| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 5.0, |
| "eval_steps": 500, |
| "global_step": 1235, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.004048582995951417, |
| "grad_norm": 8.468670396589323, |
| "learning_rate": 3.2258064516129035e-07, |
| "loss": 1.5911, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.008097165991902834, |
| "grad_norm": 8.691189397723033, |
| "learning_rate": 6.451612903225807e-07, |
| "loss": 1.6309, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.012145748987854251, |
| "grad_norm": 8.799958437715517, |
| "learning_rate": 9.67741935483871e-07, |
| "loss": 1.6228, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.016194331983805668, |
| "grad_norm": 8.429586432045182, |
| "learning_rate": 1.2903225806451614e-06, |
| "loss": 1.626, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.020242914979757085, |
| "grad_norm": 7.85236480941377, |
| "learning_rate": 1.6129032258064516e-06, |
| "loss": 1.5885, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.024291497975708502, |
| "grad_norm": 7.884862723462013, |
| "learning_rate": 1.935483870967742e-06, |
| "loss": 1.5935, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.02834008097165992, |
| "grad_norm": 6.2888603996381445, |
| "learning_rate": 2.2580645161290324e-06, |
| "loss": 1.5559, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.032388663967611336, |
| "grad_norm": 5.94909788331104, |
| "learning_rate": 2.580645161290323e-06, |
| "loss": 1.5535, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.03643724696356275, |
| "grad_norm": 3.4362437614329533, |
| "learning_rate": 2.903225806451613e-06, |
| "loss": 1.5007, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.04048582995951417, |
| "grad_norm": 3.136793342078447, |
| "learning_rate": 3.225806451612903e-06, |
| "loss": 1.4906, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.044534412955465584, |
| "grad_norm": 2.9455478480998245, |
| "learning_rate": 3.548387096774194e-06, |
| "loss": 1.4886, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.048582995951417005, |
| "grad_norm": 5.777931155774364, |
| "learning_rate": 3.870967741935484e-06, |
| "loss": 1.4634, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.05263157894736842, |
| "grad_norm": 5.726639205385707, |
| "learning_rate": 4.193548387096774e-06, |
| "loss": 1.45, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.05668016194331984, |
| "grad_norm": 5.663817557248407, |
| "learning_rate": 4.516129032258065e-06, |
| "loss": 1.4511, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.06072874493927125, |
| "grad_norm": 5.347057834886526, |
| "learning_rate": 4.838709677419355e-06, |
| "loss": 1.4423, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.06477732793522267, |
| "grad_norm": 4.137956820524769, |
| "learning_rate": 5.161290322580646e-06, |
| "loss": 1.4142, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.06882591093117409, |
| "grad_norm": 3.8908612292429767, |
| "learning_rate": 5.483870967741935e-06, |
| "loss": 1.3858, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.0728744939271255, |
| "grad_norm": 3.354431216058374, |
| "learning_rate": 5.806451612903226e-06, |
| "loss": 1.3769, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.07692307692307693, |
| "grad_norm": 2.8469722153776296, |
| "learning_rate": 6.129032258064517e-06, |
| "loss": 1.35, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.08097165991902834, |
| "grad_norm": 2.3683541260907837, |
| "learning_rate": 6.451612903225806e-06, |
| "loss": 1.3482, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.08502024291497975, |
| "grad_norm": 2.040546346876341, |
| "learning_rate": 6.774193548387097e-06, |
| "loss": 1.3513, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.08906882591093117, |
| "grad_norm": 1.854230306906381, |
| "learning_rate": 7.096774193548388e-06, |
| "loss": 1.3377, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.0931174089068826, |
| "grad_norm": 2.0863681624852495, |
| "learning_rate": 7.4193548387096784e-06, |
| "loss": 1.3263, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.09716599190283401, |
| "grad_norm": 1.9502525386841831, |
| "learning_rate": 7.741935483870968e-06, |
| "loss": 1.2962, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.10121457489878542, |
| "grad_norm": 1.5994390861322147, |
| "learning_rate": 8.064516129032258e-06, |
| "loss": 1.3124, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.10526315789473684, |
| "grad_norm": 1.1660894506695592, |
| "learning_rate": 8.387096774193549e-06, |
| "loss": 1.2945, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.10931174089068826, |
| "grad_norm": 1.5297419018567813, |
| "learning_rate": 8.70967741935484e-06, |
| "loss": 1.2819, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.11336032388663968, |
| "grad_norm": 1.6797616301389968, |
| "learning_rate": 9.03225806451613e-06, |
| "loss": 1.2324, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.11740890688259109, |
| "grad_norm": 1.1312982501973141, |
| "learning_rate": 9.35483870967742e-06, |
| "loss": 1.2524, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.1214574898785425, |
| "grad_norm": 1.2135170779902373, |
| "learning_rate": 9.67741935483871e-06, |
| "loss": 1.2637, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.12550607287449392, |
| "grad_norm": 1.447824892485478, |
| "learning_rate": 1e-05, |
| "loss": 1.2596, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.12955465587044535, |
| "grad_norm": 1.1443261131082354, |
| "learning_rate": 1.0322580645161291e-05, |
| "loss": 1.2179, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.13360323886639677, |
| "grad_norm": 1.3141489141621714, |
| "learning_rate": 1.0645161290322582e-05, |
| "loss": 1.2514, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.13765182186234817, |
| "grad_norm": 1.1881357633097882, |
| "learning_rate": 1.096774193548387e-05, |
| "loss": 1.2447, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.1417004048582996, |
| "grad_norm": 1.00144858664447, |
| "learning_rate": 1.1290322580645164e-05, |
| "loss": 1.2038, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.145748987854251, |
| "grad_norm": 1.2393155211589486, |
| "learning_rate": 1.1612903225806453e-05, |
| "loss": 1.2375, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.14979757085020243, |
| "grad_norm": 0.8498444739639609, |
| "learning_rate": 1.1935483870967743e-05, |
| "loss": 1.227, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.15384615384615385, |
| "grad_norm": 1.063921159557141, |
| "learning_rate": 1.2258064516129034e-05, |
| "loss": 1.2253, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.15789473684210525, |
| "grad_norm": 1.1439131166276195, |
| "learning_rate": 1.2580645161290324e-05, |
| "loss": 1.2596, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.16194331983805668, |
| "grad_norm": 0.9395020437493491, |
| "learning_rate": 1.2903225806451613e-05, |
| "loss": 1.2277, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.1659919028340081, |
| "grad_norm": 0.7629304829301455, |
| "learning_rate": 1.3225806451612903e-05, |
| "loss": 1.2275, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.1700404858299595, |
| "grad_norm": 1.0204249962433083, |
| "learning_rate": 1.3548387096774194e-05, |
| "loss": 1.2028, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.17408906882591094, |
| "grad_norm": 1.0229778723843508, |
| "learning_rate": 1.3870967741935486e-05, |
| "loss": 1.231, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.17813765182186234, |
| "grad_norm": 0.8625383573764686, |
| "learning_rate": 1.4193548387096776e-05, |
| "loss": 1.1859, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.18218623481781376, |
| "grad_norm": 1.007342455102793, |
| "learning_rate": 1.4516129032258066e-05, |
| "loss": 1.2022, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.1862348178137652, |
| "grad_norm": 1.1563222500369228, |
| "learning_rate": 1.4838709677419357e-05, |
| "loss": 1.1943, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.1902834008097166, |
| "grad_norm": 0.8639722868893267, |
| "learning_rate": 1.5161290322580646e-05, |
| "loss": 1.1849, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.19433198380566802, |
| "grad_norm": 1.2768814574402845, |
| "learning_rate": 1.5483870967741936e-05, |
| "loss": 1.1643, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.19838056680161945, |
| "grad_norm": 0.9901235884578591, |
| "learning_rate": 1.5806451612903226e-05, |
| "loss": 1.2096, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.20242914979757085, |
| "grad_norm": 1.6789616253186337, |
| "learning_rate": 1.6129032258064517e-05, |
| "loss": 1.1598, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.20647773279352227, |
| "grad_norm": 1.0310361097594467, |
| "learning_rate": 1.6451612903225807e-05, |
| "loss": 1.2002, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.21052631578947367, |
| "grad_norm": 1.6076182411252293, |
| "learning_rate": 1.6774193548387098e-05, |
| "loss": 1.1768, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.2145748987854251, |
| "grad_norm": 1.6422908536576966, |
| "learning_rate": 1.7096774193548388e-05, |
| "loss": 1.1762, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.21862348178137653, |
| "grad_norm": 1.2087548828262216, |
| "learning_rate": 1.741935483870968e-05, |
| "loss": 1.2064, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.22267206477732793, |
| "grad_norm": 1.4203327451162184, |
| "learning_rate": 1.774193548387097e-05, |
| "loss": 1.2108, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.22672064777327935, |
| "grad_norm": 1.2015980535031856, |
| "learning_rate": 1.806451612903226e-05, |
| "loss": 1.1955, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.23076923076923078, |
| "grad_norm": 1.2486579846419033, |
| "learning_rate": 1.838709677419355e-05, |
| "loss": 1.2089, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.23481781376518218, |
| "grad_norm": 1.0592822787926586, |
| "learning_rate": 1.870967741935484e-05, |
| "loss": 1.1865, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.2388663967611336, |
| "grad_norm": 1.4763978813139054, |
| "learning_rate": 1.903225806451613e-05, |
| "loss": 1.1705, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.242914979757085, |
| "grad_norm": 1.3324511464263518, |
| "learning_rate": 1.935483870967742e-05, |
| "loss": 1.1974, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.24696356275303644, |
| "grad_norm": 1.0651058354296739, |
| "learning_rate": 1.967741935483871e-05, |
| "loss": 1.1734, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.25101214574898784, |
| "grad_norm": 1.7789764497322447, |
| "learning_rate": 2e-05, |
| "loss": 1.1356, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.2550607287449393, |
| "grad_norm": 1.2792652225357244, |
| "learning_rate": 2.0322580645161292e-05, |
| "loss": 1.1953, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.2591093117408907, |
| "grad_norm": 1.9788327777704986, |
| "learning_rate": 2.0645161290322582e-05, |
| "loss": 1.1705, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.2631578947368421, |
| "grad_norm": 1.6898309530695694, |
| "learning_rate": 2.0967741935483873e-05, |
| "loss": 1.203, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.26720647773279355, |
| "grad_norm": 1.824595763057766, |
| "learning_rate": 2.1290322580645163e-05, |
| "loss": 1.1685, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.27125506072874495, |
| "grad_norm": 1.5322310767138214, |
| "learning_rate": 2.161290322580645e-05, |
| "loss": 1.1808, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.27530364372469635, |
| "grad_norm": 1.9731476285253464, |
| "learning_rate": 2.193548387096774e-05, |
| "loss": 1.1495, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.2793522267206478, |
| "grad_norm": 1.4331973516569336, |
| "learning_rate": 2.225806451612903e-05, |
| "loss": 1.1864, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.2834008097165992, |
| "grad_norm": 2.2839195301286357, |
| "learning_rate": 2.2580645161290328e-05, |
| "loss": 1.216, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.2874493927125506, |
| "grad_norm": 1.9445793775539835, |
| "learning_rate": 2.2903225806451615e-05, |
| "loss": 1.2004, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.291497975708502, |
| "grad_norm": 1.952468959592539, |
| "learning_rate": 2.3225806451612906e-05, |
| "loss": 1.1757, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.29554655870445345, |
| "grad_norm": 2.190103166895596, |
| "learning_rate": 2.3548387096774196e-05, |
| "loss": 1.1885, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.29959514170040485, |
| "grad_norm": 1.3894615387583877, |
| "learning_rate": 2.3870967741935486e-05, |
| "loss": 1.1662, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.30364372469635625, |
| "grad_norm": 1.4296102077248716, |
| "learning_rate": 2.4193548387096777e-05, |
| "loss": 1.1788, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.3076923076923077, |
| "grad_norm": 1.4210118528441704, |
| "learning_rate": 2.4516129032258067e-05, |
| "loss": 1.205, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.3117408906882591, |
| "grad_norm": 1.2506957239256302, |
| "learning_rate": 2.4838709677419358e-05, |
| "loss": 1.1415, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.3157894736842105, |
| "grad_norm": 1.6310132671412647, |
| "learning_rate": 2.5161290322580648e-05, |
| "loss": 1.177, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.31983805668016196, |
| "grad_norm": 1.581376421879731, |
| "learning_rate": 2.548387096774194e-05, |
| "loss": 1.1678, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.32388663967611336, |
| "grad_norm": 1.431847688600866, |
| "learning_rate": 2.5806451612903226e-05, |
| "loss": 1.1754, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.32793522267206476, |
| "grad_norm": 1.2649657954313007, |
| "learning_rate": 2.6129032258064516e-05, |
| "loss": 1.1835, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.3319838056680162, |
| "grad_norm": 1.1030784288101083, |
| "learning_rate": 2.6451612903225806e-05, |
| "loss": 1.1328, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.3360323886639676, |
| "grad_norm": 1.6325923992767415, |
| "learning_rate": 2.6774193548387097e-05, |
| "loss": 1.1509, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.340080971659919, |
| "grad_norm": 1.0937283370473252, |
| "learning_rate": 2.7096774193548387e-05, |
| "loss": 1.1697, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.3441295546558704, |
| "grad_norm": 1.7268341557987534, |
| "learning_rate": 2.7419354838709678e-05, |
| "loss": 1.193, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.3481781376518219, |
| "grad_norm": 1.2678314457755464, |
| "learning_rate": 2.774193548387097e-05, |
| "loss": 1.1652, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.3522267206477733, |
| "grad_norm": 1.6582127450791697, |
| "learning_rate": 2.8064516129032262e-05, |
| "loss": 1.1467, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.3562753036437247, |
| "grad_norm": 1.6208629141761544, |
| "learning_rate": 2.8387096774193552e-05, |
| "loss": 1.1971, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.3603238866396761, |
| "grad_norm": 1.2378264177407658, |
| "learning_rate": 2.8709677419354843e-05, |
| "loss": 1.1545, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.3643724696356275, |
| "grad_norm": 1.411442446572754, |
| "learning_rate": 2.9032258064516133e-05, |
| "loss": 1.1933, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.3684210526315789, |
| "grad_norm": 1.5395778891104268, |
| "learning_rate": 2.9354838709677423e-05, |
| "loss": 1.1933, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.3724696356275304, |
| "grad_norm": 1.4036375831720875, |
| "learning_rate": 2.9677419354838714e-05, |
| "loss": 1.1271, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.3765182186234818, |
| "grad_norm": 1.6715670073380937, |
| "learning_rate": 3.0000000000000004e-05, |
| "loss": 1.1456, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.3805668016194332, |
| "grad_norm": 1.735738490689031, |
| "learning_rate": 3.032258064516129e-05, |
| "loss": 1.1864, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.38461538461538464, |
| "grad_norm": 1.3157368413606774, |
| "learning_rate": 3.0645161290322585e-05, |
| "loss": 1.1025, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.38866396761133604, |
| "grad_norm": 1.6621824805577994, |
| "learning_rate": 3.096774193548387e-05, |
| "loss": 1.1795, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.39271255060728744, |
| "grad_norm": 1.0487816914050725, |
| "learning_rate": 3.1290322580645166e-05, |
| "loss": 1.1525, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.3967611336032389, |
| "grad_norm": 1.2825685499847348, |
| "learning_rate": 3.161290322580645e-05, |
| "loss": 1.1432, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.4008097165991903, |
| "grad_norm": 1.1063115279400753, |
| "learning_rate": 3.193548387096774e-05, |
| "loss": 1.1691, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.4048582995951417, |
| "grad_norm": 1.6103538456486248, |
| "learning_rate": 3.2258064516129034e-05, |
| "loss": 1.1058, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.4089068825910931, |
| "grad_norm": 1.2355374333275022, |
| "learning_rate": 3.258064516129033e-05, |
| "loss": 1.1877, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.41295546558704455, |
| "grad_norm": 1.3719139509402511, |
| "learning_rate": 3.2903225806451614e-05, |
| "loss": 1.1108, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.41700404858299595, |
| "grad_norm": 1.642230628074308, |
| "learning_rate": 3.322580645161291e-05, |
| "loss": 1.1475, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.42105263157894735, |
| "grad_norm": 1.481439572175682, |
| "learning_rate": 3.3548387096774195e-05, |
| "loss": 1.1696, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.4251012145748988, |
| "grad_norm": 1.3558738179883771, |
| "learning_rate": 3.387096774193549e-05, |
| "loss": 1.1962, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.4291497975708502, |
| "grad_norm": 1.284967335071505, |
| "learning_rate": 3.4193548387096776e-05, |
| "loss": 1.1512, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.4331983805668016, |
| "grad_norm": 1.1818253350467165, |
| "learning_rate": 3.451612903225807e-05, |
| "loss": 1.1863, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.43724696356275305, |
| "grad_norm": 1.4370961257452508, |
| "learning_rate": 3.483870967741936e-05, |
| "loss": 1.1688, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.44129554655870445, |
| "grad_norm": 1.6490818410039954, |
| "learning_rate": 3.516129032258065e-05, |
| "loss": 1.135, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.44534412955465585, |
| "grad_norm": 1.1414818893794374, |
| "learning_rate": 3.548387096774194e-05, |
| "loss": 1.1637, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.4493927125506073, |
| "grad_norm": 1.1863122652867433, |
| "learning_rate": 3.5806451612903225e-05, |
| "loss": 1.2007, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.4534412955465587, |
| "grad_norm": 1.832835807831948, |
| "learning_rate": 3.612903225806452e-05, |
| "loss": 1.1451, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.4574898785425101, |
| "grad_norm": 1.3587213989163611, |
| "learning_rate": 3.6451612903225805e-05, |
| "loss": 1.1507, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.46153846153846156, |
| "grad_norm": 1.4322357134641799, |
| "learning_rate": 3.67741935483871e-05, |
| "loss": 1.1474, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.46558704453441296, |
| "grad_norm": 1.9880169287057, |
| "learning_rate": 3.7096774193548386e-05, |
| "loss": 1.1837, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.46963562753036436, |
| "grad_norm": 1.5952492168649937, |
| "learning_rate": 3.741935483870968e-05, |
| "loss": 1.1877, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.47368421052631576, |
| "grad_norm": 1.5636647948571956, |
| "learning_rate": 3.7741935483870974e-05, |
| "loss": 1.1745, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.4777327935222672, |
| "grad_norm": 2.6912891180879326, |
| "learning_rate": 3.806451612903226e-05, |
| "loss": 1.1643, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.4817813765182186, |
| "grad_norm": 1.540133082488959, |
| "learning_rate": 3.8387096774193555e-05, |
| "loss": 1.1838, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.48582995951417, |
| "grad_norm": 3.097935918464308, |
| "learning_rate": 3.870967741935484e-05, |
| "loss": 1.0894, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.4898785425101215, |
| "grad_norm": 3.0781343031513573, |
| "learning_rate": 3.9032258064516135e-05, |
| "loss": 1.1585, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.4939271255060729, |
| "grad_norm": 1.794086284589536, |
| "learning_rate": 3.935483870967742e-05, |
| "loss": 1.1762, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.4979757085020243, |
| "grad_norm": 2.2877208512074243, |
| "learning_rate": 3.9677419354838716e-05, |
| "loss": 1.1631, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.5020242914979757, |
| "grad_norm": 1.4721897713952405, |
| "learning_rate": 4e-05, |
| "loss": 1.1411, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.5060728744939271, |
| "grad_norm": 2.0091553354677183, |
| "learning_rate": 3.9999920040266474e-05, |
| "loss": 1.1483, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.5101214574898786, |
| "grad_norm": 1.5556286398417398, |
| "learning_rate": 3.999968016170525e-05, |
| "loss": 1.1482, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.5141700404858299, |
| "grad_norm": 2.142360574998737, |
| "learning_rate": 3.9999280366234394e-05, |
| "loss": 1.1685, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.5182186234817814, |
| "grad_norm": 2.0753875728653615, |
| "learning_rate": 3.999872065705065e-05, |
| "loss": 1.1799, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.5222672064777328, |
| "grad_norm": 1.5753646742111036, |
| "learning_rate": 3.999800103862945e-05, |
| "loss": 1.1567, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.5263157894736842, |
| "grad_norm": 2.544009957429584, |
| "learning_rate": 3.9997121516724834e-05, |
| "loss": 1.1502, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.5303643724696356, |
| "grad_norm": 1.8755203437424368, |
| "learning_rate": 3.999608209836944e-05, |
| "loss": 1.1435, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.5344129554655871, |
| "grad_norm": 2.9298178769317005, |
| "learning_rate": 3.999488279187443e-05, |
| "loss": 1.1619, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.5384615384615384, |
| "grad_norm": 2.286358493007044, |
| "learning_rate": 3.9993523606829425e-05, |
| "loss": 1.1443, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.5425101214574899, |
| "grad_norm": 2.6804351600907785, |
| "learning_rate": 3.9992004554102435e-05, |
| "loss": 1.1704, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.5465587044534413, |
| "grad_norm": 2.200352067277791, |
| "learning_rate": 3.999032564583976e-05, |
| "loss": 1.1237, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.5506072874493927, |
| "grad_norm": 2.49717308995895, |
| "learning_rate": 3.998848689546591e-05, |
| "loss": 1.114, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.5546558704453441, |
| "grad_norm": 1.9587279267641116, |
| "learning_rate": 3.9986488317683496e-05, |
| "loss": 1.1206, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.5587044534412956, |
| "grad_norm": 2.4163230134480544, |
| "learning_rate": 3.9984329928473067e-05, |
| "loss": 1.1736, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.562753036437247, |
| "grad_norm": 1.8539132055840106, |
| "learning_rate": 3.998201174509306e-05, |
| "loss": 1.1686, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.5668016194331984, |
| "grad_norm": 2.7608102607966267, |
| "learning_rate": 3.997953378607961e-05, |
| "loss": 1.1304, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.5708502024291497, |
| "grad_norm": 1.9152187967500027, |
| "learning_rate": 3.9976896071246403e-05, |
| "loss": 1.1316, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.5748987854251012, |
| "grad_norm": 3.033075461217084, |
| "learning_rate": 3.997409862168454e-05, |
| "loss": 1.146, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.5789473684210527, |
| "grad_norm": 2.6351082003044817, |
| "learning_rate": 3.997114145976237e-05, |
| "loss": 1.1453, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.582995951417004, |
| "grad_norm": 2.1230145861135936, |
| "learning_rate": 3.9968024609125245e-05, |
| "loss": 1.1385, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.5870445344129555, |
| "grad_norm": 2.3818298577291057, |
| "learning_rate": 3.996474809469545e-05, |
| "loss": 1.104, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.5910931174089069, |
| "grad_norm": 1.6485615408029353, |
| "learning_rate": 3.996131194267189e-05, |
| "loss": 1.1522, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.5951417004048583, |
| "grad_norm": 1.8154451692564277, |
| "learning_rate": 3.995771618052995e-05, |
| "loss": 1.1126, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.5991902834008097, |
| "grad_norm": 1.0766329994853043, |
| "learning_rate": 3.9953960837021256e-05, |
| "loss": 1.1781, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.6032388663967612, |
| "grad_norm": 2.0738515031834615, |
| "learning_rate": 3.995004594217343e-05, |
| "loss": 1.1386, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.6072874493927125, |
| "grad_norm": 1.189408526132442, |
| "learning_rate": 3.994597152728987e-05, |
| "loss": 1.1554, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.611336032388664, |
| "grad_norm": 2.212318758619874, |
| "learning_rate": 3.994173762494947e-05, |
| "loss": 1.1083, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.6153846153846154, |
| "grad_norm": 1.8714434043188992, |
| "learning_rate": 3.9937344269006424e-05, |
| "loss": 1.1212, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.6194331983805668, |
| "grad_norm": 1.9505742023086288, |
| "learning_rate": 3.993279149458988e-05, |
| "loss": 1.1599, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.6234817813765182, |
| "grad_norm": 2.109725730948719, |
| "learning_rate": 3.9928079338103695e-05, |
| "loss": 1.1638, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.6275303643724697, |
| "grad_norm": 1.8547573333644882, |
| "learning_rate": 3.992320783722616e-05, |
| "loss": 1.143, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.631578947368421, |
| "grad_norm": 2.2288034821581957, |
| "learning_rate": 3.991817703090965e-05, |
| "loss": 1.1436, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.6356275303643725, |
| "grad_norm": 1.6763902511816258, |
| "learning_rate": 3.991298695938038e-05, |
| "loss": 1.1199, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.6396761133603239, |
| "grad_norm": 2.1682094799828238, |
| "learning_rate": 3.9907637664138e-05, |
| "loss": 1.1343, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.6437246963562753, |
| "grad_norm": 1.4556152223582268, |
| "learning_rate": 3.990212918795534e-05, |
| "loss": 1.1449, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.6477732793522267, |
| "grad_norm": 2.5355892592901217, |
| "learning_rate": 3.9896461574878036e-05, |
| "loss": 1.1799, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.6518218623481782, |
| "grad_norm": 1.6564025798263988, |
| "learning_rate": 3.989063487022416e-05, |
| "loss": 1.1562, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.6558704453441295, |
| "grad_norm": 2.766743472265191, |
| "learning_rate": 3.9884649120583903e-05, |
| "loss": 1.1567, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.659919028340081, |
| "grad_norm": 1.922841669508268, |
| "learning_rate": 3.987850437381915e-05, |
| "loss": 1.1732, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.6639676113360324, |
| "grad_norm": 2.785505302006914, |
| "learning_rate": 3.987220067906313e-05, |
| "loss": 1.1445, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.6680161943319838, |
| "grad_norm": 2.167054913987502, |
| "learning_rate": 3.986573808672002e-05, |
| "loss": 1.1213, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.6720647773279352, |
| "grad_norm": 2.7661054237582157, |
| "learning_rate": 3.9859116648464534e-05, |
| "loss": 1.152, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.6761133603238867, |
| "grad_norm": 2.803853945170692, |
| "learning_rate": 3.985233641724152e-05, |
| "loss": 1.1571, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.680161943319838, |
| "grad_norm": 1.9613904324405322, |
| "learning_rate": 3.984539744726554e-05, |
| "loss": 1.1273, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.6842105263157895, |
| "grad_norm": 2.407350311100867, |
| "learning_rate": 3.983829979402038e-05, |
| "loss": 1.1317, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.6882591093117408, |
| "grad_norm": 1.7409041026154204, |
| "learning_rate": 3.983104351425871e-05, |
| "loss": 1.1261, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.6923076923076923, |
| "grad_norm": 1.5862614977191676, |
| "learning_rate": 3.982362866600154e-05, |
| "loss": 1.1576, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.6963562753036437, |
| "grad_norm": 1.9883837465631475, |
| "learning_rate": 3.9816055308537806e-05, |
| "loss": 1.1686, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.7004048582995951, |
| "grad_norm": 1.580690410263307, |
| "learning_rate": 3.980832350242388e-05, |
| "loss": 1.1574, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.7044534412955465, |
| "grad_norm": 1.9821054096420336, |
| "learning_rate": 3.980043330948306e-05, |
| "loss": 1.1366, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.708502024291498, |
| "grad_norm": 1.5392503566790576, |
| "learning_rate": 3.9792384792805124e-05, |
| "loss": 1.1069, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.7125506072874493, |
| "grad_norm": 2.0349054346022966, |
| "learning_rate": 3.978417801674579e-05, |
| "loss": 1.14, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.7165991902834008, |
| "grad_norm": 1.715864841723352, |
| "learning_rate": 3.9775813046926244e-05, |
| "loss": 1.1088, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.7206477732793523, |
| "grad_norm": 1.6890491558744103, |
| "learning_rate": 3.976728995023254e-05, |
| "loss": 1.1272, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.7246963562753036, |
| "grad_norm": 1.8086672808065176, |
| "learning_rate": 3.975860879481514e-05, |
| "loss": 1.1483, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.728744939271255, |
| "grad_norm": 0.9641481826607671, |
| "learning_rate": 3.974976965008833e-05, |
| "loss": 1.1145, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.7327935222672065, |
| "grad_norm": 2.0417714422293507, |
| "learning_rate": 3.9740772586729674e-05, |
| "loss": 1.1305, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.7368421052631579, |
| "grad_norm": 1.6539493008806427, |
| "learning_rate": 3.973161767667946e-05, |
| "loss": 1.1073, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.7408906882591093, |
| "grad_norm": 2.064550936938208, |
| "learning_rate": 3.972230499314009e-05, |
| "loss": 1.1651, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.7449392712550608, |
| "grad_norm": 1.7320926973571846, |
| "learning_rate": 3.9712834610575545e-05, |
| "loss": 1.143, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.7489878542510121, |
| "grad_norm": 2.1610247778766847, |
| "learning_rate": 3.9703206604710746e-05, |
| "loss": 1.1421, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.7530364372469636, |
| "grad_norm": 1.5328142001078764, |
| "learning_rate": 3.969342105253098e-05, |
| "loss": 1.1169, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.757085020242915, |
| "grad_norm": 2.214076794638587, |
| "learning_rate": 3.9683478032281246e-05, |
| "loss": 1.1375, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.7611336032388664, |
| "grad_norm": 1.3997304137501776, |
| "learning_rate": 3.967337762346568e-05, |
| "loss": 1.1569, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.7651821862348178, |
| "grad_norm": 1.7855159418283086, |
| "learning_rate": 3.9663119906846885e-05, |
| "loss": 1.1547, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.7692307692307693, |
| "grad_norm": 1.6370859125197479, |
| "learning_rate": 3.965270496444528e-05, |
| "loss": 1.1262, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.7732793522267206, |
| "grad_norm": 1.4942813557593602, |
| "learning_rate": 3.9642132879538465e-05, |
| "loss": 1.1419, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.7773279352226721, |
| "grad_norm": 2.004226012383275, |
| "learning_rate": 3.9631403736660566e-05, |
| "loss": 1.1152, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.7813765182186235, |
| "grad_norm": 1.8848989277474746, |
| "learning_rate": 3.962051762160151e-05, |
| "loss": 1.1475, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.7854251012145749, |
| "grad_norm": 2.2158707808868514, |
| "learning_rate": 3.960947462140638e-05, |
| "loss": 1.1377, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.7894736842105263, |
| "grad_norm": 2.031869698914545, |
| "learning_rate": 3.959827482437472e-05, |
| "loss": 1.0749, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.7935222672064778, |
| "grad_norm": 1.9455903033517878, |
| "learning_rate": 3.958691832005981e-05, |
| "loss": 1.1347, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.7975708502024291, |
| "grad_norm": 1.9288369842805395, |
| "learning_rate": 3.9575405199267944e-05, |
| "loss": 1.1573, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.8016194331983806, |
| "grad_norm": 1.5245052570376225, |
| "learning_rate": 3.956373555405774e-05, |
| "loss": 1.1359, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.805668016194332, |
| "grad_norm": 1.700823149554145, |
| "learning_rate": 3.955190947773937e-05, |
| "loss": 1.1668, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.8097165991902834, |
| "grad_norm": 1.5455373112504782, |
| "learning_rate": 3.9539927064873824e-05, |
| "loss": 1.1524, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.8137651821862348, |
| "grad_norm": 1.2965234612291925, |
| "learning_rate": 3.952778841127214e-05, |
| "loss": 1.1373, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.8178137651821862, |
| "grad_norm": 1.4383559680297429, |
| "learning_rate": 3.95154936139947e-05, |
| "loss": 1.1463, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.8218623481781376, |
| "grad_norm": 1.5097828480310806, |
| "learning_rate": 3.950304277135034e-05, |
| "loss": 1.1295, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.8259109311740891, |
| "grad_norm": 1.4337992634345824, |
| "learning_rate": 3.94904359828957e-05, |
| "loss": 1.1291, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.8299595141700404, |
| "grad_norm": 1.3364806737481965, |
| "learning_rate": 3.94776733494343e-05, |
| "loss": 1.1045, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.8340080971659919, |
| "grad_norm": 1.8825851672912648, |
| "learning_rate": 3.9464754973015826e-05, |
| "loss": 1.1205, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.8380566801619433, |
| "grad_norm": 1.0378056840992604, |
| "learning_rate": 3.9451680956935275e-05, |
| "loss": 1.1064, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.8421052631578947, |
| "grad_norm": 2.539102978762906, |
| "learning_rate": 3.943845140573213e-05, |
| "loss": 1.1334, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.8461538461538461, |
| "grad_norm": 1.4228322054678486, |
| "learning_rate": 3.942506642518952e-05, |
| "loss": 1.1194, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.8502024291497976, |
| "grad_norm": 2.770624780837206, |
| "learning_rate": 3.9411526122333396e-05, |
| "loss": 1.1261, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.854251012145749, |
| "grad_norm": 2.103906808523346, |
| "learning_rate": 3.939783060543167e-05, |
| "loss": 1.1779, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.8582995951417004, |
| "grad_norm": 2.467472422250228, |
| "learning_rate": 3.938397998399333e-05, |
| "loss": 1.1379, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.8623481781376519, |
| "grad_norm": 2.390613280105405, |
| "learning_rate": 3.936997436876756e-05, |
| "loss": 1.17, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.8663967611336032, |
| "grad_norm": 1.6866103919864177, |
| "learning_rate": 3.935581387174289e-05, |
| "loss": 1.1383, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.8704453441295547, |
| "grad_norm": 1.8428249186556946, |
| "learning_rate": 3.9341498606146295e-05, |
| "loss": 1.1476, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.8744939271255061, |
| "grad_norm": 1.6085981090920765, |
| "learning_rate": 3.932702868644224e-05, |
| "loss": 1.1458, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.8785425101214575, |
| "grad_norm": 1.4923770180900617, |
| "learning_rate": 3.931240422833181e-05, |
| "loss": 1.143, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.8825910931174089, |
| "grad_norm": 1.3611333122845923, |
| "learning_rate": 3.9297625348751806e-05, |
| "loss": 1.112, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.8866396761133604, |
| "grad_norm": 1.489239383256606, |
| "learning_rate": 3.9282692165873744e-05, |
| "loss": 1.1946, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.8906882591093117, |
| "grad_norm": 1.7092832936647853, |
| "learning_rate": 3.9267604799102945e-05, |
| "loss": 1.153, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.8947368421052632, |
| "grad_norm": 1.1744737215224805, |
| "learning_rate": 3.9252363369077615e-05, |
| "loss": 1.1394, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.8987854251012146, |
| "grad_norm": 2.1179978910041464, |
| "learning_rate": 3.9236967997667796e-05, |
| "loss": 1.1127, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.902834008097166, |
| "grad_norm": 1.7771839866735364, |
| "learning_rate": 3.9221418807974496e-05, |
| "loss": 1.1222, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.9068825910931174, |
| "grad_norm": 1.7967990931656908, |
| "learning_rate": 3.92057159243286e-05, |
| "loss": 1.1391, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.9109311740890689, |
| "grad_norm": 1.9298446802300546, |
| "learning_rate": 3.9189859472289956e-05, |
| "loss": 1.1242, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.9149797570850202, |
| "grad_norm": 1.5211277807370966, |
| "learning_rate": 3.9173849578646324e-05, |
| "loss": 1.0882, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.9190283400809717, |
| "grad_norm": 2.0077025258799974, |
| "learning_rate": 3.9157686371412395e-05, |
| "loss": 1.1014, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.9230769230769231, |
| "grad_norm": 1.6611618267904193, |
| "learning_rate": 3.914136997982874e-05, |
| "loss": 1.1377, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.9271255060728745, |
| "grad_norm": 2.6746708278517572, |
| "learning_rate": 3.912490053436079e-05, |
| "loss": 1.1695, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.9311740890688259, |
| "grad_norm": 1.794296492364274, |
| "learning_rate": 3.9108278166697805e-05, |
| "loss": 1.1173, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.9352226720647774, |
| "grad_norm": 2.753577370560727, |
| "learning_rate": 3.909150300975178e-05, |
| "loss": 1.1317, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.9392712550607287, |
| "grad_norm": 2.741417788974489, |
| "learning_rate": 3.907457519765642e-05, |
| "loss": 1.1649, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.9433198380566802, |
| "grad_norm": 1.5638492583264574, |
| "learning_rate": 3.905749486576607e-05, |
| "loss": 1.1433, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.9473684210526315, |
| "grad_norm": 2.035132203685619, |
| "learning_rate": 3.90402621506546e-05, |
| "loss": 1.1321, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.951417004048583, |
| "grad_norm": 1.742026930646904, |
| "learning_rate": 3.902287719011435e-05, |
| "loss": 1.1199, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.9554655870445344, |
| "grad_norm": 1.3867858081233415, |
| "learning_rate": 3.900534012315499e-05, |
| "loss": 1.1015, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.9595141700404858, |
| "grad_norm": 1.5031079776839218, |
| "learning_rate": 3.8987651090002454e-05, |
| "loss": 1.1053, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.9635627530364372, |
| "grad_norm": 1.425252700573384, |
| "learning_rate": 3.896981023209777e-05, |
| "loss": 1.0946, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.9676113360323887, |
| "grad_norm": 0.8258979885448275, |
| "learning_rate": 3.8951817692095966e-05, |
| "loss": 1.1378, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.97165991902834, |
| "grad_norm": 1.619281593398078, |
| "learning_rate": 3.893367361386491e-05, |
| "loss": 1.1244, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.9757085020242915, |
| "grad_norm": 1.366546001068629, |
| "learning_rate": 3.891537814248417e-05, |
| "loss": 1.1203, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.979757085020243, |
| "grad_norm": 0.9322737248639651, |
| "learning_rate": 3.889693142424384e-05, |
| "loss": 1.0844, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.9838056680161943, |
| "grad_norm": 1.6252011675836653, |
| "learning_rate": 3.8878333606643405e-05, |
| "loss": 1.1482, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.9878542510121457, |
| "grad_norm": 1.4693589037213186, |
| "learning_rate": 3.885958483839049e-05, |
| "loss": 1.1422, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.9919028340080972, |
| "grad_norm": 1.112244890146087, |
| "learning_rate": 3.8840685269399786e-05, |
| "loss": 1.1083, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.9959514170040485, |
| "grad_norm": 1.764640265725053, |
| "learning_rate": 3.882163505079171e-05, |
| "loss": 1.1764, |
| "step": 246 |
| }, |
| { |
| "epoch": 1.0, |
| "grad_norm": 1.2049097845323549, |
| "learning_rate": 3.880243433489132e-05, |
| "loss": 1.105, |
| "step": 247 |
| }, |
| { |
| "epoch": 1.0040485829959513, |
| "grad_norm": 1.510212667674659, |
| "learning_rate": 3.878308327522702e-05, |
| "loss": 1.0672, |
| "step": 248 |
| }, |
| { |
| "epoch": 1.008097165991903, |
| "grad_norm": 1.2400587482336372, |
| "learning_rate": 3.876358202652936e-05, |
| "loss": 1.0568, |
| "step": 249 |
| }, |
| { |
| "epoch": 1.0121457489878543, |
| "grad_norm": 1.43574045637118, |
| "learning_rate": 3.8743930744729835e-05, |
| "loss": 1.068, |
| "step": 250 |
| }, |
| { |
| "epoch": 1.0161943319838056, |
| "grad_norm": 1.321370218982357, |
| "learning_rate": 3.8724129586959535e-05, |
| "loss": 1.0299, |
| "step": 251 |
| }, |
| { |
| "epoch": 1.0202429149797572, |
| "grad_norm": 1.4087386640776962, |
| "learning_rate": 3.870417871154801e-05, |
| "loss": 1.0637, |
| "step": 252 |
| }, |
| { |
| "epoch": 1.0242914979757085, |
| "grad_norm": 1.4616006999554387, |
| "learning_rate": 3.868407827802192e-05, |
| "loss": 1.0417, |
| "step": 253 |
| }, |
| { |
| "epoch": 1.0283400809716599, |
| "grad_norm": 1.2524451647826127, |
| "learning_rate": 3.8663828447103804e-05, |
| "loss": 1.0729, |
| "step": 254 |
| }, |
| { |
| "epoch": 1.0323886639676114, |
| "grad_norm": 1.5022341452097843, |
| "learning_rate": 3.8643429380710765e-05, |
| "loss": 1.0348, |
| "step": 255 |
| }, |
| { |
| "epoch": 1.0364372469635628, |
| "grad_norm": 1.2249236110067723, |
| "learning_rate": 3.862288124195319e-05, |
| "loss": 1.0246, |
| "step": 256 |
| }, |
| { |
| "epoch": 1.040485829959514, |
| "grad_norm": 1.3027400226251726, |
| "learning_rate": 3.8602184195133464e-05, |
| "loss": 1.0576, |
| "step": 257 |
| }, |
| { |
| "epoch": 1.0445344129554657, |
| "grad_norm": 1.1963810979752776, |
| "learning_rate": 3.8581338405744604e-05, |
| "loss": 1.0154, |
| "step": 258 |
| }, |
| { |
| "epoch": 1.048582995951417, |
| "grad_norm": 1.703832314197434, |
| "learning_rate": 3.8560344040468996e-05, |
| "loss": 1.0553, |
| "step": 259 |
| }, |
| { |
| "epoch": 1.0526315789473684, |
| "grad_norm": 1.18928876118049, |
| "learning_rate": 3.853920126717703e-05, |
| "loss": 1.0308, |
| "step": 260 |
| }, |
| { |
| "epoch": 1.05668016194332, |
| "grad_norm": 1.6703303688234696, |
| "learning_rate": 3.851791025492574e-05, |
| "loss": 1.0524, |
| "step": 261 |
| }, |
| { |
| "epoch": 1.0607287449392713, |
| "grad_norm": 1.1092683273378885, |
| "learning_rate": 3.8496471173957515e-05, |
| "loss": 1.0383, |
| "step": 262 |
| }, |
| { |
| "epoch": 1.0647773279352226, |
| "grad_norm": 1.5384660949394466, |
| "learning_rate": 3.847488419569866e-05, |
| "loss": 1.033, |
| "step": 263 |
| }, |
| { |
| "epoch": 1.0688259109311742, |
| "grad_norm": 1.449968716766826, |
| "learning_rate": 3.8453149492758077e-05, |
| "loss": 1.0527, |
| "step": 264 |
| }, |
| { |
| "epoch": 1.0728744939271255, |
| "grad_norm": 1.2664937837110435, |
| "learning_rate": 3.8431267238925886e-05, |
| "loss": 1.0182, |
| "step": 265 |
| }, |
| { |
| "epoch": 1.0769230769230769, |
| "grad_norm": 1.5630053412076979, |
| "learning_rate": 3.840923760917199e-05, |
| "loss": 1.0327, |
| "step": 266 |
| }, |
| { |
| "epoch": 1.0809716599190284, |
| "grad_norm": 1.345520161489642, |
| "learning_rate": 3.838706077964473e-05, |
| "loss": 1.0216, |
| "step": 267 |
| }, |
| { |
| "epoch": 1.0850202429149798, |
| "grad_norm": 1.352447243993805, |
| "learning_rate": 3.836473692766944e-05, |
| "loss": 1.0715, |
| "step": 268 |
| }, |
| { |
| "epoch": 1.0890688259109311, |
| "grad_norm": 1.8181792387421152, |
| "learning_rate": 3.834226623174705e-05, |
| "loss": 1.0594, |
| "step": 269 |
| }, |
| { |
| "epoch": 1.0931174089068827, |
| "grad_norm": 1.4509199327634335, |
| "learning_rate": 3.831964887155264e-05, |
| "loss": 1.0264, |
| "step": 270 |
| }, |
| { |
| "epoch": 1.097165991902834, |
| "grad_norm": 1.472047649439964, |
| "learning_rate": 3.829688502793403e-05, |
| "loss": 0.9963, |
| "step": 271 |
| }, |
| { |
| "epoch": 1.1012145748987854, |
| "grad_norm": 1.7304243908022598, |
| "learning_rate": 3.827397488291029e-05, |
| "loss": 1.0774, |
| "step": 272 |
| }, |
| { |
| "epoch": 1.1052631578947367, |
| "grad_norm": 1.301043127202309, |
| "learning_rate": 3.8250918619670343e-05, |
| "loss": 0.9913, |
| "step": 273 |
| }, |
| { |
| "epoch": 1.1093117408906883, |
| "grad_norm": 1.3871478048681323, |
| "learning_rate": 3.822771642257145e-05, |
| "loss": 1.0462, |
| "step": 274 |
| }, |
| { |
| "epoch": 1.1133603238866396, |
| "grad_norm": 1.0439586117033608, |
| "learning_rate": 3.820436847713776e-05, |
| "loss": 1.0388, |
| "step": 275 |
| }, |
| { |
| "epoch": 1.117408906882591, |
| "grad_norm": 1.1949537665826693, |
| "learning_rate": 3.8180874970058826e-05, |
| "loss": 1.0335, |
| "step": 276 |
| }, |
| { |
| "epoch": 1.1214574898785425, |
| "grad_norm": 1.0915308729929967, |
| "learning_rate": 3.8157236089188103e-05, |
| "loss": 1.0209, |
| "step": 277 |
| }, |
| { |
| "epoch": 1.125506072874494, |
| "grad_norm": 1.2416556264438208, |
| "learning_rate": 3.813345202354145e-05, |
| "loss": 1.0557, |
| "step": 278 |
| }, |
| { |
| "epoch": 1.1295546558704452, |
| "grad_norm": 1.2563830536853853, |
| "learning_rate": 3.810952296329563e-05, |
| "loss": 1.0374, |
| "step": 279 |
| }, |
| { |
| "epoch": 1.1336032388663968, |
| "grad_norm": 1.1068744690595436, |
| "learning_rate": 3.808544909978676e-05, |
| "loss": 1.0498, |
| "step": 280 |
| }, |
| { |
| "epoch": 1.1376518218623481, |
| "grad_norm": 1.3671395937103639, |
| "learning_rate": 3.806123062550882e-05, |
| "loss": 1.0625, |
| "step": 281 |
| }, |
| { |
| "epoch": 1.1417004048582995, |
| "grad_norm": 1.2697768028846346, |
| "learning_rate": 3.803686773411208e-05, |
| "loss": 1.0488, |
| "step": 282 |
| }, |
| { |
| "epoch": 1.145748987854251, |
| "grad_norm": 1.2652039352529507, |
| "learning_rate": 3.801236062040158e-05, |
| "loss": 1.0173, |
| "step": 283 |
| }, |
| { |
| "epoch": 1.1497975708502024, |
| "grad_norm": 1.0179970827484477, |
| "learning_rate": 3.798770948033553e-05, |
| "loss": 1.0532, |
| "step": 284 |
| }, |
| { |
| "epoch": 1.1538461538461537, |
| "grad_norm": 1.3844700336614149, |
| "learning_rate": 3.796291451102381e-05, |
| "loss": 1.0341, |
| "step": 285 |
| }, |
| { |
| "epoch": 1.1578947368421053, |
| "grad_norm": 1.0796107443554654, |
| "learning_rate": 3.793797591072631e-05, |
| "loss": 1.0396, |
| "step": 286 |
| }, |
| { |
| "epoch": 1.1619433198380567, |
| "grad_norm": 1.9058243745792902, |
| "learning_rate": 3.791289387885144e-05, |
| "loss": 1.0407, |
| "step": 287 |
| }, |
| { |
| "epoch": 1.165991902834008, |
| "grad_norm": 1.5223466774649232, |
| "learning_rate": 3.788766861595443e-05, |
| "loss": 1.048, |
| "step": 288 |
| }, |
| { |
| "epoch": 1.1700404858299596, |
| "grad_norm": 1.4188985870090396, |
| "learning_rate": 3.7862300323735835e-05, |
| "loss": 1.0621, |
| "step": 289 |
| }, |
| { |
| "epoch": 1.174089068825911, |
| "grad_norm": 1.4445740861138034, |
| "learning_rate": 3.783678920503983e-05, |
| "loss": 1.0467, |
| "step": 290 |
| }, |
| { |
| "epoch": 1.1781376518218623, |
| "grad_norm": 1.1259645347771476, |
| "learning_rate": 3.781113546385265e-05, |
| "loss": 1.0464, |
| "step": 291 |
| }, |
| { |
| "epoch": 1.1821862348178138, |
| "grad_norm": 1.1992580006996914, |
| "learning_rate": 3.7785339305300916e-05, |
| "loss": 1.0381, |
| "step": 292 |
| }, |
| { |
| "epoch": 1.1862348178137652, |
| "grad_norm": 1.1369790679787284, |
| "learning_rate": 3.7759400935650025e-05, |
| "loss": 1.0624, |
| "step": 293 |
| }, |
| { |
| "epoch": 1.1902834008097165, |
| "grad_norm": 1.46105487271457, |
| "learning_rate": 3.77333205623025e-05, |
| "loss": 1.0757, |
| "step": 294 |
| }, |
| { |
| "epoch": 1.194331983805668, |
| "grad_norm": 1.420046414856149, |
| "learning_rate": 3.7707098393796295e-05, |
| "loss": 1.0418, |
| "step": 295 |
| }, |
| { |
| "epoch": 1.1983805668016194, |
| "grad_norm": 1.2156552488935526, |
| "learning_rate": 3.7680734639803185e-05, |
| "loss": 1.0566, |
| "step": 296 |
| }, |
| { |
| "epoch": 1.2024291497975708, |
| "grad_norm": 1.6537303716237448, |
| "learning_rate": 3.765422951112704e-05, |
| "loss": 1.07, |
| "step": 297 |
| }, |
| { |
| "epoch": 1.2064777327935223, |
| "grad_norm": 1.6776313916948862, |
| "learning_rate": 3.762758321970216e-05, |
| "loss": 1.0433, |
| "step": 298 |
| }, |
| { |
| "epoch": 1.2105263157894737, |
| "grad_norm": 1.5076831082648623, |
| "learning_rate": 3.7600795978591584e-05, |
| "loss": 1.0317, |
| "step": 299 |
| }, |
| { |
| "epoch": 1.214574898785425, |
| "grad_norm": 1.132100707628427, |
| "learning_rate": 3.757386800198538e-05, |
| "loss": 1.0115, |
| "step": 300 |
| }, |
| { |
| "epoch": 1.2186234817813766, |
| "grad_norm": 1.4080298205307968, |
| "learning_rate": 3.7546799505198925e-05, |
| "loss": 1.0062, |
| "step": 301 |
| }, |
| { |
| "epoch": 1.222672064777328, |
| "grad_norm": 1.0540431413887559, |
| "learning_rate": 3.7519590704671197e-05, |
| "loss": 1.0348, |
| "step": 302 |
| }, |
| { |
| "epoch": 1.2267206477732793, |
| "grad_norm": 1.1582234391920345, |
| "learning_rate": 3.749224181796305e-05, |
| "loss": 1.0695, |
| "step": 303 |
| }, |
| { |
| "epoch": 1.2307692307692308, |
| "grad_norm": 1.1854390416370666, |
| "learning_rate": 3.746475306375545e-05, |
| "loss": 1.0524, |
| "step": 304 |
| }, |
| { |
| "epoch": 1.2348178137651822, |
| "grad_norm": 1.3117599528011283, |
| "learning_rate": 3.743712466184774e-05, |
| "loss": 1.0655, |
| "step": 305 |
| }, |
| { |
| "epoch": 1.2388663967611335, |
| "grad_norm": 1.1966859727747594, |
| "learning_rate": 3.7409356833155885e-05, |
| "loss": 1.0308, |
| "step": 306 |
| }, |
| { |
| "epoch": 1.242914979757085, |
| "grad_norm": 1.413438329781443, |
| "learning_rate": 3.73814497997107e-05, |
| "loss": 1.0284, |
| "step": 307 |
| }, |
| { |
| "epoch": 1.2469635627530364, |
| "grad_norm": 1.0919324676093507, |
| "learning_rate": 3.7353403784656095e-05, |
| "loss": 1.0434, |
| "step": 308 |
| }, |
| { |
| "epoch": 1.2510121457489878, |
| "grad_norm": 1.8499669428306043, |
| "learning_rate": 3.732521901224724e-05, |
| "loss": 1.0313, |
| "step": 309 |
| }, |
| { |
| "epoch": 1.2550607287449393, |
| "grad_norm": 1.2614587843434526, |
| "learning_rate": 3.7296895707848844e-05, |
| "loss": 1.0477, |
| "step": 310 |
| }, |
| { |
| "epoch": 1.2591093117408907, |
| "grad_norm": 1.7529807074299903, |
| "learning_rate": 3.7268434097933275e-05, |
| "loss": 1.0055, |
| "step": 311 |
| }, |
| { |
| "epoch": 1.263157894736842, |
| "grad_norm": 1.5717631743126275, |
| "learning_rate": 3.723983441007882e-05, |
| "loss": 1.0837, |
| "step": 312 |
| }, |
| { |
| "epoch": 1.2672064777327936, |
| "grad_norm": 1.4087896319211874, |
| "learning_rate": 3.721109687296781e-05, |
| "loss": 1.058, |
| "step": 313 |
| }, |
| { |
| "epoch": 1.271255060728745, |
| "grad_norm": 1.4555895984577636, |
| "learning_rate": 3.718222171638484e-05, |
| "loss": 1.0228, |
| "step": 314 |
| }, |
| { |
| "epoch": 1.2753036437246963, |
| "grad_norm": 1.3355432577453425, |
| "learning_rate": 3.715320917121488e-05, |
| "loss": 1.065, |
| "step": 315 |
| }, |
| { |
| "epoch": 1.2793522267206479, |
| "grad_norm": 1.3723769829491652, |
| "learning_rate": 3.712405946944149e-05, |
| "loss": 1.0422, |
| "step": 316 |
| }, |
| { |
| "epoch": 1.2834008097165992, |
| "grad_norm": 1.2214061038330084, |
| "learning_rate": 3.709477284414488e-05, |
| "loss": 1.0592, |
| "step": 317 |
| }, |
| { |
| "epoch": 1.2874493927125505, |
| "grad_norm": 1.7506104227997383, |
| "learning_rate": 3.7065349529500136e-05, |
| "loss": 1.0344, |
| "step": 318 |
| }, |
| { |
| "epoch": 1.291497975708502, |
| "grad_norm": 1.369867037622663, |
| "learning_rate": 3.7035789760775305e-05, |
| "loss": 1.0405, |
| "step": 319 |
| }, |
| { |
| "epoch": 1.2955465587044535, |
| "grad_norm": 1.652589607171243, |
| "learning_rate": 3.70060937743295e-05, |
| "loss": 1.035, |
| "step": 320 |
| }, |
| { |
| "epoch": 1.2995951417004048, |
| "grad_norm": 1.5573817104845296, |
| "learning_rate": 3.6976261807611045e-05, |
| "loss": 1.0143, |
| "step": 321 |
| }, |
| { |
| "epoch": 1.3036437246963564, |
| "grad_norm": 1.2397229948400104, |
| "learning_rate": 3.694629409915555e-05, |
| "loss": 1.0764, |
| "step": 322 |
| }, |
| { |
| "epoch": 1.3076923076923077, |
| "grad_norm": 1.6825091544529784, |
| "learning_rate": 3.691619088858401e-05, |
| "loss": 1.0548, |
| "step": 323 |
| }, |
| { |
| "epoch": 1.311740890688259, |
| "grad_norm": 1.166427799720964, |
| "learning_rate": 3.68859524166009e-05, |
| "loss": 1.071, |
| "step": 324 |
| }, |
| { |
| "epoch": 1.3157894736842106, |
| "grad_norm": 1.3521381634628526, |
| "learning_rate": 3.6855578924992226e-05, |
| "loss": 1.0238, |
| "step": 325 |
| }, |
| { |
| "epoch": 1.319838056680162, |
| "grad_norm": 1.3736817858016441, |
| "learning_rate": 3.6825070656623626e-05, |
| "loss": 1.0363, |
| "step": 326 |
| }, |
| { |
| "epoch": 1.3238866396761133, |
| "grad_norm": 1.1992156576627135, |
| "learning_rate": 3.67944278554384e-05, |
| "loss": 1.0631, |
| "step": 327 |
| }, |
| { |
| "epoch": 1.3279352226720649, |
| "grad_norm": 1.4130677979950337, |
| "learning_rate": 3.676365076645557e-05, |
| "loss": 0.9692, |
| "step": 328 |
| }, |
| { |
| "epoch": 1.3319838056680162, |
| "grad_norm": 1.1293268459777863, |
| "learning_rate": 3.673273963576791e-05, |
| "loss": 1.0325, |
| "step": 329 |
| }, |
| { |
| "epoch": 1.3360323886639676, |
| "grad_norm": 1.610718239415277, |
| "learning_rate": 3.670169471054001e-05, |
| "loss": 1.0443, |
| "step": 330 |
| }, |
| { |
| "epoch": 1.3400809716599191, |
| "grad_norm": 1.3906237225017748, |
| "learning_rate": 3.6670516239006254e-05, |
| "loss": 1.0398, |
| "step": 331 |
| }, |
| { |
| "epoch": 1.3441295546558705, |
| "grad_norm": 1.0027623838456834, |
| "learning_rate": 3.663920447046888e-05, |
| "loss": 1.0504, |
| "step": 332 |
| }, |
| { |
| "epoch": 1.3481781376518218, |
| "grad_norm": 1.3515344057293308, |
| "learning_rate": 3.6607759655295954e-05, |
| "loss": 1.0309, |
| "step": 333 |
| }, |
| { |
| "epoch": 1.3522267206477734, |
| "grad_norm": 1.0154716482482067, |
| "learning_rate": 3.657618204491937e-05, |
| "loss": 1.0189, |
| "step": 334 |
| }, |
| { |
| "epoch": 1.3562753036437247, |
| "grad_norm": 1.492673783989314, |
| "learning_rate": 3.6544471891832875e-05, |
| "loss": 1.0674, |
| "step": 335 |
| }, |
| { |
| "epoch": 1.360323886639676, |
| "grad_norm": 0.9819656009960984, |
| "learning_rate": 3.6512629449589983e-05, |
| "loss": 1.0485, |
| "step": 336 |
| }, |
| { |
| "epoch": 1.3643724696356276, |
| "grad_norm": 1.4876766276298108, |
| "learning_rate": 3.6480654972802045e-05, |
| "loss": 1.0348, |
| "step": 337 |
| }, |
| { |
| "epoch": 1.368421052631579, |
| "grad_norm": 1.1082999751923241, |
| "learning_rate": 3.644854871713611e-05, |
| "loss": 1.0322, |
| "step": 338 |
| }, |
| { |
| "epoch": 1.3724696356275303, |
| "grad_norm": 1.8428456764744772, |
| "learning_rate": 3.6416310939312924e-05, |
| "loss": 1.0267, |
| "step": 339 |
| }, |
| { |
| "epoch": 1.376518218623482, |
| "grad_norm": 1.549640168857041, |
| "learning_rate": 3.638394189710493e-05, |
| "loss": 1.0335, |
| "step": 340 |
| }, |
| { |
| "epoch": 1.3805668016194332, |
| "grad_norm": 1.3688966386566337, |
| "learning_rate": 3.635144184933412e-05, |
| "loss": 1.0417, |
| "step": 341 |
| }, |
| { |
| "epoch": 1.3846153846153846, |
| "grad_norm": 1.4563179590951778, |
| "learning_rate": 3.6318811055869994e-05, |
| "loss": 1.0833, |
| "step": 342 |
| }, |
| { |
| "epoch": 1.3886639676113361, |
| "grad_norm": 1.412739593774845, |
| "learning_rate": 3.628604977762752e-05, |
| "loss": 1.0588, |
| "step": 343 |
| }, |
| { |
| "epoch": 1.3927125506072875, |
| "grad_norm": 1.4952289865262767, |
| "learning_rate": 3.6253158276565004e-05, |
| "loss": 1.0358, |
| "step": 344 |
| }, |
| { |
| "epoch": 1.3967611336032388, |
| "grad_norm": 1.0782555764991202, |
| "learning_rate": 3.6220136815682016e-05, |
| "loss": 1.0622, |
| "step": 345 |
| }, |
| { |
| "epoch": 1.4008097165991904, |
| "grad_norm": 1.5778945121001913, |
| "learning_rate": 3.618698565901727e-05, |
| "loss": 1.0319, |
| "step": 346 |
| }, |
| { |
| "epoch": 1.4048582995951417, |
| "grad_norm": 1.1490384460452372, |
| "learning_rate": 3.615370507164654e-05, |
| "loss": 1.0635, |
| "step": 347 |
| }, |
| { |
| "epoch": 1.408906882591093, |
| "grad_norm": 1.034155590532865, |
| "learning_rate": 3.61202953196805e-05, |
| "loss": 1.0658, |
| "step": 348 |
| }, |
| { |
| "epoch": 1.4129554655870447, |
| "grad_norm": 1.2115023668080673, |
| "learning_rate": 3.6086756670262654e-05, |
| "loss": 1.0559, |
| "step": 349 |
| }, |
| { |
| "epoch": 1.417004048582996, |
| "grad_norm": 1.6183120660054175, |
| "learning_rate": 3.6053089391567135e-05, |
| "loss": 1.0378, |
| "step": 350 |
| }, |
| { |
| "epoch": 1.4210526315789473, |
| "grad_norm": 1.0212306274192449, |
| "learning_rate": 3.601929375279662e-05, |
| "loss": 1.0625, |
| "step": 351 |
| }, |
| { |
| "epoch": 1.425101214574899, |
| "grad_norm": 1.2217145390662434, |
| "learning_rate": 3.598537002418012e-05, |
| "loss": 1.0726, |
| "step": 352 |
| }, |
| { |
| "epoch": 1.4291497975708503, |
| "grad_norm": 1.0340722139813894, |
| "learning_rate": 3.5951318476970885e-05, |
| "loss": 1.0043, |
| "step": 353 |
| }, |
| { |
| "epoch": 1.4331983805668016, |
| "grad_norm": 1.1883261778029521, |
| "learning_rate": 3.591713938344416e-05, |
| "loss": 1.0588, |
| "step": 354 |
| }, |
| { |
| "epoch": 1.4372469635627532, |
| "grad_norm": 1.38833279647006, |
| "learning_rate": 3.588283301689507e-05, |
| "loss": 1.0342, |
| "step": 355 |
| }, |
| { |
| "epoch": 1.4412955465587045, |
| "grad_norm": 1.6254855577341345, |
| "learning_rate": 3.5848399651636424e-05, |
| "loss": 1.0453, |
| "step": 356 |
| }, |
| { |
| "epoch": 1.4453441295546559, |
| "grad_norm": 1.1245804845945244, |
| "learning_rate": 3.5813839562996473e-05, |
| "loss": 1.0366, |
| "step": 357 |
| }, |
| { |
| "epoch": 1.4493927125506074, |
| "grad_norm": 1.6771734102178915, |
| "learning_rate": 3.5779153027316775e-05, |
| "loss": 1.0437, |
| "step": 358 |
| }, |
| { |
| "epoch": 1.4534412955465588, |
| "grad_norm": 1.0051914017364048, |
| "learning_rate": 3.574434032194994e-05, |
| "loss": 1.0482, |
| "step": 359 |
| }, |
| { |
| "epoch": 1.45748987854251, |
| "grad_norm": 1.7342863024238457, |
| "learning_rate": 3.5709401725257436e-05, |
| "loss": 1.0706, |
| "step": 360 |
| }, |
| { |
| "epoch": 1.4615384615384617, |
| "grad_norm": 1.3430609957243949, |
| "learning_rate": 3.5674337516607346e-05, |
| "loss": 1.0363, |
| "step": 361 |
| }, |
| { |
| "epoch": 1.465587044534413, |
| "grad_norm": 1.2382309862919403, |
| "learning_rate": 3.563914797637216e-05, |
| "loss": 1.0751, |
| "step": 362 |
| }, |
| { |
| "epoch": 1.4696356275303644, |
| "grad_norm": 1.2715168335294111, |
| "learning_rate": 3.560383338592649e-05, |
| "loss": 1.0461, |
| "step": 363 |
| }, |
| { |
| "epoch": 1.4736842105263157, |
| "grad_norm": 1.1620910037088614, |
| "learning_rate": 3.556839402764487e-05, |
| "loss": 1.0063, |
| "step": 364 |
| }, |
| { |
| "epoch": 1.4777327935222673, |
| "grad_norm": 0.8850211487892274, |
| "learning_rate": 3.553283018489946e-05, |
| "loss": 1.0487, |
| "step": 365 |
| }, |
| { |
| "epoch": 1.4817813765182186, |
| "grad_norm": 1.1960278702655094, |
| "learning_rate": 3.54971421420578e-05, |
| "loss": 1.0105, |
| "step": 366 |
| }, |
| { |
| "epoch": 1.48582995951417, |
| "grad_norm": 1.064508834890321, |
| "learning_rate": 3.5461330184480524e-05, |
| "loss": 1.042, |
| "step": 367 |
| }, |
| { |
| "epoch": 1.4898785425101215, |
| "grad_norm": 1.3236796479626098, |
| "learning_rate": 3.54253945985191e-05, |
| "loss": 1.0054, |
| "step": 368 |
| }, |
| { |
| "epoch": 1.4939271255060729, |
| "grad_norm": 1.0951725995782402, |
| "learning_rate": 3.5389335671513524e-05, |
| "loss": 1.0463, |
| "step": 369 |
| }, |
| { |
| "epoch": 1.4979757085020242, |
| "grad_norm": 1.4693570350265857, |
| "learning_rate": 3.5353153691789986e-05, |
| "loss": 1.0319, |
| "step": 370 |
| }, |
| { |
| "epoch": 1.5020242914979756, |
| "grad_norm": 1.3007028266409801, |
| "learning_rate": 3.531684894865866e-05, |
| "loss": 1.0117, |
| "step": 371 |
| }, |
| { |
| "epoch": 1.5060728744939271, |
| "grad_norm": 1.5674737102932312, |
| "learning_rate": 3.528042173241128e-05, |
| "loss": 1.0769, |
| "step": 372 |
| }, |
| { |
| "epoch": 1.5101214574898787, |
| "grad_norm": 1.4341382672312004, |
| "learning_rate": 3.524387233431892e-05, |
| "loss": 1.0582, |
| "step": 373 |
| }, |
| { |
| "epoch": 1.5141700404858298, |
| "grad_norm": 1.4108373375965113, |
| "learning_rate": 3.520720104662958e-05, |
| "loss": 1.0389, |
| "step": 374 |
| }, |
| { |
| "epoch": 1.5182186234817814, |
| "grad_norm": 1.5018868530365561, |
| "learning_rate": 3.5170408162565904e-05, |
| "loss": 1.0377, |
| "step": 375 |
| }, |
| { |
| "epoch": 1.522267206477733, |
| "grad_norm": 0.897345406361951, |
| "learning_rate": 3.513349397632281e-05, |
| "loss": 1.0305, |
| "step": 376 |
| }, |
| { |
| "epoch": 1.526315789473684, |
| "grad_norm": 1.2563424189645032, |
| "learning_rate": 3.5096458783065145e-05, |
| "loss": 1.0627, |
| "step": 377 |
| }, |
| { |
| "epoch": 1.5303643724696356, |
| "grad_norm": 0.9594835847537964, |
| "learning_rate": 3.505930287892533e-05, |
| "loss": 1.0403, |
| "step": 378 |
| }, |
| { |
| "epoch": 1.5344129554655872, |
| "grad_norm": 1.1610152263486018, |
| "learning_rate": 3.502202656100099e-05, |
| "loss": 1.0403, |
| "step": 379 |
| }, |
| { |
| "epoch": 1.5384615384615383, |
| "grad_norm": 1.022298903922081, |
| "learning_rate": 3.498463012735256e-05, |
| "loss": 1.077, |
| "step": 380 |
| }, |
| { |
| "epoch": 1.54251012145749, |
| "grad_norm": 1.5120413458403201, |
| "learning_rate": 3.494711387700094e-05, |
| "loss": 1.0533, |
| "step": 381 |
| }, |
| { |
| "epoch": 1.5465587044534415, |
| "grad_norm": 1.1160750286437722, |
| "learning_rate": 3.490947810992505e-05, |
| "loss": 1.0498, |
| "step": 382 |
| }, |
| { |
| "epoch": 1.5506072874493926, |
| "grad_norm": 1.2230933127366406, |
| "learning_rate": 3.48717231270595e-05, |
| "loss": 1.0556, |
| "step": 383 |
| }, |
| { |
| "epoch": 1.5546558704453441, |
| "grad_norm": 1.3847957067314274, |
| "learning_rate": 3.483384923029211e-05, |
| "loss": 1.0175, |
| "step": 384 |
| }, |
| { |
| "epoch": 1.5587044534412957, |
| "grad_norm": 1.0029599443894253, |
| "learning_rate": 3.479585672246156e-05, |
| "loss": 1.0647, |
| "step": 385 |
| }, |
| { |
| "epoch": 1.5627530364372468, |
| "grad_norm": 1.3280079449803566, |
| "learning_rate": 3.4757745907354924e-05, |
| "loss": 1.0131, |
| "step": 386 |
| }, |
| { |
| "epoch": 1.5668016194331984, |
| "grad_norm": 0.9745265326687522, |
| "learning_rate": 3.471951708970527e-05, |
| "loss": 1.0421, |
| "step": 387 |
| }, |
| { |
| "epoch": 1.5708502024291497, |
| "grad_norm": 1.222452732481316, |
| "learning_rate": 3.468117057518921e-05, |
| "loss": 1.0541, |
| "step": 388 |
| }, |
| { |
| "epoch": 1.574898785425101, |
| "grad_norm": 1.0287600589348802, |
| "learning_rate": 3.4642706670424426e-05, |
| "loss": 1.0115, |
| "step": 389 |
| }, |
| { |
| "epoch": 1.5789473684210527, |
| "grad_norm": 1.2161645202123752, |
| "learning_rate": 3.460412568296731e-05, |
| "loss": 1.019, |
| "step": 390 |
| }, |
| { |
| "epoch": 1.582995951417004, |
| "grad_norm": 0.9690701367815727, |
| "learning_rate": 3.456542792131039e-05, |
| "loss": 1.081, |
| "step": 391 |
| }, |
| { |
| "epoch": 1.5870445344129553, |
| "grad_norm": 1.076638932036548, |
| "learning_rate": 3.4526613694879936e-05, |
| "loss": 1.1014, |
| "step": 392 |
| }, |
| { |
| "epoch": 1.591093117408907, |
| "grad_norm": 1.4045452073362146, |
| "learning_rate": 3.448768331403348e-05, |
| "loss": 1.0675, |
| "step": 393 |
| }, |
| { |
| "epoch": 1.5951417004048583, |
| "grad_norm": 1.1825435666530684, |
| "learning_rate": 3.44486370900573e-05, |
| "loss": 1.0249, |
| "step": 394 |
| }, |
| { |
| "epoch": 1.5991902834008096, |
| "grad_norm": 1.2397310211245995, |
| "learning_rate": 3.440947533516396e-05, |
| "loss": 1.0694, |
| "step": 395 |
| }, |
| { |
| "epoch": 1.6032388663967612, |
| "grad_norm": 1.3422977042516298, |
| "learning_rate": 3.4370198362489816e-05, |
| "loss": 1.0492, |
| "step": 396 |
| }, |
| { |
| "epoch": 1.6072874493927125, |
| "grad_norm": 0.9157882171515165, |
| "learning_rate": 3.43308064860925e-05, |
| "loss": 1.0583, |
| "step": 397 |
| }, |
| { |
| "epoch": 1.6113360323886639, |
| "grad_norm": 1.1715632536707488, |
| "learning_rate": 3.4291300020948393e-05, |
| "loss": 1.032, |
| "step": 398 |
| }, |
| { |
| "epoch": 1.6153846153846154, |
| "grad_norm": 0.7494632041988553, |
| "learning_rate": 3.425167928295015e-05, |
| "loss": 1.0222, |
| "step": 399 |
| }, |
| { |
| "epoch": 1.6194331983805668, |
| "grad_norm": 1.2752286275608182, |
| "learning_rate": 3.421194458890411e-05, |
| "loss": 1.0611, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.623481781376518, |
| "grad_norm": 0.9269971938859848, |
| "learning_rate": 3.4172096256527865e-05, |
| "loss": 1.0266, |
| "step": 401 |
| }, |
| { |
| "epoch": 1.6275303643724697, |
| "grad_norm": 1.3385895543232296, |
| "learning_rate": 3.4132134604447595e-05, |
| "loss": 1.0601, |
| "step": 402 |
| }, |
| { |
| "epoch": 1.631578947368421, |
| "grad_norm": 1.1797236030301175, |
| "learning_rate": 3.409205995219561e-05, |
| "loss": 1.0273, |
| "step": 403 |
| }, |
| { |
| "epoch": 1.6356275303643724, |
| "grad_norm": 1.348300643593844, |
| "learning_rate": 3.4051872620207765e-05, |
| "loss": 1.0499, |
| "step": 404 |
| }, |
| { |
| "epoch": 1.639676113360324, |
| "grad_norm": 1.163609662735727, |
| "learning_rate": 3.40115729298209e-05, |
| "loss": 1.0574, |
| "step": 405 |
| }, |
| { |
| "epoch": 1.6437246963562753, |
| "grad_norm": 1.5789245382686026, |
| "learning_rate": 3.397116120327025e-05, |
| "loss": 1.083, |
| "step": 406 |
| }, |
| { |
| "epoch": 1.6477732793522266, |
| "grad_norm": 1.166228013652488, |
| "learning_rate": 3.3930637763686916e-05, |
| "loss": 1.0878, |
| "step": 407 |
| }, |
| { |
| "epoch": 1.6518218623481782, |
| "grad_norm": 1.5851378236682954, |
| "learning_rate": 3.389000293509524e-05, |
| "loss": 1.0376, |
| "step": 408 |
| }, |
| { |
| "epoch": 1.6558704453441295, |
| "grad_norm": 0.8972321191027742, |
| "learning_rate": 3.3849257042410225e-05, |
| "loss": 1.0422, |
| "step": 409 |
| }, |
| { |
| "epoch": 1.6599190283400809, |
| "grad_norm": 1.2678364291487247, |
| "learning_rate": 3.380840041143494e-05, |
| "loss": 1.0467, |
| "step": 410 |
| }, |
| { |
| "epoch": 1.6639676113360324, |
| "grad_norm": 1.1321829107376757, |
| "learning_rate": 3.376743336885792e-05, |
| "loss": 1.0607, |
| "step": 411 |
| }, |
| { |
| "epoch": 1.6680161943319838, |
| "grad_norm": 1.276153184293288, |
| "learning_rate": 3.372635624225055e-05, |
| "loss": 1.0527, |
| "step": 412 |
| }, |
| { |
| "epoch": 1.6720647773279351, |
| "grad_norm": 1.107819823127467, |
| "learning_rate": 3.3685169360064445e-05, |
| "loss": 1.0448, |
| "step": 413 |
| }, |
| { |
| "epoch": 1.6761133603238867, |
| "grad_norm": 1.032466049431014, |
| "learning_rate": 3.3643873051628804e-05, |
| "loss": 1.0609, |
| "step": 414 |
| }, |
| { |
| "epoch": 1.680161943319838, |
| "grad_norm": 1.8119559549150255, |
| "learning_rate": 3.3602467647147814e-05, |
| "loss": 1.0305, |
| "step": 415 |
| }, |
| { |
| "epoch": 1.6842105263157894, |
| "grad_norm": 1.2111660504454058, |
| "learning_rate": 3.356095347769799e-05, |
| "loss": 1.0125, |
| "step": 416 |
| }, |
| { |
| "epoch": 1.688259109311741, |
| "grad_norm": 2.0935621558416284, |
| "learning_rate": 3.351933087522552e-05, |
| "loss": 1.0557, |
| "step": 417 |
| }, |
| { |
| "epoch": 1.6923076923076923, |
| "grad_norm": 1.7523759637721479, |
| "learning_rate": 3.3477600172543626e-05, |
| "loss": 1.0495, |
| "step": 418 |
| }, |
| { |
| "epoch": 1.6963562753036436, |
| "grad_norm": 1.9115740325874013, |
| "learning_rate": 3.3435761703329894e-05, |
| "loss": 1.0265, |
| "step": 419 |
| }, |
| { |
| "epoch": 1.7004048582995952, |
| "grad_norm": 1.571643366433319, |
| "learning_rate": 3.3393815802123606e-05, |
| "loss": 1.0731, |
| "step": 420 |
| }, |
| { |
| "epoch": 1.7044534412955465, |
| "grad_norm": 1.7623283834251537, |
| "learning_rate": 3.335176280432308e-05, |
| "loss": 1.0461, |
| "step": 421 |
| }, |
| { |
| "epoch": 1.708502024291498, |
| "grad_norm": 1.5638855452043183, |
| "learning_rate": 3.330960304618295e-05, |
| "loss": 1.0416, |
| "step": 422 |
| }, |
| { |
| "epoch": 1.7125506072874495, |
| "grad_norm": 2.0523140666938344, |
| "learning_rate": 3.326733686481153e-05, |
| "loss": 1.0127, |
| "step": 423 |
| }, |
| { |
| "epoch": 1.7165991902834008, |
| "grad_norm": 1.4740431693747993, |
| "learning_rate": 3.3224964598168084e-05, |
| "loss": 1.0443, |
| "step": 424 |
| }, |
| { |
| "epoch": 1.7206477732793521, |
| "grad_norm": 1.835077522883751, |
| "learning_rate": 3.318248658506012e-05, |
| "loss": 1.0342, |
| "step": 425 |
| }, |
| { |
| "epoch": 1.7246963562753037, |
| "grad_norm": 1.2160700980753112, |
| "learning_rate": 3.313990316514069e-05, |
| "loss": 1.0731, |
| "step": 426 |
| }, |
| { |
| "epoch": 1.728744939271255, |
| "grad_norm": 2.0726717428379047, |
| "learning_rate": 3.309721467890571e-05, |
| "loss": 1.0323, |
| "step": 427 |
| }, |
| { |
| "epoch": 1.7327935222672064, |
| "grad_norm": 1.8350309759569734, |
| "learning_rate": 3.305442146769115e-05, |
| "loss": 1.0417, |
| "step": 428 |
| }, |
| { |
| "epoch": 1.736842105263158, |
| "grad_norm": 1.8324872807012549, |
| "learning_rate": 3.30115238736704e-05, |
| "loss": 1.051, |
| "step": 429 |
| }, |
| { |
| "epoch": 1.7408906882591093, |
| "grad_norm": 1.726955825370765, |
| "learning_rate": 3.296852223985148e-05, |
| "loss": 1.0457, |
| "step": 430 |
| }, |
| { |
| "epoch": 1.7449392712550607, |
| "grad_norm": 1.7465299788974538, |
| "learning_rate": 3.2925416910074315e-05, |
| "loss": 1.0491, |
| "step": 431 |
| }, |
| { |
| "epoch": 1.7489878542510122, |
| "grad_norm": 1.3401303811979284, |
| "learning_rate": 3.288220822900796e-05, |
| "loss": 1.0532, |
| "step": 432 |
| }, |
| { |
| "epoch": 1.7530364372469636, |
| "grad_norm": 2.0902629729757396, |
| "learning_rate": 3.283889654214788e-05, |
| "loss": 1.0461, |
| "step": 433 |
| }, |
| { |
| "epoch": 1.757085020242915, |
| "grad_norm": 1.284639616669281, |
| "learning_rate": 3.279548219581317e-05, |
| "loss": 1.0325, |
| "step": 434 |
| }, |
| { |
| "epoch": 1.7611336032388665, |
| "grad_norm": 2.440702829122251, |
| "learning_rate": 3.275196553714379e-05, |
| "loss": 1.05, |
| "step": 435 |
| }, |
| { |
| "epoch": 1.7651821862348178, |
| "grad_norm": 2.2053355528497343, |
| "learning_rate": 3.270834691409778e-05, |
| "loss": 1.0243, |
| "step": 436 |
| }, |
| { |
| "epoch": 1.7692307692307692, |
| "grad_norm": 1.8434515953244732, |
| "learning_rate": 3.266462667544849e-05, |
| "loss": 1.0237, |
| "step": 437 |
| }, |
| { |
| "epoch": 1.7732793522267207, |
| "grad_norm": 1.8614329548688082, |
| "learning_rate": 3.262080517078178e-05, |
| "loss": 1.0516, |
| "step": 438 |
| }, |
| { |
| "epoch": 1.777327935222672, |
| "grad_norm": 1.7640855246155347, |
| "learning_rate": 3.257688275049323e-05, |
| "loss": 1.0706, |
| "step": 439 |
| }, |
| { |
| "epoch": 1.7813765182186234, |
| "grad_norm": 1.7808330154138954, |
| "learning_rate": 3.253285976578535e-05, |
| "loss": 1.0368, |
| "step": 440 |
| }, |
| { |
| "epoch": 1.785425101214575, |
| "grad_norm": 1.6345332495391471, |
| "learning_rate": 3.2488736568664756e-05, |
| "loss": 1.0117, |
| "step": 441 |
| }, |
| { |
| "epoch": 1.7894736842105263, |
| "grad_norm": 1.595802381323778, |
| "learning_rate": 3.244451351193935e-05, |
| "loss": 1.0537, |
| "step": 442 |
| }, |
| { |
| "epoch": 1.7935222672064777, |
| "grad_norm": 1.522974314741888, |
| "learning_rate": 3.240019094921551e-05, |
| "loss": 1.0394, |
| "step": 443 |
| }, |
| { |
| "epoch": 1.7975708502024292, |
| "grad_norm": 1.3768607596002047, |
| "learning_rate": 3.235576923489527e-05, |
| "loss": 1.0339, |
| "step": 444 |
| }, |
| { |
| "epoch": 1.8016194331983806, |
| "grad_norm": 1.6852439341104335, |
| "learning_rate": 3.2311248724173476e-05, |
| "loss": 1.0446, |
| "step": 445 |
| }, |
| { |
| "epoch": 1.805668016194332, |
| "grad_norm": 1.1261743084871483, |
| "learning_rate": 3.226662977303494e-05, |
| "loss": 1.0526, |
| "step": 446 |
| }, |
| { |
| "epoch": 1.8097165991902835, |
| "grad_norm": 2.2068408384010256, |
| "learning_rate": 3.2221912738251626e-05, |
| "loss": 1.0497, |
| "step": 447 |
| }, |
| { |
| "epoch": 1.8137651821862348, |
| "grad_norm": 2.056518485626637, |
| "learning_rate": 3.2177097977379736e-05, |
| "loss": 1.0383, |
| "step": 448 |
| }, |
| { |
| "epoch": 1.8178137651821862, |
| "grad_norm": 1.7497817365575585, |
| "learning_rate": 3.21321858487569e-05, |
| "loss": 0.995, |
| "step": 449 |
| }, |
| { |
| "epoch": 1.8218623481781377, |
| "grad_norm": 1.8571777443295205, |
| "learning_rate": 3.208717671149931e-05, |
| "loss": 1.0014, |
| "step": 450 |
| }, |
| { |
| "epoch": 1.825910931174089, |
| "grad_norm": 1.5695678256909782, |
| "learning_rate": 3.204207092549882e-05, |
| "loss": 1.04, |
| "step": 451 |
| }, |
| { |
| "epoch": 1.8299595141700404, |
| "grad_norm": 1.219882310596643, |
| "learning_rate": 3.19968688514201e-05, |
| "loss": 1.0419, |
| "step": 452 |
| }, |
| { |
| "epoch": 1.834008097165992, |
| "grad_norm": 2.1212764815455234, |
| "learning_rate": 3.195157085069774e-05, |
| "loss": 1.0359, |
| "step": 453 |
| }, |
| { |
| "epoch": 1.8380566801619433, |
| "grad_norm": 1.610659705926026, |
| "learning_rate": 3.190617728553332e-05, |
| "loss": 1.0375, |
| "step": 454 |
| }, |
| { |
| "epoch": 1.8421052631578947, |
| "grad_norm": 2.0185680956941447, |
| "learning_rate": 3.1860688518892606e-05, |
| "loss": 1.0303, |
| "step": 455 |
| }, |
| { |
| "epoch": 1.8461538461538463, |
| "grad_norm": 1.8831567692066349, |
| "learning_rate": 3.181510491450254e-05, |
| "loss": 0.9882, |
| "step": 456 |
| }, |
| { |
| "epoch": 1.8502024291497976, |
| "grad_norm": 1.3200541676666329, |
| "learning_rate": 3.176942683684842e-05, |
| "loss": 1.0413, |
| "step": 457 |
| }, |
| { |
| "epoch": 1.854251012145749, |
| "grad_norm": 1.1656763563425627, |
| "learning_rate": 3.1723654651170934e-05, |
| "loss": 1.0487, |
| "step": 458 |
| }, |
| { |
| "epoch": 1.8582995951417005, |
| "grad_norm": 2.0104919503530163, |
| "learning_rate": 3.167778872346327e-05, |
| "loss": 1.0194, |
| "step": 459 |
| }, |
| { |
| "epoch": 1.8623481781376519, |
| "grad_norm": 1.5161348026477568, |
| "learning_rate": 3.163182942046815e-05, |
| "loss": 1.0508, |
| "step": 460 |
| }, |
| { |
| "epoch": 1.8663967611336032, |
| "grad_norm": 2.0094333935591835, |
| "learning_rate": 3.158577710967494e-05, |
| "loss": 1.0442, |
| "step": 461 |
| }, |
| { |
| "epoch": 1.8704453441295548, |
| "grad_norm": 1.9716602227838338, |
| "learning_rate": 3.153963215931669e-05, |
| "loss": 1.0482, |
| "step": 462 |
| }, |
| { |
| "epoch": 1.874493927125506, |
| "grad_norm": 1.3689582415529447, |
| "learning_rate": 3.14933949383672e-05, |
| "loss": 0.9984, |
| "step": 463 |
| }, |
| { |
| "epoch": 1.8785425101214575, |
| "grad_norm": 1.7188799961915957, |
| "learning_rate": 3.144706581653805e-05, |
| "loss": 1.011, |
| "step": 464 |
| }, |
| { |
| "epoch": 1.882591093117409, |
| "grad_norm": 1.2277911571492617, |
| "learning_rate": 3.140064516427566e-05, |
| "loss": 1.0335, |
| "step": 465 |
| }, |
| { |
| "epoch": 1.8866396761133604, |
| "grad_norm": 1.23408384570321, |
| "learning_rate": 3.135413335275833e-05, |
| "loss": 1.0307, |
| "step": 466 |
| }, |
| { |
| "epoch": 1.8906882591093117, |
| "grad_norm": 1.4234912417449508, |
| "learning_rate": 3.130753075389327e-05, |
| "loss": 1.0274, |
| "step": 467 |
| }, |
| { |
| "epoch": 1.8947368421052633, |
| "grad_norm": 1.0005982034837426, |
| "learning_rate": 3.1260837740313625e-05, |
| "loss": 1.0406, |
| "step": 468 |
| }, |
| { |
| "epoch": 1.8987854251012146, |
| "grad_norm": 1.7625501066691556, |
| "learning_rate": 3.121405468537547e-05, |
| "loss": 1.0486, |
| "step": 469 |
| }, |
| { |
| "epoch": 1.902834008097166, |
| "grad_norm": 1.1517060372357544, |
| "learning_rate": 3.116718196315488e-05, |
| "loss": 1.0311, |
| "step": 470 |
| }, |
| { |
| "epoch": 1.9068825910931175, |
| "grad_norm": 2.0869969088274183, |
| "learning_rate": 3.1120219948444894e-05, |
| "loss": 1.03, |
| "step": 471 |
| }, |
| { |
| "epoch": 1.9109311740890689, |
| "grad_norm": 1.880114733020443, |
| "learning_rate": 3.107316901675252e-05, |
| "loss": 1.0522, |
| "step": 472 |
| }, |
| { |
| "epoch": 1.9149797570850202, |
| "grad_norm": 1.592800978532512, |
| "learning_rate": 3.1026029544295745e-05, |
| "loss": 1.0245, |
| "step": 473 |
| }, |
| { |
| "epoch": 1.9190283400809718, |
| "grad_norm": 1.798415066510731, |
| "learning_rate": 3.097880190800057e-05, |
| "loss": 1.0375, |
| "step": 474 |
| }, |
| { |
| "epoch": 1.9230769230769231, |
| "grad_norm": 1.1578004043901418, |
| "learning_rate": 3.093148648549788e-05, |
| "loss": 1.0131, |
| "step": 475 |
| }, |
| { |
| "epoch": 1.9271255060728745, |
| "grad_norm": 1.1825055441666752, |
| "learning_rate": 3.088408365512055e-05, |
| "loss": 1.0344, |
| "step": 476 |
| }, |
| { |
| "epoch": 1.931174089068826, |
| "grad_norm": 1.330999172028336, |
| "learning_rate": 3.083659379590034e-05, |
| "loss": 1.0448, |
| "step": 477 |
| }, |
| { |
| "epoch": 1.9352226720647774, |
| "grad_norm": 0.8237075149060147, |
| "learning_rate": 3.0789017287564913e-05, |
| "loss": 1.0504, |
| "step": 478 |
| }, |
| { |
| "epoch": 1.9392712550607287, |
| "grad_norm": 1.0636163412030821, |
| "learning_rate": 3.074135451053475e-05, |
| "loss": 1.0495, |
| "step": 479 |
| }, |
| { |
| "epoch": 1.9433198380566803, |
| "grad_norm": 1.1564161279298837, |
| "learning_rate": 3.069360584592015e-05, |
| "loss": 1.026, |
| "step": 480 |
| }, |
| { |
| "epoch": 1.9473684210526314, |
| "grad_norm": 1.0045967975426582, |
| "learning_rate": 3.064577167551816e-05, |
| "loss": 1.0336, |
| "step": 481 |
| }, |
| { |
| "epoch": 1.951417004048583, |
| "grad_norm": 1.0801416670989286, |
| "learning_rate": 3.059785238180953e-05, |
| "loss": 1.0232, |
| "step": 482 |
| }, |
| { |
| "epoch": 1.9554655870445345, |
| "grad_norm": 1.1977403406411171, |
| "learning_rate": 3.054984834795566e-05, |
| "loss": 1.0444, |
| "step": 483 |
| }, |
| { |
| "epoch": 1.9595141700404857, |
| "grad_norm": 0.8475420612903407, |
| "learning_rate": 3.0501759957795537e-05, |
| "loss": 1.0154, |
| "step": 484 |
| }, |
| { |
| "epoch": 1.9635627530364372, |
| "grad_norm": 1.4294733442626821, |
| "learning_rate": 3.0453587595842627e-05, |
| "loss": 1.03, |
| "step": 485 |
| }, |
| { |
| "epoch": 1.9676113360323888, |
| "grad_norm": 0.9564518192921299, |
| "learning_rate": 3.0405331647281856e-05, |
| "loss": 1.0581, |
| "step": 486 |
| }, |
| { |
| "epoch": 1.97165991902834, |
| "grad_norm": 1.6737718924257705, |
| "learning_rate": 3.0356992497966508e-05, |
| "loss": 1.0433, |
| "step": 487 |
| }, |
| { |
| "epoch": 1.9757085020242915, |
| "grad_norm": 1.2625901564329196, |
| "learning_rate": 3.0308570534415137e-05, |
| "loss": 1.0506, |
| "step": 488 |
| }, |
| { |
| "epoch": 1.979757085020243, |
| "grad_norm": 1.268557679679458, |
| "learning_rate": 3.0260066143808467e-05, |
| "loss": 1.0632, |
| "step": 489 |
| }, |
| { |
| "epoch": 1.9838056680161942, |
| "grad_norm": 1.4022668987340432, |
| "learning_rate": 3.021147971398632e-05, |
| "loss": 1.0281, |
| "step": 490 |
| }, |
| { |
| "epoch": 1.9878542510121457, |
| "grad_norm": 1.081315118494276, |
| "learning_rate": 3.0162811633444478e-05, |
| "loss": 1.0161, |
| "step": 491 |
| }, |
| { |
| "epoch": 1.9919028340080973, |
| "grad_norm": 1.7076777531346692, |
| "learning_rate": 3.0114062291331635e-05, |
| "loss": 1.0413, |
| "step": 492 |
| }, |
| { |
| "epoch": 1.9959514170040484, |
| "grad_norm": 1.2300374937392595, |
| "learning_rate": 3.0065232077446217e-05, |
| "loss": 1.0324, |
| "step": 493 |
| }, |
| { |
| "epoch": 2.0, |
| "grad_norm": 1.6562517282934297, |
| "learning_rate": 3.001632138223332e-05, |
| "loss": 1.0072, |
| "step": 494 |
| }, |
| { |
| "epoch": 2.0040485829959516, |
| "grad_norm": 1.7375387311013795, |
| "learning_rate": 2.9967330596781565e-05, |
| "loss": 0.9197, |
| "step": 495 |
| }, |
| { |
| "epoch": 2.0080971659919027, |
| "grad_norm": 1.2639066309221982, |
| "learning_rate": 2.9918260112819952e-05, |
| "loss": 0.8578, |
| "step": 496 |
| }, |
| { |
| "epoch": 2.0121457489878543, |
| "grad_norm": 1.6018770495579906, |
| "learning_rate": 2.9869110322714778e-05, |
| "loss": 0.8862, |
| "step": 497 |
| }, |
| { |
| "epoch": 2.016194331983806, |
| "grad_norm": 1.4045677292284346, |
| "learning_rate": 2.9819881619466447e-05, |
| "loss": 0.9106, |
| "step": 498 |
| }, |
| { |
| "epoch": 2.020242914979757, |
| "grad_norm": 1.2933042279056441, |
| "learning_rate": 2.9770574396706364e-05, |
| "loss": 0.9, |
| "step": 499 |
| }, |
| { |
| "epoch": 2.0242914979757085, |
| "grad_norm": 1.1856312250151082, |
| "learning_rate": 2.972118904869376e-05, |
| "loss": 0.8807, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.02834008097166, |
| "grad_norm": 1.2224271633863928, |
| "learning_rate": 2.9671725970312563e-05, |
| "loss": 0.9214, |
| "step": 501 |
| }, |
| { |
| "epoch": 2.032388663967611, |
| "grad_norm": 1.0921895244598434, |
| "learning_rate": 2.9622185557068238e-05, |
| "loss": 0.8915, |
| "step": 502 |
| }, |
| { |
| "epoch": 2.0364372469635628, |
| "grad_norm": 1.2951827664968256, |
| "learning_rate": 2.95725682050846e-05, |
| "loss": 0.8831, |
| "step": 503 |
| }, |
| { |
| "epoch": 2.0404858299595143, |
| "grad_norm": 1.1471446400737364, |
| "learning_rate": 2.9522874311100684e-05, |
| "loss": 0.9045, |
| "step": 504 |
| }, |
| { |
| "epoch": 2.0445344129554655, |
| "grad_norm": 1.317005501875735, |
| "learning_rate": 2.9473104272467526e-05, |
| "loss": 0.9104, |
| "step": 505 |
| }, |
| { |
| "epoch": 2.048582995951417, |
| "grad_norm": 1.080668167495605, |
| "learning_rate": 2.942325848714505e-05, |
| "loss": 0.8878, |
| "step": 506 |
| }, |
| { |
| "epoch": 2.0526315789473686, |
| "grad_norm": 1.1291979109997605, |
| "learning_rate": 2.9373337353698804e-05, |
| "loss": 0.8565, |
| "step": 507 |
| }, |
| { |
| "epoch": 2.0566801619433197, |
| "grad_norm": 0.9395474269536062, |
| "learning_rate": 2.932334127129686e-05, |
| "loss": 0.9061, |
| "step": 508 |
| }, |
| { |
| "epoch": 2.0607287449392713, |
| "grad_norm": 1.1671225820143165, |
| "learning_rate": 2.9273270639706547e-05, |
| "loss": 0.8864, |
| "step": 509 |
| }, |
| { |
| "epoch": 2.064777327935223, |
| "grad_norm": 0.8660782827969822, |
| "learning_rate": 2.922312585929131e-05, |
| "loss": 0.8746, |
| "step": 510 |
| }, |
| { |
| "epoch": 2.068825910931174, |
| "grad_norm": 1.1432952564484031, |
| "learning_rate": 2.9172907331007478e-05, |
| "loss": 0.8901, |
| "step": 511 |
| }, |
| { |
| "epoch": 2.0728744939271255, |
| "grad_norm": 0.8618293366557402, |
| "learning_rate": 2.9122615456401053e-05, |
| "loss": 0.9235, |
| "step": 512 |
| }, |
| { |
| "epoch": 2.076923076923077, |
| "grad_norm": 0.9039752283495845, |
| "learning_rate": 2.9072250637604535e-05, |
| "loss": 0.9065, |
| "step": 513 |
| }, |
| { |
| "epoch": 2.080971659919028, |
| "grad_norm": 1.0593211824027946, |
| "learning_rate": 2.9021813277333667e-05, |
| "loss": 0.9037, |
| "step": 514 |
| }, |
| { |
| "epoch": 2.08502024291498, |
| "grad_norm": 1.0044572703655736, |
| "learning_rate": 2.8971303778884244e-05, |
| "loss": 0.8988, |
| "step": 515 |
| }, |
| { |
| "epoch": 2.0890688259109313, |
| "grad_norm": 1.0470821332358036, |
| "learning_rate": 2.8920722546128863e-05, |
| "loss": 0.8929, |
| "step": 516 |
| }, |
| { |
| "epoch": 2.0931174089068825, |
| "grad_norm": 0.7749465556960055, |
| "learning_rate": 2.887006998351372e-05, |
| "loss": 0.9147, |
| "step": 517 |
| }, |
| { |
| "epoch": 2.097165991902834, |
| "grad_norm": 0.9712007004656309, |
| "learning_rate": 2.881934649605535e-05, |
| "loss": 0.8848, |
| "step": 518 |
| }, |
| { |
| "epoch": 2.1012145748987856, |
| "grad_norm": 0.9549059358853991, |
| "learning_rate": 2.8768552489337408e-05, |
| "loss": 0.8812, |
| "step": 519 |
| }, |
| { |
| "epoch": 2.1052631578947367, |
| "grad_norm": 0.8804218416983838, |
| "learning_rate": 2.8717688369507425e-05, |
| "loss": 0.8795, |
| "step": 520 |
| }, |
| { |
| "epoch": 2.1093117408906883, |
| "grad_norm": 0.9282549283358114, |
| "learning_rate": 2.8666754543273535e-05, |
| "loss": 0.889, |
| "step": 521 |
| }, |
| { |
| "epoch": 2.11336032388664, |
| "grad_norm": 1.0189180356298997, |
| "learning_rate": 2.8615751417901273e-05, |
| "loss": 0.8722, |
| "step": 522 |
| }, |
| { |
| "epoch": 2.117408906882591, |
| "grad_norm": 0.8368886145764572, |
| "learning_rate": 2.856467940121025e-05, |
| "loss": 0.8951, |
| "step": 523 |
| }, |
| { |
| "epoch": 2.1214574898785425, |
| "grad_norm": 1.1588207443885319, |
| "learning_rate": 2.8513538901570964e-05, |
| "loss": 0.8783, |
| "step": 524 |
| }, |
| { |
| "epoch": 2.125506072874494, |
| "grad_norm": 0.911472432868388, |
| "learning_rate": 2.846233032790149e-05, |
| "loss": 0.8585, |
| "step": 525 |
| }, |
| { |
| "epoch": 2.1295546558704452, |
| "grad_norm": 1.0965950785541232, |
| "learning_rate": 2.84110540896642e-05, |
| "loss": 0.8606, |
| "step": 526 |
| }, |
| { |
| "epoch": 2.133603238866397, |
| "grad_norm": 1.0080365595475678, |
| "learning_rate": 2.8359710596862555e-05, |
| "loss": 0.8721, |
| "step": 527 |
| }, |
| { |
| "epoch": 2.1376518218623484, |
| "grad_norm": 0.9320596100893715, |
| "learning_rate": 2.8308300260037734e-05, |
| "loss": 0.9079, |
| "step": 528 |
| }, |
| { |
| "epoch": 2.1417004048582995, |
| "grad_norm": 1.1528215454118658, |
| "learning_rate": 2.8256823490265424e-05, |
| "loss": 0.8999, |
| "step": 529 |
| }, |
| { |
| "epoch": 2.145748987854251, |
| "grad_norm": 0.7389914458877977, |
| "learning_rate": 2.8205280699152515e-05, |
| "loss": 0.8718, |
| "step": 530 |
| }, |
| { |
| "epoch": 2.1497975708502026, |
| "grad_norm": 0.957974084214064, |
| "learning_rate": 2.8153672298833777e-05, |
| "loss": 0.9027, |
| "step": 531 |
| }, |
| { |
| "epoch": 2.1538461538461537, |
| "grad_norm": 0.7943122676550494, |
| "learning_rate": 2.8101998701968615e-05, |
| "loss": 0.882, |
| "step": 532 |
| }, |
| { |
| "epoch": 2.1578947368421053, |
| "grad_norm": 0.8415229748895574, |
| "learning_rate": 2.805026032173773e-05, |
| "loss": 0.8899, |
| "step": 533 |
| }, |
| { |
| "epoch": 2.161943319838057, |
| "grad_norm": 0.7744452513665163, |
| "learning_rate": 2.7998457571839832e-05, |
| "loss": 0.8795, |
| "step": 534 |
| }, |
| { |
| "epoch": 2.165991902834008, |
| "grad_norm": 1.0105150711886886, |
| "learning_rate": 2.7946590866488322e-05, |
| "loss": 0.9148, |
| "step": 535 |
| }, |
| { |
| "epoch": 2.1700404858299596, |
| "grad_norm": 1.0755919805796508, |
| "learning_rate": 2.7894660620408e-05, |
| "loss": 0.9063, |
| "step": 536 |
| }, |
| { |
| "epoch": 2.174089068825911, |
| "grad_norm": 0.6795480756640717, |
| "learning_rate": 2.784266724883173e-05, |
| "loss": 0.8573, |
| "step": 537 |
| }, |
| { |
| "epoch": 2.1781376518218623, |
| "grad_norm": 0.8425463265366985, |
| "learning_rate": 2.7790611167497128e-05, |
| "loss": 0.9237, |
| "step": 538 |
| }, |
| { |
| "epoch": 2.182186234817814, |
| "grad_norm": 0.930992973519507, |
| "learning_rate": 2.7738492792643224e-05, |
| "loss": 0.9033, |
| "step": 539 |
| }, |
| { |
| "epoch": 2.1862348178137654, |
| "grad_norm": 0.7070754471615176, |
| "learning_rate": 2.768631254100716e-05, |
| "loss": 0.8627, |
| "step": 540 |
| }, |
| { |
| "epoch": 2.1902834008097165, |
| "grad_norm": 0.8221581148307489, |
| "learning_rate": 2.7634070829820836e-05, |
| "loss": 0.9088, |
| "step": 541 |
| }, |
| { |
| "epoch": 2.194331983805668, |
| "grad_norm": 0.7840565822440921, |
| "learning_rate": 2.7581768076807592e-05, |
| "loss": 0.9053, |
| "step": 542 |
| }, |
| { |
| "epoch": 2.1983805668016196, |
| "grad_norm": 0.723771017810014, |
| "learning_rate": 2.7529404700178832e-05, |
| "loss": 0.8607, |
| "step": 543 |
| }, |
| { |
| "epoch": 2.2024291497975708, |
| "grad_norm": 0.9216996590279596, |
| "learning_rate": 2.7476981118630735e-05, |
| "loss": 0.8858, |
| "step": 544 |
| }, |
| { |
| "epoch": 2.2064777327935223, |
| "grad_norm": 0.86090241416406, |
| "learning_rate": 2.742449775134085e-05, |
| "loss": 0.92, |
| "step": 545 |
| }, |
| { |
| "epoch": 2.2105263157894735, |
| "grad_norm": 0.8612283960117196, |
| "learning_rate": 2.7371955017964788e-05, |
| "loss": 0.8976, |
| "step": 546 |
| }, |
| { |
| "epoch": 2.214574898785425, |
| "grad_norm": 0.7005016869582557, |
| "learning_rate": 2.7319353338632845e-05, |
| "loss": 0.877, |
| "step": 547 |
| }, |
| { |
| "epoch": 2.2186234817813766, |
| "grad_norm": 0.9994097940058263, |
| "learning_rate": 2.7266693133946645e-05, |
| "loss": 0.9166, |
| "step": 548 |
| }, |
| { |
| "epoch": 2.2226720647773277, |
| "grad_norm": 0.7381007768146903, |
| "learning_rate": 2.7213974824975785e-05, |
| "loss": 0.886, |
| "step": 549 |
| }, |
| { |
| "epoch": 2.2267206477732793, |
| "grad_norm": 0.9476586752108841, |
| "learning_rate": 2.7161198833254464e-05, |
| "loss": 0.9337, |
| "step": 550 |
| }, |
| { |
| "epoch": 2.230769230769231, |
| "grad_norm": 0.8389117278323334, |
| "learning_rate": 2.7108365580778095e-05, |
| "loss": 0.8722, |
| "step": 551 |
| }, |
| { |
| "epoch": 2.234817813765182, |
| "grad_norm": 0.8728053429625956, |
| "learning_rate": 2.705547548999996e-05, |
| "loss": 0.8948, |
| "step": 552 |
| }, |
| { |
| "epoch": 2.2388663967611335, |
| "grad_norm": 1.0116868121728824, |
| "learning_rate": 2.7002528983827817e-05, |
| "loss": 0.9117, |
| "step": 553 |
| }, |
| { |
| "epoch": 2.242914979757085, |
| "grad_norm": 0.7567286615749226, |
| "learning_rate": 2.6949526485620517e-05, |
| "loss": 0.9064, |
| "step": 554 |
| }, |
| { |
| "epoch": 2.246963562753036, |
| "grad_norm": 0.8187157967638948, |
| "learning_rate": 2.6896468419184623e-05, |
| "loss": 0.937, |
| "step": 555 |
| }, |
| { |
| "epoch": 2.251012145748988, |
| "grad_norm": 0.9626593463389618, |
| "learning_rate": 2.6843355208771024e-05, |
| "loss": 0.8947, |
| "step": 556 |
| }, |
| { |
| "epoch": 2.2550607287449393, |
| "grad_norm": 0.9636305147678053, |
| "learning_rate": 2.6790187279071536e-05, |
| "loss": 0.9247, |
| "step": 557 |
| }, |
| { |
| "epoch": 2.2591093117408905, |
| "grad_norm": 0.748879429186165, |
| "learning_rate": 2.67369650552155e-05, |
| "loss": 0.8747, |
| "step": 558 |
| }, |
| { |
| "epoch": 2.263157894736842, |
| "grad_norm": 0.9266109923107715, |
| "learning_rate": 2.6683688962766407e-05, |
| "loss": 0.8735, |
| "step": 559 |
| }, |
| { |
| "epoch": 2.2672064777327936, |
| "grad_norm": 0.7191990739718408, |
| "learning_rate": 2.6630359427718468e-05, |
| "loss": 0.8884, |
| "step": 560 |
| }, |
| { |
| "epoch": 2.2712550607287447, |
| "grad_norm": 0.8094860868848845, |
| "learning_rate": 2.6576976876493225e-05, |
| "loss": 0.8477, |
| "step": 561 |
| }, |
| { |
| "epoch": 2.2753036437246963, |
| "grad_norm": 0.8251501361956893, |
| "learning_rate": 2.652354173593614e-05, |
| "loss": 0.8993, |
| "step": 562 |
| }, |
| { |
| "epoch": 2.279352226720648, |
| "grad_norm": 0.8218059616270373, |
| "learning_rate": 2.647005443331317e-05, |
| "loss": 0.8989, |
| "step": 563 |
| }, |
| { |
| "epoch": 2.283400809716599, |
| "grad_norm": 0.7110813535922988, |
| "learning_rate": 2.6416515396307354e-05, |
| "loss": 0.9139, |
| "step": 564 |
| }, |
| { |
| "epoch": 2.2874493927125505, |
| "grad_norm": 0.8744126375148802, |
| "learning_rate": 2.636292505301542e-05, |
| "loss": 0.9084, |
| "step": 565 |
| }, |
| { |
| "epoch": 2.291497975708502, |
| "grad_norm": 0.7260927529264088, |
| "learning_rate": 2.630928383194431e-05, |
| "loss": 0.8902, |
| "step": 566 |
| }, |
| { |
| "epoch": 2.2955465587044532, |
| "grad_norm": 0.9410983892458527, |
| "learning_rate": 2.625559216200781e-05, |
| "loss": 0.917, |
| "step": 567 |
| }, |
| { |
| "epoch": 2.299595141700405, |
| "grad_norm": 0.7480705427370594, |
| "learning_rate": 2.620185047252308e-05, |
| "loss": 0.8867, |
| "step": 568 |
| }, |
| { |
| "epoch": 2.3036437246963564, |
| "grad_norm": 0.8848244300777325, |
| "learning_rate": 2.614805919320723e-05, |
| "loss": 0.8858, |
| "step": 569 |
| }, |
| { |
| "epoch": 2.3076923076923075, |
| "grad_norm": 1.0207666966465385, |
| "learning_rate": 2.6094218754173895e-05, |
| "loss": 0.9143, |
| "step": 570 |
| }, |
| { |
| "epoch": 2.311740890688259, |
| "grad_norm": 0.783341131230456, |
| "learning_rate": 2.60403295859298e-05, |
| "loss": 0.8854, |
| "step": 571 |
| }, |
| { |
| "epoch": 2.3157894736842106, |
| "grad_norm": 0.9494359176546776, |
| "learning_rate": 2.5986392119371297e-05, |
| "loss": 0.8971, |
| "step": 572 |
| }, |
| { |
| "epoch": 2.3198380566801617, |
| "grad_norm": 0.7433302190444511, |
| "learning_rate": 2.5932406785780928e-05, |
| "loss": 0.8922, |
| "step": 573 |
| }, |
| { |
| "epoch": 2.3238866396761133, |
| "grad_norm": 1.1079621589494801, |
| "learning_rate": 2.5878374016823987e-05, |
| "loss": 0.9005, |
| "step": 574 |
| }, |
| { |
| "epoch": 2.327935222672065, |
| "grad_norm": 0.8632915734084952, |
| "learning_rate": 2.5824294244545045e-05, |
| "loss": 0.9246, |
| "step": 575 |
| }, |
| { |
| "epoch": 2.331983805668016, |
| "grad_norm": 1.0799957509848144, |
| "learning_rate": 2.577016790136453e-05, |
| "loss": 0.8924, |
| "step": 576 |
| }, |
| { |
| "epoch": 2.3360323886639676, |
| "grad_norm": 0.8398627061461672, |
| "learning_rate": 2.5715995420075225e-05, |
| "loss": 0.9231, |
| "step": 577 |
| }, |
| { |
| "epoch": 2.340080971659919, |
| "grad_norm": 1.1295533240733682, |
| "learning_rate": 2.5661777233838864e-05, |
| "loss": 0.908, |
| "step": 578 |
| }, |
| { |
| "epoch": 2.3441295546558703, |
| "grad_norm": 0.9104144947799991, |
| "learning_rate": 2.5607513776182616e-05, |
| "loss": 0.8838, |
| "step": 579 |
| }, |
| { |
| "epoch": 2.348178137651822, |
| "grad_norm": 1.0475395727817096, |
| "learning_rate": 2.5553205480995633e-05, |
| "loss": 0.9079, |
| "step": 580 |
| }, |
| { |
| "epoch": 2.3522267206477734, |
| "grad_norm": 0.7422362698720059, |
| "learning_rate": 2.5498852782525604e-05, |
| "loss": 0.9111, |
| "step": 581 |
| }, |
| { |
| "epoch": 2.3562753036437245, |
| "grad_norm": 0.9990559993858348, |
| "learning_rate": 2.5444456115375252e-05, |
| "loss": 0.9009, |
| "step": 582 |
| }, |
| { |
| "epoch": 2.360323886639676, |
| "grad_norm": 0.7408087249093794, |
| "learning_rate": 2.539001591449889e-05, |
| "loss": 0.904, |
| "step": 583 |
| }, |
| { |
| "epoch": 2.3643724696356276, |
| "grad_norm": 0.9915071868629687, |
| "learning_rate": 2.53355326151989e-05, |
| "loss": 0.8903, |
| "step": 584 |
| }, |
| { |
| "epoch": 2.3684210526315788, |
| "grad_norm": 1.0001285055566962, |
| "learning_rate": 2.5281006653122288e-05, |
| "loss": 0.9157, |
| "step": 585 |
| }, |
| { |
| "epoch": 2.3724696356275303, |
| "grad_norm": 0.78251831947872, |
| "learning_rate": 2.5226438464257217e-05, |
| "loss": 0.9118, |
| "step": 586 |
| }, |
| { |
| "epoch": 2.376518218623482, |
| "grad_norm": 0.6780268009284526, |
| "learning_rate": 2.5171828484929447e-05, |
| "loss": 0.8899, |
| "step": 587 |
| }, |
| { |
| "epoch": 2.380566801619433, |
| "grad_norm": 0.7533496622844972, |
| "learning_rate": 2.5117177151798934e-05, |
| "loss": 0.9171, |
| "step": 588 |
| }, |
| { |
| "epoch": 2.3846153846153846, |
| "grad_norm": 0.6829758445031765, |
| "learning_rate": 2.506248490185627e-05, |
| "loss": 0.8892, |
| "step": 589 |
| }, |
| { |
| "epoch": 2.388663967611336, |
| "grad_norm": 0.8288308799470322, |
| "learning_rate": 2.500775217241923e-05, |
| "loss": 0.9365, |
| "step": 590 |
| }, |
| { |
| "epoch": 2.3927125506072873, |
| "grad_norm": 0.8007751554553462, |
| "learning_rate": 2.4952979401129266e-05, |
| "loss": 0.9041, |
| "step": 591 |
| }, |
| { |
| "epoch": 2.396761133603239, |
| "grad_norm": 0.8244938162766328, |
| "learning_rate": 2.4898167025948e-05, |
| "loss": 0.9143, |
| "step": 592 |
| }, |
| { |
| "epoch": 2.4008097165991904, |
| "grad_norm": 0.939848535454733, |
| "learning_rate": 2.4843315485153703e-05, |
| "loss": 0.9287, |
| "step": 593 |
| }, |
| { |
| "epoch": 2.4048582995951415, |
| "grad_norm": 0.7352833439771145, |
| "learning_rate": 2.4788425217337856e-05, |
| "loss": 0.9485, |
| "step": 594 |
| }, |
| { |
| "epoch": 2.408906882591093, |
| "grad_norm": 0.8348625769044344, |
| "learning_rate": 2.4733496661401572e-05, |
| "loss": 0.9262, |
| "step": 595 |
| }, |
| { |
| "epoch": 2.4129554655870447, |
| "grad_norm": 0.7139182241632022, |
| "learning_rate": 2.4678530256552114e-05, |
| "loss": 0.909, |
| "step": 596 |
| }, |
| { |
| "epoch": 2.417004048582996, |
| "grad_norm": 0.8248331246944739, |
| "learning_rate": 2.46235264422994e-05, |
| "loss": 0.8943, |
| "step": 597 |
| }, |
| { |
| "epoch": 2.4210526315789473, |
| "grad_norm": 0.7678486537390845, |
| "learning_rate": 2.4568485658452452e-05, |
| "loss": 0.9109, |
| "step": 598 |
| }, |
| { |
| "epoch": 2.425101214574899, |
| "grad_norm": 0.7576913500287776, |
| "learning_rate": 2.451340834511592e-05, |
| "loss": 0.9204, |
| "step": 599 |
| }, |
| { |
| "epoch": 2.42914979757085, |
| "grad_norm": 0.8381506794917551, |
| "learning_rate": 2.4458294942686528e-05, |
| "loss": 0.8664, |
| "step": 600 |
| }, |
| { |
| "epoch": 2.4331983805668016, |
| "grad_norm": 0.7574043612674364, |
| "learning_rate": 2.440314589184958e-05, |
| "loss": 0.9214, |
| "step": 601 |
| }, |
| { |
| "epoch": 2.437246963562753, |
| "grad_norm": 0.8595876549166049, |
| "learning_rate": 2.434796163357541e-05, |
| "loss": 0.9015, |
| "step": 602 |
| }, |
| { |
| "epoch": 2.4412955465587043, |
| "grad_norm": 0.7389865112806218, |
| "learning_rate": 2.429274260911588e-05, |
| "loss": 0.9314, |
| "step": 603 |
| }, |
| { |
| "epoch": 2.445344129554656, |
| "grad_norm": 1.020739889946229, |
| "learning_rate": 2.423748926000084e-05, |
| "loss": 0.9115, |
| "step": 604 |
| }, |
| { |
| "epoch": 2.4493927125506074, |
| "grad_norm": 0.7491237121286242, |
| "learning_rate": 2.418220202803459e-05, |
| "loss": 0.9134, |
| "step": 605 |
| }, |
| { |
| "epoch": 2.4534412955465585, |
| "grad_norm": 0.8383256971461217, |
| "learning_rate": 2.412688135529237e-05, |
| "loss": 0.8881, |
| "step": 606 |
| }, |
| { |
| "epoch": 2.45748987854251, |
| "grad_norm": 0.7910520431951439, |
| "learning_rate": 2.4071527684116807e-05, |
| "loss": 0.9016, |
| "step": 607 |
| }, |
| { |
| "epoch": 2.4615384615384617, |
| "grad_norm": 0.774567995885448, |
| "learning_rate": 2.4016141457114372e-05, |
| "loss": 0.9325, |
| "step": 608 |
| }, |
| { |
| "epoch": 2.465587044534413, |
| "grad_norm": 0.9381628509722176, |
| "learning_rate": 2.396072311715187e-05, |
| "loss": 0.9047, |
| "step": 609 |
| }, |
| { |
| "epoch": 2.4696356275303644, |
| "grad_norm": 0.7664510300422627, |
| "learning_rate": 2.3905273107352863e-05, |
| "loss": 0.87, |
| "step": 610 |
| }, |
| { |
| "epoch": 2.473684210526316, |
| "grad_norm": 0.9378175099366048, |
| "learning_rate": 2.3849791871094157e-05, |
| "loss": 0.9023, |
| "step": 611 |
| }, |
| { |
| "epoch": 2.477732793522267, |
| "grad_norm": 1.0646608684744956, |
| "learning_rate": 2.379427985200223e-05, |
| "loss": 0.9136, |
| "step": 612 |
| }, |
| { |
| "epoch": 2.4817813765182186, |
| "grad_norm": 0.7724988991954904, |
| "learning_rate": 2.373873749394972e-05, |
| "loss": 0.9098, |
| "step": 613 |
| }, |
| { |
| "epoch": 2.48582995951417, |
| "grad_norm": 1.226788123914223, |
| "learning_rate": 2.3683165241051836e-05, |
| "loss": 0.8999, |
| "step": 614 |
| }, |
| { |
| "epoch": 2.4898785425101213, |
| "grad_norm": 0.716688393929572, |
| "learning_rate": 2.3627563537662823e-05, |
| "loss": 0.8937, |
| "step": 615 |
| }, |
| { |
| "epoch": 2.493927125506073, |
| "grad_norm": 0.8066367878801375, |
| "learning_rate": 2.3571932828372433e-05, |
| "loss": 0.9136, |
| "step": 616 |
| }, |
| { |
| "epoch": 2.4979757085020244, |
| "grad_norm": 0.6834256933806026, |
| "learning_rate": 2.351627355800233e-05, |
| "loss": 0.8991, |
| "step": 617 |
| }, |
| { |
| "epoch": 2.5020242914979756, |
| "grad_norm": 0.7891776546170166, |
| "learning_rate": 2.346058617160256e-05, |
| "loss": 0.9259, |
| "step": 618 |
| }, |
| { |
| "epoch": 2.506072874493927, |
| "grad_norm": 0.7194804222295433, |
| "learning_rate": 2.3404871114447976e-05, |
| "loss": 0.8916, |
| "step": 619 |
| }, |
| { |
| "epoch": 2.5101214574898787, |
| "grad_norm": 0.8440421002862636, |
| "learning_rate": 2.334912883203469e-05, |
| "loss": 0.8883, |
| "step": 620 |
| }, |
| { |
| "epoch": 2.51417004048583, |
| "grad_norm": 0.8484339418789688, |
| "learning_rate": 2.3293359770076505e-05, |
| "loss": 0.9359, |
| "step": 621 |
| }, |
| { |
| "epoch": 2.5182186234817814, |
| "grad_norm": 0.6486235049711858, |
| "learning_rate": 2.323756437450136e-05, |
| "loss": 0.914, |
| "step": 622 |
| }, |
| { |
| "epoch": 2.522267206477733, |
| "grad_norm": 0.7628021813719684, |
| "learning_rate": 2.3181743091447754e-05, |
| "loss": 0.9164, |
| "step": 623 |
| }, |
| { |
| "epoch": 2.526315789473684, |
| "grad_norm": 0.6757856915005677, |
| "learning_rate": 2.3125896367261164e-05, |
| "loss": 0.8779, |
| "step": 624 |
| }, |
| { |
| "epoch": 2.5303643724696356, |
| "grad_norm": 0.6780939912620115, |
| "learning_rate": 2.307002464849052e-05, |
| "loss": 0.9267, |
| "step": 625 |
| }, |
| { |
| "epoch": 2.534412955465587, |
| "grad_norm": 0.6899371475900762, |
| "learning_rate": 2.30141283818846e-05, |
| "loss": 0.907, |
| "step": 626 |
| }, |
| { |
| "epoch": 2.5384615384615383, |
| "grad_norm": 0.700254115745053, |
| "learning_rate": 2.2958208014388462e-05, |
| "loss": 0.9073, |
| "step": 627 |
| }, |
| { |
| "epoch": 2.54251012145749, |
| "grad_norm": 0.6889904077564973, |
| "learning_rate": 2.2902263993139857e-05, |
| "loss": 0.8985, |
| "step": 628 |
| }, |
| { |
| "epoch": 2.5465587044534415, |
| "grad_norm": 0.6382508598558657, |
| "learning_rate": 2.2846296765465708e-05, |
| "loss": 0.9044, |
| "step": 629 |
| }, |
| { |
| "epoch": 2.5506072874493926, |
| "grad_norm": 0.7349774617385413, |
| "learning_rate": 2.2790306778878465e-05, |
| "loss": 0.9028, |
| "step": 630 |
| }, |
| { |
| "epoch": 2.554655870445344, |
| "grad_norm": 0.6875079679210848, |
| "learning_rate": 2.273429448107257e-05, |
| "loss": 0.9262, |
| "step": 631 |
| }, |
| { |
| "epoch": 2.5587044534412957, |
| "grad_norm": 0.6663213821914061, |
| "learning_rate": 2.2678260319920868e-05, |
| "loss": 0.8672, |
| "step": 632 |
| }, |
| { |
| "epoch": 2.562753036437247, |
| "grad_norm": 0.6872952228404817, |
| "learning_rate": 2.262220474347102e-05, |
| "loss": 0.9286, |
| "step": 633 |
| }, |
| { |
| "epoch": 2.5668016194331984, |
| "grad_norm": 0.7565882655869771, |
| "learning_rate": 2.2566128199941918e-05, |
| "loss": 0.8919, |
| "step": 634 |
| }, |
| { |
| "epoch": 2.57085020242915, |
| "grad_norm": 0.7619754406151699, |
| "learning_rate": 2.2510031137720104e-05, |
| "loss": 0.9216, |
| "step": 635 |
| }, |
| { |
| "epoch": 2.574898785425101, |
| "grad_norm": 0.7802887018032153, |
| "learning_rate": 2.2453914005356206e-05, |
| "loss": 0.8895, |
| "step": 636 |
| }, |
| { |
| "epoch": 2.5789473684210527, |
| "grad_norm": 0.7010484794909454, |
| "learning_rate": 2.2397777251561302e-05, |
| "loss": 0.8933, |
| "step": 637 |
| }, |
| { |
| "epoch": 2.582995951417004, |
| "grad_norm": 0.7818236538133725, |
| "learning_rate": 2.2341621325203393e-05, |
| "loss": 0.9251, |
| "step": 638 |
| }, |
| { |
| "epoch": 2.5870445344129553, |
| "grad_norm": 0.7503283336566918, |
| "learning_rate": 2.2285446675303764e-05, |
| "loss": 0.894, |
| "step": 639 |
| }, |
| { |
| "epoch": 2.591093117408907, |
| "grad_norm": 0.6661043705383741, |
| "learning_rate": 2.2229253751033423e-05, |
| "loss": 0.9123, |
| "step": 640 |
| }, |
| { |
| "epoch": 2.5951417004048585, |
| "grad_norm": 0.7735185141387093, |
| "learning_rate": 2.2173043001709492e-05, |
| "loss": 0.9245, |
| "step": 641 |
| }, |
| { |
| "epoch": 2.5991902834008096, |
| "grad_norm": 0.6232786021687494, |
| "learning_rate": 2.2116814876791627e-05, |
| "loss": 0.8893, |
| "step": 642 |
| }, |
| { |
| "epoch": 2.603238866396761, |
| "grad_norm": 0.7559575564276246, |
| "learning_rate": 2.2060569825878413e-05, |
| "loss": 0.9127, |
| "step": 643 |
| }, |
| { |
| "epoch": 2.6072874493927127, |
| "grad_norm": 0.6985443758647639, |
| "learning_rate": 2.2004308298703777e-05, |
| "loss": 0.9089, |
| "step": 644 |
| }, |
| { |
| "epoch": 2.611336032388664, |
| "grad_norm": 0.7680317753109804, |
| "learning_rate": 2.1948030745133396e-05, |
| "loss": 0.8986, |
| "step": 645 |
| }, |
| { |
| "epoch": 2.6153846153846154, |
| "grad_norm": 0.8403432129382882, |
| "learning_rate": 2.1891737615161083e-05, |
| "loss": 0.904, |
| "step": 646 |
| }, |
| { |
| "epoch": 2.619433198380567, |
| "grad_norm": 0.7302216013548477, |
| "learning_rate": 2.1835429358905218e-05, |
| "loss": 0.9314, |
| "step": 647 |
| }, |
| { |
| "epoch": 2.623481781376518, |
| "grad_norm": 0.7853432163561157, |
| "learning_rate": 2.17791064266051e-05, |
| "loss": 0.9053, |
| "step": 648 |
| }, |
| { |
| "epoch": 2.6275303643724697, |
| "grad_norm": 0.6704871982876516, |
| "learning_rate": 2.172276926861741e-05, |
| "loss": 0.8825, |
| "step": 649 |
| }, |
| { |
| "epoch": 2.6315789473684212, |
| "grad_norm": 0.779836325267494, |
| "learning_rate": 2.166641833541255e-05, |
| "loss": 0.91, |
| "step": 650 |
| }, |
| { |
| "epoch": 2.6356275303643724, |
| "grad_norm": 0.7396404097082778, |
| "learning_rate": 2.161005407757109e-05, |
| "loss": 0.9286, |
| "step": 651 |
| }, |
| { |
| "epoch": 2.639676113360324, |
| "grad_norm": 0.6341195656037277, |
| "learning_rate": 2.155367694578013e-05, |
| "loss": 0.9163, |
| "step": 652 |
| }, |
| { |
| "epoch": 2.6437246963562755, |
| "grad_norm": 0.8866745042395711, |
| "learning_rate": 2.1497287390829712e-05, |
| "loss": 0.8906, |
| "step": 653 |
| }, |
| { |
| "epoch": 2.6477732793522266, |
| "grad_norm": 0.5833007228472453, |
| "learning_rate": 2.1440885863609208e-05, |
| "loss": 0.897, |
| "step": 654 |
| }, |
| { |
| "epoch": 2.651821862348178, |
| "grad_norm": 0.8638209794406689, |
| "learning_rate": 2.1384472815103737e-05, |
| "loss": 0.9149, |
| "step": 655 |
| }, |
| { |
| "epoch": 2.6558704453441297, |
| "grad_norm": 0.6789000128212279, |
| "learning_rate": 2.1328048696390534e-05, |
| "loss": 0.9348, |
| "step": 656 |
| }, |
| { |
| "epoch": 2.659919028340081, |
| "grad_norm": 0.728157942928097, |
| "learning_rate": 2.1271613958635342e-05, |
| "loss": 0.8914, |
| "step": 657 |
| }, |
| { |
| "epoch": 2.6639676113360324, |
| "grad_norm": 0.8167166624441069, |
| "learning_rate": 2.1215169053088825e-05, |
| "loss": 0.9015, |
| "step": 658 |
| }, |
| { |
| "epoch": 2.668016194331984, |
| "grad_norm": 0.669011753944312, |
| "learning_rate": 2.1158714431082937e-05, |
| "loss": 0.8956, |
| "step": 659 |
| }, |
| { |
| "epoch": 2.672064777327935, |
| "grad_norm": 0.7810694039577911, |
| "learning_rate": 2.110225054402733e-05, |
| "loss": 0.9243, |
| "step": 660 |
| }, |
| { |
| "epoch": 2.6761133603238867, |
| "grad_norm": 0.6373647345604864, |
| "learning_rate": 2.1045777843405747e-05, |
| "loss": 0.9102, |
| "step": 661 |
| }, |
| { |
| "epoch": 2.6801619433198383, |
| "grad_norm": 0.7370513257806691, |
| "learning_rate": 2.09892967807724e-05, |
| "loss": 0.9225, |
| "step": 662 |
| }, |
| { |
| "epoch": 2.6842105263157894, |
| "grad_norm": 0.7557658654822316, |
| "learning_rate": 2.093280780774835e-05, |
| "loss": 0.9399, |
| "step": 663 |
| }, |
| { |
| "epoch": 2.688259109311741, |
| "grad_norm": 0.7317875705511553, |
| "learning_rate": 2.087631137601793e-05, |
| "loss": 0.9195, |
| "step": 664 |
| }, |
| { |
| "epoch": 2.6923076923076925, |
| "grad_norm": 0.7147154447345427, |
| "learning_rate": 2.0819807937325104e-05, |
| "loss": 0.9084, |
| "step": 665 |
| }, |
| { |
| "epoch": 2.6963562753036436, |
| "grad_norm": 0.6238512393858743, |
| "learning_rate": 2.0763297943469847e-05, |
| "loss": 0.8585, |
| "step": 666 |
| }, |
| { |
| "epoch": 2.700404858299595, |
| "grad_norm": 0.7558146417722807, |
| "learning_rate": 2.070678184630458e-05, |
| "loss": 0.8941, |
| "step": 667 |
| }, |
| { |
| "epoch": 2.7044534412955468, |
| "grad_norm": 0.6298785153810263, |
| "learning_rate": 2.06502600977305e-05, |
| "loss": 0.881, |
| "step": 668 |
| }, |
| { |
| "epoch": 2.708502024291498, |
| "grad_norm": 0.6541761288123853, |
| "learning_rate": 2.0593733149694008e-05, |
| "loss": 0.8852, |
| "step": 669 |
| }, |
| { |
| "epoch": 2.7125506072874495, |
| "grad_norm": 0.5994542918184318, |
| "learning_rate": 2.0537201454183074e-05, |
| "loss": 0.8842, |
| "step": 670 |
| }, |
| { |
| "epoch": 2.716599190283401, |
| "grad_norm": 0.6190033540839754, |
| "learning_rate": 2.0480665463223626e-05, |
| "loss": 0.8958, |
| "step": 671 |
| }, |
| { |
| "epoch": 2.720647773279352, |
| "grad_norm": 0.8554291974565289, |
| "learning_rate": 2.042412562887594e-05, |
| "loss": 0.9242, |
| "step": 672 |
| }, |
| { |
| "epoch": 2.7246963562753037, |
| "grad_norm": 0.610877322757858, |
| "learning_rate": 2.0367582403231033e-05, |
| "loss": 0.9009, |
| "step": 673 |
| }, |
| { |
| "epoch": 2.7287449392712553, |
| "grad_norm": 0.7780870928602482, |
| "learning_rate": 2.0311036238407023e-05, |
| "loss": 0.9431, |
| "step": 674 |
| }, |
| { |
| "epoch": 2.7327935222672064, |
| "grad_norm": 0.6669610895297321, |
| "learning_rate": 2.0254487586545534e-05, |
| "loss": 0.9008, |
| "step": 675 |
| }, |
| { |
| "epoch": 2.736842105263158, |
| "grad_norm": 0.8099899926278401, |
| "learning_rate": 2.0197936899808088e-05, |
| "loss": 0.8794, |
| "step": 676 |
| }, |
| { |
| "epoch": 2.7408906882591095, |
| "grad_norm": 0.6694020506365862, |
| "learning_rate": 2.0141384630372466e-05, |
| "loss": 0.8833, |
| "step": 677 |
| }, |
| { |
| "epoch": 2.7449392712550607, |
| "grad_norm": 0.8053542627515762, |
| "learning_rate": 2.0084831230429106e-05, |
| "loss": 0.9113, |
| "step": 678 |
| }, |
| { |
| "epoch": 2.748987854251012, |
| "grad_norm": 0.829184703233809, |
| "learning_rate": 2.0028277152177492e-05, |
| "loss": 0.9232, |
| "step": 679 |
| }, |
| { |
| "epoch": 2.753036437246964, |
| "grad_norm": 0.9555539381486678, |
| "learning_rate": 1.9971722847822518e-05, |
| "loss": 0.9291, |
| "step": 680 |
| }, |
| { |
| "epoch": 2.757085020242915, |
| "grad_norm": 0.9687509398579465, |
| "learning_rate": 1.99151687695709e-05, |
| "loss": 0.927, |
| "step": 681 |
| }, |
| { |
| "epoch": 2.7611336032388665, |
| "grad_norm": 0.8585517822333045, |
| "learning_rate": 1.9858615369627537e-05, |
| "loss": 0.8959, |
| "step": 682 |
| }, |
| { |
| "epoch": 2.765182186234818, |
| "grad_norm": 0.970592651497807, |
| "learning_rate": 1.9802063100191916e-05, |
| "loss": 0.9324, |
| "step": 683 |
| }, |
| { |
| "epoch": 2.769230769230769, |
| "grad_norm": 0.8726976721640566, |
| "learning_rate": 1.974551241345447e-05, |
| "loss": 0.8963, |
| "step": 684 |
| }, |
| { |
| "epoch": 2.7732793522267207, |
| "grad_norm": 0.7880555474754197, |
| "learning_rate": 1.9688963761592984e-05, |
| "loss": 0.91, |
| "step": 685 |
| }, |
| { |
| "epoch": 2.7773279352226723, |
| "grad_norm": 0.7330274216276548, |
| "learning_rate": 1.963241759676897e-05, |
| "loss": 0.8991, |
| "step": 686 |
| }, |
| { |
| "epoch": 2.7813765182186234, |
| "grad_norm": 0.6466419292658644, |
| "learning_rate": 1.9575874371124066e-05, |
| "loss": 0.9177, |
| "step": 687 |
| }, |
| { |
| "epoch": 2.785425101214575, |
| "grad_norm": 0.8711521815627767, |
| "learning_rate": 1.951933453677638e-05, |
| "loss": 0.8825, |
| "step": 688 |
| }, |
| { |
| "epoch": 2.7894736842105265, |
| "grad_norm": 0.6741330996284937, |
| "learning_rate": 1.9462798545816932e-05, |
| "loss": 0.9108, |
| "step": 689 |
| }, |
| { |
| "epoch": 2.7935222672064777, |
| "grad_norm": 0.7266150153912465, |
| "learning_rate": 1.9406266850305996e-05, |
| "loss": 0.8732, |
| "step": 690 |
| }, |
| { |
| "epoch": 2.7975708502024292, |
| "grad_norm": 0.7528978370662313, |
| "learning_rate": 1.9349739902269503e-05, |
| "loss": 0.8727, |
| "step": 691 |
| }, |
| { |
| "epoch": 2.801619433198381, |
| "grad_norm": 0.6656023288413814, |
| "learning_rate": 1.9293218153695425e-05, |
| "loss": 0.9137, |
| "step": 692 |
| }, |
| { |
| "epoch": 2.805668016194332, |
| "grad_norm": 0.7025468367310848, |
| "learning_rate": 1.9236702056530153e-05, |
| "loss": 0.9, |
| "step": 693 |
| }, |
| { |
| "epoch": 2.8097165991902835, |
| "grad_norm": 0.6695610680995763, |
| "learning_rate": 1.9180192062674907e-05, |
| "loss": 0.9027, |
| "step": 694 |
| }, |
| { |
| "epoch": 2.813765182186235, |
| "grad_norm": 0.6820246964000168, |
| "learning_rate": 1.9123688623982076e-05, |
| "loss": 0.925, |
| "step": 695 |
| }, |
| { |
| "epoch": 2.817813765182186, |
| "grad_norm": 0.6389810869158864, |
| "learning_rate": 1.906719219225165e-05, |
| "loss": 0.9167, |
| "step": 696 |
| }, |
| { |
| "epoch": 2.8218623481781377, |
| "grad_norm": 0.6781198991058135, |
| "learning_rate": 1.9010703219227603e-05, |
| "loss": 0.9294, |
| "step": 697 |
| }, |
| { |
| "epoch": 2.8259109311740893, |
| "grad_norm": 0.7137593691120043, |
| "learning_rate": 1.8954222156594256e-05, |
| "loss": 0.9109, |
| "step": 698 |
| }, |
| { |
| "epoch": 2.8299595141700404, |
| "grad_norm": 0.716782587321261, |
| "learning_rate": 1.8897749455972673e-05, |
| "loss": 0.8958, |
| "step": 699 |
| }, |
| { |
| "epoch": 2.834008097165992, |
| "grad_norm": 0.7069478994090977, |
| "learning_rate": 1.884128556891707e-05, |
| "loss": 0.8886, |
| "step": 700 |
| }, |
| { |
| "epoch": 2.8380566801619436, |
| "grad_norm": 0.67303864543188, |
| "learning_rate": 1.878483094691119e-05, |
| "loss": 0.8989, |
| "step": 701 |
| }, |
| { |
| "epoch": 2.8421052631578947, |
| "grad_norm": 0.7004741877344418, |
| "learning_rate": 1.8728386041364664e-05, |
| "loss": 0.8886, |
| "step": 702 |
| }, |
| { |
| "epoch": 2.8461538461538463, |
| "grad_norm": 0.8062782394467252, |
| "learning_rate": 1.867195130360947e-05, |
| "loss": 0.896, |
| "step": 703 |
| }, |
| { |
| "epoch": 2.850202429149798, |
| "grad_norm": 0.7246207928988877, |
| "learning_rate": 1.8615527184896263e-05, |
| "loss": 0.9266, |
| "step": 704 |
| }, |
| { |
| "epoch": 2.854251012145749, |
| "grad_norm": 0.6406062746208214, |
| "learning_rate": 1.8559114136390795e-05, |
| "loss": 0.8703, |
| "step": 705 |
| }, |
| { |
| "epoch": 2.8582995951417005, |
| "grad_norm": 0.7355842537543618, |
| "learning_rate": 1.8502712609170298e-05, |
| "loss": 0.9086, |
| "step": 706 |
| }, |
| { |
| "epoch": 2.862348178137652, |
| "grad_norm": 0.6717470662759829, |
| "learning_rate": 1.8446323054219876e-05, |
| "loss": 0.8942, |
| "step": 707 |
| }, |
| { |
| "epoch": 2.866396761133603, |
| "grad_norm": 0.7167415200091812, |
| "learning_rate": 1.838994592242891e-05, |
| "loss": 0.9035, |
| "step": 708 |
| }, |
| { |
| "epoch": 2.8704453441295548, |
| "grad_norm": 0.7593417558966612, |
| "learning_rate": 1.8333581664587453e-05, |
| "loss": 0.9431, |
| "step": 709 |
| }, |
| { |
| "epoch": 2.8744939271255063, |
| "grad_norm": 0.6531769904941046, |
| "learning_rate": 1.8277230731382593e-05, |
| "loss": 0.8878, |
| "step": 710 |
| }, |
| { |
| "epoch": 2.8785425101214575, |
| "grad_norm": 0.655530443995037, |
| "learning_rate": 1.82208935733949e-05, |
| "loss": 0.8708, |
| "step": 711 |
| }, |
| { |
| "epoch": 2.882591093117409, |
| "grad_norm": 0.8797273109345312, |
| "learning_rate": 1.8164570641094793e-05, |
| "loss": 0.8978, |
| "step": 712 |
| }, |
| { |
| "epoch": 2.8866396761133606, |
| "grad_norm": 0.6277801082334746, |
| "learning_rate": 1.810826238483892e-05, |
| "loss": 0.9367, |
| "step": 713 |
| }, |
| { |
| "epoch": 2.8906882591093117, |
| "grad_norm": 0.8653685001433539, |
| "learning_rate": 1.8051969254866608e-05, |
| "loss": 0.9326, |
| "step": 714 |
| }, |
| { |
| "epoch": 2.8947368421052633, |
| "grad_norm": 0.6978420022426539, |
| "learning_rate": 1.7995691701296226e-05, |
| "loss": 0.9172, |
| "step": 715 |
| }, |
| { |
| "epoch": 2.898785425101215, |
| "grad_norm": 0.8711205540660308, |
| "learning_rate": 1.7939430174121593e-05, |
| "loss": 0.9205, |
| "step": 716 |
| }, |
| { |
| "epoch": 2.902834008097166, |
| "grad_norm": 0.863636982510114, |
| "learning_rate": 1.788318512320838e-05, |
| "loss": 0.9076, |
| "step": 717 |
| }, |
| { |
| "epoch": 2.9068825910931175, |
| "grad_norm": 0.8668977192486608, |
| "learning_rate": 1.782695699829051e-05, |
| "loss": 0.9227, |
| "step": 718 |
| }, |
| { |
| "epoch": 2.910931174089069, |
| "grad_norm": 0.8785568260321244, |
| "learning_rate": 1.7770746248966587e-05, |
| "loss": 0.9141, |
| "step": 719 |
| }, |
| { |
| "epoch": 2.91497975708502, |
| "grad_norm": 0.7545706866394984, |
| "learning_rate": 1.7714553324696243e-05, |
| "loss": 0.8898, |
| "step": 720 |
| }, |
| { |
| "epoch": 2.919028340080972, |
| "grad_norm": 0.7187371775248349, |
| "learning_rate": 1.7658378674796614e-05, |
| "loss": 0.8966, |
| "step": 721 |
| }, |
| { |
| "epoch": 2.9230769230769234, |
| "grad_norm": 0.8967511349172098, |
| "learning_rate": 1.76022227484387e-05, |
| "loss": 0.9157, |
| "step": 722 |
| }, |
| { |
| "epoch": 2.9271255060728745, |
| "grad_norm": 0.5801647242468748, |
| "learning_rate": 1.7546085994643807e-05, |
| "loss": 0.8973, |
| "step": 723 |
| }, |
| { |
| "epoch": 2.931174089068826, |
| "grad_norm": 0.9920237194361617, |
| "learning_rate": 1.7489968862279902e-05, |
| "loss": 0.9132, |
| "step": 724 |
| }, |
| { |
| "epoch": 2.9352226720647776, |
| "grad_norm": 0.7903943559580976, |
| "learning_rate": 1.743387180005809e-05, |
| "loss": 0.8971, |
| "step": 725 |
| }, |
| { |
| "epoch": 2.9392712550607287, |
| "grad_norm": 0.681377906362162, |
| "learning_rate": 1.737779525652899e-05, |
| "loss": 0.9232, |
| "step": 726 |
| }, |
| { |
| "epoch": 2.9433198380566803, |
| "grad_norm": 0.9716316391961594, |
| "learning_rate": 1.732173968007914e-05, |
| "loss": 0.9102, |
| "step": 727 |
| }, |
| { |
| "epoch": 2.9473684210526314, |
| "grad_norm": 0.6768097682472366, |
| "learning_rate": 1.7265705518927437e-05, |
| "loss": 0.9316, |
| "step": 728 |
| }, |
| { |
| "epoch": 2.951417004048583, |
| "grad_norm": 0.7297239601298781, |
| "learning_rate": 1.7209693221121542e-05, |
| "loss": 0.8816, |
| "step": 729 |
| }, |
| { |
| "epoch": 2.9554655870445345, |
| "grad_norm": 0.8538364985151601, |
| "learning_rate": 1.7153703234534302e-05, |
| "loss": 0.9109, |
| "step": 730 |
| }, |
| { |
| "epoch": 2.9595141700404857, |
| "grad_norm": 0.6268540777409978, |
| "learning_rate": 1.709773600686015e-05, |
| "loss": 0.8921, |
| "step": 731 |
| }, |
| { |
| "epoch": 2.9635627530364372, |
| "grad_norm": 0.7141776249850967, |
| "learning_rate": 1.7041791985611548e-05, |
| "loss": 0.9279, |
| "step": 732 |
| }, |
| { |
| "epoch": 2.967611336032389, |
| "grad_norm": 0.6527186294503934, |
| "learning_rate": 1.69858716181154e-05, |
| "loss": 0.8689, |
| "step": 733 |
| }, |
| { |
| "epoch": 2.97165991902834, |
| "grad_norm": 0.7169248937750502, |
| "learning_rate": 1.692997535150948e-05, |
| "loss": 0.911, |
| "step": 734 |
| }, |
| { |
| "epoch": 2.9757085020242915, |
| "grad_norm": 0.645259372347957, |
| "learning_rate": 1.687410363273884e-05, |
| "loss": 0.8905, |
| "step": 735 |
| }, |
| { |
| "epoch": 2.979757085020243, |
| "grad_norm": 0.7696111944785683, |
| "learning_rate": 1.6818256908552257e-05, |
| "loss": 0.8964, |
| "step": 736 |
| }, |
| { |
| "epoch": 2.983805668016194, |
| "grad_norm": 0.6868669687854592, |
| "learning_rate": 1.6762435625498646e-05, |
| "loss": 0.9234, |
| "step": 737 |
| }, |
| { |
| "epoch": 2.9878542510121457, |
| "grad_norm": 0.712428495308762, |
| "learning_rate": 1.67066402299235e-05, |
| "loss": 0.9218, |
| "step": 738 |
| }, |
| { |
| "epoch": 2.9919028340080973, |
| "grad_norm": 0.6155962161876348, |
| "learning_rate": 1.6650871167965313e-05, |
| "loss": 0.8974, |
| "step": 739 |
| }, |
| { |
| "epoch": 2.9959514170040484, |
| "grad_norm": 0.7763155975250593, |
| "learning_rate": 1.6595128885552028e-05, |
| "loss": 0.9203, |
| "step": 740 |
| }, |
| { |
| "epoch": 3.0, |
| "grad_norm": 0.584557587301322, |
| "learning_rate": 1.653941382839745e-05, |
| "loss": 0.849, |
| "step": 741 |
| }, |
| { |
| "epoch": 3.0040485829959516, |
| "grad_norm": 1.0069820249186228, |
| "learning_rate": 1.6483726441997673e-05, |
| "loss": 0.7182, |
| "step": 742 |
| }, |
| { |
| "epoch": 3.0080971659919027, |
| "grad_norm": 0.7175849662717889, |
| "learning_rate": 1.642806717162757e-05, |
| "loss": 0.7079, |
| "step": 743 |
| }, |
| { |
| "epoch": 3.0121457489878543, |
| "grad_norm": 1.7136417240662982, |
| "learning_rate": 1.637243646233718e-05, |
| "loss": 0.738, |
| "step": 744 |
| }, |
| { |
| "epoch": 3.016194331983806, |
| "grad_norm": 0.9942797577916082, |
| "learning_rate": 1.6316834758948174e-05, |
| "loss": 0.7186, |
| "step": 745 |
| }, |
| { |
| "epoch": 3.020242914979757, |
| "grad_norm": 0.9903588024315164, |
| "learning_rate": 1.6261262506050282e-05, |
| "loss": 0.7324, |
| "step": 746 |
| }, |
| { |
| "epoch": 3.0242914979757085, |
| "grad_norm": 1.2061024159795197, |
| "learning_rate": 1.620572014799777e-05, |
| "loss": 0.7083, |
| "step": 747 |
| }, |
| { |
| "epoch": 3.02834008097166, |
| "grad_norm": 0.9185711122886613, |
| "learning_rate": 1.6150208128905857e-05, |
| "loss": 0.7092, |
| "step": 748 |
| }, |
| { |
| "epoch": 3.032388663967611, |
| "grad_norm": 1.2308710402852432, |
| "learning_rate": 1.6094726892647147e-05, |
| "loss": 0.6886, |
| "step": 749 |
| }, |
| { |
| "epoch": 3.0364372469635628, |
| "grad_norm": 1.0603403975230439, |
| "learning_rate": 1.6039276882848135e-05, |
| "loss": 0.7097, |
| "step": 750 |
| }, |
| { |
| "epoch": 3.0404858299595143, |
| "grad_norm": 0.8843885808123224, |
| "learning_rate": 1.5983858542885635e-05, |
| "loss": 0.7303, |
| "step": 751 |
| }, |
| { |
| "epoch": 3.0445344129554655, |
| "grad_norm": 1.1100330364375783, |
| "learning_rate": 1.5928472315883203e-05, |
| "loss": 0.6941, |
| "step": 752 |
| }, |
| { |
| "epoch": 3.048582995951417, |
| "grad_norm": 0.879520836478064, |
| "learning_rate": 1.5873118644707633e-05, |
| "loss": 0.7263, |
| "step": 753 |
| }, |
| { |
| "epoch": 3.0526315789473686, |
| "grad_norm": 0.9270108218918809, |
| "learning_rate": 1.5817797971965413e-05, |
| "loss": 0.6815, |
| "step": 754 |
| }, |
| { |
| "epoch": 3.0566801619433197, |
| "grad_norm": 1.0216362551867357, |
| "learning_rate": 1.576251073999917e-05, |
| "loss": 0.6729, |
| "step": 755 |
| }, |
| { |
| "epoch": 3.0607287449392713, |
| "grad_norm": 0.9177261817189604, |
| "learning_rate": 1.5707257390884126e-05, |
| "loss": 0.7043, |
| "step": 756 |
| }, |
| { |
| "epoch": 3.064777327935223, |
| "grad_norm": 0.9915075932584778, |
| "learning_rate": 1.5652038366424595e-05, |
| "loss": 0.7166, |
| "step": 757 |
| }, |
| { |
| "epoch": 3.068825910931174, |
| "grad_norm": 0.8788835847750213, |
| "learning_rate": 1.5596854108150424e-05, |
| "loss": 0.7195, |
| "step": 758 |
| }, |
| { |
| "epoch": 3.0728744939271255, |
| "grad_norm": 0.8532324792868063, |
| "learning_rate": 1.5541705057313476e-05, |
| "loss": 0.6795, |
| "step": 759 |
| }, |
| { |
| "epoch": 3.076923076923077, |
| "grad_norm": 0.8008471416974539, |
| "learning_rate": 1.5486591654884086e-05, |
| "loss": 0.6996, |
| "step": 760 |
| }, |
| { |
| "epoch": 3.080971659919028, |
| "grad_norm": 0.846659809462749, |
| "learning_rate": 1.543151434154755e-05, |
| "loss": 0.7173, |
| "step": 761 |
| }, |
| { |
| "epoch": 3.08502024291498, |
| "grad_norm": 0.8978454812968363, |
| "learning_rate": 1.5376473557700615e-05, |
| "loss": 0.7094, |
| "step": 762 |
| }, |
| { |
| "epoch": 3.0890688259109313, |
| "grad_norm": 0.6964732606118762, |
| "learning_rate": 1.5321469743447893e-05, |
| "loss": 0.6867, |
| "step": 763 |
| }, |
| { |
| "epoch": 3.0931174089068825, |
| "grad_norm": 0.8028620837076302, |
| "learning_rate": 1.5266503338598434e-05, |
| "loss": 0.7027, |
| "step": 764 |
| }, |
| { |
| "epoch": 3.097165991902834, |
| "grad_norm": 0.7241268128948171, |
| "learning_rate": 1.5211574782662149e-05, |
| "loss": 0.7029, |
| "step": 765 |
| }, |
| { |
| "epoch": 3.1012145748987856, |
| "grad_norm": 0.6526242384053065, |
| "learning_rate": 1.5156684514846307e-05, |
| "loss": 0.6984, |
| "step": 766 |
| }, |
| { |
| "epoch": 3.1052631578947367, |
| "grad_norm": 0.7129495022462077, |
| "learning_rate": 1.5101832974052012e-05, |
| "loss": 0.7032, |
| "step": 767 |
| }, |
| { |
| "epoch": 3.1093117408906883, |
| "grad_norm": 0.761439638911344, |
| "learning_rate": 1.5047020598870737e-05, |
| "loss": 0.7162, |
| "step": 768 |
| }, |
| { |
| "epoch": 3.11336032388664, |
| "grad_norm": 0.6061956261826503, |
| "learning_rate": 1.4992247827580778e-05, |
| "loss": 0.7063, |
| "step": 769 |
| }, |
| { |
| "epoch": 3.117408906882591, |
| "grad_norm": 0.7157459267179416, |
| "learning_rate": 1.493751509814374e-05, |
| "loss": 0.7333, |
| "step": 770 |
| }, |
| { |
| "epoch": 3.1214574898785425, |
| "grad_norm": 0.6090833011248832, |
| "learning_rate": 1.4882822848201073e-05, |
| "loss": 0.7118, |
| "step": 771 |
| }, |
| { |
| "epoch": 3.125506072874494, |
| "grad_norm": 0.5570263011504701, |
| "learning_rate": 1.4828171515070553e-05, |
| "loss": 0.7078, |
| "step": 772 |
| }, |
| { |
| "epoch": 3.1295546558704452, |
| "grad_norm": 0.6855616894675642, |
| "learning_rate": 1.4773561535742793e-05, |
| "loss": 0.6925, |
| "step": 773 |
| }, |
| { |
| "epoch": 3.133603238866397, |
| "grad_norm": 0.5923136885347966, |
| "learning_rate": 1.4718993346877715e-05, |
| "loss": 0.6911, |
| "step": 774 |
| }, |
| { |
| "epoch": 3.1376518218623484, |
| "grad_norm": 0.5732834227860268, |
| "learning_rate": 1.466446738480111e-05, |
| "loss": 0.71, |
| "step": 775 |
| }, |
| { |
| "epoch": 3.1417004048582995, |
| "grad_norm": 0.6084817362876497, |
| "learning_rate": 1.460998408550112e-05, |
| "loss": 0.6969, |
| "step": 776 |
| }, |
| { |
| "epoch": 3.145748987854251, |
| "grad_norm": 0.5574738835361038, |
| "learning_rate": 1.4555543884624751e-05, |
| "loss": 0.7165, |
| "step": 777 |
| }, |
| { |
| "epoch": 3.1497975708502026, |
| "grad_norm": 0.5848646575610474, |
| "learning_rate": 1.4501147217474402e-05, |
| "loss": 0.7172, |
| "step": 778 |
| }, |
| { |
| "epoch": 3.1538461538461537, |
| "grad_norm": 0.5962855012022937, |
| "learning_rate": 1.444679451900437e-05, |
| "loss": 0.7012, |
| "step": 779 |
| }, |
| { |
| "epoch": 3.1578947368421053, |
| "grad_norm": 0.5387584634625391, |
| "learning_rate": 1.4392486223817397e-05, |
| "loss": 0.6974, |
| "step": 780 |
| }, |
| { |
| "epoch": 3.161943319838057, |
| "grad_norm": 0.5733803250036643, |
| "learning_rate": 1.4338222766161143e-05, |
| "loss": 0.7038, |
| "step": 781 |
| }, |
| { |
| "epoch": 3.165991902834008, |
| "grad_norm": 0.6622752264286887, |
| "learning_rate": 1.4284004579924777e-05, |
| "loss": 0.703, |
| "step": 782 |
| }, |
| { |
| "epoch": 3.1700404858299596, |
| "grad_norm": 0.5284643889729461, |
| "learning_rate": 1.4229832098635479e-05, |
| "loss": 0.7011, |
| "step": 783 |
| }, |
| { |
| "epoch": 3.174089068825911, |
| "grad_norm": 0.5818586579560491, |
| "learning_rate": 1.4175705755454963e-05, |
| "loss": 0.7088, |
| "step": 784 |
| }, |
| { |
| "epoch": 3.1781376518218623, |
| "grad_norm": 0.5527528797925313, |
| "learning_rate": 1.412162598317602e-05, |
| "loss": 0.7157, |
| "step": 785 |
| }, |
| { |
| "epoch": 3.182186234817814, |
| "grad_norm": 0.6155319585505107, |
| "learning_rate": 1.406759321421907e-05, |
| "loss": 0.7126, |
| "step": 786 |
| }, |
| { |
| "epoch": 3.1862348178137654, |
| "grad_norm": 0.5439910823633117, |
| "learning_rate": 1.401360788062871e-05, |
| "loss": 0.7096, |
| "step": 787 |
| }, |
| { |
| "epoch": 3.1902834008097165, |
| "grad_norm": 0.5670521588702437, |
| "learning_rate": 1.3959670414070204e-05, |
| "loss": 0.7301, |
| "step": 788 |
| }, |
| { |
| "epoch": 3.194331983805668, |
| "grad_norm": 0.5888808361082403, |
| "learning_rate": 1.3905781245826108e-05, |
| "loss": 0.714, |
| "step": 789 |
| }, |
| { |
| "epoch": 3.1983805668016196, |
| "grad_norm": 0.6057789611707872, |
| "learning_rate": 1.3851940806792778e-05, |
| "loss": 0.7367, |
| "step": 790 |
| }, |
| { |
| "epoch": 3.2024291497975708, |
| "grad_norm": 0.5409189627498653, |
| "learning_rate": 1.379814952747693e-05, |
| "loss": 0.698, |
| "step": 791 |
| }, |
| { |
| "epoch": 3.2064777327935223, |
| "grad_norm": 0.5187962427820666, |
| "learning_rate": 1.3744407837992193e-05, |
| "loss": 0.7069, |
| "step": 792 |
| }, |
| { |
| "epoch": 3.2105263157894735, |
| "grad_norm": 0.5290662938169406, |
| "learning_rate": 1.3690716168055692e-05, |
| "loss": 0.6981, |
| "step": 793 |
| }, |
| { |
| "epoch": 3.214574898785425, |
| "grad_norm": 0.5721393171943286, |
| "learning_rate": 1.363707494698459e-05, |
| "loss": 0.7137, |
| "step": 794 |
| }, |
| { |
| "epoch": 3.2186234817813766, |
| "grad_norm": 0.54925139743246, |
| "learning_rate": 1.358348460369265e-05, |
| "loss": 0.7118, |
| "step": 795 |
| }, |
| { |
| "epoch": 3.2226720647773277, |
| "grad_norm": 0.6302613399039392, |
| "learning_rate": 1.352994556668684e-05, |
| "loss": 0.7144, |
| "step": 796 |
| }, |
| { |
| "epoch": 3.2267206477732793, |
| "grad_norm": 0.5581461321186655, |
| "learning_rate": 1.3476458264063865e-05, |
| "loss": 0.742, |
| "step": 797 |
| }, |
| { |
| "epoch": 3.230769230769231, |
| "grad_norm": 0.5713116343141083, |
| "learning_rate": 1.342302312350678e-05, |
| "loss": 0.7029, |
| "step": 798 |
| }, |
| { |
| "epoch": 3.234817813765182, |
| "grad_norm": 0.6239898038615358, |
| "learning_rate": 1.3369640572281537e-05, |
| "loss": 0.7001, |
| "step": 799 |
| }, |
| { |
| "epoch": 3.2388663967611335, |
| "grad_norm": 0.566191700081629, |
| "learning_rate": 1.3316311037233596e-05, |
| "loss": 0.7063, |
| "step": 800 |
| }, |
| { |
| "epoch": 3.242914979757085, |
| "grad_norm": 0.5363454713920637, |
| "learning_rate": 1.326303494478451e-05, |
| "loss": 0.6905, |
| "step": 801 |
| }, |
| { |
| "epoch": 3.246963562753036, |
| "grad_norm": 0.6009600148347235, |
| "learning_rate": 1.3209812720928472e-05, |
| "loss": 0.7267, |
| "step": 802 |
| }, |
| { |
| "epoch": 3.251012145748988, |
| "grad_norm": 0.6543626281094772, |
| "learning_rate": 1.315664479122898e-05, |
| "loss": 0.702, |
| "step": 803 |
| }, |
| { |
| "epoch": 3.2550607287449393, |
| "grad_norm": 0.6479362728325447, |
| "learning_rate": 1.3103531580815378e-05, |
| "loss": 0.7087, |
| "step": 804 |
| }, |
| { |
| "epoch": 3.2591093117408905, |
| "grad_norm": 0.5721364616548021, |
| "learning_rate": 1.3050473514379488e-05, |
| "loss": 0.711, |
| "step": 805 |
| }, |
| { |
| "epoch": 3.263157894736842, |
| "grad_norm": 0.5482122497985201, |
| "learning_rate": 1.2997471016172188e-05, |
| "loss": 0.7322, |
| "step": 806 |
| }, |
| { |
| "epoch": 3.2672064777327936, |
| "grad_norm": 0.5694733583942042, |
| "learning_rate": 1.2944524510000042e-05, |
| "loss": 0.7026, |
| "step": 807 |
| }, |
| { |
| "epoch": 3.2712550607287447, |
| "grad_norm": 0.534101634813539, |
| "learning_rate": 1.2891634419221908e-05, |
| "loss": 0.7044, |
| "step": 808 |
| }, |
| { |
| "epoch": 3.2753036437246963, |
| "grad_norm": 0.5555740249913925, |
| "learning_rate": 1.2838801166745545e-05, |
| "loss": 0.7249, |
| "step": 809 |
| }, |
| { |
| "epoch": 3.279352226720648, |
| "grad_norm": 0.584594167110179, |
| "learning_rate": 1.2786025175024216e-05, |
| "loss": 0.7021, |
| "step": 810 |
| }, |
| { |
| "epoch": 3.283400809716599, |
| "grad_norm": 0.5304993545588648, |
| "learning_rate": 1.2733306866053357e-05, |
| "loss": 0.7276, |
| "step": 811 |
| }, |
| { |
| "epoch": 3.2874493927125505, |
| "grad_norm": 0.5956641873100074, |
| "learning_rate": 1.2680646661367163e-05, |
| "loss": 0.7091, |
| "step": 812 |
| }, |
| { |
| "epoch": 3.291497975708502, |
| "grad_norm": 0.552624639336083, |
| "learning_rate": 1.2628044982035219e-05, |
| "loss": 0.6998, |
| "step": 813 |
| }, |
| { |
| "epoch": 3.2955465587044532, |
| "grad_norm": 0.5338512569415875, |
| "learning_rate": 1.2575502248659155e-05, |
| "loss": 0.7078, |
| "step": 814 |
| }, |
| { |
| "epoch": 3.299595141700405, |
| "grad_norm": 0.5747462662076935, |
| "learning_rate": 1.2523018881369269e-05, |
| "loss": 0.6974, |
| "step": 815 |
| }, |
| { |
| "epoch": 3.3036437246963564, |
| "grad_norm": 0.5069464145006273, |
| "learning_rate": 1.2470595299821176e-05, |
| "loss": 0.6903, |
| "step": 816 |
| }, |
| { |
| "epoch": 3.3076923076923075, |
| "grad_norm": 0.5821633391514246, |
| "learning_rate": 1.2418231923192416e-05, |
| "loss": 0.72, |
| "step": 817 |
| }, |
| { |
| "epoch": 3.311740890688259, |
| "grad_norm": 0.5736407109685104, |
| "learning_rate": 1.2365929170179165e-05, |
| "loss": 0.7037, |
| "step": 818 |
| }, |
| { |
| "epoch": 3.3157894736842106, |
| "grad_norm": 0.5619817369418735, |
| "learning_rate": 1.2313687458992848e-05, |
| "loss": 0.7141, |
| "step": 819 |
| }, |
| { |
| "epoch": 3.3198380566801617, |
| "grad_norm": 0.5604447282024645, |
| "learning_rate": 1.2261507207356785e-05, |
| "loss": 0.711, |
| "step": 820 |
| }, |
| { |
| "epoch": 3.3238866396761133, |
| "grad_norm": 0.6543885237526503, |
| "learning_rate": 1.220938883250288e-05, |
| "loss": 0.7115, |
| "step": 821 |
| }, |
| { |
| "epoch": 3.327935222672065, |
| "grad_norm": 0.6188199973849695, |
| "learning_rate": 1.215733275116827e-05, |
| "loss": 0.7008, |
| "step": 822 |
| }, |
| { |
| "epoch": 3.331983805668016, |
| "grad_norm": 0.6787089626371516, |
| "learning_rate": 1.2105339379592005e-05, |
| "loss": 0.7237, |
| "step": 823 |
| }, |
| { |
| "epoch": 3.3360323886639676, |
| "grad_norm": 0.5533781019574001, |
| "learning_rate": 1.2053409133511681e-05, |
| "loss": 0.7359, |
| "step": 824 |
| }, |
| { |
| "epoch": 3.340080971659919, |
| "grad_norm": 0.641865839325486, |
| "learning_rate": 1.2001542428160173e-05, |
| "loss": 0.7163, |
| "step": 825 |
| }, |
| { |
| "epoch": 3.3441295546558703, |
| "grad_norm": 0.6371908023468423, |
| "learning_rate": 1.1949739678262278e-05, |
| "loss": 0.7207, |
| "step": 826 |
| }, |
| { |
| "epoch": 3.348178137651822, |
| "grad_norm": 0.5480173141095919, |
| "learning_rate": 1.189800129803139e-05, |
| "loss": 0.7114, |
| "step": 827 |
| }, |
| { |
| "epoch": 3.3522267206477734, |
| "grad_norm": 0.6297714206765229, |
| "learning_rate": 1.1846327701166228e-05, |
| "loss": 0.7226, |
| "step": 828 |
| }, |
| { |
| "epoch": 3.3562753036437245, |
| "grad_norm": 0.5548497793398668, |
| "learning_rate": 1.1794719300847493e-05, |
| "loss": 0.7011, |
| "step": 829 |
| }, |
| { |
| "epoch": 3.360323886639676, |
| "grad_norm": 0.5489100833854913, |
| "learning_rate": 1.1743176509734582e-05, |
| "loss": 0.7285, |
| "step": 830 |
| }, |
| { |
| "epoch": 3.3643724696356276, |
| "grad_norm": 0.5743588254416523, |
| "learning_rate": 1.1691699739962275e-05, |
| "loss": 0.7078, |
| "step": 831 |
| }, |
| { |
| "epoch": 3.3684210526315788, |
| "grad_norm": 0.5291060358992604, |
| "learning_rate": 1.1640289403137456e-05, |
| "loss": 0.7136, |
| "step": 832 |
| }, |
| { |
| "epoch": 3.3724696356275303, |
| "grad_norm": 0.5513836476382475, |
| "learning_rate": 1.1588945910335802e-05, |
| "loss": 0.7164, |
| "step": 833 |
| }, |
| { |
| "epoch": 3.376518218623482, |
| "grad_norm": 0.6276344191704487, |
| "learning_rate": 1.1537669672098517e-05, |
| "loss": 0.726, |
| "step": 834 |
| }, |
| { |
| "epoch": 3.380566801619433, |
| "grad_norm": 0.500561506211749, |
| "learning_rate": 1.1486461098429036e-05, |
| "loss": 0.7171, |
| "step": 835 |
| }, |
| { |
| "epoch": 3.3846153846153846, |
| "grad_norm": 0.6072958320403234, |
| "learning_rate": 1.1435320598789747e-05, |
| "loss": 0.6706, |
| "step": 836 |
| }, |
| { |
| "epoch": 3.388663967611336, |
| "grad_norm": 0.5361507214588519, |
| "learning_rate": 1.1384248582098736e-05, |
| "loss": 0.7337, |
| "step": 837 |
| }, |
| { |
| "epoch": 3.3927125506072873, |
| "grad_norm": 0.5404938955183634, |
| "learning_rate": 1.1333245456726467e-05, |
| "loss": 0.6654, |
| "step": 838 |
| }, |
| { |
| "epoch": 3.396761133603239, |
| "grad_norm": 0.5334155368007767, |
| "learning_rate": 1.128231163049258e-05, |
| "loss": 0.7022, |
| "step": 839 |
| }, |
| { |
| "epoch": 3.4008097165991904, |
| "grad_norm": 0.5226040792430436, |
| "learning_rate": 1.123144751066259e-05, |
| "loss": 0.7085, |
| "step": 840 |
| }, |
| { |
| "epoch": 3.4048582995951415, |
| "grad_norm": 0.548385532639748, |
| "learning_rate": 1.1180653503944657e-05, |
| "loss": 0.7144, |
| "step": 841 |
| }, |
| { |
| "epoch": 3.408906882591093, |
| "grad_norm": 0.49586339200840956, |
| "learning_rate": 1.1129930016486287e-05, |
| "loss": 0.7308, |
| "step": 842 |
| }, |
| { |
| "epoch": 3.4129554655870447, |
| "grad_norm": 0.668428778464208, |
| "learning_rate": 1.107927745387114e-05, |
| "loss": 0.7221, |
| "step": 843 |
| }, |
| { |
| "epoch": 3.417004048582996, |
| "grad_norm": 0.5393679557543677, |
| "learning_rate": 1.1028696221115766e-05, |
| "loss": 0.7012, |
| "step": 844 |
| }, |
| { |
| "epoch": 3.4210526315789473, |
| "grad_norm": 0.5467149218160963, |
| "learning_rate": 1.0978186722666341e-05, |
| "loss": 0.7241, |
| "step": 845 |
| }, |
| { |
| "epoch": 3.425101214574899, |
| "grad_norm": 0.6745667283940088, |
| "learning_rate": 1.0927749362395475e-05, |
| "loss": 0.7044, |
| "step": 846 |
| }, |
| { |
| "epoch": 3.42914979757085, |
| "grad_norm": 0.5810251183311901, |
| "learning_rate": 1.0877384543598954e-05, |
| "loss": 0.692, |
| "step": 847 |
| }, |
| { |
| "epoch": 3.4331983805668016, |
| "grad_norm": 0.5827326779634058, |
| "learning_rate": 1.0827092668992529e-05, |
| "loss": 0.7234, |
| "step": 848 |
| }, |
| { |
| "epoch": 3.437246963562753, |
| "grad_norm": 0.5585559422285465, |
| "learning_rate": 1.077687414070869e-05, |
| "loss": 0.6903, |
| "step": 849 |
| }, |
| { |
| "epoch": 3.4412955465587043, |
| "grad_norm": 0.5030990316067658, |
| "learning_rate": 1.0726729360293451e-05, |
| "loss": 0.6777, |
| "step": 850 |
| }, |
| { |
| "epoch": 3.445344129554656, |
| "grad_norm": 0.631028258259601, |
| "learning_rate": 1.0676658728703147e-05, |
| "loss": 0.7294, |
| "step": 851 |
| }, |
| { |
| "epoch": 3.4493927125506074, |
| "grad_norm": 0.5147869860410017, |
| "learning_rate": 1.0626662646301198e-05, |
| "loss": 0.7156, |
| "step": 852 |
| }, |
| { |
| "epoch": 3.4534412955465585, |
| "grad_norm": 0.5349434121712076, |
| "learning_rate": 1.0576741512854958e-05, |
| "loss": 0.7126, |
| "step": 853 |
| }, |
| { |
| "epoch": 3.45748987854251, |
| "grad_norm": 0.6319999129414904, |
| "learning_rate": 1.052689572753247e-05, |
| "loss": 0.7065, |
| "step": 854 |
| }, |
| { |
| "epoch": 3.4615384615384617, |
| "grad_norm": 0.5187455470129333, |
| "learning_rate": 1.0477125688899328e-05, |
| "loss": 0.7112, |
| "step": 855 |
| }, |
| { |
| "epoch": 3.465587044534413, |
| "grad_norm": 0.5821798110259399, |
| "learning_rate": 1.0427431794915406e-05, |
| "loss": 0.7003, |
| "step": 856 |
| }, |
| { |
| "epoch": 3.4696356275303644, |
| "grad_norm": 0.5440288428256166, |
| "learning_rate": 1.0377814442931769e-05, |
| "loss": 0.7124, |
| "step": 857 |
| }, |
| { |
| "epoch": 3.473684210526316, |
| "grad_norm": 0.5713602924793094, |
| "learning_rate": 1.0328274029687439e-05, |
| "loss": 0.7042, |
| "step": 858 |
| }, |
| { |
| "epoch": 3.477732793522267, |
| "grad_norm": 0.535233868339478, |
| "learning_rate": 1.0278810951306253e-05, |
| "loss": 0.7054, |
| "step": 859 |
| }, |
| { |
| "epoch": 3.4817813765182186, |
| "grad_norm": 0.5263213841005026, |
| "learning_rate": 1.0229425603293647e-05, |
| "loss": 0.7194, |
| "step": 860 |
| }, |
| { |
| "epoch": 3.48582995951417, |
| "grad_norm": 0.5334547906796416, |
| "learning_rate": 1.0180118380533558e-05, |
| "loss": 0.7069, |
| "step": 861 |
| }, |
| { |
| "epoch": 3.4898785425101213, |
| "grad_norm": 0.5472684881967639, |
| "learning_rate": 1.0130889677285225e-05, |
| "loss": 0.7115, |
| "step": 862 |
| }, |
| { |
| "epoch": 3.493927125506073, |
| "grad_norm": 0.527873540443911, |
| "learning_rate": 1.008173988718005e-05, |
| "loss": 0.7039, |
| "step": 863 |
| }, |
| { |
| "epoch": 3.4979757085020244, |
| "grad_norm": 0.6163794295585924, |
| "learning_rate": 1.0032669403218439e-05, |
| "loss": 0.6948, |
| "step": 864 |
| }, |
| { |
| "epoch": 3.5020242914979756, |
| "grad_norm": 0.5402286145810168, |
| "learning_rate": 9.983678617766677e-06, |
| "loss": 0.706, |
| "step": 865 |
| }, |
| { |
| "epoch": 3.506072874493927, |
| "grad_norm": 0.5892687247747453, |
| "learning_rate": 9.934767922553788e-06, |
| "loss": 0.7066, |
| "step": 866 |
| }, |
| { |
| "epoch": 3.5101214574898787, |
| "grad_norm": 0.6042856928938752, |
| "learning_rate": 9.885937708668371e-06, |
| "loss": 0.701, |
| "step": 867 |
| }, |
| { |
| "epoch": 3.51417004048583, |
| "grad_norm": 0.5361606383282979, |
| "learning_rate": 9.837188366555524e-06, |
| "loss": 0.7296, |
| "step": 868 |
| }, |
| { |
| "epoch": 3.5182186234817814, |
| "grad_norm": 0.5416266101368994, |
| "learning_rate": 9.788520286013691e-06, |
| "loss": 0.692, |
| "step": 869 |
| }, |
| { |
| "epoch": 3.522267206477733, |
| "grad_norm": 0.581393091676346, |
| "learning_rate": 9.739933856191537e-06, |
| "loss": 0.7067, |
| "step": 870 |
| }, |
| { |
| "epoch": 3.526315789473684, |
| "grad_norm": 0.536985434304934, |
| "learning_rate": 9.691429465584866e-06, |
| "loss": 0.6945, |
| "step": 871 |
| }, |
| { |
| "epoch": 3.5303643724696356, |
| "grad_norm": 0.5621814109409153, |
| "learning_rate": 9.643007502033492e-06, |
| "loss": 0.7127, |
| "step": 872 |
| }, |
| { |
| "epoch": 3.534412955465587, |
| "grad_norm": 0.5436900993713992, |
| "learning_rate": 9.594668352718152e-06, |
| "loss": 0.7105, |
| "step": 873 |
| }, |
| { |
| "epoch": 3.5384615384615383, |
| "grad_norm": 0.5769502264305841, |
| "learning_rate": 9.546412404157383e-06, |
| "loss": 0.7083, |
| "step": 874 |
| }, |
| { |
| "epoch": 3.54251012145749, |
| "grad_norm": 0.5601402173414914, |
| "learning_rate": 9.498240042204471e-06, |
| "loss": 0.7279, |
| "step": 875 |
| }, |
| { |
| "epoch": 3.5465587044534415, |
| "grad_norm": 0.5602393943425926, |
| "learning_rate": 9.450151652044339e-06, |
| "loss": 0.6965, |
| "step": 876 |
| }, |
| { |
| "epoch": 3.5506072874493926, |
| "grad_norm": 0.5231610232634745, |
| "learning_rate": 9.402147618190472e-06, |
| "loss": 0.7162, |
| "step": 877 |
| }, |
| { |
| "epoch": 3.554655870445344, |
| "grad_norm": 0.5525998206509369, |
| "learning_rate": 9.354228324481841e-06, |
| "loss": 0.7345, |
| "step": 878 |
| }, |
| { |
| "epoch": 3.5587044534412957, |
| "grad_norm": 0.6273153180141393, |
| "learning_rate": 9.306394154079852e-06, |
| "loss": 0.74, |
| "step": 879 |
| }, |
| { |
| "epoch": 3.562753036437247, |
| "grad_norm": 0.554594634543018, |
| "learning_rate": 9.258645489465254e-06, |
| "loss": 0.6972, |
| "step": 880 |
| }, |
| { |
| "epoch": 3.5668016194331984, |
| "grad_norm": 0.5428963098489195, |
| "learning_rate": 9.21098271243509e-06, |
| "loss": 0.7137, |
| "step": 881 |
| }, |
| { |
| "epoch": 3.57085020242915, |
| "grad_norm": 0.6447151734959374, |
| "learning_rate": 9.163406204099659e-06, |
| "loss": 0.7417, |
| "step": 882 |
| }, |
| { |
| "epoch": 3.574898785425101, |
| "grad_norm": 0.6136158799527647, |
| "learning_rate": 9.115916344879452e-06, |
| "loss": 0.7015, |
| "step": 883 |
| }, |
| { |
| "epoch": 3.5789473684210527, |
| "grad_norm": 0.5161291674804777, |
| "learning_rate": 9.068513514502126e-06, |
| "loss": 0.7029, |
| "step": 884 |
| }, |
| { |
| "epoch": 3.582995951417004, |
| "grad_norm": 0.5639314534205178, |
| "learning_rate": 9.021198091999439e-06, |
| "loss": 0.6836, |
| "step": 885 |
| }, |
| { |
| "epoch": 3.5870445344129553, |
| "grad_norm": 0.5924998916594758, |
| "learning_rate": 8.973970455704251e-06, |
| "loss": 0.7422, |
| "step": 886 |
| }, |
| { |
| "epoch": 3.591093117408907, |
| "grad_norm": 0.5357755109081249, |
| "learning_rate": 8.926830983247495e-06, |
| "loss": 0.6781, |
| "step": 887 |
| }, |
| { |
| "epoch": 3.5951417004048585, |
| "grad_norm": 0.489903533170762, |
| "learning_rate": 8.879780051555116e-06, |
| "loss": 0.7098, |
| "step": 888 |
| }, |
| { |
| "epoch": 3.5991902834008096, |
| "grad_norm": 0.6017401910591245, |
| "learning_rate": 8.83281803684512e-06, |
| "loss": 0.7068, |
| "step": 889 |
| }, |
| { |
| "epoch": 3.603238866396761, |
| "grad_norm": 0.5979210366717324, |
| "learning_rate": 8.785945314624532e-06, |
| "loss": 0.7212, |
| "step": 890 |
| }, |
| { |
| "epoch": 3.6072874493927127, |
| "grad_norm": 0.49845258437247153, |
| "learning_rate": 8.739162259686385e-06, |
| "loss": 0.7169, |
| "step": 891 |
| }, |
| { |
| "epoch": 3.611336032388664, |
| "grad_norm": 0.503354150021631, |
| "learning_rate": 8.692469246106734e-06, |
| "loss": 0.73, |
| "step": 892 |
| }, |
| { |
| "epoch": 3.6153846153846154, |
| "grad_norm": 0.6234452328579273, |
| "learning_rate": 8.645866647241677e-06, |
| "loss": 0.7338, |
| "step": 893 |
| }, |
| { |
| "epoch": 3.619433198380567, |
| "grad_norm": 0.61553571692953, |
| "learning_rate": 8.599354835724348e-06, |
| "loss": 0.6888, |
| "step": 894 |
| }, |
| { |
| "epoch": 3.623481781376518, |
| "grad_norm": 0.4994768417040188, |
| "learning_rate": 8.552934183461958e-06, |
| "loss": 0.7184, |
| "step": 895 |
| }, |
| { |
| "epoch": 3.6275303643724697, |
| "grad_norm": 0.5424966640602341, |
| "learning_rate": 8.506605061632805e-06, |
| "loss": 0.7034, |
| "step": 896 |
| }, |
| { |
| "epoch": 3.6315789473684212, |
| "grad_norm": 0.5567704853819222, |
| "learning_rate": 8.460367840683308e-06, |
| "loss": 0.6965, |
| "step": 897 |
| }, |
| { |
| "epoch": 3.6356275303643724, |
| "grad_norm": 0.5732175698982731, |
| "learning_rate": 8.414222890325065e-06, |
| "loss": 0.724, |
| "step": 898 |
| }, |
| { |
| "epoch": 3.639676113360324, |
| "grad_norm": 0.5083783706545664, |
| "learning_rate": 8.368170579531855e-06, |
| "loss": 0.7009, |
| "step": 899 |
| }, |
| { |
| "epoch": 3.6437246963562755, |
| "grad_norm": 0.5594254115062383, |
| "learning_rate": 8.322211276536733e-06, |
| "loss": 0.7147, |
| "step": 900 |
| }, |
| { |
| "epoch": 3.6477732793522266, |
| "grad_norm": 0.49991018704480106, |
| "learning_rate": 8.27634534882907e-06, |
| "loss": 0.7153, |
| "step": 901 |
| }, |
| { |
| "epoch": 3.651821862348178, |
| "grad_norm": 0.5196545092853145, |
| "learning_rate": 8.230573163151585e-06, |
| "loss": 0.7183, |
| "step": 902 |
| }, |
| { |
| "epoch": 3.6558704453441297, |
| "grad_norm": 0.5057358415171336, |
| "learning_rate": 8.184895085497466e-06, |
| "loss": 0.7254, |
| "step": 903 |
| }, |
| { |
| "epoch": 3.659919028340081, |
| "grad_norm": 0.5400906268633274, |
| "learning_rate": 8.139311481107401e-06, |
| "loss": 0.7262, |
| "step": 904 |
| }, |
| { |
| "epoch": 3.6639676113360324, |
| "grad_norm": 0.5453700865691128, |
| "learning_rate": 8.093822714466686e-06, |
| "loss": 0.7223, |
| "step": 905 |
| }, |
| { |
| "epoch": 3.668016194331984, |
| "grad_norm": 0.5199435881225414, |
| "learning_rate": 8.048429149302271e-06, |
| "loss": 0.7128, |
| "step": 906 |
| }, |
| { |
| "epoch": 3.672064777327935, |
| "grad_norm": 0.5123118971429416, |
| "learning_rate": 8.003131148579902e-06, |
| "loss": 0.7084, |
| "step": 907 |
| }, |
| { |
| "epoch": 3.6761133603238867, |
| "grad_norm": 0.5106368260410369, |
| "learning_rate": 7.957929074501182e-06, |
| "loss": 0.7017, |
| "step": 908 |
| }, |
| { |
| "epoch": 3.6801619433198383, |
| "grad_norm": 0.5041756318062497, |
| "learning_rate": 7.912823288500695e-06, |
| "loss": 0.7062, |
| "step": 909 |
| }, |
| { |
| "epoch": 3.6842105263157894, |
| "grad_norm": 0.4796980979810892, |
| "learning_rate": 7.867814151243102e-06, |
| "loss": 0.7038, |
| "step": 910 |
| }, |
| { |
| "epoch": 3.688259109311741, |
| "grad_norm": 0.49407400205803187, |
| "learning_rate": 7.822902022620267e-06, |
| "loss": 0.7183, |
| "step": 911 |
| }, |
| { |
| "epoch": 3.6923076923076925, |
| "grad_norm": 0.5243279049051345, |
| "learning_rate": 7.778087261748377e-06, |
| "loss": 0.719, |
| "step": 912 |
| }, |
| { |
| "epoch": 3.6963562753036436, |
| "grad_norm": 0.4804725754716955, |
| "learning_rate": 7.733370226965056e-06, |
| "loss": 0.6978, |
| "step": 913 |
| }, |
| { |
| "epoch": 3.700404858299595, |
| "grad_norm": 0.48439053238726965, |
| "learning_rate": 7.688751275826527e-06, |
| "loss": 0.7076, |
| "step": 914 |
| }, |
| { |
| "epoch": 3.7044534412955468, |
| "grad_norm": 0.49840850076975174, |
| "learning_rate": 7.644230765104735e-06, |
| "loss": 0.7101, |
| "step": 915 |
| }, |
| { |
| "epoch": 3.708502024291498, |
| "grad_norm": 0.5524842524505841, |
| "learning_rate": 7.599809050784502e-06, |
| "loss": 0.7148, |
| "step": 916 |
| }, |
| { |
| "epoch": 3.7125506072874495, |
| "grad_norm": 0.48632504989818903, |
| "learning_rate": 7.555486488060661e-06, |
| "loss": 0.7216, |
| "step": 917 |
| }, |
| { |
| "epoch": 3.716599190283401, |
| "grad_norm": 0.5269840700674241, |
| "learning_rate": 7.511263431335249e-06, |
| "loss": 0.7018, |
| "step": 918 |
| }, |
| { |
| "epoch": 3.720647773279352, |
| "grad_norm": 0.4973705671397241, |
| "learning_rate": 7.467140234214656e-06, |
| "loss": 0.6783, |
| "step": 919 |
| }, |
| { |
| "epoch": 3.7246963562753037, |
| "grad_norm": 0.536867279406637, |
| "learning_rate": 7.423117249506777e-06, |
| "loss": 0.7132, |
| "step": 920 |
| }, |
| { |
| "epoch": 3.7287449392712553, |
| "grad_norm": 0.5625974771905498, |
| "learning_rate": 7.379194829218231e-06, |
| "loss": 0.7432, |
| "step": 921 |
| }, |
| { |
| "epoch": 3.7327935222672064, |
| "grad_norm": 0.48764665633739834, |
| "learning_rate": 7.335373324551518e-06, |
| "loss": 0.7106, |
| "step": 922 |
| }, |
| { |
| "epoch": 3.736842105263158, |
| "grad_norm": 0.5012560392963378, |
| "learning_rate": 7.291653085902224e-06, |
| "loss": 0.7096, |
| "step": 923 |
| }, |
| { |
| "epoch": 3.7408906882591095, |
| "grad_norm": 0.4829262970181347, |
| "learning_rate": 7.248034462856213e-06, |
| "loss": 0.7195, |
| "step": 924 |
| }, |
| { |
| "epoch": 3.7449392712550607, |
| "grad_norm": 0.510042409829087, |
| "learning_rate": 7.20451780418683e-06, |
| "loss": 0.6822, |
| "step": 925 |
| }, |
| { |
| "epoch": 3.748987854251012, |
| "grad_norm": 0.5048727240837187, |
| "learning_rate": 7.161103457852128e-06, |
| "loss": 0.7477, |
| "step": 926 |
| }, |
| { |
| "epoch": 3.753036437246964, |
| "grad_norm": 0.4917025313590476, |
| "learning_rate": 7.117791770992046e-06, |
| "loss": 0.7102, |
| "step": 927 |
| }, |
| { |
| "epoch": 3.757085020242915, |
| "grad_norm": 0.46379669193830425, |
| "learning_rate": 7.074583089925688e-06, |
| "loss": 0.7133, |
| "step": 928 |
| }, |
| { |
| "epoch": 3.7611336032388665, |
| "grad_norm": 0.53131495134438, |
| "learning_rate": 7.0314777601485155e-06, |
| "loss": 0.7232, |
| "step": 929 |
| }, |
| { |
| "epoch": 3.765182186234818, |
| "grad_norm": 0.4916773126666677, |
| "learning_rate": 6.988476126329604e-06, |
| "loss": 0.7144, |
| "step": 930 |
| }, |
| { |
| "epoch": 3.769230769230769, |
| "grad_norm": 0.4673355150096268, |
| "learning_rate": 6.945578532308857e-06, |
| "loss": 0.7171, |
| "step": 931 |
| }, |
| { |
| "epoch": 3.7732793522267207, |
| "grad_norm": 0.5298953786837687, |
| "learning_rate": 6.902785321094301e-06, |
| "loss": 0.7199, |
| "step": 932 |
| }, |
| { |
| "epoch": 3.7773279352226723, |
| "grad_norm": 0.4677169823097187, |
| "learning_rate": 6.8600968348593115e-06, |
| "loss": 0.6997, |
| "step": 933 |
| }, |
| { |
| "epoch": 3.7813765182186234, |
| "grad_norm": 0.4813420937138145, |
| "learning_rate": 6.817513414939892e-06, |
| "loss": 0.7049, |
| "step": 934 |
| }, |
| { |
| "epoch": 3.785425101214575, |
| "grad_norm": 0.5129273008856562, |
| "learning_rate": 6.775035401831924e-06, |
| "loss": 0.7264, |
| "step": 935 |
| }, |
| { |
| "epoch": 3.7894736842105265, |
| "grad_norm": 0.4824133781199396, |
| "learning_rate": 6.732663135188473e-06, |
| "loss": 0.7199, |
| "step": 936 |
| }, |
| { |
| "epoch": 3.7935222672064777, |
| "grad_norm": 0.49848696200930287, |
| "learning_rate": 6.690396953817055e-06, |
| "loss": 0.7026, |
| "step": 937 |
| }, |
| { |
| "epoch": 3.7975708502024292, |
| "grad_norm": 0.4730055919800572, |
| "learning_rate": 6.648237195676927e-06, |
| "loss": 0.6976, |
| "step": 938 |
| }, |
| { |
| "epoch": 3.801619433198381, |
| "grad_norm": 0.47858958061279816, |
| "learning_rate": 6.606184197876395e-06, |
| "loss": 0.7213, |
| "step": 939 |
| }, |
| { |
| "epoch": 3.805668016194332, |
| "grad_norm": 0.5297134119305936, |
| "learning_rate": 6.564238296670109e-06, |
| "loss": 0.6993, |
| "step": 940 |
| }, |
| { |
| "epoch": 3.8097165991902835, |
| "grad_norm": 0.5549468953232043, |
| "learning_rate": 6.52239982745638e-06, |
| "loss": 0.7079, |
| "step": 941 |
| }, |
| { |
| "epoch": 3.813765182186235, |
| "grad_norm": 0.4752678556802537, |
| "learning_rate": 6.480669124774484e-06, |
| "loss": 0.6971, |
| "step": 942 |
| }, |
| { |
| "epoch": 3.817813765182186, |
| "grad_norm": 0.46146273132160354, |
| "learning_rate": 6.439046522302013e-06, |
| "loss": 0.6912, |
| "step": 943 |
| }, |
| { |
| "epoch": 3.8218623481781377, |
| "grad_norm": 0.47209613228706043, |
| "learning_rate": 6.397532352852191e-06, |
| "loss": 0.7376, |
| "step": 944 |
| }, |
| { |
| "epoch": 3.8259109311740893, |
| "grad_norm": 0.489408721908986, |
| "learning_rate": 6.356126948371202e-06, |
| "loss": 0.6895, |
| "step": 945 |
| }, |
| { |
| "epoch": 3.8299595141700404, |
| "grad_norm": 0.5330663090109188, |
| "learning_rate": 6.314830639935559e-06, |
| "loss": 0.7215, |
| "step": 946 |
| }, |
| { |
| "epoch": 3.834008097165992, |
| "grad_norm": 0.46009830933199153, |
| "learning_rate": 6.273643757749448e-06, |
| "loss": 0.6928, |
| "step": 947 |
| }, |
| { |
| "epoch": 3.8380566801619436, |
| "grad_norm": 0.48556032622601814, |
| "learning_rate": 6.2325666311420856e-06, |
| "loss": 0.7307, |
| "step": 948 |
| }, |
| { |
| "epoch": 3.8421052631578947, |
| "grad_norm": 0.48761484946414063, |
| "learning_rate": 6.1915995885650675e-06, |
| "loss": 0.7206, |
| "step": 949 |
| }, |
| { |
| "epoch": 3.8461538461538463, |
| "grad_norm": 0.5054257774676378, |
| "learning_rate": 6.150742957589784e-06, |
| "loss": 0.7018, |
| "step": 950 |
| }, |
| { |
| "epoch": 3.850202429149798, |
| "grad_norm": 0.5113473663263283, |
| "learning_rate": 6.109997064904774e-06, |
| "loss": 0.7306, |
| "step": 951 |
| }, |
| { |
| "epoch": 3.854251012145749, |
| "grad_norm": 0.4904166565434664, |
| "learning_rate": 6.069362236313092e-06, |
| "loss": 0.6973, |
| "step": 952 |
| }, |
| { |
| "epoch": 3.8582995951417005, |
| "grad_norm": 0.501991670209039, |
| "learning_rate": 6.0288387967297565e-06, |
| "loss": 0.6853, |
| "step": 953 |
| }, |
| { |
| "epoch": 3.862348178137652, |
| "grad_norm": 0.4637909905590731, |
| "learning_rate": 5.988427070179108e-06, |
| "loss": 0.7194, |
| "step": 954 |
| }, |
| { |
| "epoch": 3.866396761133603, |
| "grad_norm": 0.5340147586029127, |
| "learning_rate": 5.948127379792234e-06, |
| "loss": 0.7144, |
| "step": 955 |
| }, |
| { |
| "epoch": 3.8704453441295548, |
| "grad_norm": 0.49652879161614655, |
| "learning_rate": 5.90794004780439e-06, |
| "loss": 0.7174, |
| "step": 956 |
| }, |
| { |
| "epoch": 3.8744939271255063, |
| "grad_norm": 0.4938667595160264, |
| "learning_rate": 5.867865395552405e-06, |
| "loss": 0.7002, |
| "step": 957 |
| }, |
| { |
| "epoch": 3.8785425101214575, |
| "grad_norm": 0.48665694599831827, |
| "learning_rate": 5.8279037434721364e-06, |
| "loss": 0.6875, |
| "step": 958 |
| }, |
| { |
| "epoch": 3.882591093117409, |
| "grad_norm": 0.5072044708126234, |
| "learning_rate": 5.788055411095892e-06, |
| "loss": 0.7187, |
| "step": 959 |
| }, |
| { |
| "epoch": 3.8866396761133606, |
| "grad_norm": 0.5012134523625346, |
| "learning_rate": 5.7483207170498625e-06, |
| "loss": 0.7068, |
| "step": 960 |
| }, |
| { |
| "epoch": 3.8906882591093117, |
| "grad_norm": 0.4780202615921947, |
| "learning_rate": 5.708699979051613e-06, |
| "loss": 0.6984, |
| "step": 961 |
| }, |
| { |
| "epoch": 3.8947368421052633, |
| "grad_norm": 0.499097548704152, |
| "learning_rate": 5.66919351390751e-06, |
| "loss": 0.7188, |
| "step": 962 |
| }, |
| { |
| "epoch": 3.898785425101215, |
| "grad_norm": 0.49871330556582105, |
| "learning_rate": 5.629801637510188e-06, |
| "loss": 0.6982, |
| "step": 963 |
| }, |
| { |
| "epoch": 3.902834008097166, |
| "grad_norm": 0.47635354271100805, |
| "learning_rate": 5.590524664836043e-06, |
| "loss": 0.718, |
| "step": 964 |
| }, |
| { |
| "epoch": 3.9068825910931175, |
| "grad_norm": 0.5209069281160695, |
| "learning_rate": 5.551362909942708e-06, |
| "loss": 0.7206, |
| "step": 965 |
| }, |
| { |
| "epoch": 3.910931174089069, |
| "grad_norm": 0.4910437384854514, |
| "learning_rate": 5.51231668596653e-06, |
| "loss": 0.7334, |
| "step": 966 |
| }, |
| { |
| "epoch": 3.91497975708502, |
| "grad_norm": 0.47093772704990045, |
| "learning_rate": 5.47338630512007e-06, |
| "loss": 0.6868, |
| "step": 967 |
| }, |
| { |
| "epoch": 3.919028340080972, |
| "grad_norm": 0.4908712034769523, |
| "learning_rate": 5.434572078689617e-06, |
| "loss": 0.7073, |
| "step": 968 |
| }, |
| { |
| "epoch": 3.9230769230769234, |
| "grad_norm": 0.4713083161137284, |
| "learning_rate": 5.395874317032694e-06, |
| "loss": 0.7126, |
| "step": 969 |
| }, |
| { |
| "epoch": 3.9271255060728745, |
| "grad_norm": 0.45847359985534597, |
| "learning_rate": 5.357293329575573e-06, |
| "loss": 0.7024, |
| "step": 970 |
| }, |
| { |
| "epoch": 3.931174089068826, |
| "grad_norm": 0.4649773256188797, |
| "learning_rate": 5.318829424810797e-06, |
| "loss": 0.7082, |
| "step": 971 |
| }, |
| { |
| "epoch": 3.9352226720647776, |
| "grad_norm": 0.47567735387325644, |
| "learning_rate": 5.280482910294728e-06, |
| "loss": 0.7158, |
| "step": 972 |
| }, |
| { |
| "epoch": 3.9392712550607287, |
| "grad_norm": 0.4913543822887899, |
| "learning_rate": 5.242254092645078e-06, |
| "loss": 0.7027, |
| "step": 973 |
| }, |
| { |
| "epoch": 3.9433198380566803, |
| "grad_norm": 0.4674943846259301, |
| "learning_rate": 5.2041432775384445e-06, |
| "loss": 0.7173, |
| "step": 974 |
| }, |
| { |
| "epoch": 3.9473684210526314, |
| "grad_norm": 0.45935002005650555, |
| "learning_rate": 5.166150769707892e-06, |
| "loss": 0.7229, |
| "step": 975 |
| }, |
| { |
| "epoch": 3.951417004048583, |
| "grad_norm": 0.46373404509048366, |
| "learning_rate": 5.128276872940503e-06, |
| "loss": 0.691, |
| "step": 976 |
| }, |
| { |
| "epoch": 3.9554655870445345, |
| "grad_norm": 0.5067244745609577, |
| "learning_rate": 5.090521890074953e-06, |
| "loss": 0.7102, |
| "step": 977 |
| }, |
| { |
| "epoch": 3.9595141700404857, |
| "grad_norm": 0.4727908613481245, |
| "learning_rate": 5.0528861229990675e-06, |
| "loss": 0.6986, |
| "step": 978 |
| }, |
| { |
| "epoch": 3.9635627530364372, |
| "grad_norm": 0.5128330850075712, |
| "learning_rate": 5.015369872647442e-06, |
| "loss": 0.7254, |
| "step": 979 |
| }, |
| { |
| "epoch": 3.967611336032389, |
| "grad_norm": 0.4785472644114532, |
| "learning_rate": 4.9779734389990195e-06, |
| "loss": 0.7302, |
| "step": 980 |
| }, |
| { |
| "epoch": 3.97165991902834, |
| "grad_norm": 0.4873963849549054, |
| "learning_rate": 4.940697121074675e-06, |
| "loss": 0.7135, |
| "step": 981 |
| }, |
| { |
| "epoch": 3.9757085020242915, |
| "grad_norm": 0.4807472681597943, |
| "learning_rate": 4.903541216934864e-06, |
| "loss": 0.6892, |
| "step": 982 |
| }, |
| { |
| "epoch": 3.979757085020243, |
| "grad_norm": 0.4545440035039379, |
| "learning_rate": 4.8665060236772e-06, |
| "loss": 0.7196, |
| "step": 983 |
| }, |
| { |
| "epoch": 3.983805668016194, |
| "grad_norm": 0.45403181145342286, |
| "learning_rate": 4.829591837434102e-06, |
| "loss": 0.7166, |
| "step": 984 |
| }, |
| { |
| "epoch": 3.9878542510121457, |
| "grad_norm": 0.48636735993555114, |
| "learning_rate": 4.792798953370424e-06, |
| "loss": 0.6813, |
| "step": 985 |
| }, |
| { |
| "epoch": 3.9919028340080973, |
| "grad_norm": 0.503953109428849, |
| "learning_rate": 4.756127665681083e-06, |
| "loss": 0.714, |
| "step": 986 |
| }, |
| { |
| "epoch": 3.9959514170040484, |
| "grad_norm": 0.4966547976570261, |
| "learning_rate": 4.719578267588725e-06, |
| "loss": 0.7143, |
| "step": 987 |
| }, |
| { |
| "epoch": 4.0, |
| "grad_norm": 0.46466442321000284, |
| "learning_rate": 4.6831510513413525e-06, |
| "loss": 0.6957, |
| "step": 988 |
| }, |
| { |
| "epoch": 4.004048582995951, |
| "grad_norm": 0.9582839892660946, |
| "learning_rate": 4.6468463082100176e-06, |
| "loss": 0.567, |
| "step": 989 |
| }, |
| { |
| "epoch": 4.008097165991903, |
| "grad_norm": 0.7865771275261273, |
| "learning_rate": 4.610664328486485e-06, |
| "loss": 0.5666, |
| "step": 990 |
| }, |
| { |
| "epoch": 4.012145748987854, |
| "grad_norm": 0.5612879949592661, |
| "learning_rate": 4.5746054014809e-06, |
| "loss": 0.5856, |
| "step": 991 |
| }, |
| { |
| "epoch": 4.016194331983805, |
| "grad_norm": 0.771285177817603, |
| "learning_rate": 4.5386698155194766e-06, |
| "loss": 0.5785, |
| "step": 992 |
| }, |
| { |
| "epoch": 4.020242914979757, |
| "grad_norm": 1.1213335805169666, |
| "learning_rate": 4.502857857942204e-06, |
| "loss": 0.5336, |
| "step": 993 |
| }, |
| { |
| "epoch": 4.0242914979757085, |
| "grad_norm": 0.9930192403788248, |
| "learning_rate": 4.467169815100549e-06, |
| "loss": 0.539, |
| "step": 994 |
| }, |
| { |
| "epoch": 4.02834008097166, |
| "grad_norm": 0.8237973634834632, |
| "learning_rate": 4.431605972355137e-06, |
| "loss": 0.5549, |
| "step": 995 |
| }, |
| { |
| "epoch": 4.032388663967612, |
| "grad_norm": 0.8180526592531201, |
| "learning_rate": 4.396166614073516e-06, |
| "loss": 0.5427, |
| "step": 996 |
| }, |
| { |
| "epoch": 4.036437246963563, |
| "grad_norm": 0.8415892120504785, |
| "learning_rate": 4.360852023627846e-06, |
| "loss": 0.5537, |
| "step": 997 |
| }, |
| { |
| "epoch": 4.040485829959514, |
| "grad_norm": 0.8594482657791296, |
| "learning_rate": 4.325662483392656e-06, |
| "loss": 0.5439, |
| "step": 998 |
| }, |
| { |
| "epoch": 4.044534412955466, |
| "grad_norm": 0.7802470233353487, |
| "learning_rate": 4.29059827474257e-06, |
| "loss": 0.5425, |
| "step": 999 |
| }, |
| { |
| "epoch": 4.048582995951417, |
| "grad_norm": 0.6858534579065132, |
| "learning_rate": 4.255659678050063e-06, |
| "loss": 0.5447, |
| "step": 1000 |
| }, |
| { |
| "epoch": 4.052631578947368, |
| "grad_norm": 0.6279546009319623, |
| "learning_rate": 4.22084697268323e-06, |
| "loss": 0.5638, |
| "step": 1001 |
| }, |
| { |
| "epoch": 4.05668016194332, |
| "grad_norm": 0.695778420811024, |
| "learning_rate": 4.186160437003535e-06, |
| "loss": 0.5528, |
| "step": 1002 |
| }, |
| { |
| "epoch": 4.060728744939271, |
| "grad_norm": 0.7694838284792949, |
| "learning_rate": 4.151600348363582e-06, |
| "loss": 0.5615, |
| "step": 1003 |
| }, |
| { |
| "epoch": 4.064777327935222, |
| "grad_norm": 0.7109914670337336, |
| "learning_rate": 4.11716698310493e-06, |
| "loss": 0.5757, |
| "step": 1004 |
| }, |
| { |
| "epoch": 4.068825910931174, |
| "grad_norm": 0.5774468020189376, |
| "learning_rate": 4.082860616555848e-06, |
| "loss": 0.5267, |
| "step": 1005 |
| }, |
| { |
| "epoch": 4.0728744939271255, |
| "grad_norm": 0.5836584596846526, |
| "learning_rate": 4.048681523029125e-06, |
| "loss": 0.5853, |
| "step": 1006 |
| }, |
| { |
| "epoch": 4.076923076923077, |
| "grad_norm": 0.5934681776088001, |
| "learning_rate": 4.014629975819882e-06, |
| "loss": 0.5582, |
| "step": 1007 |
| }, |
| { |
| "epoch": 4.080971659919029, |
| "grad_norm": 0.6067871110987149, |
| "learning_rate": 3.9807062472033855e-06, |
| "loss": 0.5336, |
| "step": 1008 |
| }, |
| { |
| "epoch": 4.08502024291498, |
| "grad_norm": 0.6326886472019508, |
| "learning_rate": 3.94691060843287e-06, |
| "loss": 0.5702, |
| "step": 1009 |
| }, |
| { |
| "epoch": 4.089068825910931, |
| "grad_norm": 0.5996516721475017, |
| "learning_rate": 3.913243329737353e-06, |
| "loss": 0.5795, |
| "step": 1010 |
| }, |
| { |
| "epoch": 4.093117408906883, |
| "grad_norm": 0.5467278220443119, |
| "learning_rate": 3.879704680319503e-06, |
| "loss": 0.5596, |
| "step": 1011 |
| }, |
| { |
| "epoch": 4.097165991902834, |
| "grad_norm": 0.5736375349699847, |
| "learning_rate": 3.846294928353471e-06, |
| "loss": 0.5748, |
| "step": 1012 |
| }, |
| { |
| "epoch": 4.101214574898785, |
| "grad_norm": 0.5328497807699406, |
| "learning_rate": 3.8130143409827345e-06, |
| "loss": 0.5566, |
| "step": 1013 |
| }, |
| { |
| "epoch": 4.105263157894737, |
| "grad_norm": 0.5424519646041318, |
| "learning_rate": 3.7798631843179887e-06, |
| "loss": 0.5585, |
| "step": 1014 |
| }, |
| { |
| "epoch": 4.109311740890688, |
| "grad_norm": 0.5202186060227175, |
| "learning_rate": 3.7468417234349973e-06, |
| "loss": 0.551, |
| "step": 1015 |
| }, |
| { |
| "epoch": 4.113360323886639, |
| "grad_norm": 0.5361866902108628, |
| "learning_rate": 3.7139502223724823e-06, |
| "loss": 0.5728, |
| "step": 1016 |
| }, |
| { |
| "epoch": 4.117408906882591, |
| "grad_norm": 0.5286620953515059, |
| "learning_rate": 3.6811889441300095e-06, |
| "loss": 0.5526, |
| "step": 1017 |
| }, |
| { |
| "epoch": 4.1214574898785425, |
| "grad_norm": 0.5073361964988735, |
| "learning_rate": 3.648558150665886e-06, |
| "loss": 0.5373, |
| "step": 1018 |
| }, |
| { |
| "epoch": 4.125506072874494, |
| "grad_norm": 0.5147677246550312, |
| "learning_rate": 3.616058102895075e-06, |
| "loss": 0.5633, |
| "step": 1019 |
| }, |
| { |
| "epoch": 4.129554655870446, |
| "grad_norm": 0.5204274640842496, |
| "learning_rate": 3.583689060687079e-06, |
| "loss": 0.5788, |
| "step": 1020 |
| }, |
| { |
| "epoch": 4.133603238866397, |
| "grad_norm": 0.4985762178404418, |
| "learning_rate": 3.5514512828639023e-06, |
| "loss": 0.5502, |
| "step": 1021 |
| }, |
| { |
| "epoch": 4.137651821862348, |
| "grad_norm": 0.5157097627659819, |
| "learning_rate": 3.51934502719796e-06, |
| "loss": 0.5458, |
| "step": 1022 |
| }, |
| { |
| "epoch": 4.1417004048583, |
| "grad_norm": 0.5037053715866284, |
| "learning_rate": 3.4873705504100185e-06, |
| "loss": 0.5658, |
| "step": 1023 |
| }, |
| { |
| "epoch": 4.145748987854251, |
| "grad_norm": 0.48658101761598416, |
| "learning_rate": 3.4555281081671345e-06, |
| "loss": 0.5253, |
| "step": 1024 |
| }, |
| { |
| "epoch": 4.149797570850202, |
| "grad_norm": 0.4899992281263155, |
| "learning_rate": 3.4238179550806326e-06, |
| "loss": 0.5458, |
| "step": 1025 |
| }, |
| { |
| "epoch": 4.153846153846154, |
| "grad_norm": 0.4765515401060497, |
| "learning_rate": 3.3922403447040497e-06, |
| "loss": 0.5529, |
| "step": 1026 |
| }, |
| { |
| "epoch": 4.157894736842105, |
| "grad_norm": 0.47526955250721753, |
| "learning_rate": 3.3607955295311247e-06, |
| "loss": 0.5489, |
| "step": 1027 |
| }, |
| { |
| "epoch": 4.161943319838056, |
| "grad_norm": 0.4973414430994903, |
| "learning_rate": 3.32948376099375e-06, |
| "loss": 0.5491, |
| "step": 1028 |
| }, |
| { |
| "epoch": 4.165991902834008, |
| "grad_norm": 0.5260522939374747, |
| "learning_rate": 3.2983052894599976e-06, |
| "loss": 0.5526, |
| "step": 1029 |
| }, |
| { |
| "epoch": 4.17004048582996, |
| "grad_norm": 0.46896298464806235, |
| "learning_rate": 3.2672603642320945e-06, |
| "loss": 0.5264, |
| "step": 1030 |
| }, |
| { |
| "epoch": 4.174089068825911, |
| "grad_norm": 0.48205616732071066, |
| "learning_rate": 3.236349233544438e-06, |
| "loss": 0.5646, |
| "step": 1031 |
| }, |
| { |
| "epoch": 4.178137651821863, |
| "grad_norm": 0.5009506599761363, |
| "learning_rate": 3.205572144561604e-06, |
| "loss": 0.5592, |
| "step": 1032 |
| }, |
| { |
| "epoch": 4.182186234817814, |
| "grad_norm": 0.4935786052744522, |
| "learning_rate": 3.174929343376374e-06, |
| "loss": 0.5425, |
| "step": 1033 |
| }, |
| { |
| "epoch": 4.186234817813765, |
| "grad_norm": 0.4815224604736771, |
| "learning_rate": 3.1444210750077777e-06, |
| "loss": 0.5559, |
| "step": 1034 |
| }, |
| { |
| "epoch": 4.190283400809717, |
| "grad_norm": 0.48320804148655905, |
| "learning_rate": 3.1140475833991046e-06, |
| "loss": 0.5672, |
| "step": 1035 |
| }, |
| { |
| "epoch": 4.194331983805668, |
| "grad_norm": 0.46807795577589417, |
| "learning_rate": 3.0838091114159895e-06, |
| "loss": 0.5556, |
| "step": 1036 |
| }, |
| { |
| "epoch": 4.198380566801619, |
| "grad_norm": 0.47579815828401206, |
| "learning_rate": 3.0537059008444524e-06, |
| "loss": 0.556, |
| "step": 1037 |
| }, |
| { |
| "epoch": 4.202429149797571, |
| "grad_norm": 0.49546400580748595, |
| "learning_rate": 3.0237381923889563e-06, |
| "loss": 0.5598, |
| "step": 1038 |
| }, |
| { |
| "epoch": 4.206477732793522, |
| "grad_norm": 0.501670325799055, |
| "learning_rate": 2.9939062256704997e-06, |
| "loss": 0.554, |
| "step": 1039 |
| }, |
| { |
| "epoch": 4.2105263157894735, |
| "grad_norm": 0.47835079211863674, |
| "learning_rate": 2.964210239224696e-06, |
| "loss": 0.5306, |
| "step": 1040 |
| }, |
| { |
| "epoch": 4.2145748987854255, |
| "grad_norm": 0.4956678407221014, |
| "learning_rate": 2.934650470499869e-06, |
| "loss": 0.5485, |
| "step": 1041 |
| }, |
| { |
| "epoch": 4.218623481781377, |
| "grad_norm": 0.4724047488120582, |
| "learning_rate": 2.905227155855128e-06, |
| "loss": 0.5542, |
| "step": 1042 |
| }, |
| { |
| "epoch": 4.222672064777328, |
| "grad_norm": 0.4853030261305805, |
| "learning_rate": 2.875940530558519e-06, |
| "loss": 0.5337, |
| "step": 1043 |
| }, |
| { |
| "epoch": 4.22672064777328, |
| "grad_norm": 0.4814475628218854, |
| "learning_rate": 2.8467908287851154e-06, |
| "loss": 0.5633, |
| "step": 1044 |
| }, |
| { |
| "epoch": 4.230769230769231, |
| "grad_norm": 0.4782849494974031, |
| "learning_rate": 2.8177782836151603e-06, |
| "loss": 0.5544, |
| "step": 1045 |
| }, |
| { |
| "epoch": 4.234817813765182, |
| "grad_norm": 0.4771702577481492, |
| "learning_rate": 2.7889031270321874e-06, |
| "loss": 0.5423, |
| "step": 1046 |
| }, |
| { |
| "epoch": 4.238866396761134, |
| "grad_norm": 0.4609695350931223, |
| "learning_rate": 2.760165589921184e-06, |
| "loss": 0.5673, |
| "step": 1047 |
| }, |
| { |
| "epoch": 4.242914979757085, |
| "grad_norm": 0.4933693035304753, |
| "learning_rate": 2.7315659020667305e-06, |
| "loss": 0.5479, |
| "step": 1048 |
| }, |
| { |
| "epoch": 4.246963562753036, |
| "grad_norm": 0.46642781603496686, |
| "learning_rate": 2.703104292151164e-06, |
| "loss": 0.5432, |
| "step": 1049 |
| }, |
| { |
| "epoch": 4.251012145748988, |
| "grad_norm": 0.4663114785217346, |
| "learning_rate": 2.6747809877527607e-06, |
| "loss": 0.5434, |
| "step": 1050 |
| }, |
| { |
| "epoch": 4.255060728744939, |
| "grad_norm": 0.4728665258098779, |
| "learning_rate": 2.6465962153439082e-06, |
| "loss": 0.5538, |
| "step": 1051 |
| }, |
| { |
| "epoch": 4.2591093117408905, |
| "grad_norm": 0.470792115596792, |
| "learning_rate": 2.618550200289303e-06, |
| "loss": 0.5584, |
| "step": 1052 |
| }, |
| { |
| "epoch": 4.2631578947368425, |
| "grad_norm": 0.47929167969072173, |
| "learning_rate": 2.59064316684412e-06, |
| "loss": 0.5527, |
| "step": 1053 |
| }, |
| { |
| "epoch": 4.267206477732794, |
| "grad_norm": 0.4766865368166975, |
| "learning_rate": 2.562875338152264e-06, |
| "loss": 0.5461, |
| "step": 1054 |
| }, |
| { |
| "epoch": 4.271255060728745, |
| "grad_norm": 0.4768150974901715, |
| "learning_rate": 2.5352469362445552e-06, |
| "loss": 0.5396, |
| "step": 1055 |
| }, |
| { |
| "epoch": 4.275303643724697, |
| "grad_norm": 0.47829506728691434, |
| "learning_rate": 2.5077581820369523e-06, |
| "loss": 0.5225, |
| "step": 1056 |
| }, |
| { |
| "epoch": 4.279352226720648, |
| "grad_norm": 0.470902768279911, |
| "learning_rate": 2.480409295328803e-06, |
| "loss": 0.522, |
| "step": 1057 |
| }, |
| { |
| "epoch": 4.283400809716599, |
| "grad_norm": 0.4869466450988195, |
| "learning_rate": 2.45320049480108e-06, |
| "loss": 0.5369, |
| "step": 1058 |
| }, |
| { |
| "epoch": 4.287449392712551, |
| "grad_norm": 0.4873059759032473, |
| "learning_rate": 2.4261319980146293e-06, |
| "loss": 0.5756, |
| "step": 1059 |
| }, |
| { |
| "epoch": 4.291497975708502, |
| "grad_norm": 0.4873847840313206, |
| "learning_rate": 2.399204021408421e-06, |
| "loss": 0.54, |
| "step": 1060 |
| }, |
| { |
| "epoch": 4.295546558704453, |
| "grad_norm": 0.46193545771983624, |
| "learning_rate": 2.3724167802978395e-06, |
| "loss": 0.5722, |
| "step": 1061 |
| }, |
| { |
| "epoch": 4.299595141700405, |
| "grad_norm": 0.4951839147409792, |
| "learning_rate": 2.3457704888729625e-06, |
| "loss": 0.5816, |
| "step": 1062 |
| }, |
| { |
| "epoch": 4.303643724696356, |
| "grad_norm": 0.47472865114427165, |
| "learning_rate": 2.3192653601968162e-06, |
| "loss": 0.5423, |
| "step": 1063 |
| }, |
| { |
| "epoch": 4.3076923076923075, |
| "grad_norm": 0.4730504805591687, |
| "learning_rate": 2.2929016062037057e-06, |
| "loss": 0.554, |
| "step": 1064 |
| }, |
| { |
| "epoch": 4.3117408906882595, |
| "grad_norm": 0.4790436031504865, |
| "learning_rate": 2.2666794376975055e-06, |
| "loss": 0.5553, |
| "step": 1065 |
| }, |
| { |
| "epoch": 4.315789473684211, |
| "grad_norm": 0.5014024942990211, |
| "learning_rate": 2.24059906434998e-06, |
| "loss": 0.5788, |
| "step": 1066 |
| }, |
| { |
| "epoch": 4.319838056680162, |
| "grad_norm": 0.4833286426224097, |
| "learning_rate": 2.214660694699091e-06, |
| "loss": 0.5811, |
| "step": 1067 |
| }, |
| { |
| "epoch": 4.323886639676114, |
| "grad_norm": 0.47692404477588035, |
| "learning_rate": 2.188864536147355e-06, |
| "loss": 0.5564, |
| "step": 1068 |
| }, |
| { |
| "epoch": 4.327935222672065, |
| "grad_norm": 0.4772202319030718, |
| "learning_rate": 2.1632107949601757e-06, |
| "loss": 0.5702, |
| "step": 1069 |
| }, |
| { |
| "epoch": 4.331983805668016, |
| "grad_norm": 0.46564633935016675, |
| "learning_rate": 2.137699676264171e-06, |
| "loss": 0.5545, |
| "step": 1070 |
| }, |
| { |
| "epoch": 4.336032388663968, |
| "grad_norm": 0.4713858423291676, |
| "learning_rate": 2.1123313840455738e-06, |
| "loss": 0.5667, |
| "step": 1071 |
| }, |
| { |
| "epoch": 4.340080971659919, |
| "grad_norm": 0.48071352615348417, |
| "learning_rate": 2.08710612114857e-06, |
| "loss": 0.5502, |
| "step": 1072 |
| }, |
| { |
| "epoch": 4.34412955465587, |
| "grad_norm": 0.4566558384979126, |
| "learning_rate": 2.0620240892736953e-06, |
| "loss": 0.5542, |
| "step": 1073 |
| }, |
| { |
| "epoch": 4.348178137651822, |
| "grad_norm": 0.4826914713432461, |
| "learning_rate": 2.0370854889762e-06, |
| "loss": 0.5792, |
| "step": 1074 |
| }, |
| { |
| "epoch": 4.352226720647773, |
| "grad_norm": 0.47281633003279444, |
| "learning_rate": 2.0122905196644726e-06, |
| "loss": 0.5709, |
| "step": 1075 |
| }, |
| { |
| "epoch": 4.3562753036437245, |
| "grad_norm": 0.46814775449107565, |
| "learning_rate": 1.987639379598427e-06, |
| "loss": 0.5533, |
| "step": 1076 |
| }, |
| { |
| "epoch": 4.3603238866396765, |
| "grad_norm": 0.46861918565393434, |
| "learning_rate": 1.963132265887919e-06, |
| "loss": 0.5568, |
| "step": 1077 |
| }, |
| { |
| "epoch": 4.364372469635628, |
| "grad_norm": 0.4801993238143561, |
| "learning_rate": 1.9387693744911784e-06, |
| "loss": 0.5663, |
| "step": 1078 |
| }, |
| { |
| "epoch": 4.368421052631579, |
| "grad_norm": 0.46747841655508765, |
| "learning_rate": 1.9145509002132368e-06, |
| "loss": 0.5771, |
| "step": 1079 |
| }, |
| { |
| "epoch": 4.372469635627531, |
| "grad_norm": 0.46435346559136814, |
| "learning_rate": 1.8904770367043723e-06, |
| "loss": 0.5599, |
| "step": 1080 |
| }, |
| { |
| "epoch": 4.376518218623482, |
| "grad_norm": 0.461284452096678, |
| "learning_rate": 1.8665479764585504e-06, |
| "loss": 0.5603, |
| "step": 1081 |
| }, |
| { |
| "epoch": 4.380566801619433, |
| "grad_norm": 0.45902811927735876, |
| "learning_rate": 1.8427639108118977e-06, |
| "loss": 0.551, |
| "step": 1082 |
| }, |
| { |
| "epoch": 4.384615384615385, |
| "grad_norm": 0.4645438082427658, |
| "learning_rate": 1.8191250299411755e-06, |
| "loss": 0.5442, |
| "step": 1083 |
| }, |
| { |
| "epoch": 4.388663967611336, |
| "grad_norm": 0.471514668967908, |
| "learning_rate": 1.7956315228622446e-06, |
| "loss": 0.5342, |
| "step": 1084 |
| }, |
| { |
| "epoch": 4.392712550607287, |
| "grad_norm": 0.4747657859283996, |
| "learning_rate": 1.7722835774285553e-06, |
| "loss": 0.5413, |
| "step": 1085 |
| }, |
| { |
| "epoch": 4.396761133603239, |
| "grad_norm": 0.47992556082683413, |
| "learning_rate": 1.7490813803296624e-06, |
| "loss": 0.5534, |
| "step": 1086 |
| }, |
| { |
| "epoch": 4.40080971659919, |
| "grad_norm": 0.45993864801785617, |
| "learning_rate": 1.726025117089718e-06, |
| "loss": 0.5444, |
| "step": 1087 |
| }, |
| { |
| "epoch": 4.4048582995951415, |
| "grad_norm": 0.455311383734262, |
| "learning_rate": 1.70311497206598e-06, |
| "loss": 0.5392, |
| "step": 1088 |
| }, |
| { |
| "epoch": 4.4089068825910935, |
| "grad_norm": 0.4735271588115312, |
| "learning_rate": 1.6803511284473638e-06, |
| "loss": 0.5456, |
| "step": 1089 |
| }, |
| { |
| "epoch": 4.412955465587045, |
| "grad_norm": 0.45779234165980154, |
| "learning_rate": 1.6577337682529538e-06, |
| "loss": 0.5656, |
| "step": 1090 |
| }, |
| { |
| "epoch": 4.417004048582996, |
| "grad_norm": 0.4560334358775673, |
| "learning_rate": 1.6352630723305617e-06, |
| "loss": 0.5354, |
| "step": 1091 |
| }, |
| { |
| "epoch": 4.421052631578947, |
| "grad_norm": 0.47289969329903664, |
| "learning_rate": 1.612939220355272e-06, |
| "loss": 0.5581, |
| "step": 1092 |
| }, |
| { |
| "epoch": 4.425101214574899, |
| "grad_norm": 0.4754044859351733, |
| "learning_rate": 1.5907623908280112e-06, |
| "loss": 0.5686, |
| "step": 1093 |
| }, |
| { |
| "epoch": 4.42914979757085, |
| "grad_norm": 0.4730043877664583, |
| "learning_rate": 1.568732761074121e-06, |
| "loss": 0.5503, |
| "step": 1094 |
| }, |
| { |
| "epoch": 4.433198380566802, |
| "grad_norm": 0.48977643910557705, |
| "learning_rate": 1.5468505072419237e-06, |
| "loss": 0.5793, |
| "step": 1095 |
| }, |
| { |
| "epoch": 4.437246963562753, |
| "grad_norm": 0.4746327634894057, |
| "learning_rate": 1.5251158043013448e-06, |
| "loss": 0.5587, |
| "step": 1096 |
| }, |
| { |
| "epoch": 4.441295546558704, |
| "grad_norm": 0.4721734188375165, |
| "learning_rate": 1.5035288260424885e-06, |
| "loss": 0.5551, |
| "step": 1097 |
| }, |
| { |
| "epoch": 4.445344129554655, |
| "grad_norm": 0.46668975777982363, |
| "learning_rate": 1.4820897450742623e-06, |
| "loss": 0.5505, |
| "step": 1098 |
| }, |
| { |
| "epoch": 4.449392712550607, |
| "grad_norm": 0.4584392997489458, |
| "learning_rate": 1.4607987328229767e-06, |
| "loss": 0.5386, |
| "step": 1099 |
| }, |
| { |
| "epoch": 4.4534412955465585, |
| "grad_norm": 0.4649815879829264, |
| "learning_rate": 1.4396559595310056e-06, |
| "loss": 0.5512, |
| "step": 1100 |
| }, |
| { |
| "epoch": 4.4574898785425106, |
| "grad_norm": 0.48153178463589696, |
| "learning_rate": 1.4186615942553973e-06, |
| "loss": 0.5733, |
| "step": 1101 |
| }, |
| { |
| "epoch": 4.461538461538462, |
| "grad_norm": 0.4560396662000878, |
| "learning_rate": 1.3978158048665424e-06, |
| "loss": 0.5555, |
| "step": 1102 |
| }, |
| { |
| "epoch": 4.465587044534413, |
| "grad_norm": 0.4726169367397038, |
| "learning_rate": 1.3771187580468115e-06, |
| "loss": 0.5684, |
| "step": 1103 |
| }, |
| { |
| "epoch": 4.469635627530364, |
| "grad_norm": 0.4655056926486905, |
| "learning_rate": 1.3565706192892392e-06, |
| "loss": 0.5391, |
| "step": 1104 |
| }, |
| { |
| "epoch": 4.473684210526316, |
| "grad_norm": 0.4695869140067437, |
| "learning_rate": 1.3361715528961993e-06, |
| "loss": 0.5375, |
| "step": 1105 |
| }, |
| { |
| "epoch": 4.477732793522267, |
| "grad_norm": 0.46396104329642596, |
| "learning_rate": 1.3159217219780817e-06, |
| "loss": 0.5475, |
| "step": 1106 |
| }, |
| { |
| "epoch": 4.481781376518219, |
| "grad_norm": 0.481135453899786, |
| "learning_rate": 1.2958212884519949e-06, |
| "loss": 0.5738, |
| "step": 1107 |
| }, |
| { |
| "epoch": 4.48582995951417, |
| "grad_norm": 0.4536821079678787, |
| "learning_rate": 1.2758704130404675e-06, |
| "loss": 0.5435, |
| "step": 1108 |
| }, |
| { |
| "epoch": 4.489878542510121, |
| "grad_norm": 0.462182160684223, |
| "learning_rate": 1.2560692552701738e-06, |
| "loss": 0.5492, |
| "step": 1109 |
| }, |
| { |
| "epoch": 4.493927125506072, |
| "grad_norm": 0.4580729768212833, |
| "learning_rate": 1.2364179734706382e-06, |
| "loss": 0.5449, |
| "step": 1110 |
| }, |
| { |
| "epoch": 4.497975708502024, |
| "grad_norm": 0.47232473130513164, |
| "learning_rate": 1.2169167247729852e-06, |
| "loss": 0.5239, |
| "step": 1111 |
| }, |
| { |
| "epoch": 4.502024291497976, |
| "grad_norm": 0.4607261386675077, |
| "learning_rate": 1.1975656651086864e-06, |
| "loss": 0.5434, |
| "step": 1112 |
| }, |
| { |
| "epoch": 4.506072874493928, |
| "grad_norm": 0.4629452074451093, |
| "learning_rate": 1.1783649492082926e-06, |
| "loss": 0.5463, |
| "step": 1113 |
| }, |
| { |
| "epoch": 4.510121457489879, |
| "grad_norm": 0.46199743172638563, |
| "learning_rate": 1.1593147306002183e-06, |
| "loss": 0.5346, |
| "step": 1114 |
| }, |
| { |
| "epoch": 4.51417004048583, |
| "grad_norm": 0.4751959569183529, |
| "learning_rate": 1.140415161609505e-06, |
| "loss": 0.5402, |
| "step": 1115 |
| }, |
| { |
| "epoch": 4.518218623481781, |
| "grad_norm": 0.4728433315326435, |
| "learning_rate": 1.121666393356604e-06, |
| "loss": 0.5519, |
| "step": 1116 |
| }, |
| { |
| "epoch": 4.522267206477733, |
| "grad_norm": 0.4782362827356549, |
| "learning_rate": 1.1030685757561632e-06, |
| "loss": 0.5743, |
| "step": 1117 |
| }, |
| { |
| "epoch": 4.526315789473684, |
| "grad_norm": 0.4765333939842031, |
| "learning_rate": 1.0846218575158373e-06, |
| "loss": 0.5358, |
| "step": 1118 |
| }, |
| { |
| "epoch": 4.530364372469636, |
| "grad_norm": 0.4576735053164261, |
| "learning_rate": 1.066326386135097e-06, |
| "loss": 0.5555, |
| "step": 1119 |
| }, |
| { |
| "epoch": 4.534412955465587, |
| "grad_norm": 0.47573525985411286, |
| "learning_rate": 1.0481823079040398e-06, |
| "loss": 0.5352, |
| "step": 1120 |
| }, |
| { |
| "epoch": 4.538461538461538, |
| "grad_norm": 0.45986003674096043, |
| "learning_rate": 1.0301897679022344e-06, |
| "loss": 0.552, |
| "step": 1121 |
| }, |
| { |
| "epoch": 4.5425101214574894, |
| "grad_norm": 0.46608669987140816, |
| "learning_rate": 1.0123489099975491e-06, |
| "loss": 0.5737, |
| "step": 1122 |
| }, |
| { |
| "epoch": 4.5465587044534415, |
| "grad_norm": 0.4702006595696034, |
| "learning_rate": 9.9465987684501e-07, |
| "loss": 0.5449, |
| "step": 1123 |
| }, |
| { |
| "epoch": 4.550607287449393, |
| "grad_norm": 0.4610214350968514, |
| "learning_rate": 9.771228098856534e-07, |
| "loss": 0.5631, |
| "step": 1124 |
| }, |
| { |
| "epoch": 4.554655870445345, |
| "grad_norm": 0.46807279465594565, |
| "learning_rate": 9.597378493454013e-07, |
| "loss": 0.5636, |
| "step": 1125 |
| }, |
| { |
| "epoch": 4.558704453441296, |
| "grad_norm": 0.45791058525367606, |
| "learning_rate": 9.425051342339353e-07, |
| "loss": 0.5433, |
| "step": 1126 |
| }, |
| { |
| "epoch": 4.562753036437247, |
| "grad_norm": 0.47154066457280436, |
| "learning_rate": 9.254248023435864e-07, |
| "loss": 0.5653, |
| "step": 1127 |
| }, |
| { |
| "epoch": 4.566801619433198, |
| "grad_norm": 0.46603936491891407, |
| "learning_rate": 9.084969902482288e-07, |
| "loss": 0.5639, |
| "step": 1128 |
| }, |
| { |
| "epoch": 4.57085020242915, |
| "grad_norm": 0.46699118652644683, |
| "learning_rate": 8.917218333021993e-07, |
| "loss": 0.5664, |
| "step": 1129 |
| }, |
| { |
| "epoch": 4.574898785425101, |
| "grad_norm": 0.46640568261244986, |
| "learning_rate": 8.750994656392109e-07, |
| "loss": 0.5457, |
| "step": 1130 |
| }, |
| { |
| "epoch": 4.578947368421053, |
| "grad_norm": 0.46791055729676145, |
| "learning_rate": 8.58630020171265e-07, |
| "loss": 0.575, |
| "step": 1131 |
| }, |
| { |
| "epoch": 4.582995951417004, |
| "grad_norm": 0.4588524428370476, |
| "learning_rate": 8.423136285876099e-07, |
| "loss": 0.5588, |
| "step": 1132 |
| }, |
| { |
| "epoch": 4.587044534412955, |
| "grad_norm": 0.45920895976716497, |
| "learning_rate": 8.26150421353682e-07, |
| "loss": 0.5398, |
| "step": 1133 |
| }, |
| { |
| "epoch": 4.5910931174089065, |
| "grad_norm": 0.4628319482830738, |
| "learning_rate": 8.101405277100549e-07, |
| "loss": 0.5406, |
| "step": 1134 |
| }, |
| { |
| "epoch": 4.5951417004048585, |
| "grad_norm": 0.4804489992962439, |
| "learning_rate": 7.942840756714077e-07, |
| "loss": 0.5691, |
| "step": 1135 |
| }, |
| { |
| "epoch": 4.59919028340081, |
| "grad_norm": 0.4635149081368094, |
| "learning_rate": 7.785811920255093e-07, |
| "loss": 0.5559, |
| "step": 1136 |
| }, |
| { |
| "epoch": 4.603238866396762, |
| "grad_norm": 0.46485153405015156, |
| "learning_rate": 7.630320023322024e-07, |
| "loss": 0.5471, |
| "step": 1137 |
| }, |
| { |
| "epoch": 4.607287449392713, |
| "grad_norm": 0.45899528770177644, |
| "learning_rate": 7.476366309223903e-07, |
| "loss": 0.5457, |
| "step": 1138 |
| }, |
| { |
| "epoch": 4.611336032388664, |
| "grad_norm": 0.45591445679496895, |
| "learning_rate": 7.323952008970537e-07, |
| "loss": 0.5393, |
| "step": 1139 |
| }, |
| { |
| "epoch": 4.615384615384615, |
| "grad_norm": 0.47023626642452365, |
| "learning_rate": 7.173078341262618e-07, |
| "loss": 0.56, |
| "step": 1140 |
| }, |
| { |
| "epoch": 4.619433198380567, |
| "grad_norm": 0.4704540501147713, |
| "learning_rate": 7.023746512481989e-07, |
| "loss": 0.5861, |
| "step": 1141 |
| }, |
| { |
| "epoch": 4.623481781376518, |
| "grad_norm": 0.45553216445604044, |
| "learning_rate": 6.875957716681902e-07, |
| "loss": 0.5596, |
| "step": 1142 |
| }, |
| { |
| "epoch": 4.62753036437247, |
| "grad_norm": 0.460537330458832, |
| "learning_rate": 6.729713135577687e-07, |
| "loss": 0.5528, |
| "step": 1143 |
| }, |
| { |
| "epoch": 4.631578947368421, |
| "grad_norm": 0.46295824553032494, |
| "learning_rate": 6.585013938537143e-07, |
| "loss": 0.5486, |
| "step": 1144 |
| }, |
| { |
| "epoch": 4.635627530364372, |
| "grad_norm": 0.4584215341058629, |
| "learning_rate": 6.441861282571115e-07, |
| "loss": 0.5758, |
| "step": 1145 |
| }, |
| { |
| "epoch": 4.6396761133603235, |
| "grad_norm": 0.44867495638738836, |
| "learning_rate": 6.300256312324471e-07, |
| "loss": 0.5492, |
| "step": 1146 |
| }, |
| { |
| "epoch": 4.6437246963562755, |
| "grad_norm": 0.4734676032116956, |
| "learning_rate": 6.160200160066776e-07, |
| "loss": 0.5422, |
| "step": 1147 |
| }, |
| { |
| "epoch": 4.647773279352227, |
| "grad_norm": 0.4631968252938265, |
| "learning_rate": 6.021693945683327e-07, |
| "loss": 0.5881, |
| "step": 1148 |
| }, |
| { |
| "epoch": 4.651821862348179, |
| "grad_norm": 0.4459543123827127, |
| "learning_rate": 5.884738776666044e-07, |
| "loss": 0.5566, |
| "step": 1149 |
| }, |
| { |
| "epoch": 4.65587044534413, |
| "grad_norm": 0.45634113917251895, |
| "learning_rate": 5.749335748104878e-07, |
| "loss": 0.5514, |
| "step": 1150 |
| }, |
| { |
| "epoch": 4.659919028340081, |
| "grad_norm": 0.4572939155429965, |
| "learning_rate": 5.615485942678778e-07, |
| "loss": 0.5746, |
| "step": 1151 |
| }, |
| { |
| "epoch": 4.663967611336032, |
| "grad_norm": 0.46277637194292054, |
| "learning_rate": 5.483190430647289e-07, |
| "loss": 0.5648, |
| "step": 1152 |
| }, |
| { |
| "epoch": 4.668016194331984, |
| "grad_norm": 0.46597024259329484, |
| "learning_rate": 5.352450269841747e-07, |
| "loss": 0.568, |
| "step": 1153 |
| }, |
| { |
| "epoch": 4.672064777327935, |
| "grad_norm": 0.46110087605854866, |
| "learning_rate": 5.22326650565701e-07, |
| "loss": 0.5339, |
| "step": 1154 |
| }, |
| { |
| "epoch": 4.676113360323887, |
| "grad_norm": 0.46459243417106244, |
| "learning_rate": 5.095640171043048e-07, |
| "loss": 0.5777, |
| "step": 1155 |
| }, |
| { |
| "epoch": 4.680161943319838, |
| "grad_norm": 0.4545344639348187, |
| "learning_rate": 4.969572286496594e-07, |
| "loss": 0.521, |
| "step": 1156 |
| }, |
| { |
| "epoch": 4.684210526315789, |
| "grad_norm": 0.4746384042362378, |
| "learning_rate": 4.845063860053079e-07, |
| "loss": 0.5495, |
| "step": 1157 |
| }, |
| { |
| "epoch": 4.6882591093117405, |
| "grad_norm": 0.47716184881266216, |
| "learning_rate": 4.7221158872785997e-07, |
| "loss": 0.5628, |
| "step": 1158 |
| }, |
| { |
| "epoch": 4.6923076923076925, |
| "grad_norm": 0.4528306727811798, |
| "learning_rate": 4.600729351261857e-07, |
| "loss": 0.5606, |
| "step": 1159 |
| }, |
| { |
| "epoch": 4.696356275303644, |
| "grad_norm": 0.44928503093092303, |
| "learning_rate": 4.4809052226063355e-07, |
| "loss": 0.5665, |
| "step": 1160 |
| }, |
| { |
| "epoch": 4.700404858299595, |
| "grad_norm": 0.46911099214942514, |
| "learning_rate": 4.3626444594226046e-07, |
| "loss": 0.5503, |
| "step": 1161 |
| }, |
| { |
| "epoch": 4.704453441295547, |
| "grad_norm": 0.46804536002424224, |
| "learning_rate": 4.245948007320588e-07, |
| "loss": 0.5231, |
| "step": 1162 |
| }, |
| { |
| "epoch": 4.708502024291498, |
| "grad_norm": 0.4480436227392749, |
| "learning_rate": 4.130816799401971e-07, |
| "loss": 0.553, |
| "step": 1163 |
| }, |
| { |
| "epoch": 4.712550607287449, |
| "grad_norm": 0.4616888276100999, |
| "learning_rate": 4.017251756252827e-07, |
| "loss": 0.5561, |
| "step": 1164 |
| }, |
| { |
| "epoch": 4.716599190283401, |
| "grad_norm": 0.4677563445655689, |
| "learning_rate": 3.9052537859362473e-07, |
| "loss": 0.5863, |
| "step": 1165 |
| }, |
| { |
| "epoch": 4.720647773279352, |
| "grad_norm": 0.4604321614838545, |
| "learning_rate": 3.7948237839849687e-07, |
| "loss": 0.5307, |
| "step": 1166 |
| }, |
| { |
| "epoch": 4.724696356275303, |
| "grad_norm": 0.4606113195590151, |
| "learning_rate": 3.6859626333943797e-07, |
| "loss": 0.5644, |
| "step": 1167 |
| }, |
| { |
| "epoch": 4.728744939271255, |
| "grad_norm": 0.4551662593501243, |
| "learning_rate": 3.578671204615347e-07, |
| "loss": 0.5328, |
| "step": 1168 |
| }, |
| { |
| "epoch": 4.732793522267206, |
| "grad_norm": 0.4743003250177843, |
| "learning_rate": 3.472950355547267e-07, |
| "loss": 0.582, |
| "step": 1169 |
| }, |
| { |
| "epoch": 4.7368421052631575, |
| "grad_norm": 0.4493691046281069, |
| "learning_rate": 3.3688009315312244e-07, |
| "loss": 0.5353, |
| "step": 1170 |
| }, |
| { |
| "epoch": 4.7408906882591095, |
| "grad_norm": 0.46150828799074817, |
| "learning_rate": 3.266223765343224e-07, |
| "loss": 0.5526, |
| "step": 1171 |
| }, |
| { |
| "epoch": 4.744939271255061, |
| "grad_norm": 0.4619584749950087, |
| "learning_rate": 3.16521967718757e-07, |
| "loss": 0.5702, |
| "step": 1172 |
| }, |
| { |
| "epoch": 4.748987854251012, |
| "grad_norm": 0.4644031180471932, |
| "learning_rate": 3.0657894746902726e-07, |
| "loss": 0.5738, |
| "step": 1173 |
| }, |
| { |
| "epoch": 4.753036437246964, |
| "grad_norm": 0.46296825271500114, |
| "learning_rate": 2.9679339528925654e-07, |
| "loss": 0.5571, |
| "step": 1174 |
| }, |
| { |
| "epoch": 4.757085020242915, |
| "grad_norm": 0.45915186177179107, |
| "learning_rate": 2.871653894244597e-07, |
| "loss": 0.569, |
| "step": 1175 |
| }, |
| { |
| "epoch": 4.761133603238866, |
| "grad_norm": 0.46101870790498883, |
| "learning_rate": 2.7769500685991266e-07, |
| "loss": 0.5433, |
| "step": 1176 |
| }, |
| { |
| "epoch": 4.765182186234818, |
| "grad_norm": 0.466003810211752, |
| "learning_rate": 2.683823233205485e-07, |
| "loss": 0.563, |
| "step": 1177 |
| }, |
| { |
| "epoch": 4.769230769230769, |
| "grad_norm": 0.45832684444674643, |
| "learning_rate": 2.592274132703287e-07, |
| "loss": 0.5622, |
| "step": 1178 |
| }, |
| { |
| "epoch": 4.77327935222672, |
| "grad_norm": 0.44988573727340697, |
| "learning_rate": 2.502303499116754e-07, |
| "loss": 0.555, |
| "step": 1179 |
| }, |
| { |
| "epoch": 4.777327935222672, |
| "grad_norm": 0.4685622993287884, |
| "learning_rate": 2.413912051848666e-07, |
| "loss": 0.5541, |
| "step": 1180 |
| }, |
| { |
| "epoch": 4.781376518218623, |
| "grad_norm": 0.4690774280795803, |
| "learning_rate": 2.3271004976746613e-07, |
| "loss": 0.5723, |
| "step": 1181 |
| }, |
| { |
| "epoch": 4.7854251012145745, |
| "grad_norm": 0.4641043207219058, |
| "learning_rate": 2.2418695307376393e-07, |
| "loss": 0.5481, |
| "step": 1182 |
| }, |
| { |
| "epoch": 4.7894736842105265, |
| "grad_norm": 0.454668894936846, |
| "learning_rate": 2.1582198325420968e-07, |
| "loss": 0.5581, |
| "step": 1183 |
| }, |
| { |
| "epoch": 4.793522267206478, |
| "grad_norm": 0.4544358209566346, |
| "learning_rate": 2.0761520719488447e-07, |
| "loss": 0.5583, |
| "step": 1184 |
| }, |
| { |
| "epoch": 4.797570850202429, |
| "grad_norm": 0.4636071544089814, |
| "learning_rate": 1.9956669051694798e-07, |
| "loss": 0.5554, |
| "step": 1185 |
| }, |
| { |
| "epoch": 4.801619433198381, |
| "grad_norm": 0.4632115412370377, |
| "learning_rate": 1.916764975761254e-07, |
| "loss": 0.5413, |
| "step": 1186 |
| }, |
| { |
| "epoch": 4.805668016194332, |
| "grad_norm": 0.4615336822297377, |
| "learning_rate": 1.839446914621923e-07, |
| "loss": 0.57, |
| "step": 1187 |
| }, |
| { |
| "epoch": 4.809716599190283, |
| "grad_norm": 0.45222173254843456, |
| "learning_rate": 1.7637133399845962e-07, |
| "loss": 0.5534, |
| "step": 1188 |
| }, |
| { |
| "epoch": 4.813765182186235, |
| "grad_norm": 0.4691784719775113, |
| "learning_rate": 1.6895648574129397e-07, |
| "loss": 0.5447, |
| "step": 1189 |
| }, |
| { |
| "epoch": 4.817813765182186, |
| "grad_norm": 0.45376648382161217, |
| "learning_rate": 1.6170020597962465e-07, |
| "loss": 0.5383, |
| "step": 1190 |
| }, |
| { |
| "epoch": 4.821862348178137, |
| "grad_norm": 0.4493038362631467, |
| "learning_rate": 1.5460255273447078e-07, |
| "loss": 0.5766, |
| "step": 1191 |
| }, |
| { |
| "epoch": 4.825910931174089, |
| "grad_norm": 0.46162643937051845, |
| "learning_rate": 1.4766358275847715e-07, |
| "loss": 0.5543, |
| "step": 1192 |
| }, |
| { |
| "epoch": 4.82995951417004, |
| "grad_norm": 0.4561651194551197, |
| "learning_rate": 1.4088335153546573e-07, |
| "loss": 0.5663, |
| "step": 1193 |
| }, |
| { |
| "epoch": 4.834008097165992, |
| "grad_norm": 0.4652773637796262, |
| "learning_rate": 1.3426191327998496e-07, |
| "loss": 0.5513, |
| "step": 1194 |
| }, |
| { |
| "epoch": 4.838056680161944, |
| "grad_norm": 0.4518859214421965, |
| "learning_rate": 1.277993209368744e-07, |
| "loss": 0.5649, |
| "step": 1195 |
| }, |
| { |
| "epoch": 4.842105263157895, |
| "grad_norm": 0.45724158632917733, |
| "learning_rate": 1.214956261808542e-07, |
| "loss": 0.5716, |
| "step": 1196 |
| }, |
| { |
| "epoch": 4.846153846153846, |
| "grad_norm": 0.4545634083029702, |
| "learning_rate": 1.1535087941609846e-07, |
| "loss": 0.5671, |
| "step": 1197 |
| }, |
| { |
| "epoch": 4.850202429149798, |
| "grad_norm": 0.4655093911810796, |
| "learning_rate": 1.0936512977583802e-07, |
| "loss": 0.5656, |
| "step": 1198 |
| }, |
| { |
| "epoch": 4.854251012145749, |
| "grad_norm": 0.45375185626399006, |
| "learning_rate": 1.0353842512196732e-07, |
| "loss": 0.5371, |
| "step": 1199 |
| }, |
| { |
| "epoch": 4.8582995951417, |
| "grad_norm": 0.45772026866576027, |
| "learning_rate": 9.787081204466253e-08, |
| "loss": 0.5495, |
| "step": 1200 |
| }, |
| { |
| "epoch": 4.862348178137652, |
| "grad_norm": 0.45746750300936945, |
| "learning_rate": 9.236233586200627e-08, |
| "loss": 0.5514, |
| "step": 1201 |
| }, |
| { |
| "epoch": 4.866396761133603, |
| "grad_norm": 0.46430365561101855, |
| "learning_rate": 8.70130406196279e-08, |
| "loss": 0.5795, |
| "step": 1202 |
| }, |
| { |
| "epoch": 4.870445344129554, |
| "grad_norm": 0.44597917105719476, |
| "learning_rate": 8.182296909035047e-08, |
| "loss": 0.5644, |
| "step": 1203 |
| }, |
| { |
| "epoch": 4.874493927125506, |
| "grad_norm": 0.46360520841320807, |
| "learning_rate": 7.679216277384439e-08, |
| "loss": 0.5651, |
| "step": 1204 |
| }, |
| { |
| "epoch": 4.8785425101214575, |
| "grad_norm": 0.46262937734453513, |
| "learning_rate": 7.19206618963053e-08, |
| "loss": 0.5479, |
| "step": 1205 |
| }, |
| { |
| "epoch": 4.882591093117409, |
| "grad_norm": 0.4546986099111602, |
| "learning_rate": 6.720850541012347e-08, |
| "loss": 0.5501, |
| "step": 1206 |
| }, |
| { |
| "epoch": 4.886639676113361, |
| "grad_norm": 0.4654085815326181, |
| "learning_rate": 6.265573099357714e-08, |
| "loss": 0.5138, |
| "step": 1207 |
| }, |
| { |
| "epoch": 4.890688259109312, |
| "grad_norm": 0.45793684163151355, |
| "learning_rate": 5.826237505053067e-08, |
| "loss": 0.5479, |
| "step": 1208 |
| }, |
| { |
| "epoch": 4.894736842105263, |
| "grad_norm": 0.4670816483264318, |
| "learning_rate": 5.4028472710137e-08, |
| "loss": 0.5721, |
| "step": 1209 |
| }, |
| { |
| "epoch": 4.898785425101215, |
| "grad_norm": 0.44476645022864336, |
| "learning_rate": 4.9954057826571146e-08, |
| "loss": 0.5242, |
| "step": 1210 |
| }, |
| { |
| "epoch": 4.902834008097166, |
| "grad_norm": 0.45092625948469356, |
| "learning_rate": 4.6039162978743776e-08, |
| "loss": 0.5467, |
| "step": 1211 |
| }, |
| { |
| "epoch": 4.906882591093117, |
| "grad_norm": 0.45075391177281954, |
| "learning_rate": 4.228381947004812e-08, |
| "loss": 0.5293, |
| "step": 1212 |
| }, |
| { |
| "epoch": 4.910931174089069, |
| "grad_norm": 0.4549296307676341, |
| "learning_rate": 3.868805732811343e-08, |
| "loss": 0.5531, |
| "step": 1213 |
| }, |
| { |
| "epoch": 4.91497975708502, |
| "grad_norm": 0.4524899160692988, |
| "learning_rate": 3.525190530455636e-08, |
| "loss": 0.5844, |
| "step": 1214 |
| }, |
| { |
| "epoch": 4.919028340080971, |
| "grad_norm": 0.4595589495702853, |
| "learning_rate": 3.197539087475887e-08, |
| "loss": 0.558, |
| "step": 1215 |
| }, |
| { |
| "epoch": 4.923076923076923, |
| "grad_norm": 0.45350321194708293, |
| "learning_rate": 2.8858540237641786e-08, |
| "loss": 0.5451, |
| "step": 1216 |
| }, |
| { |
| "epoch": 4.9271255060728745, |
| "grad_norm": 0.4616119646471698, |
| "learning_rate": 2.5901378315458247e-08, |
| "loss": 0.5544, |
| "step": 1217 |
| }, |
| { |
| "epoch": 4.931174089068826, |
| "grad_norm": 0.45427961397780403, |
| "learning_rate": 2.3103928753598347e-08, |
| "loss": 0.5628, |
| "step": 1218 |
| }, |
| { |
| "epoch": 4.935222672064778, |
| "grad_norm": 0.4521116973836029, |
| "learning_rate": 2.0466213920393718e-08, |
| "loss": 0.55, |
| "step": 1219 |
| }, |
| { |
| "epoch": 4.939271255060729, |
| "grad_norm": 0.46592861428008847, |
| "learning_rate": 1.7988254906942115e-08, |
| "loss": 0.5377, |
| "step": 1220 |
| }, |
| { |
| "epoch": 4.94331983805668, |
| "grad_norm": 0.45082239842539656, |
| "learning_rate": 1.5670071526936447e-08, |
| "loss": 0.5597, |
| "step": 1221 |
| }, |
| { |
| "epoch": 4.947368421052632, |
| "grad_norm": 0.45785832898961903, |
| "learning_rate": 1.3511682316509344e-08, |
| "loss": 0.5592, |
| "step": 1222 |
| }, |
| { |
| "epoch": 4.951417004048583, |
| "grad_norm": 0.4555421845340565, |
| "learning_rate": 1.1513104534086605e-08, |
| "loss": 0.5207, |
| "step": 1223 |
| }, |
| { |
| "epoch": 4.955465587044534, |
| "grad_norm": 0.44496467004670387, |
| "learning_rate": 9.67435416023843e-09, |
| "loss": 0.5457, |
| "step": 1224 |
| }, |
| { |
| "epoch": 4.959514170040486, |
| "grad_norm": 0.4522410455412655, |
| "learning_rate": 7.995445897566178e-09, |
| "loss": 0.5143, |
| "step": 1225 |
| }, |
| { |
| "epoch": 4.963562753036437, |
| "grad_norm": 0.4625703188928339, |
| "learning_rate": 6.476393170578021e-09, |
| "loss": 0.5699, |
| "step": 1226 |
| }, |
| { |
| "epoch": 4.967611336032388, |
| "grad_norm": 0.46758175364599874, |
| "learning_rate": 5.11720812557348e-09, |
| "loss": 0.5434, |
| "step": 1227 |
| }, |
| { |
| "epoch": 4.97165991902834, |
| "grad_norm": 0.46284484820889604, |
| "learning_rate": 3.917901630563492e-09, |
| "loss": 0.5679, |
| "step": 1228 |
| }, |
| { |
| "epoch": 4.9757085020242915, |
| "grad_norm": 0.45339245458491356, |
| "learning_rate": 2.878483275170485e-09, |
| "loss": 0.5481, |
| "step": 1229 |
| }, |
| { |
| "epoch": 4.979757085020243, |
| "grad_norm": 0.4734672396436435, |
| "learning_rate": 1.998961370555108e-09, |
| "loss": 0.5555, |
| "step": 1230 |
| }, |
| { |
| "epoch": 4.983805668016195, |
| "grad_norm": 0.4734234237205056, |
| "learning_rate": 1.2793429493518361e-09, |
| "loss": 0.5461, |
| "step": 1231 |
| }, |
| { |
| "epoch": 4.987854251012146, |
| "grad_norm": 0.45880151438470096, |
| "learning_rate": 7.196337656112384e-10, |
| "loss": 0.5536, |
| "step": 1232 |
| }, |
| { |
| "epoch": 4.991902834008097, |
| "grad_norm": 0.4727121523765575, |
| "learning_rate": 3.198382947511292e-10, |
| "loss": 0.5769, |
| "step": 1233 |
| }, |
| { |
| "epoch": 4.995951417004049, |
| "grad_norm": 0.44632366124392686, |
| "learning_rate": 7.995973352770137e-11, |
| "loss": 0.5305, |
| "step": 1234 |
| }, |
| { |
| "epoch": 5.0, |
| "grad_norm": 0.4527298467959021, |
| "learning_rate": 0.0, |
| "loss": 0.5326, |
| "step": 1235 |
| }, |
| { |
| "epoch": 5.0, |
| "step": 1235, |
| "total_flos": 1.615206363376386e+18, |
| "train_loss": 0.8795208685311229, |
| "train_runtime": 13686.0149, |
| "train_samples_per_second": 11.545, |
| "train_steps_per_second": 0.09 |
| } |
| ], |
| "logging_steps": 1.0, |
| "max_steps": 1235, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 5, |
| "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": 1.615206363376386e+18, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
