| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 5.0, |
| "eval_steps": 500, |
| "global_step": 1235, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.004048582995951417, |
| "grad_norm": 6.563708031524433, |
| "learning_rate": 3.2258064516129035e-07, |
| "loss": 0.8324, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.008097165991902834, |
| "grad_norm": 6.491777079921468, |
| "learning_rate": 6.451612903225807e-07, |
| "loss": 0.8118, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.012145748987854251, |
| "grad_norm": 6.454949472086113, |
| "learning_rate": 9.67741935483871e-07, |
| "loss": 0.8099, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.016194331983805668, |
| "grad_norm": 6.412626644724627, |
| "learning_rate": 1.2903225806451614e-06, |
| "loss": 0.8176, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.020242914979757085, |
| "grad_norm": 6.147439194405763, |
| "learning_rate": 1.6129032258064516e-06, |
| "loss": 0.8034, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.024291497975708502, |
| "grad_norm": 5.772945240794099, |
| "learning_rate": 1.935483870967742e-06, |
| "loss": 0.7939, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.02834008097165992, |
| "grad_norm": 4.877285063645693, |
| "learning_rate": 2.2580645161290324e-06, |
| "loss": 0.7616, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.032388663967611336, |
| "grad_norm": 4.513896907301424, |
| "learning_rate": 2.580645161290323e-06, |
| "loss": 0.7495, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.03643724696356275, |
| "grad_norm": 2.620140323877515, |
| "learning_rate": 2.903225806451613e-06, |
| "loss": 0.6788, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.04048582995951417, |
| "grad_norm": 2.478026443767167, |
| "learning_rate": 3.225806451612903e-06, |
| "loss": 0.6665, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.044534412955465584, |
| "grad_norm": 2.2076556313898505, |
| "learning_rate": 3.548387096774194e-06, |
| "loss": 0.6836, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.048582995951417005, |
| "grad_norm": 3.803502801943972, |
| "learning_rate": 3.870967741935484e-06, |
| "loss": 0.6855, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.05263157894736842, |
| "grad_norm": 3.7184515657372548, |
| "learning_rate": 4.193548387096774e-06, |
| "loss": 0.6594, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.05668016194331984, |
| "grad_norm": 3.644946744143057, |
| "learning_rate": 4.516129032258065e-06, |
| "loss": 0.6394, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.06072874493927125, |
| "grad_norm": 3.597054397744173, |
| "learning_rate": 4.838709677419355e-06, |
| "loss": 0.6265, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.06477732793522267, |
| "grad_norm": 3.1373189872980727, |
| "learning_rate": 5.161290322580646e-06, |
| "loss": 0.6348, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.06882591093117409, |
| "grad_norm": 2.8575900951669664, |
| "learning_rate": 5.483870967741935e-06, |
| "loss": 0.6011, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.0728744939271255, |
| "grad_norm": 2.4964332188805267, |
| "learning_rate": 5.806451612903226e-06, |
| "loss": 0.5882, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.07692307692307693, |
| "grad_norm": 1.7431608438942332, |
| "learning_rate": 6.129032258064517e-06, |
| "loss": 0.5694, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.08097165991902834, |
| "grad_norm": 1.319228681187466, |
| "learning_rate": 6.451612903225806e-06, |
| "loss": 0.5532, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.08502024291497975, |
| "grad_norm": 1.41792608545027, |
| "learning_rate": 6.774193548387097e-06, |
| "loss": 0.5472, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.08906882591093117, |
| "grad_norm": 1.4713567642770515, |
| "learning_rate": 7.096774193548388e-06, |
| "loss": 0.5376, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.0931174089068826, |
| "grad_norm": 1.6378285536138395, |
| "learning_rate": 7.4193548387096784e-06, |
| "loss": 0.5593, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.09716599190283401, |
| "grad_norm": 1.0252078399623044, |
| "learning_rate": 7.741935483870968e-06, |
| "loss": 0.5376, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.10121457489878542, |
| "grad_norm": 0.8281396336743938, |
| "learning_rate": 8.064516129032258e-06, |
| "loss": 0.5188, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.10526315789473684, |
| "grad_norm": 1.043742200591078, |
| "learning_rate": 8.387096774193549e-06, |
| "loss": 0.5123, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.10931174089068826, |
| "grad_norm": 0.961032030244357, |
| "learning_rate": 8.70967741935484e-06, |
| "loss": 0.513, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.11336032388663968, |
| "grad_norm": 0.7181197009567644, |
| "learning_rate": 9.03225806451613e-06, |
| "loss": 0.5081, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.11740890688259109, |
| "grad_norm": 0.6932489318432652, |
| "learning_rate": 9.35483870967742e-06, |
| "loss": 0.5085, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.1214574898785425, |
| "grad_norm": 0.8219144796926592, |
| "learning_rate": 9.67741935483871e-06, |
| "loss": 0.4726, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.12550607287449392, |
| "grad_norm": 0.6999445738164941, |
| "learning_rate": 1e-05, |
| "loss": 0.4941, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.12955465587044535, |
| "grad_norm": 0.6923074090117829, |
| "learning_rate": 1.0322580645161291e-05, |
| "loss": 0.4863, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.13360323886639677, |
| "grad_norm": 0.7285883425220445, |
| "learning_rate": 1.0645161290322582e-05, |
| "loss": 0.5054, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.13765182186234817, |
| "grad_norm": 0.6720589817118815, |
| "learning_rate": 1.096774193548387e-05, |
| "loss": 0.4743, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.1417004048582996, |
| "grad_norm": 0.6222336618033134, |
| "learning_rate": 1.1290322580645164e-05, |
| "loss": 0.4824, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.145748987854251, |
| "grad_norm": 0.580681467117815, |
| "learning_rate": 1.1612903225806453e-05, |
| "loss": 0.4751, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.14979757085020243, |
| "grad_norm": 0.6513381452810239, |
| "learning_rate": 1.1935483870967743e-05, |
| "loss": 0.4587, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.15384615384615385, |
| "grad_norm": 0.7106225840718775, |
| "learning_rate": 1.2258064516129034e-05, |
| "loss": 0.4992, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.15789473684210525, |
| "grad_norm": 0.5591864692672563, |
| "learning_rate": 1.2580645161290324e-05, |
| "loss": 0.4788, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.16194331983805668, |
| "grad_norm": 0.6060244697388646, |
| "learning_rate": 1.2903225806451613e-05, |
| "loss": 0.4527, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.1659919028340081, |
| "grad_norm": 0.5810611352547224, |
| "learning_rate": 1.3225806451612903e-05, |
| "loss": 0.4707, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.1700404858299595, |
| "grad_norm": 0.6002272244302436, |
| "learning_rate": 1.3548387096774194e-05, |
| "loss": 0.4516, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.17408906882591094, |
| "grad_norm": 0.5424118212871981, |
| "learning_rate": 1.3870967741935486e-05, |
| "loss": 0.4469, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.17813765182186234, |
| "grad_norm": 0.5677963930798434, |
| "learning_rate": 1.4193548387096776e-05, |
| "loss": 0.4487, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.18218623481781376, |
| "grad_norm": 0.48399576825688306, |
| "learning_rate": 1.4516129032258066e-05, |
| "loss": 0.4613, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.1862348178137652, |
| "grad_norm": 0.7211845265363408, |
| "learning_rate": 1.4838709677419357e-05, |
| "loss": 0.4695, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.1902834008097166, |
| "grad_norm": 0.48227263867078823, |
| "learning_rate": 1.5161290322580646e-05, |
| "loss": 0.45, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.19433198380566802, |
| "grad_norm": 0.4948399076215356, |
| "learning_rate": 1.5483870967741936e-05, |
| "loss": 0.4495, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.19838056680161945, |
| "grad_norm": 0.4933097381120239, |
| "learning_rate": 1.5806451612903226e-05, |
| "loss": 0.4694, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.20242914979757085, |
| "grad_norm": 0.5000277117099814, |
| "learning_rate": 1.6129032258064517e-05, |
| "loss": 0.4552, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.20647773279352227, |
| "grad_norm": 0.5291362499613479, |
| "learning_rate": 1.6451612903225807e-05, |
| "loss": 0.4498, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.21052631578947367, |
| "grad_norm": 0.5575503306516764, |
| "learning_rate": 1.6774193548387098e-05, |
| "loss": 0.4667, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.2145748987854251, |
| "grad_norm": 0.6251186215337555, |
| "learning_rate": 1.7096774193548388e-05, |
| "loss": 0.4505, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.21862348178137653, |
| "grad_norm": 0.4580508404969374, |
| "learning_rate": 1.741935483870968e-05, |
| "loss": 0.4288, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.22267206477732793, |
| "grad_norm": 0.6397311419916276, |
| "learning_rate": 1.774193548387097e-05, |
| "loss": 0.4501, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.22672064777327935, |
| "grad_norm": 0.5279163370241632, |
| "learning_rate": 1.806451612903226e-05, |
| "loss": 0.443, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.23076923076923078, |
| "grad_norm": 0.7056335466409, |
| "learning_rate": 1.838709677419355e-05, |
| "loss": 0.4596, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.23481781376518218, |
| "grad_norm": 0.6291313133697399, |
| "learning_rate": 1.870967741935484e-05, |
| "loss": 0.452, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.2388663967611336, |
| "grad_norm": 0.5841302956072792, |
| "learning_rate": 1.903225806451613e-05, |
| "loss": 0.4512, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.242914979757085, |
| "grad_norm": 0.46330918968359724, |
| "learning_rate": 1.935483870967742e-05, |
| "loss": 0.4358, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.24696356275303644, |
| "grad_norm": 0.4903374262430884, |
| "learning_rate": 1.967741935483871e-05, |
| "loss": 0.4233, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.25101214574898784, |
| "grad_norm": 0.5007460681335626, |
| "learning_rate": 2e-05, |
| "loss": 0.4273, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.2550607287449393, |
| "grad_norm": 0.5193870387432274, |
| "learning_rate": 2.0322580645161292e-05, |
| "loss": 0.4362, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.2591093117408907, |
| "grad_norm": 0.4748397799774122, |
| "learning_rate": 2.0645161290322582e-05, |
| "loss": 0.4421, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.2631578947368421, |
| "grad_norm": 0.6361428849330221, |
| "learning_rate": 2.0967741935483873e-05, |
| "loss": 0.4347, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.26720647773279355, |
| "grad_norm": 0.5327272580086719, |
| "learning_rate": 2.1290322580645163e-05, |
| "loss": 0.4447, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.27125506072874495, |
| "grad_norm": 0.5060954027811673, |
| "learning_rate": 2.161290322580645e-05, |
| "loss": 0.4515, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.27530364372469635, |
| "grad_norm": 0.540187045509995, |
| "learning_rate": 2.193548387096774e-05, |
| "loss": 0.4181, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.2793522267206478, |
| "grad_norm": 0.5463663542345398, |
| "learning_rate": 2.225806451612903e-05, |
| "loss": 0.4243, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.2834008097165992, |
| "grad_norm": 0.6101772132801231, |
| "learning_rate": 2.2580645161290328e-05, |
| "loss": 0.4363, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.2874493927125506, |
| "grad_norm": 0.534407812101356, |
| "learning_rate": 2.2903225806451615e-05, |
| "loss": 0.4406, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.291497975708502, |
| "grad_norm": 0.6203964517752467, |
| "learning_rate": 2.3225806451612906e-05, |
| "loss": 0.4281, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.29554655870445345, |
| "grad_norm": 0.545062181928916, |
| "learning_rate": 2.3548387096774196e-05, |
| "loss": 0.4425, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.29959514170040485, |
| "grad_norm": 0.6944702741448721, |
| "learning_rate": 2.3870967741935486e-05, |
| "loss": 0.4176, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.30364372469635625, |
| "grad_norm": 0.7067843200461914, |
| "learning_rate": 2.4193548387096777e-05, |
| "loss": 0.4487, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.3076923076923077, |
| "grad_norm": 0.5766468445010243, |
| "learning_rate": 2.4516129032258067e-05, |
| "loss": 0.4412, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.3117408906882591, |
| "grad_norm": 0.6659232719013346, |
| "learning_rate": 2.4838709677419358e-05, |
| "loss": 0.4251, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.3157894736842105, |
| "grad_norm": 0.7003764241298202, |
| "learning_rate": 2.5161290322580648e-05, |
| "loss": 0.4268, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.31983805668016196, |
| "grad_norm": 0.844454415023376, |
| "learning_rate": 2.548387096774194e-05, |
| "loss": 0.4365, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.32388663967611336, |
| "grad_norm": 0.5412688524732145, |
| "learning_rate": 2.5806451612903226e-05, |
| "loss": 0.4308, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.32793522267206476, |
| "grad_norm": 0.7577027876709705, |
| "learning_rate": 2.6129032258064516e-05, |
| "loss": 0.4291, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.3319838056680162, |
| "grad_norm": 0.570036174880917, |
| "learning_rate": 2.6451612903225806e-05, |
| "loss": 0.4232, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.3360323886639676, |
| "grad_norm": 0.7650296023888453, |
| "learning_rate": 2.6774193548387097e-05, |
| "loss": 0.4173, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.340080971659919, |
| "grad_norm": 0.5878693825080787, |
| "learning_rate": 2.7096774193548387e-05, |
| "loss": 0.4287, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.3441295546558704, |
| "grad_norm": 0.6426825536799831, |
| "learning_rate": 2.7419354838709678e-05, |
| "loss": 0.4431, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.3481781376518219, |
| "grad_norm": 0.5821433346643433, |
| "learning_rate": 2.774193548387097e-05, |
| "loss": 0.4297, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.3522267206477733, |
| "grad_norm": 0.5528950833300484, |
| "learning_rate": 2.8064516129032262e-05, |
| "loss": 0.4236, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.3562753036437247, |
| "grad_norm": 0.6382067207381356, |
| "learning_rate": 2.8387096774193552e-05, |
| "loss": 0.4355, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.3603238866396761, |
| "grad_norm": 0.7737524570722991, |
| "learning_rate": 2.8709677419354843e-05, |
| "loss": 0.4328, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.3643724696356275, |
| "grad_norm": 0.6850636517361031, |
| "learning_rate": 2.9032258064516133e-05, |
| "loss": 0.4426, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.3684210526315789, |
| "grad_norm": 0.8723393701354069, |
| "learning_rate": 2.9354838709677423e-05, |
| "loss": 0.4388, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.3724696356275304, |
| "grad_norm": 0.5623350426250757, |
| "learning_rate": 2.9677419354838714e-05, |
| "loss": 0.4276, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.3765182186234818, |
| "grad_norm": 0.6961451980001087, |
| "learning_rate": 3.0000000000000004e-05, |
| "loss": 0.4208, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.3805668016194332, |
| "grad_norm": 0.736893530519594, |
| "learning_rate": 3.032258064516129e-05, |
| "loss": 0.4182, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.38461538461538464, |
| "grad_norm": 0.6663629750852842, |
| "learning_rate": 3.0645161290322585e-05, |
| "loss": 0.4184, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.38866396761133604, |
| "grad_norm": 0.9071122296698014, |
| "learning_rate": 3.096774193548387e-05, |
| "loss": 0.4357, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.39271255060728744, |
| "grad_norm": 0.8131439066627294, |
| "learning_rate": 3.1290322580645166e-05, |
| "loss": 0.4236, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.3967611336032389, |
| "grad_norm": 0.5934474566749712, |
| "learning_rate": 3.161290322580645e-05, |
| "loss": 0.4076, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.4008097165991903, |
| "grad_norm": 0.7087615330186364, |
| "learning_rate": 3.193548387096774e-05, |
| "loss": 0.443, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.4048582995951417, |
| "grad_norm": 0.5897488251004153, |
| "learning_rate": 3.2258064516129034e-05, |
| "loss": 0.4304, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.4089068825910931, |
| "grad_norm": 0.5433994277384446, |
| "learning_rate": 3.258064516129033e-05, |
| "loss": 0.4196, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.41295546558704455, |
| "grad_norm": 0.5418857744380926, |
| "learning_rate": 3.2903225806451614e-05, |
| "loss": 0.4132, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.41700404858299595, |
| "grad_norm": 0.5709693505729505, |
| "learning_rate": 3.322580645161291e-05, |
| "loss": 0.4219, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.42105263157894735, |
| "grad_norm": 0.6302532652451412, |
| "learning_rate": 3.3548387096774195e-05, |
| "loss": 0.4212, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.4251012145748988, |
| "grad_norm": 0.6192017676989524, |
| "learning_rate": 3.387096774193549e-05, |
| "loss": 0.4244, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.4291497975708502, |
| "grad_norm": 0.9723107661465855, |
| "learning_rate": 3.4193548387096776e-05, |
| "loss": 0.4384, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.4331983805668016, |
| "grad_norm": 0.816660099013038, |
| "learning_rate": 3.451612903225807e-05, |
| "loss": 0.4372, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.43724696356275305, |
| "grad_norm": 1.005901463121118, |
| "learning_rate": 3.483870967741936e-05, |
| "loss": 0.418, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.44129554655870445, |
| "grad_norm": 0.7024357857225915, |
| "learning_rate": 3.516129032258065e-05, |
| "loss": 0.4119, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.44534412955465585, |
| "grad_norm": 0.6714668597310268, |
| "learning_rate": 3.548387096774194e-05, |
| "loss": 0.4268, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.4493927125506073, |
| "grad_norm": 0.6404766639943532, |
| "learning_rate": 3.5806451612903225e-05, |
| "loss": 0.4276, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.4534412955465587, |
| "grad_norm": 0.6350096256618297, |
| "learning_rate": 3.612903225806452e-05, |
| "loss": 0.4253, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.4574898785425101, |
| "grad_norm": 0.6038152391757917, |
| "learning_rate": 3.6451612903225805e-05, |
| "loss": 0.4035, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.46153846153846156, |
| "grad_norm": 0.6129208056234016, |
| "learning_rate": 3.67741935483871e-05, |
| "loss": 0.4093, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.46558704453441296, |
| "grad_norm": 0.5616934782084458, |
| "learning_rate": 3.7096774193548386e-05, |
| "loss": 0.4115, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.46963562753036436, |
| "grad_norm": 0.6960874665244597, |
| "learning_rate": 3.741935483870968e-05, |
| "loss": 0.415, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.47368421052631576, |
| "grad_norm": 0.6026648762764232, |
| "learning_rate": 3.7741935483870974e-05, |
| "loss": 0.4118, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.4777327935222672, |
| "grad_norm": 0.7405544668723468, |
| "learning_rate": 3.806451612903226e-05, |
| "loss": 0.4132, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.4817813765182186, |
| "grad_norm": 0.5697461578522112, |
| "learning_rate": 3.8387096774193555e-05, |
| "loss": 0.4272, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.48582995951417, |
| "grad_norm": 0.7639089478756494, |
| "learning_rate": 3.870967741935484e-05, |
| "loss": 0.431, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.4898785425101215, |
| "grad_norm": 0.6424062264785787, |
| "learning_rate": 3.9032258064516135e-05, |
| "loss": 0.4415, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.4939271255060729, |
| "grad_norm": 0.6711495655670747, |
| "learning_rate": 3.935483870967742e-05, |
| "loss": 0.4052, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.4979757085020243, |
| "grad_norm": 0.782996135587719, |
| "learning_rate": 3.9677419354838716e-05, |
| "loss": 0.4185, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.5020242914979757, |
| "grad_norm": 0.6342396241071421, |
| "learning_rate": 4e-05, |
| "loss": 0.4339, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.5060728744939271, |
| "grad_norm": 0.7060230617361839, |
| "learning_rate": 3.9999920040266474e-05, |
| "loss": 0.4194, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.5101214574898786, |
| "grad_norm": 0.8605230488308427, |
| "learning_rate": 3.999968016170525e-05, |
| "loss": 0.4297, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.5141700404858299, |
| "grad_norm": 0.7164721659299549, |
| "learning_rate": 3.9999280366234394e-05, |
| "loss": 0.4074, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.5182186234817814, |
| "grad_norm": 0.9069399813167286, |
| "learning_rate": 3.999872065705065e-05, |
| "loss": 0.438, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.5222672064777328, |
| "grad_norm": 0.8766908165951781, |
| "learning_rate": 3.999800103862945e-05, |
| "loss": 0.4219, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.5263157894736842, |
| "grad_norm": 0.75251455873621, |
| "learning_rate": 3.9997121516724834e-05, |
| "loss": 0.4168, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.5303643724696356, |
| "grad_norm": 0.8150131825982871, |
| "learning_rate": 3.999608209836944e-05, |
| "loss": 0.4273, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.5344129554655871, |
| "grad_norm": 0.5820177235802553, |
| "learning_rate": 3.999488279187443e-05, |
| "loss": 0.4286, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.5384615384615384, |
| "grad_norm": 0.8017373799359105, |
| "learning_rate": 3.9993523606829425e-05, |
| "loss": 0.4248, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.5425101214574899, |
| "grad_norm": 0.5602345349430007, |
| "learning_rate": 3.9992004554102435e-05, |
| "loss": 0.4369, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.5465587044534413, |
| "grad_norm": 0.6751620794799653, |
| "learning_rate": 3.999032564583976e-05, |
| "loss": 0.4101, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.5506072874493927, |
| "grad_norm": 0.5649667376218699, |
| "learning_rate": 3.998848689546591e-05, |
| "loss": 0.4124, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.5546558704453441, |
| "grad_norm": 0.6563384178470728, |
| "learning_rate": 3.9986488317683496e-05, |
| "loss": 0.4255, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.5587044534412956, |
| "grad_norm": 0.4441009962016813, |
| "learning_rate": 3.9984329928473067e-05, |
| "loss": 0.4123, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.562753036437247, |
| "grad_norm": 0.5602163510631745, |
| "learning_rate": 3.998201174509306e-05, |
| "loss": 0.416, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.5668016194331984, |
| "grad_norm": 0.5503870647988064, |
| "learning_rate": 3.997953378607961e-05, |
| "loss": 0.4355, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.5708502024291497, |
| "grad_norm": 0.558838242995912, |
| "learning_rate": 3.9976896071246403e-05, |
| "loss": 0.4411, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.5748987854251012, |
| "grad_norm": 0.46958999429906817, |
| "learning_rate": 3.997409862168454e-05, |
| "loss": 0.4217, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.5789473684210527, |
| "grad_norm": 0.537915286192827, |
| "learning_rate": 3.997114145976237e-05, |
| "loss": 0.4233, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.582995951417004, |
| "grad_norm": 0.6056647446866671, |
| "learning_rate": 3.9968024609125245e-05, |
| "loss": 0.406, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.5870445344129555, |
| "grad_norm": 0.5411982967401662, |
| "learning_rate": 3.996474809469545e-05, |
| "loss": 0.4234, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.5910931174089069, |
| "grad_norm": 0.5156039096840958, |
| "learning_rate": 3.996131194267189e-05, |
| "loss": 0.393, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.5951417004048583, |
| "grad_norm": 0.5540730780674173, |
| "learning_rate": 3.995771618052995e-05, |
| "loss": 0.4031, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.5991902834008097, |
| "grad_norm": 0.7018067193438458, |
| "learning_rate": 3.9953960837021256e-05, |
| "loss": 0.4296, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.6032388663967612, |
| "grad_norm": 0.8077378442049692, |
| "learning_rate": 3.995004594217343e-05, |
| "loss": 0.4263, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.6072874493927125, |
| "grad_norm": 0.5335503130293905, |
| "learning_rate": 3.994597152728987e-05, |
| "loss": 0.4046, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.611336032388664, |
| "grad_norm": 0.6360248826283119, |
| "learning_rate": 3.994173762494947e-05, |
| "loss": 0.4169, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.6153846153846154, |
| "grad_norm": 0.6905221458577911, |
| "learning_rate": 3.9937344269006424e-05, |
| "loss": 0.4038, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.6194331983805668, |
| "grad_norm": 0.6602960318121116, |
| "learning_rate": 3.993279149458988e-05, |
| "loss": 0.4256, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.6234817813765182, |
| "grad_norm": 0.5635384998575663, |
| "learning_rate": 3.9928079338103695e-05, |
| "loss": 0.4079, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.6275303643724697, |
| "grad_norm": 0.4683817237781957, |
| "learning_rate": 3.992320783722616e-05, |
| "loss": 0.4221, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.631578947368421, |
| "grad_norm": 0.5828224777421622, |
| "learning_rate": 3.991817703090965e-05, |
| "loss": 0.4285, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.6356275303643725, |
| "grad_norm": 0.5946767220113172, |
| "learning_rate": 3.991298695938038e-05, |
| "loss": 0.4041, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.6396761133603239, |
| "grad_norm": 0.439299955627649, |
| "learning_rate": 3.9907637664138e-05, |
| "loss": 0.4085, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.6437246963562753, |
| "grad_norm": 0.45337870806784775, |
| "learning_rate": 3.990212918795534e-05, |
| "loss": 0.3889, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.6477732793522267, |
| "grad_norm": 0.5100647310462432, |
| "learning_rate": 3.9896461574878036e-05, |
| "loss": 0.4071, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.6518218623481782, |
| "grad_norm": 0.4694789911417236, |
| "learning_rate": 3.989063487022416e-05, |
| "loss": 0.4061, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.6558704453441295, |
| "grad_norm": 0.5479297402805408, |
| "learning_rate": 3.9884649120583903e-05, |
| "loss": 0.409, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.659919028340081, |
| "grad_norm": 0.5651956013777776, |
| "learning_rate": 3.987850437381915e-05, |
| "loss": 0.4165, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.6639676113360324, |
| "grad_norm": 0.47160237946411626, |
| "learning_rate": 3.987220067906313e-05, |
| "loss": 0.429, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.6680161943319838, |
| "grad_norm": 0.6026109343123514, |
| "learning_rate": 3.986573808672002e-05, |
| "loss": 0.4213, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.6720647773279352, |
| "grad_norm": 0.47559758719813583, |
| "learning_rate": 3.9859116648464534e-05, |
| "loss": 0.4317, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.6761133603238867, |
| "grad_norm": 0.44905976441003864, |
| "learning_rate": 3.985233641724152e-05, |
| "loss": 0.4096, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.680161943319838, |
| "grad_norm": 0.5756236914271495, |
| "learning_rate": 3.984539744726554e-05, |
| "loss": 0.4182, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.6842105263157895, |
| "grad_norm": 0.5645437126875592, |
| "learning_rate": 3.983829979402038e-05, |
| "loss": 0.4039, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.6882591093117408, |
| "grad_norm": 0.5280363199943358, |
| "learning_rate": 3.983104351425871e-05, |
| "loss": 0.4125, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.6923076923076923, |
| "grad_norm": 0.6318932054518549, |
| "learning_rate": 3.982362866600154e-05, |
| "loss": 0.4086, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.6963562753036437, |
| "grad_norm": 0.5125013679687952, |
| "learning_rate": 3.9816055308537806e-05, |
| "loss": 0.4017, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.7004048582995951, |
| "grad_norm": 0.5215096153067412, |
| "learning_rate": 3.980832350242388e-05, |
| "loss": 0.4121, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.7044534412955465, |
| "grad_norm": 0.5465865244699569, |
| "learning_rate": 3.980043330948306e-05, |
| "loss": 0.4093, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.708502024291498, |
| "grad_norm": 0.7046984232135567, |
| "learning_rate": 3.9792384792805124e-05, |
| "loss": 0.4054, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.7125506072874493, |
| "grad_norm": 0.4784739611200344, |
| "learning_rate": 3.978417801674579e-05, |
| "loss": 0.4166, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.7165991902834008, |
| "grad_norm": 0.545856331954117, |
| "learning_rate": 3.9775813046926244e-05, |
| "loss": 0.425, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.7206477732793523, |
| "grad_norm": 0.5288335123941479, |
| "learning_rate": 3.976728995023254e-05, |
| "loss": 0.4184, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.7246963562753036, |
| "grad_norm": 0.4901664814227653, |
| "learning_rate": 3.975860879481514e-05, |
| "loss": 0.3982, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.728744939271255, |
| "grad_norm": 0.573352589690754, |
| "learning_rate": 3.974976965008833e-05, |
| "loss": 0.4008, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.7327935222672065, |
| "grad_norm": 0.43357407705045037, |
| "learning_rate": 3.9740772586729674e-05, |
| "loss": 0.402, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.7368421052631579, |
| "grad_norm": 0.44028649303556666, |
| "learning_rate": 3.973161767667946e-05, |
| "loss": 0.3801, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.7408906882591093, |
| "grad_norm": 0.43386781806448627, |
| "learning_rate": 3.972230499314009e-05, |
| "loss": 0.4035, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.7449392712550608, |
| "grad_norm": 0.4330305139339875, |
| "learning_rate": 3.9712834610575545e-05, |
| "loss": 0.4133, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.7489878542510121, |
| "grad_norm": 0.47329689781923384, |
| "learning_rate": 3.9703206604710746e-05, |
| "loss": 0.4004, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.7530364372469636, |
| "grad_norm": 0.5636788983249776, |
| "learning_rate": 3.969342105253098e-05, |
| "loss": 0.4078, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.757085020242915, |
| "grad_norm": 0.4344821798002646, |
| "learning_rate": 3.9683478032281246e-05, |
| "loss": 0.3962, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.7611336032388664, |
| "grad_norm": 0.4378972618535626, |
| "learning_rate": 3.967337762346568e-05, |
| "loss": 0.4214, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.7651821862348178, |
| "grad_norm": 0.43207903392613417, |
| "learning_rate": 3.9663119906846885e-05, |
| "loss": 0.4093, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.7692307692307693, |
| "grad_norm": 0.41306586090140507, |
| "learning_rate": 3.965270496444528e-05, |
| "loss": 0.4149, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.7732793522267206, |
| "grad_norm": 0.34683571842467487, |
| "learning_rate": 3.9642132879538465e-05, |
| "loss": 0.3957, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.7773279352226721, |
| "grad_norm": 0.46021438345677174, |
| "learning_rate": 3.9631403736660566e-05, |
| "loss": 0.4064, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.7813765182186235, |
| "grad_norm": 0.37384305142115987, |
| "learning_rate": 3.962051762160151e-05, |
| "loss": 0.4093, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.7854251012145749, |
| "grad_norm": 0.46052065405719134, |
| "learning_rate": 3.960947462140638e-05, |
| "loss": 0.3966, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.7894736842105263, |
| "grad_norm": 0.37919617839932224, |
| "learning_rate": 3.959827482437472e-05, |
| "loss": 0.398, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.7935222672064778, |
| "grad_norm": 0.4809299886856657, |
| "learning_rate": 3.958691832005981e-05, |
| "loss": 0.4163, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.7975708502024291, |
| "grad_norm": 0.4302245180296029, |
| "learning_rate": 3.9575405199267944e-05, |
| "loss": 0.403, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.8016194331983806, |
| "grad_norm": 0.5346817998236933, |
| "learning_rate": 3.956373555405774e-05, |
| "loss": 0.4022, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.805668016194332, |
| "grad_norm": 0.48423918028332824, |
| "learning_rate": 3.955190947773937e-05, |
| "loss": 0.4031, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.8097165991902834, |
| "grad_norm": 0.49786486157606447, |
| "learning_rate": 3.9539927064873824e-05, |
| "loss": 0.415, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.8137651821862348, |
| "grad_norm": 0.43372278507309925, |
| "learning_rate": 3.952778841127214e-05, |
| "loss": 0.4181, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.8178137651821862, |
| "grad_norm": 0.35858339706424547, |
| "learning_rate": 3.95154936139947e-05, |
| "loss": 0.3957, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.8218623481781376, |
| "grad_norm": 0.485757788601579, |
| "learning_rate": 3.950304277135034e-05, |
| "loss": 0.4182, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.8259109311740891, |
| "grad_norm": 0.39575858269226627, |
| "learning_rate": 3.94904359828957e-05, |
| "loss": 0.4097, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.8299595141700404, |
| "grad_norm": 0.4935624582409233, |
| "learning_rate": 3.94776733494343e-05, |
| "loss": 0.4085, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.8340080971659919, |
| "grad_norm": 0.40863883305655685, |
| "learning_rate": 3.9464754973015826e-05, |
| "loss": 0.3921, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.8380566801619433, |
| "grad_norm": 0.5082512995489098, |
| "learning_rate": 3.9451680956935275e-05, |
| "loss": 0.4179, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.8421052631578947, |
| "grad_norm": 0.468556620758671, |
| "learning_rate": 3.943845140573213e-05, |
| "loss": 0.4122, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.8461538461538461, |
| "grad_norm": 0.48932049470816236, |
| "learning_rate": 3.942506642518952e-05, |
| "loss": 0.4118, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.8502024291497976, |
| "grad_norm": 0.49601754723791763, |
| "learning_rate": 3.9411526122333396e-05, |
| "loss": 0.4122, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.854251012145749, |
| "grad_norm": 0.47824166428550063, |
| "learning_rate": 3.939783060543167e-05, |
| "loss": 0.4046, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.8582995951417004, |
| "grad_norm": 0.4597313656930914, |
| "learning_rate": 3.938397998399333e-05, |
| "loss": 0.4149, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.8623481781376519, |
| "grad_norm": 0.5219668092131503, |
| "learning_rate": 3.936997436876756e-05, |
| "loss": 0.4097, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.8663967611336032, |
| "grad_norm": 0.4503569552639839, |
| "learning_rate": 3.935581387174289e-05, |
| "loss": 0.4092, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.8704453441295547, |
| "grad_norm": 0.4718008578200377, |
| "learning_rate": 3.9341498606146295e-05, |
| "loss": 0.4232, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.8744939271255061, |
| "grad_norm": 0.4846401258402992, |
| "learning_rate": 3.932702868644224e-05, |
| "loss": 0.3927, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.8785425101214575, |
| "grad_norm": 0.42620562715111837, |
| "learning_rate": 3.931240422833181e-05, |
| "loss": 0.3884, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.8825910931174089, |
| "grad_norm": 0.3637632313708226, |
| "learning_rate": 3.9297625348751806e-05, |
| "loss": 0.4092, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.8866396761133604, |
| "grad_norm": 0.3974202808826163, |
| "learning_rate": 3.9282692165873744e-05, |
| "loss": 0.4084, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.8906882591093117, |
| "grad_norm": 0.3695400971522709, |
| "learning_rate": 3.9267604799102945e-05, |
| "loss": 0.4199, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.8947368421052632, |
| "grad_norm": 0.40114344743379804, |
| "learning_rate": 3.9252363369077615e-05, |
| "loss": 0.4231, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.8987854251012146, |
| "grad_norm": 0.41407118563162076, |
| "learning_rate": 3.9236967997667796e-05, |
| "loss": 0.4041, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.902834008097166, |
| "grad_norm": 0.3954744465635331, |
| "learning_rate": 3.9221418807974496e-05, |
| "loss": 0.4046, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.9068825910931174, |
| "grad_norm": 0.3937493509020905, |
| "learning_rate": 3.92057159243286e-05, |
| "loss": 0.407, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.9109311740890689, |
| "grad_norm": 0.4548525501482176, |
| "learning_rate": 3.9189859472289956e-05, |
| "loss": 0.4087, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.9149797570850202, |
| "grad_norm": 0.37253726903358675, |
| "learning_rate": 3.9173849578646324e-05, |
| "loss": 0.4069, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.9190283400809717, |
| "grad_norm": 0.475262348455805, |
| "learning_rate": 3.9157686371412395e-05, |
| "loss": 0.396, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.9230769230769231, |
| "grad_norm": 0.38301516506198463, |
| "learning_rate": 3.914136997982874e-05, |
| "loss": 0.4043, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.9271255060728745, |
| "grad_norm": 0.49215886913673323, |
| "learning_rate": 3.912490053436079e-05, |
| "loss": 0.4155, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.9311740890688259, |
| "grad_norm": 0.5239294726598561, |
| "learning_rate": 3.9108278166697805e-05, |
| "loss": 0.4053, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.9352226720647774, |
| "grad_norm": 0.3862509850454398, |
| "learning_rate": 3.909150300975178e-05, |
| "loss": 0.4019, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.9392712550607287, |
| "grad_norm": 0.4192177716540819, |
| "learning_rate": 3.907457519765642e-05, |
| "loss": 0.409, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.9433198380566802, |
| "grad_norm": 0.5202900330675131, |
| "learning_rate": 3.905749486576607e-05, |
| "loss": 0.3986, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.9473684210526315, |
| "grad_norm": 0.42511854966801393, |
| "learning_rate": 3.90402621506546e-05, |
| "loss": 0.4019, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.951417004048583, |
| "grad_norm": 0.46706618895133944, |
| "learning_rate": 3.902287719011435e-05, |
| "loss": 0.3982, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.9554655870445344, |
| "grad_norm": 0.5287082613810972, |
| "learning_rate": 3.900534012315499e-05, |
| "loss": 0.4146, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.9595141700404858, |
| "grad_norm": 0.370039795695217, |
| "learning_rate": 3.8987651090002454e-05, |
| "loss": 0.3933, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.9635627530364372, |
| "grad_norm": 0.4127469185732051, |
| "learning_rate": 3.896981023209777e-05, |
| "loss": 0.4131, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.9676113360323887, |
| "grad_norm": 0.42008114467385965, |
| "learning_rate": 3.8951817692095966e-05, |
| "loss": 0.3986, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.97165991902834, |
| "grad_norm": 0.38946004399509976, |
| "learning_rate": 3.893367361386491e-05, |
| "loss": 0.4014, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.9757085020242915, |
| "grad_norm": 0.4544456305498687, |
| "learning_rate": 3.891537814248417e-05, |
| "loss": 0.4032, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.979757085020243, |
| "grad_norm": 0.3922254950192444, |
| "learning_rate": 3.889693142424384e-05, |
| "loss": 0.3887, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.9838056680161943, |
| "grad_norm": 0.3783335958602488, |
| "learning_rate": 3.8878333606643405e-05, |
| "loss": 0.3998, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.9878542510121457, |
| "grad_norm": 0.4136456605406144, |
| "learning_rate": 3.885958483839049e-05, |
| "loss": 0.412, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.9919028340080972, |
| "grad_norm": 0.3737708526762848, |
| "learning_rate": 3.8840685269399786e-05, |
| "loss": 0.3817, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.9959514170040485, |
| "grad_norm": 0.4433774811549586, |
| "learning_rate": 3.882163505079171e-05, |
| "loss": 0.3961, |
| "step": 246 |
| }, |
| { |
| "epoch": 1.0, |
| "grad_norm": 0.3723223546215645, |
| "learning_rate": 3.880243433489132e-05, |
| "loss": 0.3933, |
| "step": 247 |
| }, |
| { |
| "epoch": 1.0040485829959513, |
| "grad_norm": 0.5119521352859029, |
| "learning_rate": 3.878308327522702e-05, |
| "loss": 0.3667, |
| "step": 248 |
| }, |
| { |
| "epoch": 1.008097165991903, |
| "grad_norm": 0.3710478186270894, |
| "learning_rate": 3.876358202652936e-05, |
| "loss": 0.3545, |
| "step": 249 |
| }, |
| { |
| "epoch": 1.0121457489878543, |
| "grad_norm": 0.4369185190400931, |
| "learning_rate": 3.8743930744729835e-05, |
| "loss": 0.3333, |
| "step": 250 |
| }, |
| { |
| "epoch": 1.0161943319838056, |
| "grad_norm": 0.4470055637135976, |
| "learning_rate": 3.8724129586959535e-05, |
| "loss": 0.32, |
| "step": 251 |
| }, |
| { |
| "epoch": 1.0202429149797572, |
| "grad_norm": 0.451191463682643, |
| "learning_rate": 3.870417871154801e-05, |
| "loss": 0.3648, |
| "step": 252 |
| }, |
| { |
| "epoch": 1.0242914979757085, |
| "grad_norm": 0.5577591151141459, |
| "learning_rate": 3.868407827802192e-05, |
| "loss": 0.3333, |
| "step": 253 |
| }, |
| { |
| "epoch": 1.0283400809716599, |
| "grad_norm": 0.43425105473487524, |
| "learning_rate": 3.8663828447103804e-05, |
| "loss": 0.3437, |
| "step": 254 |
| }, |
| { |
| "epoch": 1.0323886639676114, |
| "grad_norm": 0.5398198906612056, |
| "learning_rate": 3.8643429380710765e-05, |
| "loss": 0.3678, |
| "step": 255 |
| }, |
| { |
| "epoch": 1.0364372469635628, |
| "grad_norm": 0.4445212034317505, |
| "learning_rate": 3.862288124195319e-05, |
| "loss": 0.3569, |
| "step": 256 |
| }, |
| { |
| "epoch": 1.040485829959514, |
| "grad_norm": 0.5275159358381312, |
| "learning_rate": 3.8602184195133464e-05, |
| "loss": 0.346, |
| "step": 257 |
| }, |
| { |
| "epoch": 1.0445344129554657, |
| "grad_norm": 0.43053712431430774, |
| "learning_rate": 3.8581338405744604e-05, |
| "loss": 0.332, |
| "step": 258 |
| }, |
| { |
| "epoch": 1.048582995951417, |
| "grad_norm": 0.5267177234630847, |
| "learning_rate": 3.8560344040468996e-05, |
| "loss": 0.3192, |
| "step": 259 |
| }, |
| { |
| "epoch": 1.0526315789473684, |
| "grad_norm": 0.4848995981836777, |
| "learning_rate": 3.853920126717703e-05, |
| "loss": 0.3356, |
| "step": 260 |
| }, |
| { |
| "epoch": 1.05668016194332, |
| "grad_norm": 0.45620497760191675, |
| "learning_rate": 3.851791025492574e-05, |
| "loss": 0.3325, |
| "step": 261 |
| }, |
| { |
| "epoch": 1.0607287449392713, |
| "grad_norm": 0.5523327843428918, |
| "learning_rate": 3.8496471173957515e-05, |
| "loss": 0.3508, |
| "step": 262 |
| }, |
| { |
| "epoch": 1.0647773279352226, |
| "grad_norm": 0.4147571204253754, |
| "learning_rate": 3.847488419569866e-05, |
| "loss": 0.3443, |
| "step": 263 |
| }, |
| { |
| "epoch": 1.0688259109311742, |
| "grad_norm": 0.5648565057921806, |
| "learning_rate": 3.8453149492758077e-05, |
| "loss": 0.341, |
| "step": 264 |
| }, |
| { |
| "epoch": 1.0728744939271255, |
| "grad_norm": 0.3890179614379519, |
| "learning_rate": 3.8431267238925886e-05, |
| "loss": 0.3434, |
| "step": 265 |
| }, |
| { |
| "epoch": 1.0769230769230769, |
| "grad_norm": 0.4818738399518093, |
| "learning_rate": 3.840923760917199e-05, |
| "loss": 0.3388, |
| "step": 266 |
| }, |
| { |
| "epoch": 1.0809716599190284, |
| "grad_norm": 0.40322648849432424, |
| "learning_rate": 3.838706077964473e-05, |
| "loss": 0.3472, |
| "step": 267 |
| }, |
| { |
| "epoch": 1.0850202429149798, |
| "grad_norm": 0.43383885430158703, |
| "learning_rate": 3.836473692766944e-05, |
| "loss": 0.3251, |
| "step": 268 |
| }, |
| { |
| "epoch": 1.0890688259109311, |
| "grad_norm": 0.48926590601260506, |
| "learning_rate": 3.834226623174705e-05, |
| "loss": 0.3533, |
| "step": 269 |
| }, |
| { |
| "epoch": 1.0931174089068827, |
| "grad_norm": 0.38406293528567614, |
| "learning_rate": 3.831964887155264e-05, |
| "loss": 0.3356, |
| "step": 270 |
| }, |
| { |
| "epoch": 1.097165991902834, |
| "grad_norm": 0.5004323555718425, |
| "learning_rate": 3.829688502793403e-05, |
| "loss": 0.3472, |
| "step": 271 |
| }, |
| { |
| "epoch": 1.1012145748987854, |
| "grad_norm": 0.4852068088318573, |
| "learning_rate": 3.827397488291029e-05, |
| "loss": 0.3509, |
| "step": 272 |
| }, |
| { |
| "epoch": 1.1052631578947367, |
| "grad_norm": 0.41856676902419865, |
| "learning_rate": 3.8250918619670343e-05, |
| "loss": 0.3178, |
| "step": 273 |
| }, |
| { |
| "epoch": 1.1093117408906883, |
| "grad_norm": 0.4980039258653882, |
| "learning_rate": 3.822771642257145e-05, |
| "loss": 0.347, |
| "step": 274 |
| }, |
| { |
| "epoch": 1.1133603238866396, |
| "grad_norm": 0.451797856771414, |
| "learning_rate": 3.820436847713776e-05, |
| "loss": 0.3274, |
| "step": 275 |
| }, |
| { |
| "epoch": 1.117408906882591, |
| "grad_norm": 0.4494968370746643, |
| "learning_rate": 3.8180874970058826e-05, |
| "loss": 0.3403, |
| "step": 276 |
| }, |
| { |
| "epoch": 1.1214574898785425, |
| "grad_norm": 0.5353227228050652, |
| "learning_rate": 3.8157236089188103e-05, |
| "loss": 0.3415, |
| "step": 277 |
| }, |
| { |
| "epoch": 1.125506072874494, |
| "grad_norm": 0.4876862284409634, |
| "learning_rate": 3.813345202354145e-05, |
| "loss": 0.3386, |
| "step": 278 |
| }, |
| { |
| "epoch": 1.1295546558704452, |
| "grad_norm": 0.4088124805593137, |
| "learning_rate": 3.810952296329563e-05, |
| "loss": 0.3281, |
| "step": 279 |
| }, |
| { |
| "epoch": 1.1336032388663968, |
| "grad_norm": 0.4579450206559793, |
| "learning_rate": 3.808544909978676e-05, |
| "loss": 0.3353, |
| "step": 280 |
| }, |
| { |
| "epoch": 1.1376518218623481, |
| "grad_norm": 0.41720943291726104, |
| "learning_rate": 3.806123062550882e-05, |
| "loss": 0.3289, |
| "step": 281 |
| }, |
| { |
| "epoch": 1.1417004048582995, |
| "grad_norm": 0.3995882438793137, |
| "learning_rate": 3.803686773411208e-05, |
| "loss": 0.3338, |
| "step": 282 |
| }, |
| { |
| "epoch": 1.145748987854251, |
| "grad_norm": 0.4900752648415133, |
| "learning_rate": 3.801236062040158e-05, |
| "loss": 0.3506, |
| "step": 283 |
| }, |
| { |
| "epoch": 1.1497975708502024, |
| "grad_norm": 0.42182824506473415, |
| "learning_rate": 3.798770948033553e-05, |
| "loss": 0.3309, |
| "step": 284 |
| }, |
| { |
| "epoch": 1.1538461538461537, |
| "grad_norm": 0.42101779929427946, |
| "learning_rate": 3.796291451102381e-05, |
| "loss": 0.3406, |
| "step": 285 |
| }, |
| { |
| "epoch": 1.1578947368421053, |
| "grad_norm": 0.4600682231483164, |
| "learning_rate": 3.793797591072631e-05, |
| "loss": 0.3374, |
| "step": 286 |
| }, |
| { |
| "epoch": 1.1619433198380567, |
| "grad_norm": 0.4059004453212378, |
| "learning_rate": 3.791289387885144e-05, |
| "loss": 0.331, |
| "step": 287 |
| }, |
| { |
| "epoch": 1.165991902834008, |
| "grad_norm": 0.4197560946997698, |
| "learning_rate": 3.788766861595443e-05, |
| "loss": 0.3335, |
| "step": 288 |
| }, |
| { |
| "epoch": 1.1700404858299596, |
| "grad_norm": 0.3948459616033019, |
| "learning_rate": 3.7862300323735835e-05, |
| "loss": 0.3646, |
| "step": 289 |
| }, |
| { |
| "epoch": 1.174089068825911, |
| "grad_norm": 0.4570794388471147, |
| "learning_rate": 3.783678920503983e-05, |
| "loss": 0.3415, |
| "step": 290 |
| }, |
| { |
| "epoch": 1.1781376518218623, |
| "grad_norm": 0.4339697496092888, |
| "learning_rate": 3.781113546385265e-05, |
| "loss": 0.3442, |
| "step": 291 |
| }, |
| { |
| "epoch": 1.1821862348178138, |
| "grad_norm": 0.46019762279738086, |
| "learning_rate": 3.7785339305300916e-05, |
| "loss": 0.3551, |
| "step": 292 |
| }, |
| { |
| "epoch": 1.1862348178137652, |
| "grad_norm": 0.4944687986708608, |
| "learning_rate": 3.7759400935650025e-05, |
| "loss": 0.3319, |
| "step": 293 |
| }, |
| { |
| "epoch": 1.1902834008097165, |
| "grad_norm": 0.40964667963633683, |
| "learning_rate": 3.77333205623025e-05, |
| "loss": 0.338, |
| "step": 294 |
| }, |
| { |
| "epoch": 1.194331983805668, |
| "grad_norm": 0.4239316469610999, |
| "learning_rate": 3.7707098393796295e-05, |
| "loss": 0.34, |
| "step": 295 |
| }, |
| { |
| "epoch": 1.1983805668016194, |
| "grad_norm": 0.4204471723022606, |
| "learning_rate": 3.7680734639803185e-05, |
| "loss": 0.3389, |
| "step": 296 |
| }, |
| { |
| "epoch": 1.2024291497975708, |
| "grad_norm": 0.39192260649722943, |
| "learning_rate": 3.765422951112704e-05, |
| "loss": 0.3606, |
| "step": 297 |
| }, |
| { |
| "epoch": 1.2064777327935223, |
| "grad_norm": 0.4536637040290487, |
| "learning_rate": 3.762758321970216e-05, |
| "loss": 0.3364, |
| "step": 298 |
| }, |
| { |
| "epoch": 1.2105263157894737, |
| "grad_norm": 0.4570542707240625, |
| "learning_rate": 3.7600795978591584e-05, |
| "loss": 0.3171, |
| "step": 299 |
| }, |
| { |
| "epoch": 1.214574898785425, |
| "grad_norm": 0.3975227588009496, |
| "learning_rate": 3.757386800198538e-05, |
| "loss": 0.3343, |
| "step": 300 |
| }, |
| { |
| "epoch": 1.2186234817813766, |
| "grad_norm": 0.36593395446202714, |
| "learning_rate": 3.7546799505198925e-05, |
| "loss": 0.334, |
| "step": 301 |
| }, |
| { |
| "epoch": 1.222672064777328, |
| "grad_norm": 0.4179855652518417, |
| "learning_rate": 3.7519590704671197e-05, |
| "loss": 0.3384, |
| "step": 302 |
| }, |
| { |
| "epoch": 1.2267206477732793, |
| "grad_norm": 0.3625673996434552, |
| "learning_rate": 3.749224181796305e-05, |
| "loss": 0.3268, |
| "step": 303 |
| }, |
| { |
| "epoch": 1.2307692307692308, |
| "grad_norm": 0.4001547147776107, |
| "learning_rate": 3.746475306375545e-05, |
| "loss": 0.3425, |
| "step": 304 |
| }, |
| { |
| "epoch": 1.2348178137651822, |
| "grad_norm": 0.3681189132616672, |
| "learning_rate": 3.743712466184774e-05, |
| "loss": 0.3257, |
| "step": 305 |
| }, |
| { |
| "epoch": 1.2388663967611335, |
| "grad_norm": 0.3814848037601224, |
| "learning_rate": 3.7409356833155885e-05, |
| "loss": 0.3106, |
| "step": 306 |
| }, |
| { |
| "epoch": 1.242914979757085, |
| "grad_norm": 0.39895730191880735, |
| "learning_rate": 3.73814497997107e-05, |
| "loss": 0.324, |
| "step": 307 |
| }, |
| { |
| "epoch": 1.2469635627530364, |
| "grad_norm": 0.41408057979817947, |
| "learning_rate": 3.7353403784656095e-05, |
| "loss": 0.3331, |
| "step": 308 |
| }, |
| { |
| "epoch": 1.2510121457489878, |
| "grad_norm": 0.3572517717573157, |
| "learning_rate": 3.732521901224724e-05, |
| "loss": 0.3443, |
| "step": 309 |
| }, |
| { |
| "epoch": 1.2550607287449393, |
| "grad_norm": 0.3616619552257627, |
| "learning_rate": 3.7296895707848844e-05, |
| "loss": 0.347, |
| "step": 310 |
| }, |
| { |
| "epoch": 1.2591093117408907, |
| "grad_norm": 0.4328674774250919, |
| "learning_rate": 3.7268434097933275e-05, |
| "loss": 0.346, |
| "step": 311 |
| }, |
| { |
| "epoch": 1.263157894736842, |
| "grad_norm": 0.381963984241296, |
| "learning_rate": 3.723983441007882e-05, |
| "loss": 0.3261, |
| "step": 312 |
| }, |
| { |
| "epoch": 1.2672064777327936, |
| "grad_norm": 0.5204199138245018, |
| "learning_rate": 3.721109687296781e-05, |
| "loss": 0.3447, |
| "step": 313 |
| }, |
| { |
| "epoch": 1.271255060728745, |
| "grad_norm": 0.40362074828826366, |
| "learning_rate": 3.718222171638484e-05, |
| "loss": 0.3322, |
| "step": 314 |
| }, |
| { |
| "epoch": 1.2753036437246963, |
| "grad_norm": 0.4360770844537081, |
| "learning_rate": 3.715320917121488e-05, |
| "loss": 0.3498, |
| "step": 315 |
| }, |
| { |
| "epoch": 1.2793522267206479, |
| "grad_norm": 0.3962129000757534, |
| "learning_rate": 3.712405946944149e-05, |
| "loss": 0.3442, |
| "step": 316 |
| }, |
| { |
| "epoch": 1.2834008097165992, |
| "grad_norm": 0.47753711586676995, |
| "learning_rate": 3.709477284414488e-05, |
| "loss": 0.3296, |
| "step": 317 |
| }, |
| { |
| "epoch": 1.2874493927125505, |
| "grad_norm": 0.4112695645287948, |
| "learning_rate": 3.7065349529500136e-05, |
| "loss": 0.3444, |
| "step": 318 |
| }, |
| { |
| "epoch": 1.291497975708502, |
| "grad_norm": 0.3638309311535185, |
| "learning_rate": 3.7035789760775305e-05, |
| "loss": 0.3436, |
| "step": 319 |
| }, |
| { |
| "epoch": 1.2955465587044535, |
| "grad_norm": 0.40749205031606284, |
| "learning_rate": 3.70060937743295e-05, |
| "loss": 0.3362, |
| "step": 320 |
| }, |
| { |
| "epoch": 1.2995951417004048, |
| "grad_norm": 0.41139954623735575, |
| "learning_rate": 3.6976261807611045e-05, |
| "loss": 0.3367, |
| "step": 321 |
| }, |
| { |
| "epoch": 1.3036437246963564, |
| "grad_norm": 0.36041585990488384, |
| "learning_rate": 3.694629409915555e-05, |
| "loss": 0.3246, |
| "step": 322 |
| }, |
| { |
| "epoch": 1.3076923076923077, |
| "grad_norm": 0.4299856048843735, |
| "learning_rate": 3.691619088858401e-05, |
| "loss": 0.3492, |
| "step": 323 |
| }, |
| { |
| "epoch": 1.311740890688259, |
| "grad_norm": 0.3591671106411351, |
| "learning_rate": 3.68859524166009e-05, |
| "loss": 0.3384, |
| "step": 324 |
| }, |
| { |
| "epoch": 1.3157894736842106, |
| "grad_norm": 0.47678836634751964, |
| "learning_rate": 3.6855578924992226e-05, |
| "loss": 0.3457, |
| "step": 325 |
| }, |
| { |
| "epoch": 1.319838056680162, |
| "grad_norm": 0.3568811453485233, |
| "learning_rate": 3.6825070656623626e-05, |
| "loss": 0.3409, |
| "step": 326 |
| }, |
| { |
| "epoch": 1.3238866396761133, |
| "grad_norm": 0.3715233527384245, |
| "learning_rate": 3.67944278554384e-05, |
| "loss": 0.3406, |
| "step": 327 |
| }, |
| { |
| "epoch": 1.3279352226720649, |
| "grad_norm": 0.34032224063236505, |
| "learning_rate": 3.676365076645557e-05, |
| "loss": 0.3479, |
| "step": 328 |
| }, |
| { |
| "epoch": 1.3319838056680162, |
| "grad_norm": 0.4002209206193757, |
| "learning_rate": 3.673273963576791e-05, |
| "loss": 0.3507, |
| "step": 329 |
| }, |
| { |
| "epoch": 1.3360323886639676, |
| "grad_norm": 0.3442595998924542, |
| "learning_rate": 3.670169471054001e-05, |
| "loss": 0.3218, |
| "step": 330 |
| }, |
| { |
| "epoch": 1.3400809716599191, |
| "grad_norm": 0.4628590850576964, |
| "learning_rate": 3.6670516239006254e-05, |
| "loss": 0.3431, |
| "step": 331 |
| }, |
| { |
| "epoch": 1.3441295546558705, |
| "grad_norm": 0.3690070927512376, |
| "learning_rate": 3.663920447046888e-05, |
| "loss": 0.3417, |
| "step": 332 |
| }, |
| { |
| "epoch": 1.3481781376518218, |
| "grad_norm": 0.37451589843790106, |
| "learning_rate": 3.6607759655295954e-05, |
| "loss": 0.317, |
| "step": 333 |
| }, |
| { |
| "epoch": 1.3522267206477734, |
| "grad_norm": 0.3777469921480051, |
| "learning_rate": 3.657618204491937e-05, |
| "loss": 0.3311, |
| "step": 334 |
| }, |
| { |
| "epoch": 1.3562753036437247, |
| "grad_norm": 0.37427584353555615, |
| "learning_rate": 3.6544471891832875e-05, |
| "loss": 0.3344, |
| "step": 335 |
| }, |
| { |
| "epoch": 1.360323886639676, |
| "grad_norm": 0.34858454401617217, |
| "learning_rate": 3.6512629449589983e-05, |
| "loss": 0.3457, |
| "step": 336 |
| }, |
| { |
| "epoch": 1.3643724696356276, |
| "grad_norm": 0.4126663386490452, |
| "learning_rate": 3.6480654972802045e-05, |
| "loss": 0.348, |
| "step": 337 |
| }, |
| { |
| "epoch": 1.368421052631579, |
| "grad_norm": 0.37569571070581154, |
| "learning_rate": 3.644854871713611e-05, |
| "loss": 0.341, |
| "step": 338 |
| }, |
| { |
| "epoch": 1.3724696356275303, |
| "grad_norm": 0.3966306372622114, |
| "learning_rate": 3.6416310939312924e-05, |
| "loss": 0.3525, |
| "step": 339 |
| }, |
| { |
| "epoch": 1.376518218623482, |
| "grad_norm": 0.3830758251055363, |
| "learning_rate": 3.638394189710493e-05, |
| "loss": 0.3485, |
| "step": 340 |
| }, |
| { |
| "epoch": 1.3805668016194332, |
| "grad_norm": 0.36052107182320253, |
| "learning_rate": 3.635144184933412e-05, |
| "loss": 0.311, |
| "step": 341 |
| }, |
| { |
| "epoch": 1.3846153846153846, |
| "grad_norm": 0.3782850120817355, |
| "learning_rate": 3.6318811055869994e-05, |
| "loss": 0.3341, |
| "step": 342 |
| }, |
| { |
| "epoch": 1.3886639676113361, |
| "grad_norm": 0.3876135985191769, |
| "learning_rate": 3.628604977762752e-05, |
| "loss": 0.3279, |
| "step": 343 |
| }, |
| { |
| "epoch": 1.3927125506072875, |
| "grad_norm": 0.35783056163975885, |
| "learning_rate": 3.6253158276565004e-05, |
| "loss": 0.3355, |
| "step": 344 |
| }, |
| { |
| "epoch": 1.3967611336032388, |
| "grad_norm": 0.4253487282187663, |
| "learning_rate": 3.6220136815682016e-05, |
| "loss": 0.3496, |
| "step": 345 |
| }, |
| { |
| "epoch": 1.4008097165991904, |
| "grad_norm": 0.36749467008509645, |
| "learning_rate": 3.618698565901727e-05, |
| "loss": 0.3375, |
| "step": 346 |
| }, |
| { |
| "epoch": 1.4048582995951417, |
| "grad_norm": 0.4112964234670882, |
| "learning_rate": 3.615370507164654e-05, |
| "loss": 0.3701, |
| "step": 347 |
| }, |
| { |
| "epoch": 1.408906882591093, |
| "grad_norm": 0.40689261004930655, |
| "learning_rate": 3.61202953196805e-05, |
| "loss": 0.3365, |
| "step": 348 |
| }, |
| { |
| "epoch": 1.4129554655870447, |
| "grad_norm": 0.37207572156293334, |
| "learning_rate": 3.6086756670262654e-05, |
| "loss": 0.3426, |
| "step": 349 |
| }, |
| { |
| "epoch": 1.417004048582996, |
| "grad_norm": 0.4201000705942993, |
| "learning_rate": 3.6053089391567135e-05, |
| "loss": 0.3378, |
| "step": 350 |
| }, |
| { |
| "epoch": 1.4210526315789473, |
| "grad_norm": 0.42182226420248203, |
| "learning_rate": 3.601929375279662e-05, |
| "loss": 0.3405, |
| "step": 351 |
| }, |
| { |
| "epoch": 1.425101214574899, |
| "grad_norm": 0.4476578667364011, |
| "learning_rate": 3.598537002418012e-05, |
| "loss": 0.3516, |
| "step": 352 |
| }, |
| { |
| "epoch": 1.4291497975708503, |
| "grad_norm": 0.48752205653785646, |
| "learning_rate": 3.5951318476970885e-05, |
| "loss": 0.3347, |
| "step": 353 |
| }, |
| { |
| "epoch": 1.4331983805668016, |
| "grad_norm": 0.39758034157873806, |
| "learning_rate": 3.591713938344416e-05, |
| "loss": 0.3388, |
| "step": 354 |
| }, |
| { |
| "epoch": 1.4372469635627532, |
| "grad_norm": 0.49251468704968565, |
| "learning_rate": 3.588283301689507e-05, |
| "loss": 0.3419, |
| "step": 355 |
| }, |
| { |
| "epoch": 1.4412955465587045, |
| "grad_norm": 0.34015239188988267, |
| "learning_rate": 3.5848399651636424e-05, |
| "loss": 0.3353, |
| "step": 356 |
| }, |
| { |
| "epoch": 1.4453441295546559, |
| "grad_norm": 0.4928674299157921, |
| "learning_rate": 3.5813839562996473e-05, |
| "loss": 0.3398, |
| "step": 357 |
| }, |
| { |
| "epoch": 1.4493927125506074, |
| "grad_norm": 0.39754888125449533, |
| "learning_rate": 3.5779153027316775e-05, |
| "loss": 0.3465, |
| "step": 358 |
| }, |
| { |
| "epoch": 1.4534412955465588, |
| "grad_norm": 0.49194561065467285, |
| "learning_rate": 3.574434032194994e-05, |
| "loss": 0.3481, |
| "step": 359 |
| }, |
| { |
| "epoch": 1.45748987854251, |
| "grad_norm": 0.46238230499907607, |
| "learning_rate": 3.5709401725257436e-05, |
| "loss": 0.3361, |
| "step": 360 |
| }, |
| { |
| "epoch": 1.4615384615384617, |
| "grad_norm": 0.4040811093685103, |
| "learning_rate": 3.5674337516607346e-05, |
| "loss": 0.3587, |
| "step": 361 |
| }, |
| { |
| "epoch": 1.465587044534413, |
| "grad_norm": 0.4531887246034381, |
| "learning_rate": 3.563914797637216e-05, |
| "loss": 0.3401, |
| "step": 362 |
| }, |
| { |
| "epoch": 1.4696356275303644, |
| "grad_norm": 0.4517838875614873, |
| "learning_rate": 3.560383338592649e-05, |
| "loss": 0.3325, |
| "step": 363 |
| }, |
| { |
| "epoch": 1.4736842105263157, |
| "grad_norm": 0.47793688649657473, |
| "learning_rate": 3.556839402764487e-05, |
| "loss": 0.3413, |
| "step": 364 |
| }, |
| { |
| "epoch": 1.4777327935222673, |
| "grad_norm": 0.429208934789864, |
| "learning_rate": 3.553283018489946e-05, |
| "loss": 0.3316, |
| "step": 365 |
| }, |
| { |
| "epoch": 1.4817813765182186, |
| "grad_norm": 0.4842727759007502, |
| "learning_rate": 3.54971421420578e-05, |
| "loss": 0.3381, |
| "step": 366 |
| }, |
| { |
| "epoch": 1.48582995951417, |
| "grad_norm": 0.5007412496113276, |
| "learning_rate": 3.5461330184480524e-05, |
| "loss": 0.3586, |
| "step": 367 |
| }, |
| { |
| "epoch": 1.4898785425101215, |
| "grad_norm": 0.44077679258173286, |
| "learning_rate": 3.54253945985191e-05, |
| "loss": 0.3458, |
| "step": 368 |
| }, |
| { |
| "epoch": 1.4939271255060729, |
| "grad_norm": 0.42093693200368243, |
| "learning_rate": 3.5389335671513524e-05, |
| "loss": 0.3542, |
| "step": 369 |
| }, |
| { |
| "epoch": 1.4979757085020242, |
| "grad_norm": 0.4456801499339308, |
| "learning_rate": 3.5353153691789986e-05, |
| "loss": 0.3355, |
| "step": 370 |
| }, |
| { |
| "epoch": 1.5020242914979756, |
| "grad_norm": 0.3784356295664434, |
| "learning_rate": 3.531684894865866e-05, |
| "loss": 0.337, |
| "step": 371 |
| }, |
| { |
| "epoch": 1.5060728744939271, |
| "grad_norm": 0.5304860799550661, |
| "learning_rate": 3.528042173241128e-05, |
| "loss": 0.3481, |
| "step": 372 |
| }, |
| { |
| "epoch": 1.5101214574898787, |
| "grad_norm": 0.3487439079568983, |
| "learning_rate": 3.524387233431892e-05, |
| "loss": 0.3369, |
| "step": 373 |
| }, |
| { |
| "epoch": 1.5141700404858298, |
| "grad_norm": 0.44578716170913846, |
| "learning_rate": 3.520720104662958e-05, |
| "loss": 0.3296, |
| "step": 374 |
| }, |
| { |
| "epoch": 1.5182186234817814, |
| "grad_norm": 0.4186487238395273, |
| "learning_rate": 3.5170408162565904e-05, |
| "loss": 0.3547, |
| "step": 375 |
| }, |
| { |
| "epoch": 1.522267206477733, |
| "grad_norm": 0.33084362680734225, |
| "learning_rate": 3.513349397632281e-05, |
| "loss": 0.349, |
| "step": 376 |
| }, |
| { |
| "epoch": 1.526315789473684, |
| "grad_norm": 0.4209994492856671, |
| "learning_rate": 3.5096458783065145e-05, |
| "loss": 0.3593, |
| "step": 377 |
| }, |
| { |
| "epoch": 1.5303643724696356, |
| "grad_norm": 0.39924443827030187, |
| "learning_rate": 3.505930287892533e-05, |
| "loss": 0.3465, |
| "step": 378 |
| }, |
| { |
| "epoch": 1.5344129554655872, |
| "grad_norm": 0.44121996010571285, |
| "learning_rate": 3.502202656100099e-05, |
| "loss": 0.335, |
| "step": 379 |
| }, |
| { |
| "epoch": 1.5384615384615383, |
| "grad_norm": 0.5026150112439011, |
| "learning_rate": 3.498463012735256e-05, |
| "loss": 0.3477, |
| "step": 380 |
| }, |
| { |
| "epoch": 1.54251012145749, |
| "grad_norm": 0.4113570944485569, |
| "learning_rate": 3.494711387700094e-05, |
| "loss": 0.3348, |
| "step": 381 |
| }, |
| { |
| "epoch": 1.5465587044534415, |
| "grad_norm": 0.5575816430800176, |
| "learning_rate": 3.490947810992505e-05, |
| "loss": 0.3457, |
| "step": 382 |
| }, |
| { |
| "epoch": 1.5506072874493926, |
| "grad_norm": 0.3874692873387124, |
| "learning_rate": 3.48717231270595e-05, |
| "loss": 0.3647, |
| "step": 383 |
| }, |
| { |
| "epoch": 1.5546558704453441, |
| "grad_norm": 0.41039970705940176, |
| "learning_rate": 3.483384923029211e-05, |
| "loss": 0.342, |
| "step": 384 |
| }, |
| { |
| "epoch": 1.5587044534412957, |
| "grad_norm": 0.3538556937132035, |
| "learning_rate": 3.479585672246156e-05, |
| "loss": 0.3424, |
| "step": 385 |
| }, |
| { |
| "epoch": 1.5627530364372468, |
| "grad_norm": 0.3824651013779483, |
| "learning_rate": 3.4757745907354924e-05, |
| "loss": 0.3512, |
| "step": 386 |
| }, |
| { |
| "epoch": 1.5668016194331984, |
| "grad_norm": 0.3759720189914231, |
| "learning_rate": 3.471951708970527e-05, |
| "loss": 0.335, |
| "step": 387 |
| }, |
| { |
| "epoch": 1.5708502024291497, |
| "grad_norm": 0.40224769845670594, |
| "learning_rate": 3.468117057518921e-05, |
| "loss": 0.3403, |
| "step": 388 |
| }, |
| { |
| "epoch": 1.574898785425101, |
| "grad_norm": 0.3669743204925936, |
| "learning_rate": 3.4642706670424426e-05, |
| "loss": 0.3398, |
| "step": 389 |
| }, |
| { |
| "epoch": 1.5789473684210527, |
| "grad_norm": 0.3701065300900419, |
| "learning_rate": 3.460412568296731e-05, |
| "loss": 0.3478, |
| "step": 390 |
| }, |
| { |
| "epoch": 1.582995951417004, |
| "grad_norm": 0.3659128348043706, |
| "learning_rate": 3.456542792131039e-05, |
| "loss": 0.333, |
| "step": 391 |
| }, |
| { |
| "epoch": 1.5870445344129553, |
| "grad_norm": 0.4704530360038611, |
| "learning_rate": 3.4526613694879936e-05, |
| "loss": 0.3462, |
| "step": 392 |
| }, |
| { |
| "epoch": 1.591093117408907, |
| "grad_norm": 0.35560428334619354, |
| "learning_rate": 3.448768331403348e-05, |
| "loss": 0.3584, |
| "step": 393 |
| }, |
| { |
| "epoch": 1.5951417004048583, |
| "grad_norm": 0.3896220868632546, |
| "learning_rate": 3.44486370900573e-05, |
| "loss": 0.3511, |
| "step": 394 |
| }, |
| { |
| "epoch": 1.5991902834008096, |
| "grad_norm": 0.38412851167861395, |
| "learning_rate": 3.440947533516396e-05, |
| "loss": 0.3554, |
| "step": 395 |
| }, |
| { |
| "epoch": 1.6032388663967612, |
| "grad_norm": 0.37762736461111096, |
| "learning_rate": 3.4370198362489816e-05, |
| "loss": 0.3361, |
| "step": 396 |
| }, |
| { |
| "epoch": 1.6072874493927125, |
| "grad_norm": 0.37623723803484876, |
| "learning_rate": 3.43308064860925e-05, |
| "loss": 0.3361, |
| "step": 397 |
| }, |
| { |
| "epoch": 1.6113360323886639, |
| "grad_norm": 0.36959308305211397, |
| "learning_rate": 3.4291300020948393e-05, |
| "loss": 0.3214, |
| "step": 398 |
| }, |
| { |
| "epoch": 1.6153846153846154, |
| "grad_norm": 0.38346423497660365, |
| "learning_rate": 3.425167928295015e-05, |
| "loss": 0.3405, |
| "step": 399 |
| }, |
| { |
| "epoch": 1.6194331983805668, |
| "grad_norm": 0.41893489311824034, |
| "learning_rate": 3.421194458890411e-05, |
| "loss": 0.352, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.623481781376518, |
| "grad_norm": 0.3265140555337351, |
| "learning_rate": 3.4172096256527865e-05, |
| "loss": 0.3475, |
| "step": 401 |
| }, |
| { |
| "epoch": 1.6275303643724697, |
| "grad_norm": 0.4117361344836061, |
| "learning_rate": 3.4132134604447595e-05, |
| "loss": 0.3488, |
| "step": 402 |
| }, |
| { |
| "epoch": 1.631578947368421, |
| "grad_norm": 0.33846518884806087, |
| "learning_rate": 3.409205995219561e-05, |
| "loss": 0.3264, |
| "step": 403 |
| }, |
| { |
| "epoch": 1.6356275303643724, |
| "grad_norm": 0.36569326537532454, |
| "learning_rate": 3.4051872620207765e-05, |
| "loss": 0.3415, |
| "step": 404 |
| }, |
| { |
| "epoch": 1.639676113360324, |
| "grad_norm": 0.3482372901791403, |
| "learning_rate": 3.40115729298209e-05, |
| "loss": 0.3384, |
| "step": 405 |
| }, |
| { |
| "epoch": 1.6437246963562753, |
| "grad_norm": 0.3986540019731291, |
| "learning_rate": 3.397116120327025e-05, |
| "loss": 0.3482, |
| "step": 406 |
| }, |
| { |
| "epoch": 1.6477732793522266, |
| "grad_norm": 0.3383340210947966, |
| "learning_rate": 3.3930637763686916e-05, |
| "loss": 0.3239, |
| "step": 407 |
| }, |
| { |
| "epoch": 1.6518218623481782, |
| "grad_norm": 0.3502712856105517, |
| "learning_rate": 3.389000293509524e-05, |
| "loss": 0.3442, |
| "step": 408 |
| }, |
| { |
| "epoch": 1.6558704453441295, |
| "grad_norm": 0.36142116793975143, |
| "learning_rate": 3.3849257042410225e-05, |
| "loss": 0.3172, |
| "step": 409 |
| }, |
| { |
| "epoch": 1.6599190283400809, |
| "grad_norm": 0.3454541321330931, |
| "learning_rate": 3.380840041143494e-05, |
| "loss": 0.3551, |
| "step": 410 |
| }, |
| { |
| "epoch": 1.6639676113360324, |
| "grad_norm": 0.3862648211470358, |
| "learning_rate": 3.376743336885792e-05, |
| "loss": 0.3487, |
| "step": 411 |
| }, |
| { |
| "epoch": 1.6680161943319838, |
| "grad_norm": 0.32820083387798515, |
| "learning_rate": 3.372635624225055e-05, |
| "loss": 0.3381, |
| "step": 412 |
| }, |
| { |
| "epoch": 1.6720647773279351, |
| "grad_norm": 0.3626558751301146, |
| "learning_rate": 3.3685169360064445e-05, |
| "loss": 0.3389, |
| "step": 413 |
| }, |
| { |
| "epoch": 1.6761133603238867, |
| "grad_norm": 0.38541952184586176, |
| "learning_rate": 3.3643873051628804e-05, |
| "loss": 0.3578, |
| "step": 414 |
| }, |
| { |
| "epoch": 1.680161943319838, |
| "grad_norm": 0.3459134174129889, |
| "learning_rate": 3.3602467647147814e-05, |
| "loss": 0.3228, |
| "step": 415 |
| }, |
| { |
| "epoch": 1.6842105263157894, |
| "grad_norm": 0.3989842924668146, |
| "learning_rate": 3.356095347769799e-05, |
| "loss": 0.3437, |
| "step": 416 |
| }, |
| { |
| "epoch": 1.688259109311741, |
| "grad_norm": 0.40245941554853176, |
| "learning_rate": 3.351933087522552e-05, |
| "loss": 0.3422, |
| "step": 417 |
| }, |
| { |
| "epoch": 1.6923076923076923, |
| "grad_norm": 0.41594148337084924, |
| "learning_rate": 3.3477600172543626e-05, |
| "loss": 0.3349, |
| "step": 418 |
| }, |
| { |
| "epoch": 1.6963562753036436, |
| "grad_norm": 0.4540518899498624, |
| "learning_rate": 3.3435761703329894e-05, |
| "loss": 0.3468, |
| "step": 419 |
| }, |
| { |
| "epoch": 1.7004048582995952, |
| "grad_norm": 0.4285802369115759, |
| "learning_rate": 3.3393815802123606e-05, |
| "loss": 0.3351, |
| "step": 420 |
| }, |
| { |
| "epoch": 1.7044534412955465, |
| "grad_norm": 0.48110300298834313, |
| "learning_rate": 3.335176280432308e-05, |
| "loss": 0.342, |
| "step": 421 |
| }, |
| { |
| "epoch": 1.708502024291498, |
| "grad_norm": 0.3833411554862261, |
| "learning_rate": 3.330960304618295e-05, |
| "loss": 0.3355, |
| "step": 422 |
| }, |
| { |
| "epoch": 1.7125506072874495, |
| "grad_norm": 0.46017371936551715, |
| "learning_rate": 3.326733686481153e-05, |
| "loss": 0.3216, |
| "step": 423 |
| }, |
| { |
| "epoch": 1.7165991902834008, |
| "grad_norm": 0.34455005386176163, |
| "learning_rate": 3.3224964598168084e-05, |
| "loss": 0.336, |
| "step": 424 |
| }, |
| { |
| "epoch": 1.7206477732793521, |
| "grad_norm": 0.3756779320429783, |
| "learning_rate": 3.318248658506012e-05, |
| "loss": 0.348, |
| "step": 425 |
| }, |
| { |
| "epoch": 1.7246963562753037, |
| "grad_norm": 0.42748768047706476, |
| "learning_rate": 3.313990316514069e-05, |
| "loss": 0.3559, |
| "step": 426 |
| }, |
| { |
| "epoch": 1.728744939271255, |
| "grad_norm": 0.3191676621298771, |
| "learning_rate": 3.309721467890571e-05, |
| "loss": 0.3513, |
| "step": 427 |
| }, |
| { |
| "epoch": 1.7327935222672064, |
| "grad_norm": 0.39827807659468634, |
| "learning_rate": 3.305442146769115e-05, |
| "loss": 0.3409, |
| "step": 428 |
| }, |
| { |
| "epoch": 1.736842105263158, |
| "grad_norm": 0.36439859712168154, |
| "learning_rate": 3.30115238736704e-05, |
| "loss": 0.3374, |
| "step": 429 |
| }, |
| { |
| "epoch": 1.7408906882591093, |
| "grad_norm": 0.40374784455728235, |
| "learning_rate": 3.296852223985148e-05, |
| "loss": 0.3364, |
| "step": 430 |
| }, |
| { |
| "epoch": 1.7449392712550607, |
| "grad_norm": 0.38291433884980747, |
| "learning_rate": 3.2925416910074315e-05, |
| "loss": 0.3462, |
| "step": 431 |
| }, |
| { |
| "epoch": 1.7489878542510122, |
| "grad_norm": 0.35417030440753566, |
| "learning_rate": 3.288220822900796e-05, |
| "loss": 0.3535, |
| "step": 432 |
| }, |
| { |
| "epoch": 1.7530364372469636, |
| "grad_norm": 0.38597550616186854, |
| "learning_rate": 3.283889654214788e-05, |
| "loss": 0.3183, |
| "step": 433 |
| }, |
| { |
| "epoch": 1.757085020242915, |
| "grad_norm": 0.3252766937057128, |
| "learning_rate": 3.279548219581317e-05, |
| "loss": 0.321, |
| "step": 434 |
| }, |
| { |
| "epoch": 1.7611336032388665, |
| "grad_norm": 0.423583797798488, |
| "learning_rate": 3.275196553714379e-05, |
| "loss": 0.3384, |
| "step": 435 |
| }, |
| { |
| "epoch": 1.7651821862348178, |
| "grad_norm": 0.3286003127569554, |
| "learning_rate": 3.270834691409778e-05, |
| "loss": 0.35, |
| "step": 436 |
| }, |
| { |
| "epoch": 1.7692307692307692, |
| "grad_norm": 0.4337524070843203, |
| "learning_rate": 3.266462667544849e-05, |
| "loss": 0.3316, |
| "step": 437 |
| }, |
| { |
| "epoch": 1.7732793522267207, |
| "grad_norm": 0.3825993068899728, |
| "learning_rate": 3.262080517078178e-05, |
| "loss": 0.3395, |
| "step": 438 |
| }, |
| { |
| "epoch": 1.777327935222672, |
| "grad_norm": 0.3500129751575795, |
| "learning_rate": 3.257688275049323e-05, |
| "loss": 0.3368, |
| "step": 439 |
| }, |
| { |
| "epoch": 1.7813765182186234, |
| "grad_norm": 0.40389637477211676, |
| "learning_rate": 3.253285976578535e-05, |
| "loss": 0.36, |
| "step": 440 |
| }, |
| { |
| "epoch": 1.785425101214575, |
| "grad_norm": 0.34683632412074455, |
| "learning_rate": 3.2488736568664756e-05, |
| "loss": 0.344, |
| "step": 441 |
| }, |
| { |
| "epoch": 1.7894736842105263, |
| "grad_norm": 0.3562367996433747, |
| "learning_rate": 3.244451351193935e-05, |
| "loss": 0.3445, |
| "step": 442 |
| }, |
| { |
| "epoch": 1.7935222672064777, |
| "grad_norm": 0.3902693143484903, |
| "learning_rate": 3.240019094921551e-05, |
| "loss": 0.3417, |
| "step": 443 |
| }, |
| { |
| "epoch": 1.7975708502024292, |
| "grad_norm": 0.4242531182909229, |
| "learning_rate": 3.235576923489527e-05, |
| "loss": 0.3454, |
| "step": 444 |
| }, |
| { |
| "epoch": 1.8016194331983806, |
| "grad_norm": 0.3765831294277979, |
| "learning_rate": 3.2311248724173476e-05, |
| "loss": 0.3488, |
| "step": 445 |
| }, |
| { |
| "epoch": 1.805668016194332, |
| "grad_norm": 0.4292773644141255, |
| "learning_rate": 3.226662977303494e-05, |
| "loss": 0.344, |
| "step": 446 |
| }, |
| { |
| "epoch": 1.8097165991902835, |
| "grad_norm": 0.39986101900533577, |
| "learning_rate": 3.2221912738251626e-05, |
| "loss": 0.3475, |
| "step": 447 |
| }, |
| { |
| "epoch": 1.8137651821862348, |
| "grad_norm": 0.3474469533584368, |
| "learning_rate": 3.2177097977379736e-05, |
| "loss": 0.3418, |
| "step": 448 |
| }, |
| { |
| "epoch": 1.8178137651821862, |
| "grad_norm": 0.3771099552336464, |
| "learning_rate": 3.21321858487569e-05, |
| "loss": 0.3448, |
| "step": 449 |
| }, |
| { |
| "epoch": 1.8218623481781377, |
| "grad_norm": 0.31954396571251253, |
| "learning_rate": 3.208717671149931e-05, |
| "loss": 0.3444, |
| "step": 450 |
| }, |
| { |
| "epoch": 1.825910931174089, |
| "grad_norm": 0.34963278929060626, |
| "learning_rate": 3.204207092549882e-05, |
| "loss": 0.3427, |
| "step": 451 |
| }, |
| { |
| "epoch": 1.8299595141700404, |
| "grad_norm": 0.3668734022912011, |
| "learning_rate": 3.19968688514201e-05, |
| "loss": 0.3392, |
| "step": 452 |
| }, |
| { |
| "epoch": 1.834008097165992, |
| "grad_norm": 0.3302949422921762, |
| "learning_rate": 3.195157085069774e-05, |
| "loss": 0.3471, |
| "step": 453 |
| }, |
| { |
| "epoch": 1.8380566801619433, |
| "grad_norm": 0.35322218255173154, |
| "learning_rate": 3.190617728553332e-05, |
| "loss": 0.3316, |
| "step": 454 |
| }, |
| { |
| "epoch": 1.8421052631578947, |
| "grad_norm": 0.38725090147123564, |
| "learning_rate": 3.1860688518892606e-05, |
| "loss": 0.3257, |
| "step": 455 |
| }, |
| { |
| "epoch": 1.8461538461538463, |
| "grad_norm": 0.32650811333726076, |
| "learning_rate": 3.181510491450254e-05, |
| "loss": 0.3605, |
| "step": 456 |
| }, |
| { |
| "epoch": 1.8502024291497976, |
| "grad_norm": 0.36509441076873944, |
| "learning_rate": 3.176942683684842e-05, |
| "loss": 0.3354, |
| "step": 457 |
| }, |
| { |
| "epoch": 1.854251012145749, |
| "grad_norm": 0.37043829509931014, |
| "learning_rate": 3.1723654651170934e-05, |
| "loss": 0.3448, |
| "step": 458 |
| }, |
| { |
| "epoch": 1.8582995951417005, |
| "grad_norm": 0.3588136016973412, |
| "learning_rate": 3.167778872346327e-05, |
| "loss": 0.3509, |
| "step": 459 |
| }, |
| { |
| "epoch": 1.8623481781376519, |
| "grad_norm": 0.3745673475700913, |
| "learning_rate": 3.163182942046815e-05, |
| "loss": 0.3293, |
| "step": 460 |
| }, |
| { |
| "epoch": 1.8663967611336032, |
| "grad_norm": 0.33843087345966616, |
| "learning_rate": 3.158577710967494e-05, |
| "loss": 0.3334, |
| "step": 461 |
| }, |
| { |
| "epoch": 1.8704453441295548, |
| "grad_norm": 0.4070006574566823, |
| "learning_rate": 3.153963215931669e-05, |
| "loss": 0.3223, |
| "step": 462 |
| }, |
| { |
| "epoch": 1.874493927125506, |
| "grad_norm": 0.3240138753972508, |
| "learning_rate": 3.14933949383672e-05, |
| "loss": 0.3252, |
| "step": 463 |
| }, |
| { |
| "epoch": 1.8785425101214575, |
| "grad_norm": 0.3970892900032312, |
| "learning_rate": 3.144706581653805e-05, |
| "loss": 0.3353, |
| "step": 464 |
| }, |
| { |
| "epoch": 1.882591093117409, |
| "grad_norm": 0.3773253355489966, |
| "learning_rate": 3.140064516427566e-05, |
| "loss": 0.3387, |
| "step": 465 |
| }, |
| { |
| "epoch": 1.8866396761133604, |
| "grad_norm": 0.34918281027356807, |
| "learning_rate": 3.135413335275833e-05, |
| "loss": 0.3293, |
| "step": 466 |
| }, |
| { |
| "epoch": 1.8906882591093117, |
| "grad_norm": 0.39597086622392014, |
| "learning_rate": 3.130753075389327e-05, |
| "loss": 0.3623, |
| "step": 467 |
| }, |
| { |
| "epoch": 1.8947368421052633, |
| "grad_norm": 0.3893497110986002, |
| "learning_rate": 3.1260837740313625e-05, |
| "loss": 0.3505, |
| "step": 468 |
| }, |
| { |
| "epoch": 1.8987854251012146, |
| "grad_norm": 0.3931254762200784, |
| "learning_rate": 3.121405468537547e-05, |
| "loss": 0.3363, |
| "step": 469 |
| }, |
| { |
| "epoch": 1.902834008097166, |
| "grad_norm": 0.38861436163266583, |
| "learning_rate": 3.116718196315488e-05, |
| "loss": 0.3277, |
| "step": 470 |
| }, |
| { |
| "epoch": 1.9068825910931175, |
| "grad_norm": 0.426366854594213, |
| "learning_rate": 3.1120219948444894e-05, |
| "loss": 0.3243, |
| "step": 471 |
| }, |
| { |
| "epoch": 1.9109311740890689, |
| "grad_norm": 0.3943809840492929, |
| "learning_rate": 3.107316901675252e-05, |
| "loss": 0.3313, |
| "step": 472 |
| }, |
| { |
| "epoch": 1.9149797570850202, |
| "grad_norm": 0.39314277638408024, |
| "learning_rate": 3.1026029544295745e-05, |
| "loss": 0.3366, |
| "step": 473 |
| }, |
| { |
| "epoch": 1.9190283400809718, |
| "grad_norm": 0.4042995741791517, |
| "learning_rate": 3.097880190800057e-05, |
| "loss": 0.3464, |
| "step": 474 |
| }, |
| { |
| "epoch": 1.9230769230769231, |
| "grad_norm": 0.4245820924762205, |
| "learning_rate": 3.093148648549788e-05, |
| "loss": 0.3463, |
| "step": 475 |
| }, |
| { |
| "epoch": 1.9271255060728745, |
| "grad_norm": 0.35665524572296525, |
| "learning_rate": 3.088408365512055e-05, |
| "loss": 0.3286, |
| "step": 476 |
| }, |
| { |
| "epoch": 1.931174089068826, |
| "grad_norm": 0.4047257660060213, |
| "learning_rate": 3.083659379590034e-05, |
| "loss": 0.3318, |
| "step": 477 |
| }, |
| { |
| "epoch": 1.9352226720647774, |
| "grad_norm": 0.34270067781602614, |
| "learning_rate": 3.0789017287564913e-05, |
| "loss": 0.3279, |
| "step": 478 |
| }, |
| { |
| "epoch": 1.9392712550607287, |
| "grad_norm": 0.36514628241661995, |
| "learning_rate": 3.074135451053475e-05, |
| "loss": 0.3449, |
| "step": 479 |
| }, |
| { |
| "epoch": 1.9433198380566803, |
| "grad_norm": 0.3435757111538838, |
| "learning_rate": 3.069360584592015e-05, |
| "loss": 0.3428, |
| "step": 480 |
| }, |
| { |
| "epoch": 1.9473684210526314, |
| "grad_norm": 0.32029259758019496, |
| "learning_rate": 3.064577167551816e-05, |
| "loss": 0.3504, |
| "step": 481 |
| }, |
| { |
| "epoch": 1.951417004048583, |
| "grad_norm": 0.3837467665944728, |
| "learning_rate": 3.059785238180953e-05, |
| "loss": 0.3527, |
| "step": 482 |
| }, |
| { |
| "epoch": 1.9554655870445345, |
| "grad_norm": 0.32884267551900104, |
| "learning_rate": 3.054984834795566e-05, |
| "loss": 0.3287, |
| "step": 483 |
| }, |
| { |
| "epoch": 1.9595141700404857, |
| "grad_norm": 0.36115073912422113, |
| "learning_rate": 3.0501759957795537e-05, |
| "loss": 0.3322, |
| "step": 484 |
| }, |
| { |
| "epoch": 1.9635627530364372, |
| "grad_norm": 0.31697058000910405, |
| "learning_rate": 3.0453587595842627e-05, |
| "loss": 0.3528, |
| "step": 485 |
| }, |
| { |
| "epoch": 1.9676113360323888, |
| "grad_norm": 0.32964742255582746, |
| "learning_rate": 3.0405331647281856e-05, |
| "loss": 0.3327, |
| "step": 486 |
| }, |
| { |
| "epoch": 1.97165991902834, |
| "grad_norm": 0.3248699196154423, |
| "learning_rate": 3.0356992497966508e-05, |
| "loss": 0.3322, |
| "step": 487 |
| }, |
| { |
| "epoch": 1.9757085020242915, |
| "grad_norm": 0.34282473210793235, |
| "learning_rate": 3.0308570534415137e-05, |
| "loss": 0.3245, |
| "step": 488 |
| }, |
| { |
| "epoch": 1.979757085020243, |
| "grad_norm": 0.34471348072092883, |
| "learning_rate": 3.0260066143808467e-05, |
| "loss": 0.3534, |
| "step": 489 |
| }, |
| { |
| "epoch": 1.9838056680161942, |
| "grad_norm": 0.3577170730106755, |
| "learning_rate": 3.021147971398632e-05, |
| "loss": 0.3239, |
| "step": 490 |
| }, |
| { |
| "epoch": 1.9878542510121457, |
| "grad_norm": 0.36079245475042154, |
| "learning_rate": 3.0162811633444478e-05, |
| "loss": 0.3395, |
| "step": 491 |
| }, |
| { |
| "epoch": 1.9919028340080973, |
| "grad_norm": 0.35580878899413304, |
| "learning_rate": 3.0114062291331635e-05, |
| "loss": 0.326, |
| "step": 492 |
| }, |
| { |
| "epoch": 1.9959514170040484, |
| "grad_norm": 0.3362093445695427, |
| "learning_rate": 3.0065232077446217e-05, |
| "loss": 0.3393, |
| "step": 493 |
| }, |
| { |
| "epoch": 2.0, |
| "grad_norm": 0.3842665208343462, |
| "learning_rate": 3.001632138223332e-05, |
| "loss": 0.3401, |
| "step": 494 |
| }, |
| { |
| "epoch": 2.0040485829959516, |
| "grad_norm": 0.4208703019041765, |
| "learning_rate": 2.9967330596781565e-05, |
| "loss": 0.2685, |
| "step": 495 |
| }, |
| { |
| "epoch": 2.0080971659919027, |
| "grad_norm": 0.38500904010100784, |
| "learning_rate": 2.9918260112819952e-05, |
| "loss": 0.255, |
| "step": 496 |
| }, |
| { |
| "epoch": 2.0121457489878543, |
| "grad_norm": 0.6124878266547644, |
| "learning_rate": 2.9869110322714778e-05, |
| "loss": 0.2428, |
| "step": 497 |
| }, |
| { |
| "epoch": 2.016194331983806, |
| "grad_norm": 0.43337534470300054, |
| "learning_rate": 2.9819881619466447e-05, |
| "loss": 0.255, |
| "step": 498 |
| }, |
| { |
| "epoch": 2.020242914979757, |
| "grad_norm": 0.45262748581023354, |
| "learning_rate": 2.9770574396706364e-05, |
| "loss": 0.241, |
| "step": 499 |
| }, |
| { |
| "epoch": 2.0242914979757085, |
| "grad_norm": 0.41880371521080534, |
| "learning_rate": 2.972118904869376e-05, |
| "loss": 0.2791, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.02834008097166, |
| "grad_norm": 0.39373543277858414, |
| "learning_rate": 2.9671725970312563e-05, |
| "loss": 0.2456, |
| "step": 501 |
| }, |
| { |
| "epoch": 2.032388663967611, |
| "grad_norm": 0.39910297929988214, |
| "learning_rate": 2.9622185557068238e-05, |
| "loss": 0.2548, |
| "step": 502 |
| }, |
| { |
| "epoch": 2.0364372469635628, |
| "grad_norm": 0.4403611039033112, |
| "learning_rate": 2.95725682050846e-05, |
| "loss": 0.2759, |
| "step": 503 |
| }, |
| { |
| "epoch": 2.0404858299595143, |
| "grad_norm": 0.3549254619901999, |
| "learning_rate": 2.9522874311100684e-05, |
| "loss": 0.2603, |
| "step": 504 |
| }, |
| { |
| "epoch": 2.0445344129554655, |
| "grad_norm": 0.35384708042008184, |
| "learning_rate": 2.9473104272467526e-05, |
| "loss": 0.2419, |
| "step": 505 |
| }, |
| { |
| "epoch": 2.048582995951417, |
| "grad_norm": 0.3849797053739274, |
| "learning_rate": 2.942325848714505e-05, |
| "loss": 0.2498, |
| "step": 506 |
| }, |
| { |
| "epoch": 2.0526315789473686, |
| "grad_norm": 0.372189272193299, |
| "learning_rate": 2.9373337353698804e-05, |
| "loss": 0.2493, |
| "step": 507 |
| }, |
| { |
| "epoch": 2.0566801619433197, |
| "grad_norm": 0.382226683101649, |
| "learning_rate": 2.932334127129686e-05, |
| "loss": 0.2493, |
| "step": 508 |
| }, |
| { |
| "epoch": 2.0607287449392713, |
| "grad_norm": 0.40478236927623656, |
| "learning_rate": 2.9273270639706547e-05, |
| "loss": 0.2581, |
| "step": 509 |
| }, |
| { |
| "epoch": 2.064777327935223, |
| "grad_norm": 0.3403554268753071, |
| "learning_rate": 2.922312585929131e-05, |
| "loss": 0.2569, |
| "step": 510 |
| }, |
| { |
| "epoch": 2.068825910931174, |
| "grad_norm": 0.3810943830077671, |
| "learning_rate": 2.9172907331007478e-05, |
| "loss": 0.2463, |
| "step": 511 |
| }, |
| { |
| "epoch": 2.0728744939271255, |
| "grad_norm": 0.3604183976340681, |
| "learning_rate": 2.9122615456401053e-05, |
| "loss": 0.2613, |
| "step": 512 |
| }, |
| { |
| "epoch": 2.076923076923077, |
| "grad_norm": 0.32124322980167525, |
| "learning_rate": 2.9072250637604535e-05, |
| "loss": 0.2389, |
| "step": 513 |
| }, |
| { |
| "epoch": 2.080971659919028, |
| "grad_norm": 0.3456024627878585, |
| "learning_rate": 2.9021813277333667e-05, |
| "loss": 0.2501, |
| "step": 514 |
| }, |
| { |
| "epoch": 2.08502024291498, |
| "grad_norm": 0.33763336215312284, |
| "learning_rate": 2.8971303778884244e-05, |
| "loss": 0.2647, |
| "step": 515 |
| }, |
| { |
| "epoch": 2.0890688259109313, |
| "grad_norm": 0.3043469824186136, |
| "learning_rate": 2.8920722546128863e-05, |
| "loss": 0.2412, |
| "step": 516 |
| }, |
| { |
| "epoch": 2.0931174089068825, |
| "grad_norm": 0.36805060467787143, |
| "learning_rate": 2.887006998351372e-05, |
| "loss": 0.2541, |
| "step": 517 |
| }, |
| { |
| "epoch": 2.097165991902834, |
| "grad_norm": 0.3170146884781653, |
| "learning_rate": 2.881934649605535e-05, |
| "loss": 0.2494, |
| "step": 518 |
| }, |
| { |
| "epoch": 2.1012145748987856, |
| "grad_norm": 0.319280639421291, |
| "learning_rate": 2.8768552489337408e-05, |
| "loss": 0.2779, |
| "step": 519 |
| }, |
| { |
| "epoch": 2.1052631578947367, |
| "grad_norm": 0.3505954045536158, |
| "learning_rate": 2.8717688369507425e-05, |
| "loss": 0.2394, |
| "step": 520 |
| }, |
| { |
| "epoch": 2.1093117408906883, |
| "grad_norm": 0.3203268001366862, |
| "learning_rate": 2.8666754543273535e-05, |
| "loss": 0.2565, |
| "step": 521 |
| }, |
| { |
| "epoch": 2.11336032388664, |
| "grad_norm": 0.33243559363195624, |
| "learning_rate": 2.8615751417901273e-05, |
| "loss": 0.251, |
| "step": 522 |
| }, |
| { |
| "epoch": 2.117408906882591, |
| "grad_norm": 0.3528565198641662, |
| "learning_rate": 2.856467940121025e-05, |
| "loss": 0.2429, |
| "step": 523 |
| }, |
| { |
| "epoch": 2.1214574898785425, |
| "grad_norm": 0.3395323796238842, |
| "learning_rate": 2.8513538901570964e-05, |
| "loss": 0.2614, |
| "step": 524 |
| }, |
| { |
| "epoch": 2.125506072874494, |
| "grad_norm": 0.33311398286810134, |
| "learning_rate": 2.846233032790149e-05, |
| "loss": 0.2557, |
| "step": 525 |
| }, |
| { |
| "epoch": 2.1295546558704452, |
| "grad_norm": 0.3606430332179728, |
| "learning_rate": 2.84110540896642e-05, |
| "loss": 0.2543, |
| "step": 526 |
| }, |
| { |
| "epoch": 2.133603238866397, |
| "grad_norm": 0.35909497634328996, |
| "learning_rate": 2.8359710596862555e-05, |
| "loss": 0.2535, |
| "step": 527 |
| }, |
| { |
| "epoch": 2.1376518218623484, |
| "grad_norm": 0.3882996449906683, |
| "learning_rate": 2.8308300260037734e-05, |
| "loss": 0.266, |
| "step": 528 |
| }, |
| { |
| "epoch": 2.1417004048582995, |
| "grad_norm": 0.36623484261163447, |
| "learning_rate": 2.8256823490265424e-05, |
| "loss": 0.2424, |
| "step": 529 |
| }, |
| { |
| "epoch": 2.145748987854251, |
| "grad_norm": 0.4049955693330238, |
| "learning_rate": 2.8205280699152515e-05, |
| "loss": 0.2464, |
| "step": 530 |
| }, |
| { |
| "epoch": 2.1497975708502026, |
| "grad_norm": 0.34765831106104433, |
| "learning_rate": 2.8153672298833777e-05, |
| "loss": 0.2585, |
| "step": 531 |
| }, |
| { |
| "epoch": 2.1538461538461537, |
| "grad_norm": 0.38078192914096026, |
| "learning_rate": 2.8101998701968615e-05, |
| "loss": 0.2491, |
| "step": 532 |
| }, |
| { |
| "epoch": 2.1578947368421053, |
| "grad_norm": 0.3330552779298403, |
| "learning_rate": 2.805026032173773e-05, |
| "loss": 0.2673, |
| "step": 533 |
| }, |
| { |
| "epoch": 2.161943319838057, |
| "grad_norm": 0.3322566740023325, |
| "learning_rate": 2.7998457571839832e-05, |
| "loss": 0.256, |
| "step": 534 |
| }, |
| { |
| "epoch": 2.165991902834008, |
| "grad_norm": 0.3192040188194537, |
| "learning_rate": 2.7946590866488322e-05, |
| "loss": 0.2537, |
| "step": 535 |
| }, |
| { |
| "epoch": 2.1700404858299596, |
| "grad_norm": 0.29027534066376764, |
| "learning_rate": 2.7894660620408e-05, |
| "loss": 0.2828, |
| "step": 536 |
| }, |
| { |
| "epoch": 2.174089068825911, |
| "grad_norm": 0.33865754000875153, |
| "learning_rate": 2.784266724883173e-05, |
| "loss": 0.2425, |
| "step": 537 |
| }, |
| { |
| "epoch": 2.1781376518218623, |
| "grad_norm": 0.3152971734341131, |
| "learning_rate": 2.7790611167497128e-05, |
| "loss": 0.2538, |
| "step": 538 |
| }, |
| { |
| "epoch": 2.182186234817814, |
| "grad_norm": 0.31464895636763957, |
| "learning_rate": 2.7738492792643224e-05, |
| "loss": 0.2489, |
| "step": 539 |
| }, |
| { |
| "epoch": 2.1862348178137654, |
| "grad_norm": 0.3303580956722188, |
| "learning_rate": 2.768631254100716e-05, |
| "loss": 0.2594, |
| "step": 540 |
| }, |
| { |
| "epoch": 2.1902834008097165, |
| "grad_norm": 0.343130240585527, |
| "learning_rate": 2.7634070829820836e-05, |
| "loss": 0.2657, |
| "step": 541 |
| }, |
| { |
| "epoch": 2.194331983805668, |
| "grad_norm": 0.31781644189396313, |
| "learning_rate": 2.7581768076807592e-05, |
| "loss": 0.236, |
| "step": 542 |
| }, |
| { |
| "epoch": 2.1983805668016196, |
| "grad_norm": 0.3278948337429747, |
| "learning_rate": 2.7529404700178832e-05, |
| "loss": 0.2611, |
| "step": 543 |
| }, |
| { |
| "epoch": 2.2024291497975708, |
| "grad_norm": 0.3055914034234545, |
| "learning_rate": 2.7476981118630735e-05, |
| "loss": 0.2702, |
| "step": 544 |
| }, |
| { |
| "epoch": 2.2064777327935223, |
| "grad_norm": 0.3337784901818513, |
| "learning_rate": 2.742449775134085e-05, |
| "loss": 0.2576, |
| "step": 545 |
| }, |
| { |
| "epoch": 2.2105263157894735, |
| "grad_norm": 0.3083860559979172, |
| "learning_rate": 2.7371955017964788e-05, |
| "loss": 0.2721, |
| "step": 546 |
| }, |
| { |
| "epoch": 2.214574898785425, |
| "grad_norm": 0.37015641160761414, |
| "learning_rate": 2.7319353338632845e-05, |
| "loss": 0.2611, |
| "step": 547 |
| }, |
| { |
| "epoch": 2.2186234817813766, |
| "grad_norm": 0.33098026632939936, |
| "learning_rate": 2.7266693133946645e-05, |
| "loss": 0.2605, |
| "step": 548 |
| }, |
| { |
| "epoch": 2.2226720647773277, |
| "grad_norm": 0.35532586736902594, |
| "learning_rate": 2.7213974824975785e-05, |
| "loss": 0.2706, |
| "step": 549 |
| }, |
| { |
| "epoch": 2.2267206477732793, |
| "grad_norm": 0.36045069527761403, |
| "learning_rate": 2.7161198833254464e-05, |
| "loss": 0.2603, |
| "step": 550 |
| }, |
| { |
| "epoch": 2.230769230769231, |
| "grad_norm": 0.3189819397099714, |
| "learning_rate": 2.7108365580778095e-05, |
| "loss": 0.2806, |
| "step": 551 |
| }, |
| { |
| "epoch": 2.234817813765182, |
| "grad_norm": 0.3252657106926538, |
| "learning_rate": 2.705547548999996e-05, |
| "loss": 0.2652, |
| "step": 552 |
| }, |
| { |
| "epoch": 2.2388663967611335, |
| "grad_norm": 0.3117371460455724, |
| "learning_rate": 2.7002528983827817e-05, |
| "loss": 0.2324, |
| "step": 553 |
| }, |
| { |
| "epoch": 2.242914979757085, |
| "grad_norm": 0.3573364504374601, |
| "learning_rate": 2.6949526485620517e-05, |
| "loss": 0.2648, |
| "step": 554 |
| }, |
| { |
| "epoch": 2.246963562753036, |
| "grad_norm": 0.31987712968828397, |
| "learning_rate": 2.6896468419184623e-05, |
| "loss": 0.2393, |
| "step": 555 |
| }, |
| { |
| "epoch": 2.251012145748988, |
| "grad_norm": 0.33168939669681224, |
| "learning_rate": 2.6843355208771024e-05, |
| "loss": 0.2621, |
| "step": 556 |
| }, |
| { |
| "epoch": 2.2550607287449393, |
| "grad_norm": 0.36012348998523536, |
| "learning_rate": 2.6790187279071536e-05, |
| "loss": 0.2597, |
| "step": 557 |
| }, |
| { |
| "epoch": 2.2591093117408905, |
| "grad_norm": 0.29929546846450494, |
| "learning_rate": 2.67369650552155e-05, |
| "loss": 0.2626, |
| "step": 558 |
| }, |
| { |
| "epoch": 2.263157894736842, |
| "grad_norm": 0.3784138450403676, |
| "learning_rate": 2.6683688962766407e-05, |
| "loss": 0.255, |
| "step": 559 |
| }, |
| { |
| "epoch": 2.2672064777327936, |
| "grad_norm": 0.34376699821043355, |
| "learning_rate": 2.6630359427718468e-05, |
| "loss": 0.2639, |
| "step": 560 |
| }, |
| { |
| "epoch": 2.2712550607287447, |
| "grad_norm": 0.3122178531265938, |
| "learning_rate": 2.6576976876493225e-05, |
| "loss": 0.2551, |
| "step": 561 |
| }, |
| { |
| "epoch": 2.2753036437246963, |
| "grad_norm": 0.34996616535764613, |
| "learning_rate": 2.652354173593614e-05, |
| "loss": 0.263, |
| "step": 562 |
| }, |
| { |
| "epoch": 2.279352226720648, |
| "grad_norm": 0.3374790677931627, |
| "learning_rate": 2.647005443331317e-05, |
| "loss": 0.2487, |
| "step": 563 |
| }, |
| { |
| "epoch": 2.283400809716599, |
| "grad_norm": 0.3709778037903167, |
| "learning_rate": 2.6416515396307354e-05, |
| "loss": 0.2678, |
| "step": 564 |
| }, |
| { |
| "epoch": 2.2874493927125505, |
| "grad_norm": 0.38309952029181554, |
| "learning_rate": 2.636292505301542e-05, |
| "loss": 0.2486, |
| "step": 565 |
| }, |
| { |
| "epoch": 2.291497975708502, |
| "grad_norm": 0.3599529757615204, |
| "learning_rate": 2.630928383194431e-05, |
| "loss": 0.2815, |
| "step": 566 |
| }, |
| { |
| "epoch": 2.2955465587044532, |
| "grad_norm": 0.39339414739555784, |
| "learning_rate": 2.625559216200781e-05, |
| "loss": 0.2565, |
| "step": 567 |
| }, |
| { |
| "epoch": 2.299595141700405, |
| "grad_norm": 0.3584893167497682, |
| "learning_rate": 2.620185047252308e-05, |
| "loss": 0.2715, |
| "step": 568 |
| }, |
| { |
| "epoch": 2.3036437246963564, |
| "grad_norm": 0.32354649106466593, |
| "learning_rate": 2.614805919320723e-05, |
| "loss": 0.2644, |
| "step": 569 |
| }, |
| { |
| "epoch": 2.3076923076923075, |
| "grad_norm": 0.40456308444496647, |
| "learning_rate": 2.6094218754173895e-05, |
| "loss": 0.2637, |
| "step": 570 |
| }, |
| { |
| "epoch": 2.311740890688259, |
| "grad_norm": 0.33935606074834834, |
| "learning_rate": 2.60403295859298e-05, |
| "loss": 0.2553, |
| "step": 571 |
| }, |
| { |
| "epoch": 2.3157894736842106, |
| "grad_norm": 0.3694625276857769, |
| "learning_rate": 2.5986392119371297e-05, |
| "loss": 0.2422, |
| "step": 572 |
| }, |
| { |
| "epoch": 2.3198380566801617, |
| "grad_norm": 0.3259881378965262, |
| "learning_rate": 2.5932406785780928e-05, |
| "loss": 0.2566, |
| "step": 573 |
| }, |
| { |
| "epoch": 2.3238866396761133, |
| "grad_norm": 0.3716829933936014, |
| "learning_rate": 2.5878374016823987e-05, |
| "loss": 0.2521, |
| "step": 574 |
| }, |
| { |
| "epoch": 2.327935222672065, |
| "grad_norm": 0.3457219444911625, |
| "learning_rate": 2.5824294244545045e-05, |
| "loss": 0.2743, |
| "step": 575 |
| }, |
| { |
| "epoch": 2.331983805668016, |
| "grad_norm": 0.30936904302162405, |
| "learning_rate": 2.577016790136453e-05, |
| "loss": 0.2541, |
| "step": 576 |
| }, |
| { |
| "epoch": 2.3360323886639676, |
| "grad_norm": 0.3334225622096724, |
| "learning_rate": 2.5715995420075225e-05, |
| "loss": 0.2601, |
| "step": 577 |
| }, |
| { |
| "epoch": 2.340080971659919, |
| "grad_norm": 0.30103386286459416, |
| "learning_rate": 2.5661777233838864e-05, |
| "loss": 0.2597, |
| "step": 578 |
| }, |
| { |
| "epoch": 2.3441295546558703, |
| "grad_norm": 0.3206037355211713, |
| "learning_rate": 2.5607513776182616e-05, |
| "loss": 0.2673, |
| "step": 579 |
| }, |
| { |
| "epoch": 2.348178137651822, |
| "grad_norm": 0.33947807279768144, |
| "learning_rate": 2.5553205480995633e-05, |
| "loss": 0.2669, |
| "step": 580 |
| }, |
| { |
| "epoch": 2.3522267206477734, |
| "grad_norm": 0.30692369549492204, |
| "learning_rate": 2.5498852782525604e-05, |
| "loss": 0.2684, |
| "step": 581 |
| }, |
| { |
| "epoch": 2.3562753036437245, |
| "grad_norm": 0.3430956531732463, |
| "learning_rate": 2.5444456115375252e-05, |
| "loss": 0.2735, |
| "step": 582 |
| }, |
| { |
| "epoch": 2.360323886639676, |
| "grad_norm": 0.3404721525734231, |
| "learning_rate": 2.539001591449889e-05, |
| "loss": 0.2538, |
| "step": 583 |
| }, |
| { |
| "epoch": 2.3643724696356276, |
| "grad_norm": 0.33656603582536493, |
| "learning_rate": 2.53355326151989e-05, |
| "loss": 0.2517, |
| "step": 584 |
| }, |
| { |
| "epoch": 2.3684210526315788, |
| "grad_norm": 0.32398318933154235, |
| "learning_rate": 2.5281006653122288e-05, |
| "loss": 0.2604, |
| "step": 585 |
| }, |
| { |
| "epoch": 2.3724696356275303, |
| "grad_norm": 0.3194071964539124, |
| "learning_rate": 2.5226438464257217e-05, |
| "loss": 0.2594, |
| "step": 586 |
| }, |
| { |
| "epoch": 2.376518218623482, |
| "grad_norm": 0.32560979959396724, |
| "learning_rate": 2.5171828484929447e-05, |
| "loss": 0.2767, |
| "step": 587 |
| }, |
| { |
| "epoch": 2.380566801619433, |
| "grad_norm": 0.29763100106948276, |
| "learning_rate": 2.5117177151798934e-05, |
| "loss": 0.2657, |
| "step": 588 |
| }, |
| { |
| "epoch": 2.3846153846153846, |
| "grad_norm": 0.31600508110780484, |
| "learning_rate": 2.506248490185627e-05, |
| "loss": 0.2611, |
| "step": 589 |
| }, |
| { |
| "epoch": 2.388663967611336, |
| "grad_norm": 0.3186288041067899, |
| "learning_rate": 2.500775217241923e-05, |
| "loss": 0.2607, |
| "step": 590 |
| }, |
| { |
| "epoch": 2.3927125506072873, |
| "grad_norm": 0.29543045821064634, |
| "learning_rate": 2.4952979401129266e-05, |
| "loss": 0.2599, |
| "step": 591 |
| }, |
| { |
| "epoch": 2.396761133603239, |
| "grad_norm": 0.29968599630995985, |
| "learning_rate": 2.4898167025948e-05, |
| "loss": 0.2763, |
| "step": 592 |
| }, |
| { |
| "epoch": 2.4008097165991904, |
| "grad_norm": 0.3182796213026839, |
| "learning_rate": 2.4843315485153703e-05, |
| "loss": 0.2675, |
| "step": 593 |
| }, |
| { |
| "epoch": 2.4048582995951415, |
| "grad_norm": 0.2807604353021131, |
| "learning_rate": 2.4788425217337856e-05, |
| "loss": 0.2504, |
| "step": 594 |
| }, |
| { |
| "epoch": 2.408906882591093, |
| "grad_norm": 0.33407106967298295, |
| "learning_rate": 2.4733496661401572e-05, |
| "loss": 0.2427, |
| "step": 595 |
| }, |
| { |
| "epoch": 2.4129554655870447, |
| "grad_norm": 0.32224033381166667, |
| "learning_rate": 2.4678530256552114e-05, |
| "loss": 0.2697, |
| "step": 596 |
| }, |
| { |
| "epoch": 2.417004048582996, |
| "grad_norm": 0.3035117148773753, |
| "learning_rate": 2.46235264422994e-05, |
| "loss": 0.2592, |
| "step": 597 |
| }, |
| { |
| "epoch": 2.4210526315789473, |
| "grad_norm": 0.3236737451483454, |
| "learning_rate": 2.4568485658452452e-05, |
| "loss": 0.2773, |
| "step": 598 |
| }, |
| { |
| "epoch": 2.425101214574899, |
| "grad_norm": 0.3184006301512246, |
| "learning_rate": 2.451340834511592e-05, |
| "loss": 0.2447, |
| "step": 599 |
| }, |
| { |
| "epoch": 2.42914979757085, |
| "grad_norm": 0.3050300346198166, |
| "learning_rate": 2.4458294942686528e-05, |
| "loss": 0.2701, |
| "step": 600 |
| }, |
| { |
| "epoch": 2.4331983805668016, |
| "grad_norm": 0.2955889798569046, |
| "learning_rate": 2.440314589184958e-05, |
| "loss": 0.268, |
| "step": 601 |
| }, |
| { |
| "epoch": 2.437246963562753, |
| "grad_norm": 0.3315081647871225, |
| "learning_rate": 2.434796163357541e-05, |
| "loss": 0.2319, |
| "step": 602 |
| }, |
| { |
| "epoch": 2.4412955465587043, |
| "grad_norm": 0.3039201105483638, |
| "learning_rate": 2.429274260911588e-05, |
| "loss": 0.2538, |
| "step": 603 |
| }, |
| { |
| "epoch": 2.445344129554656, |
| "grad_norm": 0.31908428370941955, |
| "learning_rate": 2.423748926000084e-05, |
| "loss": 0.2749, |
| "step": 604 |
| }, |
| { |
| "epoch": 2.4493927125506074, |
| "grad_norm": 0.3162920723869634, |
| "learning_rate": 2.418220202803459e-05, |
| "loss": 0.2548, |
| "step": 605 |
| }, |
| { |
| "epoch": 2.4534412955465585, |
| "grad_norm": 0.3210550022241104, |
| "learning_rate": 2.412688135529237e-05, |
| "loss": 0.2592, |
| "step": 606 |
| }, |
| { |
| "epoch": 2.45748987854251, |
| "grad_norm": 0.3209080298667877, |
| "learning_rate": 2.4071527684116807e-05, |
| "loss": 0.2562, |
| "step": 607 |
| }, |
| { |
| "epoch": 2.4615384615384617, |
| "grad_norm": 0.34223299629862175, |
| "learning_rate": 2.4016141457114372e-05, |
| "loss": 0.2794, |
| "step": 608 |
| }, |
| { |
| "epoch": 2.465587044534413, |
| "grad_norm": 0.32456199208552644, |
| "learning_rate": 2.396072311715187e-05, |
| "loss": 0.274, |
| "step": 609 |
| }, |
| { |
| "epoch": 2.4696356275303644, |
| "grad_norm": 0.30839549956878387, |
| "learning_rate": 2.3905273107352863e-05, |
| "loss": 0.2874, |
| "step": 610 |
| }, |
| { |
| "epoch": 2.473684210526316, |
| "grad_norm": 0.3399097531674187, |
| "learning_rate": 2.3849791871094157e-05, |
| "loss": 0.2615, |
| "step": 611 |
| }, |
| { |
| "epoch": 2.477732793522267, |
| "grad_norm": 0.344712295384869, |
| "learning_rate": 2.379427985200223e-05, |
| "loss": 0.2449, |
| "step": 612 |
| }, |
| { |
| "epoch": 2.4817813765182186, |
| "grad_norm": 0.3051669169549875, |
| "learning_rate": 2.373873749394972e-05, |
| "loss": 0.2465, |
| "step": 613 |
| }, |
| { |
| "epoch": 2.48582995951417, |
| "grad_norm": 0.357628810271979, |
| "learning_rate": 2.3683165241051836e-05, |
| "loss": 0.2565, |
| "step": 614 |
| }, |
| { |
| "epoch": 2.4898785425101213, |
| "grad_norm": 0.297113961497081, |
| "learning_rate": 2.3627563537662823e-05, |
| "loss": 0.2656, |
| "step": 615 |
| }, |
| { |
| "epoch": 2.493927125506073, |
| "grad_norm": 0.3365658936032585, |
| "learning_rate": 2.3571932828372433e-05, |
| "loss": 0.2407, |
| "step": 616 |
| }, |
| { |
| "epoch": 2.4979757085020244, |
| "grad_norm": 0.3450163645418969, |
| "learning_rate": 2.351627355800233e-05, |
| "loss": 0.2646, |
| "step": 617 |
| }, |
| { |
| "epoch": 2.5020242914979756, |
| "grad_norm": 0.2849127055741892, |
| "learning_rate": 2.346058617160256e-05, |
| "loss": 0.2599, |
| "step": 618 |
| }, |
| { |
| "epoch": 2.506072874493927, |
| "grad_norm": 0.3406772543686688, |
| "learning_rate": 2.3404871114447976e-05, |
| "loss": 0.2409, |
| "step": 619 |
| }, |
| { |
| "epoch": 2.5101214574898787, |
| "grad_norm": 0.3591476153780799, |
| "learning_rate": 2.334912883203469e-05, |
| "loss": 0.2552, |
| "step": 620 |
| }, |
| { |
| "epoch": 2.51417004048583, |
| "grad_norm": 0.33166714029488464, |
| "learning_rate": 2.3293359770076505e-05, |
| "loss": 0.2576, |
| "step": 621 |
| }, |
| { |
| "epoch": 2.5182186234817814, |
| "grad_norm": 0.43927102648768357, |
| "learning_rate": 2.323756437450136e-05, |
| "loss": 0.2553, |
| "step": 622 |
| }, |
| { |
| "epoch": 2.522267206477733, |
| "grad_norm": 0.3169397927965758, |
| "learning_rate": 2.3181743091447754e-05, |
| "loss": 0.2659, |
| "step": 623 |
| }, |
| { |
| "epoch": 2.526315789473684, |
| "grad_norm": 0.3396421530118784, |
| "learning_rate": 2.3125896367261164e-05, |
| "loss": 0.2641, |
| "step": 624 |
| }, |
| { |
| "epoch": 2.5303643724696356, |
| "grad_norm": 0.3736138216936975, |
| "learning_rate": 2.307002464849052e-05, |
| "loss": 0.2564, |
| "step": 625 |
| }, |
| { |
| "epoch": 2.534412955465587, |
| "grad_norm": 0.31656095191333433, |
| "learning_rate": 2.30141283818846e-05, |
| "loss": 0.2625, |
| "step": 626 |
| }, |
| { |
| "epoch": 2.5384615384615383, |
| "grad_norm": 0.39912832247971286, |
| "learning_rate": 2.2958208014388462e-05, |
| "loss": 0.2544, |
| "step": 627 |
| }, |
| { |
| "epoch": 2.54251012145749, |
| "grad_norm": 0.33099624207681816, |
| "learning_rate": 2.2902263993139857e-05, |
| "loss": 0.256, |
| "step": 628 |
| }, |
| { |
| "epoch": 2.5465587044534415, |
| "grad_norm": 0.34302198801481065, |
| "learning_rate": 2.2846296765465708e-05, |
| "loss": 0.2592, |
| "step": 629 |
| }, |
| { |
| "epoch": 2.5506072874493926, |
| "grad_norm": 0.3494839634283674, |
| "learning_rate": 2.2790306778878465e-05, |
| "loss": 0.2746, |
| "step": 630 |
| }, |
| { |
| "epoch": 2.554655870445344, |
| "grad_norm": 0.34609347249808387, |
| "learning_rate": 2.273429448107257e-05, |
| "loss": 0.2447, |
| "step": 631 |
| }, |
| { |
| "epoch": 2.5587044534412957, |
| "grad_norm": 0.3221487035018445, |
| "learning_rate": 2.2678260319920868e-05, |
| "loss": 0.264, |
| "step": 632 |
| }, |
| { |
| "epoch": 2.562753036437247, |
| "grad_norm": 0.3447831489495334, |
| "learning_rate": 2.262220474347102e-05, |
| "loss": 0.2731, |
| "step": 633 |
| }, |
| { |
| "epoch": 2.5668016194331984, |
| "grad_norm": 0.3178568358459219, |
| "learning_rate": 2.2566128199941918e-05, |
| "loss": 0.2661, |
| "step": 634 |
| }, |
| { |
| "epoch": 2.57085020242915, |
| "grad_norm": 0.31949589449810245, |
| "learning_rate": 2.2510031137720104e-05, |
| "loss": 0.2177, |
| "step": 635 |
| }, |
| { |
| "epoch": 2.574898785425101, |
| "grad_norm": 0.31717782118811905, |
| "learning_rate": 2.2453914005356206e-05, |
| "loss": 0.2623, |
| "step": 636 |
| }, |
| { |
| "epoch": 2.5789473684210527, |
| "grad_norm": 0.32799426298478057, |
| "learning_rate": 2.2397777251561302e-05, |
| "loss": 0.2467, |
| "step": 637 |
| }, |
| { |
| "epoch": 2.582995951417004, |
| "grad_norm": 0.33419756571218945, |
| "learning_rate": 2.2341621325203393e-05, |
| "loss": 0.2619, |
| "step": 638 |
| }, |
| { |
| "epoch": 2.5870445344129553, |
| "grad_norm": 0.34560109278697554, |
| "learning_rate": 2.2285446675303764e-05, |
| "loss": 0.2455, |
| "step": 639 |
| }, |
| { |
| "epoch": 2.591093117408907, |
| "grad_norm": 0.33347206881212593, |
| "learning_rate": 2.2229253751033423e-05, |
| "loss": 0.2547, |
| "step": 640 |
| }, |
| { |
| "epoch": 2.5951417004048585, |
| "grad_norm": 0.34352170519557707, |
| "learning_rate": 2.2173043001709492e-05, |
| "loss": 0.2638, |
| "step": 641 |
| }, |
| { |
| "epoch": 2.5991902834008096, |
| "grad_norm": 0.3662912361547416, |
| "learning_rate": 2.2116814876791627e-05, |
| "loss": 0.2472, |
| "step": 642 |
| }, |
| { |
| "epoch": 2.603238866396761, |
| "grad_norm": 0.32732593482128963, |
| "learning_rate": 2.2060569825878413e-05, |
| "loss": 0.2638, |
| "step": 643 |
| }, |
| { |
| "epoch": 2.6072874493927127, |
| "grad_norm": 0.33085724075142714, |
| "learning_rate": 2.2004308298703777e-05, |
| "loss": 0.2649, |
| "step": 644 |
| }, |
| { |
| "epoch": 2.611336032388664, |
| "grad_norm": 0.3220609795746063, |
| "learning_rate": 2.1948030745133396e-05, |
| "loss": 0.2734, |
| "step": 645 |
| }, |
| { |
| "epoch": 2.6153846153846154, |
| "grad_norm": 0.3398656632294484, |
| "learning_rate": 2.1891737615161083e-05, |
| "loss": 0.2765, |
| "step": 646 |
| }, |
| { |
| "epoch": 2.619433198380567, |
| "grad_norm": 0.30974907279140707, |
| "learning_rate": 2.1835429358905218e-05, |
| "loss": 0.2578, |
| "step": 647 |
| }, |
| { |
| "epoch": 2.623481781376518, |
| "grad_norm": 0.3752158487569087, |
| "learning_rate": 2.17791064266051e-05, |
| "loss": 0.2854, |
| "step": 648 |
| }, |
| { |
| "epoch": 2.6275303643724697, |
| "grad_norm": 0.29162314753702184, |
| "learning_rate": 2.172276926861741e-05, |
| "loss": 0.2363, |
| "step": 649 |
| }, |
| { |
| "epoch": 2.6315789473684212, |
| "grad_norm": 0.33345196755102124, |
| "learning_rate": 2.166641833541255e-05, |
| "loss": 0.2722, |
| "step": 650 |
| }, |
| { |
| "epoch": 2.6356275303643724, |
| "grad_norm": 0.3234961136391531, |
| "learning_rate": 2.161005407757109e-05, |
| "loss": 0.2774, |
| "step": 651 |
| }, |
| { |
| "epoch": 2.639676113360324, |
| "grad_norm": 0.313638617385981, |
| "learning_rate": 2.155367694578013e-05, |
| "loss": 0.2673, |
| "step": 652 |
| }, |
| { |
| "epoch": 2.6437246963562755, |
| "grad_norm": 0.3194294202149954, |
| "learning_rate": 2.1497287390829712e-05, |
| "loss": 0.2574, |
| "step": 653 |
| }, |
| { |
| "epoch": 2.6477732793522266, |
| "grad_norm": 0.3231801455196432, |
| "learning_rate": 2.1440885863609208e-05, |
| "loss": 0.2592, |
| "step": 654 |
| }, |
| { |
| "epoch": 2.651821862348178, |
| "grad_norm": 0.3383156461526147, |
| "learning_rate": 2.1384472815103737e-05, |
| "loss": 0.2803, |
| "step": 655 |
| }, |
| { |
| "epoch": 2.6558704453441297, |
| "grad_norm": 0.31927814044102515, |
| "learning_rate": 2.1328048696390534e-05, |
| "loss": 0.244, |
| "step": 656 |
| }, |
| { |
| "epoch": 2.659919028340081, |
| "grad_norm": 0.3170775973258476, |
| "learning_rate": 2.1271613958635342e-05, |
| "loss": 0.2612, |
| "step": 657 |
| }, |
| { |
| "epoch": 2.6639676113360324, |
| "grad_norm": 0.3269074323655446, |
| "learning_rate": 2.1215169053088825e-05, |
| "loss": 0.2653, |
| "step": 658 |
| }, |
| { |
| "epoch": 2.668016194331984, |
| "grad_norm": 0.3404802291269436, |
| "learning_rate": 2.1158714431082937e-05, |
| "loss": 0.2514, |
| "step": 659 |
| }, |
| { |
| "epoch": 2.672064777327935, |
| "grad_norm": 0.30361232668165594, |
| "learning_rate": 2.110225054402733e-05, |
| "loss": 0.2547, |
| "step": 660 |
| }, |
| { |
| "epoch": 2.6761133603238867, |
| "grad_norm": 0.33909643359501745, |
| "learning_rate": 2.1045777843405747e-05, |
| "loss": 0.2724, |
| "step": 661 |
| }, |
| { |
| "epoch": 2.6801619433198383, |
| "grad_norm": 0.3088784621305505, |
| "learning_rate": 2.09892967807724e-05, |
| "loss": 0.265, |
| "step": 662 |
| }, |
| { |
| "epoch": 2.6842105263157894, |
| "grad_norm": 0.3326986904389568, |
| "learning_rate": 2.093280780774835e-05, |
| "loss": 0.2601, |
| "step": 663 |
| }, |
| { |
| "epoch": 2.688259109311741, |
| "grad_norm": 0.3339057301706217, |
| "learning_rate": 2.087631137601793e-05, |
| "loss": 0.2667, |
| "step": 664 |
| }, |
| { |
| "epoch": 2.6923076923076925, |
| "grad_norm": 0.29215375073524663, |
| "learning_rate": 2.0819807937325104e-05, |
| "loss": 0.2651, |
| "step": 665 |
| }, |
| { |
| "epoch": 2.6963562753036436, |
| "grad_norm": 0.3081924693500854, |
| "learning_rate": 2.0763297943469847e-05, |
| "loss": 0.2678, |
| "step": 666 |
| }, |
| { |
| "epoch": 2.700404858299595, |
| "grad_norm": 0.2943047802357407, |
| "learning_rate": 2.070678184630458e-05, |
| "loss": 0.2605, |
| "step": 667 |
| }, |
| { |
| "epoch": 2.7044534412955468, |
| "grad_norm": 0.3031071621259219, |
| "learning_rate": 2.06502600977305e-05, |
| "loss": 0.2654, |
| "step": 668 |
| }, |
| { |
| "epoch": 2.708502024291498, |
| "grad_norm": 0.31599752423500593, |
| "learning_rate": 2.0593733149694008e-05, |
| "loss": 0.2634, |
| "step": 669 |
| }, |
| { |
| "epoch": 2.7125506072874495, |
| "grad_norm": 0.3010713457616555, |
| "learning_rate": 2.0537201454183074e-05, |
| "loss": 0.2536, |
| "step": 670 |
| }, |
| { |
| "epoch": 2.716599190283401, |
| "grad_norm": 0.2905173548630858, |
| "learning_rate": 2.0480665463223626e-05, |
| "loss": 0.25, |
| "step": 671 |
| }, |
| { |
| "epoch": 2.720647773279352, |
| "grad_norm": 0.30782794230546756, |
| "learning_rate": 2.042412562887594e-05, |
| "loss": 0.2656, |
| "step": 672 |
| }, |
| { |
| "epoch": 2.7246963562753037, |
| "grad_norm": 0.28890864157211943, |
| "learning_rate": 2.0367582403231033e-05, |
| "loss": 0.2625, |
| "step": 673 |
| }, |
| { |
| "epoch": 2.7287449392712553, |
| "grad_norm": 0.31270468509078153, |
| "learning_rate": 2.0311036238407023e-05, |
| "loss": 0.2569, |
| "step": 674 |
| }, |
| { |
| "epoch": 2.7327935222672064, |
| "grad_norm": 0.3205720304471892, |
| "learning_rate": 2.0254487586545534e-05, |
| "loss": 0.262, |
| "step": 675 |
| }, |
| { |
| "epoch": 2.736842105263158, |
| "grad_norm": 0.2904897557684781, |
| "learning_rate": 2.0197936899808088e-05, |
| "loss": 0.2512, |
| "step": 676 |
| }, |
| { |
| "epoch": 2.7408906882591095, |
| "grad_norm": 0.309270280661125, |
| "learning_rate": 2.0141384630372466e-05, |
| "loss": 0.2485, |
| "step": 677 |
| }, |
| { |
| "epoch": 2.7449392712550607, |
| "grad_norm": 0.3250929796841352, |
| "learning_rate": 2.0084831230429106e-05, |
| "loss": 0.2569, |
| "step": 678 |
| }, |
| { |
| "epoch": 2.748987854251012, |
| "grad_norm": 0.3116851835075163, |
| "learning_rate": 2.0028277152177492e-05, |
| "loss": 0.2469, |
| "step": 679 |
| }, |
| { |
| "epoch": 2.753036437246964, |
| "grad_norm": 0.2996242038895974, |
| "learning_rate": 1.9971722847822518e-05, |
| "loss": 0.2545, |
| "step": 680 |
| }, |
| { |
| "epoch": 2.757085020242915, |
| "grad_norm": 0.28510548256087914, |
| "learning_rate": 1.99151687695709e-05, |
| "loss": 0.2478, |
| "step": 681 |
| }, |
| { |
| "epoch": 2.7611336032388665, |
| "grad_norm": 0.3108245950555442, |
| "learning_rate": 1.9858615369627537e-05, |
| "loss": 0.256, |
| "step": 682 |
| }, |
| { |
| "epoch": 2.765182186234818, |
| "grad_norm": 0.29979545273552827, |
| "learning_rate": 1.9802063100191916e-05, |
| "loss": 0.2615, |
| "step": 683 |
| }, |
| { |
| "epoch": 2.769230769230769, |
| "grad_norm": 0.31474269284638584, |
| "learning_rate": 1.974551241345447e-05, |
| "loss": 0.2643, |
| "step": 684 |
| }, |
| { |
| "epoch": 2.7732793522267207, |
| "grad_norm": 0.2997216636326053, |
| "learning_rate": 1.9688963761592984e-05, |
| "loss": 0.2793, |
| "step": 685 |
| }, |
| { |
| "epoch": 2.7773279352226723, |
| "grad_norm": 0.32044024316248826, |
| "learning_rate": 1.963241759676897e-05, |
| "loss": 0.2703, |
| "step": 686 |
| }, |
| { |
| "epoch": 2.7813765182186234, |
| "grad_norm": 0.318296510390314, |
| "learning_rate": 1.9575874371124066e-05, |
| "loss": 0.2546, |
| "step": 687 |
| }, |
| { |
| "epoch": 2.785425101214575, |
| "grad_norm": 0.32547441826070445, |
| "learning_rate": 1.951933453677638e-05, |
| "loss": 0.2585, |
| "step": 688 |
| }, |
| { |
| "epoch": 2.7894736842105265, |
| "grad_norm": 0.31482045063488173, |
| "learning_rate": 1.9462798545816932e-05, |
| "loss": 0.2609, |
| "step": 689 |
| }, |
| { |
| "epoch": 2.7935222672064777, |
| "grad_norm": 0.3272899679656168, |
| "learning_rate": 1.9406266850305996e-05, |
| "loss": 0.2587, |
| "step": 690 |
| }, |
| { |
| "epoch": 2.7975708502024292, |
| "grad_norm": 0.28034619909961317, |
| "learning_rate": 1.9349739902269503e-05, |
| "loss": 0.2495, |
| "step": 691 |
| }, |
| { |
| "epoch": 2.801619433198381, |
| "grad_norm": 0.31447945019427775, |
| "learning_rate": 1.9293218153695425e-05, |
| "loss": 0.269, |
| "step": 692 |
| }, |
| { |
| "epoch": 2.805668016194332, |
| "grad_norm": 0.3385104333622747, |
| "learning_rate": 1.9236702056530153e-05, |
| "loss": 0.2551, |
| "step": 693 |
| }, |
| { |
| "epoch": 2.8097165991902835, |
| "grad_norm": 0.27807357649704634, |
| "learning_rate": 1.9180192062674907e-05, |
| "loss": 0.2677, |
| "step": 694 |
| }, |
| { |
| "epoch": 2.813765182186235, |
| "grad_norm": 0.3079029468598094, |
| "learning_rate": 1.9123688623982076e-05, |
| "loss": 0.2532, |
| "step": 695 |
| }, |
| { |
| "epoch": 2.817813765182186, |
| "grad_norm": 0.29590179426964475, |
| "learning_rate": 1.906719219225165e-05, |
| "loss": 0.2461, |
| "step": 696 |
| }, |
| { |
| "epoch": 2.8218623481781377, |
| "grad_norm": 0.28626012132993306, |
| "learning_rate": 1.9010703219227603e-05, |
| "loss": 0.2825, |
| "step": 697 |
| }, |
| { |
| "epoch": 2.8259109311740893, |
| "grad_norm": 0.3094751247469855, |
| "learning_rate": 1.8954222156594256e-05, |
| "loss": 0.2622, |
| "step": 698 |
| }, |
| { |
| "epoch": 2.8299595141700404, |
| "grad_norm": 0.3636085061093592, |
| "learning_rate": 1.8897749455972673e-05, |
| "loss": 0.2702, |
| "step": 699 |
| }, |
| { |
| "epoch": 2.834008097165992, |
| "grad_norm": 0.3149647046025852, |
| "learning_rate": 1.884128556891707e-05, |
| "loss": 0.2567, |
| "step": 700 |
| }, |
| { |
| "epoch": 2.8380566801619436, |
| "grad_norm": 0.3265154572718607, |
| "learning_rate": 1.878483094691119e-05, |
| "loss": 0.2714, |
| "step": 701 |
| }, |
| { |
| "epoch": 2.8421052631578947, |
| "grad_norm": 0.32935013481903586, |
| "learning_rate": 1.8728386041364664e-05, |
| "loss": 0.2295, |
| "step": 702 |
| }, |
| { |
| "epoch": 2.8461538461538463, |
| "grad_norm": 0.3195110874904098, |
| "learning_rate": 1.867195130360947e-05, |
| "loss": 0.2608, |
| "step": 703 |
| }, |
| { |
| "epoch": 2.850202429149798, |
| "grad_norm": 0.32025107892331534, |
| "learning_rate": 1.8615527184896263e-05, |
| "loss": 0.2632, |
| "step": 704 |
| }, |
| { |
| "epoch": 2.854251012145749, |
| "grad_norm": 0.32237094589181575, |
| "learning_rate": 1.8559114136390795e-05, |
| "loss": 0.2427, |
| "step": 705 |
| }, |
| { |
| "epoch": 2.8582995951417005, |
| "grad_norm": 0.30528303260337036, |
| "learning_rate": 1.8502712609170298e-05, |
| "loss": 0.2429, |
| "step": 706 |
| }, |
| { |
| "epoch": 2.862348178137652, |
| "grad_norm": 0.3323809527421542, |
| "learning_rate": 1.8446323054219876e-05, |
| "loss": 0.2561, |
| "step": 707 |
| }, |
| { |
| "epoch": 2.866396761133603, |
| "grad_norm": 0.28612198000666594, |
| "learning_rate": 1.838994592242891e-05, |
| "loss": 0.2635, |
| "step": 708 |
| }, |
| { |
| "epoch": 2.8704453441295548, |
| "grad_norm": 0.30709461955686224, |
| "learning_rate": 1.8333581664587453e-05, |
| "loss": 0.2741, |
| "step": 709 |
| }, |
| { |
| "epoch": 2.8744939271255063, |
| "grad_norm": 0.3064069548209538, |
| "learning_rate": 1.8277230731382593e-05, |
| "loss": 0.2641, |
| "step": 710 |
| }, |
| { |
| "epoch": 2.8785425101214575, |
| "grad_norm": 0.2929258377918191, |
| "learning_rate": 1.82208935733949e-05, |
| "loss": 0.2735, |
| "step": 711 |
| }, |
| { |
| "epoch": 2.882591093117409, |
| "grad_norm": 0.3013752079376017, |
| "learning_rate": 1.8164570641094793e-05, |
| "loss": 0.2637, |
| "step": 712 |
| }, |
| { |
| "epoch": 2.8866396761133606, |
| "grad_norm": 0.30854339607559444, |
| "learning_rate": 1.810826238483892e-05, |
| "loss": 0.2663, |
| "step": 713 |
| }, |
| { |
| "epoch": 2.8906882591093117, |
| "grad_norm": 0.31228346342402186, |
| "learning_rate": 1.8051969254866608e-05, |
| "loss": 0.273, |
| "step": 714 |
| }, |
| { |
| "epoch": 2.8947368421052633, |
| "grad_norm": 0.2881550985194099, |
| "learning_rate": 1.7995691701296226e-05, |
| "loss": 0.2573, |
| "step": 715 |
| }, |
| { |
| "epoch": 2.898785425101215, |
| "grad_norm": 0.3036090753107869, |
| "learning_rate": 1.7939430174121593e-05, |
| "loss": 0.2671, |
| "step": 716 |
| }, |
| { |
| "epoch": 2.902834008097166, |
| "grad_norm": 0.31137355157137014, |
| "learning_rate": 1.788318512320838e-05, |
| "loss": 0.2703, |
| "step": 717 |
| }, |
| { |
| "epoch": 2.9068825910931175, |
| "grad_norm": 0.29392232430409343, |
| "learning_rate": 1.782695699829051e-05, |
| "loss": 0.2688, |
| "step": 718 |
| }, |
| { |
| "epoch": 2.910931174089069, |
| "grad_norm": 0.3186079369314793, |
| "learning_rate": 1.7770746248966587e-05, |
| "loss": 0.2462, |
| "step": 719 |
| }, |
| { |
| "epoch": 2.91497975708502, |
| "grad_norm": 0.2890170720347342, |
| "learning_rate": 1.7714553324696243e-05, |
| "loss": 0.2705, |
| "step": 720 |
| }, |
| { |
| "epoch": 2.919028340080972, |
| "grad_norm": 0.28972498288118603, |
| "learning_rate": 1.7658378674796614e-05, |
| "loss": 0.2794, |
| "step": 721 |
| }, |
| { |
| "epoch": 2.9230769230769234, |
| "grad_norm": 0.29119460214323933, |
| "learning_rate": 1.76022227484387e-05, |
| "loss": 0.2563, |
| "step": 722 |
| }, |
| { |
| "epoch": 2.9271255060728745, |
| "grad_norm": 0.273210956190742, |
| "learning_rate": 1.7546085994643807e-05, |
| "loss": 0.2556, |
| "step": 723 |
| }, |
| { |
| "epoch": 2.931174089068826, |
| "grad_norm": 0.29833652022105583, |
| "learning_rate": 1.7489968862279902e-05, |
| "loss": 0.263, |
| "step": 724 |
| }, |
| { |
| "epoch": 2.9352226720647776, |
| "grad_norm": 0.2838749939603081, |
| "learning_rate": 1.743387180005809e-05, |
| "loss": 0.2499, |
| "step": 725 |
| }, |
| { |
| "epoch": 2.9392712550607287, |
| "grad_norm": 0.29819490435314105, |
| "learning_rate": 1.737779525652899e-05, |
| "loss": 0.2584, |
| "step": 726 |
| }, |
| { |
| "epoch": 2.9433198380566803, |
| "grad_norm": 0.28338833540135205, |
| "learning_rate": 1.732173968007914e-05, |
| "loss": 0.2477, |
| "step": 727 |
| }, |
| { |
| "epoch": 2.9473684210526314, |
| "grad_norm": 0.29773668379269186, |
| "learning_rate": 1.7265705518927437e-05, |
| "loss": 0.2609, |
| "step": 728 |
| }, |
| { |
| "epoch": 2.951417004048583, |
| "grad_norm": 0.2777445261555438, |
| "learning_rate": 1.7209693221121542e-05, |
| "loss": 0.252, |
| "step": 729 |
| }, |
| { |
| "epoch": 2.9554655870445345, |
| "grad_norm": 0.2962550063956606, |
| "learning_rate": 1.7153703234534302e-05, |
| "loss": 0.2635, |
| "step": 730 |
| }, |
| { |
| "epoch": 2.9595141700404857, |
| "grad_norm": 0.29845270963186543, |
| "learning_rate": 1.709773600686015e-05, |
| "loss": 0.2536, |
| "step": 731 |
| }, |
| { |
| "epoch": 2.9635627530364372, |
| "grad_norm": 0.30407504746915476, |
| "learning_rate": 1.7041791985611548e-05, |
| "loss": 0.2608, |
| "step": 732 |
| }, |
| { |
| "epoch": 2.967611336032389, |
| "grad_norm": 0.2720129335965621, |
| "learning_rate": 1.69858716181154e-05, |
| "loss": 0.2646, |
| "step": 733 |
| }, |
| { |
| "epoch": 2.97165991902834, |
| "grad_norm": 0.2958751636879754, |
| "learning_rate": 1.692997535150948e-05, |
| "loss": 0.2346, |
| "step": 734 |
| }, |
| { |
| "epoch": 2.9757085020242915, |
| "grad_norm": 0.26228288843324854, |
| "learning_rate": 1.687410363273884e-05, |
| "loss": 0.2576, |
| "step": 735 |
| }, |
| { |
| "epoch": 2.979757085020243, |
| "grad_norm": 0.28789263236093693, |
| "learning_rate": 1.6818256908552257e-05, |
| "loss": 0.2561, |
| "step": 736 |
| }, |
| { |
| "epoch": 2.983805668016194, |
| "grad_norm": 0.29378425661613927, |
| "learning_rate": 1.6762435625498646e-05, |
| "loss": 0.2583, |
| "step": 737 |
| }, |
| { |
| "epoch": 2.9878542510121457, |
| "grad_norm": 0.2752449513045386, |
| "learning_rate": 1.67066402299235e-05, |
| "loss": 0.267, |
| "step": 738 |
| }, |
| { |
| "epoch": 2.9919028340080973, |
| "grad_norm": 0.2958949928279527, |
| "learning_rate": 1.6650871167965313e-05, |
| "loss": 0.2572, |
| "step": 739 |
| }, |
| { |
| "epoch": 2.9959514170040484, |
| "grad_norm": 0.2861769806216303, |
| "learning_rate": 1.6595128885552028e-05, |
| "loss": 0.2544, |
| "step": 740 |
| }, |
| { |
| "epoch": 3.0, |
| "grad_norm": 0.3023860144234979, |
| "learning_rate": 1.653941382839745e-05, |
| "loss": 0.2656, |
| "step": 741 |
| }, |
| { |
| "epoch": 3.0040485829959516, |
| "grad_norm": 0.44469157080471716, |
| "learning_rate": 1.6483726441997673e-05, |
| "loss": 0.1905, |
| "step": 742 |
| }, |
| { |
| "epoch": 3.0080971659919027, |
| "grad_norm": 0.27125510245135925, |
| "learning_rate": 1.642806717162757e-05, |
| "loss": 0.1921, |
| "step": 743 |
| }, |
| { |
| "epoch": 3.0121457489878543, |
| "grad_norm": 0.46010424388428267, |
| "learning_rate": 1.637243646233718e-05, |
| "loss": 0.1914, |
| "step": 744 |
| }, |
| { |
| "epoch": 3.016194331983806, |
| "grad_norm": 0.4218166023923455, |
| "learning_rate": 1.6316834758948174e-05, |
| "loss": 0.1963, |
| "step": 745 |
| }, |
| { |
| "epoch": 3.020242914979757, |
| "grad_norm": 0.324792172813281, |
| "learning_rate": 1.6261262506050282e-05, |
| "loss": 0.1868, |
| "step": 746 |
| }, |
| { |
| "epoch": 3.0242914979757085, |
| "grad_norm": 0.40729834594047143, |
| "learning_rate": 1.620572014799777e-05, |
| "loss": 0.1856, |
| "step": 747 |
| }, |
| { |
| "epoch": 3.02834008097166, |
| "grad_norm": 0.3513686279235662, |
| "learning_rate": 1.6150208128905857e-05, |
| "loss": 0.1785, |
| "step": 748 |
| }, |
| { |
| "epoch": 3.032388663967611, |
| "grad_norm": 0.291840321958323, |
| "learning_rate": 1.6094726892647147e-05, |
| "loss": 0.1995, |
| "step": 749 |
| }, |
| { |
| "epoch": 3.0364372469635628, |
| "grad_norm": 0.32385320135012247, |
| "learning_rate": 1.6039276882848135e-05, |
| "loss": 0.1883, |
| "step": 750 |
| }, |
| { |
| "epoch": 3.0404858299595143, |
| "grad_norm": 0.34913692341542407, |
| "learning_rate": 1.5983858542885635e-05, |
| "loss": 0.193, |
| "step": 751 |
| }, |
| { |
| "epoch": 3.0445344129554655, |
| "grad_norm": 0.3201718757412167, |
| "learning_rate": 1.5928472315883203e-05, |
| "loss": 0.2039, |
| "step": 752 |
| }, |
| { |
| "epoch": 3.048582995951417, |
| "grad_norm": 0.2937257793872338, |
| "learning_rate": 1.5873118644707633e-05, |
| "loss": 0.1897, |
| "step": 753 |
| }, |
| { |
| "epoch": 3.0526315789473686, |
| "grad_norm": 0.308565804090432, |
| "learning_rate": 1.5817797971965413e-05, |
| "loss": 0.1786, |
| "step": 754 |
| }, |
| { |
| "epoch": 3.0566801619433197, |
| "grad_norm": 0.3049971232932517, |
| "learning_rate": 1.576251073999917e-05, |
| "loss": 0.1868, |
| "step": 755 |
| }, |
| { |
| "epoch": 3.0607287449392713, |
| "grad_norm": 0.29647520641181585, |
| "learning_rate": 1.5707257390884126e-05, |
| "loss": 0.2094, |
| "step": 756 |
| }, |
| { |
| "epoch": 3.064777327935223, |
| "grad_norm": 0.31233476330405074, |
| "learning_rate": 1.5652038366424595e-05, |
| "loss": 0.1915, |
| "step": 757 |
| }, |
| { |
| "epoch": 3.068825910931174, |
| "grad_norm": 0.2879417278818393, |
| "learning_rate": 1.5596854108150424e-05, |
| "loss": 0.1924, |
| "step": 758 |
| }, |
| { |
| "epoch": 3.0728744939271255, |
| "grad_norm": 0.2960811213919211, |
| "learning_rate": 1.5541705057313476e-05, |
| "loss": 0.1873, |
| "step": 759 |
| }, |
| { |
| "epoch": 3.076923076923077, |
| "grad_norm": 0.29651323974121907, |
| "learning_rate": 1.5486591654884086e-05, |
| "loss": 0.1935, |
| "step": 760 |
| }, |
| { |
| "epoch": 3.080971659919028, |
| "grad_norm": 0.294187450431988, |
| "learning_rate": 1.543151434154755e-05, |
| "loss": 0.1923, |
| "step": 761 |
| }, |
| { |
| "epoch": 3.08502024291498, |
| "grad_norm": 0.2791961988351, |
| "learning_rate": 1.5376473557700615e-05, |
| "loss": 0.209, |
| "step": 762 |
| }, |
| { |
| "epoch": 3.0890688259109313, |
| "grad_norm": 0.27865900036211183, |
| "learning_rate": 1.5321469743447893e-05, |
| "loss": 0.2053, |
| "step": 763 |
| }, |
| { |
| "epoch": 3.0931174089068825, |
| "grad_norm": 0.2947932355759639, |
| "learning_rate": 1.5266503338598434e-05, |
| "loss": 0.2133, |
| "step": 764 |
| }, |
| { |
| "epoch": 3.097165991902834, |
| "grad_norm": 0.2979515797497424, |
| "learning_rate": 1.5211574782662149e-05, |
| "loss": 0.1993, |
| "step": 765 |
| }, |
| { |
| "epoch": 3.1012145748987856, |
| "grad_norm": 0.2845471222349274, |
| "learning_rate": 1.5156684514846307e-05, |
| "loss": 0.1903, |
| "step": 766 |
| }, |
| { |
| "epoch": 3.1052631578947367, |
| "grad_norm": 0.30568777537780134, |
| "learning_rate": 1.5101832974052012e-05, |
| "loss": 0.1773, |
| "step": 767 |
| }, |
| { |
| "epoch": 3.1093117408906883, |
| "grad_norm": 0.27266115800539714, |
| "learning_rate": 1.5047020598870737e-05, |
| "loss": 0.1846, |
| "step": 768 |
| }, |
| { |
| "epoch": 3.11336032388664, |
| "grad_norm": 0.27793421457264644, |
| "learning_rate": 1.4992247827580778e-05, |
| "loss": 0.1726, |
| "step": 769 |
| }, |
| { |
| "epoch": 3.117408906882591, |
| "grad_norm": 0.28533825913262567, |
| "learning_rate": 1.493751509814374e-05, |
| "loss": 0.1877, |
| "step": 770 |
| }, |
| { |
| "epoch": 3.1214574898785425, |
| "grad_norm": 0.26317363738976446, |
| "learning_rate": 1.4882822848201073e-05, |
| "loss": 0.2156, |
| "step": 771 |
| }, |
| { |
| "epoch": 3.125506072874494, |
| "grad_norm": 0.3011466870674777, |
| "learning_rate": 1.4828171515070553e-05, |
| "loss": 0.19, |
| "step": 772 |
| }, |
| { |
| "epoch": 3.1295546558704452, |
| "grad_norm": 0.2649377172464014, |
| "learning_rate": 1.4773561535742793e-05, |
| "loss": 0.2037, |
| "step": 773 |
| }, |
| { |
| "epoch": 3.133603238866397, |
| "grad_norm": 0.2797535170885471, |
| "learning_rate": 1.4718993346877715e-05, |
| "loss": 0.2057, |
| "step": 774 |
| }, |
| { |
| "epoch": 3.1376518218623484, |
| "grad_norm": 0.27378222909700856, |
| "learning_rate": 1.466446738480111e-05, |
| "loss": 0.2001, |
| "step": 775 |
| }, |
| { |
| "epoch": 3.1417004048582995, |
| "grad_norm": 0.26783852718203527, |
| "learning_rate": 1.460998408550112e-05, |
| "loss": 0.1959, |
| "step": 776 |
| }, |
| { |
| "epoch": 3.145748987854251, |
| "grad_norm": 0.268409990470649, |
| "learning_rate": 1.4555543884624751e-05, |
| "loss": 0.2066, |
| "step": 777 |
| }, |
| { |
| "epoch": 3.1497975708502026, |
| "grad_norm": 0.2737035581238145, |
| "learning_rate": 1.4501147217474402e-05, |
| "loss": 0.1962, |
| "step": 778 |
| }, |
| { |
| "epoch": 3.1538461538461537, |
| "grad_norm": 0.25791057622234104, |
| "learning_rate": 1.444679451900437e-05, |
| "loss": 0.2058, |
| "step": 779 |
| }, |
| { |
| "epoch": 3.1578947368421053, |
| "grad_norm": 0.250516112605739, |
| "learning_rate": 1.4392486223817397e-05, |
| "loss": 0.2127, |
| "step": 780 |
| }, |
| { |
| "epoch": 3.161943319838057, |
| "grad_norm": 0.25000407714482886, |
| "learning_rate": 1.4338222766161143e-05, |
| "loss": 0.1927, |
| "step": 781 |
| }, |
| { |
| "epoch": 3.165991902834008, |
| "grad_norm": 0.2501803687555566, |
| "learning_rate": 1.4284004579924777e-05, |
| "loss": 0.2022, |
| "step": 782 |
| }, |
| { |
| "epoch": 3.1700404858299596, |
| "grad_norm": 0.24534897806565734, |
| "learning_rate": 1.4229832098635479e-05, |
| "loss": 0.2049, |
| "step": 783 |
| }, |
| { |
| "epoch": 3.174089068825911, |
| "grad_norm": 0.26326266110859264, |
| "learning_rate": 1.4175705755454963e-05, |
| "loss": 0.1978, |
| "step": 784 |
| }, |
| { |
| "epoch": 3.1781376518218623, |
| "grad_norm": 0.2420635870996033, |
| "learning_rate": 1.412162598317602e-05, |
| "loss": 0.2116, |
| "step": 785 |
| }, |
| { |
| "epoch": 3.182186234817814, |
| "grad_norm": 0.2481083608723764, |
| "learning_rate": 1.406759321421907e-05, |
| "loss": 0.2082, |
| "step": 786 |
| }, |
| { |
| "epoch": 3.1862348178137654, |
| "grad_norm": 0.24868575402061915, |
| "learning_rate": 1.401360788062871e-05, |
| "loss": 0.1915, |
| "step": 787 |
| }, |
| { |
| "epoch": 3.1902834008097165, |
| "grad_norm": 0.264889936100424, |
| "learning_rate": 1.3959670414070204e-05, |
| "loss": 0.1901, |
| "step": 788 |
| }, |
| { |
| "epoch": 3.194331983805668, |
| "grad_norm": 0.2580939017519339, |
| "learning_rate": 1.3905781245826108e-05, |
| "loss": 0.1902, |
| "step": 789 |
| }, |
| { |
| "epoch": 3.1983805668016196, |
| "grad_norm": 0.24779644931140105, |
| "learning_rate": 1.3851940806792778e-05, |
| "loss": 0.1677, |
| "step": 790 |
| }, |
| { |
| "epoch": 3.2024291497975708, |
| "grad_norm": 0.25701923621433426, |
| "learning_rate": 1.379814952747693e-05, |
| "loss": 0.1906, |
| "step": 791 |
| }, |
| { |
| "epoch": 3.2064777327935223, |
| "grad_norm": 0.2761742062927418, |
| "learning_rate": 1.3744407837992193e-05, |
| "loss": 0.187, |
| "step": 792 |
| }, |
| { |
| "epoch": 3.2105263157894735, |
| "grad_norm": 0.2680119870303698, |
| "learning_rate": 1.3690716168055692e-05, |
| "loss": 0.1688, |
| "step": 793 |
| }, |
| { |
| "epoch": 3.214574898785425, |
| "grad_norm": 0.2641119688801228, |
| "learning_rate": 1.363707494698459e-05, |
| "loss": 0.1823, |
| "step": 794 |
| }, |
| { |
| "epoch": 3.2186234817813766, |
| "grad_norm": 0.25800276826224927, |
| "learning_rate": 1.358348460369265e-05, |
| "loss": 0.1779, |
| "step": 795 |
| }, |
| { |
| "epoch": 3.2226720647773277, |
| "grad_norm": 0.270644245260025, |
| "learning_rate": 1.352994556668684e-05, |
| "loss": 0.1974, |
| "step": 796 |
| }, |
| { |
| "epoch": 3.2267206477732793, |
| "grad_norm": 0.26012385434987845, |
| "learning_rate": 1.3476458264063865e-05, |
| "loss": 0.2158, |
| "step": 797 |
| }, |
| { |
| "epoch": 3.230769230769231, |
| "grad_norm": 0.25582606041456224, |
| "learning_rate": 1.342302312350678e-05, |
| "loss": 0.1935, |
| "step": 798 |
| }, |
| { |
| "epoch": 3.234817813765182, |
| "grad_norm": 0.2832813757466281, |
| "learning_rate": 1.3369640572281537e-05, |
| "loss": 0.1879, |
| "step": 799 |
| }, |
| { |
| "epoch": 3.2388663967611335, |
| "grad_norm": 0.2633958518488864, |
| "learning_rate": 1.3316311037233596e-05, |
| "loss": 0.1909, |
| "step": 800 |
| }, |
| { |
| "epoch": 3.242914979757085, |
| "grad_norm": 0.2674805273095265, |
| "learning_rate": 1.326303494478451e-05, |
| "loss": 0.1954, |
| "step": 801 |
| }, |
| { |
| "epoch": 3.246963562753036, |
| "grad_norm": 0.2566855593392634, |
| "learning_rate": 1.3209812720928472e-05, |
| "loss": 0.19, |
| "step": 802 |
| }, |
| { |
| "epoch": 3.251012145748988, |
| "grad_norm": 0.2781919733679089, |
| "learning_rate": 1.315664479122898e-05, |
| "loss": 0.1896, |
| "step": 803 |
| }, |
| { |
| "epoch": 3.2550607287449393, |
| "grad_norm": 0.256036069813793, |
| "learning_rate": 1.3103531580815378e-05, |
| "loss": 0.1775, |
| "step": 804 |
| }, |
| { |
| "epoch": 3.2591093117408905, |
| "grad_norm": 0.2574586129345356, |
| "learning_rate": 1.3050473514379488e-05, |
| "loss": 0.1864, |
| "step": 805 |
| }, |
| { |
| "epoch": 3.263157894736842, |
| "grad_norm": 0.29589940940190773, |
| "learning_rate": 1.2997471016172188e-05, |
| "loss": 0.1708, |
| "step": 806 |
| }, |
| { |
| "epoch": 3.2672064777327936, |
| "grad_norm": 0.2857866313520732, |
| "learning_rate": 1.2944524510000042e-05, |
| "loss": 0.1895, |
| "step": 807 |
| }, |
| { |
| "epoch": 3.2712550607287447, |
| "grad_norm": 0.26196712430755503, |
| "learning_rate": 1.2891634419221908e-05, |
| "loss": 0.1756, |
| "step": 808 |
| }, |
| { |
| "epoch": 3.2753036437246963, |
| "grad_norm": 0.2754148493919365, |
| "learning_rate": 1.2838801166745545e-05, |
| "loss": 0.2056, |
| "step": 809 |
| }, |
| { |
| "epoch": 3.279352226720648, |
| "grad_norm": 0.2707067874360677, |
| "learning_rate": 1.2786025175024216e-05, |
| "loss": 0.1885, |
| "step": 810 |
| }, |
| { |
| "epoch": 3.283400809716599, |
| "grad_norm": 0.2586496685904975, |
| "learning_rate": 1.2733306866053357e-05, |
| "loss": 0.2, |
| "step": 811 |
| }, |
| { |
| "epoch": 3.2874493927125505, |
| "grad_norm": 0.2551187241030325, |
| "learning_rate": 1.2680646661367163e-05, |
| "loss": 0.1982, |
| "step": 812 |
| }, |
| { |
| "epoch": 3.291497975708502, |
| "grad_norm": 0.26410798964919713, |
| "learning_rate": 1.2628044982035219e-05, |
| "loss": 0.1887, |
| "step": 813 |
| }, |
| { |
| "epoch": 3.2955465587044532, |
| "grad_norm": 0.26872507781127997, |
| "learning_rate": 1.2575502248659155e-05, |
| "loss": 0.1785, |
| "step": 814 |
| }, |
| { |
| "epoch": 3.299595141700405, |
| "grad_norm": 0.25626296046263486, |
| "learning_rate": 1.2523018881369269e-05, |
| "loss": 0.1902, |
| "step": 815 |
| }, |
| { |
| "epoch": 3.3036437246963564, |
| "grad_norm": 0.27513233982526203, |
| "learning_rate": 1.2470595299821176e-05, |
| "loss": 0.1989, |
| "step": 816 |
| }, |
| { |
| "epoch": 3.3076923076923075, |
| "grad_norm": 0.2828915121421577, |
| "learning_rate": 1.2418231923192416e-05, |
| "loss": 0.1817, |
| "step": 817 |
| }, |
| { |
| "epoch": 3.311740890688259, |
| "grad_norm": 0.28734206680040575, |
| "learning_rate": 1.2365929170179165e-05, |
| "loss": 0.1769, |
| "step": 818 |
| }, |
| { |
| "epoch": 3.3157894736842106, |
| "grad_norm": 0.2570262517879523, |
| "learning_rate": 1.2313687458992848e-05, |
| "loss": 0.2039, |
| "step": 819 |
| }, |
| { |
| "epoch": 3.3198380566801617, |
| "grad_norm": 0.2613693291698574, |
| "learning_rate": 1.2261507207356785e-05, |
| "loss": 0.1856, |
| "step": 820 |
| }, |
| { |
| "epoch": 3.3238866396761133, |
| "grad_norm": 0.25548529110175405, |
| "learning_rate": 1.220938883250288e-05, |
| "loss": 0.2024, |
| "step": 821 |
| }, |
| { |
| "epoch": 3.327935222672065, |
| "grad_norm": 0.27810875751948194, |
| "learning_rate": 1.215733275116827e-05, |
| "loss": 0.1817, |
| "step": 822 |
| }, |
| { |
| "epoch": 3.331983805668016, |
| "grad_norm": 0.2513116392075657, |
| "learning_rate": 1.2105339379592005e-05, |
| "loss": 0.1825, |
| "step": 823 |
| }, |
| { |
| "epoch": 3.3360323886639676, |
| "grad_norm": 0.2560408137407627, |
| "learning_rate": 1.2053409133511681e-05, |
| "loss": 0.1717, |
| "step": 824 |
| }, |
| { |
| "epoch": 3.340080971659919, |
| "grad_norm": 0.26266351565930135, |
| "learning_rate": 1.2001542428160173e-05, |
| "loss": 0.1901, |
| "step": 825 |
| }, |
| { |
| "epoch": 3.3441295546558703, |
| "grad_norm": 0.2750629177543435, |
| "learning_rate": 1.1949739678262278e-05, |
| "loss": 0.1923, |
| "step": 826 |
| }, |
| { |
| "epoch": 3.348178137651822, |
| "grad_norm": 0.2728380119238748, |
| "learning_rate": 1.189800129803139e-05, |
| "loss": 0.1719, |
| "step": 827 |
| }, |
| { |
| "epoch": 3.3522267206477734, |
| "grad_norm": 0.28095307097627553, |
| "learning_rate": 1.1846327701166228e-05, |
| "loss": 0.2022, |
| "step": 828 |
| }, |
| { |
| "epoch": 3.3562753036437245, |
| "grad_norm": 0.25858457505580734, |
| "learning_rate": 1.1794719300847493e-05, |
| "loss": 0.1871, |
| "step": 829 |
| }, |
| { |
| "epoch": 3.360323886639676, |
| "grad_norm": 0.2807604000099615, |
| "learning_rate": 1.1743176509734582e-05, |
| "loss": 0.1997, |
| "step": 830 |
| }, |
| { |
| "epoch": 3.3643724696356276, |
| "grad_norm": 0.2689985293656738, |
| "learning_rate": 1.1691699739962275e-05, |
| "loss": 0.1919, |
| "step": 831 |
| }, |
| { |
| "epoch": 3.3684210526315788, |
| "grad_norm": 0.27424772555335003, |
| "learning_rate": 1.1640289403137456e-05, |
| "loss": 0.1812, |
| "step": 832 |
| }, |
| { |
| "epoch": 3.3724696356275303, |
| "grad_norm": 0.279280617421667, |
| "learning_rate": 1.1588945910335802e-05, |
| "loss": 0.1911, |
| "step": 833 |
| }, |
| { |
| "epoch": 3.376518218623482, |
| "grad_norm": 0.2666963633997466, |
| "learning_rate": 1.1537669672098517e-05, |
| "loss": 0.1931, |
| "step": 834 |
| }, |
| { |
| "epoch": 3.380566801619433, |
| "grad_norm": 0.2598209668553413, |
| "learning_rate": 1.1486461098429036e-05, |
| "loss": 0.1863, |
| "step": 835 |
| }, |
| { |
| "epoch": 3.3846153846153846, |
| "grad_norm": 0.272681826877222, |
| "learning_rate": 1.1435320598789747e-05, |
| "loss": 0.1748, |
| "step": 836 |
| }, |
| { |
| "epoch": 3.388663967611336, |
| "grad_norm": 0.2782838598816993, |
| "learning_rate": 1.1384248582098736e-05, |
| "loss": 0.1885, |
| "step": 837 |
| }, |
| { |
| "epoch": 3.3927125506072873, |
| "grad_norm": 0.2495867400954, |
| "learning_rate": 1.1333245456726467e-05, |
| "loss": 0.1856, |
| "step": 838 |
| }, |
| { |
| "epoch": 3.396761133603239, |
| "grad_norm": 0.2515303830474592, |
| "learning_rate": 1.128231163049258e-05, |
| "loss": 0.184, |
| "step": 839 |
| }, |
| { |
| "epoch": 3.4008097165991904, |
| "grad_norm": 0.26564531691910775, |
| "learning_rate": 1.123144751066259e-05, |
| "loss": 0.1908, |
| "step": 840 |
| }, |
| { |
| "epoch": 3.4048582995951415, |
| "grad_norm": 0.2580593767540626, |
| "learning_rate": 1.1180653503944657e-05, |
| "loss": 0.2004, |
| "step": 841 |
| }, |
| { |
| "epoch": 3.408906882591093, |
| "grad_norm": 0.25581527207805327, |
| "learning_rate": 1.1129930016486287e-05, |
| "loss": 0.1876, |
| "step": 842 |
| }, |
| { |
| "epoch": 3.4129554655870447, |
| "grad_norm": 0.2625239442905213, |
| "learning_rate": 1.107927745387114e-05, |
| "loss": 0.1809, |
| "step": 843 |
| }, |
| { |
| "epoch": 3.417004048582996, |
| "grad_norm": 0.24626355924829574, |
| "learning_rate": 1.1028696221115766e-05, |
| "loss": 0.1988, |
| "step": 844 |
| }, |
| { |
| "epoch": 3.4210526315789473, |
| "grad_norm": 0.25444660512377026, |
| "learning_rate": 1.0978186722666341e-05, |
| "loss": 0.1918, |
| "step": 845 |
| }, |
| { |
| "epoch": 3.425101214574899, |
| "grad_norm": 0.24978474781607254, |
| "learning_rate": 1.0927749362395475e-05, |
| "loss": 0.195, |
| "step": 846 |
| }, |
| { |
| "epoch": 3.42914979757085, |
| "grad_norm": 0.2627367161133618, |
| "learning_rate": 1.0877384543598954e-05, |
| "loss": 0.2056, |
| "step": 847 |
| }, |
| { |
| "epoch": 3.4331983805668016, |
| "grad_norm": 0.2570782615959188, |
| "learning_rate": 1.0827092668992529e-05, |
| "loss": 0.1845, |
| "step": 848 |
| }, |
| { |
| "epoch": 3.437246963562753, |
| "grad_norm": 0.2593004955355829, |
| "learning_rate": 1.077687414070869e-05, |
| "loss": 0.1951, |
| "step": 849 |
| }, |
| { |
| "epoch": 3.4412955465587043, |
| "grad_norm": 0.24269553337371325, |
| "learning_rate": 1.0726729360293451e-05, |
| "loss": 0.1961, |
| "step": 850 |
| }, |
| { |
| "epoch": 3.445344129554656, |
| "grad_norm": 0.26205391063129024, |
| "learning_rate": 1.0676658728703147e-05, |
| "loss": 0.18, |
| "step": 851 |
| }, |
| { |
| "epoch": 3.4493927125506074, |
| "grad_norm": 0.24660825853193533, |
| "learning_rate": 1.0626662646301198e-05, |
| "loss": 0.1895, |
| "step": 852 |
| }, |
| { |
| "epoch": 3.4534412955465585, |
| "grad_norm": 0.2614466965288134, |
| "learning_rate": 1.0576741512854958e-05, |
| "loss": 0.1807, |
| "step": 853 |
| }, |
| { |
| "epoch": 3.45748987854251, |
| "grad_norm": 0.25988190313664156, |
| "learning_rate": 1.052689572753247e-05, |
| "loss": 0.1834, |
| "step": 854 |
| }, |
| { |
| "epoch": 3.4615384615384617, |
| "grad_norm": 0.27693817036186114, |
| "learning_rate": 1.0477125688899328e-05, |
| "loss": 0.185, |
| "step": 855 |
| }, |
| { |
| "epoch": 3.465587044534413, |
| "grad_norm": 0.2609010703757351, |
| "learning_rate": 1.0427431794915406e-05, |
| "loss": 0.192, |
| "step": 856 |
| }, |
| { |
| "epoch": 3.4696356275303644, |
| "grad_norm": 0.26920717398753174, |
| "learning_rate": 1.0377814442931769e-05, |
| "loss": 0.187, |
| "step": 857 |
| }, |
| { |
| "epoch": 3.473684210526316, |
| "grad_norm": 0.2618278475382347, |
| "learning_rate": 1.0328274029687439e-05, |
| "loss": 0.1747, |
| "step": 858 |
| }, |
| { |
| "epoch": 3.477732793522267, |
| "grad_norm": 0.2535391767446397, |
| "learning_rate": 1.0278810951306253e-05, |
| "loss": 0.1944, |
| "step": 859 |
| }, |
| { |
| "epoch": 3.4817813765182186, |
| "grad_norm": 0.25564617279716506, |
| "learning_rate": 1.0229425603293647e-05, |
| "loss": 0.1826, |
| "step": 860 |
| }, |
| { |
| "epoch": 3.48582995951417, |
| "grad_norm": 0.2681905980273499, |
| "learning_rate": 1.0180118380533558e-05, |
| "loss": 0.2098, |
| "step": 861 |
| }, |
| { |
| "epoch": 3.4898785425101213, |
| "grad_norm": 0.26798644141469, |
| "learning_rate": 1.0130889677285225e-05, |
| "loss": 0.1932, |
| "step": 862 |
| }, |
| { |
| "epoch": 3.493927125506073, |
| "grad_norm": 0.2688759437484638, |
| "learning_rate": 1.008173988718005e-05, |
| "loss": 0.1815, |
| "step": 863 |
| }, |
| { |
| "epoch": 3.4979757085020244, |
| "grad_norm": 0.24901114011806064, |
| "learning_rate": 1.0032669403218439e-05, |
| "loss": 0.1999, |
| "step": 864 |
| }, |
| { |
| "epoch": 3.5020242914979756, |
| "grad_norm": 0.27511299015085033, |
| "learning_rate": 9.983678617766677e-06, |
| "loss": 0.1881, |
| "step": 865 |
| }, |
| { |
| "epoch": 3.506072874493927, |
| "grad_norm": 0.24427541812300746, |
| "learning_rate": 9.934767922553788e-06, |
| "loss": 0.1788, |
| "step": 866 |
| }, |
| { |
| "epoch": 3.5101214574898787, |
| "grad_norm": 0.2620028454038348, |
| "learning_rate": 9.885937708668371e-06, |
| "loss": 0.1786, |
| "step": 867 |
| }, |
| { |
| "epoch": 3.51417004048583, |
| "grad_norm": 0.2710732327703116, |
| "learning_rate": 9.837188366555524e-06, |
| "loss": 0.1926, |
| "step": 868 |
| }, |
| { |
| "epoch": 3.5182186234817814, |
| "grad_norm": 0.27364640027830656, |
| "learning_rate": 9.788520286013691e-06, |
| "loss": 0.1895, |
| "step": 869 |
| }, |
| { |
| "epoch": 3.522267206477733, |
| "grad_norm": 0.26606522992440984, |
| "learning_rate": 9.739933856191537e-06, |
| "loss": 0.1932, |
| "step": 870 |
| }, |
| { |
| "epoch": 3.526315789473684, |
| "grad_norm": 0.26070864298045765, |
| "learning_rate": 9.691429465584866e-06, |
| "loss": 0.1909, |
| "step": 871 |
| }, |
| { |
| "epoch": 3.5303643724696356, |
| "grad_norm": 0.2692121504564334, |
| "learning_rate": 9.643007502033492e-06, |
| "loss": 0.1949, |
| "step": 872 |
| }, |
| { |
| "epoch": 3.534412955465587, |
| "grad_norm": 0.24947026234779146, |
| "learning_rate": 9.594668352718152e-06, |
| "loss": 0.1902, |
| "step": 873 |
| }, |
| { |
| "epoch": 3.5384615384615383, |
| "grad_norm": 0.2594775419191058, |
| "learning_rate": 9.546412404157383e-06, |
| "loss": 0.1953, |
| "step": 874 |
| }, |
| { |
| "epoch": 3.54251012145749, |
| "grad_norm": 0.26739977818699795, |
| "learning_rate": 9.498240042204471e-06, |
| "loss": 0.1918, |
| "step": 875 |
| }, |
| { |
| "epoch": 3.5465587044534415, |
| "grad_norm": 0.25543807632323856, |
| "learning_rate": 9.450151652044339e-06, |
| "loss": 0.1957, |
| "step": 876 |
| }, |
| { |
| "epoch": 3.5506072874493926, |
| "grad_norm": 0.2631236818300369, |
| "learning_rate": 9.402147618190472e-06, |
| "loss": 0.1846, |
| "step": 877 |
| }, |
| { |
| "epoch": 3.554655870445344, |
| "grad_norm": 0.2594986376163518, |
| "learning_rate": 9.354228324481841e-06, |
| "loss": 0.1775, |
| "step": 878 |
| }, |
| { |
| "epoch": 3.5587044534412957, |
| "grad_norm": 0.27159319162995704, |
| "learning_rate": 9.306394154079852e-06, |
| "loss": 0.1676, |
| "step": 879 |
| }, |
| { |
| "epoch": 3.562753036437247, |
| "grad_norm": 0.2590305881670029, |
| "learning_rate": 9.258645489465254e-06, |
| "loss": 0.1783, |
| "step": 880 |
| }, |
| { |
| "epoch": 3.5668016194331984, |
| "grad_norm": 0.25247135865942233, |
| "learning_rate": 9.21098271243509e-06, |
| "loss": 0.1898, |
| "step": 881 |
| }, |
| { |
| "epoch": 3.57085020242915, |
| "grad_norm": 0.2617568168840038, |
| "learning_rate": 9.163406204099659e-06, |
| "loss": 0.1914, |
| "step": 882 |
| }, |
| { |
| "epoch": 3.574898785425101, |
| "grad_norm": 0.24679264262756792, |
| "learning_rate": 9.115916344879452e-06, |
| "loss": 0.1843, |
| "step": 883 |
| }, |
| { |
| "epoch": 3.5789473684210527, |
| "grad_norm": 0.25300238762793353, |
| "learning_rate": 9.068513514502126e-06, |
| "loss": 0.1887, |
| "step": 884 |
| }, |
| { |
| "epoch": 3.582995951417004, |
| "grad_norm": 0.26475591738336773, |
| "learning_rate": 9.021198091999439e-06, |
| "loss": 0.191, |
| "step": 885 |
| }, |
| { |
| "epoch": 3.5870445344129553, |
| "grad_norm": 0.24491083378364178, |
| "learning_rate": 8.973970455704251e-06, |
| "loss": 0.1936, |
| "step": 886 |
| }, |
| { |
| "epoch": 3.591093117408907, |
| "grad_norm": 0.24371403458528004, |
| "learning_rate": 8.926830983247495e-06, |
| "loss": 0.1809, |
| "step": 887 |
| }, |
| { |
| "epoch": 3.5951417004048585, |
| "grad_norm": 0.2687858074375461, |
| "learning_rate": 8.879780051555116e-06, |
| "loss": 0.1926, |
| "step": 888 |
| }, |
| { |
| "epoch": 3.5991902834008096, |
| "grad_norm": 0.25885684368249984, |
| "learning_rate": 8.83281803684512e-06, |
| "loss": 0.1848, |
| "step": 889 |
| }, |
| { |
| "epoch": 3.603238866396761, |
| "grad_norm": 0.2498444492743122, |
| "learning_rate": 8.785945314624532e-06, |
| "loss": 0.1859, |
| "step": 890 |
| }, |
| { |
| "epoch": 3.6072874493927127, |
| "grad_norm": 0.2554282827637628, |
| "learning_rate": 8.739162259686385e-06, |
| "loss": 0.172, |
| "step": 891 |
| }, |
| { |
| "epoch": 3.611336032388664, |
| "grad_norm": 0.2606034665592544, |
| "learning_rate": 8.692469246106734e-06, |
| "loss": 0.1812, |
| "step": 892 |
| }, |
| { |
| "epoch": 3.6153846153846154, |
| "grad_norm": 0.23440906659915284, |
| "learning_rate": 8.645866647241677e-06, |
| "loss": 0.178, |
| "step": 893 |
| }, |
| { |
| "epoch": 3.619433198380567, |
| "grad_norm": 0.25750848807752064, |
| "learning_rate": 8.599354835724348e-06, |
| "loss": 0.1815, |
| "step": 894 |
| }, |
| { |
| "epoch": 3.623481781376518, |
| "grad_norm": 0.2555175221394916, |
| "learning_rate": 8.552934183461958e-06, |
| "loss": 0.1891, |
| "step": 895 |
| }, |
| { |
| "epoch": 3.6275303643724697, |
| "grad_norm": 0.2538367526229815, |
| "learning_rate": 8.506605061632805e-06, |
| "loss": 0.1905, |
| "step": 896 |
| }, |
| { |
| "epoch": 3.6315789473684212, |
| "grad_norm": 0.24302917192210785, |
| "learning_rate": 8.460367840683308e-06, |
| "loss": 0.2003, |
| "step": 897 |
| }, |
| { |
| "epoch": 3.6356275303643724, |
| "grad_norm": 0.25248168324718046, |
| "learning_rate": 8.414222890325065e-06, |
| "loss": 0.1929, |
| "step": 898 |
| }, |
| { |
| "epoch": 3.639676113360324, |
| "grad_norm": 0.2535550664035451, |
| "learning_rate": 8.368170579531855e-06, |
| "loss": 0.1849, |
| "step": 899 |
| }, |
| { |
| "epoch": 3.6437246963562755, |
| "grad_norm": 0.24881623608052603, |
| "learning_rate": 8.322211276536733e-06, |
| "loss": 0.1875, |
| "step": 900 |
| }, |
| { |
| "epoch": 3.6477732793522266, |
| "grad_norm": 0.2652019311432668, |
| "learning_rate": 8.27634534882907e-06, |
| "loss": 0.2068, |
| "step": 901 |
| }, |
| { |
| "epoch": 3.651821862348178, |
| "grad_norm": 0.25190046562754165, |
| "learning_rate": 8.230573163151585e-06, |
| "loss": 0.1988, |
| "step": 902 |
| }, |
| { |
| "epoch": 3.6558704453441297, |
| "grad_norm": 0.24315075397070457, |
| "learning_rate": 8.184895085497466e-06, |
| "loss": 0.1841, |
| "step": 903 |
| }, |
| { |
| "epoch": 3.659919028340081, |
| "grad_norm": 0.2530166869297662, |
| "learning_rate": 8.139311481107401e-06, |
| "loss": 0.1856, |
| "step": 904 |
| }, |
| { |
| "epoch": 3.6639676113360324, |
| "grad_norm": 0.253158589171919, |
| "learning_rate": 8.093822714466686e-06, |
| "loss": 0.1962, |
| "step": 905 |
| }, |
| { |
| "epoch": 3.668016194331984, |
| "grad_norm": 0.23453103113446833, |
| "learning_rate": 8.048429149302271e-06, |
| "loss": 0.1793, |
| "step": 906 |
| }, |
| { |
| "epoch": 3.672064777327935, |
| "grad_norm": 0.25108865673168057, |
| "learning_rate": 8.003131148579902e-06, |
| "loss": 0.1972, |
| "step": 907 |
| }, |
| { |
| "epoch": 3.6761133603238867, |
| "grad_norm": 0.25777387192994466, |
| "learning_rate": 7.957929074501182e-06, |
| "loss": 0.1951, |
| "step": 908 |
| }, |
| { |
| "epoch": 3.6801619433198383, |
| "grad_norm": 0.2368758223900873, |
| "learning_rate": 7.912823288500695e-06, |
| "loss": 0.1845, |
| "step": 909 |
| }, |
| { |
| "epoch": 3.6842105263157894, |
| "grad_norm": 0.2417311783313552, |
| "learning_rate": 7.867814151243102e-06, |
| "loss": 0.1934, |
| "step": 910 |
| }, |
| { |
| "epoch": 3.688259109311741, |
| "grad_norm": 0.2602272308163576, |
| "learning_rate": 7.822902022620267e-06, |
| "loss": 0.183, |
| "step": 911 |
| }, |
| { |
| "epoch": 3.6923076923076925, |
| "grad_norm": 0.24875691747536283, |
| "learning_rate": 7.778087261748377e-06, |
| "loss": 0.1845, |
| "step": 912 |
| }, |
| { |
| "epoch": 3.6963562753036436, |
| "grad_norm": 0.24417795224163907, |
| "learning_rate": 7.733370226965056e-06, |
| "loss": 0.1917, |
| "step": 913 |
| }, |
| { |
| "epoch": 3.700404858299595, |
| "grad_norm": 0.25864835494889027, |
| "learning_rate": 7.688751275826527e-06, |
| "loss": 0.1852, |
| "step": 914 |
| }, |
| { |
| "epoch": 3.7044534412955468, |
| "grad_norm": 0.2495821501932974, |
| "learning_rate": 7.644230765104735e-06, |
| "loss": 0.1788, |
| "step": 915 |
| }, |
| { |
| "epoch": 3.708502024291498, |
| "grad_norm": 0.23635931193883597, |
| "learning_rate": 7.599809050784502e-06, |
| "loss": 0.186, |
| "step": 916 |
| }, |
| { |
| "epoch": 3.7125506072874495, |
| "grad_norm": 0.24557689613808192, |
| "learning_rate": 7.555486488060661e-06, |
| "loss": 0.1851, |
| "step": 917 |
| }, |
| { |
| "epoch": 3.716599190283401, |
| "grad_norm": 0.2437321507382009, |
| "learning_rate": 7.511263431335249e-06, |
| "loss": 0.2145, |
| "step": 918 |
| }, |
| { |
| "epoch": 3.720647773279352, |
| "grad_norm": 0.2421987661434455, |
| "learning_rate": 7.467140234214656e-06, |
| "loss": 0.1943, |
| "step": 919 |
| }, |
| { |
| "epoch": 3.7246963562753037, |
| "grad_norm": 0.22801170086401615, |
| "learning_rate": 7.423117249506777e-06, |
| "loss": 0.2148, |
| "step": 920 |
| }, |
| { |
| "epoch": 3.7287449392712553, |
| "grad_norm": 0.2576380172798157, |
| "learning_rate": 7.379194829218231e-06, |
| "loss": 0.1773, |
| "step": 921 |
| }, |
| { |
| "epoch": 3.7327935222672064, |
| "grad_norm": 0.2424238089089844, |
| "learning_rate": 7.335373324551518e-06, |
| "loss": 0.212, |
| "step": 922 |
| }, |
| { |
| "epoch": 3.736842105263158, |
| "grad_norm": 0.2463811676632385, |
| "learning_rate": 7.291653085902224e-06, |
| "loss": 0.1827, |
| "step": 923 |
| }, |
| { |
| "epoch": 3.7408906882591095, |
| "grad_norm": 0.229106254133146, |
| "learning_rate": 7.248034462856213e-06, |
| "loss": 0.1957, |
| "step": 924 |
| }, |
| { |
| "epoch": 3.7449392712550607, |
| "grad_norm": 0.24792778768139093, |
| "learning_rate": 7.20451780418683e-06, |
| "loss": 0.1933, |
| "step": 925 |
| }, |
| { |
| "epoch": 3.748987854251012, |
| "grad_norm": 0.24084956700253446, |
| "learning_rate": 7.161103457852128e-06, |
| "loss": 0.1985, |
| "step": 926 |
| }, |
| { |
| "epoch": 3.753036437246964, |
| "grad_norm": 0.24216206571245358, |
| "learning_rate": 7.117791770992046e-06, |
| "loss": 0.1687, |
| "step": 927 |
| }, |
| { |
| "epoch": 3.757085020242915, |
| "grad_norm": 0.24746651144115184, |
| "learning_rate": 7.074583089925688e-06, |
| "loss": 0.1759, |
| "step": 928 |
| }, |
| { |
| "epoch": 3.7611336032388665, |
| "grad_norm": 0.2436945604571861, |
| "learning_rate": 7.0314777601485155e-06, |
| "loss": 0.205, |
| "step": 929 |
| }, |
| { |
| "epoch": 3.765182186234818, |
| "grad_norm": 0.2397960746304773, |
| "learning_rate": 6.988476126329604e-06, |
| "loss": 0.1858, |
| "step": 930 |
| }, |
| { |
| "epoch": 3.769230769230769, |
| "grad_norm": 0.25447713131231164, |
| "learning_rate": 6.945578532308857e-06, |
| "loss": 0.1695, |
| "step": 931 |
| }, |
| { |
| "epoch": 3.7732793522267207, |
| "grad_norm": 0.24582981334677373, |
| "learning_rate": 6.902785321094301e-06, |
| "loss": 0.1878, |
| "step": 932 |
| }, |
| { |
| "epoch": 3.7773279352226723, |
| "grad_norm": 0.235104297042147, |
| "learning_rate": 6.8600968348593115e-06, |
| "loss": 0.1936, |
| "step": 933 |
| }, |
| { |
| "epoch": 3.7813765182186234, |
| "grad_norm": 0.23580168934063853, |
| "learning_rate": 6.817513414939892e-06, |
| "loss": 0.1867, |
| "step": 934 |
| }, |
| { |
| "epoch": 3.785425101214575, |
| "grad_norm": 0.24013217746438897, |
| "learning_rate": 6.775035401831924e-06, |
| "loss": 0.2125, |
| "step": 935 |
| }, |
| { |
| "epoch": 3.7894736842105265, |
| "grad_norm": 0.2488254780275488, |
| "learning_rate": 6.732663135188473e-06, |
| "loss": 0.1752, |
| "step": 936 |
| }, |
| { |
| "epoch": 3.7935222672064777, |
| "grad_norm": 0.23215948307458625, |
| "learning_rate": 6.690396953817055e-06, |
| "loss": 0.1802, |
| "step": 937 |
| }, |
| { |
| "epoch": 3.7975708502024292, |
| "grad_norm": 0.2610949306000269, |
| "learning_rate": 6.648237195676927e-06, |
| "loss": 0.1768, |
| "step": 938 |
| }, |
| { |
| "epoch": 3.801619433198381, |
| "grad_norm": 0.2441946920047564, |
| "learning_rate": 6.606184197876395e-06, |
| "loss": 0.1777, |
| "step": 939 |
| }, |
| { |
| "epoch": 3.805668016194332, |
| "grad_norm": 0.24082876198079567, |
| "learning_rate": 6.564238296670109e-06, |
| "loss": 0.1956, |
| "step": 940 |
| }, |
| { |
| "epoch": 3.8097165991902835, |
| "grad_norm": 0.2540306996646904, |
| "learning_rate": 6.52239982745638e-06, |
| "loss": 0.1746, |
| "step": 941 |
| }, |
| { |
| "epoch": 3.813765182186235, |
| "grad_norm": 0.23611683359849447, |
| "learning_rate": 6.480669124774484e-06, |
| "loss": 0.1991, |
| "step": 942 |
| }, |
| { |
| "epoch": 3.817813765182186, |
| "grad_norm": 0.24142281750994146, |
| "learning_rate": 6.439046522302013e-06, |
| "loss": 0.1791, |
| "step": 943 |
| }, |
| { |
| "epoch": 3.8218623481781377, |
| "grad_norm": 0.23731420512772208, |
| "learning_rate": 6.397532352852191e-06, |
| "loss": 0.1815, |
| "step": 944 |
| }, |
| { |
| "epoch": 3.8259109311740893, |
| "grad_norm": 0.23974963984373066, |
| "learning_rate": 6.356126948371202e-06, |
| "loss": 0.1889, |
| "step": 945 |
| }, |
| { |
| "epoch": 3.8299595141700404, |
| "grad_norm": 0.2545173836262436, |
| "learning_rate": 6.314830639935559e-06, |
| "loss": 0.1756, |
| "step": 946 |
| }, |
| { |
| "epoch": 3.834008097165992, |
| "grad_norm": 0.2475997952423021, |
| "learning_rate": 6.273643757749448e-06, |
| "loss": 0.1799, |
| "step": 947 |
| }, |
| { |
| "epoch": 3.8380566801619436, |
| "grad_norm": 0.23336227165300505, |
| "learning_rate": 6.2325666311420856e-06, |
| "loss": 0.2046, |
| "step": 948 |
| }, |
| { |
| "epoch": 3.8421052631578947, |
| "grad_norm": 0.235600451094766, |
| "learning_rate": 6.1915995885650675e-06, |
| "loss": 0.19, |
| "step": 949 |
| }, |
| { |
| "epoch": 3.8461538461538463, |
| "grad_norm": 0.2503232426379318, |
| "learning_rate": 6.150742957589784e-06, |
| "loss": 0.1592, |
| "step": 950 |
| }, |
| { |
| "epoch": 3.850202429149798, |
| "grad_norm": 0.24859182854858833, |
| "learning_rate": 6.109997064904774e-06, |
| "loss": 0.1828, |
| "step": 951 |
| }, |
| { |
| "epoch": 3.854251012145749, |
| "grad_norm": 0.24331898297039106, |
| "learning_rate": 6.069362236313092e-06, |
| "loss": 0.1747, |
| "step": 952 |
| }, |
| { |
| "epoch": 3.8582995951417005, |
| "grad_norm": 0.25655100355848315, |
| "learning_rate": 6.0288387967297565e-06, |
| "loss": 0.1885, |
| "step": 953 |
| }, |
| { |
| "epoch": 3.862348178137652, |
| "grad_norm": 0.23350802114821323, |
| "learning_rate": 5.988427070179108e-06, |
| "loss": 0.181, |
| "step": 954 |
| }, |
| { |
| "epoch": 3.866396761133603, |
| "grad_norm": 0.23812243557896376, |
| "learning_rate": 5.948127379792234e-06, |
| "loss": 0.1915, |
| "step": 955 |
| }, |
| { |
| "epoch": 3.8704453441295548, |
| "grad_norm": 0.25350694990209094, |
| "learning_rate": 5.90794004780439e-06, |
| "loss": 0.1765, |
| "step": 956 |
| }, |
| { |
| "epoch": 3.8744939271255063, |
| "grad_norm": 0.2480191639566577, |
| "learning_rate": 5.867865395552405e-06, |
| "loss": 0.1934, |
| "step": 957 |
| }, |
| { |
| "epoch": 3.8785425101214575, |
| "grad_norm": 0.2333883984296886, |
| "learning_rate": 5.8279037434721364e-06, |
| "loss": 0.178, |
| "step": 958 |
| }, |
| { |
| "epoch": 3.882591093117409, |
| "grad_norm": 0.2443089036323067, |
| "learning_rate": 5.788055411095892e-06, |
| "loss": 0.1895, |
| "step": 959 |
| }, |
| { |
| "epoch": 3.8866396761133606, |
| "grad_norm": 0.22976708881730634, |
| "learning_rate": 5.7483207170498625e-06, |
| "loss": 0.1965, |
| "step": 960 |
| }, |
| { |
| "epoch": 3.8906882591093117, |
| "grad_norm": 0.23344310676700789, |
| "learning_rate": 5.708699979051613e-06, |
| "loss": 0.2041, |
| "step": 961 |
| }, |
| { |
| "epoch": 3.8947368421052633, |
| "grad_norm": 0.2471798518773799, |
| "learning_rate": 5.66919351390751e-06, |
| "loss": 0.1805, |
| "step": 962 |
| }, |
| { |
| "epoch": 3.898785425101215, |
| "grad_norm": 0.24464055284171626, |
| "learning_rate": 5.629801637510188e-06, |
| "loss": 0.1888, |
| "step": 963 |
| }, |
| { |
| "epoch": 3.902834008097166, |
| "grad_norm": 0.24044524827840952, |
| "learning_rate": 5.590524664836043e-06, |
| "loss": 0.1941, |
| "step": 964 |
| }, |
| { |
| "epoch": 3.9068825910931175, |
| "grad_norm": 0.2408393218793809, |
| "learning_rate": 5.551362909942708e-06, |
| "loss": 0.1919, |
| "step": 965 |
| }, |
| { |
| "epoch": 3.910931174089069, |
| "grad_norm": 0.24315196103179293, |
| "learning_rate": 5.51231668596653e-06, |
| "loss": 0.2021, |
| "step": 966 |
| }, |
| { |
| "epoch": 3.91497975708502, |
| "grad_norm": 0.23521236262899844, |
| "learning_rate": 5.47338630512007e-06, |
| "loss": 0.1943, |
| "step": 967 |
| }, |
| { |
| "epoch": 3.919028340080972, |
| "grad_norm": 0.24202357933323568, |
| "learning_rate": 5.434572078689617e-06, |
| "loss": 0.1668, |
| "step": 968 |
| }, |
| { |
| "epoch": 3.9230769230769234, |
| "grad_norm": 0.23418289518859278, |
| "learning_rate": 5.395874317032694e-06, |
| "loss": 0.2039, |
| "step": 969 |
| }, |
| { |
| "epoch": 3.9271255060728745, |
| "grad_norm": 0.23907651311577047, |
| "learning_rate": 5.357293329575573e-06, |
| "loss": 0.1761, |
| "step": 970 |
| }, |
| { |
| "epoch": 3.931174089068826, |
| "grad_norm": 0.2459861060053384, |
| "learning_rate": 5.318829424810797e-06, |
| "loss": 0.1772, |
| "step": 971 |
| }, |
| { |
| "epoch": 3.9352226720647776, |
| "grad_norm": 0.2405968484955578, |
| "learning_rate": 5.280482910294728e-06, |
| "loss": 0.1986, |
| "step": 972 |
| }, |
| { |
| "epoch": 3.9392712550607287, |
| "grad_norm": 0.24185780579171776, |
| "learning_rate": 5.242254092645078e-06, |
| "loss": 0.2003, |
| "step": 973 |
| }, |
| { |
| "epoch": 3.9433198380566803, |
| "grad_norm": 0.23620304458042501, |
| "learning_rate": 5.2041432775384445e-06, |
| "loss": 0.2058, |
| "step": 974 |
| }, |
| { |
| "epoch": 3.9473684210526314, |
| "grad_norm": 0.23151164550490874, |
| "learning_rate": 5.166150769707892e-06, |
| "loss": 0.1762, |
| "step": 975 |
| }, |
| { |
| "epoch": 3.951417004048583, |
| "grad_norm": 0.24330828160846094, |
| "learning_rate": 5.128276872940503e-06, |
| "loss": 0.1919, |
| "step": 976 |
| }, |
| { |
| "epoch": 3.9554655870445345, |
| "grad_norm": 0.24420553023541114, |
| "learning_rate": 5.090521890074953e-06, |
| "loss": 0.1737, |
| "step": 977 |
| }, |
| { |
| "epoch": 3.9595141700404857, |
| "grad_norm": 0.23312457822019156, |
| "learning_rate": 5.0528861229990675e-06, |
| "loss": 0.174, |
| "step": 978 |
| }, |
| { |
| "epoch": 3.9635627530364372, |
| "grad_norm": 0.25817886423534936, |
| "learning_rate": 5.015369872647442e-06, |
| "loss": 0.1767, |
| "step": 979 |
| }, |
| { |
| "epoch": 3.967611336032389, |
| "grad_norm": 0.24795981581307322, |
| "learning_rate": 4.9779734389990195e-06, |
| "loss": 0.1805, |
| "step": 980 |
| }, |
| { |
| "epoch": 3.97165991902834, |
| "grad_norm": 0.23551731559932648, |
| "learning_rate": 4.940697121074675e-06, |
| "loss": 0.1997, |
| "step": 981 |
| }, |
| { |
| "epoch": 3.9757085020242915, |
| "grad_norm": 0.23517847899787137, |
| "learning_rate": 4.903541216934864e-06, |
| "loss": 0.1804, |
| "step": 982 |
| }, |
| { |
| "epoch": 3.979757085020243, |
| "grad_norm": 0.23356953728134008, |
| "learning_rate": 4.8665060236772e-06, |
| "loss": 0.1865, |
| "step": 983 |
| }, |
| { |
| "epoch": 3.983805668016194, |
| "grad_norm": 0.24173933555083155, |
| "learning_rate": 4.829591837434102e-06, |
| "loss": 0.1953, |
| "step": 984 |
| }, |
| { |
| "epoch": 3.9878542510121457, |
| "grad_norm": 0.23802657276606687, |
| "learning_rate": 4.792798953370424e-06, |
| "loss": 0.174, |
| "step": 985 |
| }, |
| { |
| "epoch": 3.9919028340080973, |
| "grad_norm": 0.2507488813454155, |
| "learning_rate": 4.756127665681083e-06, |
| "loss": 0.1837, |
| "step": 986 |
| }, |
| { |
| "epoch": 3.9959514170040484, |
| "grad_norm": 0.22992460970915712, |
| "learning_rate": 4.719578267588725e-06, |
| "loss": 0.1914, |
| "step": 987 |
| }, |
| { |
| "epoch": 4.0, |
| "grad_norm": 0.23009808138159543, |
| "learning_rate": 4.6831510513413525e-06, |
| "loss": 0.186, |
| "step": 988 |
| }, |
| { |
| "epoch": 4.004048582995951, |
| "grad_norm": 0.3493295049746174, |
| "learning_rate": 4.6468463082100176e-06, |
| "loss": 0.1636, |
| "step": 989 |
| }, |
| { |
| "epoch": 4.008097165991903, |
| "grad_norm": 0.28373988739122147, |
| "learning_rate": 4.610664328486485e-06, |
| "loss": 0.1428, |
| "step": 990 |
| }, |
| { |
| "epoch": 4.012145748987854, |
| "grad_norm": 0.21073051748790048, |
| "learning_rate": 4.5746054014809e-06, |
| "loss": 0.1605, |
| "step": 991 |
| }, |
| { |
| "epoch": 4.016194331983805, |
| "grad_norm": 0.23401410640174045, |
| "learning_rate": 4.5386698155194766e-06, |
| "loss": 0.1533, |
| "step": 992 |
| }, |
| { |
| "epoch": 4.020242914979757, |
| "grad_norm": 0.3108134435982764, |
| "learning_rate": 4.502857857942204e-06, |
| "loss": 0.1557, |
| "step": 993 |
| }, |
| { |
| "epoch": 4.0242914979757085, |
| "grad_norm": 0.36186024939896005, |
| "learning_rate": 4.467169815100549e-06, |
| "loss": 0.1641, |
| "step": 994 |
| }, |
| { |
| "epoch": 4.02834008097166, |
| "grad_norm": 0.2837340789468728, |
| "learning_rate": 4.431605972355137e-06, |
| "loss": 0.1548, |
| "step": 995 |
| }, |
| { |
| "epoch": 4.032388663967612, |
| "grad_norm": 0.23203499301541947, |
| "learning_rate": 4.396166614073516e-06, |
| "loss": 0.1503, |
| "step": 996 |
| }, |
| { |
| "epoch": 4.036437246963563, |
| "grad_norm": 0.2658535235411011, |
| "learning_rate": 4.360852023627846e-06, |
| "loss": 0.1557, |
| "step": 997 |
| }, |
| { |
| "epoch": 4.040485829959514, |
| "grad_norm": 0.25959527065832866, |
| "learning_rate": 4.325662483392656e-06, |
| "loss": 0.1349, |
| "step": 998 |
| }, |
| { |
| "epoch": 4.044534412955466, |
| "grad_norm": 0.30077003680352643, |
| "learning_rate": 4.29059827474257e-06, |
| "loss": 0.1519, |
| "step": 999 |
| }, |
| { |
| "epoch": 4.048582995951417, |
| "grad_norm": 0.23615015732568612, |
| "learning_rate": 4.255659678050063e-06, |
| "loss": 0.1433, |
| "step": 1000 |
| }, |
| { |
| "epoch": 4.052631578947368, |
| "grad_norm": 0.22338272349227997, |
| "learning_rate": 4.22084697268323e-06, |
| "loss": 0.1381, |
| "step": 1001 |
| }, |
| { |
| "epoch": 4.05668016194332, |
| "grad_norm": 0.24463510243344336, |
| "learning_rate": 4.186160437003535e-06, |
| "loss": 0.1582, |
| "step": 1002 |
| }, |
| { |
| "epoch": 4.060728744939271, |
| "grad_norm": 0.25856134826117755, |
| "learning_rate": 4.151600348363582e-06, |
| "loss": 0.1389, |
| "step": 1003 |
| }, |
| { |
| "epoch": 4.064777327935222, |
| "grad_norm": 0.25372889599699866, |
| "learning_rate": 4.11716698310493e-06, |
| "loss": 0.1358, |
| "step": 1004 |
| }, |
| { |
| "epoch": 4.068825910931174, |
| "grad_norm": 0.2270257584363236, |
| "learning_rate": 4.082860616555848e-06, |
| "loss": 0.1647, |
| "step": 1005 |
| }, |
| { |
| "epoch": 4.0728744939271255, |
| "grad_norm": 0.22428176254603252, |
| "learning_rate": 4.048681523029125e-06, |
| "loss": 0.1518, |
| "step": 1006 |
| }, |
| { |
| "epoch": 4.076923076923077, |
| "grad_norm": 0.22409331153943196, |
| "learning_rate": 4.014629975819882e-06, |
| "loss": 0.1481, |
| "step": 1007 |
| }, |
| { |
| "epoch": 4.080971659919029, |
| "grad_norm": 0.21250514996113515, |
| "learning_rate": 3.9807062472033855e-06, |
| "loss": 0.1579, |
| "step": 1008 |
| }, |
| { |
| "epoch": 4.08502024291498, |
| "grad_norm": 0.21422233855716882, |
| "learning_rate": 3.94691060843287e-06, |
| "loss": 0.1707, |
| "step": 1009 |
| }, |
| { |
| "epoch": 4.089068825910931, |
| "grad_norm": 0.21902419292044212, |
| "learning_rate": 3.913243329737353e-06, |
| "loss": 0.147, |
| "step": 1010 |
| }, |
| { |
| "epoch": 4.093117408906883, |
| "grad_norm": 0.21091289554786577, |
| "learning_rate": 3.879704680319503e-06, |
| "loss": 0.1519, |
| "step": 1011 |
| }, |
| { |
| "epoch": 4.097165991902834, |
| "grad_norm": 0.21415584572238386, |
| "learning_rate": 3.846294928353471e-06, |
| "loss": 0.145, |
| "step": 1012 |
| }, |
| { |
| "epoch": 4.101214574898785, |
| "grad_norm": 0.21361762282725044, |
| "learning_rate": 3.8130143409827345e-06, |
| "loss": 0.1519, |
| "step": 1013 |
| }, |
| { |
| "epoch": 4.105263157894737, |
| "grad_norm": 0.2059782141712376, |
| "learning_rate": 3.7798631843179887e-06, |
| "loss": 0.1504, |
| "step": 1014 |
| }, |
| { |
| "epoch": 4.109311740890688, |
| "grad_norm": 0.2091679330930599, |
| "learning_rate": 3.7468417234349973e-06, |
| "loss": 0.1454, |
| "step": 1015 |
| }, |
| { |
| "epoch": 4.113360323886639, |
| "grad_norm": 0.20281213656472372, |
| "learning_rate": 3.7139502223724823e-06, |
| "loss": 0.1507, |
| "step": 1016 |
| }, |
| { |
| "epoch": 4.117408906882591, |
| "grad_norm": 0.20869868255335672, |
| "learning_rate": 3.6811889441300095e-06, |
| "loss": 0.1475, |
| "step": 1017 |
| }, |
| { |
| "epoch": 4.1214574898785425, |
| "grad_norm": 0.2142019283332993, |
| "learning_rate": 3.648558150665886e-06, |
| "loss": 0.1544, |
| "step": 1018 |
| }, |
| { |
| "epoch": 4.125506072874494, |
| "grad_norm": 0.21200834750412284, |
| "learning_rate": 3.616058102895075e-06, |
| "loss": 0.1584, |
| "step": 1019 |
| }, |
| { |
| "epoch": 4.129554655870446, |
| "grad_norm": 0.2037493856035483, |
| "learning_rate": 3.583689060687079e-06, |
| "loss": 0.1284, |
| "step": 1020 |
| }, |
| { |
| "epoch": 4.133603238866397, |
| "grad_norm": 0.2006235761383419, |
| "learning_rate": 3.5514512828639023e-06, |
| "loss": 0.1542, |
| "step": 1021 |
| }, |
| { |
| "epoch": 4.137651821862348, |
| "grad_norm": 0.2095297147654613, |
| "learning_rate": 3.51934502719796e-06, |
| "loss": 0.1466, |
| "step": 1022 |
| }, |
| { |
| "epoch": 4.1417004048583, |
| "grad_norm": 0.2016581417602334, |
| "learning_rate": 3.4873705504100185e-06, |
| "loss": 0.1457, |
| "step": 1023 |
| }, |
| { |
| "epoch": 4.145748987854251, |
| "grad_norm": 0.20533522032575588, |
| "learning_rate": 3.4555281081671345e-06, |
| "loss": 0.1221, |
| "step": 1024 |
| }, |
| { |
| "epoch": 4.149797570850202, |
| "grad_norm": 0.1993053775712917, |
| "learning_rate": 3.4238179550806326e-06, |
| "loss": 0.1469, |
| "step": 1025 |
| }, |
| { |
| "epoch": 4.153846153846154, |
| "grad_norm": 0.22384444044191504, |
| "learning_rate": 3.3922403447040497e-06, |
| "loss": 0.1504, |
| "step": 1026 |
| }, |
| { |
| "epoch": 4.157894736842105, |
| "grad_norm": 0.19898308590723926, |
| "learning_rate": 3.3607955295311247e-06, |
| "loss": 0.1401, |
| "step": 1027 |
| }, |
| { |
| "epoch": 4.161943319838056, |
| "grad_norm": 0.20232863421651087, |
| "learning_rate": 3.32948376099375e-06, |
| "loss": 0.1396, |
| "step": 1028 |
| }, |
| { |
| "epoch": 4.165991902834008, |
| "grad_norm": 0.20996037267037201, |
| "learning_rate": 3.2983052894599976e-06, |
| "loss": 0.135, |
| "step": 1029 |
| }, |
| { |
| "epoch": 4.17004048582996, |
| "grad_norm": 0.20433303496823865, |
| "learning_rate": 3.2672603642320945e-06, |
| "loss": 0.1562, |
| "step": 1030 |
| }, |
| { |
| "epoch": 4.174089068825911, |
| "grad_norm": 0.20156473885620968, |
| "learning_rate": 3.236349233544438e-06, |
| "loss": 0.1256, |
| "step": 1031 |
| }, |
| { |
| "epoch": 4.178137651821863, |
| "grad_norm": 0.19769622315718083, |
| "learning_rate": 3.205572144561604e-06, |
| "loss": 0.1452, |
| "step": 1032 |
| }, |
| { |
| "epoch": 4.182186234817814, |
| "grad_norm": 0.20363184191599978, |
| "learning_rate": 3.174929343376374e-06, |
| "loss": 0.137, |
| "step": 1033 |
| }, |
| { |
| "epoch": 4.186234817813765, |
| "grad_norm": 0.20704680007860868, |
| "learning_rate": 3.1444210750077777e-06, |
| "loss": 0.1402, |
| "step": 1034 |
| }, |
| { |
| "epoch": 4.190283400809717, |
| "grad_norm": 0.20010893112430597, |
| "learning_rate": 3.1140475833991046e-06, |
| "loss": 0.1483, |
| "step": 1035 |
| }, |
| { |
| "epoch": 4.194331983805668, |
| "grad_norm": 0.20121354476574874, |
| "learning_rate": 3.0838091114159895e-06, |
| "loss": 0.1475, |
| "step": 1036 |
| }, |
| { |
| "epoch": 4.198380566801619, |
| "grad_norm": 0.20070529704635628, |
| "learning_rate": 3.0537059008444524e-06, |
| "loss": 0.1395, |
| "step": 1037 |
| }, |
| { |
| "epoch": 4.202429149797571, |
| "grad_norm": 0.1974965736701602, |
| "learning_rate": 3.0237381923889563e-06, |
| "loss": 0.1554, |
| "step": 1038 |
| }, |
| { |
| "epoch": 4.206477732793522, |
| "grad_norm": 0.2019563747713323, |
| "learning_rate": 2.9939062256704997e-06, |
| "loss": 0.1413, |
| "step": 1039 |
| }, |
| { |
| "epoch": 4.2105263157894735, |
| "grad_norm": 0.210582277700481, |
| "learning_rate": 2.964210239224696e-06, |
| "loss": 0.1535, |
| "step": 1040 |
| }, |
| { |
| "epoch": 4.2145748987854255, |
| "grad_norm": 0.2112700800261687, |
| "learning_rate": 2.934650470499869e-06, |
| "loss": 0.1486, |
| "step": 1041 |
| }, |
| { |
| "epoch": 4.218623481781377, |
| "grad_norm": 0.20669010643355634, |
| "learning_rate": 2.905227155855128e-06, |
| "loss": 0.1395, |
| "step": 1042 |
| }, |
| { |
| "epoch": 4.222672064777328, |
| "grad_norm": 0.2000388338193207, |
| "learning_rate": 2.875940530558519e-06, |
| "loss": 0.1362, |
| "step": 1043 |
| }, |
| { |
| "epoch": 4.22672064777328, |
| "grad_norm": 0.2028080414090865, |
| "learning_rate": 2.8467908287851154e-06, |
| "loss": 0.1604, |
| "step": 1044 |
| }, |
| { |
| "epoch": 4.230769230769231, |
| "grad_norm": 0.19831467865347488, |
| "learning_rate": 2.8177782836151603e-06, |
| "loss": 0.1595, |
| "step": 1045 |
| }, |
| { |
| "epoch": 4.234817813765182, |
| "grad_norm": 0.19594300171024379, |
| "learning_rate": 2.7889031270321874e-06, |
| "loss": 0.1653, |
| "step": 1046 |
| }, |
| { |
| "epoch": 4.238866396761134, |
| "grad_norm": 0.21041844733527512, |
| "learning_rate": 2.760165589921184e-06, |
| "loss": 0.1532, |
| "step": 1047 |
| }, |
| { |
| "epoch": 4.242914979757085, |
| "grad_norm": 0.2108760399084688, |
| "learning_rate": 2.7315659020667305e-06, |
| "loss": 0.1593, |
| "step": 1048 |
| }, |
| { |
| "epoch": 4.246963562753036, |
| "grad_norm": 0.19323719230061195, |
| "learning_rate": 2.703104292151164e-06, |
| "loss": 0.1461, |
| "step": 1049 |
| }, |
| { |
| "epoch": 4.251012145748988, |
| "grad_norm": 0.19689447073596691, |
| "learning_rate": 2.6747809877527607e-06, |
| "loss": 0.1633, |
| "step": 1050 |
| }, |
| { |
| "epoch": 4.255060728744939, |
| "grad_norm": 0.21306810782086238, |
| "learning_rate": 2.6465962153439082e-06, |
| "loss": 0.1538, |
| "step": 1051 |
| }, |
| { |
| "epoch": 4.2591093117408905, |
| "grad_norm": 0.20603918917617794, |
| "learning_rate": 2.618550200289303e-06, |
| "loss": 0.1288, |
| "step": 1052 |
| }, |
| { |
| "epoch": 4.2631578947368425, |
| "grad_norm": 0.1938787628293216, |
| "learning_rate": 2.59064316684412e-06, |
| "loss": 0.135, |
| "step": 1053 |
| }, |
| { |
| "epoch": 4.267206477732794, |
| "grad_norm": 0.21056124478270535, |
| "learning_rate": 2.562875338152264e-06, |
| "loss": 0.1418, |
| "step": 1054 |
| }, |
| { |
| "epoch": 4.271255060728745, |
| "grad_norm": 0.21632240523416343, |
| "learning_rate": 2.5352469362445552e-06, |
| "loss": 0.1404, |
| "step": 1055 |
| }, |
| { |
| "epoch": 4.275303643724697, |
| "grad_norm": 0.20844763743668396, |
| "learning_rate": 2.5077581820369523e-06, |
| "loss": 0.1377, |
| "step": 1056 |
| }, |
| { |
| "epoch": 4.279352226720648, |
| "grad_norm": 0.20371307553008045, |
| "learning_rate": 2.480409295328803e-06, |
| "loss": 0.1604, |
| "step": 1057 |
| }, |
| { |
| "epoch": 4.283400809716599, |
| "grad_norm": 0.20612127988043505, |
| "learning_rate": 2.45320049480108e-06, |
| "loss": 0.1253, |
| "step": 1058 |
| }, |
| { |
| "epoch": 4.287449392712551, |
| "grad_norm": 0.20159691472638194, |
| "learning_rate": 2.4261319980146293e-06, |
| "loss": 0.1895, |
| "step": 1059 |
| }, |
| { |
| "epoch": 4.291497975708502, |
| "grad_norm": 0.21043866623790025, |
| "learning_rate": 2.399204021408421e-06, |
| "loss": 0.1213, |
| "step": 1060 |
| }, |
| { |
| "epoch": 4.295546558704453, |
| "grad_norm": 0.2007641534284875, |
| "learning_rate": 2.3724167802978395e-06, |
| "loss": 0.1359, |
| "step": 1061 |
| }, |
| { |
| "epoch": 4.299595141700405, |
| "grad_norm": 0.19737427424735798, |
| "learning_rate": 2.3457704888729625e-06, |
| "loss": 0.1584, |
| "step": 1062 |
| }, |
| { |
| "epoch": 4.303643724696356, |
| "grad_norm": 0.19610979308953908, |
| "learning_rate": 2.3192653601968162e-06, |
| "loss": 0.1443, |
| "step": 1063 |
| }, |
| { |
| "epoch": 4.3076923076923075, |
| "grad_norm": 0.20193605274823165, |
| "learning_rate": 2.2929016062037057e-06, |
| "loss": 0.1518, |
| "step": 1064 |
| }, |
| { |
| "epoch": 4.3117408906882595, |
| "grad_norm": 0.19428678404532834, |
| "learning_rate": 2.2666794376975055e-06, |
| "loss": 0.1547, |
| "step": 1065 |
| }, |
| { |
| "epoch": 4.315789473684211, |
| "grad_norm": 0.2030770934370617, |
| "learning_rate": 2.24059906434998e-06, |
| "loss": 0.1543, |
| "step": 1066 |
| }, |
| { |
| "epoch": 4.319838056680162, |
| "grad_norm": 0.1975925189643666, |
| "learning_rate": 2.214660694699091e-06, |
| "loss": 0.1435, |
| "step": 1067 |
| }, |
| { |
| "epoch": 4.323886639676114, |
| "grad_norm": 0.2163256546918373, |
| "learning_rate": 2.188864536147355e-06, |
| "loss": 0.1513, |
| "step": 1068 |
| }, |
| { |
| "epoch": 4.327935222672065, |
| "grad_norm": 0.20207027469671504, |
| "learning_rate": 2.1632107949601757e-06, |
| "loss": 0.1517, |
| "step": 1069 |
| }, |
| { |
| "epoch": 4.331983805668016, |
| "grad_norm": 0.20530273028740825, |
| "learning_rate": 2.137699676264171e-06, |
| "loss": 0.1357, |
| "step": 1070 |
| }, |
| { |
| "epoch": 4.336032388663968, |
| "grad_norm": 0.20422106614726784, |
| "learning_rate": 2.1123313840455738e-06, |
| "loss": 0.1425, |
| "step": 1071 |
| }, |
| { |
| "epoch": 4.340080971659919, |
| "grad_norm": 0.20594508815285986, |
| "learning_rate": 2.08710612114857e-06, |
| "loss": 0.1345, |
| "step": 1072 |
| }, |
| { |
| "epoch": 4.34412955465587, |
| "grad_norm": 0.1991433578453612, |
| "learning_rate": 2.0620240892736953e-06, |
| "loss": 0.1386, |
| "step": 1073 |
| }, |
| { |
| "epoch": 4.348178137651822, |
| "grad_norm": 0.2048291900076616, |
| "learning_rate": 2.0370854889762e-06, |
| "loss": 0.1566, |
| "step": 1074 |
| }, |
| { |
| "epoch": 4.352226720647773, |
| "grad_norm": 0.21457387966046473, |
| "learning_rate": 2.0122905196644726e-06, |
| "loss": 0.1628, |
| "step": 1075 |
| }, |
| { |
| "epoch": 4.3562753036437245, |
| "grad_norm": 0.2077735760785679, |
| "learning_rate": 1.987639379598427e-06, |
| "loss": 0.1512, |
| "step": 1076 |
| }, |
| { |
| "epoch": 4.3603238866396765, |
| "grad_norm": 0.2027748105452019, |
| "learning_rate": 1.963132265887919e-06, |
| "loss": 0.1405, |
| "step": 1077 |
| }, |
| { |
| "epoch": 4.364372469635628, |
| "grad_norm": 0.1954473321532602, |
| "learning_rate": 1.9387693744911784e-06, |
| "loss": 0.1623, |
| "step": 1078 |
| }, |
| { |
| "epoch": 4.368421052631579, |
| "grad_norm": 0.23889372711854945, |
| "learning_rate": 1.9145509002132368e-06, |
| "loss": 0.1406, |
| "step": 1079 |
| }, |
| { |
| "epoch": 4.372469635627531, |
| "grad_norm": 0.20079775889694204, |
| "learning_rate": 1.8904770367043723e-06, |
| "loss": 0.1487, |
| "step": 1080 |
| }, |
| { |
| "epoch": 4.376518218623482, |
| "grad_norm": 0.19963885187256342, |
| "learning_rate": 1.8665479764585504e-06, |
| "loss": 0.1547, |
| "step": 1081 |
| }, |
| { |
| "epoch": 4.380566801619433, |
| "grad_norm": 0.2040813096304476, |
| "learning_rate": 1.8427639108118977e-06, |
| "loss": 0.1264, |
| "step": 1082 |
| }, |
| { |
| "epoch": 4.384615384615385, |
| "grad_norm": 0.2037223508827178, |
| "learning_rate": 1.8191250299411755e-06, |
| "loss": 0.1501, |
| "step": 1083 |
| }, |
| { |
| "epoch": 4.388663967611336, |
| "grad_norm": 0.20611515132825292, |
| "learning_rate": 1.7956315228622446e-06, |
| "loss": 0.1263, |
| "step": 1084 |
| }, |
| { |
| "epoch": 4.392712550607287, |
| "grad_norm": 0.1988234742041305, |
| "learning_rate": 1.7722835774285553e-06, |
| "loss": 0.1557, |
| "step": 1085 |
| }, |
| { |
| "epoch": 4.396761133603239, |
| "grad_norm": 0.194443393384881, |
| "learning_rate": 1.7490813803296624e-06, |
| "loss": 0.1612, |
| "step": 1086 |
| }, |
| { |
| "epoch": 4.40080971659919, |
| "grad_norm": 0.2044687109910247, |
| "learning_rate": 1.726025117089718e-06, |
| "loss": 0.1372, |
| "step": 1087 |
| }, |
| { |
| "epoch": 4.4048582995951415, |
| "grad_norm": 0.20573230581411622, |
| "learning_rate": 1.70311497206598e-06, |
| "loss": 0.1541, |
| "step": 1088 |
| }, |
| { |
| "epoch": 4.4089068825910935, |
| "grad_norm": 0.20275160235205505, |
| "learning_rate": 1.6803511284473638e-06, |
| "loss": 0.161, |
| "step": 1089 |
| }, |
| { |
| "epoch": 4.412955465587045, |
| "grad_norm": 0.19489838869781842, |
| "learning_rate": 1.6577337682529538e-06, |
| "loss": 0.1436, |
| "step": 1090 |
| }, |
| { |
| "epoch": 4.417004048582996, |
| "grad_norm": 0.20313715682694086, |
| "learning_rate": 1.6352630723305617e-06, |
| "loss": 0.1669, |
| "step": 1091 |
| }, |
| { |
| "epoch": 4.421052631578947, |
| "grad_norm": 0.2000808168270224, |
| "learning_rate": 1.612939220355272e-06, |
| "loss": 0.1362, |
| "step": 1092 |
| }, |
| { |
| "epoch": 4.425101214574899, |
| "grad_norm": 0.1988620045865299, |
| "learning_rate": 1.5907623908280112e-06, |
| "loss": 0.1255, |
| "step": 1093 |
| }, |
| { |
| "epoch": 4.42914979757085, |
| "grad_norm": 0.20538314104241237, |
| "learning_rate": 1.568732761074121e-06, |
| "loss": 0.1499, |
| "step": 1094 |
| }, |
| { |
| "epoch": 4.433198380566802, |
| "grad_norm": 0.20213318013211137, |
| "learning_rate": 1.5468505072419237e-06, |
| "loss": 0.1507, |
| "step": 1095 |
| }, |
| { |
| "epoch": 4.437246963562753, |
| "grad_norm": 0.21591431691814852, |
| "learning_rate": 1.5251158043013448e-06, |
| "loss": 0.1271, |
| "step": 1096 |
| }, |
| { |
| "epoch": 4.441295546558704, |
| "grad_norm": 0.19872488136323746, |
| "learning_rate": 1.5035288260424885e-06, |
| "loss": 0.1299, |
| "step": 1097 |
| }, |
| { |
| "epoch": 4.445344129554655, |
| "grad_norm": 0.20556001968320264, |
| "learning_rate": 1.4820897450742623e-06, |
| "loss": 0.1476, |
| "step": 1098 |
| }, |
| { |
| "epoch": 4.449392712550607, |
| "grad_norm": 0.20269880684608996, |
| "learning_rate": 1.4607987328229767e-06, |
| "loss": 0.1441, |
| "step": 1099 |
| }, |
| { |
| "epoch": 4.4534412955465585, |
| "grad_norm": 0.18753518685883913, |
| "learning_rate": 1.4396559595310056e-06, |
| "loss": 0.1665, |
| "step": 1100 |
| }, |
| { |
| "epoch": 4.4574898785425106, |
| "grad_norm": 0.20112025586584434, |
| "learning_rate": 1.4186615942553973e-06, |
| "loss": 0.1388, |
| "step": 1101 |
| }, |
| { |
| "epoch": 4.461538461538462, |
| "grad_norm": 0.19888297140826935, |
| "learning_rate": 1.3978158048665424e-06, |
| "loss": 0.1427, |
| "step": 1102 |
| }, |
| { |
| "epoch": 4.465587044534413, |
| "grad_norm": 0.2032965435092053, |
| "learning_rate": 1.3771187580468115e-06, |
| "loss": 0.1543, |
| "step": 1103 |
| }, |
| { |
| "epoch": 4.469635627530364, |
| "grad_norm": 0.19371555849560787, |
| "learning_rate": 1.3565706192892392e-06, |
| "loss": 0.1447, |
| "step": 1104 |
| }, |
| { |
| "epoch": 4.473684210526316, |
| "grad_norm": 0.19368628910606164, |
| "learning_rate": 1.3361715528961993e-06, |
| "loss": 0.141, |
| "step": 1105 |
| }, |
| { |
| "epoch": 4.477732793522267, |
| "grad_norm": 0.20958942583086185, |
| "learning_rate": 1.3159217219780817e-06, |
| "loss": 0.1403, |
| "step": 1106 |
| }, |
| { |
| "epoch": 4.481781376518219, |
| "grad_norm": 0.21074860340002544, |
| "learning_rate": 1.2958212884519949e-06, |
| "loss": 0.121, |
| "step": 1107 |
| }, |
| { |
| "epoch": 4.48582995951417, |
| "grad_norm": 0.20298228683740496, |
| "learning_rate": 1.2758704130404675e-06, |
| "loss": 0.147, |
| "step": 1108 |
| }, |
| { |
| "epoch": 4.489878542510121, |
| "grad_norm": 0.1984004282080862, |
| "learning_rate": 1.2560692552701738e-06, |
| "loss": 0.1437, |
| "step": 1109 |
| }, |
| { |
| "epoch": 4.493927125506072, |
| "grad_norm": 0.1963090155531576, |
| "learning_rate": 1.2364179734706382e-06, |
| "loss": 0.1488, |
| "step": 1110 |
| }, |
| { |
| "epoch": 4.497975708502024, |
| "grad_norm": 0.1952118144032855, |
| "learning_rate": 1.2169167247729852e-06, |
| "loss": 0.1115, |
| "step": 1111 |
| }, |
| { |
| "epoch": 4.502024291497976, |
| "grad_norm": 0.20516953934989346, |
| "learning_rate": 1.1975656651086864e-06, |
| "loss": 0.1334, |
| "step": 1112 |
| }, |
| { |
| "epoch": 4.506072874493928, |
| "grad_norm": 0.19269021336710804, |
| "learning_rate": 1.1783649492082926e-06, |
| "loss": 0.1444, |
| "step": 1113 |
| }, |
| { |
| "epoch": 4.510121457489879, |
| "grad_norm": 0.1965329380410411, |
| "learning_rate": 1.1593147306002183e-06, |
| "loss": 0.1468, |
| "step": 1114 |
| }, |
| { |
| "epoch": 4.51417004048583, |
| "grad_norm": 0.2075593949230797, |
| "learning_rate": 1.140415161609505e-06, |
| "loss": 0.1186, |
| "step": 1115 |
| }, |
| { |
| "epoch": 4.518218623481781, |
| "grad_norm": 0.2029061294036998, |
| "learning_rate": 1.121666393356604e-06, |
| "loss": 0.1587, |
| "step": 1116 |
| }, |
| { |
| "epoch": 4.522267206477733, |
| "grad_norm": 0.20445829003538576, |
| "learning_rate": 1.1030685757561632e-06, |
| "loss": 0.1486, |
| "step": 1117 |
| }, |
| { |
| "epoch": 4.526315789473684, |
| "grad_norm": 0.19811857363039467, |
| "learning_rate": 1.0846218575158373e-06, |
| "loss": 0.1496, |
| "step": 1118 |
| }, |
| { |
| "epoch": 4.530364372469636, |
| "grad_norm": 0.20120579801797775, |
| "learning_rate": 1.066326386135097e-06, |
| "loss": 0.1419, |
| "step": 1119 |
| }, |
| { |
| "epoch": 4.534412955465587, |
| "grad_norm": 0.19642133585828164, |
| "learning_rate": 1.0481823079040398e-06, |
| "loss": 0.1408, |
| "step": 1120 |
| }, |
| { |
| "epoch": 4.538461538461538, |
| "grad_norm": 0.19714754109047178, |
| "learning_rate": 1.0301897679022344e-06, |
| "loss": 0.1629, |
| "step": 1121 |
| }, |
| { |
| "epoch": 4.5425101214574894, |
| "grad_norm": 0.19876711082922766, |
| "learning_rate": 1.0123489099975491e-06, |
| "loss": 0.1452, |
| "step": 1122 |
| }, |
| { |
| "epoch": 4.5465587044534415, |
| "grad_norm": 0.1989904267483094, |
| "learning_rate": 9.9465987684501e-07, |
| "loss": 0.1447, |
| "step": 1123 |
| }, |
| { |
| "epoch": 4.550607287449393, |
| "grad_norm": 0.20740323024018747, |
| "learning_rate": 9.771228098856534e-07, |
| "loss": 0.1472, |
| "step": 1124 |
| }, |
| { |
| "epoch": 4.554655870445345, |
| "grad_norm": 0.20686517332763282, |
| "learning_rate": 9.597378493454013e-07, |
| "loss": 0.162, |
| "step": 1125 |
| }, |
| { |
| "epoch": 4.558704453441296, |
| "grad_norm": 0.19884047687601197, |
| "learning_rate": 9.425051342339353e-07, |
| "loss": 0.1335, |
| "step": 1126 |
| }, |
| { |
| "epoch": 4.562753036437247, |
| "grad_norm": 0.1960499866548161, |
| "learning_rate": 9.254248023435864e-07, |
| "loss": 0.1575, |
| "step": 1127 |
| }, |
| { |
| "epoch": 4.566801619433198, |
| "grad_norm": 0.19677545276845584, |
| "learning_rate": 9.084969902482288e-07, |
| "loss": 0.1497, |
| "step": 1128 |
| }, |
| { |
| "epoch": 4.57085020242915, |
| "grad_norm": 0.19116337835164776, |
| "learning_rate": 8.917218333021993e-07, |
| "loss": 0.1465, |
| "step": 1129 |
| }, |
| { |
| "epoch": 4.574898785425101, |
| "grad_norm": 0.20109039634091597, |
| "learning_rate": 8.750994656392109e-07, |
| "loss": 0.1569, |
| "step": 1130 |
| }, |
| { |
| "epoch": 4.578947368421053, |
| "grad_norm": 0.1973022260092343, |
| "learning_rate": 8.58630020171265e-07, |
| "loss": 0.1515, |
| "step": 1131 |
| }, |
| { |
| "epoch": 4.582995951417004, |
| "grad_norm": 0.19898802767609944, |
| "learning_rate": 8.423136285876099e-07, |
| "loss": 0.142, |
| "step": 1132 |
| }, |
| { |
| "epoch": 4.587044534412955, |
| "grad_norm": 0.20158577422989993, |
| "learning_rate": 8.26150421353682e-07, |
| "loss": 0.1542, |
| "step": 1133 |
| }, |
| { |
| "epoch": 4.5910931174089065, |
| "grad_norm": 0.19313674003905554, |
| "learning_rate": 8.101405277100549e-07, |
| "loss": 0.1488, |
| "step": 1134 |
| }, |
| { |
| "epoch": 4.5951417004048585, |
| "grad_norm": 0.19949454872014402, |
| "learning_rate": 7.942840756714077e-07, |
| "loss": 0.1594, |
| "step": 1135 |
| }, |
| { |
| "epoch": 4.59919028340081, |
| "grad_norm": 0.1963972902151125, |
| "learning_rate": 7.785811920255093e-07, |
| "loss": 0.1347, |
| "step": 1136 |
| }, |
| { |
| "epoch": 4.603238866396762, |
| "grad_norm": 0.18886035205309482, |
| "learning_rate": 7.630320023322024e-07, |
| "loss": 0.1539, |
| "step": 1137 |
| }, |
| { |
| "epoch": 4.607287449392713, |
| "grad_norm": 0.19232674836654656, |
| "learning_rate": 7.476366309223903e-07, |
| "loss": 0.1531, |
| "step": 1138 |
| }, |
| { |
| "epoch": 4.611336032388664, |
| "grad_norm": 0.20132943291707164, |
| "learning_rate": 7.323952008970537e-07, |
| "loss": 0.1557, |
| "step": 1139 |
| }, |
| { |
| "epoch": 4.615384615384615, |
| "grad_norm": 0.19154933222766615, |
| "learning_rate": 7.173078341262618e-07, |
| "loss": 0.1489, |
| "step": 1140 |
| }, |
| { |
| "epoch": 4.619433198380567, |
| "grad_norm": 0.19744308199381744, |
| "learning_rate": 7.023746512481989e-07, |
| "loss": 0.1209, |
| "step": 1141 |
| }, |
| { |
| "epoch": 4.623481781376518, |
| "grad_norm": 0.18801003171094857, |
| "learning_rate": 6.875957716681902e-07, |
| "loss": 0.1473, |
| "step": 1142 |
| }, |
| { |
| "epoch": 4.62753036437247, |
| "grad_norm": 0.19795490489138975, |
| "learning_rate": 6.729713135577687e-07, |
| "loss": 0.1435, |
| "step": 1143 |
| }, |
| { |
| "epoch": 4.631578947368421, |
| "grad_norm": 0.1926002728231295, |
| "learning_rate": 6.585013938537143e-07, |
| "loss": 0.148, |
| "step": 1144 |
| }, |
| { |
| "epoch": 4.635627530364372, |
| "grad_norm": 0.18884719784879547, |
| "learning_rate": 6.441861282571115e-07, |
| "loss": 0.1424, |
| "step": 1145 |
| }, |
| { |
| "epoch": 4.6396761133603235, |
| "grad_norm": 0.19739667514403006, |
| "learning_rate": 6.300256312324471e-07, |
| "loss": 0.1301, |
| "step": 1146 |
| }, |
| { |
| "epoch": 4.6437246963562755, |
| "grad_norm": 0.20185829638236458, |
| "learning_rate": 6.160200160066776e-07, |
| "loss": 0.1282, |
| "step": 1147 |
| }, |
| { |
| "epoch": 4.647773279352227, |
| "grad_norm": 0.2029747715398886, |
| "learning_rate": 6.021693945683327e-07, |
| "loss": 0.146, |
| "step": 1148 |
| }, |
| { |
| "epoch": 4.651821862348179, |
| "grad_norm": 0.200327418026789, |
| "learning_rate": 5.884738776666044e-07, |
| "loss": 0.1577, |
| "step": 1149 |
| }, |
| { |
| "epoch": 4.65587044534413, |
| "grad_norm": 0.2000870987773498, |
| "learning_rate": 5.749335748104878e-07, |
| "loss": 0.1411, |
| "step": 1150 |
| }, |
| { |
| "epoch": 4.659919028340081, |
| "grad_norm": 0.1922253680386575, |
| "learning_rate": 5.615485942678778e-07, |
| "loss": 0.1528, |
| "step": 1151 |
| }, |
| { |
| "epoch": 4.663967611336032, |
| "grad_norm": 0.1979448897154144, |
| "learning_rate": 5.483190430647289e-07, |
| "loss": 0.1403, |
| "step": 1152 |
| }, |
| { |
| "epoch": 4.668016194331984, |
| "grad_norm": 0.20351886519518553, |
| "learning_rate": 5.352450269841747e-07, |
| "loss": 0.1489, |
| "step": 1153 |
| }, |
| { |
| "epoch": 4.672064777327935, |
| "grad_norm": 0.20239701358510956, |
| "learning_rate": 5.22326650565701e-07, |
| "loss": 0.127, |
| "step": 1154 |
| }, |
| { |
| "epoch": 4.676113360323887, |
| "grad_norm": 0.19341793846116992, |
| "learning_rate": 5.095640171043048e-07, |
| "loss": 0.1485, |
| "step": 1155 |
| }, |
| { |
| "epoch": 4.680161943319838, |
| "grad_norm": 0.20156999264837647, |
| "learning_rate": 4.969572286496594e-07, |
| "loss": 0.1374, |
| "step": 1156 |
| }, |
| { |
| "epoch": 4.684210526315789, |
| "grad_norm": 0.20584456571283233, |
| "learning_rate": 4.845063860053079e-07, |
| "loss": 0.1447, |
| "step": 1157 |
| }, |
| { |
| "epoch": 4.6882591093117405, |
| "grad_norm": 0.19130370533218305, |
| "learning_rate": 4.7221158872785997e-07, |
| "loss": 0.1641, |
| "step": 1158 |
| }, |
| { |
| "epoch": 4.6923076923076925, |
| "grad_norm": 0.1910620212977208, |
| "learning_rate": 4.600729351261857e-07, |
| "loss": 0.151, |
| "step": 1159 |
| }, |
| { |
| "epoch": 4.696356275303644, |
| "grad_norm": 0.1957744519391178, |
| "learning_rate": 4.4809052226063355e-07, |
| "loss": 0.1445, |
| "step": 1160 |
| }, |
| { |
| "epoch": 4.700404858299595, |
| "grad_norm": 0.1973305458488524, |
| "learning_rate": 4.3626444594226046e-07, |
| "loss": 0.1541, |
| "step": 1161 |
| }, |
| { |
| "epoch": 4.704453441295547, |
| "grad_norm": 0.1912422759397071, |
| "learning_rate": 4.245948007320588e-07, |
| "loss": 0.1439, |
| "step": 1162 |
| }, |
| { |
| "epoch": 4.708502024291498, |
| "grad_norm": 0.19631224697463978, |
| "learning_rate": 4.130816799401971e-07, |
| "loss": 0.1506, |
| "step": 1163 |
| }, |
| { |
| "epoch": 4.712550607287449, |
| "grad_norm": 0.1986990470973571, |
| "learning_rate": 4.017251756252827e-07, |
| "loss": 0.1502, |
| "step": 1164 |
| }, |
| { |
| "epoch": 4.716599190283401, |
| "grad_norm": 0.19358467133358553, |
| "learning_rate": 3.9052537859362473e-07, |
| "loss": 0.1373, |
| "step": 1165 |
| }, |
| { |
| "epoch": 4.720647773279352, |
| "grad_norm": 0.19290534744405652, |
| "learning_rate": 3.7948237839849687e-07, |
| "loss": 0.1573, |
| "step": 1166 |
| }, |
| { |
| "epoch": 4.724696356275303, |
| "grad_norm": 0.20282680830726105, |
| "learning_rate": 3.6859626333943797e-07, |
| "loss": 0.154, |
| "step": 1167 |
| }, |
| { |
| "epoch": 4.728744939271255, |
| "grad_norm": 0.19190937513303374, |
| "learning_rate": 3.578671204615347e-07, |
| "loss": 0.1423, |
| "step": 1168 |
| }, |
| { |
| "epoch": 4.732793522267206, |
| "grad_norm": 0.1944910547594479, |
| "learning_rate": 3.472950355547267e-07, |
| "loss": 0.1296, |
| "step": 1169 |
| }, |
| { |
| "epoch": 4.7368421052631575, |
| "grad_norm": 0.19967582486144952, |
| "learning_rate": 3.3688009315312244e-07, |
| "loss": 0.1491, |
| "step": 1170 |
| }, |
| { |
| "epoch": 4.7408906882591095, |
| "grad_norm": 0.20026535636145737, |
| "learning_rate": 3.266223765343224e-07, |
| "loss": 0.134, |
| "step": 1171 |
| }, |
| { |
| "epoch": 4.744939271255061, |
| "grad_norm": 0.19026273229793272, |
| "learning_rate": 3.16521967718757e-07, |
| "loss": 0.1751, |
| "step": 1172 |
| }, |
| { |
| "epoch": 4.748987854251012, |
| "grad_norm": 0.20083612792537064, |
| "learning_rate": 3.0657894746902726e-07, |
| "loss": 0.1484, |
| "step": 1173 |
| }, |
| { |
| "epoch": 4.753036437246964, |
| "grad_norm": 0.19454273962980953, |
| "learning_rate": 2.9679339528925654e-07, |
| "loss": 0.1531, |
| "step": 1174 |
| }, |
| { |
| "epoch": 4.757085020242915, |
| "grad_norm": 0.1962169033978974, |
| "learning_rate": 2.871653894244597e-07, |
| "loss": 0.1438, |
| "step": 1175 |
| }, |
| { |
| "epoch": 4.761133603238866, |
| "grad_norm": 0.1971064809514222, |
| "learning_rate": 2.7769500685991266e-07, |
| "loss": 0.1535, |
| "step": 1176 |
| }, |
| { |
| "epoch": 4.765182186234818, |
| "grad_norm": 0.18875027743797437, |
| "learning_rate": 2.683823233205485e-07, |
| "loss": 0.1481, |
| "step": 1177 |
| }, |
| { |
| "epoch": 4.769230769230769, |
| "grad_norm": 0.2042677096065354, |
| "learning_rate": 2.592274132703287e-07, |
| "loss": 0.1486, |
| "step": 1178 |
| }, |
| { |
| "epoch": 4.77327935222672, |
| "grad_norm": 0.19748799939306919, |
| "learning_rate": 2.502303499116754e-07, |
| "loss": 0.1569, |
| "step": 1179 |
| }, |
| { |
| "epoch": 4.777327935222672, |
| "grad_norm": 0.20881000746941802, |
| "learning_rate": 2.413912051848666e-07, |
| "loss": 0.1408, |
| "step": 1180 |
| }, |
| { |
| "epoch": 4.781376518218623, |
| "grad_norm": 0.1956522690057225, |
| "learning_rate": 2.3271004976746613e-07, |
| "loss": 0.1789, |
| "step": 1181 |
| }, |
| { |
| "epoch": 4.7854251012145745, |
| "grad_norm": 0.21210606767310275, |
| "learning_rate": 2.2418695307376393e-07, |
| "loss": 0.1588, |
| "step": 1182 |
| }, |
| { |
| "epoch": 4.7894736842105265, |
| "grad_norm": 0.19719751036959318, |
| "learning_rate": 2.1582198325420968e-07, |
| "loss": 0.137, |
| "step": 1183 |
| }, |
| { |
| "epoch": 4.793522267206478, |
| "grad_norm": 0.201714023273691, |
| "learning_rate": 2.0761520719488447e-07, |
| "loss": 0.1432, |
| "step": 1184 |
| }, |
| { |
| "epoch": 4.797570850202429, |
| "grad_norm": 0.19151852395539135, |
| "learning_rate": 1.9956669051694798e-07, |
| "loss": 0.1401, |
| "step": 1185 |
| }, |
| { |
| "epoch": 4.801619433198381, |
| "grad_norm": 0.19550071756784604, |
| "learning_rate": 1.916764975761254e-07, |
| "loss": 0.1527, |
| "step": 1186 |
| }, |
| { |
| "epoch": 4.805668016194332, |
| "grad_norm": 0.19176489407325992, |
| "learning_rate": 1.839446914621923e-07, |
| "loss": 0.148, |
| "step": 1187 |
| }, |
| { |
| "epoch": 4.809716599190283, |
| "grad_norm": 0.1951199674391208, |
| "learning_rate": 1.7637133399845962e-07, |
| "loss": 0.1447, |
| "step": 1188 |
| }, |
| { |
| "epoch": 4.813765182186235, |
| "grad_norm": 0.20283258880163468, |
| "learning_rate": 1.6895648574129397e-07, |
| "loss": 0.143, |
| "step": 1189 |
| }, |
| { |
| "epoch": 4.817813765182186, |
| "grad_norm": 0.190688899654572, |
| "learning_rate": 1.6170020597962465e-07, |
| "loss": 0.142, |
| "step": 1190 |
| }, |
| { |
| "epoch": 4.821862348178137, |
| "grad_norm": 0.19919917522312963, |
| "learning_rate": 1.5460255273447078e-07, |
| "loss": 0.1586, |
| "step": 1191 |
| }, |
| { |
| "epoch": 4.825910931174089, |
| "grad_norm": 0.18454409371353872, |
| "learning_rate": 1.4766358275847715e-07, |
| "loss": 0.1518, |
| "step": 1192 |
| }, |
| { |
| "epoch": 4.82995951417004, |
| "grad_norm": 0.2001611375663235, |
| "learning_rate": 1.4088335153546573e-07, |
| "loss": 0.1475, |
| "step": 1193 |
| }, |
| { |
| "epoch": 4.834008097165992, |
| "grad_norm": 0.19575053988684005, |
| "learning_rate": 1.3426191327998496e-07, |
| "loss": 0.1497, |
| "step": 1194 |
| }, |
| { |
| "epoch": 4.838056680161944, |
| "grad_norm": 0.1995518117128797, |
| "learning_rate": 1.277993209368744e-07, |
| "loss": 0.1633, |
| "step": 1195 |
| }, |
| { |
| "epoch": 4.842105263157895, |
| "grad_norm": 0.19852087805732677, |
| "learning_rate": 1.214956261808542e-07, |
| "loss": 0.1485, |
| "step": 1196 |
| }, |
| { |
| "epoch": 4.846153846153846, |
| "grad_norm": 0.1931581637374895, |
| "learning_rate": 1.1535087941609846e-07, |
| "loss": 0.1524, |
| "step": 1197 |
| }, |
| { |
| "epoch": 4.850202429149798, |
| "grad_norm": 0.19996614136737087, |
| "learning_rate": 1.0936512977583802e-07, |
| "loss": 0.1426, |
| "step": 1198 |
| }, |
| { |
| "epoch": 4.854251012145749, |
| "grad_norm": 0.19343045375954365, |
| "learning_rate": 1.0353842512196732e-07, |
| "loss": 0.1446, |
| "step": 1199 |
| }, |
| { |
| "epoch": 4.8582995951417, |
| "grad_norm": 0.1943006709679248, |
| "learning_rate": 9.787081204466253e-08, |
| "loss": 0.1414, |
| "step": 1200 |
| }, |
| { |
| "epoch": 4.862348178137652, |
| "grad_norm": 0.19871194333471456, |
| "learning_rate": 9.236233586200627e-08, |
| "loss": 0.1553, |
| "step": 1201 |
| }, |
| { |
| "epoch": 4.866396761133603, |
| "grad_norm": 0.19410160893209802, |
| "learning_rate": 8.70130406196279e-08, |
| "loss": 0.1463, |
| "step": 1202 |
| }, |
| { |
| "epoch": 4.870445344129554, |
| "grad_norm": 0.20013023320105167, |
| "learning_rate": 8.182296909035047e-08, |
| "loss": 0.1403, |
| "step": 1203 |
| }, |
| { |
| "epoch": 4.874493927125506, |
| "grad_norm": 0.2010133546456109, |
| "learning_rate": 7.679216277384439e-08, |
| "loss": 0.1359, |
| "step": 1204 |
| }, |
| { |
| "epoch": 4.8785425101214575, |
| "grad_norm": 0.1920116157626257, |
| "learning_rate": 7.19206618963053e-08, |
| "loss": 0.1594, |
| "step": 1205 |
| }, |
| { |
| "epoch": 4.882591093117409, |
| "grad_norm": 0.19757546269286255, |
| "learning_rate": 6.720850541012347e-08, |
| "loss": 0.1507, |
| "step": 1206 |
| }, |
| { |
| "epoch": 4.886639676113361, |
| "grad_norm": 0.18723525603404925, |
| "learning_rate": 6.265573099357714e-08, |
| "loss": 0.1542, |
| "step": 1207 |
| }, |
| { |
| "epoch": 4.890688259109312, |
| "grad_norm": 0.2044383217382427, |
| "learning_rate": 5.826237505053067e-08, |
| "loss": 0.1715, |
| "step": 1208 |
| }, |
| { |
| "epoch": 4.894736842105263, |
| "grad_norm": 0.20171981473044065, |
| "learning_rate": 5.4028472710137e-08, |
| "loss": 0.1338, |
| "step": 1209 |
| }, |
| { |
| "epoch": 4.898785425101215, |
| "grad_norm": 0.19520185543865815, |
| "learning_rate": 4.9954057826571146e-08, |
| "loss": 0.1298, |
| "step": 1210 |
| }, |
| { |
| "epoch": 4.902834008097166, |
| "grad_norm": 0.19691498342570207, |
| "learning_rate": 4.6039162978743776e-08, |
| "loss": 0.1372, |
| "step": 1211 |
| }, |
| { |
| "epoch": 4.906882591093117, |
| "grad_norm": 0.2026645513108959, |
| "learning_rate": 4.228381947004812e-08, |
| "loss": 0.1209, |
| "step": 1212 |
| }, |
| { |
| "epoch": 4.910931174089069, |
| "grad_norm": 0.20087309867848033, |
| "learning_rate": 3.868805732811343e-08, |
| "loss": 0.1484, |
| "step": 1213 |
| }, |
| { |
| "epoch": 4.91497975708502, |
| "grad_norm": 0.19186324168896, |
| "learning_rate": 3.525190530455636e-08, |
| "loss": 0.1587, |
| "step": 1214 |
| }, |
| { |
| "epoch": 4.919028340080971, |
| "grad_norm": 0.1940599903855179, |
| "learning_rate": 3.197539087475887e-08, |
| "loss": 0.1594, |
| "step": 1215 |
| }, |
| { |
| "epoch": 4.923076923076923, |
| "grad_norm": 0.20227501665189748, |
| "learning_rate": 2.8858540237641786e-08, |
| "loss": 0.1506, |
| "step": 1216 |
| }, |
| { |
| "epoch": 4.9271255060728745, |
| "grad_norm": 0.20504201159710725, |
| "learning_rate": 2.5901378315458247e-08, |
| "loss": 0.1417, |
| "step": 1217 |
| }, |
| { |
| "epoch": 4.931174089068826, |
| "grad_norm": 0.19140405910444758, |
| "learning_rate": 2.3103928753598347e-08, |
| "loss": 0.1411, |
| "step": 1218 |
| }, |
| { |
| "epoch": 4.935222672064778, |
| "grad_norm": 0.19932304039162782, |
| "learning_rate": 2.0466213920393718e-08, |
| "loss": 0.1707, |
| "step": 1219 |
| }, |
| { |
| "epoch": 4.939271255060729, |
| "grad_norm": 0.19318147974442085, |
| "learning_rate": 1.7988254906942115e-08, |
| "loss": 0.1495, |
| "step": 1220 |
| }, |
| { |
| "epoch": 4.94331983805668, |
| "grad_norm": 0.19972942188289786, |
| "learning_rate": 1.5670071526936447e-08, |
| "loss": 0.1511, |
| "step": 1221 |
| }, |
| { |
| "epoch": 4.947368421052632, |
| "grad_norm": 0.20901602178266926, |
| "learning_rate": 1.3511682316509344e-08, |
| "loss": 0.137, |
| "step": 1222 |
| }, |
| { |
| "epoch": 4.951417004048583, |
| "grad_norm": 0.19639215967678214, |
| "learning_rate": 1.1513104534086605e-08, |
| "loss": 0.1569, |
| "step": 1223 |
| }, |
| { |
| "epoch": 4.955465587044534, |
| "grad_norm": 0.19272738642635864, |
| "learning_rate": 9.67435416023843e-09, |
| "loss": 0.1483, |
| "step": 1224 |
| }, |
| { |
| "epoch": 4.959514170040486, |
| "grad_norm": 0.1942463406496907, |
| "learning_rate": 7.995445897566178e-09, |
| "loss": 0.1389, |
| "step": 1225 |
| }, |
| { |
| "epoch": 4.963562753036437, |
| "grad_norm": 0.19375145218297576, |
| "learning_rate": 6.476393170578021e-09, |
| "loss": 0.154, |
| "step": 1226 |
| }, |
| { |
| "epoch": 4.967611336032388, |
| "grad_norm": 0.19056210753964573, |
| "learning_rate": 5.11720812557348e-09, |
| "loss": 0.1505, |
| "step": 1227 |
| }, |
| { |
| "epoch": 4.97165991902834, |
| "grad_norm": 0.19031809349886128, |
| "learning_rate": 3.917901630563492e-09, |
| "loss": 0.1278, |
| "step": 1228 |
| }, |
| { |
| "epoch": 4.9757085020242915, |
| "grad_norm": 0.20358210561787285, |
| "learning_rate": 2.878483275170485e-09, |
| "loss": 0.1509, |
| "step": 1229 |
| }, |
| { |
| "epoch": 4.979757085020243, |
| "grad_norm": 0.19054451956440713, |
| "learning_rate": 1.998961370555108e-09, |
| "loss": 0.1539, |
| "step": 1230 |
| }, |
| { |
| "epoch": 4.983805668016195, |
| "grad_norm": 0.20104478232904796, |
| "learning_rate": 1.2793429493518361e-09, |
| "loss": 0.1654, |
| "step": 1231 |
| }, |
| { |
| "epoch": 4.987854251012146, |
| "grad_norm": 0.20826730829951132, |
| "learning_rate": 7.196337656112384e-10, |
| "loss": 0.128, |
| "step": 1232 |
| }, |
| { |
| "epoch": 4.991902834008097, |
| "grad_norm": 0.21208179688424078, |
| "learning_rate": 3.198382947511292e-10, |
| "loss": 0.1559, |
| "step": 1233 |
| }, |
| { |
| "epoch": 4.995951417004049, |
| "grad_norm": 0.1953038055999227, |
| "learning_rate": 7.995973352770137e-11, |
| "loss": 0.1485, |
| "step": 1234 |
| }, |
| { |
| "epoch": 5.0, |
| "grad_norm": 0.2003994992677451, |
| "learning_rate": 0.0, |
| "loss": 0.1478, |
| "step": 1235 |
| }, |
| { |
| "epoch": 5.0, |
| "step": 1235, |
| "total_flos": 2.3632554217680404e+18, |
| "train_loss": 0.27695590181027346, |
| "train_runtime": 25283.7385, |
| "train_samples_per_second": 6.249, |
| "train_steps_per_second": 0.049 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 1235, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 5, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 2.3632554217680404e+18, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
