| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.991296779808529, | |
| "eval_steps": 500, | |
| "global_step": 5740, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0017406440382941688, | |
| "grad_norm": 600.0, | |
| "learning_rate": 3.4843205574912896e-07, | |
| "loss": 32.6735, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.008703220191470844, | |
| "grad_norm": 548.0, | |
| "learning_rate": 1.7421602787456445e-06, | |
| "loss": 30.8839, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.017406440382941687, | |
| "grad_norm": 338.0, | |
| "learning_rate": 3.484320557491289e-06, | |
| "loss": 28.8366, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02610966057441253, | |
| "grad_norm": 197.0, | |
| "learning_rate": 5.226480836236934e-06, | |
| "loss": 24.2042, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.034812880765883375, | |
| "grad_norm": 82.5, | |
| "learning_rate": 6.968641114982578e-06, | |
| "loss": 20.0294, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04351610095735422, | |
| "grad_norm": 52.75, | |
| "learning_rate": 8.710801393728225e-06, | |
| "loss": 18.1416, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.05221932114882506, | |
| "grad_norm": 25.375, | |
| "learning_rate": 1.0452961672473868e-05, | |
| "loss": 16.3408, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.060922541340295906, | |
| "grad_norm": 11.5625, | |
| "learning_rate": 1.2195121951219513e-05, | |
| "loss": 15.4579, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.06962576153176675, | |
| "grad_norm": 8.1875, | |
| "learning_rate": 1.3937282229965156e-05, | |
| "loss": 14.8065, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0783289817232376, | |
| "grad_norm": 6.8125, | |
| "learning_rate": 1.56794425087108e-05, | |
| "loss": 14.3068, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.08703220191470844, | |
| "grad_norm": 8.4375, | |
| "learning_rate": 1.742160278745645e-05, | |
| "loss": 13.9, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09573542210617929, | |
| "grad_norm": 12.875, | |
| "learning_rate": 1.9163763066202093e-05, | |
| "loss": 13.1213, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.10443864229765012, | |
| "grad_norm": 21.25, | |
| "learning_rate": 2.0905923344947736e-05, | |
| "loss": 12.0425, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11314186248912098, | |
| "grad_norm": 44.5, | |
| "learning_rate": 2.264808362369338e-05, | |
| "loss": 10.0004, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.12184508268059181, | |
| "grad_norm": 42.75, | |
| "learning_rate": 2.4390243902439026e-05, | |
| "loss": 5.8601, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.13054830287206268, | |
| "grad_norm": 5.03125, | |
| "learning_rate": 2.6132404181184672e-05, | |
| "loss": 2.3269, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1392515230635335, | |
| "grad_norm": 3.046875, | |
| "learning_rate": 2.7874564459930312e-05, | |
| "loss": 1.8038, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.14795474325500435, | |
| "grad_norm": 2.890625, | |
| "learning_rate": 2.9616724738675962e-05, | |
| "loss": 1.6296, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.1566579634464752, | |
| "grad_norm": 2.421875, | |
| "learning_rate": 3.13588850174216e-05, | |
| "loss": 1.4928, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.16536118363794605, | |
| "grad_norm": 8.0, | |
| "learning_rate": 3.310104529616725e-05, | |
| "loss": 1.4006, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.17406440382941687, | |
| "grad_norm": 3.328125, | |
| "learning_rate": 3.48432055749129e-05, | |
| "loss": 1.366, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.18276762402088773, | |
| "grad_norm": 1.6875, | |
| "learning_rate": 3.6585365853658535e-05, | |
| "loss": 1.2998, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.19147084421235858, | |
| "grad_norm": 5.40625, | |
| "learning_rate": 3.8327526132404185e-05, | |
| "loss": 1.2497, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.20017406440382943, | |
| "grad_norm": 2.1875, | |
| "learning_rate": 4.006968641114983e-05, | |
| "loss": 1.2106, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.20887728459530025, | |
| "grad_norm": 5.65625, | |
| "learning_rate": 4.181184668989547e-05, | |
| "loss": 1.2119, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2175805047867711, | |
| "grad_norm": 22.75, | |
| "learning_rate": 4.3554006968641115e-05, | |
| "loss": 1.1897, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.22628372497824195, | |
| "grad_norm": 16.375, | |
| "learning_rate": 4.529616724738676e-05, | |
| "loss": 1.1921, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2349869451697128, | |
| "grad_norm": 11.1875, | |
| "learning_rate": 4.703832752613241e-05, | |
| "loss": 1.1743, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.24369016536118362, | |
| "grad_norm": 9.6875, | |
| "learning_rate": 4.878048780487805e-05, | |
| "loss": 1.1545, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.2523933855526545, | |
| "grad_norm": 5.21875, | |
| "learning_rate": 5.0522648083623695e-05, | |
| "loss": 1.1263, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.26109660574412535, | |
| "grad_norm": 2.34375, | |
| "learning_rate": 5.2264808362369345e-05, | |
| "loss": 1.1232, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.26979982593559615, | |
| "grad_norm": 8.4375, | |
| "learning_rate": 5.400696864111499e-05, | |
| "loss": 1.0888, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.278503046127067, | |
| "grad_norm": 3.3125, | |
| "learning_rate": 5.5749128919860624e-05, | |
| "loss": 1.0734, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.28720626631853785, | |
| "grad_norm": 1.5078125, | |
| "learning_rate": 5.749128919860628e-05, | |
| "loss": 1.0545, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.2959094865100087, | |
| "grad_norm": 5.5625, | |
| "learning_rate": 5.9233449477351924e-05, | |
| "loss": 1.0451, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.30461270670147955, | |
| "grad_norm": 6.03125, | |
| "learning_rate": 6.097560975609756e-05, | |
| "loss": 1.023, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.3133159268929504, | |
| "grad_norm": 3.78125, | |
| "learning_rate": 6.27177700348432e-05, | |
| "loss": 1.0475, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.32201914708442125, | |
| "grad_norm": 1.8984375, | |
| "learning_rate": 6.445993031358886e-05, | |
| "loss": 1.008, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.3307223672758921, | |
| "grad_norm": 6.46875, | |
| "learning_rate": 6.62020905923345e-05, | |
| "loss": 0.9875, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.3394255874673629, | |
| "grad_norm": 3.3125, | |
| "learning_rate": 6.794425087108013e-05, | |
| "loss": 1.0139, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.34812880765883375, | |
| "grad_norm": 1.8828125, | |
| "learning_rate": 6.96864111498258e-05, | |
| "loss": 0.9675, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3568320278503046, | |
| "grad_norm": 2.25, | |
| "learning_rate": 7.142857142857143e-05, | |
| "loss": 0.9668, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.36553524804177545, | |
| "grad_norm": 2.28125, | |
| "learning_rate": 7.317073170731707e-05, | |
| "loss": 0.9589, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.3742384682332463, | |
| "grad_norm": 1.6484375, | |
| "learning_rate": 7.491289198606272e-05, | |
| "loss": 0.9686, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.38294168842471715, | |
| "grad_norm": 1.328125, | |
| "learning_rate": 7.665505226480837e-05, | |
| "loss": 0.9702, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.391644908616188, | |
| "grad_norm": 2.875, | |
| "learning_rate": 7.839721254355401e-05, | |
| "loss": 0.9665, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.40034812880765885, | |
| "grad_norm": 4.9375, | |
| "learning_rate": 8.013937282229966e-05, | |
| "loss": 0.9598, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.4090513489991297, | |
| "grad_norm": 3.359375, | |
| "learning_rate": 8.188153310104531e-05, | |
| "loss": 0.944, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.4177545691906005, | |
| "grad_norm": 2.484375, | |
| "learning_rate": 8.362369337979094e-05, | |
| "loss": 0.9368, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.42645778938207135, | |
| "grad_norm": 3.625, | |
| "learning_rate": 8.53658536585366e-05, | |
| "loss": 0.9467, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.4351610095735422, | |
| "grad_norm": 2.015625, | |
| "learning_rate": 8.710801393728223e-05, | |
| "loss": 0.9282, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.44386422976501305, | |
| "grad_norm": 5.0, | |
| "learning_rate": 8.885017421602788e-05, | |
| "loss": 0.9065, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.4525674499564839, | |
| "grad_norm": 1.6796875, | |
| "learning_rate": 9.059233449477352e-05, | |
| "loss": 0.911, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.46127067014795475, | |
| "grad_norm": 1.1328125, | |
| "learning_rate": 9.233449477351917e-05, | |
| "loss": 0.9055, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.4699738903394256, | |
| "grad_norm": 1.0625, | |
| "learning_rate": 9.407665505226482e-05, | |
| "loss": 0.9053, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.47867711053089645, | |
| "grad_norm": 2.703125, | |
| "learning_rate": 9.581881533101045e-05, | |
| "loss": 0.8986, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.48738033072236725, | |
| "grad_norm": 2.734375, | |
| "learning_rate": 9.75609756097561e-05, | |
| "loss": 0.8968, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.4960835509138381, | |
| "grad_norm": 2.53125, | |
| "learning_rate": 9.930313588850174e-05, | |
| "loss": 0.9147, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.504786771105309, | |
| "grad_norm": 2.84375, | |
| "learning_rate": 0.00010104529616724739, | |
| "loss": 0.8892, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.5134899912967799, | |
| "grad_norm": 2.296875, | |
| "learning_rate": 0.00010278745644599304, | |
| "loss": 0.8896, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.5221932114882507, | |
| "grad_norm": 0.87109375, | |
| "learning_rate": 0.00010452961672473869, | |
| "loss": 0.8926, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.5308964316797214, | |
| "grad_norm": 1.2890625, | |
| "learning_rate": 0.00010627177700348431, | |
| "loss": 0.8943, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.5395996518711923, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 0.00010801393728222998, | |
| "loss": 0.8807, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.5483028720626631, | |
| "grad_norm": 20.125, | |
| "learning_rate": 0.00010975609756097563, | |
| "loss": 0.882, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.557006092254134, | |
| "grad_norm": 3.609375, | |
| "learning_rate": 0.00011149825783972125, | |
| "loss": 0.9149, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.5657093124456049, | |
| "grad_norm": 4.1875, | |
| "learning_rate": 0.00011324041811846691, | |
| "loss": 0.8841, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.5744125326370757, | |
| "grad_norm": 2.078125, | |
| "learning_rate": 0.00011498257839721256, | |
| "loss": 0.891, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.5831157528285466, | |
| "grad_norm": 2.4375, | |
| "learning_rate": 0.00011672473867595819, | |
| "loss": 0.8778, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.5918189730200174, | |
| "grad_norm": 1.5078125, | |
| "learning_rate": 0.00011846689895470385, | |
| "loss": 0.8664, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.6005221932114883, | |
| "grad_norm": 1.6640625, | |
| "learning_rate": 0.00012020905923344947, | |
| "loss": 0.8838, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.6092254134029591, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 0.00012195121951219512, | |
| "loss": 0.8502, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.61792863359443, | |
| "grad_norm": 1.703125, | |
| "learning_rate": 0.00012369337979094077, | |
| "loss": 0.8579, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.6266318537859008, | |
| "grad_norm": 0.890625, | |
| "learning_rate": 0.0001254355400696864, | |
| "loss": 0.8601, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.6353350739773717, | |
| "grad_norm": 1.2578125, | |
| "learning_rate": 0.00012717770034843207, | |
| "loss": 0.8575, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.6440382941688425, | |
| "grad_norm": 1.9375, | |
| "learning_rate": 0.00012891986062717772, | |
| "loss": 0.8595, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.6527415143603134, | |
| "grad_norm": 1.359375, | |
| "learning_rate": 0.00013066202090592334, | |
| "loss": 0.8814, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.6614447345517842, | |
| "grad_norm": 0.93359375, | |
| "learning_rate": 0.000132404181184669, | |
| "loss": 0.8418, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.6701479547432551, | |
| "grad_norm": 0.94921875, | |
| "learning_rate": 0.00013414634146341464, | |
| "loss": 0.846, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.6788511749347258, | |
| "grad_norm": 1.0, | |
| "learning_rate": 0.00013588850174216027, | |
| "loss": 0.8663, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.6875543951261966, | |
| "grad_norm": 2.625, | |
| "learning_rate": 0.00013763066202090594, | |
| "loss": 0.8521, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.6962576153176675, | |
| "grad_norm": 3.46875, | |
| "learning_rate": 0.0001393728222996516, | |
| "loss": 0.8726, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.7049608355091384, | |
| "grad_norm": 3.0625, | |
| "learning_rate": 0.00014111498257839722, | |
| "loss": 0.8606, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.7136640557006092, | |
| "grad_norm": 1.25, | |
| "learning_rate": 0.00014285714285714287, | |
| "loss": 0.8517, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.72236727589208, | |
| "grad_norm": 2.515625, | |
| "learning_rate": 0.00014459930313588852, | |
| "loss": 0.8638, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.7310704960835509, | |
| "grad_norm": 1.4296875, | |
| "learning_rate": 0.00014634146341463414, | |
| "loss": 0.8448, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.7397737162750218, | |
| "grad_norm": 1.5390625, | |
| "learning_rate": 0.0001480836236933798, | |
| "loss": 0.8351, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.7484769364664926, | |
| "grad_norm": 3.203125, | |
| "learning_rate": 0.00014982578397212544, | |
| "loss": 0.8616, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.7571801566579635, | |
| "grad_norm": 2.109375, | |
| "learning_rate": 0.0001515679442508711, | |
| "loss": 0.8569, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.7658833768494343, | |
| "grad_norm": 3.265625, | |
| "learning_rate": 0.00015331010452961674, | |
| "loss": 0.8531, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.7745865970409052, | |
| "grad_norm": 1.2734375, | |
| "learning_rate": 0.00015505226480836236, | |
| "loss": 0.8483, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.783289817232376, | |
| "grad_norm": 1.8046875, | |
| "learning_rate": 0.00015679442508710801, | |
| "loss": 0.8444, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.7919930374238469, | |
| "grad_norm": 1.984375, | |
| "learning_rate": 0.00015853658536585366, | |
| "loss": 0.8513, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.8006962576153177, | |
| "grad_norm": 1.8125, | |
| "learning_rate": 0.00016027874564459931, | |
| "loss": 0.8326, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.8093994778067886, | |
| "grad_norm": 1.671875, | |
| "learning_rate": 0.00016202090592334496, | |
| "loss": 0.8554, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.8181026979982594, | |
| "grad_norm": 3.0, | |
| "learning_rate": 0.00016376306620209061, | |
| "loss": 0.8334, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.8268059181897301, | |
| "grad_norm": 1.2109375, | |
| "learning_rate": 0.00016550522648083624, | |
| "loss": 0.8547, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.835509138381201, | |
| "grad_norm": 1.6328125, | |
| "learning_rate": 0.0001672473867595819, | |
| "loss": 0.8496, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.8442123585726719, | |
| "grad_norm": 42.75, | |
| "learning_rate": 0.00016898954703832754, | |
| "loss": 0.8443, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.8529155787641427, | |
| "grad_norm": 1.5, | |
| "learning_rate": 0.0001707317073170732, | |
| "loss": 0.8419, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.8616187989556136, | |
| "grad_norm": 1.4375, | |
| "learning_rate": 0.00017247386759581884, | |
| "loss": 0.8447, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.8703220191470844, | |
| "grad_norm": 1.65625, | |
| "learning_rate": 0.00017421602787456446, | |
| "loss": 0.836, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.8790252393385553, | |
| "grad_norm": 2.40625, | |
| "learning_rate": 0.0001759581881533101, | |
| "loss": 0.8499, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.8877284595300261, | |
| "grad_norm": 1.7421875, | |
| "learning_rate": 0.00017770034843205576, | |
| "loss": 0.8429, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.896431679721497, | |
| "grad_norm": 1.875, | |
| "learning_rate": 0.00017944250871080138, | |
| "loss": 0.8562, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.9051348999129678, | |
| "grad_norm": 2.578125, | |
| "learning_rate": 0.00018118466898954703, | |
| "loss": 0.8553, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.9138381201044387, | |
| "grad_norm": 8.25, | |
| "learning_rate": 0.0001829268292682927, | |
| "loss": 0.8297, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.9225413402959095, | |
| "grad_norm": 0.97265625, | |
| "learning_rate": 0.00018466898954703833, | |
| "loss": 0.8288, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.9312445604873804, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 0.00018641114982578398, | |
| "loss": 0.8325, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.9399477806788512, | |
| "grad_norm": 1.171875, | |
| "learning_rate": 0.00018815331010452963, | |
| "loss": 0.8317, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.9486510008703221, | |
| "grad_norm": 2.3125, | |
| "learning_rate": 0.00018989547038327526, | |
| "loss": 0.8422, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.9573542210617929, | |
| "grad_norm": 1.53125, | |
| "learning_rate": 0.0001916376306620209, | |
| "loss": 0.8528, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.9660574412532638, | |
| "grad_norm": 1.046875, | |
| "learning_rate": 0.00019337979094076658, | |
| "loss": 0.8385, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.9747606614447345, | |
| "grad_norm": 2.28125, | |
| "learning_rate": 0.0001951219512195122, | |
| "loss": 0.8434, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.9834638816362054, | |
| "grad_norm": 2.296875, | |
| "learning_rate": 0.00019686411149825786, | |
| "loss": 0.8416, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.9921671018276762, | |
| "grad_norm": 6.90625, | |
| "learning_rate": 0.00019860627177700348, | |
| "loss": 0.8445, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.999129677980853, | |
| "eval_loss": 2.1106536388397217, | |
| "eval_runtime": 1.1027, | |
| "eval_samples_per_second": 5.441, | |
| "eval_steps_per_second": 0.907, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.000870322019147, | |
| "grad_norm": 1.0546875, | |
| "learning_rate": 0.00019999998150897728, | |
| "loss": 0.8531, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.009573542210618, | |
| "grad_norm": 2.625, | |
| "learning_rate": 0.00019999933432389942, | |
| "loss": 0.7707, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.0182767624020888, | |
| "grad_norm": 3.84375, | |
| "learning_rate": 0.00019999776259452297, | |
| "loss": 0.7908, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.0269799825935597, | |
| "grad_norm": 3.234375, | |
| "learning_rate": 0.00019999526633537938, | |
| "loss": 0.7832, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.0356832027850305, | |
| "grad_norm": 2.328125, | |
| "learning_rate": 0.00019999184556954776, | |
| "loss": 0.7502, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.0443864229765012, | |
| "grad_norm": 4.84375, | |
| "learning_rate": 0.00019998750032865483, | |
| "loss": 0.7704, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.0530896431679722, | |
| "grad_norm": 1.4765625, | |
| "learning_rate": 0.00019998223065287456, | |
| "loss": 0.7887, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.061792863359443, | |
| "grad_norm": 1.8046875, | |
| "learning_rate": 0.00019997603659092773, | |
| "loss": 0.7848, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.0704960835509139, | |
| "grad_norm": 1.515625, | |
| "learning_rate": 0.00019996891820008164, | |
| "loss": 0.7635, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.0791993037423846, | |
| "grad_norm": 1.109375, | |
| "learning_rate": 0.00019996087554614934, | |
| "loss": 0.7591, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.0879025239338556, | |
| "grad_norm": 1.2734375, | |
| "learning_rate": 0.00019995190870348922, | |
| "loss": 0.7569, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.0966057441253263, | |
| "grad_norm": 1.3125, | |
| "learning_rate": 0.0001999420177550043, | |
| "loss": 0.7677, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.1053089643167973, | |
| "grad_norm": 1.109375, | |
| "learning_rate": 0.00019993120279214135, | |
| "loss": 0.7648, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.114012184508268, | |
| "grad_norm": 0.984375, | |
| "learning_rate": 0.00019991946391489018, | |
| "loss": 0.7819, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.122715404699739, | |
| "grad_norm": 9.4375, | |
| "learning_rate": 0.00019990680123178263, | |
| "loss": 0.7606, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.1314186248912097, | |
| "grad_norm": 1.0234375, | |
| "learning_rate": 0.00019989321485989163, | |
| "loss": 0.796, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.1401218450826807, | |
| "grad_norm": 1.9296875, | |
| "learning_rate": 0.00019987870492482997, | |
| "loss": 0.7866, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.1488250652741514, | |
| "grad_norm": 1.46875, | |
| "learning_rate": 0.00019986327156074939, | |
| "loss": 0.7824, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.1575282854656224, | |
| "grad_norm": 3.234375, | |
| "learning_rate": 0.00019984691491033906, | |
| "loss": 0.7748, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.166231505657093, | |
| "grad_norm": 2.09375, | |
| "learning_rate": 0.00019982963512482453, | |
| "loss": 0.794, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.174934725848564, | |
| "grad_norm": 7.3125, | |
| "learning_rate": 0.00019981143236396612, | |
| "loss": 0.7733, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.1836379460400348, | |
| "grad_norm": 1.515625, | |
| "learning_rate": 0.00019979230679605749, | |
| "loss": 0.7919, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.1923411662315058, | |
| "grad_norm": 1.2265625, | |
| "learning_rate": 0.0001997722585979242, | |
| "loss": 0.7668, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.2010443864229765, | |
| "grad_norm": 0.578125, | |
| "learning_rate": 0.000199751287954922, | |
| "loss": 0.7746, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.2097476066144472, | |
| "grad_norm": 1.0859375, | |
| "learning_rate": 0.000199729395060935, | |
| "loss": 0.778, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.2184508268059182, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 0.00019970658011837404, | |
| "loss": 0.7742, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.227154046997389, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 0.00019968284333817486, | |
| "loss": 0.7856, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.23585726718886, | |
| "grad_norm": 0.953125, | |
| "learning_rate": 0.00019965818493979586, | |
| "loss": 0.78, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.2445604873803306, | |
| "grad_norm": 0.63671875, | |
| "learning_rate": 0.00019963260515121648, | |
| "loss": 0.804, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.2532637075718016, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 0.0001996061042089347, | |
| "loss": 0.7713, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.2619669277632724, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 0.00019957868235796514, | |
| "loss": 0.7725, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.2706701479547433, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 0.0001995503398518366, | |
| "loss": 0.7738, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.279373368146214, | |
| "grad_norm": 1.234375, | |
| "learning_rate": 0.00019952107695258992, | |
| "loss": 0.7935, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.288076588337685, | |
| "grad_norm": 0.93359375, | |
| "learning_rate": 0.0001994908939307753, | |
| "loss": 0.7573, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.2967798085291558, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 0.00019945979106545002, | |
| "loss": 0.8069, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.3054830287206267, | |
| "grad_norm": 1.6796875, | |
| "learning_rate": 0.0001994277686441758, | |
| "loss": 0.7752, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.3141862489120975, | |
| "grad_norm": 1.8671875, | |
| "learning_rate": 0.00019939482696301606, | |
| "loss": 0.7989, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.3228894691035684, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 0.00019936096632653324, | |
| "loss": 0.7946, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.3315926892950392, | |
| "grad_norm": 1.7421875, | |
| "learning_rate": 0.000199326187047786, | |
| "loss": 0.7781, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 1.34029590948651, | |
| "grad_norm": 1.859375, | |
| "learning_rate": 0.00019929048944832638, | |
| "loss": 0.7819, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.3489991296779809, | |
| "grad_norm": 1.2890625, | |
| "learning_rate": 0.00019925387385819664, | |
| "loss": 0.7702, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.3577023498694518, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 0.00019921634061592644, | |
| "loss": 0.7759, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.3664055700609226, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 0.0001991778900685295, | |
| "loss": 0.7683, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.3751087902523933, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 0.00019913852257150052, | |
| "loss": 0.7831, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.3838120104438643, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 0.0001990982384888119, | |
| "loss": 0.7823, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.392515230635335, | |
| "grad_norm": 1.046875, | |
| "learning_rate": 0.0001990570381929103, | |
| "loss": 0.7698, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.401218450826806, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 0.00019901492206471325, | |
| "loss": 0.7663, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.4099216710182767, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 0.00019897189049360557, | |
| "loss": 0.7966, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.4186248912097477, | |
| "grad_norm": 1.0625, | |
| "learning_rate": 0.00019892794387743593, | |
| "loss": 0.7792, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 1.4273281114012184, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 0.00019888308262251285, | |
| "loss": 0.7761, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.4360313315926894, | |
| "grad_norm": 1.5390625, | |
| "learning_rate": 0.00019883730714360137, | |
| "loss": 0.772, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.44473455178416, | |
| "grad_norm": 3.578125, | |
| "learning_rate": 0.00019879061786391881, | |
| "loss": 0.7705, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.453437771975631, | |
| "grad_norm": 2.03125, | |
| "learning_rate": 0.0001987430152151312, | |
| "loss": 0.7637, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.4621409921671018, | |
| "grad_norm": 0.9765625, | |
| "learning_rate": 0.00019869449963734893, | |
| "loss": 0.7647, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.4708442123585725, | |
| "grad_norm": 1.3125, | |
| "learning_rate": 0.0001986450715791231, | |
| "loss": 0.7772, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 1.4795474325500435, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 0.000198594731497441, | |
| "loss": 0.7538, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.4882506527415145, | |
| "grad_norm": 3.5, | |
| "learning_rate": 0.00019854347985772208, | |
| "loss": 0.7732, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 1.4969538729329852, | |
| "grad_norm": 0.96484375, | |
| "learning_rate": 0.00019849131713381364, | |
| "loss": 0.7777, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.505657093124456, | |
| "grad_norm": 1.5703125, | |
| "learning_rate": 0.00019843824380798633, | |
| "loss": 0.7742, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 1.514360313315927, | |
| "grad_norm": 1.3828125, | |
| "learning_rate": 0.00019838426037092988, | |
| "loss": 0.7596, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.5230635335073979, | |
| "grad_norm": 5.03125, | |
| "learning_rate": 0.00019832936732174834, | |
| "loss": 0.7668, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.5317667536988686, | |
| "grad_norm": 1.1640625, | |
| "learning_rate": 0.0001982735651679557, | |
| "loss": 0.7635, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.5404699738903394, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 0.000198216854425471, | |
| "loss": 0.7745, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 1.5491731940818103, | |
| "grad_norm": 0.9296875, | |
| "learning_rate": 0.0001981592356186137, | |
| "loss": 0.7905, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.5578764142732813, | |
| "grad_norm": 0.75, | |
| "learning_rate": 0.00019810070928009867, | |
| "loss": 0.7773, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 1.566579634464752, | |
| "grad_norm": 1.546875, | |
| "learning_rate": 0.0001980412759510315, | |
| "loss": 0.7611, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.5752828546562228, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 0.00019798093618090328, | |
| "loss": 0.7705, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 1.5839860748476937, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 0.00019791969052758562, | |
| "loss": 0.7895, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.5926892950391645, | |
| "grad_norm": 3.40625, | |
| "learning_rate": 0.0001978575395573255, | |
| "loss": 0.7738, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 1.6013925152306352, | |
| "grad_norm": 1.3515625, | |
| "learning_rate": 0.00019779448384474, | |
| "loss": 0.7661, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.6100957354221062, | |
| "grad_norm": 1.359375, | |
| "learning_rate": 0.000197730523972811, | |
| "loss": 0.7561, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.6187989556135771, | |
| "grad_norm": 1.0078125, | |
| "learning_rate": 0.00019766566053287975, | |
| "loss": 0.7742, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.6275021758050479, | |
| "grad_norm": 1.03125, | |
| "learning_rate": 0.00019759989412464153, | |
| "loss": 0.7742, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 1.6362053959965186, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 0.0001975332253561399, | |
| "loss": 0.769, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.6449086161879896, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 0.00019746565484376132, | |
| "loss": 0.7564, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 1.6536118363794605, | |
| "grad_norm": 0.87890625, | |
| "learning_rate": 0.00019739718321222928, | |
| "loss": 0.7574, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.6623150565709313, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 0.00019732781109459846, | |
| "loss": 0.7702, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 1.671018276762402, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 0.00019725753913224918, | |
| "loss": 0.7785, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.679721496953873, | |
| "grad_norm": 1.8828125, | |
| "learning_rate": 0.0001971863679748812, | |
| "loss": 0.7694, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 1.688424717145344, | |
| "grad_norm": 0.96875, | |
| "learning_rate": 0.00019711429828050769, | |
| "loss": 0.7802, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.6971279373368147, | |
| "grad_norm": 1.1171875, | |
| "learning_rate": 0.00019704133071544942, | |
| "loss": 0.7629, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.7058311575282854, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 0.00019696746595432828, | |
| "loss": 0.7739, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.7145343777197564, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 0.00019689270468006132, | |
| "loss": 0.7794, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 1.723237597911227, | |
| "grad_norm": 1.0078125, | |
| "learning_rate": 0.00019681704758385418, | |
| "loss": 0.7575, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.7319408181026978, | |
| "grad_norm": 1.0546875, | |
| "learning_rate": 0.0001967404953651949, | |
| "loss": 0.7673, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 1.7406440382941688, | |
| "grad_norm": 0.96484375, | |
| "learning_rate": 0.00019666304873184739, | |
| "loss": 0.7734, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.7493472584856398, | |
| "grad_norm": 1.6171875, | |
| "learning_rate": 0.0001965847083998448, | |
| "loss": 0.7785, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 1.7580504786771105, | |
| "grad_norm": 1.640625, | |
| "learning_rate": 0.00019650547509348306, | |
| "loss": 0.7652, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.7667536988685812, | |
| "grad_norm": 2.34375, | |
| "learning_rate": 0.0001964253495453141, | |
| "loss": 0.7631, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 1.7754569190600522, | |
| "grad_norm": 0.83203125, | |
| "learning_rate": 0.00019634433249613898, | |
| "loss": 0.7819, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.7841601392515232, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 0.0001962624246950012, | |
| "loss": 0.7774, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.792863359442994, | |
| "grad_norm": 0.609375, | |
| "learning_rate": 0.00019617962689917975, | |
| "loss": 0.7723, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.8015665796344646, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 0.00019609593987418198, | |
| "loss": 0.7645, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 1.8102697998259356, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 0.00019601136439373668, | |
| "loss": 0.7653, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.8189730200174066, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 0.0001959259012397868, | |
| "loss": 0.7756, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 1.8276762402088773, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 0.00019583955120248237, | |
| "loss": 0.7656, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.836379460400348, | |
| "grad_norm": 0.90625, | |
| "learning_rate": 0.00019575231508017307, | |
| "loss": 0.761, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 1.845082680591819, | |
| "grad_norm": 0.87890625, | |
| "learning_rate": 0.0001956641936794008, | |
| "loss": 0.7584, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.85378590078329, | |
| "grad_norm": 1.234375, | |
| "learning_rate": 0.00019557518781489238, | |
| "loss": 0.749, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 1.8624891209747607, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 0.00019548529830955196, | |
| "loss": 0.7635, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.8711923411662315, | |
| "grad_norm": 1.0703125, | |
| "learning_rate": 0.00019539452599445336, | |
| "loss": 0.7601, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.8798955613577024, | |
| "grad_norm": 0.90625, | |
| "learning_rate": 0.0001953028717088324, | |
| "loss": 0.7869, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.8885987815491732, | |
| "grad_norm": 2.046875, | |
| "learning_rate": 0.00019521033630007928, | |
| "loss": 0.766, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 1.897302001740644, | |
| "grad_norm": 1.4375, | |
| "learning_rate": 0.00019511692062373044, | |
| "loss": 0.7744, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.9060052219321149, | |
| "grad_norm": 1.03125, | |
| "learning_rate": 0.000195022625543461, | |
| "loss": 0.7749, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 1.9147084421235858, | |
| "grad_norm": 0.6328125, | |
| "learning_rate": 0.0001949274519310765, | |
| "loss": 0.7684, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.9234116623150566, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 0.00019483140066650507, | |
| "loss": 0.7596, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 1.9321148825065273, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 0.00019473447263778905, | |
| "loss": 0.768, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.9408181026979983, | |
| "grad_norm": 6.03125, | |
| "learning_rate": 0.00019463666874107704, | |
| "loss": 0.7563, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 1.9495213228894692, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 0.00019453798988061535, | |
| "loss": 0.7834, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.95822454308094, | |
| "grad_norm": 7.6875, | |
| "learning_rate": 0.00019443843696873985, | |
| "loss": 0.7471, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.9669277632724107, | |
| "grad_norm": 2.28125, | |
| "learning_rate": 0.00019433801092586742, | |
| "loss": 0.768, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.9756309834638817, | |
| "grad_norm": 1.1953125, | |
| "learning_rate": 0.00019423671268048754, | |
| "loss": 0.7806, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 1.9843342036553526, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 0.00019413454316915356, | |
| "loss": 0.7543, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.9930374238468234, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 0.00019403150333647417, | |
| "loss": 0.784, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 2.230104684829712, | |
| "eval_runtime": 0.7759, | |
| "eval_samples_per_second": 7.733, | |
| "eval_steps_per_second": 1.289, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 2.001740644038294, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 0.0001939275941351046, | |
| "loss": 0.7099, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.010443864229765, | |
| "grad_norm": 0.87890625, | |
| "learning_rate": 0.00019382281652573785, | |
| "loss": 0.6306, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 2.019147084421236, | |
| "grad_norm": 0.96484375, | |
| "learning_rate": 0.00019371717147709583, | |
| "loss": 0.6241, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.0278503046127065, | |
| "grad_norm": 1.1328125, | |
| "learning_rate": 0.0001936106599659202, | |
| "loss": 0.6167, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 2.0365535248041775, | |
| "grad_norm": 1.015625, | |
| "learning_rate": 0.00019350328297696373, | |
| "loss": 0.6173, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.0452567449956485, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 0.00019339504150298084, | |
| "loss": 0.6234, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.0539599651871194, | |
| "grad_norm": 1.265625, | |
| "learning_rate": 0.00019328593654471848, | |
| "loss": 0.6151, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.06266318537859, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 0.00019317596911090713, | |
| "loss": 0.6386, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 2.071366405570061, | |
| "grad_norm": 0.62890625, | |
| "learning_rate": 0.00019306514021825118, | |
| "loss": 0.6209, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.080069625761532, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 0.00019295345089141963, | |
| "loss": 0.625, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 2.0887728459530024, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 0.00019284090216303666, | |
| "loss": 0.6336, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.0974760661444734, | |
| "grad_norm": 0.8828125, | |
| "learning_rate": 0.00019272749507367212, | |
| "loss": 0.6266, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 2.1061792863359443, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 0.00019261323067183166, | |
| "loss": 0.6286, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.1148825065274153, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 0.0001924981100139474, | |
| "loss": 0.6458, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 2.123585726718886, | |
| "grad_norm": 2.03125, | |
| "learning_rate": 0.00019238213416436785, | |
| "loss": 0.6328, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.1322889469103568, | |
| "grad_norm": 1.1328125, | |
| "learning_rate": 0.00019226530419534833, | |
| "loss": 0.6398, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.1409921671018277, | |
| "grad_norm": 1.78125, | |
| "learning_rate": 0.00019214762118704076, | |
| "loss": 0.6361, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.1496953872932987, | |
| "grad_norm": 1.1875, | |
| "learning_rate": 0.000192029086227484, | |
| "loss": 0.6357, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 2.158398607484769, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 0.00019190970041259352, | |
| "loss": 0.6277, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.16710182767624, | |
| "grad_norm": 1.2734375, | |
| "learning_rate": 0.0001917894648461514, | |
| "loss": 0.6455, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 2.175805047867711, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 0.00019166838063979614, | |
| "loss": 0.6374, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.184508268059182, | |
| "grad_norm": 0.76953125, | |
| "learning_rate": 0.0001915464489130123, | |
| "loss": 0.6343, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 2.1932114882506526, | |
| "grad_norm": 1.0703125, | |
| "learning_rate": 0.00019142367079312021, | |
| "loss": 0.623, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.2019147084421236, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 0.00019130004741526558, | |
| "loss": 0.6359, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 2.2106179286335945, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 0.00019117557992240887, | |
| "loss": 0.6344, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.2193211488250655, | |
| "grad_norm": 0.75, | |
| "learning_rate": 0.00019105026946531482, | |
| "loss": 0.6511, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.228024369016536, | |
| "grad_norm": 0.83203125, | |
| "learning_rate": 0.0001909241172025419, | |
| "loss": 0.636, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.236727589208007, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 0.00019079712430043134, | |
| "loss": 0.6374, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 2.245430809399478, | |
| "grad_norm": 1.3671875, | |
| "learning_rate": 0.0001906692919330967, | |
| "loss": 0.6359, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.254134029590949, | |
| "grad_norm": 1.0859375, | |
| "learning_rate": 0.00019054062128241264, | |
| "loss": 0.6518, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 2.2628372497824194, | |
| "grad_norm": 1.140625, | |
| "learning_rate": 0.00019041111353800425, | |
| "loss": 0.6428, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.2715404699738904, | |
| "grad_norm": 1.0546875, | |
| "learning_rate": 0.00019028076989723597, | |
| "loss": 0.6562, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 2.2802436901653613, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 0.00019014959156520052, | |
| "loss": 0.6495, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.288946910356832, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 0.0001900175797547078, | |
| "loss": 0.6466, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 2.297650130548303, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 0.00018988473568627354, | |
| "loss": 0.6603, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.3063533507397738, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 0.00018975106058810823, | |
| "loss": 0.6352, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.3150565709312447, | |
| "grad_norm": 0.90234375, | |
| "learning_rate": 0.00018961655569610557, | |
| "loss": 0.6592, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.3237597911227152, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 0.00018948122225383114, | |
| "loss": 0.6515, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 2.332463011314186, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 0.00018934506151251093, | |
| "loss": 0.6534, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.341166231505657, | |
| "grad_norm": 1.0, | |
| "learning_rate": 0.00018920807473101964, | |
| "loss": 0.6558, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 2.349869451697128, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 0.00018907026317586923, | |
| "loss": 0.6547, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.3585726718885986, | |
| "grad_norm": 0.875, | |
| "learning_rate": 0.00018893162812119702, | |
| "loss": 0.6541, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 2.3672758920800696, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 0.00018879217084875408, | |
| "loss": 0.655, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.3759791122715406, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 0.0001886518926478932, | |
| "loss": 0.648, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 2.3846823324630115, | |
| "grad_norm": 0.60546875, | |
| "learning_rate": 0.00018851079481555714, | |
| "loss": 0.6474, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.393385552654482, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 0.00018836887865626654, | |
| "loss": 0.6543, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.402088772845953, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 0.00018822614548210797, | |
| "loss": 0.6529, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.410791993037424, | |
| "grad_norm": 1.0078125, | |
| "learning_rate": 0.00018808259661272153, | |
| "loss": 0.6612, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 2.4194952132288945, | |
| "grad_norm": 1.0078125, | |
| "learning_rate": 0.000187938233375289, | |
| "loss": 0.6519, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.4281984334203655, | |
| "grad_norm": 1.5625, | |
| "learning_rate": 0.00018779305710452132, | |
| "loss": 0.6558, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 2.4369016536118364, | |
| "grad_norm": 0.88671875, | |
| "learning_rate": 0.00018764706914264635, | |
| "loss": 0.6532, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.4456048738033074, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 0.00018750027083939654, | |
| "loss": 0.6443, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 2.454308093994778, | |
| "grad_norm": 0.8828125, | |
| "learning_rate": 0.00018735266355199618, | |
| "loss": 0.6544, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 2.463011314186249, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 0.00018720424864514913, | |
| "loss": 0.6663, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 2.47171453437772, | |
| "grad_norm": 3.359375, | |
| "learning_rate": 0.0001870550274910261, | |
| "loss": 0.6654, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 2.480417754569191, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 0.00018690500146925193, | |
| "loss": 0.6456, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.4891209747606613, | |
| "grad_norm": 0.88671875, | |
| "learning_rate": 0.00018675417196689292, | |
| "loss": 0.6495, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 2.4978241949521323, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 0.00018660254037844388, | |
| "loss": 0.6551, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 2.506527415143603, | |
| "grad_norm": 1.0703125, | |
| "learning_rate": 0.00018645010810581535, | |
| "loss": 0.6432, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 2.515230635335074, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 0.00018629687655832063, | |
| "loss": 0.6521, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 2.5239338555265447, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 0.00018614284715266264, | |
| "loss": 0.6626, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.5326370757180157, | |
| "grad_norm": 0.9296875, | |
| "learning_rate": 0.00018598802131292093, | |
| "loss": 0.6451, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 2.5413402959094866, | |
| "grad_norm": 0.94140625, | |
| "learning_rate": 0.00018583240047053863, | |
| "loss": 0.6627, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 2.550043516100957, | |
| "grad_norm": 0.8828125, | |
| "learning_rate": 0.00018567598606430882, | |
| "loss": 0.6756, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 2.558746736292428, | |
| "grad_norm": 1.0234375, | |
| "learning_rate": 0.00018551877954036162, | |
| "loss": 0.6734, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 2.567449956483899, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 0.0001853607823521507, | |
| "loss": 0.6495, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.57615317667537, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 0.00018520199596043976, | |
| "loss": 0.6459, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 2.584856396866841, | |
| "grad_norm": 1.0859375, | |
| "learning_rate": 0.0001850424218332891, | |
| "loss": 0.6665, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 2.5935596170583115, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 0.00018488206144604203, | |
| "loss": 0.6637, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 2.6022628372497825, | |
| "grad_norm": 2.015625, | |
| "learning_rate": 0.00018472091628131125, | |
| "loss": 0.6705, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 2.6109660574412534, | |
| "grad_norm": 0.76953125, | |
| "learning_rate": 0.00018455898782896511, | |
| "loss": 0.6601, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.619669277632724, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 0.00018439627758611385, | |
| "loss": 0.6591, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 2.628372497824195, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 0.00018423278705709573, | |
| "loss": 0.6574, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 2.637075718015666, | |
| "grad_norm": 0.9609375, | |
| "learning_rate": 0.00018406851775346322, | |
| "loss": 0.6665, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 2.645778938207137, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 0.0001839034711939689, | |
| "loss": 0.6591, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 2.6544821583986073, | |
| "grad_norm": 0.65625, | |
| "learning_rate": 0.00018373764890455146, | |
| "loss": 0.6505, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.6631853785900783, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 0.00018357105241832163, | |
| "loss": 0.6654, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 2.6718885987815493, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 0.000183403683275548, | |
| "loss": 0.6551, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 2.68059181897302, | |
| "grad_norm": 0.75, | |
| "learning_rate": 0.00018323554302364272, | |
| "loss": 0.6647, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 2.6892950391644908, | |
| "grad_norm": 0.9921875, | |
| "learning_rate": 0.0001830666332171473, | |
| "loss": 0.6658, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 2.6979982593559617, | |
| "grad_norm": 1.890625, | |
| "learning_rate": 0.00018289695541771802, | |
| "loss": 0.6584, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.7067014795474327, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 0.00018272651119411186, | |
| "loss": 0.6661, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 2.7154046997389036, | |
| "grad_norm": 1.9296875, | |
| "learning_rate": 0.0001825553021221716, | |
| "loss": 0.6695, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 2.724107919930374, | |
| "grad_norm": 1.453125, | |
| "learning_rate": 0.00018238332978481148, | |
| "loss": 0.6592, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 2.732811140121845, | |
| "grad_norm": 1.0859375, | |
| "learning_rate": 0.0001822105957720025, | |
| "loss": 0.6587, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 2.741514360313316, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 0.00018203710168075788, | |
| "loss": 0.6635, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.7502175805047866, | |
| "grad_norm": 0.91796875, | |
| "learning_rate": 0.00018186284911511787, | |
| "loss": 0.6567, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 2.7589208006962576, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 0.0001816878396861355, | |
| "loss": 0.6543, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 2.7676240208877285, | |
| "grad_norm": 1.2421875, | |
| "learning_rate": 0.0001815120750118611, | |
| "loss": 0.6662, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.7763272410791995, | |
| "grad_norm": 0.875, | |
| "learning_rate": 0.0001813355567173279, | |
| "loss": 0.6637, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 2.78503046127067, | |
| "grad_norm": 1.4296875, | |
| "learning_rate": 0.00018115828643453647, | |
| "loss": 0.6598, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.793733681462141, | |
| "grad_norm": 0.76953125, | |
| "learning_rate": 0.0001809802658024401, | |
| "loss": 0.6734, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 2.802436901653612, | |
| "grad_norm": 1.734375, | |
| "learning_rate": 0.0001808014964669293, | |
| "loss": 0.6547, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 2.8111401218450824, | |
| "grad_norm": 1.0859375, | |
| "learning_rate": 0.0001806219800808168, | |
| "loss": 0.6662, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 2.8198433420365534, | |
| "grad_norm": 1.015625, | |
| "learning_rate": 0.00018044171830382215, | |
| "loss": 0.658, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 2.8285465622280244, | |
| "grad_norm": 1.0078125, | |
| "learning_rate": 0.0001802607128025564, | |
| "loss": 0.6574, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.8372497824194953, | |
| "grad_norm": 1.09375, | |
| "learning_rate": 0.0001800789652505068, | |
| "loss": 0.6631, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 2.8459530026109663, | |
| "grad_norm": 0.62109375, | |
| "learning_rate": 0.00017989647732802113, | |
| "loss": 0.6606, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 2.854656222802437, | |
| "grad_norm": 0.875, | |
| "learning_rate": 0.00017971325072229226, | |
| "loss": 0.6759, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 2.8633594429939078, | |
| "grad_norm": 1.1015625, | |
| "learning_rate": 0.00017952928712734268, | |
| "loss": 0.6751, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 2.8720626631853787, | |
| "grad_norm": 0.9765625, | |
| "learning_rate": 0.00017934458824400858, | |
| "loss": 0.6604, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.8807658833768492, | |
| "grad_norm": 0.90625, | |
| "learning_rate": 0.00017915915577992433, | |
| "loss": 0.6528, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 2.88946910356832, | |
| "grad_norm": 1.109375, | |
| "learning_rate": 0.00017897299144950662, | |
| "loss": 0.653, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 2.898172323759791, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 0.00017878609697393868, | |
| "loss": 0.6757, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 2.906875543951262, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 0.00017859847408115414, | |
| "loss": 0.6608, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 2.9155787641427326, | |
| "grad_norm": 3.5625, | |
| "learning_rate": 0.00017841012450582134, | |
| "loss": 0.6624, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.9242819843342036, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 0.00017822104998932713, | |
| "loss": 0.671, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 2.9329852045256746, | |
| "grad_norm": 2.0, | |
| "learning_rate": 0.00017803125227976082, | |
| "loss": 0.6495, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 2.941688424717145, | |
| "grad_norm": 1.203125, | |
| "learning_rate": 0.00017784073313189795, | |
| "loss": 0.6729, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 2.950391644908616, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 0.00017764949430718426, | |
| "loss": 0.6656, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 2.959094865100087, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 0.00017745753757371905, | |
| "loss": 0.6674, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.967798085291558, | |
| "grad_norm": 1.375, | |
| "learning_rate": 0.00017726486470623926, | |
| "loss": 0.6585, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 2.976501305483029, | |
| "grad_norm": 1.5234375, | |
| "learning_rate": 0.00017707147748610274, | |
| "loss": 0.6659, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 2.9852045256744995, | |
| "grad_norm": 0.9296875, | |
| "learning_rate": 0.00017687737770127185, | |
| "loss": 0.67, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 2.9939077458659704, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 0.00017668256714629713, | |
| "loss": 0.6545, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 2.9991296779808527, | |
| "eval_loss": 2.432891607284546, | |
| "eval_runtime": 1.0987, | |
| "eval_samples_per_second": 5.461, | |
| "eval_steps_per_second": 0.91, | |
| "step": 1723 | |
| }, | |
| { | |
| "epoch": 3.0026109660574414, | |
| "grad_norm": 0.9609375, | |
| "learning_rate": 0.00017648704762230036, | |
| "loss": 0.6195, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 3.011314186248912, | |
| "grad_norm": 1.46875, | |
| "learning_rate": 0.00017629082093695823, | |
| "loss": 0.5228, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 3.020017406440383, | |
| "grad_norm": 1.71875, | |
| "learning_rate": 0.00017609388890448547, | |
| "loss": 0.5116, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 3.028720626631854, | |
| "grad_norm": 0.8515625, | |
| "learning_rate": 0.00017589625334561801, | |
| "loss": 0.5045, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 3.037423846823325, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 0.00017569791608759635, | |
| "loss": 0.51, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 3.0461270670147953, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 0.00017549887896414851, | |
| "loss": 0.5144, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.0548302872062663, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 0.0001752991438154731, | |
| "loss": 0.5033, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 3.063533507397737, | |
| "grad_norm": 1.0, | |
| "learning_rate": 0.00017509871248822236, | |
| "loss": 0.5268, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 3.072236727589208, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 0.00017489758683548502, | |
| "loss": 0.5163, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 3.0809399477806787, | |
| "grad_norm": 0.75, | |
| "learning_rate": 0.00017469576871676922, | |
| "loss": 0.5165, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 3.0896431679721497, | |
| "grad_norm": 0.875, | |
| "learning_rate": 0.00017449325999798528, | |
| "loss": 0.5237, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 3.0983463881636206, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 0.00017429006255142851, | |
| "loss": 0.5108, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 3.1070496083550916, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 0.0001740861782557618, | |
| "loss": 0.5086, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 3.115752828546562, | |
| "grad_norm": 0.97265625, | |
| "learning_rate": 0.0001738816089959983, | |
| "loss": 0.523, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 3.124456048738033, | |
| "grad_norm": 1.7109375, | |
| "learning_rate": 0.00017367635666348406, | |
| "loss": 0.5265, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 3.133159268929504, | |
| "grad_norm": 1.4453125, | |
| "learning_rate": 0.00017347042315588046, | |
| "loss": 0.5328, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.1418624891209745, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 0.00017326381037714668, | |
| "loss": 0.5294, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 3.1505657093124455, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 0.00017305652023752205, | |
| "loss": 0.5264, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 3.1592689295039165, | |
| "grad_norm": 0.88671875, | |
| "learning_rate": 0.00017284855465350856, | |
| "loss": 0.5164, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 3.1679721496953874, | |
| "grad_norm": 1.0078125, | |
| "learning_rate": 0.0001726399155478529, | |
| "loss": 0.5269, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 3.176675369886858, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 0.00017243060484952894, | |
| "loss": 0.5237, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 3.185378590078329, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 0.00017222062449371962, | |
| "loss": 0.5189, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 3.1940818102698, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 0.0001720099764217993, | |
| "loss": 0.5306, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 3.202785030461271, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 0.00017179866258131568, | |
| "loss": 0.5401, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 3.2114882506527413, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 0.00017158668492597186, | |
| "loss": 0.5254, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 3.2201914708442123, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 0.00017137404541560817, | |
| "loss": 0.5306, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.2288946910356833, | |
| "grad_norm": 0.91015625, | |
| "learning_rate": 0.00017116074601618417, | |
| "loss": 0.5299, | |
| "step": 1855 | |
| }, | |
| { | |
| "epoch": 3.2375979112271542, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 0.00017094678869976045, | |
| "loss": 0.53, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 3.2463011314186248, | |
| "grad_norm": 0.75, | |
| "learning_rate": 0.0001707321754444803, | |
| "loss": 0.5422, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 3.2550043516100957, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 0.00017051690823455162, | |
| "loss": 0.5357, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 3.2637075718015667, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 0.00017030098906022832, | |
| "loss": 0.5355, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.272410791993037, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 0.0001700844199177921, | |
| "loss": 0.5439, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 3.281114012184508, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 0.00016986720280953396, | |
| "loss": 0.5294, | |
| "step": 1885 | |
| }, | |
| { | |
| "epoch": 3.289817232375979, | |
| "grad_norm": 1.015625, | |
| "learning_rate": 0.0001696493397437357, | |
| "loss": 0.5485, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 3.29852045256745, | |
| "grad_norm": 1.6484375, | |
| "learning_rate": 0.0001694308327346512, | |
| "loss": 0.5429, | |
| "step": 1895 | |
| }, | |
| { | |
| "epoch": 3.307223672758921, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 0.0001692116838024881, | |
| "loss": 0.5518, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.3159268929503916, | |
| "grad_norm": 0.95703125, | |
| "learning_rate": 0.00016899189497338876, | |
| "loss": 0.5429, | |
| "step": 1905 | |
| }, | |
| { | |
| "epoch": 3.3246301131418625, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 0.00016877146827941187, | |
| "loss": 0.5392, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "grad_norm": 0.87109375, | |
| "learning_rate": 0.00016855040575851335, | |
| "loss": 0.5338, | |
| "step": 1915 | |
| }, | |
| { | |
| "epoch": 3.342036553524804, | |
| "grad_norm": 0.9609375, | |
| "learning_rate": 0.00016832870945452776, | |
| "loss": 0.545, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 3.350739773716275, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 0.00016810638141714934, | |
| "loss": 0.56, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.359442993907746, | |
| "grad_norm": 2.671875, | |
| "learning_rate": 0.0001678834237019129, | |
| "loss": 0.5483, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 3.368146214099217, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 0.00016765983837017503, | |
| "loss": 0.5448, | |
| "step": 1935 | |
| }, | |
| { | |
| "epoch": 3.3768494342906874, | |
| "grad_norm": 0.89453125, | |
| "learning_rate": 0.00016743562748909493, | |
| "loss": 0.5463, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 3.3855526544821584, | |
| "grad_norm": 1.1015625, | |
| "learning_rate": 0.00016721079313161534, | |
| "loss": 0.5518, | |
| "step": 1945 | |
| }, | |
| { | |
| "epoch": 3.3942558746736293, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 0.00016698533737644327, | |
| "loss": 0.551, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.4029590948651, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 0.000166759262308031, | |
| "loss": 0.5452, | |
| "step": 1955 | |
| }, | |
| { | |
| "epoch": 3.411662315056571, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 0.00016653257001655652, | |
| "loss": 0.5371, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 3.4203655352480418, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 0.00016630526259790455, | |
| "loss": 0.5615, | |
| "step": 1965 | |
| }, | |
| { | |
| "epoch": 3.4290687554395127, | |
| "grad_norm": 0.953125, | |
| "learning_rate": 0.00016607734215364674, | |
| "loss": 0.5466, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 3.4377719756309837, | |
| "grad_norm": 0.93359375, | |
| "learning_rate": 0.00016584881079102263, | |
| "loss": 0.554, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 3.446475195822454, | |
| "grad_norm": 0.91015625, | |
| "learning_rate": 0.00016561967062292, | |
| "loss": 0.5541, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 3.455178416013925, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 0.00016538992376785529, | |
| "loss": 0.5476, | |
| "step": 1985 | |
| }, | |
| { | |
| "epoch": 3.463881636205396, | |
| "grad_norm": 0.875, | |
| "learning_rate": 0.0001651595723499541, | |
| "loss": 0.5543, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 3.4725848563968666, | |
| "grad_norm": 1.0234375, | |
| "learning_rate": 0.0001649286184989315, | |
| "loss": 0.5547, | |
| "step": 1995 | |
| }, | |
| { | |
| "epoch": 3.4812880765883376, | |
| "grad_norm": 0.93359375, | |
| "learning_rate": 0.00016469706435007236, | |
| "loss": 0.5467, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.4899912967798086, | |
| "grad_norm": 1.5546875, | |
| "learning_rate": 0.0001644649120442116, | |
| "loss": 0.539, | |
| "step": 2005 | |
| }, | |
| { | |
| "epoch": 3.4986945169712795, | |
| "grad_norm": 0.9609375, | |
| "learning_rate": 0.00016423216372771443, | |
| "loss": 0.5448, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 3.5073977371627505, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 0.0001639988215524565, | |
| "loss": 0.5639, | |
| "step": 2015 | |
| }, | |
| { | |
| "epoch": 3.516100957354221, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 0.0001637648876758039, | |
| "loss": 0.5511, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 3.524804177545692, | |
| "grad_norm": 0.94921875, | |
| "learning_rate": 0.00016353036426059334, | |
| "loss": 0.5438, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 3.5335073977371625, | |
| "grad_norm": 0.88671875, | |
| "learning_rate": 0.0001632952534751122, | |
| "loss": 0.548, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 3.5422106179286335, | |
| "grad_norm": 0.8515625, | |
| "learning_rate": 0.00016305955749307816, | |
| "loss": 0.5532, | |
| "step": 2035 | |
| }, | |
| { | |
| "epoch": 3.5509138381201044, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 0.00016282327849361967, | |
| "loss": 0.5432, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 3.5596170583115754, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 0.00016258641866125518, | |
| "loss": 0.551, | |
| "step": 2045 | |
| }, | |
| { | |
| "epoch": 3.5683202785030463, | |
| "grad_norm": 0.75, | |
| "learning_rate": 0.00016234898018587337, | |
| "loss": 0.5454, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.577023498694517, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 0.00016211096526271273, | |
| "loss": 0.5555, | |
| "step": 2055 | |
| }, | |
| { | |
| "epoch": 3.585726718885988, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 0.00016187237609234132, | |
| "loss": 0.5503, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 3.594429939077459, | |
| "grad_norm": 0.9765625, | |
| "learning_rate": 0.00016163321488063637, | |
| "loss": 0.5432, | |
| "step": 2065 | |
| }, | |
| { | |
| "epoch": 3.6031331592689293, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 0.000161393483838764, | |
| "loss": 0.5531, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 3.6118363794604003, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 0.0001611531851831586, | |
| "loss": 0.5479, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.620539599651871, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 0.0001609123211355025, | |
| "loss": 0.553, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 3.629242819843342, | |
| "grad_norm": 1.8359375, | |
| "learning_rate": 0.00016067089392270533, | |
| "loss": 0.5554, | |
| "step": 2085 | |
| }, | |
| { | |
| "epoch": 3.637946040034813, | |
| "grad_norm": 0.75, | |
| "learning_rate": 0.00016042890577688349, | |
| "loss": 0.5501, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 3.6466492602262837, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 0.0001601863589353395, | |
| "loss": 0.5488, | |
| "step": 2095 | |
| }, | |
| { | |
| "epoch": 3.6553524804177546, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 0.00015994325564054122, | |
| "loss": 0.5618, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.664055700609225, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 0.00015969959814010132, | |
| "loss": 0.5526, | |
| "step": 2105 | |
| }, | |
| { | |
| "epoch": 3.672758920800696, | |
| "grad_norm": 0.890625, | |
| "learning_rate": 0.00015945538868675628, | |
| "loss": 0.5492, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 3.681462140992167, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 0.0001592106295383458, | |
| "loss": 0.5558, | |
| "step": 2115 | |
| }, | |
| { | |
| "epoch": 3.690165361183638, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 0.00015896532295779157, | |
| "loss": 0.5576, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 3.698868581375109, | |
| "grad_norm": 0.88671875, | |
| "learning_rate": 0.00015871947121307676, | |
| "loss": 0.5514, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.7075718015665795, | |
| "grad_norm": 0.875, | |
| "learning_rate": 0.0001584730765772248, | |
| "loss": 0.5615, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 3.7162750217580505, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 0.00015822614132827837, | |
| "loss": 0.5489, | |
| "step": 2135 | |
| }, | |
| { | |
| "epoch": 3.7249782419495214, | |
| "grad_norm": 0.91015625, | |
| "learning_rate": 0.00015797866774927848, | |
| "loss": 0.5507, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 3.733681462140992, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 0.0001577306581282432, | |
| "loss": 0.5574, | |
| "step": 2145 | |
| }, | |
| { | |
| "epoch": 3.742384682332463, | |
| "grad_norm": 0.91015625, | |
| "learning_rate": 0.00015748211475814658, | |
| "loss": 0.5579, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.751087902523934, | |
| "grad_norm": 0.89453125, | |
| "learning_rate": 0.00015723303993689754, | |
| "loss": 0.5736, | |
| "step": 2155 | |
| }, | |
| { | |
| "epoch": 3.759791122715405, | |
| "grad_norm": 0.9296875, | |
| "learning_rate": 0.0001569834359673184, | |
| "loss": 0.553, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 3.768494342906876, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 0.00015673330515712382, | |
| "loss": 0.5617, | |
| "step": 2165 | |
| }, | |
| { | |
| "epoch": 3.7771975630983463, | |
| "grad_norm": 1.0625, | |
| "learning_rate": 0.00015648264981889934, | |
| "loss": 0.5583, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 3.7859007832898173, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 0.00015623147227008006, | |
| "loss": 0.5584, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.7946040034812882, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 0.00015597977483292907, | |
| "loss": 0.5559, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 3.8033072236727588, | |
| "grad_norm": 1.046875, | |
| "learning_rate": 0.00015572755983451626, | |
| "loss": 0.5543, | |
| "step": 2185 | |
| }, | |
| { | |
| "epoch": 3.8120104438642297, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 0.00015547482960669645, | |
| "loss": 0.5554, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 3.8207136640557007, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 0.00015522158648608817, | |
| "loss": 0.5665, | |
| "step": 2195 | |
| }, | |
| { | |
| "epoch": 3.8294168842471716, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 0.00015496783281405177, | |
| "loss": 0.5614, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.838120104438642, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 0.00015471357093666804, | |
| "loss": 0.5596, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 3.846823324630113, | |
| "grad_norm": 0.90625, | |
| "learning_rate": 0.0001544588032047163, | |
| "loss": 0.553, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 3.855526544821584, | |
| "grad_norm": 0.81640625, | |
| "learning_rate": 0.0001542035319736528, | |
| "loss": 0.549, | |
| "step": 2215 | |
| }, | |
| { | |
| "epoch": 3.8642297650130546, | |
| "grad_norm": 0.75, | |
| "learning_rate": 0.0001539477596035888, | |
| "loss": 0.5562, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 3.8729329852045256, | |
| "grad_norm": 1.046875, | |
| "learning_rate": 0.00015369148845926893, | |
| "loss": 0.5658, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.8816362053959965, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 0.00015343472091004925, | |
| "loss": 0.5625, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 3.8903394255874675, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 0.00015317745932987524, | |
| "loss": 0.5613, | |
| "step": 2235 | |
| }, | |
| { | |
| "epoch": 3.8990426457789384, | |
| "grad_norm": 1.078125, | |
| "learning_rate": 0.00015291970609726007, | |
| "loss": 0.567, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 3.907745865970409, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 0.0001526614635952624, | |
| "loss": 0.568, | |
| "step": 2245 | |
| }, | |
| { | |
| "epoch": 3.91644908616188, | |
| "grad_norm": 0.90625, | |
| "learning_rate": 0.0001524027342114644, | |
| "loss": 0.5671, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.925152306353351, | |
| "grad_norm": 0.890625, | |
| "learning_rate": 0.0001521435203379498, | |
| "loss": 0.5538, | |
| "step": 2255 | |
| }, | |
| { | |
| "epoch": 3.9338555265448214, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 0.00015188382437128167, | |
| "loss": 0.5624, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 3.9425587467362924, | |
| "grad_norm": 0.99609375, | |
| "learning_rate": 0.00015162364871248023, | |
| "loss": 0.5491, | |
| "step": 2265 | |
| }, | |
| { | |
| "epoch": 3.9512619669277633, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 0.0001513629957670007, | |
| "loss": 0.5575, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 3.9599651871192343, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 0.00015110186794471103, | |
| "loss": 0.5639, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.968668407310705, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 0.00015084026765986979, | |
| "loss": 0.564, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 3.9773716275021758, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 0.00015057819733110348, | |
| "loss": 0.569, | |
| "step": 2285 | |
| }, | |
| { | |
| "epoch": 3.9860748476936467, | |
| "grad_norm": 0.8671875, | |
| "learning_rate": 0.00015031565938138458, | |
| "loss": 0.5676, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 3.9947780678851172, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 0.0001500526562380089, | |
| "loss": 0.5693, | |
| "step": 2295 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 2.70300030708313, | |
| "eval_runtime": 0.778, | |
| "eval_samples_per_second": 7.712, | |
| "eval_steps_per_second": 1.285, | |
| "step": 2298 | |
| }, | |
| { | |
| "epoch": 4.003481288076588, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 0.00014978919033257316, | |
| "loss": 0.5013, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.012184508268059, | |
| "grad_norm": 0.9453125, | |
| "learning_rate": 0.00014952526410095258, | |
| "loss": 0.412, | |
| "step": 2305 | |
| }, | |
| { | |
| "epoch": 4.02088772845953, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 0.00014926087998327837, | |
| "loss": 0.4225, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 4.029590948651001, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 0.00014899604042391506, | |
| "loss": 0.4255, | |
| "step": 2315 | |
| }, | |
| { | |
| "epoch": 4.038294168842472, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 0.000148730747871438, | |
| "loss": 0.4108, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 4.046997389033942, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 0.0001484650047786107, | |
| "loss": 0.4152, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 4.055700609225413, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 0.00014819881360236207, | |
| "loss": 0.4197, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 4.064403829416884, | |
| "grad_norm": 0.91015625, | |
| "learning_rate": 0.00014793217680376394, | |
| "loss": 0.4203, | |
| "step": 2335 | |
| }, | |
| { | |
| "epoch": 4.073107049608355, | |
| "grad_norm": 0.89453125, | |
| "learning_rate": 0.00014766509684800794, | |
| "loss": 0.4138, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 4.081810269799826, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 0.00014739757620438307, | |
| "loss": 0.4167, | |
| "step": 2345 | |
| }, | |
| { | |
| "epoch": 4.090513489991297, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 0.00014712961734625264, | |
| "loss": 0.4183, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.099216710182768, | |
| "grad_norm": 1.0, | |
| "learning_rate": 0.0001468612227510315, | |
| "loss": 0.4302, | |
| "step": 2355 | |
| }, | |
| { | |
| "epoch": 4.107919930374239, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 0.00014659239490016302, | |
| "loss": 0.4329, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 4.116623150565709, | |
| "grad_norm": 1.1328125, | |
| "learning_rate": 0.00014632313627909642, | |
| "loss": 0.4304, | |
| "step": 2365 | |
| }, | |
| { | |
| "epoch": 4.12532637075718, | |
| "grad_norm": 1.3125, | |
| "learning_rate": 0.00014605344937726345, | |
| "loss": 0.4194, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 4.134029590948651, | |
| "grad_norm": 0.8828125, | |
| "learning_rate": 0.00014578333668805558, | |
| "loss": 0.4195, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 4.142732811140122, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 0.0001455128007088009, | |
| "loss": 0.4354, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 4.151436031331593, | |
| "grad_norm": 0.96484375, | |
| "learning_rate": 0.00014524184394074102, | |
| "loss": 0.442, | |
| "step": 2385 | |
| }, | |
| { | |
| "epoch": 4.160139251523064, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 0.00014497046888900801, | |
| "loss": 0.433, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 4.168842471714535, | |
| "grad_norm": 1.2734375, | |
| "learning_rate": 0.00014469867806260115, | |
| "loss": 0.4325, | |
| "step": 2395 | |
| }, | |
| { | |
| "epoch": 4.177545691906005, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 0.00014442647397436365, | |
| "loss": 0.4255, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.186248912097476, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 0.0001441538591409598, | |
| "loss": 0.4419, | |
| "step": 2405 | |
| }, | |
| { | |
| "epoch": 4.194952132288947, | |
| "grad_norm": 1.0625, | |
| "learning_rate": 0.00014388083608285113, | |
| "loss": 0.4354, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 4.203655352480418, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 0.00014360740732427367, | |
| "loss": 0.4308, | |
| "step": 2415 | |
| }, | |
| { | |
| "epoch": 4.212358572671889, | |
| "grad_norm": 1.015625, | |
| "learning_rate": 0.00014333357539321416, | |
| "loss": 0.434, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 4.22106179286336, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 0.00014305934282138701, | |
| "loss": 0.4402, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 4.2297650130548305, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 0.00014278471214421073, | |
| "loss": 0.4298, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 4.2384682332463015, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 0.0001425096859007844, | |
| "loss": 0.4332, | |
| "step": 2435 | |
| }, | |
| { | |
| "epoch": 4.247171453437772, | |
| "grad_norm": 1.3515625, | |
| "learning_rate": 0.0001422342666338645, | |
| "loss": 0.4441, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 4.2558746736292425, | |
| "grad_norm": 0.953125, | |
| "learning_rate": 0.00014195845688984104, | |
| "loss": 0.435, | |
| "step": 2445 | |
| }, | |
| { | |
| "epoch": 4.2645778938207135, | |
| "grad_norm": 0.81640625, | |
| "learning_rate": 0.00014168225921871433, | |
| "loss": 0.4355, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 4.2732811140121845, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 0.00014140567617407105, | |
| "loss": 0.4422, | |
| "step": 2455 | |
| }, | |
| { | |
| "epoch": 4.281984334203655, | |
| "grad_norm": 0.8984375, | |
| "learning_rate": 0.00014112871031306119, | |
| "loss": 0.4347, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 4.290687554395126, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 0.00014085136419637369, | |
| "loss": 0.4353, | |
| "step": 2465 | |
| }, | |
| { | |
| "epoch": 4.299390774586597, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 0.00014057364038821347, | |
| "loss": 0.4425, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 4.308093994778067, | |
| "grad_norm": 0.87890625, | |
| "learning_rate": 0.00014029554145627714, | |
| "loss": 0.4419, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 4.316797214969538, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 0.00014001706997172973, | |
| "loss": 0.4403, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 4.325500435161009, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 0.00013973822850918055, | |
| "loss": 0.4427, | |
| "step": 2485 | |
| }, | |
| { | |
| "epoch": 4.33420365535248, | |
| "grad_norm": 0.83203125, | |
| "learning_rate": 0.0001394590196466596, | |
| "loss": 0.4351, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 4.342906875543951, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 0.00013917944596559376, | |
| "loss": 0.437, | |
| "step": 2495 | |
| }, | |
| { | |
| "epoch": 4.351610095735422, | |
| "grad_norm": 0.9375, | |
| "learning_rate": 0.0001388995100507827, | |
| "loss": 0.4383, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.360313315926893, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 0.0001386192144903752, | |
| "loss": 0.4403, | |
| "step": 2505 | |
| }, | |
| { | |
| "epoch": 4.369016536118364, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 0.00013833856187584514, | |
| "loss": 0.4474, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 4.377719756309834, | |
| "grad_norm": 1.046875, | |
| "learning_rate": 0.00013805755480196755, | |
| "loss": 0.4424, | |
| "step": 2515 | |
| }, | |
| { | |
| "epoch": 4.386422976501305, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 0.0001377761958667946, | |
| "loss": 0.4495, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 4.395126196692776, | |
| "grad_norm": 1.140625, | |
| "learning_rate": 0.00013749448767163156, | |
| "loss": 0.4468, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 4.403829416884247, | |
| "grad_norm": 1.1015625, | |
| "learning_rate": 0.0001372124328210129, | |
| "loss": 0.4472, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 4.412532637075718, | |
| "grad_norm": 0.90625, | |
| "learning_rate": 0.0001369300339226779, | |
| "loss": 0.4459, | |
| "step": 2535 | |
| }, | |
| { | |
| "epoch": 4.421235857267189, | |
| "grad_norm": 1.0546875, | |
| "learning_rate": 0.000136647293587547, | |
| "loss": 0.4462, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 4.42993907745866, | |
| "grad_norm": 0.97265625, | |
| "learning_rate": 0.00013636421442969718, | |
| "loss": 0.4439, | |
| "step": 2545 | |
| }, | |
| { | |
| "epoch": 4.438642297650131, | |
| "grad_norm": 0.921875, | |
| "learning_rate": 0.00013608079906633807, | |
| "loss": 0.4468, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 4.447345517841601, | |
| "grad_norm": 1.0234375, | |
| "learning_rate": 0.00013579705011778766, | |
| "loss": 0.4528, | |
| "step": 2555 | |
| }, | |
| { | |
| "epoch": 4.456048738033072, | |
| "grad_norm": 0.93359375, | |
| "learning_rate": 0.00013551297020744825, | |
| "loss": 0.4449, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 4.464751958224543, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 0.0001352285619617818, | |
| "loss": 0.4475, | |
| "step": 2565 | |
| }, | |
| { | |
| "epoch": 4.473455178416014, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 0.00013494382801028615, | |
| "loss": 0.4431, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 4.482158398607485, | |
| "grad_norm": 0.98046875, | |
| "learning_rate": 0.00013465877098547033, | |
| "loss": 0.4472, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 4.490861618798956, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 0.00013437339352283026, | |
| "loss": 0.4492, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 4.499564838990427, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 0.00013408769826082467, | |
| "loss": 0.46, | |
| "step": 2585 | |
| }, | |
| { | |
| "epoch": 4.508268059181898, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 0.00013380168784085027, | |
| "loss": 0.449, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 4.516971279373368, | |
| "grad_norm": 0.8515625, | |
| "learning_rate": 0.00013351536490721784, | |
| "loss": 0.4548, | |
| "step": 2595 | |
| }, | |
| { | |
| "epoch": 4.525674499564839, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 0.00013322873210712727, | |
| "loss": 0.4428, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.53437771975631, | |
| "grad_norm": 0.98828125, | |
| "learning_rate": 0.00013294179209064348, | |
| "loss": 0.4523, | |
| "step": 2605 | |
| }, | |
| { | |
| "epoch": 4.543080939947781, | |
| "grad_norm": 0.8984375, | |
| "learning_rate": 0.0001326545475106716, | |
| "loss": 0.4523, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 4.551784160139252, | |
| "grad_norm": 0.88671875, | |
| "learning_rate": 0.0001323670010229328, | |
| "loss": 0.4463, | |
| "step": 2615 | |
| }, | |
| { | |
| "epoch": 4.560487380330723, | |
| "grad_norm": 0.87109375, | |
| "learning_rate": 0.00013207915528593933, | |
| "loss": 0.4485, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 4.569190600522193, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 0.00013179101296097035, | |
| "loss": 0.4508, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 4.577893820713664, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 0.00013150257671204696, | |
| "loss": 0.446, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 4.586597040905135, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 0.00013121384920590786, | |
| "loss": 0.448, | |
| "step": 2635 | |
| }, | |
| { | |
| "epoch": 4.595300261096606, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 0.00013092483311198444, | |
| "loss": 0.4522, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 4.604003481288077, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 0.00013063553110237642, | |
| "loss": 0.4565, | |
| "step": 2645 | |
| }, | |
| { | |
| "epoch": 4.6127067014795475, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 0.00013034594585182677, | |
| "loss": 0.4575, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 4.6214099216710185, | |
| "grad_norm": 0.9140625, | |
| "learning_rate": 0.00013005608003769718, | |
| "loss": 0.4544, | |
| "step": 2655 | |
| }, | |
| { | |
| "epoch": 4.6301131418624895, | |
| "grad_norm": 1.015625, | |
| "learning_rate": 0.00012976593633994346, | |
| "loss": 0.457, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 4.63881636205396, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 0.00012947551744109043, | |
| "loss": 0.4478, | |
| "step": 2665 | |
| }, | |
| { | |
| "epoch": 4.6475195822454305, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 0.00012918482602620733, | |
| "loss": 0.4591, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 4.6562228024369015, | |
| "grad_norm": 0.98046875, | |
| "learning_rate": 0.00012889386478288299, | |
| "loss": 0.4549, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 4.664926022628372, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 0.00012860263640120085, | |
| "loss": 0.4468, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 4.673629242819843, | |
| "grad_norm": 0.92578125, | |
| "learning_rate": 0.00012831114357371426, | |
| "loss": 0.444, | |
| "step": 2685 | |
| }, | |
| { | |
| "epoch": 4.682332463011314, | |
| "grad_norm": 0.90625, | |
| "learning_rate": 0.0001280193889954215, | |
| "loss": 0.4649, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 4.691035683202785, | |
| "grad_norm": 1.125, | |
| "learning_rate": 0.0001277273753637408, | |
| "loss": 0.4608, | |
| "step": 2695 | |
| }, | |
| { | |
| "epoch": 4.699738903394256, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 0.00012743510537848555, | |
| "loss": 0.4522, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 4.708442123585726, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 0.0001271425817418392, | |
| "loss": 0.4637, | |
| "step": 2705 | |
| }, | |
| { | |
| "epoch": 4.717145343777197, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 0.00012684980715833039, | |
| "loss": 0.4589, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 4.725848563968668, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 0.0001265567843348078, | |
| "loss": 0.4552, | |
| "step": 2715 | |
| }, | |
| { | |
| "epoch": 4.734551784160139, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 0.00012626351598041532, | |
| "loss": 0.4555, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 4.74325500435161, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 0.00012597000480656684, | |
| "loss": 0.463, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 4.751958224543081, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 0.00012567625352692127, | |
| "loss": 0.462, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 4.760661444734552, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 0.00012538226485735735, | |
| "loss": 0.4553, | |
| "step": 2735 | |
| }, | |
| { | |
| "epoch": 4.769364664926023, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 0.00012508804151594867, | |
| "loss": 0.4525, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 4.778067885117493, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 0.0001247935862229385, | |
| "loss": 0.4609, | |
| "step": 2745 | |
| }, | |
| { | |
| "epoch": 4.786771105308964, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 0.00012449890170071454, | |
| "loss": 0.4491, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.795474325500435, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 0.00012420399067378392, | |
| "loss": 0.4502, | |
| "step": 2755 | |
| }, | |
| { | |
| "epoch": 4.804177545691906, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 0.00012390885586874783, | |
| "loss": 0.4527, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 4.812880765883377, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 0.0001236135000142765, | |
| "loss": 0.4531, | |
| "step": 2765 | |
| }, | |
| { | |
| "epoch": 4.821583986074848, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 0.00012331792584108374, | |
| "loss": 0.4511, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 4.830287206266319, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 0.00012302213608190202, | |
| "loss": 0.4504, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 4.838990426457789, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 0.0001227261334714568, | |
| "loss": 0.4538, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 4.84769364664926, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 0.00012242992074644162, | |
| "loss": 0.4585, | |
| "step": 2785 | |
| }, | |
| { | |
| "epoch": 4.856396866840731, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 0.0001221335006454925, | |
| "loss": 0.4518, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 4.865100087032202, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 0.00012183687590916291, | |
| "loss": 0.4534, | |
| "step": 2795 | |
| }, | |
| { | |
| "epoch": 4.873803307223673, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 0.00012154004927989815, | |
| "loss": 0.4543, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.882506527415144, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 0.00012124302350201016, | |
| "loss": 0.4549, | |
| "step": 2805 | |
| }, | |
| { | |
| "epoch": 4.891209747606615, | |
| "grad_norm": 1.078125, | |
| "learning_rate": 0.00012094580132165211, | |
| "loss": 0.4405, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 4.899912967798086, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 0.00012064838548679307, | |
| "loss": 0.4501, | |
| "step": 2815 | |
| }, | |
| { | |
| "epoch": 4.908616187989556, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 0.00012035077874719242, | |
| "loss": 0.4574, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 4.917319408181027, | |
| "grad_norm": 0.90234375, | |
| "learning_rate": 0.00012005298385437467, | |
| "loss": 0.4515, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 4.926022628372498, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 0.00011975500356160383, | |
| "loss": 0.4532, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 4.934725848563969, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 0.00011945684062385803, | |
| "loss": 0.4533, | |
| "step": 2835 | |
| }, | |
| { | |
| "epoch": 4.94342906875544, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 0.00011915849779780408, | |
| "loss": 0.4633, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 4.952132288946911, | |
| "grad_norm": 0.984375, | |
| "learning_rate": 0.00011885997784177196, | |
| "loss": 0.4568, | |
| "step": 2845 | |
| }, | |
| { | |
| "epoch": 4.960835509138382, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 0.00011856128351572921, | |
| "loss": 0.4543, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.969538729329852, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 0.00011826241758125565, | |
| "loss": 0.4576, | |
| "step": 2855 | |
| }, | |
| { | |
| "epoch": 4.978241949521323, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 0.00011796338280151756, | |
| "loss": 0.4595, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 4.986945169712794, | |
| "grad_norm": 0.75, | |
| "learning_rate": 0.0001176641819412424, | |
| "loss": 0.4549, | |
| "step": 2865 | |
| }, | |
| { | |
| "epoch": 4.9956483899042645, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 0.00011736481776669306, | |
| "loss": 0.4555, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 4.999129677980853, | |
| "eval_loss": 3.144505500793457, | |
| "eval_runtime": 1.1115, | |
| "eval_samples_per_second": 5.398, | |
| "eval_steps_per_second": 0.9, | |
| "step": 2872 | |
| }, | |
| { | |
| "epoch": 5.0043516100957355, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 0.00011706529304564235, | |
| "loss": 0.4042, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 5.013054830287206, | |
| "grad_norm": 0.890625, | |
| "learning_rate": 0.00011676561054734749, | |
| "loss": 0.3352, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 5.021758050478677, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 0.00011646577304252433, | |
| "loss": 0.3304, | |
| "step": 2885 | |
| }, | |
| { | |
| "epoch": 5.030461270670148, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 0.0001161657833033219, | |
| "loss": 0.3354, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 5.039164490861618, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 0.0001158656441032967, | |
| "loss": 0.3342, | |
| "step": 2895 | |
| }, | |
| { | |
| "epoch": 5.047867711053089, | |
| "grad_norm": 0.75, | |
| "learning_rate": 0.00011556535821738705, | |
| "loss": 0.3344, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 5.05657093124456, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 0.00011526492842188745, | |
| "loss": 0.3339, | |
| "step": 2905 | |
| }, | |
| { | |
| "epoch": 5.065274151436031, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 0.000114964357494423, | |
| "loss": 0.3343, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 5.073977371627502, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 0.00011466364821392348, | |
| "loss": 0.3391, | |
| "step": 2915 | |
| }, | |
| { | |
| "epoch": 5.082680591818973, | |
| "grad_norm": 0.88671875, | |
| "learning_rate": 0.00011436280336059799, | |
| "loss": 0.34, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 5.091383812010444, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 0.00011406182571590893, | |
| "loss": 0.3388, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 5.100087032201914, | |
| "grad_norm": 0.75, | |
| "learning_rate": 0.00011376071806254651, | |
| "loss": 0.3371, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 5.108790252393385, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 0.00011345948318440289, | |
| "loss": 0.3496, | |
| "step": 2935 | |
| }, | |
| { | |
| "epoch": 5.117493472584856, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 0.0001131581238665465, | |
| "loss": 0.3433, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 5.126196692776327, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 0.00011285664289519626, | |
| "loss": 0.3426, | |
| "step": 2945 | |
| }, | |
| { | |
| "epoch": 5.134899912967798, | |
| "grad_norm": 0.8515625, | |
| "learning_rate": 0.00011255504305769589, | |
| "loss": 0.3352, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 5.143603133159269, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 0.00011225332714248804, | |
| "loss": 0.3492, | |
| "step": 2955 | |
| }, | |
| { | |
| "epoch": 5.15230635335074, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 0.00011195149793908856, | |
| "loss": 0.338, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 5.161009573542211, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 0.00011164955823806079, | |
| "loss": 0.343, | |
| "step": 2965 | |
| }, | |
| { | |
| "epoch": 5.169712793733681, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 0.00011134751083098946, | |
| "loss": 0.3407, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 5.178416013925152, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 0.00011104535851045539, | |
| "loss": 0.3391, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 5.187119234116623, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 0.00011074310407000914, | |
| "loss": 0.3438, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 5.195822454308094, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 0.00011044075030414553, | |
| "loss": 0.3394, | |
| "step": 2985 | |
| }, | |
| { | |
| "epoch": 5.204525674499565, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 0.00011013830000827767, | |
| "loss": 0.3471, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 5.213228894691036, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 0.00010983575597871114, | |
| "loss": 0.3392, | |
| "step": 2995 | |
| }, | |
| { | |
| "epoch": 5.221932114882507, | |
| "grad_norm": 0.81640625, | |
| "learning_rate": 0.00010953312101261815, | |
| "loss": 0.3436, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.230635335073977, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 0.00010923039790801164, | |
| "loss": 0.3398, | |
| "step": 3005 | |
| }, | |
| { | |
| "epoch": 5.239338555265448, | |
| "grad_norm": 0.81640625, | |
| "learning_rate": 0.00010892758946371944, | |
| "loss": 0.3469, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 5.248041775456919, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 0.00010862469847935841, | |
| "loss": 0.3444, | |
| "step": 3015 | |
| }, | |
| { | |
| "epoch": 5.25674499564839, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 0.00010832172775530851, | |
| "loss": 0.3431, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 5.265448215839861, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 0.00010801868009268691, | |
| "loss": 0.3513, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 5.274151436031332, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 0.00010771555829332223, | |
| "loss": 0.3476, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 5.282854656222803, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 0.00010741236515972839, | |
| "loss": 0.3471, | |
| "step": 3035 | |
| }, | |
| { | |
| "epoch": 5.291557876414274, | |
| "grad_norm": 0.95703125, | |
| "learning_rate": 0.0001071091034950788, | |
| "loss": 0.3416, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 5.300261096605744, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 0.00010680577610318072, | |
| "loss": 0.3454, | |
| "step": 3045 | |
| }, | |
| { | |
| "epoch": 5.308964316797215, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 0.0001065023857884488, | |
| "loss": 0.3486, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 5.317667536988686, | |
| "grad_norm": 0.984375, | |
| "learning_rate": 0.00010619893535587964, | |
| "loss": 0.3386, | |
| "step": 3055 | |
| }, | |
| { | |
| "epoch": 5.326370757180157, | |
| "grad_norm": 0.9609375, | |
| "learning_rate": 0.00010589542761102553, | |
| "loss": 0.3418, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 5.335073977371628, | |
| "grad_norm": 0.87109375, | |
| "learning_rate": 0.00010559186535996873, | |
| "loss": 0.3522, | |
| "step": 3065 | |
| }, | |
| { | |
| "epoch": 5.3437771975630985, | |
| "grad_norm": 1.0234375, | |
| "learning_rate": 0.00010528825140929541, | |
| "loss": 0.3449, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 5.3524804177545695, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 0.00010498458856606972, | |
| "loss": 0.3473, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 5.36118363794604, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 0.00010468087963780789, | |
| "loss": 0.353, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 5.3698868581375105, | |
| "grad_norm": 0.9921875, | |
| "learning_rate": 0.00010437712743245209, | |
| "loss": 0.352, | |
| "step": 3085 | |
| }, | |
| { | |
| "epoch": 5.3785900783289815, | |
| "grad_norm": 0.8828125, | |
| "learning_rate": 0.00010407333475834487, | |
| "loss": 0.354, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 5.3872932985204525, | |
| "grad_norm": 0.89453125, | |
| "learning_rate": 0.00010376950442420259, | |
| "loss": 0.3436, | |
| "step": 3095 | |
| }, | |
| { | |
| "epoch": 5.395996518711923, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 0.00010346563923909014, | |
| "loss": 0.3511, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 5.404699738903394, | |
| "grad_norm": 0.90234375, | |
| "learning_rate": 0.00010316174201239437, | |
| "loss": 0.3472, | |
| "step": 3105 | |
| }, | |
| { | |
| "epoch": 5.413402959094865, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 0.00010285781555379852, | |
| "loss": 0.3449, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 5.422106179286336, | |
| "grad_norm": 1.015625, | |
| "learning_rate": 0.00010255386267325602, | |
| "loss": 0.3471, | |
| "step": 3115 | |
| }, | |
| { | |
| "epoch": 5.430809399477806, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 0.00010224988618096458, | |
| "loss": 0.3523, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 5.439512619669277, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 0.00010194588888734027, | |
| "loss": 0.3492, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 5.448215839860748, | |
| "grad_norm": 0.8828125, | |
| "learning_rate": 0.00010164187360299142, | |
| "loss": 0.3465, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 5.456919060052219, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 0.00010133784313869277, | |
| "loss": 0.3472, | |
| "step": 3135 | |
| }, | |
| { | |
| "epoch": 5.46562228024369, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 0.00010103380030535929, | |
| "loss": 0.3558, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 5.474325500435161, | |
| "grad_norm": 0.8828125, | |
| "learning_rate": 0.0001007297479140204, | |
| "loss": 0.3539, | |
| "step": 3145 | |
| }, | |
| { | |
| "epoch": 5.483028720626632, | |
| "grad_norm": 0.91796875, | |
| "learning_rate": 0.00010042568877579388, | |
| "loss": 0.3486, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 5.491731940818102, | |
| "grad_norm": 1.078125, | |
| "learning_rate": 0.00010012162570185983, | |
| "loss": 0.3573, | |
| "step": 3155 | |
| }, | |
| { | |
| "epoch": 5.500435161009573, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 9.981756150343485e-05, | |
| "loss": 0.3473, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 5.509138381201044, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 9.951349899174577e-05, | |
| "loss": 0.3558, | |
| "step": 3165 | |
| }, | |
| { | |
| "epoch": 5.517841601392515, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 9.920944097800398e-05, | |
| "loss": 0.3542, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 5.526544821583986, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 9.890539027337924e-05, | |
| "loss": 0.3471, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 5.535248041775457, | |
| "grad_norm": 0.9375, | |
| "learning_rate": 9.860134968897366e-05, | |
| "loss": 0.3553, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 5.543951261966928, | |
| "grad_norm": 0.81640625, | |
| "learning_rate": 9.829732203579584e-05, | |
| "loss": 0.3558, | |
| "step": 3185 | |
| }, | |
| { | |
| "epoch": 5.552654482158399, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 9.799331012473493e-05, | |
| "loss": 0.3526, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 5.56135770234987, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 9.768931676653427e-05, | |
| "loss": 0.3499, | |
| "step": 3195 | |
| }, | |
| { | |
| "epoch": 5.57006092254134, | |
| "grad_norm": 0.81640625, | |
| "learning_rate": 9.738534477176596e-05, | |
| "loss": 0.3447, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 5.578764142732811, | |
| "grad_norm": 0.88671875, | |
| "learning_rate": 9.708139695080441e-05, | |
| "loss": 0.3568, | |
| "step": 3205 | |
| }, | |
| { | |
| "epoch": 5.587467362924282, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 9.677747611380058e-05, | |
| "loss": 0.3575, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 5.596170583115753, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 9.647358507065594e-05, | |
| "loss": 0.3536, | |
| "step": 3215 | |
| }, | |
| { | |
| "epoch": 5.604873803307224, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 9.616972663099647e-05, | |
| "loss": 0.3524, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 5.613577023498695, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 9.58659036041468e-05, | |
| "loss": 0.3541, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 5.622280243690165, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 9.556211879910414e-05, | |
| "loss": 0.3519, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 5.630983463881636, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 9.52583750245122e-05, | |
| "loss": 0.3514, | |
| "step": 3235 | |
| }, | |
| { | |
| "epoch": 5.639686684073107, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 9.495467508863542e-05, | |
| "loss": 0.3485, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 5.648389904264578, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 9.465102179933302e-05, | |
| "loss": 0.3547, | |
| "step": 3245 | |
| }, | |
| { | |
| "epoch": 5.657093124456049, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 9.434741796403282e-05, | |
| "loss": 0.3549, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 5.66579634464752, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 9.404386638970542e-05, | |
| "loss": 0.3502, | |
| "step": 3255 | |
| }, | |
| { | |
| "epoch": 5.674499564838991, | |
| "grad_norm": 0.87890625, | |
| "learning_rate": 9.37403698828383e-05, | |
| "loss": 0.354, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 5.683202785030462, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 9.343693124940977e-05, | |
| "loss": 0.3499, | |
| "step": 3265 | |
| }, | |
| { | |
| "epoch": 5.691906005221933, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 9.313355329486318e-05, | |
| "loss": 0.3535, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 5.700609225413403, | |
| "grad_norm": 0.89453125, | |
| "learning_rate": 9.283023882408065e-05, | |
| "loss": 0.3487, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 5.709312445604874, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 9.252699064135758e-05, | |
| "loss": 0.3458, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 5.718015665796345, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 9.22238115503764e-05, | |
| "loss": 0.3518, | |
| "step": 3285 | |
| }, | |
| { | |
| "epoch": 5.7267188859878155, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 9.192070435418079e-05, | |
| "loss": 0.3488, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 5.7354221061792865, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 9.161767185514964e-05, | |
| "loss": 0.3529, | |
| "step": 3295 | |
| }, | |
| { | |
| "epoch": 5.7441253263707575, | |
| "grad_norm": 0.83203125, | |
| "learning_rate": 9.131471685497134e-05, | |
| "loss": 0.3553, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 5.7528285465622275, | |
| "grad_norm": 0.83203125, | |
| "learning_rate": 9.101184215461774e-05, | |
| "loss": 0.3494, | |
| "step": 3305 | |
| }, | |
| { | |
| "epoch": 5.7615317667536985, | |
| "grad_norm": 0.8984375, | |
| "learning_rate": 9.070905055431822e-05, | |
| "loss": 0.357, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 5.7702349869451695, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 9.040634485353389e-05, | |
| "loss": 0.3592, | |
| "step": 3315 | |
| }, | |
| { | |
| "epoch": 5.77893820713664, | |
| "grad_norm": 0.81640625, | |
| "learning_rate": 9.010372785093167e-05, | |
| "loss": 0.3521, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 5.787641427328111, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 8.980120234435849e-05, | |
| "loss": 0.3605, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 5.796344647519582, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 8.949877113081521e-05, | |
| "loss": 0.35, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 5.805047867711053, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 8.919643700643103e-05, | |
| "loss": 0.3483, | |
| "step": 3335 | |
| }, | |
| { | |
| "epoch": 5.813751087902524, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 8.889420276643746e-05, | |
| "loss": 0.3505, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 5.822454308093995, | |
| "grad_norm": 0.8515625, | |
| "learning_rate": 8.859207120514255e-05, | |
| "loss": 0.3468, | |
| "step": 3345 | |
| }, | |
| { | |
| "epoch": 5.831157528285465, | |
| "grad_norm": 0.88671875, | |
| "learning_rate": 8.829004511590501e-05, | |
| "loss": 0.3539, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 5.839860748476936, | |
| "grad_norm": 0.953125, | |
| "learning_rate": 8.798812729110837e-05, | |
| "loss": 0.3481, | |
| "step": 3355 | |
| }, | |
| { | |
| "epoch": 5.848563968668407, | |
| "grad_norm": 0.87109375, | |
| "learning_rate": 8.768632052213531e-05, | |
| "loss": 0.3551, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 5.857267188859878, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 8.738462759934168e-05, | |
| "loss": 0.3509, | |
| "step": 3365 | |
| }, | |
| { | |
| "epoch": 5.865970409051349, | |
| "grad_norm": 0.91796875, | |
| "learning_rate": 8.708305131203072e-05, | |
| "loss": 0.3551, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 5.87467362924282, | |
| "grad_norm": 0.8671875, | |
| "learning_rate": 8.678159444842737e-05, | |
| "loss": 0.3469, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 5.883376849434291, | |
| "grad_norm": 0.91015625, | |
| "learning_rate": 8.648025979565245e-05, | |
| "loss": 0.3544, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 5.892080069625761, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 8.617905013969688e-05, | |
| "loss": 0.3476, | |
| "step": 3385 | |
| }, | |
| { | |
| "epoch": 5.900783289817232, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 8.587796826539585e-05, | |
| "loss": 0.3531, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 5.909486510008703, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 8.557701695640321e-05, | |
| "loss": 0.3401, | |
| "step": 3395 | |
| }, | |
| { | |
| "epoch": 5.918189730200174, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 8.527619899516567e-05, | |
| "loss": 0.35, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 5.926892950391645, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 8.497551716289703e-05, | |
| "loss": 0.3474, | |
| "step": 3405 | |
| }, | |
| { | |
| "epoch": 5.935596170583116, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 8.467497423955249e-05, | |
| "loss": 0.35, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 5.944299390774587, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 8.437457300380309e-05, | |
| "loss": 0.3564, | |
| "step": 3415 | |
| }, | |
| { | |
| "epoch": 5.953002610966058, | |
| "grad_norm": 0.8671875, | |
| "learning_rate": 8.407431623300983e-05, | |
| "loss": 0.3516, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 5.961705831157528, | |
| "grad_norm": 0.9140625, | |
| "learning_rate": 8.377420670319795e-05, | |
| "loss": 0.356, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 5.970409051348999, | |
| "grad_norm": 0.87109375, | |
| "learning_rate": 8.347424718903151e-05, | |
| "loss": 0.3538, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 5.97911227154047, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 8.317444046378757e-05, | |
| "loss": 0.3491, | |
| "step": 3435 | |
| }, | |
| { | |
| "epoch": 5.987815491731941, | |
| "grad_norm": 0.9296875, | |
| "learning_rate": 8.28747892993306e-05, | |
| "loss": 0.3559, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 5.996518711923412, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 8.257529646608672e-05, | |
| "loss": 0.3504, | |
| "step": 3445 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_loss": 3.7196741104125977, | |
| "eval_runtime": 0.7785, | |
| "eval_samples_per_second": 7.707, | |
| "eval_steps_per_second": 1.285, | |
| "step": 3447 | |
| }, | |
| { | |
| "epoch": 6.005221932114883, | |
| "grad_norm": 0.62109375, | |
| "learning_rate": 8.227596473301835e-05, | |
| "loss": 0.2993, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 6.013925152306354, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 8.19767968675983e-05, | |
| "loss": 0.2552, | |
| "step": 3455 | |
| }, | |
| { | |
| "epoch": 6.022628372497824, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 8.167779563578456e-05, | |
| "loss": 0.2635, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 6.031331592689295, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 8.13789638019942e-05, | |
| "loss": 0.2613, | |
| "step": 3465 | |
| }, | |
| { | |
| "epoch": 6.040034812880766, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 8.108030412907844e-05, | |
| "loss": 0.2631, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 6.048738033072237, | |
| "grad_norm": 0.86328125, | |
| "learning_rate": 8.078181937829656e-05, | |
| "loss": 0.2646, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 6.057441253263708, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 8.048351230929074e-05, | |
| "loss": 0.2621, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 6.066144473455179, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 8.018538568006027e-05, | |
| "loss": 0.267, | |
| "step": 3485 | |
| }, | |
| { | |
| "epoch": 6.07484769364665, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 7.988744224693625e-05, | |
| "loss": 0.2599, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 6.0835509138381205, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 7.958968476455608e-05, | |
| "loss": 0.2643, | |
| "step": 3495 | |
| }, | |
| { | |
| "epoch": 6.092254134029591, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 7.929211598583794e-05, | |
| "loss": 0.269, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 6.100957354221062, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 7.899473866195526e-05, | |
| "loss": 0.2622, | |
| "step": 3505 | |
| }, | |
| { | |
| "epoch": 6.1096605744125325, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 7.869755554231145e-05, | |
| "loss": 0.2633, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 6.1183637946040035, | |
| "grad_norm": 0.76953125, | |
| "learning_rate": 7.840056937451444e-05, | |
| "loss": 0.2687, | |
| "step": 3515 | |
| }, | |
| { | |
| "epoch": 6.127067014795474, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 7.810378290435108e-05, | |
| "loss": 0.2622, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 6.135770234986945, | |
| "grad_norm": 0.87109375, | |
| "learning_rate": 7.780719887576213e-05, | |
| "loss": 0.2652, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 6.144473455178416, | |
| "grad_norm": 2.03125, | |
| "learning_rate": 7.751082003081653e-05, | |
| "loss": 0.267, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 6.153176675369886, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 7.721464910968627e-05, | |
| "loss": 0.2621, | |
| "step": 3535 | |
| }, | |
| { | |
| "epoch": 6.161879895561357, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 7.691868885062088e-05, | |
| "loss": 0.2614, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 6.170583115752828, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 7.662294198992228e-05, | |
| "loss": 0.264, | |
| "step": 3545 | |
| }, | |
| { | |
| "epoch": 6.179286335944299, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 7.632741126191947e-05, | |
| "loss": 0.267, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 6.18798955613577, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 7.603209939894312e-05, | |
| "loss": 0.2638, | |
| "step": 3555 | |
| }, | |
| { | |
| "epoch": 6.196692776327241, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 7.573700913130035e-05, | |
| "loss": 0.2614, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 6.205395996518712, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 7.544214318724961e-05, | |
| "loss": 0.2659, | |
| "step": 3565 | |
| }, | |
| { | |
| "epoch": 6.214099216710183, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 7.514750429297528e-05, | |
| "loss": 0.2686, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 6.222802436901653, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 7.485309517256267e-05, | |
| "loss": 0.268, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 6.231505657093124, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 7.455891854797256e-05, | |
| "loss": 0.2652, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 6.240208877284595, | |
| "grad_norm": 0.76953125, | |
| "learning_rate": 7.426497713901629e-05, | |
| "loss": 0.2638, | |
| "step": 3585 | |
| }, | |
| { | |
| "epoch": 6.248912097476066, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 7.397127366333048e-05, | |
| "loss": 0.2649, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 6.257615317667537, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 7.3677810836352e-05, | |
| "loss": 0.271, | |
| "step": 3595 | |
| }, | |
| { | |
| "epoch": 6.266318537859008, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 7.338459137129266e-05, | |
| "loss": 0.2661, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 6.275021758050479, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 7.309161797911441e-05, | |
| "loss": 0.2693, | |
| "step": 3605 | |
| }, | |
| { | |
| "epoch": 6.283724978241949, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 7.279889336850408e-05, | |
| "loss": 0.2668, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 6.29242819843342, | |
| "grad_norm": 0.81640625, | |
| "learning_rate": 7.250642024584835e-05, | |
| "loss": 0.2709, | |
| "step": 3615 | |
| }, | |
| { | |
| "epoch": 6.301131418624891, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 7.22142013152088e-05, | |
| "loss": 0.2682, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 6.309834638816362, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 7.192223927829689e-05, | |
| "loss": 0.264, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 6.318537859007833, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 7.163053683444901e-05, | |
| "loss": 0.2719, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 6.327241079199304, | |
| "grad_norm": 0.87890625, | |
| "learning_rate": 7.133909668060131e-05, | |
| "loss": 0.2715, | |
| "step": 3635 | |
| }, | |
| { | |
| "epoch": 6.335944299390775, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 7.104792151126515e-05, | |
| "loss": 0.263, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 6.344647519582246, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 7.075701401850183e-05, | |
| "loss": 0.2629, | |
| "step": 3645 | |
| }, | |
| { | |
| "epoch": 6.353350739773716, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 7.046637689189794e-05, | |
| "loss": 0.2674, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 6.362053959965187, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 7.017601281854027e-05, | |
| "loss": 0.2684, | |
| "step": 3655 | |
| }, | |
| { | |
| "epoch": 6.370757180156658, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 6.988592448299124e-05, | |
| "loss": 0.2652, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 6.379460400348129, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 6.959611456726387e-05, | |
| "loss": 0.2642, | |
| "step": 3665 | |
| }, | |
| { | |
| "epoch": 6.3881636205396, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 6.930658575079705e-05, | |
| "loss": 0.2696, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 6.396866840731071, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 6.901734071043071e-05, | |
| "loss": 0.27, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 6.405570060922542, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 6.872838212038122e-05, | |
| "loss": 0.2699, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 6.414273281114012, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 6.843971265221655e-05, | |
| "loss": 0.2687, | |
| "step": 3685 | |
| }, | |
| { | |
| "epoch": 6.422976501305483, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 6.815133497483157e-05, | |
| "loss": 0.2681, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 6.431679721496954, | |
| "grad_norm": 0.8828125, | |
| "learning_rate": 6.786325175442339e-05, | |
| "loss": 0.2631, | |
| "step": 3695 | |
| }, | |
| { | |
| "epoch": 6.440382941688425, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 6.75754656544667e-05, | |
| "loss": 0.2619, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 6.449086161879896, | |
| "grad_norm": 0.83203125, | |
| "learning_rate": 6.728797933568924e-05, | |
| "loss": 0.2658, | |
| "step": 3705 | |
| }, | |
| { | |
| "epoch": 6.4577893820713665, | |
| "grad_norm": 0.81640625, | |
| "learning_rate": 6.700079545604708e-05, | |
| "loss": 0.2696, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 6.4664926022628375, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 6.671391667070002e-05, | |
| "loss": 0.2707, | |
| "step": 3715 | |
| }, | |
| { | |
| "epoch": 6.4751958224543085, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 6.642734563198723e-05, | |
| "loss": 0.2653, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 6.4838990426457785, | |
| "grad_norm": 0.8984375, | |
| "learning_rate": 6.614108498940252e-05, | |
| "loss": 0.2721, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 6.4926022628372495, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 6.585513738956996e-05, | |
| "loss": 0.2674, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 6.5013054830287205, | |
| "grad_norm": 0.75, | |
| "learning_rate": 6.556950547621936e-05, | |
| "loss": 0.2689, | |
| "step": 3735 | |
| }, | |
| { | |
| "epoch": 6.510008703220191, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 6.52841918901619e-05, | |
| "loss": 0.2695, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 6.518711923411662, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 6.499919926926566e-05, | |
| "loss": 0.269, | |
| "step": 3745 | |
| }, | |
| { | |
| "epoch": 6.527415143603133, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 6.471453024843113e-05, | |
| "loss": 0.2655, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 6.536118363794604, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 6.44301874595671e-05, | |
| "loss": 0.265, | |
| "step": 3755 | |
| }, | |
| { | |
| "epoch": 6.544821583986074, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 6.414617353156605e-05, | |
| "loss": 0.2627, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 6.553524804177545, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 6.386249109028013e-05, | |
| "loss": 0.2724, | |
| "step": 3765 | |
| }, | |
| { | |
| "epoch": 6.562228024369016, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 6.357914275849652e-05, | |
| "loss": 0.2693, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 6.570931244560487, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 6.329613115591359e-05, | |
| "loss": 0.273, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 6.579634464751958, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 6.301345889911637e-05, | |
| "loss": 0.2665, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 6.588337684943429, | |
| "grad_norm": 0.76953125, | |
| "learning_rate": 6.273112860155251e-05, | |
| "loss": 0.2676, | |
| "step": 3785 | |
| }, | |
| { | |
| "epoch": 6.5970409051349, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 6.2449142873508e-05, | |
| "loss": 0.2659, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 6.605744125326371, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 6.21675043220832e-05, | |
| "loss": 0.2691, | |
| "step": 3795 | |
| }, | |
| { | |
| "epoch": 6.614447345517842, | |
| "grad_norm": 0.89453125, | |
| "learning_rate": 6.188621555116865e-05, | |
| "loss": 0.273, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 6.623150565709312, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 6.160527916142093e-05, | |
| "loss": 0.2637, | |
| "step": 3805 | |
| }, | |
| { | |
| "epoch": 6.631853785900783, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 6.132469775023867e-05, | |
| "loss": 0.2665, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 6.640557006092254, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 6.104447391173858e-05, | |
| "loss": 0.2675, | |
| "step": 3815 | |
| }, | |
| { | |
| "epoch": 6.649260226283725, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 6.0764610236731524e-05, | |
| "loss": 0.2696, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 6.657963446475196, | |
| "grad_norm": 0.81640625, | |
| "learning_rate": 6.048510931269824e-05, | |
| "loss": 0.2654, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 6.666666666666667, | |
| "grad_norm": 0.83203125, | |
| "learning_rate": 6.020597372376589e-05, | |
| "loss": 0.2746, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 6.675369886858137, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 5.992720605068378e-05, | |
| "loss": 0.2731, | |
| "step": 3835 | |
| }, | |
| { | |
| "epoch": 6.684073107049608, | |
| "grad_norm": 0.9140625, | |
| "learning_rate": 5.964880887079972e-05, | |
| "loss": 0.2694, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 6.692776327241079, | |
| "grad_norm": 0.8984375, | |
| "learning_rate": 5.937078475803607e-05, | |
| "loss": 0.2718, | |
| "step": 3845 | |
| }, | |
| { | |
| "epoch": 6.70147954743255, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 5.909313628286601e-05, | |
| "loss": 0.2679, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 6.710182767624021, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 5.881586601228983e-05, | |
| "loss": 0.2644, | |
| "step": 3855 | |
| }, | |
| { | |
| "epoch": 6.718885987815492, | |
| "grad_norm": 0.81640625, | |
| "learning_rate": 5.853897650981107e-05, | |
| "loss": 0.2712, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 6.727589208006963, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 5.8262470335412834e-05, | |
| "loss": 0.2645, | |
| "step": 3865 | |
| }, | |
| { | |
| "epoch": 6.736292428198434, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 5.798635004553421e-05, | |
| "loss": 0.2668, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 6.744995648389905, | |
| "grad_norm": 0.82421875, | |
| "learning_rate": 5.771061819304664e-05, | |
| "loss": 0.2735, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 6.753698868581375, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 5.7435277327230206e-05, | |
| "loss": 0.2721, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 6.762402088772846, | |
| "grad_norm": 0.84375, | |
| "learning_rate": 5.716032999375006e-05, | |
| "loss": 0.2654, | |
| "step": 3885 | |
| }, | |
| { | |
| "epoch": 6.771105308964317, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 5.6885778734633074e-05, | |
| "loss": 0.2701, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 6.779808529155788, | |
| "grad_norm": 0.77734375, | |
| "learning_rate": 5.6611626088244194e-05, | |
| "loss": 0.2684, | |
| "step": 3895 | |
| }, | |
| { | |
| "epoch": 6.788511749347259, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 5.6337874589262915e-05, | |
| "loss": 0.2686, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 6.79721496953873, | |
| "grad_norm": 0.83984375, | |
| "learning_rate": 5.606452676865993e-05, | |
| "loss": 0.2666, | |
| "step": 3905 | |
| }, | |
| { | |
| "epoch": 6.8059181897302, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 5.5791585153673774e-05, | |
| "loss": 0.2687, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 6.814621409921671, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 5.5519052267787444e-05, | |
| "loss": 0.2667, | |
| "step": 3915 | |
| }, | |
| { | |
| "epoch": 6.823324630113142, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 5.524693063070492e-05, | |
| "loss": 0.2689, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 6.832027850304613, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 5.497522275832799e-05, | |
| "loss": 0.2666, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 6.8407310704960835, | |
| "grad_norm": 0.79296875, | |
| "learning_rate": 5.4703931162733116e-05, | |
| "loss": 0.265, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 6.8494342906875545, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 5.4433058352147914e-05, | |
| "loss": 0.2667, | |
| "step": 3935 | |
| }, | |
| { | |
| "epoch": 6.8581375108790255, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 5.416260683092814e-05, | |
| "loss": 0.2629, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 6.866840731070496, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 5.389257909953462e-05, | |
| "loss": 0.2712, | |
| "step": 3945 | |
| }, | |
| { | |
| "epoch": 6.875543951261967, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 5.362297765450999e-05, | |
| "loss": 0.2671, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 6.8842471714534375, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 5.335380498845559e-05, | |
| "loss": 0.261, | |
| "step": 3955 | |
| }, | |
| { | |
| "epoch": 6.892950391644908, | |
| "grad_norm": 0.83203125, | |
| "learning_rate": 5.308506359000851e-05, | |
| "loss": 0.2663, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 6.901653611836379, | |
| "grad_norm": 0.75, | |
| "learning_rate": 5.281675594381859e-05, | |
| "loss": 0.2673, | |
| "step": 3965 | |
| }, | |
| { | |
| "epoch": 6.91035683202785, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 5.25488845305254e-05, | |
| "loss": 0.2691, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 6.919060052219321, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 5.228145182673532e-05, | |
| "loss": 0.2725, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 6.927763272410792, | |
| "grad_norm": 0.84765625, | |
| "learning_rate": 5.2014460304998545e-05, | |
| "loss": 0.2653, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 6.936466492602263, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 5.1747912433786497e-05, | |
| "loss": 0.2661, | |
| "step": 3985 | |
| }, | |
| { | |
| "epoch": 6.945169712793733, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 5.148181067746862e-05, | |
| "loss": 0.2707, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 6.953872932985204, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 5.121615749629003e-05, | |
| "loss": 0.267, | |
| "step": 3995 | |
| }, | |
| { | |
| "epoch": 6.962576153176675, | |
| "grad_norm": 0.796875, | |
| "learning_rate": 5.0950955346348314e-05, | |
| "loss": 0.2662, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.971279373368146, | |
| "grad_norm": 0.8359375, | |
| "learning_rate": 5.068620667957123e-05, | |
| "loss": 0.2695, | |
| "step": 4005 | |
| }, | |
| { | |
| "epoch": 6.979982593559617, | |
| "grad_norm": 0.859375, | |
| "learning_rate": 5.042191394369371e-05, | |
| "loss": 0.266, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 6.988685813751088, | |
| "grad_norm": 0.83203125, | |
| "learning_rate": 5.01580795822355e-05, | |
| "loss": 0.2737, | |
| "step": 4015 | |
| }, | |
| { | |
| "epoch": 6.997389033942559, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 4.989470603447835e-05, | |
| "loss": 0.2672, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 6.999129677980853, | |
| "eval_loss": 4.307767391204834, | |
| "eval_runtime": 1.1109, | |
| "eval_samples_per_second": 5.401, | |
| "eval_steps_per_second": 0.9, | |
| "step": 4021 | |
| }, | |
| { | |
| "epoch": 7.00609225413403, | |
| "grad_norm": 0.57421875, | |
| "learning_rate": 4.963179573544357e-05, | |
| "loss": 0.2314, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 7.0147954743255, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 4.9369351115869535e-05, | |
| "loss": 0.2146, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 7.023498694516971, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 4.9107374602189216e-05, | |
| "loss": 0.2171, | |
| "step": 4035 | |
| }, | |
| { | |
| "epoch": 7.032201914708442, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 4.8845868616507617e-05, | |
| "loss": 0.2179, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 7.040905134899913, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 4.8584835576579466e-05, | |
| "loss": 0.2184, | |
| "step": 4045 | |
| }, | |
| { | |
| "epoch": 7.049608355091384, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 4.832427789578701e-05, | |
| "loss": 0.2178, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 7.058311575282855, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 4.806419798311739e-05, | |
| "loss": 0.214, | |
| "step": 4055 | |
| }, | |
| { | |
| "epoch": 7.067014795474326, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 4.7804598243140666e-05, | |
| "loss": 0.2176, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 7.075718015665796, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 4.754548107598736e-05, | |
| "loss": 0.2158, | |
| "step": 4065 | |
| }, | |
| { | |
| "epoch": 7.084421235857267, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 4.728684887732649e-05, | |
| "loss": 0.2175, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 7.093124456048738, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 4.702870403834317e-05, | |
| "loss": 0.2162, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 7.101827676240209, | |
| "grad_norm": 0.75, | |
| "learning_rate": 4.6771048945716664e-05, | |
| "loss": 0.2189, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 7.11053089643168, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 4.65138859815983e-05, | |
| "loss": 0.2187, | |
| "step": 4085 | |
| }, | |
| { | |
| "epoch": 7.119234116623151, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 4.62572175235895e-05, | |
| "loss": 0.2207, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 7.127937336814622, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 4.60010459447196e-05, | |
| "loss": 0.2111, | |
| "step": 4095 | |
| }, | |
| { | |
| "epoch": 7.136640557006093, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 4.574537361342407e-05, | |
| "loss": 0.2194, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 7.145343777197563, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 4.5490202893522614e-05, | |
| "loss": 0.2172, | |
| "step": 4105 | |
| }, | |
| { | |
| "epoch": 7.154046997389034, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 4.5235536144197353e-05, | |
| "loss": 0.2194, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 7.162750217580505, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 4.498137571997081e-05, | |
| "loss": 0.2166, | |
| "step": 4115 | |
| }, | |
| { | |
| "epoch": 7.171453437771976, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 4.472772397068431e-05, | |
| "loss": 0.2176, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 7.180156657963447, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 4.447458324147629e-05, | |
| "loss": 0.225, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 7.188859878154918, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 4.422195587276058e-05, | |
| "loss": 0.217, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 7.1975630983463885, | |
| "grad_norm": 0.75, | |
| "learning_rate": 4.396984420020451e-05, | |
| "loss": 0.2182, | |
| "step": 4135 | |
| }, | |
| { | |
| "epoch": 7.206266318537859, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 4.3718250554707784e-05, | |
| "loss": 0.2171, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 7.21496953872933, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 4.34671772623806e-05, | |
| "loss": 0.2155, | |
| "step": 4145 | |
| }, | |
| { | |
| "epoch": 7.2236727589208005, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 4.321662664452221e-05, | |
| "loss": 0.217, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 7.2323759791122715, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 4.296660101759942e-05, | |
| "loss": 0.2158, | |
| "step": 4155 | |
| }, | |
| { | |
| "epoch": 7.241079199303742, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 4.271710269322536e-05, | |
| "loss": 0.2191, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 7.249782419495213, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 4.2468133978137945e-05, | |
| "loss": 0.2119, | |
| "step": 4165 | |
| }, | |
| { | |
| "epoch": 7.258485639686684, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 4.221969717417852e-05, | |
| "loss": 0.2125, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 7.267188859878155, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 4.1971794578270654e-05, | |
| "loss": 0.2176, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 7.275892080069625, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 4.1724428482398945e-05, | |
| "loss": 0.2171, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 7.284595300261096, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 4.1477601173587836e-05, | |
| "loss": 0.2168, | |
| "step": 4185 | |
| }, | |
| { | |
| "epoch": 7.293298520452567, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 4.1231314933880175e-05, | |
| "loss": 0.2171, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 7.302001740644038, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 4.098557204031658e-05, | |
| "loss": 0.217, | |
| "step": 4195 | |
| }, | |
| { | |
| "epoch": 7.310704960835509, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 4.0740374764914136e-05, | |
| "loss": 0.2184, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 7.31940818102698, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 4.049572537464531e-05, | |
| "loss": 0.2126, | |
| "step": 4205 | |
| }, | |
| { | |
| "epoch": 7.328111401218451, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 4.025162613141713e-05, | |
| "loss": 0.2173, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 7.336814621409921, | |
| "grad_norm": 0.78125, | |
| "learning_rate": 4.000807929205027e-05, | |
| "loss": 0.2113, | |
| "step": 4215 | |
| }, | |
| { | |
| "epoch": 7.345517841601392, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 3.9765087108258204e-05, | |
| "loss": 0.2215, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 7.354221061792863, | |
| "grad_norm": 0.75, | |
| "learning_rate": 3.95226518266262e-05, | |
| "loss": 0.2204, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 7.362924281984334, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 3.9280775688590735e-05, | |
| "loss": 0.2169, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 7.371627502175805, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 3.903946093041877e-05, | |
| "loss": 0.2188, | |
| "step": 4235 | |
| }, | |
| { | |
| "epoch": 7.380330722367276, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 3.8798709783187036e-05, | |
| "loss": 0.2162, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 7.389033942558747, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 3.85585244727613e-05, | |
| "loss": 0.2163, | |
| "step": 4245 | |
| }, | |
| { | |
| "epoch": 7.397737162750218, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 3.8318907219775935e-05, | |
| "loss": 0.2179, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 7.406440382941688, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 3.8079860239613395e-05, | |
| "loss": 0.2197, | |
| "step": 4255 | |
| }, | |
| { | |
| "epoch": 7.415143603133159, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 3.784138574238357e-05, | |
| "loss": 0.2177, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 7.42384682332463, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 3.760348593290348e-05, | |
| "loss": 0.2188, | |
| "step": 4265 | |
| }, | |
| { | |
| "epoch": 7.432550043516101, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 3.736616301067694e-05, | |
| "loss": 0.2187, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 7.441253263707572, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 3.7129419169874114e-05, | |
| "loss": 0.221, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 7.449956483899043, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 3.689325659931123e-05, | |
| "loss": 0.2236, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 7.458659704090514, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 3.6657677482430377e-05, | |
| "loss": 0.2188, | |
| "step": 4285 | |
| }, | |
| { | |
| "epoch": 7.467362924281984, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 3.642268399727941e-05, | |
| "loss": 0.2165, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 7.476066144473455, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 3.618827831649158e-05, | |
| "loss": 0.2183, | |
| "step": 4295 | |
| }, | |
| { | |
| "epoch": 7.484769364664926, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 3.595446260726576e-05, | |
| "loss": 0.2117, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 7.493472584856397, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 3.5721239031346066e-05, | |
| "loss": 0.2167, | |
| "step": 4305 | |
| }, | |
| { | |
| "epoch": 7.502175805047868, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 3.5488609745002214e-05, | |
| "loss": 0.219, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 7.510879025239339, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 3.525657689900923e-05, | |
| "loss": 0.2145, | |
| "step": 4315 | |
| }, | |
| { | |
| "epoch": 7.51958224543081, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 3.502514263862793e-05, | |
| "loss": 0.2159, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 7.528285465622281, | |
| "grad_norm": 0.76953125, | |
| "learning_rate": 3.479430910358474e-05, | |
| "loss": 0.2177, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 7.536988685813752, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 3.456407842805223e-05, | |
| "loss": 0.2154, | |
| "step": 4330 | |
| }, | |
| { | |
| "epoch": 7.545691906005222, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 3.433445274062907e-05, | |
| "loss": 0.2157, | |
| "step": 4335 | |
| }, | |
| { | |
| "epoch": 7.554395126196693, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 3.410543416432069e-05, | |
| "loss": 0.2122, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 7.563098346388164, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 3.387702481651931e-05, | |
| "loss": 0.2215, | |
| "step": 4345 | |
| }, | |
| { | |
| "epoch": 7.5718015665796345, | |
| "grad_norm": 0.75, | |
| "learning_rate": 3.364922680898458e-05, | |
| "loss": 0.2192, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 7.5805047867711055, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 3.342204224782406e-05, | |
| "loss": 0.2168, | |
| "step": 4355 | |
| }, | |
| { | |
| "epoch": 7.5892080069625765, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 3.3195473233473584e-05, | |
| "loss": 0.2163, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 7.5979112271540465, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 3.2969521860678066e-05, | |
| "loss": 0.2162, | |
| "step": 4365 | |
| }, | |
| { | |
| "epoch": 7.6066144473455175, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 3.2744190218471884e-05, | |
| "loss": 0.2178, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 7.6153176675369885, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 3.2519480390159806e-05, | |
| "loss": 0.218, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 7.624020887728459, | |
| "grad_norm": 0.8203125, | |
| "learning_rate": 3.229539445329752e-05, | |
| "loss": 0.216, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 7.63272410791993, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 3.207193447967264e-05, | |
| "loss": 0.2207, | |
| "step": 4385 | |
| }, | |
| { | |
| "epoch": 7.641427328111401, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 3.184910253528528e-05, | |
| "loss": 0.217, | |
| "step": 4390 | |
| }, | |
| { | |
| "epoch": 7.650130548302872, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 3.162690068032926e-05, | |
| "loss": 0.2183, | |
| "step": 4395 | |
| }, | |
| { | |
| "epoch": 7.658833768494343, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 3.140533096917282e-05, | |
| "loss": 0.2197, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 7.667536988685814, | |
| "grad_norm": 0.75, | |
| "learning_rate": 3.118439545033969e-05, | |
| "loss": 0.2204, | |
| "step": 4405 | |
| }, | |
| { | |
| "epoch": 7.676240208877284, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 3.096409616649023e-05, | |
| "loss": 0.2194, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 7.684943429068755, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 3.074443515440252e-05, | |
| "loss": 0.2211, | |
| "step": 4415 | |
| }, | |
| { | |
| "epoch": 7.693646649260226, | |
| "grad_norm": 0.8046875, | |
| "learning_rate": 3.0525414444953396e-05, | |
| "loss": 0.219, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 7.702349869451697, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 3.0307036063099782e-05, | |
| "loss": 0.2131, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 7.711053089643168, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 3.0089302027860044e-05, | |
| "loss": 0.2141, | |
| "step": 4430 | |
| }, | |
| { | |
| "epoch": 7.719756309834639, | |
| "grad_norm": 0.7890625, | |
| "learning_rate": 2.9872214352295213e-05, | |
| "loss": 0.2192, | |
| "step": 4435 | |
| }, | |
| { | |
| "epoch": 7.728459530026109, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 2.965577504349035e-05, | |
| "loss": 0.2214, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 7.73716275021758, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 2.9439986102536043e-05, | |
| "loss": 0.2188, | |
| "step": 4445 | |
| }, | |
| { | |
| "epoch": 7.745865970409051, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 2.9224849524509936e-05, | |
| "loss": 0.2155, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 7.754569190600522, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 2.901036729845831e-05, | |
| "loss": 0.2156, | |
| "step": 4455 | |
| }, | |
| { | |
| "epoch": 7.763272410791993, | |
| "grad_norm": 0.75, | |
| "learning_rate": 2.879654140737743e-05, | |
| "loss": 0.2161, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 7.771975630983464, | |
| "grad_norm": 0.80859375, | |
| "learning_rate": 2.8583373828195603e-05, | |
| "loss": 0.2185, | |
| "step": 4465 | |
| }, | |
| { | |
| "epoch": 7.780678851174935, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 2.837086653175468e-05, | |
| "loss": 0.2226, | |
| "step": 4470 | |
| }, | |
| { | |
| "epoch": 7.789382071366406, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 2.8159021482791802e-05, | |
| "loss": 0.2171, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 7.798085291557877, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 2.794784063992131e-05, | |
| "loss": 0.2204, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 7.806788511749347, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 2.7737325955616643e-05, | |
| "loss": 0.215, | |
| "step": 4485 | |
| }, | |
| { | |
| "epoch": 7.815491731940818, | |
| "grad_norm": 0.765625, | |
| "learning_rate": 2.7527479376192366e-05, | |
| "loss": 0.2161, | |
| "step": 4490 | |
| }, | |
| { | |
| "epoch": 7.824194952132289, | |
| "grad_norm": 0.75, | |
| "learning_rate": 2.7318302841785827e-05, | |
| "loss": 0.2187, | |
| "step": 4495 | |
| }, | |
| { | |
| "epoch": 7.83289817232376, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 2.7109798286339705e-05, | |
| "loss": 0.2214, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 7.841601392515231, | |
| "grad_norm": 0.80078125, | |
| "learning_rate": 2.6901967637583835e-05, | |
| "loss": 0.2142, | |
| "step": 4505 | |
| }, | |
| { | |
| "epoch": 7.850304612706702, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 2.669481281701739e-05, | |
| "loss": 0.2194, | |
| "step": 4510 | |
| }, | |
| { | |
| "epoch": 7.859007832898172, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 2.6488335739891178e-05, | |
| "loss": 0.2228, | |
| "step": 4515 | |
| }, | |
| { | |
| "epoch": 7.867711053089643, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 2.6282538315189974e-05, | |
| "loss": 0.2196, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 7.876414273281114, | |
| "grad_norm": 0.7734375, | |
| "learning_rate": 2.607742244561484e-05, | |
| "loss": 0.2225, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 7.885117493472585, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 2.5872990027565434e-05, | |
| "loss": 0.2163, | |
| "step": 4530 | |
| }, | |
| { | |
| "epoch": 7.893820713664056, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 2.5669242951122586e-05, | |
| "loss": 0.2155, | |
| "step": 4535 | |
| }, | |
| { | |
| "epoch": 7.902523933855527, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 2.5466183100030837e-05, | |
| "loss": 0.2167, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 7.911227154046998, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 2.5263812351680995e-05, | |
| "loss": 0.2184, | |
| "step": 4545 | |
| }, | |
| { | |
| "epoch": 7.919930374238469, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 2.50621325770927e-05, | |
| "loss": 0.2132, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 7.9286335944299395, | |
| "grad_norm": 0.75, | |
| "learning_rate": 2.4861145640897188e-05, | |
| "loss": 0.2144, | |
| "step": 4555 | |
| }, | |
| { | |
| "epoch": 7.93733681462141, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 2.466085340132014e-05, | |
| "loss": 0.2171, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 7.946040034812881, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 2.446125771016433e-05, | |
| "loss": 0.2167, | |
| "step": 4565 | |
| }, | |
| { | |
| "epoch": 7.9547432550043515, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 2.426236041279266e-05, | |
| "loss": 0.2196, | |
| "step": 4570 | |
| }, | |
| { | |
| "epoch": 7.9634464751958225, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 2.4064163348110956e-05, | |
| "loss": 0.2196, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 7.9721496953872935, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 2.3866668348551112e-05, | |
| "loss": 0.212, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 7.980852915578764, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 2.366987724005404e-05, | |
| "loss": 0.2119, | |
| "step": 4585 | |
| }, | |
| { | |
| "epoch": 7.9895561357702345, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 2.3473791842052774e-05, | |
| "loss": 0.2194, | |
| "step": 4590 | |
| }, | |
| { | |
| "epoch": 7.9982593559617055, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 2.327841396745578e-05, | |
| "loss": 0.2167, | |
| "step": 4595 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_loss": 4.845585823059082, | |
| "eval_runtime": 0.7795, | |
| "eval_samples_per_second": 7.697, | |
| "eval_steps_per_second": 1.283, | |
| "step": 4596 | |
| }, | |
| { | |
| "epoch": 8.006962576153176, | |
| "grad_norm": 0.62109375, | |
| "learning_rate": 2.3083745422630122e-05, | |
| "loss": 0.2056, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 8.015665796344647, | |
| "grad_norm": 0.6328125, | |
| "learning_rate": 2.2889788007384683e-05, | |
| "loss": 0.1978, | |
| "step": 4605 | |
| }, | |
| { | |
| "epoch": 8.024369016536118, | |
| "grad_norm": 0.63671875, | |
| "learning_rate": 2.2696543514953595e-05, | |
| "loss": 0.2014, | |
| "step": 4610 | |
| }, | |
| { | |
| "epoch": 8.03307223672759, | |
| "grad_norm": 0.62890625, | |
| "learning_rate": 2.2504013731979732e-05, | |
| "loss": 0.1991, | |
| "step": 4615 | |
| }, | |
| { | |
| "epoch": 8.04177545691906, | |
| "grad_norm": 0.6640625, | |
| "learning_rate": 2.2312200438498043e-05, | |
| "loss": 0.2006, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 8.050478677110531, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 2.212110540791924e-05, | |
| "loss": 0.2018, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 8.059181897302002, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 2.1930730407013245e-05, | |
| "loss": 0.1963, | |
| "step": 4630 | |
| }, | |
| { | |
| "epoch": 8.067885117493473, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 2.1741077195893043e-05, | |
| "loss": 0.1995, | |
| "step": 4635 | |
| }, | |
| { | |
| "epoch": 8.076588337684944, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 2.1552147527998213e-05, | |
| "loss": 0.1984, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 8.085291557876415, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 2.136394315007889e-05, | |
| "loss": 0.2005, | |
| "step": 4645 | |
| }, | |
| { | |
| "epoch": 8.093994778067884, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 2.1176465802179467e-05, | |
| "loss": 0.203, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 8.102697998259355, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 2.0989717217622652e-05, | |
| "loss": 0.1967, | |
| "step": 4655 | |
| }, | |
| { | |
| "epoch": 8.111401218450826, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 2.0803699122993293e-05, | |
| "loss": 0.2029, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 8.120104438642297, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 2.061841323812257e-05, | |
| "loss": 0.2005, | |
| "step": 4665 | |
| }, | |
| { | |
| "epoch": 8.128807658833768, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 2.0433861276071942e-05, | |
| "loss": 0.1966, | |
| "step": 4670 | |
| }, | |
| { | |
| "epoch": 8.137510879025239, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 2.0250044943117385e-05, | |
| "loss": 0.2023, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 8.14621409921671, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 2.0066965938733707e-05, | |
| "loss": 0.198, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 8.154917319408181, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 1.9884625955578594e-05, | |
| "loss": 0.196, | |
| "step": 4685 | |
| }, | |
| { | |
| "epoch": 8.163620539599652, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 1.9703026679477256e-05, | |
| "loss": 0.1954, | |
| "step": 4690 | |
| }, | |
| { | |
| "epoch": 8.172323759791123, | |
| "grad_norm": 0.640625, | |
| "learning_rate": 1.9522169789406575e-05, | |
| "loss": 0.196, | |
| "step": 4695 | |
| }, | |
| { | |
| "epoch": 8.181026979982594, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 1.934205695747978e-05, | |
| "loss": 0.2014, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 8.189730200174065, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 1.916268984893086e-05, | |
| "loss": 0.1984, | |
| "step": 4705 | |
| }, | |
| { | |
| "epoch": 8.198433420365536, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 1.8984070122099218e-05, | |
| "loss": 0.1994, | |
| "step": 4710 | |
| }, | |
| { | |
| "epoch": 8.207136640557007, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 1.880619942841435e-05, | |
| "loss": 0.2002, | |
| "step": 4715 | |
| }, | |
| { | |
| "epoch": 8.215839860748478, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 1.862907941238059e-05, | |
| "loss": 0.197, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 8.224543080939949, | |
| "grad_norm": 0.8125, | |
| "learning_rate": 1.8452711711561842e-05, | |
| "loss": 0.2023, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 8.233246301131418, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 1.8277097956566437e-05, | |
| "loss": 0.201, | |
| "step": 4730 | |
| }, | |
| { | |
| "epoch": 8.241949521322889, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 1.810223977103217e-05, | |
| "loss": 0.1982, | |
| "step": 4735 | |
| }, | |
| { | |
| "epoch": 8.25065274151436, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 1.7928138771611225e-05, | |
| "loss": 0.1983, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 8.25935596170583, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 1.7754796567955155e-05, | |
| "loss": 0.2005, | |
| "step": 4745 | |
| }, | |
| { | |
| "epoch": 8.268059181897302, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 1.7582214762700054e-05, | |
| "loss": 0.1974, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 8.276762402088773, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 1.7410394951451814e-05, | |
| "loss": 0.1993, | |
| "step": 4755 | |
| }, | |
| { | |
| "epoch": 8.285465622280244, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 1.7239338722771327e-05, | |
| "loss": 0.2046, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 8.294168842471715, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 1.706904765815963e-05, | |
| "loss": 0.2007, | |
| "step": 4765 | |
| }, | |
| { | |
| "epoch": 8.302872062663186, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 1.6899523332043586e-05, | |
| "loss": 0.2041, | |
| "step": 4770 | |
| }, | |
| { | |
| "epoch": 8.311575282854657, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 1.673076731176114e-05, | |
| "loss": 0.2024, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 8.320278503046127, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 1.6562781157546835e-05, | |
| "loss": 0.2025, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 8.328981723237598, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 1.639556642251737e-05, | |
| "loss": 0.1961, | |
| "step": 4785 | |
| }, | |
| { | |
| "epoch": 8.33768494342907, | |
| "grad_norm": 0.75, | |
| "learning_rate": 1.622912465265738e-05, | |
| "loss": 0.1966, | |
| "step": 4790 | |
| }, | |
| { | |
| "epoch": 8.34638816362054, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 1.6063457386805004e-05, | |
| "loss": 0.1987, | |
| "step": 4795 | |
| }, | |
| { | |
| "epoch": 8.35509138381201, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 1.5898566156637708e-05, | |
| "loss": 0.2005, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 8.36379460400348, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 1.573445248665806e-05, | |
| "loss": 0.1993, | |
| "step": 4805 | |
| }, | |
| { | |
| "epoch": 8.372497824194951, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 1.5571117894179754e-05, | |
| "loss": 0.2004, | |
| "step": 4810 | |
| }, | |
| { | |
| "epoch": 8.381201044386422, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 1.540856388931359e-05, | |
| "loss": 0.1989, | |
| "step": 4815 | |
| }, | |
| { | |
| "epoch": 8.389904264577893, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 1.5246791974953223e-05, | |
| "loss": 0.1935, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 8.398607484769364, | |
| "grad_norm": 0.625, | |
| "learning_rate": 1.5085803646761687e-05, | |
| "loss": 0.1989, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 8.407310704960835, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 1.4925600393157324e-05, | |
| "loss": 0.1976, | |
| "step": 4830 | |
| }, | |
| { | |
| "epoch": 8.416013925152306, | |
| "grad_norm": 0.7578125, | |
| "learning_rate": 1.4766183695300006e-05, | |
| "loss": 0.2008, | |
| "step": 4835 | |
| }, | |
| { | |
| "epoch": 8.424717145343777, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 1.4607555027077525e-05, | |
| "loss": 0.2007, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 8.433420365535248, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 1.4449715855091972e-05, | |
| "loss": 0.1992, | |
| "step": 4845 | |
| }, | |
| { | |
| "epoch": 8.44212358572672, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 1.429266763864614e-05, | |
| "loss": 0.1959, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 8.45082680591819, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 1.4136411829730023e-05, | |
| "loss": 0.1981, | |
| "step": 4855 | |
| }, | |
| { | |
| "epoch": 8.459530026109661, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 1.3980949873007364e-05, | |
| "loss": 0.2006, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 8.468233246301132, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 1.3826283205802427e-05, | |
| "loss": 0.1991, | |
| "step": 4865 | |
| }, | |
| { | |
| "epoch": 8.476936466492603, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 1.3672413258086592e-05, | |
| "loss": 0.1991, | |
| "step": 4870 | |
| }, | |
| { | |
| "epoch": 8.485639686684074, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 1.3519341452465151e-05, | |
| "loss": 0.2025, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 8.494342906875543, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 1.336706920416415e-05, | |
| "loss": 0.2, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 8.503046127067014, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 1.3215597921017387e-05, | |
| "loss": 0.2004, | |
| "step": 4885 | |
| }, | |
| { | |
| "epoch": 8.511749347258485, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 1.3064929003453286e-05, | |
| "loss": 0.1985, | |
| "step": 4890 | |
| }, | |
| { | |
| "epoch": 8.520452567449956, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 1.2915063844481989e-05, | |
| "loss": 0.1978, | |
| "step": 4895 | |
| }, | |
| { | |
| "epoch": 8.529155787641427, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 1.2766003829682505e-05, | |
| "loss": 0.1972, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 8.537859007832898, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 1.2617750337189904e-05, | |
| "loss": 0.1993, | |
| "step": 4905 | |
| }, | |
| { | |
| "epoch": 8.546562228024369, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 1.2470304737682514e-05, | |
| "loss": 0.1956, | |
| "step": 4910 | |
| }, | |
| { | |
| "epoch": 8.55526544821584, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 1.232366839436926e-05, | |
| "loss": 0.1976, | |
| "step": 4915 | |
| }, | |
| { | |
| "epoch": 8.56396866840731, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 1.2177842662977135e-05, | |
| "loss": 0.192, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 8.572671888598782, | |
| "grad_norm": 0.78515625, | |
| "learning_rate": 1.2032828891738646e-05, | |
| "loss": 0.2021, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 8.581375108790253, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 1.1888628421379221e-05, | |
| "loss": 0.1987, | |
| "step": 4930 | |
| }, | |
| { | |
| "epoch": 8.590078328981724, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 1.1745242585104955e-05, | |
| "loss": 0.2024, | |
| "step": 4935 | |
| }, | |
| { | |
| "epoch": 8.598781549173195, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 1.160267270859029e-05, | |
| "loss": 0.2027, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 8.607484769364666, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 1.1460920109965612e-05, | |
| "loss": 0.2012, | |
| "step": 4945 | |
| }, | |
| { | |
| "epoch": 8.616187989556135, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 1.1319986099805279e-05, | |
| "loss": 0.2001, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 8.624891209747606, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 1.1179871981115253e-05, | |
| "loss": 0.2014, | |
| "step": 4955 | |
| }, | |
| { | |
| "epoch": 8.633594429939077, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 1.1040579049321309e-05, | |
| "loss": 0.2014, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 8.642297650130548, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 1.0902108592256831e-05, | |
| "loss": 0.2002, | |
| "step": 4965 | |
| }, | |
| { | |
| "epoch": 8.651000870322019, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 1.0764461890151112e-05, | |
| "loss": 0.1967, | |
| "step": 4970 | |
| }, | |
| { | |
| "epoch": 8.65970409051349, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 1.062764021561733e-05, | |
| "loss": 0.2005, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 8.66840731070496, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 1.0491644833640868e-05, | |
| "loss": 0.2013, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 8.677110530896432, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 1.0356477001567677e-05, | |
| "loss": 0.197, | |
| "step": 4985 | |
| }, | |
| { | |
| "epoch": 8.685813751087903, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 1.0222137969092581e-05, | |
| "loss": 0.2012, | |
| "step": 4990 | |
| }, | |
| { | |
| "epoch": 8.694516971279374, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 1.0088628978247694e-05, | |
| "loss": 0.2006, | |
| "step": 4995 | |
| }, | |
| { | |
| "epoch": 8.703220191470844, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 9.955951263390972e-06, | |
| "loss": 0.1987, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 8.711923411662315, | |
| "grad_norm": 0.62890625, | |
| "learning_rate": 9.824106051194859e-06, | |
| "loss": 0.1977, | |
| "step": 5005 | |
| }, | |
| { | |
| "epoch": 8.720626631853786, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 9.69309456063484e-06, | |
| "loss": 0.1986, | |
| "step": 5010 | |
| }, | |
| { | |
| "epoch": 8.729329852045257, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 9.562918002978283e-06, | |
| "loss": 0.2016, | |
| "step": 5015 | |
| }, | |
| { | |
| "epoch": 8.738033072236728, | |
| "grad_norm": 0.66015625, | |
| "learning_rate": 9.43357758177309e-06, | |
| "loss": 0.1969, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 8.7467362924282, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 9.305074492836763e-06, | |
| "loss": 0.197, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 8.755439512619668, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 9.177409924245161e-06, | |
| "loss": 0.1953, | |
| "step": 5030 | |
| }, | |
| { | |
| "epoch": 8.76414273281114, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 9.050585056321626e-06, | |
| "loss": 0.1979, | |
| "step": 5035 | |
| }, | |
| { | |
| "epoch": 8.77284595300261, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 8.924601061626048e-06, | |
| "loss": 0.1969, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 8.781549173194081, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 8.799459104944064e-06, | |
| "loss": 0.1983, | |
| "step": 5045 | |
| }, | |
| { | |
| "epoch": 8.790252393385552, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 8.675160343276167e-06, | |
| "loss": 0.1982, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 8.798955613577023, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 8.551705925827103e-06, | |
| "loss": 0.1989, | |
| "step": 5055 | |
| }, | |
| { | |
| "epoch": 8.807658833768494, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 8.429096993995277e-06, | |
| "loss": 0.1958, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 8.816362053959965, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 8.307334681362133e-06, | |
| "loss": 0.1996, | |
| "step": 5065 | |
| }, | |
| { | |
| "epoch": 8.825065274151436, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 8.18642011368167e-06, | |
| "loss": 0.2031, | |
| "step": 5070 | |
| }, | |
| { | |
| "epoch": 8.833768494342907, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 8.066354408870048e-06, | |
| "loss": 0.201, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 8.842471714534378, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 7.947138676995302e-06, | |
| "loss": 0.2003, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 8.851174934725849, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 7.828774020267072e-06, | |
| "loss": 0.1989, | |
| "step": 5085 | |
| }, | |
| { | |
| "epoch": 8.85987815491732, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 7.711261533026238e-06, | |
| "loss": 0.2007, | |
| "step": 5090 | |
| }, | |
| { | |
| "epoch": 8.868581375108791, | |
| "grad_norm": 0.72265625, | |
| "learning_rate": 7.594602301735087e-06, | |
| "loss": 0.204, | |
| "step": 5095 | |
| }, | |
| { | |
| "epoch": 8.877284595300262, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 7.478797404967075e-06, | |
| "loss": 0.1964, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 8.885987815491731, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 7.363847913396882e-06, | |
| "loss": 0.1953, | |
| "step": 5105 | |
| }, | |
| { | |
| "epoch": 8.894691035683202, | |
| "grad_norm": 0.74609375, | |
| "learning_rate": 7.249754889790539e-06, | |
| "loss": 0.2054, | |
| "step": 5110 | |
| }, | |
| { | |
| "epoch": 8.903394255874673, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 7.136519388995633e-06, | |
| "loss": 0.1996, | |
| "step": 5115 | |
| }, | |
| { | |
| "epoch": 8.912097476066144, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 7.024142457931504e-06, | |
| "loss": 0.198, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 8.920800696257615, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 6.9126251355795864e-06, | |
| "loss": 0.1938, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 8.929503916449086, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 6.8019684529737505e-06, | |
| "loss": 0.2041, | |
| "step": 5130 | |
| }, | |
| { | |
| "epoch": 8.938207136640557, | |
| "grad_norm": 0.75, | |
| "learning_rate": 6.6921734331908735e-06, | |
| "loss": 0.199, | |
| "step": 5135 | |
| }, | |
| { | |
| "epoch": 8.946910356832028, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 6.583241091341353e-06, | |
| "loss": 0.1971, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 8.955613577023499, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 6.475172434559573e-06, | |
| "loss": 0.1962, | |
| "step": 5145 | |
| }, | |
| { | |
| "epoch": 8.96431679721497, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 6.367968461994833e-06, | |
| "loss": 0.1993, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 8.97302001740644, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 6.261630164801957e-06, | |
| "loss": 0.2026, | |
| "step": 5155 | |
| }, | |
| { | |
| "epoch": 8.981723237597912, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 6.156158526132139e-06, | |
| "loss": 0.1999, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 8.990426457789383, | |
| "grad_norm": 0.76953125, | |
| "learning_rate": 6.05155452112387e-06, | |
| "loss": 0.1983, | |
| "step": 5165 | |
| }, | |
| { | |
| "epoch": 8.999129677980854, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 5.947819116893971e-06, | |
| "loss": 0.2037, | |
| "step": 5170 | |
| }, | |
| { | |
| "epoch": 8.999129677980854, | |
| "eval_loss": 5.056090831756592, | |
| "eval_runtime": 1.1157, | |
| "eval_samples_per_second": 5.378, | |
| "eval_steps_per_second": 0.896, | |
| "step": 5170 | |
| }, | |
| { | |
| "epoch": 9.007832898172325, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 5.8449532725286196e-06, | |
| "loss": 0.1957, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 9.016536118363794, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 5.742957939074412e-06, | |
| "loss": 0.1967, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 9.025239338555265, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 5.641834059529661e-06, | |
| "loss": 0.1998, | |
| "step": 5185 | |
| }, | |
| { | |
| "epoch": 9.033942558746736, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 5.541582568835679e-06, | |
| "loss": 0.2032, | |
| "step": 5190 | |
| }, | |
| { | |
| "epoch": 9.042645778938207, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 5.442204393868056e-06, | |
| "loss": 0.1979, | |
| "step": 5195 | |
| }, | |
| { | |
| "epoch": 9.051348999129678, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 5.343700453428168e-06, | |
| "loss": 0.1942, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 9.060052219321149, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 5.246071658234642e-06, | |
| "loss": 0.2022, | |
| "step": 5205 | |
| }, | |
| { | |
| "epoch": 9.06875543951262, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 5.1493189109149575e-06, | |
| "loss": 0.2016, | |
| "step": 5210 | |
| }, | |
| { | |
| "epoch": 9.07745865970409, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 5.0534431059970685e-06, | |
| "loss": 0.1946, | |
| "step": 5215 | |
| }, | |
| { | |
| "epoch": 9.086161879895561, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 4.958445129901146e-06, | |
| "loss": 0.2002, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 9.094865100087032, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 4.864325860931429e-06, | |
| "loss": 0.1978, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 9.103568320278503, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 4.771086169268057e-06, | |
| "loss": 0.1992, | |
| "step": 5230 | |
| }, | |
| { | |
| "epoch": 9.112271540469974, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 4.678726916958998e-06, | |
| "loss": 0.1997, | |
| "step": 5235 | |
| }, | |
| { | |
| "epoch": 9.120974760661445, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 4.587248957912138e-06, | |
| "loss": 0.1998, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 9.129677980852916, | |
| "grad_norm": 0.66015625, | |
| "learning_rate": 4.496653137887386e-06, | |
| "loss": 0.1923, | |
| "step": 5245 | |
| }, | |
| { | |
| "epoch": 9.138381201044387, | |
| "grad_norm": 0.76171875, | |
| "learning_rate": 4.40694029448877e-06, | |
| "loss": 0.1998, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 9.147084421235856, | |
| "grad_norm": 0.6328125, | |
| "learning_rate": 4.318111257156831e-06, | |
| "loss": 0.1911, | |
| "step": 5255 | |
| }, | |
| { | |
| "epoch": 9.155787641427327, | |
| "grad_norm": 0.73828125, | |
| "learning_rate": 4.230166847160799e-06, | |
| "loss": 0.1949, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 9.164490861618798, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 4.143107877591135e-06, | |
| "loss": 0.1974, | |
| "step": 5265 | |
| }, | |
| { | |
| "epoch": 9.17319408181027, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 4.056935153351937e-06, | |
| "loss": 0.1964, | |
| "step": 5270 | |
| }, | |
| { | |
| "epoch": 9.18189730200174, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 3.971649471153516e-06, | |
| "loss": 0.1956, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 9.190600522193211, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 3.887251619505028e-06, | |
| "loss": 0.1969, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 9.199303742384682, | |
| "grad_norm": 0.65234375, | |
| "learning_rate": 3.803742378707198e-06, | |
| "loss": 0.1992, | |
| "step": 5285 | |
| }, | |
| { | |
| "epoch": 9.208006962576153, | |
| "grad_norm": 0.64453125, | |
| "learning_rate": 3.7211225208450774e-06, | |
| "loss": 0.1945, | |
| "step": 5290 | |
| }, | |
| { | |
| "epoch": 9.216710182767624, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 3.6393928097809617e-06, | |
| "loss": 0.199, | |
| "step": 5295 | |
| }, | |
| { | |
| "epoch": 9.225413402959095, | |
| "grad_norm": 0.65625, | |
| "learning_rate": 3.5585540011472516e-06, | |
| "loss": 0.1956, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 9.234116623150566, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 3.4786068423395044e-06, | |
| "loss": 0.1991, | |
| "step": 5305 | |
| }, | |
| { | |
| "epoch": 9.242819843342037, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 3.3995520725095486e-06, | |
| "loss": 0.1943, | |
| "step": 5310 | |
| }, | |
| { | |
| "epoch": 9.251523063533508, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 3.3213904225586346e-06, | |
| "loss": 0.1973, | |
| "step": 5315 | |
| }, | |
| { | |
| "epoch": 9.260226283724979, | |
| "grad_norm": 0.65234375, | |
| "learning_rate": 3.2441226151306404e-06, | |
| "loss": 0.1907, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 9.26892950391645, | |
| "grad_norm": 0.66015625, | |
| "learning_rate": 3.16774936460541e-06, | |
| "loss": 0.1968, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 9.27763272410792, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 3.092271377092215e-06, | |
| "loss": 0.1968, | |
| "step": 5330 | |
| }, | |
| { | |
| "epoch": 9.28633594429939, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 3.0176893504230807e-06, | |
| "loss": 0.1955, | |
| "step": 5335 | |
| }, | |
| { | |
| "epoch": 9.295039164490861, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 2.944003974146525e-06, | |
| "loss": 0.1939, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 9.303742384682332, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 2.8712159295209873e-06, | |
| "loss": 0.1955, | |
| "step": 5345 | |
| }, | |
| { | |
| "epoch": 9.312445604873803, | |
| "grad_norm": 0.65234375, | |
| "learning_rate": 2.7993258895086973e-06, | |
| "loss": 0.1925, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 9.321148825065274, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 2.7283345187693264e-06, | |
| "loss": 0.196, | |
| "step": 5355 | |
| }, | |
| { | |
| "epoch": 9.329852045256745, | |
| "grad_norm": 0.63671875, | |
| "learning_rate": 2.658242473653905e-06, | |
| "loss": 0.1929, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 9.338555265448216, | |
| "grad_norm": 0.65625, | |
| "learning_rate": 2.589050402198767e-06, | |
| "loss": 0.1958, | |
| "step": 5365 | |
| }, | |
| { | |
| "epoch": 9.347258485639687, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 2.520758944119539e-06, | |
| "loss": 0.1939, | |
| "step": 5370 | |
| }, | |
| { | |
| "epoch": 9.355961705831158, | |
| "grad_norm": 0.6640625, | |
| "learning_rate": 2.4533687308051835e-06, | |
| "loss": 0.1917, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 9.364664926022629, | |
| "grad_norm": 0.6953125, | |
| "learning_rate": 2.386880385312218e-06, | |
| "loss": 0.1937, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 9.3733681462141, | |
| "grad_norm": 0.6484375, | |
| "learning_rate": 2.321294522358952e-06, | |
| "loss": 0.1988, | |
| "step": 5385 | |
| }, | |
| { | |
| "epoch": 9.38207136640557, | |
| "grad_norm": 0.65625, | |
| "learning_rate": 2.256611748319792e-06, | |
| "loss": 0.1943, | |
| "step": 5390 | |
| }, | |
| { | |
| "epoch": 9.390774586597042, | |
| "grad_norm": 0.65625, | |
| "learning_rate": 2.1928326612196015e-06, | |
| "loss": 0.1964, | |
| "step": 5395 | |
| }, | |
| { | |
| "epoch": 9.399477806788513, | |
| "grad_norm": 0.640625, | |
| "learning_rate": 2.1299578507282147e-06, | |
| "loss": 0.196, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 9.408181026979982, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 2.0679878981549993e-06, | |
| "loss": 0.1921, | |
| "step": 5405 | |
| }, | |
| { | |
| "epoch": 9.416884247171453, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 2.006923376443415e-06, | |
| "loss": 0.1983, | |
| "step": 5410 | |
| }, | |
| { | |
| "epoch": 9.425587467362924, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 1.946764850165772e-06, | |
| "loss": 0.1984, | |
| "step": 5415 | |
| }, | |
| { | |
| "epoch": 9.434290687554395, | |
| "grad_norm": 0.6640625, | |
| "learning_rate": 1.8875128755179938e-06, | |
| "loss": 0.198, | |
| "step": 5420 | |
| }, | |
| { | |
| "epoch": 9.442993907745866, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 1.8291680003145073e-06, | |
| "loss": 0.1977, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 9.451697127937337, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 1.7717307639831037e-06, | |
| "loss": 0.1966, | |
| "step": 5430 | |
| }, | |
| { | |
| "epoch": 9.460400348128807, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 1.7152016975599983e-06, | |
| "loss": 0.1959, | |
| "step": 5435 | |
| }, | |
| { | |
| "epoch": 9.469103568320278, | |
| "grad_norm": 0.68359375, | |
| "learning_rate": 1.6595813236849556e-06, | |
| "loss": 0.1946, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 9.47780678851175, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 1.604870156596383e-06, | |
| "loss": 0.194, | |
| "step": 5445 | |
| }, | |
| { | |
| "epoch": 9.48651000870322, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 1.5510687021266234e-06, | |
| "loss": 0.1926, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 9.495213228894691, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 1.4981774576972584e-06, | |
| "loss": 0.1963, | |
| "step": 5455 | |
| }, | |
| { | |
| "epoch": 9.503916449086162, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 1.4461969123145457e-06, | |
| "loss": 0.1973, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 9.512619669277633, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 1.395127546564845e-06, | |
| "loss": 0.1963, | |
| "step": 5465 | |
| }, | |
| { | |
| "epoch": 9.521322889469104, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 1.344969832610199e-06, | |
| "loss": 0.1932, | |
| "step": 5470 | |
| }, | |
| { | |
| "epoch": 9.530026109660575, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 1.2957242341839927e-06, | |
| "loss": 0.197, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 9.538729329852046, | |
| "grad_norm": 0.828125, | |
| "learning_rate": 1.2473912065866345e-06, | |
| "loss": 0.1921, | |
| "step": 5480 | |
| }, | |
| { | |
| "epoch": 9.547432550043515, | |
| "grad_norm": 0.65234375, | |
| "learning_rate": 1.1999711966813377e-06, | |
| "loss": 0.1969, | |
| "step": 5485 | |
| }, | |
| { | |
| "epoch": 9.556135770234986, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 1.1534646428900232e-06, | |
| "loss": 0.1981, | |
| "step": 5490 | |
| }, | |
| { | |
| "epoch": 9.564838990426457, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 1.107871975189234e-06, | |
| "loss": 0.2015, | |
| "step": 5495 | |
| }, | |
| { | |
| "epoch": 9.573542210617928, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 1.0631936151062172e-06, | |
| "loss": 0.1953, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 9.5822454308094, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 1.019429975714914e-06, | |
| "loss": 0.1969, | |
| "step": 5505 | |
| }, | |
| { | |
| "epoch": 9.59094865100087, | |
| "grad_norm": 0.65234375, | |
| "learning_rate": 9.765814616322755e-07, | |
| "loss": 0.1956, | |
| "step": 5510 | |
| }, | |
| { | |
| "epoch": 9.599651871192341, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 9.346484690144319e-07, | |
| "loss": 0.1987, | |
| "step": 5515 | |
| }, | |
| { | |
| "epoch": 9.608355091383812, | |
| "grad_norm": 0.65234375, | |
| "learning_rate": 8.936313855530398e-07, | |
| "loss": 0.1944, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 9.617058311575283, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 8.535305904717517e-07, | |
| "loss": 0.1932, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 9.625761531766754, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 8.143464545226298e-07, | |
| "loss": 0.196, | |
| "step": 5530 | |
| }, | |
| { | |
| "epoch": 9.634464751958225, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 7.760793399827937e-07, | |
| "loss": 0.1967, | |
| "step": 5535 | |
| }, | |
| { | |
| "epoch": 9.643167972149696, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 7.387296006510225e-07, | |
| "loss": 0.1958, | |
| "step": 5540 | |
| }, | |
| { | |
| "epoch": 9.651871192341167, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 7.022975818445022e-07, | |
| "loss": 0.1933, | |
| "step": 5545 | |
| }, | |
| { | |
| "epoch": 9.660574412532638, | |
| "grad_norm": 0.73046875, | |
| "learning_rate": 6.667836203956168e-07, | |
| "loss": 0.1972, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 9.669277632724107, | |
| "grad_norm": 0.65234375, | |
| "learning_rate": 6.321880446488737e-07, | |
| "loss": 0.1932, | |
| "step": 5555 | |
| }, | |
| { | |
| "epoch": 9.677980852915578, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 5.985111744578165e-07, | |
| "loss": 0.1977, | |
| "step": 5560 | |
| }, | |
| { | |
| "epoch": 9.686684073107049, | |
| "grad_norm": 0.67578125, | |
| "learning_rate": 5.657533211820942e-07, | |
| "loss": 0.1979, | |
| "step": 5565 | |
| }, | |
| { | |
| "epoch": 9.69538729329852, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 5.339147876845974e-07, | |
| "loss": 0.1961, | |
| "step": 5570 | |
| }, | |
| { | |
| "epoch": 9.70409051348999, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 5.029958683286263e-07, | |
| "loss": 0.197, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 9.712793733681462, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 4.7299684897520456e-07, | |
| "loss": 0.193, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 9.721496953872933, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 4.4391800698038165e-07, | |
| "loss": 0.1961, | |
| "step": 5585 | |
| }, | |
| { | |
| "epoch": 9.730200174064404, | |
| "grad_norm": 0.63671875, | |
| "learning_rate": 4.157596111927342e-07, | |
| "loss": 0.1903, | |
| "step": 5590 | |
| }, | |
| { | |
| "epoch": 9.738903394255875, | |
| "grad_norm": 0.71875, | |
| "learning_rate": 3.8852192195083516e-07, | |
| "loss": 0.1948, | |
| "step": 5595 | |
| }, | |
| { | |
| "epoch": 9.747606614447346, | |
| "grad_norm": 0.70703125, | |
| "learning_rate": 3.622051910808666e-07, | |
| "loss": 0.1969, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 9.756309834638817, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 3.368096618942773e-07, | |
| "loss": 0.1948, | |
| "step": 5605 | |
| }, | |
| { | |
| "epoch": 9.765013054830288, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 3.1233556918555117e-07, | |
| "loss": 0.1982, | |
| "step": 5610 | |
| }, | |
| { | |
| "epoch": 9.773716275021759, | |
| "grad_norm": 0.625, | |
| "learning_rate": 2.8878313923002e-07, | |
| "loss": 0.1929, | |
| "step": 5615 | |
| }, | |
| { | |
| "epoch": 9.78241949521323, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 2.661525897817874e-07, | |
| "loss": 0.1987, | |
| "step": 5620 | |
| }, | |
| { | |
| "epoch": 9.7911227154047, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 2.444441300717082e-07, | |
| "loss": 0.1953, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 9.799825935596171, | |
| "grad_norm": 0.6875, | |
| "learning_rate": 2.2365796080542345e-07, | |
| "loss": 0.2007, | |
| "step": 5630 | |
| }, | |
| { | |
| "epoch": 9.80852915578764, | |
| "grad_norm": 0.85546875, | |
| "learning_rate": 2.037942741615617e-07, | |
| "loss": 0.2001, | |
| "step": 5635 | |
| }, | |
| { | |
| "epoch": 9.817232375979112, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 1.8485325378994056e-07, | |
| "loss": 0.198, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 9.825935596170583, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 1.6683507480983462e-07, | |
| "loss": 0.1958, | |
| "step": 5645 | |
| }, | |
| { | |
| "epoch": 9.834638816362054, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 1.4973990380841019e-07, | |
| "loss": 0.1938, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 9.843342036553524, | |
| "grad_norm": 0.7109375, | |
| "learning_rate": 1.3356789883914865e-07, | |
| "loss": 0.1938, | |
| "step": 5655 | |
| }, | |
| { | |
| "epoch": 9.852045256744995, | |
| "grad_norm": 0.703125, | |
| "learning_rate": 1.1831920942039221e-07, | |
| "loss": 0.1973, | |
| "step": 5660 | |
| }, | |
| { | |
| "epoch": 9.860748476936466, | |
| "grad_norm": 0.69140625, | |
| "learning_rate": 1.0399397653395593e-07, | |
| "loss": 0.2024, | |
| "step": 5665 | |
| }, | |
| { | |
| "epoch": 9.869451697127937, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 9.059233262386225e-08, | |
| "loss": 0.1995, | |
| "step": 5670 | |
| }, | |
| { | |
| "epoch": 9.878154917319408, | |
| "grad_norm": 0.71484375, | |
| "learning_rate": 7.811440159507522e-08, | |
| "loss": 0.1972, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 9.88685813751088, | |
| "grad_norm": 0.65625, | |
| "learning_rate": 6.656029881233483e-08, | |
| "loss": 0.1938, | |
| "step": 5680 | |
| }, | |
| { | |
| "epoch": 9.89556135770235, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 5.593013109917999e-08, | |
| "loss": 0.1974, | |
| "step": 5685 | |
| }, | |
| { | |
| "epoch": 9.904264577893821, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 4.6223996736860506e-08, | |
| "loss": 0.1957, | |
| "step": 5690 | |
| }, | |
| { | |
| "epoch": 9.912967798085292, | |
| "grad_norm": 0.65625, | |
| "learning_rate": 3.744198546348221e-08, | |
| "loss": 0.1971, | |
| "step": 5695 | |
| }, | |
| { | |
| "epoch": 9.921671018276763, | |
| "grad_norm": 0.66796875, | |
| "learning_rate": 2.9584178473174296e-08, | |
| "loss": 0.1977, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 9.930374238468232, | |
| "grad_norm": 0.7421875, | |
| "learning_rate": 2.2650648415334376e-08, | |
| "loss": 0.1934, | |
| "step": 5705 | |
| }, | |
| { | |
| "epoch": 9.939077458659703, | |
| "grad_norm": 0.69921875, | |
| "learning_rate": 1.664145939394013e-08, | |
| "loss": 0.1937, | |
| "step": 5710 | |
| }, | |
| { | |
| "epoch": 9.947780678851174, | |
| "grad_norm": 0.6796875, | |
| "learning_rate": 1.1556666966971997e-08, | |
| "loss": 0.1996, | |
| "step": 5715 | |
| }, | |
| { | |
| "epoch": 9.956483899042645, | |
| "grad_norm": 0.7265625, | |
| "learning_rate": 7.39631814590247e-09, | |
| "loss": 0.1984, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 9.965187119234116, | |
| "grad_norm": 0.66015625, | |
| "learning_rate": 4.160451395263109e-09, | |
| "loss": 0.1927, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 9.973890339425587, | |
| "grad_norm": 0.734375, | |
| "learning_rate": 1.8490966322670666e-09, | |
| "loss": 0.2, | |
| "step": 5730 | |
| }, | |
| { | |
| "epoch": 9.982593559617058, | |
| "grad_norm": 0.75390625, | |
| "learning_rate": 4.6227522655373223e-10, | |
| "loss": 0.1946, | |
| "step": 5735 | |
| }, | |
| { | |
| "epoch": 9.991296779808529, | |
| "grad_norm": 0.671875, | |
| "learning_rate": 0.0, | |
| "loss": 0.1899, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 9.991296779808529, | |
| "eval_loss": 5.076513767242432, | |
| "eval_runtime": 0.7783, | |
| "eval_samples_per_second": 7.709, | |
| "eval_steps_per_second": 1.285, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 9.991296779808529, | |
| "step": 5740, | |
| "total_flos": 6.645284010274587e+18, | |
| "train_loss": 0.6571785643956387, | |
| "train_runtime": 32584.0572, | |
| "train_samples_per_second": 4.229, | |
| "train_steps_per_second": 0.176 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 5740, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 6.645284010274587e+18, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |