| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "eval_steps": 500, | |
| "global_step": 1235, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004048582995951417, | |
| "grad_norm": 10.071609260973078, | |
| "learning_rate": 3.2258064516129035e-07, | |
| "loss": 1.4438, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.008097165991902834, | |
| "grad_norm": 9.821432455501931, | |
| "learning_rate": 6.451612903225807e-07, | |
| "loss": 1.4356, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.012145748987854251, | |
| "grad_norm": 9.833457684714906, | |
| "learning_rate": 9.67741935483871e-07, | |
| "loss": 1.4329, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.016194331983805668, | |
| "grad_norm": 9.65457500165867, | |
| "learning_rate": 1.2903225806451614e-06, | |
| "loss": 1.4187, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.020242914979757085, | |
| "grad_norm": 9.24901931080625, | |
| "learning_rate": 1.6129032258064516e-06, | |
| "loss": 1.3945, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.024291497975708502, | |
| "grad_norm": 8.961712793191921, | |
| "learning_rate": 1.935483870967742e-06, | |
| "loss": 1.4054, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02834008097165992, | |
| "grad_norm": 6.832523324571276, | |
| "learning_rate": 2.2580645161290324e-06, | |
| "loss": 1.3306, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.032388663967611336, | |
| "grad_norm": 6.444850857997571, | |
| "learning_rate": 2.580645161290323e-06, | |
| "loss": 1.3355, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03643724696356275, | |
| "grad_norm": 3.7587454044683244, | |
| "learning_rate": 2.903225806451613e-06, | |
| "loss": 1.2369, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.04048582995951417, | |
| "grad_norm": 3.4381877674669865, | |
| "learning_rate": 3.225806451612903e-06, | |
| "loss": 1.2522, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.044534412955465584, | |
| "grad_norm": 3.042582437937669, | |
| "learning_rate": 3.548387096774194e-06, | |
| "loss": 1.2318, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.048582995951417005, | |
| "grad_norm": 5.872656940107583, | |
| "learning_rate": 3.870967741935484e-06, | |
| "loss": 1.2059, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.05263157894736842, | |
| "grad_norm": 6.042879384459821, | |
| "learning_rate": 4.193548387096774e-06, | |
| "loss": 1.1994, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.05668016194331984, | |
| "grad_norm": 5.41905098862907, | |
| "learning_rate": 4.516129032258065e-06, | |
| "loss": 1.1792, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.06072874493927125, | |
| "grad_norm": 4.811236215857367, | |
| "learning_rate": 4.838709677419355e-06, | |
| "loss": 1.1428, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.06477732793522267, | |
| "grad_norm": 2.7613011363861504, | |
| "learning_rate": 5.161290322580646e-06, | |
| "loss": 1.0779, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06882591093117409, | |
| "grad_norm": 2.686381624297244, | |
| "learning_rate": 5.483870967741935e-06, | |
| "loss": 1.0609, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0728744939271255, | |
| "grad_norm": 2.4726028576882935, | |
| "learning_rate": 5.806451612903226e-06, | |
| "loss": 1.0442, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.07692307692307693, | |
| "grad_norm": 2.131313757630042, | |
| "learning_rate": 6.129032258064517e-06, | |
| "loss": 1.0181, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.08097165991902834, | |
| "grad_norm": 1.7821883539781014, | |
| "learning_rate": 6.451612903225806e-06, | |
| "loss": 1.0014, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08502024291497975, | |
| "grad_norm": 1.6039498896800564, | |
| "learning_rate": 6.774193548387097e-06, | |
| "loss": 0.9749, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.08906882591093117, | |
| "grad_norm": 1.3447451851826, | |
| "learning_rate": 7.096774193548388e-06, | |
| "loss": 0.9861, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0931174089068826, | |
| "grad_norm": 1.348759384011975, | |
| "learning_rate": 7.4193548387096784e-06, | |
| "loss": 0.9497, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.09716599190283401, | |
| "grad_norm": 1.2552168774211803, | |
| "learning_rate": 7.741935483870968e-06, | |
| "loss": 0.9407, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.10121457489878542, | |
| "grad_norm": 1.2124766409541914, | |
| "learning_rate": 8.064516129032258e-06, | |
| "loss": 0.9115, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.10526315789473684, | |
| "grad_norm": 1.1608986668133614, | |
| "learning_rate": 8.387096774193549e-06, | |
| "loss": 0.9114, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.10931174089068826, | |
| "grad_norm": 1.1202542719166533, | |
| "learning_rate": 8.70967741935484e-06, | |
| "loss": 0.8908, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.11336032388663968, | |
| "grad_norm": 1.0112740843194796, | |
| "learning_rate": 9.03225806451613e-06, | |
| "loss": 0.8672, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.11740890688259109, | |
| "grad_norm": 1.0282332773116032, | |
| "learning_rate": 9.35483870967742e-06, | |
| "loss": 0.8303, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.1214574898785425, | |
| "grad_norm": 0.9826121785369821, | |
| "learning_rate": 9.67741935483871e-06, | |
| "loss": 0.828, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.12550607287449392, | |
| "grad_norm": 1.199869824342117, | |
| "learning_rate": 1e-05, | |
| "loss": 0.7991, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.12955465587044535, | |
| "grad_norm": 0.9721410232423505, | |
| "learning_rate": 1.0322580645161291e-05, | |
| "loss": 0.7939, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.13360323886639677, | |
| "grad_norm": 0.95704982818329, | |
| "learning_rate": 1.0645161290322582e-05, | |
| "loss": 0.7662, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.13765182186234817, | |
| "grad_norm": 0.9706250949542301, | |
| "learning_rate": 1.096774193548387e-05, | |
| "loss": 0.7407, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.1417004048582996, | |
| "grad_norm": 1.0763602453276153, | |
| "learning_rate": 1.1290322580645164e-05, | |
| "loss": 0.7201, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.145748987854251, | |
| "grad_norm": 0.9118583839790483, | |
| "learning_rate": 1.1612903225806453e-05, | |
| "loss": 0.6853, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.14979757085020243, | |
| "grad_norm": 1.0406677633575512, | |
| "learning_rate": 1.1935483870967743e-05, | |
| "loss": 0.6779, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.15384615384615385, | |
| "grad_norm": 0.9551909507844649, | |
| "learning_rate": 1.2258064516129034e-05, | |
| "loss": 0.6557, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.15789473684210525, | |
| "grad_norm": 0.9754304182117205, | |
| "learning_rate": 1.2580645161290324e-05, | |
| "loss": 0.6384, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.16194331983805668, | |
| "grad_norm": 0.9470798676277908, | |
| "learning_rate": 1.2903225806451613e-05, | |
| "loss": 0.5957, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1659919028340081, | |
| "grad_norm": 0.8621318303769712, | |
| "learning_rate": 1.3225806451612903e-05, | |
| "loss": 0.5652, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.1700404858299595, | |
| "grad_norm": 1.1673651972297365, | |
| "learning_rate": 1.3548387096774194e-05, | |
| "loss": 0.5781, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.17408906882591094, | |
| "grad_norm": 0.993565596062832, | |
| "learning_rate": 1.3870967741935486e-05, | |
| "loss": 0.554, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.17813765182186234, | |
| "grad_norm": 0.9983277210004106, | |
| "learning_rate": 1.4193548387096776e-05, | |
| "loss": 0.5037, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.18218623481781376, | |
| "grad_norm": 1.5776518523787202, | |
| "learning_rate": 1.4516129032258066e-05, | |
| "loss": 0.534, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.1862348178137652, | |
| "grad_norm": 0.937785076137023, | |
| "learning_rate": 1.4838709677419357e-05, | |
| "loss": 0.5176, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.1902834008097166, | |
| "grad_norm": 1.3883224077789271, | |
| "learning_rate": 1.5161290322580646e-05, | |
| "loss": 0.4852, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.19433198380566802, | |
| "grad_norm": 0.9834930866723117, | |
| "learning_rate": 1.5483870967741936e-05, | |
| "loss": 0.4368, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.19838056680161945, | |
| "grad_norm": 1.8295941588269171, | |
| "learning_rate": 1.5806451612903226e-05, | |
| "loss": 0.4361, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.20242914979757085, | |
| "grad_norm": 1.071402400223168, | |
| "learning_rate": 1.6129032258064517e-05, | |
| "loss": 0.4141, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.20647773279352227, | |
| "grad_norm": 1.5535535719730926, | |
| "learning_rate": 1.6451612903225807e-05, | |
| "loss": 0.382, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.21052631578947367, | |
| "grad_norm": 1.2101028599273784, | |
| "learning_rate": 1.6774193548387098e-05, | |
| "loss": 0.3681, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.2145748987854251, | |
| "grad_norm": 1.6494554555022165, | |
| "learning_rate": 1.7096774193548388e-05, | |
| "loss": 0.329, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.21862348178137653, | |
| "grad_norm": 1.458164666687377, | |
| "learning_rate": 1.741935483870968e-05, | |
| "loss": 0.3342, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.22267206477732793, | |
| "grad_norm": 1.55732338592029, | |
| "learning_rate": 1.774193548387097e-05, | |
| "loss": 0.2981, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.22672064777327935, | |
| "grad_norm": 1.4407050872081797, | |
| "learning_rate": 1.806451612903226e-05, | |
| "loss": 0.313, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.23076923076923078, | |
| "grad_norm": 1.0765296575486252, | |
| "learning_rate": 1.838709677419355e-05, | |
| "loss": 0.2602, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.23481781376518218, | |
| "grad_norm": 1.6941328246961866, | |
| "learning_rate": 1.870967741935484e-05, | |
| "loss": 0.2602, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.2388663967611336, | |
| "grad_norm": 1.035109384650917, | |
| "learning_rate": 1.903225806451613e-05, | |
| "loss": 0.2675, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.242914979757085, | |
| "grad_norm": 1.2546219182123133, | |
| "learning_rate": 1.935483870967742e-05, | |
| "loss": 0.2337, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.24696356275303644, | |
| "grad_norm": 1.0397882400614518, | |
| "learning_rate": 1.967741935483871e-05, | |
| "loss": 0.2459, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.25101214574898784, | |
| "grad_norm": 1.1151812170458462, | |
| "learning_rate": 2e-05, | |
| "loss": 0.1964, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.2550607287449393, | |
| "grad_norm": 1.0675939309802733, | |
| "learning_rate": 2.0322580645161292e-05, | |
| "loss": 0.2099, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.2591093117408907, | |
| "grad_norm": 0.9565344899056574, | |
| "learning_rate": 2.0645161290322582e-05, | |
| "loss": 0.1789, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.2631578947368421, | |
| "grad_norm": 0.9419075371886411, | |
| "learning_rate": 2.0967741935483873e-05, | |
| "loss": 0.1848, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.26720647773279355, | |
| "grad_norm": 0.8822713394098127, | |
| "learning_rate": 2.1290322580645163e-05, | |
| "loss": 0.1465, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.27125506072874495, | |
| "grad_norm": 0.9709146716202983, | |
| "learning_rate": 2.161290322580645e-05, | |
| "loss": 0.1577, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.27530364372469635, | |
| "grad_norm": 1.2594039192045887, | |
| "learning_rate": 2.193548387096774e-05, | |
| "loss": 0.1712, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2793522267206478, | |
| "grad_norm": 0.9360206064902659, | |
| "learning_rate": 2.225806451612903e-05, | |
| "loss": 0.1148, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.2834008097165992, | |
| "grad_norm": 0.976279333372765, | |
| "learning_rate": 2.2580645161290328e-05, | |
| "loss": 0.1304, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2874493927125506, | |
| "grad_norm": 0.8796066844650432, | |
| "learning_rate": 2.2903225806451615e-05, | |
| "loss": 0.1165, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.291497975708502, | |
| "grad_norm": 1.0708617267385623, | |
| "learning_rate": 2.3225806451612906e-05, | |
| "loss": 0.1335, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.29554655870445345, | |
| "grad_norm": 0.8183758609774919, | |
| "learning_rate": 2.3548387096774196e-05, | |
| "loss": 0.0941, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.29959514170040485, | |
| "grad_norm": 0.749618074536778, | |
| "learning_rate": 2.3870967741935486e-05, | |
| "loss": 0.1179, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.30364372469635625, | |
| "grad_norm": 0.7985681331929608, | |
| "learning_rate": 2.4193548387096777e-05, | |
| "loss": 0.0986, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.3076923076923077, | |
| "grad_norm": 0.7490131868093833, | |
| "learning_rate": 2.4516129032258067e-05, | |
| "loss": 0.087, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.3117408906882591, | |
| "grad_norm": 0.6509735122422271, | |
| "learning_rate": 2.4838709677419358e-05, | |
| "loss": 0.0873, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3157894736842105, | |
| "grad_norm": 0.7571704127659357, | |
| "learning_rate": 2.5161290322580648e-05, | |
| "loss": 0.0808, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.31983805668016196, | |
| "grad_norm": 0.6950925054621598, | |
| "learning_rate": 2.548387096774194e-05, | |
| "loss": 0.0708, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.32388663967611336, | |
| "grad_norm": 0.6538067355418463, | |
| "learning_rate": 2.5806451612903226e-05, | |
| "loss": 0.0692, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.32793522267206476, | |
| "grad_norm": 0.7658680687565181, | |
| "learning_rate": 2.6129032258064516e-05, | |
| "loss": 0.0661, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.3319838056680162, | |
| "grad_norm": 0.7364896859988205, | |
| "learning_rate": 2.6451612903225806e-05, | |
| "loss": 0.0682, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.3360323886639676, | |
| "grad_norm": 0.7171278258425478, | |
| "learning_rate": 2.6774193548387097e-05, | |
| "loss": 0.0699, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.340080971659919, | |
| "grad_norm": 0.6328933510267436, | |
| "learning_rate": 2.7096774193548387e-05, | |
| "loss": 0.0493, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.3441295546558704, | |
| "grad_norm": 2.1382606874505927, | |
| "learning_rate": 2.7419354838709678e-05, | |
| "loss": 0.057, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.3481781376518219, | |
| "grad_norm": 0.785750538008199, | |
| "learning_rate": 2.774193548387097e-05, | |
| "loss": 0.0633, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.3522267206477733, | |
| "grad_norm": 0.7712633134875542, | |
| "learning_rate": 2.8064516129032262e-05, | |
| "loss": 0.0575, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.3562753036437247, | |
| "grad_norm": 0.7445017081510712, | |
| "learning_rate": 2.8387096774193552e-05, | |
| "loss": 0.0596, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.3603238866396761, | |
| "grad_norm": 0.8030420676733789, | |
| "learning_rate": 2.8709677419354843e-05, | |
| "loss": 0.0453, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.3643724696356275, | |
| "grad_norm": 0.6908472598287251, | |
| "learning_rate": 2.9032258064516133e-05, | |
| "loss": 0.0494, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3684210526315789, | |
| "grad_norm": 0.5756365015936168, | |
| "learning_rate": 2.9354838709677423e-05, | |
| "loss": 0.0419, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.3724696356275304, | |
| "grad_norm": 1.39469059194505, | |
| "learning_rate": 2.9677419354838714e-05, | |
| "loss": 0.0454, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.3765182186234818, | |
| "grad_norm": 0.5882996945783179, | |
| "learning_rate": 3.0000000000000004e-05, | |
| "loss": 0.0463, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.3805668016194332, | |
| "grad_norm": 0.49928065195555327, | |
| "learning_rate": 3.032258064516129e-05, | |
| "loss": 0.0379, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.38461538461538464, | |
| "grad_norm": 0.4961320087811067, | |
| "learning_rate": 3.0645161290322585e-05, | |
| "loss": 0.0353, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.38866396761133604, | |
| "grad_norm": 0.5562588737308852, | |
| "learning_rate": 3.096774193548387e-05, | |
| "loss": 0.0432, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.39271255060728744, | |
| "grad_norm": 0.48391063962127323, | |
| "learning_rate": 3.1290322580645166e-05, | |
| "loss": 0.0305, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.3967611336032389, | |
| "grad_norm": 0.5080877650952307, | |
| "learning_rate": 3.161290322580645e-05, | |
| "loss": 0.0406, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.4008097165991903, | |
| "grad_norm": 0.4173731014078851, | |
| "learning_rate": 3.193548387096774e-05, | |
| "loss": 0.0271, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.4048582995951417, | |
| "grad_norm": 0.44569255338507296, | |
| "learning_rate": 3.2258064516129034e-05, | |
| "loss": 0.0281, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4089068825910931, | |
| "grad_norm": 0.47524307388014675, | |
| "learning_rate": 3.258064516129033e-05, | |
| "loss": 0.0306, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.41295546558704455, | |
| "grad_norm": 0.43796210828807663, | |
| "learning_rate": 3.2903225806451614e-05, | |
| "loss": 0.0269, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.41700404858299595, | |
| "grad_norm": 0.55587998212623, | |
| "learning_rate": 3.322580645161291e-05, | |
| "loss": 0.0376, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.42105263157894735, | |
| "grad_norm": 2.3195497235399674, | |
| "learning_rate": 3.3548387096774195e-05, | |
| "loss": 0.0276, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.4251012145748988, | |
| "grad_norm": 0.4681075524291391, | |
| "learning_rate": 3.387096774193549e-05, | |
| "loss": 0.0267, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.4291497975708502, | |
| "grad_norm": 0.4068746729680895, | |
| "learning_rate": 3.4193548387096776e-05, | |
| "loss": 0.0265, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.4331983805668016, | |
| "grad_norm": 0.45445728165341526, | |
| "learning_rate": 3.451612903225807e-05, | |
| "loss": 0.0273, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.43724696356275305, | |
| "grad_norm": 0.4522954317684568, | |
| "learning_rate": 3.483870967741936e-05, | |
| "loss": 0.0312, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.44129554655870445, | |
| "grad_norm": 0.4032193008427547, | |
| "learning_rate": 3.516129032258065e-05, | |
| "loss": 0.0215, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.44534412955465585, | |
| "grad_norm": 0.45776093559426145, | |
| "learning_rate": 3.548387096774194e-05, | |
| "loss": 0.0275, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.4493927125506073, | |
| "grad_norm": 0.4432607058722638, | |
| "learning_rate": 3.5806451612903225e-05, | |
| "loss": 0.0247, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.4534412955465587, | |
| "grad_norm": 0.36690637137772736, | |
| "learning_rate": 3.612903225806452e-05, | |
| "loss": 0.0187, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.4574898785425101, | |
| "grad_norm": 0.4408448180809998, | |
| "learning_rate": 3.6451612903225805e-05, | |
| "loss": 0.0254, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.46153846153846156, | |
| "grad_norm": 0.3610986575290524, | |
| "learning_rate": 3.67741935483871e-05, | |
| "loss": 0.0214, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.46558704453441296, | |
| "grad_norm": 0.41763864819802976, | |
| "learning_rate": 3.7096774193548386e-05, | |
| "loss": 0.0285, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.46963562753036436, | |
| "grad_norm": 0.41830765650221763, | |
| "learning_rate": 3.741935483870968e-05, | |
| "loss": 0.0251, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.47368421052631576, | |
| "grad_norm": 0.44150651053579715, | |
| "learning_rate": 3.7741935483870974e-05, | |
| "loss": 0.0227, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.4777327935222672, | |
| "grad_norm": 0.3846618394157275, | |
| "learning_rate": 3.806451612903226e-05, | |
| "loss": 0.0193, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.4817813765182186, | |
| "grad_norm": 0.40177068439066044, | |
| "learning_rate": 3.8387096774193555e-05, | |
| "loss": 0.0207, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.48582995951417, | |
| "grad_norm": 0.3870446184449085, | |
| "learning_rate": 3.870967741935484e-05, | |
| "loss": 0.0195, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.4898785425101215, | |
| "grad_norm": 0.39870220186793137, | |
| "learning_rate": 3.9032258064516135e-05, | |
| "loss": 0.0209, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.4939271255060729, | |
| "grad_norm": 0.4111289508432182, | |
| "learning_rate": 3.935483870967742e-05, | |
| "loss": 0.0217, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.4979757085020243, | |
| "grad_norm": 0.39855680477589234, | |
| "learning_rate": 3.9677419354838716e-05, | |
| "loss": 0.0224, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.5020242914979757, | |
| "grad_norm": 0.33837424539582583, | |
| "learning_rate": 4e-05, | |
| "loss": 0.016, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.5060728744939271, | |
| "grad_norm": 0.32875428948791197, | |
| "learning_rate": 3.9999920040266474e-05, | |
| "loss": 0.0147, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5101214574898786, | |
| "grad_norm": 0.3780642659382347, | |
| "learning_rate": 3.999968016170525e-05, | |
| "loss": 0.0211, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.5141700404858299, | |
| "grad_norm": 0.36621987074485457, | |
| "learning_rate": 3.9999280366234394e-05, | |
| "loss": 0.0175, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.5182186234817814, | |
| "grad_norm": 0.3970386976551513, | |
| "learning_rate": 3.999872065705065e-05, | |
| "loss": 0.0199, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.5222672064777328, | |
| "grad_norm": 0.38959810112089294, | |
| "learning_rate": 3.999800103862945e-05, | |
| "loss": 0.0226, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.5263157894736842, | |
| "grad_norm": 0.36110687355713356, | |
| "learning_rate": 3.9997121516724834e-05, | |
| "loss": 0.0174, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.5303643724696356, | |
| "grad_norm": 0.33186093685379425, | |
| "learning_rate": 3.999608209836944e-05, | |
| "loss": 0.0146, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.5344129554655871, | |
| "grad_norm": 0.38011158011600793, | |
| "learning_rate": 3.999488279187443e-05, | |
| "loss": 0.0188, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.5384615384615384, | |
| "grad_norm": 0.3539488510481163, | |
| "learning_rate": 3.9993523606829425e-05, | |
| "loss": 0.0172, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.5425101214574899, | |
| "grad_norm": 0.360525575143427, | |
| "learning_rate": 3.9992004554102435e-05, | |
| "loss": 0.0167, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.5465587044534413, | |
| "grad_norm": 0.3454541123795058, | |
| "learning_rate": 3.999032564583976e-05, | |
| "loss": 0.0149, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.5506072874493927, | |
| "grad_norm": 0.3342104362520803, | |
| "learning_rate": 3.998848689546591e-05, | |
| "loss": 0.0166, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.5546558704453441, | |
| "grad_norm": 0.36521833718831326, | |
| "learning_rate": 3.9986488317683496e-05, | |
| "loss": 0.0192, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.5587044534412956, | |
| "grad_norm": 0.3279525260674068, | |
| "learning_rate": 3.9984329928473067e-05, | |
| "loss": 0.0151, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.562753036437247, | |
| "grad_norm": 0.3310779612480269, | |
| "learning_rate": 3.998201174509306e-05, | |
| "loss": 0.0159, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.5668016194331984, | |
| "grad_norm": 0.3414480299596721, | |
| "learning_rate": 3.997953378607961e-05, | |
| "loss": 0.0161, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.5708502024291497, | |
| "grad_norm": 0.3291002994510692, | |
| "learning_rate": 3.9976896071246403e-05, | |
| "loss": 0.0143, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.5748987854251012, | |
| "grad_norm": 0.3447851899063351, | |
| "learning_rate": 3.997409862168454e-05, | |
| "loss": 0.0152, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.5789473684210527, | |
| "grad_norm": 0.34017387712024694, | |
| "learning_rate": 3.997114145976237e-05, | |
| "loss": 0.0163, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.582995951417004, | |
| "grad_norm": 0.294646633462219, | |
| "learning_rate": 3.9968024609125245e-05, | |
| "loss": 0.0128, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.5870445344129555, | |
| "grad_norm": 0.2921998175699193, | |
| "learning_rate": 3.996474809469545e-05, | |
| "loss": 0.0147, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.5910931174089069, | |
| "grad_norm": 0.2869397260559656, | |
| "learning_rate": 3.996131194267189e-05, | |
| "loss": 0.0118, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.5951417004048583, | |
| "grad_norm": 0.33729920700471283, | |
| "learning_rate": 3.995771618052995e-05, | |
| "loss": 0.0179, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.5991902834008097, | |
| "grad_norm": 0.3185074391655103, | |
| "learning_rate": 3.9953960837021256e-05, | |
| "loss": 0.0157, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.6032388663967612, | |
| "grad_norm": 0.3016091135645087, | |
| "learning_rate": 3.995004594217343e-05, | |
| "loss": 0.013, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.6072874493927125, | |
| "grad_norm": 0.34263284227519836, | |
| "learning_rate": 3.994597152728987e-05, | |
| "loss": 0.0146, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.611336032388664, | |
| "grad_norm": 0.28771053373089084, | |
| "learning_rate": 3.994173762494947e-05, | |
| "loss": 0.0128, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.6153846153846154, | |
| "grad_norm": 0.29399821201588516, | |
| "learning_rate": 3.9937344269006424e-05, | |
| "loss": 0.0114, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.6194331983805668, | |
| "grad_norm": 0.2844474130061803, | |
| "learning_rate": 3.993279149458988e-05, | |
| "loss": 0.0111, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.6234817813765182, | |
| "grad_norm": 0.26249292454351897, | |
| "learning_rate": 3.9928079338103695e-05, | |
| "loss": 0.0099, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.6275303643724697, | |
| "grad_norm": 0.30283939992188796, | |
| "learning_rate": 3.992320783722616e-05, | |
| "loss": 0.0127, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.631578947368421, | |
| "grad_norm": 0.30943376834554165, | |
| "learning_rate": 3.991817703090965e-05, | |
| "loss": 0.0128, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.6356275303643725, | |
| "grad_norm": 0.27172814142823337, | |
| "learning_rate": 3.991298695938038e-05, | |
| "loss": 0.0121, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.6396761133603239, | |
| "grad_norm": 0.3027415432419124, | |
| "learning_rate": 3.9907637664138e-05, | |
| "loss": 0.0127, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.6437246963562753, | |
| "grad_norm": 0.276544309761548, | |
| "learning_rate": 3.990212918795534e-05, | |
| "loss": 0.0113, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.6477732793522267, | |
| "grad_norm": 0.2742693807499547, | |
| "learning_rate": 3.9896461574878036e-05, | |
| "loss": 0.0097, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.6518218623481782, | |
| "grad_norm": 0.2852328649455706, | |
| "learning_rate": 3.989063487022416e-05, | |
| "loss": 0.0103, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.6558704453441295, | |
| "grad_norm": 0.2562054048239581, | |
| "learning_rate": 3.9884649120583903e-05, | |
| "loss": 0.0086, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.659919028340081, | |
| "grad_norm": 0.2654288524861383, | |
| "learning_rate": 3.987850437381915e-05, | |
| "loss": 0.009, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.6639676113360324, | |
| "grad_norm": 0.26646772377654415, | |
| "learning_rate": 3.987220067906313e-05, | |
| "loss": 0.0101, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.6680161943319838, | |
| "grad_norm": 0.27853951015950207, | |
| "learning_rate": 3.986573808672002e-05, | |
| "loss": 0.0109, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.6720647773279352, | |
| "grad_norm": 0.28465555576877266, | |
| "learning_rate": 3.9859116648464534e-05, | |
| "loss": 0.0114, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.6761133603238867, | |
| "grad_norm": 0.2859975811906986, | |
| "learning_rate": 3.985233641724152e-05, | |
| "loss": 0.0121, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.680161943319838, | |
| "grad_norm": 0.25330690061562755, | |
| "learning_rate": 3.984539744726554e-05, | |
| "loss": 0.0101, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.6842105263157895, | |
| "grad_norm": 0.22950305199591525, | |
| "learning_rate": 3.983829979402038e-05, | |
| "loss": 0.0083, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.6882591093117408, | |
| "grad_norm": 0.263252322926028, | |
| "learning_rate": 3.983104351425871e-05, | |
| "loss": 0.011, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.6923076923076923, | |
| "grad_norm": 0.22259549583812543, | |
| "learning_rate": 3.982362866600154e-05, | |
| "loss": 0.0078, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.6963562753036437, | |
| "grad_norm": 0.22895407054438863, | |
| "learning_rate": 3.9816055308537806e-05, | |
| "loss": 0.0084, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.7004048582995951, | |
| "grad_norm": 0.24535047836232285, | |
| "learning_rate": 3.980832350242388e-05, | |
| "loss": 0.009, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.7044534412955465, | |
| "grad_norm": 0.2467262547863088, | |
| "learning_rate": 3.980043330948306e-05, | |
| "loss": 0.0083, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.708502024291498, | |
| "grad_norm": 0.23361515178309214, | |
| "learning_rate": 3.9792384792805124e-05, | |
| "loss": 0.0076, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.7125506072874493, | |
| "grad_norm": 0.25190189288347825, | |
| "learning_rate": 3.978417801674579e-05, | |
| "loss": 0.009, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.7165991902834008, | |
| "grad_norm": 0.2565776298010175, | |
| "learning_rate": 3.9775813046926244e-05, | |
| "loss": 0.0083, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.7206477732793523, | |
| "grad_norm": 0.23455602390357957, | |
| "learning_rate": 3.976728995023254e-05, | |
| "loss": 0.0075, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.7246963562753036, | |
| "grad_norm": 0.21553350322978082, | |
| "learning_rate": 3.975860879481514e-05, | |
| "loss": 0.0066, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.728744939271255, | |
| "grad_norm": 0.2247838617588064, | |
| "learning_rate": 3.974976965008833e-05, | |
| "loss": 0.0072, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.7327935222672065, | |
| "grad_norm": 0.23067318936736808, | |
| "learning_rate": 3.9740772586729674e-05, | |
| "loss": 0.0077, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.7368421052631579, | |
| "grad_norm": 0.22641159624031698, | |
| "learning_rate": 3.973161767667946e-05, | |
| "loss": 0.0081, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.7408906882591093, | |
| "grad_norm": 0.21614606145467977, | |
| "learning_rate": 3.972230499314009e-05, | |
| "loss": 0.008, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.7449392712550608, | |
| "grad_norm": 0.2566496658003984, | |
| "learning_rate": 3.9712834610575545e-05, | |
| "loss": 0.0079, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.7489878542510121, | |
| "grad_norm": 0.23641217139526333, | |
| "learning_rate": 3.9703206604710746e-05, | |
| "loss": 0.007, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.7530364372469636, | |
| "grad_norm": 0.25245613365169506, | |
| "learning_rate": 3.969342105253098e-05, | |
| "loss": 0.0087, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.757085020242915, | |
| "grad_norm": 0.21272079492094662, | |
| "learning_rate": 3.9683478032281246e-05, | |
| "loss": 0.0057, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.7611336032388664, | |
| "grad_norm": 0.20801162431467787, | |
| "learning_rate": 3.967337762346568e-05, | |
| "loss": 0.0072, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.7651821862348178, | |
| "grad_norm": 0.2393190785322378, | |
| "learning_rate": 3.9663119906846885e-05, | |
| "loss": 0.0094, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.7692307692307693, | |
| "grad_norm": 0.2012068015747044, | |
| "learning_rate": 3.965270496444528e-05, | |
| "loss": 0.0072, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.7732793522267206, | |
| "grad_norm": 0.19986775321427686, | |
| "learning_rate": 3.9642132879538465e-05, | |
| "loss": 0.0071, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.7773279352226721, | |
| "grad_norm": 0.22018526431524596, | |
| "learning_rate": 3.9631403736660566e-05, | |
| "loss": 0.0071, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.7813765182186235, | |
| "grad_norm": 0.21056338909082237, | |
| "learning_rate": 3.962051762160151e-05, | |
| "loss": 0.0062, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.7854251012145749, | |
| "grad_norm": 0.19581592405851692, | |
| "learning_rate": 3.960947462140638e-05, | |
| "loss": 0.0051, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.7894736842105263, | |
| "grad_norm": 0.20854458716644608, | |
| "learning_rate": 3.959827482437472e-05, | |
| "loss": 0.0063, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.7935222672064778, | |
| "grad_norm": 0.24425040452927557, | |
| "learning_rate": 3.958691832005981e-05, | |
| "loss": 0.0066, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.7975708502024291, | |
| "grad_norm": 0.2355474726734509, | |
| "learning_rate": 3.9575405199267944e-05, | |
| "loss": 0.0084, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.8016194331983806, | |
| "grad_norm": 0.17459311179218212, | |
| "learning_rate": 3.956373555405774e-05, | |
| "loss": 0.0048, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.805668016194332, | |
| "grad_norm": 0.19227786864664914, | |
| "learning_rate": 3.955190947773937e-05, | |
| "loss": 0.0061, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.8097165991902834, | |
| "grad_norm": 0.1910016426903256, | |
| "learning_rate": 3.9539927064873824e-05, | |
| "loss": 0.006, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.8137651821862348, | |
| "grad_norm": 0.1852267531411387, | |
| "learning_rate": 3.952778841127214e-05, | |
| "loss": 0.0051, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.8178137651821862, | |
| "grad_norm": 0.2047270375089176, | |
| "learning_rate": 3.95154936139947e-05, | |
| "loss": 0.0061, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.8218623481781376, | |
| "grad_norm": 0.19257999138992257, | |
| "learning_rate": 3.950304277135034e-05, | |
| "loss": 0.0052, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.8259109311740891, | |
| "grad_norm": 0.2014011105826715, | |
| "learning_rate": 3.94904359828957e-05, | |
| "loss": 0.0066, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.8299595141700404, | |
| "grad_norm": 0.18625453928984395, | |
| "learning_rate": 3.94776733494343e-05, | |
| "loss": 0.0058, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.8340080971659919, | |
| "grad_norm": 0.1897207679994044, | |
| "learning_rate": 3.9464754973015826e-05, | |
| "loss": 0.0057, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.8380566801619433, | |
| "grad_norm": 0.21856218099422953, | |
| "learning_rate": 3.9451680956935275e-05, | |
| "loss": 0.0061, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.8421052631578947, | |
| "grad_norm": 0.17114714305891596, | |
| "learning_rate": 3.943845140573213e-05, | |
| "loss": 0.0047, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.8461538461538461, | |
| "grad_norm": 0.20603466810250862, | |
| "learning_rate": 3.942506642518952e-05, | |
| "loss": 0.0065, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.8502024291497976, | |
| "grad_norm": 0.17687315475112578, | |
| "learning_rate": 3.9411526122333396e-05, | |
| "loss": 0.005, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.854251012145749, | |
| "grad_norm": 0.20904284943064566, | |
| "learning_rate": 3.939783060543167e-05, | |
| "loss": 0.0065, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.8582995951417004, | |
| "grad_norm": 0.14731596729968066, | |
| "learning_rate": 3.938397998399333e-05, | |
| "loss": 0.0038, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.8623481781376519, | |
| "grad_norm": 0.17717371671453785, | |
| "learning_rate": 3.936997436876756e-05, | |
| "loss": 0.0049, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.8663967611336032, | |
| "grad_norm": 0.19331542761256945, | |
| "learning_rate": 3.935581387174289e-05, | |
| "loss": 0.006, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.8704453441295547, | |
| "grad_norm": 0.17870045240016688, | |
| "learning_rate": 3.9341498606146295e-05, | |
| "loss": 0.0054, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.8744939271255061, | |
| "grad_norm": 0.20888009738009816, | |
| "learning_rate": 3.932702868644224e-05, | |
| "loss": 0.0059, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.8785425101214575, | |
| "grad_norm": 0.18743271695004546, | |
| "learning_rate": 3.931240422833181e-05, | |
| "loss": 0.0043, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.8825910931174089, | |
| "grad_norm": 0.1726780029594339, | |
| "learning_rate": 3.9297625348751806e-05, | |
| "loss": 0.0049, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.8866396761133604, | |
| "grad_norm": 0.17978468207904105, | |
| "learning_rate": 3.9282692165873744e-05, | |
| "loss": 0.0046, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.8906882591093117, | |
| "grad_norm": 0.20708253165975424, | |
| "learning_rate": 3.9267604799102945e-05, | |
| "loss": 0.006, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.8947368421052632, | |
| "grad_norm": 0.1713263587644036, | |
| "learning_rate": 3.9252363369077615e-05, | |
| "loss": 0.0038, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.8987854251012146, | |
| "grad_norm": 0.20030569613523352, | |
| "learning_rate": 3.9236967997667796e-05, | |
| "loss": 0.0058, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.902834008097166, | |
| "grad_norm": 0.1770107419496089, | |
| "learning_rate": 3.9221418807974496e-05, | |
| "loss": 0.0049, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.9068825910931174, | |
| "grad_norm": 0.19460144866137208, | |
| "learning_rate": 3.92057159243286e-05, | |
| "loss": 0.0052, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.9109311740890689, | |
| "grad_norm": 0.17481688136664114, | |
| "learning_rate": 3.9189859472289956e-05, | |
| "loss": 0.0053, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.9149797570850202, | |
| "grad_norm": 0.1775123070085614, | |
| "learning_rate": 3.9173849578646324e-05, | |
| "loss": 0.0051, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.9190283400809717, | |
| "grad_norm": 0.1638601384567398, | |
| "learning_rate": 3.9157686371412395e-05, | |
| "loss": 0.0045, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.9230769230769231, | |
| "grad_norm": 0.18786069954850723, | |
| "learning_rate": 3.914136997982874e-05, | |
| "loss": 0.0051, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.9271255060728745, | |
| "grad_norm": 0.1945474387204526, | |
| "learning_rate": 3.912490053436079e-05, | |
| "loss": 0.005, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.9311740890688259, | |
| "grad_norm": 0.153520002860529, | |
| "learning_rate": 3.9108278166697805e-05, | |
| "loss": 0.004, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.9352226720647774, | |
| "grad_norm": 0.1669067897917306, | |
| "learning_rate": 3.909150300975178e-05, | |
| "loss": 0.0046, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.9392712550607287, | |
| "grad_norm": 0.18350394502566975, | |
| "learning_rate": 3.907457519765642e-05, | |
| "loss": 0.0046, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.9433198380566802, | |
| "grad_norm": 0.1568883748861573, | |
| "learning_rate": 3.905749486576607e-05, | |
| "loss": 0.0037, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.9473684210526315, | |
| "grad_norm": 0.143373747112347, | |
| "learning_rate": 3.90402621506546e-05, | |
| "loss": 0.0033, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.951417004048583, | |
| "grad_norm": 0.17049093256567938, | |
| "learning_rate": 3.902287719011435e-05, | |
| "loss": 0.0037, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.9554655870445344, | |
| "grad_norm": 0.1982952449552468, | |
| "learning_rate": 3.900534012315499e-05, | |
| "loss": 0.0066, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.9595141700404858, | |
| "grad_norm": 0.16297088131270032, | |
| "learning_rate": 3.8987651090002454e-05, | |
| "loss": 0.0033, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.9635627530364372, | |
| "grad_norm": 0.17266192320971058, | |
| "learning_rate": 3.896981023209777e-05, | |
| "loss": 0.0045, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.9676113360323887, | |
| "grad_norm": 0.16859281828002737, | |
| "learning_rate": 3.8951817692095966e-05, | |
| "loss": 0.0049, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.97165991902834, | |
| "grad_norm": 0.18273526705191293, | |
| "learning_rate": 3.893367361386491e-05, | |
| "loss": 0.0048, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.9757085020242915, | |
| "grad_norm": 0.20038678606098476, | |
| "learning_rate": 3.891537814248417e-05, | |
| "loss": 0.0059, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.979757085020243, | |
| "grad_norm": 0.15361595940683612, | |
| "learning_rate": 3.889693142424384e-05, | |
| "loss": 0.0049, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.9838056680161943, | |
| "grad_norm": 0.17891334177183618, | |
| "learning_rate": 3.8878333606643405e-05, | |
| "loss": 0.0041, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.9878542510121457, | |
| "grad_norm": 0.15133912926154564, | |
| "learning_rate": 3.885958483839049e-05, | |
| "loss": 0.0036, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.9919028340080972, | |
| "grad_norm": 0.16880865331909614, | |
| "learning_rate": 3.8840685269399786e-05, | |
| "loss": 0.0044, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.9959514170040485, | |
| "grad_norm": 0.15282752080952322, | |
| "learning_rate": 3.882163505079171e-05, | |
| "loss": 0.0036, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.16892837426330073, | |
| "learning_rate": 3.880243433489132e-05, | |
| "loss": 0.0032, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 1.0040485829959513, | |
| "grad_norm": 0.15296869670725227, | |
| "learning_rate": 3.878308327522702e-05, | |
| "loss": 0.003, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 1.008097165991903, | |
| "grad_norm": 0.1633573700830128, | |
| "learning_rate": 3.876358202652936e-05, | |
| "loss": 0.004, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 1.0121457489878543, | |
| "grad_norm": 0.15216047559839407, | |
| "learning_rate": 3.8743930744729835e-05, | |
| "loss": 0.0037, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.0161943319838056, | |
| "grad_norm": 0.16025334168670333, | |
| "learning_rate": 3.8724129586959535e-05, | |
| "loss": 0.0033, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 1.0202429149797572, | |
| "grad_norm": 0.16984183401010405, | |
| "learning_rate": 3.870417871154801e-05, | |
| "loss": 0.0038, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 1.0242914979757085, | |
| "grad_norm": 0.1333090806588814, | |
| "learning_rate": 3.868407827802192e-05, | |
| "loss": 0.0031, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 1.0283400809716599, | |
| "grad_norm": 0.15913272635879652, | |
| "learning_rate": 3.8663828447103804e-05, | |
| "loss": 0.0038, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 1.0323886639676114, | |
| "grad_norm": 0.20884067250969093, | |
| "learning_rate": 3.8643429380710765e-05, | |
| "loss": 0.0037, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 1.0364372469635628, | |
| "grad_norm": 0.15287980091495998, | |
| "learning_rate": 3.862288124195319e-05, | |
| "loss": 0.0035, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 1.040485829959514, | |
| "grad_norm": 0.16006637115991698, | |
| "learning_rate": 3.8602184195133464e-05, | |
| "loss": 0.0039, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 1.0445344129554657, | |
| "grad_norm": 0.1571664430963936, | |
| "learning_rate": 3.8581338405744604e-05, | |
| "loss": 0.0046, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 1.048582995951417, | |
| "grad_norm": 0.20411026257669398, | |
| "learning_rate": 3.8560344040468996e-05, | |
| "loss": 0.0058, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 1.0526315789473684, | |
| "grad_norm": 0.14297042034672633, | |
| "learning_rate": 3.853920126717703e-05, | |
| "loss": 0.0034, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.05668016194332, | |
| "grad_norm": 0.17672718707282814, | |
| "learning_rate": 3.851791025492574e-05, | |
| "loss": 0.0044, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 1.0607287449392713, | |
| "grad_norm": 0.13748017886329147, | |
| "learning_rate": 3.8496471173957515e-05, | |
| "loss": 0.0036, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 1.0647773279352226, | |
| "grad_norm": 0.15250785828844354, | |
| "learning_rate": 3.847488419569866e-05, | |
| "loss": 0.0032, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 1.0688259109311742, | |
| "grad_norm": 0.13398797111043298, | |
| "learning_rate": 3.8453149492758077e-05, | |
| "loss": 0.0029, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 1.0728744939271255, | |
| "grad_norm": 0.16376863621090687, | |
| "learning_rate": 3.8431267238925886e-05, | |
| "loss": 0.0044, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 1.0769230769230769, | |
| "grad_norm": 0.1446213442772679, | |
| "learning_rate": 3.840923760917199e-05, | |
| "loss": 0.0031, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 1.0809716599190284, | |
| "grad_norm": 0.16210821837353515, | |
| "learning_rate": 3.838706077964473e-05, | |
| "loss": 0.0041, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 1.0850202429149798, | |
| "grad_norm": 0.13680523012834045, | |
| "learning_rate": 3.836473692766944e-05, | |
| "loss": 0.0029, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 1.0890688259109311, | |
| "grad_norm": 0.14587146707036522, | |
| "learning_rate": 3.834226623174705e-05, | |
| "loss": 0.0034, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 1.0931174089068827, | |
| "grad_norm": 0.14711076254972358, | |
| "learning_rate": 3.831964887155264e-05, | |
| "loss": 0.0031, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.097165991902834, | |
| "grad_norm": 0.13185195384325407, | |
| "learning_rate": 3.829688502793403e-05, | |
| "loss": 0.0027, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 1.1012145748987854, | |
| "grad_norm": 0.12548177788382, | |
| "learning_rate": 3.827397488291029e-05, | |
| "loss": 0.0022, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 1.1052631578947367, | |
| "grad_norm": 0.1432774280097295, | |
| "learning_rate": 3.8250918619670343e-05, | |
| "loss": 0.0031, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 1.1093117408906883, | |
| "grad_norm": 0.14733822736562613, | |
| "learning_rate": 3.822771642257145e-05, | |
| "loss": 0.0034, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 1.1133603238866396, | |
| "grad_norm": 0.1876067765105697, | |
| "learning_rate": 3.820436847713776e-05, | |
| "loss": 0.0039, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.117408906882591, | |
| "grad_norm": 0.17500220158697005, | |
| "learning_rate": 3.8180874970058826e-05, | |
| "loss": 0.0034, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.1214574898785425, | |
| "grad_norm": 0.1442904548637083, | |
| "learning_rate": 3.8157236089188103e-05, | |
| "loss": 0.0037, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 1.125506072874494, | |
| "grad_norm": 0.12913095914802433, | |
| "learning_rate": 3.813345202354145e-05, | |
| "loss": 0.0027, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 1.1295546558704452, | |
| "grad_norm": 0.16992900436847438, | |
| "learning_rate": 3.810952296329563e-05, | |
| "loss": 0.0044, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 1.1336032388663968, | |
| "grad_norm": 0.15709065883485654, | |
| "learning_rate": 3.808544909978676e-05, | |
| "loss": 0.0033, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.1376518218623481, | |
| "grad_norm": 0.13272646875458582, | |
| "learning_rate": 3.806123062550882e-05, | |
| "loss": 0.003, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 1.1417004048582995, | |
| "grad_norm": 0.16845134773364628, | |
| "learning_rate": 3.803686773411208e-05, | |
| "loss": 0.0042, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 1.145748987854251, | |
| "grad_norm": 0.14216829301676248, | |
| "learning_rate": 3.801236062040158e-05, | |
| "loss": 0.0026, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 1.1497975708502024, | |
| "grad_norm": 0.1627979461001903, | |
| "learning_rate": 3.798770948033553e-05, | |
| "loss": 0.0046, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 1.1538461538461537, | |
| "grad_norm": 0.13926867733734122, | |
| "learning_rate": 3.796291451102381e-05, | |
| "loss": 0.0029, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.1578947368421053, | |
| "grad_norm": 0.12887972228244493, | |
| "learning_rate": 3.793797591072631e-05, | |
| "loss": 0.0034, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 1.1619433198380567, | |
| "grad_norm": 0.1759204319505031, | |
| "learning_rate": 3.791289387885144e-05, | |
| "loss": 0.0041, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 1.165991902834008, | |
| "grad_norm": 0.1586776929402385, | |
| "learning_rate": 3.788766861595443e-05, | |
| "loss": 0.0037, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 1.1700404858299596, | |
| "grad_norm": 0.12303741158941259, | |
| "learning_rate": 3.7862300323735835e-05, | |
| "loss": 0.0029, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 1.174089068825911, | |
| "grad_norm": 0.16351325014713458, | |
| "learning_rate": 3.783678920503983e-05, | |
| "loss": 0.0042, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.1781376518218623, | |
| "grad_norm": 0.1371769423679162, | |
| "learning_rate": 3.781113546385265e-05, | |
| "loss": 0.0031, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 1.1821862348178138, | |
| "grad_norm": 0.1515102657111056, | |
| "learning_rate": 3.7785339305300916e-05, | |
| "loss": 0.0033, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 1.1862348178137652, | |
| "grad_norm": 0.13771399411385085, | |
| "learning_rate": 3.7759400935650025e-05, | |
| "loss": 0.0031, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 1.1902834008097165, | |
| "grad_norm": 0.14083055732349384, | |
| "learning_rate": 3.77333205623025e-05, | |
| "loss": 0.0023, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 1.194331983805668, | |
| "grad_norm": 0.14479559569826675, | |
| "learning_rate": 3.7707098393796295e-05, | |
| "loss": 0.0028, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 1.1983805668016194, | |
| "grad_norm": 0.16007812280826417, | |
| "learning_rate": 3.7680734639803185e-05, | |
| "loss": 0.0036, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 1.2024291497975708, | |
| "grad_norm": 0.16903153675057656, | |
| "learning_rate": 3.765422951112704e-05, | |
| "loss": 0.0023, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 1.2064777327935223, | |
| "grad_norm": 0.14284942213310586, | |
| "learning_rate": 3.762758321970216e-05, | |
| "loss": 0.0026, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 1.2105263157894737, | |
| "grad_norm": 0.16344723186646015, | |
| "learning_rate": 3.7600795978591584e-05, | |
| "loss": 0.0035, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 1.214574898785425, | |
| "grad_norm": 0.13470287686728272, | |
| "learning_rate": 3.757386800198538e-05, | |
| "loss": 0.0028, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.2186234817813766, | |
| "grad_norm": 0.13338851373042954, | |
| "learning_rate": 3.7546799505198925e-05, | |
| "loss": 0.0031, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 1.222672064777328, | |
| "grad_norm": 0.14369016135000626, | |
| "learning_rate": 3.7519590704671197e-05, | |
| "loss": 0.0028, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 1.2267206477732793, | |
| "grad_norm": 0.15215179135048087, | |
| "learning_rate": 3.749224181796305e-05, | |
| "loss": 0.0043, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 1.2307692307692308, | |
| "grad_norm": 0.14397974531474, | |
| "learning_rate": 3.746475306375545e-05, | |
| "loss": 0.0033, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 1.2348178137651822, | |
| "grad_norm": 0.2822775140980981, | |
| "learning_rate": 3.743712466184774e-05, | |
| "loss": 0.0026, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 1.2388663967611335, | |
| "grad_norm": 0.13695940892422198, | |
| "learning_rate": 3.7409356833155885e-05, | |
| "loss": 0.0026, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 1.242914979757085, | |
| "grad_norm": 0.1565531843043718, | |
| "learning_rate": 3.73814497997107e-05, | |
| "loss": 0.0035, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 1.2469635627530364, | |
| "grad_norm": 0.1353773206325612, | |
| "learning_rate": 3.7353403784656095e-05, | |
| "loss": 0.003, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 1.2510121457489878, | |
| "grad_norm": 0.1378829388771464, | |
| "learning_rate": 3.732521901224724e-05, | |
| "loss": 0.0033, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 1.2550607287449393, | |
| "grad_norm": 0.1558322315856395, | |
| "learning_rate": 3.7296895707848844e-05, | |
| "loss": 0.0038, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.2591093117408907, | |
| "grad_norm": 0.15711380441589803, | |
| "learning_rate": 3.7268434097933275e-05, | |
| "loss": 0.0028, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 1.263157894736842, | |
| "grad_norm": 0.15047561749379834, | |
| "learning_rate": 3.723983441007882e-05, | |
| "loss": 0.0033, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 1.2672064777327936, | |
| "grad_norm": 0.14734601763255306, | |
| "learning_rate": 3.721109687296781e-05, | |
| "loss": 0.0029, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.271255060728745, | |
| "grad_norm": 0.13450449314922142, | |
| "learning_rate": 3.718222171638484e-05, | |
| "loss": 0.0026, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 1.2753036437246963, | |
| "grad_norm": 0.17368820512001631, | |
| "learning_rate": 3.715320917121488e-05, | |
| "loss": 0.0047, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 1.2793522267206479, | |
| "grad_norm": 0.14389734758313955, | |
| "learning_rate": 3.712405946944149e-05, | |
| "loss": 0.0035, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 1.2834008097165992, | |
| "grad_norm": 0.14488940128328817, | |
| "learning_rate": 3.709477284414488e-05, | |
| "loss": 0.0037, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 1.2874493927125505, | |
| "grad_norm": 0.14158416522222675, | |
| "learning_rate": 3.7065349529500136e-05, | |
| "loss": 0.0029, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 1.291497975708502, | |
| "grad_norm": 0.1435110357695059, | |
| "learning_rate": 3.7035789760775305e-05, | |
| "loss": 0.0037, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 1.2955465587044535, | |
| "grad_norm": 0.1374791611790858, | |
| "learning_rate": 3.70060937743295e-05, | |
| "loss": 0.0028, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.2995951417004048, | |
| "grad_norm": 0.12655775471764685, | |
| "learning_rate": 3.6976261807611045e-05, | |
| "loss": 0.002, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 1.3036437246963564, | |
| "grad_norm": 0.14224524422726664, | |
| "learning_rate": 3.694629409915555e-05, | |
| "loss": 0.003, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 1.3076923076923077, | |
| "grad_norm": 0.16126792280969068, | |
| "learning_rate": 3.691619088858401e-05, | |
| "loss": 0.0051, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 1.311740890688259, | |
| "grad_norm": 0.12226350746182763, | |
| "learning_rate": 3.68859524166009e-05, | |
| "loss": 0.0033, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 1.3157894736842106, | |
| "grad_norm": 0.13924546060918114, | |
| "learning_rate": 3.6855578924992226e-05, | |
| "loss": 0.0032, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.319838056680162, | |
| "grad_norm": 0.14572208086654767, | |
| "learning_rate": 3.6825070656623626e-05, | |
| "loss": 0.0031, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 1.3238866396761133, | |
| "grad_norm": 0.20274734675404968, | |
| "learning_rate": 3.67944278554384e-05, | |
| "loss": 0.0046, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 1.3279352226720649, | |
| "grad_norm": 0.14423391260020005, | |
| "learning_rate": 3.676365076645557e-05, | |
| "loss": 0.0028, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 1.3319838056680162, | |
| "grad_norm": 0.1725944611054312, | |
| "learning_rate": 3.673273963576791e-05, | |
| "loss": 0.0042, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 1.3360323886639676, | |
| "grad_norm": 0.14425351142131615, | |
| "learning_rate": 3.670169471054001e-05, | |
| "loss": 0.0039, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.3400809716599191, | |
| "grad_norm": 0.15152232095614562, | |
| "learning_rate": 3.6670516239006254e-05, | |
| "loss": 0.0029, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 1.3441295546558705, | |
| "grad_norm": 0.15101618483552684, | |
| "learning_rate": 3.663920447046888e-05, | |
| "loss": 0.0031, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 1.3481781376518218, | |
| "grad_norm": 0.14428941493377057, | |
| "learning_rate": 3.6607759655295954e-05, | |
| "loss": 0.0035, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.3522267206477734, | |
| "grad_norm": 0.12426247418093586, | |
| "learning_rate": 3.657618204491937e-05, | |
| "loss": 0.0028, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 1.3562753036437247, | |
| "grad_norm": 0.14322314046920476, | |
| "learning_rate": 3.6544471891832875e-05, | |
| "loss": 0.0036, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 1.360323886639676, | |
| "grad_norm": 0.16156123053613633, | |
| "learning_rate": 3.6512629449589983e-05, | |
| "loss": 0.0033, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 1.3643724696356276, | |
| "grad_norm": 0.14379385131833217, | |
| "learning_rate": 3.6480654972802045e-05, | |
| "loss": 0.004, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.368421052631579, | |
| "grad_norm": 0.14189740453065536, | |
| "learning_rate": 3.644854871713611e-05, | |
| "loss": 0.0035, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.3724696356275303, | |
| "grad_norm": 0.14683431455315699, | |
| "learning_rate": 3.6416310939312924e-05, | |
| "loss": 0.0036, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.376518218623482, | |
| "grad_norm": 0.12380102633247357, | |
| "learning_rate": 3.638394189710493e-05, | |
| "loss": 0.0026, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.3805668016194332, | |
| "grad_norm": 0.13765602163058754, | |
| "learning_rate": 3.635144184933412e-05, | |
| "loss": 0.0026, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.3846153846153846, | |
| "grad_norm": 0.14187381823803735, | |
| "learning_rate": 3.6318811055869994e-05, | |
| "loss": 0.0037, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.3886639676113361, | |
| "grad_norm": 0.1427710216931439, | |
| "learning_rate": 3.628604977762752e-05, | |
| "loss": 0.0032, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.3927125506072875, | |
| "grad_norm": 0.134784331995163, | |
| "learning_rate": 3.6253158276565004e-05, | |
| "loss": 0.0033, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.3967611336032388, | |
| "grad_norm": 0.15885390024180607, | |
| "learning_rate": 3.6220136815682016e-05, | |
| "loss": 0.0027, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.4008097165991904, | |
| "grad_norm": 0.13855782378649664, | |
| "learning_rate": 3.618698565901727e-05, | |
| "loss": 0.0035, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.4048582995951417, | |
| "grad_norm": 0.16033353283756124, | |
| "learning_rate": 3.615370507164654e-05, | |
| "loss": 0.0034, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.408906882591093, | |
| "grad_norm": 0.12684475252722205, | |
| "learning_rate": 3.61202953196805e-05, | |
| "loss": 0.0024, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.4129554655870447, | |
| "grad_norm": 0.13270980682826813, | |
| "learning_rate": 3.6086756670262654e-05, | |
| "loss": 0.0023, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.417004048582996, | |
| "grad_norm": 0.1196276497123289, | |
| "learning_rate": 3.6053089391567135e-05, | |
| "loss": 0.002, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.4210526315789473, | |
| "grad_norm": 0.12682597652315722, | |
| "learning_rate": 3.601929375279662e-05, | |
| "loss": 0.0027, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.425101214574899, | |
| "grad_norm": 0.13536995304072308, | |
| "learning_rate": 3.598537002418012e-05, | |
| "loss": 0.0029, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.4291497975708503, | |
| "grad_norm": 0.13108030608015708, | |
| "learning_rate": 3.5951318476970885e-05, | |
| "loss": 0.0028, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.4331983805668016, | |
| "grad_norm": 0.09962386338407336, | |
| "learning_rate": 3.591713938344416e-05, | |
| "loss": 0.0019, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.4372469635627532, | |
| "grad_norm": 0.1457181279579853, | |
| "learning_rate": 3.588283301689507e-05, | |
| "loss": 0.0022, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.4412955465587045, | |
| "grad_norm": 0.14047476997306665, | |
| "learning_rate": 3.5848399651636424e-05, | |
| "loss": 0.0028, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.4453441295546559, | |
| "grad_norm": 0.1213348021960891, | |
| "learning_rate": 3.5813839562996473e-05, | |
| "loss": 0.0025, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.4493927125506074, | |
| "grad_norm": 0.10647618974042507, | |
| "learning_rate": 3.5779153027316775e-05, | |
| "loss": 0.0018, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.4534412955465588, | |
| "grad_norm": 0.1334945051992226, | |
| "learning_rate": 3.574434032194994e-05, | |
| "loss": 0.0025, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.45748987854251, | |
| "grad_norm": 0.17688097486316415, | |
| "learning_rate": 3.5709401725257436e-05, | |
| "loss": 0.0037, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.4615384615384617, | |
| "grad_norm": 0.13022460054976243, | |
| "learning_rate": 3.5674337516607346e-05, | |
| "loss": 0.0025, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.465587044534413, | |
| "grad_norm": 0.12680596055656657, | |
| "learning_rate": 3.563914797637216e-05, | |
| "loss": 0.0022, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.4696356275303644, | |
| "grad_norm": 0.16061828412472026, | |
| "learning_rate": 3.560383338592649e-05, | |
| "loss": 0.0032, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.4736842105263157, | |
| "grad_norm": 0.12072248826900822, | |
| "learning_rate": 3.556839402764487e-05, | |
| "loss": 0.0021, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.4777327935222673, | |
| "grad_norm": 0.1416222681575815, | |
| "learning_rate": 3.553283018489946e-05, | |
| "loss": 0.0026, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.4817813765182186, | |
| "grad_norm": 0.13328017666013878, | |
| "learning_rate": 3.54971421420578e-05, | |
| "loss": 0.0028, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.48582995951417, | |
| "grad_norm": 0.11100348617040998, | |
| "learning_rate": 3.5461330184480524e-05, | |
| "loss": 0.0018, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.4898785425101215, | |
| "grad_norm": 0.15047898546330368, | |
| "learning_rate": 3.54253945985191e-05, | |
| "loss": 0.0037, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.4939271255060729, | |
| "grad_norm": 0.14007620193263345, | |
| "learning_rate": 3.5389335671513524e-05, | |
| "loss": 0.0024, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.4979757085020242, | |
| "grad_norm": 0.1727774890508891, | |
| "learning_rate": 3.5353153691789986e-05, | |
| "loss": 0.0042, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.5020242914979756, | |
| "grad_norm": 0.12582707931149595, | |
| "learning_rate": 3.531684894865866e-05, | |
| "loss": 0.0031, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.5060728744939271, | |
| "grad_norm": 0.12863965014335846, | |
| "learning_rate": 3.528042173241128e-05, | |
| "loss": 0.0023, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.5101214574898787, | |
| "grad_norm": 0.14164475805799012, | |
| "learning_rate": 3.524387233431892e-05, | |
| "loss": 0.0031, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.5141700404858298, | |
| "grad_norm": 0.12118761517122591, | |
| "learning_rate": 3.520720104662958e-05, | |
| "loss": 0.0026, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.5182186234817814, | |
| "grad_norm": 0.11891557194133442, | |
| "learning_rate": 3.5170408162565904e-05, | |
| "loss": 0.002, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.522267206477733, | |
| "grad_norm": 0.12996574311980993, | |
| "learning_rate": 3.513349397632281e-05, | |
| "loss": 0.0029, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.526315789473684, | |
| "grad_norm": 0.12088548915287982, | |
| "learning_rate": 3.5096458783065145e-05, | |
| "loss": 0.0026, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.5303643724696356, | |
| "grad_norm": 0.1196183500566019, | |
| "learning_rate": 3.505930287892533e-05, | |
| "loss": 0.0028, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.5344129554655872, | |
| "grad_norm": 0.1372743679797849, | |
| "learning_rate": 3.502202656100099e-05, | |
| "loss": 0.0023, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.5384615384615383, | |
| "grad_norm": 0.11477149431459815, | |
| "learning_rate": 3.498463012735256e-05, | |
| "loss": 0.002, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.54251012145749, | |
| "grad_norm": 0.12874004540775028, | |
| "learning_rate": 3.494711387700094e-05, | |
| "loss": 0.0024, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.5465587044534415, | |
| "grad_norm": 0.1280014592031412, | |
| "learning_rate": 3.490947810992505e-05, | |
| "loss": 0.0028, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.5506072874493926, | |
| "grad_norm": 0.09952752039768727, | |
| "learning_rate": 3.48717231270595e-05, | |
| "loss": 0.002, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.5546558704453441, | |
| "grad_norm": 0.13695880124371165, | |
| "learning_rate": 3.483384923029211e-05, | |
| "loss": 0.002, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.5587044534412957, | |
| "grad_norm": 0.10814349044322284, | |
| "learning_rate": 3.479585672246156e-05, | |
| "loss": 0.0016, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.5627530364372468, | |
| "grad_norm": 0.11248597438004368, | |
| "learning_rate": 3.4757745907354924e-05, | |
| "loss": 0.0016, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.5668016194331984, | |
| "grad_norm": 0.1282235813582787, | |
| "learning_rate": 3.471951708970527e-05, | |
| "loss": 0.0023, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.5708502024291497, | |
| "grad_norm": 0.12582729284103647, | |
| "learning_rate": 3.468117057518921e-05, | |
| "loss": 0.0026, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.574898785425101, | |
| "grad_norm": 0.11164302661129802, | |
| "learning_rate": 3.4642706670424426e-05, | |
| "loss": 0.0019, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.5789473684210527, | |
| "grad_norm": 0.14315851771992494, | |
| "learning_rate": 3.460412568296731e-05, | |
| "loss": 0.0039, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.582995951417004, | |
| "grad_norm": 0.10972549503887527, | |
| "learning_rate": 3.456542792131039e-05, | |
| "loss": 0.0024, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.5870445344129553, | |
| "grad_norm": 0.14558905344441117, | |
| "learning_rate": 3.4526613694879936e-05, | |
| "loss": 0.0034, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.591093117408907, | |
| "grad_norm": 0.12585859649789222, | |
| "learning_rate": 3.448768331403348e-05, | |
| "loss": 0.0026, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.5951417004048583, | |
| "grad_norm": 0.10320373066258277, | |
| "learning_rate": 3.44486370900573e-05, | |
| "loss": 0.0019, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.5991902834008096, | |
| "grad_norm": 0.17108466064572664, | |
| "learning_rate": 3.440947533516396e-05, | |
| "loss": 0.0048, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.6032388663967612, | |
| "grad_norm": 0.11304845071635755, | |
| "learning_rate": 3.4370198362489816e-05, | |
| "loss": 0.003, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.6072874493927125, | |
| "grad_norm": 0.09923103938726544, | |
| "learning_rate": 3.43308064860925e-05, | |
| "loss": 0.0021, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.6113360323886639, | |
| "grad_norm": 0.13952975770070797, | |
| "learning_rate": 3.4291300020948393e-05, | |
| "loss": 0.0034, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.6153846153846154, | |
| "grad_norm": 0.13294951006354072, | |
| "learning_rate": 3.425167928295015e-05, | |
| "loss": 0.0035, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.6194331983805668, | |
| "grad_norm": 0.13632961336477245, | |
| "learning_rate": 3.421194458890411e-05, | |
| "loss": 0.0028, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.623481781376518, | |
| "grad_norm": 0.12336380103419912, | |
| "learning_rate": 3.4172096256527865e-05, | |
| "loss": 0.0019, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.6275303643724697, | |
| "grad_norm": 0.1741801984895995, | |
| "learning_rate": 3.4132134604447595e-05, | |
| "loss": 0.0033, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.631578947368421, | |
| "grad_norm": 0.07791227207199047, | |
| "learning_rate": 3.409205995219561e-05, | |
| "loss": 0.0015, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.6356275303643724, | |
| "grad_norm": 0.08826266520098448, | |
| "learning_rate": 3.4051872620207765e-05, | |
| "loss": 0.0015, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.639676113360324, | |
| "grad_norm": 0.09492885321525994, | |
| "learning_rate": 3.40115729298209e-05, | |
| "loss": 0.0017, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.6437246963562753, | |
| "grad_norm": 0.12530489765975847, | |
| "learning_rate": 3.397116120327025e-05, | |
| "loss": 0.0023, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.6477732793522266, | |
| "grad_norm": 0.13970077480367202, | |
| "learning_rate": 3.3930637763686916e-05, | |
| "loss": 0.0039, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.6518218623481782, | |
| "grad_norm": 0.1104471117777748, | |
| "learning_rate": 3.389000293509524e-05, | |
| "loss": 0.0028, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.6558704453441295, | |
| "grad_norm": 0.09727536701344452, | |
| "learning_rate": 3.3849257042410225e-05, | |
| "loss": 0.0016, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.6599190283400809, | |
| "grad_norm": 0.10887400463908042, | |
| "learning_rate": 3.380840041143494e-05, | |
| "loss": 0.0016, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.6639676113360324, | |
| "grad_norm": 0.10428093988762577, | |
| "learning_rate": 3.376743336885792e-05, | |
| "loss": 0.0023, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.6680161943319838, | |
| "grad_norm": 0.09432466490640126, | |
| "learning_rate": 3.372635624225055e-05, | |
| "loss": 0.0018, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.6720647773279351, | |
| "grad_norm": 0.09564577584597403, | |
| "learning_rate": 3.3685169360064445e-05, | |
| "loss": 0.0016, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.6761133603238867, | |
| "grad_norm": 0.11649881389948102, | |
| "learning_rate": 3.3643873051628804e-05, | |
| "loss": 0.0018, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.680161943319838, | |
| "grad_norm": 0.11138879894464296, | |
| "learning_rate": 3.3602467647147814e-05, | |
| "loss": 0.0015, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.6842105263157894, | |
| "grad_norm": 0.10763793284900779, | |
| "learning_rate": 3.356095347769799e-05, | |
| "loss": 0.0014, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.688259109311741, | |
| "grad_norm": 0.11307992883300841, | |
| "learning_rate": 3.351933087522552e-05, | |
| "loss": 0.0017, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.6923076923076923, | |
| "grad_norm": 0.11768123662101102, | |
| "learning_rate": 3.3477600172543626e-05, | |
| "loss": 0.0021, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.6963562753036436, | |
| "grad_norm": 0.09999203267117587, | |
| "learning_rate": 3.3435761703329894e-05, | |
| "loss": 0.0011, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.7004048582995952, | |
| "grad_norm": 0.08769625667921219, | |
| "learning_rate": 3.3393815802123606e-05, | |
| "loss": 0.0016, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.7044534412955465, | |
| "grad_norm": 0.11177661027631687, | |
| "learning_rate": 3.335176280432308e-05, | |
| "loss": 0.0015, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.708502024291498, | |
| "grad_norm": 0.09151591324988914, | |
| "learning_rate": 3.330960304618295e-05, | |
| "loss": 0.0016, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.7125506072874495, | |
| "grad_norm": 0.09990486591814621, | |
| "learning_rate": 3.326733686481153e-05, | |
| "loss": 0.0013, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.7165991902834008, | |
| "grad_norm": 0.16341446141072596, | |
| "learning_rate": 3.3224964598168084e-05, | |
| "loss": 0.004, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.7206477732793521, | |
| "grad_norm": 0.08999856465117015, | |
| "learning_rate": 3.318248658506012e-05, | |
| "loss": 0.0011, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.7246963562753037, | |
| "grad_norm": 0.10490810646348823, | |
| "learning_rate": 3.313990316514069e-05, | |
| "loss": 0.0014, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.728744939271255, | |
| "grad_norm": 0.10254934353816693, | |
| "learning_rate": 3.309721467890571e-05, | |
| "loss": 0.0023, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.7327935222672064, | |
| "grad_norm": 0.09768726334482332, | |
| "learning_rate": 3.305442146769115e-05, | |
| "loss": 0.0015, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.736842105263158, | |
| "grad_norm": 0.09176299774874816, | |
| "learning_rate": 3.30115238736704e-05, | |
| "loss": 0.0013, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.7408906882591093, | |
| "grad_norm": 0.10611780057646095, | |
| "learning_rate": 3.296852223985148e-05, | |
| "loss": 0.0021, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.7449392712550607, | |
| "grad_norm": 0.08419390425315525, | |
| "learning_rate": 3.2925416910074315e-05, | |
| "loss": 0.001, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.7489878542510122, | |
| "grad_norm": 0.08307787821931087, | |
| "learning_rate": 3.288220822900796e-05, | |
| "loss": 0.001, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.7530364372469636, | |
| "grad_norm": 0.08252901428340506, | |
| "learning_rate": 3.283889654214788e-05, | |
| "loss": 0.0012, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.757085020242915, | |
| "grad_norm": 0.07535678098321162, | |
| "learning_rate": 3.279548219581317e-05, | |
| "loss": 0.001, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.7611336032388665, | |
| "grad_norm": 0.11523690613435886, | |
| "learning_rate": 3.275196553714379e-05, | |
| "loss": 0.0031, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.7651821862348178, | |
| "grad_norm": 0.08788580884497403, | |
| "learning_rate": 3.270834691409778e-05, | |
| "loss": 0.0014, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.7692307692307692, | |
| "grad_norm": 0.10991260031483728, | |
| "learning_rate": 3.266462667544849e-05, | |
| "loss": 0.002, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.7732793522267207, | |
| "grad_norm": 0.07525405971840288, | |
| "learning_rate": 3.262080517078178e-05, | |
| "loss": 0.0013, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.777327935222672, | |
| "grad_norm": 0.10186270496623287, | |
| "learning_rate": 3.257688275049323e-05, | |
| "loss": 0.0012, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.7813765182186234, | |
| "grad_norm": 0.07088841685015984, | |
| "learning_rate": 3.253285976578535e-05, | |
| "loss": 0.0012, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.785425101214575, | |
| "grad_norm": 0.09257468365438105, | |
| "learning_rate": 3.2488736568664756e-05, | |
| "loss": 0.0019, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.7894736842105263, | |
| "grad_norm": 0.08181034787049496, | |
| "learning_rate": 3.244451351193935e-05, | |
| "loss": 0.0012, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.7935222672064777, | |
| "grad_norm": 0.08203949587828167, | |
| "learning_rate": 3.240019094921551e-05, | |
| "loss": 0.001, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.7975708502024292, | |
| "grad_norm": 0.0773843309500088, | |
| "learning_rate": 3.235576923489527e-05, | |
| "loss": 0.0015, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.8016194331983806, | |
| "grad_norm": 0.10017176298195839, | |
| "learning_rate": 3.2311248724173476e-05, | |
| "loss": 0.0016, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.805668016194332, | |
| "grad_norm": 0.09346514435423071, | |
| "learning_rate": 3.226662977303494e-05, | |
| "loss": 0.0018, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.8097165991902835, | |
| "grad_norm": 0.07672377801375822, | |
| "learning_rate": 3.2221912738251626e-05, | |
| "loss": 0.0009, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.8137651821862348, | |
| "grad_norm": 0.10897440403806527, | |
| "learning_rate": 3.2177097977379736e-05, | |
| "loss": 0.0022, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.8178137651821862, | |
| "grad_norm": 0.08738847574459412, | |
| "learning_rate": 3.21321858487569e-05, | |
| "loss": 0.0015, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.8218623481781377, | |
| "grad_norm": 0.11171021034494105, | |
| "learning_rate": 3.208717671149931e-05, | |
| "loss": 0.0018, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.825910931174089, | |
| "grad_norm": 0.08316060410034828, | |
| "learning_rate": 3.204207092549882e-05, | |
| "loss": 0.0011, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.8299595141700404, | |
| "grad_norm": 0.09949526317719481, | |
| "learning_rate": 3.19968688514201e-05, | |
| "loss": 0.0015, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.834008097165992, | |
| "grad_norm": 0.09540641637940303, | |
| "learning_rate": 3.195157085069774e-05, | |
| "loss": 0.0019, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.8380566801619433, | |
| "grad_norm": 0.08673009998174742, | |
| "learning_rate": 3.190617728553332e-05, | |
| "loss": 0.0015, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.8421052631578947, | |
| "grad_norm": 0.06951149612818038, | |
| "learning_rate": 3.1860688518892606e-05, | |
| "loss": 0.0008, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.8461538461538463, | |
| "grad_norm": 0.08202638406647474, | |
| "learning_rate": 3.181510491450254e-05, | |
| "loss": 0.0013, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.8502024291497976, | |
| "grad_norm": 0.08041976167529848, | |
| "learning_rate": 3.176942683684842e-05, | |
| "loss": 0.0009, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.854251012145749, | |
| "grad_norm": 0.05697611130672205, | |
| "learning_rate": 3.1723654651170934e-05, | |
| "loss": 0.0009, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.8582995951417005, | |
| "grad_norm": 0.09337257202947909, | |
| "learning_rate": 3.167778872346327e-05, | |
| "loss": 0.0016, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.8623481781376519, | |
| "grad_norm": 0.0906398325711153, | |
| "learning_rate": 3.163182942046815e-05, | |
| "loss": 0.0012, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.8663967611336032, | |
| "grad_norm": 0.06411929340077714, | |
| "learning_rate": 3.158577710967494e-05, | |
| "loss": 0.0011, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.8704453441295548, | |
| "grad_norm": 0.06257578691470869, | |
| "learning_rate": 3.153963215931669e-05, | |
| "loss": 0.001, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.874493927125506, | |
| "grad_norm": 0.09831535828836545, | |
| "learning_rate": 3.14933949383672e-05, | |
| "loss": 0.0015, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.8785425101214575, | |
| "grad_norm": 0.09623299088902462, | |
| "learning_rate": 3.144706581653805e-05, | |
| "loss": 0.0013, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.882591093117409, | |
| "grad_norm": 0.09316746117153415, | |
| "learning_rate": 3.140064516427566e-05, | |
| "loss": 0.0012, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.8866396761133604, | |
| "grad_norm": 0.11077303705028148, | |
| "learning_rate": 3.135413335275833e-05, | |
| "loss": 0.0024, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.8906882591093117, | |
| "grad_norm": 0.060652483913783264, | |
| "learning_rate": 3.130753075389327e-05, | |
| "loss": 0.0006, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.8947368421052633, | |
| "grad_norm": 0.08552369371540584, | |
| "learning_rate": 3.1260837740313625e-05, | |
| "loss": 0.001, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.8987854251012146, | |
| "grad_norm": 0.06915816338046613, | |
| "learning_rate": 3.121405468537547e-05, | |
| "loss": 0.0007, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.902834008097166, | |
| "grad_norm": 0.0578127083534263, | |
| "learning_rate": 3.116718196315488e-05, | |
| "loss": 0.0008, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.9068825910931175, | |
| "grad_norm": 0.0641805735815862, | |
| "learning_rate": 3.1120219948444894e-05, | |
| "loss": 0.0007, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.9109311740890689, | |
| "grad_norm": 0.09021202253970874, | |
| "learning_rate": 3.107316901675252e-05, | |
| "loss": 0.0014, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.9149797570850202, | |
| "grad_norm": 0.08020291914774265, | |
| "learning_rate": 3.1026029544295745e-05, | |
| "loss": 0.0014, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.9190283400809718, | |
| "grad_norm": 0.08540121169022137, | |
| "learning_rate": 3.097880190800057e-05, | |
| "loss": 0.001, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.9230769230769231, | |
| "grad_norm": 0.06346576685967675, | |
| "learning_rate": 3.093148648549788e-05, | |
| "loss": 0.0006, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.9271255060728745, | |
| "grad_norm": 0.08019625873325945, | |
| "learning_rate": 3.088408365512055e-05, | |
| "loss": 0.0013, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.931174089068826, | |
| "grad_norm": 0.06378626829259214, | |
| "learning_rate": 3.083659379590034e-05, | |
| "loss": 0.0007, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.9352226720647774, | |
| "grad_norm": 0.08807279708290554, | |
| "learning_rate": 3.0789017287564913e-05, | |
| "loss": 0.0013, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.9392712550607287, | |
| "grad_norm": 0.07154582637497753, | |
| "learning_rate": 3.074135451053475e-05, | |
| "loss": 0.0009, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.9433198380566803, | |
| "grad_norm": 0.06206128874103014, | |
| "learning_rate": 3.069360584592015e-05, | |
| "loss": 0.0007, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.9473684210526314, | |
| "grad_norm": 0.07250758799026524, | |
| "learning_rate": 3.064577167551816e-05, | |
| "loss": 0.0009, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.951417004048583, | |
| "grad_norm": 0.08442475458111322, | |
| "learning_rate": 3.059785238180953e-05, | |
| "loss": 0.0017, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.9554655870445345, | |
| "grad_norm": 0.08443126248096765, | |
| "learning_rate": 3.054984834795566e-05, | |
| "loss": 0.0013, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.9595141700404857, | |
| "grad_norm": 0.06878101491660633, | |
| "learning_rate": 3.0501759957795537e-05, | |
| "loss": 0.0013, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.9635627530364372, | |
| "grad_norm": 0.07190059838460996, | |
| "learning_rate": 3.0453587595842627e-05, | |
| "loss": 0.001, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.9676113360323888, | |
| "grad_norm": 0.0780129716835786, | |
| "learning_rate": 3.0405331647281856e-05, | |
| "loss": 0.001, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.97165991902834, | |
| "grad_norm": 0.04418089057000386, | |
| "learning_rate": 3.0356992497966508e-05, | |
| "loss": 0.0005, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.9757085020242915, | |
| "grad_norm": 0.07561666281082768, | |
| "learning_rate": 3.0308570534415137e-05, | |
| "loss": 0.0008, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.979757085020243, | |
| "grad_norm": 0.06339746715273513, | |
| "learning_rate": 3.0260066143808467e-05, | |
| "loss": 0.0009, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.9838056680161942, | |
| "grad_norm": 0.0768868233296669, | |
| "learning_rate": 3.021147971398632e-05, | |
| "loss": 0.0013, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.9878542510121457, | |
| "grad_norm": 0.057813659642229814, | |
| "learning_rate": 3.0162811633444478e-05, | |
| "loss": 0.0008, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.9919028340080973, | |
| "grad_norm": 0.08734855981759972, | |
| "learning_rate": 3.0114062291331635e-05, | |
| "loss": 0.0008, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.9959514170040484, | |
| "grad_norm": 0.08469226714247519, | |
| "learning_rate": 3.0065232077446217e-05, | |
| "loss": 0.0014, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.05933360177662548, | |
| "learning_rate": 3.001632138223332e-05, | |
| "loss": 0.0007, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 2.0040485829959516, | |
| "grad_norm": 0.0584995634810491, | |
| "learning_rate": 2.9967330596781565e-05, | |
| "loss": 0.0006, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 2.0080971659919027, | |
| "grad_norm": 0.06683306829884199, | |
| "learning_rate": 2.9918260112819952e-05, | |
| "loss": 0.0008, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 2.0121457489878543, | |
| "grad_norm": 0.04763114929352044, | |
| "learning_rate": 2.9869110322714778e-05, | |
| "loss": 0.0005, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 2.016194331983806, | |
| "grad_norm": 0.07543772846194982, | |
| "learning_rate": 2.9819881619466447e-05, | |
| "loss": 0.0011, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 2.020242914979757, | |
| "grad_norm": 0.10272459267529165, | |
| "learning_rate": 2.9770574396706364e-05, | |
| "loss": 0.0022, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 2.0242914979757085, | |
| "grad_norm": 0.08692781208553793, | |
| "learning_rate": 2.972118904869376e-05, | |
| "loss": 0.0016, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.02834008097166, | |
| "grad_norm": 0.07328356053562211, | |
| "learning_rate": 2.9671725970312563e-05, | |
| "loss": 0.001, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 2.032388663967611, | |
| "grad_norm": 0.06085697856194978, | |
| "learning_rate": 2.9622185557068238e-05, | |
| "loss": 0.0007, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 2.0364372469635628, | |
| "grad_norm": 0.10552626256235846, | |
| "learning_rate": 2.95725682050846e-05, | |
| "loss": 0.0005, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 2.0404858299595143, | |
| "grad_norm": 0.054624069032517875, | |
| "learning_rate": 2.9522874311100684e-05, | |
| "loss": 0.0005, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 2.0445344129554655, | |
| "grad_norm": 0.044508174081321634, | |
| "learning_rate": 2.9473104272467526e-05, | |
| "loss": 0.0005, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 2.048582995951417, | |
| "grad_norm": 0.06620525257857683, | |
| "learning_rate": 2.942325848714505e-05, | |
| "loss": 0.0006, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 2.0526315789473686, | |
| "grad_norm": 0.04562837709705403, | |
| "learning_rate": 2.9373337353698804e-05, | |
| "loss": 0.0004, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 2.0566801619433197, | |
| "grad_norm": 0.08805822168596888, | |
| "learning_rate": 2.932334127129686e-05, | |
| "loss": 0.001, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 2.0607287449392713, | |
| "grad_norm": 0.06177061368271423, | |
| "learning_rate": 2.9273270639706547e-05, | |
| "loss": 0.0009, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 2.064777327935223, | |
| "grad_norm": 0.07534470359013205, | |
| "learning_rate": 2.922312585929131e-05, | |
| "loss": 0.0007, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.068825910931174, | |
| "grad_norm": 0.056675682689597065, | |
| "learning_rate": 2.9172907331007478e-05, | |
| "loss": 0.0004, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 2.0728744939271255, | |
| "grad_norm": 0.05952312455047544, | |
| "learning_rate": 2.9122615456401053e-05, | |
| "loss": 0.0005, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 2.076923076923077, | |
| "grad_norm": 0.06289763831409051, | |
| "learning_rate": 2.9072250637604535e-05, | |
| "loss": 0.0006, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 2.080971659919028, | |
| "grad_norm": 0.15823279331849702, | |
| "learning_rate": 2.9021813277333667e-05, | |
| "loss": 0.0046, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 2.08502024291498, | |
| "grad_norm": 0.07818677376174547, | |
| "learning_rate": 2.8971303778884244e-05, | |
| "loss": 0.0005, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 2.0890688259109313, | |
| "grad_norm": 0.049563010979181345, | |
| "learning_rate": 2.8920722546128863e-05, | |
| "loss": 0.0004, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 2.0931174089068825, | |
| "grad_norm": 0.09680737023228878, | |
| "learning_rate": 2.887006998351372e-05, | |
| "loss": 0.0009, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 2.097165991902834, | |
| "grad_norm": 0.07259015045157695, | |
| "learning_rate": 2.881934649605535e-05, | |
| "loss": 0.0006, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 2.1012145748987856, | |
| "grad_norm": 0.06208117462712911, | |
| "learning_rate": 2.8768552489337408e-05, | |
| "loss": 0.0007, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 2.1052631578947367, | |
| "grad_norm": 0.04481208314246549, | |
| "learning_rate": 2.8717688369507425e-05, | |
| "loss": 0.0005, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.1093117408906883, | |
| "grad_norm": 0.06083086052766766, | |
| "learning_rate": 2.8666754543273535e-05, | |
| "loss": 0.0007, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 2.11336032388664, | |
| "grad_norm": 0.06561173272197103, | |
| "learning_rate": 2.8615751417901273e-05, | |
| "loss": 0.0007, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 2.117408906882591, | |
| "grad_norm": 0.04582820869457509, | |
| "learning_rate": 2.856467940121025e-05, | |
| "loss": 0.0006, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 2.1214574898785425, | |
| "grad_norm": 0.04701586973535674, | |
| "learning_rate": 2.8513538901570964e-05, | |
| "loss": 0.0005, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 2.125506072874494, | |
| "grad_norm": 0.05759068903320962, | |
| "learning_rate": 2.846233032790149e-05, | |
| "loss": 0.0007, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.1295546558704452, | |
| "grad_norm": 0.05848633881144438, | |
| "learning_rate": 2.84110540896642e-05, | |
| "loss": 0.0005, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 2.133603238866397, | |
| "grad_norm": 0.05637717529215738, | |
| "learning_rate": 2.8359710596862555e-05, | |
| "loss": 0.0008, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 2.1376518218623484, | |
| "grad_norm": 0.05376356264943993, | |
| "learning_rate": 2.8308300260037734e-05, | |
| "loss": 0.0006, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 2.1417004048582995, | |
| "grad_norm": 0.05870882829345742, | |
| "learning_rate": 2.8256823490265424e-05, | |
| "loss": 0.0004, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 2.145748987854251, | |
| "grad_norm": 0.12931940677422152, | |
| "learning_rate": 2.8205280699152515e-05, | |
| "loss": 0.0043, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 2.1497975708502026, | |
| "grad_norm": 0.04649661102692709, | |
| "learning_rate": 2.8153672298833777e-05, | |
| "loss": 0.0005, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 2.1538461538461537, | |
| "grad_norm": 0.05682629730422205, | |
| "learning_rate": 2.8101998701968615e-05, | |
| "loss": 0.0011, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 2.1578947368421053, | |
| "grad_norm": 0.037341331564916654, | |
| "learning_rate": 2.805026032173773e-05, | |
| "loss": 0.0002, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 2.161943319838057, | |
| "grad_norm": 0.06534302747764438, | |
| "learning_rate": 2.7998457571839832e-05, | |
| "loss": 0.0007, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 2.165991902834008, | |
| "grad_norm": 0.06871161545721904, | |
| "learning_rate": 2.7946590866488322e-05, | |
| "loss": 0.0008, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 2.1700404858299596, | |
| "grad_norm": 0.10228521981005415, | |
| "learning_rate": 2.7894660620408e-05, | |
| "loss": 0.0009, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 2.174089068825911, | |
| "grad_norm": 0.0675250924838126, | |
| "learning_rate": 2.784266724883173e-05, | |
| "loss": 0.0008, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 2.1781376518218623, | |
| "grad_norm": 0.06927800287039441, | |
| "learning_rate": 2.7790611167497128e-05, | |
| "loss": 0.0007, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 2.182186234817814, | |
| "grad_norm": 0.04922855438916108, | |
| "learning_rate": 2.7738492792643224e-05, | |
| "loss": 0.0007, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 2.1862348178137654, | |
| "grad_norm": 0.06394654195997744, | |
| "learning_rate": 2.768631254100716e-05, | |
| "loss": 0.0008, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.1902834008097165, | |
| "grad_norm": 0.053674087748931114, | |
| "learning_rate": 2.7634070829820836e-05, | |
| "loss": 0.0005, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 2.194331983805668, | |
| "grad_norm": 0.051205219237521624, | |
| "learning_rate": 2.7581768076807592e-05, | |
| "loss": 0.0006, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 2.1983805668016196, | |
| "grad_norm": 0.06302354701859306, | |
| "learning_rate": 2.7529404700178832e-05, | |
| "loss": 0.0004, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 2.2024291497975708, | |
| "grad_norm": 0.0465304057247363, | |
| "learning_rate": 2.7476981118630735e-05, | |
| "loss": 0.0003, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 2.2064777327935223, | |
| "grad_norm": 0.05272536635513218, | |
| "learning_rate": 2.742449775134085e-05, | |
| "loss": 0.0005, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 2.2105263157894735, | |
| "grad_norm": 0.04700882687068295, | |
| "learning_rate": 2.7371955017964788e-05, | |
| "loss": 0.0005, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 2.214574898785425, | |
| "grad_norm": 0.041931734660082536, | |
| "learning_rate": 2.7319353338632845e-05, | |
| "loss": 0.0004, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 2.2186234817813766, | |
| "grad_norm": 0.0462308040053672, | |
| "learning_rate": 2.7266693133946645e-05, | |
| "loss": 0.0003, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 2.2226720647773277, | |
| "grad_norm": 0.04820106048765466, | |
| "learning_rate": 2.7213974824975785e-05, | |
| "loss": 0.0005, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 2.2267206477732793, | |
| "grad_norm": 0.06335875942992873, | |
| "learning_rate": 2.7161198833254464e-05, | |
| "loss": 0.0005, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.230769230769231, | |
| "grad_norm": 0.045587071324496416, | |
| "learning_rate": 2.7108365580778095e-05, | |
| "loss": 0.0004, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 2.234817813765182, | |
| "grad_norm": 0.025907065702077462, | |
| "learning_rate": 2.705547548999996e-05, | |
| "loss": 0.0003, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 2.2388663967611335, | |
| "grad_norm": 0.038726000494545144, | |
| "learning_rate": 2.7002528983827817e-05, | |
| "loss": 0.0003, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 2.242914979757085, | |
| "grad_norm": 0.06767636227031389, | |
| "learning_rate": 2.6949526485620517e-05, | |
| "loss": 0.0007, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 2.246963562753036, | |
| "grad_norm": 0.07303818709846398, | |
| "learning_rate": 2.6896468419184623e-05, | |
| "loss": 0.0008, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 2.251012145748988, | |
| "grad_norm": 0.057349443123808116, | |
| "learning_rate": 2.6843355208771024e-05, | |
| "loss": 0.0006, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 2.2550607287449393, | |
| "grad_norm": 0.06673656378495676, | |
| "learning_rate": 2.6790187279071536e-05, | |
| "loss": 0.0003, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 2.2591093117408905, | |
| "grad_norm": 0.0581244142428203, | |
| "learning_rate": 2.67369650552155e-05, | |
| "loss": 0.0006, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 2.263157894736842, | |
| "grad_norm": 0.09601661750487259, | |
| "learning_rate": 2.6683688962766407e-05, | |
| "loss": 0.0021, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 2.2672064777327936, | |
| "grad_norm": 0.031726996593891776, | |
| "learning_rate": 2.6630359427718468e-05, | |
| "loss": 0.0002, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.2712550607287447, | |
| "grad_norm": 0.06173507855191347, | |
| "learning_rate": 2.6576976876493225e-05, | |
| "loss": 0.0005, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 2.2753036437246963, | |
| "grad_norm": 0.045476579796256265, | |
| "learning_rate": 2.652354173593614e-05, | |
| "loss": 0.0004, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 2.279352226720648, | |
| "grad_norm": 0.04775839617746505, | |
| "learning_rate": 2.647005443331317e-05, | |
| "loss": 0.0003, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 2.283400809716599, | |
| "grad_norm": 0.04279286094095107, | |
| "learning_rate": 2.6416515396307354e-05, | |
| "loss": 0.0003, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 2.2874493927125505, | |
| "grad_norm": 0.04712126406972628, | |
| "learning_rate": 2.636292505301542e-05, | |
| "loss": 0.0006, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 2.291497975708502, | |
| "grad_norm": 0.03821801941541125, | |
| "learning_rate": 2.630928383194431e-05, | |
| "loss": 0.0003, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 2.2955465587044532, | |
| "grad_norm": 0.04738547833235275, | |
| "learning_rate": 2.625559216200781e-05, | |
| "loss": 0.0003, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 2.299595141700405, | |
| "grad_norm": 0.032298854570616604, | |
| "learning_rate": 2.620185047252308e-05, | |
| "loss": 0.0002, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 2.3036437246963564, | |
| "grad_norm": 0.025568613180125495, | |
| "learning_rate": 2.614805919320723e-05, | |
| "loss": 0.0002, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 2.3076923076923075, | |
| "grad_norm": 0.051240156102479166, | |
| "learning_rate": 2.6094218754173895e-05, | |
| "loss": 0.0007, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.311740890688259, | |
| "grad_norm": 0.03807173726855206, | |
| "learning_rate": 2.60403295859298e-05, | |
| "loss": 0.0002, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 2.3157894736842106, | |
| "grad_norm": 0.034949110684794955, | |
| "learning_rate": 2.5986392119371297e-05, | |
| "loss": 0.0004, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 2.3198380566801617, | |
| "grad_norm": 0.03920497043544319, | |
| "learning_rate": 2.5932406785780928e-05, | |
| "loss": 0.0003, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 2.3238866396761133, | |
| "grad_norm": 0.030792136520222148, | |
| "learning_rate": 2.5878374016823987e-05, | |
| "loss": 0.0003, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 2.327935222672065, | |
| "grad_norm": 0.10929372363913369, | |
| "learning_rate": 2.5824294244545045e-05, | |
| "loss": 0.0022, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.331983805668016, | |
| "grad_norm": 0.04178182702606532, | |
| "learning_rate": 2.577016790136453e-05, | |
| "loss": 0.0008, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 2.3360323886639676, | |
| "grad_norm": 0.056030165003915285, | |
| "learning_rate": 2.5715995420075225e-05, | |
| "loss": 0.0005, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 2.340080971659919, | |
| "grad_norm": 0.056473591241361684, | |
| "learning_rate": 2.5661777233838864e-05, | |
| "loss": 0.0006, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 2.3441295546558703, | |
| "grad_norm": 0.04353373998498146, | |
| "learning_rate": 2.5607513776182616e-05, | |
| "loss": 0.0007, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 2.348178137651822, | |
| "grad_norm": 0.050894253284328284, | |
| "learning_rate": 2.5553205480995633e-05, | |
| "loss": 0.0005, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.3522267206477734, | |
| "grad_norm": 0.024478921066805277, | |
| "learning_rate": 2.5498852782525604e-05, | |
| "loss": 0.0003, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 2.3562753036437245, | |
| "grad_norm": 0.030988399241657867, | |
| "learning_rate": 2.5444456115375252e-05, | |
| "loss": 0.0004, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 2.360323886639676, | |
| "grad_norm": 0.04070614406869648, | |
| "learning_rate": 2.539001591449889e-05, | |
| "loss": 0.0002, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 2.3643724696356276, | |
| "grad_norm": 0.04374623886805484, | |
| "learning_rate": 2.53355326151989e-05, | |
| "loss": 0.0004, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 2.3684210526315788, | |
| "grad_norm": 0.03508808374439582, | |
| "learning_rate": 2.5281006653122288e-05, | |
| "loss": 0.0002, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 2.3724696356275303, | |
| "grad_norm": 0.04242837549415554, | |
| "learning_rate": 2.5226438464257217e-05, | |
| "loss": 0.0004, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 2.376518218623482, | |
| "grad_norm": 0.02757337468182366, | |
| "learning_rate": 2.5171828484929447e-05, | |
| "loss": 0.0002, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 2.380566801619433, | |
| "grad_norm": 0.04317643342120071, | |
| "learning_rate": 2.5117177151798934e-05, | |
| "loss": 0.0003, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 2.3846153846153846, | |
| "grad_norm": 0.04426212931650172, | |
| "learning_rate": 2.506248490185627e-05, | |
| "loss": 0.0004, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 2.388663967611336, | |
| "grad_norm": 0.03214913692754791, | |
| "learning_rate": 2.500775217241923e-05, | |
| "loss": 0.0003, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 2.3927125506072873, | |
| "grad_norm": 0.02939320298646272, | |
| "learning_rate": 2.4952979401129266e-05, | |
| "loss": 0.0002, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 2.396761133603239, | |
| "grad_norm": 0.03137284814727064, | |
| "learning_rate": 2.4898167025948e-05, | |
| "loss": 0.0003, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 2.4008097165991904, | |
| "grad_norm": 0.04694288833471494, | |
| "learning_rate": 2.4843315485153703e-05, | |
| "loss": 0.0004, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 2.4048582995951415, | |
| "grad_norm": 0.028564765752291516, | |
| "learning_rate": 2.4788425217337856e-05, | |
| "loss": 0.0002, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 2.408906882591093, | |
| "grad_norm": 0.04443032650158277, | |
| "learning_rate": 2.4733496661401572e-05, | |
| "loss": 0.0005, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 2.4129554655870447, | |
| "grad_norm": 0.04561769401153744, | |
| "learning_rate": 2.4678530256552114e-05, | |
| "loss": 0.0004, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 2.417004048582996, | |
| "grad_norm": 0.03923410228247591, | |
| "learning_rate": 2.46235264422994e-05, | |
| "loss": 0.0003, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 2.4210526315789473, | |
| "grad_norm": 0.05132138401505904, | |
| "learning_rate": 2.4568485658452452e-05, | |
| "loss": 0.0003, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 2.425101214574899, | |
| "grad_norm": 0.030670173128123857, | |
| "learning_rate": 2.451340834511592e-05, | |
| "loss": 0.0001, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 2.42914979757085, | |
| "grad_norm": 0.034446155516067486, | |
| "learning_rate": 2.4458294942686528e-05, | |
| "loss": 0.0002, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.4331983805668016, | |
| "grad_norm": 0.031892932960726726, | |
| "learning_rate": 2.440314589184958e-05, | |
| "loss": 0.0002, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 2.437246963562753, | |
| "grad_norm": 0.038550305457584456, | |
| "learning_rate": 2.434796163357541e-05, | |
| "loss": 0.0004, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 2.4412955465587043, | |
| "grad_norm": 0.02797696317203818, | |
| "learning_rate": 2.429274260911588e-05, | |
| "loss": 0.0002, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 2.445344129554656, | |
| "grad_norm": 0.03266632376799661, | |
| "learning_rate": 2.423748926000084e-05, | |
| "loss": 0.0002, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 2.4493927125506074, | |
| "grad_norm": 0.03388426954036973, | |
| "learning_rate": 2.418220202803459e-05, | |
| "loss": 0.0004, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 2.4534412955465585, | |
| "grad_norm": 0.015498200011975808, | |
| "learning_rate": 2.412688135529237e-05, | |
| "loss": 0.0001, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 2.45748987854251, | |
| "grad_norm": 0.06380119228551986, | |
| "learning_rate": 2.4071527684116807e-05, | |
| "loss": 0.0001, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 2.4615384615384617, | |
| "grad_norm": 0.04761746792781018, | |
| "learning_rate": 2.4016141457114372e-05, | |
| "loss": 0.0007, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 2.465587044534413, | |
| "grad_norm": 0.01566135036362417, | |
| "learning_rate": 2.396072311715187e-05, | |
| "loss": 0.0001, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 2.4696356275303644, | |
| "grad_norm": 0.024347217237093193, | |
| "learning_rate": 2.3905273107352863e-05, | |
| "loss": 0.0001, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 2.473684210526316, | |
| "grad_norm": 0.017173733214932877, | |
| "learning_rate": 2.3849791871094157e-05, | |
| "loss": 0.0001, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 2.477732793522267, | |
| "grad_norm": 0.03445708679894824, | |
| "learning_rate": 2.379427985200223e-05, | |
| "loss": 0.0002, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 2.4817813765182186, | |
| "grad_norm": 0.06682884090538065, | |
| "learning_rate": 2.373873749394972e-05, | |
| "loss": 0.0005, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 2.48582995951417, | |
| "grad_norm": 0.025450206886728208, | |
| "learning_rate": 2.3683165241051836e-05, | |
| "loss": 0.0001, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 2.4898785425101213, | |
| "grad_norm": 0.022189358131631514, | |
| "learning_rate": 2.3627563537662823e-05, | |
| "loss": 0.0001, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 2.493927125506073, | |
| "grad_norm": 0.029102857644916172, | |
| "learning_rate": 2.3571932828372433e-05, | |
| "loss": 0.0002, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 2.4979757085020244, | |
| "grad_norm": 0.07974397364247016, | |
| "learning_rate": 2.351627355800233e-05, | |
| "loss": 0.0002, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 2.5020242914979756, | |
| "grad_norm": 0.03787249844014864, | |
| "learning_rate": 2.346058617160256e-05, | |
| "loss": 0.0002, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 2.506072874493927, | |
| "grad_norm": 0.025326535686786233, | |
| "learning_rate": 2.3404871114447976e-05, | |
| "loss": 0.0001, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 2.5101214574898787, | |
| "grad_norm": 0.04384904880459098, | |
| "learning_rate": 2.334912883203469e-05, | |
| "loss": 0.0001, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.51417004048583, | |
| "grad_norm": 0.036647141271078626, | |
| "learning_rate": 2.3293359770076505e-05, | |
| "loss": 0.0003, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 2.5182186234817814, | |
| "grad_norm": 0.04243237926804928, | |
| "learning_rate": 2.323756437450136e-05, | |
| "loss": 0.0002, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 2.522267206477733, | |
| "grad_norm": 0.018136070125897893, | |
| "learning_rate": 2.3181743091447754e-05, | |
| "loss": 0.0001, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 2.526315789473684, | |
| "grad_norm": 0.05354341540837024, | |
| "learning_rate": 2.3125896367261164e-05, | |
| "loss": 0.0003, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 2.5303643724696356, | |
| "grad_norm": 0.06438369922368851, | |
| "learning_rate": 2.307002464849052e-05, | |
| "loss": 0.0003, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.534412955465587, | |
| "grad_norm": 0.037832762588638415, | |
| "learning_rate": 2.30141283818846e-05, | |
| "loss": 0.0002, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 2.5384615384615383, | |
| "grad_norm": 0.027305036357867567, | |
| "learning_rate": 2.2958208014388462e-05, | |
| "loss": 0.0002, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 2.54251012145749, | |
| "grad_norm": 0.035983677525743316, | |
| "learning_rate": 2.2902263993139857e-05, | |
| "loss": 0.0001, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 2.5465587044534415, | |
| "grad_norm": 0.07996096307287263, | |
| "learning_rate": 2.2846296765465708e-05, | |
| "loss": 0.001, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 2.5506072874493926, | |
| "grad_norm": 0.03391862603339445, | |
| "learning_rate": 2.2790306778878465e-05, | |
| "loss": 0.0003, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.554655870445344, | |
| "grad_norm": 0.022790905814330516, | |
| "learning_rate": 2.273429448107257e-05, | |
| "loss": 0.0003, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 2.5587044534412957, | |
| "grad_norm": 0.035293117882370696, | |
| "learning_rate": 2.2678260319920868e-05, | |
| "loss": 0.0004, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 2.562753036437247, | |
| "grad_norm": 0.03164502753277674, | |
| "learning_rate": 2.262220474347102e-05, | |
| "loss": 0.0002, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 2.5668016194331984, | |
| "grad_norm": 0.04923157166580477, | |
| "learning_rate": 2.2566128199941918e-05, | |
| "loss": 0.0002, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 2.57085020242915, | |
| "grad_norm": 0.02762873397080906, | |
| "learning_rate": 2.2510031137720104e-05, | |
| "loss": 0.0001, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 2.574898785425101, | |
| "grad_norm": 0.013635675255653658, | |
| "learning_rate": 2.2453914005356206e-05, | |
| "loss": 0.0001, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 2.5789473684210527, | |
| "grad_norm": 0.04641764861425799, | |
| "learning_rate": 2.2397777251561302e-05, | |
| "loss": 0.0004, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 2.582995951417004, | |
| "grad_norm": 0.03635567599914254, | |
| "learning_rate": 2.2341621325203393e-05, | |
| "loss": 0.0002, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 2.5870445344129553, | |
| "grad_norm": 0.019756209492463567, | |
| "learning_rate": 2.2285446675303764e-05, | |
| "loss": 0.0002, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 2.591093117408907, | |
| "grad_norm": 0.031133608707543465, | |
| "learning_rate": 2.2229253751033423e-05, | |
| "loss": 0.0003, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.5951417004048585, | |
| "grad_norm": 0.054306796711965095, | |
| "learning_rate": 2.2173043001709492e-05, | |
| "loss": 0.0004, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 2.5991902834008096, | |
| "grad_norm": 0.05306084789242474, | |
| "learning_rate": 2.2116814876791627e-05, | |
| "loss": 0.0003, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 2.603238866396761, | |
| "grad_norm": 0.026592066195084342, | |
| "learning_rate": 2.2060569825878413e-05, | |
| "loss": 0.0001, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 2.6072874493927127, | |
| "grad_norm": 0.02873279362311861, | |
| "learning_rate": 2.2004308298703777e-05, | |
| "loss": 0.0002, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 2.611336032388664, | |
| "grad_norm": 0.052077268014598686, | |
| "learning_rate": 2.1948030745133396e-05, | |
| "loss": 0.0004, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 2.6153846153846154, | |
| "grad_norm": 0.03798721495780667, | |
| "learning_rate": 2.1891737615161083e-05, | |
| "loss": 0.0004, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 2.619433198380567, | |
| "grad_norm": 0.014041858340527917, | |
| "learning_rate": 2.1835429358905218e-05, | |
| "loss": 0.0001, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 2.623481781376518, | |
| "grad_norm": 0.016565432472066075, | |
| "learning_rate": 2.17791064266051e-05, | |
| "loss": 0.0001, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 2.6275303643724697, | |
| "grad_norm": 0.01552675169794235, | |
| "learning_rate": 2.172276926861741e-05, | |
| "loss": 0.0001, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 2.6315789473684212, | |
| "grad_norm": 0.021781000102431353, | |
| "learning_rate": 2.166641833541255e-05, | |
| "loss": 0.0001, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.6356275303643724, | |
| "grad_norm": 0.06700925105605687, | |
| "learning_rate": 2.161005407757109e-05, | |
| "loss": 0.0008, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 2.639676113360324, | |
| "grad_norm": 0.027623471871792576, | |
| "learning_rate": 2.155367694578013e-05, | |
| "loss": 0.0002, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 2.6437246963562755, | |
| "grad_norm": 0.03472702333399782, | |
| "learning_rate": 2.1497287390829712e-05, | |
| "loss": 0.0002, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 2.6477732793522266, | |
| "grad_norm": 0.036801037255470716, | |
| "learning_rate": 2.1440885863609208e-05, | |
| "loss": 0.0001, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 2.651821862348178, | |
| "grad_norm": 0.029340377087649524, | |
| "learning_rate": 2.1384472815103737e-05, | |
| "loss": 0.0002, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 2.6558704453441297, | |
| "grad_norm": 0.024170627049205137, | |
| "learning_rate": 2.1328048696390534e-05, | |
| "loss": 0.0001, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 2.659919028340081, | |
| "grad_norm": 0.036208921840455106, | |
| "learning_rate": 2.1271613958635342e-05, | |
| "loss": 0.0001, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 2.6639676113360324, | |
| "grad_norm": 0.04195359551501685, | |
| "learning_rate": 2.1215169053088825e-05, | |
| "loss": 0.0004, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 2.668016194331984, | |
| "grad_norm": 0.04619744129258447, | |
| "learning_rate": 2.1158714431082937e-05, | |
| "loss": 0.0002, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 2.672064777327935, | |
| "grad_norm": 0.01823371948518699, | |
| "learning_rate": 2.110225054402733e-05, | |
| "loss": 0.0001, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.6761133603238867, | |
| "grad_norm": 0.040591092636746366, | |
| "learning_rate": 2.1045777843405747e-05, | |
| "loss": 0.0005, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 2.6801619433198383, | |
| "grad_norm": 0.02545601352009389, | |
| "learning_rate": 2.09892967807724e-05, | |
| "loss": 0.0002, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 2.6842105263157894, | |
| "grad_norm": 0.02790629814311321, | |
| "learning_rate": 2.093280780774835e-05, | |
| "loss": 0.0002, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 2.688259109311741, | |
| "grad_norm": 0.03875008850086678, | |
| "learning_rate": 2.087631137601793e-05, | |
| "loss": 0.0003, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 2.6923076923076925, | |
| "grad_norm": 0.03676378060278413, | |
| "learning_rate": 2.0819807937325104e-05, | |
| "loss": 0.0001, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 2.6963562753036436, | |
| "grad_norm": 0.030082118337670456, | |
| "learning_rate": 2.0763297943469847e-05, | |
| "loss": 0.0003, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 2.700404858299595, | |
| "grad_norm": 0.028986004034645328, | |
| "learning_rate": 2.070678184630458e-05, | |
| "loss": 0.0002, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 2.7044534412955468, | |
| "grad_norm": 0.018976610926731998, | |
| "learning_rate": 2.06502600977305e-05, | |
| "loss": 0.0001, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 2.708502024291498, | |
| "grad_norm": 0.00868773823103107, | |
| "learning_rate": 2.0593733149694008e-05, | |
| "loss": 0.0001, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 2.7125506072874495, | |
| "grad_norm": 0.004760393565187522, | |
| "learning_rate": 2.0537201454183074e-05, | |
| "loss": 0.0001, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.716599190283401, | |
| "grad_norm": 0.04209269680064897, | |
| "learning_rate": 2.0480665463223626e-05, | |
| "loss": 0.0005, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 2.720647773279352, | |
| "grad_norm": 0.021964840786774333, | |
| "learning_rate": 2.042412562887594e-05, | |
| "loss": 0.0001, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 2.7246963562753037, | |
| "grad_norm": 0.01727968995222453, | |
| "learning_rate": 2.0367582403231033e-05, | |
| "loss": 0.0001, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 2.7287449392712553, | |
| "grad_norm": 0.03442217131316174, | |
| "learning_rate": 2.0311036238407023e-05, | |
| "loss": 0.0003, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 2.7327935222672064, | |
| "grad_norm": 0.014595432787514948, | |
| "learning_rate": 2.0254487586545534e-05, | |
| "loss": 0.0001, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.736842105263158, | |
| "grad_norm": 0.015869234147492327, | |
| "learning_rate": 2.0197936899808088e-05, | |
| "loss": 0.0002, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 2.7408906882591095, | |
| "grad_norm": 0.02995306358344833, | |
| "learning_rate": 2.0141384630372466e-05, | |
| "loss": 0.0002, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 2.7449392712550607, | |
| "grad_norm": 0.025841656946913795, | |
| "learning_rate": 2.0084831230429106e-05, | |
| "loss": 0.0001, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 2.748987854251012, | |
| "grad_norm": 0.04183114919346923, | |
| "learning_rate": 2.0028277152177492e-05, | |
| "loss": 0.0003, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 2.753036437246964, | |
| "grad_norm": 0.020025641719375556, | |
| "learning_rate": 1.9971722847822518e-05, | |
| "loss": 0.0001, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.757085020242915, | |
| "grad_norm": 0.015061494358865868, | |
| "learning_rate": 1.99151687695709e-05, | |
| "loss": 0.0001, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 2.7611336032388665, | |
| "grad_norm": 0.005992650668186758, | |
| "learning_rate": 1.9858615369627537e-05, | |
| "loss": 0.0001, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 2.765182186234818, | |
| "grad_norm": 0.020550686005761768, | |
| "learning_rate": 1.9802063100191916e-05, | |
| "loss": 0.0002, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 2.769230769230769, | |
| "grad_norm": 0.016563839715013576, | |
| "learning_rate": 1.974551241345447e-05, | |
| "loss": 0.0003, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 2.7732793522267207, | |
| "grad_norm": 0.011408268200807975, | |
| "learning_rate": 1.9688963761592984e-05, | |
| "loss": 0.0001, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 2.7773279352226723, | |
| "grad_norm": 0.004134143294632397, | |
| "learning_rate": 1.963241759676897e-05, | |
| "loss": 0.0001, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 2.7813765182186234, | |
| "grad_norm": 0.005612940226143198, | |
| "learning_rate": 1.9575874371124066e-05, | |
| "loss": 0.0001, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 2.785425101214575, | |
| "grad_norm": 0.009197004072499362, | |
| "learning_rate": 1.951933453677638e-05, | |
| "loss": 0.0001, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 2.7894736842105265, | |
| "grad_norm": 0.012340539773013614, | |
| "learning_rate": 1.9462798545816932e-05, | |
| "loss": 0.0001, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 2.7935222672064777, | |
| "grad_norm": 0.006357159856869539, | |
| "learning_rate": 1.9406266850305996e-05, | |
| "loss": 0.0001, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.7975708502024292, | |
| "grad_norm": 0.03793377422449579, | |
| "learning_rate": 1.9349739902269503e-05, | |
| "loss": 0.0002, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 2.801619433198381, | |
| "grad_norm": 0.06448943951026576, | |
| "learning_rate": 1.9293218153695425e-05, | |
| "loss": 0.0004, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 2.805668016194332, | |
| "grad_norm": 0.013776500643993436, | |
| "learning_rate": 1.9236702056530153e-05, | |
| "loss": 0.0001, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 2.8097165991902835, | |
| "grad_norm": 0.023544758648206033, | |
| "learning_rate": 1.9180192062674907e-05, | |
| "loss": 0.0002, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 2.813765182186235, | |
| "grad_norm": 0.0038092514438469068, | |
| "learning_rate": 1.9123688623982076e-05, | |
| "loss": 0.0, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 2.817813765182186, | |
| "grad_norm": 0.00365088965976741, | |
| "learning_rate": 1.906719219225165e-05, | |
| "loss": 0.0, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 2.8218623481781377, | |
| "grad_norm": 0.0016001437562713227, | |
| "learning_rate": 1.9010703219227603e-05, | |
| "loss": 0.0001, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 2.8259109311740893, | |
| "grad_norm": 0.013484690976870235, | |
| "learning_rate": 1.8954222156594256e-05, | |
| "loss": 0.0001, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 2.8299595141700404, | |
| "grad_norm": 0.022132345664147752, | |
| "learning_rate": 1.8897749455972673e-05, | |
| "loss": 0.0001, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 2.834008097165992, | |
| "grad_norm": 0.0030919638761904714, | |
| "learning_rate": 1.884128556891707e-05, | |
| "loss": 0.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.8380566801619436, | |
| "grad_norm": 0.018630431822574515, | |
| "learning_rate": 1.878483094691119e-05, | |
| "loss": 0.0001, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 2.8421052631578947, | |
| "grad_norm": 0.004964220914335516, | |
| "learning_rate": 1.8728386041364664e-05, | |
| "loss": 0.0001, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 2.8461538461538463, | |
| "grad_norm": 0.0022260715650253685, | |
| "learning_rate": 1.867195130360947e-05, | |
| "loss": 0.0, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 2.850202429149798, | |
| "grad_norm": 0.01580630838635798, | |
| "learning_rate": 1.8615527184896263e-05, | |
| "loss": 0.0001, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 2.854251012145749, | |
| "grad_norm": 0.007929730698984212, | |
| "learning_rate": 1.8559114136390795e-05, | |
| "loss": 0.0001, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 2.8582995951417005, | |
| "grad_norm": 0.0033907481506477797, | |
| "learning_rate": 1.8502712609170298e-05, | |
| "loss": 0.0, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 2.862348178137652, | |
| "grad_norm": 0.0015706615561684775, | |
| "learning_rate": 1.8446323054219876e-05, | |
| "loss": 0.0, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 2.866396761133603, | |
| "grad_norm": 0.011145432723268324, | |
| "learning_rate": 1.838994592242891e-05, | |
| "loss": 0.0001, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 2.8704453441295548, | |
| "grad_norm": 0.0016622988963215619, | |
| "learning_rate": 1.8333581664587453e-05, | |
| "loss": 0.0, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 2.8744939271255063, | |
| "grad_norm": 0.0011736682704659442, | |
| "learning_rate": 1.8277230731382593e-05, | |
| "loss": 0.0, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.8785425101214575, | |
| "grad_norm": 0.023744798441762062, | |
| "learning_rate": 1.82208935733949e-05, | |
| "loss": 0.0002, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 2.882591093117409, | |
| "grad_norm": 0.0013517303300732365, | |
| "learning_rate": 1.8164570641094793e-05, | |
| "loss": 0.0, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 2.8866396761133606, | |
| "grad_norm": 0.013360356689729912, | |
| "learning_rate": 1.810826238483892e-05, | |
| "loss": 0.0001, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 2.8906882591093117, | |
| "grad_norm": 0.04309744241054798, | |
| "learning_rate": 1.8051969254866608e-05, | |
| "loss": 0.0001, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 2.8947368421052633, | |
| "grad_norm": 0.0007850155677844528, | |
| "learning_rate": 1.7995691701296226e-05, | |
| "loss": 0.0, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 2.898785425101215, | |
| "grad_norm": 0.0050617188702010985, | |
| "learning_rate": 1.7939430174121593e-05, | |
| "loss": 0.0, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 2.902834008097166, | |
| "grad_norm": 0.011811384099315222, | |
| "learning_rate": 1.788318512320838e-05, | |
| "loss": 0.0001, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 2.9068825910931175, | |
| "grad_norm": 0.0007904869517553299, | |
| "learning_rate": 1.782695699829051e-05, | |
| "loss": 0.0, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 2.910931174089069, | |
| "grad_norm": 0.02615354858447353, | |
| "learning_rate": 1.7770746248966587e-05, | |
| "loss": 0.0001, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 2.91497975708502, | |
| "grad_norm": 0.0007240807658345232, | |
| "learning_rate": 1.7714553324696243e-05, | |
| "loss": 0.0, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.919028340080972, | |
| "grad_norm": 0.0007534303648024434, | |
| "learning_rate": 1.7658378674796614e-05, | |
| "loss": 0.0, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 2.9230769230769234, | |
| "grad_norm": 0.0007298382607335399, | |
| "learning_rate": 1.76022227484387e-05, | |
| "loss": 0.0, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 2.9271255060728745, | |
| "grad_norm": 0.026735269508733655, | |
| "learning_rate": 1.7546085994643807e-05, | |
| "loss": 0.0002, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 2.931174089068826, | |
| "grad_norm": 0.001203163397772205, | |
| "learning_rate": 1.7489968862279902e-05, | |
| "loss": 0.0, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 2.9352226720647776, | |
| "grad_norm": 0.005372325198098058, | |
| "learning_rate": 1.743387180005809e-05, | |
| "loss": 0.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.9392712550607287, | |
| "grad_norm": 0.010010780951090942, | |
| "learning_rate": 1.737779525652899e-05, | |
| "loss": 0.0, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 2.9433198380566803, | |
| "grad_norm": 0.0038006702658664577, | |
| "learning_rate": 1.732173968007914e-05, | |
| "loss": 0.0, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 2.9473684210526314, | |
| "grad_norm": 0.0006922746120441002, | |
| "learning_rate": 1.7265705518927437e-05, | |
| "loss": 0.0, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 2.951417004048583, | |
| "grad_norm": 0.02684765844676431, | |
| "learning_rate": 1.7209693221121542e-05, | |
| "loss": 0.0001, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 2.9554655870445345, | |
| "grad_norm": 0.000906333052219522, | |
| "learning_rate": 1.7153703234534302e-05, | |
| "loss": 0.0, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.9595141700404857, | |
| "grad_norm": 0.0008701630495458088, | |
| "learning_rate": 1.709773600686015e-05, | |
| "loss": 0.0, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 2.9635627530364372, | |
| "grad_norm": 0.0007820783353298642, | |
| "learning_rate": 1.7041791985611548e-05, | |
| "loss": 0.0, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 2.967611336032389, | |
| "grad_norm": 0.00861244690686876, | |
| "learning_rate": 1.69858716181154e-05, | |
| "loss": 0.0, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 2.97165991902834, | |
| "grad_norm": 0.018262781131575246, | |
| "learning_rate": 1.692997535150948e-05, | |
| "loss": 0.0001, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 2.9757085020242915, | |
| "grad_norm": 0.008262732077901855, | |
| "learning_rate": 1.687410363273884e-05, | |
| "loss": 0.0001, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 2.979757085020243, | |
| "grad_norm": 0.000647108775000297, | |
| "learning_rate": 1.6818256908552257e-05, | |
| "loss": 0.0, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 2.983805668016194, | |
| "grad_norm": 0.0012427306029925936, | |
| "learning_rate": 1.6762435625498646e-05, | |
| "loss": 0.0, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 2.9878542510121457, | |
| "grad_norm": 0.0014335233992652848, | |
| "learning_rate": 1.67066402299235e-05, | |
| "loss": 0.0, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 2.9919028340080973, | |
| "grad_norm": 0.0008285430877394383, | |
| "learning_rate": 1.6650871167965313e-05, | |
| "loss": 0.0, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 2.9959514170040484, | |
| "grad_norm": 0.0007149941496498698, | |
| "learning_rate": 1.6595128885552028e-05, | |
| "loss": 0.0, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 0.005480385470951294, | |
| "learning_rate": 1.653941382839745e-05, | |
| "loss": 0.0, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 3.0040485829959516, | |
| "grad_norm": 0.0007249226324497209, | |
| "learning_rate": 1.6483726441997673e-05, | |
| "loss": 0.0, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 3.0080971659919027, | |
| "grad_norm": 0.0005946687691567566, | |
| "learning_rate": 1.642806717162757e-05, | |
| "loss": 0.0, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 3.0121457489878543, | |
| "grad_norm": 0.0006299084362629527, | |
| "learning_rate": 1.637243646233718e-05, | |
| "loss": 0.0, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 3.016194331983806, | |
| "grad_norm": 0.00052368332321913, | |
| "learning_rate": 1.6316834758948174e-05, | |
| "loss": 0.0, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 3.020242914979757, | |
| "grad_norm": 0.0007647005121836242, | |
| "learning_rate": 1.6261262506050282e-05, | |
| "loss": 0.0, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 3.0242914979757085, | |
| "grad_norm": 0.0007387662459894115, | |
| "learning_rate": 1.620572014799777e-05, | |
| "loss": 0.0, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 3.02834008097166, | |
| "grad_norm": 0.0004980036374615334, | |
| "learning_rate": 1.6150208128905857e-05, | |
| "loss": 0.0, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 3.032388663967611, | |
| "grad_norm": 0.0008462041724996156, | |
| "learning_rate": 1.6094726892647147e-05, | |
| "loss": 0.0, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 3.0364372469635628, | |
| "grad_norm": 0.0006398654161285608, | |
| "learning_rate": 1.6039276882848135e-05, | |
| "loss": 0.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.0404858299595143, | |
| "grad_norm": 0.0017583500722881383, | |
| "learning_rate": 1.5983858542885635e-05, | |
| "loss": 0.0, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 3.0445344129554655, | |
| "grad_norm": 0.0006424298794269876, | |
| "learning_rate": 1.5928472315883203e-05, | |
| "loss": 0.0, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 3.048582995951417, | |
| "grad_norm": 0.0005323499130128915, | |
| "learning_rate": 1.5873118644707633e-05, | |
| "loss": 0.0, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 3.0526315789473686, | |
| "grad_norm": 0.0004501849858203386, | |
| "learning_rate": 1.5817797971965413e-05, | |
| "loss": 0.0, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 3.0566801619433197, | |
| "grad_norm": 0.0004898663151333657, | |
| "learning_rate": 1.576251073999917e-05, | |
| "loss": 0.0, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 3.0607287449392713, | |
| "grad_norm": 0.00046790148258959774, | |
| "learning_rate": 1.5707257390884126e-05, | |
| "loss": 0.0, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 3.064777327935223, | |
| "grad_norm": 0.0004884145162990336, | |
| "learning_rate": 1.5652038366424595e-05, | |
| "loss": 0.0, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 3.068825910931174, | |
| "grad_norm": 0.00042051809473610746, | |
| "learning_rate": 1.5596854108150424e-05, | |
| "loss": 0.0, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 3.0728744939271255, | |
| "grad_norm": 0.00043651493648579076, | |
| "learning_rate": 1.5541705057313476e-05, | |
| "loss": 0.0, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 3.076923076923077, | |
| "grad_norm": 0.014182613140366106, | |
| "learning_rate": 1.5486591654884086e-05, | |
| "loss": 0.0, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 3.080971659919028, | |
| "grad_norm": 0.00045166228322483554, | |
| "learning_rate": 1.543151434154755e-05, | |
| "loss": 0.0, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 3.08502024291498, | |
| "grad_norm": 0.00043408103390201445, | |
| "learning_rate": 1.5376473557700615e-05, | |
| "loss": 0.0, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 3.0890688259109313, | |
| "grad_norm": 0.00043048703385947936, | |
| "learning_rate": 1.5321469743447893e-05, | |
| "loss": 0.0, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 3.0931174089068825, | |
| "grad_norm": 0.00043974238644585873, | |
| "learning_rate": 1.5266503338598434e-05, | |
| "loss": 0.0, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 3.097165991902834, | |
| "grad_norm": 0.00045818290486930585, | |
| "learning_rate": 1.5211574782662149e-05, | |
| "loss": 0.0, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 3.1012145748987856, | |
| "grad_norm": 0.0004219740997697336, | |
| "learning_rate": 1.5156684514846307e-05, | |
| "loss": 0.0, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 3.1052631578947367, | |
| "grad_norm": 0.0004124915418187473, | |
| "learning_rate": 1.5101832974052012e-05, | |
| "loss": 0.0, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 3.1093117408906883, | |
| "grad_norm": 0.00039461542326133106, | |
| "learning_rate": 1.5047020598870737e-05, | |
| "loss": 0.0, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 3.11336032388664, | |
| "grad_norm": 0.0003975858789137352, | |
| "learning_rate": 1.4992247827580778e-05, | |
| "loss": 0.0, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 3.117408906882591, | |
| "grad_norm": 0.00041406456962947847, | |
| "learning_rate": 1.493751509814374e-05, | |
| "loss": 0.0, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 3.1214574898785425, | |
| "grad_norm": 0.000408761688119374, | |
| "learning_rate": 1.4882822848201073e-05, | |
| "loss": 0.0, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 3.125506072874494, | |
| "grad_norm": 0.00044203182964878057, | |
| "learning_rate": 1.4828171515070553e-05, | |
| "loss": 0.0, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 3.1295546558704452, | |
| "grad_norm": 0.0003838992379207088, | |
| "learning_rate": 1.4773561535742793e-05, | |
| "loss": 0.0, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 3.133603238866397, | |
| "grad_norm": 0.0004040069401875941, | |
| "learning_rate": 1.4718993346877715e-05, | |
| "loss": 0.0, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 3.1376518218623484, | |
| "grad_norm": 0.00037666220247148553, | |
| "learning_rate": 1.466446738480111e-05, | |
| "loss": 0.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.1417004048582995, | |
| "grad_norm": 0.0003973714610289173, | |
| "learning_rate": 1.460998408550112e-05, | |
| "loss": 0.0, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 3.145748987854251, | |
| "grad_norm": 0.0003823065377978282, | |
| "learning_rate": 1.4555543884624751e-05, | |
| "loss": 0.0, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 3.1497975708502026, | |
| "grad_norm": 0.0003938152821369476, | |
| "learning_rate": 1.4501147217474402e-05, | |
| "loss": 0.0, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 3.1538461538461537, | |
| "grad_norm": 0.0006326194431148889, | |
| "learning_rate": 1.444679451900437e-05, | |
| "loss": 0.0, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 3.1578947368421053, | |
| "grad_norm": 0.0003975860094595032, | |
| "learning_rate": 1.4392486223817397e-05, | |
| "loss": 0.0, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 3.161943319838057, | |
| "grad_norm": 0.0003707854204401172, | |
| "learning_rate": 1.4338222766161143e-05, | |
| "loss": 0.0, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 3.165991902834008, | |
| "grad_norm": 0.0003780403722256462, | |
| "learning_rate": 1.4284004579924777e-05, | |
| "loss": 0.0, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 3.1700404858299596, | |
| "grad_norm": 0.0003807524900611682, | |
| "learning_rate": 1.4229832098635479e-05, | |
| "loss": 0.0, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 3.174089068825911, | |
| "grad_norm": 0.00037486971293124226, | |
| "learning_rate": 1.4175705755454963e-05, | |
| "loss": 0.0, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 3.1781376518218623, | |
| "grad_norm": 0.00037153579489968724, | |
| "learning_rate": 1.412162598317602e-05, | |
| "loss": 0.0, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 3.182186234817814, | |
| "grad_norm": 0.00036902448745282823, | |
| "learning_rate": 1.406759321421907e-05, | |
| "loss": 0.0, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 3.1862348178137654, | |
| "grad_norm": 0.00034910581823528806, | |
| "learning_rate": 1.401360788062871e-05, | |
| "loss": 0.0, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 3.1902834008097165, | |
| "grad_norm": 0.00036712546083107484, | |
| "learning_rate": 1.3959670414070204e-05, | |
| "loss": 0.0, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 3.194331983805668, | |
| "grad_norm": 0.0003560555075468064, | |
| "learning_rate": 1.3905781245826108e-05, | |
| "loss": 0.0, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 3.1983805668016196, | |
| "grad_norm": 0.0019014802713063802, | |
| "learning_rate": 1.3851940806792778e-05, | |
| "loss": 0.0, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 3.2024291497975708, | |
| "grad_norm": 0.00035530082868505245, | |
| "learning_rate": 1.379814952747693e-05, | |
| "loss": 0.0, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 3.2064777327935223, | |
| "grad_norm": 0.0003624261963106265, | |
| "learning_rate": 1.3744407837992193e-05, | |
| "loss": 0.0, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 3.2105263157894735, | |
| "grad_norm": 0.000356496435757262, | |
| "learning_rate": 1.3690716168055692e-05, | |
| "loss": 0.0, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 3.214574898785425, | |
| "grad_norm": 0.0003365725073684527, | |
| "learning_rate": 1.363707494698459e-05, | |
| "loss": 0.0, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 3.2186234817813766, | |
| "grad_norm": 0.00035313210863880323, | |
| "learning_rate": 1.358348460369265e-05, | |
| "loss": 0.0, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 3.2226720647773277, | |
| "grad_norm": 0.0003660532947687084, | |
| "learning_rate": 1.352994556668684e-05, | |
| "loss": 0.0, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 3.2267206477732793, | |
| "grad_norm": 0.0003538874934732405, | |
| "learning_rate": 1.3476458264063865e-05, | |
| "loss": 0.0, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 3.230769230769231, | |
| "grad_norm": 0.00034694098088413646, | |
| "learning_rate": 1.342302312350678e-05, | |
| "loss": 0.0, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 3.234817813765182, | |
| "grad_norm": 0.0003580857861588643, | |
| "learning_rate": 1.3369640572281537e-05, | |
| "loss": 0.0, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 3.2388663967611335, | |
| "grad_norm": 0.00033457511624845134, | |
| "learning_rate": 1.3316311037233596e-05, | |
| "loss": 0.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.242914979757085, | |
| "grad_norm": 0.0004221705746633671, | |
| "learning_rate": 1.326303494478451e-05, | |
| "loss": 0.0, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 3.246963562753036, | |
| "grad_norm": 0.0003784997991148486, | |
| "learning_rate": 1.3209812720928472e-05, | |
| "loss": 0.0, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 3.251012145748988, | |
| "grad_norm": 0.0003399244634889871, | |
| "learning_rate": 1.315664479122898e-05, | |
| "loss": 0.0, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 3.2550607287449393, | |
| "grad_norm": 0.00032515805334787194, | |
| "learning_rate": 1.3103531580815378e-05, | |
| "loss": 0.0, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 3.2591093117408905, | |
| "grad_norm": 0.00034605995352755554, | |
| "learning_rate": 1.3050473514379488e-05, | |
| "loss": 0.0, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 3.263157894736842, | |
| "grad_norm": 0.00033211865764783843, | |
| "learning_rate": 1.2997471016172188e-05, | |
| "loss": 0.0, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 3.2672064777327936, | |
| "grad_norm": 0.00032269524675691683, | |
| "learning_rate": 1.2944524510000042e-05, | |
| "loss": 0.0, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 3.2712550607287447, | |
| "grad_norm": 0.000328044944508132, | |
| "learning_rate": 1.2891634419221908e-05, | |
| "loss": 0.0, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 3.2753036437246963, | |
| "grad_norm": 0.0003755241085454023, | |
| "learning_rate": 1.2838801166745545e-05, | |
| "loss": 0.0, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 3.279352226720648, | |
| "grad_norm": 0.0003377276339405059, | |
| "learning_rate": 1.2786025175024216e-05, | |
| "loss": 0.0, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 3.283400809716599, | |
| "grad_norm": 0.0003287347223236657, | |
| "learning_rate": 1.2733306866053357e-05, | |
| "loss": 0.0, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 3.2874493927125505, | |
| "grad_norm": 0.0003452994883341673, | |
| "learning_rate": 1.2680646661367163e-05, | |
| "loss": 0.0, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 3.291497975708502, | |
| "grad_norm": 0.00035422413477002624, | |
| "learning_rate": 1.2628044982035219e-05, | |
| "loss": 0.0, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 3.2955465587044532, | |
| "grad_norm": 0.0003343397795237015, | |
| "learning_rate": 1.2575502248659155e-05, | |
| "loss": 0.0, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 3.299595141700405, | |
| "grad_norm": 0.00031042122362079003, | |
| "learning_rate": 1.2523018881369269e-05, | |
| "loss": 0.0, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 3.3036437246963564, | |
| "grad_norm": 0.0003366199955691758, | |
| "learning_rate": 1.2470595299821176e-05, | |
| "loss": 0.0, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 3.3076923076923075, | |
| "grad_norm": 0.00032486244972844153, | |
| "learning_rate": 1.2418231923192416e-05, | |
| "loss": 0.0, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 3.311740890688259, | |
| "grad_norm": 0.0003451271110974879, | |
| "learning_rate": 1.2365929170179165e-05, | |
| "loss": 0.0, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 3.3157894736842106, | |
| "grad_norm": 0.00032284244051969103, | |
| "learning_rate": 1.2313687458992848e-05, | |
| "loss": 0.0, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 3.3198380566801617, | |
| "grad_norm": 0.0003064095690335378, | |
| "learning_rate": 1.2261507207356785e-05, | |
| "loss": 0.0, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 3.3238866396761133, | |
| "grad_norm": 0.00032558460233846927, | |
| "learning_rate": 1.220938883250288e-05, | |
| "loss": 0.0, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 3.327935222672065, | |
| "grad_norm": 0.0003216924086717267, | |
| "learning_rate": 1.215733275116827e-05, | |
| "loss": 0.0, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 3.331983805668016, | |
| "grad_norm": 0.0003190066231467066, | |
| "learning_rate": 1.2105339379592005e-05, | |
| "loss": 0.0, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 3.3360323886639676, | |
| "grad_norm": 0.0003096915996504782, | |
| "learning_rate": 1.2053409133511681e-05, | |
| "loss": 0.0, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 3.340080971659919, | |
| "grad_norm": 0.0003064725515636213, | |
| "learning_rate": 1.2001542428160173e-05, | |
| "loss": 0.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.3441295546558703, | |
| "grad_norm": 0.00032361964354269497, | |
| "learning_rate": 1.1949739678262278e-05, | |
| "loss": 0.0, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 3.348178137651822, | |
| "grad_norm": 0.0003118903705964117, | |
| "learning_rate": 1.189800129803139e-05, | |
| "loss": 0.0, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 3.3522267206477734, | |
| "grad_norm": 0.00031650142483999024, | |
| "learning_rate": 1.1846327701166228e-05, | |
| "loss": 0.0, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 3.3562753036437245, | |
| "grad_norm": 0.0003104684592430653, | |
| "learning_rate": 1.1794719300847493e-05, | |
| "loss": 0.0, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 3.360323886639676, | |
| "grad_norm": 0.00032180371196267556, | |
| "learning_rate": 1.1743176509734582e-05, | |
| "loss": 0.0, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 3.3643724696356276, | |
| "grad_norm": 0.00031592578632985024, | |
| "learning_rate": 1.1691699739962275e-05, | |
| "loss": 0.0, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 3.3684210526315788, | |
| "grad_norm": 0.0003180661478649692, | |
| "learning_rate": 1.1640289403137456e-05, | |
| "loss": 0.0, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 3.3724696356275303, | |
| "grad_norm": 0.00030865034495492497, | |
| "learning_rate": 1.1588945910335802e-05, | |
| "loss": 0.0, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 3.376518218623482, | |
| "grad_norm": 0.00030197184849063577, | |
| "learning_rate": 1.1537669672098517e-05, | |
| "loss": 0.0, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 3.380566801619433, | |
| "grad_norm": 0.00029862406891086794, | |
| "learning_rate": 1.1486461098429036e-05, | |
| "loss": 0.0, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 3.3846153846153846, | |
| "grad_norm": 0.00029014049305374457, | |
| "learning_rate": 1.1435320598789747e-05, | |
| "loss": 0.0, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 3.388663967611336, | |
| "grad_norm": 0.00030502248443264125, | |
| "learning_rate": 1.1384248582098736e-05, | |
| "loss": 0.0, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 3.3927125506072873, | |
| "grad_norm": 0.00030415191737143826, | |
| "learning_rate": 1.1333245456726467e-05, | |
| "loss": 0.0, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 3.396761133603239, | |
| "grad_norm": 0.0003065490539937491, | |
| "learning_rate": 1.128231163049258e-05, | |
| "loss": 0.0, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 3.4008097165991904, | |
| "grad_norm": 0.0003048472646177889, | |
| "learning_rate": 1.123144751066259e-05, | |
| "loss": 0.0, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 3.4048582995951415, | |
| "grad_norm": 0.0003119703035425097, | |
| "learning_rate": 1.1180653503944657e-05, | |
| "loss": 0.0, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 3.408906882591093, | |
| "grad_norm": 0.0003311706360416777, | |
| "learning_rate": 1.1129930016486287e-05, | |
| "loss": 0.0, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 3.4129554655870447, | |
| "grad_norm": 0.00031102874557609487, | |
| "learning_rate": 1.107927745387114e-05, | |
| "loss": 0.0, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 3.417004048582996, | |
| "grad_norm": 0.00029014428709643586, | |
| "learning_rate": 1.1028696221115766e-05, | |
| "loss": 0.0, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 3.4210526315789473, | |
| "grad_norm": 0.0002979895022044208, | |
| "learning_rate": 1.0978186722666341e-05, | |
| "loss": 0.0, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 3.425101214574899, | |
| "grad_norm": 0.00029323314679978403, | |
| "learning_rate": 1.0927749362395475e-05, | |
| "loss": 0.0, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 3.42914979757085, | |
| "grad_norm": 0.0002869036539141608, | |
| "learning_rate": 1.0877384543598954e-05, | |
| "loss": 0.0, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 3.4331983805668016, | |
| "grad_norm": 0.00029978267776798676, | |
| "learning_rate": 1.0827092668992529e-05, | |
| "loss": 0.0, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 3.437246963562753, | |
| "grad_norm": 0.00030312273270743467, | |
| "learning_rate": 1.077687414070869e-05, | |
| "loss": 0.0, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 3.4412955465587043, | |
| "grad_norm": 0.00030567009465891056, | |
| "learning_rate": 1.0726729360293451e-05, | |
| "loss": 0.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.445344129554656, | |
| "grad_norm": 0.0002927202253657367, | |
| "learning_rate": 1.0676658728703147e-05, | |
| "loss": 0.0, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 3.4493927125506074, | |
| "grad_norm": 0.0002845213414723303, | |
| "learning_rate": 1.0626662646301198e-05, | |
| "loss": 0.0, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 3.4534412955465585, | |
| "grad_norm": 0.0002959075359405662, | |
| "learning_rate": 1.0576741512854958e-05, | |
| "loss": 0.0, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 3.45748987854251, | |
| "grad_norm": 0.00030301287775213553, | |
| "learning_rate": 1.052689572753247e-05, | |
| "loss": 0.0, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 3.4615384615384617, | |
| "grad_norm": 0.0002943927458299948, | |
| "learning_rate": 1.0477125688899328e-05, | |
| "loss": 0.0, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 3.465587044534413, | |
| "grad_norm": 0.0002867512827029084, | |
| "learning_rate": 1.0427431794915406e-05, | |
| "loss": 0.0, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 3.4696356275303644, | |
| "grad_norm": 0.00029641290420654874, | |
| "learning_rate": 1.0377814442931769e-05, | |
| "loss": 0.0, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 3.473684210526316, | |
| "grad_norm": 0.0002947810729409546, | |
| "learning_rate": 1.0328274029687439e-05, | |
| "loss": 0.0, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 3.477732793522267, | |
| "grad_norm": 0.00027913363279908146, | |
| "learning_rate": 1.0278810951306253e-05, | |
| "loss": 0.0, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 3.4817813765182186, | |
| "grad_norm": 0.00029621524249251506, | |
| "learning_rate": 1.0229425603293647e-05, | |
| "loss": 0.0, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 3.48582995951417, | |
| "grad_norm": 0.00028929181902963595, | |
| "learning_rate": 1.0180118380533558e-05, | |
| "loss": 0.0, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 3.4898785425101213, | |
| "grad_norm": 0.0002836170336034989, | |
| "learning_rate": 1.0130889677285225e-05, | |
| "loss": 0.0, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 3.493927125506073, | |
| "grad_norm": 0.0002901232835146565, | |
| "learning_rate": 1.008173988718005e-05, | |
| "loss": 0.0, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 3.4979757085020244, | |
| "grad_norm": 0.0002853295870957905, | |
| "learning_rate": 1.0032669403218439e-05, | |
| "loss": 0.0, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 3.5020242914979756, | |
| "grad_norm": 0.00027865492901904684, | |
| "learning_rate": 9.983678617766677e-06, | |
| "loss": 0.0, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 3.506072874493927, | |
| "grad_norm": 0.00029003436125958794, | |
| "learning_rate": 9.934767922553788e-06, | |
| "loss": 0.0, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 3.5101214574898787, | |
| "grad_norm": 0.0002884173303665325, | |
| "learning_rate": 9.885937708668371e-06, | |
| "loss": 0.0, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 3.51417004048583, | |
| "grad_norm": 0.00029914740954744683, | |
| "learning_rate": 9.837188366555524e-06, | |
| "loss": 0.0, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 3.5182186234817814, | |
| "grad_norm": 0.0002839788063740539, | |
| "learning_rate": 9.788520286013691e-06, | |
| "loss": 0.0, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 3.522267206477733, | |
| "grad_norm": 0.00027801286052142127, | |
| "learning_rate": 9.739933856191537e-06, | |
| "loss": 0.0, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 3.526315789473684, | |
| "grad_norm": 0.0002841871514600673, | |
| "learning_rate": 9.691429465584866e-06, | |
| "loss": 0.0, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 3.5303643724696356, | |
| "grad_norm": 0.00028639214188676406, | |
| "learning_rate": 9.643007502033492e-06, | |
| "loss": 0.0, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 3.534412955465587, | |
| "grad_norm": 0.00027848856202697907, | |
| "learning_rate": 9.594668352718152e-06, | |
| "loss": 0.0, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 3.5384615384615383, | |
| "grad_norm": 0.00027601871973959735, | |
| "learning_rate": 9.546412404157383e-06, | |
| "loss": 0.0, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 3.54251012145749, | |
| "grad_norm": 0.0002768997054684471, | |
| "learning_rate": 9.498240042204471e-06, | |
| "loss": 0.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.5465587044534415, | |
| "grad_norm": 0.00028770457516093684, | |
| "learning_rate": 9.450151652044339e-06, | |
| "loss": 0.0, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 3.5506072874493926, | |
| "grad_norm": 0.00028738960041192096, | |
| "learning_rate": 9.402147618190472e-06, | |
| "loss": 0.0, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 3.554655870445344, | |
| "grad_norm": 0.00027579158698104694, | |
| "learning_rate": 9.354228324481841e-06, | |
| "loss": 0.0, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 3.5587044534412957, | |
| "grad_norm": 0.0002824302469949513, | |
| "learning_rate": 9.306394154079852e-06, | |
| "loss": 0.0, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 3.562753036437247, | |
| "grad_norm": 0.00027164715962947134, | |
| "learning_rate": 9.258645489465254e-06, | |
| "loss": 0.0, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 3.5668016194331984, | |
| "grad_norm": 0.0002937912964350729, | |
| "learning_rate": 9.21098271243509e-06, | |
| "loss": 0.0, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 3.57085020242915, | |
| "grad_norm": 0.00029682169024411545, | |
| "learning_rate": 9.163406204099659e-06, | |
| "loss": 0.0, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 3.574898785425101, | |
| "grad_norm": 0.00028388056415425735, | |
| "learning_rate": 9.115916344879452e-06, | |
| "loss": 0.0, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 3.5789473684210527, | |
| "grad_norm": 0.0002711310367979457, | |
| "learning_rate": 9.068513514502126e-06, | |
| "loss": 0.0, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 3.582995951417004, | |
| "grad_norm": 0.0002816744656202141, | |
| "learning_rate": 9.021198091999439e-06, | |
| "loss": 0.0, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 3.5870445344129553, | |
| "grad_norm": 0.00026310354845524074, | |
| "learning_rate": 8.973970455704251e-06, | |
| "loss": 0.0, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 3.591093117408907, | |
| "grad_norm": 0.00028261649915065556, | |
| "learning_rate": 8.926830983247495e-06, | |
| "loss": 0.0, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 3.5951417004048585, | |
| "grad_norm": 0.0002689859586050993, | |
| "learning_rate": 8.879780051555116e-06, | |
| "loss": 0.0, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 3.5991902834008096, | |
| "grad_norm": 0.0002789584628121397, | |
| "learning_rate": 8.83281803684512e-06, | |
| "loss": 0.0, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 3.603238866396761, | |
| "grad_norm": 0.0002851336787661661, | |
| "learning_rate": 8.785945314624532e-06, | |
| "loss": 0.0, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 3.6072874493927127, | |
| "grad_norm": 0.00029077783411734294, | |
| "learning_rate": 8.739162259686385e-06, | |
| "loss": 0.0, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 3.611336032388664, | |
| "grad_norm": 0.00028855731439790435, | |
| "learning_rate": 8.692469246106734e-06, | |
| "loss": 0.0, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 3.6153846153846154, | |
| "grad_norm": 0.00027750702956590194, | |
| "learning_rate": 8.645866647241677e-06, | |
| "loss": 0.0, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 3.619433198380567, | |
| "grad_norm": 0.00027987513081329614, | |
| "learning_rate": 8.599354835724348e-06, | |
| "loss": 0.0, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 3.623481781376518, | |
| "grad_norm": 0.00027043578861230276, | |
| "learning_rate": 8.552934183461958e-06, | |
| "loss": 0.0, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 3.6275303643724697, | |
| "grad_norm": 0.00025953992440612823, | |
| "learning_rate": 8.506605061632805e-06, | |
| "loss": 0.0, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 3.6315789473684212, | |
| "grad_norm": 0.0002724113097887457, | |
| "learning_rate": 8.460367840683308e-06, | |
| "loss": 0.0, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 3.6356275303643724, | |
| "grad_norm": 0.0002725631930627984, | |
| "learning_rate": 8.414222890325065e-06, | |
| "loss": 0.0, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 3.639676113360324, | |
| "grad_norm": 0.0002780825183510319, | |
| "learning_rate": 8.368170579531855e-06, | |
| "loss": 0.0, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 3.6437246963562755, | |
| "grad_norm": 0.00029166255423317085, | |
| "learning_rate": 8.322211276536733e-06, | |
| "loss": 0.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.6477732793522266, | |
| "grad_norm": 0.0002695891256174862, | |
| "learning_rate": 8.27634534882907e-06, | |
| "loss": 0.0, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 3.651821862348178, | |
| "grad_norm": 0.00028747867383079, | |
| "learning_rate": 8.230573163151585e-06, | |
| "loss": 0.0, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 3.6558704453441297, | |
| "grad_norm": 0.0002650558065881571, | |
| "learning_rate": 8.184895085497466e-06, | |
| "loss": 0.0, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 3.659919028340081, | |
| "grad_norm": 0.00026683421636464823, | |
| "learning_rate": 8.139311481107401e-06, | |
| "loss": 0.0, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 3.6639676113360324, | |
| "grad_norm": 0.00027824030047929056, | |
| "learning_rate": 8.093822714466686e-06, | |
| "loss": 0.0, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 3.668016194331984, | |
| "grad_norm": 0.0002726867724795342, | |
| "learning_rate": 8.048429149302271e-06, | |
| "loss": 0.0, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 3.672064777327935, | |
| "grad_norm": 0.0002748051098774529, | |
| "learning_rate": 8.003131148579902e-06, | |
| "loss": 0.0, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 3.6761133603238867, | |
| "grad_norm": 0.00027178257871472423, | |
| "learning_rate": 7.957929074501182e-06, | |
| "loss": 0.0, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 3.6801619433198383, | |
| "grad_norm": 0.00027484648549629457, | |
| "learning_rate": 7.912823288500695e-06, | |
| "loss": 0.0, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 3.6842105263157894, | |
| "grad_norm": 0.00027661196191530984, | |
| "learning_rate": 7.867814151243102e-06, | |
| "loss": 0.0, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 3.688259109311741, | |
| "grad_norm": 0.0002542022299408088, | |
| "learning_rate": 7.822902022620267e-06, | |
| "loss": 0.0, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 3.6923076923076925, | |
| "grad_norm": 0.00026199873985837156, | |
| "learning_rate": 7.778087261748377e-06, | |
| "loss": 0.0, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 3.6963562753036436, | |
| "grad_norm": 0.0002760911947866933, | |
| "learning_rate": 7.733370226965056e-06, | |
| "loss": 0.0, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 3.700404858299595, | |
| "grad_norm": 0.000254018645122317, | |
| "learning_rate": 7.688751275826527e-06, | |
| "loss": 0.0, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 3.7044534412955468, | |
| "grad_norm": 0.0002642064333006633, | |
| "learning_rate": 7.644230765104735e-06, | |
| "loss": 0.0, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 3.708502024291498, | |
| "grad_norm": 0.0002635013680550276, | |
| "learning_rate": 7.599809050784502e-06, | |
| "loss": 0.0, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 3.7125506072874495, | |
| "grad_norm": 0.0002733804717741444, | |
| "learning_rate": 7.555486488060661e-06, | |
| "loss": 0.0, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 3.716599190283401, | |
| "grad_norm": 0.0002626364371654375, | |
| "learning_rate": 7.511263431335249e-06, | |
| "loss": 0.0, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 3.720647773279352, | |
| "grad_norm": 0.0002604166662967649, | |
| "learning_rate": 7.467140234214656e-06, | |
| "loss": 0.0, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 3.7246963562753037, | |
| "grad_norm": 0.0002585794535528466, | |
| "learning_rate": 7.423117249506777e-06, | |
| "loss": 0.0, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 3.7287449392712553, | |
| "grad_norm": 0.0002686830006596434, | |
| "learning_rate": 7.379194829218231e-06, | |
| "loss": 0.0, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 3.7327935222672064, | |
| "grad_norm": 0.00027629592597955716, | |
| "learning_rate": 7.335373324551518e-06, | |
| "loss": 0.0, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 3.736842105263158, | |
| "grad_norm": 0.0002564101967407936, | |
| "learning_rate": 7.291653085902224e-06, | |
| "loss": 0.0, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 3.7408906882591095, | |
| "grad_norm": 0.0002530545151549545, | |
| "learning_rate": 7.248034462856213e-06, | |
| "loss": 0.0, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 3.7449392712550607, | |
| "grad_norm": 0.00025601590989067404, | |
| "learning_rate": 7.20451780418683e-06, | |
| "loss": 0.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 3.748987854251012, | |
| "grad_norm": 0.0002593751411073882, | |
| "learning_rate": 7.161103457852128e-06, | |
| "loss": 0.0, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 3.753036437246964, | |
| "grad_norm": 0.00025887949059543363, | |
| "learning_rate": 7.117791770992046e-06, | |
| "loss": 0.0, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 3.757085020242915, | |
| "grad_norm": 0.0002576471569471408, | |
| "learning_rate": 7.074583089925688e-06, | |
| "loss": 0.0, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 3.7611336032388665, | |
| "grad_norm": 0.0002527002302393947, | |
| "learning_rate": 7.0314777601485155e-06, | |
| "loss": 0.0, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 3.765182186234818, | |
| "grad_norm": 0.00026971426655333154, | |
| "learning_rate": 6.988476126329604e-06, | |
| "loss": 0.0, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 3.769230769230769, | |
| "grad_norm": 0.00026078213122948126, | |
| "learning_rate": 6.945578532308857e-06, | |
| "loss": 0.0, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 3.7732793522267207, | |
| "grad_norm": 0.0002628165781288529, | |
| "learning_rate": 6.902785321094301e-06, | |
| "loss": 0.0, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 3.7773279352226723, | |
| "grad_norm": 0.0002661132263101758, | |
| "learning_rate": 6.8600968348593115e-06, | |
| "loss": 0.0, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 3.7813765182186234, | |
| "grad_norm": 0.00026378609536618995, | |
| "learning_rate": 6.817513414939892e-06, | |
| "loss": 0.0, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 3.785425101214575, | |
| "grad_norm": 0.00025424441012469217, | |
| "learning_rate": 6.775035401831924e-06, | |
| "loss": 0.0, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 3.7894736842105265, | |
| "grad_norm": 0.00025857691546249, | |
| "learning_rate": 6.732663135188473e-06, | |
| "loss": 0.0, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 3.7935222672064777, | |
| "grad_norm": 0.00025508852163060427, | |
| "learning_rate": 6.690396953817055e-06, | |
| "loss": 0.0, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 3.7975708502024292, | |
| "grad_norm": 0.00026211604473424046, | |
| "learning_rate": 6.648237195676927e-06, | |
| "loss": 0.0, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 3.801619433198381, | |
| "grad_norm": 0.00027283273099054764, | |
| "learning_rate": 6.606184197876395e-06, | |
| "loss": 0.0, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 3.805668016194332, | |
| "grad_norm": 0.0002526535532126408, | |
| "learning_rate": 6.564238296670109e-06, | |
| "loss": 0.0, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 3.8097165991902835, | |
| "grad_norm": 0.00025919260900448654, | |
| "learning_rate": 6.52239982745638e-06, | |
| "loss": 0.0, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 3.813765182186235, | |
| "grad_norm": 0.0002517238055993435, | |
| "learning_rate": 6.480669124774484e-06, | |
| "loss": 0.0, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 3.817813765182186, | |
| "grad_norm": 0.0002592269448360553, | |
| "learning_rate": 6.439046522302013e-06, | |
| "loss": 0.0, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 3.8218623481781377, | |
| "grad_norm": 0.0002546203192151698, | |
| "learning_rate": 6.397532352852191e-06, | |
| "loss": 0.0, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 3.8259109311740893, | |
| "grad_norm": 0.00026950310433003526, | |
| "learning_rate": 6.356126948371202e-06, | |
| "loss": 0.0, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 3.8299595141700404, | |
| "grad_norm": 0.0002463569358786061, | |
| "learning_rate": 6.314830639935559e-06, | |
| "loss": 0.0, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 3.834008097165992, | |
| "grad_norm": 0.00027671047461396765, | |
| "learning_rate": 6.273643757749448e-06, | |
| "loss": 0.0, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 3.8380566801619436, | |
| "grad_norm": 0.0002550328792217171, | |
| "learning_rate": 6.2325666311420856e-06, | |
| "loss": 0.0, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 3.8421052631578947, | |
| "grad_norm": 0.0002588127524172983, | |
| "learning_rate": 6.1915995885650675e-06, | |
| "loss": 0.0, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 3.8461538461538463, | |
| "grad_norm": 0.00025717624071059313, | |
| "learning_rate": 6.150742957589784e-06, | |
| "loss": 0.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 3.850202429149798, | |
| "grad_norm": 0.00025476535649470973, | |
| "learning_rate": 6.109997064904774e-06, | |
| "loss": 0.0, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 3.854251012145749, | |
| "grad_norm": 0.00025811848659061323, | |
| "learning_rate": 6.069362236313092e-06, | |
| "loss": 0.0, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 3.8582995951417005, | |
| "grad_norm": 0.0002601943711090709, | |
| "learning_rate": 6.0288387967297565e-06, | |
| "loss": 0.0, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 3.862348178137652, | |
| "grad_norm": 0.00025339617487938454, | |
| "learning_rate": 5.988427070179108e-06, | |
| "loss": 0.0, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 3.866396761133603, | |
| "grad_norm": 0.0002507113124015112, | |
| "learning_rate": 5.948127379792234e-06, | |
| "loss": 0.0, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 3.8704453441295548, | |
| "grad_norm": 0.00026211253142012953, | |
| "learning_rate": 5.90794004780439e-06, | |
| "loss": 0.0, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 3.8744939271255063, | |
| "grad_norm": 0.0002473769951402132, | |
| "learning_rate": 5.867865395552405e-06, | |
| "loss": 0.0, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 3.8785425101214575, | |
| "grad_norm": 0.0002476160757241943, | |
| "learning_rate": 5.8279037434721364e-06, | |
| "loss": 0.0, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 3.882591093117409, | |
| "grad_norm": 0.00026087310466288834, | |
| "learning_rate": 5.788055411095892e-06, | |
| "loss": 0.0, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 3.8866396761133606, | |
| "grad_norm": 0.0002495945540159441, | |
| "learning_rate": 5.7483207170498625e-06, | |
| "loss": 0.0, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 3.8906882591093117, | |
| "grad_norm": 0.00025149477376284053, | |
| "learning_rate": 5.708699979051613e-06, | |
| "loss": 0.0, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 3.8947368421052633, | |
| "grad_norm": 0.00025195827155969686, | |
| "learning_rate": 5.66919351390751e-06, | |
| "loss": 0.0, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 3.898785425101215, | |
| "grad_norm": 0.00027114892575921394, | |
| "learning_rate": 5.629801637510188e-06, | |
| "loss": 0.0, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 3.902834008097166, | |
| "grad_norm": 0.00024756044713484886, | |
| "learning_rate": 5.590524664836043e-06, | |
| "loss": 0.0, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 3.9068825910931175, | |
| "grad_norm": 0.00024253729786605522, | |
| "learning_rate": 5.551362909942708e-06, | |
| "loss": 0.0, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 3.910931174089069, | |
| "grad_norm": 0.00025705050170426764, | |
| "learning_rate": 5.51231668596653e-06, | |
| "loss": 0.0, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 3.91497975708502, | |
| "grad_norm": 0.0002558408145445056, | |
| "learning_rate": 5.47338630512007e-06, | |
| "loss": 0.0, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 3.919028340080972, | |
| "grad_norm": 0.0002543475244293729, | |
| "learning_rate": 5.434572078689617e-06, | |
| "loss": 0.0, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 3.9230769230769234, | |
| "grad_norm": 0.0002536698121826628, | |
| "learning_rate": 5.395874317032694e-06, | |
| "loss": 0.0, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 3.9271255060728745, | |
| "grad_norm": 0.00026310305740621833, | |
| "learning_rate": 5.357293329575573e-06, | |
| "loss": 0.0, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 3.931174089068826, | |
| "grad_norm": 0.0002603434343369588, | |
| "learning_rate": 5.318829424810797e-06, | |
| "loss": 0.0, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 3.9352226720647776, | |
| "grad_norm": 0.00026470829313472835, | |
| "learning_rate": 5.280482910294728e-06, | |
| "loss": 0.0, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 3.9392712550607287, | |
| "grad_norm": 0.00025094339264742255, | |
| "learning_rate": 5.242254092645078e-06, | |
| "loss": 0.0, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 3.9433198380566803, | |
| "grad_norm": 0.0002506618570161609, | |
| "learning_rate": 5.2041432775384445e-06, | |
| "loss": 0.0, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 3.9473684210526314, | |
| "grad_norm": 0.00026222719586907066, | |
| "learning_rate": 5.166150769707892e-06, | |
| "loss": 0.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 3.951417004048583, | |
| "grad_norm": 0.00025913332438696113, | |
| "learning_rate": 5.128276872940503e-06, | |
| "loss": 0.0, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 3.9554655870445345, | |
| "grad_norm": 0.00025316903808953546, | |
| "learning_rate": 5.090521890074953e-06, | |
| "loss": 0.0, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 3.9595141700404857, | |
| "grad_norm": 0.00024133610055776704, | |
| "learning_rate": 5.0528861229990675e-06, | |
| "loss": 0.0, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 3.9635627530364372, | |
| "grad_norm": 0.00024332887954575417, | |
| "learning_rate": 5.015369872647442e-06, | |
| "loss": 0.0, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 3.967611336032389, | |
| "grad_norm": 0.0002472276244325466, | |
| "learning_rate": 4.9779734389990195e-06, | |
| "loss": 0.0, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 3.97165991902834, | |
| "grad_norm": 0.0002619255836312224, | |
| "learning_rate": 4.940697121074675e-06, | |
| "loss": 0.0, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 3.9757085020242915, | |
| "grad_norm": 0.00025548433461400653, | |
| "learning_rate": 4.903541216934864e-06, | |
| "loss": 0.0, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 3.979757085020243, | |
| "grad_norm": 0.0002435251588170092, | |
| "learning_rate": 4.8665060236772e-06, | |
| "loss": 0.0, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 3.983805668016194, | |
| "grad_norm": 0.00025224361258247826, | |
| "learning_rate": 4.829591837434102e-06, | |
| "loss": 0.0, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 3.9878542510121457, | |
| "grad_norm": 0.00025728008422905264, | |
| "learning_rate": 4.792798953370424e-06, | |
| "loss": 0.0, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 3.9919028340080973, | |
| "grad_norm": 0.0002427097808594626, | |
| "learning_rate": 4.756127665681083e-06, | |
| "loss": 0.0, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 3.9959514170040484, | |
| "grad_norm": 0.00023948227758990376, | |
| "learning_rate": 4.719578267588725e-06, | |
| "loss": 0.0, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 0.00025322441975034947, | |
| "learning_rate": 4.6831510513413525e-06, | |
| "loss": 0.0, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 4.004048582995951, | |
| "grad_norm": 0.00024198239205662644, | |
| "learning_rate": 4.6468463082100176e-06, | |
| "loss": 0.0, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 4.008097165991903, | |
| "grad_norm": 0.0002503565649859677, | |
| "learning_rate": 4.610664328486485e-06, | |
| "loss": 0.0, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 4.012145748987854, | |
| "grad_norm": 0.00024788878549497856, | |
| "learning_rate": 4.5746054014809e-06, | |
| "loss": 0.0, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 4.016194331983805, | |
| "grad_norm": 0.00024723473212311125, | |
| "learning_rate": 4.5386698155194766e-06, | |
| "loss": 0.0, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 4.020242914979757, | |
| "grad_norm": 0.00024411986101496316, | |
| "learning_rate": 4.502857857942204e-06, | |
| "loss": 0.0, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 4.0242914979757085, | |
| "grad_norm": 0.00024368518509771124, | |
| "learning_rate": 4.467169815100549e-06, | |
| "loss": 0.0, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 4.02834008097166, | |
| "grad_norm": 0.00025386004144278574, | |
| "learning_rate": 4.431605972355137e-06, | |
| "loss": 0.0, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 4.032388663967612, | |
| "grad_norm": 0.00024124863088652174, | |
| "learning_rate": 4.396166614073516e-06, | |
| "loss": 0.0, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 4.036437246963563, | |
| "grad_norm": 0.00024261533645381, | |
| "learning_rate": 4.360852023627846e-06, | |
| "loss": 0.0, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 4.040485829959514, | |
| "grad_norm": 0.00025014906459441035, | |
| "learning_rate": 4.325662483392656e-06, | |
| "loss": 0.0, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 4.044534412955466, | |
| "grad_norm": 0.00024641961074625267, | |
| "learning_rate": 4.29059827474257e-06, | |
| "loss": 0.0, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 4.048582995951417, | |
| "grad_norm": 0.0002462645893967917, | |
| "learning_rate": 4.255659678050063e-06, | |
| "loss": 0.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.052631578947368, | |
| "grad_norm": 0.00024754851169610486, | |
| "learning_rate": 4.22084697268323e-06, | |
| "loss": 0.0, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 4.05668016194332, | |
| "grad_norm": 0.00023861772964033285, | |
| "learning_rate": 4.186160437003535e-06, | |
| "loss": 0.0, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 4.060728744939271, | |
| "grad_norm": 0.0002465201914605594, | |
| "learning_rate": 4.151600348363582e-06, | |
| "loss": 0.0, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 4.064777327935222, | |
| "grad_norm": 0.00023302080925909237, | |
| "learning_rate": 4.11716698310493e-06, | |
| "loss": 0.0, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 4.068825910931174, | |
| "grad_norm": 0.00024454676402522666, | |
| "learning_rate": 4.082860616555848e-06, | |
| "loss": 0.0, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 4.0728744939271255, | |
| "grad_norm": 0.00023299053955148468, | |
| "learning_rate": 4.048681523029125e-06, | |
| "loss": 0.0, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 4.076923076923077, | |
| "grad_norm": 0.00023461135114738796, | |
| "learning_rate": 4.014629975819882e-06, | |
| "loss": 0.0, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 4.080971659919029, | |
| "grad_norm": 0.0002495321934118724, | |
| "learning_rate": 3.9807062472033855e-06, | |
| "loss": 0.0, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 4.08502024291498, | |
| "grad_norm": 0.00024306692275914566, | |
| "learning_rate": 3.94691060843287e-06, | |
| "loss": 0.0, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 4.089068825910931, | |
| "grad_norm": 0.00023603470340644383, | |
| "learning_rate": 3.913243329737353e-06, | |
| "loss": 0.0, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 4.093117408906883, | |
| "grad_norm": 0.0002491883909705458, | |
| "learning_rate": 3.879704680319503e-06, | |
| "loss": 0.0, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 4.097165991902834, | |
| "grad_norm": 0.00023084738953777327, | |
| "learning_rate": 3.846294928353471e-06, | |
| "loss": 0.0, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 4.101214574898785, | |
| "grad_norm": 0.00023372813442129898, | |
| "learning_rate": 3.8130143409827345e-06, | |
| "loss": 0.0, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 4.105263157894737, | |
| "grad_norm": 0.00023691489896205078, | |
| "learning_rate": 3.7798631843179887e-06, | |
| "loss": 0.0, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 4.109311740890688, | |
| "grad_norm": 0.00024228411425049998, | |
| "learning_rate": 3.7468417234349973e-06, | |
| "loss": 0.0, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 4.113360323886639, | |
| "grad_norm": 0.00023967325692415898, | |
| "learning_rate": 3.7139502223724823e-06, | |
| "loss": 0.0, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 4.117408906882591, | |
| "grad_norm": 0.00024061264984285305, | |
| "learning_rate": 3.6811889441300095e-06, | |
| "loss": 0.0, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 4.1214574898785425, | |
| "grad_norm": 0.00027107865017987306, | |
| "learning_rate": 3.648558150665886e-06, | |
| "loss": 0.0, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 4.125506072874494, | |
| "grad_norm": 0.00024333516924557343, | |
| "learning_rate": 3.616058102895075e-06, | |
| "loss": 0.0, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 4.129554655870446, | |
| "grad_norm": 0.0002378730088100911, | |
| "learning_rate": 3.583689060687079e-06, | |
| "loss": 0.0, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 4.133603238866397, | |
| "grad_norm": 0.00024370092347451017, | |
| "learning_rate": 3.5514512828639023e-06, | |
| "loss": 0.0, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 4.137651821862348, | |
| "grad_norm": 0.0002415009450717022, | |
| "learning_rate": 3.51934502719796e-06, | |
| "loss": 0.0, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 4.1417004048583, | |
| "grad_norm": 0.0002415631174161534, | |
| "learning_rate": 3.4873705504100185e-06, | |
| "loss": 0.0, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 4.145748987854251, | |
| "grad_norm": 0.000235834213990549, | |
| "learning_rate": 3.4555281081671345e-06, | |
| "loss": 0.0, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 4.149797570850202, | |
| "grad_norm": 0.00023761596479585166, | |
| "learning_rate": 3.4238179550806326e-06, | |
| "loss": 0.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 4.153846153846154, | |
| "grad_norm": 0.0002404675054981725, | |
| "learning_rate": 3.3922403447040497e-06, | |
| "loss": 0.0, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 4.157894736842105, | |
| "grad_norm": 0.00024244590989855671, | |
| "learning_rate": 3.3607955295311247e-06, | |
| "loss": 0.0, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 4.161943319838056, | |
| "grad_norm": 0.00023909763972709667, | |
| "learning_rate": 3.32948376099375e-06, | |
| "loss": 0.0, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 4.165991902834008, | |
| "grad_norm": 0.0002344432512283362, | |
| "learning_rate": 3.2983052894599976e-06, | |
| "loss": 0.0, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 4.17004048582996, | |
| "grad_norm": 0.0002400328318447833, | |
| "learning_rate": 3.2672603642320945e-06, | |
| "loss": 0.0, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 4.174089068825911, | |
| "grad_norm": 0.00023595000589542886, | |
| "learning_rate": 3.236349233544438e-06, | |
| "loss": 0.0, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 4.178137651821863, | |
| "grad_norm": 0.00023739216990386296, | |
| "learning_rate": 3.205572144561604e-06, | |
| "loss": 0.0, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 4.182186234817814, | |
| "grad_norm": 0.00024099735621107998, | |
| "learning_rate": 3.174929343376374e-06, | |
| "loss": 0.0, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 4.186234817813765, | |
| "grad_norm": 0.0002371290510565098, | |
| "learning_rate": 3.1444210750077777e-06, | |
| "loss": 0.0, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 4.190283400809717, | |
| "grad_norm": 0.0002495170744665139, | |
| "learning_rate": 3.1140475833991046e-06, | |
| "loss": 0.0, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 4.194331983805668, | |
| "grad_norm": 0.0002309737723721278, | |
| "learning_rate": 3.0838091114159895e-06, | |
| "loss": 0.0, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 4.198380566801619, | |
| "grad_norm": 0.0002550047967149435, | |
| "learning_rate": 3.0537059008444524e-06, | |
| "loss": 0.0, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 4.202429149797571, | |
| "grad_norm": 0.0002360459388752921, | |
| "learning_rate": 3.0237381923889563e-06, | |
| "loss": 0.0, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 4.206477732793522, | |
| "grad_norm": 0.00024577284745500933, | |
| "learning_rate": 2.9939062256704997e-06, | |
| "loss": 0.0, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 4.2105263157894735, | |
| "grad_norm": 0.00025005036551244863, | |
| "learning_rate": 2.964210239224696e-06, | |
| "loss": 0.0, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 4.2145748987854255, | |
| "grad_norm": 0.00024122426378634377, | |
| "learning_rate": 2.934650470499869e-06, | |
| "loss": 0.0, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 4.218623481781377, | |
| "grad_norm": 0.0002371414373930687, | |
| "learning_rate": 2.905227155855128e-06, | |
| "loss": 0.0, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 4.222672064777328, | |
| "grad_norm": 0.0002314327902162553, | |
| "learning_rate": 2.875940530558519e-06, | |
| "loss": 0.0, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 4.22672064777328, | |
| "grad_norm": 0.00024588402354560474, | |
| "learning_rate": 2.8467908287851154e-06, | |
| "loss": 0.0, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 4.230769230769231, | |
| "grad_norm": 0.00025084196809891686, | |
| "learning_rate": 2.8177782836151603e-06, | |
| "loss": 0.0, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 4.234817813765182, | |
| "grad_norm": 0.00024251003172283023, | |
| "learning_rate": 2.7889031270321874e-06, | |
| "loss": 0.0, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 4.238866396761134, | |
| "grad_norm": 0.00022656557237471888, | |
| "learning_rate": 2.760165589921184e-06, | |
| "loss": 0.0, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 4.242914979757085, | |
| "grad_norm": 0.00023867336430910202, | |
| "learning_rate": 2.7315659020667305e-06, | |
| "loss": 0.0, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 4.246963562753036, | |
| "grad_norm": 0.0002411380133711942, | |
| "learning_rate": 2.703104292151164e-06, | |
| "loss": 0.0, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 4.251012145748988, | |
| "grad_norm": 0.00024171088402165694, | |
| "learning_rate": 2.6747809877527607e-06, | |
| "loss": 0.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 4.255060728744939, | |
| "grad_norm": 0.00023839570191141923, | |
| "learning_rate": 2.6465962153439082e-06, | |
| "loss": 0.0, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 4.2591093117408905, | |
| "grad_norm": 0.00024375114545322773, | |
| "learning_rate": 2.618550200289303e-06, | |
| "loss": 0.0, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 4.2631578947368425, | |
| "grad_norm": 0.0002306560298367365, | |
| "learning_rate": 2.59064316684412e-06, | |
| "loss": 0.0, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 4.267206477732794, | |
| "grad_norm": 0.00024378820777353171, | |
| "learning_rate": 2.562875338152264e-06, | |
| "loss": 0.0, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 4.271255060728745, | |
| "grad_norm": 0.00024166722991203886, | |
| "learning_rate": 2.5352469362445552e-06, | |
| "loss": 0.0, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 4.275303643724697, | |
| "grad_norm": 0.0002313585610565854, | |
| "learning_rate": 2.5077581820369523e-06, | |
| "loss": 0.0, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 4.279352226720648, | |
| "grad_norm": 0.00024477597822045627, | |
| "learning_rate": 2.480409295328803e-06, | |
| "loss": 0.0, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 4.283400809716599, | |
| "grad_norm": 0.00022732877491081754, | |
| "learning_rate": 2.45320049480108e-06, | |
| "loss": 0.0, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 4.287449392712551, | |
| "grad_norm": 0.0002346872960431557, | |
| "learning_rate": 2.4261319980146293e-06, | |
| "loss": 0.0, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 4.291497975708502, | |
| "grad_norm": 0.000245458094885919, | |
| "learning_rate": 2.399204021408421e-06, | |
| "loss": 0.0, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 4.295546558704453, | |
| "grad_norm": 0.00024497974329488586, | |
| "learning_rate": 2.3724167802978395e-06, | |
| "loss": 0.0, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 4.299595141700405, | |
| "grad_norm": 0.00024096702113710102, | |
| "learning_rate": 2.3457704888729625e-06, | |
| "loss": 0.0, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 4.303643724696356, | |
| "grad_norm": 0.0002493355862305284, | |
| "learning_rate": 2.3192653601968162e-06, | |
| "loss": 0.0, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 4.3076923076923075, | |
| "grad_norm": 0.00023573502322624373, | |
| "learning_rate": 2.2929016062037057e-06, | |
| "loss": 0.0, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 4.3117408906882595, | |
| "grad_norm": 0.00025138410648849, | |
| "learning_rate": 2.2666794376975055e-06, | |
| "loss": 0.0, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 4.315789473684211, | |
| "grad_norm": 0.0002333954406067252, | |
| "learning_rate": 2.24059906434998e-06, | |
| "loss": 0.0, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 4.319838056680162, | |
| "grad_norm": 0.00024503976906997476, | |
| "learning_rate": 2.214660694699091e-06, | |
| "loss": 0.0, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 4.323886639676114, | |
| "grad_norm": 0.00023135660970659974, | |
| "learning_rate": 2.188864536147355e-06, | |
| "loss": 0.0, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 4.327935222672065, | |
| "grad_norm": 0.00023441638310717176, | |
| "learning_rate": 2.1632107949601757e-06, | |
| "loss": 0.0, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 4.331983805668016, | |
| "grad_norm": 0.0002387250488305125, | |
| "learning_rate": 2.137699676264171e-06, | |
| "loss": 0.0, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 4.336032388663968, | |
| "grad_norm": 0.00022579165018985186, | |
| "learning_rate": 2.1123313840455738e-06, | |
| "loss": 0.0, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 4.340080971659919, | |
| "grad_norm": 0.0002328283435502402, | |
| "learning_rate": 2.08710612114857e-06, | |
| "loss": 0.0, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 4.34412955465587, | |
| "grad_norm": 0.00023220386326854278, | |
| "learning_rate": 2.0620240892736953e-06, | |
| "loss": 0.0, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 4.348178137651822, | |
| "grad_norm": 0.00023160859371602324, | |
| "learning_rate": 2.0370854889762e-06, | |
| "loss": 0.0, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 4.352226720647773, | |
| "grad_norm": 0.0002456747277199842, | |
| "learning_rate": 2.0122905196644726e-06, | |
| "loss": 0.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 4.3562753036437245, | |
| "grad_norm": 0.00023473533547879686, | |
| "learning_rate": 1.987639379598427e-06, | |
| "loss": 0.0, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 4.3603238866396765, | |
| "grad_norm": 0.0002390401628866185, | |
| "learning_rate": 1.963132265887919e-06, | |
| "loss": 0.0, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 4.364372469635628, | |
| "grad_norm": 0.0002401456592589974, | |
| "learning_rate": 1.9387693744911784e-06, | |
| "loss": 0.0, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 4.368421052631579, | |
| "grad_norm": 0.00023990261587954938, | |
| "learning_rate": 1.9145509002132368e-06, | |
| "loss": 0.0, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 4.372469635627531, | |
| "grad_norm": 0.00024283699832592525, | |
| "learning_rate": 1.8904770367043723e-06, | |
| "loss": 0.0, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 4.376518218623482, | |
| "grad_norm": 0.0002337089468060158, | |
| "learning_rate": 1.8665479764585504e-06, | |
| "loss": 0.0, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 4.380566801619433, | |
| "grad_norm": 0.0002256798834668852, | |
| "learning_rate": 1.8427639108118977e-06, | |
| "loss": 0.0, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 4.384615384615385, | |
| "grad_norm": 0.00023464805800397208, | |
| "learning_rate": 1.8191250299411755e-06, | |
| "loss": 0.0, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 4.388663967611336, | |
| "grad_norm": 0.0002366329343641748, | |
| "learning_rate": 1.7956315228622446e-06, | |
| "loss": 0.0, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 4.392712550607287, | |
| "grad_norm": 0.0002310834799989247, | |
| "learning_rate": 1.7722835774285553e-06, | |
| "loss": 0.0, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 4.396761133603239, | |
| "grad_norm": 0.000232534164655235, | |
| "learning_rate": 1.7490813803296624e-06, | |
| "loss": 0.0, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 4.40080971659919, | |
| "grad_norm": 0.0002244899530993135, | |
| "learning_rate": 1.726025117089718e-06, | |
| "loss": 0.0, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 4.4048582995951415, | |
| "grad_norm": 0.00022560968955839488, | |
| "learning_rate": 1.70311497206598e-06, | |
| "loss": 0.0, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 4.4089068825910935, | |
| "grad_norm": 0.00023647148513698956, | |
| "learning_rate": 1.6803511284473638e-06, | |
| "loss": 0.0, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 4.412955465587045, | |
| "grad_norm": 0.00023444141771290268, | |
| "learning_rate": 1.6577337682529538e-06, | |
| "loss": 0.0, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 4.417004048582996, | |
| "grad_norm": 0.00022871938246303783, | |
| "learning_rate": 1.6352630723305617e-06, | |
| "loss": 0.0, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 4.421052631578947, | |
| "grad_norm": 0.00024357588625405368, | |
| "learning_rate": 1.612939220355272e-06, | |
| "loss": 0.0, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 4.425101214574899, | |
| "grad_norm": 0.0002415021969577733, | |
| "learning_rate": 1.5907623908280112e-06, | |
| "loss": 0.0, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 4.42914979757085, | |
| "grad_norm": 0.00023098921095394628, | |
| "learning_rate": 1.568732761074121e-06, | |
| "loss": 0.0, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 4.433198380566802, | |
| "grad_norm": 0.00023501947927739023, | |
| "learning_rate": 1.5468505072419237e-06, | |
| "loss": 0.0, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 4.437246963562753, | |
| "grad_norm": 0.00023329122291141833, | |
| "learning_rate": 1.5251158043013448e-06, | |
| "loss": 0.0, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 4.441295546558704, | |
| "grad_norm": 0.0002372401030360245, | |
| "learning_rate": 1.5035288260424885e-06, | |
| "loss": 0.0, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 4.445344129554655, | |
| "grad_norm": 0.00023725184516210475, | |
| "learning_rate": 1.4820897450742623e-06, | |
| "loss": 0.0, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 4.449392712550607, | |
| "grad_norm": 0.0002400391541470882, | |
| "learning_rate": 1.4607987328229767e-06, | |
| "loss": 0.0, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 4.4534412955465585, | |
| "grad_norm": 0.00023377720568639812, | |
| "learning_rate": 1.4396559595310056e-06, | |
| "loss": 0.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.4574898785425106, | |
| "grad_norm": 0.00022361367431310308, | |
| "learning_rate": 1.4186615942553973e-06, | |
| "loss": 0.0, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 4.461538461538462, | |
| "grad_norm": 0.0002349353291144823, | |
| "learning_rate": 1.3978158048665424e-06, | |
| "loss": 0.0, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 4.465587044534413, | |
| "grad_norm": 0.00023831768990564516, | |
| "learning_rate": 1.3771187580468115e-06, | |
| "loss": 0.0, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 4.469635627530364, | |
| "grad_norm": 0.00023931268507604476, | |
| "learning_rate": 1.3565706192892392e-06, | |
| "loss": 0.0, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 4.473684210526316, | |
| "grad_norm": 0.00023329245632818858, | |
| "learning_rate": 1.3361715528961993e-06, | |
| "loss": 0.0, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 4.477732793522267, | |
| "grad_norm": 0.00023224584585324545, | |
| "learning_rate": 1.3159217219780817e-06, | |
| "loss": 0.0, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 4.481781376518219, | |
| "grad_norm": 0.00023428751982075547, | |
| "learning_rate": 1.2958212884519949e-06, | |
| "loss": 0.0, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 4.48582995951417, | |
| "grad_norm": 0.00022980904159299888, | |
| "learning_rate": 1.2758704130404675e-06, | |
| "loss": 0.0, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 4.489878542510121, | |
| "grad_norm": 0.000234512803378809, | |
| "learning_rate": 1.2560692552701738e-06, | |
| "loss": 0.0, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 4.493927125506072, | |
| "grad_norm": 0.0002206245798517386, | |
| "learning_rate": 1.2364179734706382e-06, | |
| "loss": 0.0, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 4.497975708502024, | |
| "grad_norm": 0.0002291244684503346, | |
| "learning_rate": 1.2169167247729852e-06, | |
| "loss": 0.0, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 4.502024291497976, | |
| "grad_norm": 0.0002388691033825648, | |
| "learning_rate": 1.1975656651086864e-06, | |
| "loss": 0.0, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 4.506072874493928, | |
| "grad_norm": 0.00024096089198897885, | |
| "learning_rate": 1.1783649492082926e-06, | |
| "loss": 0.0, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 4.510121457489879, | |
| "grad_norm": 0.00022451626475032872, | |
| "learning_rate": 1.1593147306002183e-06, | |
| "loss": 0.0, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 4.51417004048583, | |
| "grad_norm": 0.0002294517340469551, | |
| "learning_rate": 1.140415161609505e-06, | |
| "loss": 0.0, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 4.518218623481781, | |
| "grad_norm": 0.00022994634842360661, | |
| "learning_rate": 1.121666393356604e-06, | |
| "loss": 0.0, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 4.522267206477733, | |
| "grad_norm": 0.00024301120116363903, | |
| "learning_rate": 1.1030685757561632e-06, | |
| "loss": 0.0, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 4.526315789473684, | |
| "grad_norm": 0.0002330533659364121, | |
| "learning_rate": 1.0846218575158373e-06, | |
| "loss": 0.0, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 4.530364372469636, | |
| "grad_norm": 0.00023734005271325946, | |
| "learning_rate": 1.066326386135097e-06, | |
| "loss": 0.0, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 4.534412955465587, | |
| "grad_norm": 0.0002377105033824024, | |
| "learning_rate": 1.0481823079040398e-06, | |
| "loss": 0.0, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 4.538461538461538, | |
| "grad_norm": 0.00023140207494478217, | |
| "learning_rate": 1.0301897679022344e-06, | |
| "loss": 0.0, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 4.5425101214574894, | |
| "grad_norm": 0.00023478320792003608, | |
| "learning_rate": 1.0123489099975491e-06, | |
| "loss": 0.0, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 4.5465587044534415, | |
| "grad_norm": 0.00023438231016618247, | |
| "learning_rate": 9.9465987684501e-07, | |
| "loss": 0.0, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 4.550607287449393, | |
| "grad_norm": 0.00024726511232470237, | |
| "learning_rate": 9.771228098856534e-07, | |
| "loss": 0.0, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 4.554655870445345, | |
| "grad_norm": 0.0002323756369660634, | |
| "learning_rate": 9.597378493454013e-07, | |
| "loss": 0.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.558704453441296, | |
| "grad_norm": 0.0002280104285007604, | |
| "learning_rate": 9.425051342339353e-07, | |
| "loss": 0.0, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 4.562753036437247, | |
| "grad_norm": 0.0002378052704159675, | |
| "learning_rate": 9.254248023435864e-07, | |
| "loss": 0.0, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 4.566801619433198, | |
| "grad_norm": 0.0002289262150331279, | |
| "learning_rate": 9.084969902482288e-07, | |
| "loss": 0.0, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 4.57085020242915, | |
| "grad_norm": 0.00024470372800355347, | |
| "learning_rate": 8.917218333021993e-07, | |
| "loss": 0.0, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 4.574898785425101, | |
| "grad_norm": 0.00023845746915965467, | |
| "learning_rate": 8.750994656392109e-07, | |
| "loss": 0.0, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 4.578947368421053, | |
| "grad_norm": 0.00022802508423475188, | |
| "learning_rate": 8.58630020171265e-07, | |
| "loss": 0.0, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 4.582995951417004, | |
| "grad_norm": 0.0002327929249239758, | |
| "learning_rate": 8.423136285876099e-07, | |
| "loss": 0.0, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 4.587044534412955, | |
| "grad_norm": 0.00023561536769285013, | |
| "learning_rate": 8.26150421353682e-07, | |
| "loss": 0.0, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 4.5910931174089065, | |
| "grad_norm": 0.00022712939114275017, | |
| "learning_rate": 8.101405277100549e-07, | |
| "loss": 0.0, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 4.5951417004048585, | |
| "grad_norm": 0.00022289357786044702, | |
| "learning_rate": 7.942840756714077e-07, | |
| "loss": 0.0, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 4.59919028340081, | |
| "grad_norm": 0.00022571463614653927, | |
| "learning_rate": 7.785811920255093e-07, | |
| "loss": 0.0, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 4.603238866396762, | |
| "grad_norm": 0.00022891051208327317, | |
| "learning_rate": 7.630320023322024e-07, | |
| "loss": 0.0, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 4.607287449392713, | |
| "grad_norm": 0.00023180567277479664, | |
| "learning_rate": 7.476366309223903e-07, | |
| "loss": 0.0, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 4.611336032388664, | |
| "grad_norm": 0.00022736262965108778, | |
| "learning_rate": 7.323952008970537e-07, | |
| "loss": 0.0, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 4.615384615384615, | |
| "grad_norm": 0.00023240476465657717, | |
| "learning_rate": 7.173078341262618e-07, | |
| "loss": 0.0, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 4.619433198380567, | |
| "grad_norm": 0.00024052220751695136, | |
| "learning_rate": 7.023746512481989e-07, | |
| "loss": 0.0, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 4.623481781376518, | |
| "grad_norm": 0.00024016267663336477, | |
| "learning_rate": 6.875957716681902e-07, | |
| "loss": 0.0, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 4.62753036437247, | |
| "grad_norm": 0.00023702079965673753, | |
| "learning_rate": 6.729713135577687e-07, | |
| "loss": 0.0, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 4.631578947368421, | |
| "grad_norm": 0.0002437195790475031, | |
| "learning_rate": 6.585013938537143e-07, | |
| "loss": 0.0, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 4.635627530364372, | |
| "grad_norm": 0.00021700838380857106, | |
| "learning_rate": 6.441861282571115e-07, | |
| "loss": 0.0, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 4.6396761133603235, | |
| "grad_norm": 0.00026390040626026525, | |
| "learning_rate": 6.300256312324471e-07, | |
| "loss": 0.0, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 4.6437246963562755, | |
| "grad_norm": 0.00023580367574017357, | |
| "learning_rate": 6.160200160066776e-07, | |
| "loss": 0.0, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 4.647773279352227, | |
| "grad_norm": 0.0002295285297016369, | |
| "learning_rate": 6.021693945683327e-07, | |
| "loss": 0.0, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 4.651821862348179, | |
| "grad_norm": 0.00023531120372352095, | |
| "learning_rate": 5.884738776666044e-07, | |
| "loss": 0.0, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 4.65587044534413, | |
| "grad_norm": 0.00023009843249746463, | |
| "learning_rate": 5.749335748104878e-07, | |
| "loss": 0.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 4.659919028340081, | |
| "grad_norm": 0.00022638198666442808, | |
| "learning_rate": 5.615485942678778e-07, | |
| "loss": 0.0, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 4.663967611336032, | |
| "grad_norm": 0.0002341709735821321, | |
| "learning_rate": 5.483190430647289e-07, | |
| "loss": 0.0, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 4.668016194331984, | |
| "grad_norm": 0.00022729092597913752, | |
| "learning_rate": 5.352450269841747e-07, | |
| "loss": 0.0, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 4.672064777327935, | |
| "grad_norm": 0.0002470720955047624, | |
| "learning_rate": 5.22326650565701e-07, | |
| "loss": 0.0, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 4.676113360323887, | |
| "grad_norm": 0.00023302132353417756, | |
| "learning_rate": 5.095640171043048e-07, | |
| "loss": 0.0, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 4.680161943319838, | |
| "grad_norm": 0.00023186109512897475, | |
| "learning_rate": 4.969572286496594e-07, | |
| "loss": 0.0, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 4.684210526315789, | |
| "grad_norm": 0.00022596929926008217, | |
| "learning_rate": 4.845063860053079e-07, | |
| "loss": 0.0, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 4.6882591093117405, | |
| "grad_norm": 0.00022787549778368062, | |
| "learning_rate": 4.7221158872785997e-07, | |
| "loss": 0.0, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 4.6923076923076925, | |
| "grad_norm": 0.00022614776564073525, | |
| "learning_rate": 4.600729351261857e-07, | |
| "loss": 0.0, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 4.696356275303644, | |
| "grad_norm": 0.00022513970044033972, | |
| "learning_rate": 4.4809052226063355e-07, | |
| "loss": 0.0, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 4.700404858299595, | |
| "grad_norm": 0.00023020016895368938, | |
| "learning_rate": 4.3626444594226046e-07, | |
| "loss": 0.0, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 4.704453441295547, | |
| "grad_norm": 0.00023793801279281795, | |
| "learning_rate": 4.245948007320588e-07, | |
| "loss": 0.0, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 4.708502024291498, | |
| "grad_norm": 0.00024289699577344584, | |
| "learning_rate": 4.130816799401971e-07, | |
| "loss": 0.0, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 4.712550607287449, | |
| "grad_norm": 0.00022608350425164798, | |
| "learning_rate": 4.017251756252827e-07, | |
| "loss": 0.0, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 4.716599190283401, | |
| "grad_norm": 0.0002308800511429031, | |
| "learning_rate": 3.9052537859362473e-07, | |
| "loss": 0.0, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 4.720647773279352, | |
| "grad_norm": 0.00022674270608307573, | |
| "learning_rate": 3.7948237839849687e-07, | |
| "loss": 0.0, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 4.724696356275303, | |
| "grad_norm": 0.0002301126422106285, | |
| "learning_rate": 3.6859626333943797e-07, | |
| "loss": 0.0, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 4.728744939271255, | |
| "grad_norm": 0.00023067899577047208, | |
| "learning_rate": 3.578671204615347e-07, | |
| "loss": 0.0, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 4.732793522267206, | |
| "grad_norm": 0.00022431419548951705, | |
| "learning_rate": 3.472950355547267e-07, | |
| "loss": 0.0, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 4.7368421052631575, | |
| "grad_norm": 0.00022244163377737412, | |
| "learning_rate": 3.3688009315312244e-07, | |
| "loss": 0.0, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 4.7408906882591095, | |
| "grad_norm": 0.000235907882572876, | |
| "learning_rate": 3.266223765343224e-07, | |
| "loss": 0.0, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 4.744939271255061, | |
| "grad_norm": 0.00022590329358118558, | |
| "learning_rate": 3.16521967718757e-07, | |
| "loss": 0.0, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 4.748987854251012, | |
| "grad_norm": 0.0002206715625800822, | |
| "learning_rate": 3.0657894746902726e-07, | |
| "loss": 0.0, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 4.753036437246964, | |
| "grad_norm": 0.00022720710518262237, | |
| "learning_rate": 2.9679339528925654e-07, | |
| "loss": 0.0, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 4.757085020242915, | |
| "grad_norm": 0.00023100593371029916, | |
| "learning_rate": 2.871653894244597e-07, | |
| "loss": 0.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 4.761133603238866, | |
| "grad_norm": 0.00023223160243776645, | |
| "learning_rate": 2.7769500685991266e-07, | |
| "loss": 0.0, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 4.765182186234818, | |
| "grad_norm": 0.0002325751878464937, | |
| "learning_rate": 2.683823233205485e-07, | |
| "loss": 0.0, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 4.769230769230769, | |
| "grad_norm": 0.00023551452738661431, | |
| "learning_rate": 2.592274132703287e-07, | |
| "loss": 0.0, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 4.77327935222672, | |
| "grad_norm": 0.000226623238375117, | |
| "learning_rate": 2.502303499116754e-07, | |
| "loss": 0.0, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 4.777327935222672, | |
| "grad_norm": 0.0002243570059437583, | |
| "learning_rate": 2.413912051848666e-07, | |
| "loss": 0.0, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 4.781376518218623, | |
| "grad_norm": 0.00023590976694908635, | |
| "learning_rate": 2.3271004976746613e-07, | |
| "loss": 0.0, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 4.7854251012145745, | |
| "grad_norm": 0.00022857752166670318, | |
| "learning_rate": 2.2418695307376393e-07, | |
| "loss": 0.0, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 4.7894736842105265, | |
| "grad_norm": 0.00023165616912072687, | |
| "learning_rate": 2.1582198325420968e-07, | |
| "loss": 0.0, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 4.793522267206478, | |
| "grad_norm": 0.00022891606706724676, | |
| "learning_rate": 2.0761520719488447e-07, | |
| "loss": 0.0, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 4.797570850202429, | |
| "grad_norm": 0.00023718478567210809, | |
| "learning_rate": 1.9956669051694798e-07, | |
| "loss": 0.0, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 4.801619433198381, | |
| "grad_norm": 0.00023534186992811847, | |
| "learning_rate": 1.916764975761254e-07, | |
| "loss": 0.0, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 4.805668016194332, | |
| "grad_norm": 0.00023975334541991836, | |
| "learning_rate": 1.839446914621923e-07, | |
| "loss": 0.0, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 4.809716599190283, | |
| "grad_norm": 0.00023765762776850532, | |
| "learning_rate": 1.7637133399845962e-07, | |
| "loss": 0.0, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 4.813765182186235, | |
| "grad_norm": 0.00022741880740531067, | |
| "learning_rate": 1.6895648574129397e-07, | |
| "loss": 0.0, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 4.817813765182186, | |
| "grad_norm": 0.00022541371190664398, | |
| "learning_rate": 1.6170020597962465e-07, | |
| "loss": 0.0, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 4.821862348178137, | |
| "grad_norm": 0.0002381719896477725, | |
| "learning_rate": 1.5460255273447078e-07, | |
| "loss": 0.0, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 4.825910931174089, | |
| "grad_norm": 0.0002416318136283495, | |
| "learning_rate": 1.4766358275847715e-07, | |
| "loss": 0.0, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 4.82995951417004, | |
| "grad_norm": 0.00023157280893756205, | |
| "learning_rate": 1.4088335153546573e-07, | |
| "loss": 0.0, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 4.834008097165992, | |
| "grad_norm": 0.00023127117214541453, | |
| "learning_rate": 1.3426191327998496e-07, | |
| "loss": 0.0, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 4.838056680161944, | |
| "grad_norm": 0.00023719051010892992, | |
| "learning_rate": 1.277993209368744e-07, | |
| "loss": 0.0, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 4.842105263157895, | |
| "grad_norm": 0.00023235109922249618, | |
| "learning_rate": 1.214956261808542e-07, | |
| "loss": 0.0, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 4.846153846153846, | |
| "grad_norm": 0.0002524602935788583, | |
| "learning_rate": 1.1535087941609846e-07, | |
| "loss": 0.0, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 4.850202429149798, | |
| "grad_norm": 0.00023269565237315476, | |
| "learning_rate": 1.0936512977583802e-07, | |
| "loss": 0.0, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 4.854251012145749, | |
| "grad_norm": 0.0002251259836422741, | |
| "learning_rate": 1.0353842512196732e-07, | |
| "loss": 0.0, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 4.8582995951417, | |
| "grad_norm": 0.0002286780069872142, | |
| "learning_rate": 9.787081204466253e-08, | |
| "loss": 0.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 4.862348178137652, | |
| "grad_norm": 0.00023097920124237248, | |
| "learning_rate": 9.236233586200627e-08, | |
| "loss": 0.0, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 4.866396761133603, | |
| "grad_norm": 0.00022521712775884216, | |
| "learning_rate": 8.70130406196279e-08, | |
| "loss": 0.0, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 4.870445344129554, | |
| "grad_norm": 0.00023528480741667, | |
| "learning_rate": 8.182296909035047e-08, | |
| "loss": 0.0, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 4.874493927125506, | |
| "grad_norm": 0.00021990924646613083, | |
| "learning_rate": 7.679216277384439e-08, | |
| "loss": 0.0, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 4.8785425101214575, | |
| "grad_norm": 0.00024367888254142248, | |
| "learning_rate": 7.19206618963053e-08, | |
| "loss": 0.0, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 4.882591093117409, | |
| "grad_norm": 0.00022135495701149313, | |
| "learning_rate": 6.720850541012347e-08, | |
| "loss": 0.0, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 4.886639676113361, | |
| "grad_norm": 0.0002303010104504299, | |
| "learning_rate": 6.265573099357714e-08, | |
| "loss": 0.0, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 4.890688259109312, | |
| "grad_norm": 0.00022913339992033244, | |
| "learning_rate": 5.826237505053067e-08, | |
| "loss": 0.0, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 4.894736842105263, | |
| "grad_norm": 0.00022833706895203178, | |
| "learning_rate": 5.4028472710137e-08, | |
| "loss": 0.0, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 4.898785425101215, | |
| "grad_norm": 0.0002328721853314272, | |
| "learning_rate": 4.9954057826571146e-08, | |
| "loss": 0.0, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 4.902834008097166, | |
| "grad_norm": 0.00023073967992729615, | |
| "learning_rate": 4.6039162978743776e-08, | |
| "loss": 0.0, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 4.906882591093117, | |
| "grad_norm": 0.00023960756764466182, | |
| "learning_rate": 4.228381947004812e-08, | |
| "loss": 0.0, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 4.910931174089069, | |
| "grad_norm": 0.0002344753305520032, | |
| "learning_rate": 3.868805732811343e-08, | |
| "loss": 0.0, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 4.91497975708502, | |
| "grad_norm": 0.00022446966785116677, | |
| "learning_rate": 3.525190530455636e-08, | |
| "loss": 0.0, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 4.919028340080971, | |
| "grad_norm": 0.00023438537691939466, | |
| "learning_rate": 3.197539087475887e-08, | |
| "loss": 0.0, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 4.923076923076923, | |
| "grad_norm": 0.0002454451926763661, | |
| "learning_rate": 2.8858540237641786e-08, | |
| "loss": 0.0, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 4.9271255060728745, | |
| "grad_norm": 0.00023482980039428827, | |
| "learning_rate": 2.5901378315458247e-08, | |
| "loss": 0.0, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 4.931174089068826, | |
| "grad_norm": 0.0002284048336817537, | |
| "learning_rate": 2.3103928753598347e-08, | |
| "loss": 0.0, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 4.935222672064778, | |
| "grad_norm": 0.0002318789803189328, | |
| "learning_rate": 2.0466213920393718e-08, | |
| "loss": 0.0, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 4.939271255060729, | |
| "grad_norm": 0.00022327830795227285, | |
| "learning_rate": 1.7988254906942115e-08, | |
| "loss": 0.0, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 4.94331983805668, | |
| "grad_norm": 0.0002311592186630718, | |
| "learning_rate": 1.5670071526936447e-08, | |
| "loss": 0.0, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 4.947368421052632, | |
| "grad_norm": 0.0002248066828229642, | |
| "learning_rate": 1.3511682316509344e-08, | |
| "loss": 0.0, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 4.951417004048583, | |
| "grad_norm": 0.00023397031705196605, | |
| "learning_rate": 1.1513104534086605e-08, | |
| "loss": 0.0, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 4.955465587044534, | |
| "grad_norm": 0.00022897643671479232, | |
| "learning_rate": 9.67435416023843e-09, | |
| "loss": 0.0, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 4.959514170040486, | |
| "grad_norm": 0.00024610784134060094, | |
| "learning_rate": 7.995445897566178e-09, | |
| "loss": 0.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 4.963562753036437, | |
| "grad_norm": 0.00022530669844870934, | |
| "learning_rate": 6.476393170578021e-09, | |
| "loss": 0.0, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 4.967611336032388, | |
| "grad_norm": 0.00023440607581759316, | |
| "learning_rate": 5.11720812557348e-09, | |
| "loss": 0.0, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 4.97165991902834, | |
| "grad_norm": 0.00023001557046394605, | |
| "learning_rate": 3.917901630563492e-09, | |
| "loss": 0.0, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 4.9757085020242915, | |
| "grad_norm": 0.00023264568913031658, | |
| "learning_rate": 2.878483275170485e-09, | |
| "loss": 0.0, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 4.979757085020243, | |
| "grad_norm": 0.00022467320563054584, | |
| "learning_rate": 1.998961370555108e-09, | |
| "loss": 0.0, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 4.983805668016195, | |
| "grad_norm": 0.00023012351513973956, | |
| "learning_rate": 1.2793429493518361e-09, | |
| "loss": 0.0, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 4.987854251012146, | |
| "grad_norm": 0.00022644309716559688, | |
| "learning_rate": 7.196337656112384e-10, | |
| "loss": 0.0, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 4.991902834008097, | |
| "grad_norm": 0.00023670998919981764, | |
| "learning_rate": 3.198382947511292e-10, | |
| "loss": 0.0, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 4.995951417004049, | |
| "grad_norm": 0.0002366951430634754, | |
| "learning_rate": 7.995973352770137e-11, | |
| "loss": 0.0, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 0.00022244340788737913, | |
| "learning_rate": 0.0, | |
| "loss": 0.0, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 1235, | |
| "total_flos": 5.6976439963654554e+17, | |
| "train_loss": 0.04464531997530681, | |
| "train_runtime": 6084.8014, | |
| "train_samples_per_second": 25.966, | |
| "train_steps_per_second": 0.203 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1235, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.6976439963654554e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |