| { | |
| "best_metric": 66.26645652732091, | |
| "best_model_checkpoint": "/data/jcanete/all_results/tar/albeto_tiny/epochs_4_bs_16_lr_5e-5/checkpoint-10800", | |
| "epoch": 4.0, | |
| "global_step": 21940, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "eval_exact_match": 25.279091769157993, | |
| "eval_f1": 41.71460503665182, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.8867365542388335e-05, | |
| "loss": 3.1064, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "eval_exact_match": 33.77483443708609, | |
| "eval_f1": 51.462032531677195, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "eval_exact_match": 37.78618732261116, | |
| "eval_f1": 55.147295523401176, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.7727894257064726e-05, | |
| "loss": 2.4193, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "eval_exact_match": 39.735099337748345, | |
| "eval_f1": 57.70943484117446, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.6588422971741116e-05, | |
| "loss": 2.2494, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "eval_exact_match": 40.69063386944182, | |
| "eval_f1": 58.161449729726456, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "eval_exact_match": 41.72185430463576, | |
| "eval_f1": 59.240651392335586, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.544895168641751e-05, | |
| "loss": 2.1889, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "eval_exact_match": 41.627246925260174, | |
| "eval_f1": 60.05133975647362, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "eval_exact_match": 43.33964049195837, | |
| "eval_f1": 61.334205952553624, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.431175934366454e-05, | |
| "loss": 2.0954, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "eval_exact_match": 44.1438032166509, | |
| "eval_f1": 61.6109819191811, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.317228805834093e-05, | |
| "loss": 2.0585, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "eval_exact_match": 44.55061494796594, | |
| "eval_f1": 62.74097437748746, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_exact_match": 45.14664143803217, | |
| "eval_f1": 62.716820973684904, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.203281677301732e-05, | |
| "loss": 2.0252, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "eval_exact_match": 45.22232734153264, | |
| "eval_f1": 63.111925308146255, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "eval_exact_match": 45.13718070009461, | |
| "eval_f1": 63.12474940531198, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.0893345487693714e-05, | |
| "loss": 1.9861, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "eval_exact_match": 46.21570482497635, | |
| "eval_f1": 64.00817566022319, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.975615314494075e-05, | |
| "loss": 1.9881, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "eval_exact_match": 46.72658467360454, | |
| "eval_f1": 64.45739871415881, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "eval_exact_match": 46.31031220435194, | |
| "eval_f1": 64.03851598846974, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.861668185961714e-05, | |
| "loss": 1.9619, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "eval_exact_match": 46.57521286660359, | |
| "eval_f1": 64.71289725300291, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "eval_exact_match": 46.80227057710501, | |
| "eval_f1": 64.89896140882486, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.747721057429353e-05, | |
| "loss": 1.8961, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "eval_exact_match": 46.868495742667925, | |
| "eval_f1": 64.9566553926742, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.633773928896992e-05, | |
| "loss": 1.6613, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "eval_exact_match": 47.05771050141911, | |
| "eval_f1": 65.16060187033897, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "eval_exact_match": 47.010406811731315, | |
| "eval_f1": 64.63527459887219, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.520054694621696e-05, | |
| "loss": 1.6724, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "eval_exact_match": 46.44276253547777, | |
| "eval_f1": 64.6485660215026, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "eval_exact_match": 47.02932828760643, | |
| "eval_f1": 65.25056698142618, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.406107566089335e-05, | |
| "loss": 1.6622, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "eval_exact_match": 47.38883632923368, | |
| "eval_f1": 65.14765281416761, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 3.292160437556974e-05, | |
| "loss": 1.6765, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "eval_exact_match": 47.52128666035951, | |
| "eval_f1": 65.23959133286817, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "eval_exact_match": 47.24692526017029, | |
| "eval_f1": 64.89498224105805, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 3.178213309024613e-05, | |
| "loss": 1.6831, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "eval_exact_match": 47.42667928098392, | |
| "eval_f1": 65.43419870447086, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "eval_exact_match": 47.237464522232735, | |
| "eval_f1": 65.29424940770944, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 3.0644940747493164e-05, | |
| "loss": 1.6652, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "eval_exact_match": 47.74834437086093, | |
| "eval_f1": 65.62503101399187, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.9505469462169554e-05, | |
| "loss": 1.6622, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "eval_exact_match": 48.04162724692526, | |
| "eval_f1": 66.13531062019833, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "eval_exact_match": 47.6631977294229, | |
| "eval_f1": 65.69495765526571, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.836599817684594e-05, | |
| "loss": 1.6282, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "eval_exact_match": 48.15515610217597, | |
| "eval_f1": 65.85406198831797, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "eval_exact_match": 48.34437086092715, | |
| "eval_f1": 65.82648855586231, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 2.7226526891522335e-05, | |
| "loss": 1.6311, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "eval_exact_match": 48.17407757805109, | |
| "eval_f1": 65.95821129767383, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 2.6089334548769374e-05, | |
| "loss": 1.6879, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "eval_exact_match": 48.070009460737936, | |
| "eval_f1": 65.61674709032437, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "eval_exact_match": 48.33491012298959, | |
| "eval_f1": 66.26645652732091, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.4949863263445765e-05, | |
| "loss": 1.6127, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "eval_exact_match": 47.776726584673604, | |
| "eval_f1": 65.67716714026636, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "eval_exact_match": 46.97256385998108, | |
| "eval_f1": 65.35488377881, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 2.3810391978122152e-05, | |
| "loss": 1.3442, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "eval_exact_match": 47.25638599810785, | |
| "eval_f1": 65.11588968827414, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 2.2670920692798542e-05, | |
| "loss": 1.3419, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "eval_exact_match": 47.918637653736994, | |
| "eval_f1": 65.76384490998669, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "eval_exact_match": 47.71996215704825, | |
| "eval_f1": 65.51523648050365, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 2.1533728350045578e-05, | |
| "loss": 1.3392, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "eval_exact_match": 47.81456953642384, | |
| "eval_f1": 65.72328974412012, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "eval_exact_match": 46.868495742667925, | |
| "eval_f1": 64.95043374963731, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.0396536007292617e-05, | |
| "loss": 1.3365, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "eval_exact_match": 47.47398297067171, | |
| "eval_f1": 65.46687084562042, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 1.9257064721969007e-05, | |
| "loss": 1.3863, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "eval_exact_match": 47.64427625354778, | |
| "eval_f1": 65.54344572464974, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "eval_exact_match": 47.70104068117313, | |
| "eval_f1": 65.29014560945414, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 1.8117593436645398e-05, | |
| "loss": 1.364, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "eval_exact_match": 47.918637653736994, | |
| "eval_f1": 65.66170183373373, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "eval_exact_match": 48.33491012298959, | |
| "eval_f1": 65.94289128043252, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 1.697812215132179e-05, | |
| "loss": 1.3864, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "eval_exact_match": 48.24976348155156, | |
| "eval_f1": 65.78253411845539, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 1.583865086599818e-05, | |
| "loss": 1.3636, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "eval_exact_match": 48.438978240302745, | |
| "eval_f1": 66.07660779936401, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "eval_exact_match": 48.16461684011353, | |
| "eval_f1": 66.2498588216324, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1.469917958067457e-05, | |
| "loss": 1.3726, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "eval_exact_match": 47.95648060548723, | |
| "eval_f1": 65.7260651497257, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "eval_exact_match": 47.88079470198676, | |
| "eval_f1": 66.01379305937202, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.355970829535096e-05, | |
| "loss": 1.379, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "eval_exact_match": 48.240302743614, | |
| "eval_f1": 65.53479147043561, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 1.2422515952597995e-05, | |
| "loss": 1.3401, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "eval_exact_match": 48.221381267738884, | |
| "eval_f1": 65.86023587215384, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "eval_exact_match": 47.615894039735096, | |
| "eval_f1": 65.12853949917962, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 1.1283044667274386e-05, | |
| "loss": 1.1281, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "eval_exact_match": 47.12393566698202, | |
| "eval_f1": 65.15991467915553, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "eval_exact_match": 47.21854304635762, | |
| "eval_f1": 65.02752189426162, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 1.0143573381950776e-05, | |
| "loss": 1.1161, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "eval_exact_match": 47.13339640491959, | |
| "eval_f1": 65.2429493800908, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 9.004102096627165e-06, | |
| "loss": 1.1337, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "eval_exact_match": 47.010406811731315, | |
| "eval_f1": 65.03577782296541, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "eval_exact_match": 47.379375591296125, | |
| "eval_f1": 65.13654689416043, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 7.864630811303556e-06, | |
| "loss": 1.1477, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "eval_exact_match": 47.35099337748344, | |
| "eval_f1": 64.86398318568581, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "eval_exact_match": 47.30368968779565, | |
| "eval_f1": 65.05463221900013, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 6.725159525979946e-06, | |
| "loss": 1.1414, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "eval_exact_match": 46.97256385998108, | |
| "eval_f1": 65.11680930447929, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 5.587967183226983e-06, | |
| "loss": 1.1454, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "eval_exact_match": 47.13339640491959, | |
| "eval_f1": 65.165125839751, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "eval_exact_match": 47.20908230842006, | |
| "eval_f1": 65.09005212297234, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 4.448495897903373e-06, | |
| "loss": 1.1318, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "eval_exact_match": 46.95364238410596, | |
| "eval_f1": 65.02854138983005, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "eval_exact_match": 46.773888363292336, | |
| "eval_f1": 64.98586781544525, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 3.3090246125797635e-06, | |
| "loss": 1.124, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "eval_exact_match": 46.76442762535478, | |
| "eval_f1": 65.02603195736303, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 2.169553327256153e-06, | |
| "loss": 1.1301, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "eval_exact_match": 46.88741721854305, | |
| "eval_f1": 65.03187807653661, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "eval_exact_match": 47.086092715231786, | |
| "eval_f1": 65.13748531716237, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 1.0323609845031905e-06, | |
| "loss": 1.1298, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "eval_exact_match": 47.095553453169344, | |
| "eval_f1": 65.03629129514118, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "eval_exact_match": 47.095553453169344, | |
| "eval_f1": 65.08861259333764, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 21940, | |
| "total_flos": 859515753161088.0, | |
| "train_loss": 1.5832800059288115, | |
| "train_runtime": 4642.7737, | |
| "train_samples_per_second": 75.599, | |
| "train_steps_per_second": 4.726 | |
| } | |
| ], | |
| "max_steps": 21940, | |
| "num_train_epochs": 4, | |
| "total_flos": 859515753161088.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |