| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.4007212983370066, |
| "eval_steps": 500, |
| "global_step": 2500, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0016028851933480265, |
| "grad_norm": 678.3370361328125, |
| "learning_rate": 9e-06, |
| "loss": 9.3792, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.003205770386696053, |
| "grad_norm": 113.01063537597656, |
| "learning_rate": 1.9e-05, |
| "loss": 6.1237, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.004808655580044079, |
| "grad_norm": 38.7303352355957, |
| "learning_rate": 2.9000000000000004e-05, |
| "loss": 3.9919, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.006411540773392106, |
| "grad_norm": 11.953008651733398, |
| "learning_rate": 3.9e-05, |
| "loss": 3.2455, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.008014425966740133, |
| "grad_norm": 7.101607799530029, |
| "learning_rate": 4.9000000000000005e-05, |
| "loss": 2.8384, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.009617311160088158, |
| "grad_norm": 3.37833571434021, |
| "learning_rate": 5.9e-05, |
| "loss": 2.5245, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.011220196353436185, |
| "grad_norm": 6.2443060874938965, |
| "learning_rate": 6.900000000000001e-05, |
| "loss": 2.3123, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.012823081546784212, |
| "grad_norm": 3.346142530441284, |
| "learning_rate": 7.9e-05, |
| "loss": 2.2146, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.014425966740132239, |
| "grad_norm": 5.297983646392822, |
| "learning_rate": 8.9e-05, |
| "loss": 2.1299, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.016028851933480266, |
| "grad_norm": 3.8509230613708496, |
| "learning_rate": 9.900000000000001e-05, |
| "loss": 2.0268, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.017631737126828292, |
| "grad_norm": 3.9156150817871094, |
| "learning_rate": 0.000109, |
| "loss": 1.9719, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.019234622320176316, |
| "grad_norm": 2.475393056869507, |
| "learning_rate": 0.00011899999999999999, |
| "loss": 1.9239, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.020837507513524343, |
| "grad_norm": 1.9771146774291992, |
| "learning_rate": 0.00012900000000000002, |
| "loss": 1.9036, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.02244039270687237, |
| "grad_norm": 2.066272020339966, |
| "learning_rate": 0.00013900000000000002, |
| "loss": 1.8848, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.024043277900220397, |
| "grad_norm": 3.157252311706543, |
| "learning_rate": 0.000149, |
| "loss": 1.8243, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.025646163093568423, |
| "grad_norm": 1.3372395038604736, |
| "learning_rate": 0.00015900000000000002, |
| "loss": 1.7958, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.02724904828691645, |
| "grad_norm": 1.4542139768600464, |
| "learning_rate": 0.00016900000000000002, |
| "loss": 1.7728, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.028851933480264477, |
| "grad_norm": 1.0300779342651367, |
| "learning_rate": 0.000179, |
| "loss": 1.7621, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.030454818673612504, |
| "grad_norm": 1.7812498807907104, |
| "learning_rate": 0.000189, |
| "loss": 1.7172, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.03205770386696053, |
| "grad_norm": 2.1096291542053223, |
| "learning_rate": 0.000199, |
| "loss": 1.7046, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.033660589060308554, |
| "grad_norm": 1.7795636653900146, |
| "learning_rate": 0.00020899999999999998, |
| "loss": 1.7024, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.035263474253656585, |
| "grad_norm": 2.049923896789551, |
| "learning_rate": 0.000219, |
| "loss": 1.6741, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.03686635944700461, |
| "grad_norm": 1.7486822605133057, |
| "learning_rate": 0.000229, |
| "loss": 1.6917, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.03846924464035263, |
| "grad_norm": 0.8129467964172363, |
| "learning_rate": 0.00023899999999999998, |
| "loss": 1.611, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.04007212983370066, |
| "grad_norm": 0.798921525478363, |
| "learning_rate": 0.000249, |
| "loss": 1.6248, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.041675015027048685, |
| "grad_norm": 0.8136582374572754, |
| "learning_rate": 0.000259, |
| "loss": 1.6145, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.043277900220396716, |
| "grad_norm": 0.7742071747779846, |
| "learning_rate": 0.00026900000000000003, |
| "loss": 1.5923, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.04488078541374474, |
| "grad_norm": 1.268173098564148, |
| "learning_rate": 0.000279, |
| "loss": 1.6178, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.04648367060709277, |
| "grad_norm": 0.6928769946098328, |
| "learning_rate": 0.000289, |
| "loss": 1.5796, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.04808655580044079, |
| "grad_norm": 1.1396955251693726, |
| "learning_rate": 0.000299, |
| "loss": 1.5705, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.049689440993788817, |
| "grad_norm": 0.8991694450378418, |
| "learning_rate": 0.00030900000000000003, |
| "loss": 1.5643, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.05129232618713685, |
| "grad_norm": 0.7772015929222107, |
| "learning_rate": 0.000319, |
| "loss": 1.5568, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.05289521138048487, |
| "grad_norm": 0.5438646078109741, |
| "learning_rate": 0.00032900000000000003, |
| "loss": 1.5606, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.0544980965738329, |
| "grad_norm": 1.0528185367584229, |
| "learning_rate": 0.00033900000000000005, |
| "loss": 1.5214, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.056100981767180924, |
| "grad_norm": 0.7310357093811035, |
| "learning_rate": 0.00034899999999999997, |
| "loss": 1.522, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.057703866960528954, |
| "grad_norm": 0.7124305367469788, |
| "learning_rate": 0.000359, |
| "loss": 1.5262, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.05930675215387698, |
| "grad_norm": 0.9238534569740295, |
| "learning_rate": 0.000369, |
| "loss": 1.5308, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.06090963734722501, |
| "grad_norm": 0.6533244848251343, |
| "learning_rate": 0.000379, |
| "loss": 1.4975, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.06251252254057303, |
| "grad_norm": 0.730185329914093, |
| "learning_rate": 0.000389, |
| "loss": 1.4823, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.06411540773392106, |
| "grad_norm": 0.7585217952728271, |
| "learning_rate": 0.00039900000000000005, |
| "loss": 1.5047, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.06571829292726908, |
| "grad_norm": 1.3076744079589844, |
| "learning_rate": 0.00040899999999999997, |
| "loss": 1.4638, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.06732117812061711, |
| "grad_norm": 0.7289252877235413, |
| "learning_rate": 0.000419, |
| "loss": 1.4677, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.06892406331396514, |
| "grad_norm": 0.7431015968322754, |
| "learning_rate": 0.000429, |
| "loss": 1.4918, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.07052694850731317, |
| "grad_norm": 0.7311664819717407, |
| "learning_rate": 0.000439, |
| "loss": 1.4713, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.07212983370066119, |
| "grad_norm": 1.0821692943572998, |
| "learning_rate": 0.000449, |
| "loss": 1.4724, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.07373271889400922, |
| "grad_norm": 1.5106474161148071, |
| "learning_rate": 0.00045900000000000004, |
| "loss": 1.4655, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.07533560408735725, |
| "grad_norm": 0.5979589223861694, |
| "learning_rate": 0.00046899999999999996, |
| "loss": 1.4639, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.07693848928070526, |
| "grad_norm": 0.5786557197570801, |
| "learning_rate": 0.000479, |
| "loss": 1.4339, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.0785413744740533, |
| "grad_norm": 0.43480566143989563, |
| "learning_rate": 0.000489, |
| "loss": 1.4583, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.08014425966740132, |
| "grad_norm": 0.44123759865760803, |
| "learning_rate": 0.000499, |
| "loss": 1.4314, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.08174714486074935, |
| "grad_norm": 0.463187038898468, |
| "learning_rate": 0.0005, |
| "loss": 1.4396, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.08335003005409737, |
| "grad_norm": 1.4139055013656616, |
| "learning_rate": 0.0005, |
| "loss": 1.4437, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.0849529152474454, |
| "grad_norm": 0.3580915033817291, |
| "learning_rate": 0.0005, |
| "loss": 1.414, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.08655580044079343, |
| "grad_norm": 0.41002535820007324, |
| "learning_rate": 0.0005, |
| "loss": 1.427, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.08815868563414145, |
| "grad_norm": 0.4274766445159912, |
| "learning_rate": 0.0005, |
| "loss": 1.4381, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.08976157082748948, |
| "grad_norm": 0.46305719017982483, |
| "learning_rate": 0.0005, |
| "loss": 1.4022, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.09136445602083751, |
| "grad_norm": 0.388123482465744, |
| "learning_rate": 0.0005, |
| "loss": 1.395, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.09296734121418554, |
| "grad_norm": 0.44987061619758606, |
| "learning_rate": 0.0005, |
| "loss": 1.4009, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.09457022640753356, |
| "grad_norm": 0.3794371485710144, |
| "learning_rate": 0.0005, |
| "loss": 1.4273, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.09617311160088159, |
| "grad_norm": 0.3824334442615509, |
| "learning_rate": 0.0005, |
| "loss": 1.4076, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.09777599679422962, |
| "grad_norm": 0.40158432722091675, |
| "learning_rate": 0.0005, |
| "loss": 1.3827, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.09937888198757763, |
| "grad_norm": 0.36512696743011475, |
| "learning_rate": 0.0005, |
| "loss": 1.4304, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.10098176718092566, |
| "grad_norm": 0.5844743847846985, |
| "learning_rate": 0.0005, |
| "loss": 1.3845, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.1025846523742737, |
| "grad_norm": 0.3842383027076721, |
| "learning_rate": 0.0005, |
| "loss": 1.3851, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.10418753756762172, |
| "grad_norm": 0.34687766432762146, |
| "learning_rate": 0.0005, |
| "loss": 1.4011, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.10579042276096974, |
| "grad_norm": 0.3505879342556, |
| "learning_rate": 0.0005, |
| "loss": 1.365, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.10739330795431777, |
| "grad_norm": 0.3470805883407593, |
| "learning_rate": 0.0005, |
| "loss": 1.3882, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.1089961931476658, |
| "grad_norm": 0.35840025544166565, |
| "learning_rate": 0.0005, |
| "loss": 1.3652, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.11059907834101383, |
| "grad_norm": 0.36231186985969543, |
| "learning_rate": 0.0005, |
| "loss": 1.3841, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.11220196353436185, |
| "grad_norm": 0.3536931276321411, |
| "learning_rate": 0.0005, |
| "loss": 1.3746, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.11380484872770988, |
| "grad_norm": 0.35982581973075867, |
| "learning_rate": 0.0005, |
| "loss": 1.3875, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.11540773392105791, |
| "grad_norm": 0.4366588890552521, |
| "learning_rate": 0.0005, |
| "loss": 1.3723, |
| "step": 720 |
| }, |
| { |
| "epoch": 0.11701061911440593, |
| "grad_norm": 0.34436532855033875, |
| "learning_rate": 0.0005, |
| "loss": 1.3811, |
| "step": 730 |
| }, |
| { |
| "epoch": 0.11861350430775396, |
| "grad_norm": 0.38028818368911743, |
| "learning_rate": 0.0005, |
| "loss": 1.3894, |
| "step": 740 |
| }, |
| { |
| "epoch": 0.12021638950110199, |
| "grad_norm": 0.3495010733604431, |
| "learning_rate": 0.0005, |
| "loss": 1.3719, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.12181927469445002, |
| "grad_norm": 0.3643304109573364, |
| "learning_rate": 0.0005, |
| "loss": 1.3793, |
| "step": 760 |
| }, |
| { |
| "epoch": 0.12342215988779803, |
| "grad_norm": 0.35712549090385437, |
| "learning_rate": 0.0005, |
| "loss": 1.3459, |
| "step": 770 |
| }, |
| { |
| "epoch": 0.12502504508114606, |
| "grad_norm": 0.348839670419693, |
| "learning_rate": 0.0005, |
| "loss": 1.3794, |
| "step": 780 |
| }, |
| { |
| "epoch": 0.1266279302744941, |
| "grad_norm": 0.4368819296360016, |
| "learning_rate": 0.0005, |
| "loss": 1.3722, |
| "step": 790 |
| }, |
| { |
| "epoch": 0.12823081546784212, |
| "grad_norm": 0.5412229299545288, |
| "learning_rate": 0.0005, |
| "loss": 1.3571, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.12983370066119015, |
| "grad_norm": 0.42561405897140503, |
| "learning_rate": 0.0005, |
| "loss": 1.3667, |
| "step": 810 |
| }, |
| { |
| "epoch": 0.13143658585453816, |
| "grad_norm": 0.3597279489040375, |
| "learning_rate": 0.0005, |
| "loss": 1.3522, |
| "step": 820 |
| }, |
| { |
| "epoch": 0.1330394710478862, |
| "grad_norm": 0.4504384994506836, |
| "learning_rate": 0.0005, |
| "loss": 1.3349, |
| "step": 830 |
| }, |
| { |
| "epoch": 0.13464235624123422, |
| "grad_norm": 0.37311264872550964, |
| "learning_rate": 0.0005, |
| "loss": 1.339, |
| "step": 840 |
| }, |
| { |
| "epoch": 0.13624524143458225, |
| "grad_norm": 0.3933414816856384, |
| "learning_rate": 0.0005, |
| "loss": 1.3553, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.13784812662793028, |
| "grad_norm": 0.3546708822250366, |
| "learning_rate": 0.0005, |
| "loss": 1.3518, |
| "step": 860 |
| }, |
| { |
| "epoch": 0.1394510118212783, |
| "grad_norm": 0.4191734492778778, |
| "learning_rate": 0.0005, |
| "loss": 1.3361, |
| "step": 870 |
| }, |
| { |
| "epoch": 0.14105389701462634, |
| "grad_norm": 0.5219401717185974, |
| "learning_rate": 0.0005, |
| "loss": 1.3464, |
| "step": 880 |
| }, |
| { |
| "epoch": 0.14265678220797434, |
| "grad_norm": 0.4485955238342285, |
| "learning_rate": 0.0005, |
| "loss": 1.3602, |
| "step": 890 |
| }, |
| { |
| "epoch": 0.14425966740132237, |
| "grad_norm": 0.45198988914489746, |
| "learning_rate": 0.0005, |
| "loss": 1.3578, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.1458625525946704, |
| "grad_norm": 0.539563000202179, |
| "learning_rate": 0.0005, |
| "loss": 1.3387, |
| "step": 910 |
| }, |
| { |
| "epoch": 0.14746543778801843, |
| "grad_norm": 0.4177083671092987, |
| "learning_rate": 0.0005, |
| "loss": 1.3512, |
| "step": 920 |
| }, |
| { |
| "epoch": 0.14906832298136646, |
| "grad_norm": 0.4130881726741791, |
| "learning_rate": 0.0005, |
| "loss": 1.3481, |
| "step": 930 |
| }, |
| { |
| "epoch": 0.1506712081747145, |
| "grad_norm": 0.4453917443752289, |
| "learning_rate": 0.0005, |
| "loss": 1.371, |
| "step": 940 |
| }, |
| { |
| "epoch": 0.15227409336806252, |
| "grad_norm": 0.44530555605888367, |
| "learning_rate": 0.0005, |
| "loss": 1.3509, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.15387697856141053, |
| "grad_norm": 0.6519402265548706, |
| "learning_rate": 0.0005, |
| "loss": 1.329, |
| "step": 960 |
| }, |
| { |
| "epoch": 0.15547986375475856, |
| "grad_norm": 0.40298616886138916, |
| "learning_rate": 0.0005, |
| "loss": 1.3243, |
| "step": 970 |
| }, |
| { |
| "epoch": 0.1570827489481066, |
| "grad_norm": 0.3541094958782196, |
| "learning_rate": 0.0005, |
| "loss": 1.3278, |
| "step": 980 |
| }, |
| { |
| "epoch": 0.15868563414145462, |
| "grad_norm": 0.39235177636146545, |
| "learning_rate": 0.0005, |
| "loss": 1.3347, |
| "step": 990 |
| }, |
| { |
| "epoch": 0.16028851933480265, |
| "grad_norm": 0.3771113455295563, |
| "learning_rate": 0.0005, |
| "loss": 1.3485, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.16189140452815068, |
| "grad_norm": 0.32740509510040283, |
| "learning_rate": 0.0005, |
| "loss": 1.3404, |
| "step": 1010 |
| }, |
| { |
| "epoch": 0.1634942897214987, |
| "grad_norm": 0.3351823091506958, |
| "learning_rate": 0.0005, |
| "loss": 1.302, |
| "step": 1020 |
| }, |
| { |
| "epoch": 0.1650971749148467, |
| "grad_norm": 0.3709336519241333, |
| "learning_rate": 0.0005, |
| "loss": 1.3094, |
| "step": 1030 |
| }, |
| { |
| "epoch": 0.16670006010819474, |
| "grad_norm": 0.4003657102584839, |
| "learning_rate": 0.0005, |
| "loss": 1.3612, |
| "step": 1040 |
| }, |
| { |
| "epoch": 0.16830294530154277, |
| "grad_norm": 0.3763624429702759, |
| "learning_rate": 0.0005, |
| "loss": 1.3209, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.1699058304948908, |
| "grad_norm": 0.48239532113075256, |
| "learning_rate": 0.0005, |
| "loss": 1.3369, |
| "step": 1060 |
| }, |
| { |
| "epoch": 0.17150871568823883, |
| "grad_norm": 0.5595227479934692, |
| "learning_rate": 0.0005, |
| "loss": 1.3194, |
| "step": 1070 |
| }, |
| { |
| "epoch": 0.17311160088158686, |
| "grad_norm": 0.3660586178302765, |
| "learning_rate": 0.0005, |
| "loss": 1.3209, |
| "step": 1080 |
| }, |
| { |
| "epoch": 0.1747144860749349, |
| "grad_norm": 0.38128146529197693, |
| "learning_rate": 0.0005, |
| "loss": 1.3207, |
| "step": 1090 |
| }, |
| { |
| "epoch": 0.1763173712682829, |
| "grad_norm": 0.4592175781726837, |
| "learning_rate": 0.0005, |
| "loss": 1.3105, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.17792025646163093, |
| "grad_norm": 0.35791411995887756, |
| "learning_rate": 0.0005, |
| "loss": 1.3121, |
| "step": 1110 |
| }, |
| { |
| "epoch": 0.17952314165497896, |
| "grad_norm": 0.33707818388938904, |
| "learning_rate": 0.0005, |
| "loss": 1.3123, |
| "step": 1120 |
| }, |
| { |
| "epoch": 0.181126026848327, |
| "grad_norm": 0.37126076221466064, |
| "learning_rate": 0.0005, |
| "loss": 1.3117, |
| "step": 1130 |
| }, |
| { |
| "epoch": 0.18272891204167502, |
| "grad_norm": 0.3038891553878784, |
| "learning_rate": 0.0005, |
| "loss": 1.3189, |
| "step": 1140 |
| }, |
| { |
| "epoch": 0.18433179723502305, |
| "grad_norm": 0.31843164563179016, |
| "learning_rate": 0.0005, |
| "loss": 1.302, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.18593468242837108, |
| "grad_norm": 0.3236716687679291, |
| "learning_rate": 0.0005, |
| "loss": 1.3137, |
| "step": 1160 |
| }, |
| { |
| "epoch": 0.18753756762171908, |
| "grad_norm": 0.33301737904548645, |
| "learning_rate": 0.0005, |
| "loss": 1.3194, |
| "step": 1170 |
| }, |
| { |
| "epoch": 0.1891404528150671, |
| "grad_norm": 0.3310806453227997, |
| "learning_rate": 0.0005, |
| "loss": 1.305, |
| "step": 1180 |
| }, |
| { |
| "epoch": 0.19074333800841514, |
| "grad_norm": 0.32767626643180847, |
| "learning_rate": 0.0005, |
| "loss": 1.314, |
| "step": 1190 |
| }, |
| { |
| "epoch": 0.19234622320176317, |
| "grad_norm": 0.3089030385017395, |
| "learning_rate": 0.0005, |
| "loss": 1.2897, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.1939491083951112, |
| "grad_norm": 0.33364149928092957, |
| "learning_rate": 0.0005, |
| "loss": 1.3003, |
| "step": 1210 |
| }, |
| { |
| "epoch": 0.19555199358845923, |
| "grad_norm": 0.29430854320526123, |
| "learning_rate": 0.0005, |
| "loss": 1.3135, |
| "step": 1220 |
| }, |
| { |
| "epoch": 0.19715487878180726, |
| "grad_norm": 0.40663015842437744, |
| "learning_rate": 0.0005, |
| "loss": 1.2946, |
| "step": 1230 |
| }, |
| { |
| "epoch": 0.19875776397515527, |
| "grad_norm": 0.3090682923793793, |
| "learning_rate": 0.0005, |
| "loss": 1.3055, |
| "step": 1240 |
| }, |
| { |
| "epoch": 0.2003606491685033, |
| "grad_norm": 0.33319520950317383, |
| "learning_rate": 0.0005, |
| "loss": 1.3028, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.20196353436185133, |
| "grad_norm": 0.39136168360710144, |
| "learning_rate": 0.0005, |
| "loss": 1.3126, |
| "step": 1260 |
| }, |
| { |
| "epoch": 0.20356641955519936, |
| "grad_norm": 0.9555182456970215, |
| "learning_rate": 0.0005, |
| "loss": 1.2978, |
| "step": 1270 |
| }, |
| { |
| "epoch": 0.2051693047485474, |
| "grad_norm": 0.36503541469573975, |
| "learning_rate": 0.0005, |
| "loss": 1.3118, |
| "step": 1280 |
| }, |
| { |
| "epoch": 0.20677218994189542, |
| "grad_norm": 0.4477512240409851, |
| "learning_rate": 0.0005, |
| "loss": 1.3162, |
| "step": 1290 |
| }, |
| { |
| "epoch": 0.20837507513524345, |
| "grad_norm": 0.3497491478919983, |
| "learning_rate": 0.0005, |
| "loss": 1.2952, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.20997796032859145, |
| "grad_norm": 0.44401678442955017, |
| "learning_rate": 0.0005, |
| "loss": 1.2727, |
| "step": 1310 |
| }, |
| { |
| "epoch": 0.21158084552193948, |
| "grad_norm": 0.3597167730331421, |
| "learning_rate": 0.0005, |
| "loss": 1.3, |
| "step": 1320 |
| }, |
| { |
| "epoch": 0.2131837307152875, |
| "grad_norm": 0.3729954957962036, |
| "learning_rate": 0.0005, |
| "loss": 1.3211, |
| "step": 1330 |
| }, |
| { |
| "epoch": 0.21478661590863554, |
| "grad_norm": 0.30662021040916443, |
| "learning_rate": 0.0005, |
| "loss": 1.2768, |
| "step": 1340 |
| }, |
| { |
| "epoch": 0.21638950110198357, |
| "grad_norm": 0.31561264395713806, |
| "learning_rate": 0.0005, |
| "loss": 1.3191, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.2179923862953316, |
| "grad_norm": 0.4170988202095032, |
| "learning_rate": 0.0005, |
| "loss": 1.3091, |
| "step": 1360 |
| }, |
| { |
| "epoch": 0.21959527148867963, |
| "grad_norm": 0.35628414154052734, |
| "learning_rate": 0.0005, |
| "loss": 1.305, |
| "step": 1370 |
| }, |
| { |
| "epoch": 0.22119815668202766, |
| "grad_norm": 0.3731990456581116, |
| "learning_rate": 0.0005, |
| "loss": 1.3092, |
| "step": 1380 |
| }, |
| { |
| "epoch": 0.22280104187537567, |
| "grad_norm": 0.3259612023830414, |
| "learning_rate": 0.0005, |
| "loss": 1.293, |
| "step": 1390 |
| }, |
| { |
| "epoch": 0.2244039270687237, |
| "grad_norm": 0.320374995470047, |
| "learning_rate": 0.0005, |
| "loss": 1.304, |
| "step": 1400 |
| }, |
| { |
| "epoch": 0.22600681226207173, |
| "grad_norm": 0.4857123792171478, |
| "learning_rate": 0.0005, |
| "loss": 1.2878, |
| "step": 1410 |
| }, |
| { |
| "epoch": 0.22760969745541976, |
| "grad_norm": 1.945692539215088, |
| "learning_rate": 0.0005, |
| "loss": 1.2989, |
| "step": 1420 |
| }, |
| { |
| "epoch": 0.2292125826487678, |
| "grad_norm": 0.39001426100730896, |
| "learning_rate": 0.0005, |
| "loss": 1.2989, |
| "step": 1430 |
| }, |
| { |
| "epoch": 0.23081546784211582, |
| "grad_norm": 0.3088345229625702, |
| "learning_rate": 0.0005, |
| "loss": 1.3036, |
| "step": 1440 |
| }, |
| { |
| "epoch": 0.23241835303546385, |
| "grad_norm": 0.368073970079422, |
| "learning_rate": 0.0005, |
| "loss": 1.3009, |
| "step": 1450 |
| }, |
| { |
| "epoch": 0.23402123822881185, |
| "grad_norm": 0.37869542837142944, |
| "learning_rate": 0.0005, |
| "loss": 1.2899, |
| "step": 1460 |
| }, |
| { |
| "epoch": 0.23562412342215988, |
| "grad_norm": 0.393318235874176, |
| "learning_rate": 0.0005, |
| "loss": 1.2763, |
| "step": 1470 |
| }, |
| { |
| "epoch": 0.2372270086155079, |
| "grad_norm": 0.3757217526435852, |
| "learning_rate": 0.0005, |
| "loss": 1.2987, |
| "step": 1480 |
| }, |
| { |
| "epoch": 0.23882989380885594, |
| "grad_norm": 0.39952972531318665, |
| "learning_rate": 0.0005, |
| "loss": 1.3131, |
| "step": 1490 |
| }, |
| { |
| "epoch": 0.24043277900220397, |
| "grad_norm": 0.4175151288509369, |
| "learning_rate": 0.0005, |
| "loss": 1.2952, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.242035664195552, |
| "grad_norm": 0.4070395529270172, |
| "learning_rate": 0.0005, |
| "loss": 1.2985, |
| "step": 1510 |
| }, |
| { |
| "epoch": 0.24363854938890003, |
| "grad_norm": 0.5496709942817688, |
| "learning_rate": 0.0005, |
| "loss": 1.2822, |
| "step": 1520 |
| }, |
| { |
| "epoch": 0.24524143458224804, |
| "grad_norm": 0.3813125193119049, |
| "learning_rate": 0.0005, |
| "loss": 1.2881, |
| "step": 1530 |
| }, |
| { |
| "epoch": 0.24684431977559607, |
| "grad_norm": 0.41737911105155945, |
| "learning_rate": 0.0005, |
| "loss": 1.3007, |
| "step": 1540 |
| }, |
| { |
| "epoch": 0.2484472049689441, |
| "grad_norm": 0.3591451644897461, |
| "learning_rate": 0.0005, |
| "loss": 1.2753, |
| "step": 1550 |
| }, |
| { |
| "epoch": 0.2500500901622921, |
| "grad_norm": 0.3714822828769684, |
| "learning_rate": 0.0005, |
| "loss": 1.2949, |
| "step": 1560 |
| }, |
| { |
| "epoch": 0.25165297535564013, |
| "grad_norm": 0.4229411482810974, |
| "learning_rate": 0.0005, |
| "loss": 1.2667, |
| "step": 1570 |
| }, |
| { |
| "epoch": 0.2532558605489882, |
| "grad_norm": 0.36146730184555054, |
| "learning_rate": 0.0005, |
| "loss": 1.297, |
| "step": 1580 |
| }, |
| { |
| "epoch": 0.2548587457423362, |
| "grad_norm": 0.40604183077812195, |
| "learning_rate": 0.0005, |
| "loss": 1.2979, |
| "step": 1590 |
| }, |
| { |
| "epoch": 0.25646163093568425, |
| "grad_norm": 0.3981959819793701, |
| "learning_rate": 0.0005, |
| "loss": 1.2731, |
| "step": 1600 |
| }, |
| { |
| "epoch": 0.25806451612903225, |
| "grad_norm": 0.3921387493610382, |
| "learning_rate": 0.0005, |
| "loss": 1.2985, |
| "step": 1610 |
| }, |
| { |
| "epoch": 0.2596674013223803, |
| "grad_norm": 0.3937159776687622, |
| "learning_rate": 0.0005, |
| "loss": 1.2894, |
| "step": 1620 |
| }, |
| { |
| "epoch": 0.2612702865157283, |
| "grad_norm": 0.687377393245697, |
| "learning_rate": 0.0005, |
| "loss": 1.2857, |
| "step": 1630 |
| }, |
| { |
| "epoch": 0.2628731717090763, |
| "grad_norm": 0.35336875915527344, |
| "learning_rate": 0.0005, |
| "loss": 1.2991, |
| "step": 1640 |
| }, |
| { |
| "epoch": 0.2644760569024244, |
| "grad_norm": 0.39487507939338684, |
| "learning_rate": 0.0005, |
| "loss": 1.2848, |
| "step": 1650 |
| }, |
| { |
| "epoch": 0.2660789420957724, |
| "grad_norm": 0.33148014545440674, |
| "learning_rate": 0.0005, |
| "loss": 1.296, |
| "step": 1660 |
| }, |
| { |
| "epoch": 0.26768182728912043, |
| "grad_norm": 0.30253589153289795, |
| "learning_rate": 0.0005, |
| "loss": 1.2882, |
| "step": 1670 |
| }, |
| { |
| "epoch": 0.26928471248246844, |
| "grad_norm": 0.33182498812675476, |
| "learning_rate": 0.0005, |
| "loss": 1.2624, |
| "step": 1680 |
| }, |
| { |
| "epoch": 0.2708875976758165, |
| "grad_norm": 0.3482310473918915, |
| "learning_rate": 0.0005, |
| "loss": 1.2969, |
| "step": 1690 |
| }, |
| { |
| "epoch": 0.2724904828691645, |
| "grad_norm": 0.3354679048061371, |
| "learning_rate": 0.0005, |
| "loss": 1.2818, |
| "step": 1700 |
| }, |
| { |
| "epoch": 0.2740933680625125, |
| "grad_norm": 0.36453723907470703, |
| "learning_rate": 0.0005, |
| "loss": 1.2872, |
| "step": 1710 |
| }, |
| { |
| "epoch": 0.27569625325586056, |
| "grad_norm": 0.47449424862861633, |
| "learning_rate": 0.0005, |
| "loss": 1.2834, |
| "step": 1720 |
| }, |
| { |
| "epoch": 0.27729913844920856, |
| "grad_norm": 0.32657721638679504, |
| "learning_rate": 0.0005, |
| "loss": 1.2823, |
| "step": 1730 |
| }, |
| { |
| "epoch": 0.2789020236425566, |
| "grad_norm": 0.2980170249938965, |
| "learning_rate": 0.0005, |
| "loss": 1.2806, |
| "step": 1740 |
| }, |
| { |
| "epoch": 0.2805049088359046, |
| "grad_norm": 0.3514454662799835, |
| "learning_rate": 0.0005, |
| "loss": 1.2589, |
| "step": 1750 |
| }, |
| { |
| "epoch": 0.2821077940292527, |
| "grad_norm": 0.34434592723846436, |
| "learning_rate": 0.0005, |
| "loss": 1.295, |
| "step": 1760 |
| }, |
| { |
| "epoch": 0.2837106792226007, |
| "grad_norm": 0.2812644839286804, |
| "learning_rate": 0.0005, |
| "loss": 1.2658, |
| "step": 1770 |
| }, |
| { |
| "epoch": 0.2853135644159487, |
| "grad_norm": 0.3048274517059326, |
| "learning_rate": 0.0005, |
| "loss": 1.281, |
| "step": 1780 |
| }, |
| { |
| "epoch": 0.28691644960929674, |
| "grad_norm": 0.3719802796840668, |
| "learning_rate": 0.0005, |
| "loss": 1.2642, |
| "step": 1790 |
| }, |
| { |
| "epoch": 0.28851933480264474, |
| "grad_norm": 0.40966910123825073, |
| "learning_rate": 0.0005, |
| "loss": 1.2675, |
| "step": 1800 |
| }, |
| { |
| "epoch": 0.2901222199959928, |
| "grad_norm": 0.3720247447490692, |
| "learning_rate": 0.0005, |
| "loss": 1.2798, |
| "step": 1810 |
| }, |
| { |
| "epoch": 0.2917251051893408, |
| "grad_norm": 0.3861991763114929, |
| "learning_rate": 0.0005, |
| "loss": 1.2718, |
| "step": 1820 |
| }, |
| { |
| "epoch": 0.29332799038268886, |
| "grad_norm": 0.4071897864341736, |
| "learning_rate": 0.0005, |
| "loss": 1.288, |
| "step": 1830 |
| }, |
| { |
| "epoch": 0.29493087557603687, |
| "grad_norm": 0.31495898962020874, |
| "learning_rate": 0.0005, |
| "loss": 1.3065, |
| "step": 1840 |
| }, |
| { |
| "epoch": 0.29653376076938487, |
| "grad_norm": 0.35031652450561523, |
| "learning_rate": 0.0005, |
| "loss": 1.2732, |
| "step": 1850 |
| }, |
| { |
| "epoch": 0.2981366459627329, |
| "grad_norm": 0.38266849517822266, |
| "learning_rate": 0.0005, |
| "loss": 1.2779, |
| "step": 1860 |
| }, |
| { |
| "epoch": 0.29973953115608093, |
| "grad_norm": 0.31431642174720764, |
| "learning_rate": 0.0005, |
| "loss": 1.2761, |
| "step": 1870 |
| }, |
| { |
| "epoch": 0.301342416349429, |
| "grad_norm": 0.318367600440979, |
| "learning_rate": 0.0005, |
| "loss": 1.2688, |
| "step": 1880 |
| }, |
| { |
| "epoch": 0.302945301542777, |
| "grad_norm": 0.3060268759727478, |
| "learning_rate": 0.0005, |
| "loss": 1.2619, |
| "step": 1890 |
| }, |
| { |
| "epoch": 0.30454818673612505, |
| "grad_norm": 0.33884698152542114, |
| "learning_rate": 0.0005, |
| "loss": 1.2748, |
| "step": 1900 |
| }, |
| { |
| "epoch": 0.30615107192947305, |
| "grad_norm": 0.3074434995651245, |
| "learning_rate": 0.0005, |
| "loss": 1.2694, |
| "step": 1910 |
| }, |
| { |
| "epoch": 0.30775395712282105, |
| "grad_norm": 0.35602208971977234, |
| "learning_rate": 0.0005, |
| "loss": 1.2766, |
| "step": 1920 |
| }, |
| { |
| "epoch": 0.3093568423161691, |
| "grad_norm": 0.31432512402534485, |
| "learning_rate": 0.0005, |
| "loss": 1.2813, |
| "step": 1930 |
| }, |
| { |
| "epoch": 0.3109597275095171, |
| "grad_norm": 0.3290809094905853, |
| "learning_rate": 0.0005, |
| "loss": 1.2798, |
| "step": 1940 |
| }, |
| { |
| "epoch": 0.3125626127028652, |
| "grad_norm": 0.2725956439971924, |
| "learning_rate": 0.0005, |
| "loss": 1.2654, |
| "step": 1950 |
| }, |
| { |
| "epoch": 0.3141654978962132, |
| "grad_norm": 0.27984774112701416, |
| "learning_rate": 0.0005, |
| "loss": 1.2584, |
| "step": 1960 |
| }, |
| { |
| "epoch": 0.31576838308956123, |
| "grad_norm": 0.5261242389678955, |
| "learning_rate": 0.0005, |
| "loss": 1.2441, |
| "step": 1970 |
| }, |
| { |
| "epoch": 0.31737126828290924, |
| "grad_norm": 0.36167559027671814, |
| "learning_rate": 0.0005, |
| "loss": 1.2547, |
| "step": 1980 |
| }, |
| { |
| "epoch": 0.31897415347625724, |
| "grad_norm": 0.39349183440208435, |
| "learning_rate": 0.0005, |
| "loss": 1.2899, |
| "step": 1990 |
| }, |
| { |
| "epoch": 0.3205770386696053, |
| "grad_norm": 0.32107916474342346, |
| "learning_rate": 0.0005, |
| "loss": 1.2536, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.3221799238629533, |
| "grad_norm": 0.3739472031593323, |
| "learning_rate": 0.0005, |
| "loss": 1.2951, |
| "step": 2010 |
| }, |
| { |
| "epoch": 0.32378280905630136, |
| "grad_norm": 0.38725772500038147, |
| "learning_rate": 0.0005, |
| "loss": 1.2766, |
| "step": 2020 |
| }, |
| { |
| "epoch": 0.32538569424964936, |
| "grad_norm": 0.2980344593524933, |
| "learning_rate": 0.0005, |
| "loss": 1.2623, |
| "step": 2030 |
| }, |
| { |
| "epoch": 0.3269885794429974, |
| "grad_norm": 0.31839045882225037, |
| "learning_rate": 0.0005, |
| "loss": 1.2631, |
| "step": 2040 |
| }, |
| { |
| "epoch": 0.3285914646363454, |
| "grad_norm": 0.34590938687324524, |
| "learning_rate": 0.0005, |
| "loss": 1.2874, |
| "step": 2050 |
| }, |
| { |
| "epoch": 0.3301943498296934, |
| "grad_norm": 0.2809009253978729, |
| "learning_rate": 0.0005, |
| "loss": 1.2801, |
| "step": 2060 |
| }, |
| { |
| "epoch": 0.3317972350230415, |
| "grad_norm": 0.5641685724258423, |
| "learning_rate": 0.0005, |
| "loss": 1.2658, |
| "step": 2070 |
| }, |
| { |
| "epoch": 0.3334001202163895, |
| "grad_norm": 0.34127888083457947, |
| "learning_rate": 0.0005, |
| "loss": 1.2424, |
| "step": 2080 |
| }, |
| { |
| "epoch": 0.33500300540973754, |
| "grad_norm": 0.5770083069801331, |
| "learning_rate": 0.0005, |
| "loss": 1.2656, |
| "step": 2090 |
| }, |
| { |
| "epoch": 0.33660589060308554, |
| "grad_norm": 0.35034167766571045, |
| "learning_rate": 0.0005, |
| "loss": 1.2709, |
| "step": 2100 |
| }, |
| { |
| "epoch": 0.3382087757964336, |
| "grad_norm": 0.4296245574951172, |
| "learning_rate": 0.0005, |
| "loss": 1.2739, |
| "step": 2110 |
| }, |
| { |
| "epoch": 0.3398116609897816, |
| "grad_norm": 0.37714719772338867, |
| "learning_rate": 0.0005, |
| "loss": 1.2685, |
| "step": 2120 |
| }, |
| { |
| "epoch": 0.3414145461831296, |
| "grad_norm": 0.3473512828350067, |
| "learning_rate": 0.0005, |
| "loss": 1.2755, |
| "step": 2130 |
| }, |
| { |
| "epoch": 0.34301743137647767, |
| "grad_norm": 0.8088376522064209, |
| "learning_rate": 0.0005, |
| "loss": 1.2662, |
| "step": 2140 |
| }, |
| { |
| "epoch": 0.34462031656982567, |
| "grad_norm": 0.322681725025177, |
| "learning_rate": 0.0005, |
| "loss": 1.2548, |
| "step": 2150 |
| }, |
| { |
| "epoch": 0.3462232017631737, |
| "grad_norm": 0.42762812972068787, |
| "learning_rate": 0.0005, |
| "loss": 1.2672, |
| "step": 2160 |
| }, |
| { |
| "epoch": 0.34782608695652173, |
| "grad_norm": 0.3658334016799927, |
| "learning_rate": 0.0005, |
| "loss": 1.2565, |
| "step": 2170 |
| }, |
| { |
| "epoch": 0.3494289721498698, |
| "grad_norm": 0.37944141030311584, |
| "learning_rate": 0.0005, |
| "loss": 1.2807, |
| "step": 2180 |
| }, |
| { |
| "epoch": 0.3510318573432178, |
| "grad_norm": 0.418755441904068, |
| "learning_rate": 0.0005, |
| "loss": 1.251, |
| "step": 2190 |
| }, |
| { |
| "epoch": 0.3526347425365658, |
| "grad_norm": 0.3580203950405121, |
| "learning_rate": 0.0005, |
| "loss": 1.2621, |
| "step": 2200 |
| }, |
| { |
| "epoch": 0.35423762772991385, |
| "grad_norm": 0.39281541109085083, |
| "learning_rate": 0.0005, |
| "loss": 1.2638, |
| "step": 2210 |
| }, |
| { |
| "epoch": 0.35584051292326185, |
| "grad_norm": 0.32786354422569275, |
| "learning_rate": 0.0005, |
| "loss": 1.2498, |
| "step": 2220 |
| }, |
| { |
| "epoch": 0.3574433981166099, |
| "grad_norm": 0.3115231692790985, |
| "learning_rate": 0.0005, |
| "loss": 1.2896, |
| "step": 2230 |
| }, |
| { |
| "epoch": 0.3590462833099579, |
| "grad_norm": 0.41042885184288025, |
| "learning_rate": 0.0005, |
| "loss": 1.2748, |
| "step": 2240 |
| }, |
| { |
| "epoch": 0.360649168503306, |
| "grad_norm": 0.4438491463661194, |
| "learning_rate": 0.0005, |
| "loss": 1.2476, |
| "step": 2250 |
| }, |
| { |
| "epoch": 0.362252053696654, |
| "grad_norm": 0.3606395721435547, |
| "learning_rate": 0.0005, |
| "loss": 1.253, |
| "step": 2260 |
| }, |
| { |
| "epoch": 0.363854938890002, |
| "grad_norm": 0.3120422959327698, |
| "learning_rate": 0.0005, |
| "loss": 1.2653, |
| "step": 2270 |
| }, |
| { |
| "epoch": 0.36545782408335004, |
| "grad_norm": 0.35044899582862854, |
| "learning_rate": 0.0005, |
| "loss": 1.2663, |
| "step": 2280 |
| }, |
| { |
| "epoch": 0.36706070927669804, |
| "grad_norm": 0.3766014277935028, |
| "learning_rate": 0.0005, |
| "loss": 1.2706, |
| "step": 2290 |
| }, |
| { |
| "epoch": 0.3686635944700461, |
| "grad_norm": 0.34974873065948486, |
| "learning_rate": 0.0005, |
| "loss": 1.2598, |
| "step": 2300 |
| }, |
| { |
| "epoch": 0.3702664796633941, |
| "grad_norm": 0.43365639448165894, |
| "learning_rate": 0.0005, |
| "loss": 1.273, |
| "step": 2310 |
| }, |
| { |
| "epoch": 0.37186936485674216, |
| "grad_norm": 0.3572383522987366, |
| "learning_rate": 0.0005, |
| "loss": 1.248, |
| "step": 2320 |
| }, |
| { |
| "epoch": 0.37347225005009016, |
| "grad_norm": 0.4122643768787384, |
| "learning_rate": 0.0005, |
| "loss": 1.2399, |
| "step": 2330 |
| }, |
| { |
| "epoch": 0.37507513524343816, |
| "grad_norm": 0.3526794910430908, |
| "learning_rate": 0.0005, |
| "loss": 1.278, |
| "step": 2340 |
| }, |
| { |
| "epoch": 0.3766780204367862, |
| "grad_norm": 0.29494354128837585, |
| "learning_rate": 0.0005, |
| "loss": 1.2503, |
| "step": 2350 |
| }, |
| { |
| "epoch": 0.3782809056301342, |
| "grad_norm": 0.5473882555961609, |
| "learning_rate": 0.0005, |
| "loss": 1.2286, |
| "step": 2360 |
| }, |
| { |
| "epoch": 0.3798837908234823, |
| "grad_norm": 0.33589014410972595, |
| "learning_rate": 0.0005, |
| "loss": 1.2863, |
| "step": 2370 |
| }, |
| { |
| "epoch": 0.3814866760168303, |
| "grad_norm": 0.30862322449684143, |
| "learning_rate": 0.0005, |
| "loss": 1.2577, |
| "step": 2380 |
| }, |
| { |
| "epoch": 0.38308956121017834, |
| "grad_norm": 0.7398898005485535, |
| "learning_rate": 0.0005, |
| "loss": 1.2732, |
| "step": 2390 |
| }, |
| { |
| "epoch": 0.38469244640352634, |
| "grad_norm": 0.32832425832748413, |
| "learning_rate": 0.0005, |
| "loss": 1.2687, |
| "step": 2400 |
| }, |
| { |
| "epoch": 0.38629533159687435, |
| "grad_norm": 0.30570557713508606, |
| "learning_rate": 0.0005, |
| "loss": 1.2775, |
| "step": 2410 |
| }, |
| { |
| "epoch": 0.3878982167902224, |
| "grad_norm": 0.32628288865089417, |
| "learning_rate": 0.0005, |
| "loss": 1.2584, |
| "step": 2420 |
| }, |
| { |
| "epoch": 0.3895011019835704, |
| "grad_norm": 0.41335487365722656, |
| "learning_rate": 0.0005, |
| "loss": 1.2731, |
| "step": 2430 |
| }, |
| { |
| "epoch": 0.39110398717691847, |
| "grad_norm": 0.4485657215118408, |
| "learning_rate": 0.0005, |
| "loss": 1.2715, |
| "step": 2440 |
| }, |
| { |
| "epoch": 0.39270687237026647, |
| "grad_norm": 0.4431282877922058, |
| "learning_rate": 0.0005, |
| "loss": 1.2634, |
| "step": 2450 |
| }, |
| { |
| "epoch": 0.3943097575636145, |
| "grad_norm": 0.2843770682811737, |
| "learning_rate": 0.0005, |
| "loss": 1.2706, |
| "step": 2460 |
| }, |
| { |
| "epoch": 0.39591264275696253, |
| "grad_norm": 0.3248964250087738, |
| "learning_rate": 0.0005, |
| "loss": 1.2479, |
| "step": 2470 |
| }, |
| { |
| "epoch": 0.39751552795031053, |
| "grad_norm": 0.2866270840167999, |
| "learning_rate": 0.0005, |
| "loss": 1.2202, |
| "step": 2480 |
| }, |
| { |
| "epoch": 0.3991184131436586, |
| "grad_norm": 0.296093612909317, |
| "learning_rate": 0.0005, |
| "loss": 1.2563, |
| "step": 2490 |
| }, |
| { |
| "epoch": 0.4007212983370066, |
| "grad_norm": 0.30020371079444885, |
| "learning_rate": 0.0005, |
| "loss": 1.2569, |
| "step": 2500 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 2500, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 100, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 6.105286991832678e+19, |
| "train_batch_size": 4, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
