| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 260010, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.990384985192877e-05, | |
| "loss": 2.7022, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.980769970385754e-05, | |
| "loss": 2.1685, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.971154955578632e-05, | |
| "loss": 2.011, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.961539940771509e-05, | |
| "loss": 1.8707, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.951924925964386e-05, | |
| "loss": 1.8268, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.942309911157263e-05, | |
| "loss": 1.7562, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.932694896350141e-05, | |
| "loss": 1.7447, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.923079881543018e-05, | |
| "loss": 1.7022, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.913464866735895e-05, | |
| "loss": 1.6657, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.903849851928772e-05, | |
| "loss": 1.6436, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.89423483712165e-05, | |
| "loss": 1.6203, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.884619822314527e-05, | |
| "loss": 1.5773, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.875004807507404e-05, | |
| "loss": 1.5829, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.865389792700281e-05, | |
| "loss": 1.5611, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.8557747778931586e-05, | |
| "loss": 1.5456, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.8461597630860356e-05, | |
| "loss": 1.5471, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.836544748278913e-05, | |
| "loss": 1.5208, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.82692973347179e-05, | |
| "loss": 1.5099, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8173147186646674e-05, | |
| "loss": 1.4846, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.8076997038575445e-05, | |
| "loss": 1.4845, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.7980846890504215e-05, | |
| "loss": 1.4877, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7884696742432986e-05, | |
| "loss": 1.4947, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7788546594361756e-05, | |
| "loss": 1.4707, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.7692396446290527e-05, | |
| "loss": 1.4568, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.75962462982193e-05, | |
| "loss": 1.4249, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.7500096150148074e-05, | |
| "loss": 1.4197, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7403946002076845e-05, | |
| "loss": 1.4406, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7307795854005615e-05, | |
| "loss": 1.4452, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.7211645705934385e-05, | |
| "loss": 1.3891, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.711549555786316e-05, | |
| "loss": 1.4029, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.701934540979193e-05, | |
| "loss": 1.3949, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.6923195261720703e-05, | |
| "loss": 1.4012, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.6827045113649474e-05, | |
| "loss": 1.3649, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.673089496557825e-05, | |
| "loss": 1.3831, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.663474481750702e-05, | |
| "loss": 1.3716, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.653859466943579e-05, | |
| "loss": 1.3672, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.644244452136456e-05, | |
| "loss": 1.348, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.634629437329334e-05, | |
| "loss": 1.3364, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.625014422522211e-05, | |
| "loss": 1.3271, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.615399407715088e-05, | |
| "loss": 1.3145, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.605784392907965e-05, | |
| "loss": 1.3291, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.596169378100843e-05, | |
| "loss": 1.3227, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.58655436329372e-05, | |
| "loss": 1.335, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.576939348486597e-05, | |
| "loss": 1.3266, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.567324333679474e-05, | |
| "loss": 1.3244, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.557709318872351e-05, | |
| "loss": 1.302, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.548094304065229e-05, | |
| "loss": 1.303, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.538479289258106e-05, | |
| "loss": 1.3059, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.528864274450983e-05, | |
| "loss": 1.3215, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.51924925964386e-05, | |
| "loss": 1.3002, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.5096342448367375e-05, | |
| "loss": 1.2681, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.5000192300296146e-05, | |
| "loss": 1.2959, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.4904042152224916e-05, | |
| "loss": 1.2797, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.480789200415369e-05, | |
| "loss": 1.308, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.4711741856082464e-05, | |
| "loss": 1.2873, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.4615591708011234e-05, | |
| "loss": 1.2932, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.4519441559940005e-05, | |
| "loss": 1.2749, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.4423291411868775e-05, | |
| "loss": 1.2497, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.432714126379755e-05, | |
| "loss": 1.2735, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.423099111572632e-05, | |
| "loss": 1.2777, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.413484096765509e-05, | |
| "loss": 1.2641, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.4038690819583864e-05, | |
| "loss": 1.2625, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.394254067151264e-05, | |
| "loss": 1.2414, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.384639052344141e-05, | |
| "loss": 1.2534, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.375024037537018e-05, | |
| "loss": 1.24, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.365409022729895e-05, | |
| "loss": 1.2473, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.355794007922772e-05, | |
| "loss": 1.2415, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.346178993115649e-05, | |
| "loss": 1.2263, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.3365639783085264e-05, | |
| "loss": 1.235, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.326948963501404e-05, | |
| "loss": 1.2291, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.317333948694281e-05, | |
| "loss": 1.2161, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.307718933887158e-05, | |
| "loss": 1.2292, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.298103919080035e-05, | |
| "loss": 1.2329, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.288488904272913e-05, | |
| "loss": 1.2205, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.27887388946579e-05, | |
| "loss": 1.2117, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.269258874658667e-05, | |
| "loss": 1.2276, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.259643859851544e-05, | |
| "loss": 1.2138, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.250028845044422e-05, | |
| "loss": 1.2131, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.240413830237299e-05, | |
| "loss": 1.2088, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.230798815430176e-05, | |
| "loss": 1.2029, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.221183800623053e-05, | |
| "loss": 1.1927, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.2115687858159306e-05, | |
| "loss": 1.2093, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.2019537710088077e-05, | |
| "loss": 1.19, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.192338756201685e-05, | |
| "loss": 1.2042, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.182723741394562e-05, | |
| "loss": 1.201, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.173108726587439e-05, | |
| "loss": 1.1916, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.1634937117803165e-05, | |
| "loss": 1.1834, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.1538786969731935e-05, | |
| "loss": 1.1919, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.1442636821660706e-05, | |
| "loss": 1.1969, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.1346486673589476e-05, | |
| "loss": 1.1765, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.1250336525518254e-05, | |
| "loss": 1.2079, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.1154186377447024e-05, | |
| "loss": 1.1711, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.1058036229375794e-05, | |
| "loss": 1.1819, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.0961886081304565e-05, | |
| "loss": 1.1569, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.086573593323334e-05, | |
| "loss": 1.1762, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.076958578516211e-05, | |
| "loss": 1.1818, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.067343563709088e-05, | |
| "loss": 1.1843, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.057728548901965e-05, | |
| "loss": 1.1776, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.048113534094843e-05, | |
| "loss": 1.1853, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.03849851928772e-05, | |
| "loss": 1.1608, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.028883504480597e-05, | |
| "loss": 1.1748, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.019268489673474e-05, | |
| "loss": 1.1615, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.009653474866352e-05, | |
| "loss": 1.153, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.000038460059229e-05, | |
| "loss": 1.1643, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.990423445252106e-05, | |
| "loss": 1.154, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.980808430444983e-05, | |
| "loss": 1.1675, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.971193415637861e-05, | |
| "loss": 1.1612, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.961578400830738e-05, | |
| "loss": 1.1754, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.951963386023615e-05, | |
| "loss": 1.1577, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.942348371216492e-05, | |
| "loss": 1.1551, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.932733356409369e-05, | |
| "loss": 1.1584, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.923118341602246e-05, | |
| "loss": 1.1387, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.913503326795123e-05, | |
| "loss": 1.1469, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.903888311988001e-05, | |
| "loss": 1.1597, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.894273297180878e-05, | |
| "loss": 1.1478, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.884658282373755e-05, | |
| "loss": 1.1452, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.875043267566632e-05, | |
| "loss": 1.1199, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.8654282527595096e-05, | |
| "loss": 1.1151, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.8558132379523866e-05, | |
| "loss": 1.1492, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.8461982231452637e-05, | |
| "loss": 1.1368, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.836583208338141e-05, | |
| "loss": 1.1406, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.8269681935310184e-05, | |
| "loss": 1.124, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.8173531787238955e-05, | |
| "loss": 1.1445, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.8077381639167725e-05, | |
| "loss": 1.13, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.7981231491096496e-05, | |
| "loss": 1.1114, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.788508134302527e-05, | |
| "loss": 1.1394, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.778893119495404e-05, | |
| "loss": 1.1319, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.7692781046882814e-05, | |
| "loss": 1.1191, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.7596630898811584e-05, | |
| "loss": 1.1146, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.7500480750740354e-05, | |
| "loss": 1.1106, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.740433060266913e-05, | |
| "loss": 1.1113, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.73081804545979e-05, | |
| "loss": 1.0962, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.721203030652667e-05, | |
| "loss": 1.1239, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.711588015845544e-05, | |
| "loss": 1.1105, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.701973001038422e-05, | |
| "loss": 1.1102, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.692357986231299e-05, | |
| "loss": 1.1035, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.682742971424176e-05, | |
| "loss": 1.1265, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.673127956617053e-05, | |
| "loss": 1.1136, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.663512941809931e-05, | |
| "loss": 1.1042, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.653897927002808e-05, | |
| "loss": 1.0983, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.644282912195685e-05, | |
| "loss": 1.0823, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.634667897388562e-05, | |
| "loss": 1.1024, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.62505288258144e-05, | |
| "loss": 1.1067, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.615437867774317e-05, | |
| "loss": 1.093, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.605822852967194e-05, | |
| "loss": 1.0896, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.596207838160071e-05, | |
| "loss": 1.1115, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.5865928233529486e-05, | |
| "loss": 1.1029, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.5769778085458256e-05, | |
| "loss": 1.0828, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.5673627937387026e-05, | |
| "loss": 1.0969, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.55774777893158e-05, | |
| "loss": 1.0817, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.5481327641244574e-05, | |
| "loss": 1.0967, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.5385177493173344e-05, | |
| "loss": 1.1131, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.5289027345102115e-05, | |
| "loss": 1.0754, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.5192877197030885e-05, | |
| "loss": 1.0707, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.5096727048959656e-05, | |
| "loss": 1.0947, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.5000576900888426e-05, | |
| "loss": 1.104, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.4904426752817197e-05, | |
| "loss": 1.0834, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.4808276604745974e-05, | |
| "loss": 1.0953, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.4712126456674744e-05, | |
| "loss": 1.0776, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.4615976308603515e-05, | |
| "loss": 1.0832, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.4519826160532285e-05, | |
| "loss": 1.0796, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.442367601246106e-05, | |
| "loss": 1.0981, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.432752586438983e-05, | |
| "loss": 1.0704, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.42313757163186e-05, | |
| "loss": 1.0678, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.4135225568247374e-05, | |
| "loss": 1.0775, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.403907542017615e-05, | |
| "loss": 1.0857, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.394292527210492e-05, | |
| "loss": 1.0775, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.384677512403369e-05, | |
| "loss": 1.0674, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.375062497596246e-05, | |
| "loss": 1.0601, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.365447482789123e-05, | |
| "loss": 1.0549, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.355832467982001e-05, | |
| "loss": 1.0586, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.346217453174878e-05, | |
| "loss": 1.0681, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.336602438367755e-05, | |
| "loss": 1.0644, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.326987423560632e-05, | |
| "loss": 1.0063, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.31737240875351e-05, | |
| "loss": 0.9661, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.307757393946387e-05, | |
| "loss": 0.956, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.298142379139264e-05, | |
| "loss": 0.9734, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.288527364332141e-05, | |
| "loss": 0.9486, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.278912349525019e-05, | |
| "loss": 0.9358, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.269297334717896e-05, | |
| "loss": 0.9559, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.259682319910773e-05, | |
| "loss": 0.9637, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.25006730510365e-05, | |
| "loss": 0.9583, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.2404522902965275e-05, | |
| "loss": 0.9432, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.2308372754894046e-05, | |
| "loss": 0.9727, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.2212222606822816e-05, | |
| "loss": 0.9463, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.2116072458751586e-05, | |
| "loss": 0.9658, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.2019922310680364e-05, | |
| "loss": 0.974, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.1923772162609134e-05, | |
| "loss": 0.9577, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.1827622014537904e-05, | |
| "loss": 0.9679, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.1731471866466675e-05, | |
| "loss": 0.9531, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.163532171839545e-05, | |
| "loss": 0.9511, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.153917157032422e-05, | |
| "loss": 0.9703, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.144302142225299e-05, | |
| "loss": 0.9489, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.1346871274181763e-05, | |
| "loss": 0.9691, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.125072112611054e-05, | |
| "loss": 0.9512, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.115457097803931e-05, | |
| "loss": 0.9622, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.105842082996808e-05, | |
| "loss": 0.9493, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.096227068189685e-05, | |
| "loss": 0.9485, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.086612053382562e-05, | |
| "loss": 0.9555, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.076997038575439e-05, | |
| "loss": 0.9462, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.067382023768316e-05, | |
| "loss": 0.9637, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.057767008961194e-05, | |
| "loss": 0.9474, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.048151994154071e-05, | |
| "loss": 0.963, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.038536979346948e-05, | |
| "loss": 0.9462, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.0289219645398255e-05, | |
| "loss": 0.9531, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.0193069497327025e-05, | |
| "loss": 0.9338, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.00969193492558e-05, | |
| "loss": 0.9576, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.000076920118457e-05, | |
| "loss": 0.9486, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.9904619053113344e-05, | |
| "loss": 0.9636, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.9808468905042114e-05, | |
| "loss": 0.9499, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9712318756970888e-05, | |
| "loss": 0.9605, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9616168608899658e-05, | |
| "loss": 0.937, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.952001846082843e-05, | |
| "loss": 0.9469, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.9423868312757202e-05, | |
| "loss": 0.9519, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.9327718164685973e-05, | |
| "loss": 0.9548, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.9231568016614747e-05, | |
| "loss": 0.9411, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.9135417868543517e-05, | |
| "loss": 0.9549, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.903926772047229e-05, | |
| "loss": 0.9353, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.894311757240106e-05, | |
| "loss": 0.9543, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.8846967424329835e-05, | |
| "loss": 0.9484, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.8750817276258606e-05, | |
| "loss": 0.9491, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.865466712818738e-05, | |
| "loss": 0.9495, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.855851698011615e-05, | |
| "loss": 0.9383, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.8462366832044924e-05, | |
| "loss": 0.9243, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8366216683973694e-05, | |
| "loss": 0.9468, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.8270066535902468e-05, | |
| "loss": 0.946, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.817391638783124e-05, | |
| "loss": 0.9511, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.8077766239760012e-05, | |
| "loss": 0.9264, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.7981616091688783e-05, | |
| "loss": 0.9368, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.7885465943617556e-05, | |
| "loss": 0.9461, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.7789315795546327e-05, | |
| "loss": 0.9422, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.76931656474751e-05, | |
| "loss": 0.9478, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.759701549940387e-05, | |
| "loss": 0.9536, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.7500865351332645e-05, | |
| "loss": 0.9419, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.7404715203261415e-05, | |
| "loss": 0.9361, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.730856505519019e-05, | |
| "loss": 0.9397, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.721241490711896e-05, | |
| "loss": 0.9427, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.7116264759047733e-05, | |
| "loss": 0.9353, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.7020114610976504e-05, | |
| "loss": 0.9294, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.6923964462905278e-05, | |
| "loss": 0.9401, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.6827814314834048e-05, | |
| "loss": 0.9443, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.6731664166762815e-05, | |
| "loss": 0.9357, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.663551401869159e-05, | |
| "loss": 0.9508, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.653936387062036e-05, | |
| "loss": 0.9391, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.6443213722549133e-05, | |
| "loss": 0.9471, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.6347063574477904e-05, | |
| "loss": 0.9323, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.6250913426406677e-05, | |
| "loss": 0.9275, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.6154763278335448e-05, | |
| "loss": 0.9461, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.605861313026422e-05, | |
| "loss": 0.9252, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.5962462982192992e-05, | |
| "loss": 0.9392, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.5866312834121766e-05, | |
| "loss": 0.9449, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.5770162686050536e-05, | |
| "loss": 0.9354, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.567401253797931e-05, | |
| "loss": 0.9291, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.557786238990808e-05, | |
| "loss": 0.9141, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.548171224183685e-05, | |
| "loss": 0.9358, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.5385562093765625e-05, | |
| "loss": 0.9286, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.5289411945694395e-05, | |
| "loss": 0.9194, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.519326179762317e-05, | |
| "loss": 0.9386, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.509711164955194e-05, | |
| "loss": 0.9239, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.5000961501480713e-05, | |
| "loss": 0.9418, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.4904811353409484e-05, | |
| "loss": 0.9197, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.4808661205338257e-05, | |
| "loss": 0.9265, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.4712511057267028e-05, | |
| "loss": 0.9294, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.4616360909195802e-05, | |
| "loss": 0.9153, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.4520210761124572e-05, | |
| "loss": 0.9201, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.4424060613053346e-05, | |
| "loss": 0.9222, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.4327910464982116e-05, | |
| "loss": 0.928, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.423176031691089e-05, | |
| "loss": 0.9388, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.413561016883966e-05, | |
| "loss": 0.9244, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.4039460020768434e-05, | |
| "loss": 0.9152, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.3943309872697205e-05, | |
| "loss": 0.927, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.384715972462598e-05, | |
| "loss": 0.9179, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.375100957655475e-05, | |
| "loss": 0.9242, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.3654859428483523e-05, | |
| "loss": 0.9155, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.355870928041229e-05, | |
| "loss": 0.925, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.3462559132341064e-05, | |
| "loss": 0.9251, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.3366408984269834e-05, | |
| "loss": 0.929, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.3270258836198608e-05, | |
| "loss": 0.923, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.317410868812738e-05, | |
| "loss": 0.9419, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.3077958540056152e-05, | |
| "loss": 0.9305, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.2981808391984923e-05, | |
| "loss": 0.9138, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.2885658243913696e-05, | |
| "loss": 0.9151, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.2789508095842467e-05, | |
| "loss": 0.9246, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.269335794777124e-05, | |
| "loss": 0.9279, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.259720779970001e-05, | |
| "loss": 0.9066, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.2501057651628785e-05, | |
| "loss": 0.906, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.2404907503557555e-05, | |
| "loss": 0.9079, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.230875735548633e-05, | |
| "loss": 0.9161, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.22126072074151e-05, | |
| "loss": 0.9151, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.2116457059343873e-05, | |
| "loss": 0.9143, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.2020306911272644e-05, | |
| "loss": 0.9329, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.1924156763201418e-05, | |
| "loss": 0.9118, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.1828006615130188e-05, | |
| "loss": 0.9103, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.1731856467058962e-05, | |
| "loss": 0.9225, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.1635706318987732e-05, | |
| "loss": 0.9325, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1539556170916506e-05, | |
| "loss": 0.9267, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1443406022845273e-05, | |
| "loss": 0.916, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.1347255874774047e-05, | |
| "loss": 0.904, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.1251105726702817e-05, | |
| "loss": 0.9097, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.115495557863159e-05, | |
| "loss": 0.8927, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.1058805430560362e-05, | |
| "loss": 0.8965, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.0962655282489136e-05, | |
| "loss": 0.9112, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0866505134417906e-05, | |
| "loss": 0.9087, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.077035498634668e-05, | |
| "loss": 0.8964, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.067420483827545e-05, | |
| "loss": 0.905, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.0578054690204224e-05, | |
| "loss": 0.9104, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.0481904542132994e-05, | |
| "loss": 0.9009, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.0385754394061768e-05, | |
| "loss": 0.9134, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.028960424599054e-05, | |
| "loss": 0.8799, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0193454097919313e-05, | |
| "loss": 0.8913, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0097303949848083e-05, | |
| "loss": 0.8978, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.0001153801776857e-05, | |
| "loss": 0.9004, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.9905003653705627e-05, | |
| "loss": 0.9072, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.98088535056344e-05, | |
| "loss": 0.8987, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.971270335756317e-05, | |
| "loss": 0.8858, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.9616553209491945e-05, | |
| "loss": 0.898, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.9520403061420716e-05, | |
| "loss": 0.897, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.942425291334949e-05, | |
| "loss": 0.9073, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.9328102765278257e-05, | |
| "loss": 0.8768, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.923195261720703e-05, | |
| "loss": 0.8915, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.91358024691358e-05, | |
| "loss": 0.882, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.9039652321064575e-05, | |
| "loss": 0.9006, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.8943502172993345e-05, | |
| "loss": 0.8819, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.884735202492212e-05, | |
| "loss": 0.9125, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.875120187685089e-05, | |
| "loss": 0.8786, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.8655051728779663e-05, | |
| "loss": 0.8794, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.8558901580708433e-05, | |
| "loss": 0.8776, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.8462751432637207e-05, | |
| "loss": 0.898, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.8366601284565978e-05, | |
| "loss": 0.8917, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.827045113649475e-05, | |
| "loss": 0.9, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.8174300988423522e-05, | |
| "loss": 0.8923, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.8078150840352296e-05, | |
| "loss": 0.9107, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.7982000692281066e-05, | |
| "loss": 0.8976, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.788585054420984e-05, | |
| "loss": 0.9077, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.778970039613861e-05, | |
| "loss": 0.8855, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.7693550248067384e-05, | |
| "loss": 0.89, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.7597400099996155e-05, | |
| "loss": 0.9004, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.750124995192493e-05, | |
| "loss": 0.8804, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.74050998038537e-05, | |
| "loss": 0.9081, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.7308949655782473e-05, | |
| "loss": 0.9013, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.721279950771124e-05, | |
| "loss": 0.8874, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.7116649359640014e-05, | |
| "loss": 0.9167, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.7020499211568784e-05, | |
| "loss": 0.8841, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.6924349063497558e-05, | |
| "loss": 0.8982, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.6828198915426328e-05, | |
| "loss": 0.8789, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.6732048767355102e-05, | |
| "loss": 0.8933, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.6635898619283873e-05, | |
| "loss": 0.8458, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.6539748471212646e-05, | |
| "loss": 0.7751, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.6443598323141417e-05, | |
| "loss": 0.7772, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.634744817507019e-05, | |
| "loss": 0.7611, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.625129802699896e-05, | |
| "loss": 0.7745, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.6155147878927735e-05, | |
| "loss": 0.7699, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.6058997730856505e-05, | |
| "loss": 0.7727, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.596284758278528e-05, | |
| "loss": 0.7907, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.586669743471405e-05, | |
| "loss": 0.7675, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.5770547286642823e-05, | |
| "loss": 0.7707, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.5674397138571594e-05, | |
| "loss": 0.7717, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.5578246990500368e-05, | |
| "loss": 0.75, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.5482096842429138e-05, | |
| "loss": 0.7788, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.5385946694357912e-05, | |
| "loss": 0.7711, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.5289796546286682e-05, | |
| "loss": 0.768, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5193646398215456e-05, | |
| "loss": 0.7748, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5097496250144225e-05, | |
| "loss": 0.7892, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.5001346102072997e-05, | |
| "loss": 0.7742, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.4905195954001769e-05, | |
| "loss": 0.7751, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.4809045805930541e-05, | |
| "loss": 0.7782, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4712895657859313e-05, | |
| "loss": 0.7731, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4616745509788085e-05, | |
| "loss": 0.7739, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.4520595361716857e-05, | |
| "loss": 0.7677, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.442444521364563e-05, | |
| "loss": 0.7643, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.4328295065574402e-05, | |
| "loss": 0.7814, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.4232144917503174e-05, | |
| "loss": 0.7742, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.4135994769431946e-05, | |
| "loss": 0.7682, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.4039844621360718e-05, | |
| "loss": 0.7803, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.394369447328949e-05, | |
| "loss": 0.7869, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.3847544325218262e-05, | |
| "loss": 0.7779, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.3751394177147034e-05, | |
| "loss": 0.7697, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.3655244029075807e-05, | |
| "loss": 0.7836, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.3559093881004579e-05, | |
| "loss": 0.7681, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.346294373293335e-05, | |
| "loss": 0.7792, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.3366793584862123e-05, | |
| "loss": 0.7801, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.3270643436790895e-05, | |
| "loss": 0.7693, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3174493288719667e-05, | |
| "loss": 0.7708, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.307834314064844e-05, | |
| "loss": 0.7735, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.2982192992577208e-05, | |
| "loss": 0.7638, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.288604284450598e-05, | |
| "loss": 0.766, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.2789892696434752e-05, | |
| "loss": 0.7737, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.2693742548363524e-05, | |
| "loss": 0.7823, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.2597592400292297e-05, | |
| "loss": 0.7772, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.2501442252221069e-05, | |
| "loss": 0.7648, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.240529210414984e-05, | |
| "loss": 0.7803, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.2309141956078613e-05, | |
| "loss": 0.7615, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.2212991808007385e-05, | |
| "loss": 0.7523, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.2116841659936157e-05, | |
| "loss": 0.772, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.202069151186493e-05, | |
| "loss": 0.7659, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.1924541363793701e-05, | |
| "loss": 0.7674, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.1828391215722473e-05, | |
| "loss": 0.7698, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.1732241067651246e-05, | |
| "loss": 0.7626, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.1636090919580018e-05, | |
| "loss": 0.7697, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.153994077150879e-05, | |
| "loss": 0.7707, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.144379062343756e-05, | |
| "loss": 0.7798, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.1347640475366332e-05, | |
| "loss": 0.7816, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.1251490327295105e-05, | |
| "loss": 0.771, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.1155340179223877e-05, | |
| "loss": 0.7622, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.1059190031152649e-05, | |
| "loss": 0.7726, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.0963039883081421e-05, | |
| "loss": 0.7529, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.0866889735010193e-05, | |
| "loss": 0.7717, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.0770739586938965e-05, | |
| "loss": 0.7672, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.0674589438867737e-05, | |
| "loss": 0.7554, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.057843929079651e-05, | |
| "loss": 0.764, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.0482289142725281e-05, | |
| "loss": 0.7489, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.0386138994654052e-05, | |
| "loss": 0.7656, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.0289988846582824e-05, | |
| "loss": 0.7638, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0193838698511596e-05, | |
| "loss": 0.7664, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0097688550440368e-05, | |
| "loss": 0.7607, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.000153840236914e-05, | |
| "loss": 0.7649, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.905388254297913e-06, | |
| "loss": 0.7673, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.809238106226685e-06, | |
| "loss": 0.7509, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.713087958155457e-06, | |
| "loss": 0.7552, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.616937810084229e-06, | |
| "loss": 0.7632, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.520787662013001e-06, | |
| "loss": 0.7499, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.424637513941773e-06, | |
| "loss": 0.7441, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.328487365870544e-06, | |
| "loss": 0.7666, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.232337217799316e-06, | |
| "loss": 0.7629, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.136187069728088e-06, | |
| "loss": 0.76, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 9.04003692165686e-06, | |
| "loss": 0.7484, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 8.943886773585632e-06, | |
| "loss": 0.7486, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.847736625514404e-06, | |
| "loss": 0.769, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.751586477443176e-06, | |
| "loss": 0.7488, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 8.655436329371948e-06, | |
| "loss": 0.7639, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.55928618130072e-06, | |
| "loss": 0.7822, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.463136033229493e-06, | |
| "loss": 0.7591, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.366985885158265e-06, | |
| "loss": 0.7733, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.270835737087035e-06, | |
| "loss": 0.7594, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.174685589015807e-06, | |
| "loss": 0.7451, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 8.07853544094458e-06, | |
| "loss": 0.7477, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 7.982385292873352e-06, | |
| "loss": 0.7584, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.886235144802124e-06, | |
| "loss": 0.7598, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.790084996730896e-06, | |
| "loss": 0.7593, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.693934848659668e-06, | |
| "loss": 0.7528, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.597784700588439e-06, | |
| "loss": 0.7416, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.501634552517211e-06, | |
| "loss": 0.7163, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.4054844044459834e-06, | |
| "loss": 0.7674, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.3093342563747556e-06, | |
| "loss": 0.7475, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.213184108303527e-06, | |
| "loss": 0.7462, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.117033960232299e-06, | |
| "loss": 0.7629, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.020883812161071e-06, | |
| "loss": 0.7414, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 6.924733664089842e-06, | |
| "loss": 0.7661, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 6.8285835160186145e-06, | |
| "loss": 0.7772, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.732433367947387e-06, | |
| "loss": 0.7511, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.636283219876159e-06, | |
| "loss": 0.7655, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.540133071804931e-06, | |
| "loss": 0.7523, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.443982923733703e-06, | |
| "loss": 0.7469, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 6.347832775662475e-06, | |
| "loss": 0.7464, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 6.251682627591247e-06, | |
| "loss": 0.7576, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.1555324795200185e-06, | |
| "loss": 0.749, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 6.059382331448791e-06, | |
| "loss": 0.738, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 5.963232183377563e-06, | |
| "loss": 0.7549, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.867082035306334e-06, | |
| "loss": 0.7563, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.770931887235106e-06, | |
| "loss": 0.7334, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.674781739163878e-06, | |
| "loss": 0.7488, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.57863159109265e-06, | |
| "loss": 0.7462, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.4824814430214225e-06, | |
| "loss": 0.7757, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.386331294950194e-06, | |
| "loss": 0.7534, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.290181146878966e-06, | |
| "loss": 0.7527, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.194030998807738e-06, | |
| "loss": 0.7371, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.09788085073651e-06, | |
| "loss": 0.7477, | |
| "step": 233500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 5.001730702665282e-06, | |
| "loss": 0.7467, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.905580554594054e-06, | |
| "loss": 0.7448, | |
| "step": 234500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.809430406522826e-06, | |
| "loss": 0.7473, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.713280258451598e-06, | |
| "loss": 0.7498, | |
| "step": 235500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.61713011038037e-06, | |
| "loss": 0.73, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.520979962309142e-06, | |
| "loss": 0.7356, | |
| "step": 236500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.424829814237914e-06, | |
| "loss": 0.7358, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.328679666166685e-06, | |
| "loss": 0.7377, | |
| "step": 237500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.2325295180954575e-06, | |
| "loss": 0.7381, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.13637937002423e-06, | |
| "loss": 0.7542, | |
| "step": 238500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.040229221953002e-06, | |
| "loss": 0.7445, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 3.944079073881774e-06, | |
| "loss": 0.7493, | |
| "step": 239500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.847928925810546e-06, | |
| "loss": 0.7251, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.7517787777393177e-06, | |
| "loss": 0.7538, | |
| "step": 240500 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 3.65562862966809e-06, | |
| "loss": 0.7415, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.559478481596862e-06, | |
| "loss": 0.7514, | |
| "step": 241500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.463328333525634e-06, | |
| "loss": 0.7354, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.367178185454406e-06, | |
| "loss": 0.7417, | |
| "step": 242500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.2710280373831774e-06, | |
| "loss": 0.7465, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.1748778893119496e-06, | |
| "loss": 0.7365, | |
| "step": 243500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 3.0787277412407217e-06, | |
| "loss": 0.7406, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 2.982577593169494e-06, | |
| "loss": 0.7478, | |
| "step": 244500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.8864274450982655e-06, | |
| "loss": 0.7535, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.7902772970270376e-06, | |
| "loss": 0.7387, | |
| "step": 245500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.6941271489558097e-06, | |
| "loss": 0.7416, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.5979770008845814e-06, | |
| "loss": 0.7439, | |
| "step": 246500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.5018268528133536e-06, | |
| "loss": 0.7606, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.4056767047421257e-06, | |
| "loss": 0.7316, | |
| "step": 247500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.3095265566708974e-06, | |
| "loss": 0.7409, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.2133764085996695e-06, | |
| "loss": 0.7414, | |
| "step": 248500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.1172262605284416e-06, | |
| "loss": 0.7496, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 2.0210761124572133e-06, | |
| "loss": 0.7389, | |
| "step": 249500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 1.9249259643859854e-06, | |
| "loss": 0.7486, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.8287758163147571e-06, | |
| "loss": 0.7439, | |
| "step": 250500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.732625668243529e-06, | |
| "loss": 0.7463, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.6364755201723012e-06, | |
| "loss": 0.739, | |
| "step": 251500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.540325372101073e-06, | |
| "loss": 0.7424, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.444175224029845e-06, | |
| "loss": 0.7398, | |
| "step": 252500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.348025075958617e-06, | |
| "loss": 0.7307, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.251874927887389e-06, | |
| "loss": 0.7204, | |
| "step": 253500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.155724779816161e-06, | |
| "loss": 0.7299, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1.0595746317449328e-06, | |
| "loss": 0.7319, | |
| "step": 254500 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 9.634244836737047e-07, | |
| "loss": 0.7451, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 8.67274335602477e-07, | |
| "loss": 0.7379, | |
| "step": 255500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 7.711241875312489e-07, | |
| "loss": 0.7209, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 6.749740394600208e-07, | |
| "loss": 0.7311, | |
| "step": 256500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 5.788238913887928e-07, | |
| "loss": 0.7405, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.826737433175648e-07, | |
| "loss": 0.7318, | |
| "step": 257500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.865235952463367e-07, | |
| "loss": 0.7411, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 2.9037344717510864e-07, | |
| "loss": 0.7262, | |
| "step": 258500 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 1.9422329910388063e-07, | |
| "loss": 0.7422, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 9.80731510326526e-08, | |
| "loss": 0.7411, | |
| "step": 259500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 1.9230029614245606e-09, | |
| "loss": 0.7414, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 260010, | |
| "total_flos": 4.586582510270546e+17, | |
| "train_loss": 0.9812617931965292, | |
| "train_runtime": 72588.7542, | |
| "train_samples_per_second": 35.82, | |
| "train_steps_per_second": 3.582 | |
| } | |
| ], | |
| "max_steps": 260010, | |
| "num_train_epochs": 3, | |
| "total_flos": 4.586582510270546e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |