| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.9977761304670127, | |
| "eval_steps": 500, | |
| "global_step": 1011, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0029651593773165306, | |
| "grad_norm": 11.262527465820312, | |
| "learning_rate": 1.9607843137254904e-07, | |
| "loss": 0.9138, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.005930318754633061, | |
| "grad_norm": 10.613697052001953, | |
| "learning_rate": 3.921568627450981e-07, | |
| "loss": 0.9147, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.008895478131949592, | |
| "grad_norm": 11.271403312683105, | |
| "learning_rate": 5.882352941176471e-07, | |
| "loss": 0.9118, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.011860637509266123, | |
| "grad_norm": 11.093155860900879, | |
| "learning_rate": 7.843137254901962e-07, | |
| "loss": 0.9137, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.014825796886582653, | |
| "grad_norm": 11.175023078918457, | |
| "learning_rate": 9.80392156862745e-07, | |
| "loss": 0.8879, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.017790956263899184, | |
| "grad_norm": 10.275071144104004, | |
| "learning_rate": 1.1764705882352942e-06, | |
| "loss": 0.864, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.020756115641215715, | |
| "grad_norm": 8.285501480102539, | |
| "learning_rate": 1.3725490196078434e-06, | |
| "loss": 0.8612, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.023721275018532245, | |
| "grad_norm": 6.519635200500488, | |
| "learning_rate": 1.5686274509803923e-06, | |
| "loss": 0.8372, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.026686434395848776, | |
| "grad_norm": 6.018601894378662, | |
| "learning_rate": 1.7647058823529414e-06, | |
| "loss": 0.8244, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.029651593773165306, | |
| "grad_norm": 5.061045169830322, | |
| "learning_rate": 1.96078431372549e-06, | |
| "loss": 0.8057, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03261675315048184, | |
| "grad_norm": 5.859638214111328, | |
| "learning_rate": 2.1568627450980393e-06, | |
| "loss": 0.7734, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.03558191252779837, | |
| "grad_norm": 5.410571098327637, | |
| "learning_rate": 2.3529411764705885e-06, | |
| "loss": 0.7635, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0385470719051149, | |
| "grad_norm": 3.8421123027801514, | |
| "learning_rate": 2.549019607843137e-06, | |
| "loss": 0.7373, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.04151223128243143, | |
| "grad_norm": 2.3517632484436035, | |
| "learning_rate": 2.7450980392156867e-06, | |
| "loss": 0.7035, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.04447739065974796, | |
| "grad_norm": 2.1120362281799316, | |
| "learning_rate": 2.9411764705882355e-06, | |
| "loss": 0.6795, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.04744255003706449, | |
| "grad_norm": 2.042616605758667, | |
| "learning_rate": 3.1372549019607846e-06, | |
| "loss": 0.6596, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.050407709414381024, | |
| "grad_norm": 1.781117558479309, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.6325, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.05337286879169755, | |
| "grad_norm": 1.464235782623291, | |
| "learning_rate": 3.529411764705883e-06, | |
| "loss": 0.6265, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.056338028169014086, | |
| "grad_norm": 1.1197887659072876, | |
| "learning_rate": 3.7254901960784316e-06, | |
| "loss": 0.6251, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.05930318754633061, | |
| "grad_norm": 1.1305307149887085, | |
| "learning_rate": 3.92156862745098e-06, | |
| "loss": 0.6271, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06226834692364715, | |
| "grad_norm": 1.1442177295684814, | |
| "learning_rate": 4.11764705882353e-06, | |
| "loss": 0.605, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.06523350630096368, | |
| "grad_norm": 0.8627598881721497, | |
| "learning_rate": 4.313725490196079e-06, | |
| "loss": 0.5979, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0681986656782802, | |
| "grad_norm": 0.9222763776779175, | |
| "learning_rate": 4.509803921568628e-06, | |
| "loss": 0.6027, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.07116382505559674, | |
| "grad_norm": 0.787282407283783, | |
| "learning_rate": 4.705882352941177e-06, | |
| "loss": 0.5928, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.07412898443291327, | |
| "grad_norm": 0.8055775165557861, | |
| "learning_rate": 4.901960784313726e-06, | |
| "loss": 0.5842, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0770941438102298, | |
| "grad_norm": 0.713017463684082, | |
| "learning_rate": 5.098039215686274e-06, | |
| "loss": 0.5694, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.08005930318754632, | |
| "grad_norm": 0.7474880814552307, | |
| "learning_rate": 5.294117647058824e-06, | |
| "loss": 0.5558, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.08302446256486286, | |
| "grad_norm": 0.7316311001777649, | |
| "learning_rate": 5.4901960784313735e-06, | |
| "loss": 0.5629, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.08598962194217939, | |
| "grad_norm": 0.760550856590271, | |
| "learning_rate": 5.686274509803922e-06, | |
| "loss": 0.5574, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.08895478131949593, | |
| "grad_norm": 0.7376196384429932, | |
| "learning_rate": 5.882352941176471e-06, | |
| "loss": 0.5562, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.09191994069681246, | |
| "grad_norm": 0.7215123176574707, | |
| "learning_rate": 6.07843137254902e-06, | |
| "loss": 0.5438, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.09488510007412898, | |
| "grad_norm": 0.7079214453697205, | |
| "learning_rate": 6.274509803921569e-06, | |
| "loss": 0.5418, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.09785025945144551, | |
| "grad_norm": 0.6675574779510498, | |
| "learning_rate": 6.470588235294119e-06, | |
| "loss": 0.5402, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.10081541882876205, | |
| "grad_norm": 0.6604759693145752, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.5344, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.10378057820607858, | |
| "grad_norm": 0.7341721057891846, | |
| "learning_rate": 6.862745098039216e-06, | |
| "loss": 0.5336, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1067457375833951, | |
| "grad_norm": 0.6764810681343079, | |
| "learning_rate": 7.058823529411766e-06, | |
| "loss": 0.5327, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.10971089696071164, | |
| "grad_norm": 0.6292859315872192, | |
| "learning_rate": 7.2549019607843145e-06, | |
| "loss": 0.5275, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.11267605633802817, | |
| "grad_norm": 0.7222408652305603, | |
| "learning_rate": 7.450980392156863e-06, | |
| "loss": 0.5207, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.1156412157153447, | |
| "grad_norm": 0.592737078666687, | |
| "learning_rate": 7.647058823529411e-06, | |
| "loss": 0.5202, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.11860637509266123, | |
| "grad_norm": 0.7391071915626526, | |
| "learning_rate": 7.84313725490196e-06, | |
| "loss": 0.5088, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.12157153446997776, | |
| "grad_norm": 0.5978769659996033, | |
| "learning_rate": 8.03921568627451e-06, | |
| "loss": 0.5059, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.1245366938472943, | |
| "grad_norm": 0.7067713737487793, | |
| "learning_rate": 8.23529411764706e-06, | |
| "loss": 0.5079, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.12750185322461083, | |
| "grad_norm": 0.6121165752410889, | |
| "learning_rate": 8.43137254901961e-06, | |
| "loss": 0.4998, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.13046701260192736, | |
| "grad_norm": 0.7495785355567932, | |
| "learning_rate": 8.627450980392157e-06, | |
| "loss": 0.4877, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.1334321719792439, | |
| "grad_norm": 0.6476943492889404, | |
| "learning_rate": 8.823529411764707e-06, | |
| "loss": 0.4971, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.1363973313565604, | |
| "grad_norm": 0.7655041813850403, | |
| "learning_rate": 9.019607843137256e-06, | |
| "loss": 0.5002, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.13936249073387694, | |
| "grad_norm": 0.6622442007064819, | |
| "learning_rate": 9.215686274509804e-06, | |
| "loss": 0.484, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.14232765011119347, | |
| "grad_norm": 0.7732651233673096, | |
| "learning_rate": 9.411764705882354e-06, | |
| "loss": 0.4922, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.14529280948851, | |
| "grad_norm": 0.6692637205123901, | |
| "learning_rate": 9.607843137254903e-06, | |
| "loss": 0.4733, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.14825796886582654, | |
| "grad_norm": 0.705590546131134, | |
| "learning_rate": 9.803921568627451e-06, | |
| "loss": 0.4734, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15122312824314307, | |
| "grad_norm": 0.6731917858123779, | |
| "learning_rate": 1e-05, | |
| "loss": 0.4651, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.1541882876204596, | |
| "grad_norm": 0.6704531908035278, | |
| "learning_rate": 1.0196078431372549e-05, | |
| "loss": 0.4689, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.15715344699777614, | |
| "grad_norm": 0.6448220610618591, | |
| "learning_rate": 1.03921568627451e-05, | |
| "loss": 0.4675, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.16011860637509265, | |
| "grad_norm": 0.6441836953163147, | |
| "learning_rate": 1.0588235294117648e-05, | |
| "loss": 0.4557, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.16308376575240918, | |
| "grad_norm": 0.7347533106803894, | |
| "learning_rate": 1.0784313725490196e-05, | |
| "loss": 0.4622, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.16604892512972572, | |
| "grad_norm": 0.6999682784080505, | |
| "learning_rate": 1.0980392156862747e-05, | |
| "loss": 0.4446, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.16901408450704225, | |
| "grad_norm": 0.6985459327697754, | |
| "learning_rate": 1.1176470588235295e-05, | |
| "loss": 0.4471, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.17197924388435878, | |
| "grad_norm": 0.7167170643806458, | |
| "learning_rate": 1.1372549019607844e-05, | |
| "loss": 0.4465, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.17494440326167532, | |
| "grad_norm": 0.6770612001419067, | |
| "learning_rate": 1.1568627450980394e-05, | |
| "loss": 0.4374, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.17790956263899185, | |
| "grad_norm": 0.7454700469970703, | |
| "learning_rate": 1.1764705882352942e-05, | |
| "loss": 0.4346, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.1808747220163084, | |
| "grad_norm": 0.726898193359375, | |
| "learning_rate": 1.1960784313725491e-05, | |
| "loss": 0.4287, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.18383988139362492, | |
| "grad_norm": 0.7026724219322205, | |
| "learning_rate": 1.215686274509804e-05, | |
| "loss": 0.4242, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.18680504077094143, | |
| "grad_norm": 1.0427573919296265, | |
| "learning_rate": 1.235294117647059e-05, | |
| "loss": 0.4301, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.18977020014825796, | |
| "grad_norm": 0.9116256833076477, | |
| "learning_rate": 1.2549019607843138e-05, | |
| "loss": 0.4131, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.1927353595255745, | |
| "grad_norm": 0.7025630474090576, | |
| "learning_rate": 1.2745098039215686e-05, | |
| "loss": 0.4175, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.19570051890289103, | |
| "grad_norm": 1.24030339717865, | |
| "learning_rate": 1.2941176470588238e-05, | |
| "loss": 0.4166, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.19866567828020756, | |
| "grad_norm": 0.7674146294593811, | |
| "learning_rate": 1.3137254901960785e-05, | |
| "loss": 0.4042, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.2016308376575241, | |
| "grad_norm": 0.7968058586120605, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.4015, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.20459599703484063, | |
| "grad_norm": 0.9057684540748596, | |
| "learning_rate": 1.3529411764705885e-05, | |
| "loss": 0.3992, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.20756115641215717, | |
| "grad_norm": 0.8404118418693542, | |
| "learning_rate": 1.3725490196078432e-05, | |
| "loss": 0.3974, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.21052631578947367, | |
| "grad_norm": 0.8619468212127686, | |
| "learning_rate": 1.392156862745098e-05, | |
| "loss": 0.4023, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.2134914751667902, | |
| "grad_norm": 0.745784342288971, | |
| "learning_rate": 1.4117647058823532e-05, | |
| "loss": 0.3929, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.21645663454410674, | |
| "grad_norm": 0.8499307632446289, | |
| "learning_rate": 1.431372549019608e-05, | |
| "loss": 0.3827, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.21942179392142327, | |
| "grad_norm": 0.8255784511566162, | |
| "learning_rate": 1.4509803921568629e-05, | |
| "loss": 0.3831, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.2223869532987398, | |
| "grad_norm": 0.8738009333610535, | |
| "learning_rate": 1.4705882352941179e-05, | |
| "loss": 0.377, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.22535211267605634, | |
| "grad_norm": 0.8723142147064209, | |
| "learning_rate": 1.4901960784313726e-05, | |
| "loss": 0.3685, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.22831727205337288, | |
| "grad_norm": 0.8929502964019775, | |
| "learning_rate": 1.5098039215686276e-05, | |
| "loss": 0.3787, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.2312824314306894, | |
| "grad_norm": 1.0882786512374878, | |
| "learning_rate": 1.5294117647058822e-05, | |
| "loss": 0.3652, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.23424759080800592, | |
| "grad_norm": 0.9075109362602234, | |
| "learning_rate": 1.5490196078431373e-05, | |
| "loss": 0.3674, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.23721275018532245, | |
| "grad_norm": 1.1592175960540771, | |
| "learning_rate": 1.568627450980392e-05, | |
| "loss": 0.3644, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.24017790956263899, | |
| "grad_norm": 0.8505756258964539, | |
| "learning_rate": 1.5882352941176473e-05, | |
| "loss": 0.3642, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.24314306893995552, | |
| "grad_norm": 0.9724293947219849, | |
| "learning_rate": 1.607843137254902e-05, | |
| "loss": 0.3467, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.24610822831727205, | |
| "grad_norm": 1.0010569095611572, | |
| "learning_rate": 1.627450980392157e-05, | |
| "loss": 0.3582, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.2490733876945886, | |
| "grad_norm": 0.9776509404182434, | |
| "learning_rate": 1.647058823529412e-05, | |
| "loss": 0.3494, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.2520385470719051, | |
| "grad_norm": 0.9763832688331604, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 0.3487, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.25500370644922166, | |
| "grad_norm": 0.8749181628227234, | |
| "learning_rate": 1.686274509803922e-05, | |
| "loss": 0.3425, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.25796886582653816, | |
| "grad_norm": 0.922757089138031, | |
| "learning_rate": 1.7058823529411767e-05, | |
| "loss": 0.3431, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.2609340252038547, | |
| "grad_norm": 0.8772656321525574, | |
| "learning_rate": 1.7254901960784314e-05, | |
| "loss": 0.3424, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.26389918458117123, | |
| "grad_norm": 0.8626474738121033, | |
| "learning_rate": 1.7450980392156866e-05, | |
| "loss": 0.3351, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.2668643439584878, | |
| "grad_norm": 0.8123406767845154, | |
| "learning_rate": 1.7647058823529414e-05, | |
| "loss": 0.3274, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.2698295033358043, | |
| "grad_norm": 0.8629675507545471, | |
| "learning_rate": 1.7843137254901965e-05, | |
| "loss": 0.3332, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.2727946627131208, | |
| "grad_norm": 0.7453241944313049, | |
| "learning_rate": 1.8039215686274513e-05, | |
| "loss": 0.3264, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.27575982209043737, | |
| "grad_norm": 0.8055425882339478, | |
| "learning_rate": 1.823529411764706e-05, | |
| "loss": 0.3196, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.2787249814677539, | |
| "grad_norm": 0.8176495432853699, | |
| "learning_rate": 1.843137254901961e-05, | |
| "loss": 0.3167, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.28169014084507044, | |
| "grad_norm": 0.7777736186981201, | |
| "learning_rate": 1.862745098039216e-05, | |
| "loss": 0.318, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.28465530022238694, | |
| "grad_norm": 0.8604575395584106, | |
| "learning_rate": 1.8823529411764708e-05, | |
| "loss": 0.3231, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.2876204595997035, | |
| "grad_norm": 0.821183979511261, | |
| "learning_rate": 1.9019607843137255e-05, | |
| "loss": 0.3176, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.29058561897702, | |
| "grad_norm": 0.8958712816238403, | |
| "learning_rate": 1.9215686274509807e-05, | |
| "loss": 0.3155, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.2935507783543366, | |
| "grad_norm": 0.9813326001167297, | |
| "learning_rate": 1.9411764705882355e-05, | |
| "loss": 0.3182, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.2965159377316531, | |
| "grad_norm": 0.9215829968452454, | |
| "learning_rate": 1.9607843137254903e-05, | |
| "loss": 0.3084, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2994810971089696, | |
| "grad_norm": 0.8247601389884949, | |
| "learning_rate": 1.9803921568627454e-05, | |
| "loss": 0.3032, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.30244625648628615, | |
| "grad_norm": 0.8188148736953735, | |
| "learning_rate": 2e-05, | |
| "loss": 0.3059, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.30541141586360265, | |
| "grad_norm": 0.8999500870704651, | |
| "learning_rate": 1.9999940277008807e-05, | |
| "loss": 0.3086, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.3083765752409192, | |
| "grad_norm": 0.8770850300788879, | |
| "learning_rate": 1.99997611087486e-05, | |
| "loss": 0.299, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.3113417346182357, | |
| "grad_norm": 0.8018732070922852, | |
| "learning_rate": 1.9999462497359468e-05, | |
| "loss": 0.3034, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.3143068939955523, | |
| "grad_norm": 0.8308204412460327, | |
| "learning_rate": 1.9999044446408203e-05, | |
| "loss": 0.3009, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.3172720533728688, | |
| "grad_norm": 1.1407383680343628, | |
| "learning_rate": 1.9998506960888258e-05, | |
| "loss": 0.2982, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.3202372127501853, | |
| "grad_norm": 0.8638990521430969, | |
| "learning_rate": 1.999785004721968e-05, | |
| "loss": 0.2957, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.32320237212750186, | |
| "grad_norm": 0.8289093971252441, | |
| "learning_rate": 1.999707371324904e-05, | |
| "loss": 0.3004, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.32616753150481836, | |
| "grad_norm": 1.0886658430099487, | |
| "learning_rate": 1.9996177968249336e-05, | |
| "loss": 0.2953, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.3291326908821349, | |
| "grad_norm": 0.7261621356010437, | |
| "learning_rate": 1.999516282291988e-05, | |
| "loss": 0.2945, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.33209785025945143, | |
| "grad_norm": 0.9758589267730713, | |
| "learning_rate": 1.999402828938618e-05, | |
| "loss": 0.2946, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.335063009636768, | |
| "grad_norm": 0.7505866885185242, | |
| "learning_rate": 1.999277438119978e-05, | |
| "loss": 0.2997, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.3380281690140845, | |
| "grad_norm": 0.801395833492279, | |
| "learning_rate": 1.9991401113338103e-05, | |
| "loss": 0.2885, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.34099332839140106, | |
| "grad_norm": 0.7377513647079468, | |
| "learning_rate": 1.9989908502204295e-05, | |
| "loss": 0.2863, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.34395848776871757, | |
| "grad_norm": 0.728659987449646, | |
| "learning_rate": 1.9988296565626988e-05, | |
| "loss": 0.2863, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.3469236471460341, | |
| "grad_norm": 0.7147101759910583, | |
| "learning_rate": 1.9986565322860117e-05, | |
| "loss": 0.2813, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.34988880652335064, | |
| "grad_norm": 0.7080392837524414, | |
| "learning_rate": 1.9984714794582682e-05, | |
| "loss": 0.281, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.35285396590066714, | |
| "grad_norm": 0.7131238579750061, | |
| "learning_rate": 1.99827450028985e-05, | |
| "loss": 0.2776, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.3558191252779837, | |
| "grad_norm": 0.6940627694129944, | |
| "learning_rate": 1.9980655971335944e-05, | |
| "loss": 0.2814, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.3587842846553002, | |
| "grad_norm": 0.655299186706543, | |
| "learning_rate": 1.9978447724847655e-05, | |
| "loss": 0.2752, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.3617494440326168, | |
| "grad_norm": 0.676629900932312, | |
| "learning_rate": 1.9976120289810247e-05, | |
| "loss": 0.2818, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.3647146034099333, | |
| "grad_norm": 0.6595851182937622, | |
| "learning_rate": 1.9973673694024002e-05, | |
| "loss": 0.2801, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.36767976278724984, | |
| "grad_norm": 0.6573948860168457, | |
| "learning_rate": 1.9971107966712518e-05, | |
| "loss": 0.2829, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.37064492216456635, | |
| "grad_norm": 0.6650752425193787, | |
| "learning_rate": 1.9968423138522382e-05, | |
| "loss": 0.2774, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.37361008154188285, | |
| "grad_norm": 0.6870672106742859, | |
| "learning_rate": 1.996561924152278e-05, | |
| "loss": 0.279, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.3765752409191994, | |
| "grad_norm": 0.6355287432670593, | |
| "learning_rate": 1.9962696309205146e-05, | |
| "loss": 0.2745, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.3795404002965159, | |
| "grad_norm": 0.6908348798751831, | |
| "learning_rate": 1.995965437648273e-05, | |
| "loss": 0.2735, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.3825055596738325, | |
| "grad_norm": 0.6098238229751587, | |
| "learning_rate": 1.995649347969019e-05, | |
| "loss": 0.2658, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.385470719051149, | |
| "grad_norm": 0.6651309728622437, | |
| "learning_rate": 1.995321365658317e-05, | |
| "loss": 0.2696, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.38843587842846555, | |
| "grad_norm": 0.6579580903053284, | |
| "learning_rate": 1.994981494633784e-05, | |
| "loss": 0.2639, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.39140103780578206, | |
| "grad_norm": 0.650133490562439, | |
| "learning_rate": 1.9946297389550433e-05, | |
| "loss": 0.2664, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.39436619718309857, | |
| "grad_norm": 0.6148021221160889, | |
| "learning_rate": 1.9942661028236746e-05, | |
| "loss": 0.2691, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.3973313565604151, | |
| "grad_norm": 0.6839851140975952, | |
| "learning_rate": 1.9938905905831657e-05, | |
| "loss": 0.2619, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.40029651593773163, | |
| "grad_norm": 0.6269260048866272, | |
| "learning_rate": 1.993503206718859e-05, | |
| "loss": 0.2679, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.4032616753150482, | |
| "grad_norm": 0.6698904633522034, | |
| "learning_rate": 1.9931039558578997e-05, | |
| "loss": 0.2737, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.4062268346923647, | |
| "grad_norm": 0.6404738426208496, | |
| "learning_rate": 1.9926928427691788e-05, | |
| "loss": 0.2702, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.40919199406968126, | |
| "grad_norm": 0.6118369698524475, | |
| "learning_rate": 1.992269872363277e-05, | |
| "loss": 0.2644, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.41215715344699777, | |
| "grad_norm": 0.6277154684066772, | |
| "learning_rate": 1.991835049692405e-05, | |
| "loss": 0.2657, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.41512231282431433, | |
| "grad_norm": 0.5819361805915833, | |
| "learning_rate": 1.991388379950346e-05, | |
| "loss": 0.252, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.41808747220163084, | |
| "grad_norm": 0.672166109085083, | |
| "learning_rate": 1.9909298684723905e-05, | |
| "loss": 0.2606, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.42105263157894735, | |
| "grad_norm": 0.5884442925453186, | |
| "learning_rate": 1.9904595207352736e-05, | |
| "loss": 0.2557, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.4240177909562639, | |
| "grad_norm": 0.5893815755844116, | |
| "learning_rate": 1.9899773423571102e-05, | |
| "loss": 0.2595, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.4269829503335804, | |
| "grad_norm": 0.6969826221466064, | |
| "learning_rate": 1.9894833390973266e-05, | |
| "loss": 0.2595, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.429948109710897, | |
| "grad_norm": 0.5909337997436523, | |
| "learning_rate": 1.9889775168565942e-05, | |
| "loss": 0.2522, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.4329132690882135, | |
| "grad_norm": 0.5902915000915527, | |
| "learning_rate": 1.9884598816767563e-05, | |
| "loss": 0.2532, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.43587842846553004, | |
| "grad_norm": 0.6261239647865295, | |
| "learning_rate": 1.987930439740757e-05, | |
| "loss": 0.2566, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.43884358784284655, | |
| "grad_norm": 0.579250156879425, | |
| "learning_rate": 1.9873891973725673e-05, | |
| "loss": 0.2587, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.44180874722016306, | |
| "grad_norm": 0.5678402185440063, | |
| "learning_rate": 1.98683616103711e-05, | |
| "loss": 0.2494, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.4447739065974796, | |
| "grad_norm": 0.6142150163650513, | |
| "learning_rate": 1.986271337340182e-05, | |
| "loss": 0.2507, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.4477390659747961, | |
| "grad_norm": 0.6132687926292419, | |
| "learning_rate": 1.9856947330283752e-05, | |
| "loss": 0.2547, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.4507042253521127, | |
| "grad_norm": 0.5993427038192749, | |
| "learning_rate": 1.985106354988997e-05, | |
| "loss": 0.2478, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.4536693847294292, | |
| "grad_norm": 0.6638728380203247, | |
| "learning_rate": 1.984506210249986e-05, | |
| "loss": 0.2547, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.45663454410674575, | |
| "grad_norm": 0.6074317097663879, | |
| "learning_rate": 1.9838943059798305e-05, | |
| "loss": 0.2521, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.45959970348406226, | |
| "grad_norm": 0.6486067175865173, | |
| "learning_rate": 1.9832706494874812e-05, | |
| "loss": 0.2562, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.4625648628613788, | |
| "grad_norm": 0.6348186135292053, | |
| "learning_rate": 1.982635248222264e-05, | |
| "loss": 0.2528, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.46553002223869533, | |
| "grad_norm": 0.5568612217903137, | |
| "learning_rate": 1.9819881097737917e-05, | |
| "loss": 0.2471, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.46849518161601184, | |
| "grad_norm": 0.5930222272872925, | |
| "learning_rate": 1.9813292418718734e-05, | |
| "loss": 0.2434, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.4714603409933284, | |
| "grad_norm": 0.6412246823310852, | |
| "learning_rate": 1.9806586523864212e-05, | |
| "loss": 0.2482, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.4744255003706449, | |
| "grad_norm": 0.5488153696060181, | |
| "learning_rate": 1.9799763493273572e-05, | |
| "loss": 0.2416, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.47739065974796147, | |
| "grad_norm": 0.6217798590660095, | |
| "learning_rate": 1.9792823408445173e-05, | |
| "loss": 0.2508, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.48035581912527797, | |
| "grad_norm": 0.5728364586830139, | |
| "learning_rate": 1.978576635227554e-05, | |
| "loss": 0.2488, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.48332097850259453, | |
| "grad_norm": 0.6427583694458008, | |
| "learning_rate": 1.9778592409058376e-05, | |
| "loss": 0.2483, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.48628613787991104, | |
| "grad_norm": 0.6554081439971924, | |
| "learning_rate": 1.9771301664483548e-05, | |
| "loss": 0.2426, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.4892512972572276, | |
| "grad_norm": 0.5885781049728394, | |
| "learning_rate": 1.976389420563607e-05, | |
| "loss": 0.2551, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.4922164566345441, | |
| "grad_norm": 0.5944969058036804, | |
| "learning_rate": 1.975637012099507e-05, | |
| "loss": 0.2466, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.4951816160118606, | |
| "grad_norm": 0.6138084530830383, | |
| "learning_rate": 1.97487295004327e-05, | |
| "loss": 0.2414, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.4981467753891772, | |
| "grad_norm": 0.5585880279541016, | |
| "learning_rate": 1.9740972435213114e-05, | |
| "loss": 0.2352, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.5011119347664937, | |
| "grad_norm": 0.6357447504997253, | |
| "learning_rate": 1.9733099017991342e-05, | |
| "loss": 0.2454, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.5040770941438102, | |
| "grad_norm": 0.5774321556091309, | |
| "learning_rate": 1.972510934281218e-05, | |
| "loss": 0.2424, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.5070422535211268, | |
| "grad_norm": 0.6422623991966248, | |
| "learning_rate": 1.9717003505109097e-05, | |
| "loss": 0.2361, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.5100074128984433, | |
| "grad_norm": 0.5912754535675049, | |
| "learning_rate": 1.9708781601703066e-05, | |
| "loss": 0.243, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.5129725722757599, | |
| "grad_norm": 0.5881178379058838, | |
| "learning_rate": 1.9700443730801412e-05, | |
| "loss": 0.2394, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.5159377316530763, | |
| "grad_norm": 0.6363380551338196, | |
| "learning_rate": 1.9691989991996663e-05, | |
| "loss": 0.2407, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.5189028910303929, | |
| "grad_norm": 0.55989670753479, | |
| "learning_rate": 1.9683420486265328e-05, | |
| "loss": 0.2438, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.5218680504077094, | |
| "grad_norm": 0.6781154274940491, | |
| "learning_rate": 1.967473531596671e-05, | |
| "loss": 0.2424, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.5248332097850259, | |
| "grad_norm": 0.5050660967826843, | |
| "learning_rate": 1.966593458484168e-05, | |
| "loss": 0.2341, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.5277983691623425, | |
| "grad_norm": 0.6881943345069885, | |
| "learning_rate": 1.9657018398011435e-05, | |
| "loss": 0.2433, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.530763528539659, | |
| "grad_norm": 0.553970992565155, | |
| "learning_rate": 1.9647986861976246e-05, | |
| "loss": 0.237, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.5337286879169756, | |
| "grad_norm": 0.6539415121078491, | |
| "learning_rate": 1.9638840084614182e-05, | |
| "loss": 0.238, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.536693847294292, | |
| "grad_norm": 0.5665425658226013, | |
| "learning_rate": 1.9629578175179823e-05, | |
| "loss": 0.2399, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.5396590066716086, | |
| "grad_norm": 0.6046749949455261, | |
| "learning_rate": 1.9620201244302952e-05, | |
| "loss": 0.2359, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.5426241660489252, | |
| "grad_norm": 0.6772344708442688, | |
| "learning_rate": 1.9610709403987248e-05, | |
| "loss": 0.2382, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.5455893254262416, | |
| "grad_norm": 0.473206490278244, | |
| "learning_rate": 1.9601102767608924e-05, | |
| "loss": 0.2321, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.5485544848035582, | |
| "grad_norm": 0.6189218163490295, | |
| "learning_rate": 1.95913814499154e-05, | |
| "loss": 0.2356, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.5515196441808747, | |
| "grad_norm": 0.5345617532730103, | |
| "learning_rate": 1.95815455670239e-05, | |
| "loss": 0.2394, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.5544848035581913, | |
| "grad_norm": 0.5871132016181946, | |
| "learning_rate": 1.9571595236420103e-05, | |
| "loss": 0.2359, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.5574499629355077, | |
| "grad_norm": 0.5409566760063171, | |
| "learning_rate": 1.9561530576956703e-05, | |
| "loss": 0.2396, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.5604151223128243, | |
| "grad_norm": 0.5904874205589294, | |
| "learning_rate": 1.955135170885202e-05, | |
| "loss": 0.2361, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.5633802816901409, | |
| "grad_norm": 0.5407031178474426, | |
| "learning_rate": 1.9541058753688538e-05, | |
| "loss": 0.2368, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.5663454410674573, | |
| "grad_norm": 0.5759615302085876, | |
| "learning_rate": 1.9530651834411477e-05, | |
| "loss": 0.2358, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.5693106004447739, | |
| "grad_norm": 0.6436863541603088, | |
| "learning_rate": 1.95201310753273e-05, | |
| "loss": 0.2299, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.5722757598220904, | |
| "grad_norm": 0.5067325830459595, | |
| "learning_rate": 1.9509496602102253e-05, | |
| "loss": 0.2275, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.575240919199407, | |
| "grad_norm": 0.5916472673416138, | |
| "learning_rate": 1.9498748541760845e-05, | |
| "loss": 0.229, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.5782060785767235, | |
| "grad_norm": 0.49817144870758057, | |
| "learning_rate": 1.9487887022684336e-05, | |
| "loss": 0.2277, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.58117123795404, | |
| "grad_norm": 0.6111854910850525, | |
| "learning_rate": 1.947691217460921e-05, | |
| "loss": 0.2395, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.5841363973313566, | |
| "grad_norm": 0.524508535861969, | |
| "learning_rate": 1.946582412862562e-05, | |
| "loss": 0.2372, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.5871015567086731, | |
| "grad_norm": 0.5496771335601807, | |
| "learning_rate": 1.9454623017175814e-05, | |
| "loss": 0.2338, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.5900667160859896, | |
| "grad_norm": 0.5417652726173401, | |
| "learning_rate": 1.9443308974052574e-05, | |
| "loss": 0.2328, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.5930318754633062, | |
| "grad_norm": 0.49683743715286255, | |
| "learning_rate": 1.9431882134397596e-05, | |
| "loss": 0.2289, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5959970348406227, | |
| "grad_norm": 0.5067436099052429, | |
| "learning_rate": 1.9420342634699893e-05, | |
| "loss": 0.2303, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.5989621942179392, | |
| "grad_norm": 0.532744288444519, | |
| "learning_rate": 1.9408690612794146e-05, | |
| "loss": 0.2219, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.6019273535952557, | |
| "grad_norm": 0.5270218253135681, | |
| "learning_rate": 1.9396926207859085e-05, | |
| "loss": 0.2324, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.6048925129725723, | |
| "grad_norm": 0.4947966933250427, | |
| "learning_rate": 1.9385049560415794e-05, | |
| "loss": 0.2282, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.6078576723498889, | |
| "grad_norm": 0.5205817222595215, | |
| "learning_rate": 1.9373060812326053e-05, | |
| "loss": 0.2279, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.6108228317272053, | |
| "grad_norm": 0.5304152369499207, | |
| "learning_rate": 1.9360960106790645e-05, | |
| "loss": 0.2288, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.6137879911045219, | |
| "grad_norm": 0.49558138847351074, | |
| "learning_rate": 1.9348747588347637e-05, | |
| "loss": 0.2284, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.6167531504818384, | |
| "grad_norm": 0.48547008633613586, | |
| "learning_rate": 1.9336423402870655e-05, | |
| "loss": 0.2297, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.6197183098591549, | |
| "grad_norm": 0.5189692974090576, | |
| "learning_rate": 1.932398769756714e-05, | |
| "loss": 0.2293, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.6226834692364714, | |
| "grad_norm": 0.5088484287261963, | |
| "learning_rate": 1.9311440620976597e-05, | |
| "loss": 0.2311, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.625648628613788, | |
| "grad_norm": 0.5324704051017761, | |
| "learning_rate": 1.9298782322968817e-05, | |
| "loss": 0.2377, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.6286137879911046, | |
| "grad_norm": 0.5019773840904236, | |
| "learning_rate": 1.9286012954742078e-05, | |
| "loss": 0.2256, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.631578947368421, | |
| "grad_norm": 0.5624535083770752, | |
| "learning_rate": 1.9273132668821363e-05, | |
| "loss": 0.2291, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.6345441067457376, | |
| "grad_norm": 0.5227831602096558, | |
| "learning_rate": 1.9260141619056507e-05, | |
| "loss": 0.2268, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.6375092661230541, | |
| "grad_norm": 0.5904820561408997, | |
| "learning_rate": 1.924703996062038e-05, | |
| "loss": 0.227, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.6404744255003706, | |
| "grad_norm": 0.561266303062439, | |
| "learning_rate": 1.9233827850007028e-05, | |
| "loss": 0.2294, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.6434395848776872, | |
| "grad_norm": 0.5293812155723572, | |
| "learning_rate": 1.9220505445029803e-05, | |
| "loss": 0.2228, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.6464047442550037, | |
| "grad_norm": 0.5227711200714111, | |
| "learning_rate": 1.9207072904819484e-05, | |
| "loss": 0.2261, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.6493699036323203, | |
| "grad_norm": 0.5241237282752991, | |
| "learning_rate": 1.9193530389822364e-05, | |
| "loss": 0.2247, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.6523350630096367, | |
| "grad_norm": 0.5190705060958862, | |
| "learning_rate": 1.9179878061798347e-05, | |
| "loss": 0.2266, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.6553002223869533, | |
| "grad_norm": 0.4801787734031677, | |
| "learning_rate": 1.9166116083819002e-05, | |
| "loss": 0.2211, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.6582653817642699, | |
| "grad_norm": 0.5298479795455933, | |
| "learning_rate": 1.915224462026563e-05, | |
| "loss": 0.2145, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.6612305411415864, | |
| "grad_norm": 0.5878245830535889, | |
| "learning_rate": 1.913826383682729e-05, | |
| "loss": 0.2249, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.6641957005189029, | |
| "grad_norm": 0.4641963839530945, | |
| "learning_rate": 1.912417390049882e-05, | |
| "loss": 0.2195, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.6671608598962194, | |
| "grad_norm": 0.4989553391933441, | |
| "learning_rate": 1.9109974979578852e-05, | |
| "loss": 0.2306, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.670126019273536, | |
| "grad_norm": 0.5732155442237854, | |
| "learning_rate": 1.909566724366779e-05, | |
| "loss": 0.2246, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.6730911786508524, | |
| "grad_norm": 0.5080471038818359, | |
| "learning_rate": 1.9081250863665794e-05, | |
| "loss": 0.2253, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.676056338028169, | |
| "grad_norm": 0.5161991119384766, | |
| "learning_rate": 1.9066726011770725e-05, | |
| "loss": 0.2248, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.6790214974054856, | |
| "grad_norm": 0.5189105868339539, | |
| "learning_rate": 1.905209286147611e-05, | |
| "loss": 0.227, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.6819866567828021, | |
| "grad_norm": 0.5306798219680786, | |
| "learning_rate": 1.903735158756905e-05, | |
| "loss": 0.2253, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.6849518161601186, | |
| "grad_norm": 0.523923933506012, | |
| "learning_rate": 1.9022502366128136e-05, | |
| "loss": 0.2295, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.6879169755374351, | |
| "grad_norm": 0.5236137509346008, | |
| "learning_rate": 1.9007545374521354e-05, | |
| "loss": 0.222, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.6908821349147517, | |
| "grad_norm": 0.5138505697250366, | |
| "learning_rate": 1.8992480791403957e-05, | |
| "loss": 0.2143, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.6938472942920682, | |
| "grad_norm": 0.5385280251502991, | |
| "learning_rate": 1.897730879671634e-05, | |
| "loss": 0.2227, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.6968124536693847, | |
| "grad_norm": 0.5067414045333862, | |
| "learning_rate": 1.8962029571681887e-05, | |
| "loss": 0.2223, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.6997776130467013, | |
| "grad_norm": 0.4815332591533661, | |
| "learning_rate": 1.8946643298804794e-05, | |
| "loss": 0.2188, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.7027427724240178, | |
| "grad_norm": 0.4668591618537903, | |
| "learning_rate": 1.8931150161867917e-05, | |
| "loss": 0.2206, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.7057079318013343, | |
| "grad_norm": 0.5026832222938538, | |
| "learning_rate": 1.891555034593055e-05, | |
| "loss": 0.2228, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.7086730911786508, | |
| "grad_norm": 0.5014287233352661, | |
| "learning_rate": 1.8899844037326227e-05, | |
| "loss": 0.216, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.7116382505559674, | |
| "grad_norm": 0.4586634933948517, | |
| "learning_rate": 1.8884031423660492e-05, | |
| "loss": 0.2206, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.7146034099332839, | |
| "grad_norm": 0.500434398651123, | |
| "learning_rate": 1.8868112693808664e-05, | |
| "loss": 0.2163, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.7175685693106004, | |
| "grad_norm": 0.46279287338256836, | |
| "learning_rate": 1.8852088037913577e-05, | |
| "loss": 0.2161, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.720533728687917, | |
| "grad_norm": 0.5185891389846802, | |
| "learning_rate": 1.8835957647383304e-05, | |
| "loss": 0.2221, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.7234988880652335, | |
| "grad_norm": 0.48801976442337036, | |
| "learning_rate": 1.8819721714888878e-05, | |
| "loss": 0.225, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.72646404744255, | |
| "grad_norm": 0.4899084270000458, | |
| "learning_rate": 1.8803380434362e-05, | |
| "loss": 0.2169, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.7294292068198666, | |
| "grad_norm": 0.5264920592308044, | |
| "learning_rate": 1.878693400099269e-05, | |
| "loss": 0.2207, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.7323943661971831, | |
| "grad_norm": 0.48303139209747314, | |
| "learning_rate": 1.877038261122699e-05, | |
| "loss": 0.2244, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.7353595255744997, | |
| "grad_norm": 0.46109214425086975, | |
| "learning_rate": 1.87537264627646e-05, | |
| "loss": 0.216, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.7383246849518161, | |
| "grad_norm": 0.4971975088119507, | |
| "learning_rate": 1.8736965754556527e-05, | |
| "loss": 0.2235, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.7412898443291327, | |
| "grad_norm": 0.4700891077518463, | |
| "learning_rate": 1.8720100686802693e-05, | |
| "loss": 0.2175, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.7442550037064493, | |
| "grad_norm": 0.45833539962768555, | |
| "learning_rate": 1.8703131460949555e-05, | |
| "loss": 0.216, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.7472201630837657, | |
| "grad_norm": 0.47551876306533813, | |
| "learning_rate": 1.86860582796877e-05, | |
| "loss": 0.2222, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.7501853224610823, | |
| "grad_norm": 0.4569433629512787, | |
| "learning_rate": 1.866888134694942e-05, | |
| "loss": 0.2165, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.7531504818383988, | |
| "grad_norm": 0.43670737743377686, | |
| "learning_rate": 1.865160086790627e-05, | |
| "loss": 0.2128, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.7561156412157154, | |
| "grad_norm": 0.517746090888977, | |
| "learning_rate": 1.8634217048966638e-05, | |
| "loss": 0.2149, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.7590808005930318, | |
| "grad_norm": 0.46699458360671997, | |
| "learning_rate": 1.861673009777325e-05, | |
| "loss": 0.2187, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.7620459599703484, | |
| "grad_norm": 0.46238595247268677, | |
| "learning_rate": 1.8599140223200716e-05, | |
| "loss": 0.2137, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.765011119347665, | |
| "grad_norm": 0.47764065861701965, | |
| "learning_rate": 1.858144763535302e-05, | |
| "loss": 0.2221, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.7679762787249814, | |
| "grad_norm": 0.4717821180820465, | |
| "learning_rate": 1.8563652545561014e-05, | |
| "loss": 0.2188, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.770941438102298, | |
| "grad_norm": 0.4471701383590698, | |
| "learning_rate": 1.8545755166379898e-05, | |
| "loss": 0.2171, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.7739065974796145, | |
| "grad_norm": 0.49311378598213196, | |
| "learning_rate": 1.852775571158668e-05, | |
| "loss": 0.2157, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.7768717568569311, | |
| "grad_norm": 0.4882054924964905, | |
| "learning_rate": 1.850965439617761e-05, | |
| "loss": 0.2167, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.7798369162342476, | |
| "grad_norm": 0.45021718740463257, | |
| "learning_rate": 1.8491451436365628e-05, | |
| "loss": 0.2191, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.7828020756115641, | |
| "grad_norm": 0.5516721606254578, | |
| "learning_rate": 1.8473147049577777e-05, | |
| "loss": 0.2152, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.7857672349888807, | |
| "grad_norm": 0.4654419422149658, | |
| "learning_rate": 1.8454741454452604e-05, | |
| "loss": 0.2216, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.7887323943661971, | |
| "grad_norm": 0.4703727066516876, | |
| "learning_rate": 1.843623487083755e-05, | |
| "loss": 0.2164, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.7916975537435137, | |
| "grad_norm": 0.479714959859848, | |
| "learning_rate": 1.8417627519786317e-05, | |
| "loss": 0.2152, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.7946627131208303, | |
| "grad_norm": 0.4948756992816925, | |
| "learning_rate": 1.839891962355624e-05, | |
| "loss": 0.2219, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.7976278724981468, | |
| "grad_norm": 0.45587557554244995, | |
| "learning_rate": 1.838011140560562e-05, | |
| "loss": 0.2157, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.8005930318754633, | |
| "grad_norm": 0.46080151200294495, | |
| "learning_rate": 1.836120309059107e-05, | |
| "loss": 0.2122, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.8035581912527798, | |
| "grad_norm": 0.4493560492992401, | |
| "learning_rate": 1.8342194904364815e-05, | |
| "loss": 0.2163, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.8065233506300964, | |
| "grad_norm": 0.4825652539730072, | |
| "learning_rate": 1.8323087073971996e-05, | |
| "loss": 0.2116, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.809488510007413, | |
| "grad_norm": 0.4308413863182068, | |
| "learning_rate": 1.8303879827647977e-05, | |
| "loss": 0.2172, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.8124536693847294, | |
| "grad_norm": 0.508596658706665, | |
| "learning_rate": 1.8284573394815596e-05, | |
| "loss": 0.2186, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.815418828762046, | |
| "grad_norm": 0.4650067090988159, | |
| "learning_rate": 1.826516800608244e-05, | |
| "loss": 0.2069, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.8183839881393625, | |
| "grad_norm": 0.42739060521125793, | |
| "learning_rate": 1.8245663893238075e-05, | |
| "loss": 0.2102, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.821349147516679, | |
| "grad_norm": 0.46640655398368835, | |
| "learning_rate": 1.8226061289251297e-05, | |
| "loss": 0.2145, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.8243143068939955, | |
| "grad_norm": 0.4410681426525116, | |
| "learning_rate": 1.8206360428267332e-05, | |
| "loss": 0.2131, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.8272794662713121, | |
| "grad_norm": 0.44091495871543884, | |
| "learning_rate": 1.8186561545605055e-05, | |
| "loss": 0.2122, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.8302446256486287, | |
| "grad_norm": 0.4652099311351776, | |
| "learning_rate": 1.816666487775416e-05, | |
| "loss": 0.2179, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.8332097850259451, | |
| "grad_norm": 0.4468926787376404, | |
| "learning_rate": 1.8146670662372353e-05, | |
| "loss": 0.219, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.8361749444032617, | |
| "grad_norm": 0.4693123400211334, | |
| "learning_rate": 1.8126579138282502e-05, | |
| "loss": 0.2145, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.8391401037805782, | |
| "grad_norm": 0.43998247385025024, | |
| "learning_rate": 1.8106390545469797e-05, | |
| "loss": 0.212, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.8421052631578947, | |
| "grad_norm": 0.4576677978038788, | |
| "learning_rate": 1.8086105125078858e-05, | |
| "loss": 0.2141, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.8450704225352113, | |
| "grad_norm": 0.42104509472846985, | |
| "learning_rate": 1.8065723119410885e-05, | |
| "loss": 0.2126, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.8480355819125278, | |
| "grad_norm": 0.4544185996055603, | |
| "learning_rate": 1.804524477192075e-05, | |
| "loss": 0.2122, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.8510007412898444, | |
| "grad_norm": 0.4285774528980255, | |
| "learning_rate": 1.8024670327214084e-05, | |
| "loss": 0.211, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.8539659006671608, | |
| "grad_norm": 0.43197640776634216, | |
| "learning_rate": 1.8004000031044363e-05, | |
| "loss": 0.2103, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.8569310600444774, | |
| "grad_norm": 0.4368259906768799, | |
| "learning_rate": 1.798323413030997e-05, | |
| "loss": 0.2134, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.859896219421794, | |
| "grad_norm": 0.4898151159286499, | |
| "learning_rate": 1.796237287305125e-05, | |
| "loss": 0.2137, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.8628613787991104, | |
| "grad_norm": 0.42249011993408203, | |
| "learning_rate": 1.7941416508447537e-05, | |
| "loss": 0.2052, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.865826538176427, | |
| "grad_norm": 0.45801860094070435, | |
| "learning_rate": 1.792036528681418e-05, | |
| "loss": 0.2146, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.8687916975537435, | |
| "grad_norm": 0.44352859258651733, | |
| "learning_rate": 1.789921945959958e-05, | |
| "loss": 0.2053, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.8717568569310601, | |
| "grad_norm": 0.4158633351325989, | |
| "learning_rate": 1.7877979279382135e-05, | |
| "loss": 0.2137, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.8747220163083765, | |
| "grad_norm": 0.41102075576782227, | |
| "learning_rate": 1.7856644999867264e-05, | |
| "loss": 0.2109, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.8776871756856931, | |
| "grad_norm": 0.41784408688545227, | |
| "learning_rate": 1.783521687588437e-05, | |
| "loss": 0.2128, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.8806523350630097, | |
| "grad_norm": 0.4097442626953125, | |
| "learning_rate": 1.781369516338378e-05, | |
| "loss": 0.2116, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.8836174944403261, | |
| "grad_norm": 0.4172267019748688, | |
| "learning_rate": 1.779208011943371e-05, | |
| "loss": 0.2096, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.8865826538176427, | |
| "grad_norm": 0.4201764464378357, | |
| "learning_rate": 1.777037200221717e-05, | |
| "loss": 0.2144, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.8895478131949592, | |
| "grad_norm": 0.4283645451068878, | |
| "learning_rate": 1.77485710710289e-05, | |
| "loss": 0.2159, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.8925129725722758, | |
| "grad_norm": 0.4021233022212982, | |
| "learning_rate": 1.7726677586272263e-05, | |
| "loss": 0.2147, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.8954781319495922, | |
| "grad_norm": 0.4146812856197357, | |
| "learning_rate": 1.7704691809456142e-05, | |
| "loss": 0.2136, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.8984432913269088, | |
| "grad_norm": 0.41466352343559265, | |
| "learning_rate": 1.7682614003191807e-05, | |
| "loss": 0.2117, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.9014084507042254, | |
| "grad_norm": 0.45098355412483215, | |
| "learning_rate": 1.766044443118978e-05, | |
| "loss": 0.2141, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.9043736100815419, | |
| "grad_norm": 0.39802679419517517, | |
| "learning_rate": 1.76381833582567e-05, | |
| "loss": 0.2119, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.9073387694588584, | |
| "grad_norm": 0.4417196214199066, | |
| "learning_rate": 1.761583105029213e-05, | |
| "loss": 0.2148, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.910303928836175, | |
| "grad_norm": 0.4523768723011017, | |
| "learning_rate": 1.7593387774285412e-05, | |
| "loss": 0.2116, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.9132690882134915, | |
| "grad_norm": 0.42361876368522644, | |
| "learning_rate": 1.7570853798312462e-05, | |
| "loss": 0.2091, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.916234247590808, | |
| "grad_norm": 0.44734466075897217, | |
| "learning_rate": 1.7548229391532572e-05, | |
| "loss": 0.2098, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.9191994069681245, | |
| "grad_norm": 0.4427475333213806, | |
| "learning_rate": 1.7525514824185187e-05, | |
| "loss": 0.2159, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.9221645663454411, | |
| "grad_norm": 0.4229927659034729, | |
| "learning_rate": 1.750271036758669e-05, | |
| "loss": 0.2104, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.9251297257227576, | |
| "grad_norm": 0.4121291935443878, | |
| "learning_rate": 1.747981629412715e-05, | |
| "loss": 0.2076, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.9280948851000741, | |
| "grad_norm": 0.45084404945373535, | |
| "learning_rate": 1.7456832877267083e-05, | |
| "loss": 0.215, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.9310600444773907, | |
| "grad_norm": 0.423123836517334, | |
| "learning_rate": 1.7433760391534166e-05, | |
| "loss": 0.2082, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.9340252038547072, | |
| "grad_norm": 0.4547256827354431, | |
| "learning_rate": 1.741059911251997e-05, | |
| "loss": 0.2089, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.9369903632320237, | |
| "grad_norm": 0.4248969852924347, | |
| "learning_rate": 1.7387349316876668e-05, | |
| "loss": 0.2039, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.9399555226093402, | |
| "grad_norm": 0.46414193511009216, | |
| "learning_rate": 1.7364011282313732e-05, | |
| "loss": 0.2081, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.9429206819866568, | |
| "grad_norm": 0.4844679534435272, | |
| "learning_rate": 1.7340585287594605e-05, | |
| "loss": 0.2142, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.9458858413639734, | |
| "grad_norm": 0.4147413372993469, | |
| "learning_rate": 1.731707161253338e-05, | |
| "loss": 0.2128, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.9488510007412898, | |
| "grad_norm": 0.4431176781654358, | |
| "learning_rate": 1.7293470537991463e-05, | |
| "loss": 0.2104, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.9518161601186064, | |
| "grad_norm": 0.45323607325553894, | |
| "learning_rate": 1.7269782345874204e-05, | |
| "loss": 0.2083, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.9547813194959229, | |
| "grad_norm": 0.4210136830806732, | |
| "learning_rate": 1.7246007319127547e-05, | |
| "loss": 0.2069, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.9577464788732394, | |
| "grad_norm": 0.440244197845459, | |
| "learning_rate": 1.7222145741734625e-05, | |
| "loss": 0.2021, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.9607116382505559, | |
| "grad_norm": 0.41491949558258057, | |
| "learning_rate": 1.7198197898712402e-05, | |
| "loss": 0.2086, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.9636767976278725, | |
| "grad_norm": 0.4270980954170227, | |
| "learning_rate": 1.717416407610824e-05, | |
| "loss": 0.2063, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.9666419570051891, | |
| "grad_norm": 0.436722993850708, | |
| "learning_rate": 1.7150044560996488e-05, | |
| "loss": 0.2095, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.9696071163825055, | |
| "grad_norm": 0.42856717109680176, | |
| "learning_rate": 1.7125839641475074e-05, | |
| "loss": 0.2151, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.9725722757598221, | |
| "grad_norm": 0.4263397753238678, | |
| "learning_rate": 1.7101549606662025e-05, | |
| "loss": 0.21, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.9755374351371386, | |
| "grad_norm": 0.43046820163726807, | |
| "learning_rate": 1.7077174746692054e-05, | |
| "loss": 0.211, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.9785025945144552, | |
| "grad_norm": 0.4144728481769562, | |
| "learning_rate": 1.7052715352713076e-05, | |
| "loss": 0.2069, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.9814677538917717, | |
| "grad_norm": 0.4112738072872162, | |
| "learning_rate": 1.7028171716882714e-05, | |
| "loss": 0.209, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.9844329132690882, | |
| "grad_norm": 0.4484747052192688, | |
| "learning_rate": 1.7003544132364847e-05, | |
| "loss": 0.2118, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.9873980726464048, | |
| "grad_norm": 0.4388020634651184, | |
| "learning_rate": 1.6978832893326074e-05, | |
| "loss": 0.2069, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.9903632320237212, | |
| "grad_norm": 0.45029163360595703, | |
| "learning_rate": 1.6954038294932215e-05, | |
| "loss": 0.2153, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.9933283914010378, | |
| "grad_norm": 0.4059215486049652, | |
| "learning_rate": 1.692916063334479e-05, | |
| "loss": 0.1999, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.9962935507783544, | |
| "grad_norm": 0.430908739566803, | |
| "learning_rate": 1.690420020571747e-05, | |
| "loss": 0.2101, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.9992587101556709, | |
| "grad_norm": 0.4230971336364746, | |
| "learning_rate": 1.6879157310192537e-05, | |
| "loss": 0.2033, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 1.0022238695329875, | |
| "grad_norm": 0.37717196345329285, | |
| "learning_rate": 1.685403224589731e-05, | |
| "loss": 0.1831, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 1.005189028910304, | |
| "grad_norm": 0.4386158287525177, | |
| "learning_rate": 1.6828825312940594e-05, | |
| "loss": 0.1782, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 1.0081541882876204, | |
| "grad_norm": 0.3862016201019287, | |
| "learning_rate": 1.6803536812409077e-05, | |
| "loss": 0.1779, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.011119347664937, | |
| "grad_norm": 0.4159914553165436, | |
| "learning_rate": 1.6778167046363735e-05, | |
| "loss": 0.1699, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 1.0140845070422535, | |
| "grad_norm": 0.5072054266929626, | |
| "learning_rate": 1.675271631783623e-05, | |
| "loss": 0.1738, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 1.01704966641957, | |
| "grad_norm": 0.41934165358543396, | |
| "learning_rate": 1.672718493082529e-05, | |
| "loss": 0.1722, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 1.0200148257968866, | |
| "grad_norm": 0.4099801480770111, | |
| "learning_rate": 1.6701573190293076e-05, | |
| "loss": 0.1713, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 1.0229799851742032, | |
| "grad_norm": 0.44231241941452026, | |
| "learning_rate": 1.667588140216154e-05, | |
| "loss": 0.1675, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 1.0259451445515197, | |
| "grad_norm": 0.4088985323905945, | |
| "learning_rate": 1.6650109873308763e-05, | |
| "loss": 0.1736, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 1.028910303928836, | |
| "grad_norm": 0.4394180476665497, | |
| "learning_rate": 1.6624258911565312e-05, | |
| "loss": 0.1727, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 1.0318754633061527, | |
| "grad_norm": 0.4399167001247406, | |
| "learning_rate": 1.6598328825710536e-05, | |
| "loss": 0.1732, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 1.0348406226834692, | |
| "grad_norm": 0.46241313219070435, | |
| "learning_rate": 1.6572319925468892e-05, | |
| "loss": 0.1759, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 1.0378057820607858, | |
| "grad_norm": 0.40860143303871155, | |
| "learning_rate": 1.654623252150624e-05, | |
| "loss": 0.1711, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.0407709414381023, | |
| "grad_norm": 0.4109824597835541, | |
| "learning_rate": 1.6520066925426146e-05, | |
| "loss": 0.1799, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 1.043736100815419, | |
| "grad_norm": 0.40983447432518005, | |
| "learning_rate": 1.6493823449766137e-05, | |
| "loss": 0.1752, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 1.0467012601927355, | |
| "grad_norm": 0.4187794029712677, | |
| "learning_rate": 1.6467502407993995e-05, | |
| "loss": 0.1753, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 1.0496664195700518, | |
| "grad_norm": 0.40739187598228455, | |
| "learning_rate": 1.644110411450398e-05, | |
| "loss": 0.1771, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 1.0526315789473684, | |
| "grad_norm": 0.41065889596939087, | |
| "learning_rate": 1.6414628884613106e-05, | |
| "loss": 0.1711, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 1.055596738324685, | |
| "grad_norm": 0.43635791540145874, | |
| "learning_rate": 1.6388077034557355e-05, | |
| "loss": 0.175, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 1.0585618977020015, | |
| "grad_norm": 0.432016521692276, | |
| "learning_rate": 1.6361448881487913e-05, | |
| "loss": 0.1754, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 1.061527057079318, | |
| "grad_norm": 0.43051794171333313, | |
| "learning_rate": 1.6334744743467366e-05, | |
| "loss": 0.177, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 1.0644922164566346, | |
| "grad_norm": 0.39719873666763306, | |
| "learning_rate": 1.6307964939465914e-05, | |
| "loss": 0.1732, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 1.0674573758339512, | |
| "grad_norm": 0.40763285756111145, | |
| "learning_rate": 1.628110978935756e-05, | |
| "loss": 0.1744, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.0704225352112675, | |
| "grad_norm": 0.40124091506004333, | |
| "learning_rate": 1.625417961391628e-05, | |
| "loss": 0.1759, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 1.073387694588584, | |
| "grad_norm": 0.41654643416404724, | |
| "learning_rate": 1.62271747348122e-05, | |
| "loss": 0.1751, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 1.0763528539659006, | |
| "grad_norm": 0.39688020944595337, | |
| "learning_rate": 1.6200095474607753e-05, | |
| "loss": 0.1704, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 1.0793180133432172, | |
| "grad_norm": 0.3920522928237915, | |
| "learning_rate": 1.6172942156753822e-05, | |
| "loss": 0.168, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 1.0822831727205338, | |
| "grad_norm": 0.4264538586139679, | |
| "learning_rate": 1.614571510558588e-05, | |
| "loss": 0.174, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 1.0852483320978503, | |
| "grad_norm": 0.3995387554168701, | |
| "learning_rate": 1.6118414646320115e-05, | |
| "loss": 0.1718, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 1.0882134914751669, | |
| "grad_norm": 0.36994609236717224, | |
| "learning_rate": 1.6091041105049542e-05, | |
| "loss": 0.1726, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 1.0911786508524832, | |
| "grad_norm": 0.3809909224510193, | |
| "learning_rate": 1.6063594808740112e-05, | |
| "loss": 0.1741, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 1.0941438102297998, | |
| "grad_norm": 0.4052869975566864, | |
| "learning_rate": 1.6036076085226813e-05, | |
| "loss": 0.1728, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 1.0971089696071163, | |
| "grad_norm": 0.38783711194992065, | |
| "learning_rate": 1.6008485263209742e-05, | |
| "loss": 0.1701, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.100074128984433, | |
| "grad_norm": 0.4025594890117645, | |
| "learning_rate": 1.598082267225018e-05, | |
| "loss": 0.1743, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 1.1030392883617495, | |
| "grad_norm": 0.4071436822414398, | |
| "learning_rate": 1.595308864276666e-05, | |
| "loss": 0.1726, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 1.106004447739066, | |
| "grad_norm": 0.446532279253006, | |
| "learning_rate": 1.592528350603103e-05, | |
| "loss": 0.1708, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.1089696071163826, | |
| "grad_norm": 0.3993205726146698, | |
| "learning_rate": 1.5897407594164468e-05, | |
| "loss": 0.1805, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 1.111934766493699, | |
| "grad_norm": 0.42292505502700806, | |
| "learning_rate": 1.586946124013354e-05, | |
| "loss": 0.1823, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.1148999258710155, | |
| "grad_norm": 0.41676023602485657, | |
| "learning_rate": 1.5841444777746232e-05, | |
| "loss": 0.1756, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 1.117865085248332, | |
| "grad_norm": 0.3944017291069031, | |
| "learning_rate": 1.5813358541647915e-05, | |
| "loss": 0.1734, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 1.1208302446256486, | |
| "grad_norm": 0.38493022322654724, | |
| "learning_rate": 1.578520286731741e-05, | |
| "loss": 0.1772, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 1.1237954040029652, | |
| "grad_norm": 0.4245246350765228, | |
| "learning_rate": 1.575697809106292e-05, | |
| "loss": 0.1743, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 1.1267605633802817, | |
| "grad_norm": 0.3895925283432007, | |
| "learning_rate": 1.5728684550018066e-05, | |
| "loss": 0.1704, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.1297257227575983, | |
| "grad_norm": 0.3827330768108368, | |
| "learning_rate": 1.570032258213783e-05, | |
| "loss": 0.1746, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 1.1326908821349146, | |
| "grad_norm": 0.3874651789665222, | |
| "learning_rate": 1.5671892526194515e-05, | |
| "loss": 0.1751, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 1.1356560415122312, | |
| "grad_norm": 0.4029993712902069, | |
| "learning_rate": 1.564339472177373e-05, | |
| "loss": 0.1771, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 1.1386212008895478, | |
| "grad_norm": 0.3838706314563751, | |
| "learning_rate": 1.561482950927029e-05, | |
| "loss": 0.1732, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 1.1415863602668643, | |
| "grad_norm": 0.3896842896938324, | |
| "learning_rate": 1.5586197229884185e-05, | |
| "loss": 0.1737, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 1.144551519644181, | |
| "grad_norm": 0.4098159372806549, | |
| "learning_rate": 1.5557498225616488e-05, | |
| "loss": 0.1769, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 1.1475166790214975, | |
| "grad_norm": 0.4123744070529938, | |
| "learning_rate": 1.5528732839265272e-05, | |
| "loss": 0.177, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 1.150481838398814, | |
| "grad_norm": 0.3826339542865753, | |
| "learning_rate": 1.549990141442153e-05, | |
| "loss": 0.1708, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 1.1534469977761304, | |
| "grad_norm": 0.38323384523391724, | |
| "learning_rate": 1.5471004295465034e-05, | |
| "loss": 0.1759, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 1.156412157153447, | |
| "grad_norm": 0.3751480281352997, | |
| "learning_rate": 1.5442041827560274e-05, | |
| "loss": 0.1742, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.1593773165307635, | |
| "grad_norm": 0.42600059509277344, | |
| "learning_rate": 1.5413014356652287e-05, | |
| "loss": 0.1726, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 1.16234247590808, | |
| "grad_norm": 0.4077330529689789, | |
| "learning_rate": 1.538392222946255e-05, | |
| "loss": 0.1708, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 1.1653076352853966, | |
| "grad_norm": 0.39985400438308716, | |
| "learning_rate": 1.5354765793484834e-05, | |
| "loss": 0.1753, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 1.1682727946627132, | |
| "grad_norm": 0.4099324941635132, | |
| "learning_rate": 1.5325545396981053e-05, | |
| "loss": 0.172, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 1.1712379540400297, | |
| "grad_norm": 0.39008331298828125, | |
| "learning_rate": 1.5296261388977107e-05, | |
| "loss": 0.172, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 1.174203113417346, | |
| "grad_norm": 0.36513862013816833, | |
| "learning_rate": 1.52669141192587e-05, | |
| "loss": 0.1699, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 1.1771682727946626, | |
| "grad_norm": 0.43505406379699707, | |
| "learning_rate": 1.5237503938367186e-05, | |
| "loss": 0.1766, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 1.1801334321719792, | |
| "grad_norm": 0.4039159417152405, | |
| "learning_rate": 1.5208031197595357e-05, | |
| "loss": 0.1744, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 1.1830985915492958, | |
| "grad_norm": 0.3673771619796753, | |
| "learning_rate": 1.5178496248983254e-05, | |
| "loss": 0.1719, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 1.1860637509266123, | |
| "grad_norm": 0.3980352580547333, | |
| "learning_rate": 1.5148899445313983e-05, | |
| "loss": 0.1722, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.1890289103039289, | |
| "grad_norm": 0.39053529500961304, | |
| "learning_rate": 1.5119241140109466e-05, | |
| "loss": 0.1743, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 1.1919940696812454, | |
| "grad_norm": 0.3899192214012146, | |
| "learning_rate": 1.5089521687626243e-05, | |
| "loss": 0.1723, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 1.1949592290585618, | |
| "grad_norm": 0.4070497453212738, | |
| "learning_rate": 1.505974144285124e-05, | |
| "loss": 0.1692, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 1.1979243884358783, | |
| "grad_norm": 0.3976007103919983, | |
| "learning_rate": 1.5029900761497507e-05, | |
| "loss": 0.1781, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 1.200889547813195, | |
| "grad_norm": 0.41118377447128296, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 0.1746, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 1.2038547071905115, | |
| "grad_norm": 0.41726142168045044, | |
| "learning_rate": 1.4970039515511303e-05, | |
| "loss": 0.179, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 1.206819866567828, | |
| "grad_norm": 0.3854449391365051, | |
| "learning_rate": 1.4940019665897363e-05, | |
| "loss": 0.1737, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 1.2097850259451446, | |
| "grad_norm": 0.45727819204330444, | |
| "learning_rate": 1.4909940809733223e-05, | |
| "loss": 0.1723, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 1.2127501853224611, | |
| "grad_norm": 0.3889809250831604, | |
| "learning_rate": 1.4879803306298736e-05, | |
| "loss": 0.1714, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 1.2157153446997775, | |
| "grad_norm": 0.4237361550331116, | |
| "learning_rate": 1.4849607515574276e-05, | |
| "loss": 0.1724, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.218680504077094, | |
| "grad_norm": 0.4138452112674713, | |
| "learning_rate": 1.4819353798236427e-05, | |
| "loss": 0.1725, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 1.2216456634544106, | |
| "grad_norm": 0.4682404100894928, | |
| "learning_rate": 1.4789042515653687e-05, | |
| "loss": 0.1727, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 1.2246108228317272, | |
| "grad_norm": 0.38663214445114136, | |
| "learning_rate": 1.4758674029882152e-05, | |
| "loss": 0.176, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 1.2275759822090437, | |
| "grad_norm": 0.391353577375412, | |
| "learning_rate": 1.4728248703661183e-05, | |
| "loss": 0.1775, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 1.2305411415863603, | |
| "grad_norm": 0.4257277846336365, | |
| "learning_rate": 1.4697766900409076e-05, | |
| "loss": 0.1773, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 1.2335063009636769, | |
| "grad_norm": 0.38307616114616394, | |
| "learning_rate": 1.466722898421873e-05, | |
| "loss": 0.1739, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 1.2364714603409934, | |
| "grad_norm": 0.3973027467727661, | |
| "learning_rate": 1.4636635319853274e-05, | |
| "loss": 0.1738, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 1.2394366197183098, | |
| "grad_norm": 0.4155060052871704, | |
| "learning_rate": 1.4605986272741748e-05, | |
| "loss": 0.1737, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 1.2424017790956263, | |
| "grad_norm": 0.40221065282821655, | |
| "learning_rate": 1.4575282208974704e-05, | |
| "loss": 0.1718, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 1.2453669384729429, | |
| "grad_norm": 0.41945594549179077, | |
| "learning_rate": 1.4544523495299843e-05, | |
| "loss": 0.1772, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.2483320978502594, | |
| "grad_norm": 0.4217647612094879, | |
| "learning_rate": 1.4513710499117648e-05, | |
| "loss": 0.1816, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 1.251297257227576, | |
| "grad_norm": 0.4151117205619812, | |
| "learning_rate": 1.4482843588476976e-05, | |
| "loss": 0.1718, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 1.2542624166048926, | |
| "grad_norm": 0.38060155510902405, | |
| "learning_rate": 1.445192313207067e-05, | |
| "loss": 0.1725, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 1.257227575982209, | |
| "grad_norm": 0.4043025076389313, | |
| "learning_rate": 1.4420949499231172e-05, | |
| "loss": 0.1735, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 1.2601927353595257, | |
| "grad_norm": 0.40334248542785645, | |
| "learning_rate": 1.4389923059926064e-05, | |
| "loss": 0.1748, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.263157894736842, | |
| "grad_norm": 0.3861962854862213, | |
| "learning_rate": 1.4358844184753713e-05, | |
| "loss": 0.1751, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 1.2661230541141586, | |
| "grad_norm": 0.3862569034099579, | |
| "learning_rate": 1.432771324493879e-05, | |
| "loss": 0.1766, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 1.2690882134914752, | |
| "grad_norm": 0.3655155897140503, | |
| "learning_rate": 1.4296530612327864e-05, | |
| "loss": 0.1738, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 1.2720533728687917, | |
| "grad_norm": 0.45015332102775574, | |
| "learning_rate": 1.4265296659384956e-05, | |
| "loss": 0.1758, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 1.2750185322461083, | |
| "grad_norm": 0.40792006254196167, | |
| "learning_rate": 1.4234011759187084e-05, | |
| "loss": 0.1753, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.2779836916234246, | |
| "grad_norm": 0.3909926116466522, | |
| "learning_rate": 1.4202676285419811e-05, | |
| "loss": 0.1775, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 1.2809488510007414, | |
| "grad_norm": 0.38805529475212097, | |
| "learning_rate": 1.4171290612372781e-05, | |
| "loss": 0.1772, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 1.2839140103780577, | |
| "grad_norm": 0.3860710859298706, | |
| "learning_rate": 1.4139855114935253e-05, | |
| "loss": 0.17, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 1.2868791697553743, | |
| "grad_norm": 0.42617350816726685, | |
| "learning_rate": 1.410837016859161e-05, | |
| "loss": 0.1743, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 1.2898443291326909, | |
| "grad_norm": 0.3832889795303345, | |
| "learning_rate": 1.4076836149416889e-05, | |
| "loss": 0.1698, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 1.2928094885100074, | |
| "grad_norm": 0.4039870500564575, | |
| "learning_rate": 1.4045253434072278e-05, | |
| "loss": 0.1752, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 1.295774647887324, | |
| "grad_norm": 0.38493219017982483, | |
| "learning_rate": 1.4013622399800628e-05, | |
| "loss": 0.1737, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 1.2987398072646406, | |
| "grad_norm": 0.4500020146369934, | |
| "learning_rate": 1.3981943424421932e-05, | |
| "loss": 0.1704, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 1.3017049666419571, | |
| "grad_norm": 0.4027196764945984, | |
| "learning_rate": 1.3950216886328818e-05, | |
| "loss": 0.1699, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 1.3046701260192735, | |
| "grad_norm": 0.37555673718452454, | |
| "learning_rate": 1.3918443164482048e-05, | |
| "loss": 0.1733, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.30763528539659, | |
| "grad_norm": 0.3900480568408966, | |
| "learning_rate": 1.3886622638405953e-05, | |
| "loss": 0.168, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 1.3106004447739066, | |
| "grad_norm": 0.40044647455215454, | |
| "learning_rate": 1.3854755688183941e-05, | |
| "loss": 0.1681, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 1.3135656041512231, | |
| "grad_norm": 0.39409545063972473, | |
| "learning_rate": 1.3822842694453923e-05, | |
| "loss": 0.1731, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 1.3165307635285397, | |
| "grad_norm": 0.37648630142211914, | |
| "learning_rate": 1.3790884038403796e-05, | |
| "loss": 0.1711, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 1.3194959229058563, | |
| "grad_norm": 0.3983948826789856, | |
| "learning_rate": 1.375888010176686e-05, | |
| "loss": 0.1782, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 1.3224610822831728, | |
| "grad_norm": 0.42869091033935547, | |
| "learning_rate": 1.3726831266817278e-05, | |
| "loss": 0.1714, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 1.3254262416604892, | |
| "grad_norm": 0.43148529529571533, | |
| "learning_rate": 1.3694737916365517e-05, | |
| "loss": 0.1734, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 1.3283914010378057, | |
| "grad_norm": 0.37700700759887695, | |
| "learning_rate": 1.3662600433753746e-05, | |
| "loss": 0.1732, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 1.3313565604151223, | |
| "grad_norm": 0.3717349171638489, | |
| "learning_rate": 1.3630419202851287e-05, | |
| "loss": 0.1722, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.3343217197924389, | |
| "grad_norm": 0.401803195476532, | |
| "learning_rate": 1.3598194608050011e-05, | |
| "loss": 0.1727, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.3372868791697554, | |
| "grad_norm": 0.373855322599411, | |
| "learning_rate": 1.3565927034259757e-05, | |
| "loss": 0.1724, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.340252038547072, | |
| "grad_norm": 0.40752193331718445, | |
| "learning_rate": 1.3533616866903736e-05, | |
| "loss": 0.1741, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.3432171979243885, | |
| "grad_norm": 0.37844231724739075, | |
| "learning_rate": 1.3501264491913909e-05, | |
| "loss": 0.1759, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.3461823573017049, | |
| "grad_norm": 0.37028035521507263, | |
| "learning_rate": 1.3468870295726399e-05, | |
| "loss": 0.1743, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.3491475166790214, | |
| "grad_norm": 0.3744882047176361, | |
| "learning_rate": 1.3436434665276865e-05, | |
| "loss": 0.176, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.352112676056338, | |
| "grad_norm": 0.34571152925491333, | |
| "learning_rate": 1.3403957987995884e-05, | |
| "loss": 0.1725, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.3550778354336546, | |
| "grad_norm": 0.3648885488510132, | |
| "learning_rate": 1.3371440651804313e-05, | |
| "loss": 0.1752, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.3580429948109711, | |
| "grad_norm": 0.37405288219451904, | |
| "learning_rate": 1.3338883045108674e-05, | |
| "loss": 0.1716, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.3610081541882877, | |
| "grad_norm": 0.3600881099700928, | |
| "learning_rate": 1.3306285556796494e-05, | |
| "loss": 0.166, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.3639733135656043, | |
| "grad_norm": 0.38361856341362, | |
| "learning_rate": 1.327364857623168e-05, | |
| "loss": 0.1686, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.3669384729429206, | |
| "grad_norm": 0.4009436070919037, | |
| "learning_rate": 1.3240972493249846e-05, | |
| "loss": 0.1765, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.3699036323202372, | |
| "grad_norm": 0.3752938508987427, | |
| "learning_rate": 1.3208257698153677e-05, | |
| "loss": 0.1673, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.3728687916975537, | |
| "grad_norm": 0.3697980046272278, | |
| "learning_rate": 1.3175504581708261e-05, | |
| "loss": 0.1696, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.3758339510748703, | |
| "grad_norm": 0.4123381972312927, | |
| "learning_rate": 1.3142713535136413e-05, | |
| "loss": 0.1751, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.3787991104521868, | |
| "grad_norm": 0.3773389458656311, | |
| "learning_rate": 1.3109884950114007e-05, | |
| "loss": 0.175, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.3817642698295034, | |
| "grad_norm": 0.37522801756858826, | |
| "learning_rate": 1.3077019218765306e-05, | |
| "loss": 0.1721, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.38472942920682, | |
| "grad_norm": 0.3822220265865326, | |
| "learning_rate": 1.3044116733658261e-05, | |
| "loss": 0.1741, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.3876945885841363, | |
| "grad_norm": 0.33929958939552307, | |
| "learning_rate": 1.3011177887799846e-05, | |
| "loss": 0.1669, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.3906597479614529, | |
| "grad_norm": 0.3751008212566376, | |
| "learning_rate": 1.2978203074631335e-05, | |
| "loss": 0.173, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.3936249073387694, | |
| "grad_norm": 0.3586931526660919, | |
| "learning_rate": 1.2945192688023625e-05, | |
| "loss": 0.1707, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.396590066716086, | |
| "grad_norm": 0.3598410189151764, | |
| "learning_rate": 1.2912147122272523e-05, | |
| "loss": 0.1673, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.3995552260934025, | |
| "grad_norm": 0.37330952286720276, | |
| "learning_rate": 1.287906677209403e-05, | |
| "loss": 0.1705, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.402520385470719, | |
| "grad_norm": 0.3800138533115387, | |
| "learning_rate": 1.2845952032619651e-05, | |
| "loss": 0.1707, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.4054855448480357, | |
| "grad_norm": 0.34873542189598083, | |
| "learning_rate": 1.2812803299391629e-05, | |
| "loss": 0.1716, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.408450704225352, | |
| "grad_norm": 0.39961710572242737, | |
| "learning_rate": 1.2779620968358276e-05, | |
| "loss": 0.1713, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.4114158636026686, | |
| "grad_norm": 0.37982645630836487, | |
| "learning_rate": 1.2746405435869198e-05, | |
| "loss": 0.1713, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.4143810229799851, | |
| "grad_norm": 0.3620937764644623, | |
| "learning_rate": 1.271315709867059e-05, | |
| "loss": 0.1712, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.4173461823573017, | |
| "grad_norm": 0.36581623554229736, | |
| "learning_rate": 1.2679876353900482e-05, | |
| "loss": 0.1725, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.4203113417346183, | |
| "grad_norm": 0.36710691452026367, | |
| "learning_rate": 1.2646563599083997e-05, | |
| "loss": 0.1706, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.4232765011119348, | |
| "grad_norm": 0.3968733847141266, | |
| "learning_rate": 1.2613219232128608e-05, | |
| "loss": 0.1704, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.4262416604892514, | |
| "grad_norm": 0.38720619678497314, | |
| "learning_rate": 1.2579843651319382e-05, | |
| "loss": 0.1714, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.4292068198665677, | |
| "grad_norm": 0.36827707290649414, | |
| "learning_rate": 1.2546437255314223e-05, | |
| "loss": 0.1715, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.4321719792438843, | |
| "grad_norm": 0.37707608938217163, | |
| "learning_rate": 1.2513000443139112e-05, | |
| "loss": 0.1735, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.4351371386212008, | |
| "grad_norm": 0.40368345379829407, | |
| "learning_rate": 1.2479533614183334e-05, | |
| "loss": 0.1726, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.4381022979985174, | |
| "grad_norm": 0.3910945951938629, | |
| "learning_rate": 1.2446037168194716e-05, | |
| "loss": 0.1755, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.441067457375834, | |
| "grad_norm": 0.37151867151260376, | |
| "learning_rate": 1.2412511505274845e-05, | |
| "loss": 0.1771, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.4440326167531505, | |
| "grad_norm": 0.35527053475379944, | |
| "learning_rate": 1.23789570258743e-05, | |
| "loss": 0.1677, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.446997776130467, | |
| "grad_norm": 0.3575199544429779, | |
| "learning_rate": 1.2345374130787855e-05, | |
| "loss": 0.1715, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.4499629355077834, | |
| "grad_norm": 0.35391053557395935, | |
| "learning_rate": 1.23117632211497e-05, | |
| "loss": 0.1716, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.4529280948851, | |
| "grad_norm": 0.3692530691623688, | |
| "learning_rate": 1.2278124698428643e-05, | |
| "loss": 0.1689, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.4558932542624166, | |
| "grad_norm": 0.35716333985328674, | |
| "learning_rate": 1.2244458964423328e-05, | |
| "loss": 0.1682, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.4588584136397331, | |
| "grad_norm": 0.3617175221443176, | |
| "learning_rate": 1.221076642125742e-05, | |
| "loss": 0.1749, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.4618235730170497, | |
| "grad_norm": 0.3705756366252899, | |
| "learning_rate": 1.2177047471374808e-05, | |
| "loss": 0.1706, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.4647887323943662, | |
| "grad_norm": 0.35617804527282715, | |
| "learning_rate": 1.214330251753481e-05, | |
| "loss": 0.1704, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.4677538917716828, | |
| "grad_norm": 0.3682483732700348, | |
| "learning_rate": 1.2109531962807333e-05, | |
| "loss": 0.1688, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.4707190511489991, | |
| "grad_norm": 0.3654380738735199, | |
| "learning_rate": 1.207573621056809e-05, | |
| "loss": 0.1654, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.4736842105263157, | |
| "grad_norm": 0.39695996046066284, | |
| "learning_rate": 1.2041915664493763e-05, | |
| "loss": 0.1705, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.4766493699036323, | |
| "grad_norm": 0.3834567964076996, | |
| "learning_rate": 1.2008070728557186e-05, | |
| "loss": 0.1737, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.4796145292809488, | |
| "grad_norm": 0.3756810426712036, | |
| "learning_rate": 1.1974201807022525e-05, | |
| "loss": 0.1744, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.4825796886582654, | |
| "grad_norm": 0.43872207403182983, | |
| "learning_rate": 1.1940309304440434e-05, | |
| "loss": 0.1725, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.485544848035582, | |
| "grad_norm": 0.4155595004558563, | |
| "learning_rate": 1.1906393625643244e-05, | |
| "loss": 0.167, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.4885100074128985, | |
| "grad_norm": 0.37012434005737305, | |
| "learning_rate": 1.1872455175740111e-05, | |
| "loss": 0.1714, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.4914751667902149, | |
| "grad_norm": 0.4194466173648834, | |
| "learning_rate": 1.1838494360112185e-05, | |
| "loss": 0.1731, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 1.4944403261675316, | |
| "grad_norm": 0.38535988330841064, | |
| "learning_rate": 1.1804511584407763e-05, | |
| "loss": 0.1719, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.497405485544848, | |
| "grad_norm": 0.3790641725063324, | |
| "learning_rate": 1.1770507254537454e-05, | |
| "loss": 0.1715, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.5003706449221645, | |
| "grad_norm": 0.40725064277648926, | |
| "learning_rate": 1.1736481776669307e-05, | |
| "loss": 0.1672, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.503335804299481, | |
| "grad_norm": 0.3657318949699402, | |
| "learning_rate": 1.1702435557223988e-05, | |
| "loss": 0.1701, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 1.5063009636767977, | |
| "grad_norm": 0.41225719451904297, | |
| "learning_rate": 1.1668369002869912e-05, | |
| "loss": 0.1703, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.5092661230541142, | |
| "grad_norm": 0.38106808066368103, | |
| "learning_rate": 1.1634282520518382e-05, | |
| "loss": 0.1705, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 1.5122312824314306, | |
| "grad_norm": 0.43504565954208374, | |
| "learning_rate": 1.1600176517318742e-05, | |
| "loss": 0.1712, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.5151964418087474, | |
| "grad_norm": 0.37367385625839233, | |
| "learning_rate": 1.1566051400653486e-05, | |
| "loss": 0.1708, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 1.5181616011860637, | |
| "grad_norm": 0.3934025168418884, | |
| "learning_rate": 1.153190757813343e-05, | |
| "loss": 0.1723, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.5211267605633803, | |
| "grad_norm": 0.35954198241233826, | |
| "learning_rate": 1.1497745457592817e-05, | |
| "loss": 0.1686, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 1.5240919199406968, | |
| "grad_norm": 0.3657681345939636, | |
| "learning_rate": 1.1463565447084446e-05, | |
| "loss": 0.1715, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.5270570793180134, | |
| "grad_norm": 0.3832554817199707, | |
| "learning_rate": 1.142936795487482e-05, | |
| "loss": 0.1725, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.53002223869533, | |
| "grad_norm": 0.36780476570129395, | |
| "learning_rate": 1.1395153389439232e-05, | |
| "loss": 0.1686, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.5329873980726463, | |
| "grad_norm": 0.37948641180992126, | |
| "learning_rate": 1.1360922159456929e-05, | |
| "loss": 0.169, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 1.535952557449963, | |
| "grad_norm": 0.37667399644851685, | |
| "learning_rate": 1.1326674673806195e-05, | |
| "loss": 0.1694, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.5389177168272794, | |
| "grad_norm": 0.3817925751209259, | |
| "learning_rate": 1.129241134155949e-05, | |
| "loss": 0.1684, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 1.541882876204596, | |
| "grad_norm": 0.3880022168159485, | |
| "learning_rate": 1.1258132571978555e-05, | |
| "loss": 0.1681, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.5448480355819125, | |
| "grad_norm": 0.39235079288482666, | |
| "learning_rate": 1.1223838774509515e-05, | |
| "loss": 0.1724, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 1.547813194959229, | |
| "grad_norm": 0.3959818184375763, | |
| "learning_rate": 1.1189530358778005e-05, | |
| "loss": 0.1653, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.5507783543365457, | |
| "grad_norm": 0.3723091185092926, | |
| "learning_rate": 1.1155207734584264e-05, | |
| "loss": 0.1715, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 1.553743513713862, | |
| "grad_norm": 0.3744927644729614, | |
| "learning_rate": 1.1120871311898254e-05, | |
| "loss": 0.1709, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.5567086730911788, | |
| "grad_norm": 0.37305641174316406, | |
| "learning_rate": 1.1086521500854746e-05, | |
| "loss": 0.1705, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.5596738324684951, | |
| "grad_norm": 0.3628908693790436, | |
| "learning_rate": 1.1052158711748435e-05, | |
| "loss": 0.1703, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.5626389918458117, | |
| "grad_norm": 0.3602434992790222, | |
| "learning_rate": 1.1017783355029027e-05, | |
| "loss": 0.1733, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.5656041512231282, | |
| "grad_norm": 0.3662010133266449, | |
| "learning_rate": 1.0983395841296349e-05, | |
| "loss": 0.1722, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.5685693106004448, | |
| "grad_norm": 0.38595232367515564, | |
| "learning_rate": 1.0948996581295437e-05, | |
| "loss": 0.1722, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.5715344699777614, | |
| "grad_norm": 0.3809836804866791, | |
| "learning_rate": 1.0914585985911632e-05, | |
| "loss": 0.1704, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.5744996293550777, | |
| "grad_norm": 0.3592289686203003, | |
| "learning_rate": 1.0880164466165675e-05, | |
| "loss": 0.1732, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.5774647887323945, | |
| "grad_norm": 0.3625737428665161, | |
| "learning_rate": 1.084573243320878e-05, | |
| "loss": 0.1743, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.5804299481097108, | |
| "grad_norm": 0.3582081198692322, | |
| "learning_rate": 1.0811290298317755e-05, | |
| "loss": 0.171, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.5833951074870274, | |
| "grad_norm": 0.3777657449245453, | |
| "learning_rate": 1.0776838472890065e-05, | |
| "loss": 0.1711, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.586360266864344, | |
| "grad_norm": 0.34954240918159485, | |
| "learning_rate": 1.0742377368438915e-05, | |
| "loss": 0.1685, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.5893254262416605, | |
| "grad_norm": 0.3632443845272064, | |
| "learning_rate": 1.0707907396588362e-05, | |
| "loss": 0.1689, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.592290585618977, | |
| "grad_norm": 0.35810449719429016, | |
| "learning_rate": 1.0673428969068365e-05, | |
| "loss": 0.1714, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.5952557449962934, | |
| "grad_norm": 0.36739829182624817, | |
| "learning_rate": 1.063894249770989e-05, | |
| "loss": 0.17, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.5982209043736102, | |
| "grad_norm": 0.35011234879493713, | |
| "learning_rate": 1.0604448394439983e-05, | |
| "loss": 0.1661, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.6011860637509265, | |
| "grad_norm": 0.37619051337242126, | |
| "learning_rate": 1.0569947071276847e-05, | |
| "loss": 0.1708, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.604151223128243, | |
| "grad_norm": 0.36766669154167175, | |
| "learning_rate": 1.053543894032493e-05, | |
| "loss": 0.1699, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.6071163825055597, | |
| "grad_norm": 0.3799968361854553, | |
| "learning_rate": 1.0500924413769988e-05, | |
| "loss": 0.175, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.6100815418828762, | |
| "grad_norm": 0.35972005128860474, | |
| "learning_rate": 1.0466403903874176e-05, | |
| "loss": 0.1709, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.6130467012601928, | |
| "grad_norm": 0.38818514347076416, | |
| "learning_rate": 1.0431877822971118e-05, | |
| "loss": 0.1729, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.6160118606375091, | |
| "grad_norm": 0.35318616032600403, | |
| "learning_rate": 1.0397346583460972e-05, | |
| "loss": 0.1708, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.618977020014826, | |
| "grad_norm": 0.34682103991508484, | |
| "learning_rate": 1.0362810597805526e-05, | |
| "loss": 0.172, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.6219421793921422, | |
| "grad_norm": 0.37605708837509155, | |
| "learning_rate": 1.0328270278523256e-05, | |
| "loss": 0.1733, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.6249073387694588, | |
| "grad_norm": 0.3474465608596802, | |
| "learning_rate": 1.0293726038184393e-05, | |
| "loss": 0.1659, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.6278724981467754, | |
| "grad_norm": 0.3567797839641571, | |
| "learning_rate": 1.0259178289406011e-05, | |
| "loss": 0.1692, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.630837657524092, | |
| "grad_norm": 0.35859590768814087, | |
| "learning_rate": 1.022462744484709e-05, | |
| "loss": 0.1725, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.6338028169014085, | |
| "grad_norm": 0.4004250168800354, | |
| "learning_rate": 1.019007391720359e-05, | |
| "loss": 0.171, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.6367679762787248, | |
| "grad_norm": 0.3502226769924164, | |
| "learning_rate": 1.0155518119203511e-05, | |
| "loss": 0.1669, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.6397331356560416, | |
| "grad_norm": 0.35019659996032715, | |
| "learning_rate": 1.0120960463601977e-05, | |
| "loss": 0.162, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.642698295033358, | |
| "grad_norm": 0.3413262963294983, | |
| "learning_rate": 1.0086401363176306e-05, | |
| "loss": 0.1671, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.6456634544106745, | |
| "grad_norm": 0.3686580955982208, | |
| "learning_rate": 1.0051841230721065e-05, | |
| "loss": 0.1723, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.648628613787991, | |
| "grad_norm": 0.4102790355682373, | |
| "learning_rate": 1.0017280479043148e-05, | |
| "loss": 0.1737, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.6515937731653076, | |
| "grad_norm": 0.3648839592933655, | |
| "learning_rate": 9.982719520956856e-06, | |
| "loss": 0.1701, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.6545589325426242, | |
| "grad_norm": 0.35376548767089844, | |
| "learning_rate": 9.948158769278939e-06, | |
| "loss": 0.1665, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.6575240919199405, | |
| "grad_norm": 0.34262967109680176, | |
| "learning_rate": 9.913598636823694e-06, | |
| "loss": 0.1637, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.6604892512972573, | |
| "grad_norm": 0.3623892068862915, | |
| "learning_rate": 9.879039536398023e-06, | |
| "loss": 0.1688, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.6634544106745737, | |
| "grad_norm": 0.36795225739479065, | |
| "learning_rate": 9.844481880796492e-06, | |
| "loss": 0.1716, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.6664195700518905, | |
| "grad_norm": 0.3584054112434387, | |
| "learning_rate": 9.809926082796415e-06, | |
| "loss": 0.1717, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.6693847294292068, | |
| "grad_norm": 0.3560091555118561, | |
| "learning_rate": 9.775372555152912e-06, | |
| "loss": 0.1685, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.6723498888065234, | |
| "grad_norm": 0.36741241812705994, | |
| "learning_rate": 9.740821710593989e-06, | |
| "loss": 0.1685, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.67531504818384, | |
| "grad_norm": 0.3397235870361328, | |
| "learning_rate": 9.70627396181561e-06, | |
| "loss": 0.1613, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.6782802075611563, | |
| "grad_norm": 0.3634246289730072, | |
| "learning_rate": 9.671729721476747e-06, | |
| "loss": 0.1681, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.681245366938473, | |
| "grad_norm": 0.3582555949687958, | |
| "learning_rate": 9.637189402194477e-06, | |
| "loss": 0.1687, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.6842105263157894, | |
| "grad_norm": 0.34005481004714966, | |
| "learning_rate": 9.602653416539031e-06, | |
| "loss": 0.1689, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.6871756856931062, | |
| "grad_norm": 0.3448920249938965, | |
| "learning_rate": 9.568122177028884e-06, | |
| "loss": 0.1688, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.6901408450704225, | |
| "grad_norm": 0.3394884169101715, | |
| "learning_rate": 9.533596096125826e-06, | |
| "loss": 0.163, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.693106004447739, | |
| "grad_norm": 0.35604503750801086, | |
| "learning_rate": 9.499075586230014e-06, | |
| "loss": 0.1709, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.6960711638250556, | |
| "grad_norm": 0.34704917669296265, | |
| "learning_rate": 9.464561059675073e-06, | |
| "loss": 0.1686, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.699036323202372, | |
| "grad_norm": 0.3488229811191559, | |
| "learning_rate": 9.430052928723153e-06, | |
| "loss": 0.1705, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.7020014825796888, | |
| "grad_norm": 0.349729984998703, | |
| "learning_rate": 9.395551605560018e-06, | |
| "loss": 0.1656, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.704966641957005, | |
| "grad_norm": 0.3426892161369324, | |
| "learning_rate": 9.361057502290112e-06, | |
| "loss": 0.1652, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.7079318013343219, | |
| "grad_norm": 0.3359294533729553, | |
| "learning_rate": 9.326571030931636e-06, | |
| "loss": 0.1668, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.7108969607116382, | |
| "grad_norm": 0.32818013429641724, | |
| "learning_rate": 9.292092603411642e-06, | |
| "loss": 0.1641, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.7138621200889548, | |
| "grad_norm": 0.3587988317012787, | |
| "learning_rate": 9.257622631561085e-06, | |
| "loss": 0.1692, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.7168272794662713, | |
| "grad_norm": 0.3606449365615845, | |
| "learning_rate": 9.223161527109938e-06, | |
| "loss": 0.1732, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.7197924388435877, | |
| "grad_norm": 0.33454060554504395, | |
| "learning_rate": 9.188709701682246e-06, | |
| "loss": 0.1707, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.7227575982209045, | |
| "grad_norm": 0.3533168435096741, | |
| "learning_rate": 9.154267566791224e-06, | |
| "loss": 0.1647, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.7257227575982208, | |
| "grad_norm": 0.3588050901889801, | |
| "learning_rate": 9.119835533834332e-06, | |
| "loss": 0.1709, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.7286879169755376, | |
| "grad_norm": 0.35869184136390686, | |
| "learning_rate": 9.085414014088368e-06, | |
| "loss": 0.1721, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.731653076352854, | |
| "grad_norm": 0.33058810234069824, | |
| "learning_rate": 9.051003418704566e-06, | |
| "loss": 0.1687, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.7346182357301705, | |
| "grad_norm": 0.35373157262802124, | |
| "learning_rate": 9.016604158703654e-06, | |
| "loss": 0.1685, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.737583395107487, | |
| "grad_norm": 0.3870552182197571, | |
| "learning_rate": 8.982216644970978e-06, | |
| "loss": 0.1698, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.7405485544848036, | |
| "grad_norm": 0.35172680020332336, | |
| "learning_rate": 8.947841288251568e-06, | |
| "loss": 0.167, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.7435137138621202, | |
| "grad_norm": 0.3640024960041046, | |
| "learning_rate": 8.913478499145255e-06, | |
| "loss": 0.1659, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.7464788732394365, | |
| "grad_norm": 0.36789610981941223, | |
| "learning_rate": 8.879128688101749e-06, | |
| "loss": 0.1708, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.7494440326167533, | |
| "grad_norm": 0.3513200283050537, | |
| "learning_rate": 8.844792265415738e-06, | |
| "loss": 0.1652, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.7524091919940696, | |
| "grad_norm": 0.3880747854709625, | |
| "learning_rate": 8.810469641222001e-06, | |
| "loss": 0.1699, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.7553743513713862, | |
| "grad_norm": 0.33801934123039246, | |
| "learning_rate": 8.776161225490488e-06, | |
| "loss": 0.1675, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.7583395107487028, | |
| "grad_norm": 0.3653337359428406, | |
| "learning_rate": 8.741867428021447e-06, | |
| "loss": 0.1648, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.7613046701260193, | |
| "grad_norm": 0.36136823892593384, | |
| "learning_rate": 8.707588658440511e-06, | |
| "loss": 0.1696, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.7642698295033359, | |
| "grad_norm": 0.33816996216773987, | |
| "learning_rate": 8.673325326193806e-06, | |
| "loss": 0.1665, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.7672349888806522, | |
| "grad_norm": 0.33847707509994507, | |
| "learning_rate": 8.639077840543078e-06, | |
| "loss": 0.1678, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.770200148257969, | |
| "grad_norm": 0.3402957022190094, | |
| "learning_rate": 8.604846610560771e-06, | |
| "loss": 0.1643, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.7731653076352853, | |
| "grad_norm": 0.37062397599220276, | |
| "learning_rate": 8.570632045125185e-06, | |
| "loss": 0.1679, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.776130467012602, | |
| "grad_norm": 0.34380587935447693, | |
| "learning_rate": 8.536434552915555e-06, | |
| "loss": 0.1652, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.7790956263899185, | |
| "grad_norm": 0.33917438983917236, | |
| "learning_rate": 8.502254542407186e-06, | |
| "loss": 0.1652, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.782060785767235, | |
| "grad_norm": 0.3372032940387726, | |
| "learning_rate": 8.468092421866575e-06, | |
| "loss": 0.1629, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.7850259451445516, | |
| "grad_norm": 0.34099259972572327, | |
| "learning_rate": 8.433948599346516e-06, | |
| "loss": 0.1678, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.787991104521868, | |
| "grad_norm": 0.370136559009552, | |
| "learning_rate": 8.399823482681263e-06, | |
| "loss": 0.1671, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.7909562638991847, | |
| "grad_norm": 0.3444167375564575, | |
| "learning_rate": 8.36571747948162e-06, | |
| "loss": 0.1652, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.793921423276501, | |
| "grad_norm": 0.3237707018852234, | |
| "learning_rate": 8.331630997130091e-06, | |
| "loss": 0.1647, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.7968865826538176, | |
| "grad_norm": 0.3346817195415497, | |
| "learning_rate": 8.297564442776014e-06, | |
| "loss": 0.168, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.7998517420311342, | |
| "grad_norm": 0.3474122881889343, | |
| "learning_rate": 8.263518223330698e-06, | |
| "loss": 0.1665, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.8028169014084507, | |
| "grad_norm": 0.37336310744285583, | |
| "learning_rate": 8.229492745462551e-06, | |
| "loss": 0.1628, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.8057820607857673, | |
| "grad_norm": 0.3516935706138611, | |
| "learning_rate": 8.195488415592238e-06, | |
| "loss": 0.1669, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.8087472201630836, | |
| "grad_norm": 0.3758098781108856, | |
| "learning_rate": 8.161505639887818e-06, | |
| "loss": 0.1709, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.8117123795404004, | |
| "grad_norm": 0.34178832173347473, | |
| "learning_rate": 8.12754482425989e-06, | |
| "loss": 0.1659, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.8146775389177168, | |
| "grad_norm": 0.3256490230560303, | |
| "learning_rate": 8.09360637435676e-06, | |
| "loss": 0.1649, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.8176426982950333, | |
| "grad_norm": 0.3661201596260071, | |
| "learning_rate": 8.05969069555957e-06, | |
| "loss": 0.167, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.82060785767235, | |
| "grad_norm": 0.34554868936538696, | |
| "learning_rate": 8.025798192977482e-06, | |
| "loss": 0.1685, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.8235730170496665, | |
| "grad_norm": 0.3409639298915863, | |
| "learning_rate": 7.991929271442817e-06, | |
| "loss": 0.1674, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.826538176426983, | |
| "grad_norm": 0.36160513758659363, | |
| "learning_rate": 7.958084335506239e-06, | |
| "loss": 0.167, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.8295033358042994, | |
| "grad_norm": 0.3661399483680725, | |
| "learning_rate": 7.924263789431913e-06, | |
| "loss": 0.1658, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.8324684951816161, | |
| "grad_norm": 0.3356715738773346, | |
| "learning_rate": 7.89046803719267e-06, | |
| "loss": 0.1709, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.8354336545589325, | |
| "grad_norm": 0.3546086549758911, | |
| "learning_rate": 7.856697482465195e-06, | |
| "loss": 0.1626, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.838398813936249, | |
| "grad_norm": 0.3332943320274353, | |
| "learning_rate": 7.822952528625192e-06, | |
| "loss": 0.1678, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.8413639733135656, | |
| "grad_norm": 0.34793728590011597, | |
| "learning_rate": 7.789233578742583e-06, | |
| "loss": 0.1659, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.8443291326908822, | |
| "grad_norm": 0.33829519152641296, | |
| "learning_rate": 7.755541035576677e-06, | |
| "loss": 0.1647, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.8472942920681987, | |
| "grad_norm": 0.33728060126304626, | |
| "learning_rate": 7.721875301571359e-06, | |
| "loss": 0.169, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.850259451445515, | |
| "grad_norm": 0.34719350934028625, | |
| "learning_rate": 7.688236778850307e-06, | |
| "loss": 0.1694, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.8532246108228319, | |
| "grad_norm": 0.328850656747818, | |
| "learning_rate": 7.654625869212147e-06, | |
| "loss": 0.1627, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.8561897702001482, | |
| "grad_norm": 0.34174007177352905, | |
| "learning_rate": 7.621042974125701e-06, | |
| "loss": 0.1693, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.8591549295774648, | |
| "grad_norm": 0.3343605101108551, | |
| "learning_rate": 7.587488494725157e-06, | |
| "loss": 0.1639, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.8621200889547813, | |
| "grad_norm": 0.3540162742137909, | |
| "learning_rate": 7.553962831805291e-06, | |
| "loss": 0.1667, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.8650852483320979, | |
| "grad_norm": 0.32035985589027405, | |
| "learning_rate": 7.520466385816672e-06, | |
| "loss": 0.1632, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.8680504077094144, | |
| "grad_norm": 0.3598351776599884, | |
| "learning_rate": 7.48699955686089e-06, | |
| "loss": 0.1674, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.8710155670867308, | |
| "grad_norm": 0.35652288794517517, | |
| "learning_rate": 7.453562744685779e-06, | |
| "loss": 0.1661, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.8739807264640476, | |
| "grad_norm": 0.32800406217575073, | |
| "learning_rate": 7.420156348680621e-06, | |
| "loss": 0.1617, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.876945885841364, | |
| "grad_norm": 0.3622135818004608, | |
| "learning_rate": 7.3867807678713965e-06, | |
| "loss": 0.1656, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.8799110452186805, | |
| "grad_norm": 0.34809359908103943, | |
| "learning_rate": 7.353436400916006e-06, | |
| "loss": 0.1677, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.882876204595997, | |
| "grad_norm": 0.31972047686576843, | |
| "learning_rate": 7.32012364609952e-06, | |
| "loss": 0.1614, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.8858413639733136, | |
| "grad_norm": 0.3630245327949524, | |
| "learning_rate": 7.286842901329413e-06, | |
| "loss": 0.1674, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.8888065233506302, | |
| "grad_norm": 0.337985098361969, | |
| "learning_rate": 7.253594564130804e-06, | |
| "loss": 0.1661, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.8917716827279465, | |
| "grad_norm": 0.35212552547454834, | |
| "learning_rate": 7.22037903164173e-06, | |
| "loss": 0.1667, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.8947368421052633, | |
| "grad_norm": 0.3282301127910614, | |
| "learning_rate": 7.187196700608373e-06, | |
| "loss": 0.1642, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.8977020014825796, | |
| "grad_norm": 0.3764040470123291, | |
| "learning_rate": 7.154047967380353e-06, | |
| "loss": 0.1664, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.9006671608598962, | |
| "grad_norm": 0.35521620512008667, | |
| "learning_rate": 7.120933227905971e-06, | |
| "loss": 0.1674, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.9036323202372127, | |
| "grad_norm": 0.34446388483047485, | |
| "learning_rate": 7.0878528777274814e-06, | |
| "loss": 0.167, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.9065974796145293, | |
| "grad_norm": 0.3320685029029846, | |
| "learning_rate": 7.05480731197638e-06, | |
| "loss": 0.1617, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.9095626389918459, | |
| "grad_norm": 0.3581525385379791, | |
| "learning_rate": 7.021796925368667e-06, | |
| "loss": 0.1672, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.9125277983691622, | |
| "grad_norm": 0.3463064730167389, | |
| "learning_rate": 6.988822112200157e-06, | |
| "loss": 0.1616, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.915492957746479, | |
| "grad_norm": 0.3198210597038269, | |
| "learning_rate": 6.955883266341741e-06, | |
| "loss": 0.1644, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.9184581171237953, | |
| "grad_norm": 0.3308843672275543, | |
| "learning_rate": 6.9229807812346985e-06, | |
| "loss": 0.166, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.9214232765011119, | |
| "grad_norm": 0.357516884803772, | |
| "learning_rate": 6.890115049885995e-06, | |
| "loss": 0.1653, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.9243884358784284, | |
| "grad_norm": 0.3382551074028015, | |
| "learning_rate": 6.85728646486359e-06, | |
| "loss": 0.1643, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.927353595255745, | |
| "grad_norm": 0.34474846720695496, | |
| "learning_rate": 6.824495418291741e-06, | |
| "loss": 0.1664, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.9303187546330616, | |
| "grad_norm": 0.3264622390270233, | |
| "learning_rate": 6.791742301846325e-06, | |
| "loss": 0.1651, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.933283914010378, | |
| "grad_norm": 0.34936872124671936, | |
| "learning_rate": 6.759027506750159e-06, | |
| "loss": 0.1664, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.9362490733876947, | |
| "grad_norm": 0.34215009212493896, | |
| "learning_rate": 6.726351423768323e-06, | |
| "loss": 0.1671, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.939214232765011, | |
| "grad_norm": 0.33212119340896606, | |
| "learning_rate": 6.693714443203507e-06, | |
| "loss": 0.1642, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.9421793921423276, | |
| "grad_norm": 0.3262803256511688, | |
| "learning_rate": 6.661116954891329e-06, | |
| "loss": 0.1618, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.9451445515196442, | |
| "grad_norm": 0.3281456530094147, | |
| "learning_rate": 6.62855934819569e-06, | |
| "loss": 0.1661, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.9481097108969607, | |
| "grad_norm": 0.3286809027194977, | |
| "learning_rate": 6.59604201200412e-06, | |
| "loss": 0.1639, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.9510748702742773, | |
| "grad_norm": 0.33260786533355713, | |
| "learning_rate": 6.563565334723134e-06, | |
| "loss": 0.1623, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.9540400296515936, | |
| "grad_norm": 0.34506213665008545, | |
| "learning_rate": 6.5311297042736046e-06, | |
| "loss": 0.168, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.9570051890289104, | |
| "grad_norm": 0.3355524241924286, | |
| "learning_rate": 6.498735508086094e-06, | |
| "loss": 0.1659, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.9599703484062267, | |
| "grad_norm": 0.3466867506504059, | |
| "learning_rate": 6.466383133096268e-06, | |
| "loss": 0.1658, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 1.9629355077835435, | |
| "grad_norm": 0.32791653275489807, | |
| "learning_rate": 6.4340729657402424e-06, | |
| "loss": 0.1656, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.9659006671608599, | |
| "grad_norm": 0.3234347403049469, | |
| "learning_rate": 6.40180539194999e-06, | |
| "loss": 0.1626, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 1.9688658265381764, | |
| "grad_norm": 0.33013296127319336, | |
| "learning_rate": 6.3695807971487175e-06, | |
| "loss": 0.1666, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.971830985915493, | |
| "grad_norm": 0.3211277425289154, | |
| "learning_rate": 6.337399566246257e-06, | |
| "loss": 0.1631, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.9747961452928093, | |
| "grad_norm": 0.32406309247016907, | |
| "learning_rate": 6.305262083634488e-06, | |
| "loss": 0.1648, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.9777613046701261, | |
| "grad_norm": 0.31896913051605225, | |
| "learning_rate": 6.2731687331827214e-06, | |
| "loss": 0.1639, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 1.9807264640474425, | |
| "grad_norm": 0.32637953758239746, | |
| "learning_rate": 6.2411198982331435e-06, | |
| "loss": 0.1616, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.9836916234247592, | |
| "grad_norm": 0.3404463231563568, | |
| "learning_rate": 6.209115961596208e-06, | |
| "loss": 0.1639, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 1.9866567828020756, | |
| "grad_norm": 0.33349186182022095, | |
| "learning_rate": 6.177157305546077e-06, | |
| "loss": 0.1656, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.9896219421793921, | |
| "grad_norm": 0.33086422085762024, | |
| "learning_rate": 6.145244311816063e-06, | |
| "loss": 0.1659, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 1.9925871015567087, | |
| "grad_norm": 0.325844407081604, | |
| "learning_rate": 6.113377361594048e-06, | |
| "loss": 0.1627, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.995552260934025, | |
| "grad_norm": 0.3200128376483917, | |
| "learning_rate": 6.081556835517955e-06, | |
| "loss": 0.1648, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 1.9985174203113418, | |
| "grad_norm": 0.3372708261013031, | |
| "learning_rate": 6.049783113671184e-06, | |
| "loss": 0.1647, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 2.001482579688658, | |
| "grad_norm": 0.3364209830760956, | |
| "learning_rate": 6.018056575578075e-06, | |
| "loss": 0.1502, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.004447739065975, | |
| "grad_norm": 0.48207002878189087, | |
| "learning_rate": 5.986377600199371e-06, | |
| "loss": 0.135, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 2.0074128984432913, | |
| "grad_norm": 0.4162459671497345, | |
| "learning_rate": 5.9547465659277215e-06, | |
| "loss": 0.1348, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 2.010378057820608, | |
| "grad_norm": 0.3513208329677582, | |
| "learning_rate": 5.923163850583114e-06, | |
| "loss": 0.1341, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 2.0133432171979244, | |
| "grad_norm": 0.3508232831954956, | |
| "learning_rate": 5.891629831408392e-06, | |
| "loss": 0.1293, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 2.0163083765752408, | |
| "grad_norm": 0.3477492034435272, | |
| "learning_rate": 5.8601448850647515e-06, | |
| "loss": 0.1301, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.0192735359525575, | |
| "grad_norm": 0.4429793357849121, | |
| "learning_rate": 5.828709387627219e-06, | |
| "loss": 0.1342, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 2.022238695329874, | |
| "grad_norm": 0.45886561274528503, | |
| "learning_rate": 5.797323714580192e-06, | |
| "loss": 0.1315, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 2.0252038547071907, | |
| "grad_norm": 0.4540359377861023, | |
| "learning_rate": 5.7659882408129204e-06, | |
| "loss": 0.1331, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 2.028169014084507, | |
| "grad_norm": 0.3975986838340759, | |
| "learning_rate": 5.7347033406150494e-06, | |
| "loss": 0.1336, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 2.031134173461824, | |
| "grad_norm": 0.40435880422592163, | |
| "learning_rate": 5.703469387672138e-06, | |
| "loss": 0.1314, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 2.03409933283914, | |
| "grad_norm": 0.36956506967544556, | |
| "learning_rate": 5.672286755061212e-06, | |
| "loss": 0.1292, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 2.0370644922164565, | |
| "grad_norm": 0.3341839909553528, | |
| "learning_rate": 5.64115581524629e-06, | |
| "loss": 0.1284, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 2.0400296515937733, | |
| "grad_norm": 0.3514721989631653, | |
| "learning_rate": 5.610076940073939e-06, | |
| "loss": 0.1293, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 2.0429948109710896, | |
| "grad_norm": 0.3784818947315216, | |
| "learning_rate": 5.579050500768837e-06, | |
| "loss": 0.1317, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 2.0459599703484064, | |
| "grad_norm": 0.37682050466537476, | |
| "learning_rate": 5.548076867929331e-06, | |
| "loss": 0.1299, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.0489251297257227, | |
| "grad_norm": 0.3408771753311157, | |
| "learning_rate": 5.517156411523026e-06, | |
| "loss": 0.1294, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 2.0518902891030395, | |
| "grad_norm": 0.3548658788204193, | |
| "learning_rate": 5.486289500882355e-06, | |
| "loss": 0.1319, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 2.054855448480356, | |
| "grad_norm": 0.3722321391105652, | |
| "learning_rate": 5.455476504700161e-06, | |
| "loss": 0.1317, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 2.057820607857672, | |
| "grad_norm": 0.3734947144985199, | |
| "learning_rate": 5.424717791025302e-06, | |
| "loss": 0.1333, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 2.060785767234989, | |
| "grad_norm": 0.3377910554409027, | |
| "learning_rate": 5.3940137272582534e-06, | |
| "loss": 0.1298, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 2.0637509266123053, | |
| "grad_norm": 0.371706485748291, | |
| "learning_rate": 5.3633646801467255e-06, | |
| "loss": 0.1315, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 2.066716085989622, | |
| "grad_norm": 0.37259435653686523, | |
| "learning_rate": 5.332771015781275e-06, | |
| "loss": 0.1312, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 2.0696812453669384, | |
| "grad_norm": 0.357020765542984, | |
| "learning_rate": 5.302233099590928e-06, | |
| "loss": 0.1298, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 2.072646404744255, | |
| "grad_norm": 0.3563533425331116, | |
| "learning_rate": 5.271751296338823e-06, | |
| "loss": 0.13, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 2.0756115641215716, | |
| "grad_norm": 0.35207492113113403, | |
| "learning_rate": 5.241325970117851e-06, | |
| "loss": 0.1331, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.078576723498888, | |
| "grad_norm": 0.3274974524974823, | |
| "learning_rate": 5.210957484346314e-06, | |
| "loss": 0.1277, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 2.0815418828762047, | |
| "grad_norm": 0.36036819219589233, | |
| "learning_rate": 5.1806462017635775e-06, | |
| "loss": 0.13, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 2.084507042253521, | |
| "grad_norm": 0.36002618074417114, | |
| "learning_rate": 5.150392484425728e-06, | |
| "loss": 0.1319, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 2.087472201630838, | |
| "grad_norm": 0.3353933095932007, | |
| "learning_rate": 5.120196693701267e-06, | |
| "loss": 0.1312, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 2.090437361008154, | |
| "grad_norm": 0.36249250173568726, | |
| "learning_rate": 5.090059190266779e-06, | |
| "loss": 0.1331, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 2.093402520385471, | |
| "grad_norm": 0.3565695583820343, | |
| "learning_rate": 5.059980334102637e-06, | |
| "loss": 0.1306, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 2.0963676797627873, | |
| "grad_norm": 0.35264307260513306, | |
| "learning_rate": 5.0299604844886985e-06, | |
| "loss": 0.1288, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 2.0993328391401036, | |
| "grad_norm": 0.34676074981689453, | |
| "learning_rate": 5.000000000000003e-06, | |
| "loss": 0.1289, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 2.1022979985174204, | |
| "grad_norm": 0.34547021985054016, | |
| "learning_rate": 4.970099238502494e-06, | |
| "loss": 0.13, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 2.1052631578947367, | |
| "grad_norm": 0.3479922413825989, | |
| "learning_rate": 4.940258557148765e-06, | |
| "loss": 0.1289, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.1082283172720535, | |
| "grad_norm": 0.34653687477111816, | |
| "learning_rate": 4.910478312373757e-06, | |
| "loss": 0.1274, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 2.11119347664937, | |
| "grad_norm": 0.35768458247184753, | |
| "learning_rate": 4.8807588598905364e-06, | |
| "loss": 0.1263, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 2.1141586360266866, | |
| "grad_norm": 0.3689504563808441, | |
| "learning_rate": 4.8511005546860214e-06, | |
| "loss": 0.131, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 2.117123795404003, | |
| "grad_norm": 0.36393120884895325, | |
| "learning_rate": 4.821503751016746e-06, | |
| "loss": 0.132, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 2.1200889547813193, | |
| "grad_norm": 0.331999272108078, | |
| "learning_rate": 4.791968802404648e-06, | |
| "loss": 0.1315, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 2.123054114158636, | |
| "grad_norm": 0.3519760072231293, | |
| "learning_rate": 4.762496061632814e-06, | |
| "loss": 0.1283, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 2.1260192735359524, | |
| "grad_norm": 0.3459528088569641, | |
| "learning_rate": 4.733085880741301e-06, | |
| "loss": 0.1281, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 2.128984432913269, | |
| "grad_norm": 0.3726547956466675, | |
| "learning_rate": 4.703738611022899e-06, | |
| "loss": 0.1304, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 2.1319495922905856, | |
| "grad_norm": 0.33399152755737305, | |
| "learning_rate": 4.674454603018949e-06, | |
| "loss": 0.1281, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 2.1349147516679023, | |
| "grad_norm": 0.3605293333530426, | |
| "learning_rate": 4.645234206515171e-06, | |
| "loss": 0.1276, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.1378799110452187, | |
| "grad_norm": 0.34493979811668396, | |
| "learning_rate": 4.616077770537453e-06, | |
| "loss": 0.1297, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 2.140845070422535, | |
| "grad_norm": 0.34895074367523193, | |
| "learning_rate": 4.586985643347716e-06, | |
| "loss": 0.1325, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 2.143810229799852, | |
| "grad_norm": 0.3834850788116455, | |
| "learning_rate": 4.557958172439726e-06, | |
| "loss": 0.1307, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 2.146775389177168, | |
| "grad_norm": 0.36474162340164185, | |
| "learning_rate": 4.5289957045349655e-06, | |
| "loss": 0.1319, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 2.149740548554485, | |
| "grad_norm": 0.36195108294487, | |
| "learning_rate": 4.500098585578475e-06, | |
| "loss": 0.1291, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 2.1527057079318013, | |
| "grad_norm": 0.3537023663520813, | |
| "learning_rate": 4.471267160734731e-06, | |
| "loss": 0.1301, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 2.155670867309118, | |
| "grad_norm": 0.33723926544189453, | |
| "learning_rate": 4.4425017743835155e-06, | |
| "loss": 0.1305, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 2.1586360266864344, | |
| "grad_norm": 0.3564864993095398, | |
| "learning_rate": 4.413802770115816e-06, | |
| "loss": 0.127, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 2.1616011860637507, | |
| "grad_norm": 0.35434702038764954, | |
| "learning_rate": 4.385170490729712e-06, | |
| "loss": 0.13, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 2.1645663454410675, | |
| "grad_norm": 0.35794782638549805, | |
| "learning_rate": 4.356605278226274e-06, | |
| "loss": 0.1312, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.167531504818384, | |
| "grad_norm": 0.3571893870830536, | |
| "learning_rate": 4.328107473805487e-06, | |
| "loss": 0.1275, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 2.1704966641957006, | |
| "grad_norm": 0.34572115540504456, | |
| "learning_rate": 4.299677417862174e-06, | |
| "loss": 0.1319, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 2.173461823573017, | |
| "grad_norm": 0.33964014053344727, | |
| "learning_rate": 4.2713154499819345e-06, | |
| "loss": 0.1304, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 2.1764269829503338, | |
| "grad_norm": 0.34920188784599304, | |
| "learning_rate": 4.243021908937083e-06, | |
| "loss": 0.1294, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 2.17939214232765, | |
| "grad_norm": 0.33510822057724, | |
| "learning_rate": 4.214797132682597e-06, | |
| "loss": 0.129, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 2.1823573017049664, | |
| "grad_norm": 0.33283138275146484, | |
| "learning_rate": 4.186641458352088e-06, | |
| "loss": 0.1271, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 2.1853224610822832, | |
| "grad_norm": 0.3561367988586426, | |
| "learning_rate": 4.158555222253772e-06, | |
| "loss": 0.1311, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 2.1882876204595996, | |
| "grad_norm": 0.36620989441871643, | |
| "learning_rate": 4.130538759866457e-06, | |
| "loss": 0.1317, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 2.1912527798369164, | |
| "grad_norm": 0.337467759847641, | |
| "learning_rate": 4.102592405835536e-06, | |
| "loss": 0.1297, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 2.1942179392142327, | |
| "grad_norm": 0.3357710540294647, | |
| "learning_rate": 4.074716493968976e-06, | |
| "loss": 0.1314, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 2.1971830985915495, | |
| "grad_norm": 0.35487931966781616, | |
| "learning_rate": 4.046911357233343e-06, | |
| "loss": 0.1299, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 2.200148257968866, | |
| "grad_norm": 0.34735655784606934, | |
| "learning_rate": 4.019177327749822e-06, | |
| "loss": 0.1324, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 2.203113417346182, | |
| "grad_norm": 0.3381595313549042, | |
| "learning_rate": 3.991514736790259e-06, | |
| "loss": 0.1325, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 2.206078576723499, | |
| "grad_norm": 0.33680617809295654, | |
| "learning_rate": 3.9639239147731865e-06, | |
| "loss": 0.1299, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 2.2090437361008153, | |
| "grad_norm": 0.33319249749183655, | |
| "learning_rate": 3.936405191259891e-06, | |
| "loss": 0.1291, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 2.212008895478132, | |
| "grad_norm": 0.32937324047088623, | |
| "learning_rate": 3.908958894950465e-06, | |
| "loss": 0.1306, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 2.2149740548554484, | |
| "grad_norm": 0.3424176573753357, | |
| "learning_rate": 3.881585353679891e-06, | |
| "loss": 0.1294, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 2.217939214232765, | |
| "grad_norm": 0.33413031697273254, | |
| "learning_rate": 3.854284894414122e-06, | |
| "loss": 0.1293, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 2.2209043736100815, | |
| "grad_norm": 0.33275535702705383, | |
| "learning_rate": 3.827057843246181e-06, | |
| "loss": 0.131, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 2.223869532987398, | |
| "grad_norm": 0.33580246567726135, | |
| "learning_rate": 3.799904525392251e-06, | |
| "loss": 0.1305, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.2268346923647147, | |
| "grad_norm": 0.34102970361709595, | |
| "learning_rate": 3.7728252651878018e-06, | |
| "loss": 0.1304, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 2.229799851742031, | |
| "grad_norm": 0.3335299491882324, | |
| "learning_rate": 3.745820386083724e-06, | |
| "loss": 0.1301, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 2.2327650111193478, | |
| "grad_norm": 0.33244848251342773, | |
| "learning_rate": 3.718890210642442e-06, | |
| "loss": 0.1289, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 2.235730170496664, | |
| "grad_norm": 0.3514939248561859, | |
| "learning_rate": 3.6920350605340883e-06, | |
| "loss": 0.1292, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 2.238695329873981, | |
| "grad_norm": 0.34593474864959717, | |
| "learning_rate": 3.6652552565326382e-06, | |
| "loss": 0.1308, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 2.2416604892512972, | |
| "grad_norm": 0.3418528139591217, | |
| "learning_rate": 3.638551118512089e-06, | |
| "loss": 0.13, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 2.244625648628614, | |
| "grad_norm": 0.35408085584640503, | |
| "learning_rate": 3.611922965442648e-06, | |
| "loss": 0.1278, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 2.2475908080059304, | |
| "grad_norm": 0.33797547221183777, | |
| "learning_rate": 3.5853711153868962e-06, | |
| "loss": 0.1296, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 2.2505559673832467, | |
| "grad_norm": 0.3515971302986145, | |
| "learning_rate": 3.558895885496023e-06, | |
| "loss": 0.1311, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 2.2535211267605635, | |
| "grad_norm": 0.34564974904060364, | |
| "learning_rate": 3.53249759200601e-06, | |
| "loss": 0.1271, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.25648628613788, | |
| "grad_norm": 0.3417358696460724, | |
| "learning_rate": 3.506176550233863e-06, | |
| "loss": 0.1323, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 2.2594514455151966, | |
| "grad_norm": 0.342887282371521, | |
| "learning_rate": 3.479933074573858e-06, | |
| "loss": 0.1305, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 2.262416604892513, | |
| "grad_norm": 0.3478313386440277, | |
| "learning_rate": 3.453767478493761e-06, | |
| "loss": 0.1303, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 2.2653817642698293, | |
| "grad_norm": 0.33934473991394043, | |
| "learning_rate": 3.4276800745311135e-06, | |
| "loss": 0.1288, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 2.268346923647146, | |
| "grad_norm": 0.31560465693473816, | |
| "learning_rate": 3.401671174289469e-06, | |
| "loss": 0.128, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 2.2713120830244624, | |
| "grad_norm": 0.34385186433792114, | |
| "learning_rate": 3.37574108843469e-06, | |
| "loss": 0.127, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 2.274277242401779, | |
| "grad_norm": 0.32480913400650024, | |
| "learning_rate": 3.3498901266912397e-06, | |
| "loss": 0.128, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 2.2772424017790955, | |
| "grad_norm": 0.3512122631072998, | |
| "learning_rate": 3.3241185978384636e-06, | |
| "loss": 0.1276, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 2.2802075611564123, | |
| "grad_norm": 0.34826409816741943, | |
| "learning_rate": 3.2984268097069284e-06, | |
| "loss": 0.13, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 2.2831727205337287, | |
| "grad_norm": 0.34755197167396545, | |
| "learning_rate": 3.2728150691747117e-06, | |
| "loss": 0.1314, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.2861378799110454, | |
| "grad_norm": 0.3306916654109955, | |
| "learning_rate": 3.2472836821637744e-06, | |
| "loss": 0.1314, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 2.289103039288362, | |
| "grad_norm": 0.3324066698551178, | |
| "learning_rate": 3.22183295363627e-06, | |
| "loss": 0.1304, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 2.292068198665678, | |
| "grad_norm": 0.34940484166145325, | |
| "learning_rate": 3.196463187590929e-06, | |
| "loss": 0.1351, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 2.295033358042995, | |
| "grad_norm": 0.34311389923095703, | |
| "learning_rate": 3.1711746870594083e-06, | |
| "loss": 0.1299, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 2.2979985174203112, | |
| "grad_norm": 0.3504948318004608, | |
| "learning_rate": 3.145967754102691e-06, | |
| "loss": 0.1314, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 2.300963676797628, | |
| "grad_norm": 0.33524277806282043, | |
| "learning_rate": 3.1208426898074685e-06, | |
| "loss": 0.1326, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 2.3039288361749444, | |
| "grad_norm": 0.3370322287082672, | |
| "learning_rate": 3.0957997942825337e-06, | |
| "loss": 0.1301, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 2.3068939955522607, | |
| "grad_norm": 0.33946508169174194, | |
| "learning_rate": 3.070839366655215e-06, | |
| "loss": 0.1297, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 2.3098591549295775, | |
| "grad_norm": 0.343334436416626, | |
| "learning_rate": 3.045961705067787e-06, | |
| "loss": 0.1279, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 2.312824314306894, | |
| "grad_norm": 0.34265804290771484, | |
| "learning_rate": 3.021167106673928e-06, | |
| "loss": 0.1314, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.3157894736842106, | |
| "grad_norm": 0.33246049284935, | |
| "learning_rate": 2.996455867635155e-06, | |
| "loss": 0.1306, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 2.318754633061527, | |
| "grad_norm": 0.34323611855506897, | |
| "learning_rate": 2.9718282831172885e-06, | |
| "loss": 0.1318, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 2.3217197924388437, | |
| "grad_norm": 0.34068265557289124, | |
| "learning_rate": 2.94728464728693e-06, | |
| "loss": 0.1292, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 2.32468495181616, | |
| "grad_norm": 0.3370424807071686, | |
| "learning_rate": 2.922825253307947e-06, | |
| "loss": 0.129, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 2.327650111193477, | |
| "grad_norm": 0.3519260883331299, | |
| "learning_rate": 2.898450393337977e-06, | |
| "loss": 0.1292, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 2.330615270570793, | |
| "grad_norm": 0.33347323536872864, | |
| "learning_rate": 2.8741603585249312e-06, | |
| "loss": 0.1261, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 2.3335804299481095, | |
| "grad_norm": 0.3215949833393097, | |
| "learning_rate": 2.8499554390035144e-06, | |
| "loss": 0.1294, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 2.3365455893254263, | |
| "grad_norm": 0.32965582609176636, | |
| "learning_rate": 2.8258359238917665e-06, | |
| "loss": 0.1281, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 2.3395107487027427, | |
| "grad_norm": 0.33794647455215454, | |
| "learning_rate": 2.8018021012875994e-06, | |
| "loss": 0.1285, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 2.3424759080800595, | |
| "grad_norm": 0.32937586307525635, | |
| "learning_rate": 2.7778542582653746e-06, | |
| "loss": 0.128, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.345441067457376, | |
| "grad_norm": 0.3328467607498169, | |
| "learning_rate": 2.753992680872457e-06, | |
| "loss": 0.129, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 2.348406226834692, | |
| "grad_norm": 0.32725760340690613, | |
| "learning_rate": 2.7302176541257984e-06, | |
| "loss": 0.1294, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 2.351371386212009, | |
| "grad_norm": 0.3363383114337921, | |
| "learning_rate": 2.7065294620085425e-06, | |
| "loss": 0.129, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 2.3543365455893253, | |
| "grad_norm": 0.33696410059928894, | |
| "learning_rate": 2.6829283874666236e-06, | |
| "loss": 0.1285, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 2.357301704966642, | |
| "grad_norm": 0.33244121074676514, | |
| "learning_rate": 2.6594147124053983e-06, | |
| "loss": 0.1277, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 2.3602668643439584, | |
| "grad_norm": 0.3353787958621979, | |
| "learning_rate": 2.635988717686272e-06, | |
| "loss": 0.1304, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 2.363232023721275, | |
| "grad_norm": 0.34110620617866516, | |
| "learning_rate": 2.6126506831233343e-06, | |
| "loss": 0.1306, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 2.3661971830985915, | |
| "grad_norm": 0.33648866415023804, | |
| "learning_rate": 2.5894008874800323e-06, | |
| "loss": 0.1286, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 2.3691623424759083, | |
| "grad_norm": 0.34967437386512756, | |
| "learning_rate": 2.5662396084658383e-06, | |
| "loss": 0.133, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 2.3721275018532246, | |
| "grad_norm": 0.33198100328445435, | |
| "learning_rate": 2.543167122732918e-06, | |
| "loss": 0.1277, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 2.375092661230541, | |
| "grad_norm": 0.34363454580307007, | |
| "learning_rate": 2.5201837058728506e-06, | |
| "loss": 0.1277, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 2.3780578206078578, | |
| "grad_norm": 0.3609948456287384, | |
| "learning_rate": 2.4972896324133143e-06, | |
| "loss": 0.1295, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 2.381022979985174, | |
| "grad_norm": 0.34460243582725525, | |
| "learning_rate": 2.474485175814816e-06, | |
| "loss": 0.1319, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 2.383988139362491, | |
| "grad_norm": 0.3403383493423462, | |
| "learning_rate": 2.451770608467432e-06, | |
| "loss": 0.1284, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 2.386953298739807, | |
| "grad_norm": 0.333807110786438, | |
| "learning_rate": 2.429146201687538e-06, | |
| "loss": 0.1257, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 2.3899184581171236, | |
| "grad_norm": 0.33072689175605774, | |
| "learning_rate": 2.4066122257145898e-06, | |
| "loss": 0.1309, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 2.3928836174944403, | |
| "grad_norm": 0.32463690638542175, | |
| "learning_rate": 2.3841689497078746e-06, | |
| "loss": 0.1289, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 2.3958487768717567, | |
| "grad_norm": 0.34213897585868835, | |
| "learning_rate": 2.361816641743303e-06, | |
| "loss": 0.1286, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 2.3988139362490735, | |
| "grad_norm": 0.3414537310600281, | |
| "learning_rate": 2.339555568810221e-06, | |
| "loss": 0.126, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 2.40177909562639, | |
| "grad_norm": 0.32957902550697327, | |
| "learning_rate": 2.317385996808195e-06, | |
| "loss": 0.1302, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 2.4047442550037066, | |
| "grad_norm": 0.3390369713306427, | |
| "learning_rate": 2.295308190543859e-06, | |
| "loss": 0.132, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 2.407709414381023, | |
| "grad_norm": 0.3288882076740265, | |
| "learning_rate": 2.2733224137277366e-06, | |
| "loss": 0.1271, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 2.4106745737583397, | |
| "grad_norm": 0.3289991021156311, | |
| "learning_rate": 2.251428928971102e-06, | |
| "loss": 0.1304, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 2.413639733135656, | |
| "grad_norm": 0.33164265751838684, | |
| "learning_rate": 2.229627997782834e-06, | |
| "loss": 0.1296, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 2.4166048925129724, | |
| "grad_norm": 0.33751052618026733, | |
| "learning_rate": 2.2079198805662917e-06, | |
| "loss": 0.1282, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 2.419570051890289, | |
| "grad_norm": 0.3279024362564087, | |
| "learning_rate": 2.186304836616221e-06, | |
| "loss": 0.1295, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 2.4225352112676055, | |
| "grad_norm": 0.3452274203300476, | |
| "learning_rate": 2.1647831241156304e-06, | |
| "loss": 0.1299, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 2.4255003706449223, | |
| "grad_norm": 0.3305584788322449, | |
| "learning_rate": 2.1433550001327376e-06, | |
| "loss": 0.1285, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 2.4284655300222386, | |
| "grad_norm": 0.33620432019233704, | |
| "learning_rate": 2.122020720617869e-06, | |
| "loss": 0.1304, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 2.431430689399555, | |
| "grad_norm": 0.3142911493778229, | |
| "learning_rate": 2.1007805404004247e-06, | |
| "loss": 0.125, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 2.4343958487768718, | |
| "grad_norm": 0.3442496657371521, | |
| "learning_rate": 2.0796347131858187e-06, | |
| "loss": 0.1286, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 2.437361008154188, | |
| "grad_norm": 0.34949377179145813, | |
| "learning_rate": 2.058583491552465e-06, | |
| "loss": 0.1284, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 2.440326167531505, | |
| "grad_norm": 0.36079153418540955, | |
| "learning_rate": 2.037627126948751e-06, | |
| "loss": 0.1303, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 2.4432913269088212, | |
| "grad_norm": 0.32977890968322754, | |
| "learning_rate": 2.0167658696900317e-06, | |
| "loss": 0.1279, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 2.446256486286138, | |
| "grad_norm": 0.3395943343639374, | |
| "learning_rate": 1.9959999689556407e-06, | |
| "loss": 0.1295, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.4492216456634543, | |
| "grad_norm": 0.3250430226325989, | |
| "learning_rate": 1.9753296727859195e-06, | |
| "loss": 0.1287, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 2.452186805040771, | |
| "grad_norm": 0.3329125642776489, | |
| "learning_rate": 1.9547552280792528e-06, | |
| "loss": 0.1278, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 2.4551519644180875, | |
| "grad_norm": 0.31633639335632324, | |
| "learning_rate": 1.9342768805891176e-06, | |
| "loss": 0.1291, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 2.458117123795404, | |
| "grad_norm": 0.3292962610721588, | |
| "learning_rate": 1.9138948749211473e-06, | |
| "loss": 0.1297, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 2.4610822831727206, | |
| "grad_norm": 0.34126242995262146, | |
| "learning_rate": 1.8936094545302098e-06, | |
| "loss": 0.1293, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 2.464047442550037, | |
| "grad_norm": 0.3327971398830414, | |
| "learning_rate": 1.8734208617174986e-06, | |
| "loss": 0.1284, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 2.4670126019273537, | |
| "grad_norm": 0.340774804353714, | |
| "learning_rate": 1.8533293376276473e-06, | |
| "loss": 0.129, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 2.46997776130467, | |
| "grad_norm": 0.3464578688144684, | |
| "learning_rate": 1.8333351222458407e-06, | |
| "loss": 0.1277, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 2.472942920681987, | |
| "grad_norm": 0.340108722448349, | |
| "learning_rate": 1.813438454394948e-06, | |
| "loss": 0.1304, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 2.475908080059303, | |
| "grad_norm": 0.36126676201820374, | |
| "learning_rate": 1.7936395717326705e-06, | |
| "loss": 0.1275, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 2.4788732394366195, | |
| "grad_norm": 0.3317781388759613, | |
| "learning_rate": 1.773938710748706e-06, | |
| "loss": 0.1301, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 2.4818383988139363, | |
| "grad_norm": 0.34120678901672363, | |
| "learning_rate": 1.7543361067619269e-06, | |
| "loss": 0.1287, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 2.4848035581912526, | |
| "grad_norm": 0.3353835642337799, | |
| "learning_rate": 1.734831993917564e-06, | |
| "loss": 0.1296, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 2.4877687175685694, | |
| "grad_norm": 0.34985971450805664, | |
| "learning_rate": 1.715426605184407e-06, | |
| "loss": 0.129, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 2.4907338769458858, | |
| "grad_norm": 0.3302218019962311, | |
| "learning_rate": 1.6961201723520248e-06, | |
| "loss": 0.131, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 2.4936990363232026, | |
| "grad_norm": 0.34012821316719055, | |
| "learning_rate": 1.676912926028007e-06, | |
| "loss": 0.1301, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 2.496664195700519, | |
| "grad_norm": 0.3237687945365906, | |
| "learning_rate": 1.6578050956351887e-06, | |
| "loss": 0.1257, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 2.4996293550778352, | |
| "grad_norm": 0.3470035791397095, | |
| "learning_rate": 1.6387969094089318e-06, | |
| "loss": 0.1287, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 2.502594514455152, | |
| "grad_norm": 0.35050496459007263, | |
| "learning_rate": 1.619888594394382e-06, | |
| "loss": 0.1314, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 2.5055596738324684, | |
| "grad_norm": 0.3287401795387268, | |
| "learning_rate": 1.6010803764437633e-06, | |
| "loss": 0.1285, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 2.508524833209785, | |
| "grad_norm": 0.34805530309677124, | |
| "learning_rate": 1.5823724802136863e-06, | |
| "loss": 0.1313, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 2.5114899925871015, | |
| "grad_norm": 0.33040815591812134, | |
| "learning_rate": 1.5637651291624522e-06, | |
| "loss": 0.1284, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 2.514455151964418, | |
| "grad_norm": 0.340082049369812, | |
| "learning_rate": 1.545258545547398e-06, | |
| "loss": 0.1258, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 2.5174203113417346, | |
| "grad_norm": 0.3319970965385437, | |
| "learning_rate": 1.5268529504222262e-06, | |
| "loss": 0.1278, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 2.5203854707190514, | |
| "grad_norm": 0.327903151512146, | |
| "learning_rate": 1.5085485636343755e-06, | |
| "loss": 0.1272, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.5233506300963677, | |
| "grad_norm": 0.3466844856739044, | |
| "learning_rate": 1.4903456038223941e-06, | |
| "loss": 0.131, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 2.526315789473684, | |
| "grad_norm": 0.3274025619029999, | |
| "learning_rate": 1.4722442884133214e-06, | |
| "loss": 0.127, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 2.529280948851001, | |
| "grad_norm": 0.32809337973594666, | |
| "learning_rate": 1.4542448336201021e-06, | |
| "loss": 0.1265, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 2.532246108228317, | |
| "grad_norm": 0.3453335165977478, | |
| "learning_rate": 1.4363474544389876e-06, | |
| "loss": 0.1295, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 2.535211267605634, | |
| "grad_norm": 0.3447280824184418, | |
| "learning_rate": 1.4185523646469822e-06, | |
| "loss": 0.1312, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 2.5381764269829503, | |
| "grad_norm": 0.33509477972984314, | |
| "learning_rate": 1.4008597767992872e-06, | |
| "loss": 0.1283, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 2.5411415863602667, | |
| "grad_norm": 0.3374352753162384, | |
| "learning_rate": 1.3832699022267516e-06, | |
| "loss": 0.1277, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 2.5441067457375834, | |
| "grad_norm": 0.3189197778701782, | |
| "learning_rate": 1.3657829510333653e-06, | |
| "loss": 0.1261, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 2.5470719051149, | |
| "grad_norm": 0.34467366337776184, | |
| "learning_rate": 1.3483991320937307e-06, | |
| "loss": 0.1295, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 2.5500370644922166, | |
| "grad_norm": 0.33318278193473816, | |
| "learning_rate": 1.3311186530505838e-06, | |
| "loss": 0.1271, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 2.553002223869533, | |
| "grad_norm": 0.3337114453315735, | |
| "learning_rate": 1.313941720312303e-06, | |
| "loss": 0.133, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 2.5559673832468492, | |
| "grad_norm": 0.33227020502090454, | |
| "learning_rate": 1.2968685390504465e-06, | |
| "loss": 0.1277, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 2.558932542624166, | |
| "grad_norm": 0.3402811288833618, | |
| "learning_rate": 1.2798993131973093e-06, | |
| "loss": 0.128, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 2.561897702001483, | |
| "grad_norm": 0.32487955689430237, | |
| "learning_rate": 1.263034245443473e-06, | |
| "loss": 0.1296, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 2.564862861378799, | |
| "grad_norm": 0.3243284523487091, | |
| "learning_rate": 1.2462735372353996e-06, | |
| "loss": 0.1262, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 2.5678280207561155, | |
| "grad_norm": 0.33498314023017883, | |
| "learning_rate": 1.2296173887730122e-06, | |
| "loss": 0.1311, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 2.5707931801334323, | |
| "grad_norm": 0.32444214820861816, | |
| "learning_rate": 1.2130659990073146e-06, | |
| "loss": 0.1251, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 2.5737583395107486, | |
| "grad_norm": 0.3283936083316803, | |
| "learning_rate": 1.196619565638003e-06, | |
| "loss": 0.1266, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 2.5767234988880654, | |
| "grad_norm": 0.33177807927131653, | |
| "learning_rate": 1.1802782851111206e-06, | |
| "loss": 0.1277, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 2.5796886582653817, | |
| "grad_norm": 0.327374130487442, | |
| "learning_rate": 1.1640423526166987e-06, | |
| "loss": 0.1273, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 2.582653817642698, | |
| "grad_norm": 0.3298618495464325, | |
| "learning_rate": 1.1479119620864277e-06, | |
| "loss": 0.1278, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 2.585618977020015, | |
| "grad_norm": 0.34262576699256897, | |
| "learning_rate": 1.1318873061913405e-06, | |
| "loss": 0.1253, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 2.588584136397331, | |
| "grad_norm": 0.33369916677474976, | |
| "learning_rate": 1.1159685763395113e-06, | |
| "loss": 0.1277, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 2.591549295774648, | |
| "grad_norm": 0.32637131214141846, | |
| "learning_rate": 1.1001559626737757e-06, | |
| "loss": 0.1285, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 2.5945144551519643, | |
| "grad_norm": 0.33180394768714905, | |
| "learning_rate": 1.0844496540694515e-06, | |
| "loss": 0.1294, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.597479614529281, | |
| "grad_norm": 0.36661967635154724, | |
| "learning_rate": 1.0688498381320855e-06, | |
| "loss": 0.127, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 2.6004447739065975, | |
| "grad_norm": 0.32528406381607056, | |
| "learning_rate": 1.0533567011952094e-06, | |
| "loss": 0.1253, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 2.6034099332839142, | |
| "grad_norm": 0.33627548813819885, | |
| "learning_rate": 1.037970428318118e-06, | |
| "loss": 0.1258, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 2.6063750926612306, | |
| "grad_norm": 0.329609215259552, | |
| "learning_rate": 1.022691203283661e-06, | |
| "loss": 0.1268, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 2.609340252038547, | |
| "grad_norm": 0.3270719647407532, | |
| "learning_rate": 1.0075192085960451e-06, | |
| "loss": 0.1282, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 2.6123054114158637, | |
| "grad_norm": 0.3354145586490631, | |
| "learning_rate": 9.924546254786493e-07, | |
| "loss": 0.1285, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 2.61527057079318, | |
| "grad_norm": 0.32381850481033325, | |
| "learning_rate": 9.77497633871868e-07, | |
| "loss": 0.1294, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 2.618235730170497, | |
| "grad_norm": 0.32297268509864807, | |
| "learning_rate": 9.62648412430951e-07, | |
| "loss": 0.1268, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 2.621200889547813, | |
| "grad_norm": 0.3353489339351654, | |
| "learning_rate": 9.479071385238892e-07, | |
| "loss": 0.1263, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 2.6241660489251295, | |
| "grad_norm": 0.331815630197525, | |
| "learning_rate": 9.332739882292752e-07, | |
| "loss": 0.128, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 2.6271312083024463, | |
| "grad_norm": 0.33713892102241516, | |
| "learning_rate": 9.187491363342094e-07, | |
| "loss": 0.1269, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 2.6300963676797626, | |
| "grad_norm": 0.3313647508621216, | |
| "learning_rate": 9.043327563322113e-07, | |
| "loss": 0.1305, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 2.6330615270570794, | |
| "grad_norm": 0.32444262504577637, | |
| "learning_rate": 8.900250204211513e-07, | |
| "loss": 0.1291, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 2.6360266864343957, | |
| "grad_norm": 0.34167933464050293, | |
| "learning_rate": 8.758260995011825e-07, | |
| "loss": 0.1263, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 2.6389918458117125, | |
| "grad_norm": 0.3300521671772003, | |
| "learning_rate": 8.617361631727139e-07, | |
| "loss": 0.1258, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 2.641957005189029, | |
| "grad_norm": 0.3591514527797699, | |
| "learning_rate": 8.477553797343729e-07, | |
| "loss": 0.1268, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 2.6449221645663457, | |
| "grad_norm": 0.3284503221511841, | |
| "learning_rate": 8.338839161809997e-07, | |
| "loss": 0.127, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 2.647887323943662, | |
| "grad_norm": 0.3253602981567383, | |
| "learning_rate": 8.201219382016556e-07, | |
| "loss": 0.1259, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 2.6508524833209783, | |
| "grad_norm": 0.3226112723350525, | |
| "learning_rate": 8.06469610177636e-07, | |
| "loss": 0.1264, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 2.653817642698295, | |
| "grad_norm": 0.3329734206199646, | |
| "learning_rate": 7.92927095180518e-07, | |
| "loss": 0.1277, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 2.6567828020756115, | |
| "grad_norm": 0.36240342259407043, | |
| "learning_rate": 7.794945549701993e-07, | |
| "loss": 0.1286, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 2.6597479614529282, | |
| "grad_norm": 0.3200359642505646, | |
| "learning_rate": 7.661721499929753e-07, | |
| "loss": 0.1277, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 2.6627131208302446, | |
| "grad_norm": 0.33148688077926636, | |
| "learning_rate": 7.529600393796232e-07, | |
| "loss": 0.1277, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 2.665678280207561, | |
| "grad_norm": 0.32987260818481445, | |
| "learning_rate": 7.398583809434944e-07, | |
| "loss": 0.128, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 2.6686434395848777, | |
| "grad_norm": 0.33015844225883484, | |
| "learning_rate": 7.268673311786378e-07, | |
| "loss": 0.1307, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.6716085989621945, | |
| "grad_norm": 0.32374393939971924, | |
| "learning_rate": 7.1398704525792e-07, | |
| "loss": 0.1277, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 2.674573758339511, | |
| "grad_norm": 0.318718284368515, | |
| "learning_rate": 7.012176770311863e-07, | |
| "loss": 0.1266, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 2.677538917716827, | |
| "grad_norm": 0.3262283205986023, | |
| "learning_rate": 6.885593790234057e-07, | |
| "loss": 0.1251, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 2.680504077094144, | |
| "grad_norm": 0.3396647274494171, | |
| "learning_rate": 6.760123024328624e-07, | |
| "loss": 0.1303, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 2.6834692364714603, | |
| "grad_norm": 0.3207716643810272, | |
| "learning_rate": 6.635765971293484e-07, | |
| "loss": 0.1274, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 2.686434395848777, | |
| "grad_norm": 0.32596075534820557, | |
| "learning_rate": 6.512524116523633e-07, | |
| "loss": 0.1257, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 2.6893995552260934, | |
| "grad_norm": 0.322693407535553, | |
| "learning_rate": 6.390398932093555e-07, | |
| "loss": 0.1248, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 2.6923647146034098, | |
| "grad_norm": 0.3405155837535858, | |
| "learning_rate": 6.269391876739494e-07, | |
| "loss": 0.1291, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 2.6953298739807265, | |
| "grad_norm": 0.32777202129364014, | |
| "learning_rate": 6.149504395842087e-07, | |
| "loss": 0.1288, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 2.698295033358043, | |
| "grad_norm": 0.3245905935764313, | |
| "learning_rate": 6.030737921409169e-07, | |
| "loss": 0.1261, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.7012601927353597, | |
| "grad_norm": 0.33581435680389404, | |
| "learning_rate": 5.913093872058528e-07, | |
| "loss": 0.1302, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 2.704225352112676, | |
| "grad_norm": 0.3299258053302765, | |
| "learning_rate": 5.796573653001091e-07, | |
| "loss": 0.1264, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 2.7071905114899923, | |
| "grad_norm": 0.3272840082645416, | |
| "learning_rate": 5.681178656024055e-07, | |
| "loss": 0.1269, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 2.710155670867309, | |
| "grad_norm": 0.34318456053733826, | |
| "learning_rate": 5.56691025947429e-07, | |
| "loss": 0.1267, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 2.713120830244626, | |
| "grad_norm": 0.32936856150627136, | |
| "learning_rate": 5.453769828241872e-07, | |
| "loss": 0.1259, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 2.7160859896219423, | |
| "grad_norm": 0.32494810223579407, | |
| "learning_rate": 5.341758713743828e-07, | |
| "loss": 0.1281, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 2.7190511489992586, | |
| "grad_norm": 0.3184977173805237, | |
| "learning_rate": 5.230878253907911e-07, | |
| "loss": 0.1271, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 2.7220163083765754, | |
| "grad_norm": 0.3371221125125885, | |
| "learning_rate": 5.121129773156663e-07, | |
| "loss": 0.1322, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 2.7249814677538917, | |
| "grad_norm": 0.32113394141197205, | |
| "learning_rate": 5.012514582391592e-07, | |
| "loss": 0.1272, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 2.7279466271312085, | |
| "grad_norm": 0.34688618779182434, | |
| "learning_rate": 4.905033978977492e-07, | |
| "loss": 0.1265, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.730911786508525, | |
| "grad_norm": 0.3413783311843872, | |
| "learning_rate": 4.798689246727006e-07, | |
| "loss": 0.1307, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 2.733876945885841, | |
| "grad_norm": 0.33555951714515686, | |
| "learning_rate": 4.693481655885257e-07, | |
| "loss": 0.1269, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 2.736842105263158, | |
| "grad_norm": 0.3346193730831146, | |
| "learning_rate": 4.58941246311464e-07, | |
| "loss": 0.1268, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 2.7398072646404743, | |
| "grad_norm": 0.3286806344985962, | |
| "learning_rate": 4.4864829114798394e-07, | |
| "loss": 0.1288, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 2.742772424017791, | |
| "grad_norm": 0.33568400144577026, | |
| "learning_rate": 4.384694230432984e-07, | |
| "loss": 0.1269, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.7457375833951074, | |
| "grad_norm": 0.3334142565727234, | |
| "learning_rate": 4.2840476357989825e-07, | |
| "loss": 0.1272, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 2.7487027427724238, | |
| "grad_norm": 0.32712700963020325, | |
| "learning_rate": 4.184544329761009e-07, | |
| "loss": 0.1271, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 2.7516679021497406, | |
| "grad_norm": 0.3435976803302765, | |
| "learning_rate": 4.0861855008460403e-07, | |
| "loss": 0.1286, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 2.7546330615270573, | |
| "grad_norm": 0.3265362083911896, | |
| "learning_rate": 3.988972323910778e-07, | |
| "loss": 0.1281, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 2.7575982209043737, | |
| "grad_norm": 0.32593265175819397, | |
| "learning_rate": 3.8929059601275463e-07, | |
| "loss": 0.1273, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.76056338028169, | |
| "grad_norm": 0.3315712511539459, | |
| "learning_rate": 3.797987556970495e-07, | |
| "loss": 0.1296, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 2.763528539659007, | |
| "grad_norm": 0.32149094343185425, | |
| "learning_rate": 3.7042182482018074e-07, | |
| "loss": 0.1284, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 2.766493699036323, | |
| "grad_norm": 0.3222528100013733, | |
| "learning_rate": 3.611599153858214e-07, | |
| "loss": 0.1259, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 2.76945885841364, | |
| "grad_norm": 0.32701918482780457, | |
| "learning_rate": 3.520131380237546e-07, | |
| "loss": 0.1287, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 2.7724240177909563, | |
| "grad_norm": 0.32082316279411316, | |
| "learning_rate": 3.429816019885657e-07, | |
| "loss": 0.1279, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 2.7753891771682726, | |
| "grad_norm": 0.32540014386177063, | |
| "learning_rate": 3.3406541515832e-07, | |
| "loss": 0.1273, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 2.7783543365455894, | |
| "grad_norm": 0.32940953969955444, | |
| "learning_rate": 3.252646840332918e-07, | |
| "loss": 0.1264, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 2.7813194959229057, | |
| "grad_norm": 0.32957059144973755, | |
| "learning_rate": 3.16579513734675e-07, | |
| "loss": 0.128, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 2.7842846553002225, | |
| "grad_norm": 0.31844544410705566, | |
| "learning_rate": 3.080100080033388e-07, | |
| "loss": 0.1268, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 2.787249814677539, | |
| "grad_norm": 0.3339882493019104, | |
| "learning_rate": 2.995562691985898e-07, | |
| "loss": 0.1259, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.790214974054855, | |
| "grad_norm": 0.3345246911048889, | |
| "learning_rate": 2.9121839829693857e-07, | |
| "loss": 0.1284, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 2.793180133432172, | |
| "grad_norm": 0.34839048981666565, | |
| "learning_rate": 2.829964948909048e-07, | |
| "loss": 0.1263, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 2.7961452928094888, | |
| "grad_norm": 0.3264479637145996, | |
| "learning_rate": 2.748906571878207e-07, | |
| "loss": 0.1253, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 2.799110452186805, | |
| "grad_norm": 0.3153613805770874, | |
| "learning_rate": 2.6690098200866097e-07, | |
| "loss": 0.1242, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 2.8020756115641214, | |
| "grad_norm": 0.3332012891769409, | |
| "learning_rate": 2.5902756478688674e-07, | |
| "loss": 0.1271, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 2.805040770941438, | |
| "grad_norm": 0.3120848536491394, | |
| "learning_rate": 2.5127049956730207e-07, | |
| "loss": 0.128, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 2.8080059303187546, | |
| "grad_norm": 0.32999473810195923, | |
| "learning_rate": 2.436298790049363e-07, | |
| "loss": 0.126, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 2.8109710896960713, | |
| "grad_norm": 0.32779738306999207, | |
| "learning_rate": 2.3610579436392999e-07, | |
| "loss": 0.1259, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 2.8139362490733877, | |
| "grad_norm": 0.31936830282211304, | |
| "learning_rate": 2.2869833551645293e-07, | |
| "loss": 0.1241, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 2.816901408450704, | |
| "grad_norm": 0.33030980825424194, | |
| "learning_rate": 2.2140759094162468e-07, | |
| "loss": 0.1274, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.819866567828021, | |
| "grad_norm": 0.34059956669807434, | |
| "learning_rate": 2.1423364772445886e-07, | |
| "loss": 0.1277, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 2.822831727205337, | |
| "grad_norm": 0.3205658793449402, | |
| "learning_rate": 2.071765915548274e-07, | |
| "loss": 0.1306, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 2.825796886582654, | |
| "grad_norm": 0.34191787242889404, | |
| "learning_rate": 2.002365067264289e-07, | |
| "loss": 0.1269, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 2.8287620459599703, | |
| "grad_norm": 0.3269594609737396, | |
| "learning_rate": 1.9341347613579086e-07, | |
| "loss": 0.1283, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 2.8317272053372866, | |
| "grad_norm": 0.33148735761642456, | |
| "learning_rate": 1.867075812812691e-07, | |
| "loss": 0.1298, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 2.8346923647146034, | |
| "grad_norm": 0.3186003267765045, | |
| "learning_rate": 1.8011890226208527e-07, | |
| "loss": 0.1274, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 2.83765752409192, | |
| "grad_norm": 0.32092559337615967, | |
| "learning_rate": 1.7364751777736334e-07, | |
| "loss": 0.1242, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 2.8406226834692365, | |
| "grad_norm": 0.36179545521736145, | |
| "learning_rate": 1.6729350512519006e-07, | |
| "loss": 0.129, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 2.843587842846553, | |
| "grad_norm": 0.33006298542022705, | |
| "learning_rate": 1.6105694020169594e-07, | |
| "loss": 0.1258, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 2.8465530022238696, | |
| "grad_norm": 0.31537604331970215, | |
| "learning_rate": 1.5493789750014032e-07, | |
| "loss": 0.1283, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.849518161601186, | |
| "grad_norm": 0.33820608258247375, | |
| "learning_rate": 1.489364501100332e-07, | |
| "loss": 0.1275, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 2.8524833209785028, | |
| "grad_norm": 0.3154459297657013, | |
| "learning_rate": 1.430526697162482e-07, | |
| "loss": 0.1258, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 2.855448480355819, | |
| "grad_norm": 0.31913918256759644, | |
| "learning_rate": 1.3728662659818205e-07, | |
| "loss": 0.1253, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 2.8584136397331354, | |
| "grad_norm": 0.32766804099082947, | |
| "learning_rate": 1.3163838962890196e-07, | |
| "loss": 0.129, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 2.8613787991104522, | |
| "grad_norm": 0.3298415541648865, | |
| "learning_rate": 1.2610802627432972e-07, | |
| "loss": 0.1278, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 2.8643439584877686, | |
| "grad_norm": 0.32275769114494324, | |
| "learning_rate": 1.206956025924333e-07, | |
| "loss": 0.126, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 2.8673091178650854, | |
| "grad_norm": 0.3340933918952942, | |
| "learning_rate": 1.1540118323243866e-07, | |
| "loss": 0.1272, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 2.8702742772424017, | |
| "grad_norm": 0.33475035429000854, | |
| "learning_rate": 1.1022483143405705e-07, | |
| "loss": 0.1265, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 2.873239436619718, | |
| "grad_norm": 0.32354483008384705, | |
| "learning_rate": 1.0516660902673448e-07, | |
| "loss": 0.1255, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 2.876204595997035, | |
| "grad_norm": 0.3189190924167633, | |
| "learning_rate": 1.0022657642890232e-07, | |
| "loss": 0.1254, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.8791697553743516, | |
| "grad_norm": 0.3238016366958618, | |
| "learning_rate": 9.540479264726676e-08, | |
| "loss": 0.1274, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 2.882134914751668, | |
| "grad_norm": 0.3224412798881531, | |
| "learning_rate": 9.070131527609604e-08, | |
| "loss": 0.1271, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 2.8851000741289843, | |
| "grad_norm": 0.34490659832954407, | |
| "learning_rate": 8.61162004965388e-08, | |
| "loss": 0.1277, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 2.888065233506301, | |
| "grad_norm": 0.3256824016571045, | |
| "learning_rate": 8.16495030759501e-08, | |
| "loss": 0.1304, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 2.8910303928836174, | |
| "grad_norm": 0.326412171125412, | |
| "learning_rate": 7.730127636723539e-08, | |
| "loss": 0.1271, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.893995552260934, | |
| "grad_norm": 0.32723942399024963, | |
| "learning_rate": 7.307157230821426e-08, | |
| "loss": 0.1291, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 2.8969607116382505, | |
| "grad_norm": 0.33483996987342834, | |
| "learning_rate": 6.896044142100433e-08, | |
| "loss": 0.1271, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 2.899925871015567, | |
| "grad_norm": 0.3145699203014374, | |
| "learning_rate": 6.496793281141056e-08, | |
| "loss": 0.1257, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 2.9028910303928837, | |
| "grad_norm": 0.3338087201118469, | |
| "learning_rate": 6.109409416834689e-08, | |
| "loss": 0.1272, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 2.9058561897702, | |
| "grad_norm": 0.3297833502292633, | |
| "learning_rate": 5.7338971763256646e-08, | |
| "loss": 0.1263, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.9088213491475168, | |
| "grad_norm": 0.32836630940437317, | |
| "learning_rate": 5.37026104495697e-08, | |
| "loss": 0.1264, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 2.911786508524833, | |
| "grad_norm": 0.32583150267601013, | |
| "learning_rate": 5.0185053662161756e-08, | |
| "loss": 0.1265, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 2.91475166790215, | |
| "grad_norm": 0.32299482822418213, | |
| "learning_rate": 4.678634341683252e-08, | |
| "loss": 0.1253, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 2.9177168272794662, | |
| "grad_norm": 0.32840579748153687, | |
| "learning_rate": 4.350652030981395e-08, | |
| "loss": 0.1286, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 2.920681986656783, | |
| "grad_norm": 0.3269804120063782, | |
| "learning_rate": 4.0345623517273894e-08, | |
| "loss": 0.1284, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 2.9236471460340994, | |
| "grad_norm": 0.31736278533935547, | |
| "learning_rate": 3.7303690794854296e-08, | |
| "loss": 0.1246, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 2.9266123054114157, | |
| "grad_norm": 0.3197997212409973, | |
| "learning_rate": 3.438075847721933e-08, | |
| "loss": 0.1247, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 2.9295774647887325, | |
| "grad_norm": 0.3263581395149231, | |
| "learning_rate": 3.157686147762129e-08, | |
| "loss": 0.1273, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 2.932542624166049, | |
| "grad_norm": 0.32051053643226624, | |
| "learning_rate": 2.8892033287484245e-08, | |
| "loss": 0.1265, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 2.9355077835433656, | |
| "grad_norm": 0.33849623799324036, | |
| "learning_rate": 2.6326305976001054e-08, | |
| "loss": 0.1287, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.938472942920682, | |
| "grad_norm": 0.3170969486236572, | |
| "learning_rate": 2.3879710189753657e-08, | |
| "loss": 0.1252, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 2.9414381022979983, | |
| "grad_norm": 0.32798030972480774, | |
| "learning_rate": 2.1552275152346702e-08, | |
| "loss": 0.1282, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 2.944403261675315, | |
| "grad_norm": 0.3274080157279968, | |
| "learning_rate": 1.9344028664056715e-08, | |
| "loss": 0.1249, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 2.9473684210526314, | |
| "grad_norm": 0.3368877172470093, | |
| "learning_rate": 1.7254997101500137e-08, | |
| "loss": 0.1287, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 2.950333580429948, | |
| "grad_norm": 0.32225024700164795, | |
| "learning_rate": 1.528520541731915e-08, | |
| "loss": 0.1259, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 2.9532987398072645, | |
| "grad_norm": 0.33008435368537903, | |
| "learning_rate": 1.3434677139885222e-08, | |
| "loss": 0.1262, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 2.9562638991845813, | |
| "grad_norm": 0.3370579183101654, | |
| "learning_rate": 1.170343437301491e-08, | |
| "loss": 0.126, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 2.9592290585618977, | |
| "grad_norm": 0.31601622700691223, | |
| "learning_rate": 1.0091497795706728e-08, | |
| "loss": 0.1269, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 2.9621942179392144, | |
| "grad_norm": 0.3216618299484253, | |
| "learning_rate": 8.59888666189579e-09, | |
| "loss": 0.126, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 2.965159377316531, | |
| "grad_norm": 0.3355175852775574, | |
| "learning_rate": 7.225618800222878e-09, | |
| "loss": 0.1278, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.968124536693847, | |
| "grad_norm": 0.32904869318008423, | |
| "learning_rate": 5.971710613821291e-09, | |
| "loss": 0.1284, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 2.971089696071164, | |
| "grad_norm": 0.351557195186615, | |
| "learning_rate": 4.837177080119215e-09, | |
| "loss": 0.1265, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 2.9740548554484803, | |
| "grad_norm": 0.32986804842948914, | |
| "learning_rate": 3.8220317506654226e-09, | |
| "loss": 0.1269, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 2.977020014825797, | |
| "grad_norm": 0.3295051157474518, | |
| "learning_rate": 2.9262867509605164e-09, | |
| "loss": 0.1261, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 2.9799851742031134, | |
| "grad_norm": 0.3266933858394623, | |
| "learning_rate": 2.149952780321485e-09, | |
| "loss": 0.1248, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 2.9829503335804297, | |
| "grad_norm": 0.32243990898132324, | |
| "learning_rate": 1.4930391117451427e-09, | |
| "loss": 0.1262, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 2.9859154929577465, | |
| "grad_norm": 0.34273526072502136, | |
| "learning_rate": 9.555535917993297e-10, | |
| "loss": 0.129, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 2.9888806523350633, | |
| "grad_norm": 0.3207569718360901, | |
| "learning_rate": 5.375026405352035e-10, | |
| "loss": 0.1259, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 2.9918458117123796, | |
| "grad_norm": 0.3252420723438263, | |
| "learning_rate": 2.388912514017516e-10, | |
| "loss": 0.1273, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 2.994810971089696, | |
| "grad_norm": 0.3213896155357361, | |
| "learning_rate": 5.972299119250124e-11, | |
| "loss": 0.1258, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.9977761304670127, | |
| "grad_norm": 0.3191134035587311, | |
| "learning_rate": 0.0, | |
| "loss": 0.1258, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 2.9977761304670127, | |
| "step": 1011, | |
| "total_flos": 1.1934072323532915e+19, | |
| "train_loss": 0.20480146514054692, | |
| "train_runtime": 12349.0973, | |
| "train_samples_per_second": 10.482, | |
| "train_steps_per_second": 0.082 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 1011, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 999999, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.1934072323532915e+19, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |