| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.6307561597281224, | |
| "eval_steps": 500, | |
| "global_step": 2400, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0006796941376380628, | |
| "grad_norm": 9.122925758361816, | |
| "learning_rate": 0.0, | |
| "loss": 6.6013, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0013593882752761257, | |
| "grad_norm": 9.549845695495605, | |
| "learning_rate": 6.756756756756758e-07, | |
| "loss": 6.7658, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0027187765505522514, | |
| "grad_norm": 8.209335327148438, | |
| "learning_rate": 2.0270270270270273e-06, | |
| "loss": 6.5902, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0040781648258283775, | |
| "grad_norm": 6.113947868347168, | |
| "learning_rate": 3.3783783783783788e-06, | |
| "loss": 6.5818, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.005437553101104503, | |
| "grad_norm": 6.703476428985596, | |
| "learning_rate": 4.72972972972973e-06, | |
| "loss": 6.5232, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.006796941376380629, | |
| "grad_norm": 13.405858039855957, | |
| "learning_rate": 6.081081081081082e-06, | |
| "loss": 6.5521, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.008156329651656755, | |
| "grad_norm": 11.226860046386719, | |
| "learning_rate": 7.432432432432433e-06, | |
| "loss": 6.584, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.009515717926932881, | |
| "grad_norm": 7.006751537322998, | |
| "learning_rate": 8.783783783783785e-06, | |
| "loss": 6.518, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.010875106202209005, | |
| "grad_norm": 6.206234455108643, | |
| "learning_rate": 1.0135135135135136e-05, | |
| "loss": 6.4523, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.012234494477485132, | |
| "grad_norm": 3.75468111038208, | |
| "learning_rate": 1.1486486486486488e-05, | |
| "loss": 6.4488, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.013593882752761258, | |
| "grad_norm": 3.5509755611419678, | |
| "learning_rate": 1.2837837837837838e-05, | |
| "loss": 6.3353, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.014953271028037384, | |
| "grad_norm": 2.838531494140625, | |
| "learning_rate": 1.4189189189189189e-05, | |
| "loss": 6.2737, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.01631265930331351, | |
| "grad_norm": 2.849353790283203, | |
| "learning_rate": 1.554054054054054e-05, | |
| "loss": 6.2386, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.017672047578589634, | |
| "grad_norm": 3.192340850830078, | |
| "learning_rate": 1.6891891891891892e-05, | |
| "loss": 6.1459, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.019031435853865762, | |
| "grad_norm": 3.079922914505005, | |
| "learning_rate": 1.8243243243243244e-05, | |
| "loss": 6.1783, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.020390824129141887, | |
| "grad_norm": 3.689027786254883, | |
| "learning_rate": 1.9594594594594595e-05, | |
| "loss": 5.9851, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02175021240441801, | |
| "grad_norm": 2.39050555229187, | |
| "learning_rate": 2.0945945945945947e-05, | |
| "loss": 6.0281, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.02310960067969414, | |
| "grad_norm": 2.3905773162841797, | |
| "learning_rate": 2.2297297297297298e-05, | |
| "loss": 6.0399, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.024468988954970263, | |
| "grad_norm": 2.676403045654297, | |
| "learning_rate": 2.364864864864865e-05, | |
| "loss": 6.0026, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.025828377230246388, | |
| "grad_norm": 2.220277786254883, | |
| "learning_rate": 2.5e-05, | |
| "loss": 5.9469, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.027187765505522515, | |
| "grad_norm": 3.7453274726867676, | |
| "learning_rate": 2.635135135135135e-05, | |
| "loss": 5.804, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02854715378079864, | |
| "grad_norm": 4.522032737731934, | |
| "learning_rate": 2.7702702702702704e-05, | |
| "loss": 5.8153, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.029906542056074768, | |
| "grad_norm": 3.07928204536438, | |
| "learning_rate": 2.9054054054054052e-05, | |
| "loss": 5.8357, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.031265930331350895, | |
| "grad_norm": 3.2400898933410645, | |
| "learning_rate": 3.0405405405405407e-05, | |
| "loss": 5.8705, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.03262531860662702, | |
| "grad_norm": 5.057046890258789, | |
| "learning_rate": 3.175675675675676e-05, | |
| "loss": 5.6707, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.033984706881903144, | |
| "grad_norm": 4.462399005889893, | |
| "learning_rate": 3.310810810810811e-05, | |
| "loss": 5.7012, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03534409515717927, | |
| "grad_norm": 3.095761299133301, | |
| "learning_rate": 3.445945945945946e-05, | |
| "loss": 5.6685, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.03670348343245539, | |
| "grad_norm": 3.478303909301758, | |
| "learning_rate": 3.581081081081081e-05, | |
| "loss": 5.6353, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.038062871707731524, | |
| "grad_norm": 4.6464433670043945, | |
| "learning_rate": 3.7162162162162165e-05, | |
| "loss": 5.6277, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.03942225998300765, | |
| "grad_norm": 4.2293572425842285, | |
| "learning_rate": 3.851351351351351e-05, | |
| "loss": 5.5346, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.04078164825828377, | |
| "grad_norm": 4.188422679901123, | |
| "learning_rate": 3.986486486486487e-05, | |
| "loss": 5.5544, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0421410365335599, | |
| "grad_norm": 3.0673420429229736, | |
| "learning_rate": 4.1216216216216216e-05, | |
| "loss": 5.53, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.04350042480883602, | |
| "grad_norm": 3.3032662868499756, | |
| "learning_rate": 4.256756756756757e-05, | |
| "loss": 5.5605, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.044859813084112146, | |
| "grad_norm": 3.896825075149536, | |
| "learning_rate": 4.391891891891892e-05, | |
| "loss": 5.4221, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.04621920135938828, | |
| "grad_norm": 4.151010990142822, | |
| "learning_rate": 4.5270270270270274e-05, | |
| "loss": 5.3967, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.0475785896346644, | |
| "grad_norm": 3.938117265701294, | |
| "learning_rate": 4.662162162162162e-05, | |
| "loss": 5.4716, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.048937977909940526, | |
| "grad_norm": 3.1217191219329834, | |
| "learning_rate": 4.797297297297298e-05, | |
| "loss": 5.4567, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.05029736618521665, | |
| "grad_norm": 3.293020725250244, | |
| "learning_rate": 4.9324324324324325e-05, | |
| "loss": 5.4291, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.051656754460492775, | |
| "grad_norm": 3.9366047382354736, | |
| "learning_rate": 5.067567567567568e-05, | |
| "loss": 5.378, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.053016142735768906, | |
| "grad_norm": 4.825038909912109, | |
| "learning_rate": 5.202702702702703e-05, | |
| "loss": 5.3462, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.05437553101104503, | |
| "grad_norm": 4.513136386871338, | |
| "learning_rate": 5.337837837837838e-05, | |
| "loss": 5.4209, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.055734919286321155, | |
| "grad_norm": 4.524239540100098, | |
| "learning_rate": 5.472972972972973e-05, | |
| "loss": 5.3715, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.05709430756159728, | |
| "grad_norm": 5.1905317306518555, | |
| "learning_rate": 5.6081081081081086e-05, | |
| "loss": 5.2334, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.058453695836873404, | |
| "grad_norm": 4.657945156097412, | |
| "learning_rate": 5.7432432432432434e-05, | |
| "loss": 5.2899, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.059813084112149535, | |
| "grad_norm": 3.7982685565948486, | |
| "learning_rate": 5.878378378378379e-05, | |
| "loss": 5.2191, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.06117247238742566, | |
| "grad_norm": 3.5835001468658447, | |
| "learning_rate": 6.013513513513514e-05, | |
| "loss": 5.1858, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.06253186066270179, | |
| "grad_norm": 4.594094276428223, | |
| "learning_rate": 6.14864864864865e-05, | |
| "loss": 5.2013, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.06389124893797792, | |
| "grad_norm": 3.8048019409179688, | |
| "learning_rate": 6.283783783783784e-05, | |
| "loss": 5.1493, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.06525063721325404, | |
| "grad_norm": 3.9920341968536377, | |
| "learning_rate": 6.41891891891892e-05, | |
| "loss": 5.0612, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.06661002548853016, | |
| "grad_norm": 3.4856226444244385, | |
| "learning_rate": 6.554054054054054e-05, | |
| "loss": 5.1978, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.06796941376380629, | |
| "grad_norm": 3.485684871673584, | |
| "learning_rate": 6.68918918918919e-05, | |
| "loss": 5.2438, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06932880203908241, | |
| "grad_norm": 2.92802095413208, | |
| "learning_rate": 6.824324324324325e-05, | |
| "loss": 5.0203, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.07068819031435854, | |
| "grad_norm": 3.472078561782837, | |
| "learning_rate": 6.95945945945946e-05, | |
| "loss": 5.1175, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.07204757858963466, | |
| "grad_norm": 3.5529918670654297, | |
| "learning_rate": 7.094594594594594e-05, | |
| "loss": 5.1952, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.07340696686491079, | |
| "grad_norm": 5.627261638641357, | |
| "learning_rate": 7.229729729729731e-05, | |
| "loss": 5.0469, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.07476635514018691, | |
| "grad_norm": 4.3943305015563965, | |
| "learning_rate": 7.364864864864865e-05, | |
| "loss": 5.0147, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.07612574341546305, | |
| "grad_norm": 2.405991792678833, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 5.0281, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.07748513169073917, | |
| "grad_norm": 3.361250162124634, | |
| "learning_rate": 7.635135135135135e-05, | |
| "loss": 4.9389, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.0788445199660153, | |
| "grad_norm": 3.5558111667633057, | |
| "learning_rate": 7.77027027027027e-05, | |
| "loss": 4.9327, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.08020390824129142, | |
| "grad_norm": 3.6313676834106445, | |
| "learning_rate": 7.905405405405406e-05, | |
| "loss": 4.9751, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.08156329651656755, | |
| "grad_norm": 3.766629219055176, | |
| "learning_rate": 8.040540540540541e-05, | |
| "loss": 4.9362, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.08292268479184367, | |
| "grad_norm": 3.8239798545837402, | |
| "learning_rate": 8.175675675675675e-05, | |
| "loss": 4.9906, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.0842820730671198, | |
| "grad_norm": 3.1650514602661133, | |
| "learning_rate": 8.310810810810811e-05, | |
| "loss": 4.8465, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.08564146134239592, | |
| "grad_norm": 3.0690271854400635, | |
| "learning_rate": 8.445945945945946e-05, | |
| "loss": 4.8147, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.08700084961767204, | |
| "grad_norm": 3.1290276050567627, | |
| "learning_rate": 8.581081081081082e-05, | |
| "loss": 4.8495, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.08836023789294817, | |
| "grad_norm": 3.1333677768707275, | |
| "learning_rate": 8.716216216216216e-05, | |
| "loss": 4.8112, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.08971962616822429, | |
| "grad_norm": 2.8959381580352783, | |
| "learning_rate": 8.851351351351352e-05, | |
| "loss": 4.7989, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.09107901444350043, | |
| "grad_norm": 2.715139389038086, | |
| "learning_rate": 8.986486486486487e-05, | |
| "loss": 4.7689, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.09243840271877656, | |
| "grad_norm": 2.3525729179382324, | |
| "learning_rate": 9.121621621621623e-05, | |
| "loss": 4.7503, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.09379779099405268, | |
| "grad_norm": 2.5053319931030273, | |
| "learning_rate": 9.256756756756757e-05, | |
| "loss": 4.8267, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.0951571792693288, | |
| "grad_norm": 3.2830920219421387, | |
| "learning_rate": 9.391891891891892e-05, | |
| "loss": 4.734, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.09651656754460493, | |
| "grad_norm": 3.367637872695923, | |
| "learning_rate": 9.527027027027028e-05, | |
| "loss": 4.6487, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.09787595581988105, | |
| "grad_norm": 4.157845973968506, | |
| "learning_rate": 9.662162162162163e-05, | |
| "loss": 4.7186, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.09923534409515718, | |
| "grad_norm": 3.549011707305908, | |
| "learning_rate": 9.797297297297297e-05, | |
| "loss": 4.7013, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.1005947323704333, | |
| "grad_norm": 2.438737392425537, | |
| "learning_rate": 9.932432432432433e-05, | |
| "loss": 4.6463, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.10195412064570943, | |
| "grad_norm": 2.62125301361084, | |
| "learning_rate": 9.999996843793759e-05, | |
| "loss": 4.6266, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.10331350892098555, | |
| "grad_norm": 2.5557775497436523, | |
| "learning_rate": 9.999971594167742e-05, | |
| "loss": 4.6659, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.10467289719626169, | |
| "grad_norm": 2.435065746307373, | |
| "learning_rate": 9.999921095043215e-05, | |
| "loss": 4.6833, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.10603228547153781, | |
| "grad_norm": 2.715564012527466, | |
| "learning_rate": 9.999845346675197e-05, | |
| "loss": 4.6256, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.10739167374681394, | |
| "grad_norm": 2.129850149154663, | |
| "learning_rate": 9.999744349446207e-05, | |
| "loss": 4.4834, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.10875106202209006, | |
| "grad_norm": 2.3702259063720703, | |
| "learning_rate": 9.99961810386628e-05, | |
| "loss": 4.5664, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.11011045029736619, | |
| "grad_norm": 1.9370046854019165, | |
| "learning_rate": 9.999466610572944e-05, | |
| "loss": 4.5847, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.11146983857264231, | |
| "grad_norm": 2.4077095985412598, | |
| "learning_rate": 9.999289870331232e-05, | |
| "loss": 4.6685, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.11282922684791843, | |
| "grad_norm": 3.0132172107696533, | |
| "learning_rate": 9.999087884033666e-05, | |
| "loss": 4.5605, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.11418861512319456, | |
| "grad_norm": 3.6081573963165283, | |
| "learning_rate": 9.998860652700263e-05, | |
| "loss": 4.4315, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.11554800339847068, | |
| "grad_norm": 2.687088966369629, | |
| "learning_rate": 9.998608177478525e-05, | |
| "loss": 4.5634, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.11690739167374681, | |
| "grad_norm": 2.3163015842437744, | |
| "learning_rate": 9.998330459643437e-05, | |
| "loss": 4.3725, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.11826677994902295, | |
| "grad_norm": 2.576303720474243, | |
| "learning_rate": 9.998027500597451e-05, | |
| "loss": 4.4502, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.11962616822429907, | |
| "grad_norm": 3.0173189640045166, | |
| "learning_rate": 9.997699301870488e-05, | |
| "loss": 4.4904, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.1209855564995752, | |
| "grad_norm": 1.8845309019088745, | |
| "learning_rate": 9.99734586511993e-05, | |
| "loss": 4.4285, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.12234494477485132, | |
| "grad_norm": 1.8597114086151123, | |
| "learning_rate": 9.996967192130606e-05, | |
| "loss": 4.4114, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.12370433305012744, | |
| "grad_norm": 1.9403643608093262, | |
| "learning_rate": 9.996563284814788e-05, | |
| "loss": 4.3586, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.12506372132540358, | |
| "grad_norm": 2.1628377437591553, | |
| "learning_rate": 9.99613414521218e-05, | |
| "loss": 4.4004, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.1264231096006797, | |
| "grad_norm": 2.213683843612671, | |
| "learning_rate": 9.995679775489906e-05, | |
| "loss": 4.4017, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.12778249787595583, | |
| "grad_norm": 1.9236798286437988, | |
| "learning_rate": 9.995200177942499e-05, | |
| "loss": 4.3356, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.12914188615123195, | |
| "grad_norm": 2.8310718536376953, | |
| "learning_rate": 9.994695354991892e-05, | |
| "loss": 4.2476, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.13050127442650808, | |
| "grad_norm": 2.613215446472168, | |
| "learning_rate": 9.994165309187406e-05, | |
| "loss": 4.4249, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.1318606627017842, | |
| "grad_norm": 3.2933475971221924, | |
| "learning_rate": 9.993610043205735e-05, | |
| "loss": 4.359, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.13322005097706033, | |
| "grad_norm": 2.660553455352783, | |
| "learning_rate": 9.993029559850932e-05, | |
| "loss": 4.3591, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.13457943925233645, | |
| "grad_norm": 2.223825693130493, | |
| "learning_rate": 9.992423862054397e-05, | |
| "loss": 4.2638, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.13593882752761258, | |
| "grad_norm": 1.6391338109970093, | |
| "learning_rate": 9.991792952874857e-05, | |
| "loss": 4.2506, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1372982158028887, | |
| "grad_norm": 1.568050742149353, | |
| "learning_rate": 9.991136835498363e-05, | |
| "loss": 4.1789, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.13865760407816483, | |
| "grad_norm": 1.8366698026657104, | |
| "learning_rate": 9.990455513238257e-05, | |
| "loss": 4.2361, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.14001699235344095, | |
| "grad_norm": 2.0478951930999756, | |
| "learning_rate": 9.98974898953517e-05, | |
| "loss": 4.2613, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.14137638062871707, | |
| "grad_norm": 1.7681331634521484, | |
| "learning_rate": 9.989017267956994e-05, | |
| "loss": 4.2437, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.1427357689039932, | |
| "grad_norm": 2.2257468700408936, | |
| "learning_rate": 9.988260352198872e-05, | |
| "loss": 4.1724, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.14409515717926932, | |
| "grad_norm": 1.6590179204940796, | |
| "learning_rate": 9.987478246083175e-05, | |
| "loss": 4.1619, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.14545454545454545, | |
| "grad_norm": 2.029710292816162, | |
| "learning_rate": 9.986670953559482e-05, | |
| "loss": 4.2611, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.14681393372982157, | |
| "grad_norm": 1.7355066537857056, | |
| "learning_rate": 9.985838478704563e-05, | |
| "loss": 4.222, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.1481733220050977, | |
| "grad_norm": 1.91265869140625, | |
| "learning_rate": 9.984980825722356e-05, | |
| "loss": 4.0887, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.14953271028037382, | |
| "grad_norm": 2.1522412300109863, | |
| "learning_rate": 9.984097998943947e-05, | |
| "loss": 4.1331, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.15089209855564995, | |
| "grad_norm": 1.7838095426559448, | |
| "learning_rate": 9.983190002827546e-05, | |
| "loss": 4.0928, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.1522514868309261, | |
| "grad_norm": 1.8782153129577637, | |
| "learning_rate": 9.982256841958472e-05, | |
| "loss": 4.2071, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.15361087510620222, | |
| "grad_norm": 2.179396390914917, | |
| "learning_rate": 9.981298521049118e-05, | |
| "loss": 4.0642, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.15497026338147835, | |
| "grad_norm": 2.1441640853881836, | |
| "learning_rate": 9.980315044938939e-05, | |
| "loss": 4.0892, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.15632965165675447, | |
| "grad_norm": 2.6898701190948486, | |
| "learning_rate": 9.979306418594417e-05, | |
| "loss": 4.1155, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.1576890399320306, | |
| "grad_norm": 2.3028266429901123, | |
| "learning_rate": 9.97827264710904e-05, | |
| "loss": 4.1381, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.15904842820730672, | |
| "grad_norm": 1.8704326152801514, | |
| "learning_rate": 9.977213735703283e-05, | |
| "loss": 4.1299, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.16040781648258284, | |
| "grad_norm": 1.5334903001785278, | |
| "learning_rate": 9.976129689724574e-05, | |
| "loss": 4.1585, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.16176720475785897, | |
| "grad_norm": 1.5391136407852173, | |
| "learning_rate": 9.975020514647267e-05, | |
| "loss": 4.0774, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.1631265930331351, | |
| "grad_norm": 1.731969952583313, | |
| "learning_rate": 9.973886216072614e-05, | |
| "loss": 4.1801, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.16448598130841122, | |
| "grad_norm": 1.355950117111206, | |
| "learning_rate": 9.972726799728744e-05, | |
| "loss": 4.1208, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.16584536958368734, | |
| "grad_norm": 1.6355708837509155, | |
| "learning_rate": 9.971542271470625e-05, | |
| "loss": 4.0135, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.16720475785896347, | |
| "grad_norm": 1.612067461013794, | |
| "learning_rate": 9.970332637280041e-05, | |
| "loss": 4.008, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.1685641461342396, | |
| "grad_norm": 1.5609122514724731, | |
| "learning_rate": 9.969097903265558e-05, | |
| "loss": 3.9615, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.16992353440951571, | |
| "grad_norm": 2.1877589225769043, | |
| "learning_rate": 9.967838075662495e-05, | |
| "loss": 4.0187, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.17128292268479184, | |
| "grad_norm": 2.0836243629455566, | |
| "learning_rate": 9.966553160832889e-05, | |
| "loss": 4.0108, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.17264231096006796, | |
| "grad_norm": 1.8262373208999634, | |
| "learning_rate": 9.96524316526547e-05, | |
| "loss": 3.9729, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.1740016992353441, | |
| "grad_norm": 1.4357279539108276, | |
| "learning_rate": 9.96390809557562e-05, | |
| "loss": 3.9418, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.1753610875106202, | |
| "grad_norm": 1.4747521877288818, | |
| "learning_rate": 9.962547958505346e-05, | |
| "loss": 4.0073, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.17672047578589634, | |
| "grad_norm": 1.5109456777572632, | |
| "learning_rate": 9.961162760923244e-05, | |
| "loss": 4.0114, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.17807986406117246, | |
| "grad_norm": 1.6962803602218628, | |
| "learning_rate": 9.959752509824462e-05, | |
| "loss": 3.8997, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.17943925233644858, | |
| "grad_norm": 1.2874037027359009, | |
| "learning_rate": 9.958317212330665e-05, | |
| "loss": 3.9746, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.18079864061172474, | |
| "grad_norm": 1.4089356660842896, | |
| "learning_rate": 9.956856875690006e-05, | |
| "loss": 3.8799, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.18215802888700086, | |
| "grad_norm": 1.4761899709701538, | |
| "learning_rate": 9.95537150727708e-05, | |
| "loss": 3.9084, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.18351741716227699, | |
| "grad_norm": 1.2963216304779053, | |
| "learning_rate": 9.953861114592889e-05, | |
| "loss": 3.884, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.1848768054375531, | |
| "grad_norm": 1.2376818656921387, | |
| "learning_rate": 9.952325705264806e-05, | |
| "loss": 3.9434, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.18623619371282923, | |
| "grad_norm": 1.6393024921417236, | |
| "learning_rate": 9.950765287046543e-05, | |
| "loss": 3.9175, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.18759558198810536, | |
| "grad_norm": 1.2873233556747437, | |
| "learning_rate": 9.949179867818099e-05, | |
| "loss": 3.9513, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.18895497026338148, | |
| "grad_norm": 1.3314156532287598, | |
| "learning_rate": 9.947569455585726e-05, | |
| "loss": 3.9345, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.1903143585386576, | |
| "grad_norm": 1.392342448234558, | |
| "learning_rate": 9.945934058481892e-05, | |
| "loss": 3.8092, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.19167374681393373, | |
| "grad_norm": 1.4349101781845093, | |
| "learning_rate": 9.944273684765235e-05, | |
| "loss": 3.8548, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.19303313508920986, | |
| "grad_norm": 1.2190157175064087, | |
| "learning_rate": 9.942588342820521e-05, | |
| "loss": 3.9121, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.19439252336448598, | |
| "grad_norm": 1.4537711143493652, | |
| "learning_rate": 9.94087804115861e-05, | |
| "loss": 3.8502, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.1957519116397621, | |
| "grad_norm": 1.6733758449554443, | |
| "learning_rate": 9.939142788416398e-05, | |
| "loss": 3.8743, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.19711129991503823, | |
| "grad_norm": 1.4261025190353394, | |
| "learning_rate": 9.937382593356793e-05, | |
| "loss": 3.8947, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.19847068819031435, | |
| "grad_norm": 1.6536645889282227, | |
| "learning_rate": 9.93559746486865e-05, | |
| "loss": 3.9158, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.19983007646559048, | |
| "grad_norm": 1.71151864528656, | |
| "learning_rate": 9.933787411966742e-05, | |
| "loss": 3.8466, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.2011894647408666, | |
| "grad_norm": 1.8195589780807495, | |
| "learning_rate": 9.931952443791703e-05, | |
| "loss": 3.8113, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.20254885301614273, | |
| "grad_norm": 1.5555843114852905, | |
| "learning_rate": 9.930092569609996e-05, | |
| "loss": 3.8505, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.20390824129141885, | |
| "grad_norm": 1.402797818183899, | |
| "learning_rate": 9.928207798813849e-05, | |
| "loss": 3.8856, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.20526762956669498, | |
| "grad_norm": 1.33147394657135, | |
| "learning_rate": 9.926298140921221e-05, | |
| "loss": 3.8581, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.2066270178419711, | |
| "grad_norm": 1.1469197273254395, | |
| "learning_rate": 9.924363605575746e-05, | |
| "loss": 3.8449, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.20798640611724725, | |
| "grad_norm": 1.3006025552749634, | |
| "learning_rate": 9.922404202546691e-05, | |
| "loss": 3.8268, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.20934579439252338, | |
| "grad_norm": 1.4287155866622925, | |
| "learning_rate": 9.9204199417289e-05, | |
| "loss": 3.7724, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.2107051826677995, | |
| "grad_norm": 1.5455858707427979, | |
| "learning_rate": 9.918410833142748e-05, | |
| "loss": 3.7289, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.21206457094307563, | |
| "grad_norm": 2.112565517425537, | |
| "learning_rate": 9.91637688693409e-05, | |
| "loss": 3.7394, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.21342395921835175, | |
| "grad_norm": 1.6835887432098389, | |
| "learning_rate": 9.914318113374208e-05, | |
| "loss": 3.7487, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.21478334749362787, | |
| "grad_norm": 1.4296996593475342, | |
| "learning_rate": 9.912234522859761e-05, | |
| "loss": 3.7731, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.216142735768904, | |
| "grad_norm": 1.7142002582550049, | |
| "learning_rate": 9.910126125912733e-05, | |
| "loss": 3.8279, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.21750212404418012, | |
| "grad_norm": 1.5620222091674805, | |
| "learning_rate": 9.907992933180376e-05, | |
| "loss": 3.8279, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.21886151231945625, | |
| "grad_norm": 1.3424922227859497, | |
| "learning_rate": 9.905834955435162e-05, | |
| "loss": 3.7903, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.22022090059473237, | |
| "grad_norm": 1.4565094709396362, | |
| "learning_rate": 9.903652203574722e-05, | |
| "loss": 3.793, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.2215802888700085, | |
| "grad_norm": 1.7999119758605957, | |
| "learning_rate": 9.901444688621801e-05, | |
| "loss": 3.7354, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.22293967714528462, | |
| "grad_norm": 1.4900187253952026, | |
| "learning_rate": 9.899212421724187e-05, | |
| "loss": 3.7323, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.22429906542056074, | |
| "grad_norm": 1.4624853134155273, | |
| "learning_rate": 9.896955414154669e-05, | |
| "loss": 3.7207, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.22565845369583687, | |
| "grad_norm": 1.7633172273635864, | |
| "learning_rate": 9.894673677310972e-05, | |
| "loss": 3.7566, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.227017841971113, | |
| "grad_norm": 1.172234296798706, | |
| "learning_rate": 9.892367222715709e-05, | |
| "loss": 3.7376, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.22837723024638912, | |
| "grad_norm": 1.537023901939392, | |
| "learning_rate": 9.890036062016306e-05, | |
| "loss": 3.7157, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.22973661852166524, | |
| "grad_norm": 1.3012125492095947, | |
| "learning_rate": 9.887680206984959e-05, | |
| "loss": 3.6776, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.23109600679694137, | |
| "grad_norm": 1.1854647397994995, | |
| "learning_rate": 9.885299669518569e-05, | |
| "loss": 3.6635, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.2324553950722175, | |
| "grad_norm": 1.1112992763519287, | |
| "learning_rate": 9.882894461638676e-05, | |
| "loss": 3.7341, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.23381478334749362, | |
| "grad_norm": 1.1130858659744263, | |
| "learning_rate": 9.88046459549141e-05, | |
| "loss": 3.6422, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.23517417162276974, | |
| "grad_norm": 1.441116213798523, | |
| "learning_rate": 9.878010083347419e-05, | |
| "loss": 3.6886, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.2365335598980459, | |
| "grad_norm": 1.3626590967178345, | |
| "learning_rate": 9.875530937601816e-05, | |
| "loss": 3.7735, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.23789294817332202, | |
| "grad_norm": 1.2444162368774414, | |
| "learning_rate": 9.873027170774109e-05, | |
| "loss": 3.7312, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.23925233644859814, | |
| "grad_norm": 1.3234375715255737, | |
| "learning_rate": 9.87049879550814e-05, | |
| "loss": 3.748, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.24061172472387427, | |
| "grad_norm": 1.333979606628418, | |
| "learning_rate": 9.867945824572024e-05, | |
| "loss": 3.6207, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.2419711129991504, | |
| "grad_norm": 1.0207340717315674, | |
| "learning_rate": 9.865368270858082e-05, | |
| "loss": 3.7018, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.24333050127442651, | |
| "grad_norm": 1.098137378692627, | |
| "learning_rate": 9.862766147382774e-05, | |
| "loss": 3.6689, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.24468988954970264, | |
| "grad_norm": 1.1118202209472656, | |
| "learning_rate": 9.860139467286638e-05, | |
| "loss": 3.7185, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.24604927782497876, | |
| "grad_norm": 1.4026211500167847, | |
| "learning_rate": 9.857488243834219e-05, | |
| "loss": 3.6949, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.2474086661002549, | |
| "grad_norm": 1.528132677078247, | |
| "learning_rate": 9.85481249041401e-05, | |
| "loss": 3.5872, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.248768054375531, | |
| "grad_norm": 1.4865642786026, | |
| "learning_rate": 9.852112220538367e-05, | |
| "loss": 3.6044, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.25012744265080716, | |
| "grad_norm": 1.1037031412124634, | |
| "learning_rate": 9.849387447843467e-05, | |
| "loss": 3.7614, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.25148683092608326, | |
| "grad_norm": 1.0702588558197021, | |
| "learning_rate": 9.846638186089214e-05, | |
| "loss": 3.6226, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.2528462192013594, | |
| "grad_norm": 0.970947802066803, | |
| "learning_rate": 9.843864449159182e-05, | |
| "loss": 3.6127, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.2542056074766355, | |
| "grad_norm": 1.1656701564788818, | |
| "learning_rate": 9.841066251060543e-05, | |
| "loss": 3.694, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.25556499575191166, | |
| "grad_norm": 1.2996894121170044, | |
| "learning_rate": 9.838243605924001e-05, | |
| "loss": 3.6226, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.25692438402718776, | |
| "grad_norm": 1.392196536064148, | |
| "learning_rate": 9.835396528003707e-05, | |
| "loss": 3.6542, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.2582837723024639, | |
| "grad_norm": 1.3840879201889038, | |
| "learning_rate": 9.832525031677205e-05, | |
| "loss": 3.6416, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.25964316057774, | |
| "grad_norm": 1.5829066038131714, | |
| "learning_rate": 9.829629131445342e-05, | |
| "loss": 3.6383, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.26100254885301616, | |
| "grad_norm": 1.260533332824707, | |
| "learning_rate": 9.826708841932209e-05, | |
| "loss": 3.6034, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.26236193712829226, | |
| "grad_norm": 1.3130146265029907, | |
| "learning_rate": 9.823764177885059e-05, | |
| "loss": 3.5935, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.2637213254035684, | |
| "grad_norm": 1.4189637899398804, | |
| "learning_rate": 9.820795154174235e-05, | |
| "loss": 3.6792, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.2650807136788445, | |
| "grad_norm": 1.3872414827346802, | |
| "learning_rate": 9.817801785793092e-05, | |
| "loss": 3.6149, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.26644010195412066, | |
| "grad_norm": 1.487898826599121, | |
| "learning_rate": 9.814784087857927e-05, | |
| "loss": 3.6161, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.26779949022939675, | |
| "grad_norm": 1.374002456665039, | |
| "learning_rate": 9.8117420756079e-05, | |
| "loss": 3.6394, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.2691588785046729, | |
| "grad_norm": 0.9476630091667175, | |
| "learning_rate": 9.808675764404953e-05, | |
| "loss": 3.5447, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.270518266779949, | |
| "grad_norm": 1.2160744667053223, | |
| "learning_rate": 9.805585169733738e-05, | |
| "loss": 3.552, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.27187765505522515, | |
| "grad_norm": 1.2432382106781006, | |
| "learning_rate": 9.802470307201538e-05, | |
| "loss": 3.5518, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.27323704333050125, | |
| "grad_norm": 1.0426836013793945, | |
| "learning_rate": 9.799331192538185e-05, | |
| "loss": 3.6109, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.2745964316057774, | |
| "grad_norm": 1.1359163522720337, | |
| "learning_rate": 9.796167841595986e-05, | |
| "loss": 3.563, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.2759558198810535, | |
| "grad_norm": 1.2553869485855103, | |
| "learning_rate": 9.792980270349633e-05, | |
| "loss": 3.6323, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.27731520815632965, | |
| "grad_norm": 1.0180846452713013, | |
| "learning_rate": 9.789768494896132e-05, | |
| "loss": 3.4775, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.2786745964316058, | |
| "grad_norm": 1.2593415975570679, | |
| "learning_rate": 9.786532531454722e-05, | |
| "loss": 3.6039, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.2800339847068819, | |
| "grad_norm": 1.2103174924850464, | |
| "learning_rate": 9.783272396366784e-05, | |
| "loss": 3.6399, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.28139337298215805, | |
| "grad_norm": 1.3131142854690552, | |
| "learning_rate": 9.77998810609577e-05, | |
| "loss": 3.6405, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.28275276125743415, | |
| "grad_norm": 1.4719713926315308, | |
| "learning_rate": 9.77667967722711e-05, | |
| "loss": 3.6138, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.2841121495327103, | |
| "grad_norm": 1.4857118129730225, | |
| "learning_rate": 9.773347126468128e-05, | |
| "loss": 3.5811, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.2854715378079864, | |
| "grad_norm": 1.424742579460144, | |
| "learning_rate": 9.769990470647974e-05, | |
| "loss": 3.5766, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.28683092608326255, | |
| "grad_norm": 1.399685263633728, | |
| "learning_rate": 9.766609726717515e-05, | |
| "loss": 3.5816, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.28819031435853865, | |
| "grad_norm": 1.0692592859268188, | |
| "learning_rate": 9.763204911749267e-05, | |
| "loss": 3.5316, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.2895497026338148, | |
| "grad_norm": 0.9437915682792664, | |
| "learning_rate": 9.759776042937302e-05, | |
| "loss": 3.5464, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.2909090909090909, | |
| "grad_norm": 1.3703209161758423, | |
| "learning_rate": 9.756323137597159e-05, | |
| "loss": 3.5578, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.29226847918436705, | |
| "grad_norm": 1.31071138381958, | |
| "learning_rate": 9.752846213165767e-05, | |
| "loss": 3.6392, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.29362786745964314, | |
| "grad_norm": 0.9742053747177124, | |
| "learning_rate": 9.749345287201343e-05, | |
| "loss": 3.5328, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.2949872557349193, | |
| "grad_norm": 0.9459298253059387, | |
| "learning_rate": 9.745820377383314e-05, | |
| "loss": 3.4811, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.2963466440101954, | |
| "grad_norm": 1.5522956848144531, | |
| "learning_rate": 9.74227150151222e-05, | |
| "loss": 3.5842, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.29770603228547154, | |
| "grad_norm": 1.2024612426757812, | |
| "learning_rate": 9.738698677509632e-05, | |
| "loss": 3.5591, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.29906542056074764, | |
| "grad_norm": 1.2312219142913818, | |
| "learning_rate": 9.735101923418054e-05, | |
| "loss": 3.5347, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3004248088360238, | |
| "grad_norm": 1.215908169746399, | |
| "learning_rate": 9.731481257400838e-05, | |
| "loss": 3.5233, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.3017841971112999, | |
| "grad_norm": 1.025688886642456, | |
| "learning_rate": 9.727836697742086e-05, | |
| "loss": 3.4792, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.30314358538657604, | |
| "grad_norm": 0.8059235215187073, | |
| "learning_rate": 9.724168262846566e-05, | |
| "loss": 3.4884, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.3045029736618522, | |
| "grad_norm": 1.061319351196289, | |
| "learning_rate": 9.720475971239609e-05, | |
| "loss": 3.5429, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.3058623619371283, | |
| "grad_norm": 1.2992972135543823, | |
| "learning_rate": 9.716759841567025e-05, | |
| "loss": 3.4973, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.30722175021240444, | |
| "grad_norm": 1.206768274307251, | |
| "learning_rate": 9.713019892595003e-05, | |
| "loss": 3.5063, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.30858113848768054, | |
| "grad_norm": 0.9437035918235779, | |
| "learning_rate": 9.709256143210015e-05, | |
| "loss": 3.4601, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.3099405267629567, | |
| "grad_norm": 1.4230656623840332, | |
| "learning_rate": 9.705468612418727e-05, | |
| "loss": 3.5248, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.3112999150382328, | |
| "grad_norm": 1.3039008378982544, | |
| "learning_rate": 9.701657319347902e-05, | |
| "loss": 3.3868, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.31265930331350894, | |
| "grad_norm": 1.0053726434707642, | |
| "learning_rate": 9.69782228324429e-05, | |
| "loss": 3.4689, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.31401869158878504, | |
| "grad_norm": 1.0519505739212036, | |
| "learning_rate": 9.693963523474554e-05, | |
| "loss": 3.5104, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.3153780798640612, | |
| "grad_norm": 1.2941850423812866, | |
| "learning_rate": 9.690081059525154e-05, | |
| "loss": 3.4889, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.3167374681393373, | |
| "grad_norm": 1.2811554670333862, | |
| "learning_rate": 9.686174911002253e-05, | |
| "loss": 3.5412, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.31809685641461344, | |
| "grad_norm": 1.0608913898468018, | |
| "learning_rate": 9.682245097631622e-05, | |
| "loss": 3.5063, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.31945624468988953, | |
| "grad_norm": 0.7853614687919617, | |
| "learning_rate": 9.678291639258537e-05, | |
| "loss": 3.5436, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.3208156329651657, | |
| "grad_norm": 1.3205409049987793, | |
| "learning_rate": 9.674314555847682e-05, | |
| "loss": 3.5236, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.3221750212404418, | |
| "grad_norm": 1.1064730882644653, | |
| "learning_rate": 9.670313867483041e-05, | |
| "loss": 3.4756, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.32353440951571794, | |
| "grad_norm": 0.9121582508087158, | |
| "learning_rate": 9.666289594367803e-05, | |
| "loss": 3.5151, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.32489379779099403, | |
| "grad_norm": 0.9929459691047668, | |
| "learning_rate": 9.662241756824261e-05, | |
| "loss": 3.4621, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.3262531860662702, | |
| "grad_norm": 1.1451283693313599, | |
| "learning_rate": 9.658170375293703e-05, | |
| "loss": 3.4464, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.3276125743415463, | |
| "grad_norm": 1.161496639251709, | |
| "learning_rate": 9.654075470336317e-05, | |
| "loss": 3.4189, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.32897196261682243, | |
| "grad_norm": 1.0276612043380737, | |
| "learning_rate": 9.649957062631078e-05, | |
| "loss": 3.4705, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.33033135089209853, | |
| "grad_norm": 1.061078667640686, | |
| "learning_rate": 9.645815172975649e-05, | |
| "loss": 3.4573, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.3316907391673747, | |
| "grad_norm": 1.14658522605896, | |
| "learning_rate": 9.641649822286278e-05, | |
| "loss": 3.4265, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.33305012744265083, | |
| "grad_norm": 1.1178537607192993, | |
| "learning_rate": 9.637461031597686e-05, | |
| "loss": 3.4919, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.33440951571792693, | |
| "grad_norm": 1.1290613412857056, | |
| "learning_rate": 9.633248822062968e-05, | |
| "loss": 3.5057, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.3357689039932031, | |
| "grad_norm": 0.9476689696311951, | |
| "learning_rate": 9.629013214953478e-05, | |
| "loss": 3.4467, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.3371282922684792, | |
| "grad_norm": 1.0658568143844604, | |
| "learning_rate": 9.624754231658731e-05, | |
| "loss": 3.4675, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.33848768054375533, | |
| "grad_norm": 0.9237107634544373, | |
| "learning_rate": 9.620471893686287e-05, | |
| "loss": 3.5161, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.33984706881903143, | |
| "grad_norm": 1.1166672706604004, | |
| "learning_rate": 9.616166222661646e-05, | |
| "loss": 3.4474, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3412064570943076, | |
| "grad_norm": 1.3173065185546875, | |
| "learning_rate": 9.611837240328138e-05, | |
| "loss": 3.4247, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.3425658453695837, | |
| "grad_norm": 0.8932580947875977, | |
| "learning_rate": 9.607484968546813e-05, | |
| "loss": 3.425, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.34392523364485983, | |
| "grad_norm": 1.3925460577011108, | |
| "learning_rate": 9.603109429296333e-05, | |
| "loss": 3.4579, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.3452846219201359, | |
| "grad_norm": 1.265743374824524, | |
| "learning_rate": 9.598710644672859e-05, | |
| "loss": 3.584, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.3466440101954121, | |
| "grad_norm": 1.1315666437149048, | |
| "learning_rate": 9.594288636889936e-05, | |
| "loss": 3.3599, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.3480033984706882, | |
| "grad_norm": 1.1435672044754028, | |
| "learning_rate": 9.589843428278388e-05, | |
| "loss": 3.4324, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.3493627867459643, | |
| "grad_norm": 1.0228195190429688, | |
| "learning_rate": 9.5853750412862e-05, | |
| "loss": 3.3526, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.3507221750212404, | |
| "grad_norm": 0.9447354674339294, | |
| "learning_rate": 9.580883498478406e-05, | |
| "loss": 3.4243, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.3520815632965166, | |
| "grad_norm": 0.9552397131919861, | |
| "learning_rate": 9.576368822536976e-05, | |
| "loss": 3.3066, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.35344095157179267, | |
| "grad_norm": 1.1417309045791626, | |
| "learning_rate": 9.571831036260699e-05, | |
| "loss": 3.3925, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.3548003398470688, | |
| "grad_norm": 1.0355608463287354, | |
| "learning_rate": 9.567270162565073e-05, | |
| "loss": 3.4224, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.3561597281223449, | |
| "grad_norm": 0.9213873744010925, | |
| "learning_rate": 9.562686224482182e-05, | |
| "loss": 3.4091, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.3575191163976211, | |
| "grad_norm": 0.97687828540802, | |
| "learning_rate": 9.558079245160584e-05, | |
| "loss": 3.3404, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.35887850467289717, | |
| "grad_norm": 1.0606272220611572, | |
| "learning_rate": 9.553449247865199e-05, | |
| "loss": 3.4489, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.3602378929481733, | |
| "grad_norm": 1.0698935985565186, | |
| "learning_rate": 9.548796255977175e-05, | |
| "loss": 3.4235, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.3615972812234495, | |
| "grad_norm": 1.0898542404174805, | |
| "learning_rate": 9.544120292993795e-05, | |
| "loss": 3.4447, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.36295666949872557, | |
| "grad_norm": 1.1248620748519897, | |
| "learning_rate": 9.539421382528331e-05, | |
| "loss": 3.4496, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.3643160577740017, | |
| "grad_norm": 1.1177901029586792, | |
| "learning_rate": 9.534699548309948e-05, | |
| "loss": 3.3291, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.3656754460492778, | |
| "grad_norm": 0.8826277256011963, | |
| "learning_rate": 9.529954814183572e-05, | |
| "loss": 3.4179, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.36703483432455397, | |
| "grad_norm": 0.8299089074134827, | |
| "learning_rate": 9.525187204109767e-05, | |
| "loss": 3.2932, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.36839422259983007, | |
| "grad_norm": 1.1234545707702637, | |
| "learning_rate": 9.520396742164624e-05, | |
| "loss": 3.3519, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.3697536108751062, | |
| "grad_norm": 0.8834955096244812, | |
| "learning_rate": 9.515583452539633e-05, | |
| "loss": 3.3694, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.3711129991503823, | |
| "grad_norm": 1.1295942068099976, | |
| "learning_rate": 9.510747359541562e-05, | |
| "loss": 3.4234, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.37247238742565847, | |
| "grad_norm": 1.1070072650909424, | |
| "learning_rate": 9.505888487592333e-05, | |
| "loss": 3.2804, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.37383177570093457, | |
| "grad_norm": 0.8992867469787598, | |
| "learning_rate": 9.501006861228903e-05, | |
| "loss": 3.3935, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.3751911639762107, | |
| "grad_norm": 0.9362753033638, | |
| "learning_rate": 9.496102505103135e-05, | |
| "loss": 3.4367, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.3765505522514868, | |
| "grad_norm": 0.997207760810852, | |
| "learning_rate": 9.491175443981677e-05, | |
| "loss": 3.3499, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.37790994052676297, | |
| "grad_norm": 0.9916525483131409, | |
| "learning_rate": 9.486225702745833e-05, | |
| "loss": 3.3746, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.37926932880203906, | |
| "grad_norm": 0.9965975284576416, | |
| "learning_rate": 9.481253306391445e-05, | |
| "loss": 3.2899, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.3806287170773152, | |
| "grad_norm": 0.9485024213790894, | |
| "learning_rate": 9.476258280028753e-05, | |
| "loss": 3.2999, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.3819881053525913, | |
| "grad_norm": 0.9940890669822693, | |
| "learning_rate": 9.471240648882288e-05, | |
| "loss": 3.2989, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.38334749362786746, | |
| "grad_norm": 0.8807222843170166, | |
| "learning_rate": 9.466200438290724e-05, | |
| "loss": 3.3442, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.38470688190314356, | |
| "grad_norm": 0.923343300819397, | |
| "learning_rate": 9.461137673706768e-05, | |
| "loss": 3.3407, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.3860662701784197, | |
| "grad_norm": 1.0074143409729004, | |
| "learning_rate": 9.456052380697015e-05, | |
| "loss": 3.3164, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.3874256584536958, | |
| "grad_norm": 0.9250668883323669, | |
| "learning_rate": 9.450944584941831e-05, | |
| "loss": 3.3366, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.38878504672897196, | |
| "grad_norm": 0.9320377707481384, | |
| "learning_rate": 9.44581431223522e-05, | |
| "loss": 3.2337, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.3901444350042481, | |
| "grad_norm": 0.8707028031349182, | |
| "learning_rate": 9.440661588484691e-05, | |
| "loss": 3.3565, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.3915038232795242, | |
| "grad_norm": 0.9949326515197754, | |
| "learning_rate": 9.43548643971113e-05, | |
| "loss": 3.3029, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.39286321155480036, | |
| "grad_norm": 0.7374237775802612, | |
| "learning_rate": 9.430288892048666e-05, | |
| "loss": 3.3016, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.39422259983007646, | |
| "grad_norm": 1.2260855436325073, | |
| "learning_rate": 9.425068971744547e-05, | |
| "loss": 3.4076, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.3955819881053526, | |
| "grad_norm": 1.042262315750122, | |
| "learning_rate": 9.419826705158994e-05, | |
| "loss": 3.4558, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.3969413763806287, | |
| "grad_norm": 0.9673371911048889, | |
| "learning_rate": 9.414562118765077e-05, | |
| "loss": 3.3727, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.39830076465590486, | |
| "grad_norm": 0.8633317947387695, | |
| "learning_rate": 9.40927523914858e-05, | |
| "loss": 3.3029, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.39966015293118096, | |
| "grad_norm": 1.0391029119491577, | |
| "learning_rate": 9.40396609300787e-05, | |
| "loss": 3.3455, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.4010195412064571, | |
| "grad_norm": 0.9045621752738953, | |
| "learning_rate": 9.398634707153752e-05, | |
| "loss": 3.365, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.4023789294817332, | |
| "grad_norm": 1.0436588525772095, | |
| "learning_rate": 9.393281108509342e-05, | |
| "loss": 3.3427, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.40373831775700936, | |
| "grad_norm": 0.9874547719955444, | |
| "learning_rate": 9.387905324109934e-05, | |
| "loss": 3.2496, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.40509770603228545, | |
| "grad_norm": 0.7234767079353333, | |
| "learning_rate": 9.382507381102849e-05, | |
| "loss": 3.2806, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.4064570943075616, | |
| "grad_norm": 0.7901805639266968, | |
| "learning_rate": 9.377087306747315e-05, | |
| "loss": 3.2479, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.4078164825828377, | |
| "grad_norm": 0.8309746980667114, | |
| "learning_rate": 9.37164512841432e-05, | |
| "loss": 3.2532, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.40917587085811385, | |
| "grad_norm": 0.7645075917243958, | |
| "learning_rate": 9.366180873586475e-05, | |
| "loss": 3.3121, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.41053525913338995, | |
| "grad_norm": 0.8399624824523926, | |
| "learning_rate": 9.360694569857873e-05, | |
| "loss": 3.2899, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.4118946474086661, | |
| "grad_norm": 0.7422559857368469, | |
| "learning_rate": 9.355186244933959e-05, | |
| "loss": 3.3512, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.4132540356839422, | |
| "grad_norm": 0.8472399115562439, | |
| "learning_rate": 9.349655926631375e-05, | |
| "loss": 3.3467, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.41461342395921835, | |
| "grad_norm": 0.745278000831604, | |
| "learning_rate": 9.344103642877837e-05, | |
| "loss": 3.2806, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.4159728122344945, | |
| "grad_norm": 0.9684063196182251, | |
| "learning_rate": 9.338529421711977e-05, | |
| "loss": 3.2831, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.4173322005097706, | |
| "grad_norm": 0.8448832631111145, | |
| "learning_rate": 9.332933291283215e-05, | |
| "loss": 3.2381, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.41869158878504675, | |
| "grad_norm": 0.9104022979736328, | |
| "learning_rate": 9.327315279851605e-05, | |
| "loss": 3.3765, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.42005097706032285, | |
| "grad_norm": 0.9372128248214722, | |
| "learning_rate": 9.321675415787707e-05, | |
| "loss": 3.3079, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.421410365335599, | |
| "grad_norm": 0.8389849066734314, | |
| "learning_rate": 9.316013727572429e-05, | |
| "loss": 3.3161, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.4227697536108751, | |
| "grad_norm": 0.982002854347229, | |
| "learning_rate": 9.31033024379689e-05, | |
| "loss": 3.2549, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.42412914188615125, | |
| "grad_norm": 0.9176488518714905, | |
| "learning_rate": 9.304624993162276e-05, | |
| "loss": 3.2436, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.42548853016142735, | |
| "grad_norm": 0.8668674826622009, | |
| "learning_rate": 9.298898004479697e-05, | |
| "loss": 3.2616, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.4268479184367035, | |
| "grad_norm": 0.9916755557060242, | |
| "learning_rate": 9.293149306670032e-05, | |
| "loss": 3.2799, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.4282073067119796, | |
| "grad_norm": 0.843906819820404, | |
| "learning_rate": 9.287378928763798e-05, | |
| "loss": 3.2975, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.42956669498725575, | |
| "grad_norm": 0.9572488069534302, | |
| "learning_rate": 9.281586899900985e-05, | |
| "loss": 3.2525, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.43092608326253184, | |
| "grad_norm": 0.9162222743034363, | |
| "learning_rate": 9.275773249330927e-05, | |
| "loss": 3.3031, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.432285471537808, | |
| "grad_norm": 1.137613296508789, | |
| "learning_rate": 9.269938006412142e-05, | |
| "loss": 3.2803, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.4336448598130841, | |
| "grad_norm": 0.9166907072067261, | |
| "learning_rate": 9.26408120061219e-05, | |
| "loss": 3.3167, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.43500424808836025, | |
| "grad_norm": 0.8309205174446106, | |
| "learning_rate": 9.258202861507518e-05, | |
| "loss": 3.2414, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.43636363636363634, | |
| "grad_norm": 0.9707709550857544, | |
| "learning_rate": 9.252303018783324e-05, | |
| "loss": 3.3043, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.4377230246389125, | |
| "grad_norm": 0.7566142678260803, | |
| "learning_rate": 9.246381702233385e-05, | |
| "loss": 3.2545, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.4390824129141886, | |
| "grad_norm": 1.1095722913742065, | |
| "learning_rate": 9.240438941759926e-05, | |
| "loss": 3.3787, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.44044180118946474, | |
| "grad_norm": 0.8704729080200195, | |
| "learning_rate": 9.234474767373465e-05, | |
| "loss": 3.2712, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.44180118946474084, | |
| "grad_norm": 0.7083353400230408, | |
| "learning_rate": 9.228489209192652e-05, | |
| "loss": 3.3055, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.443160577740017, | |
| "grad_norm": 0.8583936095237732, | |
| "learning_rate": 9.222482297444131e-05, | |
| "loss": 3.2406, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.44451996601529314, | |
| "grad_norm": 0.9450501203536987, | |
| "learning_rate": 9.216454062462374e-05, | |
| "loss": 3.2526, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.44587935429056924, | |
| "grad_norm": 0.7373863458633423, | |
| "learning_rate": 9.210404534689536e-05, | |
| "loss": 3.2554, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.4472387425658454, | |
| "grad_norm": 1.0712296962738037, | |
| "learning_rate": 9.2043337446753e-05, | |
| "loss": 3.3109, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.4485981308411215, | |
| "grad_norm": 0.8983728885650635, | |
| "learning_rate": 9.198241723076719e-05, | |
| "loss": 3.2369, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.44995751911639764, | |
| "grad_norm": 0.8862040042877197, | |
| "learning_rate": 9.192128500658068e-05, | |
| "loss": 3.2532, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.45131690739167374, | |
| "grad_norm": 0.8523765206336975, | |
| "learning_rate": 9.185994108290682e-05, | |
| "loss": 3.2559, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.4526762956669499, | |
| "grad_norm": 0.9425597190856934, | |
| "learning_rate": 9.179838576952802e-05, | |
| "loss": 3.2814, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.454035683942226, | |
| "grad_norm": 0.7966018915176392, | |
| "learning_rate": 9.173661937729421e-05, | |
| "loss": 3.2376, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.45539507221750214, | |
| "grad_norm": 0.7609128952026367, | |
| "learning_rate": 9.167464221812126e-05, | |
| "loss": 3.1981, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.45675446049277824, | |
| "grad_norm": 0.7252809405326843, | |
| "learning_rate": 9.161245460498936e-05, | |
| "loss": 3.3001, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.4581138487680544, | |
| "grad_norm": 0.5679906606674194, | |
| "learning_rate": 9.155005685194152e-05, | |
| "loss": 3.2215, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.4594732370433305, | |
| "grad_norm": 0.8006565570831299, | |
| "learning_rate": 9.148744927408193e-05, | |
| "loss": 3.2696, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.46083262531860664, | |
| "grad_norm": 0.9909971952438354, | |
| "learning_rate": 9.142463218757437e-05, | |
| "loss": 3.3098, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.46219201359388273, | |
| "grad_norm": 0.729555606842041, | |
| "learning_rate": 9.136160590964063e-05, | |
| "loss": 3.188, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.4635514018691589, | |
| "grad_norm": 0.9540812969207764, | |
| "learning_rate": 9.129837075855887e-05, | |
| "loss": 3.2464, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.464910790144435, | |
| "grad_norm": 0.9037480354309082, | |
| "learning_rate": 9.123492705366212e-05, | |
| "loss": 3.3134, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.46627017841971113, | |
| "grad_norm": 0.7540014386177063, | |
| "learning_rate": 9.117127511533654e-05, | |
| "loss": 3.2704, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.46762956669498723, | |
| "grad_norm": 0.7283887267112732, | |
| "learning_rate": 9.110741526501982e-05, | |
| "loss": 3.147, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.4689889549702634, | |
| "grad_norm": 0.7288716435432434, | |
| "learning_rate": 9.104334782519969e-05, | |
| "loss": 3.2257, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.4703483432455395, | |
| "grad_norm": 0.8213950991630554, | |
| "learning_rate": 9.097907311941208e-05, | |
| "loss": 3.1973, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.47170773152081563, | |
| "grad_norm": 0.773909866809845, | |
| "learning_rate": 9.091459147223968e-05, | |
| "loss": 3.2165, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.4730671197960918, | |
| "grad_norm": 0.9411752820014954, | |
| "learning_rate": 9.08499032093102e-05, | |
| "loss": 3.2191, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.4744265080713679, | |
| "grad_norm": 0.8191618919372559, | |
| "learning_rate": 9.078500865729471e-05, | |
| "loss": 3.1823, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.47578589634664403, | |
| "grad_norm": 0.7489372491836548, | |
| "learning_rate": 9.071990814390606e-05, | |
| "loss": 3.1833, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.47714528462192013, | |
| "grad_norm": 0.7682393193244934, | |
| "learning_rate": 9.065460199789719e-05, | |
| "loss": 3.2761, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.4785046728971963, | |
| "grad_norm": 0.9797720313072205, | |
| "learning_rate": 9.058909054905946e-05, | |
| "loss": 3.2743, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.4798640611724724, | |
| "grad_norm": 0.9919414520263672, | |
| "learning_rate": 9.052337412822096e-05, | |
| "loss": 3.2964, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.48122344944774853, | |
| "grad_norm": 0.8961633443832397, | |
| "learning_rate": 9.045745306724495e-05, | |
| "loss": 3.2386, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.4825828377230246, | |
| "grad_norm": 0.78994220495224, | |
| "learning_rate": 9.039132769902808e-05, | |
| "loss": 3.3232, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.4839422259983008, | |
| "grad_norm": 0.7165075540542603, | |
| "learning_rate": 9.032499835749868e-05, | |
| "loss": 3.1898, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.4853016142735769, | |
| "grad_norm": 0.6712597608566284, | |
| "learning_rate": 9.02584653776152e-05, | |
| "loss": 3.2442, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.48666100254885303, | |
| "grad_norm": 0.7340356707572937, | |
| "learning_rate": 9.019172909536441e-05, | |
| "loss": 3.2368, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.4880203908241291, | |
| "grad_norm": 0.9872474074363708, | |
| "learning_rate": 9.012478984775977e-05, | |
| "loss": 3.2627, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.4893797790994053, | |
| "grad_norm": 0.8943246006965637, | |
| "learning_rate": 9.005764797283969e-05, | |
| "loss": 3.1771, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.4907391673746814, | |
| "grad_norm": 0.8461526036262512, | |
| "learning_rate": 8.99903038096658e-05, | |
| "loss": 3.2448, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.4920985556499575, | |
| "grad_norm": 0.9530143737792969, | |
| "learning_rate": 8.992275769832135e-05, | |
| "loss": 3.1837, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.4934579439252336, | |
| "grad_norm": 0.8631094098091125, | |
| "learning_rate": 8.985500997990933e-05, | |
| "loss": 3.2388, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.4948173322005098, | |
| "grad_norm": 0.7376362681388855, | |
| "learning_rate": 8.978706099655088e-05, | |
| "loss": 3.2585, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.49617672047578587, | |
| "grad_norm": 0.7359983325004578, | |
| "learning_rate": 8.97189110913835e-05, | |
| "loss": 3.2278, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.497536108751062, | |
| "grad_norm": 0.8055249452590942, | |
| "learning_rate": 8.965056060855931e-05, | |
| "loss": 3.2128, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.4988954970263382, | |
| "grad_norm": 0.8565118908882141, | |
| "learning_rate": 8.95820098932434e-05, | |
| "loss": 3.2337, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.5002548853016143, | |
| "grad_norm": 0.7358447313308716, | |
| "learning_rate": 8.951325929161191e-05, | |
| "loss": 3.2258, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.5016142735768904, | |
| "grad_norm": 0.8792275786399841, | |
| "learning_rate": 8.944430915085051e-05, | |
| "loss": 3.2875, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.5029736618521665, | |
| "grad_norm": 0.9467582106590271, | |
| "learning_rate": 8.937515981915245e-05, | |
| "loss": 3.2911, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5043330501274427, | |
| "grad_norm": 0.7510040998458862, | |
| "learning_rate": 8.930581164571692e-05, | |
| "loss": 3.2342, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.5056924384027188, | |
| "grad_norm": 0.9511517882347107, | |
| "learning_rate": 8.92362649807472e-05, | |
| "loss": 3.2036, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.5070518266779949, | |
| "grad_norm": 0.7870407104492188, | |
| "learning_rate": 8.916652017544899e-05, | |
| "loss": 3.2493, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.508411214953271, | |
| "grad_norm": 0.8817398548126221, | |
| "learning_rate": 8.909657758202857e-05, | |
| "loss": 3.1132, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.5097706032285472, | |
| "grad_norm": 0.8300465941429138, | |
| "learning_rate": 8.9026437553691e-05, | |
| "loss": 3.2223, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5111299915038233, | |
| "grad_norm": 0.7648670077323914, | |
| "learning_rate": 8.89561004446384e-05, | |
| "loss": 3.2196, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.5124893797790994, | |
| "grad_norm": 0.6600602865219116, | |
| "learning_rate": 8.888556661006818e-05, | |
| "loss": 3.1552, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.5138487680543755, | |
| "grad_norm": 0.7653377652168274, | |
| "learning_rate": 8.881483640617112e-05, | |
| "loss": 3.1453, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.5152081563296517, | |
| "grad_norm": 0.793511688709259, | |
| "learning_rate": 8.874391019012967e-05, | |
| "loss": 3.1663, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.5165675446049278, | |
| "grad_norm": 0.6547530889511108, | |
| "learning_rate": 8.867278832011617e-05, | |
| "loss": 3.212, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.5179269328802039, | |
| "grad_norm": 0.7647628784179688, | |
| "learning_rate": 8.860147115529096e-05, | |
| "loss": 3.1953, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.51928632115548, | |
| "grad_norm": 0.6164153218269348, | |
| "learning_rate": 8.852995905580063e-05, | |
| "loss": 3.1791, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.5206457094307562, | |
| "grad_norm": 0.7801313996315002, | |
| "learning_rate": 8.845825238277614e-05, | |
| "loss": 3.1899, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.5220050977060323, | |
| "grad_norm": 0.8477320075035095, | |
| "learning_rate": 8.838635149833106e-05, | |
| "loss": 3.2452, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.5233644859813084, | |
| "grad_norm": 0.69857257604599, | |
| "learning_rate": 8.831425676555971e-05, | |
| "loss": 3.218, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.5247238742565845, | |
| "grad_norm": 0.8784200549125671, | |
| "learning_rate": 8.824196854853533e-05, | |
| "loss": 3.1743, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.5260832625318607, | |
| "grad_norm": 0.7521477937698364, | |
| "learning_rate": 8.816948721230822e-05, | |
| "loss": 3.204, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.5274426508071368, | |
| "grad_norm": 0.7114652395248413, | |
| "learning_rate": 8.809681312290397e-05, | |
| "loss": 3.1783, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.528802039082413, | |
| "grad_norm": 0.7126603126525879, | |
| "learning_rate": 8.802394664732152e-05, | |
| "loss": 3.1708, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.530161427357689, | |
| "grad_norm": 0.7315563559532166, | |
| "learning_rate": 8.795088815353135e-05, | |
| "loss": 3.1273, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.5315208156329652, | |
| "grad_norm": 0.7203145623207092, | |
| "learning_rate": 8.78776380104736e-05, | |
| "loss": 3.1287, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.5328802039082413, | |
| "grad_norm": 0.7698382139205933, | |
| "learning_rate": 8.780419658805625e-05, | |
| "loss": 3.1671, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.5342395921835175, | |
| "grad_norm": 0.7925199866294861, | |
| "learning_rate": 8.773056425715326e-05, | |
| "loss": 3.089, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.5355989804587935, | |
| "grad_norm": 0.6636077761650085, | |
| "learning_rate": 8.765674138960261e-05, | |
| "loss": 3.2514, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.5369583687340697, | |
| "grad_norm": 0.785798966884613, | |
| "learning_rate": 8.758272835820451e-05, | |
| "loss": 3.2389, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.5383177570093458, | |
| "grad_norm": 0.7094962000846863, | |
| "learning_rate": 8.750852553671942e-05, | |
| "loss": 3.2141, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.539677145284622, | |
| "grad_norm": 0.7799323797225952, | |
| "learning_rate": 8.743413329986632e-05, | |
| "loss": 3.1314, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.541036533559898, | |
| "grad_norm": 0.6692888736724854, | |
| "learning_rate": 8.735955202332065e-05, | |
| "loss": 3.1683, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.5423959218351742, | |
| "grad_norm": 0.7615490555763245, | |
| "learning_rate": 8.728478208371256e-05, | |
| "loss": 3.2109, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.5437553101104503, | |
| "grad_norm": 0.7399694919586182, | |
| "learning_rate": 8.720982385862483e-05, | |
| "loss": 3.1026, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5451146983857265, | |
| "grad_norm": 0.6580244898796082, | |
| "learning_rate": 8.713467772659118e-05, | |
| "loss": 3.1839, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.5464740866610025, | |
| "grad_norm": 0.7819783091545105, | |
| "learning_rate": 8.705934406709417e-05, | |
| "loss": 3.132, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.5478334749362787, | |
| "grad_norm": 0.7655234336853027, | |
| "learning_rate": 8.69838232605634e-05, | |
| "loss": 3.1175, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.5491928632115548, | |
| "grad_norm": 0.7018226385116577, | |
| "learning_rate": 8.690811568837352e-05, | |
| "loss": 3.0879, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.550552251486831, | |
| "grad_norm": 0.5926535129547119, | |
| "learning_rate": 8.683222173284236e-05, | |
| "loss": 3.1408, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.551911639762107, | |
| "grad_norm": 0.7037495970726013, | |
| "learning_rate": 8.675614177722895e-05, | |
| "loss": 3.1749, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.5532710280373832, | |
| "grad_norm": 0.7136229872703552, | |
| "learning_rate": 8.667987620573163e-05, | |
| "loss": 3.1631, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.5546304163126593, | |
| "grad_norm": 0.7994447946548462, | |
| "learning_rate": 8.660342540348606e-05, | |
| "loss": 3.149, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.5559898045879355, | |
| "grad_norm": 0.5606616735458374, | |
| "learning_rate": 8.652678975656335e-05, | |
| "loss": 3.1668, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.5573491928632116, | |
| "grad_norm": 0.8005346059799194, | |
| "learning_rate": 8.644996965196797e-05, | |
| "loss": 3.1792, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.5587085811384876, | |
| "grad_norm": 0.6956243515014648, | |
| "learning_rate": 8.637296547763601e-05, | |
| "loss": 3.1572, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.5600679694137638, | |
| "grad_norm": 0.7457271218299866, | |
| "learning_rate": 8.6295777622433e-05, | |
| "loss": 3.2719, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.56142735768904, | |
| "grad_norm": 0.7460895776748657, | |
| "learning_rate": 8.621840647615207e-05, | |
| "loss": 3.1218, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.5627867459643161, | |
| "grad_norm": 0.7265388369560242, | |
| "learning_rate": 8.614085242951201e-05, | |
| "loss": 3.1595, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.5641461342395921, | |
| "grad_norm": 0.7856431007385254, | |
| "learning_rate": 8.606311587415518e-05, | |
| "loss": 3.16, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.5655055225148683, | |
| "grad_norm": 0.7534942030906677, | |
| "learning_rate": 8.598519720264562e-05, | |
| "loss": 3.1308, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.5668649107901444, | |
| "grad_norm": 0.6927157640457153, | |
| "learning_rate": 8.590709680846702e-05, | |
| "loss": 3.1319, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.5682242990654206, | |
| "grad_norm": 0.7308480739593506, | |
| "learning_rate": 8.58288150860208e-05, | |
| "loss": 3.2368, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.5695836873406966, | |
| "grad_norm": 0.691619336605072, | |
| "learning_rate": 8.575035243062407e-05, | |
| "loss": 3.0438, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.5709430756159728, | |
| "grad_norm": 0.6600204110145569, | |
| "learning_rate": 8.567170923850759e-05, | |
| "loss": 3.1309, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.572302463891249, | |
| "grad_norm": 0.6632899641990662, | |
| "learning_rate": 8.559288590681387e-05, | |
| "loss": 3.0793, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.5736618521665251, | |
| "grad_norm": 0.7121036052703857, | |
| "learning_rate": 8.551388283359507e-05, | |
| "loss": 3.1492, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.5750212404418011, | |
| "grad_norm": 0.9197178483009338, | |
| "learning_rate": 8.543470041781107e-05, | |
| "loss": 3.3029, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.5763806287170773, | |
| "grad_norm": 0.6388615369796753, | |
| "learning_rate": 8.535533905932738e-05, | |
| "loss": 3.169, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.5777400169923534, | |
| "grad_norm": 0.687680184841156, | |
| "learning_rate": 8.527579915891321e-05, | |
| "loss": 3.074, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.5790994052676296, | |
| "grad_norm": 0.6250393390655518, | |
| "learning_rate": 8.519608111823931e-05, | |
| "loss": 3.1493, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.5804587935429056, | |
| "grad_norm": 0.6668142676353455, | |
| "learning_rate": 8.511618533987613e-05, | |
| "loss": 3.236, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.5818181818181818, | |
| "grad_norm": 0.7983154058456421, | |
| "learning_rate": 8.50361122272916e-05, | |
| "loss": 3.1506, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.5831775700934579, | |
| "grad_norm": 0.8851556181907654, | |
| "learning_rate": 8.495586218484923e-05, | |
| "loss": 3.1296, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.5845369583687341, | |
| "grad_norm": 0.8044039607048035, | |
| "learning_rate": 8.487543561780595e-05, | |
| "loss": 3.1169, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.5858963466440102, | |
| "grad_norm": 0.8117210865020752, | |
| "learning_rate": 8.479483293231019e-05, | |
| "loss": 3.1419, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.5872557349192863, | |
| "grad_norm": 0.7853211164474487, | |
| "learning_rate": 8.471405453539974e-05, | |
| "loss": 3.1049, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.5886151231945624, | |
| "grad_norm": 0.7936228513717651, | |
| "learning_rate": 8.463310083499971e-05, | |
| "loss": 3.2174, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.5899745114698386, | |
| "grad_norm": 0.69145268201828, | |
| "learning_rate": 8.45519722399205e-05, | |
| "loss": 3.1685, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.5913338997451147, | |
| "grad_norm": 0.8211259841918945, | |
| "learning_rate": 8.447066915985568e-05, | |
| "loss": 3.1633, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.5926932880203908, | |
| "grad_norm": 0.7517729997634888, | |
| "learning_rate": 8.438919200538003e-05, | |
| "loss": 3.1548, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.5940526762956669, | |
| "grad_norm": 0.7419262528419495, | |
| "learning_rate": 8.43075411879473e-05, | |
| "loss": 3.1094, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.5954120645709431, | |
| "grad_norm": 0.7104995846748352, | |
| "learning_rate": 8.42257171198883e-05, | |
| "loss": 3.1426, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.5967714528462192, | |
| "grad_norm": 0.6559504270553589, | |
| "learning_rate": 8.414372021440868e-05, | |
| "loss": 3.1492, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.5981308411214953, | |
| "grad_norm": 0.734095573425293, | |
| "learning_rate": 8.406155088558698e-05, | |
| "loss": 3.1309, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.5994902293967714, | |
| "grad_norm": 0.5765071511268616, | |
| "learning_rate": 8.397920954837242e-05, | |
| "loss": 3.1223, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.6008496176720476, | |
| "grad_norm": 0.73897385597229, | |
| "learning_rate": 8.389669661858284e-05, | |
| "loss": 3.1529, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.6022090059473237, | |
| "grad_norm": 0.6814929842948914, | |
| "learning_rate": 8.381401251290264e-05, | |
| "loss": 3.0561, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.6035683942225998, | |
| "grad_norm": 0.730022132396698, | |
| "learning_rate": 8.373115764888062e-05, | |
| "loss": 3.0702, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.6049277824978759, | |
| "grad_norm": 0.6274394392967224, | |
| "learning_rate": 8.36481324449279e-05, | |
| "loss": 3.1253, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.6062871707731521, | |
| "grad_norm": 0.7467242479324341, | |
| "learning_rate": 8.356493732031586e-05, | |
| "loss": 3.0619, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.6076465590484282, | |
| "grad_norm": 0.6906052231788635, | |
| "learning_rate": 8.34815726951739e-05, | |
| "loss": 3.2207, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.6090059473237044, | |
| "grad_norm": 0.7023007273674011, | |
| "learning_rate": 8.339803899048737e-05, | |
| "loss": 3.1186, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.6103653355989804, | |
| "grad_norm": 0.644953727722168, | |
| "learning_rate": 8.331433662809555e-05, | |
| "loss": 3.1045, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.6117247238742566, | |
| "grad_norm": 0.6254658102989197, | |
| "learning_rate": 8.323046603068934e-05, | |
| "loss": 3.0299, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.6130841121495327, | |
| "grad_norm": 0.9075002074241638, | |
| "learning_rate": 8.314642762180927e-05, | |
| "loss": 3.1854, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.6144435004248089, | |
| "grad_norm": 0.8213093280792236, | |
| "learning_rate": 8.306222182584328e-05, | |
| "loss": 3.2035, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.6158028887000849, | |
| "grad_norm": 0.7945600748062134, | |
| "learning_rate": 8.297784906802462e-05, | |
| "loss": 3.1396, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.6171622769753611, | |
| "grad_norm": 0.8038851022720337, | |
| "learning_rate": 8.289330977442967e-05, | |
| "loss": 3.1608, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.6185216652506372, | |
| "grad_norm": 0.6917005777359009, | |
| "learning_rate": 8.280860437197579e-05, | |
| "loss": 3.1391, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.6198810535259134, | |
| "grad_norm": 1.0512206554412842, | |
| "learning_rate": 8.272373328841923e-05, | |
| "loss": 3.1953, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.6212404418011894, | |
| "grad_norm": 0.9130175709724426, | |
| "learning_rate": 8.263869695235285e-05, | |
| "loss": 3.1308, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.6225998300764656, | |
| "grad_norm": 0.794154942035675, | |
| "learning_rate": 8.255349579320408e-05, | |
| "loss": 3.1505, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.6239592183517417, | |
| "grad_norm": 0.7815708518028259, | |
| "learning_rate": 8.246813024123267e-05, | |
| "loss": 3.0952, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.6253186066270179, | |
| "grad_norm": 0.7132216691970825, | |
| "learning_rate": 8.238260072752855e-05, | |
| "loss": 3.1855, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.6266779949022939, | |
| "grad_norm": 0.7026944756507874, | |
| "learning_rate": 8.229690768400962e-05, | |
| "loss": 3.078, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.6280373831775701, | |
| "grad_norm": 0.809191882610321, | |
| "learning_rate": 8.221105154341964e-05, | |
| "loss": 3.1573, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.6293967714528462, | |
| "grad_norm": 0.721347987651825, | |
| "learning_rate": 8.212503273932593e-05, | |
| "loss": 3.2011, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.6307561597281224, | |
| "grad_norm": 0.7457506060600281, | |
| "learning_rate": 8.203885170611734e-05, | |
| "loss": 3.0995, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.6321155480033984, | |
| "grad_norm": 0.6334044933319092, | |
| "learning_rate": 8.195250887900188e-05, | |
| "loss": 3.1587, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.6334749362786746, | |
| "grad_norm": 0.5967175960540771, | |
| "learning_rate": 8.186600469400467e-05, | |
| "loss": 3.1125, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.6348343245539507, | |
| "grad_norm": 0.6264853477478027, | |
| "learning_rate": 8.177933958796565e-05, | |
| "loss": 3.055, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.6361937128292269, | |
| "grad_norm": 0.5743052959442139, | |
| "learning_rate": 8.169251399853737e-05, | |
| "loss": 3.0728, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.637553101104503, | |
| "grad_norm": 0.5435692667961121, | |
| "learning_rate": 8.160552836418285e-05, | |
| "loss": 3.0973, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.6389124893797791, | |
| "grad_norm": 0.6908643245697021, | |
| "learning_rate": 8.151838312417332e-05, | |
| "loss": 3.0482, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.6402718776550552, | |
| "grad_norm": 0.598528265953064, | |
| "learning_rate": 8.143107871858603e-05, | |
| "loss": 3.0724, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.6416312659303314, | |
| "grad_norm": 0.669939398765564, | |
| "learning_rate": 8.134361558830193e-05, | |
| "loss": 3.1008, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.6429906542056075, | |
| "grad_norm": 0.6620224118232727, | |
| "learning_rate": 8.125599417500359e-05, | |
| "loss": 3.0508, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.6443500424808836, | |
| "grad_norm": 0.556448221206665, | |
| "learning_rate": 8.116821492117285e-05, | |
| "loss": 3.101, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.6457094307561597, | |
| "grad_norm": 0.6877358555793762, | |
| "learning_rate": 8.108027827008871e-05, | |
| "loss": 3.1337, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.6470688190314359, | |
| "grad_norm": 0.6986392140388489, | |
| "learning_rate": 8.09921846658249e-05, | |
| "loss": 3.0152, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.648428207306712, | |
| "grad_norm": 0.6411210298538208, | |
| "learning_rate": 8.090393455324783e-05, | |
| "loss": 3.1043, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.6497875955819881, | |
| "grad_norm": 0.6838498115539551, | |
| "learning_rate": 8.081552837801427e-05, | |
| "loss": 3.1058, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.6511469838572642, | |
| "grad_norm": 0.6829114556312561, | |
| "learning_rate": 8.072696658656906e-05, | |
| "loss": 3.1245, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.6525063721325404, | |
| "grad_norm": 0.6059818863868713, | |
| "learning_rate": 8.06382496261429e-05, | |
| "loss": 3.0519, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.6538657604078165, | |
| "grad_norm": 0.6728681325912476, | |
| "learning_rate": 8.05493779447501e-05, | |
| "loss": 3.0275, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.6552251486830926, | |
| "grad_norm": 0.7594171762466431, | |
| "learning_rate": 8.046035199118626e-05, | |
| "loss": 3.1189, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.6565845369583687, | |
| "grad_norm": 0.5925278067588806, | |
| "learning_rate": 8.037117221502611e-05, | |
| "loss": 3.0737, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.6579439252336449, | |
| "grad_norm": 0.7238386273384094, | |
| "learning_rate": 8.028183906662113e-05, | |
| "loss": 3.0778, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.659303313508921, | |
| "grad_norm": 0.627905011177063, | |
| "learning_rate": 8.019235299709726e-05, | |
| "loss": 3.0713, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.6606627017841971, | |
| "grad_norm": 0.6918131113052368, | |
| "learning_rate": 8.010271445835282e-05, | |
| "loss": 3.1039, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.6620220900594732, | |
| "grad_norm": 0.6929610371589661, | |
| "learning_rate": 8.001292390305597e-05, | |
| "loss": 2.9954, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.6633814783347494, | |
| "grad_norm": 0.6240336894989014, | |
| "learning_rate": 7.992298178464261e-05, | |
| "loss": 3.0909, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.6647408666100255, | |
| "grad_norm": 0.678880512714386, | |
| "learning_rate": 7.983288855731398e-05, | |
| "loss": 3.1088, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.6661002548853017, | |
| "grad_norm": 0.716385006904602, | |
| "learning_rate": 7.974264467603442e-05, | |
| "loss": 3.0681, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.6674596431605777, | |
| "grad_norm": 0.7473633289337158, | |
| "learning_rate": 7.965225059652909e-05, | |
| "loss": 3.0411, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.6688190314358539, | |
| "grad_norm": 0.6619102954864502, | |
| "learning_rate": 7.956170677528159e-05, | |
| "loss": 3.1295, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.67017841971113, | |
| "grad_norm": 0.6102796792984009, | |
| "learning_rate": 7.947101366953177e-05, | |
| "loss": 3.0331, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.6715378079864062, | |
| "grad_norm": 0.5997730493545532, | |
| "learning_rate": 7.938017173727328e-05, | |
| "loss": 3.1051, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.6728971962616822, | |
| "grad_norm": 0.7003008127212524, | |
| "learning_rate": 7.92891814372514e-05, | |
| "loss": 3.1023, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.6742565845369584, | |
| "grad_norm": 0.6912761330604553, | |
| "learning_rate": 7.919804322896062e-05, | |
| "loss": 3.1274, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.6756159728122345, | |
| "grad_norm": 0.6718102097511292, | |
| "learning_rate": 7.910675757264238e-05, | |
| "loss": 3.0487, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.6769753610875107, | |
| "grad_norm": 0.6638507843017578, | |
| "learning_rate": 7.901532492928269e-05, | |
| "loss": 3.0132, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.6783347493627867, | |
| "grad_norm": 0.6588859558105469, | |
| "learning_rate": 7.892374576060986e-05, | |
| "loss": 3.0839, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.6796941376380629, | |
| "grad_norm": 0.7421138882637024, | |
| "learning_rate": 7.88320205290921e-05, | |
| "loss": 3.0802, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.681053525913339, | |
| "grad_norm": 0.6962835192680359, | |
| "learning_rate": 7.874014969793533e-05, | |
| "loss": 3.1173, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 0.6824129141886152, | |
| "grad_norm": 0.7081164717674255, | |
| "learning_rate": 7.864813373108061e-05, | |
| "loss": 3.0741, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 0.6837723024638912, | |
| "grad_norm": 0.7623039484024048, | |
| "learning_rate": 7.8555973093202e-05, | |
| "loss": 3.1217, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 0.6851316907391674, | |
| "grad_norm": 0.5826273560523987, | |
| "learning_rate": 7.846366824970413e-05, | |
| "loss": 3.0731, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 0.6864910790144435, | |
| "grad_norm": 0.6277479529380798, | |
| "learning_rate": 7.837121966671986e-05, | |
| "loss": 3.1101, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.6878504672897197, | |
| "grad_norm": 0.6380975842475891, | |
| "learning_rate": 7.827862781110788e-05, | |
| "loss": 3.1143, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 0.6892098555649957, | |
| "grad_norm": 0.72269207239151, | |
| "learning_rate": 7.818589315045046e-05, | |
| "loss": 3.0956, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 0.6905692438402719, | |
| "grad_norm": 0.65102219581604, | |
| "learning_rate": 7.809301615305098e-05, | |
| "loss": 3.0571, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 0.691928632115548, | |
| "grad_norm": 0.7135101556777954, | |
| "learning_rate": 7.799999728793162e-05, | |
| "loss": 3.0262, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 0.6932880203908242, | |
| "grad_norm": 0.6412085294723511, | |
| "learning_rate": 7.790683702483102e-05, | |
| "loss": 3.0946, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.6946474086661003, | |
| "grad_norm": 0.5655577182769775, | |
| "learning_rate": 7.78135358342018e-05, | |
| "loss": 3.0359, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 0.6960067969413763, | |
| "grad_norm": 0.7037889361381531, | |
| "learning_rate": 7.772009418720832e-05, | |
| "loss": 3.0498, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.6973661852166525, | |
| "grad_norm": 0.5886219143867493, | |
| "learning_rate": 7.76265125557242e-05, | |
| "loss": 3.0468, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 0.6987255734919287, | |
| "grad_norm": 0.5971575379371643, | |
| "learning_rate": 7.753279141232995e-05, | |
| "loss": 3.0748, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 0.7000849617672048, | |
| "grad_norm": 0.6181765198707581, | |
| "learning_rate": 7.743893123031068e-05, | |
| "loss": 3.0709, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.7014443500424808, | |
| "grad_norm": 0.6223234534263611, | |
| "learning_rate": 7.734493248365355e-05, | |
| "loss": 3.0975, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 0.702803738317757, | |
| "grad_norm": 0.7307218313217163, | |
| "learning_rate": 7.72507956470455e-05, | |
| "loss": 3.1162, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 0.7041631265930332, | |
| "grad_norm": 0.713442325592041, | |
| "learning_rate": 7.715652119587085e-05, | |
| "loss": 3.1112, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 0.7055225148683093, | |
| "grad_norm": 0.6621354818344116, | |
| "learning_rate": 7.706210960620878e-05, | |
| "loss": 3.1128, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 0.7068819031435853, | |
| "grad_norm": 0.6017237901687622, | |
| "learning_rate": 7.696756135483109e-05, | |
| "loss": 3.0969, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.7082412914188615, | |
| "grad_norm": 0.6268143057823181, | |
| "learning_rate": 7.687287691919968e-05, | |
| "loss": 3.0387, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 0.7096006796941376, | |
| "grad_norm": 0.6222430467605591, | |
| "learning_rate": 7.677805677746415e-05, | |
| "loss": 3.0769, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 0.7109600679694138, | |
| "grad_norm": 0.6244910359382629, | |
| "learning_rate": 7.668310140845944e-05, | |
| "loss": 2.9467, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 0.7123194562446898, | |
| "grad_norm": 0.7029285430908203, | |
| "learning_rate": 7.658801129170335e-05, | |
| "loss": 3.0942, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 0.713678844519966, | |
| "grad_norm": 0.7414494156837463, | |
| "learning_rate": 7.649278690739418e-05, | |
| "loss": 3.07, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.7150382327952421, | |
| "grad_norm": 0.6464896202087402, | |
| "learning_rate": 7.639742873640825e-05, | |
| "loss": 3.0708, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 0.7163976210705183, | |
| "grad_norm": 0.8883520364761353, | |
| "learning_rate": 7.630193726029746e-05, | |
| "loss": 3.1425, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 0.7177570093457943, | |
| "grad_norm": 0.7830668687820435, | |
| "learning_rate": 7.620631296128698e-05, | |
| "loss": 3.1458, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 0.7191163976210705, | |
| "grad_norm": 0.790330708026886, | |
| "learning_rate": 7.611055632227262e-05, | |
| "loss": 3.0844, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 0.7204757858963466, | |
| "grad_norm": 0.7381483316421509, | |
| "learning_rate": 7.601466782681859e-05, | |
| "loss": 3.075, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.7218351741716228, | |
| "grad_norm": 0.6107451319694519, | |
| "learning_rate": 7.591864795915492e-05, | |
| "loss": 3.0409, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 0.723194562446899, | |
| "grad_norm": 0.627289354801178, | |
| "learning_rate": 7.582249720417504e-05, | |
| "loss": 2.9797, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 0.724553950722175, | |
| "grad_norm": 0.63350909948349, | |
| "learning_rate": 7.57262160474334e-05, | |
| "loss": 3.083, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 0.7259133389974511, | |
| "grad_norm": 0.6966723799705505, | |
| "learning_rate": 7.562980497514294e-05, | |
| "loss": 3.0733, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 0.7272727272727273, | |
| "grad_norm": 0.6452199220657349, | |
| "learning_rate": 7.553326447417267e-05, | |
| "loss": 3.0801, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.7286321155480034, | |
| "grad_norm": 0.6325047016143799, | |
| "learning_rate": 7.543659503204522e-05, | |
| "loss": 2.959, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 0.7299915038232795, | |
| "grad_norm": 0.6129205226898193, | |
| "learning_rate": 7.533979713693433e-05, | |
| "loss": 3.0407, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 0.7313508920985556, | |
| "grad_norm": 0.6302885413169861, | |
| "learning_rate": 7.524287127766245e-05, | |
| "loss": 3.102, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 0.7327102803738318, | |
| "grad_norm": 0.6648517847061157, | |
| "learning_rate": 7.514581794369822e-05, | |
| "loss": 3.1113, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 0.7340696686491079, | |
| "grad_norm": 0.7655189633369446, | |
| "learning_rate": 7.504863762515404e-05, | |
| "loss": 3.0586, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.735429056924384, | |
| "grad_norm": 0.6116030216217041, | |
| "learning_rate": 7.495133081278356e-05, | |
| "loss": 3.0208, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 0.7367884451996601, | |
| "grad_norm": 0.6955434679985046, | |
| "learning_rate": 7.48538979979792e-05, | |
| "loss": 3.1177, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 0.7381478334749363, | |
| "grad_norm": 0.6810325384140015, | |
| "learning_rate": 7.47563396727697e-05, | |
| "loss": 3.0297, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 0.7395072217502124, | |
| "grad_norm": 0.6736805438995361, | |
| "learning_rate": 7.465865632981763e-05, | |
| "loss": 3.0411, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.7408666100254885, | |
| "grad_norm": 0.689484179019928, | |
| "learning_rate": 7.456084846241687e-05, | |
| "loss": 3.0751, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.7422259983007646, | |
| "grad_norm": 0.621320366859436, | |
| "learning_rate": 7.446291656449014e-05, | |
| "loss": 3.035, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 0.7435853865760408, | |
| "grad_norm": 0.6816211938858032, | |
| "learning_rate": 7.436486113058651e-05, | |
| "loss": 3.0209, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 0.7449447748513169, | |
| "grad_norm": 0.7227087616920471, | |
| "learning_rate": 7.426668265587892e-05, | |
| "loss": 2.9714, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 0.746304163126593, | |
| "grad_norm": 0.6621195673942566, | |
| "learning_rate": 7.416838163616162e-05, | |
| "loss": 3.0837, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 0.7476635514018691, | |
| "grad_norm": 0.5446696877479553, | |
| "learning_rate": 7.406995856784772e-05, | |
| "loss": 3.0743, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.7490229396771453, | |
| "grad_norm": 0.6033000349998474, | |
| "learning_rate": 7.397141394796667e-05, | |
| "loss": 2.9843, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 0.7503823279524214, | |
| "grad_norm": 0.5980247259140015, | |
| "learning_rate": 7.387274827416175e-05, | |
| "loss": 3.066, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 0.7517417162276976, | |
| "grad_norm": 0.6181382536888123, | |
| "learning_rate": 7.377396204468754e-05, | |
| "loss": 3.0066, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 0.7531011045029736, | |
| "grad_norm": 0.7175232172012329, | |
| "learning_rate": 7.367505575840741e-05, | |
| "loss": 3.0747, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 0.7544604927782498, | |
| "grad_norm": 0.6160265207290649, | |
| "learning_rate": 7.357602991479106e-05, | |
| "loss": 3.0461, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.7558198810535259, | |
| "grad_norm": 0.6809048652648926, | |
| "learning_rate": 7.347688501391187e-05, | |
| "loss": 3.0333, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 0.7571792693288021, | |
| "grad_norm": 0.6925624012947083, | |
| "learning_rate": 7.337762155644454e-05, | |
| "loss": 3.0453, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 0.7585386576040781, | |
| "grad_norm": 0.6381937265396118, | |
| "learning_rate": 7.327824004366237e-05, | |
| "loss": 3.0466, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 0.7598980458793543, | |
| "grad_norm": 0.6669313311576843, | |
| "learning_rate": 7.317874097743491e-05, | |
| "loss": 2.9596, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 0.7612574341546304, | |
| "grad_norm": 0.7228459119796753, | |
| "learning_rate": 7.30791248602253e-05, | |
| "loss": 3.093, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.7626168224299066, | |
| "grad_norm": 0.5597134828567505, | |
| "learning_rate": 7.297939219508781e-05, | |
| "loss": 3.0038, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 0.7639762107051826, | |
| "grad_norm": 0.6500145196914673, | |
| "learning_rate": 7.287954348566529e-05, | |
| "loss": 2.9154, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 0.7653355989804588, | |
| "grad_norm": 0.5991110801696777, | |
| "learning_rate": 7.277957923618652e-05, | |
| "loss": 3.0487, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 0.7666949872557349, | |
| "grad_norm": 0.7363496422767639, | |
| "learning_rate": 7.267949995146383e-05, | |
| "loss": 3.0155, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 0.7680543755310111, | |
| "grad_norm": 0.6778906583786011, | |
| "learning_rate": 7.257930613689043e-05, | |
| "loss": 3.0946, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.7694137638062871, | |
| "grad_norm": 0.6843693256378174, | |
| "learning_rate": 7.24789982984379e-05, | |
| "loss": 2.9874, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 0.7707731520815633, | |
| "grad_norm": 0.6377856135368347, | |
| "learning_rate": 7.237857694265368e-05, | |
| "loss": 3.0625, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 0.7721325403568394, | |
| "grad_norm": 0.6182267069816589, | |
| "learning_rate": 7.227804257665837e-05, | |
| "loss": 3.0279, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 0.7734919286321156, | |
| "grad_norm": 0.5409005284309387, | |
| "learning_rate": 7.217739570814337e-05, | |
| "loss": 3.023, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 0.7748513169073916, | |
| "grad_norm": 0.6478193402290344, | |
| "learning_rate": 7.207663684536814e-05, | |
| "loss": 3.0739, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.7762107051826678, | |
| "grad_norm": 0.646008312702179, | |
| "learning_rate": 7.197576649715771e-05, | |
| "loss": 3.0265, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 0.7775700934579439, | |
| "grad_norm": 0.6225748658180237, | |
| "learning_rate": 7.187478517290014e-05, | |
| "loss": 3.0443, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 0.7789294817332201, | |
| "grad_norm": 0.6438111066818237, | |
| "learning_rate": 7.177369338254385e-05, | |
| "loss": 3.0836, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 0.7802888700084962, | |
| "grad_norm": 0.6899060010910034, | |
| "learning_rate": 7.167249163659518e-05, | |
| "loss": 3.0378, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 0.7816482582837723, | |
| "grad_norm": 0.5617516040802002, | |
| "learning_rate": 7.157118044611569e-05, | |
| "loss": 3.0376, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.7830076465590484, | |
| "grad_norm": 0.6777760982513428, | |
| "learning_rate": 7.146976032271961e-05, | |
| "loss": 3.0566, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.7843670348343246, | |
| "grad_norm": 0.5246532559394836, | |
| "learning_rate": 7.136823177857132e-05, | |
| "loss": 3.0416, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 0.7857264231096007, | |
| "grad_norm": 0.5032817125320435, | |
| "learning_rate": 7.126659532638272e-05, | |
| "loss": 3.0473, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 0.7870858113848768, | |
| "grad_norm": 0.6064226031303406, | |
| "learning_rate": 7.116485147941059e-05, | |
| "loss": 2.974, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 0.7884451996601529, | |
| "grad_norm": 0.6149367690086365, | |
| "learning_rate": 7.106300075145408e-05, | |
| "loss": 3.0362, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.7898045879354291, | |
| "grad_norm": 0.5678831934928894, | |
| "learning_rate": 7.096104365685206e-05, | |
| "loss": 3.0259, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 0.7911639762107052, | |
| "grad_norm": 0.560962975025177, | |
| "learning_rate": 7.085898071048056e-05, | |
| "loss": 3.0255, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 0.7925233644859813, | |
| "grad_norm": 0.5633766651153564, | |
| "learning_rate": 7.075681242775017e-05, | |
| "loss": 2.9758, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 0.7938827527612574, | |
| "grad_norm": 0.5855138301849365, | |
| "learning_rate": 7.065453932460337e-05, | |
| "loss": 3.0208, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 0.7952421410365336, | |
| "grad_norm": 0.6067277789115906, | |
| "learning_rate": 7.055216191751204e-05, | |
| "loss": 3.0361, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.7966015293118097, | |
| "grad_norm": 0.5953718423843384, | |
| "learning_rate": 7.044968072347473e-05, | |
| "loss": 3.0646, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 0.7979609175870858, | |
| "grad_norm": 0.648472785949707, | |
| "learning_rate": 7.034709626001416e-05, | |
| "loss": 3.0444, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 0.7993203058623619, | |
| "grad_norm": 0.5768356919288635, | |
| "learning_rate": 7.024440904517448e-05, | |
| "loss": 3.0244, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 0.8006796941376381, | |
| "grad_norm": 0.6846725940704346, | |
| "learning_rate": 7.014161959751882e-05, | |
| "loss": 3.0442, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 0.8020390824129142, | |
| "grad_norm": 0.6223293542861938, | |
| "learning_rate": 7.00387284361265e-05, | |
| "loss": 3.0533, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.8033984706881904, | |
| "grad_norm": 0.6202958822250366, | |
| "learning_rate": 6.993573608059052e-05, | |
| "loss": 3.0495, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 0.8047578589634664, | |
| "grad_norm": 0.5777451992034912, | |
| "learning_rate": 6.983264305101491e-05, | |
| "loss": 3.0278, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 0.8061172472387426, | |
| "grad_norm": 0.6108320355415344, | |
| "learning_rate": 6.972944986801209e-05, | |
| "loss": 2.8924, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 0.8074766355140187, | |
| "grad_norm": 0.5958619713783264, | |
| "learning_rate": 6.962615705270023e-05, | |
| "loss": 3.0179, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 0.8088360237892949, | |
| "grad_norm": 0.5943612456321716, | |
| "learning_rate": 6.952276512670065e-05, | |
| "loss": 3.0178, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.8101954120645709, | |
| "grad_norm": 0.6541878581047058, | |
| "learning_rate": 6.941927461213518e-05, | |
| "loss": 3.0324, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 0.8115548003398471, | |
| "grad_norm": 0.7149216532707214, | |
| "learning_rate": 6.931568603162351e-05, | |
| "loss": 3.0708, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 0.8129141886151232, | |
| "grad_norm": 0.6609304547309875, | |
| "learning_rate": 6.921199990828055e-05, | |
| "loss": 3.0639, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 0.8142735768903994, | |
| "grad_norm": 0.661949872970581, | |
| "learning_rate": 6.910821676571381e-05, | |
| "loss": 2.9924, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 0.8156329651656754, | |
| "grad_norm": 0.5939015746116638, | |
| "learning_rate": 6.90043371280207e-05, | |
| "loss": 3.0055, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.8169923534409516, | |
| "grad_norm": 0.5653124451637268, | |
| "learning_rate": 6.890036151978598e-05, | |
| "loss": 2.9889, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 0.8183517417162277, | |
| "grad_norm": 0.5646810531616211, | |
| "learning_rate": 6.879629046607903e-05, | |
| "loss": 3.0132, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 0.8197111299915039, | |
| "grad_norm": 0.551462709903717, | |
| "learning_rate": 6.869212449245118e-05, | |
| "loss": 2.9411, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 0.8210705182667799, | |
| "grad_norm": 0.6366941332817078, | |
| "learning_rate": 6.858786412493317e-05, | |
| "loss": 2.9909, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 0.822429906542056, | |
| "grad_norm": 0.5554643869400024, | |
| "learning_rate": 6.848350989003237e-05, | |
| "loss": 2.9887, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.8237892948173322, | |
| "grad_norm": 0.5416108965873718, | |
| "learning_rate": 6.837906231473023e-05, | |
| "loss": 2.9745, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 0.8251486830926084, | |
| "grad_norm": 0.6187860369682312, | |
| "learning_rate": 6.82745219264795e-05, | |
| "loss": 2.9978, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 0.8265080713678844, | |
| "grad_norm": 0.5307159423828125, | |
| "learning_rate": 6.816988925320162e-05, | |
| "loss": 3.0299, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.8278674596431606, | |
| "grad_norm": 0.6538923382759094, | |
| "learning_rate": 6.806516482328418e-05, | |
| "loss": 3.0287, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 0.8292268479184367, | |
| "grad_norm": 0.6692806482315063, | |
| "learning_rate": 6.796034916557797e-05, | |
| "loss": 3.0234, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.8305862361937129, | |
| "grad_norm": 0.5698544383049011, | |
| "learning_rate": 6.78554428093946e-05, | |
| "loss": 3.0521, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 0.831945624468989, | |
| "grad_norm": 0.5961766839027405, | |
| "learning_rate": 6.775044628450366e-05, | |
| "loss": 2.974, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 0.833305012744265, | |
| "grad_norm": 0.8825117349624634, | |
| "learning_rate": 6.764536012113005e-05, | |
| "loss": 3.0019, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 0.8346644010195412, | |
| "grad_norm": 0.6309018135070801, | |
| "learning_rate": 6.754018484995142e-05, | |
| "loss": 3.0101, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 0.8360237892948174, | |
| "grad_norm": 0.6422116160392761, | |
| "learning_rate": 6.74349210020953e-05, | |
| "loss": 2.9641, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.8373831775700935, | |
| "grad_norm": 0.7694607377052307, | |
| "learning_rate": 6.732956910913661e-05, | |
| "loss": 2.999, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 0.8387425658453695, | |
| "grad_norm": 0.6966648697853088, | |
| "learning_rate": 6.722412970309488e-05, | |
| "loss": 3.0015, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 0.8401019541206457, | |
| "grad_norm": 0.6441610455513, | |
| "learning_rate": 6.711860331643154e-05, | |
| "loss": 3.0141, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 0.8414613423959219, | |
| "grad_norm": 0.7339099645614624, | |
| "learning_rate": 6.70129904820473e-05, | |
| "loss": 3.0003, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 0.842820730671198, | |
| "grad_norm": 0.7242470383644104, | |
| "learning_rate": 6.690729173327938e-05, | |
| "loss": 3.0713, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.844180118946474, | |
| "grad_norm": 0.6477665901184082, | |
| "learning_rate": 6.680150760389894e-05, | |
| "loss": 3.0893, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 0.8455395072217502, | |
| "grad_norm": 0.5959832072257996, | |
| "learning_rate": 6.669563862810825e-05, | |
| "loss": 3.0074, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 0.8468988954970263, | |
| "grad_norm": 0.6699416637420654, | |
| "learning_rate": 6.6589685340538e-05, | |
| "loss": 2.94, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 0.8482582837723025, | |
| "grad_norm": 0.6104918122291565, | |
| "learning_rate": 6.648364827624477e-05, | |
| "loss": 2.9702, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 0.8496176720475785, | |
| "grad_norm": 0.6065095663070679, | |
| "learning_rate": 6.63775279707081e-05, | |
| "loss": 3.0213, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.8509770603228547, | |
| "grad_norm": 0.5962669253349304, | |
| "learning_rate": 6.627132495982797e-05, | |
| "loss": 2.9429, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 0.8523364485981308, | |
| "grad_norm": 0.5762045383453369, | |
| "learning_rate": 6.616503977992197e-05, | |
| "loss": 2.9507, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 0.853695836873407, | |
| "grad_norm": 0.5952661633491516, | |
| "learning_rate": 6.605867296772261e-05, | |
| "loss": 3.0323, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 0.855055225148683, | |
| "grad_norm": 0.5376638174057007, | |
| "learning_rate": 6.595222506037472e-05, | |
| "loss": 3.0315, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 0.8564146134239592, | |
| "grad_norm": 0.6530884504318237, | |
| "learning_rate": 6.58456965954326e-05, | |
| "loss": 2.9814, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.8577740016992353, | |
| "grad_norm": 0.7119038701057434, | |
| "learning_rate": 6.573908811085734e-05, | |
| "loss": 3.0361, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 0.8591333899745115, | |
| "grad_norm": 0.6418982744216919, | |
| "learning_rate": 6.56324001450142e-05, | |
| "loss": 2.9703, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 0.8604927782497876, | |
| "grad_norm": 0.5768359303474426, | |
| "learning_rate": 6.552563323666973e-05, | |
| "loss": 3.0508, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 0.8618521665250637, | |
| "grad_norm": 0.5932447910308838, | |
| "learning_rate": 6.541878792498919e-05, | |
| "loss": 3.0444, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 0.8632115548003398, | |
| "grad_norm": 0.5155834555625916, | |
| "learning_rate": 6.531186474953375e-05, | |
| "loss": 2.9336, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.864570943075616, | |
| "grad_norm": 0.4931122064590454, | |
| "learning_rate": 6.520486425025778e-05, | |
| "loss": 3.0018, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 0.8659303313508921, | |
| "grad_norm": 0.629960834980011, | |
| "learning_rate": 6.509778696750614e-05, | |
| "loss": 3.1188, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 0.8672897196261682, | |
| "grad_norm": 0.5573826432228088, | |
| "learning_rate": 6.499063344201146e-05, | |
| "loss": 2.9807, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 0.8686491079014443, | |
| "grad_norm": 0.5982751250267029, | |
| "learning_rate": 6.488340421489136e-05, | |
| "loss": 3.0134, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 0.8700084961767205, | |
| "grad_norm": 0.5728147029876709, | |
| "learning_rate": 6.477609982764575e-05, | |
| "loss": 3.0718, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.8713678844519966, | |
| "grad_norm": 0.5682603120803833, | |
| "learning_rate": 6.46687208221541e-05, | |
| "loss": 3.044, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 0.8727272727272727, | |
| "grad_norm": 0.5737953186035156, | |
| "learning_rate": 6.45612677406727e-05, | |
| "loss": 3.013, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 0.8740866610025488, | |
| "grad_norm": 0.6498506665229797, | |
| "learning_rate": 6.445374112583196e-05, | |
| "loss": 3.0197, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 0.875446049277825, | |
| "grad_norm": 0.6554915308952332, | |
| "learning_rate": 6.434614152063352e-05, | |
| "loss": 2.9238, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 0.8768054375531011, | |
| "grad_norm": 0.6168110370635986, | |
| "learning_rate": 6.423846946844771e-05, | |
| "loss": 2.979, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.8781648258283772, | |
| "grad_norm": 0.7791323661804199, | |
| "learning_rate": 6.41307255130107e-05, | |
| "loss": 2.9561, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 0.8795242141036533, | |
| "grad_norm": 0.6808146834373474, | |
| "learning_rate": 6.402291019842171e-05, | |
| "loss": 3.0073, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 0.8808836023789295, | |
| "grad_norm": 0.6812180280685425, | |
| "learning_rate": 6.391502406914039e-05, | |
| "loss": 2.9763, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 0.8822429906542056, | |
| "grad_norm": 0.5727677345275879, | |
| "learning_rate": 6.380706766998395e-05, | |
| "loss": 2.9236, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 0.8836023789294817, | |
| "grad_norm": 0.7085919380187988, | |
| "learning_rate": 6.369904154612448e-05, | |
| "loss": 3.0897, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.8849617672047578, | |
| "grad_norm": 0.7252781391143799, | |
| "learning_rate": 6.35909462430862e-05, | |
| "loss": 2.991, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 0.886321155480034, | |
| "grad_norm": 0.5278810858726501, | |
| "learning_rate": 6.348278230674258e-05, | |
| "loss": 2.9693, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 0.8876805437553101, | |
| "grad_norm": 0.6835475564002991, | |
| "learning_rate": 6.337455028331382e-05, | |
| "loss": 3.0092, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 0.8890399320305863, | |
| "grad_norm": 0.6431845426559448, | |
| "learning_rate": 6.326625071936388e-05, | |
| "loss": 2.9966, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 0.8903993203058623, | |
| "grad_norm": 0.5237377882003784, | |
| "learning_rate": 6.315788416179775e-05, | |
| "loss": 2.8934, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.8917587085811385, | |
| "grad_norm": 0.5790330171585083, | |
| "learning_rate": 6.304945115785885e-05, | |
| "loss": 3.0384, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 0.8931180968564146, | |
| "grad_norm": 0.5629638433456421, | |
| "learning_rate": 6.294095225512603e-05, | |
| "loss": 2.9331, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 0.8944774851316908, | |
| "grad_norm": 0.5486363768577576, | |
| "learning_rate": 6.283238800151103e-05, | |
| "loss": 2.9491, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 0.8958368734069668, | |
| "grad_norm": 0.5565508008003235, | |
| "learning_rate": 6.272375894525553e-05, | |
| "loss": 2.951, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 0.897196261682243, | |
| "grad_norm": 0.609849214553833, | |
| "learning_rate": 6.261506563492848e-05, | |
| "loss": 2.9716, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.8985556499575191, | |
| "grad_norm": 0.6158019304275513, | |
| "learning_rate": 6.250630861942333e-05, | |
| "loss": 2.8781, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 0.8999150382327953, | |
| "grad_norm": 0.5898751020431519, | |
| "learning_rate": 6.239748844795521e-05, | |
| "loss": 2.9243, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 0.9012744265080713, | |
| "grad_norm": 0.5750930905342102, | |
| "learning_rate": 6.228860567005819e-05, | |
| "loss": 2.9699, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 0.9026338147833475, | |
| "grad_norm": 0.571783185005188, | |
| "learning_rate": 6.21796608355825e-05, | |
| "loss": 2.9596, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 0.9039932030586236, | |
| "grad_norm": 0.5380290746688843, | |
| "learning_rate": 6.207065449469178e-05, | |
| "loss": 3.0468, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.9053525913338998, | |
| "grad_norm": 0.5547913908958435, | |
| "learning_rate": 6.196158719786021e-05, | |
| "loss": 2.9804, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 0.9067119796091758, | |
| "grad_norm": 0.6021085381507874, | |
| "learning_rate": 6.185245949586986e-05, | |
| "loss": 2.9849, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 0.908071367884452, | |
| "grad_norm": 0.5301669239997864, | |
| "learning_rate": 6.174327193980778e-05, | |
| "loss": 2.9676, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 0.9094307561597281, | |
| "grad_norm": 0.5973559617996216, | |
| "learning_rate": 6.163402508106334e-05, | |
| "loss": 2.9417, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 0.9107901444350043, | |
| "grad_norm": 0.6148339509963989, | |
| "learning_rate": 6.152471947132532e-05, | |
| "loss": 2.951, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.9121495327102803, | |
| "grad_norm": 0.6172875165939331, | |
| "learning_rate": 6.141535566257926e-05, | |
| "loss": 2.988, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 0.9135089209855565, | |
| "grad_norm": 0.6237571835517883, | |
| "learning_rate": 6.130593420710452e-05, | |
| "loss": 3.0442, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 0.9148683092608326, | |
| "grad_norm": 0.6334381103515625, | |
| "learning_rate": 6.119645565747165e-05, | |
| "loss": 2.993, | |
| "step": 1346 | |
| }, | |
| { | |
| "epoch": 0.9162276975361088, | |
| "grad_norm": 0.5856964588165283, | |
| "learning_rate": 6.108692056653948e-05, | |
| "loss": 2.9579, | |
| "step": 1348 | |
| }, | |
| { | |
| "epoch": 0.9175870858113849, | |
| "grad_norm": 0.634194552898407, | |
| "learning_rate": 6.097732948745235e-05, | |
| "loss": 2.8742, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.918946474086661, | |
| "grad_norm": 0.6679513454437256, | |
| "learning_rate": 6.0867682973637394e-05, | |
| "loss": 3.0189, | |
| "step": 1352 | |
| }, | |
| { | |
| "epoch": 0.9203058623619371, | |
| "grad_norm": 0.6384778022766113, | |
| "learning_rate": 6.075798157880164e-05, | |
| "loss": 3.0204, | |
| "step": 1354 | |
| }, | |
| { | |
| "epoch": 0.9216652506372133, | |
| "grad_norm": 0.532374382019043, | |
| "learning_rate": 6.0648225856929275e-05, | |
| "loss": 2.996, | |
| "step": 1356 | |
| }, | |
| { | |
| "epoch": 0.9230246389124894, | |
| "grad_norm": 0.5407797694206238, | |
| "learning_rate": 6.0538416362278824e-05, | |
| "loss": 2.9277, | |
| "step": 1358 | |
| }, | |
| { | |
| "epoch": 0.9243840271877655, | |
| "grad_norm": 0.6642601490020752, | |
| "learning_rate": 6.0428553649380415e-05, | |
| "loss": 2.989, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.9257434154630416, | |
| "grad_norm": 0.6345245242118835, | |
| "learning_rate": 6.031863827303284e-05, | |
| "loss": 2.9936, | |
| "step": 1362 | |
| }, | |
| { | |
| "epoch": 0.9271028037383178, | |
| "grad_norm": 0.5593590140342712, | |
| "learning_rate": 6.020867078830089e-05, | |
| "loss": 2.9882, | |
| "step": 1364 | |
| }, | |
| { | |
| "epoch": 0.9284621920135939, | |
| "grad_norm": 0.5335503220558167, | |
| "learning_rate": 6.009865175051248e-05, | |
| "loss": 2.963, | |
| "step": 1366 | |
| }, | |
| { | |
| "epoch": 0.92982158028887, | |
| "grad_norm": 0.6434462666511536, | |
| "learning_rate": 5.9988581715255876e-05, | |
| "loss": 2.9936, | |
| "step": 1368 | |
| }, | |
| { | |
| "epoch": 0.9311809685641461, | |
| "grad_norm": 0.583660900592804, | |
| "learning_rate": 5.9878461238376904e-05, | |
| "loss": 2.9092, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.9325403568394223, | |
| "grad_norm": 0.7153940200805664, | |
| "learning_rate": 5.976829087597605e-05, | |
| "loss": 2.9744, | |
| "step": 1372 | |
| }, | |
| { | |
| "epoch": 0.9338997451146984, | |
| "grad_norm": 1.2551978826522827, | |
| "learning_rate": 5.965807118440576e-05, | |
| "loss": 2.9742, | |
| "step": 1374 | |
| }, | |
| { | |
| "epoch": 0.9352591333899745, | |
| "grad_norm": 0.6354015469551086, | |
| "learning_rate": 5.954780272026761e-05, | |
| "loss": 2.9045, | |
| "step": 1376 | |
| }, | |
| { | |
| "epoch": 0.9366185216652506, | |
| "grad_norm": 0.6488978266716003, | |
| "learning_rate": 5.94374860404094e-05, | |
| "loss": 2.9853, | |
| "step": 1378 | |
| }, | |
| { | |
| "epoch": 0.9379779099405268, | |
| "grad_norm": 0.5952944755554199, | |
| "learning_rate": 5.9327121701922516e-05, | |
| "loss": 2.9655, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.9393372982158029, | |
| "grad_norm": 0.6394802331924438, | |
| "learning_rate": 5.921671026213893e-05, | |
| "loss": 2.9707, | |
| "step": 1382 | |
| }, | |
| { | |
| "epoch": 0.940696686491079, | |
| "grad_norm": 0.7093409299850464, | |
| "learning_rate": 5.91062522786285e-05, | |
| "loss": 2.9166, | |
| "step": 1384 | |
| }, | |
| { | |
| "epoch": 0.9420560747663551, | |
| "grad_norm": 0.5998436808586121, | |
| "learning_rate": 5.8995748309196184e-05, | |
| "loss": 2.983, | |
| "step": 1386 | |
| }, | |
| { | |
| "epoch": 0.9434154630416313, | |
| "grad_norm": 0.6887815594673157, | |
| "learning_rate": 5.888519891187906e-05, | |
| "loss": 2.9567, | |
| "step": 1388 | |
| }, | |
| { | |
| "epoch": 0.9447748513169074, | |
| "grad_norm": 0.659731388092041, | |
| "learning_rate": 5.877460464494369e-05, | |
| "loss": 2.8726, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.9461342395921836, | |
| "grad_norm": 0.7009007334709167, | |
| "learning_rate": 5.8663966066883205e-05, | |
| "loss": 2.9669, | |
| "step": 1392 | |
| }, | |
| { | |
| "epoch": 0.9474936278674596, | |
| "grad_norm": 0.7139768004417419, | |
| "learning_rate": 5.855328373641449e-05, | |
| "loss": 2.9808, | |
| "step": 1394 | |
| }, | |
| { | |
| "epoch": 0.9488530161427358, | |
| "grad_norm": 0.6618505120277405, | |
| "learning_rate": 5.8442558212475416e-05, | |
| "loss": 2.9656, | |
| "step": 1396 | |
| }, | |
| { | |
| "epoch": 0.9502124044180119, | |
| "grad_norm": 0.5960121750831604, | |
| "learning_rate": 5.83317900542219e-05, | |
| "loss": 3.0878, | |
| "step": 1398 | |
| }, | |
| { | |
| "epoch": 0.9515717926932881, | |
| "grad_norm": 0.6161683797836304, | |
| "learning_rate": 5.8220979821025254e-05, | |
| "loss": 2.9296, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.9529311809685641, | |
| "grad_norm": 0.6301623582839966, | |
| "learning_rate": 5.8110128072469206e-05, | |
| "loss": 2.9037, | |
| "step": 1402 | |
| }, | |
| { | |
| "epoch": 0.9542905692438403, | |
| "grad_norm": 0.6042844653129578, | |
| "learning_rate": 5.799923536834715e-05, | |
| "loss": 3.0262, | |
| "step": 1404 | |
| }, | |
| { | |
| "epoch": 0.9556499575191164, | |
| "grad_norm": 0.5644305348396301, | |
| "learning_rate": 5.7888302268659286e-05, | |
| "loss": 2.9875, | |
| "step": 1406 | |
| }, | |
| { | |
| "epoch": 0.9570093457943926, | |
| "grad_norm": 0.567065417766571, | |
| "learning_rate": 5.7777329333609855e-05, | |
| "loss": 3.0178, | |
| "step": 1408 | |
| }, | |
| { | |
| "epoch": 0.9583687340696686, | |
| "grad_norm": 0.5456449389457703, | |
| "learning_rate": 5.76663171236042e-05, | |
| "loss": 2.9099, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.9597281223449448, | |
| "grad_norm": 0.5395920276641846, | |
| "learning_rate": 5.755526619924605e-05, | |
| "loss": 2.9196, | |
| "step": 1412 | |
| }, | |
| { | |
| "epoch": 0.9610875106202209, | |
| "grad_norm": 0.533607542514801, | |
| "learning_rate": 5.744417712133462e-05, | |
| "loss": 2.9564, | |
| "step": 1414 | |
| }, | |
| { | |
| "epoch": 0.9624468988954971, | |
| "grad_norm": 0.5302989482879639, | |
| "learning_rate": 5.733305045086179e-05, | |
| "loss": 2.9872, | |
| "step": 1416 | |
| }, | |
| { | |
| "epoch": 0.9638062871707731, | |
| "grad_norm": 0.6003543138504028, | |
| "learning_rate": 5.722188674900929e-05, | |
| "loss": 2.9267, | |
| "step": 1418 | |
| }, | |
| { | |
| "epoch": 0.9651656754460493, | |
| "grad_norm": 0.6156508326530457, | |
| "learning_rate": 5.7110686577145865e-05, | |
| "loss": 2.9157, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.9665250637213254, | |
| "grad_norm": 0.5998055934906006, | |
| "learning_rate": 5.6999450496824416e-05, | |
| "loss": 2.9314, | |
| "step": 1422 | |
| }, | |
| { | |
| "epoch": 0.9678844519966016, | |
| "grad_norm": 0.6056944131851196, | |
| "learning_rate": 5.688817906977917e-05, | |
| "loss": 2.9823, | |
| "step": 1424 | |
| }, | |
| { | |
| "epoch": 0.9692438402718777, | |
| "grad_norm": 0.5981131196022034, | |
| "learning_rate": 5.677687285792288e-05, | |
| "loss": 2.9677, | |
| "step": 1426 | |
| }, | |
| { | |
| "epoch": 0.9706032285471538, | |
| "grad_norm": 0.5991801023483276, | |
| "learning_rate": 5.666553242334394e-05, | |
| "loss": 2.984, | |
| "step": 1428 | |
| }, | |
| { | |
| "epoch": 0.9719626168224299, | |
| "grad_norm": 0.5065740942955017, | |
| "learning_rate": 5.655415832830357e-05, | |
| "loss": 2.9601, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.9733220050977061, | |
| "grad_norm": 0.5771386027336121, | |
| "learning_rate": 5.644275113523297e-05, | |
| "loss": 2.9288, | |
| "step": 1432 | |
| }, | |
| { | |
| "epoch": 0.9746813933729822, | |
| "grad_norm": 0.5725680589675903, | |
| "learning_rate": 5.63313114067305e-05, | |
| "loss": 2.9348, | |
| "step": 1434 | |
| }, | |
| { | |
| "epoch": 0.9760407816482582, | |
| "grad_norm": 0.5777420997619629, | |
| "learning_rate": 5.621983970555881e-05, | |
| "loss": 2.9346, | |
| "step": 1436 | |
| }, | |
| { | |
| "epoch": 0.9774001699235344, | |
| "grad_norm": 0.5346077680587769, | |
| "learning_rate": 5.6108336594641996e-05, | |
| "loss": 3.0165, | |
| "step": 1438 | |
| }, | |
| { | |
| "epoch": 0.9787595581988106, | |
| "grad_norm": 0.6149086952209473, | |
| "learning_rate": 5.599680263706278e-05, | |
| "loss": 2.9962, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.9801189464740867, | |
| "grad_norm": 0.5737910866737366, | |
| "learning_rate": 5.588523839605968e-05, | |
| "loss": 2.9621, | |
| "step": 1442 | |
| }, | |
| { | |
| "epoch": 0.9814783347493627, | |
| "grad_norm": 0.5987441539764404, | |
| "learning_rate": 5.577364443502412e-05, | |
| "loss": 3.0176, | |
| "step": 1444 | |
| }, | |
| { | |
| "epoch": 0.9828377230246389, | |
| "grad_norm": 0.5962843894958496, | |
| "learning_rate": 5.56620213174976e-05, | |
| "loss": 2.9302, | |
| "step": 1446 | |
| }, | |
| { | |
| "epoch": 0.984197111299915, | |
| "grad_norm": 0.5651360154151917, | |
| "learning_rate": 5.5550369607168874e-05, | |
| "loss": 2.9685, | |
| "step": 1448 | |
| }, | |
| { | |
| "epoch": 0.9855564995751912, | |
| "grad_norm": 0.5286944508552551, | |
| "learning_rate": 5.543868986787109e-05, | |
| "loss": 2.9918, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.9869158878504672, | |
| "grad_norm": 0.5632530450820923, | |
| "learning_rate": 5.532698266357892e-05, | |
| "loss": 2.9934, | |
| "step": 1452 | |
| }, | |
| { | |
| "epoch": 0.9882752761257434, | |
| "grad_norm": 0.6063055396080017, | |
| "learning_rate": 5.521524855840578e-05, | |
| "loss": 2.9871, | |
| "step": 1454 | |
| }, | |
| { | |
| "epoch": 0.9896346644010195, | |
| "grad_norm": 0.5442188382148743, | |
| "learning_rate": 5.510348811660084e-05, | |
| "loss": 2.9771, | |
| "step": 1456 | |
| }, | |
| { | |
| "epoch": 0.9909940526762957, | |
| "grad_norm": 0.5566977262496948, | |
| "learning_rate": 5.499170190254641e-05, | |
| "loss": 2.875, | |
| "step": 1458 | |
| }, | |
| { | |
| "epoch": 0.9923534409515717, | |
| "grad_norm": 0.521381139755249, | |
| "learning_rate": 5.4879890480754795e-05, | |
| "loss": 2.9807, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.9937128292268479, | |
| "grad_norm": 0.5293747186660767, | |
| "learning_rate": 5.476805441586569e-05, | |
| "loss": 2.9315, | |
| "step": 1462 | |
| }, | |
| { | |
| "epoch": 0.995072217502124, | |
| "grad_norm": 0.5370301008224487, | |
| "learning_rate": 5.465619427264323e-05, | |
| "loss": 2.9827, | |
| "step": 1464 | |
| }, | |
| { | |
| "epoch": 0.9964316057774002, | |
| "grad_norm": 0.5640749335289001, | |
| "learning_rate": 5.454431061597311e-05, | |
| "loss": 2.9615, | |
| "step": 1466 | |
| }, | |
| { | |
| "epoch": 0.9977909940526763, | |
| "grad_norm": 0.5775633454322815, | |
| "learning_rate": 5.4432404010859804e-05, | |
| "loss": 2.9576, | |
| "step": 1468 | |
| }, | |
| { | |
| "epoch": 0.9991503823279524, | |
| "grad_norm": 0.5625415444374084, | |
| "learning_rate": 5.4320475022423647e-05, | |
| "loss": 2.8864, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.88974928855896, | |
| "learning_rate": 5.4208524215897985e-05, | |
| "loss": 3.0896, | |
| "step": 1472 | |
| }, | |
| { | |
| "epoch": 1.001359388275276, | |
| "grad_norm": 0.7388569712638855, | |
| "learning_rate": 5.409655215662642e-05, | |
| "loss": 2.8507, | |
| "step": 1474 | |
| }, | |
| { | |
| "epoch": 1.0027187765505523, | |
| "grad_norm": 0.6877476572990417, | |
| "learning_rate": 5.3984559410059796e-05, | |
| "loss": 2.9142, | |
| "step": 1476 | |
| }, | |
| { | |
| "epoch": 1.0040781648258283, | |
| "grad_norm": 0.6670885682106018, | |
| "learning_rate": 5.38725465417535e-05, | |
| "loss": 2.9301, | |
| "step": 1478 | |
| }, | |
| { | |
| "epoch": 1.0054375531011046, | |
| "grad_norm": 0.6211088299751282, | |
| "learning_rate": 5.376051411736447e-05, | |
| "loss": 2.9291, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.0067969413763806, | |
| "grad_norm": 0.630263090133667, | |
| "learning_rate": 5.364846270264842e-05, | |
| "loss": 2.9646, | |
| "step": 1482 | |
| }, | |
| { | |
| "epoch": 1.0081563296516567, | |
| "grad_norm": 0.5889977812767029, | |
| "learning_rate": 5.353639286345699e-05, | |
| "loss": 2.946, | |
| "step": 1484 | |
| }, | |
| { | |
| "epoch": 1.009515717926933, | |
| "grad_norm": 0.5902710556983948, | |
| "learning_rate": 5.342430516573485e-05, | |
| "loss": 2.8681, | |
| "step": 1486 | |
| }, | |
| { | |
| "epoch": 1.010875106202209, | |
| "grad_norm": 0.5864720344543457, | |
| "learning_rate": 5.3312200175516815e-05, | |
| "loss": 2.9029, | |
| "step": 1488 | |
| }, | |
| { | |
| "epoch": 1.012234494477485, | |
| "grad_norm": 0.5937751531600952, | |
| "learning_rate": 5.320007845892509e-05, | |
| "loss": 2.9837, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.0135938827527613, | |
| "grad_norm": 0.6514058113098145, | |
| "learning_rate": 5.3087940582166287e-05, | |
| "loss": 2.8821, | |
| "step": 1492 | |
| }, | |
| { | |
| "epoch": 1.0149532710280373, | |
| "grad_norm": 0.5811493992805481, | |
| "learning_rate": 5.297578711152867e-05, | |
| "loss": 2.945, | |
| "step": 1494 | |
| }, | |
| { | |
| "epoch": 1.0163126593033136, | |
| "grad_norm": 0.5168861150741577, | |
| "learning_rate": 5.286361861337924e-05, | |
| "loss": 2.9208, | |
| "step": 1496 | |
| }, | |
| { | |
| "epoch": 1.0176720475785896, | |
| "grad_norm": 0.48943331837654114, | |
| "learning_rate": 5.275143565416086e-05, | |
| "loss": 2.9156, | |
| "step": 1498 | |
| }, | |
| { | |
| "epoch": 1.0190314358538657, | |
| "grad_norm": 0.5392592549324036, | |
| "learning_rate": 5.2639238800389465e-05, | |
| "loss": 2.9327, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.020390824129142, | |
| "grad_norm": 0.5025687217712402, | |
| "learning_rate": 5.2527028618651117e-05, | |
| "loss": 2.8723, | |
| "step": 1502 | |
| }, | |
| { | |
| "epoch": 1.021750212404418, | |
| "grad_norm": 0.5709313750267029, | |
| "learning_rate": 5.2414805675599197e-05, | |
| "loss": 2.9078, | |
| "step": 1504 | |
| }, | |
| { | |
| "epoch": 1.023109600679694, | |
| "grad_norm": 0.48659902811050415, | |
| "learning_rate": 5.230257053795155e-05, | |
| "loss": 2.9351, | |
| "step": 1506 | |
| }, | |
| { | |
| "epoch": 1.0244689889549703, | |
| "grad_norm": 0.5249725580215454, | |
| "learning_rate": 5.219032377248756e-05, | |
| "loss": 2.8758, | |
| "step": 1508 | |
| }, | |
| { | |
| "epoch": 1.0258283772302463, | |
| "grad_norm": 0.5494433045387268, | |
| "learning_rate": 5.207806594604536e-05, | |
| "loss": 2.8805, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.0271877655055226, | |
| "grad_norm": 0.5198817849159241, | |
| "learning_rate": 5.196579762551895e-05, | |
| "loss": 2.883, | |
| "step": 1512 | |
| }, | |
| { | |
| "epoch": 1.0285471537807986, | |
| "grad_norm": 0.5270273685455322, | |
| "learning_rate": 5.18535193778553e-05, | |
| "loss": 2.9508, | |
| "step": 1514 | |
| }, | |
| { | |
| "epoch": 1.0299065420560747, | |
| "grad_norm": 0.5204840898513794, | |
| "learning_rate": 5.174123177005151e-05, | |
| "loss": 2.8495, | |
| "step": 1516 | |
| }, | |
| { | |
| "epoch": 1.031265930331351, | |
| "grad_norm": 0.534113883972168, | |
| "learning_rate": 5.162893536915197e-05, | |
| "loss": 2.9264, | |
| "step": 1518 | |
| }, | |
| { | |
| "epoch": 1.032625318606627, | |
| "grad_norm": 0.5477264523506165, | |
| "learning_rate": 5.151663074224547e-05, | |
| "loss": 2.9724, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.0339847068819032, | |
| "grad_norm": 0.4859903156757355, | |
| "learning_rate": 5.140431845646233e-05, | |
| "loss": 2.8334, | |
| "step": 1522 | |
| }, | |
| { | |
| "epoch": 1.0353440951571793, | |
| "grad_norm": 0.46718671917915344, | |
| "learning_rate": 5.1291999078971545e-05, | |
| "loss": 2.8148, | |
| "step": 1524 | |
| }, | |
| { | |
| "epoch": 1.0367034834324553, | |
| "grad_norm": 0.5363844037055969, | |
| "learning_rate": 5.117967317697792e-05, | |
| "loss": 2.9378, | |
| "step": 1526 | |
| }, | |
| { | |
| "epoch": 1.0380628717077316, | |
| "grad_norm": 0.5690683126449585, | |
| "learning_rate": 5.1067341317719216e-05, | |
| "loss": 2.8649, | |
| "step": 1528 | |
| }, | |
| { | |
| "epoch": 1.0394222599830076, | |
| "grad_norm": 0.5963205099105835, | |
| "learning_rate": 5.0955004068463295e-05, | |
| "loss": 2.9794, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.0407816482582837, | |
| "grad_norm": 0.5808899998664856, | |
| "learning_rate": 5.084266199650523e-05, | |
| "loss": 2.7899, | |
| "step": 1532 | |
| }, | |
| { | |
| "epoch": 1.04214103653356, | |
| "grad_norm": 0.5939140319824219, | |
| "learning_rate": 5.0730315669164416e-05, | |
| "loss": 2.8426, | |
| "step": 1534 | |
| }, | |
| { | |
| "epoch": 1.043500424808836, | |
| "grad_norm": 0.5920867323875427, | |
| "learning_rate": 5.0617965653781763e-05, | |
| "loss": 2.9197, | |
| "step": 1536 | |
| }, | |
| { | |
| "epoch": 1.0448598130841122, | |
| "grad_norm": 0.5790732502937317, | |
| "learning_rate": 5.050561251771683e-05, | |
| "loss": 2.8707, | |
| "step": 1538 | |
| }, | |
| { | |
| "epoch": 1.0462192013593883, | |
| "grad_norm": 0.4703436493873596, | |
| "learning_rate": 5.0393256828344856e-05, | |
| "loss": 2.9444, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.0475785896346643, | |
| "grad_norm": 0.4434896409511566, | |
| "learning_rate": 5.0280899153054096e-05, | |
| "loss": 2.8934, | |
| "step": 1542 | |
| }, | |
| { | |
| "epoch": 1.0489379779099406, | |
| "grad_norm": 0.5452954769134521, | |
| "learning_rate": 5.0168540059242706e-05, | |
| "loss": 2.8756, | |
| "step": 1544 | |
| }, | |
| { | |
| "epoch": 1.0502973661852166, | |
| "grad_norm": 0.6902345418930054, | |
| "learning_rate": 5.0056180114316084e-05, | |
| "loss": 2.9603, | |
| "step": 1546 | |
| }, | |
| { | |
| "epoch": 1.0516567544604927, | |
| "grad_norm": 0.626602053642273, | |
| "learning_rate": 4.994381988568392e-05, | |
| "loss": 2.9577, | |
| "step": 1548 | |
| }, | |
| { | |
| "epoch": 1.053016142735769, | |
| "grad_norm": 0.604475200176239, | |
| "learning_rate": 4.9831459940757306e-05, | |
| "loss": 2.9578, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.054375531011045, | |
| "grad_norm": 0.5631133913993835, | |
| "learning_rate": 4.9719100846945916e-05, | |
| "loss": 2.9181, | |
| "step": 1552 | |
| }, | |
| { | |
| "epoch": 1.0557349192863212, | |
| "grad_norm": 0.610561192035675, | |
| "learning_rate": 4.9606743171655135e-05, | |
| "loss": 2.9426, | |
| "step": 1554 | |
| }, | |
| { | |
| "epoch": 1.0570943075615973, | |
| "grad_norm": 0.5366804599761963, | |
| "learning_rate": 4.949438748228318e-05, | |
| "loss": 2.8779, | |
| "step": 1556 | |
| }, | |
| { | |
| "epoch": 1.0584536958368733, | |
| "grad_norm": 0.4870561361312866, | |
| "learning_rate": 4.938203434621825e-05, | |
| "loss": 2.8848, | |
| "step": 1558 | |
| }, | |
| { | |
| "epoch": 1.0598130841121496, | |
| "grad_norm": 0.5381836295127869, | |
| "learning_rate": 4.92696843308356e-05, | |
| "loss": 2.936, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.0611724723874256, | |
| "grad_norm": 0.48091086745262146, | |
| "learning_rate": 4.915733800349477e-05, | |
| "loss": 2.8891, | |
| "step": 1562 | |
| }, | |
| { | |
| "epoch": 1.0625318606627019, | |
| "grad_norm": 0.524387001991272, | |
| "learning_rate": 4.90449959315367e-05, | |
| "loss": 2.8654, | |
| "step": 1564 | |
| }, | |
| { | |
| "epoch": 1.063891248937978, | |
| "grad_norm": 0.49348440766334534, | |
| "learning_rate": 4.893265868228079e-05, | |
| "loss": 2.896, | |
| "step": 1566 | |
| }, | |
| { | |
| "epoch": 1.065250637213254, | |
| "grad_norm": 0.4692046642303467, | |
| "learning_rate": 4.882032682302209e-05, | |
| "loss": 2.9152, | |
| "step": 1568 | |
| }, | |
| { | |
| "epoch": 1.0666100254885302, | |
| "grad_norm": 0.4998084306716919, | |
| "learning_rate": 4.870800092102849e-05, | |
| "loss": 2.8765, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.0679694137638063, | |
| "grad_norm": 0.5375500917434692, | |
| "learning_rate": 4.859568154353767e-05, | |
| "loss": 2.906, | |
| "step": 1572 | |
| }, | |
| { | |
| "epoch": 1.0693288020390823, | |
| "grad_norm": 0.5358439683914185, | |
| "learning_rate": 4.848336925775454e-05, | |
| "loss": 2.9159, | |
| "step": 1574 | |
| }, | |
| { | |
| "epoch": 1.0706881903143586, | |
| "grad_norm": 0.5062316060066223, | |
| "learning_rate": 4.837106463084803e-05, | |
| "loss": 2.8846, | |
| "step": 1576 | |
| }, | |
| { | |
| "epoch": 1.0720475785896346, | |
| "grad_norm": 0.5185168385505676, | |
| "learning_rate": 4.825876822994849e-05, | |
| "loss": 2.9401, | |
| "step": 1578 | |
| }, | |
| { | |
| "epoch": 1.0734069668649109, | |
| "grad_norm": 0.5553262233734131, | |
| "learning_rate": 4.8146480622144727e-05, | |
| "loss": 2.8719, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.074766355140187, | |
| "grad_norm": 0.5499855279922485, | |
| "learning_rate": 4.8034202374481046e-05, | |
| "loss": 2.9421, | |
| "step": 1582 | |
| }, | |
| { | |
| "epoch": 1.076125743415463, | |
| "grad_norm": 0.5234200954437256, | |
| "learning_rate": 4.792193405395464e-05, | |
| "loss": 2.9119, | |
| "step": 1584 | |
| }, | |
| { | |
| "epoch": 1.0774851316907392, | |
| "grad_norm": 0.5360822081565857, | |
| "learning_rate": 4.780967622751245e-05, | |
| "loss": 2.8529, | |
| "step": 1586 | |
| }, | |
| { | |
| "epoch": 1.0788445199660153, | |
| "grad_norm": 0.5505430102348328, | |
| "learning_rate": 4.769742946204846e-05, | |
| "loss": 2.8813, | |
| "step": 1588 | |
| }, | |
| { | |
| "epoch": 1.0802039082412915, | |
| "grad_norm": 0.5276104807853699, | |
| "learning_rate": 4.7585194324400815e-05, | |
| "loss": 2.9786, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 1.0815632965165676, | |
| "grad_norm": 0.5166751146316528, | |
| "learning_rate": 4.747297138134888e-05, | |
| "loss": 2.8943, | |
| "step": 1592 | |
| }, | |
| { | |
| "epoch": 1.0829226847918436, | |
| "grad_norm": 0.5125443935394287, | |
| "learning_rate": 4.736076119961054e-05, | |
| "loss": 2.8934, | |
| "step": 1594 | |
| }, | |
| { | |
| "epoch": 1.0842820730671199, | |
| "grad_norm": 0.5239457488059998, | |
| "learning_rate": 4.724856434583915e-05, | |
| "loss": 2.8883, | |
| "step": 1596 | |
| }, | |
| { | |
| "epoch": 1.085641461342396, | |
| "grad_norm": 0.5971853137016296, | |
| "learning_rate": 4.7136381386620775e-05, | |
| "loss": 2.8673, | |
| "step": 1598 | |
| }, | |
| { | |
| "epoch": 1.087000849617672, | |
| "grad_norm": 0.5630030632019043, | |
| "learning_rate": 4.702421288847134e-05, | |
| "loss": 2.8986, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.0883602378929482, | |
| "grad_norm": 0.5520672798156738, | |
| "learning_rate": 4.691205941783371e-05, | |
| "loss": 2.8582, | |
| "step": 1602 | |
| }, | |
| { | |
| "epoch": 1.0897196261682243, | |
| "grad_norm": 0.516333281993866, | |
| "learning_rate": 4.679992154107492e-05, | |
| "loss": 2.8671, | |
| "step": 1604 | |
| }, | |
| { | |
| "epoch": 1.0910790144435005, | |
| "grad_norm": 0.5020349621772766, | |
| "learning_rate": 4.6687799824483197e-05, | |
| "loss": 2.886, | |
| "step": 1606 | |
| }, | |
| { | |
| "epoch": 1.0924384027187766, | |
| "grad_norm": 0.4855659306049347, | |
| "learning_rate": 4.657569483426517e-05, | |
| "loss": 2.8972, | |
| "step": 1608 | |
| }, | |
| { | |
| "epoch": 1.0937977909940526, | |
| "grad_norm": 0.4451141357421875, | |
| "learning_rate": 4.646360713654302e-05, | |
| "loss": 2.8747, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 1.0951571792693289, | |
| "grad_norm": 0.5112724304199219, | |
| "learning_rate": 4.635153729735158e-05, | |
| "loss": 2.8718, | |
| "step": 1612 | |
| }, | |
| { | |
| "epoch": 1.096516567544605, | |
| "grad_norm": 0.5099813342094421, | |
| "learning_rate": 4.623948588263553e-05, | |
| "loss": 2.9246, | |
| "step": 1614 | |
| }, | |
| { | |
| "epoch": 1.097875955819881, | |
| "grad_norm": 0.5851842164993286, | |
| "learning_rate": 4.612745345824652e-05, | |
| "loss": 2.8938, | |
| "step": 1616 | |
| }, | |
| { | |
| "epoch": 1.0992353440951572, | |
| "grad_norm": 0.6044837832450867, | |
| "learning_rate": 4.601544058994021e-05, | |
| "loss": 2.8764, | |
| "step": 1618 | |
| }, | |
| { | |
| "epoch": 1.1005947323704333, | |
| "grad_norm": 0.50668865442276, | |
| "learning_rate": 4.5903447843373596e-05, | |
| "loss": 2.9252, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.1019541206457095, | |
| "grad_norm": 0.5305453538894653, | |
| "learning_rate": 4.579147578410201e-05, | |
| "loss": 2.897, | |
| "step": 1622 | |
| }, | |
| { | |
| "epoch": 1.1033135089209856, | |
| "grad_norm": 0.5726643800735474, | |
| "learning_rate": 4.5679524977576365e-05, | |
| "loss": 2.8364, | |
| "step": 1624 | |
| }, | |
| { | |
| "epoch": 1.1046728971962616, | |
| "grad_norm": 0.45511624217033386, | |
| "learning_rate": 4.556759598914021e-05, | |
| "loss": 2.8507, | |
| "step": 1626 | |
| }, | |
| { | |
| "epoch": 1.1060322854715379, | |
| "grad_norm": 0.5291525721549988, | |
| "learning_rate": 4.545568938402689e-05, | |
| "loss": 2.8679, | |
| "step": 1628 | |
| }, | |
| { | |
| "epoch": 1.107391673746814, | |
| "grad_norm": 0.5431672930717468, | |
| "learning_rate": 4.534380572735678e-05, | |
| "loss": 2.922, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 1.10875106202209, | |
| "grad_norm": 0.5468505620956421, | |
| "learning_rate": 4.523194558413431e-05, | |
| "loss": 2.8737, | |
| "step": 1632 | |
| }, | |
| { | |
| "epoch": 1.1101104502973662, | |
| "grad_norm": 0.6105744242668152, | |
| "learning_rate": 4.512010951924521e-05, | |
| "loss": 2.9068, | |
| "step": 1634 | |
| }, | |
| { | |
| "epoch": 1.1114698385726423, | |
| "grad_norm": 0.5395496487617493, | |
| "learning_rate": 4.50082980974536e-05, | |
| "loss": 2.8827, | |
| "step": 1636 | |
| }, | |
| { | |
| "epoch": 1.1128292268479185, | |
| "grad_norm": 0.5439732670783997, | |
| "learning_rate": 4.4896511883399165e-05, | |
| "loss": 2.854, | |
| "step": 1638 | |
| }, | |
| { | |
| "epoch": 1.1141886151231946, | |
| "grad_norm": 0.5708232522010803, | |
| "learning_rate": 4.478475144159425e-05, | |
| "loss": 2.913, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.1155480033984706, | |
| "grad_norm": 0.5542935729026794, | |
| "learning_rate": 4.467301733642109e-05, | |
| "loss": 2.8612, | |
| "step": 1642 | |
| }, | |
| { | |
| "epoch": 1.1169073916737469, | |
| "grad_norm": 0.5303210616111755, | |
| "learning_rate": 4.456131013212892e-05, | |
| "loss": 2.9271, | |
| "step": 1644 | |
| }, | |
| { | |
| "epoch": 1.118266779949023, | |
| "grad_norm": 0.4943715035915375, | |
| "learning_rate": 4.444963039283114e-05, | |
| "loss": 2.9189, | |
| "step": 1646 | |
| }, | |
| { | |
| "epoch": 1.1196261682242992, | |
| "grad_norm": 0.5043460130691528, | |
| "learning_rate": 4.433797868250242e-05, | |
| "loss": 2.8599, | |
| "step": 1648 | |
| }, | |
| { | |
| "epoch": 1.1209855564995752, | |
| "grad_norm": 0.5098512172698975, | |
| "learning_rate": 4.42263555649759e-05, | |
| "loss": 2.8624, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.1223449447748512, | |
| "grad_norm": 0.5372128486633301, | |
| "learning_rate": 4.4114761603940326e-05, | |
| "loss": 2.9188, | |
| "step": 1652 | |
| }, | |
| { | |
| "epoch": 1.1237043330501275, | |
| "grad_norm": 0.5066424012184143, | |
| "learning_rate": 4.4003197362937224e-05, | |
| "loss": 2.8573, | |
| "step": 1654 | |
| }, | |
| { | |
| "epoch": 1.1250637213254036, | |
| "grad_norm": 0.4755190908908844, | |
| "learning_rate": 4.3891663405358016e-05, | |
| "loss": 2.9142, | |
| "step": 1656 | |
| }, | |
| { | |
| "epoch": 1.1264231096006796, | |
| "grad_norm": 0.48332327604293823, | |
| "learning_rate": 4.378016029444121e-05, | |
| "loss": 2.9031, | |
| "step": 1658 | |
| }, | |
| { | |
| "epoch": 1.1277824978759559, | |
| "grad_norm": 0.4979184567928314, | |
| "learning_rate": 4.366868859326952e-05, | |
| "loss": 2.8946, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.129141886151232, | |
| "grad_norm": 0.5440660715103149, | |
| "learning_rate": 4.355724886476704e-05, | |
| "loss": 2.8659, | |
| "step": 1662 | |
| }, | |
| { | |
| "epoch": 1.1305012744265082, | |
| "grad_norm": 0.5121813416481018, | |
| "learning_rate": 4.3445841671696445e-05, | |
| "loss": 2.8856, | |
| "step": 1664 | |
| }, | |
| { | |
| "epoch": 1.1318606627017842, | |
| "grad_norm": 0.49314969778060913, | |
| "learning_rate": 4.333446757665608e-05, | |
| "loss": 2.9192, | |
| "step": 1666 | |
| }, | |
| { | |
| "epoch": 1.1332200509770602, | |
| "grad_norm": 0.5015016794204712, | |
| "learning_rate": 4.3223127142077136e-05, | |
| "loss": 2.7729, | |
| "step": 1668 | |
| }, | |
| { | |
| "epoch": 1.1345794392523365, | |
| "grad_norm": 0.49645712971687317, | |
| "learning_rate": 4.311182093022086e-05, | |
| "loss": 2.793, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 1.1359388275276125, | |
| "grad_norm": 0.5285964012145996, | |
| "learning_rate": 4.30005495031756e-05, | |
| "loss": 2.9351, | |
| "step": 1672 | |
| }, | |
| { | |
| "epoch": 1.1372982158028888, | |
| "grad_norm": 0.5081846714019775, | |
| "learning_rate": 4.2889313422854146e-05, | |
| "loss": 2.9178, | |
| "step": 1674 | |
| }, | |
| { | |
| "epoch": 1.1386576040781649, | |
| "grad_norm": 0.5235109329223633, | |
| "learning_rate": 4.2778113250990714e-05, | |
| "loss": 2.8663, | |
| "step": 1676 | |
| }, | |
| { | |
| "epoch": 1.140016992353441, | |
| "grad_norm": 0.5389015078544617, | |
| "learning_rate": 4.2666949549138226e-05, | |
| "loss": 2.8902, | |
| "step": 1678 | |
| }, | |
| { | |
| "epoch": 1.1413763806287172, | |
| "grad_norm": 0.531001627445221, | |
| "learning_rate": 4.255582287866538e-05, | |
| "loss": 2.9296, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.1427357689039932, | |
| "grad_norm": 0.5297266244888306, | |
| "learning_rate": 4.244473380075395e-05, | |
| "loss": 2.879, | |
| "step": 1682 | |
| }, | |
| { | |
| "epoch": 1.1440951571792692, | |
| "grad_norm": 0.5300079584121704, | |
| "learning_rate": 4.233368287639581e-05, | |
| "loss": 2.9001, | |
| "step": 1684 | |
| }, | |
| { | |
| "epoch": 1.1454545454545455, | |
| "grad_norm": 0.5117048621177673, | |
| "learning_rate": 4.2222670666390164e-05, | |
| "loss": 2.9228, | |
| "step": 1686 | |
| }, | |
| { | |
| "epoch": 1.1468139337298215, | |
| "grad_norm": 0.45803189277648926, | |
| "learning_rate": 4.211169773134072e-05, | |
| "loss": 2.8995, | |
| "step": 1688 | |
| }, | |
| { | |
| "epoch": 1.1481733220050976, | |
| "grad_norm": 0.48471149802207947, | |
| "learning_rate": 4.200076463165285e-05, | |
| "loss": 2.8796, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 1.1495327102803738, | |
| "grad_norm": 0.5804728865623474, | |
| "learning_rate": 4.188987192753079e-05, | |
| "loss": 2.9262, | |
| "step": 1692 | |
| }, | |
| { | |
| "epoch": 1.15089209855565, | |
| "grad_norm": 0.541896402835846, | |
| "learning_rate": 4.177902017897476e-05, | |
| "loss": 2.9021, | |
| "step": 1694 | |
| }, | |
| { | |
| "epoch": 1.1522514868309262, | |
| "grad_norm": 0.5527855157852173, | |
| "learning_rate": 4.166820994577811e-05, | |
| "loss": 2.8635, | |
| "step": 1696 | |
| }, | |
| { | |
| "epoch": 1.1536108751062022, | |
| "grad_norm": 0.5674459338188171, | |
| "learning_rate": 4.155744178752461e-05, | |
| "loss": 2.9249, | |
| "step": 1698 | |
| }, | |
| { | |
| "epoch": 1.1549702633814785, | |
| "grad_norm": 0.5663526058197021, | |
| "learning_rate": 4.144671626358551e-05, | |
| "loss": 2.8515, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.1563296516567545, | |
| "grad_norm": 0.559400200843811, | |
| "learning_rate": 4.13360339331168e-05, | |
| "loss": 2.8725, | |
| "step": 1702 | |
| }, | |
| { | |
| "epoch": 1.1576890399320305, | |
| "grad_norm": 0.5365723371505737, | |
| "learning_rate": 4.122539535505632e-05, | |
| "loss": 2.9244, | |
| "step": 1704 | |
| }, | |
| { | |
| "epoch": 1.1590484282073068, | |
| "grad_norm": 0.4927426278591156, | |
| "learning_rate": 4.1114801088120955e-05, | |
| "loss": 2.9094, | |
| "step": 1706 | |
| }, | |
| { | |
| "epoch": 1.1604078164825828, | |
| "grad_norm": 0.46224114298820496, | |
| "learning_rate": 4.1004251690803835e-05, | |
| "loss": 2.9315, | |
| "step": 1708 | |
| }, | |
| { | |
| "epoch": 1.1617672047578589, | |
| "grad_norm": 0.49180853366851807, | |
| "learning_rate": 4.089374772137149e-05, | |
| "loss": 2.8438, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 1.1631265930331351, | |
| "grad_norm": 0.5289267897605896, | |
| "learning_rate": 4.078328973786108e-05, | |
| "loss": 2.8533, | |
| "step": 1712 | |
| }, | |
| { | |
| "epoch": 1.1644859813084112, | |
| "grad_norm": 0.5112552046775818, | |
| "learning_rate": 4.0672878298077496e-05, | |
| "loss": 2.876, | |
| "step": 1714 | |
| }, | |
| { | |
| "epoch": 1.1658453695836872, | |
| "grad_norm": 0.5080769062042236, | |
| "learning_rate": 4.056251395959061e-05, | |
| "loss": 2.8569, | |
| "step": 1716 | |
| }, | |
| { | |
| "epoch": 1.1672047578589635, | |
| "grad_norm": 0.5041526556015015, | |
| "learning_rate": 4.045219727973242e-05, | |
| "loss": 2.8736, | |
| "step": 1718 | |
| }, | |
| { | |
| "epoch": 1.1685641461342395, | |
| "grad_norm": 0.5132370591163635, | |
| "learning_rate": 4.0341928815594246e-05, | |
| "loss": 2.8896, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.1699235344095158, | |
| "grad_norm": 0.5206835269927979, | |
| "learning_rate": 4.023170912402396e-05, | |
| "loss": 2.9236, | |
| "step": 1722 | |
| }, | |
| { | |
| "epoch": 1.1712829226847918, | |
| "grad_norm": 0.501701295375824, | |
| "learning_rate": 4.01215387616231e-05, | |
| "loss": 2.8922, | |
| "step": 1724 | |
| }, | |
| { | |
| "epoch": 1.1726423109600679, | |
| "grad_norm": 0.49321606755256653, | |
| "learning_rate": 4.001141828474413e-05, | |
| "loss": 2.9145, | |
| "step": 1726 | |
| }, | |
| { | |
| "epoch": 1.1740016992353441, | |
| "grad_norm": 0.5181498527526855, | |
| "learning_rate": 3.990134824948754e-05, | |
| "loss": 2.89, | |
| "step": 1728 | |
| }, | |
| { | |
| "epoch": 1.1753610875106202, | |
| "grad_norm": 0.5045918822288513, | |
| "learning_rate": 3.9791329211699126e-05, | |
| "loss": 2.9238, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 1.1767204757858964, | |
| "grad_norm": 0.4834674894809723, | |
| "learning_rate": 3.968136172696717e-05, | |
| "loss": 2.8589, | |
| "step": 1732 | |
| }, | |
| { | |
| "epoch": 1.1780798640611725, | |
| "grad_norm": 0.490170419216156, | |
| "learning_rate": 3.9571446350619596e-05, | |
| "loss": 2.8641, | |
| "step": 1734 | |
| }, | |
| { | |
| "epoch": 1.1794392523364485, | |
| "grad_norm": 0.49838393926620483, | |
| "learning_rate": 3.946158363772118e-05, | |
| "loss": 2.8944, | |
| "step": 1736 | |
| }, | |
| { | |
| "epoch": 1.1807986406117248, | |
| "grad_norm": 0.4819714426994324, | |
| "learning_rate": 3.935177414307074e-05, | |
| "loss": 2.9104, | |
| "step": 1738 | |
| }, | |
| { | |
| "epoch": 1.1821580288870008, | |
| "grad_norm": 0.5041332244873047, | |
| "learning_rate": 3.924201842119837e-05, | |
| "loss": 2.8782, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.1835174171622769, | |
| "grad_norm": 0.4966350197792053, | |
| "learning_rate": 3.913231702636262e-05, | |
| "loss": 2.8988, | |
| "step": 1742 | |
| }, | |
| { | |
| "epoch": 1.1848768054375531, | |
| "grad_norm": 0.48014774918556213, | |
| "learning_rate": 3.902267051254765e-05, | |
| "loss": 2.855, | |
| "step": 1744 | |
| }, | |
| { | |
| "epoch": 1.1862361937128292, | |
| "grad_norm": 0.5495843887329102, | |
| "learning_rate": 3.891307943346053e-05, | |
| "loss": 2.9295, | |
| "step": 1746 | |
| }, | |
| { | |
| "epoch": 1.1875955819881054, | |
| "grad_norm": 0.5902272462844849, | |
| "learning_rate": 3.880354434252837e-05, | |
| "loss": 2.8563, | |
| "step": 1748 | |
| }, | |
| { | |
| "epoch": 1.1889549702633815, | |
| "grad_norm": 0.603439211845398, | |
| "learning_rate": 3.869406579289547e-05, | |
| "loss": 2.8921, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.1903143585386575, | |
| "grad_norm": 0.547380805015564, | |
| "learning_rate": 3.8584644337420755e-05, | |
| "loss": 2.827, | |
| "step": 1752 | |
| }, | |
| { | |
| "epoch": 1.1916737468139338, | |
| "grad_norm": 0.5630144476890564, | |
| "learning_rate": 3.8475280528674685e-05, | |
| "loss": 2.8361, | |
| "step": 1754 | |
| }, | |
| { | |
| "epoch": 1.1930331350892098, | |
| "grad_norm": 0.5452881455421448, | |
| "learning_rate": 3.8365974918936673e-05, | |
| "loss": 2.8507, | |
| "step": 1756 | |
| }, | |
| { | |
| "epoch": 1.194392523364486, | |
| "grad_norm": 0.5179116129875183, | |
| "learning_rate": 3.825672806019224e-05, | |
| "loss": 2.8862, | |
| "step": 1758 | |
| }, | |
| { | |
| "epoch": 1.1957519116397621, | |
| "grad_norm": 0.5036887526512146, | |
| "learning_rate": 3.814754050413014e-05, | |
| "loss": 2.8467, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.1971112999150382, | |
| "grad_norm": 0.5105863213539124, | |
| "learning_rate": 3.803841280213979e-05, | |
| "loss": 2.764, | |
| "step": 1762 | |
| }, | |
| { | |
| "epoch": 1.1984706881903144, | |
| "grad_norm": 0.49122196435928345, | |
| "learning_rate": 3.792934550530823e-05, | |
| "loss": 2.7833, | |
| "step": 1764 | |
| }, | |
| { | |
| "epoch": 1.1998300764655905, | |
| "grad_norm": 0.49533551931381226, | |
| "learning_rate": 3.78203391644175e-05, | |
| "loss": 2.836, | |
| "step": 1766 | |
| }, | |
| { | |
| "epoch": 1.2011894647408665, | |
| "grad_norm": 0.4618118405342102, | |
| "learning_rate": 3.7711394329941826e-05, | |
| "loss": 2.8429, | |
| "step": 1768 | |
| }, | |
| { | |
| "epoch": 1.2025488530161428, | |
| "grad_norm": 0.49988555908203125, | |
| "learning_rate": 3.7602511552044794e-05, | |
| "loss": 2.7775, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 1.2039082412914188, | |
| "grad_norm": 0.5055059790611267, | |
| "learning_rate": 3.749369138057668e-05, | |
| "loss": 2.8176, | |
| "step": 1772 | |
| }, | |
| { | |
| "epoch": 1.2052676295666949, | |
| "grad_norm": 0.48918312788009644, | |
| "learning_rate": 3.738493436507153e-05, | |
| "loss": 2.8536, | |
| "step": 1774 | |
| }, | |
| { | |
| "epoch": 1.2066270178419711, | |
| "grad_norm": 0.46505334973335266, | |
| "learning_rate": 3.7276241054744484e-05, | |
| "loss": 2.813, | |
| "step": 1776 | |
| }, | |
| { | |
| "epoch": 1.2079864061172472, | |
| "grad_norm": 0.4389493763446808, | |
| "learning_rate": 3.7167611998488974e-05, | |
| "loss": 2.8881, | |
| "step": 1778 | |
| }, | |
| { | |
| "epoch": 1.2093457943925234, | |
| "grad_norm": 0.46267229318618774, | |
| "learning_rate": 3.705904774487396e-05, | |
| "loss": 2.9389, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.2107051826677995, | |
| "grad_norm": 0.5205557346343994, | |
| "learning_rate": 3.695054884214115e-05, | |
| "loss": 2.8476, | |
| "step": 1782 | |
| }, | |
| { | |
| "epoch": 1.2120645709430757, | |
| "grad_norm": 0.4998641014099121, | |
| "learning_rate": 3.6842115838202255e-05, | |
| "loss": 2.9114, | |
| "step": 1784 | |
| }, | |
| { | |
| "epoch": 1.2134239592183518, | |
| "grad_norm": 0.5282990336418152, | |
| "learning_rate": 3.673374928063614e-05, | |
| "loss": 2.8685, | |
| "step": 1786 | |
| }, | |
| { | |
| "epoch": 1.2147833474936278, | |
| "grad_norm": 0.48257991671562195, | |
| "learning_rate": 3.662544971668619e-05, | |
| "loss": 2.7768, | |
| "step": 1788 | |
| }, | |
| { | |
| "epoch": 1.216142735768904, | |
| "grad_norm": 0.5428197383880615, | |
| "learning_rate": 3.651721769325742e-05, | |
| "loss": 2.8578, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 1.2175021240441801, | |
| "grad_norm": 0.560145378112793, | |
| "learning_rate": 3.640905375691382e-05, | |
| "loss": 2.8684, | |
| "step": 1792 | |
| }, | |
| { | |
| "epoch": 1.2188615123194562, | |
| "grad_norm": 0.5094372630119324, | |
| "learning_rate": 3.630095845387553e-05, | |
| "loss": 2.9107, | |
| "step": 1794 | |
| }, | |
| { | |
| "epoch": 1.2202209005947324, | |
| "grad_norm": 0.5483725666999817, | |
| "learning_rate": 3.619293233001607e-05, | |
| "loss": 2.907, | |
| "step": 1796 | |
| }, | |
| { | |
| "epoch": 1.2215802888700085, | |
| "grad_norm": 0.5447711944580078, | |
| "learning_rate": 3.608497593085963e-05, | |
| "loss": 2.8597, | |
| "step": 1798 | |
| }, | |
| { | |
| "epoch": 1.2229396771452845, | |
| "grad_norm": 0.4736505150794983, | |
| "learning_rate": 3.59770898015783e-05, | |
| "loss": 2.8392, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.2242990654205608, | |
| "grad_norm": 0.4515341818332672, | |
| "learning_rate": 3.586927448698932e-05, | |
| "loss": 2.8654, | |
| "step": 1802 | |
| }, | |
| { | |
| "epoch": 1.2256584536958368, | |
| "grad_norm": 0.5168652534484863, | |
| "learning_rate": 3.576153053155231e-05, | |
| "loss": 2.8678, | |
| "step": 1804 | |
| }, | |
| { | |
| "epoch": 1.227017841971113, | |
| "grad_norm": 0.5041874647140503, | |
| "learning_rate": 3.56538584793665e-05, | |
| "loss": 2.9214, | |
| "step": 1806 | |
| }, | |
| { | |
| "epoch": 1.2283772302463891, | |
| "grad_norm": 0.5031694173812866, | |
| "learning_rate": 3.5546258874168055e-05, | |
| "loss": 2.8274, | |
| "step": 1808 | |
| }, | |
| { | |
| "epoch": 1.2297366185216652, | |
| "grad_norm": 0.48512697219848633, | |
| "learning_rate": 3.543873225932729e-05, | |
| "loss": 2.916, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 1.2310960067969414, | |
| "grad_norm": 0.48854321241378784, | |
| "learning_rate": 3.5331279177845905e-05, | |
| "loss": 2.8523, | |
| "step": 1812 | |
| }, | |
| { | |
| "epoch": 1.2324553950722175, | |
| "grad_norm": 0.48543816804885864, | |
| "learning_rate": 3.5223900172354264e-05, | |
| "loss": 2.8916, | |
| "step": 1814 | |
| }, | |
| { | |
| "epoch": 1.2338147833474937, | |
| "grad_norm": 0.5021933913230896, | |
| "learning_rate": 3.511659578510866e-05, | |
| "loss": 2.8412, | |
| "step": 1816 | |
| }, | |
| { | |
| "epoch": 1.2351741716227698, | |
| "grad_norm": 0.5450052618980408, | |
| "learning_rate": 3.5009366557988535e-05, | |
| "loss": 2.8668, | |
| "step": 1818 | |
| }, | |
| { | |
| "epoch": 1.2365335598980458, | |
| "grad_norm": 0.5432640314102173, | |
| "learning_rate": 3.4902213032493866e-05, | |
| "loss": 2.8854, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.237892948173322, | |
| "grad_norm": 0.5332628488540649, | |
| "learning_rate": 3.479513574974224e-05, | |
| "loss": 2.8601, | |
| "step": 1822 | |
| }, | |
| { | |
| "epoch": 1.2392523364485981, | |
| "grad_norm": 0.5209367275238037, | |
| "learning_rate": 3.468813525046627e-05, | |
| "loss": 2.8922, | |
| "step": 1824 | |
| }, | |
| { | |
| "epoch": 1.2406117247238742, | |
| "grad_norm": 0.49470552802085876, | |
| "learning_rate": 3.458121207501083e-05, | |
| "loss": 2.8358, | |
| "step": 1826 | |
| }, | |
| { | |
| "epoch": 1.2419711129991504, | |
| "grad_norm": 0.47799840569496155, | |
| "learning_rate": 3.4474366763330266e-05, | |
| "loss": 2.8292, | |
| "step": 1828 | |
| }, | |
| { | |
| "epoch": 1.2433305012744265, | |
| "grad_norm": 0.46769949793815613, | |
| "learning_rate": 3.436759985498581e-05, | |
| "loss": 2.8569, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 1.2446898895497027, | |
| "grad_norm": 0.4332127273082733, | |
| "learning_rate": 3.426091188914266e-05, | |
| "loss": 2.851, | |
| "step": 1832 | |
| }, | |
| { | |
| "epoch": 1.2460492778249788, | |
| "grad_norm": 0.49272555112838745, | |
| "learning_rate": 3.415430340456741e-05, | |
| "loss": 2.8041, | |
| "step": 1834 | |
| }, | |
| { | |
| "epoch": 1.2474086661002548, | |
| "grad_norm": 0.49650368094444275, | |
| "learning_rate": 3.40477749396253e-05, | |
| "loss": 2.892, | |
| "step": 1836 | |
| }, | |
| { | |
| "epoch": 1.248768054375531, | |
| "grad_norm": 0.5175567269325256, | |
| "learning_rate": 3.394132703227738e-05, | |
| "loss": 2.8535, | |
| "step": 1838 | |
| }, | |
| { | |
| "epoch": 1.250127442650807, | |
| "grad_norm": 0.44107678532600403, | |
| "learning_rate": 3.3834960220078046e-05, | |
| "loss": 2.7789, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.2514868309260834, | |
| "grad_norm": 0.454111248254776, | |
| "learning_rate": 3.372867504017203e-05, | |
| "loss": 2.8574, | |
| "step": 1842 | |
| }, | |
| { | |
| "epoch": 1.2528462192013594, | |
| "grad_norm": 0.42469438910484314, | |
| "learning_rate": 3.36224720292919e-05, | |
| "loss": 2.8232, | |
| "step": 1844 | |
| }, | |
| { | |
| "epoch": 1.2542056074766355, | |
| "grad_norm": 0.4407486915588379, | |
| "learning_rate": 3.351635172375524e-05, | |
| "loss": 2.8474, | |
| "step": 1846 | |
| }, | |
| { | |
| "epoch": 1.2555649957519117, | |
| "grad_norm": 0.48633596301078796, | |
| "learning_rate": 3.341031465946199e-05, | |
| "loss": 2.867, | |
| "step": 1848 | |
| }, | |
| { | |
| "epoch": 1.2569243840271878, | |
| "grad_norm": 0.4663158357143402, | |
| "learning_rate": 3.3304361371891766e-05, | |
| "loss": 2.9479, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.2582837723024638, | |
| "grad_norm": 0.4377327263355255, | |
| "learning_rate": 3.319849239610107e-05, | |
| "loss": 2.8796, | |
| "step": 1852 | |
| }, | |
| { | |
| "epoch": 1.25964316057774, | |
| "grad_norm": 0.47921085357666016, | |
| "learning_rate": 3.309270826672062e-05, | |
| "loss": 2.7795, | |
| "step": 1854 | |
| }, | |
| { | |
| "epoch": 1.261002548853016, | |
| "grad_norm": 0.47767359018325806, | |
| "learning_rate": 3.298700951795272e-05, | |
| "loss": 2.8866, | |
| "step": 1856 | |
| }, | |
| { | |
| "epoch": 1.2623619371282921, | |
| "grad_norm": 0.49657002091407776, | |
| "learning_rate": 3.2881396683568466e-05, | |
| "loss": 2.9269, | |
| "step": 1858 | |
| }, | |
| { | |
| "epoch": 1.2637213254035684, | |
| "grad_norm": 0.5701535940170288, | |
| "learning_rate": 3.277587029690512e-05, | |
| "loss": 2.8742, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.2650807136788444, | |
| "grad_norm": 0.5785423517227173, | |
| "learning_rate": 3.26704308908634e-05, | |
| "loss": 2.8234, | |
| "step": 1862 | |
| }, | |
| { | |
| "epoch": 1.2664401019541207, | |
| "grad_norm": 0.5667394399642944, | |
| "learning_rate": 3.256507899790472e-05, | |
| "loss": 2.8592, | |
| "step": 1864 | |
| }, | |
| { | |
| "epoch": 1.2677994902293968, | |
| "grad_norm": 0.534062922000885, | |
| "learning_rate": 3.245981515004861e-05, | |
| "loss": 2.8033, | |
| "step": 1866 | |
| }, | |
| { | |
| "epoch": 1.269158878504673, | |
| "grad_norm": 0.4877188205718994, | |
| "learning_rate": 3.2354639878869945e-05, | |
| "loss": 2.8559, | |
| "step": 1868 | |
| }, | |
| { | |
| "epoch": 1.270518266779949, | |
| "grad_norm": 0.49617844820022583, | |
| "learning_rate": 3.224955371549635e-05, | |
| "loss": 2.8251, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.271877655055225, | |
| "grad_norm": 0.49815118312835693, | |
| "learning_rate": 3.2144557190605405e-05, | |
| "loss": 2.8841, | |
| "step": 1872 | |
| }, | |
| { | |
| "epoch": 1.2732370433305014, | |
| "grad_norm": 0.5205366015434265, | |
| "learning_rate": 3.2039650834422044e-05, | |
| "loss": 2.844, | |
| "step": 1874 | |
| }, | |
| { | |
| "epoch": 1.2745964316057774, | |
| "grad_norm": 0.5351797342300415, | |
| "learning_rate": 3.193483517671585e-05, | |
| "loss": 2.8207, | |
| "step": 1876 | |
| }, | |
| { | |
| "epoch": 1.2759558198810534, | |
| "grad_norm": 0.5261284708976746, | |
| "learning_rate": 3.1830110746798374e-05, | |
| "loss": 2.9103, | |
| "step": 1878 | |
| }, | |
| { | |
| "epoch": 1.2773152081563297, | |
| "grad_norm": 0.5345544815063477, | |
| "learning_rate": 3.172547807352052e-05, | |
| "loss": 2.8577, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.2786745964316057, | |
| "grad_norm": 0.5085960626602173, | |
| "learning_rate": 3.162093768526978e-05, | |
| "loss": 2.8485, | |
| "step": 1882 | |
| }, | |
| { | |
| "epoch": 1.2800339847068818, | |
| "grad_norm": 0.5139952898025513, | |
| "learning_rate": 3.151649010996763e-05, | |
| "loss": 2.8725, | |
| "step": 1884 | |
| }, | |
| { | |
| "epoch": 1.281393372982158, | |
| "grad_norm": 0.49606403708457947, | |
| "learning_rate": 3.141213587506685e-05, | |
| "loss": 2.9127, | |
| "step": 1886 | |
| }, | |
| { | |
| "epoch": 1.282752761257434, | |
| "grad_norm": 0.5038158297538757, | |
| "learning_rate": 3.130787550754883e-05, | |
| "loss": 2.8541, | |
| "step": 1888 | |
| }, | |
| { | |
| "epoch": 1.2841121495327104, | |
| "grad_norm": 0.47365954518318176, | |
| "learning_rate": 3.120370953392099e-05, | |
| "loss": 2.9008, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 1.2854715378079864, | |
| "grad_norm": 0.4973777234554291, | |
| "learning_rate": 3.109963848021402e-05, | |
| "loss": 2.8606, | |
| "step": 1892 | |
| }, | |
| { | |
| "epoch": 1.2868309260832627, | |
| "grad_norm": 0.4743622839450836, | |
| "learning_rate": 3.0995662871979316e-05, | |
| "loss": 2.9113, | |
| "step": 1894 | |
| }, | |
| { | |
| "epoch": 1.2881903143585387, | |
| "grad_norm": 0.432390421628952, | |
| "learning_rate": 3.089178323428621e-05, | |
| "loss": 2.8505, | |
| "step": 1896 | |
| }, | |
| { | |
| "epoch": 1.2895497026338147, | |
| "grad_norm": 0.45295336842536926, | |
| "learning_rate": 3.0788000091719456e-05, | |
| "loss": 2.8676, | |
| "step": 1898 | |
| }, | |
| { | |
| "epoch": 1.290909090909091, | |
| "grad_norm": 0.5054479241371155, | |
| "learning_rate": 3.06843139683765e-05, | |
| "loss": 2.7953, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.292268479184367, | |
| "grad_norm": 0.4613514244556427, | |
| "learning_rate": 3.058072538786483e-05, | |
| "loss": 2.7688, | |
| "step": 1902 | |
| }, | |
| { | |
| "epoch": 1.293627867459643, | |
| "grad_norm": 0.46900779008865356, | |
| "learning_rate": 3.0477234873299378e-05, | |
| "loss": 2.8095, | |
| "step": 1904 | |
| }, | |
| { | |
| "epoch": 1.2949872557349194, | |
| "grad_norm": 0.495675265789032, | |
| "learning_rate": 3.0373842947299804e-05, | |
| "loss": 2.8388, | |
| "step": 1906 | |
| }, | |
| { | |
| "epoch": 1.2963466440101954, | |
| "grad_norm": 0.5001600980758667, | |
| "learning_rate": 3.027055013198793e-05, | |
| "loss": 2.8909, | |
| "step": 1908 | |
| }, | |
| { | |
| "epoch": 1.2977060322854714, | |
| "grad_norm": 0.49891334772109985, | |
| "learning_rate": 3.0167356948985094e-05, | |
| "loss": 2.8075, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 1.2990654205607477, | |
| "grad_norm": 0.4925783574581146, | |
| "learning_rate": 3.0064263919409485e-05, | |
| "loss": 2.8572, | |
| "step": 1912 | |
| }, | |
| { | |
| "epoch": 1.3004248088360237, | |
| "grad_norm": 0.4837004840373993, | |
| "learning_rate": 2.9961271563873504e-05, | |
| "loss": 2.829, | |
| "step": 1914 | |
| }, | |
| { | |
| "epoch": 1.3017841971112998, | |
| "grad_norm": 0.49941548705101013, | |
| "learning_rate": 2.98583804024812e-05, | |
| "loss": 2.8893, | |
| "step": 1916 | |
| }, | |
| { | |
| "epoch": 1.303143585386576, | |
| "grad_norm": 0.44845762848854065, | |
| "learning_rate": 2.9755590954825506e-05, | |
| "loss": 2.9031, | |
| "step": 1918 | |
| }, | |
| { | |
| "epoch": 1.3045029736618523, | |
| "grad_norm": 0.4511342942714691, | |
| "learning_rate": 2.9652903739985855e-05, | |
| "loss": 2.7632, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 1.3058623619371283, | |
| "grad_norm": 0.4593941569328308, | |
| "learning_rate": 2.9550319276525272e-05, | |
| "loss": 2.8752, | |
| "step": 1922 | |
| }, | |
| { | |
| "epoch": 1.3072217502124044, | |
| "grad_norm": 0.4910091757774353, | |
| "learning_rate": 2.9447838082487965e-05, | |
| "loss": 2.8727, | |
| "step": 1924 | |
| }, | |
| { | |
| "epoch": 1.3085811384876807, | |
| "grad_norm": 0.4901241958141327, | |
| "learning_rate": 2.9345460675396648e-05, | |
| "loss": 2.8058, | |
| "step": 1926 | |
| }, | |
| { | |
| "epoch": 1.3099405267629567, | |
| "grad_norm": 0.4703480899333954, | |
| "learning_rate": 2.924318757224984e-05, | |
| "loss": 2.8246, | |
| "step": 1928 | |
| }, | |
| { | |
| "epoch": 1.3112999150382327, | |
| "grad_norm": 0.4554821848869324, | |
| "learning_rate": 2.9141019289519456e-05, | |
| "loss": 2.8579, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 1.312659303313509, | |
| "grad_norm": 0.45331230759620667, | |
| "learning_rate": 2.903895634314795e-05, | |
| "loss": 2.8389, | |
| "step": 1932 | |
| }, | |
| { | |
| "epoch": 1.314018691588785, | |
| "grad_norm": 0.4508642554283142, | |
| "learning_rate": 2.8936999248545943e-05, | |
| "loss": 2.9092, | |
| "step": 1934 | |
| }, | |
| { | |
| "epoch": 1.315378079864061, | |
| "grad_norm": 0.44733238220214844, | |
| "learning_rate": 2.8835148520589416e-05, | |
| "loss": 2.8279, | |
| "step": 1936 | |
| }, | |
| { | |
| "epoch": 1.3167374681393373, | |
| "grad_norm": 0.4859914183616638, | |
| "learning_rate": 2.8733404673617293e-05, | |
| "loss": 2.8903, | |
| "step": 1938 | |
| }, | |
| { | |
| "epoch": 1.3180968564146134, | |
| "grad_norm": 0.4850110113620758, | |
| "learning_rate": 2.8631768221428675e-05, | |
| "loss": 2.8036, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 1.3194562446898894, | |
| "grad_norm": 0.46834543347358704, | |
| "learning_rate": 2.8530239677280402e-05, | |
| "loss": 2.9167, | |
| "step": 1942 | |
| }, | |
| { | |
| "epoch": 1.3208156329651657, | |
| "grad_norm": 0.4862910211086273, | |
| "learning_rate": 2.8428819553884344e-05, | |
| "loss": 2.8569, | |
| "step": 1944 | |
| }, | |
| { | |
| "epoch": 1.3221750212404417, | |
| "grad_norm": 0.4921141564846039, | |
| "learning_rate": 2.8327508363404813e-05, | |
| "loss": 2.8712, | |
| "step": 1946 | |
| }, | |
| { | |
| "epoch": 1.323534409515718, | |
| "grad_norm": 0.43411630392074585, | |
| "learning_rate": 2.822630661745615e-05, | |
| "loss": 2.8041, | |
| "step": 1948 | |
| }, | |
| { | |
| "epoch": 1.324893797790994, | |
| "grad_norm": 0.4472651183605194, | |
| "learning_rate": 2.8125214827099878e-05, | |
| "loss": 2.8275, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.3262531860662703, | |
| "grad_norm": 0.4626396596431732, | |
| "learning_rate": 2.802423350284229e-05, | |
| "loss": 2.8892, | |
| "step": 1952 | |
| }, | |
| { | |
| "epoch": 1.3276125743415463, | |
| "grad_norm": 0.43589314818382263, | |
| "learning_rate": 2.7923363154631875e-05, | |
| "loss": 2.8549, | |
| "step": 1954 | |
| }, | |
| { | |
| "epoch": 1.3289719626168224, | |
| "grad_norm": 0.45954057574272156, | |
| "learning_rate": 2.782260429185663e-05, | |
| "loss": 2.8644, | |
| "step": 1956 | |
| }, | |
| { | |
| "epoch": 1.3303313508920986, | |
| "grad_norm": 0.4734349846839905, | |
| "learning_rate": 2.7721957423341638e-05, | |
| "loss": 2.907, | |
| "step": 1958 | |
| }, | |
| { | |
| "epoch": 1.3316907391673747, | |
| "grad_norm": 0.48602449893951416, | |
| "learning_rate": 2.7621423057346335e-05, | |
| "loss": 2.8916, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 1.3330501274426507, | |
| "grad_norm": 0.4574413299560547, | |
| "learning_rate": 2.752100170156211e-05, | |
| "loss": 2.7806, | |
| "step": 1962 | |
| }, | |
| { | |
| "epoch": 1.334409515717927, | |
| "grad_norm": 0.47104230523109436, | |
| "learning_rate": 2.7420693863109583e-05, | |
| "loss": 2.7884, | |
| "step": 1964 | |
| }, | |
| { | |
| "epoch": 1.335768903993203, | |
| "grad_norm": 0.46158191561698914, | |
| "learning_rate": 2.7320500048536174e-05, | |
| "loss": 2.8153, | |
| "step": 1966 | |
| }, | |
| { | |
| "epoch": 1.337128292268479, | |
| "grad_norm": 0.49134361743927, | |
| "learning_rate": 2.722042076381349e-05, | |
| "loss": 2.8798, | |
| "step": 1968 | |
| }, | |
| { | |
| "epoch": 1.3384876805437553, | |
| "grad_norm": 0.451119065284729, | |
| "learning_rate": 2.712045651433473e-05, | |
| "loss": 2.7705, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 1.3398470688190314, | |
| "grad_norm": 0.4850034713745117, | |
| "learning_rate": 2.702060780491218e-05, | |
| "loss": 2.8366, | |
| "step": 1972 | |
| }, | |
| { | |
| "epoch": 1.3412064570943076, | |
| "grad_norm": 0.49492767453193665, | |
| "learning_rate": 2.6920875139774714e-05, | |
| "loss": 2.7507, | |
| "step": 1974 | |
| }, | |
| { | |
| "epoch": 1.3425658453695837, | |
| "grad_norm": 0.4679727554321289, | |
| "learning_rate": 2.6821259022565105e-05, | |
| "loss": 2.8974, | |
| "step": 1976 | |
| }, | |
| { | |
| "epoch": 1.34392523364486, | |
| "grad_norm": 0.42731571197509766, | |
| "learning_rate": 2.6721759956337633e-05, | |
| "loss": 2.8225, | |
| "step": 1978 | |
| }, | |
| { | |
| "epoch": 1.345284621920136, | |
| "grad_norm": 0.46519172191619873, | |
| "learning_rate": 2.6622378443555475e-05, | |
| "loss": 2.8715, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 1.346644010195412, | |
| "grad_norm": 0.4477517902851105, | |
| "learning_rate": 2.6523114986088132e-05, | |
| "loss": 2.8791, | |
| "step": 1982 | |
| }, | |
| { | |
| "epoch": 1.3480033984706883, | |
| "grad_norm": 0.43799683451652527, | |
| "learning_rate": 2.6423970085208947e-05, | |
| "loss": 2.8548, | |
| "step": 1984 | |
| }, | |
| { | |
| "epoch": 1.3493627867459643, | |
| "grad_norm": 0.4669962227344513, | |
| "learning_rate": 2.632494424159258e-05, | |
| "loss": 2.789, | |
| "step": 1986 | |
| }, | |
| { | |
| "epoch": 1.3507221750212404, | |
| "grad_norm": 0.4736376106739044, | |
| "learning_rate": 2.6226037955312476e-05, | |
| "loss": 2.8733, | |
| "step": 1988 | |
| }, | |
| { | |
| "epoch": 1.3520815632965166, | |
| "grad_norm": 0.49567386507987976, | |
| "learning_rate": 2.6127251725838276e-05, | |
| "loss": 2.8018, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 1.3534409515717927, | |
| "grad_norm": 0.523835301399231, | |
| "learning_rate": 2.6028586052033337e-05, | |
| "loss": 2.7986, | |
| "step": 1992 | |
| }, | |
| { | |
| "epoch": 1.3548003398470687, | |
| "grad_norm": 0.441232830286026, | |
| "learning_rate": 2.5930041432152296e-05, | |
| "loss": 2.8209, | |
| "step": 1994 | |
| }, | |
| { | |
| "epoch": 1.356159728122345, | |
| "grad_norm": 0.4717652201652527, | |
| "learning_rate": 2.5831618363838396e-05, | |
| "loss": 2.8599, | |
| "step": 1996 | |
| }, | |
| { | |
| "epoch": 1.357519116397621, | |
| "grad_norm": 0.4860689640045166, | |
| "learning_rate": 2.5733317344121077e-05, | |
| "loss": 2.8483, | |
| "step": 1998 | |
| }, | |
| { | |
| "epoch": 1.358878504672897, | |
| "grad_norm": 0.4942462146282196, | |
| "learning_rate": 2.5635138869413494e-05, | |
| "loss": 2.7951, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.3602378929481733, | |
| "grad_norm": 0.4958104193210602, | |
| "learning_rate": 2.5537083435509877e-05, | |
| "loss": 2.8055, | |
| "step": 2002 | |
| }, | |
| { | |
| "epoch": 1.3615972812234496, | |
| "grad_norm": 0.43072277307510376, | |
| "learning_rate": 2.543915153758314e-05, | |
| "loss": 2.8338, | |
| "step": 2004 | |
| }, | |
| { | |
| "epoch": 1.3629566694987256, | |
| "grad_norm": 0.4571772515773773, | |
| "learning_rate": 2.534134367018237e-05, | |
| "loss": 2.8616, | |
| "step": 2006 | |
| }, | |
| { | |
| "epoch": 1.3643160577740017, | |
| "grad_norm": 0.4508160352706909, | |
| "learning_rate": 2.5243660327230306e-05, | |
| "loss": 2.8556, | |
| "step": 2008 | |
| }, | |
| { | |
| "epoch": 1.365675446049278, | |
| "grad_norm": 0.4363895356655121, | |
| "learning_rate": 2.51461020020208e-05, | |
| "loss": 2.8408, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 1.367034834324554, | |
| "grad_norm": 0.4177226722240448, | |
| "learning_rate": 2.5048669187216455e-05, | |
| "loss": 2.9044, | |
| "step": 2012 | |
| }, | |
| { | |
| "epoch": 1.36839422259983, | |
| "grad_norm": 0.41639578342437744, | |
| "learning_rate": 2.4951362374845976e-05, | |
| "loss": 2.827, | |
| "step": 2014 | |
| }, | |
| { | |
| "epoch": 1.3697536108751063, | |
| "grad_norm": 0.43476781249046326, | |
| "learning_rate": 2.485418205630179e-05, | |
| "loss": 2.889, | |
| "step": 2016 | |
| }, | |
| { | |
| "epoch": 1.3711129991503823, | |
| "grad_norm": 0.46106597781181335, | |
| "learning_rate": 2.475712872233755e-05, | |
| "loss": 2.7914, | |
| "step": 2018 | |
| }, | |
| { | |
| "epoch": 1.3724723874256584, | |
| "grad_norm": 0.46807268261909485, | |
| "learning_rate": 2.4660202863065685e-05, | |
| "loss": 2.7833, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 1.3738317757009346, | |
| "grad_norm": 0.5006850957870483, | |
| "learning_rate": 2.4563404967954805e-05, | |
| "loss": 2.8421, | |
| "step": 2022 | |
| }, | |
| { | |
| "epoch": 1.3751911639762107, | |
| "grad_norm": 0.4208027720451355, | |
| "learning_rate": 2.446673552582734e-05, | |
| "loss": 2.823, | |
| "step": 2024 | |
| }, | |
| { | |
| "epoch": 1.3765505522514867, | |
| "grad_norm": 0.4191267192363739, | |
| "learning_rate": 2.437019502485706e-05, | |
| "loss": 2.9048, | |
| "step": 2026 | |
| }, | |
| { | |
| "epoch": 1.377909940526763, | |
| "grad_norm": 0.4592435956001282, | |
| "learning_rate": 2.4273783952566613e-05, | |
| "loss": 2.7819, | |
| "step": 2028 | |
| }, | |
| { | |
| "epoch": 1.379269328802039, | |
| "grad_norm": 0.42076575756073, | |
| "learning_rate": 2.4177502795824963e-05, | |
| "loss": 2.823, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 1.3806287170773153, | |
| "grad_norm": 0.42424866557121277, | |
| "learning_rate": 2.4081352040845096e-05, | |
| "loss": 2.954, | |
| "step": 2032 | |
| }, | |
| { | |
| "epoch": 1.3819881053525913, | |
| "grad_norm": 0.4503638446331024, | |
| "learning_rate": 2.3985332173181425e-05, | |
| "loss": 2.8155, | |
| "step": 2034 | |
| }, | |
| { | |
| "epoch": 1.3833474936278676, | |
| "grad_norm": 0.47037214040756226, | |
| "learning_rate": 2.3889443677727385e-05, | |
| "loss": 2.8223, | |
| "step": 2036 | |
| }, | |
| { | |
| "epoch": 1.3847068819031436, | |
| "grad_norm": 0.43903493881225586, | |
| "learning_rate": 2.379368703871303e-05, | |
| "loss": 2.8353, | |
| "step": 2038 | |
| }, | |
| { | |
| "epoch": 1.3860662701784197, | |
| "grad_norm": 0.4471070170402527, | |
| "learning_rate": 2.369806273970255e-05, | |
| "loss": 2.894, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 1.387425658453696, | |
| "grad_norm": 0.4600374102592468, | |
| "learning_rate": 2.360257126359176e-05, | |
| "loss": 2.7868, | |
| "step": 2042 | |
| }, | |
| { | |
| "epoch": 1.388785046728972, | |
| "grad_norm": 0.42782899737358093, | |
| "learning_rate": 2.3507213092605827e-05, | |
| "loss": 2.9069, | |
| "step": 2044 | |
| }, | |
| { | |
| "epoch": 1.390144435004248, | |
| "grad_norm": 0.44227334856987, | |
| "learning_rate": 2.3411988708296644e-05, | |
| "loss": 2.8507, | |
| "step": 2046 | |
| }, | |
| { | |
| "epoch": 1.3915038232795243, | |
| "grad_norm": 0.42961835861206055, | |
| "learning_rate": 2.331689859154057e-05, | |
| "loss": 2.8275, | |
| "step": 2048 | |
| }, | |
| { | |
| "epoch": 1.3928632115548003, | |
| "grad_norm": 0.4032466411590576, | |
| "learning_rate": 2.3221943222535853e-05, | |
| "loss": 2.8334, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.3942225998300763, | |
| "grad_norm": 0.4255675673484802, | |
| "learning_rate": 2.3127123080800333e-05, | |
| "loss": 2.8475, | |
| "step": 2052 | |
| }, | |
| { | |
| "epoch": 1.3955819881053526, | |
| "grad_norm": 0.4679095447063446, | |
| "learning_rate": 2.3032438645168923e-05, | |
| "loss": 2.8035, | |
| "step": 2054 | |
| }, | |
| { | |
| "epoch": 1.3969413763806287, | |
| "grad_norm": 0.46383097767829895, | |
| "learning_rate": 2.2937890393791213e-05, | |
| "loss": 2.8414, | |
| "step": 2056 | |
| }, | |
| { | |
| "epoch": 1.398300764655905, | |
| "grad_norm": 0.43089723587036133, | |
| "learning_rate": 2.2843478804129165e-05, | |
| "loss": 2.8987, | |
| "step": 2058 | |
| }, | |
| { | |
| "epoch": 1.399660152931181, | |
| "grad_norm": 0.44261154532432556, | |
| "learning_rate": 2.2749204352954512e-05, | |
| "loss": 2.8264, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 1.4010195412064572, | |
| "grad_norm": 0.45536670088768005, | |
| "learning_rate": 2.2655067516346462e-05, | |
| "loss": 2.8366, | |
| "step": 2062 | |
| }, | |
| { | |
| "epoch": 1.4023789294817333, | |
| "grad_norm": 0.4627995193004608, | |
| "learning_rate": 2.2561068769689324e-05, | |
| "loss": 2.7332, | |
| "step": 2064 | |
| }, | |
| { | |
| "epoch": 1.4037383177570093, | |
| "grad_norm": 0.5014654397964478, | |
| "learning_rate": 2.2467208587670052e-05, | |
| "loss": 2.8671, | |
| "step": 2066 | |
| }, | |
| { | |
| "epoch": 1.4050977060322856, | |
| "grad_norm": 0.4588980972766876, | |
| "learning_rate": 2.2373487444275824e-05, | |
| "loss": 2.8668, | |
| "step": 2068 | |
| }, | |
| { | |
| "epoch": 1.4064570943075616, | |
| "grad_norm": 0.41913726925849915, | |
| "learning_rate": 2.2279905812791685e-05, | |
| "loss": 2.9089, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 1.4078164825828376, | |
| "grad_norm": 0.4351765513420105, | |
| "learning_rate": 2.218646416579821e-05, | |
| "loss": 2.8597, | |
| "step": 2072 | |
| }, | |
| { | |
| "epoch": 1.409175870858114, | |
| "grad_norm": 0.4838959574699402, | |
| "learning_rate": 2.209316297516899e-05, | |
| "loss": 2.8911, | |
| "step": 2074 | |
| }, | |
| { | |
| "epoch": 1.41053525913339, | |
| "grad_norm": 0.3995363712310791, | |
| "learning_rate": 2.2000002712068372e-05, | |
| "loss": 2.8658, | |
| "step": 2076 | |
| }, | |
| { | |
| "epoch": 1.411894647408666, | |
| "grad_norm": 0.4259326756000519, | |
| "learning_rate": 2.1906983846949032e-05, | |
| "loss": 2.8813, | |
| "step": 2078 | |
| }, | |
| { | |
| "epoch": 1.4132540356839423, | |
| "grad_norm": 0.4843901991844177, | |
| "learning_rate": 2.1814106849549565e-05, | |
| "loss": 2.9012, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 1.4146134239592183, | |
| "grad_norm": 0.5000529289245605, | |
| "learning_rate": 2.172137218889213e-05, | |
| "loss": 2.8222, | |
| "step": 2082 | |
| }, | |
| { | |
| "epoch": 1.4159728122344946, | |
| "grad_norm": 0.46484094858169556, | |
| "learning_rate": 2.1628780333280147e-05, | |
| "loss": 2.8229, | |
| "step": 2084 | |
| }, | |
| { | |
| "epoch": 1.4173322005097706, | |
| "grad_norm": 0.48778775334358215, | |
| "learning_rate": 2.1536331750295873e-05, | |
| "loss": 2.8788, | |
| "step": 2086 | |
| }, | |
| { | |
| "epoch": 1.4186915887850469, | |
| "grad_norm": 0.46361327171325684, | |
| "learning_rate": 2.1444026906797997e-05, | |
| "loss": 2.8669, | |
| "step": 2088 | |
| }, | |
| { | |
| "epoch": 1.420050977060323, | |
| "grad_norm": 0.4668735861778259, | |
| "learning_rate": 2.1351866268919403e-05, | |
| "loss": 2.8373, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 1.421410365335599, | |
| "grad_norm": 0.40944549441337585, | |
| "learning_rate": 2.1259850302064698e-05, | |
| "loss": 2.8257, | |
| "step": 2092 | |
| }, | |
| { | |
| "epoch": 1.4227697536108752, | |
| "grad_norm": 0.41765180230140686, | |
| "learning_rate": 2.11679794709079e-05, | |
| "loss": 2.8493, | |
| "step": 2094 | |
| }, | |
| { | |
| "epoch": 1.4241291418861513, | |
| "grad_norm": 0.44578754901885986, | |
| "learning_rate": 2.1076254239390153e-05, | |
| "loss": 2.8111, | |
| "step": 2096 | |
| }, | |
| { | |
| "epoch": 1.4254885301614273, | |
| "grad_norm": 0.4859372675418854, | |
| "learning_rate": 2.0984675070717325e-05, | |
| "loss": 2.8332, | |
| "step": 2098 | |
| }, | |
| { | |
| "epoch": 1.4268479184367036, | |
| "grad_norm": 0.4398513734340668, | |
| "learning_rate": 2.089324242735764e-05, | |
| "loss": 2.876, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.4282073067119796, | |
| "grad_norm": 0.42089375853538513, | |
| "learning_rate": 2.0801956771039382e-05, | |
| "loss": 2.8038, | |
| "step": 2102 | |
| }, | |
| { | |
| "epoch": 1.4295666949872556, | |
| "grad_norm": 0.4485505521297455, | |
| "learning_rate": 2.0710818562748596e-05, | |
| "loss": 2.8652, | |
| "step": 2104 | |
| }, | |
| { | |
| "epoch": 1.430926083262532, | |
| "grad_norm": 0.4214796721935272, | |
| "learning_rate": 2.0619828262726725e-05, | |
| "loss": 2.8339, | |
| "step": 2106 | |
| }, | |
| { | |
| "epoch": 1.432285471537808, | |
| "grad_norm": 0.4246174097061157, | |
| "learning_rate": 2.0528986330468235e-05, | |
| "loss": 2.805, | |
| "step": 2108 | |
| }, | |
| { | |
| "epoch": 1.433644859813084, | |
| "grad_norm": 0.43008217215538025, | |
| "learning_rate": 2.0438293224718418e-05, | |
| "loss": 2.855, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 1.4350042480883602, | |
| "grad_norm": 0.46797874569892883, | |
| "learning_rate": 2.0347749403470933e-05, | |
| "loss": 2.8797, | |
| "step": 2112 | |
| }, | |
| { | |
| "epoch": 1.4363636363636363, | |
| "grad_norm": 0.43796199560165405, | |
| "learning_rate": 2.0257355323965573e-05, | |
| "loss": 2.8097, | |
| "step": 2114 | |
| }, | |
| { | |
| "epoch": 1.4377230246389126, | |
| "grad_norm": 0.44776514172554016, | |
| "learning_rate": 2.0167111442686027e-05, | |
| "loss": 2.8539, | |
| "step": 2116 | |
| }, | |
| { | |
| "epoch": 1.4390824129141886, | |
| "grad_norm": 0.44064590334892273, | |
| "learning_rate": 2.0077018215357402e-05, | |
| "loss": 2.8498, | |
| "step": 2118 | |
| }, | |
| { | |
| "epoch": 1.4404418011894649, | |
| "grad_norm": 0.41469806432724, | |
| "learning_rate": 1.9987076096944023e-05, | |
| "loss": 2.7579, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 1.441801189464741, | |
| "grad_norm": 0.43275830149650574, | |
| "learning_rate": 1.9897285541647194e-05, | |
| "loss": 2.8156, | |
| "step": 2122 | |
| }, | |
| { | |
| "epoch": 1.443160577740017, | |
| "grad_norm": 0.45670434832572937, | |
| "learning_rate": 1.9807647002902736e-05, | |
| "loss": 2.8503, | |
| "step": 2124 | |
| }, | |
| { | |
| "epoch": 1.4445199660152932, | |
| "grad_norm": 0.45010295510292053, | |
| "learning_rate": 1.9718160933378904e-05, | |
| "loss": 2.8501, | |
| "step": 2126 | |
| }, | |
| { | |
| "epoch": 1.4458793542905692, | |
| "grad_norm": 0.45412638783454895, | |
| "learning_rate": 1.962882778497389e-05, | |
| "loss": 2.853, | |
| "step": 2128 | |
| }, | |
| { | |
| "epoch": 1.4472387425658453, | |
| "grad_norm": 0.40923628211021423, | |
| "learning_rate": 1.953964800881375e-05, | |
| "loss": 2.8493, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 1.4485981308411215, | |
| "grad_norm": 0.42023617029190063, | |
| "learning_rate": 1.9450622055249933e-05, | |
| "loss": 2.7737, | |
| "step": 2132 | |
| }, | |
| { | |
| "epoch": 1.4499575191163976, | |
| "grad_norm": 0.38607409596443176, | |
| "learning_rate": 1.9361750373857102e-05, | |
| "loss": 2.8204, | |
| "step": 2134 | |
| }, | |
| { | |
| "epoch": 1.4513169073916736, | |
| "grad_norm": 0.45782044529914856, | |
| "learning_rate": 1.927303341343095e-05, | |
| "loss": 2.8127, | |
| "step": 2136 | |
| }, | |
| { | |
| "epoch": 1.45267629566695, | |
| "grad_norm": 0.44400888681411743, | |
| "learning_rate": 1.918447162198575e-05, | |
| "loss": 2.8133, | |
| "step": 2138 | |
| }, | |
| { | |
| "epoch": 1.454035683942226, | |
| "grad_norm": 0.44565916061401367, | |
| "learning_rate": 1.9096065446752176e-05, | |
| "loss": 2.8483, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 1.4553950722175022, | |
| "grad_norm": 0.4803503751754761, | |
| "learning_rate": 1.9007815334175124e-05, | |
| "loss": 2.8845, | |
| "step": 2142 | |
| }, | |
| { | |
| "epoch": 1.4567544604927782, | |
| "grad_norm": 0.4118395745754242, | |
| "learning_rate": 1.8919721729911305e-05, | |
| "loss": 2.7629, | |
| "step": 2144 | |
| }, | |
| { | |
| "epoch": 1.4581138487680545, | |
| "grad_norm": 0.4230281710624695, | |
| "learning_rate": 1.8831785078827136e-05, | |
| "loss": 2.8507, | |
| "step": 2146 | |
| }, | |
| { | |
| "epoch": 1.4594732370433305, | |
| "grad_norm": 0.4142739176750183, | |
| "learning_rate": 1.874400582499642e-05, | |
| "loss": 2.7689, | |
| "step": 2148 | |
| }, | |
| { | |
| "epoch": 1.4608326253186066, | |
| "grad_norm": 0.46020829677581787, | |
| "learning_rate": 1.8656384411698087e-05, | |
| "loss": 2.7747, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.4621920135938828, | |
| "grad_norm": 0.45219558477401733, | |
| "learning_rate": 1.8568921281413983e-05, | |
| "loss": 2.8217, | |
| "step": 2152 | |
| }, | |
| { | |
| "epoch": 1.4635514018691589, | |
| "grad_norm": 0.39816343784332275, | |
| "learning_rate": 1.848161687582667e-05, | |
| "loss": 2.8177, | |
| "step": 2154 | |
| }, | |
| { | |
| "epoch": 1.464910790144435, | |
| "grad_norm": 0.40539437532424927, | |
| "learning_rate": 1.8394471635817156e-05, | |
| "loss": 2.7834, | |
| "step": 2156 | |
| }, | |
| { | |
| "epoch": 1.4662701784197112, | |
| "grad_norm": 0.4099854528903961, | |
| "learning_rate": 1.8307486001462655e-05, | |
| "loss": 2.7991, | |
| "step": 2158 | |
| }, | |
| { | |
| "epoch": 1.4676295666949872, | |
| "grad_norm": 0.39933493733406067, | |
| "learning_rate": 1.822066041203437e-05, | |
| "loss": 2.9055, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 1.4689889549702633, | |
| "grad_norm": 0.3902811110019684, | |
| "learning_rate": 1.8133995305995344e-05, | |
| "loss": 2.8591, | |
| "step": 2162 | |
| }, | |
| { | |
| "epoch": 1.4703483432455395, | |
| "grad_norm": 0.40793663263320923, | |
| "learning_rate": 1.8047491120998123e-05, | |
| "loss": 2.7323, | |
| "step": 2164 | |
| }, | |
| { | |
| "epoch": 1.4717077315208156, | |
| "grad_norm": 0.5329314470291138, | |
| "learning_rate": 1.7961148293882663e-05, | |
| "loss": 2.8121, | |
| "step": 2166 | |
| }, | |
| { | |
| "epoch": 1.4730671197960918, | |
| "grad_norm": 0.4276919364929199, | |
| "learning_rate": 1.7874967260674075e-05, | |
| "loss": 2.859, | |
| "step": 2168 | |
| }, | |
| { | |
| "epoch": 1.4744265080713679, | |
| "grad_norm": 0.40856489539146423, | |
| "learning_rate": 1.778894845658039e-05, | |
| "loss": 2.8251, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 1.4757858963466441, | |
| "grad_norm": 0.41639694571495056, | |
| "learning_rate": 1.7703092315990393e-05, | |
| "loss": 2.8238, | |
| "step": 2172 | |
| }, | |
| { | |
| "epoch": 1.4771452846219202, | |
| "grad_norm": 0.40431302785873413, | |
| "learning_rate": 1.7617399272471457e-05, | |
| "loss": 2.8933, | |
| "step": 2174 | |
| }, | |
| { | |
| "epoch": 1.4785046728971962, | |
| "grad_norm": 0.4122006595134735, | |
| "learning_rate": 1.753186975876734e-05, | |
| "loss": 2.8219, | |
| "step": 2176 | |
| }, | |
| { | |
| "epoch": 1.4798640611724725, | |
| "grad_norm": 0.39818820357322693, | |
| "learning_rate": 1.7446504206795922e-05, | |
| "loss": 2.8671, | |
| "step": 2178 | |
| }, | |
| { | |
| "epoch": 1.4812234494477485, | |
| "grad_norm": 0.4175485372543335, | |
| "learning_rate": 1.7361303047647166e-05, | |
| "loss": 2.8011, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 1.4825828377230246, | |
| "grad_norm": 0.40247878432273865, | |
| "learning_rate": 1.72762667115808e-05, | |
| "loss": 2.7351, | |
| "step": 2182 | |
| }, | |
| { | |
| "epoch": 1.4839422259983008, | |
| "grad_norm": 0.38898593187332153, | |
| "learning_rate": 1.719139562802422e-05, | |
| "loss": 2.7965, | |
| "step": 2184 | |
| }, | |
| { | |
| "epoch": 1.4853016142735769, | |
| "grad_norm": 0.42152050137519836, | |
| "learning_rate": 1.710669022557034e-05, | |
| "loss": 2.8801, | |
| "step": 2186 | |
| }, | |
| { | |
| "epoch": 1.486661002548853, | |
| "grad_norm": 0.3899691700935364, | |
| "learning_rate": 1.7022150931975394e-05, | |
| "loss": 2.7555, | |
| "step": 2188 | |
| }, | |
| { | |
| "epoch": 1.4880203908241292, | |
| "grad_norm": 0.4126317799091339, | |
| "learning_rate": 1.693777817415674e-05, | |
| "loss": 2.7941, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 1.4893797790994052, | |
| "grad_norm": 0.3750491440296173, | |
| "learning_rate": 1.6853572378190725e-05, | |
| "loss": 2.819, | |
| "step": 2192 | |
| }, | |
| { | |
| "epoch": 1.4907391673746813, | |
| "grad_norm": 0.4226130545139313, | |
| "learning_rate": 1.6769533969310664e-05, | |
| "loss": 2.9102, | |
| "step": 2194 | |
| }, | |
| { | |
| "epoch": 1.4920985556499575, | |
| "grad_norm": 0.43610116839408875, | |
| "learning_rate": 1.668566337190447e-05, | |
| "loss": 2.8815, | |
| "step": 2196 | |
| }, | |
| { | |
| "epoch": 1.4934579439252336, | |
| "grad_norm": 0.44443652033805847, | |
| "learning_rate": 1.6601961009512635e-05, | |
| "loss": 2.8859, | |
| "step": 2198 | |
| }, | |
| { | |
| "epoch": 1.4948173322005098, | |
| "grad_norm": 0.38702818751335144, | |
| "learning_rate": 1.651842730482613e-05, | |
| "loss": 2.8196, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.4961767204757859, | |
| "grad_norm": 0.41439834237098694, | |
| "learning_rate": 1.6435062679684144e-05, | |
| "loss": 2.8981, | |
| "step": 2202 | |
| }, | |
| { | |
| "epoch": 1.4975361087510621, | |
| "grad_norm": 0.39517101645469666, | |
| "learning_rate": 1.6351867555072096e-05, | |
| "loss": 2.8214, | |
| "step": 2204 | |
| }, | |
| { | |
| "epoch": 1.4988954970263382, | |
| "grad_norm": 0.40214547514915466, | |
| "learning_rate": 1.626884235111939e-05, | |
| "loss": 2.8655, | |
| "step": 2206 | |
| }, | |
| { | |
| "epoch": 1.5002548853016142, | |
| "grad_norm": 0.38211849331855774, | |
| "learning_rate": 1.6185987487097377e-05, | |
| "loss": 2.8034, | |
| "step": 2208 | |
| }, | |
| { | |
| "epoch": 1.5016142735768905, | |
| "grad_norm": 0.39007025957107544, | |
| "learning_rate": 1.610330338141716e-05, | |
| "loss": 2.8185, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 1.5029736618521665, | |
| "grad_norm": 0.38158321380615234, | |
| "learning_rate": 1.6020790451627576e-05, | |
| "loss": 2.8079, | |
| "step": 2212 | |
| }, | |
| { | |
| "epoch": 1.5043330501274426, | |
| "grad_norm": 0.4224998950958252, | |
| "learning_rate": 1.5938449114413013e-05, | |
| "loss": 2.847, | |
| "step": 2214 | |
| }, | |
| { | |
| "epoch": 1.5056924384027188, | |
| "grad_norm": 0.3995256721973419, | |
| "learning_rate": 1.585627978559132e-05, | |
| "loss": 2.797, | |
| "step": 2216 | |
| }, | |
| { | |
| "epoch": 1.5070518266779949, | |
| "grad_norm": 0.42583292722702026, | |
| "learning_rate": 1.577428288011171e-05, | |
| "loss": 2.8205, | |
| "step": 2218 | |
| }, | |
| { | |
| "epoch": 1.508411214953271, | |
| "grad_norm": 0.41680529713630676, | |
| "learning_rate": 1.5692458812052714e-05, | |
| "loss": 2.8584, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 1.5097706032285472, | |
| "grad_norm": 0.40372368693351746, | |
| "learning_rate": 1.561080799461998e-05, | |
| "loss": 2.7948, | |
| "step": 2222 | |
| }, | |
| { | |
| "epoch": 1.5111299915038234, | |
| "grad_norm": 0.4342186152935028, | |
| "learning_rate": 1.552933084014431e-05, | |
| "loss": 2.8211, | |
| "step": 2224 | |
| }, | |
| { | |
| "epoch": 1.5124893797790993, | |
| "grad_norm": 0.4139116108417511, | |
| "learning_rate": 1.544802776007951e-05, | |
| "loss": 2.8423, | |
| "step": 2226 | |
| }, | |
| { | |
| "epoch": 1.5138487680543755, | |
| "grad_norm": 0.436837375164032, | |
| "learning_rate": 1.5366899165000305e-05, | |
| "loss": 2.7988, | |
| "step": 2228 | |
| }, | |
| { | |
| "epoch": 1.5152081563296518, | |
| "grad_norm": 0.4361181855201721, | |
| "learning_rate": 1.5285945464600264e-05, | |
| "loss": 2.8399, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 1.5165675446049278, | |
| "grad_norm": 0.4215008318424225, | |
| "learning_rate": 1.5205167067689802e-05, | |
| "loss": 2.8689, | |
| "step": 2232 | |
| }, | |
| { | |
| "epoch": 1.5179269328802039, | |
| "grad_norm": 0.450169175863266, | |
| "learning_rate": 1.5124564382194052e-05, | |
| "loss": 2.9058, | |
| "step": 2234 | |
| }, | |
| { | |
| "epoch": 1.5192863211554801, | |
| "grad_norm": 0.4369889497756958, | |
| "learning_rate": 1.5044137815150788e-05, | |
| "loss": 2.8181, | |
| "step": 2236 | |
| }, | |
| { | |
| "epoch": 1.5206457094307562, | |
| "grad_norm": 0.4108685255050659, | |
| "learning_rate": 1.4963887772708397e-05, | |
| "loss": 2.8799, | |
| "step": 2238 | |
| }, | |
| { | |
| "epoch": 1.5220050977060322, | |
| "grad_norm": 0.39875927567481995, | |
| "learning_rate": 1.4883814660123879e-05, | |
| "loss": 2.7797, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 1.5233644859813085, | |
| "grad_norm": 0.404422789812088, | |
| "learning_rate": 1.4803918881760692e-05, | |
| "loss": 2.8349, | |
| "step": 2242 | |
| }, | |
| { | |
| "epoch": 1.5247238742565845, | |
| "grad_norm": 0.42393773794174194, | |
| "learning_rate": 1.47242008410868e-05, | |
| "loss": 2.907, | |
| "step": 2244 | |
| }, | |
| { | |
| "epoch": 1.5260832625318606, | |
| "grad_norm": 0.3849833011627197, | |
| "learning_rate": 1.4644660940672627e-05, | |
| "loss": 2.9181, | |
| "step": 2246 | |
| }, | |
| { | |
| "epoch": 1.5274426508071368, | |
| "grad_norm": 0.3820267617702484, | |
| "learning_rate": 1.4565299582188952e-05, | |
| "loss": 2.7744, | |
| "step": 2248 | |
| }, | |
| { | |
| "epoch": 1.528802039082413, | |
| "grad_norm": 0.3930812180042267, | |
| "learning_rate": 1.4486117166404938e-05, | |
| "loss": 2.8334, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.530161427357689, | |
| "grad_norm": 0.4204096496105194, | |
| "learning_rate": 1.4407114093186136e-05, | |
| "loss": 2.8429, | |
| "step": 2252 | |
| }, | |
| { | |
| "epoch": 1.5315208156329652, | |
| "grad_norm": 0.379271924495697, | |
| "learning_rate": 1.4328290761492414e-05, | |
| "loss": 2.8559, | |
| "step": 2254 | |
| }, | |
| { | |
| "epoch": 1.5328802039082414, | |
| "grad_norm": 0.3917960226535797, | |
| "learning_rate": 1.4249647569375928e-05, | |
| "loss": 2.7823, | |
| "step": 2256 | |
| }, | |
| { | |
| "epoch": 1.5342395921835175, | |
| "grad_norm": 0.4033399820327759, | |
| "learning_rate": 1.4171184913979196e-05, | |
| "loss": 2.8707, | |
| "step": 2258 | |
| }, | |
| { | |
| "epoch": 1.5355989804587935, | |
| "grad_norm": 0.3916587233543396, | |
| "learning_rate": 1.4092903191532996e-05, | |
| "loss": 2.8717, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 1.5369583687340698, | |
| "grad_norm": 0.4255055785179138, | |
| "learning_rate": 1.4014802797354404e-05, | |
| "loss": 2.8225, | |
| "step": 2262 | |
| }, | |
| { | |
| "epoch": 1.5383177570093458, | |
| "grad_norm": 0.40315207839012146, | |
| "learning_rate": 1.3936884125844828e-05, | |
| "loss": 2.821, | |
| "step": 2264 | |
| }, | |
| { | |
| "epoch": 1.5396771452846219, | |
| "grad_norm": 0.38984397053718567, | |
| "learning_rate": 1.3859147570488e-05, | |
| "loss": 2.8274, | |
| "step": 2266 | |
| }, | |
| { | |
| "epoch": 1.5410365335598981, | |
| "grad_norm": 0.41258957982063293, | |
| "learning_rate": 1.3781593523847946e-05, | |
| "loss": 2.8562, | |
| "step": 2268 | |
| }, | |
| { | |
| "epoch": 1.5423959218351742, | |
| "grad_norm": 0.38504886627197266, | |
| "learning_rate": 1.3704222377567022e-05, | |
| "loss": 2.8142, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 1.5437553101104502, | |
| "grad_norm": 0.3893933892250061, | |
| "learning_rate": 1.3627034522364002e-05, | |
| "loss": 2.8162, | |
| "step": 2272 | |
| }, | |
| { | |
| "epoch": 1.5451146983857265, | |
| "grad_norm": 0.3827742338180542, | |
| "learning_rate": 1.355003034803204e-05, | |
| "loss": 2.8243, | |
| "step": 2274 | |
| }, | |
| { | |
| "epoch": 1.5464740866610025, | |
| "grad_norm": 0.3980223536491394, | |
| "learning_rate": 1.3473210243436673e-05, | |
| "loss": 2.84, | |
| "step": 2276 | |
| }, | |
| { | |
| "epoch": 1.5478334749362785, | |
| "grad_norm": 0.3831814229488373, | |
| "learning_rate": 1.339657459651395e-05, | |
| "loss": 2.823, | |
| "step": 2278 | |
| }, | |
| { | |
| "epoch": 1.5491928632115548, | |
| "grad_norm": 0.3754749000072479, | |
| "learning_rate": 1.3320123794268396e-05, | |
| "loss": 2.8656, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 1.550552251486831, | |
| "grad_norm": 0.4047335386276245, | |
| "learning_rate": 1.3243858222771067e-05, | |
| "loss": 2.865, | |
| "step": 2282 | |
| }, | |
| { | |
| "epoch": 1.5519116397621069, | |
| "grad_norm": 0.3802129030227661, | |
| "learning_rate": 1.3167778267157648e-05, | |
| "loss": 2.7413, | |
| "step": 2284 | |
| }, | |
| { | |
| "epoch": 1.5532710280373832, | |
| "grad_norm": 0.4164421260356903, | |
| "learning_rate": 1.3091884311626495e-05, | |
| "loss": 2.8057, | |
| "step": 2286 | |
| }, | |
| { | |
| "epoch": 1.5546304163126594, | |
| "grad_norm": 0.4302055835723877, | |
| "learning_rate": 1.3016176739436609e-05, | |
| "loss": 2.8473, | |
| "step": 2288 | |
| }, | |
| { | |
| "epoch": 1.5559898045879355, | |
| "grad_norm": 0.39841482043266296, | |
| "learning_rate": 1.2940655932905844e-05, | |
| "loss": 2.7838, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 1.5573491928632115, | |
| "grad_norm": 0.4039965271949768, | |
| "learning_rate": 1.2865322273408826e-05, | |
| "loss": 2.8177, | |
| "step": 2292 | |
| }, | |
| { | |
| "epoch": 1.5587085811384878, | |
| "grad_norm": 0.40445390343666077, | |
| "learning_rate": 1.2790176141375177e-05, | |
| "loss": 2.8335, | |
| "step": 2294 | |
| }, | |
| { | |
| "epoch": 1.5600679694137638, | |
| "grad_norm": 0.3808634579181671, | |
| "learning_rate": 1.2715217916287453e-05, | |
| "loss": 2.7455, | |
| "step": 2296 | |
| }, | |
| { | |
| "epoch": 1.5614273576890398, | |
| "grad_norm": 0.381149560213089, | |
| "learning_rate": 1.2640447976679353e-05, | |
| "loss": 2.8425, | |
| "step": 2298 | |
| }, | |
| { | |
| "epoch": 1.562786745964316, | |
| "grad_norm": 0.402127206325531, | |
| "learning_rate": 1.2565866700133705e-05, | |
| "loss": 2.8608, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.5641461342395921, | |
| "grad_norm": 0.3847315311431885, | |
| "learning_rate": 1.2491474463280579e-05, | |
| "loss": 2.8103, | |
| "step": 2302 | |
| }, | |
| { | |
| "epoch": 1.5655055225148682, | |
| "grad_norm": 0.39522722363471985, | |
| "learning_rate": 1.2417271641795508e-05, | |
| "loss": 2.7478, | |
| "step": 2304 | |
| }, | |
| { | |
| "epoch": 1.5668649107901444, | |
| "grad_norm": 0.3950273096561432, | |
| "learning_rate": 1.2343258610397396e-05, | |
| "loss": 2.8867, | |
| "step": 2306 | |
| }, | |
| { | |
| "epoch": 1.5682242990654207, | |
| "grad_norm": 0.3833482563495636, | |
| "learning_rate": 1.2269435742846735e-05, | |
| "loss": 2.8769, | |
| "step": 2308 | |
| }, | |
| { | |
| "epoch": 1.5695836873406965, | |
| "grad_norm": 0.37314730882644653, | |
| "learning_rate": 1.219580341194374e-05, | |
| "loss": 2.8027, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 1.5709430756159728, | |
| "grad_norm": 0.3930293023586273, | |
| "learning_rate": 1.2122361989526421e-05, | |
| "loss": 2.8316, | |
| "step": 2312 | |
| }, | |
| { | |
| "epoch": 1.572302463891249, | |
| "grad_norm": 0.37856975197792053, | |
| "learning_rate": 1.204911184646868e-05, | |
| "loss": 2.796, | |
| "step": 2314 | |
| }, | |
| { | |
| "epoch": 1.573661852166525, | |
| "grad_norm": 0.38083016872406006, | |
| "learning_rate": 1.197605335267849e-05, | |
| "loss": 2.8333, | |
| "step": 2316 | |
| }, | |
| { | |
| "epoch": 1.5750212404418011, | |
| "grad_norm": 0.37718525528907776, | |
| "learning_rate": 1.190318687709604e-05, | |
| "loss": 2.8358, | |
| "step": 2318 | |
| }, | |
| { | |
| "epoch": 1.5763806287170774, | |
| "grad_norm": 0.4028748869895935, | |
| "learning_rate": 1.1830512787691783e-05, | |
| "loss": 2.8045, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 1.5777400169923534, | |
| "grad_norm": 0.3694095015525818, | |
| "learning_rate": 1.175803145146468e-05, | |
| "loss": 2.8986, | |
| "step": 2322 | |
| }, | |
| { | |
| "epoch": 1.5790994052676295, | |
| "grad_norm": 0.38383910059928894, | |
| "learning_rate": 1.1685743234440305e-05, | |
| "loss": 2.8858, | |
| "step": 2324 | |
| }, | |
| { | |
| "epoch": 1.5804587935429057, | |
| "grad_norm": 0.37843042612075806, | |
| "learning_rate": 1.1613648501668961e-05, | |
| "loss": 2.8596, | |
| "step": 2326 | |
| }, | |
| { | |
| "epoch": 1.5818181818181818, | |
| "grad_norm": 0.37164831161499023, | |
| "learning_rate": 1.1541747617223874e-05, | |
| "loss": 2.8026, | |
| "step": 2328 | |
| }, | |
| { | |
| "epoch": 1.5831775700934578, | |
| "grad_norm": 0.39616233110427856, | |
| "learning_rate": 1.1470040944199373e-05, | |
| "loss": 2.8134, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 1.584536958368734, | |
| "grad_norm": 0.3910829424858093, | |
| "learning_rate": 1.1398528844709039e-05, | |
| "loss": 2.8525, | |
| "step": 2332 | |
| }, | |
| { | |
| "epoch": 1.5858963466440104, | |
| "grad_norm": 0.3922969400882721, | |
| "learning_rate": 1.1327211679883825e-05, | |
| "loss": 2.8271, | |
| "step": 2334 | |
| }, | |
| { | |
| "epoch": 1.5872557349192862, | |
| "grad_norm": 0.44064638018608093, | |
| "learning_rate": 1.1256089809870334e-05, | |
| "loss": 2.9043, | |
| "step": 2336 | |
| }, | |
| { | |
| "epoch": 1.5886151231945624, | |
| "grad_norm": 0.4038518965244293, | |
| "learning_rate": 1.1185163593828902e-05, | |
| "loss": 2.8445, | |
| "step": 2338 | |
| }, | |
| { | |
| "epoch": 1.5899745114698387, | |
| "grad_norm": 0.4049074351787567, | |
| "learning_rate": 1.1114433389931827e-05, | |
| "loss": 2.8259, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 1.5913338997451147, | |
| "grad_norm": 0.38776373863220215, | |
| "learning_rate": 1.1043899555361592e-05, | |
| "loss": 2.7772, | |
| "step": 2342 | |
| }, | |
| { | |
| "epoch": 1.5926932880203908, | |
| "grad_norm": 0.3804343640804291, | |
| "learning_rate": 1.0973562446309016e-05, | |
| "loss": 2.7711, | |
| "step": 2344 | |
| }, | |
| { | |
| "epoch": 1.594052676295667, | |
| "grad_norm": 0.37529852986335754, | |
| "learning_rate": 1.0903422417971459e-05, | |
| "loss": 2.836, | |
| "step": 2346 | |
| }, | |
| { | |
| "epoch": 1.595412064570943, | |
| "grad_norm": 0.3775539994239807, | |
| "learning_rate": 1.0833479824551019e-05, | |
| "loss": 2.7873, | |
| "step": 2348 | |
| }, | |
| { | |
| "epoch": 1.5967714528462191, | |
| "grad_norm": 0.3730499744415283, | |
| "learning_rate": 1.0763735019252802e-05, | |
| "loss": 2.7707, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.5981308411214954, | |
| "grad_norm": 0.3781737983226776, | |
| "learning_rate": 1.0694188354283096e-05, | |
| "loss": 2.854, | |
| "step": 2352 | |
| }, | |
| { | |
| "epoch": 1.5994902293967714, | |
| "grad_norm": 0.3752993941307068, | |
| "learning_rate": 1.0624840180847551e-05, | |
| "loss": 2.804, | |
| "step": 2354 | |
| }, | |
| { | |
| "epoch": 1.6008496176720475, | |
| "grad_norm": 0.36867862939834595, | |
| "learning_rate": 1.0555690849149497e-05, | |
| "loss": 2.8384, | |
| "step": 2356 | |
| }, | |
| { | |
| "epoch": 1.6022090059473237, | |
| "grad_norm": 0.37930601835250854, | |
| "learning_rate": 1.0486740708388104e-05, | |
| "loss": 2.8379, | |
| "step": 2358 | |
| }, | |
| { | |
| "epoch": 1.6035683942225998, | |
| "grad_norm": 0.371690571308136, | |
| "learning_rate": 1.0417990106756609e-05, | |
| "loss": 2.823, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 1.6049277824978758, | |
| "grad_norm": 0.36744946241378784, | |
| "learning_rate": 1.034943939144069e-05, | |
| "loss": 2.7879, | |
| "step": 2362 | |
| }, | |
| { | |
| "epoch": 1.606287170773152, | |
| "grad_norm": 0.3579390048980713, | |
| "learning_rate": 1.0281088908616521e-05, | |
| "loss": 2.8361, | |
| "step": 2364 | |
| }, | |
| { | |
| "epoch": 1.6076465590484283, | |
| "grad_norm": 0.39030399918556213, | |
| "learning_rate": 1.0212939003449128e-05, | |
| "loss": 2.8286, | |
| "step": 2366 | |
| }, | |
| { | |
| "epoch": 1.6090059473237044, | |
| "grad_norm": 0.38567525148391724, | |
| "learning_rate": 1.0144990020090683e-05, | |
| "loss": 2.8185, | |
| "step": 2368 | |
| }, | |
| { | |
| "epoch": 1.6103653355989804, | |
| "grad_norm": 0.370043009519577, | |
| "learning_rate": 1.0077242301678657e-05, | |
| "loss": 2.7847, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 1.6117247238742567, | |
| "grad_norm": 0.40155136585235596, | |
| "learning_rate": 1.0009696190334206e-05, | |
| "loss": 2.852, | |
| "step": 2372 | |
| }, | |
| { | |
| "epoch": 1.6130841121495327, | |
| "grad_norm": 0.3666565418243408, | |
| "learning_rate": 9.942352027160323e-06, | |
| "loss": 2.8063, | |
| "step": 2374 | |
| }, | |
| { | |
| "epoch": 1.6144435004248088, | |
| "grad_norm": 0.38090911507606506, | |
| "learning_rate": 9.875210152240244e-06, | |
| "loss": 2.8612, | |
| "step": 2376 | |
| }, | |
| { | |
| "epoch": 1.615802888700085, | |
| "grad_norm": 0.3603736162185669, | |
| "learning_rate": 9.808270904635613e-06, | |
| "loss": 2.8067, | |
| "step": 2378 | |
| }, | |
| { | |
| "epoch": 1.617162276975361, | |
| "grad_norm": 0.37685248255729675, | |
| "learning_rate": 9.741534622384806e-06, | |
| "loss": 2.7653, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 1.6185216652506371, | |
| "grad_norm": 0.36234524846076965, | |
| "learning_rate": 9.67500164250133e-06, | |
| "loss": 2.788, | |
| "step": 2382 | |
| }, | |
| { | |
| "epoch": 1.6198810535259134, | |
| "grad_norm": 0.3880058228969574, | |
| "learning_rate": 9.608672300971938e-06, | |
| "loss": 2.7644, | |
| "step": 2384 | |
| }, | |
| { | |
| "epoch": 1.6212404418011894, | |
| "grad_norm": 0.39096465706825256, | |
| "learning_rate": 9.542546932755043e-06, | |
| "loss": 2.8435, | |
| "step": 2386 | |
| }, | |
| { | |
| "epoch": 1.6225998300764655, | |
| "grad_norm": 0.37929651141166687, | |
| "learning_rate": 9.476625871779044e-06, | |
| "loss": 2.8432, | |
| "step": 2388 | |
| }, | |
| { | |
| "epoch": 1.6239592183517417, | |
| "grad_norm": 0.3924437463283539, | |
| "learning_rate": 9.410909450940557e-06, | |
| "loss": 2.794, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 1.625318606627018, | |
| "grad_norm": 0.38013944029808044, | |
| "learning_rate": 9.345398002102812e-06, | |
| "loss": 2.8715, | |
| "step": 2392 | |
| }, | |
| { | |
| "epoch": 1.6266779949022938, | |
| "grad_norm": 0.3609662353992462, | |
| "learning_rate": 9.280091856093941e-06, | |
| "loss": 2.8452, | |
| "step": 2394 | |
| }, | |
| { | |
| "epoch": 1.62803738317757, | |
| "grad_norm": 0.3805861175060272, | |
| "learning_rate": 9.214991342705303e-06, | |
| "loss": 2.8205, | |
| "step": 2396 | |
| }, | |
| { | |
| "epoch": 1.6293967714528463, | |
| "grad_norm": 0.37913620471954346, | |
| "learning_rate": 9.150096790689804e-06, | |
| "loss": 2.7875, | |
| "step": 2398 | |
| }, | |
| { | |
| "epoch": 1.6307561597281224, | |
| "grad_norm": 0.38181012868881226, | |
| "learning_rate": 9.085408527760308e-06, | |
| "loss": 2.7525, | |
| "step": 2400 | |
| } | |
| ], | |
| "logging_steps": 2, | |
| "max_steps": 2944, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 300, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.701195881339683e+18, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |