|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 1.0, |
|
"eval_steps": 500, |
|
"global_step": 1624, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0006157635467980296, |
|
"grad_norm": 6.913773939975575, |
|
"learning_rate": 1.8404907975460124e-08, |
|
"loss": 0.9784, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.003078817733990148, |
|
"grad_norm": 7.825438885604264, |
|
"learning_rate": 9.202453987730061e-08, |
|
"loss": 1.0022, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.006157635467980296, |
|
"grad_norm": 7.160715717206394, |
|
"learning_rate": 1.8404907975460122e-07, |
|
"loss": 1.0144, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.009236453201970444, |
|
"grad_norm": 5.381505135217186, |
|
"learning_rate": 2.7607361963190183e-07, |
|
"loss": 0.9952, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.012315270935960592, |
|
"grad_norm": 2.7183274818278274, |
|
"learning_rate": 3.6809815950920245e-07, |
|
"loss": 0.9764, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.01539408866995074, |
|
"grad_norm": 4.791530965270314, |
|
"learning_rate": 4.601226993865031e-07, |
|
"loss": 0.9583, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.01847290640394089, |
|
"grad_norm": 1.9291502548810062, |
|
"learning_rate": 5.521472392638037e-07, |
|
"loss": 0.934, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.021551724137931036, |
|
"grad_norm": 2.549720558736737, |
|
"learning_rate": 6.441717791411044e-07, |
|
"loss": 0.9266, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.024630541871921183, |
|
"grad_norm": 2.2777145227057995, |
|
"learning_rate": 7.361963190184049e-07, |
|
"loss": 0.9221, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.02770935960591133, |
|
"grad_norm": 1.65348797355226, |
|
"learning_rate": 8.282208588957056e-07, |
|
"loss": 0.9002, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.03078817733990148, |
|
"grad_norm": 1.4255646800205126, |
|
"learning_rate": 9.202453987730062e-07, |
|
"loss": 0.9173, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.033866995073891626, |
|
"grad_norm": 1.3127964948073867, |
|
"learning_rate": 1.0122699386503068e-06, |
|
"loss": 0.8722, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.03694581280788178, |
|
"grad_norm": 1.3676057719432315, |
|
"learning_rate": 1.1042944785276073e-06, |
|
"loss": 0.8642, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.04002463054187192, |
|
"grad_norm": 1.384144431804707, |
|
"learning_rate": 1.1963190184049078e-06, |
|
"loss": 0.8974, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.04310344827586207, |
|
"grad_norm": 1.3470465492733947, |
|
"learning_rate": 1.2883435582822088e-06, |
|
"loss": 0.882, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.046182266009852216, |
|
"grad_norm": 1.3217192721217494, |
|
"learning_rate": 1.3803680981595093e-06, |
|
"loss": 0.8941, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.04926108374384237, |
|
"grad_norm": 1.318630391147535, |
|
"learning_rate": 1.4723926380368098e-06, |
|
"loss": 0.872, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.05233990147783251, |
|
"grad_norm": 1.4012119971492354, |
|
"learning_rate": 1.5644171779141105e-06, |
|
"loss": 0.8807, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.05541871921182266, |
|
"grad_norm": 1.2884444463046538, |
|
"learning_rate": 1.6564417177914112e-06, |
|
"loss": 0.8643, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.058497536945812806, |
|
"grad_norm": 1.3737928683024987, |
|
"learning_rate": 1.7484662576687115e-06, |
|
"loss": 0.8771, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.06157635467980296, |
|
"grad_norm": 1.247376359958333, |
|
"learning_rate": 1.8404907975460124e-06, |
|
"loss": 0.8879, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.06465517241379311, |
|
"grad_norm": 1.39726152323821, |
|
"learning_rate": 1.932515337423313e-06, |
|
"loss": 0.8778, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.06773399014778325, |
|
"grad_norm": 1.2801758571019077, |
|
"learning_rate": 2.0245398773006137e-06, |
|
"loss": 0.8948, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.0708128078817734, |
|
"grad_norm": 1.321445350144951, |
|
"learning_rate": 2.1165644171779144e-06, |
|
"loss": 0.8757, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.07389162561576355, |
|
"grad_norm": 1.3605189305382268, |
|
"learning_rate": 2.2085889570552147e-06, |
|
"loss": 0.8726, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.0769704433497537, |
|
"grad_norm": 1.3403550813907201, |
|
"learning_rate": 2.3006134969325154e-06, |
|
"loss": 0.8556, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.08004926108374384, |
|
"grad_norm": 1.5765670477069915, |
|
"learning_rate": 2.3926380368098157e-06, |
|
"loss": 0.8632, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.08312807881773399, |
|
"grad_norm": 1.53171076235737, |
|
"learning_rate": 2.4846625766871164e-06, |
|
"loss": 0.885, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.08620689655172414, |
|
"grad_norm": 1.3517727998137348, |
|
"learning_rate": 2.5766871165644175e-06, |
|
"loss": 0.8896, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.08928571428571429, |
|
"grad_norm": 1.3321264090180198, |
|
"learning_rate": 2.668711656441718e-06, |
|
"loss": 0.8679, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.09236453201970443, |
|
"grad_norm": 1.6364655921115436, |
|
"learning_rate": 2.7607361963190186e-06, |
|
"loss": 0.8683, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.09544334975369458, |
|
"grad_norm": 1.3954805337443055, |
|
"learning_rate": 2.852760736196319e-06, |
|
"loss": 0.8917, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.09852216748768473, |
|
"grad_norm": 1.2554504879922763, |
|
"learning_rate": 2.9447852760736196e-06, |
|
"loss": 0.8641, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.10160098522167488, |
|
"grad_norm": 1.3656321404184575, |
|
"learning_rate": 2.999986128612664e-06, |
|
"loss": 0.8791, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.10467980295566502, |
|
"grad_norm": 1.3569360164716509, |
|
"learning_rate": 2.9998300784514776e-06, |
|
"loss": 0.8674, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.10775862068965517, |
|
"grad_norm": 1.3318817855448093, |
|
"learning_rate": 2.99950065699352e-06, |
|
"loss": 0.8936, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.11083743842364532, |
|
"grad_norm": 1.1968758419665766, |
|
"learning_rate": 2.9989979023179235e-06, |
|
"loss": 0.8672, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.11391625615763547, |
|
"grad_norm": 1.2612414868299464, |
|
"learning_rate": 2.9983218725400893e-06, |
|
"loss": 0.863, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.11699507389162561, |
|
"grad_norm": 1.2725738041996348, |
|
"learning_rate": 2.9974726458049776e-06, |
|
"loss": 0.8703, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.12007389162561577, |
|
"grad_norm": 1.2437787711981092, |
|
"learning_rate": 2.9964503202780676e-06, |
|
"loss": 0.8521, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.12315270935960591, |
|
"grad_norm": 1.240489837880236, |
|
"learning_rate": 2.9952550141340154e-06, |
|
"loss": 0.8838, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.12623152709359606, |
|
"grad_norm": 1.461158309047738, |
|
"learning_rate": 2.99388686554299e-06, |
|
"loss": 0.8878, |
|
"step": 205 |
|
}, |
|
{ |
|
"epoch": 0.12931034482758622, |
|
"grad_norm": 1.318848917819606, |
|
"learning_rate": 2.9923460326547038e-06, |
|
"loss": 0.8541, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.13238916256157635, |
|
"grad_norm": 1.3436997832844435, |
|
"learning_rate": 2.990632693580131e-06, |
|
"loss": 0.8717, |
|
"step": 215 |
|
}, |
|
{ |
|
"epoch": 0.1354679802955665, |
|
"grad_norm": 1.4597565690505312, |
|
"learning_rate": 2.988747046370918e-06, |
|
"loss": 0.8681, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.13854679802955666, |
|
"grad_norm": 1.3585856656212756, |
|
"learning_rate": 2.986689308996492e-06, |
|
"loss": 0.8851, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 0.1416256157635468, |
|
"grad_norm": 1.2397385011947064, |
|
"learning_rate": 2.984459719318862e-06, |
|
"loss": 0.865, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.14470443349753695, |
|
"grad_norm": 1.2509056258515312, |
|
"learning_rate": 2.9820585350651265e-06, |
|
"loss": 0.8585, |
|
"step": 235 |
|
}, |
|
{ |
|
"epoch": 0.1477832512315271, |
|
"grad_norm": 1.2285891792995178, |
|
"learning_rate": 2.9794860337976802e-06, |
|
"loss": 0.8669, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.15086206896551724, |
|
"grad_norm": 1.2936945932407728, |
|
"learning_rate": 2.9767425128821288e-06, |
|
"loss": 0.8582, |
|
"step": 245 |
|
}, |
|
{ |
|
"epoch": 0.1539408866995074, |
|
"grad_norm": 1.282555595687812, |
|
"learning_rate": 2.9738282894529177e-06, |
|
"loss": 0.8738, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.15701970443349753, |
|
"grad_norm": 1.2870084504395782, |
|
"learning_rate": 2.9707437003766704e-06, |
|
"loss": 0.8677, |
|
"step": 255 |
|
}, |
|
{ |
|
"epoch": 0.16009852216748768, |
|
"grad_norm": 1.353243025181188, |
|
"learning_rate": 2.96748910221325e-06, |
|
"loss": 0.878, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.16317733990147784, |
|
"grad_norm": 1.264413213172448, |
|
"learning_rate": 2.964064871174544e-06, |
|
"loss": 0.8678, |
|
"step": 265 |
|
}, |
|
{ |
|
"epoch": 0.16625615763546797, |
|
"grad_norm": 1.3269239709537661, |
|
"learning_rate": 2.9604714030809755e-06, |
|
"loss": 0.864, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.16933497536945813, |
|
"grad_norm": 1.2879236942347894, |
|
"learning_rate": 2.9567091133157484e-06, |
|
"loss": 0.8724, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 0.1724137931034483, |
|
"grad_norm": 1.2081051395475086, |
|
"learning_rate": 2.9527784367768305e-06, |
|
"loss": 0.8631, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.17549261083743842, |
|
"grad_norm": 1.2233596670003795, |
|
"learning_rate": 2.948679827826687e-06, |
|
"loss": 0.8773, |
|
"step": 285 |
|
}, |
|
{ |
|
"epoch": 0.17857142857142858, |
|
"grad_norm": 1.2682616646735592, |
|
"learning_rate": 2.9444137602397515e-06, |
|
"loss": 0.8876, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.1816502463054187, |
|
"grad_norm": 1.2169689320755697, |
|
"learning_rate": 2.9399807271476674e-06, |
|
"loss": 0.8654, |
|
"step": 295 |
|
}, |
|
{ |
|
"epoch": 0.18472906403940886, |
|
"grad_norm": 1.2598697861772865, |
|
"learning_rate": 2.935381240982281e-06, |
|
"loss": 0.8717, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.18780788177339902, |
|
"grad_norm": 1.254316740037268, |
|
"learning_rate": 2.9306158334164066e-06, |
|
"loss": 0.861, |
|
"step": 305 |
|
}, |
|
{ |
|
"epoch": 0.19088669950738915, |
|
"grad_norm": 1.2353415453137906, |
|
"learning_rate": 2.9256850553023724e-06, |
|
"loss": 0.8806, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.1939655172413793, |
|
"grad_norm": 1.1684667032801705, |
|
"learning_rate": 2.920589476608343e-06, |
|
"loss": 0.8385, |
|
"step": 315 |
|
}, |
|
{ |
|
"epoch": 0.19704433497536947, |
|
"grad_norm": 1.2113614584832126, |
|
"learning_rate": 2.9153296863524315e-06, |
|
"loss": 0.87, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.2001231527093596, |
|
"grad_norm": 1.350840273124045, |
|
"learning_rate": 2.90990629253462e-06, |
|
"loss": 0.87, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 0.20320197044334976, |
|
"grad_norm": 1.2887655848361137, |
|
"learning_rate": 2.9043199220664704e-06, |
|
"loss": 0.8576, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.2062807881773399, |
|
"grad_norm": 1.1361120979628074, |
|
"learning_rate": 2.8985712206986627e-06, |
|
"loss": 0.8713, |
|
"step": 335 |
|
}, |
|
{ |
|
"epoch": 0.20935960591133004, |
|
"grad_norm": 1.2566658610805421, |
|
"learning_rate": 2.8926608529463473e-06, |
|
"loss": 0.8658, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.2124384236453202, |
|
"grad_norm": 1.1938429193038493, |
|
"learning_rate": 2.8865895020123326e-06, |
|
"loss": 0.8682, |
|
"step": 345 |
|
}, |
|
{ |
|
"epoch": 0.21551724137931033, |
|
"grad_norm": 1.1927966596708883, |
|
"learning_rate": 2.880357869708111e-06, |
|
"loss": 0.8959, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.2185960591133005, |
|
"grad_norm": 1.1623715689493113, |
|
"learning_rate": 2.8739666763727315e-06, |
|
"loss": 0.8697, |
|
"step": 355 |
|
}, |
|
{ |
|
"epoch": 0.22167487684729065, |
|
"grad_norm": 1.1842338397052192, |
|
"learning_rate": 2.8674166607895357e-06, |
|
"loss": 0.8627, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.22475369458128078, |
|
"grad_norm": 1.3616371310779263, |
|
"learning_rate": 2.8607085801007565e-06, |
|
"loss": 0.8546, |
|
"step": 365 |
|
}, |
|
{ |
|
"epoch": 0.22783251231527094, |
|
"grad_norm": 1.311966905899673, |
|
"learning_rate": 2.85384320972e-06, |
|
"loss": 0.8576, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.2309113300492611, |
|
"grad_norm": 1.3414358462962774, |
|
"learning_rate": 2.846821343242608e-06, |
|
"loss": 0.8517, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 0.23399014778325122, |
|
"grad_norm": 1.2748762327701373, |
|
"learning_rate": 2.839643792353928e-06, |
|
"loss": 0.8764, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.23706896551724138, |
|
"grad_norm": 1.3230993855053796, |
|
"learning_rate": 2.832311386735483e-06, |
|
"loss": 0.8983, |
|
"step": 385 |
|
}, |
|
{ |
|
"epoch": 0.24014778325123154, |
|
"grad_norm": 1.2368249582586939, |
|
"learning_rate": 2.824824973969069e-06, |
|
"loss": 0.8652, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.24322660098522167, |
|
"grad_norm": 1.2631609761367932, |
|
"learning_rate": 2.817185419438777e-06, |
|
"loss": 0.8668, |
|
"step": 395 |
|
}, |
|
{ |
|
"epoch": 0.24630541871921183, |
|
"grad_norm": 1.1698659577223387, |
|
"learning_rate": 2.8093936062309614e-06, |
|
"loss": 0.85, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.24938423645320196, |
|
"grad_norm": 1.2098312196573524, |
|
"learning_rate": 2.8014504350321594e-06, |
|
"loss": 0.8819, |
|
"step": 405 |
|
}, |
|
{ |
|
"epoch": 0.2524630541871921, |
|
"grad_norm": 1.3035212462808752, |
|
"learning_rate": 2.7933568240249776e-06, |
|
"loss": 0.8568, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.2555418719211823, |
|
"grad_norm": 1.301195180537406, |
|
"learning_rate": 2.7851137087819562e-06, |
|
"loss": 0.8731, |
|
"step": 415 |
|
}, |
|
{ |
|
"epoch": 0.25862068965517243, |
|
"grad_norm": 1.2687993974830427, |
|
"learning_rate": 2.776722042157421e-06, |
|
"loss": 0.8712, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.2616995073891626, |
|
"grad_norm": 1.2584554723191728, |
|
"learning_rate": 2.768182794177341e-06, |
|
"loss": 0.8629, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 0.2647783251231527, |
|
"grad_norm": 1.2338984176969319, |
|
"learning_rate": 2.7594969519271988e-06, |
|
"loss": 0.8545, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.26785714285714285, |
|
"grad_norm": 1.2870854729570638, |
|
"learning_rate": 2.7506655194378874e-06, |
|
"loss": 0.87, |
|
"step": 435 |
|
}, |
|
{ |
|
"epoch": 0.270935960591133, |
|
"grad_norm": 1.2151872850885181, |
|
"learning_rate": 2.7416895175696533e-06, |
|
"loss": 0.8788, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.27401477832512317, |
|
"grad_norm": 1.193049984158674, |
|
"learning_rate": 2.7325699838940894e-06, |
|
"loss": 0.8808, |
|
"step": 445 |
|
}, |
|
{ |
|
"epoch": 0.2770935960591133, |
|
"grad_norm": 2.1512581943290816, |
|
"learning_rate": 2.723307972574199e-06, |
|
"loss": 0.8679, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.2801724137931034, |
|
"grad_norm": 1.2246096729147533, |
|
"learning_rate": 2.713904554242539e-06, |
|
"loss": 0.8661, |
|
"step": 455 |
|
}, |
|
{ |
|
"epoch": 0.2832512315270936, |
|
"grad_norm": 1.2118181362190374, |
|
"learning_rate": 2.7043608158774645e-06, |
|
"loss": 0.8598, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.28633004926108374, |
|
"grad_norm": 1.2327629256736043, |
|
"learning_rate": 2.6946778606774777e-06, |
|
"loss": 0.8589, |
|
"step": 465 |
|
}, |
|
{ |
|
"epoch": 0.2894088669950739, |
|
"grad_norm": 1.2373102093793737, |
|
"learning_rate": 2.684856807933706e-06, |
|
"loss": 0.8676, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 0.29248768472906406, |
|
"grad_norm": 1.4752818322798746, |
|
"learning_rate": 2.6748987929005187e-06, |
|
"loss": 0.8549, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 0.2955665024630542, |
|
"grad_norm": 1.2262642368471763, |
|
"learning_rate": 2.664804966664298e-06, |
|
"loss": 0.8873, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 0.2986453201970443, |
|
"grad_norm": 1.3999005157781816, |
|
"learning_rate": 2.6545764960103823e-06, |
|
"loss": 0.8699, |
|
"step": 485 |
|
}, |
|
{ |
|
"epoch": 0.3017241379310345, |
|
"grad_norm": 1.2074290958945613, |
|
"learning_rate": 2.6442145632881894e-06, |
|
"loss": 0.8686, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 0.30480295566502463, |
|
"grad_norm": 1.165013891083953, |
|
"learning_rate": 2.6337203662745465e-06, |
|
"loss": 0.8779, |
|
"step": 495 |
|
}, |
|
{ |
|
"epoch": 0.3078817733990148, |
|
"grad_norm": 1.215482555799305, |
|
"learning_rate": 2.623095118035235e-06, |
|
"loss": 0.8741, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.31096059113300495, |
|
"grad_norm": 1.2044286444458985, |
|
"learning_rate": 2.612340046784765e-06, |
|
"loss": 0.8784, |
|
"step": 505 |
|
}, |
|
{ |
|
"epoch": 0.31403940886699505, |
|
"grad_norm": 1.2266878125080043, |
|
"learning_rate": 2.601456395744403e-06, |
|
"loss": 0.8695, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 0.3171182266009852, |
|
"grad_norm": 1.2672115735011746, |
|
"learning_rate": 2.590445422998462e-06, |
|
"loss": 0.8885, |
|
"step": 515 |
|
}, |
|
{ |
|
"epoch": 0.32019704433497537, |
|
"grad_norm": 1.2135704940215861, |
|
"learning_rate": 2.579308401348876e-06, |
|
"loss": 0.8666, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 0.3232758620689655, |
|
"grad_norm": 1.2352647052778694, |
|
"learning_rate": 2.56804661816807e-06, |
|
"loss": 0.8569, |
|
"step": 525 |
|
}, |
|
{ |
|
"epoch": 0.3263546798029557, |
|
"grad_norm": 1.1481923904639533, |
|
"learning_rate": 2.556661375250149e-06, |
|
"loss": 0.8694, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 0.3294334975369458, |
|
"grad_norm": 1.1021273309473116, |
|
"learning_rate": 2.5451539886604165e-06, |
|
"loss": 0.8548, |
|
"step": 535 |
|
}, |
|
{ |
|
"epoch": 0.33251231527093594, |
|
"grad_norm": 1.2200902608890873, |
|
"learning_rate": 2.533525788583248e-06, |
|
"loss": 0.868, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 0.3355911330049261, |
|
"grad_norm": 1.183008490401881, |
|
"learning_rate": 2.5217781191683284e-06, |
|
"loss": 0.8559, |
|
"step": 545 |
|
}, |
|
{ |
|
"epoch": 0.33866995073891626, |
|
"grad_norm": 1.2172301950707558, |
|
"learning_rate": 2.509912338375275e-06, |
|
"loss": 0.8491, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.3417487684729064, |
|
"grad_norm": 1.1858763526432543, |
|
"learning_rate": 2.497929817816667e-06, |
|
"loss": 0.8727, |
|
"step": 555 |
|
}, |
|
{ |
|
"epoch": 0.3448275862068966, |
|
"grad_norm": 1.182553659695508, |
|
"learning_rate": 2.4858319425994978e-06, |
|
"loss": 0.8728, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 0.3479064039408867, |
|
"grad_norm": 1.212226883648794, |
|
"learning_rate": 2.4736201111650593e-06, |
|
"loss": 0.8749, |
|
"step": 565 |
|
}, |
|
{ |
|
"epoch": 0.35098522167487683, |
|
"grad_norm": 1.2065381166659566, |
|
"learning_rate": 2.4612957351272963e-06, |
|
"loss": 0.8642, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 0.354064039408867, |
|
"grad_norm": 1.2056593153630641, |
|
"learning_rate": 2.448860239109627e-06, |
|
"loss": 0.8602, |
|
"step": 575 |
|
}, |
|
{ |
|
"epoch": 0.35714285714285715, |
|
"grad_norm": 1.230227236910998, |
|
"learning_rate": 2.4363150605802704e-06, |
|
"loss": 0.8302, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 0.3602216748768473, |
|
"grad_norm": 1.1850041621929952, |
|
"learning_rate": 2.423661649686081e-06, |
|
"loss": 0.8736, |
|
"step": 585 |
|
}, |
|
{ |
|
"epoch": 0.3633004926108374, |
|
"grad_norm": 1.1365699274676013, |
|
"learning_rate": 2.41090146908492e-06, |
|
"loss": 0.8376, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 0.36637931034482757, |
|
"grad_norm": 1.158850798276412, |
|
"learning_rate": 2.398035993776582e-06, |
|
"loss": 0.8394, |
|
"step": 595 |
|
}, |
|
{ |
|
"epoch": 0.3694581280788177, |
|
"grad_norm": 1.1589737316795938, |
|
"learning_rate": 2.385066710932294e-06, |
|
"loss": 0.8571, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.3725369458128079, |
|
"grad_norm": 1.2404593093066263, |
|
"learning_rate": 2.3719951197228068e-06, |
|
"loss": 0.8422, |
|
"step": 605 |
|
}, |
|
{ |
|
"epoch": 0.37561576354679804, |
|
"grad_norm": 1.172803846391092, |
|
"learning_rate": 2.3588227311451007e-06, |
|
"loss": 0.8585, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 0.3786945812807882, |
|
"grad_norm": 1.2257863189413905, |
|
"learning_rate": 2.3455510678477217e-06, |
|
"loss": 0.8523, |
|
"step": 615 |
|
}, |
|
{ |
|
"epoch": 0.3817733990147783, |
|
"grad_norm": 1.1912296312950699, |
|
"learning_rate": 2.3321816639547747e-06, |
|
"loss": 0.8555, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 0.38485221674876846, |
|
"grad_norm": 1.1578973856425505, |
|
"learning_rate": 2.3187160648885864e-06, |
|
"loss": 0.8642, |
|
"step": 625 |
|
}, |
|
{ |
|
"epoch": 0.3879310344827586, |
|
"grad_norm": 1.2075267871283375, |
|
"learning_rate": 2.305155827191066e-06, |
|
"loss": 0.8592, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 0.3910098522167488, |
|
"grad_norm": 1.1892148288294622, |
|
"learning_rate": 2.291502518343774e-06, |
|
"loss": 0.8812, |
|
"step": 635 |
|
}, |
|
{ |
|
"epoch": 0.39408866995073893, |
|
"grad_norm": 1.2187696115868283, |
|
"learning_rate": 2.2777577165867354e-06, |
|
"loss": 0.8635, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 0.39716748768472904, |
|
"grad_norm": 1.2439181051613986, |
|
"learning_rate": 2.2639230107360033e-06, |
|
"loss": 0.8547, |
|
"step": 645 |
|
}, |
|
{ |
|
"epoch": 0.4002463054187192, |
|
"grad_norm": 1.2243209144902911, |
|
"learning_rate": 2.25e-06, |
|
"loss": 0.8468, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.40332512315270935, |
|
"grad_norm": 1.2511378722439845, |
|
"learning_rate": 2.235990293794659e-06, |
|
"loss": 0.8789, |
|
"step": 655 |
|
}, |
|
{ |
|
"epoch": 0.4064039408866995, |
|
"grad_norm": 1.1898754773741533, |
|
"learning_rate": 2.2218955115573864e-06, |
|
"loss": 0.8674, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 0.40948275862068967, |
|
"grad_norm": 1.2683544814556624, |
|
"learning_rate": 2.2077172825598645e-06, |
|
"loss": 0.8544, |
|
"step": 665 |
|
}, |
|
{ |
|
"epoch": 0.4125615763546798, |
|
"grad_norm": 1.1565601629897568, |
|
"learning_rate": 2.1934572457197163e-06, |
|
"loss": 0.8673, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 0.41564039408866993, |
|
"grad_norm": 1.4995919583590112, |
|
"learning_rate": 2.179117049411057e-06, |
|
"loss": 0.8632, |
|
"step": 675 |
|
}, |
|
{ |
|
"epoch": 0.4187192118226601, |
|
"grad_norm": 1.188260702571106, |
|
"learning_rate": 2.164698351273952e-06, |
|
"loss": 0.8607, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 0.42179802955665024, |
|
"grad_norm": 1.1404383396537943, |
|
"learning_rate": 2.150202818022805e-06, |
|
"loss": 0.862, |
|
"step": 685 |
|
}, |
|
{ |
|
"epoch": 0.4248768472906404, |
|
"grad_norm": 1.175977890822118, |
|
"learning_rate": 2.1356321252536947e-06, |
|
"loss": 0.8491, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 0.42795566502463056, |
|
"grad_norm": 1.1989659270703132, |
|
"learning_rate": 2.1209879572506836e-06, |
|
"loss": 0.8612, |
|
"step": 695 |
|
}, |
|
{ |
|
"epoch": 0.43103448275862066, |
|
"grad_norm": 1.245108344821411, |
|
"learning_rate": 2.10627200679113e-06, |
|
"loss": 0.8595, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.4341133004926108, |
|
"grad_norm": 1.2211796526267087, |
|
"learning_rate": 2.09148597495001e-06, |
|
"loss": 0.8757, |
|
"step": 705 |
|
}, |
|
{ |
|
"epoch": 0.437192118226601, |
|
"grad_norm": 1.265791526826937, |
|
"learning_rate": 2.0766315709032837e-06, |
|
"loss": 0.8403, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 0.44027093596059114, |
|
"grad_norm": 1.1984341397843337, |
|
"learning_rate": 2.0617105117303247e-06, |
|
"loss": 0.8569, |
|
"step": 715 |
|
}, |
|
{ |
|
"epoch": 0.4433497536945813, |
|
"grad_norm": 1.1497081905260205, |
|
"learning_rate": 2.046724522215437e-06, |
|
"loss": 0.8438, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 0.44642857142857145, |
|
"grad_norm": 1.2147041590973553, |
|
"learning_rate": 2.03167533464848e-06, |
|
"loss": 0.8634, |
|
"step": 725 |
|
}, |
|
{ |
|
"epoch": 0.44950738916256155, |
|
"grad_norm": 1.1188033176522687, |
|
"learning_rate": 2.016564688624627e-06, |
|
"loss": 0.8676, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 0.4525862068965517, |
|
"grad_norm": 1.1905270406995492, |
|
"learning_rate": 2.001394330843276e-06, |
|
"loss": 0.8372, |
|
"step": 735 |
|
}, |
|
{ |
|
"epoch": 0.45566502463054187, |
|
"grad_norm": 1.176752475031966, |
|
"learning_rate": 1.9861660149061435e-06, |
|
"loss": 0.8698, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 0.45874384236453203, |
|
"grad_norm": 1.216007273644598, |
|
"learning_rate": 1.9708815011145605e-06, |
|
"loss": 0.8607, |
|
"step": 745 |
|
}, |
|
{ |
|
"epoch": 0.4618226600985222, |
|
"grad_norm": 1.2084465000727591, |
|
"learning_rate": 1.9555425562659878e-06, |
|
"loss": 0.8633, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 0.4649014778325123, |
|
"grad_norm": 1.1831446750097752, |
|
"learning_rate": 1.940150953449791e-06, |
|
"loss": 0.8272, |
|
"step": 755 |
|
}, |
|
{ |
|
"epoch": 0.46798029556650245, |
|
"grad_norm": 1.1461713685681005, |
|
"learning_rate": 1.924708471842276e-06, |
|
"loss": 0.8567, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 0.4710591133004926, |
|
"grad_norm": 1.1398047886843539, |
|
"learning_rate": 1.90921689650103e-06, |
|
"loss": 0.8586, |
|
"step": 765 |
|
}, |
|
{ |
|
"epoch": 0.47413793103448276, |
|
"grad_norm": 1.182661874060586, |
|
"learning_rate": 1.8936780181585799e-06, |
|
"loss": 0.848, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 0.4772167487684729, |
|
"grad_norm": 1.1600436435911352, |
|
"learning_rate": 1.8780936330153935e-06, |
|
"loss": 0.8683, |
|
"step": 775 |
|
}, |
|
{ |
|
"epoch": 0.4802955665024631, |
|
"grad_norm": 1.108370916933491, |
|
"learning_rate": 1.86246554253225e-06, |
|
"loss": 0.8608, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 0.4833743842364532, |
|
"grad_norm": 1.2035299191669948, |
|
"learning_rate": 1.8467955532220017e-06, |
|
"loss": 0.85, |
|
"step": 785 |
|
}, |
|
{ |
|
"epoch": 0.48645320197044334, |
|
"grad_norm": 1.129202559563198, |
|
"learning_rate": 1.831085476440753e-06, |
|
"loss": 0.8529, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 0.4895320197044335, |
|
"grad_norm": 1.4226918512621995, |
|
"learning_rate": 1.8153371281784756e-06, |
|
"loss": 0.8566, |
|
"step": 795 |
|
}, |
|
{ |
|
"epoch": 0.49261083743842365, |
|
"grad_norm": 1.2281243961911925, |
|
"learning_rate": 1.7995523288490959e-06, |
|
"loss": 0.8712, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 0.4956896551724138, |
|
"grad_norm": 1.240241806491165, |
|
"learning_rate": 1.783732903080062e-06, |
|
"loss": 0.8636, |
|
"step": 805 |
|
}, |
|
{ |
|
"epoch": 0.4987684729064039, |
|
"grad_norm": 1.2294607557158477, |
|
"learning_rate": 1.7678806795014293e-06, |
|
"loss": 0.8594, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 0.5018472906403941, |
|
"grad_norm": 1.3745175004224655, |
|
"learning_rate": 1.751997490534482e-06, |
|
"loss": 0.8561, |
|
"step": 815 |
|
}, |
|
{ |
|
"epoch": 0.5049261083743842, |
|
"grad_norm": 1.241922432983906, |
|
"learning_rate": 1.7360851721799163e-06, |
|
"loss": 0.8189, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 0.5080049261083743, |
|
"grad_norm": 1.1305212980758934, |
|
"learning_rate": 1.7201455638056097e-06, |
|
"loss": 0.8473, |
|
"step": 825 |
|
}, |
|
{ |
|
"epoch": 0.5110837438423645, |
|
"grad_norm": 1.2269811433350686, |
|
"learning_rate": 1.7041805079340006e-06, |
|
"loss": 0.8518, |
|
"step": 830 |
|
}, |
|
{ |
|
"epoch": 0.5141625615763546, |
|
"grad_norm": 1.1609689030215362, |
|
"learning_rate": 1.6881918500291052e-06, |
|
"loss": 0.8668, |
|
"step": 835 |
|
}, |
|
{ |
|
"epoch": 0.5172413793103449, |
|
"grad_norm": 1.190209837012491, |
|
"learning_rate": 1.6721814382831911e-06, |
|
"loss": 0.8335, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 0.520320197044335, |
|
"grad_norm": 1.1974020470755453, |
|
"learning_rate": 1.6561511234031394e-06, |
|
"loss": 0.851, |
|
"step": 845 |
|
}, |
|
{ |
|
"epoch": 0.5233990147783252, |
|
"grad_norm": 1.152699709457816, |
|
"learning_rate": 1.6401027583965135e-06, |
|
"loss": 0.8547, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 0.5264778325123153, |
|
"grad_norm": 1.1460747381161807, |
|
"learning_rate": 1.624038198357361e-06, |
|
"loss": 0.8605, |
|
"step": 855 |
|
}, |
|
{ |
|
"epoch": 0.5295566502463054, |
|
"grad_norm": 1.083927796775476, |
|
"learning_rate": 1.6079593002517785e-06, |
|
"loss": 0.8874, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 0.5326354679802956, |
|
"grad_norm": 1.2069752443152426, |
|
"learning_rate": 1.5918679227032564e-06, |
|
"loss": 0.8464, |
|
"step": 865 |
|
}, |
|
{ |
|
"epoch": 0.5357142857142857, |
|
"grad_norm": 1.1237083687572078, |
|
"learning_rate": 1.575765925777834e-06, |
|
"loss": 0.8512, |
|
"step": 870 |
|
}, |
|
{ |
|
"epoch": 0.5387931034482759, |
|
"grad_norm": 1.1678972634907783, |
|
"learning_rate": 1.559655170769084e-06, |
|
"loss": 0.8504, |
|
"step": 875 |
|
}, |
|
{ |
|
"epoch": 0.541871921182266, |
|
"grad_norm": 1.1609321993586625, |
|
"learning_rate": 1.543537519982963e-06, |
|
"loss": 0.8542, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 0.5449507389162561, |
|
"grad_norm": 1.1404089632309733, |
|
"learning_rate": 1.5274148365225372e-06, |
|
"loss": 0.8514, |
|
"step": 885 |
|
}, |
|
{ |
|
"epoch": 0.5480295566502463, |
|
"grad_norm": 1.1630429155150872, |
|
"learning_rate": 1.5112889840726194e-06, |
|
"loss": 0.8716, |
|
"step": 890 |
|
}, |
|
{ |
|
"epoch": 0.5511083743842364, |
|
"grad_norm": 1.0736270999785007, |
|
"learning_rate": 1.4951618266843384e-06, |
|
"loss": 0.8354, |
|
"step": 895 |
|
}, |
|
{ |
|
"epoch": 0.5541871921182266, |
|
"grad_norm": 1.273097149325257, |
|
"learning_rate": 1.4790352285596656e-06, |
|
"loss": 0.8548, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 0.5572660098522167, |
|
"grad_norm": 1.268323524907104, |
|
"learning_rate": 1.4629110538359251e-06, |
|
"loss": 0.8431, |
|
"step": 905 |
|
}, |
|
{ |
|
"epoch": 0.5603448275862069, |
|
"grad_norm": 1.1865457508903126, |
|
"learning_rate": 1.4467911663703118e-06, |
|
"loss": 0.8447, |
|
"step": 910 |
|
}, |
|
{ |
|
"epoch": 0.5634236453201971, |
|
"grad_norm": 1.1399700864471836, |
|
"learning_rate": 1.4306774295244372e-06, |
|
"loss": 0.8575, |
|
"step": 915 |
|
}, |
|
{ |
|
"epoch": 0.5665024630541872, |
|
"grad_norm": 1.1691162634795846, |
|
"learning_rate": 1.4145717059489405e-06, |
|
"loss": 0.8554, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 0.5695812807881774, |
|
"grad_norm": 1.1199883022446742, |
|
"learning_rate": 1.3984758573681741e-06, |
|
"loss": 0.8371, |
|
"step": 925 |
|
}, |
|
{ |
|
"epoch": 0.5726600985221675, |
|
"grad_norm": 1.2188238517802348, |
|
"learning_rate": 1.3823917443649994e-06, |
|
"loss": 0.8254, |
|
"step": 930 |
|
}, |
|
{ |
|
"epoch": 0.5757389162561576, |
|
"grad_norm": 1.1444999170917183, |
|
"learning_rate": 1.3663212261657191e-06, |
|
"loss": 0.8532, |
|
"step": 935 |
|
}, |
|
{ |
|
"epoch": 0.5788177339901478, |
|
"grad_norm": 1.2197091882632716, |
|
"learning_rate": 1.3502661604251562e-06, |
|
"loss": 0.8511, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 0.5818965517241379, |
|
"grad_norm": 1.0998598595046392, |
|
"learning_rate": 1.3342284030119233e-06, |
|
"loss": 0.8385, |
|
"step": 945 |
|
}, |
|
{ |
|
"epoch": 0.5849753694581281, |
|
"grad_norm": 1.0916385771576937, |
|
"learning_rate": 1.3182098077938954e-06, |
|
"loss": 0.8397, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 0.5880541871921182, |
|
"grad_norm": 1.1536674054474005, |
|
"learning_rate": 1.3022122264239134e-06, |
|
"loss": 0.8684, |
|
"step": 955 |
|
}, |
|
{ |
|
"epoch": 0.5911330049261084, |
|
"grad_norm": 1.1341075308777682, |
|
"learning_rate": 1.286237508125744e-06, |
|
"loss": 0.8647, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 0.5942118226600985, |
|
"grad_norm": 1.1259792884659034, |
|
"learning_rate": 1.2702874994803206e-06, |
|
"loss": 0.8567, |
|
"step": 965 |
|
}, |
|
{ |
|
"epoch": 0.5972906403940886, |
|
"grad_norm": 1.2111985480502694, |
|
"learning_rate": 1.254364044212291e-06, |
|
"loss": 0.8604, |
|
"step": 970 |
|
}, |
|
{ |
|
"epoch": 0.6003694581280788, |
|
"grad_norm": 1.1728213388321937, |
|
"learning_rate": 1.2384689829768929e-06, |
|
"loss": 0.8333, |
|
"step": 975 |
|
}, |
|
{ |
|
"epoch": 0.603448275862069, |
|
"grad_norm": 1.1943182089701272, |
|
"learning_rate": 1.2226041531471835e-06, |
|
"loss": 0.8608, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 0.6065270935960592, |
|
"grad_norm": 1.2208920148774323, |
|
"learning_rate": 1.2067713886016547e-06, |
|
"loss": 0.8345, |
|
"step": 985 |
|
}, |
|
{ |
|
"epoch": 0.6096059113300493, |
|
"grad_norm": 1.1504993360515716, |
|
"learning_rate": 1.1909725195122443e-06, |
|
"loss": 0.854, |
|
"step": 990 |
|
}, |
|
{ |
|
"epoch": 0.6126847290640394, |
|
"grad_norm": 1.1040721149822834, |
|
"learning_rate": 1.1752093721327784e-06, |
|
"loss": 0.8385, |
|
"step": 995 |
|
}, |
|
{ |
|
"epoch": 0.6157635467980296, |
|
"grad_norm": 1.0838254979213422, |
|
"learning_rate": 1.1594837685878725e-06, |
|
"loss": 0.8226, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.6188423645320197, |
|
"grad_norm": 1.2581857165566288, |
|
"learning_rate": 1.143797526662299e-06, |
|
"loss": 0.8707, |
|
"step": 1005 |
|
}, |
|
{ |
|
"epoch": 0.6219211822660099, |
|
"grad_norm": 1.1364414289275864, |
|
"learning_rate": 1.1281524595908653e-06, |
|
"loss": 0.8363, |
|
"step": 1010 |
|
}, |
|
{ |
|
"epoch": 0.625, |
|
"grad_norm": 1.1368845890793131, |
|
"learning_rate": 1.1125503758488158e-06, |
|
"loss": 0.8571, |
|
"step": 1015 |
|
}, |
|
{ |
|
"epoch": 0.6280788177339901, |
|
"grad_norm": 1.183712056366396, |
|
"learning_rate": 1.0969930789427798e-06, |
|
"loss": 0.8522, |
|
"step": 1020 |
|
}, |
|
{ |
|
"epoch": 0.6311576354679803, |
|
"grad_norm": 1.162393195400783, |
|
"learning_rate": 1.0814823672023007e-06, |
|
"loss": 0.844, |
|
"step": 1025 |
|
}, |
|
{ |
|
"epoch": 0.6342364532019704, |
|
"grad_norm": 1.1672928232591524, |
|
"learning_rate": 1.0660200335719569e-06, |
|
"loss": 0.8638, |
|
"step": 1030 |
|
}, |
|
{ |
|
"epoch": 0.6373152709359606, |
|
"grad_norm": 1.1741999257753903, |
|
"learning_rate": 1.0506078654041095e-06, |
|
"loss": 0.8549, |
|
"step": 1035 |
|
}, |
|
{ |
|
"epoch": 0.6403940886699507, |
|
"grad_norm": 1.1575054446436286, |
|
"learning_rate": 1.0352476442522963e-06, |
|
"loss": 0.8667, |
|
"step": 1040 |
|
}, |
|
{ |
|
"epoch": 0.6434729064039408, |
|
"grad_norm": 1.206898852230806, |
|
"learning_rate": 1.0199411456652932e-06, |
|
"loss": 0.8347, |
|
"step": 1045 |
|
}, |
|
{ |
|
"epoch": 0.646551724137931, |
|
"grad_norm": 1.1096576965744405, |
|
"learning_rate": 1.004690138981871e-06, |
|
"loss": 0.868, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 0.6496305418719212, |
|
"grad_norm": 1.1489836338113144, |
|
"learning_rate": 9.89496387126274e-07, |
|
"loss": 0.8495, |
|
"step": 1055 |
|
}, |
|
{ |
|
"epoch": 0.6527093596059114, |
|
"grad_norm": 1.1522300404718326, |
|
"learning_rate": 9.74361646404432e-07, |
|
"loss": 0.8248, |
|
"step": 1060 |
|
}, |
|
{ |
|
"epoch": 0.6557881773399015, |
|
"grad_norm": 1.2033254283693793, |
|
"learning_rate": 9.592876663009462e-07, |
|
"loss": 0.8535, |
|
"step": 1065 |
|
}, |
|
{ |
|
"epoch": 0.6588669950738916, |
|
"grad_norm": 1.137432240293731, |
|
"learning_rate": 9.442761892768561e-07, |
|
"loss": 0.8275, |
|
"step": 1070 |
|
}, |
|
{ |
|
"epoch": 0.6619458128078818, |
|
"grad_norm": 1.15459107431455, |
|
"learning_rate": 9.293289505682223e-07, |
|
"loss": 0.862, |
|
"step": 1075 |
|
}, |
|
{ |
|
"epoch": 0.6650246305418719, |
|
"grad_norm": 1.1632426422489177, |
|
"learning_rate": 9.144476779855462e-07, |
|
"loss": 0.8722, |
|
"step": 1080 |
|
}, |
|
{ |
|
"epoch": 0.6681034482758621, |
|
"grad_norm": 1.1564452407327455, |
|
"learning_rate": 8.99634091714042e-07, |
|
"loss": 0.8572, |
|
"step": 1085 |
|
}, |
|
{ |
|
"epoch": 0.6711822660098522, |
|
"grad_norm": 1.124466204967846, |
|
"learning_rate": 8.848899041147947e-07, |
|
"loss": 0.8453, |
|
"step": 1090 |
|
}, |
|
{ |
|
"epoch": 0.6742610837438424, |
|
"grad_norm": 1.0438572933417327, |
|
"learning_rate": 8.702168195268239e-07, |
|
"loss": 0.8387, |
|
"step": 1095 |
|
}, |
|
{ |
|
"epoch": 0.6773399014778325, |
|
"grad_norm": 1.113884530794887, |
|
"learning_rate": 8.556165340700687e-07, |
|
"loss": 0.8408, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 0.6804187192118226, |
|
"grad_norm": 1.2085543214925392, |
|
"learning_rate": 8.410907354493304e-07, |
|
"loss": 0.8394, |
|
"step": 1105 |
|
}, |
|
{ |
|
"epoch": 0.6834975369458128, |
|
"grad_norm": 1.5442045796635173, |
|
"learning_rate": 8.266411027591801e-07, |
|
"loss": 0.859, |
|
"step": 1110 |
|
}, |
|
{ |
|
"epoch": 0.6865763546798029, |
|
"grad_norm": 1.1741040287340676, |
|
"learning_rate": 8.12269306289868e-07, |
|
"loss": 0.8465, |
|
"step": 1115 |
|
}, |
|
{ |
|
"epoch": 0.6896551724137931, |
|
"grad_norm": 1.2373266225797797, |
|
"learning_rate": 7.979770073342484e-07, |
|
"loss": 0.8556, |
|
"step": 1120 |
|
}, |
|
{ |
|
"epoch": 0.6927339901477833, |
|
"grad_norm": 1.182099641627804, |
|
"learning_rate": 7.837658579957422e-07, |
|
"loss": 0.842, |
|
"step": 1125 |
|
}, |
|
{ |
|
"epoch": 0.6958128078817734, |
|
"grad_norm": 1.121108609994855, |
|
"learning_rate": 7.696375009973643e-07, |
|
"loss": 0.8312, |
|
"step": 1130 |
|
}, |
|
{ |
|
"epoch": 0.6988916256157636, |
|
"grad_norm": 1.121279263938038, |
|
"learning_rate": 7.555935694918371e-07, |
|
"loss": 0.8553, |
|
"step": 1135 |
|
}, |
|
{ |
|
"epoch": 0.7019704433497537, |
|
"grad_norm": 1.1204765733762827, |
|
"learning_rate": 7.41635686872804e-07, |
|
"loss": 0.8524, |
|
"step": 1140 |
|
}, |
|
{ |
|
"epoch": 0.7050492610837439, |
|
"grad_norm": 1.0928742671497675, |
|
"learning_rate": 7.277654665871785e-07, |
|
"loss": 0.8227, |
|
"step": 1145 |
|
}, |
|
{ |
|
"epoch": 0.708128078817734, |
|
"grad_norm": 1.180402060361, |
|
"learning_rate": 7.139845119486371e-07, |
|
"loss": 0.8251, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 0.7112068965517241, |
|
"grad_norm": 1.153545166631368, |
|
"learning_rate": 7.00294415952286e-07, |
|
"loss": 0.8511, |
|
"step": 1155 |
|
}, |
|
{ |
|
"epoch": 0.7142857142857143, |
|
"grad_norm": 1.20278698818192, |
|
"learning_rate": 6.866967610905234e-07, |
|
"loss": 0.8594, |
|
"step": 1160 |
|
}, |
|
{ |
|
"epoch": 0.7173645320197044, |
|
"grad_norm": 1.1752665349603137, |
|
"learning_rate": 6.731931191701088e-07, |
|
"loss": 0.8406, |
|
"step": 1165 |
|
}, |
|
{ |
|
"epoch": 0.7204433497536946, |
|
"grad_norm": 1.1775568699552572, |
|
"learning_rate": 6.597850511304739e-07, |
|
"loss": 0.8306, |
|
"step": 1170 |
|
}, |
|
{ |
|
"epoch": 0.7235221674876847, |
|
"grad_norm": 1.1360858811762655, |
|
"learning_rate": 6.464741068632883e-07, |
|
"loss": 0.8347, |
|
"step": 1175 |
|
}, |
|
{ |
|
"epoch": 0.7266009852216748, |
|
"grad_norm": 1.1202993090837823, |
|
"learning_rate": 6.332618250332988e-07, |
|
"loss": 0.8241, |
|
"step": 1180 |
|
}, |
|
{ |
|
"epoch": 0.729679802955665, |
|
"grad_norm": 1.1521986508617748, |
|
"learning_rate": 6.201497329004718e-07, |
|
"loss": 0.8631, |
|
"step": 1185 |
|
}, |
|
{ |
|
"epoch": 0.7327586206896551, |
|
"grad_norm": 1.1544851365131765, |
|
"learning_rate": 6.071393461434488e-07, |
|
"loss": 0.8469, |
|
"step": 1190 |
|
}, |
|
{ |
|
"epoch": 0.7358374384236454, |
|
"grad_norm": 1.1602821185558192, |
|
"learning_rate": 5.942321686843444e-07, |
|
"loss": 0.8353, |
|
"step": 1195 |
|
}, |
|
{ |
|
"epoch": 0.7389162561576355, |
|
"grad_norm": 1.117263157859087, |
|
"learning_rate": 5.814296925149026e-07, |
|
"loss": 0.8668, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 0.7419950738916257, |
|
"grad_norm": 1.1493367054140813, |
|
"learning_rate": 5.687333975240304e-07, |
|
"loss": 0.8415, |
|
"step": 1205 |
|
}, |
|
{ |
|
"epoch": 0.7450738916256158, |
|
"grad_norm": 1.1925389719718977, |
|
"learning_rate": 5.561447513267311e-07, |
|
"loss": 0.8545, |
|
"step": 1210 |
|
}, |
|
{ |
|
"epoch": 0.7481527093596059, |
|
"grad_norm": 1.086104321966628, |
|
"learning_rate": 5.436652090944596e-07, |
|
"loss": 0.8336, |
|
"step": 1215 |
|
}, |
|
{ |
|
"epoch": 0.7512315270935961, |
|
"grad_norm": 1.123629920924981, |
|
"learning_rate": 5.312962133869093e-07, |
|
"loss": 0.8529, |
|
"step": 1220 |
|
}, |
|
{ |
|
"epoch": 0.7543103448275862, |
|
"grad_norm": 1.1532520106866275, |
|
"learning_rate": 5.19039193985266e-07, |
|
"loss": 0.8337, |
|
"step": 1225 |
|
}, |
|
{ |
|
"epoch": 0.7573891625615764, |
|
"grad_norm": 1.1434038160488575, |
|
"learning_rate": 5.068955677269281e-07, |
|
"loss": 0.8454, |
|
"step": 1230 |
|
}, |
|
{ |
|
"epoch": 0.7604679802955665, |
|
"grad_norm": 1.062773583023481, |
|
"learning_rate": 4.948667383417332e-07, |
|
"loss": 0.8619, |
|
"step": 1235 |
|
}, |
|
{ |
|
"epoch": 0.7635467980295566, |
|
"grad_norm": 1.1189719192133103, |
|
"learning_rate": 4.829540962896927e-07, |
|
"loss": 0.8458, |
|
"step": 1240 |
|
}, |
|
{ |
|
"epoch": 0.7666256157635468, |
|
"grad_norm": 1.1497616317100816, |
|
"learning_rate": 4.7115901860026634e-07, |
|
"loss": 0.8553, |
|
"step": 1245 |
|
}, |
|
{ |
|
"epoch": 0.7697044334975369, |
|
"grad_norm": 1.117467647853885, |
|
"learning_rate": 4.594828687131814e-07, |
|
"loss": 0.8422, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 0.7727832512315271, |
|
"grad_norm": 1.1545001867144575, |
|
"learning_rate": 4.4792699632083043e-07, |
|
"loss": 0.8341, |
|
"step": 1255 |
|
}, |
|
{ |
|
"epoch": 0.7758620689655172, |
|
"grad_norm": 1.2094683485025837, |
|
"learning_rate": 4.36492737212255e-07, |
|
"loss": 0.8498, |
|
"step": 1260 |
|
}, |
|
{ |
|
"epoch": 0.7789408866995073, |
|
"grad_norm": 1.108537189945719, |
|
"learning_rate": 4.251814131187349e-07, |
|
"loss": 0.8331, |
|
"step": 1265 |
|
}, |
|
{ |
|
"epoch": 0.7820197044334976, |
|
"grad_norm": 1.1631674970790735, |
|
"learning_rate": 4.13994331561004e-07, |
|
"loss": 0.8291, |
|
"step": 1270 |
|
}, |
|
{ |
|
"epoch": 0.7850985221674877, |
|
"grad_norm": 1.3825542298378888, |
|
"learning_rate": 4.0293278569811197e-07, |
|
"loss": 0.8358, |
|
"step": 1275 |
|
}, |
|
{ |
|
"epoch": 0.7881773399014779, |
|
"grad_norm": 1.1720833443366188, |
|
"learning_rate": 3.9199805417793833e-07, |
|
"loss": 0.8492, |
|
"step": 1280 |
|
}, |
|
{ |
|
"epoch": 0.791256157635468, |
|
"grad_norm": 1.15561431106047, |
|
"learning_rate": 3.8119140098939074e-07, |
|
"loss": 0.8387, |
|
"step": 1285 |
|
}, |
|
{ |
|
"epoch": 0.7943349753694581, |
|
"grad_norm": 1.1154866318662433, |
|
"learning_rate": 3.705140753162973e-07, |
|
"loss": 0.8674, |
|
"step": 1290 |
|
}, |
|
{ |
|
"epoch": 0.7974137931034483, |
|
"grad_norm": 1.1286513773992464, |
|
"learning_rate": 3.5996731139300476e-07, |
|
"loss": 0.8412, |
|
"step": 1295 |
|
}, |
|
{ |
|
"epoch": 0.8004926108374384, |
|
"grad_norm": 1.1405012893425726, |
|
"learning_rate": 3.495523283617106e-07, |
|
"loss": 0.851, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 0.8035714285714286, |
|
"grad_norm": 1.1507250504486406, |
|
"learning_rate": 3.39270330131538e-07, |
|
"loss": 0.8401, |
|
"step": 1305 |
|
}, |
|
{ |
|
"epoch": 0.8066502463054187, |
|
"grad_norm": 1.2824212484601696, |
|
"learning_rate": 3.2912250523937e-07, |
|
"loss": 0.8554, |
|
"step": 1310 |
|
}, |
|
{ |
|
"epoch": 0.8097290640394089, |
|
"grad_norm": 1.229837096715884, |
|
"learning_rate": 3.1911002671246164e-07, |
|
"loss": 0.8408, |
|
"step": 1315 |
|
}, |
|
{ |
|
"epoch": 0.812807881773399, |
|
"grad_norm": 1.1262808794150745, |
|
"learning_rate": 3.092340519328474e-07, |
|
"loss": 0.8846, |
|
"step": 1320 |
|
}, |
|
{ |
|
"epoch": 0.8158866995073891, |
|
"grad_norm": 1.1513720812150885, |
|
"learning_rate": 2.9949572250355176e-07, |
|
"loss": 0.8222, |
|
"step": 1325 |
|
}, |
|
{ |
|
"epoch": 0.8189655172413793, |
|
"grad_norm": 1.0774467677577528, |
|
"learning_rate": 2.8989616411662826e-07, |
|
"loss": 0.8482, |
|
"step": 1330 |
|
}, |
|
{ |
|
"epoch": 0.8220443349753694, |
|
"grad_norm": 1.117458202223298, |
|
"learning_rate": 2.8043648642303716e-07, |
|
"loss": 0.8318, |
|
"step": 1335 |
|
}, |
|
{ |
|
"epoch": 0.8251231527093597, |
|
"grad_norm": 1.1761574400752972, |
|
"learning_rate": 2.7111778290437465e-07, |
|
"loss": 0.8545, |
|
"step": 1340 |
|
}, |
|
{ |
|
"epoch": 0.8282019704433498, |
|
"grad_norm": 1.1095160101348673, |
|
"learning_rate": 2.619411307464747e-07, |
|
"loss": 0.842, |
|
"step": 1345 |
|
}, |
|
{ |
|
"epoch": 0.8312807881773399, |
|
"grad_norm": 1.1311968412403977, |
|
"learning_rate": 2.529075907148916e-07, |
|
"loss": 0.8514, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 0.8343596059113301, |
|
"grad_norm": 1.124889089284095, |
|
"learning_rate": 2.440182070322818e-07, |
|
"loss": 0.8439, |
|
"step": 1355 |
|
}, |
|
{ |
|
"epoch": 0.8374384236453202, |
|
"grad_norm": 1.2124152431980482, |
|
"learning_rate": 2.352740072577002e-07, |
|
"loss": 0.8381, |
|
"step": 1360 |
|
}, |
|
{ |
|
"epoch": 0.8405172413793104, |
|
"grad_norm": 1.1285438400669963, |
|
"learning_rate": 2.2667600216781757e-07, |
|
"loss": 0.8287, |
|
"step": 1365 |
|
}, |
|
{ |
|
"epoch": 0.8435960591133005, |
|
"grad_norm": 1.2161816958846985, |
|
"learning_rate": 2.182251856400826e-07, |
|
"loss": 0.8414, |
|
"step": 1370 |
|
}, |
|
{ |
|
"epoch": 0.8466748768472906, |
|
"grad_norm": 1.1752287036084772, |
|
"learning_rate": 2.0992253453783595e-07, |
|
"loss": 0.8653, |
|
"step": 1375 |
|
}, |
|
{ |
|
"epoch": 0.8497536945812808, |
|
"grad_norm": 1.1584226215126705, |
|
"learning_rate": 2.0176900859738906e-07, |
|
"loss": 0.8525, |
|
"step": 1380 |
|
}, |
|
{ |
|
"epoch": 0.8528325123152709, |
|
"grad_norm": 1.183555004522999, |
|
"learning_rate": 1.93765550317087e-07, |
|
"loss": 0.8287, |
|
"step": 1385 |
|
}, |
|
{ |
|
"epoch": 0.8559113300492611, |
|
"grad_norm": 1.1306173343667651, |
|
"learning_rate": 1.8591308484835833e-07, |
|
"loss": 0.8695, |
|
"step": 1390 |
|
}, |
|
{ |
|
"epoch": 0.8589901477832512, |
|
"grad_norm": 1.1008522143759358, |
|
"learning_rate": 1.7821251988877556e-07, |
|
"loss": 0.8476, |
|
"step": 1395 |
|
}, |
|
{ |
|
"epoch": 0.8620689655172413, |
|
"grad_norm": 1.180832971444619, |
|
"learning_rate": 1.706647455771302e-07, |
|
"loss": 0.8499, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 0.8651477832512315, |
|
"grad_norm": 1.1442312991912245, |
|
"learning_rate": 1.6327063439053675e-07, |
|
"loss": 0.8525, |
|
"step": 1405 |
|
}, |
|
{ |
|
"epoch": 0.8682266009852216, |
|
"grad_norm": 1.1303941912871311, |
|
"learning_rate": 1.56031041043582e-07, |
|
"loss": 0.855, |
|
"step": 1410 |
|
}, |
|
{ |
|
"epoch": 0.8713054187192119, |
|
"grad_norm": 1.205765651469266, |
|
"learning_rate": 1.4894680238952318e-07, |
|
"loss": 0.8599, |
|
"step": 1415 |
|
}, |
|
{ |
|
"epoch": 0.874384236453202, |
|
"grad_norm": 1.1362909634405434, |
|
"learning_rate": 1.4201873732355343e-07, |
|
"loss": 0.8482, |
|
"step": 1420 |
|
}, |
|
{ |
|
"epoch": 0.8774630541871922, |
|
"grad_norm": 1.1002574562968226, |
|
"learning_rate": 1.3524764668814305e-07, |
|
"loss": 0.8604, |
|
"step": 1425 |
|
}, |
|
{ |
|
"epoch": 0.8805418719211823, |
|
"grad_norm": 1.2261315487517963, |
|
"learning_rate": 1.2863431318046615e-07, |
|
"loss": 0.8797, |
|
"step": 1430 |
|
}, |
|
{ |
|
"epoch": 0.8836206896551724, |
|
"grad_norm": 1.1281285330103035, |
|
"learning_rate": 1.2217950126192523e-07, |
|
"loss": 0.8502, |
|
"step": 1435 |
|
}, |
|
{ |
|
"epoch": 0.8866995073891626, |
|
"grad_norm": 1.3067066600493065, |
|
"learning_rate": 1.158839570697861e-07, |
|
"loss": 0.8558, |
|
"step": 1440 |
|
}, |
|
{ |
|
"epoch": 0.8897783251231527, |
|
"grad_norm": 1.1074762795586004, |
|
"learning_rate": 1.0974840833092603e-07, |
|
"loss": 0.827, |
|
"step": 1445 |
|
}, |
|
{ |
|
"epoch": 0.8928571428571429, |
|
"grad_norm": 1.1515662827940316, |
|
"learning_rate": 1.0377356427771567e-07, |
|
"loss": 0.8521, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 0.895935960591133, |
|
"grad_norm": 1.182715294997636, |
|
"learning_rate": 9.796011556603418e-08, |
|
"loss": 0.8333, |
|
"step": 1455 |
|
}, |
|
{ |
|
"epoch": 0.8990147783251231, |
|
"grad_norm": 1.174648240411651, |
|
"learning_rate": 9.230873419543373e-08, |
|
"loss": 0.8414, |
|
"step": 1460 |
|
}, |
|
{ |
|
"epoch": 0.9020935960591133, |
|
"grad_norm": 1.164751938603579, |
|
"learning_rate": 8.682007343146198e-08, |
|
"loss": 0.8427, |
|
"step": 1465 |
|
}, |
|
{ |
|
"epoch": 0.9051724137931034, |
|
"grad_norm": 1.114964693831345, |
|
"learning_rate": 8.14947677301468e-08, |
|
"loss": 0.8449, |
|
"step": 1470 |
|
}, |
|
{ |
|
"epoch": 0.9082512315270936, |
|
"grad_norm": 1.1386575763510343, |
|
"learning_rate": 7.633343266465731e-08, |
|
"loss": 0.8355, |
|
"step": 1475 |
|
}, |
|
{ |
|
"epoch": 0.9113300492610837, |
|
"grad_norm": 1.0913394171845956, |
|
"learning_rate": 7.133666485414858e-08, |
|
"loss": 0.8456, |
|
"step": 1480 |
|
}, |
|
{ |
|
"epoch": 0.9144088669950738, |
|
"grad_norm": 1.127034006921905, |
|
"learning_rate": 6.650504189479417e-08, |
|
"loss": 0.8186, |
|
"step": 1485 |
|
}, |
|
{ |
|
"epoch": 0.9174876847290641, |
|
"grad_norm": 1.1487057626187083, |
|
"learning_rate": 6.183912229302135e-08, |
|
"loss": 0.8648, |
|
"step": 1490 |
|
}, |
|
{ |
|
"epoch": 0.9205665024630542, |
|
"grad_norm": 1.1340056386882922, |
|
"learning_rate": 5.7339445400949763e-08, |
|
"loss": 0.8292, |
|
"step": 1495 |
|
}, |
|
{ |
|
"epoch": 0.9236453201970444, |
|
"grad_norm": 1.1203682903661263, |
|
"learning_rate": 5.3006531354045596e-08, |
|
"loss": 0.854, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.9267241379310345, |
|
"grad_norm": 1.0928512210320804, |
|
"learning_rate": 4.8840881010998536e-08, |
|
"loss": 0.8358, |
|
"step": 1505 |
|
}, |
|
{ |
|
"epoch": 0.9298029556650246, |
|
"grad_norm": 1.1132516765401899, |
|
"learning_rate": 4.4842975895823926e-08, |
|
"loss": 0.8648, |
|
"step": 1510 |
|
}, |
|
{ |
|
"epoch": 0.9328817733990148, |
|
"grad_norm": 1.1498839085251829, |
|
"learning_rate": 4.1013278142201904e-08, |
|
"loss": 0.849, |
|
"step": 1515 |
|
}, |
|
{ |
|
"epoch": 0.9359605911330049, |
|
"grad_norm": 1.101399361627499, |
|
"learning_rate": 3.7352230440058534e-08, |
|
"loss": 0.8541, |
|
"step": 1520 |
|
}, |
|
{ |
|
"epoch": 0.9390394088669951, |
|
"grad_norm": 1.0935788387746401, |
|
"learning_rate": 3.386025598439185e-08, |
|
"loss": 0.8506, |
|
"step": 1525 |
|
}, |
|
{ |
|
"epoch": 0.9421182266009852, |
|
"grad_norm": 1.1237165898312163, |
|
"learning_rate": 3.053775842635453e-08, |
|
"loss": 0.8508, |
|
"step": 1530 |
|
}, |
|
{ |
|
"epoch": 0.9451970443349754, |
|
"grad_norm": 1.145170230124706, |
|
"learning_rate": 2.7385121826593363e-08, |
|
"loss": 0.8414, |
|
"step": 1535 |
|
}, |
|
{ |
|
"epoch": 0.9482758620689655, |
|
"grad_norm": 1.1489967926566926, |
|
"learning_rate": 2.4402710610854582e-08, |
|
"loss": 0.8342, |
|
"step": 1540 |
|
}, |
|
{ |
|
"epoch": 0.9513546798029556, |
|
"grad_norm": 1.1339560978358958, |
|
"learning_rate": 2.159086952785827e-08, |
|
"loss": 0.8282, |
|
"step": 1545 |
|
}, |
|
{ |
|
"epoch": 0.9544334975369458, |
|
"grad_norm": 1.1397207455047003, |
|
"learning_rate": 1.894992360944786e-08, |
|
"loss": 0.8322, |
|
"step": 1550 |
|
}, |
|
{ |
|
"epoch": 0.9575123152709359, |
|
"grad_norm": 1.0877530145758472, |
|
"learning_rate": 1.648017813301739e-08, |
|
"loss": 0.8425, |
|
"step": 1555 |
|
}, |
|
{ |
|
"epoch": 0.9605911330049262, |
|
"grad_norm": 1.2936654102555174, |
|
"learning_rate": 1.4181918586225029e-08, |
|
"loss": 0.8428, |
|
"step": 1560 |
|
}, |
|
{ |
|
"epoch": 0.9636699507389163, |
|
"grad_norm": 1.0579175518840687, |
|
"learning_rate": 1.2055410633990515e-08, |
|
"loss": 0.8535, |
|
"step": 1565 |
|
}, |
|
{ |
|
"epoch": 0.9667487684729064, |
|
"grad_norm": 1.1404735068405747, |
|
"learning_rate": 1.0100900087787357e-08, |
|
"loss": 0.8383, |
|
"step": 1570 |
|
}, |
|
{ |
|
"epoch": 0.9698275862068966, |
|
"grad_norm": 1.0272418184002488, |
|
"learning_rate": 8.318612877227571e-09, |
|
"loss": 0.8588, |
|
"step": 1575 |
|
}, |
|
{ |
|
"epoch": 0.9729064039408867, |
|
"grad_norm": 1.1026771910871591, |
|
"learning_rate": 6.708755023946245e-09, |
|
"loss": 0.8446, |
|
"step": 1580 |
|
}, |
|
{ |
|
"epoch": 0.9759852216748769, |
|
"grad_norm": 1.1415727609152322, |
|
"learning_rate": 5.271512617786078e-09, |
|
"loss": 0.8573, |
|
"step": 1585 |
|
}, |
|
{ |
|
"epoch": 0.979064039408867, |
|
"grad_norm": 1.122482714286765, |
|
"learning_rate": 4.007051795287098e-09, |
|
"loss": 0.8329, |
|
"step": 1590 |
|
}, |
|
{ |
|
"epoch": 0.9821428571428571, |
|
"grad_norm": 1.1191940499558126, |
|
"learning_rate": 2.9155187204820134e-09, |
|
"loss": 0.8513, |
|
"step": 1595 |
|
}, |
|
{ |
|
"epoch": 0.9852216748768473, |
|
"grad_norm": 1.118386462087675, |
|
"learning_rate": 1.997039568000403e-09, |
|
"loss": 0.8128, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 0.9883004926108374, |
|
"grad_norm": 1.0994770155637474, |
|
"learning_rate": 1.2517205084837112e-09, |
|
"loss": 0.8511, |
|
"step": 1605 |
|
}, |
|
{ |
|
"epoch": 0.9913793103448276, |
|
"grad_norm": 1.1612834899755462, |
|
"learning_rate": 6.796476963130683e-10, |
|
"loss": 0.8427, |
|
"step": 1610 |
|
}, |
|
{ |
|
"epoch": 0.9944581280788177, |
|
"grad_norm": 1.1803762333521288, |
|
"learning_rate": 2.8088725964958705e-10, |
|
"loss": 0.8614, |
|
"step": 1615 |
|
}, |
|
{ |
|
"epoch": 0.9975369458128078, |
|
"grad_norm": 1.1467994440426534, |
|
"learning_rate": 5.548529279081338e-11, |
|
"loss": 0.829, |
|
"step": 1620 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_loss": 0.8464784026145935, |
|
"eval_runtime": 1019.8993, |
|
"eval_samples_per_second": 22.658, |
|
"eval_steps_per_second": 0.709, |
|
"step": 1624 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"step": 1624, |
|
"total_flos": 1278025572745216.0, |
|
"train_loss": 0.8594843297095721, |
|
"train_runtime": 35019.5984, |
|
"train_samples_per_second": 5.936, |
|
"train_steps_per_second": 0.046 |
|
} |
|
], |
|
"logging_steps": 5, |
|
"max_steps": 1624, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 500, |
|
"stateful_callbacks": { |
|
"TrainerControl": { |
|
"args": { |
|
"should_epoch_stop": false, |
|
"should_evaluate": false, |
|
"should_log": false, |
|
"should_save": false, |
|
"should_training_stop": false |
|
}, |
|
"attributes": {} |
|
} |
|
}, |
|
"total_flos": 1278025572745216.0, |
|
"train_batch_size": 4, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|