| { | |
| "best_metric": 1.0081819295883179, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.47058823529411764, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.002352941176470588, | |
| "grad_norm": 3.062838554382324, | |
| "learning_rate": 1.0100000000000002e-05, | |
| "loss": 2.3793, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.002352941176470588, | |
| "eval_loss": 2.571444034576416, | |
| "eval_runtime": 13.5133, | |
| "eval_samples_per_second": 423.657, | |
| "eval_steps_per_second": 13.246, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004705882352941176, | |
| "grad_norm": 3.511554002761841, | |
| "learning_rate": 2.0200000000000003e-05, | |
| "loss": 2.4046, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.007058823529411765, | |
| "grad_norm": 3.7289419174194336, | |
| "learning_rate": 3.0299999999999998e-05, | |
| "loss": 2.4423, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.009411764705882352, | |
| "grad_norm": 3.826002597808838, | |
| "learning_rate": 4.0400000000000006e-05, | |
| "loss": 2.433, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.011764705882352941, | |
| "grad_norm": 3.733748435974121, | |
| "learning_rate": 5.05e-05, | |
| "loss": 2.3183, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01411764705882353, | |
| "grad_norm": 3.9309937953948975, | |
| "learning_rate": 6.0599999999999996e-05, | |
| "loss": 2.1593, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.01647058823529412, | |
| "grad_norm": 1.619845986366272, | |
| "learning_rate": 7.07e-05, | |
| "loss": 1.9091, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.018823529411764704, | |
| "grad_norm": 1.6118720769882202, | |
| "learning_rate": 8.080000000000001e-05, | |
| "loss": 1.7707, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.021176470588235293, | |
| "grad_norm": 1.6020575761795044, | |
| "learning_rate": 9.09e-05, | |
| "loss": 1.6202, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.023529411764705882, | |
| "grad_norm": 1.5743026733398438, | |
| "learning_rate": 0.000101, | |
| "loss": 1.5756, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02588235294117647, | |
| "grad_norm": 1.6522502899169922, | |
| "learning_rate": 0.00010046842105263158, | |
| "loss": 1.4119, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.02823529411764706, | |
| "grad_norm": 1.905711054801941, | |
| "learning_rate": 9.993684210526315e-05, | |
| "loss": 1.3029, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.03058823529411765, | |
| "grad_norm": 1.981130838394165, | |
| "learning_rate": 9.940526315789473e-05, | |
| "loss": 1.6009, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.03294117647058824, | |
| "grad_norm": 1.2570853233337402, | |
| "learning_rate": 9.887368421052632e-05, | |
| "loss": 1.5536, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.03529411764705882, | |
| "grad_norm": 0.8584917187690735, | |
| "learning_rate": 9.83421052631579e-05, | |
| "loss": 1.4409, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.03764705882352941, | |
| "grad_norm": 1.3553175926208496, | |
| "learning_rate": 9.781052631578948e-05, | |
| "loss": 1.3756, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.9470474123954773, | |
| "learning_rate": 9.727894736842106e-05, | |
| "loss": 1.2287, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.042352941176470586, | |
| "grad_norm": 0.9331281185150146, | |
| "learning_rate": 9.674736842105263e-05, | |
| "loss": 1.1582, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.04470588235294118, | |
| "grad_norm": 1.0281063318252563, | |
| "learning_rate": 9.621578947368421e-05, | |
| "loss": 1.3719, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.047058823529411764, | |
| "grad_norm": 1.2275820970535278, | |
| "learning_rate": 9.568421052631578e-05, | |
| "loss": 1.5648, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04941176470588235, | |
| "grad_norm": 0.8868420124053955, | |
| "learning_rate": 9.515263157894737e-05, | |
| "loss": 1.4091, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.05176470588235294, | |
| "grad_norm": 0.8574855327606201, | |
| "learning_rate": 9.462105263157895e-05, | |
| "loss": 1.3385, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.05411764705882353, | |
| "grad_norm": 1.0610370635986328, | |
| "learning_rate": 9.408947368421054e-05, | |
| "loss": 1.2697, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.05647058823529412, | |
| "grad_norm": 1.0069725513458252, | |
| "learning_rate": 9.355789473684211e-05, | |
| "loss": 1.1621, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.058823529411764705, | |
| "grad_norm": 0.9563344717025757, | |
| "learning_rate": 9.302631578947369e-05, | |
| "loss": 1.0005, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0611764705882353, | |
| "grad_norm": 0.9165502786636353, | |
| "learning_rate": 9.249473684210526e-05, | |
| "loss": 1.4816, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.06352941176470588, | |
| "grad_norm": 1.1508539915084839, | |
| "learning_rate": 9.196315789473685e-05, | |
| "loss": 1.4342, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.06588235294117648, | |
| "grad_norm": 1.090813398361206, | |
| "learning_rate": 9.143157894736843e-05, | |
| "loss": 1.3629, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.06823529411764706, | |
| "grad_norm": 0.8920236229896545, | |
| "learning_rate": 9.09e-05, | |
| "loss": 1.2553, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.07058823529411765, | |
| "grad_norm": 0.7090237736701965, | |
| "learning_rate": 9.036842105263158e-05, | |
| "loss": 1.1073, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07294117647058823, | |
| "grad_norm": 1.0130761861801147, | |
| "learning_rate": 8.983684210526316e-05, | |
| "loss": 1.0312, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.07529411764705882, | |
| "grad_norm": 1.0755921602249146, | |
| "learning_rate": 8.930526315789474e-05, | |
| "loss": 1.4467, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.07764705882352942, | |
| "grad_norm": 0.9146537184715271, | |
| "learning_rate": 8.877368421052632e-05, | |
| "loss": 1.4095, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.6506552696228027, | |
| "learning_rate": 8.82421052631579e-05, | |
| "loss": 1.3034, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.08235294117647059, | |
| "grad_norm": 0.6594511866569519, | |
| "learning_rate": 8.771052631578948e-05, | |
| "loss": 1.1873, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.08470588235294117, | |
| "grad_norm": 0.7837107181549072, | |
| "learning_rate": 8.717894736842105e-05, | |
| "loss": 1.1256, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.08705882352941176, | |
| "grad_norm": 0.8229796290397644, | |
| "learning_rate": 8.664736842105263e-05, | |
| "loss": 1.0601, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.08941176470588236, | |
| "grad_norm": 0.6585485339164734, | |
| "learning_rate": 8.61157894736842e-05, | |
| "loss": 1.3792, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.09176470588235294, | |
| "grad_norm": 0.6591870784759521, | |
| "learning_rate": 8.55842105263158e-05, | |
| "loss": 1.3891, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.09411764705882353, | |
| "grad_norm": 0.5894742012023926, | |
| "learning_rate": 8.505263157894737e-05, | |
| "loss": 1.3068, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09647058823529411, | |
| "grad_norm": 0.5963307023048401, | |
| "learning_rate": 8.452105263157896e-05, | |
| "loss": 1.1833, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0988235294117647, | |
| "grad_norm": 0.7109506726264954, | |
| "learning_rate": 8.398947368421053e-05, | |
| "loss": 1.1566, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.1011764705882353, | |
| "grad_norm": 0.7429890632629395, | |
| "learning_rate": 8.345789473684211e-05, | |
| "loss": 1.0686, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.10352941176470588, | |
| "grad_norm": 0.7084085941314697, | |
| "learning_rate": 8.292631578947368e-05, | |
| "loss": 1.2941, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.10588235294117647, | |
| "grad_norm": 0.5622501373291016, | |
| "learning_rate": 8.239473684210526e-05, | |
| "loss": 1.3853, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10823529411764705, | |
| "grad_norm": 0.5678622126579285, | |
| "learning_rate": 8.186315789473683e-05, | |
| "loss": 1.2757, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.11058823529411765, | |
| "grad_norm": 0.5646488666534424, | |
| "learning_rate": 8.133157894736842e-05, | |
| "loss": 1.2279, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.11294117647058824, | |
| "grad_norm": 0.6251313090324402, | |
| "learning_rate": 8.080000000000001e-05, | |
| "loss": 1.1053, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.11529411764705882, | |
| "grad_norm": 0.6798649430274963, | |
| "learning_rate": 8.026842105263159e-05, | |
| "loss": 1.0443, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.11764705882352941, | |
| "grad_norm": 0.7527917623519897, | |
| "learning_rate": 7.973684210526316e-05, | |
| "loss": 0.881, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11764705882352941, | |
| "eval_loss": 1.176622986793518, | |
| "eval_runtime": 13.5969, | |
| "eval_samples_per_second": 421.05, | |
| "eval_steps_per_second": 13.165, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.671344518661499, | |
| "learning_rate": 7.920526315789474e-05, | |
| "loss": 1.4253, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.1223529411764706, | |
| "grad_norm": 0.5880969166755676, | |
| "learning_rate": 7.867368421052631e-05, | |
| "loss": 1.3348, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.12470588235294118, | |
| "grad_norm": 0.6282343864440918, | |
| "learning_rate": 7.814210526315789e-05, | |
| "loss": 1.2422, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.12705882352941175, | |
| "grad_norm": 0.5805298089981079, | |
| "learning_rate": 7.761052631578946e-05, | |
| "loss": 1.1409, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.12941176470588237, | |
| "grad_norm": 0.6550160050392151, | |
| "learning_rate": 7.707894736842105e-05, | |
| "loss": 1.0416, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.13176470588235295, | |
| "grad_norm": 0.8028090596199036, | |
| "learning_rate": 7.654736842105264e-05, | |
| "loss": 0.9202, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.13411764705882354, | |
| "grad_norm": 0.5485410690307617, | |
| "learning_rate": 7.601578947368422e-05, | |
| "loss": 1.3603, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.13647058823529412, | |
| "grad_norm": 0.5518571138381958, | |
| "learning_rate": 7.548421052631579e-05, | |
| "loss": 1.293, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.1388235294117647, | |
| "grad_norm": 0.5716201066970825, | |
| "learning_rate": 7.495263157894737e-05, | |
| "loss": 1.2133, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.1411764705882353, | |
| "grad_norm": 0.6243718266487122, | |
| "learning_rate": 7.442105263157894e-05, | |
| "loss": 1.1545, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.14352941176470588, | |
| "grad_norm": 0.68055260181427, | |
| "learning_rate": 7.388947368421053e-05, | |
| "loss": 1.1208, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.14588235294117646, | |
| "grad_norm": 0.7047899961471558, | |
| "learning_rate": 7.335789473684211e-05, | |
| "loss": 0.9698, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.14823529411764705, | |
| "grad_norm": 0.6293048858642578, | |
| "learning_rate": 7.282631578947368e-05, | |
| "loss": 1.2973, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.15058823529411763, | |
| "grad_norm": 0.556554913520813, | |
| "learning_rate": 7.229473684210527e-05, | |
| "loss": 1.3389, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.15294117647058825, | |
| "grad_norm": 0.5991462469100952, | |
| "learning_rate": 7.176315789473685e-05, | |
| "loss": 1.2742, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.15529411764705883, | |
| "grad_norm": 0.6164782643318176, | |
| "learning_rate": 7.123157894736842e-05, | |
| "loss": 1.2032, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.15764705882352942, | |
| "grad_norm": 0.6064922213554382, | |
| "learning_rate": 7.07e-05, | |
| "loss": 1.056, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 0.6620113849639893, | |
| "learning_rate": 7.016842105263159e-05, | |
| "loss": 1.0083, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.1623529411764706, | |
| "grad_norm": 0.6773979067802429, | |
| "learning_rate": 6.963684210526316e-05, | |
| "loss": 1.1461, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.16470588235294117, | |
| "grad_norm": 0.5662854313850403, | |
| "learning_rate": 6.910526315789474e-05, | |
| "loss": 1.3559, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.16705882352941176, | |
| "grad_norm": 0.5798998475074768, | |
| "learning_rate": 6.857368421052631e-05, | |
| "loss": 1.2389, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.16941176470588235, | |
| "grad_norm": 0.6113397479057312, | |
| "learning_rate": 6.80421052631579e-05, | |
| "loss": 1.1883, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.17176470588235293, | |
| "grad_norm": 0.6429812908172607, | |
| "learning_rate": 6.751052631578948e-05, | |
| "loss": 1.0725, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.17411764705882352, | |
| "grad_norm": 0.669151782989502, | |
| "learning_rate": 6.697894736842105e-05, | |
| "loss": 0.9818, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.17647058823529413, | |
| "grad_norm": 0.7604007124900818, | |
| "learning_rate": 6.644736842105264e-05, | |
| "loss": 0.8495, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.17882352941176471, | |
| "grad_norm": 0.541003942489624, | |
| "learning_rate": 6.591578947368422e-05, | |
| "loss": 1.3588, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.1811764705882353, | |
| "grad_norm": 0.6061081290245056, | |
| "learning_rate": 6.538421052631579e-05, | |
| "loss": 1.2932, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.18352941176470589, | |
| "grad_norm": 0.6060153841972351, | |
| "learning_rate": 6.485263157894737e-05, | |
| "loss": 1.1908, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.18588235294117647, | |
| "grad_norm": 0.6308295726776123, | |
| "learning_rate": 6.432105263157894e-05, | |
| "loss": 1.1446, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.18823529411764706, | |
| "grad_norm": 0.6454964280128479, | |
| "learning_rate": 6.378947368421053e-05, | |
| "loss": 0.9889, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.19058823529411764, | |
| "grad_norm": 0.7258140444755554, | |
| "learning_rate": 6.32578947368421e-05, | |
| "loss": 0.886, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.19294117647058823, | |
| "grad_norm": 0.5802991986274719, | |
| "learning_rate": 6.27263157894737e-05, | |
| "loss": 1.3406, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.1952941176470588, | |
| "grad_norm": 0.6241512894630432, | |
| "learning_rate": 6.219473684210527e-05, | |
| "loss": 1.3188, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.1976470588235294, | |
| "grad_norm": 0.6080717444419861, | |
| "learning_rate": 6.166315789473685e-05, | |
| "loss": 1.1992, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.6189625263214111, | |
| "learning_rate": 6.113157894736842e-05, | |
| "loss": 1.1226, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.2023529411764706, | |
| "grad_norm": 0.6604960560798645, | |
| "learning_rate": 6.0599999999999996e-05, | |
| "loss": 1.0563, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.20470588235294118, | |
| "grad_norm": 0.7085356116294861, | |
| "learning_rate": 6.006842105263158e-05, | |
| "loss": 0.9438, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.20705882352941177, | |
| "grad_norm": 0.6556686162948608, | |
| "learning_rate": 5.953684210526315e-05, | |
| "loss": 1.3147, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.20941176470588235, | |
| "grad_norm": 0.637535572052002, | |
| "learning_rate": 5.900526315789474e-05, | |
| "loss": 1.2895, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.21176470588235294, | |
| "grad_norm": 0.5857135057449341, | |
| "learning_rate": 5.847368421052632e-05, | |
| "loss": 1.2138, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.21411764705882352, | |
| "grad_norm": 0.6470074653625488, | |
| "learning_rate": 5.79421052631579e-05, | |
| "loss": 1.1728, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.2164705882352941, | |
| "grad_norm": 0.6902785301208496, | |
| "learning_rate": 5.7410526315789475e-05, | |
| "loss": 1.0183, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.2188235294117647, | |
| "grad_norm": 0.721517026424408, | |
| "learning_rate": 5.687894736842105e-05, | |
| "loss": 0.9262, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.2211764705882353, | |
| "grad_norm": 0.7650291919708252, | |
| "learning_rate": 5.6347368421052625e-05, | |
| "loss": 1.1511, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.2235294117647059, | |
| "grad_norm": 0.6252285838127136, | |
| "learning_rate": 5.5815789473684214e-05, | |
| "loss": 1.3332, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.22588235294117648, | |
| "grad_norm": 0.6287341117858887, | |
| "learning_rate": 5.5284210526315796e-05, | |
| "loss": 1.1899, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.22823529411764706, | |
| "grad_norm": 0.7121133208274841, | |
| "learning_rate": 5.475263157894737e-05, | |
| "loss": 1.1943, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.23058823529411765, | |
| "grad_norm": 0.6616747379302979, | |
| "learning_rate": 5.422105263157895e-05, | |
| "loss": 1.0497, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.23294117647058823, | |
| "grad_norm": 0.7290382385253906, | |
| "learning_rate": 5.368947368421053e-05, | |
| "loss": 0.9758, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.23529411764705882, | |
| "grad_norm": 0.783645749092102, | |
| "learning_rate": 5.3157894736842104e-05, | |
| "loss": 0.8081, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.23529411764705882, | |
| "eval_loss": 1.0883766412734985, | |
| "eval_runtime": 13.5365, | |
| "eval_samples_per_second": 422.93, | |
| "eval_steps_per_second": 13.224, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2376470588235294, | |
| "grad_norm": 0.5632948279380798, | |
| "learning_rate": 5.262631578947368e-05, | |
| "loss": 1.2828, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.6145716309547424, | |
| "learning_rate": 5.209473684210527e-05, | |
| "loss": 1.2529, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.24235294117647058, | |
| "grad_norm": 0.6459245085716248, | |
| "learning_rate": 5.1563157894736844e-05, | |
| "loss": 1.1525, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.2447058823529412, | |
| "grad_norm": 0.6708313822746277, | |
| "learning_rate": 5.1031578947368426e-05, | |
| "loss": 1.0458, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.24705882352941178, | |
| "grad_norm": 0.73152756690979, | |
| "learning_rate": 5.05e-05, | |
| "loss": 0.9634, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.24941176470588236, | |
| "grad_norm": 0.7417388558387756, | |
| "learning_rate": 4.9968421052631576e-05, | |
| "loss": 0.8512, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.25176470588235295, | |
| "grad_norm": 0.61714106798172, | |
| "learning_rate": 4.943684210526316e-05, | |
| "loss": 1.2966, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.2541176470588235, | |
| "grad_norm": 0.6486982703208923, | |
| "learning_rate": 4.890526315789474e-05, | |
| "loss": 1.296, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.2564705882352941, | |
| "grad_norm": 0.6546630859375, | |
| "learning_rate": 4.8373684210526316e-05, | |
| "loss": 1.1757, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.25882352941176473, | |
| "grad_norm": 0.6595495939254761, | |
| "learning_rate": 4.784210526315789e-05, | |
| "loss": 1.0585, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2611764705882353, | |
| "grad_norm": 0.6996961832046509, | |
| "learning_rate": 4.731052631578947e-05, | |
| "loss": 0.926, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.2635294117647059, | |
| "grad_norm": 0.7664803266525269, | |
| "learning_rate": 4.6778947368421055e-05, | |
| "loss": 0.9221, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.26588235294117646, | |
| "grad_norm": 0.7366296648979187, | |
| "learning_rate": 4.624736842105263e-05, | |
| "loss": 1.2654, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.26823529411764707, | |
| "grad_norm": 0.5936444997787476, | |
| "learning_rate": 4.571578947368421e-05, | |
| "loss": 1.2535, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.27058823529411763, | |
| "grad_norm": 0.6593197584152222, | |
| "learning_rate": 4.518421052631579e-05, | |
| "loss": 1.2031, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.27294117647058824, | |
| "grad_norm": 0.6682748198509216, | |
| "learning_rate": 4.465263157894737e-05, | |
| "loss": 1.083, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.2752941176470588, | |
| "grad_norm": 0.723254919052124, | |
| "learning_rate": 4.412105263157895e-05, | |
| "loss": 0.9981, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.2776470588235294, | |
| "grad_norm": 0.7454279661178589, | |
| "learning_rate": 4.358947368421053e-05, | |
| "loss": 0.858, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.7310999035835266, | |
| "learning_rate": 4.30578947368421e-05, | |
| "loss": 1.0973, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.2823529411764706, | |
| "grad_norm": 0.6533228158950806, | |
| "learning_rate": 4.2526315789473685e-05, | |
| "loss": 1.2816, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2847058823529412, | |
| "grad_norm": 0.6907062530517578, | |
| "learning_rate": 4.199473684210527e-05, | |
| "loss": 1.1821, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.28705882352941176, | |
| "grad_norm": 0.655579686164856, | |
| "learning_rate": 4.146315789473684e-05, | |
| "loss": 1.0816, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.28941176470588237, | |
| "grad_norm": 0.7422165274620056, | |
| "learning_rate": 4.093157894736842e-05, | |
| "loss": 1.0255, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.2917647058823529, | |
| "grad_norm": 0.7321489453315735, | |
| "learning_rate": 4.0400000000000006e-05, | |
| "loss": 0.9031, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.29411764705882354, | |
| "grad_norm": 0.8530080318450928, | |
| "learning_rate": 3.986842105263158e-05, | |
| "loss": 0.7604, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.2964705882352941, | |
| "grad_norm": 0.6039284467697144, | |
| "learning_rate": 3.933684210526316e-05, | |
| "loss": 1.3394, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.2988235294117647, | |
| "grad_norm": 0.6608713865280151, | |
| "learning_rate": 3.880526315789473e-05, | |
| "loss": 1.2596, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.30117647058823527, | |
| "grad_norm": 0.6703007221221924, | |
| "learning_rate": 3.827368421052632e-05, | |
| "loss": 1.1305, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.3035294117647059, | |
| "grad_norm": 0.7194546461105347, | |
| "learning_rate": 3.7742105263157896e-05, | |
| "loss": 1.0206, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.3058823529411765, | |
| "grad_norm": 0.7508780360221863, | |
| "learning_rate": 3.721052631578947e-05, | |
| "loss": 0.9207, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.30823529411764705, | |
| "grad_norm": 0.747114360332489, | |
| "learning_rate": 3.6678947368421054e-05, | |
| "loss": 0.7736, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.31058823529411766, | |
| "grad_norm": 0.6456690430641174, | |
| "learning_rate": 3.6147368421052636e-05, | |
| "loss": 1.2843, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.3129411764705882, | |
| "grad_norm": 0.6609508991241455, | |
| "learning_rate": 3.561578947368421e-05, | |
| "loss": 1.2404, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.31529411764705884, | |
| "grad_norm": 0.6665840744972229, | |
| "learning_rate": 3.508421052631579e-05, | |
| "loss": 1.119, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.3176470588235294, | |
| "grad_norm": 0.6979455351829529, | |
| "learning_rate": 3.455263157894737e-05, | |
| "loss": 1.0221, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.7018398642539978, | |
| "learning_rate": 3.402105263157895e-05, | |
| "loss": 0.9381, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.32235294117647056, | |
| "grad_norm": 0.8422653675079346, | |
| "learning_rate": 3.3489473684210526e-05, | |
| "loss": 0.841, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.3247058823529412, | |
| "grad_norm": 0.7037672400474548, | |
| "learning_rate": 3.295789473684211e-05, | |
| "loss": 1.2042, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.3270588235294118, | |
| "grad_norm": 0.6488674283027649, | |
| "learning_rate": 3.242631578947368e-05, | |
| "loss": 1.2254, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.32941176470588235, | |
| "grad_norm": 0.6584794521331787, | |
| "learning_rate": 3.1894736842105265e-05, | |
| "loss": 1.1519, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.33176470588235296, | |
| "grad_norm": 0.7326436638832092, | |
| "learning_rate": 3.136315789473685e-05, | |
| "loss": 1.0655, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.3341176470588235, | |
| "grad_norm": 0.7417482137680054, | |
| "learning_rate": 3.083157894736842e-05, | |
| "loss": 0.9448, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.33647058823529413, | |
| "grad_norm": 0.7665135264396667, | |
| "learning_rate": 3.0299999999999998e-05, | |
| "loss": 0.8526, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.3388235294117647, | |
| "grad_norm": 0.7983363270759583, | |
| "learning_rate": 2.9768421052631577e-05, | |
| "loss": 1.1367, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.3411764705882353, | |
| "grad_norm": 0.6683626770973206, | |
| "learning_rate": 2.923684210526316e-05, | |
| "loss": 1.2683, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.34352941176470586, | |
| "grad_norm": 0.6849150657653809, | |
| "learning_rate": 2.8705263157894737e-05, | |
| "loss": 1.1433, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.3458823529411765, | |
| "grad_norm": 0.7283281683921814, | |
| "learning_rate": 2.8173684210526313e-05, | |
| "loss": 1.0574, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.34823529411764703, | |
| "grad_norm": 0.7541592121124268, | |
| "learning_rate": 2.7642105263157898e-05, | |
| "loss": 0.9838, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.35058823529411764, | |
| "grad_norm": 0.7645028233528137, | |
| "learning_rate": 2.7110526315789473e-05, | |
| "loss": 0.9245, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.35294117647058826, | |
| "grad_norm": 0.8516260981559753, | |
| "learning_rate": 2.6578947368421052e-05, | |
| "loss": 0.7641, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.35294117647058826, | |
| "eval_loss": 1.0314304828643799, | |
| "eval_runtime": 13.5252, | |
| "eval_samples_per_second": 423.285, | |
| "eval_steps_per_second": 13.235, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3552941176470588, | |
| "grad_norm": 0.6219114661216736, | |
| "learning_rate": 2.6047368421052634e-05, | |
| "loss": 1.2679, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.35764705882352943, | |
| "grad_norm": 0.6947327852249146, | |
| "learning_rate": 2.5515789473684213e-05, | |
| "loss": 1.1812, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.6886230707168579, | |
| "learning_rate": 2.4984210526315788e-05, | |
| "loss": 1.0995, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.3623529411764706, | |
| "grad_norm": 0.7634572982788086, | |
| "learning_rate": 2.445263157894737e-05, | |
| "loss": 1.026, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.36470588235294116, | |
| "grad_norm": 0.7814245223999023, | |
| "learning_rate": 2.3921052631578946e-05, | |
| "loss": 0.9492, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.36705882352941177, | |
| "grad_norm": 0.8378550410270691, | |
| "learning_rate": 2.3389473684210528e-05, | |
| "loss": 0.7962, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.36941176470588233, | |
| "grad_norm": 0.7198935151100159, | |
| "learning_rate": 2.2857894736842106e-05, | |
| "loss": 1.2946, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.37176470588235294, | |
| "grad_norm": 0.6716011166572571, | |
| "learning_rate": 2.2326315789473685e-05, | |
| "loss": 1.2314, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.37411764705882355, | |
| "grad_norm": 0.7164915800094604, | |
| "learning_rate": 2.1794736842105264e-05, | |
| "loss": 1.0985, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.3764705882352941, | |
| "grad_norm": 0.7440788745880127, | |
| "learning_rate": 2.1263157894736842e-05, | |
| "loss": 1.0568, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.3788235294117647, | |
| "grad_norm": 0.7702521085739136, | |
| "learning_rate": 2.073157894736842e-05, | |
| "loss": 0.9384, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.3811764705882353, | |
| "grad_norm": 0.8371986746788025, | |
| "learning_rate": 2.0200000000000003e-05, | |
| "loss": 0.7604, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.3835294117647059, | |
| "grad_norm": 0.7777345776557922, | |
| "learning_rate": 1.966842105263158e-05, | |
| "loss": 1.202, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.38588235294117645, | |
| "grad_norm": 0.6666561365127563, | |
| "learning_rate": 1.913684210526316e-05, | |
| "loss": 1.2439, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.38823529411764707, | |
| "grad_norm": 0.6794990301132202, | |
| "learning_rate": 1.8605263157894736e-05, | |
| "loss": 1.1159, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.3905882352941176, | |
| "grad_norm": 0.7484257221221924, | |
| "learning_rate": 1.8073684210526318e-05, | |
| "loss": 1.0882, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.39294117647058824, | |
| "grad_norm": 0.7433997988700867, | |
| "learning_rate": 1.7542105263157897e-05, | |
| "loss": 0.9214, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.3952941176470588, | |
| "grad_norm": 0.8234543800354004, | |
| "learning_rate": 1.7010526315789475e-05, | |
| "loss": 0.8793, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.3976470588235294, | |
| "grad_norm": 0.8275732398033142, | |
| "learning_rate": 1.6478947368421054e-05, | |
| "loss": 1.0662, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.6753153800964355, | |
| "learning_rate": 1.5947368421052633e-05, | |
| "loss": 1.2686, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.4023529411764706, | |
| "grad_norm": 0.7300885319709778, | |
| "learning_rate": 1.541578947368421e-05, | |
| "loss": 1.1619, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.4047058823529412, | |
| "grad_norm": 0.720065176486969, | |
| "learning_rate": 1.4884210526315788e-05, | |
| "loss": 1.0637, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.40705882352941175, | |
| "grad_norm": 0.7786014080047607, | |
| "learning_rate": 1.4352631578947369e-05, | |
| "loss": 0.9741, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.40941176470588236, | |
| "grad_norm": 0.8148536086082458, | |
| "learning_rate": 1.3821052631578949e-05, | |
| "loss": 0.8535, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.4117647058823529, | |
| "grad_norm": 0.8557289242744446, | |
| "learning_rate": 1.3289473684210526e-05, | |
| "loss": 0.7194, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.41411764705882353, | |
| "grad_norm": 0.6708107590675354, | |
| "learning_rate": 1.2757894736842106e-05, | |
| "loss": 1.2988, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.4164705882352941, | |
| "grad_norm": 0.695059597492218, | |
| "learning_rate": 1.2226315789473685e-05, | |
| "loss": 1.1462, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.4188235294117647, | |
| "grad_norm": 0.7289761304855347, | |
| "learning_rate": 1.1694736842105264e-05, | |
| "loss": 1.0924, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.4211764705882353, | |
| "grad_norm": 0.7350468635559082, | |
| "learning_rate": 1.1163157894736842e-05, | |
| "loss": 0.9741, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.4235294117647059, | |
| "grad_norm": 0.7673172950744629, | |
| "learning_rate": 1.0631578947368421e-05, | |
| "loss": 0.8985, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4258823529411765, | |
| "grad_norm": 0.907876193523407, | |
| "learning_rate": 1.0100000000000002e-05, | |
| "loss": 0.7691, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.42823529411764705, | |
| "grad_norm": 0.6842520236968994, | |
| "learning_rate": 9.56842105263158e-06, | |
| "loss": 1.2146, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.43058823529411766, | |
| "grad_norm": 0.6845569014549255, | |
| "learning_rate": 9.036842105263159e-06, | |
| "loss": 1.1655, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.4329411764705882, | |
| "grad_norm": 0.7404122352600098, | |
| "learning_rate": 8.505263157894738e-06, | |
| "loss": 1.1055, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.43529411764705883, | |
| "grad_norm": 0.7731722593307495, | |
| "learning_rate": 7.973684210526316e-06, | |
| "loss": 1.067, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.4376470588235294, | |
| "grad_norm": 0.8028244972229004, | |
| "learning_rate": 7.442105263157894e-06, | |
| "loss": 0.9007, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.9162750244140625, | |
| "learning_rate": 6.9105263157894745e-06, | |
| "loss": 0.8109, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.4423529411764706, | |
| "grad_norm": 0.729189932346344, | |
| "learning_rate": 6.378947368421053e-06, | |
| "loss": 1.1673, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.4447058823529412, | |
| "grad_norm": 0.691579282283783, | |
| "learning_rate": 5.847368421052632e-06, | |
| "loss": 1.2002, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.4470588235294118, | |
| "grad_norm": 0.6862315535545349, | |
| "learning_rate": 5.315789473684211e-06, | |
| "loss": 1.075, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.44941176470588234, | |
| "grad_norm": 0.795240044593811, | |
| "learning_rate": 4.78421052631579e-06, | |
| "loss": 1.0699, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.45176470588235296, | |
| "grad_norm": 0.7700913548469543, | |
| "learning_rate": 4.252631578947369e-06, | |
| "loss": 0.8609, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.4541176470588235, | |
| "grad_norm": 0.8226372003555298, | |
| "learning_rate": 3.721052631578947e-06, | |
| "loss": 0.8284, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.45647058823529413, | |
| "grad_norm": 0.7959129810333252, | |
| "learning_rate": 3.1894736842105266e-06, | |
| "loss": 0.9911, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.4588235294117647, | |
| "grad_norm": 0.6747735142707825, | |
| "learning_rate": 2.6578947368421053e-06, | |
| "loss": 1.2293, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.4611764705882353, | |
| "grad_norm": 0.729983925819397, | |
| "learning_rate": 2.1263157894736844e-06, | |
| "loss": 1.1679, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.46352941176470586, | |
| "grad_norm": 0.7446680665016174, | |
| "learning_rate": 1.5947368421052633e-06, | |
| "loss": 0.9839, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.46588235294117647, | |
| "grad_norm": 0.7689118385314941, | |
| "learning_rate": 1.0631578947368422e-06, | |
| "loss": 0.9591, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.4682352941176471, | |
| "grad_norm": 0.8310439586639404, | |
| "learning_rate": 5.315789473684211e-07, | |
| "loss": 0.8448, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.47058823529411764, | |
| "grad_norm": 0.9187881350517273, | |
| "learning_rate": 0.0, | |
| "loss": 0.7259, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.47058823529411764, | |
| "eval_loss": 1.0081819295883179, | |
| "eval_runtime": 13.5434, | |
| "eval_samples_per_second": 422.714, | |
| "eval_steps_per_second": 13.217, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.205091905167688e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |