| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0287932707861533, | |
| "eval_steps": 386, | |
| "global_step": 795, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0012940795858945326, | |
| "grad_norm": 1.7405146360397339, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 2.4269, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0012940795858945326, | |
| "eval_loss": 2.247628688812256, | |
| "eval_runtime": 189.8853, | |
| "eval_samples_per_second": 3.428, | |
| "eval_steps_per_second": 0.858, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.002588159171789065, | |
| "grad_norm": 1.6643542051315308, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 2.2583, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0038822387576835974, | |
| "grad_norm": 1.8690767288208008, | |
| "learning_rate": 6e-06, | |
| "loss": 2.2696, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00517631834357813, | |
| "grad_norm": 1.828118085861206, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 2.3646, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.006470397929472662, | |
| "grad_norm": 1.9319926500320435, | |
| "learning_rate": 1e-05, | |
| "loss": 2.4196, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.007764477515367195, | |
| "grad_norm": 1.7723782062530518, | |
| "learning_rate": 1.2e-05, | |
| "loss": 2.4177, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.009058557101261728, | |
| "grad_norm": 1.9500815868377686, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 2.3497, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.01035263668715626, | |
| "grad_norm": 2.3909075260162354, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 2.405, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.011646716273050793, | |
| "grad_norm": 2.0620856285095215, | |
| "learning_rate": 1.8e-05, | |
| "loss": 2.4098, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.012940795858945324, | |
| "grad_norm": 1.8054910898208618, | |
| "learning_rate": 2e-05, | |
| "loss": 2.1233, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.014234875444839857, | |
| "grad_norm": 2.190964937210083, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 2.3985, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.01552895503073439, | |
| "grad_norm": 1.9412921667099, | |
| "learning_rate": 2.4e-05, | |
| "loss": 2.462, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.016823034616628922, | |
| "grad_norm": 1.9161555767059326, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 2.2118, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.018117114202523456, | |
| "grad_norm": 1.7161599397659302, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 2.2175, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.019411193788417987, | |
| "grad_norm": 2.173877000808716, | |
| "learning_rate": 3e-05, | |
| "loss": 2.2521, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02070527337431252, | |
| "grad_norm": 2.0000555515289307, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 2.1615, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.021999352960207053, | |
| "grad_norm": 1.5915080308914185, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 1.9522, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.023293432546101587, | |
| "grad_norm": 1.6972448825836182, | |
| "learning_rate": 3.6e-05, | |
| "loss": 1.7224, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.024587512131996118, | |
| "grad_norm": 1.7509772777557373, | |
| "learning_rate": 3.8e-05, | |
| "loss": 2.0414, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.02588159171789065, | |
| "grad_norm": 1.697340488433838, | |
| "learning_rate": 4e-05, | |
| "loss": 2.0427, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.027175671303785183, | |
| "grad_norm": 1.8733758926391602, | |
| "learning_rate": 4.2e-05, | |
| "loss": 1.6772, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.028469750889679714, | |
| "grad_norm": 1.6085255146026611, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 1.6527, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.029763830475574248, | |
| "grad_norm": 1.5792337656021118, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 1.6567, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.03105791006146878, | |
| "grad_norm": 1.4392567873001099, | |
| "learning_rate": 4.8e-05, | |
| "loss": 1.508, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.03235198964736331, | |
| "grad_norm": 1.5222433805465698, | |
| "learning_rate": 5e-05, | |
| "loss": 1.4606, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.033646069233257844, | |
| "grad_norm": 1.5480064153671265, | |
| "learning_rate": 5.2000000000000004e-05, | |
| "loss": 1.5027, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.034940148819152375, | |
| "grad_norm": 1.6736445426940918, | |
| "learning_rate": 5.4000000000000005e-05, | |
| "loss": 1.2426, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.03623422840504691, | |
| "grad_norm": 1.7392551898956299, | |
| "learning_rate": 5.6000000000000006e-05, | |
| "loss": 1.4703, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.037528307990941444, | |
| "grad_norm": 1.6173359155654907, | |
| "learning_rate": 5.8e-05, | |
| "loss": 1.4546, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.038822387576835975, | |
| "grad_norm": 1.3955802917480469, | |
| "learning_rate": 6e-05, | |
| "loss": 1.3808, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.040116467162730506, | |
| "grad_norm": 1.353873372077942, | |
| "learning_rate": 6.2e-05, | |
| "loss": 1.229, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.04141054674862504, | |
| "grad_norm": 1.2547746896743774, | |
| "learning_rate": 6.400000000000001e-05, | |
| "loss": 1.1668, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.042704626334519574, | |
| "grad_norm": 1.3806778192520142, | |
| "learning_rate": 6.6e-05, | |
| "loss": 1.0691, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.043998705920414105, | |
| "grad_norm": 1.2815773487091064, | |
| "learning_rate": 6.800000000000001e-05, | |
| "loss": 1.2409, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.045292785506308636, | |
| "grad_norm": 1.3677266836166382, | |
| "learning_rate": 7e-05, | |
| "loss": 0.9668, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.046586865092203174, | |
| "grad_norm": 1.5457032918930054, | |
| "learning_rate": 7.2e-05, | |
| "loss": 1.1385, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.047880944678097705, | |
| "grad_norm": 1.5587060451507568, | |
| "learning_rate": 7.4e-05, | |
| "loss": 1.1707, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.049175024263992236, | |
| "grad_norm": 1.079053282737732, | |
| "learning_rate": 7.6e-05, | |
| "loss": 1.0655, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.050469103849886766, | |
| "grad_norm": 1.1773897409439087, | |
| "learning_rate": 7.800000000000001e-05, | |
| "loss": 1.0465, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0517631834357813, | |
| "grad_norm": 1.2437673807144165, | |
| "learning_rate": 8e-05, | |
| "loss": 1.2779, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.053057263021675835, | |
| "grad_norm": 1.254847526550293, | |
| "learning_rate": 8.2e-05, | |
| "loss": 1.0898, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.054351342607570366, | |
| "grad_norm": 1.1771515607833862, | |
| "learning_rate": 8.4e-05, | |
| "loss": 1.1827, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0556454221934649, | |
| "grad_norm": 1.1400648355484009, | |
| "learning_rate": 8.6e-05, | |
| "loss": 1.1066, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.05693950177935943, | |
| "grad_norm": 1.2047138214111328, | |
| "learning_rate": 8.800000000000001e-05, | |
| "loss": 0.8974, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.058233581365253966, | |
| "grad_norm": 1.1269346475601196, | |
| "learning_rate": 9e-05, | |
| "loss": 1.0146, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.059527660951148496, | |
| "grad_norm": 1.169231653213501, | |
| "learning_rate": 9.200000000000001e-05, | |
| "loss": 1.1266, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.06082174053704303, | |
| "grad_norm": 0.9771779179573059, | |
| "learning_rate": 9.4e-05, | |
| "loss": 0.8351, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.06211582012293756, | |
| "grad_norm": 1.2849314212799072, | |
| "learning_rate": 9.6e-05, | |
| "loss": 1.1822, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0634098997088321, | |
| "grad_norm": 1.023181676864624, | |
| "learning_rate": 9.8e-05, | |
| "loss": 0.9082, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.06470397929472663, | |
| "grad_norm": 1.135751724243164, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9407, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06599805888062116, | |
| "grad_norm": 0.9701154828071594, | |
| "learning_rate": 9.999998300231494e-05, | |
| "loss": 0.9423, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.06729213846651569, | |
| "grad_norm": 1.2891143560409546, | |
| "learning_rate": 9.999993200927133e-05, | |
| "loss": 0.9757, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.06858621805241022, | |
| "grad_norm": 1.3360975980758667, | |
| "learning_rate": 9.999984702090383e-05, | |
| "loss": 1.0158, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.06988029763830475, | |
| "grad_norm": 0.977446436882019, | |
| "learning_rate": 9.999972803727024e-05, | |
| "loss": 0.8175, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0711743772241993, | |
| "grad_norm": 0.9943827390670776, | |
| "learning_rate": 9.999957505845144e-05, | |
| "loss": 0.8627, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.07246845681009383, | |
| "grad_norm": 1.1531224250793457, | |
| "learning_rate": 9.999938808455145e-05, | |
| "loss": 1.143, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.07376253639598836, | |
| "grad_norm": 1.287972092628479, | |
| "learning_rate": 9.99991671156974e-05, | |
| "loss": 1.2342, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.07505661598188289, | |
| "grad_norm": 1.1554590463638306, | |
| "learning_rate": 9.999891215203949e-05, | |
| "loss": 0.9692, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.07635069556777742, | |
| "grad_norm": 1.0786008834838867, | |
| "learning_rate": 9.999862319375113e-05, | |
| "loss": 1.1254, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.07764477515367195, | |
| "grad_norm": 1.0764508247375488, | |
| "learning_rate": 9.999830024102874e-05, | |
| "loss": 0.9312, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07893885473956648, | |
| "grad_norm": 1.1909526586532593, | |
| "learning_rate": 9.999794329409194e-05, | |
| "loss": 0.9959, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.08023293432546101, | |
| "grad_norm": 0.9989166259765625, | |
| "learning_rate": 9.999755235318337e-05, | |
| "loss": 0.934, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.08152701391135554, | |
| "grad_norm": 1.0302046537399292, | |
| "learning_rate": 9.999712741856889e-05, | |
| "loss": 1.1017, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.08282109349725009, | |
| "grad_norm": 0.9583478569984436, | |
| "learning_rate": 9.999666849053738e-05, | |
| "loss": 1.1384, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.08411517308314462, | |
| "grad_norm": 1.001126766204834, | |
| "learning_rate": 9.999617556940085e-05, | |
| "loss": 0.9279, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.08540925266903915, | |
| "grad_norm": 1.0130903720855713, | |
| "learning_rate": 9.999564865549449e-05, | |
| "loss": 0.9381, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.08670333225493368, | |
| "grad_norm": 1.1210829019546509, | |
| "learning_rate": 9.999508774917652e-05, | |
| "loss": 0.9607, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.08799741184082821, | |
| "grad_norm": 1.045749545097351, | |
| "learning_rate": 9.999449285082831e-05, | |
| "loss": 1.0037, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.08929149142672274, | |
| "grad_norm": 1.1308139562606812, | |
| "learning_rate": 9.999386396085434e-05, | |
| "loss": 0.9086, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.09058557101261727, | |
| "grad_norm": 1.1013413667678833, | |
| "learning_rate": 9.999320107968219e-05, | |
| "loss": 1.0712, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0918796505985118, | |
| "grad_norm": 1.0830566883087158, | |
| "learning_rate": 9.999250420776258e-05, | |
| "loss": 1.0326, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.09317373018440635, | |
| "grad_norm": 1.0673171281814575, | |
| "learning_rate": 9.999177334556929e-05, | |
| "loss": 1.0034, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.09446780977030088, | |
| "grad_norm": 1.1546461582183838, | |
| "learning_rate": 9.999100849359926e-05, | |
| "loss": 1.059, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.09576188935619541, | |
| "grad_norm": 0.9139528870582581, | |
| "learning_rate": 9.999020965237249e-05, | |
| "loss": 0.8596, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.09705596894208994, | |
| "grad_norm": 1.1570812463760376, | |
| "learning_rate": 9.998937682243215e-05, | |
| "loss": 1.0456, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.09835004852798447, | |
| "grad_norm": 1.3232612609863281, | |
| "learning_rate": 9.998851000434448e-05, | |
| "loss": 0.9994, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.099644128113879, | |
| "grad_norm": 1.2017115354537964, | |
| "learning_rate": 9.998760919869883e-05, | |
| "loss": 1.2664, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.10093820769977353, | |
| "grad_norm": 1.0694175958633423, | |
| "learning_rate": 9.998667440610765e-05, | |
| "loss": 0.9483, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.10223228728566806, | |
| "grad_norm": 0.9963059425354004, | |
| "learning_rate": 9.998570562720654e-05, | |
| "loss": 0.9577, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.1035263668715626, | |
| "grad_norm": 0.8873535394668579, | |
| "learning_rate": 9.998470286265416e-05, | |
| "loss": 0.8498, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.10482044645745714, | |
| "grad_norm": 1.1350760459899902, | |
| "learning_rate": 9.99836661131323e-05, | |
| "loss": 1.0024, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.10611452604335167, | |
| "grad_norm": 0.8355389833450317, | |
| "learning_rate": 9.998259537934586e-05, | |
| "loss": 0.7399, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.1074086056292462, | |
| "grad_norm": 0.9935446381568909, | |
| "learning_rate": 9.998149066202284e-05, | |
| "loss": 0.9809, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.10870268521514073, | |
| "grad_norm": 1.0571558475494385, | |
| "learning_rate": 9.998035196191435e-05, | |
| "loss": 1.0144, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.10999676480103526, | |
| "grad_norm": 0.9860286116600037, | |
| "learning_rate": 9.99791792797946e-05, | |
| "loss": 1.0467, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.1112908443869298, | |
| "grad_norm": 1.1422507762908936, | |
| "learning_rate": 9.997797261646089e-05, | |
| "loss": 0.9535, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.11258492397282432, | |
| "grad_norm": 0.8561545014381409, | |
| "learning_rate": 9.997673197273365e-05, | |
| "loss": 1.007, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.11387900355871886, | |
| "grad_norm": 1.0027543306350708, | |
| "learning_rate": 9.997545734945639e-05, | |
| "loss": 0.9861, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.1151730831446134, | |
| "grad_norm": 0.8489773273468018, | |
| "learning_rate": 9.997414874749575e-05, | |
| "loss": 0.9672, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.11646716273050793, | |
| "grad_norm": 1.0517115592956543, | |
| "learning_rate": 9.997280616774147e-05, | |
| "loss": 1.1672, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11776124231640246, | |
| "grad_norm": 1.0035395622253418, | |
| "learning_rate": 9.997142961110634e-05, | |
| "loss": 0.9294, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.11905532190229699, | |
| "grad_norm": 1.1194915771484375, | |
| "learning_rate": 9.997001907852635e-05, | |
| "loss": 1.0857, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.12034940148819152, | |
| "grad_norm": 1.5234825611114502, | |
| "learning_rate": 9.996857457096047e-05, | |
| "loss": 1.027, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.12164348107408605, | |
| "grad_norm": 0.949878454208374, | |
| "learning_rate": 9.996709608939088e-05, | |
| "loss": 0.8173, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.12293756065998059, | |
| "grad_norm": 0.8736472129821777, | |
| "learning_rate": 9.996558363482277e-05, | |
| "loss": 0.855, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.12423164024587512, | |
| "grad_norm": 0.8604567050933838, | |
| "learning_rate": 9.996403720828449e-05, | |
| "loss": 0.9485, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.12552571983176966, | |
| "grad_norm": 1.020851492881775, | |
| "learning_rate": 9.996245681082748e-05, | |
| "loss": 1.0024, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.1268197994176642, | |
| "grad_norm": 1.0704892873764038, | |
| "learning_rate": 9.996084244352623e-05, | |
| "loss": 0.9246, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.12811387900355872, | |
| "grad_norm": 0.8441987037658691, | |
| "learning_rate": 9.99591941074784e-05, | |
| "loss": 1.0343, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.12940795858945325, | |
| "grad_norm": 1.0280612707138062, | |
| "learning_rate": 9.995751180380466e-05, | |
| "loss": 0.9644, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.13070203817534778, | |
| "grad_norm": 0.9827906489372253, | |
| "learning_rate": 9.995579553364887e-05, | |
| "loss": 0.9583, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.13199611776124232, | |
| "grad_norm": 1.035618543624878, | |
| "learning_rate": 9.995404529817791e-05, | |
| "loss": 1.0366, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.13329019734713685, | |
| "grad_norm": 1.2775524854660034, | |
| "learning_rate": 9.995226109858178e-05, | |
| "loss": 0.9353, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.13458427693303138, | |
| "grad_norm": 1.0101919174194336, | |
| "learning_rate": 9.995044293607355e-05, | |
| "loss": 0.9045, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.1358783565189259, | |
| "grad_norm": 0.8396942019462585, | |
| "learning_rate": 9.994859081188943e-05, | |
| "loss": 0.867, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.13717243610482044, | |
| "grad_norm": 1.04515540599823, | |
| "learning_rate": 9.99467047272887e-05, | |
| "loss": 0.9693, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.13846651569071497, | |
| "grad_norm": 1.099042534828186, | |
| "learning_rate": 9.994478468355369e-05, | |
| "loss": 0.8879, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.1397605952766095, | |
| "grad_norm": 0.8710360527038574, | |
| "learning_rate": 9.994283068198988e-05, | |
| "loss": 0.9018, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.14105467486250403, | |
| "grad_norm": 0.961025059223175, | |
| "learning_rate": 9.99408427239258e-05, | |
| "loss": 0.8806, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.1423487544483986, | |
| "grad_norm": 0.915665328502655, | |
| "learning_rate": 9.993882081071306e-05, | |
| "loss": 0.8628, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.14364283403429312, | |
| "grad_norm": 1.2776648998260498, | |
| "learning_rate": 9.993676494372642e-05, | |
| "loss": 0.9742, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.14493691362018765, | |
| "grad_norm": 1.1270071268081665, | |
| "learning_rate": 9.993467512436364e-05, | |
| "loss": 0.9729, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.14623099320608218, | |
| "grad_norm": 0.8188664317131042, | |
| "learning_rate": 9.99325513540456e-05, | |
| "loss": 0.9428, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.1475250727919767, | |
| "grad_norm": 1.0760393142700195, | |
| "learning_rate": 9.993039363421627e-05, | |
| "loss": 0.9482, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.14881915237787124, | |
| "grad_norm": 1.019920825958252, | |
| "learning_rate": 9.992820196634273e-05, | |
| "loss": 0.9785, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.15011323196376578, | |
| "grad_norm": 0.8342046737670898, | |
| "learning_rate": 9.992597635191509e-05, | |
| "loss": 0.9291, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.1514073115496603, | |
| "grad_norm": 0.8460632562637329, | |
| "learning_rate": 9.992371679244658e-05, | |
| "loss": 0.8797, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.15270139113555484, | |
| "grad_norm": 0.933060348033905, | |
| "learning_rate": 9.992142328947345e-05, | |
| "loss": 0.9657, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.15399547072144937, | |
| "grad_norm": 0.8822593688964844, | |
| "learning_rate": 9.991909584455511e-05, | |
| "loss": 0.8872, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.1552895503073439, | |
| "grad_norm": 0.9599350094795227, | |
| "learning_rate": 9.991673445927398e-05, | |
| "loss": 0.9064, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.15658362989323843, | |
| "grad_norm": 0.8505874872207642, | |
| "learning_rate": 9.99143391352356e-05, | |
| "loss": 0.9966, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.15787770947913296, | |
| "grad_norm": 1.3977786302566528, | |
| "learning_rate": 9.991190987406857e-05, | |
| "loss": 0.9145, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.1591717890650275, | |
| "grad_norm": 0.8947294354438782, | |
| "learning_rate": 9.990944667742455e-05, | |
| "loss": 0.9569, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.16046586865092202, | |
| "grad_norm": 0.7973839044570923, | |
| "learning_rate": 9.990694954697828e-05, | |
| "loss": 0.8853, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.16175994823681655, | |
| "grad_norm": 0.9481159448623657, | |
| "learning_rate": 9.99044184844276e-05, | |
| "loss": 1.04, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.16305402782271108, | |
| "grad_norm": 1.3568611145019531, | |
| "learning_rate": 9.990185349149339e-05, | |
| "loss": 1.1104, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.16434810740860564, | |
| "grad_norm": 0.900867223739624, | |
| "learning_rate": 9.98992545699196e-05, | |
| "loss": 0.8427, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.16564218699450017, | |
| "grad_norm": 0.9025059938430786, | |
| "learning_rate": 9.989662172147326e-05, | |
| "loss": 0.9671, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.1669362665803947, | |
| "grad_norm": 0.944692850112915, | |
| "learning_rate": 9.989395494794446e-05, | |
| "loss": 1.0966, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.16823034616628924, | |
| "grad_norm": 1.196311116218567, | |
| "learning_rate": 9.989125425114638e-05, | |
| "loss": 1.0888, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.16952442575218377, | |
| "grad_norm": 0.9069584608078003, | |
| "learning_rate": 9.988851963291522e-05, | |
| "loss": 0.8579, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.1708185053380783, | |
| "grad_norm": 0.8150789141654968, | |
| "learning_rate": 9.988575109511026e-05, | |
| "loss": 0.7622, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.17211258492397283, | |
| "grad_norm": 1.0844395160675049, | |
| "learning_rate": 9.988294863961387e-05, | |
| "loss": 0.9284, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.17340666450986736, | |
| "grad_norm": 1.0463049411773682, | |
| "learning_rate": 9.988011226833146e-05, | |
| "loss": 0.9185, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.1747007440957619, | |
| "grad_norm": 0.9481234550476074, | |
| "learning_rate": 9.987724198319148e-05, | |
| "loss": 0.8631, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.17599482368165642, | |
| "grad_norm": 0.882074773311615, | |
| "learning_rate": 9.987433778614549e-05, | |
| "loss": 0.8997, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.17728890326755095, | |
| "grad_norm": 0.9853332042694092, | |
| "learning_rate": 9.987139967916805e-05, | |
| "loss": 0.9226, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.17858298285344548, | |
| "grad_norm": 1.151941180229187, | |
| "learning_rate": 9.98684276642568e-05, | |
| "loss": 1.0486, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.17987706243934, | |
| "grad_norm": 1.0128459930419922, | |
| "learning_rate": 9.986542174343245e-05, | |
| "loss": 1.0797, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.18117114202523454, | |
| "grad_norm": 0.9798718094825745, | |
| "learning_rate": 9.986238191873874e-05, | |
| "loss": 0.875, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.18246522161112907, | |
| "grad_norm": 0.8143295645713806, | |
| "learning_rate": 9.985930819224247e-05, | |
| "loss": 0.8454, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.1837593011970236, | |
| "grad_norm": 0.8755755424499512, | |
| "learning_rate": 9.985620056603348e-05, | |
| "loss": 0.8029, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.18505338078291814, | |
| "grad_norm": 0.899174690246582, | |
| "learning_rate": 9.985305904222469e-05, | |
| "loss": 0.9608, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.1863474603688127, | |
| "grad_norm": 0.920137882232666, | |
| "learning_rate": 9.984988362295203e-05, | |
| "loss": 0.9022, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.18764153995470723, | |
| "grad_norm": 1.1012908220291138, | |
| "learning_rate": 9.984667431037447e-05, | |
| "loss": 0.9621, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.18893561954060176, | |
| "grad_norm": 0.8609358668327332, | |
| "learning_rate": 9.98434311066741e-05, | |
| "loss": 0.917, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.1902296991264963, | |
| "grad_norm": 0.8248727321624756, | |
| "learning_rate": 9.984015401405594e-05, | |
| "loss": 0.7864, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.19152377871239082, | |
| "grad_norm": 0.8680225610733032, | |
| "learning_rate": 9.983684303474815e-05, | |
| "loss": 0.9288, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.19281785829828535, | |
| "grad_norm": 1.0807067155838013, | |
| "learning_rate": 9.983349817100188e-05, | |
| "loss": 0.9842, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.19411193788417988, | |
| "grad_norm": 0.9310898780822754, | |
| "learning_rate": 9.983011942509131e-05, | |
| "loss": 1.0568, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.1954060174700744, | |
| "grad_norm": 0.8052242398262024, | |
| "learning_rate": 9.98267067993137e-05, | |
| "loss": 0.8235, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.19670009705596894, | |
| "grad_norm": 0.9700384140014648, | |
| "learning_rate": 9.982326029598931e-05, | |
| "loss": 0.8611, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.19799417664186347, | |
| "grad_norm": 0.8437764048576355, | |
| "learning_rate": 9.981977991746142e-05, | |
| "loss": 0.83, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.199288256227758, | |
| "grad_norm": 0.930636465549469, | |
| "learning_rate": 9.98162656660964e-05, | |
| "loss": 1.0892, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.20058233581365253, | |
| "grad_norm": 0.9111954569816589, | |
| "learning_rate": 9.98127175442836e-05, | |
| "loss": 0.9962, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.20187641539954707, | |
| "grad_norm": 0.9521974921226501, | |
| "learning_rate": 9.980913555443541e-05, | |
| "loss": 0.911, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.2031704949854416, | |
| "grad_norm": 0.8516745567321777, | |
| "learning_rate": 9.980551969898727e-05, | |
| "loss": 0.9009, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.20446457457133613, | |
| "grad_norm": 0.8302998542785645, | |
| "learning_rate": 9.98018699803976e-05, | |
| "loss": 0.8854, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.20575865415723066, | |
| "grad_norm": 0.814391016960144, | |
| "learning_rate": 9.979818640114789e-05, | |
| "loss": 0.9601, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.2070527337431252, | |
| "grad_norm": 0.8938564658164978, | |
| "learning_rate": 9.979446896374262e-05, | |
| "loss": 0.8834, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.20834681332901975, | |
| "grad_norm": 0.9066985249519348, | |
| "learning_rate": 9.979071767070932e-05, | |
| "loss": 0.7427, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.20964089291491428, | |
| "grad_norm": 0.7866595983505249, | |
| "learning_rate": 9.978693252459851e-05, | |
| "loss": 0.8556, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.2109349725008088, | |
| "grad_norm": 0.9159708023071289, | |
| "learning_rate": 9.978311352798374e-05, | |
| "loss": 0.8101, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.21222905208670334, | |
| "grad_norm": 1.1350793838500977, | |
| "learning_rate": 9.977926068346157e-05, | |
| "loss": 0.9374, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.21352313167259787, | |
| "grad_norm": 1.0535932779312134, | |
| "learning_rate": 9.977537399365159e-05, | |
| "loss": 1.0238, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.2148172112584924, | |
| "grad_norm": 0.8717033267021179, | |
| "learning_rate": 9.977145346119637e-05, | |
| "loss": 1.0265, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.21611129084438693, | |
| "grad_norm": 0.8357003927230835, | |
| "learning_rate": 9.976749908876152e-05, | |
| "loss": 0.9016, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.21740537043028146, | |
| "grad_norm": 0.8369495868682861, | |
| "learning_rate": 9.976351087903568e-05, | |
| "loss": 0.8764, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.218699450016176, | |
| "grad_norm": 0.912352979183197, | |
| "learning_rate": 9.97594888347304e-05, | |
| "loss": 0.9078, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.21999352960207053, | |
| "grad_norm": 0.8475804328918457, | |
| "learning_rate": 9.975543295858035e-05, | |
| "loss": 0.8836, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.22128760918796506, | |
| "grad_norm": 0.8391397595405579, | |
| "learning_rate": 9.97513432533431e-05, | |
| "loss": 0.9003, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.2225816887738596, | |
| "grad_norm": 0.9666821360588074, | |
| "learning_rate": 9.974721972179931e-05, | |
| "loss": 0.9528, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.22387576835975412, | |
| "grad_norm": 0.9321691393852234, | |
| "learning_rate": 9.974306236675259e-05, | |
| "loss": 0.9575, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.22516984794564865, | |
| "grad_norm": 0.8022271990776062, | |
| "learning_rate": 9.973887119102957e-05, | |
| "loss": 0.8731, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.22646392753154318, | |
| "grad_norm": 1.1056872606277466, | |
| "learning_rate": 9.973464619747983e-05, | |
| "loss": 0.9925, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.2277580071174377, | |
| "grad_norm": 0.810420036315918, | |
| "learning_rate": 9.9730387388976e-05, | |
| "loss": 1.0073, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.22905208670333224, | |
| "grad_norm": 0.9536454677581787, | |
| "learning_rate": 9.972609476841367e-05, | |
| "loss": 0.9595, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.2303461662892268, | |
| "grad_norm": 0.8205066919326782, | |
| "learning_rate": 9.972176833871142e-05, | |
| "loss": 0.8146, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.23164024587512133, | |
| "grad_norm": 0.9716495275497437, | |
| "learning_rate": 9.971740810281083e-05, | |
| "loss": 1.0377, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.23293432546101586, | |
| "grad_norm": 0.828642725944519, | |
| "learning_rate": 9.971301406367644e-05, | |
| "loss": 0.8619, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2342284050469104, | |
| "grad_norm": 0.6980477571487427, | |
| "learning_rate": 9.970858622429579e-05, | |
| "loss": 0.8271, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.23552248463280492, | |
| "grad_norm": 0.954387903213501, | |
| "learning_rate": 9.970412458767943e-05, | |
| "loss": 0.8465, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.23681656421869945, | |
| "grad_norm": 0.8425692915916443, | |
| "learning_rate": 9.969962915686083e-05, | |
| "loss": 0.8893, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.23811064380459399, | |
| "grad_norm": 0.8565071225166321, | |
| "learning_rate": 9.969509993489647e-05, | |
| "loss": 0.939, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.23940472339048852, | |
| "grad_norm": 0.8831691145896912, | |
| "learning_rate": 9.969053692486583e-05, | |
| "loss": 0.8907, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.24069880297638305, | |
| "grad_norm": 0.9661678075790405, | |
| "learning_rate": 9.96859401298713e-05, | |
| "loss": 0.9191, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.24199288256227758, | |
| "grad_norm": 0.8784729838371277, | |
| "learning_rate": 9.968130955303828e-05, | |
| "loss": 1.0393, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.2432869621481721, | |
| "grad_norm": 0.8830071091651917, | |
| "learning_rate": 9.967664519751515e-05, | |
| "loss": 0.9837, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.24458104173406664, | |
| "grad_norm": 0.862108588218689, | |
| "learning_rate": 9.967194706647322e-05, | |
| "loss": 0.7871, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.24587512131996117, | |
| "grad_norm": 1.0068063735961914, | |
| "learning_rate": 9.966721516310682e-05, | |
| "loss": 0.9526, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.2471692009058557, | |
| "grad_norm": 0.9828710556030273, | |
| "learning_rate": 9.966244949063316e-05, | |
| "loss": 0.8923, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.24846328049175023, | |
| "grad_norm": 1.0729883909225464, | |
| "learning_rate": 9.965765005229248e-05, | |
| "loss": 1.0115, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.24975736007764476, | |
| "grad_norm": 0.9844326972961426, | |
| "learning_rate": 9.965281685134796e-05, | |
| "loss": 0.9855, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.2510514396635393, | |
| "grad_norm": 1.1593172550201416, | |
| "learning_rate": 9.96479498910857e-05, | |
| "loss": 1.0912, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.2523455192494338, | |
| "grad_norm": 0.8835370540618896, | |
| "learning_rate": 9.964304917481482e-05, | |
| "loss": 0.9951, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.2536395988353284, | |
| "grad_norm": 0.9553850889205933, | |
| "learning_rate": 9.963811470586733e-05, | |
| "loss": 0.9335, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.2549336784212229, | |
| "grad_norm": 0.863814115524292, | |
| "learning_rate": 9.963314648759823e-05, | |
| "loss": 1.0203, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.25622775800711745, | |
| "grad_norm": 0.9639378786087036, | |
| "learning_rate": 9.962814452338542e-05, | |
| "loss": 1.0357, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.25752183759301195, | |
| "grad_norm": 0.880519688129425, | |
| "learning_rate": 9.96231088166298e-05, | |
| "loss": 0.9964, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.2588159171789065, | |
| "grad_norm": 0.8445360064506531, | |
| "learning_rate": 9.961803937075516e-05, | |
| "loss": 0.9766, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.260109996764801, | |
| "grad_norm": 0.8204835057258606, | |
| "learning_rate": 9.961293618920826e-05, | |
| "loss": 0.8864, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.26140407635069557, | |
| "grad_norm": 0.9315406084060669, | |
| "learning_rate": 9.960779927545883e-05, | |
| "loss": 1.0388, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.2626981559365901, | |
| "grad_norm": 0.9286762475967407, | |
| "learning_rate": 9.960262863299943e-05, | |
| "loss": 0.9653, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.26399223552248463, | |
| "grad_norm": 0.8037816882133484, | |
| "learning_rate": 9.959742426534566e-05, | |
| "loss": 0.765, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.2652863151083792, | |
| "grad_norm": 0.9435904622077942, | |
| "learning_rate": 9.9592186176036e-05, | |
| "loss": 0.8889, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.2665803946942737, | |
| "grad_norm": 1.0072762966156006, | |
| "learning_rate": 9.958691436863188e-05, | |
| "loss": 0.8358, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.26787447428016825, | |
| "grad_norm": 0.9463568329811096, | |
| "learning_rate": 9.958160884671761e-05, | |
| "loss": 0.8815, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.26916855386606275, | |
| "grad_norm": 0.9203188419342041, | |
| "learning_rate": 9.957626961390047e-05, | |
| "loss": 0.9312, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.2704626334519573, | |
| "grad_norm": 1.0614677667617798, | |
| "learning_rate": 9.957089667381064e-05, | |
| "loss": 0.9822, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.2717567130378518, | |
| "grad_norm": 0.8971818089485168, | |
| "learning_rate": 9.956549003010123e-05, | |
| "loss": 0.9421, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.2730507926237464, | |
| "grad_norm": 0.9978768825531006, | |
| "learning_rate": 9.956004968644825e-05, | |
| "loss": 0.9539, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.2743448722096409, | |
| "grad_norm": 0.7017274498939514, | |
| "learning_rate": 9.955457564655064e-05, | |
| "loss": 0.665, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.27563895179553544, | |
| "grad_norm": 0.8292055726051331, | |
| "learning_rate": 9.954906791413023e-05, | |
| "loss": 0.922, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.27693303138142994, | |
| "grad_norm": 0.978084146976471, | |
| "learning_rate": 9.954352649293178e-05, | |
| "loss": 0.9465, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.2782271109673245, | |
| "grad_norm": 1.0260313749313354, | |
| "learning_rate": 9.953795138672291e-05, | |
| "loss": 0.9093, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.279521190553219, | |
| "grad_norm": 1.162850022315979, | |
| "learning_rate": 9.95323425992942e-05, | |
| "loss": 1.0372, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.28081527013911356, | |
| "grad_norm": 0.9785279631614685, | |
| "learning_rate": 9.952670013445913e-05, | |
| "loss": 0.8818, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.28210934972500806, | |
| "grad_norm": 0.9386499524116516, | |
| "learning_rate": 9.9521023996054e-05, | |
| "loss": 0.8711, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.2834034293109026, | |
| "grad_norm": 0.8620506525039673, | |
| "learning_rate": 9.951531418793812e-05, | |
| "loss": 1.011, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.2846975088967972, | |
| "grad_norm": 0.9523435831069946, | |
| "learning_rate": 9.950957071399357e-05, | |
| "loss": 0.8541, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.2859915884826917, | |
| "grad_norm": 0.8993477821350098, | |
| "learning_rate": 9.950379357812543e-05, | |
| "loss": 1.0253, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.28728566806858624, | |
| "grad_norm": 1.073880910873413, | |
| "learning_rate": 9.949798278426158e-05, | |
| "loss": 1.115, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.28857974765448074, | |
| "grad_norm": 0.7941976189613342, | |
| "learning_rate": 9.949213833635285e-05, | |
| "loss": 0.9398, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.2898738272403753, | |
| "grad_norm": 0.798089325428009, | |
| "learning_rate": 9.948626023837291e-05, | |
| "loss": 0.8523, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.2911679068262698, | |
| "grad_norm": 1.0251280069351196, | |
| "learning_rate": 9.948034849431831e-05, | |
| "loss": 0.939, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.29246198641216437, | |
| "grad_norm": 0.9793195724487305, | |
| "learning_rate": 9.947440310820852e-05, | |
| "loss": 1.0998, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.29375606599805887, | |
| "grad_norm": 0.8190125823020935, | |
| "learning_rate": 9.946842408408583e-05, | |
| "loss": 0.9606, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.2950501455839534, | |
| "grad_norm": 0.8229602575302124, | |
| "learning_rate": 9.946241142601543e-05, | |
| "loss": 0.7944, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.29634422516984793, | |
| "grad_norm": 0.8640865683555603, | |
| "learning_rate": 9.945636513808537e-05, | |
| "loss": 1.112, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.2976383047557425, | |
| "grad_norm": 0.774501621723175, | |
| "learning_rate": 9.945028522440653e-05, | |
| "loss": 0.8986, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.298932384341637, | |
| "grad_norm": 0.9039688110351562, | |
| "learning_rate": 9.944417168911275e-05, | |
| "loss": 1.0461, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.30022646392753155, | |
| "grad_norm": 0.8048250675201416, | |
| "learning_rate": 9.943802453636065e-05, | |
| "loss": 0.89, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.30152054351342605, | |
| "grad_norm": 0.8166521787643433, | |
| "learning_rate": 9.94318437703297e-05, | |
| "loss": 0.9149, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.3028146230993206, | |
| "grad_norm": 0.7571333646774292, | |
| "learning_rate": 9.942562939522228e-05, | |
| "loss": 0.9655, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.3041087026852151, | |
| "grad_norm": 0.6913223266601562, | |
| "learning_rate": 9.941938141526354e-05, | |
| "loss": 0.869, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.3054027822711097, | |
| "grad_norm": 0.8476676344871521, | |
| "learning_rate": 9.94130998347016e-05, | |
| "loss": 0.8849, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.30669686185700423, | |
| "grad_norm": 0.8454031944274902, | |
| "learning_rate": 9.940678465780728e-05, | |
| "loss": 0.9102, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.30799094144289874, | |
| "grad_norm": 0.8514583706855774, | |
| "learning_rate": 9.940043588887438e-05, | |
| "loss": 0.9723, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.3092850210287933, | |
| "grad_norm": 0.7330415844917297, | |
| "learning_rate": 9.939405353221942e-05, | |
| "loss": 0.9537, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.3105791006146878, | |
| "grad_norm": 0.9652897715568542, | |
| "learning_rate": 9.938763759218185e-05, | |
| "loss": 0.9736, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.31187318020058236, | |
| "grad_norm": 0.7517886161804199, | |
| "learning_rate": 9.93811880731239e-05, | |
| "loss": 0.8436, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.31316725978647686, | |
| "grad_norm": 0.8159210681915283, | |
| "learning_rate": 9.937470497943064e-05, | |
| "loss": 0.7521, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.3144613393723714, | |
| "grad_norm": 0.9554911851882935, | |
| "learning_rate": 9.936818831550998e-05, | |
| "loss": 1.1076, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.3157554189582659, | |
| "grad_norm": 0.8745877742767334, | |
| "learning_rate": 9.936163808579266e-05, | |
| "loss": 0.8908, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.3170494985441605, | |
| "grad_norm": 0.8050674200057983, | |
| "learning_rate": 9.93550542947322e-05, | |
| "loss": 1.0134, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.318343578130055, | |
| "grad_norm": 1.009790062904358, | |
| "learning_rate": 9.9348436946805e-05, | |
| "loss": 1.0264, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.31963765771594954, | |
| "grad_norm": 0.8702448606491089, | |
| "learning_rate": 9.934178604651023e-05, | |
| "loss": 1.0067, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.32093173730184404, | |
| "grad_norm": 0.8105303049087524, | |
| "learning_rate": 9.933510159836989e-05, | |
| "loss": 0.8121, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.3222258168877386, | |
| "grad_norm": 0.7680085897445679, | |
| "learning_rate": 9.932838360692878e-05, | |
| "loss": 0.8951, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.3235198964736331, | |
| "grad_norm": 0.8338052034378052, | |
| "learning_rate": 9.93216320767545e-05, | |
| "loss": 0.8878, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.32481397605952766, | |
| "grad_norm": 0.8709661960601807, | |
| "learning_rate": 9.93148470124375e-05, | |
| "loss": 0.8786, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.32610805564542217, | |
| "grad_norm": 0.9593453407287598, | |
| "learning_rate": 9.930802841859095e-05, | |
| "loss": 1.0659, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.3274021352313167, | |
| "grad_norm": 1.0130974054336548, | |
| "learning_rate": 9.93011762998509e-05, | |
| "loss": 0.9626, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.3286962148172113, | |
| "grad_norm": 0.9949910640716553, | |
| "learning_rate": 9.929429066087616e-05, | |
| "loss": 1.0499, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.3299902944031058, | |
| "grad_norm": 0.927542507648468, | |
| "learning_rate": 9.92873715063483e-05, | |
| "loss": 0.9571, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.33128437398900035, | |
| "grad_norm": 1.0638468265533447, | |
| "learning_rate": 9.92804188409717e-05, | |
| "loss": 1.0293, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.33257845357489485, | |
| "grad_norm": 0.7083877325057983, | |
| "learning_rate": 9.927343266947356e-05, | |
| "loss": 0.875, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.3338725331607894, | |
| "grad_norm": 0.7915517091751099, | |
| "learning_rate": 9.92664129966038e-05, | |
| "loss": 0.8848, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.3351666127466839, | |
| "grad_norm": 0.8054295182228088, | |
| "learning_rate": 9.925935982713518e-05, | |
| "loss": 0.8981, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.33646069233257847, | |
| "grad_norm": 1.012574553489685, | |
| "learning_rate": 9.925227316586316e-05, | |
| "loss": 0.8119, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.337754771918473, | |
| "grad_norm": 0.8329979181289673, | |
| "learning_rate": 9.924515301760606e-05, | |
| "loss": 0.8467, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.33904885150436753, | |
| "grad_norm": 0.801017701625824, | |
| "learning_rate": 9.923799938720488e-05, | |
| "loss": 0.8333, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.34034293109026204, | |
| "grad_norm": 0.9083892703056335, | |
| "learning_rate": 9.923081227952347e-05, | |
| "loss": 0.8727, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.3416370106761566, | |
| "grad_norm": 0.7917154431343079, | |
| "learning_rate": 9.922359169944834e-05, | |
| "loss": 1.0341, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.3429310902620511, | |
| "grad_norm": 0.6865798234939575, | |
| "learning_rate": 9.921633765188886e-05, | |
| "loss": 0.9117, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.34422516984794566, | |
| "grad_norm": 0.9448872208595276, | |
| "learning_rate": 9.92090501417771e-05, | |
| "loss": 1.1399, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.34551924943384016, | |
| "grad_norm": 0.8711137175559998, | |
| "learning_rate": 9.920172917406789e-05, | |
| "loss": 0.8158, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.3468133290197347, | |
| "grad_norm": 0.7883780598640442, | |
| "learning_rate": 9.919437475373882e-05, | |
| "loss": 1.0259, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.3481074086056292, | |
| "grad_norm": 0.7776859402656555, | |
| "learning_rate": 9.91869868857902e-05, | |
| "loss": 0.9145, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.3494014881915238, | |
| "grad_norm": 0.8383511900901794, | |
| "learning_rate": 9.91795655752451e-05, | |
| "loss": 0.8769, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.35069556777741834, | |
| "grad_norm": 0.8562968373298645, | |
| "learning_rate": 9.917211082714933e-05, | |
| "loss": 1.0078, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.35198964736331284, | |
| "grad_norm": 0.810517430305481, | |
| "learning_rate": 9.916462264657142e-05, | |
| "loss": 0.8847, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.3532837269492074, | |
| "grad_norm": 0.8934103846549988, | |
| "learning_rate": 9.915710103860263e-05, | |
| "loss": 0.8633, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.3545778065351019, | |
| "grad_norm": 0.8963167667388916, | |
| "learning_rate": 9.914954600835699e-05, | |
| "loss": 0.8411, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.35587188612099646, | |
| "grad_norm": 0.774557888507843, | |
| "learning_rate": 9.91419575609712e-05, | |
| "loss": 0.958, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.35716596570689096, | |
| "grad_norm": 0.8483523726463318, | |
| "learning_rate": 9.913433570160469e-05, | |
| "loss": 0.905, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.3584600452927855, | |
| "grad_norm": 0.8082010746002197, | |
| "learning_rate": 9.912668043543964e-05, | |
| "loss": 0.8633, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.35975412487868, | |
| "grad_norm": 0.8654133677482605, | |
| "learning_rate": 9.911899176768091e-05, | |
| "loss": 0.8465, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.3610482044645746, | |
| "grad_norm": 0.8709694743156433, | |
| "learning_rate": 9.911126970355609e-05, | |
| "loss": 0.9286, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.3623422840504691, | |
| "grad_norm": 0.8783992528915405, | |
| "learning_rate": 9.910351424831546e-05, | |
| "loss": 0.9349, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.36363636363636365, | |
| "grad_norm": 0.8846459984779358, | |
| "learning_rate": 9.909572540723202e-05, | |
| "loss": 0.7986, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.36493044322225815, | |
| "grad_norm": 0.8271780610084534, | |
| "learning_rate": 9.908790318560146e-05, | |
| "loss": 0.8179, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.3662245228081527, | |
| "grad_norm": 0.8694506883621216, | |
| "learning_rate": 9.908004758874216e-05, | |
| "loss": 0.8453, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.3675186023940472, | |
| "grad_norm": 0.8625207543373108, | |
| "learning_rate": 9.90721586219952e-05, | |
| "loss": 0.7825, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.36881268197994177, | |
| "grad_norm": 0.8270084261894226, | |
| "learning_rate": 9.906423629072434e-05, | |
| "loss": 1.0889, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.3701067615658363, | |
| "grad_norm": 0.8891452550888062, | |
| "learning_rate": 9.905628060031605e-05, | |
| "loss": 0.8847, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.37140084115173083, | |
| "grad_norm": 0.8682214021682739, | |
| "learning_rate": 9.904829155617945e-05, | |
| "loss": 0.9311, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.3726949207376254, | |
| "grad_norm": 1.2218021154403687, | |
| "learning_rate": 9.904026916374636e-05, | |
| "loss": 0.92, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.3739890003235199, | |
| "grad_norm": 1.0069034099578857, | |
| "learning_rate": 9.903221342847125e-05, | |
| "loss": 1.0061, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.37528307990941445, | |
| "grad_norm": 0.7723405361175537, | |
| "learning_rate": 9.902412435583128e-05, | |
| "loss": 0.8627, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.37657715949530896, | |
| "grad_norm": 0.8336161971092224, | |
| "learning_rate": 9.901600195132627e-05, | |
| "loss": 0.9815, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.3778712390812035, | |
| "grad_norm": 0.8170490264892578, | |
| "learning_rate": 9.90078462204787e-05, | |
| "loss": 0.8305, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.379165318667098, | |
| "grad_norm": 0.7367318868637085, | |
| "learning_rate": 9.899965716883372e-05, | |
| "loss": 0.8314, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.3804593982529926, | |
| "grad_norm": 0.8122355937957764, | |
| "learning_rate": 9.899143480195913e-05, | |
| "loss": 0.8369, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.3817534778388871, | |
| "grad_norm": 0.7415926456451416, | |
| "learning_rate": 9.898317912544536e-05, | |
| "loss": 0.8985, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.38304755742478164, | |
| "grad_norm": 0.8654418587684631, | |
| "learning_rate": 9.897489014490553e-05, | |
| "loss": 0.8752, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.38434163701067614, | |
| "grad_norm": 0.7863161563873291, | |
| "learning_rate": 9.896656786597535e-05, | |
| "loss": 1.0138, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.3856357165965707, | |
| "grad_norm": 0.8346667885780334, | |
| "learning_rate": 9.895821229431323e-05, | |
| "loss": 0.94, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.3869297961824652, | |
| "grad_norm": 0.9577547907829285, | |
| "learning_rate": 9.894982343560016e-05, | |
| "loss": 0.9967, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.38822387576835976, | |
| "grad_norm": 0.7633039951324463, | |
| "learning_rate": 9.894140129553981e-05, | |
| "loss": 0.8469, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.38951795535425426, | |
| "grad_norm": 0.8275448083877563, | |
| "learning_rate": 9.893294587985843e-05, | |
| "loss": 0.8295, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.3908120349401488, | |
| "grad_norm": 0.9197372198104858, | |
| "learning_rate": 9.892445719430493e-05, | |
| "loss": 0.9363, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.3921061145260433, | |
| "grad_norm": 0.7137726545333862, | |
| "learning_rate": 9.891593524465083e-05, | |
| "loss": 0.8207, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.3934001941119379, | |
| "grad_norm": 0.9908462762832642, | |
| "learning_rate": 9.890738003669029e-05, | |
| "loss": 0.9287, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.39469427369783244, | |
| "grad_norm": 1.0070343017578125, | |
| "learning_rate": 9.889879157624002e-05, | |
| "loss": 1.2092, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.39598835328372695, | |
| "grad_norm": 0.7812051177024841, | |
| "learning_rate": 9.889016986913941e-05, | |
| "loss": 0.8149, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.3972824328696215, | |
| "grad_norm": 0.8907694816589355, | |
| "learning_rate": 9.888151492125039e-05, | |
| "loss": 0.9008, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.398576512455516, | |
| "grad_norm": 0.960164487361908, | |
| "learning_rate": 9.887282673845754e-05, | |
| "loss": 0.9437, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.39987059204141057, | |
| "grad_norm": 0.8578314185142517, | |
| "learning_rate": 9.886410532666805e-05, | |
| "loss": 0.8422, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.40116467162730507, | |
| "grad_norm": 0.8671669960021973, | |
| "learning_rate": 9.885535069181162e-05, | |
| "loss": 0.924, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.40245875121319963, | |
| "grad_norm": 1.037150263786316, | |
| "learning_rate": 9.884656283984062e-05, | |
| "loss": 1.0074, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.40375283079909413, | |
| "grad_norm": 0.7935649156570435, | |
| "learning_rate": 9.883774177672998e-05, | |
| "loss": 0.9804, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.4050469103849887, | |
| "grad_norm": 0.8249583840370178, | |
| "learning_rate": 9.882888750847717e-05, | |
| "loss": 0.9627, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.4063409899708832, | |
| "grad_norm": 0.7363491058349609, | |
| "learning_rate": 9.882000004110233e-05, | |
| "loss": 0.8309, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.40763506955677775, | |
| "grad_norm": 0.8029589653015137, | |
| "learning_rate": 9.881107938064806e-05, | |
| "loss": 0.8762, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.40892914914267225, | |
| "grad_norm": 0.7131720781326294, | |
| "learning_rate": 9.880212553317963e-05, | |
| "loss": 0.882, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.4102232287285668, | |
| "grad_norm": 0.7951234579086304, | |
| "learning_rate": 9.879313850478478e-05, | |
| "loss": 1.0095, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.4115173083144613, | |
| "grad_norm": 0.7714753150939941, | |
| "learning_rate": 9.87841183015739e-05, | |
| "loss": 0.8688, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.4128113879003559, | |
| "grad_norm": 0.7753300666809082, | |
| "learning_rate": 9.877506492967987e-05, | |
| "loss": 0.9669, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.4141054674862504, | |
| "grad_norm": 0.9767395853996277, | |
| "learning_rate": 9.876597839525814e-05, | |
| "loss": 1.1169, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.41539954707214494, | |
| "grad_norm": 0.7923420667648315, | |
| "learning_rate": 9.875685870448672e-05, | |
| "loss": 0.9942, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.4166936266580395, | |
| "grad_norm": 0.7265552282333374, | |
| "learning_rate": 9.874770586356616e-05, | |
| "loss": 1.0377, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.417987706243934, | |
| "grad_norm": 0.7586270570755005, | |
| "learning_rate": 9.873851987871954e-05, | |
| "loss": 0.9172, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.41928178582982856, | |
| "grad_norm": 0.782192587852478, | |
| "learning_rate": 9.872930075619249e-05, | |
| "loss": 0.9219, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.42057586541572306, | |
| "grad_norm": 0.8508116602897644, | |
| "learning_rate": 9.872004850225313e-05, | |
| "loss": 0.939, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.4218699450016176, | |
| "grad_norm": 0.8639410138130188, | |
| "learning_rate": 9.871076312319218e-05, | |
| "loss": 1.0854, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.4231640245875121, | |
| "grad_norm": 0.809862494468689, | |
| "learning_rate": 9.870144462532281e-05, | |
| "loss": 0.8283, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.4244581041734067, | |
| "grad_norm": 0.7755741477012634, | |
| "learning_rate": 9.869209301498072e-05, | |
| "loss": 0.7645, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.4257521837593012, | |
| "grad_norm": 0.8238282203674316, | |
| "learning_rate": 9.868270829852416e-05, | |
| "loss": 0.8858, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.42704626334519574, | |
| "grad_norm": 0.810691773891449, | |
| "learning_rate": 9.867329048233387e-05, | |
| "loss": 0.7913, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.42834034293109025, | |
| "grad_norm": 0.8913648128509521, | |
| "learning_rate": 9.866383957281309e-05, | |
| "loss": 0.7885, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.4296344225169848, | |
| "grad_norm": 0.9160618782043457, | |
| "learning_rate": 9.865435557638757e-05, | |
| "loss": 0.7296, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.4309285021028793, | |
| "grad_norm": 0.7498170137405396, | |
| "learning_rate": 9.864483849950553e-05, | |
| "loss": 0.9655, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.43222258168877387, | |
| "grad_norm": 0.7315449714660645, | |
| "learning_rate": 9.863528834863773e-05, | |
| "loss": 0.6886, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.43351666127466837, | |
| "grad_norm": 0.8278869390487671, | |
| "learning_rate": 9.862570513027735e-05, | |
| "loss": 0.9637, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.4348107408605629, | |
| "grad_norm": 0.8372804522514343, | |
| "learning_rate": 9.861608885094012e-05, | |
| "loss": 0.8609, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.43610482044645743, | |
| "grad_norm": 0.8712325096130371, | |
| "learning_rate": 9.860643951716421e-05, | |
| "loss": 0.9718, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.437398900032352, | |
| "grad_norm": 0.9869045615196228, | |
| "learning_rate": 9.859675713551028e-05, | |
| "loss": 0.887, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.43869297961824655, | |
| "grad_norm": 0.9166460037231445, | |
| "learning_rate": 9.858704171256145e-05, | |
| "loss": 1.0751, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.43998705920414105, | |
| "grad_norm": 1.1965091228485107, | |
| "learning_rate": 9.857729325492329e-05, | |
| "loss": 1.0093, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.4412811387900356, | |
| "grad_norm": 0.8646867275238037, | |
| "learning_rate": 9.856751176922388e-05, | |
| "loss": 0.9235, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.4425752183759301, | |
| "grad_norm": 0.7576479315757751, | |
| "learning_rate": 9.85576972621137e-05, | |
| "loss": 0.8323, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.44386929796182467, | |
| "grad_norm": 0.8257366418838501, | |
| "learning_rate": 9.854784974026572e-05, | |
| "loss": 0.8478, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.4451633775477192, | |
| "grad_norm": 0.8963577747344971, | |
| "learning_rate": 9.853796921037534e-05, | |
| "loss": 0.8943, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.44645745713361373, | |
| "grad_norm": 0.8696343898773193, | |
| "learning_rate": 9.85280556791604e-05, | |
| "loss": 0.991, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.44775153671950824, | |
| "grad_norm": 0.8321841359138489, | |
| "learning_rate": 9.851810915336119e-05, | |
| "loss": 1.0462, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.4490456163054028, | |
| "grad_norm": 0.8194176554679871, | |
| "learning_rate": 9.850812963974042e-05, | |
| "loss": 0.9484, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.4503396958912973, | |
| "grad_norm": 0.9131197333335876, | |
| "learning_rate": 9.849811714508323e-05, | |
| "loss": 0.9687, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.45163377547719186, | |
| "grad_norm": 0.891231894493103, | |
| "learning_rate": 9.848807167619721e-05, | |
| "loss": 0.9605, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.45292785506308636, | |
| "grad_norm": 0.9574115872383118, | |
| "learning_rate": 9.847799323991234e-05, | |
| "loss": 1.077, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.4542219346489809, | |
| "grad_norm": 0.9179444909095764, | |
| "learning_rate": 9.8467881843081e-05, | |
| "loss": 0.8353, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.4555160142348754, | |
| "grad_norm": 0.855802595615387, | |
| "learning_rate": 9.845773749257804e-05, | |
| "loss": 0.9764, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.45681009382077, | |
| "grad_norm": 0.9086332321166992, | |
| "learning_rate": 9.844756019530066e-05, | |
| "loss": 1.0526, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.4581041734066645, | |
| "grad_norm": 0.890271782875061, | |
| "learning_rate": 9.843734995816848e-05, | |
| "loss": 0.8905, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.45939825299255904, | |
| "grad_norm": 0.7878096699714661, | |
| "learning_rate": 9.842710678812351e-05, | |
| "loss": 0.8706, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.4606923325784536, | |
| "grad_norm": 0.9886014461517334, | |
| "learning_rate": 9.841683069213017e-05, | |
| "loss": 0.9579, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.4619864121643481, | |
| "grad_norm": 0.8265432119369507, | |
| "learning_rate": 9.840652167717526e-05, | |
| "loss": 0.8528, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.46328049175024266, | |
| "grad_norm": 0.7354372143745422, | |
| "learning_rate": 9.839617975026793e-05, | |
| "loss": 0.775, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.46457457133613717, | |
| "grad_norm": 0.8311409950256348, | |
| "learning_rate": 9.838580491843976e-05, | |
| "loss": 1.0374, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.4658686509220317, | |
| "grad_norm": 0.8180521130561829, | |
| "learning_rate": 9.837539718874464e-05, | |
| "loss": 0.818, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.4671627305079262, | |
| "grad_norm": 0.7231691479682922, | |
| "learning_rate": 9.83649565682589e-05, | |
| "loss": 0.9141, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.4684568100938208, | |
| "grad_norm": 0.7644848227500916, | |
| "learning_rate": 9.835448306408118e-05, | |
| "loss": 0.9565, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.4697508896797153, | |
| "grad_norm": 0.7103495001792908, | |
| "learning_rate": 9.83439766833325e-05, | |
| "loss": 0.7706, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.47104496926560985, | |
| "grad_norm": 0.8206077218055725, | |
| "learning_rate": 9.83334374331562e-05, | |
| "loss": 1.0923, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.47233904885150435, | |
| "grad_norm": 0.7409754991531372, | |
| "learning_rate": 9.832286532071802e-05, | |
| "loss": 0.8534, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.4736331284373989, | |
| "grad_norm": 0.8880231380462646, | |
| "learning_rate": 9.831226035320602e-05, | |
| "loss": 0.9025, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.4749272080232934, | |
| "grad_norm": 0.9805209636688232, | |
| "learning_rate": 9.830162253783058e-05, | |
| "loss": 0.9939, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.47622128760918797, | |
| "grad_norm": 0.9529579281806946, | |
| "learning_rate": 9.829095188182442e-05, | |
| "loss": 0.9645, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.4775153671950825, | |
| "grad_norm": 0.8122926354408264, | |
| "learning_rate": 9.828024839244263e-05, | |
| "loss": 0.9269, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.47880944678097703, | |
| "grad_norm": 0.8557335734367371, | |
| "learning_rate": 9.826951207696258e-05, | |
| "loss": 0.8781, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.48010352636687154, | |
| "grad_norm": 0.959077775478363, | |
| "learning_rate": 9.825874294268396e-05, | |
| "loss": 0.9448, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.4813976059527661, | |
| "grad_norm": 0.7479959726333618, | |
| "learning_rate": 9.824794099692878e-05, | |
| "loss": 0.8608, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.48269168553866065, | |
| "grad_norm": 0.8064723014831543, | |
| "learning_rate": 9.823710624704137e-05, | |
| "loss": 1.1222, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.48398576512455516, | |
| "grad_norm": 0.8007084131240845, | |
| "learning_rate": 9.822623870038838e-05, | |
| "loss": 0.8967, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.4852798447104497, | |
| "grad_norm": 0.988571286201477, | |
| "learning_rate": 9.82153383643587e-05, | |
| "loss": 0.9574, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.4865739242963442, | |
| "grad_norm": 0.9317258596420288, | |
| "learning_rate": 9.820440524636356e-05, | |
| "loss": 0.929, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.4878680038822388, | |
| "grad_norm": 0.8651334643363953, | |
| "learning_rate": 9.819343935383649e-05, | |
| "loss": 0.9483, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.4891620834681333, | |
| "grad_norm": 0.791306734085083, | |
| "learning_rate": 9.818244069423325e-05, | |
| "loss": 0.9277, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.49045616305402784, | |
| "grad_norm": 0.6706424355506897, | |
| "learning_rate": 9.817140927503192e-05, | |
| "loss": 0.9013, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.49175024263992234, | |
| "grad_norm": 0.8574838042259216, | |
| "learning_rate": 9.816034510373286e-05, | |
| "loss": 0.9889, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.4930443222258169, | |
| "grad_norm": 0.829309344291687, | |
| "learning_rate": 9.814924818785865e-05, | |
| "loss": 0.8139, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.4943384018117114, | |
| "grad_norm": 0.83943110704422, | |
| "learning_rate": 9.81381185349542e-05, | |
| "loss": 0.9202, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.49563248139760596, | |
| "grad_norm": 0.7981933355331421, | |
| "learning_rate": 9.812695615258662e-05, | |
| "loss": 0.9131, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.49692656098350046, | |
| "grad_norm": 0.7930905818939209, | |
| "learning_rate": 9.81157610483453e-05, | |
| "loss": 0.769, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.498220640569395, | |
| "grad_norm": 0.8699679970741272, | |
| "learning_rate": 9.81045332298419e-05, | |
| "loss": 0.9468, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.4995147201552895, | |
| "grad_norm": 0.7733138799667358, | |
| "learning_rate": 9.809327270471025e-05, | |
| "loss": 0.8982, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.4995147201552895, | |
| "eval_loss": 0.8872498869895935, | |
| "eval_runtime": 189.8035, | |
| "eval_samples_per_second": 3.43, | |
| "eval_steps_per_second": 0.859, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.5008087997411841, | |
| "grad_norm": 0.9202268123626709, | |
| "learning_rate": 9.808197948060651e-05, | |
| "loss": 0.9664, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.5021028793270786, | |
| "grad_norm": 0.7772640585899353, | |
| "learning_rate": 9.807065356520899e-05, | |
| "loss": 0.8825, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.5033969589129732, | |
| "grad_norm": 0.8117741346359253, | |
| "learning_rate": 9.805929496621828e-05, | |
| "loss": 0.8166, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.5046910384988677, | |
| "grad_norm": 0.7308422327041626, | |
| "learning_rate": 9.804790369135718e-05, | |
| "loss": 0.8657, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.5059851180847622, | |
| "grad_norm": 0.8373358845710754, | |
| "learning_rate": 9.80364797483707e-05, | |
| "loss": 0.9175, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.5072791976706568, | |
| "grad_norm": 0.8288902640342712, | |
| "learning_rate": 9.802502314502607e-05, | |
| "loss": 0.7463, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.5085732772565513, | |
| "grad_norm": 0.6780114769935608, | |
| "learning_rate": 9.801353388911269e-05, | |
| "loss": 0.7973, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.5098673568424458, | |
| "grad_norm": 0.9328367710113525, | |
| "learning_rate": 9.800201198844221e-05, | |
| "loss": 1.0405, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.5111614364283403, | |
| "grad_norm": 0.9010327458381653, | |
| "learning_rate": 9.799045745084847e-05, | |
| "loss": 1.1194, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.5124555160142349, | |
| "grad_norm": 0.7900609374046326, | |
| "learning_rate": 9.797887028418746e-05, | |
| "loss": 0.9747, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.5137495956001294, | |
| "grad_norm": 0.8245125412940979, | |
| "learning_rate": 9.796725049633741e-05, | |
| "loss": 0.8594, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.5150436751860239, | |
| "grad_norm": 0.7617928385734558, | |
| "learning_rate": 9.795559809519866e-05, | |
| "loss": 0.8176, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.5163377547719185, | |
| "grad_norm": 0.9844444990158081, | |
| "learning_rate": 9.79439130886938e-05, | |
| "loss": 1.0116, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.517631834357813, | |
| "grad_norm": 0.8849684000015259, | |
| "learning_rate": 9.793219548476753e-05, | |
| "loss": 0.8706, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.5189259139437076, | |
| "grad_norm": 0.893902063369751, | |
| "learning_rate": 9.792044529138674e-05, | |
| "loss": 0.8217, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.520219993529602, | |
| "grad_norm": 0.7412934899330139, | |
| "learning_rate": 9.79086625165405e-05, | |
| "loss": 0.868, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.5215140731154966, | |
| "grad_norm": 0.796435534954071, | |
| "learning_rate": 9.789684716823995e-05, | |
| "loss": 0.8691, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.5228081527013911, | |
| "grad_norm": 1.063193440437317, | |
| "learning_rate": 9.788499925451849e-05, | |
| "loss": 1.0085, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.5241022322872857, | |
| "grad_norm": 0.952882707118988, | |
| "learning_rate": 9.787311878343157e-05, | |
| "loss": 0.8378, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.5253963118731803, | |
| "grad_norm": 0.7899916768074036, | |
| "learning_rate": 9.786120576305682e-05, | |
| "loss": 0.8917, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.5266903914590747, | |
| "grad_norm": 0.8242781758308411, | |
| "learning_rate": 9.784926020149398e-05, | |
| "loss": 0.9778, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.5279844710449693, | |
| "grad_norm": 0.9736928343772888, | |
| "learning_rate": 9.783728210686496e-05, | |
| "loss": 1.0145, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.5292785506308638, | |
| "grad_norm": 0.8070263862609863, | |
| "learning_rate": 9.782527148731372e-05, | |
| "loss": 0.9923, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.5305726302167584, | |
| "grad_norm": 0.861262857913971, | |
| "learning_rate": 9.781322835100638e-05, | |
| "loss": 1.0261, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.5318667098026528, | |
| "grad_norm": 0.8204033970832825, | |
| "learning_rate": 9.780115270613115e-05, | |
| "loss": 0.8865, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.5331607893885474, | |
| "grad_norm": 0.8053429126739502, | |
| "learning_rate": 9.778904456089838e-05, | |
| "loss": 0.8606, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.5344548689744419, | |
| "grad_norm": 0.7944838404655457, | |
| "learning_rate": 9.777690392354045e-05, | |
| "loss": 0.8925, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.5357489485603365, | |
| "grad_norm": 0.9882616400718689, | |
| "learning_rate": 9.77647308023119e-05, | |
| "loss": 0.8577, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.537043028146231, | |
| "grad_norm": 0.6937118768692017, | |
| "learning_rate": 9.77525252054893e-05, | |
| "loss": 0.6972, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.5383371077321255, | |
| "grad_norm": 1.0071479082107544, | |
| "learning_rate": 9.774028714137133e-05, | |
| "loss": 0.9233, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.5396311873180201, | |
| "grad_norm": 0.916771411895752, | |
| "learning_rate": 9.772801661827874e-05, | |
| "loss": 1.0072, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.5409252669039146, | |
| "grad_norm": 0.8663278818130493, | |
| "learning_rate": 9.771571364455439e-05, | |
| "loss": 1.1011, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.5422193464898091, | |
| "grad_norm": 0.7842355370521545, | |
| "learning_rate": 9.77033782285631e-05, | |
| "loss": 1.0382, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.5435134260757036, | |
| "grad_norm": 0.8407487273216248, | |
| "learning_rate": 9.769101037869187e-05, | |
| "loss": 0.9612, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.5448075056615982, | |
| "grad_norm": 0.7557950615882874, | |
| "learning_rate": 9.767861010334962e-05, | |
| "loss": 0.9965, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.5461015852474927, | |
| "grad_norm": 0.812099814414978, | |
| "learning_rate": 9.766617741096746e-05, | |
| "loss": 0.8824, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.5473956648333873, | |
| "grad_norm": 0.8438544273376465, | |
| "learning_rate": 9.765371230999843e-05, | |
| "loss": 0.8852, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.5486897444192818, | |
| "grad_norm": 0.7006101608276367, | |
| "learning_rate": 9.764121480891765e-05, | |
| "loss": 0.808, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.5499838240051763, | |
| "grad_norm": 0.7312132716178894, | |
| "learning_rate": 9.76286849162223e-05, | |
| "loss": 0.7964, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.5512779035910709, | |
| "grad_norm": 0.939939558506012, | |
| "learning_rate": 9.76161226404315e-05, | |
| "loss": 0.9434, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.5525719831769654, | |
| "grad_norm": 0.7815152406692505, | |
| "learning_rate": 9.760352799008643e-05, | |
| "loss": 1.0487, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.5538660627628599, | |
| "grad_norm": 0.8934217691421509, | |
| "learning_rate": 9.759090097375032e-05, | |
| "loss": 1.0036, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.5551601423487544, | |
| "grad_norm": 0.7264319658279419, | |
| "learning_rate": 9.757824160000837e-05, | |
| "loss": 0.7866, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.556454221934649, | |
| "grad_norm": 0.8536062836647034, | |
| "learning_rate": 9.756554987746776e-05, | |
| "loss": 0.9949, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.5577483015205436, | |
| "grad_norm": 0.8253523111343384, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.9801, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.559042381106438, | |
| "grad_norm": 0.6686350703239441, | |
| "learning_rate": 9.754006942052936e-05, | |
| "loss": 0.8388, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.5603364606923326, | |
| "grad_norm": 0.8560882210731506, | |
| "learning_rate": 9.752728070345591e-05, | |
| "loss": 1.0167, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.5616305402782271, | |
| "grad_norm": 0.765852153301239, | |
| "learning_rate": 9.751445967223252e-05, | |
| "loss": 0.7865, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.5629246198641217, | |
| "grad_norm": 0.8869863152503967, | |
| "learning_rate": 9.750160633557627e-05, | |
| "loss": 0.9424, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.5642186994500161, | |
| "grad_norm": 0.8522469401359558, | |
| "learning_rate": 9.748872070222625e-05, | |
| "loss": 0.8151, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.5655127790359107, | |
| "grad_norm": 0.862091064453125, | |
| "learning_rate": 9.747580278094352e-05, | |
| "loss": 1.0131, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.5668068586218052, | |
| "grad_norm": 0.8809049725532532, | |
| "learning_rate": 9.746285258051104e-05, | |
| "loss": 0.9092, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.5681009382076998, | |
| "grad_norm": 0.8506961464881897, | |
| "learning_rate": 9.744987010973377e-05, | |
| "loss": 0.8756, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.5693950177935944, | |
| "grad_norm": 0.7440844774246216, | |
| "learning_rate": 9.743685537743856e-05, | |
| "loss": 0.7948, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.5706890973794888, | |
| "grad_norm": 0.8117014169692993, | |
| "learning_rate": 9.742380839247425e-05, | |
| "loss": 0.9382, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.5719831769653834, | |
| "grad_norm": 0.8515580296516418, | |
| "learning_rate": 9.741072916371157e-05, | |
| "loss": 0.9202, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.5732772565512779, | |
| "grad_norm": 0.8774899244308472, | |
| "learning_rate": 9.739761770004318e-05, | |
| "loss": 0.9129, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.5745713361371725, | |
| "grad_norm": 0.7189667820930481, | |
| "learning_rate": 9.738447401038367e-05, | |
| "loss": 0.7392, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.5758654157230669, | |
| "grad_norm": 0.7225956916809082, | |
| "learning_rate": 9.737129810366952e-05, | |
| "loss": 0.8703, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.5771594953089615, | |
| "grad_norm": 0.8351484537124634, | |
| "learning_rate": 9.735808998885915e-05, | |
| "loss": 0.9626, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.578453574894856, | |
| "grad_norm": 0.860819399356842, | |
| "learning_rate": 9.734484967493282e-05, | |
| "loss": 0.9061, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.5797476544807506, | |
| "grad_norm": 0.7928848266601562, | |
| "learning_rate": 9.733157717089277e-05, | |
| "loss": 0.9141, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.581041734066645, | |
| "grad_norm": 0.9132540822029114, | |
| "learning_rate": 9.7318272485763e-05, | |
| "loss": 0.7613, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.5823358136525396, | |
| "grad_norm": 0.7894258499145508, | |
| "learning_rate": 9.730493562858953e-05, | |
| "loss": 0.8234, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.5836298932384342, | |
| "grad_norm": 0.8427807688713074, | |
| "learning_rate": 9.729156660844017e-05, | |
| "loss": 1.1023, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.5849239728243287, | |
| "grad_norm": 0.8546382188796997, | |
| "learning_rate": 9.727816543440458e-05, | |
| "loss": 0.886, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.5862180524102232, | |
| "grad_norm": 0.6701838970184326, | |
| "learning_rate": 9.726473211559437e-05, | |
| "loss": 0.7861, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.5875121319961177, | |
| "grad_norm": 1.010311245918274, | |
| "learning_rate": 9.725126666114292e-05, | |
| "loss": 1.0393, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.5888062115820123, | |
| "grad_norm": 0.8440571427345276, | |
| "learning_rate": 9.72377690802055e-05, | |
| "loss": 0.9959, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.5901002911679069, | |
| "grad_norm": 0.8143338561058044, | |
| "learning_rate": 9.722423938195922e-05, | |
| "loss": 0.9954, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.5913943707538014, | |
| "grad_norm": 0.6839838027954102, | |
| "learning_rate": 9.721067757560303e-05, | |
| "loss": 0.7288, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.5926884503396959, | |
| "grad_norm": 0.8407920598983765, | |
| "learning_rate": 9.719708367035767e-05, | |
| "loss": 0.858, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.5939825299255904, | |
| "grad_norm": 0.8388239741325378, | |
| "learning_rate": 9.718345767546576e-05, | |
| "loss": 0.8455, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.595276609511485, | |
| "grad_norm": 0.7476726770401001, | |
| "learning_rate": 9.716979960019173e-05, | |
| "loss": 0.8261, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5965706890973795, | |
| "grad_norm": 0.766036331653595, | |
| "learning_rate": 9.715610945382177e-05, | |
| "loss": 0.9601, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.597864768683274, | |
| "grad_norm": 0.9155464172363281, | |
| "learning_rate": 9.714238724566393e-05, | |
| "loss": 0.8716, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.5991588482691685, | |
| "grad_norm": 0.8961794972419739, | |
| "learning_rate": 9.712863298504807e-05, | |
| "loss": 0.9595, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.6004529278550631, | |
| "grad_norm": 0.8102921843528748, | |
| "learning_rate": 9.711484668132575e-05, | |
| "loss": 0.7174, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.6017470074409577, | |
| "grad_norm": 0.7466800808906555, | |
| "learning_rate": 9.710102834387043e-05, | |
| "loss": 0.8707, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.6030410870268521, | |
| "grad_norm": 0.7572315335273743, | |
| "learning_rate": 9.708717798207729e-05, | |
| "loss": 1.0715, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.6043351666127467, | |
| "grad_norm": 0.944107174873352, | |
| "learning_rate": 9.707329560536328e-05, | |
| "loss": 0.9103, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.6056292461986412, | |
| "grad_norm": 0.7636998891830444, | |
| "learning_rate": 9.705938122316715e-05, | |
| "loss": 1.0246, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.6069233257845358, | |
| "grad_norm": 0.7484267950057983, | |
| "learning_rate": 9.704543484494937e-05, | |
| "loss": 1.0018, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.6082174053704302, | |
| "grad_norm": 0.89073646068573, | |
| "learning_rate": 9.70314564801922e-05, | |
| "loss": 0.8776, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.6095114849563248, | |
| "grad_norm": 0.8132034540176392, | |
| "learning_rate": 9.701744613839963e-05, | |
| "loss": 0.9575, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.6108055645422193, | |
| "grad_norm": 0.8051428198814392, | |
| "learning_rate": 9.700340382909739e-05, | |
| "loss": 0.7546, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.6120996441281139, | |
| "grad_norm": 0.7600153088569641, | |
| "learning_rate": 9.698932956183296e-05, | |
| "loss": 0.8556, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.6133937237140085, | |
| "grad_norm": 0.9171023964881897, | |
| "learning_rate": 9.697522334617555e-05, | |
| "loss": 0.8196, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.6146878032999029, | |
| "grad_norm": 1.0437992811203003, | |
| "learning_rate": 9.696108519171605e-05, | |
| "loss": 1.0997, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.6159818828857975, | |
| "grad_norm": 0.8450473546981812, | |
| "learning_rate": 9.69469151080671e-05, | |
| "loss": 1.0144, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.617275962471692, | |
| "grad_norm": 0.8718084692955017, | |
| "learning_rate": 9.693271310486307e-05, | |
| "loss": 1.0536, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.6185700420575866, | |
| "grad_norm": 0.858589768409729, | |
| "learning_rate": 9.691847919175999e-05, | |
| "loss": 0.9851, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.619864121643481, | |
| "grad_norm": 0.7903156280517578, | |
| "learning_rate": 9.69042133784356e-05, | |
| "loss": 1.2178, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.6211582012293756, | |
| "grad_norm": 0.6688570976257324, | |
| "learning_rate": 9.688991567458933e-05, | |
| "loss": 0.7721, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.6224522808152702, | |
| "grad_norm": 0.7499825954437256, | |
| "learning_rate": 9.687558608994232e-05, | |
| "loss": 0.8728, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.6237463604011647, | |
| "grad_norm": 0.871703028678894, | |
| "learning_rate": 9.686122463423732e-05, | |
| "loss": 0.88, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.6250404399870592, | |
| "grad_norm": 0.8204110860824585, | |
| "learning_rate": 9.684683131723884e-05, | |
| "loss": 0.8569, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.6263345195729537, | |
| "grad_norm": 0.8779101371765137, | |
| "learning_rate": 9.683240614873294e-05, | |
| "loss": 0.901, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.6276285991588483, | |
| "grad_norm": 0.8231368064880371, | |
| "learning_rate": 9.681794913852746e-05, | |
| "loss": 0.989, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.6289226787447428, | |
| "grad_norm": 0.7921927571296692, | |
| "learning_rate": 9.68034602964518e-05, | |
| "loss": 0.8748, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.6302167583306373, | |
| "grad_norm": 0.8911926746368408, | |
| "learning_rate": 9.678893963235704e-05, | |
| "loss": 1.0641, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.6315108379165318, | |
| "grad_norm": 0.8004783391952515, | |
| "learning_rate": 9.677438715611586e-05, | |
| "loss": 0.9681, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.6328049175024264, | |
| "grad_norm": 0.9230672717094421, | |
| "learning_rate": 9.675980287762263e-05, | |
| "loss": 0.9226, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.634098997088321, | |
| "grad_norm": 0.7298123240470886, | |
| "learning_rate": 9.67451868067933e-05, | |
| "loss": 0.7876, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.6353930766742155, | |
| "grad_norm": 0.7051340937614441, | |
| "learning_rate": 9.673053895356543e-05, | |
| "loss": 0.8534, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.63668715626011, | |
| "grad_norm": 0.9023073315620422, | |
| "learning_rate": 9.671585932789821e-05, | |
| "loss": 0.8995, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.6379812358460045, | |
| "grad_norm": 0.7804109454154968, | |
| "learning_rate": 9.670114793977243e-05, | |
| "loss": 0.788, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.6392753154318991, | |
| "grad_norm": 0.7766551375389099, | |
| "learning_rate": 9.66864047991905e-05, | |
| "loss": 0.7799, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.6405693950177936, | |
| "grad_norm": 0.8782010078430176, | |
| "learning_rate": 9.667162991617633e-05, | |
| "loss": 0.897, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.6418634746036881, | |
| "grad_norm": 0.7679833173751831, | |
| "learning_rate": 9.665682330077551e-05, | |
| "loss": 0.84, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.6431575541895826, | |
| "grad_norm": 0.9792498350143433, | |
| "learning_rate": 9.664198496305517e-05, | |
| "loss": 0.9597, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.6444516337754772, | |
| "grad_norm": 0.9134929180145264, | |
| "learning_rate": 9.662711491310398e-05, | |
| "loss": 1.0342, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.6457457133613718, | |
| "grad_norm": 0.7805169224739075, | |
| "learning_rate": 9.661221316103224e-05, | |
| "loss": 0.8463, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.6470397929472662, | |
| "grad_norm": 0.8533938527107239, | |
| "learning_rate": 9.659727971697174e-05, | |
| "loss": 0.8315, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.6483338725331608, | |
| "grad_norm": 0.7894063591957092, | |
| "learning_rate": 9.658231459107582e-05, | |
| "loss": 0.9049, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.6496279521190553, | |
| "grad_norm": 0.8634886145591736, | |
| "learning_rate": 9.656731779351939e-05, | |
| "loss": 0.9318, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.6509220317049499, | |
| "grad_norm": 0.8121569156646729, | |
| "learning_rate": 9.655228933449888e-05, | |
| "loss": 0.8475, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.6522161112908443, | |
| "grad_norm": 0.8126516342163086, | |
| "learning_rate": 9.653722922423229e-05, | |
| "loss": 0.8375, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.6535101908767389, | |
| "grad_norm": 0.7656330466270447, | |
| "learning_rate": 9.652213747295906e-05, | |
| "loss": 0.8605, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.6548042704626335, | |
| "grad_norm": 0.7796474695205688, | |
| "learning_rate": 9.650701409094018e-05, | |
| "loss": 0.803, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.656098350048528, | |
| "grad_norm": 0.8425498008728027, | |
| "learning_rate": 9.649185908845818e-05, | |
| "loss": 0.9324, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.6573924296344226, | |
| "grad_norm": 0.6857015490531921, | |
| "learning_rate": 9.647667247581703e-05, | |
| "loss": 0.8701, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.658686509220317, | |
| "grad_norm": 0.7896414995193481, | |
| "learning_rate": 9.646145426334223e-05, | |
| "loss": 0.7447, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.6599805888062116, | |
| "grad_norm": 0.7142564058303833, | |
| "learning_rate": 9.644620446138077e-05, | |
| "loss": 0.9221, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.6612746683921061, | |
| "grad_norm": 0.8938276767730713, | |
| "learning_rate": 9.643092308030109e-05, | |
| "loss": 1.0021, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.6625687479780007, | |
| "grad_norm": 1.0916978120803833, | |
| "learning_rate": 9.641561013049309e-05, | |
| "loss": 0.9354, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.6638628275638951, | |
| "grad_norm": 0.8139733672142029, | |
| "learning_rate": 9.64002656223682e-05, | |
| "loss": 0.923, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.6651569071497897, | |
| "grad_norm": 0.8234017491340637, | |
| "learning_rate": 9.638488956635925e-05, | |
| "loss": 0.8491, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.6664509867356843, | |
| "grad_norm": 0.7203138470649719, | |
| "learning_rate": 9.636948197292052e-05, | |
| "loss": 0.8313, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.6677450663215788, | |
| "grad_norm": 0.7210227251052856, | |
| "learning_rate": 9.635404285252777e-05, | |
| "loss": 0.8593, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.6690391459074733, | |
| "grad_norm": 0.7367059588432312, | |
| "learning_rate": 9.633857221567815e-05, | |
| "loss": 0.9743, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.6703332254933678, | |
| "grad_norm": 0.8669779896736145, | |
| "learning_rate": 9.632307007289027e-05, | |
| "loss": 0.8914, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.6716273050792624, | |
| "grad_norm": 0.700624406337738, | |
| "learning_rate": 9.630753643470416e-05, | |
| "loss": 0.9032, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.6729213846651569, | |
| "grad_norm": 0.81085205078125, | |
| "learning_rate": 9.629197131168124e-05, | |
| "loss": 0.8612, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.6742154642510514, | |
| "grad_norm": 0.760386049747467, | |
| "learning_rate": 9.627637471440436e-05, | |
| "loss": 0.8932, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.675509543836946, | |
| "grad_norm": 0.9040992259979248, | |
| "learning_rate": 9.626074665347778e-05, | |
| "loss": 0.8954, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.6768036234228405, | |
| "grad_norm": 0.9394627809524536, | |
| "learning_rate": 9.624508713952709e-05, | |
| "loss": 0.9641, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.6780977030087351, | |
| "grad_norm": 0.7919620871543884, | |
| "learning_rate": 9.622939618319936e-05, | |
| "loss": 1.0499, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.6793917825946296, | |
| "grad_norm": 0.7263360619544983, | |
| "learning_rate": 9.621367379516294e-05, | |
| "loss": 1.0297, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.6806858621805241, | |
| "grad_norm": 0.7512305974960327, | |
| "learning_rate": 9.619791998610763e-05, | |
| "loss": 0.7866, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.6819799417664186, | |
| "grad_norm": 0.8404532670974731, | |
| "learning_rate": 9.618213476674455e-05, | |
| "loss": 0.8134, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.6832740213523132, | |
| "grad_norm": 0.8437522053718567, | |
| "learning_rate": 9.61663181478062e-05, | |
| "loss": 0.908, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.6845681009382077, | |
| "grad_norm": 0.8053356409072876, | |
| "learning_rate": 9.61504701400464e-05, | |
| "loss": 0.8775, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.6858621805241022, | |
| "grad_norm": 0.7193287014961243, | |
| "learning_rate": 9.613459075424034e-05, | |
| "loss": 0.8122, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.6871562601099968, | |
| "grad_norm": 0.7333494424819946, | |
| "learning_rate": 9.611868000118452e-05, | |
| "loss": 0.8027, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.6884503396958913, | |
| "grad_norm": 0.7772257924079895, | |
| "learning_rate": 9.61027378916968e-05, | |
| "loss": 0.8538, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.6897444192817859, | |
| "grad_norm": 0.7605924606323242, | |
| "learning_rate": 9.60867644366163e-05, | |
| "loss": 0.875, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.6910384988676803, | |
| "grad_norm": 0.8444223999977112, | |
| "learning_rate": 9.607075964680352e-05, | |
| "loss": 1.0179, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.6923325784535749, | |
| "grad_norm": 0.7386454939842224, | |
| "learning_rate": 9.605472353314023e-05, | |
| "loss": 0.9023, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.6936266580394694, | |
| "grad_norm": 0.7770008444786072, | |
| "learning_rate": 9.603865610652952e-05, | |
| "loss": 0.9006, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.694920737625364, | |
| "grad_norm": 0.8127918243408203, | |
| "learning_rate": 9.60225573778957e-05, | |
| "loss": 0.8098, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.6962148172112584, | |
| "grad_norm": 0.7819888591766357, | |
| "learning_rate": 9.600642735818446e-05, | |
| "loss": 0.9933, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.697508896797153, | |
| "grad_norm": 0.7097641229629517, | |
| "learning_rate": 9.599026605836273e-05, | |
| "loss": 0.7463, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.6988029763830476, | |
| "grad_norm": 0.7568448781967163, | |
| "learning_rate": 9.597407348941865e-05, | |
| "loss": 0.7678, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.7000970559689421, | |
| "grad_norm": 0.7815922498703003, | |
| "learning_rate": 9.59578496623617e-05, | |
| "loss": 0.8866, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.7013911355548367, | |
| "grad_norm": 0.9351438283920288, | |
| "learning_rate": 9.594159458822257e-05, | |
| "loss": 0.9607, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.7026852151407311, | |
| "grad_norm": 0.8216230869293213, | |
| "learning_rate": 9.592530827805322e-05, | |
| "loss": 1.0062, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.7039792947266257, | |
| "grad_norm": 0.8280504941940308, | |
| "learning_rate": 9.59089907429268e-05, | |
| "loss": 0.8715, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.7052733743125202, | |
| "grad_norm": 0.9679670929908752, | |
| "learning_rate": 9.589264199393776e-05, | |
| "loss": 1.0543, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.7065674538984148, | |
| "grad_norm": 0.7370772361755371, | |
| "learning_rate": 9.587626204220171e-05, | |
| "loss": 0.8853, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.7078615334843092, | |
| "grad_norm": 0.7535960674285889, | |
| "learning_rate": 9.585985089885552e-05, | |
| "loss": 0.8419, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.7091556130702038, | |
| "grad_norm": 0.7049685716629028, | |
| "learning_rate": 9.584340857505722e-05, | |
| "loss": 0.8282, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.7104496926560984, | |
| "grad_norm": 0.7157800197601318, | |
| "learning_rate": 9.58269350819861e-05, | |
| "loss": 0.8282, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.7117437722419929, | |
| "grad_norm": 0.7660270929336548, | |
| "learning_rate": 9.581043043084259e-05, | |
| "loss": 0.9006, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.7130378518278874, | |
| "grad_norm": 0.7422590255737305, | |
| "learning_rate": 9.579389463284834e-05, | |
| "loss": 0.9591, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.7143319314137819, | |
| "grad_norm": 0.7437978982925415, | |
| "learning_rate": 9.577732769924614e-05, | |
| "loss": 0.8823, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.7156260109996765, | |
| "grad_norm": 0.7947098016738892, | |
| "learning_rate": 9.576072964129998e-05, | |
| "loss": 0.8161, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.716920090585571, | |
| "grad_norm": 0.8041345477104187, | |
| "learning_rate": 9.574410047029502e-05, | |
| "loss": 0.9831, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.7182141701714655, | |
| "grad_norm": 0.9462196826934814, | |
| "learning_rate": 9.572744019753752e-05, | |
| "loss": 0.9516, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.71950824975736, | |
| "grad_norm": 0.7633935213088989, | |
| "learning_rate": 9.571074883435496e-05, | |
| "loss": 0.8415, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.7208023293432546, | |
| "grad_norm": 0.7996200919151306, | |
| "learning_rate": 9.569402639209589e-05, | |
| "loss": 0.9336, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.7220964089291492, | |
| "grad_norm": 0.879639208316803, | |
| "learning_rate": 9.567727288213005e-05, | |
| "loss": 0.9479, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.7233904885150437, | |
| "grad_norm": 0.7905813455581665, | |
| "learning_rate": 9.566048831584826e-05, | |
| "loss": 0.7692, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.7246845681009382, | |
| "grad_norm": 0.7934820055961609, | |
| "learning_rate": 9.564367270466247e-05, | |
| "loss": 0.8735, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.7259786476868327, | |
| "grad_norm": 0.8444498181343079, | |
| "learning_rate": 9.562682606000575e-05, | |
| "loss": 0.9474, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.7272727272727273, | |
| "grad_norm": 0.7441167831420898, | |
| "learning_rate": 9.560994839333224e-05, | |
| "loss": 0.9077, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.7285668068586219, | |
| "grad_norm": 0.7276771664619446, | |
| "learning_rate": 9.55930397161172e-05, | |
| "loss": 0.8866, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.7298608864445163, | |
| "grad_norm": 0.8522891402244568, | |
| "learning_rate": 9.557610003985698e-05, | |
| "loss": 0.9961, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.7311549660304109, | |
| "grad_norm": 0.8116034269332886, | |
| "learning_rate": 9.555912937606895e-05, | |
| "loss": 0.8989, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.7324490456163054, | |
| "grad_norm": 0.8132615089416504, | |
| "learning_rate": 9.554212773629164e-05, | |
| "loss": 0.809, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.7337431252022, | |
| "grad_norm": 0.9693892002105713, | |
| "learning_rate": 9.552509513208456e-05, | |
| "loss": 0.959, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.7350372047880944, | |
| "grad_norm": 0.8740898966789246, | |
| "learning_rate": 9.550803157502831e-05, | |
| "loss": 0.9758, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.736331284373989, | |
| "grad_norm": 0.8451546430587769, | |
| "learning_rate": 9.549093707672453e-05, | |
| "loss": 0.9776, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.7376253639598835, | |
| "grad_norm": 0.8144363760948181, | |
| "learning_rate": 9.54738116487959e-05, | |
| "loss": 0.8114, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.7389194435457781, | |
| "grad_norm": 0.9224253296852112, | |
| "learning_rate": 9.545665530288612e-05, | |
| "loss": 0.8829, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.7402135231316725, | |
| "grad_norm": 0.8720259666442871, | |
| "learning_rate": 9.543946805065992e-05, | |
| "loss": 0.9658, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.7415076027175671, | |
| "grad_norm": 0.5992270708084106, | |
| "learning_rate": 9.542224990380304e-05, | |
| "loss": 0.8167, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.7428016823034617, | |
| "grad_norm": 0.7595481276512146, | |
| "learning_rate": 9.540500087402222e-05, | |
| "loss": 0.9002, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.7440957618893562, | |
| "grad_norm": 0.8269613981246948, | |
| "learning_rate": 9.538772097304521e-05, | |
| "loss": 0.8442, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.7453898414752508, | |
| "grad_norm": 0.7087002992630005, | |
| "learning_rate": 9.537041021262072e-05, | |
| "loss": 0.8004, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.7466839210611452, | |
| "grad_norm": 0.8022588491439819, | |
| "learning_rate": 9.535306860451849e-05, | |
| "loss": 0.9188, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.7479780006470398, | |
| "grad_norm": 0.8022916316986084, | |
| "learning_rate": 9.533569616052921e-05, | |
| "loss": 0.9815, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.7492720802329343, | |
| "grad_norm": 0.6670771837234497, | |
| "learning_rate": 9.531829289246452e-05, | |
| "loss": 0.812, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.7505661598188289, | |
| "grad_norm": 0.7644294500350952, | |
| "learning_rate": 9.530085881215705e-05, | |
| "loss": 0.8092, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.7518602394047234, | |
| "grad_norm": 0.8892885446548462, | |
| "learning_rate": 9.528339393146033e-05, | |
| "loss": 0.8422, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.7531543189906179, | |
| "grad_norm": 0.7766391634941101, | |
| "learning_rate": 9.526589826224887e-05, | |
| "loss": 0.9596, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.7544483985765125, | |
| "grad_norm": 0.8290911316871643, | |
| "learning_rate": 9.524837181641813e-05, | |
| "loss": 0.9624, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.755742478162407, | |
| "grad_norm": 0.7832044363021851, | |
| "learning_rate": 9.523081460588444e-05, | |
| "loss": 0.8141, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.7570365577483015, | |
| "grad_norm": 0.9279314875602722, | |
| "learning_rate": 9.521322664258508e-05, | |
| "loss": 0.8869, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.758330637334196, | |
| "grad_norm": 0.8901260495185852, | |
| "learning_rate": 9.519560793847826e-05, | |
| "loss": 1.0233, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.7596247169200906, | |
| "grad_norm": 0.7530801296234131, | |
| "learning_rate": 9.517795850554306e-05, | |
| "loss": 0.8466, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.7609187965059852, | |
| "grad_norm": 0.7969221472740173, | |
| "learning_rate": 9.516027835577943e-05, | |
| "loss": 0.789, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.7622128760918796, | |
| "grad_norm": 0.7585557103157043, | |
| "learning_rate": 9.514256750120828e-05, | |
| "loss": 0.8779, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.7635069556777742, | |
| "grad_norm": 0.8145812749862671, | |
| "learning_rate": 9.512482595387132e-05, | |
| "loss": 0.9836, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.7648010352636687, | |
| "grad_norm": 0.6959390640258789, | |
| "learning_rate": 9.510705372583118e-05, | |
| "loss": 0.8525, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.7660951148495633, | |
| "grad_norm": 0.6348171234130859, | |
| "learning_rate": 9.50892508291713e-05, | |
| "loss": 0.7776, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.7673891944354578, | |
| "grad_norm": 1.0133094787597656, | |
| "learning_rate": 9.507141727599602e-05, | |
| "loss": 0.8943, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.7686832740213523, | |
| "grad_norm": 0.8890700340270996, | |
| "learning_rate": 9.50535530784305e-05, | |
| "loss": 0.9169, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.7699773536072468, | |
| "grad_norm": 0.7059516310691833, | |
| "learning_rate": 9.503565824862076e-05, | |
| "loss": 0.7544, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.7712714331931414, | |
| "grad_norm": 0.8086109161376953, | |
| "learning_rate": 9.50177327987336e-05, | |
| "loss": 0.8457, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.772565512779036, | |
| "grad_norm": 0.8387449383735657, | |
| "learning_rate": 9.499977674095669e-05, | |
| "loss": 0.8291, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.7738595923649304, | |
| "grad_norm": 0.7841625809669495, | |
| "learning_rate": 9.498179008749847e-05, | |
| "loss": 0.8972, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.775153671950825, | |
| "grad_norm": 0.7261033058166504, | |
| "learning_rate": 9.496377285058819e-05, | |
| "loss": 0.7678, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.7764477515367195, | |
| "grad_norm": 0.8468754887580872, | |
| "learning_rate": 9.494572504247593e-05, | |
| "loss": 0.9044, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.7777418311226141, | |
| "grad_norm": 0.8177486062049866, | |
| "learning_rate": 9.492764667543252e-05, | |
| "loss": 0.8132, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.7790359107085085, | |
| "grad_norm": 1.0061938762664795, | |
| "learning_rate": 9.490953776174955e-05, | |
| "loss": 0.9204, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.7803299902944031, | |
| "grad_norm": 0.8224837779998779, | |
| "learning_rate": 9.489139831373944e-05, | |
| "loss": 0.7909, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.7816240698802976, | |
| "grad_norm": 0.7812953591346741, | |
| "learning_rate": 9.48732283437353e-05, | |
| "loss": 0.9583, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.7829181494661922, | |
| "grad_norm": 0.8498075604438782, | |
| "learning_rate": 9.485502786409107e-05, | |
| "loss": 1.0692, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.7842122290520867, | |
| "grad_norm": 0.733768105506897, | |
| "learning_rate": 9.483679688718135e-05, | |
| "loss": 0.8835, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.7855063086379812, | |
| "grad_norm": 0.7453616261482239, | |
| "learning_rate": 9.481853542540154e-05, | |
| "loss": 0.7778, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.7868003882238758, | |
| "grad_norm": 0.8596757650375366, | |
| "learning_rate": 9.480024349116771e-05, | |
| "loss": 0.9711, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.7880944678097703, | |
| "grad_norm": 0.6516830325126648, | |
| "learning_rate": 9.478192109691674e-05, | |
| "loss": 0.8342, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.7893885473956649, | |
| "grad_norm": 0.8740130662918091, | |
| "learning_rate": 9.476356825510612e-05, | |
| "loss": 0.9179, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.7906826269815593, | |
| "grad_norm": 0.7338031530380249, | |
| "learning_rate": 9.474518497821409e-05, | |
| "loss": 0.9518, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.7919767065674539, | |
| "grad_norm": 0.8368909358978271, | |
| "learning_rate": 9.472677127873958e-05, | |
| "loss": 0.875, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.7932707861533485, | |
| "grad_norm": 0.7648499011993408, | |
| "learning_rate": 9.47083271692022e-05, | |
| "loss": 1.0374, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.794564865739243, | |
| "grad_norm": 0.767672598361969, | |
| "learning_rate": 9.468985266214223e-05, | |
| "loss": 0.8341, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.7958589453251375, | |
| "grad_norm": 0.9221040606498718, | |
| "learning_rate": 9.467134777012063e-05, | |
| "loss": 1.0149, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.797153024911032, | |
| "grad_norm": 0.8752298355102539, | |
| "learning_rate": 9.465281250571902e-05, | |
| "loss": 0.9948, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.7984471044969266, | |
| "grad_norm": 0.8295148611068726, | |
| "learning_rate": 9.463424688153966e-05, | |
| "loss": 0.8181, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.7997411840828211, | |
| "grad_norm": 0.7995513677597046, | |
| "learning_rate": 9.461565091020543e-05, | |
| "loss": 0.8851, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.8010352636687156, | |
| "grad_norm": 0.7693185210227966, | |
| "learning_rate": 9.459702460435989e-05, | |
| "loss": 0.8141, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.8023293432546101, | |
| "grad_norm": 0.8453662991523743, | |
| "learning_rate": 9.457836797666722e-05, | |
| "loss": 0.9608, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.8036234228405047, | |
| "grad_norm": 0.8325058817863464, | |
| "learning_rate": 9.455968103981219e-05, | |
| "loss": 0.8521, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.8049175024263993, | |
| "grad_norm": 0.7803249359130859, | |
| "learning_rate": 9.454096380650014e-05, | |
| "loss": 0.8598, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.8062115820122937, | |
| "grad_norm": 0.731028139591217, | |
| "learning_rate": 9.452221628945713e-05, | |
| "loss": 0.8091, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.8075056615981883, | |
| "grad_norm": 0.8708882927894592, | |
| "learning_rate": 9.450343850142969e-05, | |
| "loss": 0.8467, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.8087997411840828, | |
| "grad_norm": 0.7343994975090027, | |
| "learning_rate": 9.448463045518499e-05, | |
| "loss": 0.8496, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.8100938207699774, | |
| "grad_norm": 0.8136367797851562, | |
| "learning_rate": 9.446579216351074e-05, | |
| "loss": 0.8387, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.8113879003558719, | |
| "grad_norm": 0.7161028981208801, | |
| "learning_rate": 9.444692363921528e-05, | |
| "loss": 0.839, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.8126819799417664, | |
| "grad_norm": 0.9956607222557068, | |
| "learning_rate": 9.442802489512741e-05, | |
| "loss": 1.0051, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.813976059527661, | |
| "grad_norm": 0.8381032943725586, | |
| "learning_rate": 9.440909594409655e-05, | |
| "loss": 0.9681, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.8152701391135555, | |
| "grad_norm": 0.7452998757362366, | |
| "learning_rate": 9.439013679899262e-05, | |
| "loss": 0.8106, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.8165642186994501, | |
| "grad_norm": 0.7175205945968628, | |
| "learning_rate": 9.437114747270612e-05, | |
| "loss": 0.8622, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.8178582982853445, | |
| "grad_norm": 0.8083619475364685, | |
| "learning_rate": 9.435212797814798e-05, | |
| "loss": 0.8608, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.8191523778712391, | |
| "grad_norm": 0.9113181233406067, | |
| "learning_rate": 9.433307832824974e-05, | |
| "loss": 0.9445, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.8204464574571336, | |
| "grad_norm": 0.7915831208229065, | |
| "learning_rate": 9.431399853596336e-05, | |
| "loss": 0.8651, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.8217405370430282, | |
| "grad_norm": 0.7277971506118774, | |
| "learning_rate": 9.429488861426137e-05, | |
| "loss": 0.8799, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.8230346166289226, | |
| "grad_norm": 0.7446788549423218, | |
| "learning_rate": 9.427574857613672e-05, | |
| "loss": 0.8071, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.8243286962148172, | |
| "grad_norm": 0.8236122131347656, | |
| "learning_rate": 9.425657843460288e-05, | |
| "loss": 0.9889, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.8256227758007118, | |
| "grad_norm": 0.8189204335212708, | |
| "learning_rate": 9.423737820269376e-05, | |
| "loss": 0.9607, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.8269168553866063, | |
| "grad_norm": 0.9449727535247803, | |
| "learning_rate": 9.421814789346375e-05, | |
| "loss": 0.9581, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.8282109349725008, | |
| "grad_norm": 0.7527281045913696, | |
| "learning_rate": 9.419888751998767e-05, | |
| "loss": 0.7984, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.8295050145583953, | |
| "grad_norm": 0.8296061754226685, | |
| "learning_rate": 9.417959709536078e-05, | |
| "loss": 0.8597, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.8307990941442899, | |
| "grad_norm": 0.8495222330093384, | |
| "learning_rate": 9.416027663269881e-05, | |
| "loss": 0.8396, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.8320931737301844, | |
| "grad_norm": 0.8501962423324585, | |
| "learning_rate": 9.414092614513787e-05, | |
| "loss": 0.9911, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.833387253316079, | |
| "grad_norm": 0.7546764612197876, | |
| "learning_rate": 9.412154564583448e-05, | |
| "loss": 0.9043, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.8346813329019734, | |
| "grad_norm": 0.8009942770004272, | |
| "learning_rate": 9.410213514796564e-05, | |
| "loss": 0.8242, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.835975412487868, | |
| "grad_norm": 0.8321415185928345, | |
| "learning_rate": 9.408269466472864e-05, | |
| "loss": 0.9094, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.8372694920737626, | |
| "grad_norm": 0.7805325984954834, | |
| "learning_rate": 9.406322420934123e-05, | |
| "loss": 0.837, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.8385635716596571, | |
| "grad_norm": 0.8147938251495361, | |
| "learning_rate": 9.404372379504151e-05, | |
| "loss": 0.9177, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.8398576512455516, | |
| "grad_norm": 0.7518514394760132, | |
| "learning_rate": 9.402419343508797e-05, | |
| "loss": 0.7562, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.8411517308314461, | |
| "grad_norm": 0.7858056426048279, | |
| "learning_rate": 9.400463314275943e-05, | |
| "loss": 0.9865, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.8424458104173407, | |
| "grad_norm": 0.7086622714996338, | |
| "learning_rate": 9.398504293135507e-05, | |
| "loss": 0.8451, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.8437398900032352, | |
| "grad_norm": 0.8999130725860596, | |
| "learning_rate": 9.396542281419445e-05, | |
| "loss": 0.919, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.8450339695891297, | |
| "grad_norm": 0.8439355492591858, | |
| "learning_rate": 9.39457728046174e-05, | |
| "loss": 0.808, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.8463280491750242, | |
| "grad_norm": 0.863634467124939, | |
| "learning_rate": 9.392609291598413e-05, | |
| "loss": 1.0594, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.8476221287609188, | |
| "grad_norm": 0.8126616477966309, | |
| "learning_rate": 9.390638316167512e-05, | |
| "loss": 1.0526, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.8489162083468134, | |
| "grad_norm": 0.6746686697006226, | |
| "learning_rate": 9.388664355509122e-05, | |
| "loss": 0.8609, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.8502102879327078, | |
| "grad_norm": 0.6596627831459045, | |
| "learning_rate": 9.386687410965349e-05, | |
| "loss": 0.8474, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.8515043675186024, | |
| "grad_norm": 0.7229893207550049, | |
| "learning_rate": 9.384707483880334e-05, | |
| "loss": 0.8095, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.8527984471044969, | |
| "grad_norm": 0.678260326385498, | |
| "learning_rate": 9.382724575600243e-05, | |
| "loss": 0.9453, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.8540925266903915, | |
| "grad_norm": 0.7702340483665466, | |
| "learning_rate": 9.380738687473272e-05, | |
| "loss": 0.9527, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.855386606276286, | |
| "grad_norm": 0.7638357877731323, | |
| "learning_rate": 9.378749820849642e-05, | |
| "loss": 1.0131, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.8566806858621805, | |
| "grad_norm": 0.7495226263999939, | |
| "learning_rate": 9.376757977081594e-05, | |
| "loss": 0.8511, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.857974765448075, | |
| "grad_norm": 0.8003675937652588, | |
| "learning_rate": 9.374763157523399e-05, | |
| "loss": 0.9291, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.8592688450339696, | |
| "grad_norm": 0.7893322706222534, | |
| "learning_rate": 9.372765363531351e-05, | |
| "loss": 0.8206, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.8605629246198642, | |
| "grad_norm": 0.7351086735725403, | |
| "learning_rate": 9.370764596463763e-05, | |
| "loss": 0.7185, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.8618570042057586, | |
| "grad_norm": 0.8211125135421753, | |
| "learning_rate": 9.368760857680974e-05, | |
| "loss": 0.9613, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.8631510837916532, | |
| "grad_norm": 0.9151694178581238, | |
| "learning_rate": 9.366754148545338e-05, | |
| "loss": 0.9788, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.8644451633775477, | |
| "grad_norm": 0.84197998046875, | |
| "learning_rate": 9.364744470421234e-05, | |
| "loss": 0.9891, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.8657392429634423, | |
| "grad_norm": 0.6799368858337402, | |
| "learning_rate": 9.362731824675056e-05, | |
| "loss": 0.7988, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.8670333225493367, | |
| "grad_norm": 0.722326934337616, | |
| "learning_rate": 9.360716212675214e-05, | |
| "loss": 0.8399, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.8683274021352313, | |
| "grad_norm": 0.7735809683799744, | |
| "learning_rate": 9.35869763579214e-05, | |
| "loss": 0.8677, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.8696214817211259, | |
| "grad_norm": 0.7918348908424377, | |
| "learning_rate": 9.35667609539828e-05, | |
| "loss": 0.8792, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.8709155613070204, | |
| "grad_norm": 0.7758587002754211, | |
| "learning_rate": 9.354651592868094e-05, | |
| "loss": 0.8844, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.8722096408929149, | |
| "grad_norm": 0.8733750581741333, | |
| "learning_rate": 9.352624129578054e-05, | |
| "loss": 0.811, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.8735037204788094, | |
| "grad_norm": 0.8485933542251587, | |
| "learning_rate": 9.350593706906651e-05, | |
| "loss": 0.787, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.874797800064704, | |
| "grad_norm": 0.7540378570556641, | |
| "learning_rate": 9.348560326234381e-05, | |
| "loss": 0.8578, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.8760918796505985, | |
| "grad_norm": 0.7626041173934937, | |
| "learning_rate": 9.346523988943758e-05, | |
| "loss": 0.9294, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.8773859592364931, | |
| "grad_norm": 0.8360442519187927, | |
| "learning_rate": 9.3444846964193e-05, | |
| "loss": 0.8635, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.8786800388223875, | |
| "grad_norm": 0.9039386510848999, | |
| "learning_rate": 9.342442450047537e-05, | |
| "loss": 0.83, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.8799741184082821, | |
| "grad_norm": 0.7554466724395752, | |
| "learning_rate": 9.340397251217009e-05, | |
| "loss": 0.8103, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.8812681979941767, | |
| "grad_norm": 0.8201668858528137, | |
| "learning_rate": 9.338349101318261e-05, | |
| "loss": 0.7959, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.8825622775800712, | |
| "grad_norm": 0.814172625541687, | |
| "learning_rate": 9.336298001743846e-05, | |
| "loss": 0.8074, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.8838563571659657, | |
| "grad_norm": 0.7825469374656677, | |
| "learning_rate": 9.33424395388832e-05, | |
| "loss": 0.91, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.8851504367518602, | |
| "grad_norm": 0.7842270135879517, | |
| "learning_rate": 9.332186959148248e-05, | |
| "loss": 0.9331, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.8864445163377548, | |
| "grad_norm": 0.8008681535720825, | |
| "learning_rate": 9.330127018922194e-05, | |
| "loss": 0.7166, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.8877385959236493, | |
| "grad_norm": 0.8031767010688782, | |
| "learning_rate": 9.328064134610727e-05, | |
| "loss": 0.9718, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.8890326755095438, | |
| "grad_norm": 0.7503944635391235, | |
| "learning_rate": 9.325998307616417e-05, | |
| "loss": 0.9732, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.8903267550954383, | |
| "grad_norm": 0.8711128234863281, | |
| "learning_rate": 9.323929539343837e-05, | |
| "loss": 1.05, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.8916208346813329, | |
| "grad_norm": 0.7816237211227417, | |
| "learning_rate": 9.321857831199554e-05, | |
| "loss": 0.9514, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.8929149142672275, | |
| "grad_norm": 0.8090843558311462, | |
| "learning_rate": 9.319783184592142e-05, | |
| "loss": 0.9691, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.8942089938531219, | |
| "grad_norm": 0.6824143528938293, | |
| "learning_rate": 9.317705600932164e-05, | |
| "loss": 0.7027, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.8955030734390165, | |
| "grad_norm": 0.8860282301902771, | |
| "learning_rate": 9.315625081632191e-05, | |
| "loss": 0.9419, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.896797153024911, | |
| "grad_norm": 0.9011827707290649, | |
| "learning_rate": 9.313541628106777e-05, | |
| "loss": 0.8468, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.8980912326108056, | |
| "grad_norm": 0.836565375328064, | |
| "learning_rate": 9.311455241772482e-05, | |
| "loss": 1.0456, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.8993853121967001, | |
| "grad_norm": 0.819817841053009, | |
| "learning_rate": 9.309365924047853e-05, | |
| "loss": 0.9606, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.9006793917825946, | |
| "grad_norm": 0.7008129954338074, | |
| "learning_rate": 9.307273676353432e-05, | |
| "loss": 0.8531, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.9019734713684892, | |
| "grad_norm": 0.8026110529899597, | |
| "learning_rate": 9.305178500111755e-05, | |
| "loss": 0.7784, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.9032675509543837, | |
| "grad_norm": 0.7309970855712891, | |
| "learning_rate": 9.30308039674735e-05, | |
| "loss": 0.9284, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.9045616305402783, | |
| "grad_norm": 0.801511824131012, | |
| "learning_rate": 9.300979367686729e-05, | |
| "loss": 0.8111, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.9058557101261727, | |
| "grad_norm": 0.8487135767936707, | |
| "learning_rate": 9.298875414358399e-05, | |
| "loss": 0.9095, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.9071497897120673, | |
| "grad_norm": 0.6941442489624023, | |
| "learning_rate": 9.296768538192853e-05, | |
| "loss": 0.8285, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.9084438692979618, | |
| "grad_norm": 0.7753114700317383, | |
| "learning_rate": 9.294658740622573e-05, | |
| "loss": 0.9726, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.9097379488838564, | |
| "grad_norm": 0.8224105834960938, | |
| "learning_rate": 9.292546023082025e-05, | |
| "loss": 0.8728, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.9110320284697508, | |
| "grad_norm": 0.9305656552314758, | |
| "learning_rate": 9.29043038700766e-05, | |
| "loss": 0.891, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.9123261080556454, | |
| "grad_norm": 0.7868937849998474, | |
| "learning_rate": 9.288311833837917e-05, | |
| "loss": 0.8883, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.91362018764154, | |
| "grad_norm": 0.6708902716636658, | |
| "learning_rate": 9.286190365013217e-05, | |
| "loss": 0.8315, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.9149142672274345, | |
| "grad_norm": 0.802245557308197, | |
| "learning_rate": 9.284065981975958e-05, | |
| "loss": 0.9128, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.916208346813329, | |
| "grad_norm": 0.8574107885360718, | |
| "learning_rate": 9.281938686170526e-05, | |
| "loss": 0.9122, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.9175024263992235, | |
| "grad_norm": 1.060194730758667, | |
| "learning_rate": 9.279808479043286e-05, | |
| "loss": 1.1447, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.9187965059851181, | |
| "grad_norm": 0.8753707408905029, | |
| "learning_rate": 9.277675362042581e-05, | |
| "loss": 0.8524, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.9200905855710126, | |
| "grad_norm": 1.0536813735961914, | |
| "learning_rate": 9.275539336618732e-05, | |
| "loss": 1.014, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.9213846651569072, | |
| "grad_norm": 0.779331624507904, | |
| "learning_rate": 9.27340040422404e-05, | |
| "loss": 0.913, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.9226787447428016, | |
| "grad_norm": 1.0297069549560547, | |
| "learning_rate": 9.27125856631278e-05, | |
| "loss": 1.0926, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.9239728243286962, | |
| "grad_norm": 0.7252759337425232, | |
| "learning_rate": 9.269113824341205e-05, | |
| "loss": 0.8096, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.9252669039145908, | |
| "grad_norm": 0.900393009185791, | |
| "learning_rate": 9.266966179767538e-05, | |
| "loss": 0.7856, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.9265609835004853, | |
| "grad_norm": 0.8092474937438965, | |
| "learning_rate": 9.264815634051982e-05, | |
| "loss": 1.0298, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.9278550630863798, | |
| "grad_norm": 0.8919221758842468, | |
| "learning_rate": 9.262662188656705e-05, | |
| "loss": 0.7962, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.9291491426722743, | |
| "grad_norm": 0.771659791469574, | |
| "learning_rate": 9.260505845045854e-05, | |
| "loss": 1.0239, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.9304432222581689, | |
| "grad_norm": 0.7577182054519653, | |
| "learning_rate": 9.258346604685542e-05, | |
| "loss": 0.9672, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.9317373018440634, | |
| "grad_norm": 0.7373968362808228, | |
| "learning_rate": 9.256184469043851e-05, | |
| "loss": 0.7484, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.9330313814299579, | |
| "grad_norm": 0.7698187232017517, | |
| "learning_rate": 9.254019439590835e-05, | |
| "loss": 0.7563, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.9343254610158525, | |
| "grad_norm": 0.8331950306892395, | |
| "learning_rate": 9.251851517798514e-05, | |
| "loss": 0.8826, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.935619540601747, | |
| "grad_norm": 0.7643804550170898, | |
| "learning_rate": 9.24968070514087e-05, | |
| "loss": 1.0688, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.9369136201876416, | |
| "grad_norm": 0.802943229675293, | |
| "learning_rate": 9.247507003093858e-05, | |
| "loss": 0.8872, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.938207699773536, | |
| "grad_norm": 0.8977981209754944, | |
| "learning_rate": 9.245330413135395e-05, | |
| "loss": 0.8946, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.9395017793594306, | |
| "grad_norm": 0.7683425545692444, | |
| "learning_rate": 9.243150936745357e-05, | |
| "loss": 0.8221, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.9407958589453251, | |
| "grad_norm": 0.7723820209503174, | |
| "learning_rate": 9.240968575405589e-05, | |
| "loss": 0.9058, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.9420899385312197, | |
| "grad_norm": 0.7846856117248535, | |
| "learning_rate": 9.238783330599893e-05, | |
| "loss": 0.8775, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.9433840181171143, | |
| "grad_norm": 0.6629738211631775, | |
| "learning_rate": 9.236595203814037e-05, | |
| "loss": 0.8753, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.9446780977030087, | |
| "grad_norm": 0.7508718371391296, | |
| "learning_rate": 9.23440419653574e-05, | |
| "loss": 0.9403, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.9459721772889033, | |
| "grad_norm": 0.8904352784156799, | |
| "learning_rate": 9.232210310254685e-05, | |
| "loss": 0.8514, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.9472662568747978, | |
| "grad_norm": 1.004438877105713, | |
| "learning_rate": 9.230013546462512e-05, | |
| "loss": 0.9824, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.9485603364606924, | |
| "grad_norm": 0.7370134592056274, | |
| "learning_rate": 9.227813906652818e-05, | |
| "loss": 0.9869, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.9498544160465868, | |
| "grad_norm": 0.8597853183746338, | |
| "learning_rate": 9.225611392321153e-05, | |
| "loss": 0.8758, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.9511484956324814, | |
| "grad_norm": 0.7227449417114258, | |
| "learning_rate": 9.223406004965023e-05, | |
| "loss": 0.8208, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.9524425752183759, | |
| "grad_norm": 0.8053308725357056, | |
| "learning_rate": 9.221197746083887e-05, | |
| "loss": 0.8555, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.9537366548042705, | |
| "grad_norm": 0.7663565278053284, | |
| "learning_rate": 9.21898661717916e-05, | |
| "loss": 0.7754, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.955030734390165, | |
| "grad_norm": 0.7501769661903381, | |
| "learning_rate": 9.216772619754199e-05, | |
| "loss": 0.8798, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.9563248139760595, | |
| "grad_norm": 0.7159123420715332, | |
| "learning_rate": 9.21455575531432e-05, | |
| "loss": 0.963, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.9576188935619541, | |
| "grad_norm": 0.8228005766868591, | |
| "learning_rate": 9.212336025366788e-05, | |
| "loss": 0.9521, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.9589129731478486, | |
| "grad_norm": 0.7884674072265625, | |
| "learning_rate": 9.21011343142081e-05, | |
| "loss": 0.9046, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.9602070527337431, | |
| "grad_norm": 0.8630079030990601, | |
| "learning_rate": 9.207887974987546e-05, | |
| "loss": 0.9627, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.9615011323196376, | |
| "grad_norm": 0.707506537437439, | |
| "learning_rate": 9.205659657580101e-05, | |
| "loss": 0.8428, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.9627952119055322, | |
| "grad_norm": 0.700090765953064, | |
| "learning_rate": 9.203428480713521e-05, | |
| "loss": 0.8938, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.9640892914914267, | |
| "grad_norm": 0.813056468963623, | |
| "learning_rate": 9.201194445904803e-05, | |
| "loss": 1.0089, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.9653833710773213, | |
| "grad_norm": 0.747582733631134, | |
| "learning_rate": 9.198957554672885e-05, | |
| "loss": 0.9392, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.9666774506632158, | |
| "grad_norm": 0.7720314264297485, | |
| "learning_rate": 9.196717808538641e-05, | |
| "loss": 0.9572, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.9679715302491103, | |
| "grad_norm": 0.802967369556427, | |
| "learning_rate": 9.194475209024895e-05, | |
| "loss": 0.697, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.9692656098350049, | |
| "grad_norm": 0.7272825837135315, | |
| "learning_rate": 9.192229757656406e-05, | |
| "loss": 0.7704, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.9705596894208994, | |
| "grad_norm": 0.7247483730316162, | |
| "learning_rate": 9.189981455959874e-05, | |
| "loss": 0.8788, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.9718537690067939, | |
| "grad_norm": 0.830154538154602, | |
| "learning_rate": 9.187730305463933e-05, | |
| "loss": 0.8609, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.9731478485926884, | |
| "grad_norm": 0.7187288403511047, | |
| "learning_rate": 9.185476307699159e-05, | |
| "loss": 0.8538, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.974441928178583, | |
| "grad_norm": 0.8054041862487793, | |
| "learning_rate": 9.18321946419806e-05, | |
| "loss": 0.8907, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.9757360077644776, | |
| "grad_norm": 0.7722858190536499, | |
| "learning_rate": 9.180959776495085e-05, | |
| "loss": 0.82, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.977030087350372, | |
| "grad_norm": 0.7097930312156677, | |
| "learning_rate": 9.178697246126607e-05, | |
| "loss": 0.855, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.9783241669362666, | |
| "grad_norm": 0.894588828086853, | |
| "learning_rate": 9.176431874630941e-05, | |
| "loss": 0.8531, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.9796182465221611, | |
| "grad_norm": 0.7806453704833984, | |
| "learning_rate": 9.174163663548329e-05, | |
| "loss": 0.9423, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.9809123261080557, | |
| "grad_norm": 0.8258002400398254, | |
| "learning_rate": 9.171892614420942e-05, | |
| "loss": 0.6763, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.9822064056939501, | |
| "grad_norm": 0.7445089221000671, | |
| "learning_rate": 9.169618728792886e-05, | |
| "loss": 0.7886, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.9835004852798447, | |
| "grad_norm": 0.8452383279800415, | |
| "learning_rate": 9.167342008210192e-05, | |
| "loss": 1.0001, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.9847945648657392, | |
| "grad_norm": 0.7532628178596497, | |
| "learning_rate": 9.165062454220818e-05, | |
| "loss": 0.9464, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.9860886444516338, | |
| "grad_norm": 0.835107684135437, | |
| "learning_rate": 9.162780068374651e-05, | |
| "loss": 0.8498, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.9873827240375284, | |
| "grad_norm": 0.8420085310935974, | |
| "learning_rate": 9.160494852223503e-05, | |
| "loss": 1.0078, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.9886768036234228, | |
| "grad_norm": 0.8478022813796997, | |
| "learning_rate": 9.158206807321105e-05, | |
| "loss": 0.7799, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.9899708832093174, | |
| "grad_norm": 0.7231602072715759, | |
| "learning_rate": 9.155915935223121e-05, | |
| "loss": 0.8196, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.9912649627952119, | |
| "grad_norm": 0.6537997722625732, | |
| "learning_rate": 9.153622237487127e-05, | |
| "loss": 0.6911, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.9925590423811065, | |
| "grad_norm": 0.6952023506164551, | |
| "learning_rate": 9.15132571567263e-05, | |
| "loss": 0.7354, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.9938531219670009, | |
| "grad_norm": 0.7851712703704834, | |
| "learning_rate": 9.149026371341049e-05, | |
| "loss": 1.0092, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.9951472015528955, | |
| "grad_norm": 0.8476317524909973, | |
| "learning_rate": 9.146724206055725e-05, | |
| "loss": 1.0262, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.99644128113879, | |
| "grad_norm": 0.7637205123901367, | |
| "learning_rate": 9.144419221381919e-05, | |
| "loss": 0.879, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.9977353607246846, | |
| "grad_norm": 0.9002841711044312, | |
| "learning_rate": 9.142111418886807e-05, | |
| "loss": 0.8887, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.999029440310579, | |
| "grad_norm": 0.7710572481155396, | |
| "learning_rate": 9.13980080013948e-05, | |
| "loss": 0.868, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.999029440310579, | |
| "eval_loss": 0.8607526421546936, | |
| "eval_runtime": 189.9676, | |
| "eval_samples_per_second": 3.427, | |
| "eval_steps_per_second": 0.858, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 1.0003235198964737, | |
| "grad_norm": 0.7324461936950684, | |
| "learning_rate": 9.137487366710945e-05, | |
| "loss": 0.7605, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 1.0016175994823682, | |
| "grad_norm": 0.7000962495803833, | |
| "learning_rate": 9.135171120174122e-05, | |
| "loss": 0.7624, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 1.0029116790682626, | |
| "grad_norm": 0.711403489112854, | |
| "learning_rate": 9.132852062103844e-05, | |
| "loss": 0.868, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.0042057586541573, | |
| "grad_norm": 0.6421488523483276, | |
| "learning_rate": 9.130530194076858e-05, | |
| "loss": 0.6891, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 1.0054998382400517, | |
| "grad_norm": 0.7791386246681213, | |
| "learning_rate": 9.12820551767182e-05, | |
| "loss": 0.8292, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 1.0067939178259464, | |
| "grad_norm": 0.7175559401512146, | |
| "learning_rate": 9.125878034469289e-05, | |
| "loss": 0.7888, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 1.0080879974118409, | |
| "grad_norm": 0.75743567943573, | |
| "learning_rate": 9.123547746051743e-05, | |
| "loss": 0.675, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 1.0093820769977353, | |
| "grad_norm": 0.7583343386650085, | |
| "learning_rate": 9.12121465400356e-05, | |
| "loss": 0.7507, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.01067615658363, | |
| "grad_norm": 0.6894446611404419, | |
| "learning_rate": 9.118878759911029e-05, | |
| "loss": 0.8629, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 1.0119702361695244, | |
| "grad_norm": 0.8267707228660583, | |
| "learning_rate": 9.116540065362338e-05, | |
| "loss": 0.8126, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 1.0132643157554189, | |
| "grad_norm": 0.7307786345481873, | |
| "learning_rate": 9.114198571947587e-05, | |
| "loss": 0.729, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 1.0145583953413135, | |
| "grad_norm": 0.7283042669296265, | |
| "learning_rate": 9.111854281258772e-05, | |
| "loss": 0.8176, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 1.015852474927208, | |
| "grad_norm": 0.8333950042724609, | |
| "learning_rate": 9.109507194889792e-05, | |
| "loss": 0.8758, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.0171465545131027, | |
| "grad_norm": 0.7236875891685486, | |
| "learning_rate": 9.107157314436452e-05, | |
| "loss": 0.7658, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 1.018440634098997, | |
| "grad_norm": 0.6792603731155396, | |
| "learning_rate": 9.104804641496454e-05, | |
| "loss": 0.6965, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 1.0197347136848915, | |
| "grad_norm": 0.7678570747375488, | |
| "learning_rate": 9.102449177669393e-05, | |
| "loss": 0.6608, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 1.0210287932707862, | |
| "grad_norm": 0.8103516697883606, | |
| "learning_rate": 9.100090924556769e-05, | |
| "loss": 1.0342, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 1.0223228728566807, | |
| "grad_norm": 0.6715088486671448, | |
| "learning_rate": 9.097729883761976e-05, | |
| "loss": 0.6884, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.0236169524425751, | |
| "grad_norm": 0.7473916411399841, | |
| "learning_rate": 9.095366056890302e-05, | |
| "loss": 0.7346, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 1.0249110320284698, | |
| "grad_norm": 0.7097285985946655, | |
| "learning_rate": 9.09299944554893e-05, | |
| "loss": 0.825, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 1.0262051116143642, | |
| "grad_norm": 0.7450020909309387, | |
| "learning_rate": 9.09063005134694e-05, | |
| "loss": 0.8589, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 1.027499191200259, | |
| "grad_norm": 0.828708291053772, | |
| "learning_rate": 9.088257875895297e-05, | |
| "loss": 0.8284, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 1.0287932707861533, | |
| "grad_norm": 0.7440184354782104, | |
| "learning_rate": 9.085882920806861e-05, | |
| "loss": 0.8285, | |
| "step": 795 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 3860, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 5, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 8.886901804231557e+17, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |