| { | |
| "best_metric": 69.66985602287733, | |
| "best_model_checkpoint": "/data/jcanete/all_results/tar/distillbeto/epochs_4_bs_16_lr_5e-5/checkpoint-16800", | |
| "epoch": 4.0, | |
| "global_step": 21932, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "eval_exact_match": 26.310312204351938, | |
| "eval_f1": 40.606556876208344, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.886695239832209e-05, | |
| "loss": 3.3838, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "eval_exact_match": 35.042573320719015, | |
| "eval_f1": 51.1874637199927, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "eval_exact_match": 37.59697256385998, | |
| "eval_f1": 55.70379412517118, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.772706547510487e-05, | |
| "loss": 2.6487, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "eval_exact_match": 40.30274361400189, | |
| "eval_f1": 57.845372827145646, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.658717855188766e-05, | |
| "loss": 2.4073, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "eval_exact_match": 41.5042573320719, | |
| "eval_f1": 59.344242314747255, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "eval_exact_match": 43.528855250709555, | |
| "eval_f1": 60.57412156026134, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.544729162867044e-05, | |
| "loss": 2.3006, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "eval_exact_match": 43.85998107852412, | |
| "eval_f1": 60.74218457977161, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "eval_exact_match": 45.32639545884579, | |
| "eval_f1": 62.60100745933504, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.4309684479299654e-05, | |
| "loss": 2.2202, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "eval_exact_match": 45.87511825922422, | |
| "eval_f1": 62.66596281773377, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.3169797556082435e-05, | |
| "loss": 2.154, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "eval_exact_match": 45.53453169347209, | |
| "eval_f1": 62.82975014751129, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_exact_match": 46.395458845789975, | |
| "eval_f1": 63.47874761462123, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.202991063286522e-05, | |
| "loss": 2.1242, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "eval_exact_match": 47.44560075685904, | |
| "eval_f1": 64.43571994761247, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "eval_exact_match": 47.62535477767266, | |
| "eval_f1": 65.01568330950501, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.0890023709648004e-05, | |
| "loss": 2.0919, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "eval_exact_match": 48.58088930936613, | |
| "eval_f1": 65.844186710796, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.9750136786430785e-05, | |
| "loss": 2.0242, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "eval_exact_match": 48.618732261116364, | |
| "eval_f1": 65.93288102675687, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "eval_exact_match": 47.62535477767266, | |
| "eval_f1": 65.07422958403791, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.861252963706001e-05, | |
| "loss": 2.0193, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "eval_exact_match": 49.09176915799432, | |
| "eval_f1": 66.2146996747605, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "eval_exact_match": 49.26206244087039, | |
| "eval_f1": 67.13615087977189, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.7474922487689224e-05, | |
| "loss": 1.9651, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "eval_exact_match": 49.92431409649953, | |
| "eval_f1": 66.56942877421614, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.633503556447201e-05, | |
| "loss": 1.6733, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "eval_exact_match": 50.08514664143803, | |
| "eval_f1": 66.7344480216334, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "eval_exact_match": 49.85808893093661, | |
| "eval_f1": 67.17831265861523, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.5195148641254786e-05, | |
| "loss": 1.7062, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "eval_exact_match": 50.07568590350047, | |
| "eval_f1": 67.34676690188185, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "eval_exact_match": 49.92431409649953, | |
| "eval_f1": 66.86831954211718, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.4055261718037574e-05, | |
| "loss": 1.6685, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "eval_exact_match": 50.3689687795648, | |
| "eval_f1": 67.58117044489109, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 3.2915374794820355e-05, | |
| "loss": 1.683, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "eval_exact_match": 51.116367076631974, | |
| "eval_f1": 67.65638675710437, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "eval_exact_match": 50.879848628193, | |
| "eval_f1": 67.74322775380313, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 3.177776764544958e-05, | |
| "loss": 1.6271, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "eval_exact_match": 51.14474929044466, | |
| "eval_f1": 67.41847561216078, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "eval_exact_match": 50.90823084200568, | |
| "eval_f1": 67.83244584175503, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 3.063788072223235e-05, | |
| "loss": 1.6509, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "eval_exact_match": 51.485335856196784, | |
| "eval_f1": 68.03877774930996, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.949799379901514e-05, | |
| "loss": 1.6591, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "eval_exact_match": 51.390728476821195, | |
| "eval_f1": 68.12396413437372, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "eval_exact_match": 51.91106906338695, | |
| "eval_f1": 68.28274234884077, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.8358106875797924e-05, | |
| "loss": 1.6597, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "eval_exact_match": 52.147587511825925, | |
| "eval_f1": 68.44926769664524, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "eval_exact_match": 51.920529801324506, | |
| "eval_f1": 68.25478495844789, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 2.7218219952580705e-05, | |
| "loss": 1.6526, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "eval_exact_match": 51.88268684957427, | |
| "eval_f1": 68.59555473771334, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 2.607833302936349e-05, | |
| "loss": 1.6083, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "eval_exact_match": 52.223273415326396, | |
| "eval_f1": 68.3894855383881, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "eval_exact_match": 52.04351939451277, | |
| "eval_f1": 68.85931291757578, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 2.493844610614627e-05, | |
| "loss": 1.5584, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "eval_exact_match": 51.41911069063387, | |
| "eval_f1": 68.45020620158613, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "eval_exact_match": 51.42857142857143, | |
| "eval_f1": 68.42914280017077, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 2.3798559182929054e-05, | |
| "loss": 1.2913, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "eval_exact_match": 51.788079470198674, | |
| "eval_f1": 68.62435982099952, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 2.265867225971184e-05, | |
| "loss": 1.2903, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "eval_exact_match": 51.83538315988647, | |
| "eval_f1": 68.70784934153268, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "eval_exact_match": 52.29895931882687, | |
| "eval_f1": 68.91609842633096, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 2.1521065110341055e-05, | |
| "loss": 1.3203, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "eval_exact_match": 51.9678334910123, | |
| "eval_f1": 68.89606884437242, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "eval_exact_match": 52.21381267738884, | |
| "eval_f1": 68.5166390817109, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.0383457960970274e-05, | |
| "loss": 1.2923, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "eval_exact_match": 52.11920529801324, | |
| "eval_f1": 68.80154847964492, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 1.9243571037753055e-05, | |
| "loss": 1.3007, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "eval_exact_match": 51.93945127719962, | |
| "eval_f1": 68.56070274331668, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "eval_exact_match": 52.450331125827816, | |
| "eval_f1": 68.92160672281842, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 1.8103684114535836e-05, | |
| "loss": 1.3022, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "eval_exact_match": 52.081362346263006, | |
| "eval_f1": 68.734639590144, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "eval_exact_match": 52.56385998107852, | |
| "eval_f1": 69.14264904059371, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 1.6963797191318624e-05, | |
| "loss": 1.2735, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "eval_exact_match": 52.65846736045412, | |
| "eval_f1": 69.27808981447744, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 1.5823910268101405e-05, | |
| "loss": 1.3057, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "eval_exact_match": 52.37464522232734, | |
| "eval_f1": 69.28041873050952, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "eval_exact_match": 52.639545884579, | |
| "eval_f1": 69.10864509041865, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1.4684023344884188e-05, | |
| "loss": 1.2689, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "eval_exact_match": 52.157048249763484, | |
| "eval_f1": 68.73617410296806, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "eval_exact_match": 52.58278145695364, | |
| "eval_f1": 69.30018384216147, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.3544136421666972e-05, | |
| "loss": 1.2489, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "eval_exact_match": 52.639545884579, | |
| "eval_f1": 69.01439480212419, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 1.2406529272296188e-05, | |
| "loss": 1.2788, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "eval_exact_match": 52.61116367076632, | |
| "eval_f1": 69.30278936364587, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "eval_exact_match": 53.074739829706715, | |
| "eval_f1": 69.66985602287733, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 1.1266642349078972e-05, | |
| "loss": 1.0115, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "eval_exact_match": 52.12866603595081, | |
| "eval_f1": 68.93516419404564, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "eval_exact_match": 52.544938505203405, | |
| "eval_f1": 68.96323841158566, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 1.0126755425861755e-05, | |
| "loss": 1.0276, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "eval_exact_match": 52.49763481551561, | |
| "eval_f1": 69.07220845517458, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 8.986868502644538e-06, | |
| "loss": 1.0218, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "eval_exact_match": 52.29895931882687, | |
| "eval_f1": 69.02811306482906, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "eval_exact_match": 52.639545884579, | |
| "eval_f1": 69.2185250484871, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 7.846981579427322e-06, | |
| "loss": 1.0386, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "eval_exact_match": 52.459791863765375, | |
| "eval_f1": 69.36180052659964, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "eval_exact_match": 52.57332071901608, | |
| "eval_f1": 69.30319952546826, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 6.707094656210104e-06, | |
| "loss": 1.0086, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "eval_exact_match": 52.47871333964049, | |
| "eval_f1": 69.30902776116545, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 5.5694875068393215e-06, | |
| "loss": 1.0143, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "eval_exact_match": 52.28949858088931, | |
| "eval_f1": 69.21311662285441, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "eval_exact_match": 52.61116367076632, | |
| "eval_f1": 69.32649279145429, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 4.429600583622105e-06, | |
| "loss": 1.0071, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "eval_exact_match": 52.544938505203405, | |
| "eval_f1": 68.97807013082176, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "eval_exact_match": 52.5922421948912, | |
| "eval_f1": 69.18859635059812, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 3.289713660404888e-06, | |
| "loss": 1.0236, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "eval_exact_match": 52.55439924314096, | |
| "eval_f1": 69.08256750898333, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 2.149826737187671e-06, | |
| "loss": 1.0015, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "eval_exact_match": 52.70577105014191, | |
| "eval_f1": 69.22107041576008, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "eval_exact_match": 52.73415326395459, | |
| "eval_f1": 69.36938667664833, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 1.0099398139704543e-06, | |
| "loss": 1.0189, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "eval_exact_match": 52.63008514664144, | |
| "eval_f1": 69.28051735265512, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "eval_exact_match": 52.71523178807947, | |
| "eval_f1": 69.35772325822661, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 21932, | |
| "total_flos": 2.986644942216499e+16, | |
| "train_loss": 1.5664932693461966, | |
| "train_runtime": 16788.5182, | |
| "train_samples_per_second": 20.902, | |
| "train_steps_per_second": 1.306 | |
| } | |
| ], | |
| "max_steps": 21932, | |
| "num_train_epochs": 4, | |
| "total_flos": 2.986644942216499e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |