| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.9968652037617556, | |
| "eval_steps": 500, | |
| "global_step": 424, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004702194357366771, | |
| "grad_norm": 3.308954954147339, | |
| "learning_rate": 5.0000000000000004e-08, | |
| "loss": 0.9758, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.009404388714733543, | |
| "grad_norm": 3.0945961475372314, | |
| "learning_rate": 1.0000000000000001e-07, | |
| "loss": 0.9784, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.014106583072100314, | |
| "grad_norm": 2.6266396045684814, | |
| "learning_rate": 1.5000000000000002e-07, | |
| "loss": 0.9407, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.018808777429467086, | |
| "grad_norm": 2.781346321105957, | |
| "learning_rate": 2.0000000000000002e-07, | |
| "loss": 0.983, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.023510971786833857, | |
| "grad_norm": 2.8360986709594727, | |
| "learning_rate": 2.5000000000000004e-07, | |
| "loss": 0.941, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02821316614420063, | |
| "grad_norm": 2.814222812652588, | |
| "learning_rate": 3.0000000000000004e-07, | |
| "loss": 0.9675, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.032915360501567396, | |
| "grad_norm": 2.821528196334839, | |
| "learning_rate": 3.5000000000000004e-07, | |
| "loss": 0.8856, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03761755485893417, | |
| "grad_norm": 2.7562029361724854, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.95, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.04231974921630094, | |
| "grad_norm": 2.9743621349334717, | |
| "learning_rate": 4.5000000000000003e-07, | |
| "loss": 0.9654, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.047021943573667714, | |
| "grad_norm": 3.0444681644439697, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.9133, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05172413793103448, | |
| "grad_norm": 2.9696431159973145, | |
| "learning_rate": 5.5e-07, | |
| "loss": 0.9425, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.05642633228840126, | |
| "grad_norm": 2.8478922843933105, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 0.9743, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.061128526645768025, | |
| "grad_norm": 2.664818525314331, | |
| "learning_rate": 6.5e-07, | |
| "loss": 0.9276, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.06583072100313479, | |
| "grad_norm": 2.727060556411743, | |
| "learning_rate": 7.000000000000001e-07, | |
| "loss": 0.9221, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.07053291536050156, | |
| "grad_norm": 2.8166537284851074, | |
| "learning_rate": 7.5e-07, | |
| "loss": 0.9406, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.07523510971786834, | |
| "grad_norm": 2.8412551879882812, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 0.901, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.07993730407523511, | |
| "grad_norm": 2.6464271545410156, | |
| "learning_rate": 8.500000000000001e-07, | |
| "loss": 0.9586, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.08463949843260188, | |
| "grad_norm": 2.506002187728882, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.8895, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.08934169278996865, | |
| "grad_norm": 2.663196086883545, | |
| "learning_rate": 9.500000000000001e-07, | |
| "loss": 0.8833, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.09404388714733543, | |
| "grad_norm": 2.5434982776641846, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.8737, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0987460815047022, | |
| "grad_norm": 2.2520358562469482, | |
| "learning_rate": 1.0500000000000001e-06, | |
| "loss": 0.8895, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.10344827586206896, | |
| "grad_norm": 2.3946690559387207, | |
| "learning_rate": 1.1e-06, | |
| "loss": 0.8405, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.10815047021943573, | |
| "grad_norm": 2.209516763687134, | |
| "learning_rate": 1.1500000000000002e-06, | |
| "loss": 0.8441, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.11285266457680251, | |
| "grad_norm": 2.3084263801574707, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 0.8103, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.11755485893416928, | |
| "grad_norm": 2.4108471870422363, | |
| "learning_rate": 1.25e-06, | |
| "loss": 0.808, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.12225705329153605, | |
| "grad_norm": 2.116116762161255, | |
| "learning_rate": 1.3e-06, | |
| "loss": 0.7901, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.12695924764890282, | |
| "grad_norm": 2.2457115650177, | |
| "learning_rate": 1.3500000000000002e-06, | |
| "loss": 0.7737, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.13166144200626959, | |
| "grad_norm": 1.9436711072921753, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.7465, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.13636363636363635, | |
| "grad_norm": 2.1678073406219482, | |
| "learning_rate": 1.45e-06, | |
| "loss": 0.7819, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.14106583072100312, | |
| "grad_norm": 1.8869282007217407, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.7653, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.14576802507836992, | |
| "grad_norm": 1.6957731246948242, | |
| "learning_rate": 1.5500000000000002e-06, | |
| "loss": 0.7659, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.15047021943573669, | |
| "grad_norm": 1.4701424837112427, | |
| "learning_rate": 1.6000000000000001e-06, | |
| "loss": 0.7553, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.15517241379310345, | |
| "grad_norm": 1.3234986066818237, | |
| "learning_rate": 1.6500000000000003e-06, | |
| "loss": 0.6875, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.15987460815047022, | |
| "grad_norm": 1.3325831890106201, | |
| "learning_rate": 1.7000000000000002e-06, | |
| "loss": 0.6681, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.164576802507837, | |
| "grad_norm": 1.3700138330459595, | |
| "learning_rate": 1.75e-06, | |
| "loss": 0.7635, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.16927899686520376, | |
| "grad_norm": 1.2316628694534302, | |
| "learning_rate": 1.8000000000000001e-06, | |
| "loss": 0.6776, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.17398119122257052, | |
| "grad_norm": 1.170616865158081, | |
| "learning_rate": 1.85e-06, | |
| "loss": 0.7073, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.1786833855799373, | |
| "grad_norm": 1.1797659397125244, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.7572, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.1833855799373041, | |
| "grad_norm": 1.0123047828674316, | |
| "learning_rate": 1.9500000000000004e-06, | |
| "loss": 0.6798, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.18808777429467086, | |
| "grad_norm": 0.975169837474823, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.7088, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.19278996865203762, | |
| "grad_norm": 0.891377866268158, | |
| "learning_rate": 2.05e-06, | |
| "loss": 0.6696, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.1974921630094044, | |
| "grad_norm": 0.8997182250022888, | |
| "learning_rate": 2.1000000000000002e-06, | |
| "loss": 0.7135, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.20219435736677116, | |
| "grad_norm": 0.9378617405891418, | |
| "learning_rate": 2.15e-06, | |
| "loss": 0.6827, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.20689655172413793, | |
| "grad_norm": 0.8060838580131531, | |
| "learning_rate": 2.2e-06, | |
| "loss": 0.6704, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.2115987460815047, | |
| "grad_norm": 0.7708892822265625, | |
| "learning_rate": 2.25e-06, | |
| "loss": 0.6499, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.21630094043887146, | |
| "grad_norm": 0.7236770987510681, | |
| "learning_rate": 2.3000000000000004e-06, | |
| "loss": 0.657, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.22100313479623823, | |
| "grad_norm": 0.6666117906570435, | |
| "learning_rate": 2.35e-06, | |
| "loss": 0.634, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.22570532915360503, | |
| "grad_norm": 0.6475918292999268, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.6237, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.2304075235109718, | |
| "grad_norm": 0.7277692556381226, | |
| "learning_rate": 2.4500000000000003e-06, | |
| "loss": 0.6707, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.23510971786833856, | |
| "grad_norm": 0.6572731733322144, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.6812, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.23981191222570533, | |
| "grad_norm": 0.6716486811637878, | |
| "learning_rate": 2.55e-06, | |
| "loss": 0.6803, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.2445141065830721, | |
| "grad_norm": 0.6065218448638916, | |
| "learning_rate": 2.6e-06, | |
| "loss": 0.6353, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.24921630094043887, | |
| "grad_norm": 0.6749820709228516, | |
| "learning_rate": 2.6500000000000005e-06, | |
| "loss": 0.6393, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.25391849529780564, | |
| "grad_norm": 0.6187207102775574, | |
| "learning_rate": 2.7000000000000004e-06, | |
| "loss": 0.6318, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.25862068965517243, | |
| "grad_norm": 0.6700596809387207, | |
| "learning_rate": 2.7500000000000004e-06, | |
| "loss": 0.6762, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.26332288401253917, | |
| "grad_norm": 0.6460264921188354, | |
| "learning_rate": 2.8000000000000003e-06, | |
| "loss": 0.6388, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.26802507836990597, | |
| "grad_norm": 0.7263085842132568, | |
| "learning_rate": 2.85e-06, | |
| "loss": 0.6413, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.2727272727272727, | |
| "grad_norm": 0.674784779548645, | |
| "learning_rate": 2.9e-06, | |
| "loss": 0.6134, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.2774294670846395, | |
| "grad_norm": 0.6552851796150208, | |
| "learning_rate": 2.95e-06, | |
| "loss": 0.6508, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.28213166144200624, | |
| "grad_norm": 0.7160854339599609, | |
| "learning_rate": 3e-06, | |
| "loss": 0.677, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.28683385579937304, | |
| "grad_norm": 0.682802677154541, | |
| "learning_rate": 3.05e-06, | |
| "loss": 0.5983, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.29153605015673983, | |
| "grad_norm": 0.5360945463180542, | |
| "learning_rate": 3.1000000000000004e-06, | |
| "loss": 0.5928, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.2962382445141066, | |
| "grad_norm": 0.6875145435333252, | |
| "learning_rate": 3.1500000000000003e-06, | |
| "loss": 0.6366, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.30094043887147337, | |
| "grad_norm": 0.5564191937446594, | |
| "learning_rate": 3.2000000000000003e-06, | |
| "loss": 0.6466, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.3056426332288401, | |
| "grad_norm": 0.8959600925445557, | |
| "learning_rate": 3.2500000000000002e-06, | |
| "loss": 0.5875, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.3103448275862069, | |
| "grad_norm": 0.5513110160827637, | |
| "learning_rate": 3.3000000000000006e-06, | |
| "loss": 0.6042, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.31504702194357365, | |
| "grad_norm": 0.5638250112533569, | |
| "learning_rate": 3.3500000000000005e-06, | |
| "loss": 0.6343, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.31974921630094044, | |
| "grad_norm": 0.5723473429679871, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.6007, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.32445141065830724, | |
| "grad_norm": 0.5193734765052795, | |
| "learning_rate": 3.45e-06, | |
| "loss": 0.5877, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.329153605015674, | |
| "grad_norm": 0.7437443733215332, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.6579, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.3338557993730408, | |
| "grad_norm": 0.5810403823852539, | |
| "learning_rate": 3.5500000000000003e-06, | |
| "loss": 0.625, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.3385579937304075, | |
| "grad_norm": 0.5594515204429626, | |
| "learning_rate": 3.6000000000000003e-06, | |
| "loss": 0.6039, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.3432601880877743, | |
| "grad_norm": 0.5781659483909607, | |
| "learning_rate": 3.65e-06, | |
| "loss": 0.5936, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.34796238244514105, | |
| "grad_norm": 0.5736168622970581, | |
| "learning_rate": 3.7e-06, | |
| "loss": 0.61, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.35266457680250785, | |
| "grad_norm": 0.5082181096076965, | |
| "learning_rate": 3.7500000000000005e-06, | |
| "loss": 0.6031, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.3573667711598746, | |
| "grad_norm": 0.5188243985176086, | |
| "learning_rate": 3.8000000000000005e-06, | |
| "loss": 0.6048, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.3620689655172414, | |
| "grad_norm": 0.678931713104248, | |
| "learning_rate": 3.85e-06, | |
| "loss": 0.615, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3667711598746082, | |
| "grad_norm": 0.5580578446388245, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.5622, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.3714733542319749, | |
| "grad_norm": 0.5222985744476318, | |
| "learning_rate": 3.95e-06, | |
| "loss": 0.6233, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.3761755485893417, | |
| "grad_norm": 0.6105766892433167, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.5448, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.38087774294670845, | |
| "grad_norm": 0.4837906062602997, | |
| "learning_rate": 4.05e-06, | |
| "loss": 0.6058, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.38557993730407525, | |
| "grad_norm": 0.5260081887245178, | |
| "learning_rate": 4.1e-06, | |
| "loss": 0.5753, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.390282131661442, | |
| "grad_norm": 0.5424107909202576, | |
| "learning_rate": 4.15e-06, | |
| "loss": 0.5836, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.3949843260188088, | |
| "grad_norm": 0.5307353734970093, | |
| "learning_rate": 4.2000000000000004e-06, | |
| "loss": 0.5891, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.3996865203761755, | |
| "grad_norm": 0.5606550574302673, | |
| "learning_rate": 4.25e-06, | |
| "loss": 0.6097, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.4043887147335423, | |
| "grad_norm": 0.5380442142486572, | |
| "learning_rate": 4.3e-06, | |
| "loss": 0.6161, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.4090909090909091, | |
| "grad_norm": 0.7110352516174316, | |
| "learning_rate": 4.350000000000001e-06, | |
| "loss": 0.5681, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.41379310344827586, | |
| "grad_norm": 0.525680422782898, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.5942, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.41849529780564265, | |
| "grad_norm": 0.5567077398300171, | |
| "learning_rate": 4.450000000000001e-06, | |
| "loss": 0.5929, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.4231974921630094, | |
| "grad_norm": 0.5358735918998718, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.5717, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.4278996865203762, | |
| "grad_norm": 0.48081454634666443, | |
| "learning_rate": 4.5500000000000005e-06, | |
| "loss": 0.5984, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.43260188087774293, | |
| "grad_norm": 0.4900401830673218, | |
| "learning_rate": 4.600000000000001e-06, | |
| "loss": 0.6017, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.4373040752351097, | |
| "grad_norm": 0.48369699716567993, | |
| "learning_rate": 4.65e-06, | |
| "loss": 0.5295, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.44200626959247646, | |
| "grad_norm": 0.4992128312587738, | |
| "learning_rate": 4.7e-06, | |
| "loss": 0.5495, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.44670846394984326, | |
| "grad_norm": 0.5139058828353882, | |
| "learning_rate": 4.75e-06, | |
| "loss": 0.5528, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.45141065830721006, | |
| "grad_norm": 0.485097199678421, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 0.5528, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.4561128526645768, | |
| "grad_norm": 0.5944838523864746, | |
| "learning_rate": 4.85e-06, | |
| "loss": 0.5577, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.4608150470219436, | |
| "grad_norm": 0.5313893556594849, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.5744, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.46551724137931033, | |
| "grad_norm": 0.49062076210975647, | |
| "learning_rate": 4.95e-06, | |
| "loss": 0.5563, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.4702194357366771, | |
| "grad_norm": 0.5196151733398438, | |
| "learning_rate": 5e-06, | |
| "loss": 0.5817, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.47492163009404387, | |
| "grad_norm": 0.507022500038147, | |
| "learning_rate": 4.9999910183883085e-06, | |
| "loss": 0.5608, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.47962382445141066, | |
| "grad_norm": 0.494018018245697, | |
| "learning_rate": 4.999964073617768e-06, | |
| "loss": 0.57, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.4843260188087774, | |
| "grad_norm": 0.48901236057281494, | |
| "learning_rate": 4.999919165881985e-06, | |
| "loss": 0.5568, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.4890282131661442, | |
| "grad_norm": 0.5166047811508179, | |
| "learning_rate": 4.999856295503635e-06, | |
| "loss": 0.5676, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.493730407523511, | |
| "grad_norm": 0.4952854812145233, | |
| "learning_rate": 4.9997754629344596e-06, | |
| "loss": 0.5454, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.49843260188087773, | |
| "grad_norm": 0.48603224754333496, | |
| "learning_rate": 4.999676668755263e-06, | |
| "loss": 0.5351, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.5031347962382445, | |
| "grad_norm": 0.4922390878200531, | |
| "learning_rate": 4.999559913675912e-06, | |
| "loss": 0.5672, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.5078369905956113, | |
| "grad_norm": 0.4846023619174957, | |
| "learning_rate": 4.999425198535325e-06, | |
| "loss": 0.5472, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.512539184952978, | |
| "grad_norm": 0.4974515736103058, | |
| "learning_rate": 4.999272524301469e-06, | |
| "loss": 0.5507, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.5172413793103449, | |
| "grad_norm": 0.5394945740699768, | |
| "learning_rate": 4.9991018920713505e-06, | |
| "loss": 0.5667, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.5219435736677116, | |
| "grad_norm": 0.5393733978271484, | |
| "learning_rate": 4.9989133030710154e-06, | |
| "loss": 0.5505, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.5266457680250783, | |
| "grad_norm": 0.7595535516738892, | |
| "learning_rate": 4.9987067586555275e-06, | |
| "loss": 0.5484, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.5313479623824452, | |
| "grad_norm": 0.5122578144073486, | |
| "learning_rate": 4.998482260308969e-06, | |
| "loss": 0.5669, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.5360501567398119, | |
| "grad_norm": 0.5233834981918335, | |
| "learning_rate": 4.998239809644427e-06, | |
| "loss": 0.5613, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.5407523510971787, | |
| "grad_norm": 3.08868408203125, | |
| "learning_rate": 4.9979794084039755e-06, | |
| "loss": 0.5746, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.5454545454545454, | |
| "grad_norm": 0.49716249108314514, | |
| "learning_rate": 4.997701058458677e-06, | |
| "loss": 0.5349, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.5501567398119123, | |
| "grad_norm": 0.5552968978881836, | |
| "learning_rate": 4.997404761808554e-06, | |
| "loss": 0.569, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.554858934169279, | |
| "grad_norm": 0.5029860734939575, | |
| "learning_rate": 4.9970905205825845e-06, | |
| "loss": 0.5755, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.5595611285266457, | |
| "grad_norm": 0.5006077885627747, | |
| "learning_rate": 4.996758337038683e-06, | |
| "loss": 0.5654, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.5642633228840125, | |
| "grad_norm": 0.4900127053260803, | |
| "learning_rate": 4.996408213563684e-06, | |
| "loss": 0.5575, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.5689655172413793, | |
| "grad_norm": 0.5142834186553955, | |
| "learning_rate": 4.996040152673326e-06, | |
| "loss": 0.5247, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.5736677115987461, | |
| "grad_norm": 0.5097452998161316, | |
| "learning_rate": 4.995654157012233e-06, | |
| "loss": 0.538, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.5783699059561128, | |
| "grad_norm": 0.5581697225570679, | |
| "learning_rate": 4.995250229353895e-06, | |
| "loss": 0.5324, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.5830721003134797, | |
| "grad_norm": 0.5943841934204102, | |
| "learning_rate": 4.99482837260065e-06, | |
| "loss": 0.547, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.5877742946708464, | |
| "grad_norm": 0.541631281375885, | |
| "learning_rate": 4.99438858978366e-06, | |
| "loss": 0.577, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5924764890282131, | |
| "grad_norm": 0.6008974313735962, | |
| "learning_rate": 4.993930884062892e-06, | |
| "loss": 0.5579, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.5971786833855799, | |
| "grad_norm": 0.7009158730506897, | |
| "learning_rate": 4.993455258727094e-06, | |
| "loss": 0.5579, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.6018808777429467, | |
| "grad_norm": 0.5942498445510864, | |
| "learning_rate": 4.992961717193773e-06, | |
| "loss": 0.5599, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.6065830721003135, | |
| "grad_norm": 0.4936768710613251, | |
| "learning_rate": 4.9924502630091655e-06, | |
| "loss": 0.5676, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.6112852664576802, | |
| "grad_norm": 0.5035172700881958, | |
| "learning_rate": 4.99192089984822e-06, | |
| "loss": 0.5519, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.6159874608150471, | |
| "grad_norm": 0.8178919553756714, | |
| "learning_rate": 4.9913736315145614e-06, | |
| "loss": 0.5594, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.6206896551724138, | |
| "grad_norm": 0.4977147877216339, | |
| "learning_rate": 4.990808461940474e-06, | |
| "loss": 0.54, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.6253918495297806, | |
| "grad_norm": 0.7395517826080322, | |
| "learning_rate": 4.990225395186862e-06, | |
| "loss": 0.5467, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.6300940438871473, | |
| "grad_norm": 1.1424250602722168, | |
| "learning_rate": 4.9896244354432314e-06, | |
| "loss": 0.5432, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.6347962382445141, | |
| "grad_norm": 0.5002970695495605, | |
| "learning_rate": 4.98900558702765e-06, | |
| "loss": 0.5526, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.6394984326018809, | |
| "grad_norm": 0.49464455246925354, | |
| "learning_rate": 4.9883688543867225e-06, | |
| "loss": 0.563, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.6442006269592476, | |
| "grad_norm": 0.48660922050476074, | |
| "learning_rate": 4.987714242095558e-06, | |
| "loss": 0.5328, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.6489028213166145, | |
| "grad_norm": 0.5086668133735657, | |
| "learning_rate": 4.9870417548577355e-06, | |
| "loss": 0.5373, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.6536050156739812, | |
| "grad_norm": 0.4638510048389435, | |
| "learning_rate": 4.9863513975052696e-06, | |
| "loss": 0.5459, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.658307210031348, | |
| "grad_norm": 0.5742124915122986, | |
| "learning_rate": 4.985643174998578e-06, | |
| "loss": 0.5681, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.6630094043887147, | |
| "grad_norm": 0.5262806415557861, | |
| "learning_rate": 4.984917092426445e-06, | |
| "loss": 0.5549, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.6677115987460815, | |
| "grad_norm": 0.4765531122684479, | |
| "learning_rate": 4.984173155005982e-06, | |
| "loss": 0.5435, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.6724137931034483, | |
| "grad_norm": 0.4731754660606384, | |
| "learning_rate": 4.983411368082597e-06, | |
| "loss": 0.5484, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.677115987460815, | |
| "grad_norm": 0.5015071630477905, | |
| "learning_rate": 4.982631737129948e-06, | |
| "loss": 0.5314, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.6818181818181818, | |
| "grad_norm": 0.5013841986656189, | |
| "learning_rate": 4.98183426774991e-06, | |
| "loss": 0.5705, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.6865203761755486, | |
| "grad_norm": 0.5398444533348083, | |
| "learning_rate": 4.981018965672529e-06, | |
| "loss": 0.5713, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.6912225705329154, | |
| "grad_norm": 0.5247324109077454, | |
| "learning_rate": 4.98018583675599e-06, | |
| "loss": 0.5512, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.6959247648902821, | |
| "grad_norm": 0.488935649394989, | |
| "learning_rate": 4.979334886986562e-06, | |
| "loss": 0.5532, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.700626959247649, | |
| "grad_norm": 0.5056779384613037, | |
| "learning_rate": 4.978466122478567e-06, | |
| "loss": 0.5668, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.7053291536050157, | |
| "grad_norm": 0.5102982521057129, | |
| "learning_rate": 4.97757954947433e-06, | |
| "loss": 0.5346, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.7100313479623824, | |
| "grad_norm": 0.4989202320575714, | |
| "learning_rate": 4.976675174344132e-06, | |
| "loss": 0.5488, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.7147335423197492, | |
| "grad_norm": 0.4996865689754486, | |
| "learning_rate": 4.975753003586172e-06, | |
| "loss": 0.528, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.719435736677116, | |
| "grad_norm": 0.5011430978775024, | |
| "learning_rate": 4.974813043826513e-06, | |
| "loss": 0.5221, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.7241379310344828, | |
| "grad_norm": 0.530082643032074, | |
| "learning_rate": 4.973855301819039e-06, | |
| "loss": 0.5398, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.7288401253918495, | |
| "grad_norm": 0.5243843197822571, | |
| "learning_rate": 4.972879784445402e-06, | |
| "loss": 0.5215, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.7335423197492164, | |
| "grad_norm": 0.5321754217147827, | |
| "learning_rate": 4.971886498714978e-06, | |
| "loss": 0.5432, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.7382445141065831, | |
| "grad_norm": 0.5054258704185486, | |
| "learning_rate": 4.97087545176481e-06, | |
| "loss": 0.5197, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.7429467084639498, | |
| "grad_norm": 0.5209529995918274, | |
| "learning_rate": 4.9698466508595655e-06, | |
| "loss": 0.5486, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.7476489028213166, | |
| "grad_norm": 0.47057369351387024, | |
| "learning_rate": 4.9688001033914756e-06, | |
| "loss": 0.5358, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.7523510971786834, | |
| "grad_norm": 0.5286029577255249, | |
| "learning_rate": 4.967735816880286e-06, | |
| "loss": 0.5471, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.7570532915360502, | |
| "grad_norm": 0.7403479218482971, | |
| "learning_rate": 4.966653798973205e-06, | |
| "loss": 0.5264, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.7617554858934169, | |
| "grad_norm": 0.5945819020271301, | |
| "learning_rate": 4.965554057444842e-06, | |
| "loss": 0.5194, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.7664576802507836, | |
| "grad_norm": 0.48147913813591003, | |
| "learning_rate": 4.964436600197161e-06, | |
| "loss": 0.5425, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.7711598746081505, | |
| "grad_norm": 0.5892655849456787, | |
| "learning_rate": 4.963301435259413e-06, | |
| "loss": 0.5134, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.7758620689655172, | |
| "grad_norm": 0.546521782875061, | |
| "learning_rate": 4.962148570788088e-06, | |
| "loss": 0.5334, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.780564263322884, | |
| "grad_norm": 0.5432559251785278, | |
| "learning_rate": 4.96097801506685e-06, | |
| "loss": 0.5214, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.7852664576802508, | |
| "grad_norm": 0.4892060458660126, | |
| "learning_rate": 4.959789776506482e-06, | |
| "loss": 0.5307, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.7899686520376176, | |
| "grad_norm": 0.5296177268028259, | |
| "learning_rate": 4.958583863644821e-06, | |
| "loss": 0.5539, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.7946708463949843, | |
| "grad_norm": 0.4970422685146332, | |
| "learning_rate": 4.9573602851466985e-06, | |
| "loss": 0.5193, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.799373040752351, | |
| "grad_norm": 0.5201449990272522, | |
| "learning_rate": 4.9561190498038815e-06, | |
| "loss": 0.542, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.8040752351097179, | |
| "grad_norm": 0.6067378520965576, | |
| "learning_rate": 4.954860166535005e-06, | |
| "loss": 0.5319, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.8087774294670846, | |
| "grad_norm": 0.4754677712917328, | |
| "learning_rate": 4.95358364438551e-06, | |
| "loss": 0.5366, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.8134796238244514, | |
| "grad_norm": 0.4970039129257202, | |
| "learning_rate": 4.952289492527576e-06, | |
| "loss": 0.5626, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.8181818181818182, | |
| "grad_norm": 0.5602061748504639, | |
| "learning_rate": 4.9509777202600605e-06, | |
| "loss": 0.4951, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.822884012539185, | |
| "grad_norm": 0.5069930553436279, | |
| "learning_rate": 4.949648337008425e-06, | |
| "loss": 0.5419, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.8275862068965517, | |
| "grad_norm": 2.1186537742614746, | |
| "learning_rate": 4.948301352324674e-06, | |
| "loss": 0.5361, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.8322884012539185, | |
| "grad_norm": 0.4984792470932007, | |
| "learning_rate": 4.946936775887281e-06, | |
| "loss": 0.5315, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.8369905956112853, | |
| "grad_norm": 0.49590349197387695, | |
| "learning_rate": 4.945554617501124e-06, | |
| "loss": 0.5258, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.841692789968652, | |
| "grad_norm": 0.48108625411987305, | |
| "learning_rate": 4.944154887097411e-06, | |
| "loss": 0.5566, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.8463949843260188, | |
| "grad_norm": 0.5167516469955444, | |
| "learning_rate": 4.942737594733608e-06, | |
| "loss": 0.5285, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.8510971786833855, | |
| "grad_norm": 0.4649943709373474, | |
| "learning_rate": 4.941302750593373e-06, | |
| "loss": 0.5443, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.8557993730407524, | |
| "grad_norm": 0.5184140205383301, | |
| "learning_rate": 4.939850364986475e-06, | |
| "loss": 0.4859, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.8605015673981191, | |
| "grad_norm": 0.6746440529823303, | |
| "learning_rate": 4.938380448348725e-06, | |
| "loss": 0.4947, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.8652037617554859, | |
| "grad_norm": 0.5150278806686401, | |
| "learning_rate": 4.9368930112419e-06, | |
| "loss": 0.5351, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.8699059561128527, | |
| "grad_norm": 0.47614121437072754, | |
| "learning_rate": 4.935388064353665e-06, | |
| "loss": 0.5362, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.8746081504702194, | |
| "grad_norm": 0.5044179558753967, | |
| "learning_rate": 4.9338656184975e-06, | |
| "loss": 0.532, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.8793103448275862, | |
| "grad_norm": 0.4887774884700775, | |
| "learning_rate": 4.932325684612618e-06, | |
| "loss": 0.5448, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.8840125391849529, | |
| "grad_norm": 0.49574583768844604, | |
| "learning_rate": 4.93076827376389e-06, | |
| "loss": 0.5485, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.8887147335423198, | |
| "grad_norm": 0.5294933915138245, | |
| "learning_rate": 4.9291933971417635e-06, | |
| "loss": 0.5441, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.8934169278996865, | |
| "grad_norm": 0.4594694972038269, | |
| "learning_rate": 4.9276010660621835e-06, | |
| "loss": 0.5322, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.8981191222570533, | |
| "grad_norm": 0.47507232427597046, | |
| "learning_rate": 4.925991291966508e-06, | |
| "loss": 0.5103, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.9028213166144201, | |
| "grad_norm": 0.4808727204799652, | |
| "learning_rate": 4.92436408642143e-06, | |
| "loss": 0.5479, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.9075235109717869, | |
| "grad_norm": 0.481921523809433, | |
| "learning_rate": 4.9227194611188934e-06, | |
| "loss": 0.5242, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.9122257053291536, | |
| "grad_norm": 0.6387396454811096, | |
| "learning_rate": 4.921057427876007e-06, | |
| "loss": 0.4937, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.9169278996865203, | |
| "grad_norm": 0.5185182094573975, | |
| "learning_rate": 4.919377998634959e-06, | |
| "loss": 0.5515, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.9216300940438872, | |
| "grad_norm": 0.5029557943344116, | |
| "learning_rate": 4.917681185462934e-06, | |
| "loss": 0.5371, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.9263322884012539, | |
| "grad_norm": 0.5311322808265686, | |
| "learning_rate": 4.915967000552028e-06, | |
| "loss": 0.5275, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.9310344827586207, | |
| "grad_norm": 0.46653756499290466, | |
| "learning_rate": 4.914235456219154e-06, | |
| "loss": 0.5208, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.9357366771159875, | |
| "grad_norm": 0.4970494508743286, | |
| "learning_rate": 4.912486564905959e-06, | |
| "loss": 0.549, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.9404388714733543, | |
| "grad_norm": 0.5315521359443665, | |
| "learning_rate": 4.910720339178735e-06, | |
| "loss": 0.5314, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.945141065830721, | |
| "grad_norm": 0.5144534707069397, | |
| "learning_rate": 4.908936791728323e-06, | |
| "loss": 0.5372, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.9498432601880877, | |
| "grad_norm": 0.6076200008392334, | |
| "learning_rate": 4.907135935370027e-06, | |
| "loss": 0.5353, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.9545454545454546, | |
| "grad_norm": 0.5259833931922913, | |
| "learning_rate": 4.905317783043523e-06, | |
| "loss": 0.5442, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.9592476489028213, | |
| "grad_norm": 0.4931281507015228, | |
| "learning_rate": 4.9034823478127605e-06, | |
| "loss": 0.5246, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.9639498432601881, | |
| "grad_norm": 0.4909270405769348, | |
| "learning_rate": 4.901629642865872e-06, | |
| "loss": 0.5011, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.9686520376175548, | |
| "grad_norm": 0.581427276134491, | |
| "learning_rate": 4.89975968151508e-06, | |
| "loss": 0.5253, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.9733542319749217, | |
| "grad_norm": 0.5055932998657227, | |
| "learning_rate": 4.8978724771965965e-06, | |
| "loss": 0.5152, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.9780564263322884, | |
| "grad_norm": 0.5337440371513367, | |
| "learning_rate": 4.895968043470532e-06, | |
| "loss": 0.5321, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.9827586206896551, | |
| "grad_norm": 0.5171258449554443, | |
| "learning_rate": 4.894046394020794e-06, | |
| "loss": 0.4973, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.987460815047022, | |
| "grad_norm": 0.4880523085594177, | |
| "learning_rate": 4.892107542654988e-06, | |
| "loss": 0.529, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.9921630094043887, | |
| "grad_norm": 0.4594219923019409, | |
| "learning_rate": 4.890151503304325e-06, | |
| "loss": 0.5502, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.9968652037617555, | |
| "grad_norm": 0.5071013569831848, | |
| "learning_rate": 4.88817829002351e-06, | |
| "loss": 0.5258, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 1.0047021943573669, | |
| "grad_norm": 0.953562319278717, | |
| "learning_rate": 4.886187916990653e-06, | |
| "loss": 1.0585, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 1.0094043887147335, | |
| "grad_norm": 0.46260955929756165, | |
| "learning_rate": 4.884180398507163e-06, | |
| "loss": 0.5036, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 1.0141065830721003, | |
| "grad_norm": 0.46361738443374634, | |
| "learning_rate": 4.882155748997636e-06, | |
| "loss": 0.4975, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.0188087774294672, | |
| "grad_norm": 0.47986194491386414, | |
| "learning_rate": 4.8801139830097685e-06, | |
| "loss": 0.5039, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 1.0235109717868338, | |
| "grad_norm": 0.49743711948394775, | |
| "learning_rate": 4.878055115214238e-06, | |
| "loss": 0.5139, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 1.0282131661442007, | |
| "grad_norm": 0.5503370761871338, | |
| "learning_rate": 4.875979160404607e-06, | |
| "loss": 0.5117, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 1.0329153605015673, | |
| "grad_norm": 0.47498467564582825, | |
| "learning_rate": 4.873886133497209e-06, | |
| "loss": 0.5195, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 1.0376175548589341, | |
| "grad_norm": 0.4586140513420105, | |
| "learning_rate": 4.87177604953105e-06, | |
| "loss": 0.5168, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.042319749216301, | |
| "grad_norm": 0.5042305588722229, | |
| "learning_rate": 4.869648923667694e-06, | |
| "loss": 0.4693, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 1.0470219435736676, | |
| "grad_norm": 0.6340722441673279, | |
| "learning_rate": 4.867504771191154e-06, | |
| "loss": 0.4945, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 1.0517241379310345, | |
| "grad_norm": 0.4741908311843872, | |
| "learning_rate": 4.865343607507788e-06, | |
| "loss": 0.5027, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 1.0564263322884013, | |
| "grad_norm": 0.6638283133506775, | |
| "learning_rate": 4.86316544814618e-06, | |
| "loss": 0.5191, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 1.061128526645768, | |
| "grad_norm": 0.7720419764518738, | |
| "learning_rate": 4.860970308757038e-06, | |
| "loss": 0.4634, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.0658307210031348, | |
| "grad_norm": 0.5258729457855225, | |
| "learning_rate": 4.858758205113072e-06, | |
| "loss": 0.4924, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 1.0705329153605017, | |
| "grad_norm": 0.5001716017723083, | |
| "learning_rate": 4.856529153108888e-06, | |
| "loss": 0.5241, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 1.0752351097178683, | |
| "grad_norm": 0.5368038415908813, | |
| "learning_rate": 4.854283168760868e-06, | |
| "loss": 0.5031, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 1.0799373040752351, | |
| "grad_norm": 0.5357080101966858, | |
| "learning_rate": 4.85202026820706e-06, | |
| "loss": 0.5003, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 1.084639498432602, | |
| "grad_norm": 0.5052481293678284, | |
| "learning_rate": 4.84974046770706e-06, | |
| "loss": 0.5375, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.0893416927899686, | |
| "grad_norm": 0.5544849634170532, | |
| "learning_rate": 4.847443783641893e-06, | |
| "loss": 0.4472, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 1.0940438871473355, | |
| "grad_norm": 0.49668335914611816, | |
| "learning_rate": 4.845130232513901e-06, | |
| "loss": 0.4934, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 1.098746081504702, | |
| "grad_norm": 0.5318028330802917, | |
| "learning_rate": 4.842799830946615e-06, | |
| "loss": 0.4903, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 1.103448275862069, | |
| "grad_norm": 0.48978278040885925, | |
| "learning_rate": 4.840452595684646e-06, | |
| "loss": 0.4864, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 1.1081504702194358, | |
| "grad_norm": 0.47105321288108826, | |
| "learning_rate": 4.83808854359356e-06, | |
| "loss": 0.4805, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 1.1128526645768024, | |
| "grad_norm": 0.6040552854537964, | |
| "learning_rate": 4.835707691659753e-06, | |
| "loss": 0.4839, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 1.1175548589341693, | |
| "grad_norm": 0.501312255859375, | |
| "learning_rate": 4.8333100569903365e-06, | |
| "loss": 0.495, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 1.1222570532915361, | |
| "grad_norm": 0.48165130615234375, | |
| "learning_rate": 4.8308956568130094e-06, | |
| "loss": 0.5144, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 1.1269592476489028, | |
| "grad_norm": 0.5097485780715942, | |
| "learning_rate": 4.828464508475934e-06, | |
| "loss": 0.5081, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 1.1316614420062696, | |
| "grad_norm": 0.49503275752067566, | |
| "learning_rate": 4.826016629447616e-06, | |
| "loss": 0.5103, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.1363636363636362, | |
| "grad_norm": 0.5472146272659302, | |
| "learning_rate": 4.823552037316775e-06, | |
| "loss": 0.4857, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 1.141065830721003, | |
| "grad_norm": 0.5358877182006836, | |
| "learning_rate": 4.821070749792218e-06, | |
| "loss": 0.5418, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 1.14576802507837, | |
| "grad_norm": 0.4832223951816559, | |
| "learning_rate": 4.818572784702713e-06, | |
| "loss": 0.5121, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 1.1504702194357366, | |
| "grad_norm": 0.928979218006134, | |
| "learning_rate": 4.816058159996863e-06, | |
| "loss": 0.5218, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 1.1551724137931034, | |
| "grad_norm": 0.46563178300857544, | |
| "learning_rate": 4.813526893742972e-06, | |
| "loss": 0.5045, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 1.1598746081504703, | |
| "grad_norm": 0.5262567400932312, | |
| "learning_rate": 4.810979004128924e-06, | |
| "loss": 0.4984, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 1.164576802507837, | |
| "grad_norm": 0.4928685426712036, | |
| "learning_rate": 4.808414509462042e-06, | |
| "loss": 0.5183, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 1.1692789968652038, | |
| "grad_norm": 0.8799586892127991, | |
| "learning_rate": 4.80583342816896e-06, | |
| "loss": 0.4851, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 1.1739811912225706, | |
| "grad_norm": 0.5468968749046326, | |
| "learning_rate": 4.803235778795496e-06, | |
| "loss": 0.5242, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 1.1786833855799372, | |
| "grad_norm": 0.49233874678611755, | |
| "learning_rate": 4.800621580006511e-06, | |
| "loss": 0.4687, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.183385579937304, | |
| "grad_norm": 0.4705159366130829, | |
| "learning_rate": 4.797990850585782e-06, | |
| "loss": 0.5166, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 1.188087774294671, | |
| "grad_norm": 0.49235984683036804, | |
| "learning_rate": 4.79534360943586e-06, | |
| "loss": 0.4964, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.1927899686520376, | |
| "grad_norm": 0.5998479127883911, | |
| "learning_rate": 4.792679875577937e-06, | |
| "loss": 0.4816, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 1.1974921630094044, | |
| "grad_norm": 0.6658573746681213, | |
| "learning_rate": 4.789999668151714e-06, | |
| "loss": 0.5157, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.2021943573667713, | |
| "grad_norm": 0.47048208117485046, | |
| "learning_rate": 4.7873030064152545e-06, | |
| "loss": 0.4957, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.206896551724138, | |
| "grad_norm": 0.47910359501838684, | |
| "learning_rate": 4.784589909744856e-06, | |
| "loss": 0.4906, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 1.2115987460815048, | |
| "grad_norm": 0.6204367876052856, | |
| "learning_rate": 4.7818603976349005e-06, | |
| "loss": 0.5032, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.2163009404388714, | |
| "grad_norm": 0.48487767577171326, | |
| "learning_rate": 4.779114489697724e-06, | |
| "loss": 0.4986, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.2210031347962382, | |
| "grad_norm": 0.5075157284736633, | |
| "learning_rate": 4.776352205663469e-06, | |
| "loss": 0.5021, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 1.225705329153605, | |
| "grad_norm": 0.4977814257144928, | |
| "learning_rate": 4.773573565379947e-06, | |
| "loss": 0.5132, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.2304075235109717, | |
| "grad_norm": 0.515428900718689, | |
| "learning_rate": 4.770778588812489e-06, | |
| "loss": 0.4783, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 1.2351097178683386, | |
| "grad_norm": 0.5740492939949036, | |
| "learning_rate": 4.7679672960438135e-06, | |
| "loss": 0.505, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 1.2398119122257054, | |
| "grad_norm": 0.46476778388023376, | |
| "learning_rate": 4.765139707273872e-06, | |
| "loss": 0.4925, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 1.244514106583072, | |
| "grad_norm": 0.4873107373714447, | |
| "learning_rate": 4.762295842819707e-06, | |
| "loss": 0.5057, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 1.249216300940439, | |
| "grad_norm": 0.5337681174278259, | |
| "learning_rate": 4.759435723115308e-06, | |
| "loss": 0.4709, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.2539184952978055, | |
| "grad_norm": 0.5014109015464783, | |
| "learning_rate": 4.756559368711463e-06, | |
| "loss": 0.5074, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 1.2586206896551724, | |
| "grad_norm": 0.4641067385673523, | |
| "learning_rate": 4.75366680027561e-06, | |
| "loss": 0.4882, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 1.2633228840125392, | |
| "grad_norm": 0.5014533996582031, | |
| "learning_rate": 4.7507580385916906e-06, | |
| "loss": 0.4911, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.2680250783699059, | |
| "grad_norm": 0.7282389402389526, | |
| "learning_rate": 4.747833104559999e-06, | |
| "loss": 0.4674, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 1.2727272727272727, | |
| "grad_norm": 0.44998112320899963, | |
| "learning_rate": 4.744892019197033e-06, | |
| "loss": 0.4818, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.2774294670846396, | |
| "grad_norm": 0.48868605494499207, | |
| "learning_rate": 4.74193480363534e-06, | |
| "loss": 0.4891, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 1.2821316614420062, | |
| "grad_norm": 0.49108532071113586, | |
| "learning_rate": 4.738961479123373e-06, | |
| "loss": 0.497, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.286833855799373, | |
| "grad_norm": 0.49707749485969543, | |
| "learning_rate": 4.735972067025326e-06, | |
| "loss": 0.5044, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 1.29153605015674, | |
| "grad_norm": 0.5466123223304749, | |
| "learning_rate": 4.732966588820991e-06, | |
| "loss": 0.4966, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 1.2962382445141065, | |
| "grad_norm": 0.4684564769268036, | |
| "learning_rate": 4.729945066105599e-06, | |
| "loss": 0.475, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.3009404388714734, | |
| "grad_norm": 0.5143007636070251, | |
| "learning_rate": 4.726907520589664e-06, | |
| "loss": 0.4669, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.3056426332288402, | |
| "grad_norm": 0.5112866163253784, | |
| "learning_rate": 4.72385397409883e-06, | |
| "loss": 0.5085, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 1.3103448275862069, | |
| "grad_norm": 0.5432227849960327, | |
| "learning_rate": 4.720784448573712e-06, | |
| "loss": 0.5003, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 1.3150470219435737, | |
| "grad_norm": 0.5111671686172485, | |
| "learning_rate": 4.717698966069739e-06, | |
| "loss": 0.5305, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 1.3197492163009406, | |
| "grad_norm": 0.5037922859191895, | |
| "learning_rate": 4.7145975487569965e-06, | |
| "loss": 0.5081, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.3244514106583072, | |
| "grad_norm": 0.7806075215339661, | |
| "learning_rate": 4.711480218920064e-06, | |
| "loss": 0.4698, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.329153605015674, | |
| "grad_norm": 0.5198766589164734, | |
| "learning_rate": 4.708346998957859e-06, | |
| "loss": 0.5188, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 1.3338557993730409, | |
| "grad_norm": 0.7277726531028748, | |
| "learning_rate": 4.705197911383473e-06, | |
| "loss": 0.4893, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 1.3385579937304075, | |
| "grad_norm": 0.4980437457561493, | |
| "learning_rate": 4.7020329788240115e-06, | |
| "loss": 0.476, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 1.3432601880877744, | |
| "grad_norm": 0.5318494439125061, | |
| "learning_rate": 4.6988522240204325e-06, | |
| "loss": 0.4853, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.347962382445141, | |
| "grad_norm": 0.5113736391067505, | |
| "learning_rate": 4.695655669827377e-06, | |
| "loss": 0.5005, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 1.3526645768025078, | |
| "grad_norm": 0.7522029876708984, | |
| "learning_rate": 4.6924433392130135e-06, | |
| "loss": 0.4924, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 1.3573667711598745, | |
| "grad_norm": 0.4861883521080017, | |
| "learning_rate": 4.689215255258866e-06, | |
| "loss": 0.5099, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 1.3620689655172413, | |
| "grad_norm": 0.46011000871658325, | |
| "learning_rate": 4.685971441159653e-06, | |
| "loss": 0.4792, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 1.3667711598746082, | |
| "grad_norm": 0.5154278874397278, | |
| "learning_rate": 4.682711920223115e-06, | |
| "loss": 0.4781, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.3714733542319748, | |
| "grad_norm": 0.45991218090057373, | |
| "learning_rate": 4.679436715869856e-06, | |
| "loss": 0.4989, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 1.3761755485893417, | |
| "grad_norm": 0.49241772294044495, | |
| "learning_rate": 4.676145851633166e-06, | |
| "loss": 0.5159, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 1.3808777429467085, | |
| "grad_norm": 0.523045003414154, | |
| "learning_rate": 4.672839351158856e-06, | |
| "loss": 0.5012, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 1.3855799373040751, | |
| "grad_norm": 0.5598923563957214, | |
| "learning_rate": 4.669517238205089e-06, | |
| "loss": 0.4855, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 1.390282131661442, | |
| "grad_norm": 0.507128894329071, | |
| "learning_rate": 4.666179536642208e-06, | |
| "loss": 0.4845, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.3949843260188088, | |
| "grad_norm": 0.721315860748291, | |
| "learning_rate": 4.662826270452565e-06, | |
| "loss": 0.4817, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 1.3996865203761755, | |
| "grad_norm": 0.560897171497345, | |
| "learning_rate": 4.659457463730347e-06, | |
| "loss": 0.4912, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 1.4043887147335423, | |
| "grad_norm": 0.5121816396713257, | |
| "learning_rate": 4.6560731406814056e-06, | |
| "loss": 0.5061, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 1.4090909090909092, | |
| "grad_norm": 0.4652915596961975, | |
| "learning_rate": 4.65267332562308e-06, | |
| "loss": 0.5088, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 1.4137931034482758, | |
| "grad_norm": 0.5149025321006775, | |
| "learning_rate": 4.649258042984026e-06, | |
| "loss": 0.5071, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.4184952978056427, | |
| "grad_norm": 0.4650140106678009, | |
| "learning_rate": 4.6458273173040395e-06, | |
| "loss": 0.4631, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 1.4231974921630095, | |
| "grad_norm": 0.9061777591705322, | |
| "learning_rate": 4.642381173233874e-06, | |
| "loss": 0.5008, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 1.4278996865203761, | |
| "grad_norm": 0.484729140996933, | |
| "learning_rate": 4.638919635535073e-06, | |
| "loss": 0.4564, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 1.432601880877743, | |
| "grad_norm": 1.1061334609985352, | |
| "learning_rate": 4.635442729079788e-06, | |
| "loss": 0.4843, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 1.4373040752351098, | |
| "grad_norm": 0.5028228163719177, | |
| "learning_rate": 4.6319504788505956e-06, | |
| "loss": 0.4794, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.4420062695924765, | |
| "grad_norm": 0.49323561787605286, | |
| "learning_rate": 4.628442909940325e-06, | |
| "loss": 0.4901, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 1.4467084639498433, | |
| "grad_norm": 0.5237039923667908, | |
| "learning_rate": 4.624920047551874e-06, | |
| "loss": 0.5079, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 1.4514106583072102, | |
| "grad_norm": 0.5407230854034424, | |
| "learning_rate": 4.621381916998029e-06, | |
| "loss": 0.4742, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 1.4561128526645768, | |
| "grad_norm": 0.4688400328159332, | |
| "learning_rate": 4.6178285437012806e-06, | |
| "loss": 0.5096, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 1.4608150470219436, | |
| "grad_norm": 0.6071098446846008, | |
| "learning_rate": 4.6142599531936435e-06, | |
| "loss": 0.4721, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.4655172413793103, | |
| "grad_norm": 0.47543954849243164, | |
| "learning_rate": 4.610676171116475e-06, | |
| "loss": 0.4924, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 1.4702194357366771, | |
| "grad_norm": 0.47207143902778625, | |
| "learning_rate": 4.607077223220286e-06, | |
| "loss": 0.4965, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 1.4749216300940438, | |
| "grad_norm": 0.5051419734954834, | |
| "learning_rate": 4.603463135364556e-06, | |
| "loss": 0.4671, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.4796238244514106, | |
| "grad_norm": 0.5427581667900085, | |
| "learning_rate": 4.5998339335175555e-06, | |
| "loss": 0.4896, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 1.4843260188087775, | |
| "grad_norm": 0.8648138642311096, | |
| "learning_rate": 4.596189643756147e-06, | |
| "loss": 0.4666, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.489028213166144, | |
| "grad_norm": 0.5034400224685669, | |
| "learning_rate": 4.592530292265609e-06, | |
| "loss": 0.4849, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 1.493730407523511, | |
| "grad_norm": 0.5227445960044861, | |
| "learning_rate": 4.58885590533944e-06, | |
| "loss": 0.4942, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 1.4984326018808778, | |
| "grad_norm": 0.4640982151031494, | |
| "learning_rate": 4.585166509379173e-06, | |
| "loss": 0.5165, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 1.5031347962382444, | |
| "grad_norm": 0.5094525814056396, | |
| "learning_rate": 4.581462130894186e-06, | |
| "loss": 0.4934, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 1.5078369905956113, | |
| "grad_norm": 0.5064478516578674, | |
| "learning_rate": 4.57774279650151e-06, | |
| "loss": 0.4847, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.5125391849529781, | |
| "grad_norm": 0.5131570100784302, | |
| "learning_rate": 4.574008532925638e-06, | |
| "loss": 0.5111, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 1.5172413793103448, | |
| "grad_norm": 0.5142849087715149, | |
| "learning_rate": 4.570259366998336e-06, | |
| "loss": 0.4953, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 1.5219435736677116, | |
| "grad_norm": 0.4943976402282715, | |
| "learning_rate": 4.566495325658445e-06, | |
| "loss": 0.5235, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 1.5266457680250785, | |
| "grad_norm": 0.46294504404067993, | |
| "learning_rate": 4.5627164359516915e-06, | |
| "loss": 0.505, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.531347962382445, | |
| "grad_norm": 0.6234234571456909, | |
| "learning_rate": 4.558922725030491e-06, | |
| "loss": 0.4776, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.536050156739812, | |
| "grad_norm": 0.5395891666412354, | |
| "learning_rate": 4.555114220153755e-06, | |
| "loss": 0.4313, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 1.5407523510971788, | |
| "grad_norm": 0.8332454562187195, | |
| "learning_rate": 4.551290948686693e-06, | |
| "loss": 0.5141, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 1.5454545454545454, | |
| "grad_norm": 0.4640675485134125, | |
| "learning_rate": 4.547452938100615e-06, | |
| "loss": 0.5178, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 1.5501567398119123, | |
| "grad_norm": 0.48269912600517273, | |
| "learning_rate": 4.54360021597274e-06, | |
| "loss": 0.4955, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 1.5548589341692791, | |
| "grad_norm": 0.5193467140197754, | |
| "learning_rate": 4.539732809985989e-06, | |
| "loss": 0.4864, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.5595611285266457, | |
| "grad_norm": 0.4845351576805115, | |
| "learning_rate": 4.535850747928796e-06, | |
| "loss": 0.5003, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 1.5642633228840124, | |
| "grad_norm": 0.5264540314674377, | |
| "learning_rate": 4.531954057694897e-06, | |
| "loss": 0.4828, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.5689655172413794, | |
| "grad_norm": 0.4795820415019989, | |
| "learning_rate": 4.5280427672831414e-06, | |
| "loss": 0.4904, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.573667711598746, | |
| "grad_norm": 0.5169357061386108, | |
| "learning_rate": 4.524116904797281e-06, | |
| "loss": 0.4661, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.5783699059561127, | |
| "grad_norm": 0.4765758514404297, | |
| "learning_rate": 4.520176498445774e-06, | |
| "loss": 0.4793, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.5830721003134798, | |
| "grad_norm": 0.5529272556304932, | |
| "learning_rate": 4.516221576541581e-06, | |
| "loss": 0.4793, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.5877742946708464, | |
| "grad_norm": 0.594944179058075, | |
| "learning_rate": 4.512252167501959e-06, | |
| "loss": 0.4786, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.592476489028213, | |
| "grad_norm": 0.4921863079071045, | |
| "learning_rate": 4.508268299848262e-06, | |
| "loss": 0.4854, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.59717868338558, | |
| "grad_norm": 0.4983494281768799, | |
| "learning_rate": 4.50427000220573e-06, | |
| "loss": 0.4999, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.6018808777429467, | |
| "grad_norm": 0.48530513048171997, | |
| "learning_rate": 4.50025730330329e-06, | |
| "loss": 0.477, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.6065830721003134, | |
| "grad_norm": 0.47819897532463074, | |
| "learning_rate": 4.4962302319733445e-06, | |
| "loss": 0.4933, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.6112852664576802, | |
| "grad_norm": 0.608711838722229, | |
| "learning_rate": 4.492188817151565e-06, | |
| "loss": 0.5285, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.615987460815047, | |
| "grad_norm": 0.5040555000305176, | |
| "learning_rate": 4.488133087876688e-06, | |
| "loss": 0.467, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.6206896551724137, | |
| "grad_norm": 0.4925503432750702, | |
| "learning_rate": 4.484063073290301e-06, | |
| "loss": 0.4651, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.6253918495297806, | |
| "grad_norm": 0.5364562273025513, | |
| "learning_rate": 4.479978802636637e-06, | |
| "loss": 0.5002, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.6300940438871474, | |
| "grad_norm": 0.5685595870018005, | |
| "learning_rate": 4.475880305262362e-06, | |
| "loss": 0.504, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.634796238244514, | |
| "grad_norm": 0.4894132912158966, | |
| "learning_rate": 4.471767610616366e-06, | |
| "loss": 0.4961, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.6394984326018809, | |
| "grad_norm": 0.5791006088256836, | |
| "learning_rate": 4.467640748249549e-06, | |
| "loss": 0.4706, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.6442006269592477, | |
| "grad_norm": 0.4814227521419525, | |
| "learning_rate": 4.4634997478146125e-06, | |
| "loss": 0.4895, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.6489028213166144, | |
| "grad_norm": 0.4842629134654999, | |
| "learning_rate": 4.459344639065842e-06, | |
| "loss": 0.482, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.6536050156739812, | |
| "grad_norm": 0.44065847992897034, | |
| "learning_rate": 4.455175451858897e-06, | |
| "loss": 0.4912, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.658307210031348, | |
| "grad_norm": 0.5212785601615906, | |
| "learning_rate": 4.450992216150592e-06, | |
| "loss": 0.4995, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.6630094043887147, | |
| "grad_norm": 0.6393409967422485, | |
| "learning_rate": 4.446794961998689e-06, | |
| "loss": 0.4675, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.6677115987460815, | |
| "grad_norm": 0.4786463975906372, | |
| "learning_rate": 4.442583719561671e-06, | |
| "loss": 0.4938, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.6724137931034484, | |
| "grad_norm": 0.5183210968971252, | |
| "learning_rate": 4.438358519098536e-06, | |
| "loss": 0.5032, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.677115987460815, | |
| "grad_norm": 0.7055342197418213, | |
| "learning_rate": 4.4341193909685685e-06, | |
| "loss": 0.4861, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.6818181818181817, | |
| "grad_norm": 0.4798656105995178, | |
| "learning_rate": 4.429866365631134e-06, | |
| "loss": 0.4917, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.6865203761755487, | |
| "grad_norm": 0.5035173296928406, | |
| "learning_rate": 4.425599473645447e-06, | |
| "loss": 0.4834, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.6912225705329154, | |
| "grad_norm": 0.5660416483879089, | |
| "learning_rate": 4.421318745670364e-06, | |
| "loss": 0.4829, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.695924764890282, | |
| "grad_norm": 0.7084828019142151, | |
| "learning_rate": 4.4170242124641524e-06, | |
| "loss": 0.4603, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.700626959247649, | |
| "grad_norm": 0.4513323903083801, | |
| "learning_rate": 4.412715904884277e-06, | |
| "loss": 0.4896, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.7053291536050157, | |
| "grad_norm": 0.48268529772758484, | |
| "learning_rate": 4.4083938538871735e-06, | |
| "loss": 0.4684, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.7100313479623823, | |
| "grad_norm": 0.4861423671245575, | |
| "learning_rate": 4.4040580905280295e-06, | |
| "loss": 0.4878, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.7147335423197492, | |
| "grad_norm": 0.48606055974960327, | |
| "learning_rate": 4.3997086459605586e-06, | |
| "loss": 0.4849, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.719435736677116, | |
| "grad_norm": 0.4912388026714325, | |
| "learning_rate": 4.395345551436779e-06, | |
| "loss": 0.509, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.7241379310344827, | |
| "grad_norm": 0.47986406087875366, | |
| "learning_rate": 4.390968838306788e-06, | |
| "loss": 0.4635, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.7288401253918495, | |
| "grad_norm": 0.5145572423934937, | |
| "learning_rate": 4.386578538018535e-06, | |
| "loss": 0.4631, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.7335423197492164, | |
| "grad_norm": 0.5538941621780396, | |
| "learning_rate": 4.382174682117598e-06, | |
| "loss": 0.5091, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.738244514106583, | |
| "grad_norm": 0.49918535351753235, | |
| "learning_rate": 4.377757302246956e-06, | |
| "loss": 0.4454, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.7429467084639498, | |
| "grad_norm": 0.48130425810813904, | |
| "learning_rate": 4.373326430146762e-06, | |
| "loss": 0.5011, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.7476489028213167, | |
| "grad_norm": 0.8833332657814026, | |
| "learning_rate": 4.368882097654113e-06, | |
| "loss": 0.4961, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.7523510971786833, | |
| "grad_norm": 0.45491841435432434, | |
| "learning_rate": 4.364424336702825e-06, | |
| "loss": 0.4708, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.7570532915360502, | |
| "grad_norm": 0.6433991193771362, | |
| "learning_rate": 4.3599531793232e-06, | |
| "loss": 0.488, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.761755485893417, | |
| "grad_norm": 0.49767547845840454, | |
| "learning_rate": 4.355468657641797e-06, | |
| "loss": 0.4836, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.7664576802507836, | |
| "grad_norm": 0.5577245950698853, | |
| "learning_rate": 4.3509708038812035e-06, | |
| "loss": 0.4879, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.7711598746081505, | |
| "grad_norm": 0.4875679314136505, | |
| "learning_rate": 4.346459650359798e-06, | |
| "loss": 0.4858, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.7758620689655173, | |
| "grad_norm": 0.47319531440734863, | |
| "learning_rate": 4.341935229491525e-06, | |
| "loss": 0.4554, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.780564263322884, | |
| "grad_norm": 0.7505528330802917, | |
| "learning_rate": 4.337397573785659e-06, | |
| "loss": 0.5043, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.7852664576802508, | |
| "grad_norm": 0.49199047684669495, | |
| "learning_rate": 4.332846715846566e-06, | |
| "loss": 0.4712, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.7899686520376177, | |
| "grad_norm": 0.5161461234092712, | |
| "learning_rate": 4.328282688373479e-06, | |
| "loss": 0.4927, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.7946708463949843, | |
| "grad_norm": 0.5262524485588074, | |
| "learning_rate": 4.323705524160258e-06, | |
| "loss": 0.4904, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.799373040752351, | |
| "grad_norm": 0.5441488027572632, | |
| "learning_rate": 4.319115256095149e-06, | |
| "loss": 0.4662, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.804075235109718, | |
| "grad_norm": 0.6795123815536499, | |
| "learning_rate": 4.314511917160557e-06, | |
| "loss": 0.4681, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.8087774294670846, | |
| "grad_norm": 0.47761398553848267, | |
| "learning_rate": 4.3098955404328045e-06, | |
| "loss": 0.4611, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.8134796238244513, | |
| "grad_norm": 0.7907284498214722, | |
| "learning_rate": 4.305266159081895e-06, | |
| "loss": 0.4826, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.8181818181818183, | |
| "grad_norm": 0.52130126953125, | |
| "learning_rate": 4.3006238063712725e-06, | |
| "loss": 0.4671, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.822884012539185, | |
| "grad_norm": 0.8447866439819336, | |
| "learning_rate": 4.295968515657583e-06, | |
| "loss": 0.5014, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.8275862068965516, | |
| "grad_norm": 0.4711958169937134, | |
| "learning_rate": 4.29130032039044e-06, | |
| "loss": 0.4844, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.8322884012539185, | |
| "grad_norm": 0.5012089610099792, | |
| "learning_rate": 4.2866192541121755e-06, | |
| "loss": 0.4747, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.8369905956112853, | |
| "grad_norm": 0.5260939598083496, | |
| "learning_rate": 4.281925350457606e-06, | |
| "loss": 0.4778, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.841692789968652, | |
| "grad_norm": 0.4744386672973633, | |
| "learning_rate": 4.277218643153787e-06, | |
| "loss": 0.4795, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.8463949843260188, | |
| "grad_norm": 0.4848107695579529, | |
| "learning_rate": 4.272499166019771e-06, | |
| "loss": 0.4783, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.8510971786833856, | |
| "grad_norm": 0.48980602622032166, | |
| "learning_rate": 4.267766952966369e-06, | |
| "loss": 0.4665, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.8557993730407523, | |
| "grad_norm": 0.49879398941993713, | |
| "learning_rate": 4.2630220379959006e-06, | |
| "loss": 0.444, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.8605015673981191, | |
| "grad_norm": 0.5382682681083679, | |
| "learning_rate": 4.258264455201953e-06, | |
| "loss": 0.466, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.865203761755486, | |
| "grad_norm": 0.47293514013290405, | |
| "learning_rate": 4.2534942387691335e-06, | |
| "loss": 0.4915, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.8699059561128526, | |
| "grad_norm": 0.4993828237056732, | |
| "learning_rate": 4.248711422972829e-06, | |
| "loss": 0.4771, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.8746081504702194, | |
| "grad_norm": 0.4950089156627655, | |
| "learning_rate": 4.243916042178954e-06, | |
| "loss": 0.4614, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.8793103448275863, | |
| "grad_norm": 0.48008930683135986, | |
| "learning_rate": 4.239108130843709e-06, | |
| "loss": 0.4684, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.884012539184953, | |
| "grad_norm": 0.5422408580780029, | |
| "learning_rate": 4.234287723513326e-06, | |
| "loss": 0.4933, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.8887147335423198, | |
| "grad_norm": 0.5405403971672058, | |
| "learning_rate": 4.229454854823827e-06, | |
| "loss": 0.4912, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.8934169278996866, | |
| "grad_norm": 0.6024321913719177, | |
| "learning_rate": 4.224609559500772e-06, | |
| "loss": 0.504, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.8981191222570533, | |
| "grad_norm": 0.4791870415210724, | |
| "learning_rate": 4.21975187235901e-06, | |
| "loss": 0.4547, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.90282131661442, | |
| "grad_norm": 0.9783840775489807, | |
| "learning_rate": 4.21488182830243e-06, | |
| "loss": 0.483, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.907523510971787, | |
| "grad_norm": 1.0899996757507324, | |
| "learning_rate": 4.209999462323706e-06, | |
| "loss": 0.4603, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.9122257053291536, | |
| "grad_norm": 0.6221780180931091, | |
| "learning_rate": 4.20510480950405e-06, | |
| "loss": 0.4926, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.9169278996865202, | |
| "grad_norm": 0.671617329120636, | |
| "learning_rate": 4.200197905012961e-06, | |
| "loss": 0.4536, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.9216300940438873, | |
| "grad_norm": 0.48824113607406616, | |
| "learning_rate": 4.195278784107965e-06, | |
| "loss": 0.4709, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.926332288401254, | |
| "grad_norm": 0.4700049161911011, | |
| "learning_rate": 4.19034748213437e-06, | |
| "loss": 0.4721, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.9310344827586206, | |
| "grad_norm": 0.5228457450866699, | |
| "learning_rate": 4.185404034525008e-06, | |
| "loss": 0.4644, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.9357366771159876, | |
| "grad_norm": 0.5192087888717651, | |
| "learning_rate": 4.180448476799981e-06, | |
| "loss": 0.5018, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.9404388714733543, | |
| "grad_norm": 0.49069511890411377, | |
| "learning_rate": 4.175480844566404e-06, | |
| "loss": 0.4747, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.9451410658307209, | |
| "grad_norm": 0.45442187786102295, | |
| "learning_rate": 4.170501173518152e-06, | |
| "loss": 0.4683, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.9498432601880877, | |
| "grad_norm": 0.5145336985588074, | |
| "learning_rate": 4.165509499435604e-06, | |
| "loss": 0.4677, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.9545454545454546, | |
| "grad_norm": 0.514231264591217, | |
| "learning_rate": 4.16050585818538e-06, | |
| "loss": 0.4812, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.9592476489028212, | |
| "grad_norm": 0.5052164793014526, | |
| "learning_rate": 4.155490285720092e-06, | |
| "loss": 0.5056, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.963949843260188, | |
| "grad_norm": 0.5336954593658447, | |
| "learning_rate": 4.150462818078079e-06, | |
| "loss": 0.4916, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.968652037617555, | |
| "grad_norm": 0.5100321769714355, | |
| "learning_rate": 4.145423491383153e-06, | |
| "loss": 0.4869, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.9733542319749215, | |
| "grad_norm": 0.5358601212501526, | |
| "learning_rate": 4.14037234184433e-06, | |
| "loss": 0.5041, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.9780564263322884, | |
| "grad_norm": 0.4950590133666992, | |
| "learning_rate": 4.135309405755583e-06, | |
| "loss": 0.4889, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.9827586206896552, | |
| "grad_norm": 0.5121294856071472, | |
| "learning_rate": 4.130234719495574e-06, | |
| "loss": 0.4736, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.9874608150470219, | |
| "grad_norm": 0.5014020800590515, | |
| "learning_rate": 4.125148319527391e-06, | |
| "loss": 0.4635, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.9921630094043887, | |
| "grad_norm": 0.5898067355155945, | |
| "learning_rate": 4.1200502423982904e-06, | |
| "loss": 0.4842, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.9968652037617556, | |
| "grad_norm": 0.535094141960144, | |
| "learning_rate": 4.1149405247394295e-06, | |
| "loss": 0.4885, | |
| "step": 424 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1272, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 6, | |
| "save_steps": 212, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.7298140993196392e+19, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |