epoch": 27.136996336996337, "grad_norm": 1.015625, "learning_rate": 1.404614489826801e-05, "loss": 0.1181, "step": 37042 }, { "epoch": 27.137728937728937, "grad_norm": 0.66015625, "learning_rate": 1.4043827047988406e-05, "loss": 0.1213, "step": 37043 }, { "epoch": 27.138461538461538, "grad_norm": 1.2578125, "learning_rate": 1.4041509372382312e-05, "loss": 0.1244, "step": 37044 }, { "epoch": 27.13919413919414, "grad_norm": 0.94140625, "learning_rate": 1.4039191871460803e-05, "loss": 0.1195, "step": 37045 }, { "epoch": 27.13992673992674, "grad_norm": 1.046875, "learning_rate": 1.4036874545235005e-05, "loss": 0.1476, "step": 37046 }, { "epoch": 27.14065934065934, "grad_norm": 0.828125, "learning_rate": 1.4034557393715986e-05, "loss": 0.0955, "step": 37047 }, { "epoch": 27.14139194139194, "grad_norm": 0.7734375, "learning_rate": 1.4032240416914899e-05, "loss": 0.1213, "step": 37048 }, { "epoch": 27.14212454212454, "grad_norm": 0.8203125, "learning_rate": 1.4029923614842811e-05, "loss": 0.107, "step": 37049 }, { "epoch": 27.142857142857142, "grad_norm": 0.8984375, "learning_rate": 1.4027606987510842e-05, "loss": 0.1071, "step": 37050 }, { "epoch": 27.143589743589743, "grad_norm": 0.71875, "learning_rate": 1.4025290534930063e-05, "loss": 0.0955, "step": 37051 }, { "epoch": 27.144322344322344, "grad_norm": 0.7265625, "learning_rate": 1.4022974257111601e-05, "loss": 0.119, "step": 37052 }, { "epoch": 27.145054945054945, "grad_norm": 1.2265625, "learning_rate": 1.4020658154066538e-05, "loss": 0.1425, "step": 37053 }, { "epoch": 27.145787545787545, "grad_norm": 0.69921875, "learning_rate": 1.4018342225805969e-05, "loss": 0.1169, "step": 37054 }, { "epoch": 27.146520146520146, "grad_norm": 2.015625, "learning_rate": 1.4016026472341024e-05, "loss": 0.1532, "step": 37055 }, { "epoch": 27.147252747252747, "grad_norm": 1.15625, "learning_rate": 1.4013710893682755e-05, "loss": 0.1085, "step": 37056 }, { "epoch": 27.147985347985347, "grad_norm": 1.328125, "learning_rate": 1.4011395489842275e-05, "loss": 0.1245, "step": 37057 }, { "epoch": 27.148717948717948, "grad_norm": 0.78515625, "learning_rate": 1.4009080260830679e-05, "loss": 0.119, "step": 37058 }, { "epoch": 27.14945054945055, "grad_norm": 1.5234375, "learning_rate": 1.4006765206659075e-05, "loss": 0.1226, "step": 37059 }, { "epoch": 27.15018315018315, "grad_norm": 1.015625, "learning_rate": 1.400445032733852e-05, "loss": 0.1293, "step": 37060 }, { "epoch": 27.15091575091575, "grad_norm": 1.296875, "learning_rate": 1.4002135622880137e-05, "loss": 0.1162, "step": 37061 }, { "epoch": 27.15164835164835, "grad_norm": 0.86328125, "learning_rate": 1.3999821093294977e-05, "loss": 0.132, "step": 37062 }, { "epoch": 27.152380952380952, "grad_norm": 1.125, "learning_rate": 1.3997506738594191e-05, "loss": 0.1063, "step": 37063 }, { "epoch": 27.153113553113553, "grad_norm": 1.015625, "learning_rate": 1.3995192558788814e-05, "loss": 0.1291, "step": 37064 }, { "epoch": 27.153846153846153, "grad_norm": 0.7890625, "learning_rate": 1.3992878553889951e-05, "loss": 0.1163, "step": 37065 }, { "epoch": 27.154578754578754, "grad_norm": 0.7109375, "learning_rate": 1.3990564723908705e-05, "loss": 0.1059, "step": 37066 }, { "epoch": 27.155311355311355, "grad_norm": 0.765625, "learning_rate": 1.3988251068856137e-05, "loss": 0.1119, "step": 37067 }, { "epoch": 27.156043956043955, "grad_norm": 1.3359375, "learning_rate": 1.3985937588743347e-05, "loss": 0.1189, "step": 37068 }, { "epoch": 27.156776556776556, "grad_norm": 1.1875, "learning_rate": 1.3983624283581411e-05, "loss": 0.1018, "step": 37069 }, { "epoch": 27.157509157509157, "grad_norm": 1.1953125, "learning_rate": 1.3981311153381437e-05, "loss": 0.1609, "step": 37070 }, { "epoch": 27.158241758241758, "grad_norm": 0.69140625, "learning_rate": 1.3978998198154475e-05, "loss": 0.1115, "step": 37071 }, { "epoch": 27.15897435897436, "grad_norm": 0.91796875, "learning_rate": 1.3976685417911633e-05, "loss": 0.1218, "step": 37072 }, { "epoch": 27.15970695970696, "grad_norm": 1.0390625, "learning_rate": 1.3974372812663955e-05, "loss": 0.1214, "step": 37073 }, { "epoch": 27.16043956043956, "grad_norm": 0.94140625, "learning_rate": 1.3972060382422577e-05, "loss": 0.1473, "step": 37074 }, { "epoch": 27.16117216117216, "grad_norm": 0.92578125, "learning_rate": 1.3969748127198528e-05, "loss": 0.0981, "step": 37075 }, { "epoch": 27.16190476190476, "grad_norm": 1.359375, "learning_rate": 1.3967436047002932e-05, "loss": 0.0952, "step": 37076 }, { "epoch": 27.162637362637362, "grad_norm": 0.98828125, "learning_rate": 1.3965124141846824e-05, "loss": 0.1336, "step": 37077 }, { "epoch": 27.163369963369963, "grad_norm": 0.91015625, "learning_rate": 1.3962812411741297e-05, "loss": 0.1273, "step": 37078 }, { "epoch": 27.164102564102564, "grad_norm": 0.94921875, "learning_rate": 1.3960500856697434e-05, "loss": 0.1328, "step": 37079 }, { "epoch": 27.164835164835164, "grad_norm": 1.109375, "learning_rate": 1.3958189476726305e-05, "loss": 0.1274, "step": 37080 }, { "epoch": 27.165567765567765, "grad_norm": 1.0625, "learning_rate": 1.3955878271839005e-05, "loss": 0.1284, "step": 37081 }, { "epoch": 27.166300366300366, "grad_norm": 0.66796875, "learning_rate": 1.3953567242046569e-05, "loss": 0.1176, "step": 37082 }, { "epoch": 27.167032967032966, "grad_norm": 1.4140625, "learning_rate": 1.395125638736009e-05, "loss": 0.1174, "step": 37083 }, { "epoch": 27.167765567765567, "grad_norm": 0.9375, "learning_rate": 1.3948945707790642e-05, "loss": 0.1301, "step": 37084 }, { "epoch": 27.168498168498168, "grad_norm": 0.83203125, "learning_rate": 1.3946635203349304e-05, "loss": 0.1338, "step": 37085 }, { "epoch": 27.16923076923077, "grad_norm": 1.078125, "learning_rate": 1.3944324874047124e-05, "loss": 0.1034, "step": 37086 }, { "epoch": 27.16996336996337, "grad_norm": 0.62890625, "learning_rate": 1.3942014719895206e-05, "loss": 0.1225, "step": 37087 }, { "epoch": 27.17069597069597, "grad_norm": 1.078125, "learning_rate": 1.3939704740904558e-05, "loss": 0.1289, "step": 37088 }, { "epoch": 27.17142857142857, "grad_norm": 1.0859375, "learning_rate": 1.3937394937086317e-05, "loss": 0.1063, "step": 37089 }, { "epoch": 27.17216117216117, "grad_norm": 1.0, "learning_rate": 1.3935085308451509e-05, "loss": 0.1237, "step": 37090 }, { "epoch": 27.172893772893772, "grad_norm": 0.81640625, "learning_rate": 1.393277585501121e-05, "loss": 0.1192, "step": 37091 }, { "epoch": 27.173626373626373, "grad_norm": 0.828125, "learning_rate": 1.3930466576776496e-05, "loss": 0.1223, "step": 37092 }, { "epoch": 27.174358974358974, "grad_norm": 0.70703125, "learning_rate": 1.392815747375841e-05, "loss": 0.137, "step": 37093 }, { "epoch": 27.175091575091574, "grad_norm": 0.76171875, "learning_rate": 1.3925848545968028e-05, "loss": 0.1046, "step": 37094 }, { "epoch": 27.175824175824175, "grad_norm": 0.87109375, "learning_rate": 1.3923539793416412e-05, "loss": 0.1214, "step": 37095 }, { "epoch": 27.176556776556776, "grad_norm": 0.96875, "learning_rate": 1.3921231216114639e-05, "loss": 0.1355, "step": 37096 }, { "epoch": 27.177289377289377, "grad_norm": 0.95703125, "learning_rate": 1.3918922814073743e-05, "loss": 0.1199, "step": 37097 }, { "epoch": 27.178021978021977, "grad_norm": 1.53125, "learning_rate": 1.391661458730481e-05, "loss": 0.1261, "step": 37098 }, { "epoch": 27.178754578754578, "grad_norm": 1.234375, "learning_rate": 1.3914306535818853e-05, "loss": 0.1105, "step": 37099 }, { "epoch": 27.17948717948718, "grad_norm": 1.0546875, "learning_rate": 1.3911998659626999e-05, "loss": 0.1587, "step": 37100 }, { "epoch": 27.18021978021978, "grad_norm": 0.84375, "learning_rate": 1.3909690958740251e-05, "loss": 0.1102, "step": 37101 }, { "epoch": 27.18095238095238, "grad_norm": 1.125, "learning_rate": 1.3907383433169698e-05, "loss": 0.1362, "step": 37102 }, { "epoch": 27.18168498168498, "grad_norm": 0.8046875, "learning_rate": 1.3905076082926375e-05, "loss": 0.1229, "step": 37103 }, { "epoch": 27.18241758241758, "grad_norm": 1.21875, "learning_rate": 1.3902768908021338e-05, "loss": 0.1327, "step": 37104 }, { "epoch": 27.183150183150182, "grad_norm": 1.609375, "learning_rate": 1.3900461908465657e-05, "loss": 0.1059, "step": 37105 }, { "epoch": 27.183882783882783, "grad_norm": 0.9609375, "learning_rate": 1.3898155084270373e-05, "loss": 0.1374, "step": 37106 }, { "epoch": 27.184615384615384, "grad_norm": 0.84765625, "learning_rate": 1.389584843544656e-05, "loss": 0.1239, "step": 37107 }, { "epoch": 27.185347985347985, "grad_norm": 1.8046875, "learning_rate": 1.3893541962005238e-05, "loss": 0.1221, "step": 37108 }, { "epoch": 27.186080586080585, "grad_norm": 1.078125, "learning_rate": 1.3891235663957478e-05, "loss": 0.1302, "step": 37109 }, { "epoch": 27.186813186813186, "grad_norm": 1.1484375, "learning_rate": 1.388892954131432e-05, "loss": 0.1152, "step": 37110 }, { "epoch": 27.187545787545787, "grad_norm": 0.796875, "learning_rate": 1.3886623594086837e-05, "loss": 0.1074, "step": 37111 }, { "epoch": 27.188278388278388, "grad_norm": 0.99609375, "learning_rate": 1.3884317822286045e-05, "loss": 0.1247, "step": 37112 }, { "epoch": 27.18901098901099, "grad_norm": 1.625, "learning_rate": 1.388201222592302e-05, "loss": 0.1293, "step": 37113 }, { "epoch": 27.18974358974359, "grad_norm": 1.125, "learning_rate": 1.3879706805008768e-05, "loss": 0.1527, "step": 37114 }, { "epoch": 27.19047619047619, "grad_norm": 0.9453125, "learning_rate": 1.3877401559554388e-05, "loss": 0.1259, "step": 37115 }, { "epoch": 27.19120879120879, "grad_norm": 1.0546875, "learning_rate": 1.3875096489570886e-05, "loss": 0.1091, "step": 37116 }, { "epoch": 27.19194139194139, "grad_norm": 0.79296875, "learning_rate": 1.3872791595069323e-05, "loss": 0.1101, "step": 37117 }, { "epoch": 27.192673992673992, "grad_norm": 1.390625, "learning_rate": 1.3870486876060752e-05, "loss": 0.1207, "step": 37118 }, { "epoch": 27.193406593406593, "grad_norm": 0.96484375, "learning_rate": 1.3868182332556187e-05, "loss": 0.1017, "step": 37119 }, { "epoch": 27.194139194139193, "grad_norm": 0.8203125, "learning_rate": 1.386587796456669e-05, "loss": 0.1063, "step": 37120 }, { "epoch": 27.194871794871794, "grad_norm": 1.0078125, "learning_rate": 1.3863573772103291e-05, "loss": 0.1332, "step": 37121 }, { "epoch": 27.195604395604395, "grad_norm": 1.578125, "learning_rate": 1.3861269755177061e-05, "loss": 0.1189, "step": 37122 }, { "epoch": 27.196336996336996, "grad_norm": 1.1796875, "learning_rate": 1.3858965913798993e-05, "loss": 0.1106, "step": 37123 }, { "epoch": 27.197069597069596, "grad_norm": 0.65234375, "learning_rate": 1.3856662247980168e-05, "loss": 0.1033, "step": 37124 }, { "epoch": 27.197802197802197, "grad_norm": 1.03125, "learning_rate": 1.385435875773157e-05, "loss": 0.1156, "step": 37125 }, { "epoch": 27.198534798534798, "grad_norm": 0.87109375, "learning_rate": 1.3852055443064296e-05, "loss": 0.1389, "step": 37126 }, { "epoch": 27.1992673992674, "grad_norm": 1.2421875, "learning_rate": 1.3849752303989343e-05, "loss": 0.126, "step": 37127 }, { "epoch": 27.2, "grad_norm": 1.0859375, "learning_rate": 1.3847449340517776e-05, "loss": 0.1091, "step": 37128 }, { "epoch": 27.2007326007326, "grad_norm": 0.78515625, "learning_rate": 1.3845146552660593e-05, "loss": 0.0886, "step": 37129 }, { "epoch": 27.2014652014652, "grad_norm": 0.87109375, "learning_rate": 1.3842843940428848e-05, "loss": 0.1114, "step": 37130 }, { "epoch": 27.2021978021978, "grad_norm": 0.953125, "learning_rate": 1.3840541503833574e-05, "loss": 0.1064, "step": 37131 }, { "epoch": 27.202930402930402, "grad_norm": 0.828125, "learning_rate": 1.3838239242885792e-05, "loss": 0.1119, "step": 37132 }, { "epoch": 27.203663003663003, "grad_norm": 1.6015625, "learning_rate": 1.3835937157596562e-05, "loss": 0.1275, "step": 37133 }, { "epoch": 27.204395604395604, "grad_norm": 1.53125, "learning_rate": 1.3833635247976881e-05, "loss": 0.1371, "step": 37134 }, { "epoch": 27.205128205128204, "grad_norm": 1.421875, "learning_rate": 1.3831333514037786e-05, "loss": 0.1423, "step": 37135 }, { "epoch": 27.205860805860805, "grad_norm": 1.078125, "learning_rate": 1.3829031955790308e-05, "loss": 0.1241, "step": 37136 }, { "epoch": 27.206593406593406, "grad_norm": 1.28125, "learning_rate": 1.3826730573245495e-05, "loss": 0.1406, "step": 37137 }, { "epoch": 27.207326007326007, "grad_norm": 0.8515625, "learning_rate": 1.3824429366414337e-05, "loss": 0.1244, "step": 37138 }, { "epoch": 27.208058608058607, "grad_norm": 0.90234375, "learning_rate": 1.382212833530789e-05, "loss": 0.1166, "step": 37139 }, { "epoch": 27.208791208791208, "grad_norm": 1.3671875, "learning_rate": 1.3819827479937142e-05, "loss": 0.1163, "step": 37140 }, { "epoch": 27.20952380952381, "grad_norm": 1.0, "learning_rate": 1.3817526800313172e-05, "loss": 0.1395, "step": 37141 }, { "epoch": 27.21025641025641, "grad_norm": 0.78515625, "learning_rate": 1.381522629644696e-05, "loss": 0.109, "step": 37142 }, { "epoch": 27.21098901098901, "grad_norm": 0.8671875, "learning_rate": 1.3812925968349532e-05, "loss": 0.1307, "step": 37143 }, { "epoch": 27.21172161172161, "grad_norm": 0.75, "learning_rate": 1.3810625816031944e-05, "loss": 0.1166, "step": 37144 }, { "epoch": 27.21245421245421, "grad_norm": 1.171875, "learning_rate": 1.380832583950517e-05, "loss": 0.1318, "step": 37145 }, { "epoch": 27.213186813186812, "grad_norm": 1.7734375, "learning_rate": 1.3806026038780256e-05, "loss": 0.1357, "step": 37146 }, { "epoch": 27.213919413919413, "grad_norm": 1.1171875, "learning_rate": 1.380372641386822e-05, "loss": 0.0952, "step": 37147 }, { "epoch": 27.214652014652014, "grad_norm": 0.6015625, "learning_rate": 1.3801426964780087e-05, "loss": 0.116, "step": 37148 }, { "epoch": 27.215384615384615, "grad_norm": 1.4921875, "learning_rate": 1.379912769152685e-05, "loss": 0.1447, "step": 37149 }, { "epoch": 27.216117216117215, "grad_norm": 1.125, "learning_rate": 1.3796828594119563e-05, "loss": 0.1222, "step": 37150 }, { "epoch": 27.216849816849816, "grad_norm": 1.4453125, "learning_rate": 1.379452967256918e-05, "loss": 0.1078, "step": 37151 }, { "epoch": 27.217582417582417, "grad_norm": 1.2578125, "learning_rate": 1.379223092688679e-05, "loss": 0.0933, "step": 37152 }, { "epoch": 27.218315018315018, "grad_norm": 1.109375, "learning_rate": 1.3789932357083355e-05, "loss": 0.123, "step": 37153 }, { "epoch": 27.21904761904762, "grad_norm": 1.1328125, "learning_rate": 1.3787633963169916e-05, "loss": 0.1402, "step": 37154 }, { "epoch": 27.21978021978022, "grad_norm": 0.9375, "learning_rate": 1.3785335745157463e-05, "loss": 0.1208, "step": 37155 }, { "epoch": 27.22051282051282, "grad_norm": 0.875, "learning_rate": 1.3783037703057026e-05, "loss": 0.1038, "step": 37156 }, { "epoch": 27.22124542124542, "grad_norm": 1.875, "learning_rate": 1.3780739836879599e-05, "loss": 0.1351, "step": 37157 }, { "epoch": 27.22197802197802, "grad_norm": 0.79296875, "learning_rate": 1.3778442146636206e-05, "loss": 0.1052, "step": 37158 }, { "epoch": 27.222710622710622, "grad_norm": 0.90234375, "learning_rate": 1.3776144632337865e-05, "loss": 0.1053, "step": 37159 }, { "epoch": 27.223443223443223, "grad_norm": 1.671875, "learning_rate": 1.3773847293995553e-05, "loss": 0.1203, "step": 37160 }, { "epoch": 27.224175824175823, "grad_norm": 1.4609375, "learning_rate": 1.3771550131620303e-05, "loss": 0.1191, "step": 37161 }, { "epoch": 27.224908424908424, "grad_norm": 0.6640625, "learning_rate": 1.376925314522311e-05, "loss": 0.1197, "step": 37162 }, { "epoch": 27.225641025641025, "grad_norm": 1.09375, "learning_rate": 1.3766956334814999e-05, "loss": 0.0911, "step": 37163 }, { "epoch": 27.226373626373626, "grad_norm": 1.0078125, "learning_rate": 1.3764659700406938e-05, "loss": 0.1343, "step": 37164 }, { "epoch": 27.227106227106226, "grad_norm": 1.0703125, "learning_rate": 1.376236324200997e-05, "loss": 0.1121, "step": 37165 }, { "epoch": 27.227838827838827, "grad_norm": 0.875, "learning_rate": 1.3760066959635052e-05, "loss": 0.105, "step": 37166 }, { "epoch": 27.228571428571428, "grad_norm": 1.421875, "learning_rate": 1.3757770853293244e-05, "loss": 0.1448, "step": 37167 }, { "epoch": 27.22930402930403, "grad_norm": 1.4609375, "learning_rate": 1.37554749229955e-05, "loss": 0.1531, "step": 37168 }, { "epoch": 27.23003663003663, "grad_norm": 1.03125, "learning_rate": 1.3753179168752836e-05, "loss": 0.1073, "step": 37169 }, { "epoch": 27.23076923076923, "grad_norm": 1.453125, "learning_rate": 1.3750883590576278e-05, "loss": 0.1331, "step": 37170 }, { "epoch": 27.23150183150183, "grad_norm": 0.83203125, "learning_rate": 1.3748588188476777e-05, "loss": 0.115, "step": 37171 }, { "epoch": 27.23223443223443, "grad_norm": 0.80859375, "learning_rate": 1.3746292962465356e-05, "loss": 0.105, "step": 37172 }, { "epoch": 27.232967032967032, "grad_norm": 1.109375, "learning_rate": 1.3743997912553017e-05, "loss": 0.1125, "step": 37173 }, { "epoch": 27.233699633699633, "grad_norm": 2.21875, "learning_rate": 1.374170303875076e-05, "loss": 0.1219, "step": 37174 }, { "epoch": 27.234432234432234, "grad_norm": 1.25, "learning_rate": 1.3739408341069557e-05, "loss": 0.1272, "step": 37175 }, { "epoch": 27.235164835164834, "grad_norm": 0.99609375, "learning_rate": 1.3737113819520431e-05, "loss": 0.0952, "step": 37176 }, { "epoch": 27.235897435897435, "grad_norm": 0.796875, "learning_rate": 1.373481947411433e-05, "loss": 0.1098, "step": 37177 }, { "epoch": 27.236630036630036, "grad_norm": 1.7421875, "learning_rate": 1.3732525304862309e-05, "loss": 0.1326, "step": 37178 }, { "epoch": 27.237362637362637, "grad_norm": 0.8828125, "learning_rate": 1.3730231311775312e-05, "loss": 0.1045, "step": 37179 }, { "epoch": 27.238095238095237, "grad_norm": 0.90234375, "learning_rate": 1.3727937494864352e-05, "loss": 0.1134, "step": 37180 }, { "epoch": 27.238827838827838, "grad_norm": 1.0234375, "learning_rate": 1.3725643854140425e-05, "loss": 0.1357, "step": 37181 }, { "epoch": 27.23956043956044, "grad_norm": 0.95703125, "learning_rate": 1.3723350389614503e-05, "loss": 0.0952, "step": 37182 }, { "epoch": 27.24029304029304, "grad_norm": 0.7421875, "learning_rate": 1.3721057101297577e-05, "loss": 0.1041, "step": 37183 }, { "epoch": 27.24102564102564, "grad_norm": 1.0859375, "learning_rate": 1.3718763989200636e-05, "loss": 0.1439, "step": 37184 }, { "epoch": 27.24175824175824, "grad_norm": 0.88671875, "learning_rate": 1.3716471053334692e-05, "loss": 0.1209, "step": 37185 }, { "epoch": 27.24249084249084, "grad_norm": 0.85546875, "learning_rate": 1.3714178293710695e-05, "loss": 0.1349, "step": 37186 }, { "epoch": 27.243223443223442, "grad_norm": 1.234375, "learning_rate": 1.3711885710339642e-05, "loss": 0.1228, "step": 37187 }, { "epoch": 27.243956043956043, "grad_norm": 0.6796875, "learning_rate": 1.3709593303232526e-05, "loss": 0.0817, "step": 37188 }, { "epoch": 27.244688644688644, "grad_norm": 1.1015625, "learning_rate": 1.3707301072400342e-05, "loss": 0.1248, "step": 37189 }, { "epoch": 27.245421245421245, "grad_norm": 0.9296875, "learning_rate": 1.3705009017854035e-05, "loss": 0.1236, "step": 37190 }, { "epoch": 27.246153846153845, "grad_norm": 1.296875, "learning_rate": 1.370271713960463e-05, "loss": 0.1442, "step": 37191 }, { "epoch": 27.246886446886446, "grad_norm": 0.87890625, "learning_rate": 1.370042543766307e-05, "loss": 0.0967, "step": 37192 }, { "epoch": 27.247619047619047, "grad_norm": 1.125, "learning_rate": 1.3698133912040351e-05, "loss": 0.1172, "step": 37193 }, { "epoch": 27.248351648351647, "grad_norm": 0.88671875, "learning_rate": 1.3695842562747457e-05, "loss": 0.135, "step": 37194 }, { "epoch": 27.249084249084248, "grad_norm": 1.2890625, "learning_rate": 1.3693551389795362e-05, "loss": 0.1095, "step": 37195 }, { "epoch": 27.24981684981685, "grad_norm": 1.2578125, "learning_rate": 1.3691260393195055e-05, "loss": 0.1349, "step": 37196 }, { "epoch": 27.25054945054945, "grad_norm": 1.375, "learning_rate": 1.3688969572957494e-05, "loss": 0.1275, "step": 37197 }, { "epoch": 27.25128205128205, "grad_norm": 0.58984375, "learning_rate": 1.3686678929093666e-05, "loss": 0.1004, "step": 37198 }, { "epoch": 27.25201465201465, "grad_norm": 0.90234375, "learning_rate": 1.368438846161454e-05, "loss": 0.1283, "step": 37199 }, { "epoch": 27.252747252747252, "grad_norm": 1.1015625, "learning_rate": 1.368209817053111e-05, "loss": 0.1262, "step": 37200 }, { "epoch": 27.253479853479853, "grad_norm": 0.75390625, "learning_rate": 1.367980805585432e-05, "loss": 0.1216, "step": 37201 }, { "epoch": 27.254212454212453, "grad_norm": 0.890625, "learning_rate": 1.3677518117595168e-05, "loss": 0.118, "step": 37202 }, { "epoch": 27.254945054945054, "grad_norm": 0.81640625, "learning_rate": 1.3675228355764586e-05, "loss": 0.1229, "step": 37203 }, { "epoch": 27.255677655677655, "grad_norm": 1.078125, "learning_rate": 1.3672938770373609e-05, "loss": 0.1568, "step": 37204 }, { "epoch": 27.256410256410255, "grad_norm": 0.8984375, "learning_rate": 1.3670649361433151e-05, "loss": 0.1139, "step": 37205 }, { "epoch": 27.257142857142856, "grad_norm": 0.83203125, "learning_rate": 1.3668360128954205e-05, "loss": 0.1171, "step": 37206 }, { "epoch": 27.257875457875457, "grad_norm": 0.7890625, "learning_rate": 1.366607107294775e-05, "loss": 0.1169, "step": 37207 }, { "epoch": 27.258608058608058, "grad_norm": 0.80078125, "learning_rate": 1.3663782193424732e-05, "loss": 0.1158, "step": 37208 }, { "epoch": 27.25934065934066, "grad_norm": 0.86328125, "learning_rate": 1.3661493490396119e-05, "loss": 0.0908, "step": 37209 }, { "epoch": 27.26007326007326, "grad_norm": 1.03125, "learning_rate": 1.3659204963872883e-05, "loss": 0.1257, "step": 37210 }, { "epoch": 27.26080586080586, "grad_norm": 0.71875, "learning_rate": 1.3656916613866009e-05, "loss": 0.1138, "step": 37211 }, { "epoch": 27.26153846153846, "grad_norm": 0.66796875, "learning_rate": 1.3654628440386427e-05, "loss": 0.1095, "step": 37212 }, { "epoch": 27.26227106227106, "grad_norm": 1.328125, "learning_rate": 1.3652340443445113e-05, "loss": 0.1203, "step": 37213 }, { "epoch": 27.263003663003662, "grad_norm": 0.93359375, "learning_rate": 1.3650052623053038e-05, "loss": 0.1275, "step": 37214 }, { "epoch": 27.263736263736263, "grad_norm": 1.375, "learning_rate": 1.3647764979221166e-05, "loss": 0.1289, "step": 37215 }, { "epoch": 27.264468864468864, "grad_norm": 1.2890625, "learning_rate": 1.3645477511960443e-05, "loss": 0.1365, "step": 37216 }, { "epoch": 27.265201465201464, "grad_norm": 0.60546875, "learning_rate": 1.3643190221281841e-05, "loss": 0.1273, "step": 37217 }, { "epoch": 27.265934065934065, "grad_norm": 1.03125, "learning_rate": 1.36409031071963e-05, "loss": 0.1489, "step": 37218 }, { "epoch": 27.266666666666666, "grad_norm": 1.296875, "learning_rate": 1.3638616169714798e-05, "loss": 0.129, "step": 37219 }, { "epoch": 27.267399267399266, "grad_norm": 0.91796875, "learning_rate": 1.3636329408848283e-05, "loss": 0.1157, "step": 37220 }, { "epoch": 27.268131868131867, "grad_norm": 1.53125, "learning_rate": 1.3634042824607718e-05, "loss": 0.1385, "step": 37221 }, { "epoch": 27.268864468864468, "grad_norm": 0.81640625, "learning_rate": 1.3631756417004068e-05, "loss": 0.1409, "step": 37222 }, { "epoch": 27.26959706959707, "grad_norm": 0.94140625, "learning_rate": 1.3629470186048266e-05, "loss": 0.1193, "step": 37223 }, { "epoch": 27.27032967032967, "grad_norm": 1.0234375, "learning_rate": 1.3627184131751276e-05, "loss": 0.1155, "step": 37224 }, { "epoch": 27.27106227106227, "grad_norm": 0.76953125, "learning_rate": 1.3624898254124047e-05, "loss": 0.1064, "step": 37225 }, { "epoch": 27.27179487179487, "grad_norm": 1.1015625, "learning_rate": 1.3622612553177547e-05, "loss": 0.1373, "step": 37226 }, { "epoch": 27.27252747252747, "grad_norm": 1.1875, "learning_rate": 1.362032702892271e-05, "loss": 0.1126, "step": 37227 }, { "epoch": 27.273260073260072, "grad_norm": 0.9375, "learning_rate": 1.3618041681370502e-05, "loss": 0.1321, "step": 37228 }, { "epoch": 27.273992673992673, "grad_norm": 0.99609375, "learning_rate": 1.3615756510531835e-05, "loss": 0.1093, "step": 37229 }, { "epoch": 27.274725274725274, "grad_norm": 1.1953125, "learning_rate": 1.3613471516417726e-05, "loss": 0.1407, "step": 37230 }, { "epoch": 27.275457875457874, "grad_norm": 0.7421875, "learning_rate": 1.361118669903906e-05, "loss": 0.0947, "step": 37231 }, { "epoch": 27.276190476190475, "grad_norm": 0.87109375, "learning_rate": 1.3608902058406808e-05, "loss": 0.1048, "step": 37232 }, { "epoch": 27.276923076923076, "grad_norm": 0.7421875, "learning_rate": 1.3606617594531934e-05, "loss": 0.1216, "step": 37233 }, { "epoch": 27.277655677655677, "grad_norm": 0.99609375, "learning_rate": 1.3604333307425348e-05, "loss": 0.1534, "step": 37234 }, { "epoch": 27.278388278388277, "grad_norm": 1.0, "learning_rate": 1.3602049197098018e-05, "loss": 0.1078, "step": 37235 }, { "epoch": 27.279120879120878, "grad_norm": 1.1875, "learning_rate": 1.3599765263560882e-05, "loss": 0.1176, "step": 37236 }, { "epoch": 27.27985347985348, "grad_norm": 0.984375, "learning_rate": 1.3597481506824896e-05, "loss": 0.1042, "step": 37237 }, { "epoch": 27.28058608058608, "grad_norm": 1.375, "learning_rate": 1.359519792690098e-05, "loss": 0.0896, "step": 37238 }, { "epoch": 27.28131868131868, "grad_norm": 1.015625, "learning_rate": 1.3592914523800082e-05, "loss": 0.126, "step": 37239 }, { "epoch": 27.28205128205128, "grad_norm": 0.8125, "learning_rate": 1.3590631297533147e-05, "loss": 0.1084, "step": 37240 }, { "epoch": 27.282783882783882, "grad_norm": 0.82421875, "learning_rate": 1.358834824811113e-05, "loss": 0.1169, "step": 37241 }, { "epoch": 27.283516483516483, "grad_norm": 1.09375, "learning_rate": 1.3586065375544943e-05, "loss": 0.1516, "step": 37242 }, { "epoch": 27.284249084249083, "grad_norm": 0.984375, "learning_rate": 1.3583782679845541e-05, "loss": 0.1311, "step": 37243 }, { "epoch": 27.284981684981684, "grad_norm": 1.3046875, "learning_rate": 1.3581500161023847e-05, "loss": 0.1236, "step": 37244 }, { "epoch": 27.285714285714285, "grad_norm": 0.87109375, "learning_rate": 1.3579217819090801e-05, "loss": 0.1393, "step": 37245 }, { "epoch": 27.286446886446885, "grad_norm": 1.2890625, "learning_rate": 1.3576935654057344e-05, "loss": 0.126, "step": 37246 }, { "epoch": 27.287179487179486, "grad_norm": 1.1875, "learning_rate": 1.3574653665934413e-05, "loss": 0.1089, "step": 37247 }, { "epoch": 27.287912087912087, "grad_norm": 1.3125, "learning_rate": 1.3572371854732946e-05, "loss": 0.1188, "step": 37248 }, { "epoch": 27.288644688644688, "grad_norm": 0.75, "learning_rate": 1.357009022046386e-05, "loss": 0.1196, "step": 37249 }, { "epoch": 27.28937728937729, "grad_norm": 0.69140625, "learning_rate": 1.3567808763138092e-05, "loss": 0.106, "step": 37250 }, { "epoch": 27.29010989010989, "grad_norm": 0.76953125, "learning_rate": 1.3565527482766574e-05, "loss": 0.1003, "step": 37251 }, { "epoch": 27.29084249084249, "grad_norm": 0.96484375, "learning_rate": 1.3563246379360258e-05, "loss": 0.1264, "step": 37252 }, { "epoch": 27.29157509157509, "grad_norm": 1.1875, "learning_rate": 1.3560965452930039e-05, "loss": 0.104, "step": 37253 }, { "epoch": 27.29230769230769, "grad_norm": 1.3046875, "learning_rate": 1.3558684703486867e-05, "loss": 0.1185, "step": 37254 }, { "epoch": 27.293040293040292, "grad_norm": 0.9140625, "learning_rate": 1.3556404131041644e-05, "loss": 0.1301, "step": 37255 }, { "epoch": 27.293772893772893, "grad_norm": 0.90234375, "learning_rate": 1.3554123735605345e-05, "loss": 0.1228, "step": 37256 }, { "epoch": 27.294505494505493, "grad_norm": 1.0625, "learning_rate": 1.3551843517188848e-05, "loss": 0.1464, "step": 37257 }, { "epoch": 27.295238095238094, "grad_norm": 0.9296875, "learning_rate": 1.3549563475803101e-05, "loss": 0.13, "step": 37258 }, { "epoch": 27.295970695970695, "grad_norm": 1.125, "learning_rate": 1.3547283611459041e-05, "loss": 0.1096, "step": 37259 }, { "epoch": 27.296703296703296, "grad_norm": 1.140625, "learning_rate": 1.3545003924167557e-05, "loss": 0.1112, "step": 37260 }, { "epoch": 27.297435897435896, "grad_norm": 1.1640625, "learning_rate": 1.3542724413939598e-05, "loss": 0.1073, "step": 37261 }, { "epoch": 27.298168498168497, "grad_norm": 0.86328125, "learning_rate": 1.354044508078607e-05, "loss": 0.1372, "step": 37262 }, { "epoch": 27.298901098901098, "grad_norm": 0.84375, "learning_rate": 1.3538165924717921e-05, "loss": 0.1095, "step": 37263 }, { "epoch": 27.2996336996337, "grad_norm": 0.76953125, "learning_rate": 1.3535886945746044e-05, "loss": 0.0906, "step": 37264 }, { "epoch": 27.3003663003663, "grad_norm": 1.015625, "learning_rate": 1.3533608143881357e-05, "loss": 0.1316, "step": 37265 }, { "epoch": 27.3010989010989, "grad_norm": 1.0703125, "learning_rate": 1.3531329519134797e-05, "loss": 0.1172, "step": 37266 }, { "epoch": 27.3018315018315, "grad_norm": 0.7421875, "learning_rate": 1.3529051071517285e-05, "loss": 0.1026, "step": 37267 }, { "epoch": 27.3025641025641, "grad_norm": 1.234375, "learning_rate": 1.3526772801039711e-05, "loss": 0.1033, "step": 37268 }, { "epoch": 27.303296703296702, "grad_norm": 1.1875, "learning_rate": 1.352449470771302e-05, "loss": 0.1358, "step": 37269 }, { "epoch": 27.304029304029303, "grad_norm": 0.7265625, "learning_rate": 1.3522216791548101e-05, "loss": 0.1045, "step": 37270 }, { "epoch": 27.304761904761904, "grad_norm": 1.0078125, "learning_rate": 1.3519939052555878e-05, "loss": 0.1438, "step": 37271 }, { "epoch": 27.305494505494504, "grad_norm": 1.90625, "learning_rate": 1.3517661490747269e-05, "loss": 0.108, "step": 37272 }, { "epoch": 27.306227106227105, "grad_norm": 1.265625, "learning_rate": 1.3515384106133183e-05, "loss": 0.145, "step": 37273 }, { "epoch": 27.306959706959706, "grad_norm": 0.99609375, "learning_rate": 1.351310689872455e-05, "loss": 0.1155, "step": 37274 }, { "epoch": 27.307692307692307, "grad_norm": 1.390625, "learning_rate": 1.3510829868532247e-05, "loss": 0.1296, "step": 37275 }, { "epoch": 27.308424908424907, "grad_norm": 0.73828125, "learning_rate": 1.3508553015567196e-05, "loss": 0.1029, "step": 37276 }, { "epoch": 27.309157509157508, "grad_norm": 1.1953125, "learning_rate": 1.3506276339840316e-05, "loss": 0.1224, "step": 37277 }, { "epoch": 27.30989010989011, "grad_norm": 1.1171875, "learning_rate": 1.3503999841362525e-05, "loss": 0.1238, "step": 37278 }, { "epoch": 27.31062271062271, "grad_norm": 0.890625, "learning_rate": 1.3501723520144702e-05, "loss": 0.13, "step": 37279 }, { "epoch": 27.31135531135531, "grad_norm": 0.828125, "learning_rate": 1.3499447376197776e-05, "loss": 0.1208, "step": 37280 }, { "epoch": 27.31208791208791, "grad_norm": 0.953125, "learning_rate": 1.3497171409532621e-05, "loss": 0.1082, "step": 37281 }, { "epoch": 27.31282051282051, "grad_norm": 0.65234375, "learning_rate": 1.3494895620160195e-05, "loss": 0.1114, "step": 37282 }, { "epoch": 27.313553113553112, "grad_norm": 1.34375, "learning_rate": 1.3492620008091353e-05, "loss": 0.1337, "step": 37283 }, { "epoch": 27.314285714285713, "grad_norm": 0.76953125, "learning_rate": 1.3490344573337021e-05, "loss": 0.1253, "step": 37284 }, { "epoch": 27.315018315018314, "grad_norm": 1.203125, "learning_rate": 1.3488069315908108e-05, "loss": 0.1132, "step": 37285 }, { "epoch": 27.315750915750915, "grad_norm": 1.1171875, "learning_rate": 1.348579423581548e-05, "loss": 0.1296, "step": 37286 }, { "epoch": 27.316483516483515, "grad_norm": 1.046875, "learning_rate": 1.3483519333070093e-05, "loss": 0.1122, "step": 37287 }, { "epoch": 27.317216117216116, "grad_norm": 1.2265625, "learning_rate": 1.34812446076828e-05, "loss": 0.1248, "step": 37288 }, { "epoch": 27.317948717948717, "grad_norm": 0.84375, "learning_rate": 1.347897005966453e-05, "loss": 0.1378, "step": 37289 }, { "epoch": 27.318681318681318, "grad_norm": 1.03125, "learning_rate": 1.3476695689026156e-05, "loss": 0.1273, "step": 37290 }, { "epoch": 27.31941391941392, "grad_norm": 0.76953125, "learning_rate": 1.3474421495778577e-05, "loss": 0.1078, "step": 37291 }, { "epoch": 27.32014652014652, "grad_norm": 0.96484375, "learning_rate": 1.347214747993271e-05, "loss": 0.1063, "step": 37292 }, { "epoch": 27.32087912087912, "grad_norm": 0.69140625, "learning_rate": 1.3469873641499456e-05, "loss": 0.1078, "step": 37293 }, { "epoch": 27.32161172161172, "grad_norm": 1.0703125, "learning_rate": 1.346759998048967e-05, "loss": 0.1205, "step": 37294 }, { "epoch": 27.32234432234432, "grad_norm": 0.83984375, "learning_rate": 1.3465326496914275e-05, "loss": 0.1074, "step": 37295 }, { "epoch": 27.323076923076922, "grad_norm": 1.0390625, "learning_rate": 1.3463053190784173e-05, "loss": 0.1161, "step": 37296 }, { "epoch": 27.323809523809523, "grad_norm": 0.92578125, "learning_rate": 1.3460780062110226e-05, "loss": 0.1459, "step": 37297 }, { "epoch": 27.324542124542123, "grad_norm": 0.99609375, "learning_rate": 1.3458507110903341e-05, "loss": 0.1299, "step": 37298 }, { "epoch": 27.325274725274724, "grad_norm": 0.65234375, "learning_rate": 1.3456234337174409e-05, "loss": 0.1187, "step": 37299 }, { "epoch": 27.326007326007325, "grad_norm": 1.25, "learning_rate": 1.3453961740934332e-05, "loss": 0.1002, "step": 37300 }, { "epoch": 27.326739926739926, "grad_norm": 0.99609375, "learning_rate": 1.345168932219397e-05, "loss": 0.1064, "step": 37301 }, { "epoch": 27.327472527472526, "grad_norm": 1.3203125, "learning_rate": 1.3449417080964227e-05, "loss": 0.1333, "step": 37302 }, { "epoch": 27.328205128205127, "grad_norm": 1.21875, "learning_rate": 1.344714501725599e-05, "loss": 0.1047, "step": 37303 }, { "epoch": 27.328937728937728, "grad_norm": 0.96484375, "learning_rate": 1.3444873131080163e-05, "loss": 0.0974, "step": 37304 }, { "epoch": 27.32967032967033, "grad_norm": 1.3671875, "learning_rate": 1.3442601422447598e-05, "loss": 0.1022, "step": 37305 }, { "epoch": 27.33040293040293, "grad_norm": 1.296875, "learning_rate": 1.3440329891369205e-05, "loss": 0.1581, "step": 37306 }, { "epoch": 27.33113553113553, "grad_norm": 0.8203125, "learning_rate": 1.343805853785583e-05, "loss": 0.1336, "step": 37307 }, { "epoch": 27.33186813186813, "grad_norm": 0.86328125, "learning_rate": 1.3435787361918417e-05, "loss": 0.1103, "step": 37308 }, { "epoch": 27.33260073260073, "grad_norm": 1.015625, "learning_rate": 1.3433516363567791e-05, "loss": 0.1219, "step": 37309 }, { "epoch": 27.333333333333332, "grad_norm": 0.859375, "learning_rate": 1.3431245542814863e-05, "loss": 0.1292, "step": 37310 }, { "epoch": 27.334065934065933, "grad_norm": 0.953125, "learning_rate": 1.3428974899670519e-05, "loss": 0.107, "step": 37311 }, { "epoch": 27.334798534798534, "grad_norm": 1.140625, "learning_rate": 1.3426704434145601e-05, "loss": 0.1368, "step": 37312 }, { "epoch": 27.335531135531134, "grad_norm": 1.609375, "learning_rate": 1.3424434146251032e-05, "loss": 0.1166, "step": 37313 }, { "epoch": 27.336263736263735, "grad_norm": 0.87109375, "learning_rate": 1.3422164035997663e-05, "loss": 0.0955, "step": 37314 }, { "epoch": 27.336996336996336, "grad_norm": 1.4140625, "learning_rate": 1.3419894103396389e-05, "loss": 0.1026, "step": 37315 }, { "epoch": 27.337728937728937, "grad_norm": 0.98046875, "learning_rate": 1.3417624348458063e-05, "loss": 0.1203, "step": 37316 }, { "epoch": 27.338461538461537, "grad_norm": 0.8125, "learning_rate": 1.3415354771193572e-05, "loss": 0.115, "step": 37317 }, { "epoch": 27.339194139194138, "grad_norm": 1.296875, "learning_rate": 1.341308537161379e-05, "loss": 0.1199, "step": 37318 }, { "epoch": 27.33992673992674, "grad_norm": 0.90625, "learning_rate": 1.3410816149729602e-05, "loss": 0.1116, "step": 37319 }, { "epoch": 27.34065934065934, "grad_norm": 0.9765625, "learning_rate": 1.3408547105551856e-05, "loss": 0.1345, "step": 37320 }, { "epoch": 27.34139194139194, "grad_norm": 0.86328125, "learning_rate": 1.3406278239091442e-05, "loss": 0.1103, "step": 37321 }, { "epoch": 27.34212454212454, "grad_norm": 1.0390625, "learning_rate": 1.3404009550359234e-05, "loss": 0.1157, "step": 37322 }, { "epoch": 27.34285714285714, "grad_norm": 0.73046875, "learning_rate": 1.3401741039366086e-05, "loss": 0.1137, "step": 37323 }, { "epoch": 27.343589743589742, "grad_norm": 0.94140625, "learning_rate": 1.3399472706122868e-05, "loss": 0.1422, "step": 37324 }, { "epoch": 27.344322344322343, "grad_norm": 1.2265625, "learning_rate": 1.3397204550640463e-05, "loss": 0.1142, "step": 37325 }, { "epoch": 27.345054945054944, "grad_norm": 1.375, "learning_rate": 1.3394936572929736e-05, "loss": 0.108, "step": 37326 }, { "epoch": 27.345787545787545, "grad_norm": 1.078125, "learning_rate": 1.3392668773001543e-05, "loss": 0.1097, "step": 37327 }, { "epoch": 27.346520146520145, "grad_norm": 0.9375, "learning_rate": 1.3390401150866755e-05, "loss": 0.1052, "step": 37328 }, { "epoch": 27.347252747252746, "grad_norm": 1.2890625, "learning_rate": 1.3388133706536233e-05, "loss": 0.1125, "step": 37329 }, { "epoch": 27.347985347985347, "grad_norm": 0.7265625, "learning_rate": 1.3385866440020862e-05, "loss": 0.1127, "step": 37330 }, { "epoch": 27.348717948717947, "grad_norm": 0.87109375, "learning_rate": 1.3383599351331474e-05, "loss": 0.1138, "step": 37331 }, { "epoch": 27.349450549450548, "grad_norm": 0.9296875, "learning_rate": 1.3381332440478963e-05, "loss": 0.1151, "step": 37332 }, { "epoch": 27.35018315018315, "grad_norm": 0.9921875, "learning_rate": 1.337906570747414e-05, "loss": 0.0986, "step": 37333 }, { "epoch": 27.35091575091575, "grad_norm": 1.6328125, "learning_rate": 1.3376799152327931e-05, "loss": 0.1369, "step": 37334 }, { "epoch": 27.35164835164835, "grad_norm": 1.09375, "learning_rate": 1.3374532775051152e-05, "loss": 0.1158, "step": 37335 }, { "epoch": 27.35238095238095, "grad_norm": 1.3984375, "learning_rate": 1.3372266575654668e-05, "loss": 0.1348, "step": 37336 }, { "epoch": 27.353113553113552, "grad_norm": 0.76171875, "learning_rate": 1.3370000554149362e-05, "loss": 0.1194, "step": 37337 }, { "epoch": 27.353846153846153, "grad_norm": 0.9296875, "learning_rate": 1.336773471054604e-05, "loss": 0.1166, "step": 37338 }, { "epoch": 27.354578754578753, "grad_norm": 1.2109375, "learning_rate": 1.3365469044855625e-05, "loss": 0.1053, "step": 37339 }, { "epoch": 27.355311355311354, "grad_norm": 1.296875, "learning_rate": 1.3363203557088915e-05, "loss": 0.1155, "step": 37340 }, { "epoch": 27.356043956043955, "grad_norm": 1.3203125, "learning_rate": 1.336093824725681e-05, "loss": 0.1704, "step": 37341 }, { "epoch": 27.356776556776556, "grad_norm": 1.0703125, "learning_rate": 1.3358673115370127e-05, "loss": 0.1216, "step": 37342 }, { "epoch": 27.357509157509156, "grad_norm": 1.0625, "learning_rate": 1.335640816143973e-05, "loss": 0.1551, "step": 37343 }, { "epoch": 27.358241758241757, "grad_norm": 1.25, "learning_rate": 1.3354143385476479e-05, "loss": 0.1363, "step": 37344 }, { "epoch": 27.358974358974358, "grad_norm": 1.28125, "learning_rate": 1.3351878787491232e-05, "loss": 0.0999, "step": 37345 }, { "epoch": 27.35970695970696, "grad_norm": 1.0, "learning_rate": 1.3349614367494814e-05, "loss": 0.1124, "step": 37346 }, { "epoch": 27.36043956043956, "grad_norm": 1.5, "learning_rate": 1.3347350125498092e-05, "loss": 0.1193, "step": 37347 }, { "epoch": 27.36117216117216, "grad_norm": 1.6953125, "learning_rate": 1.3345086061511924e-05, "loss": 0.1341, "step": 37348 }, { "epoch": 27.36190476190476, "grad_norm": 1.3125, "learning_rate": 1.3342822175547137e-05, "loss": 0.1371, "step": 37349 }, { "epoch": 27.36263736263736, "grad_norm": 0.93359375, "learning_rate": 1.3340558467614583e-05, "loss": 0.114, "step": 37350 }, { "epoch": 27.363369963369962, "grad_norm": 0.91796875, "learning_rate": 1.333829493772511e-05, "loss": 0.117, "step": 37351 }, { "epoch": 27.364102564102563, "grad_norm": 1.125, "learning_rate": 1.333603158588958e-05, "loss": 0.1284, "step": 37352 }, { "epoch": 27.364835164835164, "grad_norm": 0.97265625, "learning_rate": 1.333376841211881e-05, "loss": 0.1186, "step": 37353 }, { "epoch": 27.365567765567764, "grad_norm": 1.015625, "learning_rate": 1.3331505416423653e-05, "loss": 0.1192, "step": 37354 }, { "epoch": 27.366300366300365, "grad_norm": 1.3359375, "learning_rate": 1.332924259881496e-05, "loss": 0.1215, "step": 37355 }, { "epoch": 27.367032967032966, "grad_norm": 1.0234375, "learning_rate": 1.3326979959303582e-05, "loss": 0.1319, "step": 37356 }, { "epoch": 27.367765567765566, "grad_norm": 0.9453125, "learning_rate": 1.332471749790033e-05, "loss": 0.0983, "step": 37357 }, { "epoch": 27.368498168498167, "grad_norm": 1.03125, "learning_rate": 1.3322455214616079e-05, "loss": 0.1061, "step": 37358 }, { "epoch": 27.369230769230768, "grad_norm": 1.3828125, "learning_rate": 1.3320193109461618e-05, "loss": 0.1463, "step": 37359 }, { "epoch": 27.36996336996337, "grad_norm": 1.53125, "learning_rate": 1.331793118244784e-05, "loss": 0.1281, "step": 37360 }, { "epoch": 27.37069597069597, "grad_norm": 0.796875, "learning_rate": 1.3315669433585551e-05, "loss": 0.0984, "step": 37361 }, { "epoch": 27.37142857142857, "grad_norm": 0.87109375, "learning_rate": 1.3313407862885599e-05, "loss": 0.1306, "step": 37362 }, { "epoch": 27.37216117216117, "grad_norm": 1.234375, "learning_rate": 1.3311146470358828e-05, "loss": 0.1328, "step": 37363 }, { "epoch": 27.37289377289377, "grad_norm": 1.2890625, "learning_rate": 1.3308885256016035e-05, "loss": 0.1324, "step": 37364 }, { "epoch": 27.373626373626372, "grad_norm": 0.7421875, "learning_rate": 1.3306624219868108e-05, "loss": 0.1013, "step": 37365 }, { "epoch": 27.374358974358973, "grad_norm": 1.4609375, "learning_rate": 1.3304363361925845e-05, "loss": 0.1109, "step": 37366 }, { "epoch": 27.375091575091574, "grad_norm": 0.8671875, "learning_rate": 1.3302102682200098e-05, "loss": 0.1193, "step": 37367 }, { "epoch": 27.375824175824174, "grad_norm": 1.1015625, "learning_rate": 1.3299842180701675e-05, "loss": 0.0937, "step": 37368 }, { "epoch": 27.376556776556775, "grad_norm": 0.98046875, "learning_rate": 1.3297581857441418e-05, "loss": 0.127, "step": 37369 }, { "epoch": 27.377289377289376, "grad_norm": 1.984375, "learning_rate": 1.3295321712430161e-05, "loss": 0.1423, "step": 37370 }, { "epoch": 27.378021978021977, "grad_norm": 1.046875, "learning_rate": 1.3293061745678748e-05, "loss": 0.1342, "step": 37371 }, { "epoch": 27.378754578754577, "grad_norm": 1.265625, "learning_rate": 1.3290801957197977e-05, "loss": 0.1348, "step": 37372 }, { "epoch": 27.379487179487178, "grad_norm": 0.68359375, "learning_rate": 1.3288542346998681e-05, "loss": 0.1208, "step": 37373 }, { "epoch": 27.38021978021978, "grad_norm": 1.09375, "learning_rate": 1.3286282915091717e-05, "loss": 0.1513, "step": 37374 }, { "epoch": 27.38095238095238, "grad_norm": 0.82421875, "learning_rate": 1.3284023661487868e-05, "loss": 0.1459, "step": 37375 }, { "epoch": 27.38168498168498, "grad_norm": 0.96875, "learning_rate": 1.3281764586197981e-05, "loss": 0.1065, "step": 37376 }, { "epoch": 27.38241758241758, "grad_norm": 1.6171875, "learning_rate": 1.3279505689232877e-05, "loss": 0.1532, "step": 37377 }, { "epoch": 27.383150183150182, "grad_norm": 1.5703125, "learning_rate": 1.3277246970603394e-05, "loss": 0.1256, "step": 37378 }, { "epoch": 27.383882783882783, "grad_norm": 0.703125, "learning_rate": 1.3274988430320326e-05, "loss": 0.1312, "step": 37379 }, { "epoch": 27.384615384615383, "grad_norm": 1.0390625, "learning_rate": 1.3272730068394512e-05, "loss": 0.1227, "step": 37380 }, { "epoch": 27.385347985347984, "grad_norm": 1.1328125, "learning_rate": 1.3270471884836768e-05, "loss": 0.1257, "step": 37381 }, { "epoch": 27.386080586080585, "grad_norm": 1.078125, "learning_rate": 1.3268213879657926e-05, "loss": 0.1193, "step": 37382 }, { "epoch": 27.386813186813185, "grad_norm": 0.73828125, "learning_rate": 1.326595605286878e-05, "loss": 0.0907, "step": 37383 }, { "epoch": 27.387545787545786, "grad_norm": 1.1015625, "learning_rate": 1.3263698404480177e-05, "loss": 0.1119, "step": 37384 }, { "epoch": 27.388278388278387, "grad_norm": 1.078125, "learning_rate": 1.3261440934502891e-05, "loss": 0.1234, "step": 37385 }, { "epoch": 27.389010989010988, "grad_norm": 1.03125, "learning_rate": 1.3259183642947796e-05, "loss": 0.1305, "step": 37386 }, { "epoch": 27.38974358974359, "grad_norm": 1.1953125, "learning_rate": 1.3256926529825656e-05, "loss": 0.1562, "step": 37387 }, { "epoch": 27.39047619047619, "grad_norm": 0.6640625, "learning_rate": 1.3254669595147318e-05, "loss": 0.1081, "step": 37388 }, { "epoch": 27.39120879120879, "grad_norm": 1.3671875, "learning_rate": 1.3252412838923593e-05, "loss": 0.1298, "step": 37389 }, { "epoch": 27.39194139194139, "grad_norm": 0.6484375, "learning_rate": 1.3250156261165256e-05, "loss": 0.1144, "step": 37390 }, { "epoch": 27.39267399267399, "grad_norm": 0.78515625, "learning_rate": 1.3247899861883178e-05, "loss": 0.1076, "step": 37391 }, { "epoch": 27.393406593406592, "grad_norm": 1.21875, "learning_rate": 1.3245643641088126e-05, "loss": 0.1574, "step": 37392 }, { "epoch": 27.394139194139193, "grad_norm": 0.80078125, "learning_rate": 1.3243387598790938e-05, "loss": 0.105, "step": 37393 }, { "epoch": 27.394871794871793, "grad_norm": 0.86328125, "learning_rate": 1.3241131735002404e-05, "loss": 0.1238, "step": 37394 }, { "epoch": 27.395604395604394, "grad_norm": 1.09375, "learning_rate": 1.3238876049733333e-05, "loss": 0.1227, "step": 37395 }, { "epoch": 27.396336996336995, "grad_norm": 1.078125, "learning_rate": 1.3236620542994541e-05, "loss": 0.1254, "step": 37396 }, { "epoch": 27.397069597069596, "grad_norm": 1.1015625, "learning_rate": 1.3234365214796843e-05, "loss": 0.0978, "step": 37397 }, { "epoch": 27.397802197802196, "grad_norm": 0.83203125, "learning_rate": 1.3232110065151027e-05, "loss": 0.1032, "step": 37398 }, { "epoch": 27.398534798534797, "grad_norm": 1.109375, "learning_rate": 1.3229855094067903e-05, "loss": 0.1137, "step": 37399 }, { "epoch": 27.399267399267398, "grad_norm": 0.67578125, "learning_rate": 1.322760030155829e-05, "loss": 0.1296, "step": 37400 }, { "epoch": 27.4, "grad_norm": 1.0234375, "learning_rate": 1.3225345687632968e-05, "loss": 0.113, "step": 37401 }, { "epoch": 27.4007326007326, "grad_norm": 1.015625, "learning_rate": 1.3223091252302749e-05, "loss": 0.1243, "step": 37402 }, { "epoch": 27.4014652014652, "grad_norm": 1.0234375, "learning_rate": 1.3220836995578439e-05, "loss": 0.1259, "step": 37403 }, { "epoch": 27.4021978021978, "grad_norm": 0.94921875, "learning_rate": 1.3218582917470848e-05, "loss": 0.1048, "step": 37404 }, { "epoch": 27.4029304029304, "grad_norm": 0.62109375, "learning_rate": 1.3216329017990753e-05, "loss": 0.0944, "step": 37405 }, { "epoch": 27.403663003663002, "grad_norm": 0.7421875, "learning_rate": 1.321407529714897e-05, "loss": 0.117, "step": 37406 }, { "epoch": 27.404395604395603, "grad_norm": 0.98828125, "learning_rate": 1.3211821754956285e-05, "loss": 0.0841, "step": 37407 }, { "epoch": 27.405128205128204, "grad_norm": 1.2265625, "learning_rate": 1.320956839142352e-05, "loss": 0.1307, "step": 37408 }, { "epoch": 27.405860805860804, "grad_norm": 0.9375, "learning_rate": 1.3207315206561443e-05, "loss": 0.1123, "step": 37409 }, { "epoch": 27.406593406593405, "grad_norm": 1.1171875, "learning_rate": 1.320506220038086e-05, "loss": 0.1013, "step": 37410 }, { "epoch": 27.407326007326006, "grad_norm": 0.91015625, "learning_rate": 1.3202809372892567e-05, "loss": 0.0882, "step": 37411 }, { "epoch": 27.408058608058607, "grad_norm": 1.0, "learning_rate": 1.3200556724107369e-05, "loss": 0.1219, "step": 37412 }, { "epoch": 27.408791208791207, "grad_norm": 0.67578125, "learning_rate": 1.3198304254036034e-05, "loss": 0.1052, "step": 37413 }, { "epoch": 27.409523809523808, "grad_norm": 0.77734375, "learning_rate": 1.319605196268937e-05, "loss": 0.124, "step": 37414 }, { "epoch": 27.41025641025641, "grad_norm": 0.9453125, "learning_rate": 1.3193799850078184e-05, "loss": 0.118, "step": 37415 }, { "epoch": 27.41098901098901, "grad_norm": 0.73046875, "learning_rate": 1.3191547916213217e-05, "loss": 0.1304, "step": 37416 }, { "epoch": 27.41172161172161, "grad_norm": 0.98046875, "learning_rate": 1.3189296161105324e-05, "loss": 0.1314, "step": 37417 }, { "epoch": 27.41245421245421, "grad_norm": 0.78515625, "learning_rate": 1.3187044584765246e-05, "loss": 0.1079, "step": 37418 }, { "epoch": 27.41318681318681, "grad_norm": 1.0859375, "learning_rate": 1.31847931872038e-05, "loss": 0.1, "step": 37419 }, { "epoch": 27.413919413919412, "grad_norm": 0.89453125, "learning_rate": 1.3182541968431749e-05, "loss": 0.1082, "step": 37420 }, { "epoch": 27.414652014652013, "grad_norm": 1.3203125, "learning_rate": 1.3180290928459896e-05, "loss": 0.1184, "step": 37421 }, { "epoch": 27.415384615384614, "grad_norm": 0.90234375, "learning_rate": 1.3178040067299e-05, "loss": 0.1203, "step": 37422 }, { "epoch": 27.416117216117215, "grad_norm": 0.92578125, "learning_rate": 1.3175789384959891e-05, "loss": 0.129, "step": 37423 }, { "epoch": 27.416849816849815, "grad_norm": 1.3671875, "learning_rate": 1.3173538881453314e-05, "loss": 0.1552, "step": 37424 }, { "epoch": 27.417582417582416, "grad_norm": 1.15625, "learning_rate": 1.3171288556790068e-05, "loss": 0.1529, "step": 37425 }, { "epoch": 27.418315018315017, "grad_norm": 1.03125, "learning_rate": 1.3169038410980941e-05, "loss": 0.1126, "step": 37426 }, { "epoch": 27.419047619047618, "grad_norm": 1.375, "learning_rate": 1.3166788444036681e-05, "loss": 0.1306, "step": 37427 }, { "epoch": 27.41978021978022, "grad_norm": 1.1328125, "learning_rate": 1.3164538655968123e-05, "loss": 0.1544, "step": 37428 }, { "epoch": 27.42051282051282, "grad_norm": 1.140625, "learning_rate": 1.3162289046786003e-05, "loss": 0.1209, "step": 37429 }, { "epoch": 27.42124542124542, "grad_norm": 0.80078125, "learning_rate": 1.3160039616501127e-05, "loss": 0.1158, "step": 37430 }, { "epoch": 27.42197802197802, "grad_norm": 0.98828125, "learning_rate": 1.315779036512424e-05, "loss": 0.0969, "step": 37431 }, { "epoch": 27.42271062271062, "grad_norm": 1.1015625, "learning_rate": 1.3155541292666143e-05, "loss": 0.118, "step": 37432 }, { "epoch": 27.423443223443222, "grad_norm": 1.1328125, "learning_rate": 1.3153292399137605e-05, "loss": 0.1267, "step": 37433 }, { "epoch": 27.424175824175823, "grad_norm": 1.1171875, "learning_rate": 1.3151043684549418e-05, "loss": 0.1325, "step": 37434 }, { "epoch": 27.424908424908423, "grad_norm": 1.2890625, "learning_rate": 1.3148795148912327e-05, "loss": 0.1154, "step": 37435 }, { "epoch": 27.425641025641024, "grad_norm": 1.2421875, "learning_rate": 1.3146546792237125e-05, "loss": 0.1311, "step": 37436 }, { "epoch": 27.426373626373625, "grad_norm": 1.6015625, "learning_rate": 1.3144298614534572e-05, "loss": 0.1366, "step": 37437 }, { "epoch": 27.427106227106226, "grad_norm": 1.390625, "learning_rate": 1.3142050615815466e-05, "loss": 0.15, "step": 37438 }, { "epoch": 27.427838827838826, "grad_norm": 1.2421875, "learning_rate": 1.3139802796090544e-05, "loss": 0.1472, "step": 37439 }, { "epoch": 27.428571428571427, "grad_norm": 0.6640625, "learning_rate": 1.313755515537059e-05, "loss": 0.1188, "step": 37440 }, { "epoch": 27.429304029304028, "grad_norm": 0.71875, "learning_rate": 1.313530769366639e-05, "loss": 0.0993, "step": 37441 }, { "epoch": 27.43003663003663, "grad_norm": 1.3359375, "learning_rate": 1.3133060410988669e-05, "loss": 0.12, "step": 37442 }, { "epoch": 27.43076923076923, "grad_norm": 1.6328125, "learning_rate": 1.3130813307348253e-05, "loss": 0.1183, "step": 37443 }, { "epoch": 27.43150183150183, "grad_norm": 0.97265625, "learning_rate": 1.3128566382755858e-05, "loss": 0.1231, "step": 37444 }, { "epoch": 27.43223443223443, "grad_norm": 1.15625, "learning_rate": 1.3126319637222287e-05, "loss": 0.1379, "step": 37445 }, { "epoch": 27.43296703296703, "grad_norm": 0.84375, "learning_rate": 1.3124073070758273e-05, "loss": 0.1135, "step": 37446 }, { "epoch": 27.433699633699632, "grad_norm": 0.76953125, "learning_rate": 1.3121826683374611e-05, "loss": 0.1431, "step": 37447 }, { "epoch": 27.434432234432233, "grad_norm": 0.8515625, "learning_rate": 1.3119580475082014e-05, "loss": 0.1279, "step": 37448 }, { "epoch": 27.435164835164834, "grad_norm": 1.109375, "learning_rate": 1.3117334445891313e-05, "loss": 0.1328, "step": 37449 }, { "epoch": 27.435897435897434, "grad_norm": 1.0, "learning_rate": 1.3115088595813219e-05, "loss": 0.1468, "step": 37450 }, { "epoch": 27.436630036630035, "grad_norm": 1.25, "learning_rate": 1.3112842924858501e-05, "loss": 0.1125, "step": 37451 }, { "epoch": 27.437362637362636, "grad_norm": 0.78125, "learning_rate": 1.311059743303795e-05, "loss": 0.1067, "step": 37452 }, { "epoch": 27.438095238095237, "grad_norm": 1.4140625, "learning_rate": 1.3108352120362266e-05, "loss": 0.1176, "step": 37453 }, { "epoch": 27.438827838827837, "grad_norm": 1.2109375, "learning_rate": 1.3106106986842275e-05, "loss": 0.1303, "step": 37454 }, { "epoch": 27.439560439560438, "grad_norm": 1.234375, "learning_rate": 1.3103862032488682e-05, "loss": 0.1276, "step": 37455 }, { "epoch": 27.44029304029304, "grad_norm": 0.68359375, "learning_rate": 1.3101617257312282e-05, "loss": 0.1357, "step": 37456 }, { "epoch": 27.44102564102564, "grad_norm": 1.25, "learning_rate": 1.309937266132379e-05, "loss": 0.0992, "step": 37457 }, { "epoch": 27.44175824175824, "grad_norm": 0.99609375, "learning_rate": 1.3097128244533986e-05, "loss": 0.1101, "step": 37458 }, { "epoch": 27.44249084249084, "grad_norm": 0.80859375, "learning_rate": 1.3094884006953617e-05, "loss": 0.109, "step": 37459 }, { "epoch": 27.44322344322344, "grad_norm": 1.0859375, "learning_rate": 1.3092639948593457e-05, "loss": 0.111, "step": 37460 }, { "epoch": 27.443956043956042, "grad_norm": 1.1015625, "learning_rate": 1.3090396069464223e-05, "loss": 0.1226, "step": 37461 }, { "epoch": 27.444688644688643, "grad_norm": 1.28125, "learning_rate": 1.3088152369576675e-05, "loss": 0.1361, "step": 37462 }, { "epoch": 27.445421245421244, "grad_norm": 1.3125, "learning_rate": 1.308590884894158e-05, "loss": 0.1002, "step": 37463 }, { "epoch": 27.446153846153845, "grad_norm": 0.9765625, "learning_rate": 1.3083665507569685e-05, "loss": 0.1598, "step": 37464 }, { "epoch": 27.446886446886445, "grad_norm": 0.71875, "learning_rate": 1.3081422345471722e-05, "loss": 0.1258, "step": 37465 }, { "epoch": 27.447619047619046, "grad_norm": 1.203125, "learning_rate": 1.3079179362658445e-05, "loss": 0.1138, "step": 37466 }, { "epoch": 27.448351648351647, "grad_norm": 1.3984375, "learning_rate": 1.3076936559140623e-05, "loss": 0.1525, "step": 37467 }, { "epoch": 27.449084249084247, "grad_norm": 0.81640625, "learning_rate": 1.307469393492895e-05, "loss": 0.1052, "step": 37468 }, { "epoch": 27.449816849816848, "grad_norm": 1.21875, "learning_rate": 1.3072451490034232e-05, "loss": 0.1014, "step": 37469 }, { "epoch": 27.45054945054945, "grad_norm": 1.0390625, "learning_rate": 1.3070209224467175e-05, "loss": 0.1314, "step": 37470 }, { "epoch": 27.45128205128205, "grad_norm": 0.9375, "learning_rate": 1.3067967138238546e-05, "loss": 0.158, "step": 37471 }, { "epoch": 27.45201465201465, "grad_norm": 0.890625, "learning_rate": 1.3065725231359061e-05, "loss": 0.155, "step": 37472 }, { "epoch": 27.45274725274725, "grad_norm": 1.46875, "learning_rate": 1.3063483503839488e-05, "loss": 0.1123, "step": 37473 }, { "epoch": 27.453479853479852, "grad_norm": 1.328125, "learning_rate": 1.3061241955690525e-05, "loss": 0.1204, "step": 37474 }, { "epoch": 27.454212454212453, "grad_norm": 1.0859375, "learning_rate": 1.3059000586922976e-05, "loss": 0.1163, "step": 37475 }, { "epoch": 27.454945054945053, "grad_norm": 1.015625, "learning_rate": 1.305675939754753e-05, "loss": 0.1391, "step": 37476 }, { "epoch": 27.455677655677654, "grad_norm": 1.3828125, "learning_rate": 1.3054518387574937e-05, "loss": 0.0993, "step": 37477 }, { "epoch": 27.456410256410255, "grad_norm": 0.859375, "learning_rate": 1.305227755701596e-05, "loss": 0.093, "step": 37478 }, { "epoch": 27.457142857142856, "grad_norm": 1.4609375, "learning_rate": 1.3050036905881285e-05, "loss": 0.1268, "step": 37479 }, { "epoch": 27.457875457875456, "grad_norm": 1.2578125, "learning_rate": 1.3047796434181702e-05, "loss": 0.1043, "step": 37480 }, { "epoch": 27.458608058608057, "grad_norm": 1.078125, "learning_rate": 1.3045556141927906e-05, "loss": 0.1379, "step": 37481 }, { "epoch": 27.459340659340658, "grad_norm": 1.1328125, "learning_rate": 1.3043316029130664e-05, "loss": 0.1348, "step": 37482 }, { "epoch": 27.46007326007326, "grad_norm": 1.1953125, "learning_rate": 1.3041076095800675e-05, "loss": 0.1574, "step": 37483 }, { "epoch": 27.46080586080586, "grad_norm": 0.64453125, "learning_rate": 1.3038836341948688e-05, "loss": 0.1124, "step": 37484 }, { "epoch": 27.46153846153846, "grad_norm": 1.109375, "learning_rate": 1.3036596767585435e-05, "loss": 0.1322, "step": 37485 }, { "epoch": 27.46227106227106, "grad_norm": 1.1484375, "learning_rate": 1.3034357372721659e-05, "loss": 0.1439, "step": 37486 }, { "epoch": 27.46300366300366, "grad_norm": 0.9296875, "learning_rate": 1.3032118157368059e-05, "loss": 0.1049, "step": 37487 }, { "epoch": 27.463736263736262, "grad_norm": 1.03125, "learning_rate": 1.3029879121535382e-05, "loss": 0.0923, "step": 37488 }, { "epoch": 27.464468864468863, "grad_norm": 1.046875, "learning_rate": 1.3027640265234356e-05, "loss": 0.089, "step": 37489 }, { "epoch": 27.465201465201464, "grad_norm": 1.2578125, "learning_rate": 1.3025401588475722e-05, "loss": 0.1228, "step": 37490 }, { "epoch": 27.465934065934064, "grad_norm": 0.953125, "learning_rate": 1.3023163091270175e-05, "loss": 0.1401, "step": 37491 }, { "epoch": 27.466666666666665, "grad_norm": 1.453125, "learning_rate": 1.3020924773628457e-05, "loss": 0.158, "step": 37492 }, { "epoch": 27.467399267399266, "grad_norm": 0.875, "learning_rate": 1.3018686635561305e-05, "loss": 0.11, "step": 37493 }, { "epoch": 27.468131868131866, "grad_norm": 1.3671875, "learning_rate": 1.30164486770794e-05, "loss": 0.1123, "step": 37494 }, { "epoch": 27.468864468864467, "grad_norm": 1.3671875, "learning_rate": 1.3014210898193526e-05, "loss": 0.1626, "step": 37495 }, { "epoch": 27.469597069597068, "grad_norm": 0.84765625, "learning_rate": 1.3011973298914362e-05, "loss": 0.1179, "step": 37496 }, { "epoch": 27.47032967032967, "grad_norm": 0.94140625, "learning_rate": 1.3009735879252647e-05, "loss": 0.1085, "step": 37497 }, { "epoch": 27.47106227106227, "grad_norm": 1.0, "learning_rate": 1.3007498639219085e-05, "loss": 0.0945, "step": 37498 }, { "epoch": 27.47179487179487, "grad_norm": 0.6640625, "learning_rate": 1.3005261578824419e-05, "loss": 0.1279, "step": 37499 }, { "epoch": 27.47252747252747, "grad_norm": 1.296875, "learning_rate": 1.3003024698079325e-05, "loss": 0.1266, "step": 37500 }, { "epoch": 27.47326007326007, "grad_norm": 1.3984375, "learning_rate": 1.300078799699458e-05, "loss": 0.1226, "step": 37501 }, { "epoch": 27.473992673992672, "grad_norm": 1.078125, "learning_rate": 1.2998551475580853e-05, "loss": 0.1042, "step": 37502 }, { "epoch": 27.474725274725273, "grad_norm": 0.96484375, "learning_rate": 1.2996315133848874e-05, "loss": 0.0893, "step": 37503 }, { "epoch": 27.475457875457874, "grad_norm": 0.609375, "learning_rate": 1.2994078971809379e-05, "loss": 0.1278, "step": 37504 }, { "epoch": 27.476190476190474, "grad_norm": 0.96484375, "learning_rate": 1.2991842989473038e-05, "loss": 0.1134, "step": 37505 }, { "epoch": 27.476923076923075, "grad_norm": 1.328125, "learning_rate": 1.2989607186850614e-05, "loss": 0.1225, "step": 37506 }, { "epoch": 27.477655677655676, "grad_norm": 0.7421875, "learning_rate": 1.2987371563952782e-05, "loss": 0.1074, "step": 37507 }, { "epoch": 27.478388278388277, "grad_norm": 1.390625, "learning_rate": 1.2985136120790282e-05, "loss": 0.1259, "step": 37508 }, { "epoch": 27.479120879120877, "grad_norm": 1.890625, "learning_rate": 1.2982900857373792e-05, "loss": 0.1094, "step": 37509 }, { "epoch": 27.479853479853478, "grad_norm": 1.0234375, "learning_rate": 1.2980665773714052e-05, "loss": 0.129, "step": 37510 }, { "epoch": 27.48058608058608, "grad_norm": 0.8359375, "learning_rate": 1.297843086982175e-05, "loss": 0.13, "step": 37511 }, { "epoch": 27.48131868131868, "grad_norm": 1.28125, "learning_rate": 1.2976196145707618e-05, "loss": 0.138, "step": 37512 }, { "epoch": 27.48205128205128, "grad_norm": 0.78515625, "learning_rate": 1.2973961601382337e-05, "loss": 0.1214, "step": 37513 }, { "epoch": 27.48278388278388, "grad_norm": 1.4453125, "learning_rate": 1.2971727236856621e-05, "loss": 0.1263, "step": 37514 }, { "epoch": 27.483516483516482, "grad_norm": 0.734375, "learning_rate": 1.2969493052141182e-05, "loss": 0.1142, "step": 37515 }, { "epoch": 27.484249084249083, "grad_norm": 0.99609375, "learning_rate": 1.2967259047246735e-05, "loss": 0.1117, "step": 37516 }, { "epoch": 27.484981684981683, "grad_norm": 0.7578125, "learning_rate": 1.2965025222183961e-05, "loss": 0.1233, "step": 37517 }, { "epoch": 27.485714285714284, "grad_norm": 0.72265625, "learning_rate": 1.2962791576963566e-05, "loss": 0.1004, "step": 37518 }, { "epoch": 27.486446886446885, "grad_norm": 0.74609375, "learning_rate": 1.296055811159628e-05, "loss": 0.1035, "step": 37519 }, { "epoch": 27.487179487179485, "grad_norm": 0.91796875, "learning_rate": 1.2958324826092754e-05, "loss": 0.117, "step": 37520 }, { "epoch": 27.487912087912086, "grad_norm": 1.3125, "learning_rate": 1.2956091720463743e-05, "loss": 0.1533, "step": 37521 }, { "epoch": 27.488644688644687, "grad_norm": 1.109375, "learning_rate": 1.295385879471991e-05, "loss": 0.1221, "step": 37522 }, { "epoch": 27.489377289377288, "grad_norm": 0.72265625, "learning_rate": 1.2951626048871981e-05, "loss": 0.0899, "step": 37523 }, { "epoch": 27.49010989010989, "grad_norm": 1.0234375, "learning_rate": 1.2949393482930616e-05, "loss": 0.1305, "step": 37524 }, { "epoch": 27.49084249084249, "grad_norm": 1.2578125, "learning_rate": 1.2947161096906541e-05, "loss": 0.1439, "step": 37525 }, { "epoch": 27.49157509157509, "grad_norm": 1.0859375, "learning_rate": 1.294492889081044e-05, "loss": 0.1112, "step": 37526 }, { "epoch": 27.49230769230769, "grad_norm": 0.8984375, "learning_rate": 1.2942696864653035e-05, "loss": 0.1094, "step": 37527 }, { "epoch": 27.493040293040295, "grad_norm": 1.421875, "learning_rate": 1.2940465018444978e-05, "loss": 0.1089, "step": 37528 }, { "epoch": 27.493772893772892, "grad_norm": 0.83984375, "learning_rate": 1.2938233352196988e-05, "loss": 0.1222, "step": 37529 }, { "epoch": 27.494505494505496, "grad_norm": 1.1796875, "learning_rate": 1.2936001865919762e-05, "loss": 0.1158, "step": 37530 }, { "epoch": 27.495238095238093, "grad_norm": 0.9296875, "learning_rate": 1.2933770559623953e-05, "loss": 0.1004, "step": 37531 }, { "epoch": 27.495970695970698, "grad_norm": 1.171875, "learning_rate": 1.2931539433320319e-05, "loss": 0.1078, "step": 37532 }, { "epoch": 27.496703296703295, "grad_norm": 1.40625, "learning_rate": 1.292930848701949e-05, "loss": 0.1534, "step": 37533 }, { "epoch": 27.4974358974359, "grad_norm": 0.5390625, "learning_rate": 1.2927077720732196e-05, "loss": 0.0927, "step": 37534 }, { "epoch": 27.498168498168496, "grad_norm": 0.734375, "learning_rate": 1.292484713446909e-05, "loss": 0.1117, "step": 37535 }, { "epoch": 27.4989010989011, "grad_norm": 0.73828125, "learning_rate": 1.2922616728240875e-05, "loss": 0.117, "step": 37536 }, { "epoch": 27.499633699633698, "grad_norm": 1.140625, "learning_rate": 1.2920386502058243e-05, "loss": 0.1486, "step": 37537 }, { "epoch": 27.500366300366302, "grad_norm": 1.15625, "learning_rate": 1.291815645593189e-05, "loss": 0.0885, "step": 37538 }, { "epoch": 27.5010989010989, "grad_norm": 0.953125, "learning_rate": 1.2915926589872468e-05, "loss": 0.1265, "step": 37539 }, { "epoch": 27.501831501831504, "grad_norm": 1.1328125, "learning_rate": 1.2913696903890685e-05, "loss": 0.131, "step": 37540 }, { "epoch": 27.5025641025641, "grad_norm": 1.0, "learning_rate": 1.2911467397997215e-05, "loss": 0.1352, "step": 37541 }, { "epoch": 27.503296703296705, "grad_norm": 1.203125, "learning_rate": 1.2909238072202742e-05, "loss": 0.1389, "step": 37542 }, { "epoch": 27.504029304029302, "grad_norm": 1.21875, "learning_rate": 1.2907008926517972e-05, "loss": 0.1283, "step": 37543 }, { "epoch": 27.504761904761907, "grad_norm": 1.046875, "learning_rate": 1.2904779960953542e-05, "loss": 0.1273, "step": 37544 }, { "epoch": 27.505494505494504, "grad_norm": 0.7109375, "learning_rate": 1.2902551175520175e-05, "loss": 0.1126, "step": 37545 }, { "epoch": 27.506227106227108, "grad_norm": 1.2109375, "learning_rate": 1.290032257022849e-05, "loss": 0.1336, "step": 37546 }, { "epoch": 27.506959706959705, "grad_norm": 0.76953125, "learning_rate": 1.2898094145089237e-05, "loss": 0.1086, "step": 37547 }, { "epoch": 27.50769230769231, "grad_norm": 0.80859375, "learning_rate": 1.2895865900113043e-05, "loss": 0.1194, "step": 37548 }, { "epoch": 27.508424908424907, "grad_norm": 0.70703125, "learning_rate": 1.2893637835310615e-05, "loss": 0.0963, "step": 37549 }, { "epoch": 27.50915750915751, "grad_norm": 1.5078125, "learning_rate": 1.2891409950692602e-05, "loss": 0.1046, "step": 37550 }, { "epoch": 27.509890109890108, "grad_norm": 1.5859375, "learning_rate": 1.2889182246269682e-05, "loss": 0.113, "step": 37551 }, { "epoch": 27.510622710622712, "grad_norm": 1.109375, "learning_rate": 1.2886954722052541e-05, "loss": 0.1331, "step": 37552 }, { "epoch": 27.51135531135531, "grad_norm": 1.140625, "learning_rate": 1.2884727378051861e-05, "loss": 0.1244, "step": 37553 }, { "epoch": 27.512087912087914, "grad_norm": 0.8046875, "learning_rate": 1.2882500214278288e-05, "loss": 0.128, "step": 37554 }, { "epoch": 27.51282051282051, "grad_norm": 0.79296875, "learning_rate": 1.2880273230742507e-05, "loss": 0.1335, "step": 37555 }, { "epoch": 27.513553113553115, "grad_norm": 1.359375, "learning_rate": 1.2878046427455197e-05, "loss": 0.1399, "step": 37556 }, { "epoch": 27.514285714285712, "grad_norm": 0.6953125, "learning_rate": 1.2875819804426992e-05, "loss": 0.1297, "step": 37557 }, { "epoch": 27.515018315018317, "grad_norm": 0.859375, "learning_rate": 1.2873593361668609e-05, "loss": 0.1065, "step": 37558 }, { "epoch": 27.515750915750917, "grad_norm": 0.72265625, "learning_rate": 1.2871367099190684e-05, "loss": 0.1032, "step": 37559 }, { "epoch": 27.516483516483518, "grad_norm": 1.59375, "learning_rate": 1.2869141017003902e-05, "loss": 0.1364, "step": 37560 }, { "epoch": 27.51721611721612, "grad_norm": 0.90234375, "learning_rate": 1.2866915115118904e-05, "loss": 0.1029, "step": 37561 }, { "epoch": 27.51794871794872, "grad_norm": 0.85546875, "learning_rate": 1.2864689393546373e-05, "loss": 0.1089, "step": 37562 }, { "epoch": 27.51868131868132, "grad_norm": 0.8515625, "learning_rate": 1.2862463852296969e-05, "loss": 0.1242, "step": 37563 }, { "epoch": 27.51941391941392, "grad_norm": 1.3203125, "learning_rate": 1.2860238491381372e-05, "loss": 0.1327, "step": 37564 }, { "epoch": 27.520146520146522, "grad_norm": 0.97265625, "learning_rate": 1.2858013310810212e-05, "loss": 0.1239, "step": 37565 }, { "epoch": 27.520879120879123, "grad_norm": 0.86328125, "learning_rate": 1.2855788310594168e-05, "loss": 0.1102, "step": 37566 }, { "epoch": 27.521611721611723, "grad_norm": 1.765625, "learning_rate": 1.28535634907439e-05, "loss": 0.1686, "step": 37567 }, { "epoch": 27.522344322344324, "grad_norm": 0.73828125, "learning_rate": 1.2851338851270068e-05, "loss": 0.136, "step": 37568 }, { "epoch": 27.523076923076925, "grad_norm": 1.59375, "learning_rate": 1.2849114392183345e-05, "loss": 0.1442, "step": 37569 }, { "epoch": 27.523809523809526, "grad_norm": 1.6953125, "learning_rate": 1.2846890113494358e-05, "loss": 0.133, "step": 37570 }, { "epoch": 27.524542124542126, "grad_norm": 0.81640625, "learning_rate": 1.2844666015213802e-05, "loss": 0.1218, "step": 37571 }, { "epoch": 27.525274725274727, "grad_norm": 0.63671875, "learning_rate": 1.2842442097352277e-05, "loss": 0.0993, "step": 37572 }, { "epoch": 27.526007326007328, "grad_norm": 0.78125, "learning_rate": 1.2840218359920509e-05, "loss": 0.0892, "step": 37573 }, { "epoch": 27.52673992673993, "grad_norm": 0.73828125, "learning_rate": 1.2837994802929094e-05, "loss": 0.1161, "step": 37574 }, { "epoch": 27.52747252747253, "grad_norm": 1.46875, "learning_rate": 1.2835771426388734e-05, "loss": 0.136, "step": 37575 }, { "epoch": 27.52820512820513, "grad_norm": 1.125, "learning_rate": 1.2833548230310038e-05, "loss": 0.1095, "step": 37576 }, { "epoch": 27.52893772893773, "grad_norm": 1.046875, "learning_rate": 1.2831325214703679e-05, "loss": 0.0988, "step": 37577 }, { "epoch": 27.52967032967033, "grad_norm": 1.171875, "learning_rate": 1.2829102379580301e-05, "loss": 0.0931, "step": 37578 }, { "epoch": 27.530402930402932, "grad_norm": 1.21875, "learning_rate": 1.282687972495058e-05, "loss": 0.1436, "step": 37579 }, { "epoch": 27.531135531135533, "grad_norm": 0.88671875, "learning_rate": 1.2824657250825136e-05, "loss": 0.1242, "step": 37580 }, { "epoch": 27.531868131868134, "grad_norm": 1.1796875, "learning_rate": 1.2822434957214618e-05, "loss": 0.131, "step": 37581 }, { "epoch": 27.532600732600734, "grad_norm": 0.921875, "learning_rate": 1.2820212844129704e-05, "loss": 0.1121, "step": 37582 }, { "epoch": 27.533333333333335, "grad_norm": 0.93359375, "learning_rate": 1.2817990911580987e-05, "loss": 0.1046, "step": 37583 }, { "epoch": 27.534065934065936, "grad_norm": 0.69921875, "learning_rate": 1.2815769159579175e-05, "loss": 0.122, "step": 37584 }, { "epoch": 27.534798534798536, "grad_norm": 0.8671875, "learning_rate": 1.281354758813487e-05, "loss": 0.1344, "step": 37585 }, { "epoch": 27.535531135531137, "grad_norm": 0.77734375, "learning_rate": 1.2811326197258747e-05, "loss": 0.1245, "step": 37586 }, { "epoch": 27.536263736263738, "grad_norm": 1.2734375, "learning_rate": 1.2809104986961413e-05, "loss": 0.1317, "step": 37587 }, { "epoch": 27.53699633699634, "grad_norm": 0.859375, "learning_rate": 1.2806883957253534e-05, "loss": 0.126, "step": 37588 }, { "epoch": 27.53772893772894, "grad_norm": 1.4609375, "learning_rate": 1.2804663108145745e-05, "loss": 0.1395, "step": 37589 }, { "epoch": 27.53846153846154, "grad_norm": 1.296875, "learning_rate": 1.2802442439648707e-05, "loss": 0.1194, "step": 37590 }, { "epoch": 27.53919413919414, "grad_norm": 0.8359375, "learning_rate": 1.2800221951773029e-05, "loss": 0.118, "step": 37591 }, { "epoch": 27.53992673992674, "grad_norm": 1.2578125, "learning_rate": 1.2798001644529361e-05, "loss": 0.1293, "step": 37592 }, { "epoch": 27.540659340659342, "grad_norm": 0.9140625, "learning_rate": 1.2795781517928343e-05, "loss": 0.1333, "step": 37593 }, { "epoch": 27.541391941391943, "grad_norm": 1.1171875, "learning_rate": 1.2793561571980611e-05, "loss": 0.1346, "step": 37594 }, { "epoch": 27.542124542124544, "grad_norm": 0.96875, "learning_rate": 1.2791341806696821e-05, "loss": 0.1053, "step": 37595 }, { "epoch": 27.542857142857144, "grad_norm": 0.95703125, "learning_rate": 1.278912222208758e-05, "loss": 0.1213, "step": 37596 }, { "epoch": 27.543589743589745, "grad_norm": 1.546875, "learning_rate": 1.2786902818163545e-05, "loss": 0.0988, "step": 37597 }, { "epoch": 27.544322344322346, "grad_norm": 1.1953125, "learning_rate": 1.2784683594935312e-05, "loss": 0.1516, "step": 37598 }, { "epoch": 27.545054945054947, "grad_norm": 1.265625, "learning_rate": 1.2782464552413573e-05, "loss": 0.1221, "step": 37599 }, { "epoch": 27.545787545787547, "grad_norm": 1.5234375, "learning_rate": 1.2780245690608904e-05, "loss": 0.1367, "step": 37600 }, { "epoch": 27.546520146520148, "grad_norm": 0.9140625, "learning_rate": 1.2778027009531984e-05, "loss": 0.1413, "step": 37601 }, { "epoch": 27.54725274725275, "grad_norm": 1.546875, "learning_rate": 1.2775808509193405e-05, "loss": 0.129, "step": 37602 }, { "epoch": 27.54798534798535, "grad_norm": 0.828125, "learning_rate": 1.2773590189603809e-05, "loss": 0.1207, "step": 37603 }, { "epoch": 27.54871794871795, "grad_norm": 0.90625, "learning_rate": 1.2771372050773828e-05, "loss": 0.1217, "step": 37604 }, { "epoch": 27.54945054945055, "grad_norm": 1.0390625, "learning_rate": 1.2769154092714101e-05, "loss": 0.1359, "step": 37605 }, { "epoch": 27.550183150183152, "grad_norm": 0.875, "learning_rate": 1.2766936315435227e-05, "loss": 0.0882, "step": 37606 }, { "epoch": 27.550915750915753, "grad_norm": 0.9375, "learning_rate": 1.2764718718947855e-05, "loss": 0.1236, "step": 37607 }, { "epoch": 27.551648351648353, "grad_norm": 0.93359375, "learning_rate": 1.2762501303262611e-05, "loss": 0.11, "step": 37608 }, { "epoch": 27.552380952380954, "grad_norm": 0.68359375, "learning_rate": 1.2760284068390086e-05, "loss": 0.1039, "step": 37609 }, { "epoch": 27.553113553113555, "grad_norm": 0.72265625, "learning_rate": 1.2758067014340961e-05, "loss": 0.1033, "step": 37610 }, { "epoch": 27.553846153846155, "grad_norm": 0.73828125, "learning_rate": 1.2755850141125805e-05, "loss": 0.1013, "step": 37611 }, { "epoch": 27.554578754578756, "grad_norm": 0.69921875, "learning_rate": 1.2753633448755284e-05, "loss": 0.1069, "step": 37612 }, { "epoch": 27.555311355311357, "grad_norm": 0.8515625, "learning_rate": 1.2751416937239974e-05, "loss": 0.123, "step": 37613 }, { "epoch": 27.556043956043958, "grad_norm": 1.2890625, "learning_rate": 1.2749200606590522e-05, "loss": 0.1148, "step": 37614 }, { "epoch": 27.55677655677656, "grad_norm": 1.1484375, "learning_rate": 1.2746984456817543e-05, "loss": 0.1322, "step": 37615 }, { "epoch": 27.55750915750916, "grad_norm": 1.1875, "learning_rate": 1.2744768487931665e-05, "loss": 0.1181, "step": 37616 }, { "epoch": 27.55824175824176, "grad_norm": 0.9140625, "learning_rate": 1.2742552699943488e-05, "loss": 0.1126, "step": 37617 }, { "epoch": 27.55897435897436, "grad_norm": 0.83984375, "learning_rate": 1.2740337092863632e-05, "loss": 0.1286, "step": 37618 }, { "epoch": 27.55970695970696, "grad_norm": 0.7109375, "learning_rate": 1.2738121666702716e-05, "loss": 0.1259, "step": 37619 }, { "epoch": 27.560439560439562, "grad_norm": 1.7109375, "learning_rate": 1.2735906421471358e-05, "loss": 0.1125, "step": 37620 }, { "epoch": 27.561172161172163, "grad_norm": 1.5234375, "learning_rate": 1.2733691357180184e-05, "loss": 0.136, "step": 37621 }, { "epoch": 27.561904761904763, "grad_norm": 0.9453125, "learning_rate": 1.2731476473839776e-05, "loss": 0.0999, "step": 37622 }, { "epoch": 27.562637362637364, "grad_norm": 0.85546875, "learning_rate": 1.2729261771460776e-05, "loss": 0.1267, "step": 37623 }, { "epoch": 27.563369963369965, "grad_norm": 1.390625, "learning_rate": 1.2727047250053773e-05, "loss": 0.1051, "step": 37624 }, { "epoch": 27.564102564102566, "grad_norm": 0.984375, "learning_rate": 1.272483290962938e-05, "loss": 0.1523, "step": 37625 }, { "epoch": 27.564835164835166, "grad_norm": 0.8359375, "learning_rate": 1.272261875019822e-05, "loss": 0.0976, "step": 37626 }, { "epoch": 27.565567765567767, "grad_norm": 1.109375, "learning_rate": 1.2720404771770902e-05, "loss": 0.1324, "step": 37627 }, { "epoch": 27.566300366300368, "grad_norm": 0.80859375, "learning_rate": 1.271819097435802e-05, "loss": 0.089, "step": 37628 }, { "epoch": 27.56703296703297, "grad_norm": 0.95703125, "learning_rate": 1.2715977357970188e-05, "loss": 0.1113, "step": 37629 }, { "epoch": 27.56776556776557, "grad_norm": 1.1171875, "learning_rate": 1.2713763922618006e-05, "loss": 0.1365, "step": 37630 }, { "epoch": 27.56849816849817, "grad_norm": 1.796875, "learning_rate": 1.2711550668312103e-05, "loss": 0.125, "step": 37631 }, { "epoch": 27.56923076923077, "grad_norm": 0.9609375, "learning_rate": 1.2709337595063053e-05, "loss": 0.1072, "step": 37632 }, { "epoch": 27.56996336996337, "grad_norm": 1.0625, "learning_rate": 1.2707124702881468e-05, "loss": 0.115, "step": 37633 }, { "epoch": 27.570695970695972, "grad_norm": 0.9921875, "learning_rate": 1.2704911991777974e-05, "loss": 0.1131, "step": 37634 }, { "epoch": 27.571428571428573, "grad_norm": 0.7890625, "learning_rate": 1.270269946176312e-05, "loss": 0.122, "step": 37635 }, { "epoch": 27.572161172161174, "grad_norm": 1.53125, "learning_rate": 1.2700487112847578e-05, "loss": 0.1203, "step": 37636 }, { "epoch": 27.572893772893774, "grad_norm": 1.265625, "learning_rate": 1.2698274945041888e-05, "loss": 0.1147, "step": 37637 }, { "epoch": 27.573626373626375, "grad_norm": 1.6640625, "learning_rate": 1.2696062958356695e-05, "loss": 0.1543, "step": 37638 }, { "epoch": 27.574358974358976, "grad_norm": 0.93359375, "learning_rate": 1.2693851152802555e-05, "loss": 0.1233, "step": 37639 }, { "epoch": 27.575091575091577, "grad_norm": 1.2734375, "learning_rate": 1.2691639528390085e-05, "loss": 0.1399, "step": 37640 }, { "epoch": 27.575824175824177, "grad_norm": 0.984375, "learning_rate": 1.2689428085129878e-05, "loss": 0.1146, "step": 37641 }, { "epoch": 27.576556776556778, "grad_norm": 0.97265625, "learning_rate": 1.2687216823032555e-05, "loss": 0.1215, "step": 37642 }, { "epoch": 27.57728937728938, "grad_norm": 0.828125, "learning_rate": 1.2685005742108674e-05, "loss": 0.1236, "step": 37643 }, { "epoch": 27.57802197802198, "grad_norm": 1.0546875, "learning_rate": 1.2682794842368838e-05, "loss": 0.1354, "step": 37644 }, { "epoch": 27.57875457875458, "grad_norm": 1.1953125, "learning_rate": 1.2680584123823652e-05, "loss": 0.136, "step": 37645 }, { "epoch": 27.57948717948718, "grad_norm": 0.94140625, "learning_rate": 1.2678373586483705e-05, "loss": 0.1156, "step": 37646 }, { "epoch": 27.58021978021978, "grad_norm": 1.1796875, "learning_rate": 1.2676163230359595e-05, "loss": 0.0967, "step": 37647 }, { "epoch": 27.580952380952382, "grad_norm": 0.7421875, "learning_rate": 1.2673953055461894e-05, "loss": 0.114, "step": 37648 }, { "epoch": 27.581684981684983, "grad_norm": 1.0625, "learning_rate": 1.2671743061801211e-05, "loss": 0.1042, "step": 37649 }, { "epoch": 27.582417582417584, "grad_norm": 1.1484375, "learning_rate": 1.2669533249388116e-05, "loss": 0.1215, "step": 37650 }, { "epoch": 27.583150183150185, "grad_norm": 1.2890625, "learning_rate": 1.2667323618233201e-05, "loss": 0.1352, "step": 37651 }, { "epoch": 27.583882783882785, "grad_norm": 0.6953125, "learning_rate": 1.2665114168347065e-05, "loss": 0.1201, "step": 37652 }, { "epoch": 27.584615384615386, "grad_norm": 0.63671875, "learning_rate": 1.266290489974029e-05, "loss": 0.1289, "step": 37653 }, { "epoch": 27.585347985347987, "grad_norm": 1.578125, "learning_rate": 1.2660695812423454e-05, "loss": 0.1011, "step": 37654 }, { "epoch": 27.586080586080588, "grad_norm": 1.0625, "learning_rate": 1.265848690640714e-05, "loss": 0.1105, "step": 37655 }, { "epoch": 27.58681318681319, "grad_norm": 0.81640625, "learning_rate": 1.2656278181701944e-05, "loss": 0.1185, "step": 37656 }, { "epoch": 27.58754578754579, "grad_norm": 0.875, "learning_rate": 1.2654069638318442e-05, "loss": 0.0956, "step": 37657 }, { "epoch": 27.58827838827839, "grad_norm": 0.96875, "learning_rate": 1.265186127626723e-05, "loss": 0.1088, "step": 37658 }, { "epoch": 27.58901098901099, "grad_norm": 0.8984375, "learning_rate": 1.2649653095558855e-05, "loss": 0.1343, "step": 37659 }, { "epoch": 27.58974358974359, "grad_norm": 1.46875, "learning_rate": 1.2647445096203942e-05, "loss": 0.1267, "step": 37660 }, { "epoch": 27.590476190476192, "grad_norm": 0.8671875, "learning_rate": 1.2645237278213013e-05, "loss": 0.1144, "step": 37661 }, { "epoch": 27.591208791208793, "grad_norm": 0.88671875, "learning_rate": 1.2643029641596707e-05, "loss": 0.1361, "step": 37662 }, { "epoch": 27.591941391941393, "grad_norm": 1.2734375, "learning_rate": 1.2640822186365559e-05, "loss": 0.1285, "step": 37663 }, { "epoch": 27.592673992673994, "grad_norm": 1.1171875, "learning_rate": 1.2638614912530173e-05, "loss": 0.0913, "step": 37664 }, { "epoch": 27.593406593406595, "grad_norm": 1.0546875, "learning_rate": 1.2636407820101095e-05, "loss": 0.1104, "step": 37665 }, { "epoch": 27.594139194139196, "grad_norm": 0.96875, "learning_rate": 1.2634200909088922e-05, "loss": 0.1351, "step": 37666 }, { "epoch": 27.594871794871796, "grad_norm": 1.125, "learning_rate": 1.2631994179504226e-05, "loss": 0.1451, "step": 37667 }, { "epoch": 27.595604395604397, "grad_norm": 1.0234375, "learning_rate": 1.2629787631357585e-05, "loss": 0.1012, "step": 37668 }, { "epoch": 27.596336996336998, "grad_norm": 0.984375, "learning_rate": 1.2627581264659551e-05, "loss": 0.1469, "step": 37669 }, { "epoch": 27.5970695970696, "grad_norm": 1.0234375, "learning_rate": 1.2625375079420709e-05, "loss": 0.091, "step": 37670 }, { "epoch": 27.5978021978022, "grad_norm": 0.8984375, "learning_rate": 1.2623169075651628e-05, "loss": 0.1195, "step": 37671 }, { "epoch": 27.5985347985348, "grad_norm": 1.4453125, "learning_rate": 1.262096325336288e-05, "loss": 0.1172, "step": 37672 }, { "epoch": 27.5992673992674, "grad_norm": 1.265625, "learning_rate": 1.2618757612565043e-05, "loss": 0.1502, "step": 37673 }, { "epoch": 27.6, "grad_norm": 0.92578125, "learning_rate": 1.261655215326866e-05, "loss": 0.1193, "step": 37674 }, { "epoch": 27.600732600732602, "grad_norm": 0.9609375, "learning_rate": 1.2614346875484328e-05, "loss": 0.1316, "step": 37675 }, { "epoch": 27.601465201465203, "grad_norm": 1.3046875, "learning_rate": 1.261214177922258e-05, "loss": 0.1319, "step": 37676 }, { "epoch": 27.602197802197804, "grad_norm": 1.0390625, "learning_rate": 1.2609936864494004e-05, "loss": 0.1457, "step": 37677 }, { "epoch": 27.602930402930404, "grad_norm": 1.359375, "learning_rate": 1.2607732131309152e-05, "loss": 0.097, "step": 37678 }, { "epoch": 27.603663003663005, "grad_norm": 0.99609375, "learning_rate": 1.2605527579678614e-05, "loss": 0.1311, "step": 37679 }, { "epoch": 27.604395604395606, "grad_norm": 0.953125, "learning_rate": 1.2603323209612919e-05, "loss": 0.1104, "step": 37680 }, { "epoch": 27.605128205128207, "grad_norm": 0.95703125, "learning_rate": 1.260111902112264e-05, "loss": 0.1368, "step": 37681 }, { "epoch": 27.605860805860807, "grad_norm": 1.0703125, "learning_rate": 1.259891501421834e-05, "loss": 0.133, "step": 37682 }, { "epoch": 27.606593406593408, "grad_norm": 1.2109375, "learning_rate": 1.2596711188910588e-05, "loss": 0.1337, "step": 37683 }, { "epoch": 27.60732600732601, "grad_norm": 1.25, "learning_rate": 1.2594507545209941e-05, "loss": 0.1248, "step": 37684 }, { "epoch": 27.60805860805861, "grad_norm": 0.88671875, "learning_rate": 1.259230408312694e-05, "loss": 0.1353, "step": 37685 }, { "epoch": 27.60879120879121, "grad_norm": 1.5859375, "learning_rate": 1.259010080267217e-05, "loss": 0.1024, "step": 37686 }, { "epoch": 27.60952380952381, "grad_norm": 0.51953125, "learning_rate": 1.2587897703856146e-05, "loss": 0.091, "step": 37687 }, { "epoch": 27.61025641025641, "grad_norm": 1.078125, "learning_rate": 1.2585694786689472e-05, "loss": 0.1098, "step": 37688 }, { "epoch": 27.610989010989012, "grad_norm": 0.83984375, "learning_rate": 1.2583492051182674e-05, "loss": 0.1047, "step": 37689 }, { "epoch": 27.611721611721613, "grad_norm": 1.3125, "learning_rate": 1.2581289497346321e-05, "loss": 0.1027, "step": 37690 }, { "epoch": 27.612454212454214, "grad_norm": 0.85546875, "learning_rate": 1.2579087125190943e-05, "loss": 0.088, "step": 37691 }, { "epoch": 27.613186813186815, "grad_norm": 1.34375, "learning_rate": 1.2576884934727118e-05, "loss": 0.1393, "step": 37692 }, { "epoch": 27.613919413919415, "grad_norm": 1.2578125, "learning_rate": 1.2574682925965375e-05, "loss": 0.1412, "step": 37693 }, { "epoch": 27.614652014652016, "grad_norm": 1.1484375, "learning_rate": 1.2572481098916298e-05, "loss": 0.1266, "step": 37694 }, { "epoch": 27.615384615384617, "grad_norm": 1.4921875, "learning_rate": 1.2570279453590395e-05, "loss": 0.1143, "step": 37695 }, { "epoch": 27.616117216117217, "grad_norm": 1.1328125, "learning_rate": 1.2568077989998234e-05, "loss": 0.1129, "step": 37696 }, { "epoch": 27.616849816849818, "grad_norm": 1.2109375, "learning_rate": 1.2565876708150368e-05, "loss": 0.1033, "step": 37697 }, { "epoch": 27.61758241758242, "grad_norm": 0.953125, "learning_rate": 1.2563675608057337e-05, "loss": 0.1518, "step": 37698 }, { "epoch": 27.61831501831502, "grad_norm": 0.61328125, "learning_rate": 1.2561474689729705e-05, "loss": 0.0925, "step": 37699 }, { "epoch": 27.61904761904762, "grad_norm": 1.046875, "learning_rate": 1.2559273953177988e-05, "loss": 0.1325, "step": 37700 }, { "epoch": 27.61978021978022, "grad_norm": 0.82421875, "learning_rate": 1.2557073398412762e-05, "loss": 0.0988, "step": 37701 }, { "epoch": 27.620512820512822, "grad_norm": 1.046875, "learning_rate": 1.2554873025444533e-05, "loss": 0.0923, "step": 37702 }, { "epoch": 27.621245421245423, "grad_norm": 1.34375, "learning_rate": 1.2552672834283869e-05, "loss": 0.1279, "step": 37703 }, { "epoch": 27.621978021978023, "grad_norm": 0.89453125, "learning_rate": 1.2550472824941301e-05, "loss": 0.1218, "step": 37704 }, { "epoch": 27.622710622710624, "grad_norm": 1.0625, "learning_rate": 1.2548272997427392e-05, "loss": 0.1085, "step": 37705 }, { "epoch": 27.623443223443225, "grad_norm": 1.109375, "learning_rate": 1.2546073351752655e-05, "loss": 0.1055, "step": 37706 }, { "epoch": 27.624175824175826, "grad_norm": 0.9453125, "learning_rate": 1.2543873887927637e-05, "loss": 0.1316, "step": 37707 }, { "epoch": 27.624908424908426, "grad_norm": 1.0546875, "learning_rate": 1.254167460596288e-05, "loss": 0.1369, "step": 37708 }, { "epoch": 27.625641025641027, "grad_norm": 1.6953125, "learning_rate": 1.2539475505868926e-05, "loss": 0.1565, "step": 37709 }, { "epoch": 27.626373626373628, "grad_norm": 1.0390625, "learning_rate": 1.2537276587656311e-05, "loss": 0.1435, "step": 37710 }, { "epoch": 27.62710622710623, "grad_norm": 0.9921875, "learning_rate": 1.2535077851335561e-05, "loss": 0.0936, "step": 37711 }, { "epoch": 27.62783882783883, "grad_norm": 1.4375, "learning_rate": 1.2532879296917224e-05, "loss": 0.1807, "step": 37712 }, { "epoch": 27.62857142857143, "grad_norm": 0.7109375, "learning_rate": 1.2530680924411798e-05, "loss": 0.1107, "step": 37713 }, { "epoch": 27.62930402930403, "grad_norm": 1.46875, "learning_rate": 1.2528482733829872e-05, "loss": 0.1561, "step": 37714 }, { "epoch": 27.63003663003663, "grad_norm": 0.921875, "learning_rate": 1.2526284725181939e-05, "loss": 0.1164, "step": 37715 }, { "epoch": 27.630769230769232, "grad_norm": 1.171875, "learning_rate": 1.2524086898478555e-05, "loss": 0.1237, "step": 37716 }, { "epoch": 27.631501831501833, "grad_norm": 1.40625, "learning_rate": 1.2521889253730226e-05, "loss": 0.1156, "step": 37717 }, { "epoch": 27.632234432234434, "grad_norm": 0.84765625, "learning_rate": 1.2519691790947491e-05, "loss": 0.1064, "step": 37718 }, { "epoch": 27.632967032967034, "grad_norm": 1.2421875, "learning_rate": 1.2517494510140882e-05, "loss": 0.1117, "step": 37719 }, { "epoch": 27.633699633699635, "grad_norm": 0.6953125, "learning_rate": 1.251529741132094e-05, "loss": 0.1263, "step": 37720 }, { "epoch": 27.634432234432236, "grad_norm": 0.87890625, "learning_rate": 1.2513100494498162e-05, "loss": 0.1138, "step": 37721 }, { "epoch": 27.635164835164836, "grad_norm": 1.3125, "learning_rate": 1.2510903759683087e-05, "loss": 0.114, "step": 37722 }, { "epoch": 27.635897435897437, "grad_norm": 1.2421875, "learning_rate": 1.2508707206886247e-05, "loss": 0.1183, "step": 37723 }, { "epoch": 27.636630036630038, "grad_norm": 1.4453125, "learning_rate": 1.250651083611816e-05, "loss": 0.1572, "step": 37724 }, { "epoch": 27.63736263736264, "grad_norm": 0.84765625, "learning_rate": 1.2504314647389365e-05, "loss": 0.1253, "step": 37725 }, { "epoch": 27.63809523809524, "grad_norm": 1.203125, "learning_rate": 1.250211864071036e-05, "loss": 0.1081, "step": 37726 }, { "epoch": 27.63882783882784, "grad_norm": 1.0234375, "learning_rate": 1.2499922816091691e-05, "loss": 0.103, "step": 37727 }, { "epoch": 27.63956043956044, "grad_norm": 0.70703125, "learning_rate": 1.2497727173543855e-05, "loss": 0.1277, "step": 37728 }, { "epoch": 27.64029304029304, "grad_norm": 0.98046875, "learning_rate": 1.2495531713077386e-05, "loss": 0.1078, "step": 37729 }, { "epoch": 27.641025641025642, "grad_norm": 1.3828125, "learning_rate": 1.2493336434702794e-05, "loss": 0.1218, "step": 37730 }, { "epoch": 27.641758241758243, "grad_norm": 0.8125, "learning_rate": 1.249114133843062e-05, "loss": 0.1262, "step": 37731 }, { "epoch": 27.642490842490844, "grad_norm": 1.1875, "learning_rate": 1.2488946424271354e-05, "loss": 0.1169, "step": 37732 }, { "epoch": 27.643223443223444, "grad_norm": 0.63671875, "learning_rate": 1.248675169223552e-05, "loss": 0.1001, "step": 37733 }, { "epoch": 27.643956043956045, "grad_norm": 0.609375, "learning_rate": 1.2484557142333639e-05, "loss": 0.1253, "step": 37734 }, { "epoch": 27.644688644688646, "grad_norm": 1.28125, "learning_rate": 1.2482362774576226e-05, "loss": 0.1227, "step": 37735 }, { "epoch": 27.645421245421247, "grad_norm": 0.7734375, "learning_rate": 1.2480168588973806e-05, "loss": 0.1157, "step": 37736 }, { "epoch": 27.646153846153847, "grad_norm": 1.1953125, "learning_rate": 1.247797458553687e-05, "loss": 0.1213, "step": 37737 }, { "epoch": 27.646886446886448, "grad_norm": 1.4296875, "learning_rate": 1.2475780764275947e-05, "loss": 0.1294, "step": 37738 }, { "epoch": 27.64761904761905, "grad_norm": 1.109375, "learning_rate": 1.2473587125201515e-05, "loss": 0.1153, "step": 37739 }, { "epoch": 27.64835164835165, "grad_norm": 1.2734375, "learning_rate": 1.2471393668324138e-05, "loss": 0.1091, "step": 37740 }, { "epoch": 27.64908424908425, "grad_norm": 0.9140625, "learning_rate": 1.2469200393654282e-05, "loss": 0.1369, "step": 37741 }, { "epoch": 27.64981684981685, "grad_norm": 1.03125, "learning_rate": 1.246700730120249e-05, "loss": 0.1005, "step": 37742 }, { "epoch": 27.650549450549452, "grad_norm": 1.34375, "learning_rate": 1.2464814390979234e-05, "loss": 0.1156, "step": 37743 }, { "epoch": 27.651282051282053, "grad_norm": 0.98046875, "learning_rate": 1.246262166299504e-05, "loss": 0.1146, "step": 37744 }, { "epoch": 27.652014652014653, "grad_norm": 0.75, "learning_rate": 1.2460429117260412e-05, "loss": 0.1179, "step": 37745 }, { "epoch": 27.652747252747254, "grad_norm": 0.859375, "learning_rate": 1.2458236753785876e-05, "loss": 0.1175, "step": 37746 }, { "epoch": 27.653479853479855, "grad_norm": 1.265625, "learning_rate": 1.2456044572581891e-05, "loss": 0.1021, "step": 37747 }, { "epoch": 27.654212454212455, "grad_norm": 1.0703125, "learning_rate": 1.2453852573658991e-05, "loss": 0.0964, "step": 37748 }, { "epoch": 27.654945054945056, "grad_norm": 1.125, "learning_rate": 1.2451660757027677e-05, "loss": 0.136, "step": 37749 }, { "epoch": 27.655677655677657, "grad_norm": 0.83203125, "learning_rate": 1.2449469122698437e-05, "loss": 0.1232, "step": 37750 }, { "epoch": 27.656410256410258, "grad_norm": 1.5, "learning_rate": 1.2447277670681805e-05, "loss": 0.1588, "step": 37751 }, { "epoch": 27.65714285714286, "grad_norm": 0.6796875, "learning_rate": 1.2445086400988235e-05, "loss": 0.1242, "step": 37752 }, { "epoch": 27.65787545787546, "grad_norm": 0.68359375, "learning_rate": 1.2442895313628272e-05, "loss": 0.1176, "step": 37753 }, { "epoch": 27.65860805860806, "grad_norm": 1.4140625, "learning_rate": 1.244070440861237e-05, "loss": 0.1286, "step": 37754 }, { "epoch": 27.65934065934066, "grad_norm": 0.80078125, "learning_rate": 1.243851368595105e-05, "loss": 0.1213, "step": 37755 }, { "epoch": 27.66007326007326, "grad_norm": 1.171875, "learning_rate": 1.2436323145654802e-05, "loss": 0.1416, "step": 37756 }, { "epoch": 27.660805860805862, "grad_norm": 1.34375, "learning_rate": 1.2434132787734143e-05, "loss": 0.131, "step": 37757 }, { "epoch": 27.661538461538463, "grad_norm": 0.7578125, "learning_rate": 1.2431942612199536e-05, "loss": 0.1435, "step": 37758 }, { "epoch": 27.662271062271063, "grad_norm": 1.0703125, "learning_rate": 1.2429752619061486e-05, "loss": 0.1198, "step": 37759 }, { "epoch": 27.663003663003664, "grad_norm": 0.8984375, "learning_rate": 1.2427562808330494e-05, "loss": 0.1264, "step": 37760 }, { "epoch": 27.663736263736265, "grad_norm": 0.6953125, "learning_rate": 1.2425373180017043e-05, "loss": 0.1297, "step": 37761 }, { "epoch": 27.664468864468866, "grad_norm": 1.1484375, "learning_rate": 1.2423183734131643e-05, "loss": 0.1451, "step": 37762 }, { "epoch": 27.665201465201466, "grad_norm": 0.9296875, "learning_rate": 1.2420994470684754e-05, "loss": 0.1194, "step": 37763 }, { "epoch": 27.665934065934067, "grad_norm": 0.9921875, "learning_rate": 1.24188053896869e-05, "loss": 0.1356, "step": 37764 }, { "epoch": 27.666666666666668, "grad_norm": 0.94921875, "learning_rate": 1.241661649114852e-05, "loss": 0.1319, "step": 37765 }, { "epoch": 27.66739926739927, "grad_norm": 1.046875, "learning_rate": 1.2414427775080158e-05, "loss": 0.119, "step": 37766 }, { "epoch": 27.66813186813187, "grad_norm": 1.15625, "learning_rate": 1.2412239241492264e-05, "loss": 0.1289, "step": 37767 }, { "epoch": 27.66886446886447, "grad_norm": 0.97265625, "learning_rate": 1.2410050890395349e-05, "loss": 0.1109, "step": 37768 }, { "epoch": 27.66959706959707, "grad_norm": 1.4375, "learning_rate": 1.2407862721799866e-05, "loss": 0.1272, "step": 37769 }, { "epoch": 27.67032967032967, "grad_norm": 0.7734375, "learning_rate": 1.2405674735716322e-05, "loss": 0.101, "step": 37770 }, { "epoch": 27.671062271062272, "grad_norm": 1.0703125, "learning_rate": 1.2403486932155195e-05, "loss": 0.1468, "step": 37771 }, { "epoch": 27.671794871794873, "grad_norm": 0.66015625, "learning_rate": 1.2401299311126965e-05, "loss": 0.0955, "step": 37772 }, { "epoch": 27.672527472527474, "grad_norm": 0.7109375, "learning_rate": 1.2399111872642136e-05, "loss": 0.1384, "step": 37773 }, { "epoch": 27.673260073260074, "grad_norm": 0.94921875, "learning_rate": 1.2396924616711152e-05, "loss": 0.1107, "step": 37774 }, { "epoch": 27.673992673992675, "grad_norm": 1.8515625, "learning_rate": 1.2394737543344517e-05, "loss": 0.1201, "step": 37775 }, { "epoch": 27.674725274725276, "grad_norm": 1.0625, "learning_rate": 1.2392550652552699e-05, "loss": 0.1201, "step": 37776 }, { "epoch": 27.675457875457877, "grad_norm": 0.6328125, "learning_rate": 1.239036394434619e-05, "loss": 0.1191, "step": 37777 }, { "epoch": 27.676190476190477, "grad_norm": 1.421875, "learning_rate": 1.2388177418735448e-05, "loss": 0.1249, "step": 37778 }, { "epoch": 27.676923076923078, "grad_norm": 1.1484375, "learning_rate": 1.2385991075730976e-05, "loss": 0.1085, "step": 37779 }, { "epoch": 27.67765567765568, "grad_norm": 1.390625, "learning_rate": 1.2383804915343216e-05, "loss": 0.1252, "step": 37780 }, { "epoch": 27.67838827838828, "grad_norm": 0.80859375, "learning_rate": 1.2381618937582661e-05, "loss": 0.1054, "step": 37781 }, { "epoch": 27.67912087912088, "grad_norm": 0.578125, "learning_rate": 1.2379433142459784e-05, "loss": 0.101, "step": 37782 }, { "epoch": 27.67985347985348, "grad_norm": 1.03125, "learning_rate": 1.2377247529985067e-05, "loss": 0.1011, "step": 37783 }, { "epoch": 27.68058608058608, "grad_norm": 1.5546875, "learning_rate": 1.237506210016896e-05, "loss": 0.1599, "step": 37784 }, { "epoch": 27.681318681318682, "grad_norm": 1.171875, "learning_rate": 1.2372876853021949e-05, "loss": 0.1295, "step": 37785 }, { "epoch": 27.682051282051283, "grad_norm": 0.75390625, "learning_rate": 1.2370691788554494e-05, "loss": 0.1, "step": 37786 }, { "epoch": 27.682783882783884, "grad_norm": 1.5625, "learning_rate": 1.236850690677708e-05, "loss": 0.1039, "step": 37787 }, { "epoch": 27.683516483516485, "grad_norm": 1.703125, "learning_rate": 1.2366322207700174e-05, "loss": 0.14, "step": 37788 }, { "epoch": 27.684249084249085, "grad_norm": 1.203125, "learning_rate": 1.2364137691334226e-05, "loss": 0.0947, "step": 37789 }, { "epoch": 27.684981684981686, "grad_norm": 0.84375, "learning_rate": 1.2361953357689726e-05, "loss": 0.1172, "step": 37790 }, { "epoch": 27.685714285714287, "grad_norm": 0.9765625, "learning_rate": 1.2359769206777102e-05, "loss": 0.1201, "step": 37791 }, { "epoch": 27.686446886446888, "grad_norm": 0.88671875, "learning_rate": 1.2357585238606875e-05, "loss": 0.118, "step": 37792 }, { "epoch": 27.68717948717949, "grad_norm": 0.85546875, "learning_rate": 1.235540145318946e-05, "loss": 0.1064, "step": 37793 }, { "epoch": 27.68791208791209, "grad_norm": 1.5078125, "learning_rate": 1.2353217850535353e-05, "loss": 0.1118, "step": 37794 }, { "epoch": 27.68864468864469, "grad_norm": 0.6328125, "learning_rate": 1.2351034430654987e-05, "loss": 0.0995, "step": 37795 }, { "epoch": 27.68937728937729, "grad_norm": 1.046875, "learning_rate": 1.2348851193558843e-05, "loss": 0.1187, "step": 37796 }, { "epoch": 27.69010989010989, "grad_norm": 2.015625, "learning_rate": 1.2346668139257369e-05, "loss": 0.1234, "step": 37797 }, { "epoch": 27.690842490842492, "grad_norm": 0.57421875, "learning_rate": 1.234448526776104e-05, "loss": 0.1144, "step": 37798 }, { "epoch": 27.691575091575093, "grad_norm": 0.77734375, "learning_rate": 1.2342302579080323e-05, "loss": 0.1252, "step": 37799 }, { "epoch": 27.692307692307693, "grad_norm": 0.83203125, "learning_rate": 1.234012007322564e-05, "loss": 0.1049, "step": 37800 }, { "epoch": 27.693040293040294, "grad_norm": 1.140625, "learning_rate": 1.2337937750207473e-05, "loss": 0.1454, "step": 37801 }, { "epoch": 27.693772893772895, "grad_norm": 1.15625, "learning_rate": 1.2335755610036273e-05, "loss": 0.1222, "step": 37802 }, { "epoch": 27.694505494505496, "grad_norm": 1.1875, "learning_rate": 1.2333573652722513e-05, "loss": 0.1034, "step": 37803 }, { "epoch": 27.695238095238096, "grad_norm": 1.328125, "learning_rate": 1.2331391878276614e-05, "loss": 0.1544, "step": 37804 }, { "epoch": 27.695970695970697, "grad_norm": 1.2890625, "learning_rate": 1.2329210286709058e-05, "loss": 0.1159, "step": 37805 }, { "epoch": 27.696703296703298, "grad_norm": 1.078125, "learning_rate": 1.2327028878030274e-05, "loss": 0.1204, "step": 37806 }, { "epoch": 27.6974358974359, "grad_norm": 1.0625, "learning_rate": 1.2324847652250731e-05, "loss": 0.1094, "step": 37807 }, { "epoch": 27.6981684981685, "grad_norm": 1.2109375, "learning_rate": 1.2322666609380868e-05, "loss": 0.1167, "step": 37808 }, { "epoch": 27.6989010989011, "grad_norm": 1.0390625, "learning_rate": 1.232048574943116e-05, "loss": 0.1147, "step": 37809 }, { "epoch": 27.6996336996337, "grad_norm": 0.88671875, "learning_rate": 1.2318305072412024e-05, "loss": 0.1183, "step": 37810 }, { "epoch": 27.7003663003663, "grad_norm": 1.265625, "learning_rate": 1.2316124578333923e-05, "loss": 0.1268, "step": 37811 }, { "epoch": 27.701098901098902, "grad_norm": 0.76953125, "learning_rate": 1.23139442672073e-05, "loss": 0.1099, "step": 37812 }, { "epoch": 27.701831501831503, "grad_norm": 1.421875, "learning_rate": 1.2311764139042611e-05, "loss": 0.152, "step": 37813 }, { "epoch": 27.702564102564104, "grad_norm": 1.40625, "learning_rate": 1.230958419385031e-05, "loss": 0.113, "step": 37814 }, { "epoch": 27.703296703296704, "grad_norm": 1.0546875, "learning_rate": 1.2307404431640808e-05, "loss": 0.1223, "step": 37815 }, { "epoch": 27.704029304029305, "grad_norm": 1.0234375, "learning_rate": 1.2305224852424587e-05, "loss": 0.1032, "step": 37816 }, { "epoch": 27.704761904761906, "grad_norm": 0.91015625, "learning_rate": 1.2303045456212049e-05, "loss": 0.1181, "step": 37817 }, { "epoch": 27.705494505494507, "grad_norm": 1.140625, "learning_rate": 1.2300866243013683e-05, "loss": 0.1419, "step": 37818 }, { "epoch": 27.706227106227107, "grad_norm": 0.93359375, "learning_rate": 1.2298687212839892e-05, "loss": 0.1248, "step": 37819 }, { "epoch": 27.706959706959708, "grad_norm": 1.171875, "learning_rate": 1.2296508365701149e-05, "loss": 0.1016, "step": 37820 }, { "epoch": 27.70769230769231, "grad_norm": 1.25, "learning_rate": 1.2294329701607857e-05, "loss": 0.1801, "step": 37821 }, { "epoch": 27.70842490842491, "grad_norm": 1.34375, "learning_rate": 1.2292151220570473e-05, "loss": 0.1307, "step": 37822 }, { "epoch": 27.70915750915751, "grad_norm": 0.78515625, "learning_rate": 1.2289972922599437e-05, "loss": 0.1337, "step": 37823 }, { "epoch": 27.70989010989011, "grad_norm": 0.90234375, "learning_rate": 1.2287794807705192e-05, "loss": 0.1028, "step": 37824 }, { "epoch": 27.71062271062271, "grad_norm": 0.9765625, "learning_rate": 1.228561687589817e-05, "loss": 0.1087, "step": 37825 }, { "epoch": 27.711355311355312, "grad_norm": 1.109375, "learning_rate": 1.2283439127188794e-05, "loss": 0.1111, "step": 37826 }, { "epoch": 27.712087912087913, "grad_norm": 0.9921875, "learning_rate": 1.2281261561587503e-05, "loss": 0.1473, "step": 37827 }, { "epoch": 27.712820512820514, "grad_norm": 1.3671875, "learning_rate": 1.2279084179104736e-05, "loss": 0.1229, "step": 37828 }, { "epoch": 27.713553113553115, "grad_norm": 1.2265625, "learning_rate": 1.2276906979750939e-05, "loss": 0.1432, "step": 37829 }, { "epoch": 27.714285714285715, "grad_norm": 1.203125, "learning_rate": 1.2274729963536514e-05, "loss": 0.1111, "step": 37830 }, { "epoch": 27.715018315018316, "grad_norm": 1.140625, "learning_rate": 1.227255313047192e-05, "loss": 0.0945, "step": 37831 }, { "epoch": 27.715750915750917, "grad_norm": 1.2578125, "learning_rate": 1.2270376480567554e-05, "loss": 0.0972, "step": 37832 }, { "epoch": 27.716483516483517, "grad_norm": 1.7109375, "learning_rate": 1.2268200013833871e-05, "loss": 0.1606, "step": 37833 }, { "epoch": 27.717216117216118, "grad_norm": 1.28125, "learning_rate": 1.2266023730281294e-05, "loss": 0.1142, "step": 37834 }, { "epoch": 27.71794871794872, "grad_norm": 1.109375, "learning_rate": 1.2263847629920256e-05, "loss": 0.1111, "step": 37835 }, { "epoch": 27.71868131868132, "grad_norm": 0.9921875, "learning_rate": 1.2261671712761165e-05, "loss": 0.1027, "step": 37836 }, { "epoch": 27.71941391941392, "grad_norm": 1.2890625, "learning_rate": 1.225949597881446e-05, "loss": 0.1333, "step": 37837 }, { "epoch": 27.72014652014652, "grad_norm": 1.625, "learning_rate": 1.2257320428090558e-05, "loss": 0.1292, "step": 37838 }, { "epoch": 27.720879120879122, "grad_norm": 1.8046875, "learning_rate": 1.2255145060599893e-05, "loss": 0.1408, "step": 37839 }, { "epoch": 27.721611721611723, "grad_norm": 1.21875, "learning_rate": 1.2252969876352894e-05, "loss": 0.1222, "step": 37840 }, { "epoch": 27.722344322344323, "grad_norm": 1.59375, "learning_rate": 1.2250794875359962e-05, "loss": 0.1494, "step": 37841 }, { "epoch": 27.723076923076924, "grad_norm": 1.0234375, "learning_rate": 1.2248620057631537e-05, "loss": 0.098, "step": 37842 }, { "epoch": 27.723809523809525, "grad_norm": 1.5859375, "learning_rate": 1.2246445423178008e-05, "loss": 0.1539, "step": 37843 }, { "epoch": 27.724542124542126, "grad_norm": 0.90234375, "learning_rate": 1.224427097200984e-05, "loss": 0.1257, "step": 37844 }, { "epoch": 27.725274725274726, "grad_norm": 1.2109375, "learning_rate": 1.2242096704137413e-05, "loss": 0.123, "step": 37845 }, { "epoch": 27.726007326007327, "grad_norm": 1.34375, "learning_rate": 1.2239922619571177e-05, "loss": 0.1195, "step": 37846 }, { "epoch": 27.726739926739928, "grad_norm": 1.140625, "learning_rate": 1.2237748718321523e-05, "loss": 0.1273, "step": 37847 }, { "epoch": 27.72747252747253, "grad_norm": 1.1484375, "learning_rate": 1.223557500039887e-05, "loss": 0.1347, "step": 37848 }, { "epoch": 27.72820512820513, "grad_norm": 0.80859375, "learning_rate": 1.2233401465813636e-05, "loss": 0.122, "step": 37849 }, { "epoch": 27.72893772893773, "grad_norm": 1.2890625, "learning_rate": 1.2231228114576243e-05, "loss": 0.1277, "step": 37850 }, { "epoch": 27.72967032967033, "grad_norm": 0.8203125, "learning_rate": 1.2229054946697111e-05, "loss": 0.1109, "step": 37851 }, { "epoch": 27.73040293040293, "grad_norm": 1.3828125, "learning_rate": 1.2226881962186625e-05, "loss": 0.1164, "step": 37852 }, { "epoch": 27.731135531135532, "grad_norm": 0.90234375, "learning_rate": 1.2224709161055227e-05, "loss": 0.1397, "step": 37853 }, { "epoch": 27.731868131868133, "grad_norm": 0.58203125, "learning_rate": 1.2222536543313282e-05, "loss": 0.1101, "step": 37854 }, { "epoch": 27.732600732600734, "grad_norm": 0.82421875, "learning_rate": 1.2220364108971259e-05, "loss": 0.1012, "step": 37855 }, { "epoch": 27.733333333333334, "grad_norm": 1.0703125, "learning_rate": 1.2218191858039525e-05, "loss": 0.1441, "step": 37856 }, { "epoch": 27.734065934065935, "grad_norm": 1.0390625, "learning_rate": 1.2216019790528513e-05, "loss": 0.1476, "step": 37857 }, { "epoch": 27.734798534798536, "grad_norm": 0.890625, "learning_rate": 1.2213847906448604e-05, "loss": 0.118, "step": 37858 }, { "epoch": 27.735531135531136, "grad_norm": 1.0390625, "learning_rate": 1.2211676205810222e-05, "loss": 0.1403, "step": 37859 }, { "epoch": 27.736263736263737, "grad_norm": 1.0078125, "learning_rate": 1.220950468862377e-05, "loss": 0.1025, "step": 37860 }, { "epoch": 27.736996336996338, "grad_norm": 1.375, "learning_rate": 1.220733335489966e-05, "loss": 0.1374, "step": 37861 }, { "epoch": 27.73772893772894, "grad_norm": 1.1328125, "learning_rate": 1.2205162204648272e-05, "loss": 0.1155, "step": 37862 }, { "epoch": 27.73846153846154, "grad_norm": 0.76171875, "learning_rate": 1.2202991237880028e-05, "loss": 0.0924, "step": 37863 }, { "epoch": 27.73919413919414, "grad_norm": 1.078125, "learning_rate": 1.2200820454605322e-05, "loss": 0.1128, "step": 37864 }, { "epoch": 27.73992673992674, "grad_norm": 1.0234375, "learning_rate": 1.2198649854834559e-05, "loss": 0.1142, "step": 37865 }, { "epoch": 27.74065934065934, "grad_norm": 1.1328125, "learning_rate": 1.219647943857815e-05, "loss": 0.1394, "step": 37866 }, { "epoch": 27.741391941391942, "grad_norm": 0.64453125, "learning_rate": 1.2194309205846471e-05, "loss": 0.1273, "step": 37867 }, { "epoch": 27.742124542124543, "grad_norm": 0.67578125, "learning_rate": 1.2192139156649941e-05, "loss": 0.1121, "step": 37868 }, { "epoch": 27.742857142857144, "grad_norm": 0.75390625, "learning_rate": 1.2189969290998922e-05, "loss": 0.1078, "step": 37869 }, { "epoch": 27.743589743589745, "grad_norm": 1.09375, "learning_rate": 1.2187799608903867e-05, "loss": 0.1097, "step": 37870 }, { "epoch": 27.744322344322345, "grad_norm": 0.7578125, "learning_rate": 1.2185630110375126e-05, "loss": 0.1027, "step": 37871 }, { "epoch": 27.745054945054946, "grad_norm": 1.21875, "learning_rate": 1.2183460795423123e-05, "loss": 0.1322, "step": 37872 }, { "epoch": 27.745787545787547, "grad_norm": 1.09375, "learning_rate": 1.2181291664058222e-05, "loss": 0.1217, "step": 37873 }, { "epoch": 27.746520146520147, "grad_norm": 1.1328125, "learning_rate": 1.2179122716290835e-05, "loss": 0.1177, "step": 37874 }, { "epoch": 27.747252747252748, "grad_norm": 1.1015625, "learning_rate": 1.2176953952131349e-05, "loss": 0.1, "step": 37875 }, { "epoch": 27.74798534798535, "grad_norm": 1.3828125, "learning_rate": 1.2174785371590159e-05, "loss": 0.1061, "step": 37876 }, { "epoch": 27.74871794871795, "grad_norm": 1.28125, "learning_rate": 1.2172616974677666e-05, "loss": 0.0906, "step": 37877 }, { "epoch": 27.74945054945055, "grad_norm": 0.79296875, "learning_rate": 1.2170448761404231e-05, "loss": 0.0908, "step": 37878 }, { "epoch": 27.75018315018315, "grad_norm": 0.609375, "learning_rate": 1.2168280731780275e-05, "loss": 0.0857, "step": 37879 }, { "epoch": 27.750915750915752, "grad_norm": 1.015625, "learning_rate": 1.2166112885816143e-05, "loss": 0.1186, "step": 37880 }, { "epoch": 27.751648351648353, "grad_norm": 1.5703125, "learning_rate": 1.2163945223522275e-05, "loss": 0.1072, "step": 37881 }, { "epoch": 27.752380952380953, "grad_norm": 1.25, "learning_rate": 1.2161777744909015e-05, "loss": 0.1475, "step": 37882 }, { "epoch": 27.753113553113554, "grad_norm": 0.89453125, "learning_rate": 1.215961044998678e-05, "loss": 0.1027, "step": 37883 }, { "epoch": 27.753846153846155, "grad_norm": 1.171875, "learning_rate": 1.2157443338765919e-05, "loss": 0.1243, "step": 37884 }, { "epoch": 27.754578754578755, "grad_norm": 1.25, "learning_rate": 1.2155276411256839e-05, "loss": 0.1315, "step": 37885 }, { "epoch": 27.755311355311356, "grad_norm": 1.0, "learning_rate": 1.2153109667469915e-05, "loss": 0.1062, "step": 37886 }, { "epoch": 27.756043956043957, "grad_norm": 1.21875, "learning_rate": 1.2150943107415536e-05, "loss": 0.1299, "step": 37887 }, { "epoch": 27.756776556776558, "grad_norm": 1.1640625, "learning_rate": 1.2148776731104085e-05, "loss": 0.1211, "step": 37888 }, { "epoch": 27.75750915750916, "grad_norm": 0.80859375, "learning_rate": 1.2146610538545925e-05, "loss": 0.1158, "step": 37889 }, { "epoch": 27.75824175824176, "grad_norm": 1.5234375, "learning_rate": 1.2144444529751441e-05, "loss": 0.1158, "step": 37890 }, { "epoch": 27.75897435897436, "grad_norm": 1.0859375, "learning_rate": 1.2142278704731014e-05, "loss": 0.0945, "step": 37891 }, { "epoch": 27.75970695970696, "grad_norm": 1.1171875, "learning_rate": 1.2140113063495038e-05, "loss": 0.1278, "step": 37892 }, { "epoch": 27.76043956043956, "grad_norm": 0.90234375, "learning_rate": 1.213794760605386e-05, "loss": 0.1377, "step": 37893 }, { "epoch": 27.761172161172162, "grad_norm": 2.109375, "learning_rate": 1.2135782332417883e-05, "loss": 0.1639, "step": 37894 }, { "epoch": 27.761904761904763, "grad_norm": 1.3515625, "learning_rate": 1.2133617242597438e-05, "loss": 0.1563, "step": 37895 }, { "epoch": 27.762637362637363, "grad_norm": 1.015625, "learning_rate": 1.213145233660295e-05, "loss": 0.1405, "step": 37896 }, { "epoch": 27.763369963369964, "grad_norm": 1.140625, "learning_rate": 1.2129287614444765e-05, "loss": 0.1164, "step": 37897 }, { "epoch": 27.764102564102565, "grad_norm": 1.359375, "learning_rate": 1.2127123076133267e-05, "loss": 0.1585, "step": 37898 }, { "epoch": 27.764835164835166, "grad_norm": 1.3203125, "learning_rate": 1.2124958721678808e-05, "loss": 0.1065, "step": 37899 }, { "epoch": 27.765567765567766, "grad_norm": 1.578125, "learning_rate": 1.2122794551091767e-05, "loss": 0.1552, "step": 37900 }, { "epoch": 27.766300366300367, "grad_norm": 0.90234375, "learning_rate": 1.212063056438252e-05, "loss": 0.1201, "step": 37901 }, { "epoch": 27.767032967032968, "grad_norm": 1.0, "learning_rate": 1.2118466761561429e-05, "loss": 0.1, "step": 37902 }, { "epoch": 27.76776556776557, "grad_norm": 1.0234375, "learning_rate": 1.2116303142638876e-05, "loss": 0.1026, "step": 37903 }, { "epoch": 27.76849816849817, "grad_norm": 0.74609375, "learning_rate": 1.2114139707625203e-05, "loss": 0.1251, "step": 37904 }, { "epoch": 27.76923076923077, "grad_norm": 0.984375, "learning_rate": 1.21119764565308e-05, "loss": 0.1168, "step": 37905 }, { "epoch": 27.76996336996337, "grad_norm": 1.1875, "learning_rate": 1.210981338936599e-05, "loss": 0.1374, "step": 37906 }, { "epoch": 27.77069597069597, "grad_norm": 1.1328125, "learning_rate": 1.21076505061412e-05, "loss": 0.1199, "step": 37907 }, { "epoch": 27.771428571428572, "grad_norm": 0.66796875, "learning_rate": 1.2105487806866742e-05, "loss": 0.1069, "step": 37908 }, { "epoch": 27.772161172161173, "grad_norm": 0.75390625, "learning_rate": 1.2103325291553014e-05, "loss": 0.1111, "step": 37909 }, { "epoch": 27.772893772893774, "grad_norm": 1.234375, "learning_rate": 1.2101162960210342e-05, "loss": 0.1387, "step": 37910 }, { "epoch": 27.773626373626374, "grad_norm": 1.359375, "learning_rate": 1.2099000812849105e-05, "loss": 0.0965, "step": 37911 }, { "epoch": 27.774358974358975, "grad_norm": 1.09375, "learning_rate": 1.2096838849479663e-05, "loss": 0.1159, "step": 37912 }, { "epoch": 27.775091575091576, "grad_norm": 1.1328125, "learning_rate": 1.2094677070112366e-05, "loss": 0.113, "step": 37913 }, { "epoch": 27.775824175824177, "grad_norm": 1.0390625, "learning_rate": 1.2092515474757597e-05, "loss": 0.1312, "step": 37914 }, { "epoch": 27.776556776556777, "grad_norm": 1.0625, "learning_rate": 1.2090354063425684e-05, "loss": 0.114, "step": 37915 }, { "epoch": 27.777289377289378, "grad_norm": 0.60546875, "learning_rate": 1.2088192836126994e-05, "loss": 0.1271, "step": 37916 }, { "epoch": 27.77802197802198, "grad_norm": 1.09375, "learning_rate": 1.2086031792871877e-05, "loss": 0.124, "step": 37917 }, { "epoch": 27.77875457875458, "grad_norm": 0.875, "learning_rate": 1.208387093367071e-05, "loss": 0.1228, "step": 37918 }, { "epoch": 27.77948717948718, "grad_norm": 0.84765625, "learning_rate": 1.208171025853381e-05, "loss": 0.1144, "step": 37919 }, { "epoch": 27.78021978021978, "grad_norm": 0.953125, "learning_rate": 1.2079549767471562e-05, "loss": 0.123, "step": 37920 }, { "epoch": 27.78095238095238, "grad_norm": 0.9921875, "learning_rate": 1.207738946049428e-05, "loss": 0.1449, "step": 37921 }, { "epoch": 27.781684981684982, "grad_norm": 1.0859375, "learning_rate": 1.2075229337612368e-05, "loss": 0.1232, "step": 37922 }, { "epoch": 27.782417582417583, "grad_norm": 0.890625, "learning_rate": 1.2073069398836127e-05, "loss": 0.1131, "step": 37923 }, { "epoch": 27.783150183150184, "grad_norm": 1.1015625, "learning_rate": 1.2070909644175942e-05, "loss": 0.1253, "step": 37924 }, { "epoch": 27.783882783882785, "grad_norm": 0.7265625, "learning_rate": 1.206875007364213e-05, "loss": 0.1457, "step": 37925 }, { "epoch": 27.784615384615385, "grad_norm": 0.984375, "learning_rate": 1.2066590687245049e-05, "loss": 0.1241, "step": 37926 }, { "epoch": 27.785347985347986, "grad_norm": 1.46875, "learning_rate": 1.2064431484995056e-05, "loss": 0.0857, "step": 37927 }, { "epoch": 27.786080586080587, "grad_norm": 1.4765625, "learning_rate": 1.2062272466902482e-05, "loss": 0.1411, "step": 37928 }, { "epoch": 27.786813186813188, "grad_norm": 1.2421875, "learning_rate": 1.2060113632977698e-05, "loss": 0.0981, "step": 37929 }, { "epoch": 27.78754578754579, "grad_norm": 0.6640625, "learning_rate": 1.2057954983231013e-05, "loss": 0.1115, "step": 37930 }, { "epoch": 27.78827838827839, "grad_norm": 0.9921875, "learning_rate": 1.20557965176728e-05, "loss": 0.1043, "step": 37931 }, { "epoch": 27.78901098901099, "grad_norm": 1.2265625, "learning_rate": 1.2053638236313356e-05, "loss": 0.1209, "step": 37932 }, { "epoch": 27.78974358974359, "grad_norm": 0.94921875, "learning_rate": 1.2051480139163082e-05, "loss": 0.1046, "step": 37933 }, { "epoch": 27.79047619047619, "grad_norm": 0.890625, "learning_rate": 1.2049322226232273e-05, "loss": 0.1217, "step": 37934 }, { "epoch": 27.791208791208792, "grad_norm": 1.9375, "learning_rate": 1.20471644975313e-05, "loss": 0.1309, "step": 37935 }, { "epoch": 27.791941391941393, "grad_norm": 0.96484375, "learning_rate": 1.2045006953070473e-05, "loss": 0.1244, "step": 37936 }, { "epoch": 27.792673992673993, "grad_norm": 0.796875, "learning_rate": 1.2042849592860139e-05, "loss": 0.146, "step": 37937 }, { "epoch": 27.793406593406594, "grad_norm": 1.578125, "learning_rate": 1.2040692416910634e-05, "loss": 0.1348, "step": 37938 }, { "epoch": 27.794139194139195, "grad_norm": 0.8984375, "learning_rate": 1.2038535425232304e-05, "loss": 0.1324, "step": 37939 }, { "epoch": 27.794871794871796, "grad_norm": 0.703125, "learning_rate": 1.2036378617835485e-05, "loss": 0.1197, "step": 37940 }, { "epoch": 27.795604395604396, "grad_norm": 1.3046875, "learning_rate": 1.203422199473049e-05, "loss": 0.1292, "step": 37941 }, { "epoch": 27.796336996336997, "grad_norm": 1.5, "learning_rate": 1.2032065555927665e-05, "loss": 0.1383, "step": 37942 }, { "epoch": 27.797069597069598, "grad_norm": 0.66796875, "learning_rate": 1.2029909301437345e-05, "loss": 0.1169, "step": 37943 }, { "epoch": 27.7978021978022, "grad_norm": 0.80078125, "learning_rate": 1.2027753231269868e-05, "loss": 0.1324, "step": 37944 }, { "epoch": 27.7985347985348, "grad_norm": 1.125, "learning_rate": 1.2025597345435546e-05, "loss": 0.1421, "step": 37945 }, { "epoch": 27.7992673992674, "grad_norm": 1.171875, "learning_rate": 1.2023441643944728e-05, "loss": 0.1305, "step": 37946 }, { "epoch": 27.8, "grad_norm": 0.74609375, "learning_rate": 1.2021286126807703e-05, "loss": 0.1203, "step": 37947 }, { "epoch": 27.8007326007326, "grad_norm": 1.03125, "learning_rate": 1.201913079403486e-05, "loss": 0.1108, "step": 37948 }, { "epoch": 27.801465201465202, "grad_norm": 0.9765625, "learning_rate": 1.2016975645636477e-05, "loss": 0.0982, "step": 37949 }, { "epoch": 27.802197802197803, "grad_norm": 0.5546875, "learning_rate": 1.2014820681622909e-05, "loss": 0.1108, "step": 37950 }, { "epoch": 27.802930402930404, "grad_norm": 1.1953125, "learning_rate": 1.2012665902004456e-05, "loss": 0.1519, "step": 37951 }, { "epoch": 27.803663003663004, "grad_norm": 1.1953125, "learning_rate": 1.2010511306791457e-05, "loss": 0.1371, "step": 37952 }, { "epoch": 27.804395604395605, "grad_norm": 1.109375, "learning_rate": 1.2008356895994236e-05, "loss": 0.1369, "step": 37953 }, { "epoch": 27.805128205128206, "grad_norm": 0.79296875, "learning_rate": 1.2006202669623112e-05, "loss": 0.1293, "step": 37954 }, { "epoch": 27.805860805860807, "grad_norm": 0.7265625, "learning_rate": 1.2004048627688418e-05, "loss": 0.1065, "step": 37955 }, { "epoch": 27.806593406593407, "grad_norm": 1.453125, "learning_rate": 1.2001894770200455e-05, "loss": 0.1528, "step": 37956 }, { "epoch": 27.807326007326008, "grad_norm": 0.9296875, "learning_rate": 1.1999741097169558e-05, "loss": 0.1037, "step": 37957 }, { "epoch": 27.80805860805861, "grad_norm": 0.87890625, "learning_rate": 1.1997587608606023e-05, "loss": 0.0981, "step": 37958 }, { "epoch": 27.80879120879121, "grad_norm": 1.578125, "learning_rate": 1.1995434304520208e-05, "loss": 0.1114, "step": 37959 }, { "epoch": 27.80952380952381, "grad_norm": 0.9609375, "learning_rate": 1.1993281184922389e-05, "loss": 0.1284, "step": 37960 }, { "epoch": 27.81025641025641, "grad_norm": 0.8984375, "learning_rate": 1.1991128249822923e-05, "loss": 0.1062, "step": 37961 }, { "epoch": 27.81098901098901, "grad_norm": 1.3515625, "learning_rate": 1.1988975499232081e-05, "loss": 0.1046, "step": 37962 }, { "epoch": 27.811721611721612, "grad_norm": 1.0859375, "learning_rate": 1.1986822933160206e-05, "loss": 0.1049, "step": 37963 }, { "epoch": 27.812454212454213, "grad_norm": 1.3515625, "learning_rate": 1.1984670551617605e-05, "loss": 0.1533, "step": 37964 }, { "epoch": 27.813186813186814, "grad_norm": 1.078125, "learning_rate": 1.1982518354614587e-05, "loss": 0.14, "step": 37965 }, { "epoch": 27.813919413919415, "grad_norm": 1.0625, "learning_rate": 1.1980366342161487e-05, "loss": 0.1525, "step": 37966 }, { "epoch": 27.814652014652015, "grad_norm": 1.9453125, "learning_rate": 1.1978214514268578e-05, "loss": 0.1319, "step": 37967 }, { "epoch": 27.815384615384616, "grad_norm": 0.828125, "learning_rate": 1.1976062870946193e-05, "loss": 0.1496, "step": 37968 }, { "epoch": 27.816117216117217, "grad_norm": 1.0078125, "learning_rate": 1.1973911412204639e-05, "loss": 0.1096, "step": 37969 }, { "epoch": 27.816849816849818, "grad_norm": 1.0234375, "learning_rate": 1.1971760138054232e-05, "loss": 0.1124, "step": 37970 }, { "epoch": 27.81758241758242, "grad_norm": 0.73046875, "learning_rate": 1.1969609048505256e-05, "loss": 0.112, "step": 37971 }, { "epoch": 27.81831501831502, "grad_norm": 1.6171875, "learning_rate": 1.1967458143568051e-05, "loss": 0.1466, "step": 37972 }, { "epoch": 27.81904761904762, "grad_norm": 0.97265625, "learning_rate": 1.196530742325287e-05, "loss": 0.1097, "step": 37973 }, { "epoch": 27.81978021978022, "grad_norm": 1.3203125, "learning_rate": 1.1963156887570081e-05, "loss": 0.1101, "step": 37974 }, { "epoch": 27.82051282051282, "grad_norm": 1.0625, "learning_rate": 1.1961006536529946e-05, "loss": 0.1125, "step": 37975 }, { "epoch": 27.821245421245422, "grad_norm": 0.62890625, "learning_rate": 1.1958856370142793e-05, "loss": 0.1181, "step": 37976 }, { "epoch": 27.821978021978023, "grad_norm": 1.25, "learning_rate": 1.1956706388418896e-05, "loss": 0.1219, "step": 37977 }, { "epoch": 27.822710622710623, "grad_norm": 0.546875, "learning_rate": 1.1954556591368572e-05, "loss": 0.1313, "step": 37978 }, { "epoch": 27.823443223443224, "grad_norm": 1.2265625, "learning_rate": 1.1952406979002126e-05, "loss": 0.1651, "step": 37979 }, { "epoch": 27.824175824175825, "grad_norm": 0.9453125, "learning_rate": 1.1950257551329848e-05, "loss": 0.1201, "step": 37980 }, { "epoch": 27.824908424908426, "grad_norm": 0.7421875, "learning_rate": 1.1948108308362057e-05, "loss": 0.1077, "step": 37981 }, { "epoch": 27.825641025641026, "grad_norm": 1.0859375, "learning_rate": 1.194595925010902e-05, "loss": 0.1119, "step": 37982 }, { "epoch": 27.826373626373627, "grad_norm": 0.8515625, "learning_rate": 1.1943810376581061e-05, "loss": 0.1153, "step": 37983 }, { "epoch": 27.827106227106228, "grad_norm": 0.828125, "learning_rate": 1.1941661687788435e-05, "loss": 0.1169, "step": 37984 }, { "epoch": 27.82783882783883, "grad_norm": 0.97265625, "learning_rate": 1.1939513183741494e-05, "loss": 0.1037, "step": 37985 }, { "epoch": 27.82857142857143, "grad_norm": 0.6484375, "learning_rate": 1.1937364864450494e-05, "loss": 0.105, "step": 37986 }, { "epoch": 27.82930402930403, "grad_norm": 1.203125, "learning_rate": 1.193521672992575e-05, "loss": 0.0941, "step": 37987 }, { "epoch": 27.83003663003663, "grad_norm": 1.3203125, "learning_rate": 1.1933068780177528e-05, "loss": 0.1207, "step": 37988 }, { "epoch": 27.83076923076923, "grad_norm": 1.140625, "learning_rate": 1.1930921015216136e-05, "loss": 0.1301, "step": 37989 }, { "epoch": 27.831501831501832, "grad_norm": 1.125, "learning_rate": 1.1928773435051859e-05, "loss": 0.1642, "step": 37990 }, { "epoch": 27.832234432234433, "grad_norm": 1.3046875, "learning_rate": 1.1926626039694993e-05, "loss": 0.1279, "step": 37991 }, { "epoch": 27.832967032967034, "grad_norm": 0.93359375, "learning_rate": 1.1924478829155838e-05, "loss": 0.1093, "step": 37992 }, { "epoch": 27.833699633699634, "grad_norm": 1.21875, "learning_rate": 1.1922331803444656e-05, "loss": 0.1341, "step": 37993 }, { "epoch": 27.834432234432235, "grad_norm": 0.83984375, "learning_rate": 1.1920184962571742e-05, "loss": 0.1252, "step": 37994 }, { "epoch": 27.835164835164836, "grad_norm": 1.2890625, "learning_rate": 1.1918038306547394e-05, "loss": 0.1174, "step": 37995 }, { "epoch": 27.835897435897436, "grad_norm": 0.796875, "learning_rate": 1.1915891835381902e-05, "loss": 0.1081, "step": 37996 }, { "epoch": 27.836630036630037, "grad_norm": 1.5703125, "learning_rate": 1.191374554908552e-05, "loss": 0.1336, "step": 37997 }, { "epoch": 27.837362637362638, "grad_norm": 0.8046875, "learning_rate": 1.1911599447668566e-05, "loss": 0.1425, "step": 37998 }, { "epoch": 27.83809523809524, "grad_norm": 0.91015625, "learning_rate": 1.1909453531141283e-05, "loss": 0.1216, "step": 37999 }, { "epoch": 27.83882783882784, "grad_norm": 0.9296875, "learning_rate": 1.1907307799513997e-05, "loss": 0.1207, "step": 38000 }, { "epoch": 27.83956043956044, "grad_norm": 1.78125, "learning_rate": 1.190516225279696e-05, "loss": 0.1188, "step": 38001 }, { "epoch": 27.84029304029304, "grad_norm": 0.75390625, "learning_rate": 1.190301689100047e-05, "loss": 0.1163, "step": 38002 }, { "epoch": 27.84102564102564, "grad_norm": 1.203125, "learning_rate": 1.1900871714134791e-05, "loss": 0.0971, "step": 38003 }, { "epoch": 27.841758241758242, "grad_norm": 0.72265625, "learning_rate": 1.1898726722210199e-05, "loss": 0.1232, "step": 38004 }, { "epoch": 27.842490842490843, "grad_norm": 1.359375, "learning_rate": 1.1896581915236982e-05, "loss": 0.1509, "step": 38005 }, { "epoch": 27.843223443223444, "grad_norm": 1.1328125, "learning_rate": 1.1894437293225415e-05, "loss": 0.0952, "step": 38006 }, { "epoch": 27.843956043956045, "grad_norm": 1.1328125, "learning_rate": 1.1892292856185785e-05, "loss": 0.1247, "step": 38007 }, { "epoch": 27.844688644688645, "grad_norm": 1.1875, "learning_rate": 1.1890148604128338e-05, "loss": 0.1178, "step": 38008 }, { "epoch": 27.845421245421246, "grad_norm": 0.9921875, "learning_rate": 1.188800453706338e-05, "loss": 0.1009, "step": 38009 }, { "epoch": 27.846153846153847, "grad_norm": 0.98828125, "learning_rate": 1.188586065500114e-05, "loss": 0.1494, "step": 38010 }, { "epoch": 27.846886446886447, "grad_norm": 0.9765625, "learning_rate": 1.1883716957951949e-05, "loss": 0.1382, "step": 38011 }, { "epoch": 27.847619047619048, "grad_norm": 1.0078125, "learning_rate": 1.1881573445926026e-05, "loss": 0.1156, "step": 38012 }, { "epoch": 27.84835164835165, "grad_norm": 1.21875, "learning_rate": 1.187943011893368e-05, "loss": 0.1005, "step": 38013 }, { "epoch": 27.84908424908425, "grad_norm": 1.265625, "learning_rate": 1.1877286976985148e-05, "loss": 0.1253, "step": 38014 }, { "epoch": 27.84981684981685, "grad_norm": 1.2109375, "learning_rate": 1.1875144020090713e-05, "loss": 0.1195, "step": 38015 }, { "epoch": 27.85054945054945, "grad_norm": 1.515625, "learning_rate": 1.1873001248260643e-05, "loss": 0.1171, "step": 38016 }, { "epoch": 27.851282051282052, "grad_norm": 0.6328125, "learning_rate": 1.1870858661505205e-05, "loss": 0.1115, "step": 38017 }, { "epoch": 27.852014652014653, "grad_norm": 0.75, "learning_rate": 1.186871625983468e-05, "loss": 0.1071, "step": 38018 }, { "epoch": 27.852747252747253, "grad_norm": 0.71484375, "learning_rate": 1.1866574043259299e-05, "loss": 0.0944, "step": 38019 }, { "epoch": 27.853479853479854, "grad_norm": 0.96484375, "learning_rate": 1.1864432011789345e-05, "loss": 0.1105, "step": 38020 }, { "epoch": 27.854212454212455, "grad_norm": 0.84375, "learning_rate": 1.1862290165435082e-05, "loss": 0.1113, "step": 38021 }, { "epoch": 27.854945054945055, "grad_norm": 0.6796875, "learning_rate": 1.1860148504206784e-05, "loss": 0.0882, "step": 38022 }, { "epoch": 27.855677655677656, "grad_norm": 1.5390625, "learning_rate": 1.1858007028114687e-05, "loss": 0.1154, "step": 38023 }, { "epoch": 27.856410256410257, "grad_norm": 1.2265625, "learning_rate": 1.1855865737169073e-05, "loss": 0.1255, "step": 38024 }, { "epoch": 27.857142857142858, "grad_norm": 1.15625, "learning_rate": 1.185372463138017e-05, "loss": 0.1522, "step": 38025 }, { "epoch": 27.85787545787546, "grad_norm": 0.98046875, "learning_rate": 1.1851583710758286e-05, "loss": 0.1007, "step": 38026 }, { "epoch": 27.85860805860806, "grad_norm": 1.1015625, "learning_rate": 1.1849442975313633e-05, "loss": 0.1099, "step": 38027 }, { "epoch": 27.85934065934066, "grad_norm": 1.359375, "learning_rate": 1.1847302425056491e-05, "loss": 0.1339, "step": 38028 }, { "epoch": 27.86007326007326, "grad_norm": 1.1640625, "learning_rate": 1.1845162059997127e-05, "loss": 0.0954, "step": 38029 }, { "epoch": 27.86080586080586, "grad_norm": 0.98046875, "learning_rate": 1.1843021880145765e-05, "loss": 0.0958, "step": 38030 }, { "epoch": 27.861538461538462, "grad_norm": 1.0078125, "learning_rate": 1.1840881885512677e-05, "loss": 0.1108, "step": 38031 }, { "epoch": 27.862271062271063, "grad_norm": 1.125, "learning_rate": 1.1838742076108108e-05, "loss": 0.1439, "step": 38032 }, { "epoch": 27.863003663003663, "grad_norm": 1.3359375, "learning_rate": 1.1836602451942342e-05, "loss": 0.1542, "step": 38033 }, { "epoch": 27.863736263736264, "grad_norm": 1.1875, "learning_rate": 1.1834463013025585e-05, "loss": 0.112, "step": 38034 }, { "epoch": 27.864468864468865, "grad_norm": 0.63671875, "learning_rate": 1.1832323759368122e-05, "loss": 0.1407, "step": 38035 }, { "epoch": 27.865201465201466, "grad_norm": 0.8671875, "learning_rate": 1.1830184690980164e-05, "loss": 0.1293, "step": 38036 }, { "epoch": 27.865934065934066, "grad_norm": 1.3125, "learning_rate": 1.1828045807872016e-05, "loss": 0.1274, "step": 38037 }, { "epoch": 27.866666666666667, "grad_norm": 0.85546875, "learning_rate": 1.1825907110053883e-05, "loss": 0.1075, "step": 38038 }, { "epoch": 27.867399267399268, "grad_norm": 0.734375, "learning_rate": 1.1823768597536034e-05, "loss": 0.1124, "step": 38039 }, { "epoch": 27.86813186813187, "grad_norm": 0.84375, "learning_rate": 1.1821630270328692e-05, "loss": 0.103, "step": 38040 }, { "epoch": 27.86886446886447, "grad_norm": 0.7421875, "learning_rate": 1.181949212844212e-05, "loss": 0.1025, "step": 38041 }, { "epoch": 27.86959706959707, "grad_norm": 0.9609375, "learning_rate": 1.1817354171886558e-05, "loss": 0.1046, "step": 38042 }, { "epoch": 27.87032967032967, "grad_norm": 1.2890625, "learning_rate": 1.1815216400672251e-05, "loss": 0.1268, "step": 38043 }, { "epoch": 27.87106227106227, "grad_norm": 0.75, "learning_rate": 1.1813078814809454e-05, "loss": 0.1115, "step": 38044 }, { "epoch": 27.871794871794872, "grad_norm": 0.984375, "learning_rate": 1.1810941414308383e-05, "loss": 0.0934, "step": 38045 }, { "epoch": 27.872527472527473, "grad_norm": 0.8046875, "learning_rate": 1.1808804199179295e-05, "loss": 0.1364, "step": 38046 }, { "epoch": 27.873260073260074, "grad_norm": 0.7578125, "learning_rate": 1.1806667169432427e-05, "loss": 0.12, "step": 38047 }, { "epoch": 27.873992673992674, "grad_norm": 1.875, "learning_rate": 1.1804530325078027e-05, "loss": 0.158, "step": 38048 }, { "epoch": 27.874725274725275, "grad_norm": 1.2578125, "learning_rate": 1.1802393666126317e-05, "loss": 0.1327, "step": 38049 }, { "epoch": 27.875457875457876, "grad_norm": 1.078125, "learning_rate": 1.1800257192587552e-05, "loss": 0.107, "step": 38050 }, { "epoch": 27.876190476190477, "grad_norm": 1.046875, "learning_rate": 1.179812090447194e-05, "loss": 0.1157, "step": 38051 }, { "epoch": 27.876923076923077, "grad_norm": 0.953125, "learning_rate": 1.1795984801789755e-05, "loss": 0.1332, "step": 38052 }, { "epoch": 27.877655677655678, "grad_norm": 1.1484375, "learning_rate": 1.1793848884551201e-05, "loss": 0.1248, "step": 38053 }, { "epoch": 27.87838827838828, "grad_norm": 2.078125, "learning_rate": 1.1791713152766525e-05, "loss": 0.1624, "step": 38054 }, { "epoch": 27.87912087912088, "grad_norm": 1.125, "learning_rate": 1.178957760644597e-05, "loss": 0.1425, "step": 38055 }, { "epoch": 27.87985347985348, "grad_norm": 1.1796875, "learning_rate": 1.1787442245599746e-05, "loss": 0.1044, "step": 38056 }, { "epoch": 27.88058608058608, "grad_norm": 0.6875, "learning_rate": 1.1785307070238093e-05, "loss": 0.1009, "step": 38057 }, { "epoch": 27.88131868131868, "grad_norm": 1.2890625, "learning_rate": 1.178317208037124e-05, "loss": 0.1175, "step": 38058 }, { "epoch": 27.882051282051282, "grad_norm": 0.69921875, "learning_rate": 1.1781037276009436e-05, "loss": 0.1036, "step": 38059 }, { "epoch": 27.882783882783883, "grad_norm": 0.98046875, "learning_rate": 1.1778902657162876e-05, "loss": 0.1205, "step": 38060 }, { "epoch": 27.883516483516484, "grad_norm": 0.94140625, "learning_rate": 1.1776768223841824e-05, "loss": 0.1181, "step": 38061 }, { "epoch": 27.884249084249085, "grad_norm": 1.3515625, "learning_rate": 1.1774633976056456e-05, "loss": 0.1125, "step": 38062 }, { "epoch": 27.884981684981685, "grad_norm": 1.484375, "learning_rate": 1.1772499913817059e-05, "loss": 0.1252, "step": 38063 }, { "epoch": 27.885714285714286, "grad_norm": 1.2734375, "learning_rate": 1.1770366037133811e-05, "loss": 0.1603, "step": 38064 }, { "epoch": 27.886446886446887, "grad_norm": 1.0703125, "learning_rate": 1.176823234601697e-05, "loss": 0.1146, "step": 38065 }, { "epoch": 27.887179487179488, "grad_norm": 1.1328125, "learning_rate": 1.1766098840476733e-05, "loss": 0.1156, "step": 38066 }, { "epoch": 27.88791208791209, "grad_norm": 0.93359375, "learning_rate": 1.1763965520523327e-05, "loss": 0.1106, "step": 38067 }, { "epoch": 27.88864468864469, "grad_norm": 1.7421875, "learning_rate": 1.1761832386166977e-05, "loss": 0.1137, "step": 38068 }, { "epoch": 27.88937728937729, "grad_norm": 1.0, "learning_rate": 1.1759699437417906e-05, "loss": 0.1248, "step": 38069 }, { "epoch": 27.89010989010989, "grad_norm": 1.4296875, "learning_rate": 1.1757566674286349e-05, "loss": 0.1132, "step": 38070 }, { "epoch": 27.89084249084249, "grad_norm": 1.015625, "learning_rate": 1.1755434096782495e-05, "loss": 0.1145, "step": 38071 }, { "epoch": 27.891575091575092, "grad_norm": 0.8828125, "learning_rate": 1.1753301704916573e-05, "loss": 0.0923, "step": 38072 }, { "epoch": 27.892307692307693, "grad_norm": 1.0390625, "learning_rate": 1.17511694986988e-05, "loss": 0.0934, "step": 38073 }, { "epoch": 27.893040293040293, "grad_norm": 1.2421875, "learning_rate": 1.1749037478139413e-05, "loss": 0.1511, "step": 38074 }, { "epoch": 27.893772893772894, "grad_norm": 1.125, "learning_rate": 1.1746905643248594e-05, "loss": 0.1344, "step": 38075 }, { "epoch": 27.894505494505495, "grad_norm": 2.46875, "learning_rate": 1.1744773994036582e-05, "loss": 0.1898, "step": 38076 }, { "epoch": 27.895238095238096, "grad_norm": 1.1328125, "learning_rate": 1.1742642530513557e-05, "loss": 0.1483, "step": 38077 }, { "epoch": 27.895970695970696, "grad_norm": 1.125, "learning_rate": 1.1740511252689778e-05, "loss": 0.1159, "step": 38078 }, { "epoch": 27.896703296703297, "grad_norm": 0.89453125, "learning_rate": 1.1738380160575422e-05, "loss": 0.1136, "step": 38079 }, { "epoch": 27.897435897435898, "grad_norm": 1.078125, "learning_rate": 1.173624925418071e-05, "loss": 0.1442, "step": 38080 }, { "epoch": 27.8981684981685, "grad_norm": 1.03125, "learning_rate": 1.173411853351587e-05, "loss": 0.1107, "step": 38081 }, { "epoch": 27.8989010989011, "grad_norm": 1.1015625, "learning_rate": 1.1731987998591077e-05, "loss": 0.1144, "step": 38082 }, { "epoch": 27.8996336996337, "grad_norm": 0.984375, "learning_rate": 1.1729857649416557e-05, "loss": 0.1254, "step": 38083 }, { "epoch": 27.9003663003663, "grad_norm": 1.1015625, "learning_rate": 1.172772748600252e-05, "loss": 0.1256, "step": 38084 }, { "epoch": 27.9010989010989, "grad_norm": 1.0546875, "learning_rate": 1.1725597508359178e-05, "loss": 0.1129, "step": 38085 }, { "epoch": 27.901831501831502, "grad_norm": 0.8203125, "learning_rate": 1.1723467716496715e-05, "loss": 0.1503, "step": 38086 }, { "epoch": 27.902564102564103, "grad_norm": 0.7421875, "learning_rate": 1.1721338110425362e-05, "loss": 0.1106, "step": 38087 }, { "epoch": 27.903296703296704, "grad_norm": 1.3359375, "learning_rate": 1.1719208690155283e-05, "loss": 0.1198, "step": 38088 }, { "epoch": 27.904029304029304, "grad_norm": 1.0625, "learning_rate": 1.1717079455696728e-05, "loss": 0.1051, "step": 38089 }, { "epoch": 27.904761904761905, "grad_norm": 0.765625, "learning_rate": 1.1714950407059867e-05, "loss": 0.1174, "step": 38090 }, { "epoch": 27.905494505494506, "grad_norm": 0.91015625, "learning_rate": 1.171282154425492e-05, "loss": 0.162, "step": 38091 }, { "epoch": 27.906227106227107, "grad_norm": 0.921875, "learning_rate": 1.1710692867292069e-05, "loss": 0.128, "step": 38092 }, { "epoch": 27.906959706959707, "grad_norm": 1.7890625, "learning_rate": 1.170856437618152e-05, "loss": 0.1449, "step": 38093 }, { "epoch": 27.907692307692308, "grad_norm": 0.984375, "learning_rate": 1.1706436070933474e-05, "loss": 0.1206, "step": 38094 }, { "epoch": 27.90842490842491, "grad_norm": 0.83984375, "learning_rate": 1.170430795155813e-05, "loss": 0.1183, "step": 38095 }, { "epoch": 27.90915750915751, "grad_norm": 1.15625, "learning_rate": 1.1702180018065695e-05, "loss": 0.112, "step": 38096 }, { "epoch": 27.90989010989011, "grad_norm": 1.21875, "learning_rate": 1.170005227046634e-05, "loss": 0.138, "step": 38097 }, { "epoch": 27.91062271062271, "grad_norm": 0.8515625, "learning_rate": 1.1697924708770268e-05, "loss": 0.1456, "step": 38098 }, { "epoch": 27.91135531135531, "grad_norm": 0.75, "learning_rate": 1.1695797332987678e-05, "loss": 0.1283, "step": 38099 }, { "epoch": 27.912087912087912, "grad_norm": 0.6875, "learning_rate": 1.1693670143128777e-05, "loss": 0.1183, "step": 38100 }, { "epoch": 27.912820512820513, "grad_norm": 0.71875, "learning_rate": 1.1691543139203723e-05, "loss": 0.1362, "step": 38101 }, { "epoch": 27.913553113553114, "grad_norm": 0.890625, "learning_rate": 1.1689416321222741e-05, "loss": 0.1168, "step": 38102 }, { "epoch": 27.914285714285715, "grad_norm": 1.046875, "learning_rate": 1.1687289689195985e-05, "loss": 0.1052, "step": 38103 }, { "epoch": 27.915018315018315, "grad_norm": 1.2734375, "learning_rate": 1.1685163243133686e-05, "loss": 0.1442, "step": 38104 }, { "epoch": 27.915750915750916, "grad_norm": 0.7734375, "learning_rate": 1.1683036983045998e-05, "loss": 0.1056, "step": 38105 }, { "epoch": 27.916483516483517, "grad_norm": 1.4296875, "learning_rate": 1.168091090894313e-05, "loss": 0.1612, "step": 38106 }, { "epoch": 27.917216117216118, "grad_norm": 1.0546875, "learning_rate": 1.1678785020835267e-05, "loss": 0.1119, "step": 38107 }, { "epoch": 27.91794871794872, "grad_norm": 1.0390625, "learning_rate": 1.1676659318732578e-05, "loss": 0.1322, "step": 38108 }, { "epoch": 27.91868131868132, "grad_norm": 1.0703125, "learning_rate": 1.1674533802645265e-05, "loss": 0.142, "step": 38109 }, { "epoch": 27.91941391941392, "grad_norm": 1.53125, "learning_rate": 1.1672408472583495e-05, "loss": 0.1338, "step": 38110 }, { "epoch": 27.92014652014652, "grad_norm": 0.78515625, "learning_rate": 1.1670283328557483e-05, "loss": 0.116, "step": 38111 }, { "epoch": 27.92087912087912, "grad_norm": 0.91796875, "learning_rate": 1.1668158370577378e-05, "loss": 0.1234, "step": 38112 }, { "epoch": 27.921611721611722, "grad_norm": 1.2265625, "learning_rate": 1.1666033598653385e-05, "loss": 0.1262, "step": 38113 }, { "epoch": 27.922344322344323, "grad_norm": 1.34375, "learning_rate": 1.1663909012795644e-05, "loss": 0.1669, "step": 38114 }, { "epoch": 27.923076923076923, "grad_norm": 1.34375, "learning_rate": 1.1661784613014395e-05, "loss": 0.1122, "step": 38115 }, { "epoch": 27.923809523809524, "grad_norm": 1.1953125, "learning_rate": 1.165966039931977e-05, "loss": 0.1177, "step": 38116 }, { "epoch": 27.924542124542125, "grad_norm": 2.03125, "learning_rate": 1.1657536371721978e-05, "loss": 0.1553, "step": 38117 }, { "epoch": 27.925274725274726, "grad_norm": 0.703125, "learning_rate": 1.1655412530231166e-05, "loss": 0.1129, "step": 38118 }, { "epoch": 27.926007326007326, "grad_norm": 1.296875, "learning_rate": 1.1653288874857523e-05, "loss": 0.1452, "step": 38119 }, { "epoch": 27.926739926739927, "grad_norm": 1.0390625, "learning_rate": 1.1651165405611228e-05, "loss": 0.1081, "step": 38120 }, { "epoch": 27.927472527472528, "grad_norm": 0.8359375, "learning_rate": 1.164904212250245e-05, "loss": 0.144, "step": 38121 }, { "epoch": 27.92820512820513, "grad_norm": 1.03125, "learning_rate": 1.1646919025541381e-05, "loss": 0.1727, "step": 38122 }, { "epoch": 27.92893772893773, "grad_norm": 1.4609375, "learning_rate": 1.1644796114738165e-05, "loss": 0.1238, "step": 38123 }, { "epoch": 27.92967032967033, "grad_norm": 0.6796875, "learning_rate": 1.1642673390102983e-05, "loss": 0.1175, "step": 38124 }, { "epoch": 27.93040293040293, "grad_norm": 0.8046875, "learning_rate": 1.1640550851646005e-05, "loss": 0.0846, "step": 38125 }, { "epoch": 27.93113553113553, "grad_norm": 0.83984375, "learning_rate": 1.1638428499377424e-05, "loss": 0.1199, "step": 38126 }, { "epoch": 27.931868131868132, "grad_norm": 1.1171875, "learning_rate": 1.1636306333307374e-05, "loss": 0.1365, "step": 38127 }, { "epoch": 27.932600732600733, "grad_norm": 1.15625, "learning_rate": 1.1634184353446055e-05, "loss": 0.1765, "step": 38128 }, { "epoch": 27.933333333333334, "grad_norm": 0.89453125, "learning_rate": 1.163206255980359e-05, "loss": 0.1138, "step": 38129 }, { "epoch": 27.934065934065934, "grad_norm": 1.203125, "learning_rate": 1.1629940952390196e-05, "loss": 0.1435, "step": 38130 }, { "epoch": 27.934798534798535, "grad_norm": 0.94921875, "learning_rate": 1.1627819531216006e-05, "loss": 0.116, "step": 38131 }, { "epoch": 27.935531135531136, "grad_norm": 1.3359375, "learning_rate": 1.162569829629119e-05, "loss": 0.1303, "step": 38132 }, { "epoch": 27.936263736263736, "grad_norm": 0.5546875, "learning_rate": 1.162357724762593e-05, "loss": 0.0915, "step": 38133 }, { "epoch": 27.936996336996337, "grad_norm": 0.8203125, "learning_rate": 1.162145638523036e-05, "loss": 0.1229, "step": 38134 }, { "epoch": 27.937728937728938, "grad_norm": 1.453125, "learning_rate": 1.1619335709114658e-05, "loss": 0.135, "step": 38135 }, { "epoch": 27.93846153846154, "grad_norm": 1.203125, "learning_rate": 1.161721521928898e-05, "loss": 0.1421, "step": 38136 }, { "epoch": 27.93919413919414, "grad_norm": 1.34375, "learning_rate": 1.1615094915763501e-05, "loss": 0.1152, "step": 38137 }, { "epoch": 27.93992673992674, "grad_norm": 0.88671875, "learning_rate": 1.1612974798548357e-05, "loss": 0.1195, "step": 38138 }, { "epoch": 27.94065934065934, "grad_norm": 1.1015625, "learning_rate": 1.1610854867653722e-05, "loss": 0.1211, "step": 38139 }, { "epoch": 27.94139194139194, "grad_norm": 1.3203125, "learning_rate": 1.160873512308973e-05, "loss": 0.1442, "step": 38140 }, { "epoch": 27.942124542124542, "grad_norm": 1.3515625, "learning_rate": 1.1606615564866575e-05, "loss": 0.1292, "step": 38141 }, { "epoch": 27.942857142857143, "grad_norm": 1.21875, "learning_rate": 1.1604496192994385e-05, "loss": 0.1337, "step": 38142 }, { "epoch": 27.943589743589744, "grad_norm": 0.88671875, "learning_rate": 1.1602377007483322e-05, "loss": 0.1074, "step": 38143 }, { "epoch": 27.944322344322345, "grad_norm": 1.3203125, "learning_rate": 1.1600258008343552e-05, "loss": 0.1402, "step": 38144 }, { "epoch": 27.945054945054945, "grad_norm": 1.453125, "learning_rate": 1.15981391955852e-05, "loss": 0.1376, "step": 38145 }, { "epoch": 27.945787545787546, "grad_norm": 1.03125, "learning_rate": 1.1596020569218435e-05, "loss": 0.1046, "step": 38146 }, { "epoch": 27.946520146520147, "grad_norm": 0.98046875, "learning_rate": 1.1593902129253407e-05, "loss": 0.1517, "step": 38147 }, { "epoch": 27.947252747252747, "grad_norm": 1.6796875, "learning_rate": 1.1591783875700283e-05, "loss": 0.1559, "step": 38148 }, { "epoch": 27.947985347985348, "grad_norm": 1.4140625, "learning_rate": 1.1589665808569177e-05, "loss": 0.1426, "step": 38149 }, { "epoch": 27.94871794871795, "grad_norm": 1.171875, "learning_rate": 1.1587547927870254e-05, "loss": 0.113, "step": 38150 }, { "epoch": 27.94945054945055, "grad_norm": 1.03125, "learning_rate": 1.158543023361367e-05, "loss": 0.1255, "step": 38151 }, { "epoch": 27.95018315018315, "grad_norm": 1.0390625, "learning_rate": 1.1583312725809575e-05, "loss": 0.1087, "step": 38152 }, { "epoch": 27.95091575091575, "grad_norm": 1.046875, "learning_rate": 1.158119540446809e-05, "loss": 0.0927, "step": 38153 }, { "epoch": 27.951648351648352, "grad_norm": 0.921875, "learning_rate": 1.1579078269599385e-05, "loss": 0.1348, "step": 38154 }, { "epoch": 27.952380952380953, "grad_norm": 0.703125, "learning_rate": 1.1576961321213568e-05, "loss": 0.1328, "step": 38155 }, { "epoch": 27.953113553113553, "grad_norm": 1.3828125, "learning_rate": 1.1574844559320838e-05, "loss": 0.1261, "step": 38156 }, { "epoch": 27.953846153846154, "grad_norm": 0.68359375, "learning_rate": 1.1572727983931285e-05, "loss": 0.112, "step": 38157 }, { "epoch": 27.954578754578755, "grad_norm": 1.046875, "learning_rate": 1.1570611595055075e-05, "loss": 0.0953, "step": 38158 }, { "epoch": 27.955311355311355, "grad_norm": 1.5390625, "learning_rate": 1.1568495392702361e-05, "loss": 0.1102, "step": 38159 }, { "epoch": 27.956043956043956, "grad_norm": 0.94140625, "learning_rate": 1.1566379376883247e-05, "loss": 0.1287, "step": 38160 }, { "epoch": 27.956776556776557, "grad_norm": 1.1015625, "learning_rate": 1.1564263547607896e-05, "loss": 0.0955, "step": 38161 }, { "epoch": 27.957509157509158, "grad_norm": 1.2421875, "learning_rate": 1.1562147904886433e-05, "loss": 0.1401, "step": 38162 }, { "epoch": 27.95824175824176, "grad_norm": 0.859375, "learning_rate": 1.1560032448729021e-05, "loss": 0.1161, "step": 38163 }, { "epoch": 27.95897435897436, "grad_norm": 1.171875, "learning_rate": 1.1557917179145762e-05, "loss": 0.1193, "step": 38164 }, { "epoch": 27.95970695970696, "grad_norm": 0.6953125, "learning_rate": 1.1555802096146817e-05, "loss": 0.0967, "step": 38165 }, { "epoch": 27.96043956043956, "grad_norm": 1.265625, "learning_rate": 1.1553687199742284e-05, "loss": 0.1008, "step": 38166 }, { "epoch": 27.96117216117216, "grad_norm": 0.6875, "learning_rate": 1.155157248994235e-05, "loss": 0.1141, "step": 38167 }, { "epoch": 27.961904761904762, "grad_norm": 0.890625, "learning_rate": 1.15494579667571e-05, "loss": 0.1219, "step": 38168 }, { "epoch": 27.962637362637363, "grad_norm": 1.21875, "learning_rate": 1.1547343630196687e-05, "loss": 0.1399, "step": 38169 }, { "epoch": 27.963369963369964, "grad_norm": 0.8671875, "learning_rate": 1.1545229480271247e-05, "loss": 0.129, "step": 38170 }, { "epoch": 27.964102564102564, "grad_norm": 1.1015625, "learning_rate": 1.1543115516990893e-05, "loss": 0.0985, "step": 38171 }, { "epoch": 27.964835164835165, "grad_norm": 0.85546875, "learning_rate": 1.1541001740365756e-05, "loss": 0.1152, "step": 38172 }, { "epoch": 27.965567765567766, "grad_norm": 0.84375, "learning_rate": 1.1538888150405975e-05, "loss": 0.0916, "step": 38173 }, { "epoch": 27.966300366300366, "grad_norm": 1.109375, "learning_rate": 1.1536774747121676e-05, "loss": 0.1124, "step": 38174 }, { "epoch": 27.967032967032967, "grad_norm": 0.6328125, "learning_rate": 1.1534661530522972e-05, "loss": 0.1178, "step": 38175 }, { "epoch": 27.967765567765568, "grad_norm": 1.2734375, "learning_rate": 1.1532548500619999e-05, "loss": 0.1219, "step": 38176 }, { "epoch": 27.96849816849817, "grad_norm": 1.4140625, "learning_rate": 1.153043565742287e-05, "loss": 0.1301, "step": 38177 }, { "epoch": 27.96923076923077, "grad_norm": 1.0390625, "learning_rate": 1.1528323000941737e-05, "loss": 0.1055, "step": 38178 }, { "epoch": 27.96996336996337, "grad_norm": 1.7734375, "learning_rate": 1.1526210531186688e-05, "loss": 0.1482, "step": 38179 }, { "epoch": 27.97069597069597, "grad_norm": 0.93359375, "learning_rate": 1.152409824816787e-05, "loss": 0.1265, "step": 38180 }, { "epoch": 27.97142857142857, "grad_norm": 0.73828125, "learning_rate": 1.1521986151895365e-05, "loss": 0.1215, "step": 38181 }, { "epoch": 27.972161172161172, "grad_norm": 0.9375, "learning_rate": 1.151987424237935e-05, "loss": 0.116, "step": 38182 }, { "epoch": 27.972893772893773, "grad_norm": 1.1875, "learning_rate": 1.1517762519629895e-05, "loss": 0.1751, "step": 38183 }, { "epoch": 27.973626373626374, "grad_norm": 0.65234375, "learning_rate": 1.1515650983657141e-05, "loss": 0.0883, "step": 38184 }, { "epoch": 27.974358974358974, "grad_norm": 1.0703125, "learning_rate": 1.1513539634471214e-05, "loss": 0.1382, "step": 38185 }, { "epoch": 27.975091575091575, "grad_norm": 1.203125, "learning_rate": 1.15114284720822e-05, "loss": 0.1177, "step": 38186 }, { "epoch": 27.975824175824176, "grad_norm": 1.0859375, "learning_rate": 1.1509317496500234e-05, "loss": 0.091, "step": 38187 }, { "epoch": 27.976556776556777, "grad_norm": 0.6953125, "learning_rate": 1.150720670773543e-05, "loss": 0.1283, "step": 38188 }, { "epoch": 27.977289377289377, "grad_norm": 0.9296875, "learning_rate": 1.1505096105797912e-05, "loss": 0.0883, "step": 38189 }, { "epoch": 27.978021978021978, "grad_norm": 0.8046875, "learning_rate": 1.1502985690697764e-05, "loss": 0.1244, "step": 38190 }, { "epoch": 27.97875457875458, "grad_norm": 1.1015625, "learning_rate": 1.1500875462445127e-05, "loss": 0.1382, "step": 38191 }, { "epoch": 27.97948717948718, "grad_norm": 1.7109375, "learning_rate": 1.149876542105007e-05, "loss": 0.1691, "step": 38192 }, { "epoch": 27.98021978021978, "grad_norm": 1.0625, "learning_rate": 1.1496655566522762e-05, "loss": 0.1093, "step": 38193 }, { "epoch": 27.98095238095238, "grad_norm": 0.87109375, "learning_rate": 1.1494545898873262e-05, "loss": 0.1051, "step": 38194 }, { "epoch": 27.98168498168498, "grad_norm": 1.2578125, "learning_rate": 1.1492436418111703e-05, "loss": 0.1037, "step": 38195 }, { "epoch": 27.982417582417582, "grad_norm": 0.89453125, "learning_rate": 1.1490327124248196e-05, "loss": 0.1386, "step": 38196 }, { "epoch": 27.983150183150183, "grad_norm": 1.25, "learning_rate": 1.148821801729283e-05, "loss": 0.129, "step": 38197 }, { "epoch": 27.983882783882784, "grad_norm": 0.80078125, "learning_rate": 1.1486109097255711e-05, "loss": 0.1115, "step": 38198 }, { "epoch": 27.984615384615385, "grad_norm": 0.5859375, "learning_rate": 1.1484000364146956e-05, "loss": 0.0799, "step": 38199 }, { "epoch": 27.985347985347985, "grad_norm": 1.0859375, "learning_rate": 1.1481891817976676e-05, "loss": 0.1459, "step": 38200 }, { "epoch": 27.986080586080586, "grad_norm": 0.75, "learning_rate": 1.1479783458754945e-05, "loss": 0.1399, "step": 38201 }, { "epoch": 27.986813186813187, "grad_norm": 1.0859375, "learning_rate": 1.1477675286491881e-05, "loss": 0.1522, "step": 38202 }, { "epoch": 27.987545787545788, "grad_norm": 1.7734375, "learning_rate": 1.147556730119759e-05, "loss": 0.1523, "step": 38203 }, { "epoch": 27.98827838827839, "grad_norm": 0.67578125, "learning_rate": 1.1473459502882172e-05, "loss": 0.1045, "step": 38204 }, { "epoch": 27.98901098901099, "grad_norm": 1.5, "learning_rate": 1.1471351891555714e-05, "loss": 0.1237, "step": 38205 }, { "epoch": 27.98974358974359, "grad_norm": 1.09375, "learning_rate": 1.1469244467228336e-05, "loss": 0.1295, "step": 38206 }, { "epoch": 27.99047619047619, "grad_norm": 1.1171875, "learning_rate": 1.146713722991009e-05, "loss": 0.1432, "step": 38207 }, { "epoch": 27.99120879120879, "grad_norm": 0.96484375, "learning_rate": 1.1465030179611129e-05, "loss": 0.1625, "step": 38208 }, { "epoch": 27.991941391941392, "grad_norm": 0.9296875, "learning_rate": 1.1462923316341513e-05, "loss": 0.1341, "step": 38209 }, { "epoch": 27.992673992673993, "grad_norm": 1.078125, "learning_rate": 1.1460816640111342e-05, "loss": 0.1659, "step": 38210 }, { "epoch": 27.993406593406593, "grad_norm": 1.234375, "learning_rate": 1.145871015093073e-05, "loss": 0.0961, "step": 38211 }, { "epoch": 27.994139194139194, "grad_norm": 1.21875, "learning_rate": 1.1456603848809746e-05, "loss": 0.1234, "step": 38212 }, { "epoch": 27.994871794871795, "grad_norm": 1.21875, "learning_rate": 1.1454497733758481e-05, "loss": 0.1258, "step": 38213 }, { "epoch": 27.995604395604396, "grad_norm": 0.8515625, "learning_rate": 1.1452391805787041e-05, "loss": 0.1189, "step": 38214 }, { "epoch": 27.996336996336996, "grad_norm": 0.7890625, "learning_rate": 1.1450286064905524e-05, "loss": 0.1177, "step": 38215 }, { "epoch": 27.997069597069597, "grad_norm": 1.1171875, "learning_rate": 1.1448180511123987e-05, "loss": 0.1285, "step": 38216 }, { "epoch": 27.997802197802198, "grad_norm": 0.69921875, "learning_rate": 1.1446075144452559e-05, "loss": 0.1222, "step": 38217 }, { "epoch": 27.9985347985348, "grad_norm": 0.74609375, "learning_rate": 1.144396996490127e-05, "loss": 0.1031, "step": 38218 }, { "epoch": 27.9992673992674, "grad_norm": 0.98046875, "learning_rate": 1.1441864972480273e-05, "loss": 0.128, "step": 38219 }, { "epoch": 28.0, "grad_norm": 0.734375, "learning_rate": 1.1439760167199607e-05, "loss": 0.1152, "step": 38220 }, { "epoch": 28.0007326007326, "grad_norm": 1.0703125, "learning_rate": 1.1437655549069378e-05, "loss": 0.1042, "step": 38221 }, { "epoch": 28.0014652014652, "grad_norm": 1.6328125, "learning_rate": 1.1435551118099673e-05, "loss": 0.1378, "step": 38222 }, { "epoch": 28.002197802197802, "grad_norm": 1.3359375, "learning_rate": 1.1433446874300556e-05, "loss": 0.1218, "step": 38223 }, { "epoch": 28.002930402930403, "grad_norm": 0.57421875, "learning_rate": 1.1431342817682117e-05, "loss": 0.1203, "step": 38224 }, { "epoch": 28.003663003663004, "grad_norm": 1.0546875, "learning_rate": 1.1429238948254444e-05, "loss": 0.1055, "step": 38225 }, { "epoch": 28.004395604395604, "grad_norm": 1.515625, "learning_rate": 1.1427135266027621e-05, "loss": 0.1075, "step": 38226 }, { "epoch": 28.005128205128205, "grad_norm": 0.734375, "learning_rate": 1.1425031771011711e-05, "loss": 0.1301, "step": 38227 }, { "epoch": 28.005860805860806, "grad_norm": 1.0703125, "learning_rate": 1.1422928463216801e-05, "loss": 0.119, "step": 38228 }, { "epoch": 28.006593406593407, "grad_norm": 1.0390625, "learning_rate": 1.142082534265297e-05, "loss": 0.1202, "step": 38229 }, { "epoch": 28.007326007326007, "grad_norm": 1.0546875, "learning_rate": 1.14187224093303e-05, "loss": 0.1318, "step": 38230 }, { "epoch": 28.008058608058608, "grad_norm": 0.9765625, "learning_rate": 1.1416619663258855e-05, "loss": 0.0929, "step": 38231 }, { "epoch": 28.00879120879121, "grad_norm": 1.7734375, "learning_rate": 1.1414517104448726e-05, "loss": 0.1125, "step": 38232 }, { "epoch": 28.00952380952381, "grad_norm": 1.078125, "learning_rate": 1.1412414732909946e-05, "loss": 0.1228, "step": 38233 }, { "epoch": 28.01025641025641, "grad_norm": 0.6640625, "learning_rate": 1.141031254865265e-05, "loss": 0.121, "step": 38234 }, { "epoch": 28.01098901098901, "grad_norm": 1.0390625, "learning_rate": 1.1408210551686863e-05, "loss": 0.1193, "step": 38235 }, { "epoch": 28.01172161172161, "grad_norm": 1.0078125, "learning_rate": 1.1406108742022677e-05, "loss": 0.1099, "step": 38236 }, { "epoch": 28.012454212454212, "grad_norm": 1.6015625, "learning_rate": 1.1404007119670164e-05, "loss": 0.0971, "step": 38237 }, { "epoch": 28.013186813186813, "grad_norm": 1.2109375, "learning_rate": 1.140190568463938e-05, "loss": 0.0985, "step": 38238 }, { "epoch": 28.013919413919414, "grad_norm": 1.359375, "learning_rate": 1.13998044369404e-05, "loss": 0.1196, "step": 38239 }, { "epoch": 28.014652014652015, "grad_norm": 1.625, "learning_rate": 1.139770337658329e-05, "loss": 0.1282, "step": 38240 }, { "epoch": 28.015384615384615, "grad_norm": 0.8125, "learning_rate": 1.139560250357813e-05, "loss": 0.1101, "step": 38241 }, { "epoch": 28.016117216117216, "grad_norm": 0.953125, "learning_rate": 1.1393501817934971e-05, "loss": 0.1196, "step": 38242 }, { "epoch": 28.016849816849817, "grad_norm": 1.4375, "learning_rate": 1.1391401319663888e-05, "loss": 0.1251, "step": 38243 }, { "epoch": 28.017582417582418, "grad_norm": 0.859375, "learning_rate": 1.1389301008774915e-05, "loss": 0.0976, "step": 38244 }, { "epoch": 28.01831501831502, "grad_norm": 0.84375, "learning_rate": 1.1387200885278174e-05, "loss": 0.1116, "step": 38245 }, { "epoch": 28.01904761904762, "grad_norm": 0.65625, "learning_rate": 1.1385100949183671e-05, "loss": 0.1092, "step": 38246 }, { "epoch": 28.01978021978022, "grad_norm": 0.92578125, "learning_rate": 1.1383001200501495e-05, "loss": 0.0927, "step": 38247 }, { "epoch": 28.02051282051282, "grad_norm": 1.3125, "learning_rate": 1.1380901639241712e-05, "loss": 0.1394, "step": 38248 }, { "epoch": 28.02124542124542, "grad_norm": 1.375, "learning_rate": 1.137880226541436e-05, "loss": 0.1433, "step": 38249 }, { "epoch": 28.021978021978022, "grad_norm": 1.2265625, "learning_rate": 1.1376703079029508e-05, "loss": 0.1191, "step": 38250 }, { "epoch": 28.022710622710623, "grad_norm": 1.03125, "learning_rate": 1.1374604080097211e-05, "loss": 0.1343, "step": 38251 }, { "epoch": 28.023443223443223, "grad_norm": 1.71875, "learning_rate": 1.1372505268627551e-05, "loss": 0.1142, "step": 38252 }, { "epoch": 28.024175824175824, "grad_norm": 0.84765625, "learning_rate": 1.1370406644630544e-05, "loss": 0.0994, "step": 38253 }, { "epoch": 28.024908424908425, "grad_norm": 0.8515625, "learning_rate": 1.1368308208116264e-05, "loss": 0.1176, "step": 38254 }, { "epoch": 28.025641025641026, "grad_norm": 1.125, "learning_rate": 1.1366209959094769e-05, "loss": 0.1046, "step": 38255 }, { "epoch": 28.026373626373626, "grad_norm": 1.109375, "learning_rate": 1.1364111897576122e-05, "loss": 0.1473, "step": 38256 }, { "epoch": 28.027106227106227, "grad_norm": 0.8203125, "learning_rate": 1.1362014023570342e-05, "loss": 0.0858, "step": 38257 }, { "epoch": 28.027838827838828, "grad_norm": 1.2734375, "learning_rate": 1.135991633708751e-05, "loss": 0.1672, "step": 38258 }, { "epoch": 28.02857142857143, "grad_norm": 0.98046875, "learning_rate": 1.135781883813766e-05, "loss": 0.1312, "step": 38259 }, { "epoch": 28.02930402930403, "grad_norm": 0.75390625, "learning_rate": 1.135572152673087e-05, "loss": 0.0968, "step": 38260 }, { "epoch": 28.03003663003663, "grad_norm": 1.0546875, "learning_rate": 1.1353624402877152e-05, "loss": 0.1352, "step": 38261 }, { "epoch": 28.03076923076923, "grad_norm": 0.61328125, "learning_rate": 1.135152746658657e-05, "loss": 0.1079, "step": 38262 }, { "epoch": 28.03150183150183, "grad_norm": 1.09375, "learning_rate": 1.1349430717869187e-05, "loss": 0.1022, "step": 38263 }, { "epoch": 28.032234432234432, "grad_norm": 0.84765625, "learning_rate": 1.1347334156735016e-05, "loss": 0.1203, "step": 38264 }, { "epoch": 28.032967032967033, "grad_norm": 0.91015625, "learning_rate": 1.1345237783194123e-05, "loss": 0.1157, "step": 38265 }, { "epoch": 28.033699633699634, "grad_norm": 0.99609375, "learning_rate": 1.1343141597256552e-05, "loss": 0.1019, "step": 38266 }, { "epoch": 28.034432234432234, "grad_norm": 1.3984375, "learning_rate": 1.134104559893235e-05, "loss": 0.1332, "step": 38267 }, { "epoch": 28.035164835164835, "grad_norm": 1.2578125, "learning_rate": 1.1338949788231547e-05, "loss": 0.1287, "step": 38268 }, { "epoch": 28.035897435897436, "grad_norm": 1.0625, "learning_rate": 1.1336854165164204e-05, "loss": 0.1045, "step": 38269 }, { "epoch": 28.036630036630036, "grad_norm": 1.0625, "learning_rate": 1.133475872974032e-05, "loss": 0.1195, "step": 38270 }, { "epoch": 28.037362637362637, "grad_norm": 1.125, "learning_rate": 1.1332663481969992e-05, "loss": 0.1224, "step": 38271 }, { "epoch": 28.038095238095238, "grad_norm": 0.88671875, "learning_rate": 1.1330568421863215e-05, "loss": 0.1242, "step": 38272 }, { "epoch": 28.03882783882784, "grad_norm": 1.078125, "learning_rate": 1.1328473549430048e-05, "loss": 0.1019, "step": 38273 }, { "epoch": 28.03956043956044, "grad_norm": 1.3046875, "learning_rate": 1.1326378864680533e-05, "loss": 0.1008, "step": 38274 }, { "epoch": 28.04029304029304, "grad_norm": 0.9609375, "learning_rate": 1.1324284367624668e-05, "loss": 0.1163, "step": 38275 }, { "epoch": 28.04102564102564, "grad_norm": 1.34375, "learning_rate": 1.1322190058272547e-05, "loss": 0.104, "step": 38276 }, { "epoch": 28.04175824175824, "grad_norm": 0.79296875, "learning_rate": 1.1320095936634162e-05, "loss": 0.11, "step": 38277 }, { "epoch": 28.042490842490842, "grad_norm": 1.1015625, "learning_rate": 1.1318002002719573e-05, "loss": 0.1605, "step": 38278 }, { "epoch": 28.043223443223443, "grad_norm": 0.828125, "learning_rate": 1.1315908256538783e-05, "loss": 0.1326, "step": 38279 }, { "epoch": 28.043956043956044, "grad_norm": 0.94140625, "learning_rate": 1.1313814698101844e-05, "loss": 0.1348, "step": 38280 }, { "epoch": 28.044688644688645, "grad_norm": 0.7734375, "learning_rate": 1.1311721327418776e-05, "loss": 0.1104, "step": 38281 }, { "epoch": 28.045421245421245, "grad_norm": 1.1015625, "learning_rate": 1.1309628144499632e-05, "loss": 0.0981, "step": 38282 }, { "epoch": 28.046153846153846, "grad_norm": 1.03125, "learning_rate": 1.1307535149354415e-05, "loss": 0.1122, "step": 38283 }, { "epoch": 28.046886446886447, "grad_norm": 0.63671875, "learning_rate": 1.1305442341993164e-05, "loss": 0.0974, "step": 38284 }, { "epoch": 28.047619047619047, "grad_norm": 0.8515625, "learning_rate": 1.1303349722425917e-05, "loss": 0.1196, "step": 38285 }, { "epoch": 28.048351648351648, "grad_norm": 0.92578125, "learning_rate": 1.130125729066267e-05, "loss": 0.1297, "step": 38286 }, { "epoch": 28.04908424908425, "grad_norm": 1.0078125, "learning_rate": 1.1299165046713475e-05, "loss": 0.1229, "step": 38287 }, { "epoch": 28.04981684981685, "grad_norm": 1.109375, "learning_rate": 1.1297072990588344e-05, "loss": 0.1414, "step": 38288 }, { "epoch": 28.05054945054945, "grad_norm": 1.1171875, "learning_rate": 1.1294981122297321e-05, "loss": 0.1305, "step": 38289 }, { "epoch": 28.05128205128205, "grad_norm": 1.203125, "learning_rate": 1.1292889441850398e-05, "loss": 0.1213, "step": 38290 }, { "epoch": 28.052014652014652, "grad_norm": 2.0625, "learning_rate": 1.1290797949257617e-05, "loss": 0.1741, "step": 38291 }, { "epoch": 28.052747252747253, "grad_norm": 1.046875, "learning_rate": 1.1288706644528992e-05, "loss": 0.1101, "step": 38292 }, { "epoch": 28.053479853479853, "grad_norm": 0.625, "learning_rate": 1.1286615527674557e-05, "loss": 0.1155, "step": 38293 }, { "epoch": 28.054212454212454, "grad_norm": 1.6484375, "learning_rate": 1.1284524598704314e-05, "loss": 0.1377, "step": 38294 }, { "epoch": 28.054945054945055, "grad_norm": 0.76953125, "learning_rate": 1.1282433857628292e-05, "loss": 0.1042, "step": 38295 }, { "epoch": 28.055677655677655, "grad_norm": 0.95703125, "learning_rate": 1.1280343304456484e-05, "loss": 0.0999, "step": 38296 }, { "epoch": 28.056410256410256, "grad_norm": 0.8203125, "learning_rate": 1.1278252939198945e-05, "loss": 0.1001, "step": 38297 }, { "epoch": 28.057142857142857, "grad_norm": 0.71875, "learning_rate": 1.1276162761865668e-05, "loss": 0.1069, "step": 38298 }, { "epoch": 28.057875457875458, "grad_norm": 0.69921875, "learning_rate": 1.1274072772466663e-05, "loss": 0.1103, "step": 38299 }, { "epoch": 28.05860805860806, "grad_norm": 0.6953125, "learning_rate": 1.1271982971011973e-05, "loss": 0.13, "step": 38300 }, { "epoch": 28.05934065934066, "grad_norm": 0.9765625, "learning_rate": 1.1269893357511553e-05, "loss": 0.1259, "step": 38301 }, { "epoch": 28.06007326007326, "grad_norm": 0.7265625, "learning_rate": 1.1267803931975485e-05, "loss": 0.1242, "step": 38302 }, { "epoch": 28.06080586080586, "grad_norm": 0.96484375, "learning_rate": 1.126571469441374e-05, "loss": 0.1226, "step": 38303 }, { "epoch": 28.06153846153846, "grad_norm": 0.703125, "learning_rate": 1.126362564483634e-05, "loss": 0.098, "step": 38304 }, { "epoch": 28.062271062271062, "grad_norm": 1.1875, "learning_rate": 1.1261536783253285e-05, "loss": 0.1336, "step": 38305 }, { "epoch": 28.063003663003663, "grad_norm": 0.703125, "learning_rate": 1.125944810967458e-05, "loss": 0.0986, "step": 38306 }, { "epoch": 28.063736263736264, "grad_norm": 1.1796875, "learning_rate": 1.1257359624110245e-05, "loss": 0.1408, "step": 38307 }, { "epoch": 28.064468864468864, "grad_norm": 0.92578125, "learning_rate": 1.1255271326570298e-05, "loss": 0.1228, "step": 38308 }, { "epoch": 28.065201465201465, "grad_norm": 1.0703125, "learning_rate": 1.1253183217064711e-05, "loss": 0.1346, "step": 38309 }, { "epoch": 28.065934065934066, "grad_norm": 0.875, "learning_rate": 1.1251095295603514e-05, "loss": 0.1559, "step": 38310 }, { "epoch": 28.066666666666666, "grad_norm": 0.91796875, "learning_rate": 1.1249007562196714e-05, "loss": 0.124, "step": 38311 }, { "epoch": 28.067399267399267, "grad_norm": 1.390625, "learning_rate": 1.1246920016854289e-05, "loss": 0.1305, "step": 38312 }, { "epoch": 28.068131868131868, "grad_norm": 1.1796875, "learning_rate": 1.1244832659586254e-05, "loss": 0.1098, "step": 38313 }, { "epoch": 28.06886446886447, "grad_norm": 1.5625, "learning_rate": 1.1242745490402616e-05, "loss": 0.1348, "step": 38314 }, { "epoch": 28.06959706959707, "grad_norm": 0.7265625, "learning_rate": 1.1240658509313382e-05, "loss": 0.1066, "step": 38315 }, { "epoch": 28.07032967032967, "grad_norm": 0.890625, "learning_rate": 1.1238571716328533e-05, "loss": 0.1416, "step": 38316 }, { "epoch": 28.07106227106227, "grad_norm": 0.953125, "learning_rate": 1.1236485111458071e-05, "loss": 0.1133, "step": 38317 }, { "epoch": 28.07179487179487, "grad_norm": 1.1640625, "learning_rate": 1.1234398694712003e-05, "loss": 0.1177, "step": 38318 }, { "epoch": 28.072527472527472, "grad_norm": 1.078125, "learning_rate": 1.123231246610033e-05, "loss": 0.0981, "step": 38319 }, { "epoch": 28.073260073260073, "grad_norm": 1.2890625, "learning_rate": 1.1230226425633023e-05, "loss": 0.1005, "step": 38320 }, { "epoch": 28.073992673992674, "grad_norm": 0.76953125, "learning_rate": 1.1228140573320111e-05, "loss": 0.1153, "step": 38321 }, { "epoch": 28.074725274725274, "grad_norm": 1.0390625, "learning_rate": 1.122605490917154e-05, "loss": 0.128, "step": 38322 }, { "epoch": 28.075457875457875, "grad_norm": 1.6015625, "learning_rate": 1.1223969433197364e-05, "loss": 0.1195, "step": 38323 }, { "epoch": 28.076190476190476, "grad_norm": 0.65234375, "learning_rate": 1.1221884145407525e-05, "loss": 0.098, "step": 38324 }, { "epoch": 28.076923076923077, "grad_norm": 0.72265625, "learning_rate": 1.1219799045812036e-05, "loss": 0.1111, "step": 38325 }, { "epoch": 28.077655677655677, "grad_norm": 0.80078125, "learning_rate": 1.12177141344209e-05, "loss": 0.1086, "step": 38326 }, { "epoch": 28.078388278388278, "grad_norm": 0.8984375, "learning_rate": 1.1215629411244061e-05, "loss": 0.1295, "step": 38327 }, { "epoch": 28.07912087912088, "grad_norm": 1.578125, "learning_rate": 1.1213544876291567e-05, "loss": 0.1213, "step": 38328 }, { "epoch": 28.07985347985348, "grad_norm": 0.451171875, "learning_rate": 1.121146052957336e-05, "loss": 0.1046, "step": 38329 }, { "epoch": 28.08058608058608, "grad_norm": 1.0390625, "learning_rate": 1.1209376371099457e-05, "loss": 0.095, "step": 38330 }, { "epoch": 28.08131868131868, "grad_norm": 0.99609375, "learning_rate": 1.120729240087982e-05, "loss": 0.1367, "step": 38331 }, { "epoch": 28.08205128205128, "grad_norm": 1.2265625, "learning_rate": 1.120520861892444e-05, "loss": 0.1286, "step": 38332 }, { "epoch": 28.082783882783882, "grad_norm": 1.0546875, "learning_rate": 1.1203125025243308e-05, "loss": 0.1395, "step": 38333 }, { "epoch": 28.083516483516483, "grad_norm": 0.8046875, "learning_rate": 1.1201041619846416e-05, "loss": 0.1343, "step": 38334 }, { "epoch": 28.084249084249084, "grad_norm": 1.0625, "learning_rate": 1.119895840274372e-05, "loss": 0.1106, "step": 38335 }, { "epoch": 28.084981684981685, "grad_norm": 1.3671875, "learning_rate": 1.1196875373945218e-05, "loss": 0.1264, "step": 38336 }, { "epoch": 28.085714285714285, "grad_norm": 1.625, "learning_rate": 1.1194792533460901e-05, "loss": 0.1504, "step": 38337 }, { "epoch": 28.086446886446886, "grad_norm": 1.4140625, "learning_rate": 1.1192709881300727e-05, "loss": 0.1496, "step": 38338 }, { "epoch": 28.087179487179487, "grad_norm": 1.4453125, "learning_rate": 1.1190627417474677e-05, "loss": 0.1138, "step": 38339 }, { "epoch": 28.087912087912088, "grad_norm": 1.203125, "learning_rate": 1.1188545141992739e-05, "loss": 0.114, "step": 38340 }, { "epoch": 28.08864468864469, "grad_norm": 1.3515625, "learning_rate": 1.11864630548649e-05, "loss": 0.1363, "step": 38341 }, { "epoch": 28.08937728937729, "grad_norm": 1.8125, "learning_rate": 1.1184381156101108e-05, "loss": 0.1222, "step": 38342 }, { "epoch": 28.09010989010989, "grad_norm": 1.078125, "learning_rate": 1.1182299445711354e-05, "loss": 0.1296, "step": 38343 }, { "epoch": 28.09084249084249, "grad_norm": 1.3828125, "learning_rate": 1.1180217923705606e-05, "loss": 0.1083, "step": 38344 }, { "epoch": 28.09157509157509, "grad_norm": 0.578125, "learning_rate": 1.1178136590093862e-05, "loss": 0.1111, "step": 38345 }, { "epoch": 28.092307692307692, "grad_norm": 0.96875, "learning_rate": 1.1176055444886054e-05, "loss": 0.135, "step": 38346 }, { "epoch": 28.093040293040293, "grad_norm": 1.171875, "learning_rate": 1.117397448809219e-05, "loss": 0.1221, "step": 38347 }, { "epoch": 28.093772893772893, "grad_norm": 1.2109375, "learning_rate": 1.1171893719722196e-05, "loss": 0.1443, "step": 38348 }, { "epoch": 28.094505494505494, "grad_norm": 1.3359375, "learning_rate": 1.1169813139786098e-05, "loss": 0.1503, "step": 38349 }, { "epoch": 28.095238095238095, "grad_norm": 0.87109375, "learning_rate": 1.1167732748293826e-05, "loss": 0.1145, "step": 38350 }, { "epoch": 28.095970695970696, "grad_norm": 1.0078125, "learning_rate": 1.1165652545255353e-05, "loss": 0.129, "step": 38351 }, { "epoch": 28.096703296703296, "grad_norm": 0.96484375, "learning_rate": 1.1163572530680667e-05, "loss": 0.0984, "step": 38352 }, { "epoch": 28.097435897435897, "grad_norm": 0.82421875, "learning_rate": 1.1161492704579693e-05, "loss": 0.1022, "step": 38353 }, { "epoch": 28.098168498168498, "grad_norm": 0.859375, "learning_rate": 1.1159413066962443e-05, "loss": 0.0985, "step": 38354 }, { "epoch": 28.0989010989011, "grad_norm": 0.9453125, "learning_rate": 1.1157333617838853e-05, "loss": 0.1262, "step": 38355 }, { "epoch": 28.0996336996337, "grad_norm": 1.015625, "learning_rate": 1.1155254357218902e-05, "loss": 0.1362, "step": 38356 }, { "epoch": 28.1003663003663, "grad_norm": 1.5703125, "learning_rate": 1.115317528511253e-05, "loss": 0.1308, "step": 38357 }, { "epoch": 28.1010989010989, "grad_norm": 1.2734375, "learning_rate": 1.1151096401529712e-05, "loss": 0.1198, "step": 38358 }, { "epoch": 28.1018315018315, "grad_norm": 1.828125, "learning_rate": 1.1149017706480407e-05, "loss": 0.1296, "step": 38359 }, { "epoch": 28.102564102564102, "grad_norm": 0.6796875, "learning_rate": 1.1146939199974594e-05, "loss": 0.1036, "step": 38360 }, { "epoch": 28.103296703296703, "grad_norm": 0.78125, "learning_rate": 1.1144860882022199e-05, "loss": 0.145, "step": 38361 }, { "epoch": 28.104029304029304, "grad_norm": 0.86328125, "learning_rate": 1.1142782752633199e-05, "loss": 0.1292, "step": 38362 }, { "epoch": 28.104761904761904, "grad_norm": 1.0078125, "learning_rate": 1.1140704811817555e-05, "loss": 0.1041, "step": 38363 }, { "epoch": 28.105494505494505, "grad_norm": 0.77734375, "learning_rate": 1.1138627059585204e-05, "loss": 0.1145, "step": 38364 }, { "epoch": 28.106227106227106, "grad_norm": 1.5546875, "learning_rate": 1.113654949594611e-05, "loss": 0.136, "step": 38365 }, { "epoch": 28.106959706959707, "grad_norm": 1.390625, "learning_rate": 1.1134472120910233e-05, "loss": 0.1198, "step": 38366 }, { "epoch": 28.107692307692307, "grad_norm": 1.4140625, "learning_rate": 1.1132394934487533e-05, "loss": 0.1273, "step": 38367 }, { "epoch": 28.108424908424908, "grad_norm": 0.765625, "learning_rate": 1.1130317936687938e-05, "loss": 0.1029, "step": 38368 }, { "epoch": 28.10915750915751, "grad_norm": 0.96484375, "learning_rate": 1.1128241127521415e-05, "loss": 0.1235, "step": 38369 }, { "epoch": 28.10989010989011, "grad_norm": 1.8359375, "learning_rate": 1.1126164506997917e-05, "loss": 0.1577, "step": 38370 }, { "epoch": 28.11062271062271, "grad_norm": 1.0546875, "learning_rate": 1.1124088075127402e-05, "loss": 0.1143, "step": 38371 }, { "epoch": 28.11135531135531, "grad_norm": 0.8125, "learning_rate": 1.112201183191979e-05, "loss": 0.1031, "step": 38372 }, { "epoch": 28.11208791208791, "grad_norm": 0.79296875, "learning_rate": 1.111993577738505e-05, "loss": 0.0985, "step": 38373 }, { "epoch": 28.112820512820512, "grad_norm": 1.2421875, "learning_rate": 1.1117859911533129e-05, "loss": 0.1286, "step": 38374 }, { "epoch": 28.113553113553113, "grad_norm": 1.15625, "learning_rate": 1.1115784234373978e-05, "loss": 0.1058, "step": 38375 }, { "epoch": 28.114285714285714, "grad_norm": 1.046875, "learning_rate": 1.111370874591752e-05, "loss": 0.1136, "step": 38376 }, { "epoch": 28.115018315018315, "grad_norm": 1.21875, "learning_rate": 1.1111633446173715e-05, "loss": 0.1255, "step": 38377 }, { "epoch": 28.115750915750915, "grad_norm": 0.96484375, "learning_rate": 1.1109558335152519e-05, "loss": 0.1443, "step": 38378 }, { "epoch": 28.116483516483516, "grad_norm": 1.1328125, "learning_rate": 1.1107483412863832e-05, "loss": 0.1102, "step": 38379 }, { "epoch": 28.117216117216117, "grad_norm": 0.7421875, "learning_rate": 1.1105408679317646e-05, "loss": 0.1042, "step": 38380 }, { "epoch": 28.117948717948718, "grad_norm": 0.94921875, "learning_rate": 1.1103334134523874e-05, "loss": 0.1115, "step": 38381 }, { "epoch": 28.11868131868132, "grad_norm": 1.125, "learning_rate": 1.1101259778492465e-05, "loss": 0.1072, "step": 38382 }, { "epoch": 28.11941391941392, "grad_norm": 1.5, "learning_rate": 1.109918561123335e-05, "loss": 0.1789, "step": 38383 }, { "epoch": 28.12014652014652, "grad_norm": 1.234375, "learning_rate": 1.1097111632756469e-05, "loss": 0.1418, "step": 38384 }, { "epoch": 28.12087912087912, "grad_norm": 0.71875, "learning_rate": 1.1095037843071756e-05, "loss": 0.1062, "step": 38385 }, { "epoch": 28.12161172161172, "grad_norm": 0.703125, "learning_rate": 1.1092964242189175e-05, "loss": 0.1154, "step": 38386 }, { "epoch": 28.122344322344322, "grad_norm": 0.91015625, "learning_rate": 1.1090890830118618e-05, "loss": 0.1085, "step": 38387 }, { "epoch": 28.123076923076923, "grad_norm": 1.3515625, "learning_rate": 1.108881760687004e-05, "loss": 0.1453, "step": 38388 }, { "epoch": 28.123809523809523, "grad_norm": 1.234375, "learning_rate": 1.1086744572453393e-05, "loss": 0.1492, "step": 38389 }, { "epoch": 28.124542124542124, "grad_norm": 0.9609375, "learning_rate": 1.1084671726878564e-05, "loss": 0.1187, "step": 38390 }, { "epoch": 28.125274725274725, "grad_norm": 0.765625, "learning_rate": 1.1082599070155531e-05, "loss": 0.1201, "step": 38391 }, { "epoch": 28.126007326007326, "grad_norm": 1.3671875, "learning_rate": 1.10805266022942e-05, "loss": 0.15, "step": 38392 }, { "epoch": 28.126739926739926, "grad_norm": 0.8359375, "learning_rate": 1.1078454323304517e-05, "loss": 0.134, "step": 38393 }, { "epoch": 28.127472527472527, "grad_norm": 0.7734375, "learning_rate": 1.107638223319639e-05, "loss": 0.1097, "step": 38394 }, { "epoch": 28.128205128205128, "grad_norm": 0.9921875, "learning_rate": 1.1074310331979757e-05, "loss": 0.136, "step": 38395 }, { "epoch": 28.12893772893773, "grad_norm": 1.1328125, "learning_rate": 1.1072238619664547e-05, "loss": 0.1132, "step": 38396 }, { "epoch": 28.12967032967033, "grad_norm": 1.125, "learning_rate": 1.1070167096260693e-05, "loss": 0.1549, "step": 38397 }, { "epoch": 28.13040293040293, "grad_norm": 0.87890625, "learning_rate": 1.1068095761778103e-05, "loss": 0.1072, "step": 38398 }, { "epoch": 28.13113553113553, "grad_norm": 0.9609375, "learning_rate": 1.1066024616226706e-05, "loss": 0.0835, "step": 38399 }, { "epoch": 28.13186813186813, "grad_norm": 1.296875, "learning_rate": 1.1063953659616438e-05, "loss": 0.1087, "step": 38400 }, { "epoch": 28.132600732600732, "grad_norm": 1.3125, "learning_rate": 1.1061882891957222e-05, "loss": 0.0852, "step": 38401 }, { "epoch": 28.133333333333333, "grad_norm": 0.796875, "learning_rate": 1.1059812313258959e-05, "loss": 0.1041, "step": 38402 }, { "epoch": 28.134065934065934, "grad_norm": 0.82421875, "learning_rate": 1.1057741923531584e-05, "loss": 0.1243, "step": 38403 }, { "epoch": 28.134798534798534, "grad_norm": 1.1015625, "learning_rate": 1.105567172278503e-05, "loss": 0.1125, "step": 38404 }, { "epoch": 28.135531135531135, "grad_norm": 1.1015625, "learning_rate": 1.1053601711029173e-05, "loss": 0.1176, "step": 38405 }, { "epoch": 28.136263736263736, "grad_norm": 0.91796875, "learning_rate": 1.1051531888273988e-05, "loss": 0.1212, "step": 38406 }, { "epoch": 28.136996336996337, "grad_norm": 0.93359375, "learning_rate": 1.1049462254529346e-05, "loss": 0.1018, "step": 38407 }, { "epoch": 28.137728937728937, "grad_norm": 1.046875, "learning_rate": 1.10473928098052e-05, "loss": 0.1289, "step": 38408 }, { "epoch": 28.138461538461538, "grad_norm": 1.34375, "learning_rate": 1.1045323554111433e-05, "loss": 0.1411, "step": 38409 }, { "epoch": 28.13919413919414, "grad_norm": 1.078125, "learning_rate": 1.1043254487457967e-05, "loss": 0.128, "step": 38410 }, { "epoch": 28.13992673992674, "grad_norm": 0.83984375, "learning_rate": 1.1041185609854729e-05, "loss": 0.1339, "step": 38411 }, { "epoch": 28.14065934065934, "grad_norm": 0.83203125, "learning_rate": 1.1039116921311633e-05, "loss": 0.103, "step": 38412 }, { "epoch": 28.14139194139194, "grad_norm": 1.1484375, "learning_rate": 1.103704842183857e-05, "loss": 0.1299, "step": 38413 }, { "epoch": 28.14212454212454, "grad_norm": 0.8515625, "learning_rate": 1.103498011144546e-05, "loss": 0.1374, "step": 38414 }, { "epoch": 28.142857142857142, "grad_norm": 0.875, "learning_rate": 1.1032911990142234e-05, "loss": 0.1191, "step": 38415 }, { "epoch": 28.143589743589743, "grad_norm": 1.296875, "learning_rate": 1.1030844057938751e-05, "loss": 0.1365, "step": 38416 }, { "epoch": 28.144322344322344, "grad_norm": 1.203125, "learning_rate": 1.102877631484498e-05, "loss": 0.1201, "step": 38417 }, { "epoch": 28.145054945054945, "grad_norm": 0.8828125, "learning_rate": 1.1026708760870785e-05, "loss": 0.1116, "step": 38418 }, { "epoch": 28.145787545787545, "grad_norm": 1.0546875, "learning_rate": 1.1024641396026102e-05, "loss": 0.1237, "step": 38419 }, { "epoch": 28.146520146520146, "grad_norm": 0.7265625, "learning_rate": 1.1022574220320809e-05, "loss": 0.0956, "step": 38420 }, { "epoch": 28.147252747252747, "grad_norm": 1.25, "learning_rate": 1.1020507233764817e-05, "loss": 0.1411, "step": 38421 }, { "epoch": 28.147985347985347, "grad_norm": 0.73046875, "learning_rate": 1.1018440436368041e-05, "loss": 0.1266, "step": 38422 }, { "epoch": 28.148717948717948, "grad_norm": 1.1328125, "learning_rate": 1.1016373828140385e-05, "loss": 0.1203, "step": 38423 }, { "epoch": 28.14945054945055, "grad_norm": 1.3828125, "learning_rate": 1.1014307409091736e-05, "loss": 0.1353, "step": 38424 }, { "epoch": 28.15018315018315, "grad_norm": 0.76171875, "learning_rate": 1.1012241179232e-05, "loss": 0.12, "step": 38425 }, { "epoch": 28.15091575091575, "grad_norm": 1.2109375, "learning_rate": 1.1010175138571082e-05, "loss": 0.1163, "step": 38426 }, { "epoch": 28.15164835164835, "grad_norm": 1.046875, "learning_rate": 1.1008109287118889e-05, "loss": 0.1202, "step": 38427 }, { "epoch": 28.152380952380952, "grad_norm": 1.5546875, "learning_rate": 1.1006043624885294e-05, "loss": 0.155, "step": 38428 }, { "epoch": 28.153113553113553, "grad_norm": 1.0234375, "learning_rate": 1.1003978151880216e-05, "loss": 0.128, "step": 38429 }, { "epoch": 28.153846153846153, "grad_norm": 0.59375, "learning_rate": 1.1001912868113547e-05, "loss": 0.1156, "step": 38430 }, { "epoch": 28.154578754578754, "grad_norm": 1.2890625, "learning_rate": 1.0999847773595161e-05, "loss": 0.1169, "step": 38431 }, { "epoch": 28.155311355311355, "grad_norm": 0.82421875, "learning_rate": 1.0997782868334998e-05, "loss": 0.1091, "step": 38432 }, { "epoch": 28.156043956043955, "grad_norm": 1.8203125, "learning_rate": 1.0995718152342907e-05, "loss": 0.1408, "step": 38433 }, { "epoch": 28.156776556776556, "grad_norm": 1.2265625, "learning_rate": 1.0993653625628817e-05, "loss": 0.1026, "step": 38434 }, { "epoch": 28.157509157509157, "grad_norm": 0.875, "learning_rate": 1.0991589288202587e-05, "loss": 0.1315, "step": 38435 }, { "epoch": 28.158241758241758, "grad_norm": 0.79296875, "learning_rate": 1.098952514007412e-05, "loss": 0.1232, "step": 38436 }, { "epoch": 28.15897435897436, "grad_norm": 1.171875, "learning_rate": 1.0987461181253309e-05, "loss": 0.133, "step": 38437 }, { "epoch": 28.15970695970696, "grad_norm": 1.234375, "learning_rate": 1.098539741175006e-05, "loss": 0.1271, "step": 38438 }, { "epoch": 28.16043956043956, "grad_norm": 0.98828125, "learning_rate": 1.0983333831574226e-05, "loss": 0.1294, "step": 38439 }, { "epoch": 28.16117216117216, "grad_norm": 1.125, "learning_rate": 1.0981270440735716e-05, "loss": 0.1329, "step": 38440 }, { "epoch": 28.16190476190476, "grad_norm": 1.2265625, "learning_rate": 1.0979207239244418e-05, "loss": 0.0941, "step": 38441 }, { "epoch": 28.162637362637362, "grad_norm": 1.3203125, "learning_rate": 1.0977144227110191e-05, "loss": 0.142, "step": 38442 }, { "epoch": 28.163369963369963, "grad_norm": 0.90625, "learning_rate": 1.0975081404342965e-05, "loss": 0.118, "step": 38443 }, { "epoch": 28.164102564102564, "grad_norm": 1.0234375, "learning_rate": 1.0973018770952593e-05, "loss": 0.113, "step": 38444 }, { "epoch": 28.164835164835164, "grad_norm": 1.109375, "learning_rate": 1.0970956326948968e-05, "loss": 0.1131, "step": 38445 }, { "epoch": 28.165567765567765, "grad_norm": 1.1953125, "learning_rate": 1.0968894072341961e-05, "loss": 0.1218, "step": 38446 }, { "epoch": 28.166300366300366, "grad_norm": 1.21875, "learning_rate": 1.0966832007141456e-05, "loss": 0.1047, "step": 38447 }, { "epoch": 28.167032967032966, "grad_norm": 1.5625, "learning_rate": 1.0964770131357338e-05, "loss": 0.1428, "step": 38448 }, { "epoch": 28.167765567765567, "grad_norm": 1.03125, "learning_rate": 1.0962708444999497e-05, "loss": 0.102, "step": 38449 }, { "epoch": 28.168498168498168, "grad_norm": 1.0546875, "learning_rate": 1.096064694807779e-05, "loss": 0.1074, "step": 38450 }, { "epoch": 28.16923076923077, "grad_norm": 1.234375, "learning_rate": 1.0958585640602096e-05, "loss": 0.113, "step": 38451 }, { "epoch": 28.16996336996337, "grad_norm": 0.99609375, "learning_rate": 1.0956524522582301e-05, "loss": 0.1033, "step": 38452 }, { "epoch": 28.17069597069597, "grad_norm": 0.75390625, "learning_rate": 1.0954463594028294e-05, "loss": 0.0981, "step": 38453 }, { "epoch": 28.17142857142857, "grad_norm": 0.8515625, "learning_rate": 1.0952402854949925e-05, "loss": 0.1021, "step": 38454 }, { "epoch": 28.17216117216117, "grad_norm": 1.171875, "learning_rate": 1.095034230535707e-05, "loss": 0.1473, "step": 38455 }, { "epoch": 28.172893772893772, "grad_norm": 0.52734375, "learning_rate": 1.0948281945259622e-05, "loss": 0.1096, "step": 38456 }, { "epoch": 28.173626373626373, "grad_norm": 1.7265625, "learning_rate": 1.0946221774667415e-05, "loss": 0.1208, "step": 38457 }, { "epoch": 28.174358974358974, "grad_norm": 1.0078125, "learning_rate": 1.094416179359037e-05, "loss": 0.1312, "step": 38458 }, { "epoch": 28.175091575091574, "grad_norm": 1.2578125, "learning_rate": 1.0942102002038317e-05, "loss": 0.1303, "step": 38459 }, { "epoch": 28.175824175824175, "grad_norm": 1.4765625, "learning_rate": 1.0940042400021154e-05, "loss": 0.1592, "step": 38460 }, { "epoch": 28.176556776556776, "grad_norm": 1.296875, "learning_rate": 1.0937982987548718e-05, "loss": 0.1265, "step": 38461 }, { "epoch": 28.177289377289377, "grad_norm": 1.265625, "learning_rate": 1.0935923764630896e-05, "loss": 0.118, "step": 38462 }, { "epoch": 28.178021978021977, "grad_norm": 0.90234375, "learning_rate": 1.0933864731277549e-05, "loss": 0.1031, "step": 38463 }, { "epoch": 28.178754578754578, "grad_norm": 0.90625, "learning_rate": 1.093180588749856e-05, "loss": 0.1361, "step": 38464 }, { "epoch": 28.17948717948718, "grad_norm": 1.296875, "learning_rate": 1.0929747233303762e-05, "loss": 0.1151, "step": 38465 }, { "epoch": 28.18021978021978, "grad_norm": 1.171875, "learning_rate": 1.092768876870304e-05, "loss": 0.121, "step": 38466 }, { "epoch": 28.18095238095238, "grad_norm": 0.72265625, "learning_rate": 1.0925630493706262e-05, "loss": 0.1226, "step": 38467 }, { "epoch": 28.18168498168498, "grad_norm": 0.64453125, "learning_rate": 1.0923572408323254e-05, "loss": 0.1207, "step": 38468 }, { "epoch": 28.18241758241758, "grad_norm": 0.8359375, "learning_rate": 1.0921514512563928e-05, "loss": 0.1419, "step": 38469 }, { "epoch": 28.183150183150182, "grad_norm": 0.6484375, "learning_rate": 1.0919456806438105e-05, "loss": 0.1007, "step": 38470 }, { "epoch": 28.183882783882783, "grad_norm": 0.8203125, "learning_rate": 1.0917399289955674e-05, "loss": 0.1196, "step": 38471 }, { "epoch": 28.184615384615384, "grad_norm": 1.1328125, "learning_rate": 1.0915341963126464e-05, "loss": 0.1347, "step": 38472 }, { "epoch": 28.185347985347985, "grad_norm": 0.97265625, "learning_rate": 1.0913284825960345e-05, "loss": 0.1324, "step": 38473 }, { "epoch": 28.186080586080585, "grad_norm": 0.76953125, "learning_rate": 1.0911227878467172e-05, "loss": 0.1104, "step": 38474 }, { "epoch": 28.186813186813186, "grad_norm": 1.203125, "learning_rate": 1.0909171120656819e-05, "loss": 0.1422, "step": 38475 }, { "epoch": 28.187545787545787, "grad_norm": 1.2890625, "learning_rate": 1.090711455253911e-05, "loss": 0.1468, "step": 38476 }, { "epoch": 28.188278388278388, "grad_norm": 1.3984375, "learning_rate": 1.090505817412391e-05, "loss": 0.1217, "step": 38477 }, { "epoch": 28.18901098901099, "grad_norm": 1.328125, "learning_rate": 1.0903001985421071e-05, "loss": 0.1231, "step": 38478 }, { "epoch": 28.18974358974359, "grad_norm": 0.83203125, "learning_rate": 1.0900945986440466e-05, "loss": 0.1208, "step": 38479 }, { "epoch": 28.19047619047619, "grad_norm": 1.0, "learning_rate": 1.089889017719192e-05, "loss": 0.1102, "step": 38480 }, { "epoch": 28.19120879120879, "grad_norm": 1.0, "learning_rate": 1.0896834557685281e-05, "loss": 0.1381, "step": 38481 }, { "epoch": 28.19194139194139, "grad_norm": 1.1484375, "learning_rate": 1.0894779127930429e-05, "loss": 0.1191, "step": 38482 }, { "epoch": 28.192673992673992, "grad_norm": 1.5390625, "learning_rate": 1.0892723887937166e-05, "loss": 0.1328, "step": 38483 }, { "epoch": 28.193406593406593, "grad_norm": 0.9765625, "learning_rate": 1.089066883771539e-05, "loss": 0.1211, "step": 38484 }, { "epoch": 28.194139194139193, "grad_norm": 0.7890625, "learning_rate": 1.0888613977274905e-05, "loss": 0.1078, "step": 38485 }, { "epoch": 28.194871794871794, "grad_norm": 0.97265625, "learning_rate": 1.0886559306625593e-05, "loss": 0.1149, "step": 38486 }, { "epoch": 28.195604395604395, "grad_norm": 0.69921875, "learning_rate": 1.0884504825777262e-05, "loss": 0.1031, "step": 38487 }, { "epoch": 28.196336996336996, "grad_norm": 0.9609375, "learning_rate": 1.0882450534739776e-05, "loss": 0.1419, "step": 38488 }, { "epoch": 28.197069597069596, "grad_norm": 0.78125, "learning_rate": 1.088039643352297e-05, "loss": 0.1244, "step": 38489 }, { "epoch": 28.197802197802197, "grad_norm": 1.0234375, "learning_rate": 1.0878342522136713e-05, "loss": 0.1259, "step": 38490 }, { "epoch": 28.198534798534798, "grad_norm": 1.109375, "learning_rate": 1.0876288800590802e-05, "loss": 0.122, "step": 38491 }, { "epoch": 28.1992673992674, "grad_norm": 1.2265625, "learning_rate": 1.0874235268895108e-05, "loss": 0.1413, "step": 38492 }, { "epoch": 28.2, "grad_norm": 1.234375, "learning_rate": 1.0872181927059469e-05, "loss": 0.1289, "step": 38493 }, { "epoch": 28.2007326007326, "grad_norm": 0.73828125, "learning_rate": 1.0870128775093686e-05, "loss": 0.1392, "step": 38494 }, { "epoch": 28.2014652014652, "grad_norm": 1.03125, "learning_rate": 1.0868075813007656e-05, "loss": 0.14, "step": 38495 }, { "epoch": 28.2021978021978, "grad_norm": 0.84375, "learning_rate": 1.0866023040811174e-05, "loss": 0.1246, "step": 38496 }, { "epoch": 28.202930402930402, "grad_norm": 0.8046875, "learning_rate": 1.0863970458514095e-05, "loss": 0.1111, "step": 38497 }, { "epoch": 28.203663003663003, "grad_norm": 0.921875, "learning_rate": 1.0861918066126234e-05, "loss": 0.1534, "step": 38498 }, { "epoch": 28.204395604395604, "grad_norm": 1.1875, "learning_rate": 1.0859865863657433e-05, "loss": 0.1295, "step": 38499 }, { "epoch": 28.205128205128204, "grad_norm": 1.21875, "learning_rate": 1.0857813851117535e-05, "loss": 0.1156, "step": 38500 }, { "epoch": 28.205860805860805, "grad_norm": 1.28125, "learning_rate": 1.0855762028516372e-05, "loss": 0.159, "step": 38501 }, { "epoch": 28.206593406593406, "grad_norm": 1.0625, "learning_rate": 1.0853710395863755e-05, "loss": 0.1322, "step": 38502 }, { "epoch": 28.207326007326007, "grad_norm": 1.515625, "learning_rate": 1.0851658953169528e-05, "loss": 0.1097, "step": 38503 }, { "epoch": 28.208058608058607, "grad_norm": 0.921875, "learning_rate": 1.0849607700443516e-05, "loss": 0.136, "step": 38504 }, { "epoch": 28.208791208791208, "grad_norm": 1.5625, "learning_rate": 1.0847556637695552e-05, "loss": 0.1197, "step": 38505 }, { "epoch": 28.20952380952381, "grad_norm": 1.2734375, "learning_rate": 1.0845505764935478e-05, "loss": 0.1476, "step": 38506 }, { "epoch": 28.21025641025641, "grad_norm": 1.5859375, "learning_rate": 1.0843455082173083e-05, "loss": 0.1173, "step": 38507 }, { "epoch": 28.21098901098901, "grad_norm": 0.875, "learning_rate": 1.0841404589418227e-05, "loss": 0.1276, "step": 38508 }, { "epoch": 28.21172161172161, "grad_norm": 1.28125, "learning_rate": 1.08393542866807e-05, "loss": 0.1187, "step": 38509 }, { "epoch": 28.21245421245421, "grad_norm": 1.265625, "learning_rate": 1.083730417397037e-05, "loss": 0.1331, "step": 38510 }, { "epoch": 28.213186813186812, "grad_norm": 1.0078125, "learning_rate": 1.083525425129702e-05, "loss": 0.1236, "step": 38511 }, { "epoch": 28.213919413919413, "grad_norm": 1.296875, "learning_rate": 1.0833204518670509e-05, "loss": 0.1613, "step": 38512 }, { "epoch": 28.214652014652014, "grad_norm": 1.421875, "learning_rate": 1.0831154976100617e-05, "loss": 0.1315, "step": 38513 }, { "epoch": 28.215384615384615, "grad_norm": 1.0, "learning_rate": 1.0829105623597185e-05, "loss": 0.1231, "step": 38514 }, { "epoch": 28.216117216117215, "grad_norm": 0.78515625, "learning_rate": 1.0827056461170032e-05, "loss": 0.1331, "step": 38515 }, { "epoch": 28.216849816849816, "grad_norm": 0.86328125, "learning_rate": 1.0825007488828988e-05, "loss": 0.0947, "step": 38516 }, { "epoch": 28.217582417582417, "grad_norm": 1.03125, "learning_rate": 1.0822958706583842e-05, "loss": 0.1274, "step": 38517 }, { "epoch": 28.218315018315018, "grad_norm": 0.8515625, "learning_rate": 1.0820910114444433e-05, "loss": 0.1206, "step": 38518 }, { "epoch": 28.21904761904762, "grad_norm": 1.0859375, "learning_rate": 1.0818861712420573e-05, "loss": 0.1301, "step": 38519 }, { "epoch": 28.21978021978022, "grad_norm": 1.2109375, "learning_rate": 1.0816813500522055e-05, "loss": 0.1091, "step": 38520 }, { "epoch": 28.22051282051282, "grad_norm": 1.046875, "learning_rate": 1.081476547875873e-05, "loss": 0.0986, "step": 38521 }, { "epoch": 28.22124542124542, "grad_norm": 0.62890625, "learning_rate": 1.0812717647140384e-05, "loss": 0.1188, "step": 38522 }, { "epoch": 28.22197802197802, "grad_norm": 1.953125, "learning_rate": 1.0810670005676845e-05, "loss": 0.1225, "step": 38523 }, { "epoch": 28.222710622710622, "grad_norm": 0.7109375, "learning_rate": 1.0808622554377908e-05, "loss": 0.108, "step": 38524 }, { "epoch": 28.223443223443223, "grad_norm": 1.078125, "learning_rate": 1.0806575293253387e-05, "loss": 0.1123, "step": 38525 }, { "epoch": 28.224175824175823, "grad_norm": 1.2421875, "learning_rate": 1.0804528222313092e-05, "loss": 0.1504, "step": 38526 }, { "epoch": 28.224908424908424, "grad_norm": 1.0, "learning_rate": 1.0802481341566847e-05, "loss": 0.0963, "step": 38527 }, { "epoch": 28.225641025641025, "grad_norm": 0.85546875, "learning_rate": 1.0800434651024435e-05, "loss": 0.1333, "step": 38528 }, { "epoch": 28.226373626373626, "grad_norm": 0.9453125, "learning_rate": 1.079838815069567e-05, "loss": 0.1129, "step": 38529 }, { "epoch": 28.227106227106226, "grad_norm": 0.96875, "learning_rate": 1.0796341840590366e-05, "loss": 0.1174, "step": 38530 }, { "epoch": 28.227838827838827, "grad_norm": 1.640625, "learning_rate": 1.0794295720718318e-05, "loss": 0.131, "step": 38531 }, { "epoch": 28.228571428571428, "grad_norm": 0.80859375, "learning_rate": 1.0792249791089346e-05, "loss": 0.1184, "step": 38532 }, { "epoch": 28.22930402930403, "grad_norm": 1.0546875, "learning_rate": 1.079020405171323e-05, "loss": 0.1121, "step": 38533 }, { "epoch": 28.23003663003663, "grad_norm": 1.28125, "learning_rate": 1.0788158502599794e-05, "loss": 0.1384, "step": 38534 }, { "epoch": 28.23076923076923, "grad_norm": 0.72265625, "learning_rate": 1.0786113143758803e-05, "loss": 0.115, "step": 38535 }, { "epoch": 28.23150183150183, "grad_norm": 1.3359375, "learning_rate": 1.0784067975200103e-05, "loss": 0.1187, "step": 38536 }, { "epoch": 28.23223443223443, "grad_norm": 0.8203125, "learning_rate": 1.0782022996933462e-05, "loss": 0.1059, "step": 38537 }, { "epoch": 28.232967032967032, "grad_norm": 0.72265625, "learning_rate": 1.07799782089687e-05, "loss": 0.1135, "step": 38538 }, { "epoch": 28.233699633699633, "grad_norm": 1.3125, "learning_rate": 1.0777933611315587e-05, "loss": 0.1545, "step": 38539 }, { "epoch": 28.234432234432234, "grad_norm": 1.0390625, "learning_rate": 1.0775889203983934e-05, "loss": 0.1344, "step": 38540 }, { "epoch": 28.235164835164834, "grad_norm": 1.3984375, "learning_rate": 1.0773844986983533e-05, "loss": 0.165, "step": 38541 }, { "epoch": 28.235897435897435, "grad_norm": 0.97265625, "learning_rate": 1.0771800960324201e-05, "loss": 0.1422, "step": 38542 }, { "epoch": 28.236630036630036, "grad_norm": 0.8671875, "learning_rate": 1.0769757124015697e-05, "loss": 0.1179, "step": 38543 }, { "epoch": 28.237362637362637, "grad_norm": 0.78125, "learning_rate": 1.076771347806783e-05, "loss": 0.1233, "step": 38544 }, { "epoch": 28.238095238095237, "grad_norm": 0.83984375, "learning_rate": 1.0765670022490406e-05, "loss": 0.1148, "step": 38545 }, { "epoch": 28.238827838827838, "grad_norm": 0.7734375, "learning_rate": 1.0763626757293172e-05, "loss": 0.0918, "step": 38546 }, { "epoch": 28.23956043956044, "grad_norm": 1.25, "learning_rate": 1.0761583682485973e-05, "loss": 0.1049, "step": 38547 }, { "epoch": 28.24029304029304, "grad_norm": 1.234375, "learning_rate": 1.0759540798078556e-05, "loss": 0.1147, "step": 38548 }, { "epoch": 28.24102564102564, "grad_norm": 1.3359375, "learning_rate": 1.0757498104080742e-05, "loss": 0.1118, "step": 38549 }, { "epoch": 28.24175824175824, "grad_norm": 1.1015625, "learning_rate": 1.0755455600502288e-05, "loss": 0.1201, "step": 38550 }, { "epoch": 28.24249084249084, "grad_norm": 0.9921875, "learning_rate": 1.0753413287352995e-05, "loss": 0.1166, "step": 38551 }, { "epoch": 28.243223443223442, "grad_norm": 0.78515625, "learning_rate": 1.0751371164642641e-05, "loss": 0.1059, "step": 38552 }, { "epoch": 28.243956043956043, "grad_norm": 1.109375, "learning_rate": 1.0749329232381036e-05, "loss": 0.1668, "step": 38553 }, { "epoch": 28.244688644688644, "grad_norm": 1.0703125, "learning_rate": 1.0747287490577924e-05, "loss": 0.1289, "step": 38554 }, { "epoch": 28.245421245421245, "grad_norm": 1.1171875, "learning_rate": 1.0745245939243115e-05, "loss": 0.118, "step": 38555 }, { "epoch": 28.246153846153845, "grad_norm": 0.734375, "learning_rate": 1.0743204578386379e-05, "loss": 0.1262, "step": 38556 }, { "epoch": 28.246886446886446, "grad_norm": 1.09375, "learning_rate": 1.0741163408017502e-05, "loss": 0.1003, "step": 38557 }, { "epoch": 28.247619047619047, "grad_norm": 0.875, "learning_rate": 1.0739122428146275e-05, "loss": 0.1347, "step": 38558 }, { "epoch": 28.248351648351647, "grad_norm": 1.28125, "learning_rate": 1.0737081638782452e-05, "loss": 0.1236, "step": 38559 }, { "epoch": 28.249084249084248, "grad_norm": 1.171875, "learning_rate": 1.073504103993584e-05, "loss": 0.155, "step": 38560 }, { "epoch": 28.24981684981685, "grad_norm": 1.7109375, "learning_rate": 1.0733000631616176e-05, "loss": 0.1142, "step": 38561 }, { "epoch": 28.25054945054945, "grad_norm": 0.91015625, "learning_rate": 1.0730960413833283e-05, "loss": 0.139, "step": 38562 }, { "epoch": 28.25128205128205, "grad_norm": 1.078125, "learning_rate": 1.0728920386596899e-05, "loss": 0.1606, "step": 38563 }, { "epoch": 28.25201465201465, "grad_norm": 1.5859375, "learning_rate": 1.0726880549916831e-05, "loss": 0.1251, "step": 38564 }, { "epoch": 28.252747252747252, "grad_norm": 1.0546875, "learning_rate": 1.072484090380282e-05, "loss": 0.1069, "step": 38565 }, { "epoch": 28.253479853479853, "grad_norm": 1.28125, "learning_rate": 1.072280144826465e-05, "loss": 0.1191, "step": 38566 }, { "epoch": 28.254212454212453, "grad_norm": 1.2578125, "learning_rate": 1.07207621833121e-05, "loss": 0.097, "step": 38567 }, { "epoch": 28.254945054945054, "grad_norm": 1.328125, "learning_rate": 1.0718723108954952e-05, "loss": 0.1066, "step": 38568 }, { "epoch": 28.255677655677655, "grad_norm": 1.0625, "learning_rate": 1.0716684225202945e-05, "loss": 0.1084, "step": 38569 }, { "epoch": 28.256410256410255, "grad_norm": 1.390625, "learning_rate": 1.0714645532065864e-05, "loss": 0.1171, "step": 38570 }, { "epoch": 28.257142857142856, "grad_norm": 1.5234375, "learning_rate": 1.0712607029553488e-05, "loss": 0.143, "step": 38571 }, { "epoch": 28.257875457875457, "grad_norm": 0.859375, "learning_rate": 1.0710568717675552e-05, "loss": 0.0962, "step": 38572 }, { "epoch": 28.258608058608058, "grad_norm": 1.3125, "learning_rate": 1.0708530596441866e-05, "loss": 0.1343, "step": 38573 }, { "epoch": 28.25934065934066, "grad_norm": 1.0859375, "learning_rate": 1.0706492665862163e-05, "loss": 0.0852, "step": 38574 }, { "epoch": 28.26007326007326, "grad_norm": 0.80078125, "learning_rate": 1.0704454925946224e-05, "loss": 0.1256, "step": 38575 }, { "epoch": 28.26080586080586, "grad_norm": 1.0625, "learning_rate": 1.0702417376703798e-05, "loss": 0.1293, "step": 38576 }, { "epoch": 28.26153846153846, "grad_norm": 0.8125, "learning_rate": 1.070038001814465e-05, "loss": 0.1172, "step": 38577 }, { "epoch": 28.26227106227106, "grad_norm": 1.515625, "learning_rate": 1.069834285027855e-05, "loss": 0.1308, "step": 38578 }, { "epoch": 28.263003663003662, "grad_norm": 0.9921875, "learning_rate": 1.0696305873115268e-05, "loss": 0.1168, "step": 38579 }, { "epoch": 28.263736263736263, "grad_norm": 0.546875, "learning_rate": 1.0694269086664537e-05, "loss": 0.106, "step": 38580 }, { "epoch": 28.264468864468864, "grad_norm": 1.0390625, "learning_rate": 1.0692232490936136e-05, "loss": 0.1088, "step": 38581 }, { "epoch": 28.265201465201464, "grad_norm": 1.078125, "learning_rate": 1.069019608593981e-05, "loss": 0.0885, "step": 38582 }, { "epoch": 28.265934065934065, "grad_norm": 0.7734375, "learning_rate": 1.0688159871685324e-05, "loss": 0.0966, "step": 38583 }, { "epoch": 28.266666666666666, "grad_norm": 0.7265625, "learning_rate": 1.068612384818245e-05, "loss": 0.1047, "step": 38584 }, { "epoch": 28.267399267399266, "grad_norm": 1.03125, "learning_rate": 1.0684088015440915e-05, "loss": 0.1249, "step": 38585 }, { "epoch": 28.268131868131867, "grad_norm": 1.703125, "learning_rate": 1.0682052373470493e-05, "loss": 0.1722, "step": 38586 }, { "epoch": 28.268864468864468, "grad_norm": 1.03125, "learning_rate": 1.0680016922280906e-05, "loss": 0.1025, "step": 38587 }, { "epoch": 28.26959706959707, "grad_norm": 0.86328125, "learning_rate": 1.0677981661881953e-05, "loss": 0.148, "step": 38588 }, { "epoch": 28.27032967032967, "grad_norm": 0.73828125, "learning_rate": 1.0675946592283351e-05, "loss": 0.1155, "step": 38589 }, { "epoch": 28.27106227106227, "grad_norm": 0.99609375, "learning_rate": 1.067391171349488e-05, "loss": 0.1153, "step": 38590 }, { "epoch": 28.27179487179487, "grad_norm": 0.97265625, "learning_rate": 1.0671877025526253e-05, "loss": 0.1045, "step": 38591 }, { "epoch": 28.27252747252747, "grad_norm": 0.73828125, "learning_rate": 1.0669842528387242e-05, "loss": 0.1112, "step": 38592 }, { "epoch": 28.273260073260072, "grad_norm": 0.9453125, "learning_rate": 1.0667808222087585e-05, "loss": 0.1356, "step": 38593 }, { "epoch": 28.273992673992673, "grad_norm": 1.1171875, "learning_rate": 1.0665774106637055e-05, "loss": 0.1002, "step": 38594 }, { "epoch": 28.274725274725274, "grad_norm": 0.9765625, "learning_rate": 1.066374018204536e-05, "loss": 0.1211, "step": 38595 }, { "epoch": 28.275457875457874, "grad_norm": 0.90625, "learning_rate": 1.0661706448322265e-05, "loss": 0.1163, "step": 38596 }, { "epoch": 28.276190476190475, "grad_norm": 0.91015625, "learning_rate": 1.0659672905477525e-05, "loss": 0.1108, "step": 38597 }, { "epoch": 28.276923076923076, "grad_norm": 1.4140625, "learning_rate": 1.0657639553520845e-05, "loss": 0.129, "step": 38598 }, { "epoch": 28.277655677655677, "grad_norm": 0.9453125, "learning_rate": 1.0655606392462018e-05, "loss": 0.0972, "step": 38599 }, { "epoch": 28.278388278388277, "grad_norm": 0.61328125, "learning_rate": 1.0653573422310755e-05, "loss": 0.1224, "step": 38600 }, { "epoch": 28.279120879120878, "grad_norm": 0.84375, "learning_rate": 1.065154064307681e-05, "loss": 0.1039, "step": 38601 }, { "epoch": 28.27985347985348, "grad_norm": 1.6875, "learning_rate": 1.0649508054769896e-05, "loss": 0.1404, "step": 38602 }, { "epoch": 28.28058608058608, "grad_norm": 0.9765625, "learning_rate": 1.064747565739978e-05, "loss": 0.1256, "step": 38603 }, { "epoch": 28.28131868131868, "grad_norm": 0.94140625, "learning_rate": 1.064544345097619e-05, "loss": 0.1199, "step": 38604 }, { "epoch": 28.28205128205128, "grad_norm": 1.546875, "learning_rate": 1.0643411435508875e-05, "loss": 0.1295, "step": 38605 }, { "epoch": 28.282783882783882, "grad_norm": 0.94921875, "learning_rate": 1.0641379611007546e-05, "loss": 0.1141, "step": 38606 }, { "epoch": 28.283516483516483, "grad_norm": 0.66015625, "learning_rate": 1.0639347977481949e-05, "loss": 0.1077, "step": 38607 }, { "epoch": 28.284249084249083, "grad_norm": 1.9375, "learning_rate": 1.0637316534941826e-05, "loss": 0.1459, "step": 38608 }, { "epoch": 28.284981684981684, "grad_norm": 1.6640625, "learning_rate": 1.0635285283396904e-05, "loss": 0.1404, "step": 38609 }, { "epoch": 28.285714285714285, "grad_norm": 0.8828125, "learning_rate": 1.0633254222856933e-05, "loss": 0.1274, "step": 38610 }, { "epoch": 28.286446886446885, "grad_norm": 1.8984375, "learning_rate": 1.0631223353331612e-05, "loss": 0.1302, "step": 38611 }, { "epoch": 28.287179487179486, "grad_norm": 1.2421875, "learning_rate": 1.06291926748307e-05, "loss": 0.1065, "step": 38612 }, { "epoch": 28.287912087912087, "grad_norm": 1.1171875, "learning_rate": 1.0627162187363891e-05, "loss": 0.1433, "step": 38613 }, { "epoch": 28.288644688644688, "grad_norm": 1.9453125, "learning_rate": 1.0625131890940964e-05, "loss": 0.1227, "step": 38614 }, { "epoch": 28.28937728937729, "grad_norm": 1.5703125, "learning_rate": 1.0623101785571607e-05, "loss": 0.1382, "step": 38615 }, { "epoch": 28.29010989010989, "grad_norm": 1.4609375, "learning_rate": 1.0621071871265572e-05, "loss": 0.1267, "step": 38616 }, { "epoch": 28.29084249084249, "grad_norm": 1.1015625, "learning_rate": 1.0619042148032558e-05, "loss": 0.1536, "step": 38617 }, { "epoch": 28.29157509157509, "grad_norm": 1.0546875, "learning_rate": 1.061701261588231e-05, "loss": 0.1322, "step": 38618 }, { "epoch": 28.29230769230769, "grad_norm": 0.7109375, "learning_rate": 1.0614983274824543e-05, "loss": 0.1189, "step": 38619 }, { "epoch": 28.293040293040292, "grad_norm": 1.2890625, "learning_rate": 1.0612954124868985e-05, "loss": 0.1123, "step": 38620 }, { "epoch": 28.293772893772893, "grad_norm": 0.93359375, "learning_rate": 1.061092516602537e-05, "loss": 0.1402, "step": 38621 }, { "epoch": 28.294505494505493, "grad_norm": 0.90234375, "learning_rate": 1.0608896398303395e-05, "loss": 0.1177, "step": 38622 }, { "epoch": 28.295238095238094, "grad_norm": 1.734375, "learning_rate": 1.0606867821712806e-05, "loss": 0.1315, "step": 38623 }, { "epoch": 28.295970695970695, "grad_norm": 1.75, "learning_rate": 1.0604839436263287e-05, "loss": 0.1296, "step": 38624 }, { "epoch": 28.296703296703296, "grad_norm": 1.2265625, "learning_rate": 1.06028112419646e-05, "loss": 0.1688, "step": 38625 }, { "epoch": 28.297435897435896, "grad_norm": 1.2265625, "learning_rate": 1.060078323882643e-05, "loss": 0.1162, "step": 38626 }, { "epoch": 28.298168498168497, "grad_norm": 1.6796875, "learning_rate": 1.0598755426858523e-05, "loss": 0.1298, "step": 38627 }, { "epoch": 28.298901098901098, "grad_norm": 0.625, "learning_rate": 1.059672780607056e-05, "loss": 0.1147, "step": 38628 }, { "epoch": 28.2996336996337, "grad_norm": 0.7890625, "learning_rate": 1.0594700376472273e-05, "loss": 0.1064, "step": 38629 }, { "epoch": 28.3003663003663, "grad_norm": 0.90234375, "learning_rate": 1.059267313807338e-05, "loss": 0.1284, "step": 38630 }, { "epoch": 28.3010989010989, "grad_norm": 0.80859375, "learning_rate": 1.0590646090883602e-05, "loss": 0.1398, "step": 38631 }, { "epoch": 28.3018315018315, "grad_norm": 1.3515625, "learning_rate": 1.0588619234912622e-05, "loss": 0.111, "step": 38632 }, { "epoch": 28.3025641025641, "grad_norm": 0.87109375, "learning_rate": 1.0586592570170179e-05, "loss": 0.116, "step": 38633 }, { "epoch": 28.303296703296702, "grad_norm": 0.7734375, "learning_rate": 1.0584566096665967e-05, "loss": 0.1157, "step": 38634 }, { "epoch": 28.304029304029303, "grad_norm": 1.03125, "learning_rate": 1.0582539814409701e-05, "loss": 0.1607, "step": 38635 }, { "epoch": 28.304761904761904, "grad_norm": 0.97265625, "learning_rate": 1.058051372341111e-05, "loss": 0.1399, "step": 38636 }, { "epoch": 28.305494505494504, "grad_norm": 0.8203125, "learning_rate": 1.0578487823679866e-05, "loss": 0.0945, "step": 38637 }, { "epoch": 28.306227106227105, "grad_norm": 0.66796875, "learning_rate": 1.0576462115225706e-05, "loss": 0.0992, "step": 38638 }, { "epoch": 28.306959706959706, "grad_norm": 1.5859375, "learning_rate": 1.057443659805829e-05, "loss": 0.1184, "step": 38639 }, { "epoch": 28.307692307692307, "grad_norm": 0.9765625, "learning_rate": 1.0572411272187387e-05, "loss": 0.1347, "step": 38640 }, { "epoch": 28.308424908424907, "grad_norm": 0.94140625, "learning_rate": 1.0570386137622659e-05, "loss": 0.1124, "step": 38641 }, { "epoch": 28.309157509157508, "grad_norm": 0.63671875, "learning_rate": 1.0568361194373824e-05, "loss": 0.1111, "step": 38642 }, { "epoch": 28.30989010989011, "grad_norm": 1.0, "learning_rate": 1.0566336442450573e-05, "loss": 0.1246, "step": 38643 }, { "epoch": 28.31062271062271, "grad_norm": 1.1328125, "learning_rate": 1.056431188186261e-05, "loss": 0.105, "step": 38644 }, { "epoch": 28.31135531135531, "grad_norm": 0.93359375, "learning_rate": 1.056228751261964e-05, "loss": 0.1444, "step": 38645 }, { "epoch": 28.31208791208791, "grad_norm": 1.40625, "learning_rate": 1.056026333473136e-05, "loss": 0.1353, "step": 38646 }, { "epoch": 28.31282051282051, "grad_norm": 1.40625, "learning_rate": 1.0558239348207487e-05, "loss": 0.1443, "step": 38647 }, { "epoch": 28.313553113553112, "grad_norm": 0.79296875, "learning_rate": 1.0556215553057681e-05, "loss": 0.126, "step": 38648 }, { "epoch": 28.314285714285713, "grad_norm": 0.84765625, "learning_rate": 1.055419194929168e-05, "loss": 0.0998, "step": 38649 }, { "epoch": 28.315018315018314, "grad_norm": 0.84375, "learning_rate": 1.0552168536919126e-05, "loss": 0.1063, "step": 38650 }, { "epoch": 28.315750915750915, "grad_norm": 0.78515625, "learning_rate": 1.0550145315949777e-05, "loss": 0.1065, "step": 38651 }, { "epoch": 28.316483516483515, "grad_norm": 0.7734375, "learning_rate": 1.0548122286393281e-05, "loss": 0.1548, "step": 38652 }, { "epoch": 28.317216117216116, "grad_norm": 1.671875, "learning_rate": 1.054609944825936e-05, "loss": 0.1301, "step": 38653 }, { "epoch": 28.317948717948717, "grad_norm": 1.0859375, "learning_rate": 1.054407680155768e-05, "loss": 0.1001, "step": 38654 }, { "epoch": 28.318681318681318, "grad_norm": 1.109375, "learning_rate": 1.0542054346297943e-05, "loss": 0.1304, "step": 38655 }, { "epoch": 28.31941391941392, "grad_norm": 0.94921875, "learning_rate": 1.0540032082489842e-05, "loss": 0.1321, "step": 38656 }, { "epoch": 28.32014652014652, "grad_norm": 1.2578125, "learning_rate": 1.0538010010143081e-05, "loss": 0.1265, "step": 38657 }, { "epoch": 28.32087912087912, "grad_norm": 1.0859375, "learning_rate": 1.0535988129267313e-05, "loss": 0.1101, "step": 38658 }, { "epoch": 28.32161172161172, "grad_norm": 0.76953125, "learning_rate": 1.0533966439872253e-05, "loss": 0.1335, "step": 38659 }, { "epoch": 28.32234432234432, "grad_norm": 0.984375, "learning_rate": 1.0531944941967579e-05, "loss": 0.1059, "step": 38660 }, { "epoch": 28.323076923076922, "grad_norm": 1.515625, "learning_rate": 1.0529923635562971e-05, "loss": 0.1288, "step": 38661 }, { "epoch": 28.323809523809523, "grad_norm": 1.0078125, "learning_rate": 1.0527902520668139e-05, "loss": 0.1019, "step": 38662 }, { "epoch": 28.324542124542123, "grad_norm": 1.015625, "learning_rate": 1.0525881597292732e-05, "loss": 0.1047, "step": 38663 }, { "epoch": 28.325274725274724, "grad_norm": 1.1484375, "learning_rate": 1.0523860865446464e-05, "loss": 0.1284, "step": 38664 }, { "epoch": 28.326007326007325, "grad_norm": 0.8515625, "learning_rate": 1.0521840325138971e-05, "loss": 0.1097, "step": 38665 }, { "epoch": 28.326739926739926, "grad_norm": 0.85546875, "learning_rate": 1.0519819976379997e-05, "loss": 0.1228, "step": 38666 }, { "epoch": 28.327472527472526, "grad_norm": 1.5078125, "learning_rate": 1.0517799819179174e-05, "loss": 0.1254, "step": 38667 }, { "epoch": 28.328205128205127, "grad_norm": 1.0390625, "learning_rate": 1.0515779853546212e-05, "loss": 0.1073, "step": 38668 }, { "epoch": 28.328937728937728, "grad_norm": 0.8203125, "learning_rate": 1.0513760079490759e-05, "loss": 0.1331, "step": 38669 }, { "epoch": 28.32967032967033, "grad_norm": 0.77734375, "learning_rate": 1.0511740497022514e-05, "loss": 0.1229, "step": 38670 }, { "epoch": 28.33040293040293, "grad_norm": 0.73046875, "learning_rate": 1.050972110615114e-05, "loss": 0.0959, "step": 38671 }, { "epoch": 28.33113553113553, "grad_norm": 1.0078125, "learning_rate": 1.0507701906886327e-05, "loss": 0.1343, "step": 38672 }, { "epoch": 28.33186813186813, "grad_norm": 0.98046875, "learning_rate": 1.050568289923775e-05, "loss": 0.1212, "step": 38673 }, { "epoch": 28.33260073260073, "grad_norm": 1.359375, "learning_rate": 1.0503664083215067e-05, "loss": 0.1265, "step": 38674 }, { "epoch": 28.333333333333332, "grad_norm": 1.3984375, "learning_rate": 1.0501645458827967e-05, "loss": 0.1377, "step": 38675 }, { "epoch": 28.334065934065933, "grad_norm": 1.5546875, "learning_rate": 1.0499627026086093e-05, "loss": 0.0978, "step": 38676 }, { "epoch": 28.334798534798534, "grad_norm": 1.1796875, "learning_rate": 1.0497608784999164e-05, "loss": 0.1187, "step": 38677 }, { "epoch": 28.335531135531134, "grad_norm": 0.953125, "learning_rate": 1.0495590735576804e-05, "loss": 0.1139, "step": 38678 }, { "epoch": 28.336263736263735, "grad_norm": 1.15625, "learning_rate": 1.0493572877828723e-05, "loss": 0.1109, "step": 38679 }, { "epoch": 28.336996336996336, "grad_norm": 0.7734375, "learning_rate": 1.0491555211764549e-05, "loss": 0.1247, "step": 38680 }, { "epoch": 28.337728937728937, "grad_norm": 1.640625, "learning_rate": 1.0489537737393968e-05, "loss": 0.1301, "step": 38681 }, { "epoch": 28.338461538461537, "grad_norm": 0.69921875, "learning_rate": 1.0487520454726648e-05, "loss": 0.1078, "step": 38682 }, { "epoch": 28.339194139194138, "grad_norm": 1.6328125, "learning_rate": 1.0485503363772266e-05, "loss": 0.1297, "step": 38683 }, { "epoch": 28.33992673992674, "grad_norm": 0.984375, "learning_rate": 1.0483486464540457e-05, "loss": 0.1115, "step": 38684 }, { "epoch": 28.34065934065934, "grad_norm": 0.6640625, "learning_rate": 1.04814697570409e-05, "loss": 0.1048, "step": 38685 }, { "epoch": 28.34139194139194, "grad_norm": 0.7734375, "learning_rate": 1.0479453241283264e-05, "loss": 0.1123, "step": 38686 }, { "epoch": 28.34212454212454, "grad_norm": 0.859375, "learning_rate": 1.0477436917277202e-05, "loss": 0.1276, "step": 38687 }, { "epoch": 28.34285714285714, "grad_norm": 0.78515625, "learning_rate": 1.0475420785032395e-05, "loss": 0.0949, "step": 38688 }, { "epoch": 28.343589743589742, "grad_norm": 1.5703125, "learning_rate": 1.0473404844558467e-05, "loss": 0.1137, "step": 38689 }, { "epoch": 28.344322344322343, "grad_norm": 1.1875, "learning_rate": 1.0471389095865115e-05, "loss": 0.1098, "step": 38690 }, { "epoch": 28.345054945054944, "grad_norm": 1.265625, "learning_rate": 1.0469373538961951e-05, "loss": 0.1165, "step": 38691 }, { "epoch": 28.345787545787545, "grad_norm": 0.8828125, "learning_rate": 1.0467358173858682e-05, "loss": 0.1201, "step": 38692 }, { "epoch": 28.346520146520145, "grad_norm": 0.91796875, "learning_rate": 1.0465343000564931e-05, "loss": 0.1337, "step": 38693 }, { "epoch": 28.347252747252746, "grad_norm": 0.98046875, "learning_rate": 1.0463328019090378e-05, "loss": 0.1168, "step": 38694 }, { "epoch": 28.347985347985347, "grad_norm": 0.77734375, "learning_rate": 1.0461313229444652e-05, "loss": 0.0973, "step": 38695 }, { "epoch": 28.348717948717947, "grad_norm": 1.3203125, "learning_rate": 1.0459298631637413e-05, "loss": 0.1234, "step": 38696 }, { "epoch": 28.349450549450548, "grad_norm": 1.5859375, "learning_rate": 1.0457284225678326e-05, "loss": 0.1334, "step": 38697 }, { "epoch": 28.35018315018315, "grad_norm": 1.25, "learning_rate": 1.0455270011577032e-05, "loss": 0.1104, "step": 38698 }, { "epoch": 28.35091575091575, "grad_norm": 1.0703125, "learning_rate": 1.0453255989343197e-05, "loss": 0.1351, "step": 38699 }, { "epoch": 28.35164835164835, "grad_norm": 1.125, "learning_rate": 1.0451242158986447e-05, "loss": 0.1007, "step": 38700 }, { "epoch": 28.35238095238095, "grad_norm": 1.34375, "learning_rate": 1.0449228520516459e-05, "loss": 0.107, "step": 38701 }, { "epoch": 28.353113553113552, "grad_norm": 1.125, "learning_rate": 1.0447215073942834e-05, "loss": 0.1418, "step": 38702 }, { "epoch": 28.353846153846153, "grad_norm": 1.109375, "learning_rate": 1.044520181927528e-05, "loss": 0.107, "step": 38703 }, { "epoch": 28.354578754578753, "grad_norm": 1.34375, "learning_rate": 1.0443188756523395e-05, "loss": 0.1419, "step": 38704 }, { "epoch": 28.355311355311354, "grad_norm": 0.84375, "learning_rate": 1.0441175885696861e-05, "loss": 0.1177, "step": 38705 }, { "epoch": 28.356043956043955, "grad_norm": 1.421875, "learning_rate": 1.0439163206805287e-05, "loss": 0.1445, "step": 38706 }, { "epoch": 28.356776556776556, "grad_norm": 0.82421875, "learning_rate": 1.0437150719858335e-05, "loss": 0.1486, "step": 38707 }, { "epoch": 28.357509157509156, "grad_norm": 1.0390625, "learning_rate": 1.0435138424865645e-05, "loss": 0.1085, "step": 38708 }, { "epoch": 28.358241758241757, "grad_norm": 1.28125, "learning_rate": 1.0433126321836871e-05, "loss": 0.118, "step": 38709 }, { "epoch": 28.358974358974358, "grad_norm": 1.5859375, "learning_rate": 1.043111441078163e-05, "loss": 0.1073, "step": 38710 }, { "epoch": 28.35970695970696, "grad_norm": 1.484375, "learning_rate": 1.0429102691709577e-05, "loss": 0.1207, "step": 38711 }, { "epoch": 28.36043956043956, "grad_norm": 0.9140625, "learning_rate": 1.0427091164630346e-05, "loss": 0.1347, "step": 38712 }, { "epoch": 28.36117216117216, "grad_norm": 0.96484375, "learning_rate": 1.0425079829553572e-05, "loss": 0.1327, "step": 38713 }, { "epoch": 28.36190476190476, "grad_norm": 2.0625, "learning_rate": 1.0423068686488912e-05, "loss": 0.1575, "step": 38714 }, { "epoch": 28.36263736263736, "grad_norm": 1.0078125, "learning_rate": 1.0421057735445974e-05, "loss": 0.1212, "step": 38715 }, { "epoch": 28.363369963369962, "grad_norm": 1.15625, "learning_rate": 1.0419046976434418e-05, "loss": 0.0976, "step": 38716 }, { "epoch": 28.364102564102563, "grad_norm": 0.91015625, "learning_rate": 1.0417036409463854e-05, "loss": 0.1233, "step": 38717 }, { "epoch": 28.364835164835164, "grad_norm": 0.8984375, "learning_rate": 1.0415026034543925e-05, "loss": 0.0936, "step": 38718 }, { "epoch": 28.365567765567764, "grad_norm": 1.578125, "learning_rate": 1.0413015851684264e-05, "loss": 0.1188, "step": 38719 }, { "epoch": 28.366300366300365, "grad_norm": 1.140625, "learning_rate": 1.0411005860894516e-05, "loss": 0.0923, "step": 38720 }, { "epoch": 28.367032967032966, "grad_norm": 1.078125, "learning_rate": 1.0408996062184292e-05, "loss": 0.1209, "step": 38721 }, { "epoch": 28.367765567765566, "grad_norm": 1.125, "learning_rate": 1.0406986455563224e-05, "loss": 0.1126, "step": 38722 }, { "epoch": 28.368498168498167, "grad_norm": 0.8671875, "learning_rate": 1.0404977041040948e-05, "loss": 0.1107, "step": 38723 }, { "epoch": 28.369230769230768, "grad_norm": 1.6171875, "learning_rate": 1.0402967818627087e-05, "loss": 0.1158, "step": 38724 }, { "epoch": 28.36996336996337, "grad_norm": 0.82421875, "learning_rate": 1.0400958788331288e-05, "loss": 0.1059, "step": 38725 }, { "epoch": 28.37069597069597, "grad_norm": 0.89453125, "learning_rate": 1.039894995016314e-05, "loss": 0.1074, "step": 38726 }, { "epoch": 28.37142857142857, "grad_norm": 1.328125, "learning_rate": 1.0396941304132302e-05, "loss": 0.1001, "step": 38727 }, { "epoch": 28.37216117216117, "grad_norm": 1.375, "learning_rate": 1.0394932850248359e-05, "loss": 0.1295, "step": 38728 }, { "epoch": 28.37289377289377, "grad_norm": 1.125, "learning_rate": 1.0392924588520986e-05, "loss": 0.1502, "step": 38729 }, { "epoch": 28.373626373626372, "grad_norm": 1.546875, "learning_rate": 1.039091651895976e-05, "loss": 0.1561, "step": 38730 }, { "epoch": 28.374358974358973, "grad_norm": 0.61328125, "learning_rate": 1.0388908641574333e-05, "loss": 0.1007, "step": 38731 }, { "epoch": 28.375091575091574, "grad_norm": 1.8203125, "learning_rate": 1.0386900956374303e-05, "loss": 0.1623, "step": 38732 }, { "epoch": 28.375824175824174, "grad_norm": 1.15625, "learning_rate": 1.0384893463369298e-05, "loss": 0.1508, "step": 38733 }, { "epoch": 28.376556776556775, "grad_norm": 0.81640625, "learning_rate": 1.0382886162568937e-05, "loss": 0.1134, "step": 38734 }, { "epoch": 28.377289377289376, "grad_norm": 0.80859375, "learning_rate": 1.0380879053982848e-05, "loss": 0.1118, "step": 38735 }, { "epoch": 28.378021978021977, "grad_norm": 1.1875, "learning_rate": 1.0378872137620631e-05, "loss": 0.1382, "step": 38736 }, { "epoch": 28.378754578754577, "grad_norm": 1.0859375, "learning_rate": 1.0376865413491905e-05, "loss": 0.1318, "step": 38737 }, { "epoch": 28.379487179487178, "grad_norm": 1.421875, "learning_rate": 1.0374858881606286e-05, "loss": 0.1099, "step": 38738 }, { "epoch": 28.38021978021978, "grad_norm": 0.94921875, "learning_rate": 1.0372852541973392e-05, "loss": 0.1423, "step": 38739 }, { "epoch": 28.38095238095238, "grad_norm": 1.2578125, "learning_rate": 1.0370846394602845e-05, "loss": 0.1449, "step": 38740 }, { "epoch": 28.38168498168498, "grad_norm": 1.1640625, "learning_rate": 1.0368840439504235e-05, "loss": 0.1102, "step": 38741 }, { "epoch": 28.38241758241758, "grad_norm": 1.109375, "learning_rate": 1.0366834676687202e-05, "loss": 0.1373, "step": 38742 }, { "epoch": 28.383150183150182, "grad_norm": 0.53125, "learning_rate": 1.036482910616132e-05, "loss": 0.1141, "step": 38743 }, { "epoch": 28.383882783882783, "grad_norm": 1.4140625, "learning_rate": 1.036282372793622e-05, "loss": 0.1046, "step": 38744 }, { "epoch": 28.384615384615383, "grad_norm": 0.97265625, "learning_rate": 1.0360818542021508e-05, "loss": 0.0988, "step": 38745 }, { "epoch": 28.385347985347984, "grad_norm": 0.890625, "learning_rate": 1.0358813548426805e-05, "loss": 0.1298, "step": 38746 }, { "epoch": 28.386080586080585, "grad_norm": 0.828125, "learning_rate": 1.0356808747161689e-05, "loss": 0.1283, "step": 38747 }, { "epoch": 28.386813186813185, "grad_norm": 0.5546875, "learning_rate": 1.0354804138235784e-05, "loss": 0.1002, "step": 38748 }, { "epoch": 28.387545787545786, "grad_norm": 1.21875, "learning_rate": 1.0352799721658687e-05, "loss": 0.1356, "step": 38749 }, { "epoch": 28.388278388278387, "grad_norm": 1.2734375, "learning_rate": 1.0350795497440008e-05, "loss": 0.1084, "step": 38750 }, { "epoch": 28.389010989010988, "grad_norm": 1.421875, "learning_rate": 1.0348791465589356e-05, "loss": 0.1147, "step": 38751 }, { "epoch": 28.38974358974359, "grad_norm": 1.0703125, "learning_rate": 1.0346787626116316e-05, "loss": 0.1149, "step": 38752 }, { "epoch": 28.39047619047619, "grad_norm": 0.77734375, "learning_rate": 1.0344783979030511e-05, "loss": 0.1151, "step": 38753 }, { "epoch": 28.39120879120879, "grad_norm": 1.671875, "learning_rate": 1.0342780524341505e-05, "loss": 0.1223, "step": 38754 }, { "epoch": 28.39194139194139, "grad_norm": 1.125, "learning_rate": 1.0340777262058939e-05, "loss": 0.1351, "step": 38755 }, { "epoch": 28.39267399267399, "grad_norm": 0.62109375, "learning_rate": 1.0338774192192384e-05, "loss": 0.0995, "step": 38756 }, { "epoch": 28.393406593406592, "grad_norm": 1.3984375, "learning_rate": 1.0336771314751458e-05, "loss": 0.1311, "step": 38757 }, { "epoch": 28.394139194139193, "grad_norm": 0.92578125, "learning_rate": 1.0334768629745737e-05, "loss": 0.1222, "step": 38758 }, { "epoch": 28.394871794871793, "grad_norm": 0.8046875, "learning_rate": 1.0332766137184818e-05, "loss": 0.1368, "step": 38759 }, { "epoch": 28.395604395604394, "grad_norm": 0.71875, "learning_rate": 1.0330763837078308e-05, "loss": 0.1012, "step": 38760 }, { "epoch": 28.396336996336995, "grad_norm": 1.2109375, "learning_rate": 1.0328761729435792e-05, "loss": 0.1441, "step": 38761 }, { "epoch": 28.397069597069596, "grad_norm": 0.83203125, "learning_rate": 1.0326759814266877e-05, "loss": 0.1372, "step": 38762 }, { "epoch": 28.397802197802196, "grad_norm": 0.68359375, "learning_rate": 1.0324758091581136e-05, "loss": 0.1021, "step": 38763 }, { "epoch": 28.398534798534797, "grad_norm": 0.87890625, "learning_rate": 1.0322756561388164e-05, "loss": 0.113, "step": 38764 }, { "epoch": 28.399267399267398, "grad_norm": 1.3984375, "learning_rate": 1.0320755223697556e-05, "loss": 0.1316, "step": 38765 }, { "epoch": 28.4, "grad_norm": 1.28125, "learning_rate": 1.0318754078518912e-05, "loss": 0.1183, "step": 38766 }, { "epoch": 28.4007326007326, "grad_norm": 0.80859375, "learning_rate": 1.0316753125861794e-05, "loss": 0.1096, "step": 38767 }, { "epoch": 28.4014652014652, "grad_norm": 1.21875, "learning_rate": 1.0314752365735818e-05, "loss": 0.1343, "step": 38768 }, { "epoch": 28.4021978021978, "grad_norm": 1.09375, "learning_rate": 1.0312751798150542e-05, "loss": 0.1067, "step": 38769 }, { "epoch": 28.4029304029304, "grad_norm": 1.4140625, "learning_rate": 1.0310751423115562e-05, "loss": 0.1885, "step": 38770 }, { "epoch": 28.403663003663002, "grad_norm": 1.5546875, "learning_rate": 1.0308751240640466e-05, "loss": 0.1423, "step": 38771 }, { "epoch": 28.404395604395603, "grad_norm": 0.9765625, "learning_rate": 1.0306751250734856e-05, "loss": 0.1155, "step": 38772 }, { "epoch": 28.405128205128204, "grad_norm": 1.125, "learning_rate": 1.0304751453408272e-05, "loss": 0.1093, "step": 38773 }, { "epoch": 28.405860805860804, "grad_norm": 0.796875, "learning_rate": 1.0302751848670326e-05, "loss": 0.1039, "step": 38774 }, { "epoch": 28.406593406593405, "grad_norm": 0.9296875, "learning_rate": 1.0300752436530591e-05, "loss": 0.1156, "step": 38775 }, { "epoch": 28.407326007326006, "grad_norm": 0.91015625, "learning_rate": 1.0298753216998649e-05, "loss": 0.116, "step": 38776 }, { "epoch": 28.408058608058607, "grad_norm": 0.8125, "learning_rate": 1.0296754190084085e-05, "loss": 0.1163, "step": 38777 }, { "epoch": 28.408791208791207, "grad_norm": 0.7734375, "learning_rate": 1.0294755355796463e-05, "loss": 0.105, "step": 38778 }, { "epoch": 28.409523809523808, "grad_norm": 1.09375, "learning_rate": 1.0292756714145374e-05, "loss": 0.1086, "step": 38779 }, { "epoch": 28.41025641025641, "grad_norm": 1.2265625, "learning_rate": 1.0290758265140362e-05, "loss": 0.1614, "step": 38780 }, { "epoch": 28.41098901098901, "grad_norm": 0.765625, "learning_rate": 1.0288760008791056e-05, "loss": 0.1223, "step": 38781 }, { "epoch": 28.41172161172161, "grad_norm": 1.1015625, "learning_rate": 1.0286761945106989e-05, "loss": 0.1309, "step": 38782 }, { "epoch": 28.41245421245421, "grad_norm": 1.40625, "learning_rate": 1.0284764074097756e-05, "loss": 0.1212, "step": 38783 }, { "epoch": 28.41318681318681, "grad_norm": 0.859375, "learning_rate": 1.0282766395772907e-05, "loss": 0.1124, "step": 38784 }, { "epoch": 28.413919413919412, "grad_norm": 1.328125, "learning_rate": 1.0280768910142029e-05, "loss": 0.1301, "step": 38785 }, { "epoch": 28.414652014652013, "grad_norm": 1.2734375, "learning_rate": 1.0278771617214692e-05, "loss": 0.1578, "step": 38786 }, { "epoch": 28.415384615384614, "grad_norm": 0.87109375, "learning_rate": 1.0276774517000454e-05, "loss": 0.1249, "step": 38787 }, { "epoch": 28.416117216117215, "grad_norm": 1.0546875, "learning_rate": 1.027477760950892e-05, "loss": 0.1575, "step": 38788 }, { "epoch": 28.416849816849815, "grad_norm": 1.1953125, "learning_rate": 1.0272780894749608e-05, "loss": 0.1282, "step": 38789 }, { "epoch": 28.417582417582416, "grad_norm": 1.0, "learning_rate": 1.0270784372732115e-05, "loss": 0.1008, "step": 38790 }, { "epoch": 28.418315018315017, "grad_norm": 0.91796875, "learning_rate": 1.0268788043465997e-05, "loss": 0.1528, "step": 38791 }, { "epoch": 28.419047619047618, "grad_norm": 0.953125, "learning_rate": 1.026679190696084e-05, "loss": 0.1059, "step": 38792 }, { "epoch": 28.41978021978022, "grad_norm": 0.79296875, "learning_rate": 1.0264795963226172e-05, "loss": 0.0981, "step": 38793 }, { "epoch": 28.42051282051282, "grad_norm": 1.0625, "learning_rate": 1.0262800212271592e-05, "loss": 0.1344, "step": 38794 }, { "epoch": 28.42124542124542, "grad_norm": 1.3515625, "learning_rate": 1.0260804654106632e-05, "loss": 0.1086, "step": 38795 }, { "epoch": 28.42197802197802, "grad_norm": 0.73046875, "learning_rate": 1.0258809288740866e-05, "loss": 0.1354, "step": 38796 }, { "epoch": 28.42271062271062, "grad_norm": 1.046875, "learning_rate": 1.0256814116183853e-05, "loss": 0.1176, "step": 38797 }, { "epoch": 28.423443223443222, "grad_norm": 0.8515625, "learning_rate": 1.0254819136445169e-05, "loss": 0.1134, "step": 38798 }, { "epoch": 28.424175824175823, "grad_norm": 1.0703125, "learning_rate": 1.0252824349534346e-05, "loss": 0.1064, "step": 38799 }, { "epoch": 28.424908424908423, "grad_norm": 0.92578125, "learning_rate": 1.0250829755460952e-05, "loss": 0.1124, "step": 38800 }, { "epoch": 28.425641025641024, "grad_norm": 0.8515625, "learning_rate": 1.0248835354234544e-05, "loss": 0.1068, "step": 38801 }, { "epoch": 28.426373626373625, "grad_norm": 0.8125, "learning_rate": 1.024684114586468e-05, "loss": 0.1171, "step": 38802 }, { "epoch": 28.427106227106226, "grad_norm": 0.9140625, "learning_rate": 1.0244847130360933e-05, "loss": 0.1114, "step": 38803 }, { "epoch": 28.427838827838826, "grad_norm": 1.2109375, "learning_rate": 1.024285330773282e-05, "loss": 0.0994, "step": 38804 }, { "epoch": 28.428571428571427, "grad_norm": 0.9765625, "learning_rate": 1.0240859677989926e-05, "loss": 0.118, "step": 38805 }, { "epoch": 28.429304029304028, "grad_norm": 1.546875, "learning_rate": 1.0238866241141766e-05, "loss": 0.0983, "step": 38806 }, { "epoch": 28.43003663003663, "grad_norm": 1.2109375, "learning_rate": 1.0236872997197938e-05, "loss": 0.1176, "step": 38807 }, { "epoch": 28.43076923076923, "grad_norm": 1.1171875, "learning_rate": 1.0234879946167956e-05, "loss": 0.16, "step": 38808 }, { "epoch": 28.43150183150183, "grad_norm": 0.94921875, "learning_rate": 1.0232887088061399e-05, "loss": 0.1404, "step": 38809 }, { "epoch": 28.43223443223443, "grad_norm": 1.1171875, "learning_rate": 1.0230894422887778e-05, "loss": 0.1406, "step": 38810 }, { "epoch": 28.43296703296703, "grad_norm": 1.0625, "learning_rate": 1.0228901950656668e-05, "loss": 0.125, "step": 38811 }, { "epoch": 28.433699633699632, "grad_norm": 0.93359375, "learning_rate": 1.0226909671377606e-05, "loss": 0.1111, "step": 38812 }, { "epoch": 28.434432234432233, "grad_norm": 0.88671875, "learning_rate": 1.0224917585060147e-05, "loss": 0.1321, "step": 38813 }, { "epoch": 28.435164835164834, "grad_norm": 0.80859375, "learning_rate": 1.0222925691713835e-05, "loss": 0.1149, "step": 38814 }, { "epoch": 28.435897435897434, "grad_norm": 0.9140625, "learning_rate": 1.0220933991348194e-05, "loss": 0.1108, "step": 38815 }, { "epoch": 28.436630036630035, "grad_norm": 1.0703125, "learning_rate": 1.0218942483972784e-05, "loss": 0.1131, "step": 38816 }, { "epoch": 28.437362637362636, "grad_norm": 1.25, "learning_rate": 1.0216951169597145e-05, "loss": 0.1439, "step": 38817 }, { "epoch": 28.438095238095237, "grad_norm": 0.95703125, "learning_rate": 1.021496004823083e-05, "loss": 0.13, "step": 38818 }, { "epoch": 28.438827838827837, "grad_norm": 1.2734375, "learning_rate": 1.0212969119883355e-05, "loss": 0.1107, "step": 38819 }, { "epoch": 28.439560439560438, "grad_norm": 1.4609375, "learning_rate": 1.0210978384564279e-05, "loss": 0.1437, "step": 38820 }, { "epoch": 28.44029304029304, "grad_norm": 1.1796875, "learning_rate": 1.020898784228312e-05, "loss": 0.1274, "step": 38821 }, { "epoch": 28.44102564102564, "grad_norm": 1.0078125, "learning_rate": 1.0206997493049428e-05, "loss": 0.1193, "step": 38822 }, { "epoch": 28.44175824175824, "grad_norm": 0.671875, "learning_rate": 1.0205007336872737e-05, "loss": 0.1126, "step": 38823 }, { "epoch": 28.44249084249084, "grad_norm": 1.078125, "learning_rate": 1.0203017373762599e-05, "loss": 0.1353, "step": 38824 }, { "epoch": 28.44322344322344, "grad_norm": 1.15625, "learning_rate": 1.0201027603728518e-05, "loss": 0.1201, "step": 38825 }, { "epoch": 28.443956043956042, "grad_norm": 1.2890625, "learning_rate": 1.0199038026780045e-05, "loss": 0.0948, "step": 38826 }, { "epoch": 28.444688644688643, "grad_norm": 1.234375, "learning_rate": 1.0197048642926705e-05, "loss": 0.1084, "step": 38827 }, { "epoch": 28.445421245421244, "grad_norm": 0.79296875, "learning_rate": 1.0195059452178042e-05, "loss": 0.1022, "step": 38828 }, { "epoch": 28.446153846153845, "grad_norm": 1.0078125, "learning_rate": 1.0193070454543589e-05, "loss": 0.1437, "step": 38829 }, { "epoch": 28.446886446886445, "grad_norm": 1.1328125, "learning_rate": 1.019108165003286e-05, "loss": 0.1212, "step": 38830 }, { "epoch": 28.447619047619046, "grad_norm": 1.046875, "learning_rate": 1.0189093038655402e-05, "loss": 0.1533, "step": 38831 }, { "epoch": 28.448351648351647, "grad_norm": 0.8359375, "learning_rate": 1.0187104620420707e-05, "loss": 0.1189, "step": 38832 }, { "epoch": 28.449084249084247, "grad_norm": 1.6328125, "learning_rate": 1.0185116395338348e-05, "loss": 0.1153, "step": 38833 }, { "epoch": 28.449816849816848, "grad_norm": 0.98828125, "learning_rate": 1.0183128363417827e-05, "loss": 0.134, "step": 38834 }, { "epoch": 28.45054945054945, "grad_norm": 1.0, "learning_rate": 1.0181140524668678e-05, "loss": 0.0955, "step": 38835 }, { "epoch": 28.45128205128205, "grad_norm": 0.8203125, "learning_rate": 1.017915287910041e-05, "loss": 0.1289, "step": 38836 }, { "epoch": 28.45201465201465, "grad_norm": 0.95703125, "learning_rate": 1.0177165426722557e-05, "loss": 0.1066, "step": 38837 }, { "epoch": 28.45274725274725, "grad_norm": 1.0703125, "learning_rate": 1.0175178167544645e-05, "loss": 0.1357, "step": 38838 }, { "epoch": 28.453479853479852, "grad_norm": 0.82421875, "learning_rate": 1.017319110157619e-05, "loss": 0.1165, "step": 38839 }, { "epoch": 28.454212454212453, "grad_norm": 0.72265625, "learning_rate": 1.0171204228826728e-05, "loss": 0.1182, "step": 38840 }, { "epoch": 28.454945054945053, "grad_norm": 0.9140625, "learning_rate": 1.0169217549305751e-05, "loss": 0.1122, "step": 38841 }, { "epoch": 28.455677655677654, "grad_norm": 0.99609375, "learning_rate": 1.0167231063022791e-05, "loss": 0.0999, "step": 38842 }, { "epoch": 28.456410256410255, "grad_norm": 1.46875, "learning_rate": 1.016524476998737e-05, "loss": 0.1345, "step": 38843 }, { "epoch": 28.457142857142856, "grad_norm": 0.76953125, "learning_rate": 1.0163258670209017e-05, "loss": 0.0852, "step": 38844 }, { "epoch": 28.457875457875456, "grad_norm": 0.96484375, "learning_rate": 1.0161272763697217e-05, "loss": 0.1177, "step": 38845 }, { "epoch": 28.458608058608057, "grad_norm": 1.1015625, "learning_rate": 1.0159287050461516e-05, "loss": 0.1232, "step": 38846 }, { "epoch": 28.459340659340658, "grad_norm": 1.65625, "learning_rate": 1.0157301530511402e-05, "loss": 0.1156, "step": 38847 }, { "epoch": 28.46007326007326, "grad_norm": 0.7734375, "learning_rate": 1.01553162038564e-05, "loss": 0.0968, "step": 38848 }, { "epoch": 28.46080586080586, "grad_norm": 1.1796875, "learning_rate": 1.0153331070506023e-05, "loss": 0.1119, "step": 38849 }, { "epoch": 28.46153846153846, "grad_norm": 1.1796875, "learning_rate": 1.0151346130469789e-05, "loss": 0.1276, "step": 38850 }, { "epoch": 28.46227106227106, "grad_norm": 1.875, "learning_rate": 1.0149361383757192e-05, "loss": 0.0949, "step": 38851 }, { "epoch": 28.46300366300366, "grad_norm": 0.9296875, "learning_rate": 1.0147376830377752e-05, "loss": 0.1269, "step": 38852 }, { "epoch": 28.463736263736262, "grad_norm": 1.4609375, "learning_rate": 1.0145392470340978e-05, "loss": 0.143, "step": 38853 }, { "epoch": 28.464468864468863, "grad_norm": 0.6953125, "learning_rate": 1.0143408303656374e-05, "loss": 0.0959, "step": 38854 }, { "epoch": 28.465201465201464, "grad_norm": 0.84765625, "learning_rate": 1.0141424330333459e-05, "loss": 0.0911, "step": 38855 }, { "epoch": 28.465934065934064, "grad_norm": 1.0234375, "learning_rate": 1.013944055038172e-05, "loss": 0.1163, "step": 38856 }, { "epoch": 28.466666666666665, "grad_norm": 1.03125, "learning_rate": 1.0137456963810681e-05, "loss": 0.1331, "step": 38857 }, { "epoch": 28.467399267399266, "grad_norm": 1.515625, "learning_rate": 1.0135473570629816e-05, "loss": 0.1104, "step": 38858 }, { "epoch": 28.468131868131866, "grad_norm": 1.140625, "learning_rate": 1.0133490370848666e-05, "loss": 0.1176, "step": 38859 }, { "epoch": 28.468864468864467, "grad_norm": 1.21875, "learning_rate": 1.0131507364476709e-05, "loss": 0.128, "step": 38860 }, { "epoch": 28.469597069597068, "grad_norm": 1.078125, "learning_rate": 1.0129524551523461e-05, "loss": 0.1027, "step": 38861 }, { "epoch": 28.47032967032967, "grad_norm": 0.85546875, "learning_rate": 1.01275419319984e-05, "loss": 0.1229, "step": 38862 }, { "epoch": 28.47106227106227, "grad_norm": 1.0234375, "learning_rate": 1.012555950591104e-05, "loss": 0.1063, "step": 38863 }, { "epoch": 28.47179487179487, "grad_norm": 1.2109375, "learning_rate": 1.0123577273270877e-05, "loss": 0.1594, "step": 38864 }, { "epoch": 28.47252747252747, "grad_norm": 0.94140625, "learning_rate": 1.0121595234087414e-05, "loss": 0.1365, "step": 38865 }, { "epoch": 28.47326007326007, "grad_norm": 1.3984375, "learning_rate": 1.0119613388370155e-05, "loss": 0.1484, "step": 38866 }, { "epoch": 28.473992673992672, "grad_norm": 0.65625, "learning_rate": 1.0117631736128568e-05, "loss": 0.1254, "step": 38867 }, { "epoch": 28.474725274725273, "grad_norm": 1.5390625, "learning_rate": 1.0115650277372167e-05, "loss": 0.1622, "step": 38868 }, { "epoch": 28.475457875457874, "grad_norm": 1.09375, "learning_rate": 1.0113669012110443e-05, "loss": 0.105, "step": 38869 }, { "epoch": 28.476190476190474, "grad_norm": 0.86328125, "learning_rate": 1.0111687940352901e-05, "loss": 0.1211, "step": 38870 }, { "epoch": 28.476923076923075, "grad_norm": 0.75390625, "learning_rate": 1.0109707062109013e-05, "loss": 0.1271, "step": 38871 }, { "epoch": 28.477655677655676, "grad_norm": 0.62109375, "learning_rate": 1.0107726377388283e-05, "loss": 0.1093, "step": 38872 }, { "epoch": 28.478388278388277, "grad_norm": 1.0078125, "learning_rate": 1.0105745886200185e-05, "loss": 0.1209, "step": 38873 }, { "epoch": 28.479120879120877, "grad_norm": 0.7734375, "learning_rate": 1.0103765588554222e-05, "loss": 0.0971, "step": 38874 }, { "epoch": 28.479853479853478, "grad_norm": 1.2265625, "learning_rate": 1.0101785484459874e-05, "loss": 0.1141, "step": 38875 }, { "epoch": 28.48058608058608, "grad_norm": 1.328125, "learning_rate": 1.009980557392664e-05, "loss": 0.1322, "step": 38876 }, { "epoch": 28.48131868131868, "grad_norm": 1.109375, "learning_rate": 1.0097825856964009e-05, "loss": 0.13, "step": 38877 }, { "epoch": 28.48205128205128, "grad_norm": 1.078125, "learning_rate": 1.0095846333581444e-05, "loss": 0.1243, "step": 38878 }, { "epoch": 28.48278388278388, "grad_norm": 0.859375, "learning_rate": 1.0093867003788444e-05, "loss": 0.1046, "step": 38879 }, { "epoch": 28.483516483516482, "grad_norm": 1.453125, "learning_rate": 1.0091887867594493e-05, "loss": 0.1242, "step": 38880 }, { "epoch": 28.484249084249083, "grad_norm": 1.015625, "learning_rate": 1.008990892500908e-05, "loss": 0.1328, "step": 38881 }, { "epoch": 28.484981684981683, "grad_norm": 1.546875, "learning_rate": 1.008793017604167e-05, "loss": 0.1324, "step": 38882 }, { "epoch": 28.485714285714284, "grad_norm": 0.8046875, "learning_rate": 1.0085951620701767e-05, "loss": 0.1025, "step": 38883 }, { "epoch": 28.486446886446885, "grad_norm": 1.3359375, "learning_rate": 1.0083973258998809e-05, "loss": 0.1315, "step": 38884 }, { "epoch": 28.487179487179485, "grad_norm": 1.3203125, "learning_rate": 1.0081995090942333e-05, "loss": 0.111, "step": 38885 }, { "epoch": 28.487912087912086, "grad_norm": 1.0546875, "learning_rate": 1.0080017116541769e-05, "loss": 0.1519, "step": 38886 }, { "epoch": 28.488644688644687, "grad_norm": 1.40625, "learning_rate": 1.007803933580663e-05, "loss": 0.1332, "step": 38887 }, { "epoch": 28.489377289377288, "grad_norm": 0.76171875, "learning_rate": 1.0076061748746356e-05, "loss": 0.1271, "step": 38888 }, { "epoch": 28.49010989010989, "grad_norm": 0.875, "learning_rate": 1.007408435537044e-05, "loss": 0.1253, "step": 38889 }, { "epoch": 28.49084249084249, "grad_norm": 0.890625, "learning_rate": 1.0072107155688361e-05, "loss": 0.0966, "step": 38890 }, { "epoch": 28.49157509157509, "grad_norm": 1.4375, "learning_rate": 1.0070130149709588e-05, "loss": 0.1161, "step": 38891 }, { "epoch": 28.49230769230769, "grad_norm": 1.203125, "learning_rate": 1.0068153337443604e-05, "loss": 0.1583, "step": 38892 }, { "epoch": 28.493040293040295, "grad_norm": 0.74609375, "learning_rate": 1.006617671889986e-05, "loss": 0.1424, "step": 38893 }, { "epoch": 28.493772893772892, "grad_norm": 1.0859375, "learning_rate": 1.0064200294087834e-05, "loss": 0.1314, "step": 38894 }, { "epoch": 28.494505494505496, "grad_norm": 0.94921875, "learning_rate": 1.0062224063017002e-05, "loss": 0.1032, "step": 38895 }, { "epoch": 28.495238095238093, "grad_norm": 1.1640625, "learning_rate": 1.0060248025696835e-05, "loss": 0.1114, "step": 38896 }, { "epoch": 28.495970695970698, "grad_norm": 0.99609375, "learning_rate": 1.0058272182136788e-05, "loss": 0.1429, "step": 38897 }, { "epoch": 28.496703296703295, "grad_norm": 0.96484375, "learning_rate": 1.0056296532346348e-05, "loss": 0.1232, "step": 38898 }, { "epoch": 28.4974358974359, "grad_norm": 1.3359375, "learning_rate": 1.0054321076334957e-05, "loss": 0.1255, "step": 38899 }, { "epoch": 28.498168498168496, "grad_norm": 0.86328125, "learning_rate": 1.0052345814112089e-05, "loss": 0.1347, "step": 38900 }, { "epoch": 28.4989010989011, "grad_norm": 1.0390625, "learning_rate": 1.0050370745687208e-05, "loss": 0.1249, "step": 38901 }, { "epoch": 28.499633699633698, "grad_norm": 0.8984375, "learning_rate": 1.004839587106978e-05, "loss": 0.1077, "step": 38902 }, { "epoch": 28.500366300366302, "grad_norm": 1.109375, "learning_rate": 1.0046421190269282e-05, "loss": 0.1348, "step": 38903 }, { "epoch": 28.5010989010989, "grad_norm": 0.6484375, "learning_rate": 1.0044446703295147e-05, "loss": 0.1117, "step": 38904 }, { "epoch": 28.501831501831504, "grad_norm": 0.9609375, "learning_rate": 1.0042472410156849e-05, "loss": 0.1299, "step": 38905 }, { "epoch": 28.5025641025641, "grad_norm": 0.734375, "learning_rate": 1.0040498310863847e-05, "loss": 0.1159, "step": 38906 }, { "epoch": 28.503296703296705, "grad_norm": 1.0390625, "learning_rate": 1.003852440542561e-05, "loss": 0.1338, "step": 38907 }, { "epoch": 28.504029304029302, "grad_norm": 0.96875, "learning_rate": 1.0036550693851578e-05, "loss": 0.1248, "step": 38908 }, { "epoch": 28.504761904761907, "grad_norm": 1.421875, "learning_rate": 1.0034577176151222e-05, "loss": 0.1455, "step": 38909 }, { "epoch": 28.505494505494504, "grad_norm": 0.65625, "learning_rate": 1.0032603852333967e-05, "loss": 0.1139, "step": 38910 }, { "epoch": 28.506227106227108, "grad_norm": 1.1328125, "learning_rate": 1.0030630722409317e-05, "loss": 0.1232, "step": 38911 }, { "epoch": 28.506959706959705, "grad_norm": 1.375, "learning_rate": 1.0028657786386686e-05, "loss": 0.1099, "step": 38912 }, { "epoch": 28.50769230769231, "grad_norm": 1.078125, "learning_rate": 1.0026685044275555e-05, "loss": 0.1274, "step": 38913 }, { "epoch": 28.508424908424907, "grad_norm": 0.72265625, "learning_rate": 1.0024712496085349e-05, "loss": 0.1247, "step": 38914 }, { "epoch": 28.50915750915751, "grad_norm": 0.890625, "learning_rate": 1.0022740141825534e-05, "loss": 0.1309, "step": 38915 }, { "epoch": 28.509890109890108, "grad_norm": 0.66015625, "learning_rate": 1.0020767981505562e-05, "loss": 0.1134, "step": 38916 }, { "epoch": 28.510622710622712, "grad_norm": 1.03125, "learning_rate": 1.0018796015134872e-05, "loss": 0.1278, "step": 38917 }, { "epoch": 28.51135531135531, "grad_norm": 0.73046875, "learning_rate": 1.0016824242722942e-05, "loss": 0.1344, "step": 38918 }, { "epoch": 28.512087912087914, "grad_norm": 1.171875, "learning_rate": 1.0014852664279177e-05, "loss": 0.1245, "step": 38919 }, { "epoch": 28.51282051282051, "grad_norm": 0.99609375, "learning_rate": 1.0012881279813053e-05, "loss": 0.0937, "step": 38920 }, { "epoch": 28.513553113553115, "grad_norm": 0.84765625, "learning_rate": 1.0010910089333985e-05, "loss": 0.13, "step": 38921 }, { "epoch": 28.514285714285712, "grad_norm": 1.9296875, "learning_rate": 1.0008939092851467e-05, "loss": 0.152, "step": 38922 }, { "epoch": 28.515018315018317, "grad_norm": 0.89453125, "learning_rate": 1.0006968290374899e-05, "loss": 0.1143, "step": 38923 }, { "epoch": 28.515750915750917, "grad_norm": 0.70703125, "learning_rate": 1.0004997681913749e-05, "loss": 0.123, "step": 38924 }, { "epoch": 28.516483516483518, "grad_norm": 1.2421875, "learning_rate": 1.0003027267477434e-05, "loss": 0.1171, "step": 38925 }, { "epoch": 28.51721611721612, "grad_norm": 1.203125, "learning_rate": 1.0001057047075415e-05, "loss": 0.1291, "step": 38926 }, { "epoch": 28.51794871794872, "grad_norm": 1.2421875, "learning_rate": 9.999087020717123e-06, "loss": 0.1432, "step": 38927 }, { "epoch": 28.51868131868132, "grad_norm": 1.265625, "learning_rate": 9.997117188412002e-06, "loss": 0.1415, "step": 38928 }, { "epoch": 28.51941391941392, "grad_norm": 1.1171875, "learning_rate": 9.995147550169506e-06, "loss": 0.1103, "step": 38929 }, { "epoch": 28.520146520146522, "grad_norm": 1.234375, "learning_rate": 9.993178105999039e-06, "loss": 0.1183, "step": 38930 }, { "epoch": 28.520879120879123, "grad_norm": 0.6484375, "learning_rate": 9.991208855910051e-06, "loss": 0.1212, "step": 38931 }, { "epoch": 28.521611721611723, "grad_norm": 0.96875, "learning_rate": 9.989239799911982e-06, "loss": 0.106, "step": 38932 }, { "epoch": 28.522344322344324, "grad_norm": 1.1484375, "learning_rate": 9.987270938014274e-06, "loss": 0.1269, "step": 38933 }, { "epoch": 28.523076923076925, "grad_norm": 0.99609375, "learning_rate": 9.985302270226337e-06, "loss": 0.1148, "step": 38934 }, { "epoch": 28.523809523809526, "grad_norm": 0.80078125, "learning_rate": 9.98333379655763e-06, "loss": 0.1265, "step": 38935 }, { "epoch": 28.524542124542126, "grad_norm": 1.015625, "learning_rate": 9.981365517017554e-06, "loss": 0.0943, "step": 38936 }, { "epoch": 28.525274725274727, "grad_norm": 1.3359375, "learning_rate": 9.979397431615575e-06, "loss": 0.1767, "step": 38937 }, { "epoch": 28.526007326007328, "grad_norm": 0.8125, "learning_rate": 9.977429540361092e-06, "loss": 0.1288, "step": 38938 }, { "epoch": 28.52673992673993, "grad_norm": 1.0859375, "learning_rate": 9.97546184326356e-06, "loss": 0.1232, "step": 38939 }, { "epoch": 28.52747252747253, "grad_norm": 0.9375, "learning_rate": 9.973494340332382e-06, "loss": 0.137, "step": 38940 }, { "epoch": 28.52820512820513, "grad_norm": 1.265625, "learning_rate": 9.971527031577e-06, "loss": 0.1121, "step": 38941 }, { "epoch": 28.52893772893773, "grad_norm": 1.46875, "learning_rate": 9.969559917006833e-06, "loss": 0.1212, "step": 38942 }, { "epoch": 28.52967032967033, "grad_norm": 0.65234375, "learning_rate": 9.96759299663131e-06, "loss": 0.1061, "step": 38943 }, { "epoch": 28.530402930402932, "grad_norm": 1.0625, "learning_rate": 9.965626270459869e-06, "loss": 0.1593, "step": 38944 }, { "epoch": 28.531135531135533, "grad_norm": 0.97265625, "learning_rate": 9.963659738501905e-06, "loss": 0.1291, "step": 38945 }, { "epoch": 28.531868131868134, "grad_norm": 0.84765625, "learning_rate": 9.961693400766865e-06, "loss": 0.1133, "step": 38946 }, { "epoch": 28.532600732600734, "grad_norm": 1.0, "learning_rate": 9.95972725726414e-06, "loss": 0.1048, "step": 38947 }, { "epoch": 28.533333333333335, "grad_norm": 0.83984375, "learning_rate": 9.95776130800319e-06, "loss": 0.1392, "step": 38948 }, { "epoch": 28.534065934065936, "grad_norm": 1.484375, "learning_rate": 9.955795552993403e-06, "loss": 0.1205, "step": 38949 }, { "epoch": 28.534798534798536, "grad_norm": 1.25, "learning_rate": 9.953829992244226e-06, "loss": 0.1239, "step": 38950 }, { "epoch": 28.535531135531137, "grad_norm": 1.015625, "learning_rate": 9.951864625765045e-06, "loss": 0.118, "step": 38951 }, { "epoch": 28.536263736263738, "grad_norm": 1.296875, "learning_rate": 9.94989945356529e-06, "loss": 0.1021, "step": 38952 }, { "epoch": 28.53699633699634, "grad_norm": 0.703125, "learning_rate": 9.94793447565438e-06, "loss": 0.1363, "step": 38953 }, { "epoch": 28.53772893772894, "grad_norm": 1.0, "learning_rate": 9.945969692041726e-06, "loss": 0.108, "step": 38954 }, { "epoch": 28.53846153846154, "grad_norm": 0.66015625, "learning_rate": 9.944005102736757e-06, "loss": 0.0921, "step": 38955 }, { "epoch": 28.53919413919414, "grad_norm": 1.3671875, "learning_rate": 9.942040707748865e-06, "loss": 0.1723, "step": 38956 }, { "epoch": 28.53992673992674, "grad_norm": 0.81640625, "learning_rate": 9.94007650708747e-06, "loss": 0.1017, "step": 38957 }, { "epoch": 28.540659340659342, "grad_norm": 1.046875, "learning_rate": 9.938112500761977e-06, "loss": 0.1181, "step": 38958 }, { "epoch": 28.541391941391943, "grad_norm": 0.7734375, "learning_rate": 9.936148688781822e-06, "loss": 0.0996, "step": 38959 }, { "epoch": 28.542124542124544, "grad_norm": 0.69140625, "learning_rate": 9.934185071156376e-06, "loss": 0.1123, "step": 38960 }, { "epoch": 28.542857142857144, "grad_norm": 1.0078125, "learning_rate": 9.932221647895083e-06, "loss": 0.0939, "step": 38961 }, { "epoch": 28.543589743589745, "grad_norm": 0.77734375, "learning_rate": 9.930258419007308e-06, "loss": 0.0893, "step": 38962 }, { "epoch": 28.544322344322346, "grad_norm": 1.1796875, "learning_rate": 9.928295384502511e-06, "loss": 0.1134, "step": 38963 }, { "epoch": 28.545054945054947, "grad_norm": 1.4609375, "learning_rate": 9.92633254439005e-06, "loss": 0.1188, "step": 38964 }, { "epoch": 28.545787545787547, "grad_norm": 0.953125, "learning_rate": 9.924369898679368e-06, "loss": 0.1361, "step": 38965 }, { "epoch": 28.546520146520148, "grad_norm": 1.0078125, "learning_rate": 9.922407447379837e-06, "loss": 0.1242, "step": 38966 }, { "epoch": 28.54725274725275, "grad_norm": 1.2421875, "learning_rate": 9.920445190500871e-06, "loss": 0.113, "step": 38967 }, { "epoch": 28.54798534798535, "grad_norm": 1.5, "learning_rate": 9.918483128051873e-06, "loss": 0.1443, "step": 38968 }, { "epoch": 28.54871794871795, "grad_norm": 0.89453125, "learning_rate": 9.916521260042245e-06, "loss": 0.1251, "step": 38969 }, { "epoch": 28.54945054945055, "grad_norm": 0.703125, "learning_rate": 9.914559586481397e-06, "loss": 0.1063, "step": 38970 }, { "epoch": 28.550183150183152, "grad_norm": 1.375, "learning_rate": 9.912598107378707e-06, "loss": 0.1069, "step": 38971 }, { "epoch": 28.550915750915753, "grad_norm": 1.4140625, "learning_rate": 9.910636822743593e-06, "loss": 0.1106, "step": 38972 }, { "epoch": 28.551648351648353, "grad_norm": 0.71875, "learning_rate": 9.908675732585422e-06, "loss": 0.0914, "step": 38973 }, { "epoch": 28.552380952380954, "grad_norm": 1.0078125, "learning_rate": 9.906714836913632e-06, "loss": 0.1016, "step": 38974 }, { "epoch": 28.553113553113555, "grad_norm": 1.078125, "learning_rate": 9.904754135737585e-06, "loss": 0.1353, "step": 38975 }, { "epoch": 28.553846153846155, "grad_norm": 0.68359375, "learning_rate": 9.902793629066703e-06, "loss": 0.1187, "step": 38976 }, { "epoch": 28.554578754578756, "grad_norm": 1.1796875, "learning_rate": 9.90083331691035e-06, "loss": 0.1178, "step": 38977 }, { "epoch": 28.555311355311357, "grad_norm": 1.53125, "learning_rate": 9.898873199277934e-06, "loss": 0.1272, "step": 38978 }, { "epoch": 28.556043956043958, "grad_norm": 0.99609375, "learning_rate": 9.896913276178845e-06, "loss": 0.1238, "step": 38979 }, { "epoch": 28.55677655677656, "grad_norm": 0.93359375, "learning_rate": 9.894953547622468e-06, "loss": 0.1189, "step": 38980 }, { "epoch": 28.55750915750916, "grad_norm": 0.98828125, "learning_rate": 9.892994013618215e-06, "loss": 0.1398, "step": 38981 }, { "epoch": 28.55824175824176, "grad_norm": 0.86328125, "learning_rate": 9.89103467417545e-06, "loss": 0.1037, "step": 38982 }, { "epoch": 28.55897435897436, "grad_norm": 1.234375, "learning_rate": 9.889075529303563e-06, "loss": 0.1185, "step": 38983 }, { "epoch": 28.55970695970696, "grad_norm": 1.421875, "learning_rate": 9.887116579011946e-06, "loss": 0.1401, "step": 38984 }, { "epoch": 28.560439560439562, "grad_norm": 1.0625, "learning_rate": 9.885157823310003e-06, "loss": 0.1195, "step": 38985 }, { "epoch": 28.561172161172163, "grad_norm": 0.96875, "learning_rate": 9.883199262207083e-06, "loss": 0.1015, "step": 38986 }, { "epoch": 28.561904761904763, "grad_norm": 0.78125, "learning_rate": 9.881240895712607e-06, "loss": 0.0853, "step": 38987 }, { "epoch": 28.562637362637364, "grad_norm": 1.1640625, "learning_rate": 9.87928272383591e-06, "loss": 0.1566, "step": 38988 }, { "epoch": 28.563369963369965, "grad_norm": 1.375, "learning_rate": 9.877324746586434e-06, "loss": 0.134, "step": 38989 }, { "epoch": 28.564102564102566, "grad_norm": 1.015625, "learning_rate": 9.875366963973517e-06, "loss": 0.1087, "step": 38990 }, { "epoch": 28.564835164835166, "grad_norm": 0.9296875, "learning_rate": 9.87340937600655e-06, "loss": 0.0952, "step": 38991 }, { "epoch": 28.565567765567767, "grad_norm": 1.2890625, "learning_rate": 9.871451982694933e-06, "loss": 0.117, "step": 38992 }, { "epoch": 28.566300366300368, "grad_norm": 0.76953125, "learning_rate": 9.869494784048016e-06, "loss": 0.0999, "step": 38993 }, { "epoch": 28.56703296703297, "grad_norm": 0.88671875, "learning_rate": 9.86753778007518e-06, "loss": 0.0897, "step": 38994 }, { "epoch": 28.56776556776557, "grad_norm": 1.125, "learning_rate": 9.865580970785816e-06, "loss": 0.1191, "step": 38995 }, { "epoch": 28.56849816849817, "grad_norm": 1.3125, "learning_rate": 9.863624356189303e-06, "loss": 0.1147, "step": 38996 }, { "epoch": 28.56923076923077, "grad_norm": 1.15625, "learning_rate": 9.86166793629499e-06, "loss": 0.1286, "step": 38997 }, { "epoch": 28.56996336996337, "grad_norm": 1.4453125, "learning_rate": 9.859711711112284e-06, "loss": 0.0963, "step": 38998 }, { "epoch": 28.570695970695972, "grad_norm": 0.91796875, "learning_rate": 9.857755680650515e-06, "loss": 0.1052, "step": 38999 }, { "epoch": 28.571428571428573, "grad_norm": 1.5859375, "learning_rate": 9.855799844919104e-06, "loss": 0.1236, "step": 39000 }, { "epoch": 28.572161172161174, "grad_norm": 0.984375, "learning_rate": 9.853844203927388e-06, "loss": 0.1252, "step": 39001 }, { "epoch": 28.572893772893774, "grad_norm": 0.91796875, "learning_rate": 9.851888757684756e-06, "loss": 0.121, "step": 39002 }, { "epoch": 28.573626373626375, "grad_norm": 1.0, "learning_rate": 9.84993350620056e-06, "loss": 0.1076, "step": 39003 }, { "epoch": 28.574358974358976, "grad_norm": 1.3203125, "learning_rate": 9.847978449484172e-06, "loss": 0.1254, "step": 39004 }, { "epoch": 28.575091575091577, "grad_norm": 1.4921875, "learning_rate": 9.846023587544973e-06, "loss": 0.0998, "step": 39005 }, { "epoch": 28.575824175824177, "grad_norm": 1.2109375, "learning_rate": 9.844068920392316e-06, "loss": 0.1544, "step": 39006 }, { "epoch": 28.576556776556778, "grad_norm": 0.84765625, "learning_rate": 9.842114448035584e-06, "loss": 0.1443, "step": 39007 }, { "epoch": 28.57728937728938, "grad_norm": 0.953125, "learning_rate": 9.840160170484115e-06, "loss": 0.1381, "step": 39008 }, { "epoch": 28.57802197802198, "grad_norm": 0.66796875, "learning_rate": 9.83820608774729e-06, "loss": 0.1277, "step": 39009 }, { "epoch": 28.57875457875458, "grad_norm": 1.234375, "learning_rate": 9.836252199834462e-06, "loss": 0.1617, "step": 39010 }, { "epoch": 28.57948717948718, "grad_norm": 1.34375, "learning_rate": 9.834298506755018e-06, "loss": 0.1377, "step": 39011 }, { "epoch": 28.58021978021978, "grad_norm": 1.015625, "learning_rate": 9.832345008518282e-06, "loss": 0.1279, "step": 39012 }, { "epoch": 28.580952380952382, "grad_norm": 1.125, "learning_rate": 9.830391705133648e-06, "loss": 0.1214, "step": 39013 }, { "epoch": 28.581684981684983, "grad_norm": 1.6796875, "learning_rate": 9.82843859661043e-06, "loss": 0.1498, "step": 39014 }, { "epoch": 28.582417582417584, "grad_norm": 0.88671875, "learning_rate": 9.826485682958043e-06, "loss": 0.1212, "step": 39015 }, { "epoch": 28.583150183150185, "grad_norm": 1.1015625, "learning_rate": 9.824532964185796e-06, "loss": 0.1209, "step": 39016 }, { "epoch": 28.583882783882785, "grad_norm": 0.90234375, "learning_rate": 9.82258044030307e-06, "loss": 0.1302, "step": 39017 }, { "epoch": 28.584615384615386, "grad_norm": 0.8203125, "learning_rate": 9.820628111319223e-06, "loss": 0.1108, "step": 39018 }, { "epoch": 28.585347985347987, "grad_norm": 1.015625, "learning_rate": 9.818675977243593e-06, "loss": 0.1147, "step": 39019 }, { "epoch": 28.586080586080588, "grad_norm": 1.3203125, "learning_rate": 9.816724038085542e-06, "loss": 0.1095, "step": 39020 }, { "epoch": 28.58681318681319, "grad_norm": 0.7421875, "learning_rate": 9.814772293854418e-06, "loss": 0.1109, "step": 39021 }, { "epoch": 28.58754578754579, "grad_norm": 1.1796875, "learning_rate": 9.812820744559589e-06, "loss": 0.0846, "step": 39022 }, { "epoch": 28.58827838827839, "grad_norm": 1.4921875, "learning_rate": 9.810869390210379e-06, "loss": 0.1366, "step": 39023 }, { "epoch": 28.58901098901099, "grad_norm": 1.21875, "learning_rate": 9.808918230816166e-06, "loss": 0.0967, "step": 39024 }, { "epoch": 28.58974358974359, "grad_norm": 1.328125, "learning_rate": 9.806967266386257e-06, "loss": 0.1233, "step": 39025 }, { "epoch": 28.590476190476192, "grad_norm": 0.84375, "learning_rate": 9.805016496930053e-06, "loss": 0.1079, "step": 39026 }, { "epoch": 28.591208791208793, "grad_norm": 0.87109375, "learning_rate": 9.803065922456859e-06, "loss": 0.1321, "step": 39027 }, { "epoch": 28.591941391941393, "grad_norm": 1.5546875, "learning_rate": 9.801115542976054e-06, "loss": 0.1509, "step": 39028 }, { "epoch": 28.592673992673994, "grad_norm": 1.0390625, "learning_rate": 9.799165358496951e-06, "loss": 0.1484, "step": 39029 }, { "epoch": 28.593406593406595, "grad_norm": 1.4453125, "learning_rate": 9.797215369028905e-06, "loss": 0.1332, "step": 39030 }, { "epoch": 28.594139194139196, "grad_norm": 1.359375, "learning_rate": 9.795265574581263e-06, "loss": 0.1118, "step": 39031 }, { "epoch": 28.594871794871796, "grad_norm": 1.1015625, "learning_rate": 9.793315975163366e-06, "loss": 0.1317, "step": 39032 }, { "epoch": 28.595604395604397, "grad_norm": 0.4765625, "learning_rate": 9.791366570784566e-06, "loss": 0.0952, "step": 39033 }, { "epoch": 28.596336996336998, "grad_norm": 1.078125, "learning_rate": 9.789417361454181e-06, "loss": 0.1157, "step": 39034 }, { "epoch": 28.5970695970696, "grad_norm": 0.85546875, "learning_rate": 9.787468347181564e-06, "loss": 0.1261, "step": 39035 }, { "epoch": 28.5978021978022, "grad_norm": 0.91796875, "learning_rate": 9.78551952797605e-06, "loss": 0.1298, "step": 39036 }, { "epoch": 28.5985347985348, "grad_norm": 1.0625, "learning_rate": 9.783570903846993e-06, "loss": 0.1245, "step": 39037 }, { "epoch": 28.5992673992674, "grad_norm": 0.7890625, "learning_rate": 9.7816224748037e-06, "loss": 0.1153, "step": 39038 }, { "epoch": 28.6, "grad_norm": 0.81640625, "learning_rate": 9.77967424085553e-06, "loss": 0.1234, "step": 39039 }, { "epoch": 28.600732600732602, "grad_norm": 0.96484375, "learning_rate": 9.777726202011787e-06, "loss": 0.1222, "step": 39040 }, { "epoch": 28.601465201465203, "grad_norm": 1.125, "learning_rate": 9.775778358281847e-06, "loss": 0.1196, "step": 39041 }, { "epoch": 28.602197802197804, "grad_norm": 1.2265625, "learning_rate": 9.773830709675011e-06, "loss": 0.1423, "step": 39042 }, { "epoch": 28.602930402930404, "grad_norm": 1.171875, "learning_rate": 9.771883256200625e-06, "loss": 0.1087, "step": 39043 }, { "epoch": 28.603663003663005, "grad_norm": 1.21875, "learning_rate": 9.769935997868024e-06, "loss": 0.1337, "step": 39044 }, { "epoch": 28.604395604395606, "grad_norm": 0.9609375, "learning_rate": 9.767988934686525e-06, "loss": 0.1033, "step": 39045 }, { "epoch": 28.605128205128207, "grad_norm": 0.94921875, "learning_rate": 9.766042066665456e-06, "loss": 0.1214, "step": 39046 }, { "epoch": 28.605860805860807, "grad_norm": 1.1484375, "learning_rate": 9.764095393814154e-06, "loss": 0.1417, "step": 39047 }, { "epoch": 28.606593406593408, "grad_norm": 0.7890625, "learning_rate": 9.762148916141956e-06, "loss": 0.1121, "step": 39048 }, { "epoch": 28.60732600732601, "grad_norm": 1.015625, "learning_rate": 9.760202633658164e-06, "loss": 0.128, "step": 39049 }, { "epoch": 28.60805860805861, "grad_norm": 1.421875, "learning_rate": 9.758256546372124e-06, "loss": 0.1115, "step": 39050 }, { "epoch": 28.60879120879121, "grad_norm": 0.890625, "learning_rate": 9.75631065429313e-06, "loss": 0.1056, "step": 39051 }, { "epoch": 28.60952380952381, "grad_norm": 1.4921875, "learning_rate": 9.754364957430549e-06, "loss": 0.1364, "step": 39052 }, { "epoch": 28.61025641025641, "grad_norm": 0.82421875, "learning_rate": 9.752419455793674e-06, "loss": 0.1257, "step": 39053 }, { "epoch": 28.610989010989012, "grad_norm": 0.87890625, "learning_rate": 9.750474149391841e-06, "loss": 0.1266, "step": 39054 }, { "epoch": 28.611721611721613, "grad_norm": 1.2421875, "learning_rate": 9.748529038234353e-06, "loss": 0.1241, "step": 39055 }, { "epoch": 28.612454212454214, "grad_norm": 1.5078125, "learning_rate": 9.74658412233054e-06, "loss": 0.1587, "step": 39056 }, { "epoch": 28.613186813186815, "grad_norm": 1.6171875, "learning_rate": 9.744639401689724e-06, "loss": 0.1468, "step": 39057 }, { "epoch": 28.613919413919415, "grad_norm": 0.71484375, "learning_rate": 9.742694876321215e-06, "loss": 0.0967, "step": 39058 }, { "epoch": 28.614652014652016, "grad_norm": 0.89453125, "learning_rate": 9.74075054623435e-06, "loss": 0.1191, "step": 39059 }, { "epoch": 28.615384615384617, "grad_norm": 1.234375, "learning_rate": 9.738806411438414e-06, "loss": 0.1219, "step": 39060 }, { "epoch": 28.616117216117217, "grad_norm": 0.62890625, "learning_rate": 9.736862471942742e-06, "loss": 0.1082, "step": 39061 }, { "epoch": 28.616849816849818, "grad_norm": 0.83984375, "learning_rate": 9.734918727756642e-06, "loss": 0.1098, "step": 39062 }, { "epoch": 28.61758241758242, "grad_norm": 1.015625, "learning_rate": 9.73297517888944e-06, "loss": 0.1095, "step": 39063 }, { "epoch": 28.61831501831502, "grad_norm": 1.0859375, "learning_rate": 9.731031825350429e-06, "loss": 0.1104, "step": 39064 }, { "epoch": 28.61904761904762, "grad_norm": 1.0, "learning_rate": 9.729088667148935e-06, "loss": 0.11, "step": 39065 }, { "epoch": 28.61978021978022, "grad_norm": 0.81640625, "learning_rate": 9.727145704294235e-06, "loss": 0.1141, "step": 39066 }, { "epoch": 28.620512820512822, "grad_norm": 1.3515625, "learning_rate": 9.725202936795692e-06, "loss": 0.1308, "step": 39067 }, { "epoch": 28.621245421245423, "grad_norm": 1.28125, "learning_rate": 9.72326036466258e-06, "loss": 0.0905, "step": 39068 }, { "epoch": 28.621978021978023, "grad_norm": 0.953125, "learning_rate": 9.721317987904206e-06, "loss": 0.1174, "step": 39069 }, { "epoch": 28.622710622710624, "grad_norm": 0.890625, "learning_rate": 9.7193758065299e-06, "loss": 0.1237, "step": 39070 }, { "epoch": 28.623443223443225, "grad_norm": 1.25, "learning_rate": 9.717433820548939e-06, "loss": 0.111, "step": 39071 }, { "epoch": 28.624175824175826, "grad_norm": 1.125, "learning_rate": 9.71549202997064e-06, "loss": 0.1213, "step": 39072 }, { "epoch": 28.624908424908426, "grad_norm": 1.0390625, "learning_rate": 9.713550434804312e-06, "loss": 0.1246, "step": 39073 }, { "epoch": 28.625641025641027, "grad_norm": 0.9140625, "learning_rate": 9.71160903505926e-06, "loss": 0.1311, "step": 39074 }, { "epoch": 28.626373626373628, "grad_norm": 1.4453125, "learning_rate": 9.70966783074477e-06, "loss": 0.1179, "step": 39075 }, { "epoch": 28.62710622710623, "grad_norm": 1.09375, "learning_rate": 9.707726821870165e-06, "loss": 0.1065, "step": 39076 }, { "epoch": 28.62783882783883, "grad_norm": 1.453125, "learning_rate": 9.705786008444703e-06, "loss": 0.1439, "step": 39077 }, { "epoch": 28.62857142857143, "grad_norm": 0.765625, "learning_rate": 9.703845390477738e-06, "loss": 0.1274, "step": 39078 }, { "epoch": 28.62930402930403, "grad_norm": 0.828125, "learning_rate": 9.70190496797853e-06, "loss": 0.1141, "step": 39079 }, { "epoch": 28.63003663003663, "grad_norm": 0.7890625, "learning_rate": 9.699964740956403e-06, "loss": 0.119, "step": 39080 }, { "epoch": 28.630769230769232, "grad_norm": 0.953125, "learning_rate": 9.698024709420627e-06, "loss": 0.1127, "step": 39081 }, { "epoch": 28.631501831501833, "grad_norm": 1.140625, "learning_rate": 9.696084873380508e-06, "loss": 0.1142, "step": 39082 }, { "epoch": 28.632234432234434, "grad_norm": 1.109375, "learning_rate": 9.694145232845338e-06, "loss": 0.1375, "step": 39083 }, { "epoch": 28.632967032967034, "grad_norm": 0.8828125, "learning_rate": 9.69220578782442e-06, "loss": 0.1343, "step": 39084 }, { "epoch": 28.633699633699635, "grad_norm": 1.6640625, "learning_rate": 9.69026653832705e-06, "loss": 0.1228, "step": 39085 }, { "epoch": 28.634432234432236, "grad_norm": 0.83203125, "learning_rate": 9.688327484362493e-06, "loss": 0.0963, "step": 39086 }, { "epoch": 28.635164835164836, "grad_norm": 1.5078125, "learning_rate": 9.686388625940062e-06, "loss": 0.1597, "step": 39087 }, { "epoch": 28.635897435897437, "grad_norm": 0.921875, "learning_rate": 9.68444996306904e-06, "loss": 0.1331, "step": 39088 }, { "epoch": 28.636630036630038, "grad_norm": 1.5078125, "learning_rate": 9.68251149575873e-06, "loss": 0.1247, "step": 39089 }, { "epoch": 28.63736263736264, "grad_norm": 0.93359375, "learning_rate": 9.680573224018391e-06, "loss": 0.1458, "step": 39090 }, { "epoch": 28.63809523809524, "grad_norm": 0.70703125, "learning_rate": 9.678635147857343e-06, "loss": 0.1367, "step": 39091 }, { "epoch": 28.63882783882784, "grad_norm": 0.96875, "learning_rate": 9.676697267284828e-06, "loss": 0.1211, "step": 39092 }, { "epoch": 28.63956043956044, "grad_norm": 0.984375, "learning_rate": 9.674759582310185e-06, "loss": 0.1072, "step": 39093 }, { "epoch": 28.64029304029304, "grad_norm": 0.796875, "learning_rate": 9.672822092942655e-06, "loss": 0.1057, "step": 39094 }, { "epoch": 28.641025641025642, "grad_norm": 0.95703125, "learning_rate": 9.670884799191536e-06, "loss": 0.1337, "step": 39095 }, { "epoch": 28.641758241758243, "grad_norm": 1.171875, "learning_rate": 9.668947701066128e-06, "loss": 0.137, "step": 39096 }, { "epoch": 28.642490842490844, "grad_norm": 1.0625, "learning_rate": 9.667010798575681e-06, "loss": 0.1363, "step": 39097 }, { "epoch": 28.643223443223444, "grad_norm": 1.328125, "learning_rate": 9.665074091729488e-06, "loss": 0.111, "step": 39098 }, { "epoch": 28.643956043956045, "grad_norm": 1.1640625, "learning_rate": 9.663137580536833e-06, "loss": 0.1089, "step": 39099 }, { "epoch": 28.644688644688646, "grad_norm": 1.28125, "learning_rate": 9.661201265007005e-06, "loss": 0.1428, "step": 39100 }, { "epoch": 28.645421245421247, "grad_norm": 1.0, "learning_rate": 9.659265145149252e-06, "loss": 0.1429, "step": 39101 }, { "epoch": 28.646153846153847, "grad_norm": 0.96484375, "learning_rate": 9.65732922097288e-06, "loss": 0.1612, "step": 39102 }, { "epoch": 28.646886446886448, "grad_norm": 1.9140625, "learning_rate": 9.65539349248713e-06, "loss": 0.1577, "step": 39103 }, { "epoch": 28.64761904761905, "grad_norm": 1.3125, "learning_rate": 9.65345795970132e-06, "loss": 0.1033, "step": 39104 }, { "epoch": 28.64835164835165, "grad_norm": 0.828125, "learning_rate": 9.651522622624691e-06, "loss": 0.1097, "step": 39105 }, { "epoch": 28.64908424908425, "grad_norm": 0.81640625, "learning_rate": 9.649587481266529e-06, "loss": 0.1381, "step": 39106 }, { "epoch": 28.64981684981685, "grad_norm": 1.0078125, "learning_rate": 9.647652535636108e-06, "loss": 0.1009, "step": 39107 }, { "epoch": 28.650549450549452, "grad_norm": 1.140625, "learning_rate": 9.645717785742693e-06, "loss": 0.1124, "step": 39108 }, { "epoch": 28.651282051282053, "grad_norm": 0.73046875, "learning_rate": 9.643783231595546e-06, "loss": 0.1313, "step": 39109 }, { "epoch": 28.652014652014653, "grad_norm": 1.1640625, "learning_rate": 9.64184887320395e-06, "loss": 0.1518, "step": 39110 }, { "epoch": 28.652747252747254, "grad_norm": 1.6328125, "learning_rate": 9.639914710577185e-06, "loss": 0.165, "step": 39111 }, { "epoch": 28.653479853479855, "grad_norm": 1.1640625, "learning_rate": 9.637980743724483e-06, "loss": 0.1181, "step": 39112 }, { "epoch": 28.654212454212455, "grad_norm": 0.828125, "learning_rate": 9.636046972655135e-06, "loss": 0.1239, "step": 39113 }, { "epoch": 28.654945054945056, "grad_norm": 0.515625, "learning_rate": 9.634113397378394e-06, "loss": 0.1043, "step": 39114 }, { "epoch": 28.655677655677657, "grad_norm": 1.53125, "learning_rate": 9.63218001790355e-06, "loss": 0.1342, "step": 39115 }, { "epoch": 28.656410256410258, "grad_norm": 0.6640625, "learning_rate": 9.630246834239835e-06, "loss": 0.1072, "step": 39116 }, { "epoch": 28.65714285714286, "grad_norm": 1.328125, "learning_rate": 9.628313846396535e-06, "loss": 0.1393, "step": 39117 }, { "epoch": 28.65787545787546, "grad_norm": 0.6328125, "learning_rate": 9.626381054382872e-06, "loss": 0.1123, "step": 39118 }, { "epoch": 28.65860805860806, "grad_norm": 1.03125, "learning_rate": 9.624448458208163e-06, "loss": 0.124, "step": 39119 }, { "epoch": 28.65934065934066, "grad_norm": 1.5390625, "learning_rate": 9.622516057881627e-06, "loss": 0.1267, "step": 39120 }, { "epoch": 28.66007326007326, "grad_norm": 1.0390625, "learning_rate": 9.620583853412536e-06, "loss": 0.1062, "step": 39121 }, { "epoch": 28.660805860805862, "grad_norm": 0.89453125, "learning_rate": 9.618651844810163e-06, "loss": 0.1273, "step": 39122 }, { "epoch": 28.661538461538463, "grad_norm": 1.015625, "learning_rate": 9.61672003208372e-06, "loss": 0.1349, "step": 39123 }, { "epoch": 28.662271062271063, "grad_norm": 0.69921875, "learning_rate": 9.61478841524252e-06, "loss": 0.1113, "step": 39124 }, { "epoch": 28.663003663003664, "grad_norm": 0.96875, "learning_rate": 9.61285699429578e-06, "loss": 0.1057, "step": 39125 }, { "epoch": 28.663736263736265, "grad_norm": 1.5859375, "learning_rate": 9.610925769252773e-06, "loss": 0.1304, "step": 39126 }, { "epoch": 28.664468864468866, "grad_norm": 0.84375, "learning_rate": 9.608994740122735e-06, "loss": 0.1227, "step": 39127 }, { "epoch": 28.665201465201466, "grad_norm": 1.0078125, "learning_rate": 9.607063906914938e-06, "loss": 0.1064, "step": 39128 }, { "epoch": 28.665934065934067, "grad_norm": 0.796875, "learning_rate": 9.605133269638591e-06, "loss": 0.1247, "step": 39129 }, { "epoch": 28.666666666666668, "grad_norm": 1.2265625, "learning_rate": 9.60320282830301e-06, "loss": 0.1437, "step": 39130 }, { "epoch": 28.66739926739927, "grad_norm": 1.21875, "learning_rate": 9.60127258291739e-06, "loss": 0.1487, "step": 39131 }, { "epoch": 28.66813186813187, "grad_norm": 1.0234375, "learning_rate": 9.599342533491001e-06, "loss": 0.1277, "step": 39132 }, { "epoch": 28.66886446886447, "grad_norm": 0.87890625, "learning_rate": 9.597412680033102e-06, "loss": 0.1312, "step": 39133 }, { "epoch": 28.66959706959707, "grad_norm": 0.76953125, "learning_rate": 9.595483022552918e-06, "loss": 0.1403, "step": 39134 }, { "epoch": 28.67032967032967, "grad_norm": 0.6796875, "learning_rate": 9.593553561059695e-06, "loss": 0.107, "step": 39135 }, { "epoch": 28.671062271062272, "grad_norm": 1.4765625, "learning_rate": 9.591624295562693e-06, "loss": 0.1196, "step": 39136 }, { "epoch": 28.671794871794873, "grad_norm": 1.328125, "learning_rate": 9.589695226071154e-06, "loss": 0.1648, "step": 39137 }, { "epoch": 28.672527472527474, "grad_norm": 1.125, "learning_rate": 9.587766352594307e-06, "loss": 0.1201, "step": 39138 }, { "epoch": 28.673260073260074, "grad_norm": 1.2734375, "learning_rate": 9.585837675141404e-06, "loss": 0.1696, "step": 39139 }, { "epoch": 28.673992673992675, "grad_norm": 0.953125, "learning_rate": 9.58390919372168e-06, "loss": 0.1057, "step": 39140 }, { "epoch": 28.674725274725276, "grad_norm": 1.046875, "learning_rate": 9.581980908344395e-06, "loss": 0.1026, "step": 39141 }, { "epoch": 28.675457875457877, "grad_norm": 0.76953125, "learning_rate": 9.580052819018752e-06, "loss": 0.1087, "step": 39142 }, { "epoch": 28.676190476190477, "grad_norm": 1.6875, "learning_rate": 9.578124925754029e-06, "loss": 0.1348, "step": 39143 }, { "epoch": 28.676923076923078, "grad_norm": 1.03125, "learning_rate": 9.576197228559415e-06, "loss": 0.1402, "step": 39144 }, { "epoch": 28.67765567765568, "grad_norm": 0.7578125, "learning_rate": 9.574269727444204e-06, "loss": 0.1142, "step": 39145 }, { "epoch": 28.67838827838828, "grad_norm": 1.46875, "learning_rate": 9.572342422417582e-06, "loss": 0.1258, "step": 39146 }, { "epoch": 28.67912087912088, "grad_norm": 1.046875, "learning_rate": 9.570415313488805e-06, "loss": 0.1157, "step": 39147 }, { "epoch": 28.67985347985348, "grad_norm": 0.6171875, "learning_rate": 9.568488400667122e-06, "loss": 0.125, "step": 39148 }, { "epoch": 28.68058608058608, "grad_norm": 1.03125, "learning_rate": 9.566561683961733e-06, "loss": 0.1358, "step": 39149 }, { "epoch": 28.681318681318682, "grad_norm": 0.90234375, "learning_rate": 9.564635163381882e-06, "loss": 0.1238, "step": 39150 }, { "epoch": 28.682051282051283, "grad_norm": 0.97265625, "learning_rate": 9.562708838936806e-06, "loss": 0.1218, "step": 39151 }, { "epoch": 28.682783882783884, "grad_norm": 1.46875, "learning_rate": 9.560782710635737e-06, "loss": 0.122, "step": 39152 }, { "epoch": 28.683516483516485, "grad_norm": 0.80078125, "learning_rate": 9.558856778487892e-06, "loss": 0.121, "step": 39153 }, { "epoch": 28.684249084249085, "grad_norm": 1.078125, "learning_rate": 9.556931042502512e-06, "loss": 0.1164, "step": 39154 }, { "epoch": 28.684981684981686, "grad_norm": 1.34375, "learning_rate": 9.555005502688792e-06, "loss": 0.158, "step": 39155 }, { "epoch": 28.685714285714287, "grad_norm": 0.87890625, "learning_rate": 9.55308015905601e-06, "loss": 0.127, "step": 39156 }, { "epoch": 28.686446886446888, "grad_norm": 1.3828125, "learning_rate": 9.551155011613351e-06, "loss": 0.1261, "step": 39157 }, { "epoch": 28.68717948717949, "grad_norm": 1.0546875, "learning_rate": 9.549230060370046e-06, "loss": 0.113, "step": 39158 }, { "epoch": 28.68791208791209, "grad_norm": 0.734375, "learning_rate": 9.547305305335338e-06, "loss": 0.1034, "step": 39159 }, { "epoch": 28.68864468864469, "grad_norm": 0.9921875, "learning_rate": 9.545380746518421e-06, "loss": 0.1184, "step": 39160 }, { "epoch": 28.68937728937729, "grad_norm": 0.93359375, "learning_rate": 9.543456383928531e-06, "loss": 0.1194, "step": 39161 }, { "epoch": 28.69010989010989, "grad_norm": 1.859375, "learning_rate": 9.541532217574886e-06, "loss": 0.1463, "step": 39162 }, { "epoch": 28.690842490842492, "grad_norm": 1.078125, "learning_rate": 9.539608247466714e-06, "loss": 0.1276, "step": 39163 }, { "epoch": 28.691575091575093, "grad_norm": 0.69140625, "learning_rate": 9.537684473613224e-06, "loss": 0.1058, "step": 39164 }, { "epoch": 28.692307692307693, "grad_norm": 0.5390625, "learning_rate": 9.535760896023624e-06, "loss": 0.1133, "step": 39165 }, { "epoch": 28.693040293040294, "grad_norm": 1.0078125, "learning_rate": 9.533837514707143e-06, "loss": 0.1047, "step": 39166 }, { "epoch": 28.693772893772895, "grad_norm": 0.76171875, "learning_rate": 9.53191432967301e-06, "loss": 0.09, "step": 39167 }, { "epoch": 28.694505494505496, "grad_norm": 1.2421875, "learning_rate": 9.529991340930413e-06, "loss": 0.1241, "step": 39168 }, { "epoch": 28.695238095238096, "grad_norm": 0.88671875, "learning_rate": 9.528068548488589e-06, "loss": 0.1193, "step": 39169 }, { "epoch": 28.695970695970697, "grad_norm": 0.73828125, "learning_rate": 9.526145952356712e-06, "loss": 0.0962, "step": 39170 }, { "epoch": 28.696703296703298, "grad_norm": 0.78125, "learning_rate": 9.524223552544041e-06, "loss": 0.1045, "step": 39171 }, { "epoch": 28.6974358974359, "grad_norm": 0.6484375, "learning_rate": 9.522301349059763e-06, "loss": 0.1211, "step": 39172 }, { "epoch": 28.6981684981685, "grad_norm": 0.9140625, "learning_rate": 9.520379341913086e-06, "loss": 0.0989, "step": 39173 }, { "epoch": 28.6989010989011, "grad_norm": 0.953125, "learning_rate": 9.518457531113239e-06, "loss": 0.1117, "step": 39174 }, { "epoch": 28.6996336996337, "grad_norm": 1.125, "learning_rate": 9.516535916669401e-06, "loss": 0.0965, "step": 39175 }, { "epoch": 28.7003663003663, "grad_norm": 0.9375, "learning_rate": 9.514614498590793e-06, "loss": 0.1128, "step": 39176 }, { "epoch": 28.701098901098902, "grad_norm": 0.640625, "learning_rate": 9.512693276886624e-06, "loss": 0.1168, "step": 39177 }, { "epoch": 28.701831501831503, "grad_norm": 0.9765625, "learning_rate": 9.51077225156611e-06, "loss": 0.1108, "step": 39178 }, { "epoch": 28.702564102564104, "grad_norm": 1.0625, "learning_rate": 9.508851422638428e-06, "loss": 0.0913, "step": 39179 }, { "epoch": 28.703296703296704, "grad_norm": 0.77734375, "learning_rate": 9.506930790112805e-06, "loss": 0.1194, "step": 39180 }, { "epoch": 28.704029304029305, "grad_norm": 0.90625, "learning_rate": 9.505010353998408e-06, "loss": 0.1431, "step": 39181 }, { "epoch": 28.704761904761906, "grad_norm": 0.8046875, "learning_rate": 9.503090114304492e-06, "loss": 0.1108, "step": 39182 }, { "epoch": 28.705494505494507, "grad_norm": 1.109375, "learning_rate": 9.50117007104022e-06, "loss": 0.1224, "step": 39183 }, { "epoch": 28.706227106227107, "grad_norm": 1.0703125, "learning_rate": 9.499250224214799e-06, "loss": 0.1195, "step": 39184 }, { "epoch": 28.706959706959708, "grad_norm": 1.0078125, "learning_rate": 9.497330573837442e-06, "loss": 0.1294, "step": 39185 }, { "epoch": 28.70769230769231, "grad_norm": 0.8828125, "learning_rate": 9.49541111991732e-06, "loss": 0.1064, "step": 39186 }, { "epoch": 28.70842490842491, "grad_norm": 0.8671875, "learning_rate": 9.49349186246365e-06, "loss": 0.1178, "step": 39187 }, { "epoch": 28.70915750915751, "grad_norm": 1.1640625, "learning_rate": 9.491572801485622e-06, "loss": 0.15, "step": 39188 }, { "epoch": 28.70989010989011, "grad_norm": 1.1953125, "learning_rate": 9.489653936992437e-06, "loss": 0.1114, "step": 39189 }, { "epoch": 28.71062271062271, "grad_norm": 0.8046875, "learning_rate": 9.487735268993268e-06, "loss": 0.118, "step": 39190 }, { "epoch": 28.711355311355312, "grad_norm": 1.1484375, "learning_rate": 9.485816797497331e-06, "loss": 0.137, "step": 39191 }, { "epoch": 28.712087912087913, "grad_norm": 1.3359375, "learning_rate": 9.483898522513809e-06, "loss": 0.1327, "step": 39192 }, { "epoch": 28.712820512820514, "grad_norm": 0.87890625, "learning_rate": 9.481980444051902e-06, "loss": 0.1284, "step": 39193 }, { "epoch": 28.713553113553115, "grad_norm": 0.79296875, "learning_rate": 9.480062562120781e-06, "loss": 0.0955, "step": 39194 }, { "epoch": 28.714285714285715, "grad_norm": 0.98046875, "learning_rate": 9.478144876729657e-06, "loss": 0.1069, "step": 39195 }, { "epoch": 28.715018315018316, "grad_norm": 0.77734375, "learning_rate": 9.476227387887691e-06, "loss": 0.1275, "step": 39196 }, { "epoch": 28.715750915750917, "grad_norm": 0.890625, "learning_rate": 9.474310095604105e-06, "loss": 0.1298, "step": 39197 }, { "epoch": 28.716483516483517, "grad_norm": 1.328125, "learning_rate": 9.472392999888055e-06, "loss": 0.1394, "step": 39198 }, { "epoch": 28.717216117216118, "grad_norm": 1.5625, "learning_rate": 9.47047610074874e-06, "loss": 0.1192, "step": 39199 }, { "epoch": 28.71794871794872, "grad_norm": 1.0078125, "learning_rate": 9.468559398195348e-06, "loss": 0.1105, "step": 39200 }, { "epoch": 28.71868131868132, "grad_norm": 1.3046875, "learning_rate": 9.466642892237056e-06, "loss": 0.1106, "step": 39201 }, { "epoch": 28.71941391941392, "grad_norm": 0.9609375, "learning_rate": 9.46472658288304e-06, "loss": 0.1219, "step": 39202 }, { "epoch": 28.72014652014652, "grad_norm": 1.640625, "learning_rate": 9.462810470142494e-06, "loss": 0.1354, "step": 39203 }, { "epoch": 28.720879120879122, "grad_norm": 1.71875, "learning_rate": 9.460894554024611e-06, "loss": 0.1323, "step": 39204 }, { "epoch": 28.721611721611723, "grad_norm": 0.8125, "learning_rate": 9.458978834538539e-06, "loss": 0.1193, "step": 39205 }, { "epoch": 28.722344322344323, "grad_norm": 1.0859375, "learning_rate": 9.45706331169348e-06, "loss": 0.137, "step": 39206 }, { "epoch": 28.723076923076924, "grad_norm": 1.0, "learning_rate": 9.455147985498584e-06, "loss": 0.1084, "step": 39207 }, { "epoch": 28.723809523809525, "grad_norm": 1.234375, "learning_rate": 9.453232855963075e-06, "loss": 0.1434, "step": 39208 }, { "epoch": 28.724542124542126, "grad_norm": 1.03125, "learning_rate": 9.451317923096084e-06, "loss": 0.1315, "step": 39209 }, { "epoch": 28.725274725274726, "grad_norm": 0.8203125, "learning_rate": 9.449403186906808e-06, "loss": 0.1, "step": 39210 }, { "epoch": 28.726007326007327, "grad_norm": 1.1640625, "learning_rate": 9.447488647404428e-06, "loss": 0.1172, "step": 39211 }, { "epoch": 28.726739926739928, "grad_norm": 1.1484375, "learning_rate": 9.445574304598097e-06, "loss": 0.0951, "step": 39212 }, { "epoch": 28.72747252747253, "grad_norm": 1.0234375, "learning_rate": 9.443660158496993e-06, "loss": 0.1367, "step": 39213 }, { "epoch": 28.72820512820513, "grad_norm": 1.3515625, "learning_rate": 9.441746209110294e-06, "loss": 0.143, "step": 39214 }, { "epoch": 28.72893772893773, "grad_norm": 0.65234375, "learning_rate": 9.439832456447178e-06, "loss": 0.1193, "step": 39215 }, { "epoch": 28.72967032967033, "grad_norm": 1.140625, "learning_rate": 9.437918900516789e-06, "loss": 0.1253, "step": 39216 }, { "epoch": 28.73040293040293, "grad_norm": 1.4375, "learning_rate": 9.436005541328317e-06, "loss": 0.1346, "step": 39217 }, { "epoch": 28.731135531135532, "grad_norm": 1.203125, "learning_rate": 9.43409237889092e-06, "loss": 0.1198, "step": 39218 }, { "epoch": 28.731868131868133, "grad_norm": 0.796875, "learning_rate": 9.432179413213775e-06, "loss": 0.1152, "step": 39219 }, { "epoch": 28.732600732600734, "grad_norm": 0.6484375, "learning_rate": 9.430266644306032e-06, "loss": 0.1064, "step": 39220 }, { "epoch": 28.733333333333334, "grad_norm": 0.93359375, "learning_rate": 9.428354072176865e-06, "loss": 0.1345, "step": 39221 }, { "epoch": 28.734065934065935, "grad_norm": 1.0, "learning_rate": 9.426441696835438e-06, "loss": 0.1019, "step": 39222 }, { "epoch": 28.734798534798536, "grad_norm": 0.7421875, "learning_rate": 9.42452951829092e-06, "loss": 0.1237, "step": 39223 }, { "epoch": 28.735531135531136, "grad_norm": 0.95703125, "learning_rate": 9.422617536552456e-06, "loss": 0.1268, "step": 39224 }, { "epoch": 28.736263736263737, "grad_norm": 0.7734375, "learning_rate": 9.420705751629216e-06, "loss": 0.0982, "step": 39225 }, { "epoch": 28.736996336996338, "grad_norm": 0.98046875, "learning_rate": 9.418794163530368e-06, "loss": 0.1228, "step": 39226 }, { "epoch": 28.73772893772894, "grad_norm": 0.8828125, "learning_rate": 9.41688277226506e-06, "loss": 0.0931, "step": 39227 }, { "epoch": 28.73846153846154, "grad_norm": 1.0390625, "learning_rate": 9.414971577842447e-06, "loss": 0.1026, "step": 39228 }, { "epoch": 28.73919413919414, "grad_norm": 0.7421875, "learning_rate": 9.413060580271693e-06, "loss": 0.1231, "step": 39229 }, { "epoch": 28.73992673992674, "grad_norm": 0.77734375, "learning_rate": 9.411149779561965e-06, "loss": 0.0968, "step": 39230 }, { "epoch": 28.74065934065934, "grad_norm": 1.1953125, "learning_rate": 9.409239175722399e-06, "loss": 0.1445, "step": 39231 }, { "epoch": 28.741391941391942, "grad_norm": 0.98828125, "learning_rate": 9.407328768762165e-06, "loss": 0.1567, "step": 39232 }, { "epoch": 28.742124542124543, "grad_norm": 1.109375, "learning_rate": 9.405418558690385e-06, "loss": 0.1164, "step": 39233 }, { "epoch": 28.742857142857144, "grad_norm": 1.0, "learning_rate": 9.403508545516263e-06, "loss": 0.1329, "step": 39234 }, { "epoch": 28.743589743589745, "grad_norm": 0.6640625, "learning_rate": 9.401598729248906e-06, "loss": 0.112, "step": 39235 }, { "epoch": 28.744322344322345, "grad_norm": 0.60546875, "learning_rate": 9.399689109897486e-06, "loss": 0.0884, "step": 39236 }, { "epoch": 28.745054945054946, "grad_norm": 0.89453125, "learning_rate": 9.397779687471153e-06, "loss": 0.138, "step": 39237 }, { "epoch": 28.745787545787547, "grad_norm": 0.828125, "learning_rate": 9.39587046197903e-06, "loss": 0.1356, "step": 39238 }, { "epoch": 28.746520146520147, "grad_norm": 0.6875, "learning_rate": 9.393961433430312e-06, "loss": 0.1038, "step": 39239 }, { "epoch": 28.747252747252748, "grad_norm": 1.1015625, "learning_rate": 9.392052601834105e-06, "loss": 0.1136, "step": 39240 }, { "epoch": 28.74798534798535, "grad_norm": 1.0546875, "learning_rate": 9.390143967199585e-06, "loss": 0.1641, "step": 39241 }, { "epoch": 28.74871794871795, "grad_norm": 0.80859375, "learning_rate": 9.388235529535862e-06, "loss": 0.1008, "step": 39242 }, { "epoch": 28.74945054945055, "grad_norm": 0.90625, "learning_rate": 9.386327288852098e-06, "loss": 0.1186, "step": 39243 }, { "epoch": 28.75018315018315, "grad_norm": 0.9140625, "learning_rate": 9.384419245157444e-06, "loss": 0.1198, "step": 39244 }, { "epoch": 28.750915750915752, "grad_norm": 1.09375, "learning_rate": 9.38251139846104e-06, "loss": 0.1028, "step": 39245 }, { "epoch": 28.751648351648353, "grad_norm": 1.0625, "learning_rate": 9.380603748772014e-06, "loss": 0.1029, "step": 39246 }, { "epoch": 28.752380952380953, "grad_norm": 0.66015625, "learning_rate": 9.378696296099511e-06, "loss": 0.1134, "step": 39247 }, { "epoch": 28.753113553113554, "grad_norm": 1.0859375, "learning_rate": 9.376789040452675e-06, "loss": 0.104, "step": 39248 }, { "epoch": 28.753846153846155, "grad_norm": 0.90625, "learning_rate": 9.374881981840654e-06, "loss": 0.1368, "step": 39249 }, { "epoch": 28.754578754578755, "grad_norm": 1.15625, "learning_rate": 9.372975120272561e-06, "loss": 0.1307, "step": 39250 }, { "epoch": 28.755311355311356, "grad_norm": 1.109375, "learning_rate": 9.371068455757548e-06, "loss": 0.1125, "step": 39251 }, { "epoch": 28.756043956043957, "grad_norm": 1.1796875, "learning_rate": 9.36916198830476e-06, "loss": 0.1453, "step": 39252 }, { "epoch": 28.756776556776558, "grad_norm": 0.72265625, "learning_rate": 9.367255717923308e-06, "loss": 0.1137, "step": 39253 }, { "epoch": 28.75750915750916, "grad_norm": 0.89453125, "learning_rate": 9.365349644622331e-06, "loss": 0.1434, "step": 39254 }, { "epoch": 28.75824175824176, "grad_norm": 1.0546875, "learning_rate": 9.36344376841097e-06, "loss": 0.1091, "step": 39255 }, { "epoch": 28.75897435897436, "grad_norm": 1.359375, "learning_rate": 9.361538089298368e-06, "loss": 0.1205, "step": 39256 }, { "epoch": 28.75970695970696, "grad_norm": 0.97265625, "learning_rate": 9.35963260729363e-06, "loss": 0.1297, "step": 39257 }, { "epoch": 28.76043956043956, "grad_norm": 0.88671875, "learning_rate": 9.35772732240591e-06, "loss": 0.1202, "step": 39258 }, { "epoch": 28.761172161172162, "grad_norm": 0.60546875, "learning_rate": 9.3558222346443e-06, "loss": 0.0932, "step": 39259 }, { "epoch": 28.761904761904763, "grad_norm": 1.0234375, "learning_rate": 9.353917344017976e-06, "loss": 0.1072, "step": 39260 }, { "epoch": 28.762637362637363, "grad_norm": 0.75390625, "learning_rate": 9.352012650536027e-06, "loss": 0.0975, "step": 39261 }, { "epoch": 28.763369963369964, "grad_norm": 0.82421875, "learning_rate": 9.350108154207595e-06, "loss": 0.1036, "step": 39262 }, { "epoch": 28.764102564102565, "grad_norm": 1.3671875, "learning_rate": 9.348203855041814e-06, "loss": 0.1209, "step": 39263 }, { "epoch": 28.764835164835166, "grad_norm": 0.9296875, "learning_rate": 9.346299753047775e-06, "loss": 0.1246, "step": 39264 }, { "epoch": 28.765567765567766, "grad_norm": 0.9140625, "learning_rate": 9.34439584823465e-06, "loss": 0.1306, "step": 39265 }, { "epoch": 28.766300366300367, "grad_norm": 0.82421875, "learning_rate": 9.34249214061152e-06, "loss": 0.1, "step": 39266 }, { "epoch": 28.767032967032968, "grad_norm": 1.3359375, "learning_rate": 9.340588630187538e-06, "loss": 0.1148, "step": 39267 }, { "epoch": 28.76776556776557, "grad_norm": 1.421875, "learning_rate": 9.338685316971789e-06, "loss": 0.1136, "step": 39268 }, { "epoch": 28.76849816849817, "grad_norm": 1.0234375, "learning_rate": 9.336782200973416e-06, "loss": 0.1338, "step": 39269 }, { "epoch": 28.76923076923077, "grad_norm": 0.9921875, "learning_rate": 9.33487928220153e-06, "loss": 0.103, "step": 39270 }, { "epoch": 28.76996336996337, "grad_norm": 1.453125, "learning_rate": 9.332976560665266e-06, "loss": 0.1169, "step": 39271 }, { "epoch": 28.77069597069597, "grad_norm": 1.0546875, "learning_rate": 9.33107403637371e-06, "loss": 0.1086, "step": 39272 }, { "epoch": 28.771428571428572, "grad_norm": 1.078125, "learning_rate": 9.329171709335993e-06, "loss": 0.1197, "step": 39273 }, { "epoch": 28.772161172161173, "grad_norm": 1.375, "learning_rate": 9.327269579561232e-06, "loss": 0.1351, "step": 39274 }, { "epoch": 28.772893772893774, "grad_norm": 0.90625, "learning_rate": 9.325367647058552e-06, "loss": 0.1053, "step": 39275 }, { "epoch": 28.773626373626374, "grad_norm": 1.203125, "learning_rate": 9.323465911837035e-06, "loss": 0.1336, "step": 39276 }, { "epoch": 28.774358974358975, "grad_norm": 1.03125, "learning_rate": 9.321564373905815e-06, "loss": 0.1067, "step": 39277 }, { "epoch": 28.775091575091576, "grad_norm": 0.80078125, "learning_rate": 9.319663033274011e-06, "loss": 0.1106, "step": 39278 }, { "epoch": 28.775824175824177, "grad_norm": 0.77734375, "learning_rate": 9.317761889950705e-06, "loss": 0.0896, "step": 39279 }, { "epoch": 28.776556776556777, "grad_norm": 1.28125, "learning_rate": 9.315860943945021e-06, "loss": 0.1335, "step": 39280 }, { "epoch": 28.777289377289378, "grad_norm": 0.984375, "learning_rate": 9.313960195266065e-06, "loss": 0.1151, "step": 39281 }, { "epoch": 28.77802197802198, "grad_norm": 0.7890625, "learning_rate": 9.31205964392296e-06, "loss": 0.1131, "step": 39282 }, { "epoch": 28.77875457875458, "grad_norm": 1.3671875, "learning_rate": 9.310159289924789e-06, "loss": 0.1209, "step": 39283 }, { "epoch": 28.77948717948718, "grad_norm": 0.5234375, "learning_rate": 9.308259133280678e-06, "loss": 0.0986, "step": 39284 }, { "epoch": 28.78021978021978, "grad_norm": 1.28125, "learning_rate": 9.306359173999696e-06, "loss": 0.1212, "step": 39285 }, { "epoch": 28.78095238095238, "grad_norm": 1.0234375, "learning_rate": 9.304459412090988e-06, "loss": 0.1139, "step": 39286 }, { "epoch": 28.781684981684982, "grad_norm": 0.828125, "learning_rate": 9.302559847563633e-06, "loss": 0.1168, "step": 39287 }, { "epoch": 28.782417582417583, "grad_norm": 2.25, "learning_rate": 9.300660480426735e-06, "loss": 0.1576, "step": 39288 }, { "epoch": 28.783150183150184, "grad_norm": 0.83203125, "learning_rate": 9.298761310689411e-06, "loss": 0.1064, "step": 39289 }, { "epoch": 28.783882783882785, "grad_norm": 0.90625, "learning_rate": 9.296862338360724e-06, "loss": 0.137, "step": 39290 }, { "epoch": 28.784615384615385, "grad_norm": 0.703125, "learning_rate": 9.29496356344982e-06, "loss": 0.1194, "step": 39291 }, { "epoch": 28.785347985347986, "grad_norm": 1.125, "learning_rate": 9.293064985965755e-06, "loss": 0.1297, "step": 39292 }, { "epoch": 28.786080586080587, "grad_norm": 0.9296875, "learning_rate": 9.29116660591766e-06, "loss": 0.1185, "step": 39293 }, { "epoch": 28.786813186813188, "grad_norm": 1.390625, "learning_rate": 9.289268423314602e-06, "loss": 0.1131, "step": 39294 }, { "epoch": 28.78754578754579, "grad_norm": 1.0390625, "learning_rate": 9.28737043816569e-06, "loss": 0.1184, "step": 39295 }, { "epoch": 28.78827838827839, "grad_norm": 0.7890625, "learning_rate": 9.285472650480008e-06, "loss": 0.1192, "step": 39296 }, { "epoch": 28.78901098901099, "grad_norm": 1.4453125, "learning_rate": 9.283575060266672e-06, "loss": 0.1341, "step": 39297 }, { "epoch": 28.78974358974359, "grad_norm": 0.8515625, "learning_rate": 9.281677667534745e-06, "loss": 0.1037, "step": 39298 }, { "epoch": 28.79047619047619, "grad_norm": 0.90625, "learning_rate": 9.279780472293334e-06, "loss": 0.1144, "step": 39299 }, { "epoch": 28.791208791208792, "grad_norm": 0.859375, "learning_rate": 9.277883474551523e-06, "loss": 0.1452, "step": 39300 }, { "epoch": 28.791941391941393, "grad_norm": 1.0078125, "learning_rate": 9.27598667431842e-06, "loss": 0.1107, "step": 39301 }, { "epoch": 28.792673992673993, "grad_norm": 1.3203125, "learning_rate": 9.274090071603084e-06, "loss": 0.1193, "step": 39302 }, { "epoch": 28.793406593406594, "grad_norm": 1.03125, "learning_rate": 9.27219366641462e-06, "loss": 0.1229, "step": 39303 }, { "epoch": 28.794139194139195, "grad_norm": 1.03125, "learning_rate": 9.270297458762118e-06, "loss": 0.1217, "step": 39304 }, { "epoch": 28.794871794871796, "grad_norm": 1.9765625, "learning_rate": 9.268401448654645e-06, "loss": 0.1424, "step": 39305 }, { "epoch": 28.795604395604396, "grad_norm": 1.0859375, "learning_rate": 9.266505636101293e-06, "loss": 0.1376, "step": 39306 }, { "epoch": 28.796336996336997, "grad_norm": 1.2734375, "learning_rate": 9.264610021111156e-06, "loss": 0.1238, "step": 39307 }, { "epoch": 28.797069597069598, "grad_norm": 1.390625, "learning_rate": 9.262714603693311e-06, "loss": 0.108, "step": 39308 }, { "epoch": 28.7978021978022, "grad_norm": 1.03125, "learning_rate": 9.260819383856834e-06, "loss": 0.1256, "step": 39309 }, { "epoch": 28.7985347985348, "grad_norm": 1.171875, "learning_rate": 9.258924361610818e-06, "loss": 0.1177, "step": 39310 }, { "epoch": 28.7992673992674, "grad_norm": 0.8515625, "learning_rate": 9.257029536964311e-06, "loss": 0.109, "step": 39311 }, { "epoch": 28.8, "grad_norm": 0.85546875, "learning_rate": 9.255134909926438e-06, "loss": 0.1175, "step": 39312 }, { "epoch": 28.8007326007326, "grad_norm": 0.71875, "learning_rate": 9.253240480506244e-06, "loss": 0.1039, "step": 39313 }, { "epoch": 28.801465201465202, "grad_norm": 1.5234375, "learning_rate": 9.251346248712809e-06, "loss": 0.1616, "step": 39314 }, { "epoch": 28.802197802197803, "grad_norm": 1.4921875, "learning_rate": 9.24945221455523e-06, "loss": 0.1632, "step": 39315 }, { "epoch": 28.802930402930404, "grad_norm": 0.9921875, "learning_rate": 9.247558378042545e-06, "loss": 0.1182, "step": 39316 }, { "epoch": 28.803663003663004, "grad_norm": 1.140625, "learning_rate": 9.245664739183874e-06, "loss": 0.127, "step": 39317 }, { "epoch": 28.804395604395605, "grad_norm": 1.1640625, "learning_rate": 9.243771297988253e-06, "loss": 0.1263, "step": 39318 }, { "epoch": 28.805128205128206, "grad_norm": 1.3203125, "learning_rate": 9.24187805446478e-06, "loss": 0.1353, "step": 39319 }, { "epoch": 28.805860805860807, "grad_norm": 0.69921875, "learning_rate": 9.2399850086225e-06, "loss": 0.103, "step": 39320 }, { "epoch": 28.806593406593407, "grad_norm": 1.515625, "learning_rate": 9.238092160470499e-06, "loss": 0.1725, "step": 39321 }, { "epoch": 28.807326007326008, "grad_norm": 0.88671875, "learning_rate": 9.236199510017838e-06, "loss": 0.1479, "step": 39322 }, { "epoch": 28.80805860805861, "grad_norm": 0.93359375, "learning_rate": 9.23430705727361e-06, "loss": 0.098, "step": 39323 }, { "epoch": 28.80879120879121, "grad_norm": 0.8125, "learning_rate": 9.232414802246847e-06, "loss": 0.1358, "step": 39324 }, { "epoch": 28.80952380952381, "grad_norm": 0.87890625, "learning_rate": 9.230522744946637e-06, "loss": 0.0995, "step": 39325 }, { "epoch": 28.81025641025641, "grad_norm": 1.734375, "learning_rate": 9.228630885382036e-06, "loss": 0.1116, "step": 39326 }, { "epoch": 28.81098901098901, "grad_norm": 1.140625, "learning_rate": 9.226739223562125e-06, "loss": 0.1151, "step": 39327 }, { "epoch": 28.811721611721612, "grad_norm": 0.875, "learning_rate": 9.224847759495948e-06, "loss": 0.1128, "step": 39328 }, { "epoch": 28.812454212454213, "grad_norm": 0.875, "learning_rate": 9.222956493192571e-06, "loss": 0.1182, "step": 39329 }, { "epoch": 28.813186813186814, "grad_norm": 0.796875, "learning_rate": 9.221065424661072e-06, "loss": 0.1099, "step": 39330 }, { "epoch": 28.813919413919415, "grad_norm": 1.4296875, "learning_rate": 9.219174553910486e-06, "loss": 0.1401, "step": 39331 }, { "epoch": 28.814652014652015, "grad_norm": 1.15625, "learning_rate": 9.217283880949889e-06, "loss": 0.1337, "step": 39332 }, { "epoch": 28.815384615384616, "grad_norm": 1.03125, "learning_rate": 9.215393405788342e-06, "loss": 0.1301, "step": 39333 }, { "epoch": 28.816117216117217, "grad_norm": 0.7734375, "learning_rate": 9.213503128434903e-06, "loss": 0.1224, "step": 39334 }, { "epoch": 28.816849816849818, "grad_norm": 1.03125, "learning_rate": 9.211613048898615e-06, "loss": 0.1303, "step": 39335 }, { "epoch": 28.81758241758242, "grad_norm": 0.984375, "learning_rate": 9.20972316718855e-06, "loss": 0.1348, "step": 39336 }, { "epoch": 28.81831501831502, "grad_norm": 1.2265625, "learning_rate": 9.207833483313737e-06, "loss": 0.1441, "step": 39337 }, { "epoch": 28.81904761904762, "grad_norm": 0.78125, "learning_rate": 9.205943997283272e-06, "loss": 0.1405, "step": 39338 }, { "epoch": 28.81978021978022, "grad_norm": 0.65625, "learning_rate": 9.204054709106174e-06, "loss": 0.0994, "step": 39339 }, { "epoch": 28.82051282051282, "grad_norm": 1.1328125, "learning_rate": 9.202165618791504e-06, "loss": 0.1202, "step": 39340 }, { "epoch": 28.821245421245422, "grad_norm": 0.5859375, "learning_rate": 9.200276726348332e-06, "loss": 0.1007, "step": 39341 }, { "epoch": 28.821978021978023, "grad_norm": 0.98828125, "learning_rate": 9.19838803178566e-06, "loss": 0.1165, "step": 39342 }, { "epoch": 28.822710622710623, "grad_norm": 0.89453125, "learning_rate": 9.1964995351126e-06, "loss": 0.0994, "step": 39343 }, { "epoch": 28.823443223443224, "grad_norm": 0.96484375, "learning_rate": 9.194611236338156e-06, "loss": 0.1298, "step": 39344 }, { "epoch": 28.824175824175825, "grad_norm": 0.79296875, "learning_rate": 9.192723135471404e-06, "loss": 0.1369, "step": 39345 }, { "epoch": 28.824908424908426, "grad_norm": 1.3984375, "learning_rate": 9.190835232521363e-06, "loss": 0.1093, "step": 39346 }, { "epoch": 28.825641025641026, "grad_norm": 1.03125, "learning_rate": 9.188947527497091e-06, "loss": 0.1187, "step": 39347 }, { "epoch": 28.826373626373627, "grad_norm": 0.765625, "learning_rate": 9.187060020407627e-06, "loss": 0.1055, "step": 39348 }, { "epoch": 28.827106227106228, "grad_norm": 1.1015625, "learning_rate": 9.185172711262038e-06, "loss": 0.1069, "step": 39349 }, { "epoch": 28.82783882783883, "grad_norm": 0.76953125, "learning_rate": 9.183285600069339e-06, "loss": 0.1265, "step": 39350 }, { "epoch": 28.82857142857143, "grad_norm": 1.3984375, "learning_rate": 9.181398686838579e-06, "loss": 0.1057, "step": 39351 }, { "epoch": 28.82930402930403, "grad_norm": 0.54296875, "learning_rate": 9.179511971578813e-06, "loss": 0.0994, "step": 39352 }, { "epoch": 28.83003663003663, "grad_norm": 1.515625, "learning_rate": 9.177625454299049e-06, "loss": 0.1288, "step": 39353 }, { "epoch": 28.83076923076923, "grad_norm": 1.0625, "learning_rate": 9.175739135008369e-06, "loss": 0.1352, "step": 39354 }, { "epoch": 28.831501831501832, "grad_norm": 0.85546875, "learning_rate": 9.173853013715776e-06, "loss": 0.1144, "step": 39355 }, { "epoch": 28.832234432234433, "grad_norm": 0.71875, "learning_rate": 9.171967090430333e-06, "loss": 0.1089, "step": 39356 }, { "epoch": 28.832967032967034, "grad_norm": 0.69140625, "learning_rate": 9.170081365161052e-06, "loss": 0.0854, "step": 39357 }, { "epoch": 28.833699633699634, "grad_norm": 1.2734375, "learning_rate": 9.168195837916976e-06, "loss": 0.1344, "step": 39358 }, { "epoch": 28.834432234432235, "grad_norm": 0.6328125, "learning_rate": 9.166310508707142e-06, "loss": 0.1035, "step": 39359 }, { "epoch": 28.835164835164836, "grad_norm": 0.7578125, "learning_rate": 9.164425377540596e-06, "loss": 0.116, "step": 39360 }, { "epoch": 28.835897435897436, "grad_norm": 1.0703125, "learning_rate": 9.162540444426348e-06, "loss": 0.13, "step": 39361 }, { "epoch": 28.836630036630037, "grad_norm": 1.2578125, "learning_rate": 9.16065570937344e-06, "loss": 0.1316, "step": 39362 }, { "epoch": 28.837362637362638, "grad_norm": 0.78515625, "learning_rate": 9.158771172390897e-06, "loss": 0.12, "step": 39363 }, { "epoch": 28.83809523809524, "grad_norm": 0.78125, "learning_rate": 9.156886833487768e-06, "loss": 0.1135, "step": 39364 }, { "epoch": 28.83882783882784, "grad_norm": 1.1640625, "learning_rate": 9.15500269267305e-06, "loss": 0.1409, "step": 39365 }, { "epoch": 28.83956043956044, "grad_norm": 1.0390625, "learning_rate": 9.153118749955785e-06, "loss": 0.1114, "step": 39366 }, { "epoch": 28.84029304029304, "grad_norm": 0.6640625, "learning_rate": 9.15123500534502e-06, "loss": 0.0907, "step": 39367 }, { "epoch": 28.84102564102564, "grad_norm": 1.0, "learning_rate": 9.149351458849735e-06, "loss": 0.1351, "step": 39368 }, { "epoch": 28.841758241758242, "grad_norm": 0.9140625, "learning_rate": 9.147468110479008e-06, "loss": 0.1375, "step": 39369 }, { "epoch": 28.842490842490843, "grad_norm": 1.6171875, "learning_rate": 9.145584960241822e-06, "loss": 0.1364, "step": 39370 }, { "epoch": 28.843223443223444, "grad_norm": 2.015625, "learning_rate": 9.143702008147226e-06, "loss": 0.12, "step": 39371 }, { "epoch": 28.843956043956045, "grad_norm": 0.8359375, "learning_rate": 9.14181925420422e-06, "loss": 0.1154, "step": 39372 }, { "epoch": 28.844688644688645, "grad_norm": 1.4296875, "learning_rate": 9.139936698421835e-06, "loss": 0.1314, "step": 39373 }, { "epoch": 28.845421245421246, "grad_norm": 0.875, "learning_rate": 9.138054340809087e-06, "loss": 0.1134, "step": 39374 }, { "epoch": 28.846153846153847, "grad_norm": 0.828125, "learning_rate": 9.136172181375013e-06, "loss": 0.1012, "step": 39375 }, { "epoch": 28.846886446886447, "grad_norm": 0.93359375, "learning_rate": 9.134290220128607e-06, "loss": 0.104, "step": 39376 }, { "epoch": 28.847619047619048, "grad_norm": 0.8125, "learning_rate": 9.132408457078897e-06, "loss": 0.1094, "step": 39377 }, { "epoch": 28.84835164835165, "grad_norm": 1.125, "learning_rate": 9.130526892234906e-06, "loss": 0.1557, "step": 39378 }, { "epoch": 28.84908424908425, "grad_norm": 0.91796875, "learning_rate": 9.128645525605618e-06, "loss": 0.0893, "step": 39379 }, { "epoch": 28.84981684981685, "grad_norm": 0.9609375, "learning_rate": 9.126764357200094e-06, "loss": 0.1225, "step": 39380 }, { "epoch": 28.85054945054945, "grad_norm": 1.140625, "learning_rate": 9.124883387027317e-06, "loss": 0.1175, "step": 39381 }, { "epoch": 28.851282051282052, "grad_norm": 1.46875, "learning_rate": 9.123002615096317e-06, "loss": 0.1113, "step": 39382 }, { "epoch": 28.852014652014653, "grad_norm": 1.0234375, "learning_rate": 9.121122041416074e-06, "loss": 0.1429, "step": 39383 }, { "epoch": 28.852747252747253, "grad_norm": 0.7109375, "learning_rate": 9.119241665995628e-06, "loss": 0.0984, "step": 39384 }, { "epoch": 28.853479853479854, "grad_norm": 0.6875, "learning_rate": 9.117361488843981e-06, "loss": 0.1125, "step": 39385 }, { "epoch": 28.854212454212455, "grad_norm": 1.140625, "learning_rate": 9.115481509970147e-06, "loss": 0.0978, "step": 39386 }, { "epoch": 28.854945054945055, "grad_norm": 0.90234375, "learning_rate": 9.113601729383115e-06, "loss": 0.1228, "step": 39387 }, { "epoch": 28.855677655677656, "grad_norm": 0.75, "learning_rate": 9.111722147091903e-06, "loss": 0.132, "step": 39388 }, { "epoch": 28.856410256410257, "grad_norm": 1.2421875, "learning_rate": 9.109842763105521e-06, "loss": 0.1158, "step": 39389 }, { "epoch": 28.857142857142858, "grad_norm": 1.015625, "learning_rate": 9.107963577432977e-06, "loss": 0.1086, "step": 39390 }, { "epoch": 28.85787545787546, "grad_norm": 0.890625, "learning_rate": 9.106084590083262e-06, "loss": 0.108, "step": 39391 }, { "epoch": 28.85860805860806, "grad_norm": 1.015625, "learning_rate": 9.104205801065376e-06, "loss": 0.1289, "step": 39392 }, { "epoch": 28.85934065934066, "grad_norm": 1.3125, "learning_rate": 9.102327210388349e-06, "loss": 0.1035, "step": 39393 }, { "epoch": 28.86007326007326, "grad_norm": 0.85546875, "learning_rate": 9.100448818061134e-06, "loss": 0.1051, "step": 39394 }, { "epoch": 28.86080586080586, "grad_norm": 0.9453125, "learning_rate": 9.098570624092782e-06, "loss": 0.1052, "step": 39395 }, { "epoch": 28.861538461538462, "grad_norm": 1.109375, "learning_rate": 9.096692628492262e-06, "loss": 0.1247, "step": 39396 }, { "epoch": 28.862271062271063, "grad_norm": 0.95703125, "learning_rate": 9.094814831268591e-06, "loss": 0.1255, "step": 39397 }, { "epoch": 28.863003663003663, "grad_norm": 0.91796875, "learning_rate": 9.09293723243074e-06, "loss": 0.1031, "step": 39398 }, { "epoch": 28.863736263736264, "grad_norm": 1.140625, "learning_rate": 9.091059831987719e-06, "loss": 0.1149, "step": 39399 }, { "epoch": 28.864468864468865, "grad_norm": 1.4296875, "learning_rate": 9.089182629948527e-06, "loss": 0.15, "step": 39400 }, { "epoch": 28.865201465201466, "grad_norm": 0.88671875, "learning_rate": 9.087305626322166e-06, "loss": 0.0844, "step": 39401 }, { "epoch": 28.865934065934066, "grad_norm": 1.21875, "learning_rate": 9.085428821117606e-06, "loss": 0.1393, "step": 39402 }, { "epoch": 28.866666666666667, "grad_norm": 1.0, "learning_rate": 9.083552214343853e-06, "loss": 0.1004, "step": 39403 }, { "epoch": 28.867399267399268, "grad_norm": 1.515625, "learning_rate": 9.081675806009908e-06, "loss": 0.0988, "step": 39404 }, { "epoch": 28.86813186813187, "grad_norm": 1.234375, "learning_rate": 9.079799596124729e-06, "loss": 0.119, "step": 39405 }, { "epoch": 28.86886446886447, "grad_norm": 1.0078125, "learning_rate": 9.077923584697348e-06, "loss": 0.1096, "step": 39406 }, { "epoch": 28.86959706959707, "grad_norm": 0.93359375, "learning_rate": 9.076047771736727e-06, "loss": 0.1223, "step": 39407 }, { "epoch": 28.87032967032967, "grad_norm": 1.59375, "learning_rate": 9.074172157251865e-06, "loss": 0.1035, "step": 39408 }, { "epoch": 28.87106227106227, "grad_norm": 1.0703125, "learning_rate": 9.072296741251738e-06, "loss": 0.1185, "step": 39409 }, { "epoch": 28.871794871794872, "grad_norm": 1.5, "learning_rate": 9.070421523745335e-06, "loss": 0.1189, "step": 39410 }, { "epoch": 28.872527472527473, "grad_norm": 1.3046875, "learning_rate": 9.068546504741638e-06, "loss": 0.112, "step": 39411 }, { "epoch": 28.873260073260074, "grad_norm": 0.73828125, "learning_rate": 9.066671684249652e-06, "loss": 0.1241, "step": 39412 }, { "epoch": 28.873992673992674, "grad_norm": 0.94140625, "learning_rate": 9.06479706227833e-06, "loss": 0.1196, "step": 39413 }, { "epoch": 28.874725274725275, "grad_norm": 0.890625, "learning_rate": 9.06292263883667e-06, "loss": 0.1146, "step": 39414 }, { "epoch": 28.875457875457876, "grad_norm": 1.515625, "learning_rate": 9.06104841393365e-06, "loss": 0.126, "step": 39415 }, { "epoch": 28.876190476190477, "grad_norm": 1.34375, "learning_rate": 9.05917438757826e-06, "loss": 0.1164, "step": 39416 }, { "epoch": 28.876923076923077, "grad_norm": 1.2265625, "learning_rate": 9.05730055977946e-06, "loss": 0.1412, "step": 39417 }, { "epoch": 28.877655677655678, "grad_norm": 1.125, "learning_rate": 9.05542693054624e-06, "loss": 0.1356, "step": 39418 }, { "epoch": 28.87838827838828, "grad_norm": 0.7421875, "learning_rate": 9.05355349988758e-06, "loss": 0.1111, "step": 39419 }, { "epoch": 28.87912087912088, "grad_norm": 0.93359375, "learning_rate": 9.051680267812433e-06, "loss": 0.1237, "step": 39420 }, { "epoch": 28.87985347985348, "grad_norm": 0.703125, "learning_rate": 9.049807234329815e-06, "loss": 0.1126, "step": 39421 }, { "epoch": 28.88058608058608, "grad_norm": 1.09375, "learning_rate": 9.047934399448664e-06, "loss": 0.0935, "step": 39422 }, { "epoch": 28.88131868131868, "grad_norm": 0.83203125, "learning_rate": 9.046061763177983e-06, "loss": 0.1013, "step": 39423 }, { "epoch": 28.882051282051282, "grad_norm": 0.984375, "learning_rate": 9.044189325526714e-06, "loss": 0.1247, "step": 39424 }, { "epoch": 28.882783882783883, "grad_norm": 0.80078125, "learning_rate": 9.042317086503846e-06, "loss": 0.1144, "step": 39425 }, { "epoch": 28.883516483516484, "grad_norm": 0.796875, "learning_rate": 9.04044504611834e-06, "loss": 0.1164, "step": 39426 }, { "epoch": 28.884249084249085, "grad_norm": 0.8828125, "learning_rate": 9.038573204379185e-06, "loss": 0.112, "step": 39427 }, { "epoch": 28.884981684981685, "grad_norm": 1.296875, "learning_rate": 9.036701561295331e-06, "loss": 0.0933, "step": 39428 }, { "epoch": 28.885714285714286, "grad_norm": 0.76171875, "learning_rate": 9.034830116875748e-06, "loss": 0.1031, "step": 39429 }, { "epoch": 28.886446886446887, "grad_norm": 1.65625, "learning_rate": 9.032958871129418e-06, "loss": 0.1345, "step": 39430 }, { "epoch": 28.887179487179488, "grad_norm": 0.85546875, "learning_rate": 9.03108782406527e-06, "loss": 0.1384, "step": 39431 }, { "epoch": 28.88791208791209, "grad_norm": 1.1640625, "learning_rate": 9.02921697569232e-06, "loss": 0.1134, "step": 39432 }, { "epoch": 28.88864468864469, "grad_norm": 0.890625, "learning_rate": 9.027346326019486e-06, "loss": 0.1214, "step": 39433 }, { "epoch": 28.88937728937729, "grad_norm": 1.3515625, "learning_rate": 9.025475875055769e-06, "loss": 0.123, "step": 39434 }, { "epoch": 28.89010989010989, "grad_norm": 2.109375, "learning_rate": 9.0236056228101e-06, "loss": 0.1374, "step": 39435 }, { "epoch": 28.89084249084249, "grad_norm": 1.1015625, "learning_rate": 9.021735569291451e-06, "loss": 0.1324, "step": 39436 }, { "epoch": 28.891575091575092, "grad_norm": 0.86328125, "learning_rate": 9.019865714508785e-06, "loss": 0.1341, "step": 39437 }, { "epoch": 28.892307692307693, "grad_norm": 0.8671875, "learning_rate": 9.017996058471069e-06, "loss": 0.1358, "step": 39438 }, { "epoch": 28.893040293040293, "grad_norm": 1.015625, "learning_rate": 9.016126601187246e-06, "loss": 0.1063, "step": 39439 }, { "epoch": 28.893772893772894, "grad_norm": 1.3515625, "learning_rate": 9.014257342666271e-06, "loss": 0.1401, "step": 39440 }, { "epoch": 28.894505494505495, "grad_norm": 1.203125, "learning_rate": 9.012388282917111e-06, "loss": 0.1098, "step": 39441 }, { "epoch": 28.895238095238096, "grad_norm": 1.734375, "learning_rate": 9.010519421948727e-06, "loss": 0.1165, "step": 39442 }, { "epoch": 28.895970695970696, "grad_norm": 0.79296875, "learning_rate": 9.008650759770057e-06, "loss": 0.1096, "step": 39443 }, { "epoch": 28.896703296703297, "grad_norm": 0.9140625, "learning_rate": 9.006782296390058e-06, "loss": 0.0906, "step": 39444 }, { "epoch": 28.897435897435898, "grad_norm": 1.03125, "learning_rate": 9.004914031817702e-06, "loss": 0.1247, "step": 39445 }, { "epoch": 28.8981684981685, "grad_norm": 1.109375, "learning_rate": 9.0030459660619e-06, "loss": 0.1309, "step": 39446 }, { "epoch": 28.8989010989011, "grad_norm": 1.078125, "learning_rate": 9.001178099131648e-06, "loss": 0.1193, "step": 39447 }, { "epoch": 28.8996336996337, "grad_norm": 1.1015625, "learning_rate": 8.999310431035866e-06, "loss": 0.1074, "step": 39448 }, { "epoch": 28.9003663003663, "grad_norm": 1.203125, "learning_rate": 8.997442961783524e-06, "loss": 0.1328, "step": 39449 }, { "epoch": 28.9010989010989, "grad_norm": 0.68359375, "learning_rate": 8.99557569138354e-06, "loss": 0.1191, "step": 39450 }, { "epoch": 28.901831501831502, "grad_norm": 1.1171875, "learning_rate": 8.993708619844882e-06, "loss": 0.1188, "step": 39451 }, { "epoch": 28.902564102564103, "grad_norm": 0.83984375, "learning_rate": 8.991841747176494e-06, "loss": 0.0903, "step": 39452 }, { "epoch": 28.903296703296704, "grad_norm": 0.828125, "learning_rate": 8.989975073387328e-06, "loss": 0.1143, "step": 39453 }, { "epoch": 28.904029304029304, "grad_norm": 0.80859375, "learning_rate": 8.98810859848631e-06, "loss": 0.1055, "step": 39454 }, { "epoch": 28.904761904761905, "grad_norm": 1.140625, "learning_rate": 8.986242322482384e-06, "loss": 0.1498, "step": 39455 }, { "epoch": 28.905494505494506, "grad_norm": 0.69921875, "learning_rate": 8.984376245384516e-06, "loss": 0.1177, "step": 39456 }, { "epoch": 28.906227106227107, "grad_norm": 1.1171875, "learning_rate": 8.982510367201603e-06, "loss": 0.122, "step": 39457 }, { "epoch": 28.906959706959707, "grad_norm": 0.67578125, "learning_rate": 8.98064468794264e-06, "loss": 0.1104, "step": 39458 }, { "epoch": 28.907692307692308, "grad_norm": 0.765625, "learning_rate": 8.978779207616522e-06, "loss": 0.1295, "step": 39459 }, { "epoch": 28.90842490842491, "grad_norm": 0.64453125, "learning_rate": 8.97691392623222e-06, "loss": 0.1083, "step": 39460 }, { "epoch": 28.90915750915751, "grad_norm": 1.1328125, "learning_rate": 8.97504884379864e-06, "loss": 0.1194, "step": 39461 }, { "epoch": 28.90989010989011, "grad_norm": 0.92578125, "learning_rate": 8.973183960324728e-06, "loss": 0.1048, "step": 39462 }, { "epoch": 28.91062271062271, "grad_norm": 0.6796875, "learning_rate": 8.97131927581943e-06, "loss": 0.1234, "step": 39463 }, { "epoch": 28.91135531135531, "grad_norm": 0.9140625, "learning_rate": 8.969454790291688e-06, "loss": 0.1304, "step": 39464 }, { "epoch": 28.912087912087912, "grad_norm": 1.1796875, "learning_rate": 8.967590503750407e-06, "loss": 0.1076, "step": 39465 }, { "epoch": 28.912820512820513, "grad_norm": 0.9765625, "learning_rate": 8.965726416204534e-06, "loss": 0.1291, "step": 39466 }, { "epoch": 28.913553113553114, "grad_norm": 1.734375, "learning_rate": 8.963862527663e-06, "loss": 0.1376, "step": 39467 }, { "epoch": 28.914285714285715, "grad_norm": 1.03125, "learning_rate": 8.961998838134733e-06, "loss": 0.1245, "step": 39468 }, { "epoch": 28.915018315018315, "grad_norm": 1.40625, "learning_rate": 8.960135347628681e-06, "loss": 0.1123, "step": 39469 }, { "epoch": 28.915750915750916, "grad_norm": 1.203125, "learning_rate": 8.958272056153748e-06, "loss": 0.1301, "step": 39470 }, { "epoch": 28.916483516483517, "grad_norm": 0.85546875, "learning_rate": 8.95640896371888e-06, "loss": 0.1286, "step": 39471 }, { "epoch": 28.917216117216118, "grad_norm": 0.92578125, "learning_rate": 8.954546070332969e-06, "loss": 0.1462, "step": 39472 }, { "epoch": 28.91794871794872, "grad_norm": 0.578125, "learning_rate": 8.952683376004993e-06, "loss": 0.1166, "step": 39473 }, { "epoch": 28.91868131868132, "grad_norm": 1.484375, "learning_rate": 8.950820880743836e-06, "loss": 0.1676, "step": 39474 }, { "epoch": 28.91941391941392, "grad_norm": 0.64453125, "learning_rate": 8.948958584558447e-06, "loss": 0.1242, "step": 39475 }, { "epoch": 28.92014652014652, "grad_norm": 1.2578125, "learning_rate": 8.947096487457727e-06, "loss": 0.111, "step": 39476 }, { "epoch": 28.92087912087912, "grad_norm": 0.93359375, "learning_rate": 8.945234589450611e-06, "loss": 0.1089, "step": 39477 }, { "epoch": 28.921611721611722, "grad_norm": 0.96875, "learning_rate": 8.943372890546016e-06, "loss": 0.1008, "step": 39478 }, { "epoch": 28.922344322344323, "grad_norm": 1.296875, "learning_rate": 8.941511390752874e-06, "loss": 0.1217, "step": 39479 }, { "epoch": 28.923076923076923, "grad_norm": 1.109375, "learning_rate": 8.939650090080084e-06, "loss": 0.0902, "step": 39480 }, { "epoch": 28.923809523809524, "grad_norm": 0.86328125, "learning_rate": 8.937788988536572e-06, "loss": 0.1244, "step": 39481 }, { "epoch": 28.924542124542125, "grad_norm": 1.59375, "learning_rate": 8.935928086131272e-06, "loss": 0.1541, "step": 39482 }, { "epoch": 28.925274725274726, "grad_norm": 0.97265625, "learning_rate": 8.934067382873057e-06, "loss": 0.1206, "step": 39483 }, { "epoch": 28.926007326007326, "grad_norm": 1.171875, "learning_rate": 8.9322068787709e-06, "loss": 0.1191, "step": 39484 }, { "epoch": 28.926739926739927, "grad_norm": 1.4296875, "learning_rate": 8.930346573833665e-06, "loss": 0.1471, "step": 39485 }, { "epoch": 28.927472527472528, "grad_norm": 1.0625, "learning_rate": 8.928486468070303e-06, "loss": 0.135, "step": 39486 }, { "epoch": 28.92820512820513, "grad_norm": 0.90625, "learning_rate": 8.926626561489699e-06, "loss": 0.1254, "step": 39487 }, { "epoch": 28.92893772893773, "grad_norm": 1.3359375, "learning_rate": 8.924766854100775e-06, "loss": 0.147, "step": 39488 }, { "epoch": 28.92967032967033, "grad_norm": 0.921875, "learning_rate": 8.92290734591244e-06, "loss": 0.1298, "step": 39489 }, { "epoch": 28.93040293040293, "grad_norm": 0.81640625, "learning_rate": 8.92104803693361e-06, "loss": 0.099, "step": 39490 }, { "epoch": 28.93113553113553, "grad_norm": 0.828125, "learning_rate": 8.919188927173184e-06, "loss": 0.1218, "step": 39491 }, { "epoch": 28.931868131868132, "grad_norm": 1.1640625, "learning_rate": 8.917330016640076e-06, "loss": 0.1089, "step": 39492 }, { "epoch": 28.932600732600733, "grad_norm": 0.94921875, "learning_rate": 8.915471305343186e-06, "loss": 0.149, "step": 39493 }, { "epoch": 28.933333333333334, "grad_norm": 1.1484375, "learning_rate": 8.913612793291422e-06, "loss": 0.1353, "step": 39494 }, { "epoch": 28.934065934065934, "grad_norm": 0.8828125, "learning_rate": 8.91175448049371e-06, "loss": 0.1074, "step": 39495 }, { "epoch": 28.934798534798535, "grad_norm": 0.72265625, "learning_rate": 8.909896366958915e-06, "loss": 0.1065, "step": 39496 }, { "epoch": 28.935531135531136, "grad_norm": 0.92578125, "learning_rate": 8.908038452695977e-06, "loss": 0.1045, "step": 39497 }, { "epoch": 28.936263736263736, "grad_norm": 0.8828125, "learning_rate": 8.906180737713755e-06, "loss": 0.1292, "step": 39498 }, { "epoch": 28.936996336996337, "grad_norm": 0.72265625, "learning_rate": 8.904323222021203e-06, "loss": 0.1088, "step": 39499 }, { "epoch": 28.937728937728938, "grad_norm": 0.7734375, "learning_rate": 8.902465905627176e-06, "loss": 0.1169, "step": 39500 }, { "epoch": 28.93846153846154, "grad_norm": 0.75390625, "learning_rate": 8.900608788540605e-06, "loss": 0.0995, "step": 39501 }, { "epoch": 28.93919413919414, "grad_norm": 1.0, "learning_rate": 8.898751870770359e-06, "loss": 0.1382, "step": 39502 }, { "epoch": 28.93992673992674, "grad_norm": 0.97265625, "learning_rate": 8.896895152325354e-06, "loss": 0.1515, "step": 39503 }, { "epoch": 28.94065934065934, "grad_norm": 1.046875, "learning_rate": 8.89503863321448e-06, "loss": 0.1158, "step": 39504 }, { "epoch": 28.94139194139194, "grad_norm": 1.0703125, "learning_rate": 8.893182313446645e-06, "loss": 0.1294, "step": 39505 }, { "epoch": 28.942124542124542, "grad_norm": 1.9921875, "learning_rate": 8.89132619303072e-06, "loss": 0.152, "step": 39506 }, { "epoch": 28.942857142857143, "grad_norm": 1.140625, "learning_rate": 8.889470271975606e-06, "loss": 0.1241, "step": 39507 }, { "epoch": 28.943589743589744, "grad_norm": 0.84765625, "learning_rate": 8.887614550290219e-06, "loss": 0.1137, "step": 39508 }, { "epoch": 28.944322344322345, "grad_norm": 1.3515625, "learning_rate": 8.885759027983406e-06, "loss": 0.135, "step": 39509 }, { "epoch": 28.945054945054945, "grad_norm": 0.97265625, "learning_rate": 8.883903705064103e-06, "loss": 0.108, "step": 39510 }, { "epoch": 28.945787545787546, "grad_norm": 1.1953125, "learning_rate": 8.882048581541172e-06, "loss": 0.1353, "step": 39511 }, { "epoch": 28.946520146520147, "grad_norm": 1.0546875, "learning_rate": 8.880193657423517e-06, "loss": 0.122, "step": 39512 }, { "epoch": 28.947252747252747, "grad_norm": 1.0703125, "learning_rate": 8.878338932720005e-06, "loss": 0.0962, "step": 39513 }, { "epoch": 28.947985347985348, "grad_norm": 1.125, "learning_rate": 8.876484407439538e-06, "loss": 0.1215, "step": 39514 }, { "epoch": 28.94871794871795, "grad_norm": 0.65234375, "learning_rate": 8.874630081590993e-06, "loss": 0.145, "step": 39515 }, { "epoch": 28.94945054945055, "grad_norm": 1.4609375, "learning_rate": 8.872775955183275e-06, "loss": 0.1611, "step": 39516 }, { "epoch": 28.95018315018315, "grad_norm": 0.90625, "learning_rate": 8.870922028225242e-06, "loss": 0.1241, "step": 39517 }, { "epoch": 28.95091575091575, "grad_norm": 1.0859375, "learning_rate": 8.869068300725787e-06, "loss": 0.1145, "step": 39518 }, { "epoch": 28.951648351648352, "grad_norm": 0.91796875, "learning_rate": 8.867214772693789e-06, "loss": 0.1224, "step": 39519 }, { "epoch": 28.952380952380953, "grad_norm": 1.140625, "learning_rate": 8.865361444138136e-06, "loss": 0.1081, "step": 39520 }, { "epoch": 28.953113553113553, "grad_norm": 1.03125, "learning_rate": 8.863508315067712e-06, "loss": 0.1506, "step": 39521 }, { "epoch": 28.953846153846154, "grad_norm": 1.3984375, "learning_rate": 8.861655385491381e-06, "loss": 0.1464, "step": 39522 }, { "epoch": 28.954578754578755, "grad_norm": 1.21875, "learning_rate": 8.859802655418042e-06, "loss": 0.1464, "step": 39523 }, { "epoch": 28.955311355311355, "grad_norm": 0.96875, "learning_rate": 8.857950124856524e-06, "loss": 0.1095, "step": 39524 }, { "epoch": 28.956043956043956, "grad_norm": 1.0625, "learning_rate": 8.856097793815768e-06, "loss": 0.1136, "step": 39525 }, { "epoch": 28.956776556776557, "grad_norm": 0.91796875, "learning_rate": 8.854245662304605e-06, "loss": 0.131, "step": 39526 }, { "epoch": 28.957509157509158, "grad_norm": 0.99609375, "learning_rate": 8.852393730331934e-06, "loss": 0.1184, "step": 39527 }, { "epoch": 28.95824175824176, "grad_norm": 1.421875, "learning_rate": 8.850541997906606e-06, "loss": 0.1173, "step": 39528 }, { "epoch": 28.95897435897436, "grad_norm": 1.2578125, "learning_rate": 8.848690465037505e-06, "loss": 0.1191, "step": 39529 }, { "epoch": 28.95970695970696, "grad_norm": 0.625, "learning_rate": 8.8468391317335e-06, "loss": 0.1197, "step": 39530 }, { "epoch": 28.96043956043956, "grad_norm": 0.62890625, "learning_rate": 8.84498799800347e-06, "loss": 0.1091, "step": 39531 }, { "epoch": 28.96117216117216, "grad_norm": 0.8828125, "learning_rate": 8.84313706385627e-06, "loss": 0.1486, "step": 39532 }, { "epoch": 28.961904761904762, "grad_norm": 0.86328125, "learning_rate": 8.841286329300773e-06, "loss": 0.117, "step": 39533 }, { "epoch": 28.962637362637363, "grad_norm": 1.0390625, "learning_rate": 8.839435794345864e-06, "loss": 0.1264, "step": 39534 }, { "epoch": 28.963369963369964, "grad_norm": 0.81640625, "learning_rate": 8.83758545900037e-06, "loss": 0.1206, "step": 39535 }, { "epoch": 28.964102564102564, "grad_norm": 0.66015625, "learning_rate": 8.835735323273202e-06, "loss": 0.1154, "step": 39536 }, { "epoch": 28.964835164835165, "grad_norm": 1.09375, "learning_rate": 8.833885387173191e-06, "loss": 0.1065, "step": 39537 }, { "epoch": 28.965567765567766, "grad_norm": 0.9140625, "learning_rate": 8.832035650709228e-06, "loss": 0.1268, "step": 39538 }, { "epoch": 28.966300366300366, "grad_norm": 1.296875, "learning_rate": 8.830186113890148e-06, "loss": 0.1455, "step": 39539 }, { "epoch": 28.967032967032967, "grad_norm": 1.2578125, "learning_rate": 8.828336776724828e-06, "loss": 0.1153, "step": 39540 }, { "epoch": 28.967765567765568, "grad_norm": 0.89453125, "learning_rate": 8.826487639222125e-06, "loss": 0.1283, "step": 39541 }, { "epoch": 28.96849816849817, "grad_norm": 1.2890625, "learning_rate": 8.824638701390911e-06, "loss": 0.1222, "step": 39542 }, { "epoch": 28.96923076923077, "grad_norm": 0.828125, "learning_rate": 8.822789963240021e-06, "loss": 0.1428, "step": 39543 }, { "epoch": 28.96996336996337, "grad_norm": 0.9375, "learning_rate": 8.820941424778325e-06, "loss": 0.1287, "step": 39544 }, { "epoch": 28.97069597069597, "grad_norm": 1.3515625, "learning_rate": 8.819093086014686e-06, "loss": 0.1259, "step": 39545 }, { "epoch": 28.97142857142857, "grad_norm": 1.2421875, "learning_rate": 8.817244946957954e-06, "loss": 0.1193, "step": 39546 }, { "epoch": 28.972161172161172, "grad_norm": 1.5078125, "learning_rate": 8.815397007616997e-06, "loss": 0.1107, "step": 39547 }, { "epoch": 28.972893772893773, "grad_norm": 0.953125, "learning_rate": 8.813549268000642e-06, "loss": 0.1225, "step": 39548 }, { "epoch": 28.973626373626374, "grad_norm": 1.109375, "learning_rate": 8.811701728117772e-06, "loss": 0.1166, "step": 39549 }, { "epoch": 28.974358974358974, "grad_norm": 1.1015625, "learning_rate": 8.809854387977198e-06, "loss": 0.1188, "step": 39550 }, { "epoch": 28.975091575091575, "grad_norm": 0.6875, "learning_rate": 8.808007247587822e-06, "loss": 0.0998, "step": 39551 }, { "epoch": 28.975824175824176, "grad_norm": 1.2265625, "learning_rate": 8.806160306958456e-06, "loss": 0.1075, "step": 39552 }, { "epoch": 28.976556776556777, "grad_norm": 0.83203125, "learning_rate": 8.804313566097979e-06, "loss": 0.0963, "step": 39553 }, { "epoch": 28.977289377289377, "grad_norm": 1.6875, "learning_rate": 8.802467025015213e-06, "loss": 0.1432, "step": 39554 }, { "epoch": 28.978021978021978, "grad_norm": 1.6640625, "learning_rate": 8.800620683719011e-06, "loss": 0.1553, "step": 39555 }, { "epoch": 28.97875457875458, "grad_norm": 0.91015625, "learning_rate": 8.79877454221823e-06, "loss": 0.1119, "step": 39556 }, { "epoch": 28.97948717948718, "grad_norm": 1.515625, "learning_rate": 8.796928600521716e-06, "loss": 0.1235, "step": 39557 }, { "epoch": 28.98021978021978, "grad_norm": 0.7265625, "learning_rate": 8.7950828586383e-06, "loss": 0.1235, "step": 39558 }, { "epoch": 28.98095238095238, "grad_norm": 1.15625, "learning_rate": 8.79323731657683e-06, "loss": 0.1214, "step": 39559 }, { "epoch": 28.98168498168498, "grad_norm": 1.0859375, "learning_rate": 8.79139197434616e-06, "loss": 0.1235, "step": 39560 }, { "epoch": 28.982417582417582, "grad_norm": 0.84375, "learning_rate": 8.789546831955105e-06, "loss": 0.1232, "step": 39561 }, { "epoch": 28.983150183150183, "grad_norm": 0.97265625, "learning_rate": 8.787701889412547e-06, "loss": 0.1134, "step": 39562 }, { "epoch": 28.983882783882784, "grad_norm": 1.109375, "learning_rate": 8.785857146727288e-06, "loss": 0.119, "step": 39563 }, { "epoch": 28.984615384615385, "grad_norm": 0.82421875, "learning_rate": 8.784012603908193e-06, "loss": 0.1346, "step": 39564 }, { "epoch": 28.985347985347985, "grad_norm": 0.8203125, "learning_rate": 8.782168260964078e-06, "loss": 0.1125, "step": 39565 }, { "epoch": 28.986080586080586, "grad_norm": 1.0546875, "learning_rate": 8.78032411790379e-06, "loss": 0.1367, "step": 39566 }, { "epoch": 28.986813186813187, "grad_norm": 0.94921875, "learning_rate": 8.778480174736162e-06, "loss": 0.1452, "step": 39567 }, { "epoch": 28.987545787545788, "grad_norm": 1.46875, "learning_rate": 8.77663643147005e-06, "loss": 0.1758, "step": 39568 }, { "epoch": 28.98827838827839, "grad_norm": 0.9375, "learning_rate": 8.774792888114253e-06, "loss": 0.1158, "step": 39569 }, { "epoch": 28.98901098901099, "grad_norm": 0.92578125, "learning_rate": 8.77294954467762e-06, "loss": 0.1056, "step": 39570 }, { "epoch": 28.98974358974359, "grad_norm": 1.203125, "learning_rate": 8.771106401168986e-06, "loss": 0.1547, "step": 39571 }, { "epoch": 28.99047619047619, "grad_norm": 1.0546875, "learning_rate": 8.769263457597181e-06, "loss": 0.1241, "step": 39572 }, { "epoch": 28.99120879120879, "grad_norm": 0.81640625, "learning_rate": 8.767420713971044e-06, "loss": 0.1135, "step": 39573 }, { "epoch": 28.991941391941392, "grad_norm": 0.98046875, "learning_rate": 8.765578170299384e-06, "loss": 0.0998, "step": 39574 }, { "epoch": 28.992673992673993, "grad_norm": 1.2421875, "learning_rate": 8.763735826591047e-06, "loss": 0.1233, "step": 39575 }, { "epoch": 28.993406593406593, "grad_norm": 0.74609375, "learning_rate": 8.761893682854832e-06, "loss": 0.1044, "step": 39576 }, { "epoch": 28.994139194139194, "grad_norm": 1.7734375, "learning_rate": 8.760051739099608e-06, "loss": 0.1177, "step": 39577 }, { "epoch": 28.994871794871795, "grad_norm": 0.87890625, "learning_rate": 8.758209995334172e-06, "loss": 0.1136, "step": 39578 }, { "epoch": 28.995604395604396, "grad_norm": 1.0703125, "learning_rate": 8.756368451567361e-06, "loss": 0.1197, "step": 39579 }, { "epoch": 28.996336996336996, "grad_norm": 1.2265625, "learning_rate": 8.754527107807985e-06, "loss": 0.1312, "step": 39580 }, { "epoch": 28.997069597069597, "grad_norm": 1.09375, "learning_rate": 8.752685964064871e-06, "loss": 0.1191, "step": 39581 }, { "epoch": 28.997802197802198, "grad_norm": 1.9296875, "learning_rate": 8.750845020346847e-06, "loss": 0.1607, "step": 39582 }, { "epoch": 28.9985347985348, "grad_norm": 0.74609375, "learning_rate": 8.749004276662737e-06, "loss": 0.1009, "step": 39583 }, { "epoch": 28.9992673992674, "grad_norm": 1.390625, "learning_rate": 8.74716373302134e-06, "loss": 0.1162, "step": 39584 }, { "epoch": 29.0, "grad_norm": 1.1953125, "learning_rate": 8.745323389431495e-06, "loss": 0.1096, "step": 39585 }, { "epoch": 29.0007326007326, "grad_norm": 1.140625, "learning_rate": 8.743483245902018e-06, "loss": 0.1125, "step": 39586 }, { "epoch": 29.0014652014652, "grad_norm": 1.453125, "learning_rate": 8.741643302441701e-06, "loss": 0.1235, "step": 39587 }, { "epoch": 29.002197802197802, "grad_norm": 0.71875, "learning_rate": 8.739803559059403e-06, "loss": 0.1179, "step": 39588 }, { "epoch": 29.002930402930403, "grad_norm": 1.0546875, "learning_rate": 8.737964015763898e-06, "loss": 0.112, "step": 39589 }, { "epoch": 29.003663003663004, "grad_norm": 0.66796875, "learning_rate": 8.736124672564037e-06, "loss": 0.1159, "step": 39590 }, { "epoch": 29.004395604395604, "grad_norm": 0.78125, "learning_rate": 8.734285529468594e-06, "loss": 0.1426, "step": 39591 }, { "epoch": 29.005128205128205, "grad_norm": 0.94921875, "learning_rate": 8.732446586486404e-06, "loss": 0.1245, "step": 39592 }, { "epoch": 29.005860805860806, "grad_norm": 1.265625, "learning_rate": 8.73060784362627e-06, "loss": 0.1205, "step": 39593 }, { "epoch": 29.006593406593407, "grad_norm": 1.09375, "learning_rate": 8.728769300897023e-06, "loss": 0.1293, "step": 39594 }, { "epoch": 29.007326007326007, "grad_norm": 1.125, "learning_rate": 8.726930958307442e-06, "loss": 0.1033, "step": 39595 }, { "epoch": 29.008058608058608, "grad_norm": 1.3046875, "learning_rate": 8.725092815866343e-06, "loss": 0.0904, "step": 39596 }, { "epoch": 29.00879120879121, "grad_norm": 0.98828125, "learning_rate": 8.723254873582544e-06, "loss": 0.0905, "step": 39597 }, { "epoch": 29.00952380952381, "grad_norm": 0.953125, "learning_rate": 8.721417131464842e-06, "loss": 0.1258, "step": 39598 }, { "epoch": 29.01025641025641, "grad_norm": 1.3046875, "learning_rate": 8.71957958952206e-06, "loss": 0.1424, "step": 39599 }, { "epoch": 29.01098901098901, "grad_norm": 1.4140625, "learning_rate": 8.71774224776297e-06, "loss": 0.1228, "step": 39600 }, { "epoch": 29.01172161172161, "grad_norm": 1.015625, "learning_rate": 8.715905106196408e-06, "loss": 0.1062, "step": 39601 }, { "epoch": 29.012454212454212, "grad_norm": 1.0234375, "learning_rate": 8.714068164831142e-06, "loss": 0.0856, "step": 39602 }, { "epoch": 29.013186813186813, "grad_norm": 1.3046875, "learning_rate": 8.712231423676011e-06, "loss": 0.1274, "step": 39603 }, { "epoch": 29.013919413919414, "grad_norm": 0.625, "learning_rate": 8.71039488273979e-06, "loss": 0.1159, "step": 39604 }, { "epoch": 29.014652014652015, "grad_norm": 1.0234375, "learning_rate": 8.708558542031294e-06, "loss": 0.123, "step": 39605 }, { "epoch": 29.015384615384615, "grad_norm": 0.72265625, "learning_rate": 8.706722401559299e-06, "loss": 0.1056, "step": 39606 }, { "epoch": 29.016117216117216, "grad_norm": 0.9921875, "learning_rate": 8.704886461332622e-06, "loss": 0.0926, "step": 39607 }, { "epoch": 29.016849816849817, "grad_norm": 0.8828125, "learning_rate": 8.703050721360056e-06, "loss": 0.1214, "step": 39608 }, { "epoch": 29.017582417582418, "grad_norm": 0.875, "learning_rate": 8.701215181650391e-06, "loss": 0.1138, "step": 39609 }, { "epoch": 29.01831501831502, "grad_norm": 1.3984375, "learning_rate": 8.699379842212438e-06, "loss": 0.1319, "step": 39610 }, { "epoch": 29.01904761904762, "grad_norm": 0.84765625, "learning_rate": 8.697544703054964e-06, "loss": 0.1219, "step": 39611 }, { "epoch": 29.01978021978022, "grad_norm": 0.80859375, "learning_rate": 8.695709764186794e-06, "loss": 0.1389, "step": 39612 }, { "epoch": 29.02051282051282, "grad_norm": 0.97265625, "learning_rate": 8.693875025616673e-06, "loss": 0.1401, "step": 39613 }, { "epoch": 29.02124542124542, "grad_norm": 0.89453125, "learning_rate": 8.692040487353446e-06, "loss": 0.1186, "step": 39614 }, { "epoch": 29.021978021978022, "grad_norm": 0.80078125, "learning_rate": 8.690206149405866e-06, "loss": 0.0991, "step": 39615 }, { "epoch": 29.022710622710623, "grad_norm": 1.6328125, "learning_rate": 8.688372011782746e-06, "loss": 0.0985, "step": 39616 }, { "epoch": 29.023443223443223, "grad_norm": 1.0078125, "learning_rate": 8.68653807449285e-06, "loss": 0.1438, "step": 39617 }, { "epoch": 29.024175824175824, "grad_norm": 0.9609375, "learning_rate": 8.684704337544978e-06, "loss": 0.1285, "step": 39618 }, { "epoch": 29.024908424908425, "grad_norm": 0.8359375, "learning_rate": 8.682870800947911e-06, "loss": 0.1118, "step": 39619 }, { "epoch": 29.025641025641026, "grad_norm": 0.9140625, "learning_rate": 8.681037464710452e-06, "loss": 0.1166, "step": 39620 }, { "epoch": 29.026373626373626, "grad_norm": 0.9765625, "learning_rate": 8.679204328841362e-06, "loss": 0.1199, "step": 39621 }, { "epoch": 29.027106227106227, "grad_norm": 1.3828125, "learning_rate": 8.677371393349429e-06, "loss": 0.1145, "step": 39622 }, { "epoch": 29.027838827838828, "grad_norm": 0.94921875, "learning_rate": 8.67553865824344e-06, "loss": 0.1433, "step": 39623 }, { "epoch": 29.02857142857143, "grad_norm": 1.2578125, "learning_rate": 8.673706123532174e-06, "loss": 0.1187, "step": 39624 }, { "epoch": 29.02930402930403, "grad_norm": 0.9296875, "learning_rate": 8.671873789224424e-06, "loss": 0.1128, "step": 39625 }, { "epoch": 29.03003663003663, "grad_norm": 0.58203125, "learning_rate": 8.670041655328948e-06, "loss": 0.0783, "step": 39626 }, { "epoch": 29.03076923076923, "grad_norm": 0.90234375, "learning_rate": 8.668209721854541e-06, "loss": 0.1035, "step": 39627 }, { "epoch": 29.03150183150183, "grad_norm": 1.125, "learning_rate": 8.666377988809957e-06, "loss": 0.1284, "step": 39628 }, { "epoch": 29.032234432234432, "grad_norm": 0.98828125, "learning_rate": 8.664546456204013e-06, "loss": 0.1259, "step": 39629 }, { "epoch": 29.032967032967033, "grad_norm": 1.0546875, "learning_rate": 8.662715124045443e-06, "loss": 0.1292, "step": 39630 }, { "epoch": 29.033699633699634, "grad_norm": 0.890625, "learning_rate": 8.660883992343056e-06, "loss": 0.1151, "step": 39631 }, { "epoch": 29.034432234432234, "grad_norm": 0.56640625, "learning_rate": 8.659053061105595e-06, "loss": 0.1124, "step": 39632 }, { "epoch": 29.035164835164835, "grad_norm": 1.0, "learning_rate": 8.657222330341846e-06, "loss": 0.115, "step": 39633 }, { "epoch": 29.035897435897436, "grad_norm": 1.7578125, "learning_rate": 8.655391800060578e-06, "loss": 0.1153, "step": 39634 }, { "epoch": 29.036630036630036, "grad_norm": 1.53125, "learning_rate": 8.653561470270568e-06, "loss": 0.1273, "step": 39635 }, { "epoch": 29.037362637362637, "grad_norm": 0.87890625, "learning_rate": 8.651731340980596e-06, "loss": 0.1144, "step": 39636 }, { "epoch": 29.038095238095238, "grad_norm": 1.5, "learning_rate": 8.649901412199405e-06, "loss": 0.1545, "step": 39637 }, { "epoch": 29.03882783882784, "grad_norm": 0.953125, "learning_rate": 8.648071683935784e-06, "loss": 0.112, "step": 39638 }, { "epoch": 29.03956043956044, "grad_norm": 1.2890625, "learning_rate": 8.646242156198463e-06, "loss": 0.1224, "step": 39639 }, { "epoch": 29.04029304029304, "grad_norm": 1.203125, "learning_rate": 8.644412828996264e-06, "loss": 0.1277, "step": 39640 }, { "epoch": 29.04102564102564, "grad_norm": 1.125, "learning_rate": 8.642583702337912e-06, "loss": 0.1101, "step": 39641 }, { "epoch": 29.04175824175824, "grad_norm": 0.94140625, "learning_rate": 8.640754776232194e-06, "loss": 0.1245, "step": 39642 }, { "epoch": 29.042490842490842, "grad_norm": 0.72265625, "learning_rate": 8.63892605068785e-06, "loss": 0.1207, "step": 39643 }, { "epoch": 29.043223443223443, "grad_norm": 0.88671875, "learning_rate": 8.63709752571365e-06, "loss": 0.0978, "step": 39644 }, { "epoch": 29.043956043956044, "grad_norm": 1.2265625, "learning_rate": 8.635269201318365e-06, "loss": 0.1314, "step": 39645 }, { "epoch": 29.044688644688645, "grad_norm": 1.3125, "learning_rate": 8.633441077510753e-06, "loss": 0.1407, "step": 39646 }, { "epoch": 29.045421245421245, "grad_norm": 0.74609375, "learning_rate": 8.631613154299563e-06, "loss": 0.1276, "step": 39647 }, { "epoch": 29.046153846153846, "grad_norm": 0.98828125, "learning_rate": 8.629785431693552e-06, "loss": 0.1184, "step": 39648 }, { "epoch": 29.046886446886447, "grad_norm": 0.8203125, "learning_rate": 8.627957909701493e-06, "loss": 0.1056, "step": 39649 }, { "epoch": 29.047619047619047, "grad_norm": 1.15625, "learning_rate": 8.626130588332128e-06, "loss": 0.1061, "step": 39650 }, { "epoch": 29.048351648351648, "grad_norm": 0.9140625, "learning_rate": 8.624303467594228e-06, "loss": 0.1183, "step": 39651 }, { "epoch": 29.04908424908425, "grad_norm": 1.3515625, "learning_rate": 8.622476547496525e-06, "loss": 0.1518, "step": 39652 }, { "epoch": 29.04981684981685, "grad_norm": 1.2734375, "learning_rate": 8.620649828047799e-06, "loss": 0.126, "step": 39653 }, { "epoch": 29.05054945054945, "grad_norm": 0.87109375, "learning_rate": 8.618823309256766e-06, "loss": 0.1207, "step": 39654 }, { "epoch": 29.05128205128205, "grad_norm": 1.359375, "learning_rate": 8.616996991132218e-06, "loss": 0.1663, "step": 39655 }, { "epoch": 29.052014652014652, "grad_norm": 1.2578125, "learning_rate": 8.615170873682877e-06, "loss": 0.1283, "step": 39656 }, { "epoch": 29.052747252747253, "grad_norm": 0.9375, "learning_rate": 8.613344956917515e-06, "loss": 0.115, "step": 39657 }, { "epoch": 29.053479853479853, "grad_norm": 1.2734375, "learning_rate": 8.611519240844856e-06, "loss": 0.16, "step": 39658 }, { "epoch": 29.054212454212454, "grad_norm": 0.75, "learning_rate": 8.60969372547366e-06, "loss": 0.1162, "step": 39659 }, { "epoch": 29.054945054945055, "grad_norm": 0.8671875, "learning_rate": 8.60786841081267e-06, "loss": 0.121, "step": 39660 }, { "epoch": 29.055677655677655, "grad_norm": 1.0078125, "learning_rate": 8.606043296870638e-06, "loss": 0.1226, "step": 39661 }, { "epoch": 29.056410256410256, "grad_norm": 1.015625, "learning_rate": 8.604218383656313e-06, "loss": 0.0879, "step": 39662 }, { "epoch": 29.057142857142857, "grad_norm": 0.9609375, "learning_rate": 8.602393671178421e-06, "loss": 0.1246, "step": 39663 }, { "epoch": 29.057875457875458, "grad_norm": 1.5078125, "learning_rate": 8.600569159445732e-06, "loss": 0.1377, "step": 39664 }, { "epoch": 29.05860805860806, "grad_norm": 1.78125, "learning_rate": 8.598744848466938e-06, "loss": 0.1481, "step": 39665 }, { "epoch": 29.05934065934066, "grad_norm": 0.89453125, "learning_rate": 8.596920738250846e-06, "loss": 0.1198, "step": 39666 }, { "epoch": 29.06007326007326, "grad_norm": 1.0546875, "learning_rate": 8.59509682880614e-06, "loss": 0.1291, "step": 39667 }, { "epoch": 29.06080586080586, "grad_norm": 0.87890625, "learning_rate": 8.593273120141602e-06, "loss": 0.1239, "step": 39668 }, { "epoch": 29.06153846153846, "grad_norm": 1.0546875, "learning_rate": 8.591449612265936e-06, "loss": 0.1184, "step": 39669 }, { "epoch": 29.062271062271062, "grad_norm": 1.2109375, "learning_rate": 8.589626305187893e-06, "loss": 0.1072, "step": 39670 }, { "epoch": 29.063003663003663, "grad_norm": 0.9609375, "learning_rate": 8.587803198916207e-06, "loss": 0.09, "step": 39671 }, { "epoch": 29.063736263736264, "grad_norm": 1.1015625, "learning_rate": 8.585980293459627e-06, "loss": 0.1129, "step": 39672 }, { "epoch": 29.064468864468864, "grad_norm": 1.0546875, "learning_rate": 8.58415758882686e-06, "loss": 0.1135, "step": 39673 }, { "epoch": 29.065201465201465, "grad_norm": 1.015625, "learning_rate": 8.58233508502666e-06, "loss": 0.1206, "step": 39674 }, { "epoch": 29.065934065934066, "grad_norm": 0.91015625, "learning_rate": 8.580512782067751e-06, "loss": 0.1126, "step": 39675 }, { "epoch": 29.066666666666666, "grad_norm": 2.09375, "learning_rate": 8.578690679958864e-06, "loss": 0.1678, "step": 39676 }, { "epoch": 29.067399267399267, "grad_norm": 0.91015625, "learning_rate": 8.576868778708746e-06, "loss": 0.1449, "step": 39677 }, { "epoch": 29.068131868131868, "grad_norm": 1.765625, "learning_rate": 8.575047078326101e-06, "loss": 0.1061, "step": 39678 }, { "epoch": 29.06886446886447, "grad_norm": 1.0390625, "learning_rate": 8.573225578819678e-06, "loss": 0.124, "step": 39679 }, { "epoch": 29.06959706959707, "grad_norm": 0.80078125, "learning_rate": 8.571404280198175e-06, "loss": 0.1027, "step": 39680 }, { "epoch": 29.07032967032967, "grad_norm": 0.66796875, "learning_rate": 8.569583182470359e-06, "loss": 0.1117, "step": 39681 }, { "epoch": 29.07106227106227, "grad_norm": 1.6015625, "learning_rate": 8.567762285644925e-06, "loss": 0.1207, "step": 39682 }, { "epoch": 29.07179487179487, "grad_norm": 1.1015625, "learning_rate": 8.565941589730617e-06, "loss": 0.1424, "step": 39683 }, { "epoch": 29.072527472527472, "grad_norm": 0.9140625, "learning_rate": 8.56412109473614e-06, "loss": 0.1261, "step": 39684 }, { "epoch": 29.073260073260073, "grad_norm": 0.79296875, "learning_rate": 8.56230080067022e-06, "loss": 0.116, "step": 39685 }, { "epoch": 29.073992673992674, "grad_norm": 1.1171875, "learning_rate": 8.56048070754159e-06, "loss": 0.1074, "step": 39686 }, { "epoch": 29.074725274725274, "grad_norm": 1.0078125, "learning_rate": 8.558660815358962e-06, "loss": 0.1353, "step": 39687 }, { "epoch": 29.075457875457875, "grad_norm": 0.80078125, "learning_rate": 8.556841124131067e-06, "loss": 0.1334, "step": 39688 }, { "epoch": 29.076190476190476, "grad_norm": 1.7734375, "learning_rate": 8.555021633866609e-06, "loss": 0.1154, "step": 39689 }, { "epoch": 29.076923076923077, "grad_norm": 1.21875, "learning_rate": 8.55320234457432e-06, "loss": 0.1462, "step": 39690 }, { "epoch": 29.077655677655677, "grad_norm": 1.28125, "learning_rate": 8.551383256262887e-06, "loss": 0.1207, "step": 39691 }, { "epoch": 29.078388278388278, "grad_norm": 0.95703125, "learning_rate": 8.549564368941075e-06, "loss": 0.1197, "step": 39692 }, { "epoch": 29.07912087912088, "grad_norm": 1.4609375, "learning_rate": 8.547745682617555e-06, "loss": 0.1473, "step": 39693 }, { "epoch": 29.07985347985348, "grad_norm": 1.3828125, "learning_rate": 8.545927197301068e-06, "loss": 0.1312, "step": 39694 }, { "epoch": 29.08058608058608, "grad_norm": 1.1953125, "learning_rate": 8.544108913000307e-06, "loss": 0.1252, "step": 39695 }, { "epoch": 29.08131868131868, "grad_norm": 1.0703125, "learning_rate": 8.542290829723996e-06, "loss": 0.1574, "step": 39696 }, { "epoch": 29.08205128205128, "grad_norm": 0.5859375, "learning_rate": 8.540472947480839e-06, "loss": 0.1111, "step": 39697 }, { "epoch": 29.082783882783882, "grad_norm": 1.0625, "learning_rate": 8.538655266279564e-06, "loss": 0.1285, "step": 39698 }, { "epoch": 29.083516483516483, "grad_norm": 0.921875, "learning_rate": 8.536837786128851e-06, "loss": 0.1001, "step": 39699 }, { "epoch": 29.084249084249084, "grad_norm": 0.83984375, "learning_rate": 8.535020507037432e-06, "loss": 0.1201, "step": 39700 }, { "epoch": 29.084981684981685, "grad_norm": 1.359375, "learning_rate": 8.533203429013997e-06, "loss": 0.1257, "step": 39701 }, { "epoch": 29.085714285714285, "grad_norm": 0.9453125, "learning_rate": 8.531386552067267e-06, "loss": 0.1364, "step": 39702 }, { "epoch": 29.086446886446886, "grad_norm": 1.1640625, "learning_rate": 8.529569876205955e-06, "loss": 0.1477, "step": 39703 }, { "epoch": 29.087179487179487, "grad_norm": 0.609375, "learning_rate": 8.527753401438732e-06, "loss": 0.1264, "step": 39704 }, { "epoch": 29.087912087912088, "grad_norm": 1.640625, "learning_rate": 8.525937127774341e-06, "loss": 0.1518, "step": 39705 }, { "epoch": 29.08864468864469, "grad_norm": 1.71875, "learning_rate": 8.524121055221438e-06, "loss": 0.1371, "step": 39706 }, { "epoch": 29.08937728937729, "grad_norm": 1.0390625, "learning_rate": 8.522305183788775e-06, "loss": 0.1198, "step": 39707 }, { "epoch": 29.09010989010989, "grad_norm": 1.171875, "learning_rate": 8.520489513485012e-06, "loss": 0.1349, "step": 39708 }, { "epoch": 29.09084249084249, "grad_norm": 0.92578125, "learning_rate": 8.51867404431888e-06, "loss": 0.0905, "step": 39709 }, { "epoch": 29.09157509157509, "grad_norm": 0.875, "learning_rate": 8.51685877629905e-06, "loss": 0.1145, "step": 39710 }, { "epoch": 29.092307692307692, "grad_norm": 1.1875, "learning_rate": 8.515043709434232e-06, "loss": 0.1127, "step": 39711 }, { "epoch": 29.093040293040293, "grad_norm": 0.9140625, "learning_rate": 8.513228843733124e-06, "loss": 0.1105, "step": 39712 }, { "epoch": 29.093772893772893, "grad_norm": 1.0, "learning_rate": 8.511414179204416e-06, "loss": 0.1193, "step": 39713 }, { "epoch": 29.094505494505494, "grad_norm": 1.0703125, "learning_rate": 8.50959971585682e-06, "loss": 0.1309, "step": 39714 }, { "epoch": 29.095238095238095, "grad_norm": 1.25, "learning_rate": 8.50778545369901e-06, "loss": 0.1269, "step": 39715 }, { "epoch": 29.095970695970696, "grad_norm": 0.7109375, "learning_rate": 8.505971392739691e-06, "loss": 0.093, "step": 39716 }, { "epoch": 29.096703296703296, "grad_norm": 0.984375, "learning_rate": 8.504157532987523e-06, "loss": 0.1454, "step": 39717 }, { "epoch": 29.097435897435897, "grad_norm": 0.99609375, "learning_rate": 8.502343874451252e-06, "loss": 0.0973, "step": 39718 }, { "epoch": 29.098168498168498, "grad_norm": 1.34375, "learning_rate": 8.500530417139525e-06, "loss": 0.1325, "step": 39719 }, { "epoch": 29.0989010989011, "grad_norm": 0.96875, "learning_rate": 8.498717161061055e-06, "loss": 0.1133, "step": 39720 }, { "epoch": 29.0996336996337, "grad_norm": 1.0234375, "learning_rate": 8.496904106224507e-06, "loss": 0.1095, "step": 39721 }, { "epoch": 29.1003663003663, "grad_norm": 0.9921875, "learning_rate": 8.495091252638581e-06, "loss": 0.1198, "step": 39722 }, { "epoch": 29.1010989010989, "grad_norm": 0.80078125, "learning_rate": 8.493278600311962e-06, "loss": 0.117, "step": 39723 }, { "epoch": 29.1018315018315, "grad_norm": 0.81640625, "learning_rate": 8.491466149253335e-06, "loss": 0.0995, "step": 39724 }, { "epoch": 29.102564102564102, "grad_norm": 0.8359375, "learning_rate": 8.489653899471396e-06, "loss": 0.1229, "step": 39725 }, { "epoch": 29.103296703296703, "grad_norm": 0.84765625, "learning_rate": 8.4878418509748e-06, "loss": 0.121, "step": 39726 }, { "epoch": 29.104029304029304, "grad_norm": 0.9140625, "learning_rate": 8.486030003772255e-06, "loss": 0.1488, "step": 39727 }, { "epoch": 29.104761904761904, "grad_norm": 1.0, "learning_rate": 8.484218357872423e-06, "loss": 0.1346, "step": 39728 }, { "epoch": 29.105494505494505, "grad_norm": 1.5078125, "learning_rate": 8.48240691328401e-06, "loss": 0.1108, "step": 39729 }, { "epoch": 29.106227106227106, "grad_norm": 0.92578125, "learning_rate": 8.48059567001567e-06, "loss": 0.1241, "step": 39730 }, { "epoch": 29.106959706959707, "grad_norm": 0.9296875, "learning_rate": 8.478784628076098e-06, "loss": 0.1392, "step": 39731 }, { "epoch": 29.107692307692307, "grad_norm": 0.734375, "learning_rate": 8.476973787473939e-06, "loss": 0.1178, "step": 39732 }, { "epoch": 29.108424908424908, "grad_norm": 1.4140625, "learning_rate": 8.475163148217923e-06, "loss": 0.1417, "step": 39733 }, { "epoch": 29.10915750915751, "grad_norm": 0.85546875, "learning_rate": 8.473352710316676e-06, "loss": 0.1097, "step": 39734 }, { "epoch": 29.10989010989011, "grad_norm": 0.72265625, "learning_rate": 8.471542473778914e-06, "loss": 0.1038, "step": 39735 }, { "epoch": 29.11062271062271, "grad_norm": 0.6796875, "learning_rate": 8.469732438613271e-06, "loss": 0.1205, "step": 39736 }, { "epoch": 29.11135531135531, "grad_norm": 1.0078125, "learning_rate": 8.467922604828436e-06, "loss": 0.1259, "step": 39737 }, { "epoch": 29.11208791208791, "grad_norm": 1.796875, "learning_rate": 8.466112972433086e-06, "loss": 0.1129, "step": 39738 }, { "epoch": 29.112820512820512, "grad_norm": 1.390625, "learning_rate": 8.464303541435887e-06, "loss": 0.1368, "step": 39739 }, { "epoch": 29.113553113553113, "grad_norm": 0.71875, "learning_rate": 8.46249431184552e-06, "loss": 0.1098, "step": 39740 }, { "epoch": 29.114285714285714, "grad_norm": 0.765625, "learning_rate": 8.460685283670635e-06, "loss": 0.1, "step": 39741 }, { "epoch": 29.115018315018315, "grad_norm": 1.265625, "learning_rate": 8.458876456919911e-06, "loss": 0.1281, "step": 39742 }, { "epoch": 29.115750915750915, "grad_norm": 1.21875, "learning_rate": 8.457067831601996e-06, "loss": 0.1126, "step": 39743 }, { "epoch": 29.116483516483516, "grad_norm": 0.95703125, "learning_rate": 8.455259407725589e-06, "loss": 0.1373, "step": 39744 }, { "epoch": 29.117216117216117, "grad_norm": 1.4296875, "learning_rate": 8.453451185299324e-06, "loss": 0.1604, "step": 39745 }, { "epoch": 29.117948717948718, "grad_norm": 1.0625, "learning_rate": 8.451643164331896e-06, "loss": 0.1092, "step": 39746 }, { "epoch": 29.11868131868132, "grad_norm": 0.66015625, "learning_rate": 8.449835344831928e-06, "loss": 0.0976, "step": 39747 }, { "epoch": 29.11941391941392, "grad_norm": 1.1640625, "learning_rate": 8.448027726808113e-06, "loss": 0.1301, "step": 39748 }, { "epoch": 29.12014652014652, "grad_norm": 1.09375, "learning_rate": 8.446220310269092e-06, "loss": 0.1135, "step": 39749 }, { "epoch": 29.12087912087912, "grad_norm": 1.046875, "learning_rate": 8.444413095223544e-06, "loss": 0.1418, "step": 39750 }, { "epoch": 29.12161172161172, "grad_norm": 1.34375, "learning_rate": 8.442606081680125e-06, "loss": 0.122, "step": 39751 }, { "epoch": 29.122344322344322, "grad_norm": 0.93359375, "learning_rate": 8.440799269647473e-06, "loss": 0.1293, "step": 39752 }, { "epoch": 29.123076923076923, "grad_norm": 0.93359375, "learning_rate": 8.438992659134266e-06, "loss": 0.1227, "step": 39753 }, { "epoch": 29.123809523809523, "grad_norm": 0.96484375, "learning_rate": 8.437186250149149e-06, "loss": 0.103, "step": 39754 }, { "epoch": 29.124542124542124, "grad_norm": 0.76953125, "learning_rate": 8.435380042700788e-06, "loss": 0.1252, "step": 39755 }, { "epoch": 29.125274725274725, "grad_norm": 1.5, "learning_rate": 8.433574036797824e-06, "loss": 0.1736, "step": 39756 }, { "epoch": 29.126007326007326, "grad_norm": 1.109375, "learning_rate": 8.431768232448931e-06, "loss": 0.1196, "step": 39757 }, { "epoch": 29.126739926739926, "grad_norm": 0.90625, "learning_rate": 8.429962629662717e-06, "loss": 0.1047, "step": 39758 }, { "epoch": 29.127472527472527, "grad_norm": 1.1796875, "learning_rate": 8.428157228447888e-06, "loss": 0.1367, "step": 39759 }, { "epoch": 29.128205128205128, "grad_norm": 0.78515625, "learning_rate": 8.426352028813062e-06, "loss": 0.113, "step": 39760 }, { "epoch": 29.12893772893773, "grad_norm": 0.81640625, "learning_rate": 8.4245470307669e-06, "loss": 0.1245, "step": 39761 }, { "epoch": 29.12967032967033, "grad_norm": 0.69140625, "learning_rate": 8.422742234318036e-06, "loss": 0.1035, "step": 39762 }, { "epoch": 29.13040293040293, "grad_norm": 0.9609375, "learning_rate": 8.420937639475129e-06, "loss": 0.1106, "step": 39763 }, { "epoch": 29.13113553113553, "grad_norm": 1.0078125, "learning_rate": 8.419133246246822e-06, "loss": 0.1389, "step": 39764 }, { "epoch": 29.13186813186813, "grad_norm": 1.0546875, "learning_rate": 8.417329054641757e-06, "loss": 0.1222, "step": 39765 }, { "epoch": 29.132600732600732, "grad_norm": 0.72265625, "learning_rate": 8.415525064668605e-06, "loss": 0.0944, "step": 39766 }, { "epoch": 29.133333333333333, "grad_norm": 1.109375, "learning_rate": 8.413721276335972e-06, "loss": 0.1037, "step": 39767 }, { "epoch": 29.134065934065934, "grad_norm": 0.8515625, "learning_rate": 8.411917689652523e-06, "loss": 0.1325, "step": 39768 }, { "epoch": 29.134798534798534, "grad_norm": 1.078125, "learning_rate": 8.410114304626876e-06, "loss": 0.1314, "step": 39769 }, { "epoch": 29.135531135531135, "grad_norm": 1.265625, "learning_rate": 8.40831112126771e-06, "loss": 0.1055, "step": 39770 }, { "epoch": 29.136263736263736, "grad_norm": 1.34375, "learning_rate": 8.40650813958363e-06, "loss": 0.1118, "step": 39771 }, { "epoch": 29.136996336996337, "grad_norm": 1.1171875, "learning_rate": 8.404705359583304e-06, "loss": 0.1237, "step": 39772 }, { "epoch": 29.137728937728937, "grad_norm": 1.0546875, "learning_rate": 8.402902781275334e-06, "loss": 0.1119, "step": 39773 }, { "epoch": 29.138461538461538, "grad_norm": 0.90625, "learning_rate": 8.40110040466838e-06, "loss": 0.1181, "step": 39774 }, { "epoch": 29.13919413919414, "grad_norm": 0.7578125, "learning_rate": 8.399298229771079e-06, "loss": 0.11, "step": 39775 }, { "epoch": 29.13992673992674, "grad_norm": 0.7421875, "learning_rate": 8.397496256592054e-06, "loss": 0.1104, "step": 39776 }, { "epoch": 29.14065934065934, "grad_norm": 0.5703125, "learning_rate": 8.395694485139962e-06, "loss": 0.1204, "step": 39777 }, { "epoch": 29.14139194139194, "grad_norm": 0.953125, "learning_rate": 8.393892915423408e-06, "loss": 0.1244, "step": 39778 }, { "epoch": 29.14212454212454, "grad_norm": 1.484375, "learning_rate": 8.392091547451035e-06, "loss": 0.1429, "step": 39779 }, { "epoch": 29.142857142857142, "grad_norm": 1.1953125, "learning_rate": 8.390290381231473e-06, "loss": 0.1164, "step": 39780 }, { "epoch": 29.143589743589743, "grad_norm": 1.0703125, "learning_rate": 8.388489416773366e-06, "loss": 0.1251, "step": 39781 }, { "epoch": 29.144322344322344, "grad_norm": 0.9921875, "learning_rate": 8.38668865408532e-06, "loss": 0.143, "step": 39782 }, { "epoch": 29.145054945054945, "grad_norm": 0.9609375, "learning_rate": 8.38488809317598e-06, "loss": 0.108, "step": 39783 }, { "epoch": 29.145787545787545, "grad_norm": 1.9609375, "learning_rate": 8.383087734053962e-06, "loss": 0.1232, "step": 39784 }, { "epoch": 29.146520146520146, "grad_norm": 0.87890625, "learning_rate": 8.38128757672789e-06, "loss": 0.1393, "step": 39785 }, { "epoch": 29.147252747252747, "grad_norm": 0.90234375, "learning_rate": 8.379487621206402e-06, "loss": 0.1136, "step": 39786 }, { "epoch": 29.147985347985347, "grad_norm": 1.28125, "learning_rate": 8.377687867498123e-06, "loss": 0.1283, "step": 39787 }, { "epoch": 29.148717948717948, "grad_norm": 0.99609375, "learning_rate": 8.37588831561166e-06, "loss": 0.1223, "step": 39788 }, { "epoch": 29.14945054945055, "grad_norm": 1.0, "learning_rate": 8.374088965555646e-06, "loss": 0.1327, "step": 39789 }, { "epoch": 29.15018315018315, "grad_norm": 1.0234375, "learning_rate": 8.3722898173387e-06, "loss": 0.1076, "step": 39790 }, { "epoch": 29.15091575091575, "grad_norm": 0.8515625, "learning_rate": 8.370490870969443e-06, "loss": 0.1218, "step": 39791 }, { "epoch": 29.15164835164835, "grad_norm": 1.109375, "learning_rate": 8.368692126456505e-06, "loss": 0.1236, "step": 39792 }, { "epoch": 29.152380952380952, "grad_norm": 0.76953125, "learning_rate": 8.36689358380848e-06, "loss": 0.1102, "step": 39793 }, { "epoch": 29.153113553113553, "grad_norm": 1.375, "learning_rate": 8.365095243034012e-06, "loss": 0.1269, "step": 39794 }, { "epoch": 29.153846153846153, "grad_norm": 1.1328125, "learning_rate": 8.363297104141688e-06, "loss": 0.1417, "step": 39795 }, { "epoch": 29.154578754578754, "grad_norm": 1.0, "learning_rate": 8.361499167140156e-06, "loss": 0.1163, "step": 39796 }, { "epoch": 29.155311355311355, "grad_norm": 1.2734375, "learning_rate": 8.359701432038003e-06, "loss": 0.1209, "step": 39797 }, { "epoch": 29.156043956043955, "grad_norm": 1.03125, "learning_rate": 8.357903898843867e-06, "loss": 0.1451, "step": 39798 }, { "epoch": 29.156776556776556, "grad_norm": 1.7265625, "learning_rate": 8.35610656756634e-06, "loss": 0.1092, "step": 39799 }, { "epoch": 29.157509157509157, "grad_norm": 0.74609375, "learning_rate": 8.354309438214037e-06, "loss": 0.128, "step": 39800 }, { "epoch": 29.158241758241758, "grad_norm": 1.3359375, "learning_rate": 8.352512510795578e-06, "loss": 0.1196, "step": 39801 }, { "epoch": 29.15897435897436, "grad_norm": 1.2578125, "learning_rate": 8.350715785319562e-06, "loss": 0.1031, "step": 39802 }, { "epoch": 29.15970695970696, "grad_norm": 0.765625, "learning_rate": 8.348919261794615e-06, "loss": 0.1287, "step": 39803 }, { "epoch": 29.16043956043956, "grad_norm": 1.1640625, "learning_rate": 8.347122940229323e-06, "loss": 0.1418, "step": 39804 }, { "epoch": 29.16117216117216, "grad_norm": 1.40625, "learning_rate": 8.345326820632308e-06, "loss": 0.1027, "step": 39805 }, { "epoch": 29.16190476190476, "grad_norm": 1.578125, "learning_rate": 8.343530903012167e-06, "loss": 0.1472, "step": 39806 }, { "epoch": 29.162637362637362, "grad_norm": 0.9765625, "learning_rate": 8.341735187377516e-06, "loss": 0.1442, "step": 39807 }, { "epoch": 29.163369963369963, "grad_norm": 0.7734375, "learning_rate": 8.339939673736945e-06, "loss": 0.1152, "step": 39808 }, { "epoch": 29.164102564102564, "grad_norm": 0.86328125, "learning_rate": 8.33814436209907e-06, "loss": 0.1127, "step": 39809 }, { "epoch": 29.164835164835164, "grad_norm": 0.8671875, "learning_rate": 8.336349252472479e-06, "loss": 0.1163, "step": 39810 }, { "epoch": 29.165567765567765, "grad_norm": 1.0625, "learning_rate": 8.334554344865782e-06, "loss": 0.1372, "step": 39811 }, { "epoch": 29.166300366300366, "grad_norm": 1.109375, "learning_rate": 8.332759639287574e-06, "loss": 0.1054, "step": 39812 }, { "epoch": 29.167032967032966, "grad_norm": 0.90625, "learning_rate": 8.330965135746465e-06, "loss": 0.1011, "step": 39813 }, { "epoch": 29.167765567765567, "grad_norm": 0.84375, "learning_rate": 8.32917083425104e-06, "loss": 0.1383, "step": 39814 }, { "epoch": 29.168498168498168, "grad_norm": 1.2421875, "learning_rate": 8.327376734809896e-06, "loss": 0.1293, "step": 39815 }, { "epoch": 29.16923076923077, "grad_norm": 0.8671875, "learning_rate": 8.32558283743164e-06, "loss": 0.1494, "step": 39816 }, { "epoch": 29.16996336996337, "grad_norm": 0.8984375, "learning_rate": 8.323789142124859e-06, "loss": 0.1049, "step": 39817 }, { "epoch": 29.17069597069597, "grad_norm": 0.6171875, "learning_rate": 8.32199564889816e-06, "loss": 0.1079, "step": 39818 }, { "epoch": 29.17142857142857, "grad_norm": 1.3359375, "learning_rate": 8.320202357760116e-06, "loss": 0.118, "step": 39819 }, { "epoch": 29.17216117216117, "grad_norm": 0.98828125, "learning_rate": 8.31840926871934e-06, "loss": 0.1143, "step": 39820 }, { "epoch": 29.172893772893772, "grad_norm": 1.046875, "learning_rate": 8.316616381784387e-06, "loss": 0.0911, "step": 39821 }, { "epoch": 29.173626373626373, "grad_norm": 1.3828125, "learning_rate": 8.314823696963903e-06, "loss": 0.1247, "step": 39822 }, { "epoch": 29.174358974358974, "grad_norm": 1.03125, "learning_rate": 8.313031214266428e-06, "loss": 0.1304, "step": 39823 }, { "epoch": 29.175091575091574, "grad_norm": 1.140625, "learning_rate": 8.311238933700584e-06, "loss": 0.1151, "step": 39824 }, { "epoch": 29.175824175824175, "grad_norm": 1.015625, "learning_rate": 8.309446855274935e-06, "loss": 0.0845, "step": 39825 }, { "epoch": 29.176556776556776, "grad_norm": 0.90625, "learning_rate": 8.307654978998069e-06, "loss": 0.1111, "step": 39826 }, { "epoch": 29.177289377289377, "grad_norm": 0.95703125, "learning_rate": 8.305863304878584e-06, "loss": 0.1299, "step": 39827 }, { "epoch": 29.178021978021977, "grad_norm": 0.94921875, "learning_rate": 8.30407183292506e-06, "loss": 0.1205, "step": 39828 }, { "epoch": 29.178754578754578, "grad_norm": 1.25, "learning_rate": 8.302280563146093e-06, "loss": 0.125, "step": 39829 }, { "epoch": 29.17948717948718, "grad_norm": 1.1484375, "learning_rate": 8.300489495550239e-06, "loss": 0.1285, "step": 39830 }, { "epoch": 29.18021978021978, "grad_norm": 1.640625, "learning_rate": 8.29869863014609e-06, "loss": 0.1417, "step": 39831 }, { "epoch": 29.18095238095238, "grad_norm": 0.73046875, "learning_rate": 8.29690796694223e-06, "loss": 0.1221, "step": 39832 }, { "epoch": 29.18168498168498, "grad_norm": 0.80078125, "learning_rate": 8.295117505947258e-06, "loss": 0.1315, "step": 39833 }, { "epoch": 29.18241758241758, "grad_norm": 1.3828125, "learning_rate": 8.293327247169712e-06, "loss": 0.1249, "step": 39834 }, { "epoch": 29.183150183150182, "grad_norm": 0.9921875, "learning_rate": 8.291537190618207e-06, "loss": 0.139, "step": 39835 }, { "epoch": 29.183882783882783, "grad_norm": 1.109375, "learning_rate": 8.289747336301289e-06, "loss": 0.1378, "step": 39836 }, { "epoch": 29.184615384615384, "grad_norm": 1.09375, "learning_rate": 8.287957684227548e-06, "loss": 0.1259, "step": 39837 }, { "epoch": 29.185347985347985, "grad_norm": 1.3359375, "learning_rate": 8.286168234405565e-06, "loss": 0.1463, "step": 39838 }, { "epoch": 29.186080586080585, "grad_norm": 1.1015625, "learning_rate": 8.284378986843899e-06, "loss": 0.1148, "step": 39839 }, { "epoch": 29.186813186813186, "grad_norm": 1.03125, "learning_rate": 8.282589941551152e-06, "loss": 0.1253, "step": 39840 }, { "epoch": 29.187545787545787, "grad_norm": 1.0546875, "learning_rate": 8.28080109853586e-06, "loss": 0.1137, "step": 39841 }, { "epoch": 29.188278388278388, "grad_norm": 1.4375, "learning_rate": 8.279012457806605e-06, "loss": 0.1343, "step": 39842 }, { "epoch": 29.18901098901099, "grad_norm": 0.87109375, "learning_rate": 8.277224019371967e-06, "loss": 0.1233, "step": 39843 }, { "epoch": 29.18974358974359, "grad_norm": 0.65234375, "learning_rate": 8.275435783240524e-06, "loss": 0.1158, "step": 39844 }, { "epoch": 29.19047619047619, "grad_norm": 1.21875, "learning_rate": 8.273647749420811e-06, "loss": 0.1312, "step": 39845 }, { "epoch": 29.19120879120879, "grad_norm": 1.2109375, "learning_rate": 8.271859917921428e-06, "loss": 0.1265, "step": 39846 }, { "epoch": 29.19194139194139, "grad_norm": 0.96875, "learning_rate": 8.270072288750905e-06, "loss": 0.1186, "step": 39847 }, { "epoch": 29.192673992673992, "grad_norm": 0.9609375, "learning_rate": 8.268284861917854e-06, "loss": 0.1459, "step": 39848 }, { "epoch": 29.193406593406593, "grad_norm": 1.390625, "learning_rate": 8.266497637430797e-06, "loss": 0.156, "step": 39849 }, { "epoch": 29.194139194139193, "grad_norm": 0.97265625, "learning_rate": 8.264710615298329e-06, "loss": 0.128, "step": 39850 }, { "epoch": 29.194871794871794, "grad_norm": 1.171875, "learning_rate": 8.262923795528987e-06, "loss": 0.1227, "step": 39851 }, { "epoch": 29.195604395604395, "grad_norm": 0.89453125, "learning_rate": 8.261137178131338e-06, "loss": 0.1201, "step": 39852 }, { "epoch": 29.196336996336996, "grad_norm": 1.0078125, "learning_rate": 8.259350763113948e-06, "loss": 0.0934, "step": 39853 }, { "epoch": 29.197069597069596, "grad_norm": 0.92578125, "learning_rate": 8.257564550485381e-06, "loss": 0.1032, "step": 39854 }, { "epoch": 29.197802197802197, "grad_norm": 1.1015625, "learning_rate": 8.255778540254193e-06, "loss": 0.1282, "step": 39855 }, { "epoch": 29.198534798534798, "grad_norm": 1.046875, "learning_rate": 8.253992732428934e-06, "loss": 0.1186, "step": 39856 }, { "epoch": 29.1992673992674, "grad_norm": 1.234375, "learning_rate": 8.252207127018158e-06, "loss": 0.1372, "step": 39857 }, { "epoch": 29.2, "grad_norm": 1.5859375, "learning_rate": 8.25042172403043e-06, "loss": 0.1567, "step": 39858 }, { "epoch": 29.2007326007326, "grad_norm": 0.96484375, "learning_rate": 8.248636523474313e-06, "loss": 0.1202, "step": 39859 }, { "epoch": 29.2014652014652, "grad_norm": 1.640625, "learning_rate": 8.246851525358335e-06, "loss": 0.1415, "step": 39860 }, { "epoch": 29.2021978021978, "grad_norm": 1.046875, "learning_rate": 8.245066729691076e-06, "loss": 0.082, "step": 39861 }, { "epoch": 29.202930402930402, "grad_norm": 0.8203125, "learning_rate": 8.24328213648106e-06, "loss": 0.1409, "step": 39862 }, { "epoch": 29.203663003663003, "grad_norm": 0.63671875, "learning_rate": 8.241497745736849e-06, "loss": 0.1198, "step": 39863 }, { "epoch": 29.204395604395604, "grad_norm": 0.828125, "learning_rate": 8.239713557467e-06, "loss": 0.1203, "step": 39864 }, { "epoch": 29.205128205128204, "grad_norm": 0.89453125, "learning_rate": 8.237929571680057e-06, "loss": 0.1214, "step": 39865 }, { "epoch": 29.205860805860805, "grad_norm": 1.3671875, "learning_rate": 8.236145788384576e-06, "loss": 0.1157, "step": 39866 }, { "epoch": 29.206593406593406, "grad_norm": 1.1796875, "learning_rate": 8.234362207589083e-06, "loss": 0.1027, "step": 39867 }, { "epoch": 29.207326007326007, "grad_norm": 1.046875, "learning_rate": 8.232578829302137e-06, "loss": 0.1074, "step": 39868 }, { "epoch": 29.208058608058607, "grad_norm": 0.83203125, "learning_rate": 8.230795653532276e-06, "loss": 0.1478, "step": 39869 }, { "epoch": 29.208791208791208, "grad_norm": 1.5546875, "learning_rate": 8.229012680288064e-06, "loss": 0.1439, "step": 39870 }, { "epoch": 29.20952380952381, "grad_norm": 0.8203125, "learning_rate": 8.227229909578022e-06, "loss": 0.107, "step": 39871 }, { "epoch": 29.21025641025641, "grad_norm": 1.4765625, "learning_rate": 8.225447341410705e-06, "loss": 0.1507, "step": 39872 }, { "epoch": 29.21098901098901, "grad_norm": 0.6796875, "learning_rate": 8.223664975794629e-06, "loss": 0.113, "step": 39873 }, { "epoch": 29.21172161172161, "grad_norm": 1.109375, "learning_rate": 8.22188281273837e-06, "loss": 0.1435, "step": 39874 }, { "epoch": 29.21245421245421, "grad_norm": 1.6796875, "learning_rate": 8.220100852250441e-06, "loss": 0.1423, "step": 39875 }, { "epoch": 29.213186813186812, "grad_norm": 0.8046875, "learning_rate": 8.218319094339402e-06, "loss": 0.1231, "step": 39876 }, { "epoch": 29.213919413919413, "grad_norm": 0.703125, "learning_rate": 8.21653753901376e-06, "loss": 0.0962, "step": 39877 }, { "epoch": 29.214652014652014, "grad_norm": 0.609375, "learning_rate": 8.21475618628207e-06, "loss": 0.1345, "step": 39878 }, { "epoch": 29.215384615384615, "grad_norm": 1.296875, "learning_rate": 8.212975036152866e-06, "loss": 0.1306, "step": 39879 }, { "epoch": 29.216117216117215, "grad_norm": 1.203125, "learning_rate": 8.211194088634677e-06, "loss": 0.1219, "step": 39880 }, { "epoch": 29.216849816849816, "grad_norm": 1.046875, "learning_rate": 8.209413343736053e-06, "loss": 0.1155, "step": 39881 }, { "epoch": 29.217582417582417, "grad_norm": 0.953125, "learning_rate": 8.207632801465499e-06, "loss": 0.1208, "step": 39882 }, { "epoch": 29.218315018315018, "grad_norm": 0.77734375, "learning_rate": 8.205852461831563e-06, "loss": 0.1155, "step": 39883 }, { "epoch": 29.21904761904762, "grad_norm": 1.21875, "learning_rate": 8.20407232484277e-06, "loss": 0.1442, "step": 39884 }, { "epoch": 29.21978021978022, "grad_norm": 0.56640625, "learning_rate": 8.202292390507663e-06, "loss": 0.1072, "step": 39885 }, { "epoch": 29.22051282051282, "grad_norm": 0.83203125, "learning_rate": 8.200512658834749e-06, "loss": 0.1409, "step": 39886 }, { "epoch": 29.22124542124542, "grad_norm": 1.25, "learning_rate": 8.198733129832575e-06, "loss": 0.1201, "step": 39887 }, { "epoch": 29.22197802197802, "grad_norm": 0.84375, "learning_rate": 8.19695380350965e-06, "loss": 0.1011, "step": 39888 }, { "epoch": 29.222710622710622, "grad_norm": 1.015625, "learning_rate": 8.195174679874501e-06, "loss": 0.1014, "step": 39889 }, { "epoch": 29.223443223443223, "grad_norm": 0.7265625, "learning_rate": 8.193395758935659e-06, "loss": 0.1036, "step": 39890 }, { "epoch": 29.224175824175823, "grad_norm": 0.9921875, "learning_rate": 8.191617040701649e-06, "loss": 0.1325, "step": 39891 }, { "epoch": 29.224908424908424, "grad_norm": 1.3828125, "learning_rate": 8.189838525180998e-06, "loss": 0.1045, "step": 39892 }, { "epoch": 29.225641025641025, "grad_norm": 1.0, "learning_rate": 8.188060212382208e-06, "loss": 0.1285, "step": 39893 }, { "epoch": 29.226373626373626, "grad_norm": 1.0, "learning_rate": 8.186282102313818e-06, "loss": 0.1191, "step": 39894 }, { "epoch": 29.227106227106226, "grad_norm": 1.3359375, "learning_rate": 8.184504194984334e-06, "loss": 0.135, "step": 39895 }, { "epoch": 29.227838827838827, "grad_norm": 1.6796875, "learning_rate": 8.182726490402298e-06, "loss": 0.1454, "step": 39896 }, { "epoch": 29.228571428571428, "grad_norm": 2.046875, "learning_rate": 8.180948988576201e-06, "loss": 0.1794, "step": 39897 }, { "epoch": 29.22930402930403, "grad_norm": 0.96484375, "learning_rate": 8.179171689514579e-06, "loss": 0.1246, "step": 39898 }, { "epoch": 29.23003663003663, "grad_norm": 0.78515625, "learning_rate": 8.177394593225916e-06, "loss": 0.1076, "step": 39899 }, { "epoch": 29.23076923076923, "grad_norm": 1.3984375, "learning_rate": 8.175617699718768e-06, "loss": 0.1136, "step": 39900 }, { "epoch": 29.23150183150183, "grad_norm": 0.96484375, "learning_rate": 8.173841009001622e-06, "loss": 0.1074, "step": 39901 }, { "epoch": 29.23223443223443, "grad_norm": 0.80859375, "learning_rate": 8.172064521083006e-06, "loss": 0.1332, "step": 39902 }, { "epoch": 29.232967032967032, "grad_norm": 1.0546875, "learning_rate": 8.170288235971418e-06, "loss": 0.1468, "step": 39903 }, { "epoch": 29.233699633699633, "grad_norm": 0.9453125, "learning_rate": 8.168512153675374e-06, "loss": 0.1223, "step": 39904 }, { "epoch": 29.234432234432234, "grad_norm": 1.5078125, "learning_rate": 8.166736274203382e-06, "loss": 0.1477, "step": 39905 }, { "epoch": 29.235164835164834, "grad_norm": 1.8671875, "learning_rate": 8.16496059756396e-06, "loss": 0.1625, "step": 39906 }, { "epoch": 29.235897435897435, "grad_norm": 0.91015625, "learning_rate": 8.163185123765617e-06, "loss": 0.0941, "step": 39907 }, { "epoch": 29.236630036630036, "grad_norm": 0.83203125, "learning_rate": 8.16140985281684e-06, "loss": 0.1292, "step": 39908 }, { "epoch": 29.237362637362637, "grad_norm": 1.328125, "learning_rate": 8.159634784726144e-06, "loss": 0.1304, "step": 39909 }, { "epoch": 29.238095238095237, "grad_norm": 1.0234375, "learning_rate": 8.157859919502043e-06, "loss": 0.1232, "step": 39910 }, { "epoch": 29.238827838827838, "grad_norm": 1.0, "learning_rate": 8.156085257153044e-06, "loss": 0.1112, "step": 39911 }, { "epoch": 29.23956043956044, "grad_norm": 0.9609375, "learning_rate": 8.154310797687632e-06, "loss": 0.115, "step": 39912 }, { "epoch": 29.24029304029304, "grad_norm": 1.390625, "learning_rate": 8.152536541114327e-06, "loss": 0.1473, "step": 39913 }, { "epoch": 29.24102564102564, "grad_norm": 1.0234375, "learning_rate": 8.15076248744161e-06, "loss": 0.1339, "step": 39914 }, { "epoch": 29.24175824175824, "grad_norm": 0.921875, "learning_rate": 8.148988636677992e-06, "loss": 0.1234, "step": 39915 }, { "epoch": 29.24249084249084, "grad_norm": 1.28125, "learning_rate": 8.147214988831976e-06, "loss": 0.1789, "step": 39916 }, { "epoch": 29.243223443223442, "grad_norm": 1.109375, "learning_rate": 8.145441543912048e-06, "loss": 0.1022, "step": 39917 }, { "epoch": 29.243956043956043, "grad_norm": 1.0546875, "learning_rate": 8.143668301926734e-06, "loss": 0.1055, "step": 39918 }, { "epoch": 29.244688644688644, "grad_norm": 0.90625, "learning_rate": 8.141895262884495e-06, "loss": 0.1147, "step": 39919 }, { "epoch": 29.245421245421245, "grad_norm": 0.98046875, "learning_rate": 8.140122426793841e-06, "loss": 0.1356, "step": 39920 }, { "epoch": 29.246153846153845, "grad_norm": 1.03125, "learning_rate": 8.138349793663265e-06, "loss": 0.1019, "step": 39921 }, { "epoch": 29.246886446886446, "grad_norm": 0.96875, "learning_rate": 8.136577363501274e-06, "loss": 0.1338, "step": 39922 }, { "epoch": 29.247619047619047, "grad_norm": 1.0546875, "learning_rate": 8.13480513631634e-06, "loss": 0.1101, "step": 39923 }, { "epoch": 29.248351648351647, "grad_norm": 0.9453125, "learning_rate": 8.133033112116967e-06, "loss": 0.1429, "step": 39924 }, { "epoch": 29.249084249084248, "grad_norm": 0.64453125, "learning_rate": 8.13126129091162e-06, "loss": 0.1306, "step": 39925 }, { "epoch": 29.24981684981685, "grad_norm": 0.8671875, "learning_rate": 8.129489672708834e-06, "loss": 0.1299, "step": 39926 }, { "epoch": 29.25054945054945, "grad_norm": 1.109375, "learning_rate": 8.127718257517062e-06, "loss": 0.1373, "step": 39927 }, { "epoch": 29.25128205128205, "grad_norm": 0.94921875, "learning_rate": 8.125947045344811e-06, "loss": 0.127, "step": 39928 }, { "epoch": 29.25201465201465, "grad_norm": 0.859375, "learning_rate": 8.124176036200547e-06, "loss": 0.115, "step": 39929 }, { "epoch": 29.252747252747252, "grad_norm": 1.171875, "learning_rate": 8.122405230092765e-06, "loss": 0.127, "step": 39930 }, { "epoch": 29.253479853479853, "grad_norm": 0.71484375, "learning_rate": 8.120634627029954e-06, "loss": 0.0985, "step": 39931 }, { "epoch": 29.254212454212453, "grad_norm": 0.796875, "learning_rate": 8.118864227020592e-06, "loss": 0.1396, "step": 39932 }, { "epoch": 29.254945054945054, "grad_norm": 1.3671875, "learning_rate": 8.117094030073176e-06, "loss": 0.1053, "step": 39933 }, { "epoch": 29.255677655677655, "grad_norm": 1.171875, "learning_rate": 8.115324036196168e-06, "loss": 0.1144, "step": 39934 }, { "epoch": 29.256410256410255, "grad_norm": 1.21875, "learning_rate": 8.113554245398051e-06, "loss": 0.1268, "step": 39935 }, { "epoch": 29.257142857142856, "grad_norm": 1.28125, "learning_rate": 8.111784657687318e-06, "loss": 0.0803, "step": 39936 }, { "epoch": 29.257875457875457, "grad_norm": 0.7265625, "learning_rate": 8.110015273072444e-06, "loss": 0.0931, "step": 39937 }, { "epoch": 29.258608058608058, "grad_norm": 0.765625, "learning_rate": 8.10824609156189e-06, "loss": 0.0818, "step": 39938 }, { "epoch": 29.25934065934066, "grad_norm": 0.81640625, "learning_rate": 8.106477113164164e-06, "loss": 0.1282, "step": 39939 }, { "epoch": 29.26007326007326, "grad_norm": 1.171875, "learning_rate": 8.104708337887706e-06, "loss": 0.1068, "step": 39940 }, { "epoch": 29.26080586080586, "grad_norm": 0.94921875, "learning_rate": 8.10293976574101e-06, "loss": 0.1196, "step": 39941 }, { "epoch": 29.26153846153846, "grad_norm": 1.4921875, "learning_rate": 8.101171396732545e-06, "loss": 0.1397, "step": 39942 }, { "epoch": 29.26227106227106, "grad_norm": 1.078125, "learning_rate": 8.09940323087079e-06, "loss": 0.1237, "step": 39943 }, { "epoch": 29.263003663003662, "grad_norm": 1.484375, "learning_rate": 8.097635268164225e-06, "loss": 0.1174, "step": 39944 }, { "epoch": 29.263736263736263, "grad_norm": 1.0234375, "learning_rate": 8.095867508621295e-06, "loss": 0.1336, "step": 39945 }, { "epoch": 29.264468864468864, "grad_norm": 0.97265625, "learning_rate": 8.094099952250487e-06, "loss": 0.1313, "step": 39946 }, { "epoch": 29.265201465201464, "grad_norm": 1.171875, "learning_rate": 8.092332599060265e-06, "loss": 0.1244, "step": 39947 }, { "epoch": 29.265934065934065, "grad_norm": 1.34375, "learning_rate": 8.090565449059112e-06, "loss": 0.1381, "step": 39948 }, { "epoch": 29.266666666666666, "grad_norm": 1.1796875, "learning_rate": 8.088798502255471e-06, "loss": 0.1079, "step": 39949 }, { "epoch": 29.267399267399266, "grad_norm": 0.7109375, "learning_rate": 8.087031758657826e-06, "loss": 0.1142, "step": 39950 }, { "epoch": 29.268131868131867, "grad_norm": 1.1953125, "learning_rate": 8.08526521827462e-06, "loss": 0.0965, "step": 39951 }, { "epoch": 29.268864468864468, "grad_norm": 1.0625, "learning_rate": 8.083498881114345e-06, "loss": 0.1177, "step": 39952 }, { "epoch": 29.26959706959707, "grad_norm": 1.28125, "learning_rate": 8.081732747185449e-06, "loss": 0.1416, "step": 39953 }, { "epoch": 29.27032967032967, "grad_norm": 1.375, "learning_rate": 8.079966816496394e-06, "loss": 0.1362, "step": 39954 }, { "epoch": 29.27106227106227, "grad_norm": 0.7890625, "learning_rate": 8.078201089055654e-06, "loss": 0.1253, "step": 39955 }, { "epoch": 29.27179487179487, "grad_norm": 1.0625, "learning_rate": 8.076435564871666e-06, "loss": 0.1147, "step": 39956 }, { "epoch": 29.27252747252747, "grad_norm": 1.0859375, "learning_rate": 8.074670243952902e-06, "loss": 0.1134, "step": 39957 }, { "epoch": 29.273260073260072, "grad_norm": 1.2421875, "learning_rate": 8.072905126307822e-06, "loss": 0.1517, "step": 39958 }, { "epoch": 29.273992673992673, "grad_norm": 1.1953125, "learning_rate": 8.071140211944893e-06, "loss": 0.1287, "step": 39959 }, { "epoch": 29.274725274725274, "grad_norm": 1.0390625, "learning_rate": 8.069375500872543e-06, "loss": 0.1449, "step": 39960 }, { "epoch": 29.275457875457874, "grad_norm": 1.3515625, "learning_rate": 8.067610993099252e-06, "loss": 0.1137, "step": 39961 }, { "epoch": 29.276190476190475, "grad_norm": 0.87890625, "learning_rate": 8.06584668863346e-06, "loss": 0.1105, "step": 39962 }, { "epoch": 29.276923076923076, "grad_norm": 0.875, "learning_rate": 8.064082587483645e-06, "loss": 0.1081, "step": 39963 }, { "epoch": 29.277655677655677, "grad_norm": 0.9609375, "learning_rate": 8.062318689658222e-06, "loss": 0.1347, "step": 39964 }, { "epoch": 29.278388278388277, "grad_norm": 0.62109375, "learning_rate": 8.060554995165679e-06, "loss": 0.0882, "step": 39965 }, { "epoch": 29.279120879120878, "grad_norm": 1.078125, "learning_rate": 8.058791504014433e-06, "loss": 0.1217, "step": 39966 }, { "epoch": 29.27985347985348, "grad_norm": 0.875, "learning_rate": 8.05702821621295e-06, "loss": 0.1155, "step": 39967 }, { "epoch": 29.28058608058608, "grad_norm": 1.4921875, "learning_rate": 8.05526513176968e-06, "loss": 0.1315, "step": 39968 }, { "epoch": 29.28131868131868, "grad_norm": 0.87890625, "learning_rate": 8.05350225069307e-06, "loss": 0.1175, "step": 39969 }, { "epoch": 29.28205128205128, "grad_norm": 1.1796875, "learning_rate": 8.051739572991573e-06, "loss": 0.0805, "step": 39970 }, { "epoch": 29.282783882783882, "grad_norm": 1.375, "learning_rate": 8.049977098673615e-06, "loss": 0.1381, "step": 39971 }, { "epoch": 29.283516483516483, "grad_norm": 0.84375, "learning_rate": 8.048214827747656e-06, "loss": 0.0921, "step": 39972 }, { "epoch": 29.284249084249083, "grad_norm": 1.1484375, "learning_rate": 8.046452760222135e-06, "loss": 0.152, "step": 39973 }, { "epoch": 29.284981684981684, "grad_norm": 0.64453125, "learning_rate": 8.044690896105502e-06, "loss": 0.1241, "step": 39974 }, { "epoch": 29.285714285714285, "grad_norm": 0.77734375, "learning_rate": 8.042929235406189e-06, "loss": 0.1034, "step": 39975 }, { "epoch": 29.286446886446885, "grad_norm": 0.87109375, "learning_rate": 8.041167778132644e-06, "loss": 0.1302, "step": 39976 }, { "epoch": 29.287179487179486, "grad_norm": 1.3203125, "learning_rate": 8.039406524293282e-06, "loss": 0.1203, "step": 39977 }, { "epoch": 29.287912087912087, "grad_norm": 0.93359375, "learning_rate": 8.037645473896583e-06, "loss": 0.1327, "step": 39978 }, { "epoch": 29.288644688644688, "grad_norm": 0.99609375, "learning_rate": 8.035884626950957e-06, "loss": 0.1195, "step": 39979 }, { "epoch": 29.28937728937729, "grad_norm": 0.8125, "learning_rate": 8.034123983464847e-06, "loss": 0.0991, "step": 39980 }, { "epoch": 29.29010989010989, "grad_norm": 0.9921875, "learning_rate": 8.032363543446696e-06, "loss": 0.1105, "step": 39981 }, { "epoch": 29.29084249084249, "grad_norm": 1.2734375, "learning_rate": 8.030603306904925e-06, "loss": 0.1424, "step": 39982 }, { "epoch": 29.29157509157509, "grad_norm": 1.328125, "learning_rate": 8.02884327384797e-06, "loss": 0.1065, "step": 39983 }, { "epoch": 29.29230769230769, "grad_norm": 0.76953125, "learning_rate": 8.02708344428428e-06, "loss": 0.1194, "step": 39984 }, { "epoch": 29.293040293040292, "grad_norm": 1.1328125, "learning_rate": 8.02532381822228e-06, "loss": 0.1544, "step": 39985 }, { "epoch": 29.293772893772893, "grad_norm": 1.234375, "learning_rate": 8.023564395670384e-06, "loss": 0.1684, "step": 39986 }, { "epoch": 29.294505494505493, "grad_norm": 0.984375, "learning_rate": 8.02180517663704e-06, "loss": 0.1195, "step": 39987 }, { "epoch": 29.295238095238094, "grad_norm": 1.0546875, "learning_rate": 8.02004616113067e-06, "loss": 0.1273, "step": 39988 }, { "epoch": 29.295970695970695, "grad_norm": 0.73046875, "learning_rate": 8.018287349159717e-06, "loss": 0.1047, "step": 39989 }, { "epoch": 29.296703296703296, "grad_norm": 0.9140625, "learning_rate": 8.016528740732583e-06, "loss": 0.1163, "step": 39990 }, { "epoch": 29.297435897435896, "grad_norm": 1.359375, "learning_rate": 8.014770335857718e-06, "loss": 0.1113, "step": 39991 }, { "epoch": 29.298168498168497, "grad_norm": 1.2734375, "learning_rate": 8.013012134543524e-06, "loss": 0.1085, "step": 39992 }, { "epoch": 29.298901098901098, "grad_norm": 1.4375, "learning_rate": 8.011254136798433e-06, "loss": 0.1229, "step": 39993 }, { "epoch": 29.2996336996337, "grad_norm": 1.046875, "learning_rate": 8.009496342630874e-06, "loss": 0.1217, "step": 39994 }, { "epoch": 29.3003663003663, "grad_norm": 1.09375, "learning_rate": 8.007738752049265e-06, "loss": 0.1189, "step": 39995 }, { "epoch": 29.3010989010989, "grad_norm": 1.171875, "learning_rate": 8.005981365062044e-06, "loss": 0.1345, "step": 39996 }, { "epoch": 29.3018315018315, "grad_norm": 0.74609375, "learning_rate": 8.004224181677604e-06, "loss": 0.1073, "step": 39997 }, { "epoch": 29.3025641025641, "grad_norm": 0.96484375, "learning_rate": 8.002467201904377e-06, "loss": 0.136, "step": 39998 }, { "epoch": 29.303296703296702, "grad_norm": 1.4921875, "learning_rate": 8.000710425750779e-06, "loss": 0.1031, "step": 39999 }, { "epoch": 29.304029304029303, "grad_norm": 1.03125, "learning_rate": 7.99895385322524e-06, "loss": 0.108, "step": 40000 }, { "epoch": 29.304761904761904, "grad_norm": 1.0546875, "learning_rate": 7.997197484336152e-06, "loss": 0.1201, "step": 40001 }, { "epoch": 29.305494505494504, "grad_norm": 0.90234375, "learning_rate": 7.995441319091957e-06, "loss": 0.1336, "step": 40002 }, { "epoch": 29.306227106227105, "grad_norm": 1.3515625, "learning_rate": 7.993685357501033e-06, "loss": 0.1055, "step": 40003 }, { "epoch": 29.306959706959706, "grad_norm": 1.546875, "learning_rate": 7.99192959957184e-06, "loss": 0.0946, "step": 40004 }, { "epoch": 29.307692307692307, "grad_norm": 0.98046875, "learning_rate": 7.990174045312758e-06, "loss": 0.1187, "step": 40005 }, { "epoch": 29.308424908424907, "grad_norm": 0.87890625, "learning_rate": 7.9884186947322e-06, "loss": 0.1218, "step": 40006 }, { "epoch": 29.309157509157508, "grad_norm": 0.7578125, "learning_rate": 7.986663547838601e-06, "loss": 0.1075, "step": 40007 }, { "epoch": 29.30989010989011, "grad_norm": 0.92578125, "learning_rate": 7.984908604640342e-06, "loss": 0.135, "step": 40008 }, { "epoch": 29.31062271062271, "grad_norm": 1.1875, "learning_rate": 7.983153865145835e-06, "loss": 0.1043, "step": 40009 }, { "epoch": 29.31135531135531, "grad_norm": 0.6953125, "learning_rate": 7.981399329363507e-06, "loss": 0.1107, "step": 40010 }, { "epoch": 29.31208791208791, "grad_norm": 1.1484375, "learning_rate": 7.979644997301753e-06, "loss": 0.1442, "step": 40011 }, { "epoch": 29.31282051282051, "grad_norm": 0.96875, "learning_rate": 7.977890868968976e-06, "loss": 0.131, "step": 40012 }, { "epoch": 29.313553113553112, "grad_norm": 0.99609375, "learning_rate": 7.976136944373587e-06, "loss": 0.1273, "step": 40013 }, { "epoch": 29.314285714285713, "grad_norm": 1.203125, "learning_rate": 7.974383223523968e-06, "loss": 0.149, "step": 40014 }, { "epoch": 29.315018315018314, "grad_norm": 1.609375, "learning_rate": 7.972629706428558e-06, "loss": 0.106, "step": 40015 }, { "epoch": 29.315750915750915, "grad_norm": 0.66796875, "learning_rate": 7.970876393095734e-06, "loss": 0.107, "step": 40016 }, { "epoch": 29.316483516483515, "grad_norm": 1.4453125, "learning_rate": 7.96912328353391e-06, "loss": 0.1193, "step": 40017 }, { "epoch": 29.317216117216116, "grad_norm": 0.7578125, "learning_rate": 7.967370377751467e-06, "loss": 0.1179, "step": 40018 }, { "epoch": 29.317948717948717, "grad_norm": 0.90234375, "learning_rate": 7.965617675756817e-06, "loss": 0.1042, "step": 40019 }, { "epoch": 29.318681318681318, "grad_norm": 0.69140625, "learning_rate": 7.963865177558354e-06, "loss": 0.0931, "step": 40020 }, { "epoch": 29.31941391941392, "grad_norm": 0.90234375, "learning_rate": 7.962112883164475e-06, "loss": 0.0964, "step": 40021 }, { "epoch": 29.32014652014652, "grad_norm": 1.453125, "learning_rate": 7.96036079258359e-06, "loss": 0.1511, "step": 40022 }, { "epoch": 29.32087912087912, "grad_norm": 1.171875, "learning_rate": 7.958608905824073e-06, "loss": 0.1482, "step": 40023 }, { "epoch": 29.32161172161172, "grad_norm": 1.0703125, "learning_rate": 7.956857222894325e-06, "loss": 0.1266, "step": 40024 }, { "epoch": 29.32234432234432, "grad_norm": 1.4609375, "learning_rate": 7.955105743802734e-06, "loss": 0.1423, "step": 40025 }, { "epoch": 29.323076923076922, "grad_norm": 0.8515625, "learning_rate": 7.95335446855772e-06, "loss": 0.0971, "step": 40026 }, { "epoch": 29.323809523809523, "grad_norm": 0.9765625, "learning_rate": 7.951603397167632e-06, "loss": 0.1158, "step": 40027 }, { "epoch": 29.324542124542123, "grad_norm": 0.90234375, "learning_rate": 7.949852529640896e-06, "loss": 0.147, "step": 40028 }, { "epoch": 29.325274725274724, "grad_norm": 1.15625, "learning_rate": 7.948101865985859e-06, "loss": 0.1399, "step": 40029 }, { "epoch": 29.326007326007325, "grad_norm": 0.94140625, "learning_rate": 7.946351406210962e-06, "loss": 0.1119, "step": 40030 }, { "epoch": 29.326739926739926, "grad_norm": 1.03125, "learning_rate": 7.94460115032455e-06, "loss": 0.1124, "step": 40031 }, { "epoch": 29.327472527472526, "grad_norm": 0.8359375, "learning_rate": 7.942851098335019e-06, "loss": 0.1081, "step": 40032 }, { "epoch": 29.328205128205127, "grad_norm": 0.9140625, "learning_rate": 7.941101250250779e-06, "loss": 0.1191, "step": 40033 }, { "epoch": 29.328937728937728, "grad_norm": 1.078125, "learning_rate": 7.939351606080178e-06, "loss": 0.1585, "step": 40034 }, { "epoch": 29.32967032967033, "grad_norm": 1.234375, "learning_rate": 7.937602165831616e-06, "loss": 0.1365, "step": 40035 }, { "epoch": 29.33040293040293, "grad_norm": 1.0859375, "learning_rate": 7.935852929513477e-06, "loss": 0.1233, "step": 40036 }, { "epoch": 29.33113553113553, "grad_norm": 1.46875, "learning_rate": 7.93410389713415e-06, "loss": 0.139, "step": 40037 }, { "epoch": 29.33186813186813, "grad_norm": 0.9609375, "learning_rate": 7.932355068701991e-06, "loss": 0.1196, "step": 40038 }, { "epoch": 29.33260073260073, "grad_norm": 0.83203125, "learning_rate": 7.930606444225406e-06, "loss": 0.1548, "step": 40039 }, { "epoch": 29.333333333333332, "grad_norm": 1.1328125, "learning_rate": 7.928858023712733e-06, "loss": 0.1291, "step": 40040 }, { "epoch": 29.334065934065933, "grad_norm": 0.890625, "learning_rate": 7.927109807172402e-06, "loss": 0.1214, "step": 40041 }, { "epoch": 29.334798534798534, "grad_norm": 1.0546875, "learning_rate": 7.925361794612755e-06, "loss": 0.1168, "step": 40042 }, { "epoch": 29.335531135531134, "grad_norm": 0.6875, "learning_rate": 7.923613986042184e-06, "loss": 0.0931, "step": 40043 }, { "epoch": 29.336263736263735, "grad_norm": 1.2265625, "learning_rate": 7.92186638146905e-06, "loss": 0.149, "step": 40044 }, { "epoch": 29.336996336996336, "grad_norm": 1.09375, "learning_rate": 7.920118980901724e-06, "loss": 0.1119, "step": 40045 }, { "epoch": 29.337728937728937, "grad_norm": 0.88671875, "learning_rate": 7.91837178434859e-06, "loss": 0.1187, "step": 40046 }, { "epoch": 29.338461538461537, "grad_norm": 0.97265625, "learning_rate": 7.916624791818016e-06, "loss": 0.1117, "step": 40047 }, { "epoch": 29.339194139194138, "grad_norm": 1.1796875, "learning_rate": 7.914878003318378e-06, "loss": 0.1177, "step": 40048 }, { "epoch": 29.33992673992674, "grad_norm": 1.21875, "learning_rate": 7.913131418858035e-06, "loss": 0.1506, "step": 40049 }, { "epoch": 29.34065934065934, "grad_norm": 1.1328125, "learning_rate": 7.911385038445352e-06, "loss": 0.1266, "step": 40050 }, { "epoch": 29.34139194139194, "grad_norm": 1.4765625, "learning_rate": 7.909638862088709e-06, "loss": 0.1245, "step": 40051 }, { "epoch": 29.34212454212454, "grad_norm": 1.1875, "learning_rate": 7.907892889796482e-06, "loss": 0.1493, "step": 40052 }, { "epoch": 29.34285714285714, "grad_norm": 1.0390625, "learning_rate": 7.906147121577006e-06, "loss": 0.1321, "step": 40053 }, { "epoch": 29.343589743589742, "grad_norm": 1.3203125, "learning_rate": 7.904401557438674e-06, "loss": 0.1569, "step": 40054 }, { "epoch": 29.344322344322343, "grad_norm": 1.0078125, "learning_rate": 7.902656197389818e-06, "loss": 0.115, "step": 40055 }, { "epoch": 29.345054945054944, "grad_norm": 0.890625, "learning_rate": 7.900911041438835e-06, "loss": 0.1172, "step": 40056 }, { "epoch": 29.345787545787545, "grad_norm": 0.87109375, "learning_rate": 7.899166089594067e-06, "loss": 0.1029, "step": 40057 }, { "epoch": 29.346520146520145, "grad_norm": 0.76171875, "learning_rate": 7.897421341863877e-06, "loss": 0.1054, "step": 40058 }, { "epoch": 29.347252747252746, "grad_norm": 1.40625, "learning_rate": 7.895676798256644e-06, "loss": 0.115, "step": 40059 }, { "epoch": 29.347985347985347, "grad_norm": 1.1875, "learning_rate": 7.893932458780688e-06, "loss": 0.1205, "step": 40060 }, { "epoch": 29.348717948717947, "grad_norm": 1.15625, "learning_rate": 7.892188323444396e-06, "loss": 0.1252, "step": 40061 }, { "epoch": 29.349450549450548, "grad_norm": 0.78515625, "learning_rate": 7.89044439225611e-06, "loss": 0.1012, "step": 40062 }, { "epoch": 29.35018315018315, "grad_norm": 1.0859375, "learning_rate": 7.888700665224211e-06, "loss": 0.1003, "step": 40063 }, { "epoch": 29.35091575091575, "grad_norm": 1.0859375, "learning_rate": 7.88695714235702e-06, "loss": 0.1161, "step": 40064 }, { "epoch": 29.35164835164835, "grad_norm": 1.5078125, "learning_rate": 7.885213823662923e-06, "loss": 0.1549, "step": 40065 }, { "epoch": 29.35238095238095, "grad_norm": 1.0234375, "learning_rate": 7.883470709150223e-06, "loss": 0.1541, "step": 40066 }, { "epoch": 29.353113553113552, "grad_norm": 1.28125, "learning_rate": 7.881727798827338e-06, "loss": 0.0928, "step": 40067 }, { "epoch": 29.353846153846153, "grad_norm": 0.9296875, "learning_rate": 7.879985092702565e-06, "loss": 0.0977, "step": 40068 }, { "epoch": 29.354578754578753, "grad_norm": 0.93359375, "learning_rate": 7.878242590784283e-06, "loss": 0.1272, "step": 40069 }, { "epoch": 29.355311355311354, "grad_norm": 0.91796875, "learning_rate": 7.876500293080843e-06, "loss": 0.1236, "step": 40070 }, { "epoch": 29.356043956043955, "grad_norm": 0.87890625, "learning_rate": 7.874758199600568e-06, "loss": 0.1369, "step": 40071 }, { "epoch": 29.356776556776556, "grad_norm": 1.3203125, "learning_rate": 7.873016310351821e-06, "loss": 0.1219, "step": 40072 }, { "epoch": 29.357509157509156, "grad_norm": 0.83203125, "learning_rate": 7.871274625342945e-06, "loss": 0.1054, "step": 40073 }, { "epoch": 29.358241758241757, "grad_norm": 0.9375, "learning_rate": 7.8695331445823e-06, "loss": 0.1412, "step": 40074 }, { "epoch": 29.358974358974358, "grad_norm": 0.83984375, "learning_rate": 7.867791868078202e-06, "loss": 0.1065, "step": 40075 }, { "epoch": 29.35970695970696, "grad_norm": 1.765625, "learning_rate": 7.86605079583901e-06, "loss": 0.1242, "step": 40076 }, { "epoch": 29.36043956043956, "grad_norm": 0.6796875, "learning_rate": 7.86430992787306e-06, "loss": 0.0885, "step": 40077 }, { "epoch": 29.36117216117216, "grad_norm": 1.1171875, "learning_rate": 7.862569264188717e-06, "loss": 0.1021, "step": 40078 }, { "epoch": 29.36190476190476, "grad_norm": 1.21875, "learning_rate": 7.860828804794287e-06, "loss": 0.13, "step": 40079 }, { "epoch": 29.36263736263736, "grad_norm": 0.94921875, "learning_rate": 7.859088549698133e-06, "loss": 0.1184, "step": 40080 }, { "epoch": 29.363369963369962, "grad_norm": 1.3828125, "learning_rate": 7.857348498908562e-06, "loss": 0.0991, "step": 40081 }, { "epoch": 29.364102564102563, "grad_norm": 1.15625, "learning_rate": 7.855608652433956e-06, "loss": 0.1098, "step": 40082 }, { "epoch": 29.364835164835164, "grad_norm": 1.015625, "learning_rate": 7.85386901028262e-06, "loss": 0.1189, "step": 40083 }, { "epoch": 29.365567765567764, "grad_norm": 1.1875, "learning_rate": 7.852129572462894e-06, "loss": 0.1356, "step": 40084 }, { "epoch": 29.366300366300365, "grad_norm": 1.53125, "learning_rate": 7.850390338983124e-06, "loss": 0.1237, "step": 40085 }, { "epoch": 29.367032967032966, "grad_norm": 0.765625, "learning_rate": 7.848651309851617e-06, "loss": 0.1314, "step": 40086 }, { "epoch": 29.367765567765566, "grad_norm": 0.890625, "learning_rate": 7.846912485076748e-06, "loss": 0.1084, "step": 40087 }, { "epoch": 29.368498168498167, "grad_norm": 0.671875, "learning_rate": 7.845173864666806e-06, "loss": 0.1244, "step": 40088 }, { "epoch": 29.369230769230768, "grad_norm": 0.67578125, "learning_rate": 7.843435448630153e-06, "loss": 0.0998, "step": 40089 }, { "epoch": 29.36996336996337, "grad_norm": 1.1328125, "learning_rate": 7.84169723697509e-06, "loss": 0.1374, "step": 40090 }, { "epoch": 29.37069597069597, "grad_norm": 1.2734375, "learning_rate": 7.839959229709973e-06, "loss": 0.1168, "step": 40091 }, { "epoch": 29.37142857142857, "grad_norm": 1.2265625, "learning_rate": 7.838221426843097e-06, "loss": 0.1331, "step": 40092 }, { "epoch": 29.37216117216117, "grad_norm": 1.234375, "learning_rate": 7.836483828382825e-06, "loss": 0.0958, "step": 40093 }, { "epoch": 29.37289377289377, "grad_norm": 1.234375, "learning_rate": 7.834746434337452e-06, "loss": 0.1226, "step": 40094 }, { "epoch": 29.373626373626372, "grad_norm": 0.54296875, "learning_rate": 7.833009244715319e-06, "loss": 0.1056, "step": 40095 }, { "epoch": 29.374358974358973, "grad_norm": 1.0546875, "learning_rate": 7.83127225952475e-06, "loss": 0.1084, "step": 40096 }, { "epoch": 29.375091575091574, "grad_norm": 0.95703125, "learning_rate": 7.829535478774056e-06, "loss": 0.1278, "step": 40097 }, { "epoch": 29.375824175824174, "grad_norm": 0.8671875, "learning_rate": 7.82779890247157e-06, "loss": 0.1149, "step": 40098 }, { "epoch": 29.376556776556775, "grad_norm": 1.1953125, "learning_rate": 7.826062530625598e-06, "loss": 0.121, "step": 40099 }, { "epoch": 29.377289377289376, "grad_norm": 0.71875, "learning_rate": 7.82432636324449e-06, "loss": 0.1386, "step": 40100 }, { "epoch": 29.378021978021977, "grad_norm": 1.359375, "learning_rate": 7.822590400336529e-06, "loss": 0.118, "step": 40101 }, { "epoch": 29.378754578754577, "grad_norm": 0.84375, "learning_rate": 7.820854641910042e-06, "loss": 0.1214, "step": 40102 }, { "epoch": 29.379487179487178, "grad_norm": 0.88671875, "learning_rate": 7.81911908797336e-06, "loss": 0.111, "step": 40103 }, { "epoch": 29.38021978021978, "grad_norm": 0.7578125, "learning_rate": 7.817383738534795e-06, "loss": 0.1197, "step": 40104 }, { "epoch": 29.38095238095238, "grad_norm": 0.90234375, "learning_rate": 7.815648593602646e-06, "loss": 0.1186, "step": 40105 }, { "epoch": 29.38168498168498, "grad_norm": 1.2890625, "learning_rate": 7.813913653185248e-06, "loss": 0.1209, "step": 40106 }, { "epoch": 29.38241758241758, "grad_norm": 0.875, "learning_rate": 7.812178917290886e-06, "loss": 0.1335, "step": 40107 }, { "epoch": 29.383150183150182, "grad_norm": 0.83984375, "learning_rate": 7.810444385927905e-06, "loss": 0.139, "step": 40108 }, { "epoch": 29.383882783882783, "grad_norm": 0.91796875, "learning_rate": 7.80871005910459e-06, "loss": 0.1226, "step": 40109 }, { "epoch": 29.384615384615383, "grad_norm": 1.515625, "learning_rate": 7.806975936829258e-06, "loss": 0.1, "step": 40110 }, { "epoch": 29.385347985347984, "grad_norm": 0.765625, "learning_rate": 7.805242019110231e-06, "loss": 0.1271, "step": 40111 }, { "epoch": 29.386080586080585, "grad_norm": 0.65234375, "learning_rate": 7.80350830595578e-06, "loss": 0.1131, "step": 40112 }, { "epoch": 29.386813186813185, "grad_norm": 1.2578125, "learning_rate": 7.801774797374264e-06, "loss": 0.1429, "step": 40113 }, { "epoch": 29.387545787545786, "grad_norm": 0.98046875, "learning_rate": 7.800041493373948e-06, "loss": 0.1146, "step": 40114 }, { "epoch": 29.388278388278387, "grad_norm": 1.2578125, "learning_rate": 7.798308393963163e-06, "loss": 0.1649, "step": 40115 }, { "epoch": 29.389010989010988, "grad_norm": 1.2109375, "learning_rate": 7.796575499150193e-06, "loss": 0.131, "step": 40116 }, { "epoch": 29.38974358974359, "grad_norm": 1.46875, "learning_rate": 7.794842808943353e-06, "loss": 0.1206, "step": 40117 }, { "epoch": 29.39047619047619, "grad_norm": 0.8984375, "learning_rate": 7.79311032335092e-06, "loss": 0.0915, "step": 40118 }, { "epoch": 29.39120879120879, "grad_norm": 0.74609375, "learning_rate": 7.791378042381239e-06, "loss": 0.1048, "step": 40119 }, { "epoch": 29.39194139194139, "grad_norm": 1.2109375, "learning_rate": 7.789645966042574e-06, "loss": 0.1171, "step": 40120 }, { "epoch": 29.39267399267399, "grad_norm": 0.87109375, "learning_rate": 7.787914094343238e-06, "loss": 0.1166, "step": 40121 }, { "epoch": 29.393406593406592, "grad_norm": 0.8984375, "learning_rate": 7.786182427291538e-06, "loss": 0.1206, "step": 40122 }, { "epoch": 29.394139194139193, "grad_norm": 1.125, "learning_rate": 7.784450964895743e-06, "loss": 0.1413, "step": 40123 }, { "epoch": 29.394871794871793, "grad_norm": 1.171875, "learning_rate": 7.782719707164173e-06, "loss": 0.1263, "step": 40124 }, { "epoch": 29.395604395604394, "grad_norm": 0.8671875, "learning_rate": 7.780988654105111e-06, "loss": 0.1224, "step": 40125 }, { "epoch": 29.396336996336995, "grad_norm": 0.88671875, "learning_rate": 7.779257805726873e-06, "loss": 0.1262, "step": 40126 }, { "epoch": 29.397069597069596, "grad_norm": 1.296875, "learning_rate": 7.77752716203772e-06, "loss": 0.1052, "step": 40127 }, { "epoch": 29.397802197802196, "grad_norm": 1.6484375, "learning_rate": 7.77579672304596e-06, "loss": 0.0991, "step": 40128 }, { "epoch": 29.398534798534797, "grad_norm": 1.140625, "learning_rate": 7.774066488759882e-06, "loss": 0.0961, "step": 40129 }, { "epoch": 29.399267399267398, "grad_norm": 1.4921875, "learning_rate": 7.772336459187793e-06, "loss": 0.1033, "step": 40130 }, { "epoch": 29.4, "grad_norm": 0.9453125, "learning_rate": 7.770606634337949e-06, "loss": 0.1285, "step": 40131 }, { "epoch": 29.4007326007326, "grad_norm": 1.046875, "learning_rate": 7.76887701421867e-06, "loss": 0.1453, "step": 40132 }, { "epoch": 29.4014652014652, "grad_norm": 1.140625, "learning_rate": 7.76714759883821e-06, "loss": 0.1045, "step": 40133 }, { "epoch": 29.4021978021978, "grad_norm": 0.8359375, "learning_rate": 7.765418388204891e-06, "loss": 0.1158, "step": 40134 }, { "epoch": 29.4029304029304, "grad_norm": 0.8984375, "learning_rate": 7.763689382326976e-06, "loss": 0.1287, "step": 40135 }, { "epoch": 29.403663003663002, "grad_norm": 1.0234375, "learning_rate": 7.761960581212748e-06, "loss": 0.127, "step": 40136 }, { "epoch": 29.404395604395603, "grad_norm": 1.140625, "learning_rate": 7.76023198487051e-06, "loss": 0.1065, "step": 40137 }, { "epoch": 29.405128205128204, "grad_norm": 0.7734375, "learning_rate": 7.758503593308514e-06, "loss": 0.1218, "step": 40138 }, { "epoch": 29.405860805860804, "grad_norm": 1.109375, "learning_rate": 7.756775406535072e-06, "loss": 0.1097, "step": 40139 }, { "epoch": 29.406593406593405, "grad_norm": 1.09375, "learning_rate": 7.75504742455845e-06, "loss": 0.1235, "step": 40140 }, { "epoch": 29.407326007326006, "grad_norm": 1.0390625, "learning_rate": 7.753319647386929e-06, "loss": 0.129, "step": 40141 }, { "epoch": 29.408058608058607, "grad_norm": 0.8984375, "learning_rate": 7.751592075028781e-06, "loss": 0.1205, "step": 40142 }, { "epoch": 29.408791208791207, "grad_norm": 1.2734375, "learning_rate": 7.749864707492299e-06, "loss": 0.1216, "step": 40143 }, { "epoch": 29.409523809523808, "grad_norm": 0.77734375, "learning_rate": 7.748137544785726e-06, "loss": 0.0966, "step": 40144 }, { "epoch": 29.41025641025641, "grad_norm": 1.203125, "learning_rate": 7.746410586917383e-06, "loss": 0.1119, "step": 40145 }, { "epoch": 29.41098901098901, "grad_norm": 0.62109375, "learning_rate": 7.744683833895511e-06, "loss": 0.109, "step": 40146 }, { "epoch": 29.41172161172161, "grad_norm": 1.0078125, "learning_rate": 7.742957285728395e-06, "loss": 0.1161, "step": 40147 }, { "epoch": 29.41245421245421, "grad_norm": 1.4453125, "learning_rate": 7.741230942424314e-06, "loss": 0.0991, "step": 40148 }, { "epoch": 29.41318681318681, "grad_norm": 1.4375, "learning_rate": 7.739504803991526e-06, "loss": 0.1342, "step": 40149 }, { "epoch": 29.413919413919412, "grad_norm": 0.8828125, "learning_rate": 7.73777887043831e-06, "loss": 0.0981, "step": 40150 }, { "epoch": 29.414652014652013, "grad_norm": 1.171875, "learning_rate": 7.736053141772927e-06, "loss": 0.1402, "step": 40151 }, { "epoch": 29.415384615384614, "grad_norm": 0.8515625, "learning_rate": 7.734327618003668e-06, "loss": 0.1085, "step": 40152 }, { "epoch": 29.416117216117215, "grad_norm": 0.7421875, "learning_rate": 7.732602299138769e-06, "loss": 0.1103, "step": 40153 }, { "epoch": 29.416849816849815, "grad_norm": 1.046875, "learning_rate": 7.730877185186515e-06, "loss": 0.1163, "step": 40154 }, { "epoch": 29.417582417582416, "grad_norm": 1.4453125, "learning_rate": 7.729152276155166e-06, "loss": 0.099, "step": 40155 }, { "epoch": 29.418315018315017, "grad_norm": 0.7890625, "learning_rate": 7.727427572053003e-06, "loss": 0.1288, "step": 40156 }, { "epoch": 29.419047619047618, "grad_norm": 1.7265625, "learning_rate": 7.725703072888265e-06, "loss": 0.1217, "step": 40157 }, { "epoch": 29.41978021978022, "grad_norm": 0.9375, "learning_rate": 7.723978778669236e-06, "loss": 0.1253, "step": 40158 }, { "epoch": 29.42051282051282, "grad_norm": 0.84375, "learning_rate": 7.722254689404146e-06, "loss": 0.1226, "step": 40159 }, { "epoch": 29.42124542124542, "grad_norm": 0.76953125, "learning_rate": 7.720530805101291e-06, "loss": 0.11, "step": 40160 }, { "epoch": 29.42197802197802, "grad_norm": 1.0390625, "learning_rate": 7.718807125768915e-06, "loss": 0.1289, "step": 40161 }, { "epoch": 29.42271062271062, "grad_norm": 1.03125, "learning_rate": 7.717083651415273e-06, "loss": 0.1267, "step": 40162 }, { "epoch": 29.423443223443222, "grad_norm": 0.91796875, "learning_rate": 7.715360382048642e-06, "loss": 0.1102, "step": 40163 }, { "epoch": 29.424175824175823, "grad_norm": 0.76953125, "learning_rate": 7.71363731767724e-06, "loss": 0.1239, "step": 40164 }, { "epoch": 29.424908424908423, "grad_norm": 1.2265625, "learning_rate": 7.711914458309375e-06, "loss": 0.1231, "step": 40165 }, { "epoch": 29.425641025641024, "grad_norm": 0.765625, "learning_rate": 7.710191803953256e-06, "loss": 0.1108, "step": 40166 }, { "epoch": 29.426373626373625, "grad_norm": 0.73828125, "learning_rate": 7.70846935461717e-06, "loss": 0.1159, "step": 40167 }, { "epoch": 29.427106227106226, "grad_norm": 1.03125, "learning_rate": 7.706747110309342e-06, "loss": 0.1276, "step": 40168 }, { "epoch": 29.427838827838826, "grad_norm": 0.94140625, "learning_rate": 7.70502507103805e-06, "loss": 0.1163, "step": 40169 }, { "epoch": 29.428571428571427, "grad_norm": 2.5625, "learning_rate": 7.703303236811509e-06, "loss": 0.1708, "step": 40170 }, { "epoch": 29.429304029304028, "grad_norm": 0.9453125, "learning_rate": 7.701581607638011e-06, "loss": 0.1204, "step": 40171 }, { "epoch": 29.43003663003663, "grad_norm": 1.046875, "learning_rate": 7.699860183525778e-06, "loss": 0.1126, "step": 40172 }, { "epoch": 29.43076923076923, "grad_norm": 1.1953125, "learning_rate": 7.698138964483065e-06, "loss": 0.123, "step": 40173 }, { "epoch": 29.43150183150183, "grad_norm": 0.8828125, "learning_rate": 7.69641795051813e-06, "loss": 0.1118, "step": 40174 }, { "epoch": 29.43223443223443, "grad_norm": 0.439453125, "learning_rate": 7.694697141639194e-06, "loss": 0.0945, "step": 40175 }, { "epoch": 29.43296703296703, "grad_norm": 0.8828125, "learning_rate": 7.692976537854517e-06, "loss": 0.0984, "step": 40176 }, { "epoch": 29.433699633699632, "grad_norm": 1.0859375, "learning_rate": 7.691256139172341e-06, "loss": 0.1138, "step": 40177 }, { "epoch": 29.434432234432233, "grad_norm": 1.0546875, "learning_rate": 7.689535945600922e-06, "loss": 0.125, "step": 40178 }, { "epoch": 29.435164835164834, "grad_norm": 0.59765625, "learning_rate": 7.68781595714848e-06, "loss": 0.1137, "step": 40179 }, { "epoch": 29.435897435897434, "grad_norm": 1.171875, "learning_rate": 7.686096173823265e-06, "loss": 0.092, "step": 40180 }, { "epoch": 29.436630036630035, "grad_norm": 0.9921875, "learning_rate": 7.684376595633521e-06, "loss": 0.1444, "step": 40181 }, { "epoch": 29.437362637362636, "grad_norm": 0.91796875, "learning_rate": 7.682657222587496e-06, "loss": 0.0921, "step": 40182 }, { "epoch": 29.438095238095237, "grad_norm": 1.5234375, "learning_rate": 7.6809380546934e-06, "loss": 0.1133, "step": 40183 }, { "epoch": 29.438827838827837, "grad_norm": 1.5390625, "learning_rate": 7.679219091959499e-06, "loss": 0.1115, "step": 40184 }, { "epoch": 29.439560439560438, "grad_norm": 0.91796875, "learning_rate": 7.677500334393995e-06, "loss": 0.1025, "step": 40185 }, { "epoch": 29.44029304029304, "grad_norm": 1.2421875, "learning_rate": 7.675781782005166e-06, "loss": 0.099, "step": 40186 }, { "epoch": 29.44102564102564, "grad_norm": 0.953125, "learning_rate": 7.674063434801214e-06, "loss": 0.106, "step": 40187 }, { "epoch": 29.44175824175824, "grad_norm": 1.7109375, "learning_rate": 7.672345292790385e-06, "loss": 0.1382, "step": 40188 }, { "epoch": 29.44249084249084, "grad_norm": 1.09375, "learning_rate": 7.670627355980917e-06, "loss": 0.1171, "step": 40189 }, { "epoch": 29.44322344322344, "grad_norm": 1.140625, "learning_rate": 7.668909624381015e-06, "loss": 0.1324, "step": 40190 }, { "epoch": 29.443956043956042, "grad_norm": 1.328125, "learning_rate": 7.667192097998947e-06, "loss": 0.1385, "step": 40191 }, { "epoch": 29.444688644688643, "grad_norm": 1.3359375, "learning_rate": 7.665474776842912e-06, "loss": 0.128, "step": 40192 }, { "epoch": 29.445421245421244, "grad_norm": 1.1484375, "learning_rate": 7.663757660921162e-06, "loss": 0.1031, "step": 40193 }, { "epoch": 29.446153846153845, "grad_norm": 1.03125, "learning_rate": 7.6620407502419e-06, "loss": 0.1284, "step": 40194 }, { "epoch": 29.446886446886445, "grad_norm": 1.4375, "learning_rate": 7.660324044813368e-06, "loss": 0.1598, "step": 40195 }, { "epoch": 29.447619047619046, "grad_norm": 1.09375, "learning_rate": 7.65860754464377e-06, "loss": 0.1115, "step": 40196 }, { "epoch": 29.448351648351647, "grad_norm": 1.0078125, "learning_rate": 7.65689124974137e-06, "loss": 0.1125, "step": 40197 }, { "epoch": 29.449084249084247, "grad_norm": 1.1875, "learning_rate": 7.655175160114358e-06, "loss": 0.1204, "step": 40198 }, { "epoch": 29.449816849816848, "grad_norm": 1.234375, "learning_rate": 7.653459275770966e-06, "loss": 0.14, "step": 40199 }, { "epoch": 29.45054945054945, "grad_norm": 0.921875, "learning_rate": 7.651743596719423e-06, "loss": 0.1167, "step": 40200 }, { "epoch": 29.45128205128205, "grad_norm": 1.0078125, "learning_rate": 7.650028122967925e-06, "loss": 0.1012, "step": 40201 }, { "epoch": 29.45201465201465, "grad_norm": 0.76953125, "learning_rate": 7.64831285452473e-06, "loss": 0.1039, "step": 40202 }, { "epoch": 29.45274725274725, "grad_norm": 1.15625, "learning_rate": 7.646597791398025e-06, "loss": 0.1111, "step": 40203 }, { "epoch": 29.453479853479852, "grad_norm": 0.9375, "learning_rate": 7.644882933596046e-06, "loss": 0.1341, "step": 40204 }, { "epoch": 29.454212454212453, "grad_norm": 1.5, "learning_rate": 7.643168281126995e-06, "loss": 0.1527, "step": 40205 }, { "epoch": 29.454945054945053, "grad_norm": 0.78125, "learning_rate": 7.641453833999095e-06, "loss": 0.1116, "step": 40206 }, { "epoch": 29.455677655677654, "grad_norm": 1.296875, "learning_rate": 7.639739592220552e-06, "loss": 0.1336, "step": 40207 }, { "epoch": 29.456410256410255, "grad_norm": 1.265625, "learning_rate": 7.638025555799605e-06, "loss": 0.1262, "step": 40208 }, { "epoch": 29.457142857142856, "grad_norm": 1.21875, "learning_rate": 7.636311724744433e-06, "loss": 0.135, "step": 40209 }, { "epoch": 29.457875457875456, "grad_norm": 0.8203125, "learning_rate": 7.634598099063264e-06, "loss": 0.1287, "step": 40210 }, { "epoch": 29.458608058608057, "grad_norm": 1.09375, "learning_rate": 7.632884678764305e-06, "loss": 0.1264, "step": 40211 }, { "epoch": 29.459340659340658, "grad_norm": 0.65625, "learning_rate": 7.631171463855779e-06, "loss": 0.1123, "step": 40212 }, { "epoch": 29.46007326007326, "grad_norm": 1.421875, "learning_rate": 7.629458454345873e-06, "loss": 0.127, "step": 40213 }, { "epoch": 29.46080586080586, "grad_norm": 1.34375, "learning_rate": 7.627745650242808e-06, "loss": 0.1311, "step": 40214 }, { "epoch": 29.46153846153846, "grad_norm": 0.7578125, "learning_rate": 7.626033051554794e-06, "loss": 0.1004, "step": 40215 }, { "epoch": 29.46227106227106, "grad_norm": 0.92578125, "learning_rate": 7.624320658290024e-06, "loss": 0.1419, "step": 40216 }, { "epoch": 29.46300366300366, "grad_norm": 1.3828125, "learning_rate": 7.622608470456705e-06, "loss": 0.1575, "step": 40217 }, { "epoch": 29.463736263736262, "grad_norm": 1.328125, "learning_rate": 7.620896488063049e-06, "loss": 0.1315, "step": 40218 }, { "epoch": 29.464468864468863, "grad_norm": 1.234375, "learning_rate": 7.619184711117261e-06, "loss": 0.1469, "step": 40219 }, { "epoch": 29.465201465201464, "grad_norm": 1.046875, "learning_rate": 7.617473139627523e-06, "loss": 0.1152, "step": 40220 }, { "epoch": 29.465934065934064, "grad_norm": 0.97265625, "learning_rate": 7.6157617736020635e-06, "loss": 0.1199, "step": 40221 }, { "epoch": 29.466666666666665, "grad_norm": 1.234375, "learning_rate": 7.614050613049048e-06, "loss": 0.1115, "step": 40222 }, { "epoch": 29.467399267399266, "grad_norm": 0.84765625, "learning_rate": 7.612339657976714e-06, "loss": 0.1372, "step": 40223 }, { "epoch": 29.468131868131866, "grad_norm": 1.2265625, "learning_rate": 7.610628908393227e-06, "loss": 0.1214, "step": 40224 }, { "epoch": 29.468864468864467, "grad_norm": 1.7109375, "learning_rate": 7.6089183643067956e-06, "loss": 0.1377, "step": 40225 }, { "epoch": 29.469597069597068, "grad_norm": 0.83984375, "learning_rate": 7.607208025725633e-06, "loss": 0.1421, "step": 40226 }, { "epoch": 29.47032967032967, "grad_norm": 0.890625, "learning_rate": 7.6054978926578895e-06, "loss": 0.1241, "step": 40227 }, { "epoch": 29.47106227106227, "grad_norm": 0.859375, "learning_rate": 7.603787965111814e-06, "loss": 0.1089, "step": 40228 }, { "epoch": 29.47179487179487, "grad_norm": 0.7578125, "learning_rate": 7.602078243095558e-06, "loss": 0.1091, "step": 40229 }, { "epoch": 29.47252747252747, "grad_norm": 1.28125, "learning_rate": 7.600368726617339e-06, "loss": 0.1284, "step": 40230 }, { "epoch": 29.47326007326007, "grad_norm": 1.3984375, "learning_rate": 7.598659415685327e-06, "loss": 0.132, "step": 40231 }, { "epoch": 29.473992673992672, "grad_norm": 1.0234375, "learning_rate": 7.596950310307717e-06, "loss": 0.1169, "step": 40232 }, { "epoch": 29.474725274725273, "grad_norm": 1.0, "learning_rate": 7.595241410492704e-06, "loss": 0.1256, "step": 40233 }, { "epoch": 29.475457875457874, "grad_norm": 1.34375, "learning_rate": 7.593532716248491e-06, "loss": 0.1378, "step": 40234 }, { "epoch": 29.476190476190474, "grad_norm": 1.3203125, "learning_rate": 7.591824227583232e-06, "loss": 0.1221, "step": 40235 }, { "epoch": 29.476923076923075, "grad_norm": 0.90234375, "learning_rate": 7.590115944505127e-06, "loss": 0.1007, "step": 40236 }, { "epoch": 29.477655677655676, "grad_norm": 0.828125, "learning_rate": 7.588407867022364e-06, "loss": 0.1114, "step": 40237 }, { "epoch": 29.478388278388277, "grad_norm": 0.69921875, "learning_rate": 7.5866999951431424e-06, "loss": 0.1256, "step": 40238 }, { "epoch": 29.479120879120877, "grad_norm": 1.203125, "learning_rate": 7.584992328875613e-06, "loss": 0.1375, "step": 40239 }, { "epoch": 29.479853479853478, "grad_norm": 0.95703125, "learning_rate": 7.583284868227976e-06, "loss": 0.094, "step": 40240 }, { "epoch": 29.48058608058608, "grad_norm": 0.93359375, "learning_rate": 7.581577613208417e-06, "loss": 0.1075, "step": 40241 }, { "epoch": 29.48131868131868, "grad_norm": 1.1328125, "learning_rate": 7.579870563825099e-06, "loss": 0.1287, "step": 40242 }, { "epoch": 29.48205128205128, "grad_norm": 0.953125, "learning_rate": 7.5781637200862096e-06, "loss": 0.0944, "step": 40243 }, { "epoch": 29.48278388278388, "grad_norm": 1.546875, "learning_rate": 7.576457081999932e-06, "loss": 0.1418, "step": 40244 }, { "epoch": 29.483516483516482, "grad_norm": 0.91796875, "learning_rate": 7.57475064957445e-06, "loss": 0.1592, "step": 40245 }, { "epoch": 29.484249084249083, "grad_norm": 1.234375, "learning_rate": 7.5730444228179185e-06, "loss": 0.1473, "step": 40246 }, { "epoch": 29.484981684981683, "grad_norm": 0.7109375, "learning_rate": 7.57133840173853e-06, "loss": 0.094, "step": 40247 }, { "epoch": 29.485714285714284, "grad_norm": 1.5625, "learning_rate": 7.569632586344431e-06, "loss": 0.1624, "step": 40248 }, { "epoch": 29.486446886446885, "grad_norm": 0.66015625, "learning_rate": 7.567926976643837e-06, "loss": 0.1213, "step": 40249 }, { "epoch": 29.487179487179485, "grad_norm": 1.5, "learning_rate": 7.566221572644892e-06, "loss": 0.0962, "step": 40250 }, { "epoch": 29.487912087912086, "grad_norm": 1.1328125, "learning_rate": 7.564516374355768e-06, "loss": 0.1408, "step": 40251 }, { "epoch": 29.488644688644687, "grad_norm": 0.9453125, "learning_rate": 7.562811381784654e-06, "loss": 0.0882, "step": 40252 }, { "epoch": 29.489377289377288, "grad_norm": 0.953125, "learning_rate": 7.561106594939682e-06, "loss": 0.101, "step": 40253 }, { "epoch": 29.49010989010989, "grad_norm": 0.953125, "learning_rate": 7.55940201382907e-06, "loss": 0.1461, "step": 40254 }, { "epoch": 29.49084249084249, "grad_norm": 0.7109375, "learning_rate": 7.557697638460952e-06, "loss": 0.0989, "step": 40255 }, { "epoch": 29.49157509157509, "grad_norm": 0.8671875, "learning_rate": 7.555993468843507e-06, "loss": 0.1217, "step": 40256 }, { "epoch": 29.49230769230769, "grad_norm": 1.1640625, "learning_rate": 7.554289504984883e-06, "loss": 0.0952, "step": 40257 }, { "epoch": 29.493040293040295, "grad_norm": 0.921875, "learning_rate": 7.5525857468932565e-06, "loss": 0.1265, "step": 40258 }, { "epoch": 29.493772893772892, "grad_norm": 0.98046875, "learning_rate": 7.550882194576794e-06, "loss": 0.1284, "step": 40259 }, { "epoch": 29.494505494505496, "grad_norm": 1.28125, "learning_rate": 7.549178848043661e-06, "loss": 0.1254, "step": 40260 }, { "epoch": 29.495238095238093, "grad_norm": 1.140625, "learning_rate": 7.547475707302004e-06, "loss": 0.0812, "step": 40261 }, { "epoch": 29.495970695970698, "grad_norm": 0.79296875, "learning_rate": 7.545772772359988e-06, "loss": 0.1114, "step": 40262 }, { "epoch": 29.496703296703295, "grad_norm": 1.046875, "learning_rate": 7.54407004322578e-06, "loss": 0.1426, "step": 40263 }, { "epoch": 29.4974358974359, "grad_norm": 1.7265625, "learning_rate": 7.542367519907541e-06, "loss": 0.1347, "step": 40264 }, { "epoch": 29.498168498168496, "grad_norm": 1.5546875, "learning_rate": 7.540665202413415e-06, "loss": 0.1163, "step": 40265 }, { "epoch": 29.4989010989011, "grad_norm": 0.921875, "learning_rate": 7.538963090751556e-06, "loss": 0.1065, "step": 40266 }, { "epoch": 29.499633699633698, "grad_norm": 1.0859375, "learning_rate": 7.537261184930146e-06, "loss": 0.1117, "step": 40267 }, { "epoch": 29.500366300366302, "grad_norm": 0.828125, "learning_rate": 7.535559484957308e-06, "loss": 0.1238, "step": 40268 }, { "epoch": 29.5010989010989, "grad_norm": 1.109375, "learning_rate": 7.5338579908412116e-06, "loss": 0.1048, "step": 40269 }, { "epoch": 29.501831501831504, "grad_norm": 1.609375, "learning_rate": 7.532156702590005e-06, "loss": 0.0998, "step": 40270 }, { "epoch": 29.5025641025641, "grad_norm": 1.484375, "learning_rate": 7.530455620211851e-06, "loss": 0.1386, "step": 40271 }, { "epoch": 29.503296703296705, "grad_norm": 0.86328125, "learning_rate": 7.528754743714883e-06, "loss": 0.1125, "step": 40272 }, { "epoch": 29.504029304029302, "grad_norm": 1.1484375, "learning_rate": 7.527054073107265e-06, "loss": 0.1163, "step": 40273 }, { "epoch": 29.504761904761907, "grad_norm": 1.140625, "learning_rate": 7.525353608397116e-06, "loss": 0.1026, "step": 40274 }, { "epoch": 29.505494505494504, "grad_norm": 2.078125, "learning_rate": 7.523653349592625e-06, "loss": 0.1529, "step": 40275 }, { "epoch": 29.506227106227108, "grad_norm": 0.9609375, "learning_rate": 7.521953296701911e-06, "loss": 0.1011, "step": 40276 }, { "epoch": 29.506959706959705, "grad_norm": 1.2578125, "learning_rate": 7.520253449733129e-06, "loss": 0.124, "step": 40277 }, { "epoch": 29.50769230769231, "grad_norm": 0.984375, "learning_rate": 7.518553808694431e-06, "loss": 0.1151, "step": 40278 }, { "epoch": 29.508424908424907, "grad_norm": 1.125, "learning_rate": 7.516854373593929e-06, "loss": 0.1232, "step": 40279 }, { "epoch": 29.50915750915751, "grad_norm": 0.8828125, "learning_rate": 7.515155144439813e-06, "loss": 0.0969, "step": 40280 }, { "epoch": 29.509890109890108, "grad_norm": 0.8515625, "learning_rate": 7.513456121240187e-06, "loss": 0.1264, "step": 40281 }, { "epoch": 29.510622710622712, "grad_norm": 0.9609375, "learning_rate": 7.5117573040032144e-06, "loss": 0.109, "step": 40282 }, { "epoch": 29.51135531135531, "grad_norm": 1.4140625, "learning_rate": 7.510058692737014e-06, "loss": 0.1249, "step": 40283 }, { "epoch": 29.512087912087914, "grad_norm": 1.1484375, "learning_rate": 7.5083602874497334e-06, "loss": 0.1181, "step": 40284 }, { "epoch": 29.51282051282051, "grad_norm": 1.7265625, "learning_rate": 7.506662088149518e-06, "loss": 0.1348, "step": 40285 }, { "epoch": 29.513553113553115, "grad_norm": 0.953125, "learning_rate": 7.504964094844502e-06, "loss": 0.1228, "step": 40286 }, { "epoch": 29.514285714285712, "grad_norm": 1.109375, "learning_rate": 7.503266307542812e-06, "loss": 0.1397, "step": 40287 }, { "epoch": 29.515018315018317, "grad_norm": 0.76953125, "learning_rate": 7.5015687262525875e-06, "loss": 0.1064, "step": 40288 }, { "epoch": 29.515750915750917, "grad_norm": 1.09375, "learning_rate": 7.499871350981959e-06, "loss": 0.1013, "step": 40289 }, { "epoch": 29.516483516483518, "grad_norm": 0.84765625, "learning_rate": 7.4981741817390785e-06, "loss": 0.1053, "step": 40290 }, { "epoch": 29.51721611721612, "grad_norm": 1.125, "learning_rate": 7.496477218532045e-06, "loss": 0.1179, "step": 40291 }, { "epoch": 29.51794871794872, "grad_norm": 1.65625, "learning_rate": 7.49478046136901e-06, "loss": 0.1404, "step": 40292 }, { "epoch": 29.51868131868132, "grad_norm": 1.78125, "learning_rate": 7.493083910258111e-06, "loss": 0.1671, "step": 40293 }, { "epoch": 29.51941391941392, "grad_norm": 0.8984375, "learning_rate": 7.49138756520745e-06, "loss": 0.1465, "step": 40294 }, { "epoch": 29.520146520146522, "grad_norm": 0.91015625, "learning_rate": 7.4896914262251774e-06, "loss": 0.1307, "step": 40295 }, { "epoch": 29.520879120879123, "grad_norm": 1.21875, "learning_rate": 7.487995493319408e-06, "loss": 0.1002, "step": 40296 }, { "epoch": 29.521611721611723, "grad_norm": 0.9140625, "learning_rate": 7.486299766498282e-06, "loss": 0.1397, "step": 40297 }, { "epoch": 29.522344322344324, "grad_norm": 0.625, "learning_rate": 7.484604245769905e-06, "loss": 0.1203, "step": 40298 }, { "epoch": 29.523076923076925, "grad_norm": 1.0078125, "learning_rate": 7.482908931142422e-06, "loss": 0.1304, "step": 40299 }, { "epoch": 29.523809523809526, "grad_norm": 0.9140625, "learning_rate": 7.481213822623924e-06, "loss": 0.1182, "step": 40300 }, { "epoch": 29.524542124542126, "grad_norm": 0.765625, "learning_rate": 7.479518920222568e-06, "loss": 0.1019, "step": 40301 }, { "epoch": 29.525274725274727, "grad_norm": 0.89453125, "learning_rate": 7.477824223946453e-06, "loss": 0.1273, "step": 40302 }, { "epoch": 29.526007326007328, "grad_norm": 1.0078125, "learning_rate": 7.476129733803705e-06, "loss": 0.1023, "step": 40303 }, { "epoch": 29.52673992673993, "grad_norm": 0.71484375, "learning_rate": 7.474435449802451e-06, "loss": 0.1188, "step": 40304 }, { "epoch": 29.52747252747253, "grad_norm": 1.15625, "learning_rate": 7.472741371950784e-06, "loss": 0.1193, "step": 40305 }, { "epoch": 29.52820512820513, "grad_norm": 1.0390625, "learning_rate": 7.471047500256856e-06, "loss": 0.1237, "step": 40306 }, { "epoch": 29.52893772893773, "grad_norm": 0.83984375, "learning_rate": 7.469353834728758e-06, "loss": 0.1168, "step": 40307 }, { "epoch": 29.52967032967033, "grad_norm": 0.8671875, "learning_rate": 7.4676603753746186e-06, "loss": 0.111, "step": 40308 }, { "epoch": 29.530402930402932, "grad_norm": 0.84375, "learning_rate": 7.465967122202537e-06, "loss": 0.1202, "step": 40309 }, { "epoch": 29.531135531135533, "grad_norm": 1.6484375, "learning_rate": 7.464274075220632e-06, "loss": 0.14, "step": 40310 }, { "epoch": 29.531868131868134, "grad_norm": 1.046875, "learning_rate": 7.462581234437019e-06, "loss": 0.1435, "step": 40311 }, { "epoch": 29.532600732600734, "grad_norm": 1.265625, "learning_rate": 7.46088859985982e-06, "loss": 0.101, "step": 40312 }, { "epoch": 29.533333333333335, "grad_norm": 1.421875, "learning_rate": 7.4591961714971205e-06, "loss": 0.1396, "step": 40313 }, { "epoch": 29.534065934065936, "grad_norm": 0.77734375, "learning_rate": 7.457503949357042e-06, "loss": 0.1208, "step": 40314 }, { "epoch": 29.534798534798536, "grad_norm": 1.1328125, "learning_rate": 7.4558119334476855e-06, "loss": 0.1329, "step": 40315 }, { "epoch": 29.535531135531137, "grad_norm": 0.734375, "learning_rate": 7.454120123777175e-06, "loss": 0.1207, "step": 40316 }, { "epoch": 29.536263736263738, "grad_norm": 1.4140625, "learning_rate": 7.452428520353609e-06, "loss": 0.1238, "step": 40317 }, { "epoch": 29.53699633699634, "grad_norm": 0.62109375, "learning_rate": 7.450737123185082e-06, "loss": 0.1215, "step": 40318 }, { "epoch": 29.53772893772894, "grad_norm": 1.3984375, "learning_rate": 7.449045932279718e-06, "loss": 0.1345, "step": 40319 }, { "epoch": 29.53846153846154, "grad_norm": 0.91796875, "learning_rate": 7.447354947645595e-06, "loss": 0.1003, "step": 40320 }, { "epoch": 29.53919413919414, "grad_norm": 1.2109375, "learning_rate": 7.445664169290825e-06, "loss": 0.1182, "step": 40321 }, { "epoch": 29.53992673992674, "grad_norm": 0.9375, "learning_rate": 7.443973597223513e-06, "loss": 0.1383, "step": 40322 }, { "epoch": 29.540659340659342, "grad_norm": 0.94140625, "learning_rate": 7.4422832314517715e-06, "loss": 0.1158, "step": 40323 }, { "epoch": 29.541391941391943, "grad_norm": 0.7265625, "learning_rate": 7.440593071983673e-06, "loss": 0.1038, "step": 40324 }, { "epoch": 29.542124542124544, "grad_norm": 1.1953125, "learning_rate": 7.43890311882733e-06, "loss": 0.1165, "step": 40325 }, { "epoch": 29.542857142857144, "grad_norm": 0.99609375, "learning_rate": 7.437213371990836e-06, "loss": 0.1135, "step": 40326 }, { "epoch": 29.543589743589745, "grad_norm": 1.671875, "learning_rate": 7.435523831482304e-06, "loss": 0.1413, "step": 40327 }, { "epoch": 29.544322344322346, "grad_norm": 0.74609375, "learning_rate": 7.4338344973098e-06, "loss": 0.1404, "step": 40328 }, { "epoch": 29.545054945054947, "grad_norm": 1.0625, "learning_rate": 7.432145369481436e-06, "loss": 0.1084, "step": 40329 }, { "epoch": 29.545787545787547, "grad_norm": 0.87109375, "learning_rate": 7.430456448005307e-06, "loss": 0.0876, "step": 40330 }, { "epoch": 29.546520146520148, "grad_norm": 0.984375, "learning_rate": 7.428767732889481e-06, "loss": 0.1334, "step": 40331 }, { "epoch": 29.54725274725275, "grad_norm": 0.77734375, "learning_rate": 7.427079224142093e-06, "loss": 0.1037, "step": 40332 }, { "epoch": 29.54798534798535, "grad_norm": 0.96484375, "learning_rate": 7.425390921771192e-06, "loss": 0.152, "step": 40333 }, { "epoch": 29.54871794871795, "grad_norm": 1.03125, "learning_rate": 7.423702825784899e-06, "loss": 0.154, "step": 40334 }, { "epoch": 29.54945054945055, "grad_norm": 0.859375, "learning_rate": 7.422014936191274e-06, "loss": 0.1255, "step": 40335 }, { "epoch": 29.550183150183152, "grad_norm": 1.2109375, "learning_rate": 7.420327252998416e-06, "loss": 0.1287, "step": 40336 }, { "epoch": 29.550915750915753, "grad_norm": 1.3671875, "learning_rate": 7.418639776214413e-06, "loss": 0.1031, "step": 40337 }, { "epoch": 29.551648351648353, "grad_norm": 0.96484375, "learning_rate": 7.4169525058473636e-06, "loss": 0.1248, "step": 40338 }, { "epoch": 29.552380952380954, "grad_norm": 1.109375, "learning_rate": 7.415265441905327e-06, "loss": 0.1035, "step": 40339 }, { "epoch": 29.553113553113555, "grad_norm": 0.93359375, "learning_rate": 7.413578584396393e-06, "loss": 0.1034, "step": 40340 }, { "epoch": 29.553846153846155, "grad_norm": 1.109375, "learning_rate": 7.411891933328652e-06, "loss": 0.1188, "step": 40341 }, { "epoch": 29.554578754578756, "grad_norm": 0.9296875, "learning_rate": 7.4102054887101755e-06, "loss": 0.1256, "step": 40342 }, { "epoch": 29.555311355311357, "grad_norm": 1.171875, "learning_rate": 7.408519250549066e-06, "loss": 0.1376, "step": 40343 }, { "epoch": 29.556043956043958, "grad_norm": 1.4453125, "learning_rate": 7.406833218853375e-06, "loss": 0.1724, "step": 40344 }, { "epoch": 29.55677655677656, "grad_norm": 0.91796875, "learning_rate": 7.405147393631206e-06, "loss": 0.0916, "step": 40345 }, { "epoch": 29.55750915750916, "grad_norm": 0.9921875, "learning_rate": 7.403461774890606e-06, "loss": 0.1145, "step": 40346 }, { "epoch": 29.55824175824176, "grad_norm": 0.71875, "learning_rate": 7.4017763626396755e-06, "loss": 0.1108, "step": 40347 }, { "epoch": 29.55897435897436, "grad_norm": 1.4140625, "learning_rate": 7.400091156886477e-06, "loss": 0.1472, "step": 40348 }, { "epoch": 29.55970695970696, "grad_norm": 1.0234375, "learning_rate": 7.398406157639104e-06, "loss": 0.1183, "step": 40349 }, { "epoch": 29.560439560439562, "grad_norm": 1.1796875, "learning_rate": 7.396721364905604e-06, "loss": 0.1234, "step": 40350 }, { "epoch": 29.561172161172163, "grad_norm": 0.67578125, "learning_rate": 7.3950367786940644e-06, "loss": 0.1036, "step": 40351 }, { "epoch": 29.561904761904763, "grad_norm": 1.2265625, "learning_rate": 7.393352399012554e-06, "loss": 0.1255, "step": 40352 }, { "epoch": 29.562637362637364, "grad_norm": 0.7890625, "learning_rate": 7.3916682258691525e-06, "loss": 0.1171, "step": 40353 }, { "epoch": 29.563369963369965, "grad_norm": 0.875, "learning_rate": 7.389984259271908e-06, "loss": 0.1193, "step": 40354 }, { "epoch": 29.564102564102566, "grad_norm": 1.0, "learning_rate": 7.3883004992289025e-06, "loss": 0.0906, "step": 40355 }, { "epoch": 29.564835164835166, "grad_norm": 0.65234375, "learning_rate": 7.386616945748211e-06, "loss": 0.102, "step": 40356 }, { "epoch": 29.565567765567767, "grad_norm": 1.828125, "learning_rate": 7.38493359883787e-06, "loss": 0.1508, "step": 40357 }, { "epoch": 29.566300366300368, "grad_norm": 1.1875, "learning_rate": 7.383250458505993e-06, "loss": 0.1272, "step": 40358 }, { "epoch": 29.56703296703297, "grad_norm": 1.0625, "learning_rate": 7.381567524760599e-06, "loss": 0.1219, "step": 40359 }, { "epoch": 29.56776556776557, "grad_norm": 0.8671875, "learning_rate": 7.379884797609786e-06, "loss": 0.1145, "step": 40360 }, { "epoch": 29.56849816849817, "grad_norm": 1.375, "learning_rate": 7.378202277061583e-06, "loss": 0.1445, "step": 40361 }, { "epoch": 29.56923076923077, "grad_norm": 0.73828125, "learning_rate": 7.3765199631240735e-06, "loss": 0.1318, "step": 40362 }, { "epoch": 29.56996336996337, "grad_norm": 1.3515625, "learning_rate": 7.374837855805314e-06, "loss": 0.1223, "step": 40363 }, { "epoch": 29.570695970695972, "grad_norm": 1.3125, "learning_rate": 7.373155955113372e-06, "loss": 0.1489, "step": 40364 }, { "epoch": 29.571428571428573, "grad_norm": 0.8046875, "learning_rate": 7.371474261056288e-06, "loss": 0.1628, "step": 40365 }, { "epoch": 29.572161172161174, "grad_norm": 0.640625, "learning_rate": 7.36979277364213e-06, "loss": 0.113, "step": 40366 }, { "epoch": 29.572893772893774, "grad_norm": 1.3203125, "learning_rate": 7.368111492878953e-06, "loss": 0.1265, "step": 40367 }, { "epoch": 29.573626373626375, "grad_norm": 1.0234375, "learning_rate": 7.366430418774814e-06, "loss": 0.1313, "step": 40368 }, { "epoch": 29.574358974358976, "grad_norm": 0.92578125, "learning_rate": 7.364749551337775e-06, "loss": 0.1158, "step": 40369 }, { "epoch": 29.575091575091577, "grad_norm": 1.171875, "learning_rate": 7.363068890575872e-06, "loss": 0.1398, "step": 40370 }, { "epoch": 29.575824175824177, "grad_norm": 0.66796875, "learning_rate": 7.3613884364971765e-06, "loss": 0.1278, "step": 40371 }, { "epoch": 29.576556776556778, "grad_norm": 1.171875, "learning_rate": 7.3597081891097195e-06, "loss": 0.121, "step": 40372 }, { "epoch": 29.57728937728938, "grad_norm": 0.84375, "learning_rate": 7.3580281484215655e-06, "loss": 0.0997, "step": 40373 }, { "epoch": 29.57802197802198, "grad_norm": 0.69921875, "learning_rate": 7.356348314440757e-06, "loss": 0.1098, "step": 40374 }, { "epoch": 29.57875457875458, "grad_norm": 0.796875, "learning_rate": 7.35466868717536e-06, "loss": 0.0996, "step": 40375 }, { "epoch": 29.57948717948718, "grad_norm": 1.2265625, "learning_rate": 7.352989266633401e-06, "loss": 0.1191, "step": 40376 }, { "epoch": 29.58021978021978, "grad_norm": 0.78515625, "learning_rate": 7.351310052822927e-06, "loss": 0.1098, "step": 40377 }, { "epoch": 29.580952380952382, "grad_norm": 1.546875, "learning_rate": 7.3496310457519965e-06, "loss": 0.1246, "step": 40378 }, { "epoch": 29.581684981684983, "grad_norm": 1.5625, "learning_rate": 7.3479522454286614e-06, "loss": 0.1333, "step": 40379 }, { "epoch": 29.582417582417584, "grad_norm": 1.140625, "learning_rate": 7.346273651860938e-06, "loss": 0.1141, "step": 40380 }, { "epoch": 29.583150183150185, "grad_norm": 0.80078125, "learning_rate": 7.344595265056885e-06, "loss": 0.1244, "step": 40381 }, { "epoch": 29.583882783882785, "grad_norm": 1.1796875, "learning_rate": 7.342917085024558e-06, "loss": 0.1113, "step": 40382 }, { "epoch": 29.584615384615386, "grad_norm": 0.6953125, "learning_rate": 7.341239111771956e-06, "loss": 0.1243, "step": 40383 }, { "epoch": 29.585347985347987, "grad_norm": 1.46875, "learning_rate": 7.339561345307166e-06, "loss": 0.1543, "step": 40384 }, { "epoch": 29.586080586080588, "grad_norm": 1.359375, "learning_rate": 7.337883785638202e-06, "loss": 0.0961, "step": 40385 }, { "epoch": 29.58681318681319, "grad_norm": 0.98046875, "learning_rate": 7.336206432773111e-06, "loss": 0.121, "step": 40386 }, { "epoch": 29.58754578754579, "grad_norm": 0.71484375, "learning_rate": 7.33452928671992e-06, "loss": 0.104, "step": 40387 }, { "epoch": 29.58827838827839, "grad_norm": 0.7265625, "learning_rate": 7.332852347486667e-06, "loss": 0.1002, "step": 40388 }, { "epoch": 29.58901098901099, "grad_norm": 1.9453125, "learning_rate": 7.331175615081387e-06, "loss": 0.1346, "step": 40389 }, { "epoch": 29.58974358974359, "grad_norm": 1.3203125, "learning_rate": 7.3294990895121266e-06, "loss": 0.1294, "step": 40390 }, { "epoch": 29.590476190476192, "grad_norm": 1.3125, "learning_rate": 7.327822770786896e-06, "loss": 0.1127, "step": 40391 }, { "epoch": 29.591208791208793, "grad_norm": 0.69921875, "learning_rate": 7.326146658913745e-06, "loss": 0.1338, "step": 40392 }, { "epoch": 29.591941391941393, "grad_norm": 1.140625, "learning_rate": 7.324470753900692e-06, "loss": 0.1271, "step": 40393 }, { "epoch": 29.592673992673994, "grad_norm": 1.2578125, "learning_rate": 7.322795055755775e-06, "loss": 0.1046, "step": 40394 }, { "epoch": 29.593406593406595, "grad_norm": 0.91796875, "learning_rate": 7.321119564487033e-06, "loss": 0.1356, "step": 40395 }, { "epoch": 29.594139194139196, "grad_norm": 0.9765625, "learning_rate": 7.319444280102475e-06, "loss": 0.1143, "step": 40396 }, { "epoch": 29.594871794871796, "grad_norm": 1.125, "learning_rate": 7.317769202610138e-06, "loss": 0.1097, "step": 40397 }, { "epoch": 29.595604395604397, "grad_norm": 0.96484375, "learning_rate": 7.3160943320180404e-06, "loss": 0.1066, "step": 40398 }, { "epoch": 29.596336996336998, "grad_norm": 0.71484375, "learning_rate": 7.3144196683342074e-06, "loss": 0.1207, "step": 40399 }, { "epoch": 29.5970695970696, "grad_norm": 0.875, "learning_rate": 7.3127452115666685e-06, "loss": 0.1028, "step": 40400 }, { "epoch": 29.5978021978022, "grad_norm": 1.1640625, "learning_rate": 7.311070961723451e-06, "loss": 0.1444, "step": 40401 }, { "epoch": 29.5985347985348, "grad_norm": 0.7109375, "learning_rate": 7.309396918812564e-06, "loss": 0.115, "step": 40402 }, { "epoch": 29.5992673992674, "grad_norm": 1.7734375, "learning_rate": 7.307723082842035e-06, "loss": 0.1484, "step": 40403 }, { "epoch": 29.6, "grad_norm": 1.390625, "learning_rate": 7.30604945381988e-06, "loss": 0.1064, "step": 40404 }, { "epoch": 29.600732600732602, "grad_norm": 0.796875, "learning_rate": 7.304376031754139e-06, "loss": 0.1277, "step": 40405 }, { "epoch": 29.601465201465203, "grad_norm": 0.87890625, "learning_rate": 7.302702816652797e-06, "loss": 0.1449, "step": 40406 }, { "epoch": 29.602197802197804, "grad_norm": 0.8828125, "learning_rate": 7.301029808523888e-06, "loss": 0.1074, "step": 40407 }, { "epoch": 29.602930402930404, "grad_norm": 1.3203125, "learning_rate": 7.2993570073754376e-06, "loss": 0.1441, "step": 40408 }, { "epoch": 29.603663003663005, "grad_norm": 1.1875, "learning_rate": 7.2976844132154264e-06, "loss": 0.1195, "step": 40409 }, { "epoch": 29.604395604395606, "grad_norm": 0.8203125, "learning_rate": 7.296012026051912e-06, "loss": 0.1096, "step": 40410 }, { "epoch": 29.605128205128207, "grad_norm": 1.046875, "learning_rate": 7.294339845892878e-06, "loss": 0.0997, "step": 40411 }, { "epoch": 29.605860805860807, "grad_norm": 1.2734375, "learning_rate": 7.2926678727463605e-06, "loss": 0.1379, "step": 40412 }, { "epoch": 29.606593406593408, "grad_norm": 1.3671875, "learning_rate": 7.2909961066203375e-06, "loss": 0.1624, "step": 40413 }, { "epoch": 29.60732600732601, "grad_norm": 1.2578125, "learning_rate": 7.289324547522839e-06, "loss": 0.1484, "step": 40414 }, { "epoch": 29.60805860805861, "grad_norm": 0.77734375, "learning_rate": 7.287653195461871e-06, "loss": 0.1088, "step": 40415 }, { "epoch": 29.60879120879121, "grad_norm": 0.8125, "learning_rate": 7.285982050445454e-06, "loss": 0.1234, "step": 40416 }, { "epoch": 29.60952380952381, "grad_norm": 1.3828125, "learning_rate": 7.284311112481576e-06, "loss": 0.1036, "step": 40417 }, { "epoch": 29.61025641025641, "grad_norm": 1.09375, "learning_rate": 7.282640381578246e-06, "loss": 0.1412, "step": 40418 }, { "epoch": 29.610989010989012, "grad_norm": 1.3125, "learning_rate": 7.280969857743476e-06, "loss": 0.1147, "step": 40419 }, { "epoch": 29.611721611721613, "grad_norm": 1.5703125, "learning_rate": 7.2792995409852665e-06, "loss": 0.117, "step": 40420 }, { "epoch": 29.612454212454214, "grad_norm": 0.98828125, "learning_rate": 7.277629431311626e-06, "loss": 0.1343, "step": 40421 }, { "epoch": 29.613186813186815, "grad_norm": 1.0625, "learning_rate": 7.27595952873055e-06, "loss": 0.1193, "step": 40422 }, { "epoch": 29.613919413919415, "grad_norm": 0.90625, "learning_rate": 7.2742898332500465e-06, "loss": 0.1201, "step": 40423 }, { "epoch": 29.614652014652016, "grad_norm": 1.2890625, "learning_rate": 7.2726203448781035e-06, "loss": 0.1517, "step": 40424 }, { "epoch": 29.615384615384617, "grad_norm": 1.0390625, "learning_rate": 7.270951063622726e-06, "loss": 0.135, "step": 40425 }, { "epoch": 29.616117216117217, "grad_norm": 0.734375, "learning_rate": 7.269281989491912e-06, "loss": 0.0991, "step": 40426 }, { "epoch": 29.616849816849818, "grad_norm": 0.91796875, "learning_rate": 7.267613122493667e-06, "loss": 0.1018, "step": 40427 }, { "epoch": 29.61758241758242, "grad_norm": 0.75, "learning_rate": 7.265944462635973e-06, "loss": 0.1226, "step": 40428 }, { "epoch": 29.61831501831502, "grad_norm": 0.58984375, "learning_rate": 7.264276009926835e-06, "loss": 0.132, "step": 40429 }, { "epoch": 29.61904761904762, "grad_norm": 0.92578125, "learning_rate": 7.262607764374239e-06, "loss": 0.1201, "step": 40430 }, { "epoch": 29.61978021978022, "grad_norm": 1.1796875, "learning_rate": 7.2609397259861965e-06, "loss": 0.1456, "step": 40431 }, { "epoch": 29.620512820512822, "grad_norm": 0.66796875, "learning_rate": 7.259271894770672e-06, "loss": 0.1167, "step": 40432 }, { "epoch": 29.621245421245423, "grad_norm": 1.25, "learning_rate": 7.2576042707356754e-06, "loss": 0.1032, "step": 40433 }, { "epoch": 29.621978021978023, "grad_norm": 1.1015625, "learning_rate": 7.255936853889203e-06, "loss": 0.1284, "step": 40434 }, { "epoch": 29.622710622710624, "grad_norm": 1.5703125, "learning_rate": 7.254269644239206e-06, "loss": 0.1191, "step": 40435 }, { "epoch": 29.623443223443225, "grad_norm": 0.94921875, "learning_rate": 7.252602641793724e-06, "loss": 0.1048, "step": 40436 }, { "epoch": 29.624175824175826, "grad_norm": 1.1015625, "learning_rate": 7.250935846560715e-06, "loss": 0.1457, "step": 40437 }, { "epoch": 29.624908424908426, "grad_norm": 1.2578125, "learning_rate": 7.249269258548176e-06, "loss": 0.13, "step": 40438 }, { "epoch": 29.625641025641027, "grad_norm": 1.09375, "learning_rate": 7.2476028777640746e-06, "loss": 0.1285, "step": 40439 }, { "epoch": 29.626373626373628, "grad_norm": 1.421875, "learning_rate": 7.245936704216408e-06, "loss": 0.149, "step": 40440 }, { "epoch": 29.62710622710623, "grad_norm": 0.90625, "learning_rate": 7.2442707379131586e-06, "loss": 0.1166, "step": 40441 }, { "epoch": 29.62783882783883, "grad_norm": 1.03125, "learning_rate": 7.242604978862315e-06, "loss": 0.1187, "step": 40442 }, { "epoch": 29.62857142857143, "grad_norm": 0.91796875, "learning_rate": 7.240939427071844e-06, "loss": 0.0968, "step": 40443 }, { "epoch": 29.62930402930403, "grad_norm": 1.5703125, "learning_rate": 7.239274082549731e-06, "loss": 0.1283, "step": 40444 }, { "epoch": 29.63003663003663, "grad_norm": 1.03125, "learning_rate": 7.237608945303971e-06, "loss": 0.1222, "step": 40445 }, { "epoch": 29.630769230769232, "grad_norm": 1.4296875, "learning_rate": 7.235944015342508e-06, "loss": 0.1189, "step": 40446 }, { "epoch": 29.631501831501833, "grad_norm": 1.1015625, "learning_rate": 7.2342792926733575e-06, "loss": 0.1401, "step": 40447 }, { "epoch": 29.632234432234434, "grad_norm": 0.98828125, "learning_rate": 7.232614777304471e-06, "loss": 0.1201, "step": 40448 }, { "epoch": 29.632967032967034, "grad_norm": 1.0, "learning_rate": 7.230950469243837e-06, "loss": 0.0899, "step": 40449 }, { "epoch": 29.633699633699635, "grad_norm": 0.984375, "learning_rate": 7.229286368499416e-06, "loss": 0.1121, "step": 40450 }, { "epoch": 29.634432234432236, "grad_norm": 1.09375, "learning_rate": 7.227622475079188e-06, "loss": 0.1259, "step": 40451 }, { "epoch": 29.635164835164836, "grad_norm": 1.1015625, "learning_rate": 7.2259587889911284e-06, "loss": 0.1374, "step": 40452 }, { "epoch": 29.635897435897437, "grad_norm": 1.5703125, "learning_rate": 7.224295310243215e-06, "loss": 0.1322, "step": 40453 }, { "epoch": 29.636630036630038, "grad_norm": 0.9296875, "learning_rate": 7.2226320388434005e-06, "loss": 0.1178, "step": 40454 }, { "epoch": 29.63736263736264, "grad_norm": 1.03125, "learning_rate": 7.22096897479966e-06, "loss": 0.1111, "step": 40455 }, { "epoch": 29.63809523809524, "grad_norm": 0.81640625, "learning_rate": 7.219306118119967e-06, "loss": 0.1301, "step": 40456 }, { "epoch": 29.63882783882784, "grad_norm": 1.5078125, "learning_rate": 7.217643468812285e-06, "loss": 0.0973, "step": 40457 }, { "epoch": 29.63956043956044, "grad_norm": 1.078125, "learning_rate": 7.2159810268845925e-06, "loss": 0.1099, "step": 40458 }, { "epoch": 29.64029304029304, "grad_norm": 1.15625, "learning_rate": 7.214318792344836e-06, "loss": 0.1141, "step": 40459 }, { "epoch": 29.641025641025642, "grad_norm": 0.84375, "learning_rate": 7.212656765200998e-06, "loss": 0.1355, "step": 40460 }, { "epoch": 29.641758241758243, "grad_norm": 0.7734375, "learning_rate": 7.210994945461005e-06, "loss": 0.1272, "step": 40461 }, { "epoch": 29.642490842490844, "grad_norm": 0.8984375, "learning_rate": 7.209333333132871e-06, "loss": 0.091, "step": 40462 }, { "epoch": 29.643223443223444, "grad_norm": 1.3203125, "learning_rate": 7.207671928224522e-06, "loss": 0.1385, "step": 40463 }, { "epoch": 29.643956043956045, "grad_norm": 1.0859375, "learning_rate": 7.206010730743941e-06, "loss": 0.1002, "step": 40464 }, { "epoch": 29.644688644688646, "grad_norm": 1.125, "learning_rate": 7.20434974069906e-06, "loss": 0.1029, "step": 40465 }, { "epoch": 29.645421245421247, "grad_norm": 0.66015625, "learning_rate": 7.202688958097857e-06, "loss": 0.1073, "step": 40466 }, { "epoch": 29.646153846153847, "grad_norm": 1.8671875, "learning_rate": 7.2010283829482856e-06, "loss": 0.12, "step": 40467 }, { "epoch": 29.646886446886448, "grad_norm": 0.765625, "learning_rate": 7.1993680152583075e-06, "loss": 0.1104, "step": 40468 }, { "epoch": 29.64761904761905, "grad_norm": 0.81640625, "learning_rate": 7.197707855035865e-06, "loss": 0.1133, "step": 40469 }, { "epoch": 29.64835164835165, "grad_norm": 1.0078125, "learning_rate": 7.196047902288915e-06, "loss": 0.1371, "step": 40470 }, { "epoch": 29.64908424908425, "grad_norm": 1.0703125, "learning_rate": 7.1943881570254265e-06, "loss": 0.1042, "step": 40471 }, { "epoch": 29.64981684981685, "grad_norm": 1.0390625, "learning_rate": 7.192728619253323e-06, "loss": 0.1371, "step": 40472 }, { "epoch": 29.650549450549452, "grad_norm": 1.0703125, "learning_rate": 7.191069288980592e-06, "loss": 0.1489, "step": 40473 }, { "epoch": 29.651282051282053, "grad_norm": 1.1875, "learning_rate": 7.189410166215157e-06, "loss": 0.1108, "step": 40474 }, { "epoch": 29.652014652014653, "grad_norm": 1.1015625, "learning_rate": 7.187751250964986e-06, "loss": 0.1534, "step": 40475 }, { "epoch": 29.652747252747254, "grad_norm": 0.94921875, "learning_rate": 7.186092543238005e-06, "loss": 0.1047, "step": 40476 }, { "epoch": 29.653479853479855, "grad_norm": 0.80078125, "learning_rate": 7.184434043042175e-06, "loss": 0.101, "step": 40477 }, { "epoch": 29.654212454212455, "grad_norm": 1.078125, "learning_rate": 7.1827757503854385e-06, "loss": 0.1458, "step": 40478 }, { "epoch": 29.654945054945056, "grad_norm": 1.2734375, "learning_rate": 7.181117665275759e-06, "loss": 0.1053, "step": 40479 }, { "epoch": 29.655677655677657, "grad_norm": 1.390625, "learning_rate": 7.179459787721051e-06, "loss": 0.1165, "step": 40480 }, { "epoch": 29.656410256410258, "grad_norm": 1.4921875, "learning_rate": 7.177802117729273e-06, "loss": 0.1409, "step": 40481 }, { "epoch": 29.65714285714286, "grad_norm": 1.09375, "learning_rate": 7.176144655308367e-06, "loss": 0.1077, "step": 40482 }, { "epoch": 29.65787545787546, "grad_norm": 1.78125, "learning_rate": 7.174487400466276e-06, "loss": 0.1147, "step": 40483 }, { "epoch": 29.65860805860806, "grad_norm": 0.73828125, "learning_rate": 7.172830353210946e-06, "loss": 0.0956, "step": 40484 }, { "epoch": 29.65934065934066, "grad_norm": 0.97265625, "learning_rate": 7.171173513550304e-06, "loss": 0.1273, "step": 40485 }, { "epoch": 29.66007326007326, "grad_norm": 1.234375, "learning_rate": 7.169516881492301e-06, "loss": 0.1204, "step": 40486 }, { "epoch": 29.660805860805862, "grad_norm": 0.8671875, "learning_rate": 7.167860457044845e-06, "loss": 0.1059, "step": 40487 }, { "epoch": 29.661538461538463, "grad_norm": 1.046875, "learning_rate": 7.1662042402159166e-06, "loss": 0.1325, "step": 40488 }, { "epoch": 29.662271062271063, "grad_norm": 0.9453125, "learning_rate": 7.164548231013417e-06, "loss": 0.1329, "step": 40489 }, { "epoch": 29.663003663003664, "grad_norm": 0.82421875, "learning_rate": 7.162892429445305e-06, "loss": 0.1221, "step": 40490 }, { "epoch": 29.663736263736265, "grad_norm": 0.92578125, "learning_rate": 7.161236835519493e-06, "loss": 0.1035, "step": 40491 }, { "epoch": 29.664468864468866, "grad_norm": 1.015625, "learning_rate": 7.159581449243918e-06, "loss": 0.1174, "step": 40492 }, { "epoch": 29.665201465201466, "grad_norm": 0.84375, "learning_rate": 7.157926270626521e-06, "loss": 0.1061, "step": 40493 }, { "epoch": 29.665934065934067, "grad_norm": 1.03125, "learning_rate": 7.156271299675232e-06, "loss": 0.1121, "step": 40494 }, { "epoch": 29.666666666666668, "grad_norm": 1.4140625, "learning_rate": 7.154616536397969e-06, "loss": 0.1341, "step": 40495 }, { "epoch": 29.66739926739927, "grad_norm": 0.765625, "learning_rate": 7.152961980802669e-06, "loss": 0.105, "step": 40496 }, { "epoch": 29.66813186813187, "grad_norm": 0.734375, "learning_rate": 7.151307632897268e-06, "loss": 0.1213, "step": 40497 }, { "epoch": 29.66886446886447, "grad_norm": 1.015625, "learning_rate": 7.149653492689662e-06, "loss": 0.1113, "step": 40498 }, { "epoch": 29.66959706959707, "grad_norm": 1.203125, "learning_rate": 7.1479995601878136e-06, "loss": 0.1327, "step": 40499 }, { "epoch": 29.67032967032967, "grad_norm": 0.79296875, "learning_rate": 7.1463458353996285e-06, "loss": 0.1291, "step": 40500 }, { "epoch": 29.671062271062272, "grad_norm": 1.140625, "learning_rate": 7.144692318333039e-06, "loss": 0.1412, "step": 40501 }, { "epoch": 29.671794871794873, "grad_norm": 1.8828125, "learning_rate": 7.14303900899595e-06, "loss": 0.1213, "step": 40502 }, { "epoch": 29.672527472527474, "grad_norm": 0.91015625, "learning_rate": 7.141385907396294e-06, "loss": 0.1165, "step": 40503 }, { "epoch": 29.673260073260074, "grad_norm": 1.046875, "learning_rate": 7.13973301354199e-06, "loss": 0.1193, "step": 40504 }, { "epoch": 29.673992673992675, "grad_norm": 1.015625, "learning_rate": 7.1380803274409745e-06, "loss": 0.1248, "step": 40505 }, { "epoch": 29.674725274725276, "grad_norm": 1.015625, "learning_rate": 7.1364278491011346e-06, "loss": 0.1225, "step": 40506 }, { "epoch": 29.675457875457877, "grad_norm": 0.94921875, "learning_rate": 7.134775578530408e-06, "loss": 0.1295, "step": 40507 }, { "epoch": 29.676190476190477, "grad_norm": 0.8984375, "learning_rate": 7.133123515736706e-06, "loss": 0.097, "step": 40508 }, { "epoch": 29.676923076923078, "grad_norm": 1.046875, "learning_rate": 7.13147166072795e-06, "loss": 0.1344, "step": 40509 }, { "epoch": 29.67765567765568, "grad_norm": 0.75, "learning_rate": 7.129820013512056e-06, "loss": 0.1133, "step": 40510 }, { "epoch": 29.67838827838828, "grad_norm": 1.09375, "learning_rate": 7.128168574096925e-06, "loss": 0.1387, "step": 40511 }, { "epoch": 29.67912087912088, "grad_norm": 1.109375, "learning_rate": 7.126517342490484e-06, "loss": 0.1139, "step": 40512 }, { "epoch": 29.67985347985348, "grad_norm": 1.09375, "learning_rate": 7.124866318700614e-06, "loss": 0.1286, "step": 40513 }, { "epoch": 29.68058608058608, "grad_norm": 1.1171875, "learning_rate": 7.123215502735276e-06, "loss": 0.101, "step": 40514 }, { "epoch": 29.681318681318682, "grad_norm": 1.0703125, "learning_rate": 7.121564894602335e-06, "loss": 0.134, "step": 40515 }, { "epoch": 29.682051282051283, "grad_norm": 1.171875, "learning_rate": 7.119914494309731e-06, "loss": 0.1051, "step": 40516 }, { "epoch": 29.682783882783884, "grad_norm": 0.921875, "learning_rate": 7.118264301865348e-06, "loss": 0.1331, "step": 40517 }, { "epoch": 29.683516483516485, "grad_norm": 0.859375, "learning_rate": 7.116614317277098e-06, "loss": 0.1259, "step": 40518 }, { "epoch": 29.684249084249085, "grad_norm": 0.56640625, "learning_rate": 7.114964540552892e-06, "loss": 0.1055, "step": 40519 }, { "epoch": 29.684981684981686, "grad_norm": 1.125, "learning_rate": 7.113314971700648e-06, "loss": 0.1377, "step": 40520 }, { "epoch": 29.685714285714287, "grad_norm": 1.2578125, "learning_rate": 7.111665610728237e-06, "loss": 0.1195, "step": 40521 }, { "epoch": 29.686446886446888, "grad_norm": 1.1328125, "learning_rate": 7.110016457643587e-06, "loss": 0.1313, "step": 40522 }, { "epoch": 29.68717948717949, "grad_norm": 0.859375, "learning_rate": 7.108367512454596e-06, "loss": 0.1195, "step": 40523 }, { "epoch": 29.68791208791209, "grad_norm": 1.171875, "learning_rate": 7.106718775169145e-06, "loss": 0.1241, "step": 40524 }, { "epoch": 29.68864468864469, "grad_norm": 1.234375, "learning_rate": 7.105070245795169e-06, "loss": 0.1167, "step": 40525 }, { "epoch": 29.68937728937729, "grad_norm": 0.84375, "learning_rate": 7.1034219243405335e-06, "loss": 0.1199, "step": 40526 }, { "epoch": 29.69010989010989, "grad_norm": 1.1796875, "learning_rate": 7.101773810813163e-06, "loss": 0.1284, "step": 40527 }, { "epoch": 29.690842490842492, "grad_norm": 1.609375, "learning_rate": 7.100125905220931e-06, "loss": 0.115, "step": 40528 }, { "epoch": 29.691575091575093, "grad_norm": 0.81640625, "learning_rate": 7.0984782075717414e-06, "loss": 0.1155, "step": 40529 }, { "epoch": 29.692307692307693, "grad_norm": 1.1171875, "learning_rate": 7.096830717873492e-06, "loss": 0.1033, "step": 40530 }, { "epoch": 29.693040293040294, "grad_norm": 0.92578125, "learning_rate": 7.095183436134084e-06, "loss": 0.1296, "step": 40531 }, { "epoch": 29.693772893772895, "grad_norm": 0.875, "learning_rate": 7.0935363623613946e-06, "loss": 0.1274, "step": 40532 }, { "epoch": 29.694505494505496, "grad_norm": 0.6796875, "learning_rate": 7.091889496563318e-06, "loss": 0.1172, "step": 40533 }, { "epoch": 29.695238095238096, "grad_norm": 1.0546875, "learning_rate": 7.090242838747749e-06, "loss": 0.1224, "step": 40534 }, { "epoch": 29.695970695970697, "grad_norm": 0.859375, "learning_rate": 7.088596388922576e-06, "loss": 0.0962, "step": 40535 }, { "epoch": 29.696703296703298, "grad_norm": 1.078125, "learning_rate": 7.0869501470957e-06, "loss": 0.1346, "step": 40536 }, { "epoch": 29.6974358974359, "grad_norm": 1.4375, "learning_rate": 7.085304113274987e-06, "loss": 0.1304, "step": 40537 }, { "epoch": 29.6981684981685, "grad_norm": 1.21875, "learning_rate": 7.083658287468344e-06, "loss": 0.1212, "step": 40538 }, { "epoch": 29.6989010989011, "grad_norm": 1.4609375, "learning_rate": 7.082012669683625e-06, "loss": 0.1114, "step": 40539 }, { "epoch": 29.6996336996337, "grad_norm": 1.65625, "learning_rate": 7.080367259928763e-06, "loss": 0.1361, "step": 40540 }, { "epoch": 29.7003663003663, "grad_norm": 1.015625, "learning_rate": 7.0787220582116005e-06, "loss": 0.0977, "step": 40541 }, { "epoch": 29.701098901098902, "grad_norm": 1.0390625, "learning_rate": 7.077077064540044e-06, "loss": 0.122, "step": 40542 }, { "epoch": 29.701831501831503, "grad_norm": 1.0234375, "learning_rate": 7.075432278921959e-06, "loss": 0.1305, "step": 40543 }, { "epoch": 29.702564102564104, "grad_norm": 0.9296875, "learning_rate": 7.07378770136523e-06, "loss": 0.1145, "step": 40544 }, { "epoch": 29.703296703296704, "grad_norm": 0.890625, "learning_rate": 7.072143331877745e-06, "loss": 0.1201, "step": 40545 }, { "epoch": 29.704029304029305, "grad_norm": 0.95703125, "learning_rate": 7.070499170467385e-06, "loss": 0.1322, "step": 40546 }, { "epoch": 29.704761904761906, "grad_norm": 0.91796875, "learning_rate": 7.068855217142011e-06, "loss": 0.1196, "step": 40547 }, { "epoch": 29.705494505494507, "grad_norm": 1.3984375, "learning_rate": 7.067211471909507e-06, "loss": 0.1386, "step": 40548 }, { "epoch": 29.706227106227107, "grad_norm": 0.71875, "learning_rate": 7.065567934777764e-06, "loss": 0.1235, "step": 40549 }, { "epoch": 29.706959706959708, "grad_norm": 1.109375, "learning_rate": 7.063924605754618e-06, "loss": 0.1134, "step": 40550 }, { "epoch": 29.70769230769231, "grad_norm": 0.984375, "learning_rate": 7.062281484847995e-06, "loss": 0.1126, "step": 40551 }, { "epoch": 29.70842490842491, "grad_norm": 0.9609375, "learning_rate": 7.060638572065724e-06, "loss": 0.0946, "step": 40552 }, { "epoch": 29.70915750915751, "grad_norm": 1.140625, "learning_rate": 7.0589958674157086e-06, "loss": 0.1385, "step": 40553 }, { "epoch": 29.70989010989011, "grad_norm": 1.09375, "learning_rate": 7.057353370905798e-06, "loss": 0.1276, "step": 40554 }, { "epoch": 29.71062271062271, "grad_norm": 1.734375, "learning_rate": 7.055711082543861e-06, "loss": 0.1113, "step": 40555 }, { "epoch": 29.711355311355312, "grad_norm": 1.0859375, "learning_rate": 7.054069002337782e-06, "loss": 0.1092, "step": 40556 }, { "epoch": 29.712087912087913, "grad_norm": 1.3828125, "learning_rate": 7.052427130295422e-06, "loss": 0.1508, "step": 40557 }, { "epoch": 29.712820512820514, "grad_norm": 1.3828125, "learning_rate": 7.050785466424645e-06, "loss": 0.1513, "step": 40558 }, { "epoch": 29.713553113553115, "grad_norm": 1.2109375, "learning_rate": 7.049144010733313e-06, "loss": 0.116, "step": 40559 }, { "epoch": 29.714285714285715, "grad_norm": 1.3671875, "learning_rate": 7.047502763229299e-06, "loss": 0.1035, "step": 40560 }, { "epoch": 29.715018315018316, "grad_norm": 1.1953125, "learning_rate": 7.0458617239204605e-06, "loss": 0.1297, "step": 40561 }, { "epoch": 29.715750915750917, "grad_norm": 1.078125, "learning_rate": 7.0442208928146765e-06, "loss": 0.1094, "step": 40562 }, { "epoch": 29.716483516483517, "grad_norm": 1.125, "learning_rate": 7.042580269919787e-06, "loss": 0.113, "step": 40563 }, { "epoch": 29.717216117216118, "grad_norm": 1.078125, "learning_rate": 7.040939855243668e-06, "loss": 0.1335, "step": 40564 }, { "epoch": 29.71794871794872, "grad_norm": 1.34375, "learning_rate": 7.039299648794149e-06, "loss": 0.1276, "step": 40565 }, { "epoch": 29.71868131868132, "grad_norm": 0.7109375, "learning_rate": 7.037659650579139e-06, "loss": 0.0938, "step": 40566 }, { "epoch": 29.71941391941392, "grad_norm": 0.66796875, "learning_rate": 7.036019860606459e-06, "loss": 0.0934, "step": 40567 }, { "epoch": 29.72014652014652, "grad_norm": 1.4375, "learning_rate": 7.034380278883988e-06, "loss": 0.1333, "step": 40568 }, { "epoch": 29.720879120879122, "grad_norm": 0.90625, "learning_rate": 7.0327409054195575e-06, "loss": 0.1179, "step": 40569 }, { "epoch": 29.721611721611723, "grad_norm": 1.1796875, "learning_rate": 7.03110174022103e-06, "loss": 0.1509, "step": 40570 }, { "epoch": 29.722344322344323, "grad_norm": 1.390625, "learning_rate": 7.02946278329627e-06, "loss": 0.1152, "step": 40571 }, { "epoch": 29.723076923076924, "grad_norm": 1.4453125, "learning_rate": 7.027824034653119e-06, "loss": 0.1302, "step": 40572 }, { "epoch": 29.723809523809525, "grad_norm": 1.5625, "learning_rate": 7.026185494299449e-06, "loss": 0.1318, "step": 40573 }, { "epoch": 29.724542124542126, "grad_norm": 1.21875, "learning_rate": 7.024547162243083e-06, "loss": 0.0906, "step": 40574 }, { "epoch": 29.725274725274726, "grad_norm": 0.79296875, "learning_rate": 7.0229090384919e-06, "loss": 0.1242, "step": 40575 }, { "epoch": 29.726007326007327, "grad_norm": 1.3125, "learning_rate": 7.021271123053705e-06, "loss": 0.1447, "step": 40576 }, { "epoch": 29.726739926739928, "grad_norm": 1.03125, "learning_rate": 7.019633415936397e-06, "loss": 0.0809, "step": 40577 }, { "epoch": 29.72747252747253, "grad_norm": 2.09375, "learning_rate": 7.017995917147786e-06, "loss": 0.1466, "step": 40578 }, { "epoch": 29.72820512820513, "grad_norm": 0.89453125, "learning_rate": 7.01635862669574e-06, "loss": 0.1137, "step": 40579 }, { "epoch": 29.72893772893773, "grad_norm": 0.94921875, "learning_rate": 7.0147215445880875e-06, "loss": 0.1304, "step": 40580 }, { "epoch": 29.72967032967033, "grad_norm": 0.69140625, "learning_rate": 7.013084670832679e-06, "loss": 0.1049, "step": 40581 }, { "epoch": 29.73040293040293, "grad_norm": 1.4296875, "learning_rate": 7.011448005437357e-06, "loss": 0.1503, "step": 40582 }, { "epoch": 29.731135531135532, "grad_norm": 0.87109375, "learning_rate": 7.009811548409977e-06, "loss": 0.1184, "step": 40583 }, { "epoch": 29.731868131868133, "grad_norm": 0.9765625, "learning_rate": 7.00817529975835e-06, "loss": 0.1235, "step": 40584 }, { "epoch": 29.732600732600734, "grad_norm": 0.78515625, "learning_rate": 7.006539259490338e-06, "loss": 0.1151, "step": 40585 }, { "epoch": 29.733333333333334, "grad_norm": 0.90625, "learning_rate": 7.004903427613772e-06, "loss": 0.1392, "step": 40586 }, { "epoch": 29.734065934065935, "grad_norm": 0.94140625, "learning_rate": 7.003267804136497e-06, "loss": 0.1279, "step": 40587 }, { "epoch": 29.734798534798536, "grad_norm": 1.0859375, "learning_rate": 7.001632389066349e-06, "loss": 0.1112, "step": 40588 }, { "epoch": 29.735531135531136, "grad_norm": 1.390625, "learning_rate": 6.999997182411152e-06, "loss": 0.1235, "step": 40589 }, { "epoch": 29.736263736263737, "grad_norm": 0.734375, "learning_rate": 6.998362184178761e-06, "loss": 0.1067, "step": 40590 }, { "epoch": 29.736996336996338, "grad_norm": 1.0390625, "learning_rate": 6.9967273943769715e-06, "loss": 0.0966, "step": 40591 }, { "epoch": 29.73772893772894, "grad_norm": 1.2578125, "learning_rate": 6.995092813013668e-06, "loss": 0.1649, "step": 40592 }, { "epoch": 29.73846153846154, "grad_norm": 1.265625, "learning_rate": 6.993458440096645e-06, "loss": 0.1232, "step": 40593 }, { "epoch": 29.73919413919414, "grad_norm": 0.8515625, "learning_rate": 6.99182427563375e-06, "loss": 0.1281, "step": 40594 }, { "epoch": 29.73992673992674, "grad_norm": 1.59375, "learning_rate": 6.9901903196328045e-06, "loss": 0.1323, "step": 40595 }, { "epoch": 29.74065934065934, "grad_norm": 0.71484375, "learning_rate": 6.988556572101637e-06, "loss": 0.1254, "step": 40596 }, { "epoch": 29.741391941391942, "grad_norm": 0.7734375, "learning_rate": 6.98692303304808e-06, "loss": 0.1236, "step": 40597 }, { "epoch": 29.742124542124543, "grad_norm": 1.3515625, "learning_rate": 6.985289702479961e-06, "loss": 0.121, "step": 40598 }, { "epoch": 29.742857142857144, "grad_norm": 1.2421875, "learning_rate": 6.983656580405108e-06, "loss": 0.1171, "step": 40599 }, { "epoch": 29.743589743589745, "grad_norm": 0.7734375, "learning_rate": 6.982023666831339e-06, "loss": 0.1261, "step": 40600 }, { "epoch": 29.744322344322345, "grad_norm": 0.8203125, "learning_rate": 6.9803909617664876e-06, "loss": 0.1032, "step": 40601 }, { "epoch": 29.745054945054946, "grad_norm": 0.94921875, "learning_rate": 6.978758465218349e-06, "loss": 0.1244, "step": 40602 }, { "epoch": 29.745787545787547, "grad_norm": 1.4375, "learning_rate": 6.9771261771947855e-06, "loss": 0.1493, "step": 40603 }, { "epoch": 29.746520146520147, "grad_norm": 0.62109375, "learning_rate": 6.975494097703586e-06, "loss": 0.0993, "step": 40604 }, { "epoch": 29.747252747252748, "grad_norm": 1.09375, "learning_rate": 6.9738622267525975e-06, "loss": 0.1135, "step": 40605 }, { "epoch": 29.74798534798535, "grad_norm": 1.015625, "learning_rate": 6.972230564349613e-06, "loss": 0.1014, "step": 40606 }, { "epoch": 29.74871794871795, "grad_norm": 0.90625, "learning_rate": 6.9705991105024584e-06, "loss": 0.1056, "step": 40607 }, { "epoch": 29.74945054945055, "grad_norm": 1.1171875, "learning_rate": 6.968967865218953e-06, "loss": 0.1372, "step": 40608 }, { "epoch": 29.75018315018315, "grad_norm": 0.98828125, "learning_rate": 6.967336828506925e-06, "loss": 0.1254, "step": 40609 }, { "epoch": 29.750915750915752, "grad_norm": 1.15625, "learning_rate": 6.965706000374168e-06, "loss": 0.1299, "step": 40610 }, { "epoch": 29.751648351648353, "grad_norm": 1.03125, "learning_rate": 6.964075380828502e-06, "loss": 0.1235, "step": 40611 }, { "epoch": 29.752380952380953, "grad_norm": 0.734375, "learning_rate": 6.962444969877743e-06, "loss": 0.1075, "step": 40612 }, { "epoch": 29.753113553113554, "grad_norm": 0.97265625, "learning_rate": 6.960814767529704e-06, "loss": 0.1333, "step": 40613 }, { "epoch": 29.753846153846155, "grad_norm": 1.4453125, "learning_rate": 6.959184773792202e-06, "loss": 0.1386, "step": 40614 }, { "epoch": 29.754578754578755, "grad_norm": 1.03125, "learning_rate": 6.957554988673032e-06, "loss": 0.1094, "step": 40615 }, { "epoch": 29.755311355311356, "grad_norm": 1.234375, "learning_rate": 6.955925412180018e-06, "loss": 0.128, "step": 40616 }, { "epoch": 29.756043956043957, "grad_norm": 1.3046875, "learning_rate": 6.9542960443209434e-06, "loss": 0.1178, "step": 40617 }, { "epoch": 29.756776556776558, "grad_norm": 0.765625, "learning_rate": 6.9526668851036484e-06, "loss": 0.098, "step": 40618 }, { "epoch": 29.75750915750916, "grad_norm": 0.89453125, "learning_rate": 6.951037934535912e-06, "loss": 0.1314, "step": 40619 }, { "epoch": 29.75824175824176, "grad_norm": 0.89453125, "learning_rate": 6.949409192625563e-06, "loss": 0.1108, "step": 40620 }, { "epoch": 29.75897435897436, "grad_norm": 1.0546875, "learning_rate": 6.947780659380379e-06, "loss": 0.1273, "step": 40621 }, { "epoch": 29.75970695970696, "grad_norm": 0.87109375, "learning_rate": 6.946152334808174e-06, "loss": 0.1234, "step": 40622 }, { "epoch": 29.76043956043956, "grad_norm": 1.6484375, "learning_rate": 6.944524218916752e-06, "loss": 0.109, "step": 40623 }, { "epoch": 29.761172161172162, "grad_norm": 0.80859375, "learning_rate": 6.9428963117139115e-06, "loss": 0.1151, "step": 40624 }, { "epoch": 29.761904761904763, "grad_norm": 0.68359375, "learning_rate": 6.941268613207463e-06, "loss": 0.1078, "step": 40625 }, { "epoch": 29.762637362637363, "grad_norm": 1.6171875, "learning_rate": 6.939641123405186e-06, "loss": 0.1236, "step": 40626 }, { "epoch": 29.763369963369964, "grad_norm": 1.21875, "learning_rate": 6.938013842314901e-06, "loss": 0.1149, "step": 40627 }, { "epoch": 29.764102564102565, "grad_norm": 0.75, "learning_rate": 6.93638676994437e-06, "loss": 0.0987, "step": 40628 }, { "epoch": 29.764835164835166, "grad_norm": 0.7578125, "learning_rate": 6.934759906301433e-06, "loss": 0.0927, "step": 40629 }, { "epoch": 29.765567765567766, "grad_norm": 0.90625, "learning_rate": 6.93313325139385e-06, "loss": 0.1251, "step": 40630 }, { "epoch": 29.766300366300367, "grad_norm": 1.03125, "learning_rate": 6.931506805229441e-06, "loss": 0.1292, "step": 40631 }, { "epoch": 29.767032967032968, "grad_norm": 1.0546875, "learning_rate": 6.929880567815975e-06, "loss": 0.1161, "step": 40632 }, { "epoch": 29.76776556776557, "grad_norm": 1.3984375, "learning_rate": 6.928254539161249e-06, "loss": 0.1045, "step": 40633 }, { "epoch": 29.76849816849817, "grad_norm": 0.6875, "learning_rate": 6.926628719273068e-06, "loss": 0.1114, "step": 40634 }, { "epoch": 29.76923076923077, "grad_norm": 0.8515625, "learning_rate": 6.9250031081592155e-06, "loss": 0.1048, "step": 40635 }, { "epoch": 29.76996336996337, "grad_norm": 1.1484375, "learning_rate": 6.9233777058274694e-06, "loss": 0.11, "step": 40636 }, { "epoch": 29.77069597069597, "grad_norm": 0.77734375, "learning_rate": 6.9217525122856314e-06, "loss": 0.1118, "step": 40637 }, { "epoch": 29.771428571428572, "grad_norm": 1.1640625, "learning_rate": 6.920127527541473e-06, "loss": 0.1279, "step": 40638 }, { "epoch": 29.772161172161173, "grad_norm": 0.94921875, "learning_rate": 6.9185027516028e-06, "loss": 0.1313, "step": 40639 }, { "epoch": 29.772893772893774, "grad_norm": 1.28125, "learning_rate": 6.916878184477392e-06, "loss": 0.1362, "step": 40640 }, { "epoch": 29.773626373626374, "grad_norm": 1.1875, "learning_rate": 6.91525382617302e-06, "loss": 0.1349, "step": 40641 }, { "epoch": 29.774358974358975, "grad_norm": 1.0625, "learning_rate": 6.913629676697483e-06, "loss": 0.1041, "step": 40642 }, { "epoch": 29.775091575091576, "grad_norm": 0.68359375, "learning_rate": 6.912005736058536e-06, "loss": 0.1278, "step": 40643 }, { "epoch": 29.775824175824177, "grad_norm": 0.78125, "learning_rate": 6.910382004263998e-06, "loss": 0.1043, "step": 40644 }, { "epoch": 29.776556776556777, "grad_norm": 1.125, "learning_rate": 6.908758481321617e-06, "loss": 0.1351, "step": 40645 }, { "epoch": 29.777289377289378, "grad_norm": 1.3515625, "learning_rate": 6.907135167239198e-06, "loss": 0.1038, "step": 40646 }, { "epoch": 29.77802197802198, "grad_norm": 1.0625, "learning_rate": 6.9055120620244895e-06, "loss": 0.1203, "step": 40647 }, { "epoch": 29.77875457875458, "grad_norm": 1.078125, "learning_rate": 6.903889165685292e-06, "loss": 0.1265, "step": 40648 }, { "epoch": 29.77948717948718, "grad_norm": 1.09375, "learning_rate": 6.902266478229366e-06, "loss": 0.1118, "step": 40649 }, { "epoch": 29.78021978021978, "grad_norm": 0.6640625, "learning_rate": 6.900643999664497e-06, "loss": 0.1292, "step": 40650 }, { "epoch": 29.78095238095238, "grad_norm": 0.76953125, "learning_rate": 6.899021729998466e-06, "loss": 0.1068, "step": 40651 }, { "epoch": 29.781684981684982, "grad_norm": 0.92578125, "learning_rate": 6.8973996692390265e-06, "loss": 0.1241, "step": 40652 }, { "epoch": 29.782417582417583, "grad_norm": 0.88671875, "learning_rate": 6.8957778173939665e-06, "loss": 0.1104, "step": 40653 }, { "epoch": 29.783150183150184, "grad_norm": 1.4375, "learning_rate": 6.894156174471024e-06, "loss": 0.1088, "step": 40654 }, { "epoch": 29.783882783882785, "grad_norm": 0.8984375, "learning_rate": 6.892534740478024e-06, "loss": 0.1402, "step": 40655 }, { "epoch": 29.784615384615385, "grad_norm": 1.078125, "learning_rate": 6.8909135154226885e-06, "loss": 0.1427, "step": 40656 }, { "epoch": 29.785347985347986, "grad_norm": 0.77734375, "learning_rate": 6.889292499312815e-06, "loss": 0.1004, "step": 40657 }, { "epoch": 29.786080586080587, "grad_norm": 1.09375, "learning_rate": 6.8876716921561475e-06, "loss": 0.1282, "step": 40658 }, { "epoch": 29.786813186813188, "grad_norm": 1.359375, "learning_rate": 6.886051093960462e-06, "loss": 0.1362, "step": 40659 }, { "epoch": 29.78754578754579, "grad_norm": 0.5625, "learning_rate": 6.884430704733523e-06, "loss": 0.1202, "step": 40660 }, { "epoch": 29.78827838827839, "grad_norm": 0.80859375, "learning_rate": 6.882810524483104e-06, "loss": 0.101, "step": 40661 }, { "epoch": 29.78901098901099, "grad_norm": 1.90625, "learning_rate": 6.88119055321695e-06, "loss": 0.1195, "step": 40662 }, { "epoch": 29.78974358974359, "grad_norm": 0.96875, "learning_rate": 6.8795707909428284e-06, "loss": 0.1162, "step": 40663 }, { "epoch": 29.79047619047619, "grad_norm": 1.7421875, "learning_rate": 6.877951237668505e-06, "loss": 0.1318, "step": 40664 }, { "epoch": 29.791208791208792, "grad_norm": 1.015625, "learning_rate": 6.8763318934017385e-06, "loss": 0.1325, "step": 40665 }, { "epoch": 29.791941391941393, "grad_norm": 1.1171875, "learning_rate": 6.874712758150295e-06, "loss": 0.1161, "step": 40666 }, { "epoch": 29.792673992673993, "grad_norm": 1.03125, "learning_rate": 6.87309383192191e-06, "loss": 0.1355, "step": 40667 }, { "epoch": 29.793406593406594, "grad_norm": 1.171875, "learning_rate": 6.871475114724369e-06, "loss": 0.1299, "step": 40668 }, { "epoch": 29.794139194139195, "grad_norm": 0.8671875, "learning_rate": 6.8698566065653915e-06, "loss": 0.0989, "step": 40669 }, { "epoch": 29.794871794871796, "grad_norm": 0.87890625, "learning_rate": 6.868238307452776e-06, "loss": 0.1193, "step": 40670 }, { "epoch": 29.795604395604396, "grad_norm": 1.21875, "learning_rate": 6.866620217394243e-06, "loss": 0.1432, "step": 40671 }, { "epoch": 29.796336996336997, "grad_norm": 0.79296875, "learning_rate": 6.865002336397564e-06, "loss": 0.112, "step": 40672 }, { "epoch": 29.797069597069598, "grad_norm": 1.0, "learning_rate": 6.863384664470475e-06, "loss": 0.1326, "step": 40673 }, { "epoch": 29.7978021978022, "grad_norm": 0.90625, "learning_rate": 6.861767201620734e-06, "loss": 0.153, "step": 40674 }, { "epoch": 29.7985347985348, "grad_norm": 1.0234375, "learning_rate": 6.860149947856093e-06, "loss": 0.1188, "step": 40675 }, { "epoch": 29.7992673992674, "grad_norm": 0.8671875, "learning_rate": 6.858532903184299e-06, "loss": 0.1115, "step": 40676 }, { "epoch": 29.8, "grad_norm": 1.015625, "learning_rate": 6.856916067613116e-06, "loss": 0.1243, "step": 40677 }, { "epoch": 29.8007326007326, "grad_norm": 1.4296875, "learning_rate": 6.855299441150258e-06, "loss": 0.1388, "step": 40678 }, { "epoch": 29.801465201465202, "grad_norm": 1.28125, "learning_rate": 6.853683023803499e-06, "loss": 0.1102, "step": 40679 }, { "epoch": 29.802197802197803, "grad_norm": 1.4765625, "learning_rate": 6.852066815580552e-06, "loss": 0.1253, "step": 40680 }, { "epoch": 29.802930402930404, "grad_norm": 1.140625, "learning_rate": 6.850450816489209e-06, "loss": 0.1212, "step": 40681 }, { "epoch": 29.803663003663004, "grad_norm": 1.0, "learning_rate": 6.848835026537171e-06, "loss": 0.1181, "step": 40682 }, { "epoch": 29.804395604395605, "grad_norm": 1.0390625, "learning_rate": 6.8472194457321986e-06, "loss": 0.1252, "step": 40683 }, { "epoch": 29.805128205128206, "grad_norm": 1.4453125, "learning_rate": 6.845604074082024e-06, "loss": 0.1497, "step": 40684 }, { "epoch": 29.805860805860807, "grad_norm": 0.66796875, "learning_rate": 6.843988911594387e-06, "loss": 0.1081, "step": 40685 }, { "epoch": 29.806593406593407, "grad_norm": 0.828125, "learning_rate": 6.842373958277034e-06, "loss": 0.1169, "step": 40686 }, { "epoch": 29.807326007326008, "grad_norm": 0.609375, "learning_rate": 6.840759214137698e-06, "loss": 0.1006, "step": 40687 }, { "epoch": 29.80805860805861, "grad_norm": 0.80078125, "learning_rate": 6.8391446791841274e-06, "loss": 0.1025, "step": 40688 }, { "epoch": 29.80879120879121, "grad_norm": 1.3671875, "learning_rate": 6.837530353424042e-06, "loss": 0.1119, "step": 40689 }, { "epoch": 29.80952380952381, "grad_norm": 1.171875, "learning_rate": 6.8359162368651765e-06, "loss": 0.1291, "step": 40690 }, { "epoch": 29.81025641025641, "grad_norm": 0.7890625, "learning_rate": 6.834302329515272e-06, "loss": 0.1268, "step": 40691 }, { "epoch": 29.81098901098901, "grad_norm": 0.87890625, "learning_rate": 6.832688631382066e-06, "loss": 0.1089, "step": 40692 }, { "epoch": 29.811721611721612, "grad_norm": 0.81640625, "learning_rate": 6.83107514247328e-06, "loss": 0.1132, "step": 40693 }, { "epoch": 29.812454212454213, "grad_norm": 1.28125, "learning_rate": 6.829461862796654e-06, "loss": 0.1322, "step": 40694 }, { "epoch": 29.813186813186814, "grad_norm": 1.40625, "learning_rate": 6.8278487923598925e-06, "loss": 0.1276, "step": 40695 }, { "epoch": 29.813919413919415, "grad_norm": 0.78125, "learning_rate": 6.826235931170769e-06, "loss": 0.1159, "step": 40696 }, { "epoch": 29.814652014652015, "grad_norm": 1.0546875, "learning_rate": 6.82462327923697e-06, "loss": 0.1446, "step": 40697 }, { "epoch": 29.815384615384616, "grad_norm": 0.9453125, "learning_rate": 6.823010836566249e-06, "loss": 0.1009, "step": 40698 }, { "epoch": 29.816117216117217, "grad_norm": 0.99609375, "learning_rate": 6.821398603166318e-06, "loss": 0.126, "step": 40699 }, { "epoch": 29.816849816849818, "grad_norm": 0.5390625, "learning_rate": 6.819786579044899e-06, "loss": 0.1166, "step": 40700 }, { "epoch": 29.81758241758242, "grad_norm": 1.0859375, "learning_rate": 6.81817476420972e-06, "loss": 0.1307, "step": 40701 }, { "epoch": 29.81831501831502, "grad_norm": 1.1171875, "learning_rate": 6.816563158668513e-06, "loss": 0.1265, "step": 40702 }, { "epoch": 29.81904761904762, "grad_norm": 0.93359375, "learning_rate": 6.814951762428998e-06, "loss": 0.1003, "step": 40703 }, { "epoch": 29.81978021978022, "grad_norm": 0.8828125, "learning_rate": 6.813340575498875e-06, "loss": 0.1034, "step": 40704 }, { "epoch": 29.82051282051282, "grad_norm": 0.8046875, "learning_rate": 6.811729597885895e-06, "loss": 0.1001, "step": 40705 }, { "epoch": 29.821245421245422, "grad_norm": 0.76953125, "learning_rate": 6.8101188295977374e-06, "loss": 0.12, "step": 40706 }, { "epoch": 29.821978021978023, "grad_norm": 0.92578125, "learning_rate": 6.808508270642169e-06, "loss": 0.1329, "step": 40707 }, { "epoch": 29.822710622710623, "grad_norm": 1.03125, "learning_rate": 6.806897921026862e-06, "loss": 0.1347, "step": 40708 }, { "epoch": 29.823443223443224, "grad_norm": 1.203125, "learning_rate": 6.805287780759566e-06, "loss": 0.1165, "step": 40709 }, { "epoch": 29.824175824175825, "grad_norm": 1.234375, "learning_rate": 6.8036778498479725e-06, "loss": 0.1416, "step": 40710 }, { "epoch": 29.824908424908426, "grad_norm": 1.46875, "learning_rate": 6.802068128299799e-06, "loss": 0.1299, "step": 40711 }, { "epoch": 29.825641025641026, "grad_norm": 1.2265625, "learning_rate": 6.800458616122761e-06, "loss": 0.1291, "step": 40712 }, { "epoch": 29.826373626373627, "grad_norm": 0.98828125, "learning_rate": 6.798849313324575e-06, "loss": 0.1516, "step": 40713 }, { "epoch": 29.827106227106228, "grad_norm": 0.79296875, "learning_rate": 6.797240219912955e-06, "loss": 0.1038, "step": 40714 }, { "epoch": 29.82783882783883, "grad_norm": 0.8828125, "learning_rate": 6.7956313358955955e-06, "loss": 0.1252, "step": 40715 }, { "epoch": 29.82857142857143, "grad_norm": 0.83984375, "learning_rate": 6.794022661280208e-06, "loss": 0.134, "step": 40716 }, { "epoch": 29.82930402930403, "grad_norm": 1.03125, "learning_rate": 6.792414196074513e-06, "loss": 0.14, "step": 40717 }, { "epoch": 29.83003663003663, "grad_norm": 1.578125, "learning_rate": 6.790805940286216e-06, "loss": 0.1066, "step": 40718 }, { "epoch": 29.83076923076923, "grad_norm": 1.109375, "learning_rate": 6.78919789392301e-06, "loss": 0.091, "step": 40719 }, { "epoch": 29.831501831501832, "grad_norm": 1.359375, "learning_rate": 6.787590056992617e-06, "loss": 0.128, "step": 40720 }, { "epoch": 29.832234432234433, "grad_norm": 1.3515625, "learning_rate": 6.785982429502702e-06, "loss": 0.1078, "step": 40721 }, { "epoch": 29.832967032967034, "grad_norm": 0.7421875, "learning_rate": 6.7843750114610204e-06, "loss": 0.0982, "step": 40722 }, { "epoch": 29.833699633699634, "grad_norm": 0.91796875, "learning_rate": 6.782767802875238e-06, "loss": 0.1092, "step": 40723 }, { "epoch": 29.834432234432235, "grad_norm": 1.4296875, "learning_rate": 6.781160803753082e-06, "loss": 0.1232, "step": 40724 }, { "epoch": 29.835164835164836, "grad_norm": 1.453125, "learning_rate": 6.77955401410222e-06, "loss": 0.1171, "step": 40725 }, { "epoch": 29.835897435897436, "grad_norm": 0.8046875, "learning_rate": 6.777947433930372e-06, "loss": 0.0995, "step": 40726 }, { "epoch": 29.836630036630037, "grad_norm": 0.8671875, "learning_rate": 6.776341063245226e-06, "loss": 0.1249, "step": 40727 }, { "epoch": 29.837362637362638, "grad_norm": 0.76953125, "learning_rate": 6.774734902054487e-06, "loss": 0.1195, "step": 40728 }, { "epoch": 29.83809523809524, "grad_norm": 1.1640625, "learning_rate": 6.773128950365856e-06, "loss": 0.1388, "step": 40729 }, { "epoch": 29.83882783882784, "grad_norm": 1.1796875, "learning_rate": 6.7715232081870086e-06, "loss": 0.1598, "step": 40730 }, { "epoch": 29.83956043956044, "grad_norm": 0.859375, "learning_rate": 6.769917675525659e-06, "loss": 0.1191, "step": 40731 }, { "epoch": 29.84029304029304, "grad_norm": 1.0546875, "learning_rate": 6.7683123523894666e-06, "loss": 0.1045, "step": 40732 }, { "epoch": 29.84102564102564, "grad_norm": 1.1796875, "learning_rate": 6.766707238786167e-06, "loss": 0.1408, "step": 40733 }, { "epoch": 29.841758241758242, "grad_norm": 0.890625, "learning_rate": 6.76510233472342e-06, "loss": 0.1217, "step": 40734 }, { "epoch": 29.842490842490843, "grad_norm": 1.546875, "learning_rate": 6.763497640208939e-06, "loss": 0.1294, "step": 40735 }, { "epoch": 29.843223443223444, "grad_norm": 0.69140625, "learning_rate": 6.761893155250388e-06, "loss": 0.1217, "step": 40736 }, { "epoch": 29.843956043956045, "grad_norm": 1.265625, "learning_rate": 6.760288879855461e-06, "loss": 0.1279, "step": 40737 }, { "epoch": 29.844688644688645, "grad_norm": 0.71484375, "learning_rate": 6.758684814031849e-06, "loss": 0.1073, "step": 40738 }, { "epoch": 29.845421245421246, "grad_norm": 1.0234375, "learning_rate": 6.757080957787241e-06, "loss": 0.1328, "step": 40739 }, { "epoch": 29.846153846153847, "grad_norm": 0.68359375, "learning_rate": 6.755477311129324e-06, "loss": 0.1266, "step": 40740 }, { "epoch": 29.846886446886447, "grad_norm": 0.66796875, "learning_rate": 6.753873874065768e-06, "loss": 0.1194, "step": 40741 }, { "epoch": 29.847619047619048, "grad_norm": 0.83984375, "learning_rate": 6.752270646604267e-06, "loss": 0.1265, "step": 40742 }, { "epoch": 29.84835164835165, "grad_norm": 1.109375, "learning_rate": 6.7506676287524925e-06, "loss": 0.1279, "step": 40743 }, { "epoch": 29.84908424908425, "grad_norm": 1.09375, "learning_rate": 6.749064820518145e-06, "loss": 0.1141, "step": 40744 }, { "epoch": 29.84981684981685, "grad_norm": 1.203125, "learning_rate": 6.747462221908879e-06, "loss": 0.1294, "step": 40745 }, { "epoch": 29.85054945054945, "grad_norm": 1.171875, "learning_rate": 6.7458598329323944e-06, "loss": 0.1092, "step": 40746 }, { "epoch": 29.851282051282052, "grad_norm": 1.203125, "learning_rate": 6.744257653596335e-06, "loss": 0.1385, "step": 40747 }, { "epoch": 29.852014652014653, "grad_norm": 1.1484375, "learning_rate": 6.742655683908429e-06, "loss": 0.1321, "step": 40748 }, { "epoch": 29.852747252747253, "grad_norm": 0.8203125, "learning_rate": 6.741053923876308e-06, "loss": 0.1143, "step": 40749 }, { "epoch": 29.853479853479854, "grad_norm": 1.125, "learning_rate": 6.739452373507676e-06, "loss": 0.144, "step": 40750 }, { "epoch": 29.854212454212455, "grad_norm": 0.765625, "learning_rate": 6.737851032810182e-06, "loss": 0.1237, "step": 40751 }, { "epoch": 29.854945054945055, "grad_norm": 0.9296875, "learning_rate": 6.7362499017915096e-06, "loss": 0.0887, "step": 40752 }, { "epoch": 29.855677655677656, "grad_norm": 1.4453125, "learning_rate": 6.7346489804593325e-06, "loss": 0.0914, "step": 40753 }, { "epoch": 29.856410256410257, "grad_norm": 0.8828125, "learning_rate": 6.733048268821317e-06, "loss": 0.1124, "step": 40754 }, { "epoch": 29.857142857142858, "grad_norm": 1.1328125, "learning_rate": 6.731447766885146e-06, "loss": 0.1178, "step": 40755 }, { "epoch": 29.85787545787546, "grad_norm": 1.609375, "learning_rate": 6.7298474746584686e-06, "loss": 0.1574, "step": 40756 }, { "epoch": 29.85860805860806, "grad_norm": 0.78125, "learning_rate": 6.728247392148969e-06, "loss": 0.0926, "step": 40757 }, { "epoch": 29.85934065934066, "grad_norm": 0.81640625, "learning_rate": 6.7266475193642845e-06, "loss": 0.1285, "step": 40758 }, { "epoch": 29.86007326007326, "grad_norm": 1.21875, "learning_rate": 6.725047856312122e-06, "loss": 0.1382, "step": 40759 }, { "epoch": 29.86080586080586, "grad_norm": 0.7734375, "learning_rate": 6.72344840300012e-06, "loss": 0.129, "step": 40760 }, { "epoch": 29.861538461538462, "grad_norm": 1.4921875, "learning_rate": 6.7218491594359525e-06, "loss": 0.1414, "step": 40761 }, { "epoch": 29.862271062271063, "grad_norm": 1.1171875, "learning_rate": 6.720250125627269e-06, "loss": 0.1315, "step": 40762 }, { "epoch": 29.863003663003663, "grad_norm": 1.203125, "learning_rate": 6.718651301581738e-06, "loss": 0.1079, "step": 40763 }, { "epoch": 29.863736263736264, "grad_norm": 1.484375, "learning_rate": 6.717052687307021e-06, "loss": 0.1242, "step": 40764 }, { "epoch": 29.864468864468865, "grad_norm": 0.65234375, "learning_rate": 6.715454282810779e-06, "loss": 0.1118, "step": 40765 }, { "epoch": 29.865201465201466, "grad_norm": 0.92578125, "learning_rate": 6.7138560881006785e-06, "loss": 0.1252, "step": 40766 }, { "epoch": 29.865934065934066, "grad_norm": 0.98828125, "learning_rate": 6.712258103184359e-06, "loss": 0.1199, "step": 40767 }, { "epoch": 29.866666666666667, "grad_norm": 1.0859375, "learning_rate": 6.710660328069485e-06, "loss": 0.1357, "step": 40768 }, { "epoch": 29.867399267399268, "grad_norm": 1.328125, "learning_rate": 6.709062762763709e-06, "loss": 0.1068, "step": 40769 }, { "epoch": 29.86813186813187, "grad_norm": 0.6953125, "learning_rate": 6.7074654072747005e-06, "loss": 0.1203, "step": 40770 }, { "epoch": 29.86886446886447, "grad_norm": 1.1171875, "learning_rate": 6.705868261610093e-06, "loss": 0.1599, "step": 40771 }, { "epoch": 29.86959706959707, "grad_norm": 0.83984375, "learning_rate": 6.704271325777555e-06, "loss": 0.1071, "step": 40772 }, { "epoch": 29.87032967032967, "grad_norm": 1.734375, "learning_rate": 6.702674599784713e-06, "loss": 0.1207, "step": 40773 }, { "epoch": 29.87106227106227, "grad_norm": 1.3046875, "learning_rate": 6.701078083639254e-06, "loss": 0.1435, "step": 40774 }, { "epoch": 29.871794871794872, "grad_norm": 0.92578125, "learning_rate": 6.6994817773488e-06, "loss": 0.1092, "step": 40775 }, { "epoch": 29.872527472527473, "grad_norm": 0.828125, "learning_rate": 6.697885680921017e-06, "loss": 0.1092, "step": 40776 }, { "epoch": 29.873260073260074, "grad_norm": 0.8828125, "learning_rate": 6.696289794363529e-06, "loss": 0.0916, "step": 40777 }, { "epoch": 29.873992673992674, "grad_norm": 1.21875, "learning_rate": 6.694694117683997e-06, "loss": 0.1369, "step": 40778 }, { "epoch": 29.874725274725275, "grad_norm": 0.59765625, "learning_rate": 6.693098650890068e-06, "loss": 0.1091, "step": 40779 }, { "epoch": 29.875457875457876, "grad_norm": 1.359375, "learning_rate": 6.691503393989383e-06, "loss": 0.1411, "step": 40780 }, { "epoch": 29.876190476190477, "grad_norm": 1.4140625, "learning_rate": 6.6899083469895984e-06, "loss": 0.1233, "step": 40781 }, { "epoch": 29.876923076923077, "grad_norm": 0.90625, "learning_rate": 6.688313509898332e-06, "loss": 0.1207, "step": 40782 }, { "epoch": 29.877655677655678, "grad_norm": 1.2109375, "learning_rate": 6.6867188827232505e-06, "loss": 0.15, "step": 40783 }, { "epoch": 29.87838827838828, "grad_norm": 1.1015625, "learning_rate": 6.685124465471963e-06, "loss": 0.1184, "step": 40784 }, { "epoch": 29.87912087912088, "grad_norm": 0.93359375, "learning_rate": 6.683530258152143e-06, "loss": 0.1135, "step": 40785 }, { "epoch": 29.87985347985348, "grad_norm": 1.171875, "learning_rate": 6.681936260771407e-06, "loss": 0.1336, "step": 40786 }, { "epoch": 29.88058608058608, "grad_norm": 1.625, "learning_rate": 6.680342473337412e-06, "loss": 0.1307, "step": 40787 }, { "epoch": 29.88131868131868, "grad_norm": 0.72265625, "learning_rate": 6.678748895857766e-06, "loss": 0.1142, "step": 40788 }, { "epoch": 29.882051282051282, "grad_norm": 1.0390625, "learning_rate": 6.677155528340121e-06, "loss": 0.1092, "step": 40789 }, { "epoch": 29.882783882783883, "grad_norm": 1.0546875, "learning_rate": 6.67556237079211e-06, "loss": 0.1279, "step": 40790 }, { "epoch": 29.883516483516484, "grad_norm": 1.0078125, "learning_rate": 6.673969423221367e-06, "loss": 0.1413, "step": 40791 }, { "epoch": 29.884249084249085, "grad_norm": 0.875, "learning_rate": 6.672376685635531e-06, "loss": 0.1247, "step": 40792 }, { "epoch": 29.884981684981685, "grad_norm": 0.9609375, "learning_rate": 6.670784158042218e-06, "loss": 0.1293, "step": 40793 }, { "epoch": 29.885714285714286, "grad_norm": 1.25, "learning_rate": 6.669191840449067e-06, "loss": 0.1306, "step": 40794 }, { "epoch": 29.886446886446887, "grad_norm": 1.2265625, "learning_rate": 6.6675997328637066e-06, "loss": 0.1305, "step": 40795 }, { "epoch": 29.887179487179488, "grad_norm": 1.03125, "learning_rate": 6.666007835293775e-06, "loss": 0.1094, "step": 40796 }, { "epoch": 29.88791208791209, "grad_norm": 2.140625, "learning_rate": 6.664416147746878e-06, "loss": 0.1428, "step": 40797 }, { "epoch": 29.88864468864469, "grad_norm": 0.9921875, "learning_rate": 6.6628246702306664e-06, "loss": 0.1096, "step": 40798 }, { "epoch": 29.88937728937729, "grad_norm": 1.3515625, "learning_rate": 6.661233402752728e-06, "loss": 0.1138, "step": 40799 }, { "epoch": 29.89010989010989, "grad_norm": 1.21875, "learning_rate": 6.659642345320738e-06, "loss": 0.1195, "step": 40800 }, { "epoch": 29.89084249084249, "grad_norm": 1.5703125, "learning_rate": 6.658051497942276e-06, "loss": 0.156, "step": 40801 }, { "epoch": 29.891575091575092, "grad_norm": 0.921875, "learning_rate": 6.65646086062499e-06, "loss": 0.1542, "step": 40802 }, { "epoch": 29.892307692307693, "grad_norm": 1.046875, "learning_rate": 6.654870433376494e-06, "loss": 0.115, "step": 40803 }, { "epoch": 29.893040293040293, "grad_norm": 1.0234375, "learning_rate": 6.6532802162044065e-06, "loss": 0.0945, "step": 40804 }, { "epoch": 29.893772893772894, "grad_norm": 1.2890625, "learning_rate": 6.651690209116339e-06, "loss": 0.13, "step": 40805 }, { "epoch": 29.894505494505495, "grad_norm": 0.7734375, "learning_rate": 6.650100412119924e-06, "loss": 0.1125, "step": 40806 }, { "epoch": 29.895238095238096, "grad_norm": 1.015625, "learning_rate": 6.6485108252227796e-06, "loss": 0.1339, "step": 40807 }, { "epoch": 29.895970695970696, "grad_norm": 0.9765625, "learning_rate": 6.646921448432506e-06, "loss": 0.1138, "step": 40808 }, { "epoch": 29.896703296703297, "grad_norm": 1.5078125, "learning_rate": 6.645332281756736e-06, "loss": 0.1286, "step": 40809 }, { "epoch": 29.897435897435898, "grad_norm": 1.2421875, "learning_rate": 6.643743325203058e-06, "loss": 0.1145, "step": 40810 }, { "epoch": 29.8981684981685, "grad_norm": 1.015625, "learning_rate": 6.642154578779125e-06, "loss": 0.0848, "step": 40811 }, { "epoch": 29.8989010989011, "grad_norm": 1.296875, "learning_rate": 6.640566042492511e-06, "loss": 0.1294, "step": 40812 }, { "epoch": 29.8996336996337, "grad_norm": 0.875, "learning_rate": 6.638977716350859e-06, "loss": 0.1145, "step": 40813 }, { "epoch": 29.9003663003663, "grad_norm": 0.56640625, "learning_rate": 6.637389600361753e-06, "loss": 0.1173, "step": 40814 }, { "epoch": 29.9010989010989, "grad_norm": 1.59375, "learning_rate": 6.635801694532806e-06, "loss": 0.1254, "step": 40815 }, { "epoch": 29.901831501831502, "grad_norm": 0.9609375, "learning_rate": 6.6342139988716385e-06, "loss": 0.1269, "step": 40816 }, { "epoch": 29.902564102564103, "grad_norm": 0.8671875, "learning_rate": 6.632626513385854e-06, "loss": 0.1092, "step": 40817 }, { "epoch": 29.903296703296704, "grad_norm": 0.89453125, "learning_rate": 6.63103923808306e-06, "loss": 0.1027, "step": 40818 }, { "epoch": 29.904029304029304, "grad_norm": 0.81640625, "learning_rate": 6.629452172970856e-06, "loss": 0.1257, "step": 40819 }, { "epoch": 29.904761904761905, "grad_norm": 1.34375, "learning_rate": 6.627865318056843e-06, "loss": 0.1034, "step": 40820 }, { "epoch": 29.905494505494506, "grad_norm": 1.375, "learning_rate": 6.626278673348628e-06, "loss": 0.1219, "step": 40821 }, { "epoch": 29.906227106227107, "grad_norm": 0.89453125, "learning_rate": 6.62469223885383e-06, "loss": 0.1146, "step": 40822 }, { "epoch": 29.906959706959707, "grad_norm": 0.90234375, "learning_rate": 6.62310601458002e-06, "loss": 0.1064, "step": 40823 }, { "epoch": 29.907692307692308, "grad_norm": 0.65234375, "learning_rate": 6.621520000534824e-06, "loss": 0.1188, "step": 40824 }, { "epoch": 29.90842490842491, "grad_norm": 0.92578125, "learning_rate": 6.6199341967258106e-06, "loss": 0.144, "step": 40825 }, { "epoch": 29.90915750915751, "grad_norm": 1.390625, "learning_rate": 6.618348603160618e-06, "loss": 0.1179, "step": 40826 }, { "epoch": 29.90989010989011, "grad_norm": 1.2734375, "learning_rate": 6.616763219846814e-06, "loss": 0.1424, "step": 40827 }, { "epoch": 29.91062271062271, "grad_norm": 0.765625, "learning_rate": 6.615178046792003e-06, "loss": 0.122, "step": 40828 }, { "epoch": 29.91135531135531, "grad_norm": 1.1328125, "learning_rate": 6.6135930840037875e-06, "loss": 0.1019, "step": 40829 }, { "epoch": 29.912087912087912, "grad_norm": 1.2578125, "learning_rate": 6.612008331489744e-06, "loss": 0.0999, "step": 40830 }, { "epoch": 29.912820512820513, "grad_norm": 0.91015625, "learning_rate": 6.61042378925748e-06, "loss": 0.1322, "step": 40831 }, { "epoch": 29.913553113553114, "grad_norm": 0.859375, "learning_rate": 6.608839457314581e-06, "loss": 0.1063, "step": 40832 }, { "epoch": 29.914285714285715, "grad_norm": 0.81640625, "learning_rate": 6.6072553356686534e-06, "loss": 0.1061, "step": 40833 }, { "epoch": 29.915018315018315, "grad_norm": 0.99609375, "learning_rate": 6.605671424327263e-06, "loss": 0.138, "step": 40834 }, { "epoch": 29.915750915750916, "grad_norm": 0.6796875, "learning_rate": 6.6040877232980205e-06, "loss": 0.0989, "step": 40835 }, { "epoch": 29.916483516483517, "grad_norm": 1.0078125, "learning_rate": 6.602504232588483e-06, "loss": 0.1252, "step": 40836 }, { "epoch": 29.917216117216118, "grad_norm": 1.015625, "learning_rate": 6.600920952206277e-06, "loss": 0.1114, "step": 40837 }, { "epoch": 29.91794871794872, "grad_norm": 1.1875, "learning_rate": 6.599337882158964e-06, "loss": 0.1311, "step": 40838 }, { "epoch": 29.91868131868132, "grad_norm": 0.88671875, "learning_rate": 6.597755022454141e-06, "loss": 0.1252, "step": 40839 }, { "epoch": 29.91941391941392, "grad_norm": 1.0390625, "learning_rate": 6.59617237309938e-06, "loss": 0.124, "step": 40840 }, { "epoch": 29.92014652014652, "grad_norm": 0.81640625, "learning_rate": 6.5945899341022645e-06, "loss": 0.1222, "step": 40841 }, { "epoch": 29.92087912087912, "grad_norm": 1.0859375, "learning_rate": 6.593007705470383e-06, "loss": 0.0938, "step": 40842 }, { "epoch": 29.921611721611722, "grad_norm": 1.125, "learning_rate": 6.591425687211317e-06, "loss": 0.1267, "step": 40843 }, { "epoch": 29.922344322344323, "grad_norm": 1.359375, "learning_rate": 6.589843879332654e-06, "loss": 0.122, "step": 40844 }, { "epoch": 29.923076923076923, "grad_norm": 0.8203125, "learning_rate": 6.588262281841955e-06, "loss": 0.1087, "step": 40845 }, { "epoch": 29.923809523809524, "grad_norm": 1.0390625, "learning_rate": 6.586680894746805e-06, "loss": 0.1272, "step": 40846 }, { "epoch": 29.924542124542125, "grad_norm": 0.6640625, "learning_rate": 6.585099718054783e-06, "loss": 0.0957, "step": 40847 }, { "epoch": 29.925274725274726, "grad_norm": 0.83984375, "learning_rate": 6.583518751773478e-06, "loss": 0.115, "step": 40848 }, { "epoch": 29.926007326007326, "grad_norm": 0.94921875, "learning_rate": 6.58193799591044e-06, "loss": 0.1087, "step": 40849 }, { "epoch": 29.926739926739927, "grad_norm": 1.8671875, "learning_rate": 6.5803574504732595e-06, "loss": 0.129, "step": 40850 }, { "epoch": 29.927472527472528, "grad_norm": 0.71484375, "learning_rate": 6.578777115469488e-06, "loss": 0.0993, "step": 40851 }, { "epoch": 29.92820512820513, "grad_norm": 1.4140625, "learning_rate": 6.577196990906735e-06, "loss": 0.1084, "step": 40852 }, { "epoch": 29.92893772893773, "grad_norm": 1.3203125, "learning_rate": 6.5756170767925364e-06, "loss": 0.1242, "step": 40853 }, { "epoch": 29.92967032967033, "grad_norm": 0.90625, "learning_rate": 6.574037373134481e-06, "loss": 0.1039, "step": 40854 }, { "epoch": 29.93040293040293, "grad_norm": 1.203125, "learning_rate": 6.5724578799401375e-06, "loss": 0.1131, "step": 40855 }, { "epoch": 29.93113553113553, "grad_norm": 1.2265625, "learning_rate": 6.570878597217062e-06, "loss": 0.1288, "step": 40856 }, { "epoch": 29.931868131868132, "grad_norm": 1.0078125, "learning_rate": 6.569299524972826e-06, "loss": 0.1412, "step": 40857 }, { "epoch": 29.932600732600733, "grad_norm": 1.1796875, "learning_rate": 6.5677206632150014e-06, "loss": 0.1084, "step": 40858 }, { "epoch": 29.933333333333334, "grad_norm": 1.4609375, "learning_rate": 6.566142011951155e-06, "loss": 0.1185, "step": 40859 }, { "epoch": 29.934065934065934, "grad_norm": 0.7890625, "learning_rate": 6.564563571188834e-06, "loss": 0.1424, "step": 40860 }, { "epoch": 29.934798534798535, "grad_norm": 1.09375, "learning_rate": 6.562985340935628e-06, "loss": 0.1095, "step": 40861 }, { "epoch": 29.935531135531136, "grad_norm": 1.3125, "learning_rate": 6.561407321199056e-06, "loss": 0.1388, "step": 40862 }, { "epoch": 29.936263736263736, "grad_norm": 0.7890625, "learning_rate": 6.5598295119867285e-06, "loss": 0.105, "step": 40863 }, { "epoch": 29.936996336996337, "grad_norm": 1.4609375, "learning_rate": 6.558251913306175e-06, "loss": 0.1245, "step": 40864 }, { "epoch": 29.937728937728938, "grad_norm": 0.73046875, "learning_rate": 6.55667452516497e-06, "loss": 0.102, "step": 40865 }, { "epoch": 29.93846153846154, "grad_norm": 2.046875, "learning_rate": 6.55509734757065e-06, "loss": 0.125, "step": 40866 }, { "epoch": 29.93919413919414, "grad_norm": 1.109375, "learning_rate": 6.553520380530785e-06, "loss": 0.1301, "step": 40867 }, { "epoch": 29.93992673992674, "grad_norm": 1.03125, "learning_rate": 6.551943624052931e-06, "loss": 0.1288, "step": 40868 }, { "epoch": 29.94065934065934, "grad_norm": 1.09375, "learning_rate": 6.5503670781446455e-06, "loss": 0.1347, "step": 40869 }, { "epoch": 29.94139194139194, "grad_norm": 1.40625, "learning_rate": 6.5487907428134874e-06, "loss": 0.1497, "step": 40870 }, { "epoch": 29.942124542124542, "grad_norm": 0.99609375, "learning_rate": 6.547214618066994e-06, "loss": 0.1017, "step": 40871 }, { "epoch": 29.942857142857143, "grad_norm": 0.796875, "learning_rate": 6.5456387039127195e-06, "loss": 0.1254, "step": 40872 }, { "epoch": 29.943589743589744, "grad_norm": 1.3671875, "learning_rate": 6.544063000358218e-06, "loss": 0.0996, "step": 40873 }, { "epoch": 29.944322344322345, "grad_norm": 0.64453125, "learning_rate": 6.542487507411053e-06, "loss": 0.0988, "step": 40874 }, { "epoch": 29.945054945054945, "grad_norm": 0.75, "learning_rate": 6.540912225078752e-06, "loss": 0.1155, "step": 40875 }, { "epoch": 29.945787545787546, "grad_norm": 1.03125, "learning_rate": 6.539337153368879e-06, "loss": 0.1079, "step": 40876 }, { "epoch": 29.946520146520147, "grad_norm": 0.84375, "learning_rate": 6.537762292288966e-06, "loss": 0.1011, "step": 40877 }, { "epoch": 29.947252747252747, "grad_norm": 1.46875, "learning_rate": 6.536187641846567e-06, "loss": 0.1015, "step": 40878 }, { "epoch": 29.947985347985348, "grad_norm": 1.03125, "learning_rate": 6.534613202049222e-06, "loss": 0.1466, "step": 40879 }, { "epoch": 29.94871794871795, "grad_norm": 1.1171875, "learning_rate": 6.533038972904479e-06, "loss": 0.1066, "step": 40880 }, { "epoch": 29.94945054945055, "grad_norm": 0.80859375, "learning_rate": 6.531464954419886e-06, "loss": 0.1158, "step": 40881 }, { "epoch": 29.95018315018315, "grad_norm": 1.171875, "learning_rate": 6.529891146602975e-06, "loss": 0.141, "step": 40882 }, { "epoch": 29.95091575091575, "grad_norm": 0.875, "learning_rate": 6.528317549461282e-06, "loss": 0.1012, "step": 40883 }, { "epoch": 29.951648351648352, "grad_norm": 1.2109375, "learning_rate": 6.5267441630023614e-06, "loss": 0.1099, "step": 40884 }, { "epoch": 29.952380952380953, "grad_norm": 1.265625, "learning_rate": 6.525170987233753e-06, "loss": 0.1053, "step": 40885 }, { "epoch": 29.953113553113553, "grad_norm": 1.109375, "learning_rate": 6.523598022162972e-06, "loss": 0.1592, "step": 40886 }, { "epoch": 29.953846153846154, "grad_norm": 0.6015625, "learning_rate": 6.522025267797588e-06, "loss": 0.1113, "step": 40887 }, { "epoch": 29.954578754578755, "grad_norm": 0.859375, "learning_rate": 6.520452724145091e-06, "loss": 0.1186, "step": 40888 }, { "epoch": 29.955311355311355, "grad_norm": 1.21875, "learning_rate": 6.518880391213067e-06, "loss": 0.1402, "step": 40889 }, { "epoch": 29.956043956043956, "grad_norm": 1.03125, "learning_rate": 6.517308269009008e-06, "loss": 0.115, "step": 40890 }, { "epoch": 29.956776556776557, "grad_norm": 1.046875, "learning_rate": 6.51573635754048e-06, "loss": 0.1022, "step": 40891 }, { "epoch": 29.957509157509158, "grad_norm": 1.375, "learning_rate": 6.514164656814986e-06, "loss": 0.1221, "step": 40892 }, { "epoch": 29.95824175824176, "grad_norm": 0.8046875, "learning_rate": 6.512593166840074e-06, "loss": 0.1026, "step": 40893 }, { "epoch": 29.95897435897436, "grad_norm": 2.125, "learning_rate": 6.5110218876232615e-06, "loss": 0.1331, "step": 40894 }, { "epoch": 29.95970695970696, "grad_norm": 1.0859375, "learning_rate": 6.509450819172083e-06, "loss": 0.1088, "step": 40895 }, { "epoch": 29.96043956043956, "grad_norm": 1.2578125, "learning_rate": 6.507879961494084e-06, "loss": 0.1551, "step": 40896 }, { "epoch": 29.96117216117216, "grad_norm": 0.9453125, "learning_rate": 6.506309314596762e-06, "loss": 0.1383, "step": 40897 }, { "epoch": 29.961904761904762, "grad_norm": 0.65234375, "learning_rate": 6.504738878487647e-06, "loss": 0.1019, "step": 40898 }, { "epoch": 29.962637362637363, "grad_norm": 1.171875, "learning_rate": 6.503168653174278e-06, "loss": 0.1406, "step": 40899 }, { "epoch": 29.963369963369964, "grad_norm": 1.171875, "learning_rate": 6.501598638664177e-06, "loss": 0.0934, "step": 40900 }, { "epoch": 29.964102564102564, "grad_norm": 0.90234375, "learning_rate": 6.500028834964855e-06, "loss": 0.1241, "step": 40901 }, { "epoch": 29.964835164835165, "grad_norm": 0.86328125, "learning_rate": 6.498459242083844e-06, "loss": 0.1184, "step": 40902 }, { "epoch": 29.965567765567766, "grad_norm": 1.765625, "learning_rate": 6.496889860028654e-06, "loss": 0.1387, "step": 40903 }, { "epoch": 29.966300366300366, "grad_norm": 1.2421875, "learning_rate": 6.495320688806809e-06, "loss": 0.1385, "step": 40904 }, { "epoch": 29.967032967032967, "grad_norm": 0.8984375, "learning_rate": 6.493751728425829e-06, "loss": 0.135, "step": 40905 }, { "epoch": 29.967765567765568, "grad_norm": 0.96875, "learning_rate": 6.492182978893229e-06, "loss": 0.1215, "step": 40906 }, { "epoch": 29.96849816849817, "grad_norm": 1.46875, "learning_rate": 6.490614440216533e-06, "loss": 0.1558, "step": 40907 }, { "epoch": 29.96923076923077, "grad_norm": 0.98828125, "learning_rate": 6.489046112403249e-06, "loss": 0.1173, "step": 40908 }, { "epoch": 29.96996336996337, "grad_norm": 1.40625, "learning_rate": 6.487477995460883e-06, "loss": 0.1258, "step": 40909 }, { "epoch": 29.97069597069597, "grad_norm": 1.0390625, "learning_rate": 6.485910089396967e-06, "loss": 0.1028, "step": 40910 }, { "epoch": 29.97142857142857, "grad_norm": 1.4375, "learning_rate": 6.484342394219015e-06, "loss": 0.1158, "step": 40911 }, { "epoch": 29.972161172161172, "grad_norm": 1.4296875, "learning_rate": 6.482774909934514e-06, "loss": 0.1424, "step": 40912 }, { "epoch": 29.972893772893773, "grad_norm": 1.46875, "learning_rate": 6.481207636550998e-06, "loss": 0.1288, "step": 40913 }, { "epoch": 29.973626373626374, "grad_norm": 1.015625, "learning_rate": 6.479640574075947e-06, "loss": 0.0964, "step": 40914 }, { "epoch": 29.974358974358974, "grad_norm": 0.875, "learning_rate": 6.478073722516911e-06, "loss": 0.1291, "step": 40915 }, { "epoch": 29.975091575091575, "grad_norm": 0.6796875, "learning_rate": 6.476507081881361e-06, "loss": 0.1223, "step": 40916 }, { "epoch": 29.975824175824176, "grad_norm": 1.0234375, "learning_rate": 6.4749406521768335e-06, "loss": 0.1275, "step": 40917 }, { "epoch": 29.976556776556777, "grad_norm": 1.1796875, "learning_rate": 6.473374433410805e-06, "loss": 0.1142, "step": 40918 }, { "epoch": 29.977289377289377, "grad_norm": 1.5, "learning_rate": 6.471808425590792e-06, "loss": 0.0981, "step": 40919 }, { "epoch": 29.978021978021978, "grad_norm": 1.0078125, "learning_rate": 6.470242628724297e-06, "loss": 0.0789, "step": 40920 }, { "epoch": 29.97875457875458, "grad_norm": 0.62890625, "learning_rate": 6.468677042818827e-06, "loss": 0.1146, "step": 40921 }, { "epoch": 29.97948717948718, "grad_norm": 0.99609375, "learning_rate": 6.467111667881888e-06, "loss": 0.0978, "step": 40922 }, { "epoch": 29.98021978021978, "grad_norm": 1.3984375, "learning_rate": 6.46554650392096e-06, "loss": 0.1351, "step": 40923 }, { "epoch": 29.98095238095238, "grad_norm": 0.83984375, "learning_rate": 6.463981550943557e-06, "loss": 0.1273, "step": 40924 }, { "epoch": 29.98168498168498, "grad_norm": 1.1328125, "learning_rate": 6.462416808957173e-06, "loss": 0.1337, "step": 40925 }, { "epoch": 29.982417582417582, "grad_norm": 0.58984375, "learning_rate": 6.460852277969321e-06, "loss": 0.1184, "step": 40926 }, { "epoch": 29.983150183150183, "grad_norm": 1.328125, "learning_rate": 6.459287957987469e-06, "loss": 0.1089, "step": 40927 }, { "epoch": 29.983882783882784, "grad_norm": 1.2109375, "learning_rate": 6.457723849019136e-06, "loss": 0.1424, "step": 40928 }, { "epoch": 29.984615384615385, "grad_norm": 1.1796875, "learning_rate": 6.4561599510717956e-06, "loss": 0.149, "step": 40929 }, { "epoch": 29.985347985347985, "grad_norm": 0.8359375, "learning_rate": 6.45459626415295e-06, "loss": 0.1156, "step": 40930 }, { "epoch": 29.986080586080586, "grad_norm": 0.97265625, "learning_rate": 6.4530327882701e-06, "loss": 0.1082, "step": 40931 }, { "epoch": 29.986813186813187, "grad_norm": 0.69921875, "learning_rate": 6.451469523430724e-06, "loss": 0.1079, "step": 40932 }, { "epoch": 29.987545787545788, "grad_norm": 0.78515625, "learning_rate": 6.449906469642327e-06, "loss": 0.1114, "step": 40933 }, { "epoch": 29.98827838827839, "grad_norm": 0.5625, "learning_rate": 6.448343626912378e-06, "loss": 0.1038, "step": 40934 }, { "epoch": 29.98901098901099, "grad_norm": 1.03125, "learning_rate": 6.4467809952483836e-06, "loss": 0.1184, "step": 40935 }, { "epoch": 29.98974358974359, "grad_norm": 1.1171875, "learning_rate": 6.445218574657813e-06, "loss": 0.1262, "step": 40936 }, { "epoch": 29.99047619047619, "grad_norm": 0.9609375, "learning_rate": 6.443656365148178e-06, "loss": 0.146, "step": 40937 }, { "epoch": 29.99120879120879, "grad_norm": 1.1328125, "learning_rate": 6.442094366726938e-06, "loss": 0.103, "step": 40938 }, { "epoch": 29.991941391941392, "grad_norm": 1.0078125, "learning_rate": 6.440532579401595e-06, "loss": 0.1274, "step": 40939 }, { "epoch": 29.992673992673993, "grad_norm": 1.4296875, "learning_rate": 6.4389710031796096e-06, "loss": 0.1294, "step": 40940 }, { "epoch": 29.993406593406593, "grad_norm": 1.234375, "learning_rate": 6.437409638068496e-06, "loss": 0.1126, "step": 40941 }, { "epoch": 29.994139194139194, "grad_norm": 0.83203125, "learning_rate": 6.435848484075703e-06, "loss": 0.0919, "step": 40942 }, { "epoch": 29.994871794871795, "grad_norm": 0.9609375, "learning_rate": 6.434287541208731e-06, "loss": 0.1195, "step": 40943 }, { "epoch": 29.995604395604396, "grad_norm": 1.71875, "learning_rate": 6.432726809475059e-06, "loss": 0.1268, "step": 40944 }, { "epoch": 29.996336996336996, "grad_norm": 0.80078125, "learning_rate": 6.431166288882151e-06, "loss": 0.1104, "step": 40945 }, { "epoch": 29.997069597069597, "grad_norm": 0.8671875, "learning_rate": 6.429605979437493e-06, "loss": 0.1507, "step": 40946 }, { "epoch": 29.997802197802198, "grad_norm": 0.94921875, "learning_rate": 6.428045881148558e-06, "loss": 0.111, "step": 40947 }, { "epoch": 29.9985347985348, "grad_norm": 1.015625, "learning_rate": 6.426485994022835e-06, "loss": 0.1478, "step": 40948 }, { "epoch": 29.9992673992674, "grad_norm": 1.46875, "learning_rate": 6.424926318067773e-06, "loss": 0.1386, "step": 40949 }, { "epoch": 30.0, "grad_norm": 0.57421875, "learning_rate": 6.42336685329086e-06, "loss": 0.1088, "step": 40950 }, { "epoch": 30.0007326007326, "grad_norm": 1.6328125, "learning_rate": 6.421807599699567e-06, "loss": 0.159, "step": 40951 }, { "epoch": 30.0014652014652, "grad_norm": 1.03125, "learning_rate": 6.4202485573013724e-06, "loss": 0.1296, "step": 40952 }, { "epoch": 30.002197802197802, "grad_norm": 1.1015625, "learning_rate": 6.41868972610373e-06, "loss": 0.104, "step": 40953 }, { "epoch": 30.002930402930403, "grad_norm": 1.3828125, "learning_rate": 6.4171311061141215e-06, "loss": 0.1247, "step": 40954 }, { "epoch": 30.003663003663004, "grad_norm": 1.015625, "learning_rate": 6.41557269734e-06, "loss": 0.1689, "step": 40955 }, { "epoch": 30.004395604395604, "grad_norm": 0.765625, "learning_rate": 6.414014499788847e-06, "loss": 0.1296, "step": 40956 }, { "epoch": 30.005128205128205, "grad_norm": 1.2109375, "learning_rate": 6.412456513468118e-06, "loss": 0.1416, "step": 40957 }, { "epoch": 30.005860805860806, "grad_norm": 0.734375, "learning_rate": 6.410898738385282e-06, "loss": 0.1252, "step": 40958 }, { "epoch": 30.006593406593407, "grad_norm": 1.15625, "learning_rate": 6.409341174547819e-06, "loss": 0.1543, "step": 40959 }, { "epoch": 30.007326007326007, "grad_norm": 1.1484375, "learning_rate": 6.407783821963165e-06, "loss": 0.0994, "step": 40960 }, { "epoch": 30.008058608058608, "grad_norm": 0.7578125, "learning_rate": 6.4062266806387865e-06, "loss": 0.0937, "step": 40961 }, { "epoch": 30.00879120879121, "grad_norm": 0.82421875, "learning_rate": 6.404669750582159e-06, "loss": 0.1367, "step": 40962 }, { "epoch": 30.00952380952381, "grad_norm": 1.15625, "learning_rate": 6.403113031800741e-06, "loss": 0.1449, "step": 40963 }, { "epoch": 30.01025641025641, "grad_norm": 0.9140625, "learning_rate": 6.401556524301976e-06, "loss": 0.1212, "step": 40964 }, { "epoch": 30.01098901098901, "grad_norm": 1.2734375, "learning_rate": 6.400000228093335e-06, "loss": 0.1259, "step": 40965 }, { "epoch": 30.01172161172161, "grad_norm": 1.0078125, "learning_rate": 6.3984441431822545e-06, "loss": 0.137, "step": 40966 }, { "epoch": 30.012454212454212, "grad_norm": 1.109375, "learning_rate": 6.3968882695762274e-06, "loss": 0.1434, "step": 40967 }, { "epoch": 30.013186813186813, "grad_norm": 0.828125, "learning_rate": 6.395332607282674e-06, "loss": 0.1285, "step": 40968 }, { "epoch": 30.013919413919414, "grad_norm": 0.9296875, "learning_rate": 6.39377715630906e-06, "loss": 0.1271, "step": 40969 }, { "epoch": 30.014652014652015, "grad_norm": 1.328125, "learning_rate": 6.39222191666285e-06, "loss": 0.1264, "step": 40970 }, { "epoch": 30.015384615384615, "grad_norm": 0.81640625, "learning_rate": 6.390666888351475e-06, "loss": 0.0973, "step": 40971 }, { "epoch": 30.016117216117216, "grad_norm": 1.1171875, "learning_rate": 6.389112071382393e-06, "loss": 0.1162, "step": 40972 }, { "epoch": 30.016849816849817, "grad_norm": 0.765625, "learning_rate": 6.387557465763057e-06, "loss": 0.1034, "step": 40973 }, { "epoch": 30.017582417582418, "grad_norm": 0.7265625, "learning_rate": 6.386003071500929e-06, "loss": 0.0899, "step": 40974 }, { "epoch": 30.01831501831502, "grad_norm": 1.0078125, "learning_rate": 6.384448888603423e-06, "loss": 0.12, "step": 40975 }, { "epoch": 30.01904761904762, "grad_norm": 1.34375, "learning_rate": 6.382894917078012e-06, "loss": 0.1419, "step": 40976 }, { "epoch": 30.01978021978022, "grad_norm": 0.96484375, "learning_rate": 6.381341156932128e-06, "loss": 0.1315, "step": 40977 }, { "epoch": 30.02051282051282, "grad_norm": 1.328125, "learning_rate": 6.379787608173239e-06, "loss": 0.1156, "step": 40978 }, { "epoch": 30.02124542124542, "grad_norm": 1.1875, "learning_rate": 6.378234270808757e-06, "loss": 0.1295, "step": 40979 }, { "epoch": 30.021978021978022, "grad_norm": 1.2421875, "learning_rate": 6.376681144846146e-06, "loss": 0.1286, "step": 40980 }, { "epoch": 30.022710622710623, "grad_norm": 0.9296875, "learning_rate": 6.375128230292836e-06, "loss": 0.1193, "step": 40981 }, { "epoch": 30.023443223443223, "grad_norm": 1.015625, "learning_rate": 6.373575527156268e-06, "loss": 0.1306, "step": 40982 }, { "epoch": 30.024175824175824, "grad_norm": 1.0234375, "learning_rate": 6.372023035443888e-06, "loss": 0.119, "step": 40983 }, { "epoch": 30.024908424908425, "grad_norm": 1.015625, "learning_rate": 6.3704707551631285e-06, "loss": 0.1076, "step": 40984 }, { "epoch": 30.025641025641026, "grad_norm": 0.703125, "learning_rate": 6.368918686321444e-06, "loss": 0.0987, "step": 40985 }, { "epoch": 30.026373626373626, "grad_norm": 1.15625, "learning_rate": 6.3673668289262455e-06, "loss": 0.1293, "step": 40986 }, { "epoch": 30.027106227106227, "grad_norm": 0.73828125, "learning_rate": 6.365815182984981e-06, "loss": 0.0965, "step": 40987 }, { "epoch": 30.027838827838828, "grad_norm": 0.84765625, "learning_rate": 6.364263748505084e-06, "loss": 0.1148, "step": 40988 }, { "epoch": 30.02857142857143, "grad_norm": 0.65234375, "learning_rate": 6.362712525494001e-06, "loss": 0.1242, "step": 40989 }, { "epoch": 30.02930402930403, "grad_norm": 1.2109375, "learning_rate": 6.3611615139591376e-06, "loss": 0.1082, "step": 40990 }, { "epoch": 30.03003663003663, "grad_norm": 1.1484375, "learning_rate": 6.359610713907948e-06, "loss": 0.1374, "step": 40991 }, { "epoch": 30.03076923076923, "grad_norm": 0.91796875, "learning_rate": 6.358060125347838e-06, "loss": 0.1484, "step": 40992 }, { "epoch": 30.03150183150183, "grad_norm": 0.62109375, "learning_rate": 6.356509748286269e-06, "loss": 0.1146, "step": 40993 }, { "epoch": 30.032234432234432, "grad_norm": 0.7578125, "learning_rate": 6.354959582730645e-06, "loss": 0.1118, "step": 40994 }, { "epoch": 30.032967032967033, "grad_norm": 1.46875, "learning_rate": 6.353409628688405e-06, "loss": 0.1277, "step": 40995 }, { "epoch": 30.033699633699634, "grad_norm": 0.79296875, "learning_rate": 6.351859886166974e-06, "loss": 0.1357, "step": 40996 }, { "epoch": 30.034432234432234, "grad_norm": 1.109375, "learning_rate": 6.350310355173772e-06, "loss": 0.1197, "step": 40997 }, { "epoch": 30.035164835164835, "grad_norm": 0.90234375, "learning_rate": 6.348761035716226e-06, "loss": 0.1235, "step": 40998 }, { "epoch": 30.035897435897436, "grad_norm": 1.0625, "learning_rate": 6.347211927801754e-06, "loss": 0.1066, "step": 40999 }, { "epoch": 30.036630036630036, "grad_norm": 0.81640625, "learning_rate": 6.345663031437799e-06, "loss": 0.1101, "step": 41000 }, { "epoch": 30.037362637362637, "grad_norm": 1.515625, "learning_rate": 6.344114346631756e-06, "loss": 0.1134, "step": 41001 }, { "epoch": 30.038095238095238, "grad_norm": 0.70703125, "learning_rate": 6.342565873391057e-06, "loss": 0.1045, "step": 41002 }, { "epoch": 30.03882783882784, "grad_norm": 1.2578125, "learning_rate": 6.341017611723125e-06, "loss": 0.1153, "step": 41003 }, { "epoch": 30.03956043956044, "grad_norm": 0.734375, "learning_rate": 6.339469561635375e-06, "loss": 0.1063, "step": 41004 }, { "epoch": 30.04029304029304, "grad_norm": 0.98828125, "learning_rate": 6.33792172313522e-06, "loss": 0.097, "step": 41005 }, { "epoch": 30.04102564102564, "grad_norm": 0.7578125, "learning_rate": 6.33637409623009e-06, "loss": 0.1262, "step": 41006 }, { "epoch": 30.04175824175824, "grad_norm": 1.0859375, "learning_rate": 6.334826680927376e-06, "loss": 0.122, "step": 41007 }, { "epoch": 30.042490842490842, "grad_norm": 1.4609375, "learning_rate": 6.333279477234505e-06, "loss": 0.1224, "step": 41008 }, { "epoch": 30.043223443223443, "grad_norm": 1.5625, "learning_rate": 6.331732485158897e-06, "loss": 0.1167, "step": 41009 }, { "epoch": 30.043956043956044, "grad_norm": 1.109375, "learning_rate": 6.330185704707954e-06, "loss": 0.1131, "step": 41010 }, { "epoch": 30.044688644688645, "grad_norm": 1.0546875, "learning_rate": 6.3286391358890975e-06, "loss": 0.1161, "step": 41011 }, { "epoch": 30.045421245421245, "grad_norm": 0.828125, "learning_rate": 6.327092778709725e-06, "loss": 0.1395, "step": 41012 }, { "epoch": 30.046153846153846, "grad_norm": 1.6953125, "learning_rate": 6.325546633177257e-06, "loss": 0.1219, "step": 41013 }, { "epoch": 30.046886446886447, "grad_norm": 0.78515625, "learning_rate": 6.324000699299089e-06, "loss": 0.0912, "step": 41014 }, { "epoch": 30.047619047619047, "grad_norm": 1.0234375, "learning_rate": 6.322454977082652e-06, "loss": 0.0997, "step": 41015 }, { "epoch": 30.048351648351648, "grad_norm": 0.94921875, "learning_rate": 6.32090946653532e-06, "loss": 0.1163, "step": 41016 }, { "epoch": 30.04908424908425, "grad_norm": 0.77734375, "learning_rate": 6.3193641676645305e-06, "loss": 0.0958, "step": 41017 }, { "epoch": 30.04981684981685, "grad_norm": 1.296875, "learning_rate": 6.317819080477645e-06, "loss": 0.126, "step": 41018 }, { "epoch": 30.05054945054945, "grad_norm": 1.5859375, "learning_rate": 6.316274204982113e-06, "loss": 0.135, "step": 41019 }, { "epoch": 30.05128205128205, "grad_norm": 0.73046875, "learning_rate": 6.3147295411853064e-06, "loss": 0.1043, "step": 41020 }, { "epoch": 30.052014652014652, "grad_norm": 0.96484375, "learning_rate": 6.313185089094637e-06, "loss": 0.1233, "step": 41021 }, { "epoch": 30.052747252747253, "grad_norm": 0.9765625, "learning_rate": 6.311640848717514e-06, "loss": 0.1246, "step": 41022 }, { "epoch": 30.053479853479853, "grad_norm": 1.234375, "learning_rate": 6.310096820061311e-06, "loss": 0.1425, "step": 41023 }, { "epoch": 30.054212454212454, "grad_norm": 1.0078125, "learning_rate": 6.308553003133444e-06, "loss": 0.1231, "step": 41024 }, { "epoch": 30.054945054945055, "grad_norm": 1.0859375, "learning_rate": 6.307009397941306e-06, "loss": 0.1378, "step": 41025 }, { "epoch": 30.055677655677655, "grad_norm": 0.97265625, "learning_rate": 6.305466004492306e-06, "loss": 0.1219, "step": 41026 }, { "epoch": 30.056410256410256, "grad_norm": 1.2578125, "learning_rate": 6.303922822793817e-06, "loss": 0.0943, "step": 41027 }, { "epoch": 30.057142857142857, "grad_norm": 0.91796875, "learning_rate": 6.302379852853238e-06, "loss": 0.1231, "step": 41028 }, { "epoch": 30.057875457875458, "grad_norm": 1.3828125, "learning_rate": 6.300837094677971e-06, "loss": 0.1154, "step": 41029 }, { "epoch": 30.05860805860806, "grad_norm": 1.2734375, "learning_rate": 6.299294548275407e-06, "loss": 0.1382, "step": 41030 }, { "epoch": 30.05934065934066, "grad_norm": 1.046875, "learning_rate": 6.297752213652927e-06, "loss": 0.1185, "step": 41031 }, { "epoch": 30.06007326007326, "grad_norm": 1.0078125, "learning_rate": 6.296210090817939e-06, "loss": 0.1046, "step": 41032 }, { "epoch": 30.06080586080586, "grad_norm": 0.859375, "learning_rate": 6.294668179777807e-06, "loss": 0.1226, "step": 41033 }, { "epoch": 30.06153846153846, "grad_norm": 0.63671875, "learning_rate": 6.2931264805399296e-06, "loss": 0.1109, "step": 41034 }, { "epoch": 30.062271062271062, "grad_norm": 0.89453125, "learning_rate": 6.291584993111696e-06, "loss": 0.1225, "step": 41035 }, { "epoch": 30.063003663003663, "grad_norm": 1.0234375, "learning_rate": 6.290043717500493e-06, "loss": 0.0943, "step": 41036 }, { "epoch": 30.063736263736264, "grad_norm": 1.7890625, "learning_rate": 6.288502653713713e-06, "loss": 0.1528, "step": 41037 }, { "epoch": 30.064468864468864, "grad_norm": 1.1328125, "learning_rate": 6.286961801758725e-06, "loss": 0.1086, "step": 41038 }, { "epoch": 30.065201465201465, "grad_norm": 1.015625, "learning_rate": 6.285421161642908e-06, "loss": 0.105, "step": 41039 }, { "epoch": 30.065934065934066, "grad_norm": 0.671875, "learning_rate": 6.283880733373658e-06, "loss": 0.0824, "step": 41040 }, { "epoch": 30.066666666666666, "grad_norm": 1.4921875, "learning_rate": 6.282340516958363e-06, "loss": 0.1544, "step": 41041 }, { "epoch": 30.067399267399267, "grad_norm": 0.94140625, "learning_rate": 6.280800512404376e-06, "loss": 0.1017, "step": 41042 }, { "epoch": 30.068131868131868, "grad_norm": 1.0703125, "learning_rate": 6.279260719719102e-06, "loss": 0.1415, "step": 41043 }, { "epoch": 30.06886446886447, "grad_norm": 1.046875, "learning_rate": 6.2777211389098865e-06, "loss": 0.1318, "step": 41044 }, { "epoch": 30.06959706959707, "grad_norm": 0.63671875, "learning_rate": 6.276181769984149e-06, "loss": 0.1003, "step": 41045 }, { "epoch": 30.07032967032967, "grad_norm": 0.8828125, "learning_rate": 6.274642612949226e-06, "loss": 0.0978, "step": 41046 }, { "epoch": 30.07106227106227, "grad_norm": 1.171875, "learning_rate": 6.27310366781251e-06, "loss": 0.1185, "step": 41047 }, { "epoch": 30.07179487179487, "grad_norm": 1.09375, "learning_rate": 6.271564934581388e-06, "loss": 0.1129, "step": 41048 }, { "epoch": 30.072527472527472, "grad_norm": 1.171875, "learning_rate": 6.270026413263193e-06, "loss": 0.1089, "step": 41049 }, { "epoch": 30.073260073260073, "grad_norm": 1.2578125, "learning_rate": 6.2684881038653415e-06, "loss": 0.1378, "step": 41050 }, { "epoch": 30.073992673992674, "grad_norm": 0.9453125, "learning_rate": 6.266950006395175e-06, "loss": 0.1156, "step": 41051 }, { "epoch": 30.074725274725274, "grad_norm": 0.98828125, "learning_rate": 6.265412120860086e-06, "loss": 0.1218, "step": 41052 }, { "epoch": 30.075457875457875, "grad_norm": 1.1328125, "learning_rate": 6.263874447267414e-06, "loss": 0.1315, "step": 41053 }, { "epoch": 30.076190476190476, "grad_norm": 0.83984375, "learning_rate": 6.262336985624542e-06, "loss": 0.0951, "step": 41054 }, { "epoch": 30.076923076923077, "grad_norm": 0.83203125, "learning_rate": 6.2607997359388325e-06, "loss": 0.1074, "step": 41055 }, { "epoch": 30.077655677655677, "grad_norm": 1.125, "learning_rate": 6.259262698217669e-06, "loss": 0.1187, "step": 41056 }, { "epoch": 30.078388278388278, "grad_norm": 0.69140625, "learning_rate": 6.25772587246839e-06, "loss": 0.1083, "step": 41057 }, { "epoch": 30.07912087912088, "grad_norm": 1.0859375, "learning_rate": 6.256189258698372e-06, "loss": 0.1235, "step": 41058 }, { "epoch": 30.07985347985348, "grad_norm": 1.1015625, "learning_rate": 6.254652856914979e-06, "loss": 0.1177, "step": 41059 }, { "epoch": 30.08058608058608, "grad_norm": 0.90625, "learning_rate": 6.253116667125565e-06, "loss": 0.1039, "step": 41060 }, { "epoch": 30.08131868131868, "grad_norm": 1.03125, "learning_rate": 6.251580689337494e-06, "loss": 0.1048, "step": 41061 }, { "epoch": 30.08205128205128, "grad_norm": 1.09375, "learning_rate": 6.2500449235581234e-06, "loss": 0.1277, "step": 41062 }, { "epoch": 30.082783882783882, "grad_norm": 1.8046875, "learning_rate": 6.2485093697948235e-06, "loss": 0.1107, "step": 41063 }, { "epoch": 30.083516483516483, "grad_norm": 1.1875, "learning_rate": 6.246974028054936e-06, "loss": 0.1076, "step": 41064 }, { "epoch": 30.084249084249084, "grad_norm": 1.09375, "learning_rate": 6.2454388983458205e-06, "loss": 0.1149, "step": 41065 }, { "epoch": 30.084981684981685, "grad_norm": 1.0625, "learning_rate": 6.2439039806748355e-06, "loss": 0.1489, "step": 41066 }, { "epoch": 30.085714285714285, "grad_norm": 1.1484375, "learning_rate": 6.242369275049344e-06, "loss": 0.145, "step": 41067 }, { "epoch": 30.086446886446886, "grad_norm": 0.76171875, "learning_rate": 6.240834781476679e-06, "loss": 0.1049, "step": 41068 }, { "epoch": 30.087179487179487, "grad_norm": 0.9765625, "learning_rate": 6.239300499964218e-06, "loss": 0.1203, "step": 41069 }, { "epoch": 30.087912087912088, "grad_norm": 1.359375, "learning_rate": 6.237766430519274e-06, "loss": 0.1399, "step": 41070 }, { "epoch": 30.08864468864469, "grad_norm": 0.96875, "learning_rate": 6.2362325731492404e-06, "loss": 0.0939, "step": 41071 }, { "epoch": 30.08937728937729, "grad_norm": 1.6015625, "learning_rate": 6.234698927861439e-06, "loss": 0.1098, "step": 41072 }, { "epoch": 30.09010989010989, "grad_norm": 1.046875, "learning_rate": 6.2331654946632305e-06, "loss": 0.109, "step": 41073 }, { "epoch": 30.09084249084249, "grad_norm": 1.125, "learning_rate": 6.231632273561964e-06, "loss": 0.0887, "step": 41074 }, { "epoch": 30.09157509157509, "grad_norm": 0.65234375, "learning_rate": 6.230099264564962e-06, "loss": 0.1054, "step": 41075 }, { "epoch": 30.092307692307692, "grad_norm": 0.69140625, "learning_rate": 6.2285664676796054e-06, "loss": 0.1041, "step": 41076 }, { "epoch": 30.093040293040293, "grad_norm": 1.1171875, "learning_rate": 6.227033882913208e-06, "loss": 0.1464, "step": 41077 }, { "epoch": 30.093772893772893, "grad_norm": 0.94921875, "learning_rate": 6.225501510273139e-06, "loss": 0.1582, "step": 41078 }, { "epoch": 30.094505494505494, "grad_norm": 0.984375, "learning_rate": 6.2239693497667145e-06, "loss": 0.0921, "step": 41079 }, { "epoch": 30.095238095238095, "grad_norm": 1.328125, "learning_rate": 6.2224374014013026e-06, "loss": 0.1429, "step": 41080 }, { "epoch": 30.095970695970696, "grad_norm": 0.74609375, "learning_rate": 6.220905665184204e-06, "loss": 0.1133, "step": 41081 }, { "epoch": 30.096703296703296, "grad_norm": 0.63671875, "learning_rate": 6.219374141122801e-06, "loss": 0.0834, "step": 41082 }, { "epoch": 30.097435897435897, "grad_norm": 0.734375, "learning_rate": 6.217842829224405e-06, "loss": 0.134, "step": 41083 }, { "epoch": 30.098168498168498, "grad_norm": 0.6796875, "learning_rate": 6.216311729496363e-06, "loss": 0.1087, "step": 41084 }, { "epoch": 30.0989010989011, "grad_norm": 1.0078125, "learning_rate": 6.214780841946012e-06, "loss": 0.1273, "step": 41085 }, { "epoch": 30.0996336996337, "grad_norm": 1.21875, "learning_rate": 6.213250166580678e-06, "loss": 0.1294, "step": 41086 }, { "epoch": 30.1003663003663, "grad_norm": 0.7421875, "learning_rate": 6.211719703407702e-06, "loss": 0.0929, "step": 41087 }, { "epoch": 30.1010989010989, "grad_norm": 0.7734375, "learning_rate": 6.2101894524344175e-06, "loss": 0.1091, "step": 41088 }, { "epoch": 30.1018315018315, "grad_norm": 0.88671875, "learning_rate": 6.208659413668164e-06, "loss": 0.107, "step": 41089 }, { "epoch": 30.102564102564102, "grad_norm": 1.234375, "learning_rate": 6.20712958711625e-06, "loss": 0.1207, "step": 41090 }, { "epoch": 30.103296703296703, "grad_norm": 0.7734375, "learning_rate": 6.205599972786015e-06, "loss": 0.107, "step": 41091 }, { "epoch": 30.104029304029304, "grad_norm": 1.1171875, "learning_rate": 6.2040705706848e-06, "loss": 0.1322, "step": 41092 }, { "epoch": 30.104761904761904, "grad_norm": 1.9140625, "learning_rate": 6.202541380819931e-06, "loss": 0.1156, "step": 41093 }, { "epoch": 30.105494505494505, "grad_norm": 0.66015625, "learning_rate": 6.201012403198716e-06, "loss": 0.0919, "step": 41094 }, { "epoch": 30.106227106227106, "grad_norm": 1.3515625, "learning_rate": 6.1994836378285045e-06, "loss": 0.1539, "step": 41095 }, { "epoch": 30.106959706959707, "grad_norm": 0.89453125, "learning_rate": 6.1979550847165915e-06, "loss": 0.1069, "step": 41096 }, { "epoch": 30.107692307692307, "grad_norm": 1.25, "learning_rate": 6.196426743870339e-06, "loss": 0.1148, "step": 41097 }, { "epoch": 30.108424908424908, "grad_norm": 1.6171875, "learning_rate": 6.194898615297037e-06, "loss": 0.1417, "step": 41098 }, { "epoch": 30.10915750915751, "grad_norm": 1.1640625, "learning_rate": 6.193370699004018e-06, "loss": 0.1329, "step": 41099 }, { "epoch": 30.10989010989011, "grad_norm": 1.0234375, "learning_rate": 6.1918429949986184e-06, "loss": 0.1125, "step": 41100 }, { "epoch": 30.11062271062271, "grad_norm": 0.71875, "learning_rate": 6.1903155032881274e-06, "loss": 0.1231, "step": 41101 }, { "epoch": 30.11135531135531, "grad_norm": 0.98828125, "learning_rate": 6.1887882238799005e-06, "loss": 0.1042, "step": 41102 }, { "epoch": 30.11208791208791, "grad_norm": 1.171875, "learning_rate": 6.1872611567812195e-06, "loss": 0.1279, "step": 41103 }, { "epoch": 30.112820512820512, "grad_norm": 1.5234375, "learning_rate": 6.185734301999428e-06, "loss": 0.1133, "step": 41104 }, { "epoch": 30.113553113553113, "grad_norm": 0.80859375, "learning_rate": 6.184207659541824e-06, "loss": 0.1088, "step": 41105 }, { "epoch": 30.114285714285714, "grad_norm": 1.109375, "learning_rate": 6.182681229415739e-06, "loss": 0.1369, "step": 41106 }, { "epoch": 30.115018315018315, "grad_norm": 1.0078125, "learning_rate": 6.181155011628459e-06, "loss": 0.1237, "step": 41107 }, { "epoch": 30.115750915750915, "grad_norm": 1.4453125, "learning_rate": 6.179629006187332e-06, "loss": 0.125, "step": 41108 }, { "epoch": 30.116483516483516, "grad_norm": 1.0625, "learning_rate": 6.17810321309964e-06, "loss": 0.1449, "step": 41109 }, { "epoch": 30.117216117216117, "grad_norm": 0.90234375, "learning_rate": 6.1765776323727085e-06, "loss": 0.1178, "step": 41110 }, { "epoch": 30.117948717948718, "grad_norm": 1.0703125, "learning_rate": 6.1750522640138516e-06, "loss": 0.1376, "step": 41111 }, { "epoch": 30.11868131868132, "grad_norm": 1.09375, "learning_rate": 6.1735271080303584e-06, "loss": 0.0949, "step": 41112 }, { "epoch": 30.11941391941392, "grad_norm": 1.0390625, "learning_rate": 6.172002164429553e-06, "loss": 0.122, "step": 41113 }, { "epoch": 30.12014652014652, "grad_norm": 1.1015625, "learning_rate": 6.170477433218737e-06, "loss": 0.1361, "step": 41114 }, { "epoch": 30.12087912087912, "grad_norm": 0.69140625, "learning_rate": 6.168952914405224e-06, "loss": 0.1217, "step": 41115 }, { "epoch": 30.12161172161172, "grad_norm": 0.9765625, "learning_rate": 6.167428607996298e-06, "loss": 0.099, "step": 41116 }, { "epoch": 30.122344322344322, "grad_norm": 1.09375, "learning_rate": 6.16590451399928e-06, "loss": 0.1327, "step": 41117 }, { "epoch": 30.123076923076923, "grad_norm": 0.91796875, "learning_rate": 6.164380632421462e-06, "loss": 0.1301, "step": 41118 }, { "epoch": 30.123809523809523, "grad_norm": 1.140625, "learning_rate": 6.162856963270167e-06, "loss": 0.1382, "step": 41119 }, { "epoch": 30.124542124542124, "grad_norm": 1.15625, "learning_rate": 6.161333506552661e-06, "loss": 0.0849, "step": 41120 }, { "epoch": 30.125274725274725, "grad_norm": 0.85546875, "learning_rate": 6.159810262276279e-06, "loss": 0.1455, "step": 41121 }, { "epoch": 30.126007326007326, "grad_norm": 0.69921875, "learning_rate": 6.15828723044828e-06, "loss": 0.1004, "step": 41122 }, { "epoch": 30.126739926739926, "grad_norm": 0.56640625, "learning_rate": 6.156764411075999e-06, "loss": 0.0917, "step": 41123 }, { "epoch": 30.127472527472527, "grad_norm": 1.0390625, "learning_rate": 6.155241804166714e-06, "loss": 0.1033, "step": 41124 }, { "epoch": 30.128205128205128, "grad_norm": 0.78515625, "learning_rate": 6.153719409727716e-06, "loss": 0.121, "step": 41125 }, { "epoch": 30.12893772893773, "grad_norm": 0.859375, "learning_rate": 6.152197227766321e-06, "loss": 0.1091, "step": 41126 }, { "epoch": 30.12967032967033, "grad_norm": 0.9765625, "learning_rate": 6.150675258289789e-06, "loss": 0.1015, "step": 41127 }, { "epoch": 30.13040293040293, "grad_norm": 1.1328125, "learning_rate": 6.149153501305449e-06, "loss": 0.1324, "step": 41128 }, { "epoch": 30.13113553113553, "grad_norm": 0.6796875, "learning_rate": 6.147631956820563e-06, "loss": 0.1228, "step": 41129 }, { "epoch": 30.13186813186813, "grad_norm": 1.1328125, "learning_rate": 6.1461106248424385e-06, "loss": 0.1296, "step": 41130 }, { "epoch": 30.132600732600732, "grad_norm": 1.2109375, "learning_rate": 6.144589505378353e-06, "loss": 0.1109, "step": 41131 }, { "epoch": 30.133333333333333, "grad_norm": 1.09375, "learning_rate": 6.143068598435608e-06, "loss": 0.1187, "step": 41132 }, { "epoch": 30.134065934065934, "grad_norm": 1.0078125, "learning_rate": 6.141547904021466e-06, "loss": 0.1065, "step": 41133 }, { "epoch": 30.134798534798534, "grad_norm": 0.890625, "learning_rate": 6.140027422143246e-06, "loss": 0.1338, "step": 41134 }, { "epoch": 30.135531135531135, "grad_norm": 1.2890625, "learning_rate": 6.138507152808207e-06, "loss": 0.1058, "step": 41135 }, { "epoch": 30.136263736263736, "grad_norm": 0.86328125, "learning_rate": 6.136987096023648e-06, "loss": 0.122, "step": 41136 }, { "epoch": 30.136996336996337, "grad_norm": 1.5859375, "learning_rate": 6.135467251796856e-06, "loss": 0.1122, "step": 41137 }, { "epoch": 30.137728937728937, "grad_norm": 0.96875, "learning_rate": 6.13394762013509e-06, "loss": 0.1185, "step": 41138 }, { "epoch": 30.138461538461538, "grad_norm": 0.9609375, "learning_rate": 6.132428201045649e-06, "loss": 0.1337, "step": 41139 }, { "epoch": 30.13919413919414, "grad_norm": 0.96484375, "learning_rate": 6.13090899453581e-06, "loss": 0.1069, "step": 41140 }, { "epoch": 30.13992673992674, "grad_norm": 1.2421875, "learning_rate": 6.129390000612858e-06, "loss": 0.1548, "step": 41141 }, { "epoch": 30.14065934065934, "grad_norm": 0.74609375, "learning_rate": 6.1278712192840535e-06, "loss": 0.097, "step": 41142 }, { "epoch": 30.14139194139194, "grad_norm": 0.8515625, "learning_rate": 6.126352650556691e-06, "loss": 0.1321, "step": 41143 }, { "epoch": 30.14212454212454, "grad_norm": 1.0703125, "learning_rate": 6.124834294438032e-06, "loss": 0.1161, "step": 41144 }, { "epoch": 30.142857142857142, "grad_norm": 1.046875, "learning_rate": 6.1233161509353774e-06, "loss": 0.1055, "step": 41145 }, { "epoch": 30.143589743589743, "grad_norm": 1.1953125, "learning_rate": 6.121798220055966e-06, "loss": 0.1211, "step": 41146 }, { "epoch": 30.144322344322344, "grad_norm": 0.81640625, "learning_rate": 6.120280501807103e-06, "loss": 0.1016, "step": 41147 }, { "epoch": 30.145054945054945, "grad_norm": 1.40625, "learning_rate": 6.118762996196024e-06, "loss": 0.1184, "step": 41148 }, { "epoch": 30.145787545787545, "grad_norm": 0.8984375, "learning_rate": 6.117245703230037e-06, "loss": 0.107, "step": 41149 }, { "epoch": 30.146520146520146, "grad_norm": 1.2734375, "learning_rate": 6.115728622916392e-06, "loss": 0.1302, "step": 41150 }, { "epoch": 30.147252747252747, "grad_norm": 0.85546875, "learning_rate": 6.114211755262357e-06, "loss": 0.111, "step": 41151 }, { "epoch": 30.147985347985347, "grad_norm": 1.203125, "learning_rate": 6.112695100275215e-06, "loss": 0.1373, "step": 41152 }, { "epoch": 30.148717948717948, "grad_norm": 0.984375, "learning_rate": 6.111178657962203e-06, "loss": 0.1266, "step": 41153 }, { "epoch": 30.14945054945055, "grad_norm": 1.328125, "learning_rate": 6.109662428330621e-06, "loss": 0.1199, "step": 41154 }, { "epoch": 30.15018315018315, "grad_norm": 1.7890625, "learning_rate": 6.108146411387717e-06, "loss": 0.126, "step": 41155 }, { "epoch": 30.15091575091575, "grad_norm": 0.8359375, "learning_rate": 6.106630607140762e-06, "loss": 0.0966, "step": 41156 }, { "epoch": 30.15164835164835, "grad_norm": 1.9296875, "learning_rate": 6.105115015597003e-06, "loss": 0.092, "step": 41157 }, { "epoch": 30.152380952380952, "grad_norm": 1.265625, "learning_rate": 6.1035996367637224e-06, "loss": 0.1145, "step": 41158 }, { "epoch": 30.153113553113553, "grad_norm": 0.8828125, "learning_rate": 6.102084470648153e-06, "loss": 0.1126, "step": 41159 }, { "epoch": 30.153846153846153, "grad_norm": 0.84765625, "learning_rate": 6.100569517257587e-06, "loss": 0.1189, "step": 41160 }, { "epoch": 30.154578754578754, "grad_norm": 0.796875, "learning_rate": 6.099054776599258e-06, "loss": 0.1135, "step": 41161 }, { "epoch": 30.155311355311355, "grad_norm": 1.4140625, "learning_rate": 6.09754024868044e-06, "loss": 0.1177, "step": 41162 }, { "epoch": 30.156043956043955, "grad_norm": 0.9609375, "learning_rate": 6.096025933508388e-06, "loss": 0.1344, "step": 41163 }, { "epoch": 30.156776556776556, "grad_norm": 0.78515625, "learning_rate": 6.094511831090341e-06, "loss": 0.101, "step": 41164 }, { "epoch": 30.157509157509157, "grad_norm": 0.94140625, "learning_rate": 6.092997941433567e-06, "loss": 0.1089, "step": 41165 }, { "epoch": 30.158241758241758, "grad_norm": 0.84765625, "learning_rate": 6.091484264545318e-06, "loss": 0.1282, "step": 41166 }, { "epoch": 30.15897435897436, "grad_norm": 0.64453125, "learning_rate": 6.089970800432857e-06, "loss": 0.1021, "step": 41167 }, { "epoch": 30.15970695970696, "grad_norm": 1.125, "learning_rate": 6.088457549103415e-06, "loss": 0.1076, "step": 41168 }, { "epoch": 30.16043956043956, "grad_norm": 0.8359375, "learning_rate": 6.086944510564251e-06, "loss": 0.0908, "step": 41169 }, { "epoch": 30.16117216117216, "grad_norm": 0.7109375, "learning_rate": 6.085431684822613e-06, "loss": 0.1085, "step": 41170 }, { "epoch": 30.16190476190476, "grad_norm": 0.6875, "learning_rate": 6.083919071885766e-06, "loss": 0.115, "step": 41171 }, { "epoch": 30.162637362637362, "grad_norm": 0.79296875, "learning_rate": 6.0824066717609335e-06, "loss": 0.1229, "step": 41172 }, { "epoch": 30.163369963369963, "grad_norm": 1.046875, "learning_rate": 6.080894484455371e-06, "loss": 0.114, "step": 41173 }, { "epoch": 30.164102564102564, "grad_norm": 0.90625, "learning_rate": 6.079382509976326e-06, "loss": 0.1289, "step": 41174 }, { "epoch": 30.164835164835164, "grad_norm": 1.15625, "learning_rate": 6.077870748331049e-06, "loss": 0.1735, "step": 41175 }, { "epoch": 30.165567765567765, "grad_norm": 1.6484375, "learning_rate": 6.076359199526772e-06, "loss": 0.1062, "step": 41176 }, { "epoch": 30.166300366300366, "grad_norm": 1.2734375, "learning_rate": 6.074847863570737e-06, "loss": 0.1014, "step": 41177 }, { "epoch": 30.167032967032966, "grad_norm": 1.0859375, "learning_rate": 6.073336740470198e-06, "loss": 0.146, "step": 41178 }, { "epoch": 30.167765567765567, "grad_norm": 1.5859375, "learning_rate": 6.071825830232372e-06, "loss": 0.1267, "step": 41179 }, { "epoch": 30.168498168498168, "grad_norm": 0.82421875, "learning_rate": 6.070315132864529e-06, "loss": 0.0966, "step": 41180 }, { "epoch": 30.16923076923077, "grad_norm": 0.6640625, "learning_rate": 6.06880464837388e-06, "loss": 0.1318, "step": 41181 }, { "epoch": 30.16996336996337, "grad_norm": 1.0546875, "learning_rate": 6.067294376767689e-06, "loss": 0.1016, "step": 41182 }, { "epoch": 30.17069597069597, "grad_norm": 1.3828125, "learning_rate": 6.065784318053161e-06, "loss": 0.1082, "step": 41183 }, { "epoch": 30.17142857142857, "grad_norm": 1.4609375, "learning_rate": 6.064274472237562e-06, "loss": 0.1452, "step": 41184 }, { "epoch": 30.17216117216117, "grad_norm": 1.0078125, "learning_rate": 6.062764839328092e-06, "loss": 0.1121, "step": 41185 }, { "epoch": 30.172893772893772, "grad_norm": 1.640625, "learning_rate": 6.061255419332023e-06, "loss": 0.1454, "step": 41186 }, { "epoch": 30.173626373626373, "grad_norm": 1.234375, "learning_rate": 6.059746212256554e-06, "loss": 0.133, "step": 41187 }, { "epoch": 30.174358974358974, "grad_norm": 1.1953125, "learning_rate": 6.058237218108929e-06, "loss": 0.1411, "step": 41188 }, { "epoch": 30.175091575091574, "grad_norm": 1.0546875, "learning_rate": 6.056728436896396e-06, "loss": 0.1227, "step": 41189 }, { "epoch": 30.175824175824175, "grad_norm": 1.0234375, "learning_rate": 6.05521986862614e-06, "loss": 0.1002, "step": 41190 }, { "epoch": 30.176556776556776, "grad_norm": 0.98828125, "learning_rate": 6.0537115133054374e-06, "loss": 0.1071, "step": 41191 }, { "epoch": 30.177289377289377, "grad_norm": 0.9375, "learning_rate": 6.0522033709414835e-06, "loss": 0.124, "step": 41192 }, { "epoch": 30.178021978021977, "grad_norm": 1.0859375, "learning_rate": 6.0506954415415245e-06, "loss": 0.1252, "step": 41193 }, { "epoch": 30.178754578754578, "grad_norm": 1.0390625, "learning_rate": 6.049187725112769e-06, "loss": 0.1316, "step": 41194 }, { "epoch": 30.17948717948718, "grad_norm": 1.046875, "learning_rate": 6.047680221662446e-06, "loss": 0.1148, "step": 41195 }, { "epoch": 30.18021978021978, "grad_norm": 0.94140625, "learning_rate": 6.046172931197777e-06, "loss": 0.1376, "step": 41196 }, { "epoch": 30.18095238095238, "grad_norm": 0.59375, "learning_rate": 6.044665853726e-06, "loss": 0.1183, "step": 41197 }, { "epoch": 30.18168498168498, "grad_norm": 1.46875, "learning_rate": 6.043158989254309e-06, "loss": 0.1356, "step": 41198 }, { "epoch": 30.18241758241758, "grad_norm": 1.125, "learning_rate": 6.041652337789938e-06, "loss": 0.1373, "step": 41199 }, { "epoch": 30.183150183150182, "grad_norm": 0.96484375, "learning_rate": 6.0401458993401106e-06, "loss": 0.1184, "step": 41200 }, { "epoch": 30.183882783882783, "grad_norm": 1.0390625, "learning_rate": 6.038639673912045e-06, "loss": 0.0945, "step": 41201 }, { "epoch": 30.184615384615384, "grad_norm": 1.0390625, "learning_rate": 6.037133661512943e-06, "loss": 0.1344, "step": 41202 }, { "epoch": 30.185347985347985, "grad_norm": 1.3046875, "learning_rate": 6.035627862150028e-06, "loss": 0.1262, "step": 41203 }, { "epoch": 30.186080586080585, "grad_norm": 0.7734375, "learning_rate": 6.0341222758305285e-06, "loss": 0.1113, "step": 41204 }, { "epoch": 30.186813186813186, "grad_norm": 1.171875, "learning_rate": 6.032616902561624e-06, "loss": 0.121, "step": 41205 }, { "epoch": 30.187545787545787, "grad_norm": 0.98828125, "learning_rate": 6.031111742350574e-06, "loss": 0.1278, "step": 41206 }, { "epoch": 30.188278388278388, "grad_norm": 1.1328125, "learning_rate": 6.029606795204553e-06, "loss": 0.1259, "step": 41207 }, { "epoch": 30.18901098901099, "grad_norm": 0.9453125, "learning_rate": 6.028102061130797e-06, "loss": 0.1276, "step": 41208 }, { "epoch": 30.18974358974359, "grad_norm": 1.234375, "learning_rate": 6.02659754013649e-06, "loss": 0.1176, "step": 41209 }, { "epoch": 30.19047619047619, "grad_norm": 0.71484375, "learning_rate": 6.0250932322288625e-06, "loss": 0.1132, "step": 41210 }, { "epoch": 30.19120879120879, "grad_norm": 1.15625, "learning_rate": 6.023589137415098e-06, "loss": 0.1395, "step": 41211 }, { "epoch": 30.19194139194139, "grad_norm": 1.046875, "learning_rate": 6.022085255702439e-06, "loss": 0.1267, "step": 41212 }, { "epoch": 30.192673992673992, "grad_norm": 1.453125, "learning_rate": 6.020581587098059e-06, "loss": 0.1104, "step": 41213 }, { "epoch": 30.193406593406593, "grad_norm": 1.15625, "learning_rate": 6.0190781316091735e-06, "loss": 0.1105, "step": 41214 }, { "epoch": 30.194139194139193, "grad_norm": 0.98828125, "learning_rate": 6.0175748892429965e-06, "loss": 0.0908, "step": 41215 }, { "epoch": 30.194871794871794, "grad_norm": 1.046875, "learning_rate": 6.016071860006702e-06, "loss": 0.1189, "step": 41216 }, { "epoch": 30.195604395604395, "grad_norm": 0.98828125, "learning_rate": 6.014569043907529e-06, "loss": 0.1046, "step": 41217 }, { "epoch": 30.196336996336996, "grad_norm": 1.421875, "learning_rate": 6.013066440952649e-06, "loss": 0.1366, "step": 41218 }, { "epoch": 30.197069597069596, "grad_norm": 1.1171875, "learning_rate": 6.0115640511492806e-06, "loss": 0.1168, "step": 41219 }, { "epoch": 30.197802197802197, "grad_norm": 1.265625, "learning_rate": 6.010061874504606e-06, "loss": 0.104, "step": 41220 }, { "epoch": 30.198534798534798, "grad_norm": 0.98828125, "learning_rate": 6.008559911025832e-06, "loss": 0.1023, "step": 41221 }, { "epoch": 30.1992673992674, "grad_norm": 0.90234375, "learning_rate": 6.007058160720151e-06, "loss": 0.1195, "step": 41222 }, { "epoch": 30.2, "grad_norm": 0.62890625, "learning_rate": 6.0055566235947675e-06, "loss": 0.1034, "step": 41223 }, { "epoch": 30.2007326007326, "grad_norm": 0.859375, "learning_rate": 6.004055299656864e-06, "loss": 0.0911, "step": 41224 }, { "epoch": 30.2014652014652, "grad_norm": 0.63671875, "learning_rate": 6.002554188913637e-06, "loss": 0.1135, "step": 41225 }, { "epoch": 30.2021978021978, "grad_norm": 1.1171875, "learning_rate": 6.001053291372277e-06, "loss": 0.1096, "step": 41226 }, { "epoch": 30.202930402930402, "grad_norm": 0.7890625, "learning_rate": 5.999552607039989e-06, "loss": 0.1173, "step": 41227 }, { "epoch": 30.203663003663003, "grad_norm": 0.73828125, "learning_rate": 5.998052135923947e-06, "loss": 0.1397, "step": 41228 }, { "epoch": 30.204395604395604, "grad_norm": 1.1796875, "learning_rate": 5.996551878031349e-06, "loss": 0.1137, "step": 41229 }, { "epoch": 30.205128205128204, "grad_norm": 0.7265625, "learning_rate": 5.995051833369382e-06, "loss": 0.1084, "step": 41230 }, { "epoch": 30.205860805860805, "grad_norm": 0.703125, "learning_rate": 5.993552001945215e-06, "loss": 0.1232, "step": 41231 }, { "epoch": 30.206593406593406, "grad_norm": 0.80859375, "learning_rate": 5.992052383766071e-06, "loss": 0.1175, "step": 41232 }, { "epoch": 30.207326007326007, "grad_norm": 0.8515625, "learning_rate": 5.990552978839109e-06, "loss": 0.1227, "step": 41233 }, { "epoch": 30.208058608058607, "grad_norm": 0.9609375, "learning_rate": 5.989053787171525e-06, "loss": 0.0957, "step": 41234 }, { "epoch": 30.208791208791208, "grad_norm": 1.140625, "learning_rate": 5.987554808770489e-06, "loss": 0.12, "step": 41235 }, { "epoch": 30.20952380952381, "grad_norm": 1.1171875, "learning_rate": 5.986056043643197e-06, "loss": 0.1404, "step": 41236 }, { "epoch": 30.21025641025641, "grad_norm": 1.4609375, "learning_rate": 5.984557491796807e-06, "loss": 0.0897, "step": 41237 }, { "epoch": 30.21098901098901, "grad_norm": 1.109375, "learning_rate": 5.983059153238531e-06, "loss": 0.1195, "step": 41238 }, { "epoch": 30.21172161172161, "grad_norm": 1.15625, "learning_rate": 5.981561027975522e-06, "loss": 0.1108, "step": 41239 }, { "epoch": 30.21245421245421, "grad_norm": 2.21875, "learning_rate": 5.980063116014972e-06, "loss": 0.132, "step": 41240 }, { "epoch": 30.213186813186812, "grad_norm": 1.0625, "learning_rate": 5.978565417364065e-06, "loss": 0.146, "step": 41241 }, { "epoch": 30.213919413919413, "grad_norm": 1.2109375, "learning_rate": 5.977067932029945e-06, "loss": 0.1203, "step": 41242 }, { "epoch": 30.214652014652014, "grad_norm": 1.1875, "learning_rate": 5.975570660019833e-06, "loss": 0.1294, "step": 41243 }, { "epoch": 30.215384615384615, "grad_norm": 0.58203125, "learning_rate": 5.974073601340862e-06, "loss": 0.1184, "step": 41244 }, { "epoch": 30.216117216117215, "grad_norm": 1.2109375, "learning_rate": 5.972576756000231e-06, "loss": 0.0983, "step": 41245 }, { "epoch": 30.216849816849816, "grad_norm": 1.140625, "learning_rate": 5.971080124005096e-06, "loss": 0.1443, "step": 41246 }, { "epoch": 30.217582417582417, "grad_norm": 1.3359375, "learning_rate": 5.969583705362634e-06, "loss": 0.1075, "step": 41247 }, { "epoch": 30.218315018315018, "grad_norm": 1.0234375, "learning_rate": 5.968087500080012e-06, "loss": 0.1191, "step": 41248 }, { "epoch": 30.21904761904762, "grad_norm": 0.51171875, "learning_rate": 5.966591508164417e-06, "loss": 0.1062, "step": 41249 }, { "epoch": 30.21978021978022, "grad_norm": 1.359375, "learning_rate": 5.9650957296229884e-06, "loss": 0.1473, "step": 41250 }, { "epoch": 30.22051282051282, "grad_norm": 1.0234375, "learning_rate": 5.963600164462903e-06, "loss": 0.119, "step": 41251 }, { "epoch": 30.22124542124542, "grad_norm": 1.5546875, "learning_rate": 5.962104812691332e-06, "loss": 0.1375, "step": 41252 }, { "epoch": 30.22197802197802, "grad_norm": 1.5078125, "learning_rate": 5.9606096743154455e-06, "loss": 0.1383, "step": 41253 }, { "epoch": 30.222710622710622, "grad_norm": 1.2734375, "learning_rate": 5.959114749342394e-06, "loss": 0.1025, "step": 41254 }, { "epoch": 30.223443223443223, "grad_norm": 0.86328125, "learning_rate": 5.957620037779341e-06, "loss": 0.1341, "step": 41255 }, { "epoch": 30.224175824175823, "grad_norm": 1.3359375, "learning_rate": 5.956125539633467e-06, "loss": 0.0957, "step": 41256 }, { "epoch": 30.224908424908424, "grad_norm": 1.140625, "learning_rate": 5.954631254911897e-06, "loss": 0.1344, "step": 41257 }, { "epoch": 30.225641025641025, "grad_norm": 0.9609375, "learning_rate": 5.9531371836218294e-06, "loss": 0.1318, "step": 41258 }, { "epoch": 30.226373626373626, "grad_norm": 0.7421875, "learning_rate": 5.951643325770395e-06, "loss": 0.1115, "step": 41259 }, { "epoch": 30.227106227106226, "grad_norm": 0.6640625, "learning_rate": 5.950149681364775e-06, "loss": 0.1204, "step": 41260 }, { "epoch": 30.227838827838827, "grad_norm": 1.5859375, "learning_rate": 5.948656250412101e-06, "loss": 0.1102, "step": 41261 }, { "epoch": 30.228571428571428, "grad_norm": 1.6015625, "learning_rate": 5.947163032919554e-06, "loss": 0.1569, "step": 41262 }, { "epoch": 30.22930402930403, "grad_norm": 1.0703125, "learning_rate": 5.945670028894249e-06, "loss": 0.1014, "step": 41263 }, { "epoch": 30.23003663003663, "grad_norm": 0.6953125, "learning_rate": 5.944177238343386e-06, "loss": 0.1148, "step": 41264 }, { "epoch": 30.23076923076923, "grad_norm": 1.0078125, "learning_rate": 5.942684661274092e-06, "loss": 0.1018, "step": 41265 }, { "epoch": 30.23150183150183, "grad_norm": 1.546875, "learning_rate": 5.9411922976935195e-06, "loss": 0.1124, "step": 41266 }, { "epoch": 30.23223443223443, "grad_norm": 1.1328125, "learning_rate": 5.93970014760883e-06, "loss": 0.1511, "step": 41267 }, { "epoch": 30.232967032967032, "grad_norm": 0.87890625, "learning_rate": 5.938208211027145e-06, "loss": 0.1176, "step": 41268 }, { "epoch": 30.233699633699633, "grad_norm": 0.8515625, "learning_rate": 5.936716487955654e-06, "loss": 0.1022, "step": 41269 }, { "epoch": 30.234432234432234, "grad_norm": 1.4921875, "learning_rate": 5.935224978401474e-06, "loss": 0.1572, "step": 41270 }, { "epoch": 30.235164835164834, "grad_norm": 0.84765625, "learning_rate": 5.933733682371775e-06, "loss": 0.1175, "step": 41271 }, { "epoch": 30.235897435897435, "grad_norm": 1.453125, "learning_rate": 5.93224259987367e-06, "loss": 0.1276, "step": 41272 }, { "epoch": 30.236630036630036, "grad_norm": 1.421875, "learning_rate": 5.9307517309143315e-06, "loss": 0.1312, "step": 41273 }, { "epoch": 30.237362637362637, "grad_norm": 1.03125, "learning_rate": 5.929261075500885e-06, "loss": 0.1331, "step": 41274 }, { "epoch": 30.238095238095237, "grad_norm": 0.55859375, "learning_rate": 5.927770633640491e-06, "loss": 0.1066, "step": 41275 }, { "epoch": 30.238827838827838, "grad_norm": 1.3046875, "learning_rate": 5.926280405340275e-06, "loss": 0.1268, "step": 41276 }, { "epoch": 30.23956043956044, "grad_norm": 1.0234375, "learning_rate": 5.924790390607378e-06, "loss": 0.1254, "step": 41277 }, { "epoch": 30.24029304029304, "grad_norm": 1.4296875, "learning_rate": 5.923300589448941e-06, "loss": 0.1213, "step": 41278 }, { "epoch": 30.24102564102564, "grad_norm": 0.69921875, "learning_rate": 5.921811001872117e-06, "loss": 0.1077, "step": 41279 }, { "epoch": 30.24175824175824, "grad_norm": 0.8125, "learning_rate": 5.920321627884019e-06, "loss": 0.1077, "step": 41280 }, { "epoch": 30.24249084249084, "grad_norm": 0.78515625, "learning_rate": 5.918832467491798e-06, "loss": 0.1102, "step": 41281 }, { "epoch": 30.243223443223442, "grad_norm": 1.2578125, "learning_rate": 5.917343520702595e-06, "loss": 0.13, "step": 41282 }, { "epoch": 30.243956043956043, "grad_norm": 0.9296875, "learning_rate": 5.915854787523513e-06, "loss": 0.124, "step": 41283 }, { "epoch": 30.244688644688644, "grad_norm": 1.078125, "learning_rate": 5.914366267961728e-06, "loss": 0.1258, "step": 41284 }, { "epoch": 30.245421245421245, "grad_norm": 0.6953125, "learning_rate": 5.91287796202434e-06, "loss": 0.1227, "step": 41285 }, { "epoch": 30.246153846153845, "grad_norm": 1.0234375, "learning_rate": 5.911389869718502e-06, "loss": 0.1441, "step": 41286 }, { "epoch": 30.246886446886446, "grad_norm": 1.0625, "learning_rate": 5.909901991051321e-06, "loss": 0.128, "step": 41287 }, { "epoch": 30.247619047619047, "grad_norm": 1.1015625, "learning_rate": 5.908414326029937e-06, "loss": 0.1154, "step": 41288 }, { "epoch": 30.248351648351647, "grad_norm": 1.2734375, "learning_rate": 5.906926874661479e-06, "loss": 0.1211, "step": 41289 }, { "epoch": 30.249084249084248, "grad_norm": 0.93359375, "learning_rate": 5.905439636953086e-06, "loss": 0.0998, "step": 41290 }, { "epoch": 30.24981684981685, "grad_norm": 0.91015625, "learning_rate": 5.903952612911856e-06, "loss": 0.1069, "step": 41291 }, { "epoch": 30.25054945054945, "grad_norm": 0.99609375, "learning_rate": 5.902465802544936e-06, "loss": 0.1026, "step": 41292 }, { "epoch": 30.25128205128205, "grad_norm": 0.9609375, "learning_rate": 5.900979205859446e-06, "loss": 0.1383, "step": 41293 }, { "epoch": 30.25201465201465, "grad_norm": 0.83203125, "learning_rate": 5.8994928228624875e-06, "loss": 0.1168, "step": 41294 }, { "epoch": 30.252747252747252, "grad_norm": 0.78515625, "learning_rate": 5.8980066535612224e-06, "loss": 0.1022, "step": 41295 }, { "epoch": 30.253479853479853, "grad_norm": 0.89453125, "learning_rate": 5.896520697962737e-06, "loss": 0.1393, "step": 41296 }, { "epoch": 30.254212454212453, "grad_norm": 1.234375, "learning_rate": 5.8950349560741735e-06, "loss": 0.1087, "step": 41297 }, { "epoch": 30.254945054945054, "grad_norm": 1.71875, "learning_rate": 5.89354942790263e-06, "loss": 0.1137, "step": 41298 }, { "epoch": 30.255677655677655, "grad_norm": 0.76953125, "learning_rate": 5.892064113455239e-06, "loss": 0.1255, "step": 41299 }, { "epoch": 30.256410256410255, "grad_norm": 1.140625, "learning_rate": 5.89057901273911e-06, "loss": 0.1376, "step": 41300 }, { "epoch": 30.257142857142856, "grad_norm": 0.8671875, "learning_rate": 5.889094125761367e-06, "loss": 0.0915, "step": 41301 }, { "epoch": 30.257875457875457, "grad_norm": 0.73046875, "learning_rate": 5.887609452529112e-06, "loss": 0.124, "step": 41302 }, { "epoch": 30.258608058608058, "grad_norm": 1.453125, "learning_rate": 5.886124993049467e-06, "loss": 0.1304, "step": 41303 }, { "epoch": 30.25934065934066, "grad_norm": 0.6015625, "learning_rate": 5.884640747329543e-06, "loss": 0.133, "step": 41304 }, { "epoch": 30.26007326007326, "grad_norm": 1.0703125, "learning_rate": 5.883156715376447e-06, "loss": 0.1046, "step": 41305 }, { "epoch": 30.26080586080586, "grad_norm": 0.7890625, "learning_rate": 5.8816728971973055e-06, "loss": 0.1072, "step": 41306 }, { "epoch": 30.26153846153846, "grad_norm": 0.91796875, "learning_rate": 5.88018929279921e-06, "loss": 0.1106, "step": 41307 }, { "epoch": 30.26227106227106, "grad_norm": 1.2734375, "learning_rate": 5.878705902189285e-06, "loss": 0.1322, "step": 41308 }, { "epoch": 30.263003663003662, "grad_norm": 0.828125, "learning_rate": 5.877222725374616e-06, "loss": 0.123, "step": 41309 }, { "epoch": 30.263736263736263, "grad_norm": 1.4609375, "learning_rate": 5.875739762362321e-06, "loss": 0.1473, "step": 41310 }, { "epoch": 30.264468864468864, "grad_norm": 0.71484375, "learning_rate": 5.8742570131595095e-06, "loss": 0.1091, "step": 41311 }, { "epoch": 30.265201465201464, "grad_norm": 1.3203125, "learning_rate": 5.872774477773288e-06, "loss": 0.0955, "step": 41312 }, { "epoch": 30.265934065934065, "grad_norm": 1.34375, "learning_rate": 5.871292156210745e-06, "loss": 0.1254, "step": 41313 }, { "epoch": 30.266666666666666, "grad_norm": 0.86328125, "learning_rate": 5.869810048478994e-06, "loss": 0.1317, "step": 41314 }, { "epoch": 30.267399267399266, "grad_norm": 1.0546875, "learning_rate": 5.8683281545851325e-06, "loss": 0.1101, "step": 41315 }, { "epoch": 30.268131868131867, "grad_norm": 1.1953125, "learning_rate": 5.866846474536274e-06, "loss": 0.1168, "step": 41316 }, { "epoch": 30.268864468864468, "grad_norm": 0.90625, "learning_rate": 5.865365008339496e-06, "loss": 0.0948, "step": 41317 }, { "epoch": 30.26959706959707, "grad_norm": 0.90234375, "learning_rate": 5.863883756001903e-06, "loss": 0.1006, "step": 41318 }, { "epoch": 30.27032967032967, "grad_norm": 0.89453125, "learning_rate": 5.86240271753061e-06, "loss": 0.1328, "step": 41319 }, { "epoch": 30.27106227106227, "grad_norm": 1.5703125, "learning_rate": 5.86092189293268e-06, "loss": 0.116, "step": 41320 }, { "epoch": 30.27179487179487, "grad_norm": 1.171875, "learning_rate": 5.859441282215244e-06, "loss": 0.095, "step": 41321 }, { "epoch": 30.27252747252747, "grad_norm": 1.2734375, "learning_rate": 5.857960885385373e-06, "loss": 0.1608, "step": 41322 }, { "epoch": 30.273260073260072, "grad_norm": 1.03125, "learning_rate": 5.8564807024501746e-06, "loss": 0.1186, "step": 41323 }, { "epoch": 30.273992673992673, "grad_norm": 1.2734375, "learning_rate": 5.85500073341672e-06, "loss": 0.1251, "step": 41324 }, { "epoch": 30.274725274725274, "grad_norm": 1.2109375, "learning_rate": 5.853520978292115e-06, "loss": 0.1314, "step": 41325 }, { "epoch": 30.275457875457874, "grad_norm": 0.859375, "learning_rate": 5.85204143708345e-06, "loss": 0.1087, "step": 41326 }, { "epoch": 30.276190476190475, "grad_norm": 0.90625, "learning_rate": 5.850562109797818e-06, "loss": 0.1205, "step": 41327 }, { "epoch": 30.276923076923076, "grad_norm": 1.3671875, "learning_rate": 5.849082996442289e-06, "loss": 0.1577, "step": 41328 }, { "epoch": 30.277655677655677, "grad_norm": 1.28125, "learning_rate": 5.847604097023967e-06, "loss": 0.1447, "step": 41329 }, { "epoch": 30.278388278388277, "grad_norm": 1.171875, "learning_rate": 5.846125411549926e-06, "loss": 0.1046, "step": 41330 }, { "epoch": 30.279120879120878, "grad_norm": 1.1640625, "learning_rate": 5.8446469400272596e-06, "loss": 0.0855, "step": 41331 }, { "epoch": 30.27985347985348, "grad_norm": 0.8203125, "learning_rate": 5.843168682463063e-06, "loss": 0.0998, "step": 41332 }, { "epoch": 30.28058608058608, "grad_norm": 0.6875, "learning_rate": 5.841690638864392e-06, "loss": 0.1195, "step": 41333 }, { "epoch": 30.28131868131868, "grad_norm": 0.7734375, "learning_rate": 5.840212809238349e-06, "loss": 0.1382, "step": 41334 }, { "epoch": 30.28205128205128, "grad_norm": 0.67578125, "learning_rate": 5.838735193591999e-06, "loss": 0.1039, "step": 41335 }, { "epoch": 30.282783882783882, "grad_norm": 0.86328125, "learning_rate": 5.837257791932432e-06, "loss": 0.128, "step": 41336 }, { "epoch": 30.283516483516483, "grad_norm": 0.8125, "learning_rate": 5.835780604266722e-06, "loss": 0.1063, "step": 41337 }, { "epoch": 30.284249084249083, "grad_norm": 1.09375, "learning_rate": 5.834303630601963e-06, "loss": 0.1091, "step": 41338 }, { "epoch": 30.284981684981684, "grad_norm": 0.80859375, "learning_rate": 5.8328268709452055e-06, "loss": 0.1213, "step": 41339 }, { "epoch": 30.285714285714285, "grad_norm": 0.89453125, "learning_rate": 5.831350325303533e-06, "loss": 0.1113, "step": 41340 }, { "epoch": 30.286446886446885, "grad_norm": 1.0234375, "learning_rate": 5.8298739936840315e-06, "loss": 0.0973, "step": 41341 }, { "epoch": 30.287179487179486, "grad_norm": 0.8671875, "learning_rate": 5.828397876093777e-06, "loss": 0.1222, "step": 41342 }, { "epoch": 30.287912087912087, "grad_norm": 0.8359375, "learning_rate": 5.8269219725398214e-06, "loss": 0.1175, "step": 41343 }, { "epoch": 30.288644688644688, "grad_norm": 1.3203125, "learning_rate": 5.8254462830292485e-06, "loss": 0.1514, "step": 41344 }, { "epoch": 30.28937728937729, "grad_norm": 0.98828125, "learning_rate": 5.82397080756914e-06, "loss": 0.1046, "step": 41345 }, { "epoch": 30.29010989010989, "grad_norm": 1.078125, "learning_rate": 5.822495546166535e-06, "loss": 0.1045, "step": 41346 }, { "epoch": 30.29084249084249, "grad_norm": 1.359375, "learning_rate": 5.821020498828535e-06, "loss": 0.1123, "step": 41347 }, { "epoch": 30.29157509157509, "grad_norm": 0.984375, "learning_rate": 5.8195456655621875e-06, "loss": 0.1257, "step": 41348 }, { "epoch": 30.29230769230769, "grad_norm": 1.1953125, "learning_rate": 5.8180710463745694e-06, "loss": 0.1258, "step": 41349 }, { "epoch": 30.293040293040292, "grad_norm": 1.3828125, "learning_rate": 5.816596641272733e-06, "loss": 0.1256, "step": 41350 }, { "epoch": 30.293772893772893, "grad_norm": 0.859375, "learning_rate": 5.81512245026375e-06, "loss": 0.097, "step": 41351 }, { "epoch": 30.294505494505493, "grad_norm": 1.0546875, "learning_rate": 5.81364847335469e-06, "loss": 0.1296, "step": 41352 }, { "epoch": 30.295238095238094, "grad_norm": 1.75, "learning_rate": 5.81217471055261e-06, "loss": 0.1429, "step": 41353 }, { "epoch": 30.295970695970695, "grad_norm": 1.15625, "learning_rate": 5.810701161864572e-06, "loss": 0.1346, "step": 41354 }, { "epoch": 30.296703296703296, "grad_norm": 1.3515625, "learning_rate": 5.809227827297629e-06, "loss": 0.1458, "step": 41355 }, { "epoch": 30.297435897435896, "grad_norm": 0.96875, "learning_rate": 5.807754706858845e-06, "loss": 0.11, "step": 41356 }, { "epoch": 30.298168498168497, "grad_norm": 1.015625, "learning_rate": 5.806281800555286e-06, "loss": 0.1384, "step": 41357 }, { "epoch": 30.298901098901098, "grad_norm": 0.83203125, "learning_rate": 5.804809108394006e-06, "loss": 0.1198, "step": 41358 }, { "epoch": 30.2996336996337, "grad_norm": 0.93359375, "learning_rate": 5.803336630382049e-06, "loss": 0.1164, "step": 41359 }, { "epoch": 30.3003663003663, "grad_norm": 0.95703125, "learning_rate": 5.801864366526491e-06, "loss": 0.1239, "step": 41360 }, { "epoch": 30.3010989010989, "grad_norm": 0.87109375, "learning_rate": 5.800392316834367e-06, "loss": 0.1304, "step": 41361 }, { "epoch": 30.3018315018315, "grad_norm": 0.7265625, "learning_rate": 5.7989204813127345e-06, "loss": 0.1067, "step": 41362 }, { "epoch": 30.3025641025641, "grad_norm": 0.87109375, "learning_rate": 5.7974488599686475e-06, "loss": 0.0984, "step": 41363 }, { "epoch": 30.303296703296702, "grad_norm": 0.73046875, "learning_rate": 5.795977452809171e-06, "loss": 0.1059, "step": 41364 }, { "epoch": 30.304029304029303, "grad_norm": 0.92578125, "learning_rate": 5.794506259841328e-06, "loss": 0.1147, "step": 41365 }, { "epoch": 30.304761904761904, "grad_norm": 0.890625, "learning_rate": 5.793035281072187e-06, "loss": 0.129, "step": 41366 }, { "epoch": 30.305494505494504, "grad_norm": 0.9453125, "learning_rate": 5.791564516508785e-06, "loss": 0.1204, "step": 41367 }, { "epoch": 30.306227106227105, "grad_norm": 1.2109375, "learning_rate": 5.7900939661581844e-06, "loss": 0.1118, "step": 41368 }, { "epoch": 30.306959706959706, "grad_norm": 0.90234375, "learning_rate": 5.788623630027414e-06, "loss": 0.1467, "step": 41369 }, { "epoch": 30.307692307692307, "grad_norm": 0.8515625, "learning_rate": 5.787153508123522e-06, "loss": 0.1293, "step": 41370 }, { "epoch": 30.308424908424907, "grad_norm": 0.7578125, "learning_rate": 5.7856836004535685e-06, "loss": 0.1064, "step": 41371 }, { "epoch": 30.309157509157508, "grad_norm": 0.96875, "learning_rate": 5.784213907024565e-06, "loss": 0.1305, "step": 41372 }, { "epoch": 30.30989010989011, "grad_norm": 1.015625, "learning_rate": 5.782744427843588e-06, "loss": 0.1191, "step": 41373 }, { "epoch": 30.31062271062271, "grad_norm": 1.3046875, "learning_rate": 5.781275162917653e-06, "loss": 0.1188, "step": 41374 }, { "epoch": 30.31135531135531, "grad_norm": 1.078125, "learning_rate": 5.779806112253819e-06, "loss": 0.111, "step": 41375 }, { "epoch": 30.31208791208791, "grad_norm": 1.1875, "learning_rate": 5.778337275859109e-06, "loss": 0.1301, "step": 41376 }, { "epoch": 30.31282051282051, "grad_norm": 1.2109375, "learning_rate": 5.776868653740561e-06, "loss": 0.1327, "step": 41377 }, { "epoch": 30.313553113553112, "grad_norm": 0.828125, "learning_rate": 5.775400245905217e-06, "loss": 0.1085, "step": 41378 }, { "epoch": 30.314285714285713, "grad_norm": 0.89453125, "learning_rate": 5.773932052360124e-06, "loss": 0.1084, "step": 41379 }, { "epoch": 30.315018315018314, "grad_norm": 0.74609375, "learning_rate": 5.7724640731123e-06, "loss": 0.1395, "step": 41380 }, { "epoch": 30.315750915750915, "grad_norm": 1.359375, "learning_rate": 5.770996308168779e-06, "loss": 0.1449, "step": 41381 }, { "epoch": 30.316483516483515, "grad_norm": 1.2890625, "learning_rate": 5.769528757536593e-06, "loss": 0.1583, "step": 41382 }, { "epoch": 30.317216117216116, "grad_norm": 1.2109375, "learning_rate": 5.768061421222787e-06, "loss": 0.1112, "step": 41383 }, { "epoch": 30.317948717948717, "grad_norm": 0.96875, "learning_rate": 5.76659429923439e-06, "loss": 0.1437, "step": 41384 }, { "epoch": 30.318681318681318, "grad_norm": 0.84375, "learning_rate": 5.765127391578418e-06, "loss": 0.1104, "step": 41385 }, { "epoch": 30.31941391941392, "grad_norm": 1.5859375, "learning_rate": 5.763660698261909e-06, "loss": 0.1298, "step": 41386 }, { "epoch": 30.32014652014652, "grad_norm": 0.765625, "learning_rate": 5.762194219291882e-06, "loss": 0.1009, "step": 41387 }, { "epoch": 30.32087912087912, "grad_norm": 1.65625, "learning_rate": 5.760727954675371e-06, "loss": 0.1227, "step": 41388 }, { "epoch": 30.32161172161172, "grad_norm": 0.8203125, "learning_rate": 5.759261904419394e-06, "loss": 0.0955, "step": 41389 }, { "epoch": 30.32234432234432, "grad_norm": 0.84375, "learning_rate": 5.757796068530996e-06, "loss": 0.121, "step": 41390 }, { "epoch": 30.323076923076922, "grad_norm": 1.6875, "learning_rate": 5.756330447017173e-06, "loss": 0.132, "step": 41391 }, { "epoch": 30.323809523809523, "grad_norm": 0.99609375, "learning_rate": 5.7548650398849585e-06, "loss": 0.1081, "step": 41392 }, { "epoch": 30.324542124542123, "grad_norm": 0.93359375, "learning_rate": 5.753399847141373e-06, "loss": 0.1398, "step": 41393 }, { "epoch": 30.325274725274724, "grad_norm": 1.6015625, "learning_rate": 5.7519348687934545e-06, "loss": 0.1256, "step": 41394 }, { "epoch": 30.326007326007325, "grad_norm": 0.78515625, "learning_rate": 5.750470104848188e-06, "loss": 0.1014, "step": 41395 }, { "epoch": 30.326739926739926, "grad_norm": 1.0390625, "learning_rate": 5.749005555312613e-06, "loss": 0.1362, "step": 41396 }, { "epoch": 30.327472527472526, "grad_norm": 0.62890625, "learning_rate": 5.7475412201937525e-06, "loss": 0.1123, "step": 41397 }, { "epoch": 30.328205128205127, "grad_norm": 0.59375, "learning_rate": 5.746077099498594e-06, "loss": 0.1119, "step": 41398 }, { "epoch": 30.328937728937728, "grad_norm": 1.1015625, "learning_rate": 5.744613193234191e-06, "loss": 0.0889, "step": 41399 }, { "epoch": 30.32967032967033, "grad_norm": 0.890625, "learning_rate": 5.74314950140753e-06, "loss": 0.1098, "step": 41400 }, { "epoch": 30.33040293040293, "grad_norm": 1.2265625, "learning_rate": 5.74168602402564e-06, "loss": 0.1541, "step": 41401 }, { "epoch": 30.33113553113553, "grad_norm": 1.6953125, "learning_rate": 5.74022276109552e-06, "loss": 0.1413, "step": 41402 }, { "epoch": 30.33186813186813, "grad_norm": 1.0546875, "learning_rate": 5.738759712624181e-06, "loss": 0.1291, "step": 41403 }, { "epoch": 30.33260073260073, "grad_norm": 0.7890625, "learning_rate": 5.737296878618643e-06, "loss": 0.1252, "step": 41404 }, { "epoch": 30.333333333333332, "grad_norm": 1.296875, "learning_rate": 5.7358342590859206e-06, "loss": 0.1215, "step": 41405 }, { "epoch": 30.334065934065933, "grad_norm": 1.3125, "learning_rate": 5.7343718540329965e-06, "loss": 0.1285, "step": 41406 }, { "epoch": 30.334798534798534, "grad_norm": 1.03125, "learning_rate": 5.732909663466895e-06, "loss": 0.1072, "step": 41407 }, { "epoch": 30.335531135531134, "grad_norm": 0.90234375, "learning_rate": 5.731447687394619e-06, "loss": 0.1234, "step": 41408 }, { "epoch": 30.336263736263735, "grad_norm": 1.0625, "learning_rate": 5.729985925823173e-06, "loss": 0.1233, "step": 41409 }, { "epoch": 30.336996336996336, "grad_norm": 1.15625, "learning_rate": 5.7285243787595755e-06, "loss": 0.1404, "step": 41410 }, { "epoch": 30.337728937728937, "grad_norm": 0.8515625, "learning_rate": 5.7270630462108025e-06, "loss": 0.1232, "step": 41411 }, { "epoch": 30.338461538461537, "grad_norm": 1.15625, "learning_rate": 5.725601928183874e-06, "loss": 0.1204, "step": 41412 }, { "epoch": 30.339194139194138, "grad_norm": 1.1796875, "learning_rate": 5.724141024685784e-06, "loss": 0.1225, "step": 41413 }, { "epoch": 30.33992673992674, "grad_norm": 1.765625, "learning_rate": 5.722680335723525e-06, "loss": 0.1438, "step": 41414 }, { "epoch": 30.34065934065934, "grad_norm": 0.86328125, "learning_rate": 5.7212198613041055e-06, "loss": 0.1138, "step": 41415 }, { "epoch": 30.34139194139194, "grad_norm": 1.0, "learning_rate": 5.719759601434534e-06, "loss": 0.124, "step": 41416 }, { "epoch": 30.34212454212454, "grad_norm": 1.7265625, "learning_rate": 5.718299556121787e-06, "loss": 0.1387, "step": 41417 }, { "epoch": 30.34285714285714, "grad_norm": 0.77734375, "learning_rate": 5.716839725372862e-06, "loss": 0.1548, "step": 41418 }, { "epoch": 30.343589743589742, "grad_norm": 0.87890625, "learning_rate": 5.715380109194767e-06, "loss": 0.1272, "step": 41419 }, { "epoch": 30.344322344322343, "grad_norm": 1.109375, "learning_rate": 5.713920707594478e-06, "loss": 0.1131, "step": 41420 }, { "epoch": 30.345054945054944, "grad_norm": 0.58984375, "learning_rate": 5.712461520579016e-06, "loss": 0.106, "step": 41421 }, { "epoch": 30.345787545787545, "grad_norm": 1.0625, "learning_rate": 5.711002548155336e-06, "loss": 0.1551, "step": 41422 }, { "epoch": 30.346520146520145, "grad_norm": 1.2109375, "learning_rate": 5.709543790330459e-06, "loss": 0.1066, "step": 41423 }, { "epoch": 30.347252747252746, "grad_norm": 0.73046875, "learning_rate": 5.708085247111344e-06, "loss": 0.1245, "step": 41424 }, { "epoch": 30.347985347985347, "grad_norm": 1.3984375, "learning_rate": 5.706626918505013e-06, "loss": 0.1354, "step": 41425 }, { "epoch": 30.348717948717947, "grad_norm": 0.984375, "learning_rate": 5.705168804518432e-06, "loss": 0.1295, "step": 41426 }, { "epoch": 30.349450549450548, "grad_norm": 0.70703125, "learning_rate": 5.703710905158598e-06, "loss": 0.0962, "step": 41427 }, { "epoch": 30.35018315018315, "grad_norm": 1.046875, "learning_rate": 5.702253220432482e-06, "loss": 0.1389, "step": 41428 }, { "epoch": 30.35091575091575, "grad_norm": 1.2578125, "learning_rate": 5.700795750347079e-06, "loss": 0.1614, "step": 41429 }, { "epoch": 30.35164835164835, "grad_norm": 0.8203125, "learning_rate": 5.699338494909365e-06, "loss": 0.1098, "step": 41430 }, { "epoch": 30.35238095238095, "grad_norm": 1.1328125, "learning_rate": 5.6978814541263426e-06, "loss": 0.1157, "step": 41431 }, { "epoch": 30.353113553113552, "grad_norm": 1.140625, "learning_rate": 5.696424628004965e-06, "loss": 0.1124, "step": 41432 }, { "epoch": 30.353846153846153, "grad_norm": 0.72265625, "learning_rate": 5.694968016552225e-06, "loss": 0.1082, "step": 41433 }, { "epoch": 30.354578754578753, "grad_norm": 1.2265625, "learning_rate": 5.693511619775105e-06, "loss": 0.1464, "step": 41434 }, { "epoch": 30.355311355311354, "grad_norm": 1.2109375, "learning_rate": 5.692055437680577e-06, "loss": 0.1339, "step": 41435 }, { "epoch": 30.356043956043955, "grad_norm": 1.171875, "learning_rate": 5.690599470275633e-06, "loss": 0.15, "step": 41436 }, { "epoch": 30.356776556776556, "grad_norm": 1.5859375, "learning_rate": 5.689143717567224e-06, "loss": 0.1235, "step": 41437 }, { "epoch": 30.357509157509156, "grad_norm": 1.2734375, "learning_rate": 5.687688179562349e-06, "loss": 0.1321, "step": 41438 }, { "epoch": 30.358241758241757, "grad_norm": 1.2421875, "learning_rate": 5.686232856267959e-06, "loss": 0.1406, "step": 41439 }, { "epoch": 30.358974358974358, "grad_norm": 0.85546875, "learning_rate": 5.684777747691042e-06, "loss": 0.1123, "step": 41440 }, { "epoch": 30.35970695970696, "grad_norm": 1.1640625, "learning_rate": 5.683322853838567e-06, "loss": 0.1094, "step": 41441 }, { "epoch": 30.36043956043956, "grad_norm": 1.0, "learning_rate": 5.681868174717513e-06, "loss": 0.129, "step": 41442 }, { "epoch": 30.36117216117216, "grad_norm": 1.3125, "learning_rate": 5.680413710334832e-06, "loss": 0.1246, "step": 41443 }, { "epoch": 30.36190476190476, "grad_norm": 1.5859375, "learning_rate": 5.678959460697503e-06, "loss": 0.1178, "step": 41444 }, { "epoch": 30.36263736263736, "grad_norm": 0.98046875, "learning_rate": 5.67750542581249e-06, "loss": 0.1329, "step": 41445 }, { "epoch": 30.363369963369962, "grad_norm": 1.0390625, "learning_rate": 5.676051605686766e-06, "loss": 0.1105, "step": 41446 }, { "epoch": 30.364102564102563, "grad_norm": 0.6484375, "learning_rate": 5.6745980003273e-06, "loss": 0.118, "step": 41447 }, { "epoch": 30.364835164835164, "grad_norm": 1.203125, "learning_rate": 5.673144609741043e-06, "loss": 0.1274, "step": 41448 }, { "epoch": 30.365567765567764, "grad_norm": 0.7578125, "learning_rate": 5.671691433934971e-06, "loss": 0.1191, "step": 41449 }, { "epoch": 30.366300366300365, "grad_norm": 1.203125, "learning_rate": 5.670238472916028e-06, "loss": 0.1262, "step": 41450 }, { "epoch": 30.367032967032966, "grad_norm": 0.8203125, "learning_rate": 5.668785726691208e-06, "loss": 0.1186, "step": 41451 }, { "epoch": 30.367765567765566, "grad_norm": 0.7890625, "learning_rate": 5.667333195267441e-06, "loss": 0.1139, "step": 41452 }, { "epoch": 30.368498168498167, "grad_norm": 0.8046875, "learning_rate": 5.665880878651709e-06, "loss": 0.093, "step": 41453 }, { "epoch": 30.369230769230768, "grad_norm": 0.8203125, "learning_rate": 5.664428776850949e-06, "loss": 0.1023, "step": 41454 }, { "epoch": 30.36996336996337, "grad_norm": 1.140625, "learning_rate": 5.662976889872129e-06, "loss": 0.0904, "step": 41455 }, { "epoch": 30.37069597069597, "grad_norm": 1.015625, "learning_rate": 5.661525217722205e-06, "loss": 0.1424, "step": 41456 }, { "epoch": 30.37142857142857, "grad_norm": 0.890625, "learning_rate": 5.660073760408147e-06, "loss": 0.1227, "step": 41457 }, { "epoch": 30.37216117216117, "grad_norm": 1.515625, "learning_rate": 5.658622517936883e-06, "loss": 0.1538, "step": 41458 }, { "epoch": 30.37289377289377, "grad_norm": 1.3125, "learning_rate": 5.657171490315377e-06, "loss": 0.0993, "step": 41459 }, { "epoch": 30.373626373626372, "grad_norm": 1.1171875, "learning_rate": 5.655720677550585e-06, "loss": 0.1333, "step": 41460 }, { "epoch": 30.374358974358973, "grad_norm": 1.421875, "learning_rate": 5.654270079649457e-06, "loss": 0.1309, "step": 41461 }, { "epoch": 30.375091575091574, "grad_norm": 0.8828125, "learning_rate": 5.652819696618954e-06, "loss": 0.124, "step": 41462 }, { "epoch": 30.375824175824174, "grad_norm": 0.7578125, "learning_rate": 5.651369528466005e-06, "loss": 0.1183, "step": 41463 }, { "epoch": 30.376556776556775, "grad_norm": 1.0703125, "learning_rate": 5.649919575197573e-06, "loss": 0.1275, "step": 41464 }, { "epoch": 30.377289377289376, "grad_norm": 1.1875, "learning_rate": 5.648469836820596e-06, "loss": 0.145, "step": 41465 }, { "epoch": 30.378021978021977, "grad_norm": 0.9609375, "learning_rate": 5.6470203133420215e-06, "loss": 0.1034, "step": 41466 }, { "epoch": 30.378754578754577, "grad_norm": 0.72265625, "learning_rate": 5.645571004768794e-06, "loss": 0.0976, "step": 41467 }, { "epoch": 30.379487179487178, "grad_norm": 1.1796875, "learning_rate": 5.644121911107871e-06, "loss": 0.1231, "step": 41468 }, { "epoch": 30.38021978021978, "grad_norm": 0.80078125, "learning_rate": 5.642673032366175e-06, "loss": 0.1267, "step": 41469 }, { "epoch": 30.38095238095238, "grad_norm": 1.03125, "learning_rate": 5.641224368550658e-06, "loss": 0.1233, "step": 41470 }, { "epoch": 30.38168498168498, "grad_norm": 1.171875, "learning_rate": 5.6397759196682634e-06, "loss": 0.1264, "step": 41471 }, { "epoch": 30.38241758241758, "grad_norm": 1.1015625, "learning_rate": 5.63832768572593e-06, "loss": 0.135, "step": 41472 }, { "epoch": 30.383150183150182, "grad_norm": 0.76953125, "learning_rate": 5.636879666730606e-06, "loss": 0.1019, "step": 41473 }, { "epoch": 30.383882783882783, "grad_norm": 1.1953125, "learning_rate": 5.635431862689207e-06, "loss": 0.1079, "step": 41474 }, { "epoch": 30.384615384615383, "grad_norm": 0.91015625, "learning_rate": 5.633984273608695e-06, "loss": 0.1018, "step": 41475 }, { "epoch": 30.385347985347984, "grad_norm": 0.75390625, "learning_rate": 5.632536899495967e-06, "loss": 0.1216, "step": 41476 }, { "epoch": 30.386080586080585, "grad_norm": 1.6328125, "learning_rate": 5.631089740358008e-06, "loss": 0.1177, "step": 41477 }, { "epoch": 30.386813186813185, "grad_norm": 1.1640625, "learning_rate": 5.629642796201718e-06, "loss": 0.1076, "step": 41478 }, { "epoch": 30.387545787545786, "grad_norm": 0.78515625, "learning_rate": 5.628196067034047e-06, "loss": 0.1268, "step": 41479 }, { "epoch": 30.388278388278387, "grad_norm": 1.359375, "learning_rate": 5.626749552861911e-06, "loss": 0.1163, "step": 41480 }, { "epoch": 30.389010989010988, "grad_norm": 0.8828125, "learning_rate": 5.6253032536922445e-06, "loss": 0.0943, "step": 41481 }, { "epoch": 30.38974358974359, "grad_norm": 1.328125, "learning_rate": 5.6238571695319846e-06, "loss": 0.1629, "step": 41482 }, { "epoch": 30.39047619047619, "grad_norm": 1.125, "learning_rate": 5.622411300388065e-06, "loss": 0.1325, "step": 41483 }, { "epoch": 30.39120879120879, "grad_norm": 0.87890625, "learning_rate": 5.6209656462674e-06, "loss": 0.113, "step": 41484 }, { "epoch": 30.39194139194139, "grad_norm": 1.359375, "learning_rate": 5.6195202071769175e-06, "loss": 0.1261, "step": 41485 }, { "epoch": 30.39267399267399, "grad_norm": 0.8515625, "learning_rate": 5.61807498312355e-06, "loss": 0.0922, "step": 41486 }, { "epoch": 30.393406593406592, "grad_norm": 1.078125, "learning_rate": 5.616629974114217e-06, "loss": 0.121, "step": 41487 }, { "epoch": 30.394139194139193, "grad_norm": 1.6171875, "learning_rate": 5.615185180155856e-06, "loss": 0.116, "step": 41488 }, { "epoch": 30.394871794871793, "grad_norm": 0.8125, "learning_rate": 5.613740601255361e-06, "loss": 0.1485, "step": 41489 }, { "epoch": 30.395604395604394, "grad_norm": 1.328125, "learning_rate": 5.61229623741969e-06, "loss": 0.1405, "step": 41490 }, { "epoch": 30.396336996336995, "grad_norm": 0.83203125, "learning_rate": 5.610852088655727e-06, "loss": 0.0893, "step": 41491 }, { "epoch": 30.397069597069596, "grad_norm": 0.8359375, "learning_rate": 5.609408154970408e-06, "loss": 0.1168, "step": 41492 }, { "epoch": 30.397802197802196, "grad_norm": 1.2578125, "learning_rate": 5.607964436370654e-06, "loss": 0.1035, "step": 41493 }, { "epoch": 30.398534798534797, "grad_norm": 1.6171875, "learning_rate": 5.606520932863386e-06, "loss": 0.1103, "step": 41494 }, { "epoch": 30.399267399267398, "grad_norm": 0.9453125, "learning_rate": 5.605077644455503e-06, "loss": 0.1341, "step": 41495 }, { "epoch": 30.4, "grad_norm": 1.1953125, "learning_rate": 5.603634571153936e-06, "loss": 0.0934, "step": 41496 }, { "epoch": 30.4007326007326, "grad_norm": 1.1875, "learning_rate": 5.60219171296559e-06, "loss": 0.1231, "step": 41497 }, { "epoch": 30.4014652014652, "grad_norm": 1.2109375, "learning_rate": 5.600749069897382e-06, "loss": 0.1094, "step": 41498 }, { "epoch": 30.4021978021978, "grad_norm": 1.359375, "learning_rate": 5.59930664195624e-06, "loss": 0.1249, "step": 41499 }, { "epoch": 30.4029304029304, "grad_norm": 1.296875, "learning_rate": 5.597864429149044e-06, "loss": 0.1233, "step": 41500 }, { "epoch": 30.403663003663002, "grad_norm": 1.2109375, "learning_rate": 5.596422431482735e-06, "loss": 0.1294, "step": 41501 }, { "epoch": 30.404395604395603, "grad_norm": 1.1015625, "learning_rate": 5.594980648964185e-06, "loss": 0.1147, "step": 41502 }, { "epoch": 30.405128205128204, "grad_norm": 0.97265625, "learning_rate": 5.593539081600342e-06, "loss": 0.104, "step": 41503 }, { "epoch": 30.405860805860804, "grad_norm": 1.125, "learning_rate": 5.592097729398089e-06, "loss": 0.1207, "step": 41504 }, { "epoch": 30.406593406593405, "grad_norm": 0.7421875, "learning_rate": 5.590656592364345e-06, "loss": 0.1241, "step": 41505 }, { "epoch": 30.407326007326006, "grad_norm": 0.78515625, "learning_rate": 5.5892156705059955e-06, "loss": 0.1351, "step": 41506 }, { "epoch": 30.408058608058607, "grad_norm": 1.109375, "learning_rate": 5.58777496382996e-06, "loss": 0.114, "step": 41507 }, { "epoch": 30.408791208791207, "grad_norm": 1.203125, "learning_rate": 5.58633447234314e-06, "loss": 0.119, "step": 41508 }, { "epoch": 30.409523809523808, "grad_norm": 0.55859375, "learning_rate": 5.5848941960524426e-06, "loss": 0.1103, "step": 41509 }, { "epoch": 30.41025641025641, "grad_norm": 1.171875, "learning_rate": 5.583454134964757e-06, "loss": 0.1316, "step": 41510 }, { "epoch": 30.41098901098901, "grad_norm": 1.28125, "learning_rate": 5.582014289086983e-06, "loss": 0.1095, "step": 41511 }, { "epoch": 30.41172161172161, "grad_norm": 1.0, "learning_rate": 5.580574658426035e-06, "loss": 0.1271, "step": 41512 }, { "epoch": 30.41245421245421, "grad_norm": 1.0859375, "learning_rate": 5.5791352429887824e-06, "loss": 0.1176, "step": 41513 }, { "epoch": 30.41318681318681, "grad_norm": 0.640625, "learning_rate": 5.577696042782154e-06, "loss": 0.1107, "step": 41514 }, { "epoch": 30.413919413919412, "grad_norm": 1.125, "learning_rate": 5.576257057813028e-06, "loss": 0.1453, "step": 41515 }, { "epoch": 30.414652014652013, "grad_norm": 1.25, "learning_rate": 5.574818288088303e-06, "loss": 0.1329, "step": 41516 }, { "epoch": 30.415384615384614, "grad_norm": 0.83203125, "learning_rate": 5.573379733614868e-06, "loss": 0.1165, "step": 41517 }, { "epoch": 30.416117216117215, "grad_norm": 0.9140625, "learning_rate": 5.571941394399618e-06, "loss": 0.1265, "step": 41518 }, { "epoch": 30.416849816849815, "grad_norm": 1.2890625, "learning_rate": 5.5705032704494455e-06, "loss": 0.1, "step": 41519 }, { "epoch": 30.417582417582416, "grad_norm": 1.03125, "learning_rate": 5.569065361771253e-06, "loss": 0.1085, "step": 41520 }, { "epoch": 30.418315018315017, "grad_norm": 0.9296875, "learning_rate": 5.567627668371904e-06, "loss": 0.1292, "step": 41521 }, { "epoch": 30.419047619047618, "grad_norm": 0.61328125, "learning_rate": 5.5661901902583065e-06, "loss": 0.1184, "step": 41522 }, { "epoch": 30.41978021978022, "grad_norm": 1.3046875, "learning_rate": 5.564752927437341e-06, "loss": 0.1282, "step": 41523 }, { "epoch": 30.42051282051282, "grad_norm": 1.3515625, "learning_rate": 5.563315879915895e-06, "loss": 0.1263, "step": 41524 }, { "epoch": 30.42124542124542, "grad_norm": 0.95703125, "learning_rate": 5.561879047700863e-06, "loss": 0.1243, "step": 41525 }, { "epoch": 30.42197802197802, "grad_norm": 0.65234375, "learning_rate": 5.5604424307991105e-06, "loss": 0.1225, "step": 41526 }, { "epoch": 30.42271062271062, "grad_norm": 1.7109375, "learning_rate": 5.559006029217541e-06, "loss": 0.1714, "step": 41527 }, { "epoch": 30.423443223443222, "grad_norm": 0.78125, "learning_rate": 5.557569842963009e-06, "loss": 0.1237, "step": 41528 }, { "epoch": 30.424175824175823, "grad_norm": 1.0859375, "learning_rate": 5.556133872042432e-06, "loss": 0.1043, "step": 41529 }, { "epoch": 30.424908424908423, "grad_norm": 0.89453125, "learning_rate": 5.554698116462663e-06, "loss": 0.1453, "step": 41530 }, { "epoch": 30.425641025641024, "grad_norm": 1.453125, "learning_rate": 5.553262576230596e-06, "loss": 0.095, "step": 41531 }, { "epoch": 30.426373626373625, "grad_norm": 1.046875, "learning_rate": 5.551827251353098e-06, "loss": 0.1179, "step": 41532 }, { "epoch": 30.427106227106226, "grad_norm": 0.921875, "learning_rate": 5.550392141837046e-06, "loss": 0.0989, "step": 41533 }, { "epoch": 30.427838827838826, "grad_norm": 0.9453125, "learning_rate": 5.548957247689325e-06, "loss": 0.1544, "step": 41534 }, { "epoch": 30.428571428571427, "grad_norm": 1.4609375, "learning_rate": 5.547522568916806e-06, "loss": 0.1116, "step": 41535 }, { "epoch": 30.429304029304028, "grad_norm": 0.82421875, "learning_rate": 5.546088105526371e-06, "loss": 0.1037, "step": 41536 }, { "epoch": 30.43003663003663, "grad_norm": 1.2890625, "learning_rate": 5.544653857524881e-06, "loss": 0.1132, "step": 41537 }, { "epoch": 30.43076923076923, "grad_norm": 1.15625, "learning_rate": 5.543219824919218e-06, "loss": 0.1242, "step": 41538 }, { "epoch": 30.43150183150183, "grad_norm": 0.9140625, "learning_rate": 5.541786007716226e-06, "loss": 0.145, "step": 41539 }, { "epoch": 30.43223443223443, "grad_norm": 0.80859375, "learning_rate": 5.540352405922817e-06, "loss": 0.1283, "step": 41540 }, { "epoch": 30.43296703296703, "grad_norm": 0.875, "learning_rate": 5.538919019545831e-06, "loss": 0.1163, "step": 41541 }, { "epoch": 30.433699633699632, "grad_norm": 1.0, "learning_rate": 5.53748584859215e-06, "loss": 0.1076, "step": 41542 }, { "epoch": 30.434432234432233, "grad_norm": 1.078125, "learning_rate": 5.53605289306863e-06, "loss": 0.1283, "step": 41543 }, { "epoch": 30.435164835164834, "grad_norm": 1.140625, "learning_rate": 5.534620152982138e-06, "loss": 0.1348, "step": 41544 }, { "epoch": 30.435897435897434, "grad_norm": 0.953125, "learning_rate": 5.533187628339539e-06, "loss": 0.1403, "step": 41545 }, { "epoch": 30.436630036630035, "grad_norm": 0.75, "learning_rate": 5.5317553191477144e-06, "loss": 0.1329, "step": 41546 }, { "epoch": 30.437362637362636, "grad_norm": 1.046875, "learning_rate": 5.5303232254135e-06, "loss": 0.1335, "step": 41547 }, { "epoch": 30.438095238095237, "grad_norm": 0.8828125, "learning_rate": 5.528891347143772e-06, "loss": 0.1049, "step": 41548 }, { "epoch": 30.438827838827837, "grad_norm": 1.0234375, "learning_rate": 5.527459684345386e-06, "loss": 0.1467, "step": 41549 }, { "epoch": 30.439560439560438, "grad_norm": 1.734375, "learning_rate": 5.526028237025203e-06, "loss": 0.136, "step": 41550 }, { "epoch": 30.44029304029304, "grad_norm": 1.3203125, "learning_rate": 5.5245970051900975e-06, "loss": 0.1138, "step": 41551 }, { "epoch": 30.44102564102564, "grad_norm": 0.765625, "learning_rate": 5.523165988846895e-06, "loss": 0.0881, "step": 41552 }, { "epoch": 30.44175824175824, "grad_norm": 0.8359375, "learning_rate": 5.521735188002486e-06, "loss": 0.1085, "step": 41553 }, { "epoch": 30.44249084249084, "grad_norm": 1.5703125, "learning_rate": 5.520304602663686e-06, "loss": 0.1538, "step": 41554 }, { "epoch": 30.44322344322344, "grad_norm": 1.3984375, "learning_rate": 5.5188742328373915e-06, "loss": 0.1179, "step": 41555 }, { "epoch": 30.443956043956042, "grad_norm": 1.265625, "learning_rate": 5.517444078530429e-06, "loss": 0.1522, "step": 41556 }, { "epoch": 30.444688644688643, "grad_norm": 0.828125, "learning_rate": 5.516014139749667e-06, "loss": 0.1026, "step": 41557 }, { "epoch": 30.445421245421244, "grad_norm": 0.98046875, "learning_rate": 5.514584416501943e-06, "loss": 0.1294, "step": 41558 }, { "epoch": 30.446153846153845, "grad_norm": 1.453125, "learning_rate": 5.513154908794109e-06, "loss": 0.1348, "step": 41559 }, { "epoch": 30.446886446886445, "grad_norm": 0.84765625, "learning_rate": 5.511725616633019e-06, "loss": 0.1117, "step": 41560 }, { "epoch": 30.447619047619046, "grad_norm": 0.953125, "learning_rate": 5.5102965400255196e-06, "loss": 0.1056, "step": 41561 }, { "epoch": 30.448351648351647, "grad_norm": 1.3125, "learning_rate": 5.50886767897847e-06, "loss": 0.1317, "step": 41562 }, { "epoch": 30.449084249084247, "grad_norm": 0.73046875, "learning_rate": 5.5074390334986886e-06, "loss": 0.092, "step": 41563 }, { "epoch": 30.449816849816848, "grad_norm": 0.9453125, "learning_rate": 5.506010603593053e-06, "loss": 0.1234, "step": 41564 }, { "epoch": 30.45054945054945, "grad_norm": 0.90625, "learning_rate": 5.5045823892683715e-06, "loss": 0.1266, "step": 41565 }, { "epoch": 30.45128205128205, "grad_norm": 1.03125, "learning_rate": 5.5031543905315285e-06, "loss": 0.1037, "step": 41566 }, { "epoch": 30.45201465201465, "grad_norm": 1.109375, "learning_rate": 5.501726607389329e-06, "loss": 0.1264, "step": 41567 }, { "epoch": 30.45274725274725, "grad_norm": 0.90625, "learning_rate": 5.5002990398486416e-06, "loss": 0.1161, "step": 41568 }, { "epoch": 30.453479853479852, "grad_norm": 0.73046875, "learning_rate": 5.4988716879162824e-06, "loss": 0.1294, "step": 41569 }, { "epoch": 30.454212454212453, "grad_norm": 0.82421875, "learning_rate": 5.4974445515991085e-06, "loss": 0.1319, "step": 41570 }, { "epoch": 30.454945054945053, "grad_norm": 1.375, "learning_rate": 5.496017630903946e-06, "loss": 0.0949, "step": 41571 }, { "epoch": 30.455677655677654, "grad_norm": 0.86328125, "learning_rate": 5.494590925837642e-06, "loss": 0.0939, "step": 41572 }, { "epoch": 30.456410256410255, "grad_norm": 0.9375, "learning_rate": 5.493164436407028e-06, "loss": 0.1132, "step": 41573 }, { "epoch": 30.457142857142856, "grad_norm": 0.9609375, "learning_rate": 5.491738162618928e-06, "loss": 0.0953, "step": 41574 }, { "epoch": 30.457875457875456, "grad_norm": 0.78515625, "learning_rate": 5.490312104480193e-06, "loss": 0.1229, "step": 41575 }, { "epoch": 30.458608058608057, "grad_norm": 1.390625, "learning_rate": 5.488886261997645e-06, "loss": 0.1389, "step": 41576 }, { "epoch": 30.459340659340658, "grad_norm": 1.15625, "learning_rate": 5.487460635178128e-06, "loss": 0.1168, "step": 41577 }, { "epoch": 30.46007326007326, "grad_norm": 0.80078125, "learning_rate": 5.486035224028456e-06, "loss": 0.1188, "step": 41578 }, { "epoch": 30.46080586080586, "grad_norm": 1.0546875, "learning_rate": 5.484610028555476e-06, "loss": 0.1223, "step": 41579 }, { "epoch": 30.46153846153846, "grad_norm": 1.0703125, "learning_rate": 5.483185048765987e-06, "loss": 0.1112, "step": 41580 }, { "epoch": 30.46227106227106, "grad_norm": 0.87109375, "learning_rate": 5.481760284666852e-06, "loss": 0.1182, "step": 41581 }, { "epoch": 30.46300366300366, "grad_norm": 1.1796875, "learning_rate": 5.4803357362648756e-06, "loss": 0.1193, "step": 41582 }, { "epoch": 30.463736263736262, "grad_norm": 1.296875, "learning_rate": 5.4789114035669e-06, "loss": 0.1344, "step": 41583 }, { "epoch": 30.464468864468863, "grad_norm": 1.1328125, "learning_rate": 5.4774872865797325e-06, "loss": 0.1147, "step": 41584 }, { "epoch": 30.465201465201464, "grad_norm": 0.98828125, "learning_rate": 5.476063385310201e-06, "loss": 0.1266, "step": 41585 }, { "epoch": 30.465934065934064, "grad_norm": 1.234375, "learning_rate": 5.4746396997651284e-06, "loss": 0.1302, "step": 41586 }, { "epoch": 30.466666666666665, "grad_norm": 1.1171875, "learning_rate": 5.473216229951337e-06, "loss": 0.1534, "step": 41587 }, { "epoch": 30.467399267399266, "grad_norm": 1.2265625, "learning_rate": 5.471792975875663e-06, "loss": 0.1164, "step": 41588 }, { "epoch": 30.468131868131866, "grad_norm": 0.87109375, "learning_rate": 5.470369937544898e-06, "loss": 0.1069, "step": 41589 }, { "epoch": 30.468864468864467, "grad_norm": 1.171875, "learning_rate": 5.4689471149658775e-06, "loss": 0.1173, "step": 41590 }, { "epoch": 30.469597069597068, "grad_norm": 0.9453125, "learning_rate": 5.467524508145403e-06, "loss": 0.1286, "step": 41591 }, { "epoch": 30.47032967032967, "grad_norm": 0.77734375, "learning_rate": 5.466102117090313e-06, "loss": 0.1021, "step": 41592 }, { "epoch": 30.47106227106227, "grad_norm": 0.6328125, "learning_rate": 5.464679941807409e-06, "loss": 0.1175, "step": 41593 }, { "epoch": 30.47179487179487, "grad_norm": 0.97265625, "learning_rate": 5.463257982303513e-06, "loss": 0.1124, "step": 41594 }, { "epoch": 30.47252747252747, "grad_norm": 0.75390625, "learning_rate": 5.4618362385854255e-06, "loss": 0.095, "step": 41595 }, { "epoch": 30.47326007326007, "grad_norm": 1.2421875, "learning_rate": 5.460414710659965e-06, "loss": 0.1211, "step": 41596 }, { "epoch": 30.473992673992672, "grad_norm": 0.83984375, "learning_rate": 5.458993398533941e-06, "loss": 0.1235, "step": 41597 }, { "epoch": 30.474725274725273, "grad_norm": 0.859375, "learning_rate": 5.457572302214175e-06, "loss": 0.1227, "step": 41598 }, { "epoch": 30.475457875457874, "grad_norm": 0.91796875, "learning_rate": 5.456151421707459e-06, "loss": 0.1237, "step": 41599 }, { "epoch": 30.476190476190474, "grad_norm": 0.69140625, "learning_rate": 5.454730757020611e-06, "loss": 0.1067, "step": 41600 }, { "epoch": 30.476923076923075, "grad_norm": 0.77734375, "learning_rate": 5.453310308160432e-06, "loss": 0.1236, "step": 41601 }, { "epoch": 30.477655677655676, "grad_norm": 1.0078125, "learning_rate": 5.451890075133734e-06, "loss": 0.1184, "step": 41602 }, { "epoch": 30.478388278388277, "grad_norm": 0.984375, "learning_rate": 5.4504700579473215e-06, "loss": 0.1185, "step": 41603 }, { "epoch": 30.479120879120877, "grad_norm": 1.2578125, "learning_rate": 5.4490502566079935e-06, "loss": 0.1198, "step": 41604 }, { "epoch": 30.479853479853478, "grad_norm": 1.015625, "learning_rate": 5.447630671122564e-06, "loss": 0.1041, "step": 41605 }, { "epoch": 30.48058608058608, "grad_norm": 1.1484375, "learning_rate": 5.446211301497801e-06, "loss": 0.1222, "step": 41606 }, { "epoch": 30.48131868131868, "grad_norm": 0.96484375, "learning_rate": 5.444792147740557e-06, "loss": 0.1393, "step": 41607 }, { "epoch": 30.48205128205128, "grad_norm": 1.796875, "learning_rate": 5.443373209857589e-06, "loss": 0.1409, "step": 41608 }, { "epoch": 30.48278388278388, "grad_norm": 1.40625, "learning_rate": 5.4419544878557205e-06, "loss": 0.1136, "step": 41609 }, { "epoch": 30.483516483516482, "grad_norm": 0.80859375, "learning_rate": 5.4405359817417305e-06, "loss": 0.1064, "step": 41610 }, { "epoch": 30.484249084249083, "grad_norm": 0.6640625, "learning_rate": 5.439117691522427e-06, "loss": 0.0945, "step": 41611 }, { "epoch": 30.484981684981683, "grad_norm": 1.5625, "learning_rate": 5.437699617204603e-06, "loss": 0.1575, "step": 41612 }, { "epoch": 30.485714285714284, "grad_norm": 1.1015625, "learning_rate": 5.436281758795051e-06, "loss": 0.1177, "step": 41613 }, { "epoch": 30.486446886446885, "grad_norm": 0.91015625, "learning_rate": 5.43486411630058e-06, "loss": 0.1201, "step": 41614 }, { "epoch": 30.487179487179485, "grad_norm": 0.671875, "learning_rate": 5.433446689727954e-06, "loss": 0.1092, "step": 41615 }, { "epoch": 30.487912087912086, "grad_norm": 0.6796875, "learning_rate": 5.432029479083994e-06, "loss": 0.12, "step": 41616 }, { "epoch": 30.488644688644687, "grad_norm": 0.7265625, "learning_rate": 5.430612484375454e-06, "loss": 0.1152, "step": 41617 }, { "epoch": 30.489377289377288, "grad_norm": 1.0, "learning_rate": 5.429195705609161e-06, "loss": 0.128, "step": 41618 }, { "epoch": 30.49010989010989, "grad_norm": 0.98828125, "learning_rate": 5.427779142791882e-06, "loss": 0.1359, "step": 41619 }, { "epoch": 30.49084249084249, "grad_norm": 0.87890625, "learning_rate": 5.426362795930419e-06, "loss": 0.1015, "step": 41620 }, { "epoch": 30.49157509157509, "grad_norm": 0.953125, "learning_rate": 5.424946665031534e-06, "loss": 0.1059, "step": 41621 }, { "epoch": 30.49230769230769, "grad_norm": 1.0390625, "learning_rate": 5.423530750102033e-06, "loss": 0.1158, "step": 41622 }, { "epoch": 30.493040293040295, "grad_norm": 0.9296875, "learning_rate": 5.4221150511486905e-06, "loss": 0.1265, "step": 41623 }, { "epoch": 30.493772893772892, "grad_norm": 1.6953125, "learning_rate": 5.4206995681783e-06, "loss": 0.102, "step": 41624 }, { "epoch": 30.494505494505496, "grad_norm": 1.0625, "learning_rate": 5.419284301197632e-06, "loss": 0.1109, "step": 41625 }, { "epoch": 30.495238095238093, "grad_norm": 1.046875, "learning_rate": 5.417869250213471e-06, "loss": 0.1294, "step": 41626 }, { "epoch": 30.495970695970698, "grad_norm": 1.0234375, "learning_rate": 5.416454415232597e-06, "loss": 0.1311, "step": 41627 }, { "epoch": 30.496703296703295, "grad_norm": 1.125, "learning_rate": 5.41503979626179e-06, "loss": 0.111, "step": 41628 }, { "epoch": 30.4974358974359, "grad_norm": 1.0546875, "learning_rate": 5.413625393307837e-06, "loss": 0.1098, "step": 41629 }, { "epoch": 30.498168498168496, "grad_norm": 1.1015625, "learning_rate": 5.412211206377496e-06, "loss": 0.1035, "step": 41630 }, { "epoch": 30.4989010989011, "grad_norm": 0.9609375, "learning_rate": 5.410797235477562e-06, "loss": 0.1443, "step": 41631 }, { "epoch": 30.499633699633698, "grad_norm": 1.328125, "learning_rate": 5.409383480614783e-06, "loss": 0.1252, "step": 41632 }, { "epoch": 30.500366300366302, "grad_norm": 1.828125, "learning_rate": 5.407969941795966e-06, "loss": 0.1541, "step": 41633 }, { "epoch": 30.5010989010989, "grad_norm": 0.77734375, "learning_rate": 5.406556619027857e-06, "loss": 0.1273, "step": 41634 }, { "epoch": 30.501831501831504, "grad_norm": 1.828125, "learning_rate": 5.405143512317248e-06, "loss": 0.1567, "step": 41635 }, { "epoch": 30.5025641025641, "grad_norm": 0.9140625, "learning_rate": 5.403730621670896e-06, "loss": 0.1145, "step": 41636 }, { "epoch": 30.503296703296705, "grad_norm": 1.046875, "learning_rate": 5.4023179470955715e-06, "loss": 0.1268, "step": 41637 }, { "epoch": 30.504029304029302, "grad_norm": 1.609375, "learning_rate": 5.400905488598045e-06, "loss": 0.1688, "step": 41638 }, { "epoch": 30.504761904761907, "grad_norm": 0.95703125, "learning_rate": 5.3994932461850886e-06, "loss": 0.1403, "step": 41639 }, { "epoch": 30.505494505494504, "grad_norm": 1.8359375, "learning_rate": 5.398081219863474e-06, "loss": 0.1346, "step": 41640 }, { "epoch": 30.506227106227108, "grad_norm": 1.140625, "learning_rate": 5.396669409639944e-06, "loss": 0.1005, "step": 41641 }, { "epoch": 30.506959706959705, "grad_norm": 1.5859375, "learning_rate": 5.395257815521292e-06, "loss": 0.1381, "step": 41642 }, { "epoch": 30.50769230769231, "grad_norm": 0.96484375, "learning_rate": 5.393846437514248e-06, "loss": 0.1144, "step": 41643 }, { "epoch": 30.508424908424907, "grad_norm": 0.5859375, "learning_rate": 5.392435275625611e-06, "loss": 0.0781, "step": 41644 }, { "epoch": 30.50915750915751, "grad_norm": 1.3671875, "learning_rate": 5.391024329862114e-06, "loss": 0.1596, "step": 41645 }, { "epoch": 30.509890109890108, "grad_norm": 0.828125, "learning_rate": 5.389613600230538e-06, "loss": 0.1145, "step": 41646 }, { "epoch": 30.510622710622712, "grad_norm": 1.8203125, "learning_rate": 5.388203086737626e-06, "loss": 0.1526, "step": 41647 }, { "epoch": 30.51135531135531, "grad_norm": 0.83203125, "learning_rate": 5.386792789390136e-06, "loss": 0.1164, "step": 41648 }, { "epoch": 30.512087912087914, "grad_norm": 0.98046875, "learning_rate": 5.385382708194836e-06, "loss": 0.1118, "step": 41649 }, { "epoch": 30.51282051282051, "grad_norm": 1.125, "learning_rate": 5.383972843158473e-06, "loss": 0.1051, "step": 41650 }, { "epoch": 30.513553113553115, "grad_norm": 1.265625, "learning_rate": 5.382563194287818e-06, "loss": 0.1057, "step": 41651 }, { "epoch": 30.514285714285712, "grad_norm": 0.90625, "learning_rate": 5.381153761589609e-06, "loss": 0.1356, "step": 41652 }, { "epoch": 30.515018315018317, "grad_norm": 0.69140625, "learning_rate": 5.379744545070599e-06, "loss": 0.1099, "step": 41653 }, { "epoch": 30.515750915750917, "grad_norm": 1.3515625, "learning_rate": 5.378335544737548e-06, "loss": 0.1431, "step": 41654 }, { "epoch": 30.516483516483518, "grad_norm": 1.2890625, "learning_rate": 5.376926760597209e-06, "loss": 0.1326, "step": 41655 }, { "epoch": 30.51721611721612, "grad_norm": 0.86328125, "learning_rate": 5.375518192656323e-06, "loss": 0.1178, "step": 41656 }, { "epoch": 30.51794871794872, "grad_norm": 1.203125, "learning_rate": 5.374109840921646e-06, "loss": 0.1073, "step": 41657 }, { "epoch": 30.51868131868132, "grad_norm": 0.79296875, "learning_rate": 5.372701705399907e-06, "loss": 0.1059, "step": 41658 }, { "epoch": 30.51941391941392, "grad_norm": 1.734375, "learning_rate": 5.371293786097886e-06, "loss": 0.1038, "step": 41659 }, { "epoch": 30.520146520146522, "grad_norm": 1.21875, "learning_rate": 5.3698860830223046e-06, "loss": 0.1309, "step": 41660 }, { "epoch": 30.520879120879123, "grad_norm": 1.1875, "learning_rate": 5.368478596179923e-06, "loss": 0.1188, "step": 41661 }, { "epoch": 30.521611721611723, "grad_norm": 0.765625, "learning_rate": 5.367071325577468e-06, "loss": 0.1257, "step": 41662 }, { "epoch": 30.522344322344324, "grad_norm": 0.6953125, "learning_rate": 5.365664271221686e-06, "loss": 0.1332, "step": 41663 }, { "epoch": 30.523076923076925, "grad_norm": 1.046875, "learning_rate": 5.364257433119323e-06, "loss": 0.1112, "step": 41664 }, { "epoch": 30.523809523809526, "grad_norm": 0.78515625, "learning_rate": 5.362850811277126e-06, "loss": 0.1368, "step": 41665 }, { "epoch": 30.524542124542126, "grad_norm": 1.015625, "learning_rate": 5.361444405701832e-06, "loss": 0.0963, "step": 41666 }, { "epoch": 30.525274725274727, "grad_norm": 1.8359375, "learning_rate": 5.360038216400169e-06, "loss": 0.1661, "step": 41667 }, { "epoch": 30.526007326007328, "grad_norm": 0.9140625, "learning_rate": 5.3586322433788905e-06, "loss": 0.1147, "step": 41668 }, { "epoch": 30.52673992673993, "grad_norm": 1.234375, "learning_rate": 5.357226486644702e-06, "loss": 0.1172, "step": 41669 }, { "epoch": 30.52747252747253, "grad_norm": 0.8828125, "learning_rate": 5.355820946204383e-06, "loss": 0.1195, "step": 41670 }, { "epoch": 30.52820512820513, "grad_norm": 0.8359375, "learning_rate": 5.354415622064636e-06, "loss": 0.1099, "step": 41671 }, { "epoch": 30.52893772893773, "grad_norm": 1.1171875, "learning_rate": 5.353010514232211e-06, "loss": 0.132, "step": 41672 }, { "epoch": 30.52967032967033, "grad_norm": 0.80859375, "learning_rate": 5.351605622713826e-06, "loss": 0.1307, "step": 41673 }, { "epoch": 30.530402930402932, "grad_norm": 1.171875, "learning_rate": 5.350200947516219e-06, "loss": 0.1145, "step": 41674 }, { "epoch": 30.531135531135533, "grad_norm": 1.359375, "learning_rate": 5.348796488646121e-06, "loss": 0.1354, "step": 41675 }, { "epoch": 30.531868131868134, "grad_norm": 0.7421875, "learning_rate": 5.347392246110265e-06, "loss": 0.0992, "step": 41676 }, { "epoch": 30.532600732600734, "grad_norm": 1.2734375, "learning_rate": 5.3459882199153824e-06, "loss": 0.1199, "step": 41677 }, { "epoch": 30.533333333333335, "grad_norm": 1.046875, "learning_rate": 5.344584410068183e-06, "loss": 0.1221, "step": 41678 }, { "epoch": 30.534065934065936, "grad_norm": 1.015625, "learning_rate": 5.343180816575401e-06, "loss": 0.1102, "step": 41679 }, { "epoch": 30.534798534798536, "grad_norm": 0.66796875, "learning_rate": 5.341777439443765e-06, "loss": 0.1041, "step": 41680 }, { "epoch": 30.535531135531137, "grad_norm": 1.296875, "learning_rate": 5.3403742786800106e-06, "loss": 0.1156, "step": 41681 }, { "epoch": 30.536263736263738, "grad_norm": 0.91796875, "learning_rate": 5.338971334290835e-06, "loss": 0.1494, "step": 41682 }, { "epoch": 30.53699633699634, "grad_norm": 0.94140625, "learning_rate": 5.337568606282982e-06, "loss": 0.1241, "step": 41683 }, { "epoch": 30.53772893772894, "grad_norm": 1.1953125, "learning_rate": 5.3361660946631504e-06, "loss": 0.1186, "step": 41684 }, { "epoch": 30.53846153846154, "grad_norm": 0.91015625, "learning_rate": 5.334763799438089e-06, "loss": 0.1286, "step": 41685 }, { "epoch": 30.53919413919414, "grad_norm": 1.015625, "learning_rate": 5.333361720614491e-06, "loss": 0.1124, "step": 41686 }, { "epoch": 30.53992673992674, "grad_norm": 1.1015625, "learning_rate": 5.331959858199094e-06, "loss": 0.15, "step": 41687 }, { "epoch": 30.540659340659342, "grad_norm": 0.84375, "learning_rate": 5.330558212198593e-06, "loss": 0.1117, "step": 41688 }, { "epoch": 30.541391941391943, "grad_norm": 0.671875, "learning_rate": 5.329156782619718e-06, "loss": 0.1232, "step": 41689 }, { "epoch": 30.542124542124544, "grad_norm": 1.140625, "learning_rate": 5.327755569469181e-06, "loss": 0.117, "step": 41690 }, { "epoch": 30.542857142857144, "grad_norm": 1.3203125, "learning_rate": 5.326354572753696e-06, "loss": 0.0839, "step": 41691 }, { "epoch": 30.543589743589745, "grad_norm": 0.81640625, "learning_rate": 5.324953792479986e-06, "loss": 0.1042, "step": 41692 }, { "epoch": 30.544322344322346, "grad_norm": 1.125, "learning_rate": 5.323553228654743e-06, "loss": 0.139, "step": 41693 }, { "epoch": 30.545054945054947, "grad_norm": 1.4453125, "learning_rate": 5.322152881284695e-06, "loss": 0.1274, "step": 41694 }, { "epoch": 30.545787545787547, "grad_norm": 1.3515625, "learning_rate": 5.320752750376529e-06, "loss": 0.118, "step": 41695 }, { "epoch": 30.546520146520148, "grad_norm": 0.828125, "learning_rate": 5.319352835936983e-06, "loss": 0.1251, "step": 41696 }, { "epoch": 30.54725274725275, "grad_norm": 1.6171875, "learning_rate": 5.31795313797274e-06, "loss": 0.1428, "step": 41697 }, { "epoch": 30.54798534798535, "grad_norm": 0.90625, "learning_rate": 5.316553656490527e-06, "loss": 0.117, "step": 41698 }, { "epoch": 30.54871794871795, "grad_norm": 1.0234375, "learning_rate": 5.3151543914970235e-06, "loss": 0.1197, "step": 41699 }, { "epoch": 30.54945054945055, "grad_norm": 1.046875, "learning_rate": 5.313755342998955e-06, "loss": 0.1494, "step": 41700 }, { "epoch": 30.550183150183152, "grad_norm": 1.015625, "learning_rate": 5.312356511003018e-06, "loss": 0.1151, "step": 41701 }, { "epoch": 30.550915750915753, "grad_norm": 0.75390625, "learning_rate": 5.310957895515912e-06, "loss": 0.1014, "step": 41702 }, { "epoch": 30.551648351648353, "grad_norm": 1.1015625, "learning_rate": 5.309559496544349e-06, "loss": 0.148, "step": 41703 }, { "epoch": 30.552380952380954, "grad_norm": 0.875, "learning_rate": 5.3081613140950146e-06, "loss": 0.1134, "step": 41704 }, { "epoch": 30.553113553113555, "grad_norm": 1.125, "learning_rate": 5.306763348174615e-06, "loss": 0.1209, "step": 41705 }, { "epoch": 30.553846153846155, "grad_norm": 0.8515625, "learning_rate": 5.305365598789848e-06, "loss": 0.0931, "step": 41706 }, { "epoch": 30.554578754578756, "grad_norm": 1.3203125, "learning_rate": 5.303968065947425e-06, "loss": 0.1229, "step": 41707 }, { "epoch": 30.555311355311357, "grad_norm": 0.96875, "learning_rate": 5.302570749654012e-06, "loss": 0.1283, "step": 41708 }, { "epoch": 30.556043956043958, "grad_norm": 0.90234375, "learning_rate": 5.30117364991633e-06, "loss": 0.1251, "step": 41709 }, { "epoch": 30.55677655677656, "grad_norm": 0.71875, "learning_rate": 5.299776766741044e-06, "loss": 0.0897, "step": 41710 }, { "epoch": 30.55750915750916, "grad_norm": 0.89453125, "learning_rate": 5.298380100134888e-06, "loss": 0.0997, "step": 41711 }, { "epoch": 30.55824175824176, "grad_norm": 0.9140625, "learning_rate": 5.296983650104521e-06, "loss": 0.1323, "step": 41712 }, { "epoch": 30.55897435897436, "grad_norm": 0.90234375, "learning_rate": 5.29558741665665e-06, "loss": 0.1362, "step": 41713 }, { "epoch": 30.55970695970696, "grad_norm": 0.97265625, "learning_rate": 5.294191399797954e-06, "loss": 0.1273, "step": 41714 }, { "epoch": 30.560439560439562, "grad_norm": 1.140625, "learning_rate": 5.292795599535125e-06, "loss": 0.1497, "step": 41715 }, { "epoch": 30.561172161172163, "grad_norm": 0.9609375, "learning_rate": 5.291400015874855e-06, "loss": 0.1356, "step": 41716 }, { "epoch": 30.561904761904763, "grad_norm": 1.171875, "learning_rate": 5.29000464882383e-06, "loss": 0.1547, "step": 41717 }, { "epoch": 30.562637362637364, "grad_norm": 0.85546875, "learning_rate": 5.288609498388737e-06, "loss": 0.1468, "step": 41718 }, { "epoch": 30.563369963369965, "grad_norm": 0.83203125, "learning_rate": 5.287214564576255e-06, "loss": 0.1271, "step": 41719 }, { "epoch": 30.564102564102566, "grad_norm": 1.0703125, "learning_rate": 5.285819847393079e-06, "loss": 0.1019, "step": 41720 }, { "epoch": 30.564835164835166, "grad_norm": 1.015625, "learning_rate": 5.284425346845859e-06, "loss": 0.1396, "step": 41721 }, { "epoch": 30.565567765567767, "grad_norm": 0.98828125, "learning_rate": 5.283031062941321e-06, "loss": 0.1269, "step": 41722 }, { "epoch": 30.566300366300368, "grad_norm": 1.265625, "learning_rate": 5.28163699568612e-06, "loss": 0.1639, "step": 41723 }, { "epoch": 30.56703296703297, "grad_norm": 1.015625, "learning_rate": 5.280243145086946e-06, "loss": 0.1344, "step": 41724 }, { "epoch": 30.56776556776557, "grad_norm": 1.3046875, "learning_rate": 5.2788495111504634e-06, "loss": 0.1305, "step": 41725 }, { "epoch": 30.56849816849817, "grad_norm": 0.90234375, "learning_rate": 5.277456093883361e-06, "loss": 0.1162, "step": 41726 }, { "epoch": 30.56923076923077, "grad_norm": 0.90234375, "learning_rate": 5.2760628932923095e-06, "loss": 0.1376, "step": 41727 }, { "epoch": 30.56996336996337, "grad_norm": 0.73046875, "learning_rate": 5.274669909383982e-06, "loss": 0.1087, "step": 41728 }, { "epoch": 30.570695970695972, "grad_norm": 0.9609375, "learning_rate": 5.27327714216507e-06, "loss": 0.1185, "step": 41729 }, { "epoch": 30.571428571428573, "grad_norm": 1.2890625, "learning_rate": 5.271884591642226e-06, "loss": 0.103, "step": 41730 }, { "epoch": 30.572161172161174, "grad_norm": 0.85546875, "learning_rate": 5.270492257822131e-06, "loss": 0.1006, "step": 41731 }, { "epoch": 30.572893772893774, "grad_norm": 1.1015625, "learning_rate": 5.269100140711457e-06, "loss": 0.1542, "step": 41732 }, { "epoch": 30.573626373626375, "grad_norm": 1.2734375, "learning_rate": 5.267708240316879e-06, "loss": 0.1273, "step": 41733 }, { "epoch": 30.574358974358976, "grad_norm": 1.1015625, "learning_rate": 5.266316556645051e-06, "loss": 0.1359, "step": 41734 }, { "epoch": 30.575091575091577, "grad_norm": 0.76171875, "learning_rate": 5.264925089702659e-06, "loss": 0.1036, "step": 41735 }, { "epoch": 30.575824175824177, "grad_norm": 0.79296875, "learning_rate": 5.26353383949634e-06, "loss": 0.1035, "step": 41736 }, { "epoch": 30.576556776556778, "grad_norm": 1.1328125, "learning_rate": 5.262142806032802e-06, "loss": 0.114, "step": 41737 }, { "epoch": 30.57728937728938, "grad_norm": 0.83984375, "learning_rate": 5.260751989318676e-06, "loss": 0.1204, "step": 41738 }, { "epoch": 30.57802197802198, "grad_norm": 1.171875, "learning_rate": 5.259361389360649e-06, "loss": 0.1006, "step": 41739 }, { "epoch": 30.57875457875458, "grad_norm": 1.1484375, "learning_rate": 5.257971006165368e-06, "loss": 0.1152, "step": 41740 }, { "epoch": 30.57948717948718, "grad_norm": 0.9375, "learning_rate": 5.256580839739493e-06, "loss": 0.1144, "step": 41741 }, { "epoch": 30.58021978021978, "grad_norm": 0.83984375, "learning_rate": 5.255190890089694e-06, "loss": 0.0909, "step": 41742 }, { "epoch": 30.580952380952382, "grad_norm": 1.40625, "learning_rate": 5.2538011572226336e-06, "loss": 0.141, "step": 41743 }, { "epoch": 30.581684981684983, "grad_norm": 1.578125, "learning_rate": 5.252411641144971e-06, "loss": 0.1256, "step": 41744 }, { "epoch": 30.582417582417584, "grad_norm": 0.79296875, "learning_rate": 5.251022341863346e-06, "loss": 0.1188, "step": 41745 }, { "epoch": 30.583150183150185, "grad_norm": 0.8515625, "learning_rate": 5.24963325938444e-06, "loss": 0.113, "step": 41746 }, { "epoch": 30.583882783882785, "grad_norm": 1.4140625, "learning_rate": 5.248244393714871e-06, "loss": 0.158, "step": 41747 }, { "epoch": 30.584615384615386, "grad_norm": 1.265625, "learning_rate": 5.246855744861343e-06, "loss": 0.1039, "step": 41748 }, { "epoch": 30.585347985347987, "grad_norm": 1.015625, "learning_rate": 5.245467312830473e-06, "loss": 0.102, "step": 41749 }, { "epoch": 30.586080586080588, "grad_norm": 1.3984375, "learning_rate": 5.244079097628937e-06, "loss": 0.1178, "step": 41750 }, { "epoch": 30.58681318681319, "grad_norm": 0.9375, "learning_rate": 5.242691099263365e-06, "loss": 0.1521, "step": 41751 }, { "epoch": 30.58754578754579, "grad_norm": 1.3828125, "learning_rate": 5.241303317740416e-06, "loss": 0.1221, "step": 41752 }, { "epoch": 30.58827838827839, "grad_norm": 1.03125, "learning_rate": 5.2399157530667406e-06, "loss": 0.0927, "step": 41753 }, { "epoch": 30.58901098901099, "grad_norm": 1.0859375, "learning_rate": 5.2385284052489814e-06, "loss": 0.1211, "step": 41754 }, { "epoch": 30.58974358974359, "grad_norm": 1.03125, "learning_rate": 5.237141274293807e-06, "loss": 0.144, "step": 41755 }, { "epoch": 30.590476190476192, "grad_norm": 0.8671875, "learning_rate": 5.235754360207837e-06, "loss": 0.1003, "step": 41756 }, { "epoch": 30.591208791208793, "grad_norm": 0.9453125, "learning_rate": 5.234367662997727e-06, "loss": 0.1214, "step": 41757 }, { "epoch": 30.591941391941393, "grad_norm": 1.4375, "learning_rate": 5.232981182670117e-06, "loss": 0.1449, "step": 41758 }, { "epoch": 30.592673992673994, "grad_norm": 0.67578125, "learning_rate": 5.231594919231668e-06, "loss": 0.0796, "step": 41759 }, { "epoch": 30.593406593406595, "grad_norm": 0.82421875, "learning_rate": 5.230208872689001e-06, "loss": 0.1129, "step": 41760 }, { "epoch": 30.594139194139196, "grad_norm": 1.2890625, "learning_rate": 5.228823043048773e-06, "loss": 0.1369, "step": 41761 }, { "epoch": 30.594871794871796, "grad_norm": 1.0859375, "learning_rate": 5.2274374303175905e-06, "loss": 0.1288, "step": 41762 }, { "epoch": 30.595604395604397, "grad_norm": 1.3984375, "learning_rate": 5.226052034502141e-06, "loss": 0.1175, "step": 41763 }, { "epoch": 30.596336996336998, "grad_norm": 0.84765625, "learning_rate": 5.224666855609028e-06, "loss": 0.114, "step": 41764 }, { "epoch": 30.5970695970696, "grad_norm": 1.1953125, "learning_rate": 5.223281893644909e-06, "loss": 0.1421, "step": 41765 }, { "epoch": 30.5978021978022, "grad_norm": 1.0625, "learning_rate": 5.221897148616403e-06, "loss": 0.1078, "step": 41766 }, { "epoch": 30.5985347985348, "grad_norm": 0.9140625, "learning_rate": 5.22051262053015e-06, "loss": 0.1209, "step": 41767 }, { "epoch": 30.5992673992674, "grad_norm": 1.3125, "learning_rate": 5.219128309392784e-06, "loss": 0.1434, "step": 41768 }, { "epoch": 30.6, "grad_norm": 0.953125, "learning_rate": 5.2177442152109375e-06, "loss": 0.1238, "step": 41769 }, { "epoch": 30.600732600732602, "grad_norm": 1.328125, "learning_rate": 5.216360337991255e-06, "loss": 0.136, "step": 41770 }, { "epoch": 30.601465201465203, "grad_norm": 0.80078125, "learning_rate": 5.214976677740347e-06, "loss": 0.1183, "step": 41771 }, { "epoch": 30.602197802197804, "grad_norm": 0.88671875, "learning_rate": 5.213593234464859e-06, "loss": 0.1036, "step": 41772 }, { "epoch": 30.602930402930404, "grad_norm": 0.87890625, "learning_rate": 5.212210008171395e-06, "loss": 0.1119, "step": 41773 }, { "epoch": 30.603663003663005, "grad_norm": 0.99609375, "learning_rate": 5.210826998866622e-06, "loss": 0.1182, "step": 41774 }, { "epoch": 30.604395604395606, "grad_norm": 1.359375, "learning_rate": 5.209444206557129e-06, "loss": 0.1078, "step": 41775 }, { "epoch": 30.605128205128207, "grad_norm": 1.2890625, "learning_rate": 5.208061631249565e-06, "loss": 0.1511, "step": 41776 }, { "epoch": 30.605860805860807, "grad_norm": 0.6640625, "learning_rate": 5.2066792729505444e-06, "loss": 0.0922, "step": 41777 }, { "epoch": 30.606593406593408, "grad_norm": 1.171875, "learning_rate": 5.205297131666687e-06, "loss": 0.1042, "step": 41778 }, { "epoch": 30.60732600732601, "grad_norm": 1.046875, "learning_rate": 5.203915207404625e-06, "loss": 0.1279, "step": 41779 }, { "epoch": 30.60805860805861, "grad_norm": 1.0390625, "learning_rate": 5.20253350017097e-06, "loss": 0.1058, "step": 41780 }, { "epoch": 30.60879120879121, "grad_norm": 1.6328125, "learning_rate": 5.201152009972357e-06, "loss": 0.127, "step": 41781 }, { "epoch": 30.60952380952381, "grad_norm": 1.1328125, "learning_rate": 5.199770736815385e-06, "loss": 0.1143, "step": 41782 }, { "epoch": 30.61025641025641, "grad_norm": 1.015625, "learning_rate": 5.198389680706688e-06, "loss": 0.1391, "step": 41783 }, { "epoch": 30.610989010989012, "grad_norm": 0.703125, "learning_rate": 5.197008841652876e-06, "loss": 0.1011, "step": 41784 }, { "epoch": 30.611721611721613, "grad_norm": 1.125, "learning_rate": 5.195628219660571e-06, "loss": 0.1336, "step": 41785 }, { "epoch": 30.612454212454214, "grad_norm": 0.8046875, "learning_rate": 5.194247814736375e-06, "loss": 0.1126, "step": 41786 }, { "epoch": 30.613186813186815, "grad_norm": 1.453125, "learning_rate": 5.192867626886926e-06, "loss": 0.1327, "step": 41787 }, { "epoch": 30.613919413919415, "grad_norm": 0.859375, "learning_rate": 5.191487656118802e-06, "loss": 0.108, "step": 41788 }, { "epoch": 30.614652014652016, "grad_norm": 1.453125, "learning_rate": 5.190107902438646e-06, "loss": 0.122, "step": 41789 }, { "epoch": 30.615384615384617, "grad_norm": 1.109375, "learning_rate": 5.188728365853055e-06, "loss": 0.1398, "step": 41790 }, { "epoch": 30.616117216117217, "grad_norm": 1.0703125, "learning_rate": 5.187349046368638e-06, "loss": 0.144, "step": 41791 }, { "epoch": 30.616849816849818, "grad_norm": 0.96484375, "learning_rate": 5.185969943992018e-06, "loss": 0.1034, "step": 41792 }, { "epoch": 30.61758241758242, "grad_norm": 0.98046875, "learning_rate": 5.184591058729779e-06, "loss": 0.1313, "step": 41793 }, { "epoch": 30.61831501831502, "grad_norm": 1.546875, "learning_rate": 5.183212390588545e-06, "loss": 0.1057, "step": 41794 }, { "epoch": 30.61904761904762, "grad_norm": 0.9375, "learning_rate": 5.181833939574912e-06, "loss": 0.1233, "step": 41795 }, { "epoch": 30.61978021978022, "grad_norm": 0.87109375, "learning_rate": 5.1804557056955e-06, "loss": 0.1241, "step": 41796 }, { "epoch": 30.620512820512822, "grad_norm": 1.140625, "learning_rate": 5.179077688956893e-06, "loss": 0.1103, "step": 41797 }, { "epoch": 30.621245421245423, "grad_norm": 0.91015625, "learning_rate": 5.177699889365715e-06, "loss": 0.1236, "step": 41798 }, { "epoch": 30.621978021978023, "grad_norm": 1.25, "learning_rate": 5.176322306928533e-06, "loss": 0.1157, "step": 41799 }, { "epoch": 30.622710622710624, "grad_norm": 1.1171875, "learning_rate": 5.174944941651993e-06, "loss": 0.1165, "step": 41800 }, { "epoch": 30.623443223443225, "grad_norm": 0.9453125, "learning_rate": 5.173567793542658e-06, "loss": 0.102, "step": 41801 }, { "epoch": 30.624175824175826, "grad_norm": 1.3515625, "learning_rate": 5.1721908626071484e-06, "loss": 0.1109, "step": 41802 }, { "epoch": 30.624908424908426, "grad_norm": 1.0234375, "learning_rate": 5.170814148852046e-06, "loss": 0.108, "step": 41803 }, { "epoch": 30.625641025641027, "grad_norm": 0.83203125, "learning_rate": 5.169437652283948e-06, "loss": 0.126, "step": 41804 }, { "epoch": 30.626373626373628, "grad_norm": 1.09375, "learning_rate": 5.168061372909461e-06, "loss": 0.1299, "step": 41805 }, { "epoch": 30.62710622710623, "grad_norm": 1.3671875, "learning_rate": 5.166685310735168e-06, "loss": 0.1204, "step": 41806 }, { "epoch": 30.62783882783883, "grad_norm": 0.921875, "learning_rate": 5.165309465767683e-06, "loss": 0.1331, "step": 41807 }, { "epoch": 30.62857142857143, "grad_norm": 1.1328125, "learning_rate": 5.163933838013569e-06, "loss": 0.1324, "step": 41808 }, { "epoch": 30.62930402930403, "grad_norm": 0.828125, "learning_rate": 5.1625584274794355e-06, "loss": 0.1234, "step": 41809 }, { "epoch": 30.63003663003663, "grad_norm": 0.62109375, "learning_rate": 5.161183234171865e-06, "loss": 0.1059, "step": 41810 }, { "epoch": 30.630769230769232, "grad_norm": 1.1875, "learning_rate": 5.159808258097458e-06, "loss": 0.1526, "step": 41811 }, { "epoch": 30.631501831501833, "grad_norm": 0.953125, "learning_rate": 5.158433499262785e-06, "loss": 0.1144, "step": 41812 }, { "epoch": 30.632234432234434, "grad_norm": 1.2890625, "learning_rate": 5.1570589576744485e-06, "loss": 0.1349, "step": 41813 }, { "epoch": 30.632967032967034, "grad_norm": 1.0546875, "learning_rate": 5.155684633339016e-06, "loss": 0.108, "step": 41814 }, { "epoch": 30.633699633699635, "grad_norm": 0.8046875, "learning_rate": 5.154310526263096e-06, "loss": 0.1158, "step": 41815 }, { "epoch": 30.634432234432236, "grad_norm": 1.3359375, "learning_rate": 5.152936636453254e-06, "loss": 0.1204, "step": 41816 }, { "epoch": 30.635164835164836, "grad_norm": 0.796875, "learning_rate": 5.151562963916076e-06, "loss": 0.1232, "step": 41817 }, { "epoch": 30.635897435897437, "grad_norm": 0.9375, "learning_rate": 5.15018950865816e-06, "loss": 0.1351, "step": 41818 }, { "epoch": 30.636630036630038, "grad_norm": 1.328125, "learning_rate": 5.148816270686064e-06, "loss": 0.1345, "step": 41819 }, { "epoch": 30.63736263736264, "grad_norm": 0.8046875, "learning_rate": 5.147443250006374e-06, "loss": 0.1061, "step": 41820 }, { "epoch": 30.63809523809524, "grad_norm": 0.91015625, "learning_rate": 5.146070446625674e-06, "loss": 0.1521, "step": 41821 }, { "epoch": 30.63882783882784, "grad_norm": 0.69140625, "learning_rate": 5.1446978605505455e-06, "loss": 0.1107, "step": 41822 }, { "epoch": 30.63956043956044, "grad_norm": 0.94140625, "learning_rate": 5.143325491787552e-06, "loss": 0.1153, "step": 41823 }, { "epoch": 30.64029304029304, "grad_norm": 1.1171875, "learning_rate": 5.1419533403432845e-06, "loss": 0.1771, "step": 41824 }, { "epoch": 30.641025641025642, "grad_norm": 1.171875, "learning_rate": 5.140581406224296e-06, "loss": 0.119, "step": 41825 }, { "epoch": 30.641758241758243, "grad_norm": 1.203125, "learning_rate": 5.139209689437183e-06, "loss": 0.1135, "step": 41826 }, { "epoch": 30.642490842490844, "grad_norm": 1.1796875, "learning_rate": 5.137838189988501e-06, "loss": 0.1302, "step": 41827 }, { "epoch": 30.643223443223444, "grad_norm": 1.3671875, "learning_rate": 5.136466907884834e-06, "loss": 0.131, "step": 41828 }, { "epoch": 30.643956043956045, "grad_norm": 0.703125, "learning_rate": 5.135095843132742e-06, "loss": 0.1122, "step": 41829 }, { "epoch": 30.644688644688646, "grad_norm": 1.53125, "learning_rate": 5.133724995738796e-06, "loss": 0.1432, "step": 41830 }, { "epoch": 30.645421245421247, "grad_norm": 1.1640625, "learning_rate": 5.132354365709562e-06, "loss": 0.1325, "step": 41831 }, { "epoch": 30.646153846153847, "grad_norm": 0.828125, "learning_rate": 5.130983953051616e-06, "loss": 0.1212, "step": 41832 }, { "epoch": 30.646886446886448, "grad_norm": 0.54296875, "learning_rate": 5.129613757771526e-06, "loss": 0.1002, "step": 41833 }, { "epoch": 30.64761904761905, "grad_norm": 0.96875, "learning_rate": 5.128243779875845e-06, "loss": 0.1125, "step": 41834 }, { "epoch": 30.64835164835165, "grad_norm": 0.9375, "learning_rate": 5.1268740193711435e-06, "loss": 0.1172, "step": 41835 }, { "epoch": 30.64908424908425, "grad_norm": 1.4140625, "learning_rate": 5.125504476263979e-06, "loss": 0.1449, "step": 41836 }, { "epoch": 30.64981684981685, "grad_norm": 1.1640625, "learning_rate": 5.1241351505609285e-06, "loss": 0.1247, "step": 41837 }, { "epoch": 30.650549450549452, "grad_norm": 0.9453125, "learning_rate": 5.122766042268533e-06, "loss": 0.1054, "step": 41838 }, { "epoch": 30.651282051282053, "grad_norm": 1.25, "learning_rate": 5.121397151393374e-06, "loss": 0.0943, "step": 41839 }, { "epoch": 30.652014652014653, "grad_norm": 0.60546875, "learning_rate": 5.120028477941979e-06, "loss": 0.1087, "step": 41840 }, { "epoch": 30.652747252747254, "grad_norm": 0.6796875, "learning_rate": 5.118660021920939e-06, "loss": 0.1344, "step": 41841 }, { "epoch": 30.653479853479855, "grad_norm": 0.98046875, "learning_rate": 5.117291783336789e-06, "loss": 0.1248, "step": 41842 }, { "epoch": 30.654212454212455, "grad_norm": 0.8515625, "learning_rate": 5.115923762196092e-06, "loss": 0.1076, "step": 41843 }, { "epoch": 30.654945054945056, "grad_norm": 1.0859375, "learning_rate": 5.114555958505415e-06, "loss": 0.1365, "step": 41844 }, { "epoch": 30.655677655677657, "grad_norm": 1.0234375, "learning_rate": 5.113188372271291e-06, "loss": 0.1087, "step": 41845 }, { "epoch": 30.656410256410258, "grad_norm": 0.92578125, "learning_rate": 5.111821003500278e-06, "loss": 0.1063, "step": 41846 }, { "epoch": 30.65714285714286, "grad_norm": 1.3671875, "learning_rate": 5.110453852198929e-06, "loss": 0.1295, "step": 41847 }, { "epoch": 30.65787545787546, "grad_norm": 1.5078125, "learning_rate": 5.109086918373811e-06, "loss": 0.1226, "step": 41848 }, { "epoch": 30.65860805860806, "grad_norm": 0.8203125, "learning_rate": 5.107720202031442e-06, "loss": 0.1398, "step": 41849 }, { "epoch": 30.65934065934066, "grad_norm": 0.921875, "learning_rate": 5.106353703178399e-06, "loss": 0.1335, "step": 41850 }, { "epoch": 30.66007326007326, "grad_norm": 1.1953125, "learning_rate": 5.104987421821201e-06, "loss": 0.1302, "step": 41851 }, { "epoch": 30.660805860805862, "grad_norm": 1.1328125, "learning_rate": 5.103621357966425e-06, "loss": 0.1219, "step": 41852 }, { "epoch": 30.661538461538463, "grad_norm": 1.546875, "learning_rate": 5.102255511620596e-06, "loss": 0.1409, "step": 41853 }, { "epoch": 30.662271062271063, "grad_norm": 1.0625, "learning_rate": 5.100889882790274e-06, "loss": 0.1309, "step": 41854 }, { "epoch": 30.663003663003664, "grad_norm": 1.0625, "learning_rate": 5.09952447148198e-06, "loss": 0.1133, "step": 41855 }, { "epoch": 30.663736263736265, "grad_norm": 1.078125, "learning_rate": 5.09815927770227e-06, "loss": 0.1536, "step": 41856 }, { "epoch": 30.664468864468866, "grad_norm": 0.83984375, "learning_rate": 5.096794301457679e-06, "loss": 0.0932, "step": 41857 }, { "epoch": 30.665201465201466, "grad_norm": 1.34375, "learning_rate": 5.095429542754758e-06, "loss": 0.1186, "step": 41858 }, { "epoch": 30.665934065934067, "grad_norm": 0.9921875, "learning_rate": 5.094065001600045e-06, "loss": 0.1289, "step": 41859 }, { "epoch": 30.666666666666668, "grad_norm": 0.9296875, "learning_rate": 5.0927006780000615e-06, "loss": 0.1069, "step": 41860 }, { "epoch": 30.66739926739927, "grad_norm": 1.2421875, "learning_rate": 5.091336571961361e-06, "loss": 0.1239, "step": 41861 }, { "epoch": 30.66813186813187, "grad_norm": 1.046875, "learning_rate": 5.0899726834904665e-06, "loss": 0.1424, "step": 41862 }, { "epoch": 30.66886446886447, "grad_norm": 0.87109375, "learning_rate": 5.088609012593935e-06, "loss": 0.1357, "step": 41863 }, { "epoch": 30.66959706959707, "grad_norm": 1.125, "learning_rate": 5.087245559278274e-06, "loss": 0.1045, "step": 41864 }, { "epoch": 30.67032967032967, "grad_norm": 1.0859375, "learning_rate": 5.085882323550033e-06, "loss": 0.1393, "step": 41865 }, { "epoch": 30.671062271062272, "grad_norm": 0.9921875, "learning_rate": 5.084519305415725e-06, "loss": 0.1237, "step": 41866 }, { "epoch": 30.671794871794873, "grad_norm": 0.72265625, "learning_rate": 5.0831565048819126e-06, "loss": 0.0976, "step": 41867 }, { "epoch": 30.672527472527474, "grad_norm": 1.3203125, "learning_rate": 5.0817939219550945e-06, "loss": 0.11, "step": 41868 }, { "epoch": 30.673260073260074, "grad_norm": 0.9921875, "learning_rate": 5.080431556641812e-06, "loss": 0.1041, "step": 41869 }, { "epoch": 30.673992673992675, "grad_norm": 0.85546875, "learning_rate": 5.079069408948602e-06, "loss": 0.1347, "step": 41870 }, { "epoch": 30.674725274725276, "grad_norm": 1.171875, "learning_rate": 5.077707478881973e-06, "loss": 0.1053, "step": 41871 }, { "epoch": 30.675457875457877, "grad_norm": 1.4921875, "learning_rate": 5.076345766448459e-06, "loss": 0.1231, "step": 41872 }, { "epoch": 30.676190476190477, "grad_norm": 0.95703125, "learning_rate": 5.074984271654585e-06, "loss": 0.1093, "step": 41873 }, { "epoch": 30.676923076923078, "grad_norm": 0.70703125, "learning_rate": 5.073622994506876e-06, "loss": 0.1063, "step": 41874 }, { "epoch": 30.67765567765568, "grad_norm": 1.046875, "learning_rate": 5.072261935011853e-06, "loss": 0.0971, "step": 41875 }, { "epoch": 30.67838827838828, "grad_norm": 1.1015625, "learning_rate": 5.070901093176038e-06, "loss": 0.1172, "step": 41876 }, { "epoch": 30.67912087912088, "grad_norm": 0.9375, "learning_rate": 5.0695404690059306e-06, "loss": 0.133, "step": 41877 }, { "epoch": 30.67985347985348, "grad_norm": 1.1015625, "learning_rate": 5.068180062508089e-06, "loss": 0.1046, "step": 41878 }, { "epoch": 30.68058608058608, "grad_norm": 0.67578125, "learning_rate": 5.066819873688999e-06, "loss": 0.0988, "step": 41879 }, { "epoch": 30.681318681318682, "grad_norm": 1.0625, "learning_rate": 5.065459902555202e-06, "loss": 0.1248, "step": 41880 }, { "epoch": 30.682051282051283, "grad_norm": 1.0546875, "learning_rate": 5.064100149113192e-06, "loss": 0.1271, "step": 41881 }, { "epoch": 30.682783882783884, "grad_norm": 1.203125, "learning_rate": 5.062740613369493e-06, "loss": 0.1167, "step": 41882 }, { "epoch": 30.683516483516485, "grad_norm": 0.63671875, "learning_rate": 5.061381295330622e-06, "loss": 0.108, "step": 41883 }, { "epoch": 30.684249084249085, "grad_norm": 1.3046875, "learning_rate": 5.060022195003087e-06, "loss": 0.1256, "step": 41884 }, { "epoch": 30.684981684981686, "grad_norm": 0.9609375, "learning_rate": 5.0586633123934174e-06, "loss": 0.1085, "step": 41885 }, { "epoch": 30.685714285714287, "grad_norm": 0.98046875, "learning_rate": 5.0573046475080955e-06, "loss": 0.1207, "step": 41886 }, { "epoch": 30.686446886446888, "grad_norm": 0.7421875, "learning_rate": 5.0559462003536405e-06, "loss": 0.1406, "step": 41887 }, { "epoch": 30.68717948717949, "grad_norm": 1.140625, "learning_rate": 5.054587970936571e-06, "loss": 0.1395, "step": 41888 }, { "epoch": 30.68791208791209, "grad_norm": 1.1171875, "learning_rate": 5.053229959263398e-06, "loss": 0.1194, "step": 41889 }, { "epoch": 30.68864468864469, "grad_norm": 0.921875, "learning_rate": 5.051872165340611e-06, "loss": 0.103, "step": 41890 }, { "epoch": 30.68937728937729, "grad_norm": 1.125, "learning_rate": 5.050514589174731e-06, "loss": 0.1045, "step": 41891 }, { "epoch": 30.69010989010989, "grad_norm": 1.1484375, "learning_rate": 5.049157230772239e-06, "loss": 0.1144, "step": 41892 }, { "epoch": 30.690842490842492, "grad_norm": 0.78125, "learning_rate": 5.047800090139672e-06, "loss": 0.1451, "step": 41893 }, { "epoch": 30.691575091575093, "grad_norm": 1.6796875, "learning_rate": 5.046443167283509e-06, "loss": 0.1398, "step": 41894 }, { "epoch": 30.692307692307693, "grad_norm": 0.859375, "learning_rate": 5.04508646221026e-06, "loss": 0.1355, "step": 41895 }, { "epoch": 30.693040293040294, "grad_norm": 1.296875, "learning_rate": 5.04372997492643e-06, "loss": 0.1276, "step": 41896 }, { "epoch": 30.693772893772895, "grad_norm": 1.0, "learning_rate": 5.0423737054385e-06, "loss": 0.1067, "step": 41897 }, { "epoch": 30.694505494505496, "grad_norm": 1.375, "learning_rate": 5.041017653752986e-06, "loss": 0.1088, "step": 41898 }, { "epoch": 30.695238095238096, "grad_norm": 0.76953125, "learning_rate": 5.039661819876375e-06, "loss": 0.1137, "step": 41899 }, { "epoch": 30.695970695970697, "grad_norm": 0.98046875, "learning_rate": 5.0383062038151755e-06, "loss": 0.1085, "step": 41900 }, { "epoch": 30.696703296703298, "grad_norm": 1.21875, "learning_rate": 5.036950805575871e-06, "loss": 0.1144, "step": 41901 }, { "epoch": 30.6974358974359, "grad_norm": 1.4140625, "learning_rate": 5.035595625164966e-06, "loss": 0.1068, "step": 41902 }, { "epoch": 30.6981684981685, "grad_norm": 1.3125, "learning_rate": 5.0342406625889305e-06, "loss": 0.1463, "step": 41903 }, { "epoch": 30.6989010989011, "grad_norm": 1.1875, "learning_rate": 5.03288591785429e-06, "loss": 0.1208, "step": 41904 }, { "epoch": 30.6996336996337, "grad_norm": 1.4609375, "learning_rate": 5.031531390967511e-06, "loss": 0.129, "step": 41905 }, { "epoch": 30.7003663003663, "grad_norm": 1.046875, "learning_rate": 5.030177081935093e-06, "loss": 0.1242, "step": 41906 }, { "epoch": 30.701098901098902, "grad_norm": 0.77734375, "learning_rate": 5.028822990763533e-06, "loss": 0.1092, "step": 41907 }, { "epoch": 30.701831501831503, "grad_norm": 1.109375, "learning_rate": 5.027469117459296e-06, "loss": 0.1366, "step": 41908 }, { "epoch": 30.702564102564104, "grad_norm": 0.8984375, "learning_rate": 5.026115462028884e-06, "loss": 0.107, "step": 41909 }, { "epoch": 30.703296703296704, "grad_norm": 1.15625, "learning_rate": 5.0247620244787795e-06, "loss": 0.1278, "step": 41910 }, { "epoch": 30.704029304029305, "grad_norm": 1.15625, "learning_rate": 5.023408804815481e-06, "loss": 0.1171, "step": 41911 }, { "epoch": 30.704761904761906, "grad_norm": 1.21875, "learning_rate": 5.022055803045449e-06, "loss": 0.1077, "step": 41912 }, { "epoch": 30.705494505494507, "grad_norm": 1.1953125, "learning_rate": 5.020703019175181e-06, "loss": 0.1135, "step": 41913 }, { "epoch": 30.706227106227107, "grad_norm": 1.0703125, "learning_rate": 5.019350453211148e-06, "loss": 0.142, "step": 41914 }, { "epoch": 30.706959706959708, "grad_norm": 1.0703125, "learning_rate": 5.017998105159851e-06, "loss": 0.1361, "step": 41915 }, { "epoch": 30.70769230769231, "grad_norm": 1.0546875, "learning_rate": 5.016645975027741e-06, "loss": 0.1371, "step": 41916 }, { "epoch": 30.70842490842491, "grad_norm": 0.8515625, "learning_rate": 5.015294062821324e-06, "loss": 0.1157, "step": 41917 }, { "epoch": 30.70915750915751, "grad_norm": 0.796875, "learning_rate": 5.01394236854705e-06, "loss": 0.1451, "step": 41918 }, { "epoch": 30.70989010989011, "grad_norm": 0.8125, "learning_rate": 5.0125908922114276e-06, "loss": 0.1148, "step": 41919 }, { "epoch": 30.71062271062271, "grad_norm": 1.359375, "learning_rate": 5.011239633820899e-06, "loss": 0.1112, "step": 41920 }, { "epoch": 30.711355311355312, "grad_norm": 1.015625, "learning_rate": 5.009888593381961e-06, "loss": 0.1474, "step": 41921 }, { "epoch": 30.712087912087913, "grad_norm": 0.8046875, "learning_rate": 5.0085377709010855e-06, "loss": 0.1333, "step": 41922 }, { "epoch": 30.712820512820514, "grad_norm": 1.125, "learning_rate": 5.0071871663847346e-06, "loss": 0.133, "step": 41923 }, { "epoch": 30.713553113553115, "grad_norm": 1.359375, "learning_rate": 5.00583677983938e-06, "loss": 0.116, "step": 41924 }, { "epoch": 30.714285714285715, "grad_norm": 0.96875, "learning_rate": 5.0044866112714986e-06, "loss": 0.1429, "step": 41925 }, { "epoch": 30.715018315018316, "grad_norm": 1.1796875, "learning_rate": 5.0031366606875654e-06, "loss": 0.1123, "step": 41926 }, { "epoch": 30.715750915750917, "grad_norm": 1.234375, "learning_rate": 5.001786928094031e-06, "loss": 0.1226, "step": 41927 }, { "epoch": 30.716483516483517, "grad_norm": 1.453125, "learning_rate": 5.000437413497383e-06, "loss": 0.1263, "step": 41928 }, { "epoch": 30.717216117216118, "grad_norm": 0.9609375, "learning_rate": 4.999088116904056e-06, "loss": 0.1236, "step": 41929 }, { "epoch": 30.71794871794872, "grad_norm": 1.015625, "learning_rate": 4.997739038320549e-06, "loss": 0.1321, "step": 41930 }, { "epoch": 30.71868131868132, "grad_norm": 0.984375, "learning_rate": 4.996390177753306e-06, "loss": 0.128, "step": 41931 }, { "epoch": 30.71941391941392, "grad_norm": 1.4140625, "learning_rate": 4.995041535208795e-06, "loss": 0.11, "step": 41932 }, { "epoch": 30.72014652014652, "grad_norm": 1.4765625, "learning_rate": 4.993693110693491e-06, "loss": 0.13, "step": 41933 }, { "epoch": 30.720879120879122, "grad_norm": 1.140625, "learning_rate": 4.992344904213831e-06, "loss": 0.1242, "step": 41934 }, { "epoch": 30.721611721611723, "grad_norm": 1.0390625, "learning_rate": 4.990996915776285e-06, "loss": 0.1484, "step": 41935 }, { "epoch": 30.722344322344323, "grad_norm": 0.9609375, "learning_rate": 4.989649145387312e-06, "loss": 0.1142, "step": 41936 }, { "epoch": 30.723076923076924, "grad_norm": 0.8515625, "learning_rate": 4.988301593053375e-06, "loss": 0.1353, "step": 41937 }, { "epoch": 30.723809523809525, "grad_norm": 1.453125, "learning_rate": 4.986954258780921e-06, "loss": 0.1016, "step": 41938 }, { "epoch": 30.724542124542126, "grad_norm": 1.4765625, "learning_rate": 4.985607142576412e-06, "loss": 0.1086, "step": 41939 }, { "epoch": 30.725274725274726, "grad_norm": 0.9609375, "learning_rate": 4.984260244446296e-06, "loss": 0.1201, "step": 41940 }, { "epoch": 30.726007326007327, "grad_norm": 1.421875, "learning_rate": 4.982913564397043e-06, "loss": 0.1022, "step": 41941 }, { "epoch": 30.726739926739928, "grad_norm": 1.421875, "learning_rate": 4.981567102435086e-06, "loss": 0.1659, "step": 41942 }, { "epoch": 30.72747252747253, "grad_norm": 0.91015625, "learning_rate": 4.980220858566888e-06, "loss": 0.1425, "step": 41943 }, { "epoch": 30.72820512820513, "grad_norm": 1.484375, "learning_rate": 4.978874832798887e-06, "loss": 0.1492, "step": 41944 }, { "epoch": 30.72893772893773, "grad_norm": 0.8125, "learning_rate": 4.977529025137544e-06, "loss": 0.137, "step": 41945 }, { "epoch": 30.72967032967033, "grad_norm": 1.7578125, "learning_rate": 4.976183435589303e-06, "loss": 0.1385, "step": 41946 }, { "epoch": 30.73040293040293, "grad_norm": 0.7265625, "learning_rate": 4.974838064160606e-06, "loss": 0.1133, "step": 41947 }, { "epoch": 30.731135531135532, "grad_norm": 1.4296875, "learning_rate": 4.973492910857919e-06, "loss": 0.1514, "step": 41948 }, { "epoch": 30.731868131868133, "grad_norm": 0.76171875, "learning_rate": 4.972147975687665e-06, "loss": 0.1214, "step": 41949 }, { "epoch": 30.732600732600734, "grad_norm": 1.046875, "learning_rate": 4.970803258656291e-06, "loss": 0.14, "step": 41950 }, { "epoch": 30.733333333333334, "grad_norm": 0.734375, "learning_rate": 4.969458759770247e-06, "loss": 0.109, "step": 41951 }, { "epoch": 30.734065934065935, "grad_norm": 1.4140625, "learning_rate": 4.968114479035983e-06, "loss": 0.1558, "step": 41952 }, { "epoch": 30.734798534798536, "grad_norm": 1.0234375, "learning_rate": 4.96677041645992e-06, "loss": 0.1115, "step": 41953 }, { "epoch": 30.735531135531136, "grad_norm": 1.5390625, "learning_rate": 4.965426572048513e-06, "loss": 0.1307, "step": 41954 }, { "epoch": 30.736263736263737, "grad_norm": 0.7890625, "learning_rate": 4.964082945808182e-06, "loss": 0.1146, "step": 41955 }, { "epoch": 30.736996336996338, "grad_norm": 0.86328125, "learning_rate": 4.962739537745395e-06, "loss": 0.1088, "step": 41956 }, { "epoch": 30.73772893772894, "grad_norm": 1.4453125, "learning_rate": 4.96139634786656e-06, "loss": 0.1534, "step": 41957 }, { "epoch": 30.73846153846154, "grad_norm": 0.7890625, "learning_rate": 4.960053376178128e-06, "loss": 0.1259, "step": 41958 }, { "epoch": 30.73919413919414, "grad_norm": 0.8984375, "learning_rate": 4.958710622686536e-06, "loss": 0.1179, "step": 41959 }, { "epoch": 30.73992673992674, "grad_norm": 0.97265625, "learning_rate": 4.957368087398206e-06, "loss": 0.1337, "step": 41960 }, { "epoch": 30.74065934065934, "grad_norm": 1.046875, "learning_rate": 4.956025770319575e-06, "loss": 0.1169, "step": 41961 }, { "epoch": 30.741391941391942, "grad_norm": 0.953125, "learning_rate": 4.954683671457073e-06, "loss": 0.1041, "step": 41962 }, { "epoch": 30.742124542124543, "grad_norm": 0.96484375, "learning_rate": 4.953341790817148e-06, "loss": 0.1415, "step": 41963 }, { "epoch": 30.742857142857144, "grad_norm": 0.99609375, "learning_rate": 4.952000128406198e-06, "loss": 0.1363, "step": 41964 }, { "epoch": 30.743589743589745, "grad_norm": 1.0703125, "learning_rate": 4.950658684230673e-06, "loss": 0.113, "step": 41965 }, { "epoch": 30.744322344322345, "grad_norm": 1.1015625, "learning_rate": 4.949317458296995e-06, "loss": 0.1034, "step": 41966 }, { "epoch": 30.745054945054946, "grad_norm": 0.83984375, "learning_rate": 4.9479764506115966e-06, "loss": 0.112, "step": 41967 }, { "epoch": 30.745787545787547, "grad_norm": 1.2578125, "learning_rate": 4.946635661180889e-06, "loss": 0.1502, "step": 41968 }, { "epoch": 30.746520146520147, "grad_norm": 0.8046875, "learning_rate": 4.945295090011312e-06, "loss": 0.1188, "step": 41969 }, { "epoch": 30.747252747252748, "grad_norm": 0.91015625, "learning_rate": 4.943954737109276e-06, "loss": 0.1116, "step": 41970 }, { "epoch": 30.74798534798535, "grad_norm": 0.9296875, "learning_rate": 4.942614602481205e-06, "loss": 0.1097, "step": 41971 }, { "epoch": 30.74871794871795, "grad_norm": 1.15625, "learning_rate": 4.941274686133522e-06, "loss": 0.1488, "step": 41972 }, { "epoch": 30.74945054945055, "grad_norm": 1.984375, "learning_rate": 4.939934988072653e-06, "loss": 0.1564, "step": 41973 }, { "epoch": 30.75018315018315, "grad_norm": 0.58984375, "learning_rate": 4.938595508305015e-06, "loss": 0.1075, "step": 41974 }, { "epoch": 30.750915750915752, "grad_norm": 0.95703125, "learning_rate": 4.937256246837018e-06, "loss": 0.1053, "step": 41975 }, { "epoch": 30.751648351648353, "grad_norm": 0.64453125, "learning_rate": 4.935917203675078e-06, "loss": 0.1182, "step": 41976 }, { "epoch": 30.752380952380953, "grad_norm": 1.2734375, "learning_rate": 4.934578378825622e-06, "loss": 0.1416, "step": 41977 }, { "epoch": 30.753113553113554, "grad_norm": 0.87109375, "learning_rate": 4.933239772295063e-06, "loss": 0.1442, "step": 41978 }, { "epoch": 30.753846153846155, "grad_norm": 1.265625, "learning_rate": 4.931901384089802e-06, "loss": 0.1243, "step": 41979 }, { "epoch": 30.754578754578755, "grad_norm": 1.3828125, "learning_rate": 4.930563214216276e-06, "loss": 0.1376, "step": 41980 }, { "epoch": 30.755311355311356, "grad_norm": 1.0234375, "learning_rate": 4.929225262680859e-06, "loss": 0.1251, "step": 41981 }, { "epoch": 30.756043956043957, "grad_norm": 0.859375, "learning_rate": 4.9278875294900025e-06, "loss": 0.1055, "step": 41982 }, { "epoch": 30.756776556776558, "grad_norm": 0.72265625, "learning_rate": 4.926550014650087e-06, "loss": 0.1145, "step": 41983 }, { "epoch": 30.75750915750916, "grad_norm": 0.640625, "learning_rate": 4.925212718167533e-06, "loss": 0.1106, "step": 41984 }, { "epoch": 30.75824175824176, "grad_norm": 1.109375, "learning_rate": 4.92387564004875e-06, "loss": 0.118, "step": 41985 }, { "epoch": 30.75897435897436, "grad_norm": 0.9375, "learning_rate": 4.922538780300135e-06, "loss": 0.1289, "step": 41986 }, { "epoch": 30.75970695970696, "grad_norm": 1.203125, "learning_rate": 4.921202138928102e-06, "loss": 0.126, "step": 41987 }, { "epoch": 30.76043956043956, "grad_norm": 1.1953125, "learning_rate": 4.91986571593905e-06, "loss": 0.1239, "step": 41988 }, { "epoch": 30.761172161172162, "grad_norm": 0.578125, "learning_rate": 4.918529511339392e-06, "loss": 0.1082, "step": 41989 }, { "epoch": 30.761904761904763, "grad_norm": 1.53125, "learning_rate": 4.9171935251355145e-06, "loss": 0.1448, "step": 41990 }, { "epoch": 30.762637362637363, "grad_norm": 0.8046875, "learning_rate": 4.915857757333825e-06, "loss": 0.1085, "step": 41991 }, { "epoch": 30.763369963369964, "grad_norm": 1.078125, "learning_rate": 4.914522207940725e-06, "loss": 0.1267, "step": 41992 }, { "epoch": 30.764102564102565, "grad_norm": 1.1171875, "learning_rate": 4.913186876962623e-06, "loss": 0.1076, "step": 41993 }, { "epoch": 30.764835164835166, "grad_norm": 0.7265625, "learning_rate": 4.911851764405906e-06, "loss": 0.0823, "step": 41994 }, { "epoch": 30.765567765567766, "grad_norm": 0.9921875, "learning_rate": 4.910516870276974e-06, "loss": 0.1402, "step": 41995 }, { "epoch": 30.766300366300367, "grad_norm": 0.84375, "learning_rate": 4.909182194582216e-06, "loss": 0.0941, "step": 41996 }, { "epoch": 30.767032967032968, "grad_norm": 1.0234375, "learning_rate": 4.907847737328031e-06, "loss": 0.1245, "step": 41997 }, { "epoch": 30.76776556776557, "grad_norm": 1.0859375, "learning_rate": 4.906513498520813e-06, "loss": 0.1246, "step": 41998 }, { "epoch": 30.76849816849817, "grad_norm": 0.921875, "learning_rate": 4.905179478166956e-06, "loss": 0.1178, "step": 41999 }, { "epoch": 30.76923076923077, "grad_norm": 0.8359375, "learning_rate": 4.903845676272865e-06, "loss": 0.1108, "step": 42000 }, { "epoch": 30.76996336996337, "grad_norm": 1.1015625, "learning_rate": 4.902512092844905e-06, "loss": 0.1398, "step": 42001 }, { "epoch": 30.77069597069597, "grad_norm": 1.109375, "learning_rate": 4.901178727889481e-06, "loss": 0.1304, "step": 42002 }, { "epoch": 30.771428571428572, "grad_norm": 1.21875, "learning_rate": 4.8998455814129744e-06, "loss": 0.1037, "step": 42003 }, { "epoch": 30.772161172161173, "grad_norm": 1.640625, "learning_rate": 4.898512653421793e-06, "loss": 0.1549, "step": 42004 }, { "epoch": 30.772893772893774, "grad_norm": 0.671875, "learning_rate": 4.897179943922296e-06, "loss": 0.112, "step": 42005 }, { "epoch": 30.773626373626374, "grad_norm": 0.890625, "learning_rate": 4.8958474529208844e-06, "loss": 0.1394, "step": 42006 }, { "epoch": 30.774358974358975, "grad_norm": 1.6328125, "learning_rate": 4.894515180423923e-06, "loss": 0.12, "step": 42007 }, { "epoch": 30.775091575091576, "grad_norm": 1.2734375, "learning_rate": 4.8931831264378305e-06, "loss": 0.1182, "step": 42008 }, { "epoch": 30.775824175824177, "grad_norm": 0.9296875, "learning_rate": 4.89185129096896e-06, "loss": 0.1072, "step": 42009 }, { "epoch": 30.776556776556777, "grad_norm": 1.109375, "learning_rate": 4.8905196740237025e-06, "loss": 0.1138, "step": 42010 }, { "epoch": 30.777289377289378, "grad_norm": 1.2734375, "learning_rate": 4.88918827560845e-06, "loss": 0.1337, "step": 42011 }, { "epoch": 30.77802197802198, "grad_norm": 0.9921875, "learning_rate": 4.887857095729555e-06, "loss": 0.1163, "step": 42012 }, { "epoch": 30.77875457875458, "grad_norm": 1.140625, "learning_rate": 4.8865261343934145e-06, "loss": 0.0971, "step": 42013 }, { "epoch": 30.77948717948718, "grad_norm": 1.265625, "learning_rate": 4.885195391606397e-06, "loss": 0.1071, "step": 42014 }, { "epoch": 30.78021978021978, "grad_norm": 1.171875, "learning_rate": 4.883864867374889e-06, "loss": 0.1138, "step": 42015 }, { "epoch": 30.78095238095238, "grad_norm": 1.0859375, "learning_rate": 4.8825345617052555e-06, "loss": 0.125, "step": 42016 }, { "epoch": 30.781684981684982, "grad_norm": 0.90625, "learning_rate": 4.881204474603876e-06, "loss": 0.0951, "step": 42017 }, { "epoch": 30.782417582417583, "grad_norm": 0.67578125, "learning_rate": 4.879874606077119e-06, "loss": 0.1093, "step": 42018 }, { "epoch": 30.783150183150184, "grad_norm": 1.0859375, "learning_rate": 4.878544956131366e-06, "loss": 0.1209, "step": 42019 }, { "epoch": 30.783882783882785, "grad_norm": 1.296875, "learning_rate": 4.877215524772976e-06, "loss": 0.1129, "step": 42020 }, { "epoch": 30.784615384615385, "grad_norm": 0.89453125, "learning_rate": 4.875886312008317e-06, "loss": 0.1048, "step": 42021 }, { "epoch": 30.785347985347986, "grad_norm": 0.62109375, "learning_rate": 4.874557317843775e-06, "loss": 0.1097, "step": 42022 }, { "epoch": 30.786080586080587, "grad_norm": 0.86328125, "learning_rate": 4.873228542285704e-06, "loss": 0.1109, "step": 42023 }, { "epoch": 30.786813186813188, "grad_norm": 0.7734375, "learning_rate": 4.8718999853404684e-06, "loss": 0.1005, "step": 42024 }, { "epoch": 30.78754578754579, "grad_norm": 1.0234375, "learning_rate": 4.870571647014435e-06, "loss": 0.1181, "step": 42025 }, { "epoch": 30.78827838827839, "grad_norm": 1.421875, "learning_rate": 4.8692435273139866e-06, "loss": 0.1179, "step": 42026 }, { "epoch": 30.78901098901099, "grad_norm": 0.94140625, "learning_rate": 4.867915626245464e-06, "loss": 0.0992, "step": 42027 }, { "epoch": 30.78974358974359, "grad_norm": 1.078125, "learning_rate": 4.866587943815237e-06, "loss": 0.1398, "step": 42028 }, { "epoch": 30.79047619047619, "grad_norm": 1.171875, "learning_rate": 4.8652604800296635e-06, "loss": 0.104, "step": 42029 }, { "epoch": 30.791208791208792, "grad_norm": 0.8125, "learning_rate": 4.86393323489512e-06, "loss": 0.0995, "step": 42030 }, { "epoch": 30.791941391941393, "grad_norm": 1.0625, "learning_rate": 4.862606208417946e-06, "loss": 0.1181, "step": 42031 }, { "epoch": 30.792673992673993, "grad_norm": 1.5, "learning_rate": 4.861279400604512e-06, "loss": 0.1284, "step": 42032 }, { "epoch": 30.793406593406594, "grad_norm": 0.578125, "learning_rate": 4.8599528114611545e-06, "loss": 0.124, "step": 42033 }, { "epoch": 30.794139194139195, "grad_norm": 1.484375, "learning_rate": 4.858626440994266e-06, "loss": 0.1538, "step": 42034 }, { "epoch": 30.794871794871796, "grad_norm": 1.015625, "learning_rate": 4.857300289210171e-06, "loss": 0.1293, "step": 42035 }, { "epoch": 30.795604395604396, "grad_norm": 1.7109375, "learning_rate": 4.855974356115231e-06, "loss": 0.13, "step": 42036 }, { "epoch": 30.796336996336997, "grad_norm": 1.4609375, "learning_rate": 4.854648641715815e-06, "loss": 0.1162, "step": 42037 }, { "epoch": 30.797069597069598, "grad_norm": 1.359375, "learning_rate": 4.853323146018243e-06, "loss": 0.1427, "step": 42038 }, { "epoch": 30.7978021978022, "grad_norm": 0.6015625, "learning_rate": 4.8519978690289e-06, "loss": 0.1137, "step": 42039 }, { "epoch": 30.7985347985348, "grad_norm": 0.9609375, "learning_rate": 4.8506728107541145e-06, "loss": 0.1248, "step": 42040 }, { "epoch": 30.7992673992674, "grad_norm": 0.9140625, "learning_rate": 4.849347971200245e-06, "loss": 0.1152, "step": 42041 }, { "epoch": 30.8, "grad_norm": 1.1796875, "learning_rate": 4.848023350373634e-06, "loss": 0.1039, "step": 42042 }, { "epoch": 30.8007326007326, "grad_norm": 0.83203125, "learning_rate": 4.846698948280623e-06, "loss": 0.1137, "step": 42043 }, { "epoch": 30.801465201465202, "grad_norm": 0.734375, "learning_rate": 4.845374764927571e-06, "loss": 0.1084, "step": 42044 }, { "epoch": 30.802197802197803, "grad_norm": 1.09375, "learning_rate": 4.844050800320819e-06, "loss": 0.1121, "step": 42045 }, { "epoch": 30.802930402930404, "grad_norm": 1.1015625, "learning_rate": 4.8427270544666985e-06, "loss": 0.1258, "step": 42046 }, { "epoch": 30.803663003663004, "grad_norm": 0.73828125, "learning_rate": 4.841403527371564e-06, "loss": 0.1063, "step": 42047 }, { "epoch": 30.804395604395605, "grad_norm": 1.5859375, "learning_rate": 4.840080219041765e-06, "loss": 0.1245, "step": 42048 }, { "epoch": 30.805128205128206, "grad_norm": 1.0546875, "learning_rate": 4.838757129483619e-06, "loss": 0.104, "step": 42049 }, { "epoch": 30.805860805860807, "grad_norm": 1.1875, "learning_rate": 4.837434258703477e-06, "loss": 0.1397, "step": 42050 }, { "epoch": 30.806593406593407, "grad_norm": 1.015625, "learning_rate": 4.8361116067076775e-06, "loss": 0.0977, "step": 42051 }, { "epoch": 30.807326007326008, "grad_norm": 1.3515625, "learning_rate": 4.834789173502564e-06, "loss": 0.1446, "step": 42052 }, { "epoch": 30.80805860805861, "grad_norm": 0.53515625, "learning_rate": 4.833466959094461e-06, "loss": 0.1047, "step": 42053 }, { "epoch": 30.80879120879121, "grad_norm": 0.984375, "learning_rate": 4.832144963489714e-06, "loss": 0.1352, "step": 42054 }, { "epoch": 30.80952380952381, "grad_norm": 1.375, "learning_rate": 4.830823186694641e-06, "loss": 0.1037, "step": 42055 }, { "epoch": 30.81025641025641, "grad_norm": 0.98828125, "learning_rate": 4.829501628715604e-06, "loss": 0.1024, "step": 42056 }, { "epoch": 30.81098901098901, "grad_norm": 0.96875, "learning_rate": 4.828180289558905e-06, "loss": 0.1039, "step": 42057 }, { "epoch": 30.811721611721612, "grad_norm": 0.875, "learning_rate": 4.826859169230896e-06, "loss": 0.1296, "step": 42058 }, { "epoch": 30.812454212454213, "grad_norm": 0.5703125, "learning_rate": 4.825538267737881e-06, "loss": 0.099, "step": 42059 }, { "epoch": 30.813186813186814, "grad_norm": 1.4296875, "learning_rate": 4.824217585086225e-06, "loss": 0.1105, "step": 42060 }, { "epoch": 30.813919413919415, "grad_norm": 0.90625, "learning_rate": 4.822897121282229e-06, "loss": 0.1257, "step": 42061 }, { "epoch": 30.814652014652015, "grad_norm": 0.83984375, "learning_rate": 4.821576876332226e-06, "loss": 0.1058, "step": 42062 }, { "epoch": 30.815384615384616, "grad_norm": 1.328125, "learning_rate": 4.820256850242552e-06, "loss": 0.1517, "step": 42063 }, { "epoch": 30.816117216117217, "grad_norm": 1.3125, "learning_rate": 4.818937043019504e-06, "loss": 0.1404, "step": 42064 }, { "epoch": 30.816849816849818, "grad_norm": 1.421875, "learning_rate": 4.817617454669447e-06, "loss": 0.0889, "step": 42065 }, { "epoch": 30.81758241758242, "grad_norm": 0.921875, "learning_rate": 4.816298085198669e-06, "loss": 0.1287, "step": 42066 }, { "epoch": 30.81831501831502, "grad_norm": 1.203125, "learning_rate": 4.814978934613509e-06, "loss": 0.1281, "step": 42067 }, { "epoch": 30.81904761904762, "grad_norm": 1.75, "learning_rate": 4.81366000292028e-06, "loss": 0.1294, "step": 42068 }, { "epoch": 30.81978021978022, "grad_norm": 1.0703125, "learning_rate": 4.8123412901253026e-06, "loss": 0.1305, "step": 42069 }, { "epoch": 30.82051282051282, "grad_norm": 1.09375, "learning_rate": 4.811022796234894e-06, "loss": 0.1081, "step": 42070 }, { "epoch": 30.821245421245422, "grad_norm": 0.8828125, "learning_rate": 4.809704521255386e-06, "loss": 0.0967, "step": 42071 }, { "epoch": 30.821978021978023, "grad_norm": 1.2734375, "learning_rate": 4.808386465193068e-06, "loss": 0.1507, "step": 42072 }, { "epoch": 30.822710622710623, "grad_norm": 0.79296875, "learning_rate": 4.807068628054275e-06, "loss": 0.1236, "step": 42073 }, { "epoch": 30.823443223443224, "grad_norm": 1.1796875, "learning_rate": 4.805751009845323e-06, "loss": 0.1263, "step": 42074 }, { "epoch": 30.824175824175825, "grad_norm": 0.765625, "learning_rate": 4.804433610572514e-06, "loss": 0.1369, "step": 42075 }, { "epoch": 30.824908424908426, "grad_norm": 1.125, "learning_rate": 4.803116430242158e-06, "loss": 0.116, "step": 42076 }, { "epoch": 30.825641025641026, "grad_norm": 0.77734375, "learning_rate": 4.801799468860578e-06, "loss": 0.1263, "step": 42077 }, { "epoch": 30.826373626373627, "grad_norm": 1.1640625, "learning_rate": 4.800482726434087e-06, "loss": 0.1244, "step": 42078 }, { "epoch": 30.827106227106228, "grad_norm": 1.359375, "learning_rate": 4.799166202968973e-06, "loss": 0.1217, "step": 42079 }, { "epoch": 30.82783882783883, "grad_norm": 1.2578125, "learning_rate": 4.797849898471554e-06, "loss": 0.1303, "step": 42080 }, { "epoch": 30.82857142857143, "grad_norm": 1.015625, "learning_rate": 4.796533812948148e-06, "loss": 0.1053, "step": 42081 }, { "epoch": 30.82930402930403, "grad_norm": 1.296875, "learning_rate": 4.79521794640505e-06, "loss": 0.1255, "step": 42082 }, { "epoch": 30.83003663003663, "grad_norm": 0.78125, "learning_rate": 4.793902298848564e-06, "loss": 0.1223, "step": 42083 }, { "epoch": 30.83076923076923, "grad_norm": 1.078125, "learning_rate": 4.792586870285005e-06, "loss": 0.1291, "step": 42084 }, { "epoch": 30.831501831501832, "grad_norm": 1.3671875, "learning_rate": 4.79127166072065e-06, "loss": 0.1397, "step": 42085 }, { "epoch": 30.832234432234433, "grad_norm": 1.2109375, "learning_rate": 4.789956670161833e-06, "loss": 0.1266, "step": 42086 }, { "epoch": 30.832967032967034, "grad_norm": 1.078125, "learning_rate": 4.788641898614837e-06, "loss": 0.1178, "step": 42087 }, { "epoch": 30.833699633699634, "grad_norm": 0.796875, "learning_rate": 4.787327346085954e-06, "loss": 0.0928, "step": 42088 }, { "epoch": 30.834432234432235, "grad_norm": 0.74609375, "learning_rate": 4.78601301258151e-06, "loss": 0.1268, "step": 42089 }, { "epoch": 30.835164835164836, "grad_norm": 0.89453125, "learning_rate": 4.784698898107763e-06, "loss": 0.1352, "step": 42090 }, { "epoch": 30.835897435897436, "grad_norm": 0.71875, "learning_rate": 4.783385002671047e-06, "loss": 0.1118, "step": 42091 }, { "epoch": 30.836630036630037, "grad_norm": 1.1484375, "learning_rate": 4.7820713262776336e-06, "loss": 0.1178, "step": 42092 }, { "epoch": 30.837362637362638, "grad_norm": 0.95703125, "learning_rate": 4.78075786893383e-06, "loss": 0.1282, "step": 42093 }, { "epoch": 30.83809523809524, "grad_norm": 0.6875, "learning_rate": 4.779444630645921e-06, "loss": 0.1262, "step": 42094 }, { "epoch": 30.83882783882784, "grad_norm": 0.7421875, "learning_rate": 4.778131611420202e-06, "loss": 0.1109, "step": 42095 }, { "epoch": 30.83956043956044, "grad_norm": 1.2890625, "learning_rate": 4.776818811262961e-06, "loss": 0.1466, "step": 42096 }, { "epoch": 30.84029304029304, "grad_norm": 1.046875, "learning_rate": 4.775506230180499e-06, "loss": 0.1343, "step": 42097 }, { "epoch": 30.84102564102564, "grad_norm": 0.953125, "learning_rate": 4.774193868179086e-06, "loss": 0.1423, "step": 42098 }, { "epoch": 30.841758241758242, "grad_norm": 0.90234375, "learning_rate": 4.772881725265026e-06, "loss": 0.1391, "step": 42099 }, { "epoch": 30.842490842490843, "grad_norm": 1.0234375, "learning_rate": 4.771569801444609e-06, "loss": 0.1201, "step": 42100 }, { "epoch": 30.843223443223444, "grad_norm": 1.5390625, "learning_rate": 4.770258096724102e-06, "loss": 0.1296, "step": 42101 }, { "epoch": 30.843956043956045, "grad_norm": 1.5078125, "learning_rate": 4.768946611109803e-06, "loss": 0.1235, "step": 42102 }, { "epoch": 30.844688644688645, "grad_norm": 1.296875, "learning_rate": 4.7676353446079945e-06, "loss": 0.1339, "step": 42103 }, { "epoch": 30.845421245421246, "grad_norm": 0.9921875, "learning_rate": 4.766324297224961e-06, "loss": 0.1431, "step": 42104 }, { "epoch": 30.846153846153847, "grad_norm": 0.8046875, "learning_rate": 4.765013468966974e-06, "loss": 0.1381, "step": 42105 }, { "epoch": 30.846886446886447, "grad_norm": 1.265625, "learning_rate": 4.763702859840327e-06, "loss": 0.1197, "step": 42106 }, { "epoch": 30.847619047619048, "grad_norm": 0.875, "learning_rate": 4.7623924698512885e-06, "loss": 0.1168, "step": 42107 }, { "epoch": 30.84835164835165, "grad_norm": 0.90234375, "learning_rate": 4.7610822990061515e-06, "loss": 0.102, "step": 42108 }, { "epoch": 30.84908424908425, "grad_norm": 0.81640625, "learning_rate": 4.759772347311178e-06, "loss": 0.1016, "step": 42109 }, { "epoch": 30.84981684981685, "grad_norm": 0.93359375, "learning_rate": 4.758462614772657e-06, "loss": 0.131, "step": 42110 }, { "epoch": 30.85054945054945, "grad_norm": 0.9921875, "learning_rate": 4.757153101396844e-06, "loss": 0.1215, "step": 42111 }, { "epoch": 30.851282051282052, "grad_norm": 1.0546875, "learning_rate": 4.755843807190037e-06, "loss": 0.1158, "step": 42112 }, { "epoch": 30.852014652014653, "grad_norm": 1.0703125, "learning_rate": 4.754534732158497e-06, "loss": 0.1475, "step": 42113 }, { "epoch": 30.852747252747253, "grad_norm": 1.6640625, "learning_rate": 4.753225876308497e-06, "loss": 0.1119, "step": 42114 }, { "epoch": 30.853479853479854, "grad_norm": 1.0625, "learning_rate": 4.751917239646314e-06, "loss": 0.1099, "step": 42115 }, { "epoch": 30.854212454212455, "grad_norm": 1.1796875, "learning_rate": 4.750608822178203e-06, "loss": 0.1409, "step": 42116 }, { "epoch": 30.854945054945055, "grad_norm": 0.62109375, "learning_rate": 4.749300623910456e-06, "loss": 0.0971, "step": 42117 }, { "epoch": 30.855677655677656, "grad_norm": 1.1015625, "learning_rate": 4.747992644849321e-06, "loss": 0.1135, "step": 42118 }, { "epoch": 30.856410256410257, "grad_norm": 1.078125, "learning_rate": 4.746684885001086e-06, "loss": 0.1231, "step": 42119 }, { "epoch": 30.857142857142858, "grad_norm": 1.2109375, "learning_rate": 4.7453773443719875e-06, "loss": 0.1558, "step": 42120 }, { "epoch": 30.85787545787546, "grad_norm": 0.796875, "learning_rate": 4.744070022968313e-06, "loss": 0.1058, "step": 42121 }, { "epoch": 30.85860805860806, "grad_norm": 0.9765625, "learning_rate": 4.742762920796315e-06, "loss": 0.1087, "step": 42122 }, { "epoch": 30.85934065934066, "grad_norm": 1.0234375, "learning_rate": 4.741456037862271e-06, "loss": 0.1193, "step": 42123 }, { "epoch": 30.86007326007326, "grad_norm": 0.765625, "learning_rate": 4.740149374172426e-06, "loss": 0.118, "step": 42124 }, { "epoch": 30.86080586080586, "grad_norm": 1.5, "learning_rate": 4.7388429297330406e-06, "loss": 0.1422, "step": 42125 }, { "epoch": 30.861538461538462, "grad_norm": 1.265625, "learning_rate": 4.7375367045504e-06, "loss": 0.1399, "step": 42126 }, { "epoch": 30.862271062271063, "grad_norm": 1.078125, "learning_rate": 4.7362306986307285e-06, "loss": 0.102, "step": 42127 }, { "epoch": 30.863003663003663, "grad_norm": 0.98046875, "learning_rate": 4.734924911980295e-06, "loss": 0.113, "step": 42128 }, { "epoch": 30.863736263736264, "grad_norm": 0.7890625, "learning_rate": 4.733619344605371e-06, "loss": 0.1051, "step": 42129 }, { "epoch": 30.864468864468865, "grad_norm": 0.71875, "learning_rate": 4.732313996512204e-06, "loss": 0.0951, "step": 42130 }, { "epoch": 30.865201465201466, "grad_norm": 1.0625, "learning_rate": 4.731008867707033e-06, "loss": 0.1057, "step": 42131 }, { "epoch": 30.865934065934066, "grad_norm": 0.73828125, "learning_rate": 4.72970395819613e-06, "loss": 0.1265, "step": 42132 }, { "epoch": 30.866666666666667, "grad_norm": 0.75, "learning_rate": 4.7283992679857375e-06, "loss": 0.11, "step": 42133 }, { "epoch": 30.867399267399268, "grad_norm": 0.9609375, "learning_rate": 4.727094797082115e-06, "loss": 0.0936, "step": 42134 }, { "epoch": 30.86813186813187, "grad_norm": 1.15625, "learning_rate": 4.725790545491506e-06, "loss": 0.1283, "step": 42135 }, { "epoch": 30.86886446886447, "grad_norm": 0.921875, "learning_rate": 4.724486513220156e-06, "loss": 0.1322, "step": 42136 }, { "epoch": 30.86959706959707, "grad_norm": 1.265625, "learning_rate": 4.723182700274323e-06, "loss": 0.1359, "step": 42137 }, { "epoch": 30.87032967032967, "grad_norm": 1.296875, "learning_rate": 4.721879106660253e-06, "loss": 0.1302, "step": 42138 }, { "epoch": 30.87106227106227, "grad_norm": 0.828125, "learning_rate": 4.720575732384179e-06, "loss": 0.1153, "step": 42139 }, { "epoch": 30.871794871794872, "grad_norm": 1.109375, "learning_rate": 4.7192725774523594e-06, "loss": 0.1137, "step": 42140 }, { "epoch": 30.872527472527473, "grad_norm": 0.875, "learning_rate": 4.717969641871039e-06, "loss": 0.132, "step": 42141 }, { "epoch": 30.873260073260074, "grad_norm": 1.0859375, "learning_rate": 4.716666925646437e-06, "loss": 0.1316, "step": 42142 }, { "epoch": 30.873992673992674, "grad_norm": 0.9296875, "learning_rate": 4.715364428784838e-06, "loss": 0.1234, "step": 42143 }, { "epoch": 30.874725274725275, "grad_norm": 1.515625, "learning_rate": 4.714062151292442e-06, "loss": 0.1515, "step": 42144 }, { "epoch": 30.875457875457876, "grad_norm": 0.8671875, "learning_rate": 4.712760093175517e-06, "loss": 0.1173, "step": 42145 }, { "epoch": 30.876190476190477, "grad_norm": 1.09375, "learning_rate": 4.711458254440278e-06, "loss": 0.1304, "step": 42146 }, { "epoch": 30.876923076923077, "grad_norm": 1.125, "learning_rate": 4.7101566350929785e-06, "loss": 0.1524, "step": 42147 }, { "epoch": 30.877655677655678, "grad_norm": 1.2890625, "learning_rate": 4.708855235139846e-06, "loss": 0.1458, "step": 42148 }, { "epoch": 30.87838827838828, "grad_norm": 0.6171875, "learning_rate": 4.707554054587134e-06, "loss": 0.1211, "step": 42149 }, { "epoch": 30.87912087912088, "grad_norm": 1.0859375, "learning_rate": 4.706253093441046e-06, "loss": 0.138, "step": 42150 }, { "epoch": 30.87985347985348, "grad_norm": 1.2421875, "learning_rate": 4.704952351707842e-06, "loss": 0.1312, "step": 42151 }, { "epoch": 30.88058608058608, "grad_norm": 0.73046875, "learning_rate": 4.703651829393744e-06, "loss": 0.1025, "step": 42152 }, { "epoch": 30.88131868131868, "grad_norm": 1.2109375, "learning_rate": 4.702351526504972e-06, "loss": 0.1038, "step": 42153 }, { "epoch": 30.882051282051282, "grad_norm": 1.25, "learning_rate": 4.701051443047787e-06, "loss": 0.1218, "step": 42154 }, { "epoch": 30.882783882783883, "grad_norm": 0.8671875, "learning_rate": 4.699751579028391e-06, "loss": 0.1113, "step": 42155 }, { "epoch": 30.883516483516484, "grad_norm": 1.1640625, "learning_rate": 4.698451934453024e-06, "loss": 0.131, "step": 42156 }, { "epoch": 30.884249084249085, "grad_norm": 1.1953125, "learning_rate": 4.697152509327907e-06, "loss": 0.1339, "step": 42157 }, { "epoch": 30.884981684981685, "grad_norm": 1.3828125, "learning_rate": 4.6958533036592634e-06, "loss": 0.1236, "step": 42158 }, { "epoch": 30.885714285714286, "grad_norm": 1.375, "learning_rate": 4.694554317453329e-06, "loss": 0.1479, "step": 42159 }, { "epoch": 30.886446886446887, "grad_norm": 1.3359375, "learning_rate": 4.693255550716322e-06, "loss": 0.1295, "step": 42160 }, { "epoch": 30.887179487179488, "grad_norm": 1.109375, "learning_rate": 4.691957003454464e-06, "loss": 0.1328, "step": 42161 }, { "epoch": 30.88791208791209, "grad_norm": 1.1328125, "learning_rate": 4.690658675673971e-06, "loss": 0.1087, "step": 42162 }, { "epoch": 30.88864468864469, "grad_norm": 1.21875, "learning_rate": 4.689360567381076e-06, "loss": 0.1264, "step": 42163 }, { "epoch": 30.88937728937729, "grad_norm": 1.140625, "learning_rate": 4.688062678582e-06, "loss": 0.1334, "step": 42164 }, { "epoch": 30.89010989010989, "grad_norm": 0.66015625, "learning_rate": 4.686765009282945e-06, "loss": 0.0945, "step": 42165 }, { "epoch": 30.89084249084249, "grad_norm": 1.3515625, "learning_rate": 4.685467559490141e-06, "loss": 0.1133, "step": 42166 }, { "epoch": 30.891575091575092, "grad_norm": 0.9765625, "learning_rate": 4.684170329209811e-06, "loss": 0.1259, "step": 42167 }, { "epoch": 30.892307692307693, "grad_norm": 1.015625, "learning_rate": 4.68287331844814e-06, "loss": 0.1265, "step": 42168 }, { "epoch": 30.893040293040293, "grad_norm": 1.078125, "learning_rate": 4.681576527211387e-06, "loss": 0.11, "step": 42169 }, { "epoch": 30.893772893772894, "grad_norm": 0.89453125, "learning_rate": 4.680279955505727e-06, "loss": 0.1279, "step": 42170 }, { "epoch": 30.894505494505495, "grad_norm": 0.8671875, "learning_rate": 4.678983603337396e-06, "loss": 0.1066, "step": 42171 }, { "epoch": 30.895238095238096, "grad_norm": 0.98828125, "learning_rate": 4.67768747071259e-06, "loss": 0.1164, "step": 42172 }, { "epoch": 30.895970695970696, "grad_norm": 1.390625, "learning_rate": 4.676391557637537e-06, "loss": 0.1207, "step": 42173 }, { "epoch": 30.896703296703297, "grad_norm": 0.73046875, "learning_rate": 4.6750958641184175e-06, "loss": 0.1173, "step": 42174 }, { "epoch": 30.897435897435898, "grad_norm": 0.9453125, "learning_rate": 4.6738003901614716e-06, "loss": 0.1238, "step": 42175 }, { "epoch": 30.8981684981685, "grad_norm": 0.63671875, "learning_rate": 4.672505135772881e-06, "loss": 0.1031, "step": 42176 }, { "epoch": 30.8989010989011, "grad_norm": 1.625, "learning_rate": 4.671210100958866e-06, "loss": 0.1141, "step": 42177 }, { "epoch": 30.8996336996337, "grad_norm": 1.46875, "learning_rate": 4.669915285725636e-06, "loss": 0.1212, "step": 42178 }, { "epoch": 30.9003663003663, "grad_norm": 0.80078125, "learning_rate": 4.66862069007937e-06, "loss": 0.0993, "step": 42179 }, { "epoch": 30.9010989010989, "grad_norm": 0.76953125, "learning_rate": 4.667326314026304e-06, "loss": 0.1213, "step": 42180 }, { "epoch": 30.901831501831502, "grad_norm": 0.8046875, "learning_rate": 4.666032157572617e-06, "loss": 0.0969, "step": 42181 }, { "epoch": 30.902564102564103, "grad_norm": 1.34375, "learning_rate": 4.664738220724522e-06, "loss": 0.1058, "step": 42182 }, { "epoch": 30.903296703296704, "grad_norm": 0.96484375, "learning_rate": 4.66344450348821e-06, "loss": 0.1301, "step": 42183 }, { "epoch": 30.904029304029304, "grad_norm": 1.640625, "learning_rate": 4.662151005869876e-06, "loss": 0.134, "step": 42184 }, { "epoch": 30.904761904761905, "grad_norm": 1.1875, "learning_rate": 4.660857727875728e-06, "loss": 0.1097, "step": 42185 }, { "epoch": 30.905494505494506, "grad_norm": 1.890625, "learning_rate": 4.659564669511965e-06, "loss": 0.1309, "step": 42186 }, { "epoch": 30.906227106227107, "grad_norm": 1.2421875, "learning_rate": 4.658271830784766e-06, "loss": 0.1461, "step": 42187 }, { "epoch": 30.906959706959707, "grad_norm": 0.7421875, "learning_rate": 4.656979211700339e-06, "loss": 0.1115, "step": 42188 }, { "epoch": 30.907692307692308, "grad_norm": 0.82421875, "learning_rate": 4.655686812264872e-06, "loss": 0.1088, "step": 42189 }, { "epoch": 30.90842490842491, "grad_norm": 1.59375, "learning_rate": 4.654394632484568e-06, "loss": 0.1465, "step": 42190 }, { "epoch": 30.90915750915751, "grad_norm": 0.91015625, "learning_rate": 4.653102672365599e-06, "loss": 0.0999, "step": 42191 }, { "epoch": 30.90989010989011, "grad_norm": 1.34375, "learning_rate": 4.651810931914169e-06, "loss": 0.1654, "step": 42192 }, { "epoch": 30.91062271062271, "grad_norm": 1.21875, "learning_rate": 4.650519411136472e-06, "loss": 0.1267, "step": 42193 }, { "epoch": 30.91135531135531, "grad_norm": 0.98046875, "learning_rate": 4.649228110038671e-06, "loss": 0.1217, "step": 42194 }, { "epoch": 30.912087912087912, "grad_norm": 0.7265625, "learning_rate": 4.647937028626983e-06, "loss": 0.1057, "step": 42195 }, { "epoch": 30.912820512820513, "grad_norm": 1.84375, "learning_rate": 4.646646166907576e-06, "loss": 0.136, "step": 42196 }, { "epoch": 30.913553113553114, "grad_norm": 0.73046875, "learning_rate": 4.645355524886649e-06, "loss": 0.1259, "step": 42197 }, { "epoch": 30.914285714285715, "grad_norm": 0.72265625, "learning_rate": 4.644065102570367e-06, "loss": 0.1293, "step": 42198 }, { "epoch": 30.915018315018315, "grad_norm": 0.9453125, "learning_rate": 4.642774899964933e-06, "loss": 0.1423, "step": 42199 }, { "epoch": 30.915750915750916, "grad_norm": 0.859375, "learning_rate": 4.641484917076497e-06, "loss": 0.1349, "step": 42200 }, { "epoch": 30.916483516483517, "grad_norm": 1.7109375, "learning_rate": 4.640195153911283e-06, "loss": 0.115, "step": 42201 }, { "epoch": 30.917216117216118, "grad_norm": 0.89453125, "learning_rate": 4.638905610475438e-06, "loss": 0.1048, "step": 42202 }, { "epoch": 30.91794871794872, "grad_norm": 0.78125, "learning_rate": 4.637616286775147e-06, "loss": 0.1044, "step": 42203 }, { "epoch": 30.91868131868132, "grad_norm": 0.9921875, "learning_rate": 4.636327182816609e-06, "loss": 0.1228, "step": 42204 }, { "epoch": 30.91941391941392, "grad_norm": 1.1796875, "learning_rate": 4.635038298605961e-06, "loss": 0.1112, "step": 42205 }, { "epoch": 30.92014652014652, "grad_norm": 0.69140625, "learning_rate": 4.6337496341494205e-06, "loss": 0.1028, "step": 42206 }, { "epoch": 30.92087912087912, "grad_norm": 0.875, "learning_rate": 4.63246118945313e-06, "loss": 0.11, "step": 42207 }, { "epoch": 30.921611721611722, "grad_norm": 1.6953125, "learning_rate": 4.631172964523288e-06, "loss": 0.1289, "step": 42208 }, { "epoch": 30.922344322344323, "grad_norm": 1.171875, "learning_rate": 4.629884959366046e-06, "loss": 0.126, "step": 42209 }, { "epoch": 30.923076923076923, "grad_norm": 1.3359375, "learning_rate": 4.628597173987581e-06, "loss": 0.1433, "step": 42210 }, { "epoch": 30.923809523809524, "grad_norm": 0.7265625, "learning_rate": 4.6273096083940675e-06, "loss": 0.1085, "step": 42211 }, { "epoch": 30.924542124542125, "grad_norm": 1.0234375, "learning_rate": 4.626022262591681e-06, "loss": 0.1238, "step": 42212 }, { "epoch": 30.925274725274726, "grad_norm": 1.5078125, "learning_rate": 4.6247351365865664e-06, "loss": 0.1213, "step": 42213 }, { "epoch": 30.926007326007326, "grad_norm": 0.87890625, "learning_rate": 4.623448230384911e-06, "loss": 0.1121, "step": 42214 }, { "epoch": 30.926739926739927, "grad_norm": 1.671875, "learning_rate": 4.622161543992877e-06, "loss": 0.1287, "step": 42215 }, { "epoch": 30.927472527472528, "grad_norm": 0.8359375, "learning_rate": 4.620875077416632e-06, "loss": 0.1364, "step": 42216 }, { "epoch": 30.92820512820513, "grad_norm": 1.375, "learning_rate": 4.619588830662333e-06, "loss": 0.1101, "step": 42217 }, { "epoch": 30.92893772893773, "grad_norm": 0.69921875, "learning_rate": 4.618302803736142e-06, "loss": 0.0983, "step": 42218 }, { "epoch": 30.92967032967033, "grad_norm": 1.09375, "learning_rate": 4.617016996644234e-06, "loss": 0.1196, "step": 42219 }, { "epoch": 30.93040293040293, "grad_norm": 0.80078125, "learning_rate": 4.615731409392742e-06, "loss": 0.0782, "step": 42220 }, { "epoch": 30.93113553113553, "grad_norm": 0.8671875, "learning_rate": 4.614446041987856e-06, "loss": 0.1235, "step": 42221 }, { "epoch": 30.931868131868132, "grad_norm": 0.61328125, "learning_rate": 4.613160894435718e-06, "loss": 0.0914, "step": 42222 }, { "epoch": 30.932600732600733, "grad_norm": 0.76953125, "learning_rate": 4.611875966742499e-06, "loss": 0.1171, "step": 42223 }, { "epoch": 30.933333333333334, "grad_norm": 1.25, "learning_rate": 4.610591258914336e-06, "loss": 0.1297, "step": 42224 }, { "epoch": 30.934065934065934, "grad_norm": 0.76171875, "learning_rate": 4.609306770957408e-06, "loss": 0.0869, "step": 42225 }, { "epoch": 30.934798534798535, "grad_norm": 1.1328125, "learning_rate": 4.608022502877837e-06, "loss": 0.1178, "step": 42226 }, { "epoch": 30.935531135531136, "grad_norm": 1.2265625, "learning_rate": 4.606738454681809e-06, "loss": 0.1222, "step": 42227 }, { "epoch": 30.936263736263736, "grad_norm": 1.203125, "learning_rate": 4.605454626375454e-06, "loss": 0.1287, "step": 42228 }, { "epoch": 30.936996336996337, "grad_norm": 1.2421875, "learning_rate": 4.604171017964937e-06, "loss": 0.1229, "step": 42229 }, { "epoch": 30.937728937728938, "grad_norm": 1.375, "learning_rate": 4.602887629456406e-06, "loss": 0.1188, "step": 42230 }, { "epoch": 30.93846153846154, "grad_norm": 1.1328125, "learning_rate": 4.6016044608559986e-06, "loss": 0.1227, "step": 42231 }, { "epoch": 30.93919413919414, "grad_norm": 1.28125, "learning_rate": 4.600321512169882e-06, "loss": 0.1118, "step": 42232 }, { "epoch": 30.93992673992674, "grad_norm": 1.859375, "learning_rate": 4.59903878340419e-06, "loss": 0.1266, "step": 42233 }, { "epoch": 30.94065934065934, "grad_norm": 0.83984375, "learning_rate": 4.597756274565074e-06, "loss": 0.1068, "step": 42234 }, { "epoch": 30.94139194139194, "grad_norm": 0.91796875, "learning_rate": 4.596473985658671e-06, "loss": 0.1455, "step": 42235 }, { "epoch": 30.942124542124542, "grad_norm": 0.87890625, "learning_rate": 4.595191916691131e-06, "loss": 0.1298, "step": 42236 }, { "epoch": 30.942857142857143, "grad_norm": 0.87890625, "learning_rate": 4.593910067668594e-06, "loss": 0.0985, "step": 42237 }, { "epoch": 30.943589743589744, "grad_norm": 0.828125, "learning_rate": 4.592628438597213e-06, "loss": 0.1281, "step": 42238 }, { "epoch": 30.944322344322345, "grad_norm": 0.9296875, "learning_rate": 4.59134702948311e-06, "loss": 0.1119, "step": 42239 }, { "epoch": 30.945054945054945, "grad_norm": 0.85546875, "learning_rate": 4.590065840332436e-06, "loss": 0.1019, "step": 42240 }, { "epoch": 30.945787545787546, "grad_norm": 0.859375, "learning_rate": 4.588784871151323e-06, "loss": 0.1142, "step": 42241 }, { "epoch": 30.946520146520147, "grad_norm": 0.91015625, "learning_rate": 4.587504121945928e-06, "loss": 0.1179, "step": 42242 }, { "epoch": 30.947252747252747, "grad_norm": 0.71875, "learning_rate": 4.586223592722361e-06, "loss": 0.1086, "step": 42243 }, { "epoch": 30.947985347985348, "grad_norm": 0.9140625, "learning_rate": 4.58494328348677e-06, "loss": 0.0973, "step": 42244 }, { "epoch": 30.94871794871795, "grad_norm": 1.328125, "learning_rate": 4.583663194245301e-06, "loss": 0.1396, "step": 42245 }, { "epoch": 30.94945054945055, "grad_norm": 0.6875, "learning_rate": 4.582383325004053e-06, "loss": 0.1394, "step": 42246 }, { "epoch": 30.95018315018315, "grad_norm": 0.984375, "learning_rate": 4.581103675769199e-06, "loss": 0.138, "step": 42247 }, { "epoch": 30.95091575091575, "grad_norm": 1.0, "learning_rate": 4.5798242465468435e-06, "loss": 0.1121, "step": 42248 }, { "epoch": 30.951648351648352, "grad_norm": 1.0703125, "learning_rate": 4.578545037343131e-06, "loss": 0.1128, "step": 42249 }, { "epoch": 30.952380952380953, "grad_norm": 1.21875, "learning_rate": 4.577266048164176e-06, "loss": 0.1289, "step": 42250 }, { "epoch": 30.953113553113553, "grad_norm": 1.0546875, "learning_rate": 4.575987279016121e-06, "loss": 0.1224, "step": 42251 }, { "epoch": 30.953846153846154, "grad_norm": 0.890625, "learning_rate": 4.57470872990508e-06, "loss": 0.1172, "step": 42252 }, { "epoch": 30.954578754578755, "grad_norm": 1.0078125, "learning_rate": 4.5734304008371975e-06, "loss": 0.1145, "step": 42253 }, { "epoch": 30.955311355311355, "grad_norm": 0.859375, "learning_rate": 4.572152291818575e-06, "loss": 0.1385, "step": 42254 }, { "epoch": 30.956043956043956, "grad_norm": 1.09375, "learning_rate": 4.570874402855349e-06, "loss": 0.1049, "step": 42255 }, { "epoch": 30.956776556776557, "grad_norm": 0.984375, "learning_rate": 4.569596733953653e-06, "loss": 0.1334, "step": 42256 }, { "epoch": 30.957509157509158, "grad_norm": 1.171875, "learning_rate": 4.568319285119577e-06, "loss": 0.1414, "step": 42257 }, { "epoch": 30.95824175824176, "grad_norm": 0.65234375, "learning_rate": 4.5670420563592815e-06, "loss": 0.0997, "step": 42258 }, { "epoch": 30.95897435897436, "grad_norm": 1.1171875, "learning_rate": 4.565765047678855e-06, "loss": 0.1378, "step": 42259 }, { "epoch": 30.95970695970696, "grad_norm": 1.046875, "learning_rate": 4.564488259084437e-06, "loss": 0.114, "step": 42260 }, { "epoch": 30.96043956043956, "grad_norm": 0.84765625, "learning_rate": 4.563211690582126e-06, "loss": 0.1077, "step": 42261 }, { "epoch": 30.96117216117216, "grad_norm": 0.6640625, "learning_rate": 4.561935342178046e-06, "loss": 0.1022, "step": 42262 }, { "epoch": 30.961904761904762, "grad_norm": 0.83984375, "learning_rate": 4.5606592138783185e-06, "loss": 0.1, "step": 42263 }, { "epoch": 30.962637362637363, "grad_norm": 0.87109375, "learning_rate": 4.559383305689059e-06, "loss": 0.118, "step": 42264 }, { "epoch": 30.963369963369964, "grad_norm": 0.7578125, "learning_rate": 4.5581076176163685e-06, "loss": 0.1494, "step": 42265 }, { "epoch": 30.964102564102564, "grad_norm": 0.75390625, "learning_rate": 4.556832149666364e-06, "loss": 0.1141, "step": 42266 }, { "epoch": 30.964835164835165, "grad_norm": 1.0625, "learning_rate": 4.555556901845162e-06, "loss": 0.1158, "step": 42267 }, { "epoch": 30.965567765567766, "grad_norm": 0.99609375, "learning_rate": 4.554281874158867e-06, "loss": 0.1231, "step": 42268 }, { "epoch": 30.966300366300366, "grad_norm": 1.6640625, "learning_rate": 4.553007066613603e-06, "loss": 0.1255, "step": 42269 }, { "epoch": 30.967032967032967, "grad_norm": 1.09375, "learning_rate": 4.551732479215451e-06, "loss": 0.1415, "step": 42270 }, { "epoch": 30.967765567765568, "grad_norm": 0.59375, "learning_rate": 4.5504581119705414e-06, "loss": 0.1034, "step": 42271 }, { "epoch": 30.96849816849817, "grad_norm": 1.3203125, "learning_rate": 4.549183964884961e-06, "loss": 0.1279, "step": 42272 }, { "epoch": 30.96923076923077, "grad_norm": 1.0625, "learning_rate": 4.547910037964834e-06, "loss": 0.1035, "step": 42273 }, { "epoch": 30.96996336996337, "grad_norm": 0.9140625, "learning_rate": 4.546636331216253e-06, "loss": 0.1214, "step": 42274 }, { "epoch": 30.97069597069597, "grad_norm": 1.40625, "learning_rate": 4.5453628446453265e-06, "loss": 0.1248, "step": 42275 }, { "epoch": 30.97142857142857, "grad_norm": 1.2734375, "learning_rate": 4.5440895782581455e-06, "loss": 0.125, "step": 42276 }, { "epoch": 30.972161172161172, "grad_norm": 1.1640625, "learning_rate": 4.542816532060823e-06, "loss": 0.1336, "step": 42277 }, { "epoch": 30.972893772893773, "grad_norm": 1.4375, "learning_rate": 4.541543706059448e-06, "loss": 0.139, "step": 42278 }, { "epoch": 30.973626373626374, "grad_norm": 1.1171875, "learning_rate": 4.5402711002601325e-06, "loss": 0.1217, "step": 42279 }, { "epoch": 30.974358974358974, "grad_norm": 1.4921875, "learning_rate": 4.538998714668959e-06, "loss": 0.1284, "step": 42280 }, { "epoch": 30.975091575091575, "grad_norm": 1.171875, "learning_rate": 4.537726549292026e-06, "loss": 0.1157, "step": 42281 }, { "epoch": 30.975824175824176, "grad_norm": 0.84765625, "learning_rate": 4.536454604135448e-06, "loss": 0.1191, "step": 42282 }, { "epoch": 30.976556776556777, "grad_norm": 1.1171875, "learning_rate": 4.535182879205286e-06, "loss": 0.1183, "step": 42283 }, { "epoch": 30.977289377289377, "grad_norm": 1.125, "learning_rate": 4.533911374507671e-06, "loss": 0.1337, "step": 42284 }, { "epoch": 30.978021978021978, "grad_norm": 1.1015625, "learning_rate": 4.532640090048666e-06, "loss": 0.1192, "step": 42285 }, { "epoch": 30.97875457875458, "grad_norm": 1.09375, "learning_rate": 4.531369025834376e-06, "loss": 0.1309, "step": 42286 }, { "epoch": 30.97948717948718, "grad_norm": 1.03125, "learning_rate": 4.5300981818708855e-06, "loss": 0.1173, "step": 42287 }, { "epoch": 30.98021978021978, "grad_norm": 1.234375, "learning_rate": 4.528827558164284e-06, "loss": 0.102, "step": 42288 }, { "epoch": 30.98095238095238, "grad_norm": 1.1484375, "learning_rate": 4.527557154720659e-06, "loss": 0.1197, "step": 42289 }, { "epoch": 30.98168498168498, "grad_norm": 0.87890625, "learning_rate": 4.5262869715461106e-06, "loss": 0.1238, "step": 42290 }, { "epoch": 30.982417582417582, "grad_norm": 0.828125, "learning_rate": 4.5250170086467035e-06, "loss": 0.1086, "step": 42291 }, { "epoch": 30.983150183150183, "grad_norm": 1.0078125, "learning_rate": 4.5237472660285346e-06, "loss": 0.1121, "step": 42292 }, { "epoch": 30.983882783882784, "grad_norm": 1.0234375, "learning_rate": 4.522477743697683e-06, "loss": 0.1276, "step": 42293 }, { "epoch": 30.984615384615385, "grad_norm": 1.1171875, "learning_rate": 4.521208441660232e-06, "loss": 0.0993, "step": 42294 }, { "epoch": 30.985347985347985, "grad_norm": 1.4375, "learning_rate": 4.519939359922277e-06, "loss": 0.14, "step": 42295 }, { "epoch": 30.986080586080586, "grad_norm": 1.0078125, "learning_rate": 4.51867049848988e-06, "loss": 0.1381, "step": 42296 }, { "epoch": 30.986813186813187, "grad_norm": 0.9296875, "learning_rate": 4.517401857369134e-06, "loss": 0.1167, "step": 42297 }, { "epoch": 30.987545787545788, "grad_norm": 0.8984375, "learning_rate": 4.51613343656609e-06, "loss": 0.106, "step": 42298 }, { "epoch": 30.98827838827839, "grad_norm": 0.90234375, "learning_rate": 4.514865236086872e-06, "loss": 0.1399, "step": 42299 }, { "epoch": 30.98901098901099, "grad_norm": 1.21875, "learning_rate": 4.513597255937513e-06, "loss": 0.1133, "step": 42300 }, { "epoch": 30.98974358974359, "grad_norm": 1.28125, "learning_rate": 4.512329496124118e-06, "loss": 0.1415, "step": 42301 }, { "epoch": 30.99047619047619, "grad_norm": 1.3046875, "learning_rate": 4.511061956652743e-06, "loss": 0.138, "step": 42302 }, { "epoch": 30.99120879120879, "grad_norm": 1.1953125, "learning_rate": 4.509794637529462e-06, "loss": 0.1188, "step": 42303 }, { "epoch": 30.991941391941392, "grad_norm": 1.1015625, "learning_rate": 4.508527538760357e-06, "loss": 0.1054, "step": 42304 }, { "epoch": 30.992673992673993, "grad_norm": 1.375, "learning_rate": 4.5072606603515035e-06, "loss": 0.1321, "step": 42305 }, { "epoch": 30.993406593406593, "grad_norm": 1.28125, "learning_rate": 4.505994002308954e-06, "loss": 0.1342, "step": 42306 }, { "epoch": 30.994139194139194, "grad_norm": 1.0546875, "learning_rate": 4.504727564638783e-06, "loss": 0.1107, "step": 42307 }, { "epoch": 30.994871794871795, "grad_norm": 0.9140625, "learning_rate": 4.503461347347068e-06, "loss": 0.1349, "step": 42308 }, { "epoch": 30.995604395604396, "grad_norm": 1.0546875, "learning_rate": 4.502195350439856e-06, "loss": 0.156, "step": 42309 }, { "epoch": 30.996336996336996, "grad_norm": 1.2734375, "learning_rate": 4.5009295739232456e-06, "loss": 0.1119, "step": 42310 }, { "epoch": 30.997069597069597, "grad_norm": 0.765625, "learning_rate": 4.499664017803271e-06, "loss": 0.1145, "step": 42311 }, { "epoch": 30.997802197802198, "grad_norm": 1.0390625, "learning_rate": 4.498398682086015e-06, "loss": 0.1381, "step": 42312 }, { "epoch": 30.9985347985348, "grad_norm": 1.140625, "learning_rate": 4.497133566777524e-06, "loss": 0.1038, "step": 42313 }, { "epoch": 30.9992673992674, "grad_norm": 0.84765625, "learning_rate": 4.495868671883865e-06, "loss": 0.1085, "step": 42314 }, { "epoch": 31.0, "grad_norm": 1.4140625, "learning_rate": 4.494603997411107e-06, "loss": 0.1115, "step": 42315 }, { "epoch": 31.0007326007326, "grad_norm": 1.2265625, "learning_rate": 4.493339543365313e-06, "loss": 0.1012, "step": 42316 }, { "epoch": 31.0014652014652, "grad_norm": 1.2265625, "learning_rate": 4.492075309752519e-06, "loss": 0.1117, "step": 42317 }, { "epoch": 31.002197802197802, "grad_norm": 1.5390625, "learning_rate": 4.4908112965787934e-06, "loss": 0.1237, "step": 42318 }, { "epoch": 31.002930402930403, "grad_norm": 1.1796875, "learning_rate": 4.4895475038502026e-06, "loss": 0.1095, "step": 42319 }, { "epoch": 31.003663003663004, "grad_norm": 0.78125, "learning_rate": 4.488283931572787e-06, "loss": 0.1149, "step": 42320 }, { "epoch": 31.004395604395604, "grad_norm": 1.265625, "learning_rate": 4.4870205797526224e-06, "loss": 0.1443, "step": 42321 }, { "epoch": 31.005128205128205, "grad_norm": 1.40625, "learning_rate": 4.485757448395733e-06, "loss": 0.1375, "step": 42322 }, { "epoch": 31.005860805860806, "grad_norm": 0.8203125, "learning_rate": 4.4844945375081935e-06, "loss": 0.0832, "step": 42323 }, { "epoch": 31.006593406593407, "grad_norm": 1.0234375, "learning_rate": 4.4832318470960355e-06, "loss": 0.1263, "step": 42324 }, { "epoch": 31.007326007326007, "grad_norm": 0.828125, "learning_rate": 4.481969377165334e-06, "loss": 0.1029, "step": 42325 }, { "epoch": 31.008058608058608, "grad_norm": 1.0625, "learning_rate": 4.4807071277221145e-06, "loss": 0.116, "step": 42326 }, { "epoch": 31.00879120879121, "grad_norm": 0.86328125, "learning_rate": 4.4794450987724404e-06, "loss": 0.1337, "step": 42327 }, { "epoch": 31.00952380952381, "grad_norm": 1.5234375, "learning_rate": 4.478183290322347e-06, "loss": 0.0966, "step": 42328 }, { "epoch": 31.01025641025641, "grad_norm": 1.296875, "learning_rate": 4.476921702377888e-06, "loss": 0.1528, "step": 42329 }, { "epoch": 31.01098901098901, "grad_norm": 1.046875, "learning_rate": 4.4756603349451e-06, "loss": 0.1115, "step": 42330 }, { "epoch": 31.01172161172161, "grad_norm": 1.0390625, "learning_rate": 4.4743991880300425e-06, "loss": 0.1246, "step": 42331 }, { "epoch": 31.012454212454212, "grad_norm": 1.2265625, "learning_rate": 4.473138261638742e-06, "loss": 0.1039, "step": 42332 }, { "epoch": 31.013186813186813, "grad_norm": 0.7109375, "learning_rate": 4.4718775557772445e-06, "loss": 0.102, "step": 42333 }, { "epoch": 31.013919413919414, "grad_norm": 1.1015625, "learning_rate": 4.470617070451599e-06, "loss": 0.126, "step": 42334 }, { "epoch": 31.014652014652015, "grad_norm": 1.1640625, "learning_rate": 4.469356805667821e-06, "loss": 0.132, "step": 42335 }, { "epoch": 31.015384615384615, "grad_norm": 0.8984375, "learning_rate": 4.468096761431985e-06, "loss": 0.1161, "step": 42336 }, { "epoch": 31.016117216117216, "grad_norm": 0.96875, "learning_rate": 4.466836937750095e-06, "loss": 0.1057, "step": 42337 }, { "epoch": 31.016849816849817, "grad_norm": 1.3125, "learning_rate": 4.465577334628216e-06, "loss": 0.1277, "step": 42338 }, { "epoch": 31.017582417582418, "grad_norm": 1.65625, "learning_rate": 4.4643179520723515e-06, "loss": 0.1233, "step": 42339 }, { "epoch": 31.01831501831502, "grad_norm": 1.1796875, "learning_rate": 4.46305879008856e-06, "loss": 0.1359, "step": 42340 }, { "epoch": 31.01904761904762, "grad_norm": 1.28125, "learning_rate": 4.461799848682862e-06, "loss": 0.1052, "step": 42341 }, { "epoch": 31.01978021978022, "grad_norm": 1.171875, "learning_rate": 4.460541127861302e-06, "loss": 0.0918, "step": 42342 }, { "epoch": 31.02051282051282, "grad_norm": 1.3671875, "learning_rate": 4.459282627629899e-06, "loss": 0.0994, "step": 42343 }, { "epoch": 31.02124542124542, "grad_norm": 0.83984375, "learning_rate": 4.458024347994687e-06, "loss": 0.0929, "step": 42344 }, { "epoch": 31.021978021978022, "grad_norm": 1.1484375, "learning_rate": 4.4567662889616915e-06, "loss": 0.1221, "step": 42345 }, { "epoch": 31.022710622710623, "grad_norm": 1.25, "learning_rate": 4.455508450536945e-06, "loss": 0.1361, "step": 42346 }, { "epoch": 31.023443223443223, "grad_norm": 0.75390625, "learning_rate": 4.4542508327264834e-06, "loss": 0.1103, "step": 42347 }, { "epoch": 31.024175824175824, "grad_norm": 1.4296875, "learning_rate": 4.452993435536307e-06, "loss": 0.1135, "step": 42348 }, { "epoch": 31.024908424908425, "grad_norm": 0.875, "learning_rate": 4.451736258972471e-06, "loss": 0.1186, "step": 42349 }, { "epoch": 31.025641025641026, "grad_norm": 0.91015625, "learning_rate": 4.450479303040961e-06, "loss": 0.1158, "step": 42350 }, { "epoch": 31.026373626373626, "grad_norm": 0.91015625, "learning_rate": 4.449222567747844e-06, "loss": 0.1383, "step": 42351 }, { "epoch": 31.027106227106227, "grad_norm": 0.8046875, "learning_rate": 4.4479660530991046e-06, "loss": 0.1228, "step": 42352 }, { "epoch": 31.027838827838828, "grad_norm": 1.546875, "learning_rate": 4.446709759100794e-06, "loss": 0.1444, "step": 42353 }, { "epoch": 31.02857142857143, "grad_norm": 0.5625, "learning_rate": 4.445453685758902e-06, "loss": 0.1286, "step": 42354 }, { "epoch": 31.02930402930403, "grad_norm": 1.40625, "learning_rate": 4.444197833079457e-06, "loss": 0.1196, "step": 42355 }, { "epoch": 31.03003663003663, "grad_norm": 1.046875, "learning_rate": 4.442942201068483e-06, "loss": 0.1121, "step": 42356 }, { "epoch": 31.03076923076923, "grad_norm": 0.875, "learning_rate": 4.441686789732002e-06, "loss": 0.1497, "step": 42357 }, { "epoch": 31.03150183150183, "grad_norm": 1.1328125, "learning_rate": 4.44043159907601e-06, "loss": 0.1624, "step": 42358 }, { "epoch": 31.032234432234432, "grad_norm": 1.4140625, "learning_rate": 4.439176629106533e-06, "loss": 0.1243, "step": 42359 }, { "epoch": 31.032967032967033, "grad_norm": 1.078125, "learning_rate": 4.437921879829583e-06, "loss": 0.1305, "step": 42360 }, { "epoch": 31.033699633699634, "grad_norm": 1.234375, "learning_rate": 4.4366673512511614e-06, "loss": 0.1364, "step": 42361 }, { "epoch": 31.034432234432234, "grad_norm": 1.2109375, "learning_rate": 4.435413043377298e-06, "loss": 0.0847, "step": 42362 }, { "epoch": 31.035164835164835, "grad_norm": 0.82421875, "learning_rate": 4.434158956213986e-06, "loss": 0.1117, "step": 42363 }, { "epoch": 31.035897435897436, "grad_norm": 0.9453125, "learning_rate": 4.432905089767255e-06, "loss": 0.1313, "step": 42364 }, { "epoch": 31.036630036630036, "grad_norm": 1.203125, "learning_rate": 4.431651444043085e-06, "loss": 0.1013, "step": 42365 }, { "epoch": 31.037362637362637, "grad_norm": 0.87890625, "learning_rate": 4.430398019047496e-06, "loss": 0.1362, "step": 42366 }, { "epoch": 31.038095238095238, "grad_norm": 0.82421875, "learning_rate": 4.429144814786497e-06, "loss": 0.1057, "step": 42367 }, { "epoch": 31.03882783882784, "grad_norm": 1.4453125, "learning_rate": 4.427891831266093e-06, "loss": 0.1092, "step": 42368 }, { "epoch": 31.03956043956044, "grad_norm": 0.80859375, "learning_rate": 4.426639068492278e-06, "loss": 0.1183, "step": 42369 }, { "epoch": 31.04029304029304, "grad_norm": 1.0078125, "learning_rate": 4.425386526471061e-06, "loss": 0.1382, "step": 42370 }, { "epoch": 31.04102564102564, "grad_norm": 0.9453125, "learning_rate": 4.424134205208441e-06, "loss": 0.1324, "step": 42371 }, { "epoch": 31.04175824175824, "grad_norm": 1.0078125, "learning_rate": 4.4228821047104225e-06, "loss": 0.1201, "step": 42372 }, { "epoch": 31.042490842490842, "grad_norm": 1.2421875, "learning_rate": 4.4216302249830085e-06, "loss": 0.1193, "step": 42373 }, { "epoch": 31.043223443223443, "grad_norm": 1.0546875, "learning_rate": 4.420378566032183e-06, "loss": 0.1367, "step": 42374 }, { "epoch": 31.043956043956044, "grad_norm": 1.3046875, "learning_rate": 4.419127127863958e-06, "loss": 0.1251, "step": 42375 }, { "epoch": 31.044688644688645, "grad_norm": 1.5, "learning_rate": 4.41787591048431e-06, "loss": 0.1293, "step": 42376 }, { "epoch": 31.045421245421245, "grad_norm": 1.4765625, "learning_rate": 4.416624913899252e-06, "loss": 0.1268, "step": 42377 }, { "epoch": 31.046153846153846, "grad_norm": 1.0625, "learning_rate": 4.415374138114774e-06, "loss": 0.1044, "step": 42378 }, { "epoch": 31.046886446886447, "grad_norm": 1.265625, "learning_rate": 4.41412358313687e-06, "loss": 0.1339, "step": 42379 }, { "epoch": 31.047619047619047, "grad_norm": 1.390625, "learning_rate": 4.412873248971521e-06, "loss": 0.1378, "step": 42380 }, { "epoch": 31.048351648351648, "grad_norm": 0.8828125, "learning_rate": 4.411623135624726e-06, "loss": 0.1039, "step": 42381 }, { "epoch": 31.04908424908425, "grad_norm": 1.1640625, "learning_rate": 4.410373243102478e-06, "loss": 0.1196, "step": 42382 }, { "epoch": 31.04981684981685, "grad_norm": 0.890625, "learning_rate": 4.409123571410758e-06, "loss": 0.1054, "step": 42383 }, { "epoch": 31.05054945054945, "grad_norm": 1.015625, "learning_rate": 4.407874120555568e-06, "loss": 0.1292, "step": 42384 }, { "epoch": 31.05128205128205, "grad_norm": 1.09375, "learning_rate": 4.406624890542875e-06, "loss": 0.1442, "step": 42385 }, { "epoch": 31.052014652014652, "grad_norm": 0.91015625, "learning_rate": 4.405375881378684e-06, "loss": 0.1224, "step": 42386 }, { "epoch": 31.052747252747253, "grad_norm": 0.6875, "learning_rate": 4.404127093068951e-06, "loss": 0.1334, "step": 42387 }, { "epoch": 31.053479853479853, "grad_norm": 0.66015625, "learning_rate": 4.402878525619694e-06, "loss": 0.1199, "step": 42388 }, { "epoch": 31.054212454212454, "grad_norm": 1.4453125, "learning_rate": 4.4016301790368695e-06, "loss": 0.1021, "step": 42389 }, { "epoch": 31.054945054945055, "grad_norm": 1.015625, "learning_rate": 4.400382053326474e-06, "loss": 0.123, "step": 42390 }, { "epoch": 31.055677655677655, "grad_norm": 0.94140625, "learning_rate": 4.3991341484944755e-06, "loss": 0.1004, "step": 42391 }, { "epoch": 31.056410256410256, "grad_norm": 0.7265625, "learning_rate": 4.3978864645468655e-06, "loss": 0.1178, "step": 42392 }, { "epoch": 31.057142857142857, "grad_norm": 0.87890625, "learning_rate": 4.3966390014896095e-06, "loss": 0.1037, "step": 42393 }, { "epoch": 31.057875457875458, "grad_norm": 1.046875, "learning_rate": 4.395391759328694e-06, "loss": 0.1061, "step": 42394 }, { "epoch": 31.05860805860806, "grad_norm": 1.515625, "learning_rate": 4.394144738070094e-06, "loss": 0.1424, "step": 42395 }, { "epoch": 31.05934065934066, "grad_norm": 0.69921875, "learning_rate": 4.392897937719775e-06, "loss": 0.1141, "step": 42396 }, { "epoch": 31.06007326007326, "grad_norm": 0.92578125, "learning_rate": 4.391651358283723e-06, "loss": 0.1198, "step": 42397 }, { "epoch": 31.06080586080586, "grad_norm": 1.09375, "learning_rate": 4.3904049997679e-06, "loss": 0.1235, "step": 42398 }, { "epoch": 31.06153846153846, "grad_norm": 0.7109375, "learning_rate": 4.389158862178297e-06, "loss": 0.1148, "step": 42399 }, { "epoch": 31.062271062271062, "grad_norm": 0.80078125, "learning_rate": 4.3879129455208615e-06, "loss": 0.1206, "step": 42400 }, { "epoch": 31.063003663003663, "grad_norm": 1.1953125, "learning_rate": 4.38666724980158e-06, "loss": 0.1413, "step": 42401 }, { "epoch": 31.063736263736264, "grad_norm": 1.2578125, "learning_rate": 4.385421775026402e-06, "loss": 0.1257, "step": 42402 }, { "epoch": 31.064468864468864, "grad_norm": 1.1171875, "learning_rate": 4.384176521201312e-06, "loss": 0.1493, "step": 42403 }, { "epoch": 31.065201465201465, "grad_norm": 1.6328125, "learning_rate": 4.382931488332271e-06, "loss": 0.1245, "step": 42404 }, { "epoch": 31.065934065934066, "grad_norm": 0.984375, "learning_rate": 4.381686676425249e-06, "loss": 0.1123, "step": 42405 }, { "epoch": 31.066666666666666, "grad_norm": 1.5859375, "learning_rate": 4.380442085486196e-06, "loss": 0.1262, "step": 42406 }, { "epoch": 31.067399267399267, "grad_norm": 0.68359375, "learning_rate": 4.379197715521093e-06, "loss": 0.0938, "step": 42407 }, { "epoch": 31.068131868131868, "grad_norm": 0.9921875, "learning_rate": 4.37795356653589e-06, "loss": 0.1184, "step": 42408 }, { "epoch": 31.06886446886447, "grad_norm": 1.4921875, "learning_rate": 4.376709638536551e-06, "loss": 0.1489, "step": 42409 }, { "epoch": 31.06959706959707, "grad_norm": 0.625, "learning_rate": 4.375465931529045e-06, "loss": 0.107, "step": 42410 }, { "epoch": 31.07032967032967, "grad_norm": 0.98046875, "learning_rate": 4.374222445519316e-06, "loss": 0.1386, "step": 42411 }, { "epoch": 31.07106227106227, "grad_norm": 1.3046875, "learning_rate": 4.372979180513338e-06, "loss": 0.1288, "step": 42412 }, { "epoch": 31.07179487179487, "grad_norm": 0.98046875, "learning_rate": 4.371736136517047e-06, "loss": 0.1182, "step": 42413 }, { "epoch": 31.072527472527472, "grad_norm": 0.9296875, "learning_rate": 4.370493313536424e-06, "loss": 0.1239, "step": 42414 }, { "epoch": 31.073260073260073, "grad_norm": 0.95703125, "learning_rate": 4.369250711577402e-06, "loss": 0.0873, "step": 42415 }, { "epoch": 31.073992673992674, "grad_norm": 1.1015625, "learning_rate": 4.36800833064595e-06, "loss": 0.1412, "step": 42416 }, { "epoch": 31.074725274725274, "grad_norm": 1.28125, "learning_rate": 4.366766170748007e-06, "loss": 0.1415, "step": 42417 }, { "epoch": 31.075457875457875, "grad_norm": 1.78125, "learning_rate": 4.365524231889538e-06, "loss": 0.1494, "step": 42418 }, { "epoch": 31.076190476190476, "grad_norm": 1.1484375, "learning_rate": 4.3642825140764775e-06, "loss": 0.086, "step": 42419 }, { "epoch": 31.076923076923077, "grad_norm": 1.1171875, "learning_rate": 4.363041017314799e-06, "loss": 0.1353, "step": 42420 }, { "epoch": 31.077655677655677, "grad_norm": 0.7265625, "learning_rate": 4.361799741610428e-06, "loss": 0.112, "step": 42421 }, { "epoch": 31.078388278388278, "grad_norm": 0.77734375, "learning_rate": 4.36055868696932e-06, "loss": 0.111, "step": 42422 }, { "epoch": 31.07912087912088, "grad_norm": 1.0390625, "learning_rate": 4.359317853397421e-06, "loss": 0.1172, "step": 42423 }, { "epoch": 31.07985347985348, "grad_norm": 0.69140625, "learning_rate": 4.358077240900679e-06, "loss": 0.1198, "step": 42424 }, { "epoch": 31.08058608058608, "grad_norm": 0.94140625, "learning_rate": 4.3568368494850455e-06, "loss": 0.1549, "step": 42425 }, { "epoch": 31.08131868131868, "grad_norm": 1.0859375, "learning_rate": 4.355596679156444e-06, "loss": 0.1313, "step": 42426 }, { "epoch": 31.08205128205128, "grad_norm": 0.71875, "learning_rate": 4.354356729920838e-06, "loss": 0.1043, "step": 42427 }, { "epoch": 31.082783882783882, "grad_norm": 0.7109375, "learning_rate": 4.353117001784143e-06, "loss": 0.1152, "step": 42428 }, { "epoch": 31.083516483516483, "grad_norm": 1.7109375, "learning_rate": 4.351877494752321e-06, "loss": 0.1034, "step": 42429 }, { "epoch": 31.084249084249084, "grad_norm": 0.5703125, "learning_rate": 4.350638208831299e-06, "loss": 0.1265, "step": 42430 }, { "epoch": 31.084981684981685, "grad_norm": 1.0078125, "learning_rate": 4.34939914402703e-06, "loss": 0.1389, "step": 42431 }, { "epoch": 31.085714285714285, "grad_norm": 0.7734375, "learning_rate": 4.348160300345426e-06, "loss": 0.1259, "step": 42432 }, { "epoch": 31.086446886446886, "grad_norm": 1.28125, "learning_rate": 4.3469216777924455e-06, "loss": 0.1332, "step": 42433 }, { "epoch": 31.087179487179487, "grad_norm": 0.6015625, "learning_rate": 4.345683276374007e-06, "loss": 0.0805, "step": 42434 }, { "epoch": 31.087912087912088, "grad_norm": 1.4765625, "learning_rate": 4.344445096096056e-06, "loss": 0.0876, "step": 42435 }, { "epoch": 31.08864468864469, "grad_norm": 1.09375, "learning_rate": 4.343207136964522e-06, "loss": 0.1235, "step": 42436 }, { "epoch": 31.08937728937729, "grad_norm": 1.1015625, "learning_rate": 4.341969398985337e-06, "loss": 0.1028, "step": 42437 }, { "epoch": 31.09010989010989, "grad_norm": 1.28125, "learning_rate": 4.340731882164429e-06, "loss": 0.1451, "step": 42438 }, { "epoch": 31.09084249084249, "grad_norm": 0.84765625, "learning_rate": 4.339494586507715e-06, "loss": 0.1085, "step": 42439 }, { "epoch": 31.09157509157509, "grad_norm": 1.484375, "learning_rate": 4.33825751202115e-06, "loss": 0.1502, "step": 42440 }, { "epoch": 31.092307692307692, "grad_norm": 1.0546875, "learning_rate": 4.33702065871064e-06, "loss": 0.1488, "step": 42441 }, { "epoch": 31.093040293040293, "grad_norm": 1.2734375, "learning_rate": 4.335784026582126e-06, "loss": 0.1017, "step": 42442 }, { "epoch": 31.093772893772893, "grad_norm": 0.9296875, "learning_rate": 4.334547615641517e-06, "loss": 0.0947, "step": 42443 }, { "epoch": 31.094505494505494, "grad_norm": 1.734375, "learning_rate": 4.333311425894747e-06, "loss": 0.1155, "step": 42444 }, { "epoch": 31.095238095238095, "grad_norm": 0.828125, "learning_rate": 4.332075457347741e-06, "loss": 0.1216, "step": 42445 }, { "epoch": 31.095970695970696, "grad_norm": 0.953125, "learning_rate": 4.330839710006421e-06, "loss": 0.0965, "step": 42446 }, { "epoch": 31.096703296703296, "grad_norm": 1.3671875, "learning_rate": 4.329604183876698e-06, "loss": 0.1142, "step": 42447 }, { "epoch": 31.097435897435897, "grad_norm": 1.1484375, "learning_rate": 4.328368878964495e-06, "loss": 0.1551, "step": 42448 }, { "epoch": 31.098168498168498, "grad_norm": 1.2578125, "learning_rate": 4.327133795275736e-06, "loss": 0.1318, "step": 42449 }, { "epoch": 31.0989010989011, "grad_norm": 1.1328125, "learning_rate": 4.3258989328163385e-06, "loss": 0.1357, "step": 42450 }, { "epoch": 31.0996336996337, "grad_norm": 1.5703125, "learning_rate": 4.324664291592223e-06, "loss": 0.1456, "step": 42451 }, { "epoch": 31.1003663003663, "grad_norm": 1.015625, "learning_rate": 4.323429871609286e-06, "loss": 0.0937, "step": 42452 }, { "epoch": 31.1010989010989, "grad_norm": 1.046875, "learning_rate": 4.322195672873469e-06, "loss": 0.1484, "step": 42453 }, { "epoch": 31.1018315018315, "grad_norm": 0.83203125, "learning_rate": 4.320961695390667e-06, "loss": 0.091, "step": 42454 }, { "epoch": 31.102564102564102, "grad_norm": 0.79296875, "learning_rate": 4.3197279391667945e-06, "loss": 0.1047, "step": 42455 }, { "epoch": 31.103296703296703, "grad_norm": 1.546875, "learning_rate": 4.318494404207765e-06, "loss": 0.1014, "step": 42456 }, { "epoch": 31.104029304029304, "grad_norm": 0.9609375, "learning_rate": 4.3172610905194966e-06, "loss": 0.1052, "step": 42457 }, { "epoch": 31.104761904761904, "grad_norm": 1.1171875, "learning_rate": 4.31602799810788e-06, "loss": 0.1527, "step": 42458 }, { "epoch": 31.105494505494505, "grad_norm": 1.046875, "learning_rate": 4.314795126978839e-06, "loss": 0.0867, "step": 42459 }, { "epoch": 31.106227106227106, "grad_norm": 0.95703125, "learning_rate": 4.313562477138274e-06, "loss": 0.1228, "step": 42460 }, { "epoch": 31.106959706959707, "grad_norm": 0.8671875, "learning_rate": 4.312330048592095e-06, "loss": 0.1235, "step": 42461 }, { "epoch": 31.107692307692307, "grad_norm": 0.73828125, "learning_rate": 4.31109784134621e-06, "loss": 0.132, "step": 42462 }, { "epoch": 31.108424908424908, "grad_norm": 1.140625, "learning_rate": 4.309865855406511e-06, "loss": 0.1251, "step": 42463 }, { "epoch": 31.10915750915751, "grad_norm": 0.89453125, "learning_rate": 4.308634090778917e-06, "loss": 0.1218, "step": 42464 }, { "epoch": 31.10989010989011, "grad_norm": 1.2265625, "learning_rate": 4.307402547469308e-06, "loss": 0.1027, "step": 42465 }, { "epoch": 31.11062271062271, "grad_norm": 0.67578125, "learning_rate": 4.306171225483607e-06, "loss": 0.1243, "step": 42466 }, { "epoch": 31.11135531135531, "grad_norm": 1.46875, "learning_rate": 4.304940124827697e-06, "loss": 0.1315, "step": 42467 }, { "epoch": 31.11208791208791, "grad_norm": 1.4375, "learning_rate": 4.303709245507495e-06, "loss": 0.1265, "step": 42468 }, { "epoch": 31.112820512820512, "grad_norm": 1.1328125, "learning_rate": 4.302478587528881e-06, "loss": 0.1156, "step": 42469 }, { "epoch": 31.113553113553113, "grad_norm": 0.9375, "learning_rate": 4.301248150897753e-06, "loss": 0.1344, "step": 42470 }, { "epoch": 31.114285714285714, "grad_norm": 0.81640625, "learning_rate": 4.300017935620013e-06, "loss": 0.1105, "step": 42471 }, { "epoch": 31.115018315018315, "grad_norm": 1.34375, "learning_rate": 4.298787941701561e-06, "loss": 0.1251, "step": 42472 }, { "epoch": 31.115750915750915, "grad_norm": 1.046875, "learning_rate": 4.297558169148276e-06, "loss": 0.1278, "step": 42473 }, { "epoch": 31.116483516483516, "grad_norm": 1.25, "learning_rate": 4.296328617966057e-06, "loss": 0.1164, "step": 42474 }, { "epoch": 31.117216117216117, "grad_norm": 0.8125, "learning_rate": 4.295099288160796e-06, "loss": 0.1173, "step": 42475 }, { "epoch": 31.117948717948718, "grad_norm": 0.9140625, "learning_rate": 4.293870179738378e-06, "loss": 0.1013, "step": 42476 }, { "epoch": 31.11868131868132, "grad_norm": 0.81640625, "learning_rate": 4.292641292704707e-06, "loss": 0.1017, "step": 42477 }, { "epoch": 31.11941391941392, "grad_norm": 1.7890625, "learning_rate": 4.29141262706566e-06, "loss": 0.1665, "step": 42478 }, { "epoch": 31.12014652014652, "grad_norm": 0.890625, "learning_rate": 4.290184182827123e-06, "loss": 0.1088, "step": 42479 }, { "epoch": 31.12087912087912, "grad_norm": 1.0546875, "learning_rate": 4.288955959994978e-06, "loss": 0.1365, "step": 42480 }, { "epoch": 31.12161172161172, "grad_norm": 0.83984375, "learning_rate": 4.287727958575112e-06, "loss": 0.1033, "step": 42481 }, { "epoch": 31.122344322344322, "grad_norm": 1.4609375, "learning_rate": 4.286500178573416e-06, "loss": 0.1097, "step": 42482 }, { "epoch": 31.123076923076923, "grad_norm": 0.8359375, "learning_rate": 4.2852726199957755e-06, "loss": 0.1191, "step": 42483 }, { "epoch": 31.123809523809523, "grad_norm": 0.796875, "learning_rate": 4.284045282848055e-06, "loss": 0.1007, "step": 42484 }, { "epoch": 31.124542124542124, "grad_norm": 1.4765625, "learning_rate": 4.282818167136151e-06, "loss": 0.1198, "step": 42485 }, { "epoch": 31.125274725274725, "grad_norm": 0.8125, "learning_rate": 4.2815912728659314e-06, "loss": 0.123, "step": 42486 }, { "epoch": 31.126007326007326, "grad_norm": 1.328125, "learning_rate": 4.2803646000432834e-06, "loss": 0.1561, "step": 42487 }, { "epoch": 31.126739926739926, "grad_norm": 1.2265625, "learning_rate": 4.279138148674088e-06, "loss": 0.1106, "step": 42488 }, { "epoch": 31.127472527472527, "grad_norm": 1.28125, "learning_rate": 4.277911918764206e-06, "loss": 0.1392, "step": 42489 }, { "epoch": 31.128205128205128, "grad_norm": 1.1796875, "learning_rate": 4.276685910319532e-06, "loss": 0.1494, "step": 42490 }, { "epoch": 31.12893772893773, "grad_norm": 0.98828125, "learning_rate": 4.2754601233459175e-06, "loss": 0.1045, "step": 42491 }, { "epoch": 31.12967032967033, "grad_norm": 1.0, "learning_rate": 4.274234557849259e-06, "loss": 0.1045, "step": 42492 }, { "epoch": 31.13040293040293, "grad_norm": 0.66796875, "learning_rate": 4.273009213835409e-06, "loss": 0.1183, "step": 42493 }, { "epoch": 31.13113553113553, "grad_norm": 0.7578125, "learning_rate": 4.271784091310255e-06, "loss": 0.145, "step": 42494 }, { "epoch": 31.13186813186813, "grad_norm": 1.109375, "learning_rate": 4.270559190279654e-06, "loss": 0.1093, "step": 42495 }, { "epoch": 31.132600732600732, "grad_norm": 1.0859375, "learning_rate": 4.26933451074948e-06, "loss": 0.127, "step": 42496 }, { "epoch": 31.133333333333333, "grad_norm": 1.3828125, "learning_rate": 4.2681100527255975e-06, "loss": 0.113, "step": 42497 }, { "epoch": 31.134065934065934, "grad_norm": 1.0625, "learning_rate": 4.266885816213883e-06, "loss": 0.1154, "step": 42498 }, { "epoch": 31.134798534798534, "grad_norm": 0.78515625, "learning_rate": 4.265661801220198e-06, "loss": 0.1241, "step": 42499 }, { "epoch": 31.135531135531135, "grad_norm": 1.9375, "learning_rate": 4.264438007750397e-06, "loss": 0.1395, "step": 42500 }, { "epoch": 31.136263736263736, "grad_norm": 1.265625, "learning_rate": 4.263214435810355e-06, "loss": 0.1355, "step": 42501 }, { "epoch": 31.136996336996337, "grad_norm": 0.7890625, "learning_rate": 4.261991085405932e-06, "loss": 0.1101, "step": 42502 }, { "epoch": 31.137728937728937, "grad_norm": 0.984375, "learning_rate": 4.260767956542991e-06, "loss": 0.1413, "step": 42503 }, { "epoch": 31.138461538461538, "grad_norm": 0.68359375, "learning_rate": 4.259545049227388e-06, "loss": 0.118, "step": 42504 }, { "epoch": 31.13919413919414, "grad_norm": 1.2109375, "learning_rate": 4.258322363464988e-06, "loss": 0.1677, "step": 42505 }, { "epoch": 31.13992673992674, "grad_norm": 1.1875, "learning_rate": 4.257099899261643e-06, "loss": 0.1252, "step": 42506 }, { "epoch": 31.14065934065934, "grad_norm": 1.109375, "learning_rate": 4.255877656623206e-06, "loss": 0.1049, "step": 42507 }, { "epoch": 31.14139194139194, "grad_norm": 0.9296875, "learning_rate": 4.2546556355555446e-06, "loss": 0.1419, "step": 42508 }, { "epoch": 31.14212454212454, "grad_norm": 0.6484375, "learning_rate": 4.253433836064514e-06, "loss": 0.0982, "step": 42509 }, { "epoch": 31.142857142857142, "grad_norm": 1.046875, "learning_rate": 4.2522122581559616e-06, "loss": 0.1618, "step": 42510 }, { "epoch": 31.143589743589743, "grad_norm": 1.03125, "learning_rate": 4.25099090183574e-06, "loss": 0.1312, "step": 42511 }, { "epoch": 31.144322344322344, "grad_norm": 1.109375, "learning_rate": 4.2497697671097e-06, "loss": 0.1349, "step": 42512 }, { "epoch": 31.145054945054945, "grad_norm": 0.8515625, "learning_rate": 4.248548853983701e-06, "loss": 0.1135, "step": 42513 }, { "epoch": 31.145787545787545, "grad_norm": 1.0078125, "learning_rate": 4.2473281624635916e-06, "loss": 0.1028, "step": 42514 }, { "epoch": 31.146520146520146, "grad_norm": 0.6328125, "learning_rate": 4.2461076925552126e-06, "loss": 0.1082, "step": 42515 }, { "epoch": 31.147252747252747, "grad_norm": 1.265625, "learning_rate": 4.244887444264423e-06, "loss": 0.1205, "step": 42516 }, { "epoch": 31.147985347985347, "grad_norm": 0.7109375, "learning_rate": 4.24366741759704e-06, "loss": 0.1016, "step": 42517 }, { "epoch": 31.148717948717948, "grad_norm": 1.0625, "learning_rate": 4.2424476125589525e-06, "loss": 0.1218, "step": 42518 }, { "epoch": 31.14945054945055, "grad_norm": 0.75, "learning_rate": 4.241228029155974e-06, "loss": 0.1137, "step": 42519 }, { "epoch": 31.15018315018315, "grad_norm": 1.078125, "learning_rate": 4.240008667393969e-06, "loss": 0.1259, "step": 42520 }, { "epoch": 31.15091575091575, "grad_norm": 1.3125, "learning_rate": 4.238789527278761e-06, "loss": 0.1272, "step": 42521 }, { "epoch": 31.15164835164835, "grad_norm": 2.0625, "learning_rate": 4.2375706088161975e-06, "loss": 0.135, "step": 42522 }, { "epoch": 31.152380952380952, "grad_norm": 1.1640625, "learning_rate": 4.236351912012117e-06, "loss": 0.1254, "step": 42523 }, { "epoch": 31.153113553113553, "grad_norm": 0.8984375, "learning_rate": 4.235133436872364e-06, "loss": 0.1485, "step": 42524 }, { "epoch": 31.153846153846153, "grad_norm": 0.94140625, "learning_rate": 4.233915183402783e-06, "loss": 0.1493, "step": 42525 }, { "epoch": 31.154578754578754, "grad_norm": 1.0078125, "learning_rate": 4.232697151609197e-06, "loss": 0.0986, "step": 42526 }, { "epoch": 31.155311355311355, "grad_norm": 0.921875, "learning_rate": 4.231479341497443e-06, "loss": 0.134, "step": 42527 }, { "epoch": 31.156043956043955, "grad_norm": 0.765625, "learning_rate": 4.230261753073364e-06, "loss": 0.1106, "step": 42528 }, { "epoch": 31.156776556776556, "grad_norm": 0.8984375, "learning_rate": 4.2290443863427954e-06, "loss": 0.1117, "step": 42529 }, { "epoch": 31.157509157509157, "grad_norm": 1.0, "learning_rate": 4.227827241311558e-06, "loss": 0.1325, "step": 42530 }, { "epoch": 31.158241758241758, "grad_norm": 1.1640625, "learning_rate": 4.226610317985497e-06, "loss": 0.1112, "step": 42531 }, { "epoch": 31.15897435897436, "grad_norm": 0.56640625, "learning_rate": 4.225393616370434e-06, "loss": 0.1219, "step": 42532 }, { "epoch": 31.15970695970696, "grad_norm": 0.69921875, "learning_rate": 4.224177136472195e-06, "loss": 0.0984, "step": 42533 }, { "epoch": 31.16043956043956, "grad_norm": 1.0234375, "learning_rate": 4.22296087829662e-06, "loss": 0.1196, "step": 42534 }, { "epoch": 31.16117216117216, "grad_norm": 0.94921875, "learning_rate": 4.221744841849534e-06, "loss": 0.1008, "step": 42535 }, { "epoch": 31.16190476190476, "grad_norm": 0.71484375, "learning_rate": 4.2205290271367535e-06, "loss": 0.119, "step": 42536 }, { "epoch": 31.162637362637362, "grad_norm": 1.03125, "learning_rate": 4.2193134341641124e-06, "loss": 0.104, "step": 42537 }, { "epoch": 31.163369963369963, "grad_norm": 1.3984375, "learning_rate": 4.218098062937434e-06, "loss": 0.1236, "step": 42538 }, { "epoch": 31.164102564102564, "grad_norm": 0.89453125, "learning_rate": 4.2168829134625405e-06, "loss": 0.0975, "step": 42539 }, { "epoch": 31.164835164835164, "grad_norm": 1.875, "learning_rate": 4.215667985745264e-06, "loss": 0.1515, "step": 42540 }, { "epoch": 31.165567765567765, "grad_norm": 1.296875, "learning_rate": 4.214453279791407e-06, "loss": 0.1159, "step": 42541 }, { "epoch": 31.166300366300366, "grad_norm": 1.015625, "learning_rate": 4.213238795606807e-06, "loss": 0.0908, "step": 42542 }, { "epoch": 31.167032967032966, "grad_norm": 1.484375, "learning_rate": 4.212024533197262e-06, "loss": 0.138, "step": 42543 }, { "epoch": 31.167765567765567, "grad_norm": 1.171875, "learning_rate": 4.210810492568612e-06, "loss": 0.1103, "step": 42544 }, { "epoch": 31.168498168498168, "grad_norm": 1.125, "learning_rate": 4.209596673726662e-06, "loss": 0.1231, "step": 42545 }, { "epoch": 31.16923076923077, "grad_norm": 1.171875, "learning_rate": 4.208383076677237e-06, "loss": 0.0926, "step": 42546 }, { "epoch": 31.16996336996337, "grad_norm": 0.640625, "learning_rate": 4.207169701426133e-06, "loss": 0.1054, "step": 42547 }, { "epoch": 31.17069597069597, "grad_norm": 0.625, "learning_rate": 4.205956547979185e-06, "loss": 0.0999, "step": 42548 }, { "epoch": 31.17142857142857, "grad_norm": 1.328125, "learning_rate": 4.2047436163421865e-06, "loss": 0.1156, "step": 42549 }, { "epoch": 31.17216117216117, "grad_norm": 1.0625, "learning_rate": 4.2035309065209665e-06, "loss": 0.1018, "step": 42550 }, { "epoch": 31.172893772893772, "grad_norm": 0.98046875, "learning_rate": 4.202318418521336e-06, "loss": 0.1211, "step": 42551 }, { "epoch": 31.173626373626373, "grad_norm": 0.61328125, "learning_rate": 4.201106152349085e-06, "loss": 0.1132, "step": 42552 }, { "epoch": 31.174358974358974, "grad_norm": 1.0859375, "learning_rate": 4.199894108010037e-06, "loss": 0.097, "step": 42553 }, { "epoch": 31.175091575091574, "grad_norm": 0.984375, "learning_rate": 4.1986822855099885e-06, "loss": 0.1155, "step": 42554 }, { "epoch": 31.175824175824175, "grad_norm": 1.2109375, "learning_rate": 4.197470684854768e-06, "loss": 0.1355, "step": 42555 }, { "epoch": 31.176556776556776, "grad_norm": 1.0078125, "learning_rate": 4.196259306050156e-06, "loss": 0.0936, "step": 42556 }, { "epoch": 31.177289377289377, "grad_norm": 1.078125, "learning_rate": 4.19504814910197e-06, "loss": 0.1629, "step": 42557 }, { "epoch": 31.178021978021977, "grad_norm": 1.2265625, "learning_rate": 4.193837214016006e-06, "loss": 0.1273, "step": 42558 }, { "epoch": 31.178754578754578, "grad_norm": 1.1796875, "learning_rate": 4.192626500798069e-06, "loss": 0.1261, "step": 42559 }, { "epoch": 31.17948717948718, "grad_norm": 1.3359375, "learning_rate": 4.191416009453959e-06, "loss": 0.1367, "step": 42560 }, { "epoch": 31.18021978021978, "grad_norm": 0.70703125, "learning_rate": 4.190205739989486e-06, "loss": 0.1058, "step": 42561 }, { "epoch": 31.18095238095238, "grad_norm": 0.8515625, "learning_rate": 4.18899569241043e-06, "loss": 0.1227, "step": 42562 }, { "epoch": 31.18168498168498, "grad_norm": 0.9765625, "learning_rate": 4.187785866722599e-06, "loss": 0.1238, "step": 42563 }, { "epoch": 31.18241758241758, "grad_norm": 0.65234375, "learning_rate": 4.186576262931792e-06, "loss": 0.1064, "step": 42564 }, { "epoch": 31.183150183150182, "grad_norm": 0.62890625, "learning_rate": 4.185366881043795e-06, "loss": 0.1043, "step": 42565 }, { "epoch": 31.183882783882783, "grad_norm": 1.265625, "learning_rate": 4.1841577210644244e-06, "loss": 0.113, "step": 42566 }, { "epoch": 31.184615384615384, "grad_norm": 1.2109375, "learning_rate": 4.1829487829994476e-06, "loss": 0.1321, "step": 42567 }, { "epoch": 31.185347985347985, "grad_norm": 1.171875, "learning_rate": 4.181740066854677e-06, "loss": 0.1247, "step": 42568 }, { "epoch": 31.186080586080585, "grad_norm": 1.046875, "learning_rate": 4.180531572635879e-06, "loss": 0.1218, "step": 42569 }, { "epoch": 31.186813186813186, "grad_norm": 0.93359375, "learning_rate": 4.179323300348877e-06, "loss": 0.1111, "step": 42570 }, { "epoch": 31.187545787545787, "grad_norm": 1.0703125, "learning_rate": 4.178115249999432e-06, "loss": 0.1242, "step": 42571 }, { "epoch": 31.188278388278388, "grad_norm": 1.1015625, "learning_rate": 4.176907421593353e-06, "loss": 0.1202, "step": 42572 }, { "epoch": 31.18901098901099, "grad_norm": 0.796875, "learning_rate": 4.175699815136409e-06, "loss": 0.1255, "step": 42573 }, { "epoch": 31.18974358974359, "grad_norm": 1.2734375, "learning_rate": 4.174492430634392e-06, "loss": 0.1168, "step": 42574 }, { "epoch": 31.19047619047619, "grad_norm": 0.921875, "learning_rate": 4.173285268093094e-06, "loss": 0.1311, "step": 42575 }, { "epoch": 31.19120879120879, "grad_norm": 0.84375, "learning_rate": 4.1720783275182945e-06, "loss": 0.1414, "step": 42576 }, { "epoch": 31.19194139194139, "grad_norm": 0.9609375, "learning_rate": 4.17087160891578e-06, "loss": 0.117, "step": 42577 }, { "epoch": 31.192673992673992, "grad_norm": 0.875, "learning_rate": 4.169665112291324e-06, "loss": 0.1116, "step": 42578 }, { "epoch": 31.193406593406593, "grad_norm": 1.265625, "learning_rate": 4.16845883765071e-06, "loss": 0.1273, "step": 42579 }, { "epoch": 31.194139194139193, "grad_norm": 1.0078125, "learning_rate": 4.16725278499972e-06, "loss": 0.122, "step": 42580 }, { "epoch": 31.194871794871794, "grad_norm": 1.2265625, "learning_rate": 4.166046954344137e-06, "loss": 0.1347, "step": 42581 }, { "epoch": 31.195604395604395, "grad_norm": 0.97265625, "learning_rate": 4.164841345689731e-06, "loss": 0.1153, "step": 42582 }, { "epoch": 31.196336996336996, "grad_norm": 1.3984375, "learning_rate": 4.163635959042286e-06, "loss": 0.1272, "step": 42583 }, { "epoch": 31.197069597069596, "grad_norm": 1.2578125, "learning_rate": 4.162430794407567e-06, "loss": 0.109, "step": 42584 }, { "epoch": 31.197802197802197, "grad_norm": 1.265625, "learning_rate": 4.161225851791349e-06, "loss": 0.153, "step": 42585 }, { "epoch": 31.198534798534798, "grad_norm": 0.8515625, "learning_rate": 4.160021131199412e-06, "loss": 0.1267, "step": 42586 }, { "epoch": 31.1992673992674, "grad_norm": 1.078125, "learning_rate": 4.158816632637531e-06, "loss": 0.1204, "step": 42587 }, { "epoch": 31.2, "grad_norm": 1.15625, "learning_rate": 4.157612356111472e-06, "loss": 0.1233, "step": 42588 }, { "epoch": 31.2007326007326, "grad_norm": 1.1328125, "learning_rate": 4.1564083016269976e-06, "loss": 0.1167, "step": 42589 }, { "epoch": 31.2014652014652, "grad_norm": 0.640625, "learning_rate": 4.15520446918989e-06, "loss": 0.1072, "step": 42590 }, { "epoch": 31.2021978021978, "grad_norm": 1.09375, "learning_rate": 4.154000858805908e-06, "loss": 0.1004, "step": 42591 }, { "epoch": 31.202930402930402, "grad_norm": 0.9140625, "learning_rate": 4.152797470480835e-06, "loss": 0.112, "step": 42592 }, { "epoch": 31.203663003663003, "grad_norm": 0.6171875, "learning_rate": 4.151594304220411e-06, "loss": 0.1113, "step": 42593 }, { "epoch": 31.204395604395604, "grad_norm": 1.3515625, "learning_rate": 4.150391360030427e-06, "loss": 0.1284, "step": 42594 }, { "epoch": 31.205128205128204, "grad_norm": 1.1015625, "learning_rate": 4.149188637916611e-06, "loss": 0.0989, "step": 42595 }, { "epoch": 31.205860805860805, "grad_norm": 0.8671875, "learning_rate": 4.147986137884768e-06, "loss": 0.1066, "step": 42596 }, { "epoch": 31.206593406593406, "grad_norm": 1.2421875, "learning_rate": 4.146783859940635e-06, "loss": 0.1384, "step": 42597 }, { "epoch": 31.207326007326007, "grad_norm": 1.25, "learning_rate": 4.145581804089983e-06, "loss": 0.1541, "step": 42598 }, { "epoch": 31.208058608058607, "grad_norm": 1.1015625, "learning_rate": 4.144379970338555e-06, "loss": 0.0978, "step": 42599 }, { "epoch": 31.208791208791208, "grad_norm": 1.2109375, "learning_rate": 4.143178358692121e-06, "loss": 0.1226, "step": 42600 }, { "epoch": 31.20952380952381, "grad_norm": 1.578125, "learning_rate": 4.141976969156439e-06, "loss": 0.1184, "step": 42601 }, { "epoch": 31.21025641025641, "grad_norm": 0.7734375, "learning_rate": 4.140775801737265e-06, "loss": 0.1436, "step": 42602 }, { "epoch": 31.21098901098901, "grad_norm": 0.9765625, "learning_rate": 4.139574856440358e-06, "loss": 0.1462, "step": 42603 }, { "epoch": 31.21172161172161, "grad_norm": 1.375, "learning_rate": 4.13837413327146e-06, "loss": 0.1772, "step": 42604 }, { "epoch": 31.21245421245421, "grad_norm": 0.87109375, "learning_rate": 4.137173632236341e-06, "loss": 0.1231, "step": 42605 }, { "epoch": 31.213186813186812, "grad_norm": 1.3515625, "learning_rate": 4.135973353340723e-06, "loss": 0.1107, "step": 42606 }, { "epoch": 31.213919413919413, "grad_norm": 1.125, "learning_rate": 4.134773296590394e-06, "loss": 0.1362, "step": 42607 }, { "epoch": 31.214652014652014, "grad_norm": 1.03125, "learning_rate": 4.1335734619910776e-06, "loss": 0.1234, "step": 42608 }, { "epoch": 31.215384615384615, "grad_norm": 0.92578125, "learning_rate": 4.132373849548547e-06, "loss": 0.1228, "step": 42609 }, { "epoch": 31.216117216117215, "grad_norm": 0.91015625, "learning_rate": 4.131174459268521e-06, "loss": 0.1249, "step": 42610 }, { "epoch": 31.216849816849816, "grad_norm": 0.88671875, "learning_rate": 4.129975291156759e-06, "loss": 0.124, "step": 42611 }, { "epoch": 31.217582417582417, "grad_norm": 1.125, "learning_rate": 4.128776345219011e-06, "loss": 0.138, "step": 42612 }, { "epoch": 31.218315018315018, "grad_norm": 0.9375, "learning_rate": 4.127577621461027e-06, "loss": 0.1015, "step": 42613 }, { "epoch": 31.21904761904762, "grad_norm": 0.99609375, "learning_rate": 4.1263791198885276e-06, "loss": 0.1165, "step": 42614 }, { "epoch": 31.21978021978022, "grad_norm": 1.0703125, "learning_rate": 4.125180840507277e-06, "loss": 0.1213, "step": 42615 }, { "epoch": 31.22051282051282, "grad_norm": 1.1171875, "learning_rate": 4.123982783323005e-06, "loss": 0.1528, "step": 42616 }, { "epoch": 31.22124542124542, "grad_norm": 1.3125, "learning_rate": 4.122784948341456e-06, "loss": 0.1401, "step": 42617 }, { "epoch": 31.22197802197802, "grad_norm": 0.734375, "learning_rate": 4.121587335568372e-06, "loss": 0.1447, "step": 42618 }, { "epoch": 31.222710622710622, "grad_norm": 0.80078125, "learning_rate": 4.120389945009484e-06, "loss": 0.1172, "step": 42619 }, { "epoch": 31.223443223443223, "grad_norm": 0.81640625, "learning_rate": 4.119192776670543e-06, "loss": 0.1164, "step": 42620 }, { "epoch": 31.224175824175823, "grad_norm": 1.09375, "learning_rate": 4.117995830557261e-06, "loss": 0.1009, "step": 42621 }, { "epoch": 31.224908424908424, "grad_norm": 1.515625, "learning_rate": 4.1167991066754e-06, "loss": 0.1466, "step": 42622 }, { "epoch": 31.225641025641025, "grad_norm": 1.0859375, "learning_rate": 4.11560260503067e-06, "loss": 0.1261, "step": 42623 }, { "epoch": 31.226373626373626, "grad_norm": 1.9296875, "learning_rate": 4.1144063256288295e-06, "loss": 0.1654, "step": 42624 }, { "epoch": 31.227106227106226, "grad_norm": 0.91015625, "learning_rate": 4.113210268475583e-06, "loss": 0.1189, "step": 42625 }, { "epoch": 31.227838827838827, "grad_norm": 0.6796875, "learning_rate": 4.112014433576679e-06, "loss": 0.097, "step": 42626 }, { "epoch": 31.228571428571428, "grad_norm": 1.28125, "learning_rate": 4.110818820937837e-06, "loss": 0.115, "step": 42627 }, { "epoch": 31.22930402930403, "grad_norm": 0.91015625, "learning_rate": 4.109623430564792e-06, "loss": 0.1176, "step": 42628 }, { "epoch": 31.23003663003663, "grad_norm": 1.3203125, "learning_rate": 4.108428262463279e-06, "loss": 0.1259, "step": 42629 }, { "epoch": 31.23076923076923, "grad_norm": 0.61328125, "learning_rate": 4.107233316639009e-06, "loss": 0.0923, "step": 42630 }, { "epoch": 31.23150183150183, "grad_norm": 1.0390625, "learning_rate": 4.106038593097718e-06, "loss": 0.1111, "step": 42631 }, { "epoch": 31.23223443223443, "grad_norm": 1.796875, "learning_rate": 4.104844091845112e-06, "loss": 0.1109, "step": 42632 }, { "epoch": 31.232967032967032, "grad_norm": 0.890625, "learning_rate": 4.103649812886944e-06, "loss": 0.1232, "step": 42633 }, { "epoch": 31.233699633699633, "grad_norm": 1.3671875, "learning_rate": 4.102455756228913e-06, "loss": 0.1444, "step": 42634 }, { "epoch": 31.234432234432234, "grad_norm": 0.95703125, "learning_rate": 4.101261921876753e-06, "loss": 0.1246, "step": 42635 }, { "epoch": 31.235164835164834, "grad_norm": 1.3359375, "learning_rate": 4.100068309836176e-06, "loss": 0.1486, "step": 42636 }, { "epoch": 31.235897435897435, "grad_norm": 0.9921875, "learning_rate": 4.0988749201129005e-06, "loss": 0.1171, "step": 42637 }, { "epoch": 31.236630036630036, "grad_norm": 1.703125, "learning_rate": 4.09768175271265e-06, "loss": 0.1231, "step": 42638 }, { "epoch": 31.237362637362637, "grad_norm": 0.94921875, "learning_rate": 4.096488807641137e-06, "loss": 0.1323, "step": 42639 }, { "epoch": 31.238095238095237, "grad_norm": 1.0703125, "learning_rate": 4.095296084904092e-06, "loss": 0.1437, "step": 42640 }, { "epoch": 31.238827838827838, "grad_norm": 0.97265625, "learning_rate": 4.094103584507203e-06, "loss": 0.1212, "step": 42641 }, { "epoch": 31.23956043956044, "grad_norm": 1.0625, "learning_rate": 4.092911306456203e-06, "loss": 0.1162, "step": 42642 }, { "epoch": 31.24029304029304, "grad_norm": 1.25, "learning_rate": 4.091719250756794e-06, "loss": 0.1345, "step": 42643 }, { "epoch": 31.24102564102564, "grad_norm": 0.703125, "learning_rate": 4.0905274174147055e-06, "loss": 0.1197, "step": 42644 }, { "epoch": 31.24175824175824, "grad_norm": 1.28125, "learning_rate": 4.089335806435624e-06, "loss": 0.1409, "step": 42645 }, { "epoch": 31.24249084249084, "grad_norm": 1.1171875, "learning_rate": 4.088144417825281e-06, "loss": 0.1508, "step": 42646 }, { "epoch": 31.243223443223442, "grad_norm": 1.359375, "learning_rate": 4.086953251589353e-06, "loss": 0.1302, "step": 42647 }, { "epoch": 31.243956043956043, "grad_norm": 0.98046875, "learning_rate": 4.08576230773359e-06, "loss": 0.1162, "step": 42648 }, { "epoch": 31.244688644688644, "grad_norm": 1.171875, "learning_rate": 4.084571586263669e-06, "loss": 0.1202, "step": 42649 }, { "epoch": 31.245421245421245, "grad_norm": 1.4296875, "learning_rate": 4.083381087185303e-06, "loss": 0.1208, "step": 42650 }, { "epoch": 31.246153846153845, "grad_norm": 1.15625, "learning_rate": 4.082190810504195e-06, "loss": 0.1334, "step": 42651 }, { "epoch": 31.246886446886446, "grad_norm": 0.75390625, "learning_rate": 4.08100075622605e-06, "loss": 0.1139, "step": 42652 }, { "epoch": 31.247619047619047, "grad_norm": 0.8046875, "learning_rate": 4.079810924356566e-06, "loss": 0.1339, "step": 42653 }, { "epoch": 31.248351648351647, "grad_norm": 1.390625, "learning_rate": 4.078621314901444e-06, "loss": 0.1064, "step": 42654 }, { "epoch": 31.249084249084248, "grad_norm": 1.1015625, "learning_rate": 4.0774319278663965e-06, "loss": 0.129, "step": 42655 }, { "epoch": 31.24981684981685, "grad_norm": 1.0078125, "learning_rate": 4.076242763257104e-06, "loss": 0.1206, "step": 42656 }, { "epoch": 31.25054945054945, "grad_norm": 1.4296875, "learning_rate": 4.075053821079283e-06, "loss": 0.1367, "step": 42657 }, { "epoch": 31.25128205128205, "grad_norm": 1.625, "learning_rate": 4.073865101338599e-06, "loss": 0.134, "step": 42658 }, { "epoch": 31.25201465201465, "grad_norm": 1.0078125, "learning_rate": 4.072676604040791e-06, "loss": 0.1107, "step": 42659 }, { "epoch": 31.252747252747252, "grad_norm": 1.09375, "learning_rate": 4.071488329191517e-06, "loss": 0.1255, "step": 42660 }, { "epoch": 31.253479853479853, "grad_norm": 1.046875, "learning_rate": 4.070300276796493e-06, "loss": 0.1159, "step": 42661 }, { "epoch": 31.254212454212453, "grad_norm": 1.2265625, "learning_rate": 4.069112446861391e-06, "loss": 0.0905, "step": 42662 }, { "epoch": 31.254945054945054, "grad_norm": 0.68359375, "learning_rate": 4.067924839391917e-06, "loss": 0.1343, "step": 42663 }, { "epoch": 31.255677655677655, "grad_norm": 1.2109375, "learning_rate": 4.066737454393755e-06, "loss": 0.1354, "step": 42664 }, { "epoch": 31.256410256410255, "grad_norm": 0.8046875, "learning_rate": 4.065550291872602e-06, "loss": 0.0946, "step": 42665 }, { "epoch": 31.257142857142856, "grad_norm": 0.875, "learning_rate": 4.0643633518341455e-06, "loss": 0.1234, "step": 42666 }, { "epoch": 31.257875457875457, "grad_norm": 1.171875, "learning_rate": 4.063176634284061e-06, "loss": 0.108, "step": 42667 }, { "epoch": 31.258608058608058, "grad_norm": 0.84765625, "learning_rate": 4.0619901392280434e-06, "loss": 0.1109, "step": 42668 }, { "epoch": 31.25934065934066, "grad_norm": 1.1640625, "learning_rate": 4.060803866671774e-06, "loss": 0.1104, "step": 42669 }, { "epoch": 31.26007326007326, "grad_norm": 1.1171875, "learning_rate": 4.059617816620941e-06, "loss": 0.1118, "step": 42670 }, { "epoch": 31.26080586080586, "grad_norm": 0.74609375, "learning_rate": 4.058431989081221e-06, "loss": 0.1411, "step": 42671 }, { "epoch": 31.26153846153846, "grad_norm": 0.82421875, "learning_rate": 4.057246384058308e-06, "loss": 0.1068, "step": 42672 }, { "epoch": 31.26227106227106, "grad_norm": 0.953125, "learning_rate": 4.056061001557858e-06, "loss": 0.1357, "step": 42673 }, { "epoch": 31.263003663003662, "grad_norm": 0.734375, "learning_rate": 4.054875841585578e-06, "loss": 0.1163, "step": 42674 }, { "epoch": 31.263736263736263, "grad_norm": 1.40625, "learning_rate": 4.053690904147132e-06, "loss": 0.1459, "step": 42675 }, { "epoch": 31.264468864468864, "grad_norm": 1.3515625, "learning_rate": 4.052506189248206e-06, "loss": 0.1216, "step": 42676 }, { "epoch": 31.265201465201464, "grad_norm": 1.6328125, "learning_rate": 4.051321696894461e-06, "loss": 0.1229, "step": 42677 }, { "epoch": 31.265934065934065, "grad_norm": 0.890625, "learning_rate": 4.0501374270915845e-06, "loss": 0.0839, "step": 42678 }, { "epoch": 31.266666666666666, "grad_norm": 1.1640625, "learning_rate": 4.048953379845248e-06, "loss": 0.1017, "step": 42679 }, { "epoch": 31.267399267399266, "grad_norm": 0.98046875, "learning_rate": 4.047769555161124e-06, "loss": 0.1381, "step": 42680 }, { "epoch": 31.268131868131867, "grad_norm": 0.96875, "learning_rate": 4.046585953044893e-06, "loss": 0.0864, "step": 42681 }, { "epoch": 31.268864468864468, "grad_norm": 0.82421875, "learning_rate": 4.045402573502212e-06, "loss": 0.1341, "step": 42682 }, { "epoch": 31.26959706959707, "grad_norm": 1.109375, "learning_rate": 4.044219416538762e-06, "loss": 0.1434, "step": 42683 }, { "epoch": 31.27032967032967, "grad_norm": 1.125, "learning_rate": 4.0430364821601935e-06, "loss": 0.1472, "step": 42684 }, { "epoch": 31.27106227106227, "grad_norm": 1.3515625, "learning_rate": 4.041853770372205e-06, "loss": 0.1228, "step": 42685 }, { "epoch": 31.27179487179487, "grad_norm": 0.75390625, "learning_rate": 4.040671281180434e-06, "loss": 0.1147, "step": 42686 }, { "epoch": 31.27252747252747, "grad_norm": 1.0625, "learning_rate": 4.039489014590569e-06, "loss": 0.119, "step": 42687 }, { "epoch": 31.273260073260072, "grad_norm": 0.81640625, "learning_rate": 4.038306970608262e-06, "loss": 0.128, "step": 42688 }, { "epoch": 31.273992673992673, "grad_norm": 1.1015625, "learning_rate": 4.037125149239169e-06, "loss": 0.1139, "step": 42689 }, { "epoch": 31.274725274725274, "grad_norm": 0.69140625, "learning_rate": 4.035943550488966e-06, "loss": 0.0962, "step": 42690 }, { "epoch": 31.275457875457874, "grad_norm": 0.8046875, "learning_rate": 4.034762174363315e-06, "loss": 0.1153, "step": 42691 }, { "epoch": 31.276190476190475, "grad_norm": 1.1953125, "learning_rate": 4.033581020867874e-06, "loss": 0.1463, "step": 42692 }, { "epoch": 31.276923076923076, "grad_norm": 1.28125, "learning_rate": 4.0324000900082965e-06, "loss": 0.1226, "step": 42693 }, { "epoch": 31.277655677655677, "grad_norm": 0.9609375, "learning_rate": 4.031219381790237e-06, "loss": 0.1382, "step": 42694 }, { "epoch": 31.278388278388277, "grad_norm": 1.4296875, "learning_rate": 4.03003889621937e-06, "loss": 0.167, "step": 42695 }, { "epoch": 31.279120879120878, "grad_norm": 1.2421875, "learning_rate": 4.028858633301345e-06, "loss": 0.153, "step": 42696 }, { "epoch": 31.27985347985348, "grad_norm": 0.7890625, "learning_rate": 4.0276785930418055e-06, "loss": 0.1214, "step": 42697 }, { "epoch": 31.28058608058608, "grad_norm": 0.87109375, "learning_rate": 4.02649877544642e-06, "loss": 0.1159, "step": 42698 }, { "epoch": 31.28131868131868, "grad_norm": 0.94921875, "learning_rate": 4.025319180520824e-06, "loss": 0.1226, "step": 42699 }, { "epoch": 31.28205128205128, "grad_norm": 0.87109375, "learning_rate": 4.024139808270693e-06, "loss": 0.1203, "step": 42700 }, { "epoch": 31.282783882783882, "grad_norm": 0.875, "learning_rate": 4.022960658701659e-06, "loss": 0.1224, "step": 42701 }, { "epoch": 31.283516483516483, "grad_norm": 1.1484375, "learning_rate": 4.021781731819383e-06, "loss": 0.1094, "step": 42702 }, { "epoch": 31.284249084249083, "grad_norm": 0.84375, "learning_rate": 4.0206030276295e-06, "loss": 0.1169, "step": 42703 }, { "epoch": 31.284981684981684, "grad_norm": 0.60546875, "learning_rate": 4.019424546137668e-06, "loss": 0.0718, "step": 42704 }, { "epoch": 31.285714285714285, "grad_norm": 1.8515625, "learning_rate": 4.018246287349533e-06, "loss": 0.1152, "step": 42705 }, { "epoch": 31.286446886446885, "grad_norm": 1.5234375, "learning_rate": 4.017068251270738e-06, "loss": 0.1097, "step": 42706 }, { "epoch": 31.287179487179486, "grad_norm": 1.1171875, "learning_rate": 4.015890437906937e-06, "loss": 0.114, "step": 42707 }, { "epoch": 31.287912087912087, "grad_norm": 1.1484375, "learning_rate": 4.014712847263753e-06, "loss": 0.112, "step": 42708 }, { "epoch": 31.288644688644688, "grad_norm": 1.3359375, "learning_rate": 4.013535479346853e-06, "loss": 0.1448, "step": 42709 }, { "epoch": 31.28937728937729, "grad_norm": 1.421875, "learning_rate": 4.012358334161846e-06, "loss": 0.0853, "step": 42710 }, { "epoch": 31.29010989010989, "grad_norm": 0.9765625, "learning_rate": 4.01118141171441e-06, "loss": 0.1168, "step": 42711 }, { "epoch": 31.29084249084249, "grad_norm": 1.125, "learning_rate": 4.010004712010156e-06, "loss": 0.1632, "step": 42712 }, { "epoch": 31.29157509157509, "grad_norm": 0.8046875, "learning_rate": 4.008828235054739e-06, "loss": 0.1209, "step": 42713 }, { "epoch": 31.29230769230769, "grad_norm": 1.0859375, "learning_rate": 4.007651980853774e-06, "loss": 0.0954, "step": 42714 }, { "epoch": 31.293040293040292, "grad_norm": 0.6484375, "learning_rate": 4.006475949412919e-06, "loss": 0.1012, "step": 42715 }, { "epoch": 31.293772893772893, "grad_norm": 1.1953125, "learning_rate": 4.005300140737794e-06, "loss": 0.1464, "step": 42716 }, { "epoch": 31.294505494505493, "grad_norm": 1.0625, "learning_rate": 4.004124554834043e-06, "loss": 0.1281, "step": 42717 }, { "epoch": 31.295238095238094, "grad_norm": 1.359375, "learning_rate": 4.002949191707306e-06, "loss": 0.1566, "step": 42718 }, { "epoch": 31.295970695970695, "grad_norm": 1.625, "learning_rate": 4.001774051363191e-06, "loss": 0.1236, "step": 42719 }, { "epoch": 31.296703296703296, "grad_norm": 1.1484375, "learning_rate": 4.00059913380734e-06, "loss": 0.1435, "step": 42720 }, { "epoch": 31.297435897435896, "grad_norm": 2.03125, "learning_rate": 3.9994244390453895e-06, "loss": 0.1424, "step": 42721 }, { "epoch": 31.298168498168497, "grad_norm": 0.765625, "learning_rate": 3.998249967082961e-06, "loss": 0.1015, "step": 42722 }, { "epoch": 31.298901098901098, "grad_norm": 0.8984375, "learning_rate": 3.997075717925683e-06, "loss": 0.1247, "step": 42723 }, { "epoch": 31.2996336996337, "grad_norm": 1.015625, "learning_rate": 3.995901691579183e-06, "loss": 0.1184, "step": 42724 }, { "epoch": 31.3003663003663, "grad_norm": 1.1640625, "learning_rate": 3.9947278880490724e-06, "loss": 0.1217, "step": 42725 }, { "epoch": 31.3010989010989, "grad_norm": 0.84375, "learning_rate": 3.993554307340999e-06, "loss": 0.1326, "step": 42726 }, { "epoch": 31.3018315018315, "grad_norm": 1.421875, "learning_rate": 3.992380949460565e-06, "loss": 0.1273, "step": 42727 }, { "epoch": 31.3025641025641, "grad_norm": 0.8671875, "learning_rate": 3.991207814413413e-06, "loss": 0.1111, "step": 42728 }, { "epoch": 31.303296703296702, "grad_norm": 1.34375, "learning_rate": 3.9900349022051434e-06, "loss": 0.0916, "step": 42729 }, { "epoch": 31.304029304029303, "grad_norm": 0.734375, "learning_rate": 3.988862212841384e-06, "loss": 0.1081, "step": 42730 }, { "epoch": 31.304761904761904, "grad_norm": 1.5625, "learning_rate": 3.987689746327755e-06, "loss": 0.1239, "step": 42731 }, { "epoch": 31.305494505494504, "grad_norm": 0.93359375, "learning_rate": 3.986517502669876e-06, "loss": 0.146, "step": 42732 }, { "epoch": 31.306227106227105, "grad_norm": 1.1328125, "learning_rate": 3.985345481873366e-06, "loss": 0.0918, "step": 42733 }, { "epoch": 31.306959706959706, "grad_norm": 0.97265625, "learning_rate": 3.984173683943827e-06, "loss": 0.1356, "step": 42734 }, { "epoch": 31.307692307692307, "grad_norm": 1.0390625, "learning_rate": 3.983002108886892e-06, "loss": 0.1115, "step": 42735 }, { "epoch": 31.308424908424907, "grad_norm": 0.875, "learning_rate": 3.981830756708149e-06, "loss": 0.1105, "step": 42736 }, { "epoch": 31.309157509157508, "grad_norm": 1.0703125, "learning_rate": 3.980659627413242e-06, "loss": 0.1406, "step": 42737 }, { "epoch": 31.30989010989011, "grad_norm": 1.2265625, "learning_rate": 3.979488721007755e-06, "loss": 0.1298, "step": 42738 }, { "epoch": 31.31062271062271, "grad_norm": 1.203125, "learning_rate": 3.978318037497316e-06, "loss": 0.1206, "step": 42739 }, { "epoch": 31.31135531135531, "grad_norm": 1.0625, "learning_rate": 3.977147576887525e-06, "loss": 0.1023, "step": 42740 }, { "epoch": 31.31208791208791, "grad_norm": 1.2734375, "learning_rate": 3.975977339183991e-06, "loss": 0.1684, "step": 42741 }, { "epoch": 31.31282051282051, "grad_norm": 0.91796875, "learning_rate": 3.9748073243923215e-06, "loss": 0.1094, "step": 42742 }, { "epoch": 31.313553113553112, "grad_norm": 1.390625, "learning_rate": 3.9736375325181205e-06, "loss": 0.1157, "step": 42743 }, { "epoch": 31.314285714285713, "grad_norm": 0.85546875, "learning_rate": 3.972467963567007e-06, "loss": 0.1028, "step": 42744 }, { "epoch": 31.315018315018314, "grad_norm": 1.0, "learning_rate": 3.9712986175445625e-06, "loss": 0.0999, "step": 42745 }, { "epoch": 31.315750915750915, "grad_norm": 1.484375, "learning_rate": 3.970129494456404e-06, "loss": 0.116, "step": 42746 }, { "epoch": 31.316483516483515, "grad_norm": 1.0703125, "learning_rate": 3.968960594308126e-06, "loss": 0.129, "step": 42747 }, { "epoch": 31.317216117216116, "grad_norm": 0.78125, "learning_rate": 3.967791917105344e-06, "loss": 0.1079, "step": 42748 }, { "epoch": 31.317948717948717, "grad_norm": 1.3984375, "learning_rate": 3.966623462853639e-06, "loss": 0.1407, "step": 42749 }, { "epoch": 31.318681318681318, "grad_norm": 1.078125, "learning_rate": 3.965455231558625e-06, "loss": 0.1021, "step": 42750 }, { "epoch": 31.31941391941392, "grad_norm": 0.91796875, "learning_rate": 3.964287223225874e-06, "loss": 0.1109, "step": 42751 }, { "epoch": 31.32014652014652, "grad_norm": 0.9453125, "learning_rate": 3.963119437861011e-06, "loss": 0.1146, "step": 42752 }, { "epoch": 31.32087912087912, "grad_norm": 0.953125, "learning_rate": 3.961951875469616e-06, "loss": 0.1228, "step": 42753 }, { "epoch": 31.32161172161172, "grad_norm": 1.0078125, "learning_rate": 3.9607845360572925e-06, "loss": 0.113, "step": 42754 }, { "epoch": 31.32234432234432, "grad_norm": 1.0078125, "learning_rate": 3.959617419629627e-06, "loss": 0.1576, "step": 42755 }, { "epoch": 31.323076923076922, "grad_norm": 0.8515625, "learning_rate": 3.9584505261922085e-06, "loss": 0.1114, "step": 42756 }, { "epoch": 31.323809523809523, "grad_norm": 0.9609375, "learning_rate": 3.957283855750633e-06, "loss": 0.1132, "step": 42757 }, { "epoch": 31.324542124542123, "grad_norm": 0.96875, "learning_rate": 3.956117408310487e-06, "loss": 0.1297, "step": 42758 }, { "epoch": 31.325274725274724, "grad_norm": 0.8125, "learning_rate": 3.954951183877375e-06, "loss": 0.1139, "step": 42759 }, { "epoch": 31.326007326007325, "grad_norm": 0.63671875, "learning_rate": 3.953785182456861e-06, "loss": 0.1076, "step": 42760 }, { "epoch": 31.326739926739926, "grad_norm": 0.90625, "learning_rate": 3.95261940405455e-06, "loss": 0.1094, "step": 42761 }, { "epoch": 31.327472527472526, "grad_norm": 1.0703125, "learning_rate": 3.951453848676005e-06, "loss": 0.0985, "step": 42762 }, { "epoch": 31.328205128205127, "grad_norm": 1.234375, "learning_rate": 3.950288516326836e-06, "loss": 0.1042, "step": 42763 }, { "epoch": 31.328937728937728, "grad_norm": 1.25, "learning_rate": 3.9491234070126105e-06, "loss": 0.1399, "step": 42764 }, { "epoch": 31.32967032967033, "grad_norm": 1.0390625, "learning_rate": 3.947958520738928e-06, "loss": 0.1081, "step": 42765 }, { "epoch": 31.33040293040293, "grad_norm": 1.34375, "learning_rate": 3.946793857511346e-06, "loss": 0.1306, "step": 42766 }, { "epoch": 31.33113553113553, "grad_norm": 0.87890625, "learning_rate": 3.945629417335464e-06, "loss": 0.1419, "step": 42767 }, { "epoch": 31.33186813186813, "grad_norm": 0.57421875, "learning_rate": 3.944465200216848e-06, "loss": 0.0941, "step": 42768 }, { "epoch": 31.33260073260073, "grad_norm": 1.15625, "learning_rate": 3.943301206161082e-06, "loss": 0.1165, "step": 42769 }, { "epoch": 31.333333333333332, "grad_norm": 1.25, "learning_rate": 3.9421374351737555e-06, "loss": 0.0987, "step": 42770 }, { "epoch": 31.334065934065933, "grad_norm": 1.0, "learning_rate": 3.9409738872604185e-06, "loss": 0.1028, "step": 42771 }, { "epoch": 31.334798534798534, "grad_norm": 1.1328125, "learning_rate": 3.9398105624266665e-06, "loss": 0.1285, "step": 42772 }, { "epoch": 31.335531135531134, "grad_norm": 0.87890625, "learning_rate": 3.938647460678066e-06, "loss": 0.1244, "step": 42773 }, { "epoch": 31.336263736263735, "grad_norm": 1.3125, "learning_rate": 3.937484582020194e-06, "loss": 0.1621, "step": 42774 }, { "epoch": 31.336996336996336, "grad_norm": 1.4140625, "learning_rate": 3.936321926458613e-06, "loss": 0.1272, "step": 42775 }, { "epoch": 31.337728937728937, "grad_norm": 1.4296875, "learning_rate": 3.935159493998906e-06, "loss": 0.105, "step": 42776 }, { "epoch": 31.338461538461537, "grad_norm": 1.40625, "learning_rate": 3.933997284646626e-06, "loss": 0.1201, "step": 42777 }, { "epoch": 31.339194139194138, "grad_norm": 0.921875, "learning_rate": 3.932835298407364e-06, "loss": 0.1171, "step": 42778 }, { "epoch": 31.33992673992674, "grad_norm": 0.9296875, "learning_rate": 3.931673535286667e-06, "loss": 0.1178, "step": 42779 }, { "epoch": 31.34065934065934, "grad_norm": 1.1640625, "learning_rate": 3.9305119952901066e-06, "loss": 0.1333, "step": 42780 }, { "epoch": 31.34139194139194, "grad_norm": 1.265625, "learning_rate": 3.9293506784232616e-06, "loss": 0.1123, "step": 42781 }, { "epoch": 31.34212454212454, "grad_norm": 1.453125, "learning_rate": 3.928189584691678e-06, "loss": 0.123, "step": 42782 }, { "epoch": 31.34285714285714, "grad_norm": 1.5703125, "learning_rate": 3.927028714100926e-06, "loss": 0.125, "step": 42783 }, { "epoch": 31.343589743589742, "grad_norm": 0.61328125, "learning_rate": 3.925868066656569e-06, "loss": 0.1018, "step": 42784 }, { "epoch": 31.344322344322343, "grad_norm": 0.7265625, "learning_rate": 3.924707642364178e-06, "loss": 0.1373, "step": 42785 }, { "epoch": 31.345054945054944, "grad_norm": 1.5625, "learning_rate": 3.923547441229289e-06, "loss": 0.1397, "step": 42786 }, { "epoch": 31.345787545787545, "grad_norm": 1.21875, "learning_rate": 3.922387463257484e-06, "loss": 0.1379, "step": 42787 }, { "epoch": 31.346520146520145, "grad_norm": 1.234375, "learning_rate": 3.9212277084542985e-06, "loss": 0.155, "step": 42788 }, { "epoch": 31.347252747252746, "grad_norm": 0.734375, "learning_rate": 3.920068176825315e-06, "loss": 0.1047, "step": 42789 }, { "epoch": 31.347985347985347, "grad_norm": 0.77734375, "learning_rate": 3.918908868376066e-06, "loss": 0.141, "step": 42790 }, { "epoch": 31.348717948717947, "grad_norm": 0.96484375, "learning_rate": 3.917749783112127e-06, "loss": 0.1158, "step": 42791 }, { "epoch": 31.349450549450548, "grad_norm": 0.5859375, "learning_rate": 3.91659092103903e-06, "loss": 0.1217, "step": 42792 }, { "epoch": 31.35018315018315, "grad_norm": 0.6953125, "learning_rate": 3.915432282162345e-06, "loss": 0.1355, "step": 42793 }, { "epoch": 31.35091575091575, "grad_norm": 0.94140625, "learning_rate": 3.914273866487612e-06, "loss": 0.123, "step": 42794 }, { "epoch": 31.35164835164835, "grad_norm": 0.7890625, "learning_rate": 3.913115674020386e-06, "loss": 0.1046, "step": 42795 }, { "epoch": 31.35238095238095, "grad_norm": 1.609375, "learning_rate": 3.911957704766221e-06, "loss": 0.1226, "step": 42796 }, { "epoch": 31.353113553113552, "grad_norm": 0.95703125, "learning_rate": 3.910799958730658e-06, "loss": 0.1155, "step": 42797 }, { "epoch": 31.353846153846153, "grad_norm": 1.1640625, "learning_rate": 3.909642435919246e-06, "loss": 0.1332, "step": 42798 }, { "epoch": 31.354578754578753, "grad_norm": 0.87109375, "learning_rate": 3.908485136337531e-06, "loss": 0.1385, "step": 42799 }, { "epoch": 31.355311355311354, "grad_norm": 1.234375, "learning_rate": 3.907328059991063e-06, "loss": 0.1123, "step": 42800 }, { "epoch": 31.356043956043955, "grad_norm": 1.265625, "learning_rate": 3.906171206885377e-06, "loss": 0.1204, "step": 42801 }, { "epoch": 31.356776556776556, "grad_norm": 1.0859375, "learning_rate": 3.905014577026028e-06, "loss": 0.1107, "step": 42802 }, { "epoch": 31.357509157509156, "grad_norm": 0.9296875, "learning_rate": 3.903858170418534e-06, "loss": 0.1125, "step": 42803 }, { "epoch": 31.358241758241757, "grad_norm": 1.0625, "learning_rate": 3.902701987068468e-06, "loss": 0.118, "step": 42804 }, { "epoch": 31.358974358974358, "grad_norm": 1.09375, "learning_rate": 3.901546026981345e-06, "loss": 0.1205, "step": 42805 }, { "epoch": 31.35970695970696, "grad_norm": 0.88671875, "learning_rate": 3.900390290162714e-06, "loss": 0.1103, "step": 42806 }, { "epoch": 31.36043956043956, "grad_norm": 1.09375, "learning_rate": 3.899234776618119e-06, "loss": 0.1055, "step": 42807 }, { "epoch": 31.36117216117216, "grad_norm": 1.109375, "learning_rate": 3.898079486353081e-06, "loss": 0.1336, "step": 42808 }, { "epoch": 31.36190476190476, "grad_norm": 0.8046875, "learning_rate": 3.896924419373136e-06, "loss": 0.104, "step": 42809 }, { "epoch": 31.36263736263736, "grad_norm": 1.28125, "learning_rate": 3.895769575683834e-06, "loss": 0.1115, "step": 42810 }, { "epoch": 31.363369963369962, "grad_norm": 1.3046875, "learning_rate": 3.894614955290703e-06, "loss": 0.1179, "step": 42811 }, { "epoch": 31.364102564102563, "grad_norm": 0.80078125, "learning_rate": 3.893460558199262e-06, "loss": 0.1151, "step": 42812 }, { "epoch": 31.364835164835164, "grad_norm": 0.953125, "learning_rate": 3.89230638441506e-06, "loss": 0.1117, "step": 42813 }, { "epoch": 31.365567765567764, "grad_norm": 1.3515625, "learning_rate": 3.891152433943604e-06, "loss": 0.1428, "step": 42814 }, { "epoch": 31.366300366300365, "grad_norm": 1.3828125, "learning_rate": 3.889998706790451e-06, "loss": 0.148, "step": 42815 }, { "epoch": 31.367032967032966, "grad_norm": 0.95703125, "learning_rate": 3.888845202961107e-06, "loss": 0.0885, "step": 42816 }, { "epoch": 31.367765567765566, "grad_norm": 1.1953125, "learning_rate": 3.8876919224611165e-06, "loss": 0.1497, "step": 42817 }, { "epoch": 31.368498168498167, "grad_norm": 1.0625, "learning_rate": 3.886538865295987e-06, "loss": 0.1167, "step": 42818 }, { "epoch": 31.369230769230768, "grad_norm": 1.125, "learning_rate": 3.885386031471258e-06, "loss": 0.1269, "step": 42819 }, { "epoch": 31.36996336996337, "grad_norm": 1.21875, "learning_rate": 3.88423342099244e-06, "loss": 0.1202, "step": 42820 }, { "epoch": 31.37069597069597, "grad_norm": 1.0703125, "learning_rate": 3.883081033865065e-06, "loss": 0.1069, "step": 42821 }, { "epoch": 31.37142857142857, "grad_norm": 1.0625, "learning_rate": 3.881928870094658e-06, "loss": 0.1235, "step": 42822 }, { "epoch": 31.37216117216117, "grad_norm": 1.1484375, "learning_rate": 3.880776929686726e-06, "loss": 0.139, "step": 42823 }, { "epoch": 31.37289377289377, "grad_norm": 1.3671875, "learning_rate": 3.879625212646801e-06, "loss": 0.1212, "step": 42824 }, { "epoch": 31.373626373626372, "grad_norm": 0.66796875, "learning_rate": 3.878473718980391e-06, "loss": 0.1298, "step": 42825 }, { "epoch": 31.374358974358973, "grad_norm": 1.15625, "learning_rate": 3.8773224486930265e-06, "loss": 0.1207, "step": 42826 }, { "epoch": 31.375091575091574, "grad_norm": 1.0, "learning_rate": 3.876171401790209e-06, "loss": 0.0928, "step": 42827 }, { "epoch": 31.375824175824174, "grad_norm": 0.9140625, "learning_rate": 3.875020578277465e-06, "loss": 0.1315, "step": 42828 }, { "epoch": 31.376556776556775, "grad_norm": 1.0703125, "learning_rate": 3.873869978160292e-06, "loss": 0.129, "step": 42829 }, { "epoch": 31.377289377289376, "grad_norm": 0.9375, "learning_rate": 3.872719601444224e-06, "loss": 0.1223, "step": 42830 }, { "epoch": 31.378021978021977, "grad_norm": 0.9765625, "learning_rate": 3.8715694481347626e-06, "loss": 0.1334, "step": 42831 }, { "epoch": 31.378754578754577, "grad_norm": 0.8125, "learning_rate": 3.870419518237418e-06, "loss": 0.1146, "step": 42832 }, { "epoch": 31.379487179487178, "grad_norm": 1.0390625, "learning_rate": 3.869269811757704e-06, "loss": 0.158, "step": 42833 }, { "epoch": 31.38021978021978, "grad_norm": 1.2421875, "learning_rate": 3.868120328701121e-06, "loss": 0.1331, "step": 42834 }, { "epoch": 31.38095238095238, "grad_norm": 1.3515625, "learning_rate": 3.866971069073181e-06, "loss": 0.121, "step": 42835 }, { "epoch": 31.38168498168498, "grad_norm": 0.859375, "learning_rate": 3.865822032879389e-06, "loss": 0.1082, "step": 42836 }, { "epoch": 31.38241758241758, "grad_norm": 1.21875, "learning_rate": 3.8646732201252655e-06, "loss": 0.103, "step": 42837 }, { "epoch": 31.383150183150182, "grad_norm": 0.68359375, "learning_rate": 3.863524630816294e-06, "loss": 0.1143, "step": 42838 }, { "epoch": 31.383882783882783, "grad_norm": 1.03125, "learning_rate": 3.862376264957987e-06, "loss": 0.0997, "step": 42839 }, { "epoch": 31.384615384615383, "grad_norm": 1.234375, "learning_rate": 3.861228122555839e-06, "loss": 0.1089, "step": 42840 }, { "epoch": 31.385347985347984, "grad_norm": 0.87890625, "learning_rate": 3.8600802036153686e-06, "loss": 0.0886, "step": 42841 }, { "epoch": 31.386080586080585, "grad_norm": 1.3359375, "learning_rate": 3.8589325081420536e-06, "loss": 0.1124, "step": 42842 }, { "epoch": 31.386813186813185, "grad_norm": 1.03125, "learning_rate": 3.857785036141418e-06, "loss": 0.1252, "step": 42843 }, { "epoch": 31.387545787545786, "grad_norm": 1.15625, "learning_rate": 3.856637787618936e-06, "loss": 0.124, "step": 42844 }, { "epoch": 31.388278388278387, "grad_norm": 0.515625, "learning_rate": 3.855490762580113e-06, "loss": 0.0983, "step": 42845 }, { "epoch": 31.389010989010988, "grad_norm": 1.0, "learning_rate": 3.854343961030446e-06, "loss": 0.0902, "step": 42846 }, { "epoch": 31.38974358974359, "grad_norm": 0.98828125, "learning_rate": 3.853197382975429e-06, "loss": 0.1345, "step": 42847 }, { "epoch": 31.39047619047619, "grad_norm": 1.1328125, "learning_rate": 3.852051028420564e-06, "loss": 0.1191, "step": 42848 }, { "epoch": 31.39120879120879, "grad_norm": 1.1015625, "learning_rate": 3.850904897371331e-06, "loss": 0.13, "step": 42849 }, { "epoch": 31.39194139194139, "grad_norm": 0.78515625, "learning_rate": 3.849758989833229e-06, "loss": 0.1117, "step": 42850 }, { "epoch": 31.39267399267399, "grad_norm": 0.9375, "learning_rate": 3.848613305811744e-06, "loss": 0.1227, "step": 42851 }, { "epoch": 31.393406593406592, "grad_norm": 1.0703125, "learning_rate": 3.8474678453123715e-06, "loss": 0.1237, "step": 42852 }, { "epoch": 31.394139194139193, "grad_norm": 1.2421875, "learning_rate": 3.84632260834059e-06, "loss": 0.1314, "step": 42853 }, { "epoch": 31.394871794871793, "grad_norm": 0.84765625, "learning_rate": 3.845177594901901e-06, "loss": 0.1026, "step": 42854 }, { "epoch": 31.395604395604394, "grad_norm": 1.1328125, "learning_rate": 3.844032805001768e-06, "loss": 0.1445, "step": 42855 }, { "epoch": 31.396336996336995, "grad_norm": 1.0546875, "learning_rate": 3.842888238645702e-06, "loss": 0.1457, "step": 42856 }, { "epoch": 31.397069597069596, "grad_norm": 1.0, "learning_rate": 3.841743895839165e-06, "loss": 0.1147, "step": 42857 }, { "epoch": 31.397802197802196, "grad_norm": 1.0, "learning_rate": 3.840599776587654e-06, "loss": 0.1124, "step": 42858 }, { "epoch": 31.398534798534797, "grad_norm": 1.1015625, "learning_rate": 3.839455880896654e-06, "loss": 0.1535, "step": 42859 }, { "epoch": 31.399267399267398, "grad_norm": 1.0234375, "learning_rate": 3.838312208771631e-06, "loss": 0.1324, "step": 42860 }, { "epoch": 31.4, "grad_norm": 0.8984375, "learning_rate": 3.837168760218071e-06, "loss": 0.1157, "step": 42861 }, { "epoch": 31.4007326007326, "grad_norm": 0.9921875, "learning_rate": 3.836025535241454e-06, "loss": 0.1175, "step": 42862 }, { "epoch": 31.4014652014652, "grad_norm": 1.2734375, "learning_rate": 3.834882533847269e-06, "loss": 0.1639, "step": 42863 }, { "epoch": 31.4021978021978, "grad_norm": 0.84765625, "learning_rate": 3.833739756040966e-06, "loss": 0.1502, "step": 42864 }, { "epoch": 31.4029304029304, "grad_norm": 1.2109375, "learning_rate": 3.832597201828049e-06, "loss": 0.1276, "step": 42865 }, { "epoch": 31.403663003663002, "grad_norm": 1.1484375, "learning_rate": 3.831454871213963e-06, "loss": 0.1038, "step": 42866 }, { "epoch": 31.404395604395603, "grad_norm": 1.2109375, "learning_rate": 3.830312764204208e-06, "loss": 0.131, "step": 42867 }, { "epoch": 31.405128205128204, "grad_norm": 1.21875, "learning_rate": 3.829170880804243e-06, "loss": 0.1219, "step": 42868 }, { "epoch": 31.405860805860804, "grad_norm": 0.8515625, "learning_rate": 3.8280292210195364e-06, "loss": 0.1316, "step": 42869 }, { "epoch": 31.406593406593405, "grad_norm": 0.72265625, "learning_rate": 3.8268877848555764e-06, "loss": 0.0999, "step": 42870 }, { "epoch": 31.407326007326006, "grad_norm": 1.0234375, "learning_rate": 3.825746572317804e-06, "loss": 0.1491, "step": 42871 }, { "epoch": 31.408058608058607, "grad_norm": 0.8515625, "learning_rate": 3.824605583411704e-06, "loss": 0.1225, "step": 42872 }, { "epoch": 31.408791208791207, "grad_norm": 0.64453125, "learning_rate": 3.823464818142745e-06, "loss": 0.1071, "step": 42873 }, { "epoch": 31.409523809523808, "grad_norm": 0.984375, "learning_rate": 3.8223242765163955e-06, "loss": 0.1115, "step": 42874 }, { "epoch": 31.41025641025641, "grad_norm": 0.9453125, "learning_rate": 3.821183958538106e-06, "loss": 0.1213, "step": 42875 }, { "epoch": 31.41098901098901, "grad_norm": 1.1171875, "learning_rate": 3.820043864213346e-06, "loss": 0.096, "step": 42876 }, { "epoch": 31.41172161172161, "grad_norm": 1.3984375, "learning_rate": 3.818903993547583e-06, "loss": 0.1423, "step": 42877 }, { "epoch": 31.41245421245421, "grad_norm": 1.75, "learning_rate": 3.817764346546279e-06, "loss": 0.1073, "step": 42878 }, { "epoch": 31.41318681318681, "grad_norm": 1.28125, "learning_rate": 3.816624923214885e-06, "loss": 0.1193, "step": 42879 }, { "epoch": 31.413919413919412, "grad_norm": 0.94921875, "learning_rate": 3.815485723558872e-06, "loss": 0.1019, "step": 42880 }, { "epoch": 31.414652014652013, "grad_norm": 1.3125, "learning_rate": 3.8143467475836776e-06, "loss": 0.122, "step": 42881 }, { "epoch": 31.415384615384614, "grad_norm": 0.83984375, "learning_rate": 3.813207995294793e-06, "loss": 0.1119, "step": 42882 }, { "epoch": 31.416117216117215, "grad_norm": 1.1640625, "learning_rate": 3.812069466697643e-06, "loss": 0.1067, "step": 42883 }, { "epoch": 31.416849816849815, "grad_norm": 0.9609375, "learning_rate": 3.8109311617976955e-06, "loss": 0.0973, "step": 42884 }, { "epoch": 31.417582417582416, "grad_norm": 1.671875, "learning_rate": 3.8097930806004128e-06, "loss": 0.1268, "step": 42885 }, { "epoch": 31.418315018315017, "grad_norm": 0.8125, "learning_rate": 3.808655223111236e-06, "loss": 0.1125, "step": 42886 }, { "epoch": 31.419047619047618, "grad_norm": 1.1640625, "learning_rate": 3.8075175893356103e-06, "loss": 0.1105, "step": 42887 }, { "epoch": 31.41978021978022, "grad_norm": 1.03125, "learning_rate": 3.8063801792790045e-06, "loss": 0.1446, "step": 42888 }, { "epoch": 31.42051282051282, "grad_norm": 1.6953125, "learning_rate": 3.8052429929468593e-06, "loss": 0.1088, "step": 42889 }, { "epoch": 31.42124542124542, "grad_norm": 0.921875, "learning_rate": 3.8041060303446202e-06, "loss": 0.1181, "step": 42890 }, { "epoch": 31.42197802197802, "grad_norm": 1.0390625, "learning_rate": 3.8029692914777456e-06, "loss": 0.1051, "step": 42891 }, { "epoch": 31.42271062271062, "grad_norm": 0.89453125, "learning_rate": 3.8018327763516644e-06, "loss": 0.1037, "step": 42892 }, { "epoch": 31.423443223443222, "grad_norm": 1.3828125, "learning_rate": 3.800696484971844e-06, "loss": 0.1314, "step": 42893 }, { "epoch": 31.424175824175823, "grad_norm": 0.89453125, "learning_rate": 3.7995604173437047e-06, "loss": 0.1175, "step": 42894 }, { "epoch": 31.424908424908423, "grad_norm": 1.046875, "learning_rate": 3.798424573472707e-06, "loss": 0.1069, "step": 42895 }, { "epoch": 31.425641025641024, "grad_norm": 1.234375, "learning_rate": 3.7972889533642933e-06, "loss": 0.1464, "step": 42896 }, { "epoch": 31.426373626373625, "grad_norm": 0.65625, "learning_rate": 3.7961535570238988e-06, "loss": 0.1115, "step": 42897 }, { "epoch": 31.427106227106226, "grad_norm": 1.5390625, "learning_rate": 3.7950183844569576e-06, "loss": 0.1205, "step": 42898 }, { "epoch": 31.427838827838826, "grad_norm": 1.171875, "learning_rate": 3.7938834356689163e-06, "loss": 0.1316, "step": 42899 }, { "epoch": 31.428571428571427, "grad_norm": 1.015625, "learning_rate": 3.792748710665221e-06, "loss": 0.1185, "step": 42900 }, { "epoch": 31.429304029304028, "grad_norm": 1.015625, "learning_rate": 3.7916142094512907e-06, "loss": 0.1283, "step": 42901 }, { "epoch": 31.43003663003663, "grad_norm": 0.78125, "learning_rate": 3.790479932032571e-06, "loss": 0.1046, "step": 42902 }, { "epoch": 31.43076923076923, "grad_norm": 0.953125, "learning_rate": 3.7893458784144978e-06, "loss": 0.0931, "step": 42903 }, { "epoch": 31.43150183150183, "grad_norm": 1.0546875, "learning_rate": 3.7882120486025103e-06, "loss": 0.1011, "step": 42904 }, { "epoch": 31.43223443223443, "grad_norm": 1.6484375, "learning_rate": 3.7870784426020232e-06, "loss": 0.1164, "step": 42905 }, { "epoch": 31.43296703296703, "grad_norm": 1.25, "learning_rate": 3.7859450604184872e-06, "loss": 0.13, "step": 42906 }, { "epoch": 31.433699633699632, "grad_norm": 0.78515625, "learning_rate": 3.784811902057306e-06, "loss": 0.1115, "step": 42907 }, { "epoch": 31.434432234432233, "grad_norm": 0.84375, "learning_rate": 3.7836789675239457e-06, "loss": 0.1004, "step": 42908 }, { "epoch": 31.435164835164834, "grad_norm": 1.0078125, "learning_rate": 3.7825462568238034e-06, "loss": 0.1584, "step": 42909 }, { "epoch": 31.435897435897434, "grad_norm": 0.98828125, "learning_rate": 3.7814137699623207e-06, "loss": 0.1165, "step": 42910 }, { "epoch": 31.436630036630035, "grad_norm": 1.046875, "learning_rate": 3.780281506944927e-06, "loss": 0.1393, "step": 42911 }, { "epoch": 31.437362637362636, "grad_norm": 0.9765625, "learning_rate": 3.779149467777035e-06, "loss": 0.1293, "step": 42912 }, { "epoch": 31.438095238095237, "grad_norm": 0.73046875, "learning_rate": 3.77801765246408e-06, "loss": 0.0938, "step": 42913 }, { "epoch": 31.438827838827837, "grad_norm": 0.82421875, "learning_rate": 3.7768860610114755e-06, "loss": 0.0978, "step": 42914 }, { "epoch": 31.439560439560438, "grad_norm": 1.3203125, "learning_rate": 3.775754693424656e-06, "loss": 0.1393, "step": 42915 }, { "epoch": 31.44029304029304, "grad_norm": 1.109375, "learning_rate": 3.774623549709025e-06, "loss": 0.1116, "step": 42916 }, { "epoch": 31.44102564102564, "grad_norm": 1.8359375, "learning_rate": 3.7734926298700212e-06, "loss": 0.1415, "step": 42917 }, { "epoch": 31.44175824175824, "grad_norm": 1.0546875, "learning_rate": 3.7723619339130435e-06, "loss": 0.1456, "step": 42918 }, { "epoch": 31.44249084249084, "grad_norm": 0.91796875, "learning_rate": 3.7712314618435246e-06, "loss": 0.1329, "step": 42919 }, { "epoch": 31.44322344322344, "grad_norm": 1.625, "learning_rate": 3.770101213666874e-06, "loss": 0.1476, "step": 42920 }, { "epoch": 31.443956043956042, "grad_norm": 0.98046875, "learning_rate": 3.7689711893885093e-06, "loss": 0.1391, "step": 42921 }, { "epoch": 31.444688644688643, "grad_norm": 1.328125, "learning_rate": 3.7678413890138505e-06, "loss": 0.1247, "step": 42922 }, { "epoch": 31.445421245421244, "grad_norm": 1.1015625, "learning_rate": 3.7667118125482988e-06, "loss": 0.1181, "step": 42923 }, { "epoch": 31.446153846153845, "grad_norm": 1.6171875, "learning_rate": 3.765582459997273e-06, "loss": 0.1082, "step": 42924 }, { "epoch": 31.446886446886445, "grad_norm": 0.875, "learning_rate": 3.76445333136618e-06, "loss": 0.1176, "step": 42925 }, { "epoch": 31.447619047619046, "grad_norm": 0.93359375, "learning_rate": 3.7633244266604404e-06, "loss": 0.133, "step": 42926 }, { "epoch": 31.448351648351647, "grad_norm": 1.03125, "learning_rate": 3.762195745885455e-06, "loss": 0.111, "step": 42927 }, { "epoch": 31.449084249084247, "grad_norm": 1.0625, "learning_rate": 3.7610672890466255e-06, "loss": 0.1238, "step": 42928 }, { "epoch": 31.449816849816848, "grad_norm": 0.61328125, "learning_rate": 3.7599390561493717e-06, "loss": 0.088, "step": 42929 }, { "epoch": 31.45054945054945, "grad_norm": 1.0390625, "learning_rate": 3.758811047199096e-06, "loss": 0.1326, "step": 42930 }, { "epoch": 31.45128205128205, "grad_norm": 1.875, "learning_rate": 3.757683262201199e-06, "loss": 0.1513, "step": 42931 }, { "epoch": 31.45201465201465, "grad_norm": 1.296875, "learning_rate": 3.7565557011610895e-06, "loss": 0.1354, "step": 42932 }, { "epoch": 31.45274725274725, "grad_norm": 0.78125, "learning_rate": 3.7554283640841473e-06, "loss": 0.1176, "step": 42933 }, { "epoch": 31.453479853479852, "grad_norm": 0.90234375, "learning_rate": 3.754301250975814e-06, "loss": 0.1203, "step": 42934 }, { "epoch": 31.454212454212453, "grad_norm": 1.2421875, "learning_rate": 3.7531743618414633e-06, "loss": 0.1093, "step": 42935 }, { "epoch": 31.454945054945053, "grad_norm": 0.8203125, "learning_rate": 3.7520476966864974e-06, "loss": 0.1019, "step": 42936 }, { "epoch": 31.455677655677654, "grad_norm": 0.84765625, "learning_rate": 3.750921255516324e-06, "loss": 0.1264, "step": 42937 }, { "epoch": 31.456410256410255, "grad_norm": 1.2109375, "learning_rate": 3.7497950383363243e-06, "loss": 0.0928, "step": 42938 }, { "epoch": 31.457142857142856, "grad_norm": 1.1640625, "learning_rate": 3.74866904515191e-06, "loss": 0.1104, "step": 42939 }, { "epoch": 31.457875457875456, "grad_norm": 0.8125, "learning_rate": 3.7475432759684624e-06, "loss": 0.1169, "step": 42940 }, { "epoch": 31.458608058608057, "grad_norm": 1.015625, "learning_rate": 3.7464177307913934e-06, "loss": 0.1013, "step": 42941 }, { "epoch": 31.459340659340658, "grad_norm": 0.828125, "learning_rate": 3.745292409626079e-06, "loss": 0.1281, "step": 42942 }, { "epoch": 31.46007326007326, "grad_norm": 1.046875, "learning_rate": 3.7441673124779266e-06, "loss": 0.1435, "step": 42943 }, { "epoch": 31.46080586080586, "grad_norm": 0.66015625, "learning_rate": 3.7430424393523e-06, "loss": 0.1185, "step": 42944 }, { "epoch": 31.46153846153846, "grad_norm": 1.2109375, "learning_rate": 3.7419177902546233e-06, "loss": 0.1126, "step": 42945 }, { "epoch": 31.46227106227106, "grad_norm": 1.28125, "learning_rate": 3.740793365190254e-06, "loss": 0.1123, "step": 42946 }, { "epoch": 31.46300366300366, "grad_norm": 0.98046875, "learning_rate": 3.7396691641645996e-06, "loss": 0.1114, "step": 42947 }, { "epoch": 31.463736263736262, "grad_norm": 0.84765625, "learning_rate": 3.7385451871830476e-06, "loss": 0.0979, "step": 42948 }, { "epoch": 31.464468864468863, "grad_norm": 0.92578125, "learning_rate": 3.7374214342509657e-06, "loss": 0.0917, "step": 42949 }, { "epoch": 31.465201465201464, "grad_norm": 1.2421875, "learning_rate": 3.7362979053737563e-06, "loss": 0.1123, "step": 42950 }, { "epoch": 31.465934065934064, "grad_norm": 0.8046875, "learning_rate": 3.7351746005567887e-06, "loss": 0.1151, "step": 42951 }, { "epoch": 31.466666666666665, "grad_norm": 0.640625, "learning_rate": 3.73405151980546e-06, "loss": 0.115, "step": 42952 }, { "epoch": 31.467399267399266, "grad_norm": 0.76953125, "learning_rate": 3.7329286631251332e-06, "loss": 0.0953, "step": 42953 }, { "epoch": 31.468131868131866, "grad_norm": 1.0234375, "learning_rate": 3.7318060305212e-06, "loss": 0.1251, "step": 42954 }, { "epoch": 31.468864468864467, "grad_norm": 0.9765625, "learning_rate": 3.7306836219990414e-06, "loss": 0.0944, "step": 42955 }, { "epoch": 31.469597069597068, "grad_norm": 0.71484375, "learning_rate": 3.729561437564035e-06, "loss": 0.1439, "step": 42956 }, { "epoch": 31.47032967032967, "grad_norm": 0.78515625, "learning_rate": 3.7284394772215416e-06, "loss": 0.115, "step": 42957 }, { "epoch": 31.47106227106227, "grad_norm": 0.72265625, "learning_rate": 3.727317740976962e-06, "loss": 0.0902, "step": 42958 }, { "epoch": 31.47179487179487, "grad_norm": 1.2109375, "learning_rate": 3.7261962288356385e-06, "loss": 0.1387, "step": 42959 }, { "epoch": 31.47252747252747, "grad_norm": 1.4921875, "learning_rate": 3.725074940802984e-06, "loss": 0.1294, "step": 42960 }, { "epoch": 31.47326007326007, "grad_norm": 0.6875, "learning_rate": 3.723953876884341e-06, "loss": 0.1103, "step": 42961 }, { "epoch": 31.473992673992672, "grad_norm": 1.078125, "learning_rate": 3.7228330370850883e-06, "loss": 0.1315, "step": 42962 }, { "epoch": 31.474725274725273, "grad_norm": 1.1875, "learning_rate": 3.721712421410607e-06, "loss": 0.123, "step": 42963 }, { "epoch": 31.475457875457874, "grad_norm": 0.91796875, "learning_rate": 3.72059202986625e-06, "loss": 0.1238, "step": 42964 }, { "epoch": 31.476190476190474, "grad_norm": 0.88671875, "learning_rate": 3.7194718624573964e-06, "loss": 0.1227, "step": 42965 }, { "epoch": 31.476923076923075, "grad_norm": 1.3671875, "learning_rate": 3.7183519191894055e-06, "loss": 0.1257, "step": 42966 }, { "epoch": 31.477655677655676, "grad_norm": 1.140625, "learning_rate": 3.717232200067658e-06, "loss": 0.1526, "step": 42967 }, { "epoch": 31.478388278388277, "grad_norm": 0.72265625, "learning_rate": 3.716112705097499e-06, "loss": 0.1126, "step": 42968 }, { "epoch": 31.479120879120877, "grad_norm": 1.140625, "learning_rate": 3.7149934342843105e-06, "loss": 0.105, "step": 42969 }, { "epoch": 31.479853479853478, "grad_norm": 0.9921875, "learning_rate": 3.713874387633434e-06, "loss": 0.0973, "step": 42970 }, { "epoch": 31.48058608058608, "grad_norm": 1.2890625, "learning_rate": 3.7127555651502597e-06, "loss": 0.1417, "step": 42971 }, { "epoch": 31.48131868131868, "grad_norm": 0.734375, "learning_rate": 3.7116369668401256e-06, "loss": 0.101, "step": 42972 }, { "epoch": 31.48205128205128, "grad_norm": 0.68359375, "learning_rate": 3.710518592708394e-06, "loss": 0.0903, "step": 42973 }, { "epoch": 31.48278388278388, "grad_norm": 0.859375, "learning_rate": 3.7094004427604396e-06, "loss": 0.1323, "step": 42974 }, { "epoch": 31.483516483516482, "grad_norm": 0.72265625, "learning_rate": 3.708282517001599e-06, "loss": 0.114, "step": 42975 }, { "epoch": 31.484249084249083, "grad_norm": 1.09375, "learning_rate": 3.707164815437236e-06, "loss": 0.1307, "step": 42976 }, { "epoch": 31.484981684981683, "grad_norm": 1.1015625, "learning_rate": 3.7060473380727086e-06, "loss": 0.1277, "step": 42977 }, { "epoch": 31.485714285714284, "grad_norm": 0.98046875, "learning_rate": 3.7049300849133753e-06, "loss": 0.1338, "step": 42978 }, { "epoch": 31.486446886446885, "grad_norm": 0.98828125, "learning_rate": 3.7038130559645825e-06, "loss": 0.1355, "step": 42979 }, { "epoch": 31.487179487179485, "grad_norm": 0.97265625, "learning_rate": 3.7026962512316785e-06, "loss": 0.1198, "step": 42980 }, { "epoch": 31.487912087912086, "grad_norm": 0.91796875, "learning_rate": 3.701579670720021e-06, "loss": 0.1012, "step": 42981 }, { "epoch": 31.488644688644687, "grad_norm": 0.7734375, "learning_rate": 3.7004633144349634e-06, "loss": 0.0978, "step": 42982 }, { "epoch": 31.489377289377288, "grad_norm": 0.8515625, "learning_rate": 3.6993471823818405e-06, "loss": 0.109, "step": 42983 }, { "epoch": 31.49010989010989, "grad_norm": 0.83203125, "learning_rate": 3.6982312745660177e-06, "loss": 0.1182, "step": 42984 }, { "epoch": 31.49084249084249, "grad_norm": 0.86328125, "learning_rate": 3.6971155909928236e-06, "loss": 0.1053, "step": 42985 }, { "epoch": 31.49157509157509, "grad_norm": 1.078125, "learning_rate": 3.696000131667625e-06, "loss": 0.1256, "step": 42986 }, { "epoch": 31.49230769230769, "grad_norm": 0.83984375, "learning_rate": 3.6948848965957435e-06, "loss": 0.1245, "step": 42987 }, { "epoch": 31.493040293040295, "grad_norm": 1.6015625, "learning_rate": 3.69376988578254e-06, "loss": 0.1489, "step": 42988 }, { "epoch": 31.493772893772892, "grad_norm": 1.40625, "learning_rate": 3.6926550992333496e-06, "loss": 0.135, "step": 42989 }, { "epoch": 31.494505494505496, "grad_norm": 1.15625, "learning_rate": 3.6915405369535145e-06, "loss": 0.1128, "step": 42990 }, { "epoch": 31.495238095238093, "grad_norm": 2.265625, "learning_rate": 3.6904261989483705e-06, "loss": 0.1403, "step": 42991 }, { "epoch": 31.495970695970698, "grad_norm": 0.6171875, "learning_rate": 3.689312085223265e-06, "loss": 0.1269, "step": 42992 }, { "epoch": 31.496703296703295, "grad_norm": 1.0, "learning_rate": 3.6881981957835347e-06, "loss": 0.0993, "step": 42993 }, { "epoch": 31.4974358974359, "grad_norm": 1.1875, "learning_rate": 3.6870845306345097e-06, "loss": 0.1252, "step": 42994 }, { "epoch": 31.498168498168496, "grad_norm": 1.25, "learning_rate": 3.685971089781537e-06, "loss": 0.1407, "step": 42995 }, { "epoch": 31.4989010989011, "grad_norm": 1.6484375, "learning_rate": 3.684857873229932e-06, "loss": 0.1049, "step": 42996 }, { "epoch": 31.499633699633698, "grad_norm": 1.3671875, "learning_rate": 3.6837448809850516e-06, "loss": 0.1038, "step": 42997 }, { "epoch": 31.500366300366302, "grad_norm": 0.91015625, "learning_rate": 3.6826321130522166e-06, "loss": 0.121, "step": 42998 }, { "epoch": 31.5010989010989, "grad_norm": 0.8984375, "learning_rate": 3.6815195694367573e-06, "loss": 0.1005, "step": 42999 }, { "epoch": 31.501831501831504, "grad_norm": 1.1015625, "learning_rate": 3.680407250144015e-06, "loss": 0.1182, "step": 43000 }, { "epoch": 31.5025641025641, "grad_norm": 0.87109375, "learning_rate": 3.679295155179299e-06, "loss": 0.1121, "step": 43001 }, { "epoch": 31.503296703296705, "grad_norm": 1.015625, "learning_rate": 3.678183284547962e-06, "loss": 0.1233, "step": 43002 }, { "epoch": 31.504029304029302, "grad_norm": 0.63671875, "learning_rate": 3.6770716382553124e-06, "loss": 0.1082, "step": 43003 }, { "epoch": 31.504761904761907, "grad_norm": 1.2265625, "learning_rate": 3.675960216306687e-06, "loss": 0.1181, "step": 43004 }, { "epoch": 31.505494505494504, "grad_norm": 0.66015625, "learning_rate": 3.6748490187074047e-06, "loss": 0.1133, "step": 43005 }, { "epoch": 31.506227106227108, "grad_norm": 1.2421875, "learning_rate": 3.673738045462792e-06, "loss": 0.1161, "step": 43006 }, { "epoch": 31.506959706959705, "grad_norm": 0.984375, "learning_rate": 3.6726272965781725e-06, "loss": 0.115, "step": 43007 }, { "epoch": 31.50769230769231, "grad_norm": 1.4765625, "learning_rate": 3.671516772058878e-06, "loss": 0.1211, "step": 43008 }, { "epoch": 31.508424908424907, "grad_norm": 0.9921875, "learning_rate": 3.6704064719102067e-06, "loss": 0.1257, "step": 43009 }, { "epoch": 31.50915750915751, "grad_norm": 1.140625, "learning_rate": 3.6692963961374936e-06, "loss": 0.1103, "step": 43010 }, { "epoch": 31.509890109890108, "grad_norm": 1.1640625, "learning_rate": 3.6681865447460537e-06, "loss": 0.1244, "step": 43011 }, { "epoch": 31.510622710622712, "grad_norm": 0.97265625, "learning_rate": 3.6670769177412174e-06, "loss": 0.1119, "step": 43012 }, { "epoch": 31.51135531135531, "grad_norm": 0.796875, "learning_rate": 3.665967515128278e-06, "loss": 0.1066, "step": 43013 }, { "epoch": 31.512087912087914, "grad_norm": 0.671875, "learning_rate": 3.664858336912565e-06, "loss": 0.1033, "step": 43014 }, { "epoch": 31.51282051282051, "grad_norm": 1.0078125, "learning_rate": 3.663749383099398e-06, "loss": 0.1431, "step": 43015 }, { "epoch": 31.513553113553115, "grad_norm": 0.60546875, "learning_rate": 3.6626406536940756e-06, "loss": 0.1184, "step": 43016 }, { "epoch": 31.514285714285712, "grad_norm": 1.46875, "learning_rate": 3.6615321487019177e-06, "loss": 0.1242, "step": 43017 }, { "epoch": 31.515018315018317, "grad_norm": 0.94140625, "learning_rate": 3.6604238681282322e-06, "loss": 0.1137, "step": 43018 }, { "epoch": 31.515750915750917, "grad_norm": 1.1875, "learning_rate": 3.6593158119783448e-06, "loss": 0.1746, "step": 43019 }, { "epoch": 31.516483516483518, "grad_norm": 0.953125, "learning_rate": 3.658207980257542e-06, "loss": 0.0964, "step": 43020 }, { "epoch": 31.51721611721612, "grad_norm": 1.2734375, "learning_rate": 3.6571003729711493e-06, "loss": 0.1193, "step": 43021 }, { "epoch": 31.51794871794872, "grad_norm": 1.0078125, "learning_rate": 3.6559929901244546e-06, "loss": 0.1111, "step": 43022 }, { "epoch": 31.51868131868132, "grad_norm": 1.453125, "learning_rate": 3.6548858317227862e-06, "loss": 0.1408, "step": 43023 }, { "epoch": 31.51941391941392, "grad_norm": 0.859375, "learning_rate": 3.6537788977714315e-06, "loss": 0.1273, "step": 43024 }, { "epoch": 31.520146520146522, "grad_norm": 0.609375, "learning_rate": 3.6526721882757013e-06, "loss": 0.0927, "step": 43025 }, { "epoch": 31.520879120879123, "grad_norm": 0.765625, "learning_rate": 3.651565703240903e-06, "loss": 0.1071, "step": 43026 }, { "epoch": 31.521611721611723, "grad_norm": 0.96875, "learning_rate": 3.6504594426723223e-06, "loss": 0.1376, "step": 43027 }, { "epoch": 31.522344322344324, "grad_norm": 1.2109375, "learning_rate": 3.64935340657528e-06, "loss": 0.1182, "step": 43028 }, { "epoch": 31.523076923076925, "grad_norm": 0.71875, "learning_rate": 3.6482475949550637e-06, "loss": 0.1258, "step": 43029 }, { "epoch": 31.523809523809526, "grad_norm": 1.1875, "learning_rate": 3.6471420078169754e-06, "loss": 0.1441, "step": 43030 }, { "epoch": 31.524542124542126, "grad_norm": 1.65625, "learning_rate": 3.646036645166308e-06, "loss": 0.1208, "step": 43031 }, { "epoch": 31.525274725274727, "grad_norm": 1.1484375, "learning_rate": 3.644931507008352e-06, "loss": 0.1245, "step": 43032 }, { "epoch": 31.526007326007328, "grad_norm": 0.7578125, "learning_rate": 3.6438265933484193e-06, "loss": 0.112, "step": 43033 }, { "epoch": 31.52673992673993, "grad_norm": 1.078125, "learning_rate": 3.6427219041918e-06, "loss": 0.1111, "step": 43034 }, { "epoch": 31.52747252747253, "grad_norm": 1.0390625, "learning_rate": 3.6416174395437764e-06, "loss": 0.1006, "step": 43035 }, { "epoch": 31.52820512820513, "grad_norm": 1.015625, "learning_rate": 3.6405131994096398e-06, "loss": 0.1001, "step": 43036 }, { "epoch": 31.52893772893773, "grad_norm": 1.078125, "learning_rate": 3.6394091837946997e-06, "loss": 0.1274, "step": 43037 }, { "epoch": 31.52967032967033, "grad_norm": 0.80859375, "learning_rate": 3.6383053927042258e-06, "loss": 0.1107, "step": 43038 }, { "epoch": 31.530402930402932, "grad_norm": 1.5234375, "learning_rate": 3.637201826143516e-06, "loss": 0.1603, "step": 43039 }, { "epoch": 31.531135531135533, "grad_norm": 1.375, "learning_rate": 3.6360984841178572e-06, "loss": 0.1407, "step": 43040 }, { "epoch": 31.531868131868134, "grad_norm": 0.9765625, "learning_rate": 3.634995366632537e-06, "loss": 0.0922, "step": 43041 }, { "epoch": 31.532600732600734, "grad_norm": 1.1953125, "learning_rate": 3.6338924736928354e-06, "loss": 0.12, "step": 43042 }, { "epoch": 31.533333333333335, "grad_norm": 0.85546875, "learning_rate": 3.63278980530404e-06, "loss": 0.0977, "step": 43043 }, { "epoch": 31.534065934065936, "grad_norm": 1.125, "learning_rate": 3.6316873614714315e-06, "loss": 0.1484, "step": 43044 }, { "epoch": 31.534798534798536, "grad_norm": 1.03125, "learning_rate": 3.6305851422003027e-06, "loss": 0.1126, "step": 43045 }, { "epoch": 31.535531135531137, "grad_norm": 0.8359375, "learning_rate": 3.6294831474959226e-06, "loss": 0.1035, "step": 43046 }, { "epoch": 31.536263736263738, "grad_norm": 1.5625, "learning_rate": 3.62838137736358e-06, "loss": 0.1203, "step": 43047 }, { "epoch": 31.53699633699634, "grad_norm": 0.9453125, "learning_rate": 3.627279831808539e-06, "loss": 0.1236, "step": 43048 }, { "epoch": 31.53772893772894, "grad_norm": 0.8203125, "learning_rate": 3.626178510836096e-06, "loss": 0.1269, "step": 43049 }, { "epoch": 31.53846153846154, "grad_norm": 0.890625, "learning_rate": 3.6250774144515107e-06, "loss": 0.1049, "step": 43050 }, { "epoch": 31.53919413919414, "grad_norm": 1.234375, "learning_rate": 3.6239765426600722e-06, "loss": 0.1191, "step": 43051 }, { "epoch": 31.53992673992674, "grad_norm": 0.9609375, "learning_rate": 3.6228758954670595e-06, "loss": 0.1184, "step": 43052 }, { "epoch": 31.540659340659342, "grad_norm": 0.7421875, "learning_rate": 3.621775472877721e-06, "loss": 0.099, "step": 43053 }, { "epoch": 31.541391941391943, "grad_norm": 1.125, "learning_rate": 3.6206752748973554e-06, "loss": 0.1194, "step": 43054 }, { "epoch": 31.542124542124544, "grad_norm": 0.69140625, "learning_rate": 3.619575301531221e-06, "loss": 0.132, "step": 43055 }, { "epoch": 31.542857142857144, "grad_norm": 0.85546875, "learning_rate": 3.6184755527846003e-06, "loss": 0.1098, "step": 43056 }, { "epoch": 31.543589743589745, "grad_norm": 0.984375, "learning_rate": 3.6173760286627403e-06, "loss": 0.0974, "step": 43057 }, { "epoch": 31.544322344322346, "grad_norm": 1.609375, "learning_rate": 3.616276729170928e-06, "loss": 0.1471, "step": 43058 }, { "epoch": 31.545054945054947, "grad_norm": 1.1171875, "learning_rate": 3.615177654314418e-06, "loss": 0.1522, "step": 43059 }, { "epoch": 31.545787545787547, "grad_norm": 0.91796875, "learning_rate": 3.6140788040984973e-06, "loss": 0.1064, "step": 43060 }, { "epoch": 31.546520146520148, "grad_norm": 0.80859375, "learning_rate": 3.612980178528408e-06, "loss": 0.1156, "step": 43061 }, { "epoch": 31.54725274725275, "grad_norm": 1.0546875, "learning_rate": 3.611881777609419e-06, "loss": 0.1452, "step": 43062 }, { "epoch": 31.54798534798535, "grad_norm": 0.7578125, "learning_rate": 3.6107836013468087e-06, "loss": 0.1062, "step": 43063 }, { "epoch": 31.54871794871795, "grad_norm": 1.078125, "learning_rate": 3.6096856497458183e-06, "loss": 0.1247, "step": 43064 }, { "epoch": 31.54945054945055, "grad_norm": 1.0625, "learning_rate": 3.608587922811713e-06, "loss": 0.1441, "step": 43065 }, { "epoch": 31.550183150183152, "grad_norm": 0.87890625, "learning_rate": 3.607490420549763e-06, "loss": 0.1226, "step": 43066 }, { "epoch": 31.550915750915753, "grad_norm": 1.3984375, "learning_rate": 3.6063931429652226e-06, "loss": 0.12, "step": 43067 }, { "epoch": 31.551648351648353, "grad_norm": 1.4296875, "learning_rate": 3.60529609006334e-06, "loss": 0.1228, "step": 43068 }, { "epoch": 31.552380952380954, "grad_norm": 0.890625, "learning_rate": 3.6041992618493787e-06, "loss": 0.1312, "step": 43069 }, { "epoch": 31.553113553113555, "grad_norm": 0.79296875, "learning_rate": 3.603102658328594e-06, "loss": 0.0997, "step": 43070 }, { "epoch": 31.553846153846155, "grad_norm": 1.21875, "learning_rate": 3.6020062795062496e-06, "loss": 0.1476, "step": 43071 }, { "epoch": 31.554578754578756, "grad_norm": 1.015625, "learning_rate": 3.6009101253875785e-06, "loss": 0.1263, "step": 43072 }, { "epoch": 31.555311355311357, "grad_norm": 0.84375, "learning_rate": 3.5998141959778506e-06, "loss": 0.147, "step": 43073 }, { "epoch": 31.556043956043958, "grad_norm": 0.9921875, "learning_rate": 3.5987184912822964e-06, "loss": 0.1312, "step": 43074 }, { "epoch": 31.55677655677656, "grad_norm": 0.91796875, "learning_rate": 3.597623011306188e-06, "loss": 0.1182, "step": 43075 }, { "epoch": 31.55750915750916, "grad_norm": 0.6171875, "learning_rate": 3.596527756054761e-06, "loss": 0.0903, "step": 43076 }, { "epoch": 31.55824175824176, "grad_norm": 1.6640625, "learning_rate": 3.595432725533267e-06, "loss": 0.1081, "step": 43077 }, { "epoch": 31.55897435897436, "grad_norm": 1.7578125, "learning_rate": 3.594337919746961e-06, "loss": 0.1238, "step": 43078 }, { "epoch": 31.55970695970696, "grad_norm": 1.0546875, "learning_rate": 3.593243338701066e-06, "loss": 0.1247, "step": 43079 }, { "epoch": 31.560439560439562, "grad_norm": 0.9453125, "learning_rate": 3.5921489824008523e-06, "loss": 0.0939, "step": 43080 }, { "epoch": 31.561172161172163, "grad_norm": 1.1796875, "learning_rate": 3.591054850851546e-06, "loss": 0.1231, "step": 43081 }, { "epoch": 31.561904761904763, "grad_norm": 1.3046875, "learning_rate": 3.589960944058401e-06, "loss": 0.1171, "step": 43082 }, { "epoch": 31.562637362637364, "grad_norm": 0.68359375, "learning_rate": 3.588867262026649e-06, "loss": 0.1164, "step": 43083 }, { "epoch": 31.563369963369965, "grad_norm": 0.98828125, "learning_rate": 3.587773804761532e-06, "loss": 0.1332, "step": 43084 }, { "epoch": 31.564102564102566, "grad_norm": 0.66796875, "learning_rate": 3.586680572268289e-06, "loss": 0.1148, "step": 43085 }, { "epoch": 31.564835164835166, "grad_norm": 0.78515625, "learning_rate": 3.5855875645521675e-06, "loss": 0.11, "step": 43086 }, { "epoch": 31.565567765567767, "grad_norm": 0.828125, "learning_rate": 3.584494781618393e-06, "loss": 0.134, "step": 43087 }, { "epoch": 31.566300366300368, "grad_norm": 1.0, "learning_rate": 3.5834022234722043e-06, "loss": 0.1149, "step": 43088 }, { "epoch": 31.56703296703297, "grad_norm": 0.65234375, "learning_rate": 3.582309890118843e-06, "loss": 0.0903, "step": 43089 }, { "epoch": 31.56776556776557, "grad_norm": 1.0546875, "learning_rate": 3.5812177815635303e-06, "loss": 0.1336, "step": 43090 }, { "epoch": 31.56849816849817, "grad_norm": 0.8203125, "learning_rate": 3.580125897811504e-06, "loss": 0.1217, "step": 43091 }, { "epoch": 31.56923076923077, "grad_norm": 1.296875, "learning_rate": 3.579034238868001e-06, "loss": 0.1315, "step": 43092 }, { "epoch": 31.56996336996337, "grad_norm": 0.9921875, "learning_rate": 3.577942804738253e-06, "loss": 0.1016, "step": 43093 }, { "epoch": 31.570695970695972, "grad_norm": 0.875, "learning_rate": 3.5768515954274754e-06, "loss": 0.1044, "step": 43094 }, { "epoch": 31.571428571428573, "grad_norm": 0.9453125, "learning_rate": 3.5757606109409106e-06, "loss": 0.095, "step": 43095 }, { "epoch": 31.572161172161174, "grad_norm": 1.7578125, "learning_rate": 3.5746698512837754e-06, "loss": 0.1163, "step": 43096 }, { "epoch": 31.572893772893774, "grad_norm": 1.09375, "learning_rate": 3.5735793164613106e-06, "loss": 0.1177, "step": 43097 }, { "epoch": 31.573626373626375, "grad_norm": 1.234375, "learning_rate": 3.572489006478727e-06, "loss": 0.1045, "step": 43098 }, { "epoch": 31.574358974358976, "grad_norm": 0.7734375, "learning_rate": 3.5713989213412624e-06, "loss": 0.1121, "step": 43099 }, { "epoch": 31.575091575091577, "grad_norm": 1.21875, "learning_rate": 3.57030906105411e-06, "loss": 0.1106, "step": 43100 }, { "epoch": 31.575824175824177, "grad_norm": 1.2109375, "learning_rate": 3.569219425622533e-06, "loss": 0.156, "step": 43101 }, { "epoch": 31.576556776556778, "grad_norm": 0.58984375, "learning_rate": 3.5681300150517274e-06, "loss": 0.1059, "step": 43102 }, { "epoch": 31.57728937728938, "grad_norm": 0.91015625, "learning_rate": 3.5670408293469118e-06, "loss": 0.1285, "step": 43103 }, { "epoch": 31.57802197802198, "grad_norm": 1.140625, "learning_rate": 3.565951868513319e-06, "loss": 0.1268, "step": 43104 }, { "epoch": 31.57875457875458, "grad_norm": 1.2734375, "learning_rate": 3.564863132556143e-06, "loss": 0.1305, "step": 43105 }, { "epoch": 31.57948717948718, "grad_norm": 1.0703125, "learning_rate": 3.563774621480635e-06, "loss": 0.0949, "step": 43106 }, { "epoch": 31.58021978021978, "grad_norm": 1.359375, "learning_rate": 3.56268633529198e-06, "loss": 0.1371, "step": 43107 }, { "epoch": 31.580952380952382, "grad_norm": 0.765625, "learning_rate": 3.5615982739954097e-06, "loss": 0.0977, "step": 43108 }, { "epoch": 31.581684981684983, "grad_norm": 1.1015625, "learning_rate": 3.5605104375961275e-06, "loss": 0.1239, "step": 43109 }, { "epoch": 31.582417582417584, "grad_norm": 0.875, "learning_rate": 3.5594228260993427e-06, "loss": 0.1189, "step": 43110 }, { "epoch": 31.583150183150185, "grad_norm": 0.921875, "learning_rate": 3.558335439510277e-06, "loss": 0.1234, "step": 43111 }, { "epoch": 31.583882783882785, "grad_norm": 0.75390625, "learning_rate": 3.5572482778341463e-06, "loss": 0.1034, "step": 43112 }, { "epoch": 31.584615384615386, "grad_norm": 1.5078125, "learning_rate": 3.556161341076137e-06, "loss": 0.1321, "step": 43113 }, { "epoch": 31.585347985347987, "grad_norm": 1.171875, "learning_rate": 3.555074629241471e-06, "loss": 0.1187, "step": 43114 }, { "epoch": 31.586080586080588, "grad_norm": 1.0703125, "learning_rate": 3.553988142335357e-06, "loss": 0.0899, "step": 43115 }, { "epoch": 31.58681318681319, "grad_norm": 1.5234375, "learning_rate": 3.5529018803629904e-06, "loss": 0.1572, "step": 43116 }, { "epoch": 31.58754578754579, "grad_norm": 1.140625, "learning_rate": 3.5518158433295894e-06, "loss": 0.1222, "step": 43117 }, { "epoch": 31.58827838827839, "grad_norm": 0.7890625, "learning_rate": 3.5507300312403498e-06, "loss": 0.1166, "step": 43118 }, { "epoch": 31.58901098901099, "grad_norm": 1.2109375, "learning_rate": 3.5496444441004807e-06, "loss": 0.1284, "step": 43119 }, { "epoch": 31.58974358974359, "grad_norm": 0.55078125, "learning_rate": 3.548559081915168e-06, "loss": 0.1062, "step": 43120 }, { "epoch": 31.590476190476192, "grad_norm": 1.0703125, "learning_rate": 3.5474739446896244e-06, "loss": 0.1117, "step": 43121 }, { "epoch": 31.591208791208793, "grad_norm": 0.76953125, "learning_rate": 3.546389032429047e-06, "loss": 0.104, "step": 43122 }, { "epoch": 31.591941391941393, "grad_norm": 1.0703125, "learning_rate": 3.5453043451386404e-06, "loss": 0.1297, "step": 43123 }, { "epoch": 31.592673992673994, "grad_norm": 1.03125, "learning_rate": 3.5442198828235877e-06, "loss": 0.1456, "step": 43124 }, { "epoch": 31.593406593406595, "grad_norm": 0.78515625, "learning_rate": 3.5431356454890866e-06, "loss": 0.128, "step": 43125 }, { "epoch": 31.594139194139196, "grad_norm": 0.93359375, "learning_rate": 3.542051633140342e-06, "loss": 0.1199, "step": 43126 }, { "epoch": 31.594871794871796, "grad_norm": 1.1328125, "learning_rate": 3.5409678457825528e-06, "loss": 0.1147, "step": 43127 }, { "epoch": 31.595604395604397, "grad_norm": 0.86328125, "learning_rate": 3.5398842834208966e-06, "loss": 0.1284, "step": 43128 }, { "epoch": 31.596336996336998, "grad_norm": 0.8515625, "learning_rate": 3.538800946060565e-06, "loss": 0.1197, "step": 43129 }, { "epoch": 31.5970695970696, "grad_norm": 1.0703125, "learning_rate": 3.537717833706764e-06, "loss": 0.0962, "step": 43130 }, { "epoch": 31.5978021978022, "grad_norm": 0.9140625, "learning_rate": 3.5366349463646588e-06, "loss": 0.1143, "step": 43131 }, { "epoch": 31.5985347985348, "grad_norm": 1.296875, "learning_rate": 3.53555228403947e-06, "loss": 0.1111, "step": 43132 }, { "epoch": 31.5992673992674, "grad_norm": 1.0234375, "learning_rate": 3.5344698467363574e-06, "loss": 0.1158, "step": 43133 }, { "epoch": 31.6, "grad_norm": 1.0, "learning_rate": 3.5333876344605222e-06, "loss": 0.1051, "step": 43134 }, { "epoch": 31.600732600732602, "grad_norm": 1.0625, "learning_rate": 3.5323056472171436e-06, "loss": 0.1172, "step": 43135 }, { "epoch": 31.601465201465203, "grad_norm": 0.6328125, "learning_rate": 3.5312238850114005e-06, "loss": 0.1177, "step": 43136 }, { "epoch": 31.602197802197804, "grad_norm": 1.015625, "learning_rate": 3.5301423478484863e-06, "loss": 0.0994, "step": 43137 }, { "epoch": 31.602930402930404, "grad_norm": 0.77734375, "learning_rate": 3.5290610357335886e-06, "loss": 0.1068, "step": 43138 }, { "epoch": 31.603663003663005, "grad_norm": 0.99609375, "learning_rate": 3.5279799486718725e-06, "loss": 0.1221, "step": 43139 }, { "epoch": 31.604395604395606, "grad_norm": 1.3203125, "learning_rate": 3.52689908666852e-06, "loss": 0.1109, "step": 43140 }, { "epoch": 31.605128205128207, "grad_norm": 0.72265625, "learning_rate": 3.5258184497287203e-06, "loss": 0.0916, "step": 43141 }, { "epoch": 31.605860805860807, "grad_norm": 2.015625, "learning_rate": 3.5247380378576333e-06, "loss": 0.1271, "step": 43142 }, { "epoch": 31.606593406593408, "grad_norm": 0.734375, "learning_rate": 3.5236578510604633e-06, "loss": 0.0934, "step": 43143 }, { "epoch": 31.60732600732601, "grad_norm": 0.99609375, "learning_rate": 3.52257788934236e-06, "loss": 0.1077, "step": 43144 }, { "epoch": 31.60805860805861, "grad_norm": 1.2578125, "learning_rate": 3.521498152708517e-06, "loss": 0.1728, "step": 43145 }, { "epoch": 31.60879120879121, "grad_norm": 1.1875, "learning_rate": 3.5204186411640876e-06, "loss": 0.1239, "step": 43146 }, { "epoch": 31.60952380952381, "grad_norm": 1.078125, "learning_rate": 3.5193393547142615e-06, "loss": 0.1023, "step": 43147 }, { "epoch": 31.61025641025641, "grad_norm": 1.140625, "learning_rate": 3.518260293364197e-06, "loss": 0.1373, "step": 43148 }, { "epoch": 31.610989010989012, "grad_norm": 0.92578125, "learning_rate": 3.517181457119079e-06, "loss": 0.1069, "step": 43149 }, { "epoch": 31.611721611721613, "grad_norm": 0.98828125, "learning_rate": 3.5161028459840657e-06, "loss": 0.1121, "step": 43150 }, { "epoch": 31.612454212454214, "grad_norm": 0.7578125, "learning_rate": 3.515024459964324e-06, "loss": 0.1086, "step": 43151 }, { "epoch": 31.613186813186815, "grad_norm": 0.6640625, "learning_rate": 3.5139462990650253e-06, "loss": 0.1209, "step": 43152 }, { "epoch": 31.613919413919415, "grad_norm": 1.15625, "learning_rate": 3.512868363291335e-06, "loss": 0.1286, "step": 43153 }, { "epoch": 31.614652014652016, "grad_norm": 1.25, "learning_rate": 3.5117906526484193e-06, "loss": 0.1103, "step": 43154 }, { "epoch": 31.615384615384617, "grad_norm": 0.890625, "learning_rate": 3.510713167141433e-06, "loss": 0.1145, "step": 43155 }, { "epoch": 31.616117216117217, "grad_norm": 0.98046875, "learning_rate": 3.509635906775559e-06, "loss": 0.1085, "step": 43156 }, { "epoch": 31.616849816849818, "grad_norm": 1.984375, "learning_rate": 3.5085588715559227e-06, "loss": 0.1317, "step": 43157 }, { "epoch": 31.61758241758242, "grad_norm": 1.2265625, "learning_rate": 3.507482061487731e-06, "loss": 0.1357, "step": 43158 }, { "epoch": 31.61831501831502, "grad_norm": 0.95703125, "learning_rate": 3.506405476576105e-06, "loss": 0.1275, "step": 43159 }, { "epoch": 31.61904761904762, "grad_norm": 0.8984375, "learning_rate": 3.505329116826226e-06, "loss": 0.105, "step": 43160 }, { "epoch": 31.61978021978022, "grad_norm": 1.4140625, "learning_rate": 3.504252982243239e-06, "loss": 0.1383, "step": 43161 }, { "epoch": 31.620512820512822, "grad_norm": 0.953125, "learning_rate": 3.5031770728322984e-06, "loss": 0.0917, "step": 43162 }, { "epoch": 31.621245421245423, "grad_norm": 1.9375, "learning_rate": 3.50210138859857e-06, "loss": 0.0985, "step": 43163 }, { "epoch": 31.621978021978023, "grad_norm": 0.8828125, "learning_rate": 3.501025929547204e-06, "loss": 0.1117, "step": 43164 }, { "epoch": 31.622710622710624, "grad_norm": 1.109375, "learning_rate": 3.499950695683348e-06, "loss": 0.1198, "step": 43165 }, { "epoch": 31.623443223443225, "grad_norm": 1.3515625, "learning_rate": 3.4988756870121533e-06, "loss": 0.1606, "step": 43166 }, { "epoch": 31.624175824175826, "grad_norm": 1.078125, "learning_rate": 3.497800903538785e-06, "loss": 0.109, "step": 43167 }, { "epoch": 31.624908424908426, "grad_norm": 1.640625, "learning_rate": 3.49672634526837e-06, "loss": 0.1455, "step": 43168 }, { "epoch": 31.625641025641027, "grad_norm": 1.0078125, "learning_rate": 3.4956520122060812e-06, "loss": 0.1197, "step": 43169 }, { "epoch": 31.626373626373628, "grad_norm": 0.80859375, "learning_rate": 3.4945779043570492e-06, "loss": 0.1061, "step": 43170 }, { "epoch": 31.62710622710623, "grad_norm": 0.87109375, "learning_rate": 3.4935040217264305e-06, "loss": 0.1228, "step": 43171 }, { "epoch": 31.62783882783883, "grad_norm": 1.0703125, "learning_rate": 3.492430364319358e-06, "loss": 0.1088, "step": 43172 }, { "epoch": 31.62857142857143, "grad_norm": 0.875, "learning_rate": 3.4913569321409808e-06, "loss": 0.135, "step": 43173 }, { "epoch": 31.62930402930403, "grad_norm": 0.8203125, "learning_rate": 3.4902837251964477e-06, "loss": 0.1129, "step": 43174 }, { "epoch": 31.63003663003663, "grad_norm": 1.0078125, "learning_rate": 3.489210743490903e-06, "loss": 0.141, "step": 43175 }, { "epoch": 31.630769230769232, "grad_norm": 0.9609375, "learning_rate": 3.4881379870294746e-06, "loss": 0.1259, "step": 43176 }, { "epoch": 31.631501831501833, "grad_norm": 0.96484375, "learning_rate": 3.4870654558173108e-06, "loss": 0.1299, "step": 43177 }, { "epoch": 31.632234432234434, "grad_norm": 0.71484375, "learning_rate": 3.485993149859551e-06, "loss": 0.1122, "step": 43178 }, { "epoch": 31.632967032967034, "grad_norm": 1.171875, "learning_rate": 3.484921069161339e-06, "loss": 0.1076, "step": 43179 }, { "epoch": 31.633699633699635, "grad_norm": 1.109375, "learning_rate": 3.4838492137277967e-06, "loss": 0.1118, "step": 43180 }, { "epoch": 31.634432234432236, "grad_norm": 1.0625, "learning_rate": 3.482777583564067e-06, "loss": 0.1134, "step": 43181 }, { "epoch": 31.635164835164836, "grad_norm": 1.359375, "learning_rate": 3.4817061786752947e-06, "loss": 0.0925, "step": 43182 }, { "epoch": 31.635897435897437, "grad_norm": 1.03125, "learning_rate": 3.480634999066584e-06, "loss": 0.1139, "step": 43183 }, { "epoch": 31.636630036630038, "grad_norm": 1.125, "learning_rate": 3.4795640447431034e-06, "loss": 0.1073, "step": 43184 }, { "epoch": 31.63736263736264, "grad_norm": 1.21875, "learning_rate": 3.478493315709956e-06, "loss": 0.0841, "step": 43185 }, { "epoch": 31.63809523809524, "grad_norm": 1.0078125, "learning_rate": 3.4774228119722975e-06, "loss": 0.0907, "step": 43186 }, { "epoch": 31.63882783882784, "grad_norm": 0.890625, "learning_rate": 3.4763525335352278e-06, "loss": 0.1131, "step": 43187 }, { "epoch": 31.63956043956044, "grad_norm": 1.5078125, "learning_rate": 3.4752824804038958e-06, "loss": 0.1341, "step": 43188 }, { "epoch": 31.64029304029304, "grad_norm": 1.1171875, "learning_rate": 3.4742126525834184e-06, "loss": 0.0995, "step": 43189 }, { "epoch": 31.641025641025642, "grad_norm": 0.96875, "learning_rate": 3.473143050078939e-06, "loss": 0.1115, "step": 43190 }, { "epoch": 31.641758241758243, "grad_norm": 0.8359375, "learning_rate": 3.4720736728955586e-06, "loss": 0.1148, "step": 43191 }, { "epoch": 31.642490842490844, "grad_norm": 0.80859375, "learning_rate": 3.4710045210384096e-06, "loss": 0.1098, "step": 43192 }, { "epoch": 31.643223443223444, "grad_norm": 1.265625, "learning_rate": 3.4699355945126237e-06, "loss": 0.1201, "step": 43193 }, { "epoch": 31.643956043956045, "grad_norm": 2.46875, "learning_rate": 3.4688668933233028e-06, "loss": 0.1354, "step": 43194 }, { "epoch": 31.644688644688646, "grad_norm": 0.7421875, "learning_rate": 3.4677984174755943e-06, "loss": 0.1165, "step": 43195 }, { "epoch": 31.645421245421247, "grad_norm": 0.796875, "learning_rate": 3.4667301669745992e-06, "loss": 0.1279, "step": 43196 }, { "epoch": 31.646153846153847, "grad_norm": 1.671875, "learning_rate": 3.465662141825439e-06, "loss": 0.1454, "step": 43197 }, { "epoch": 31.646886446886448, "grad_norm": 1.1171875, "learning_rate": 3.464594342033231e-06, "loss": 0.0875, "step": 43198 }, { "epoch": 31.64761904761905, "grad_norm": 0.9453125, "learning_rate": 3.463526767603091e-06, "loss": 0.133, "step": 43199 }, { "epoch": 31.64835164835165, "grad_norm": 1.1953125, "learning_rate": 3.462459418540136e-06, "loss": 0.0984, "step": 43200 }, { "epoch": 31.64908424908425, "grad_norm": 0.89453125, "learning_rate": 3.4613922948494858e-06, "loss": 0.1141, "step": 43201 }, { "epoch": 31.64981684981685, "grad_norm": 0.66015625, "learning_rate": 3.4603253965362374e-06, "loss": 0.1072, "step": 43202 }, { "epoch": 31.650549450549452, "grad_norm": 1.375, "learning_rate": 3.459258723605517e-06, "loss": 0.1188, "step": 43203 }, { "epoch": 31.651282051282053, "grad_norm": 0.86328125, "learning_rate": 3.4581922760624248e-06, "loss": 0.1352, "step": 43204 }, { "epoch": 31.652014652014653, "grad_norm": 0.875, "learning_rate": 3.457126053912088e-06, "loss": 0.128, "step": 43205 }, { "epoch": 31.652747252747254, "grad_norm": 0.94921875, "learning_rate": 3.4560600571595955e-06, "loss": 0.1154, "step": 43206 }, { "epoch": 31.653479853479855, "grad_norm": 0.9765625, "learning_rate": 3.4549942858100577e-06, "loss": 0.1204, "step": 43207 }, { "epoch": 31.654212454212455, "grad_norm": 1.046875, "learning_rate": 3.4539287398685982e-06, "loss": 0.1319, "step": 43208 }, { "epoch": 31.654945054945056, "grad_norm": 0.7578125, "learning_rate": 3.4528634193402993e-06, "loss": 0.1224, "step": 43209 }, { "epoch": 31.655677655677657, "grad_norm": 0.953125, "learning_rate": 3.451798324230282e-06, "loss": 0.107, "step": 43210 }, { "epoch": 31.656410256410258, "grad_norm": 0.7109375, "learning_rate": 3.450733454543642e-06, "loss": 0.1263, "step": 43211 }, { "epoch": 31.65714285714286, "grad_norm": 1.3046875, "learning_rate": 3.4496688102854903e-06, "loss": 0.1546, "step": 43212 }, { "epoch": 31.65787545787546, "grad_norm": 0.91015625, "learning_rate": 3.448604391460915e-06, "loss": 0.1032, "step": 43213 }, { "epoch": 31.65860805860806, "grad_norm": 1.3671875, "learning_rate": 3.4475401980750166e-06, "loss": 0.1121, "step": 43214 }, { "epoch": 31.65934065934066, "grad_norm": 0.9453125, "learning_rate": 3.4464762301329057e-06, "loss": 0.1057, "step": 43215 }, { "epoch": 31.66007326007326, "grad_norm": 0.80078125, "learning_rate": 3.4454124876396765e-06, "loss": 0.1007, "step": 43216 }, { "epoch": 31.660805860805862, "grad_norm": 1.5234375, "learning_rate": 3.4443489706004195e-06, "loss": 0.1302, "step": 43217 }, { "epoch": 31.661538461538463, "grad_norm": 0.765625, "learning_rate": 3.4432856790202318e-06, "loss": 0.1109, "step": 43218 }, { "epoch": 31.662271062271063, "grad_norm": 0.88671875, "learning_rate": 3.4422226129042167e-06, "loss": 0.11, "step": 43219 }, { "epoch": 31.663003663003664, "grad_norm": 1.09375, "learning_rate": 3.441159772257449e-06, "loss": 0.1408, "step": 43220 }, { "epoch": 31.663736263736265, "grad_norm": 0.9296875, "learning_rate": 3.440097157085047e-06, "loss": 0.1288, "step": 43221 }, { "epoch": 31.664468864468866, "grad_norm": 0.89453125, "learning_rate": 3.4390347673920768e-06, "loss": 0.1072, "step": 43222 }, { "epoch": 31.665201465201466, "grad_norm": 1.046875, "learning_rate": 3.4379726031836556e-06, "loss": 0.1216, "step": 43223 }, { "epoch": 31.665934065934067, "grad_norm": 1.265625, "learning_rate": 3.4369106644648443e-06, "loss": 0.1654, "step": 43224 }, { "epoch": 31.666666666666668, "grad_norm": 0.73046875, "learning_rate": 3.435848951240743e-06, "loss": 0.1189, "step": 43225 }, { "epoch": 31.66739926739927, "grad_norm": 1.21875, "learning_rate": 3.4347874635164395e-06, "loss": 0.105, "step": 43226 }, { "epoch": 31.66813186813187, "grad_norm": 1.375, "learning_rate": 3.4337262012970302e-06, "loss": 0.128, "step": 43227 }, { "epoch": 31.66886446886447, "grad_norm": 0.76171875, "learning_rate": 3.4326651645875813e-06, "loss": 0.1041, "step": 43228 }, { "epoch": 31.66959706959707, "grad_norm": 1.015625, "learning_rate": 3.4316043533931862e-06, "loss": 0.1361, "step": 43229 }, { "epoch": 31.67032967032967, "grad_norm": 0.953125, "learning_rate": 3.4305437677189244e-06, "loss": 0.1418, "step": 43230 }, { "epoch": 31.671062271062272, "grad_norm": 1.34375, "learning_rate": 3.4294834075698787e-06, "loss": 0.128, "step": 43231 }, { "epoch": 31.671794871794873, "grad_norm": 0.70703125, "learning_rate": 3.4284232729511373e-06, "loss": 0.0939, "step": 43232 }, { "epoch": 31.672527472527474, "grad_norm": 0.828125, "learning_rate": 3.427363363867768e-06, "loss": 0.107, "step": 43233 }, { "epoch": 31.673260073260074, "grad_norm": 0.953125, "learning_rate": 3.4263036803248544e-06, "loss": 0.0831, "step": 43234 }, { "epoch": 31.673992673992675, "grad_norm": 1.2109375, "learning_rate": 3.425244222327469e-06, "loss": 0.1305, "step": 43235 }, { "epoch": 31.674725274725276, "grad_norm": 1.2265625, "learning_rate": 3.4241849898806985e-06, "loss": 0.1543, "step": 43236 }, { "epoch": 31.675457875457877, "grad_norm": 0.67578125, "learning_rate": 3.423125982989607e-06, "loss": 0.1254, "step": 43237 }, { "epoch": 31.676190476190477, "grad_norm": 0.734375, "learning_rate": 3.4220672016592777e-06, "loss": 0.1141, "step": 43238 }, { "epoch": 31.676923076923078, "grad_norm": 1.3046875, "learning_rate": 3.4210086458947762e-06, "loss": 0.1117, "step": 43239 }, { "epoch": 31.67765567765568, "grad_norm": 1.0234375, "learning_rate": 3.4199503157011757e-06, "loss": 0.1093, "step": 43240 }, { "epoch": 31.67838827838828, "grad_norm": 1.2890625, "learning_rate": 3.418892211083549e-06, "loss": 0.1295, "step": 43241 }, { "epoch": 31.67912087912088, "grad_norm": 1.421875, "learning_rate": 3.4178343320469736e-06, "loss": 0.1206, "step": 43242 }, { "epoch": 31.67985347985348, "grad_norm": 1.4140625, "learning_rate": 3.4167766785965e-06, "loss": 0.1401, "step": 43243 }, { "epoch": 31.68058608058608, "grad_norm": 1.140625, "learning_rate": 3.4157192507372057e-06, "loss": 0.1313, "step": 43244 }, { "epoch": 31.681318681318682, "grad_norm": 1.4140625, "learning_rate": 3.414662048474164e-06, "loss": 0.1127, "step": 43245 }, { "epoch": 31.682051282051283, "grad_norm": 1.0625, "learning_rate": 3.413605071812424e-06, "loss": 0.1177, "step": 43246 }, { "epoch": 31.682783882783884, "grad_norm": 0.91015625, "learning_rate": 3.412548320757071e-06, "loss": 0.1031, "step": 43247 }, { "epoch": 31.683516483516485, "grad_norm": 1.3203125, "learning_rate": 3.411491795313149e-06, "loss": 0.17, "step": 43248 }, { "epoch": 31.684249084249085, "grad_norm": 0.7265625, "learning_rate": 3.4104354954857367e-06, "loss": 0.1311, "step": 43249 }, { "epoch": 31.684981684981686, "grad_norm": 1.171875, "learning_rate": 3.409379421279877e-06, "loss": 0.1442, "step": 43250 }, { "epoch": 31.685714285714287, "grad_norm": 1.2578125, "learning_rate": 3.408323572700639e-06, "loss": 0.1209, "step": 43251 }, { "epoch": 31.686446886446888, "grad_norm": 1.046875, "learning_rate": 3.407267949753083e-06, "loss": 0.147, "step": 43252 }, { "epoch": 31.68717948717949, "grad_norm": 0.9296875, "learning_rate": 3.4062125524422765e-06, "loss": 0.1298, "step": 43253 }, { "epoch": 31.68791208791209, "grad_norm": 0.6953125, "learning_rate": 3.405157380773252e-06, "loss": 0.1319, "step": 43254 }, { "epoch": 31.68864468864469, "grad_norm": 1.15625, "learning_rate": 3.404102434751085e-06, "loss": 0.1201, "step": 43255 }, { "epoch": 31.68937728937729, "grad_norm": 0.99609375, "learning_rate": 3.403047714380818e-06, "loss": 0.1088, "step": 43256 }, { "epoch": 31.69010989010989, "grad_norm": 0.79296875, "learning_rate": 3.401993219667513e-06, "loss": 0.1199, "step": 43257 }, { "epoch": 31.690842490842492, "grad_norm": 1.0546875, "learning_rate": 3.4009389506162317e-06, "loss": 0.1427, "step": 43258 }, { "epoch": 31.691575091575093, "grad_norm": 0.8046875, "learning_rate": 3.399884907231997e-06, "loss": 0.109, "step": 43259 }, { "epoch": 31.692307692307693, "grad_norm": 0.96484375, "learning_rate": 3.3988310895198923e-06, "loss": 0.1302, "step": 43260 }, { "epoch": 31.693040293040294, "grad_norm": 1.15625, "learning_rate": 3.397777497484929e-06, "loss": 0.114, "step": 43261 }, { "epoch": 31.693772893772895, "grad_norm": 1.3046875, "learning_rate": 3.3967241311321918e-06, "loss": 0.1267, "step": 43262 }, { "epoch": 31.694505494505496, "grad_norm": 0.765625, "learning_rate": 3.395670990466708e-06, "loss": 0.1032, "step": 43263 }, { "epoch": 31.695238095238096, "grad_norm": 1.484375, "learning_rate": 3.394618075493529e-06, "loss": 0.1323, "step": 43264 }, { "epoch": 31.695970695970697, "grad_norm": 1.40625, "learning_rate": 3.3935653862176993e-06, "loss": 0.1356, "step": 43265 }, { "epoch": 31.696703296703298, "grad_norm": 0.796875, "learning_rate": 3.392512922644258e-06, "loss": 0.129, "step": 43266 }, { "epoch": 31.6974358974359, "grad_norm": 1.34375, "learning_rate": 3.3914606847782506e-06, "loss": 0.1443, "step": 43267 }, { "epoch": 31.6981684981685, "grad_norm": 0.66796875, "learning_rate": 3.3904086726247273e-06, "loss": 0.1172, "step": 43268 }, { "epoch": 31.6989010989011, "grad_norm": 0.8203125, "learning_rate": 3.3893568861887174e-06, "loss": 0.1157, "step": 43269 }, { "epoch": 31.6996336996337, "grad_norm": 1.1171875, "learning_rate": 3.3883053254752583e-06, "loss": 0.1278, "step": 43270 }, { "epoch": 31.7003663003663, "grad_norm": 0.9140625, "learning_rate": 3.387253990489403e-06, "loss": 0.09, "step": 43271 }, { "epoch": 31.701098901098902, "grad_norm": 0.58984375, "learning_rate": 3.386202881236167e-06, "loss": 0.1202, "step": 43272 }, { "epoch": 31.701831501831503, "grad_norm": 1.0703125, "learning_rate": 3.3851519977206133e-06, "loss": 0.1353, "step": 43273 }, { "epoch": 31.702564102564104, "grad_norm": 0.95703125, "learning_rate": 3.384101339947758e-06, "loss": 0.113, "step": 43274 }, { "epoch": 31.703296703296704, "grad_norm": 0.98046875, "learning_rate": 3.383050907922647e-06, "loss": 0.1172, "step": 43275 }, { "epoch": 31.704029304029305, "grad_norm": 0.94140625, "learning_rate": 3.382000701650297e-06, "loss": 0.1222, "step": 43276 }, { "epoch": 31.704761904761906, "grad_norm": 0.96484375, "learning_rate": 3.380950721135749e-06, "loss": 0.1116, "step": 43277 }, { "epoch": 31.705494505494507, "grad_norm": 0.96875, "learning_rate": 3.3799009663840404e-06, "loss": 0.1301, "step": 43278 }, { "epoch": 31.706227106227107, "grad_norm": 1.09375, "learning_rate": 3.3788514374001963e-06, "loss": 0.1181, "step": 43279 }, { "epoch": 31.706959706959708, "grad_norm": 1.28125, "learning_rate": 3.377802134189237e-06, "loss": 0.1192, "step": 43280 }, { "epoch": 31.70769230769231, "grad_norm": 0.80859375, "learning_rate": 3.376753056756199e-06, "loss": 0.1055, "step": 43281 }, { "epoch": 31.70842490842491, "grad_norm": 1.03125, "learning_rate": 3.375704205106105e-06, "loss": 0.1156, "step": 43282 }, { "epoch": 31.70915750915751, "grad_norm": 0.8515625, "learning_rate": 3.3746555792439815e-06, "loss": 0.1162, "step": 43283 }, { "epoch": 31.70989010989011, "grad_norm": 0.57421875, "learning_rate": 3.373607179174859e-06, "loss": 0.1242, "step": 43284 }, { "epoch": 31.71062271062271, "grad_norm": 1.28125, "learning_rate": 3.3725590049037537e-06, "loss": 0.1103, "step": 43285 }, { "epoch": 31.711355311355312, "grad_norm": 1.15625, "learning_rate": 3.37151105643569e-06, "loss": 0.1285, "step": 43286 }, { "epoch": 31.712087912087913, "grad_norm": 0.99609375, "learning_rate": 3.3704633337756785e-06, "loss": 0.1299, "step": 43287 }, { "epoch": 31.712820512820514, "grad_norm": 0.86328125, "learning_rate": 3.3694158369287595e-06, "loss": 0.113, "step": 43288 }, { "epoch": 31.713553113553115, "grad_norm": 0.796875, "learning_rate": 3.368368565899939e-06, "loss": 0.1153, "step": 43289 }, { "epoch": 31.714285714285715, "grad_norm": 0.90234375, "learning_rate": 3.367321520694241e-06, "loss": 0.1151, "step": 43290 }, { "epoch": 31.715018315018316, "grad_norm": 1.265625, "learning_rate": 3.36627470131667e-06, "loss": 0.1387, "step": 43291 }, { "epoch": 31.715750915750917, "grad_norm": 0.8828125, "learning_rate": 3.365228107772251e-06, "loss": 0.1295, "step": 43292 }, { "epoch": 31.716483516483517, "grad_norm": 1.3828125, "learning_rate": 3.3641817400660005e-06, "loss": 0.1123, "step": 43293 }, { "epoch": 31.717216117216118, "grad_norm": 1.265625, "learning_rate": 3.3631355982029307e-06, "loss": 0.1197, "step": 43294 }, { "epoch": 31.71794871794872, "grad_norm": 1.0078125, "learning_rate": 3.362089682188048e-06, "loss": 0.1057, "step": 43295 }, { "epoch": 31.71868131868132, "grad_norm": 1.0546875, "learning_rate": 3.3610439920263694e-06, "loss": 0.102, "step": 43296 }, { "epoch": 31.71941391941392, "grad_norm": 0.76953125, "learning_rate": 3.359998527722908e-06, "loss": 0.1017, "step": 43297 }, { "epoch": 31.72014652014652, "grad_norm": 0.9375, "learning_rate": 3.358953289282653e-06, "loss": 0.1347, "step": 43298 }, { "epoch": 31.720879120879122, "grad_norm": 0.73828125, "learning_rate": 3.3579082767106446e-06, "loss": 0.1081, "step": 43299 }, { "epoch": 31.721611721611723, "grad_norm": 0.6953125, "learning_rate": 3.356863490011866e-06, "loss": 0.1211, "step": 43300 }, { "epoch": 31.722344322344323, "grad_norm": 1.2109375, "learning_rate": 3.3558189291913414e-06, "loss": 0.1356, "step": 43301 }, { "epoch": 31.723076923076924, "grad_norm": 0.984375, "learning_rate": 3.3547745942540553e-06, "loss": 0.122, "step": 43302 }, { "epoch": 31.723809523809525, "grad_norm": 1.046875, "learning_rate": 3.3537304852050185e-06, "loss": 0.1351, "step": 43303 }, { "epoch": 31.724542124542126, "grad_norm": 0.5625, "learning_rate": 3.352686602049239e-06, "loss": 0.1026, "step": 43304 }, { "epoch": 31.725274725274726, "grad_norm": 1.3671875, "learning_rate": 3.3516429447917163e-06, "loss": 0.1338, "step": 43305 }, { "epoch": 31.726007326007327, "grad_norm": 1.3125, "learning_rate": 3.350599513437454e-06, "loss": 0.1039, "step": 43306 }, { "epoch": 31.726739926739928, "grad_norm": 1.0078125, "learning_rate": 3.349556307991439e-06, "loss": 0.1415, "step": 43307 }, { "epoch": 31.72747252747253, "grad_norm": 1.3359375, "learning_rate": 3.3485133284586788e-06, "loss": 0.1325, "step": 43308 }, { "epoch": 31.72820512820513, "grad_norm": 0.796875, "learning_rate": 3.347470574844175e-06, "loss": 0.1157, "step": 43309 }, { "epoch": 31.72893772893773, "grad_norm": 1.1484375, "learning_rate": 3.3464280471529225e-06, "loss": 0.1094, "step": 43310 }, { "epoch": 31.72967032967033, "grad_norm": 1.09375, "learning_rate": 3.3453857453899007e-06, "loss": 0.121, "step": 43311 }, { "epoch": 31.73040293040293, "grad_norm": 1.3359375, "learning_rate": 3.3443436695601334e-06, "loss": 0.1121, "step": 43312 }, { "epoch": 31.731135531135532, "grad_norm": 0.9921875, "learning_rate": 3.343301819668576e-06, "loss": 0.1382, "step": 43313 }, { "epoch": 31.731868131868133, "grad_norm": 0.921875, "learning_rate": 3.3422601957202523e-06, "loss": 0.108, "step": 43314 }, { "epoch": 31.732600732600734, "grad_norm": 1.453125, "learning_rate": 3.3412187977201357e-06, "loss": 0.1037, "step": 43315 }, { "epoch": 31.733333333333334, "grad_norm": 0.83203125, "learning_rate": 3.3401776256732276e-06, "loss": 0.1082, "step": 43316 }, { "epoch": 31.734065934065935, "grad_norm": 1.125, "learning_rate": 3.339136679584506e-06, "loss": 0.1384, "step": 43317 }, { "epoch": 31.734798534798536, "grad_norm": 0.90625, "learning_rate": 3.338095959458957e-06, "loss": 0.1421, "step": 43318 }, { "epoch": 31.735531135531136, "grad_norm": 1.703125, "learning_rate": 3.3370554653015746e-06, "loss": 0.1257, "step": 43319 }, { "epoch": 31.736263736263737, "grad_norm": 1.640625, "learning_rate": 3.336015197117355e-06, "loss": 0.1353, "step": 43320 }, { "epoch": 31.736996336996338, "grad_norm": 1.2421875, "learning_rate": 3.3349751549112547e-06, "loss": 0.1303, "step": 43321 }, { "epoch": 31.73772893772894, "grad_norm": 0.80859375, "learning_rate": 3.3339353386882757e-06, "loss": 0.1094, "step": 43322 }, { "epoch": 31.73846153846154, "grad_norm": 1.265625, "learning_rate": 3.3328957484534016e-06, "loss": 0.1104, "step": 43323 }, { "epoch": 31.73919413919414, "grad_norm": 0.75, "learning_rate": 3.3318563842115956e-06, "loss": 0.1174, "step": 43324 }, { "epoch": 31.73992673992674, "grad_norm": 1.3359375, "learning_rate": 3.3308172459678636e-06, "loss": 0.0953, "step": 43325 }, { "epoch": 31.74065934065934, "grad_norm": 1.2578125, "learning_rate": 3.3297783337271625e-06, "loss": 0.1215, "step": 43326 }, { "epoch": 31.741391941391942, "grad_norm": 1.046875, "learning_rate": 3.3287396474944834e-06, "loss": 0.079, "step": 43327 }, { "epoch": 31.742124542124543, "grad_norm": 0.74609375, "learning_rate": 3.3277011872747933e-06, "loss": 0.1257, "step": 43328 }, { "epoch": 31.742857142857144, "grad_norm": 0.8359375, "learning_rate": 3.326662953073072e-06, "loss": 0.1337, "step": 43329 }, { "epoch": 31.743589743589745, "grad_norm": 1.0234375, "learning_rate": 3.3256249448942968e-06, "loss": 0.1122, "step": 43330 }, { "epoch": 31.744322344322345, "grad_norm": 0.95703125, "learning_rate": 3.3245871627434435e-06, "loss": 0.1139, "step": 43331 }, { "epoch": 31.745054945054946, "grad_norm": 1.375, "learning_rate": 3.3235496066254743e-06, "loss": 0.1069, "step": 43332 }, { "epoch": 31.745787545787547, "grad_norm": 0.8515625, "learning_rate": 3.322512276545361e-06, "loss": 0.1108, "step": 43333 }, { "epoch": 31.746520146520147, "grad_norm": 1.140625, "learning_rate": 3.321475172508083e-06, "loss": 0.1466, "step": 43334 }, { "epoch": 31.747252747252748, "grad_norm": 0.9296875, "learning_rate": 3.3204382945186046e-06, "loss": 0.136, "step": 43335 }, { "epoch": 31.74798534798535, "grad_norm": 0.90625, "learning_rate": 3.3194016425819033e-06, "loss": 0.1373, "step": 43336 }, { "epoch": 31.74871794871795, "grad_norm": 1.4296875, "learning_rate": 3.3183652167029318e-06, "loss": 0.1114, "step": 43337 }, { "epoch": 31.74945054945055, "grad_norm": 0.7578125, "learning_rate": 3.3173290168866616e-06, "loss": 0.1013, "step": 43338 }, { "epoch": 31.75018315018315, "grad_norm": 1.1953125, "learning_rate": 3.3162930431380462e-06, "loss": 0.1269, "step": 43339 }, { "epoch": 31.750915750915752, "grad_norm": 1.2265625, "learning_rate": 3.315257295462075e-06, "loss": 0.1452, "step": 43340 }, { "epoch": 31.751648351648353, "grad_norm": 0.890625, "learning_rate": 3.314221773863688e-06, "loss": 0.118, "step": 43341 }, { "epoch": 31.752380952380953, "grad_norm": 1.1171875, "learning_rate": 3.3131864783478592e-06, "loss": 0.1221, "step": 43342 }, { "epoch": 31.753113553113554, "grad_norm": 0.609375, "learning_rate": 3.312151408919541e-06, "loss": 0.1218, "step": 43343 }, { "epoch": 31.753846153846155, "grad_norm": 1.203125, "learning_rate": 3.311116565583694e-06, "loss": 0.1235, "step": 43344 }, { "epoch": 31.754578754578755, "grad_norm": 1.171875, "learning_rate": 3.3100819483452824e-06, "loss": 0.1065, "step": 43345 }, { "epoch": 31.755311355311356, "grad_norm": 0.85546875, "learning_rate": 3.309047557209262e-06, "loss": 0.1447, "step": 43346 }, { "epoch": 31.756043956043957, "grad_norm": 1.1640625, "learning_rate": 3.308013392180585e-06, "loss": 0.1219, "step": 43347 }, { "epoch": 31.756776556776558, "grad_norm": 0.94921875, "learning_rate": 3.3069794532642037e-06, "loss": 0.1176, "step": 43348 }, { "epoch": 31.75750915750916, "grad_norm": 0.9375, "learning_rate": 3.305945740465084e-06, "loss": 0.1143, "step": 43349 }, { "epoch": 31.75824175824176, "grad_norm": 0.9375, "learning_rate": 3.304912253788158e-06, "loss": 0.1386, "step": 43350 }, { "epoch": 31.75897435897436, "grad_norm": 1.390625, "learning_rate": 3.303878993238403e-06, "loss": 0.1277, "step": 43351 }, { "epoch": 31.75970695970696, "grad_norm": 1.1015625, "learning_rate": 3.3028459588207495e-06, "loss": 0.0978, "step": 43352 }, { "epoch": 31.76043956043956, "grad_norm": 1.0078125, "learning_rate": 3.3018131505401643e-06, "loss": 0.1295, "step": 43353 }, { "epoch": 31.761172161172162, "grad_norm": 1.2578125, "learning_rate": 3.300780568401578e-06, "loss": 0.1126, "step": 43354 }, { "epoch": 31.761904761904763, "grad_norm": 1.03125, "learning_rate": 3.2997482124099478e-06, "loss": 0.1286, "step": 43355 }, { "epoch": 31.762637362637363, "grad_norm": 1.2890625, "learning_rate": 3.298716082570219e-06, "loss": 0.1103, "step": 43356 }, { "epoch": 31.763369963369964, "grad_norm": 0.84375, "learning_rate": 3.297684178887338e-06, "loss": 0.0906, "step": 43357 }, { "epoch": 31.764102564102565, "grad_norm": 1.3828125, "learning_rate": 3.2966525013662453e-06, "loss": 0.1132, "step": 43358 }, { "epoch": 31.764835164835166, "grad_norm": 0.7734375, "learning_rate": 3.2956210500118866e-06, "loss": 0.1183, "step": 43359 }, { "epoch": 31.765567765567766, "grad_norm": 1.15625, "learning_rate": 3.294589824829203e-06, "loss": 0.1336, "step": 43360 }, { "epoch": 31.766300366300367, "grad_norm": 0.7734375, "learning_rate": 3.293558825823135e-06, "loss": 0.1072, "step": 43361 }, { "epoch": 31.767032967032968, "grad_norm": 0.73046875, "learning_rate": 3.2925280529986336e-06, "loss": 0.1029, "step": 43362 }, { "epoch": 31.76776556776557, "grad_norm": 0.73046875, "learning_rate": 3.2914975063606177e-06, "loss": 0.15, "step": 43363 }, { "epoch": 31.76849816849817, "grad_norm": 1.359375, "learning_rate": 3.290467185914045e-06, "loss": 0.1451, "step": 43364 }, { "epoch": 31.76923076923077, "grad_norm": 1.3046875, "learning_rate": 3.2894370916638216e-06, "loss": 0.1231, "step": 43365 }, { "epoch": 31.76996336996337, "grad_norm": 0.96875, "learning_rate": 3.288407223614923e-06, "loss": 0.1085, "step": 43366 }, { "epoch": 31.77069597069597, "grad_norm": 1.3046875, "learning_rate": 3.287377581772255e-06, "loss": 0.1209, "step": 43367 }, { "epoch": 31.771428571428572, "grad_norm": 1.0625, "learning_rate": 3.2863481661407695e-06, "loss": 0.1279, "step": 43368 }, { "epoch": 31.772161172161173, "grad_norm": 1.2109375, "learning_rate": 3.2853189767253807e-06, "loss": 0.1026, "step": 43369 }, { "epoch": 31.772893772893774, "grad_norm": 1.0234375, "learning_rate": 3.2842900135310216e-06, "loss": 0.1259, "step": 43370 }, { "epoch": 31.773626373626374, "grad_norm": 0.875, "learning_rate": 3.2832612765626354e-06, "loss": 0.0961, "step": 43371 }, { "epoch": 31.774358974358975, "grad_norm": 1.3359375, "learning_rate": 3.2822327658251438e-06, "loss": 0.124, "step": 43372 }, { "epoch": 31.775091575091576, "grad_norm": 1.2734375, "learning_rate": 3.281204481323483e-06, "loss": 0.1032, "step": 43373 }, { "epoch": 31.775824175824177, "grad_norm": 1.03125, "learning_rate": 3.2801764230625657e-06, "loss": 0.1445, "step": 43374 }, { "epoch": 31.776556776556777, "grad_norm": 0.9921875, "learning_rate": 3.2791485910473325e-06, "loss": 0.0996, "step": 43375 }, { "epoch": 31.777289377289378, "grad_norm": 1.1875, "learning_rate": 3.2781209852826856e-06, "loss": 0.1125, "step": 43376 }, { "epoch": 31.77802197802198, "grad_norm": 1.3671875, "learning_rate": 3.277093605773577e-06, "loss": 0.1242, "step": 43377 }, { "epoch": 31.77875457875458, "grad_norm": 1.1640625, "learning_rate": 3.2760664525249022e-06, "loss": 0.12, "step": 43378 }, { "epoch": 31.77948717948718, "grad_norm": 1.125, "learning_rate": 3.2750395255416082e-06, "loss": 0.1073, "step": 43379 }, { "epoch": 31.78021978021978, "grad_norm": 1.0625, "learning_rate": 3.2740128248285963e-06, "loss": 0.0936, "step": 43380 }, { "epoch": 31.78095238095238, "grad_norm": 1.25, "learning_rate": 3.2729863503907914e-06, "loss": 0.1105, "step": 43381 }, { "epoch": 31.781684981684982, "grad_norm": 0.62109375, "learning_rate": 3.271960102233111e-06, "loss": 0.0977, "step": 43382 }, { "epoch": 31.782417582417583, "grad_norm": 1.6953125, "learning_rate": 3.2709340803604806e-06, "loss": 0.1787, "step": 43383 }, { "epoch": 31.783150183150184, "grad_norm": 0.84765625, "learning_rate": 3.269908284777807e-06, "loss": 0.1169, "step": 43384 }, { "epoch": 31.783882783882785, "grad_norm": 0.71484375, "learning_rate": 3.268882715490003e-06, "loss": 0.0985, "step": 43385 }, { "epoch": 31.784615384615385, "grad_norm": 1.0, "learning_rate": 3.267857372501987e-06, "loss": 0.1057, "step": 43386 }, { "epoch": 31.785347985347986, "grad_norm": 0.86328125, "learning_rate": 3.2668322558186726e-06, "loss": 0.125, "step": 43387 }, { "epoch": 31.786080586080587, "grad_norm": 1.0078125, "learning_rate": 3.2658073654449782e-06, "loss": 0.1154, "step": 43388 }, { "epoch": 31.786813186813188, "grad_norm": 0.86328125, "learning_rate": 3.2647827013857954e-06, "loss": 0.1374, "step": 43389 }, { "epoch": 31.78754578754579, "grad_norm": 0.96484375, "learning_rate": 3.2637582636460586e-06, "loss": 0.1219, "step": 43390 }, { "epoch": 31.78827838827839, "grad_norm": 1.1015625, "learning_rate": 3.262734052230643e-06, "loss": 0.1288, "step": 43391 }, { "epoch": 31.78901098901099, "grad_norm": 0.94921875, "learning_rate": 3.261710067144493e-06, "loss": 0.1295, "step": 43392 }, { "epoch": 31.78974358974359, "grad_norm": 1.015625, "learning_rate": 3.2606863083924853e-06, "loss": 0.1066, "step": 43393 }, { "epoch": 31.79047619047619, "grad_norm": 1.0234375, "learning_rate": 3.259662775979548e-06, "loss": 0.1301, "step": 43394 }, { "epoch": 31.791208791208792, "grad_norm": 1.0703125, "learning_rate": 3.2586394699105684e-06, "loss": 0.1236, "step": 43395 }, { "epoch": 31.791941391941393, "grad_norm": 0.9375, "learning_rate": 3.2576163901904574e-06, "loss": 0.1314, "step": 43396 }, { "epoch": 31.792673992673993, "grad_norm": 1.1640625, "learning_rate": 3.2565935368241136e-06, "loss": 0.1227, "step": 43397 }, { "epoch": 31.793406593406594, "grad_norm": 1.53125, "learning_rate": 3.255570909816437e-06, "loss": 0.1378, "step": 43398 }, { "epoch": 31.794139194139195, "grad_norm": 1.2109375, "learning_rate": 3.2545485091723365e-06, "loss": 0.1164, "step": 43399 }, { "epoch": 31.794871794871796, "grad_norm": 1.21875, "learning_rate": 3.2535263348966927e-06, "loss": 0.1327, "step": 43400 }, { "epoch": 31.795604395604396, "grad_norm": 0.7265625, "learning_rate": 3.2525043869944288e-06, "loss": 0.1124, "step": 43401 }, { "epoch": 31.796336996336997, "grad_norm": 0.95703125, "learning_rate": 3.251482665470408e-06, "loss": 0.1109, "step": 43402 }, { "epoch": 31.797069597069598, "grad_norm": 0.66796875, "learning_rate": 3.2504611703295606e-06, "loss": 0.09, "step": 43403 }, { "epoch": 31.7978021978022, "grad_norm": 1.0546875, "learning_rate": 3.249439901576756e-06, "loss": 0.1054, "step": 43404 }, { "epoch": 31.7985347985348, "grad_norm": 0.94140625, "learning_rate": 3.248418859216906e-06, "loss": 0.1094, "step": 43405 }, { "epoch": 31.7992673992674, "grad_norm": 0.8359375, "learning_rate": 3.2473980432548864e-06, "loss": 0.1291, "step": 43406 }, { "epoch": 31.8, "grad_norm": 1.3125, "learning_rate": 3.2463774536955934e-06, "loss": 0.1311, "step": 43407 }, { "epoch": 31.8007326007326, "grad_norm": 0.875, "learning_rate": 3.2453570905439183e-06, "loss": 0.1132, "step": 43408 }, { "epoch": 31.801465201465202, "grad_norm": 1.3203125, "learning_rate": 3.244336953804751e-06, "loss": 0.1508, "step": 43409 }, { "epoch": 31.802197802197803, "grad_norm": 0.921875, "learning_rate": 3.243317043482979e-06, "loss": 0.1039, "step": 43410 }, { "epoch": 31.802930402930404, "grad_norm": 0.78125, "learning_rate": 3.2422973595834864e-06, "loss": 0.1141, "step": 43411 }, { "epoch": 31.803663003663004, "grad_norm": 1.015625, "learning_rate": 3.2412779021111594e-06, "loss": 0.1058, "step": 43412 }, { "epoch": 31.804395604395605, "grad_norm": 1.2578125, "learning_rate": 3.2402586710708893e-06, "loss": 0.1344, "step": 43413 }, { "epoch": 31.805128205128206, "grad_norm": 0.8984375, "learning_rate": 3.2392396664675558e-06, "loss": 0.1001, "step": 43414 }, { "epoch": 31.805860805860807, "grad_norm": 0.75390625, "learning_rate": 3.238220888306033e-06, "loss": 0.1303, "step": 43415 }, { "epoch": 31.806593406593407, "grad_norm": 0.91015625, "learning_rate": 3.237202336591219e-06, "loss": 0.124, "step": 43416 }, { "epoch": 31.807326007326008, "grad_norm": 1.4140625, "learning_rate": 3.2361840113279654e-06, "loss": 0.0993, "step": 43417 }, { "epoch": 31.80805860805861, "grad_norm": 1.5859375, "learning_rate": 3.2351659125211904e-06, "loss": 0.1386, "step": 43418 }, { "epoch": 31.80879120879121, "grad_norm": 1.1953125, "learning_rate": 3.2341480401757355e-06, "loss": 0.1491, "step": 43419 }, { "epoch": 31.80952380952381, "grad_norm": 1.21875, "learning_rate": 3.23313039429651e-06, "loss": 0.1276, "step": 43420 }, { "epoch": 31.81025641025641, "grad_norm": 1.4921875, "learning_rate": 3.23211297488836e-06, "loss": 0.1332, "step": 43421 }, { "epoch": 31.81098901098901, "grad_norm": 1.40625, "learning_rate": 3.2310957819561754e-06, "loss": 0.1109, "step": 43422 }, { "epoch": 31.811721611721612, "grad_norm": 0.66796875, "learning_rate": 3.2300788155048335e-06, "loss": 0.1091, "step": 43423 }, { "epoch": 31.812454212454213, "grad_norm": 1.4921875, "learning_rate": 3.2290620755391952e-06, "loss": 0.1499, "step": 43424 }, { "epoch": 31.813186813186814, "grad_norm": 1.109375, "learning_rate": 3.228045562064147e-06, "loss": 0.123, "step": 43425 }, { "epoch": 31.813919413919415, "grad_norm": 0.50390625, "learning_rate": 3.2270292750845476e-06, "loss": 0.1005, "step": 43426 }, { "epoch": 31.814652014652015, "grad_norm": 0.859375, "learning_rate": 3.226013214605271e-06, "loss": 0.1113, "step": 43427 }, { "epoch": 31.815384615384616, "grad_norm": 1.015625, "learning_rate": 3.2249973806311703e-06, "loss": 0.1188, "step": 43428 }, { "epoch": 31.816117216117217, "grad_norm": 1.09375, "learning_rate": 3.2239817731671465e-06, "loss": 0.1486, "step": 43429 }, { "epoch": 31.816849816849818, "grad_norm": 0.703125, "learning_rate": 3.2229663922180313e-06, "loss": 0.104, "step": 43430 }, { "epoch": 31.81758241758242, "grad_norm": 1.28125, "learning_rate": 3.221951237788716e-06, "loss": 0.1723, "step": 43431 }, { "epoch": 31.81831501831502, "grad_norm": 1.2265625, "learning_rate": 3.220936309884047e-06, "loss": 0.0929, "step": 43432 }, { "epoch": 31.81904761904762, "grad_norm": 0.76953125, "learning_rate": 3.2199216085088884e-06, "loss": 0.1155, "step": 43433 }, { "epoch": 31.81978021978022, "grad_norm": 0.81640625, "learning_rate": 3.218907133668109e-06, "loss": 0.1449, "step": 43434 }, { "epoch": 31.82051282051282, "grad_norm": 0.88671875, "learning_rate": 3.2178928853665718e-06, "loss": 0.1139, "step": 43435 }, { "epoch": 31.821245421245422, "grad_norm": 0.76171875, "learning_rate": 3.2168788636091253e-06, "loss": 0.1255, "step": 43436 }, { "epoch": 31.821978021978023, "grad_norm": 0.94140625, "learning_rate": 3.2158650684006376e-06, "loss": 0.119, "step": 43437 }, { "epoch": 31.822710622710623, "grad_norm": 0.95703125, "learning_rate": 3.214851499745962e-06, "loss": 0.1291, "step": 43438 }, { "epoch": 31.823443223443224, "grad_norm": 0.94140625, "learning_rate": 3.2138381576499503e-06, "loss": 0.1115, "step": 43439 }, { "epoch": 31.824175824175825, "grad_norm": 1.03125, "learning_rate": 3.212825042117472e-06, "loss": 0.1183, "step": 43440 }, { "epoch": 31.824908424908426, "grad_norm": 0.9921875, "learning_rate": 3.211812153153369e-06, "loss": 0.1143, "step": 43441 }, { "epoch": 31.825641025641026, "grad_norm": 1.34375, "learning_rate": 3.210799490762505e-06, "loss": 0.1274, "step": 43442 }, { "epoch": 31.826373626373627, "grad_norm": 1.046875, "learning_rate": 3.2097870549497048e-06, "loss": 0.1073, "step": 43443 }, { "epoch": 31.827106227106228, "grad_norm": 0.99609375, "learning_rate": 3.2087748457198584e-06, "loss": 0.1168, "step": 43444 }, { "epoch": 31.82783882783883, "grad_norm": 0.9921875, "learning_rate": 3.2077628630777874e-06, "loss": 0.1258, "step": 43445 }, { "epoch": 31.82857142857143, "grad_norm": 0.80078125, "learning_rate": 3.2067511070283546e-06, "loss": 0.1217, "step": 43446 }, { "epoch": 31.82930402930403, "grad_norm": 1.0, "learning_rate": 3.2057395775763957e-06, "loss": 0.1073, "step": 43447 }, { "epoch": 31.83003663003663, "grad_norm": 0.62890625, "learning_rate": 3.204728274726769e-06, "loss": 0.0903, "step": 43448 }, { "epoch": 31.83076923076923, "grad_norm": 0.8125, "learning_rate": 3.2037171984843113e-06, "loss": 0.118, "step": 43449 }, { "epoch": 31.831501831501832, "grad_norm": 1.296875, "learning_rate": 3.202706348853875e-06, "loss": 0.1239, "step": 43450 }, { "epoch": 31.832234432234433, "grad_norm": 1.1875, "learning_rate": 3.2016957258403006e-06, "loss": 0.0944, "step": 43451 }, { "epoch": 31.832967032967034, "grad_norm": 1.0234375, "learning_rate": 3.200685329448426e-06, "loss": 0.1387, "step": 43452 }, { "epoch": 31.833699633699634, "grad_norm": 0.8046875, "learning_rate": 3.199675159683097e-06, "loss": 0.1194, "step": 43453 }, { "epoch": 31.834432234432235, "grad_norm": 1.3046875, "learning_rate": 3.1986652165491446e-06, "loss": 0.1148, "step": 43454 }, { "epoch": 31.835164835164836, "grad_norm": 0.734375, "learning_rate": 3.197655500051422e-06, "loss": 0.0942, "step": 43455 }, { "epoch": 31.835897435897436, "grad_norm": 0.90625, "learning_rate": 3.1966460101947594e-06, "loss": 0.1196, "step": 43456 }, { "epoch": 31.836630036630037, "grad_norm": 1.0234375, "learning_rate": 3.1956367469839974e-06, "loss": 0.1116, "step": 43457 }, { "epoch": 31.837362637362638, "grad_norm": 0.65625, "learning_rate": 3.194627710423968e-06, "loss": 0.1076, "step": 43458 }, { "epoch": 31.83809523809524, "grad_norm": 1.03125, "learning_rate": 3.1936189005195016e-06, "loss": 0.1217, "step": 43459 }, { "epoch": 31.83882783882784, "grad_norm": 1.140625, "learning_rate": 3.1926103172754392e-06, "loss": 0.1193, "step": 43460 }, { "epoch": 31.83956043956044, "grad_norm": 0.8671875, "learning_rate": 3.191601960696612e-06, "loss": 0.1108, "step": 43461 }, { "epoch": 31.84029304029304, "grad_norm": 1.59375, "learning_rate": 3.190593830787851e-06, "loss": 0.1306, "step": 43462 }, { "epoch": 31.84102564102564, "grad_norm": 0.9921875, "learning_rate": 3.1895859275539814e-06, "loss": 0.1216, "step": 43463 }, { "epoch": 31.841758241758242, "grad_norm": 1.03125, "learning_rate": 3.188578250999838e-06, "loss": 0.1244, "step": 43464 }, { "epoch": 31.842490842490843, "grad_norm": 1.046875, "learning_rate": 3.187570801130246e-06, "loss": 0.1142, "step": 43465 }, { "epoch": 31.843223443223444, "grad_norm": 0.95703125, "learning_rate": 3.1865635779500436e-06, "loss": 0.1148, "step": 43466 }, { "epoch": 31.843956043956045, "grad_norm": 0.9140625, "learning_rate": 3.1855565814640372e-06, "loss": 0.1174, "step": 43467 }, { "epoch": 31.844688644688645, "grad_norm": 0.83984375, "learning_rate": 3.184549811677069e-06, "loss": 0.1141, "step": 43468 }, { "epoch": 31.845421245421246, "grad_norm": 1.3984375, "learning_rate": 3.183543268593949e-06, "loss": 0.139, "step": 43469 }, { "epoch": 31.846153846153847, "grad_norm": 1.3828125, "learning_rate": 3.1825369522195016e-06, "loss": 0.1765, "step": 43470 }, { "epoch": 31.846886446886447, "grad_norm": 0.921875, "learning_rate": 3.1815308625585614e-06, "loss": 0.1153, "step": 43471 }, { "epoch": 31.847619047619048, "grad_norm": 1.078125, "learning_rate": 3.1805249996159395e-06, "loss": 0.1165, "step": 43472 }, { "epoch": 31.84835164835165, "grad_norm": 1.1328125, "learning_rate": 3.1795193633964536e-06, "loss": 0.1189, "step": 43473 }, { "epoch": 31.84908424908425, "grad_norm": 1.65625, "learning_rate": 3.178513953904924e-06, "loss": 0.1224, "step": 43474 }, { "epoch": 31.84981684981685, "grad_norm": 0.98828125, "learning_rate": 3.177508771146171e-06, "loss": 0.1193, "step": 43475 }, { "epoch": 31.85054945054945, "grad_norm": 1.7421875, "learning_rate": 3.176503815125008e-06, "loss": 0.1328, "step": 43476 }, { "epoch": 31.851282051282052, "grad_norm": 1.546875, "learning_rate": 3.175499085846255e-06, "loss": 0.1402, "step": 43477 }, { "epoch": 31.852014652014653, "grad_norm": 0.8671875, "learning_rate": 3.174494583314715e-06, "loss": 0.1158, "step": 43478 }, { "epoch": 31.852747252747253, "grad_norm": 1.1875, "learning_rate": 3.1734903075352125e-06, "loss": 0.1297, "step": 43479 }, { "epoch": 31.853479853479854, "grad_norm": 1.3046875, "learning_rate": 3.1724862585125464e-06, "loss": 0.14, "step": 43480 }, { "epoch": 31.854212454212455, "grad_norm": 1.6875, "learning_rate": 3.171482436251541e-06, "loss": 0.1608, "step": 43481 }, { "epoch": 31.854945054945055, "grad_norm": 0.71875, "learning_rate": 3.1704788407570004e-06, "loss": 0.1562, "step": 43482 }, { "epoch": 31.855677655677656, "grad_norm": 1.1796875, "learning_rate": 3.1694754720337368e-06, "loss": 0.1253, "step": 43483 }, { "epoch": 31.856410256410257, "grad_norm": 1.546875, "learning_rate": 3.1684723300865444e-06, "loss": 0.189, "step": 43484 }, { "epoch": 31.857142857142858, "grad_norm": 0.609375, "learning_rate": 3.167469414920242e-06, "loss": 0.0806, "step": 43485 }, { "epoch": 31.85787545787546, "grad_norm": 1.078125, "learning_rate": 3.166466726539633e-06, "loss": 0.1394, "step": 43486 }, { "epoch": 31.85860805860806, "grad_norm": 0.80078125, "learning_rate": 3.1654642649495205e-06, "loss": 0.1451, "step": 43487 }, { "epoch": 31.85934065934066, "grad_norm": 0.83984375, "learning_rate": 3.164462030154712e-06, "loss": 0.1284, "step": 43488 }, { "epoch": 31.86007326007326, "grad_norm": 1.1171875, "learning_rate": 3.1634600221600015e-06, "loss": 0.1365, "step": 43489 }, { "epoch": 31.86080586080586, "grad_norm": 1.3359375, "learning_rate": 3.162458240970192e-06, "loss": 0.1411, "step": 43490 }, { "epoch": 31.861538461538462, "grad_norm": 1.3203125, "learning_rate": 3.1614566865900795e-06, "loss": 0.1017, "step": 43491 }, { "epoch": 31.862271062271063, "grad_norm": 1.2890625, "learning_rate": 3.1604553590244793e-06, "loss": 0.148, "step": 43492 }, { "epoch": 31.863003663003663, "grad_norm": 1.0390625, "learning_rate": 3.159454258278173e-06, "loss": 0.0923, "step": 43493 }, { "epoch": 31.863736263736264, "grad_norm": 0.8203125, "learning_rate": 3.158453384355968e-06, "loss": 0.1043, "step": 43494 }, { "epoch": 31.864468864468865, "grad_norm": 1.8671875, "learning_rate": 3.1574527372626465e-06, "loss": 0.1235, "step": 43495 }, { "epoch": 31.865201465201466, "grad_norm": 1.0390625, "learning_rate": 3.1564523170030115e-06, "loss": 0.1244, "step": 43496 }, { "epoch": 31.865934065934066, "grad_norm": 0.9765625, "learning_rate": 3.155452123581855e-06, "loss": 0.1136, "step": 43497 }, { "epoch": 31.866666666666667, "grad_norm": 1.2734375, "learning_rate": 3.1544521570039802e-06, "loss": 0.1008, "step": 43498 }, { "epoch": 31.867399267399268, "grad_norm": 1.453125, "learning_rate": 3.1534524172741567e-06, "loss": 0.1392, "step": 43499 }, { "epoch": 31.86813186813187, "grad_norm": 1.75, "learning_rate": 3.1524529043971832e-06, "loss": 0.1606, "step": 43500 }, { "epoch": 31.86886446886447, "grad_norm": 0.921875, "learning_rate": 3.1514536183778566e-06, "loss": 0.1111, "step": 43501 }, { "epoch": 31.86959706959707, "grad_norm": 0.953125, "learning_rate": 3.150454559220953e-06, "loss": 0.1605, "step": 43502 }, { "epoch": 31.87032967032967, "grad_norm": 1.15625, "learning_rate": 3.1494557269312752e-06, "loss": 0.1147, "step": 43503 }, { "epoch": 31.87106227106227, "grad_norm": 1.109375, "learning_rate": 3.148457121513594e-06, "loss": 0.1253, "step": 43504 }, { "epoch": 31.871794871794872, "grad_norm": 1.3515625, "learning_rate": 3.147458742972706e-06, "loss": 0.1503, "step": 43505 }, { "epoch": 31.872527472527473, "grad_norm": 0.83203125, "learning_rate": 3.1464605913133714e-06, "loss": 0.1318, "step": 43506 }, { "epoch": 31.873260073260074, "grad_norm": 1.6796875, "learning_rate": 3.1454626665404046e-06, "loss": 0.1287, "step": 43507 }, { "epoch": 31.873992673992674, "grad_norm": 0.8671875, "learning_rate": 3.1444649686585636e-06, "loss": 0.1015, "step": 43508 }, { "epoch": 31.874725274725275, "grad_norm": 1.2734375, "learning_rate": 3.143467497672647e-06, "loss": 0.1241, "step": 43509 }, { "epoch": 31.875457875457876, "grad_norm": 0.89453125, "learning_rate": 3.142470253587413e-06, "loss": 0.1149, "step": 43510 }, { "epoch": 31.876190476190477, "grad_norm": 0.921875, "learning_rate": 3.1414732364076547e-06, "loss": 0.1225, "step": 43511 }, { "epoch": 31.876923076923077, "grad_norm": 1.5, "learning_rate": 3.1404764461381423e-06, "loss": 0.1058, "step": 43512 }, { "epoch": 31.877655677655678, "grad_norm": 0.9453125, "learning_rate": 3.139479882783651e-06, "loss": 0.1181, "step": 43513 }, { "epoch": 31.87838827838828, "grad_norm": 1.3046875, "learning_rate": 3.138483546348973e-06, "loss": 0.1096, "step": 43514 }, { "epoch": 31.87912087912088, "grad_norm": 1.015625, "learning_rate": 3.1374874368388567e-06, "loss": 0.1276, "step": 43515 }, { "epoch": 31.87985347985348, "grad_norm": 0.62109375, "learning_rate": 3.1364915542580884e-06, "loss": 0.0969, "step": 43516 }, { "epoch": 31.88058608058608, "grad_norm": 1.1484375, "learning_rate": 3.1354958986114333e-06, "loss": 0.1258, "step": 43517 }, { "epoch": 31.88131868131868, "grad_norm": 1.1640625, "learning_rate": 3.1345004699036784e-06, "loss": 0.1365, "step": 43518 }, { "epoch": 31.882051282051282, "grad_norm": 1.09375, "learning_rate": 3.133505268139571e-06, "loss": 0.1189, "step": 43519 }, { "epoch": 31.882783882783883, "grad_norm": 0.76171875, "learning_rate": 3.1325102933238927e-06, "loss": 0.13, "step": 43520 }, { "epoch": 31.883516483516484, "grad_norm": 1.28125, "learning_rate": 3.131515545461408e-06, "loss": 0.1246, "step": 43521 }, { "epoch": 31.884249084249085, "grad_norm": 1.4140625, "learning_rate": 3.1305210245568763e-06, "loss": 0.1262, "step": 43522 }, { "epoch": 31.884981684981685, "grad_norm": 0.63671875, "learning_rate": 3.1295267306150734e-06, "loss": 0.1231, "step": 43523 }, { "epoch": 31.885714285714286, "grad_norm": 0.8046875, "learning_rate": 3.1285326636407633e-06, "loss": 0.1163, "step": 43524 }, { "epoch": 31.886446886446887, "grad_norm": 1.53125, "learning_rate": 3.1275388236386946e-06, "loss": 0.1257, "step": 43525 }, { "epoch": 31.887179487179488, "grad_norm": 1.421875, "learning_rate": 3.1265452106136435e-06, "loss": 0.1513, "step": 43526 }, { "epoch": 31.88791208791209, "grad_norm": 0.91796875, "learning_rate": 3.1255518245703635e-06, "loss": 0.1096, "step": 43527 }, { "epoch": 31.88864468864469, "grad_norm": 0.98828125, "learning_rate": 3.1245586655136183e-06, "loss": 0.0945, "step": 43528 }, { "epoch": 31.88937728937729, "grad_norm": 0.9375, "learning_rate": 3.1235657334481687e-06, "loss": 0.0834, "step": 43529 }, { "epoch": 31.89010989010989, "grad_norm": 1.0078125, "learning_rate": 3.1225730283787618e-06, "loss": 0.123, "step": 43530 }, { "epoch": 31.89084249084249, "grad_norm": 0.97265625, "learning_rate": 3.121580550310168e-06, "loss": 0.1175, "step": 43531 }, { "epoch": 31.891575091575092, "grad_norm": 1.109375, "learning_rate": 3.1205882992471247e-06, "loss": 0.0956, "step": 43532 }, { "epoch": 31.892307692307693, "grad_norm": 1.328125, "learning_rate": 3.119596275194407e-06, "loss": 0.1018, "step": 43533 }, { "epoch": 31.893040293040293, "grad_norm": 1.046875, "learning_rate": 3.1186044781567532e-06, "loss": 0.1309, "step": 43534 }, { "epoch": 31.893772893772894, "grad_norm": 1.15625, "learning_rate": 3.1176129081389213e-06, "loss": 0.1281, "step": 43535 }, { "epoch": 31.894505494505495, "grad_norm": 1.125, "learning_rate": 3.1166215651456597e-06, "loss": 0.1311, "step": 43536 }, { "epoch": 31.895238095238096, "grad_norm": 1.0703125, "learning_rate": 3.115630449181718e-06, "loss": 0.1069, "step": 43537 }, { "epoch": 31.895970695970696, "grad_norm": 1.03125, "learning_rate": 3.1146395602518425e-06, "loss": 0.1001, "step": 43538 }, { "epoch": 31.896703296703297, "grad_norm": 1.1015625, "learning_rate": 3.113648898360788e-06, "loss": 0.1134, "step": 43539 }, { "epoch": 31.897435897435898, "grad_norm": 1.2421875, "learning_rate": 3.1126584635133025e-06, "loss": 0.1364, "step": 43540 }, { "epoch": 31.8981684981685, "grad_norm": 0.83203125, "learning_rate": 3.1116682557141177e-06, "loss": 0.0949, "step": 43541 }, { "epoch": 31.8989010989011, "grad_norm": 1.3671875, "learning_rate": 3.1106782749679925e-06, "loss": 0.1328, "step": 43542 }, { "epoch": 31.8996336996337, "grad_norm": 1.15625, "learning_rate": 3.109688521279659e-06, "loss": 0.124, "step": 43543 }, { "epoch": 31.9003663003663, "grad_norm": 0.90234375, "learning_rate": 3.1086989946538768e-06, "loss": 0.1416, "step": 43544 }, { "epoch": 31.9010989010989, "grad_norm": 0.73828125, "learning_rate": 3.1077096950953658e-06, "loss": 0.1033, "step": 43545 }, { "epoch": 31.901831501831502, "grad_norm": 1.2421875, "learning_rate": 3.1067206226088794e-06, "loss": 0.1227, "step": 43546 }, { "epoch": 31.902564102564103, "grad_norm": 1.5, "learning_rate": 3.1057317771991503e-06, "loss": 0.1485, "step": 43547 }, { "epoch": 31.903296703296704, "grad_norm": 1.234375, "learning_rate": 3.1047431588709152e-06, "loss": 0.137, "step": 43548 }, { "epoch": 31.904029304029304, "grad_norm": 0.72265625, "learning_rate": 3.1037547676289175e-06, "loss": 0.1058, "step": 43549 }, { "epoch": 31.904761904761905, "grad_norm": 1.1953125, "learning_rate": 3.1027666034778984e-06, "loss": 0.1188, "step": 43550 }, { "epoch": 31.905494505494506, "grad_norm": 0.9921875, "learning_rate": 3.1017786664225746e-06, "loss": 0.1265, "step": 43551 }, { "epoch": 31.906227106227107, "grad_norm": 0.9921875, "learning_rate": 3.1007909564676882e-06, "loss": 0.1391, "step": 43552 }, { "epoch": 31.906959706959707, "grad_norm": 1.265625, "learning_rate": 3.0998034736179763e-06, "loss": 0.1097, "step": 43553 }, { "epoch": 31.907692307692308, "grad_norm": 0.99609375, "learning_rate": 3.0988162178781603e-06, "loss": 0.1149, "step": 43554 }, { "epoch": 31.90842490842491, "grad_norm": 0.99609375, "learning_rate": 3.0978291892529873e-06, "loss": 0.1153, "step": 43555 }, { "epoch": 31.90915750915751, "grad_norm": 1.140625, "learning_rate": 3.096842387747168e-06, "loss": 0.1664, "step": 43556 }, { "epoch": 31.90989010989011, "grad_norm": 1.171875, "learning_rate": 3.0958558133654443e-06, "loss": 0.119, "step": 43557 }, { "epoch": 31.91062271062271, "grad_norm": 0.88671875, "learning_rate": 3.0948694661125264e-06, "loss": 0.1408, "step": 43558 }, { "epoch": 31.91135531135531, "grad_norm": 1.09375, "learning_rate": 3.093883345993163e-06, "loss": 0.12, "step": 43559 }, { "epoch": 31.912087912087912, "grad_norm": 0.87109375, "learning_rate": 3.0928974530120576e-06, "loss": 0.1121, "step": 43560 }, { "epoch": 31.912820512820513, "grad_norm": 0.92578125, "learning_rate": 3.0919117871739527e-06, "loss": 0.0891, "step": 43561 }, { "epoch": 31.913553113553114, "grad_norm": 0.96875, "learning_rate": 3.0909263484835535e-06, "loss": 0.1223, "step": 43562 }, { "epoch": 31.914285714285715, "grad_norm": 0.90625, "learning_rate": 3.0899411369455907e-06, "loss": 0.1138, "step": 43563 }, { "epoch": 31.915018315018315, "grad_norm": 1.6640625, "learning_rate": 3.0889561525647856e-06, "loss": 0.1254, "step": 43564 }, { "epoch": 31.915750915750916, "grad_norm": 1.09375, "learning_rate": 3.0879713953458537e-06, "loss": 0.1232, "step": 43565 }, { "epoch": 31.916483516483517, "grad_norm": 0.98046875, "learning_rate": 3.08698686529352e-06, "loss": 0.1413, "step": 43566 }, { "epoch": 31.917216117216118, "grad_norm": 1.15625, "learning_rate": 3.0860025624124906e-06, "loss": 0.1123, "step": 43567 }, { "epoch": 31.91794871794872, "grad_norm": 1.265625, "learning_rate": 3.08501848670749e-06, "loss": 0.1339, "step": 43568 }, { "epoch": 31.91868131868132, "grad_norm": 1.203125, "learning_rate": 3.0840346381832286e-06, "loss": 0.1048, "step": 43569 }, { "epoch": 31.91941391941392, "grad_norm": 1.3828125, "learning_rate": 3.0830510168444333e-06, "loss": 0.143, "step": 43570 }, { "epoch": 31.92014652014652, "grad_norm": 1.3046875, "learning_rate": 3.082067622695796e-06, "loss": 0.1369, "step": 43571 }, { "epoch": 31.92087912087912, "grad_norm": 1.4453125, "learning_rate": 3.0810844557420492e-06, "loss": 0.1307, "step": 43572 }, { "epoch": 31.921611721611722, "grad_norm": 1.125, "learning_rate": 3.080101515987886e-06, "loss": 0.1264, "step": 43573 }, { "epoch": 31.922344322344323, "grad_norm": 0.6796875, "learning_rate": 3.079118803438022e-06, "loss": 0.12, "step": 43574 }, { "epoch": 31.923076923076923, "grad_norm": 0.8046875, "learning_rate": 3.0781363180971666e-06, "loss": 0.1141, "step": 43575 }, { "epoch": 31.923809523809524, "grad_norm": 1.234375, "learning_rate": 3.077154059970035e-06, "loss": 0.0965, "step": 43576 }, { "epoch": 31.924542124542125, "grad_norm": 1.28125, "learning_rate": 3.07617202906132e-06, "loss": 0.122, "step": 43577 }, { "epoch": 31.925274725274726, "grad_norm": 0.84375, "learning_rate": 3.075190225375732e-06, "loss": 0.0989, "step": 43578 }, { "epoch": 31.926007326007326, "grad_norm": 1.2109375, "learning_rate": 3.0742086489179755e-06, "loss": 0.0921, "step": 43579 }, { "epoch": 31.926739926739927, "grad_norm": 0.91015625, "learning_rate": 3.0732272996927545e-06, "loss": 0.0952, "step": 43580 }, { "epoch": 31.927472527472528, "grad_norm": 1.140625, "learning_rate": 3.0722461777047777e-06, "loss": 0.1302, "step": 43581 }, { "epoch": 31.92820512820513, "grad_norm": 0.99609375, "learning_rate": 3.071265282958734e-06, "loss": 0.1302, "step": 43582 }, { "epoch": 31.92893772893773, "grad_norm": 1.25, "learning_rate": 3.0702846154593327e-06, "loss": 0.1256, "step": 43583 }, { "epoch": 31.92967032967033, "grad_norm": 1.7265625, "learning_rate": 3.0693041752112564e-06, "loss": 0.1445, "step": 43584 }, { "epoch": 31.93040293040293, "grad_norm": 0.640625, "learning_rate": 3.068323962219231e-06, "loss": 0.1049, "step": 43585 }, { "epoch": 31.93113553113553, "grad_norm": 1.1640625, "learning_rate": 3.0673439764879224e-06, "loss": 0.095, "step": 43586 }, { "epoch": 31.931868131868132, "grad_norm": 1.4140625, "learning_rate": 3.0663642180220512e-06, "loss": 0.1105, "step": 43587 }, { "epoch": 31.932600732600733, "grad_norm": 1.3125, "learning_rate": 3.0653846868262942e-06, "loss": 0.1215, "step": 43588 }, { "epoch": 31.933333333333334, "grad_norm": 0.8828125, "learning_rate": 3.06440538290535e-06, "loss": 0.0986, "step": 43589 }, { "epoch": 31.934065934065934, "grad_norm": 2.28125, "learning_rate": 3.0634263062639068e-06, "loss": 0.1305, "step": 43590 }, { "epoch": 31.934798534798535, "grad_norm": 1.1171875, "learning_rate": 3.062447456906668e-06, "loss": 0.1439, "step": 43591 }, { "epoch": 31.935531135531136, "grad_norm": 0.8984375, "learning_rate": 3.061468834838317e-06, "loss": 0.0933, "step": 43592 }, { "epoch": 31.936263736263736, "grad_norm": 1.203125, "learning_rate": 3.0604904400635412e-06, "loss": 0.126, "step": 43593 }, { "epoch": 31.936996336996337, "grad_norm": 0.91796875, "learning_rate": 3.0595122725870224e-06, "loss": 0.1301, "step": 43594 }, { "epoch": 31.937728937728938, "grad_norm": 0.79296875, "learning_rate": 3.058534332413459e-06, "loss": 0.0953, "step": 43595 }, { "epoch": 31.93846153846154, "grad_norm": 1.1171875, "learning_rate": 3.0575566195475344e-06, "loss": 0.1253, "step": 43596 }, { "epoch": 31.93919413919414, "grad_norm": 1.171875, "learning_rate": 3.0565791339939242e-06, "loss": 0.1212, "step": 43597 }, { "epoch": 31.93992673992674, "grad_norm": 1.140625, "learning_rate": 3.0556018757573285e-06, "loss": 0.1605, "step": 43598 }, { "epoch": 31.94065934065934, "grad_norm": 1.1796875, "learning_rate": 3.054624844842412e-06, "loss": 0.1389, "step": 43599 }, { "epoch": 31.94139194139194, "grad_norm": 1.0625, "learning_rate": 3.053648041253857e-06, "loss": 0.1029, "step": 43600 }, { "epoch": 31.942124542124542, "grad_norm": 0.71875, "learning_rate": 3.052671464996357e-06, "loss": 0.1103, "step": 43601 }, { "epoch": 31.942857142857143, "grad_norm": 1.28125, "learning_rate": 3.0516951160745834e-06, "loss": 0.1316, "step": 43602 }, { "epoch": 31.943589743589744, "grad_norm": 1.2734375, "learning_rate": 3.050718994493218e-06, "loss": 0.1395, "step": 43603 }, { "epoch": 31.944322344322345, "grad_norm": 0.8046875, "learning_rate": 3.049743100256927e-06, "loss": 0.1029, "step": 43604 }, { "epoch": 31.945054945054945, "grad_norm": 0.95703125, "learning_rate": 3.048767433370398e-06, "loss": 0.1199, "step": 43605 }, { "epoch": 31.945787545787546, "grad_norm": 1.1171875, "learning_rate": 3.0477919938382974e-06, "loss": 0.135, "step": 43606 }, { "epoch": 31.946520146520147, "grad_norm": 1.234375, "learning_rate": 3.0468167816653118e-06, "loss": 0.144, "step": 43607 }, { "epoch": 31.947252747252747, "grad_norm": 2.03125, "learning_rate": 3.0458417968560963e-06, "loss": 0.1108, "step": 43608 }, { "epoch": 31.947985347985348, "grad_norm": 0.8828125, "learning_rate": 3.0448670394153395e-06, "loss": 0.102, "step": 43609 }, { "epoch": 31.94871794871795, "grad_norm": 0.8671875, "learning_rate": 3.0438925093476893e-06, "loss": 0.1348, "step": 43610 }, { "epoch": 31.94945054945055, "grad_norm": 0.78125, "learning_rate": 3.0429182066578455e-06, "loss": 0.1162, "step": 43611 }, { "epoch": 31.95018315018315, "grad_norm": 1.0703125, "learning_rate": 3.041944131350446e-06, "loss": 0.1447, "step": 43612 }, { "epoch": 31.95091575091575, "grad_norm": 1.0625, "learning_rate": 3.0409702834301844e-06, "loss": 0.1269, "step": 43613 }, { "epoch": 31.951648351648352, "grad_norm": 1.2734375, "learning_rate": 3.039996662901704e-06, "loss": 0.1176, "step": 43614 }, { "epoch": 31.952380952380953, "grad_norm": 0.77734375, "learning_rate": 3.039023269769682e-06, "loss": 0.1054, "step": 43615 }, { "epoch": 31.953113553113553, "grad_norm": 1.3359375, "learning_rate": 3.038050104038778e-06, "loss": 0.1353, "step": 43616 }, { "epoch": 31.953846153846154, "grad_norm": 0.734375, "learning_rate": 3.0370771657136585e-06, "loss": 0.1215, "step": 43617 }, { "epoch": 31.954578754578755, "grad_norm": 0.8125, "learning_rate": 3.0361044547989887e-06, "loss": 0.1084, "step": 43618 }, { "epoch": 31.955311355311355, "grad_norm": 1.15625, "learning_rate": 3.0351319712994185e-06, "loss": 0.1174, "step": 43619 }, { "epoch": 31.956043956043956, "grad_norm": 1.3203125, "learning_rate": 3.0341597152196075e-06, "loss": 0.1279, "step": 43620 }, { "epoch": 31.956776556776557, "grad_norm": 1.2578125, "learning_rate": 3.0331876865642273e-06, "loss": 0.1137, "step": 43621 }, { "epoch": 31.957509157509158, "grad_norm": 1.234375, "learning_rate": 3.0322158853379275e-06, "loss": 0.1029, "step": 43622 }, { "epoch": 31.95824175824176, "grad_norm": 1.1953125, "learning_rate": 3.0312443115453624e-06, "loss": 0.1236, "step": 43623 }, { "epoch": 31.95897435897436, "grad_norm": 1.3515625, "learning_rate": 3.0302729651911924e-06, "loss": 0.1148, "step": 43624 }, { "epoch": 31.95970695970696, "grad_norm": 1.125, "learning_rate": 3.0293018462800614e-06, "loss": 0.1507, "step": 43625 }, { "epoch": 31.96043956043956, "grad_norm": 1.046875, "learning_rate": 3.0283309548166293e-06, "loss": 0.1189, "step": 43626 }, { "epoch": 31.96117216117216, "grad_norm": 1.078125, "learning_rate": 3.027360290805546e-06, "loss": 0.1205, "step": 43627 }, { "epoch": 31.961904761904762, "grad_norm": 0.765625, "learning_rate": 3.0263898542514656e-06, "loss": 0.1227, "step": 43628 }, { "epoch": 31.962637362637363, "grad_norm": 1.09375, "learning_rate": 3.0254196451590434e-06, "loss": 0.1363, "step": 43629 }, { "epoch": 31.963369963369964, "grad_norm": 0.9375, "learning_rate": 3.024449663532912e-06, "loss": 0.1127, "step": 43630 }, { "epoch": 31.964102564102564, "grad_norm": 0.91796875, "learning_rate": 3.023479909377726e-06, "loss": 0.1118, "step": 43631 }, { "epoch": 31.964835164835165, "grad_norm": 0.85546875, "learning_rate": 3.0225103826981403e-06, "loss": 0.1121, "step": 43632 }, { "epoch": 31.965567765567766, "grad_norm": 1.03125, "learning_rate": 3.021541083498794e-06, "loss": 0.1338, "step": 43633 }, { "epoch": 31.966300366300366, "grad_norm": 0.8828125, "learning_rate": 3.020572011784324e-06, "loss": 0.1161, "step": 43634 }, { "epoch": 31.967032967032967, "grad_norm": 1.3515625, "learning_rate": 3.0196031675593857e-06, "loss": 0.1217, "step": 43635 }, { "epoch": 31.967765567765568, "grad_norm": 1.15625, "learning_rate": 3.0186345508286066e-06, "loss": 0.1137, "step": 43636 }, { "epoch": 31.96849816849817, "grad_norm": 1.1484375, "learning_rate": 3.017666161596647e-06, "loss": 0.1387, "step": 43637 }, { "epoch": 31.96923076923077, "grad_norm": 0.91015625, "learning_rate": 3.0166979998681278e-06, "loss": 0.1352, "step": 43638 }, { "epoch": 31.96996336996337, "grad_norm": 1.03125, "learning_rate": 3.015730065647705e-06, "loss": 0.1293, "step": 43639 }, { "epoch": 31.97069597069597, "grad_norm": 1.0390625, "learning_rate": 3.0147623589399994e-06, "loss": 0.1284, "step": 43640 }, { "epoch": 31.97142857142857, "grad_norm": 1.1875, "learning_rate": 3.013794879749661e-06, "loss": 0.1411, "step": 43641 }, { "epoch": 31.972161172161172, "grad_norm": 1.578125, "learning_rate": 3.0128276280813173e-06, "loss": 0.1552, "step": 43642 }, { "epoch": 31.972893772893773, "grad_norm": 0.90234375, "learning_rate": 3.0118606039396e-06, "loss": 0.1098, "step": 43643 }, { "epoch": 31.973626373626374, "grad_norm": 0.6640625, "learning_rate": 3.010893807329159e-06, "loss": 0.1141, "step": 43644 }, { "epoch": 31.974358974358974, "grad_norm": 0.74609375, "learning_rate": 3.009927238254611e-06, "loss": 0.1205, "step": 43645 }, { "epoch": 31.975091575091575, "grad_norm": 1.1875, "learning_rate": 3.0089608967205882e-06, "loss": 0.1093, "step": 43646 }, { "epoch": 31.975824175824176, "grad_norm": 0.9765625, "learning_rate": 3.007994782731724e-06, "loss": 0.1232, "step": 43647 }, { "epoch": 31.976556776556777, "grad_norm": 1.15625, "learning_rate": 3.007028896292656e-06, "loss": 0.1119, "step": 43648 }, { "epoch": 31.977289377289377, "grad_norm": 1.1796875, "learning_rate": 3.0060632374079953e-06, "loss": 0.1201, "step": 43649 }, { "epoch": 31.978021978021978, "grad_norm": 1.28125, "learning_rate": 3.0050978060823855e-06, "loss": 0.1626, "step": 43650 }, { "epoch": 31.97875457875458, "grad_norm": 0.93359375, "learning_rate": 3.0041326023204326e-06, "loss": 0.0944, "step": 43651 }, { "epoch": 31.97948717948718, "grad_norm": 1.265625, "learning_rate": 3.0031676261267796e-06, "loss": 0.1318, "step": 43652 }, { "epoch": 31.98021978021978, "grad_norm": 1.2265625, "learning_rate": 3.002202877506038e-06, "loss": 0.1327, "step": 43653 }, { "epoch": 31.98095238095238, "grad_norm": 0.82421875, "learning_rate": 3.001238356462834e-06, "loss": 0.1083, "step": 43654 }, { "epoch": 31.98168498168498, "grad_norm": 1.59375, "learning_rate": 3.0002740630017965e-06, "loss": 0.1684, "step": 43655 }, { "epoch": 31.982417582417582, "grad_norm": 0.94140625, "learning_rate": 2.9993099971275343e-06, "loss": 0.1073, "step": 43656 }, { "epoch": 31.983150183150183, "grad_norm": 1.1640625, "learning_rate": 2.998346158844671e-06, "loss": 0.1497, "step": 43657 }, { "epoch": 31.983882783882784, "grad_norm": 0.6640625, "learning_rate": 2.9973825481578278e-06, "loss": 0.0897, "step": 43658 }, { "epoch": 31.984615384615385, "grad_norm": 1.2265625, "learning_rate": 2.996419165071621e-06, "loss": 0.1255, "step": 43659 }, { "epoch": 31.985347985347985, "grad_norm": 1.140625, "learning_rate": 2.9954560095906555e-06, "loss": 0.1133, "step": 43660 }, { "epoch": 31.986080586080586, "grad_norm": 1.2578125, "learning_rate": 2.9944930817195702e-06, "loss": 0.0898, "step": 43661 }, { "epoch": 31.986813186813187, "grad_norm": 1.1640625, "learning_rate": 2.9935303814629427e-06, "loss": 0.1087, "step": 43662 }, { "epoch": 31.987545787545788, "grad_norm": 1.328125, "learning_rate": 2.992567908825422e-06, "loss": 0.1175, "step": 43663 }, { "epoch": 31.98827838827839, "grad_norm": 0.796875, "learning_rate": 2.9916056638115975e-06, "loss": 0.0966, "step": 43664 }, { "epoch": 31.98901098901099, "grad_norm": 1.1875, "learning_rate": 2.990643646426096e-06, "loss": 0.1427, "step": 43665 }, { "epoch": 31.98974358974359, "grad_norm": 1.921875, "learning_rate": 2.9896818566735066e-06, "loss": 0.124, "step": 43666 }, { "epoch": 31.99047619047619, "grad_norm": 0.984375, "learning_rate": 2.988720294558451e-06, "loss": 0.1155, "step": 43667 }, { "epoch": 31.99120879120879, "grad_norm": 1.09375, "learning_rate": 2.987758960085529e-06, "loss": 0.0998, "step": 43668 }, { "epoch": 31.991941391941392, "grad_norm": 0.84765625, "learning_rate": 2.9867978532593573e-06, "loss": 0.1244, "step": 43669 }, { "epoch": 31.992673992673993, "grad_norm": 0.84375, "learning_rate": 2.9858369740845347e-06, "loss": 0.1426, "step": 43670 }, { "epoch": 31.993406593406593, "grad_norm": 1.0390625, "learning_rate": 2.984876322565662e-06, "loss": 0.1469, "step": 43671 }, { "epoch": 31.994139194139194, "grad_norm": 0.89453125, "learning_rate": 2.98391589870735e-06, "loss": 0.1225, "step": 43672 }, { "epoch": 31.994871794871795, "grad_norm": 1.2890625, "learning_rate": 2.982955702514187e-06, "loss": 0.1293, "step": 43673 }, { "epoch": 31.995604395604396, "grad_norm": 1.5390625, "learning_rate": 2.9819957339907898e-06, "loss": 0.1111, "step": 43674 }, { "epoch": 31.996336996336996, "grad_norm": 1.0703125, "learning_rate": 2.9810359931417462e-06, "loss": 0.1175, "step": 43675 }, { "epoch": 31.997069597069597, "grad_norm": 1.7109375, "learning_rate": 2.980076479971663e-06, "loss": 0.1438, "step": 43676 }, { "epoch": 31.997802197802198, "grad_norm": 0.84765625, "learning_rate": 2.979117194485128e-06, "loss": 0.124, "step": 43677 }, { "epoch": 31.9985347985348, "grad_norm": 0.89453125, "learning_rate": 2.978158136686747e-06, "loss": 0.1269, "step": 43678 }, { "epoch": 31.9992673992674, "grad_norm": 1.2578125, "learning_rate": 2.977199306581103e-06, "loss": 0.0913, "step": 43679 }, { "epoch": 32.0, "grad_norm": 1.1484375, "learning_rate": 2.9762407041728017e-06, "loss": 0.1556, "step": 43680 }, { "epoch": 32.000732600732604, "grad_norm": 1.421875, "learning_rate": 2.975282329466437e-06, "loss": 0.1125, "step": 43681 }, { "epoch": 32.0014652014652, "grad_norm": 0.78125, "learning_rate": 2.9743241824665922e-06, "loss": 0.0984, "step": 43682 }, { "epoch": 32.002197802197806, "grad_norm": 1.1953125, "learning_rate": 2.9733662631778624e-06, "loss": 0.1213, "step": 43683 }, { "epoch": 32.0029304029304, "grad_norm": 1.3515625, "learning_rate": 2.9724085716048357e-06, "loss": 0.1572, "step": 43684 }, { "epoch": 32.00366300366301, "grad_norm": 1.0, "learning_rate": 2.9714511077521064e-06, "loss": 0.1385, "step": 43685 }, { "epoch": 32.004395604395604, "grad_norm": 1.078125, "learning_rate": 2.970493871624253e-06, "loss": 0.1109, "step": 43686 }, { "epoch": 32.00512820512821, "grad_norm": 1.3359375, "learning_rate": 2.969536863225869e-06, "loss": 0.1478, "step": 43687 }, { "epoch": 32.005860805860806, "grad_norm": 1.0546875, "learning_rate": 2.9685800825615276e-06, "loss": 0.1236, "step": 43688 }, { "epoch": 32.00659340659341, "grad_norm": 0.5546875, "learning_rate": 2.967623529635833e-06, "loss": 0.1027, "step": 43689 }, { "epoch": 32.00732600732601, "grad_norm": 0.90625, "learning_rate": 2.966667204453353e-06, "loss": 0.1339, "step": 43690 }, { "epoch": 32.00805860805861, "grad_norm": 1.2890625, "learning_rate": 2.9657111070186817e-06, "loss": 0.1081, "step": 43691 }, { "epoch": 32.00879120879121, "grad_norm": 0.6796875, "learning_rate": 2.964755237336386e-06, "loss": 0.1171, "step": 43692 }, { "epoch": 32.00952380952381, "grad_norm": 1.015625, "learning_rate": 2.9637995954110543e-06, "loss": 0.12, "step": 43693 }, { "epoch": 32.01025641025641, "grad_norm": 0.79296875, "learning_rate": 2.962844181247265e-06, "loss": 0.1011, "step": 43694 }, { "epoch": 32.010989010989015, "grad_norm": 0.96484375, "learning_rate": 2.9618889948495898e-06, "loss": 0.1046, "step": 43695 }, { "epoch": 32.01172161172161, "grad_norm": 1.1796875, "learning_rate": 2.9609340362226232e-06, "loss": 0.124, "step": 43696 }, { "epoch": 32.012454212454216, "grad_norm": 1.0859375, "learning_rate": 2.9599793053709164e-06, "loss": 0.1188, "step": 43697 }, { "epoch": 32.01318681318681, "grad_norm": 1.453125, "learning_rate": 2.9590248022990623e-06, "loss": 0.179, "step": 43698 }, { "epoch": 32.01391941391942, "grad_norm": 0.91796875, "learning_rate": 2.958070527011613e-06, "loss": 0.1027, "step": 43699 }, { "epoch": 32.014652014652015, "grad_norm": 0.71875, "learning_rate": 2.9571164795131666e-06, "loss": 0.0876, "step": 43700 }, { "epoch": 32.01538461538462, "grad_norm": 0.71484375, "learning_rate": 2.95616265980828e-06, "loss": 0.115, "step": 43701 }, { "epoch": 32.016117216117216, "grad_norm": 1.78125, "learning_rate": 2.9552090679015307e-06, "loss": 0.1576, "step": 43702 }, { "epoch": 32.01684981684982, "grad_norm": 0.9921875, "learning_rate": 2.9542557037974745e-06, "loss": 0.1279, "step": 43703 }, { "epoch": 32.01758241758242, "grad_norm": 1.0546875, "learning_rate": 2.9533025675006895e-06, "loss": 0.1095, "step": 43704 }, { "epoch": 32.01831501831502, "grad_norm": 0.70703125, "learning_rate": 2.952349659015743e-06, "loss": 0.1281, "step": 43705 }, { "epoch": 32.01904761904762, "grad_norm": 1.1171875, "learning_rate": 2.9513969783471946e-06, "loss": 0.1158, "step": 43706 }, { "epoch": 32.01978021978022, "grad_norm": 0.71875, "learning_rate": 2.9504445254996188e-06, "loss": 0.1049, "step": 43707 }, { "epoch": 32.02051282051282, "grad_norm": 1.2421875, "learning_rate": 2.9494923004775643e-06, "loss": 0.1407, "step": 43708 }, { "epoch": 32.021245421245425, "grad_norm": 1.2265625, "learning_rate": 2.9485403032856043e-06, "loss": 0.1344, "step": 43709 }, { "epoch": 32.02197802197802, "grad_norm": 1.171875, "learning_rate": 2.9475885339282996e-06, "loss": 0.1213, "step": 43710 }, { "epoch": 32.022710622710626, "grad_norm": 1.21875, "learning_rate": 2.9466369924102124e-06, "loss": 0.1733, "step": 43711 }, { "epoch": 32.02344322344322, "grad_norm": 1.3125, "learning_rate": 2.9456856787358924e-06, "loss": 0.1278, "step": 43712 }, { "epoch": 32.02417582417583, "grad_norm": 0.71875, "learning_rate": 2.9447345929099117e-06, "loss": 0.1154, "step": 43713 }, { "epoch": 32.024908424908425, "grad_norm": 0.9609375, "learning_rate": 2.943783734936805e-06, "loss": 0.109, "step": 43714 }, { "epoch": 32.02564102564103, "grad_norm": 1.21875, "learning_rate": 2.9428331048211555e-06, "loss": 0.095, "step": 43715 }, { "epoch": 32.026373626373626, "grad_norm": 2.109375, "learning_rate": 2.9418827025674963e-06, "loss": 0.1181, "step": 43716 }, { "epoch": 32.02710622710623, "grad_norm": 0.984375, "learning_rate": 2.940932528180389e-06, "loss": 0.1398, "step": 43717 }, { "epoch": 32.02783882783883, "grad_norm": 1.3828125, "learning_rate": 2.9399825816643957e-06, "loss": 0.1352, "step": 43718 }, { "epoch": 32.02857142857143, "grad_norm": 0.80859375, "learning_rate": 2.9390328630240494e-06, "loss": 0.1054, "step": 43719 }, { "epoch": 32.02930402930403, "grad_norm": 0.625, "learning_rate": 2.938083372263912e-06, "loss": 0.1035, "step": 43720 }, { "epoch": 32.03003663003663, "grad_norm": 0.7734375, "learning_rate": 2.9371341093885338e-06, "loss": 0.1094, "step": 43721 }, { "epoch": 32.03076923076923, "grad_norm": 1.359375, "learning_rate": 2.936185074402459e-06, "loss": 0.1273, "step": 43722 }, { "epoch": 32.031501831501835, "grad_norm": 0.78515625, "learning_rate": 2.9352362673102337e-06, "loss": 0.1083, "step": 43723 }, { "epoch": 32.03223443223443, "grad_norm": 0.94921875, "learning_rate": 2.9342876881164126e-06, "loss": 0.1281, "step": 43724 }, { "epoch": 32.032967032967036, "grad_norm": 0.71875, "learning_rate": 2.933339336825519e-06, "loss": 0.1018, "step": 43725 }, { "epoch": 32.033699633699634, "grad_norm": 0.94921875, "learning_rate": 2.9323912134421252e-06, "loss": 0.1249, "step": 43726 }, { "epoch": 32.03443223443224, "grad_norm": 1.328125, "learning_rate": 2.93144331797076e-06, "loss": 0.0999, "step": 43727 }, { "epoch": 32.035164835164835, "grad_norm": 0.83203125, "learning_rate": 2.930495650415962e-06, "loss": 0.1225, "step": 43728 }, { "epoch": 32.03589743589744, "grad_norm": 0.86328125, "learning_rate": 2.929548210782277e-06, "loss": 0.1261, "step": 43729 }, { "epoch": 32.03663003663004, "grad_norm": 0.640625, "learning_rate": 2.928600999074238e-06, "loss": 0.1359, "step": 43730 }, { "epoch": 32.03736263736264, "grad_norm": 1.03125, "learning_rate": 2.9276540152963905e-06, "loss": 0.11, "step": 43731 }, { "epoch": 32.03809523809524, "grad_norm": 0.87890625, "learning_rate": 2.926707259453268e-06, "loss": 0.1143, "step": 43732 }, { "epoch": 32.03882783882784, "grad_norm": 0.94140625, "learning_rate": 2.925760731549415e-06, "loss": 0.1202, "step": 43733 }, { "epoch": 32.03956043956044, "grad_norm": 1.0859375, "learning_rate": 2.9248144315893552e-06, "loss": 0.1086, "step": 43734 }, { "epoch": 32.040293040293044, "grad_norm": 1.1171875, "learning_rate": 2.923868359577628e-06, "loss": 0.1168, "step": 43735 }, { "epoch": 32.04102564102564, "grad_norm": 1.8046875, "learning_rate": 2.9229225155187662e-06, "loss": 0.1151, "step": 43736 }, { "epoch": 32.041758241758245, "grad_norm": 1.2890625, "learning_rate": 2.9219768994173046e-06, "loss": 0.1197, "step": 43737 }, { "epoch": 32.04249084249084, "grad_norm": 0.7109375, "learning_rate": 2.921031511277766e-06, "loss": 0.1217, "step": 43738 }, { "epoch": 32.04322344322345, "grad_norm": 1.2421875, "learning_rate": 2.9200863511046894e-06, "loss": 0.1186, "step": 43739 }, { "epoch": 32.043956043956044, "grad_norm": 0.98828125, "learning_rate": 2.919141418902587e-06, "loss": 0.1219, "step": 43740 }, { "epoch": 32.04468864468865, "grad_norm": 0.953125, "learning_rate": 2.918196714676009e-06, "loss": 0.1161, "step": 43741 }, { "epoch": 32.045421245421245, "grad_norm": 1.203125, "learning_rate": 2.917252238429462e-06, "loss": 0.1095, "step": 43742 }, { "epoch": 32.04615384615385, "grad_norm": 0.80859375, "learning_rate": 2.9163079901674854e-06, "loss": 0.1288, "step": 43743 }, { "epoch": 32.04688644688645, "grad_norm": 1.0, "learning_rate": 2.9153639698946016e-06, "loss": 0.1293, "step": 43744 }, { "epoch": 32.04761904761905, "grad_norm": 1.2265625, "learning_rate": 2.9144201776153227e-06, "loss": 0.1122, "step": 43745 }, { "epoch": 32.04835164835165, "grad_norm": 1.34375, "learning_rate": 2.9134766133341828e-06, "loss": 0.1405, "step": 43746 }, { "epoch": 32.04908424908425, "grad_norm": 0.921875, "learning_rate": 2.9125332770556937e-06, "loss": 0.1266, "step": 43747 }, { "epoch": 32.04981684981685, "grad_norm": 1.1640625, "learning_rate": 2.911590168784389e-06, "loss": 0.1115, "step": 43748 }, { "epoch": 32.050549450549454, "grad_norm": 1.0078125, "learning_rate": 2.910647288524765e-06, "loss": 0.1246, "step": 43749 }, { "epoch": 32.05128205128205, "grad_norm": 0.66015625, "learning_rate": 2.909704636281365e-06, "loss": 0.0965, "step": 43750 }, { "epoch": 32.052014652014655, "grad_norm": 0.8515625, "learning_rate": 2.9087622120586805e-06, "loss": 0.1227, "step": 43751 }, { "epoch": 32.05274725274725, "grad_norm": 0.8515625, "learning_rate": 2.9078200158612498e-06, "loss": 0.1139, "step": 43752 }, { "epoch": 32.05347985347986, "grad_norm": 1.421875, "learning_rate": 2.906878047693569e-06, "loss": 0.1216, "step": 43753 }, { "epoch": 32.054212454212454, "grad_norm": 1.171875, "learning_rate": 2.9059363075601714e-06, "loss": 0.1257, "step": 43754 }, { "epoch": 32.05494505494506, "grad_norm": 0.76953125, "learning_rate": 2.904994795465547e-06, "loss": 0.1155, "step": 43755 }, { "epoch": 32.055677655677655, "grad_norm": 1.3359375, "learning_rate": 2.904053511414214e-06, "loss": 0.1498, "step": 43756 }, { "epoch": 32.05641025641026, "grad_norm": 0.984375, "learning_rate": 2.9031124554106883e-06, "loss": 0.1155, "step": 43757 }, { "epoch": 32.05714285714286, "grad_norm": 0.76953125, "learning_rate": 2.9021716274594773e-06, "loss": 0.0972, "step": 43758 }, { "epoch": 32.05787545787546, "grad_norm": 0.81640625, "learning_rate": 2.9012310275650932e-06, "loss": 0.1143, "step": 43759 }, { "epoch": 32.05860805860806, "grad_norm": 1.0078125, "learning_rate": 2.900290655732031e-06, "loss": 0.0998, "step": 43760 }, { "epoch": 32.05934065934066, "grad_norm": 1.2578125, "learning_rate": 2.8993505119647967e-06, "loss": 0.1169, "step": 43761 }, { "epoch": 32.06007326007326, "grad_norm": 0.78515625, "learning_rate": 2.898410596267908e-06, "loss": 0.0995, "step": 43762 }, { "epoch": 32.060805860805864, "grad_norm": 1.0234375, "learning_rate": 2.8974709086458602e-06, "loss": 0.1162, "step": 43763 }, { "epoch": 32.06153846153846, "grad_norm": 1.0390625, "learning_rate": 2.8965314491031544e-06, "loss": 0.1231, "step": 43764 }, { "epoch": 32.062271062271066, "grad_norm": 0.90625, "learning_rate": 2.8955922176443023e-06, "loss": 0.1164, "step": 43765 }, { "epoch": 32.06300366300366, "grad_norm": 1.0390625, "learning_rate": 2.894653214273778e-06, "loss": 0.1191, "step": 43766 }, { "epoch": 32.06373626373627, "grad_norm": 0.78515625, "learning_rate": 2.8937144389961086e-06, "loss": 0.1276, "step": 43767 }, { "epoch": 32.064468864468864, "grad_norm": 1.015625, "learning_rate": 2.8927758918157795e-06, "loss": 0.1322, "step": 43768 }, { "epoch": 32.06520146520147, "grad_norm": 1.28125, "learning_rate": 2.8918375727372913e-06, "loss": 0.1205, "step": 43769 }, { "epoch": 32.065934065934066, "grad_norm": 1.140625, "learning_rate": 2.8908994817651395e-06, "loss": 0.1366, "step": 43770 }, { "epoch": 32.06666666666667, "grad_norm": 0.88671875, "learning_rate": 2.889961618903814e-06, "loss": 0.0909, "step": 43771 }, { "epoch": 32.06739926739927, "grad_norm": 0.71875, "learning_rate": 2.88902398415781e-06, "loss": 0.1164, "step": 43772 }, { "epoch": 32.06813186813187, "grad_norm": 0.98046875, "learning_rate": 2.8880865775316227e-06, "loss": 0.1102, "step": 43773 }, { "epoch": 32.06886446886447, "grad_norm": 1.1171875, "learning_rate": 2.8871493990297482e-06, "loss": 0.1412, "step": 43774 }, { "epoch": 32.06959706959707, "grad_norm": 0.65625, "learning_rate": 2.886212448656665e-06, "loss": 0.1367, "step": 43775 }, { "epoch": 32.07032967032967, "grad_norm": 0.90234375, "learning_rate": 2.8852757264168744e-06, "loss": 0.1087, "step": 43776 }, { "epoch": 32.071062271062274, "grad_norm": 1.21875, "learning_rate": 2.884339232314849e-06, "loss": 0.1189, "step": 43777 }, { "epoch": 32.07179487179487, "grad_norm": 0.91796875, "learning_rate": 2.883402966355096e-06, "loss": 0.104, "step": 43778 }, { "epoch": 32.072527472527476, "grad_norm": 1.125, "learning_rate": 2.8824669285420827e-06, "loss": 0.1332, "step": 43779 }, { "epoch": 32.07326007326007, "grad_norm": 0.89453125, "learning_rate": 2.8815311188803106e-06, "loss": 0.13, "step": 43780 }, { "epoch": 32.07399267399268, "grad_norm": 1.0390625, "learning_rate": 2.880595537374247e-06, "loss": 0.1257, "step": 43781 }, { "epoch": 32.074725274725274, "grad_norm": 1.4375, "learning_rate": 2.879660184028388e-06, "loss": 0.1196, "step": 43782 }, { "epoch": 32.07545787545788, "grad_norm": 0.94921875, "learning_rate": 2.8787250588472063e-06, "loss": 0.1047, "step": 43783 }, { "epoch": 32.076190476190476, "grad_norm": 0.65625, "learning_rate": 2.8777901618351868e-06, "loss": 0.1171, "step": 43784 }, { "epoch": 32.07692307692308, "grad_norm": 1.0, "learning_rate": 2.8768554929968085e-06, "loss": 0.151, "step": 43785 }, { "epoch": 32.07765567765568, "grad_norm": 0.97265625, "learning_rate": 2.87592105233655e-06, "loss": 0.1084, "step": 43786 }, { "epoch": 32.07838827838828, "grad_norm": 0.6796875, "learning_rate": 2.8749868398588844e-06, "loss": 0.1053, "step": 43787 }, { "epoch": 32.07912087912088, "grad_norm": 1.453125, "learning_rate": 2.8740528555682917e-06, "loss": 0.1376, "step": 43788 }, { "epoch": 32.07985347985348, "grad_norm": 0.6640625, "learning_rate": 2.873119099469255e-06, "loss": 0.0822, "step": 43789 }, { "epoch": 32.08058608058608, "grad_norm": 0.90625, "learning_rate": 2.8721855715662268e-06, "loss": 0.119, "step": 43790 }, { "epoch": 32.081318681318685, "grad_norm": 1.3203125, "learning_rate": 2.8712522718637073e-06, "loss": 0.1443, "step": 43791 }, { "epoch": 32.08205128205128, "grad_norm": 0.66796875, "learning_rate": 2.8703192003661372e-06, "loss": 0.114, "step": 43792 }, { "epoch": 32.082783882783886, "grad_norm": 0.83984375, "learning_rate": 2.8693863570780125e-06, "loss": 0.1157, "step": 43793 }, { "epoch": 32.08351648351648, "grad_norm": 1.109375, "learning_rate": 2.8684537420037946e-06, "loss": 0.1239, "step": 43794 }, { "epoch": 32.08424908424909, "grad_norm": 0.85546875, "learning_rate": 2.8675213551479467e-06, "loss": 0.1353, "step": 43795 }, { "epoch": 32.084981684981685, "grad_norm": 0.87890625, "learning_rate": 2.866589196514947e-06, "loss": 0.1261, "step": 43796 }, { "epoch": 32.08571428571429, "grad_norm": 1.25, "learning_rate": 2.865657266109248e-06, "loss": 0.1049, "step": 43797 }, { "epoch": 32.086446886446886, "grad_norm": 1.59375, "learning_rate": 2.864725563935327e-06, "loss": 0.105, "step": 43798 }, { "epoch": 32.08717948717949, "grad_norm": 1.0390625, "learning_rate": 2.8637940899976427e-06, "loss": 0.1184, "step": 43799 }, { "epoch": 32.08791208791209, "grad_norm": 1.3984375, "learning_rate": 2.862862844300662e-06, "loss": 0.1185, "step": 43800 }, { "epoch": 32.08864468864469, "grad_norm": 0.921875, "learning_rate": 2.861931826848837e-06, "loss": 0.1022, "step": 43801 }, { "epoch": 32.08937728937729, "grad_norm": 1.78125, "learning_rate": 2.8610010376466453e-06, "loss": 0.1457, "step": 43802 }, { "epoch": 32.09010989010989, "grad_norm": 0.98828125, "learning_rate": 2.8600704766985235e-06, "loss": 0.108, "step": 43803 }, { "epoch": 32.09084249084249, "grad_norm": 1.5234375, "learning_rate": 2.8591401440089495e-06, "loss": 0.1586, "step": 43804 }, { "epoch": 32.091575091575095, "grad_norm": 0.90625, "learning_rate": 2.858210039582375e-06, "loss": 0.1199, "step": 43805 }, { "epoch": 32.09230769230769, "grad_norm": 1.265625, "learning_rate": 2.8572801634232622e-06, "loss": 0.1212, "step": 43806 }, { "epoch": 32.093040293040296, "grad_norm": 1.1640625, "learning_rate": 2.8563505155360525e-06, "loss": 0.1282, "step": 43807 }, { "epoch": 32.09377289377289, "grad_norm": 1.3046875, "learning_rate": 2.855421095925207e-06, "loss": 0.1159, "step": 43808 }, { "epoch": 32.0945054945055, "grad_norm": 1.4453125, "learning_rate": 2.854491904595178e-06, "loss": 0.1335, "step": 43809 }, { "epoch": 32.095238095238095, "grad_norm": 0.8359375, "learning_rate": 2.8535629415504217e-06, "loss": 0.1193, "step": 43810 }, { "epoch": 32.0959706959707, "grad_norm": 0.84765625, "learning_rate": 2.852634206795391e-06, "loss": 0.1216, "step": 43811 }, { "epoch": 32.096703296703296, "grad_norm": 0.97265625, "learning_rate": 2.85170570033453e-06, "loss": 0.1033, "step": 43812 }, { "epoch": 32.0974358974359, "grad_norm": 1.125, "learning_rate": 2.8507774221722856e-06, "loss": 0.1147, "step": 43813 }, { "epoch": 32.0981684981685, "grad_norm": 0.9765625, "learning_rate": 2.8498493723131093e-06, "loss": 0.1266, "step": 43814 }, { "epoch": 32.0989010989011, "grad_norm": 0.93359375, "learning_rate": 2.8489215507614525e-06, "loss": 0.1251, "step": 43815 }, { "epoch": 32.0996336996337, "grad_norm": 1.4609375, "learning_rate": 2.84799395752175e-06, "loss": 0.1129, "step": 43816 }, { "epoch": 32.100366300366304, "grad_norm": 1.0703125, "learning_rate": 2.847066592598453e-06, "loss": 0.1264, "step": 43817 }, { "epoch": 32.1010989010989, "grad_norm": 1.5390625, "learning_rate": 2.8461394559959975e-06, "loss": 0.1174, "step": 43818 }, { "epoch": 32.101831501831505, "grad_norm": 1.0, "learning_rate": 2.8452125477188444e-06, "loss": 0.1195, "step": 43819 }, { "epoch": 32.1025641025641, "grad_norm": 0.859375, "learning_rate": 2.844285867771413e-06, "loss": 0.1536, "step": 43820 }, { "epoch": 32.10329670329671, "grad_norm": 1.015625, "learning_rate": 2.8433594161581494e-06, "loss": 0.1409, "step": 43821 }, { "epoch": 32.104029304029304, "grad_norm": 0.80078125, "learning_rate": 2.8424331928835047e-06, "loss": 0.1128, "step": 43822 }, { "epoch": 32.10476190476191, "grad_norm": 0.80078125, "learning_rate": 2.8415071979519032e-06, "loss": 0.0975, "step": 43823 }, { "epoch": 32.105494505494505, "grad_norm": 0.76171875, "learning_rate": 2.840581431367785e-06, "loss": 0.0872, "step": 43824 }, { "epoch": 32.10622710622711, "grad_norm": 0.8515625, "learning_rate": 2.839655893135585e-06, "loss": 0.1152, "step": 43825 }, { "epoch": 32.10695970695971, "grad_norm": 1.078125, "learning_rate": 2.8387305832597495e-06, "loss": 0.1443, "step": 43826 }, { "epoch": 32.10769230769231, "grad_norm": 1.515625, "learning_rate": 2.837805501744691e-06, "loss": 0.1291, "step": 43827 }, { "epoch": 32.10842490842491, "grad_norm": 1.484375, "learning_rate": 2.8368806485948615e-06, "loss": 0.1034, "step": 43828 }, { "epoch": 32.10915750915751, "grad_norm": 1.4375, "learning_rate": 2.8359560238146685e-06, "loss": 0.1243, "step": 43829 }, { "epoch": 32.10989010989011, "grad_norm": 1.15625, "learning_rate": 2.8350316274085736e-06, "loss": 0.1215, "step": 43830 }, { "epoch": 32.110622710622714, "grad_norm": 1.4140625, "learning_rate": 2.8341074593809795e-06, "loss": 0.1049, "step": 43831 }, { "epoch": 32.11135531135531, "grad_norm": 1.921875, "learning_rate": 2.8331835197363265e-06, "loss": 0.1439, "step": 43832 }, { "epoch": 32.112087912087915, "grad_norm": 1.0703125, "learning_rate": 2.832259808479044e-06, "loss": 0.116, "step": 43833 }, { "epoch": 32.11282051282051, "grad_norm": 1.296875, "learning_rate": 2.8313363256135503e-06, "loss": 0.1311, "step": 43834 }, { "epoch": 32.11355311355312, "grad_norm": 0.98046875, "learning_rate": 2.8304130711442697e-06, "loss": 0.1208, "step": 43835 }, { "epoch": 32.114285714285714, "grad_norm": 1.140625, "learning_rate": 2.8294900450756315e-06, "loss": 0.1144, "step": 43836 }, { "epoch": 32.11501831501832, "grad_norm": 1.3203125, "learning_rate": 2.82856724741206e-06, "loss": 0.1406, "step": 43837 }, { "epoch": 32.115750915750915, "grad_norm": 1.0390625, "learning_rate": 2.8276446781579673e-06, "loss": 0.1242, "step": 43838 }, { "epoch": 32.11648351648352, "grad_norm": 1.5234375, "learning_rate": 2.826722337317778e-06, "loss": 0.1337, "step": 43839 }, { "epoch": 32.11721611721612, "grad_norm": 1.0234375, "learning_rate": 2.8258002248959106e-06, "loss": 0.1211, "step": 43840 }, { "epoch": 32.11794871794872, "grad_norm": 1.1875, "learning_rate": 2.8248783408967945e-06, "loss": 0.1305, "step": 43841 }, { "epoch": 32.11868131868132, "grad_norm": 0.74609375, "learning_rate": 2.823956685324826e-06, "loss": 0.1352, "step": 43842 }, { "epoch": 32.11941391941392, "grad_norm": 0.9140625, "learning_rate": 2.8230352581844403e-06, "loss": 0.1097, "step": 43843 }, { "epoch": 32.12014652014652, "grad_norm": 1.3671875, "learning_rate": 2.822114059480033e-06, "loss": 0.1306, "step": 43844 }, { "epoch": 32.120879120879124, "grad_norm": 1.375, "learning_rate": 2.8211930892160404e-06, "loss": 0.1307, "step": 43845 }, { "epoch": 32.12161172161172, "grad_norm": 0.90625, "learning_rate": 2.8202723473968583e-06, "loss": 0.115, "step": 43846 }, { "epoch": 32.122344322344325, "grad_norm": 1.0625, "learning_rate": 2.819351834026905e-06, "loss": 0.1249, "step": 43847 }, { "epoch": 32.12307692307692, "grad_norm": 0.84375, "learning_rate": 2.8184315491105945e-06, "loss": 0.1248, "step": 43848 }, { "epoch": 32.12380952380953, "grad_norm": 1.375, "learning_rate": 2.8175114926523165e-06, "loss": 0.1366, "step": 43849 }, { "epoch": 32.124542124542124, "grad_norm": 1.359375, "learning_rate": 2.8165916646565057e-06, "loss": 0.1359, "step": 43850 }, { "epoch": 32.12527472527473, "grad_norm": 1.1484375, "learning_rate": 2.815672065127555e-06, "loss": 0.1434, "step": 43851 }, { "epoch": 32.126007326007326, "grad_norm": 0.80078125, "learning_rate": 2.814752694069881e-06, "loss": 0.1146, "step": 43852 }, { "epoch": 32.12673992673993, "grad_norm": 0.83984375, "learning_rate": 2.81383355148787e-06, "loss": 0.1086, "step": 43853 }, { "epoch": 32.12747252747253, "grad_norm": 0.94921875, "learning_rate": 2.8129146373859457e-06, "loss": 0.1191, "step": 43854 }, { "epoch": 32.12820512820513, "grad_norm": 1.0625, "learning_rate": 2.8119959517684886e-06, "loss": 0.1266, "step": 43855 }, { "epoch": 32.12893772893773, "grad_norm": 1.21875, "learning_rate": 2.811077494639928e-06, "loss": 0.1447, "step": 43856 }, { "epoch": 32.12967032967033, "grad_norm": 1.4140625, "learning_rate": 2.810159266004638e-06, "loss": 0.114, "step": 43857 }, { "epoch": 32.13040293040293, "grad_norm": 0.828125, "learning_rate": 2.8092412658670367e-06, "loss": 0.131, "step": 43858 }, { "epoch": 32.131135531135534, "grad_norm": 1.203125, "learning_rate": 2.808323494231522e-06, "loss": 0.1163, "step": 43859 }, { "epoch": 32.13186813186813, "grad_norm": 1.03125, "learning_rate": 2.807405951102478e-06, "loss": 0.1234, "step": 43860 }, { "epoch": 32.132600732600736, "grad_norm": 1.109375, "learning_rate": 2.806488636484307e-06, "loss": 0.1144, "step": 43861 }, { "epoch": 32.13333333333333, "grad_norm": 1.2421875, "learning_rate": 2.805571550381412e-06, "loss": 0.1453, "step": 43862 }, { "epoch": 32.13406593406594, "grad_norm": 1.2421875, "learning_rate": 2.804654692798183e-06, "loss": 0.1134, "step": 43863 }, { "epoch": 32.134798534798534, "grad_norm": 0.89453125, "learning_rate": 2.803738063739e-06, "loss": 0.133, "step": 43864 }, { "epoch": 32.13553113553114, "grad_norm": 1.3515625, "learning_rate": 2.80282166320827e-06, "loss": 0.1452, "step": 43865 }, { "epoch": 32.136263736263736, "grad_norm": 1.0546875, "learning_rate": 2.8019054912103796e-06, "loss": 0.1114, "step": 43866 }, { "epoch": 32.13699633699634, "grad_norm": 1.109375, "learning_rate": 2.8009895477497256e-06, "loss": 0.1106, "step": 43867 }, { "epoch": 32.13772893772894, "grad_norm": 1.3046875, "learning_rate": 2.8000738328306806e-06, "loss": 0.1685, "step": 43868 }, { "epoch": 32.13846153846154, "grad_norm": 1.0859375, "learning_rate": 2.7991583464576486e-06, "loss": 0.13, "step": 43869 }, { "epoch": 32.13919413919414, "grad_norm": 0.63671875, "learning_rate": 2.7982430886349916e-06, "loss": 0.1175, "step": 43870 }, { "epoch": 32.13992673992674, "grad_norm": 1.0703125, "learning_rate": 2.797328059367129e-06, "loss": 0.1361, "step": 43871 }, { "epoch": 32.14065934065934, "grad_norm": 0.87109375, "learning_rate": 2.7964132586584187e-06, "loss": 0.1146, "step": 43872 }, { "epoch": 32.141391941391944, "grad_norm": 1.53125, "learning_rate": 2.7954986865132514e-06, "loss": 0.1284, "step": 43873 }, { "epoch": 32.14212454212454, "grad_norm": 0.9609375, "learning_rate": 2.794584342936018e-06, "loss": 0.1342, "step": 43874 }, { "epoch": 32.142857142857146, "grad_norm": 1.0234375, "learning_rate": 2.793670227931082e-06, "loss": 0.1022, "step": 43875 }, { "epoch": 32.14358974358974, "grad_norm": 1.1484375, "learning_rate": 2.7927563415028404e-06, "loss": 0.13, "step": 43876 }, { "epoch": 32.14432234432235, "grad_norm": 0.87109375, "learning_rate": 2.7918426836556616e-06, "loss": 0.1213, "step": 43877 }, { "epoch": 32.145054945054945, "grad_norm": 0.76171875, "learning_rate": 2.7909292543939316e-06, "loss": 0.1284, "step": 43878 }, { "epoch": 32.14578754578755, "grad_norm": 0.98046875, "learning_rate": 2.790016053722014e-06, "loss": 0.1048, "step": 43879 }, { "epoch": 32.146520146520146, "grad_norm": 0.95703125, "learning_rate": 2.7891030816442987e-06, "loss": 0.1005, "step": 43880 }, { "epoch": 32.14725274725275, "grad_norm": 0.953125, "learning_rate": 2.7881903381651394e-06, "loss": 0.1623, "step": 43881 }, { "epoch": 32.14798534798535, "grad_norm": 0.5703125, "learning_rate": 2.787277823288938e-06, "loss": 0.0927, "step": 43882 }, { "epoch": 32.14871794871795, "grad_norm": 1.890625, "learning_rate": 2.786365537020046e-06, "loss": 0.1042, "step": 43883 }, { "epoch": 32.14945054945055, "grad_norm": 1.1484375, "learning_rate": 2.785453479362839e-06, "loss": 0.1349, "step": 43884 }, { "epoch": 32.15018315018315, "grad_norm": 0.70703125, "learning_rate": 2.784541650321691e-06, "loss": 0.0978, "step": 43885 }, { "epoch": 32.15091575091575, "grad_norm": 1.15625, "learning_rate": 2.7836300499009656e-06, "loss": 0.1252, "step": 43886 }, { "epoch": 32.151648351648355, "grad_norm": 1.0546875, "learning_rate": 2.7827186781050315e-06, "loss": 0.1119, "step": 43887 }, { "epoch": 32.15238095238095, "grad_norm": 1.4765625, "learning_rate": 2.781807534938253e-06, "loss": 0.1114, "step": 43888 }, { "epoch": 32.153113553113556, "grad_norm": 1.296875, "learning_rate": 2.7808966204050096e-06, "loss": 0.121, "step": 43889 }, { "epoch": 32.15384615384615, "grad_norm": 1.234375, "learning_rate": 2.779985934509648e-06, "loss": 0.1155, "step": 43890 }, { "epoch": 32.15457875457876, "grad_norm": 0.91015625, "learning_rate": 2.779075477256543e-06, "loss": 0.133, "step": 43891 }, { "epoch": 32.155311355311355, "grad_norm": 0.85546875, "learning_rate": 2.7781652486500475e-06, "loss": 0.109, "step": 43892 }, { "epoch": 32.15604395604396, "grad_norm": 0.75390625, "learning_rate": 2.7772552486945407e-06, "loss": 0.1128, "step": 43893 }, { "epoch": 32.156776556776556, "grad_norm": 0.96875, "learning_rate": 2.7763454773943596e-06, "loss": 0.0982, "step": 43894 }, { "epoch": 32.15750915750916, "grad_norm": 0.94921875, "learning_rate": 2.7754359347538833e-06, "loss": 0.1342, "step": 43895 }, { "epoch": 32.15824175824176, "grad_norm": 0.77734375, "learning_rate": 2.7745266207774427e-06, "loss": 0.094, "step": 43896 }, { "epoch": 32.15897435897436, "grad_norm": 0.6875, "learning_rate": 2.7736175354694292e-06, "loss": 0.1031, "step": 43897 }, { "epoch": 32.15970695970696, "grad_norm": 0.78125, "learning_rate": 2.7727086788341725e-06, "loss": 0.1188, "step": 43898 }, { "epoch": 32.16043956043956, "grad_norm": 1.171875, "learning_rate": 2.771800050876042e-06, "loss": 0.1512, "step": 43899 }, { "epoch": 32.16117216117216, "grad_norm": 1.21875, "learning_rate": 2.770891651599385e-06, "loss": 0.1525, "step": 43900 }, { "epoch": 32.161904761904765, "grad_norm": 0.96484375, "learning_rate": 2.7699834810085536e-06, "loss": 0.124, "step": 43901 }, { "epoch": 32.16263736263736, "grad_norm": 1.0, "learning_rate": 2.7690755391078953e-06, "loss": 0.1392, "step": 43902 }, { "epoch": 32.163369963369966, "grad_norm": 1.53125, "learning_rate": 2.7681678259017675e-06, "loss": 0.1153, "step": 43903 }, { "epoch": 32.16410256410256, "grad_norm": 1.0, "learning_rate": 2.767260341394523e-06, "loss": 0.1231, "step": 43904 }, { "epoch": 32.16483516483517, "grad_norm": 0.7890625, "learning_rate": 2.7663530855905034e-06, "loss": 0.105, "step": 43905 }, { "epoch": 32.165567765567765, "grad_norm": 1.0, "learning_rate": 2.765446058494056e-06, "loss": 0.1186, "step": 43906 }, { "epoch": 32.16630036630037, "grad_norm": 1.2109375, "learning_rate": 2.7645392601095164e-06, "loss": 0.1057, "step": 43907 }, { "epoch": 32.167032967032966, "grad_norm": 0.90625, "learning_rate": 2.763632690441248e-06, "loss": 0.1029, "step": 43908 }, { "epoch": 32.16776556776557, "grad_norm": 1.125, "learning_rate": 2.7627263494935814e-06, "loss": 0.0806, "step": 43909 }, { "epoch": 32.16849816849817, "grad_norm": 1.375, "learning_rate": 2.7618202372708697e-06, "loss": 0.1211, "step": 43910 }, { "epoch": 32.16923076923077, "grad_norm": 0.9140625, "learning_rate": 2.7609143537774483e-06, "loss": 0.1208, "step": 43911 }, { "epoch": 32.16996336996337, "grad_norm": 0.83984375, "learning_rate": 2.7600086990176536e-06, "loss": 0.0845, "step": 43912 }, { "epoch": 32.170695970695974, "grad_norm": 1.3828125, "learning_rate": 2.759103272995832e-06, "loss": 0.1389, "step": 43913 }, { "epoch": 32.17142857142857, "grad_norm": 0.8828125, "learning_rate": 2.758198075716315e-06, "loss": 0.1025, "step": 43914 }, { "epoch": 32.172161172161175, "grad_norm": 0.8125, "learning_rate": 2.7572931071834546e-06, "loss": 0.086, "step": 43915 }, { "epoch": 32.17289377289377, "grad_norm": 0.93359375, "learning_rate": 2.7563883674015646e-06, "loss": 0.1136, "step": 43916 }, { "epoch": 32.17362637362638, "grad_norm": 0.890625, "learning_rate": 2.7554838563749925e-06, "loss": 0.1299, "step": 43917 }, { "epoch": 32.174358974358974, "grad_norm": 1.15625, "learning_rate": 2.7545795741080743e-06, "loss": 0.1113, "step": 43918 }, { "epoch": 32.17509157509158, "grad_norm": 0.94921875, "learning_rate": 2.75367552060514e-06, "loss": 0.1198, "step": 43919 }, { "epoch": 32.175824175824175, "grad_norm": 1.4453125, "learning_rate": 2.752771695870515e-06, "loss": 0.1306, "step": 43920 }, { "epoch": 32.17655677655678, "grad_norm": 0.69921875, "learning_rate": 2.751868099908546e-06, "loss": 0.1199, "step": 43921 }, { "epoch": 32.17728937728938, "grad_norm": 1.2265625, "learning_rate": 2.750964732723537e-06, "loss": 0.1188, "step": 43922 }, { "epoch": 32.17802197802198, "grad_norm": 1.5859375, "learning_rate": 2.75006159431984e-06, "loss": 0.1541, "step": 43923 }, { "epoch": 32.17875457875458, "grad_norm": 0.8671875, "learning_rate": 2.7491586847017684e-06, "loss": 0.1069, "step": 43924 }, { "epoch": 32.17948717948718, "grad_norm": 0.78125, "learning_rate": 2.7482560038736537e-06, "loss": 0.1099, "step": 43925 }, { "epoch": 32.18021978021978, "grad_norm": 0.70703125, "learning_rate": 2.7473535518398314e-06, "loss": 0.1105, "step": 43926 }, { "epoch": 32.180952380952384, "grad_norm": 0.93359375, "learning_rate": 2.746451328604604e-06, "loss": 0.1228, "step": 43927 }, { "epoch": 32.18168498168498, "grad_norm": 0.796875, "learning_rate": 2.7455493341723037e-06, "loss": 0.0924, "step": 43928 }, { "epoch": 32.182417582417585, "grad_norm": 0.93359375, "learning_rate": 2.7446475685472537e-06, "loss": 0.121, "step": 43929 }, { "epoch": 32.18315018315018, "grad_norm": 1.96875, "learning_rate": 2.7437460317337807e-06, "loss": 0.1461, "step": 43930 }, { "epoch": 32.18388278388279, "grad_norm": 1.0546875, "learning_rate": 2.7428447237361923e-06, "loss": 0.1313, "step": 43931 }, { "epoch": 32.184615384615384, "grad_norm": 0.92578125, "learning_rate": 2.741943644558819e-06, "loss": 0.1209, "step": 43932 }, { "epoch": 32.18534798534799, "grad_norm": 1.3203125, "learning_rate": 2.7410427942059583e-06, "loss": 0.1554, "step": 43933 }, { "epoch": 32.186080586080585, "grad_norm": 1.03125, "learning_rate": 2.7401421726819516e-06, "loss": 0.1218, "step": 43934 }, { "epoch": 32.18681318681319, "grad_norm": 1.3125, "learning_rate": 2.7392417799910976e-06, "loss": 0.1321, "step": 43935 }, { "epoch": 32.18754578754579, "grad_norm": 0.890625, "learning_rate": 2.738341616137715e-06, "loss": 0.1117, "step": 43936 }, { "epoch": 32.18827838827839, "grad_norm": 1.265625, "learning_rate": 2.7374416811261234e-06, "loss": 0.1252, "step": 43937 }, { "epoch": 32.18901098901099, "grad_norm": 1.078125, "learning_rate": 2.73654197496062e-06, "loss": 0.1025, "step": 43938 }, { "epoch": 32.18974358974359, "grad_norm": 1.25, "learning_rate": 2.7356424976455247e-06, "loss": 0.1545, "step": 43939 }, { "epoch": 32.19047619047619, "grad_norm": 1.109375, "learning_rate": 2.7347432491851465e-06, "loss": 0.1321, "step": 43940 }, { "epoch": 32.191208791208794, "grad_norm": 1.203125, "learning_rate": 2.7338442295837986e-06, "loss": 0.1252, "step": 43941 }, { "epoch": 32.19194139194139, "grad_norm": 0.9296875, "learning_rate": 2.7329454388457798e-06, "loss": 0.1124, "step": 43942 }, { "epoch": 32.192673992673996, "grad_norm": 0.859375, "learning_rate": 2.732046876975397e-06, "loss": 0.0978, "step": 43943 }, { "epoch": 32.19340659340659, "grad_norm": 0.94921875, "learning_rate": 2.73114854397696e-06, "loss": 0.1257, "step": 43944 }, { "epoch": 32.1941391941392, "grad_norm": 1.296875, "learning_rate": 2.7302504398547823e-06, "loss": 0.1407, "step": 43945 }, { "epoch": 32.194871794871794, "grad_norm": 1.109375, "learning_rate": 2.729352564613145e-06, "loss": 0.1123, "step": 43946 }, { "epoch": 32.1956043956044, "grad_norm": 1.0, "learning_rate": 2.7284549182563684e-06, "loss": 0.127, "step": 43947 }, { "epoch": 32.196336996336996, "grad_norm": 1.171875, "learning_rate": 2.7275575007887375e-06, "loss": 0.1206, "step": 43948 }, { "epoch": 32.1970695970696, "grad_norm": 1.2578125, "learning_rate": 2.726660312214568e-06, "loss": 0.1468, "step": 43949 }, { "epoch": 32.1978021978022, "grad_norm": 0.96875, "learning_rate": 2.7257633525381512e-06, "loss": 0.1131, "step": 43950 }, { "epoch": 32.1985347985348, "grad_norm": 1.171875, "learning_rate": 2.724866621763784e-06, "loss": 0.1272, "step": 43951 }, { "epoch": 32.1992673992674, "grad_norm": 1.5546875, "learning_rate": 2.7239701198957706e-06, "loss": 0.1419, "step": 43952 }, { "epoch": 32.2, "grad_norm": 1.015625, "learning_rate": 2.723073846938392e-06, "loss": 0.108, "step": 43953 }, { "epoch": 32.2007326007326, "grad_norm": 0.87890625, "learning_rate": 2.722177802895956e-06, "loss": 0.1, "step": 43954 }, { "epoch": 32.201465201465204, "grad_norm": 0.8359375, "learning_rate": 2.7212819877727444e-06, "loss": 0.1091, "step": 43955 }, { "epoch": 32.2021978021978, "grad_norm": 1.1640625, "learning_rate": 2.720386401573065e-06, "loss": 0.0947, "step": 43956 }, { "epoch": 32.202930402930406, "grad_norm": 1.0859375, "learning_rate": 2.7194910443011944e-06, "loss": 0.125, "step": 43957 }, { "epoch": 32.203663003663, "grad_norm": 1.3828125, "learning_rate": 2.718595915961429e-06, "loss": 0.1184, "step": 43958 }, { "epoch": 32.20439560439561, "grad_norm": 0.984375, "learning_rate": 2.7177010165580453e-06, "loss": 0.1219, "step": 43959 }, { "epoch": 32.205128205128204, "grad_norm": 1.3515625, "learning_rate": 2.7168063460953568e-06, "loss": 0.1479, "step": 43960 }, { "epoch": 32.20586080586081, "grad_norm": 1.1484375, "learning_rate": 2.7159119045776283e-06, "loss": 0.1402, "step": 43961 }, { "epoch": 32.206593406593406, "grad_norm": 1.1328125, "learning_rate": 2.715017692009152e-06, "loss": 0.1032, "step": 43962 }, { "epoch": 32.20732600732601, "grad_norm": 0.62109375, "learning_rate": 2.7141237083942197e-06, "loss": 0.1031, "step": 43963 }, { "epoch": 32.20805860805861, "grad_norm": 0.69140625, "learning_rate": 2.713229953737096e-06, "loss": 0.1148, "step": 43964 }, { "epoch": 32.20879120879121, "grad_norm": 0.97265625, "learning_rate": 2.7123364280420843e-06, "loss": 0.1311, "step": 43965 }, { "epoch": 32.20952380952381, "grad_norm": 1.84375, "learning_rate": 2.711443131313449e-06, "loss": 0.1484, "step": 43966 }, { "epoch": 32.21025641025641, "grad_norm": 0.890625, "learning_rate": 2.7105500635554873e-06, "loss": 0.0925, "step": 43967 }, { "epoch": 32.21098901098901, "grad_norm": 0.94140625, "learning_rate": 2.7096572247724643e-06, "loss": 0.1256, "step": 43968 }, { "epoch": 32.211721611721615, "grad_norm": 1.0, "learning_rate": 2.7087646149686554e-06, "loss": 0.1309, "step": 43969 }, { "epoch": 32.21245421245421, "grad_norm": 0.65234375, "learning_rate": 2.707872234148353e-06, "loss": 0.1159, "step": 43970 }, { "epoch": 32.213186813186816, "grad_norm": 1.5859375, "learning_rate": 2.706980082315821e-06, "loss": 0.1155, "step": 43971 }, { "epoch": 32.21391941391941, "grad_norm": 0.87109375, "learning_rate": 2.706088159475335e-06, "loss": 0.0903, "step": 43972 }, { "epoch": 32.21465201465202, "grad_norm": 0.75390625, "learning_rate": 2.7051964656311714e-06, "loss": 0.1117, "step": 43973 }, { "epoch": 32.215384615384615, "grad_norm": 1.3515625, "learning_rate": 2.7043050007875996e-06, "loss": 0.1184, "step": 43974 }, { "epoch": 32.21611721611722, "grad_norm": 1.0078125, "learning_rate": 2.7034137649489007e-06, "loss": 0.1094, "step": 43975 }, { "epoch": 32.216849816849816, "grad_norm": 0.62890625, "learning_rate": 2.702522758119329e-06, "loss": 0.1001, "step": 43976 }, { "epoch": 32.21758241758242, "grad_norm": 1.1484375, "learning_rate": 2.7016319803031646e-06, "loss": 0.1076, "step": 43977 }, { "epoch": 32.21831501831502, "grad_norm": 1.1953125, "learning_rate": 2.7007414315046727e-06, "loss": 0.1222, "step": 43978 }, { "epoch": 32.21904761904762, "grad_norm": 0.77734375, "learning_rate": 2.6998511117281175e-06, "loss": 0.1001, "step": 43979 }, { "epoch": 32.21978021978022, "grad_norm": 1.0546875, "learning_rate": 2.6989610209777695e-06, "loss": 0.1278, "step": 43980 }, { "epoch": 32.22051282051282, "grad_norm": 0.7421875, "learning_rate": 2.698071159257893e-06, "loss": 0.1248, "step": 43981 }, { "epoch": 32.22124542124542, "grad_norm": 1.140625, "learning_rate": 2.697181526572747e-06, "loss": 0.1227, "step": 43982 }, { "epoch": 32.221978021978025, "grad_norm": 1.203125, "learning_rate": 2.696292122926597e-06, "loss": 0.1157, "step": 43983 }, { "epoch": 32.22271062271062, "grad_norm": 1.09375, "learning_rate": 2.6954029483237116e-06, "loss": 0.1342, "step": 43984 }, { "epoch": 32.223443223443226, "grad_norm": 0.99609375, "learning_rate": 2.6945140027683287e-06, "loss": 0.1479, "step": 43985 }, { "epoch": 32.22417582417582, "grad_norm": 1.2734375, "learning_rate": 2.693625286264735e-06, "loss": 0.1611, "step": 43986 }, { "epoch": 32.22490842490843, "grad_norm": 1.40625, "learning_rate": 2.692736798817173e-06, "loss": 0.1319, "step": 43987 }, { "epoch": 32.225641025641025, "grad_norm": 0.88671875, "learning_rate": 2.691848540429902e-06, "loss": 0.1168, "step": 43988 }, { "epoch": 32.22637362637363, "grad_norm": 1.265625, "learning_rate": 2.690960511107186e-06, "loss": 0.1205, "step": 43989 }, { "epoch": 32.227106227106226, "grad_norm": 0.53515625, "learning_rate": 2.6900727108532567e-06, "loss": 0.1176, "step": 43990 }, { "epoch": 32.22783882783883, "grad_norm": 1.09375, "learning_rate": 2.689185139672401e-06, "loss": 0.1152, "step": 43991 }, { "epoch": 32.22857142857143, "grad_norm": 0.98046875, "learning_rate": 2.6882977975688454e-06, "loss": 0.1019, "step": 43992 }, { "epoch": 32.22930402930403, "grad_norm": 1.1953125, "learning_rate": 2.687410684546859e-06, "loss": 0.1062, "step": 43993 }, { "epoch": 32.23003663003663, "grad_norm": 1.0078125, "learning_rate": 2.6865238006106743e-06, "loss": 0.1004, "step": 43994 }, { "epoch": 32.23076923076923, "grad_norm": 1.015625, "learning_rate": 2.6856371457645554e-06, "loss": 0.1088, "step": 43995 }, { "epoch": 32.23150183150183, "grad_norm": 0.66015625, "learning_rate": 2.6847507200127394e-06, "loss": 0.102, "step": 43996 }, { "epoch": 32.232234432234435, "grad_norm": 0.671875, "learning_rate": 2.683864523359491e-06, "loss": 0.1205, "step": 43997 }, { "epoch": 32.23296703296703, "grad_norm": 1.5859375, "learning_rate": 2.6829785558090365e-06, "loss": 0.0935, "step": 43998 }, { "epoch": 32.233699633699636, "grad_norm": 1.53125, "learning_rate": 2.6820928173656237e-06, "loss": 0.1513, "step": 43999 }, { "epoch": 32.234432234432234, "grad_norm": 1.0703125, "learning_rate": 2.681207308033512e-06, "loss": 0.1443, "step": 44000 }, { "epoch": 32.23516483516484, "grad_norm": 0.70703125, "learning_rate": 2.6803220278169327e-06, "loss": 0.1041, "step": 44001 }, { "epoch": 32.235897435897435, "grad_norm": 0.94921875, "learning_rate": 2.6794369767201233e-06, "loss": 0.1079, "step": 44002 }, { "epoch": 32.23663003663004, "grad_norm": 0.8828125, "learning_rate": 2.6785521547473316e-06, "loss": 0.103, "step": 44003 }, { "epoch": 32.23736263736264, "grad_norm": 0.7734375, "learning_rate": 2.6776675619028e-06, "loss": 0.1159, "step": 44004 }, { "epoch": 32.23809523809524, "grad_norm": 1.390625, "learning_rate": 2.6767831981907605e-06, "loss": 0.112, "step": 44005 }, { "epoch": 32.23882783882784, "grad_norm": 1.5625, "learning_rate": 2.6758990636154446e-06, "loss": 0.116, "step": 44006 }, { "epoch": 32.23956043956044, "grad_norm": 0.85546875, "learning_rate": 2.6750151581811002e-06, "loss": 0.1157, "step": 44007 }, { "epoch": 32.24029304029304, "grad_norm": 1.28125, "learning_rate": 2.674131481891965e-06, "loss": 0.1564, "step": 44008 }, { "epoch": 32.241025641025644, "grad_norm": 0.859375, "learning_rate": 2.6732480347522593e-06, "loss": 0.1242, "step": 44009 }, { "epoch": 32.24175824175824, "grad_norm": 1.3515625, "learning_rate": 2.672364816766225e-06, "loss": 0.1147, "step": 44010 }, { "epoch": 32.242490842490845, "grad_norm": 0.94140625, "learning_rate": 2.671481827938084e-06, "loss": 0.1342, "step": 44011 }, { "epoch": 32.24322344322344, "grad_norm": 0.953125, "learning_rate": 2.670599068272089e-06, "loss": 0.1374, "step": 44012 }, { "epoch": 32.24395604395605, "grad_norm": 1.21875, "learning_rate": 2.6697165377724443e-06, "loss": 0.0934, "step": 44013 }, { "epoch": 32.244688644688644, "grad_norm": 0.7265625, "learning_rate": 2.6688342364433924e-06, "loss": 0.1172, "step": 44014 }, { "epoch": 32.24542124542125, "grad_norm": 0.671875, "learning_rate": 2.66795216428916e-06, "loss": 0.1202, "step": 44015 }, { "epoch": 32.246153846153845, "grad_norm": 1.3671875, "learning_rate": 2.6670703213139617e-06, "loss": 0.1427, "step": 44016 }, { "epoch": 32.24688644688645, "grad_norm": 1.4140625, "learning_rate": 2.6661887075220406e-06, "loss": 0.1288, "step": 44017 }, { "epoch": 32.24761904761905, "grad_norm": 0.75390625, "learning_rate": 2.6653073229176113e-06, "loss": 0.1312, "step": 44018 }, { "epoch": 32.24835164835165, "grad_norm": 1.21875, "learning_rate": 2.6644261675049e-06, "loss": 0.1284, "step": 44019 }, { "epoch": 32.24908424908425, "grad_norm": 1.09375, "learning_rate": 2.663545241288122e-06, "loss": 0.1345, "step": 44020 }, { "epoch": 32.24981684981685, "grad_norm": 0.8671875, "learning_rate": 2.6626645442715033e-06, "loss": 0.1181, "step": 44021 }, { "epoch": 32.25054945054945, "grad_norm": 0.65234375, "learning_rate": 2.661784076459259e-06, "loss": 0.1056, "step": 44022 }, { "epoch": 32.251282051282054, "grad_norm": 0.71484375, "learning_rate": 2.660903837855621e-06, "loss": 0.1233, "step": 44023 }, { "epoch": 32.25201465201465, "grad_norm": 0.67578125, "learning_rate": 2.660023828464793e-06, "loss": 0.1143, "step": 44024 }, { "epoch": 32.252747252747255, "grad_norm": 1.0234375, "learning_rate": 2.6591440482909903e-06, "loss": 0.1264, "step": 44025 }, { "epoch": 32.25347985347985, "grad_norm": 0.9375, "learning_rate": 2.6582644973384395e-06, "loss": 0.0905, "step": 44026 }, { "epoch": 32.25421245421246, "grad_norm": 1.046875, "learning_rate": 2.6573851756113497e-06, "loss": 0.1394, "step": 44027 }, { "epoch": 32.254945054945054, "grad_norm": 1.328125, "learning_rate": 2.65650608311393e-06, "loss": 0.141, "step": 44028 }, { "epoch": 32.25567765567766, "grad_norm": 0.96484375, "learning_rate": 2.6556272198503916e-06, "loss": 0.1123, "step": 44029 }, { "epoch": 32.256410256410255, "grad_norm": 0.76171875, "learning_rate": 2.6547485858249594e-06, "loss": 0.1151, "step": 44030 }, { "epoch": 32.25714285714286, "grad_norm": 1.8046875, "learning_rate": 2.6538701810418268e-06, "loss": 0.1308, "step": 44031 }, { "epoch": 32.25787545787546, "grad_norm": 1.046875, "learning_rate": 2.652992005505209e-06, "loss": 0.1228, "step": 44032 }, { "epoch": 32.25860805860806, "grad_norm": 1.171875, "learning_rate": 2.6521140592193153e-06, "loss": 0.1213, "step": 44033 }, { "epoch": 32.25934065934066, "grad_norm": 1.203125, "learning_rate": 2.65123634218835e-06, "loss": 0.1115, "step": 44034 }, { "epoch": 32.26007326007326, "grad_norm": 1.015625, "learning_rate": 2.6503588544165235e-06, "loss": 0.127, "step": 44035 }, { "epoch": 32.26080586080586, "grad_norm": 0.640625, "learning_rate": 2.6494815959080333e-06, "loss": 0.0893, "step": 44036 }, { "epoch": 32.261538461538464, "grad_norm": 1.09375, "learning_rate": 2.6486045666670734e-06, "loss": 0.135, "step": 44037 }, { "epoch": 32.26227106227106, "grad_norm": 0.96875, "learning_rate": 2.647727766697875e-06, "loss": 0.1163, "step": 44038 }, { "epoch": 32.263003663003666, "grad_norm": 1.1484375, "learning_rate": 2.6468511960046146e-06, "loss": 0.1443, "step": 44039 }, { "epoch": 32.26373626373626, "grad_norm": 0.92578125, "learning_rate": 2.645974854591497e-06, "loss": 0.1069, "step": 44040 }, { "epoch": 32.26446886446887, "grad_norm": 1.21875, "learning_rate": 2.645098742462726e-06, "loss": 0.1084, "step": 44041 }, { "epoch": 32.265201465201464, "grad_norm": 1.2109375, "learning_rate": 2.6442228596224884e-06, "loss": 0.1397, "step": 44042 }, { "epoch": 32.26593406593407, "grad_norm": 1.0390625, "learning_rate": 2.6433472060750006e-06, "loss": 0.1148, "step": 44043 }, { "epoch": 32.266666666666666, "grad_norm": 0.8515625, "learning_rate": 2.6424717818244444e-06, "loss": 0.1272, "step": 44044 }, { "epoch": 32.26739926739927, "grad_norm": 1.0390625, "learning_rate": 2.641596586875018e-06, "loss": 0.1256, "step": 44045 }, { "epoch": 32.26813186813187, "grad_norm": 0.7109375, "learning_rate": 2.640721621230909e-06, "loss": 0.1279, "step": 44046 }, { "epoch": 32.26886446886447, "grad_norm": 2.03125, "learning_rate": 2.6398468848963116e-06, "loss": 0.1436, "step": 44047 }, { "epoch": 32.26959706959707, "grad_norm": 1.015625, "learning_rate": 2.638972377875423e-06, "loss": 0.1465, "step": 44048 }, { "epoch": 32.27032967032967, "grad_norm": 0.9296875, "learning_rate": 2.6380981001724323e-06, "loss": 0.1134, "step": 44049 }, { "epoch": 32.27106227106227, "grad_norm": 1.078125, "learning_rate": 2.6372240517915206e-06, "loss": 0.1411, "step": 44050 }, { "epoch": 32.271794871794874, "grad_norm": 0.90625, "learning_rate": 2.6363502327368816e-06, "loss": 0.1, "step": 44051 }, { "epoch": 32.27252747252747, "grad_norm": 1.21875, "learning_rate": 2.635476643012702e-06, "loss": 0.1231, "step": 44052 }, { "epoch": 32.273260073260076, "grad_norm": 0.73828125, "learning_rate": 2.6346032826231706e-06, "loss": 0.1303, "step": 44053 }, { "epoch": 32.27399267399267, "grad_norm": 1.96875, "learning_rate": 2.633730151572464e-06, "loss": 0.1661, "step": 44054 }, { "epoch": 32.27472527472528, "grad_norm": 0.890625, "learning_rate": 2.632857249864769e-06, "loss": 0.1256, "step": 44055 }, { "epoch": 32.275457875457874, "grad_norm": 1.40625, "learning_rate": 2.631984577504274e-06, "loss": 0.0956, "step": 44056 }, { "epoch": 32.27619047619048, "grad_norm": 1.203125, "learning_rate": 2.63111213449515e-06, "loss": 0.1209, "step": 44057 }, { "epoch": 32.276923076923076, "grad_norm": 1.3359375, "learning_rate": 2.6302399208415787e-06, "loss": 0.1377, "step": 44058 }, { "epoch": 32.27765567765568, "grad_norm": 0.9609375, "learning_rate": 2.629367936547748e-06, "loss": 0.1124, "step": 44059 }, { "epoch": 32.27838827838828, "grad_norm": 1.421875, "learning_rate": 2.6284961816178296e-06, "loss": 0.125, "step": 44060 }, { "epoch": 32.27912087912088, "grad_norm": 1.046875, "learning_rate": 2.6276246560559996e-06, "loss": 0.1263, "step": 44061 }, { "epoch": 32.27985347985348, "grad_norm": 1.625, "learning_rate": 2.6267533598664405e-06, "loss": 0.1145, "step": 44062 }, { "epoch": 32.28058608058608, "grad_norm": 0.984375, "learning_rate": 2.625882293053312e-06, "loss": 0.1324, "step": 44063 }, { "epoch": 32.28131868131868, "grad_norm": 0.8515625, "learning_rate": 2.6250114556208074e-06, "loss": 0.1168, "step": 44064 }, { "epoch": 32.282051282051285, "grad_norm": 0.828125, "learning_rate": 2.624140847573081e-06, "loss": 0.1061, "step": 44065 }, { "epoch": 32.28278388278388, "grad_norm": 1.296875, "learning_rate": 2.623270468914316e-06, "loss": 0.0961, "step": 44066 }, { "epoch": 32.283516483516486, "grad_norm": 1.078125, "learning_rate": 2.622400319648683e-06, "loss": 0.1138, "step": 44067 }, { "epoch": 32.28424908424908, "grad_norm": 0.8203125, "learning_rate": 2.621530399780337e-06, "loss": 0.1323, "step": 44068 }, { "epoch": 32.28498168498169, "grad_norm": 0.90234375, "learning_rate": 2.62066070931347e-06, "loss": 0.1447, "step": 44069 }, { "epoch": 32.285714285714285, "grad_norm": 1.5625, "learning_rate": 2.619791248252227e-06, "loss": 0.1256, "step": 44070 }, { "epoch": 32.28644688644689, "grad_norm": 0.75, "learning_rate": 2.6189220166007842e-06, "loss": 0.1097, "step": 44071 }, { "epoch": 32.287179487179486, "grad_norm": 1.0390625, "learning_rate": 2.6180530143633072e-06, "loss": 0.1481, "step": 44072 }, { "epoch": 32.28791208791209, "grad_norm": 1.0078125, "learning_rate": 2.6171842415439506e-06, "loss": 0.105, "step": 44073 }, { "epoch": 32.28864468864469, "grad_norm": 1.4296875, "learning_rate": 2.6163156981468855e-06, "loss": 0.0994, "step": 44074 }, { "epoch": 32.28937728937729, "grad_norm": 0.8125, "learning_rate": 2.6154473841762726e-06, "loss": 0.1137, "step": 44075 }, { "epoch": 32.29010989010989, "grad_norm": 0.7734375, "learning_rate": 2.614579299636271e-06, "loss": 0.1102, "step": 44076 }, { "epoch": 32.29084249084249, "grad_norm": 0.57421875, "learning_rate": 2.613711444531037e-06, "loss": 0.0867, "step": 44077 }, { "epoch": 32.29157509157509, "grad_norm": 1.03125, "learning_rate": 2.6128438188647293e-06, "loss": 0.1127, "step": 44078 }, { "epoch": 32.292307692307695, "grad_norm": 1.359375, "learning_rate": 2.611976422641509e-06, "loss": 0.1257, "step": 44079 }, { "epoch": 32.29304029304029, "grad_norm": 1.0078125, "learning_rate": 2.6111092558655356e-06, "loss": 0.123, "step": 44080 }, { "epoch": 32.293772893772896, "grad_norm": 1.046875, "learning_rate": 2.6102423185409534e-06, "loss": 0.1318, "step": 44081 }, { "epoch": 32.29450549450549, "grad_norm": 1.0859375, "learning_rate": 2.6093756106719273e-06, "loss": 0.0937, "step": 44082 }, { "epoch": 32.2952380952381, "grad_norm": 1.015625, "learning_rate": 2.6085091322626016e-06, "loss": 0.1228, "step": 44083 }, { "epoch": 32.295970695970695, "grad_norm": 1.1015625, "learning_rate": 2.6076428833171254e-06, "loss": 0.1164, "step": 44084 }, { "epoch": 32.2967032967033, "grad_norm": 0.82421875, "learning_rate": 2.606776863839659e-06, "loss": 0.1149, "step": 44085 }, { "epoch": 32.297435897435896, "grad_norm": 1.1171875, "learning_rate": 2.6059110738343454e-06, "loss": 0.1013, "step": 44086 }, { "epoch": 32.2981684981685, "grad_norm": 1.3984375, "learning_rate": 2.605045513305335e-06, "loss": 0.1161, "step": 44087 }, { "epoch": 32.2989010989011, "grad_norm": 0.9296875, "learning_rate": 2.60418018225677e-06, "loss": 0.1162, "step": 44088 }, { "epoch": 32.2996336996337, "grad_norm": 1.3515625, "learning_rate": 2.603315080692806e-06, "loss": 0.1251, "step": 44089 }, { "epoch": 32.3003663003663, "grad_norm": 0.9453125, "learning_rate": 2.6024502086175864e-06, "loss": 0.1086, "step": 44090 }, { "epoch": 32.301098901098904, "grad_norm": 0.97265625, "learning_rate": 2.6015855660352444e-06, "loss": 0.1106, "step": 44091 }, { "epoch": 32.3018315018315, "grad_norm": 1.4453125, "learning_rate": 2.600721152949934e-06, "loss": 0.1222, "step": 44092 }, { "epoch": 32.302564102564105, "grad_norm": 0.89453125, "learning_rate": 2.5998569693657937e-06, "loss": 0.1249, "step": 44093 }, { "epoch": 32.3032967032967, "grad_norm": 1.203125, "learning_rate": 2.598993015286951e-06, "loss": 0.0799, "step": 44094 }, { "epoch": 32.30402930402931, "grad_norm": 1.125, "learning_rate": 2.598129290717576e-06, "loss": 0.1167, "step": 44095 }, { "epoch": 32.304761904761904, "grad_norm": 1.28125, "learning_rate": 2.5972657956617756e-06, "loss": 0.146, "step": 44096 }, { "epoch": 32.30549450549451, "grad_norm": 0.953125, "learning_rate": 2.596402530123708e-06, "loss": 0.1156, "step": 44097 }, { "epoch": 32.306227106227105, "grad_norm": 1.1171875, "learning_rate": 2.5955394941074964e-06, "loss": 0.1033, "step": 44098 }, { "epoch": 32.30695970695971, "grad_norm": 0.78125, "learning_rate": 2.5946766876172843e-06, "loss": 0.1126, "step": 44099 }, { "epoch": 32.30769230769231, "grad_norm": 0.828125, "learning_rate": 2.5938141106571982e-06, "loss": 0.1223, "step": 44100 }, { "epoch": 32.30842490842491, "grad_norm": 1.046875, "learning_rate": 2.5929517632313823e-06, "loss": 0.1226, "step": 44101 }, { "epoch": 32.30915750915751, "grad_norm": 0.8984375, "learning_rate": 2.5920896453439586e-06, "loss": 0.1247, "step": 44102 }, { "epoch": 32.30989010989011, "grad_norm": 0.8125, "learning_rate": 2.591227756999059e-06, "loss": 0.1014, "step": 44103 }, { "epoch": 32.31062271062271, "grad_norm": 0.81640625, "learning_rate": 2.5903660982008165e-06, "loss": 0.1031, "step": 44104 }, { "epoch": 32.311355311355314, "grad_norm": 1.0703125, "learning_rate": 2.589504668953353e-06, "loss": 0.1246, "step": 44105 }, { "epoch": 32.31208791208791, "grad_norm": 0.875, "learning_rate": 2.5886434692608066e-06, "loss": 0.1373, "step": 44106 }, { "epoch": 32.312820512820515, "grad_norm": 1.3125, "learning_rate": 2.5877824991272936e-06, "loss": 0.1292, "step": 44107 }, { "epoch": 32.31355311355311, "grad_norm": 1.3046875, "learning_rate": 2.5869217585569463e-06, "loss": 0.1484, "step": 44108 }, { "epoch": 32.31428571428572, "grad_norm": 1.0078125, "learning_rate": 2.5860612475538865e-06, "loss": 0.1183, "step": 44109 }, { "epoch": 32.315018315018314, "grad_norm": 1.203125, "learning_rate": 2.585200966122231e-06, "loss": 0.1239, "step": 44110 }, { "epoch": 32.31575091575092, "grad_norm": 0.89453125, "learning_rate": 2.5843409142661064e-06, "loss": 0.1172, "step": 44111 }, { "epoch": 32.316483516483515, "grad_norm": 0.94140625, "learning_rate": 2.5834810919896454e-06, "loss": 0.1136, "step": 44112 }, { "epoch": 32.31721611721612, "grad_norm": 1.03125, "learning_rate": 2.582621499296943e-06, "loss": 0.1059, "step": 44113 }, { "epoch": 32.31794871794872, "grad_norm": 1.4375, "learning_rate": 2.5817621361921366e-06, "loss": 0.1348, "step": 44114 }, { "epoch": 32.31868131868132, "grad_norm": 1.2109375, "learning_rate": 2.5809030026793367e-06, "loss": 0.1024, "step": 44115 }, { "epoch": 32.31941391941392, "grad_norm": 0.75390625, "learning_rate": 2.5800440987626714e-06, "loss": 0.1283, "step": 44116 }, { "epoch": 32.32014652014652, "grad_norm": 1.109375, "learning_rate": 2.579185424446234e-06, "loss": 0.1171, "step": 44117 }, { "epoch": 32.32087912087912, "grad_norm": 0.953125, "learning_rate": 2.5783269797341523e-06, "loss": 0.1224, "step": 44118 }, { "epoch": 32.321611721611724, "grad_norm": 1.375, "learning_rate": 2.5774687646305424e-06, "loss": 0.1144, "step": 44119 }, { "epoch": 32.32234432234432, "grad_norm": 0.8984375, "learning_rate": 2.5766107791395036e-06, "loss": 0.1039, "step": 44120 }, { "epoch": 32.323076923076925, "grad_norm": 1.109375, "learning_rate": 2.575753023265164e-06, "loss": 0.1341, "step": 44121 }, { "epoch": 32.32380952380952, "grad_norm": 1.1328125, "learning_rate": 2.5748954970116228e-06, "loss": 0.1162, "step": 44122 }, { "epoch": 32.32454212454213, "grad_norm": 0.64453125, "learning_rate": 2.57403820038299e-06, "loss": 0.1075, "step": 44123 }, { "epoch": 32.325274725274724, "grad_norm": 1.328125, "learning_rate": 2.5731811333833665e-06, "loss": 0.1267, "step": 44124 }, { "epoch": 32.32600732600733, "grad_norm": 0.84765625, "learning_rate": 2.5723242960168678e-06, "loss": 0.0955, "step": 44125 }, { "epoch": 32.326739926739926, "grad_norm": 1.21875, "learning_rate": 2.5714676882876e-06, "loss": 0.1017, "step": 44126 }, { "epoch": 32.32747252747253, "grad_norm": 1.15625, "learning_rate": 2.570611310199667e-06, "loss": 0.1247, "step": 44127 }, { "epoch": 32.32820512820513, "grad_norm": 1.5703125, "learning_rate": 2.569755161757164e-06, "loss": 0.1347, "step": 44128 }, { "epoch": 32.32893772893773, "grad_norm": 0.66796875, "learning_rate": 2.568899242964201e-06, "loss": 0.1307, "step": 44129 }, { "epoch": 32.32967032967033, "grad_norm": 1.375, "learning_rate": 2.568043553824878e-06, "loss": 0.1213, "step": 44130 }, { "epoch": 32.33040293040293, "grad_norm": 0.84765625, "learning_rate": 2.567188094343284e-06, "loss": 0.1059, "step": 44131 }, { "epoch": 32.33113553113553, "grad_norm": 1.0, "learning_rate": 2.5663328645235406e-06, "loss": 0.124, "step": 44132 }, { "epoch": 32.331868131868134, "grad_norm": 1.125, "learning_rate": 2.565477864369726e-06, "loss": 0.1429, "step": 44133 }, { "epoch": 32.33260073260073, "grad_norm": 1.015625, "learning_rate": 2.564623093885944e-06, "loss": 0.1292, "step": 44134 }, { "epoch": 32.333333333333336, "grad_norm": 1.0703125, "learning_rate": 2.5637685530762847e-06, "loss": 0.1107, "step": 44135 }, { "epoch": 32.33406593406593, "grad_norm": 1.40625, "learning_rate": 2.562914241944853e-06, "loss": 0.1054, "step": 44136 }, { "epoch": 32.33479853479854, "grad_norm": 1.1796875, "learning_rate": 2.5620601604957264e-06, "loss": 0.1194, "step": 44137 }, { "epoch": 32.335531135531134, "grad_norm": 1.109375, "learning_rate": 2.561206308733021e-06, "loss": 0.1386, "step": 44138 }, { "epoch": 32.33626373626374, "grad_norm": 1.0078125, "learning_rate": 2.5603526866608036e-06, "loss": 0.1281, "step": 44139 }, { "epoch": 32.336996336996336, "grad_norm": 1.1640625, "learning_rate": 2.559499294283174e-06, "loss": 0.1061, "step": 44140 }, { "epoch": 32.33772893772894, "grad_norm": 1.09375, "learning_rate": 2.5586461316042213e-06, "loss": 0.1166, "step": 44141 }, { "epoch": 32.33846153846154, "grad_norm": 1.21875, "learning_rate": 2.5577931986280343e-06, "loss": 0.1269, "step": 44142 }, { "epoch": 32.33919413919414, "grad_norm": 0.984375, "learning_rate": 2.5569404953587007e-06, "loss": 0.093, "step": 44143 }, { "epoch": 32.33992673992674, "grad_norm": 1.1953125, "learning_rate": 2.5560880218002988e-06, "loss": 0.1433, "step": 44144 }, { "epoch": 32.34065934065934, "grad_norm": 1.1015625, "learning_rate": 2.555235777956923e-06, "loss": 0.1285, "step": 44145 }, { "epoch": 32.34139194139194, "grad_norm": 0.84375, "learning_rate": 2.5543837638326454e-06, "loss": 0.1316, "step": 44146 }, { "epoch": 32.342124542124544, "grad_norm": 0.88671875, "learning_rate": 2.5535319794315665e-06, "loss": 0.13, "step": 44147 }, { "epoch": 32.34285714285714, "grad_norm": 0.78125, "learning_rate": 2.552680424757746e-06, "loss": 0.1106, "step": 44148 }, { "epoch": 32.343589743589746, "grad_norm": 1.171875, "learning_rate": 2.551829099815279e-06, "loss": 0.1098, "step": 44149 }, { "epoch": 32.34432234432234, "grad_norm": 1.2421875, "learning_rate": 2.5509780046082324e-06, "loss": 0.121, "step": 44150 }, { "epoch": 32.34505494505495, "grad_norm": 0.96484375, "learning_rate": 2.550127139140695e-06, "loss": 0.1021, "step": 44151 }, { "epoch": 32.345787545787545, "grad_norm": 0.9765625, "learning_rate": 2.5492765034167384e-06, "loss": 0.1288, "step": 44152 }, { "epoch": 32.34652014652015, "grad_norm": 1.1796875, "learning_rate": 2.5484260974404408e-06, "loss": 0.1795, "step": 44153 }, { "epoch": 32.347252747252746, "grad_norm": 0.8671875, "learning_rate": 2.5475759212158746e-06, "loss": 0.1207, "step": 44154 }, { "epoch": 32.34798534798535, "grad_norm": 0.96875, "learning_rate": 2.546725974747111e-06, "loss": 0.0975, "step": 44155 }, { "epoch": 32.34871794871795, "grad_norm": 0.98828125, "learning_rate": 2.5458762580382343e-06, "loss": 0.1046, "step": 44156 }, { "epoch": 32.34945054945055, "grad_norm": 0.69921875, "learning_rate": 2.5450267710932948e-06, "loss": 0.116, "step": 44157 }, { "epoch": 32.35018315018315, "grad_norm": 1.265625, "learning_rate": 2.5441775139163806e-06, "loss": 0.0984, "step": 44158 }, { "epoch": 32.35091575091575, "grad_norm": 1.1484375, "learning_rate": 2.5433284865115538e-06, "loss": 0.1102, "step": 44159 }, { "epoch": 32.35164835164835, "grad_norm": 0.66796875, "learning_rate": 2.542479688882891e-06, "loss": 0.127, "step": 44160 }, { "epoch": 32.352380952380955, "grad_norm": 1.1328125, "learning_rate": 2.5416311210344434e-06, "loss": 0.1172, "step": 44161 }, { "epoch": 32.35311355311355, "grad_norm": 1.1015625, "learning_rate": 2.540782782970283e-06, "loss": 0.1194, "step": 44162 }, { "epoch": 32.353846153846156, "grad_norm": 0.98046875, "learning_rate": 2.539934674694476e-06, "loss": 0.1514, "step": 44163 }, { "epoch": 32.35457875457875, "grad_norm": 2.046875, "learning_rate": 2.5390867962110954e-06, "loss": 0.1251, "step": 44164 }, { "epoch": 32.35531135531136, "grad_norm": 1.2578125, "learning_rate": 2.5382391475241854e-06, "loss": 0.1131, "step": 44165 }, { "epoch": 32.356043956043955, "grad_norm": 0.52734375, "learning_rate": 2.537391728637819e-06, "loss": 0.093, "step": 44166 }, { "epoch": 32.35677655677656, "grad_norm": 0.88671875, "learning_rate": 2.536544539556051e-06, "loss": 0.12, "step": 44167 }, { "epoch": 32.357509157509156, "grad_norm": 1.3125, "learning_rate": 2.5356975802829537e-06, "loss": 0.1434, "step": 44168 }, { "epoch": 32.35824175824176, "grad_norm": 0.8203125, "learning_rate": 2.5348508508225675e-06, "loss": 0.1099, "step": 44169 }, { "epoch": 32.35897435897436, "grad_norm": 1.515625, "learning_rate": 2.5340043511789527e-06, "loss": 0.1124, "step": 44170 }, { "epoch": 32.35970695970696, "grad_norm": 0.828125, "learning_rate": 2.5331580813561816e-06, "loss": 0.1104, "step": 44171 }, { "epoch": 32.36043956043956, "grad_norm": 0.828125, "learning_rate": 2.5323120413582825e-06, "loss": 0.1376, "step": 44172 }, { "epoch": 32.36117216117216, "grad_norm": 0.9296875, "learning_rate": 2.531466231189334e-06, "loss": 0.0989, "step": 44173 }, { "epoch": 32.36190476190476, "grad_norm": 0.94140625, "learning_rate": 2.5306206508533684e-06, "loss": 0.1057, "step": 44174 }, { "epoch": 32.362637362637365, "grad_norm": 1.359375, "learning_rate": 2.5297753003544592e-06, "loss": 0.0995, "step": 44175 }, { "epoch": 32.36336996336996, "grad_norm": 0.875, "learning_rate": 2.5289301796966337e-06, "loss": 0.1264, "step": 44176 }, { "epoch": 32.364102564102566, "grad_norm": 1.1640625, "learning_rate": 2.5280852888839542e-06, "loss": 0.122, "step": 44177 }, { "epoch": 32.36483516483516, "grad_norm": 0.9765625, "learning_rate": 2.5272406279204646e-06, "loss": 0.1416, "step": 44178 }, { "epoch": 32.36556776556777, "grad_norm": 1.1171875, "learning_rate": 2.5263961968102207e-06, "loss": 0.1067, "step": 44179 }, { "epoch": 32.366300366300365, "grad_norm": 1.34375, "learning_rate": 2.5255519955572507e-06, "loss": 0.0859, "step": 44180 }, { "epoch": 32.36703296703297, "grad_norm": 1.015625, "learning_rate": 2.524708024165617e-06, "loss": 0.1415, "step": 44181 }, { "epoch": 32.36776556776557, "grad_norm": 1.046875, "learning_rate": 2.5238642826393577e-06, "loss": 0.1173, "step": 44182 }, { "epoch": 32.36849816849817, "grad_norm": 1.6796875, "learning_rate": 2.523020770982507e-06, "loss": 0.1308, "step": 44183 }, { "epoch": 32.36923076923077, "grad_norm": 0.83984375, "learning_rate": 2.5221774891991202e-06, "loss": 0.1074, "step": 44184 }, { "epoch": 32.36996336996337, "grad_norm": 1.9609375, "learning_rate": 2.521334437293226e-06, "loss": 0.1212, "step": 44185 }, { "epoch": 32.37069597069597, "grad_norm": 1.4140625, "learning_rate": 2.5204916152688754e-06, "loss": 0.1918, "step": 44186 }, { "epoch": 32.371428571428574, "grad_norm": 0.828125, "learning_rate": 2.519649023130095e-06, "loss": 0.1267, "step": 44187 }, { "epoch": 32.37216117216117, "grad_norm": 0.90234375, "learning_rate": 2.5188066608809303e-06, "loss": 0.1399, "step": 44188 }, { "epoch": 32.372893772893775, "grad_norm": 0.765625, "learning_rate": 2.5179645285254107e-06, "loss": 0.098, "step": 44189 }, { "epoch": 32.37362637362637, "grad_norm": 0.78125, "learning_rate": 2.5171226260675794e-06, "loss": 0.0975, "step": 44190 }, { "epoch": 32.37435897435898, "grad_norm": 2.203125, "learning_rate": 2.5162809535114595e-06, "loss": 0.1363, "step": 44191 }, { "epoch": 32.375091575091574, "grad_norm": 0.83984375, "learning_rate": 2.5154395108610908e-06, "loss": 0.1119, "step": 44192 }, { "epoch": 32.37582417582418, "grad_norm": 1.375, "learning_rate": 2.514598298120507e-06, "loss": 0.1631, "step": 44193 }, { "epoch": 32.376556776556775, "grad_norm": 0.6875, "learning_rate": 2.513757315293736e-06, "loss": 0.1069, "step": 44194 }, { "epoch": 32.37728937728938, "grad_norm": 0.99609375, "learning_rate": 2.5129165623848006e-06, "loss": 0.1064, "step": 44195 }, { "epoch": 32.37802197802198, "grad_norm": 1.2578125, "learning_rate": 2.5120760393977405e-06, "loss": 0.1077, "step": 44196 }, { "epoch": 32.37875457875458, "grad_norm": 1.0546875, "learning_rate": 2.5112357463365783e-06, "loss": 0.1294, "step": 44197 }, { "epoch": 32.37948717948718, "grad_norm": 0.796875, "learning_rate": 2.5103956832053257e-06, "loss": 0.1265, "step": 44198 }, { "epoch": 32.38021978021978, "grad_norm": 1.1328125, "learning_rate": 2.509555850008038e-06, "loss": 0.1204, "step": 44199 }, { "epoch": 32.38095238095238, "grad_norm": 1.1640625, "learning_rate": 2.508716246748717e-06, "loss": 0.1223, "step": 44200 }, { "epoch": 32.381684981684984, "grad_norm": 1.234375, "learning_rate": 2.507876873431396e-06, "loss": 0.1472, "step": 44201 }, { "epoch": 32.38241758241758, "grad_norm": 0.7890625, "learning_rate": 2.5070377300600806e-06, "loss": 0.1215, "step": 44202 }, { "epoch": 32.383150183150185, "grad_norm": 1.0859375, "learning_rate": 2.5061988166388058e-06, "loss": 0.1227, "step": 44203 }, { "epoch": 32.38388278388278, "grad_norm": 0.9921875, "learning_rate": 2.5053601331715935e-06, "loss": 0.1318, "step": 44204 }, { "epoch": 32.38461538461539, "grad_norm": 0.734375, "learning_rate": 2.5045216796624563e-06, "loss": 0.1114, "step": 44205 }, { "epoch": 32.385347985347984, "grad_norm": 0.88671875, "learning_rate": 2.5036834561154044e-06, "loss": 0.1429, "step": 44206 }, { "epoch": 32.38608058608059, "grad_norm": 0.9140625, "learning_rate": 2.5028454625344625e-06, "loss": 0.1062, "step": 44207 }, { "epoch": 32.386813186813185, "grad_norm": 1.0546875, "learning_rate": 2.5020076989236523e-06, "loss": 0.1173, "step": 44208 }, { "epoch": 32.38754578754579, "grad_norm": 1.4765625, "learning_rate": 2.5011701652869696e-06, "loss": 0.114, "step": 44209 }, { "epoch": 32.38827838827839, "grad_norm": 1.875, "learning_rate": 2.5003328616284473e-06, "loss": 0.1084, "step": 44210 }, { "epoch": 32.38901098901099, "grad_norm": 1.3828125, "learning_rate": 2.4994957879520816e-06, "loss": 0.1582, "step": 44211 }, { "epoch": 32.38974358974359, "grad_norm": 0.9609375, "learning_rate": 2.4986589442618954e-06, "loss": 0.1335, "step": 44212 }, { "epoch": 32.39047619047619, "grad_norm": 1.09375, "learning_rate": 2.497822330561883e-06, "loss": 0.1236, "step": 44213 }, { "epoch": 32.39120879120879, "grad_norm": 1.265625, "learning_rate": 2.4969859468560628e-06, "loss": 0.1235, "step": 44214 }, { "epoch": 32.391941391941394, "grad_norm": 0.7578125, "learning_rate": 2.496149793148446e-06, "loss": 0.0857, "step": 44215 }, { "epoch": 32.39267399267399, "grad_norm": 1.2265625, "learning_rate": 2.4953138694430333e-06, "loss": 0.1079, "step": 44216 }, { "epoch": 32.393406593406596, "grad_norm": 1.0, "learning_rate": 2.4944781757438314e-06, "loss": 0.1109, "step": 44217 }, { "epoch": 32.39413919413919, "grad_norm": 0.94921875, "learning_rate": 2.493642712054841e-06, "loss": 0.1053, "step": 44218 }, { "epoch": 32.3948717948718, "grad_norm": 1.171875, "learning_rate": 2.492807478380068e-06, "loss": 0.1164, "step": 44219 }, { "epoch": 32.395604395604394, "grad_norm": 1.25, "learning_rate": 2.4919724747235134e-06, "loss": 0.1489, "step": 44220 }, { "epoch": 32.396336996337, "grad_norm": 1.640625, "learning_rate": 2.4911377010891837e-06, "loss": 0.1342, "step": 44221 }, { "epoch": 32.397069597069596, "grad_norm": 0.703125, "learning_rate": 2.490303157481068e-06, "loss": 0.1065, "step": 44222 }, { "epoch": 32.3978021978022, "grad_norm": 1.0390625, "learning_rate": 2.489468843903179e-06, "loss": 0.1145, "step": 44223 }, { "epoch": 32.3985347985348, "grad_norm": 0.8984375, "learning_rate": 2.4886347603594957e-06, "loss": 0.1207, "step": 44224 }, { "epoch": 32.3992673992674, "grad_norm": 1.0, "learning_rate": 2.4878009068540293e-06, "loss": 0.1124, "step": 44225 }, { "epoch": 32.4, "grad_norm": 1.1328125, "learning_rate": 2.486967283390769e-06, "loss": 0.1386, "step": 44226 }, { "epoch": 32.4007326007326, "grad_norm": 0.921875, "learning_rate": 2.486133889973711e-06, "loss": 0.1256, "step": 44227 }, { "epoch": 32.4014652014652, "grad_norm": 1.0703125, "learning_rate": 2.4853007266068458e-06, "loss": 0.1117, "step": 44228 }, { "epoch": 32.402197802197804, "grad_norm": 0.94140625, "learning_rate": 2.4844677932941665e-06, "loss": 0.1048, "step": 44229 }, { "epoch": 32.4029304029304, "grad_norm": 0.73046875, "learning_rate": 2.483635090039665e-06, "loss": 0.103, "step": 44230 }, { "epoch": 32.403663003663006, "grad_norm": 1.2734375, "learning_rate": 2.4828026168473357e-06, "loss": 0.1109, "step": 44231 }, { "epoch": 32.4043956043956, "grad_norm": 1.1015625, "learning_rate": 2.4819703737211576e-06, "loss": 0.1023, "step": 44232 }, { "epoch": 32.40512820512821, "grad_norm": 0.90625, "learning_rate": 2.48113836066512e-06, "loss": 0.0994, "step": 44233 }, { "epoch": 32.405860805860804, "grad_norm": 1.0, "learning_rate": 2.4803065776832187e-06, "loss": 0.0902, "step": 44234 }, { "epoch": 32.40659340659341, "grad_norm": 0.82421875, "learning_rate": 2.479475024779422e-06, "loss": 0.1091, "step": 44235 }, { "epoch": 32.407326007326006, "grad_norm": 0.796875, "learning_rate": 2.478643701957736e-06, "loss": 0.103, "step": 44236 }, { "epoch": 32.40805860805861, "grad_norm": 1.203125, "learning_rate": 2.477812609222127e-06, "loss": 0.1225, "step": 44237 }, { "epoch": 32.40879120879121, "grad_norm": 1.1875, "learning_rate": 2.4769817465765866e-06, "loss": 0.135, "step": 44238 }, { "epoch": 32.40952380952381, "grad_norm": 1.109375, "learning_rate": 2.4761511140250874e-06, "loss": 0.1148, "step": 44239 }, { "epoch": 32.41025641025641, "grad_norm": 1.1015625, "learning_rate": 2.4753207115716084e-06, "loss": 0.0991, "step": 44240 }, { "epoch": 32.41098901098901, "grad_norm": 1.3203125, "learning_rate": 2.4744905392201395e-06, "loss": 0.1081, "step": 44241 }, { "epoch": 32.41172161172161, "grad_norm": 0.85546875, "learning_rate": 2.4736605969746536e-06, "loss": 0.137, "step": 44242 }, { "epoch": 32.412454212454215, "grad_norm": 0.97265625, "learning_rate": 2.4728308848391243e-06, "loss": 0.1143, "step": 44243 }, { "epoch": 32.41318681318681, "grad_norm": 1.0625, "learning_rate": 2.4720014028175253e-06, "loss": 0.1197, "step": 44244 }, { "epoch": 32.413919413919416, "grad_norm": 1.140625, "learning_rate": 2.4711721509138348e-06, "loss": 0.1366, "step": 44245 }, { "epoch": 32.41465201465201, "grad_norm": 1.359375, "learning_rate": 2.470343129132026e-06, "loss": 0.1273, "step": 44246 }, { "epoch": 32.41538461538462, "grad_norm": 1.03125, "learning_rate": 2.4695143374760787e-06, "loss": 0.1353, "step": 44247 }, { "epoch": 32.416117216117215, "grad_norm": 1.1796875, "learning_rate": 2.4686857759499487e-06, "loss": 0.1072, "step": 44248 }, { "epoch": 32.41684981684982, "grad_norm": 0.94140625, "learning_rate": 2.4678574445576153e-06, "loss": 0.1518, "step": 44249 }, { "epoch": 32.417582417582416, "grad_norm": 0.78515625, "learning_rate": 2.467029343303035e-06, "loss": 0.0926, "step": 44250 }, { "epoch": 32.41831501831502, "grad_norm": 1.234375, "learning_rate": 2.4662014721901983e-06, "loss": 0.1507, "step": 44251 }, { "epoch": 32.41904761904762, "grad_norm": 1.03125, "learning_rate": 2.4653738312230506e-06, "loss": 0.1312, "step": 44252 }, { "epoch": 32.41978021978022, "grad_norm": 1.546875, "learning_rate": 2.4645464204055705e-06, "loss": 0.1442, "step": 44253 }, { "epoch": 32.42051282051282, "grad_norm": 1.6484375, "learning_rate": 2.4637192397417152e-06, "loss": 0.1046, "step": 44254 }, { "epoch": 32.42124542124542, "grad_norm": 1.515625, "learning_rate": 2.462892289235447e-06, "loss": 0.1562, "step": 44255 }, { "epoch": 32.42197802197802, "grad_norm": 0.99609375, "learning_rate": 2.462065568890733e-06, "loss": 0.1529, "step": 44256 }, { "epoch": 32.422710622710625, "grad_norm": 1.3359375, "learning_rate": 2.4612390787115364e-06, "loss": 0.1301, "step": 44257 }, { "epoch": 32.42344322344322, "grad_norm": 1.1640625, "learning_rate": 2.460412818701814e-06, "loss": 0.1102, "step": 44258 }, { "epoch": 32.424175824175826, "grad_norm": 0.9609375, "learning_rate": 2.459586788865517e-06, "loss": 0.1053, "step": 44259 }, { "epoch": 32.42490842490842, "grad_norm": 1.140625, "learning_rate": 2.4587609892066184e-06, "loss": 0.1025, "step": 44260 }, { "epoch": 32.42564102564103, "grad_norm": 0.84375, "learning_rate": 2.457935419729059e-06, "loss": 0.1352, "step": 44261 }, { "epoch": 32.426373626373625, "grad_norm": 1.2890625, "learning_rate": 2.457110080436812e-06, "loss": 0.1248, "step": 44262 }, { "epoch": 32.42710622710623, "grad_norm": 0.91015625, "learning_rate": 2.4562849713338118e-06, "loss": 0.1073, "step": 44263 }, { "epoch": 32.427838827838826, "grad_norm": 1.484375, "learning_rate": 2.455460092424032e-06, "loss": 0.1238, "step": 44264 }, { "epoch": 32.42857142857143, "grad_norm": 1.078125, "learning_rate": 2.454635443711408e-06, "loss": 0.114, "step": 44265 }, { "epoch": 32.42930402930403, "grad_norm": 0.85546875, "learning_rate": 2.4538110251998957e-06, "loss": 0.1311, "step": 44266 }, { "epoch": 32.43003663003663, "grad_norm": 0.96484375, "learning_rate": 2.4529868368934526e-06, "loss": 0.1073, "step": 44267 }, { "epoch": 32.43076923076923, "grad_norm": 1.328125, "learning_rate": 2.4521628787960244e-06, "loss": 0.0947, "step": 44268 }, { "epoch": 32.43150183150183, "grad_norm": 1.0, "learning_rate": 2.4513391509115515e-06, "loss": 0.1292, "step": 44269 }, { "epoch": 32.43223443223443, "grad_norm": 1.0546875, "learning_rate": 2.450515653243985e-06, "loss": 0.1217, "step": 44270 }, { "epoch": 32.432967032967035, "grad_norm": 1.1875, "learning_rate": 2.4496923857972768e-06, "loss": 0.1058, "step": 44271 }, { "epoch": 32.43369963369963, "grad_norm": 0.92578125, "learning_rate": 2.4488693485753604e-06, "loss": 0.1075, "step": 44272 }, { "epoch": 32.434432234432236, "grad_norm": 1.0625, "learning_rate": 2.4480465415821944e-06, "loss": 0.1339, "step": 44273 }, { "epoch": 32.435164835164834, "grad_norm": 0.94921875, "learning_rate": 2.4472239648217017e-06, "loss": 0.1192, "step": 44274 }, { "epoch": 32.43589743589744, "grad_norm": 0.8359375, "learning_rate": 2.4464016182978445e-06, "loss": 0.1132, "step": 44275 }, { "epoch": 32.436630036630035, "grad_norm": 1.1015625, "learning_rate": 2.4455795020145363e-06, "loss": 0.1256, "step": 44276 }, { "epoch": 32.43736263736264, "grad_norm": 1.140625, "learning_rate": 2.4447576159757493e-06, "loss": 0.1062, "step": 44277 }, { "epoch": 32.43809523809524, "grad_norm": 1.421875, "learning_rate": 2.4439359601853922e-06, "loss": 0.101, "step": 44278 }, { "epoch": 32.43882783882784, "grad_norm": 1.0078125, "learning_rate": 2.443114534647421e-06, "loss": 0.1406, "step": 44279 }, { "epoch": 32.43956043956044, "grad_norm": 1.015625, "learning_rate": 2.4422933393657602e-06, "loss": 0.1258, "step": 44280 }, { "epoch": 32.44029304029304, "grad_norm": 1.515625, "learning_rate": 2.4414723743443496e-06, "loss": 0.1253, "step": 44281 }, { "epoch": 32.44102564102564, "grad_norm": 0.86328125, "learning_rate": 2.4406516395871186e-06, "loss": 0.1243, "step": 44282 }, { "epoch": 32.441758241758244, "grad_norm": 1.0625, "learning_rate": 2.439831135098008e-06, "loss": 0.1359, "step": 44283 }, { "epoch": 32.44249084249084, "grad_norm": 1.4453125, "learning_rate": 2.4390108608809353e-06, "loss": 0.1251, "step": 44284 }, { "epoch": 32.443223443223445, "grad_norm": 1.4140625, "learning_rate": 2.4381908169398414e-06, "loss": 0.1337, "step": 44285 }, { "epoch": 32.44395604395604, "grad_norm": 0.78125, "learning_rate": 2.4373710032786617e-06, "loss": 0.1181, "step": 44286 }, { "epoch": 32.44468864468865, "grad_norm": 0.9296875, "learning_rate": 2.436551419901303e-06, "loss": 0.1132, "step": 44287 }, { "epoch": 32.445421245421244, "grad_norm": 0.80859375, "learning_rate": 2.4357320668117162e-06, "loss": 0.0964, "step": 44288 }, { "epoch": 32.44615384615385, "grad_norm": 1.109375, "learning_rate": 2.434912944013809e-06, "loss": 0.1316, "step": 44289 }, { "epoch": 32.446886446886445, "grad_norm": 1.15625, "learning_rate": 2.4340940515115167e-06, "loss": 0.097, "step": 44290 }, { "epoch": 32.44761904761905, "grad_norm": 0.65625, "learning_rate": 2.433275389308752e-06, "loss": 0.1036, "step": 44291 }, { "epoch": 32.44835164835165, "grad_norm": 1.0703125, "learning_rate": 2.4324569574094494e-06, "loss": 0.1071, "step": 44292 }, { "epoch": 32.44908424908425, "grad_norm": 0.9375, "learning_rate": 2.4316387558175217e-06, "loss": 0.1195, "step": 44293 }, { "epoch": 32.44981684981685, "grad_norm": 0.8203125, "learning_rate": 2.430820784536899e-06, "loss": 0.1034, "step": 44294 }, { "epoch": 32.45054945054945, "grad_norm": 1.09375, "learning_rate": 2.430003043571488e-06, "loss": 0.1404, "step": 44295 }, { "epoch": 32.45128205128205, "grad_norm": 0.94140625, "learning_rate": 2.429185532925213e-06, "loss": 0.1487, "step": 44296 }, { "epoch": 32.452014652014654, "grad_norm": 1.140625, "learning_rate": 2.4283682526019863e-06, "loss": 0.1336, "step": 44297 }, { "epoch": 32.45274725274725, "grad_norm": 0.921875, "learning_rate": 2.4275512026057324e-06, "loss": 0.0991, "step": 44298 }, { "epoch": 32.453479853479855, "grad_norm": 0.7265625, "learning_rate": 2.4267343829403688e-06, "loss": 0.1034, "step": 44299 }, { "epoch": 32.45421245421245, "grad_norm": 1.1875, "learning_rate": 2.4259177936097985e-06, "loss": 0.1349, "step": 44300 }, { "epoch": 32.45494505494506, "grad_norm": 1.0078125, "learning_rate": 2.425101434617939e-06, "loss": 0.1085, "step": 44301 }, { "epoch": 32.455677655677654, "grad_norm": 1.3828125, "learning_rate": 2.424285305968687e-06, "loss": 0.1266, "step": 44302 }, { "epoch": 32.45641025641026, "grad_norm": 1.0, "learning_rate": 2.423469407665983e-06, "loss": 0.1265, "step": 44303 }, { "epoch": 32.457142857142856, "grad_norm": 1.015625, "learning_rate": 2.422653739713712e-06, "loss": 0.1103, "step": 44304 }, { "epoch": 32.45787545787546, "grad_norm": 1.859375, "learning_rate": 2.421838302115793e-06, "loss": 0.1296, "step": 44305 }, { "epoch": 32.45860805860806, "grad_norm": 0.890625, "learning_rate": 2.421023094876127e-06, "loss": 0.1294, "step": 44306 }, { "epoch": 32.45934065934066, "grad_norm": 0.99609375, "learning_rate": 2.420208117998622e-06, "loss": 0.1268, "step": 44307 }, { "epoch": 32.46007326007326, "grad_norm": 1.1875, "learning_rate": 2.4193933714871853e-06, "loss": 0.1619, "step": 44308 }, { "epoch": 32.46080586080586, "grad_norm": 0.95703125, "learning_rate": 2.418578855345719e-06, "loss": 0.1519, "step": 44309 }, { "epoch": 32.46153846153846, "grad_norm": 1.0078125, "learning_rate": 2.417764569578124e-06, "loss": 0.1014, "step": 44310 }, { "epoch": 32.462271062271064, "grad_norm": 1.3984375, "learning_rate": 2.4169505141883036e-06, "loss": 0.1093, "step": 44311 }, { "epoch": 32.46300366300366, "grad_norm": 0.875, "learning_rate": 2.416136689180164e-06, "loss": 0.1414, "step": 44312 }, { "epoch": 32.463736263736266, "grad_norm": 1.2890625, "learning_rate": 2.415323094557586e-06, "loss": 0.1192, "step": 44313 }, { "epoch": 32.46446886446886, "grad_norm": 0.90234375, "learning_rate": 2.4145097303244863e-06, "loss": 0.1277, "step": 44314 }, { "epoch": 32.46520146520147, "grad_norm": 0.82421875, "learning_rate": 2.413696596484758e-06, "loss": 0.106, "step": 44315 }, { "epoch": 32.465934065934064, "grad_norm": 0.75390625, "learning_rate": 2.4128836930422964e-06, "loss": 0.115, "step": 44316 }, { "epoch": 32.46666666666667, "grad_norm": 1.046875, "learning_rate": 2.4120710200009924e-06, "loss": 0.1132, "step": 44317 }, { "epoch": 32.467399267399266, "grad_norm": 1.1171875, "learning_rate": 2.4112585773647422e-06, "loss": 0.1404, "step": 44318 }, { "epoch": 32.46813186813187, "grad_norm": 1.2421875, "learning_rate": 2.4104463651374372e-06, "loss": 0.1128, "step": 44319 }, { "epoch": 32.46886446886447, "grad_norm": 0.859375, "learning_rate": 2.4096343833229733e-06, "loss": 0.112, "step": 44320 }, { "epoch": 32.46959706959707, "grad_norm": 1.234375, "learning_rate": 2.408822631925236e-06, "loss": 0.1129, "step": 44321 }, { "epoch": 32.47032967032967, "grad_norm": 1.1484375, "learning_rate": 2.408011110948116e-06, "loss": 0.1092, "step": 44322 }, { "epoch": 32.47106227106227, "grad_norm": 1.09375, "learning_rate": 2.4071998203955053e-06, "loss": 0.1264, "step": 44323 }, { "epoch": 32.47179487179487, "grad_norm": 0.7265625, "learning_rate": 2.4063887602712825e-06, "loss": 0.1217, "step": 44324 }, { "epoch": 32.472527472527474, "grad_norm": 0.75, "learning_rate": 2.4055779305793448e-06, "loss": 0.1409, "step": 44325 }, { "epoch": 32.47326007326007, "grad_norm": 1.0703125, "learning_rate": 2.4047673313235717e-06, "loss": 0.1278, "step": 44326 }, { "epoch": 32.473992673992676, "grad_norm": 0.74609375, "learning_rate": 2.403956962507848e-06, "loss": 0.1141, "step": 44327 }, { "epoch": 32.47472527472527, "grad_norm": 0.984375, "learning_rate": 2.4031468241360495e-06, "loss": 0.1553, "step": 44328 }, { "epoch": 32.47545787545788, "grad_norm": 0.85546875, "learning_rate": 2.402336916212071e-06, "loss": 0.1237, "step": 44329 }, { "epoch": 32.476190476190474, "grad_norm": 0.79296875, "learning_rate": 2.4015272387397775e-06, "loss": 0.1272, "step": 44330 }, { "epoch": 32.47692307692308, "grad_norm": 0.69140625, "learning_rate": 2.4007177917230646e-06, "loss": 0.1, "step": 44331 }, { "epoch": 32.477655677655676, "grad_norm": 1.453125, "learning_rate": 2.3999085751658007e-06, "loss": 0.1073, "step": 44332 }, { "epoch": 32.47838827838828, "grad_norm": 1.3359375, "learning_rate": 2.399099589071861e-06, "loss": 0.0963, "step": 44333 }, { "epoch": 32.47912087912088, "grad_norm": 0.87109375, "learning_rate": 2.3982908334451248e-06, "loss": 0.1064, "step": 44334 }, { "epoch": 32.47985347985348, "grad_norm": 0.6796875, "learning_rate": 2.3974823082894728e-06, "loss": 0.1071, "step": 44335 }, { "epoch": 32.48058608058608, "grad_norm": 0.9375, "learning_rate": 2.396674013608773e-06, "loss": 0.1206, "step": 44336 }, { "epoch": 32.48131868131868, "grad_norm": 1.109375, "learning_rate": 2.395865949406901e-06, "loss": 0.133, "step": 44337 }, { "epoch": 32.48205128205128, "grad_norm": 2.015625, "learning_rate": 2.395058115687724e-06, "loss": 0.1142, "step": 44338 }, { "epoch": 32.482783882783885, "grad_norm": 1.3203125, "learning_rate": 2.3942505124551126e-06, "loss": 0.1324, "step": 44339 }, { "epoch": 32.48351648351648, "grad_norm": 1.1328125, "learning_rate": 2.3934431397129456e-06, "loss": 0.1245, "step": 44340 }, { "epoch": 32.484249084249086, "grad_norm": 0.69921875, "learning_rate": 2.3926359974650814e-06, "loss": 0.1078, "step": 44341 }, { "epoch": 32.48498168498168, "grad_norm": 0.828125, "learning_rate": 2.391829085715394e-06, "loss": 0.0944, "step": 44342 }, { "epoch": 32.48571428571429, "grad_norm": 0.921875, "learning_rate": 2.3910224044677418e-06, "loss": 0.1285, "step": 44343 }, { "epoch": 32.486446886446885, "grad_norm": 1.1015625, "learning_rate": 2.390215953725993e-06, "loss": 0.1038, "step": 44344 }, { "epoch": 32.48717948717949, "grad_norm": 1.2578125, "learning_rate": 2.389409733494017e-06, "loss": 0.1114, "step": 44345 }, { "epoch": 32.487912087912086, "grad_norm": 0.89453125, "learning_rate": 2.3886037437756776e-06, "loss": 0.1068, "step": 44346 }, { "epoch": 32.48864468864469, "grad_norm": 1.328125, "learning_rate": 2.3877979845748213e-06, "loss": 0.1606, "step": 44347 }, { "epoch": 32.48937728937729, "grad_norm": 1.265625, "learning_rate": 2.3869924558953223e-06, "loss": 0.1253, "step": 44348 }, { "epoch": 32.49010989010989, "grad_norm": 1.296875, "learning_rate": 2.3861871577410387e-06, "loss": 0.1121, "step": 44349 }, { "epoch": 32.49084249084249, "grad_norm": 0.7734375, "learning_rate": 2.385382090115823e-06, "loss": 0.1332, "step": 44350 }, { "epoch": 32.49157509157509, "grad_norm": 0.98046875, "learning_rate": 2.384577253023544e-06, "loss": 0.1352, "step": 44351 }, { "epoch": 32.49230769230769, "grad_norm": 1.453125, "learning_rate": 2.3837726464680484e-06, "loss": 0.1608, "step": 44352 }, { "epoch": 32.493040293040295, "grad_norm": 0.84375, "learning_rate": 2.3829682704531954e-06, "loss": 0.129, "step": 44353 }, { "epoch": 32.49377289377289, "grad_norm": 1.4765625, "learning_rate": 2.3821641249828243e-06, "loss": 0.1192, "step": 44354 }, { "epoch": 32.494505494505496, "grad_norm": 0.86328125, "learning_rate": 2.3813602100608166e-06, "loss": 0.1093, "step": 44355 }, { "epoch": 32.49523809523809, "grad_norm": 0.88671875, "learning_rate": 2.380556525691002e-06, "loss": 0.1096, "step": 44356 }, { "epoch": 32.4959706959707, "grad_norm": 0.8125, "learning_rate": 2.3797530718772388e-06, "loss": 0.1037, "step": 44357 }, { "epoch": 32.496703296703295, "grad_norm": 1.09375, "learning_rate": 2.3789498486233737e-06, "loss": 0.0954, "step": 44358 }, { "epoch": 32.4974358974359, "grad_norm": 1.1953125, "learning_rate": 2.378146855933259e-06, "loss": 0.1218, "step": 44359 }, { "epoch": 32.498168498168496, "grad_norm": 1.40625, "learning_rate": 2.377344093810737e-06, "loss": 0.1513, "step": 44360 }, { "epoch": 32.4989010989011, "grad_norm": 1.078125, "learning_rate": 2.3765415622596586e-06, "loss": 0.1144, "step": 44361 }, { "epoch": 32.4996336996337, "grad_norm": 0.796875, "learning_rate": 2.375739261283872e-06, "loss": 0.1179, "step": 44362 }, { "epoch": 32.5003663003663, "grad_norm": 1.2109375, "learning_rate": 2.3749371908872072e-06, "loss": 0.1149, "step": 44363 }, { "epoch": 32.5010989010989, "grad_norm": 1.046875, "learning_rate": 2.3741353510735274e-06, "loss": 0.1313, "step": 44364 }, { "epoch": 32.501831501831504, "grad_norm": 1.1875, "learning_rate": 2.3733337418466527e-06, "loss": 0.1227, "step": 44365 }, { "epoch": 32.5025641025641, "grad_norm": 0.99609375, "learning_rate": 2.3725323632104457e-06, "loss": 0.1091, "step": 44366 }, { "epoch": 32.503296703296705, "grad_norm": 0.96875, "learning_rate": 2.3717312151687264e-06, "loss": 0.1165, "step": 44367 }, { "epoch": 32.5040293040293, "grad_norm": 1.1328125, "learning_rate": 2.3709302977253526e-06, "loss": 0.1152, "step": 44368 }, { "epoch": 32.50476190476191, "grad_norm": 0.890625, "learning_rate": 2.370129610884143e-06, "loss": 0.0989, "step": 44369 }, { "epoch": 32.505494505494504, "grad_norm": 0.93359375, "learning_rate": 2.369329154648946e-06, "loss": 0.1323, "step": 44370 }, { "epoch": 32.50622710622711, "grad_norm": 1.1796875, "learning_rate": 2.3685289290235904e-06, "loss": 0.1089, "step": 44371 }, { "epoch": 32.506959706959705, "grad_norm": 1.3125, "learning_rate": 2.367728934011919e-06, "loss": 0.1518, "step": 44372 }, { "epoch": 32.50769230769231, "grad_norm": 1.0859375, "learning_rate": 2.366929169617751e-06, "loss": 0.1525, "step": 44373 }, { "epoch": 32.50842490842491, "grad_norm": 0.98046875, "learning_rate": 2.3661296358449324e-06, "loss": 0.0946, "step": 44374 }, { "epoch": 32.50915750915751, "grad_norm": 0.9375, "learning_rate": 2.3653303326972834e-06, "loss": 0.1082, "step": 44375 }, { "epoch": 32.50989010989011, "grad_norm": 0.96484375, "learning_rate": 2.36453126017864e-06, "loss": 0.1081, "step": 44376 }, { "epoch": 32.51062271062271, "grad_norm": 1.375, "learning_rate": 2.363732418292832e-06, "loss": 0.117, "step": 44377 }, { "epoch": 32.51135531135531, "grad_norm": 0.7421875, "learning_rate": 2.3629338070436796e-06, "loss": 0.113, "step": 44378 }, { "epoch": 32.512087912087914, "grad_norm": 0.796875, "learning_rate": 2.362135426435024e-06, "loss": 0.0948, "step": 44379 }, { "epoch": 32.51282051282051, "grad_norm": 1.0078125, "learning_rate": 2.3613372764706626e-06, "loss": 0.119, "step": 44380 }, { "epoch": 32.513553113553115, "grad_norm": 1.25, "learning_rate": 2.360539357154453e-06, "loss": 0.1415, "step": 44381 }, { "epoch": 32.51428571428571, "grad_norm": 1.28125, "learning_rate": 2.3597416684901987e-06, "loss": 0.1224, "step": 44382 }, { "epoch": 32.51501831501832, "grad_norm": 1.21875, "learning_rate": 2.358944210481724e-06, "loss": 0.1357, "step": 44383 }, { "epoch": 32.515750915750914, "grad_norm": 0.9609375, "learning_rate": 2.358146983132849e-06, "loss": 0.1101, "step": 44384 }, { "epoch": 32.51648351648352, "grad_norm": 0.7109375, "learning_rate": 2.3573499864473983e-06, "loss": 0.1075, "step": 44385 }, { "epoch": 32.517216117216115, "grad_norm": 1.2421875, "learning_rate": 2.356553220429186e-06, "loss": 0.0998, "step": 44386 }, { "epoch": 32.51794871794872, "grad_norm": 1.4140625, "learning_rate": 2.355756685082031e-06, "loss": 0.1408, "step": 44387 }, { "epoch": 32.51868131868132, "grad_norm": 0.93359375, "learning_rate": 2.354960380409754e-06, "loss": 0.114, "step": 44388 }, { "epoch": 32.51941391941392, "grad_norm": 1.1015625, "learning_rate": 2.3541643064161678e-06, "loss": 0.1294, "step": 44389 }, { "epoch": 32.52014652014652, "grad_norm": 1.4765625, "learning_rate": 2.3533684631050866e-06, "loss": 0.1131, "step": 44390 }, { "epoch": 32.52087912087912, "grad_norm": 1.28125, "learning_rate": 2.3525728504803076e-06, "loss": 0.1349, "step": 44391 }, { "epoch": 32.52161172161172, "grad_norm": 1.015625, "learning_rate": 2.351777468545673e-06, "loss": 0.1312, "step": 44392 }, { "epoch": 32.522344322344324, "grad_norm": 1.46875, "learning_rate": 2.3509823173049744e-06, "loss": 0.1146, "step": 44393 }, { "epoch": 32.52307692307692, "grad_norm": 1.609375, "learning_rate": 2.350187396762031e-06, "loss": 0.125, "step": 44394 }, { "epoch": 32.523809523809526, "grad_norm": 1.03125, "learning_rate": 2.349392706920641e-06, "loss": 0.1207, "step": 44395 }, { "epoch": 32.52454212454212, "grad_norm": 0.8828125, "learning_rate": 2.348598247784617e-06, "loss": 0.0916, "step": 44396 }, { "epoch": 32.52527472527473, "grad_norm": 1.234375, "learning_rate": 2.3478040193577635e-06, "loss": 0.0923, "step": 44397 }, { "epoch": 32.526007326007324, "grad_norm": 0.98046875, "learning_rate": 2.3470100216438987e-06, "loss": 0.1244, "step": 44398 }, { "epoch": 32.52673992673993, "grad_norm": 1.328125, "learning_rate": 2.346216254646804e-06, "loss": 0.1301, "step": 44399 }, { "epoch": 32.527472527472526, "grad_norm": 1.0546875, "learning_rate": 2.345422718370299e-06, "loss": 0.104, "step": 44400 }, { "epoch": 32.52820512820513, "grad_norm": 1.015625, "learning_rate": 2.344629412818182e-06, "loss": 0.1145, "step": 44401 }, { "epoch": 32.52893772893773, "grad_norm": 0.91015625, "learning_rate": 2.343836337994254e-06, "loss": 0.1417, "step": 44402 }, { "epoch": 32.52967032967033, "grad_norm": 0.79296875, "learning_rate": 2.343043493902319e-06, "loss": 0.1246, "step": 44403 }, { "epoch": 32.53040293040293, "grad_norm": 0.91015625, "learning_rate": 2.34225088054617e-06, "loss": 0.1226, "step": 44404 }, { "epoch": 32.53113553113553, "grad_norm": 0.9609375, "learning_rate": 2.341458497929608e-06, "loss": 0.1309, "step": 44405 }, { "epoch": 32.53186813186813, "grad_norm": 0.94140625, "learning_rate": 2.3406663460564153e-06, "loss": 0.1145, "step": 44406 }, { "epoch": 32.532600732600734, "grad_norm": 1.21875, "learning_rate": 2.339874424930416e-06, "loss": 0.1029, "step": 44407 }, { "epoch": 32.53333333333333, "grad_norm": 1.359375, "learning_rate": 2.339082734555376e-06, "loss": 0.1486, "step": 44408 }, { "epoch": 32.534065934065936, "grad_norm": 1.328125, "learning_rate": 2.338291274935113e-06, "loss": 0.1107, "step": 44409 }, { "epoch": 32.53479853479853, "grad_norm": 1.09375, "learning_rate": 2.337500046073398e-06, "loss": 0.1414, "step": 44410 }, { "epoch": 32.53553113553114, "grad_norm": 1.40625, "learning_rate": 2.336709047974033e-06, "loss": 0.1518, "step": 44411 }, { "epoch": 32.536263736263734, "grad_norm": 1.015625, "learning_rate": 2.335918280640801e-06, "loss": 0.1604, "step": 44412 }, { "epoch": 32.53699633699634, "grad_norm": 1.484375, "learning_rate": 2.335127744077503e-06, "loss": 0.1413, "step": 44413 }, { "epoch": 32.537728937728936, "grad_norm": 0.98046875, "learning_rate": 2.3343374382879204e-06, "loss": 0.1402, "step": 44414 }, { "epoch": 32.53846153846154, "grad_norm": 0.80078125, "learning_rate": 2.3335473632758292e-06, "loss": 0.1243, "step": 44415 }, { "epoch": 32.53919413919414, "grad_norm": 0.89453125, "learning_rate": 2.3327575190450377e-06, "loss": 0.0853, "step": 44416 }, { "epoch": 32.53992673992674, "grad_norm": 1.3828125, "learning_rate": 2.331967905599305e-06, "loss": 0.1077, "step": 44417 }, { "epoch": 32.54065934065934, "grad_norm": 1.6171875, "learning_rate": 2.331178522942433e-06, "loss": 0.1605, "step": 44418 }, { "epoch": 32.54139194139194, "grad_norm": 0.98046875, "learning_rate": 2.3303893710781983e-06, "loss": 0.1497, "step": 44419 }, { "epoch": 32.54212454212454, "grad_norm": 1.234375, "learning_rate": 2.329600450010382e-06, "loss": 0.1065, "step": 44420 }, { "epoch": 32.542857142857144, "grad_norm": 1.3359375, "learning_rate": 2.3288117597427586e-06, "loss": 0.1274, "step": 44421 }, { "epoch": 32.54358974358974, "grad_norm": 0.703125, "learning_rate": 2.32802330027911e-06, "loss": 0.1112, "step": 44422 }, { "epoch": 32.544322344322346, "grad_norm": 1.03125, "learning_rate": 2.3272350716232225e-06, "loss": 0.1049, "step": 44423 }, { "epoch": 32.54505494505494, "grad_norm": 0.88671875, "learning_rate": 2.3264470737788653e-06, "loss": 0.1347, "step": 44424 }, { "epoch": 32.54578754578755, "grad_norm": 0.70703125, "learning_rate": 2.3256593067498092e-06, "loss": 0.1293, "step": 44425 }, { "epoch": 32.546520146520145, "grad_norm": 1.265625, "learning_rate": 2.324871770539835e-06, "loss": 0.1132, "step": 44426 }, { "epoch": 32.54725274725275, "grad_norm": 1.4453125, "learning_rate": 2.324084465152713e-06, "loss": 0.0863, "step": 44427 }, { "epoch": 32.547985347985346, "grad_norm": 1.0546875, "learning_rate": 2.3232973905922233e-06, "loss": 0.1241, "step": 44428 }, { "epoch": 32.54871794871795, "grad_norm": 1.515625, "learning_rate": 2.3225105468621314e-06, "loss": 0.1444, "step": 44429 }, { "epoch": 32.54945054945055, "grad_norm": 0.97265625, "learning_rate": 2.3217239339662004e-06, "loss": 0.103, "step": 44430 }, { "epoch": 32.55018315018315, "grad_norm": 1.2578125, "learning_rate": 2.3209375519082184e-06, "loss": 0.099, "step": 44431 }, { "epoch": 32.55091575091575, "grad_norm": 0.828125, "learning_rate": 2.3201514006919214e-06, "loss": 0.1098, "step": 44432 }, { "epoch": 32.55164835164835, "grad_norm": 1.4921875, "learning_rate": 2.3193654803211126e-06, "loss": 0.1305, "step": 44433 }, { "epoch": 32.55238095238095, "grad_norm": 1.1171875, "learning_rate": 2.3185797907995346e-06, "loss": 0.1385, "step": 44434 }, { "epoch": 32.553113553113555, "grad_norm": 1.140625, "learning_rate": 2.3177943321309575e-06, "loss": 0.1206, "step": 44435 }, { "epoch": 32.55384615384615, "grad_norm": 0.796875, "learning_rate": 2.3170091043191455e-06, "loss": 0.1264, "step": 44436 }, { "epoch": 32.554578754578756, "grad_norm": 0.921875, "learning_rate": 2.3162241073678573e-06, "loss": 0.102, "step": 44437 }, { "epoch": 32.55531135531135, "grad_norm": 1.1875, "learning_rate": 2.315439341280858e-06, "loss": 0.131, "step": 44438 }, { "epoch": 32.55604395604396, "grad_norm": 1.0, "learning_rate": 2.314654806061907e-06, "loss": 0.1313, "step": 44439 }, { "epoch": 32.556776556776555, "grad_norm": 1.0, "learning_rate": 2.3138705017147675e-06, "loss": 0.1157, "step": 44440 }, { "epoch": 32.55750915750916, "grad_norm": 1.03125, "learning_rate": 2.3130864282431885e-06, "loss": 0.1404, "step": 44441 }, { "epoch": 32.558241758241756, "grad_norm": 1.0625, "learning_rate": 2.312302585650939e-06, "loss": 0.1323, "step": 44442 }, { "epoch": 32.55897435897436, "grad_norm": 0.62890625, "learning_rate": 2.311518973941752e-06, "loss": 0.1124, "step": 44443 }, { "epoch": 32.55970695970696, "grad_norm": 1.015625, "learning_rate": 2.310735593119407e-06, "loss": 0.1304, "step": 44444 }, { "epoch": 32.56043956043956, "grad_norm": 0.90234375, "learning_rate": 2.3099524431876477e-06, "loss": 0.1152, "step": 44445 }, { "epoch": 32.56117216117216, "grad_norm": 0.74609375, "learning_rate": 2.3091695241502266e-06, "loss": 0.0934, "step": 44446 }, { "epoch": 32.56190476190476, "grad_norm": 1.1484375, "learning_rate": 2.3083868360108872e-06, "loss": 0.0957, "step": 44447 }, { "epoch": 32.56263736263736, "grad_norm": 1.0859375, "learning_rate": 2.3076043787733933e-06, "loss": 0.1218, "step": 44448 }, { "epoch": 32.563369963369965, "grad_norm": 0.8828125, "learning_rate": 2.306822152441482e-06, "loss": 0.1128, "step": 44449 }, { "epoch": 32.56410256410256, "grad_norm": 0.66015625, "learning_rate": 2.3060401570189065e-06, "loss": 0.0905, "step": 44450 }, { "epoch": 32.564835164835166, "grad_norm": 1.328125, "learning_rate": 2.3052583925094205e-06, "loss": 0.1293, "step": 44451 }, { "epoch": 32.565567765567764, "grad_norm": 1.7109375, "learning_rate": 2.304476858916762e-06, "loss": 0.146, "step": 44452 }, { "epoch": 32.56630036630037, "grad_norm": 1.796875, "learning_rate": 2.303695556244672e-06, "loss": 0.1453, "step": 44453 }, { "epoch": 32.567032967032965, "grad_norm": 0.77734375, "learning_rate": 2.3029144844968937e-06, "loss": 0.1131, "step": 44454 }, { "epoch": 32.56776556776557, "grad_norm": 1.0390625, "learning_rate": 2.3021336436771864e-06, "loss": 0.1085, "step": 44455 }, { "epoch": 32.56849816849817, "grad_norm": 0.96484375, "learning_rate": 2.30135303378927e-06, "loss": 0.1237, "step": 44456 }, { "epoch": 32.56923076923077, "grad_norm": 1.21875, "learning_rate": 2.300572654836904e-06, "loss": 0.1661, "step": 44457 }, { "epoch": 32.56996336996337, "grad_norm": 1.1484375, "learning_rate": 2.299792506823803e-06, "loss": 0.1442, "step": 44458 }, { "epoch": 32.57069597069597, "grad_norm": 0.77734375, "learning_rate": 2.299012589753726e-06, "loss": 0.109, "step": 44459 }, { "epoch": 32.57142857142857, "grad_norm": 1.53125, "learning_rate": 2.2982329036304054e-06, "loss": 0.1045, "step": 44460 }, { "epoch": 32.572161172161174, "grad_norm": 0.94140625, "learning_rate": 2.297453448457577e-06, "loss": 0.124, "step": 44461 }, { "epoch": 32.57289377289377, "grad_norm": 0.95703125, "learning_rate": 2.2966742242389675e-06, "loss": 0.1286, "step": 44462 }, { "epoch": 32.573626373626375, "grad_norm": 1.0390625, "learning_rate": 2.295895230978314e-06, "loss": 0.0987, "step": 44463 }, { "epoch": 32.57435897435897, "grad_norm": 1.5546875, "learning_rate": 2.295116468679353e-06, "loss": 0.1521, "step": 44464 }, { "epoch": 32.57509157509158, "grad_norm": 1.0234375, "learning_rate": 2.2943379373458106e-06, "loss": 0.1301, "step": 44465 }, { "epoch": 32.575824175824174, "grad_norm": 0.83984375, "learning_rate": 2.29355963698143e-06, "loss": 0.101, "step": 44466 }, { "epoch": 32.57655677655678, "grad_norm": 1.3828125, "learning_rate": 2.2927815675899195e-06, "loss": 0.1392, "step": 44467 }, { "epoch": 32.577289377289375, "grad_norm": 0.65234375, "learning_rate": 2.292003729175029e-06, "loss": 0.1104, "step": 44468 }, { "epoch": 32.57802197802198, "grad_norm": 1.21875, "learning_rate": 2.2912261217404605e-06, "loss": 0.1353, "step": 44469 }, { "epoch": 32.57875457875458, "grad_norm": 1.0625, "learning_rate": 2.290448745289963e-06, "loss": 0.1494, "step": 44470 }, { "epoch": 32.57948717948718, "grad_norm": 1.234375, "learning_rate": 2.289671599827251e-06, "loss": 0.1365, "step": 44471 }, { "epoch": 32.58021978021978, "grad_norm": 1.15625, "learning_rate": 2.2888946853560515e-06, "loss": 0.1311, "step": 44472 }, { "epoch": 32.58095238095238, "grad_norm": 1.3359375, "learning_rate": 2.288118001880079e-06, "loss": 0.0931, "step": 44473 }, { "epoch": 32.58168498168498, "grad_norm": 0.83984375, "learning_rate": 2.287341549403059e-06, "loss": 0.1233, "step": 44474 }, { "epoch": 32.582417582417584, "grad_norm": 1.4296875, "learning_rate": 2.2865653279287127e-06, "loss": 0.1012, "step": 44475 }, { "epoch": 32.58315018315018, "grad_norm": 1.2421875, "learning_rate": 2.285789337460761e-06, "loss": 0.1503, "step": 44476 }, { "epoch": 32.583882783882785, "grad_norm": 0.95703125, "learning_rate": 2.285013578002923e-06, "loss": 0.1282, "step": 44477 }, { "epoch": 32.58461538461538, "grad_norm": 1.4453125, "learning_rate": 2.2842380495589095e-06, "loss": 0.1198, "step": 44478 }, { "epoch": 32.58534798534799, "grad_norm": 1.6171875, "learning_rate": 2.2834627521324406e-06, "loss": 0.1583, "step": 44479 }, { "epoch": 32.586080586080584, "grad_norm": 1.0703125, "learning_rate": 2.2826876857272255e-06, "loss": 0.13, "step": 44480 }, { "epoch": 32.58681318681319, "grad_norm": 1.21875, "learning_rate": 2.281912850346991e-06, "loss": 0.1184, "step": 44481 }, { "epoch": 32.587545787545785, "grad_norm": 1.0078125, "learning_rate": 2.28113824599544e-06, "loss": 0.126, "step": 44482 }, { "epoch": 32.58827838827839, "grad_norm": 1.2890625, "learning_rate": 2.2803638726762834e-06, "loss": 0.1465, "step": 44483 }, { "epoch": 32.58901098901099, "grad_norm": 0.9609375, "learning_rate": 2.279589730393224e-06, "loss": 0.1305, "step": 44484 }, { "epoch": 32.58974358974359, "grad_norm": 1.15625, "learning_rate": 2.2788158191499943e-06, "loss": 0.108, "step": 44485 }, { "epoch": 32.59047619047619, "grad_norm": 0.89453125, "learning_rate": 2.278042138950281e-06, "loss": 0.1243, "step": 44486 }, { "epoch": 32.59120879120879, "grad_norm": 0.63671875, "learning_rate": 2.2772686897978e-06, "loss": 0.0981, "step": 44487 }, { "epoch": 32.59194139194139, "grad_norm": 1.15625, "learning_rate": 2.276495471696255e-06, "loss": 0.1257, "step": 44488 }, { "epoch": 32.592673992673994, "grad_norm": 1.5, "learning_rate": 2.275722484649344e-06, "loss": 0.1298, "step": 44489 }, { "epoch": 32.59340659340659, "grad_norm": 1.03125, "learning_rate": 2.2749497286607823e-06, "loss": 0.1139, "step": 44490 }, { "epoch": 32.594139194139196, "grad_norm": 0.93359375, "learning_rate": 2.2741772037342693e-06, "loss": 0.0909, "step": 44491 }, { "epoch": 32.59487179487179, "grad_norm": 1.609375, "learning_rate": 2.2734049098735074e-06, "loss": 0.149, "step": 44492 }, { "epoch": 32.5956043956044, "grad_norm": 0.65234375, "learning_rate": 2.2726328470821913e-06, "loss": 0.1245, "step": 44493 }, { "epoch": 32.596336996336994, "grad_norm": 1.296875, "learning_rate": 2.271861015364024e-06, "loss": 0.1406, "step": 44494 }, { "epoch": 32.5970695970696, "grad_norm": 0.91015625, "learning_rate": 2.2710894147226937e-06, "loss": 0.1077, "step": 44495 }, { "epoch": 32.597802197802196, "grad_norm": 0.796875, "learning_rate": 2.2703180451619203e-06, "loss": 0.1166, "step": 44496 }, { "epoch": 32.5985347985348, "grad_norm": 0.66015625, "learning_rate": 2.2695469066853807e-06, "loss": 0.1269, "step": 44497 }, { "epoch": 32.5992673992674, "grad_norm": 0.8671875, "learning_rate": 2.268775999296779e-06, "loss": 0.1165, "step": 44498 }, { "epoch": 32.6, "grad_norm": 0.82421875, "learning_rate": 2.268005322999797e-06, "loss": 0.1229, "step": 44499 }, { "epoch": 32.6007326007326, "grad_norm": 1.5703125, "learning_rate": 2.267234877798139e-06, "loss": 0.1528, "step": 44500 }, { "epoch": 32.6014652014652, "grad_norm": 0.9140625, "learning_rate": 2.266464663695487e-06, "loss": 0.1297, "step": 44501 }, { "epoch": 32.6021978021978, "grad_norm": 1.078125, "learning_rate": 2.265694680695539e-06, "loss": 0.1127, "step": 44502 }, { "epoch": 32.602930402930404, "grad_norm": 0.8203125, "learning_rate": 2.264924928801989e-06, "loss": 0.1284, "step": 44503 }, { "epoch": 32.603663003663, "grad_norm": 0.8671875, "learning_rate": 2.2641554080185126e-06, "loss": 0.1219, "step": 44504 }, { "epoch": 32.604395604395606, "grad_norm": 1.2421875, "learning_rate": 2.2633861183488036e-06, "loss": 0.123, "step": 44505 }, { "epoch": 32.6051282051282, "grad_norm": 0.86328125, "learning_rate": 2.262617059796544e-06, "loss": 0.1065, "step": 44506 }, { "epoch": 32.60586080586081, "grad_norm": 1.078125, "learning_rate": 2.2618482323654267e-06, "loss": 0.1133, "step": 44507 }, { "epoch": 32.606593406593404, "grad_norm": 0.78125, "learning_rate": 2.2610796360591276e-06, "loss": 0.1167, "step": 44508 }, { "epoch": 32.60732600732601, "grad_norm": 1.109375, "learning_rate": 2.260311270881335e-06, "loss": 0.1161, "step": 44509 }, { "epoch": 32.608058608058606, "grad_norm": 1.5, "learning_rate": 2.2595431368357203e-06, "loss": 0.1241, "step": 44510 }, { "epoch": 32.60879120879121, "grad_norm": 1.171875, "learning_rate": 2.2587752339259757e-06, "loss": 0.1093, "step": 44511 }, { "epoch": 32.60952380952381, "grad_norm": 0.84765625, "learning_rate": 2.2580075621557722e-06, "loss": 0.1347, "step": 44512 }, { "epoch": 32.61025641025641, "grad_norm": 1.296875, "learning_rate": 2.2572401215287984e-06, "loss": 0.1629, "step": 44513 }, { "epoch": 32.61098901098901, "grad_norm": 0.74609375, "learning_rate": 2.256472912048719e-06, "loss": 0.1124, "step": 44514 }, { "epoch": 32.61172161172161, "grad_norm": 0.9765625, "learning_rate": 2.2557059337192107e-06, "loss": 0.1197, "step": 44515 }, { "epoch": 32.61245421245421, "grad_norm": 0.91796875, "learning_rate": 2.2549391865439555e-06, "loss": 0.1203, "step": 44516 }, { "epoch": 32.613186813186815, "grad_norm": 1.1953125, "learning_rate": 2.2541726705266255e-06, "loss": 0.1406, "step": 44517 }, { "epoch": 32.61391941391941, "grad_norm": 1.1015625, "learning_rate": 2.2534063856708955e-06, "loss": 0.1226, "step": 44518 }, { "epoch": 32.614652014652016, "grad_norm": 0.88671875, "learning_rate": 2.252640331980432e-06, "loss": 0.114, "step": 44519 }, { "epoch": 32.61538461538461, "grad_norm": 1.1171875, "learning_rate": 2.2518745094589116e-06, "loss": 0.1045, "step": 44520 }, { "epoch": 32.61611721611722, "grad_norm": 0.7421875, "learning_rate": 2.251108918109989e-06, "loss": 0.1173, "step": 44521 }, { "epoch": 32.616849816849815, "grad_norm": 0.53515625, "learning_rate": 2.250343557937351e-06, "loss": 0.122, "step": 44522 }, { "epoch": 32.61758241758242, "grad_norm": 1.4140625, "learning_rate": 2.2495784289446528e-06, "loss": 0.1452, "step": 44523 }, { "epoch": 32.618315018315016, "grad_norm": 1.34375, "learning_rate": 2.2488135311355655e-06, "loss": 0.1002, "step": 44524 }, { "epoch": 32.61904761904762, "grad_norm": 0.73046875, "learning_rate": 2.2480488645137493e-06, "loss": 0.1167, "step": 44525 }, { "epoch": 32.61978021978022, "grad_norm": 1.53125, "learning_rate": 2.2472844290828748e-06, "loss": 0.1315, "step": 44526 }, { "epoch": 32.62051282051282, "grad_norm": 0.94921875, "learning_rate": 2.246520224846597e-06, "loss": 0.1095, "step": 44527 }, { "epoch": 32.62124542124542, "grad_norm": 1.015625, "learning_rate": 2.2457562518085806e-06, "loss": 0.117, "step": 44528 }, { "epoch": 32.62197802197802, "grad_norm": 0.91015625, "learning_rate": 2.244992509972492e-06, "loss": 0.1108, "step": 44529 }, { "epoch": 32.62271062271062, "grad_norm": 1.1015625, "learning_rate": 2.244228999341986e-06, "loss": 0.0953, "step": 44530 }, { "epoch": 32.623443223443225, "grad_norm": 1.5, "learning_rate": 2.2434657199207164e-06, "loss": 0.1143, "step": 44531 }, { "epoch": 32.62417582417582, "grad_norm": 1.171875, "learning_rate": 2.2427026717123435e-06, "loss": 0.1029, "step": 44532 }, { "epoch": 32.624908424908426, "grad_norm": 0.87109375, "learning_rate": 2.2419398547205273e-06, "loss": 0.1318, "step": 44533 }, { "epoch": 32.62564102564102, "grad_norm": 1.3125, "learning_rate": 2.241177268948918e-06, "loss": 0.1142, "step": 44534 }, { "epoch": 32.62637362637363, "grad_norm": 0.82421875, "learning_rate": 2.2404149144011737e-06, "loss": 0.1141, "step": 44535 }, { "epoch": 32.627106227106225, "grad_norm": 1.4375, "learning_rate": 2.2396527910809402e-06, "loss": 0.1433, "step": 44536 }, { "epoch": 32.62783882783883, "grad_norm": 0.9140625, "learning_rate": 2.23889089899187e-06, "loss": 0.1131, "step": 44537 }, { "epoch": 32.628571428571426, "grad_norm": 0.90234375, "learning_rate": 2.238129238137619e-06, "loss": 0.1116, "step": 44538 }, { "epoch": 32.62930402930403, "grad_norm": 1.0234375, "learning_rate": 2.2373678085218355e-06, "loss": 0.1264, "step": 44539 }, { "epoch": 32.63003663003663, "grad_norm": 1.0078125, "learning_rate": 2.2366066101481697e-06, "loss": 0.1197, "step": 44540 }, { "epoch": 32.63076923076923, "grad_norm": 1.1953125, "learning_rate": 2.235845643020258e-06, "loss": 0.1173, "step": 44541 }, { "epoch": 32.63150183150183, "grad_norm": 1.34375, "learning_rate": 2.2350849071417565e-06, "loss": 0.1166, "step": 44542 }, { "epoch": 32.63223443223443, "grad_norm": 1.0859375, "learning_rate": 2.2343244025163082e-06, "loss": 0.1133, "step": 44543 }, { "epoch": 32.63296703296703, "grad_norm": 0.8671875, "learning_rate": 2.233564129147562e-06, "loss": 0.112, "step": 44544 }, { "epoch": 32.633699633699635, "grad_norm": 1.1875, "learning_rate": 2.2328040870391453e-06, "loss": 0.1183, "step": 44545 }, { "epoch": 32.63443223443223, "grad_norm": 0.8671875, "learning_rate": 2.232044276194718e-06, "loss": 0.103, "step": 44546 }, { "epoch": 32.63516483516484, "grad_norm": 1.0078125, "learning_rate": 2.2312846966179017e-06, "loss": 0.1176, "step": 44547 }, { "epoch": 32.635897435897434, "grad_norm": 1.3828125, "learning_rate": 2.2305253483123566e-06, "loss": 0.1391, "step": 44548 }, { "epoch": 32.63663003663004, "grad_norm": 1.21875, "learning_rate": 2.2297662312817043e-06, "loss": 0.1153, "step": 44549 }, { "epoch": 32.637362637362635, "grad_norm": 1.203125, "learning_rate": 2.2290073455295943e-06, "loss": 0.116, "step": 44550 }, { "epoch": 32.63809523809524, "grad_norm": 0.97265625, "learning_rate": 2.2282486910596525e-06, "loss": 0.1347, "step": 44551 }, { "epoch": 32.63882783882784, "grad_norm": 0.73046875, "learning_rate": 2.227490267875512e-06, "loss": 0.1244, "step": 44552 }, { "epoch": 32.63956043956044, "grad_norm": 0.63671875, "learning_rate": 2.2267320759808173e-06, "loss": 0.0986, "step": 44553 }, { "epoch": 32.64029304029304, "grad_norm": 0.765625, "learning_rate": 2.2259741153791944e-06, "loss": 0.1092, "step": 44554 }, { "epoch": 32.64102564102564, "grad_norm": 1.296875, "learning_rate": 2.2252163860742816e-06, "loss": 0.1461, "step": 44555 }, { "epoch": 32.64175824175824, "grad_norm": 1.109375, "learning_rate": 2.224458888069701e-06, "loss": 0.1117, "step": 44556 }, { "epoch": 32.642490842490844, "grad_norm": 1.4921875, "learning_rate": 2.223701621369085e-06, "loss": 0.148, "step": 44557 }, { "epoch": 32.64322344322344, "grad_norm": 1.4375, "learning_rate": 2.222944585976066e-06, "loss": 0.1303, "step": 44558 }, { "epoch": 32.643956043956045, "grad_norm": 0.73046875, "learning_rate": 2.2221877818942713e-06, "loss": 0.1303, "step": 44559 }, { "epoch": 32.64468864468864, "grad_norm": 1.375, "learning_rate": 2.2214312091273163e-06, "loss": 0.117, "step": 44560 }, { "epoch": 32.64542124542125, "grad_norm": 0.7890625, "learning_rate": 2.2206748676788403e-06, "loss": 0.088, "step": 44561 }, { "epoch": 32.646153846153844, "grad_norm": 0.94140625, "learning_rate": 2.2199187575524525e-06, "loss": 0.1253, "step": 44562 }, { "epoch": 32.64688644688645, "grad_norm": 1.1484375, "learning_rate": 2.2191628787517816e-06, "loss": 0.1187, "step": 44563 }, { "epoch": 32.647619047619045, "grad_norm": 1.0, "learning_rate": 2.2184072312804538e-06, "loss": 0.1116, "step": 44564 }, { "epoch": 32.64835164835165, "grad_norm": 0.78515625, "learning_rate": 2.2176518151420904e-06, "loss": 0.1233, "step": 44565 }, { "epoch": 32.64908424908425, "grad_norm": 1.1015625, "learning_rate": 2.216896630340308e-06, "loss": 0.1039, "step": 44566 }, { "epoch": 32.64981684981685, "grad_norm": 1.1640625, "learning_rate": 2.2161416768787173e-06, "loss": 0.1608, "step": 44567 }, { "epoch": 32.65054945054945, "grad_norm": 0.7890625, "learning_rate": 2.215386954760945e-06, "loss": 0.1178, "step": 44568 }, { "epoch": 32.65128205128205, "grad_norm": 0.67578125, "learning_rate": 2.2146324639906076e-06, "loss": 0.1022, "step": 44569 }, { "epoch": 32.65201465201465, "grad_norm": 1.09375, "learning_rate": 2.213878204571315e-06, "loss": 0.1242, "step": 44570 }, { "epoch": 32.652747252747254, "grad_norm": 0.6953125, "learning_rate": 2.2131241765066846e-06, "loss": 0.0987, "step": 44571 }, { "epoch": 32.65347985347985, "grad_norm": 1.0234375, "learning_rate": 2.2123703798003314e-06, "loss": 0.1094, "step": 44572 }, { "epoch": 32.654212454212455, "grad_norm": 0.7734375, "learning_rate": 2.2116168144558498e-06, "loss": 0.1103, "step": 44573 }, { "epoch": 32.65494505494505, "grad_norm": 1.03125, "learning_rate": 2.2108634804768782e-06, "loss": 0.1386, "step": 44574 }, { "epoch": 32.65567765567766, "grad_norm": 0.81640625, "learning_rate": 2.210110377867005e-06, "loss": 0.0968, "step": 44575 }, { "epoch": 32.656410256410254, "grad_norm": 1.2890625, "learning_rate": 2.2093575066298514e-06, "loss": 0.1378, "step": 44576 }, { "epoch": 32.65714285714286, "grad_norm": 1.46875, "learning_rate": 2.2086048667690115e-06, "loss": 0.113, "step": 44577 }, { "epoch": 32.657875457875456, "grad_norm": 0.9609375, "learning_rate": 2.2078524582881028e-06, "loss": 0.1418, "step": 44578 }, { "epoch": 32.65860805860806, "grad_norm": 0.8671875, "learning_rate": 2.2071002811907235e-06, "loss": 0.1272, "step": 44579 }, { "epoch": 32.65934065934066, "grad_norm": 0.92578125, "learning_rate": 2.206348335480479e-06, "loss": 0.1094, "step": 44580 }, { "epoch": 32.66007326007326, "grad_norm": 1.03125, "learning_rate": 2.2055966211609803e-06, "loss": 0.1594, "step": 44581 }, { "epoch": 32.66080586080586, "grad_norm": 1.1953125, "learning_rate": 2.2048451382358158e-06, "loss": 0.1527, "step": 44582 }, { "epoch": 32.66153846153846, "grad_norm": 1.4375, "learning_rate": 2.204093886708596e-06, "loss": 0.1394, "step": 44583 }, { "epoch": 32.66227106227106, "grad_norm": 1.234375, "learning_rate": 2.203342866582915e-06, "loss": 0.1147, "step": 44584 }, { "epoch": 32.663003663003664, "grad_norm": 1.7265625, "learning_rate": 2.202592077862378e-06, "loss": 0.1392, "step": 44585 }, { "epoch": 32.66373626373626, "grad_norm": 1.0546875, "learning_rate": 2.201841520550569e-06, "loss": 0.1413, "step": 44586 }, { "epoch": 32.664468864468866, "grad_norm": 1.921875, "learning_rate": 2.2010911946511024e-06, "loss": 0.1231, "step": 44587 }, { "epoch": 32.66520146520146, "grad_norm": 1.2734375, "learning_rate": 2.200341100167557e-06, "loss": 0.1181, "step": 44588 }, { "epoch": 32.66593406593407, "grad_norm": 1.234375, "learning_rate": 2.199591237103532e-06, "loss": 0.0923, "step": 44589 }, { "epoch": 32.666666666666664, "grad_norm": 0.9453125, "learning_rate": 2.198841605462622e-06, "loss": 0.1267, "step": 44590 }, { "epoch": 32.66739926739927, "grad_norm": 1.015625, "learning_rate": 2.1980922052484205e-06, "loss": 0.1275, "step": 44591 }, { "epoch": 32.668131868131866, "grad_norm": 1.296875, "learning_rate": 2.1973430364645166e-06, "loss": 0.1424, "step": 44592 }, { "epoch": 32.66886446886447, "grad_norm": 0.98046875, "learning_rate": 2.1965940991144987e-06, "loss": 0.0965, "step": 44593 }, { "epoch": 32.66959706959707, "grad_norm": 0.8359375, "learning_rate": 2.19584539320195e-06, "loss": 0.097, "step": 44594 }, { "epoch": 32.67032967032967, "grad_norm": 1.1953125, "learning_rate": 2.1950969187304647e-06, "loss": 0.0967, "step": 44595 }, { "epoch": 32.67106227106227, "grad_norm": 1.015625, "learning_rate": 2.194348675703636e-06, "loss": 0.1095, "step": 44596 }, { "epoch": 32.67179487179487, "grad_norm": 0.765625, "learning_rate": 2.1936006641250326e-06, "loss": 0.1115, "step": 44597 }, { "epoch": 32.67252747252747, "grad_norm": 0.9453125, "learning_rate": 2.1928528839982525e-06, "loss": 0.1322, "step": 44598 }, { "epoch": 32.673260073260074, "grad_norm": 0.83203125, "learning_rate": 2.1921053353268625e-06, "loss": 0.1044, "step": 44599 }, { "epoch": 32.67399267399267, "grad_norm": 0.9296875, "learning_rate": 2.191358018114463e-06, "loss": 0.1047, "step": 44600 }, { "epoch": 32.674725274725276, "grad_norm": 1.1484375, "learning_rate": 2.1906109323646194e-06, "loss": 0.1322, "step": 44601 }, { "epoch": 32.67545787545787, "grad_norm": 0.62890625, "learning_rate": 2.189864078080927e-06, "loss": 0.1104, "step": 44602 }, { "epoch": 32.67619047619048, "grad_norm": 1.1015625, "learning_rate": 2.1891174552669465e-06, "loss": 0.112, "step": 44603 }, { "epoch": 32.676923076923075, "grad_norm": 0.79296875, "learning_rate": 2.1883710639262664e-06, "loss": 0.1199, "step": 44604 }, { "epoch": 32.67765567765568, "grad_norm": 1.1875, "learning_rate": 2.187624904062459e-06, "loss": 0.1004, "step": 44605 }, { "epoch": 32.678388278388276, "grad_norm": 0.69140625, "learning_rate": 2.1868789756790962e-06, "loss": 0.1245, "step": 44606 }, { "epoch": 32.67912087912088, "grad_norm": 1.1796875, "learning_rate": 2.1861332787797674e-06, "loss": 0.1127, "step": 44607 }, { "epoch": 32.67985347985348, "grad_norm": 0.83984375, "learning_rate": 2.1853878133680278e-06, "loss": 0.1128, "step": 44608 }, { "epoch": 32.68058608058608, "grad_norm": 1.7890625, "learning_rate": 2.184642579447454e-06, "loss": 0.1509, "step": 44609 }, { "epoch": 32.68131868131868, "grad_norm": 0.78515625, "learning_rate": 2.183897577021625e-06, "loss": 0.1243, "step": 44610 }, { "epoch": 32.68205128205128, "grad_norm": 1.1953125, "learning_rate": 2.183152806094102e-06, "loss": 0.1546, "step": 44611 }, { "epoch": 32.68278388278388, "grad_norm": 1.0390625, "learning_rate": 2.1824082666684506e-06, "loss": 0.1363, "step": 44612 }, { "epoch": 32.683516483516485, "grad_norm": 1.1953125, "learning_rate": 2.181663958748249e-06, "loss": 0.097, "step": 44613 }, { "epoch": 32.68424908424908, "grad_norm": 1.15625, "learning_rate": 2.1809198823370535e-06, "loss": 0.1502, "step": 44614 }, { "epoch": 32.684981684981686, "grad_norm": 1.0390625, "learning_rate": 2.1801760374384345e-06, "loss": 0.1373, "step": 44615 }, { "epoch": 32.68571428571428, "grad_norm": 1.6796875, "learning_rate": 2.1794324240559496e-06, "loss": 0.1362, "step": 44616 }, { "epoch": 32.68644688644689, "grad_norm": 1.328125, "learning_rate": 2.1786890421931704e-06, "loss": 0.103, "step": 44617 }, { "epoch": 32.687179487179485, "grad_norm": 1.2265625, "learning_rate": 2.177945891853657e-06, "loss": 0.126, "step": 44618 }, { "epoch": 32.68791208791209, "grad_norm": 1.140625, "learning_rate": 2.177202973040966e-06, "loss": 0.1093, "step": 44619 }, { "epoch": 32.688644688644686, "grad_norm": 0.66796875, "learning_rate": 2.1764602857586587e-06, "loss": 0.1067, "step": 44620 }, { "epoch": 32.68937728937729, "grad_norm": 0.76171875, "learning_rate": 2.17571783001029e-06, "loss": 0.126, "step": 44621 }, { "epoch": 32.69010989010989, "grad_norm": 1.046875, "learning_rate": 2.174975605799433e-06, "loss": 0.131, "step": 44622 }, { "epoch": 32.69084249084249, "grad_norm": 1.328125, "learning_rate": 2.1742336131296203e-06, "loss": 0.1213, "step": 44623 }, { "epoch": 32.69157509157509, "grad_norm": 0.859375, "learning_rate": 2.1734918520044246e-06, "loss": 0.0988, "step": 44624 }, { "epoch": 32.69230769230769, "grad_norm": 1.078125, "learning_rate": 2.1727503224273855e-06, "loss": 0.1169, "step": 44625 }, { "epoch": 32.69304029304029, "grad_norm": 0.99609375, "learning_rate": 2.172009024402074e-06, "loss": 0.0885, "step": 44626 }, { "epoch": 32.693772893772895, "grad_norm": 1.0546875, "learning_rate": 2.1712679579320306e-06, "loss": 0.1146, "step": 44627 }, { "epoch": 32.69450549450549, "grad_norm": 1.046875, "learning_rate": 2.1705271230208044e-06, "loss": 0.1288, "step": 44628 }, { "epoch": 32.695238095238096, "grad_norm": 1.0078125, "learning_rate": 2.169786519671952e-06, "loss": 0.1124, "step": 44629 }, { "epoch": 32.69597069597069, "grad_norm": 0.73046875, "learning_rate": 2.1690461478890116e-06, "loss": 0.1594, "step": 44630 }, { "epoch": 32.6967032967033, "grad_norm": 1.046875, "learning_rate": 2.1683060076755447e-06, "loss": 0.0986, "step": 44631 }, { "epoch": 32.697435897435895, "grad_norm": 0.93359375, "learning_rate": 2.1675660990350855e-06, "loss": 0.1336, "step": 44632 }, { "epoch": 32.6981684981685, "grad_norm": 0.921875, "learning_rate": 2.166826421971183e-06, "loss": 0.1428, "step": 44633 }, { "epoch": 32.698901098901096, "grad_norm": 1.15625, "learning_rate": 2.1660869764873837e-06, "loss": 0.1482, "step": 44634 }, { "epoch": 32.6996336996337, "grad_norm": 0.8203125, "learning_rate": 2.1653477625872302e-06, "loss": 0.11, "step": 44635 }, { "epoch": 32.7003663003663, "grad_norm": 0.91015625, "learning_rate": 2.1646087802742575e-06, "loss": 0.1234, "step": 44636 }, { "epoch": 32.7010989010989, "grad_norm": 0.9453125, "learning_rate": 2.163870029552021e-06, "loss": 0.1312, "step": 44637 }, { "epoch": 32.7018315018315, "grad_norm": 0.73046875, "learning_rate": 2.1631315104240436e-06, "loss": 0.102, "step": 44638 }, { "epoch": 32.702564102564104, "grad_norm": 0.67578125, "learning_rate": 2.1623932228938743e-06, "loss": 0.0975, "step": 44639 }, { "epoch": 32.7032967032967, "grad_norm": 1.0, "learning_rate": 2.161655166965048e-06, "loss": 0.1409, "step": 44640 }, { "epoch": 32.704029304029305, "grad_norm": 0.8984375, "learning_rate": 2.1609173426410925e-06, "loss": 0.1117, "step": 44641 }, { "epoch": 32.7047619047619, "grad_norm": 0.68359375, "learning_rate": 2.160179749925558e-06, "loss": 0.1154, "step": 44642 }, { "epoch": 32.70549450549451, "grad_norm": 1.40625, "learning_rate": 2.159442388821967e-06, "loss": 0.1056, "step": 44643 }, { "epoch": 32.706227106227104, "grad_norm": 1.046875, "learning_rate": 2.1587052593338647e-06, "loss": 0.1399, "step": 44644 }, { "epoch": 32.70695970695971, "grad_norm": 1.015625, "learning_rate": 2.157968361464767e-06, "loss": 0.1285, "step": 44645 }, { "epoch": 32.707692307692305, "grad_norm": 1.3671875, "learning_rate": 2.157231695218215e-06, "loss": 0.1779, "step": 44646 }, { "epoch": 32.70842490842491, "grad_norm": 1.0546875, "learning_rate": 2.1564952605977297e-06, "loss": 0.1227, "step": 44647 }, { "epoch": 32.70915750915751, "grad_norm": 1.0, "learning_rate": 2.1557590576068565e-06, "loss": 0.1246, "step": 44648 }, { "epoch": 32.70989010989011, "grad_norm": 1.421875, "learning_rate": 2.1550230862491067e-06, "loss": 0.1237, "step": 44649 }, { "epoch": 32.71062271062271, "grad_norm": 1.40625, "learning_rate": 2.1542873465280144e-06, "loss": 0.1225, "step": 44650 }, { "epoch": 32.71135531135531, "grad_norm": 0.8984375, "learning_rate": 2.1535518384470966e-06, "loss": 0.1544, "step": 44651 }, { "epoch": 32.71208791208791, "grad_norm": 0.86328125, "learning_rate": 2.152816562009887e-06, "loss": 0.1345, "step": 44652 }, { "epoch": 32.712820512820514, "grad_norm": 1.0625, "learning_rate": 2.152081517219903e-06, "loss": 0.1279, "step": 44653 }, { "epoch": 32.71355311355311, "grad_norm": 0.51953125, "learning_rate": 2.1513467040806728e-06, "loss": 0.0989, "step": 44654 }, { "epoch": 32.714285714285715, "grad_norm": 1.09375, "learning_rate": 2.1506121225957076e-06, "loss": 0.1053, "step": 44655 }, { "epoch": 32.71501831501831, "grad_norm": 0.98046875, "learning_rate": 2.1498777727685356e-06, "loss": 0.1497, "step": 44656 }, { "epoch": 32.71575091575092, "grad_norm": 0.9296875, "learning_rate": 2.149143654602669e-06, "loss": 0.1198, "step": 44657 }, { "epoch": 32.716483516483514, "grad_norm": 0.97265625, "learning_rate": 2.148409768101625e-06, "loss": 0.1124, "step": 44658 }, { "epoch": 32.71721611721612, "grad_norm": 0.98828125, "learning_rate": 2.1476761132689256e-06, "loss": 0.1091, "step": 44659 }, { "epoch": 32.717948717948715, "grad_norm": 0.8515625, "learning_rate": 2.146942690108083e-06, "loss": 0.1275, "step": 44660 }, { "epoch": 32.71868131868132, "grad_norm": 1.140625, "learning_rate": 2.146209498622609e-06, "loss": 0.1364, "step": 44661 }, { "epoch": 32.71941391941392, "grad_norm": 1.0390625, "learning_rate": 2.1454765388160204e-06, "loss": 0.1241, "step": 44662 }, { "epoch": 32.72014652014652, "grad_norm": 1.34375, "learning_rate": 2.144743810691829e-06, "loss": 0.1112, "step": 44663 }, { "epoch": 32.72087912087912, "grad_norm": 0.79296875, "learning_rate": 2.144011314253547e-06, "loss": 0.0992, "step": 44664 }, { "epoch": 32.72161172161172, "grad_norm": 1.2109375, "learning_rate": 2.143279049504679e-06, "loss": 0.1519, "step": 44665 }, { "epoch": 32.72234432234432, "grad_norm": 0.80859375, "learning_rate": 2.142547016448728e-06, "loss": 0.0873, "step": 44666 }, { "epoch": 32.723076923076924, "grad_norm": 0.82421875, "learning_rate": 2.141815215089215e-06, "loss": 0.0971, "step": 44667 }, { "epoch": 32.72380952380952, "grad_norm": 0.6640625, "learning_rate": 2.1410836454296418e-06, "loss": 0.088, "step": 44668 }, { "epoch": 32.724542124542126, "grad_norm": 1.046875, "learning_rate": 2.140352307473508e-06, "loss": 0.1509, "step": 44669 }, { "epoch": 32.72527472527472, "grad_norm": 1.1640625, "learning_rate": 2.1396212012243256e-06, "loss": 0.0932, "step": 44670 }, { "epoch": 32.72600732600733, "grad_norm": 0.94140625, "learning_rate": 2.138890326685591e-06, "loss": 0.1256, "step": 44671 }, { "epoch": 32.726739926739924, "grad_norm": 1.125, "learning_rate": 2.1381596838608083e-06, "loss": 0.1185, "step": 44672 }, { "epoch": 32.72747252747253, "grad_norm": 1.265625, "learning_rate": 2.1374292727534743e-06, "loss": 0.1504, "step": 44673 }, { "epoch": 32.728205128205126, "grad_norm": 1.171875, "learning_rate": 2.136699093367106e-06, "loss": 0.1218, "step": 44674 }, { "epoch": 32.72893772893773, "grad_norm": 0.796875, "learning_rate": 2.1359691457051813e-06, "loss": 0.1096, "step": 44675 }, { "epoch": 32.72967032967033, "grad_norm": 1.5546875, "learning_rate": 2.1352394297712075e-06, "loss": 0.1356, "step": 44676 }, { "epoch": 32.73040293040293, "grad_norm": 1.453125, "learning_rate": 2.1345099455686675e-06, "loss": 0.1381, "step": 44677 }, { "epoch": 32.73113553113553, "grad_norm": 1.1328125, "learning_rate": 2.13378069310108e-06, "loss": 0.1102, "step": 44678 }, { "epoch": 32.73186813186813, "grad_norm": 0.9453125, "learning_rate": 2.1330516723719224e-06, "loss": 0.1006, "step": 44679 }, { "epoch": 32.73260073260073, "grad_norm": 1.1171875, "learning_rate": 2.1323228833846965e-06, "loss": 0.116, "step": 44680 }, { "epoch": 32.733333333333334, "grad_norm": 0.75, "learning_rate": 2.131594326142891e-06, "loss": 0.1077, "step": 44681 }, { "epoch": 32.73406593406593, "grad_norm": 1.1484375, "learning_rate": 2.130866000649985e-06, "loss": 0.1471, "step": 44682 }, { "epoch": 32.734798534798536, "grad_norm": 0.86328125, "learning_rate": 2.130137906909485e-06, "loss": 0.1037, "step": 44683 }, { "epoch": 32.73553113553113, "grad_norm": 1.375, "learning_rate": 2.1294100449248697e-06, "loss": 0.1061, "step": 44684 }, { "epoch": 32.73626373626374, "grad_norm": 0.69140625, "learning_rate": 2.128682414699639e-06, "loss": 0.1032, "step": 44685 }, { "epoch": 32.736996336996334, "grad_norm": 1.1640625, "learning_rate": 2.1279550162372616e-06, "loss": 0.1022, "step": 44686 }, { "epoch": 32.73772893772894, "grad_norm": 0.9609375, "learning_rate": 2.1272278495412327e-06, "loss": 0.1123, "step": 44687 }, { "epoch": 32.738461538461536, "grad_norm": 1.046875, "learning_rate": 2.1265009146150353e-06, "loss": 0.1207, "step": 44688 }, { "epoch": 32.73919413919414, "grad_norm": 1.203125, "learning_rate": 2.1257742114621553e-06, "loss": 0.1103, "step": 44689 }, { "epoch": 32.73992673992674, "grad_norm": 0.77734375, "learning_rate": 2.1250477400860695e-06, "loss": 0.1273, "step": 44690 }, { "epoch": 32.74065934065934, "grad_norm": 0.85546875, "learning_rate": 2.1243215004902586e-06, "loss": 0.125, "step": 44691 }, { "epoch": 32.74139194139194, "grad_norm": 1.1171875, "learning_rate": 2.1235954926782058e-06, "loss": 0.1229, "step": 44692 }, { "epoch": 32.74212454212454, "grad_norm": 1.1875, "learning_rate": 2.1228697166533837e-06, "loss": 0.1219, "step": 44693 }, { "epoch": 32.74285714285714, "grad_norm": 1.03125, "learning_rate": 2.1221441724192776e-06, "loss": 0.1121, "step": 44694 }, { "epoch": 32.743589743589745, "grad_norm": 1.0625, "learning_rate": 2.1214188599793607e-06, "loss": 0.1197, "step": 44695 }, { "epoch": 32.74432234432234, "grad_norm": 0.8828125, "learning_rate": 2.120693779337111e-06, "loss": 0.1044, "step": 44696 }, { "epoch": 32.745054945054946, "grad_norm": 0.96875, "learning_rate": 2.1199689304959913e-06, "loss": 0.1331, "step": 44697 }, { "epoch": 32.74578754578754, "grad_norm": 0.89453125, "learning_rate": 2.1192443134594854e-06, "loss": 0.1306, "step": 44698 }, { "epoch": 32.74652014652015, "grad_norm": 1.0390625, "learning_rate": 2.1185199282310613e-06, "loss": 0.1416, "step": 44699 }, { "epoch": 32.747252747252745, "grad_norm": 1.375, "learning_rate": 2.1177957748141977e-06, "loss": 0.1404, "step": 44700 }, { "epoch": 32.74798534798535, "grad_norm": 1.2265625, "learning_rate": 2.117071853212351e-06, "loss": 0.1235, "step": 44701 }, { "epoch": 32.748717948717946, "grad_norm": 0.84375, "learning_rate": 2.1163481634290005e-06, "loss": 0.1132, "step": 44702 }, { "epoch": 32.74945054945055, "grad_norm": 0.8984375, "learning_rate": 2.1156247054676023e-06, "loss": 0.1227, "step": 44703 }, { "epoch": 32.75018315018315, "grad_norm": 0.875, "learning_rate": 2.1149014793316357e-06, "loss": 0.1122, "step": 44704 }, { "epoch": 32.75091575091575, "grad_norm": 1.0390625, "learning_rate": 2.1141784850245574e-06, "loss": 0.1255, "step": 44705 }, { "epoch": 32.75164835164835, "grad_norm": 1.3828125, "learning_rate": 2.1134557225498348e-06, "loss": 0.1501, "step": 44706 }, { "epoch": 32.75238095238095, "grad_norm": 0.8359375, "learning_rate": 2.11273319191093e-06, "loss": 0.1334, "step": 44707 }, { "epoch": 32.75311355311355, "grad_norm": 1.1953125, "learning_rate": 2.1120108931113055e-06, "loss": 0.1234, "step": 44708 }, { "epoch": 32.753846153846155, "grad_norm": 1.4375, "learning_rate": 2.1112888261544183e-06, "loss": 0.1186, "step": 44709 }, { "epoch": 32.75457875457875, "grad_norm": 0.8359375, "learning_rate": 2.110566991043736e-06, "loss": 0.1097, "step": 44710 }, { "epoch": 32.755311355311356, "grad_norm": 1.296875, "learning_rate": 2.109845387782709e-06, "loss": 0.1224, "step": 44711 }, { "epoch": 32.75604395604395, "grad_norm": 1.8984375, "learning_rate": 2.109124016374801e-06, "loss": 0.1184, "step": 44712 }, { "epoch": 32.75677655677656, "grad_norm": 1.078125, "learning_rate": 2.108402876823461e-06, "loss": 0.1105, "step": 44713 }, { "epoch": 32.757509157509155, "grad_norm": 1.4921875, "learning_rate": 2.1076819691321544e-06, "loss": 0.1145, "step": 44714 }, { "epoch": 32.75824175824176, "grad_norm": 1.0234375, "learning_rate": 2.1069612933043296e-06, "loss": 0.1167, "step": 44715 }, { "epoch": 32.758974358974356, "grad_norm": 1.1640625, "learning_rate": 2.1062408493434343e-06, "loss": 0.104, "step": 44716 }, { "epoch": 32.75970695970696, "grad_norm": 1.265625, "learning_rate": 2.105520637252935e-06, "loss": 0.119, "step": 44717 }, { "epoch": 32.76043956043956, "grad_norm": 0.890625, "learning_rate": 2.1048006570362605e-06, "loss": 0.1393, "step": 44718 }, { "epoch": 32.76117216117216, "grad_norm": 1.03125, "learning_rate": 2.1040809086968804e-06, "loss": 0.1182, "step": 44719 }, { "epoch": 32.76190476190476, "grad_norm": 1.078125, "learning_rate": 2.1033613922382337e-06, "loss": 0.0958, "step": 44720 }, { "epoch": 32.76263736263736, "grad_norm": 0.9765625, "learning_rate": 2.102642107663772e-06, "loss": 0.142, "step": 44721 }, { "epoch": 32.76336996336996, "grad_norm": 1.390625, "learning_rate": 2.1019230549769404e-06, "loss": 0.14, "step": 44722 }, { "epoch": 32.764102564102565, "grad_norm": 0.95703125, "learning_rate": 2.10120423418118e-06, "loss": 0.1204, "step": 44723 }, { "epoch": 32.76483516483516, "grad_norm": 0.5703125, "learning_rate": 2.100485645279942e-06, "loss": 0.1016, "step": 44724 }, { "epoch": 32.765567765567766, "grad_norm": 1.21875, "learning_rate": 2.0997672882766652e-06, "loss": 0.1024, "step": 44725 }, { "epoch": 32.766300366300364, "grad_norm": 0.9609375, "learning_rate": 2.0990491631747923e-06, "loss": 0.1315, "step": 44726 }, { "epoch": 32.76703296703297, "grad_norm": 0.8359375, "learning_rate": 2.0983312699777616e-06, "loss": 0.1138, "step": 44727 }, { "epoch": 32.767765567765565, "grad_norm": 1.640625, "learning_rate": 2.0976136086890197e-06, "loss": 0.1143, "step": 44728 }, { "epoch": 32.76849816849817, "grad_norm": 0.734375, "learning_rate": 2.0968961793119953e-06, "loss": 0.1421, "step": 44729 }, { "epoch": 32.76923076923077, "grad_norm": 1.15625, "learning_rate": 2.0961789818501346e-06, "loss": 0.139, "step": 44730 }, { "epoch": 32.76996336996337, "grad_norm": 0.70703125, "learning_rate": 2.095462016306867e-06, "loss": 0.0933, "step": 44731 }, { "epoch": 32.77069597069597, "grad_norm": 1.0703125, "learning_rate": 2.094745282685632e-06, "loss": 0.1251, "step": 44732 }, { "epoch": 32.77142857142857, "grad_norm": 1.21875, "learning_rate": 2.094028780989865e-06, "loss": 0.1297, "step": 44733 }, { "epoch": 32.77216117216117, "grad_norm": 1.15625, "learning_rate": 2.093312511222995e-06, "loss": 0.1331, "step": 44734 }, { "epoch": 32.772893772893774, "grad_norm": 1.0078125, "learning_rate": 2.0925964733884513e-06, "loss": 0.1299, "step": 44735 }, { "epoch": 32.77362637362637, "grad_norm": 0.61328125, "learning_rate": 2.091880667489675e-06, "loss": 0.114, "step": 44736 }, { "epoch": 32.774358974358975, "grad_norm": 1.3203125, "learning_rate": 2.091165093530094e-06, "loss": 0.1359, "step": 44737 }, { "epoch": 32.77509157509157, "grad_norm": 1.109375, "learning_rate": 2.090449751513121e-06, "loss": 0.1297, "step": 44738 }, { "epoch": 32.77582417582418, "grad_norm": 1.15625, "learning_rate": 2.089734641442203e-06, "loss": 0.126, "step": 44739 }, { "epoch": 32.776556776556774, "grad_norm": 1.1875, "learning_rate": 2.089019763320752e-06, "loss": 0.1311, "step": 44740 }, { "epoch": 32.77728937728938, "grad_norm": 1.234375, "learning_rate": 2.088305117152209e-06, "loss": 0.1009, "step": 44741 }, { "epoch": 32.778021978021975, "grad_norm": 0.93359375, "learning_rate": 2.0875907029399856e-06, "loss": 0.1212, "step": 44742 }, { "epoch": 32.77875457875458, "grad_norm": 0.8203125, "learning_rate": 2.0868765206875056e-06, "loss": 0.0871, "step": 44743 }, { "epoch": 32.77948717948718, "grad_norm": 0.890625, "learning_rate": 2.086162570398194e-06, "loss": 0.1415, "step": 44744 }, { "epoch": 32.78021978021978, "grad_norm": 0.85546875, "learning_rate": 2.0854488520754682e-06, "loss": 0.1312, "step": 44745 }, { "epoch": 32.78095238095238, "grad_norm": 1.0859375, "learning_rate": 2.0847353657227524e-06, "loss": 0.1144, "step": 44746 }, { "epoch": 32.78168498168498, "grad_norm": 1.609375, "learning_rate": 2.0840221113434635e-06, "loss": 0.1177, "step": 44747 }, { "epoch": 32.78241758241758, "grad_norm": 1.2265625, "learning_rate": 2.083309088941022e-06, "loss": 0.1107, "step": 44748 }, { "epoch": 32.783150183150184, "grad_norm": 1.65625, "learning_rate": 2.082596298518833e-06, "loss": 0.1338, "step": 44749 }, { "epoch": 32.78388278388278, "grad_norm": 0.8046875, "learning_rate": 2.0818837400803222e-06, "loss": 0.1287, "step": 44750 }, { "epoch": 32.784615384615385, "grad_norm": 0.98828125, "learning_rate": 2.0811714136289012e-06, "loss": 0.1294, "step": 44751 }, { "epoch": 32.78534798534798, "grad_norm": 0.98046875, "learning_rate": 2.080459319167988e-06, "loss": 0.131, "step": 44752 }, { "epoch": 32.78608058608059, "grad_norm": 1.4375, "learning_rate": 2.0797474567009854e-06, "loss": 0.1173, "step": 44753 }, { "epoch": 32.786813186813184, "grad_norm": 0.8828125, "learning_rate": 2.079035826231311e-06, "loss": 0.1334, "step": 44754 }, { "epoch": 32.78754578754579, "grad_norm": 1.03125, "learning_rate": 2.0783244277623613e-06, "loss": 0.1108, "step": 44755 }, { "epoch": 32.788278388278385, "grad_norm": 1.25, "learning_rate": 2.077613261297565e-06, "loss": 0.112, "step": 44756 }, { "epoch": 32.78901098901099, "grad_norm": 1.265625, "learning_rate": 2.0769023268403195e-06, "loss": 0.1218, "step": 44757 }, { "epoch": 32.78974358974359, "grad_norm": 1.0703125, "learning_rate": 2.0761916243940256e-06, "loss": 0.1435, "step": 44758 }, { "epoch": 32.79047619047619, "grad_norm": 1.015625, "learning_rate": 2.075481153962102e-06, "loss": 0.1398, "step": 44759 }, { "epoch": 32.79120879120879, "grad_norm": 0.80859375, "learning_rate": 2.0747709155479398e-06, "loss": 0.1138, "step": 44760 }, { "epoch": 32.79194139194139, "grad_norm": 0.8984375, "learning_rate": 2.0740609091549453e-06, "loss": 0.1234, "step": 44761 }, { "epoch": 32.79267399267399, "grad_norm": 0.515625, "learning_rate": 2.0733511347865212e-06, "loss": 0.0847, "step": 44762 }, { "epoch": 32.793406593406594, "grad_norm": 1.046875, "learning_rate": 2.072641592446074e-06, "loss": 0.1227, "step": 44763 }, { "epoch": 32.79413919413919, "grad_norm": 1.0, "learning_rate": 2.0719322821369953e-06, "loss": 0.1119, "step": 44764 }, { "epoch": 32.794871794871796, "grad_norm": 0.80078125, "learning_rate": 2.071223203862692e-06, "loss": 0.1272, "step": 44765 }, { "epoch": 32.79560439560439, "grad_norm": 1.0703125, "learning_rate": 2.0705143576265492e-06, "loss": 0.1105, "step": 44766 }, { "epoch": 32.796336996337, "grad_norm": 1.0390625, "learning_rate": 2.0698057434319803e-06, "loss": 0.1035, "step": 44767 }, { "epoch": 32.797069597069594, "grad_norm": 0.9765625, "learning_rate": 2.0690973612823587e-06, "loss": 0.1163, "step": 44768 }, { "epoch": 32.7978021978022, "grad_norm": 1.0, "learning_rate": 2.0683892111811027e-06, "loss": 0.1157, "step": 44769 }, { "epoch": 32.798534798534796, "grad_norm": 1.6328125, "learning_rate": 2.0676812931315827e-06, "loss": 0.1797, "step": 44770 }, { "epoch": 32.7992673992674, "grad_norm": 0.86328125, "learning_rate": 2.066973607137204e-06, "loss": 0.1082, "step": 44771 }, { "epoch": 32.8, "grad_norm": 0.79296875, "learning_rate": 2.0662661532013583e-06, "loss": 0.1147, "step": 44772 }, { "epoch": 32.8007326007326, "grad_norm": 1.2890625, "learning_rate": 2.065558931327425e-06, "loss": 0.1418, "step": 44773 }, { "epoch": 32.8014652014652, "grad_norm": 0.6328125, "learning_rate": 2.064851941518807e-06, "loss": 0.0926, "step": 44774 }, { "epoch": 32.8021978021978, "grad_norm": 1.4609375, "learning_rate": 2.064145183778877e-06, "loss": 0.1642, "step": 44775 }, { "epoch": 32.8029304029304, "grad_norm": 1.0859375, "learning_rate": 2.063438658111027e-06, "loss": 0.1167, "step": 44776 }, { "epoch": 32.803663003663004, "grad_norm": 1.1171875, "learning_rate": 2.0627323645186415e-06, "loss": 0.154, "step": 44777 }, { "epoch": 32.8043956043956, "grad_norm": 0.8671875, "learning_rate": 2.0620263030051175e-06, "loss": 0.1209, "step": 44778 }, { "epoch": 32.805128205128206, "grad_norm": 1.1875, "learning_rate": 2.0613204735738175e-06, "loss": 0.1293, "step": 44779 }, { "epoch": 32.8058608058608, "grad_norm": 1.0703125, "learning_rate": 2.0606148762281386e-06, "loss": 0.1086, "step": 44780 }, { "epoch": 32.80659340659341, "grad_norm": 0.875, "learning_rate": 2.059909510971444e-06, "loss": 0.139, "step": 44781 }, { "epoch": 32.807326007326004, "grad_norm": 0.94921875, "learning_rate": 2.059204377807134e-06, "loss": 0.0999, "step": 44782 }, { "epoch": 32.80805860805861, "grad_norm": 1.2265625, "learning_rate": 2.0584994767385743e-06, "loss": 0.0947, "step": 44783 }, { "epoch": 32.808791208791206, "grad_norm": 0.91015625, "learning_rate": 2.0577948077691486e-06, "loss": 0.1323, "step": 44784 }, { "epoch": 32.80952380952381, "grad_norm": 1.28125, "learning_rate": 2.0570903709022317e-06, "loss": 0.1049, "step": 44785 }, { "epoch": 32.81025641025641, "grad_norm": 1.34375, "learning_rate": 2.0563861661411923e-06, "loss": 0.1634, "step": 44786 }, { "epoch": 32.81098901098901, "grad_norm": 1.2109375, "learning_rate": 2.0556821934894154e-06, "loss": 0.1502, "step": 44787 }, { "epoch": 32.81172161172161, "grad_norm": 0.9921875, "learning_rate": 2.0549784529502596e-06, "loss": 0.1217, "step": 44788 }, { "epoch": 32.81245421245421, "grad_norm": 0.94921875, "learning_rate": 2.054274944527115e-06, "loss": 0.1139, "step": 44789 }, { "epoch": 32.81318681318681, "grad_norm": 1.71875, "learning_rate": 2.05357166822334e-06, "loss": 0.1424, "step": 44790 }, { "epoch": 32.813919413919415, "grad_norm": 1.2109375, "learning_rate": 2.0528686240423076e-06, "loss": 0.1236, "step": 44791 }, { "epoch": 32.81465201465201, "grad_norm": 1.2734375, "learning_rate": 2.0521658119873827e-06, "loss": 0.1179, "step": 44792 }, { "epoch": 32.815384615384616, "grad_norm": 1.453125, "learning_rate": 2.051463232061939e-06, "loss": 0.1415, "step": 44793 }, { "epoch": 32.81611721611721, "grad_norm": 1.359375, "learning_rate": 2.050760884269339e-06, "loss": 0.1464, "step": 44794 }, { "epoch": 32.81684981684982, "grad_norm": 1.015625, "learning_rate": 2.0500587686129478e-06, "loss": 0.1116, "step": 44795 }, { "epoch": 32.817582417582415, "grad_norm": 1.015625, "learning_rate": 2.0493568850961327e-06, "loss": 0.1299, "step": 44796 }, { "epoch": 32.81831501831502, "grad_norm": 1.03125, "learning_rate": 2.0486552337222465e-06, "loss": 0.107, "step": 44797 }, { "epoch": 32.819047619047616, "grad_norm": 0.92578125, "learning_rate": 2.0479538144946635e-06, "loss": 0.1558, "step": 44798 }, { "epoch": 32.81978021978022, "grad_norm": 0.85546875, "learning_rate": 2.047252627416741e-06, "loss": 0.1239, "step": 44799 }, { "epoch": 32.82051282051282, "grad_norm": 0.6015625, "learning_rate": 2.0465516724918378e-06, "loss": 0.1078, "step": 44800 }, { "epoch": 32.82124542124542, "grad_norm": 0.96875, "learning_rate": 2.045850949723306e-06, "loss": 0.1314, "step": 44801 }, { "epoch": 32.82197802197802, "grad_norm": 0.451171875, "learning_rate": 2.0451504591145137e-06, "loss": 0.0859, "step": 44802 }, { "epoch": 32.82271062271062, "grad_norm": 1.734375, "learning_rate": 2.0444502006688086e-06, "loss": 0.1457, "step": 44803 }, { "epoch": 32.82344322344322, "grad_norm": 1.2421875, "learning_rate": 2.043750174389553e-06, "loss": 0.1368, "step": 44804 }, { "epoch": 32.824175824175825, "grad_norm": 1.4296875, "learning_rate": 2.0430503802801002e-06, "loss": 0.1412, "step": 44805 }, { "epoch": 32.82490842490842, "grad_norm": 1.5234375, "learning_rate": 2.042350818343796e-06, "loss": 0.1068, "step": 44806 }, { "epoch": 32.825641025641026, "grad_norm": 0.8359375, "learning_rate": 2.041651488583994e-06, "loss": 0.139, "step": 44807 }, { "epoch": 32.82637362637362, "grad_norm": 1.3828125, "learning_rate": 2.040952391004051e-06, "loss": 0.1482, "step": 44808 }, { "epoch": 32.82710622710623, "grad_norm": 0.87890625, "learning_rate": 2.040253525607314e-06, "loss": 0.1249, "step": 44809 }, { "epoch": 32.827838827838825, "grad_norm": 1.265625, "learning_rate": 2.0395548923971252e-06, "loss": 0.1367, "step": 44810 }, { "epoch": 32.82857142857143, "grad_norm": 1.1171875, "learning_rate": 2.0388564913768466e-06, "loss": 0.1085, "step": 44811 }, { "epoch": 32.829304029304026, "grad_norm": 1.4296875, "learning_rate": 2.038158322549804e-06, "loss": 0.1483, "step": 44812 }, { "epoch": 32.83003663003663, "grad_norm": 0.734375, "learning_rate": 2.0374603859193655e-06, "loss": 0.1372, "step": 44813 }, { "epoch": 32.83076923076923, "grad_norm": 0.9921875, "learning_rate": 2.0367626814888563e-06, "loss": 0.1551, "step": 44814 }, { "epoch": 32.83150183150183, "grad_norm": 0.8828125, "learning_rate": 2.0360652092616347e-06, "loss": 0.1171, "step": 44815 }, { "epoch": 32.83223443223443, "grad_norm": 1.1484375, "learning_rate": 2.0353679692410243e-06, "loss": 0.1327, "step": 44816 }, { "epoch": 32.832967032967034, "grad_norm": 1.078125, "learning_rate": 2.0346709614303844e-06, "loss": 0.11, "step": 44817 }, { "epoch": 32.83369963369963, "grad_norm": 1.0859375, "learning_rate": 2.0339741858330325e-06, "loss": 0.1174, "step": 44818 }, { "epoch": 32.834432234432235, "grad_norm": 0.7421875, "learning_rate": 2.033277642452334e-06, "loss": 0.0925, "step": 44819 }, { "epoch": 32.83516483516483, "grad_norm": 0.98828125, "learning_rate": 2.0325813312916067e-06, "loss": 0.1092, "step": 44820 }, { "epoch": 32.83589743589744, "grad_norm": 0.94140625, "learning_rate": 2.031885252354193e-06, "loss": 0.1345, "step": 44821 }, { "epoch": 32.836630036630034, "grad_norm": 1.171875, "learning_rate": 2.0311894056434335e-06, "loss": 0.1514, "step": 44822 }, { "epoch": 32.83736263736264, "grad_norm": 1.71875, "learning_rate": 2.030493791162653e-06, "loss": 0.1113, "step": 44823 }, { "epoch": 32.838095238095235, "grad_norm": 0.8828125, "learning_rate": 2.029798408915188e-06, "loss": 0.1303, "step": 44824 }, { "epoch": 32.83882783882784, "grad_norm": 0.8125, "learning_rate": 2.029103258904368e-06, "loss": 0.1174, "step": 44825 }, { "epoch": 32.83956043956044, "grad_norm": 1.2734375, "learning_rate": 2.028408341133535e-06, "loss": 0.138, "step": 44826 }, { "epoch": 32.84029304029304, "grad_norm": 1.0234375, "learning_rate": 2.0277136556060027e-06, "loss": 0.1548, "step": 44827 }, { "epoch": 32.84102564102564, "grad_norm": 1.1171875, "learning_rate": 2.0270192023251063e-06, "loss": 0.1254, "step": 44828 }, { "epoch": 32.84175824175824, "grad_norm": 1.15625, "learning_rate": 2.026324981294177e-06, "loss": 0.128, "step": 44829 }, { "epoch": 32.84249084249084, "grad_norm": 0.92578125, "learning_rate": 2.0256309925165392e-06, "loss": 0.1087, "step": 44830 }, { "epoch": 32.843223443223444, "grad_norm": 0.6796875, "learning_rate": 2.024937235995507e-06, "loss": 0.1364, "step": 44831 }, { "epoch": 32.84395604395604, "grad_norm": 1.484375, "learning_rate": 2.024243711734427e-06, "loss": 0.1311, "step": 44832 }, { "epoch": 32.844688644688645, "grad_norm": 1.078125, "learning_rate": 2.0235504197365966e-06, "loss": 0.137, "step": 44833 }, { "epoch": 32.84542124542124, "grad_norm": 0.9765625, "learning_rate": 2.0228573600053575e-06, "loss": 0.1114, "step": 44834 }, { "epoch": 32.84615384615385, "grad_norm": 0.6328125, "learning_rate": 2.022164532544017e-06, "loss": 0.0979, "step": 44835 }, { "epoch": 32.846886446886444, "grad_norm": 1.0078125, "learning_rate": 2.021471937355907e-06, "loss": 0.1307, "step": 44836 }, { "epoch": 32.84761904761905, "grad_norm": 0.87890625, "learning_rate": 2.020779574444334e-06, "loss": 0.101, "step": 44837 }, { "epoch": 32.848351648351645, "grad_norm": 1.140625, "learning_rate": 2.0200874438126195e-06, "loss": 0.1601, "step": 44838 }, { "epoch": 32.84908424908425, "grad_norm": 0.69921875, "learning_rate": 2.0193955454640863e-06, "loss": 0.1094, "step": 44839 }, { "epoch": 32.84981684981685, "grad_norm": 1.1953125, "learning_rate": 2.0187038794020442e-06, "loss": 0.1322, "step": 44840 }, { "epoch": 32.85054945054945, "grad_norm": 1.046875, "learning_rate": 2.0180124456298064e-06, "loss": 0.114, "step": 44841 }, { "epoch": 32.85128205128205, "grad_norm": 0.953125, "learning_rate": 2.0173212441506864e-06, "loss": 0.1313, "step": 44842 }, { "epoch": 32.85201465201465, "grad_norm": 1.2265625, "learning_rate": 2.016630274967999e-06, "loss": 0.1125, "step": 44843 }, { "epoch": 32.85274725274725, "grad_norm": 1.078125, "learning_rate": 2.0159395380850405e-06, "loss": 0.1221, "step": 44844 }, { "epoch": 32.853479853479854, "grad_norm": 0.92578125, "learning_rate": 2.015249033505142e-06, "loss": 0.1063, "step": 44845 }, { "epoch": 32.85421245421245, "grad_norm": 0.67578125, "learning_rate": 2.0145587612316012e-06, "loss": 0.0858, "step": 44846 }, { "epoch": 32.854945054945055, "grad_norm": 1.34375, "learning_rate": 2.0138687212677253e-06, "loss": 0.1274, "step": 44847 }, { "epoch": 32.85567765567765, "grad_norm": 0.9453125, "learning_rate": 2.0131789136168235e-06, "loss": 0.1405, "step": 44848 }, { "epoch": 32.85641025641026, "grad_norm": 1.015625, "learning_rate": 2.0124893382821927e-06, "loss": 0.1136, "step": 44849 }, { "epoch": 32.857142857142854, "grad_norm": 0.66796875, "learning_rate": 2.011799995267147e-06, "loss": 0.1081, "step": 44850 }, { "epoch": 32.85787545787546, "grad_norm": 0.83203125, "learning_rate": 2.0111108845749784e-06, "loss": 0.1233, "step": 44851 }, { "epoch": 32.858608058608056, "grad_norm": 1.2734375, "learning_rate": 2.010422006209006e-06, "loss": 0.148, "step": 44852 }, { "epoch": 32.85934065934066, "grad_norm": 0.75390625, "learning_rate": 2.0097333601725164e-06, "loss": 0.1074, "step": 44853 }, { "epoch": 32.86007326007326, "grad_norm": 1.46875, "learning_rate": 2.009044946468807e-06, "loss": 0.1337, "step": 44854 }, { "epoch": 32.86080586080586, "grad_norm": 1.015625, "learning_rate": 2.008356765101185e-06, "loss": 0.1283, "step": 44855 }, { "epoch": 32.86153846153846, "grad_norm": 0.94140625, "learning_rate": 2.0076688160729497e-06, "loss": 0.1035, "step": 44856 }, { "epoch": 32.86227106227106, "grad_norm": 0.9453125, "learning_rate": 2.0069810993873864e-06, "loss": 0.1241, "step": 44857 }, { "epoch": 32.86300366300366, "grad_norm": 1.0703125, "learning_rate": 2.006293615047803e-06, "loss": 0.128, "step": 44858 }, { "epoch": 32.863736263736264, "grad_norm": 0.90234375, "learning_rate": 2.00560636305747e-06, "loss": 0.102, "step": 44859 }, { "epoch": 32.86446886446886, "grad_norm": 1.0390625, "learning_rate": 2.004919343419713e-06, "loss": 0.1151, "step": 44860 }, { "epoch": 32.865201465201466, "grad_norm": 0.8515625, "learning_rate": 2.0042325561378006e-06, "loss": 0.1328, "step": 44861 }, { "epoch": 32.86593406593406, "grad_norm": 1.171875, "learning_rate": 2.0035460012150306e-06, "loss": 0.0957, "step": 44862 }, { "epoch": 32.86666666666667, "grad_norm": 0.953125, "learning_rate": 2.002859678654695e-06, "loss": 0.1058, "step": 44863 }, { "epoch": 32.867399267399264, "grad_norm": 1.9921875, "learning_rate": 2.002173588460074e-06, "loss": 0.2003, "step": 44864 }, { "epoch": 32.86813186813187, "grad_norm": 1.328125, "learning_rate": 2.0014877306344658e-06, "loss": 0.1463, "step": 44865 }, { "epoch": 32.868864468864466, "grad_norm": 1.6484375, "learning_rate": 2.000802105181146e-06, "loss": 0.1171, "step": 44866 }, { "epoch": 32.86959706959707, "grad_norm": 1.609375, "learning_rate": 2.000116712103411e-06, "loss": 0.1373, "step": 44867 }, { "epoch": 32.87032967032967, "grad_norm": 1.015625, "learning_rate": 1.999431551404531e-06, "loss": 0.1254, "step": 44868 }, { "epoch": 32.87106227106227, "grad_norm": 0.85546875, "learning_rate": 1.998746623087804e-06, "loss": 0.1261, "step": 44869 }, { "epoch": 32.87179487179487, "grad_norm": 1.8203125, "learning_rate": 1.998061927156489e-06, "loss": 0.1275, "step": 44870 }, { "epoch": 32.87252747252747, "grad_norm": 1.25, "learning_rate": 1.9973774636138984e-06, "loss": 0.1357, "step": 44871 }, { "epoch": 32.87326007326007, "grad_norm": 1.2265625, "learning_rate": 1.9966932324632863e-06, "loss": 0.1407, "step": 44872 }, { "epoch": 32.873992673992674, "grad_norm": 1.140625, "learning_rate": 1.996009233707939e-06, "loss": 0.112, "step": 44873 }, { "epoch": 32.87472527472527, "grad_norm": 0.984375, "learning_rate": 1.9953254673511383e-06, "loss": 0.1272, "step": 44874 }, { "epoch": 32.875457875457876, "grad_norm": 1.0, "learning_rate": 1.9946419333961534e-06, "loss": 0.1541, "step": 44875 }, { "epoch": 32.87619047619047, "grad_norm": 1.3984375, "learning_rate": 1.9939586318462595e-06, "loss": 0.1149, "step": 44876 }, { "epoch": 32.87692307692308, "grad_norm": 1.421875, "learning_rate": 1.9932755627047375e-06, "loss": 0.1299, "step": 44877 }, { "epoch": 32.877655677655675, "grad_norm": 1.375, "learning_rate": 1.9925927259748574e-06, "loss": 0.1099, "step": 44878 }, { "epoch": 32.87838827838828, "grad_norm": 1.015625, "learning_rate": 1.9919101216598896e-06, "loss": 0.1342, "step": 44879 }, { "epoch": 32.879120879120876, "grad_norm": 0.609375, "learning_rate": 1.9912277497630978e-06, "loss": 0.1029, "step": 44880 }, { "epoch": 32.87985347985348, "grad_norm": 1.3046875, "learning_rate": 1.9905456102877624e-06, "loss": 0.1465, "step": 44881 }, { "epoch": 32.88058608058608, "grad_norm": 1.0859375, "learning_rate": 1.9898637032371548e-06, "loss": 0.1453, "step": 44882 }, { "epoch": 32.88131868131868, "grad_norm": 1.1796875, "learning_rate": 1.9891820286145275e-06, "loss": 0.1198, "step": 44883 }, { "epoch": 32.88205128205128, "grad_norm": 0.97265625, "learning_rate": 1.988500586423155e-06, "loss": 0.1294, "step": 44884 }, { "epoch": 32.88278388278388, "grad_norm": 0.9765625, "learning_rate": 1.9878193766662984e-06, "loss": 0.1117, "step": 44885 }, { "epoch": 32.88351648351648, "grad_norm": 1.5546875, "learning_rate": 1.987138399347231e-06, "loss": 0.1398, "step": 44886 }, { "epoch": 32.884249084249085, "grad_norm": 1.2109375, "learning_rate": 1.986457654469202e-06, "loss": 0.1492, "step": 44887 }, { "epoch": 32.88498168498168, "grad_norm": 0.87109375, "learning_rate": 1.985777142035486e-06, "loss": 0.1059, "step": 44888 }, { "epoch": 32.885714285714286, "grad_norm": 1.7109375, "learning_rate": 1.985096862049337e-06, "loss": 0.0922, "step": 44889 }, { "epoch": 32.88644688644688, "grad_norm": 1.4375, "learning_rate": 1.9844168145140078e-06, "loss": 0.1151, "step": 44890 }, { "epoch": 32.88717948717949, "grad_norm": 1.15625, "learning_rate": 1.983736999432774e-06, "loss": 0.1331, "step": 44891 }, { "epoch": 32.887912087912085, "grad_norm": 1.125, "learning_rate": 1.983057416808878e-06, "loss": 0.1467, "step": 44892 }, { "epoch": 32.88864468864469, "grad_norm": 0.921875, "learning_rate": 1.9823780666455904e-06, "loss": 0.1297, "step": 44893 }, { "epoch": 32.889377289377286, "grad_norm": 0.80078125, "learning_rate": 1.981698948946147e-06, "loss": 0.1371, "step": 44894 }, { "epoch": 32.89010989010989, "grad_norm": 0.828125, "learning_rate": 1.981020063713818e-06, "loss": 0.1077, "step": 44895 }, { "epoch": 32.89084249084249, "grad_norm": 0.8203125, "learning_rate": 1.9803414109518406e-06, "loss": 0.1001, "step": 44896 }, { "epoch": 32.89157509157509, "grad_norm": 1.2265625, "learning_rate": 1.9796629906634793e-06, "loss": 0.1598, "step": 44897 }, { "epoch": 32.89230769230769, "grad_norm": 1.09375, "learning_rate": 1.978984802851982e-06, "loss": 0.1197, "step": 44898 }, { "epoch": 32.89304029304029, "grad_norm": 1.5390625, "learning_rate": 1.9783068475205955e-06, "loss": 0.148, "step": 44899 }, { "epoch": 32.89377289377289, "grad_norm": 1.2265625, "learning_rate": 1.9776291246725748e-06, "loss": 0.107, "step": 44900 }, { "epoch": 32.894505494505495, "grad_norm": 1.0078125, "learning_rate": 1.9769516343111554e-06, "loss": 0.1032, "step": 44901 }, { "epoch": 32.89523809523809, "grad_norm": 0.68359375, "learning_rate": 1.976274376439592e-06, "loss": 0.1151, "step": 44902 }, { "epoch": 32.895970695970696, "grad_norm": 0.6796875, "learning_rate": 1.975597351061126e-06, "loss": 0.1102, "step": 44903 }, { "epoch": 32.89670329670329, "grad_norm": 1.140625, "learning_rate": 1.974920558179006e-06, "loss": 0.1362, "step": 44904 }, { "epoch": 32.8974358974359, "grad_norm": 0.7578125, "learning_rate": 1.9742439977964684e-06, "loss": 0.1106, "step": 44905 }, { "epoch": 32.898168498168495, "grad_norm": 0.6875, "learning_rate": 1.9735676699167615e-06, "loss": 0.1054, "step": 44906 }, { "epoch": 32.8989010989011, "grad_norm": 1.1796875, "learning_rate": 1.972891574543116e-06, "loss": 0.1259, "step": 44907 }, { "epoch": 32.899633699633696, "grad_norm": 1.015625, "learning_rate": 1.9722157116787862e-06, "loss": 0.1302, "step": 44908 }, { "epoch": 32.9003663003663, "grad_norm": 0.92578125, "learning_rate": 1.9715400813269917e-06, "loss": 0.1501, "step": 44909 }, { "epoch": 32.9010989010989, "grad_norm": 0.8828125, "learning_rate": 1.970864683490987e-06, "loss": 0.1201, "step": 44910 }, { "epoch": 32.9018315018315, "grad_norm": 1.1015625, "learning_rate": 1.9701895181739912e-06, "loss": 0.1246, "step": 44911 }, { "epoch": 32.9025641025641, "grad_norm": 0.52734375, "learning_rate": 1.969514585379259e-06, "loss": 0.0982, "step": 44912 }, { "epoch": 32.903296703296704, "grad_norm": 0.91796875, "learning_rate": 1.9688398851100095e-06, "loss": 0.116, "step": 44913 }, { "epoch": 32.9040293040293, "grad_norm": 1.5, "learning_rate": 1.9681654173694807e-06, "loss": 0.139, "step": 44914 }, { "epoch": 32.904761904761905, "grad_norm": 1.0390625, "learning_rate": 1.967491182160909e-06, "loss": 0.0981, "step": 44915 }, { "epoch": 32.9054945054945, "grad_norm": 1.1875, "learning_rate": 1.966817179487504e-06, "loss": 0.1176, "step": 44916 }, { "epoch": 32.90622710622711, "grad_norm": 1.015625, "learning_rate": 1.9661434093525245e-06, "loss": 0.1434, "step": 44917 }, { "epoch": 32.906959706959704, "grad_norm": 0.8828125, "learning_rate": 1.9654698717591797e-06, "loss": 0.1031, "step": 44918 }, { "epoch": 32.90769230769231, "grad_norm": 1.546875, "learning_rate": 1.964796566710701e-06, "loss": 0.1061, "step": 44919 }, { "epoch": 32.908424908424905, "grad_norm": 1.125, "learning_rate": 1.9641234942103156e-06, "loss": 0.1389, "step": 44920 }, { "epoch": 32.90915750915751, "grad_norm": 1.375, "learning_rate": 1.9634506542612528e-06, "loss": 0.1329, "step": 44921 }, { "epoch": 32.90989010989011, "grad_norm": 1.03125, "learning_rate": 1.962778046866718e-06, "loss": 0.0962, "step": 44922 }, { "epoch": 32.91062271062271, "grad_norm": 1.296875, "learning_rate": 1.9621056720299588e-06, "loss": 0.1131, "step": 44923 }, { "epoch": 32.91135531135531, "grad_norm": 0.90625, "learning_rate": 1.961433529754185e-06, "loss": 0.0996, "step": 44924 }, { "epoch": 32.91208791208791, "grad_norm": 1.2265625, "learning_rate": 1.960761620042611e-06, "loss": 0.1327, "step": 44925 }, { "epoch": 32.91282051282051, "grad_norm": 1.09375, "learning_rate": 1.9600899428984676e-06, "loss": 0.1189, "step": 44926 }, { "epoch": 32.913553113553114, "grad_norm": 1.7890625, "learning_rate": 1.9594184983249657e-06, "loss": 0.138, "step": 44927 }, { "epoch": 32.91428571428571, "grad_norm": 0.6328125, "learning_rate": 1.9587472863253244e-06, "loss": 0.1001, "step": 44928 }, { "epoch": 32.915018315018315, "grad_norm": 1.4296875, "learning_rate": 1.9580763069027594e-06, "loss": 0.1008, "step": 44929 }, { "epoch": 32.91575091575091, "grad_norm": 1.359375, "learning_rate": 1.9574055600604912e-06, "loss": 0.1176, "step": 44930 }, { "epoch": 32.91648351648352, "grad_norm": 0.86328125, "learning_rate": 1.9567350458017233e-06, "loss": 0.1177, "step": 44931 }, { "epoch": 32.917216117216114, "grad_norm": 0.9609375, "learning_rate": 1.956064764129676e-06, "loss": 0.1082, "step": 44932 }, { "epoch": 32.91794871794872, "grad_norm": 0.75, "learning_rate": 1.9553947150475536e-06, "loss": 0.1252, "step": 44933 }, { "epoch": 32.918681318681315, "grad_norm": 0.86328125, "learning_rate": 1.954724898558577e-06, "loss": 0.0946, "step": 44934 }, { "epoch": 32.91941391941392, "grad_norm": 0.8515625, "learning_rate": 1.954055314665949e-06, "loss": 0.1156, "step": 44935 }, { "epoch": 32.92014652014652, "grad_norm": 0.87109375, "learning_rate": 1.9533859633728743e-06, "loss": 0.119, "step": 44936 }, { "epoch": 32.92087912087912, "grad_norm": 1.1640625, "learning_rate": 1.9527168446825676e-06, "loss": 0.1198, "step": 44937 }, { "epoch": 32.92161172161172, "grad_norm": 1.3671875, "learning_rate": 1.9520479585982325e-06, "loss": 0.1279, "step": 44938 }, { "epoch": 32.92234432234432, "grad_norm": 0.98046875, "learning_rate": 1.951379305123074e-06, "loss": 0.1126, "step": 44939 }, { "epoch": 32.92307692307692, "grad_norm": 1.03125, "learning_rate": 1.9507108842602897e-06, "loss": 0.1077, "step": 44940 }, { "epoch": 32.923809523809524, "grad_norm": 0.671875, "learning_rate": 1.9500426960130945e-06, "loss": 0.1012, "step": 44941 }, { "epoch": 32.92454212454212, "grad_norm": 1.0625, "learning_rate": 1.9493747403846705e-06, "loss": 0.1333, "step": 44942 }, { "epoch": 32.925274725274726, "grad_norm": 1.3046875, "learning_rate": 1.9487070173782388e-06, "loss": 0.1005, "step": 44943 }, { "epoch": 32.92600732600732, "grad_norm": 1.4453125, "learning_rate": 1.948039526996991e-06, "loss": 0.1476, "step": 44944 }, { "epoch": 32.92673992673993, "grad_norm": 1.46875, "learning_rate": 1.947372269244126e-06, "loss": 0.1369, "step": 44945 }, { "epoch": 32.927472527472524, "grad_norm": 1.8125, "learning_rate": 1.9467052441228324e-06, "loss": 0.1467, "step": 44946 }, { "epoch": 32.92820512820513, "grad_norm": 1.203125, "learning_rate": 1.946038451636318e-06, "loss": 0.1079, "step": 44947 }, { "epoch": 32.928937728937726, "grad_norm": 1.0703125, "learning_rate": 1.9453718917877663e-06, "loss": 0.1144, "step": 44948 }, { "epoch": 32.92967032967033, "grad_norm": 1.1171875, "learning_rate": 1.9447055645803853e-06, "loss": 0.1137, "step": 44949 }, { "epoch": 32.93040293040293, "grad_norm": 0.90625, "learning_rate": 1.944039470017358e-06, "loss": 0.1159, "step": 44950 }, { "epoch": 32.93113553113553, "grad_norm": 1.375, "learning_rate": 1.943373608101871e-06, "loss": 0.1447, "step": 44951 }, { "epoch": 32.93186813186813, "grad_norm": 0.90234375, "learning_rate": 1.9427079788371286e-06, "loss": 0.0984, "step": 44952 }, { "epoch": 32.93260073260073, "grad_norm": 1.0078125, "learning_rate": 1.9420425822263076e-06, "loss": 0.1309, "step": 44953 }, { "epoch": 32.93333333333333, "grad_norm": 1.203125, "learning_rate": 1.941377418272606e-06, "loss": 0.11, "step": 44954 }, { "epoch": 32.934065934065934, "grad_norm": 0.71484375, "learning_rate": 1.9407124869792007e-06, "loss": 0.1078, "step": 44955 }, { "epoch": 32.93479853479853, "grad_norm": 1.0, "learning_rate": 1.94004778834929e-06, "loss": 0.1182, "step": 44956 }, { "epoch": 32.935531135531136, "grad_norm": 0.84375, "learning_rate": 1.9393833223860445e-06, "loss": 0.1179, "step": 44957 }, { "epoch": 32.93626373626373, "grad_norm": 0.62109375, "learning_rate": 1.938719089092657e-06, "loss": 0.1036, "step": 44958 }, { "epoch": 32.93699633699634, "grad_norm": 0.97265625, "learning_rate": 1.938055088472305e-06, "loss": 0.1064, "step": 44959 }, { "epoch": 32.937728937728934, "grad_norm": 0.84375, "learning_rate": 1.9373913205281802e-06, "loss": 0.1108, "step": 44960 }, { "epoch": 32.93846153846154, "grad_norm": 1.2578125, "learning_rate": 1.9367277852634488e-06, "loss": 0.1069, "step": 44961 }, { "epoch": 32.939194139194136, "grad_norm": 0.9765625, "learning_rate": 1.9360644826812983e-06, "loss": 0.1185, "step": 44962 }, { "epoch": 32.93992673992674, "grad_norm": 0.97265625, "learning_rate": 1.935401412784905e-06, "loss": 0.1207, "step": 44963 }, { "epoch": 32.94065934065934, "grad_norm": 1.03125, "learning_rate": 1.934738575577451e-06, "loss": 0.1088, "step": 44964 }, { "epoch": 32.94139194139194, "grad_norm": 1.1015625, "learning_rate": 1.934075971062106e-06, "loss": 0.1067, "step": 44965 }, { "epoch": 32.94212454212454, "grad_norm": 1.8203125, "learning_rate": 1.9334135992420433e-06, "loss": 0.1497, "step": 44966 }, { "epoch": 32.94285714285714, "grad_norm": 1.40625, "learning_rate": 1.932751460120443e-06, "loss": 0.1083, "step": 44967 }, { "epoch": 32.94358974358974, "grad_norm": 1.453125, "learning_rate": 1.9320895537004715e-06, "loss": 0.1075, "step": 44968 }, { "epoch": 32.944322344322345, "grad_norm": 0.8203125, "learning_rate": 1.9314278799852986e-06, "loss": 0.1158, "step": 44969 }, { "epoch": 32.94505494505494, "grad_norm": 0.8515625, "learning_rate": 1.9307664389781026e-06, "loss": 0.1191, "step": 44970 }, { "epoch": 32.945787545787546, "grad_norm": 0.58984375, "learning_rate": 1.930105230682053e-06, "loss": 0.101, "step": 44971 }, { "epoch": 32.94652014652014, "grad_norm": 0.609375, "learning_rate": 1.9294442551003105e-06, "loss": 0.1075, "step": 44972 }, { "epoch": 32.94725274725275, "grad_norm": 0.875, "learning_rate": 1.928783512236045e-06, "loss": 0.1475, "step": 44973 }, { "epoch": 32.947985347985345, "grad_norm": 0.94921875, "learning_rate": 1.9281230020924185e-06, "loss": 0.1141, "step": 44974 }, { "epoch": 32.94871794871795, "grad_norm": 0.640625, "learning_rate": 1.927462724672605e-06, "loss": 0.104, "step": 44975 }, { "epoch": 32.949450549450546, "grad_norm": 1.2421875, "learning_rate": 1.9268026799797613e-06, "loss": 0.1308, "step": 44976 }, { "epoch": 32.95018315018315, "grad_norm": 0.89453125, "learning_rate": 1.926142868017051e-06, "loss": 0.1099, "step": 44977 }, { "epoch": 32.95091575091575, "grad_norm": 1.9296875, "learning_rate": 1.9254832887876407e-06, "loss": 0.1448, "step": 44978 }, { "epoch": 32.95164835164835, "grad_norm": 0.9765625, "learning_rate": 1.9248239422946785e-06, "loss": 0.093, "step": 44979 }, { "epoch": 32.95238095238095, "grad_norm": 1.4296875, "learning_rate": 1.9241648285413363e-06, "loss": 0.1029, "step": 44980 }, { "epoch": 32.95311355311355, "grad_norm": 0.91015625, "learning_rate": 1.9235059475307627e-06, "loss": 0.1197, "step": 44981 }, { "epoch": 32.95384615384615, "grad_norm": 1.3359375, "learning_rate": 1.9228472992661227e-06, "loss": 0.1322, "step": 44982 }, { "epoch": 32.954578754578755, "grad_norm": 1.1875, "learning_rate": 1.922188883750566e-06, "loss": 0.1436, "step": 44983 }, { "epoch": 32.95531135531135, "grad_norm": 0.76953125, "learning_rate": 1.921530700987253e-06, "loss": 0.1017, "step": 44984 }, { "epoch": 32.956043956043956, "grad_norm": 1.7265625, "learning_rate": 1.9208727509793252e-06, "loss": 0.1225, "step": 44985 }, { "epoch": 32.95677655677655, "grad_norm": 1.40625, "learning_rate": 1.9202150337299557e-06, "loss": 0.1137, "step": 44986 }, { "epoch": 32.95750915750916, "grad_norm": 0.73828125, "learning_rate": 1.9195575492422754e-06, "loss": 0.1111, "step": 44987 }, { "epoch": 32.958241758241755, "grad_norm": 0.8515625, "learning_rate": 1.91890029751944e-06, "loss": 0.1298, "step": 44988 }, { "epoch": 32.95897435897436, "grad_norm": 0.640625, "learning_rate": 1.9182432785646036e-06, "loss": 0.0991, "step": 44989 }, { "epoch": 32.959706959706956, "grad_norm": 1.5625, "learning_rate": 1.9175864923809207e-06, "loss": 0.1229, "step": 44990 }, { "epoch": 32.96043956043956, "grad_norm": 0.89453125, "learning_rate": 1.916929938971523e-06, "loss": 0.1208, "step": 44991 }, { "epoch": 32.96117216117216, "grad_norm": 1.0703125, "learning_rate": 1.9162736183395607e-06, "loss": 0.1049, "step": 44992 }, { "epoch": 32.96190476190476, "grad_norm": 0.80078125, "learning_rate": 1.9156175304881817e-06, "loss": 0.1149, "step": 44993 }, { "epoch": 32.96263736263736, "grad_norm": 1.109375, "learning_rate": 1.91496167542053e-06, "loss": 0.1333, "step": 44994 }, { "epoch": 32.96336996336996, "grad_norm": 0.953125, "learning_rate": 1.9143060531397438e-06, "loss": 0.1153, "step": 44995 }, { "epoch": 32.96410256410256, "grad_norm": 0.63671875, "learning_rate": 1.9136506636489664e-06, "loss": 0.1179, "step": 44996 }, { "epoch": 32.964835164835165, "grad_norm": 1.1953125, "learning_rate": 1.912995506951347e-06, "loss": 0.1193, "step": 44997 }, { "epoch": 32.96556776556776, "grad_norm": 1.0859375, "learning_rate": 1.912340583050006e-06, "loss": 0.1142, "step": 44998 }, { "epoch": 32.966300366300366, "grad_norm": 0.76953125, "learning_rate": 1.9116858919481e-06, "loss": 0.1262, "step": 44999 }, { "epoch": 32.967032967032964, "grad_norm": 0.97265625, "learning_rate": 1.911031433648748e-06, "loss": 0.137, "step": 45000 }, { "epoch": 32.96776556776557, "grad_norm": 0.83203125, "learning_rate": 1.910377208155101e-06, "loss": 0.1318, "step": 45001 }, { "epoch": 32.968498168498165, "grad_norm": 0.8984375, "learning_rate": 1.9097232154702847e-06, "loss": 0.1042, "step": 45002 }, { "epoch": 32.96923076923077, "grad_norm": 0.9140625, "learning_rate": 1.9090694555974385e-06, "loss": 0.1, "step": 45003 }, { "epoch": 32.96996336996337, "grad_norm": 0.73046875, "learning_rate": 1.9084159285396934e-06, "loss": 0.1221, "step": 45004 }, { "epoch": 32.97069597069597, "grad_norm": 1.4140625, "learning_rate": 1.907762634300171e-06, "loss": 0.1301, "step": 45005 }, { "epoch": 32.97142857142857, "grad_norm": 0.984375, "learning_rate": 1.9071095728820162e-06, "loss": 0.1041, "step": 45006 }, { "epoch": 32.97216117216117, "grad_norm": 1.1796875, "learning_rate": 1.906456744288349e-06, "loss": 0.1482, "step": 45007 }, { "epoch": 32.97289377289377, "grad_norm": 0.73828125, "learning_rate": 1.9058041485223079e-06, "loss": 0.1, "step": 45008 }, { "epoch": 32.973626373626374, "grad_norm": 0.9609375, "learning_rate": 1.9051517855870039e-06, "loss": 0.1198, "step": 45009 }, { "epoch": 32.97435897435897, "grad_norm": 1.2421875, "learning_rate": 1.904499655485569e-06, "loss": 0.1347, "step": 45010 }, { "epoch": 32.975091575091575, "grad_norm": 0.65625, "learning_rate": 1.90384775822113e-06, "loss": 0.1143, "step": 45011 }, { "epoch": 32.97582417582417, "grad_norm": 1.0859375, "learning_rate": 1.9031960937968142e-06, "loss": 0.1277, "step": 45012 }, { "epoch": 32.97655677655678, "grad_norm": 1.359375, "learning_rate": 1.902544662215732e-06, "loss": 0.1005, "step": 45013 }, { "epoch": 32.977289377289374, "grad_norm": 1.046875, "learning_rate": 1.9018934634810165e-06, "loss": 0.1454, "step": 45014 }, { "epoch": 32.97802197802198, "grad_norm": 1.0, "learning_rate": 1.9012424975957829e-06, "loss": 0.1191, "step": 45015 }, { "epoch": 32.978754578754575, "grad_norm": 0.94921875, "learning_rate": 1.9005917645631533e-06, "loss": 0.1176, "step": 45016 }, { "epoch": 32.97948717948718, "grad_norm": 1.09375, "learning_rate": 1.8999412643862375e-06, "loss": 0.1242, "step": 45017 }, { "epoch": 32.98021978021978, "grad_norm": 1.25, "learning_rate": 1.8992909970681576e-06, "loss": 0.1168, "step": 45018 }, { "epoch": 32.98095238095238, "grad_norm": 0.984375, "learning_rate": 1.8986409626120294e-06, "loss": 0.1667, "step": 45019 }, { "epoch": 32.98168498168498, "grad_norm": 1.4140625, "learning_rate": 1.897991161020969e-06, "loss": 0.1077, "step": 45020 }, { "epoch": 32.98241758241758, "grad_norm": 0.84375, "learning_rate": 1.8973415922980867e-06, "loss": 0.1198, "step": 45021 }, { "epoch": 32.98315018315018, "grad_norm": 1.7109375, "learning_rate": 1.8966922564464987e-06, "loss": 0.1206, "step": 45022 }, { "epoch": 32.983882783882784, "grad_norm": 1.21875, "learning_rate": 1.89604315346931e-06, "loss": 0.1188, "step": 45023 }, { "epoch": 32.98461538461538, "grad_norm": 1.40625, "learning_rate": 1.8953942833696365e-06, "loss": 0.0905, "step": 45024 }, { "epoch": 32.985347985347985, "grad_norm": 1.90625, "learning_rate": 1.8947456461505887e-06, "loss": 0.1266, "step": 45025 }, { "epoch": 32.98608058608059, "grad_norm": 1.65625, "learning_rate": 1.894097241815261e-06, "loss": 0.1232, "step": 45026 }, { "epoch": 32.98681318681319, "grad_norm": 0.94140625, "learning_rate": 1.8934490703667742e-06, "loss": 0.1111, "step": 45027 }, { "epoch": 32.987545787545784, "grad_norm": 0.765625, "learning_rate": 1.8928011318082285e-06, "loss": 0.1304, "step": 45028 }, { "epoch": 32.98827838827839, "grad_norm": 1.3828125, "learning_rate": 1.8921534261427288e-06, "loss": 0.1179, "step": 45029 }, { "epoch": 32.98901098901099, "grad_norm": 1.0625, "learning_rate": 1.8915059533733854e-06, "loss": 0.1391, "step": 45030 }, { "epoch": 32.98974358974359, "grad_norm": 0.83203125, "learning_rate": 1.8908587135032865e-06, "loss": 0.1421, "step": 45031 }, { "epoch": 32.99047619047619, "grad_norm": 0.86328125, "learning_rate": 1.8902117065355434e-06, "loss": 0.0945, "step": 45032 }, { "epoch": 32.99120879120879, "grad_norm": 1.2890625, "learning_rate": 1.889564932473255e-06, "loss": 0.1369, "step": 45033 }, { "epoch": 32.991941391941396, "grad_norm": 0.91796875, "learning_rate": 1.888918391319521e-06, "loss": 0.1165, "step": 45034 }, { "epoch": 32.99267399267399, "grad_norm": 1.125, "learning_rate": 1.8882720830774303e-06, "loss": 0.142, "step": 45035 }, { "epoch": 32.99340659340659, "grad_norm": 1.0234375, "learning_rate": 1.887626007750093e-06, "loss": 0.1136, "step": 45036 }, { "epoch": 32.994139194139194, "grad_norm": 1.078125, "learning_rate": 1.886980165340592e-06, "loss": 0.1144, "step": 45037 }, { "epoch": 32.9948717948718, "grad_norm": 1.28125, "learning_rate": 1.8863345558520325e-06, "loss": 0.1166, "step": 45038 }, { "epoch": 32.995604395604396, "grad_norm": 1.046875, "learning_rate": 1.8856891792874978e-06, "loss": 0.1165, "step": 45039 }, { "epoch": 32.99633699633699, "grad_norm": 1.359375, "learning_rate": 1.8850440356500875e-06, "loss": 0.1416, "step": 45040 }, { "epoch": 32.9970695970696, "grad_norm": 0.6796875, "learning_rate": 1.8843991249428892e-06, "loss": 0.1137, "step": 45041 }, { "epoch": 32.9978021978022, "grad_norm": 0.92578125, "learning_rate": 1.8837544471689978e-06, "loss": 0.1257, "step": 45042 }, { "epoch": 32.9985347985348, "grad_norm": 0.70703125, "learning_rate": 1.8831100023314958e-06, "loss": 0.1361, "step": 45043 }, { "epoch": 32.999267399267396, "grad_norm": 1.5, "learning_rate": 1.882465790433478e-06, "loss": 0.123, "step": 45044 }, { "epoch": 33.0, "grad_norm": 0.80078125, "learning_rate": 1.8818218114780268e-06, "loss": 0.1074, "step": 45045 }, { "epoch": 33.000732600732604, "grad_norm": 0.828125, "learning_rate": 1.8811780654682195e-06, "loss": 0.1231, "step": 45046 }, { "epoch": 33.0014652014652, "grad_norm": 0.703125, "learning_rate": 1.880534552407151e-06, "loss": 0.1101, "step": 45047 }, { "epoch": 33.002197802197806, "grad_norm": 1.0859375, "learning_rate": 1.8798912722979033e-06, "loss": 0.1304, "step": 45048 }, { "epoch": 33.0029304029304, "grad_norm": 0.88671875, "learning_rate": 1.8792482251435658e-06, "loss": 0.1344, "step": 45049 }, { "epoch": 33.00366300366301, "grad_norm": 0.87109375, "learning_rate": 1.878605410947199e-06, "loss": 0.1207, "step": 45050 }, { "epoch": 33.004395604395604, "grad_norm": 0.9921875, "learning_rate": 1.8779628297118975e-06, "loss": 0.126, "step": 45051 }, { "epoch": 33.00512820512821, "grad_norm": 1.109375, "learning_rate": 1.877320481440744e-06, "loss": 0.1015, "step": 45052 }, { "epoch": 33.005860805860806, "grad_norm": 1.4375, "learning_rate": 1.8766783661368048e-06, "loss": 0.1053, "step": 45053 }, { "epoch": 33.00659340659341, "grad_norm": 1.3046875, "learning_rate": 1.876036483803163e-06, "loss": 0.1396, "step": 45054 }, { "epoch": 33.00732600732601, "grad_norm": 1.5546875, "learning_rate": 1.8753948344428908e-06, "loss": 0.1297, "step": 45055 }, { "epoch": 33.00805860805861, "grad_norm": 0.62109375, "learning_rate": 1.8747534180590713e-06, "loss": 0.1129, "step": 45056 }, { "epoch": 33.00879120879121, "grad_norm": 0.91015625, "learning_rate": 1.8741122346547604e-06, "loss": 0.1306, "step": 45057 }, { "epoch": 33.00952380952381, "grad_norm": 1.09375, "learning_rate": 1.8734712842330456e-06, "loss": 0.1312, "step": 45058 }, { "epoch": 33.01025641025641, "grad_norm": 1.0390625, "learning_rate": 1.8728305667969945e-06, "loss": 0.0964, "step": 45059 }, { "epoch": 33.010989010989015, "grad_norm": 1.359375, "learning_rate": 1.872190082349673e-06, "loss": 0.1221, "step": 45060 }, { "epoch": 33.01172161172161, "grad_norm": 1.421875, "learning_rate": 1.8715498308941533e-06, "loss": 0.1265, "step": 45061 }, { "epoch": 33.012454212454216, "grad_norm": 1.171875, "learning_rate": 1.8709098124335023e-06, "loss": 0.119, "step": 45062 }, { "epoch": 33.01318681318681, "grad_norm": 1.0703125, "learning_rate": 1.8702700269707858e-06, "loss": 0.1197, "step": 45063 }, { "epoch": 33.01391941391942, "grad_norm": 1.4140625, "learning_rate": 1.869630474509071e-06, "loss": 0.1395, "step": 45064 }, { "epoch": 33.014652014652015, "grad_norm": 1.0390625, "learning_rate": 1.8689911550514187e-06, "loss": 0.1307, "step": 45065 }, { "epoch": 33.01538461538462, "grad_norm": 0.80078125, "learning_rate": 1.8683520686008957e-06, "loss": 0.1138, "step": 45066 }, { "epoch": 33.016117216117216, "grad_norm": 0.91796875, "learning_rate": 1.8677132151605575e-06, "loss": 0.1305, "step": 45067 }, { "epoch": 33.01684981684982, "grad_norm": 0.84765625, "learning_rate": 1.8670745947334755e-06, "loss": 0.1386, "step": 45068 }, { "epoch": 33.01758241758242, "grad_norm": 0.68359375, "learning_rate": 1.8664362073227062e-06, "loss": 0.1082, "step": 45069 }, { "epoch": 33.01831501831502, "grad_norm": 1.1328125, "learning_rate": 1.8657980529313044e-06, "loss": 0.1362, "step": 45070 }, { "epoch": 33.01904761904762, "grad_norm": 1.2578125, "learning_rate": 1.8651601315623313e-06, "loss": 0.1301, "step": 45071 }, { "epoch": 33.01978021978022, "grad_norm": 1.1953125, "learning_rate": 1.8645224432188372e-06, "loss": 0.144, "step": 45072 }, { "epoch": 33.02051282051282, "grad_norm": 0.9296875, "learning_rate": 1.8638849879038834e-06, "loss": 0.1019, "step": 45073 }, { "epoch": 33.021245421245425, "grad_norm": 0.91015625, "learning_rate": 1.8632477656205249e-06, "loss": 0.1342, "step": 45074 }, { "epoch": 33.02197802197802, "grad_norm": 0.6640625, "learning_rate": 1.862610776371812e-06, "loss": 0.1186, "step": 45075 }, { "epoch": 33.022710622710626, "grad_norm": 0.73828125, "learning_rate": 1.8619740201608002e-06, "loss": 0.128, "step": 45076 }, { "epoch": 33.02344322344322, "grad_norm": 0.88671875, "learning_rate": 1.8613374969905283e-06, "loss": 0.1032, "step": 45077 }, { "epoch": 33.02417582417583, "grad_norm": 1.6875, "learning_rate": 1.8607012068640636e-06, "loss": 0.1675, "step": 45078 }, { "epoch": 33.024908424908425, "grad_norm": 1.5, "learning_rate": 1.8600651497844502e-06, "loss": 0.1394, "step": 45079 }, { "epoch": 33.02564102564103, "grad_norm": 1.3828125, "learning_rate": 1.859429325754722e-06, "loss": 0.1363, "step": 45080 }, { "epoch": 33.026373626373626, "grad_norm": 0.86328125, "learning_rate": 1.8587937347779388e-06, "loss": 0.0955, "step": 45081 }, { "epoch": 33.02710622710623, "grad_norm": 1.15625, "learning_rate": 1.858158376857147e-06, "loss": 0.1388, "step": 45082 }, { "epoch": 33.02783882783883, "grad_norm": 1.1953125, "learning_rate": 1.8575232519953787e-06, "loss": 0.1353, "step": 45083 }, { "epoch": 33.02857142857143, "grad_norm": 0.859375, "learning_rate": 1.8568883601956958e-06, "loss": 0.1034, "step": 45084 }, { "epoch": 33.02930402930403, "grad_norm": 0.87890625, "learning_rate": 1.8562537014611205e-06, "loss": 0.126, "step": 45085 }, { "epoch": 33.03003663003663, "grad_norm": 1.09375, "learning_rate": 1.8556192757947088e-06, "loss": 0.1319, "step": 45086 }, { "epoch": 33.03076923076923, "grad_norm": 0.93359375, "learning_rate": 1.8549850831994877e-06, "loss": 0.1306, "step": 45087 }, { "epoch": 33.031501831501835, "grad_norm": 0.91796875, "learning_rate": 1.8543511236785033e-06, "loss": 0.1306, "step": 45088 }, { "epoch": 33.03223443223443, "grad_norm": 1.1328125, "learning_rate": 1.8537173972347935e-06, "loss": 0.1178, "step": 45089 }, { "epoch": 33.032967032967036, "grad_norm": 0.90234375, "learning_rate": 1.8530839038713977e-06, "loss": 0.129, "step": 45090 }, { "epoch": 33.033699633699634, "grad_norm": 1.0078125, "learning_rate": 1.8524506435913385e-06, "loss": 0.0998, "step": 45091 }, { "epoch": 33.03443223443224, "grad_norm": 1.4140625, "learning_rate": 1.8518176163976607e-06, "loss": 0.1298, "step": 45092 }, { "epoch": 33.035164835164835, "grad_norm": 0.953125, "learning_rate": 1.8511848222933976e-06, "loss": 0.123, "step": 45093 }, { "epoch": 33.03589743589744, "grad_norm": 0.87890625, "learning_rate": 1.8505522612815775e-06, "loss": 0.1419, "step": 45094 }, { "epoch": 33.03663003663004, "grad_norm": 0.75, "learning_rate": 1.8499199333652337e-06, "loss": 0.1085, "step": 45095 }, { "epoch": 33.03736263736264, "grad_norm": 0.703125, "learning_rate": 1.8492878385473886e-06, "loss": 0.1055, "step": 45096 }, { "epoch": 33.03809523809524, "grad_norm": 0.83984375, "learning_rate": 1.8486559768310822e-06, "loss": 0.1348, "step": 45097 }, { "epoch": 33.03882783882784, "grad_norm": 1.0, "learning_rate": 1.848024348219336e-06, "loss": 0.1172, "step": 45098 }, { "epoch": 33.03956043956044, "grad_norm": 1.125, "learning_rate": 1.8473929527151734e-06, "loss": 0.1299, "step": 45099 }, { "epoch": 33.040293040293044, "grad_norm": 0.75390625, "learning_rate": 1.8467617903216217e-06, "loss": 0.1189, "step": 45100 }, { "epoch": 33.04102564102564, "grad_norm": 0.8671875, "learning_rate": 1.8461308610417096e-06, "loss": 0.1144, "step": 45101 }, { "epoch": 33.041758241758245, "grad_norm": 0.7734375, "learning_rate": 1.845500164878454e-06, "loss": 0.1216, "step": 45102 }, { "epoch": 33.04249084249084, "grad_norm": 1.4609375, "learning_rate": 1.8448697018348827e-06, "loss": 0.1173, "step": 45103 }, { "epoch": 33.04322344322345, "grad_norm": 1.1796875, "learning_rate": 1.8442394719140076e-06, "loss": 0.1561, "step": 45104 }, { "epoch": 33.043956043956044, "grad_norm": 0.91796875, "learning_rate": 1.8436094751188626e-06, "loss": 0.1006, "step": 45105 }, { "epoch": 33.04468864468865, "grad_norm": 1.265625, "learning_rate": 1.8429797114524475e-06, "loss": 0.1399, "step": 45106 }, { "epoch": 33.045421245421245, "grad_norm": 1.2578125, "learning_rate": 1.8423501809177962e-06, "loss": 0.1281, "step": 45107 }, { "epoch": 33.04615384615385, "grad_norm": 0.95703125, "learning_rate": 1.8417208835179153e-06, "loss": 0.1165, "step": 45108 }, { "epoch": 33.04688644688645, "grad_norm": 1.0390625, "learning_rate": 1.8410918192558214e-06, "loss": 0.1146, "step": 45109 }, { "epoch": 33.04761904761905, "grad_norm": 0.73828125, "learning_rate": 1.8404629881345372e-06, "loss": 0.1245, "step": 45110 }, { "epoch": 33.04835164835165, "grad_norm": 1.1796875, "learning_rate": 1.839834390157063e-06, "loss": 0.1406, "step": 45111 }, { "epoch": 33.04908424908425, "grad_norm": 1.015625, "learning_rate": 1.8392060253264218e-06, "loss": 0.1083, "step": 45112 }, { "epoch": 33.04981684981685, "grad_norm": 1.15625, "learning_rate": 1.838577893645614e-06, "loss": 0.1069, "step": 45113 }, { "epoch": 33.050549450549454, "grad_norm": 1.1015625, "learning_rate": 1.8379499951176514e-06, "loss": 0.1147, "step": 45114 }, { "epoch": 33.05128205128205, "grad_norm": 1.1953125, "learning_rate": 1.8373223297455507e-06, "loss": 0.1319, "step": 45115 }, { "epoch": 33.052014652014655, "grad_norm": 0.9140625, "learning_rate": 1.836694897532313e-06, "loss": 0.1345, "step": 45116 }, { "epoch": 33.05274725274725, "grad_norm": 0.859375, "learning_rate": 1.8360676984809436e-06, "loss": 0.0978, "step": 45117 }, { "epoch": 33.05347985347986, "grad_norm": 1.296875, "learning_rate": 1.8354407325944498e-06, "loss": 0.1137, "step": 45118 }, { "epoch": 33.054212454212454, "grad_norm": 1.453125, "learning_rate": 1.8348139998758309e-06, "loss": 0.0976, "step": 45119 }, { "epoch": 33.05494505494506, "grad_norm": 0.64453125, "learning_rate": 1.834187500328094e-06, "loss": 0.1226, "step": 45120 }, { "epoch": 33.055677655677655, "grad_norm": 1.0625, "learning_rate": 1.83356123395425e-06, "loss": 0.1422, "step": 45121 }, { "epoch": 33.05641025641026, "grad_norm": 0.8671875, "learning_rate": 1.8329352007572833e-06, "loss": 0.1126, "step": 45122 }, { "epoch": 33.05714285714286, "grad_norm": 0.91015625, "learning_rate": 1.8323094007402058e-06, "loss": 0.1259, "step": 45123 }, { "epoch": 33.05787545787546, "grad_norm": 0.92578125, "learning_rate": 1.8316838339060011e-06, "loss": 0.0859, "step": 45124 }, { "epoch": 33.05860805860806, "grad_norm": 0.78125, "learning_rate": 1.831058500257681e-06, "loss": 0.1221, "step": 45125 }, { "epoch": 33.05934065934066, "grad_norm": 0.53515625, "learning_rate": 1.830433399798235e-06, "loss": 0.1114, "step": 45126 }, { "epoch": 33.06007326007326, "grad_norm": 1.3828125, "learning_rate": 1.8298085325306643e-06, "loss": 0.1262, "step": 45127 }, { "epoch": 33.060805860805864, "grad_norm": 1.515625, "learning_rate": 1.8291838984579573e-06, "loss": 0.1382, "step": 45128 }, { "epoch": 33.06153846153846, "grad_norm": 0.9765625, "learning_rate": 1.8285594975831048e-06, "loss": 0.1299, "step": 45129 }, { "epoch": 33.062271062271066, "grad_norm": 1.203125, "learning_rate": 1.8279353299091014e-06, "loss": 0.1026, "step": 45130 }, { "epoch": 33.06300366300366, "grad_norm": 0.90625, "learning_rate": 1.827311395438942e-06, "loss": 0.1389, "step": 45131 }, { "epoch": 33.06373626373627, "grad_norm": 0.953125, "learning_rate": 1.8266876941756057e-06, "loss": 0.1237, "step": 45132 }, { "epoch": 33.064468864468864, "grad_norm": 1.09375, "learning_rate": 1.8260642261220928e-06, "loss": 0.1522, "step": 45133 }, { "epoch": 33.06520146520147, "grad_norm": 1.0234375, "learning_rate": 1.8254409912813819e-06, "loss": 0.0964, "step": 45134 }, { "epoch": 33.065934065934066, "grad_norm": 0.8984375, "learning_rate": 1.824817989656457e-06, "loss": 0.1263, "step": 45135 }, { "epoch": 33.06666666666667, "grad_norm": 0.55859375, "learning_rate": 1.8241952212503196e-06, "loss": 0.0953, "step": 45136 }, { "epoch": 33.06739926739927, "grad_norm": 0.9140625, "learning_rate": 1.8235726860659358e-06, "loss": 0.1328, "step": 45137 }, { "epoch": 33.06813186813187, "grad_norm": 1.3046875, "learning_rate": 1.8229503841062965e-06, "loss": 0.1504, "step": 45138 }, { "epoch": 33.06886446886447, "grad_norm": 0.921875, "learning_rate": 1.8223283153743797e-06, "loss": 0.1252, "step": 45139 }, { "epoch": 33.06959706959707, "grad_norm": 0.83984375, "learning_rate": 1.8217064798731639e-06, "loss": 0.1339, "step": 45140 }, { "epoch": 33.07032967032967, "grad_norm": 1.125, "learning_rate": 1.8210848776056393e-06, "loss": 0.1042, "step": 45141 }, { "epoch": 33.071062271062274, "grad_norm": 0.81640625, "learning_rate": 1.8204635085747785e-06, "loss": 0.1249, "step": 45142 }, { "epoch": 33.07179487179487, "grad_norm": 0.96875, "learning_rate": 1.8198423727835492e-06, "loss": 0.129, "step": 45143 }, { "epoch": 33.072527472527476, "grad_norm": 0.71484375, "learning_rate": 1.8192214702349414e-06, "loss": 0.1184, "step": 45144 }, { "epoch": 33.07326007326007, "grad_norm": 0.69921875, "learning_rate": 1.8186008009319224e-06, "loss": 0.1117, "step": 45145 }, { "epoch": 33.07399267399268, "grad_norm": 1.1171875, "learning_rate": 1.8179803648774655e-06, "loss": 0.1344, "step": 45146 }, { "epoch": 33.074725274725274, "grad_norm": 0.96484375, "learning_rate": 1.8173601620745544e-06, "loss": 0.1063, "step": 45147 }, { "epoch": 33.07545787545788, "grad_norm": 1.2421875, "learning_rate": 1.8167401925261457e-06, "loss": 0.1075, "step": 45148 }, { "epoch": 33.076190476190476, "grad_norm": 0.99609375, "learning_rate": 1.8161204562352187e-06, "loss": 0.113, "step": 45149 }, { "epoch": 33.07692307692308, "grad_norm": 1.6328125, "learning_rate": 1.8155009532047408e-06, "loss": 0.1351, "step": 45150 }, { "epoch": 33.07765567765568, "grad_norm": 1.1875, "learning_rate": 1.8148816834376741e-06, "loss": 0.0906, "step": 45151 }, { "epoch": 33.07838827838828, "grad_norm": 0.79296875, "learning_rate": 1.8142626469369965e-06, "loss": 0.1183, "step": 45152 }, { "epoch": 33.07912087912088, "grad_norm": 1.2734375, "learning_rate": 1.8136438437056712e-06, "loss": 0.1543, "step": 45153 }, { "epoch": 33.07985347985348, "grad_norm": 1.484375, "learning_rate": 1.81302527374666e-06, "loss": 0.1131, "step": 45154 }, { "epoch": 33.08058608058608, "grad_norm": 1.1171875, "learning_rate": 1.8124069370629242e-06, "loss": 0.109, "step": 45155 }, { "epoch": 33.081318681318685, "grad_norm": 0.88671875, "learning_rate": 1.8117888336574276e-06, "loss": 0.117, "step": 45156 }, { "epoch": 33.08205128205128, "grad_norm": 0.65234375, "learning_rate": 1.8111709635331417e-06, "loss": 0.1158, "step": 45157 }, { "epoch": 33.082783882783886, "grad_norm": 1.6015625, "learning_rate": 1.8105533266930129e-06, "loss": 0.1394, "step": 45158 }, { "epoch": 33.08351648351648, "grad_norm": 0.82421875, "learning_rate": 1.8099359231400086e-06, "loss": 0.1022, "step": 45159 }, { "epoch": 33.08424908424909, "grad_norm": 1.65625, "learning_rate": 1.8093187528770914e-06, "loss": 0.1358, "step": 45160 }, { "epoch": 33.084981684981685, "grad_norm": 0.57421875, "learning_rate": 1.808701815907201e-06, "loss": 0.0985, "step": 45161 }, { "epoch": 33.08571428571429, "grad_norm": 1.3671875, "learning_rate": 1.8080851122333105e-06, "loss": 0.1339, "step": 45162 }, { "epoch": 33.086446886446886, "grad_norm": 1.203125, "learning_rate": 1.807468641858371e-06, "loss": 0.1193, "step": 45163 }, { "epoch": 33.08717948717949, "grad_norm": 0.89453125, "learning_rate": 1.806852404785334e-06, "loss": 0.1067, "step": 45164 }, { "epoch": 33.08791208791209, "grad_norm": 0.72265625, "learning_rate": 1.8062364010171446e-06, "loss": 0.1402, "step": 45165 }, { "epoch": 33.08864468864469, "grad_norm": 0.9140625, "learning_rate": 1.8056206305567658e-06, "loss": 0.1298, "step": 45166 }, { "epoch": 33.08937728937729, "grad_norm": 0.828125, "learning_rate": 1.8050050934071478e-06, "loss": 0.109, "step": 45167 }, { "epoch": 33.09010989010989, "grad_norm": 1.0078125, "learning_rate": 1.8043897895712313e-06, "loss": 0.1137, "step": 45168 }, { "epoch": 33.09084249084249, "grad_norm": 1.25, "learning_rate": 1.803774719051973e-06, "loss": 0.1253, "step": 45169 }, { "epoch": 33.091575091575095, "grad_norm": 0.859375, "learning_rate": 1.8031598818523127e-06, "loss": 0.1138, "step": 45170 }, { "epoch": 33.09230769230769, "grad_norm": 0.8515625, "learning_rate": 1.8025452779751964e-06, "loss": 0.1191, "step": 45171 }, { "epoch": 33.093040293040296, "grad_norm": 1.3046875, "learning_rate": 1.8019309074235748e-06, "loss": 0.1237, "step": 45172 }, { "epoch": 33.09377289377289, "grad_norm": 1.53125, "learning_rate": 1.8013167702003935e-06, "loss": 0.1233, "step": 45173 }, { "epoch": 33.0945054945055, "grad_norm": 1.2890625, "learning_rate": 1.8007028663085876e-06, "loss": 0.1178, "step": 45174 }, { "epoch": 33.095238095238095, "grad_norm": 1.1171875, "learning_rate": 1.8000891957511027e-06, "loss": 0.1289, "step": 45175 }, { "epoch": 33.0959706959707, "grad_norm": 1.140625, "learning_rate": 1.7994757585308733e-06, "loss": 0.1068, "step": 45176 }, { "epoch": 33.096703296703296, "grad_norm": 0.80859375, "learning_rate": 1.7988625546508452e-06, "loss": 0.1099, "step": 45177 }, { "epoch": 33.0974358974359, "grad_norm": 1.09375, "learning_rate": 1.7982495841139525e-06, "loss": 0.137, "step": 45178 }, { "epoch": 33.0981684981685, "grad_norm": 0.8828125, "learning_rate": 1.7976368469231363e-06, "loss": 0.0934, "step": 45179 }, { "epoch": 33.0989010989011, "grad_norm": 1.171875, "learning_rate": 1.79702434308133e-06, "loss": 0.1235, "step": 45180 }, { "epoch": 33.0996336996337, "grad_norm": 1.2421875, "learning_rate": 1.7964120725914643e-06, "loss": 0.1228, "step": 45181 }, { "epoch": 33.100366300366304, "grad_norm": 1.03125, "learning_rate": 1.7958000354564775e-06, "loss": 0.0987, "step": 45182 }, { "epoch": 33.1010989010989, "grad_norm": 1.1328125, "learning_rate": 1.7951882316793005e-06, "loss": 0.1242, "step": 45183 }, { "epoch": 33.101831501831505, "grad_norm": 1.0703125, "learning_rate": 1.7945766612628727e-06, "loss": 0.1344, "step": 45184 }, { "epoch": 33.1025641025641, "grad_norm": 0.79296875, "learning_rate": 1.7939653242101126e-06, "loss": 0.1241, "step": 45185 }, { "epoch": 33.10329670329671, "grad_norm": 1.3828125, "learning_rate": 1.7933542205239538e-06, "loss": 0.1262, "step": 45186 }, { "epoch": 33.104029304029304, "grad_norm": 0.9921875, "learning_rate": 1.792743350207321e-06, "loss": 0.1185, "step": 45187 }, { "epoch": 33.10476190476191, "grad_norm": 0.75, "learning_rate": 1.7921327132631483e-06, "loss": 0.1204, "step": 45188 }, { "epoch": 33.105494505494505, "grad_norm": 0.98046875, "learning_rate": 1.7915223096943545e-06, "loss": 0.1194, "step": 45189 }, { "epoch": 33.10622710622711, "grad_norm": 1.15625, "learning_rate": 1.7909121395038738e-06, "loss": 0.0997, "step": 45190 }, { "epoch": 33.10695970695971, "grad_norm": 1.1953125, "learning_rate": 1.7903022026946193e-06, "loss": 0.1188, "step": 45191 }, { "epoch": 33.10769230769231, "grad_norm": 1.1015625, "learning_rate": 1.7896924992695136e-06, "loss": 0.1323, "step": 45192 }, { "epoch": 33.10842490842491, "grad_norm": 1.0234375, "learning_rate": 1.7890830292314815e-06, "loss": 0.0991, "step": 45193 }, { "epoch": 33.10915750915751, "grad_norm": 0.6171875, "learning_rate": 1.7884737925834519e-06, "loss": 0.1172, "step": 45194 }, { "epoch": 33.10989010989011, "grad_norm": 0.6328125, "learning_rate": 1.7878647893283258e-06, "loss": 0.1057, "step": 45195 }, { "epoch": 33.110622710622714, "grad_norm": 1.234375, "learning_rate": 1.7872560194690386e-06, "loss": 0.1029, "step": 45196 }, { "epoch": 33.11135531135531, "grad_norm": 0.8515625, "learning_rate": 1.7866474830084974e-06, "loss": 0.1183, "step": 45197 }, { "epoch": 33.112087912087915, "grad_norm": 1.109375, "learning_rate": 1.7860391799496091e-06, "loss": 0.1188, "step": 45198 }, { "epoch": 33.11282051282051, "grad_norm": 0.8359375, "learning_rate": 1.7854311102953137e-06, "loss": 0.1098, "step": 45199 }, { "epoch": 33.11355311355312, "grad_norm": 1.3671875, "learning_rate": 1.7848232740485024e-06, "loss": 0.1012, "step": 45200 }, { "epoch": 33.114285714285714, "grad_norm": 0.75390625, "learning_rate": 1.7842156712120985e-06, "loss": 0.1139, "step": 45201 }, { "epoch": 33.11501831501832, "grad_norm": 0.83984375, "learning_rate": 1.783608301789004e-06, "loss": 0.1243, "step": 45202 }, { "epoch": 33.115750915750915, "grad_norm": 1.328125, "learning_rate": 1.7830011657821365e-06, "loss": 0.1181, "step": 45203 }, { "epoch": 33.11648351648352, "grad_norm": 1.0546875, "learning_rate": 1.7823942631944035e-06, "loss": 0.164, "step": 45204 }, { "epoch": 33.11721611721612, "grad_norm": 0.9921875, "learning_rate": 1.781787594028718e-06, "loss": 0.154, "step": 45205 }, { "epoch": 33.11794871794872, "grad_norm": 1.6484375, "learning_rate": 1.7811811582879756e-06, "loss": 0.162, "step": 45206 }, { "epoch": 33.11868131868132, "grad_norm": 0.953125, "learning_rate": 1.7805749559750895e-06, "loss": 0.1386, "step": 45207 }, { "epoch": 33.11941391941392, "grad_norm": 1.0625, "learning_rate": 1.7799689870929612e-06, "loss": 0.1007, "step": 45208 }, { "epoch": 33.12014652014652, "grad_norm": 1.546875, "learning_rate": 1.779363251644492e-06, "loss": 0.1142, "step": 45209 }, { "epoch": 33.120879120879124, "grad_norm": 1.296875, "learning_rate": 1.7787577496325949e-06, "loss": 0.1342, "step": 45210 }, { "epoch": 33.12161172161172, "grad_norm": 1.0546875, "learning_rate": 1.7781524810601553e-06, "loss": 0.093, "step": 45211 }, { "epoch": 33.122344322344325, "grad_norm": 0.90234375, "learning_rate": 1.777547445930091e-06, "loss": 0.1251, "step": 45212 }, { "epoch": 33.12307692307692, "grad_norm": 0.75390625, "learning_rate": 1.7769426442452817e-06, "loss": 0.1121, "step": 45213 }, { "epoch": 33.12380952380953, "grad_norm": 0.87109375, "learning_rate": 1.7763380760086403e-06, "loss": 0.1059, "step": 45214 }, { "epoch": 33.124542124542124, "grad_norm": 0.9921875, "learning_rate": 1.7757337412230575e-06, "loss": 0.1081, "step": 45215 }, { "epoch": 33.12527472527473, "grad_norm": 1.0078125, "learning_rate": 1.7751296398914346e-06, "loss": 0.1072, "step": 45216 }, { "epoch": 33.126007326007326, "grad_norm": 1.0859375, "learning_rate": 1.7745257720166518e-06, "loss": 0.0981, "step": 45217 }, { "epoch": 33.12673992673993, "grad_norm": 0.9453125, "learning_rate": 1.7739221376016155e-06, "loss": 0.1145, "step": 45218 }, { "epoch": 33.12747252747253, "grad_norm": 0.9453125, "learning_rate": 1.773318736649212e-06, "loss": 0.1109, "step": 45219 }, { "epoch": 33.12820512820513, "grad_norm": 1.2890625, "learning_rate": 1.7727155691623369e-06, "loss": 0.1042, "step": 45220 }, { "epoch": 33.12893772893773, "grad_norm": 0.96484375, "learning_rate": 1.7721126351438805e-06, "loss": 0.1124, "step": 45221 }, { "epoch": 33.12967032967033, "grad_norm": 1.2109375, "learning_rate": 1.7715099345967232e-06, "loss": 0.1411, "step": 45222 }, { "epoch": 33.13040293040293, "grad_norm": 1.1796875, "learning_rate": 1.770907467523761e-06, "loss": 0.1035, "step": 45223 }, { "epoch": 33.131135531135534, "grad_norm": 1.0625, "learning_rate": 1.7703052339278734e-06, "loss": 0.1195, "step": 45224 }, { "epoch": 33.13186813186813, "grad_norm": 1.1015625, "learning_rate": 1.7697032338119566e-06, "loss": 0.1102, "step": 45225 }, { "epoch": 33.132600732600736, "grad_norm": 1.109375, "learning_rate": 1.769101467178885e-06, "loss": 0.1461, "step": 45226 }, { "epoch": 33.13333333333333, "grad_norm": 0.68359375, "learning_rate": 1.7684999340315489e-06, "loss": 0.0981, "step": 45227 }, { "epoch": 33.13406593406594, "grad_norm": 1.0390625, "learning_rate": 1.7678986343728228e-06, "loss": 0.1102, "step": 45228 }, { "epoch": 33.134798534798534, "grad_norm": 0.609375, "learning_rate": 1.7672975682055976e-06, "loss": 0.091, "step": 45229 }, { "epoch": 33.13553113553114, "grad_norm": 1.625, "learning_rate": 1.7666967355327412e-06, "loss": 0.1288, "step": 45230 }, { "epoch": 33.136263736263736, "grad_norm": 0.9296875, "learning_rate": 1.7660961363571452e-06, "loss": 0.1147, "step": 45231 }, { "epoch": 33.13699633699634, "grad_norm": 0.83984375, "learning_rate": 1.765495770681678e-06, "loss": 0.1332, "step": 45232 }, { "epoch": 33.13772893772894, "grad_norm": 1.15625, "learning_rate": 1.764895638509219e-06, "loss": 0.1023, "step": 45233 }, { "epoch": 33.13846153846154, "grad_norm": 1.4296875, "learning_rate": 1.764295739842637e-06, "loss": 0.1436, "step": 45234 }, { "epoch": 33.13919413919414, "grad_norm": 0.765625, "learning_rate": 1.763696074684823e-06, "loss": 0.1013, "step": 45235 }, { "epoch": 33.13992673992674, "grad_norm": 1.1015625, "learning_rate": 1.7630966430386403e-06, "loss": 0.118, "step": 45236 }, { "epoch": 33.14065934065934, "grad_norm": 1.046875, "learning_rate": 1.7624974449069517e-06, "loss": 0.1079, "step": 45237 }, { "epoch": 33.141391941391944, "grad_norm": 1.109375, "learning_rate": 1.761898480292648e-06, "loss": 0.1222, "step": 45238 }, { "epoch": 33.14212454212454, "grad_norm": 0.7578125, "learning_rate": 1.7612997491985757e-06, "loss": 0.1198, "step": 45239 }, { "epoch": 33.142857142857146, "grad_norm": 1.1484375, "learning_rate": 1.7607012516276261e-06, "loss": 0.1188, "step": 45240 }, { "epoch": 33.14358974358974, "grad_norm": 1.34375, "learning_rate": 1.7601029875826562e-06, "loss": 0.1008, "step": 45241 }, { "epoch": 33.14432234432235, "grad_norm": 0.95703125, "learning_rate": 1.7595049570665354e-06, "loss": 0.1348, "step": 45242 }, { "epoch": 33.145054945054945, "grad_norm": 0.6328125, "learning_rate": 1.7589071600821208e-06, "loss": 0.1117, "step": 45243 }, { "epoch": 33.14578754578755, "grad_norm": 0.95703125, "learning_rate": 1.7583095966322813e-06, "loss": 0.1382, "step": 45244 }, { "epoch": 33.146520146520146, "grad_norm": 1.546875, "learning_rate": 1.757712266719886e-06, "loss": 0.113, "step": 45245 }, { "epoch": 33.14725274725275, "grad_norm": 1.34375, "learning_rate": 1.7571151703477922e-06, "loss": 0.1249, "step": 45246 }, { "epoch": 33.14798534798535, "grad_norm": 0.82421875, "learning_rate": 1.7565183075188577e-06, "loss": 0.1143, "step": 45247 }, { "epoch": 33.14871794871795, "grad_norm": 1.078125, "learning_rate": 1.7559216782359458e-06, "loss": 0.1166, "step": 45248 }, { "epoch": 33.14945054945055, "grad_norm": 1.015625, "learning_rate": 1.7553252825019195e-06, "loss": 0.1364, "step": 45249 }, { "epoch": 33.15018315018315, "grad_norm": 1.078125, "learning_rate": 1.7547291203196258e-06, "loss": 0.1325, "step": 45250 }, { "epoch": 33.15091575091575, "grad_norm": 0.8984375, "learning_rate": 1.754133191691933e-06, "loss": 0.1408, "step": 45251 }, { "epoch": 33.151648351648355, "grad_norm": 0.5859375, "learning_rate": 1.7535374966216829e-06, "loss": 0.1407, "step": 45252 }, { "epoch": 33.15238095238095, "grad_norm": 1.3125, "learning_rate": 1.7529420351117439e-06, "loss": 0.1476, "step": 45253 }, { "epoch": 33.153113553113556, "grad_norm": 1.1171875, "learning_rate": 1.752346807164957e-06, "loss": 0.1006, "step": 45254 }, { "epoch": 33.15384615384615, "grad_norm": 1.53125, "learning_rate": 1.7517518127841854e-06, "loss": 0.1121, "step": 45255 }, { "epoch": 33.15457875457876, "grad_norm": 1.1484375, "learning_rate": 1.7511570519722655e-06, "loss": 0.1201, "step": 45256 }, { "epoch": 33.155311355311355, "grad_norm": 1.046875, "learning_rate": 1.7505625247320654e-06, "loss": 0.1302, "step": 45257 }, { "epoch": 33.15604395604396, "grad_norm": 0.8046875, "learning_rate": 1.7499682310664155e-06, "loss": 0.1147, "step": 45258 }, { "epoch": 33.156776556776556, "grad_norm": 0.76953125, "learning_rate": 1.7493741709781734e-06, "loss": 0.1022, "step": 45259 }, { "epoch": 33.15750915750916, "grad_norm": 0.68359375, "learning_rate": 1.7487803444701807e-06, "loss": 0.1112, "step": 45260 }, { "epoch": 33.15824175824176, "grad_norm": 1.1484375, "learning_rate": 1.748186751545289e-06, "loss": 0.1005, "step": 45261 }, { "epoch": 33.15897435897436, "grad_norm": 1.75, "learning_rate": 1.7475933922063404e-06, "loss": 0.1556, "step": 45262 }, { "epoch": 33.15970695970696, "grad_norm": 0.9140625, "learning_rate": 1.7470002664561752e-06, "loss": 0.1078, "step": 45263 }, { "epoch": 33.16043956043956, "grad_norm": 1.2421875, "learning_rate": 1.7464073742976408e-06, "loss": 0.1234, "step": 45264 }, { "epoch": 33.16117216117216, "grad_norm": 0.703125, "learning_rate": 1.7458147157335614e-06, "loss": 0.0916, "step": 45265 }, { "epoch": 33.161904761904765, "grad_norm": 1.0234375, "learning_rate": 1.7452222907667952e-06, "loss": 0.1271, "step": 45266 }, { "epoch": 33.16263736263736, "grad_norm": 1.171875, "learning_rate": 1.7446300994001776e-06, "loss": 0.0999, "step": 45267 }, { "epoch": 33.163369963369966, "grad_norm": 0.70703125, "learning_rate": 1.7440381416365392e-06, "loss": 0.0974, "step": 45268 }, { "epoch": 33.16410256410256, "grad_norm": 0.91796875, "learning_rate": 1.7434464174787206e-06, "loss": 0.1223, "step": 45269 }, { "epoch": 33.16483516483517, "grad_norm": 1.375, "learning_rate": 1.7428549269295526e-06, "loss": 0.1319, "step": 45270 }, { "epoch": 33.165567765567765, "grad_norm": 1.015625, "learning_rate": 1.7422636699918765e-06, "loss": 0.1207, "step": 45271 }, { "epoch": 33.16630036630037, "grad_norm": 0.7109375, "learning_rate": 1.7416726466685219e-06, "loss": 0.0964, "step": 45272 }, { "epoch": 33.167032967032966, "grad_norm": 0.9921875, "learning_rate": 1.7410818569623143e-06, "loss": 0.1167, "step": 45273 }, { "epoch": 33.16776556776557, "grad_norm": 1.0390625, "learning_rate": 1.7404913008760941e-06, "loss": 0.1047, "step": 45274 }, { "epoch": 33.16849816849817, "grad_norm": 1.71875, "learning_rate": 1.7399009784126867e-06, "loss": 0.1329, "step": 45275 }, { "epoch": 33.16923076923077, "grad_norm": 0.91796875, "learning_rate": 1.7393108895749166e-06, "loss": 0.1296, "step": 45276 }, { "epoch": 33.16996336996337, "grad_norm": 1.3125, "learning_rate": 1.7387210343656201e-06, "loss": 0.1202, "step": 45277 }, { "epoch": 33.170695970695974, "grad_norm": 0.99609375, "learning_rate": 1.7381314127876154e-06, "loss": 0.1075, "step": 45278 }, { "epoch": 33.17142857142857, "grad_norm": 0.85546875, "learning_rate": 1.737542024843739e-06, "loss": 0.0917, "step": 45279 }, { "epoch": 33.172161172161175, "grad_norm": 0.9609375, "learning_rate": 1.7369528705367986e-06, "loss": 0.1161, "step": 45280 }, { "epoch": 33.17289377289377, "grad_norm": 0.81640625, "learning_rate": 1.7363639498696193e-06, "loss": 0.1016, "step": 45281 }, { "epoch": 33.17362637362638, "grad_norm": 0.74609375, "learning_rate": 1.7357752628450365e-06, "loss": 0.1063, "step": 45282 }, { "epoch": 33.174358974358974, "grad_norm": 1.0, "learning_rate": 1.735186809465865e-06, "loss": 0.116, "step": 45283 }, { "epoch": 33.17509157509158, "grad_norm": 0.87890625, "learning_rate": 1.7345985897349169e-06, "loss": 0.1427, "step": 45284 }, { "epoch": 33.175824175824175, "grad_norm": 0.9765625, "learning_rate": 1.734010603655013e-06, "loss": 0.1337, "step": 45285 }, { "epoch": 33.17655677655678, "grad_norm": 1.0625, "learning_rate": 1.7334228512289771e-06, "loss": 0.1204, "step": 45286 }, { "epoch": 33.17728937728938, "grad_norm": 0.76953125, "learning_rate": 1.7328353324596174e-06, "loss": 0.1082, "step": 45287 }, { "epoch": 33.17802197802198, "grad_norm": 1.21875, "learning_rate": 1.7322480473497594e-06, "loss": 0.1508, "step": 45288 }, { "epoch": 33.17875457875458, "grad_norm": 1.1171875, "learning_rate": 1.731660995902211e-06, "loss": 0.1282, "step": 45289 }, { "epoch": 33.17948717948718, "grad_norm": 1.078125, "learning_rate": 1.73107417811978e-06, "loss": 0.1479, "step": 45290 }, { "epoch": 33.18021978021978, "grad_norm": 0.85546875, "learning_rate": 1.7304875940052805e-06, "loss": 0.1132, "step": 45291 }, { "epoch": 33.180952380952384, "grad_norm": 1.0, "learning_rate": 1.7299012435615317e-06, "loss": 0.1036, "step": 45292 }, { "epoch": 33.18168498168498, "grad_norm": 1.109375, "learning_rate": 1.729315126791331e-06, "loss": 0.1349, "step": 45293 }, { "epoch": 33.182417582417585, "grad_norm": 0.9296875, "learning_rate": 1.7287292436974922e-06, "loss": 0.1382, "step": 45294 }, { "epoch": 33.18315018315018, "grad_norm": 1.3671875, "learning_rate": 1.7281435942828234e-06, "loss": 0.1502, "step": 45295 }, { "epoch": 33.18388278388279, "grad_norm": 0.8515625, "learning_rate": 1.7275581785501275e-06, "loss": 0.1114, "step": 45296 }, { "epoch": 33.184615384615384, "grad_norm": 0.95703125, "learning_rate": 1.7269729965022122e-06, "loss": 0.1236, "step": 45297 }, { "epoch": 33.18534798534799, "grad_norm": 2.140625, "learning_rate": 1.7263880481418807e-06, "loss": 0.1517, "step": 45298 }, { "epoch": 33.186080586080585, "grad_norm": 1.6875, "learning_rate": 1.725803333471936e-06, "loss": 0.1583, "step": 45299 }, { "epoch": 33.18681318681319, "grad_norm": 0.6796875, "learning_rate": 1.7252188524951808e-06, "loss": 0.1065, "step": 45300 }, { "epoch": 33.18754578754579, "grad_norm": 0.94140625, "learning_rate": 1.7246346052144118e-06, "loss": 0.1341, "step": 45301 }, { "epoch": 33.18827838827839, "grad_norm": 1.59375, "learning_rate": 1.7240505916324267e-06, "loss": 0.1553, "step": 45302 }, { "epoch": 33.18901098901099, "grad_norm": 0.84375, "learning_rate": 1.7234668117520336e-06, "loss": 0.1567, "step": 45303 }, { "epoch": 33.18974358974359, "grad_norm": 0.9375, "learning_rate": 1.722883265576019e-06, "loss": 0.1281, "step": 45304 }, { "epoch": 33.19047619047619, "grad_norm": 0.8515625, "learning_rate": 1.7222999531071856e-06, "loss": 0.1275, "step": 45305 }, { "epoch": 33.191208791208794, "grad_norm": 1.0859375, "learning_rate": 1.721716874348325e-06, "loss": 0.1098, "step": 45306 }, { "epoch": 33.19194139194139, "grad_norm": 1.78125, "learning_rate": 1.7211340293022343e-06, "loss": 0.1133, "step": 45307 }, { "epoch": 33.192673992673996, "grad_norm": 0.9921875, "learning_rate": 1.720551417971695e-06, "loss": 0.1269, "step": 45308 }, { "epoch": 33.19340659340659, "grad_norm": 1.109375, "learning_rate": 1.7199690403595197e-06, "loss": 0.1089, "step": 45309 }, { "epoch": 33.1941391941392, "grad_norm": 1.125, "learning_rate": 1.7193868964684792e-06, "loss": 0.1387, "step": 45310 }, { "epoch": 33.194871794871794, "grad_norm": 0.87890625, "learning_rate": 1.718804986301365e-06, "loss": 0.1229, "step": 45311 }, { "epoch": 33.1956043956044, "grad_norm": 0.8203125, "learning_rate": 1.7182233098609741e-06, "loss": 0.1051, "step": 45312 }, { "epoch": 33.196336996336996, "grad_norm": 0.984375, "learning_rate": 1.7176418671500875e-06, "loss": 0.1347, "step": 45313 }, { "epoch": 33.1970695970696, "grad_norm": 1.03125, "learning_rate": 1.7170606581715026e-06, "loss": 0.1214, "step": 45314 }, { "epoch": 33.1978021978022, "grad_norm": 1.1875, "learning_rate": 1.7164796829279832e-06, "loss": 0.113, "step": 45315 }, { "epoch": 33.1985347985348, "grad_norm": 0.5859375, "learning_rate": 1.7158989414223326e-06, "loss": 0.0902, "step": 45316 }, { "epoch": 33.1992673992674, "grad_norm": 1.2890625, "learning_rate": 1.7153184336573147e-06, "loss": 0.1182, "step": 45317 }, { "epoch": 33.2, "grad_norm": 0.97265625, "learning_rate": 1.7147381596357324e-06, "loss": 0.1066, "step": 45318 }, { "epoch": 33.2007326007326, "grad_norm": 0.83984375, "learning_rate": 1.714158119360345e-06, "loss": 0.1077, "step": 45319 }, { "epoch": 33.201465201465204, "grad_norm": 0.6015625, "learning_rate": 1.7135783128339487e-06, "loss": 0.1001, "step": 45320 }, { "epoch": 33.2021978021978, "grad_norm": 1.0234375, "learning_rate": 1.712998740059314e-06, "loss": 0.0993, "step": 45321 }, { "epoch": 33.202930402930406, "grad_norm": 1.2890625, "learning_rate": 1.7124194010392103e-06, "loss": 0.0945, "step": 45322 }, { "epoch": 33.203663003663, "grad_norm": 0.96484375, "learning_rate": 1.7118402957764298e-06, "loss": 0.1107, "step": 45323 }, { "epoch": 33.20439560439561, "grad_norm": 0.9921875, "learning_rate": 1.7112614242737313e-06, "loss": 0.1737, "step": 45324 }, { "epoch": 33.205128205128204, "grad_norm": 1.2578125, "learning_rate": 1.7106827865339008e-06, "loss": 0.1018, "step": 45325 }, { "epoch": 33.20586080586081, "grad_norm": 1.0, "learning_rate": 1.7101043825597077e-06, "loss": 0.169, "step": 45326 }, { "epoch": 33.206593406593406, "grad_norm": 0.953125, "learning_rate": 1.7095262123539174e-06, "loss": 0.1381, "step": 45327 }, { "epoch": 33.20732600732601, "grad_norm": 1.203125, "learning_rate": 1.7089482759192984e-06, "loss": 0.13, "step": 45328 }, { "epoch": 33.20805860805861, "grad_norm": 1.2421875, "learning_rate": 1.7083705732586378e-06, "loss": 0.1376, "step": 45329 }, { "epoch": 33.20879120879121, "grad_norm": 0.640625, "learning_rate": 1.7077931043746835e-06, "loss": 0.1004, "step": 45330 }, { "epoch": 33.20952380952381, "grad_norm": 1.875, "learning_rate": 1.70721586927021e-06, "loss": 0.1541, "step": 45331 }, { "epoch": 33.21025641025641, "grad_norm": 0.875, "learning_rate": 1.7066388679479823e-06, "loss": 0.1016, "step": 45332 }, { "epoch": 33.21098901098901, "grad_norm": 0.734375, "learning_rate": 1.7060621004107644e-06, "loss": 0.1044, "step": 45333 }, { "epoch": 33.211721611721615, "grad_norm": 0.88671875, "learning_rate": 1.7054855666613262e-06, "loss": 0.1437, "step": 45334 }, { "epoch": 33.21245421245421, "grad_norm": 1.171875, "learning_rate": 1.7049092667024208e-06, "loss": 0.1394, "step": 45335 }, { "epoch": 33.213186813186816, "grad_norm": 0.6875, "learning_rate": 1.7043332005368184e-06, "loss": 0.115, "step": 45336 }, { "epoch": 33.21391941391941, "grad_norm": 0.69140625, "learning_rate": 1.7037573681672664e-06, "loss": 0.1129, "step": 45337 }, { "epoch": 33.21465201465202, "grad_norm": 0.97265625, "learning_rate": 1.7031817695965347e-06, "loss": 0.114, "step": 45338 }, { "epoch": 33.215384615384615, "grad_norm": 1.0234375, "learning_rate": 1.7026064048273764e-06, "loss": 0.1251, "step": 45339 }, { "epoch": 33.21611721611722, "grad_norm": 1.4609375, "learning_rate": 1.7020312738625561e-06, "loss": 0.1147, "step": 45340 }, { "epoch": 33.216849816849816, "grad_norm": 0.86328125, "learning_rate": 1.7014563767048217e-06, "loss": 0.1404, "step": 45341 }, { "epoch": 33.21758241758242, "grad_norm": 1.0625, "learning_rate": 1.7008817133569316e-06, "loss": 0.1243, "step": 45342 }, { "epoch": 33.21831501831502, "grad_norm": 1.1484375, "learning_rate": 1.7003072838216227e-06, "loss": 0.1193, "step": 45343 }, { "epoch": 33.21904761904762, "grad_norm": 1.1875, "learning_rate": 1.699733088101676e-06, "loss": 0.0999, "step": 45344 }, { "epoch": 33.21978021978022, "grad_norm": 1.015625, "learning_rate": 1.6991591261998226e-06, "loss": 0.0933, "step": 45345 }, { "epoch": 33.22051282051282, "grad_norm": 1.109375, "learning_rate": 1.6985853981188217e-06, "loss": 0.1288, "step": 45346 }, { "epoch": 33.22124542124542, "grad_norm": 1.140625, "learning_rate": 1.698011903861415e-06, "loss": 0.0995, "step": 45347 }, { "epoch": 33.221978021978025, "grad_norm": 0.70703125, "learning_rate": 1.697438643430356e-06, "loss": 0.1177, "step": 45348 }, { "epoch": 33.22271062271062, "grad_norm": 0.984375, "learning_rate": 1.6968656168283872e-06, "loss": 0.1517, "step": 45349 }, { "epoch": 33.223443223443226, "grad_norm": 0.71875, "learning_rate": 1.6962928240582564e-06, "loss": 0.1366, "step": 45350 }, { "epoch": 33.22417582417582, "grad_norm": 1.2578125, "learning_rate": 1.6957202651227112e-06, "loss": 0.1238, "step": 45351 }, { "epoch": 33.22490842490843, "grad_norm": 1.3125, "learning_rate": 1.6951479400244885e-06, "loss": 0.1452, "step": 45352 }, { "epoch": 33.225641025641025, "grad_norm": 0.9375, "learning_rate": 1.694575848766336e-06, "loss": 0.1132, "step": 45353 }, { "epoch": 33.22637362637363, "grad_norm": 0.859375, "learning_rate": 1.6940039913509852e-06, "loss": 0.1431, "step": 45354 }, { "epoch": 33.227106227106226, "grad_norm": 0.9375, "learning_rate": 1.6934323677811951e-06, "loss": 0.1079, "step": 45355 }, { "epoch": 33.22783882783883, "grad_norm": 1.2578125, "learning_rate": 1.6928609780596852e-06, "loss": 0.161, "step": 45356 }, { "epoch": 33.22857142857143, "grad_norm": 0.921875, "learning_rate": 1.6922898221892043e-06, "loss": 0.1173, "step": 45357 }, { "epoch": 33.22930402930403, "grad_norm": 1.234375, "learning_rate": 1.6917189001724777e-06, "loss": 0.1288, "step": 45358 }, { "epoch": 33.23003663003663, "grad_norm": 0.8515625, "learning_rate": 1.691148212012253e-06, "loss": 0.1232, "step": 45359 }, { "epoch": 33.23076923076923, "grad_norm": 1.171875, "learning_rate": 1.6905777577112619e-06, "loss": 0.1172, "step": 45360 }, { "epoch": 33.23150183150183, "grad_norm": 1.03125, "learning_rate": 1.6900075372722358e-06, "loss": 0.1027, "step": 45361 }, { "epoch": 33.232234432234435, "grad_norm": 1.4140625, "learning_rate": 1.6894375506979057e-06, "loss": 0.1527, "step": 45362 }, { "epoch": 33.23296703296703, "grad_norm": 1.3671875, "learning_rate": 1.6888677979910027e-06, "loss": 0.1072, "step": 45363 }, { "epoch": 33.233699633699636, "grad_norm": 0.80859375, "learning_rate": 1.6882982791542587e-06, "loss": 0.1307, "step": 45364 }, { "epoch": 33.234432234432234, "grad_norm": 1.171875, "learning_rate": 1.6877289941903993e-06, "loss": 0.1035, "step": 45365 }, { "epoch": 33.23516483516484, "grad_norm": 1.109375, "learning_rate": 1.6871599431021613e-06, "loss": 0.1364, "step": 45366 }, { "epoch": 33.235897435897435, "grad_norm": 0.890625, "learning_rate": 1.6865911258922594e-06, "loss": 0.1185, "step": 45367 }, { "epoch": 33.23663003663004, "grad_norm": 1.0546875, "learning_rate": 1.686022542563425e-06, "loss": 0.1315, "step": 45368 }, { "epoch": 33.23736263736264, "grad_norm": 1.2734375, "learning_rate": 1.6854541931183785e-06, "loss": 0.1083, "step": 45369 }, { "epoch": 33.23809523809524, "grad_norm": 1.0234375, "learning_rate": 1.684886077559851e-06, "loss": 0.1475, "step": 45370 }, { "epoch": 33.23882783882784, "grad_norm": 1.0234375, "learning_rate": 1.6843181958905524e-06, "loss": 0.1228, "step": 45371 }, { "epoch": 33.23956043956044, "grad_norm": 1.2109375, "learning_rate": 1.6837505481132187e-06, "loss": 0.1439, "step": 45372 }, { "epoch": 33.24029304029304, "grad_norm": 0.703125, "learning_rate": 1.6831831342305544e-06, "loss": 0.1254, "step": 45373 }, { "epoch": 33.241025641025644, "grad_norm": 1.484375, "learning_rate": 1.6826159542452847e-06, "loss": 0.1414, "step": 45374 }, { "epoch": 33.24175824175824, "grad_norm": 1.7109375, "learning_rate": 1.6820490081601304e-06, "loss": 0.1805, "step": 45375 }, { "epoch": 33.242490842490845, "grad_norm": 0.7890625, "learning_rate": 1.6814822959778063e-06, "loss": 0.1151, "step": 45376 }, { "epoch": 33.24322344322344, "grad_norm": 1.5546875, "learning_rate": 1.6809158177010273e-06, "loss": 0.1115, "step": 45377 }, { "epoch": 33.24395604395605, "grad_norm": 0.68359375, "learning_rate": 1.6803495733325021e-06, "loss": 0.0877, "step": 45378 }, { "epoch": 33.244688644688644, "grad_norm": 0.62890625, "learning_rate": 1.6797835628749574e-06, "loss": 0.1171, "step": 45379 }, { "epoch": 33.24542124542125, "grad_norm": 1.40625, "learning_rate": 1.6792177863310856e-06, "loss": 0.1222, "step": 45380 }, { "epoch": 33.246153846153845, "grad_norm": 1.125, "learning_rate": 1.678652243703612e-06, "loss": 0.1248, "step": 45381 }, { "epoch": 33.24688644688645, "grad_norm": 1.0625, "learning_rate": 1.6780869349952418e-06, "loss": 0.1198, "step": 45382 }, { "epoch": 33.24761904761905, "grad_norm": 1.2734375, "learning_rate": 1.6775218602086885e-06, "loss": 0.0979, "step": 45383 }, { "epoch": 33.24835164835165, "grad_norm": 0.71484375, "learning_rate": 1.6769570193466512e-06, "loss": 0.1163, "step": 45384 }, { "epoch": 33.24908424908425, "grad_norm": 1.2890625, "learning_rate": 1.6763924124118445e-06, "loss": 0.1382, "step": 45385 }, { "epoch": 33.24981684981685, "grad_norm": 0.73046875, "learning_rate": 1.6758280394069668e-06, "loss": 0.0799, "step": 45386 }, { "epoch": 33.25054945054945, "grad_norm": 1.1796875, "learning_rate": 1.6752639003347267e-06, "loss": 0.1357, "step": 45387 }, { "epoch": 33.251282051282054, "grad_norm": 1.3515625, "learning_rate": 1.6746999951978238e-06, "loss": 0.1466, "step": 45388 }, { "epoch": 33.25201465201465, "grad_norm": 1.0078125, "learning_rate": 1.6741363239989559e-06, "loss": 0.148, "step": 45389 }, { "epoch": 33.252747252747255, "grad_norm": 1.4765625, "learning_rate": 1.6735728867408373e-06, "loss": 0.1217, "step": 45390 }, { "epoch": 33.25347985347985, "grad_norm": 0.91796875, "learning_rate": 1.6730096834261562e-06, "loss": 0.111, "step": 45391 }, { "epoch": 33.25421245421246, "grad_norm": 1.0625, "learning_rate": 1.6724467140576161e-06, "loss": 0.1207, "step": 45392 }, { "epoch": 33.254945054945054, "grad_norm": 0.76171875, "learning_rate": 1.6718839786379153e-06, "loss": 0.1488, "step": 45393 }, { "epoch": 33.25567765567766, "grad_norm": 0.96484375, "learning_rate": 1.6713214771697463e-06, "loss": 0.1293, "step": 45394 }, { "epoch": 33.256410256410255, "grad_norm": 0.9140625, "learning_rate": 1.6707592096557972e-06, "loss": 0.1258, "step": 45395 }, { "epoch": 33.25714285714286, "grad_norm": 1.109375, "learning_rate": 1.670197176098782e-06, "loss": 0.1316, "step": 45396 }, { "epoch": 33.25787545787546, "grad_norm": 1.2421875, "learning_rate": 1.6696353765013724e-06, "loss": 0.1211, "step": 45397 }, { "epoch": 33.25860805860806, "grad_norm": 1.0546875, "learning_rate": 1.669073810866277e-06, "loss": 0.1274, "step": 45398 }, { "epoch": 33.25934065934066, "grad_norm": 1.0859375, "learning_rate": 1.6685124791961784e-06, "loss": 0.144, "step": 45399 }, { "epoch": 33.26007326007326, "grad_norm": 1.2890625, "learning_rate": 1.6679513814937631e-06, "loss": 0.1091, "step": 45400 }, { "epoch": 33.26080586080586, "grad_norm": 1.0703125, "learning_rate": 1.6673905177617248e-06, "loss": 0.1588, "step": 45401 }, { "epoch": 33.261538461538464, "grad_norm": 1.21875, "learning_rate": 1.6668298880027445e-06, "loss": 0.1175, "step": 45402 }, { "epoch": 33.26227106227106, "grad_norm": 0.88671875, "learning_rate": 1.6662694922195212e-06, "loss": 0.1089, "step": 45403 }, { "epoch": 33.263003663003666, "grad_norm": 0.97265625, "learning_rate": 1.6657093304147255e-06, "loss": 0.1453, "step": 45404 }, { "epoch": 33.26373626373626, "grad_norm": 0.69921875, "learning_rate": 1.6651494025910556e-06, "loss": 0.1138, "step": 45405 }, { "epoch": 33.26446886446887, "grad_norm": 1.234375, "learning_rate": 1.6645897087511773e-06, "loss": 0.1156, "step": 45406 }, { "epoch": 33.265201465201464, "grad_norm": 1.2421875, "learning_rate": 1.6640302488977885e-06, "loss": 0.126, "step": 45407 }, { "epoch": 33.26593406593407, "grad_norm": 1.078125, "learning_rate": 1.6634710230335546e-06, "loss": 0.121, "step": 45408 }, { "epoch": 33.266666666666666, "grad_norm": 0.90234375, "learning_rate": 1.6629120311611744e-06, "loss": 0.1002, "step": 45409 }, { "epoch": 33.26739926739927, "grad_norm": 1.2109375, "learning_rate": 1.6623532732833077e-06, "loss": 0.1839, "step": 45410 }, { "epoch": 33.26813186813187, "grad_norm": 1.0078125, "learning_rate": 1.6617947494026357e-06, "loss": 0.1482, "step": 45411 }, { "epoch": 33.26886446886447, "grad_norm": 1.2890625, "learning_rate": 1.6612364595218408e-06, "loss": 0.1206, "step": 45412 }, { "epoch": 33.26959706959707, "grad_norm": 0.78125, "learning_rate": 1.6606784036435938e-06, "loss": 0.1373, "step": 45413 }, { "epoch": 33.27032967032967, "grad_norm": 1.0390625, "learning_rate": 1.6601205817705762e-06, "loss": 0.1299, "step": 45414 }, { "epoch": 33.27106227106227, "grad_norm": 0.76953125, "learning_rate": 1.6595629939054482e-06, "loss": 0.1136, "step": 45415 }, { "epoch": 33.271794871794874, "grad_norm": 0.77734375, "learning_rate": 1.6590056400508866e-06, "loss": 0.115, "step": 45416 }, { "epoch": 33.27252747252747, "grad_norm": 0.9609375, "learning_rate": 1.6584485202095616e-06, "loss": 0.1261, "step": 45417 }, { "epoch": 33.273260073260076, "grad_norm": 0.7578125, "learning_rate": 1.6578916343841438e-06, "loss": 0.0916, "step": 45418 }, { "epoch": 33.27399267399267, "grad_norm": 0.95703125, "learning_rate": 1.6573349825773049e-06, "loss": 0.1329, "step": 45419 }, { "epoch": 33.27472527472528, "grad_norm": 1.0625, "learning_rate": 1.6567785647917098e-06, "loss": 0.1344, "step": 45420 }, { "epoch": 33.275457875457874, "grad_norm": 1.171875, "learning_rate": 1.656222381030013e-06, "loss": 0.1506, "step": 45421 }, { "epoch": 33.27619047619048, "grad_norm": 0.99609375, "learning_rate": 1.6556664312948964e-06, "loss": 0.1348, "step": 45422 }, { "epoch": 33.276923076923076, "grad_norm": 1.1015625, "learning_rate": 1.655110715589014e-06, "loss": 0.1402, "step": 45423 }, { "epoch": 33.27765567765568, "grad_norm": 1.4765625, "learning_rate": 1.6545552339150318e-06, "loss": 0.1273, "step": 45424 }, { "epoch": 33.27838827838828, "grad_norm": 1.0390625, "learning_rate": 1.6539999862756092e-06, "loss": 0.1237, "step": 45425 }, { "epoch": 33.27912087912088, "grad_norm": 1.171875, "learning_rate": 1.6534449726734117e-06, "loss": 0.1465, "step": 45426 }, { "epoch": 33.27985347985348, "grad_norm": 1.078125, "learning_rate": 1.6528901931110881e-06, "loss": 0.1086, "step": 45427 }, { "epoch": 33.28058608058608, "grad_norm": 1.2734375, "learning_rate": 1.6523356475913042e-06, "loss": 0.1085, "step": 45428 }, { "epoch": 33.28131868131868, "grad_norm": 0.6796875, "learning_rate": 1.6517813361167246e-06, "loss": 0.1101, "step": 45429 }, { "epoch": 33.282051282051285, "grad_norm": 1.09375, "learning_rate": 1.651227258689988e-06, "loss": 0.1175, "step": 45430 }, { "epoch": 33.28278388278388, "grad_norm": 1.8671875, "learning_rate": 1.6506734153137645e-06, "loss": 0.1322, "step": 45431 }, { "epoch": 33.283516483516486, "grad_norm": 0.75390625, "learning_rate": 1.6501198059906866e-06, "loss": 0.1195, "step": 45432 }, { "epoch": 33.28424908424908, "grad_norm": 0.85546875, "learning_rate": 1.6495664307234371e-06, "loss": 0.1034, "step": 45433 }, { "epoch": 33.28498168498169, "grad_norm": 1.390625, "learning_rate": 1.6490132895146418e-06, "loss": 0.1394, "step": 45434 }, { "epoch": 33.285714285714285, "grad_norm": 1.203125, "learning_rate": 1.6484603823669665e-06, "loss": 0.1429, "step": 45435 }, { "epoch": 33.28644688644689, "grad_norm": 1.1484375, "learning_rate": 1.6479077092830492e-06, "loss": 0.1052, "step": 45436 }, { "epoch": 33.287179487179486, "grad_norm": 0.90625, "learning_rate": 1.6473552702655445e-06, "loss": 0.1303, "step": 45437 }, { "epoch": 33.28791208791209, "grad_norm": 0.9921875, "learning_rate": 1.6468030653171002e-06, "loss": 0.1112, "step": 45438 }, { "epoch": 33.28864468864469, "grad_norm": 1.3125, "learning_rate": 1.6462510944403556e-06, "loss": 0.1497, "step": 45439 }, { "epoch": 33.28937728937729, "grad_norm": 1.375, "learning_rate": 1.6456993576379694e-06, "loss": 0.1059, "step": 45440 }, { "epoch": 33.29010989010989, "grad_norm": 0.91015625, "learning_rate": 1.6451478549125637e-06, "loss": 0.1168, "step": 45441 }, { "epoch": 33.29084249084249, "grad_norm": 1.2109375, "learning_rate": 1.644596586266798e-06, "loss": 0.1216, "step": 45442 }, { "epoch": 33.29157509157509, "grad_norm": 0.9609375, "learning_rate": 1.6440455517033052e-06, "loss": 0.1375, "step": 45443 }, { "epoch": 33.292307692307695, "grad_norm": 0.94140625, "learning_rate": 1.6434947512247334e-06, "loss": 0.1094, "step": 45444 }, { "epoch": 33.29304029304029, "grad_norm": 0.73828125, "learning_rate": 1.6429441848337158e-06, "loss": 0.1345, "step": 45445 }, { "epoch": 33.293772893772896, "grad_norm": 0.76171875, "learning_rate": 1.6423938525328954e-06, "loss": 0.1109, "step": 45446 }, { "epoch": 33.29450549450549, "grad_norm": 0.76953125, "learning_rate": 1.6418437543248935e-06, "loss": 0.0973, "step": 45447 }, { "epoch": 33.2952380952381, "grad_norm": 1.3515625, "learning_rate": 1.6412938902123648e-06, "loss": 0.1242, "step": 45448 }, { "epoch": 33.295970695970695, "grad_norm": 1.1484375, "learning_rate": 1.6407442601979358e-06, "loss": 0.1251, "step": 45449 }, { "epoch": 33.2967032967033, "grad_norm": 0.91796875, "learning_rate": 1.640194864284239e-06, "loss": 0.137, "step": 45450 }, { "epoch": 33.297435897435896, "grad_norm": 1.3671875, "learning_rate": 1.6396457024739072e-06, "loss": 0.117, "step": 45451 }, { "epoch": 33.2981684981685, "grad_norm": 1.28125, "learning_rate": 1.639096774769578e-06, "loss": 0.1283, "step": 45452 }, { "epoch": 33.2989010989011, "grad_norm": 0.859375, "learning_rate": 1.6385480811738672e-06, "loss": 0.1102, "step": 45453 }, { "epoch": 33.2996336996337, "grad_norm": 1.2890625, "learning_rate": 1.6379996216894181e-06, "loss": 0.1465, "step": 45454 }, { "epoch": 33.3003663003663, "grad_norm": 0.74609375, "learning_rate": 1.6374513963188578e-06, "loss": 0.1033, "step": 45455 }, { "epoch": 33.301098901098904, "grad_norm": 0.96875, "learning_rate": 1.6369034050648022e-06, "loss": 0.1132, "step": 45456 }, { "epoch": 33.3018315018315, "grad_norm": 0.99609375, "learning_rate": 1.6363556479298893e-06, "loss": 0.0978, "step": 45457 }, { "epoch": 33.302564102564105, "grad_norm": 1.203125, "learning_rate": 1.635808124916724e-06, "loss": 0.1098, "step": 45458 }, { "epoch": 33.3032967032967, "grad_norm": 1.484375, "learning_rate": 1.6352608360279547e-06, "loss": 0.1177, "step": 45459 }, { "epoch": 33.30402930402931, "grad_norm": 0.84765625, "learning_rate": 1.634713781266187e-06, "loss": 0.0918, "step": 45460 }, { "epoch": 33.304761904761904, "grad_norm": 0.90625, "learning_rate": 1.634166960634053e-06, "loss": 0.1406, "step": 45461 }, { "epoch": 33.30549450549451, "grad_norm": 0.578125, "learning_rate": 1.6336203741341627e-06, "loss": 0.1082, "step": 45462 }, { "epoch": 33.306227106227105, "grad_norm": 0.78125, "learning_rate": 1.6330740217691383e-06, "loss": 0.1229, "step": 45463 }, { "epoch": 33.30695970695971, "grad_norm": 0.60546875, "learning_rate": 1.6325279035415953e-06, "loss": 0.1166, "step": 45464 }, { "epoch": 33.30769230769231, "grad_norm": 0.96875, "learning_rate": 1.6319820194541548e-06, "loss": 0.12, "step": 45465 }, { "epoch": 33.30842490842491, "grad_norm": 0.859375, "learning_rate": 1.6314363695094387e-06, "loss": 0.114, "step": 45466 }, { "epoch": 33.30915750915751, "grad_norm": 1.28125, "learning_rate": 1.6308909537100462e-06, "loss": 0.1283, "step": 45467 }, { "epoch": 33.30989010989011, "grad_norm": 1.265625, "learning_rate": 1.6303457720585987e-06, "loss": 0.1288, "step": 45468 }, { "epoch": 33.31062271062271, "grad_norm": 1.6015625, "learning_rate": 1.629800824557707e-06, "loss": 0.1203, "step": 45469 }, { "epoch": 33.311355311355314, "grad_norm": 1.1171875, "learning_rate": 1.6292561112099862e-06, "loss": 0.1447, "step": 45470 }, { "epoch": 33.31208791208791, "grad_norm": 1.265625, "learning_rate": 1.6287116320180363e-06, "loss": 0.1232, "step": 45471 }, { "epoch": 33.312820512820515, "grad_norm": 0.56640625, "learning_rate": 1.6281673869844784e-06, "loss": 0.0997, "step": 45472 }, { "epoch": 33.31355311355311, "grad_norm": 1.234375, "learning_rate": 1.6276233761119068e-06, "loss": 0.0997, "step": 45473 }, { "epoch": 33.31428571428572, "grad_norm": 1.734375, "learning_rate": 1.6270795994029427e-06, "loss": 0.1592, "step": 45474 }, { "epoch": 33.315018315018314, "grad_norm": 1.6328125, "learning_rate": 1.6265360568601801e-06, "loss": 0.1433, "step": 45475 }, { "epoch": 33.31575091575092, "grad_norm": 1.4296875, "learning_rate": 1.6259927484862345e-06, "loss": 0.1192, "step": 45476 }, { "epoch": 33.316483516483515, "grad_norm": 1.328125, "learning_rate": 1.6254496742836952e-06, "loss": 0.1222, "step": 45477 }, { "epoch": 33.31721611721612, "grad_norm": 0.90625, "learning_rate": 1.624906834255172e-06, "loss": 0.1073, "step": 45478 }, { "epoch": 33.31794871794872, "grad_norm": 0.58984375, "learning_rate": 1.624364228403264e-06, "loss": 0.1052, "step": 45479 }, { "epoch": 33.31868131868132, "grad_norm": 1.0234375, "learning_rate": 1.6238218567305713e-06, "loss": 0.1103, "step": 45480 }, { "epoch": 33.31941391941392, "grad_norm": 1.59375, "learning_rate": 1.623279719239693e-06, "loss": 0.1276, "step": 45481 }, { "epoch": 33.32014652014652, "grad_norm": 0.91015625, "learning_rate": 1.6227378159332285e-06, "loss": 0.1198, "step": 45482 }, { "epoch": 33.32087912087912, "grad_norm": 0.8046875, "learning_rate": 1.6221961468137773e-06, "loss": 0.105, "step": 45483 }, { "epoch": 33.321611721611724, "grad_norm": 0.76171875, "learning_rate": 1.6216547118839166e-06, "loss": 0.0903, "step": 45484 }, { "epoch": 33.32234432234432, "grad_norm": 0.9296875, "learning_rate": 1.6211135111462684e-06, "loss": 0.1137, "step": 45485 }, { "epoch": 33.323076923076925, "grad_norm": 1.359375, "learning_rate": 1.6205725446034042e-06, "loss": 0.1498, "step": 45486 }, { "epoch": 33.32380952380952, "grad_norm": 0.96484375, "learning_rate": 1.6200318122579234e-06, "loss": 0.1109, "step": 45487 }, { "epoch": 33.32454212454213, "grad_norm": 1.0546875, "learning_rate": 1.6194913141124146e-06, "loss": 0.1375, "step": 45488 }, { "epoch": 33.325274725274724, "grad_norm": 1.265625, "learning_rate": 1.6189510501694716e-06, "loss": 0.1573, "step": 45489 }, { "epoch": 33.32600732600733, "grad_norm": 0.97265625, "learning_rate": 1.6184110204316827e-06, "loss": 0.1317, "step": 45490 }, { "epoch": 33.326739926739926, "grad_norm": 0.71875, "learning_rate": 1.617871224901631e-06, "loss": 0.0948, "step": 45491 }, { "epoch": 33.32747252747253, "grad_norm": 1.0703125, "learning_rate": 1.617331663581905e-06, "loss": 0.1217, "step": 45492 }, { "epoch": 33.32820512820513, "grad_norm": 1.359375, "learning_rate": 1.6167923364750932e-06, "loss": 0.1359, "step": 45493 }, { "epoch": 33.32893772893773, "grad_norm": 1.0234375, "learning_rate": 1.6162532435837728e-06, "loss": 0.1202, "step": 45494 }, { "epoch": 33.32967032967033, "grad_norm": 0.90234375, "learning_rate": 1.615714384910527e-06, "loss": 0.1562, "step": 45495 }, { "epoch": 33.33040293040293, "grad_norm": 0.77734375, "learning_rate": 1.6151757604579494e-06, "loss": 0.1165, "step": 45496 }, { "epoch": 33.33113553113553, "grad_norm": 0.94921875, "learning_rate": 1.6146373702286066e-06, "loss": 0.1089, "step": 45497 }, { "epoch": 33.331868131868134, "grad_norm": 0.8125, "learning_rate": 1.6140992142250927e-06, "loss": 0.1294, "step": 45498 }, { "epoch": 33.33260073260073, "grad_norm": 1.1015625, "learning_rate": 1.6135612924499683e-06, "loss": 0.1053, "step": 45499 }, { "epoch": 33.333333333333336, "grad_norm": 1.1484375, "learning_rate": 1.6130236049058218e-06, "loss": 0.1551, "step": 45500 }, { "epoch": 33.33406593406593, "grad_norm": 1.5078125, "learning_rate": 1.6124861515952313e-06, "loss": 0.1422, "step": 45501 }, { "epoch": 33.33479853479854, "grad_norm": 0.86328125, "learning_rate": 1.6119489325207682e-06, "loss": 0.1118, "step": 45502 }, { "epoch": 33.335531135531134, "grad_norm": 0.65625, "learning_rate": 1.6114119476850042e-06, "loss": 0.0803, "step": 45503 }, { "epoch": 33.33626373626374, "grad_norm": 1.3671875, "learning_rate": 1.6108751970905174e-06, "loss": 0.1228, "step": 45504 }, { "epoch": 33.336996336996336, "grad_norm": 1.3125, "learning_rate": 1.610338680739874e-06, "loss": 0.1314, "step": 45505 }, { "epoch": 33.33772893772894, "grad_norm": 1.1953125, "learning_rate": 1.6098023986356458e-06, "loss": 0.1339, "step": 45506 }, { "epoch": 33.33846153846154, "grad_norm": 0.95703125, "learning_rate": 1.60926635078041e-06, "loss": 0.1129, "step": 45507 }, { "epoch": 33.33919413919414, "grad_norm": 1.375, "learning_rate": 1.6087305371767225e-06, "loss": 0.1201, "step": 45508 }, { "epoch": 33.33992673992674, "grad_norm": 0.73046875, "learning_rate": 1.608194957827161e-06, "loss": 0.1014, "step": 45509 }, { "epoch": 33.34065934065934, "grad_norm": 0.95703125, "learning_rate": 1.6076596127342854e-06, "loss": 0.1192, "step": 45510 }, { "epoch": 33.34139194139194, "grad_norm": 1.421875, "learning_rate": 1.6071245019006631e-06, "loss": 0.1403, "step": 45511 }, { "epoch": 33.342124542124544, "grad_norm": 1.328125, "learning_rate": 1.6065896253288603e-06, "loss": 0.1252, "step": 45512 }, { "epoch": 33.34285714285714, "grad_norm": 0.6171875, "learning_rate": 1.6060549830214378e-06, "loss": 0.1055, "step": 45513 }, { "epoch": 33.343589743589746, "grad_norm": 1.1171875, "learning_rate": 1.6055205749809514e-06, "loss": 0.1226, "step": 45514 }, { "epoch": 33.34432234432234, "grad_norm": 1.0703125, "learning_rate": 1.6049864012099725e-06, "loss": 0.1341, "step": 45515 }, { "epoch": 33.34505494505495, "grad_norm": 0.56640625, "learning_rate": 1.6044524617110514e-06, "loss": 0.0956, "step": 45516 }, { "epoch": 33.345787545787545, "grad_norm": 1.0625, "learning_rate": 1.603918756486749e-06, "loss": 0.1272, "step": 45517 }, { "epoch": 33.34652014652015, "grad_norm": 0.83984375, "learning_rate": 1.6033852855396263e-06, "loss": 0.0996, "step": 45518 }, { "epoch": 33.347252747252746, "grad_norm": 1.359375, "learning_rate": 1.602852048872233e-06, "loss": 0.1108, "step": 45519 }, { "epoch": 33.34798534798535, "grad_norm": 1.046875, "learning_rate": 1.602319046487131e-06, "loss": 0.0984, "step": 45520 }, { "epoch": 33.34871794871795, "grad_norm": 0.86328125, "learning_rate": 1.6017862783868686e-06, "loss": 0.1245, "step": 45521 }, { "epoch": 33.34945054945055, "grad_norm": 0.98046875, "learning_rate": 1.6012537445740028e-06, "loss": 0.1229, "step": 45522 }, { "epoch": 33.35018315018315, "grad_norm": 0.921875, "learning_rate": 1.600721445051083e-06, "loss": 0.1333, "step": 45523 }, { "epoch": 33.35091575091575, "grad_norm": 1.0234375, "learning_rate": 1.6001893798206587e-06, "loss": 0.132, "step": 45524 }, { "epoch": 33.35164835164835, "grad_norm": 1.09375, "learning_rate": 1.599657548885275e-06, "loss": 0.1529, "step": 45525 }, { "epoch": 33.352380952380955, "grad_norm": 0.9453125, "learning_rate": 1.5991259522474871e-06, "loss": 0.1134, "step": 45526 }, { "epoch": 33.35311355311355, "grad_norm": 1.1953125, "learning_rate": 1.5985945899098392e-06, "loss": 0.1232, "step": 45527 }, { "epoch": 33.353846153846156, "grad_norm": 1.109375, "learning_rate": 1.5980634618748873e-06, "loss": 0.14, "step": 45528 }, { "epoch": 33.35457875457875, "grad_norm": 0.9765625, "learning_rate": 1.5975325681451536e-06, "loss": 0.1408, "step": 45529 }, { "epoch": 33.35531135531136, "grad_norm": 1.21875, "learning_rate": 1.5970019087232046e-06, "loss": 0.1322, "step": 45530 }, { "epoch": 33.356043956043955, "grad_norm": 1.1640625, "learning_rate": 1.5964714836115683e-06, "loss": 0.0905, "step": 45531 }, { "epoch": 33.35677655677656, "grad_norm": 1.015625, "learning_rate": 1.595941292812789e-06, "loss": 0.1195, "step": 45532 }, { "epoch": 33.357509157509156, "grad_norm": 1.4921875, "learning_rate": 1.5954113363294169e-06, "loss": 0.1234, "step": 45533 }, { "epoch": 33.35824175824176, "grad_norm": 0.70703125, "learning_rate": 1.594881614163974e-06, "loss": 0.1044, "step": 45534 }, { "epoch": 33.35897435897436, "grad_norm": 0.80078125, "learning_rate": 1.5943521263190167e-06, "loss": 0.12, "step": 45535 }, { "epoch": 33.35970695970696, "grad_norm": 0.796875, "learning_rate": 1.5938228727970665e-06, "loss": 0.1011, "step": 45536 }, { "epoch": 33.36043956043956, "grad_norm": 1.21875, "learning_rate": 1.5932938536006684e-06, "loss": 0.1289, "step": 45537 }, { "epoch": 33.36117216117216, "grad_norm": 1.390625, "learning_rate": 1.5927650687323554e-06, "loss": 0.1496, "step": 45538 }, { "epoch": 33.36190476190476, "grad_norm": 1.0546875, "learning_rate": 1.5922365181946618e-06, "loss": 0.1217, "step": 45539 }, { "epoch": 33.362637362637365, "grad_norm": 1.1796875, "learning_rate": 1.591708201990115e-06, "loss": 0.1491, "step": 45540 }, { "epoch": 33.36336996336996, "grad_norm": 0.89453125, "learning_rate": 1.5911801201212485e-06, "loss": 0.1067, "step": 45541 }, { "epoch": 33.364102564102566, "grad_norm": 1.53125, "learning_rate": 1.5906522725905958e-06, "loss": 0.1321, "step": 45542 }, { "epoch": 33.36483516483516, "grad_norm": 1.171875, "learning_rate": 1.5901246594006794e-06, "loss": 0.1203, "step": 45543 }, { "epoch": 33.36556776556777, "grad_norm": 0.96484375, "learning_rate": 1.5895972805540384e-06, "loss": 0.1049, "step": 45544 }, { "epoch": 33.366300366300365, "grad_norm": 1.140625, "learning_rate": 1.5890701360531893e-06, "loss": 0.1149, "step": 45545 }, { "epoch": 33.36703296703297, "grad_norm": 0.8515625, "learning_rate": 1.5885432259006661e-06, "loss": 0.1205, "step": 45546 }, { "epoch": 33.36776556776557, "grad_norm": 0.984375, "learning_rate": 1.588016550098986e-06, "loss": 0.101, "step": 45547 }, { "epoch": 33.36849816849817, "grad_norm": 1.328125, "learning_rate": 1.587490108650676e-06, "loss": 0.1012, "step": 45548 }, { "epoch": 33.36923076923077, "grad_norm": 1.609375, "learning_rate": 1.586963901558254e-06, "loss": 0.1195, "step": 45549 }, { "epoch": 33.36996336996337, "grad_norm": 0.81640625, "learning_rate": 1.5864379288242532e-06, "loss": 0.0865, "step": 45550 }, { "epoch": 33.37069597069597, "grad_norm": 0.84765625, "learning_rate": 1.5859121904511745e-06, "loss": 0.109, "step": 45551 }, { "epoch": 33.371428571428574, "grad_norm": 0.96875, "learning_rate": 1.5853866864415558e-06, "loss": 0.1181, "step": 45552 }, { "epoch": 33.37216117216117, "grad_norm": 1.5, "learning_rate": 1.5848614167979038e-06, "loss": 0.1466, "step": 45553 }, { "epoch": 33.372893772893775, "grad_norm": 0.86328125, "learning_rate": 1.5843363815227413e-06, "loss": 0.1181, "step": 45554 }, { "epoch": 33.37362637362637, "grad_norm": 0.78125, "learning_rate": 1.5838115806185848e-06, "loss": 0.1372, "step": 45555 }, { "epoch": 33.37435897435898, "grad_norm": 0.859375, "learning_rate": 1.5832870140879346e-06, "loss": 0.1328, "step": 45556 }, { "epoch": 33.375091575091574, "grad_norm": 1.328125, "learning_rate": 1.5827626819333189e-06, "loss": 0.1286, "step": 45557 }, { "epoch": 33.37582417582418, "grad_norm": 0.859375, "learning_rate": 1.5822385841572498e-06, "loss": 0.1196, "step": 45558 }, { "epoch": 33.376556776556775, "grad_norm": 0.921875, "learning_rate": 1.5817147207622327e-06, "loss": 0.1048, "step": 45559 }, { "epoch": 33.37728937728938, "grad_norm": 0.82421875, "learning_rate": 1.5811910917507788e-06, "loss": 0.1015, "step": 45560 }, { "epoch": 33.37802197802198, "grad_norm": 1.0546875, "learning_rate": 1.5806676971254e-06, "loss": 0.1033, "step": 45561 }, { "epoch": 33.37875457875458, "grad_norm": 1.3046875, "learning_rate": 1.5801445368885912e-06, "loss": 0.0987, "step": 45562 }, { "epoch": 33.37948717948718, "grad_norm": 1.421875, "learning_rate": 1.5796216110428809e-06, "loss": 0.1342, "step": 45563 }, { "epoch": 33.38021978021978, "grad_norm": 1.3203125, "learning_rate": 1.5790989195907577e-06, "loss": 0.1265, "step": 45564 }, { "epoch": 33.38095238095238, "grad_norm": 1.109375, "learning_rate": 1.5785764625347338e-06, "loss": 0.1203, "step": 45565 }, { "epoch": 33.381684981684984, "grad_norm": 0.953125, "learning_rate": 1.5780542398773037e-06, "loss": 0.0957, "step": 45566 }, { "epoch": 33.38241758241758, "grad_norm": 0.96875, "learning_rate": 1.5775322516209793e-06, "loss": 0.1259, "step": 45567 }, { "epoch": 33.383150183150185, "grad_norm": 0.68359375, "learning_rate": 1.5770104977682605e-06, "loss": 0.102, "step": 45568 }, { "epoch": 33.38388278388278, "grad_norm": 0.94140625, "learning_rate": 1.5764889783216378e-06, "loss": 0.136, "step": 45569 }, { "epoch": 33.38461538461539, "grad_norm": 0.97265625, "learning_rate": 1.5759676932836275e-06, "loss": 0.1008, "step": 45570 }, { "epoch": 33.385347985347984, "grad_norm": 1.2109375, "learning_rate": 1.575446642656708e-06, "loss": 0.1391, "step": 45571 }, { "epoch": 33.38608058608059, "grad_norm": 1.09375, "learning_rate": 1.574925826443391e-06, "loss": 0.108, "step": 45572 }, { "epoch": 33.386813186813185, "grad_norm": 1.2578125, "learning_rate": 1.5744052446461607e-06, "loss": 0.1291, "step": 45573 }, { "epoch": 33.38754578754579, "grad_norm": 0.9453125, "learning_rate": 1.5738848972675176e-06, "loss": 0.1186, "step": 45574 }, { "epoch": 33.38827838827839, "grad_norm": 1.015625, "learning_rate": 1.5733647843099507e-06, "loss": 0.1139, "step": 45575 }, { "epoch": 33.38901098901099, "grad_norm": 1.4609375, "learning_rate": 1.5728449057759613e-06, "loss": 0.1296, "step": 45576 }, { "epoch": 33.38974358974359, "grad_norm": 0.9765625, "learning_rate": 1.572325261668027e-06, "loss": 0.1152, "step": 45577 }, { "epoch": 33.39047619047619, "grad_norm": 0.74609375, "learning_rate": 1.5718058519886489e-06, "loss": 0.1014, "step": 45578 }, { "epoch": 33.39120879120879, "grad_norm": 0.90625, "learning_rate": 1.5712866767403058e-06, "loss": 0.1235, "step": 45579 }, { "epoch": 33.391941391941394, "grad_norm": 0.50390625, "learning_rate": 1.5707677359254974e-06, "loss": 0.1073, "step": 45580 }, { "epoch": 33.39267399267399, "grad_norm": 1.078125, "learning_rate": 1.5702490295467027e-06, "loss": 0.1267, "step": 45581 }, { "epoch": 33.393406593406596, "grad_norm": 0.859375, "learning_rate": 1.5697305576063998e-06, "loss": 0.113, "step": 45582 }, { "epoch": 33.39413919413919, "grad_norm": 0.75, "learning_rate": 1.5692123201070844e-06, "loss": 0.1234, "step": 45583 }, { "epoch": 33.3948717948718, "grad_norm": 1.25, "learning_rate": 1.5686943170512342e-06, "loss": 0.1457, "step": 45584 }, { "epoch": 33.395604395604394, "grad_norm": 1.25, "learning_rate": 1.5681765484413399e-06, "loss": 0.1182, "step": 45585 }, { "epoch": 33.396336996337, "grad_norm": 0.9140625, "learning_rate": 1.5676590142798678e-06, "loss": 0.1206, "step": 45586 }, { "epoch": 33.397069597069596, "grad_norm": 1.1328125, "learning_rate": 1.567141714569308e-06, "loss": 0.1066, "step": 45587 }, { "epoch": 33.3978021978022, "grad_norm": 0.8125, "learning_rate": 1.566624649312133e-06, "loss": 0.1086, "step": 45588 }, { "epoch": 33.3985347985348, "grad_norm": 0.89453125, "learning_rate": 1.566107818510828e-06, "loss": 0.1141, "step": 45589 }, { "epoch": 33.3992673992674, "grad_norm": 0.7265625, "learning_rate": 1.565591222167859e-06, "loss": 0.0978, "step": 45590 }, { "epoch": 33.4, "grad_norm": 1.8046875, "learning_rate": 1.5650748602857111e-06, "loss": 0.1372, "step": 45591 }, { "epoch": 33.4007326007326, "grad_norm": 1.265625, "learning_rate": 1.5645587328668516e-06, "loss": 0.1216, "step": 45592 }, { "epoch": 33.4014652014652, "grad_norm": 1.078125, "learning_rate": 1.5640428399137531e-06, "loss": 0.1225, "step": 45593 }, { "epoch": 33.402197802197804, "grad_norm": 1.484375, "learning_rate": 1.5635271814288949e-06, "loss": 0.1196, "step": 45594 }, { "epoch": 33.4029304029304, "grad_norm": 0.89453125, "learning_rate": 1.5630117574147381e-06, "loss": 0.1126, "step": 45595 }, { "epoch": 33.403663003663006, "grad_norm": 0.81640625, "learning_rate": 1.5624965678737617e-06, "loss": 0.1038, "step": 45596 }, { "epoch": 33.4043956043956, "grad_norm": 0.92578125, "learning_rate": 1.5619816128084278e-06, "loss": 0.1088, "step": 45597 }, { "epoch": 33.40512820512821, "grad_norm": 1.171875, "learning_rate": 1.5614668922212039e-06, "loss": 0.1432, "step": 45598 }, { "epoch": 33.405860805860804, "grad_norm": 0.78515625, "learning_rate": 1.5609524061145573e-06, "loss": 0.1199, "step": 45599 }, { "epoch": 33.40659340659341, "grad_norm": 1.1015625, "learning_rate": 1.5604381544909558e-06, "loss": 0.1163, "step": 45600 }, { "epoch": 33.407326007326006, "grad_norm": 1.2265625, "learning_rate": 1.559924137352856e-06, "loss": 0.1289, "step": 45601 }, { "epoch": 33.40805860805861, "grad_norm": 1.6015625, "learning_rate": 1.5594103547027304e-06, "loss": 0.1253, "step": 45602 }, { "epoch": 33.40879120879121, "grad_norm": 1.1796875, "learning_rate": 1.5588968065430304e-06, "loss": 0.1441, "step": 45603 }, { "epoch": 33.40952380952381, "grad_norm": 0.92578125, "learning_rate": 1.5583834928762291e-06, "loss": 0.103, "step": 45604 }, { "epoch": 33.41025641025641, "grad_norm": 0.8984375, "learning_rate": 1.5578704137047774e-06, "loss": 0.1251, "step": 45605 }, { "epoch": 33.41098901098901, "grad_norm": 0.98828125, "learning_rate": 1.557357569031132e-06, "loss": 0.1296, "step": 45606 }, { "epoch": 33.41172161172161, "grad_norm": 1.2734375, "learning_rate": 1.5568449588577603e-06, "loss": 0.1085, "step": 45607 }, { "epoch": 33.412454212454215, "grad_norm": 1.3125, "learning_rate": 1.5563325831871021e-06, "loss": 0.1164, "step": 45608 }, { "epoch": 33.41318681318681, "grad_norm": 1.2890625, "learning_rate": 1.5558204420216254e-06, "loss": 0.1145, "step": 45609 }, { "epoch": 33.413919413919416, "grad_norm": 0.96875, "learning_rate": 1.5553085353637809e-06, "loss": 0.1072, "step": 45610 }, { "epoch": 33.41465201465201, "grad_norm": 0.85546875, "learning_rate": 1.55479686321602e-06, "loss": 0.1068, "step": 45611 }, { "epoch": 33.41538461538462, "grad_norm": 0.94921875, "learning_rate": 1.5542854255807987e-06, "loss": 0.1365, "step": 45612 }, { "epoch": 33.416117216117215, "grad_norm": 0.94921875, "learning_rate": 1.5537742224605629e-06, "loss": 0.1113, "step": 45613 }, { "epoch": 33.41684981684982, "grad_norm": 1.109375, "learning_rate": 1.5532632538577524e-06, "loss": 0.1319, "step": 45614 }, { "epoch": 33.417582417582416, "grad_norm": 0.796875, "learning_rate": 1.5527525197748353e-06, "loss": 0.0996, "step": 45615 }, { "epoch": 33.41831501831502, "grad_norm": 1.09375, "learning_rate": 1.552242020214251e-06, "loss": 0.1145, "step": 45616 }, { "epoch": 33.41904761904762, "grad_norm": 0.859375, "learning_rate": 1.55173175517844e-06, "loss": 0.1213, "step": 45617 }, { "epoch": 33.41978021978022, "grad_norm": 0.68359375, "learning_rate": 1.5512217246698528e-06, "loss": 0.1063, "step": 45618 }, { "epoch": 33.42051282051282, "grad_norm": 0.83203125, "learning_rate": 1.5507119286909247e-06, "loss": 0.1235, "step": 45619 }, { "epoch": 33.42124542124542, "grad_norm": 0.78125, "learning_rate": 1.5502023672441115e-06, "loss": 0.1009, "step": 45620 }, { "epoch": 33.42197802197802, "grad_norm": 0.921875, "learning_rate": 1.5496930403318426e-06, "loss": 0.1356, "step": 45621 }, { "epoch": 33.422710622710625, "grad_norm": 0.75, "learning_rate": 1.5491839479565694e-06, "loss": 0.1232, "step": 45622 }, { "epoch": 33.42344322344322, "grad_norm": 0.640625, "learning_rate": 1.5486750901207256e-06, "loss": 0.0833, "step": 45623 }, { "epoch": 33.424175824175826, "grad_norm": 0.9296875, "learning_rate": 1.5481664668267465e-06, "loss": 0.1038, "step": 45624 }, { "epoch": 33.42490842490842, "grad_norm": 0.953125, "learning_rate": 1.5476580780770717e-06, "loss": 0.1134, "step": 45625 }, { "epoch": 33.42564102564103, "grad_norm": 0.67578125, "learning_rate": 1.547149923874136e-06, "loss": 0.1075, "step": 45626 }, { "epoch": 33.426373626373625, "grad_norm": 0.796875, "learning_rate": 1.546642004220374e-06, "loss": 0.12, "step": 45627 }, { "epoch": 33.42710622710623, "grad_norm": 1.40625, "learning_rate": 1.5461343191182257e-06, "loss": 0.1275, "step": 45628 }, { "epoch": 33.427838827838826, "grad_norm": 1.2890625, "learning_rate": 1.5456268685701145e-06, "loss": 0.096, "step": 45629 }, { "epoch": 33.42857142857143, "grad_norm": 1.203125, "learning_rate": 1.5451196525784805e-06, "loss": 0.1214, "step": 45630 }, { "epoch": 33.42930402930403, "grad_norm": 1.078125, "learning_rate": 1.5446126711457417e-06, "loss": 0.1465, "step": 45631 }, { "epoch": 33.43003663003663, "grad_norm": 0.93359375, "learning_rate": 1.5441059242743384e-06, "loss": 0.1008, "step": 45632 }, { "epoch": 33.43076923076923, "grad_norm": 0.51171875, "learning_rate": 1.5435994119666937e-06, "loss": 0.0943, "step": 45633 }, { "epoch": 33.43150183150183, "grad_norm": 1.2109375, "learning_rate": 1.5430931342252375e-06, "loss": 0.1333, "step": 45634 }, { "epoch": 33.43223443223443, "grad_norm": 0.72265625, "learning_rate": 1.5425870910523925e-06, "loss": 0.1099, "step": 45635 }, { "epoch": 33.432967032967035, "grad_norm": 0.890625, "learning_rate": 1.5420812824505829e-06, "loss": 0.119, "step": 45636 }, { "epoch": 33.43369963369963, "grad_norm": 1.5078125, "learning_rate": 1.5415757084222372e-06, "loss": 0.1303, "step": 45637 }, { "epoch": 33.434432234432236, "grad_norm": 0.95703125, "learning_rate": 1.5410703689697738e-06, "loss": 0.1192, "step": 45638 }, { "epoch": 33.435164835164834, "grad_norm": 1.125, "learning_rate": 1.540565264095616e-06, "loss": 0.1198, "step": 45639 }, { "epoch": 33.43589743589744, "grad_norm": 0.8671875, "learning_rate": 1.5400603938021712e-06, "loss": 0.1298, "step": 45640 }, { "epoch": 33.436630036630035, "grad_norm": 0.81640625, "learning_rate": 1.5395557580918794e-06, "loss": 0.125, "step": 45641 }, { "epoch": 33.43736263736264, "grad_norm": 1.40625, "learning_rate": 1.5390513569671473e-06, "loss": 0.1282, "step": 45642 }, { "epoch": 33.43809523809524, "grad_norm": 1.140625, "learning_rate": 1.5385471904303934e-06, "loss": 0.1289, "step": 45643 }, { "epoch": 33.43882783882784, "grad_norm": 1.1875, "learning_rate": 1.5380432584840356e-06, "loss": 0.111, "step": 45644 }, { "epoch": 33.43956043956044, "grad_norm": 1.0625, "learning_rate": 1.537539561130481e-06, "loss": 0.0927, "step": 45645 }, { "epoch": 33.44029304029304, "grad_norm": 0.68359375, "learning_rate": 1.5370360983721477e-06, "loss": 0.0985, "step": 45646 }, { "epoch": 33.44102564102564, "grad_norm": 1.546875, "learning_rate": 1.5365328702114483e-06, "loss": 0.1227, "step": 45647 }, { "epoch": 33.441758241758244, "grad_norm": 0.70703125, "learning_rate": 1.536029876650795e-06, "loss": 0.1037, "step": 45648 }, { "epoch": 33.44249084249084, "grad_norm": 0.7734375, "learning_rate": 1.5355271176925953e-06, "loss": 0.1265, "step": 45649 }, { "epoch": 33.443223443223445, "grad_norm": 0.921875, "learning_rate": 1.5350245933392563e-06, "loss": 0.1288, "step": 45650 }, { "epoch": 33.44395604395604, "grad_norm": 1.0625, "learning_rate": 1.5345223035931903e-06, "loss": 0.132, "step": 45651 }, { "epoch": 33.44468864468865, "grad_norm": 0.92578125, "learning_rate": 1.53402024845681e-06, "loss": 0.1288, "step": 45652 }, { "epoch": 33.445421245421244, "grad_norm": 1.109375, "learning_rate": 1.5335184279325058e-06, "loss": 0.1185, "step": 45653 }, { "epoch": 33.44615384615385, "grad_norm": 0.7890625, "learning_rate": 1.5330168420226904e-06, "loss": 0.1084, "step": 45654 }, { "epoch": 33.446886446886445, "grad_norm": 0.9296875, "learning_rate": 1.5325154907297657e-06, "loss": 0.1271, "step": 45655 }, { "epoch": 33.44761904761905, "grad_norm": 0.74609375, "learning_rate": 1.5320143740561383e-06, "loss": 0.1244, "step": 45656 }, { "epoch": 33.44835164835165, "grad_norm": 1.078125, "learning_rate": 1.531513492004199e-06, "loss": 0.1209, "step": 45657 }, { "epoch": 33.44908424908425, "grad_norm": 1.625, "learning_rate": 1.531012844576355e-06, "loss": 0.1226, "step": 45658 }, { "epoch": 33.44981684981685, "grad_norm": 0.84765625, "learning_rate": 1.5305124317750075e-06, "loss": 0.1018, "step": 45659 }, { "epoch": 33.45054945054945, "grad_norm": 1.1484375, "learning_rate": 1.5300122536025473e-06, "loss": 0.1215, "step": 45660 }, { "epoch": 33.45128205128205, "grad_norm": 1.40625, "learning_rate": 1.5295123100613761e-06, "loss": 0.1044, "step": 45661 }, { "epoch": 33.452014652014654, "grad_norm": 1.1015625, "learning_rate": 1.5290126011538847e-06, "loss": 0.1073, "step": 45662 }, { "epoch": 33.45274725274725, "grad_norm": 1.078125, "learning_rate": 1.5285131268824741e-06, "loss": 0.132, "step": 45663 }, { "epoch": 33.453479853479855, "grad_norm": 0.859375, "learning_rate": 1.5280138872495297e-06, "loss": 0.1288, "step": 45664 }, { "epoch": 33.45421245421245, "grad_norm": 1.421875, "learning_rate": 1.5275148822574529e-06, "loss": 0.1515, "step": 45665 }, { "epoch": 33.45494505494506, "grad_norm": 1.0703125, "learning_rate": 1.5270161119086235e-06, "loss": 0.1284, "step": 45666 }, { "epoch": 33.455677655677654, "grad_norm": 1.2578125, "learning_rate": 1.5265175762054374e-06, "loss": 0.1226, "step": 45667 }, { "epoch": 33.45641025641026, "grad_norm": 1.3984375, "learning_rate": 1.5260192751502857e-06, "loss": 0.1407, "step": 45668 }, { "epoch": 33.457142857142856, "grad_norm": 1.0546875, "learning_rate": 1.525521208745547e-06, "loss": 0.1228, "step": 45669 }, { "epoch": 33.45787545787546, "grad_norm": 1.1953125, "learning_rate": 1.5250233769936182e-06, "loss": 0.1382, "step": 45670 }, { "epoch": 33.45860805860806, "grad_norm": 1.0234375, "learning_rate": 1.524525779896879e-06, "loss": 0.1036, "step": 45671 }, { "epoch": 33.45934065934066, "grad_norm": 0.99609375, "learning_rate": 1.524028417457714e-06, "loss": 0.1114, "step": 45672 }, { "epoch": 33.46007326007326, "grad_norm": 1.5546875, "learning_rate": 1.523531289678503e-06, "loss": 0.1039, "step": 45673 }, { "epoch": 33.46080586080586, "grad_norm": 0.9296875, "learning_rate": 1.5230343965616418e-06, "loss": 0.1366, "step": 45674 }, { "epoch": 33.46153846153846, "grad_norm": 1.0078125, "learning_rate": 1.5225377381094885e-06, "loss": 0.1158, "step": 45675 }, { "epoch": 33.462271062271064, "grad_norm": 1.0703125, "learning_rate": 1.5220413143244443e-06, "loss": 0.1149, "step": 45676 }, { "epoch": 33.46300366300366, "grad_norm": 0.88671875, "learning_rate": 1.5215451252088727e-06, "loss": 0.0831, "step": 45677 }, { "epoch": 33.463736263736266, "grad_norm": 1.234375, "learning_rate": 1.5210491707651638e-06, "loss": 0.135, "step": 45678 }, { "epoch": 33.46446886446886, "grad_norm": 0.97265625, "learning_rate": 1.5205534509956809e-06, "loss": 0.1546, "step": 45679 }, { "epoch": 33.46520146520147, "grad_norm": 0.921875, "learning_rate": 1.5200579659028096e-06, "loss": 0.1292, "step": 45680 }, { "epoch": 33.465934065934064, "grad_norm": 1.0859375, "learning_rate": 1.519562715488912e-06, "loss": 0.1358, "step": 45681 }, { "epoch": 33.46666666666667, "grad_norm": 0.9453125, "learning_rate": 1.5190676997563742e-06, "loss": 0.125, "step": 45682 }, { "epoch": 33.467399267399266, "grad_norm": 0.86328125, "learning_rate": 1.518572918707564e-06, "loss": 0.0992, "step": 45683 }, { "epoch": 33.46813186813187, "grad_norm": 1.3046875, "learning_rate": 1.5180783723448451e-06, "loss": 0.1222, "step": 45684 }, { "epoch": 33.46886446886447, "grad_norm": 0.953125, "learning_rate": 1.5175840606705967e-06, "loss": 0.1184, "step": 45685 }, { "epoch": 33.46959706959707, "grad_norm": 0.9453125, "learning_rate": 1.517089983687182e-06, "loss": 0.1351, "step": 45686 }, { "epoch": 33.47032967032967, "grad_norm": 0.828125, "learning_rate": 1.5165961413969697e-06, "loss": 0.1019, "step": 45687 }, { "epoch": 33.47106227106227, "grad_norm": 0.9375, "learning_rate": 1.5161025338023231e-06, "loss": 0.1196, "step": 45688 }, { "epoch": 33.47179487179487, "grad_norm": 0.859375, "learning_rate": 1.515609160905616e-06, "loss": 0.0989, "step": 45689 }, { "epoch": 33.472527472527474, "grad_norm": 0.96484375, "learning_rate": 1.5151160227092006e-06, "loss": 0.1105, "step": 45690 }, { "epoch": 33.47326007326007, "grad_norm": 1.421875, "learning_rate": 1.5146231192154453e-06, "loss": 0.1345, "step": 45691 }, { "epoch": 33.473992673992676, "grad_norm": 1.0546875, "learning_rate": 1.514130450426708e-06, "loss": 0.1115, "step": 45692 }, { "epoch": 33.47472527472527, "grad_norm": 0.97265625, "learning_rate": 1.5136380163453631e-06, "loss": 0.1308, "step": 45693 }, { "epoch": 33.47545787545788, "grad_norm": 1.4609375, "learning_rate": 1.513145816973751e-06, "loss": 0.104, "step": 45694 }, { "epoch": 33.476190476190474, "grad_norm": 0.80859375, "learning_rate": 1.5126538523142408e-06, "loss": 0.0876, "step": 45695 }, { "epoch": 33.47692307692308, "grad_norm": 0.86328125, "learning_rate": 1.5121621223691898e-06, "loss": 0.108, "step": 45696 }, { "epoch": 33.477655677655676, "grad_norm": 0.84375, "learning_rate": 1.5116706271409447e-06, "loss": 0.1411, "step": 45697 }, { "epoch": 33.47838827838828, "grad_norm": 1.515625, "learning_rate": 1.5111793666318747e-06, "loss": 0.1353, "step": 45698 }, { "epoch": 33.47912087912088, "grad_norm": 1.640625, "learning_rate": 1.5106883408443197e-06, "loss": 0.1325, "step": 45699 }, { "epoch": 33.47985347985348, "grad_norm": 0.72265625, "learning_rate": 1.5101975497806496e-06, "loss": 0.1181, "step": 45700 }, { "epoch": 33.48058608058608, "grad_norm": 0.9765625, "learning_rate": 1.5097069934431992e-06, "loss": 0.1117, "step": 45701 }, { "epoch": 33.48131868131868, "grad_norm": 1.15625, "learning_rate": 1.509216671834321e-06, "loss": 0.1562, "step": 45702 }, { "epoch": 33.48205128205128, "grad_norm": 0.7578125, "learning_rate": 1.5087265849563725e-06, "loss": 0.131, "step": 45703 }, { "epoch": 33.482783882783885, "grad_norm": 0.73828125, "learning_rate": 1.5082367328117003e-06, "loss": 0.1228, "step": 45704 }, { "epoch": 33.48351648351648, "grad_norm": 0.96484375, "learning_rate": 1.5077471154026455e-06, "loss": 0.1112, "step": 45705 }, { "epoch": 33.484249084249086, "grad_norm": 0.9375, "learning_rate": 1.5072577327315603e-06, "loss": 0.1031, "step": 45706 }, { "epoch": 33.48498168498168, "grad_norm": 1.015625, "learning_rate": 1.5067685848007802e-06, "loss": 0.133, "step": 45707 }, { "epoch": 33.48571428571429, "grad_norm": 1.0625, "learning_rate": 1.5062796716126576e-06, "loss": 0.1446, "step": 45708 }, { "epoch": 33.486446886446885, "grad_norm": 0.765625, "learning_rate": 1.5057909931695333e-06, "loss": 0.1128, "step": 45709 }, { "epoch": 33.48717948717949, "grad_norm": 0.94140625, "learning_rate": 1.5053025494737485e-06, "loss": 0.1267, "step": 45710 }, { "epoch": 33.487912087912086, "grad_norm": 0.73046875, "learning_rate": 1.50481434052765e-06, "loss": 0.1055, "step": 45711 }, { "epoch": 33.48864468864469, "grad_norm": 0.76953125, "learning_rate": 1.5043263663335623e-06, "loss": 0.115, "step": 45712 }, { "epoch": 33.48937728937729, "grad_norm": 0.66015625, "learning_rate": 1.5038386268938322e-06, "loss": 0.1112, "step": 45713 }, { "epoch": 33.49010989010989, "grad_norm": 0.9765625, "learning_rate": 1.503351122210795e-06, "loss": 0.1216, "step": 45714 }, { "epoch": 33.49084249084249, "grad_norm": 1.7578125, "learning_rate": 1.5028638522867865e-06, "loss": 0.1306, "step": 45715 }, { "epoch": 33.49157509157509, "grad_norm": 0.69921875, "learning_rate": 1.5023768171241425e-06, "loss": 0.1028, "step": 45716 }, { "epoch": 33.49230769230769, "grad_norm": 1.0703125, "learning_rate": 1.5018900167252037e-06, "loss": 0.1301, "step": 45717 }, { "epoch": 33.493040293040295, "grad_norm": 0.890625, "learning_rate": 1.5014034510922843e-06, "loss": 0.1234, "step": 45718 }, { "epoch": 33.49377289377289, "grad_norm": 0.609375, "learning_rate": 1.5009171202277305e-06, "loss": 0.1184, "step": 45719 }, { "epoch": 33.494505494505496, "grad_norm": 0.6640625, "learning_rate": 1.5004310241338672e-06, "loss": 0.1091, "step": 45720 }, { "epoch": 33.49523809523809, "grad_norm": 0.9375, "learning_rate": 1.4999451628130243e-06, "loss": 0.1236, "step": 45721 }, { "epoch": 33.4959706959707, "grad_norm": 0.93359375, "learning_rate": 1.4994595362675321e-06, "loss": 0.1464, "step": 45722 }, { "epoch": 33.496703296703295, "grad_norm": 1.1640625, "learning_rate": 1.4989741444997098e-06, "loss": 0.125, "step": 45723 }, { "epoch": 33.4974358974359, "grad_norm": 1.015625, "learning_rate": 1.4984889875118928e-06, "loss": 0.1174, "step": 45724 }, { "epoch": 33.498168498168496, "grad_norm": 1.1875, "learning_rate": 1.4980040653064007e-06, "loss": 0.1385, "step": 45725 }, { "epoch": 33.4989010989011, "grad_norm": 0.92578125, "learning_rate": 1.4975193778855574e-06, "loss": 0.1115, "step": 45726 }, { "epoch": 33.4996336996337, "grad_norm": 0.81640625, "learning_rate": 1.497034925251688e-06, "loss": 0.1365, "step": 45727 }, { "epoch": 33.5003663003663, "grad_norm": 0.8125, "learning_rate": 1.4965507074071065e-06, "loss": 0.1426, "step": 45728 }, { "epoch": 33.5010989010989, "grad_norm": 1.0, "learning_rate": 1.4960667243541368e-06, "loss": 0.1143, "step": 45729 }, { "epoch": 33.501831501831504, "grad_norm": 1.015625, "learning_rate": 1.495582976095104e-06, "loss": 0.1302, "step": 45730 }, { "epoch": 33.5025641025641, "grad_norm": 0.890625, "learning_rate": 1.4950994626323163e-06, "loss": 0.1094, "step": 45731 }, { "epoch": 33.503296703296705, "grad_norm": 1.0390625, "learning_rate": 1.4946161839680924e-06, "loss": 0.1363, "step": 45732 }, { "epoch": 33.5040293040293, "grad_norm": 0.765625, "learning_rate": 1.4941331401047517e-06, "loss": 0.1121, "step": 45733 }, { "epoch": 33.50476190476191, "grad_norm": 0.64453125, "learning_rate": 1.4936503310446083e-06, "loss": 0.1115, "step": 45734 }, { "epoch": 33.505494505494504, "grad_norm": 0.73046875, "learning_rate": 1.4931677567899697e-06, "loss": 0.1125, "step": 45735 }, { "epoch": 33.50622710622711, "grad_norm": 0.66796875, "learning_rate": 1.4926854173431551e-06, "loss": 0.1145, "step": 45736 }, { "epoch": 33.506959706959705, "grad_norm": 1.046875, "learning_rate": 1.4922033127064732e-06, "loss": 0.1136, "step": 45737 }, { "epoch": 33.50769230769231, "grad_norm": 1.0078125, "learning_rate": 1.4917214428822317e-06, "loss": 0.1011, "step": 45738 }, { "epoch": 33.50842490842491, "grad_norm": 1.0703125, "learning_rate": 1.4912398078727443e-06, "loss": 0.1426, "step": 45739 }, { "epoch": 33.50915750915751, "grad_norm": 0.73828125, "learning_rate": 1.4907584076803136e-06, "loss": 0.134, "step": 45740 }, { "epoch": 33.50989010989011, "grad_norm": 1.375, "learning_rate": 1.4902772423072478e-06, "loss": 0.1021, "step": 45741 }, { "epoch": 33.51062271062271, "grad_norm": 1.3671875, "learning_rate": 1.4897963117558551e-06, "loss": 0.1544, "step": 45742 }, { "epoch": 33.51135531135531, "grad_norm": 0.80859375, "learning_rate": 1.489315616028433e-06, "loss": 0.1053, "step": 45743 }, { "epoch": 33.512087912087914, "grad_norm": 1.484375, "learning_rate": 1.488835155127289e-06, "loss": 0.1674, "step": 45744 }, { "epoch": 33.51282051282051, "grad_norm": 0.83203125, "learning_rate": 1.4883549290547316e-06, "loss": 0.1074, "step": 45745 }, { "epoch": 33.513553113553115, "grad_norm": 1.4140625, "learning_rate": 1.487874937813047e-06, "loss": 0.0932, "step": 45746 }, { "epoch": 33.51428571428571, "grad_norm": 1.0859375, "learning_rate": 1.4873951814045435e-06, "loss": 0.1463, "step": 45747 }, { "epoch": 33.51501831501832, "grad_norm": 1.140625, "learning_rate": 1.486915659831529e-06, "loss": 0.1328, "step": 45748 }, { "epoch": 33.515750915750914, "grad_norm": 1.2421875, "learning_rate": 1.4864363730962837e-06, "loss": 0.101, "step": 45749 }, { "epoch": 33.51648351648352, "grad_norm": 0.953125, "learning_rate": 1.4859573212011114e-06, "loss": 0.1512, "step": 45750 }, { "epoch": 33.517216117216115, "grad_norm": 1.1484375, "learning_rate": 1.4854785041483084e-06, "loss": 0.1228, "step": 45751 }, { "epoch": 33.51794871794872, "grad_norm": 0.94140625, "learning_rate": 1.4849999219401723e-06, "loss": 0.1139, "step": 45752 }, { "epoch": 33.51868131868132, "grad_norm": 1.140625, "learning_rate": 1.4845215745789888e-06, "loss": 0.1545, "step": 45753 }, { "epoch": 33.51941391941392, "grad_norm": 1.1015625, "learning_rate": 1.4840434620670556e-06, "loss": 0.144, "step": 45754 }, { "epoch": 33.52014652014652, "grad_norm": 1.0390625, "learning_rate": 1.4835655844066584e-06, "loss": 0.1372, "step": 45755 }, { "epoch": 33.52087912087912, "grad_norm": 1.1953125, "learning_rate": 1.4830879416000892e-06, "loss": 0.1238, "step": 45756 }, { "epoch": 33.52161172161172, "grad_norm": 0.984375, "learning_rate": 1.4826105336496397e-06, "loss": 0.1196, "step": 45757 }, { "epoch": 33.522344322344324, "grad_norm": 1.21875, "learning_rate": 1.4821333605575957e-06, "loss": 0.1213, "step": 45758 }, { "epoch": 33.52307692307692, "grad_norm": 1.1484375, "learning_rate": 1.4816564223262382e-06, "loss": 0.1186, "step": 45759 }, { "epoch": 33.523809523809526, "grad_norm": 0.9765625, "learning_rate": 1.4811797189578588e-06, "loss": 0.105, "step": 45760 }, { "epoch": 33.52454212454212, "grad_norm": 0.890625, "learning_rate": 1.4807032504547326e-06, "loss": 0.0992, "step": 45761 }, { "epoch": 33.52527472527473, "grad_norm": 1.0078125, "learning_rate": 1.4802270168191567e-06, "loss": 0.0996, "step": 45762 }, { "epoch": 33.526007326007324, "grad_norm": 1.171875, "learning_rate": 1.4797510180534065e-06, "loss": 0.1439, "step": 45763 }, { "epoch": 33.52673992673993, "grad_norm": 1.3046875, "learning_rate": 1.479275254159757e-06, "loss": 0.124, "step": 45764 }, { "epoch": 33.527472527472526, "grad_norm": 1.328125, "learning_rate": 1.4787997251404945e-06, "loss": 0.1368, "step": 45765 }, { "epoch": 33.52820512820513, "grad_norm": 0.9375, "learning_rate": 1.4783244309978887e-06, "loss": 0.117, "step": 45766 }, { "epoch": 33.52893772893773, "grad_norm": 0.9140625, "learning_rate": 1.477849371734231e-06, "loss": 0.1251, "step": 45767 }, { "epoch": 33.52967032967033, "grad_norm": 0.8359375, "learning_rate": 1.4773745473517913e-06, "loss": 0.1154, "step": 45768 }, { "epoch": 33.53040293040293, "grad_norm": 1.2421875, "learning_rate": 1.4768999578528393e-06, "loss": 0.1585, "step": 45769 }, { "epoch": 33.53113553113553, "grad_norm": 1.140625, "learning_rate": 1.4764256032396504e-06, "loss": 0.1246, "step": 45770 }, { "epoch": 33.53186813186813, "grad_norm": 1.3046875, "learning_rate": 1.4759514835145048e-06, "loss": 0.1128, "step": 45771 }, { "epoch": 33.532600732600734, "grad_norm": 1.2265625, "learning_rate": 1.4754775986796668e-06, "loss": 0.1205, "step": 45772 }, { "epoch": 33.53333333333333, "grad_norm": 0.953125, "learning_rate": 1.4750039487374063e-06, "loss": 0.1078, "step": 45773 }, { "epoch": 33.534065934065936, "grad_norm": 1.53125, "learning_rate": 1.4745305336899985e-06, "loss": 0.1003, "step": 45774 }, { "epoch": 33.53479853479853, "grad_norm": 0.921875, "learning_rate": 1.4740573535397127e-06, "loss": 0.1168, "step": 45775 }, { "epoch": 33.53553113553114, "grad_norm": 0.6796875, "learning_rate": 1.4735844082888079e-06, "loss": 0.1201, "step": 45776 }, { "epoch": 33.536263736263734, "grad_norm": 1.0859375, "learning_rate": 1.4731116979395538e-06, "loss": 0.1125, "step": 45777 }, { "epoch": 33.53699633699634, "grad_norm": 0.83203125, "learning_rate": 1.4726392224942199e-06, "loss": 0.1156, "step": 45778 }, { "epoch": 33.537728937728936, "grad_norm": 1.8125, "learning_rate": 1.4721669819550598e-06, "loss": 0.1459, "step": 45779 }, { "epoch": 33.53846153846154, "grad_norm": 1.15625, "learning_rate": 1.4716949763243482e-06, "loss": 0.1296, "step": 45780 }, { "epoch": 33.53919413919414, "grad_norm": 1.796875, "learning_rate": 1.4712232056043385e-06, "loss": 0.1206, "step": 45781 }, { "epoch": 33.53992673992674, "grad_norm": 0.8359375, "learning_rate": 1.4707516697972948e-06, "loss": 0.1004, "step": 45782 }, { "epoch": 33.54065934065934, "grad_norm": 0.99609375, "learning_rate": 1.470280368905476e-06, "loss": 0.1028, "step": 45783 }, { "epoch": 33.54139194139194, "grad_norm": 1.234375, "learning_rate": 1.4698093029311352e-06, "loss": 0.111, "step": 45784 }, { "epoch": 33.54212454212454, "grad_norm": 0.6171875, "learning_rate": 1.469338471876542e-06, "loss": 0.1002, "step": 45785 }, { "epoch": 33.542857142857144, "grad_norm": 0.96484375, "learning_rate": 1.468867875743933e-06, "loss": 0.1031, "step": 45786 }, { "epoch": 33.54358974358974, "grad_norm": 1.4375, "learning_rate": 1.4683975145355782e-06, "loss": 0.1346, "step": 45787 }, { "epoch": 33.544322344322346, "grad_norm": 1.078125, "learning_rate": 1.4679273882537308e-06, "loss": 0.1138, "step": 45788 }, { "epoch": 33.54505494505494, "grad_norm": 0.7890625, "learning_rate": 1.467457496900638e-06, "loss": 0.0973, "step": 45789 }, { "epoch": 33.54578754578755, "grad_norm": 1.0703125, "learning_rate": 1.466987840478548e-06, "loss": 0.1182, "step": 45790 }, { "epoch": 33.546520146520145, "grad_norm": 0.96875, "learning_rate": 1.4665184189897194e-06, "loss": 0.1635, "step": 45791 }, { "epoch": 33.54725274725275, "grad_norm": 0.65625, "learning_rate": 1.4660492324363942e-06, "loss": 0.1058, "step": 45792 }, { "epoch": 33.547985347985346, "grad_norm": 1.1484375, "learning_rate": 1.4655802808208315e-06, "loss": 0.1058, "step": 45793 }, { "epoch": 33.54871794871795, "grad_norm": 0.8203125, "learning_rate": 1.4651115641452676e-06, "loss": 0.1096, "step": 45794 }, { "epoch": 33.54945054945055, "grad_norm": 0.9609375, "learning_rate": 1.4646430824119504e-06, "loss": 0.1018, "step": 45795 }, { "epoch": 33.55018315018315, "grad_norm": 1.109375, "learning_rate": 1.464174835623122e-06, "loss": 0.1216, "step": 45796 }, { "epoch": 33.55091575091575, "grad_norm": 1.4765625, "learning_rate": 1.4637068237810304e-06, "loss": 0.1339, "step": 45797 }, { "epoch": 33.55164835164835, "grad_norm": 1.015625, "learning_rate": 1.4632390468879175e-06, "loss": 0.106, "step": 45798 }, { "epoch": 33.55238095238095, "grad_norm": 1.5390625, "learning_rate": 1.4627715049460254e-06, "loss": 0.117, "step": 45799 }, { "epoch": 33.553113553113555, "grad_norm": 0.59765625, "learning_rate": 1.4623041979575966e-06, "loss": 0.1365, "step": 45800 }, { "epoch": 33.55384615384615, "grad_norm": 0.92578125, "learning_rate": 1.4618371259248567e-06, "loss": 0.115, "step": 45801 }, { "epoch": 33.554578754578756, "grad_norm": 1.2421875, "learning_rate": 1.4613702888500642e-06, "loss": 0.1635, "step": 45802 }, { "epoch": 33.55531135531135, "grad_norm": 1.703125, "learning_rate": 1.4609036867354396e-06, "loss": 0.1349, "step": 45803 }, { "epoch": 33.55604395604396, "grad_norm": 1.734375, "learning_rate": 1.4604373195832252e-06, "loss": 0.1397, "step": 45804 }, { "epoch": 33.556776556776555, "grad_norm": 0.84375, "learning_rate": 1.459971187395652e-06, "loss": 0.1307, "step": 45805 }, { "epoch": 33.55750915750916, "grad_norm": 1.4296875, "learning_rate": 1.4595052901749567e-06, "loss": 0.1338, "step": 45806 }, { "epoch": 33.558241758241756, "grad_norm": 1.1171875, "learning_rate": 1.4590396279233705e-06, "loss": 0.0982, "step": 45807 }, { "epoch": 33.55897435897436, "grad_norm": 1.2265625, "learning_rate": 1.4585742006431247e-06, "loss": 0.1122, "step": 45808 }, { "epoch": 33.55970695970696, "grad_norm": 0.921875, "learning_rate": 1.458109008336451e-06, "loss": 0.1217, "step": 45809 }, { "epoch": 33.56043956043956, "grad_norm": 1.0625, "learning_rate": 1.457644051005574e-06, "loss": 0.1113, "step": 45810 }, { "epoch": 33.56117216117216, "grad_norm": 1.109375, "learning_rate": 1.4571793286527257e-06, "loss": 0.133, "step": 45811 }, { "epoch": 33.56190476190476, "grad_norm": 0.7578125, "learning_rate": 1.4567148412801262e-06, "loss": 0.1149, "step": 45812 }, { "epoch": 33.56263736263736, "grad_norm": 1.40625, "learning_rate": 1.456250588890012e-06, "loss": 0.1306, "step": 45813 }, { "epoch": 33.563369963369965, "grad_norm": 1.640625, "learning_rate": 1.4557865714845981e-06, "loss": 0.1162, "step": 45814 }, { "epoch": 33.56410256410256, "grad_norm": 1.2421875, "learning_rate": 1.4553227890661103e-06, "loss": 0.1229, "step": 45815 }, { "epoch": 33.564835164835166, "grad_norm": 0.93359375, "learning_rate": 1.4548592416367738e-06, "loss": 0.1374, "step": 45816 }, { "epoch": 33.565567765567764, "grad_norm": 0.98828125, "learning_rate": 1.454395929198804e-06, "loss": 0.0925, "step": 45817 }, { "epoch": 33.56630036630037, "grad_norm": 1.78125, "learning_rate": 1.4539328517544262e-06, "loss": 0.1088, "step": 45818 }, { "epoch": 33.567032967032965, "grad_norm": 0.99609375, "learning_rate": 1.4534700093058552e-06, "loss": 0.1254, "step": 45819 }, { "epoch": 33.56776556776557, "grad_norm": 1.2109375, "learning_rate": 1.4530074018553112e-06, "loss": 0.1297, "step": 45820 }, { "epoch": 33.56849816849817, "grad_norm": 1.1015625, "learning_rate": 1.4525450294050146e-06, "loss": 0.1301, "step": 45821 }, { "epoch": 33.56923076923077, "grad_norm": 1.4921875, "learning_rate": 1.4520828919571685e-06, "loss": 0.145, "step": 45822 }, { "epoch": 33.56996336996337, "grad_norm": 1.0625, "learning_rate": 1.4516209895139997e-06, "loss": 0.13, "step": 45823 }, { "epoch": 33.57069597069597, "grad_norm": 0.859375, "learning_rate": 1.4511593220777113e-06, "loss": 0.0957, "step": 45824 }, { "epoch": 33.57142857142857, "grad_norm": 0.94921875, "learning_rate": 1.4506978896505232e-06, "loss": 0.09, "step": 45825 }, { "epoch": 33.572161172161174, "grad_norm": 0.87890625, "learning_rate": 1.4502366922346453e-06, "loss": 0.1334, "step": 45826 }, { "epoch": 33.57289377289377, "grad_norm": 0.73046875, "learning_rate": 1.449775729832281e-06, "loss": 0.1126, "step": 45827 }, { "epoch": 33.573626373626375, "grad_norm": 0.94140625, "learning_rate": 1.449315002445645e-06, "loss": 0.1486, "step": 45828 }, { "epoch": 33.57435897435897, "grad_norm": 1.3515625, "learning_rate": 1.4488545100769468e-06, "loss": 0.1248, "step": 45829 }, { "epoch": 33.57509157509158, "grad_norm": 0.91796875, "learning_rate": 1.4483942527283843e-06, "loss": 0.1399, "step": 45830 }, { "epoch": 33.575824175824174, "grad_norm": 1.453125, "learning_rate": 1.4479342304021725e-06, "loss": 0.1196, "step": 45831 }, { "epoch": 33.57655677655678, "grad_norm": 1.1171875, "learning_rate": 1.4474744431005092e-06, "loss": 0.1606, "step": 45832 }, { "epoch": 33.577289377289375, "grad_norm": 1.328125, "learning_rate": 1.4470148908255932e-06, "loss": 0.1385, "step": 45833 }, { "epoch": 33.57802197802198, "grad_norm": 0.8046875, "learning_rate": 1.4465555735796335e-06, "loss": 0.1395, "step": 45834 }, { "epoch": 33.57875457875458, "grad_norm": 1.171875, "learning_rate": 1.4460964913648337e-06, "loss": 0.1425, "step": 45835 }, { "epoch": 33.57948717948718, "grad_norm": 0.92578125, "learning_rate": 1.4456376441833868e-06, "loss": 0.101, "step": 45836 }, { "epoch": 33.58021978021978, "grad_norm": 1.171875, "learning_rate": 1.445179032037491e-06, "loss": 0.169, "step": 45837 }, { "epoch": 33.58095238095238, "grad_norm": 1.0078125, "learning_rate": 1.44472065492935e-06, "loss": 0.1202, "step": 45838 }, { "epoch": 33.58168498168498, "grad_norm": 0.85546875, "learning_rate": 1.444262512861151e-06, "loss": 0.109, "step": 45839 }, { "epoch": 33.582417582417584, "grad_norm": 0.921875, "learning_rate": 1.4438046058350973e-06, "loss": 0.1154, "step": 45840 }, { "epoch": 33.58315018315018, "grad_norm": 1.8828125, "learning_rate": 1.4433469338533826e-06, "loss": 0.1136, "step": 45841 }, { "epoch": 33.583882783882785, "grad_norm": 2.234375, "learning_rate": 1.4428894969181935e-06, "loss": 0.1343, "step": 45842 }, { "epoch": 33.58461538461538, "grad_norm": 0.87109375, "learning_rate": 1.4424322950317285e-06, "loss": 0.1456, "step": 45843 }, { "epoch": 33.58534798534799, "grad_norm": 1.0078125, "learning_rate": 1.441975328196169e-06, "loss": 0.1058, "step": 45844 }, { "epoch": 33.586080586080584, "grad_norm": 0.9765625, "learning_rate": 1.441518596413719e-06, "loss": 0.1214, "step": 45845 }, { "epoch": 33.58681318681319, "grad_norm": 1.09375, "learning_rate": 1.4410620996865493e-06, "loss": 0.0944, "step": 45846 }, { "epoch": 33.587545787545785, "grad_norm": 1.4609375, "learning_rate": 1.4406058380168636e-06, "loss": 0.1231, "step": 45847 }, { "epoch": 33.58827838827839, "grad_norm": 1.1953125, "learning_rate": 1.4401498114068326e-06, "loss": 0.1316, "step": 45848 }, { "epoch": 33.58901098901099, "grad_norm": 1.1953125, "learning_rate": 1.4396940198586601e-06, "loss": 0.1317, "step": 45849 }, { "epoch": 33.58974358974359, "grad_norm": 1.6875, "learning_rate": 1.4392384633745114e-06, "loss": 0.1291, "step": 45850 }, { "epoch": 33.59047619047619, "grad_norm": 1.0859375, "learning_rate": 1.4387831419565738e-06, "loss": 0.1158, "step": 45851 }, { "epoch": 33.59120879120879, "grad_norm": 1.0546875, "learning_rate": 1.4383280556070398e-06, "loss": 0.1111, "step": 45852 }, { "epoch": 33.59194139194139, "grad_norm": 1.1171875, "learning_rate": 1.4378732043280745e-06, "loss": 0.1203, "step": 45853 }, { "epoch": 33.592673992673994, "grad_norm": 1.2734375, "learning_rate": 1.4374185881218714e-06, "loss": 0.1017, "step": 45854 }, { "epoch": 33.59340659340659, "grad_norm": 0.88671875, "learning_rate": 1.4369642069905952e-06, "loss": 0.1409, "step": 45855 }, { "epoch": 33.594139194139196, "grad_norm": 0.8125, "learning_rate": 1.4365100609364388e-06, "loss": 0.1051, "step": 45856 }, { "epoch": 33.59487179487179, "grad_norm": 0.89453125, "learning_rate": 1.4360561499615617e-06, "loss": 0.1079, "step": 45857 }, { "epoch": 33.5956043956044, "grad_norm": 1.2421875, "learning_rate": 1.4356024740681518e-06, "loss": 0.1227, "step": 45858 }, { "epoch": 33.596336996336994, "grad_norm": 0.7578125, "learning_rate": 1.435149033258368e-06, "loss": 0.1189, "step": 45859 }, { "epoch": 33.5970695970696, "grad_norm": 0.54296875, "learning_rate": 1.4346958275344039e-06, "loss": 0.1158, "step": 45860 }, { "epoch": 33.597802197802196, "grad_norm": 1.140625, "learning_rate": 1.4342428568984131e-06, "loss": 0.1324, "step": 45861 }, { "epoch": 33.5985347985348, "grad_norm": 1.828125, "learning_rate": 1.433790121352567e-06, "loss": 0.1431, "step": 45862 }, { "epoch": 33.5992673992674, "grad_norm": 0.88671875, "learning_rate": 1.433337620899047e-06, "loss": 0.1467, "step": 45863 }, { "epoch": 33.6, "grad_norm": 0.98828125, "learning_rate": 1.4328853555400125e-06, "loss": 0.1191, "step": 45864 }, { "epoch": 33.6007326007326, "grad_norm": 1.15625, "learning_rate": 1.4324333252776294e-06, "loss": 0.1284, "step": 45865 }, { "epoch": 33.6014652014652, "grad_norm": 1.3828125, "learning_rate": 1.4319815301140682e-06, "loss": 0.1303, "step": 45866 }, { "epoch": 33.6021978021978, "grad_norm": 0.9765625, "learning_rate": 1.431529970051494e-06, "loss": 0.114, "step": 45867 }, { "epoch": 33.602930402930404, "grad_norm": 0.8359375, "learning_rate": 1.4310786450920669e-06, "loss": 0.1135, "step": 45868 }, { "epoch": 33.603663003663, "grad_norm": 1.4296875, "learning_rate": 1.4306275552379466e-06, "loss": 0.1214, "step": 45869 }, { "epoch": 33.604395604395606, "grad_norm": 1.2734375, "learning_rate": 1.430176700491298e-06, "loss": 0.1147, "step": 45870 }, { "epoch": 33.6051282051282, "grad_norm": 0.69140625, "learning_rate": 1.429726080854287e-06, "loss": 0.1169, "step": 45871 }, { "epoch": 33.60586080586081, "grad_norm": 0.71875, "learning_rate": 1.4292756963290619e-06, "loss": 0.1011, "step": 45872 }, { "epoch": 33.606593406593404, "grad_norm": 0.875, "learning_rate": 1.4288255469177936e-06, "loss": 0.1207, "step": 45873 }, { "epoch": 33.60732600732601, "grad_norm": 1.21875, "learning_rate": 1.4283756326226197e-06, "loss": 0.1112, "step": 45874 }, { "epoch": 33.608058608058606, "grad_norm": 1.34375, "learning_rate": 1.4279259534457114e-06, "loss": 0.0991, "step": 45875 }, { "epoch": 33.60879120879121, "grad_norm": 0.66015625, "learning_rate": 1.4274765093892226e-06, "loss": 0.1049, "step": 45876 }, { "epoch": 33.60952380952381, "grad_norm": 0.875, "learning_rate": 1.4270273004553024e-06, "loss": 0.1095, "step": 45877 }, { "epoch": 33.61025641025641, "grad_norm": 1.8828125, "learning_rate": 1.4265783266461048e-06, "loss": 0.1261, "step": 45878 }, { "epoch": 33.61098901098901, "grad_norm": 1.296875, "learning_rate": 1.4261295879637732e-06, "loss": 0.1347, "step": 45879 }, { "epoch": 33.61172161172161, "grad_norm": 1.109375, "learning_rate": 1.4256810844104728e-06, "loss": 0.0962, "step": 45880 }, { "epoch": 33.61245421245421, "grad_norm": 0.83203125, "learning_rate": 1.4252328159883359e-06, "loss": 0.1086, "step": 45881 }, { "epoch": 33.613186813186815, "grad_norm": 1.1953125, "learning_rate": 1.424784782699528e-06, "loss": 0.1083, "step": 45882 }, { "epoch": 33.61391941391941, "grad_norm": 1.203125, "learning_rate": 1.424336984546181e-06, "loss": 0.1247, "step": 45883 }, { "epoch": 33.614652014652016, "grad_norm": 1.8359375, "learning_rate": 1.4238894215304494e-06, "loss": 0.1263, "step": 45884 }, { "epoch": 33.61538461538461, "grad_norm": 0.8515625, "learning_rate": 1.423442093654471e-06, "loss": 0.1245, "step": 45885 }, { "epoch": 33.61611721611722, "grad_norm": 0.6875, "learning_rate": 1.4229950009203945e-06, "loss": 0.1187, "step": 45886 }, { "epoch": 33.616849816849815, "grad_norm": 1.1640625, "learning_rate": 1.4225481433303579e-06, "loss": 0.0969, "step": 45887 }, { "epoch": 33.61758241758242, "grad_norm": 0.82421875, "learning_rate": 1.422101520886504e-06, "loss": 0.1068, "step": 45888 }, { "epoch": 33.618315018315016, "grad_norm": 1.1875, "learning_rate": 1.4216551335909766e-06, "loss": 0.1238, "step": 45889 }, { "epoch": 33.61904761904762, "grad_norm": 1.1796875, "learning_rate": 1.421208981445908e-06, "loss": 0.1432, "step": 45890 }, { "epoch": 33.61978021978022, "grad_norm": 0.470703125, "learning_rate": 1.4207630644534412e-06, "loss": 0.1133, "step": 45891 }, { "epoch": 33.62051282051282, "grad_norm": 1.2890625, "learning_rate": 1.4203173826157084e-06, "loss": 0.1152, "step": 45892 }, { "epoch": 33.62124542124542, "grad_norm": 0.9296875, "learning_rate": 1.4198719359348477e-06, "loss": 0.1541, "step": 45893 }, { "epoch": 33.62197802197802, "grad_norm": 1.1796875, "learning_rate": 1.4194267244129911e-06, "loss": 0.115, "step": 45894 }, { "epoch": 33.62271062271062, "grad_norm": 1.3515625, "learning_rate": 1.4189817480522712e-06, "loss": 0.1409, "step": 45895 }, { "epoch": 33.623443223443225, "grad_norm": 1.9375, "learning_rate": 1.4185370068548257e-06, "loss": 0.158, "step": 45896 }, { "epoch": 33.62417582417582, "grad_norm": 1.1171875, "learning_rate": 1.4180925008227815e-06, "loss": 0.1316, "step": 45897 }, { "epoch": 33.624908424908426, "grad_norm": 1.0703125, "learning_rate": 1.417648229958265e-06, "loss": 0.1283, "step": 45898 }, { "epoch": 33.62564102564102, "grad_norm": 0.74609375, "learning_rate": 1.417204194263409e-06, "loss": 0.1224, "step": 45899 }, { "epoch": 33.62637362637363, "grad_norm": 0.6171875, "learning_rate": 1.416760393740345e-06, "loss": 0.0939, "step": 45900 }, { "epoch": 33.627106227106225, "grad_norm": 1.234375, "learning_rate": 1.4163168283911953e-06, "loss": 0.1283, "step": 45901 }, { "epoch": 33.62783882783883, "grad_norm": 1.15625, "learning_rate": 1.4158734982180808e-06, "loss": 0.1516, "step": 45902 }, { "epoch": 33.628571428571426, "grad_norm": 0.76171875, "learning_rate": 1.415430403223128e-06, "loss": 0.1221, "step": 45903 }, { "epoch": 33.62930402930403, "grad_norm": 0.8671875, "learning_rate": 1.4149875434084696e-06, "loss": 0.1146, "step": 45904 }, { "epoch": 33.63003663003663, "grad_norm": 0.9140625, "learning_rate": 1.4145449187762103e-06, "loss": 0.1009, "step": 45905 }, { "epoch": 33.63076923076923, "grad_norm": 0.6875, "learning_rate": 1.4141025293284822e-06, "loss": 0.1158, "step": 45906 }, { "epoch": 33.63150183150183, "grad_norm": 0.6171875, "learning_rate": 1.4136603750674067e-06, "loss": 0.1059, "step": 45907 }, { "epoch": 33.63223443223443, "grad_norm": 1.453125, "learning_rate": 1.4132184559950998e-06, "loss": 0.1268, "step": 45908 }, { "epoch": 33.63296703296703, "grad_norm": 1.03125, "learning_rate": 1.4127767721136717e-06, "loss": 0.1115, "step": 45909 }, { "epoch": 33.633699633699635, "grad_norm": 1.71875, "learning_rate": 1.412335323425249e-06, "loss": 0.1181, "step": 45910 }, { "epoch": 33.63443223443223, "grad_norm": 0.7734375, "learning_rate": 1.4118941099319367e-06, "loss": 0.1292, "step": 45911 }, { "epoch": 33.63516483516484, "grad_norm": 1.109375, "learning_rate": 1.411453131635856e-06, "loss": 0.1495, "step": 45912 }, { "epoch": 33.635897435897434, "grad_norm": 1.171875, "learning_rate": 1.4110123885391176e-06, "loss": 0.1361, "step": 45913 }, { "epoch": 33.63663003663004, "grad_norm": 1.375, "learning_rate": 1.4105718806438369e-06, "loss": 0.112, "step": 45914 }, { "epoch": 33.637362637362635, "grad_norm": 1.109375, "learning_rate": 1.4101316079521189e-06, "loss": 0.1304, "step": 45915 }, { "epoch": 33.63809523809524, "grad_norm": 0.78515625, "learning_rate": 1.4096915704660739e-06, "loss": 0.1152, "step": 45916 }, { "epoch": 33.63882783882784, "grad_norm": 1.234375, "learning_rate": 1.4092517681878172e-06, "loss": 0.0989, "step": 45917 }, { "epoch": 33.63956043956044, "grad_norm": 0.984375, "learning_rate": 1.4088122011194489e-06, "loss": 0.1176, "step": 45918 }, { "epoch": 33.64029304029304, "grad_norm": 0.859375, "learning_rate": 1.408372869263079e-06, "loss": 0.1084, "step": 45919 }, { "epoch": 33.64102564102564, "grad_norm": 0.66796875, "learning_rate": 1.4079337726208068e-06, "loss": 0.1171, "step": 45920 }, { "epoch": 33.64175824175824, "grad_norm": 0.890625, "learning_rate": 1.407494911194737e-06, "loss": 0.1248, "step": 45921 }, { "epoch": 33.642490842490844, "grad_norm": 1.203125, "learning_rate": 1.4070562849869802e-06, "loss": 0.1277, "step": 45922 }, { "epoch": 33.64322344322344, "grad_norm": 0.8671875, "learning_rate": 1.40661789399963e-06, "loss": 0.111, "step": 45923 }, { "epoch": 33.643956043956045, "grad_norm": 0.60546875, "learning_rate": 1.406179738234791e-06, "loss": 0.1015, "step": 45924 }, { "epoch": 33.64468864468864, "grad_norm": 0.8984375, "learning_rate": 1.4057418176945631e-06, "loss": 0.1164, "step": 45925 }, { "epoch": 33.64542124542125, "grad_norm": 0.78515625, "learning_rate": 1.4053041323810395e-06, "loss": 0.1053, "step": 45926 }, { "epoch": 33.646153846153844, "grad_norm": 0.74609375, "learning_rate": 1.4048666822963202e-06, "loss": 0.1143, "step": 45927 }, { "epoch": 33.64688644688645, "grad_norm": 0.79296875, "learning_rate": 1.4044294674425039e-06, "loss": 0.1238, "step": 45928 }, { "epoch": 33.647619047619045, "grad_norm": 0.73828125, "learning_rate": 1.4039924878216793e-06, "loss": 0.1228, "step": 45929 }, { "epoch": 33.64835164835165, "grad_norm": 0.96484375, "learning_rate": 1.4035557434359512e-06, "loss": 0.1526, "step": 45930 }, { "epoch": 33.64908424908425, "grad_norm": 1.6640625, "learning_rate": 1.4031192342873966e-06, "loss": 0.135, "step": 45931 }, { "epoch": 33.64981684981685, "grad_norm": 1.015625, "learning_rate": 1.4026829603781208e-06, "loss": 0.0934, "step": 45932 }, { "epoch": 33.65054945054945, "grad_norm": 0.890625, "learning_rate": 1.4022469217102062e-06, "loss": 0.1487, "step": 45933 }, { "epoch": 33.65128205128205, "grad_norm": 0.95703125, "learning_rate": 1.401811118285747e-06, "loss": 0.1194, "step": 45934 }, { "epoch": 33.65201465201465, "grad_norm": 0.75390625, "learning_rate": 1.4013755501068255e-06, "loss": 0.1263, "step": 45935 }, { "epoch": 33.652747252747254, "grad_norm": 1.3828125, "learning_rate": 1.400940217175536e-06, "loss": 0.1067, "step": 45936 }, { "epoch": 33.65347985347985, "grad_norm": 0.76953125, "learning_rate": 1.4005051194939505e-06, "loss": 0.1155, "step": 45937 }, { "epoch": 33.654212454212455, "grad_norm": 0.8359375, "learning_rate": 1.4000702570641733e-06, "loss": 0.1499, "step": 45938 }, { "epoch": 33.65494505494505, "grad_norm": 1.0859375, "learning_rate": 1.3996356298882765e-06, "loss": 0.1131, "step": 45939 }, { "epoch": 33.65567765567766, "grad_norm": 1.4296875, "learning_rate": 1.3992012379683376e-06, "loss": 0.1547, "step": 45940 }, { "epoch": 33.656410256410254, "grad_norm": 0.8125, "learning_rate": 1.3987670813064501e-06, "loss": 0.1131, "step": 45941 }, { "epoch": 33.65714285714286, "grad_norm": 1.2265625, "learning_rate": 1.398333159904686e-06, "loss": 0.1282, "step": 45942 }, { "epoch": 33.657875457875456, "grad_norm": 0.84375, "learning_rate": 1.3978994737651281e-06, "loss": 0.1445, "step": 45943 }, { "epoch": 33.65860805860806, "grad_norm": 1.1015625, "learning_rate": 1.3974660228898536e-06, "loss": 0.1224, "step": 45944 }, { "epoch": 33.65934065934066, "grad_norm": 1.1953125, "learning_rate": 1.3970328072809403e-06, "loss": 0.1295, "step": 45945 }, { "epoch": 33.66007326007326, "grad_norm": 0.8203125, "learning_rate": 1.3965998269404595e-06, "loss": 0.1078, "step": 45946 }, { "epoch": 33.66080586080586, "grad_norm": 0.79296875, "learning_rate": 1.3961670818704833e-06, "loss": 0.0937, "step": 45947 }, { "epoch": 33.66153846153846, "grad_norm": 1.328125, "learning_rate": 1.3957345720731e-06, "loss": 0.1264, "step": 45948 }, { "epoch": 33.66227106227106, "grad_norm": 0.96875, "learning_rate": 1.3953022975503651e-06, "loss": 0.1305, "step": 45949 }, { "epoch": 33.663003663003664, "grad_norm": 0.91015625, "learning_rate": 1.3948702583043613e-06, "loss": 0.119, "step": 45950 }, { "epoch": 33.66373626373626, "grad_norm": 1.28125, "learning_rate": 1.3944384543371493e-06, "loss": 0.1404, "step": 45951 }, { "epoch": 33.664468864468866, "grad_norm": 0.85546875, "learning_rate": 1.3940068856508065e-06, "loss": 0.0931, "step": 45952 }, { "epoch": 33.66520146520146, "grad_norm": 1.015625, "learning_rate": 1.3935755522473996e-06, "loss": 0.1084, "step": 45953 }, { "epoch": 33.66593406593407, "grad_norm": 1.125, "learning_rate": 1.393144454128989e-06, "loss": 0.1349, "step": 45954 }, { "epoch": 33.666666666666664, "grad_norm": 0.69921875, "learning_rate": 1.392713591297647e-06, "loss": 0.0968, "step": 45955 }, { "epoch": 33.66739926739927, "grad_norm": 1.1484375, "learning_rate": 1.3922829637554339e-06, "loss": 0.086, "step": 45956 }, { "epoch": 33.668131868131866, "grad_norm": 0.828125, "learning_rate": 1.3918525715044108e-06, "loss": 0.1146, "step": 45957 }, { "epoch": 33.66886446886447, "grad_norm": 0.7578125, "learning_rate": 1.3914224145466498e-06, "loss": 0.1072, "step": 45958 }, { "epoch": 33.66959706959707, "grad_norm": 0.609375, "learning_rate": 1.3909924928842005e-06, "loss": 0.105, "step": 45959 }, { "epoch": 33.67032967032967, "grad_norm": 1.375, "learning_rate": 1.3905628065191292e-06, "loss": 0.1484, "step": 45960 }, { "epoch": 33.67106227106227, "grad_norm": 1.5703125, "learning_rate": 1.3901333554534913e-06, "loss": 0.1507, "step": 45961 }, { "epoch": 33.67179487179487, "grad_norm": 1.125, "learning_rate": 1.3897041396893477e-06, "loss": 0.134, "step": 45962 }, { "epoch": 33.67252747252747, "grad_norm": 0.99609375, "learning_rate": 1.3892751592287484e-06, "loss": 0.1151, "step": 45963 }, { "epoch": 33.673260073260074, "grad_norm": 0.93359375, "learning_rate": 1.3888464140737594e-06, "loss": 0.1152, "step": 45964 }, { "epoch": 33.67399267399267, "grad_norm": 1.046875, "learning_rate": 1.3884179042264308e-06, "loss": 0.1024, "step": 45965 }, { "epoch": 33.674725274725276, "grad_norm": 1.578125, "learning_rate": 1.3879896296888125e-06, "loss": 0.1522, "step": 45966 }, { "epoch": 33.67545787545787, "grad_norm": 1.1015625, "learning_rate": 1.38756159046296e-06, "loss": 0.1277, "step": 45967 }, { "epoch": 33.67619047619048, "grad_norm": 0.890625, "learning_rate": 1.387133786550917e-06, "loss": 0.1106, "step": 45968 }, { "epoch": 33.676923076923075, "grad_norm": 1.0625, "learning_rate": 1.3867062179547395e-06, "loss": 0.1313, "step": 45969 }, { "epoch": 33.67765567765568, "grad_norm": 0.7578125, "learning_rate": 1.3862788846764772e-06, "loss": 0.1065, "step": 45970 }, { "epoch": 33.678388278388276, "grad_norm": 0.87109375, "learning_rate": 1.3858517867181797e-06, "loss": 0.1213, "step": 45971 }, { "epoch": 33.67912087912088, "grad_norm": 1.09375, "learning_rate": 1.3854249240818862e-06, "loss": 0.116, "step": 45972 }, { "epoch": 33.67985347985348, "grad_norm": 0.8046875, "learning_rate": 1.3849982967696409e-06, "loss": 0.103, "step": 45973 }, { "epoch": 33.68058608058608, "grad_norm": 0.83984375, "learning_rate": 1.3845719047834993e-06, "loss": 0.0895, "step": 45974 }, { "epoch": 33.68131868131868, "grad_norm": 0.953125, "learning_rate": 1.3841457481254947e-06, "loss": 0.0929, "step": 45975 }, { "epoch": 33.68205128205128, "grad_norm": 1.421875, "learning_rate": 1.3837198267976714e-06, "loss": 0.1368, "step": 45976 }, { "epoch": 33.68278388278388, "grad_norm": 0.5390625, "learning_rate": 1.383294140802068e-06, "loss": 0.1114, "step": 45977 }, { "epoch": 33.683516483516485, "grad_norm": 1.15625, "learning_rate": 1.3828686901407297e-06, "loss": 0.1374, "step": 45978 }, { "epoch": 33.68424908424908, "grad_norm": 1.109375, "learning_rate": 1.3824434748156892e-06, "loss": 0.148, "step": 45979 }, { "epoch": 33.684981684981686, "grad_norm": 0.91796875, "learning_rate": 1.3820184948289852e-06, "loss": 0.1155, "step": 45980 }, { "epoch": 33.68571428571428, "grad_norm": 0.91015625, "learning_rate": 1.3815937501826572e-06, "loss": 0.1293, "step": 45981 }, { "epoch": 33.68644688644689, "grad_norm": 1.578125, "learning_rate": 1.3811692408787383e-06, "loss": 0.1533, "step": 45982 }, { "epoch": 33.687179487179485, "grad_norm": 0.78515625, "learning_rate": 1.3807449669192616e-06, "loss": 0.1337, "step": 45983 }, { "epoch": 33.68791208791209, "grad_norm": 0.94921875, "learning_rate": 1.380320928306261e-06, "loss": 0.1003, "step": 45984 }, { "epoch": 33.688644688644686, "grad_norm": 1.0546875, "learning_rate": 1.3798971250417638e-06, "loss": 0.1106, "step": 45985 }, { "epoch": 33.68937728937729, "grad_norm": 1.140625, "learning_rate": 1.3794735571278145e-06, "loss": 0.1023, "step": 45986 }, { "epoch": 33.69010989010989, "grad_norm": 0.8828125, "learning_rate": 1.379050224566425e-06, "loss": 0.0912, "step": 45987 }, { "epoch": 33.69084249084249, "grad_norm": 1.109375, "learning_rate": 1.3786271273596337e-06, "loss": 0.1096, "step": 45988 }, { "epoch": 33.69157509157509, "grad_norm": 1.0546875, "learning_rate": 1.3782042655094578e-06, "loss": 0.1149, "step": 45989 }, { "epoch": 33.69230769230769, "grad_norm": 0.90234375, "learning_rate": 1.3777816390179412e-06, "loss": 0.1149, "step": 45990 }, { "epoch": 33.69304029304029, "grad_norm": 1.25, "learning_rate": 1.3773592478870955e-06, "loss": 0.1078, "step": 45991 }, { "epoch": 33.693772893772895, "grad_norm": 0.95703125, "learning_rate": 1.3769370921189433e-06, "loss": 0.128, "step": 45992 }, { "epoch": 33.69450549450549, "grad_norm": 1.0625, "learning_rate": 1.3765151717155177e-06, "loss": 0.1112, "step": 45993 }, { "epoch": 33.695238095238096, "grad_norm": 0.69140625, "learning_rate": 1.3760934866788248e-06, "loss": 0.1013, "step": 45994 }, { "epoch": 33.69597069597069, "grad_norm": 0.6171875, "learning_rate": 1.3756720370109033e-06, "loss": 0.1012, "step": 45995 }, { "epoch": 33.6967032967033, "grad_norm": 1.0703125, "learning_rate": 1.3752508227137594e-06, "loss": 0.1507, "step": 45996 }, { "epoch": 33.697435897435895, "grad_norm": 0.79296875, "learning_rate": 1.3748298437894204e-06, "loss": 0.1122, "step": 45997 }, { "epoch": 33.6981684981685, "grad_norm": 1.1171875, "learning_rate": 1.374409100239893e-06, "loss": 0.1296, "step": 45998 }, { "epoch": 33.698901098901096, "grad_norm": 0.8125, "learning_rate": 1.373988592067204e-06, "loss": 0.0906, "step": 45999 }, { "epoch": 33.6996336996337, "grad_norm": 1.4140625, "learning_rate": 1.373568319273355e-06, "loss": 0.1176, "step": 46000 }, { "epoch": 33.7003663003663, "grad_norm": 0.875, "learning_rate": 1.3731482818603728e-06, "loss": 0.1233, "step": 46001 }, { "epoch": 33.7010989010989, "grad_norm": 0.8515625, "learning_rate": 1.372728479830264e-06, "loss": 0.1195, "step": 46002 }, { "epoch": 33.7018315018315, "grad_norm": 0.67578125, "learning_rate": 1.37230891318504e-06, "loss": 0.1105, "step": 46003 }, { "epoch": 33.702564102564104, "grad_norm": 1.0859375, "learning_rate": 1.3718895819267063e-06, "loss": 0.1347, "step": 46004 }, { "epoch": 33.7032967032967, "grad_norm": 0.85546875, "learning_rate": 1.3714704860572855e-06, "loss": 0.1215, "step": 46005 }, { "epoch": 33.704029304029305, "grad_norm": 1.234375, "learning_rate": 1.3710516255787722e-06, "loss": 0.1354, "step": 46006 }, { "epoch": 33.7047619047619, "grad_norm": 1.328125, "learning_rate": 1.3706330004931838e-06, "loss": 0.1431, "step": 46007 }, { "epoch": 33.70549450549451, "grad_norm": 0.81640625, "learning_rate": 1.370214610802515e-06, "loss": 0.1183, "step": 46008 }, { "epoch": 33.706227106227104, "grad_norm": 2.0, "learning_rate": 1.369796456508777e-06, "loss": 0.1388, "step": 46009 }, { "epoch": 33.70695970695971, "grad_norm": 1.046875, "learning_rate": 1.3693785376139758e-06, "loss": 0.1577, "step": 46010 }, { "epoch": 33.707692307692305, "grad_norm": 0.94140625, "learning_rate": 1.3689608541201065e-06, "loss": 0.118, "step": 46011 }, { "epoch": 33.70842490842491, "grad_norm": 1.3828125, "learning_rate": 1.3685434060291803e-06, "loss": 0.1573, "step": 46012 }, { "epoch": 33.70915750915751, "grad_norm": 1.2265625, "learning_rate": 1.3681261933431867e-06, "loss": 0.1282, "step": 46013 }, { "epoch": 33.70989010989011, "grad_norm": 1.8203125, "learning_rate": 1.3677092160641314e-06, "loss": 0.1369, "step": 46014 }, { "epoch": 33.71062271062271, "grad_norm": 0.72265625, "learning_rate": 1.3672924741940095e-06, "loss": 0.102, "step": 46015 }, { "epoch": 33.71135531135531, "grad_norm": 1.09375, "learning_rate": 1.3668759677348216e-06, "loss": 0.1301, "step": 46016 }, { "epoch": 33.71208791208791, "grad_norm": 0.9453125, "learning_rate": 1.3664596966885566e-06, "loss": 0.127, "step": 46017 }, { "epoch": 33.712820512820514, "grad_norm": 0.7734375, "learning_rate": 1.3660436610572097e-06, "loss": 0.1271, "step": 46018 }, { "epoch": 33.71355311355311, "grad_norm": 0.86328125, "learning_rate": 1.365627860842787e-06, "loss": 0.1076, "step": 46019 }, { "epoch": 33.714285714285715, "grad_norm": 1.453125, "learning_rate": 1.3652122960472608e-06, "loss": 0.1384, "step": 46020 }, { "epoch": 33.71501831501831, "grad_norm": 1.2890625, "learning_rate": 1.3647969666726374e-06, "loss": 0.1188, "step": 46021 }, { "epoch": 33.71575091575092, "grad_norm": 0.69921875, "learning_rate": 1.3643818727209007e-06, "loss": 0.1172, "step": 46022 }, { "epoch": 33.716483516483514, "grad_norm": 1.7578125, "learning_rate": 1.3639670141940402e-06, "loss": 0.1311, "step": 46023 }, { "epoch": 33.71721611721612, "grad_norm": 1.515625, "learning_rate": 1.3635523910940452e-06, "loss": 0.1708, "step": 46024 }, { "epoch": 33.717948717948715, "grad_norm": 1.140625, "learning_rate": 1.3631380034228996e-06, "loss": 0.1066, "step": 46025 }, { "epoch": 33.71868131868132, "grad_norm": 0.8828125, "learning_rate": 1.3627238511825929e-06, "loss": 0.0996, "step": 46026 }, { "epoch": 33.71941391941392, "grad_norm": 1.2109375, "learning_rate": 1.3623099343751089e-06, "loss": 0.1555, "step": 46027 }, { "epoch": 33.72014652014652, "grad_norm": 1.0, "learning_rate": 1.3618962530024205e-06, "loss": 0.1152, "step": 46028 }, { "epoch": 33.72087912087912, "grad_norm": 1.34375, "learning_rate": 1.3614828070665226e-06, "loss": 0.1217, "step": 46029 }, { "epoch": 33.72161172161172, "grad_norm": 1.2109375, "learning_rate": 1.3610695965693938e-06, "loss": 0.1355, "step": 46030 }, { "epoch": 33.72234432234432, "grad_norm": 1.7109375, "learning_rate": 1.360656621513007e-06, "loss": 0.1147, "step": 46031 }, { "epoch": 33.723076923076924, "grad_norm": 0.95703125, "learning_rate": 1.3602438818993512e-06, "loss": 0.108, "step": 46032 }, { "epoch": 33.72380952380952, "grad_norm": 1.140625, "learning_rate": 1.3598313777303945e-06, "loss": 0.0945, "step": 46033 }, { "epoch": 33.724542124542126, "grad_norm": 2.078125, "learning_rate": 1.3594191090081256e-06, "loss": 0.1383, "step": 46034 }, { "epoch": 33.72527472527472, "grad_norm": 1.046875, "learning_rate": 1.3590070757345014e-06, "loss": 0.1076, "step": 46035 }, { "epoch": 33.72600732600733, "grad_norm": 1.4765625, "learning_rate": 1.3585952779115113e-06, "loss": 0.1158, "step": 46036 }, { "epoch": 33.726739926739924, "grad_norm": 0.9140625, "learning_rate": 1.358183715541117e-06, "loss": 0.1394, "step": 46037 }, { "epoch": 33.72747252747253, "grad_norm": 0.578125, "learning_rate": 1.3577723886253083e-06, "loss": 0.118, "step": 46038 }, { "epoch": 33.728205128205126, "grad_norm": 0.890625, "learning_rate": 1.3573612971660356e-06, "loss": 0.1214, "step": 46039 }, { "epoch": 33.72893772893773, "grad_norm": 0.5625, "learning_rate": 1.3569504411652775e-06, "loss": 0.0989, "step": 46040 }, { "epoch": 33.72967032967033, "grad_norm": 1.078125, "learning_rate": 1.3565398206250018e-06, "loss": 0.1308, "step": 46041 }, { "epoch": 33.73040293040293, "grad_norm": 0.734375, "learning_rate": 1.356129435547181e-06, "loss": 0.0944, "step": 46042 }, { "epoch": 33.73113553113553, "grad_norm": 0.99609375, "learning_rate": 1.3557192859337771e-06, "loss": 0.1147, "step": 46043 }, { "epoch": 33.73186813186813, "grad_norm": 1.1640625, "learning_rate": 1.355309371786752e-06, "loss": 0.1125, "step": 46044 }, { "epoch": 33.73260073260073, "grad_norm": 1.234375, "learning_rate": 1.3548996931080736e-06, "loss": 0.1853, "step": 46045 }, { "epoch": 33.733333333333334, "grad_norm": 1.375, "learning_rate": 1.354490249899703e-06, "loss": 0.125, "step": 46046 }, { "epoch": 33.73406593406593, "grad_norm": 0.96484375, "learning_rate": 1.3540810421636084e-06, "loss": 0.1049, "step": 46047 }, { "epoch": 33.734798534798536, "grad_norm": 1.046875, "learning_rate": 1.3536720699017403e-06, "loss": 0.1133, "step": 46048 }, { "epoch": 33.73553113553113, "grad_norm": 0.890625, "learning_rate": 1.353263333116066e-06, "loss": 0.1184, "step": 46049 }, { "epoch": 33.73626373626374, "grad_norm": 0.8515625, "learning_rate": 1.3528548318085371e-06, "loss": 0.1167, "step": 46050 }, { "epoch": 33.736996336996334, "grad_norm": 0.90234375, "learning_rate": 1.352446565981115e-06, "loss": 0.1344, "step": 46051 }, { "epoch": 33.73772893772894, "grad_norm": 1.1015625, "learning_rate": 1.352038535635756e-06, "loss": 0.1226, "step": 46052 }, { "epoch": 33.738461538461536, "grad_norm": 1.5703125, "learning_rate": 1.351630740774417e-06, "loss": 0.1572, "step": 46053 }, { "epoch": 33.73919413919414, "grad_norm": 1.1015625, "learning_rate": 1.3512231813990489e-06, "loss": 0.1351, "step": 46054 }, { "epoch": 33.73992673992674, "grad_norm": 1.28125, "learning_rate": 1.3508158575116023e-06, "loss": 0.116, "step": 46055 }, { "epoch": 33.74065934065934, "grad_norm": 0.875, "learning_rate": 1.3504087691140339e-06, "loss": 0.0947, "step": 46056 }, { "epoch": 33.74139194139194, "grad_norm": 0.8515625, "learning_rate": 1.3500019162082887e-06, "loss": 0.1214, "step": 46057 }, { "epoch": 33.74212454212454, "grad_norm": 0.609375, "learning_rate": 1.3495952987963237e-06, "loss": 0.1185, "step": 46058 }, { "epoch": 33.74285714285714, "grad_norm": 0.91015625, "learning_rate": 1.3491889168800786e-06, "loss": 0.123, "step": 46059 }, { "epoch": 33.743589743589745, "grad_norm": 0.828125, "learning_rate": 1.3487827704615098e-06, "loss": 0.0908, "step": 46060 }, { "epoch": 33.74432234432234, "grad_norm": 1.1015625, "learning_rate": 1.3483768595425518e-06, "loss": 0.0982, "step": 46061 }, { "epoch": 33.745054945054946, "grad_norm": 0.796875, "learning_rate": 1.3479711841251613e-06, "loss": 0.1193, "step": 46062 }, { "epoch": 33.74578754578754, "grad_norm": 0.76171875, "learning_rate": 1.3475657442112723e-06, "loss": 0.1234, "step": 46063 }, { "epoch": 33.74652014652015, "grad_norm": 0.80859375, "learning_rate": 1.347160539802836e-06, "loss": 0.1215, "step": 46064 }, { "epoch": 33.747252747252745, "grad_norm": 0.84375, "learning_rate": 1.3467555709017813e-06, "loss": 0.1351, "step": 46065 }, { "epoch": 33.74798534798535, "grad_norm": 1.4765625, "learning_rate": 1.3463508375100644e-06, "loss": 0.1236, "step": 46066 }, { "epoch": 33.748717948717946, "grad_norm": 1.046875, "learning_rate": 1.3459463396296093e-06, "loss": 0.1038, "step": 46067 }, { "epoch": 33.74945054945055, "grad_norm": 0.80078125, "learning_rate": 1.345542077262372e-06, "loss": 0.1017, "step": 46068 }, { "epoch": 33.75018315018315, "grad_norm": 1.234375, "learning_rate": 1.3451380504102708e-06, "loss": 0.1075, "step": 46069 }, { "epoch": 33.75091575091575, "grad_norm": 1.03125, "learning_rate": 1.3447342590752505e-06, "loss": 0.143, "step": 46070 }, { "epoch": 33.75164835164835, "grad_norm": 0.7421875, "learning_rate": 1.3443307032592517e-06, "loss": 0.1278, "step": 46071 }, { "epoch": 33.75238095238095, "grad_norm": 0.89453125, "learning_rate": 1.3439273829641924e-06, "loss": 0.1106, "step": 46072 }, { "epoch": 33.75311355311355, "grad_norm": 1.1015625, "learning_rate": 1.3435242981920175e-06, "loss": 0.113, "step": 46073 }, { "epoch": 33.753846153846155, "grad_norm": 1.078125, "learning_rate": 1.3431214489446564e-06, "loss": 0.1189, "step": 46074 }, { "epoch": 33.75457875457875, "grad_norm": 1.078125, "learning_rate": 1.3427188352240378e-06, "loss": 0.1169, "step": 46075 }, { "epoch": 33.755311355311356, "grad_norm": 0.96875, "learning_rate": 1.3423164570320851e-06, "loss": 0.1204, "step": 46076 }, { "epoch": 33.75604395604395, "grad_norm": 0.62109375, "learning_rate": 1.3419143143707333e-06, "loss": 0.1148, "step": 46077 }, { "epoch": 33.75677655677656, "grad_norm": 1.390625, "learning_rate": 1.341512407241911e-06, "loss": 0.1589, "step": 46078 }, { "epoch": 33.757509157509155, "grad_norm": 1.2421875, "learning_rate": 1.3411107356475413e-06, "loss": 0.1493, "step": 46079 }, { "epoch": 33.75824175824176, "grad_norm": 0.7109375, "learning_rate": 1.3407092995895427e-06, "loss": 0.1292, "step": 46080 }, { "epoch": 33.758974358974356, "grad_norm": 0.83203125, "learning_rate": 1.3403080990698495e-06, "loss": 0.1352, "step": 46081 }, { "epoch": 33.75970695970696, "grad_norm": 1.2265625, "learning_rate": 1.3399071340903739e-06, "loss": 0.1243, "step": 46082 }, { "epoch": 33.76043956043956, "grad_norm": 1.640625, "learning_rate": 1.3395064046530397e-06, "loss": 0.1202, "step": 46083 }, { "epoch": 33.76117216117216, "grad_norm": 1.0, "learning_rate": 1.3391059107597757e-06, "loss": 0.1345, "step": 46084 }, { "epoch": 33.76190476190476, "grad_norm": 0.85546875, "learning_rate": 1.3387056524124891e-06, "loss": 0.1412, "step": 46085 }, { "epoch": 33.76263736263736, "grad_norm": 1.0546875, "learning_rate": 1.3383056296131031e-06, "loss": 0.1274, "step": 46086 }, { "epoch": 33.76336996336996, "grad_norm": 0.984375, "learning_rate": 1.3379058423635357e-06, "loss": 0.1098, "step": 46087 }, { "epoch": 33.764102564102565, "grad_norm": 0.9453125, "learning_rate": 1.337506290665694e-06, "loss": 0.1121, "step": 46088 }, { "epoch": 33.76483516483516, "grad_norm": 1.1328125, "learning_rate": 1.3371069745215014e-06, "loss": 0.1182, "step": 46089 }, { "epoch": 33.765567765567766, "grad_norm": 0.75, "learning_rate": 1.336707893932876e-06, "loss": 0.0933, "step": 46090 }, { "epoch": 33.766300366300364, "grad_norm": 0.80078125, "learning_rate": 1.3363090489017132e-06, "loss": 0.1189, "step": 46091 }, { "epoch": 33.76703296703297, "grad_norm": 1.0390625, "learning_rate": 1.3359104394299373e-06, "loss": 0.1403, "step": 46092 }, { "epoch": 33.767765567765565, "grad_norm": 0.79296875, "learning_rate": 1.335512065519449e-06, "loss": 0.112, "step": 46093 }, { "epoch": 33.76849816849817, "grad_norm": 1.0, "learning_rate": 1.335113927172167e-06, "loss": 0.1067, "step": 46094 }, { "epoch": 33.76923076923077, "grad_norm": 0.953125, "learning_rate": 1.3347160243899922e-06, "loss": 0.1202, "step": 46095 }, { "epoch": 33.76996336996337, "grad_norm": 1.25, "learning_rate": 1.3343183571748374e-06, "loss": 0.1308, "step": 46096 }, { "epoch": 33.77069597069597, "grad_norm": 1.1953125, "learning_rate": 1.3339209255285986e-06, "loss": 0.1808, "step": 46097 }, { "epoch": 33.77142857142857, "grad_norm": 1.1015625, "learning_rate": 1.3335237294531823e-06, "loss": 0.1034, "step": 46098 }, { "epoch": 33.77216117216117, "grad_norm": 1.0625, "learning_rate": 1.3331267689505017e-06, "loss": 0.1053, "step": 46099 }, { "epoch": 33.772893772893774, "grad_norm": 1.1328125, "learning_rate": 1.3327300440224465e-06, "loss": 0.1476, "step": 46100 }, { "epoch": 33.77362637362637, "grad_norm": 0.97265625, "learning_rate": 1.3323335546709243e-06, "loss": 0.0897, "step": 46101 }, { "epoch": 33.774358974358975, "grad_norm": 1.03125, "learning_rate": 1.3319373008978256e-06, "loss": 0.1089, "step": 46102 }, { "epoch": 33.77509157509157, "grad_norm": 1.171875, "learning_rate": 1.3315412827050623e-06, "loss": 0.0953, "step": 46103 }, { "epoch": 33.77582417582418, "grad_norm": 1.0625, "learning_rate": 1.3311455000945252e-06, "loss": 0.107, "step": 46104 }, { "epoch": 33.776556776556774, "grad_norm": 1.3515625, "learning_rate": 1.3307499530681106e-06, "loss": 0.1484, "step": 46105 }, { "epoch": 33.77728937728938, "grad_norm": 0.8125, "learning_rate": 1.330354641627714e-06, "loss": 0.1223, "step": 46106 }, { "epoch": 33.778021978021975, "grad_norm": 0.93359375, "learning_rate": 1.3299595657752316e-06, "loss": 0.1048, "step": 46107 }, { "epoch": 33.77875457875458, "grad_norm": 0.91015625, "learning_rate": 1.329564725512554e-06, "loss": 0.1253, "step": 46108 }, { "epoch": 33.77948717948718, "grad_norm": 0.828125, "learning_rate": 1.3291701208415714e-06, "loss": 0.1377, "step": 46109 }, { "epoch": 33.78021978021978, "grad_norm": 1.1171875, "learning_rate": 1.3287757517641744e-06, "loss": 0.1352, "step": 46110 }, { "epoch": 33.78095238095238, "grad_norm": 1.03125, "learning_rate": 1.3283816182822592e-06, "loss": 0.1116, "step": 46111 }, { "epoch": 33.78168498168498, "grad_norm": 1.0703125, "learning_rate": 1.3279877203977103e-06, "loss": 0.0934, "step": 46112 }, { "epoch": 33.78241758241758, "grad_norm": 1.5859375, "learning_rate": 1.3275940581124134e-06, "loss": 0.1176, "step": 46113 }, { "epoch": 33.783150183150184, "grad_norm": 0.82421875, "learning_rate": 1.327200631428253e-06, "loss": 0.1221, "step": 46114 }, { "epoch": 33.78388278388278, "grad_norm": 0.98046875, "learning_rate": 1.326807440347114e-06, "loss": 0.1154, "step": 46115 }, { "epoch": 33.784615384615385, "grad_norm": 0.97265625, "learning_rate": 1.3264144848708927e-06, "loss": 0.1059, "step": 46116 }, { "epoch": 33.78534798534798, "grad_norm": 0.9921875, "learning_rate": 1.326021765001452e-06, "loss": 0.0912, "step": 46117 }, { "epoch": 33.78608058608059, "grad_norm": 0.94921875, "learning_rate": 1.3256292807406877e-06, "loss": 0.1111, "step": 46118 }, { "epoch": 33.786813186813184, "grad_norm": 1.3515625, "learning_rate": 1.3252370320904795e-06, "loss": 0.1303, "step": 46119 }, { "epoch": 33.78754578754579, "grad_norm": 1.3828125, "learning_rate": 1.3248450190527017e-06, "loss": 0.1226, "step": 46120 }, { "epoch": 33.788278388278385, "grad_norm": 0.87890625, "learning_rate": 1.3244532416292331e-06, "loss": 0.1009, "step": 46121 }, { "epoch": 33.78901098901099, "grad_norm": 0.86328125, "learning_rate": 1.3240616998219538e-06, "loss": 0.1218, "step": 46122 }, { "epoch": 33.78974358974359, "grad_norm": 0.859375, "learning_rate": 1.3236703936327374e-06, "loss": 0.116, "step": 46123 }, { "epoch": 33.79047619047619, "grad_norm": 0.9921875, "learning_rate": 1.3232793230634582e-06, "loss": 0.1295, "step": 46124 }, { "epoch": 33.79120879120879, "grad_norm": 0.6640625, "learning_rate": 1.3228884881159955e-06, "loss": 0.0713, "step": 46125 }, { "epoch": 33.79194139194139, "grad_norm": 0.890625, "learning_rate": 1.3224978887922123e-06, "loss": 0.1013, "step": 46126 }, { "epoch": 33.79267399267399, "grad_norm": 1.6015625, "learning_rate": 1.3221075250939937e-06, "loss": 0.1307, "step": 46127 }, { "epoch": 33.793406593406594, "grad_norm": 1.375, "learning_rate": 1.3217173970231971e-06, "loss": 0.1215, "step": 46128 }, { "epoch": 33.79413919413919, "grad_norm": 0.90625, "learning_rate": 1.3213275045816967e-06, "loss": 0.1151, "step": 46129 }, { "epoch": 33.794871794871796, "grad_norm": 0.9609375, "learning_rate": 1.3209378477713608e-06, "loss": 0.1182, "step": 46130 }, { "epoch": 33.79560439560439, "grad_norm": 1.1953125, "learning_rate": 1.320548426594058e-06, "loss": 0.1515, "step": 46131 }, { "epoch": 33.796336996337, "grad_norm": 1.15625, "learning_rate": 1.3201592410516459e-06, "loss": 0.1238, "step": 46132 }, { "epoch": 33.797069597069594, "grad_norm": 1.09375, "learning_rate": 1.3197702911459984e-06, "loss": 0.136, "step": 46133 }, { "epoch": 33.7978021978022, "grad_norm": 0.96875, "learning_rate": 1.3193815768789783e-06, "loss": 0.1206, "step": 46134 }, { "epoch": 33.798534798534796, "grad_norm": 0.79296875, "learning_rate": 1.3189930982524435e-06, "loss": 0.104, "step": 46135 }, { "epoch": 33.7992673992674, "grad_norm": 1.7109375, "learning_rate": 1.3186048552682622e-06, "loss": 0.1517, "step": 46136 }, { "epoch": 33.8, "grad_norm": 0.60546875, "learning_rate": 1.3182168479282867e-06, "loss": 0.1043, "step": 46137 }, { "epoch": 33.8007326007326, "grad_norm": 1.140625, "learning_rate": 1.3178290762343852e-06, "loss": 0.1322, "step": 46138 }, { "epoch": 33.8014652014652, "grad_norm": 0.6875, "learning_rate": 1.3174415401884047e-06, "loss": 0.1039, "step": 46139 }, { "epoch": 33.8021978021978, "grad_norm": 1.0546875, "learning_rate": 1.3170542397922074e-06, "loss": 0.1334, "step": 46140 }, { "epoch": 33.8029304029304, "grad_norm": 0.77734375, "learning_rate": 1.316667175047646e-06, "loss": 0.127, "step": 46141 }, { "epoch": 33.803663003663004, "grad_norm": 0.89453125, "learning_rate": 1.3162803459565835e-06, "loss": 0.1273, "step": 46142 }, { "epoch": 33.8043956043956, "grad_norm": 1.28125, "learning_rate": 1.3158937525208661e-06, "loss": 0.09, "step": 46143 }, { "epoch": 33.805128205128206, "grad_norm": 0.88671875, "learning_rate": 1.315507394742346e-06, "loss": 0.1191, "step": 46144 }, { "epoch": 33.8058608058608, "grad_norm": 0.94140625, "learning_rate": 1.3151212726228805e-06, "loss": 0.1028, "step": 46145 }, { "epoch": 33.80659340659341, "grad_norm": 0.9609375, "learning_rate": 1.314735386164311e-06, "loss": 0.1196, "step": 46146 }, { "epoch": 33.807326007326004, "grad_norm": 1.0390625, "learning_rate": 1.3143497353684946e-06, "loss": 0.1161, "step": 46147 }, { "epoch": 33.80805860805861, "grad_norm": 1.203125, "learning_rate": 1.3139643202372728e-06, "loss": 0.1087, "step": 46148 }, { "epoch": 33.808791208791206, "grad_norm": 1.703125, "learning_rate": 1.3135791407724974e-06, "loss": 0.1115, "step": 46149 }, { "epoch": 33.80952380952381, "grad_norm": 1.4140625, "learning_rate": 1.3131941969760033e-06, "loss": 0.1624, "step": 46150 }, { "epoch": 33.81025641025641, "grad_norm": 1.09375, "learning_rate": 1.3128094888496493e-06, "loss": 0.1044, "step": 46151 }, { "epoch": 33.81098901098901, "grad_norm": 1.765625, "learning_rate": 1.31242501639527e-06, "loss": 0.1504, "step": 46152 }, { "epoch": 33.81172161172161, "grad_norm": 0.6328125, "learning_rate": 1.3120407796147175e-06, "loss": 0.1088, "step": 46153 }, { "epoch": 33.81245421245421, "grad_norm": 1.15625, "learning_rate": 1.3116567785098166e-06, "loss": 0.1491, "step": 46154 }, { "epoch": 33.81318681318681, "grad_norm": 1.28125, "learning_rate": 1.3112730130824139e-06, "loss": 0.1159, "step": 46155 }, { "epoch": 33.813919413919415, "grad_norm": 1.21875, "learning_rate": 1.3108894833343553e-06, "loss": 0.1338, "step": 46156 }, { "epoch": 33.81465201465201, "grad_norm": 0.88671875, "learning_rate": 1.3105061892674717e-06, "loss": 0.126, "step": 46157 }, { "epoch": 33.815384615384616, "grad_norm": 1.390625, "learning_rate": 1.3101231308835979e-06, "loss": 0.1195, "step": 46158 }, { "epoch": 33.81611721611721, "grad_norm": 0.71875, "learning_rate": 1.3097403081845755e-06, "loss": 0.1221, "step": 46159 }, { "epoch": 33.81684981684982, "grad_norm": 0.97265625, "learning_rate": 1.3093577211722289e-06, "loss": 0.1253, "step": 46160 }, { "epoch": 33.817582417582415, "grad_norm": 1.234375, "learning_rate": 1.308975369848404e-06, "loss": 0.1292, "step": 46161 }, { "epoch": 33.81831501831502, "grad_norm": 1.5703125, "learning_rate": 1.308593254214926e-06, "loss": 0.1676, "step": 46162 }, { "epoch": 33.819047619047616, "grad_norm": 0.9453125, "learning_rate": 1.308211374273625e-06, "loss": 0.1412, "step": 46163 }, { "epoch": 33.81978021978022, "grad_norm": 1.0703125, "learning_rate": 1.3078297300263362e-06, "loss": 0.0983, "step": 46164 }, { "epoch": 33.82051282051282, "grad_norm": 0.9921875, "learning_rate": 1.3074483214748789e-06, "loss": 0.1254, "step": 46165 }, { "epoch": 33.82124542124542, "grad_norm": 0.91015625, "learning_rate": 1.3070671486210882e-06, "loss": 0.108, "step": 46166 }, { "epoch": 33.82197802197802, "grad_norm": 0.76953125, "learning_rate": 1.3066862114667835e-06, "loss": 0.1264, "step": 46167 }, { "epoch": 33.82271062271062, "grad_norm": 1.046875, "learning_rate": 1.3063055100138003e-06, "loss": 0.1077, "step": 46168 }, { "epoch": 33.82344322344322, "grad_norm": 1.2734375, "learning_rate": 1.305925044263952e-06, "loss": 0.134, "step": 46169 }, { "epoch": 33.824175824175825, "grad_norm": 0.78125, "learning_rate": 1.3055448142190637e-06, "loss": 0.0867, "step": 46170 }, { "epoch": 33.82490842490842, "grad_norm": 1.0703125, "learning_rate": 1.3051648198809646e-06, "loss": 0.1134, "step": 46171 }, { "epoch": 33.825641025641026, "grad_norm": 0.59765625, "learning_rate": 1.3047850612514685e-06, "loss": 0.087, "step": 46172 }, { "epoch": 33.82637362637362, "grad_norm": 1.4609375, "learning_rate": 1.3044055383324004e-06, "loss": 0.0974, "step": 46173 }, { "epoch": 33.82710622710623, "grad_norm": 0.9296875, "learning_rate": 1.3040262511255732e-06, "loss": 0.1139, "step": 46174 }, { "epoch": 33.827838827838825, "grad_norm": 0.6640625, "learning_rate": 1.3036471996328066e-06, "loss": 0.1047, "step": 46175 }, { "epoch": 33.82857142857143, "grad_norm": 1.1328125, "learning_rate": 1.3032683838559082e-06, "loss": 0.1353, "step": 46176 }, { "epoch": 33.829304029304026, "grad_norm": 0.96875, "learning_rate": 1.3028898037967137e-06, "loss": 0.1304, "step": 46177 }, { "epoch": 33.83003663003663, "grad_norm": 1.375, "learning_rate": 1.30251145945702e-06, "loss": 0.1056, "step": 46178 }, { "epoch": 33.83076923076923, "grad_norm": 0.9375, "learning_rate": 1.302133350838641e-06, "loss": 0.1163, "step": 46179 }, { "epoch": 33.83150183150183, "grad_norm": 1.453125, "learning_rate": 1.3017554779433954e-06, "loss": 0.1328, "step": 46180 }, { "epoch": 33.83223443223443, "grad_norm": 1.1484375, "learning_rate": 1.3013778407730857e-06, "loss": 0.1188, "step": 46181 }, { "epoch": 33.832967032967034, "grad_norm": 0.93359375, "learning_rate": 1.3010004393295313e-06, "loss": 0.1105, "step": 46182 }, { "epoch": 33.83369963369963, "grad_norm": 1.1015625, "learning_rate": 1.3006232736145342e-06, "loss": 0.1144, "step": 46183 }, { "epoch": 33.834432234432235, "grad_norm": 0.765625, "learning_rate": 1.3002463436298977e-06, "loss": 0.1202, "step": 46184 }, { "epoch": 33.83516483516483, "grad_norm": 1.453125, "learning_rate": 1.2998696493774349e-06, "loss": 0.1468, "step": 46185 }, { "epoch": 33.83589743589744, "grad_norm": 0.60546875, "learning_rate": 1.2994931908589485e-06, "loss": 0.1073, "step": 46186 }, { "epoch": 33.836630036630034, "grad_norm": 1.0625, "learning_rate": 1.299116968076241e-06, "loss": 0.1131, "step": 46187 }, { "epoch": 33.83736263736264, "grad_norm": 1.1796875, "learning_rate": 1.298740981031115e-06, "loss": 0.1442, "step": 46188 }, { "epoch": 33.838095238095235, "grad_norm": 1.5234375, "learning_rate": 1.2983652297253727e-06, "loss": 0.1107, "step": 46189 }, { "epoch": 33.83882783882784, "grad_norm": 1.3203125, "learning_rate": 1.2979897141608175e-06, "loss": 0.1137, "step": 46190 }, { "epoch": 33.83956043956044, "grad_norm": 1.453125, "learning_rate": 1.29761443433924e-06, "loss": 0.1312, "step": 46191 }, { "epoch": 33.84029304029304, "grad_norm": 0.84765625, "learning_rate": 1.297239390262449e-06, "loss": 0.1186, "step": 46192 }, { "epoch": 33.84102564102564, "grad_norm": 0.890625, "learning_rate": 1.2968645819322298e-06, "loss": 0.0969, "step": 46193 }, { "epoch": 33.84175824175824, "grad_norm": 0.9375, "learning_rate": 1.2964900093503913e-06, "loss": 0.1181, "step": 46194 }, { "epoch": 33.84249084249084, "grad_norm": 0.86328125, "learning_rate": 1.296115672518719e-06, "loss": 0.1074, "step": 46195 }, { "epoch": 33.843223443223444, "grad_norm": 0.80859375, "learning_rate": 1.2957415714390103e-06, "loss": 0.1115, "step": 46196 }, { "epoch": 33.84395604395604, "grad_norm": 0.92578125, "learning_rate": 1.2953677061130567e-06, "loss": 0.1307, "step": 46197 }, { "epoch": 33.844688644688645, "grad_norm": 0.9921875, "learning_rate": 1.294994076542644e-06, "loss": 0.118, "step": 46198 }, { "epoch": 33.84542124542124, "grad_norm": 1.109375, "learning_rate": 1.294620682729575e-06, "loss": 0.1396, "step": 46199 }, { "epoch": 33.84615384615385, "grad_norm": 0.82421875, "learning_rate": 1.2942475246756302e-06, "loss": 0.1114, "step": 46200 }, { "epoch": 33.846886446886444, "grad_norm": 1.265625, "learning_rate": 1.2938746023825957e-06, "loss": 0.1192, "step": 46201 }, { "epoch": 33.84761904761905, "grad_norm": 0.83203125, "learning_rate": 1.2935019158522577e-06, "loss": 0.1344, "step": 46202 }, { "epoch": 33.848351648351645, "grad_norm": 1.0390625, "learning_rate": 1.2931294650864128e-06, "loss": 0.124, "step": 46203 }, { "epoch": 33.84908424908425, "grad_norm": 1.078125, "learning_rate": 1.2927572500868366e-06, "loss": 0.1357, "step": 46204 }, { "epoch": 33.84981684981685, "grad_norm": 1.6328125, "learning_rate": 1.292385270855315e-06, "loss": 0.1563, "step": 46205 }, { "epoch": 33.85054945054945, "grad_norm": 1.421875, "learning_rate": 1.2920135273936231e-06, "loss": 0.1511, "step": 46206 }, { "epoch": 33.85128205128205, "grad_norm": 0.96484375, "learning_rate": 1.2916420197035525e-06, "loss": 0.1267, "step": 46207 }, { "epoch": 33.85201465201465, "grad_norm": 1.1953125, "learning_rate": 1.291270747786878e-06, "loss": 0.1295, "step": 46208 }, { "epoch": 33.85274725274725, "grad_norm": 1.0859375, "learning_rate": 1.2908997116453859e-06, "loss": 0.1318, "step": 46209 }, { "epoch": 33.853479853479854, "grad_norm": 1.3671875, "learning_rate": 1.2905289112808405e-06, "loss": 0.1174, "step": 46210 }, { "epoch": 33.85421245421245, "grad_norm": 0.62890625, "learning_rate": 1.2901583466950274e-06, "loss": 0.1183, "step": 46211 }, { "epoch": 33.854945054945055, "grad_norm": 1.1875, "learning_rate": 1.2897880178897166e-06, "loss": 0.1064, "step": 46212 }, { "epoch": 33.85567765567765, "grad_norm": 1.671875, "learning_rate": 1.2894179248666882e-06, "loss": 0.1418, "step": 46213 }, { "epoch": 33.85641025641026, "grad_norm": 0.83984375, "learning_rate": 1.2890480676277177e-06, "loss": 0.0975, "step": 46214 }, { "epoch": 33.857142857142854, "grad_norm": 1.1875, "learning_rate": 1.2886784461745693e-06, "loss": 0.1217, "step": 46215 }, { "epoch": 33.85787545787546, "grad_norm": 0.890625, "learning_rate": 1.2883090605090177e-06, "loss": 0.135, "step": 46216 }, { "epoch": 33.858608058608056, "grad_norm": 0.90234375, "learning_rate": 1.2879399106328326e-06, "loss": 0.1239, "step": 46217 }, { "epoch": 33.85934065934066, "grad_norm": 1.515625, "learning_rate": 1.2875709965477836e-06, "loss": 0.1525, "step": 46218 }, { "epoch": 33.86007326007326, "grad_norm": 0.77734375, "learning_rate": 1.2872023182556348e-06, "loss": 0.0975, "step": 46219 }, { "epoch": 33.86080586080586, "grad_norm": 1.3359375, "learning_rate": 1.2868338757581615e-06, "loss": 0.1352, "step": 46220 }, { "epoch": 33.86153846153846, "grad_norm": 0.90234375, "learning_rate": 1.2864656690571165e-06, "loss": 0.1063, "step": 46221 }, { "epoch": 33.86227106227106, "grad_norm": 1.0078125, "learning_rate": 1.286097698154269e-06, "loss": 0.1664, "step": 46222 }, { "epoch": 33.86300366300366, "grad_norm": 0.75390625, "learning_rate": 1.285729963051384e-06, "loss": 0.1549, "step": 46223 }, { "epoch": 33.863736263736264, "grad_norm": 1.0625, "learning_rate": 1.2853624637502252e-06, "loss": 0.1334, "step": 46224 }, { "epoch": 33.86446886446886, "grad_norm": 1.8125, "learning_rate": 1.2849952002525566e-06, "loss": 0.0995, "step": 46225 }, { "epoch": 33.865201465201466, "grad_norm": 0.65625, "learning_rate": 1.2846281725601256e-06, "loss": 0.1133, "step": 46226 }, { "epoch": 33.86593406593406, "grad_norm": 0.78515625, "learning_rate": 1.2842613806746967e-06, "loss": 0.1117, "step": 46227 }, { "epoch": 33.86666666666667, "grad_norm": 0.76953125, "learning_rate": 1.2838948245980283e-06, "loss": 0.1373, "step": 46228 }, { "epoch": 33.867399267399264, "grad_norm": 0.81640625, "learning_rate": 1.283528504331879e-06, "loss": 0.1173, "step": 46229 }, { "epoch": 33.86813186813187, "grad_norm": 0.828125, "learning_rate": 1.2831624198779964e-06, "loss": 0.1147, "step": 46230 }, { "epoch": 33.868864468864466, "grad_norm": 1.0546875, "learning_rate": 1.2827965712381446e-06, "loss": 0.0982, "step": 46231 }, { "epoch": 33.86959706959707, "grad_norm": 0.8125, "learning_rate": 1.2824309584140713e-06, "loss": 0.1194, "step": 46232 }, { "epoch": 33.87032967032967, "grad_norm": 0.82421875, "learning_rate": 1.2820655814075238e-06, "loss": 0.1176, "step": 46233 }, { "epoch": 33.87106227106227, "grad_norm": 1.1328125, "learning_rate": 1.2817004402202612e-06, "loss": 0.1215, "step": 46234 }, { "epoch": 33.87179487179487, "grad_norm": 0.56640625, "learning_rate": 1.2813355348540306e-06, "loss": 0.0968, "step": 46235 }, { "epoch": 33.87252747252747, "grad_norm": 1.1171875, "learning_rate": 1.2809708653105743e-06, "loss": 0.1077, "step": 46236 }, { "epoch": 33.87326007326007, "grad_norm": 1.046875, "learning_rate": 1.2806064315916453e-06, "loss": 0.1211, "step": 46237 }, { "epoch": 33.873992673992674, "grad_norm": 0.91796875, "learning_rate": 1.280242233698986e-06, "loss": 0.1029, "step": 46238 }, { "epoch": 33.87472527472527, "grad_norm": 1.21875, "learning_rate": 1.2798782716343494e-06, "loss": 0.1264, "step": 46239 }, { "epoch": 33.875457875457876, "grad_norm": 1.359375, "learning_rate": 1.2795145453994719e-06, "loss": 0.0967, "step": 46240 }, { "epoch": 33.87619047619047, "grad_norm": 0.9296875, "learning_rate": 1.279151054996101e-06, "loss": 0.112, "step": 46241 }, { "epoch": 33.87692307692308, "grad_norm": 1.6640625, "learning_rate": 1.2787878004259739e-06, "loss": 0.157, "step": 46242 }, { "epoch": 33.877655677655675, "grad_norm": 0.984375, "learning_rate": 1.2784247816908321e-06, "loss": 0.1167, "step": 46243 }, { "epoch": 33.87838827838828, "grad_norm": 0.953125, "learning_rate": 1.2780619987924125e-06, "loss": 0.1151, "step": 46244 }, { "epoch": 33.879120879120876, "grad_norm": 0.859375, "learning_rate": 1.2776994517324572e-06, "loss": 0.1299, "step": 46245 }, { "epoch": 33.87985347985348, "grad_norm": 1.28125, "learning_rate": 1.2773371405127085e-06, "loss": 0.1389, "step": 46246 }, { "epoch": 33.88058608058608, "grad_norm": 0.9609375, "learning_rate": 1.2769750651348914e-06, "loss": 0.1111, "step": 46247 }, { "epoch": 33.88131868131868, "grad_norm": 0.984375, "learning_rate": 1.276613225600743e-06, "loss": 0.1225, "step": 46248 }, { "epoch": 33.88205128205128, "grad_norm": 1.5078125, "learning_rate": 1.2762516219119999e-06, "loss": 0.1177, "step": 46249 }, { "epoch": 33.88278388278388, "grad_norm": 0.9921875, "learning_rate": 1.2758902540703985e-06, "loss": 0.1095, "step": 46250 }, { "epoch": 33.88351648351648, "grad_norm": 0.95703125, "learning_rate": 1.2755291220776645e-06, "loss": 0.1347, "step": 46251 }, { "epoch": 33.884249084249085, "grad_norm": 1.234375, "learning_rate": 1.2751682259355347e-06, "loss": 0.1023, "step": 46252 }, { "epoch": 33.88498168498168, "grad_norm": 1.6171875, "learning_rate": 1.2748075656457289e-06, "loss": 0.1602, "step": 46253 }, { "epoch": 33.885714285714286, "grad_norm": 0.9140625, "learning_rate": 1.274447141209973e-06, "loss": 0.1249, "step": 46254 }, { "epoch": 33.88644688644688, "grad_norm": 1.375, "learning_rate": 1.274086952630009e-06, "loss": 0.0968, "step": 46255 }, { "epoch": 33.88717948717949, "grad_norm": 0.74609375, "learning_rate": 1.2737269999075517e-06, "loss": 0.0745, "step": 46256 }, { "epoch": 33.887912087912085, "grad_norm": 0.9765625, "learning_rate": 1.2733672830443318e-06, "loss": 0.158, "step": 46257 }, { "epoch": 33.88864468864469, "grad_norm": 1.09375, "learning_rate": 1.2730078020420699e-06, "loss": 0.1018, "step": 46258 }, { "epoch": 33.889377289377286, "grad_norm": 0.890625, "learning_rate": 1.2726485569024802e-06, "loss": 0.0835, "step": 46259 }, { "epoch": 33.89010989010989, "grad_norm": 1.484375, "learning_rate": 1.2722895476272937e-06, "loss": 0.1034, "step": 46260 }, { "epoch": 33.89084249084249, "grad_norm": 0.9921875, "learning_rate": 1.2719307742182314e-06, "loss": 0.1353, "step": 46261 }, { "epoch": 33.89157509157509, "grad_norm": 1.28125, "learning_rate": 1.2715722366770126e-06, "loss": 0.1097, "step": 46262 }, { "epoch": 33.89230769230769, "grad_norm": 1.2421875, "learning_rate": 1.2712139350053467e-06, "loss": 0.1162, "step": 46263 }, { "epoch": 33.89304029304029, "grad_norm": 0.796875, "learning_rate": 1.270855869204954e-06, "loss": 0.1149, "step": 46264 }, { "epoch": 33.89377289377289, "grad_norm": 0.76953125, "learning_rate": 1.270498039277554e-06, "loss": 0.1136, "step": 46265 }, { "epoch": 33.894505494505495, "grad_norm": 1.3984375, "learning_rate": 1.270140445224862e-06, "loss": 0.118, "step": 46266 }, { "epoch": 33.89523809523809, "grad_norm": 0.8984375, "learning_rate": 1.2697830870485814e-06, "loss": 0.1153, "step": 46267 }, { "epoch": 33.895970695970696, "grad_norm": 1.171875, "learning_rate": 1.269425964750438e-06, "loss": 0.1349, "step": 46268 }, { "epoch": 33.89670329670329, "grad_norm": 1.046875, "learning_rate": 1.2690690783321296e-06, "loss": 0.1365, "step": 46269 }, { "epoch": 33.8974358974359, "grad_norm": 0.921875, "learning_rate": 1.2687124277953763e-06, "loss": 0.1186, "step": 46270 }, { "epoch": 33.898168498168495, "grad_norm": 1.4296875, "learning_rate": 1.2683560131418823e-06, "loss": 0.1298, "step": 46271 }, { "epoch": 33.8989010989011, "grad_norm": 0.7890625, "learning_rate": 1.2679998343733615e-06, "loss": 0.0955, "step": 46272 }, { "epoch": 33.899633699633696, "grad_norm": 0.8671875, "learning_rate": 1.2676438914915074e-06, "loss": 0.1236, "step": 46273 }, { "epoch": 33.9003663003663, "grad_norm": 0.78515625, "learning_rate": 1.2672881844980391e-06, "loss": 0.102, "step": 46274 }, { "epoch": 33.9010989010989, "grad_norm": 1.109375, "learning_rate": 1.2669327133946499e-06, "loss": 0.1261, "step": 46275 }, { "epoch": 33.9018315018315, "grad_norm": 0.8359375, "learning_rate": 1.2665774781830545e-06, "loss": 0.114, "step": 46276 }, { "epoch": 33.9025641025641, "grad_norm": 1.0703125, "learning_rate": 1.2662224788649452e-06, "loss": 0.1129, "step": 46277 }, { "epoch": 33.903296703296704, "grad_norm": 1.609375, "learning_rate": 1.265867715442026e-06, "loss": 0.1337, "step": 46278 }, { "epoch": 33.9040293040293, "grad_norm": 0.9375, "learning_rate": 1.265513187916e-06, "loss": 0.137, "step": 46279 }, { "epoch": 33.904761904761905, "grad_norm": 1.1875, "learning_rate": 1.2651588962885601e-06, "loss": 0.0946, "step": 46280 }, { "epoch": 33.9054945054945, "grad_norm": 1.015625, "learning_rate": 1.2648048405614049e-06, "loss": 0.1442, "step": 46281 }, { "epoch": 33.90622710622711, "grad_norm": 0.73828125, "learning_rate": 1.2644510207362371e-06, "loss": 0.1106, "step": 46282 }, { "epoch": 33.906959706959704, "grad_norm": 1.4453125, "learning_rate": 1.264097436814745e-06, "loss": 0.1659, "step": 46283 }, { "epoch": 33.90769230769231, "grad_norm": 0.99609375, "learning_rate": 1.2637440887986258e-06, "loss": 0.1067, "step": 46284 }, { "epoch": 33.908424908424905, "grad_norm": 1.109375, "learning_rate": 1.2633909766895724e-06, "loss": 0.1281, "step": 46285 }, { "epoch": 33.90915750915751, "grad_norm": 0.875, "learning_rate": 1.2630381004892724e-06, "loss": 0.1268, "step": 46286 }, { "epoch": 33.90989010989011, "grad_norm": 0.890625, "learning_rate": 1.2626854601994233e-06, "loss": 0.125, "step": 46287 }, { "epoch": 33.91062271062271, "grad_norm": 1.1171875, "learning_rate": 1.2623330558217125e-06, "loss": 0.1328, "step": 46288 }, { "epoch": 33.91135531135531, "grad_norm": 1.140625, "learning_rate": 1.2619808873578273e-06, "loss": 0.1623, "step": 46289 }, { "epoch": 33.91208791208791, "grad_norm": 0.9140625, "learning_rate": 1.2616289548094602e-06, "loss": 0.0983, "step": 46290 }, { "epoch": 33.91282051282051, "grad_norm": 1.328125, "learning_rate": 1.2612772581782818e-06, "loss": 0.1015, "step": 46291 }, { "epoch": 33.913553113553114, "grad_norm": 1.1640625, "learning_rate": 1.260925797465996e-06, "loss": 0.1406, "step": 46292 }, { "epoch": 33.91428571428571, "grad_norm": 0.91796875, "learning_rate": 1.2605745726742786e-06, "loss": 0.1257, "step": 46293 }, { "epoch": 33.915018315018315, "grad_norm": 0.828125, "learning_rate": 1.2602235838048114e-06, "loss": 0.1165, "step": 46294 }, { "epoch": 33.91575091575091, "grad_norm": 1.046875, "learning_rate": 1.2598728308592763e-06, "loss": 0.1282, "step": 46295 }, { "epoch": 33.91648351648352, "grad_norm": 1.3046875, "learning_rate": 1.2595223138393547e-06, "loss": 0.1303, "step": 46296 }, { "epoch": 33.917216117216114, "grad_norm": 1.6171875, "learning_rate": 1.2591720327467226e-06, "loss": 0.1498, "step": 46297 }, { "epoch": 33.91794871794872, "grad_norm": 1.3671875, "learning_rate": 1.2588219875830676e-06, "loss": 0.1473, "step": 46298 }, { "epoch": 33.918681318681315, "grad_norm": 1.1640625, "learning_rate": 1.2584721783500598e-06, "loss": 0.1464, "step": 46299 }, { "epoch": 33.91941391941392, "grad_norm": 1.125, "learning_rate": 1.258122605049376e-06, "loss": 0.1268, "step": 46300 }, { "epoch": 33.92014652014652, "grad_norm": 1.3515625, "learning_rate": 1.257773267682692e-06, "loss": 0.1145, "step": 46301 }, { "epoch": 33.92087912087912, "grad_norm": 1.375, "learning_rate": 1.2574241662516838e-06, "loss": 0.1275, "step": 46302 }, { "epoch": 33.92161172161172, "grad_norm": 0.89453125, "learning_rate": 1.2570753007580225e-06, "loss": 0.1227, "step": 46303 }, { "epoch": 33.92234432234432, "grad_norm": 1.671875, "learning_rate": 1.2567266712033729e-06, "loss": 0.1293, "step": 46304 }, { "epoch": 33.92307692307692, "grad_norm": 1.0625, "learning_rate": 1.2563782775894166e-06, "loss": 0.1067, "step": 46305 }, { "epoch": 33.923809523809524, "grad_norm": 1.265625, "learning_rate": 1.2560301199178138e-06, "loss": 0.1355, "step": 46306 }, { "epoch": 33.92454212454212, "grad_norm": 0.93359375, "learning_rate": 1.2556821981902401e-06, "loss": 0.1255, "step": 46307 }, { "epoch": 33.925274725274726, "grad_norm": 0.9375, "learning_rate": 1.2553345124083553e-06, "loss": 0.1309, "step": 46308 }, { "epoch": 33.92600732600732, "grad_norm": 0.75, "learning_rate": 1.254987062573836e-06, "loss": 0.0976, "step": 46309 }, { "epoch": 33.92673992673993, "grad_norm": 1.6640625, "learning_rate": 1.2546398486883303e-06, "loss": 0.1029, "step": 46310 }, { "epoch": 33.927472527472524, "grad_norm": 1.78125, "learning_rate": 1.2542928707535198e-06, "loss": 0.1512, "step": 46311 }, { "epoch": 33.92820512820513, "grad_norm": 0.765625, "learning_rate": 1.2539461287710484e-06, "loss": 0.1021, "step": 46312 }, { "epoch": 33.928937728937726, "grad_norm": 1.1171875, "learning_rate": 1.2535996227425972e-06, "loss": 0.1056, "step": 46313 }, { "epoch": 33.92967032967033, "grad_norm": 1.046875, "learning_rate": 1.2532533526698151e-06, "loss": 0.1153, "step": 46314 }, { "epoch": 33.93040293040293, "grad_norm": 1.0703125, "learning_rate": 1.252907318554356e-06, "loss": 0.1209, "step": 46315 }, { "epoch": 33.93113553113553, "grad_norm": 1.265625, "learning_rate": 1.252561520397891e-06, "loss": 0.1238, "step": 46316 }, { "epoch": 33.93186813186813, "grad_norm": 0.76171875, "learning_rate": 1.252215958202068e-06, "loss": 0.1269, "step": 46317 }, { "epoch": 33.93260073260073, "grad_norm": 0.9296875, "learning_rate": 1.2518706319685478e-06, "loss": 0.1189, "step": 46318 }, { "epoch": 33.93333333333333, "grad_norm": 1.390625, "learning_rate": 1.2515255416989782e-06, "loss": 0.1491, "step": 46319 }, { "epoch": 33.934065934065934, "grad_norm": 0.8359375, "learning_rate": 1.2511806873950247e-06, "loss": 0.1278, "step": 46320 }, { "epoch": 33.93479853479853, "grad_norm": 1.171875, "learning_rate": 1.2508360690583248e-06, "loss": 0.1147, "step": 46321 }, { "epoch": 33.935531135531136, "grad_norm": 1.546875, "learning_rate": 1.2504916866905384e-06, "loss": 0.1274, "step": 46322 }, { "epoch": 33.93626373626373, "grad_norm": 0.94140625, "learning_rate": 1.2501475402933141e-06, "loss": 0.124, "step": 46323 }, { "epoch": 33.93699633699634, "grad_norm": 1.1484375, "learning_rate": 1.2498036298683007e-06, "loss": 0.0893, "step": 46324 }, { "epoch": 33.937728937728934, "grad_norm": 0.83203125, "learning_rate": 1.2494599554171465e-06, "loss": 0.1082, "step": 46325 }, { "epoch": 33.93846153846154, "grad_norm": 1.1015625, "learning_rate": 1.2491165169414951e-06, "loss": 0.1138, "step": 46326 }, { "epoch": 33.939194139194136, "grad_norm": 0.921875, "learning_rate": 1.248773314442995e-06, "loss": 0.109, "step": 46327 }, { "epoch": 33.93992673992674, "grad_norm": 0.859375, "learning_rate": 1.2484303479232895e-06, "loss": 0.1493, "step": 46328 }, { "epoch": 33.94065934065934, "grad_norm": 0.6484375, "learning_rate": 1.248087617384027e-06, "loss": 0.1128, "step": 46329 }, { "epoch": 33.94139194139194, "grad_norm": 0.78125, "learning_rate": 1.2477451228268453e-06, "loss": 0.1359, "step": 46330 }, { "epoch": 33.94212454212454, "grad_norm": 0.93359375, "learning_rate": 1.2474028642533824e-06, "loss": 0.1269, "step": 46331 }, { "epoch": 33.94285714285714, "grad_norm": 0.609375, "learning_rate": 1.2470608416652758e-06, "loss": 0.1164, "step": 46332 }, { "epoch": 33.94358974358974, "grad_norm": 0.8203125, "learning_rate": 1.2467190550641795e-06, "loss": 0.1315, "step": 46333 }, { "epoch": 33.944322344322345, "grad_norm": 0.92578125, "learning_rate": 1.2463775044517148e-06, "loss": 0.1214, "step": 46334 }, { "epoch": 33.94505494505494, "grad_norm": 1.1171875, "learning_rate": 1.2460361898295249e-06, "loss": 0.1021, "step": 46335 }, { "epoch": 33.945787545787546, "grad_norm": 1.109375, "learning_rate": 1.2456951111992475e-06, "loss": 0.1033, "step": 46336 }, { "epoch": 33.94652014652014, "grad_norm": 0.7265625, "learning_rate": 1.2453542685625096e-06, "loss": 0.1044, "step": 46337 }, { "epoch": 33.94725274725275, "grad_norm": 0.62109375, "learning_rate": 1.2450136619209483e-06, "loss": 0.1258, "step": 46338 }, { "epoch": 33.947985347985345, "grad_norm": 0.90234375, "learning_rate": 1.2446732912762017e-06, "loss": 0.1335, "step": 46339 }, { "epoch": 33.94871794871795, "grad_norm": 0.8125, "learning_rate": 1.2443331566298908e-06, "loss": 0.114, "step": 46340 }, { "epoch": 33.949450549450546, "grad_norm": 1.078125, "learning_rate": 1.2439932579836478e-06, "loss": 0.1359, "step": 46341 }, { "epoch": 33.95018315018315, "grad_norm": 1.2109375, "learning_rate": 1.2436535953391109e-06, "loss": 0.1103, "step": 46342 }, { "epoch": 33.95091575091575, "grad_norm": 1.4453125, "learning_rate": 1.2433141686978897e-06, "loss": 0.1437, "step": 46343 }, { "epoch": 33.95164835164835, "grad_norm": 0.6015625, "learning_rate": 1.242974978061622e-06, "loss": 0.1155, "step": 46344 }, { "epoch": 33.95238095238095, "grad_norm": 1.0859375, "learning_rate": 1.2426360234319346e-06, "loss": 0.1438, "step": 46345 }, { "epoch": 33.95311355311355, "grad_norm": 0.8671875, "learning_rate": 1.2422973048104484e-06, "loss": 0.1052, "step": 46346 }, { "epoch": 33.95384615384615, "grad_norm": 1.2890625, "learning_rate": 1.2419588221987853e-06, "loss": 0.1195, "step": 46347 }, { "epoch": 33.954578754578755, "grad_norm": 1.390625, "learning_rate": 1.2416205755985659e-06, "loss": 0.14, "step": 46348 }, { "epoch": 33.95531135531135, "grad_norm": 1.0390625, "learning_rate": 1.2412825650114117e-06, "loss": 0.1386, "step": 46349 }, { "epoch": 33.956043956043956, "grad_norm": 1.328125, "learning_rate": 1.240944790438949e-06, "loss": 0.1189, "step": 46350 }, { "epoch": 33.95677655677655, "grad_norm": 1.5625, "learning_rate": 1.2406072518827832e-06, "loss": 0.116, "step": 46351 }, { "epoch": 33.95750915750916, "grad_norm": 0.7578125, "learning_rate": 1.2402699493445458e-06, "loss": 0.0931, "step": 46352 }, { "epoch": 33.958241758241755, "grad_norm": 0.94921875, "learning_rate": 1.239932882825842e-06, "loss": 0.1163, "step": 46353 }, { "epoch": 33.95897435897436, "grad_norm": 0.91796875, "learning_rate": 1.2395960523282873e-06, "loss": 0.1049, "step": 46354 }, { "epoch": 33.959706959706956, "grad_norm": 1.0859375, "learning_rate": 1.2392594578535084e-06, "loss": 0.1251, "step": 46355 }, { "epoch": 33.96043956043956, "grad_norm": 1.1328125, "learning_rate": 1.2389230994031044e-06, "loss": 0.0991, "step": 46356 }, { "epoch": 33.96117216117216, "grad_norm": 0.71875, "learning_rate": 1.2385869769786914e-06, "loss": 0.1149, "step": 46357 }, { "epoch": 33.96190476190476, "grad_norm": 0.765625, "learning_rate": 1.2382510905818738e-06, "loss": 0.1327, "step": 46358 }, { "epoch": 33.96263736263736, "grad_norm": 1.140625, "learning_rate": 1.2379154402142728e-06, "loss": 0.0957, "step": 46359 }, { "epoch": 33.96336996336996, "grad_norm": 0.984375, "learning_rate": 1.2375800258774877e-06, "loss": 0.1143, "step": 46360 }, { "epoch": 33.96410256410256, "grad_norm": 1.0859375, "learning_rate": 1.2372448475731344e-06, "loss": 0.111, "step": 46361 }, { "epoch": 33.964835164835165, "grad_norm": 0.984375, "learning_rate": 1.2369099053028062e-06, "loss": 0.1169, "step": 46362 }, { "epoch": 33.96556776556776, "grad_norm": 0.6875, "learning_rate": 1.2365751990681189e-06, "loss": 0.093, "step": 46363 }, { "epoch": 33.966300366300366, "grad_norm": 1.1484375, "learning_rate": 1.2362407288706665e-06, "loss": 0.1146, "step": 46364 }, { "epoch": 33.967032967032964, "grad_norm": 0.98046875, "learning_rate": 1.2359064947120648e-06, "loss": 0.117, "step": 46365 }, { "epoch": 33.96776556776557, "grad_norm": 1.140625, "learning_rate": 1.2355724965939014e-06, "loss": 0.0981, "step": 46366 }, { "epoch": 33.968498168498165, "grad_norm": 1.1640625, "learning_rate": 1.2352387345177872e-06, "loss": 0.1239, "step": 46367 }, { "epoch": 33.96923076923077, "grad_norm": 0.94140625, "learning_rate": 1.2349052084853156e-06, "loss": 0.1054, "step": 46368 }, { "epoch": 33.96996336996337, "grad_norm": 1.2421875, "learning_rate": 1.2345719184980801e-06, "loss": 0.1195, "step": 46369 }, { "epoch": 33.97069597069597, "grad_norm": 1.0703125, "learning_rate": 1.2342388645576915e-06, "loss": 0.1234, "step": 46370 }, { "epoch": 33.97142857142857, "grad_norm": 0.890625, "learning_rate": 1.233906046665732e-06, "loss": 0.1258, "step": 46371 }, { "epoch": 33.97216117216117, "grad_norm": 1.1484375, "learning_rate": 1.2335734648238012e-06, "loss": 0.1058, "step": 46372 }, { "epoch": 33.97289377289377, "grad_norm": 1.4765625, "learning_rate": 1.233241119033498e-06, "loss": 0.1351, "step": 46373 }, { "epoch": 33.973626373626374, "grad_norm": 1.3828125, "learning_rate": 1.2329090092964053e-06, "loss": 0.1213, "step": 46374 }, { "epoch": 33.97435897435897, "grad_norm": 1.1953125, "learning_rate": 1.2325771356141224e-06, "loss": 0.1033, "step": 46375 }, { "epoch": 33.975091575091575, "grad_norm": 1.0390625, "learning_rate": 1.2322454979882319e-06, "loss": 0.117, "step": 46376 }, { "epoch": 33.97582417582417, "grad_norm": 0.9375, "learning_rate": 1.2319140964203273e-06, "loss": 0.1287, "step": 46377 }, { "epoch": 33.97655677655678, "grad_norm": 1.09375, "learning_rate": 1.2315829309119973e-06, "loss": 0.128, "step": 46378 }, { "epoch": 33.977289377289374, "grad_norm": 1.0, "learning_rate": 1.2312520014648299e-06, "loss": 0.1307, "step": 46379 }, { "epoch": 33.97802197802198, "grad_norm": 0.91796875, "learning_rate": 1.2309213080804022e-06, "loss": 0.1286, "step": 46380 }, { "epoch": 33.978754578754575, "grad_norm": 1.0859375, "learning_rate": 1.2305908507603084e-06, "loss": 0.1124, "step": 46381 }, { "epoch": 33.97948717948718, "grad_norm": 0.8984375, "learning_rate": 1.2302606295061309e-06, "loss": 0.1046, "step": 46382 }, { "epoch": 33.98021978021978, "grad_norm": 0.58203125, "learning_rate": 1.2299306443194467e-06, "loss": 0.0957, "step": 46383 }, { "epoch": 33.98095238095238, "grad_norm": 1.203125, "learning_rate": 1.2296008952018334e-06, "loss": 0.1267, "step": 46384 }, { "epoch": 33.98168498168498, "grad_norm": 0.85546875, "learning_rate": 1.2292713821548846e-06, "loss": 0.1253, "step": 46385 }, { "epoch": 33.98241758241758, "grad_norm": 1.6875, "learning_rate": 1.2289421051801666e-06, "loss": 0.1604, "step": 46386 }, { "epoch": 33.98315018315018, "grad_norm": 1.0859375, "learning_rate": 1.228613064279262e-06, "loss": 0.1263, "step": 46387 }, { "epoch": 33.983882783882784, "grad_norm": 1.1015625, "learning_rate": 1.2282842594537483e-06, "loss": 0.1452, "step": 46388 }, { "epoch": 33.98461538461538, "grad_norm": 0.71875, "learning_rate": 1.227955690705197e-06, "loss": 0.1113, "step": 46389 }, { "epoch": 33.985347985347985, "grad_norm": 0.9375, "learning_rate": 1.2276273580351856e-06, "loss": 0.1291, "step": 46390 }, { "epoch": 33.98608058608059, "grad_norm": 0.83984375, "learning_rate": 1.2272992614452853e-06, "loss": 0.1009, "step": 46391 }, { "epoch": 33.98681318681319, "grad_norm": 1.390625, "learning_rate": 1.226971400937074e-06, "loss": 0.1188, "step": 46392 }, { "epoch": 33.987545787545784, "grad_norm": 0.94921875, "learning_rate": 1.2266437765121174e-06, "loss": 0.1094, "step": 46393 }, { "epoch": 33.98827838827839, "grad_norm": 0.64453125, "learning_rate": 1.2263163881719822e-06, "loss": 0.1022, "step": 46394 }, { "epoch": 33.98901098901099, "grad_norm": 1.171875, "learning_rate": 1.2259892359182398e-06, "loss": 0.146, "step": 46395 }, { "epoch": 33.98974358974359, "grad_norm": 1.109375, "learning_rate": 1.2256623197524676e-06, "loss": 0.0831, "step": 46396 }, { "epoch": 33.99047619047619, "grad_norm": 0.84765625, "learning_rate": 1.2253356396762154e-06, "loss": 0.118, "step": 46397 }, { "epoch": 33.99120879120879, "grad_norm": 0.71875, "learning_rate": 1.2250091956910547e-06, "loss": 0.1104, "step": 46398 }, { "epoch": 33.991941391941396, "grad_norm": 0.921875, "learning_rate": 1.2246829877985576e-06, "loss": 0.1236, "step": 46399 }, { "epoch": 33.99267399267399, "grad_norm": 1.1640625, "learning_rate": 1.2243570160002736e-06, "loss": 0.1572, "step": 46400 }, { "epoch": 33.99340659340659, "grad_norm": 0.6796875, "learning_rate": 1.2240312802977745e-06, "loss": 0.1411, "step": 46401 }, { "epoch": 33.994139194139194, "grad_norm": 0.875, "learning_rate": 1.2237057806926157e-06, "loss": 0.1093, "step": 46402 }, { "epoch": 33.9948717948718, "grad_norm": 0.65625, "learning_rate": 1.223380517186363e-06, "loss": 0.1068, "step": 46403 }, { "epoch": 33.995604395604396, "grad_norm": 1.546875, "learning_rate": 1.2230554897805671e-06, "loss": 0.1427, "step": 46404 }, { "epoch": 33.99633699633699, "grad_norm": 1.0859375, "learning_rate": 1.2227306984767876e-06, "loss": 0.1159, "step": 46405 }, { "epoch": 33.9970695970696, "grad_norm": 0.890625, "learning_rate": 1.222406143276586e-06, "loss": 0.1193, "step": 46406 }, { "epoch": 33.9978021978022, "grad_norm": 0.96484375, "learning_rate": 1.2220818241815118e-06, "loss": 0.1099, "step": 46407 }, { "epoch": 33.9985347985348, "grad_norm": 1.125, "learning_rate": 1.2217577411931203e-06, "loss": 0.1233, "step": 46408 }, { "epoch": 33.999267399267396, "grad_norm": 0.796875, "learning_rate": 1.2214338943129668e-06, "loss": 0.1173, "step": 46409 }, { "epoch": 34.0, "grad_norm": 1.1875, "learning_rate": 1.2211102835425955e-06, "loss": 0.1282, "step": 46410 }, { "epoch": 34.000732600732604, "grad_norm": 0.9765625, "learning_rate": 1.2207869088835616e-06, "loss": 0.1093, "step": 46411 }, { "epoch": 34.0014652014652, "grad_norm": 1.0625, "learning_rate": 1.2204637703374205e-06, "loss": 0.1265, "step": 46412 }, { "epoch": 34.002197802197806, "grad_norm": 0.8984375, "learning_rate": 1.220140867905711e-06, "loss": 0.1134, "step": 46413 }, { "epoch": 34.0029304029304, "grad_norm": 0.6171875, "learning_rate": 1.2198182015899883e-06, "loss": 0.106, "step": 46414 }, { "epoch": 34.00366300366301, "grad_norm": 0.66796875, "learning_rate": 1.219495771391791e-06, "loss": 0.0978, "step": 46415 }, { "epoch": 34.004395604395604, "grad_norm": 0.9140625, "learning_rate": 1.2191735773126635e-06, "loss": 0.0918, "step": 46416 }, { "epoch": 34.00512820512821, "grad_norm": 0.890625, "learning_rate": 1.218851619354161e-06, "loss": 0.1175, "step": 46417 }, { "epoch": 34.005860805860806, "grad_norm": 1.1171875, "learning_rate": 1.2185298975178169e-06, "loss": 0.1225, "step": 46418 }, { "epoch": 34.00659340659341, "grad_norm": 1.109375, "learning_rate": 1.21820841180517e-06, "loss": 0.1281, "step": 46419 }, { "epoch": 34.00732600732601, "grad_norm": 0.83203125, "learning_rate": 1.2178871622177642e-06, "loss": 0.1047, "step": 46420 }, { "epoch": 34.00805860805861, "grad_norm": 1.5703125, "learning_rate": 1.2175661487571443e-06, "loss": 0.1229, "step": 46421 }, { "epoch": 34.00879120879121, "grad_norm": 0.83984375, "learning_rate": 1.2172453714248375e-06, "loss": 0.1056, "step": 46422 }, { "epoch": 34.00952380952381, "grad_norm": 0.91796875, "learning_rate": 1.2169248302223942e-06, "loss": 0.1052, "step": 46423 }, { "epoch": 34.01025641025641, "grad_norm": 0.76171875, "learning_rate": 1.2166045251513361e-06, "loss": 0.1377, "step": 46424 }, { "epoch": 34.010989010989015, "grad_norm": 1.0625, "learning_rate": 1.2162844562132078e-06, "loss": 0.1276, "step": 46425 }, { "epoch": 34.01172161172161, "grad_norm": 1.2578125, "learning_rate": 1.215964623409537e-06, "loss": 0.1144, "step": 46426 }, { "epoch": 34.012454212454216, "grad_norm": 0.84765625, "learning_rate": 1.215645026741857e-06, "loss": 0.1022, "step": 46427 }, { "epoch": 34.01318681318681, "grad_norm": 1.359375, "learning_rate": 1.2153256662117065e-06, "loss": 0.1313, "step": 46428 }, { "epoch": 34.01391941391942, "grad_norm": 1.109375, "learning_rate": 1.2150065418206078e-06, "loss": 0.1512, "step": 46429 }, { "epoch": 34.014652014652015, "grad_norm": 1.3046875, "learning_rate": 1.2146876535700885e-06, "loss": 0.1195, "step": 46430 }, { "epoch": 34.01538461538462, "grad_norm": 1.3203125, "learning_rate": 1.2143690014616824e-06, "loss": 0.1296, "step": 46431 }, { "epoch": 34.016117216117216, "grad_norm": 1.03125, "learning_rate": 1.214050585496917e-06, "loss": 0.1325, "step": 46432 }, { "epoch": 34.01684981684982, "grad_norm": 0.7734375, "learning_rate": 1.2137324056773086e-06, "loss": 0.1103, "step": 46433 }, { "epoch": 34.01758241758242, "grad_norm": 1.2265625, "learning_rate": 1.213414462004391e-06, "loss": 0.1339, "step": 46434 }, { "epoch": 34.01831501831502, "grad_norm": 1.09375, "learning_rate": 1.2130967544796866e-06, "loss": 0.1237, "step": 46435 }, { "epoch": 34.01904761904762, "grad_norm": 0.90625, "learning_rate": 1.2127792831047119e-06, "loss": 0.135, "step": 46436 }, { "epoch": 34.01978021978022, "grad_norm": 1.1484375, "learning_rate": 1.2124620478809946e-06, "loss": 0.12, "step": 46437 }, { "epoch": 34.02051282051282, "grad_norm": 0.828125, "learning_rate": 1.2121450488100461e-06, "loss": 0.1296, "step": 46438 }, { "epoch": 34.021245421245425, "grad_norm": 1.3046875, "learning_rate": 1.2118282858934002e-06, "loss": 0.1639, "step": 46439 }, { "epoch": 34.02197802197802, "grad_norm": 1.0625, "learning_rate": 1.2115117591325564e-06, "loss": 0.1289, "step": 46440 }, { "epoch": 34.022710622710626, "grad_norm": 0.8984375, "learning_rate": 1.2111954685290426e-06, "loss": 0.1107, "step": 46441 }, { "epoch": 34.02344322344322, "grad_norm": 0.77734375, "learning_rate": 1.210879414084376e-06, "loss": 0.1128, "step": 46442 }, { "epoch": 34.02417582417583, "grad_norm": 1.0390625, "learning_rate": 1.210563595800062e-06, "loss": 0.0928, "step": 46443 }, { "epoch": 34.024908424908425, "grad_norm": 0.87890625, "learning_rate": 1.2102480136776177e-06, "loss": 0.1114, "step": 46444 }, { "epoch": 34.02564102564103, "grad_norm": 1.0234375, "learning_rate": 1.20993266771856e-06, "loss": 0.1276, "step": 46445 }, { "epoch": 34.026373626373626, "grad_norm": 1.28125, "learning_rate": 1.209617557924394e-06, "loss": 0.1404, "step": 46446 }, { "epoch": 34.02710622710623, "grad_norm": 1.15625, "learning_rate": 1.209302684296626e-06, "loss": 0.133, "step": 46447 }, { "epoch": 34.02783882783883, "grad_norm": 1.2109375, "learning_rate": 1.2089880468367786e-06, "loss": 0.1454, "step": 46448 }, { "epoch": 34.02857142857143, "grad_norm": 0.984375, "learning_rate": 1.2086736455463458e-06, "loss": 0.1074, "step": 46449 }, { "epoch": 34.02930402930403, "grad_norm": 1.15625, "learning_rate": 1.208359480426845e-06, "loss": 0.1123, "step": 46450 }, { "epoch": 34.03003663003663, "grad_norm": 1.0703125, "learning_rate": 1.2080455514797653e-06, "loss": 0.1424, "step": 46451 }, { "epoch": 34.03076923076923, "grad_norm": 1.0, "learning_rate": 1.207731858706629e-06, "loss": 0.1082, "step": 46452 }, { "epoch": 34.031501831501835, "grad_norm": 0.96484375, "learning_rate": 1.2074184021089254e-06, "loss": 0.1084, "step": 46453 }, { "epoch": 34.03223443223443, "grad_norm": 0.98828125, "learning_rate": 1.2071051816881657e-06, "loss": 0.1262, "step": 46454 }, { "epoch": 34.032967032967036, "grad_norm": 1.9921875, "learning_rate": 1.2067921974458451e-06, "loss": 0.1274, "step": 46455 }, { "epoch": 34.033699633699634, "grad_norm": 1.1796875, "learning_rate": 1.2064794493834634e-06, "loss": 0.1031, "step": 46456 }, { "epoch": 34.03443223443224, "grad_norm": 1.40625, "learning_rate": 1.2061669375025264e-06, "loss": 0.1541, "step": 46457 }, { "epoch": 34.035164835164835, "grad_norm": 1.1328125, "learning_rate": 1.2058546618045184e-06, "loss": 0.1244, "step": 46458 }, { "epoch": 34.03589743589744, "grad_norm": 1.3828125, "learning_rate": 1.2055426222909505e-06, "loss": 0.1131, "step": 46459 }, { "epoch": 34.03663003663004, "grad_norm": 1.4609375, "learning_rate": 1.2052308189633061e-06, "loss": 0.1382, "step": 46460 }, { "epoch": 34.03736263736264, "grad_norm": 0.88671875, "learning_rate": 1.2049192518230807e-06, "loss": 0.1129, "step": 46461 }, { "epoch": 34.03809523809524, "grad_norm": 0.984375, "learning_rate": 1.204607920871774e-06, "loss": 0.1158, "step": 46462 }, { "epoch": 34.03882783882784, "grad_norm": 0.83203125, "learning_rate": 1.2042968261108698e-06, "loss": 0.0994, "step": 46463 }, { "epoch": 34.03956043956044, "grad_norm": 1.71875, "learning_rate": 1.2039859675418636e-06, "loss": 0.1035, "step": 46464 }, { "epoch": 34.040293040293044, "grad_norm": 0.640625, "learning_rate": 1.2036753451662493e-06, "loss": 0.1015, "step": 46465 }, { "epoch": 34.04102564102564, "grad_norm": 1.0859375, "learning_rate": 1.2033649589855005e-06, "loss": 0.1158, "step": 46466 }, { "epoch": 34.041758241758245, "grad_norm": 1.3046875, "learning_rate": 1.2030548090011169e-06, "loss": 0.1255, "step": 46467 }, { "epoch": 34.04249084249084, "grad_norm": 1.3984375, "learning_rate": 1.2027448952145827e-06, "loss": 0.1318, "step": 46468 }, { "epoch": 34.04322344322345, "grad_norm": 1.109375, "learning_rate": 1.2024352176273762e-06, "loss": 0.1048, "step": 46469 }, { "epoch": 34.043956043956044, "grad_norm": 1.203125, "learning_rate": 1.202125776240992e-06, "loss": 0.1063, "step": 46470 }, { "epoch": 34.04468864468865, "grad_norm": 1.4375, "learning_rate": 1.2018165710569032e-06, "loss": 0.1368, "step": 46471 }, { "epoch": 34.045421245421245, "grad_norm": 0.8671875, "learning_rate": 1.2015076020765933e-06, "loss": 0.1099, "step": 46472 }, { "epoch": 34.04615384615385, "grad_norm": 0.75, "learning_rate": 1.2011988693015463e-06, "loss": 0.1323, "step": 46473 }, { "epoch": 34.04688644688645, "grad_norm": 0.8671875, "learning_rate": 1.2008903727332402e-06, "loss": 0.1185, "step": 46474 }, { "epoch": 34.04761904761905, "grad_norm": 1.65625, "learning_rate": 1.2005821123731484e-06, "loss": 0.1491, "step": 46475 }, { "epoch": 34.04835164835165, "grad_norm": 1.1171875, "learning_rate": 1.2002740882227542e-06, "loss": 0.1483, "step": 46476 }, { "epoch": 34.04908424908425, "grad_norm": 0.84765625, "learning_rate": 1.1999663002835308e-06, "loss": 0.1049, "step": 46477 }, { "epoch": 34.04981684981685, "grad_norm": 0.9375, "learning_rate": 1.1996587485569508e-06, "loss": 0.1241, "step": 46478 }, { "epoch": 34.050549450549454, "grad_norm": 1.28125, "learning_rate": 1.1993514330444925e-06, "loss": 0.1482, "step": 46479 }, { "epoch": 34.05128205128205, "grad_norm": 0.68359375, "learning_rate": 1.1990443537476234e-06, "loss": 0.1115, "step": 46480 }, { "epoch": 34.052014652014655, "grad_norm": 0.796875, "learning_rate": 1.1987375106678215e-06, "loss": 0.1402, "step": 46481 }, { "epoch": 34.05274725274725, "grad_norm": 1.0234375, "learning_rate": 1.198430903806549e-06, "loss": 0.1057, "step": 46482 }, { "epoch": 34.05347985347986, "grad_norm": 0.7421875, "learning_rate": 1.1981245331652785e-06, "loss": 0.1194, "step": 46483 }, { "epoch": 34.054212454212454, "grad_norm": 1.265625, "learning_rate": 1.197818398745483e-06, "loss": 0.1429, "step": 46484 }, { "epoch": 34.05494505494506, "grad_norm": 0.83203125, "learning_rate": 1.1975125005486184e-06, "loss": 0.0917, "step": 46485 }, { "epoch": 34.055677655677655, "grad_norm": 0.69140625, "learning_rate": 1.1972068385761581e-06, "loss": 0.1108, "step": 46486 }, { "epoch": 34.05641025641026, "grad_norm": 1.1015625, "learning_rate": 1.1969014128295692e-06, "loss": 0.1235, "step": 46487 }, { "epoch": 34.05714285714286, "grad_norm": 0.98828125, "learning_rate": 1.1965962233103023e-06, "loss": 0.1181, "step": 46488 }, { "epoch": 34.05787545787546, "grad_norm": 0.84765625, "learning_rate": 1.196291270019836e-06, "loss": 0.1059, "step": 46489 }, { "epoch": 34.05860805860806, "grad_norm": 1.015625, "learning_rate": 1.1959865529596214e-06, "loss": 0.104, "step": 46490 }, { "epoch": 34.05934065934066, "grad_norm": 0.71875, "learning_rate": 1.1956820721311195e-06, "loss": 0.105, "step": 46491 }, { "epoch": 34.06007326007326, "grad_norm": 1.34375, "learning_rate": 1.195377827535793e-06, "loss": 0.1181, "step": 46492 }, { "epoch": 34.060805860805864, "grad_norm": 1.359375, "learning_rate": 1.195073819175092e-06, "loss": 0.1102, "step": 46493 }, { "epoch": 34.06153846153846, "grad_norm": 1.8515625, "learning_rate": 1.1947700470504845e-06, "loss": 0.1526, "step": 46494 }, { "epoch": 34.062271062271066, "grad_norm": 0.8359375, "learning_rate": 1.1944665111634152e-06, "loss": 0.1222, "step": 46495 }, { "epoch": 34.06300366300366, "grad_norm": 1.2890625, "learning_rate": 1.194163211515341e-06, "loss": 0.119, "step": 46496 }, { "epoch": 34.06373626373627, "grad_norm": 0.88671875, "learning_rate": 1.193860148107718e-06, "loss": 0.115, "step": 46497 }, { "epoch": 34.064468864468864, "grad_norm": 0.796875, "learning_rate": 1.1935573209420026e-06, "loss": 0.1361, "step": 46498 }, { "epoch": 34.06520146520147, "grad_norm": 1.1171875, "learning_rate": 1.1932547300196344e-06, "loss": 0.1178, "step": 46499 }, { "epoch": 34.065934065934066, "grad_norm": 0.53125, "learning_rate": 1.19295237534207e-06, "loss": 0.1263, "step": 46500 }, { "epoch": 34.06666666666667, "grad_norm": 1.484375, "learning_rate": 1.1926502569107603e-06, "loss": 0.1247, "step": 46501 }, { "epoch": 34.06739926739927, "grad_norm": 1.0390625, "learning_rate": 1.1923483747271502e-06, "loss": 0.1324, "step": 46502 }, { "epoch": 34.06813186813187, "grad_norm": 1.1328125, "learning_rate": 1.19204672879268e-06, "loss": 0.0977, "step": 46503 }, { "epoch": 34.06886446886447, "grad_norm": 0.8125, "learning_rate": 1.191745319108806e-06, "loss": 0.1055, "step": 46504 }, { "epoch": 34.06959706959707, "grad_norm": 1.1953125, "learning_rate": 1.1914441456769622e-06, "loss": 0.1212, "step": 46505 }, { "epoch": 34.07032967032967, "grad_norm": 0.8203125, "learning_rate": 1.1911432084986e-06, "loss": 0.1199, "step": 46506 }, { "epoch": 34.071062271062274, "grad_norm": 1.3984375, "learning_rate": 1.1908425075751587e-06, "loss": 0.1304, "step": 46507 }, { "epoch": 34.07179487179487, "grad_norm": 1.140625, "learning_rate": 1.1905420429080728e-06, "loss": 0.1246, "step": 46508 }, { "epoch": 34.072527472527476, "grad_norm": 0.87890625, "learning_rate": 1.1902418144987933e-06, "loss": 0.1127, "step": 46509 }, { "epoch": 34.07326007326007, "grad_norm": 1.6796875, "learning_rate": 1.1899418223487492e-06, "loss": 0.1135, "step": 46510 }, { "epoch": 34.07399267399268, "grad_norm": 1.0078125, "learning_rate": 1.1896420664593798e-06, "loss": 0.1411, "step": 46511 }, { "epoch": 34.074725274725274, "grad_norm": 0.8359375, "learning_rate": 1.1893425468321255e-06, "loss": 0.1227, "step": 46512 }, { "epoch": 34.07545787545788, "grad_norm": 1.03125, "learning_rate": 1.1890432634684203e-06, "loss": 0.1256, "step": 46513 }, { "epoch": 34.076190476190476, "grad_norm": 0.92578125, "learning_rate": 1.1887442163696933e-06, "loss": 0.1111, "step": 46514 }, { "epoch": 34.07692307692308, "grad_norm": 1.1171875, "learning_rate": 1.1884454055373842e-06, "loss": 0.114, "step": 46515 }, { "epoch": 34.07765567765568, "grad_norm": 0.96484375, "learning_rate": 1.1881468309729162e-06, "loss": 0.1285, "step": 46516 }, { "epoch": 34.07838827838828, "grad_norm": 0.83203125, "learning_rate": 1.1878484926777295e-06, "loss": 0.126, "step": 46517 }, { "epoch": 34.07912087912088, "grad_norm": 0.77734375, "learning_rate": 1.1875503906532474e-06, "loss": 0.1356, "step": 46518 }, { "epoch": 34.07985347985348, "grad_norm": 0.86328125, "learning_rate": 1.1872525249008985e-06, "loss": 0.1179, "step": 46519 }, { "epoch": 34.08058608058608, "grad_norm": 0.734375, "learning_rate": 1.1869548954221064e-06, "loss": 0.0901, "step": 46520 }, { "epoch": 34.081318681318685, "grad_norm": 1.3203125, "learning_rate": 1.1866575022183107e-06, "loss": 0.1223, "step": 46521 }, { "epoch": 34.08205128205128, "grad_norm": 0.859375, "learning_rate": 1.186360345290924e-06, "loss": 0.1304, "step": 46522 }, { "epoch": 34.082783882783886, "grad_norm": 0.88671875, "learning_rate": 1.1860634246413698e-06, "loss": 0.0963, "step": 46523 }, { "epoch": 34.08351648351648, "grad_norm": 0.86328125, "learning_rate": 1.185766740271082e-06, "loss": 0.122, "step": 46524 }, { "epoch": 34.08424908424909, "grad_norm": 1.28125, "learning_rate": 1.1854702921814678e-06, "loss": 0.1142, "step": 46525 }, { "epoch": 34.084981684981685, "grad_norm": 0.8984375, "learning_rate": 1.185174080373956e-06, "loss": 0.121, "step": 46526 }, { "epoch": 34.08571428571429, "grad_norm": 1.7109375, "learning_rate": 1.1848781048499646e-06, "loss": 0.152, "step": 46527 }, { "epoch": 34.086446886446886, "grad_norm": 0.984375, "learning_rate": 1.1845823656109112e-06, "loss": 0.1191, "step": 46528 }, { "epoch": 34.08717948717949, "grad_norm": 0.87890625, "learning_rate": 1.1842868626582139e-06, "loss": 0.1046, "step": 46529 }, { "epoch": 34.08791208791209, "grad_norm": 0.76953125, "learning_rate": 1.1839915959932847e-06, "loss": 0.1122, "step": 46530 }, { "epoch": 34.08864468864469, "grad_norm": 0.921875, "learning_rate": 1.1836965656175418e-06, "loss": 0.1043, "step": 46531 }, { "epoch": 34.08937728937729, "grad_norm": 1.171875, "learning_rate": 1.1834017715324031e-06, "loss": 0.1278, "step": 46532 }, { "epoch": 34.09010989010989, "grad_norm": 1.0234375, "learning_rate": 1.1831072137392698e-06, "loss": 0.1494, "step": 46533 }, { "epoch": 34.09084249084249, "grad_norm": 1.2265625, "learning_rate": 1.1828128922395654e-06, "loss": 0.1074, "step": 46534 }, { "epoch": 34.091575091575095, "grad_norm": 0.9140625, "learning_rate": 1.182518807034691e-06, "loss": 0.1257, "step": 46535 }, { "epoch": 34.09230769230769, "grad_norm": 1.203125, "learning_rate": 1.1822249581260595e-06, "loss": 0.0926, "step": 46536 }, { "epoch": 34.093040293040296, "grad_norm": 1.84375, "learning_rate": 1.1819313455150769e-06, "loss": 0.1112, "step": 46537 }, { "epoch": 34.09377289377289, "grad_norm": 1.1015625, "learning_rate": 1.1816379692031511e-06, "loss": 0.1277, "step": 46538 }, { "epoch": 34.0945054945055, "grad_norm": 0.94921875, "learning_rate": 1.1813448291916883e-06, "loss": 0.1245, "step": 46539 }, { "epoch": 34.095238095238095, "grad_norm": 0.88671875, "learning_rate": 1.1810519254820903e-06, "loss": 0.1392, "step": 46540 }, { "epoch": 34.0959706959707, "grad_norm": 1.265625, "learning_rate": 1.180759258075769e-06, "loss": 0.1219, "step": 46541 }, { "epoch": 34.096703296703296, "grad_norm": 0.9296875, "learning_rate": 1.1804668269741151e-06, "loss": 0.1162, "step": 46542 }, { "epoch": 34.0974358974359, "grad_norm": 1.21875, "learning_rate": 1.180174632178541e-06, "loss": 0.1041, "step": 46543 }, { "epoch": 34.0981684981685, "grad_norm": 0.71484375, "learning_rate": 1.1798826736904367e-06, "loss": 0.0998, "step": 46544 }, { "epoch": 34.0989010989011, "grad_norm": 1.6484375, "learning_rate": 1.1795909515112043e-06, "loss": 0.1402, "step": 46545 }, { "epoch": 34.0996336996337, "grad_norm": 1.0, "learning_rate": 1.1792994656422445e-06, "loss": 0.1229, "step": 46546 }, { "epoch": 34.100366300366304, "grad_norm": 1.0625, "learning_rate": 1.1790082160849534e-06, "loss": 0.112, "step": 46547 }, { "epoch": 34.1010989010989, "grad_norm": 1.78125, "learning_rate": 1.1787172028407215e-06, "loss": 0.1584, "step": 46548 }, { "epoch": 34.101831501831505, "grad_norm": 1.6484375, "learning_rate": 1.1784264259109501e-06, "loss": 0.135, "step": 46549 }, { "epoch": 34.1025641025641, "grad_norm": 1.03125, "learning_rate": 1.178135885297024e-06, "loss": 0.1275, "step": 46550 }, { "epoch": 34.10329670329671, "grad_norm": 1.78125, "learning_rate": 1.1778455810003448e-06, "loss": 0.1427, "step": 46551 }, { "epoch": 34.104029304029304, "grad_norm": 1.21875, "learning_rate": 1.1775555130222971e-06, "loss": 0.1125, "step": 46552 }, { "epoch": 34.10476190476191, "grad_norm": 1.1484375, "learning_rate": 1.1772656813642717e-06, "loss": 0.1082, "step": 46553 }, { "epoch": 34.105494505494505, "grad_norm": 0.57421875, "learning_rate": 1.1769760860276586e-06, "loss": 0.1149, "step": 46554 }, { "epoch": 34.10622710622711, "grad_norm": 0.99609375, "learning_rate": 1.1766867270138483e-06, "loss": 0.1319, "step": 46555 }, { "epoch": 34.10695970695971, "grad_norm": 1.390625, "learning_rate": 1.1763976043242204e-06, "loss": 0.0923, "step": 46556 }, { "epoch": 34.10769230769231, "grad_norm": 1.0546875, "learning_rate": 1.1761087179601593e-06, "loss": 0.1416, "step": 46557 }, { "epoch": 34.10842490842491, "grad_norm": 0.7734375, "learning_rate": 1.1758200679230557e-06, "loss": 0.1108, "step": 46558 }, { "epoch": 34.10915750915751, "grad_norm": 0.83203125, "learning_rate": 1.1755316542142947e-06, "loss": 0.1024, "step": 46559 }, { "epoch": 34.10989010989011, "grad_norm": 0.8046875, "learning_rate": 1.17524347683525e-06, "loss": 0.1207, "step": 46560 }, { "epoch": 34.110622710622714, "grad_norm": 1.109375, "learning_rate": 1.1749555357873065e-06, "loss": 0.1284, "step": 46561 }, { "epoch": 34.11135531135531, "grad_norm": 0.76953125, "learning_rate": 1.1746678310718437e-06, "loss": 0.1112, "step": 46562 }, { "epoch": 34.112087912087915, "grad_norm": 1.1875, "learning_rate": 1.1743803626902406e-06, "loss": 0.1107, "step": 46563 }, { "epoch": 34.11282051282051, "grad_norm": 1.21875, "learning_rate": 1.1740931306438714e-06, "loss": 0.1373, "step": 46564 }, { "epoch": 34.11355311355312, "grad_norm": 1.0859375, "learning_rate": 1.173806134934121e-06, "loss": 0.1247, "step": 46565 }, { "epoch": 34.114285714285714, "grad_norm": 0.6796875, "learning_rate": 1.1735193755623523e-06, "loss": 0.1052, "step": 46566 }, { "epoch": 34.11501831501832, "grad_norm": 0.8984375, "learning_rate": 1.1732328525299449e-06, "loss": 0.1196, "step": 46567 }, { "epoch": 34.115750915750915, "grad_norm": 0.68359375, "learning_rate": 1.1729465658382724e-06, "loss": 0.1119, "step": 46568 }, { "epoch": 34.11648351648352, "grad_norm": 1.09375, "learning_rate": 1.1726605154887088e-06, "loss": 0.1307, "step": 46569 }, { "epoch": 34.11721611721612, "grad_norm": 0.92578125, "learning_rate": 1.1723747014826167e-06, "loss": 0.1172, "step": 46570 }, { "epoch": 34.11794871794872, "grad_norm": 1.1171875, "learning_rate": 1.1720891238213761e-06, "loss": 0.1424, "step": 46571 }, { "epoch": 34.11868131868132, "grad_norm": 1.0234375, "learning_rate": 1.171803782506344e-06, "loss": 0.1256, "step": 46572 }, { "epoch": 34.11941391941392, "grad_norm": 1.8125, "learning_rate": 1.1715186775389e-06, "loss": 0.1382, "step": 46573 }, { "epoch": 34.12014652014652, "grad_norm": 0.8984375, "learning_rate": 1.1712338089204012e-06, "loss": 0.1011, "step": 46574 }, { "epoch": 34.120879120879124, "grad_norm": 0.953125, "learning_rate": 1.1709491766522107e-06, "loss": 0.1289, "step": 46575 }, { "epoch": 34.12161172161172, "grad_norm": 1.328125, "learning_rate": 1.1706647807357025e-06, "loss": 0.0989, "step": 46576 }, { "epoch": 34.122344322344325, "grad_norm": 0.98828125, "learning_rate": 1.1703806211722286e-06, "loss": 0.1102, "step": 46577 }, { "epoch": 34.12307692307692, "grad_norm": 1.140625, "learning_rate": 1.1700966979631572e-06, "loss": 0.1201, "step": 46578 }, { "epoch": 34.12380952380953, "grad_norm": 1.046875, "learning_rate": 1.1698130111098456e-06, "loss": 0.1082, "step": 46579 }, { "epoch": 34.124542124542124, "grad_norm": 1.03125, "learning_rate": 1.1695295606136517e-06, "loss": 0.1261, "step": 46580 }, { "epoch": 34.12527472527473, "grad_norm": 0.82421875, "learning_rate": 1.1692463464759378e-06, "loss": 0.1222, "step": 46581 }, { "epoch": 34.126007326007326, "grad_norm": 1.3828125, "learning_rate": 1.1689633686980617e-06, "loss": 0.114, "step": 46582 }, { "epoch": 34.12673992673993, "grad_norm": 1.09375, "learning_rate": 1.1686806272813696e-06, "loss": 0.132, "step": 46583 }, { "epoch": 34.12747252747253, "grad_norm": 1.65625, "learning_rate": 1.1683981222272302e-06, "loss": 0.1419, "step": 46584 }, { "epoch": 34.12820512820513, "grad_norm": 0.890625, "learning_rate": 1.168115853536984e-06, "loss": 0.1249, "step": 46585 }, { "epoch": 34.12893772893773, "grad_norm": 1.5390625, "learning_rate": 1.1678338212119946e-06, "loss": 0.1079, "step": 46586 }, { "epoch": 34.12967032967033, "grad_norm": 0.6171875, "learning_rate": 1.1675520252536023e-06, "loss": 0.1205, "step": 46587 }, { "epoch": 34.13040293040293, "grad_norm": 1.0, "learning_rate": 1.1672704656631702e-06, "loss": 0.1003, "step": 46588 }, { "epoch": 34.131135531135534, "grad_norm": 1.015625, "learning_rate": 1.1669891424420337e-06, "loss": 0.1012, "step": 46589 }, { "epoch": 34.13186813186813, "grad_norm": 1.1640625, "learning_rate": 1.1667080555915501e-06, "loss": 0.1007, "step": 46590 }, { "epoch": 34.132600732600736, "grad_norm": 0.75, "learning_rate": 1.1664272051130663e-06, "loss": 0.1337, "step": 46591 }, { "epoch": 34.13333333333333, "grad_norm": 1.046875, "learning_rate": 1.166146591007917e-06, "loss": 0.1211, "step": 46592 }, { "epoch": 34.13406593406594, "grad_norm": 1.2890625, "learning_rate": 1.1658662132774605e-06, "loss": 0.1442, "step": 46593 }, { "epoch": 34.134798534798534, "grad_norm": 1.109375, "learning_rate": 1.1655860719230369e-06, "loss": 0.1203, "step": 46594 }, { "epoch": 34.13553113553114, "grad_norm": 0.9140625, "learning_rate": 1.165306166945982e-06, "loss": 0.1351, "step": 46595 }, { "epoch": 34.136263736263736, "grad_norm": 0.79296875, "learning_rate": 1.1650264983476423e-06, "loss": 0.1145, "step": 46596 }, { "epoch": 34.13699633699634, "grad_norm": 1.25, "learning_rate": 1.164747066129353e-06, "loss": 0.1335, "step": 46597 }, { "epoch": 34.13772893772894, "grad_norm": 0.9375, "learning_rate": 1.1644678702924554e-06, "loss": 0.1202, "step": 46598 }, { "epoch": 34.13846153846154, "grad_norm": 1.390625, "learning_rate": 1.16418891083829e-06, "loss": 0.1183, "step": 46599 }, { "epoch": 34.13919413919414, "grad_norm": 1.390625, "learning_rate": 1.1639101877681926e-06, "loss": 0.1488, "step": 46600 }, { "epoch": 34.13992673992674, "grad_norm": 1.1015625, "learning_rate": 1.1636317010834983e-06, "loss": 0.1192, "step": 46601 }, { "epoch": 34.14065934065934, "grad_norm": 1.03125, "learning_rate": 1.1633534507855427e-06, "loss": 0.093, "step": 46602 }, { "epoch": 34.141391941391944, "grad_norm": 1.1640625, "learning_rate": 1.1630754368756501e-06, "loss": 0.1327, "step": 46603 }, { "epoch": 34.14212454212454, "grad_norm": 0.73046875, "learning_rate": 1.1627976593551617e-06, "loss": 0.1088, "step": 46604 }, { "epoch": 34.142857142857146, "grad_norm": 0.859375, "learning_rate": 1.1625201182254073e-06, "loss": 0.1007, "step": 46605 }, { "epoch": 34.14358974358974, "grad_norm": 1.0, "learning_rate": 1.162242813487717e-06, "loss": 0.1729, "step": 46606 }, { "epoch": 34.14432234432235, "grad_norm": 1.8203125, "learning_rate": 1.1619657451434147e-06, "loss": 0.1446, "step": 46607 }, { "epoch": 34.145054945054945, "grad_norm": 0.90625, "learning_rate": 1.161688913193831e-06, "loss": 0.123, "step": 46608 }, { "epoch": 34.14578754578755, "grad_norm": 1.2890625, "learning_rate": 1.1614123176402954e-06, "loss": 0.1431, "step": 46609 }, { "epoch": 34.146520146520146, "grad_norm": 0.875, "learning_rate": 1.161135958484127e-06, "loss": 0.1146, "step": 46610 }, { "epoch": 34.14725274725275, "grad_norm": 0.890625, "learning_rate": 1.1608598357266557e-06, "loss": 0.1048, "step": 46611 }, { "epoch": 34.14798534798535, "grad_norm": 1.1640625, "learning_rate": 1.1605839493692003e-06, "loss": 0.1271, "step": 46612 }, { "epoch": 34.14871794871795, "grad_norm": 0.94921875, "learning_rate": 1.1603082994130858e-06, "loss": 0.1139, "step": 46613 }, { "epoch": 34.14945054945055, "grad_norm": 1.3671875, "learning_rate": 1.1600328858596303e-06, "loss": 0.1226, "step": 46614 }, { "epoch": 34.15018315018315, "grad_norm": 0.98046875, "learning_rate": 1.159757708710159e-06, "loss": 0.1184, "step": 46615 }, { "epoch": 34.15091575091575, "grad_norm": 1.21875, "learning_rate": 1.1594827679659794e-06, "loss": 0.1221, "step": 46616 }, { "epoch": 34.151648351648355, "grad_norm": 1.4375, "learning_rate": 1.1592080636284163e-06, "loss": 0.1318, "step": 46617 }, { "epoch": 34.15238095238095, "grad_norm": 1.2578125, "learning_rate": 1.158933595698788e-06, "loss": 0.1165, "step": 46618 }, { "epoch": 34.153113553113556, "grad_norm": 0.8671875, "learning_rate": 1.1586593641784031e-06, "loss": 0.1241, "step": 46619 }, { "epoch": 34.15384615384615, "grad_norm": 0.6484375, "learning_rate": 1.1583853690685806e-06, "loss": 0.0989, "step": 46620 }, { "epoch": 34.15457875457876, "grad_norm": 1.1640625, "learning_rate": 1.1581116103706334e-06, "loss": 0.1355, "step": 46621 }, { "epoch": 34.155311355311355, "grad_norm": 0.94921875, "learning_rate": 1.15783808808587e-06, "loss": 0.0898, "step": 46622 }, { "epoch": 34.15604395604396, "grad_norm": 1.1328125, "learning_rate": 1.1575648022155976e-06, "loss": 0.1106, "step": 46623 }, { "epoch": 34.156776556776556, "grad_norm": 0.97265625, "learning_rate": 1.1572917527611358e-06, "loss": 0.1152, "step": 46624 }, { "epoch": 34.15750915750916, "grad_norm": 1.1640625, "learning_rate": 1.157018939723787e-06, "loss": 0.1219, "step": 46625 }, { "epoch": 34.15824175824176, "grad_norm": 0.63671875, "learning_rate": 1.156746363104859e-06, "loss": 0.1303, "step": 46626 }, { "epoch": 34.15897435897436, "grad_norm": 1.0078125, "learning_rate": 1.1564740229056545e-06, "loss": 0.1167, "step": 46627 }, { "epoch": 34.15970695970696, "grad_norm": 0.796875, "learning_rate": 1.1562019191274864e-06, "loss": 0.1146, "step": 46628 }, { "epoch": 34.16043956043956, "grad_norm": 1.5859375, "learning_rate": 1.1559300517716467e-06, "loss": 0.1049, "step": 46629 }, { "epoch": 34.16117216117216, "grad_norm": 1.25, "learning_rate": 1.1556584208394486e-06, "loss": 0.1196, "step": 46630 }, { "epoch": 34.161904761904765, "grad_norm": 0.8828125, "learning_rate": 1.155387026332189e-06, "loss": 0.1112, "step": 46631 }, { "epoch": 34.16263736263736, "grad_norm": 0.86328125, "learning_rate": 1.1551158682511655e-06, "loss": 0.111, "step": 46632 }, { "epoch": 34.163369963369966, "grad_norm": 0.97265625, "learning_rate": 1.154844946597685e-06, "loss": 0.1053, "step": 46633 }, { "epoch": 34.16410256410256, "grad_norm": 0.74609375, "learning_rate": 1.1545742613730452e-06, "loss": 0.1115, "step": 46634 }, { "epoch": 34.16483516483517, "grad_norm": 1.328125, "learning_rate": 1.154303812578532e-06, "loss": 0.1278, "step": 46635 }, { "epoch": 34.165567765567765, "grad_norm": 0.92578125, "learning_rate": 1.1540336002154475e-06, "loss": 0.1403, "step": 46636 }, { "epoch": 34.16630036630037, "grad_norm": 1.203125, "learning_rate": 1.1537636242850946e-06, "loss": 0.134, "step": 46637 }, { "epoch": 34.167032967032966, "grad_norm": 1.125, "learning_rate": 1.1534938847887536e-06, "loss": 0.1352, "step": 46638 }, { "epoch": 34.16776556776557, "grad_norm": 1.296875, "learning_rate": 1.1532243817277267e-06, "loss": 0.1038, "step": 46639 }, { "epoch": 34.16849816849817, "grad_norm": 1.1640625, "learning_rate": 1.1529551151033003e-06, "loss": 0.1368, "step": 46640 }, { "epoch": 34.16923076923077, "grad_norm": 1.078125, "learning_rate": 1.1526860849167653e-06, "loss": 0.0984, "step": 46641 }, { "epoch": 34.16996336996337, "grad_norm": 1.3671875, "learning_rate": 1.1524172911694084e-06, "loss": 0.1273, "step": 46642 }, { "epoch": 34.170695970695974, "grad_norm": 0.9765625, "learning_rate": 1.1521487338625205e-06, "loss": 0.1311, "step": 46643 }, { "epoch": 34.17142857142857, "grad_norm": 1.0625, "learning_rate": 1.151880412997388e-06, "loss": 0.0964, "step": 46644 }, { "epoch": 34.172161172161175, "grad_norm": 0.9609375, "learning_rate": 1.1516123285752965e-06, "loss": 0.0838, "step": 46645 }, { "epoch": 34.17289377289377, "grad_norm": 0.98046875, "learning_rate": 1.151344480597532e-06, "loss": 0.1361, "step": 46646 }, { "epoch": 34.17362637362638, "grad_norm": 1.1015625, "learning_rate": 1.1510768690653754e-06, "loss": 0.1142, "step": 46647 }, { "epoch": 34.174358974358974, "grad_norm": 1.5078125, "learning_rate": 1.150809493980112e-06, "loss": 0.1419, "step": 46648 }, { "epoch": 34.17509157509158, "grad_norm": 1.265625, "learning_rate": 1.1505423553430175e-06, "loss": 0.1504, "step": 46649 }, { "epoch": 34.175824175824175, "grad_norm": 0.91796875, "learning_rate": 1.1502754531553774e-06, "loss": 0.1281, "step": 46650 }, { "epoch": 34.17655677655678, "grad_norm": 0.984375, "learning_rate": 1.1500087874184667e-06, "loss": 0.1129, "step": 46651 }, { "epoch": 34.17728937728938, "grad_norm": 1.0390625, "learning_rate": 1.149742358133566e-06, "loss": 0.1234, "step": 46652 }, { "epoch": 34.17802197802198, "grad_norm": 1.0390625, "learning_rate": 1.1494761653019448e-06, "loss": 0.1233, "step": 46653 }, { "epoch": 34.17875457875458, "grad_norm": 1.2734375, "learning_rate": 1.1492102089248892e-06, "loss": 0.1212, "step": 46654 }, { "epoch": 34.17948717948718, "grad_norm": 1.375, "learning_rate": 1.1489444890036681e-06, "loss": 0.1038, "step": 46655 }, { "epoch": 34.18021978021978, "grad_norm": 0.78515625, "learning_rate": 1.1486790055395516e-06, "loss": 0.124, "step": 46656 }, { "epoch": 34.180952380952384, "grad_norm": 1.453125, "learning_rate": 1.1484137585338146e-06, "loss": 0.1274, "step": 46657 }, { "epoch": 34.18168498168498, "grad_norm": 0.75, "learning_rate": 1.1481487479877319e-06, "loss": 0.1218, "step": 46658 }, { "epoch": 34.182417582417585, "grad_norm": 0.66796875, "learning_rate": 1.147883973902573e-06, "loss": 0.0972, "step": 46659 }, { "epoch": 34.18315018315018, "grad_norm": 1.953125, "learning_rate": 1.1476194362795965e-06, "loss": 0.148, "step": 46660 }, { "epoch": 34.18388278388279, "grad_norm": 1.5234375, "learning_rate": 1.1473551351200773e-06, "loss": 0.1305, "step": 46661 }, { "epoch": 34.184615384615384, "grad_norm": 0.78515625, "learning_rate": 1.1470910704252847e-06, "loss": 0.1183, "step": 46662 }, { "epoch": 34.18534798534799, "grad_norm": 1.0390625, "learning_rate": 1.1468272421964777e-06, "loss": 0.1227, "step": 46663 }, { "epoch": 34.186080586080585, "grad_norm": 1.0078125, "learning_rate": 1.1465636504349252e-06, "loss": 0.1299, "step": 46664 }, { "epoch": 34.18681318681319, "grad_norm": 0.98828125, "learning_rate": 1.1463002951418861e-06, "loss": 0.1085, "step": 46665 }, { "epoch": 34.18754578754579, "grad_norm": 1.1484375, "learning_rate": 1.1460371763186187e-06, "loss": 0.1379, "step": 46666 }, { "epoch": 34.18827838827839, "grad_norm": 0.60546875, "learning_rate": 1.145774293966398e-06, "loss": 0.114, "step": 46667 }, { "epoch": 34.18901098901099, "grad_norm": 1.375, "learning_rate": 1.1455116480864719e-06, "loss": 0.1141, "step": 46668 }, { "epoch": 34.18974358974359, "grad_norm": 1.2734375, "learning_rate": 1.1452492386800982e-06, "loss": 0.1473, "step": 46669 }, { "epoch": 34.19047619047619, "grad_norm": 0.78515625, "learning_rate": 1.144987065748541e-06, "loss": 0.1285, "step": 46670 }, { "epoch": 34.191208791208794, "grad_norm": 0.6953125, "learning_rate": 1.144725129293054e-06, "loss": 0.1055, "step": 46671 }, { "epoch": 34.19194139194139, "grad_norm": 0.77734375, "learning_rate": 1.1444634293148896e-06, "loss": 0.1351, "step": 46672 }, { "epoch": 34.192673992673996, "grad_norm": 0.7578125, "learning_rate": 1.144201965815301e-06, "loss": 0.1143, "step": 46673 }, { "epoch": 34.19340659340659, "grad_norm": 0.83984375, "learning_rate": 1.1439407387955464e-06, "loss": 0.1324, "step": 46674 }, { "epoch": 34.1941391941392, "grad_norm": 0.765625, "learning_rate": 1.1436797482568735e-06, "loss": 0.1228, "step": 46675 }, { "epoch": 34.194871794871794, "grad_norm": 1.265625, "learning_rate": 1.14341899420053e-06, "loss": 0.1206, "step": 46676 }, { "epoch": 34.1956043956044, "grad_norm": 1.0234375, "learning_rate": 1.143158476627774e-06, "loss": 0.139, "step": 46677 }, { "epoch": 34.196336996336996, "grad_norm": 1.1953125, "learning_rate": 1.1428981955398425e-06, "loss": 0.1154, "step": 46678 }, { "epoch": 34.1970695970696, "grad_norm": 0.90234375, "learning_rate": 1.1426381509379935e-06, "loss": 0.1252, "step": 46679 }, { "epoch": 34.1978021978022, "grad_norm": 0.9609375, "learning_rate": 1.1423783428234694e-06, "loss": 0.1339, "step": 46680 }, { "epoch": 34.1985347985348, "grad_norm": 0.68359375, "learning_rate": 1.1421187711975063e-06, "loss": 0.1041, "step": 46681 }, { "epoch": 34.1992673992674, "grad_norm": 0.89453125, "learning_rate": 1.1418594360613631e-06, "loss": 0.1089, "step": 46682 }, { "epoch": 34.2, "grad_norm": 0.82421875, "learning_rate": 1.1416003374162708e-06, "loss": 0.1193, "step": 46683 }, { "epoch": 34.2007326007326, "grad_norm": 1.8203125, "learning_rate": 1.1413414752634764e-06, "loss": 0.1754, "step": 46684 }, { "epoch": 34.201465201465204, "grad_norm": 0.73828125, "learning_rate": 1.1410828496042169e-06, "loss": 0.1274, "step": 46685 }, { "epoch": 34.2021978021978, "grad_norm": 1.1953125, "learning_rate": 1.1408244604397343e-06, "loss": 0.1276, "step": 46686 }, { "epoch": 34.202930402930406, "grad_norm": 1.0390625, "learning_rate": 1.1405663077712648e-06, "loss": 0.1102, "step": 46687 }, { "epoch": 34.203663003663, "grad_norm": 1.046875, "learning_rate": 1.140308391600045e-06, "loss": 0.1218, "step": 46688 }, { "epoch": 34.20439560439561, "grad_norm": 0.94140625, "learning_rate": 1.1400507119273117e-06, "loss": 0.1145, "step": 46689 }, { "epoch": 34.205128205128204, "grad_norm": 1.3828125, "learning_rate": 1.1397932687543013e-06, "loss": 0.1252, "step": 46690 }, { "epoch": 34.20586080586081, "grad_norm": 0.7890625, "learning_rate": 1.1395360620822444e-06, "loss": 0.1151, "step": 46691 }, { "epoch": 34.206593406593406, "grad_norm": 0.73828125, "learning_rate": 1.1392790919123725e-06, "loss": 0.0845, "step": 46692 }, { "epoch": 34.20732600732601, "grad_norm": 1.2109375, "learning_rate": 1.1390223582459163e-06, "loss": 0.1145, "step": 46693 }, { "epoch": 34.20805860805861, "grad_norm": 0.953125, "learning_rate": 1.1387658610841126e-06, "loss": 0.1414, "step": 46694 }, { "epoch": 34.20879120879121, "grad_norm": 1.09375, "learning_rate": 1.1385096004281869e-06, "loss": 0.123, "step": 46695 }, { "epoch": 34.20952380952381, "grad_norm": 0.79296875, "learning_rate": 1.138253576279359e-06, "loss": 0.1051, "step": 46696 }, { "epoch": 34.21025641025641, "grad_norm": 0.8125, "learning_rate": 1.137997788638871e-06, "loss": 0.1074, "step": 46697 }, { "epoch": 34.21098901098901, "grad_norm": 1.2578125, "learning_rate": 1.137742237507932e-06, "loss": 0.1324, "step": 46698 }, { "epoch": 34.211721611721615, "grad_norm": 0.70703125, "learning_rate": 1.1374869228877783e-06, "loss": 0.1182, "step": 46699 }, { "epoch": 34.21245421245421, "grad_norm": 0.78515625, "learning_rate": 1.13723184477963e-06, "loss": 0.1334, "step": 46700 }, { "epoch": 34.213186813186816, "grad_norm": 0.87890625, "learning_rate": 1.136977003184713e-06, "loss": 0.1387, "step": 46701 }, { "epoch": 34.21391941391941, "grad_norm": 0.8125, "learning_rate": 1.136722398104236e-06, "loss": 0.1184, "step": 46702 }, { "epoch": 34.21465201465202, "grad_norm": 1.3125, "learning_rate": 1.13646802953943e-06, "loss": 0.148, "step": 46703 }, { "epoch": 34.215384615384615, "grad_norm": 1.03125, "learning_rate": 1.136213897491515e-06, "loss": 0.1067, "step": 46704 }, { "epoch": 34.21611721611722, "grad_norm": 1.3671875, "learning_rate": 1.1359600019617e-06, "loss": 0.1487, "step": 46705 }, { "epoch": 34.216849816849816, "grad_norm": 1.5, "learning_rate": 1.1357063429512054e-06, "loss": 0.1364, "step": 46706 }, { "epoch": 34.21758241758242, "grad_norm": 0.82421875, "learning_rate": 1.1354529204612453e-06, "loss": 0.1126, "step": 46707 }, { "epoch": 34.21831501831502, "grad_norm": 1.265625, "learning_rate": 1.13519973449304e-06, "loss": 0.1305, "step": 46708 }, { "epoch": 34.21904761904762, "grad_norm": 1.4609375, "learning_rate": 1.1349467850477927e-06, "loss": 0.1322, "step": 46709 }, { "epoch": 34.21978021978022, "grad_norm": 1.0234375, "learning_rate": 1.1346940721267238e-06, "loss": 0.1221, "step": 46710 }, { "epoch": 34.22051282051282, "grad_norm": 0.8359375, "learning_rate": 1.134441595731042e-06, "loss": 0.1238, "step": 46711 }, { "epoch": 34.22124542124542, "grad_norm": 0.9296875, "learning_rate": 1.1341893558619567e-06, "loss": 0.1273, "step": 46712 }, { "epoch": 34.221978021978025, "grad_norm": 0.92578125, "learning_rate": 1.1339373525206767e-06, "loss": 0.1171, "step": 46713 }, { "epoch": 34.22271062271062, "grad_norm": 1.078125, "learning_rate": 1.1336855857084056e-06, "loss": 0.1506, "step": 46714 }, { "epoch": 34.223443223443226, "grad_norm": 1.109375, "learning_rate": 1.133434055426358e-06, "loss": 0.1177, "step": 46715 }, { "epoch": 34.22417582417582, "grad_norm": 1.4921875, "learning_rate": 1.133182761675726e-06, "loss": 0.146, "step": 46716 }, { "epoch": 34.22490842490843, "grad_norm": 0.6875, "learning_rate": 1.132931704457725e-06, "loss": 0.0837, "step": 46717 }, { "epoch": 34.225641025641025, "grad_norm": 0.921875, "learning_rate": 1.1326808837735577e-06, "loss": 0.1225, "step": 46718 }, { "epoch": 34.22637362637363, "grad_norm": 0.91015625, "learning_rate": 1.1324302996244169e-06, "loss": 0.1224, "step": 46719 }, { "epoch": 34.227106227106226, "grad_norm": 1.21875, "learning_rate": 1.1321799520115119e-06, "loss": 0.0945, "step": 46720 }, { "epoch": 34.22783882783883, "grad_norm": 1.328125, "learning_rate": 1.1319298409360404e-06, "loss": 0.1358, "step": 46721 }, { "epoch": 34.22857142857143, "grad_norm": 0.65625, "learning_rate": 1.131679966399195e-06, "loss": 0.1123, "step": 46722 }, { "epoch": 34.22930402930403, "grad_norm": 1.125, "learning_rate": 1.1314303284021795e-06, "loss": 0.1225, "step": 46723 }, { "epoch": 34.23003663003663, "grad_norm": 0.83984375, "learning_rate": 1.1311809269461918e-06, "loss": 0.1123, "step": 46724 }, { "epoch": 34.23076923076923, "grad_norm": 0.984375, "learning_rate": 1.1309317620324188e-06, "loss": 0.1077, "step": 46725 }, { "epoch": 34.23150183150183, "grad_norm": 0.6953125, "learning_rate": 1.1306828336620587e-06, "loss": 0.0997, "step": 46726 }, { "epoch": 34.232234432234435, "grad_norm": 0.59375, "learning_rate": 1.130434141836304e-06, "loss": 0.1032, "step": 46727 }, { "epoch": 34.23296703296703, "grad_norm": 0.98046875, "learning_rate": 1.1301856865563473e-06, "loss": 0.1324, "step": 46728 }, { "epoch": 34.233699633699636, "grad_norm": 1.25, "learning_rate": 1.1299374678233809e-06, "loss": 0.1281, "step": 46729 }, { "epoch": 34.234432234432234, "grad_norm": 1.25, "learning_rate": 1.1296894856385866e-06, "loss": 0.1492, "step": 46730 }, { "epoch": 34.23516483516484, "grad_norm": 1.203125, "learning_rate": 1.1294417400031625e-06, "loss": 0.1571, "step": 46731 }, { "epoch": 34.235897435897435, "grad_norm": 0.84765625, "learning_rate": 1.1291942309182847e-06, "loss": 0.1357, "step": 46732 }, { "epoch": 34.23663003663004, "grad_norm": 0.88671875, "learning_rate": 1.1289469583851454e-06, "loss": 0.1412, "step": 46733 }, { "epoch": 34.23736263736264, "grad_norm": 1.546875, "learning_rate": 1.1286999224049317e-06, "loss": 0.1706, "step": 46734 }, { "epoch": 34.23809523809524, "grad_norm": 0.80078125, "learning_rate": 1.1284531229788201e-06, "loss": 0.1267, "step": 46735 }, { "epoch": 34.23882783882784, "grad_norm": 1.40625, "learning_rate": 1.1282065601080028e-06, "loss": 0.1083, "step": 46736 }, { "epoch": 34.23956043956044, "grad_norm": 1.15625, "learning_rate": 1.1279602337936503e-06, "loss": 0.1398, "step": 46737 }, { "epoch": 34.24029304029304, "grad_norm": 1.03125, "learning_rate": 1.1277141440369498e-06, "loss": 0.0893, "step": 46738 }, { "epoch": 34.241025641025644, "grad_norm": 0.8671875, "learning_rate": 1.1274682908390774e-06, "loss": 0.108, "step": 46739 }, { "epoch": 34.24175824175824, "grad_norm": 1.3125, "learning_rate": 1.1272226742012144e-06, "loss": 0.1239, "step": 46740 }, { "epoch": 34.242490842490845, "grad_norm": 1.3203125, "learning_rate": 1.1269772941245373e-06, "loss": 0.1294, "step": 46741 }, { "epoch": 34.24322344322344, "grad_norm": 0.87890625, "learning_rate": 1.1267321506102162e-06, "loss": 0.102, "step": 46742 }, { "epoch": 34.24395604395605, "grad_norm": 0.8046875, "learning_rate": 1.1264872436594329e-06, "loss": 0.1106, "step": 46743 }, { "epoch": 34.244688644688644, "grad_norm": 0.90625, "learning_rate": 1.1262425732733581e-06, "loss": 0.1023, "step": 46744 }, { "epoch": 34.24542124542125, "grad_norm": 1.0, "learning_rate": 1.1259981394531618e-06, "loss": 0.1126, "step": 46745 }, { "epoch": 34.246153846153845, "grad_norm": 1.1171875, "learning_rate": 1.1257539422000153e-06, "loss": 0.1449, "step": 46746 }, { "epoch": 34.24688644688645, "grad_norm": 1.140625, "learning_rate": 1.1255099815150887e-06, "loss": 0.1069, "step": 46747 }, { "epoch": 34.24761904761905, "grad_norm": 1.3046875, "learning_rate": 1.125266257399558e-06, "loss": 0.1168, "step": 46748 }, { "epoch": 34.24835164835165, "grad_norm": 1.1953125, "learning_rate": 1.1250227698545833e-06, "loss": 0.0918, "step": 46749 }, { "epoch": 34.24908424908425, "grad_norm": 1.015625, "learning_rate": 1.1247795188813291e-06, "loss": 0.1245, "step": 46750 }, { "epoch": 34.24981684981685, "grad_norm": 0.86328125, "learning_rate": 1.1245365044809661e-06, "loss": 0.1199, "step": 46751 }, { "epoch": 34.25054945054945, "grad_norm": 0.73046875, "learning_rate": 1.124293726654665e-06, "loss": 0.1153, "step": 46752 }, { "epoch": 34.251282051282054, "grad_norm": 1.078125, "learning_rate": 1.1240511854035746e-06, "loss": 0.1083, "step": 46753 }, { "epoch": 34.25201465201465, "grad_norm": 0.96484375, "learning_rate": 1.1238088807288648e-06, "loss": 0.1149, "step": 46754 }, { "epoch": 34.252747252747255, "grad_norm": 1.046875, "learning_rate": 1.1235668126316958e-06, "loss": 0.139, "step": 46755 }, { "epoch": 34.25347985347985, "grad_norm": 0.5703125, "learning_rate": 1.1233249811132322e-06, "loss": 0.1101, "step": 46756 }, { "epoch": 34.25421245421246, "grad_norm": 1.203125, "learning_rate": 1.1230833861746232e-06, "loss": 0.1034, "step": 46757 }, { "epoch": 34.254945054945054, "grad_norm": 0.765625, "learning_rate": 1.1228420278170331e-06, "loss": 0.1082, "step": 46758 }, { "epoch": 34.25567765567766, "grad_norm": 1.640625, "learning_rate": 1.1226009060416109e-06, "loss": 0.1094, "step": 46759 }, { "epoch": 34.256410256410255, "grad_norm": 1.4375, "learning_rate": 1.1223600208495218e-06, "loss": 0.1184, "step": 46760 }, { "epoch": 34.25714285714286, "grad_norm": 0.95703125, "learning_rate": 1.122119372241914e-06, "loss": 0.1046, "step": 46761 }, { "epoch": 34.25787545787546, "grad_norm": 1.015625, "learning_rate": 1.1218789602199475e-06, "loss": 0.1154, "step": 46762 }, { "epoch": 34.25860805860806, "grad_norm": 0.94140625, "learning_rate": 1.121638784784765e-06, "loss": 0.1261, "step": 46763 }, { "epoch": 34.25934065934066, "grad_norm": 1.203125, "learning_rate": 1.1213988459375152e-06, "loss": 0.1773, "step": 46764 }, { "epoch": 34.26007326007326, "grad_norm": 1.375, "learning_rate": 1.1211591436793633e-06, "loss": 0.1072, "step": 46765 }, { "epoch": 34.26080586080586, "grad_norm": 1.078125, "learning_rate": 1.1209196780114415e-06, "loss": 0.1185, "step": 46766 }, { "epoch": 34.261538461538464, "grad_norm": 0.875, "learning_rate": 1.1206804489349032e-06, "loss": 0.1166, "step": 46767 }, { "epoch": 34.26227106227106, "grad_norm": 1.1796875, "learning_rate": 1.120441456450898e-06, "loss": 0.1297, "step": 46768 }, { "epoch": 34.263003663003666, "grad_norm": 1.109375, "learning_rate": 1.120202700560568e-06, "loss": 0.1329, "step": 46769 }, { "epoch": 34.26373626373626, "grad_norm": 1.2421875, "learning_rate": 1.119964181265057e-06, "loss": 0.1269, "step": 46770 }, { "epoch": 34.26446886446887, "grad_norm": 1.0625, "learning_rate": 1.119725898565508e-06, "loss": 0.1225, "step": 46771 }, { "epoch": 34.265201465201464, "grad_norm": 0.98046875, "learning_rate": 1.1194878524630581e-06, "loss": 0.1006, "step": 46772 }, { "epoch": 34.26593406593407, "grad_norm": 0.88671875, "learning_rate": 1.1192500429588562e-06, "loss": 0.113, "step": 46773 }, { "epoch": 34.266666666666666, "grad_norm": 0.76171875, "learning_rate": 1.11901247005404e-06, "loss": 0.0976, "step": 46774 }, { "epoch": 34.26739926739927, "grad_norm": 0.87890625, "learning_rate": 1.1187751337497469e-06, "loss": 0.1146, "step": 46775 }, { "epoch": 34.26813186813187, "grad_norm": 1.0859375, "learning_rate": 1.118538034047109e-06, "loss": 0.1049, "step": 46776 }, { "epoch": 34.26886446886447, "grad_norm": 1.3359375, "learning_rate": 1.1183011709472642e-06, "loss": 0.1368, "step": 46777 }, { "epoch": 34.26959706959707, "grad_norm": 0.8359375, "learning_rate": 1.1180645444513551e-06, "loss": 0.113, "step": 46778 }, { "epoch": 34.27032967032967, "grad_norm": 1.0390625, "learning_rate": 1.1178281545605088e-06, "loss": 0.1344, "step": 46779 }, { "epoch": 34.27106227106227, "grad_norm": 1.2578125, "learning_rate": 1.117592001275857e-06, "loss": 0.1245, "step": 46780 }, { "epoch": 34.271794871794874, "grad_norm": 0.890625, "learning_rate": 1.1173560845985317e-06, "loss": 0.1021, "step": 46781 }, { "epoch": 34.27252747252747, "grad_norm": 1.1171875, "learning_rate": 1.1171204045296656e-06, "loss": 0.1338, "step": 46782 }, { "epoch": 34.273260073260076, "grad_norm": 0.78125, "learning_rate": 1.1168849610703904e-06, "loss": 0.1101, "step": 46783 }, { "epoch": 34.27399267399267, "grad_norm": 1.0703125, "learning_rate": 1.1166497542218268e-06, "loss": 0.112, "step": 46784 }, { "epoch": 34.27472527472528, "grad_norm": 0.74609375, "learning_rate": 1.1164147839851078e-06, "loss": 0.0939, "step": 46785 }, { "epoch": 34.275457875457874, "grad_norm": 0.8125, "learning_rate": 1.1161800503613536e-06, "loss": 0.1287, "step": 46786 }, { "epoch": 34.27619047619048, "grad_norm": 1.1015625, "learning_rate": 1.1159455533516971e-06, "loss": 0.0923, "step": 46787 }, { "epoch": 34.276923076923076, "grad_norm": 1.359375, "learning_rate": 1.1157112929572532e-06, "loss": 0.134, "step": 46788 }, { "epoch": 34.27765567765568, "grad_norm": 0.94921875, "learning_rate": 1.1154772691791544e-06, "loss": 0.1272, "step": 46789 }, { "epoch": 34.27838827838828, "grad_norm": 1.25, "learning_rate": 1.115243482018511e-06, "loss": 0.1148, "step": 46790 }, { "epoch": 34.27912087912088, "grad_norm": 1.2421875, "learning_rate": 1.115009931476449e-06, "loss": 0.1025, "step": 46791 }, { "epoch": 34.27985347985348, "grad_norm": 0.91796875, "learning_rate": 1.1147766175540846e-06, "loss": 0.0853, "step": 46792 }, { "epoch": 34.28058608058608, "grad_norm": 0.890625, "learning_rate": 1.114543540252544e-06, "loss": 0.1099, "step": 46793 }, { "epoch": 34.28131868131868, "grad_norm": 1.1640625, "learning_rate": 1.1143106995729318e-06, "loss": 0.1283, "step": 46794 }, { "epoch": 34.282051282051285, "grad_norm": 1.328125, "learning_rate": 1.1140780955163694e-06, "loss": 0.1459, "step": 46795 }, { "epoch": 34.28278388278388, "grad_norm": 1.21875, "learning_rate": 1.1138457280839724e-06, "loss": 0.113, "step": 46796 }, { "epoch": 34.283516483516486, "grad_norm": 1.3359375, "learning_rate": 1.1136135972768507e-06, "loss": 0.1503, "step": 46797 }, { "epoch": 34.28424908424908, "grad_norm": 1.5625, "learning_rate": 1.1133817030961198e-06, "loss": 0.1252, "step": 46798 }, { "epoch": 34.28498168498169, "grad_norm": 0.95703125, "learning_rate": 1.1131500455428902e-06, "loss": 0.1421, "step": 46799 }, { "epoch": 34.285714285714285, "grad_norm": 0.74609375, "learning_rate": 1.1129186246182718e-06, "loss": 0.1088, "step": 46800 }, { "epoch": 34.28644688644689, "grad_norm": 1.28125, "learning_rate": 1.1126874403233747e-06, "loss": 0.141, "step": 46801 }, { "epoch": 34.287179487179486, "grad_norm": 1.2890625, "learning_rate": 1.1124564926593034e-06, "loss": 0.1418, "step": 46802 }, { "epoch": 34.28791208791209, "grad_norm": 1.0078125, "learning_rate": 1.1122257816271626e-06, "loss": 0.0929, "step": 46803 }, { "epoch": 34.28864468864469, "grad_norm": 0.90234375, "learning_rate": 1.1119953072280682e-06, "loss": 0.1187, "step": 46804 }, { "epoch": 34.28937728937729, "grad_norm": 1.1484375, "learning_rate": 1.1117650694631135e-06, "loss": 0.1108, "step": 46805 }, { "epoch": 34.29010989010989, "grad_norm": 1.09375, "learning_rate": 1.1115350683334032e-06, "loss": 0.132, "step": 46806 }, { "epoch": 34.29084249084249, "grad_norm": 1.34375, "learning_rate": 1.1113053038400419e-06, "loss": 0.1246, "step": 46807 }, { "epoch": 34.29157509157509, "grad_norm": 1.21875, "learning_rate": 1.111075775984134e-06, "loss": 0.1315, "step": 46808 }, { "epoch": 34.292307692307695, "grad_norm": 1.5546875, "learning_rate": 1.1108464847667737e-06, "loss": 0.1317, "step": 46809 }, { "epoch": 34.29304029304029, "grad_norm": 1.3046875, "learning_rate": 1.1106174301890596e-06, "loss": 0.1316, "step": 46810 }, { "epoch": 34.293772893772896, "grad_norm": 0.73046875, "learning_rate": 1.1103886122520912e-06, "loss": 0.0945, "step": 46811 }, { "epoch": 34.29450549450549, "grad_norm": 0.83203125, "learning_rate": 1.1101600309569673e-06, "loss": 0.1064, "step": 46812 }, { "epoch": 34.2952380952381, "grad_norm": 1.21875, "learning_rate": 1.1099316863047761e-06, "loss": 0.102, "step": 46813 }, { "epoch": 34.295970695970695, "grad_norm": 1.625, "learning_rate": 1.1097035782966165e-06, "loss": 0.1023, "step": 46814 }, { "epoch": 34.2967032967033, "grad_norm": 0.71875, "learning_rate": 1.1094757069335884e-06, "loss": 0.1193, "step": 46815 }, { "epoch": 34.297435897435896, "grad_norm": 0.76953125, "learning_rate": 1.1092480722167682e-06, "loss": 0.1213, "step": 46816 }, { "epoch": 34.2981684981685, "grad_norm": 0.625, "learning_rate": 1.1090206741472553e-06, "loss": 0.1018, "step": 46817 }, { "epoch": 34.2989010989011, "grad_norm": 0.6328125, "learning_rate": 1.1087935127261375e-06, "loss": 0.1091, "step": 46818 }, { "epoch": 34.2996336996337, "grad_norm": 0.76171875, "learning_rate": 1.1085665879545092e-06, "loss": 0.095, "step": 46819 }, { "epoch": 34.3003663003663, "grad_norm": 1.3828125, "learning_rate": 1.1083398998334468e-06, "loss": 0.0917, "step": 46820 }, { "epoch": 34.301098901098904, "grad_norm": 0.9140625, "learning_rate": 1.108113448364044e-06, "loss": 0.1239, "step": 46821 }, { "epoch": 34.3018315018315, "grad_norm": 0.59375, "learning_rate": 1.1078872335473838e-06, "loss": 0.1162, "step": 46822 }, { "epoch": 34.302564102564105, "grad_norm": 0.78125, "learning_rate": 1.107661255384554e-06, "loss": 0.1062, "step": 46823 }, { "epoch": 34.3032967032967, "grad_norm": 1.328125, "learning_rate": 1.1074355138766265e-06, "loss": 0.1546, "step": 46824 }, { "epoch": 34.30402930402931, "grad_norm": 1.0390625, "learning_rate": 1.1072100090246948e-06, "loss": 0.1362, "step": 46825 }, { "epoch": 34.304761904761904, "grad_norm": 0.859375, "learning_rate": 1.106984740829836e-06, "loss": 0.1067, "step": 46826 }, { "epoch": 34.30549450549451, "grad_norm": 1.15625, "learning_rate": 1.1067597092931272e-06, "loss": 0.1519, "step": 46827 }, { "epoch": 34.306227106227105, "grad_norm": 1.09375, "learning_rate": 1.1065349144156458e-06, "loss": 0.1441, "step": 46828 }, { "epoch": 34.30695970695971, "grad_norm": 0.83203125, "learning_rate": 1.1063103561984687e-06, "loss": 0.1334, "step": 46829 }, { "epoch": 34.30769230769231, "grad_norm": 1.265625, "learning_rate": 1.106086034642673e-06, "loss": 0.1422, "step": 46830 }, { "epoch": 34.30842490842491, "grad_norm": 0.85546875, "learning_rate": 1.105861949749336e-06, "loss": 0.1167, "step": 46831 }, { "epoch": 34.30915750915751, "grad_norm": 1.109375, "learning_rate": 1.1056381015195345e-06, "loss": 0.131, "step": 46832 }, { "epoch": 34.30989010989011, "grad_norm": 0.8046875, "learning_rate": 1.1054144899543298e-06, "loss": 0.1156, "step": 46833 }, { "epoch": 34.31062271062271, "grad_norm": 0.77734375, "learning_rate": 1.1051911150547985e-06, "loss": 0.1094, "step": 46834 }, { "epoch": 34.311355311355314, "grad_norm": 1.1640625, "learning_rate": 1.104967976822018e-06, "loss": 0.1316, "step": 46835 }, { "epoch": 34.31208791208791, "grad_norm": 1.015625, "learning_rate": 1.1047450752570488e-06, "loss": 0.1243, "step": 46836 }, { "epoch": 34.312820512820515, "grad_norm": 0.73828125, "learning_rate": 1.1045224103609571e-06, "loss": 0.1143, "step": 46837 }, { "epoch": 34.31355311355311, "grad_norm": 0.859375, "learning_rate": 1.10429998213482e-06, "loss": 0.1253, "step": 46838 }, { "epoch": 34.31428571428572, "grad_norm": 0.76171875, "learning_rate": 1.1040777905796927e-06, "loss": 0.1172, "step": 46839 }, { "epoch": 34.315018315018314, "grad_norm": 1.6640625, "learning_rate": 1.1038558356966468e-06, "loss": 0.1054, "step": 46840 }, { "epoch": 34.31575091575092, "grad_norm": 1.359375, "learning_rate": 1.103634117486743e-06, "loss": 0.1392, "step": 46841 }, { "epoch": 34.316483516483515, "grad_norm": 1.0703125, "learning_rate": 1.103412635951042e-06, "loss": 0.1086, "step": 46842 }, { "epoch": 34.31721611721612, "grad_norm": 0.82421875, "learning_rate": 1.103191391090604e-06, "loss": 0.1132, "step": 46843 }, { "epoch": 34.31794871794872, "grad_norm": 0.99609375, "learning_rate": 1.102970382906496e-06, "loss": 0.1323, "step": 46844 }, { "epoch": 34.31868131868132, "grad_norm": 1.1640625, "learning_rate": 1.102749611399778e-06, "loss": 0.1304, "step": 46845 }, { "epoch": 34.31941391941392, "grad_norm": 0.828125, "learning_rate": 1.1025290765714944e-06, "loss": 0.0973, "step": 46846 }, { "epoch": 34.32014652014652, "grad_norm": 0.81640625, "learning_rate": 1.1023087784227113e-06, "loss": 0.1465, "step": 46847 }, { "epoch": 34.32087912087912, "grad_norm": 1.15625, "learning_rate": 1.1020887169544894e-06, "loss": 0.1435, "step": 46848 }, { "epoch": 34.321611721611724, "grad_norm": 0.91796875, "learning_rate": 1.1018688921678728e-06, "loss": 0.121, "step": 46849 }, { "epoch": 34.32234432234432, "grad_norm": 1.640625, "learning_rate": 1.1016493040639166e-06, "loss": 0.1314, "step": 46850 }, { "epoch": 34.323076923076925, "grad_norm": 0.984375, "learning_rate": 1.1014299526436762e-06, "loss": 0.122, "step": 46851 }, { "epoch": 34.32380952380952, "grad_norm": 0.83203125, "learning_rate": 1.1012108379082008e-06, "loss": 0.1281, "step": 46852 }, { "epoch": 34.32454212454213, "grad_norm": 1.140625, "learning_rate": 1.100991959858546e-06, "loss": 0.1207, "step": 46853 }, { "epoch": 34.325274725274724, "grad_norm": 1.0078125, "learning_rate": 1.1007733184957503e-06, "loss": 0.11, "step": 46854 }, { "epoch": 34.32600732600733, "grad_norm": 0.875, "learning_rate": 1.1005549138208634e-06, "loss": 0.1556, "step": 46855 }, { "epoch": 34.326739926739926, "grad_norm": 1.6640625, "learning_rate": 1.1003367458349404e-06, "loss": 0.129, "step": 46856 }, { "epoch": 34.32747252747253, "grad_norm": 0.95703125, "learning_rate": 1.10011881453902e-06, "loss": 0.1131, "step": 46857 }, { "epoch": 34.32820512820513, "grad_norm": 0.81640625, "learning_rate": 1.0999011199341465e-06, "loss": 0.1082, "step": 46858 }, { "epoch": 34.32893772893773, "grad_norm": 1.3671875, "learning_rate": 1.0996836620213635e-06, "loss": 0.1163, "step": 46859 }, { "epoch": 34.32967032967033, "grad_norm": 1.0, "learning_rate": 1.099466440801716e-06, "loss": 0.1085, "step": 46860 }, { "epoch": 34.33040293040293, "grad_norm": 0.9296875, "learning_rate": 1.0992494562762368e-06, "loss": 0.1025, "step": 46861 }, { "epoch": 34.33113553113553, "grad_norm": 1.1953125, "learning_rate": 1.0990327084459752e-06, "loss": 0.102, "step": 46862 }, { "epoch": 34.331868131868134, "grad_norm": 0.9765625, "learning_rate": 1.0988161973119594e-06, "loss": 0.1237, "step": 46863 }, { "epoch": 34.33260073260073, "grad_norm": 1.4453125, "learning_rate": 1.0985999228752388e-06, "loss": 0.0998, "step": 46864 }, { "epoch": 34.333333333333336, "grad_norm": 1.1953125, "learning_rate": 1.0983838851368412e-06, "loss": 0.1663, "step": 46865 }, { "epoch": 34.33406593406593, "grad_norm": 0.99609375, "learning_rate": 1.0981680840977997e-06, "loss": 0.121, "step": 46866 }, { "epoch": 34.33479853479854, "grad_norm": 0.71875, "learning_rate": 1.0979525197591585e-06, "loss": 0.1053, "step": 46867 }, { "epoch": 34.335531135531134, "grad_norm": 1.0234375, "learning_rate": 1.0977371921219396e-06, "loss": 0.0907, "step": 46868 }, { "epoch": 34.33626373626374, "grad_norm": 0.796875, "learning_rate": 1.097522101187182e-06, "loss": 0.1158, "step": 46869 }, { "epoch": 34.336996336996336, "grad_norm": 1.5390625, "learning_rate": 1.0973072469559187e-06, "loss": 0.1234, "step": 46870 }, { "epoch": 34.33772893772894, "grad_norm": 0.92578125, "learning_rate": 1.0970926294291663e-06, "loss": 0.116, "step": 46871 }, { "epoch": 34.33846153846154, "grad_norm": 1.09375, "learning_rate": 1.0968782486079637e-06, "loss": 0.116, "step": 46872 }, { "epoch": 34.33919413919414, "grad_norm": 1.609375, "learning_rate": 1.0966641044933383e-06, "loss": 0.1643, "step": 46873 }, { "epoch": 34.33992673992674, "grad_norm": 1.3046875, "learning_rate": 1.096450197086307e-06, "loss": 0.1407, "step": 46874 }, { "epoch": 34.34065934065934, "grad_norm": 0.9375, "learning_rate": 1.0962365263879084e-06, "loss": 0.111, "step": 46875 }, { "epoch": 34.34139194139194, "grad_norm": 1.1953125, "learning_rate": 1.0960230923991536e-06, "loss": 0.1353, "step": 46876 }, { "epoch": 34.342124542124544, "grad_norm": 1.2421875, "learning_rate": 1.0958098951210706e-06, "loss": 0.1254, "step": 46877 }, { "epoch": 34.34285714285714, "grad_norm": 0.76953125, "learning_rate": 1.0955969345546866e-06, "loss": 0.1144, "step": 46878 }, { "epoch": 34.343589743589746, "grad_norm": 0.921875, "learning_rate": 1.0953842107010077e-06, "loss": 0.1215, "step": 46879 }, { "epoch": 34.34432234432234, "grad_norm": 0.8359375, "learning_rate": 1.0951717235610667e-06, "loss": 0.1127, "step": 46880 }, { "epoch": 34.34505494505495, "grad_norm": 0.90625, "learning_rate": 1.0949594731358753e-06, "loss": 0.0895, "step": 46881 }, { "epoch": 34.345787545787545, "grad_norm": 0.7109375, "learning_rate": 1.0947474594264497e-06, "loss": 0.1105, "step": 46882 }, { "epoch": 34.34652014652015, "grad_norm": 0.8828125, "learning_rate": 1.0945356824338123e-06, "loss": 0.0945, "step": 46883 }, { "epoch": 34.347252747252746, "grad_norm": 0.84765625, "learning_rate": 1.0943241421589688e-06, "loss": 0.1248, "step": 46884 }, { "epoch": 34.34798534798535, "grad_norm": 0.6171875, "learning_rate": 1.0941128386029356e-06, "loss": 0.1137, "step": 46885 }, { "epoch": 34.34871794871795, "grad_norm": 1.4296875, "learning_rate": 1.0939017717667353e-06, "loss": 0.1149, "step": 46886 }, { "epoch": 34.34945054945055, "grad_norm": 0.91796875, "learning_rate": 1.0936909416513627e-06, "loss": 0.1175, "step": 46887 }, { "epoch": 34.35018315018315, "grad_norm": 1.0234375, "learning_rate": 1.0934803482578394e-06, "loss": 0.097, "step": 46888 }, { "epoch": 34.35091575091575, "grad_norm": 1.21875, "learning_rate": 1.0932699915871658e-06, "loss": 0.1305, "step": 46889 }, { "epoch": 34.35164835164835, "grad_norm": 0.91015625, "learning_rate": 1.0930598716403647e-06, "loss": 0.1362, "step": 46890 }, { "epoch": 34.352380952380955, "grad_norm": 1.671875, "learning_rate": 1.0928499884184245e-06, "loss": 0.1121, "step": 46891 }, { "epoch": 34.35311355311355, "grad_norm": 1.359375, "learning_rate": 1.0926403419223624e-06, "loss": 0.128, "step": 46892 }, { "epoch": 34.353846153846156, "grad_norm": 1.28125, "learning_rate": 1.0924309321531782e-06, "loss": 0.1178, "step": 46893 }, { "epoch": 34.35457875457875, "grad_norm": 0.921875, "learning_rate": 1.0922217591118778e-06, "loss": 0.1224, "step": 46894 }, { "epoch": 34.35531135531136, "grad_norm": 1.28125, "learning_rate": 1.0920128227994613e-06, "loss": 0.1176, "step": 46895 }, { "epoch": 34.356043956043955, "grad_norm": 0.8515625, "learning_rate": 1.0918041232169346e-06, "loss": 0.1199, "step": 46896 }, { "epoch": 34.35677655677656, "grad_norm": 0.90234375, "learning_rate": 1.0915956603652925e-06, "loss": 0.1173, "step": 46897 }, { "epoch": 34.357509157509156, "grad_norm": 0.52734375, "learning_rate": 1.0913874342455347e-06, "loss": 0.1014, "step": 46898 }, { "epoch": 34.35824175824176, "grad_norm": 0.80078125, "learning_rate": 1.0911794448586564e-06, "loss": 0.1516, "step": 46899 }, { "epoch": 34.35897435897436, "grad_norm": 0.9921875, "learning_rate": 1.0909716922056573e-06, "loss": 0.1343, "step": 46900 }, { "epoch": 34.35970695970696, "grad_norm": 0.7578125, "learning_rate": 1.0907641762875383e-06, "loss": 0.1135, "step": 46901 }, { "epoch": 34.36043956043956, "grad_norm": 1.71875, "learning_rate": 1.0905568971052881e-06, "loss": 0.1204, "step": 46902 }, { "epoch": 34.36117216117216, "grad_norm": 1.25, "learning_rate": 1.0903498546598958e-06, "loss": 0.1136, "step": 46903 }, { "epoch": 34.36190476190476, "grad_norm": 0.609375, "learning_rate": 1.0901430489523565e-06, "loss": 0.1354, "step": 46904 }, { "epoch": 34.362637362637365, "grad_norm": 1.015625, "learning_rate": 1.0899364799836647e-06, "loss": 0.1206, "step": 46905 }, { "epoch": 34.36336996336996, "grad_norm": 1.09375, "learning_rate": 1.0897301477548096e-06, "loss": 0.1195, "step": 46906 }, { "epoch": 34.364102564102566, "grad_norm": 0.85546875, "learning_rate": 1.089524052266775e-06, "loss": 0.1217, "step": 46907 }, { "epoch": 34.36483516483516, "grad_norm": 0.79296875, "learning_rate": 1.08931819352055e-06, "loss": 0.1247, "step": 46908 }, { "epoch": 34.36556776556777, "grad_norm": 1.2578125, "learning_rate": 1.0891125715171238e-06, "loss": 0.1085, "step": 46909 }, { "epoch": 34.366300366300365, "grad_norm": 1.09375, "learning_rate": 1.0889071862574803e-06, "loss": 0.1028, "step": 46910 }, { "epoch": 34.36703296703297, "grad_norm": 0.73046875, "learning_rate": 1.0887020377425975e-06, "loss": 0.1072, "step": 46911 }, { "epoch": 34.36776556776557, "grad_norm": 1.359375, "learning_rate": 1.0884971259734703e-06, "loss": 0.1076, "step": 46912 }, { "epoch": 34.36849816849817, "grad_norm": 1.1640625, "learning_rate": 1.0882924509510714e-06, "loss": 0.1337, "step": 46913 }, { "epoch": 34.36923076923077, "grad_norm": 1.265625, "learning_rate": 1.0880880126763841e-06, "loss": 0.1597, "step": 46914 }, { "epoch": 34.36996336996337, "grad_norm": 0.93359375, "learning_rate": 1.0878838111503928e-06, "loss": 0.1099, "step": 46915 }, { "epoch": 34.37069597069597, "grad_norm": 1.078125, "learning_rate": 1.0876798463740643e-06, "loss": 0.1253, "step": 46916 }, { "epoch": 34.371428571428574, "grad_norm": 1.203125, "learning_rate": 1.0874761183483878e-06, "loss": 0.137, "step": 46917 }, { "epoch": 34.37216117216117, "grad_norm": 1.1953125, "learning_rate": 1.0872726270743306e-06, "loss": 0.1171, "step": 46918 }, { "epoch": 34.372893772893775, "grad_norm": 0.9609375, "learning_rate": 1.0870693725528764e-06, "loss": 0.1102, "step": 46919 }, { "epoch": 34.37362637362637, "grad_norm": 1.0234375, "learning_rate": 1.086866354784987e-06, "loss": 0.1122, "step": 46920 }, { "epoch": 34.37435897435898, "grad_norm": 1.015625, "learning_rate": 1.0866635737716462e-06, "loss": 0.0938, "step": 46921 }, { "epoch": 34.375091575091574, "grad_norm": 1.5625, "learning_rate": 1.086461029513821e-06, "loss": 0.1252, "step": 46922 }, { "epoch": 34.37582417582418, "grad_norm": 1.1953125, "learning_rate": 1.086258722012484e-06, "loss": 0.1287, "step": 46923 }, { "epoch": 34.376556776556775, "grad_norm": 0.75, "learning_rate": 1.0860566512685974e-06, "loss": 0.098, "step": 46924 }, { "epoch": 34.37728937728938, "grad_norm": 1.0078125, "learning_rate": 1.0858548172831391e-06, "loss": 0.1481, "step": 46925 }, { "epoch": 34.37802197802198, "grad_norm": 1.1796875, "learning_rate": 1.085653220057071e-06, "loss": 0.1434, "step": 46926 }, { "epoch": 34.37875457875458, "grad_norm": 1.0, "learning_rate": 1.0854518595913656e-06, "loss": 0.0889, "step": 46927 }, { "epoch": 34.37948717948718, "grad_norm": 1.34375, "learning_rate": 1.0852507358869736e-06, "loss": 0.1249, "step": 46928 }, { "epoch": 34.38021978021978, "grad_norm": 1.15625, "learning_rate": 1.0850498489448735e-06, "loss": 0.1557, "step": 46929 }, { "epoch": 34.38095238095238, "grad_norm": 1.046875, "learning_rate": 1.084849198766016e-06, "loss": 0.1313, "step": 46930 }, { "epoch": 34.381684981684984, "grad_norm": 0.96484375, "learning_rate": 1.0846487853513683e-06, "loss": 0.1071, "step": 46931 }, { "epoch": 34.38241758241758, "grad_norm": 1.171875, "learning_rate": 1.0844486087018973e-06, "loss": 0.1448, "step": 46932 }, { "epoch": 34.383150183150185, "grad_norm": 1.21875, "learning_rate": 1.0842486688185486e-06, "loss": 0.1298, "step": 46933 }, { "epoch": 34.38388278388278, "grad_norm": 1.0078125, "learning_rate": 1.0840489657022896e-06, "loss": 0.1006, "step": 46934 }, { "epoch": 34.38461538461539, "grad_norm": 1.078125, "learning_rate": 1.0838494993540703e-06, "loss": 0.1343, "step": 46935 }, { "epoch": 34.385347985347984, "grad_norm": 1.0859375, "learning_rate": 1.0836502697748533e-06, "loss": 0.1659, "step": 46936 }, { "epoch": 34.38608058608059, "grad_norm": 0.76953125, "learning_rate": 1.0834512769655887e-06, "loss": 0.0885, "step": 46937 }, { "epoch": 34.386813186813185, "grad_norm": 0.625, "learning_rate": 1.0832525209272332e-06, "loss": 0.1029, "step": 46938 }, { "epoch": 34.38754578754579, "grad_norm": 0.9375, "learning_rate": 1.0830540016607371e-06, "loss": 0.1211, "step": 46939 }, { "epoch": 34.38827838827839, "grad_norm": 0.75390625, "learning_rate": 1.0828557191670517e-06, "loss": 0.0968, "step": 46940 }, { "epoch": 34.38901098901099, "grad_norm": 1.359375, "learning_rate": 1.082657673447127e-06, "loss": 0.1045, "step": 46941 }, { "epoch": 34.38974358974359, "grad_norm": 0.94140625, "learning_rate": 1.0824598645019141e-06, "loss": 0.1199, "step": 46942 }, { "epoch": 34.39047619047619, "grad_norm": 0.80078125, "learning_rate": 1.082262292332353e-06, "loss": 0.1095, "step": 46943 }, { "epoch": 34.39120879120879, "grad_norm": 0.8671875, "learning_rate": 1.0820649569394e-06, "loss": 0.159, "step": 46944 }, { "epoch": 34.391941391941394, "grad_norm": 1.2109375, "learning_rate": 1.0818678583239997e-06, "loss": 0.1424, "step": 46945 }, { "epoch": 34.39267399267399, "grad_norm": 0.81640625, "learning_rate": 1.0816709964870867e-06, "loss": 0.0947, "step": 46946 }, { "epoch": 34.393406593406596, "grad_norm": 0.6796875, "learning_rate": 1.0814743714296173e-06, "loss": 0.1119, "step": 46947 }, { "epoch": 34.39413919413919, "grad_norm": 0.89453125, "learning_rate": 1.0812779831525203e-06, "loss": 0.1087, "step": 46948 }, { "epoch": 34.3948717948718, "grad_norm": 0.60546875, "learning_rate": 1.0810818316567518e-06, "loss": 0.0916, "step": 46949 }, { "epoch": 34.395604395604394, "grad_norm": 0.72265625, "learning_rate": 1.0808859169432352e-06, "loss": 0.1369, "step": 46950 }, { "epoch": 34.396336996337, "grad_norm": 1.046875, "learning_rate": 1.080690239012921e-06, "loss": 0.1102, "step": 46951 }, { "epoch": 34.397069597069596, "grad_norm": 0.78515625, "learning_rate": 1.0804947978667436e-06, "loss": 0.1209, "step": 46952 }, { "epoch": 34.3978021978022, "grad_norm": 1.40625, "learning_rate": 1.0802995935056428e-06, "loss": 0.1295, "step": 46953 }, { "epoch": 34.3985347985348, "grad_norm": 0.8203125, "learning_rate": 1.0801046259305417e-06, "loss": 0.1097, "step": 46954 }, { "epoch": 34.3992673992674, "grad_norm": 0.9921875, "learning_rate": 1.0799098951423856e-06, "loss": 0.1191, "step": 46955 }, { "epoch": 34.4, "grad_norm": 0.59765625, "learning_rate": 1.079715401142109e-06, "loss": 0.1113, "step": 46956 }, { "epoch": 34.4007326007326, "grad_norm": 0.92578125, "learning_rate": 1.079521143930635e-06, "loss": 0.1186, "step": 46957 }, { "epoch": 34.4014652014652, "grad_norm": 1.2109375, "learning_rate": 1.0793271235089031e-06, "loss": 0.1048, "step": 46958 }, { "epoch": 34.402197802197804, "grad_norm": 0.95703125, "learning_rate": 1.079133339877837e-06, "loss": 0.1331, "step": 46959 }, { "epoch": 34.4029304029304, "grad_norm": 1.1328125, "learning_rate": 1.078939793038365e-06, "loss": 0.1105, "step": 46960 }, { "epoch": 34.403663003663006, "grad_norm": 1.203125, "learning_rate": 1.0787464829914165e-06, "loss": 0.1064, "step": 46961 }, { "epoch": 34.4043956043956, "grad_norm": 1.046875, "learning_rate": 1.0785534097379194e-06, "loss": 0.1229, "step": 46962 }, { "epoch": 34.40512820512821, "grad_norm": 1.4296875, "learning_rate": 1.0783605732787977e-06, "loss": 0.1181, "step": 46963 }, { "epoch": 34.405860805860804, "grad_norm": 1.453125, "learning_rate": 1.0781679736149687e-06, "loss": 0.133, "step": 46964 }, { "epoch": 34.40659340659341, "grad_norm": 0.82421875, "learning_rate": 1.0779756107473667e-06, "loss": 0.1104, "step": 46965 }, { "epoch": 34.407326007326006, "grad_norm": 1.046875, "learning_rate": 1.0777834846769044e-06, "loss": 0.1676, "step": 46966 }, { "epoch": 34.40805860805861, "grad_norm": 0.95703125, "learning_rate": 1.07759159540451e-06, "loss": 0.1103, "step": 46967 }, { "epoch": 34.40879120879121, "grad_norm": 0.90234375, "learning_rate": 1.0773999429310964e-06, "loss": 0.1314, "step": 46968 }, { "epoch": 34.40952380952381, "grad_norm": 0.94921875, "learning_rate": 1.0772085272575808e-06, "loss": 0.1154, "step": 46969 }, { "epoch": 34.41025641025641, "grad_norm": 1.3359375, "learning_rate": 1.0770173483848863e-06, "loss": 0.1159, "step": 46970 }, { "epoch": 34.41098901098901, "grad_norm": 0.7890625, "learning_rate": 1.0768264063139312e-06, "loss": 0.1097, "step": 46971 }, { "epoch": 34.41172161172161, "grad_norm": 0.8984375, "learning_rate": 1.0766357010456168e-06, "loss": 0.0898, "step": 46972 }, { "epoch": 34.412454212454215, "grad_norm": 1.0390625, "learning_rate": 1.0764452325808716e-06, "loss": 0.1104, "step": 46973 }, { "epoch": 34.41318681318681, "grad_norm": 0.69921875, "learning_rate": 1.076255000920597e-06, "loss": 0.1218, "step": 46974 }, { "epoch": 34.413919413919416, "grad_norm": 0.984375, "learning_rate": 1.076065006065716e-06, "loss": 0.1079, "step": 46975 }, { "epoch": 34.41465201465201, "grad_norm": 1.0703125, "learning_rate": 1.0758752480171246e-06, "loss": 0.1001, "step": 46976 }, { "epoch": 34.41538461538462, "grad_norm": 0.7421875, "learning_rate": 1.0756857267757463e-06, "loss": 0.1219, "step": 46977 }, { "epoch": 34.416117216117215, "grad_norm": 0.6875, "learning_rate": 1.0754964423424763e-06, "loss": 0.1271, "step": 46978 }, { "epoch": 34.41684981684982, "grad_norm": 1.9765625, "learning_rate": 1.0753073947182327e-06, "loss": 0.1107, "step": 46979 }, { "epoch": 34.417582417582416, "grad_norm": 0.7578125, "learning_rate": 1.075118583903917e-06, "loss": 0.1185, "step": 46980 }, { "epoch": 34.41831501831502, "grad_norm": 1.2734375, "learning_rate": 1.07493000990043e-06, "loss": 0.1213, "step": 46981 }, { "epoch": 34.41904761904762, "grad_norm": 0.87890625, "learning_rate": 1.074741672708679e-06, "loss": 0.0954, "step": 46982 }, { "epoch": 34.41978021978022, "grad_norm": 1.1875, "learning_rate": 1.0745535723295703e-06, "loss": 0.1175, "step": 46983 }, { "epoch": 34.42051282051282, "grad_norm": 2.421875, "learning_rate": 1.0743657087639944e-06, "loss": 0.1413, "step": 46984 }, { "epoch": 34.42124542124542, "grad_norm": 0.88671875, "learning_rate": 1.0741780820128636e-06, "loss": 0.1069, "step": 46985 }, { "epoch": 34.42197802197802, "grad_norm": 0.80078125, "learning_rate": 1.073990692077068e-06, "loss": 0.1184, "step": 46986 }, { "epoch": 34.422710622710625, "grad_norm": 1.25, "learning_rate": 1.0738035389575092e-06, "loss": 0.1622, "step": 46987 }, { "epoch": 34.42344322344322, "grad_norm": 1.1171875, "learning_rate": 1.073616622655083e-06, "loss": 0.1361, "step": 46988 }, { "epoch": 34.424175824175826, "grad_norm": 1.1796875, "learning_rate": 1.073429943170679e-06, "loss": 0.1195, "step": 46989 }, { "epoch": 34.42490842490842, "grad_norm": 1.2421875, "learning_rate": 1.0732435005052047e-06, "loss": 0.1167, "step": 46990 }, { "epoch": 34.42564102564103, "grad_norm": 0.95703125, "learning_rate": 1.0730572946595503e-06, "loss": 0.1083, "step": 46991 }, { "epoch": 34.426373626373625, "grad_norm": 1.0546875, "learning_rate": 1.0728713256345947e-06, "loss": 0.1722, "step": 46992 }, { "epoch": 34.42710622710623, "grad_norm": 1.015625, "learning_rate": 1.072685593431245e-06, "loss": 0.1336, "step": 46993 }, { "epoch": 34.427838827838826, "grad_norm": 0.87109375, "learning_rate": 1.0725000980503805e-06, "loss": 0.1156, "step": 46994 }, { "epoch": 34.42857142857143, "grad_norm": 0.98046875, "learning_rate": 1.0723148394928968e-06, "loss": 0.0866, "step": 46995 }, { "epoch": 34.42930402930403, "grad_norm": 0.953125, "learning_rate": 1.072129817759679e-06, "loss": 0.1298, "step": 46996 }, { "epoch": 34.43003663003663, "grad_norm": 0.89453125, "learning_rate": 1.0719450328516115e-06, "loss": 0.1523, "step": 46997 }, { "epoch": 34.43076923076923, "grad_norm": 0.84765625, "learning_rate": 1.0717604847695848e-06, "loss": 0.1206, "step": 46998 }, { "epoch": 34.43150183150183, "grad_norm": 1.484375, "learning_rate": 1.0715761735144784e-06, "loss": 0.1131, "step": 46999 }, { "epoch": 34.43223443223443, "grad_norm": 0.83984375, "learning_rate": 1.071392099087171e-06, "loss": 0.1026, "step": 47000 }, { "epoch": 34.432967032967035, "grad_norm": 1.1953125, "learning_rate": 1.071208261488559e-06, "loss": 0.1359, "step": 47001 }, { "epoch": 34.43369963369963, "grad_norm": 1.09375, "learning_rate": 1.0710246607195103e-06, "loss": 0.1158, "step": 47002 }, { "epoch": 34.434432234432236, "grad_norm": 1.171875, "learning_rate": 1.07084129678091e-06, "loss": 0.1282, "step": 47003 }, { "epoch": 34.435164835164834, "grad_norm": 0.66796875, "learning_rate": 1.0706581696736372e-06, "loss": 0.1054, "step": 47004 }, { "epoch": 34.43589743589744, "grad_norm": 1.234375, "learning_rate": 1.0704752793985656e-06, "loss": 0.1052, "step": 47005 }, { "epoch": 34.436630036630035, "grad_norm": 0.8984375, "learning_rate": 1.0702926259565693e-06, "loss": 0.103, "step": 47006 }, { "epoch": 34.43736263736264, "grad_norm": 0.9140625, "learning_rate": 1.070110209348533e-06, "loss": 0.0949, "step": 47007 }, { "epoch": 34.43809523809524, "grad_norm": 1.2265625, "learning_rate": 1.069928029575325e-06, "loss": 0.1287, "step": 47008 }, { "epoch": 34.43882783882784, "grad_norm": 1.265625, "learning_rate": 1.069746086637819e-06, "loss": 0.0934, "step": 47009 }, { "epoch": 34.43956043956044, "grad_norm": 0.67578125, "learning_rate": 1.0695643805368832e-06, "loss": 0.0923, "step": 47010 }, { "epoch": 34.44029304029304, "grad_norm": 1.21875, "learning_rate": 1.0693829112733973e-06, "loss": 0.1379, "step": 47011 }, { "epoch": 34.44102564102564, "grad_norm": 0.859375, "learning_rate": 1.0692016788482235e-06, "loss": 0.1302, "step": 47012 }, { "epoch": 34.441758241758244, "grad_norm": 1.1328125, "learning_rate": 1.0690206832622251e-06, "loss": 0.1074, "step": 47013 }, { "epoch": 34.44249084249084, "grad_norm": 1.578125, "learning_rate": 1.0688399245162816e-06, "loss": 0.1345, "step": 47014 }, { "epoch": 34.443223443223445, "grad_norm": 0.69921875, "learning_rate": 1.0686594026112496e-06, "loss": 0.1171, "step": 47015 }, { "epoch": 34.44395604395604, "grad_norm": 0.7109375, "learning_rate": 1.0684791175480034e-06, "loss": 0.1158, "step": 47016 }, { "epoch": 34.44468864468865, "grad_norm": 0.92578125, "learning_rate": 1.0682990693273947e-06, "loss": 0.1062, "step": 47017 }, { "epoch": 34.445421245421244, "grad_norm": 1.0703125, "learning_rate": 1.0681192579502973e-06, "loss": 0.1184, "step": 47018 }, { "epoch": 34.44615384615385, "grad_norm": 0.9140625, "learning_rate": 1.0679396834175631e-06, "loss": 0.1407, "step": 47019 }, { "epoch": 34.446886446886445, "grad_norm": 0.7421875, "learning_rate": 1.0677603457300602e-06, "loss": 0.103, "step": 47020 }, { "epoch": 34.44761904761905, "grad_norm": 0.91015625, "learning_rate": 1.067581244888646e-06, "loss": 0.1024, "step": 47021 }, { "epoch": 34.44835164835165, "grad_norm": 1.453125, "learning_rate": 1.0674023808941783e-06, "loss": 0.1458, "step": 47022 }, { "epoch": 34.44908424908425, "grad_norm": 0.71484375, "learning_rate": 1.0672237537475134e-06, "loss": 0.1223, "step": 47023 }, { "epoch": 34.44981684981685, "grad_norm": 1.7109375, "learning_rate": 1.0670453634495041e-06, "loss": 0.1169, "step": 47024 }, { "epoch": 34.45054945054945, "grad_norm": 1.5390625, "learning_rate": 1.066867210001007e-06, "loss": 0.1303, "step": 47025 }, { "epoch": 34.45128205128205, "grad_norm": 1.1484375, "learning_rate": 1.06668929340288e-06, "loss": 0.1489, "step": 47026 }, { "epoch": 34.452014652014654, "grad_norm": 1.0859375, "learning_rate": 1.0665116136559745e-06, "loss": 0.1051, "step": 47027 }, { "epoch": 34.45274725274725, "grad_norm": 1.03125, "learning_rate": 1.0663341707611367e-06, "loss": 0.1243, "step": 47028 }, { "epoch": 34.453479853479855, "grad_norm": 1.1875, "learning_rate": 1.066156964719219e-06, "loss": 0.1483, "step": 47029 }, { "epoch": 34.45421245421245, "grad_norm": 1.0625, "learning_rate": 1.0659799955310727e-06, "loss": 0.1081, "step": 47030 }, { "epoch": 34.45494505494506, "grad_norm": 1.0546875, "learning_rate": 1.0658032631975445e-06, "loss": 0.1194, "step": 47031 }, { "epoch": 34.455677655677654, "grad_norm": 0.94921875, "learning_rate": 1.0656267677194804e-06, "loss": 0.1098, "step": 47032 }, { "epoch": 34.45641025641026, "grad_norm": 0.953125, "learning_rate": 1.065450509097727e-06, "loss": 0.1464, "step": 47033 }, { "epoch": 34.457142857142856, "grad_norm": 0.984375, "learning_rate": 1.0652744873331303e-06, "loss": 0.1155, "step": 47034 }, { "epoch": 34.45787545787546, "grad_norm": 1.0390625, "learning_rate": 1.0650987024265314e-06, "loss": 0.1064, "step": 47035 }, { "epoch": 34.45860805860806, "grad_norm": 1.71875, "learning_rate": 1.0649231543787712e-06, "loss": 0.1197, "step": 47036 }, { "epoch": 34.45934065934066, "grad_norm": 1.71875, "learning_rate": 1.0647478431906957e-06, "loss": 0.1242, "step": 47037 }, { "epoch": 34.46007326007326, "grad_norm": 0.88671875, "learning_rate": 1.0645727688631404e-06, "loss": 0.1138, "step": 47038 }, { "epoch": 34.46080586080586, "grad_norm": 0.984375, "learning_rate": 1.0643979313969463e-06, "loss": 0.1211, "step": 47039 }, { "epoch": 34.46153846153846, "grad_norm": 0.97265625, "learning_rate": 1.0642233307929485e-06, "loss": 0.1179, "step": 47040 }, { "epoch": 34.462271062271064, "grad_norm": 1.421875, "learning_rate": 1.0640489670519822e-06, "loss": 0.1321, "step": 47041 }, { "epoch": 34.46300366300366, "grad_norm": 1.09375, "learning_rate": 1.063874840174894e-06, "loss": 0.1264, "step": 47042 }, { "epoch": 34.463736263736266, "grad_norm": 1.4140625, "learning_rate": 1.0637009501625028e-06, "loss": 0.167, "step": 47043 }, { "epoch": 34.46446886446886, "grad_norm": 1.0078125, "learning_rate": 1.0635272970156545e-06, "loss": 0.1206, "step": 47044 }, { "epoch": 34.46520146520147, "grad_norm": 0.8046875, "learning_rate": 1.063353880735174e-06, "loss": 0.1364, "step": 47045 }, { "epoch": 34.465934065934064, "grad_norm": 0.9765625, "learning_rate": 1.0631807013218908e-06, "loss": 0.1161, "step": 47046 }, { "epoch": 34.46666666666667, "grad_norm": 1.15625, "learning_rate": 1.0630077587766403e-06, "loss": 0.0998, "step": 47047 }, { "epoch": 34.467399267399266, "grad_norm": 1.5, "learning_rate": 1.062835053100252e-06, "loss": 0.1588, "step": 47048 }, { "epoch": 34.46813186813187, "grad_norm": 0.94921875, "learning_rate": 1.0626625842935454e-06, "loss": 0.1195, "step": 47049 }, { "epoch": 34.46886446886447, "grad_norm": 1.140625, "learning_rate": 1.0624903523573555e-06, "loss": 0.1276, "step": 47050 }, { "epoch": 34.46959706959707, "grad_norm": 1.2734375, "learning_rate": 1.062318357292501e-06, "loss": 0.1323, "step": 47051 }, { "epoch": 34.47032967032967, "grad_norm": 0.90234375, "learning_rate": 1.062146599099812e-06, "loss": 0.1045, "step": 47052 }, { "epoch": 34.47106227106227, "grad_norm": 0.76953125, "learning_rate": 1.0619750777801015e-06, "loss": 0.1177, "step": 47053 }, { "epoch": 34.47179487179487, "grad_norm": 0.87890625, "learning_rate": 1.0618037933342054e-06, "loss": 0.1113, "step": 47054 }, { "epoch": 34.472527472527474, "grad_norm": 1.1796875, "learning_rate": 1.061632745762931e-06, "loss": 0.1347, "step": 47055 }, { "epoch": 34.47326007326007, "grad_norm": 1.078125, "learning_rate": 1.0614619350671083e-06, "loss": 0.1147, "step": 47056 }, { "epoch": 34.473992673992676, "grad_norm": 1.09375, "learning_rate": 1.0612913612475454e-06, "loss": 0.1272, "step": 47057 }, { "epoch": 34.47472527472527, "grad_norm": 0.66796875, "learning_rate": 1.0611210243050718e-06, "loss": 0.118, "step": 47058 }, { "epoch": 34.47545787545788, "grad_norm": 0.6484375, "learning_rate": 1.0609509242404902e-06, "loss": 0.0953, "step": 47059 }, { "epoch": 34.476190476190474, "grad_norm": 0.75, "learning_rate": 1.0607810610546301e-06, "loss": 0.1038, "step": 47060 }, { "epoch": 34.47692307692308, "grad_norm": 1.3125, "learning_rate": 1.0606114347482884e-06, "loss": 0.1123, "step": 47061 }, { "epoch": 34.477655677655676, "grad_norm": 1.0390625, "learning_rate": 1.0604420453222952e-06, "loss": 0.1314, "step": 47062 }, { "epoch": 34.47838827838828, "grad_norm": 1.15625, "learning_rate": 1.0602728927774528e-06, "loss": 0.1283, "step": 47063 }, { "epoch": 34.47912087912088, "grad_norm": 1.4609375, "learning_rate": 1.0601039771145686e-06, "loss": 0.1233, "step": 47064 }, { "epoch": 34.47985347985348, "grad_norm": 1.15625, "learning_rate": 1.059935298334462e-06, "loss": 0.1127, "step": 47065 }, { "epoch": 34.48058608058608, "grad_norm": 0.828125, "learning_rate": 1.0597668564379296e-06, "loss": 0.1129, "step": 47066 }, { "epoch": 34.48131868131868, "grad_norm": 0.90234375, "learning_rate": 1.0595986514257904e-06, "loss": 0.1234, "step": 47067 }, { "epoch": 34.48205128205128, "grad_norm": 1.1015625, "learning_rate": 1.059430683298841e-06, "loss": 0.1478, "step": 47068 }, { "epoch": 34.482783882783885, "grad_norm": 0.8671875, "learning_rate": 1.0592629520578897e-06, "loss": 0.1152, "step": 47069 }, { "epoch": 34.48351648351648, "grad_norm": 0.921875, "learning_rate": 1.0590954577037384e-06, "loss": 0.0969, "step": 47070 }, { "epoch": 34.484249084249086, "grad_norm": 1.78125, "learning_rate": 1.058928200237195e-06, "loss": 0.1256, "step": 47071 }, { "epoch": 34.48498168498168, "grad_norm": 1.4765625, "learning_rate": 1.0587611796590565e-06, "loss": 0.1294, "step": 47072 }, { "epoch": 34.48571428571429, "grad_norm": 0.7578125, "learning_rate": 1.05859439597012e-06, "loss": 0.1257, "step": 47073 }, { "epoch": 34.486446886446885, "grad_norm": 0.9453125, "learning_rate": 1.058427849171193e-06, "loss": 0.1421, "step": 47074 }, { "epoch": 34.48717948717949, "grad_norm": 1.09375, "learning_rate": 1.0582615392630669e-06, "loss": 0.1209, "step": 47075 }, { "epoch": 34.487912087912086, "grad_norm": 0.89453125, "learning_rate": 1.0580954662465386e-06, "loss": 0.13, "step": 47076 }, { "epoch": 34.48864468864469, "grad_norm": 1.359375, "learning_rate": 1.0579296301224051e-06, "loss": 0.1592, "step": 47077 }, { "epoch": 34.48937728937729, "grad_norm": 1.1640625, "learning_rate": 1.0577640308914632e-06, "loss": 0.1228, "step": 47078 }, { "epoch": 34.49010989010989, "grad_norm": 1.1953125, "learning_rate": 1.0575986685545043e-06, "loss": 0.1445, "step": 47079 }, { "epoch": 34.49084249084249, "grad_norm": 0.9375, "learning_rate": 1.0574335431123194e-06, "loss": 0.1171, "step": 47080 }, { "epoch": 34.49157509157509, "grad_norm": 0.6953125, "learning_rate": 1.057268654565706e-06, "loss": 0.0959, "step": 47081 }, { "epoch": 34.49230769230769, "grad_norm": 0.7578125, "learning_rate": 1.057104002915444e-06, "loss": 0.1082, "step": 47082 }, { "epoch": 34.493040293040295, "grad_norm": 1.15625, "learning_rate": 1.0569395881623358e-06, "loss": 0.0951, "step": 47083 }, { "epoch": 34.49377289377289, "grad_norm": 1.484375, "learning_rate": 1.0567754103071508e-06, "loss": 0.1641, "step": 47084 }, { "epoch": 34.494505494505496, "grad_norm": 1.1640625, "learning_rate": 1.0566114693506917e-06, "loss": 0.108, "step": 47085 }, { "epoch": 34.49523809523809, "grad_norm": 1.1953125, "learning_rate": 1.0564477652937384e-06, "loss": 0.1312, "step": 47086 }, { "epoch": 34.4959706959707, "grad_norm": 1.140625, "learning_rate": 1.0562842981370717e-06, "loss": 0.1355, "step": 47087 }, { "epoch": 34.496703296703295, "grad_norm": 1.078125, "learning_rate": 1.0561210678814826e-06, "loss": 0.1535, "step": 47088 }, { "epoch": 34.4974358974359, "grad_norm": 1.0859375, "learning_rate": 1.055958074527746e-06, "loss": 0.1317, "step": 47089 }, { "epoch": 34.498168498168496, "grad_norm": 1.0234375, "learning_rate": 1.0557953180766483e-06, "loss": 0.1374, "step": 47090 }, { "epoch": 34.4989010989011, "grad_norm": 1.078125, "learning_rate": 1.0556327985289637e-06, "loss": 0.1433, "step": 47091 }, { "epoch": 34.4996336996337, "grad_norm": 2.421875, "learning_rate": 1.0554705158854785e-06, "loss": 0.1575, "step": 47092 }, { "epoch": 34.5003663003663, "grad_norm": 1.0625, "learning_rate": 1.0553084701469618e-06, "loss": 0.1112, "step": 47093 }, { "epoch": 34.5010989010989, "grad_norm": 1.2265625, "learning_rate": 1.0551466613141942e-06, "loss": 0.1243, "step": 47094 }, { "epoch": 34.501831501831504, "grad_norm": 0.578125, "learning_rate": 1.0549850893879505e-06, "loss": 0.0962, "step": 47095 }, { "epoch": 34.5025641025641, "grad_norm": 0.73046875, "learning_rate": 1.0548237543690055e-06, "loss": 0.1104, "step": 47096 }, { "epoch": 34.503296703296705, "grad_norm": 1.140625, "learning_rate": 1.0546626562581342e-06, "loss": 0.0995, "step": 47097 }, { "epoch": 34.5040293040293, "grad_norm": 0.87109375, "learning_rate": 1.0545017950561006e-06, "loss": 0.1122, "step": 47098 }, { "epoch": 34.50476190476191, "grad_norm": 0.984375, "learning_rate": 1.0543411707636845e-06, "loss": 0.1197, "step": 47099 }, { "epoch": 34.505494505494504, "grad_norm": 0.79296875, "learning_rate": 1.0541807833816507e-06, "loss": 0.1115, "step": 47100 }, { "epoch": 34.50622710622711, "grad_norm": 1.2578125, "learning_rate": 1.0540206329107678e-06, "loss": 0.1047, "step": 47101 }, { "epoch": 34.506959706959705, "grad_norm": 1.328125, "learning_rate": 1.0538607193518057e-06, "loss": 0.116, "step": 47102 }, { "epoch": 34.50769230769231, "grad_norm": 1.546875, "learning_rate": 1.0537010427055278e-06, "loss": 0.1533, "step": 47103 }, { "epoch": 34.50842490842491, "grad_norm": 0.82421875, "learning_rate": 1.053541602972704e-06, "loss": 0.1033, "step": 47104 }, { "epoch": 34.50915750915751, "grad_norm": 0.87109375, "learning_rate": 1.0533824001540923e-06, "loss": 0.1019, "step": 47105 }, { "epoch": 34.50989010989011, "grad_norm": 0.96875, "learning_rate": 1.0532234342504569e-06, "loss": 0.1357, "step": 47106 }, { "epoch": 34.51062271062271, "grad_norm": 1.1171875, "learning_rate": 1.0530647052625613e-06, "loss": 0.1234, "step": 47107 }, { "epoch": 34.51135531135531, "grad_norm": 1.4296875, "learning_rate": 1.052906213191159e-06, "loss": 0.1329, "step": 47108 }, { "epoch": 34.512087912087914, "grad_norm": 1.859375, "learning_rate": 1.052747958037024e-06, "loss": 0.1098, "step": 47109 }, { "epoch": 34.51282051282051, "grad_norm": 1.328125, "learning_rate": 1.0525899398009045e-06, "loss": 0.1432, "step": 47110 }, { "epoch": 34.513553113553115, "grad_norm": 0.9609375, "learning_rate": 1.0524321584835532e-06, "loss": 0.136, "step": 47111 }, { "epoch": 34.51428571428571, "grad_norm": 0.97265625, "learning_rate": 1.052274614085739e-06, "loss": 0.1166, "step": 47112 }, { "epoch": 34.51501831501832, "grad_norm": 1.4921875, "learning_rate": 1.0521173066082045e-06, "loss": 0.1383, "step": 47113 }, { "epoch": 34.515750915750914, "grad_norm": 0.92578125, "learning_rate": 1.051960236051713e-06, "loss": 0.1168, "step": 47114 }, { "epoch": 34.51648351648352, "grad_norm": 1.0390625, "learning_rate": 1.0518034024170126e-06, "loss": 0.1443, "step": 47115 }, { "epoch": 34.517216117216115, "grad_norm": 1.7890625, "learning_rate": 1.05164680570485e-06, "loss": 0.1203, "step": 47116 }, { "epoch": 34.51794871794872, "grad_norm": 0.7578125, "learning_rate": 1.0514904459159842e-06, "loss": 0.1151, "step": 47117 }, { "epoch": 34.51868131868132, "grad_norm": 1.2578125, "learning_rate": 1.051334323051162e-06, "loss": 0.1525, "step": 47118 }, { "epoch": 34.51941391941392, "grad_norm": 1.265625, "learning_rate": 1.0511784371111257e-06, "loss": 0.1259, "step": 47119 }, { "epoch": 34.52014652014652, "grad_norm": 1.28125, "learning_rate": 1.051022788096628e-06, "loss": 0.1175, "step": 47120 }, { "epoch": 34.52087912087912, "grad_norm": 1.0234375, "learning_rate": 1.0508673760084163e-06, "loss": 0.1057, "step": 47121 }, { "epoch": 34.52161172161172, "grad_norm": 1.0, "learning_rate": 1.0507122008472273e-06, "loss": 0.1201, "step": 47122 }, { "epoch": 34.522344322344324, "grad_norm": 1.1171875, "learning_rate": 1.0505572626138136e-06, "loss": 0.1542, "step": 47123 }, { "epoch": 34.52307692307692, "grad_norm": 1.078125, "learning_rate": 1.0504025613089119e-06, "loss": 0.1294, "step": 47124 }, { "epoch": 34.523809523809526, "grad_norm": 1.203125, "learning_rate": 1.0502480969332638e-06, "loss": 0.1439, "step": 47125 }, { "epoch": 34.52454212454212, "grad_norm": 0.875, "learning_rate": 1.0500938694876113e-06, "loss": 0.1286, "step": 47126 }, { "epoch": 34.52527472527473, "grad_norm": 0.92578125, "learning_rate": 1.0499398789726906e-06, "loss": 0.1277, "step": 47127 }, { "epoch": 34.526007326007324, "grad_norm": 1.34375, "learning_rate": 1.049786125389244e-06, "loss": 0.1255, "step": 47128 }, { "epoch": 34.52673992673993, "grad_norm": 0.7265625, "learning_rate": 1.0496326087380024e-06, "loss": 0.1199, "step": 47129 }, { "epoch": 34.527472527472526, "grad_norm": 1.3828125, "learning_rate": 1.0494793290197074e-06, "loss": 0.1197, "step": 47130 }, { "epoch": 34.52820512820513, "grad_norm": 1.015625, "learning_rate": 1.0493262862350903e-06, "loss": 0.1294, "step": 47131 }, { "epoch": 34.52893772893773, "grad_norm": 0.65625, "learning_rate": 1.0491734803848816e-06, "loss": 0.1095, "step": 47132 }, { "epoch": 34.52967032967033, "grad_norm": 0.91796875, "learning_rate": 1.0490209114698178e-06, "loss": 0.1134, "step": 47133 }, { "epoch": 34.53040293040293, "grad_norm": 1.0625, "learning_rate": 1.0488685794906302e-06, "loss": 0.1095, "step": 47134 }, { "epoch": 34.53113553113553, "grad_norm": 1.03125, "learning_rate": 1.0487164844480437e-06, "loss": 0.0984, "step": 47135 }, { "epoch": 34.53186813186813, "grad_norm": 0.7578125, "learning_rate": 1.048564626342795e-06, "loss": 0.0973, "step": 47136 }, { "epoch": 34.532600732600734, "grad_norm": 1.0078125, "learning_rate": 1.0484130051756042e-06, "loss": 0.1184, "step": 47137 }, { "epoch": 34.53333333333333, "grad_norm": 0.921875, "learning_rate": 1.0482616209472019e-06, "loss": 0.1216, "step": 47138 }, { "epoch": 34.534065934065936, "grad_norm": 0.80078125, "learning_rate": 1.048110473658308e-06, "loss": 0.0864, "step": 47139 }, { "epoch": 34.53479853479853, "grad_norm": 1.2421875, "learning_rate": 1.0479595633096536e-06, "loss": 0.152, "step": 47140 }, { "epoch": 34.53553113553114, "grad_norm": 0.90625, "learning_rate": 1.0478088899019583e-06, "loss": 0.0929, "step": 47141 }, { "epoch": 34.536263736263734, "grad_norm": 0.93359375, "learning_rate": 1.047658453435948e-06, "loss": 0.1251, "step": 47142 }, { "epoch": 34.53699633699634, "grad_norm": 0.7421875, "learning_rate": 1.047508253912337e-06, "loss": 0.1131, "step": 47143 }, { "epoch": 34.537728937728936, "grad_norm": 0.8515625, "learning_rate": 1.0473582913318505e-06, "loss": 0.1345, "step": 47144 }, { "epoch": 34.53846153846154, "grad_norm": 0.8046875, "learning_rate": 1.047208565695203e-06, "loss": 0.1116, "step": 47145 }, { "epoch": 34.53919413919414, "grad_norm": 1.3203125, "learning_rate": 1.0470590770031145e-06, "loss": 0.1193, "step": 47146 }, { "epoch": 34.53992673992674, "grad_norm": 1.2109375, "learning_rate": 1.0469098252562995e-06, "loss": 0.1289, "step": 47147 }, { "epoch": 34.54065934065934, "grad_norm": 0.921875, "learning_rate": 1.0467608104554723e-06, "loss": 0.1134, "step": 47148 }, { "epoch": 34.54139194139194, "grad_norm": 0.83984375, "learning_rate": 1.0466120326013473e-06, "loss": 0.116, "step": 47149 }, { "epoch": 34.54212454212454, "grad_norm": 0.6328125, "learning_rate": 1.0464634916946446e-06, "loss": 0.1083, "step": 47150 }, { "epoch": 34.542857142857144, "grad_norm": 1.1484375, "learning_rate": 1.0463151877360674e-06, "loss": 0.1256, "step": 47151 }, { "epoch": 34.54358974358974, "grad_norm": 1.1171875, "learning_rate": 1.0461671207263246e-06, "loss": 0.1279, "step": 47152 }, { "epoch": 34.544322344322346, "grad_norm": 1.578125, "learning_rate": 1.0460192906661365e-06, "loss": 0.1313, "step": 47153 }, { "epoch": 34.54505494505494, "grad_norm": 0.9765625, "learning_rate": 1.0458716975562007e-06, "loss": 0.1123, "step": 47154 }, { "epoch": 34.54578754578755, "grad_norm": 0.765625, "learning_rate": 1.0457243413972264e-06, "loss": 0.1312, "step": 47155 }, { "epoch": 34.546520146520145, "grad_norm": 0.703125, "learning_rate": 1.045577222189922e-06, "loss": 0.1061, "step": 47156 }, { "epoch": 34.54725274725275, "grad_norm": 0.96875, "learning_rate": 1.0454303399349973e-06, "loss": 0.1183, "step": 47157 }, { "epoch": 34.547985347985346, "grad_norm": 0.90625, "learning_rate": 1.0452836946331443e-06, "loss": 0.1446, "step": 47158 }, { "epoch": 34.54871794871795, "grad_norm": 1.25, "learning_rate": 1.045137286285077e-06, "loss": 0.1373, "step": 47159 }, { "epoch": 34.54945054945055, "grad_norm": 1.359375, "learning_rate": 1.0449911148914885e-06, "loss": 0.1255, "step": 47160 }, { "epoch": 34.55018315018315, "grad_norm": 0.81640625, "learning_rate": 1.0448451804530874e-06, "loss": 0.108, "step": 47161 }, { "epoch": 34.55091575091575, "grad_norm": 0.9375, "learning_rate": 1.0446994829705608e-06, "loss": 0.113, "step": 47162 }, { "epoch": 34.55164835164835, "grad_norm": 1.3046875, "learning_rate": 1.0445540224446173e-06, "loss": 0.1127, "step": 47163 }, { "epoch": 34.55238095238095, "grad_norm": 1.3125, "learning_rate": 1.0444087988759554e-06, "loss": 0.1351, "step": 47164 }, { "epoch": 34.553113553113555, "grad_norm": 1.2734375, "learning_rate": 1.0442638122652614e-06, "loss": 0.1268, "step": 47165 }, { "epoch": 34.55384615384615, "grad_norm": 0.890625, "learning_rate": 1.044119062613234e-06, "loss": 0.0982, "step": 47166 }, { "epoch": 34.554578754578756, "grad_norm": 0.828125, "learning_rate": 1.0439745499205706e-06, "loss": 0.1188, "step": 47167 }, { "epoch": 34.55531135531135, "grad_norm": 0.56640625, "learning_rate": 1.043830274187958e-06, "loss": 0.1077, "step": 47168 }, { "epoch": 34.55604395604396, "grad_norm": 0.94921875, "learning_rate": 1.0436862354160893e-06, "loss": 0.1393, "step": 47169 }, { "epoch": 34.556776556776555, "grad_norm": 0.66015625, "learning_rate": 1.0435424336056564e-06, "loss": 0.0999, "step": 47170 }, { "epoch": 34.55750915750916, "grad_norm": 1.0703125, "learning_rate": 1.043398868757352e-06, "loss": 0.1019, "step": 47171 }, { "epoch": 34.558241758241756, "grad_norm": 1.078125, "learning_rate": 1.0432555408718518e-06, "loss": 0.112, "step": 47172 }, { "epoch": 34.55897435897436, "grad_norm": 0.81640625, "learning_rate": 1.043112449949854e-06, "loss": 0.102, "step": 47173 }, { "epoch": 34.55970695970696, "grad_norm": 1.0078125, "learning_rate": 1.04296959599204e-06, "loss": 0.1201, "step": 47174 }, { "epoch": 34.56043956043956, "grad_norm": 0.91015625, "learning_rate": 1.0428269789990913e-06, "loss": 0.1156, "step": 47175 }, { "epoch": 34.56117216117216, "grad_norm": 0.9375, "learning_rate": 1.0426845989716946e-06, "loss": 0.0971, "step": 47176 }, { "epoch": 34.56190476190476, "grad_norm": 1.703125, "learning_rate": 1.0425424559105373e-06, "loss": 0.1167, "step": 47177 }, { "epoch": 34.56263736263736, "grad_norm": 0.9140625, "learning_rate": 1.0424005498162894e-06, "loss": 0.1344, "step": 47178 }, { "epoch": 34.563369963369965, "grad_norm": 1.0546875, "learning_rate": 1.042258880689638e-06, "loss": 0.1148, "step": 47179 }, { "epoch": 34.56410256410256, "grad_norm": 1.5, "learning_rate": 1.0421174485312592e-06, "loss": 0.1205, "step": 47180 }, { "epoch": 34.564835164835166, "grad_norm": 0.90234375, "learning_rate": 1.0419762533418344e-06, "loss": 0.1205, "step": 47181 }, { "epoch": 34.565567765567764, "grad_norm": 0.58203125, "learning_rate": 1.0418352951220339e-06, "loss": 0.1122, "step": 47182 }, { "epoch": 34.56630036630037, "grad_norm": 0.90625, "learning_rate": 1.0416945738725395e-06, "loss": 0.1216, "step": 47183 }, { "epoch": 34.567032967032965, "grad_norm": 1.125, "learning_rate": 1.0415540895940212e-06, "loss": 0.127, "step": 47184 }, { "epoch": 34.56776556776557, "grad_norm": 1.1953125, "learning_rate": 1.0414138422871553e-06, "loss": 0.1558, "step": 47185 }, { "epoch": 34.56849816849817, "grad_norm": 1.0625, "learning_rate": 1.0412738319526066e-06, "loss": 0.1185, "step": 47186 }, { "epoch": 34.56923076923077, "grad_norm": 1.2734375, "learning_rate": 1.0411340585910568e-06, "loss": 0.1449, "step": 47187 }, { "epoch": 34.56996336996337, "grad_norm": 1.15625, "learning_rate": 1.040994522203165e-06, "loss": 0.0948, "step": 47188 }, { "epoch": 34.57069597069597, "grad_norm": 0.8125, "learning_rate": 1.0408552227896076e-06, "loss": 0.1208, "step": 47189 }, { "epoch": 34.57142857142857, "grad_norm": 1.53125, "learning_rate": 1.040716160351049e-06, "loss": 0.1287, "step": 47190 }, { "epoch": 34.572161172161174, "grad_norm": 0.76171875, "learning_rate": 1.0405773348881547e-06, "loss": 0.1095, "step": 47191 }, { "epoch": 34.57289377289377, "grad_norm": 1.0546875, "learning_rate": 1.0404387464015948e-06, "loss": 0.0915, "step": 47192 }, { "epoch": 34.573626373626375, "grad_norm": 0.8671875, "learning_rate": 1.0403003948920292e-06, "loss": 0.0855, "step": 47193 }, { "epoch": 34.57435897435897, "grad_norm": 0.84375, "learning_rate": 1.0401622803601169e-06, "loss": 0.1224, "step": 47194 }, { "epoch": 34.57509157509158, "grad_norm": 1.21875, "learning_rate": 1.040024402806523e-06, "loss": 0.1523, "step": 47195 }, { "epoch": 34.575824175824174, "grad_norm": 1.28125, "learning_rate": 1.0398867622319128e-06, "loss": 0.1108, "step": 47196 }, { "epoch": 34.57655677655678, "grad_norm": 0.60546875, "learning_rate": 1.0397493586369398e-06, "loss": 0.1309, "step": 47197 }, { "epoch": 34.577289377289375, "grad_norm": 0.84765625, "learning_rate": 1.0396121920222638e-06, "loss": 0.085, "step": 47198 }, { "epoch": 34.57802197802198, "grad_norm": 0.6484375, "learning_rate": 1.039475262388544e-06, "loss": 0.0962, "step": 47199 }, { "epoch": 34.57875457875458, "grad_norm": 1.015625, "learning_rate": 1.0393385697364348e-06, "loss": 0.1246, "step": 47200 }, { "epoch": 34.57948717948718, "grad_norm": 0.953125, "learning_rate": 1.0392021140665897e-06, "loss": 0.0933, "step": 47201 }, { "epoch": 34.58021978021978, "grad_norm": 1.078125, "learning_rate": 1.0390658953796685e-06, "loss": 0.1357, "step": 47202 }, { "epoch": 34.58095238095238, "grad_norm": 1.140625, "learning_rate": 1.038929913676314e-06, "loss": 0.1151, "step": 47203 }, { "epoch": 34.58168498168498, "grad_norm": 1.2578125, "learning_rate": 1.0387941689571857e-06, "loss": 0.0991, "step": 47204 }, { "epoch": 34.582417582417584, "grad_norm": 1.2265625, "learning_rate": 1.038658661222932e-06, "loss": 0.1298, "step": 47205 }, { "epoch": 34.58315018315018, "grad_norm": 1.0625, "learning_rate": 1.0385233904742019e-06, "loss": 0.1142, "step": 47206 }, { "epoch": 34.583882783882785, "grad_norm": 0.921875, "learning_rate": 1.0383883567116432e-06, "loss": 0.1188, "step": 47207 }, { "epoch": 34.58461538461538, "grad_norm": 0.81640625, "learning_rate": 1.0382535599359048e-06, "loss": 0.1348, "step": 47208 }, { "epoch": 34.58534798534799, "grad_norm": 1.0, "learning_rate": 1.0381190001476293e-06, "loss": 0.1018, "step": 47209 }, { "epoch": 34.586080586080584, "grad_norm": 1.1171875, "learning_rate": 1.0379846773474602e-06, "loss": 0.1261, "step": 47210 }, { "epoch": 34.58681318681319, "grad_norm": 1.078125, "learning_rate": 1.0378505915360457e-06, "loss": 0.1366, "step": 47211 }, { "epoch": 34.587545787545785, "grad_norm": 0.75390625, "learning_rate": 1.0377167427140286e-06, "loss": 0.1286, "step": 47212 }, { "epoch": 34.58827838827839, "grad_norm": 1.6640625, "learning_rate": 1.0375831308820468e-06, "loss": 0.115, "step": 47213 }, { "epoch": 34.58901098901099, "grad_norm": 1.25, "learning_rate": 1.0374497560407377e-06, "loss": 0.1361, "step": 47214 }, { "epoch": 34.58974358974359, "grad_norm": 0.91796875, "learning_rate": 1.0373166181907494e-06, "loss": 0.1178, "step": 47215 }, { "epoch": 34.59047619047619, "grad_norm": 0.88671875, "learning_rate": 1.0371837173327142e-06, "loss": 0.1107, "step": 47216 }, { "epoch": 34.59120879120879, "grad_norm": 1.1015625, "learning_rate": 1.0370510534672642e-06, "loss": 0.1172, "step": 47217 }, { "epoch": 34.59194139194139, "grad_norm": 1.015625, "learning_rate": 1.0369186265950476e-06, "loss": 0.118, "step": 47218 }, { "epoch": 34.592673992673994, "grad_norm": 1.2109375, "learning_rate": 1.0367864367166857e-06, "loss": 0.1172, "step": 47219 }, { "epoch": 34.59340659340659, "grad_norm": 1.5078125, "learning_rate": 1.0366544838328213e-06, "loss": 0.1129, "step": 47220 }, { "epoch": 34.594139194139196, "grad_norm": 1.0234375, "learning_rate": 1.0365227679440807e-06, "loss": 0.127, "step": 47221 }, { "epoch": 34.59487179487179, "grad_norm": 1.8125, "learning_rate": 1.036391289051102e-06, "loss": 0.1381, "step": 47222 }, { "epoch": 34.5956043956044, "grad_norm": 1.0390625, "learning_rate": 1.0362600471545055e-06, "loss": 0.1043, "step": 47223 }, { "epoch": 34.596336996336994, "grad_norm": 1.328125, "learning_rate": 1.036129042254929e-06, "loss": 0.1235, "step": 47224 }, { "epoch": 34.5970695970696, "grad_norm": 0.91015625, "learning_rate": 1.0359982743529936e-06, "loss": 0.11, "step": 47225 }, { "epoch": 34.597802197802196, "grad_norm": 1.25, "learning_rate": 1.0358677434493259e-06, "loss": 0.1392, "step": 47226 }, { "epoch": 34.5985347985348, "grad_norm": 1.1640625, "learning_rate": 1.0357374495445574e-06, "loss": 0.1447, "step": 47227 }, { "epoch": 34.5992673992674, "grad_norm": 0.8828125, "learning_rate": 1.035607392639304e-06, "loss": 0.1103, "step": 47228 }, { "epoch": 34.6, "grad_norm": 1.25, "learning_rate": 1.0354775727342031e-06, "loss": 0.1323, "step": 47229 }, { "epoch": 34.6007326007326, "grad_norm": 0.87109375, "learning_rate": 1.0353479898298593e-06, "loss": 0.1273, "step": 47230 }, { "epoch": 34.6014652014652, "grad_norm": 0.94140625, "learning_rate": 1.0352186439269044e-06, "loss": 0.1052, "step": 47231 }, { "epoch": 34.6021978021978, "grad_norm": 0.74609375, "learning_rate": 1.035089535025954e-06, "loss": 0.1138, "step": 47232 }, { "epoch": 34.602930402930404, "grad_norm": 0.91015625, "learning_rate": 1.0349606631276294e-06, "loss": 0.1299, "step": 47233 }, { "epoch": 34.603663003663, "grad_norm": 1.390625, "learning_rate": 1.0348320282325457e-06, "loss": 0.1132, "step": 47234 }, { "epoch": 34.604395604395606, "grad_norm": 1.015625, "learning_rate": 1.0347036303413183e-06, "loss": 0.1025, "step": 47235 }, { "epoch": 34.6051282051282, "grad_norm": 1.2265625, "learning_rate": 1.0345754694545688e-06, "loss": 0.113, "step": 47236 }, { "epoch": 34.60586080586081, "grad_norm": 0.82421875, "learning_rate": 1.0344475455729013e-06, "loss": 0.1092, "step": 47237 }, { "epoch": 34.606593406593404, "grad_norm": 0.71875, "learning_rate": 1.0343198586969368e-06, "loss": 0.089, "step": 47238 }, { "epoch": 34.60732600732601, "grad_norm": 0.80078125, "learning_rate": 1.0341924088272852e-06, "loss": 0.1368, "step": 47239 }, { "epoch": 34.608058608058606, "grad_norm": 1.0234375, "learning_rate": 1.0340651959645567e-06, "loss": 0.1238, "step": 47240 }, { "epoch": 34.60879120879121, "grad_norm": 1.03125, "learning_rate": 1.033938220109356e-06, "loss": 0.1055, "step": 47241 }, { "epoch": 34.60952380952381, "grad_norm": 1.171875, "learning_rate": 1.0338114812623039e-06, "loss": 0.1245, "step": 47242 }, { "epoch": 34.61025641025641, "grad_norm": 1.0546875, "learning_rate": 1.0336849794239938e-06, "loss": 0.1341, "step": 47243 }, { "epoch": 34.61098901098901, "grad_norm": 0.953125, "learning_rate": 1.033558714595036e-06, "loss": 0.0936, "step": 47244 }, { "epoch": 34.61172161172161, "grad_norm": 0.69921875, "learning_rate": 1.0334326867760403e-06, "loss": 0.1122, "step": 47245 }, { "epoch": 34.61245421245421, "grad_norm": 0.96484375, "learning_rate": 1.0333068959676113e-06, "loss": 0.1307, "step": 47246 }, { "epoch": 34.613186813186815, "grad_norm": 1.5546875, "learning_rate": 1.0331813421703423e-06, "loss": 0.1328, "step": 47247 }, { "epoch": 34.61391941391941, "grad_norm": 1.125, "learning_rate": 1.0330560253848381e-06, "loss": 0.1085, "step": 47248 }, { "epoch": 34.614652014652016, "grad_norm": 0.81640625, "learning_rate": 1.0329309456117088e-06, "loss": 0.1217, "step": 47249 }, { "epoch": 34.61538461538461, "grad_norm": 1.03125, "learning_rate": 1.0328061028515421e-06, "loss": 0.1029, "step": 47250 }, { "epoch": 34.61611721611722, "grad_norm": 1.25, "learning_rate": 1.0326814971049372e-06, "loss": 0.1188, "step": 47251 }, { "epoch": 34.616849816849815, "grad_norm": 1.078125, "learning_rate": 1.0325571283724986e-06, "loss": 0.1162, "step": 47252 }, { "epoch": 34.61758241758242, "grad_norm": 1.5234375, "learning_rate": 1.0324329966548199e-06, "loss": 0.1056, "step": 47253 }, { "epoch": 34.618315018315016, "grad_norm": 1.3203125, "learning_rate": 1.032309101952489e-06, "loss": 0.1391, "step": 47254 }, { "epoch": 34.61904761904762, "grad_norm": 0.8984375, "learning_rate": 1.0321854442661102e-06, "loss": 0.1023, "step": 47255 }, { "epoch": 34.61978021978022, "grad_norm": 0.98828125, "learning_rate": 1.0320620235962612e-06, "loss": 0.1142, "step": 47256 }, { "epoch": 34.62051282051282, "grad_norm": 1.8203125, "learning_rate": 1.0319388399435513e-06, "loss": 0.1148, "step": 47257 }, { "epoch": 34.62124542124542, "grad_norm": 0.62109375, "learning_rate": 1.031815893308558e-06, "loss": 0.0949, "step": 47258 }, { "epoch": 34.62197802197802, "grad_norm": 0.80078125, "learning_rate": 1.0316931836918749e-06, "loss": 0.119, "step": 47259 }, { "epoch": 34.62271062271062, "grad_norm": 2.078125, "learning_rate": 1.0315707110940841e-06, "loss": 0.1514, "step": 47260 }, { "epoch": 34.623443223443225, "grad_norm": 0.69140625, "learning_rate": 1.0314484755157848e-06, "loss": 0.1227, "step": 47261 }, { "epoch": 34.62417582417582, "grad_norm": 1.015625, "learning_rate": 1.031326476957554e-06, "loss": 0.1093, "step": 47262 }, { "epoch": 34.624908424908426, "grad_norm": 0.671875, "learning_rate": 1.0312047154199744e-06, "loss": 0.1399, "step": 47263 }, { "epoch": 34.62564102564102, "grad_norm": 0.94140625, "learning_rate": 1.031083190903634e-06, "loss": 0.1153, "step": 47264 }, { "epoch": 34.62637362637363, "grad_norm": 0.67578125, "learning_rate": 1.0309619034091099e-06, "loss": 0.1033, "step": 47265 }, { "epoch": 34.627106227106225, "grad_norm": 1.3671875, "learning_rate": 1.0308408529369952e-06, "loss": 0.1423, "step": 47266 }, { "epoch": 34.62783882783883, "grad_norm": 1.59375, "learning_rate": 1.0307200394878563e-06, "loss": 0.1338, "step": 47267 }, { "epoch": 34.628571428571426, "grad_norm": 1.1015625, "learning_rate": 1.0305994630622758e-06, "loss": 0.1372, "step": 47268 }, { "epoch": 34.62930402930403, "grad_norm": 1.6875, "learning_rate": 1.0304791236608361e-06, "loss": 0.1295, "step": 47269 }, { "epoch": 34.63003663003663, "grad_norm": 1.1171875, "learning_rate": 1.0303590212841086e-06, "loss": 0.1139, "step": 47270 }, { "epoch": 34.63076923076923, "grad_norm": 0.890625, "learning_rate": 1.0302391559326762e-06, "loss": 0.1132, "step": 47271 }, { "epoch": 34.63150183150183, "grad_norm": 1.046875, "learning_rate": 1.0301195276070991e-06, "loss": 0.1043, "step": 47272 }, { "epoch": 34.63223443223443, "grad_norm": 0.859375, "learning_rate": 1.0300001363079655e-06, "loss": 0.1151, "step": 47273 }, { "epoch": 34.63296703296703, "grad_norm": 1.1328125, "learning_rate": 1.0298809820358415e-06, "loss": 0.1246, "step": 47274 }, { "epoch": 34.633699633699635, "grad_norm": 1.78125, "learning_rate": 1.029762064791293e-06, "loss": 0.1283, "step": 47275 }, { "epoch": 34.63443223443223, "grad_norm": 0.9296875, "learning_rate": 1.029643384574897e-06, "loss": 0.1125, "step": 47276 }, { "epoch": 34.63516483516484, "grad_norm": 0.93359375, "learning_rate": 1.0295249413872198e-06, "loss": 0.136, "step": 47277 }, { "epoch": 34.635897435897434, "grad_norm": 0.94921875, "learning_rate": 1.0294067352288327e-06, "loss": 0.0888, "step": 47278 }, { "epoch": 34.63663003663004, "grad_norm": 0.94921875, "learning_rate": 1.029288766100291e-06, "loss": 0.096, "step": 47279 }, { "epoch": 34.637362637362635, "grad_norm": 0.79296875, "learning_rate": 1.0291710340021716e-06, "loss": 0.1291, "step": 47280 }, { "epoch": 34.63809523809524, "grad_norm": 1.9140625, "learning_rate": 1.0290535389350348e-06, "loss": 0.1621, "step": 47281 }, { "epoch": 34.63882783882784, "grad_norm": 0.93359375, "learning_rate": 1.0289362808994362e-06, "loss": 0.1235, "step": 47282 }, { "epoch": 34.63956043956044, "grad_norm": 0.74609375, "learning_rate": 1.0288192598959524e-06, "loss": 0.0852, "step": 47283 }, { "epoch": 34.64029304029304, "grad_norm": 0.890625, "learning_rate": 1.0287024759251277e-06, "loss": 0.1154, "step": 47284 }, { "epoch": 34.64102564102564, "grad_norm": 1.6484375, "learning_rate": 1.0285859289875387e-06, "loss": 0.1326, "step": 47285 }, { "epoch": 34.64175824175824, "grad_norm": 0.69921875, "learning_rate": 1.02846961908373e-06, "loss": 0.114, "step": 47286 }, { "epoch": 34.642490842490844, "grad_norm": 0.98046875, "learning_rate": 1.0283535462142619e-06, "loss": 0.1226, "step": 47287 }, { "epoch": 34.64322344322344, "grad_norm": 1.15625, "learning_rate": 1.0282377103796952e-06, "loss": 0.1007, "step": 47288 }, { "epoch": 34.643956043956045, "grad_norm": 1.421875, "learning_rate": 1.028122111580579e-06, "loss": 0.0971, "step": 47289 }, { "epoch": 34.64468864468864, "grad_norm": 1.4765625, "learning_rate": 1.0280067498174744e-06, "loss": 0.1238, "step": 47290 }, { "epoch": 34.64542124542125, "grad_norm": 0.83984375, "learning_rate": 1.0278916250909304e-06, "loss": 0.1304, "step": 47291 }, { "epoch": 34.646153846153844, "grad_norm": 1.6328125, "learning_rate": 1.0277767374014971e-06, "loss": 0.1088, "step": 47292 }, { "epoch": 34.64688644688645, "grad_norm": 0.69921875, "learning_rate": 1.0276620867497237e-06, "loss": 0.1239, "step": 47293 }, { "epoch": 34.647619047619045, "grad_norm": 0.7578125, "learning_rate": 1.0275476731361654e-06, "loss": 0.1049, "step": 47294 }, { "epoch": 34.64835164835165, "grad_norm": 1.1953125, "learning_rate": 1.0274334965613606e-06, "loss": 0.1196, "step": 47295 }, { "epoch": 34.64908424908425, "grad_norm": 0.9140625, "learning_rate": 1.02731955702587e-06, "loss": 0.1178, "step": 47296 }, { "epoch": 34.64981684981685, "grad_norm": 1.9609375, "learning_rate": 1.027205854530227e-06, "loss": 0.1647, "step": 47297 }, { "epoch": 34.65054945054945, "grad_norm": 1.75, "learning_rate": 1.027092389074986e-06, "loss": 0.1055, "step": 47298 }, { "epoch": 34.65128205128205, "grad_norm": 0.73828125, "learning_rate": 1.0269791606606858e-06, "loss": 0.1157, "step": 47299 }, { "epoch": 34.65201465201465, "grad_norm": 1.0703125, "learning_rate": 1.026866169287865e-06, "loss": 0.1517, "step": 47300 }, { "epoch": 34.652747252747254, "grad_norm": 1.265625, "learning_rate": 1.0267534149570734e-06, "loss": 0.1047, "step": 47301 }, { "epoch": 34.65347985347985, "grad_norm": 1.0390625, "learning_rate": 1.0266408976688437e-06, "loss": 0.1249, "step": 47302 }, { "epoch": 34.654212454212455, "grad_norm": 1.2734375, "learning_rate": 1.0265286174237259e-06, "loss": 0.1275, "step": 47303 }, { "epoch": 34.65494505494505, "grad_norm": 1.296875, "learning_rate": 1.0264165742222416e-06, "loss": 0.139, "step": 47304 }, { "epoch": 34.65567765567766, "grad_norm": 0.765625, "learning_rate": 1.0263047680649407e-06, "loss": 0.1495, "step": 47305 }, { "epoch": 34.656410256410254, "grad_norm": 1.1328125, "learning_rate": 1.0261931989523561e-06, "loss": 0.1328, "step": 47306 }, { "epoch": 34.65714285714286, "grad_norm": 0.84375, "learning_rate": 1.026081866885021e-06, "loss": 0.1234, "step": 47307 }, { "epoch": 34.657875457875456, "grad_norm": 0.91015625, "learning_rate": 1.0259707718634683e-06, "loss": 0.1217, "step": 47308 }, { "epoch": 34.65860805860806, "grad_norm": 1.3515625, "learning_rate": 1.0258599138882258e-06, "loss": 0.1154, "step": 47309 }, { "epoch": 34.65934065934066, "grad_norm": 1.1015625, "learning_rate": 1.0257492929598373e-06, "loss": 0.1055, "step": 47310 }, { "epoch": 34.66007326007326, "grad_norm": 0.734375, "learning_rate": 1.0256389090788196e-06, "loss": 0.1082, "step": 47311 }, { "epoch": 34.66080586080586, "grad_norm": 0.8515625, "learning_rate": 1.0255287622457112e-06, "loss": 0.122, "step": 47312 }, { "epoch": 34.66153846153846, "grad_norm": 1.1953125, "learning_rate": 1.0254188524610343e-06, "loss": 0.1642, "step": 47313 }, { "epoch": 34.66227106227106, "grad_norm": 1.203125, "learning_rate": 1.0253091797253161e-06, "loss": 0.1688, "step": 47314 }, { "epoch": 34.663003663003664, "grad_norm": 0.91796875, "learning_rate": 1.0251997440390847e-06, "loss": 0.1113, "step": 47315 }, { "epoch": 34.66373626373626, "grad_norm": 1.2109375, "learning_rate": 1.0250905454028618e-06, "loss": 0.1216, "step": 47316 }, { "epoch": 34.664468864468866, "grad_norm": 1.34375, "learning_rate": 1.024981583817175e-06, "loss": 0.1375, "step": 47317 }, { "epoch": 34.66520146520146, "grad_norm": 0.95703125, "learning_rate": 1.024872859282541e-06, "loss": 0.1344, "step": 47318 }, { "epoch": 34.66593406593407, "grad_norm": 1.1484375, "learning_rate": 1.0247643717994767e-06, "loss": 0.108, "step": 47319 }, { "epoch": 34.666666666666664, "grad_norm": 1.0859375, "learning_rate": 1.0246561213685147e-06, "loss": 0.1152, "step": 47320 }, { "epoch": 34.66739926739927, "grad_norm": 0.54296875, "learning_rate": 1.024548107990166e-06, "loss": 0.0942, "step": 47321 }, { "epoch": 34.668131868131866, "grad_norm": 1.046875, "learning_rate": 1.0244403316649477e-06, "loss": 0.169, "step": 47322 }, { "epoch": 34.66886446886447, "grad_norm": 1.140625, "learning_rate": 1.0243327923933814e-06, "loss": 0.0925, "step": 47323 }, { "epoch": 34.66959706959707, "grad_norm": 1.109375, "learning_rate": 1.024225490175973e-06, "loss": 0.1211, "step": 47324 }, { "epoch": 34.67032967032967, "grad_norm": 1.1875, "learning_rate": 1.0241184250132445e-06, "loss": 0.1015, "step": 47325 }, { "epoch": 34.67106227106227, "grad_norm": 0.75390625, "learning_rate": 1.0240115969057071e-06, "loss": 0.1161, "step": 47326 }, { "epoch": 34.67179487179487, "grad_norm": 0.78515625, "learning_rate": 1.0239050058538716e-06, "loss": 0.0998, "step": 47327 }, { "epoch": 34.67252747252747, "grad_norm": 1.0625, "learning_rate": 1.0237986518582493e-06, "loss": 0.1383, "step": 47328 }, { "epoch": 34.673260073260074, "grad_norm": 0.90625, "learning_rate": 1.0236925349193512e-06, "loss": 0.1269, "step": 47329 }, { "epoch": 34.67399267399267, "grad_norm": 0.80859375, "learning_rate": 1.023586655037683e-06, "loss": 0.1202, "step": 47330 }, { "epoch": 34.674725274725276, "grad_norm": 1.3046875, "learning_rate": 1.0234810122137502e-06, "loss": 0.1398, "step": 47331 }, { "epoch": 34.67545787545787, "grad_norm": 1.1875, "learning_rate": 1.0233756064480637e-06, "loss": 0.1423, "step": 47332 }, { "epoch": 34.67619047619048, "grad_norm": 1.1015625, "learning_rate": 1.0232704377411297e-06, "loss": 0.1085, "step": 47333 }, { "epoch": 34.676923076923075, "grad_norm": 1.1328125, "learning_rate": 1.0231655060934424e-06, "loss": 0.1221, "step": 47334 }, { "epoch": 34.67765567765568, "grad_norm": 0.78515625, "learning_rate": 1.0230608115055184e-06, "loss": 0.126, "step": 47335 }, { "epoch": 34.678388278388276, "grad_norm": 0.77734375, "learning_rate": 1.0229563539778468e-06, "loss": 0.1262, "step": 47336 }, { "epoch": 34.67912087912088, "grad_norm": 1.1640625, "learning_rate": 1.0228521335109386e-06, "loss": 0.119, "step": 47337 }, { "epoch": 34.67985347985348, "grad_norm": 1.265625, "learning_rate": 1.0227481501052832e-06, "loss": 0.1096, "step": 47338 }, { "epoch": 34.68058608058608, "grad_norm": 0.9453125, "learning_rate": 1.0226444037613913e-06, "loss": 0.1315, "step": 47339 }, { "epoch": 34.68131868131868, "grad_norm": 1.046875, "learning_rate": 1.022540894479747e-06, "loss": 0.1461, "step": 47340 }, { "epoch": 34.68205128205128, "grad_norm": 1.03125, "learning_rate": 1.02243762226085e-06, "loss": 0.1084, "step": 47341 }, { "epoch": 34.68278388278388, "grad_norm": 1.078125, "learning_rate": 1.0223345871052006e-06, "loss": 0.1248, "step": 47342 }, { "epoch": 34.683516483516485, "grad_norm": 1.3984375, "learning_rate": 1.0222317890132876e-06, "loss": 0.1219, "step": 47343 }, { "epoch": 34.68424908424908, "grad_norm": 0.88671875, "learning_rate": 1.0221292279856005e-06, "loss": 0.1229, "step": 47344 }, { "epoch": 34.684981684981686, "grad_norm": 1.0859375, "learning_rate": 1.0220269040226392e-06, "loss": 0.0967, "step": 47345 }, { "epoch": 34.68571428571428, "grad_norm": 0.8828125, "learning_rate": 1.0219248171248874e-06, "loss": 0.1249, "step": 47346 }, { "epoch": 34.68644688644689, "grad_norm": 0.90234375, "learning_rate": 1.02182296729284e-06, "loss": 0.1114, "step": 47347 }, { "epoch": 34.687179487179485, "grad_norm": 1.15625, "learning_rate": 1.02172135452698e-06, "loss": 0.112, "step": 47348 }, { "epoch": 34.68791208791209, "grad_norm": 1.109375, "learning_rate": 1.021619978827791e-06, "loss": 0.1004, "step": 47349 }, { "epoch": 34.688644688644686, "grad_norm": 1.0234375, "learning_rate": 1.0215188401957682e-06, "loss": 0.1151, "step": 47350 }, { "epoch": 34.68937728937729, "grad_norm": 1.1015625, "learning_rate": 1.0214179386313895e-06, "loss": 0.1039, "step": 47351 }, { "epoch": 34.69010989010989, "grad_norm": 1.1484375, "learning_rate": 1.021317274135144e-06, "loss": 0.1021, "step": 47352 }, { "epoch": 34.69084249084249, "grad_norm": 1.8828125, "learning_rate": 1.021216846707504e-06, "loss": 0.1259, "step": 47353 }, { "epoch": 34.69157509157509, "grad_norm": 1.0546875, "learning_rate": 1.0211166563489644e-06, "loss": 0.1342, "step": 47354 }, { "epoch": 34.69230769230769, "grad_norm": 1.328125, "learning_rate": 1.0210167030599921e-06, "loss": 0.1256, "step": 47355 }, { "epoch": 34.69304029304029, "grad_norm": 1.21875, "learning_rate": 1.0209169868410711e-06, "loss": 0.1234, "step": 47356 }, { "epoch": 34.693772893772895, "grad_norm": 0.828125, "learning_rate": 1.0208175076926848e-06, "loss": 0.1171, "step": 47357 }, { "epoch": 34.69450549450549, "grad_norm": 1.2109375, "learning_rate": 1.0207182656153005e-06, "loss": 0.1315, "step": 47358 }, { "epoch": 34.695238095238096, "grad_norm": 1.0234375, "learning_rate": 1.0206192606093957e-06, "loss": 0.118, "step": 47359 }, { "epoch": 34.69597069597069, "grad_norm": 1.1640625, "learning_rate": 1.020520492675455e-06, "loss": 0.0958, "step": 47360 }, { "epoch": 34.6967032967033, "grad_norm": 0.82421875, "learning_rate": 1.020421961813939e-06, "loss": 0.13, "step": 47361 }, { "epoch": 34.697435897435895, "grad_norm": 0.96484375, "learning_rate": 1.020323668025321e-06, "loss": 0.1494, "step": 47362 }, { "epoch": 34.6981684981685, "grad_norm": 0.58984375, "learning_rate": 1.0202256113100791e-06, "loss": 0.1045, "step": 47363 }, { "epoch": 34.698901098901096, "grad_norm": 1.15625, "learning_rate": 1.02012779166868e-06, "loss": 0.0847, "step": 47364 }, { "epoch": 34.6996336996337, "grad_norm": 0.9453125, "learning_rate": 1.0200302091015912e-06, "loss": 0.1218, "step": 47365 }, { "epoch": 34.7003663003663, "grad_norm": 1.0078125, "learning_rate": 1.0199328636092798e-06, "loss": 0.1243, "step": 47366 }, { "epoch": 34.7010989010989, "grad_norm": 0.83203125, "learning_rate": 1.0198357551922182e-06, "loss": 0.1216, "step": 47367 }, { "epoch": 34.7018315018315, "grad_norm": 1.2421875, "learning_rate": 1.0197388838508627e-06, "loss": 0.1238, "step": 47368 }, { "epoch": 34.702564102564104, "grad_norm": 1.171875, "learning_rate": 1.0196422495856805e-06, "loss": 0.1012, "step": 47369 }, { "epoch": 34.7032967032967, "grad_norm": 0.78125, "learning_rate": 1.0195458523971384e-06, "loss": 0.1126, "step": 47370 }, { "epoch": 34.704029304029305, "grad_norm": 1.046875, "learning_rate": 1.019449692285693e-06, "loss": 0.1401, "step": 47371 }, { "epoch": 34.7047619047619, "grad_norm": 0.9765625, "learning_rate": 1.0193537692518054e-06, "loss": 0.1181, "step": 47372 }, { "epoch": 34.70549450549451, "grad_norm": 0.76171875, "learning_rate": 1.0192580832959432e-06, "loss": 0.1062, "step": 47373 }, { "epoch": 34.706227106227104, "grad_norm": 1.234375, "learning_rate": 1.0191626344185568e-06, "loss": 0.1411, "step": 47374 }, { "epoch": 34.70695970695971, "grad_norm": 1.1171875, "learning_rate": 1.0190674226201023e-06, "loss": 0.1078, "step": 47375 }, { "epoch": 34.707692307692305, "grad_norm": 0.8203125, "learning_rate": 1.0189724479010417e-06, "loss": 0.1077, "step": 47376 }, { "epoch": 34.70842490842491, "grad_norm": 1.015625, "learning_rate": 1.0188777102618307e-06, "loss": 0.1238, "step": 47377 }, { "epoch": 34.70915750915751, "grad_norm": 0.7890625, "learning_rate": 1.0187832097029144e-06, "loss": 0.1007, "step": 47378 }, { "epoch": 34.70989010989011, "grad_norm": 0.984375, "learning_rate": 1.018688946224755e-06, "loss": 0.1296, "step": 47379 }, { "epoch": 34.71062271062271, "grad_norm": 1.3828125, "learning_rate": 1.0185949198278025e-06, "loss": 0.122, "step": 47380 }, { "epoch": 34.71135531135531, "grad_norm": 1.171875, "learning_rate": 1.0185011305125023e-06, "loss": 0.0966, "step": 47381 }, { "epoch": 34.71208791208791, "grad_norm": 0.81640625, "learning_rate": 1.0184075782793051e-06, "loss": 0.1145, "step": 47382 }, { "epoch": 34.712820512820514, "grad_norm": 2.453125, "learning_rate": 1.018314263128667e-06, "loss": 0.141, "step": 47383 }, { "epoch": 34.71355311355311, "grad_norm": 1.3203125, "learning_rate": 1.018221185061022e-06, "loss": 0.107, "step": 47384 }, { "epoch": 34.714285714285715, "grad_norm": 1.4453125, "learning_rate": 1.0181283440768267e-06, "loss": 0.1054, "step": 47385 }, { "epoch": 34.71501831501831, "grad_norm": 0.8359375, "learning_rate": 1.0180357401765258e-06, "loss": 0.1101, "step": 47386 }, { "epoch": 34.71575091575092, "grad_norm": 1.125, "learning_rate": 1.0179433733605535e-06, "loss": 0.1254, "step": 47387 }, { "epoch": 34.716483516483514, "grad_norm": 0.85546875, "learning_rate": 1.0178512436293604e-06, "loss": 0.1311, "step": 47388 }, { "epoch": 34.71721611721612, "grad_norm": 0.64453125, "learning_rate": 1.0177593509833864e-06, "loss": 0.1242, "step": 47389 }, { "epoch": 34.717948717948715, "grad_norm": 0.97265625, "learning_rate": 1.017667695423071e-06, "loss": 0.1275, "step": 47390 }, { "epoch": 34.71868131868132, "grad_norm": 0.67578125, "learning_rate": 1.0175762769488536e-06, "loss": 0.086, "step": 47391 }, { "epoch": 34.71941391941392, "grad_norm": 1.140625, "learning_rate": 1.0174850955611746e-06, "loss": 0.1443, "step": 47392 }, { "epoch": 34.72014652014652, "grad_norm": 1.1171875, "learning_rate": 1.0173941512604673e-06, "loss": 0.1041, "step": 47393 }, { "epoch": 34.72087912087912, "grad_norm": 1.109375, "learning_rate": 1.0173034440471662e-06, "loss": 0.1235, "step": 47394 }, { "epoch": 34.72161172161172, "grad_norm": 0.9765625, "learning_rate": 1.017212973921711e-06, "loss": 0.137, "step": 47395 }, { "epoch": 34.72234432234432, "grad_norm": 0.8671875, "learning_rate": 1.0171227408845305e-06, "loss": 0.117, "step": 47396 }, { "epoch": 34.723076923076924, "grad_norm": 0.875, "learning_rate": 1.0170327449360586e-06, "loss": 0.1264, "step": 47397 }, { "epoch": 34.72380952380952, "grad_norm": 1.03125, "learning_rate": 1.0169429860767294e-06, "loss": 0.1354, "step": 47398 }, { "epoch": 34.724542124542126, "grad_norm": 0.91796875, "learning_rate": 1.0168534643069716e-06, "loss": 0.1289, "step": 47399 }, { "epoch": 34.72527472527472, "grad_norm": 1.4140625, "learning_rate": 1.0167641796272088e-06, "loss": 0.1348, "step": 47400 }, { "epoch": 34.72600732600733, "grad_norm": 0.97265625, "learning_rate": 1.0166751320378744e-06, "loss": 0.1717, "step": 47401 }, { "epoch": 34.726739926739924, "grad_norm": 0.9140625, "learning_rate": 1.0165863215393975e-06, "loss": 0.136, "step": 47402 }, { "epoch": 34.72747252747253, "grad_norm": 0.88671875, "learning_rate": 1.0164977481321958e-06, "loss": 0.1165, "step": 47403 }, { "epoch": 34.728205128205126, "grad_norm": 0.91796875, "learning_rate": 1.0164094118166978e-06, "loss": 0.1005, "step": 47404 }, { "epoch": 34.72893772893773, "grad_norm": 1.1171875, "learning_rate": 1.0163213125933323e-06, "loss": 0.1216, "step": 47405 }, { "epoch": 34.72967032967033, "grad_norm": 0.94921875, "learning_rate": 1.0162334504625114e-06, "loss": 0.0953, "step": 47406 }, { "epoch": 34.73040293040293, "grad_norm": 1.53125, "learning_rate": 1.0161458254246585e-06, "loss": 0.1337, "step": 47407 }, { "epoch": 34.73113553113553, "grad_norm": 1.265625, "learning_rate": 1.0160584374802016e-06, "loss": 0.1425, "step": 47408 }, { "epoch": 34.73186813186813, "grad_norm": 1.0546875, "learning_rate": 1.015971286629548e-06, "loss": 0.1189, "step": 47409 }, { "epoch": 34.73260073260073, "grad_norm": 0.82421875, "learning_rate": 1.0158843728731207e-06, "loss": 0.0996, "step": 47410 }, { "epoch": 34.733333333333334, "grad_norm": 1.0390625, "learning_rate": 1.015797696211337e-06, "loss": 0.1554, "step": 47411 }, { "epoch": 34.73406593406593, "grad_norm": 0.8359375, "learning_rate": 1.01571125664461e-06, "loss": 0.1136, "step": 47412 }, { "epoch": 34.734798534798536, "grad_norm": 1.0625, "learning_rate": 1.0156250541733562e-06, "loss": 0.1191, "step": 47413 }, { "epoch": 34.73553113553113, "grad_norm": 0.703125, "learning_rate": 1.0155390887979884e-06, "loss": 0.1264, "step": 47414 }, { "epoch": 34.73626373626374, "grad_norm": 1.3203125, "learning_rate": 1.0154533605189187e-06, "loss": 0.1311, "step": 47415 }, { "epoch": 34.736996336996334, "grad_norm": 0.8359375, "learning_rate": 1.015367869336554e-06, "loss": 0.1212, "step": 47416 }, { "epoch": 34.73772893772894, "grad_norm": 1.65625, "learning_rate": 1.015282615251306e-06, "loss": 0.1329, "step": 47417 }, { "epoch": 34.738461538461536, "grad_norm": 1.375, "learning_rate": 1.0151975982635873e-06, "loss": 0.1302, "step": 47418 }, { "epoch": 34.73919413919414, "grad_norm": 1.2890625, "learning_rate": 1.0151128183737988e-06, "loss": 0.1405, "step": 47419 }, { "epoch": 34.73992673992674, "grad_norm": 1.0, "learning_rate": 1.0150282755823472e-06, "loss": 0.1186, "step": 47420 }, { "epoch": 34.74065934065934, "grad_norm": 1.0703125, "learning_rate": 1.0149439698896448e-06, "loss": 0.1095, "step": 47421 }, { "epoch": 34.74139194139194, "grad_norm": 1.59375, "learning_rate": 1.0148599012960873e-06, "loss": 0.164, "step": 47422 }, { "epoch": 34.74212454212454, "grad_norm": 1.2421875, "learning_rate": 1.0147760698020865e-06, "loss": 0.1453, "step": 47423 }, { "epoch": 34.74285714285714, "grad_norm": 1.21875, "learning_rate": 1.0146924754080333e-06, "loss": 0.1042, "step": 47424 }, { "epoch": 34.743589743589745, "grad_norm": 1.1171875, "learning_rate": 1.0146091181143336e-06, "loss": 0.1391, "step": 47425 }, { "epoch": 34.74432234432234, "grad_norm": 1.3828125, "learning_rate": 1.014525997921389e-06, "loss": 0.1052, "step": 47426 }, { "epoch": 34.745054945054946, "grad_norm": 1.234375, "learning_rate": 1.0144431148295954e-06, "loss": 0.1473, "step": 47427 }, { "epoch": 34.74578754578754, "grad_norm": 0.765625, "learning_rate": 1.014360468839348e-06, "loss": 0.1217, "step": 47428 }, { "epoch": 34.74652014652015, "grad_norm": 1.1796875, "learning_rate": 1.0142780599510426e-06, "loss": 0.125, "step": 47429 }, { "epoch": 34.747252747252745, "grad_norm": 1.40625, "learning_rate": 1.0141958881650805e-06, "loss": 0.1223, "step": 47430 }, { "epoch": 34.74798534798535, "grad_norm": 0.734375, "learning_rate": 1.0141139534818462e-06, "loss": 0.082, "step": 47431 }, { "epoch": 34.748717948717946, "grad_norm": 0.66015625, "learning_rate": 1.0140322559017413e-06, "loss": 0.1051, "step": 47432 }, { "epoch": 34.74945054945055, "grad_norm": 0.9609375, "learning_rate": 1.01395079542515e-06, "loss": 0.1059, "step": 47433 }, { "epoch": 34.75018315018315, "grad_norm": 1.3125, "learning_rate": 1.0138695720524683e-06, "loss": 0.1291, "step": 47434 }, { "epoch": 34.75091575091575, "grad_norm": 0.953125, "learning_rate": 1.013788585784081e-06, "loss": 0.1065, "step": 47435 }, { "epoch": 34.75164835164835, "grad_norm": 1.46875, "learning_rate": 1.0137078366203834e-06, "loss": 0.1226, "step": 47436 }, { "epoch": 34.75238095238095, "grad_norm": 1.0234375, "learning_rate": 1.0136273245617494e-06, "loss": 0.1117, "step": 47437 }, { "epoch": 34.75311355311355, "grad_norm": 0.85546875, "learning_rate": 1.0135470496085745e-06, "loss": 0.1129, "step": 47438 }, { "epoch": 34.753846153846155, "grad_norm": 0.921875, "learning_rate": 1.0134670117612437e-06, "loss": 0.1103, "step": 47439 }, { "epoch": 34.75457875457875, "grad_norm": 0.9921875, "learning_rate": 1.013387211020136e-06, "loss": 0.1494, "step": 47440 }, { "epoch": 34.755311355311356, "grad_norm": 0.66015625, "learning_rate": 1.0133076473856308e-06, "loss": 0.0915, "step": 47441 }, { "epoch": 34.75604395604395, "grad_norm": 0.81640625, "learning_rate": 1.0132283208581237e-06, "loss": 0.1147, "step": 47442 }, { "epoch": 34.75677655677656, "grad_norm": 0.6328125, "learning_rate": 1.013149231437977e-06, "loss": 0.1064, "step": 47443 }, { "epoch": 34.757509157509155, "grad_norm": 0.83203125, "learning_rate": 1.0130703791255814e-06, "loss": 0.1096, "step": 47444 }, { "epoch": 34.75824175824176, "grad_norm": 1.0546875, "learning_rate": 1.0129917639213105e-06, "loss": 0.1281, "step": 47445 }, { "epoch": 34.758974358974356, "grad_norm": 1.453125, "learning_rate": 1.0129133858255435e-06, "loss": 0.1195, "step": 47446 }, { "epoch": 34.75970695970696, "grad_norm": 1.1328125, "learning_rate": 1.0128352448386484e-06, "loss": 0.1058, "step": 47447 }, { "epoch": 34.76043956043956, "grad_norm": 0.78515625, "learning_rate": 1.01275734096101e-06, "loss": 0.111, "step": 47448 }, { "epoch": 34.76117216117216, "grad_norm": 1.0078125, "learning_rate": 1.0126796741929967e-06, "loss": 0.1167, "step": 47449 }, { "epoch": 34.76190476190476, "grad_norm": 1.15625, "learning_rate": 1.0126022445349764e-06, "loss": 0.09, "step": 47450 }, { "epoch": 34.76263736263736, "grad_norm": 0.88671875, "learning_rate": 1.0125250519873285e-06, "loss": 0.1123, "step": 47451 }, { "epoch": 34.76336996336996, "grad_norm": 1.375, "learning_rate": 1.012448096550421e-06, "loss": 0.1397, "step": 47452 }, { "epoch": 34.764102564102565, "grad_norm": 1.5859375, "learning_rate": 1.0123713782246168e-06, "loss": 0.1538, "step": 47453 }, { "epoch": 34.76483516483516, "grad_norm": 0.98046875, "learning_rate": 1.0122948970102843e-06, "loss": 0.1327, "step": 47454 }, { "epoch": 34.765567765567766, "grad_norm": 1.421875, "learning_rate": 1.0122186529077967e-06, "loss": 0.1359, "step": 47455 }, { "epoch": 34.766300366300364, "grad_norm": 1.765625, "learning_rate": 1.012142645917517e-06, "loss": 0.1388, "step": 47456 }, { "epoch": 34.76703296703297, "grad_norm": 0.78515625, "learning_rate": 1.0120668760398026e-06, "loss": 0.14, "step": 47457 }, { "epoch": 34.767765567765565, "grad_norm": 1.21875, "learning_rate": 1.0119913432750268e-06, "loss": 0.1415, "step": 47458 }, { "epoch": 34.76849816849817, "grad_norm": 1.1171875, "learning_rate": 1.0119160476235415e-06, "loss": 0.1229, "step": 47459 }, { "epoch": 34.76923076923077, "grad_norm": 1.3125, "learning_rate": 1.011840989085715e-06, "loss": 0.1546, "step": 47460 }, { "epoch": 34.76996336996337, "grad_norm": 0.99609375, "learning_rate": 1.0117661676619043e-06, "loss": 0.1148, "step": 47461 }, { "epoch": 34.77069597069597, "grad_norm": 0.94140625, "learning_rate": 1.0116915833524668e-06, "loss": 0.1058, "step": 47462 }, { "epoch": 34.77142857142857, "grad_norm": 1.0234375, "learning_rate": 1.0116172361577595e-06, "loss": 0.1282, "step": 47463 }, { "epoch": 34.77216117216117, "grad_norm": 0.61328125, "learning_rate": 1.0115431260781453e-06, "loss": 0.1033, "step": 47464 }, { "epoch": 34.772893772893774, "grad_norm": 1.1640625, "learning_rate": 1.0114692531139705e-06, "loss": 0.1354, "step": 47465 }, { "epoch": 34.77362637362637, "grad_norm": 1.0, "learning_rate": 1.011395617265592e-06, "loss": 0.0999, "step": 47466 }, { "epoch": 34.774358974358975, "grad_norm": 1.796875, "learning_rate": 1.0113222185333621e-06, "loss": 0.1285, "step": 47467 }, { "epoch": 34.77509157509157, "grad_norm": 0.859375, "learning_rate": 1.0112490569176374e-06, "loss": 0.1262, "step": 47468 }, { "epoch": 34.77582417582418, "grad_norm": 0.92578125, "learning_rate": 1.0111761324187589e-06, "loss": 0.081, "step": 47469 }, { "epoch": 34.776556776556774, "grad_norm": 1.4140625, "learning_rate": 1.0111034450370893e-06, "loss": 0.136, "step": 47470 }, { "epoch": 34.77728937728938, "grad_norm": 1.2421875, "learning_rate": 1.0110309947729639e-06, "loss": 0.1237, "step": 47471 }, { "epoch": 34.778021978021975, "grad_norm": 0.8203125, "learning_rate": 1.010958781626734e-06, "loss": 0.0998, "step": 47472 }, { "epoch": 34.77875457875458, "grad_norm": 1.203125, "learning_rate": 1.0108868055987517e-06, "loss": 0.1434, "step": 47473 }, { "epoch": 34.77948717948718, "grad_norm": 0.953125, "learning_rate": 1.0108150666893524e-06, "loss": 0.1584, "step": 47474 }, { "epoch": 34.78021978021978, "grad_norm": 0.9921875, "learning_rate": 1.0107435648988874e-06, "loss": 0.0762, "step": 47475 }, { "epoch": 34.78095238095238, "grad_norm": 0.765625, "learning_rate": 1.0106723002276922e-06, "loss": 0.1244, "step": 47476 }, { "epoch": 34.78168498168498, "grad_norm": 1.265625, "learning_rate": 1.0106012726761185e-06, "loss": 0.1093, "step": 47477 }, { "epoch": 34.78241758241758, "grad_norm": 1.0859375, "learning_rate": 1.0105304822444958e-06, "loss": 0.1139, "step": 47478 }, { "epoch": 34.783150183150184, "grad_norm": 1.2265625, "learning_rate": 1.0104599289331705e-06, "loss": 0.1325, "step": 47479 }, { "epoch": 34.78388278388278, "grad_norm": 0.91796875, "learning_rate": 1.0103896127424726e-06, "loss": 0.0938, "step": 47480 }, { "epoch": 34.784615384615385, "grad_norm": 1.0234375, "learning_rate": 1.0103195336727478e-06, "loss": 0.1124, "step": 47481 }, { "epoch": 34.78534798534798, "grad_norm": 1.1484375, "learning_rate": 1.0102496917243316e-06, "loss": 0.1428, "step": 47482 }, { "epoch": 34.78608058608059, "grad_norm": 1.2421875, "learning_rate": 1.0101800868975536e-06, "loss": 0.1295, "step": 47483 }, { "epoch": 34.786813186813184, "grad_norm": 0.98046875, "learning_rate": 1.0101107191927494e-06, "loss": 0.137, "step": 47484 }, { "epoch": 34.78754578754579, "grad_norm": 0.92578125, "learning_rate": 1.0100415886102484e-06, "loss": 0.1239, "step": 47485 }, { "epoch": 34.788278388278385, "grad_norm": 1.03125, "learning_rate": 1.0099726951503863e-06, "loss": 0.1348, "step": 47486 }, { "epoch": 34.78901098901099, "grad_norm": 1.109375, "learning_rate": 1.0099040388134918e-06, "loss": 0.1039, "step": 47487 }, { "epoch": 34.78974358974359, "grad_norm": 1.078125, "learning_rate": 1.0098356195998958e-06, "loss": 0.138, "step": 47488 }, { "epoch": 34.79047619047619, "grad_norm": 0.953125, "learning_rate": 1.0097674375099219e-06, "loss": 0.1313, "step": 47489 }, { "epoch": 34.79120879120879, "grad_norm": 1.234375, "learning_rate": 1.0096994925438999e-06, "loss": 0.1398, "step": 47490 }, { "epoch": 34.79194139194139, "grad_norm": 1.0078125, "learning_rate": 1.009631784702154e-06, "loss": 0.1437, "step": 47491 }, { "epoch": 34.79267399267399, "grad_norm": 0.67578125, "learning_rate": 1.0095643139850088e-06, "loss": 0.0971, "step": 47492 }, { "epoch": 34.793406593406594, "grad_norm": 0.86328125, "learning_rate": 1.0094970803927828e-06, "loss": 0.1084, "step": 47493 }, { "epoch": 34.79413919413919, "grad_norm": 1.109375, "learning_rate": 1.009430083925811e-06, "loss": 0.1315, "step": 47494 }, { "epoch": 34.794871794871796, "grad_norm": 1.6171875, "learning_rate": 1.0093633245844015e-06, "loss": 0.1168, "step": 47495 }, { "epoch": 34.79560439560439, "grad_norm": 1.328125, "learning_rate": 1.0092968023688785e-06, "loss": 0.1016, "step": 47496 }, { "epoch": 34.796336996337, "grad_norm": 1.1171875, "learning_rate": 1.0092305172795605e-06, "loss": 0.1426, "step": 47497 }, { "epoch": 34.797069597069594, "grad_norm": 1.0234375, "learning_rate": 1.009164469316772e-06, "loss": 0.1384, "step": 47498 }, { "epoch": 34.7978021978022, "grad_norm": 0.83203125, "learning_rate": 1.0090986584808153e-06, "loss": 0.1238, "step": 47499 }, { "epoch": 34.798534798534796, "grad_norm": 1.15625, "learning_rate": 1.00903308477202e-06, "loss": 0.1126, "step": 47500 }, { "epoch": 34.7992673992674, "grad_norm": 1.1875, "learning_rate": 1.0089677481906937e-06, "loss": 0.118, "step": 47501 }, { "epoch": 34.8, "grad_norm": 1.0390625, "learning_rate": 1.0089026487371443e-06, "loss": 0.1526, "step": 47502 }, { "epoch": 34.8007326007326, "grad_norm": 1.1484375, "learning_rate": 1.0088377864116909e-06, "loss": 0.1057, "step": 47503 }, { "epoch": 34.8014652014652, "grad_norm": 0.890625, "learning_rate": 1.008773161214646e-06, "loss": 0.0992, "step": 47504 }, { "epoch": 34.8021978021978, "grad_norm": 1.734375, "learning_rate": 1.0087087731463179e-06, "loss": 0.1602, "step": 47505 }, { "epoch": 34.8029304029304, "grad_norm": 0.78515625, "learning_rate": 1.0086446222070088e-06, "loss": 0.1275, "step": 47506 }, { "epoch": 34.803663003663004, "grad_norm": 0.86328125, "learning_rate": 1.0085807083970262e-06, "loss": 0.1133, "step": 47507 }, { "epoch": 34.8043956043956, "grad_norm": 0.9375, "learning_rate": 1.0085170317166892e-06, "loss": 0.1076, "step": 47508 }, { "epoch": 34.805128205128206, "grad_norm": 0.6328125, "learning_rate": 1.0084535921662888e-06, "loss": 0.1289, "step": 47509 }, { "epoch": 34.8058608058608, "grad_norm": 1.2265625, "learning_rate": 1.0083903897461328e-06, "loss": 0.1339, "step": 47510 }, { "epoch": 34.80659340659341, "grad_norm": 1.0625, "learning_rate": 1.008327424456529e-06, "loss": 0.0985, "step": 47511 }, { "epoch": 34.807326007326004, "grad_norm": 1.046875, "learning_rate": 1.0082646962977742e-06, "loss": 0.1252, "step": 47512 }, { "epoch": 34.80805860805861, "grad_norm": 1.015625, "learning_rate": 1.0082022052701707e-06, "loss": 0.1116, "step": 47513 }, { "epoch": 34.808791208791206, "grad_norm": 0.90234375, "learning_rate": 1.008139951374015e-06, "loss": 0.1339, "step": 47514 }, { "epoch": 34.80952380952381, "grad_norm": 0.8515625, "learning_rate": 1.0080779346096153e-06, "loss": 0.0929, "step": 47515 }, { "epoch": 34.81025641025641, "grad_norm": 0.96875, "learning_rate": 1.0080161549772514e-06, "loss": 0.1124, "step": 47516 }, { "epoch": 34.81098901098901, "grad_norm": 1.2265625, "learning_rate": 1.0079546124772366e-06, "loss": 0.1302, "step": 47517 }, { "epoch": 34.81172161172161, "grad_norm": 1.6484375, "learning_rate": 1.007893307109857e-06, "loss": 0.1417, "step": 47518 }, { "epoch": 34.81245421245421, "grad_norm": 0.71484375, "learning_rate": 1.0078322388754039e-06, "loss": 0.105, "step": 47519 }, { "epoch": 34.81318681318681, "grad_norm": 1.4140625, "learning_rate": 1.007771407774179e-06, "loss": 0.1127, "step": 47520 }, { "epoch": 34.813919413919415, "grad_norm": 1.15625, "learning_rate": 1.007710813806463e-06, "loss": 0.119, "step": 47521 }, { "epoch": 34.81465201465201, "grad_norm": 0.8359375, "learning_rate": 1.0076504569725583e-06, "loss": 0.1116, "step": 47522 }, { "epoch": 34.815384615384616, "grad_norm": 0.91796875, "learning_rate": 1.007590337272745e-06, "loss": 0.1269, "step": 47523 }, { "epoch": 34.81611721611721, "grad_norm": 0.94140625, "learning_rate": 1.007530454707314e-06, "loss": 0.1092, "step": 47524 }, { "epoch": 34.81684981684982, "grad_norm": 0.90234375, "learning_rate": 1.007470809276546e-06, "loss": 0.1138, "step": 47525 }, { "epoch": 34.817582417582415, "grad_norm": 0.83984375, "learning_rate": 1.0074114009807379e-06, "loss": 0.1296, "step": 47526 }, { "epoch": 34.81831501831502, "grad_norm": 1.1640625, "learning_rate": 1.0073522298201696e-06, "loss": 0.1373, "step": 47527 }, { "epoch": 34.819047619047616, "grad_norm": 1.21875, "learning_rate": 1.007293295795127e-06, "loss": 0.1031, "step": 47528 }, { "epoch": 34.81978021978022, "grad_norm": 0.91796875, "learning_rate": 1.007234598905885e-06, "loss": 0.1108, "step": 47529 }, { "epoch": 34.82051282051282, "grad_norm": 1.375, "learning_rate": 1.007176139152729e-06, "loss": 0.0859, "step": 47530 }, { "epoch": 34.82124542124542, "grad_norm": 0.8671875, "learning_rate": 1.0071179165359398e-06, "loss": 0.1374, "step": 47531 }, { "epoch": 34.82197802197802, "grad_norm": 1.171875, "learning_rate": 1.0070599310557973e-06, "loss": 0.1369, "step": 47532 }, { "epoch": 34.82271062271062, "grad_norm": 0.9296875, "learning_rate": 1.0070021827125763e-06, "loss": 0.1138, "step": 47533 }, { "epoch": 34.82344322344322, "grad_norm": 1.2421875, "learning_rate": 1.0069446715065575e-06, "loss": 0.1089, "step": 47534 }, { "epoch": 34.824175824175825, "grad_norm": 1.0546875, "learning_rate": 1.006887397438015e-06, "loss": 0.1092, "step": 47535 }, { "epoch": 34.82490842490842, "grad_norm": 1.3125, "learning_rate": 1.0068303605072188e-06, "loss": 0.1426, "step": 47536 }, { "epoch": 34.825641025641026, "grad_norm": 0.92578125, "learning_rate": 1.0067735607144486e-06, "loss": 0.1044, "step": 47537 }, { "epoch": 34.82637362637362, "grad_norm": 0.79296875, "learning_rate": 1.0067169980599738e-06, "loss": 0.1303, "step": 47538 }, { "epoch": 34.82710622710623, "grad_norm": 0.75390625, "learning_rate": 1.0066606725440695e-06, "loss": 0.1001, "step": 47539 }, { "epoch": 34.827838827838825, "grad_norm": 0.6875, "learning_rate": 1.0066045841669937e-06, "loss": 0.1354, "step": 47540 }, { "epoch": 34.82857142857143, "grad_norm": 1.7890625, "learning_rate": 1.0065487329290268e-06, "loss": 0.1458, "step": 47541 }, { "epoch": 34.829304029304026, "grad_norm": 0.89453125, "learning_rate": 1.0064931188304324e-06, "loss": 0.1247, "step": 47542 }, { "epoch": 34.83003663003663, "grad_norm": 0.70703125, "learning_rate": 1.0064377418714798e-06, "loss": 0.1092, "step": 47543 }, { "epoch": 34.83076923076923, "grad_norm": 1.046875, "learning_rate": 1.0063826020524277e-06, "loss": 0.1392, "step": 47544 }, { "epoch": 34.83150183150183, "grad_norm": 1.078125, "learning_rate": 1.006327699373545e-06, "loss": 0.1045, "step": 47545 }, { "epoch": 34.83223443223443, "grad_norm": 0.6484375, "learning_rate": 1.0062730338350956e-06, "loss": 0.1099, "step": 47546 }, { "epoch": 34.832967032967034, "grad_norm": 1.0078125, "learning_rate": 1.0062186054373378e-06, "loss": 0.123, "step": 47547 }, { "epoch": 34.83369963369963, "grad_norm": 1.234375, "learning_rate": 1.0061644141805354e-06, "loss": 0.121, "step": 47548 }, { "epoch": 34.834432234432235, "grad_norm": 1.1796875, "learning_rate": 1.0061104600649465e-06, "loss": 0.1292, "step": 47549 }, { "epoch": 34.83516483516483, "grad_norm": 1.453125, "learning_rate": 1.0060567430908298e-06, "loss": 0.1173, "step": 47550 }, { "epoch": 34.83589743589744, "grad_norm": 1.125, "learning_rate": 1.0060032632584432e-06, "loss": 0.1135, "step": 47551 }, { "epoch": 34.836630036630034, "grad_norm": 1.0703125, "learning_rate": 1.0059500205680452e-06, "loss": 0.119, "step": 47552 }, { "epoch": 34.83736263736264, "grad_norm": 1.3671875, "learning_rate": 1.0058970150198886e-06, "loss": 0.1341, "step": 47553 }, { "epoch": 34.838095238095235, "grad_norm": 1.359375, "learning_rate": 1.0058442466142204e-06, "loss": 0.1027, "step": 47554 }, { "epoch": 34.83882783882784, "grad_norm": 0.89453125, "learning_rate": 1.0057917153513047e-06, "loss": 0.1208, "step": 47555 }, { "epoch": 34.83956043956044, "grad_norm": 0.97265625, "learning_rate": 1.0057394212313945e-06, "loss": 0.1089, "step": 47556 }, { "epoch": 34.84029304029304, "grad_norm": 1.1640625, "learning_rate": 1.0056873642547255e-06, "loss": 0.1267, "step": 47557 }, { "epoch": 34.84102564102564, "grad_norm": 0.609375, "learning_rate": 1.0056355444215623e-06, "loss": 0.1074, "step": 47558 }, { "epoch": 34.84175824175824, "grad_norm": 0.8046875, "learning_rate": 1.005583961732146e-06, "loss": 0.1085, "step": 47559 }, { "epoch": 34.84249084249084, "grad_norm": 0.85546875, "learning_rate": 1.0055326161867239e-06, "loss": 0.1427, "step": 47560 }, { "epoch": 34.843223443223444, "grad_norm": 1.1484375, "learning_rate": 1.0054815077855437e-06, "loss": 0.1256, "step": 47561 }, { "epoch": 34.84395604395604, "grad_norm": 1.21875, "learning_rate": 1.005430636528847e-06, "loss": 0.117, "step": 47562 }, { "epoch": 34.844688644688645, "grad_norm": 0.87109375, "learning_rate": 1.0053800024168813e-06, "loss": 0.1346, "step": 47563 }, { "epoch": 34.84542124542124, "grad_norm": 1.015625, "learning_rate": 1.0053296054498883e-06, "loss": 0.1186, "step": 47564 }, { "epoch": 34.84615384615385, "grad_norm": 1.1015625, "learning_rate": 1.0052794456281097e-06, "loss": 0.1052, "step": 47565 }, { "epoch": 34.846886446886444, "grad_norm": 0.765625, "learning_rate": 1.0052295229517873e-06, "loss": 0.1162, "step": 47566 }, { "epoch": 34.84761904761905, "grad_norm": 0.8359375, "learning_rate": 1.0051798374211574e-06, "loss": 0.1064, "step": 47567 }, { "epoch": 34.848351648351645, "grad_norm": 1.40625, "learning_rate": 1.0051303890364567e-06, "loss": 0.1321, "step": 47568 }, { "epoch": 34.84908424908425, "grad_norm": 0.75, "learning_rate": 1.0050811777979266e-06, "loss": 0.1378, "step": 47569 }, { "epoch": 34.84981684981685, "grad_norm": 1.375, "learning_rate": 1.0050322037057978e-06, "loss": 0.1492, "step": 47570 }, { "epoch": 34.85054945054945, "grad_norm": 0.89453125, "learning_rate": 1.0049834667603123e-06, "loss": 0.1214, "step": 47571 }, { "epoch": 34.85128205128205, "grad_norm": 1.140625, "learning_rate": 1.0049349669616957e-06, "loss": 0.1048, "step": 47572 }, { "epoch": 34.85201465201465, "grad_norm": 0.73046875, "learning_rate": 1.004886704310184e-06, "loss": 0.1318, "step": 47573 }, { "epoch": 34.85274725274725, "grad_norm": 1.1171875, "learning_rate": 1.004838678806008e-06, "loss": 0.1252, "step": 47574 }, { "epoch": 34.853479853479854, "grad_norm": 1.0234375, "learning_rate": 1.0047908904493985e-06, "loss": 0.1288, "step": 47575 }, { "epoch": 34.85421245421245, "grad_norm": 0.8515625, "learning_rate": 1.0047433392405864e-06, "loss": 0.1267, "step": 47576 }, { "epoch": 34.854945054945055, "grad_norm": 1.40625, "learning_rate": 1.0046960251797916e-06, "loss": 0.124, "step": 47577 }, { "epoch": 34.85567765567765, "grad_norm": 0.89453125, "learning_rate": 1.0046489482672505e-06, "loss": 0.1188, "step": 47578 }, { "epoch": 34.85641025641026, "grad_norm": 1.7421875, "learning_rate": 1.004602108503188e-06, "loss": 0.1239, "step": 47579 }, { "epoch": 34.857142857142854, "grad_norm": 0.82421875, "learning_rate": 1.0045555058878188e-06, "loss": 0.109, "step": 47580 }, { "epoch": 34.85787545787546, "grad_norm": 0.76953125, "learning_rate": 1.0045091404213737e-06, "loss": 0.1097, "step": 47581 }, { "epoch": 34.858608058608056, "grad_norm": 0.59375, "learning_rate": 1.0044630121040776e-06, "loss": 0.1219, "step": 47582 }, { "epoch": 34.85934065934066, "grad_norm": 1.0234375, "learning_rate": 1.0044171209361456e-06, "loss": 0.1205, "step": 47583 }, { "epoch": 34.86007326007326, "grad_norm": 1.078125, "learning_rate": 1.0043714669177967e-06, "loss": 0.1259, "step": 47584 }, { "epoch": 34.86080586080586, "grad_norm": 1.03125, "learning_rate": 1.0043260500492568e-06, "loss": 0.1093, "step": 47585 }, { "epoch": 34.86153846153846, "grad_norm": 0.90234375, "learning_rate": 1.0042808703307343e-06, "loss": 0.1427, "step": 47586 }, { "epoch": 34.86227106227106, "grad_norm": 0.98046875, "learning_rate": 1.0042359277624548e-06, "loss": 0.1169, "step": 47587 }, { "epoch": 34.86300366300366, "grad_norm": 0.89453125, "learning_rate": 1.0041912223446274e-06, "loss": 0.1395, "step": 47588 }, { "epoch": 34.863736263736264, "grad_norm": 1.265625, "learning_rate": 1.0041467540774714e-06, "loss": 0.109, "step": 47589 }, { "epoch": 34.86446886446886, "grad_norm": 1.671875, "learning_rate": 1.0041025229611904e-06, "loss": 0.1379, "step": 47590 }, { "epoch": 34.865201465201466, "grad_norm": 1.6171875, "learning_rate": 1.0040585289960095e-06, "loss": 0.1366, "step": 47591 }, { "epoch": 34.86593406593406, "grad_norm": 0.703125, "learning_rate": 1.0040147721821265e-06, "loss": 0.1058, "step": 47592 }, { "epoch": 34.86666666666667, "grad_norm": 1.0546875, "learning_rate": 1.0039712525197618e-06, "loss": 0.1154, "step": 47593 }, { "epoch": 34.867399267399264, "grad_norm": 1.2421875, "learning_rate": 1.0039279700091184e-06, "loss": 0.1211, "step": 47594 }, { "epoch": 34.86813186813187, "grad_norm": 1.3671875, "learning_rate": 1.0038849246503993e-06, "loss": 0.1387, "step": 47595 }, { "epoch": 34.868864468864466, "grad_norm": 1.0234375, "learning_rate": 1.0038421164438246e-06, "loss": 0.1438, "step": 47596 }, { "epoch": 34.86959706959707, "grad_norm": 1.2578125, "learning_rate": 1.003799545389587e-06, "loss": 0.1156, "step": 47597 }, { "epoch": 34.87032967032967, "grad_norm": 1.6171875, "learning_rate": 1.0037572114878894e-06, "loss": 0.1439, "step": 47598 }, { "epoch": 34.87106227106227, "grad_norm": 1.296875, "learning_rate": 1.0037151147389463e-06, "loss": 0.1232, "step": 47599 }, { "epoch": 34.87179487179487, "grad_norm": 1.078125, "learning_rate": 1.0036732551429496e-06, "loss": 0.1449, "step": 47600 }, { "epoch": 34.87252747252747, "grad_norm": 0.9765625, "learning_rate": 1.0036316327001032e-06, "loss": 0.1022, "step": 47601 }, { "epoch": 34.87326007326007, "grad_norm": 1.1796875, "learning_rate": 1.0035902474106105e-06, "loss": 0.1496, "step": 47602 }, { "epoch": 34.873992673992674, "grad_norm": 1.2109375, "learning_rate": 1.003549099274658e-06, "loss": 0.1282, "step": 47603 }, { "epoch": 34.87472527472527, "grad_norm": 1.0546875, "learning_rate": 1.0035081882924545e-06, "loss": 0.153, "step": 47604 }, { "epoch": 34.875457875457876, "grad_norm": 1.1640625, "learning_rate": 1.0034675144641926e-06, "loss": 0.1262, "step": 47605 }, { "epoch": 34.87619047619047, "grad_norm": 1.3671875, "learning_rate": 1.0034270777900643e-06, "loss": 0.1216, "step": 47606 }, { "epoch": 34.87692307692308, "grad_norm": 0.95703125, "learning_rate": 1.003386878270268e-06, "loss": 0.1308, "step": 47607 }, { "epoch": 34.877655677655675, "grad_norm": 1.078125, "learning_rate": 1.0033469159049954e-06, "loss": 0.1253, "step": 47608 }, { "epoch": 34.87838827838828, "grad_norm": 1.3671875, "learning_rate": 1.0033071906944338e-06, "loss": 0.1431, "step": 47609 }, { "epoch": 34.879120879120876, "grad_norm": 0.78125, "learning_rate": 1.0032677026387756e-06, "loss": 0.1185, "step": 47610 }, { "epoch": 34.87985347985348, "grad_norm": 0.859375, "learning_rate": 1.0032284517382072e-06, "loss": 0.1269, "step": 47611 }, { "epoch": 34.88058608058608, "grad_norm": 0.7734375, "learning_rate": 1.0031894379929268e-06, "loss": 0.1144, "step": 47612 }, { "epoch": 34.88131868131868, "grad_norm": 0.9140625, "learning_rate": 1.0031506614031102e-06, "loss": 0.1266, "step": 47613 }, { "epoch": 34.88205128205128, "grad_norm": 1.046875, "learning_rate": 1.0031121219689498e-06, "loss": 0.1083, "step": 47614 }, { "epoch": 34.88278388278388, "grad_norm": 1.078125, "learning_rate": 1.0030738196906268e-06, "loss": 0.1216, "step": 47615 }, { "epoch": 34.88351648351648, "grad_norm": 1.296875, "learning_rate": 1.003035754568328e-06, "loss": 0.14, "step": 47616 }, { "epoch": 34.884249084249085, "grad_norm": 1.0859375, "learning_rate": 1.0029979266022297e-06, "loss": 0.1152, "step": 47617 }, { "epoch": 34.88498168498168, "grad_norm": 1.0859375, "learning_rate": 1.002960335792518e-06, "loss": 0.1412, "step": 47618 }, { "epoch": 34.885714285714286, "grad_norm": 0.91796875, "learning_rate": 1.002922982139375e-06, "loss": 0.149, "step": 47619 }, { "epoch": 34.88644688644688, "grad_norm": 0.87890625, "learning_rate": 1.0028858656429762e-06, "loss": 0.1296, "step": 47620 }, { "epoch": 34.88717948717949, "grad_norm": 0.7890625, "learning_rate": 1.0028489863034974e-06, "loss": 0.1092, "step": 47621 }, { "epoch": 34.887912087912085, "grad_norm": 1.109375, "learning_rate": 1.00281234412112e-06, "loss": 0.1204, "step": 47622 }, { "epoch": 34.88864468864469, "grad_norm": 1.15625, "learning_rate": 1.00277593909602e-06, "loss": 0.0984, "step": 47623 }, { "epoch": 34.889377289377286, "grad_norm": 1.03125, "learning_rate": 1.002739771228362e-06, "loss": 0.1375, "step": 47624 }, { "epoch": 34.89010989010989, "grad_norm": 0.83203125, "learning_rate": 1.0027038405183335e-06, "loss": 0.0959, "step": 47625 }, { "epoch": 34.89084249084249, "grad_norm": 1.0390625, "learning_rate": 1.0026681469660985e-06, "loss": 0.1237, "step": 47626 }, { "epoch": 34.89157509157509, "grad_norm": 1.0234375, "learning_rate": 1.0026326905718279e-06, "loss": 0.1241, "step": 47627 }, { "epoch": 34.89230769230769, "grad_norm": 0.79296875, "learning_rate": 1.0025974713356918e-06, "loss": 0.1154, "step": 47628 }, { "epoch": 34.89304029304029, "grad_norm": 1.0234375, "learning_rate": 1.002562489257861e-06, "loss": 0.1037, "step": 47629 }, { "epoch": 34.89377289377289, "grad_norm": 1.0703125, "learning_rate": 1.0025277443385055e-06, "loss": 0.1054, "step": 47630 }, { "epoch": 34.894505494505495, "grad_norm": 1.09375, "learning_rate": 1.0024932365777844e-06, "loss": 0.1143, "step": 47631 }, { "epoch": 34.89523809523809, "grad_norm": 1.390625, "learning_rate": 1.0024589659758688e-06, "loss": 0.1374, "step": 47632 }, { "epoch": 34.895970695970696, "grad_norm": 1.2578125, "learning_rate": 1.0024249325329228e-06, "loss": 0.1187, "step": 47633 }, { "epoch": 34.89670329670329, "grad_norm": 1.203125, "learning_rate": 1.0023911362491064e-06, "loss": 0.1493, "step": 47634 }, { "epoch": 34.8974358974359, "grad_norm": 1.625, "learning_rate": 1.002357577124584e-06, "loss": 0.1242, "step": 47635 }, { "epoch": 34.898168498168495, "grad_norm": 0.828125, "learning_rate": 1.0023242551595153e-06, "loss": 0.1061, "step": 47636 }, { "epoch": 34.8989010989011, "grad_norm": 0.81640625, "learning_rate": 1.0022911703540594e-06, "loss": 0.1215, "step": 47637 }, { "epoch": 34.899633699633696, "grad_norm": 0.79296875, "learning_rate": 1.0022583227083706e-06, "loss": 0.1235, "step": 47638 }, { "epoch": 34.9003663003663, "grad_norm": 1.53125, "learning_rate": 1.002225712222619e-06, "loss": 0.1265, "step": 47639 }, { "epoch": 34.9010989010989, "grad_norm": 1.09375, "learning_rate": 1.0021933388969472e-06, "loss": 0.1248, "step": 47640 }, { "epoch": 34.9018315018315, "grad_norm": 0.921875, "learning_rate": 1.0021612027315203e-06, "loss": 0.137, "step": 47641 }, { "epoch": 34.9025641025641, "grad_norm": 1.171875, "learning_rate": 1.002129303726487e-06, "loss": 0.1592, "step": 47642 }, { "epoch": 34.903296703296704, "grad_norm": 1.2578125, "learning_rate": 1.0020976418820007e-06, "loss": 0.1338, "step": 47643 }, { "epoch": 34.9040293040293, "grad_norm": 0.90234375, "learning_rate": 1.0020662171982155e-06, "loss": 0.1324, "step": 47644 }, { "epoch": 34.904761904761905, "grad_norm": 1.0390625, "learning_rate": 1.0020350296752799e-06, "loss": 0.1328, "step": 47645 }, { "epoch": 34.9054945054945, "grad_norm": 0.7890625, "learning_rate": 1.0020040793133422e-06, "loss": 0.1121, "step": 47646 }, { "epoch": 34.90622710622711, "grad_norm": 1.1640625, "learning_rate": 1.0019733661125504e-06, "loss": 0.1332, "step": 47647 }, { "epoch": 34.906959706959704, "grad_norm": 0.86328125, "learning_rate": 1.001942890073059e-06, "loss": 0.1237, "step": 47648 }, { "epoch": 34.90769230769231, "grad_norm": 1.09375, "learning_rate": 1.0019126511950051e-06, "loss": 0.1009, "step": 47649 }, { "epoch": 34.908424908424905, "grad_norm": 1.0234375, "learning_rate": 1.0018826494785367e-06, "loss": 0.1088, "step": 47650 }, { "epoch": 34.90915750915751, "grad_norm": 0.83984375, "learning_rate": 1.0018528849237975e-06, "loss": 0.1372, "step": 47651 }, { "epoch": 34.90989010989011, "grad_norm": 0.82421875, "learning_rate": 1.0018233575309298e-06, "loss": 0.1396, "step": 47652 }, { "epoch": 34.91062271062271, "grad_norm": 0.93359375, "learning_rate": 1.0017940673000765e-06, "loss": 0.1005, "step": 47653 }, { "epoch": 34.91135531135531, "grad_norm": 1.234375, "learning_rate": 1.0017650142313751e-06, "loss": 0.1257, "step": 47654 }, { "epoch": 34.91208791208791, "grad_norm": 1.3828125, "learning_rate": 1.0017361983249685e-06, "loss": 0.1541, "step": 47655 }, { "epoch": 34.91282051282051, "grad_norm": 1.0859375, "learning_rate": 1.0017076195809942e-06, "loss": 0.118, "step": 47656 }, { "epoch": 34.913553113553114, "grad_norm": 0.82421875, "learning_rate": 1.001679277999584e-06, "loss": 0.1422, "step": 47657 }, { "epoch": 34.91428571428571, "grad_norm": 1.265625, "learning_rate": 1.0016511735808807e-06, "loss": 0.1138, "step": 47658 }, { "epoch": 34.915018315018315, "grad_norm": 0.62109375, "learning_rate": 1.0016233063250164e-06, "loss": 0.1112, "step": 47659 }, { "epoch": 34.91575091575091, "grad_norm": 0.8203125, "learning_rate": 1.0015956762321227e-06, "loss": 0.0928, "step": 47660 }, { "epoch": 34.91648351648352, "grad_norm": 1.140625, "learning_rate": 1.0015682833023318e-06, "loss": 0.1408, "step": 47661 }, { "epoch": 34.917216117216114, "grad_norm": 0.76953125, "learning_rate": 1.0015411275357806e-06, "loss": 0.1437, "step": 47662 }, { "epoch": 34.91794871794872, "grad_norm": 1.0390625, "learning_rate": 1.0015142089325908e-06, "loss": 0.1259, "step": 47663 }, { "epoch": 34.918681318681315, "grad_norm": 1.28125, "learning_rate": 1.001487527492899e-06, "loss": 0.1424, "step": 47664 }, { "epoch": 34.91941391941392, "grad_norm": 1.0234375, "learning_rate": 1.0014610832168322e-06, "loss": 0.108, "step": 47665 }, { "epoch": 34.92014652014652, "grad_norm": 1.6171875, "learning_rate": 1.0014348761045109e-06, "loss": 0.1356, "step": 47666 }, { "epoch": 34.92087912087912, "grad_norm": 1.4453125, "learning_rate": 1.0014089061560618e-06, "loss": 0.1362, "step": 47667 }, { "epoch": 34.92161172161172, "grad_norm": 0.953125, "learning_rate": 1.0013831733716163e-06, "loss": 0.1295, "step": 47668 }, { "epoch": 34.92234432234432, "grad_norm": 1.265625, "learning_rate": 1.0013576777512904e-06, "loss": 0.1391, "step": 47669 }, { "epoch": 34.92307692307692, "grad_norm": 1.1953125, "learning_rate": 1.0013324192952103e-06, "loss": 0.1155, "step": 47670 }, { "epoch": 34.923809523809524, "grad_norm": 1.6015625, "learning_rate": 1.0013073980034968e-06, "loss": 0.1337, "step": 47671 }, { "epoch": 34.92454212454212, "grad_norm": 0.92578125, "learning_rate": 1.0012826138762708e-06, "loss": 0.1351, "step": 47672 }, { "epoch": 34.925274725274726, "grad_norm": 1.328125, "learning_rate": 1.0012580669136421e-06, "loss": 0.1401, "step": 47673 }, { "epoch": 34.92600732600732, "grad_norm": 1.0859375, "learning_rate": 1.001233757115743e-06, "loss": 0.1432, "step": 47674 }, { "epoch": 34.92673992673993, "grad_norm": 0.84765625, "learning_rate": 1.0012096844826775e-06, "loss": 0.12, "step": 47675 }, { "epoch": 34.927472527472524, "grad_norm": 0.86328125, "learning_rate": 1.0011858490145667e-06, "loss": 0.1269, "step": 47676 }, { "epoch": 34.92820512820513, "grad_norm": 0.77734375, "learning_rate": 1.0011622507115262e-06, "loss": 0.1139, "step": 47677 }, { "epoch": 34.928937728937726, "grad_norm": 1.0390625, "learning_rate": 1.0011388895736655e-06, "loss": 0.1071, "step": 47678 }, { "epoch": 34.92967032967033, "grad_norm": 1.2890625, "learning_rate": 1.0011157656011003e-06, "loss": 0.1042, "step": 47679 }, { "epoch": 34.93040293040293, "grad_norm": 1.3203125, "learning_rate": 1.0010928787939348e-06, "loss": 0.1206, "step": 47680 }, { "epoch": 34.93113553113553, "grad_norm": 1.34375, "learning_rate": 1.0010702291522846e-06, "loss": 0.1218, "step": 47681 }, { "epoch": 34.93186813186813, "grad_norm": 1.1796875, "learning_rate": 1.001047816676254e-06, "loss": 0.155, "step": 47682 }, { "epoch": 34.93260073260073, "grad_norm": 1.3046875, "learning_rate": 1.0010256413659584e-06, "loss": 0.1184, "step": 47683 }, { "epoch": 34.93333333333333, "grad_norm": 0.85546875, "learning_rate": 1.0010037032214915e-06, "loss": 0.1083, "step": 47684 }, { "epoch": 34.934065934065934, "grad_norm": 1.2890625, "learning_rate": 1.0009820022429682e-06, "loss": 0.1347, "step": 47685 }, { "epoch": 34.93479853479853, "grad_norm": 0.64453125, "learning_rate": 1.0009605384304933e-06, "loss": 0.1069, "step": 47686 }, { "epoch": 34.935531135531136, "grad_norm": 1.609375, "learning_rate": 1.0009393117841602e-06, "loss": 0.0966, "step": 47687 }, { "epoch": 34.93626373626373, "grad_norm": 0.9375, "learning_rate": 1.0009183223040787e-06, "loss": 0.1176, "step": 47688 }, { "epoch": 34.93699633699634, "grad_norm": 0.94140625, "learning_rate": 1.0008975699903477e-06, "loss": 0.1137, "step": 47689 }, { "epoch": 34.937728937728934, "grad_norm": 1.46875, "learning_rate": 1.0008770548430663e-06, "loss": 0.1062, "step": 47690 }, { "epoch": 34.93846153846154, "grad_norm": 1.171875, "learning_rate": 1.0008567768623333e-06, "loss": 0.1398, "step": 47691 }, { "epoch": 34.939194139194136, "grad_norm": 1.046875, "learning_rate": 1.0008367360482422e-06, "loss": 0.117, "step": 47692 }, { "epoch": 34.93992673992674, "grad_norm": 0.93359375, "learning_rate": 1.0008169324008914e-06, "loss": 0.1313, "step": 47693 }, { "epoch": 34.94065934065934, "grad_norm": 1.2265625, "learning_rate": 1.0007973659203753e-06, "loss": 0.1276, "step": 47694 }, { "epoch": 34.94139194139194, "grad_norm": 0.734375, "learning_rate": 1.0007780366067921e-06, "loss": 0.1324, "step": 47695 }, { "epoch": 34.94212454212454, "grad_norm": 1.0234375, "learning_rate": 1.0007589444602298e-06, "loss": 0.0983, "step": 47696 }, { "epoch": 34.94285714285714, "grad_norm": 0.8046875, "learning_rate": 1.000740089480782e-06, "loss": 0.1226, "step": 47697 }, { "epoch": 34.94358974358974, "grad_norm": 0.7890625, "learning_rate": 1.0007214716685367e-06, "loss": 0.0935, "step": 47698 }, { "epoch": 34.944322344322345, "grad_norm": 1.4765625, "learning_rate": 1.0007030910235817e-06, "loss": 0.2108, "step": 47699 }, { "epoch": 34.94505494505494, "grad_norm": 0.6484375, "learning_rate": 1.0006849475460103e-06, "loss": 0.1036, "step": 47700 }, { "epoch": 34.945787545787546, "grad_norm": 2.125, "learning_rate": 1.0006670412359105e-06, "loss": 0.1409, "step": 47701 }, { "epoch": 34.94652014652014, "grad_norm": 0.94140625, "learning_rate": 1.0006493720933595e-06, "loss": 0.1218, "step": 47702 }, { "epoch": 34.94725274725275, "grad_norm": 0.84765625, "learning_rate": 1.0006319401184449e-06, "loss": 0.1298, "step": 47703 }, { "epoch": 34.947985347985345, "grad_norm": 1.0859375, "learning_rate": 1.0006147453112603e-06, "loss": 0.1154, "step": 47704 }, { "epoch": 34.94871794871795, "grad_norm": 0.94921875, "learning_rate": 1.0005977876718717e-06, "loss": 0.1223, "step": 47705 }, { "epoch": 34.949450549450546, "grad_norm": 0.8359375, "learning_rate": 1.0005810672003722e-06, "loss": 0.1314, "step": 47706 }, { "epoch": 34.95018315018315, "grad_norm": 1.3046875, "learning_rate": 1.0005645838968393e-06, "loss": 0.1335, "step": 47707 }, { "epoch": 34.95091575091575, "grad_norm": 0.98046875, "learning_rate": 1.000548337761349e-06, "loss": 0.1232, "step": 47708 }, { "epoch": 34.95164835164835, "grad_norm": 0.8046875, "learning_rate": 1.000532328793985e-06, "loss": 0.1191, "step": 47709 }, { "epoch": 34.95238095238095, "grad_norm": 0.9453125, "learning_rate": 1.0005165569948179e-06, "loss": 0.1071, "step": 47710 }, { "epoch": 34.95311355311355, "grad_norm": 1.2578125, "learning_rate": 1.0005010223639246e-06, "loss": 0.1315, "step": 47711 }, { "epoch": 34.95384615384615, "grad_norm": 1.3984375, "learning_rate": 1.0004857249013825e-06, "loss": 0.1116, "step": 47712 }, { "epoch": 34.954578754578755, "grad_norm": 1.3671875, "learning_rate": 1.0004706646072626e-06, "loss": 0.124, "step": 47713 }, { "epoch": 34.95531135531135, "grad_norm": 1.0234375, "learning_rate": 1.0004558414816367e-06, "loss": 0.1306, "step": 47714 }, { "epoch": 34.956043956043956, "grad_norm": 0.86328125, "learning_rate": 1.000441255524576e-06, "loss": 0.1006, "step": 47715 }, { "epoch": 34.95677655677655, "grad_norm": 1.1875, "learning_rate": 1.000426906736152e-06, "loss": 0.1363, "step": 47716 }, { "epoch": 34.95750915750916, "grad_norm": 0.5234375, "learning_rate": 1.000412795116431e-06, "loss": 0.1099, "step": 47717 }, { "epoch": 34.958241758241755, "grad_norm": 1.25, "learning_rate": 1.0003989206654838e-06, "loss": 0.098, "step": 47718 }, { "epoch": 34.95897435897436, "grad_norm": 0.8125, "learning_rate": 1.0003852833833712e-06, "loss": 0.1363, "step": 47719 }, { "epoch": 34.959706959706956, "grad_norm": 0.89453125, "learning_rate": 1.0003718832701645e-06, "loss": 0.1313, "step": 47720 }, { "epoch": 34.96043956043956, "grad_norm": 0.984375, "learning_rate": 1.00035872032593e-06, "loss": 0.1113, "step": 47721 }, { "epoch": 34.96117216117216, "grad_norm": 0.984375, "learning_rate": 1.0003457945507223e-06, "loss": 0.1079, "step": 47722 }, { "epoch": 34.96190476190476, "grad_norm": 0.75, "learning_rate": 1.0003331059446076e-06, "loss": 0.104, "step": 47723 }, { "epoch": 34.96263736263736, "grad_norm": 1.296875, "learning_rate": 1.000320654507646e-06, "loss": 0.0868, "step": 47724 }, { "epoch": 34.96336996336996, "grad_norm": 0.8203125, "learning_rate": 1.0003084402398983e-06, "loss": 0.094, "step": 47725 }, { "epoch": 34.96410256410256, "grad_norm": 0.84765625, "learning_rate": 1.0002964631414248e-06, "loss": 0.0979, "step": 47726 }, { "epoch": 34.964835164835165, "grad_norm": 1.09375, "learning_rate": 1.0002847232122804e-06, "loss": 0.1065, "step": 47727 }, { "epoch": 34.96556776556776, "grad_norm": 0.9609375, "learning_rate": 1.0002732204525204e-06, "loss": 0.1204, "step": 47728 }, { "epoch": 34.966300366300366, "grad_norm": 0.8828125, "learning_rate": 1.0002619548621993e-06, "loss": 0.1051, "step": 47729 }, { "epoch": 34.967032967032964, "grad_norm": 0.85546875, "learning_rate": 1.0002509264413722e-06, "loss": 0.1232, "step": 47730 }, { "epoch": 34.96776556776557, "grad_norm": 1.3515625, "learning_rate": 1.000240135190094e-06, "loss": 0.1586, "step": 47731 }, { "epoch": 34.968498168498165, "grad_norm": 1.078125, "learning_rate": 1.00022958110842e-06, "loss": 0.1085, "step": 47732 }, { "epoch": 34.96923076923077, "grad_norm": 1.203125, "learning_rate": 1.0002192641963884e-06, "loss": 0.0988, "step": 47733 }, { "epoch": 34.96996336996337, "grad_norm": 0.984375, "learning_rate": 1.0002091844540595e-06, "loss": 0.1205, "step": 47734 }, { "epoch": 34.97069597069597, "grad_norm": 1.28125, "learning_rate": 1.0001993418814776e-06, "loss": 0.147, "step": 47735 }, { "epoch": 34.97142857142857, "grad_norm": 0.83984375, "learning_rate": 1.000189736478692e-06, "loss": 0.1127, "step": 47736 }, { "epoch": 34.97216117216117, "grad_norm": 0.6796875, "learning_rate": 1.0001803682457464e-06, "loss": 0.1116, "step": 47737 }, { "epoch": 34.97289377289377, "grad_norm": 1.0234375, "learning_rate": 1.0001712371826854e-06, "loss": 0.0995, "step": 47738 }, { "epoch": 34.973626373626374, "grad_norm": 1.46875, "learning_rate": 1.0001623432895578e-06, "loss": 0.1193, "step": 47739 }, { "epoch": 34.97435897435897, "grad_norm": 0.8046875, "learning_rate": 1.000153686566397e-06, "loss": 0.13, "step": 47740 }, { "epoch": 34.975091575091575, "grad_norm": 0.984375, "learning_rate": 1.0001452670132524e-06, "loss": 0.1075, "step": 47741 }, { "epoch": 34.97582417582417, "grad_norm": 1.0859375, "learning_rate": 1.0001370846301567e-06, "loss": 0.1218, "step": 47742 }, { "epoch": 34.97655677655678, "grad_norm": 1.6328125, "learning_rate": 1.0001291394171598e-06, "loss": 0.1387, "step": 47743 }, { "epoch": 34.977289377289374, "grad_norm": 0.921875, "learning_rate": 1.0001214313742887e-06, "loss": 0.0982, "step": 47744 }, { "epoch": 34.97802197802198, "grad_norm": 1.078125, "learning_rate": 1.0001139605015932e-06, "loss": 0.1418, "step": 47745 }, { "epoch": 34.978754578754575, "grad_norm": 1.09375, "learning_rate": 1.0001067267990953e-06, "loss": 0.1122, "step": 47746 }, { "epoch": 34.97948717948718, "grad_norm": 1.0703125, "learning_rate": 1.0000997302668387e-06, "loss": 0.1194, "step": 47747 }, { "epoch": 34.98021978021978, "grad_norm": 1.0546875, "learning_rate": 1.000092970904851e-06, "loss": 0.1371, "step": 47748 }, { "epoch": 34.98095238095238, "grad_norm": 1.015625, "learning_rate": 1.000086448713171e-06, "loss": 0.1539, "step": 47749 }, { "epoch": 34.98168498168498, "grad_norm": 0.57421875, "learning_rate": 1.00008016369182e-06, "loss": 0.112, "step": 47750 }, { "epoch": 34.98241758241758, "grad_norm": 1.109375, "learning_rate": 1.0000741158408424e-06, "loss": 0.136, "step": 47751 }, { "epoch": 34.98315018315018, "grad_norm": 1.078125, "learning_rate": 1.000068305160255e-06, "loss": 0.1085, "step": 47752 }, { "epoch": 34.983882783882784, "grad_norm": 0.953125, "learning_rate": 1.0000627316500954e-06, "loss": 0.1185, "step": 47753 }, { "epoch": 34.98461538461538, "grad_norm": 0.921875, "learning_rate": 1.000057395310381e-06, "loss": 0.1036, "step": 47754 }, { "epoch": 34.985347985347985, "grad_norm": 0.6640625, "learning_rate": 1.0000522961411385e-06, "loss": 0.1073, "step": 47755 }, { "epoch": 34.98608058608059, "grad_norm": 1.09375, "learning_rate": 1.0000474341424013e-06, "loss": 0.1507, "step": 47756 }, { "epoch": 34.98681318681319, "grad_norm": 0.68359375, "learning_rate": 1.0000428093141803e-06, "loss": 0.1184, "step": 47757 }, { "epoch": 34.987545787545784, "grad_norm": 0.8671875, "learning_rate": 1.0000384216565087e-06, "loss": 0.1164, "step": 47758 }, { "epoch": 34.98827838827839, "grad_norm": 0.9609375, "learning_rate": 1.0000342711694029e-06, "loss": 0.1129, "step": 47759 }, { "epoch": 34.98901098901099, "grad_norm": 1.0703125, "learning_rate": 1.0000303578528843e-06, "loss": 0.1643, "step": 47760 }, { "epoch": 34.98974358974359, "grad_norm": 1.1328125, "learning_rate": 1.0000266817069647e-06, "loss": 0.113, "step": 47761 }, { "epoch": 34.99047619047619, "grad_norm": 1.3046875, "learning_rate": 1.000023242731671e-06, "loss": 0.134, "step": 47762 }, { "epoch": 34.99120879120879, "grad_norm": 1.140625, "learning_rate": 1.000020040927015e-06, "loss": 0.1082, "step": 47763 }, { "epoch": 34.991941391941396, "grad_norm": 1.2578125, "learning_rate": 1.000017076293012e-06, "loss": 0.1071, "step": 47764 }, { "epoch": 34.99267399267399, "grad_norm": 0.66796875, "learning_rate": 1.0000143488296738e-06, "loss": 0.1151, "step": 47765 }, { "epoch": 34.99340659340659, "grad_norm": 0.93359375, "learning_rate": 1.0000118585370223e-06, "loss": 0.1041, "step": 47766 }, { "epoch": 34.994139194139194, "grad_norm": 0.94921875, "learning_rate": 1.0000096054150571e-06, "loss": 0.1206, "step": 47767 }, { "epoch": 34.9948717948718, "grad_norm": 0.72265625, "learning_rate": 1.0000075894638008e-06, "loss": 0.1021, "step": 47768 }, { "epoch": 34.995604395604396, "grad_norm": 0.8828125, "learning_rate": 1.0000058106832583e-06, "loss": 0.1227, "step": 47769 }, { "epoch": 34.99633699633699, "grad_norm": 1.7109375, "learning_rate": 1.0000042690734357e-06, "loss": 0.1355, "step": 47770 }, { "epoch": 34.9970695970696, "grad_norm": 1.21875, "learning_rate": 1.0000029646343435e-06, "loss": 0.1194, "step": 47771 }, { "epoch": 34.9978021978022, "grad_norm": 1.2734375, "learning_rate": 1.0000018973659872e-06, "loss": 0.1696, "step": 47772 }, { "epoch": 34.9985347985348, "grad_norm": 0.72265625, "learning_rate": 1.0000010672683726e-06, "loss": 0.1245, "step": 47773 }, { "epoch": 34.999267399267396, "grad_norm": 0.578125, "learning_rate": 1.0000004743414995e-06, "loss": 0.1239, "step": 47774 }, { "epoch": 35.0, "grad_norm": 0.8515625, "learning_rate": 1.0000001185853734e-06, "loss": 0.1143, "step": 47775 }, { "epoch": 35.0, "step": 47775, "total_flos": 0.0, "train_loss": 0.1590119619053861, "train_runtime": 39640.5809, "train_samples_per_second": 308.564, "train_steps_per_second": 1.205 } ], "logging_steps": 1, "max_steps": 47775, "num_input_tokens_seen": 0, "num_train_epochs": 35, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 32, "trial_name": null, "trial_params": null }