| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 3.0, |
| "eval_steps": 500, |
| "global_step": 1356, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0022123893805309734, |
| "grad_norm": 2.9161036014556885, |
| "learning_rate": 7.352941176470589e-08, |
| "loss": 0.7653, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.004424778761061947, |
| "grad_norm": 2.9344425201416016, |
| "learning_rate": 1.4705882352941178e-07, |
| "loss": 0.7756, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.00663716814159292, |
| "grad_norm": 2.8798561096191406, |
| "learning_rate": 2.2058823529411768e-07, |
| "loss": 0.7784, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.008849557522123894, |
| "grad_norm": 2.898063898086548, |
| "learning_rate": 2.9411764705882356e-07, |
| "loss": 0.7798, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.011061946902654867, |
| "grad_norm": 2.915313959121704, |
| "learning_rate": 3.6764705882352943e-07, |
| "loss": 0.7844, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.01327433628318584, |
| "grad_norm": 2.7517311573028564, |
| "learning_rate": 4.4117647058823536e-07, |
| "loss": 0.7519, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.015486725663716814, |
| "grad_norm": 2.8670687675476074, |
| "learning_rate": 5.147058823529412e-07, |
| "loss": 0.7639, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.017699115044247787, |
| "grad_norm": 2.838557243347168, |
| "learning_rate": 5.882352941176471e-07, |
| "loss": 0.7794, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.01991150442477876, |
| "grad_norm": 2.6696882247924805, |
| "learning_rate": 6.61764705882353e-07, |
| "loss": 0.7539, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.022123893805309734, |
| "grad_norm": 2.657489061355591, |
| "learning_rate": 7.352941176470589e-07, |
| "loss": 0.7437, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.024336283185840708, |
| "grad_norm": 2.6917357444763184, |
| "learning_rate": 8.088235294117648e-07, |
| "loss": 0.763, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.02654867256637168, |
| "grad_norm": 2.2191965579986572, |
| "learning_rate": 8.823529411764707e-07, |
| "loss": 0.7424, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.028761061946902654, |
| "grad_norm": 2.2132158279418945, |
| "learning_rate": 9.558823529411764e-07, |
| "loss": 0.7238, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.030973451327433628, |
| "grad_norm": 2.1964025497436523, |
| "learning_rate": 1.0294117647058825e-06, |
| "loss": 0.7384, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.033185840707964605, |
| "grad_norm": 2.0652434825897217, |
| "learning_rate": 1.1029411764705884e-06, |
| "loss": 0.7366, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.035398230088495575, |
| "grad_norm": 1.6799640655517578, |
| "learning_rate": 1.1764705882352942e-06, |
| "loss": 0.7022, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.03761061946902655, |
| "grad_norm": 1.4400370121002197, |
| "learning_rate": 1.25e-06, |
| "loss": 0.6861, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.03982300884955752, |
| "grad_norm": 1.4010155200958252, |
| "learning_rate": 1.323529411764706e-06, |
| "loss": 0.6904, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.0420353982300885, |
| "grad_norm": 1.3582905530929565, |
| "learning_rate": 1.3970588235294119e-06, |
| "loss": 0.6876, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.04424778761061947, |
| "grad_norm": 1.3372609615325928, |
| "learning_rate": 1.4705882352941177e-06, |
| "loss": 0.6809, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.046460176991150445, |
| "grad_norm": 1.3210690021514893, |
| "learning_rate": 1.5441176470588238e-06, |
| "loss": 0.687, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.048672566371681415, |
| "grad_norm": 1.0419282913208008, |
| "learning_rate": 1.6176470588235297e-06, |
| "loss": 0.6743, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.05088495575221239, |
| "grad_norm": 0.9881656765937805, |
| "learning_rate": 1.6911764705882356e-06, |
| "loss": 0.6483, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.05309734513274336, |
| "grad_norm": 0.9939265251159668, |
| "learning_rate": 1.7647058823529414e-06, |
| "loss": 0.6527, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.05530973451327434, |
| "grad_norm": 0.9658608436584473, |
| "learning_rate": 1.8382352941176473e-06, |
| "loss": 0.6544, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.05752212389380531, |
| "grad_norm": 0.897885262966156, |
| "learning_rate": 1.9117647058823528e-06, |
| "loss": 0.6358, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.059734513274336286, |
| "grad_norm": 0.8456088900566101, |
| "learning_rate": 1.985294117647059e-06, |
| "loss": 0.6403, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.061946902654867256, |
| "grad_norm": 0.8407159447669983, |
| "learning_rate": 2.058823529411765e-06, |
| "loss": 0.6366, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.06415929203539823, |
| "grad_norm": 0.7394031286239624, |
| "learning_rate": 2.132352941176471e-06, |
| "loss": 0.6277, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.06637168141592921, |
| "grad_norm": 0.6359221339225769, |
| "learning_rate": 2.2058823529411767e-06, |
| "loss": 0.6088, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.06858407079646017, |
| "grad_norm": 0.6923512816429138, |
| "learning_rate": 2.2794117647058826e-06, |
| "loss": 0.6072, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.07079646017699115, |
| "grad_norm": 0.6872782707214355, |
| "learning_rate": 2.3529411764705885e-06, |
| "loss": 0.5939, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.07300884955752213, |
| "grad_norm": 0.6121441721916199, |
| "learning_rate": 2.4264705882352943e-06, |
| "loss": 0.5936, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.0752212389380531, |
| "grad_norm": 0.5648632645606995, |
| "learning_rate": 2.5e-06, |
| "loss": 0.5925, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.07743362831858407, |
| "grad_norm": 0.527875542640686, |
| "learning_rate": 2.5735294117647057e-06, |
| "loss": 0.5788, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.07964601769911504, |
| "grad_norm": 0.48248472809791565, |
| "learning_rate": 2.647058823529412e-06, |
| "loss": 0.584, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.08185840707964602, |
| "grad_norm": 0.5289769172668457, |
| "learning_rate": 2.720588235294118e-06, |
| "loss": 0.5811, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.084070796460177, |
| "grad_norm": 0.5214428901672363, |
| "learning_rate": 2.7941176470588237e-06, |
| "loss": 0.5592, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.08628318584070796, |
| "grad_norm": 0.4773428440093994, |
| "learning_rate": 2.8676470588235296e-06, |
| "loss": 0.558, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.08849557522123894, |
| "grad_norm": 0.4300067722797394, |
| "learning_rate": 2.9411764705882355e-06, |
| "loss": 0.5592, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.09070796460176991, |
| "grad_norm": 0.4010540246963501, |
| "learning_rate": 3.0147058823529413e-06, |
| "loss": 0.559, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.09292035398230089, |
| "grad_norm": 0.3986721336841583, |
| "learning_rate": 3.0882352941176476e-06, |
| "loss": 0.5653, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.09513274336283185, |
| "grad_norm": 0.37803158164024353, |
| "learning_rate": 3.161764705882353e-06, |
| "loss": 0.5421, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.09734513274336283, |
| "grad_norm": 0.3909413516521454, |
| "learning_rate": 3.2352941176470594e-06, |
| "loss": 0.5347, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.09955752212389381, |
| "grad_norm": 0.373319536447525, |
| "learning_rate": 3.308823529411765e-06, |
| "loss": 0.5262, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.10176991150442478, |
| "grad_norm": 0.38008248805999756, |
| "learning_rate": 3.382352941176471e-06, |
| "loss": 0.5285, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.10398230088495575, |
| "grad_norm": 0.37302571535110474, |
| "learning_rate": 3.4558823529411766e-06, |
| "loss": 0.5132, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.10619469026548672, |
| "grad_norm": 0.3702549636363983, |
| "learning_rate": 3.529411764705883e-06, |
| "loss": 0.5228, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.1084070796460177, |
| "grad_norm": 0.3243614733219147, |
| "learning_rate": 3.6029411764705883e-06, |
| "loss": 0.5126, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.11061946902654868, |
| "grad_norm": 0.2833242118358612, |
| "learning_rate": 3.6764705882352946e-06, |
| "loss": 0.5061, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.11283185840707964, |
| "grad_norm": 0.27726179361343384, |
| "learning_rate": 3.7500000000000005e-06, |
| "loss": 0.5054, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.11504424778761062, |
| "grad_norm": 0.2611929774284363, |
| "learning_rate": 3.8235294117647055e-06, |
| "loss": 0.5136, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.1172566371681416, |
| "grad_norm": 0.2536194324493408, |
| "learning_rate": 3.897058823529412e-06, |
| "loss": 0.5016, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.11946902654867257, |
| "grad_norm": 0.24586400389671326, |
| "learning_rate": 3.970588235294118e-06, |
| "loss": 0.5017, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.12168141592920353, |
| "grad_norm": 0.22390125691890717, |
| "learning_rate": 4.044117647058824e-06, |
| "loss": 0.5015, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.12389380530973451, |
| "grad_norm": 0.2428012639284134, |
| "learning_rate": 4.11764705882353e-06, |
| "loss": 0.5125, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.1261061946902655, |
| "grad_norm": 0.2092347890138626, |
| "learning_rate": 4.191176470588236e-06, |
| "loss": 0.4978, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.12831858407079647, |
| "grad_norm": 0.2112482786178589, |
| "learning_rate": 4.264705882352942e-06, |
| "loss": 0.5117, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.13053097345132744, |
| "grad_norm": 0.2061765491962433, |
| "learning_rate": 4.3382352941176475e-06, |
| "loss": 0.4917, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.13274336283185842, |
| "grad_norm": 0.20188196003437042, |
| "learning_rate": 4.411764705882353e-06, |
| "loss": 0.4958, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.13495575221238937, |
| "grad_norm": 0.21565476059913635, |
| "learning_rate": 4.485294117647059e-06, |
| "loss": 0.4839, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.13716814159292035, |
| "grad_norm": 0.22550469636917114, |
| "learning_rate": 4.558823529411765e-06, |
| "loss": 0.4919, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.13938053097345132, |
| "grad_norm": 0.20743007957935333, |
| "learning_rate": 4.632352941176471e-06, |
| "loss": 0.4937, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.1415929203539823, |
| "grad_norm": 0.2033708542585373, |
| "learning_rate": 4.705882352941177e-06, |
| "loss": 0.4858, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.14380530973451328, |
| "grad_norm": 0.21126723289489746, |
| "learning_rate": 4.779411764705883e-06, |
| "loss": 0.487, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.14601769911504425, |
| "grad_norm": 0.2027079313993454, |
| "learning_rate": 4.852941176470589e-06, |
| "loss": 0.4975, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.14823008849557523, |
| "grad_norm": 0.18106848001480103, |
| "learning_rate": 4.9264705882352945e-06, |
| "loss": 0.4894, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.1504424778761062, |
| "grad_norm": 0.17369899153709412, |
| "learning_rate": 5e-06, |
| "loss": 0.4825, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.15265486725663716, |
| "grad_norm": 0.1895466446876526, |
| "learning_rate": 5.073529411764706e-06, |
| "loss": 0.4792, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.15486725663716813, |
| "grad_norm": 0.17305336892604828, |
| "learning_rate": 5.147058823529411e-06, |
| "loss": 0.4715, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.1570796460176991, |
| "grad_norm": 0.1629684716463089, |
| "learning_rate": 5.220588235294118e-06, |
| "loss": 0.4696, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.1592920353982301, |
| "grad_norm": 0.1580497771501541, |
| "learning_rate": 5.294117647058824e-06, |
| "loss": 0.4647, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.16150442477876106, |
| "grad_norm": 0.16177481412887573, |
| "learning_rate": 5.36764705882353e-06, |
| "loss": 0.4667, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.16371681415929204, |
| "grad_norm": 0.14660590887069702, |
| "learning_rate": 5.441176470588236e-06, |
| "loss": 0.4702, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.16592920353982302, |
| "grad_norm": 0.16958165168762207, |
| "learning_rate": 5.514705882352942e-06, |
| "loss": 0.4664, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.168141592920354, |
| "grad_norm": 0.18446360528469086, |
| "learning_rate": 5.588235294117647e-06, |
| "loss": 0.4634, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.17035398230088494, |
| "grad_norm": 0.18088003993034363, |
| "learning_rate": 5.661764705882353e-06, |
| "loss": 0.4782, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.17256637168141592, |
| "grad_norm": 0.16989891231060028, |
| "learning_rate": 5.735294117647059e-06, |
| "loss": 0.4751, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.1747787610619469, |
| "grad_norm": 0.15071384608745575, |
| "learning_rate": 5.808823529411766e-06, |
| "loss": 0.4679, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.17699115044247787, |
| "grad_norm": 0.16606192290782928, |
| "learning_rate": 5.882352941176471e-06, |
| "loss": 0.4613, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.17920353982300885, |
| "grad_norm": 0.16193555295467377, |
| "learning_rate": 5.955882352941177e-06, |
| "loss": 0.4656, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.18141592920353983, |
| "grad_norm": 0.14852063357830048, |
| "learning_rate": 6.029411764705883e-06, |
| "loss": 0.4508, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.1836283185840708, |
| "grad_norm": 0.16259585320949554, |
| "learning_rate": 6.102941176470589e-06, |
| "loss": 0.453, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.18584070796460178, |
| "grad_norm": 0.17288607358932495, |
| "learning_rate": 6.176470588235295e-06, |
| "loss": 0.4597, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.18805309734513273, |
| "grad_norm": 0.15836317837238312, |
| "learning_rate": 6.25e-06, |
| "loss": 0.4548, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.1902654867256637, |
| "grad_norm": 0.17005616426467896, |
| "learning_rate": 6.323529411764706e-06, |
| "loss": 0.4608, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.19247787610619468, |
| "grad_norm": 0.148678258061409, |
| "learning_rate": 6.397058823529412e-06, |
| "loss": 0.4599, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.19469026548672566, |
| "grad_norm": 0.16987866163253784, |
| "learning_rate": 6.470588235294119e-06, |
| "loss": 0.4512, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.19690265486725664, |
| "grad_norm": 0.15269418060779572, |
| "learning_rate": 6.544117647058824e-06, |
| "loss": 0.4579, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.19911504424778761, |
| "grad_norm": 0.1564721018075943, |
| "learning_rate": 6.61764705882353e-06, |
| "loss": 0.4577, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.2013274336283186, |
| "grad_norm": 0.15829169750213623, |
| "learning_rate": 6.6911764705882356e-06, |
| "loss": 0.4628, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.20353982300884957, |
| "grad_norm": 0.16252678632736206, |
| "learning_rate": 6.764705882352942e-06, |
| "loss": 0.4508, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.20575221238938052, |
| "grad_norm": 0.16816888749599457, |
| "learning_rate": 6.838235294117648e-06, |
| "loss": 0.4569, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.2079646017699115, |
| "grad_norm": 0.16243158280849457, |
| "learning_rate": 6.911764705882353e-06, |
| "loss": 0.4492, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.21017699115044247, |
| "grad_norm": 0.1554814875125885, |
| "learning_rate": 6.985294117647059e-06, |
| "loss": 0.4501, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.21238938053097345, |
| "grad_norm": 0.15875332057476044, |
| "learning_rate": 7.058823529411766e-06, |
| "loss": 0.4581, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.21460176991150443, |
| "grad_norm": 0.16014939546585083, |
| "learning_rate": 7.132352941176472e-06, |
| "loss": 0.448, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.2168141592920354, |
| "grad_norm": 0.18041987717151642, |
| "learning_rate": 7.205882352941177e-06, |
| "loss": 0.4553, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.21902654867256638, |
| "grad_norm": 0.16102494299411774, |
| "learning_rate": 7.2794117647058826e-06, |
| "loss": 0.4553, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.22123893805309736, |
| "grad_norm": 0.1710263043642044, |
| "learning_rate": 7.352941176470589e-06, |
| "loss": 0.4426, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.2234513274336283, |
| "grad_norm": 0.17998401820659637, |
| "learning_rate": 7.426470588235295e-06, |
| "loss": 0.439, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.22566371681415928, |
| "grad_norm": 0.17915406823158264, |
| "learning_rate": 7.500000000000001e-06, |
| "loss": 0.4467, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.22787610619469026, |
| "grad_norm": 0.16851037740707397, |
| "learning_rate": 7.573529411764706e-06, |
| "loss": 0.4363, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.23008849557522124, |
| "grad_norm": 0.17610934376716614, |
| "learning_rate": 7.647058823529411e-06, |
| "loss": 0.4493, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.2323008849557522, |
| "grad_norm": 0.18994346261024475, |
| "learning_rate": 7.720588235294119e-06, |
| "loss": 0.4485, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.2345132743362832, |
| "grad_norm": 0.1732485294342041, |
| "learning_rate": 7.794117647058825e-06, |
| "loss": 0.4396, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.23672566371681417, |
| "grad_norm": 0.16453640162944794, |
| "learning_rate": 7.86764705882353e-06, |
| "loss": 0.4442, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.23893805309734514, |
| "grad_norm": 0.1798321008682251, |
| "learning_rate": 7.941176470588236e-06, |
| "loss": 0.44, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.2411504424778761, |
| "grad_norm": 0.17025865614414215, |
| "learning_rate": 8.014705882352942e-06, |
| "loss": 0.436, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.24336283185840707, |
| "grad_norm": 0.17872163653373718, |
| "learning_rate": 8.088235294117648e-06, |
| "loss": 0.4393, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.24557522123893805, |
| "grad_norm": 0.17654740810394287, |
| "learning_rate": 8.161764705882354e-06, |
| "loss": 0.4453, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.24778761061946902, |
| "grad_norm": 0.1653042584657669, |
| "learning_rate": 8.23529411764706e-06, |
| "loss": 0.429, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.25, |
| "grad_norm": 0.16623114049434662, |
| "learning_rate": 8.308823529411766e-06, |
| "loss": 0.4378, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.252212389380531, |
| "grad_norm": 0.17289873957633972, |
| "learning_rate": 8.382352941176472e-06, |
| "loss": 0.4386, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.25442477876106195, |
| "grad_norm": 0.17378394305706024, |
| "learning_rate": 8.455882352941177e-06, |
| "loss": 0.4337, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.25663716814159293, |
| "grad_norm": 0.15947799384593964, |
| "learning_rate": 8.529411764705883e-06, |
| "loss": 0.4426, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.2588495575221239, |
| "grad_norm": 0.20766223967075348, |
| "learning_rate": 8.60294117647059e-06, |
| "loss": 0.4345, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.2610619469026549, |
| "grad_norm": 0.19195608794689178, |
| "learning_rate": 8.676470588235295e-06, |
| "loss": 0.4288, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.26327433628318586, |
| "grad_norm": 0.1608441323041916, |
| "learning_rate": 8.750000000000001e-06, |
| "loss": 0.4431, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.26548672566371684, |
| "grad_norm": 0.21266116201877594, |
| "learning_rate": 8.823529411764707e-06, |
| "loss": 0.4312, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.2676991150442478, |
| "grad_norm": 0.21518948674201965, |
| "learning_rate": 8.897058823529413e-06, |
| "loss": 0.4296, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.26991150442477874, |
| "grad_norm": 0.1727646440267563, |
| "learning_rate": 8.970588235294119e-06, |
| "loss": 0.432, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.2721238938053097, |
| "grad_norm": 0.25698140263557434, |
| "learning_rate": 9.044117647058824e-06, |
| "loss": 0.4245, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.2743362831858407, |
| "grad_norm": 0.20740661025047302, |
| "learning_rate": 9.11764705882353e-06, |
| "loss": 0.4407, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.27654867256637167, |
| "grad_norm": 0.19939358532428741, |
| "learning_rate": 9.191176470588236e-06, |
| "loss": 0.4478, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.27876106194690264, |
| "grad_norm": 0.24062572419643402, |
| "learning_rate": 9.264705882352942e-06, |
| "loss": 0.4259, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.2809734513274336, |
| "grad_norm": 0.2283308357000351, |
| "learning_rate": 9.338235294117648e-06, |
| "loss": 0.4377, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.2831858407079646, |
| "grad_norm": 0.20251670479774475, |
| "learning_rate": 9.411764705882354e-06, |
| "loss": 0.4399, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.2853982300884956, |
| "grad_norm": 0.23135575652122498, |
| "learning_rate": 9.48529411764706e-06, |
| "loss": 0.4374, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.28761061946902655, |
| "grad_norm": 0.1677524596452713, |
| "learning_rate": 9.558823529411766e-06, |
| "loss": 0.4382, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.28982300884955753, |
| "grad_norm": 0.19608592987060547, |
| "learning_rate": 9.632352941176471e-06, |
| "loss": 0.4319, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.2920353982300885, |
| "grad_norm": 0.1991558074951172, |
| "learning_rate": 9.705882352941177e-06, |
| "loss": 0.4294, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.2942477876106195, |
| "grad_norm": 0.17177461087703705, |
| "learning_rate": 9.779411764705883e-06, |
| "loss": 0.4349, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.29646017699115046, |
| "grad_norm": 0.1729421466588974, |
| "learning_rate": 9.852941176470589e-06, |
| "loss": 0.4283, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.29867256637168144, |
| "grad_norm": 0.2021639496088028, |
| "learning_rate": 9.926470588235295e-06, |
| "loss": 0.4377, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.3008849557522124, |
| "grad_norm": 0.18062114715576172, |
| "learning_rate": 1e-05, |
| "loss": 0.4427, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.3030973451327434, |
| "grad_norm": 0.16362692415714264, |
| "learning_rate": 9.999983422468849e-06, |
| "loss": 0.4395, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.3053097345132743, |
| "grad_norm": 0.20908989012241364, |
| "learning_rate": 9.999933689985315e-06, |
| "loss": 0.4363, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.3075221238938053, |
| "grad_norm": 0.16169682145118713, |
| "learning_rate": 9.999850802879177e-06, |
| "loss": 0.4303, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.30973451327433627, |
| "grad_norm": 0.17690162360668182, |
| "learning_rate": 9.999734761700061e-06, |
| "loss": 0.4266, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.31194690265486724, |
| "grad_norm": 0.17501935362815857, |
| "learning_rate": 9.99958556721744e-06, |
| "loss": 0.4336, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.3141592920353982, |
| "grad_norm": 0.16887253522872925, |
| "learning_rate": 9.999403220420619e-06, |
| "loss": 0.4351, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.3163716814159292, |
| "grad_norm": 0.1647050976753235, |
| "learning_rate": 9.999187722518747e-06, |
| "loss": 0.426, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.3185840707964602, |
| "grad_norm": 0.18983641266822815, |
| "learning_rate": 9.998939074940788e-06, |
| "loss": 0.4219, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.32079646017699115, |
| "grad_norm": 0.19264842569828033, |
| "learning_rate": 9.99865727933553e-06, |
| "loss": 0.4237, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.3230088495575221, |
| "grad_norm": 0.18118713796138763, |
| "learning_rate": 9.998342337571566e-06, |
| "loss": 0.4174, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.3252212389380531, |
| "grad_norm": 0.22882233560085297, |
| "learning_rate": 9.997994251737276e-06, |
| "loss": 0.4277, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.3274336283185841, |
| "grad_norm": 0.17980162799358368, |
| "learning_rate": 9.99761302414082e-06, |
| "loss": 0.4257, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.32964601769911506, |
| "grad_norm": 0.2404302954673767, |
| "learning_rate": 9.997198657310126e-06, |
| "loss": 0.4305, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.33185840707964603, |
| "grad_norm": 0.19852545857429504, |
| "learning_rate": 9.996751153992861e-06, |
| "loss": 0.4212, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.334070796460177, |
| "grad_norm": 0.22487014532089233, |
| "learning_rate": 9.996270517156431e-06, |
| "loss": 0.4257, |
| "step": 151 |
| }, |
| { |
| "epoch": 0.336283185840708, |
| "grad_norm": 0.22307410836219788, |
| "learning_rate": 9.995756749987942e-06, |
| "loss": 0.4208, |
| "step": 152 |
| }, |
| { |
| "epoch": 0.33849557522123896, |
| "grad_norm": 0.2162560373544693, |
| "learning_rate": 9.995209855894191e-06, |
| "loss": 0.4206, |
| "step": 153 |
| }, |
| { |
| "epoch": 0.3407079646017699, |
| "grad_norm": 0.20444706082344055, |
| "learning_rate": 9.994629838501637e-06, |
| "loss": 0.4295, |
| "step": 154 |
| }, |
| { |
| "epoch": 0.34292035398230086, |
| "grad_norm": 0.1971275806427002, |
| "learning_rate": 9.994016701656384e-06, |
| "loss": 0.4299, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.34513274336283184, |
| "grad_norm": 0.19772182404994965, |
| "learning_rate": 9.993370449424153e-06, |
| "loss": 0.4201, |
| "step": 156 |
| }, |
| { |
| "epoch": 0.3473451327433628, |
| "grad_norm": 0.1802293360233307, |
| "learning_rate": 9.992691086090249e-06, |
| "loss": 0.4254, |
| "step": 157 |
| }, |
| { |
| "epoch": 0.3495575221238938, |
| "grad_norm": 0.1828855574131012, |
| "learning_rate": 9.991978616159535e-06, |
| "loss": 0.4279, |
| "step": 158 |
| }, |
| { |
| "epoch": 0.35176991150442477, |
| "grad_norm": 0.16842177510261536, |
| "learning_rate": 9.991233044356414e-06, |
| "loss": 0.4176, |
| "step": 159 |
| }, |
| { |
| "epoch": 0.35398230088495575, |
| "grad_norm": 0.1759248673915863, |
| "learning_rate": 9.990454375624778e-06, |
| "loss": 0.4266, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.3561946902654867, |
| "grad_norm": 0.18696172535419464, |
| "learning_rate": 9.98964261512799e-06, |
| "loss": 0.4195, |
| "step": 161 |
| }, |
| { |
| "epoch": 0.3584070796460177, |
| "grad_norm": 0.16631002724170685, |
| "learning_rate": 9.988797768248844e-06, |
| "loss": 0.4272, |
| "step": 162 |
| }, |
| { |
| "epoch": 0.3606194690265487, |
| "grad_norm": 0.16732241213321686, |
| "learning_rate": 9.987919840589529e-06, |
| "loss": 0.4195, |
| "step": 163 |
| }, |
| { |
| "epoch": 0.36283185840707965, |
| "grad_norm": 0.18925903737545013, |
| "learning_rate": 9.987008837971595e-06, |
| "loss": 0.4202, |
| "step": 164 |
| }, |
| { |
| "epoch": 0.36504424778761063, |
| "grad_norm": 0.17854173481464386, |
| "learning_rate": 9.986064766435915e-06, |
| "loss": 0.4247, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.3672566371681416, |
| "grad_norm": 0.20543624460697174, |
| "learning_rate": 9.985087632242634e-06, |
| "loss": 0.4163, |
| "step": 166 |
| }, |
| { |
| "epoch": 0.3694690265486726, |
| "grad_norm": 0.17206773161888123, |
| "learning_rate": 9.984077441871144e-06, |
| "loss": 0.4245, |
| "step": 167 |
| }, |
| { |
| "epoch": 0.37168141592920356, |
| "grad_norm": 0.19772297143936157, |
| "learning_rate": 9.98303420202003e-06, |
| "loss": 0.406, |
| "step": 168 |
| }, |
| { |
| "epoch": 0.37389380530973454, |
| "grad_norm": 0.21683964133262634, |
| "learning_rate": 9.981957919607026e-06, |
| "loss": 0.4302, |
| "step": 169 |
| }, |
| { |
| "epoch": 0.37610619469026546, |
| "grad_norm": 0.17397964000701904, |
| "learning_rate": 9.980848601768976e-06, |
| "loss": 0.4155, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.37831858407079644, |
| "grad_norm": 0.19775860011577606, |
| "learning_rate": 9.97970625586178e-06, |
| "loss": 0.4223, |
| "step": 171 |
| }, |
| { |
| "epoch": 0.3805309734513274, |
| "grad_norm": 0.18166320025920868, |
| "learning_rate": 9.978530889460351e-06, |
| "loss": 0.4205, |
| "step": 172 |
| }, |
| { |
| "epoch": 0.3827433628318584, |
| "grad_norm": 0.20108245313167572, |
| "learning_rate": 9.977322510358552e-06, |
| "loss": 0.4262, |
| "step": 173 |
| }, |
| { |
| "epoch": 0.38495575221238937, |
| "grad_norm": 0.17232555150985718, |
| "learning_rate": 9.976081126569164e-06, |
| "loss": 0.4257, |
| "step": 174 |
| }, |
| { |
| "epoch": 0.38716814159292035, |
| "grad_norm": 0.17979106307029724, |
| "learning_rate": 9.97480674632382e-06, |
| "loss": 0.4164, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.3893805309734513, |
| "grad_norm": 0.21929936110973358, |
| "learning_rate": 9.973499378072947e-06, |
| "loss": 0.4156, |
| "step": 176 |
| }, |
| { |
| "epoch": 0.3915929203539823, |
| "grad_norm": 0.19628576934337616, |
| "learning_rate": 9.972159030485722e-06, |
| "loss": 0.4176, |
| "step": 177 |
| }, |
| { |
| "epoch": 0.3938053097345133, |
| "grad_norm": 0.20714552700519562, |
| "learning_rate": 9.970785712450007e-06, |
| "loss": 0.429, |
| "step": 178 |
| }, |
| { |
| "epoch": 0.39601769911504425, |
| "grad_norm": 0.2554994821548462, |
| "learning_rate": 9.96937943307229e-06, |
| "loss": 0.4183, |
| "step": 179 |
| }, |
| { |
| "epoch": 0.39823008849557523, |
| "grad_norm": 0.21966539323329926, |
| "learning_rate": 9.967940201677628e-06, |
| "loss": 0.4148, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.4004424778761062, |
| "grad_norm": 0.19780422747135162, |
| "learning_rate": 9.966468027809582e-06, |
| "loss": 0.4086, |
| "step": 181 |
| }, |
| { |
| "epoch": 0.4026548672566372, |
| "grad_norm": 0.19103385508060455, |
| "learning_rate": 9.964962921230155e-06, |
| "loss": 0.4123, |
| "step": 182 |
| }, |
| { |
| "epoch": 0.40486725663716816, |
| "grad_norm": 0.20896346867084503, |
| "learning_rate": 9.963424891919728e-06, |
| "loss": 0.4138, |
| "step": 183 |
| }, |
| { |
| "epoch": 0.40707964601769914, |
| "grad_norm": 0.1816549003124237, |
| "learning_rate": 9.961853950076992e-06, |
| "loss": 0.4175, |
| "step": 184 |
| }, |
| { |
| "epoch": 0.4092920353982301, |
| "grad_norm": 0.1906682550907135, |
| "learning_rate": 9.960250106118883e-06, |
| "loss": 0.4146, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.41150442477876104, |
| "grad_norm": 0.19309958815574646, |
| "learning_rate": 9.958613370680507e-06, |
| "loss": 0.422, |
| "step": 186 |
| }, |
| { |
| "epoch": 0.413716814159292, |
| "grad_norm": 0.19457145035266876, |
| "learning_rate": 9.956943754615082e-06, |
| "loss": 0.4138, |
| "step": 187 |
| }, |
| { |
| "epoch": 0.415929203539823, |
| "grad_norm": 0.1757512390613556, |
| "learning_rate": 9.955241268993852e-06, |
| "loss": 0.414, |
| "step": 188 |
| }, |
| { |
| "epoch": 0.41814159292035397, |
| "grad_norm": 0.16454994678497314, |
| "learning_rate": 9.953505925106016e-06, |
| "loss": 0.4199, |
| "step": 189 |
| }, |
| { |
| "epoch": 0.42035398230088494, |
| "grad_norm": 0.1722830832004547, |
| "learning_rate": 9.951737734458665e-06, |
| "loss": 0.4219, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.4225663716814159, |
| "grad_norm": 0.1671864539384842, |
| "learning_rate": 9.949936708776692e-06, |
| "loss": 0.4226, |
| "step": 191 |
| }, |
| { |
| "epoch": 0.4247787610619469, |
| "grad_norm": 0.16959279775619507, |
| "learning_rate": 9.94810286000272e-06, |
| "loss": 0.4188, |
| "step": 192 |
| }, |
| { |
| "epoch": 0.4269911504424779, |
| "grad_norm": 0.18185065686702728, |
| "learning_rate": 9.946236200297025e-06, |
| "loss": 0.4225, |
| "step": 193 |
| }, |
| { |
| "epoch": 0.42920353982300885, |
| "grad_norm": 0.1568220853805542, |
| "learning_rate": 9.944336742037451e-06, |
| "loss": 0.4064, |
| "step": 194 |
| }, |
| { |
| "epoch": 0.4314159292035398, |
| "grad_norm": 0.19281834363937378, |
| "learning_rate": 9.942404497819324e-06, |
| "loss": 0.4135, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.4336283185840708, |
| "grad_norm": 0.16964174807071686, |
| "learning_rate": 9.940439480455386e-06, |
| "loss": 0.4196, |
| "step": 196 |
| }, |
| { |
| "epoch": 0.4358407079646018, |
| "grad_norm": 0.17745113372802734, |
| "learning_rate": 9.938441702975689e-06, |
| "loss": 0.4167, |
| "step": 197 |
| }, |
| { |
| "epoch": 0.43805309734513276, |
| "grad_norm": 0.1645510345697403, |
| "learning_rate": 9.936411178627521e-06, |
| "loss": 0.4093, |
| "step": 198 |
| }, |
| { |
| "epoch": 0.44026548672566373, |
| "grad_norm": 0.16442488133907318, |
| "learning_rate": 9.934347920875314e-06, |
| "loss": 0.4248, |
| "step": 199 |
| }, |
| { |
| "epoch": 0.4424778761061947, |
| "grad_norm": 0.17704367637634277, |
| "learning_rate": 9.932251943400554e-06, |
| "loss": 0.4075, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.4446902654867257, |
| "grad_norm": 0.1849382370710373, |
| "learning_rate": 9.930123260101697e-06, |
| "loss": 0.4182, |
| "step": 201 |
| }, |
| { |
| "epoch": 0.4469026548672566, |
| "grad_norm": 0.1769019365310669, |
| "learning_rate": 9.927961885094065e-06, |
| "loss": 0.4044, |
| "step": 202 |
| }, |
| { |
| "epoch": 0.4491150442477876, |
| "grad_norm": 0.17679619789123535, |
| "learning_rate": 9.925767832709765e-06, |
| "loss": 0.4161, |
| "step": 203 |
| }, |
| { |
| "epoch": 0.45132743362831856, |
| "grad_norm": 0.19563840329647064, |
| "learning_rate": 9.923541117497586e-06, |
| "loss": 0.4014, |
| "step": 204 |
| }, |
| { |
| "epoch": 0.45353982300884954, |
| "grad_norm": 0.20635181665420532, |
| "learning_rate": 9.921281754222903e-06, |
| "loss": 0.434, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.4557522123893805, |
| "grad_norm": 0.17097140848636627, |
| "learning_rate": 9.918989757867584e-06, |
| "loss": 0.4147, |
| "step": 206 |
| }, |
| { |
| "epoch": 0.4579646017699115, |
| "grad_norm": 0.17044438421726227, |
| "learning_rate": 9.916665143629881e-06, |
| "loss": 0.4138, |
| "step": 207 |
| }, |
| { |
| "epoch": 0.46017699115044247, |
| "grad_norm": 0.18953591585159302, |
| "learning_rate": 9.914307926924344e-06, |
| "loss": 0.4185, |
| "step": 208 |
| }, |
| { |
| "epoch": 0.46238938053097345, |
| "grad_norm": 0.17236648499965668, |
| "learning_rate": 9.911918123381706e-06, |
| "loss": 0.4079, |
| "step": 209 |
| }, |
| { |
| "epoch": 0.4646017699115044, |
| "grad_norm": 0.15649288892745972, |
| "learning_rate": 9.909495748848783e-06, |
| "loss": 0.4114, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.4668141592920354, |
| "grad_norm": 0.18650922179222107, |
| "learning_rate": 9.907040819388372e-06, |
| "loss": 0.4231, |
| "step": 211 |
| }, |
| { |
| "epoch": 0.4690265486725664, |
| "grad_norm": 0.16785378754138947, |
| "learning_rate": 9.904553351279139e-06, |
| "loss": 0.409, |
| "step": 212 |
| }, |
| { |
| "epoch": 0.47123893805309736, |
| "grad_norm": 0.17052291333675385, |
| "learning_rate": 9.902033361015515e-06, |
| "loss": 0.415, |
| "step": 213 |
| }, |
| { |
| "epoch": 0.47345132743362833, |
| "grad_norm": 0.1584978699684143, |
| "learning_rate": 9.899480865307591e-06, |
| "loss": 0.4014, |
| "step": 214 |
| }, |
| { |
| "epoch": 0.4756637168141593, |
| "grad_norm": 0.16302478313446045, |
| "learning_rate": 9.896895881080995e-06, |
| "loss": 0.4103, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.4778761061946903, |
| "grad_norm": 0.16019950807094574, |
| "learning_rate": 9.89427842547679e-06, |
| "loss": 0.4149, |
| "step": 216 |
| }, |
| { |
| "epoch": 0.48008849557522126, |
| "grad_norm": 0.17266441881656647, |
| "learning_rate": 9.891628515851358e-06, |
| "loss": 0.4079, |
| "step": 217 |
| }, |
| { |
| "epoch": 0.4823008849557522, |
| "grad_norm": 0.17257748544216156, |
| "learning_rate": 9.88894616977628e-06, |
| "loss": 0.4055, |
| "step": 218 |
| }, |
| { |
| "epoch": 0.48451327433628316, |
| "grad_norm": 0.17442859709262848, |
| "learning_rate": 9.88623140503823e-06, |
| "loss": 0.4194, |
| "step": 219 |
| }, |
| { |
| "epoch": 0.48672566371681414, |
| "grad_norm": 0.1760149896144867, |
| "learning_rate": 9.883484239638842e-06, |
| "loss": 0.4103, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.4889380530973451, |
| "grad_norm": 0.16283486783504486, |
| "learning_rate": 9.880704691794608e-06, |
| "loss": 0.4081, |
| "step": 221 |
| }, |
| { |
| "epoch": 0.4911504424778761, |
| "grad_norm": 0.17161579430103302, |
| "learning_rate": 9.877892779936744e-06, |
| "loss": 0.4105, |
| "step": 222 |
| }, |
| { |
| "epoch": 0.49336283185840707, |
| "grad_norm": 0.17453111708164215, |
| "learning_rate": 9.87504852271107e-06, |
| "loss": 0.4186, |
| "step": 223 |
| }, |
| { |
| "epoch": 0.49557522123893805, |
| "grad_norm": 0.17772270739078522, |
| "learning_rate": 9.872171938977895e-06, |
| "loss": 0.4196, |
| "step": 224 |
| }, |
| { |
| "epoch": 0.497787610619469, |
| "grad_norm": 0.1604829877614975, |
| "learning_rate": 9.869263047811877e-06, |
| "loss": 0.4138, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.5, |
| "grad_norm": 0.1991002857685089, |
| "learning_rate": 9.866321868501914e-06, |
| "loss": 0.4094, |
| "step": 226 |
| }, |
| { |
| "epoch": 0.5022123893805309, |
| "grad_norm": 0.1912929266691208, |
| "learning_rate": 9.863348420550998e-06, |
| "loss": 0.4073, |
| "step": 227 |
| }, |
| { |
| "epoch": 0.504424778761062, |
| "grad_norm": 0.1809907853603363, |
| "learning_rate": 9.860342723676105e-06, |
| "loss": 0.417, |
| "step": 228 |
| }, |
| { |
| "epoch": 0.5066371681415929, |
| "grad_norm": 0.20652230083942413, |
| "learning_rate": 9.857304797808043e-06, |
| "loss": 0.4094, |
| "step": 229 |
| }, |
| { |
| "epoch": 0.5088495575221239, |
| "grad_norm": 0.16395345330238342, |
| "learning_rate": 9.85423466309134e-06, |
| "loss": 0.4024, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.5110619469026548, |
| "grad_norm": 0.20369473099708557, |
| "learning_rate": 9.851132339884097e-06, |
| "loss": 0.4139, |
| "step": 231 |
| }, |
| { |
| "epoch": 0.5132743362831859, |
| "grad_norm": 0.169134721159935, |
| "learning_rate": 9.847997848757855e-06, |
| "loss": 0.4113, |
| "step": 232 |
| }, |
| { |
| "epoch": 0.5154867256637168, |
| "grad_norm": 0.19118903577327728, |
| "learning_rate": 9.844831210497468e-06, |
| "loss": 0.4089, |
| "step": 233 |
| }, |
| { |
| "epoch": 0.5176991150442478, |
| "grad_norm": 0.19465133547782898, |
| "learning_rate": 9.84163244610095e-06, |
| "loss": 0.4092, |
| "step": 234 |
| }, |
| { |
| "epoch": 0.5199115044247787, |
| "grad_norm": 0.18870508670806885, |
| "learning_rate": 9.83840157677935e-06, |
| "loss": 0.4108, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.5221238938053098, |
| "grad_norm": 0.18022289872169495, |
| "learning_rate": 9.835138623956603e-06, |
| "loss": 0.4244, |
| "step": 236 |
| }, |
| { |
| "epoch": 0.5243362831858407, |
| "grad_norm": 0.18440207839012146, |
| "learning_rate": 9.831843609269387e-06, |
| "loss": 0.4164, |
| "step": 237 |
| }, |
| { |
| "epoch": 0.5265486725663717, |
| "grad_norm": 0.17794957756996155, |
| "learning_rate": 9.828516554566988e-06, |
| "loss": 0.4069, |
| "step": 238 |
| }, |
| { |
| "epoch": 0.5287610619469026, |
| "grad_norm": 0.1812417358160019, |
| "learning_rate": 9.825157481911146e-06, |
| "loss": 0.4149, |
| "step": 239 |
| }, |
| { |
| "epoch": 0.5309734513274337, |
| "grad_norm": 0.18725480139255524, |
| "learning_rate": 9.821766413575915e-06, |
| "loss": 0.4303, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.5331858407079646, |
| "grad_norm": 0.17442503571510315, |
| "learning_rate": 9.818343372047509e-06, |
| "loss": 0.4036, |
| "step": 241 |
| }, |
| { |
| "epoch": 0.5353982300884956, |
| "grad_norm": 0.21173766255378723, |
| "learning_rate": 9.814888380024161e-06, |
| "loss": 0.4179, |
| "step": 242 |
| }, |
| { |
| "epoch": 0.5376106194690266, |
| "grad_norm": 0.18231187760829926, |
| "learning_rate": 9.811401460415966e-06, |
| "loss": 0.4173, |
| "step": 243 |
| }, |
| { |
| "epoch": 0.5398230088495575, |
| "grad_norm": 0.1679583489894867, |
| "learning_rate": 9.80788263634473e-06, |
| "loss": 0.4115, |
| "step": 244 |
| }, |
| { |
| "epoch": 0.5420353982300885, |
| "grad_norm": 0.16705110669136047, |
| "learning_rate": 9.804331931143822e-06, |
| "loss": 0.4025, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.5442477876106194, |
| "grad_norm": 0.17078952491283417, |
| "learning_rate": 9.80074936835801e-06, |
| "loss": 0.4208, |
| "step": 246 |
| }, |
| { |
| "epoch": 0.5464601769911505, |
| "grad_norm": 0.16935208439826965, |
| "learning_rate": 9.797134971743312e-06, |
| "loss": 0.4036, |
| "step": 247 |
| }, |
| { |
| "epoch": 0.5486725663716814, |
| "grad_norm": 0.1650349199771881, |
| "learning_rate": 9.793488765266838e-06, |
| "loss": 0.4172, |
| "step": 248 |
| }, |
| { |
| "epoch": 0.5508849557522124, |
| "grad_norm": 0.16523662209510803, |
| "learning_rate": 9.789810773106632e-06, |
| "loss": 0.4077, |
| "step": 249 |
| }, |
| { |
| "epoch": 0.5530973451327433, |
| "grad_norm": 0.20408877730369568, |
| "learning_rate": 9.786101019651499e-06, |
| "loss": 0.4043, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.5553097345132744, |
| "grad_norm": 0.1925540715456009, |
| "learning_rate": 9.782359529500867e-06, |
| "loss": 0.4166, |
| "step": 251 |
| }, |
| { |
| "epoch": 0.5575221238938053, |
| "grad_norm": 0.18965932726860046, |
| "learning_rate": 9.7785863274646e-06, |
| "loss": 0.4016, |
| "step": 252 |
| }, |
| { |
| "epoch": 0.5597345132743363, |
| "grad_norm": 0.18967291712760925, |
| "learning_rate": 9.774781438562846e-06, |
| "loss": 0.407, |
| "step": 253 |
| }, |
| { |
| "epoch": 0.5619469026548672, |
| "grad_norm": 0.17891788482666016, |
| "learning_rate": 9.770944888025874e-06, |
| "loss": 0.4078, |
| "step": 254 |
| }, |
| { |
| "epoch": 0.5641592920353983, |
| "grad_norm": 0.1903616040945053, |
| "learning_rate": 9.767076701293898e-06, |
| "loss": 0.4012, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.5663716814159292, |
| "grad_norm": 0.1644955724477768, |
| "learning_rate": 9.763176904016914e-06, |
| "loss": 0.4009, |
| "step": 256 |
| }, |
| { |
| "epoch": 0.5685840707964602, |
| "grad_norm": 0.1887083798646927, |
| "learning_rate": 9.759245522054523e-06, |
| "loss": 0.3965, |
| "step": 257 |
| }, |
| { |
| "epoch": 0.5707964601769911, |
| "grad_norm": 0.16408571600914001, |
| "learning_rate": 9.755282581475769e-06, |
| "loss": 0.4008, |
| "step": 258 |
| }, |
| { |
| "epoch": 0.5730088495575221, |
| "grad_norm": 0.2137821614742279, |
| "learning_rate": 9.751288108558961e-06, |
| "loss": 0.4029, |
| "step": 259 |
| }, |
| { |
| "epoch": 0.5752212389380531, |
| "grad_norm": 0.17689137160778046, |
| "learning_rate": 9.747262129791497e-06, |
| "loss": 0.4045, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.577433628318584, |
| "grad_norm": 0.1932871788740158, |
| "learning_rate": 9.743204671869694e-06, |
| "loss": 0.4082, |
| "step": 261 |
| }, |
| { |
| "epoch": 0.5796460176991151, |
| "grad_norm": 0.2326955646276474, |
| "learning_rate": 9.739115761698607e-06, |
| "loss": 0.4043, |
| "step": 262 |
| }, |
| { |
| "epoch": 0.581858407079646, |
| "grad_norm": 0.22756756842136383, |
| "learning_rate": 9.73499542639185e-06, |
| "loss": 0.4132, |
| "step": 263 |
| }, |
| { |
| "epoch": 0.584070796460177, |
| "grad_norm": 0.18425188958644867, |
| "learning_rate": 9.730843693271413e-06, |
| "loss": 0.4026, |
| "step": 264 |
| }, |
| { |
| "epoch": 0.5862831858407079, |
| "grad_norm": 0.24644114077091217, |
| "learning_rate": 9.726660589867495e-06, |
| "loss": 0.3966, |
| "step": 265 |
| }, |
| { |
| "epoch": 0.588495575221239, |
| "grad_norm": 0.20994241535663605, |
| "learning_rate": 9.722446143918307e-06, |
| "loss": 0.4068, |
| "step": 266 |
| }, |
| { |
| "epoch": 0.5907079646017699, |
| "grad_norm": 0.17565971612930298, |
| "learning_rate": 9.718200383369891e-06, |
| "loss": 0.402, |
| "step": 267 |
| }, |
| { |
| "epoch": 0.5929203539823009, |
| "grad_norm": 0.1879590004682541, |
| "learning_rate": 9.713923336375936e-06, |
| "loss": 0.4109, |
| "step": 268 |
| }, |
| { |
| "epoch": 0.5951327433628318, |
| "grad_norm": 0.16611909866333008, |
| "learning_rate": 9.709615031297598e-06, |
| "loss": 0.3957, |
| "step": 269 |
| }, |
| { |
| "epoch": 0.5973451327433629, |
| "grad_norm": 0.19158808887004852, |
| "learning_rate": 9.705275496703302e-06, |
| "loss": 0.4125, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.5995575221238938, |
| "grad_norm": 0.19456025958061218, |
| "learning_rate": 9.70090476136855e-06, |
| "loss": 0.4079, |
| "step": 271 |
| }, |
| { |
| "epoch": 0.6017699115044248, |
| "grad_norm": 0.17634005844593048, |
| "learning_rate": 9.69650285427575e-06, |
| "loss": 0.4025, |
| "step": 272 |
| }, |
| { |
| "epoch": 0.6039823008849557, |
| "grad_norm": 0.20455583930015564, |
| "learning_rate": 9.692069804613995e-06, |
| "loss": 0.4101, |
| "step": 273 |
| }, |
| { |
| "epoch": 0.6061946902654868, |
| "grad_norm": 0.18841254711151123, |
| "learning_rate": 9.687605641778899e-06, |
| "loss": 0.4026, |
| "step": 274 |
| }, |
| { |
| "epoch": 0.6084070796460177, |
| "grad_norm": 0.19969072937965393, |
| "learning_rate": 9.683110395372379e-06, |
| "loss": 0.4004, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.6106194690265486, |
| "grad_norm": 0.2100944221019745, |
| "learning_rate": 9.678584095202468e-06, |
| "loss": 0.4103, |
| "step": 276 |
| }, |
| { |
| "epoch": 0.6128318584070797, |
| "grad_norm": 0.16795648634433746, |
| "learning_rate": 9.674026771283123e-06, |
| "loss": 0.4046, |
| "step": 277 |
| }, |
| { |
| "epoch": 0.6150442477876106, |
| "grad_norm": 0.19852401316165924, |
| "learning_rate": 9.669438453834014e-06, |
| "loss": 0.4053, |
| "step": 278 |
| }, |
| { |
| "epoch": 0.6172566371681416, |
| "grad_norm": 0.18384751677513123, |
| "learning_rate": 9.664819173280328e-06, |
| "loss": 0.4106, |
| "step": 279 |
| }, |
| { |
| "epoch": 0.6194690265486725, |
| "grad_norm": 0.16662909090518951, |
| "learning_rate": 9.660168960252575e-06, |
| "loss": 0.4158, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.6216814159292036, |
| "grad_norm": 0.22212032973766327, |
| "learning_rate": 9.655487845586378e-06, |
| "loss": 0.4156, |
| "step": 281 |
| }, |
| { |
| "epoch": 0.6238938053097345, |
| "grad_norm": 0.17248587310314178, |
| "learning_rate": 9.650775860322263e-06, |
| "loss": 0.3941, |
| "step": 282 |
| }, |
| { |
| "epoch": 0.6261061946902655, |
| "grad_norm": 0.2107280045747757, |
| "learning_rate": 9.646033035705462e-06, |
| "loss": 0.4202, |
| "step": 283 |
| }, |
| { |
| "epoch": 0.6283185840707964, |
| "grad_norm": 0.16596029698848724, |
| "learning_rate": 9.641259403185706e-06, |
| "loss": 0.4061, |
| "step": 284 |
| }, |
| { |
| "epoch": 0.6305309734513275, |
| "grad_norm": 0.18452270328998566, |
| "learning_rate": 9.636454994417013e-06, |
| "loss": 0.4023, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.6327433628318584, |
| "grad_norm": 0.17428122460842133, |
| "learning_rate": 9.631619841257477e-06, |
| "loss": 0.3974, |
| "step": 286 |
| }, |
| { |
| "epoch": 0.6349557522123894, |
| "grad_norm": 0.18349966406822205, |
| "learning_rate": 9.626753975769054e-06, |
| "loss": 0.4051, |
| "step": 287 |
| }, |
| { |
| "epoch": 0.6371681415929203, |
| "grad_norm": 0.18694283068180084, |
| "learning_rate": 9.621857430217366e-06, |
| "loss": 0.4141, |
| "step": 288 |
| }, |
| { |
| "epoch": 0.6393805309734514, |
| "grad_norm": 0.196182981133461, |
| "learning_rate": 9.616930237071464e-06, |
| "loss": 0.4037, |
| "step": 289 |
| }, |
| { |
| "epoch": 0.6415929203539823, |
| "grad_norm": 0.1949911266565323, |
| "learning_rate": 9.611972429003626e-06, |
| "loss": 0.411, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.6438053097345132, |
| "grad_norm": 0.21375317871570587, |
| "learning_rate": 9.606984038889142e-06, |
| "loss": 0.4025, |
| "step": 291 |
| }, |
| { |
| "epoch": 0.6460176991150443, |
| "grad_norm": 0.2112773060798645, |
| "learning_rate": 9.601965099806085e-06, |
| "loss": 0.3962, |
| "step": 292 |
| }, |
| { |
| "epoch": 0.6482300884955752, |
| "grad_norm": 0.1982385218143463, |
| "learning_rate": 9.596915645035107e-06, |
| "loss": 0.4116, |
| "step": 293 |
| }, |
| { |
| "epoch": 0.6504424778761062, |
| "grad_norm": 0.18961909413337708, |
| "learning_rate": 9.591835708059202e-06, |
| "loss": 0.4016, |
| "step": 294 |
| }, |
| { |
| "epoch": 0.6526548672566371, |
| "grad_norm": 0.205304354429245, |
| "learning_rate": 9.5867253225635e-06, |
| "loss": 0.4107, |
| "step": 295 |
| }, |
| { |
| "epoch": 0.6548672566371682, |
| "grad_norm": 0.1793058067560196, |
| "learning_rate": 9.581584522435025e-06, |
| "loss": 0.4091, |
| "step": 296 |
| }, |
| { |
| "epoch": 0.6570796460176991, |
| "grad_norm": 0.2273833453655243, |
| "learning_rate": 9.57641334176249e-06, |
| "loss": 0.4079, |
| "step": 297 |
| }, |
| { |
| "epoch": 0.6592920353982301, |
| "grad_norm": 0.19448456168174744, |
| "learning_rate": 9.571211814836059e-06, |
| "loss": 0.4126, |
| "step": 298 |
| }, |
| { |
| "epoch": 0.661504424778761, |
| "grad_norm": 0.18413153290748596, |
| "learning_rate": 9.56597997614712e-06, |
| "loss": 0.4023, |
| "step": 299 |
| }, |
| { |
| "epoch": 0.6637168141592921, |
| "grad_norm": 0.20027752220630646, |
| "learning_rate": 9.560717860388061e-06, |
| "loss": 0.4111, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.665929203539823, |
| "grad_norm": 0.1603672206401825, |
| "learning_rate": 9.555425502452038e-06, |
| "loss": 0.4036, |
| "step": 301 |
| }, |
| { |
| "epoch": 0.668141592920354, |
| "grad_norm": 0.18299739062786102, |
| "learning_rate": 9.550102937432743e-06, |
| "loss": 0.3942, |
| "step": 302 |
| }, |
| { |
| "epoch": 0.6703539823008849, |
| "grad_norm": 0.1947639435529709, |
| "learning_rate": 9.544750200624169e-06, |
| "loss": 0.4054, |
| "step": 303 |
| }, |
| { |
| "epoch": 0.672566371681416, |
| "grad_norm": 0.16402120888233185, |
| "learning_rate": 9.539367327520382e-06, |
| "loss": 0.4048, |
| "step": 304 |
| }, |
| { |
| "epoch": 0.6747787610619469, |
| "grad_norm": 0.17655643820762634, |
| "learning_rate": 9.533954353815279e-06, |
| "loss": 0.4068, |
| "step": 305 |
| }, |
| { |
| "epoch": 0.6769911504424779, |
| "grad_norm": 0.1705683320760727, |
| "learning_rate": 9.528511315402358e-06, |
| "loss": 0.3953, |
| "step": 306 |
| }, |
| { |
| "epoch": 0.6792035398230089, |
| "grad_norm": 0.16182245314121246, |
| "learning_rate": 9.523038248374474e-06, |
| "loss": 0.408, |
| "step": 307 |
| }, |
| { |
| "epoch": 0.6814159292035398, |
| "grad_norm": 0.20871901512145996, |
| "learning_rate": 9.517535189023602e-06, |
| "loss": 0.4031, |
| "step": 308 |
| }, |
| { |
| "epoch": 0.6836283185840708, |
| "grad_norm": 0.13933664560317993, |
| "learning_rate": 9.512002173840597e-06, |
| "loss": 0.405, |
| "step": 309 |
| }, |
| { |
| "epoch": 0.6858407079646017, |
| "grad_norm": 0.20702913403511047, |
| "learning_rate": 9.506439239514954e-06, |
| "loss": 0.4012, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.6880530973451328, |
| "grad_norm": 0.1594098061323166, |
| "learning_rate": 9.500846422934557e-06, |
| "loss": 0.3926, |
| "step": 311 |
| }, |
| { |
| "epoch": 0.6902654867256637, |
| "grad_norm": 0.19512666761875153, |
| "learning_rate": 9.495223761185443e-06, |
| "loss": 0.4031, |
| "step": 312 |
| }, |
| { |
| "epoch": 0.6924778761061947, |
| "grad_norm": 0.14987067878246307, |
| "learning_rate": 9.489571291551553e-06, |
| "loss": 0.4138, |
| "step": 313 |
| }, |
| { |
| "epoch": 0.6946902654867256, |
| "grad_norm": 0.18819770216941833, |
| "learning_rate": 9.483889051514483e-06, |
| "loss": 0.4022, |
| "step": 314 |
| }, |
| { |
| "epoch": 0.6969026548672567, |
| "grad_norm": 0.16024132072925568, |
| "learning_rate": 9.478177078753236e-06, |
| "loss": 0.4044, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.6991150442477876, |
| "grad_norm": 0.18151608109474182, |
| "learning_rate": 9.472435411143979e-06, |
| "loss": 0.4081, |
| "step": 316 |
| }, |
| { |
| "epoch": 0.7013274336283186, |
| "grad_norm": 0.17617082595825195, |
| "learning_rate": 9.466664086759777e-06, |
| "loss": 0.4134, |
| "step": 317 |
| }, |
| { |
| "epoch": 0.7035398230088495, |
| "grad_norm": 0.17612871527671814, |
| "learning_rate": 9.460863143870355e-06, |
| "loss": 0.4089, |
| "step": 318 |
| }, |
| { |
| "epoch": 0.7057522123893806, |
| "grad_norm": 0.17769289016723633, |
| "learning_rate": 9.45503262094184e-06, |
| "loss": 0.4047, |
| "step": 319 |
| }, |
| { |
| "epoch": 0.7079646017699115, |
| "grad_norm": 0.1573716551065445, |
| "learning_rate": 9.4491725566365e-06, |
| "loss": 0.4029, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.7101769911504425, |
| "grad_norm": 0.17037460207939148, |
| "learning_rate": 9.443282989812494e-06, |
| "loss": 0.4059, |
| "step": 321 |
| }, |
| { |
| "epoch": 0.7123893805309734, |
| "grad_norm": 0.1654326617717743, |
| "learning_rate": 9.437363959523613e-06, |
| "loss": 0.4099, |
| "step": 322 |
| }, |
| { |
| "epoch": 0.7146017699115044, |
| "grad_norm": 0.1739991158246994, |
| "learning_rate": 9.431415505019024e-06, |
| "loss": 0.3916, |
| "step": 323 |
| }, |
| { |
| "epoch": 0.7168141592920354, |
| "grad_norm": 0.156654492020607, |
| "learning_rate": 9.425437665742998e-06, |
| "loss": 0.3912, |
| "step": 324 |
| }, |
| { |
| "epoch": 0.7190265486725663, |
| "grad_norm": 0.16226203739643097, |
| "learning_rate": 9.419430481334663e-06, |
| "loss": 0.3995, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.7212389380530974, |
| "grad_norm": 0.14988061785697937, |
| "learning_rate": 9.413393991627737e-06, |
| "loss": 0.3959, |
| "step": 326 |
| }, |
| { |
| "epoch": 0.7234513274336283, |
| "grad_norm": 0.1781560331583023, |
| "learning_rate": 9.407328236650257e-06, |
| "loss": 0.3957, |
| "step": 327 |
| }, |
| { |
| "epoch": 0.7256637168141593, |
| "grad_norm": 0.17818742990493774, |
| "learning_rate": 9.401233256624318e-06, |
| "loss": 0.3949, |
| "step": 328 |
| }, |
| { |
| "epoch": 0.7278761061946902, |
| "grad_norm": 0.1621919423341751, |
| "learning_rate": 9.395109091965808e-06, |
| "loss": 0.3978, |
| "step": 329 |
| }, |
| { |
| "epoch": 0.7300884955752213, |
| "grad_norm": 0.1838427484035492, |
| "learning_rate": 9.388955783284141e-06, |
| "loss": 0.3998, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.7323008849557522, |
| "grad_norm": 0.17227084934711456, |
| "learning_rate": 9.382773371381986e-06, |
| "loss": 0.3989, |
| "step": 331 |
| }, |
| { |
| "epoch": 0.7345132743362832, |
| "grad_norm": 0.18801696598529816, |
| "learning_rate": 9.376561897254987e-06, |
| "loss": 0.3956, |
| "step": 332 |
| }, |
| { |
| "epoch": 0.7367256637168141, |
| "grad_norm": 0.15474942326545715, |
| "learning_rate": 9.370321402091514e-06, |
| "loss": 0.4035, |
| "step": 333 |
| }, |
| { |
| "epoch": 0.7389380530973452, |
| "grad_norm": 0.16465848684310913, |
| "learning_rate": 9.36405192727236e-06, |
| "loss": 0.4016, |
| "step": 334 |
| }, |
| { |
| "epoch": 0.7411504424778761, |
| "grad_norm": 0.19732873141765594, |
| "learning_rate": 9.357753514370497e-06, |
| "loss": 0.4038, |
| "step": 335 |
| }, |
| { |
| "epoch": 0.7433628318584071, |
| "grad_norm": 0.16492311656475067, |
| "learning_rate": 9.351426205150778e-06, |
| "loss": 0.4015, |
| "step": 336 |
| }, |
| { |
| "epoch": 0.745575221238938, |
| "grad_norm": 0.18818548321723938, |
| "learning_rate": 9.345070041569666e-06, |
| "loss": 0.399, |
| "step": 337 |
| }, |
| { |
| "epoch": 0.7477876106194691, |
| "grad_norm": 0.17031817138195038, |
| "learning_rate": 9.338685065774964e-06, |
| "loss": 0.3937, |
| "step": 338 |
| }, |
| { |
| "epoch": 0.75, |
| "grad_norm": 0.24733419716358185, |
| "learning_rate": 9.332271320105527e-06, |
| "loss": 0.4081, |
| "step": 339 |
| }, |
| { |
| "epoch": 0.7522123893805309, |
| "grad_norm": 0.18530051410198212, |
| "learning_rate": 9.32582884709098e-06, |
| "loss": 0.4007, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.754424778761062, |
| "grad_norm": 0.18777970969676971, |
| "learning_rate": 9.319357689451444e-06, |
| "loss": 0.3989, |
| "step": 341 |
| }, |
| { |
| "epoch": 0.7566371681415929, |
| "grad_norm": 0.18476152420043945, |
| "learning_rate": 9.312857890097243e-06, |
| "loss": 0.399, |
| "step": 342 |
| }, |
| { |
| "epoch": 0.7588495575221239, |
| "grad_norm": 0.17613354325294495, |
| "learning_rate": 9.30632949212863e-06, |
| "loss": 0.4026, |
| "step": 343 |
| }, |
| { |
| "epoch": 0.7610619469026548, |
| "grad_norm": 0.17226579785346985, |
| "learning_rate": 9.299772538835492e-06, |
| "loss": 0.4051, |
| "step": 344 |
| }, |
| { |
| "epoch": 0.7632743362831859, |
| "grad_norm": 0.18831267952919006, |
| "learning_rate": 9.29318707369707e-06, |
| "loss": 0.3927, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.7654867256637168, |
| "grad_norm": 0.15509851276874542, |
| "learning_rate": 9.286573140381663e-06, |
| "loss": 0.3956, |
| "step": 346 |
| }, |
| { |
| "epoch": 0.7676991150442478, |
| "grad_norm": 0.18900151550769806, |
| "learning_rate": 9.279930782746346e-06, |
| "loss": 0.3975, |
| "step": 347 |
| }, |
| { |
| "epoch": 0.7699115044247787, |
| "grad_norm": 0.16996918618679047, |
| "learning_rate": 9.273260044836675e-06, |
| "loss": 0.3972, |
| "step": 348 |
| }, |
| { |
| "epoch": 0.7721238938053098, |
| "grad_norm": 0.17968420684337616, |
| "learning_rate": 9.266560970886397e-06, |
| "loss": 0.4026, |
| "step": 349 |
| }, |
| { |
| "epoch": 0.7743362831858407, |
| "grad_norm": 0.1665457934141159, |
| "learning_rate": 9.259833605317155e-06, |
| "loss": 0.3946, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.7765486725663717, |
| "grad_norm": 0.18998298048973083, |
| "learning_rate": 9.253077992738193e-06, |
| "loss": 0.3964, |
| "step": 351 |
| }, |
| { |
| "epoch": 0.7787610619469026, |
| "grad_norm": 0.17083673179149628, |
| "learning_rate": 9.246294177946062e-06, |
| "loss": 0.4026, |
| "step": 352 |
| }, |
| { |
| "epoch": 0.7809734513274337, |
| "grad_norm": 0.1637433022260666, |
| "learning_rate": 9.239482205924322e-06, |
| "loss": 0.3922, |
| "step": 353 |
| }, |
| { |
| "epoch": 0.7831858407079646, |
| "grad_norm": 0.1800343543291092, |
| "learning_rate": 9.232642121843247e-06, |
| "loss": 0.407, |
| "step": 354 |
| }, |
| { |
| "epoch": 0.7853982300884956, |
| "grad_norm": 0.16393564641475677, |
| "learning_rate": 9.225773971059518e-06, |
| "loss": 0.4057, |
| "step": 355 |
| }, |
| { |
| "epoch": 0.7876106194690266, |
| "grad_norm": 0.18692757189273834, |
| "learning_rate": 9.218877799115929e-06, |
| "loss": 0.3979, |
| "step": 356 |
| }, |
| { |
| "epoch": 0.7898230088495575, |
| "grad_norm": 0.17437385022640228, |
| "learning_rate": 9.21195365174108e-06, |
| "loss": 0.4057, |
| "step": 357 |
| }, |
| { |
| "epoch": 0.7920353982300885, |
| "grad_norm": 0.1898239552974701, |
| "learning_rate": 9.205001574849081e-06, |
| "loss": 0.4039, |
| "step": 358 |
| }, |
| { |
| "epoch": 0.7942477876106194, |
| "grad_norm": 0.1568395495414734, |
| "learning_rate": 9.19802161453924e-06, |
| "loss": 0.4039, |
| "step": 359 |
| }, |
| { |
| "epoch": 0.7964601769911505, |
| "grad_norm": 0.19088448584079742, |
| "learning_rate": 9.191013817095762e-06, |
| "loss": 0.3953, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.7986725663716814, |
| "grad_norm": 0.17273734509944916, |
| "learning_rate": 9.183978228987436e-06, |
| "loss": 0.4104, |
| "step": 361 |
| }, |
| { |
| "epoch": 0.8008849557522124, |
| "grad_norm": 0.17058604955673218, |
| "learning_rate": 9.176914896867335e-06, |
| "loss": 0.398, |
| "step": 362 |
| }, |
| { |
| "epoch": 0.8030973451327433, |
| "grad_norm": 0.18058565258979797, |
| "learning_rate": 9.169823867572505e-06, |
| "loss": 0.4076, |
| "step": 363 |
| }, |
| { |
| "epoch": 0.8053097345132744, |
| "grad_norm": 0.1585603505373001, |
| "learning_rate": 9.162705188123647e-06, |
| "loss": 0.3988, |
| "step": 364 |
| }, |
| { |
| "epoch": 0.8075221238938053, |
| "grad_norm": 0.1937795877456665, |
| "learning_rate": 9.155558905724815e-06, |
| "loss": 0.4081, |
| "step": 365 |
| }, |
| { |
| "epoch": 0.8097345132743363, |
| "grad_norm": 0.1497035026550293, |
| "learning_rate": 9.148385067763094e-06, |
| "loss": 0.3921, |
| "step": 366 |
| }, |
| { |
| "epoch": 0.8119469026548672, |
| "grad_norm": 0.184015691280365, |
| "learning_rate": 9.141183721808298e-06, |
| "loss": 0.3959, |
| "step": 367 |
| }, |
| { |
| "epoch": 0.8141592920353983, |
| "grad_norm": 0.16560928523540497, |
| "learning_rate": 9.133954915612635e-06, |
| "loss": 0.4016, |
| "step": 368 |
| }, |
| { |
| "epoch": 0.8163716814159292, |
| "grad_norm": 0.17013601958751678, |
| "learning_rate": 9.126698697110414e-06, |
| "loss": 0.4052, |
| "step": 369 |
| }, |
| { |
| "epoch": 0.8185840707964602, |
| "grad_norm": 0.16712944209575653, |
| "learning_rate": 9.119415114417709e-06, |
| "loss": 0.3995, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.8207964601769911, |
| "grad_norm": 0.16210488975048065, |
| "learning_rate": 9.112104215832047e-06, |
| "loss": 0.4094, |
| "step": 371 |
| }, |
| { |
| "epoch": 0.8230088495575221, |
| "grad_norm": 0.1718660295009613, |
| "learning_rate": 9.104766049832088e-06, |
| "loss": 0.3852, |
| "step": 372 |
| }, |
| { |
| "epoch": 0.8252212389380531, |
| "grad_norm": 0.1845821589231491, |
| "learning_rate": 9.0974006650773e-06, |
| "loss": 0.4091, |
| "step": 373 |
| }, |
| { |
| "epoch": 0.827433628318584, |
| "grad_norm": 0.15908625721931458, |
| "learning_rate": 9.090008110407646e-06, |
| "loss": 0.4134, |
| "step": 374 |
| }, |
| { |
| "epoch": 0.8296460176991151, |
| "grad_norm": 0.159079447388649, |
| "learning_rate": 9.082588434843244e-06, |
| "loss": 0.4003, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.831858407079646, |
| "grad_norm": 0.1775507628917694, |
| "learning_rate": 9.075141687584056e-06, |
| "loss": 0.4059, |
| "step": 376 |
| }, |
| { |
| "epoch": 0.834070796460177, |
| "grad_norm": 0.1623350828886032, |
| "learning_rate": 9.067667918009559e-06, |
| "loss": 0.3968, |
| "step": 377 |
| }, |
| { |
| "epoch": 0.8362831858407079, |
| "grad_norm": 0.17421099543571472, |
| "learning_rate": 9.060167175678407e-06, |
| "loss": 0.3928, |
| "step": 378 |
| }, |
| { |
| "epoch": 0.838495575221239, |
| "grad_norm": 0.16807816922664642, |
| "learning_rate": 9.05263951032812e-06, |
| "loss": 0.4035, |
| "step": 379 |
| }, |
| { |
| "epoch": 0.8407079646017699, |
| "grad_norm": 0.16292251646518707, |
| "learning_rate": 9.045084971874738e-06, |
| "loss": 0.4017, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.8429203539823009, |
| "grad_norm": 0.16168609261512756, |
| "learning_rate": 9.037503610412502e-06, |
| "loss": 0.3929, |
| "step": 381 |
| }, |
| { |
| "epoch": 0.8451327433628318, |
| "grad_norm": 0.1535169780254364, |
| "learning_rate": 9.02989547621351e-06, |
| "loss": 0.3944, |
| "step": 382 |
| }, |
| { |
| "epoch": 0.8473451327433629, |
| "grad_norm": 0.16936345398426056, |
| "learning_rate": 9.022260619727401e-06, |
| "loss": 0.3989, |
| "step": 383 |
| }, |
| { |
| "epoch": 0.8495575221238938, |
| "grad_norm": 0.1527484804391861, |
| "learning_rate": 9.014599091581e-06, |
| "loss": 0.3952, |
| "step": 384 |
| }, |
| { |
| "epoch": 0.8517699115044248, |
| "grad_norm": 0.17519870400428772, |
| "learning_rate": 9.006910942577995e-06, |
| "loss": 0.401, |
| "step": 385 |
| }, |
| { |
| "epoch": 0.8539823008849557, |
| "grad_norm": 0.1557687520980835, |
| "learning_rate": 8.999196223698599e-06, |
| "loss": 0.4076, |
| "step": 386 |
| }, |
| { |
| "epoch": 0.8561946902654868, |
| "grad_norm": 0.16832038760185242, |
| "learning_rate": 8.991454986099207e-06, |
| "loss": 0.3916, |
| "step": 387 |
| }, |
| { |
| "epoch": 0.8584070796460177, |
| "grad_norm": 0.16987960040569305, |
| "learning_rate": 8.983687281112066e-06, |
| "loss": 0.3921, |
| "step": 388 |
| }, |
| { |
| "epoch": 0.8606194690265486, |
| "grad_norm": 0.1476796269416809, |
| "learning_rate": 8.975893160244921e-06, |
| "loss": 0.3886, |
| "step": 389 |
| }, |
| { |
| "epoch": 0.8628318584070797, |
| "grad_norm": 0.16488701105117798, |
| "learning_rate": 8.968072675180686e-06, |
| "loss": 0.3944, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.8650442477876106, |
| "grad_norm": 0.1628437340259552, |
| "learning_rate": 8.960225877777095e-06, |
| "loss": 0.3867, |
| "step": 391 |
| }, |
| { |
| "epoch": 0.8672566371681416, |
| "grad_norm": 0.1530245542526245, |
| "learning_rate": 8.952352820066359e-06, |
| "loss": 0.4099, |
| "step": 392 |
| }, |
| { |
| "epoch": 0.8694690265486725, |
| "grad_norm": 0.16439391672611237, |
| "learning_rate": 8.944453554254823e-06, |
| "loss": 0.3991, |
| "step": 393 |
| }, |
| { |
| "epoch": 0.8716814159292036, |
| "grad_norm": 0.17328006029129028, |
| "learning_rate": 8.936528132722616e-06, |
| "loss": 0.3924, |
| "step": 394 |
| }, |
| { |
| "epoch": 0.8738938053097345, |
| "grad_norm": 0.16984131932258606, |
| "learning_rate": 8.928576608023305e-06, |
| "loss": 0.4048, |
| "step": 395 |
| }, |
| { |
| "epoch": 0.8761061946902655, |
| "grad_norm": 0.14963462948799133, |
| "learning_rate": 8.920599032883553e-06, |
| "loss": 0.3918, |
| "step": 396 |
| }, |
| { |
| "epoch": 0.8783185840707964, |
| "grad_norm": 0.18854303658008575, |
| "learning_rate": 8.912595460202758e-06, |
| "loss": 0.3981, |
| "step": 397 |
| }, |
| { |
| "epoch": 0.8805309734513275, |
| "grad_norm": 0.19335384666919708, |
| "learning_rate": 8.90456594305271e-06, |
| "loss": 0.3984, |
| "step": 398 |
| }, |
| { |
| "epoch": 0.8827433628318584, |
| "grad_norm": 0.17484664916992188, |
| "learning_rate": 8.896510534677238e-06, |
| "loss": 0.392, |
| "step": 399 |
| }, |
| { |
| "epoch": 0.8849557522123894, |
| "grad_norm": 0.19245024025440216, |
| "learning_rate": 8.888429288491857e-06, |
| "loss": 0.3949, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.8871681415929203, |
| "grad_norm": 0.15718188881874084, |
| "learning_rate": 8.880322258083408e-06, |
| "loss": 0.3973, |
| "step": 401 |
| }, |
| { |
| "epoch": 0.8893805309734514, |
| "grad_norm": 0.17654351890087128, |
| "learning_rate": 8.872189497209712e-06, |
| "loss": 0.3996, |
| "step": 402 |
| }, |
| { |
| "epoch": 0.8915929203539823, |
| "grad_norm": 0.16744081676006317, |
| "learning_rate": 8.864031059799208e-06, |
| "loss": 0.3879, |
| "step": 403 |
| }, |
| { |
| "epoch": 0.8938053097345132, |
| "grad_norm": 0.16980023682117462, |
| "learning_rate": 8.855846999950595e-06, |
| "loss": 0.4029, |
| "step": 404 |
| }, |
| { |
| "epoch": 0.8960176991150443, |
| "grad_norm": 0.17948906123638153, |
| "learning_rate": 8.847637371932478e-06, |
| "loss": 0.3938, |
| "step": 405 |
| }, |
| { |
| "epoch": 0.8982300884955752, |
| "grad_norm": 0.1663169413805008, |
| "learning_rate": 8.839402230183e-06, |
| "loss": 0.404, |
| "step": 406 |
| }, |
| { |
| "epoch": 0.9004424778761062, |
| "grad_norm": 0.1661122441291809, |
| "learning_rate": 8.831141629309492e-06, |
| "loss": 0.3923, |
| "step": 407 |
| }, |
| { |
| "epoch": 0.9026548672566371, |
| "grad_norm": 0.17008569836616516, |
| "learning_rate": 8.822855624088099e-06, |
| "loss": 0.3911, |
| "step": 408 |
| }, |
| { |
| "epoch": 0.9048672566371682, |
| "grad_norm": 0.16437990963459015, |
| "learning_rate": 8.814544269463422e-06, |
| "loss": 0.3952, |
| "step": 409 |
| }, |
| { |
| "epoch": 0.9070796460176991, |
| "grad_norm": 0.18122254312038422, |
| "learning_rate": 8.806207620548165e-06, |
| "loss": 0.3912, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.9092920353982301, |
| "grad_norm": 0.1973719596862793, |
| "learning_rate": 8.797845732622742e-06, |
| "loss": 0.3978, |
| "step": 411 |
| }, |
| { |
| "epoch": 0.911504424778761, |
| "grad_norm": 0.1716318130493164, |
| "learning_rate": 8.789458661134943e-06, |
| "loss": 0.3966, |
| "step": 412 |
| }, |
| { |
| "epoch": 0.9137168141592921, |
| "grad_norm": 0.1809050291776657, |
| "learning_rate": 8.781046461699538e-06, |
| "loss": 0.3933, |
| "step": 413 |
| }, |
| { |
| "epoch": 0.915929203539823, |
| "grad_norm": 0.1711595505475998, |
| "learning_rate": 8.772609190097932e-06, |
| "loss": 0.3912, |
| "step": 414 |
| }, |
| { |
| "epoch": 0.918141592920354, |
| "grad_norm": 0.16632473468780518, |
| "learning_rate": 8.764146902277773e-06, |
| "loss": 0.3851, |
| "step": 415 |
| }, |
| { |
| "epoch": 0.9203539823008849, |
| "grad_norm": 0.15847158432006836, |
| "learning_rate": 8.755659654352599e-06, |
| "loss": 0.3958, |
| "step": 416 |
| }, |
| { |
| "epoch": 0.922566371681416, |
| "grad_norm": 0.16290074586868286, |
| "learning_rate": 8.747147502601458e-06, |
| "loss": 0.3965, |
| "step": 417 |
| }, |
| { |
| "epoch": 0.9247787610619469, |
| "grad_norm": 0.15800520777702332, |
| "learning_rate": 8.738610503468534e-06, |
| "loss": 0.3952, |
| "step": 418 |
| }, |
| { |
| "epoch": 0.9269911504424779, |
| "grad_norm": 0.13968175649642944, |
| "learning_rate": 8.730048713562771e-06, |
| "loss": 0.3935, |
| "step": 419 |
| }, |
| { |
| "epoch": 0.9292035398230089, |
| "grad_norm": 0.15484796464443207, |
| "learning_rate": 8.72146218965751e-06, |
| "loss": 0.396, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.9314159292035398, |
| "grad_norm": 0.14914363622665405, |
| "learning_rate": 8.712850988690094e-06, |
| "loss": 0.4041, |
| "step": 421 |
| }, |
| { |
| "epoch": 0.9336283185840708, |
| "grad_norm": 0.16192984580993652, |
| "learning_rate": 8.704215167761506e-06, |
| "loss": 0.3897, |
| "step": 422 |
| }, |
| { |
| "epoch": 0.9358407079646017, |
| "grad_norm": 0.17002323269844055, |
| "learning_rate": 8.695554784135982e-06, |
| "loss": 0.3902, |
| "step": 423 |
| }, |
| { |
| "epoch": 0.9380530973451328, |
| "grad_norm": 0.15230736136436462, |
| "learning_rate": 8.686869895240631e-06, |
| "loss": 0.398, |
| "step": 424 |
| }, |
| { |
| "epoch": 0.9402654867256637, |
| "grad_norm": 0.2072356790304184, |
| "learning_rate": 8.678160558665063e-06, |
| "loss": 0.4063, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.9424778761061947, |
| "grad_norm": 0.16368243098258972, |
| "learning_rate": 8.669426832160997e-06, |
| "loss": 0.3988, |
| "step": 426 |
| }, |
| { |
| "epoch": 0.9446902654867256, |
| "grad_norm": 0.16881467401981354, |
| "learning_rate": 8.66066877364188e-06, |
| "loss": 0.3986, |
| "step": 427 |
| }, |
| { |
| "epoch": 0.9469026548672567, |
| "grad_norm": 0.17835210263729095, |
| "learning_rate": 8.651886441182509e-06, |
| "loss": 0.3969, |
| "step": 428 |
| }, |
| { |
| "epoch": 0.9491150442477876, |
| "grad_norm": 0.15917813777923584, |
| "learning_rate": 8.64307989301864e-06, |
| "loss": 0.3793, |
| "step": 429 |
| }, |
| { |
| "epoch": 0.9513274336283186, |
| "grad_norm": 0.1801290661096573, |
| "learning_rate": 8.634249187546601e-06, |
| "loss": 0.3959, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.9535398230088495, |
| "grad_norm": 0.16084758937358856, |
| "learning_rate": 8.625394383322914e-06, |
| "loss": 0.4085, |
| "step": 431 |
| }, |
| { |
| "epoch": 0.9557522123893806, |
| "grad_norm": 0.17122162878513336, |
| "learning_rate": 8.616515539063894e-06, |
| "loss": 0.3996, |
| "step": 432 |
| }, |
| { |
| "epoch": 0.9579646017699115, |
| "grad_norm": 0.16651393473148346, |
| "learning_rate": 8.60761271364527e-06, |
| "loss": 0.4013, |
| "step": 433 |
| }, |
| { |
| "epoch": 0.9601769911504425, |
| "grad_norm": 0.15684399008750916, |
| "learning_rate": 8.598685966101783e-06, |
| "loss": 0.3875, |
| "step": 434 |
| }, |
| { |
| "epoch": 0.9623893805309734, |
| "grad_norm": 0.15404202044010162, |
| "learning_rate": 8.589735355626814e-06, |
| "loss": 0.3929, |
| "step": 435 |
| }, |
| { |
| "epoch": 0.9646017699115044, |
| "grad_norm": 0.1662074774503708, |
| "learning_rate": 8.580760941571968e-06, |
| "loss": 0.3957, |
| "step": 436 |
| }, |
| { |
| "epoch": 0.9668141592920354, |
| "grad_norm": 0.15725825726985931, |
| "learning_rate": 8.571762783446696e-06, |
| "loss": 0.3914, |
| "step": 437 |
| }, |
| { |
| "epoch": 0.9690265486725663, |
| "grad_norm": 0.1751137673854828, |
| "learning_rate": 8.562740940917901e-06, |
| "loss": 0.4001, |
| "step": 438 |
| }, |
| { |
| "epoch": 0.9712389380530974, |
| "grad_norm": 0.1850391924381256, |
| "learning_rate": 8.55369547380953e-06, |
| "loss": 0.39, |
| "step": 439 |
| }, |
| { |
| "epoch": 0.9734513274336283, |
| "grad_norm": 0.16561464965343475, |
| "learning_rate": 8.544626442102188e-06, |
| "loss": 0.3903, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.9756637168141593, |
| "grad_norm": 0.16364803910255432, |
| "learning_rate": 8.535533905932739e-06, |
| "loss": 0.3862, |
| "step": 441 |
| }, |
| { |
| "epoch": 0.9778761061946902, |
| "grad_norm": 0.1534828543663025, |
| "learning_rate": 8.526417925593901e-06, |
| "loss": 0.3986, |
| "step": 442 |
| }, |
| { |
| "epoch": 0.9800884955752213, |
| "grad_norm": 0.17045795917510986, |
| "learning_rate": 8.517278561533857e-06, |
| "loss": 0.393, |
| "step": 443 |
| }, |
| { |
| "epoch": 0.9823008849557522, |
| "grad_norm": 0.16603973507881165, |
| "learning_rate": 8.50811587435584e-06, |
| "loss": 0.3915, |
| "step": 444 |
| }, |
| { |
| "epoch": 0.9845132743362832, |
| "grad_norm": 0.17057141661643982, |
| "learning_rate": 8.498929924817745e-06, |
| "loss": 0.3839, |
| "step": 445 |
| }, |
| { |
| "epoch": 0.9867256637168141, |
| "grad_norm": 0.16480037569999695, |
| "learning_rate": 8.489720773831717e-06, |
| "loss": 0.3931, |
| "step": 446 |
| }, |
| { |
| "epoch": 0.9889380530973452, |
| "grad_norm": 0.15113525092601776, |
| "learning_rate": 8.480488482463753e-06, |
| "loss": 0.4149, |
| "step": 447 |
| }, |
| { |
| "epoch": 0.9911504424778761, |
| "grad_norm": 0.17928218841552734, |
| "learning_rate": 8.471233111933291e-06, |
| "loss": 0.3984, |
| "step": 448 |
| }, |
| { |
| "epoch": 0.9933628318584071, |
| "grad_norm": 0.1759202629327774, |
| "learning_rate": 8.461954723612807e-06, |
| "loss": 0.3989, |
| "step": 449 |
| }, |
| { |
| "epoch": 0.995575221238938, |
| "grad_norm": 0.15595725178718567, |
| "learning_rate": 8.45265337902741e-06, |
| "loss": 0.3943, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.9977876106194691, |
| "grad_norm": 0.17772260308265686, |
| "learning_rate": 8.443329139854434e-06, |
| "loss": 0.3925, |
| "step": 451 |
| }, |
| { |
| "epoch": 1.0, |
| "grad_norm": 0.17362035810947418, |
| "learning_rate": 8.433982067923021e-06, |
| "loss": 0.3919, |
| "step": 452 |
| }, |
| { |
| "epoch": 1.002212389380531, |
| "grad_norm": 0.1918303370475769, |
| "learning_rate": 8.424612225213726e-06, |
| "loss": 0.3747, |
| "step": 453 |
| }, |
| { |
| "epoch": 1.0044247787610618, |
| "grad_norm": 0.1567537784576416, |
| "learning_rate": 8.41521967385809e-06, |
| "loss": 0.3705, |
| "step": 454 |
| }, |
| { |
| "epoch": 1.0066371681415929, |
| "grad_norm": 0.20494352281093597, |
| "learning_rate": 8.405804476138239e-06, |
| "loss": 0.3764, |
| "step": 455 |
| }, |
| { |
| "epoch": 1.008849557522124, |
| "grad_norm": 0.1830655187368393, |
| "learning_rate": 8.396366694486466e-06, |
| "loss": 0.3826, |
| "step": 456 |
| }, |
| { |
| "epoch": 1.011061946902655, |
| "grad_norm": 0.17687253654003143, |
| "learning_rate": 8.386906391484819e-06, |
| "loss": 0.3913, |
| "step": 457 |
| }, |
| { |
| "epoch": 1.0132743362831858, |
| "grad_norm": 0.21283870935440063, |
| "learning_rate": 8.377423629864686e-06, |
| "loss": 0.3771, |
| "step": 458 |
| }, |
| { |
| "epoch": 1.0154867256637168, |
| "grad_norm": 0.1914159059524536, |
| "learning_rate": 8.367918472506375e-06, |
| "loss": 0.3738, |
| "step": 459 |
| }, |
| { |
| "epoch": 1.0176991150442478, |
| "grad_norm": 0.17665503919124603, |
| "learning_rate": 8.358390982438706e-06, |
| "loss": 0.3784, |
| "step": 460 |
| }, |
| { |
| "epoch": 1.0199115044247788, |
| "grad_norm": 0.1868385225534439, |
| "learning_rate": 8.348841222838579e-06, |
| "loss": 0.3876, |
| "step": 461 |
| }, |
| { |
| "epoch": 1.0221238938053097, |
| "grad_norm": 0.17088955640792847, |
| "learning_rate": 8.339269257030576e-06, |
| "loss": 0.3828, |
| "step": 462 |
| }, |
| { |
| "epoch": 1.0243362831858407, |
| "grad_norm": 0.174570694565773, |
| "learning_rate": 8.329675148486518e-06, |
| "loss": 0.3789, |
| "step": 463 |
| }, |
| { |
| "epoch": 1.0265486725663717, |
| "grad_norm": 0.16205543279647827, |
| "learning_rate": 8.32005896082506e-06, |
| "loss": 0.3784, |
| "step": 464 |
| }, |
| { |
| "epoch": 1.0287610619469028, |
| "grad_norm": 0.18252155184745789, |
| "learning_rate": 8.310420757811258e-06, |
| "loss": 0.3797, |
| "step": 465 |
| }, |
| { |
| "epoch": 1.0309734513274336, |
| "grad_norm": 0.2198188453912735, |
| "learning_rate": 8.30076060335616e-06, |
| "loss": 0.3797, |
| "step": 466 |
| }, |
| { |
| "epoch": 1.0331858407079646, |
| "grad_norm": 0.15900884568691254, |
| "learning_rate": 8.291078561516368e-06, |
| "loss": 0.3737, |
| "step": 467 |
| }, |
| { |
| "epoch": 1.0353982300884956, |
| "grad_norm": 0.21332651376724243, |
| "learning_rate": 8.281374696493628e-06, |
| "loss": 0.3728, |
| "step": 468 |
| }, |
| { |
| "epoch": 1.0376106194690267, |
| "grad_norm": 0.18220320343971252, |
| "learning_rate": 8.271649072634381e-06, |
| "loss": 0.3669, |
| "step": 469 |
| }, |
| { |
| "epoch": 1.0398230088495575, |
| "grad_norm": 0.16027416288852692, |
| "learning_rate": 8.261901754429367e-06, |
| "loss": 0.3791, |
| "step": 470 |
| }, |
| { |
| "epoch": 1.0420353982300885, |
| "grad_norm": 0.19321398437023163, |
| "learning_rate": 8.25213280651317e-06, |
| "loss": 0.3803, |
| "step": 471 |
| }, |
| { |
| "epoch": 1.0442477876106195, |
| "grad_norm": 0.14267107844352722, |
| "learning_rate": 8.24234229366381e-06, |
| "loss": 0.3769, |
| "step": 472 |
| }, |
| { |
| "epoch": 1.0464601769911503, |
| "grad_norm": 0.16832539439201355, |
| "learning_rate": 8.232530280802296e-06, |
| "loss": 0.3792, |
| "step": 473 |
| }, |
| { |
| "epoch": 1.0486725663716814, |
| "grad_norm": 0.15776118636131287, |
| "learning_rate": 8.222696832992208e-06, |
| "loss": 0.3741, |
| "step": 474 |
| }, |
| { |
| "epoch": 1.0508849557522124, |
| "grad_norm": 0.16222868859767914, |
| "learning_rate": 8.212842015439263e-06, |
| "loss": 0.3758, |
| "step": 475 |
| }, |
| { |
| "epoch": 1.0530973451327434, |
| "grad_norm": 0.188007190823555, |
| "learning_rate": 8.202965893490877e-06, |
| "loss": 0.3845, |
| "step": 476 |
| }, |
| { |
| "epoch": 1.0553097345132743, |
| "grad_norm": 0.17090556025505066, |
| "learning_rate": 8.193068532635737e-06, |
| "loss": 0.3944, |
| "step": 477 |
| }, |
| { |
| "epoch": 1.0575221238938053, |
| "grad_norm": 0.17572391033172607, |
| "learning_rate": 8.18314999850337e-06, |
| "loss": 0.3736, |
| "step": 478 |
| }, |
| { |
| "epoch": 1.0597345132743363, |
| "grad_norm": 0.1866428256034851, |
| "learning_rate": 8.173210356863696e-06, |
| "loss": 0.3803, |
| "step": 479 |
| }, |
| { |
| "epoch": 1.0619469026548674, |
| "grad_norm": 0.15502867102622986, |
| "learning_rate": 8.163249673626603e-06, |
| "loss": 0.375, |
| "step": 480 |
| }, |
| { |
| "epoch": 1.0641592920353982, |
| "grad_norm": 0.1912487894296646, |
| "learning_rate": 8.153268014841507e-06, |
| "loss": 0.3697, |
| "step": 481 |
| }, |
| { |
| "epoch": 1.0663716814159292, |
| "grad_norm": 0.179610013961792, |
| "learning_rate": 8.143265446696909e-06, |
| "loss": 0.3882, |
| "step": 482 |
| }, |
| { |
| "epoch": 1.0685840707964602, |
| "grad_norm": 0.1993560791015625, |
| "learning_rate": 8.133242035519968e-06, |
| "loss": 0.3825, |
| "step": 483 |
| }, |
| { |
| "epoch": 1.0707964601769913, |
| "grad_norm": 0.153557687997818, |
| "learning_rate": 8.123197847776043e-06, |
| "loss": 0.3835, |
| "step": 484 |
| }, |
| { |
| "epoch": 1.073008849557522, |
| "grad_norm": 0.1919250637292862, |
| "learning_rate": 8.113132950068272e-06, |
| "loss": 0.3769, |
| "step": 485 |
| }, |
| { |
| "epoch": 1.075221238938053, |
| "grad_norm": 0.19700811803340912, |
| "learning_rate": 8.103047409137114e-06, |
| "loss": 0.384, |
| "step": 486 |
| }, |
| { |
| "epoch": 1.0774336283185841, |
| "grad_norm": 0.16748955845832825, |
| "learning_rate": 8.09294129185992e-06, |
| "loss": 0.3746, |
| "step": 487 |
| }, |
| { |
| "epoch": 1.079646017699115, |
| "grad_norm": 0.19604726135730743, |
| "learning_rate": 8.082814665250476e-06, |
| "loss": 0.38, |
| "step": 488 |
| }, |
| { |
| "epoch": 1.081858407079646, |
| "grad_norm": 0.16203738749027252, |
| "learning_rate": 8.072667596458573e-06, |
| "loss": 0.3702, |
| "step": 489 |
| }, |
| { |
| "epoch": 1.084070796460177, |
| "grad_norm": 0.1730918437242508, |
| "learning_rate": 8.062500152769547e-06, |
| "loss": 0.3838, |
| "step": 490 |
| }, |
| { |
| "epoch": 1.086283185840708, |
| "grad_norm": 0.1520654857158661, |
| "learning_rate": 8.052312401603848e-06, |
| "loss": 0.3749, |
| "step": 491 |
| }, |
| { |
| "epoch": 1.0884955752212389, |
| "grad_norm": 0.1743493527173996, |
| "learning_rate": 8.042104410516576e-06, |
| "loss": 0.3961, |
| "step": 492 |
| }, |
| { |
| "epoch": 1.0907079646017699, |
| "grad_norm": 0.1602836549282074, |
| "learning_rate": 8.03187624719705e-06, |
| "loss": 0.39, |
| "step": 493 |
| }, |
| { |
| "epoch": 1.092920353982301, |
| "grad_norm": 0.15004900097846985, |
| "learning_rate": 8.021627979468348e-06, |
| "loss": 0.3783, |
| "step": 494 |
| }, |
| { |
| "epoch": 1.095132743362832, |
| "grad_norm": 0.17503239214420319, |
| "learning_rate": 8.01135967528686e-06, |
| "loss": 0.3767, |
| "step": 495 |
| }, |
| { |
| "epoch": 1.0973451327433628, |
| "grad_norm": 0.15622718632221222, |
| "learning_rate": 8.001071402741843e-06, |
| "loss": 0.3758, |
| "step": 496 |
| }, |
| { |
| "epoch": 1.0995575221238938, |
| "grad_norm": 0.16066192090511322, |
| "learning_rate": 7.990763230054953e-06, |
| "loss": 0.3781, |
| "step": 497 |
| }, |
| { |
| "epoch": 1.1017699115044248, |
| "grad_norm": 0.1617273986339569, |
| "learning_rate": 7.980435225579819e-06, |
| "loss": 0.3826, |
| "step": 498 |
| }, |
| { |
| "epoch": 1.1039823008849559, |
| "grad_norm": 0.16359879076480865, |
| "learning_rate": 7.970087457801563e-06, |
| "loss": 0.3831, |
| "step": 499 |
| }, |
| { |
| "epoch": 1.1061946902654867, |
| "grad_norm": 0.1616351306438446, |
| "learning_rate": 7.959719995336364e-06, |
| "loss": 0.3809, |
| "step": 500 |
| }, |
| { |
| "epoch": 1.1084070796460177, |
| "grad_norm": 0.14531832933425903, |
| "learning_rate": 7.949332906930995e-06, |
| "loss": 0.3766, |
| "step": 501 |
| }, |
| { |
| "epoch": 1.1106194690265487, |
| "grad_norm": 0.17033183574676514, |
| "learning_rate": 7.938926261462366e-06, |
| "loss": 0.3779, |
| "step": 502 |
| }, |
| { |
| "epoch": 1.1128318584070795, |
| "grad_norm": 0.16740630567073822, |
| "learning_rate": 7.928500127937075e-06, |
| "loss": 0.3788, |
| "step": 503 |
| }, |
| { |
| "epoch": 1.1150442477876106, |
| "grad_norm": 0.15742437541484833, |
| "learning_rate": 7.918054575490943e-06, |
| "loss": 0.3778, |
| "step": 504 |
| }, |
| { |
| "epoch": 1.1172566371681416, |
| "grad_norm": 0.17109255492687225, |
| "learning_rate": 7.90758967338856e-06, |
| "loss": 0.3772, |
| "step": 505 |
| }, |
| { |
| "epoch": 1.1194690265486726, |
| "grad_norm": 0.15209320187568665, |
| "learning_rate": 7.897105491022819e-06, |
| "loss": 0.3812, |
| "step": 506 |
| }, |
| { |
| "epoch": 1.1216814159292035, |
| "grad_norm": 0.16745400428771973, |
| "learning_rate": 7.886602097914466e-06, |
| "loss": 0.3747, |
| "step": 507 |
| }, |
| { |
| "epoch": 1.1238938053097345, |
| "grad_norm": 0.1631808876991272, |
| "learning_rate": 7.876079563711631e-06, |
| "loss": 0.3791, |
| "step": 508 |
| }, |
| { |
| "epoch": 1.1261061946902655, |
| "grad_norm": 0.15150891244411469, |
| "learning_rate": 7.86553795818937e-06, |
| "loss": 0.382, |
| "step": 509 |
| }, |
| { |
| "epoch": 1.1283185840707965, |
| "grad_norm": 0.17548954486846924, |
| "learning_rate": 7.854977351249199e-06, |
| "loss": 0.3692, |
| "step": 510 |
| }, |
| { |
| "epoch": 1.1305309734513274, |
| "grad_norm": 0.15667495131492615, |
| "learning_rate": 7.844397812918637e-06, |
| "loss": 0.3848, |
| "step": 511 |
| }, |
| { |
| "epoch": 1.1327433628318584, |
| "grad_norm": 0.1640884280204773, |
| "learning_rate": 7.833799413350732e-06, |
| "loss": 0.3812, |
| "step": 512 |
| }, |
| { |
| "epoch": 1.1349557522123894, |
| "grad_norm": 0.1685960441827774, |
| "learning_rate": 7.823182222823603e-06, |
| "loss": 0.3851, |
| "step": 513 |
| }, |
| { |
| "epoch": 1.1371681415929205, |
| "grad_norm": 0.16258198022842407, |
| "learning_rate": 7.812546311739976e-06, |
| "loss": 0.376, |
| "step": 514 |
| }, |
| { |
| "epoch": 1.1393805309734513, |
| "grad_norm": 0.18257202208042145, |
| "learning_rate": 7.801891750626706e-06, |
| "loss": 0.3805, |
| "step": 515 |
| }, |
| { |
| "epoch": 1.1415929203539823, |
| "grad_norm": 0.18744589388370514, |
| "learning_rate": 7.791218610134324e-06, |
| "loss": 0.38, |
| "step": 516 |
| }, |
| { |
| "epoch": 1.1438053097345133, |
| "grad_norm": 0.17237654328346252, |
| "learning_rate": 7.780526961036556e-06, |
| "loss": 0.377, |
| "step": 517 |
| }, |
| { |
| "epoch": 1.1460176991150441, |
| "grad_norm": 0.18574164807796478, |
| "learning_rate": 7.769816874229862e-06, |
| "loss": 0.3794, |
| "step": 518 |
| }, |
| { |
| "epoch": 1.1482300884955752, |
| "grad_norm": 0.1659688502550125, |
| "learning_rate": 7.759088420732958e-06, |
| "loss": 0.3836, |
| "step": 519 |
| }, |
| { |
| "epoch": 1.1504424778761062, |
| "grad_norm": 0.18307393789291382, |
| "learning_rate": 7.748341671686355e-06, |
| "loss": 0.3849, |
| "step": 520 |
| }, |
| { |
| "epoch": 1.1526548672566372, |
| "grad_norm": 0.17125365138053894, |
| "learning_rate": 7.737576698351878e-06, |
| "loss": 0.3718, |
| "step": 521 |
| }, |
| { |
| "epoch": 1.154867256637168, |
| "grad_norm": 0.1512458175420761, |
| "learning_rate": 7.726793572112203e-06, |
| "loss": 0.363, |
| "step": 522 |
| }, |
| { |
| "epoch": 1.157079646017699, |
| "grad_norm": 0.16603504121303558, |
| "learning_rate": 7.715992364470371e-06, |
| "loss": 0.3753, |
| "step": 523 |
| }, |
| { |
| "epoch": 1.1592920353982301, |
| "grad_norm": 0.15968966484069824, |
| "learning_rate": 7.705173147049326e-06, |
| "loss": 0.3811, |
| "step": 524 |
| }, |
| { |
| "epoch": 1.1615044247787611, |
| "grad_norm": 0.1410856544971466, |
| "learning_rate": 7.694335991591431e-06, |
| "loss": 0.3915, |
| "step": 525 |
| }, |
| { |
| "epoch": 1.163716814159292, |
| "grad_norm": 0.14456801116466522, |
| "learning_rate": 7.683480969958005e-06, |
| "loss": 0.3813, |
| "step": 526 |
| }, |
| { |
| "epoch": 1.165929203539823, |
| "grad_norm": 0.15530700981616974, |
| "learning_rate": 7.672608154128824e-06, |
| "loss": 0.3871, |
| "step": 527 |
| }, |
| { |
| "epoch": 1.168141592920354, |
| "grad_norm": 0.14297249913215637, |
| "learning_rate": 7.66171761620167e-06, |
| "loss": 0.3807, |
| "step": 528 |
| }, |
| { |
| "epoch": 1.170353982300885, |
| "grad_norm": 0.13215981423854828, |
| "learning_rate": 7.650809428391834e-06, |
| "loss": 0.379, |
| "step": 529 |
| }, |
| { |
| "epoch": 1.1725663716814159, |
| "grad_norm": 0.15017202496528625, |
| "learning_rate": 7.63988366303165e-06, |
| "loss": 0.3731, |
| "step": 530 |
| }, |
| { |
| "epoch": 1.174778761061947, |
| "grad_norm": 0.15315097570419312, |
| "learning_rate": 7.628940392569995e-06, |
| "loss": 0.376, |
| "step": 531 |
| }, |
| { |
| "epoch": 1.176991150442478, |
| "grad_norm": 0.15044665336608887, |
| "learning_rate": 7.61797968957184e-06, |
| "loss": 0.3816, |
| "step": 532 |
| }, |
| { |
| "epoch": 1.1792035398230087, |
| "grad_norm": 0.15934717655181885, |
| "learning_rate": 7.607001626717738e-06, |
| "loss": 0.3741, |
| "step": 533 |
| }, |
| { |
| "epoch": 1.1814159292035398, |
| "grad_norm": 0.14937704801559448, |
| "learning_rate": 7.596006276803365e-06, |
| "loss": 0.3849, |
| "step": 534 |
| }, |
| { |
| "epoch": 1.1836283185840708, |
| "grad_norm": 0.14274275302886963, |
| "learning_rate": 7.58499371273902e-06, |
| "loss": 0.3697, |
| "step": 535 |
| }, |
| { |
| "epoch": 1.1858407079646018, |
| "grad_norm": 0.14679598808288574, |
| "learning_rate": 7.5739640075491546e-06, |
| "loss": 0.3769, |
| "step": 536 |
| }, |
| { |
| "epoch": 1.1880530973451326, |
| "grad_norm": 0.1407579779624939, |
| "learning_rate": 7.562917234371879e-06, |
| "loss": 0.3774, |
| "step": 537 |
| }, |
| { |
| "epoch": 1.1902654867256637, |
| "grad_norm": 0.14973343908786774, |
| "learning_rate": 7.551853466458486e-06, |
| "loss": 0.3842, |
| "step": 538 |
| }, |
| { |
| "epoch": 1.1924778761061947, |
| "grad_norm": 0.17621995508670807, |
| "learning_rate": 7.540772777172958e-06, |
| "loss": 0.3706, |
| "step": 539 |
| }, |
| { |
| "epoch": 1.1946902654867257, |
| "grad_norm": 0.17068910598754883, |
| "learning_rate": 7.529675239991483e-06, |
| "loss": 0.3837, |
| "step": 540 |
| }, |
| { |
| "epoch": 1.1969026548672566, |
| "grad_norm": 0.1598779559135437, |
| "learning_rate": 7.518560928501969e-06, |
| "loss": 0.365, |
| "step": 541 |
| }, |
| { |
| "epoch": 1.1991150442477876, |
| "grad_norm": 0.1503140926361084, |
| "learning_rate": 7.507429916403553e-06, |
| "loss": 0.3714, |
| "step": 542 |
| }, |
| { |
| "epoch": 1.2013274336283186, |
| "grad_norm": 0.16745896637439728, |
| "learning_rate": 7.496282277506115e-06, |
| "loss": 0.3772, |
| "step": 543 |
| }, |
| { |
| "epoch": 1.2035398230088497, |
| "grad_norm": 0.1463339626789093, |
| "learning_rate": 7.48511808572979e-06, |
| "loss": 0.3675, |
| "step": 544 |
| }, |
| { |
| "epoch": 1.2057522123893805, |
| "grad_norm": 0.16788248717784882, |
| "learning_rate": 7.473937415104471e-06, |
| "loss": 0.3768, |
| "step": 545 |
| }, |
| { |
| "epoch": 1.2079646017699115, |
| "grad_norm": 0.1352882832288742, |
| "learning_rate": 7.462740339769323e-06, |
| "loss": 0.3761, |
| "step": 546 |
| }, |
| { |
| "epoch": 1.2101769911504425, |
| "grad_norm": 0.15344858169555664, |
| "learning_rate": 7.451526933972294e-06, |
| "loss": 0.377, |
| "step": 547 |
| }, |
| { |
| "epoch": 1.2123893805309733, |
| "grad_norm": 0.16328397393226624, |
| "learning_rate": 7.440297272069615e-06, |
| "loss": 0.3877, |
| "step": 548 |
| }, |
| { |
| "epoch": 1.2146017699115044, |
| "grad_norm": 0.16266027092933655, |
| "learning_rate": 7.429051428525318e-06, |
| "loss": 0.378, |
| "step": 549 |
| }, |
| { |
| "epoch": 1.2168141592920354, |
| "grad_norm": 0.15220795571804047, |
| "learning_rate": 7.417789477910728e-06, |
| "loss": 0.3779, |
| "step": 550 |
| }, |
| { |
| "epoch": 1.2190265486725664, |
| "grad_norm": 0.15698575973510742, |
| "learning_rate": 7.406511494903982e-06, |
| "loss": 0.3748, |
| "step": 551 |
| }, |
| { |
| "epoch": 1.2212389380530975, |
| "grad_norm": 0.17737598717212677, |
| "learning_rate": 7.395217554289524e-06, |
| "loss": 0.3764, |
| "step": 552 |
| }, |
| { |
| "epoch": 1.2234513274336283, |
| "grad_norm": 0.1444133073091507, |
| "learning_rate": 7.383907730957618e-06, |
| "loss": 0.3822, |
| "step": 553 |
| }, |
| { |
| "epoch": 1.2256637168141593, |
| "grad_norm": 0.16357633471488953, |
| "learning_rate": 7.372582099903841e-06, |
| "loss": 0.3647, |
| "step": 554 |
| }, |
| { |
| "epoch": 1.2278761061946903, |
| "grad_norm": 0.14997902512550354, |
| "learning_rate": 7.361240736228594e-06, |
| "loss": 0.3735, |
| "step": 555 |
| }, |
| { |
| "epoch": 1.2300884955752212, |
| "grad_norm": 0.12847259640693665, |
| "learning_rate": 7.349883715136601e-06, |
| "loss": 0.368, |
| "step": 556 |
| }, |
| { |
| "epoch": 1.2323008849557522, |
| "grad_norm": 0.15516282618045807, |
| "learning_rate": 7.3385111119364105e-06, |
| "loss": 0.3819, |
| "step": 557 |
| }, |
| { |
| "epoch": 1.2345132743362832, |
| "grad_norm": 0.16119252145290375, |
| "learning_rate": 7.327123002039897e-06, |
| "loss": 0.3759, |
| "step": 558 |
| }, |
| { |
| "epoch": 1.2367256637168142, |
| "grad_norm": 0.15966367721557617, |
| "learning_rate": 7.315719460961757e-06, |
| "loss": 0.3806, |
| "step": 559 |
| }, |
| { |
| "epoch": 1.238938053097345, |
| "grad_norm": 0.16308777034282684, |
| "learning_rate": 7.304300564319013e-06, |
| "loss": 0.3883, |
| "step": 560 |
| }, |
| { |
| "epoch": 1.241150442477876, |
| "grad_norm": 0.14293457567691803, |
| "learning_rate": 7.292866387830515e-06, |
| "loss": 0.3748, |
| "step": 561 |
| }, |
| { |
| "epoch": 1.2433628318584071, |
| "grad_norm": 0.17340628802776337, |
| "learning_rate": 7.281417007316427e-06, |
| "loss": 0.3832, |
| "step": 562 |
| }, |
| { |
| "epoch": 1.245575221238938, |
| "grad_norm": 0.15052464604377747, |
| "learning_rate": 7.269952498697734e-06, |
| "loss": 0.3714, |
| "step": 563 |
| }, |
| { |
| "epoch": 1.247787610619469, |
| "grad_norm": 0.18350891768932343, |
| "learning_rate": 7.258472937995736e-06, |
| "loss": 0.3789, |
| "step": 564 |
| }, |
| { |
| "epoch": 1.25, |
| "grad_norm": 0.16274268925189972, |
| "learning_rate": 7.246978401331543e-06, |
| "loss": 0.3766, |
| "step": 565 |
| }, |
| { |
| "epoch": 1.252212389380531, |
| "grad_norm": 0.1731940507888794, |
| "learning_rate": 7.235468964925571e-06, |
| "loss": 0.3794, |
| "step": 566 |
| }, |
| { |
| "epoch": 1.254424778761062, |
| "grad_norm": 0.145197331905365, |
| "learning_rate": 7.223944705097035e-06, |
| "loss": 0.3771, |
| "step": 567 |
| }, |
| { |
| "epoch": 1.2566371681415929, |
| "grad_norm": 0.14675655961036682, |
| "learning_rate": 7.212405698263446e-06, |
| "loss": 0.3639, |
| "step": 568 |
| }, |
| { |
| "epoch": 1.258849557522124, |
| "grad_norm": 0.14033639430999756, |
| "learning_rate": 7.200852020940102e-06, |
| "loss": 0.3658, |
| "step": 569 |
| }, |
| { |
| "epoch": 1.261061946902655, |
| "grad_norm": 0.15826043486595154, |
| "learning_rate": 7.189283749739584e-06, |
| "loss": 0.3772, |
| "step": 570 |
| }, |
| { |
| "epoch": 1.2632743362831858, |
| "grad_norm": 0.14263145625591278, |
| "learning_rate": 7.177700961371239e-06, |
| "loss": 0.3824, |
| "step": 571 |
| }, |
| { |
| "epoch": 1.2654867256637168, |
| "grad_norm": 0.16336825489997864, |
| "learning_rate": 7.1661037326406825e-06, |
| "loss": 0.3687, |
| "step": 572 |
| }, |
| { |
| "epoch": 1.2676991150442478, |
| "grad_norm": 0.148985356092453, |
| "learning_rate": 7.154492140449283e-06, |
| "loss": 0.3687, |
| "step": 573 |
| }, |
| { |
| "epoch": 1.2699115044247788, |
| "grad_norm": 0.14789626002311707, |
| "learning_rate": 7.142866261793651e-06, |
| "loss": 0.3791, |
| "step": 574 |
| }, |
| { |
| "epoch": 1.2721238938053097, |
| "grad_norm": 0.13839952647686005, |
| "learning_rate": 7.1312261737651354e-06, |
| "loss": 0.3785, |
| "step": 575 |
| }, |
| { |
| "epoch": 1.2743362831858407, |
| "grad_norm": 0.15205146372318268, |
| "learning_rate": 7.119571953549305e-06, |
| "loss": 0.3842, |
| "step": 576 |
| }, |
| { |
| "epoch": 1.2765486725663717, |
| "grad_norm": 0.14949627220630646, |
| "learning_rate": 7.107903678425436e-06, |
| "loss": 0.3729, |
| "step": 577 |
| }, |
| { |
| "epoch": 1.2787610619469025, |
| "grad_norm": 0.14976395666599274, |
| "learning_rate": 7.09622142576601e-06, |
| "loss": 0.3747, |
| "step": 578 |
| }, |
| { |
| "epoch": 1.2809734513274336, |
| "grad_norm": 0.15081347525119781, |
| "learning_rate": 7.084525273036187e-06, |
| "loss": 0.3937, |
| "step": 579 |
| }, |
| { |
| "epoch": 1.2831858407079646, |
| "grad_norm": 0.16921454668045044, |
| "learning_rate": 7.072815297793303e-06, |
| "loss": 0.3827, |
| "step": 580 |
| }, |
| { |
| "epoch": 1.2853982300884956, |
| "grad_norm": 0.16270343959331512, |
| "learning_rate": 7.061091577686349e-06, |
| "loss": 0.3797, |
| "step": 581 |
| }, |
| { |
| "epoch": 1.2876106194690267, |
| "grad_norm": 0.15536241233348846, |
| "learning_rate": 7.0493541904554605e-06, |
| "loss": 0.3727, |
| "step": 582 |
| }, |
| { |
| "epoch": 1.2898230088495575, |
| "grad_norm": 0.1545332819223404, |
| "learning_rate": 7.037603213931397e-06, |
| "loss": 0.3826, |
| "step": 583 |
| }, |
| { |
| "epoch": 1.2920353982300885, |
| "grad_norm": 0.1420442909002304, |
| "learning_rate": 7.025838726035032e-06, |
| "loss": 0.3786, |
| "step": 584 |
| }, |
| { |
| "epoch": 1.2942477876106195, |
| "grad_norm": 0.15322910249233246, |
| "learning_rate": 7.014060804776831e-06, |
| "loss": 0.3814, |
| "step": 585 |
| }, |
| { |
| "epoch": 1.2964601769911503, |
| "grad_norm": 0.1383352428674698, |
| "learning_rate": 7.002269528256334e-06, |
| "loss": 0.3838, |
| "step": 586 |
| }, |
| { |
| "epoch": 1.2986725663716814, |
| "grad_norm": 0.15499426424503326, |
| "learning_rate": 6.990464974661644e-06, |
| "loss": 0.3803, |
| "step": 587 |
| }, |
| { |
| "epoch": 1.3008849557522124, |
| "grad_norm": 0.13908816874027252, |
| "learning_rate": 6.978647222268904e-06, |
| "loss": 0.3804, |
| "step": 588 |
| }, |
| { |
| "epoch": 1.3030973451327434, |
| "grad_norm": 0.15479522943496704, |
| "learning_rate": 6.9668163494417775e-06, |
| "loss": 0.3747, |
| "step": 589 |
| }, |
| { |
| "epoch": 1.3053097345132743, |
| "grad_norm": 0.139928936958313, |
| "learning_rate": 6.954972434630928e-06, |
| "loss": 0.3735, |
| "step": 590 |
| }, |
| { |
| "epoch": 1.3075221238938053, |
| "grad_norm": 0.1611955463886261, |
| "learning_rate": 6.943115556373503e-06, |
| "loss": 0.3719, |
| "step": 591 |
| }, |
| { |
| "epoch": 1.3097345132743363, |
| "grad_norm": 0.16825099289417267, |
| "learning_rate": 6.93124579329261e-06, |
| "loss": 0.3768, |
| "step": 592 |
| }, |
| { |
| "epoch": 1.3119469026548671, |
| "grad_norm": 0.15945830941200256, |
| "learning_rate": 6.919363224096797e-06, |
| "loss": 0.3772, |
| "step": 593 |
| }, |
| { |
| "epoch": 1.3141592920353982, |
| "grad_norm": 0.17050500214099884, |
| "learning_rate": 6.907467927579528e-06, |
| "loss": 0.3768, |
| "step": 594 |
| }, |
| { |
| "epoch": 1.3163716814159292, |
| "grad_norm": 0.15621913969516754, |
| "learning_rate": 6.8955599826186606e-06, |
| "loss": 0.3966, |
| "step": 595 |
| }, |
| { |
| "epoch": 1.3185840707964602, |
| "grad_norm": 0.15994904935359955, |
| "learning_rate": 6.883639468175926e-06, |
| "loss": 0.3798, |
| "step": 596 |
| }, |
| { |
| "epoch": 1.3207964601769913, |
| "grad_norm": 0.14877241849899292, |
| "learning_rate": 6.871706463296407e-06, |
| "loss": 0.3649, |
| "step": 597 |
| }, |
| { |
| "epoch": 1.323008849557522, |
| "grad_norm": 0.1415458768606186, |
| "learning_rate": 6.859761047108007e-06, |
| "loss": 0.3721, |
| "step": 598 |
| }, |
| { |
| "epoch": 1.325221238938053, |
| "grad_norm": 0.15687789022922516, |
| "learning_rate": 6.847803298820927e-06, |
| "loss": 0.3731, |
| "step": 599 |
| }, |
| { |
| "epoch": 1.3274336283185841, |
| "grad_norm": 0.14746886491775513, |
| "learning_rate": 6.835833297727148e-06, |
| "loss": 0.3756, |
| "step": 600 |
| }, |
| { |
| "epoch": 1.329646017699115, |
| "grad_norm": 0.16469167172908783, |
| "learning_rate": 6.823851123199894e-06, |
| "loss": 0.3807, |
| "step": 601 |
| }, |
| { |
| "epoch": 1.331858407079646, |
| "grad_norm": 0.14631612598896027, |
| "learning_rate": 6.811856854693114e-06, |
| "loss": 0.3753, |
| "step": 602 |
| }, |
| { |
| "epoch": 1.334070796460177, |
| "grad_norm": 0.13799001276493073, |
| "learning_rate": 6.799850571740955e-06, |
| "loss": 0.3635, |
| "step": 603 |
| }, |
| { |
| "epoch": 1.336283185840708, |
| "grad_norm": 0.14614099264144897, |
| "learning_rate": 6.787832353957225e-06, |
| "loss": 0.3819, |
| "step": 604 |
| }, |
| { |
| "epoch": 1.338495575221239, |
| "grad_norm": 0.14514301717281342, |
| "learning_rate": 6.775802281034876e-06, |
| "loss": 0.3881, |
| "step": 605 |
| }, |
| { |
| "epoch": 1.3407079646017699, |
| "grad_norm": 0.149121955037117, |
| "learning_rate": 6.763760432745475e-06, |
| "loss": 0.3737, |
| "step": 606 |
| }, |
| { |
| "epoch": 1.342920353982301, |
| "grad_norm": 0.14145825803279877, |
| "learning_rate": 6.751706888938665e-06, |
| "loss": 0.3786, |
| "step": 607 |
| }, |
| { |
| "epoch": 1.3451327433628317, |
| "grad_norm": 0.15232518315315247, |
| "learning_rate": 6.739641729541645e-06, |
| "loss": 0.3735, |
| "step": 608 |
| }, |
| { |
| "epoch": 1.3473451327433628, |
| "grad_norm": 0.15066224336624146, |
| "learning_rate": 6.72756503455864e-06, |
| "loss": 0.3678, |
| "step": 609 |
| }, |
| { |
| "epoch": 1.3495575221238938, |
| "grad_norm": 0.14223569631576538, |
| "learning_rate": 6.715476884070362e-06, |
| "loss": 0.3834, |
| "step": 610 |
| }, |
| { |
| "epoch": 1.3517699115044248, |
| "grad_norm": 0.13839849829673767, |
| "learning_rate": 6.703377358233489e-06, |
| "loss": 0.3744, |
| "step": 611 |
| }, |
| { |
| "epoch": 1.3539823008849559, |
| "grad_norm": 0.14612992107868195, |
| "learning_rate": 6.691266537280128e-06, |
| "loss": 0.3845, |
| "step": 612 |
| }, |
| { |
| "epoch": 1.3561946902654867, |
| "grad_norm": 0.14740517735481262, |
| "learning_rate": 6.679144501517283e-06, |
| "loss": 0.3871, |
| "step": 613 |
| }, |
| { |
| "epoch": 1.3584070796460177, |
| "grad_norm": 0.13791871070861816, |
| "learning_rate": 6.667011331326324e-06, |
| "loss": 0.3644, |
| "step": 614 |
| }, |
| { |
| "epoch": 1.3606194690265487, |
| "grad_norm": 0.15612854063510895, |
| "learning_rate": 6.654867107162454e-06, |
| "loss": 0.3771, |
| "step": 615 |
| }, |
| { |
| "epoch": 1.3628318584070795, |
| "grad_norm": 0.14010357856750488, |
| "learning_rate": 6.6427119095541745e-06, |
| "loss": 0.3658, |
| "step": 616 |
| }, |
| { |
| "epoch": 1.3650442477876106, |
| "grad_norm": 0.14571578800678253, |
| "learning_rate": 6.6305458191027525e-06, |
| "loss": 0.3761, |
| "step": 617 |
| }, |
| { |
| "epoch": 1.3672566371681416, |
| "grad_norm": 0.1459769606590271, |
| "learning_rate": 6.618368916481686e-06, |
| "loss": 0.3761, |
| "step": 618 |
| }, |
| { |
| "epoch": 1.3694690265486726, |
| "grad_norm": 0.154235377907753, |
| "learning_rate": 6.606181282436166e-06, |
| "loss": 0.3721, |
| "step": 619 |
| }, |
| { |
| "epoch": 1.3716814159292037, |
| "grad_norm": 0.1487996131181717, |
| "learning_rate": 6.593982997782549e-06, |
| "loss": 0.3735, |
| "step": 620 |
| }, |
| { |
| "epoch": 1.3738938053097345, |
| "grad_norm": 0.1267055720090866, |
| "learning_rate": 6.58177414340781e-06, |
| "loss": 0.3798, |
| "step": 621 |
| }, |
| { |
| "epoch": 1.3761061946902655, |
| "grad_norm": 0.18154090642929077, |
| "learning_rate": 6.569554800269014e-06, |
| "loss": 0.3767, |
| "step": 622 |
| }, |
| { |
| "epoch": 1.3783185840707963, |
| "grad_norm": 0.14018799364566803, |
| "learning_rate": 6.557325049392781e-06, |
| "loss": 0.3635, |
| "step": 623 |
| }, |
| { |
| "epoch": 1.3805309734513274, |
| "grad_norm": 0.18179184198379517, |
| "learning_rate": 6.545084971874738e-06, |
| "loss": 0.3787, |
| "step": 624 |
| }, |
| { |
| "epoch": 1.3827433628318584, |
| "grad_norm": 0.13475830852985382, |
| "learning_rate": 6.532834648878993e-06, |
| "loss": 0.3721, |
| "step": 625 |
| }, |
| { |
| "epoch": 1.3849557522123894, |
| "grad_norm": 0.16122294962406158, |
| "learning_rate": 6.520574161637591e-06, |
| "loss": 0.3839, |
| "step": 626 |
| }, |
| { |
| "epoch": 1.3871681415929205, |
| "grad_norm": 0.13646018505096436, |
| "learning_rate": 6.5083035914499736e-06, |
| "loss": 0.3766, |
| "step": 627 |
| }, |
| { |
| "epoch": 1.3893805309734513, |
| "grad_norm": 0.17333966493606567, |
| "learning_rate": 6.496023019682447e-06, |
| "loss": 0.3701, |
| "step": 628 |
| }, |
| { |
| "epoch": 1.3915929203539823, |
| "grad_norm": 0.14083491265773773, |
| "learning_rate": 6.483732527767633e-06, |
| "loss": 0.3803, |
| "step": 629 |
| }, |
| { |
| "epoch": 1.3938053097345133, |
| "grad_norm": 0.16099579632282257, |
| "learning_rate": 6.4714321972039395e-06, |
| "loss": 0.3743, |
| "step": 630 |
| }, |
| { |
| "epoch": 1.3960176991150441, |
| "grad_norm": 0.1497550904750824, |
| "learning_rate": 6.459122109555011e-06, |
| "loss": 0.3649, |
| "step": 631 |
| }, |
| { |
| "epoch": 1.3982300884955752, |
| "grad_norm": 0.16616369783878326, |
| "learning_rate": 6.4468023464491906e-06, |
| "loss": 0.3748, |
| "step": 632 |
| }, |
| { |
| "epoch": 1.4004424778761062, |
| "grad_norm": 0.15206293761730194, |
| "learning_rate": 6.434472989578983e-06, |
| "loss": 0.3732, |
| "step": 633 |
| }, |
| { |
| "epoch": 1.4026548672566372, |
| "grad_norm": 0.15567412972450256, |
| "learning_rate": 6.422134120700506e-06, |
| "loss": 0.3752, |
| "step": 634 |
| }, |
| { |
| "epoch": 1.4048672566371683, |
| "grad_norm": 0.1447472721338272, |
| "learning_rate": 6.409785821632952e-06, |
| "loss": 0.3775, |
| "step": 635 |
| }, |
| { |
| "epoch": 1.407079646017699, |
| "grad_norm": 0.1411687433719635, |
| "learning_rate": 6.397428174258048e-06, |
| "loss": 0.3827, |
| "step": 636 |
| }, |
| { |
| "epoch": 1.4092920353982301, |
| "grad_norm": 0.17297664284706116, |
| "learning_rate": 6.385061260519507e-06, |
| "loss": 0.3826, |
| "step": 637 |
| }, |
| { |
| "epoch": 1.411504424778761, |
| "grad_norm": 0.1365060657262802, |
| "learning_rate": 6.3726851624224875e-06, |
| "loss": 0.3818, |
| "step": 638 |
| }, |
| { |
| "epoch": 1.413716814159292, |
| "grad_norm": 0.15566328167915344, |
| "learning_rate": 6.360299962033051e-06, |
| "loss": 0.3718, |
| "step": 639 |
| }, |
| { |
| "epoch": 1.415929203539823, |
| "grad_norm": 0.13370470702648163, |
| "learning_rate": 6.347905741477613e-06, |
| "loss": 0.3708, |
| "step": 640 |
| }, |
| { |
| "epoch": 1.418141592920354, |
| "grad_norm": 0.15631158649921417, |
| "learning_rate": 6.335502582942409e-06, |
| "loss": 0.381, |
| "step": 641 |
| }, |
| { |
| "epoch": 1.420353982300885, |
| "grad_norm": 0.15640929341316223, |
| "learning_rate": 6.323090568672935e-06, |
| "loss": 0.3673, |
| "step": 642 |
| }, |
| { |
| "epoch": 1.4225663716814159, |
| "grad_norm": 0.13334697484970093, |
| "learning_rate": 6.310669780973414e-06, |
| "loss": 0.3827, |
| "step": 643 |
| }, |
| { |
| "epoch": 1.424778761061947, |
| "grad_norm": 0.1540810763835907, |
| "learning_rate": 6.298240302206242e-06, |
| "loss": 0.3776, |
| "step": 644 |
| }, |
| { |
| "epoch": 1.426991150442478, |
| "grad_norm": 0.1316145956516266, |
| "learning_rate": 6.285802214791448e-06, |
| "loss": 0.3844, |
| "step": 645 |
| }, |
| { |
| "epoch": 1.4292035398230087, |
| "grad_norm": 0.14876599609851837, |
| "learning_rate": 6.273355601206143e-06, |
| "loss": 0.3691, |
| "step": 646 |
| }, |
| { |
| "epoch": 1.4314159292035398, |
| "grad_norm": 0.1518520712852478, |
| "learning_rate": 6.260900543983982e-06, |
| "loss": 0.378, |
| "step": 647 |
| }, |
| { |
| "epoch": 1.4336283185840708, |
| "grad_norm": 0.1387074887752533, |
| "learning_rate": 6.2484371257146e-06, |
| "loss": 0.3791, |
| "step": 648 |
| }, |
| { |
| "epoch": 1.4358407079646018, |
| "grad_norm": 0.16339018940925598, |
| "learning_rate": 6.235965429043082e-06, |
| "loss": 0.3655, |
| "step": 649 |
| }, |
| { |
| "epoch": 1.4380530973451329, |
| "grad_norm": 0.15789134800434113, |
| "learning_rate": 6.223485536669401e-06, |
| "loss": 0.3773, |
| "step": 650 |
| }, |
| { |
| "epoch": 1.4402654867256637, |
| "grad_norm": 0.145791694521904, |
| "learning_rate": 6.210997531347879e-06, |
| "loss": 0.3766, |
| "step": 651 |
| }, |
| { |
| "epoch": 1.4424778761061947, |
| "grad_norm": 0.16261202096939087, |
| "learning_rate": 6.1985014958866386e-06, |
| "loss": 0.3708, |
| "step": 652 |
| }, |
| { |
| "epoch": 1.4446902654867257, |
| "grad_norm": 0.14985325932502747, |
| "learning_rate": 6.185997513147043e-06, |
| "loss": 0.3868, |
| "step": 653 |
| }, |
| { |
| "epoch": 1.4469026548672566, |
| "grad_norm": 0.1505439281463623, |
| "learning_rate": 6.173485666043159e-06, |
| "loss": 0.3803, |
| "step": 654 |
| }, |
| { |
| "epoch": 1.4491150442477876, |
| "grad_norm": 0.13515354692935944, |
| "learning_rate": 6.160966037541201e-06, |
| "loss": 0.3704, |
| "step": 655 |
| }, |
| { |
| "epoch": 1.4513274336283186, |
| "grad_norm": 0.14188633859157562, |
| "learning_rate": 6.148438710658979e-06, |
| "loss": 0.3761, |
| "step": 656 |
| }, |
| { |
| "epoch": 1.4535398230088497, |
| "grad_norm": 0.15125387907028198, |
| "learning_rate": 6.135903768465356e-06, |
| "loss": 0.3812, |
| "step": 657 |
| }, |
| { |
| "epoch": 1.4557522123893805, |
| "grad_norm": 0.13983094692230225, |
| "learning_rate": 6.123361294079691e-06, |
| "loss": 0.3757, |
| "step": 658 |
| }, |
| { |
| "epoch": 1.4579646017699115, |
| "grad_norm": 0.13178189098834991, |
| "learning_rate": 6.110811370671286e-06, |
| "loss": 0.3757, |
| "step": 659 |
| }, |
| { |
| "epoch": 1.4601769911504425, |
| "grad_norm": 0.13429959118366241, |
| "learning_rate": 6.098254081458839e-06, |
| "loss": 0.378, |
| "step": 660 |
| }, |
| { |
| "epoch": 1.4623893805309733, |
| "grad_norm": 0.14196458458900452, |
| "learning_rate": 6.085689509709893e-06, |
| "loss": 0.3817, |
| "step": 661 |
| }, |
| { |
| "epoch": 1.4646017699115044, |
| "grad_norm": 0.13938362896442413, |
| "learning_rate": 6.07311773874028e-06, |
| "loss": 0.366, |
| "step": 662 |
| }, |
| { |
| "epoch": 1.4668141592920354, |
| "grad_norm": 0.13635776937007904, |
| "learning_rate": 6.060538851913568e-06, |
| "loss": 0.3853, |
| "step": 663 |
| }, |
| { |
| "epoch": 1.4690265486725664, |
| "grad_norm": 0.15285412967205048, |
| "learning_rate": 6.047952932640513e-06, |
| "loss": 0.3793, |
| "step": 664 |
| }, |
| { |
| "epoch": 1.4712389380530975, |
| "grad_norm": 0.14070792496204376, |
| "learning_rate": 6.035360064378504e-06, |
| "loss": 0.3772, |
| "step": 665 |
| }, |
| { |
| "epoch": 1.4734513274336283, |
| "grad_norm": 0.14458097517490387, |
| "learning_rate": 6.022760330631006e-06, |
| "loss": 0.3866, |
| "step": 666 |
| }, |
| { |
| "epoch": 1.4756637168141593, |
| "grad_norm": 0.15231792628765106, |
| "learning_rate": 6.01015381494701e-06, |
| "loss": 0.3714, |
| "step": 667 |
| }, |
| { |
| "epoch": 1.4778761061946903, |
| "grad_norm": 0.16281698644161224, |
| "learning_rate": 5.997540600920479e-06, |
| "loss": 0.3855, |
| "step": 668 |
| }, |
| { |
| "epoch": 1.4800884955752212, |
| "grad_norm": 0.1445644348859787, |
| "learning_rate": 5.984920772189793e-06, |
| "loss": 0.3861, |
| "step": 669 |
| }, |
| { |
| "epoch": 1.4823008849557522, |
| "grad_norm": 0.15494373440742493, |
| "learning_rate": 5.972294412437194e-06, |
| "loss": 0.3752, |
| "step": 670 |
| }, |
| { |
| "epoch": 1.4845132743362832, |
| "grad_norm": 0.1489405632019043, |
| "learning_rate": 5.959661605388229e-06, |
| "loss": 0.3674, |
| "step": 671 |
| }, |
| { |
| "epoch": 1.4867256637168142, |
| "grad_norm": 0.13303735852241516, |
| "learning_rate": 5.947022434811202e-06, |
| "loss": 0.3668, |
| "step": 672 |
| }, |
| { |
| "epoch": 1.488938053097345, |
| "grad_norm": 0.16998596489429474, |
| "learning_rate": 5.934376984516608e-06, |
| "loss": 0.3856, |
| "step": 673 |
| }, |
| { |
| "epoch": 1.491150442477876, |
| "grad_norm": 0.1437830924987793, |
| "learning_rate": 5.921725338356585e-06, |
| "loss": 0.3686, |
| "step": 674 |
| }, |
| { |
| "epoch": 1.4933628318584071, |
| "grad_norm": 0.15047383308410645, |
| "learning_rate": 5.909067580224359e-06, |
| "loss": 0.3714, |
| "step": 675 |
| }, |
| { |
| "epoch": 1.495575221238938, |
| "grad_norm": 0.15299569070339203, |
| "learning_rate": 5.896403794053679e-06, |
| "loss": 0.368, |
| "step": 676 |
| }, |
| { |
| "epoch": 1.497787610619469, |
| "grad_norm": 0.14748515188694, |
| "learning_rate": 5.883734063818272e-06, |
| "loss": 0.3693, |
| "step": 677 |
| }, |
| { |
| "epoch": 1.5, |
| "grad_norm": 0.1613292545080185, |
| "learning_rate": 5.871058473531273e-06, |
| "loss": 0.3722, |
| "step": 678 |
| }, |
| { |
| "epoch": 1.502212389380531, |
| "grad_norm": 0.14528776705265045, |
| "learning_rate": 5.858377107244683e-06, |
| "loss": 0.372, |
| "step": 679 |
| }, |
| { |
| "epoch": 1.504424778761062, |
| "grad_norm": 0.15699909627437592, |
| "learning_rate": 5.845690049048799e-06, |
| "loss": 0.3737, |
| "step": 680 |
| }, |
| { |
| "epoch": 1.5066371681415929, |
| "grad_norm": 0.15869326889514923, |
| "learning_rate": 5.83299738307166e-06, |
| "loss": 0.3801, |
| "step": 681 |
| }, |
| { |
| "epoch": 1.508849557522124, |
| "grad_norm": 0.1448500007390976, |
| "learning_rate": 5.820299193478496e-06, |
| "loss": 0.3838, |
| "step": 682 |
| }, |
| { |
| "epoch": 1.5110619469026547, |
| "grad_norm": 0.16063882410526276, |
| "learning_rate": 5.807595564471157e-06, |
| "loss": 0.3716, |
| "step": 683 |
| }, |
| { |
| "epoch": 1.5132743362831858, |
| "grad_norm": 0.14111746847629547, |
| "learning_rate": 5.794886580287565e-06, |
| "loss": 0.3769, |
| "step": 684 |
| }, |
| { |
| "epoch": 1.5154867256637168, |
| "grad_norm": 0.16199442744255066, |
| "learning_rate": 5.782172325201155e-06, |
| "loss": 0.374, |
| "step": 685 |
| }, |
| { |
| "epoch": 1.5176991150442478, |
| "grad_norm": 0.14267931878566742, |
| "learning_rate": 5.76945288352031e-06, |
| "loss": 0.369, |
| "step": 686 |
| }, |
| { |
| "epoch": 1.5199115044247788, |
| "grad_norm": 0.14546652138233185, |
| "learning_rate": 5.756728339587806e-06, |
| "loss": 0.3732, |
| "step": 687 |
| }, |
| { |
| "epoch": 1.5221238938053099, |
| "grad_norm": 0.17248912155628204, |
| "learning_rate": 5.743998777780252e-06, |
| "loss": 0.3788, |
| "step": 688 |
| }, |
| { |
| "epoch": 1.5243362831858407, |
| "grad_norm": 0.16043299436569214, |
| "learning_rate": 5.731264282507531e-06, |
| "loss": 0.3834, |
| "step": 689 |
| }, |
| { |
| "epoch": 1.5265486725663717, |
| "grad_norm": 0.1646987795829773, |
| "learning_rate": 5.71852493821224e-06, |
| "loss": 0.3778, |
| "step": 690 |
| }, |
| { |
| "epoch": 1.5287610619469025, |
| "grad_norm": 0.1607351303100586, |
| "learning_rate": 5.7057808293691305e-06, |
| "loss": 0.3764, |
| "step": 691 |
| }, |
| { |
| "epoch": 1.5309734513274336, |
| "grad_norm": 0.16537661850452423, |
| "learning_rate": 5.6930320404845475e-06, |
| "loss": 0.3757, |
| "step": 692 |
| }, |
| { |
| "epoch": 1.5331858407079646, |
| "grad_norm": 0.16083936393260956, |
| "learning_rate": 5.680278656095868e-06, |
| "loss": 0.3768, |
| "step": 693 |
| }, |
| { |
| "epoch": 1.5353982300884956, |
| "grad_norm": 0.1676769256591797, |
| "learning_rate": 5.6675207607709426e-06, |
| "loss": 0.3799, |
| "step": 694 |
| }, |
| { |
| "epoch": 1.5376106194690267, |
| "grad_norm": 0.1764930635690689, |
| "learning_rate": 5.654758439107533e-06, |
| "loss": 0.3736, |
| "step": 695 |
| }, |
| { |
| "epoch": 1.5398230088495575, |
| "grad_norm": 0.14429877698421478, |
| "learning_rate": 5.641991775732756e-06, |
| "loss": 0.3689, |
| "step": 696 |
| }, |
| { |
| "epoch": 1.5420353982300885, |
| "grad_norm": 0.17751586437225342, |
| "learning_rate": 5.629220855302513e-06, |
| "loss": 0.3798, |
| "step": 697 |
| }, |
| { |
| "epoch": 1.5442477876106193, |
| "grad_norm": 0.15950946509838104, |
| "learning_rate": 5.6164457625009386e-06, |
| "loss": 0.3776, |
| "step": 698 |
| }, |
| { |
| "epoch": 1.5464601769911503, |
| "grad_norm": 0.1793259084224701, |
| "learning_rate": 5.60366658203983e-06, |
| "loss": 0.3851, |
| "step": 699 |
| }, |
| { |
| "epoch": 1.5486725663716814, |
| "grad_norm": 0.1719927191734314, |
| "learning_rate": 5.590883398658095e-06, |
| "loss": 0.3795, |
| "step": 700 |
| }, |
| { |
| "epoch": 1.5508849557522124, |
| "grad_norm": 0.15601767599582672, |
| "learning_rate": 5.5780962971211795e-06, |
| "loss": 0.3781, |
| "step": 701 |
| }, |
| { |
| "epoch": 1.5530973451327434, |
| "grad_norm": 0.15840476751327515, |
| "learning_rate": 5.565305362220515e-06, |
| "loss": 0.3767, |
| "step": 702 |
| }, |
| { |
| "epoch": 1.5553097345132745, |
| "grad_norm": 0.16765564680099487, |
| "learning_rate": 5.552510678772949e-06, |
| "loss": 0.3825, |
| "step": 703 |
| }, |
| { |
| "epoch": 1.5575221238938053, |
| "grad_norm": 0.15824854373931885, |
| "learning_rate": 5.539712331620186e-06, |
| "loss": 0.3646, |
| "step": 704 |
| }, |
| { |
| "epoch": 1.5597345132743363, |
| "grad_norm": 0.18552853167057037, |
| "learning_rate": 5.526910405628227e-06, |
| "loss": 0.3633, |
| "step": 705 |
| }, |
| { |
| "epoch": 1.5619469026548671, |
| "grad_norm": 0.14328765869140625, |
| "learning_rate": 5.514104985686802e-06, |
| "loss": 0.3725, |
| "step": 706 |
| }, |
| { |
| "epoch": 1.5641592920353982, |
| "grad_norm": 0.18817439675331116, |
| "learning_rate": 5.501296156708812e-06, |
| "loss": 0.3777, |
| "step": 707 |
| }, |
| { |
| "epoch": 1.5663716814159292, |
| "grad_norm": 0.13829733431339264, |
| "learning_rate": 5.488484003629759e-06, |
| "loss": 0.3648, |
| "step": 708 |
| }, |
| { |
| "epoch": 1.5685840707964602, |
| "grad_norm": 0.1701672077178955, |
| "learning_rate": 5.475668611407191e-06, |
| "loss": 0.3723, |
| "step": 709 |
| }, |
| { |
| "epoch": 1.5707964601769913, |
| "grad_norm": 0.14328286051750183, |
| "learning_rate": 5.462850065020133e-06, |
| "loss": 0.3709, |
| "step": 710 |
| }, |
| { |
| "epoch": 1.573008849557522, |
| "grad_norm": 0.1651555746793747, |
| "learning_rate": 5.4500284494685275e-06, |
| "loss": 0.3755, |
| "step": 711 |
| }, |
| { |
| "epoch": 1.575221238938053, |
| "grad_norm": 0.16392961144447327, |
| "learning_rate": 5.437203849772664e-06, |
| "loss": 0.3764, |
| "step": 712 |
| }, |
| { |
| "epoch": 1.577433628318584, |
| "grad_norm": 0.14715544879436493, |
| "learning_rate": 5.424376350972625e-06, |
| "loss": 0.3737, |
| "step": 713 |
| }, |
| { |
| "epoch": 1.579646017699115, |
| "grad_norm": 0.16308023035526276, |
| "learning_rate": 5.411546038127715e-06, |
| "loss": 0.3772, |
| "step": 714 |
| }, |
| { |
| "epoch": 1.581858407079646, |
| "grad_norm": 0.18041308224201202, |
| "learning_rate": 5.398712996315898e-06, |
| "loss": 0.3732, |
| "step": 715 |
| }, |
| { |
| "epoch": 1.584070796460177, |
| "grad_norm": 0.1642719805240631, |
| "learning_rate": 5.385877310633233e-06, |
| "loss": 0.3868, |
| "step": 716 |
| }, |
| { |
| "epoch": 1.586283185840708, |
| "grad_norm": 0.17112895846366882, |
| "learning_rate": 5.373039066193312e-06, |
| "loss": 0.3741, |
| "step": 717 |
| }, |
| { |
| "epoch": 1.588495575221239, |
| "grad_norm": 0.1910698413848877, |
| "learning_rate": 5.360198348126696e-06, |
| "loss": 0.3843, |
| "step": 718 |
| }, |
| { |
| "epoch": 1.5907079646017699, |
| "grad_norm": 0.16182135045528412, |
| "learning_rate": 5.347355241580344e-06, |
| "loss": 0.3707, |
| "step": 719 |
| }, |
| { |
| "epoch": 1.592920353982301, |
| "grad_norm": 0.15010400116443634, |
| "learning_rate": 5.334509831717058e-06, |
| "loss": 0.3743, |
| "step": 720 |
| }, |
| { |
| "epoch": 1.5951327433628317, |
| "grad_norm": 0.1402837634086609, |
| "learning_rate": 5.321662203714909e-06, |
| "loss": 0.3779, |
| "step": 721 |
| }, |
| { |
| "epoch": 1.5973451327433628, |
| "grad_norm": 0.14762374758720398, |
| "learning_rate": 5.308812442766679e-06, |
| "loss": 0.3666, |
| "step": 722 |
| }, |
| { |
| "epoch": 1.5995575221238938, |
| "grad_norm": 0.1661965399980545, |
| "learning_rate": 5.295960634079292e-06, |
| "loss": 0.3711, |
| "step": 723 |
| }, |
| { |
| "epoch": 1.6017699115044248, |
| "grad_norm": 0.1451561152935028, |
| "learning_rate": 5.283106862873253e-06, |
| "loss": 0.3804, |
| "step": 724 |
| }, |
| { |
| "epoch": 1.6039823008849559, |
| "grad_norm": 0.16984255611896515, |
| "learning_rate": 5.270251214382078e-06, |
| "loss": 0.3844, |
| "step": 725 |
| }, |
| { |
| "epoch": 1.606194690265487, |
| "grad_norm": 0.14532634615898132, |
| "learning_rate": 5.257393773851733e-06, |
| "loss": 0.3637, |
| "step": 726 |
| }, |
| { |
| "epoch": 1.6084070796460177, |
| "grad_norm": 0.149654358625412, |
| "learning_rate": 5.244534626540067e-06, |
| "loss": 0.3722, |
| "step": 727 |
| }, |
| { |
| "epoch": 1.6106194690265485, |
| "grad_norm": 0.14878515899181366, |
| "learning_rate": 5.231673857716244e-06, |
| "loss": 0.3775, |
| "step": 728 |
| }, |
| { |
| "epoch": 1.6128318584070795, |
| "grad_norm": 0.1508331298828125, |
| "learning_rate": 5.218811552660184e-06, |
| "loss": 0.372, |
| "step": 729 |
| }, |
| { |
| "epoch": 1.6150442477876106, |
| "grad_norm": 0.14183960855007172, |
| "learning_rate": 5.205947796661991e-06, |
| "loss": 0.3736, |
| "step": 730 |
| }, |
| { |
| "epoch": 1.6172566371681416, |
| "grad_norm": 0.14773423969745636, |
| "learning_rate": 5.193082675021393e-06, |
| "loss": 0.3699, |
| "step": 731 |
| }, |
| { |
| "epoch": 1.6194690265486726, |
| "grad_norm": 0.1459917426109314, |
| "learning_rate": 5.1802162730471704e-06, |
| "loss": 0.3742, |
| "step": 732 |
| }, |
| { |
| "epoch": 1.6216814159292037, |
| "grad_norm": 0.14351701736450195, |
| "learning_rate": 5.167348676056596e-06, |
| "loss": 0.375, |
| "step": 733 |
| }, |
| { |
| "epoch": 1.6238938053097345, |
| "grad_norm": 0.14323677122592926, |
| "learning_rate": 5.154479969374865e-06, |
| "loss": 0.381, |
| "step": 734 |
| }, |
| { |
| "epoch": 1.6261061946902655, |
| "grad_norm": 0.14793291687965393, |
| "learning_rate": 5.1416102383345315e-06, |
| "loss": 0.3756, |
| "step": 735 |
| }, |
| { |
| "epoch": 1.6283185840707963, |
| "grad_norm": 0.14177453517913818, |
| "learning_rate": 5.1287395682749444e-06, |
| "loss": 0.3766, |
| "step": 736 |
| }, |
| { |
| "epoch": 1.6305309734513274, |
| "grad_norm": 0.1743677854537964, |
| "learning_rate": 5.115868044541674e-06, |
| "loss": 0.3707, |
| "step": 737 |
| }, |
| { |
| "epoch": 1.6327433628318584, |
| "grad_norm": 0.14689166843891144, |
| "learning_rate": 5.102995752485956e-06, |
| "loss": 0.3672, |
| "step": 738 |
| }, |
| { |
| "epoch": 1.6349557522123894, |
| "grad_norm": 0.14640650153160095, |
| "learning_rate": 5.090122777464121e-06, |
| "loss": 0.3774, |
| "step": 739 |
| }, |
| { |
| "epoch": 1.6371681415929205, |
| "grad_norm": 0.1713770627975464, |
| "learning_rate": 5.077249204837026e-06, |
| "loss": 0.3658, |
| "step": 740 |
| }, |
| { |
| "epoch": 1.6393805309734515, |
| "grad_norm": 0.1488184779882431, |
| "learning_rate": 5.064375119969491e-06, |
| "loss": 0.3702, |
| "step": 741 |
| }, |
| { |
| "epoch": 1.6415929203539823, |
| "grad_norm": 0.14465579390525818, |
| "learning_rate": 5.051500608229734e-06, |
| "loss": 0.3652, |
| "step": 742 |
| }, |
| { |
| "epoch": 1.643805309734513, |
| "grad_norm": 0.1728200763463974, |
| "learning_rate": 5.038625754988802e-06, |
| "loss": 0.3762, |
| "step": 743 |
| }, |
| { |
| "epoch": 1.6460176991150441, |
| "grad_norm": 0.14261655509471893, |
| "learning_rate": 5.025750645620004e-06, |
| "loss": 0.3812, |
| "step": 744 |
| }, |
| { |
| "epoch": 1.6482300884955752, |
| "grad_norm": 0.14125920832157135, |
| "learning_rate": 5.012875365498357e-06, |
| "loss": 0.3794, |
| "step": 745 |
| }, |
| { |
| "epoch": 1.6504424778761062, |
| "grad_norm": 0.15553857386112213, |
| "learning_rate": 5e-06, |
| "loss": 0.3711, |
| "step": 746 |
| }, |
| { |
| "epoch": 1.6526548672566372, |
| "grad_norm": 0.14529219269752502, |
| "learning_rate": 4.9871246345016445e-06, |
| "loss": 0.3743, |
| "step": 747 |
| }, |
| { |
| "epoch": 1.6548672566371683, |
| "grad_norm": 0.13652414083480835, |
| "learning_rate": 4.974249354379997e-06, |
| "loss": 0.3773, |
| "step": 748 |
| }, |
| { |
| "epoch": 1.657079646017699, |
| "grad_norm": 0.1763504445552826, |
| "learning_rate": 4.961374245011201e-06, |
| "loss": 0.3791, |
| "step": 749 |
| }, |
| { |
| "epoch": 1.6592920353982301, |
| "grad_norm": 0.14034205675125122, |
| "learning_rate": 4.948499391770268e-06, |
| "loss": 0.3848, |
| "step": 750 |
| }, |
| { |
| "epoch": 1.661504424778761, |
| "grad_norm": 0.14464423060417175, |
| "learning_rate": 4.93562488003051e-06, |
| "loss": 0.3716, |
| "step": 751 |
| }, |
| { |
| "epoch": 1.663716814159292, |
| "grad_norm": 0.17310793697834015, |
| "learning_rate": 4.922750795162974e-06, |
| "loss": 0.3722, |
| "step": 752 |
| }, |
| { |
| "epoch": 1.665929203539823, |
| "grad_norm": 0.14925870299339294, |
| "learning_rate": 4.909877222535879e-06, |
| "loss": 0.3732, |
| "step": 753 |
| }, |
| { |
| "epoch": 1.668141592920354, |
| "grad_norm": 0.14656174182891846, |
| "learning_rate": 4.897004247514044e-06, |
| "loss": 0.3779, |
| "step": 754 |
| }, |
| { |
| "epoch": 1.670353982300885, |
| "grad_norm": 0.17445078492164612, |
| "learning_rate": 4.884131955458327e-06, |
| "loss": 0.375, |
| "step": 755 |
| }, |
| { |
| "epoch": 1.672566371681416, |
| "grad_norm": 0.17037415504455566, |
| "learning_rate": 4.871260431725058e-06, |
| "loss": 0.3852, |
| "step": 756 |
| }, |
| { |
| "epoch": 1.674778761061947, |
| "grad_norm": 0.15467847883701324, |
| "learning_rate": 4.858389761665469e-06, |
| "loss": 0.3716, |
| "step": 757 |
| }, |
| { |
| "epoch": 1.676991150442478, |
| "grad_norm": 0.1854114532470703, |
| "learning_rate": 4.845520030625136e-06, |
| "loss": 0.379, |
| "step": 758 |
| }, |
| { |
| "epoch": 1.6792035398230087, |
| "grad_norm": 0.14063851535320282, |
| "learning_rate": 4.832651323943406e-06, |
| "loss": 0.3794, |
| "step": 759 |
| }, |
| { |
| "epoch": 1.6814159292035398, |
| "grad_norm": 0.15527218580245972, |
| "learning_rate": 4.819783726952831e-06, |
| "loss": 0.3817, |
| "step": 760 |
| }, |
| { |
| "epoch": 1.6836283185840708, |
| "grad_norm": 0.14957115054130554, |
| "learning_rate": 4.806917324978608e-06, |
| "loss": 0.3724, |
| "step": 761 |
| }, |
| { |
| "epoch": 1.6858407079646018, |
| "grad_norm": 0.13799116015434265, |
| "learning_rate": 4.79405220333801e-06, |
| "loss": 0.3667, |
| "step": 762 |
| }, |
| { |
| "epoch": 1.6880530973451329, |
| "grad_norm": 0.13975948095321655, |
| "learning_rate": 4.781188447339817e-06, |
| "loss": 0.3765, |
| "step": 763 |
| }, |
| { |
| "epoch": 1.6902654867256637, |
| "grad_norm": 0.1460770219564438, |
| "learning_rate": 4.768326142283757e-06, |
| "loss": 0.3709, |
| "step": 764 |
| }, |
| { |
| "epoch": 1.6924778761061947, |
| "grad_norm": 0.1463456004858017, |
| "learning_rate": 4.755465373459934e-06, |
| "loss": 0.383, |
| "step": 765 |
| }, |
| { |
| "epoch": 1.6946902654867255, |
| "grad_norm": 0.14482566714286804, |
| "learning_rate": 4.742606226148268e-06, |
| "loss": 0.3766, |
| "step": 766 |
| }, |
| { |
| "epoch": 1.6969026548672566, |
| "grad_norm": 0.15000230073928833, |
| "learning_rate": 4.7297487856179224e-06, |
| "loss": 0.3657, |
| "step": 767 |
| }, |
| { |
| "epoch": 1.6991150442477876, |
| "grad_norm": 0.15064841508865356, |
| "learning_rate": 4.716893137126748e-06, |
| "loss": 0.369, |
| "step": 768 |
| }, |
| { |
| "epoch": 1.7013274336283186, |
| "grad_norm": 0.14898055791854858, |
| "learning_rate": 4.704039365920709e-06, |
| "loss": 0.3826, |
| "step": 769 |
| }, |
| { |
| "epoch": 1.7035398230088497, |
| "grad_norm": 0.1572868973016739, |
| "learning_rate": 4.691187557233323e-06, |
| "loss": 0.3764, |
| "step": 770 |
| }, |
| { |
| "epoch": 1.7057522123893807, |
| "grad_norm": 0.12325187027454376, |
| "learning_rate": 4.678337796285093e-06, |
| "loss": 0.3803, |
| "step": 771 |
| }, |
| { |
| "epoch": 1.7079646017699115, |
| "grad_norm": 0.1512259542942047, |
| "learning_rate": 4.665490168282943e-06, |
| "loss": 0.377, |
| "step": 772 |
| }, |
| { |
| "epoch": 1.7101769911504425, |
| "grad_norm": 0.1448717713356018, |
| "learning_rate": 4.6526447584196575e-06, |
| "loss": 0.3773, |
| "step": 773 |
| }, |
| { |
| "epoch": 1.7123893805309733, |
| "grad_norm": 0.16558457911014557, |
| "learning_rate": 4.639801651873305e-06, |
| "loss": 0.3772, |
| "step": 774 |
| }, |
| { |
| "epoch": 1.7146017699115044, |
| "grad_norm": 0.14137940108776093, |
| "learning_rate": 4.6269609338066875e-06, |
| "loss": 0.3747, |
| "step": 775 |
| }, |
| { |
| "epoch": 1.7168141592920354, |
| "grad_norm": 0.14484670758247375, |
| "learning_rate": 4.614122689366769e-06, |
| "loss": 0.3832, |
| "step": 776 |
| }, |
| { |
| "epoch": 1.7190265486725664, |
| "grad_norm": 0.1538202315568924, |
| "learning_rate": 4.601287003684104e-06, |
| "loss": 0.377, |
| "step": 777 |
| }, |
| { |
| "epoch": 1.7212389380530975, |
| "grad_norm": 0.15519271790981293, |
| "learning_rate": 4.588453961872286e-06, |
| "loss": 0.3653, |
| "step": 778 |
| }, |
| { |
| "epoch": 1.7234513274336283, |
| "grad_norm": 0.1556425839662552, |
| "learning_rate": 4.575623649027376e-06, |
| "loss": 0.3729, |
| "step": 779 |
| }, |
| { |
| "epoch": 1.7256637168141593, |
| "grad_norm": 0.15341098606586456, |
| "learning_rate": 4.562796150227337e-06, |
| "loss": 0.3722, |
| "step": 780 |
| }, |
| { |
| "epoch": 1.7278761061946901, |
| "grad_norm": 0.1618175208568573, |
| "learning_rate": 4.549971550531474e-06, |
| "loss": 0.3715, |
| "step": 781 |
| }, |
| { |
| "epoch": 1.7300884955752212, |
| "grad_norm": 0.1501229852437973, |
| "learning_rate": 4.537149934979869e-06, |
| "loss": 0.3617, |
| "step": 782 |
| }, |
| { |
| "epoch": 1.7323008849557522, |
| "grad_norm": 0.1526263803243637, |
| "learning_rate": 4.524331388592812e-06, |
| "loss": 0.3816, |
| "step": 783 |
| }, |
| { |
| "epoch": 1.7345132743362832, |
| "grad_norm": 0.14755792915821075, |
| "learning_rate": 4.511515996370244e-06, |
| "loss": 0.3719, |
| "step": 784 |
| }, |
| { |
| "epoch": 1.7367256637168142, |
| "grad_norm": 0.16713263094425201, |
| "learning_rate": 4.498703843291189e-06, |
| "loss": 0.3833, |
| "step": 785 |
| }, |
| { |
| "epoch": 1.7389380530973453, |
| "grad_norm": 0.14121297001838684, |
| "learning_rate": 4.485895014313198e-06, |
| "loss": 0.3729, |
| "step": 786 |
| }, |
| { |
| "epoch": 1.741150442477876, |
| "grad_norm": 0.1821858286857605, |
| "learning_rate": 4.4730895943717735e-06, |
| "loss": 0.3866, |
| "step": 787 |
| }, |
| { |
| "epoch": 1.7433628318584071, |
| "grad_norm": 0.150221586227417, |
| "learning_rate": 4.460287668379815e-06, |
| "loss": 0.3715, |
| "step": 788 |
| }, |
| { |
| "epoch": 1.745575221238938, |
| "grad_norm": 0.16274835169315338, |
| "learning_rate": 4.447489321227052e-06, |
| "loss": 0.3703, |
| "step": 789 |
| }, |
| { |
| "epoch": 1.747787610619469, |
| "grad_norm": 0.14972063899040222, |
| "learning_rate": 4.434694637779486e-06, |
| "loss": 0.3592, |
| "step": 790 |
| }, |
| { |
| "epoch": 1.75, |
| "grad_norm": 0.15914012491703033, |
| "learning_rate": 4.421903702878822e-06, |
| "loss": 0.3754, |
| "step": 791 |
| }, |
| { |
| "epoch": 1.752212389380531, |
| "grad_norm": 0.1527598351240158, |
| "learning_rate": 4.409116601341908e-06, |
| "loss": 0.3754, |
| "step": 792 |
| }, |
| { |
| "epoch": 1.754424778761062, |
| "grad_norm": 0.17452898621559143, |
| "learning_rate": 4.396333417960172e-06, |
| "loss": 0.3781, |
| "step": 793 |
| }, |
| { |
| "epoch": 1.7566371681415929, |
| "grad_norm": 0.1623452603816986, |
| "learning_rate": 4.383554237499064e-06, |
| "loss": 0.3787, |
| "step": 794 |
| }, |
| { |
| "epoch": 1.758849557522124, |
| "grad_norm": 0.14795322716236115, |
| "learning_rate": 4.37077914469749e-06, |
| "loss": 0.3723, |
| "step": 795 |
| }, |
| { |
| "epoch": 1.7610619469026547, |
| "grad_norm": 0.18765774369239807, |
| "learning_rate": 4.358008224267245e-06, |
| "loss": 0.3643, |
| "step": 796 |
| }, |
| { |
| "epoch": 1.7632743362831858, |
| "grad_norm": 0.1436903327703476, |
| "learning_rate": 4.345241560892467e-06, |
| "loss": 0.3757, |
| "step": 797 |
| }, |
| { |
| "epoch": 1.7654867256637168, |
| "grad_norm": 0.17049679160118103, |
| "learning_rate": 4.332479239229059e-06, |
| "loss": 0.3775, |
| "step": 798 |
| }, |
| { |
| "epoch": 1.7676991150442478, |
| "grad_norm": 0.15550793707370758, |
| "learning_rate": 4.319721343904133e-06, |
| "loss": 0.3825, |
| "step": 799 |
| }, |
| { |
| "epoch": 1.7699115044247788, |
| "grad_norm": 0.1573866903781891, |
| "learning_rate": 4.306967959515454e-06, |
| "loss": 0.3696, |
| "step": 800 |
| }, |
| { |
| "epoch": 1.7721238938053099, |
| "grad_norm": 0.16475972533226013, |
| "learning_rate": 4.29421917063087e-06, |
| "loss": 0.3789, |
| "step": 801 |
| }, |
| { |
| "epoch": 1.7743362831858407, |
| "grad_norm": 0.13793496787548065, |
| "learning_rate": 4.2814750617877615e-06, |
| "loss": 0.3718, |
| "step": 802 |
| }, |
| { |
| "epoch": 1.7765486725663717, |
| "grad_norm": 0.1852518618106842, |
| "learning_rate": 4.268735717492472e-06, |
| "loss": 0.3672, |
| "step": 803 |
| }, |
| { |
| "epoch": 1.7787610619469025, |
| "grad_norm": 0.14557519555091858, |
| "learning_rate": 4.256001222219751e-06, |
| "loss": 0.3802, |
| "step": 804 |
| }, |
| { |
| "epoch": 1.7809734513274336, |
| "grad_norm": 0.16771559417247772, |
| "learning_rate": 4.243271660412197e-06, |
| "loss": 0.365, |
| "step": 805 |
| }, |
| { |
| "epoch": 1.7831858407079646, |
| "grad_norm": 0.15564660727977753, |
| "learning_rate": 4.230547116479691e-06, |
| "loss": 0.3758, |
| "step": 806 |
| }, |
| { |
| "epoch": 1.7853982300884956, |
| "grad_norm": 0.18923735618591309, |
| "learning_rate": 4.217827674798845e-06, |
| "loss": 0.3865, |
| "step": 807 |
| }, |
| { |
| "epoch": 1.7876106194690267, |
| "grad_norm": 0.14485545456409454, |
| "learning_rate": 4.2051134197124354e-06, |
| "loss": 0.3771, |
| "step": 808 |
| }, |
| { |
| "epoch": 1.7898230088495575, |
| "grad_norm": 0.17624424397945404, |
| "learning_rate": 4.192404435528844e-06, |
| "loss": 0.3785, |
| "step": 809 |
| }, |
| { |
| "epoch": 1.7920353982300885, |
| "grad_norm": 0.16202057898044586, |
| "learning_rate": 4.179700806521506e-06, |
| "loss": 0.3815, |
| "step": 810 |
| }, |
| { |
| "epoch": 1.7942477876106193, |
| "grad_norm": 0.13739414513111115, |
| "learning_rate": 4.167002616928341e-06, |
| "loss": 0.369, |
| "step": 811 |
| }, |
| { |
| "epoch": 1.7964601769911503, |
| "grad_norm": 0.1619684100151062, |
| "learning_rate": 4.154309950951203e-06, |
| "loss": 0.3843, |
| "step": 812 |
| }, |
| { |
| "epoch": 1.7986725663716814, |
| "grad_norm": 0.14789964258670807, |
| "learning_rate": 4.141622892755318e-06, |
| "loss": 0.3842, |
| "step": 813 |
| }, |
| { |
| "epoch": 1.8008849557522124, |
| "grad_norm": 0.16962085664272308, |
| "learning_rate": 4.128941526468728e-06, |
| "loss": 0.3713, |
| "step": 814 |
| }, |
| { |
| "epoch": 1.8030973451327434, |
| "grad_norm": 0.16202415525913239, |
| "learning_rate": 4.116265936181731e-06, |
| "loss": 0.3629, |
| "step": 815 |
| }, |
| { |
| "epoch": 1.8053097345132745, |
| "grad_norm": 0.1684337556362152, |
| "learning_rate": 4.103596205946323e-06, |
| "loss": 0.383, |
| "step": 816 |
| }, |
| { |
| "epoch": 1.8075221238938053, |
| "grad_norm": 0.14394833147525787, |
| "learning_rate": 4.090932419775642e-06, |
| "loss": 0.3619, |
| "step": 817 |
| }, |
| { |
| "epoch": 1.8097345132743363, |
| "grad_norm": 0.15720577538013458, |
| "learning_rate": 4.078274661643415e-06, |
| "loss": 0.3705, |
| "step": 818 |
| }, |
| { |
| "epoch": 1.8119469026548671, |
| "grad_norm": 0.1556907743215561, |
| "learning_rate": 4.065623015483394e-06, |
| "loss": 0.3794, |
| "step": 819 |
| }, |
| { |
| "epoch": 1.8141592920353982, |
| "grad_norm": 0.16392920911312103, |
| "learning_rate": 4.0529775651888e-06, |
| "loss": 0.3671, |
| "step": 820 |
| }, |
| { |
| "epoch": 1.8163716814159292, |
| "grad_norm": 0.13912513852119446, |
| "learning_rate": 4.040338394611772e-06, |
| "loss": 0.3662, |
| "step": 821 |
| }, |
| { |
| "epoch": 1.8185840707964602, |
| "grad_norm": 0.16647394001483917, |
| "learning_rate": 4.027705587562808e-06, |
| "loss": 0.3716, |
| "step": 822 |
| }, |
| { |
| "epoch": 1.8207964601769913, |
| "grad_norm": 0.15679362416267395, |
| "learning_rate": 4.015079227810208e-06, |
| "loss": 0.3732, |
| "step": 823 |
| }, |
| { |
| "epoch": 1.823008849557522, |
| "grad_norm": 0.16492421925067902, |
| "learning_rate": 4.002459399079523e-06, |
| "loss": 0.3749, |
| "step": 824 |
| }, |
| { |
| "epoch": 1.825221238938053, |
| "grad_norm": 0.1708700805902481, |
| "learning_rate": 3.9898461850529925e-06, |
| "loss": 0.3775, |
| "step": 825 |
| }, |
| { |
| "epoch": 1.827433628318584, |
| "grad_norm": 0.14519424736499786, |
| "learning_rate": 3.977239669368998e-06, |
| "loss": 0.3708, |
| "step": 826 |
| }, |
| { |
| "epoch": 1.829646017699115, |
| "grad_norm": 0.17130383849143982, |
| "learning_rate": 3.964639935621498e-06, |
| "loss": 0.361, |
| "step": 827 |
| }, |
| { |
| "epoch": 1.831858407079646, |
| "grad_norm": 0.1383930891752243, |
| "learning_rate": 3.952047067359488e-06, |
| "loss": 0.3771, |
| "step": 828 |
| }, |
| { |
| "epoch": 1.834070796460177, |
| "grad_norm": 0.16033463180065155, |
| "learning_rate": 3.939461148086434e-06, |
| "loss": 0.378, |
| "step": 829 |
| }, |
| { |
| "epoch": 1.836283185840708, |
| "grad_norm": 0.1625954806804657, |
| "learning_rate": 3.926882261259723e-06, |
| "loss": 0.3696, |
| "step": 830 |
| }, |
| { |
| "epoch": 1.838495575221239, |
| "grad_norm": 0.14928433299064636, |
| "learning_rate": 3.9143104902901085e-06, |
| "loss": 0.3771, |
| "step": 831 |
| }, |
| { |
| "epoch": 1.8407079646017699, |
| "grad_norm": 0.14306089282035828, |
| "learning_rate": 3.901745918541162e-06, |
| "loss": 0.3773, |
| "step": 832 |
| }, |
| { |
| "epoch": 1.842920353982301, |
| "grad_norm": 0.1791837066411972, |
| "learning_rate": 3.889188629328716e-06, |
| "loss": 0.3767, |
| "step": 833 |
| }, |
| { |
| "epoch": 1.8451327433628317, |
| "grad_norm": 0.14630135893821716, |
| "learning_rate": 3.876638705920312e-06, |
| "loss": 0.3726, |
| "step": 834 |
| }, |
| { |
| "epoch": 1.8473451327433628, |
| "grad_norm": 0.15009824931621552, |
| "learning_rate": 3.864096231534645e-06, |
| "loss": 0.3656, |
| "step": 835 |
| }, |
| { |
| "epoch": 1.8495575221238938, |
| "grad_norm": 0.14151619374752045, |
| "learning_rate": 3.851561289341023e-06, |
| "loss": 0.3777, |
| "step": 836 |
| }, |
| { |
| "epoch": 1.8517699115044248, |
| "grad_norm": 0.15849147737026215, |
| "learning_rate": 3.839033962458802e-06, |
| "loss": 0.3706, |
| "step": 837 |
| }, |
| { |
| "epoch": 1.8539823008849559, |
| "grad_norm": 0.14633609354496002, |
| "learning_rate": 3.826514333956843e-06, |
| "loss": 0.3643, |
| "step": 838 |
| }, |
| { |
| "epoch": 1.856194690265487, |
| "grad_norm": 0.1464887112379074, |
| "learning_rate": 3.8140024868529585e-06, |
| "loss": 0.3688, |
| "step": 839 |
| }, |
| { |
| "epoch": 1.8584070796460177, |
| "grad_norm": 0.16905274987220764, |
| "learning_rate": 3.8014985041133627e-06, |
| "loss": 0.3862, |
| "step": 840 |
| }, |
| { |
| "epoch": 1.8606194690265485, |
| "grad_norm": 0.1511407345533371, |
| "learning_rate": 3.789002468652121e-06, |
| "loss": 0.3659, |
| "step": 841 |
| }, |
| { |
| "epoch": 1.8628318584070795, |
| "grad_norm": 0.13423565030097961, |
| "learning_rate": 3.7765144633306006e-06, |
| "loss": 0.3716, |
| "step": 842 |
| }, |
| { |
| "epoch": 1.8650442477876106, |
| "grad_norm": 0.1653309166431427, |
| "learning_rate": 3.76403457095692e-06, |
| "loss": 0.3645, |
| "step": 843 |
| }, |
| { |
| "epoch": 1.8672566371681416, |
| "grad_norm": 0.1364339143037796, |
| "learning_rate": 3.7515628742854006e-06, |
| "loss": 0.3665, |
| "step": 844 |
| }, |
| { |
| "epoch": 1.8694690265486726, |
| "grad_norm": 0.16295693814754486, |
| "learning_rate": 3.7390994560160187e-06, |
| "loss": 0.3732, |
| "step": 845 |
| }, |
| { |
| "epoch": 1.8716814159292037, |
| "grad_norm": 0.13839292526245117, |
| "learning_rate": 3.726644398793857e-06, |
| "loss": 0.3687, |
| "step": 846 |
| }, |
| { |
| "epoch": 1.8738938053097345, |
| "grad_norm": 0.15138795971870422, |
| "learning_rate": 3.714197785208554e-06, |
| "loss": 0.3795, |
| "step": 847 |
| }, |
| { |
| "epoch": 1.8761061946902655, |
| "grad_norm": 0.14297796785831451, |
| "learning_rate": 3.701759697793761e-06, |
| "loss": 0.3879, |
| "step": 848 |
| }, |
| { |
| "epoch": 1.8783185840707963, |
| "grad_norm": 0.15279804170131683, |
| "learning_rate": 3.689330219026588e-06, |
| "loss": 0.3725, |
| "step": 849 |
| }, |
| { |
| "epoch": 1.8805309734513274, |
| "grad_norm": 0.1471881866455078, |
| "learning_rate": 3.6769094313270647e-06, |
| "loss": 0.3709, |
| "step": 850 |
| }, |
| { |
| "epoch": 1.8827433628318584, |
| "grad_norm": 0.153276726603508, |
| "learning_rate": 3.6644974170575907e-06, |
| "loss": 0.3666, |
| "step": 851 |
| }, |
| { |
| "epoch": 1.8849557522123894, |
| "grad_norm": 0.13241925835609436, |
| "learning_rate": 3.652094258522387e-06, |
| "loss": 0.3726, |
| "step": 852 |
| }, |
| { |
| "epoch": 1.8871681415929205, |
| "grad_norm": 0.14609576761722565, |
| "learning_rate": 3.6397000379669513e-06, |
| "loss": 0.3795, |
| "step": 853 |
| }, |
| { |
| "epoch": 1.8893805309734515, |
| "grad_norm": 0.13605985045433044, |
| "learning_rate": 3.627314837577514e-06, |
| "loss": 0.3688, |
| "step": 854 |
| }, |
| { |
| "epoch": 1.8915929203539823, |
| "grad_norm": 0.14954599738121033, |
| "learning_rate": 3.6149387394804946e-06, |
| "loss": 0.3654, |
| "step": 855 |
| }, |
| { |
| "epoch": 1.893805309734513, |
| "grad_norm": 0.1630832552909851, |
| "learning_rate": 3.6025718257419532e-06, |
| "loss": 0.3837, |
| "step": 856 |
| }, |
| { |
| "epoch": 1.8960176991150441, |
| "grad_norm": 0.15406356751918793, |
| "learning_rate": 3.590214178367049e-06, |
| "loss": 0.3705, |
| "step": 857 |
| }, |
| { |
| "epoch": 1.8982300884955752, |
| "grad_norm": 0.14090083539485931, |
| "learning_rate": 3.5778658792994957e-06, |
| "loss": 0.3754, |
| "step": 858 |
| }, |
| { |
| "epoch": 1.9004424778761062, |
| "grad_norm": 0.1341380476951599, |
| "learning_rate": 3.565527010421019e-06, |
| "loss": 0.3688, |
| "step": 859 |
| }, |
| { |
| "epoch": 1.9026548672566372, |
| "grad_norm": 0.17473238706588745, |
| "learning_rate": 3.5531976535508107e-06, |
| "loss": 0.3743, |
| "step": 860 |
| }, |
| { |
| "epoch": 1.9048672566371683, |
| "grad_norm": 0.14425218105316162, |
| "learning_rate": 3.540877890444989e-06, |
| "loss": 0.3699, |
| "step": 861 |
| }, |
| { |
| "epoch": 1.907079646017699, |
| "grad_norm": 0.1527353972196579, |
| "learning_rate": 3.528567802796061e-06, |
| "loss": 0.3648, |
| "step": 862 |
| }, |
| { |
| "epoch": 1.9092920353982301, |
| "grad_norm": 0.142634317278862, |
| "learning_rate": 3.5162674722323677e-06, |
| "loss": 0.3673, |
| "step": 863 |
| }, |
| { |
| "epoch": 1.911504424778761, |
| "grad_norm": 0.19563040137290955, |
| "learning_rate": 3.5039769803175545e-06, |
| "loss": 0.3703, |
| "step": 864 |
| }, |
| { |
| "epoch": 1.913716814159292, |
| "grad_norm": 0.16823236644268036, |
| "learning_rate": 3.4916964085500277e-06, |
| "loss": 0.3762, |
| "step": 865 |
| }, |
| { |
| "epoch": 1.915929203539823, |
| "grad_norm": 0.15324640274047852, |
| "learning_rate": 3.4794258383624115e-06, |
| "loss": 0.3818, |
| "step": 866 |
| }, |
| { |
| "epoch": 1.918141592920354, |
| "grad_norm": 0.15772469341754913, |
| "learning_rate": 3.4671653511210086e-06, |
| "loss": 0.3761, |
| "step": 867 |
| }, |
| { |
| "epoch": 1.920353982300885, |
| "grad_norm": 0.15935544669628143, |
| "learning_rate": 3.4549150281252635e-06, |
| "loss": 0.3738, |
| "step": 868 |
| }, |
| { |
| "epoch": 1.922566371681416, |
| "grad_norm": 0.15252330899238586, |
| "learning_rate": 3.442674950607221e-06, |
| "loss": 0.3828, |
| "step": 869 |
| }, |
| { |
| "epoch": 1.924778761061947, |
| "grad_norm": 0.1612301915884018, |
| "learning_rate": 3.430445199730987e-06, |
| "loss": 0.3762, |
| "step": 870 |
| }, |
| { |
| "epoch": 1.926991150442478, |
| "grad_norm": 0.16883090138435364, |
| "learning_rate": 3.4182258565921933e-06, |
| "loss": 0.371, |
| "step": 871 |
| }, |
| { |
| "epoch": 1.9292035398230087, |
| "grad_norm": 0.14610210061073303, |
| "learning_rate": 3.406017002217452e-06, |
| "loss": 0.3766, |
| "step": 872 |
| }, |
| { |
| "epoch": 1.9314159292035398, |
| "grad_norm": 0.15019983053207397, |
| "learning_rate": 3.393818717563834e-06, |
| "loss": 0.3646, |
| "step": 873 |
| }, |
| { |
| "epoch": 1.9336283185840708, |
| "grad_norm": 0.1793092042207718, |
| "learning_rate": 3.3816310835183153e-06, |
| "loss": 0.3751, |
| "step": 874 |
| }, |
| { |
| "epoch": 1.9358407079646018, |
| "grad_norm": 0.1681625097990036, |
| "learning_rate": 3.369454180897248e-06, |
| "loss": 0.3729, |
| "step": 875 |
| }, |
| { |
| "epoch": 1.9380530973451329, |
| "grad_norm": 0.13819968700408936, |
| "learning_rate": 3.3572880904458267e-06, |
| "loss": 0.3644, |
| "step": 876 |
| }, |
| { |
| "epoch": 1.9402654867256637, |
| "grad_norm": 0.14391469955444336, |
| "learning_rate": 3.345132892837547e-06, |
| "loss": 0.374, |
| "step": 877 |
| }, |
| { |
| "epoch": 1.9424778761061947, |
| "grad_norm": 0.1573283076286316, |
| "learning_rate": 3.332988668673677e-06, |
| "loss": 0.3648, |
| "step": 878 |
| }, |
| { |
| "epoch": 1.9446902654867255, |
| "grad_norm": 0.1618097424507141, |
| "learning_rate": 3.320855498482718e-06, |
| "loss": 0.37, |
| "step": 879 |
| }, |
| { |
| "epoch": 1.9469026548672566, |
| "grad_norm": 0.14835193753242493, |
| "learning_rate": 3.308733462719873e-06, |
| "loss": 0.3658, |
| "step": 880 |
| }, |
| { |
| "epoch": 1.9491150442477876, |
| "grad_norm": 0.16873405873775482, |
| "learning_rate": 3.2966226417665125e-06, |
| "loss": 0.3727, |
| "step": 881 |
| }, |
| { |
| "epoch": 1.9513274336283186, |
| "grad_norm": 0.17028756439685822, |
| "learning_rate": 3.2845231159296404e-06, |
| "loss": 0.3718, |
| "step": 882 |
| }, |
| { |
| "epoch": 1.9535398230088497, |
| "grad_norm": 0.1492159515619278, |
| "learning_rate": 3.2724349654413612e-06, |
| "loss": 0.3769, |
| "step": 883 |
| }, |
| { |
| "epoch": 1.9557522123893807, |
| "grad_norm": 0.1774129718542099, |
| "learning_rate": 3.2603582704583547e-06, |
| "loss": 0.3734, |
| "step": 884 |
| }, |
| { |
| "epoch": 1.9579646017699115, |
| "grad_norm": 0.16402071714401245, |
| "learning_rate": 3.2482931110613358e-06, |
| "loss": 0.3791, |
| "step": 885 |
| }, |
| { |
| "epoch": 1.9601769911504425, |
| "grad_norm": 0.16409912705421448, |
| "learning_rate": 3.236239567254526e-06, |
| "loss": 0.3689, |
| "step": 886 |
| }, |
| { |
| "epoch": 1.9623893805309733, |
| "grad_norm": 0.15910659730434418, |
| "learning_rate": 3.224197718965124e-06, |
| "loss": 0.367, |
| "step": 887 |
| }, |
| { |
| "epoch": 1.9646017699115044, |
| "grad_norm": 0.16524411737918854, |
| "learning_rate": 3.2121676460427765e-06, |
| "loss": 0.3767, |
| "step": 888 |
| }, |
| { |
| "epoch": 1.9668141592920354, |
| "grad_norm": 0.1652010977268219, |
| "learning_rate": 3.2001494282590466e-06, |
| "loss": 0.3797, |
| "step": 889 |
| }, |
| { |
| "epoch": 1.9690265486725664, |
| "grad_norm": 0.17844292521476746, |
| "learning_rate": 3.188143145306888e-06, |
| "loss": 0.3864, |
| "step": 890 |
| }, |
| { |
| "epoch": 1.9712389380530975, |
| "grad_norm": 0.1592842936515808, |
| "learning_rate": 3.176148876800109e-06, |
| "loss": 0.3792, |
| "step": 891 |
| }, |
| { |
| "epoch": 1.9734513274336283, |
| "grad_norm": 0.1572018712759018, |
| "learning_rate": 3.164166702272855e-06, |
| "loss": 0.3733, |
| "step": 892 |
| }, |
| { |
| "epoch": 1.9756637168141593, |
| "grad_norm": 0.15086941421031952, |
| "learning_rate": 3.1521967011790753e-06, |
| "loss": 0.3731, |
| "step": 893 |
| }, |
| { |
| "epoch": 1.9778761061946901, |
| "grad_norm": 0.15987545251846313, |
| "learning_rate": 3.140238952891994e-06, |
| "loss": 0.3718, |
| "step": 894 |
| }, |
| { |
| "epoch": 1.9800884955752212, |
| "grad_norm": 0.19604964554309845, |
| "learning_rate": 3.1282935367035935e-06, |
| "loss": 0.3693, |
| "step": 895 |
| }, |
| { |
| "epoch": 1.9823008849557522, |
| "grad_norm": 0.14786747097969055, |
| "learning_rate": 3.116360531824074e-06, |
| "loss": 0.3707, |
| "step": 896 |
| }, |
| { |
| "epoch": 1.9845132743362832, |
| "grad_norm": 0.15440818667411804, |
| "learning_rate": 3.1044400173813415e-06, |
| "loss": 0.3782, |
| "step": 897 |
| }, |
| { |
| "epoch": 1.9867256637168142, |
| "grad_norm": 0.1537942886352539, |
| "learning_rate": 3.0925320724204743e-06, |
| "loss": 0.3859, |
| "step": 898 |
| }, |
| { |
| "epoch": 1.9889380530973453, |
| "grad_norm": 0.14648617804050446, |
| "learning_rate": 3.080636775903205e-06, |
| "loss": 0.3701, |
| "step": 899 |
| }, |
| { |
| "epoch": 1.991150442477876, |
| "grad_norm": 0.14667072892189026, |
| "learning_rate": 3.068754206707392e-06, |
| "loss": 0.37, |
| "step": 900 |
| }, |
| { |
| "epoch": 1.9933628318584071, |
| "grad_norm": 0.1474722921848297, |
| "learning_rate": 3.056884443626499e-06, |
| "loss": 0.3815, |
| "step": 901 |
| }, |
| { |
| "epoch": 1.995575221238938, |
| "grad_norm": 0.14442405104637146, |
| "learning_rate": 3.0450275653690743e-06, |
| "loss": 0.376, |
| "step": 902 |
| }, |
| { |
| "epoch": 1.997787610619469, |
| "grad_norm": 0.1478486955165863, |
| "learning_rate": 3.033183650558225e-06, |
| "loss": 0.3692, |
| "step": 903 |
| }, |
| { |
| "epoch": 2.0, |
| "grad_norm": 0.15566988289356232, |
| "learning_rate": 3.021352777731096e-06, |
| "loss": 0.3595, |
| "step": 904 |
| }, |
| { |
| "epoch": 2.002212389380531, |
| "grad_norm": 0.1503528654575348, |
| "learning_rate": 3.0095350253383567e-06, |
| "loss": 0.3539, |
| "step": 905 |
| }, |
| { |
| "epoch": 2.004424778761062, |
| "grad_norm": 0.14453503489494324, |
| "learning_rate": 2.997730471743667e-06, |
| "loss": 0.3616, |
| "step": 906 |
| }, |
| { |
| "epoch": 2.006637168141593, |
| "grad_norm": 0.14768028259277344, |
| "learning_rate": 2.985939195223171e-06, |
| "loss": 0.3451, |
| "step": 907 |
| }, |
| { |
| "epoch": 2.0088495575221237, |
| "grad_norm": 0.17169658839702606, |
| "learning_rate": 2.9741612739649694e-06, |
| "loss": 0.3583, |
| "step": 908 |
| }, |
| { |
| "epoch": 2.0110619469026547, |
| "grad_norm": 0.1610146015882492, |
| "learning_rate": 2.9623967860686035e-06, |
| "loss": 0.3563, |
| "step": 909 |
| }, |
| { |
| "epoch": 2.0132743362831858, |
| "grad_norm": 0.1743859350681305, |
| "learning_rate": 2.9506458095445408e-06, |
| "loss": 0.3538, |
| "step": 910 |
| }, |
| { |
| "epoch": 2.015486725663717, |
| "grad_norm": 0.16596271097660065, |
| "learning_rate": 2.9389084223136523e-06, |
| "loss": 0.3465, |
| "step": 911 |
| }, |
| { |
| "epoch": 2.017699115044248, |
| "grad_norm": 0.16805611550807953, |
| "learning_rate": 2.9271847022066992e-06, |
| "loss": 0.3568, |
| "step": 912 |
| }, |
| { |
| "epoch": 2.019911504424779, |
| "grad_norm": 0.2030407339334488, |
| "learning_rate": 2.915474726963815e-06, |
| "loss": 0.3584, |
| "step": 913 |
| }, |
| { |
| "epoch": 2.02212389380531, |
| "grad_norm": 0.1419757902622223, |
| "learning_rate": 2.903778574233992e-06, |
| "loss": 0.3556, |
| "step": 914 |
| }, |
| { |
| "epoch": 2.024336283185841, |
| "grad_norm": 0.19129696488380432, |
| "learning_rate": 2.892096321574564e-06, |
| "loss": 0.3629, |
| "step": 915 |
| }, |
| { |
| "epoch": 2.0265486725663715, |
| "grad_norm": 0.14490734040737152, |
| "learning_rate": 2.880428046450697e-06, |
| "loss": 0.3446, |
| "step": 916 |
| }, |
| { |
| "epoch": 2.0287610619469025, |
| "grad_norm": 0.19220589101314545, |
| "learning_rate": 2.8687738262348645e-06, |
| "loss": 0.3534, |
| "step": 917 |
| }, |
| { |
| "epoch": 2.0309734513274336, |
| "grad_norm": 0.1297258734703064, |
| "learning_rate": 2.8571337382063503e-06, |
| "loss": 0.3588, |
| "step": 918 |
| }, |
| { |
| "epoch": 2.0331858407079646, |
| "grad_norm": 0.18526272475719452, |
| "learning_rate": 2.845507859550718e-06, |
| "loss": 0.3543, |
| "step": 919 |
| }, |
| { |
| "epoch": 2.0353982300884956, |
| "grad_norm": 0.13665147125720978, |
| "learning_rate": 2.8338962673593196e-06, |
| "loss": 0.3411, |
| "step": 920 |
| }, |
| { |
| "epoch": 2.0376106194690267, |
| "grad_norm": 0.15971770882606506, |
| "learning_rate": 2.822299038628762e-06, |
| "loss": 0.3489, |
| "step": 921 |
| }, |
| { |
| "epoch": 2.0398230088495577, |
| "grad_norm": 0.14711672067642212, |
| "learning_rate": 2.810716250260418e-06, |
| "loss": 0.355, |
| "step": 922 |
| }, |
| { |
| "epoch": 2.0420353982300883, |
| "grad_norm": 0.15557117760181427, |
| "learning_rate": 2.799147979059898e-06, |
| "loss": 0.3549, |
| "step": 923 |
| }, |
| { |
| "epoch": 2.0442477876106193, |
| "grad_norm": 0.1388644129037857, |
| "learning_rate": 2.787594301736556e-06, |
| "loss": 0.3549, |
| "step": 924 |
| }, |
| { |
| "epoch": 2.0464601769911503, |
| "grad_norm": 0.1850692480802536, |
| "learning_rate": 2.7760552949029683e-06, |
| "loss": 0.3578, |
| "step": 925 |
| }, |
| { |
| "epoch": 2.0486725663716814, |
| "grad_norm": 0.15951766073703766, |
| "learning_rate": 2.7645310350744296e-06, |
| "loss": 0.3405, |
| "step": 926 |
| }, |
| { |
| "epoch": 2.0508849557522124, |
| "grad_norm": 0.16866743564605713, |
| "learning_rate": 2.7530215986684584e-06, |
| "loss": 0.3567, |
| "step": 927 |
| }, |
| { |
| "epoch": 2.0530973451327434, |
| "grad_norm": 0.15932750701904297, |
| "learning_rate": 2.741527062004264e-06, |
| "loss": 0.3509, |
| "step": 928 |
| }, |
| { |
| "epoch": 2.0553097345132745, |
| "grad_norm": 0.15171128511428833, |
| "learning_rate": 2.7300475013022666e-06, |
| "loss": 0.362, |
| "step": 929 |
| }, |
| { |
| "epoch": 2.0575221238938055, |
| "grad_norm": 0.15653663873672485, |
| "learning_rate": 2.7185829926835728e-06, |
| "loss": 0.362, |
| "step": 930 |
| }, |
| { |
| "epoch": 2.059734513274336, |
| "grad_norm": 0.14654697477817535, |
| "learning_rate": 2.7071336121694856e-06, |
| "loss": 0.3612, |
| "step": 931 |
| }, |
| { |
| "epoch": 2.061946902654867, |
| "grad_norm": 0.14019131660461426, |
| "learning_rate": 2.695699435680986e-06, |
| "loss": 0.3618, |
| "step": 932 |
| }, |
| { |
| "epoch": 2.064159292035398, |
| "grad_norm": 0.16346430778503418, |
| "learning_rate": 2.6842805390382453e-06, |
| "loss": 0.3476, |
| "step": 933 |
| }, |
| { |
| "epoch": 2.066371681415929, |
| "grad_norm": 0.12734098732471466, |
| "learning_rate": 2.6728769979601044e-06, |
| "loss": 0.3576, |
| "step": 934 |
| }, |
| { |
| "epoch": 2.0685840707964602, |
| "grad_norm": 0.15421155095100403, |
| "learning_rate": 2.661488888063591e-06, |
| "loss": 0.3605, |
| "step": 935 |
| }, |
| { |
| "epoch": 2.0707964601769913, |
| "grad_norm": 0.14653554558753967, |
| "learning_rate": 2.6501162848634023e-06, |
| "loss": 0.3525, |
| "step": 936 |
| }, |
| { |
| "epoch": 2.0730088495575223, |
| "grad_norm": 0.15689359605312347, |
| "learning_rate": 2.6387592637714062e-06, |
| "loss": 0.3531, |
| "step": 937 |
| }, |
| { |
| "epoch": 2.0752212389380533, |
| "grad_norm": 0.1751505434513092, |
| "learning_rate": 2.6274179000961604e-06, |
| "loss": 0.3489, |
| "step": 938 |
| }, |
| { |
| "epoch": 2.077433628318584, |
| "grad_norm": 0.1466556340456009, |
| "learning_rate": 2.6160922690423818e-06, |
| "loss": 0.3594, |
| "step": 939 |
| }, |
| { |
| "epoch": 2.079646017699115, |
| "grad_norm": 0.17935603857040405, |
| "learning_rate": 2.6047824457104766e-06, |
| "loss": 0.3536, |
| "step": 940 |
| }, |
| { |
| "epoch": 2.081858407079646, |
| "grad_norm": 0.13906565308570862, |
| "learning_rate": 2.5934885050960183e-06, |
| "loss": 0.366, |
| "step": 941 |
| }, |
| { |
| "epoch": 2.084070796460177, |
| "grad_norm": 0.16516318917274475, |
| "learning_rate": 2.5822105220892733e-06, |
| "loss": 0.3612, |
| "step": 942 |
| }, |
| { |
| "epoch": 2.086283185840708, |
| "grad_norm": 0.1419246345758438, |
| "learning_rate": 2.5709485714746827e-06, |
| "loss": 0.3484, |
| "step": 943 |
| }, |
| { |
| "epoch": 2.088495575221239, |
| "grad_norm": 0.1455831378698349, |
| "learning_rate": 2.559702727930386e-06, |
| "loss": 0.3614, |
| "step": 944 |
| }, |
| { |
| "epoch": 2.09070796460177, |
| "grad_norm": 0.15348093211650848, |
| "learning_rate": 2.548473066027709e-06, |
| "loss": 0.3595, |
| "step": 945 |
| }, |
| { |
| "epoch": 2.0929203539823007, |
| "grad_norm": 0.14580413699150085, |
| "learning_rate": 2.537259660230679e-06, |
| "loss": 0.3576, |
| "step": 946 |
| }, |
| { |
| "epoch": 2.0951327433628317, |
| "grad_norm": 0.16142421960830688, |
| "learning_rate": 2.5260625848955324e-06, |
| "loss": 0.3595, |
| "step": 947 |
| }, |
| { |
| "epoch": 2.0973451327433628, |
| "grad_norm": 0.1431436985731125, |
| "learning_rate": 2.5148819142702095e-06, |
| "loss": 0.361, |
| "step": 948 |
| }, |
| { |
| "epoch": 2.099557522123894, |
| "grad_norm": 0.18138659000396729, |
| "learning_rate": 2.5037177224938846e-06, |
| "loss": 0.3591, |
| "step": 949 |
| }, |
| { |
| "epoch": 2.101769911504425, |
| "grad_norm": 0.1631578505039215, |
| "learning_rate": 2.492570083596447e-06, |
| "loss": 0.3576, |
| "step": 950 |
| }, |
| { |
| "epoch": 2.103982300884956, |
| "grad_norm": 0.1554289162158966, |
| "learning_rate": 2.4814390714980325e-06, |
| "loss": 0.347, |
| "step": 951 |
| }, |
| { |
| "epoch": 2.106194690265487, |
| "grad_norm": 0.15791471302509308, |
| "learning_rate": 2.4703247600085173e-06, |
| "loss": 0.3632, |
| "step": 952 |
| }, |
| { |
| "epoch": 2.1084070796460175, |
| "grad_norm": 0.1508740782737732, |
| "learning_rate": 2.459227222827043e-06, |
| "loss": 0.3548, |
| "step": 953 |
| }, |
| { |
| "epoch": 2.1106194690265485, |
| "grad_norm": 0.1677217036485672, |
| "learning_rate": 2.4481465335415162e-06, |
| "loss": 0.3544, |
| "step": 954 |
| }, |
| { |
| "epoch": 2.1128318584070795, |
| "grad_norm": 0.13547132909297943, |
| "learning_rate": 2.437082765628122e-06, |
| "loss": 0.3438, |
| "step": 955 |
| }, |
| { |
| "epoch": 2.1150442477876106, |
| "grad_norm": 0.15041135251522064, |
| "learning_rate": 2.426035992450848e-06, |
| "loss": 0.3618, |
| "step": 956 |
| }, |
| { |
| "epoch": 2.1172566371681416, |
| "grad_norm": 0.16150957345962524, |
| "learning_rate": 2.4150062872609812e-06, |
| "loss": 0.3565, |
| "step": 957 |
| }, |
| { |
| "epoch": 2.1194690265486726, |
| "grad_norm": 0.16388878226280212, |
| "learning_rate": 2.4039937231966374e-06, |
| "loss": 0.3576, |
| "step": 958 |
| }, |
| { |
| "epoch": 2.1216814159292037, |
| "grad_norm": 0.1527947336435318, |
| "learning_rate": 2.3929983732822607e-06, |
| "loss": 0.3599, |
| "step": 959 |
| }, |
| { |
| "epoch": 2.1238938053097347, |
| "grad_norm": 0.1690537929534912, |
| "learning_rate": 2.3820203104281616e-06, |
| "loss": 0.3512, |
| "step": 960 |
| }, |
| { |
| "epoch": 2.1261061946902653, |
| "grad_norm": 0.14079472422599792, |
| "learning_rate": 2.3710596074300045e-06, |
| "loss": 0.3641, |
| "step": 961 |
| }, |
| { |
| "epoch": 2.1283185840707963, |
| "grad_norm": 0.14791461825370789, |
| "learning_rate": 2.360116336968353e-06, |
| "loss": 0.3528, |
| "step": 962 |
| }, |
| { |
| "epoch": 2.1305309734513274, |
| "grad_norm": 0.1540174037218094, |
| "learning_rate": 2.3491905716081668e-06, |
| "loss": 0.3546, |
| "step": 963 |
| }, |
| { |
| "epoch": 2.1327433628318584, |
| "grad_norm": 0.14626696705818176, |
| "learning_rate": 2.3382823837983314e-06, |
| "loss": 0.3511, |
| "step": 964 |
| }, |
| { |
| "epoch": 2.1349557522123894, |
| "grad_norm": 0.15072183310985565, |
| "learning_rate": 2.327391845871179e-06, |
| "loss": 0.3621, |
| "step": 965 |
| }, |
| { |
| "epoch": 2.1371681415929205, |
| "grad_norm": 0.14171431958675385, |
| "learning_rate": 2.316519030041998e-06, |
| "loss": 0.3499, |
| "step": 966 |
| }, |
| { |
| "epoch": 2.1393805309734515, |
| "grad_norm": 0.134790301322937, |
| "learning_rate": 2.3056640084085707e-06, |
| "loss": 0.3564, |
| "step": 967 |
| }, |
| { |
| "epoch": 2.1415929203539825, |
| "grad_norm": 0.1500406712293625, |
| "learning_rate": 2.2948268529506768e-06, |
| "loss": 0.3546, |
| "step": 968 |
| }, |
| { |
| "epoch": 2.143805309734513, |
| "grad_norm": 0.1394817978143692, |
| "learning_rate": 2.284007635529632e-06, |
| "loss": 0.3521, |
| "step": 969 |
| }, |
| { |
| "epoch": 2.146017699115044, |
| "grad_norm": 0.16197897493839264, |
| "learning_rate": 2.2732064278877975e-06, |
| "loss": 0.3573, |
| "step": 970 |
| }, |
| { |
| "epoch": 2.148230088495575, |
| "grad_norm": 0.14796555042266846, |
| "learning_rate": 2.2624233016481224e-06, |
| "loss": 0.3676, |
| "step": 971 |
| }, |
| { |
| "epoch": 2.150442477876106, |
| "grad_norm": 0.13100607693195343, |
| "learning_rate": 2.2516583283136474e-06, |
| "loss": 0.3575, |
| "step": 972 |
| }, |
| { |
| "epoch": 2.1526548672566372, |
| "grad_norm": 0.1775142401456833, |
| "learning_rate": 2.2409115792670434e-06, |
| "loss": 0.3584, |
| "step": 973 |
| }, |
| { |
| "epoch": 2.1548672566371683, |
| "grad_norm": 0.17847150564193726, |
| "learning_rate": 2.2301831257701405e-06, |
| "loss": 0.3592, |
| "step": 974 |
| }, |
| { |
| "epoch": 2.1570796460176993, |
| "grad_norm": 0.14856776595115662, |
| "learning_rate": 2.2194730389634444e-06, |
| "loss": 0.3571, |
| "step": 975 |
| }, |
| { |
| "epoch": 2.15929203539823, |
| "grad_norm": 0.17716825008392334, |
| "learning_rate": 2.2087813898656775e-06, |
| "loss": 0.3626, |
| "step": 976 |
| }, |
| { |
| "epoch": 2.161504424778761, |
| "grad_norm": 0.13400061428546906, |
| "learning_rate": 2.1981082493732945e-06, |
| "loss": 0.3591, |
| "step": 977 |
| }, |
| { |
| "epoch": 2.163716814159292, |
| "grad_norm": 0.14946198463439941, |
| "learning_rate": 2.187453688260027e-06, |
| "loss": 0.3581, |
| "step": 978 |
| }, |
| { |
| "epoch": 2.165929203539823, |
| "grad_norm": 0.15970373153686523, |
| "learning_rate": 2.176817777176398e-06, |
| "loss": 0.357, |
| "step": 979 |
| }, |
| { |
| "epoch": 2.168141592920354, |
| "grad_norm": 0.1397649049758911, |
| "learning_rate": 2.166200586649272e-06, |
| "loss": 0.3541, |
| "step": 980 |
| }, |
| { |
| "epoch": 2.170353982300885, |
| "grad_norm": 0.1510685831308365, |
| "learning_rate": 2.1556021870813653e-06, |
| "loss": 0.357, |
| "step": 981 |
| }, |
| { |
| "epoch": 2.172566371681416, |
| "grad_norm": 0.15586382150650024, |
| "learning_rate": 2.1450226487508017e-06, |
| "loss": 0.3425, |
| "step": 982 |
| }, |
| { |
| "epoch": 2.1747787610619467, |
| "grad_norm": 0.14433304965496063, |
| "learning_rate": 2.134462041810632e-06, |
| "loss": 0.3635, |
| "step": 983 |
| }, |
| { |
| "epoch": 2.1769911504424777, |
| "grad_norm": 0.16394202411174774, |
| "learning_rate": 2.12392043628837e-06, |
| "loss": 0.3509, |
| "step": 984 |
| }, |
| { |
| "epoch": 2.1792035398230087, |
| "grad_norm": 0.1602417677640915, |
| "learning_rate": 2.1133979020855357e-06, |
| "loss": 0.3595, |
| "step": 985 |
| }, |
| { |
| "epoch": 2.1814159292035398, |
| "grad_norm": 0.14802055060863495, |
| "learning_rate": 2.102894508977182e-06, |
| "loss": 0.3518, |
| "step": 986 |
| }, |
| { |
| "epoch": 2.183628318584071, |
| "grad_norm": 0.13788118958473206, |
| "learning_rate": 2.0924103266114422e-06, |
| "loss": 0.3537, |
| "step": 987 |
| }, |
| { |
| "epoch": 2.185840707964602, |
| "grad_norm": 0.16340523958206177, |
| "learning_rate": 2.081945424509057e-06, |
| "loss": 0.3588, |
| "step": 988 |
| }, |
| { |
| "epoch": 2.188053097345133, |
| "grad_norm": 0.14665019512176514, |
| "learning_rate": 2.0714998720629264e-06, |
| "loss": 0.3485, |
| "step": 989 |
| }, |
| { |
| "epoch": 2.190265486725664, |
| "grad_norm": 0.14981283247470856, |
| "learning_rate": 2.061073738537635e-06, |
| "loss": 0.3615, |
| "step": 990 |
| }, |
| { |
| "epoch": 2.1924778761061945, |
| "grad_norm": 0.13499760627746582, |
| "learning_rate": 2.0506670930690074e-06, |
| "loss": 0.3551, |
| "step": 991 |
| }, |
| { |
| "epoch": 2.1946902654867255, |
| "grad_norm": 0.15379314124584198, |
| "learning_rate": 2.040280004663637e-06, |
| "loss": 0.3471, |
| "step": 992 |
| }, |
| { |
| "epoch": 2.1969026548672566, |
| "grad_norm": 0.14276005327701569, |
| "learning_rate": 2.0299125421984367e-06, |
| "loss": 0.3565, |
| "step": 993 |
| }, |
| { |
| "epoch": 2.1991150442477876, |
| "grad_norm": 0.15052393078804016, |
| "learning_rate": 2.0195647744201826e-06, |
| "loss": 0.3552, |
| "step": 994 |
| }, |
| { |
| "epoch": 2.2013274336283186, |
| "grad_norm": 0.16427332162857056, |
| "learning_rate": 2.0092367699450466e-06, |
| "loss": 0.3588, |
| "step": 995 |
| }, |
| { |
| "epoch": 2.2035398230088497, |
| "grad_norm": 0.14771218597888947, |
| "learning_rate": 1.9989285972581595e-06, |
| "loss": 0.3583, |
| "step": 996 |
| }, |
| { |
| "epoch": 2.2057522123893807, |
| "grad_norm": 0.13689512014389038, |
| "learning_rate": 1.9886403247131395e-06, |
| "loss": 0.3525, |
| "step": 997 |
| }, |
| { |
| "epoch": 2.2079646017699117, |
| "grad_norm": 0.15630333125591278, |
| "learning_rate": 1.9783720205316535e-06, |
| "loss": 0.3492, |
| "step": 998 |
| }, |
| { |
| "epoch": 2.2101769911504423, |
| "grad_norm": 0.13555830717086792, |
| "learning_rate": 1.9681237528029513e-06, |
| "loss": 0.3368, |
| "step": 999 |
| }, |
| { |
| "epoch": 2.2123893805309733, |
| "grad_norm": 0.16636361181735992, |
| "learning_rate": 1.957895589483426e-06, |
| "loss": 0.3607, |
| "step": 1000 |
| }, |
| { |
| "epoch": 2.2146017699115044, |
| "grad_norm": 0.1358216553926468, |
| "learning_rate": 1.947687598396154e-06, |
| "loss": 0.3587, |
| "step": 1001 |
| }, |
| { |
| "epoch": 2.2168141592920354, |
| "grad_norm": 0.15513230860233307, |
| "learning_rate": 1.9374998472304523e-06, |
| "loss": 0.3537, |
| "step": 1002 |
| }, |
| { |
| "epoch": 2.2190265486725664, |
| "grad_norm": 0.16106431186199188, |
| "learning_rate": 1.927332403541428e-06, |
| "loss": 0.3582, |
| "step": 1003 |
| }, |
| { |
| "epoch": 2.2212389380530975, |
| "grad_norm": 0.16195151209831238, |
| "learning_rate": 1.9171853347495234e-06, |
| "loss": 0.356, |
| "step": 1004 |
| }, |
| { |
| "epoch": 2.2234513274336285, |
| "grad_norm": 0.1492871195077896, |
| "learning_rate": 1.9070587081400815e-06, |
| "loss": 0.3611, |
| "step": 1005 |
| }, |
| { |
| "epoch": 2.225663716814159, |
| "grad_norm": 0.15143686532974243, |
| "learning_rate": 1.896952590862886e-06, |
| "loss": 0.3542, |
| "step": 1006 |
| }, |
| { |
| "epoch": 2.22787610619469, |
| "grad_norm": 0.1420651376247406, |
| "learning_rate": 1.8868670499317298e-06, |
| "loss": 0.3589, |
| "step": 1007 |
| }, |
| { |
| "epoch": 2.230088495575221, |
| "grad_norm": 0.15692518651485443, |
| "learning_rate": 1.8768021522239576e-06, |
| "loss": 0.3606, |
| "step": 1008 |
| }, |
| { |
| "epoch": 2.232300884955752, |
| "grad_norm": 0.14444686472415924, |
| "learning_rate": 1.8667579644800344e-06, |
| "loss": 0.3584, |
| "step": 1009 |
| }, |
| { |
| "epoch": 2.234513274336283, |
| "grad_norm": 0.13391245901584625, |
| "learning_rate": 1.856734553303091e-06, |
| "loss": 0.3522, |
| "step": 1010 |
| }, |
| { |
| "epoch": 2.2367256637168142, |
| "grad_norm": 0.15556755661964417, |
| "learning_rate": 1.8467319851584952e-06, |
| "loss": 0.3591, |
| "step": 1011 |
| }, |
| { |
| "epoch": 2.2389380530973453, |
| "grad_norm": 0.14630340039730072, |
| "learning_rate": 1.8367503263733983e-06, |
| "loss": 0.3556, |
| "step": 1012 |
| }, |
| { |
| "epoch": 2.241150442477876, |
| "grad_norm": 0.13597136735916138, |
| "learning_rate": 1.8267896431363048e-06, |
| "loss": 0.3519, |
| "step": 1013 |
| }, |
| { |
| "epoch": 2.243362831858407, |
| "grad_norm": 0.14077752828598022, |
| "learning_rate": 1.8168500014966316e-06, |
| "loss": 0.349, |
| "step": 1014 |
| }, |
| { |
| "epoch": 2.245575221238938, |
| "grad_norm": 0.15375056862831116, |
| "learning_rate": 1.8069314673642624e-06, |
| "loss": 0.3575, |
| "step": 1015 |
| }, |
| { |
| "epoch": 2.247787610619469, |
| "grad_norm": 0.1405712366104126, |
| "learning_rate": 1.7970341065091246e-06, |
| "loss": 0.3543, |
| "step": 1016 |
| }, |
| { |
| "epoch": 2.25, |
| "grad_norm": 0.14993660151958466, |
| "learning_rate": 1.7871579845607378e-06, |
| "loss": 0.3564, |
| "step": 1017 |
| }, |
| { |
| "epoch": 2.252212389380531, |
| "grad_norm": 0.14628523588180542, |
| "learning_rate": 1.7773031670077934e-06, |
| "loss": 0.3488, |
| "step": 1018 |
| }, |
| { |
| "epoch": 2.254424778761062, |
| "grad_norm": 0.1479613482952118, |
| "learning_rate": 1.7674697191977053e-06, |
| "loss": 0.3619, |
| "step": 1019 |
| }, |
| { |
| "epoch": 2.256637168141593, |
| "grad_norm": 0.14518770575523376, |
| "learning_rate": 1.757657706336192e-06, |
| "loss": 0.3695, |
| "step": 1020 |
| }, |
| { |
| "epoch": 2.258849557522124, |
| "grad_norm": 0.14480751752853394, |
| "learning_rate": 1.7478671934868302e-06, |
| "loss": 0.3556, |
| "step": 1021 |
| }, |
| { |
| "epoch": 2.2610619469026547, |
| "grad_norm": 0.15581965446472168, |
| "learning_rate": 1.7380982455706353e-06, |
| "loss": 0.3559, |
| "step": 1022 |
| }, |
| { |
| "epoch": 2.2632743362831858, |
| "grad_norm": 0.14438295364379883, |
| "learning_rate": 1.72835092736562e-06, |
| "loss": 0.3697, |
| "step": 1023 |
| }, |
| { |
| "epoch": 2.265486725663717, |
| "grad_norm": 0.1663781851530075, |
| "learning_rate": 1.7186253035063738e-06, |
| "loss": 0.3517, |
| "step": 1024 |
| }, |
| { |
| "epoch": 2.267699115044248, |
| "grad_norm": 0.14601710438728333, |
| "learning_rate": 1.7089214384836322e-06, |
| "loss": 0.3467, |
| "step": 1025 |
| }, |
| { |
| "epoch": 2.269911504424779, |
| "grad_norm": 0.15470395982265472, |
| "learning_rate": 1.699239396643841e-06, |
| "loss": 0.3529, |
| "step": 1026 |
| }, |
| { |
| "epoch": 2.27212389380531, |
| "grad_norm": 0.15689963102340698, |
| "learning_rate": 1.6895792421887437e-06, |
| "loss": 0.3575, |
| "step": 1027 |
| }, |
| { |
| "epoch": 2.274336283185841, |
| "grad_norm": 0.1612994223833084, |
| "learning_rate": 1.6799410391749416e-06, |
| "loss": 0.356, |
| "step": 1028 |
| }, |
| { |
| "epoch": 2.2765486725663715, |
| "grad_norm": 0.15020525455474854, |
| "learning_rate": 1.670324851513483e-06, |
| "loss": 0.3403, |
| "step": 1029 |
| }, |
| { |
| "epoch": 2.2787610619469025, |
| "grad_norm": 0.16059263050556183, |
| "learning_rate": 1.6607307429694237e-06, |
| "loss": 0.3604, |
| "step": 1030 |
| }, |
| { |
| "epoch": 2.2809734513274336, |
| "grad_norm": 0.1457761973142624, |
| "learning_rate": 1.6511587771614208e-06, |
| "loss": 0.3573, |
| "step": 1031 |
| }, |
| { |
| "epoch": 2.2831858407079646, |
| "grad_norm": 0.14527295529842377, |
| "learning_rate": 1.6416090175612958e-06, |
| "loss": 0.3563, |
| "step": 1032 |
| }, |
| { |
| "epoch": 2.2853982300884956, |
| "grad_norm": 0.145360067486763, |
| "learning_rate": 1.6320815274936269e-06, |
| "loss": 0.356, |
| "step": 1033 |
| }, |
| { |
| "epoch": 2.2876106194690267, |
| "grad_norm": 0.14848755300045013, |
| "learning_rate": 1.622576370135317e-06, |
| "loss": 0.3522, |
| "step": 1034 |
| }, |
| { |
| "epoch": 2.2898230088495577, |
| "grad_norm": 0.1375618577003479, |
| "learning_rate": 1.613093608515181e-06, |
| "loss": 0.3513, |
| "step": 1035 |
| }, |
| { |
| "epoch": 2.2920353982300883, |
| "grad_norm": 0.13908326625823975, |
| "learning_rate": 1.6036333055135345e-06, |
| "loss": 0.3604, |
| "step": 1036 |
| }, |
| { |
| "epoch": 2.2942477876106193, |
| "grad_norm": 0.15508785843849182, |
| "learning_rate": 1.5941955238617612e-06, |
| "loss": 0.3604, |
| "step": 1037 |
| }, |
| { |
| "epoch": 2.2964601769911503, |
| "grad_norm": 0.14073608815670013, |
| "learning_rate": 1.5847803261419109e-06, |
| "loss": 0.3624, |
| "step": 1038 |
| }, |
| { |
| "epoch": 2.2986725663716814, |
| "grad_norm": 0.1389189213514328, |
| "learning_rate": 1.575387774786274e-06, |
| "loss": 0.3512, |
| "step": 1039 |
| }, |
| { |
| "epoch": 2.3008849557522124, |
| "grad_norm": 0.1345841884613037, |
| "learning_rate": 1.5660179320769792e-06, |
| "loss": 0.3625, |
| "step": 1040 |
| }, |
| { |
| "epoch": 2.3030973451327434, |
| "grad_norm": 0.14056994020938873, |
| "learning_rate": 1.556670860145567e-06, |
| "loss": 0.3559, |
| "step": 1041 |
| }, |
| { |
| "epoch": 2.3053097345132745, |
| "grad_norm": 0.13505183160305023, |
| "learning_rate": 1.5473466209725907e-06, |
| "loss": 0.3525, |
| "step": 1042 |
| }, |
| { |
| "epoch": 2.307522123893805, |
| "grad_norm": 0.13815952837467194, |
| "learning_rate": 1.5380452763871951e-06, |
| "loss": 0.3681, |
| "step": 1043 |
| }, |
| { |
| "epoch": 2.309734513274336, |
| "grad_norm": 0.1454608291387558, |
| "learning_rate": 1.5287668880667107e-06, |
| "loss": 0.3542, |
| "step": 1044 |
| }, |
| { |
| "epoch": 2.311946902654867, |
| "grad_norm": 0.1534843146800995, |
| "learning_rate": 1.5195115175362485e-06, |
| "loss": 0.3575, |
| "step": 1045 |
| }, |
| { |
| "epoch": 2.314159292035398, |
| "grad_norm": 0.16305556893348694, |
| "learning_rate": 1.5102792261682813e-06, |
| "loss": 0.3586, |
| "step": 1046 |
| }, |
| { |
| "epoch": 2.316371681415929, |
| "grad_norm": 0.16194356977939606, |
| "learning_rate": 1.5010700751822555e-06, |
| "loss": 0.351, |
| "step": 1047 |
| }, |
| { |
| "epoch": 2.3185840707964602, |
| "grad_norm": 0.17614226043224335, |
| "learning_rate": 1.4918841256441603e-06, |
| "loss": 0.3603, |
| "step": 1048 |
| }, |
| { |
| "epoch": 2.3207964601769913, |
| "grad_norm": 0.1571769267320633, |
| "learning_rate": 1.4827214384661447e-06, |
| "loss": 0.3493, |
| "step": 1049 |
| }, |
| { |
| "epoch": 2.3230088495575223, |
| "grad_norm": 0.1548829972743988, |
| "learning_rate": 1.473582074406099e-06, |
| "loss": 0.363, |
| "step": 1050 |
| }, |
| { |
| "epoch": 2.3252212389380533, |
| "grad_norm": 0.1630624383687973, |
| "learning_rate": 1.4644660940672628e-06, |
| "loss": 0.348, |
| "step": 1051 |
| }, |
| { |
| "epoch": 2.327433628318584, |
| "grad_norm": 0.18259887397289276, |
| "learning_rate": 1.455373557897814e-06, |
| "loss": 0.3582, |
| "step": 1052 |
| }, |
| { |
| "epoch": 2.329646017699115, |
| "grad_norm": 0.1488334983587265, |
| "learning_rate": 1.4463045261904718e-06, |
| "loss": 0.3559, |
| "step": 1053 |
| }, |
| { |
| "epoch": 2.331858407079646, |
| "grad_norm": 0.16422435641288757, |
| "learning_rate": 1.4372590590821012e-06, |
| "loss": 0.3586, |
| "step": 1054 |
| }, |
| { |
| "epoch": 2.334070796460177, |
| "grad_norm": 0.13306331634521484, |
| "learning_rate": 1.4282372165533042e-06, |
| "loss": 0.3465, |
| "step": 1055 |
| }, |
| { |
| "epoch": 2.336283185840708, |
| "grad_norm": 0.1893102079629898, |
| "learning_rate": 1.4192390584280347e-06, |
| "loss": 0.3616, |
| "step": 1056 |
| }, |
| { |
| "epoch": 2.338495575221239, |
| "grad_norm": 0.15167655050754547, |
| "learning_rate": 1.4102646443731866e-06, |
| "loss": 0.3515, |
| "step": 1057 |
| }, |
| { |
| "epoch": 2.34070796460177, |
| "grad_norm": 0.17918401956558228, |
| "learning_rate": 1.4013140338982168e-06, |
| "loss": 0.3634, |
| "step": 1058 |
| }, |
| { |
| "epoch": 2.3429203539823007, |
| "grad_norm": 0.152436301112175, |
| "learning_rate": 1.392387286354731e-06, |
| "loss": 0.3541, |
| "step": 1059 |
| }, |
| { |
| "epoch": 2.3451327433628317, |
| "grad_norm": 0.20484967529773712, |
| "learning_rate": 1.3834844609361064e-06, |
| "loss": 0.3529, |
| "step": 1060 |
| }, |
| { |
| "epoch": 2.3473451327433628, |
| "grad_norm": 0.15942007303237915, |
| "learning_rate": 1.3746056166770872e-06, |
| "loss": 0.3576, |
| "step": 1061 |
| }, |
| { |
| "epoch": 2.349557522123894, |
| "grad_norm": 0.17121823132038116, |
| "learning_rate": 1.3657508124533992e-06, |
| "loss": 0.3464, |
| "step": 1062 |
| }, |
| { |
| "epoch": 2.351769911504425, |
| "grad_norm": 0.16647647321224213, |
| "learning_rate": 1.3569201069813626e-06, |
| "loss": 0.36, |
| "step": 1063 |
| }, |
| { |
| "epoch": 2.353982300884956, |
| "grad_norm": 0.14579357206821442, |
| "learning_rate": 1.3481135588174926e-06, |
| "loss": 0.3429, |
| "step": 1064 |
| }, |
| { |
| "epoch": 2.356194690265487, |
| "grad_norm": 0.15810784697532654, |
| "learning_rate": 1.3393312263581222e-06, |
| "loss": 0.3518, |
| "step": 1065 |
| }, |
| { |
| "epoch": 2.3584070796460175, |
| "grad_norm": 0.15230241417884827, |
| "learning_rate": 1.330573167839005e-06, |
| "loss": 0.3682, |
| "step": 1066 |
| }, |
| { |
| "epoch": 2.3606194690265485, |
| "grad_norm": 0.15005375444889069, |
| "learning_rate": 1.3218394413349389e-06, |
| "loss": 0.3484, |
| "step": 1067 |
| }, |
| { |
| "epoch": 2.3628318584070795, |
| "grad_norm": 0.13220618665218353, |
| "learning_rate": 1.3131301047593685e-06, |
| "loss": 0.3527, |
| "step": 1068 |
| }, |
| { |
| "epoch": 2.3650442477876106, |
| "grad_norm": 0.15161649882793427, |
| "learning_rate": 1.3044452158640197e-06, |
| "loss": 0.3679, |
| "step": 1069 |
| }, |
| { |
| "epoch": 2.3672566371681416, |
| "grad_norm": 0.14612559974193573, |
| "learning_rate": 1.2957848322384959e-06, |
| "loss": 0.3595, |
| "step": 1070 |
| }, |
| { |
| "epoch": 2.3694690265486726, |
| "grad_norm": 0.14127154648303986, |
| "learning_rate": 1.2871490113099066e-06, |
| "loss": 0.3536, |
| "step": 1071 |
| }, |
| { |
| "epoch": 2.3716814159292037, |
| "grad_norm": 0.13997747004032135, |
| "learning_rate": 1.2785378103424917e-06, |
| "loss": 0.3549, |
| "step": 1072 |
| }, |
| { |
| "epoch": 2.3738938053097347, |
| "grad_norm": 0.13644693791866302, |
| "learning_rate": 1.2699512864372287e-06, |
| "loss": 0.3596, |
| "step": 1073 |
| }, |
| { |
| "epoch": 2.3761061946902653, |
| "grad_norm": 0.14478059113025665, |
| "learning_rate": 1.2613894965314682e-06, |
| "loss": 0.3588, |
| "step": 1074 |
| }, |
| { |
| "epoch": 2.3783185840707963, |
| "grad_norm": 0.15023688971996307, |
| "learning_rate": 1.2528524973985424e-06, |
| "loss": 0.3606, |
| "step": 1075 |
| }, |
| { |
| "epoch": 2.3805309734513274, |
| "grad_norm": 0.14657047390937805, |
| "learning_rate": 1.2443403456474017e-06, |
| "loss": 0.3647, |
| "step": 1076 |
| }, |
| { |
| "epoch": 2.3827433628318584, |
| "grad_norm": 0.1500697135925293, |
| "learning_rate": 1.2358530977222276e-06, |
| "loss": 0.3551, |
| "step": 1077 |
| }, |
| { |
| "epoch": 2.3849557522123894, |
| "grad_norm": 0.14721229672431946, |
| "learning_rate": 1.2273908099020703e-06, |
| "loss": 0.3595, |
| "step": 1078 |
| }, |
| { |
| "epoch": 2.3871681415929205, |
| "grad_norm": 0.15616820752620697, |
| "learning_rate": 1.218953538300462e-06, |
| "loss": 0.3498, |
| "step": 1079 |
| }, |
| { |
| "epoch": 2.3893805309734515, |
| "grad_norm": 0.15405981242656708, |
| "learning_rate": 1.2105413388650577e-06, |
| "loss": 0.3482, |
| "step": 1080 |
| }, |
| { |
| "epoch": 2.3915929203539825, |
| "grad_norm": 0.14757871627807617, |
| "learning_rate": 1.2021542673772584e-06, |
| "loss": 0.3592, |
| "step": 1081 |
| }, |
| { |
| "epoch": 2.393805309734513, |
| "grad_norm": 0.16278573870658875, |
| "learning_rate": 1.193792379451837e-06, |
| "loss": 0.3515, |
| "step": 1082 |
| }, |
| { |
| "epoch": 2.396017699115044, |
| "grad_norm": 0.15779441595077515, |
| "learning_rate": 1.1854557305365783e-06, |
| "loss": 0.351, |
| "step": 1083 |
| }, |
| { |
| "epoch": 2.398230088495575, |
| "grad_norm": 0.16500075161457062, |
| "learning_rate": 1.1771443759119028e-06, |
| "loss": 0.3538, |
| "step": 1084 |
| }, |
| { |
| "epoch": 2.400442477876106, |
| "grad_norm": 0.15716195106506348, |
| "learning_rate": 1.1688583706905099e-06, |
| "loss": 0.3522, |
| "step": 1085 |
| }, |
| { |
| "epoch": 2.4026548672566372, |
| "grad_norm": 0.1632101982831955, |
| "learning_rate": 1.1605977698170001e-06, |
| "loss": 0.3612, |
| "step": 1086 |
| }, |
| { |
| "epoch": 2.4048672566371683, |
| "grad_norm": 0.14858639240264893, |
| "learning_rate": 1.1523626280675237e-06, |
| "loss": 0.3575, |
| "step": 1087 |
| }, |
| { |
| "epoch": 2.4070796460176993, |
| "grad_norm": 0.16127295792102814, |
| "learning_rate": 1.1441530000494055e-06, |
| "loss": 0.3557, |
| "step": 1088 |
| }, |
| { |
| "epoch": 2.40929203539823, |
| "grad_norm": 0.15704701840877533, |
| "learning_rate": 1.135968940200794e-06, |
| "loss": 0.3547, |
| "step": 1089 |
| }, |
| { |
| "epoch": 2.411504424778761, |
| "grad_norm": 0.14434731006622314, |
| "learning_rate": 1.1278105027902898e-06, |
| "loss": 0.3561, |
| "step": 1090 |
| }, |
| { |
| "epoch": 2.413716814159292, |
| "grad_norm": 0.17251616716384888, |
| "learning_rate": 1.1196777419165927e-06, |
| "loss": 0.3646, |
| "step": 1091 |
| }, |
| { |
| "epoch": 2.415929203539823, |
| "grad_norm": 0.17024458944797516, |
| "learning_rate": 1.1115707115081448e-06, |
| "loss": 0.3536, |
| "step": 1092 |
| }, |
| { |
| "epoch": 2.418141592920354, |
| "grad_norm": 0.14864687621593475, |
| "learning_rate": 1.1034894653227618e-06, |
| "loss": 0.3526, |
| "step": 1093 |
| }, |
| { |
| "epoch": 2.420353982300885, |
| "grad_norm": 0.18286971747875214, |
| "learning_rate": 1.095434056947291e-06, |
| "loss": 0.356, |
| "step": 1094 |
| }, |
| { |
| "epoch": 2.422566371681416, |
| "grad_norm": 0.1528891921043396, |
| "learning_rate": 1.0874045397972433e-06, |
| "loss": 0.352, |
| "step": 1095 |
| }, |
| { |
| "epoch": 2.4247787610619467, |
| "grad_norm": 0.15347851812839508, |
| "learning_rate": 1.0794009671164484e-06, |
| "loss": 0.3591, |
| "step": 1096 |
| }, |
| { |
| "epoch": 2.4269911504424777, |
| "grad_norm": 0.16304871439933777, |
| "learning_rate": 1.0714233919766953e-06, |
| "loss": 0.3557, |
| "step": 1097 |
| }, |
| { |
| "epoch": 2.4292035398230087, |
| "grad_norm": 0.13769623637199402, |
| "learning_rate": 1.0634718672773863e-06, |
| "loss": 0.3534, |
| "step": 1098 |
| }, |
| { |
| "epoch": 2.4314159292035398, |
| "grad_norm": 0.1583521068096161, |
| "learning_rate": 1.055546445745178e-06, |
| "loss": 0.3591, |
| "step": 1099 |
| }, |
| { |
| "epoch": 2.433628318584071, |
| "grad_norm": 0.17005863785743713, |
| "learning_rate": 1.0476471799336424e-06, |
| "loss": 0.3532, |
| "step": 1100 |
| }, |
| { |
| "epoch": 2.435840707964602, |
| "grad_norm": 0.1480191946029663, |
| "learning_rate": 1.0397741222229057e-06, |
| "loss": 0.3556, |
| "step": 1101 |
| }, |
| { |
| "epoch": 2.438053097345133, |
| "grad_norm": 0.16332688927650452, |
| "learning_rate": 1.0319273248193145e-06, |
| "loss": 0.3595, |
| "step": 1102 |
| }, |
| { |
| "epoch": 2.440265486725664, |
| "grad_norm": 0.17268969118595123, |
| "learning_rate": 1.0241068397550807e-06, |
| "loss": 0.3522, |
| "step": 1103 |
| }, |
| { |
| "epoch": 2.442477876106195, |
| "grad_norm": 0.15803387761116028, |
| "learning_rate": 1.0163127188879352e-06, |
| "loss": 0.3526, |
| "step": 1104 |
| }, |
| { |
| "epoch": 2.4446902654867255, |
| "grad_norm": 0.15174445509910583, |
| "learning_rate": 1.008545013900794e-06, |
| "loss": 0.3521, |
| "step": 1105 |
| }, |
| { |
| "epoch": 2.4469026548672566, |
| "grad_norm": 0.1590649038553238, |
| "learning_rate": 1.0008037763014033e-06, |
| "loss": 0.3529, |
| "step": 1106 |
| }, |
| { |
| "epoch": 2.4491150442477876, |
| "grad_norm": 0.15736490488052368, |
| "learning_rate": 9.930890574220076e-07, |
| "loss": 0.3549, |
| "step": 1107 |
| }, |
| { |
| "epoch": 2.4513274336283186, |
| "grad_norm": 0.16248001158237457, |
| "learning_rate": 9.85400908419002e-07, |
| "loss": 0.3503, |
| "step": 1108 |
| }, |
| { |
| "epoch": 2.4535398230088497, |
| "grad_norm": 0.16862428188323975, |
| "learning_rate": 9.77739380272601e-07, |
| "loss": 0.3513, |
| "step": 1109 |
| }, |
| { |
| "epoch": 2.4557522123893807, |
| "grad_norm": 0.16417399048805237, |
| "learning_rate": 9.701045237864898e-07, |
| "loss": 0.3505, |
| "step": 1110 |
| }, |
| { |
| "epoch": 2.4579646017699117, |
| "grad_norm": 0.14535874128341675, |
| "learning_rate": 9.624963895874995e-07, |
| "loss": 0.3623, |
| "step": 1111 |
| }, |
| { |
| "epoch": 2.4601769911504423, |
| "grad_norm": 0.1544649451971054, |
| "learning_rate": 9.549150281252633e-07, |
| "loss": 0.3524, |
| "step": 1112 |
| }, |
| { |
| "epoch": 2.4623893805309733, |
| "grad_norm": 0.157157301902771, |
| "learning_rate": 9.473604896718808e-07, |
| "loss": 0.3542, |
| "step": 1113 |
| }, |
| { |
| "epoch": 2.4646017699115044, |
| "grad_norm": 0.16022609174251556, |
| "learning_rate": 9.398328243215937e-07, |
| "loss": 0.3486, |
| "step": 1114 |
| }, |
| { |
| "epoch": 2.4668141592920354, |
| "grad_norm": 0.14643922448158264, |
| "learning_rate": 9.323320819904419e-07, |
| "loss": 0.3588, |
| "step": 1115 |
| }, |
| { |
| "epoch": 2.4690265486725664, |
| "grad_norm": 0.15757547318935394, |
| "learning_rate": 9.248583124159438e-07, |
| "loss": 0.3619, |
| "step": 1116 |
| }, |
| { |
| "epoch": 2.4712389380530975, |
| "grad_norm": 0.14214399456977844, |
| "learning_rate": 9.174115651567561e-07, |
| "loss": 0.3559, |
| "step": 1117 |
| }, |
| { |
| "epoch": 2.4734513274336285, |
| "grad_norm": 0.1586948037147522, |
| "learning_rate": 9.099918895923554e-07, |
| "loss": 0.3504, |
| "step": 1118 |
| }, |
| { |
| "epoch": 2.475663716814159, |
| "grad_norm": 0.14957429468631744, |
| "learning_rate": 9.025993349226997e-07, |
| "loss": 0.3714, |
| "step": 1119 |
| }, |
| { |
| "epoch": 2.47787610619469, |
| "grad_norm": 0.15308675169944763, |
| "learning_rate": 8.952339501679142e-07, |
| "loss": 0.3639, |
| "step": 1120 |
| }, |
| { |
| "epoch": 2.480088495575221, |
| "grad_norm": 0.1536092609167099, |
| "learning_rate": 8.878957841679542e-07, |
| "loss": 0.3543, |
| "step": 1121 |
| }, |
| { |
| "epoch": 2.482300884955752, |
| "grad_norm": 0.14460043609142303, |
| "learning_rate": 8.805848855822918e-07, |
| "loss": 0.3573, |
| "step": 1122 |
| }, |
| { |
| "epoch": 2.484513274336283, |
| "grad_norm": 0.1362440139055252, |
| "learning_rate": 8.733013028895864e-07, |
| "loss": 0.3557, |
| "step": 1123 |
| }, |
| { |
| "epoch": 2.4867256637168142, |
| "grad_norm": 0.15375365316867828, |
| "learning_rate": 8.660450843873647e-07, |
| "loss": 0.3555, |
| "step": 1124 |
| }, |
| { |
| "epoch": 2.4889380530973453, |
| "grad_norm": 0.14579614996910095, |
| "learning_rate": 8.588162781917042e-07, |
| "loss": 0.358, |
| "step": 1125 |
| }, |
| { |
| "epoch": 2.491150442477876, |
| "grad_norm": 0.18101723492145538, |
| "learning_rate": 8.516149322369055e-07, |
| "loss": 0.3595, |
| "step": 1126 |
| }, |
| { |
| "epoch": 2.493362831858407, |
| "grad_norm": 0.16425468027591705, |
| "learning_rate": 8.444410942751863e-07, |
| "loss": 0.3459, |
| "step": 1127 |
| }, |
| { |
| "epoch": 2.495575221238938, |
| "grad_norm": 0.17149604856967926, |
| "learning_rate": 8.372948118763536e-07, |
| "loss": 0.3601, |
| "step": 1128 |
| }, |
| { |
| "epoch": 2.497787610619469, |
| "grad_norm": 0.1419076919555664, |
| "learning_rate": 8.301761324274965e-07, |
| "loss": 0.3441, |
| "step": 1129 |
| }, |
| { |
| "epoch": 2.5, |
| "grad_norm": 0.14477266371250153, |
| "learning_rate": 8.230851031326653e-07, |
| "loss": 0.3539, |
| "step": 1130 |
| }, |
| { |
| "epoch": 2.502212389380531, |
| "grad_norm": 0.1530475616455078, |
| "learning_rate": 8.160217710125661e-07, |
| "loss": 0.3502, |
| "step": 1131 |
| }, |
| { |
| "epoch": 2.504424778761062, |
| "grad_norm": 0.14788125455379486, |
| "learning_rate": 8.089861829042406e-07, |
| "loss": 0.3591, |
| "step": 1132 |
| }, |
| { |
| "epoch": 2.5066371681415927, |
| "grad_norm": 0.15901798009872437, |
| "learning_rate": 8.019783854607593e-07, |
| "loss": 0.3551, |
| "step": 1133 |
| }, |
| { |
| "epoch": 2.508849557522124, |
| "grad_norm": 0.1555824875831604, |
| "learning_rate": 7.949984251509185e-07, |
| "loss": 0.3448, |
| "step": 1134 |
| }, |
| { |
| "epoch": 2.5110619469026547, |
| "grad_norm": 0.15857146680355072, |
| "learning_rate": 7.880463482589196e-07, |
| "loss": 0.3433, |
| "step": 1135 |
| }, |
| { |
| "epoch": 2.5132743362831858, |
| "grad_norm": 0.16051232814788818, |
| "learning_rate": 7.811222008840719e-07, |
| "loss": 0.3563, |
| "step": 1136 |
| }, |
| { |
| "epoch": 2.515486725663717, |
| "grad_norm": 0.13366073369979858, |
| "learning_rate": 7.742260289404819e-07, |
| "loss": 0.3552, |
| "step": 1137 |
| }, |
| { |
| "epoch": 2.517699115044248, |
| "grad_norm": 0.16636747121810913, |
| "learning_rate": 7.673578781567537e-07, |
| "loss": 0.3549, |
| "step": 1138 |
| }, |
| { |
| "epoch": 2.519911504424779, |
| "grad_norm": 0.16012783348560333, |
| "learning_rate": 7.605177940756774e-07, |
| "loss": 0.3505, |
| "step": 1139 |
| }, |
| { |
| "epoch": 2.52212389380531, |
| "grad_norm": 0.1498229205608368, |
| "learning_rate": 7.537058220539395e-07, |
| "loss": 0.3579, |
| "step": 1140 |
| }, |
| { |
| "epoch": 2.524336283185841, |
| "grad_norm": 0.17187750339508057, |
| "learning_rate": 7.469220072618094e-07, |
| "loss": 0.3548, |
| "step": 1141 |
| }, |
| { |
| "epoch": 2.5265486725663715, |
| "grad_norm": 0.15075713396072388, |
| "learning_rate": 7.401663946828469e-07, |
| "loss": 0.3568, |
| "step": 1142 |
| }, |
| { |
| "epoch": 2.5287610619469025, |
| "grad_norm": 0.18497484922409058, |
| "learning_rate": 7.334390291136051e-07, |
| "loss": 0.3579, |
| "step": 1143 |
| }, |
| { |
| "epoch": 2.5309734513274336, |
| "grad_norm": 0.1425379067659378, |
| "learning_rate": 7.267399551633253e-07, |
| "loss": 0.3569, |
| "step": 1144 |
| }, |
| { |
| "epoch": 2.5331858407079646, |
| "grad_norm": 0.17833374440670013, |
| "learning_rate": 7.200692172536555e-07, |
| "loss": 0.3476, |
| "step": 1145 |
| }, |
| { |
| "epoch": 2.5353982300884956, |
| "grad_norm": 0.15675534307956696, |
| "learning_rate": 7.13426859618338e-07, |
| "loss": 0.3683, |
| "step": 1146 |
| }, |
| { |
| "epoch": 2.5376106194690267, |
| "grad_norm": 0.16931159794330597, |
| "learning_rate": 7.068129263029316e-07, |
| "loss": 0.3558, |
| "step": 1147 |
| }, |
| { |
| "epoch": 2.5398230088495577, |
| "grad_norm": 0.1581314355134964, |
| "learning_rate": 7.002274611645083e-07, |
| "loss": 0.36, |
| "step": 1148 |
| }, |
| { |
| "epoch": 2.5420353982300883, |
| "grad_norm": 0.1542506217956543, |
| "learning_rate": 6.936705078713713e-07, |
| "loss": 0.3541, |
| "step": 1149 |
| }, |
| { |
| "epoch": 2.5442477876106193, |
| "grad_norm": 0.16385367512702942, |
| "learning_rate": 6.871421099027586e-07, |
| "loss": 0.3587, |
| "step": 1150 |
| }, |
| { |
| "epoch": 2.5464601769911503, |
| "grad_norm": 0.14884746074676514, |
| "learning_rate": 6.806423105485576e-07, |
| "loss": 0.3542, |
| "step": 1151 |
| }, |
| { |
| "epoch": 2.5486725663716814, |
| "grad_norm": 0.15693628787994385, |
| "learning_rate": 6.741711529090212e-07, |
| "loss": 0.3573, |
| "step": 1152 |
| }, |
| { |
| "epoch": 2.5508849557522124, |
| "grad_norm": 0.13773605227470398, |
| "learning_rate": 6.677286798944743e-07, |
| "loss": 0.3537, |
| "step": 1153 |
| }, |
| { |
| "epoch": 2.5530973451327434, |
| "grad_norm": 0.15046106278896332, |
| "learning_rate": 6.61314934225037e-07, |
| "loss": 0.3533, |
| "step": 1154 |
| }, |
| { |
| "epoch": 2.5553097345132745, |
| "grad_norm": 0.15596453845500946, |
| "learning_rate": 6.549299584303343e-07, |
| "loss": 0.3541, |
| "step": 1155 |
| }, |
| { |
| "epoch": 2.557522123893805, |
| "grad_norm": 0.1392877697944641, |
| "learning_rate": 6.485737948492237e-07, |
| "loss": 0.3671, |
| "step": 1156 |
| }, |
| { |
| "epoch": 2.5597345132743365, |
| "grad_norm": 0.16301456093788147, |
| "learning_rate": 6.422464856295035e-07, |
| "loss": 0.3551, |
| "step": 1157 |
| }, |
| { |
| "epoch": 2.561946902654867, |
| "grad_norm": 0.14623309671878815, |
| "learning_rate": 6.359480727276407e-07, |
| "loss": 0.3566, |
| "step": 1158 |
| }, |
| { |
| "epoch": 2.564159292035398, |
| "grad_norm": 0.15128614008426666, |
| "learning_rate": 6.296785979084891e-07, |
| "loss": 0.3573, |
| "step": 1159 |
| }, |
| { |
| "epoch": 2.566371681415929, |
| "grad_norm": 0.1874575912952423, |
| "learning_rate": 6.234381027450132e-07, |
| "loss": 0.3497, |
| "step": 1160 |
| }, |
| { |
| "epoch": 2.5685840707964602, |
| "grad_norm": 0.15378333628177643, |
| "learning_rate": 6.172266286180162e-07, |
| "loss": 0.3462, |
| "step": 1161 |
| }, |
| { |
| "epoch": 2.5707964601769913, |
| "grad_norm": 0.1589403748512268, |
| "learning_rate": 6.110442167158592e-07, |
| "loss": 0.3496, |
| "step": 1162 |
| }, |
| { |
| "epoch": 2.573008849557522, |
| "grad_norm": 0.1497460901737213, |
| "learning_rate": 6.048909080341936e-07, |
| "loss": 0.3509, |
| "step": 1163 |
| }, |
| { |
| "epoch": 2.5752212389380533, |
| "grad_norm": 0.15763600170612335, |
| "learning_rate": 5.987667433756844e-07, |
| "loss": 0.3511, |
| "step": 1164 |
| }, |
| { |
| "epoch": 2.577433628318584, |
| "grad_norm": 0.16659215092658997, |
| "learning_rate": 5.926717633497453e-07, |
| "loss": 0.364, |
| "step": 1165 |
| }, |
| { |
| "epoch": 2.579646017699115, |
| "grad_norm": 0.15350916981697083, |
| "learning_rate": 5.866060083722624e-07, |
| "loss": 0.3604, |
| "step": 1166 |
| }, |
| { |
| "epoch": 2.581858407079646, |
| "grad_norm": 0.1593090444803238, |
| "learning_rate": 5.805695186653365e-07, |
| "loss": 0.3554, |
| "step": 1167 |
| }, |
| { |
| "epoch": 2.584070796460177, |
| "grad_norm": 0.13886038959026337, |
| "learning_rate": 5.745623342570039e-07, |
| "loss": 0.3567, |
| "step": 1168 |
| }, |
| { |
| "epoch": 2.586283185840708, |
| "grad_norm": 0.17093704640865326, |
| "learning_rate": 5.685844949809777e-07, |
| "loss": 0.3569, |
| "step": 1169 |
| }, |
| { |
| "epoch": 2.588495575221239, |
| "grad_norm": 0.11451248824596405, |
| "learning_rate": 5.626360404763875e-07, |
| "loss": 0.3468, |
| "step": 1170 |
| }, |
| { |
| "epoch": 2.59070796460177, |
| "grad_norm": 0.15386511385440826, |
| "learning_rate": 5.567170101875074e-07, |
| "loss": 0.3639, |
| "step": 1171 |
| }, |
| { |
| "epoch": 2.5929203539823007, |
| "grad_norm": 0.14580263197422028, |
| "learning_rate": 5.508274433635019e-07, |
| "loss": 0.3555, |
| "step": 1172 |
| }, |
| { |
| "epoch": 2.5951327433628317, |
| "grad_norm": 0.14180666208267212, |
| "learning_rate": 5.449673790581611e-07, |
| "loss": 0.3593, |
| "step": 1173 |
| }, |
| { |
| "epoch": 2.5973451327433628, |
| "grad_norm": 0.13817204535007477, |
| "learning_rate": 5.391368561296456e-07, |
| "loss": 0.3412, |
| "step": 1174 |
| }, |
| { |
| "epoch": 2.599557522123894, |
| "grad_norm": 0.14034853875637054, |
| "learning_rate": 5.333359132402238e-07, |
| "loss": 0.3539, |
| "step": 1175 |
| }, |
| { |
| "epoch": 2.601769911504425, |
| "grad_norm": 0.14221332967281342, |
| "learning_rate": 5.275645888560233e-07, |
| "loss": 0.3541, |
| "step": 1176 |
| }, |
| { |
| "epoch": 2.603982300884956, |
| "grad_norm": 0.15998932719230652, |
| "learning_rate": 5.218229212467635e-07, |
| "loss": 0.3696, |
| "step": 1177 |
| }, |
| { |
| "epoch": 2.606194690265487, |
| "grad_norm": 0.15716540813446045, |
| "learning_rate": 5.161109484855182e-07, |
| "loss": 0.3518, |
| "step": 1178 |
| }, |
| { |
| "epoch": 2.6084070796460175, |
| "grad_norm": 0.13059420883655548, |
| "learning_rate": 5.104287084484489e-07, |
| "loss": 0.3468, |
| "step": 1179 |
| }, |
| { |
| "epoch": 2.6106194690265485, |
| "grad_norm": 0.15905797481536865, |
| "learning_rate": 5.047762388145582e-07, |
| "loss": 0.3515, |
| "step": 1180 |
| }, |
| { |
| "epoch": 2.6128318584070795, |
| "grad_norm": 0.1637641042470932, |
| "learning_rate": 4.991535770654449e-07, |
| "loss": 0.3589, |
| "step": 1181 |
| }, |
| { |
| "epoch": 2.6150442477876106, |
| "grad_norm": 0.14523540437221527, |
| "learning_rate": 4.935607604850473e-07, |
| "loss": 0.3532, |
| "step": 1182 |
| }, |
| { |
| "epoch": 2.6172566371681416, |
| "grad_norm": 0.15000000596046448, |
| "learning_rate": 4.879978261594037e-07, |
| "loss": 0.3557, |
| "step": 1183 |
| }, |
| { |
| "epoch": 2.6194690265486726, |
| "grad_norm": 0.155274897813797, |
| "learning_rate": 4.824648109763991e-07, |
| "loss": 0.3578, |
| "step": 1184 |
| }, |
| { |
| "epoch": 2.6216814159292037, |
| "grad_norm": 0.1417737454175949, |
| "learning_rate": 4.769617516255276e-07, |
| "loss": 0.3631, |
| "step": 1185 |
| }, |
| { |
| "epoch": 2.6238938053097343, |
| "grad_norm": 0.1490320861339569, |
| "learning_rate": 4.71488684597643e-07, |
| "loss": 0.3526, |
| "step": 1186 |
| }, |
| { |
| "epoch": 2.6261061946902657, |
| "grad_norm": 0.1615634262561798, |
| "learning_rate": 4.660456461847224e-07, |
| "loss": 0.3572, |
| "step": 1187 |
| }, |
| { |
| "epoch": 2.6283185840707963, |
| "grad_norm": 0.14165712893009186, |
| "learning_rate": 4.606326724796195e-07, |
| "loss": 0.3653, |
| "step": 1188 |
| }, |
| { |
| "epoch": 2.6305309734513274, |
| "grad_norm": 0.14493530988693237, |
| "learning_rate": 4.5524979937583203e-07, |
| "loss": 0.3527, |
| "step": 1189 |
| }, |
| { |
| "epoch": 2.6327433628318584, |
| "grad_norm": 0.13274651765823364, |
| "learning_rate": 4.498970625672588e-07, |
| "loss": 0.3544, |
| "step": 1190 |
| }, |
| { |
| "epoch": 2.6349557522123894, |
| "grad_norm": 0.16449962556362152, |
| "learning_rate": 4.445744975479627e-07, |
| "loss": 0.3589, |
| "step": 1191 |
| }, |
| { |
| "epoch": 2.6371681415929205, |
| "grad_norm": 0.1559862345457077, |
| "learning_rate": 4.392821396119407e-07, |
| "loss": 0.3507, |
| "step": 1192 |
| }, |
| { |
| "epoch": 2.6393805309734515, |
| "grad_norm": 0.15669386088848114, |
| "learning_rate": 4.3402002385288133e-07, |
| "loss": 0.3505, |
| "step": 1193 |
| }, |
| { |
| "epoch": 2.6415929203539825, |
| "grad_norm": 0.17834536731243134, |
| "learning_rate": 4.2878818516394307e-07, |
| "loss": 0.3508, |
| "step": 1194 |
| }, |
| { |
| "epoch": 2.643805309734513, |
| "grad_norm": 0.1579374521970749, |
| "learning_rate": 4.235866582375114e-07, |
| "loss": 0.3627, |
| "step": 1195 |
| }, |
| { |
| "epoch": 2.646017699115044, |
| "grad_norm": 0.16911913454532623, |
| "learning_rate": 4.184154775649768e-07, |
| "loss": 0.36, |
| "step": 1196 |
| }, |
| { |
| "epoch": 2.648230088495575, |
| "grad_norm": 0.17118600010871887, |
| "learning_rate": 4.132746774365021e-07, |
| "loss": 0.3571, |
| "step": 1197 |
| }, |
| { |
| "epoch": 2.650442477876106, |
| "grad_norm": 0.1715717762708664, |
| "learning_rate": 4.0816429194079857e-07, |
| "loss": 0.361, |
| "step": 1198 |
| }, |
| { |
| "epoch": 2.6526548672566372, |
| "grad_norm": 0.17212989926338196, |
| "learning_rate": 4.030843549648944e-07, |
| "loss": 0.3517, |
| "step": 1199 |
| }, |
| { |
| "epoch": 2.6548672566371683, |
| "grad_norm": 0.17287541925907135, |
| "learning_rate": 3.9803490019391545e-07, |
| "loss": 0.354, |
| "step": 1200 |
| }, |
| { |
| "epoch": 2.6570796460176993, |
| "grad_norm": 0.18660354614257812, |
| "learning_rate": 3.930159611108603e-07, |
| "loss": 0.3559, |
| "step": 1201 |
| }, |
| { |
| "epoch": 2.65929203539823, |
| "grad_norm": 0.13577118515968323, |
| "learning_rate": 3.880275709963749e-07, |
| "loss": 0.3612, |
| "step": 1202 |
| }, |
| { |
| "epoch": 2.661504424778761, |
| "grad_norm": 0.17823448777198792, |
| "learning_rate": 3.8306976292853794e-07, |
| "loss": 0.3542, |
| "step": 1203 |
| }, |
| { |
| "epoch": 2.663716814159292, |
| "grad_norm": 0.14912089705467224, |
| "learning_rate": 3.7814256978263465e-07, |
| "loss": 0.3525, |
| "step": 1204 |
| }, |
| { |
| "epoch": 2.665929203539823, |
| "grad_norm": 0.16290871798992157, |
| "learning_rate": 3.7324602423094603e-07, |
| "loss": 0.3601, |
| "step": 1205 |
| }, |
| { |
| "epoch": 2.668141592920354, |
| "grad_norm": 0.15941615402698517, |
| "learning_rate": 3.683801587425251e-07, |
| "loss": 0.3627, |
| "step": 1206 |
| }, |
| { |
| "epoch": 2.670353982300885, |
| "grad_norm": 0.14429469406604767, |
| "learning_rate": 3.635450055829881e-07, |
| "loss": 0.3604, |
| "step": 1207 |
| }, |
| { |
| "epoch": 2.672566371681416, |
| "grad_norm": 0.15333802998065948, |
| "learning_rate": 3.587405968142943e-07, |
| "loss": 0.3507, |
| "step": 1208 |
| }, |
| { |
| "epoch": 2.6747787610619467, |
| "grad_norm": 0.1445317417383194, |
| "learning_rate": 3.539669642945387e-07, |
| "loss": 0.3567, |
| "step": 1209 |
| }, |
| { |
| "epoch": 2.676991150442478, |
| "grad_norm": 0.16064175963401794, |
| "learning_rate": 3.4922413967773905e-07, |
| "loss": 0.3623, |
| "step": 1210 |
| }, |
| { |
| "epoch": 2.6792035398230087, |
| "grad_norm": 0.1633787751197815, |
| "learning_rate": 3.4451215441362264e-07, |
| "loss": 0.3509, |
| "step": 1211 |
| }, |
| { |
| "epoch": 2.6814159292035398, |
| "grad_norm": 0.14114798605442047, |
| "learning_rate": 3.3983103974742415e-07, |
| "loss": 0.3658, |
| "step": 1212 |
| }, |
| { |
| "epoch": 2.683628318584071, |
| "grad_norm": 0.15633976459503174, |
| "learning_rate": 3.351808267196721e-07, |
| "loss": 0.3668, |
| "step": 1213 |
| }, |
| { |
| "epoch": 2.685840707964602, |
| "grad_norm": 0.14381296932697296, |
| "learning_rate": 3.3056154616598777e-07, |
| "loss": 0.3581, |
| "step": 1214 |
| }, |
| { |
| "epoch": 2.688053097345133, |
| "grad_norm": 0.14406733214855194, |
| "learning_rate": 3.2597322871687766e-07, |
| "loss": 0.3543, |
| "step": 1215 |
| }, |
| { |
| "epoch": 2.6902654867256635, |
| "grad_norm": 0.156460240483284, |
| "learning_rate": 3.214159047975324e-07, |
| "loss": 0.352, |
| "step": 1216 |
| }, |
| { |
| "epoch": 2.692477876106195, |
| "grad_norm": 0.1374759078025818, |
| "learning_rate": 3.1688960462762263e-07, |
| "loss": 0.3601, |
| "step": 1217 |
| }, |
| { |
| "epoch": 2.6946902654867255, |
| "grad_norm": 0.16220209002494812, |
| "learning_rate": 3.1239435822110253e-07, |
| "loss": 0.3614, |
| "step": 1218 |
| }, |
| { |
| "epoch": 2.6969026548672566, |
| "grad_norm": 0.15912321209907532, |
| "learning_rate": 3.079301953860059e-07, |
| "loss": 0.3573, |
| "step": 1219 |
| }, |
| { |
| "epoch": 2.6991150442477876, |
| "grad_norm": 0.16917406022548676, |
| "learning_rate": 3.034971457242525e-07, |
| "loss": 0.3601, |
| "step": 1220 |
| }, |
| { |
| "epoch": 2.7013274336283186, |
| "grad_norm": 0.16305102407932281, |
| "learning_rate": 2.990952386314505e-07, |
| "loss": 0.3594, |
| "step": 1221 |
| }, |
| { |
| "epoch": 2.7035398230088497, |
| "grad_norm": 0.1461646407842636, |
| "learning_rate": 2.947245032967e-07, |
| "loss": 0.3517, |
| "step": 1222 |
| }, |
| { |
| "epoch": 2.7057522123893807, |
| "grad_norm": 0.16588905453681946, |
| "learning_rate": 2.9038496870240187e-07, |
| "loss": 0.3564, |
| "step": 1223 |
| }, |
| { |
| "epoch": 2.7079646017699117, |
| "grad_norm": 0.14253613352775574, |
| "learning_rate": 2.860766636240636e-07, |
| "loss": 0.3633, |
| "step": 1224 |
| }, |
| { |
| "epoch": 2.7101769911504423, |
| "grad_norm": 0.15777863562107086, |
| "learning_rate": 2.817996166301107e-07, |
| "loss": 0.3465, |
| "step": 1225 |
| }, |
| { |
| "epoch": 2.7123893805309733, |
| "grad_norm": 0.1424233764410019, |
| "learning_rate": 2.7755385608169374e-07, |
| "loss": 0.3523, |
| "step": 1226 |
| }, |
| { |
| "epoch": 2.7146017699115044, |
| "grad_norm": 0.1586133986711502, |
| "learning_rate": 2.733394101325054e-07, |
| "loss": 0.3579, |
| "step": 1227 |
| }, |
| { |
| "epoch": 2.7168141592920354, |
| "grad_norm": 0.14557451009750366, |
| "learning_rate": 2.691563067285874e-07, |
| "loss": 0.3539, |
| "step": 1228 |
| }, |
| { |
| "epoch": 2.7190265486725664, |
| "grad_norm": 0.15484808385372162, |
| "learning_rate": 2.65004573608153e-07, |
| "loss": 0.3577, |
| "step": 1229 |
| }, |
| { |
| "epoch": 2.7212389380530975, |
| "grad_norm": 0.14703123271465302, |
| "learning_rate": 2.608842383013943e-07, |
| "loss": 0.3586, |
| "step": 1230 |
| }, |
| { |
| "epoch": 2.7234513274336285, |
| "grad_norm": 0.1443336308002472, |
| "learning_rate": 2.567953281303059e-07, |
| "loss": 0.3571, |
| "step": 1231 |
| }, |
| { |
| "epoch": 2.725663716814159, |
| "grad_norm": 0.14558246731758118, |
| "learning_rate": 2.527378702085037e-07, |
| "loss": 0.3584, |
| "step": 1232 |
| }, |
| { |
| "epoch": 2.72787610619469, |
| "grad_norm": 0.14799098670482635, |
| "learning_rate": 2.4871189144104025e-07, |
| "loss": 0.3612, |
| "step": 1233 |
| }, |
| { |
| "epoch": 2.730088495575221, |
| "grad_norm": 0.14273111522197723, |
| "learning_rate": 2.447174185242324e-07, |
| "loss": 0.3557, |
| "step": 1234 |
| }, |
| { |
| "epoch": 2.732300884955752, |
| "grad_norm": 0.15998686850070953, |
| "learning_rate": 2.407544779454779e-07, |
| "loss": 0.3445, |
| "step": 1235 |
| }, |
| { |
| "epoch": 2.734513274336283, |
| "grad_norm": 0.16166724264621735, |
| "learning_rate": 2.368230959830875e-07, |
| "loss": 0.3381, |
| "step": 1236 |
| }, |
| { |
| "epoch": 2.7367256637168142, |
| "grad_norm": 0.15504410862922668, |
| "learning_rate": 2.32923298706102e-07, |
| "loss": 0.3548, |
| "step": 1237 |
| }, |
| { |
| "epoch": 2.7389380530973453, |
| "grad_norm": 0.15463854372501373, |
| "learning_rate": 2.2905511197412634e-07, |
| "loss": 0.354, |
| "step": 1238 |
| }, |
| { |
| "epoch": 2.741150442477876, |
| "grad_norm": 0.15159274637699127, |
| "learning_rate": 2.2521856143715492e-07, |
| "loss": 0.353, |
| "step": 1239 |
| }, |
| { |
| "epoch": 2.7433628318584073, |
| "grad_norm": 0.16764990985393524, |
| "learning_rate": 2.2141367253540325e-07, |
| "loss": 0.3575, |
| "step": 1240 |
| }, |
| { |
| "epoch": 2.745575221238938, |
| "grad_norm": 0.15683040022850037, |
| "learning_rate": 2.1764047049913528e-07, |
| "loss": 0.3505, |
| "step": 1241 |
| }, |
| { |
| "epoch": 2.747787610619469, |
| "grad_norm": 0.15731599926948547, |
| "learning_rate": 2.1389898034850086e-07, |
| "loss": 0.3592, |
| "step": 1242 |
| }, |
| { |
| "epoch": 2.75, |
| "grad_norm": 0.15384253859519958, |
| "learning_rate": 2.1018922689336973e-07, |
| "loss": 0.3599, |
| "step": 1243 |
| }, |
| { |
| "epoch": 2.752212389380531, |
| "grad_norm": 0.1472090780735016, |
| "learning_rate": 2.0651123473316103e-07, |
| "loss": 0.358, |
| "step": 1244 |
| }, |
| { |
| "epoch": 2.754424778761062, |
| "grad_norm": 0.144203320145607, |
| "learning_rate": 2.0286502825668852e-07, |
| "loss": 0.3519, |
| "step": 1245 |
| }, |
| { |
| "epoch": 2.7566371681415927, |
| "grad_norm": 0.1510661542415619, |
| "learning_rate": 1.992506316419912e-07, |
| "loss": 0.3585, |
| "step": 1246 |
| }, |
| { |
| "epoch": 2.758849557522124, |
| "grad_norm": 0.1556992530822754, |
| "learning_rate": 1.9566806885617906e-07, |
| "loss": 0.3556, |
| "step": 1247 |
| }, |
| { |
| "epoch": 2.7610619469026547, |
| "grad_norm": 0.14370451867580414, |
| "learning_rate": 1.921173636552698e-07, |
| "loss": 0.3595, |
| "step": 1248 |
| }, |
| { |
| "epoch": 2.7632743362831858, |
| "grad_norm": 0.14656773209571838, |
| "learning_rate": 1.8859853958403507e-07, |
| "loss": 0.3483, |
| "step": 1249 |
| }, |
| { |
| "epoch": 2.765486725663717, |
| "grad_norm": 0.13567841053009033, |
| "learning_rate": 1.8511161997584015e-07, |
| "loss": 0.354, |
| "step": 1250 |
| }, |
| { |
| "epoch": 2.767699115044248, |
| "grad_norm": 0.1580413579940796, |
| "learning_rate": 1.8165662795249172e-07, |
| "loss": 0.3644, |
| "step": 1251 |
| }, |
| { |
| "epoch": 2.769911504424779, |
| "grad_norm": 0.14433561265468597, |
| "learning_rate": 1.7823358642408694e-07, |
| "loss": 0.3485, |
| "step": 1252 |
| }, |
| { |
| "epoch": 2.77212389380531, |
| "grad_norm": 0.1432889699935913, |
| "learning_rate": 1.748425180888541e-07, |
| "loss": 0.3645, |
| "step": 1253 |
| }, |
| { |
| "epoch": 2.774336283185841, |
| "grad_norm": 0.14327970147132874, |
| "learning_rate": 1.7148344543301277e-07, |
| "loss": 0.3495, |
| "step": 1254 |
| }, |
| { |
| "epoch": 2.7765486725663715, |
| "grad_norm": 0.1495339274406433, |
| "learning_rate": 1.681563907306133e-07, |
| "loss": 0.3439, |
| "step": 1255 |
| }, |
| { |
| "epoch": 2.7787610619469025, |
| "grad_norm": 0.14024530351161957, |
| "learning_rate": 1.6486137604339813e-07, |
| "loss": 0.3537, |
| "step": 1256 |
| }, |
| { |
| "epoch": 2.7809734513274336, |
| "grad_norm": 0.16141167283058167, |
| "learning_rate": 1.6159842322065022e-07, |
| "loss": 0.3502, |
| "step": 1257 |
| }, |
| { |
| "epoch": 2.7831858407079646, |
| "grad_norm": 0.14094969630241394, |
| "learning_rate": 1.5836755389905035e-07, |
| "loss": 0.3506, |
| "step": 1258 |
| }, |
| { |
| "epoch": 2.7853982300884956, |
| "grad_norm": 0.15461289882659912, |
| "learning_rate": 1.5516878950253333e-07, |
| "loss": 0.3616, |
| "step": 1259 |
| }, |
| { |
| "epoch": 2.7876106194690267, |
| "grad_norm": 0.14529578387737274, |
| "learning_rate": 1.5200215124214491e-07, |
| "loss": 0.3474, |
| "step": 1260 |
| }, |
| { |
| "epoch": 2.7898230088495577, |
| "grad_norm": 0.13706517219543457, |
| "learning_rate": 1.4886766011590449e-07, |
| "loss": 0.3509, |
| "step": 1261 |
| }, |
| { |
| "epoch": 2.7920353982300883, |
| "grad_norm": 0.15313966572284698, |
| "learning_rate": 1.4576533690866035e-07, |
| "loss": 0.3551, |
| "step": 1262 |
| }, |
| { |
| "epoch": 2.7942477876106193, |
| "grad_norm": 0.1422608494758606, |
| "learning_rate": 1.4269520219195753e-07, |
| "loss": 0.3437, |
| "step": 1263 |
| }, |
| { |
| "epoch": 2.7964601769911503, |
| "grad_norm": 0.13179507851600647, |
| "learning_rate": 1.396572763238957e-07, |
| "loss": 0.3562, |
| "step": 1264 |
| }, |
| { |
| "epoch": 2.7986725663716814, |
| "grad_norm": 0.14244070649147034, |
| "learning_rate": 1.3665157944900198e-07, |
| "loss": 0.3641, |
| "step": 1265 |
| }, |
| { |
| "epoch": 2.8008849557522124, |
| "grad_norm": 0.14658492803573608, |
| "learning_rate": 1.3367813149808728e-07, |
| "loss": 0.3567, |
| "step": 1266 |
| }, |
| { |
| "epoch": 2.8030973451327434, |
| "grad_norm": 0.1432352066040039, |
| "learning_rate": 1.3073695218812356e-07, |
| "loss": 0.3483, |
| "step": 1267 |
| }, |
| { |
| "epoch": 2.8053097345132745, |
| "grad_norm": 0.1332673579454422, |
| "learning_rate": 1.278280610221072e-07, |
| "loss": 0.3594, |
| "step": 1268 |
| }, |
| { |
| "epoch": 2.807522123893805, |
| "grad_norm": 0.15009014308452606, |
| "learning_rate": 1.2495147728893043e-07, |
| "loss": 0.3592, |
| "step": 1269 |
| }, |
| { |
| "epoch": 2.8097345132743365, |
| "grad_norm": 0.13184407353401184, |
| "learning_rate": 1.2210722006325782e-07, |
| "loss": 0.3542, |
| "step": 1270 |
| }, |
| { |
| "epoch": 2.811946902654867, |
| "grad_norm": 0.13673901557922363, |
| "learning_rate": 1.192953082053927e-07, |
| "loss": 0.3622, |
| "step": 1271 |
| }, |
| { |
| "epoch": 2.814159292035398, |
| "grad_norm": 0.1729208379983902, |
| "learning_rate": 1.1651576036115942e-07, |
| "loss": 0.3544, |
| "step": 1272 |
| }, |
| { |
| "epoch": 2.816371681415929, |
| "grad_norm": 0.139160618185997, |
| "learning_rate": 1.1376859496177228e-07, |
| "loss": 0.3499, |
| "step": 1273 |
| }, |
| { |
| "epoch": 2.8185840707964602, |
| "grad_norm": 0.15454445779323578, |
| "learning_rate": 1.110538302237213e-07, |
| "loss": 0.3572, |
| "step": 1274 |
| }, |
| { |
| "epoch": 2.8207964601769913, |
| "grad_norm": 0.1475721150636673, |
| "learning_rate": 1.0837148414864329e-07, |
| "loss": 0.3636, |
| "step": 1275 |
| }, |
| { |
| "epoch": 2.823008849557522, |
| "grad_norm": 0.1556055098772049, |
| "learning_rate": 1.0572157452321097e-07, |
| "loss": 0.348, |
| "step": 1276 |
| }, |
| { |
| "epoch": 2.8252212389380533, |
| "grad_norm": 0.14078259468078613, |
| "learning_rate": 1.0310411891900629e-07, |
| "loss": 0.3546, |
| "step": 1277 |
| }, |
| { |
| "epoch": 2.827433628318584, |
| "grad_norm": 0.15832266211509705, |
| "learning_rate": 1.0051913469241003e-07, |
| "loss": 0.3639, |
| "step": 1278 |
| }, |
| { |
| "epoch": 2.829646017699115, |
| "grad_norm": 0.16918309032917023, |
| "learning_rate": 9.796663898448578e-08, |
| "loss": 0.3579, |
| "step": 1279 |
| }, |
| { |
| "epoch": 2.831858407079646, |
| "grad_norm": 0.13976359367370605, |
| "learning_rate": 9.544664872086329e-08, |
| "loss": 0.3632, |
| "step": 1280 |
| }, |
| { |
| "epoch": 2.834070796460177, |
| "grad_norm": 0.14802858233451843, |
| "learning_rate": 9.295918061163034e-08, |
| "loss": 0.3548, |
| "step": 1281 |
| }, |
| { |
| "epoch": 2.836283185840708, |
| "grad_norm": 0.15616321563720703, |
| "learning_rate": 9.050425115121775e-08, |
| "loss": 0.3521, |
| "step": 1282 |
| }, |
| { |
| "epoch": 2.838495575221239, |
| "grad_norm": 0.13561472296714783, |
| "learning_rate": 8.808187661829504e-08, |
| "loss": 0.3601, |
| "step": 1283 |
| }, |
| { |
| "epoch": 2.84070796460177, |
| "grad_norm": 0.16056296229362488, |
| "learning_rate": 8.569207307565664e-08, |
| "loss": 0.3596, |
| "step": 1284 |
| }, |
| { |
| "epoch": 2.8429203539823007, |
| "grad_norm": 0.1590995341539383, |
| "learning_rate": 8.333485637012029e-08, |
| "loss": 0.3573, |
| "step": 1285 |
| }, |
| { |
| "epoch": 2.8451327433628317, |
| "grad_norm": 0.15869586169719696, |
| "learning_rate": 8.101024213241826e-08, |
| "loss": 0.3526, |
| "step": 1286 |
| }, |
| { |
| "epoch": 2.8473451327433628, |
| "grad_norm": 0.15596012771129608, |
| "learning_rate": 7.871824577709797e-08, |
| "loss": 0.3504, |
| "step": 1287 |
| }, |
| { |
| "epoch": 2.849557522123894, |
| "grad_norm": 0.1455894559621811, |
| "learning_rate": 7.645888250241485e-08, |
| "loss": 0.3554, |
| "step": 1288 |
| }, |
| { |
| "epoch": 2.851769911504425, |
| "grad_norm": 0.1496601700782776, |
| "learning_rate": 7.423216729023574e-08, |
| "loss": 0.3666, |
| "step": 1289 |
| }, |
| { |
| "epoch": 2.853982300884956, |
| "grad_norm": 0.13408933579921722, |
| "learning_rate": 7.203811490593626e-08, |
| "loss": 0.3636, |
| "step": 1290 |
| }, |
| { |
| "epoch": 2.856194690265487, |
| "grad_norm": 0.1369679570198059, |
| "learning_rate": 6.987673989830523e-08, |
| "loss": 0.3587, |
| "step": 1291 |
| }, |
| { |
| "epoch": 2.8584070796460175, |
| "grad_norm": 0.15344880521297455, |
| "learning_rate": 6.774805659944761e-08, |
| "loss": 0.3522, |
| "step": 1292 |
| }, |
| { |
| "epoch": 2.8606194690265485, |
| "grad_norm": 0.1609773337841034, |
| "learning_rate": 6.565207912468785e-08, |
| "loss": 0.3535, |
| "step": 1293 |
| }, |
| { |
| "epoch": 2.8628318584070795, |
| "grad_norm": 0.14126433432102203, |
| "learning_rate": 6.358882137248001e-08, |
| "loss": 0.3452, |
| "step": 1294 |
| }, |
| { |
| "epoch": 2.8650442477876106, |
| "grad_norm": 0.13827714323997498, |
| "learning_rate": 6.15582970243117e-08, |
| "loss": 0.3488, |
| "step": 1295 |
| }, |
| { |
| "epoch": 2.8672566371681416, |
| "grad_norm": 0.1625031977891922, |
| "learning_rate": 5.9560519544614725e-08, |
| "loss": 0.3541, |
| "step": 1296 |
| }, |
| { |
| "epoch": 2.8694690265486726, |
| "grad_norm": 0.14330527186393738, |
| "learning_rate": 5.759550218067622e-08, |
| "loss": 0.3535, |
| "step": 1297 |
| }, |
| { |
| "epoch": 2.8716814159292037, |
| "grad_norm": 0.16162008047103882, |
| "learning_rate": 5.566325796255101e-08, |
| "loss": 0.3562, |
| "step": 1298 |
| }, |
| { |
| "epoch": 2.8738938053097343, |
| "grad_norm": 0.1462399661540985, |
| "learning_rate": 5.3763799702975516e-08, |
| "loss": 0.3531, |
| "step": 1299 |
| }, |
| { |
| "epoch": 2.8761061946902657, |
| "grad_norm": 0.14997854828834534, |
| "learning_rate": 5.1897139997280074e-08, |
| "loss": 0.3561, |
| "step": 1300 |
| }, |
| { |
| "epoch": 2.8783185840707963, |
| "grad_norm": 0.1586892157793045, |
| "learning_rate": 5.0063291223308993e-08, |
| "loss": 0.3636, |
| "step": 1301 |
| }, |
| { |
| "epoch": 2.8805309734513274, |
| "grad_norm": 0.1484476923942566, |
| "learning_rate": 4.826226554133562e-08, |
| "loss": 0.3637, |
| "step": 1302 |
| }, |
| { |
| "epoch": 2.8827433628318584, |
| "grad_norm": 0.1511233150959015, |
| "learning_rate": 4.649407489398461e-08, |
| "loss": 0.3618, |
| "step": 1303 |
| }, |
| { |
| "epoch": 2.8849557522123894, |
| "grad_norm": 0.14198242127895355, |
| "learning_rate": 4.4758731006149804e-08, |
| "loss": 0.3595, |
| "step": 1304 |
| }, |
| { |
| "epoch": 2.8871681415929205, |
| "grad_norm": 0.15620240569114685, |
| "learning_rate": 4.305624538491815e-08, |
| "loss": 0.3599, |
| "step": 1305 |
| }, |
| { |
| "epoch": 2.8893805309734515, |
| "grad_norm": 0.15184152126312256, |
| "learning_rate": 4.138662931949255e-08, |
| "loss": 0.3552, |
| "step": 1306 |
| }, |
| { |
| "epoch": 2.8915929203539825, |
| "grad_norm": 0.14311784505844116, |
| "learning_rate": 3.974989388111861e-08, |
| "loss": 0.3568, |
| "step": 1307 |
| }, |
| { |
| "epoch": 2.893805309734513, |
| "grad_norm": 0.14970344305038452, |
| "learning_rate": 3.814604992300908e-08, |
| "loss": 0.3564, |
| "step": 1308 |
| }, |
| { |
| "epoch": 2.896017699115044, |
| "grad_norm": 0.13278964161872864, |
| "learning_rate": 3.657510808027343e-08, |
| "loss": 0.362, |
| "step": 1309 |
| }, |
| { |
| "epoch": 2.898230088495575, |
| "grad_norm": 0.15549008548259735, |
| "learning_rate": 3.503707876984674e-08, |
| "loss": 0.353, |
| "step": 1310 |
| }, |
| { |
| "epoch": 2.900442477876106, |
| "grad_norm": 0.15713003277778625, |
| "learning_rate": 3.353197219041981e-08, |
| "loss": 0.3599, |
| "step": 1311 |
| }, |
| { |
| "epoch": 2.9026548672566372, |
| "grad_norm": 0.17100383341312408, |
| "learning_rate": 3.205979832237416e-08, |
| "loss": 0.36, |
| "step": 1312 |
| }, |
| { |
| "epoch": 2.9048672566371683, |
| "grad_norm": 0.15952634811401367, |
| "learning_rate": 3.062056692771154e-08, |
| "loss": 0.3562, |
| "step": 1313 |
| }, |
| { |
| "epoch": 2.9070796460176993, |
| "grad_norm": 0.13878712058067322, |
| "learning_rate": 2.9214287549995114e-08, |
| "loss": 0.3486, |
| "step": 1314 |
| }, |
| { |
| "epoch": 2.90929203539823, |
| "grad_norm": 0.14025112986564636, |
| "learning_rate": 2.7840969514279503e-08, |
| "loss": 0.3565, |
| "step": 1315 |
| }, |
| { |
| "epoch": 2.911504424778761, |
| "grad_norm": 0.15756577253341675, |
| "learning_rate": 2.6500621927054716e-08, |
| "loss": 0.3475, |
| "step": 1316 |
| }, |
| { |
| "epoch": 2.913716814159292, |
| "grad_norm": 0.14622363448143005, |
| "learning_rate": 2.519325367618175e-08, |
| "loss": 0.3631, |
| "step": 1317 |
| }, |
| { |
| "epoch": 2.915929203539823, |
| "grad_norm": 0.15228000283241272, |
| "learning_rate": 2.3918873430835986e-08, |
| "loss": 0.3489, |
| "step": 1318 |
| }, |
| { |
| "epoch": 2.918141592920354, |
| "grad_norm": 0.1473086029291153, |
| "learning_rate": 2.2677489641448335e-08, |
| "loss": 0.3459, |
| "step": 1319 |
| }, |
| { |
| "epoch": 2.920353982300885, |
| "grad_norm": 0.13773617148399353, |
| "learning_rate": 2.1469110539650283e-08, |
| "loss": 0.3674, |
| "step": 1320 |
| }, |
| { |
| "epoch": 2.922566371681416, |
| "grad_norm": 0.15642863512039185, |
| "learning_rate": 2.0293744138219495e-08, |
| "loss": 0.3493, |
| "step": 1321 |
| }, |
| { |
| "epoch": 2.9247787610619467, |
| "grad_norm": 0.1433519721031189, |
| "learning_rate": 1.9151398231024297e-08, |
| "loss": 0.3517, |
| "step": 1322 |
| }, |
| { |
| "epoch": 2.926991150442478, |
| "grad_norm": 0.14097705483436584, |
| "learning_rate": 1.8042080392974837e-08, |
| "loss": 0.3654, |
| "step": 1323 |
| }, |
| { |
| "epoch": 2.9292035398230087, |
| "grad_norm": 0.15193448960781097, |
| "learning_rate": 1.6965797979971442e-08, |
| "loss": 0.3556, |
| "step": 1324 |
| }, |
| { |
| "epoch": 2.9314159292035398, |
| "grad_norm": 0.14747443795204163, |
| "learning_rate": 1.5922558128856903e-08, |
| "loss": 0.3521, |
| "step": 1325 |
| }, |
| { |
| "epoch": 2.933628318584071, |
| "grad_norm": 0.16591234505176544, |
| "learning_rate": 1.4912367757366485e-08, |
| "loss": 0.3628, |
| "step": 1326 |
| }, |
| { |
| "epoch": 2.935840707964602, |
| "grad_norm": 0.15506120026111603, |
| "learning_rate": 1.3935233564086326e-08, |
| "loss": 0.3529, |
| "step": 1327 |
| }, |
| { |
| "epoch": 2.938053097345133, |
| "grad_norm": 0.16547641158103943, |
| "learning_rate": 1.2991162028405113e-08, |
| "loss": 0.3603, |
| "step": 1328 |
| }, |
| { |
| "epoch": 2.9402654867256635, |
| "grad_norm": 0.17579355835914612, |
| "learning_rate": 1.2080159410471914e-08, |
| "loss": 0.3652, |
| "step": 1329 |
| }, |
| { |
| "epoch": 2.942477876106195, |
| "grad_norm": 0.13282181322574615, |
| "learning_rate": 1.1202231751157866e-08, |
| "loss": 0.3529, |
| "step": 1330 |
| }, |
| { |
| "epoch": 2.9446902654867255, |
| "grad_norm": 0.14515455067157745, |
| "learning_rate": 1.0357384872011767e-08, |
| "loss": 0.3637, |
| "step": 1331 |
| }, |
| { |
| "epoch": 2.9469026548672566, |
| "grad_norm": 0.1463669091463089, |
| "learning_rate": 9.545624375223439e-09, |
| "loss": 0.3551, |
| "step": 1332 |
| }, |
| { |
| "epoch": 2.9491150442477876, |
| "grad_norm": 0.1455363929271698, |
| "learning_rate": 8.766955643587094e-09, |
| "loss": 0.3561, |
| "step": 1333 |
| }, |
| { |
| "epoch": 2.9513274336283186, |
| "grad_norm": 0.15577441453933716, |
| "learning_rate": 8.021383840465247e-09, |
| "loss": 0.3589, |
| "step": 1334 |
| }, |
| { |
| "epoch": 2.9535398230088497, |
| "grad_norm": 0.14496655762195587, |
| "learning_rate": 7.308913909752635e-09, |
| "loss": 0.3571, |
| "step": 1335 |
| }, |
| { |
| "epoch": 2.9557522123893807, |
| "grad_norm": 0.15788689255714417, |
| "learning_rate": 6.629550575847355e-09, |
| "loss": 0.3575, |
| "step": 1336 |
| }, |
| { |
| "epoch": 2.9579646017699117, |
| "grad_norm": 0.1579669564962387, |
| "learning_rate": 5.983298343615884e-09, |
| "loss": 0.3494, |
| "step": 1337 |
| }, |
| { |
| "epoch": 2.9601769911504423, |
| "grad_norm": 0.15941593050956726, |
| "learning_rate": 5.3701614983647786e-09, |
| "loss": 0.3584, |
| "step": 1338 |
| }, |
| { |
| "epoch": 2.9623893805309733, |
| "grad_norm": 0.16959086060523987, |
| "learning_rate": 4.7901441058118006e-09, |
| "loss": 0.3517, |
| "step": 1339 |
| }, |
| { |
| "epoch": 2.9646017699115044, |
| "grad_norm": 0.16718046367168427, |
| "learning_rate": 4.243250012059275e-09, |
| "loss": 0.3584, |
| "step": 1340 |
| }, |
| { |
| "epoch": 2.9668141592920354, |
| "grad_norm": 0.1667267084121704, |
| "learning_rate": 3.729482843569665e-09, |
| "loss": 0.3652, |
| "step": 1341 |
| }, |
| { |
| "epoch": 2.9690265486725664, |
| "grad_norm": 0.15754228830337524, |
| "learning_rate": 3.2488460071389285e-09, |
| "loss": 0.3578, |
| "step": 1342 |
| }, |
| { |
| "epoch": 2.9712389380530975, |
| "grad_norm": 0.15564315021038055, |
| "learning_rate": 2.801342689875974e-09, |
| "loss": 0.3509, |
| "step": 1343 |
| }, |
| { |
| "epoch": 2.9734513274336285, |
| "grad_norm": 0.16392084956169128, |
| "learning_rate": 2.3869758591810177e-09, |
| "loss": 0.3574, |
| "step": 1344 |
| }, |
| { |
| "epoch": 2.975663716814159, |
| "grad_norm": 0.147927388548851, |
| "learning_rate": 2.005748262725593e-09, |
| "loss": 0.3557, |
| "step": 1345 |
| }, |
| { |
| "epoch": 2.97787610619469, |
| "grad_norm": 0.14830511808395386, |
| "learning_rate": 1.657662428434792e-09, |
| "loss": 0.3524, |
| "step": 1346 |
| }, |
| { |
| "epoch": 2.980088495575221, |
| "grad_norm": 0.15716874599456787, |
| "learning_rate": 1.342720664469499e-09, |
| "loss": 0.3549, |
| "step": 1347 |
| }, |
| { |
| "epoch": 2.982300884955752, |
| "grad_norm": 0.17521563172340393, |
| "learning_rate": 1.0609250592130693e-09, |
| "loss": 0.3629, |
| "step": 1348 |
| }, |
| { |
| "epoch": 2.984513274336283, |
| "grad_norm": 0.15542881190776825, |
| "learning_rate": 8.122774812552303e-10, |
| "loss": 0.3501, |
| "step": 1349 |
| }, |
| { |
| "epoch": 2.9867256637168142, |
| "grad_norm": 0.20576362311840057, |
| "learning_rate": 5.967795793820896e-10, |
| "loss": 0.3627, |
| "step": 1350 |
| }, |
| { |
| "epoch": 2.9889380530973453, |
| "grad_norm": 0.16061358153820038, |
| "learning_rate": 4.1443278256170227e-10, |
| "loss": 0.3524, |
| "step": 1351 |
| }, |
| { |
| "epoch": 2.991150442477876, |
| "grad_norm": 0.1754293441772461, |
| "learning_rate": 2.6523829993907456e-10, |
| "loss": 0.3595, |
| "step": 1352 |
| }, |
| { |
| "epoch": 2.9933628318584073, |
| "grad_norm": 0.17124155163764954, |
| "learning_rate": 1.4919712082339665e-10, |
| "loss": 0.3549, |
| "step": 1353 |
| }, |
| { |
| "epoch": 2.995575221238938, |
| "grad_norm": 0.14956091344356537, |
| "learning_rate": 6.63100146863771e-11, |
| "loss": 0.353, |
| "step": 1354 |
| }, |
| { |
| "epoch": 2.997787610619469, |
| "grad_norm": 0.1708720326423645, |
| "learning_rate": 1.6577531152806027e-11, |
| "loss": 0.3521, |
| "step": 1355 |
| }, |
| { |
| "epoch": 3.0, |
| "grad_norm": 0.17063909769058228, |
| "learning_rate": 0.0, |
| "loss": 0.3461, |
| "step": 1356 |
| }, |
| { |
| "epoch": 3.0, |
| "step": 1356, |
| "total_flos": 2.751475469831373e+16, |
| "train_loss": 0.20257837947315172, |
| "train_runtime": 61703.9023, |
| "train_samples_per_second": 8.428, |
| "train_steps_per_second": 0.022 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 1356, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 100, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 2.751475469831373e+16, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
