| { | |
| "best_global_step": 12292, | |
| "best_metric": 0.9563834240267894, | |
| "best_model_checkpoint": "C:\\Users\\Shara\\projects\\models\\run_16-lr_2e-05-acc_1-wd_0.01-bs_8-ep_7\\checkpoint-12292", | |
| "epoch": 7.0, | |
| "eval_steps": 500, | |
| "global_step": 12292, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02847380410022779, | |
| "grad_norm": 4.8302321434021, | |
| "learning_rate": 1.992840872111943e-05, | |
| "loss": 0.9994, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05694760820045558, | |
| "grad_norm": 3.4978649616241455, | |
| "learning_rate": 1.9847054995118778e-05, | |
| "loss": 0.3939, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08542141230068337, | |
| "grad_norm": 8.353672981262207, | |
| "learning_rate": 1.9765701269118128e-05, | |
| "loss": 0.2882, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.11389521640091116, | |
| "grad_norm": 9.822871208190918, | |
| "learning_rate": 1.9684347543117477e-05, | |
| "loss": 0.1728, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.14236902050113895, | |
| "grad_norm": 3.1411221027374268, | |
| "learning_rate": 1.9602993817116827e-05, | |
| "loss": 0.1251, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.17084282460136674, | |
| "grad_norm": 8.493424415588379, | |
| "learning_rate": 1.9521640091116173e-05, | |
| "loss": 0.1405, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.19931662870159453, | |
| "grad_norm": 4.757007122039795, | |
| "learning_rate": 1.9440286365115523e-05, | |
| "loss": 0.1431, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.22779043280182232, | |
| "grad_norm": 4.339049816131592, | |
| "learning_rate": 1.9358932639114873e-05, | |
| "loss": 0.0912, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.25626423690205014, | |
| "grad_norm": 14.540772438049316, | |
| "learning_rate": 1.9277578913114222e-05, | |
| "loss": 0.1048, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.2847380410022779, | |
| "grad_norm": 10.503073692321777, | |
| "learning_rate": 1.9196225187113572e-05, | |
| "loss": 0.1224, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3132118451025057, | |
| "grad_norm": 2.738131284713745, | |
| "learning_rate": 1.911487146111292e-05, | |
| "loss": 0.1025, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.3416856492027335, | |
| "grad_norm": 6.681102752685547, | |
| "learning_rate": 1.903351773511227e-05, | |
| "loss": 0.0869, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.3701594533029613, | |
| "grad_norm": 1.4330339431762695, | |
| "learning_rate": 1.895216400911162e-05, | |
| "loss": 0.052, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.39863325740318906, | |
| "grad_norm": 2.581470012664795, | |
| "learning_rate": 1.8870810283110967e-05, | |
| "loss": 0.0808, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.4271070615034169, | |
| "grad_norm": 5.825100898742676, | |
| "learning_rate": 1.8789456557110317e-05, | |
| "loss": 0.0716, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.45558086560364464, | |
| "grad_norm": 1.74166738986969, | |
| "learning_rate": 1.8708102831109666e-05, | |
| "loss": 0.0806, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.48405466970387245, | |
| "grad_norm": 7.3097405433654785, | |
| "learning_rate": 1.8626749105109016e-05, | |
| "loss": 0.0666, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.5125284738041003, | |
| "grad_norm": 2.220766067504883, | |
| "learning_rate": 1.8545395379108362e-05, | |
| "loss": 0.0768, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.541002277904328, | |
| "grad_norm": 4.0566511154174805, | |
| "learning_rate": 1.8464041653107715e-05, | |
| "loss": 0.0707, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.5694760820045558, | |
| "grad_norm": 0.2833240330219269, | |
| "learning_rate": 1.8382687927107065e-05, | |
| "loss": 0.0858, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5979498861047836, | |
| "grad_norm": 21.131481170654297, | |
| "learning_rate": 1.8301334201106415e-05, | |
| "loss": 0.0595, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.6264236902050114, | |
| "grad_norm": 27.655433654785156, | |
| "learning_rate": 1.821998047510576e-05, | |
| "loss": 0.0466, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.6548974943052391, | |
| "grad_norm": 3.4721717834472656, | |
| "learning_rate": 1.813862674910511e-05, | |
| "loss": 0.0776, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.683371298405467, | |
| "grad_norm": 2.4811441898345947, | |
| "learning_rate": 1.805727302310446e-05, | |
| "loss": 0.0647, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.7118451025056948, | |
| "grad_norm": 6.169066429138184, | |
| "learning_rate": 1.797591929710381e-05, | |
| "loss": 0.0724, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.7403189066059226, | |
| "grad_norm": 8.984748840332031, | |
| "learning_rate": 1.7894565571103156e-05, | |
| "loss": 0.0838, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.7687927107061503, | |
| "grad_norm": 5.944505214691162, | |
| "learning_rate": 1.7813211845102506e-05, | |
| "loss": 0.0683, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.7972665148063781, | |
| "grad_norm": 0.28812381625175476, | |
| "learning_rate": 1.7731858119101856e-05, | |
| "loss": 0.0614, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.8257403189066059, | |
| "grad_norm": 6.176011085510254, | |
| "learning_rate": 1.7650504393101205e-05, | |
| "loss": 0.0615, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.8542141230068337, | |
| "grad_norm": 2.5244405269622803, | |
| "learning_rate": 1.7569150667100555e-05, | |
| "loss": 0.0714, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.8826879271070615, | |
| "grad_norm": 3.3508074283599854, | |
| "learning_rate": 1.7487796941099904e-05, | |
| "loss": 0.0563, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.9111617312072893, | |
| "grad_norm": 9.980842590332031, | |
| "learning_rate": 1.7406443215099254e-05, | |
| "loss": 0.0647, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.9396355353075171, | |
| "grad_norm": 3.6959774494171143, | |
| "learning_rate": 1.7325089489098604e-05, | |
| "loss": 0.0682, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.9681093394077449, | |
| "grad_norm": 0.626620888710022, | |
| "learning_rate": 1.724373576309795e-05, | |
| "loss": 0.0499, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.9965831435079726, | |
| "grad_norm": 7.145672798156738, | |
| "learning_rate": 1.71623820370973e-05, | |
| "loss": 0.0569, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_f1": 0.9242246437552389, | |
| "eval_loss": 0.05600914731621742, | |
| "eval_precision": 0.9206746826987308, | |
| "eval_recall": 0.927802086839448, | |
| "eval_runtime": 3.7538, | |
| "eval_samples_per_second": 865.798, | |
| "eval_steps_per_second": 108.425, | |
| "step": 1756 | |
| }, | |
| { | |
| "epoch": 1.0250569476082005, | |
| "grad_norm": 4.280463695526123, | |
| "learning_rate": 1.708102831109665e-05, | |
| "loss": 0.0673, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.0535307517084282, | |
| "grad_norm": 0.06692535430192947, | |
| "learning_rate": 1.6999674585096e-05, | |
| "loss": 0.0403, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.082004555808656, | |
| "grad_norm": 0.03789375349879265, | |
| "learning_rate": 1.691832085909535e-05, | |
| "loss": 0.0496, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.1104783599088839, | |
| "grad_norm": 3.698348045349121, | |
| "learning_rate": 1.6836967133094698e-05, | |
| "loss": 0.0381, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.1389521640091116, | |
| "grad_norm": 2.753032684326172, | |
| "learning_rate": 1.6755613407094048e-05, | |
| "loss": 0.0432, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.1674259681093395, | |
| "grad_norm": 0.06336130946874619, | |
| "learning_rate": 1.6674259681093398e-05, | |
| "loss": 0.0482, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.1958997722095672, | |
| "grad_norm": 3.9503254890441895, | |
| "learning_rate": 1.6592905955092744e-05, | |
| "loss": 0.0436, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.224373576309795, | |
| "grad_norm": 0.08679741621017456, | |
| "learning_rate": 1.6511552229092093e-05, | |
| "loss": 0.0346, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.2528473804100229, | |
| "grad_norm": 2.1995127201080322, | |
| "learning_rate": 1.6430198503091443e-05, | |
| "loss": 0.0312, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.2813211845102506, | |
| "grad_norm": 5.7709808349609375, | |
| "learning_rate": 1.634884477709079e-05, | |
| "loss": 0.032, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.3097949886104785, | |
| "grad_norm": 0.2031829059123993, | |
| "learning_rate": 1.626749105109014e-05, | |
| "loss": 0.053, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.3382687927107062, | |
| "grad_norm": 4.292157173156738, | |
| "learning_rate": 1.6186137325089492e-05, | |
| "loss": 0.0413, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.366742596810934, | |
| "grad_norm": 0.07821401953697205, | |
| "learning_rate": 1.6104783599088842e-05, | |
| "loss": 0.0304, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.3952164009111616, | |
| "grad_norm": 0.5930206179618835, | |
| "learning_rate": 1.6023429873088188e-05, | |
| "loss": 0.0525, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.4236902050113895, | |
| "grad_norm": 1.6114712953567505, | |
| "learning_rate": 1.5942076147087538e-05, | |
| "loss": 0.0391, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.4521640091116172, | |
| "grad_norm": 0.28632065653800964, | |
| "learning_rate": 1.5860722421086887e-05, | |
| "loss": 0.0491, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.4806378132118452, | |
| "grad_norm": 1.0017669200897217, | |
| "learning_rate": 1.5779368695086237e-05, | |
| "loss": 0.0395, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.5091116173120729, | |
| "grad_norm": 0.03409272059798241, | |
| "learning_rate": 1.5698014969085583e-05, | |
| "loss": 0.0399, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.5375854214123006, | |
| "grad_norm": 3.5108702182769775, | |
| "learning_rate": 1.5616661243084933e-05, | |
| "loss": 0.0424, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.5660592255125285, | |
| "grad_norm": 3.5875349044799805, | |
| "learning_rate": 1.5535307517084283e-05, | |
| "loss": 0.0497, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.5945330296127562, | |
| "grad_norm": 0.10498251020908356, | |
| "learning_rate": 1.5453953791083632e-05, | |
| "loss": 0.0461, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.6230068337129842, | |
| "grad_norm": 7.618963718414307, | |
| "learning_rate": 1.5374227139602996e-05, | |
| "loss": 0.0394, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.6514806378132119, | |
| "grad_norm": 3.3920793533325195, | |
| "learning_rate": 1.5292873413602346e-05, | |
| "loss": 0.0339, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.6799544419134396, | |
| "grad_norm": 0.04736129939556122, | |
| "learning_rate": 1.5211519687601692e-05, | |
| "loss": 0.0358, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.7084282460136673, | |
| "grad_norm": 0.17463508248329163, | |
| "learning_rate": 1.5130165961601042e-05, | |
| "loss": 0.034, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.7369020501138952, | |
| "grad_norm": 0.058148205280303955, | |
| "learning_rate": 1.5048812235600392e-05, | |
| "loss": 0.0317, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.7653758542141231, | |
| "grad_norm": 3.5262815952301025, | |
| "learning_rate": 1.4967458509599741e-05, | |
| "loss": 0.0477, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.7938496583143508, | |
| "grad_norm": 2.971937894821167, | |
| "learning_rate": 1.488610478359909e-05, | |
| "loss": 0.0364, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.8223234624145785, | |
| "grad_norm": 0.05181724950671196, | |
| "learning_rate": 1.4804751057598439e-05, | |
| "loss": 0.0394, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.8507972665148062, | |
| "grad_norm": 0.2712370455265045, | |
| "learning_rate": 1.4723397331597789e-05, | |
| "loss": 0.0456, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.8792710706150342, | |
| "grad_norm": 1.608622431755066, | |
| "learning_rate": 1.4642043605597138e-05, | |
| "loss": 0.05, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.907744874715262, | |
| "grad_norm": 5.2423529624938965, | |
| "learning_rate": 1.4560689879596486e-05, | |
| "loss": 0.0371, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.9362186788154898, | |
| "grad_norm": 0.04786211624741554, | |
| "learning_rate": 1.4479336153595836e-05, | |
| "loss": 0.0329, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.9646924829157175, | |
| "grad_norm": 0.5264931917190552, | |
| "learning_rate": 1.4397982427595185e-05, | |
| "loss": 0.0379, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.9931662870159452, | |
| "grad_norm": 0.236006498336792, | |
| "learning_rate": 1.4316628701594535e-05, | |
| "loss": 0.0253, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_f1": 0.9406438631790744, | |
| "eval_loss": 0.05223705992102623, | |
| "eval_precision": 0.9371867691279653, | |
| "eval_recall": 0.9441265567149109, | |
| "eval_runtime": 3.7109, | |
| "eval_samples_per_second": 875.791, | |
| "eval_steps_per_second": 109.676, | |
| "step": 3512 | |
| }, | |
| { | |
| "epoch": 2.021640091116173, | |
| "grad_norm": 4.225963115692139, | |
| "learning_rate": 1.4235274975593883e-05, | |
| "loss": 0.0373, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.050113895216401, | |
| "grad_norm": 0.15388111770153046, | |
| "learning_rate": 1.4153921249593233e-05, | |
| "loss": 0.0258, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.078587699316629, | |
| "grad_norm": 6.3911895751953125, | |
| "learning_rate": 1.4072567523592582e-05, | |
| "loss": 0.0217, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 2.1070615034168565, | |
| "grad_norm": 0.03075530007481575, | |
| "learning_rate": 1.3991213797591932e-05, | |
| "loss": 0.019, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.135535307517084, | |
| "grad_norm": 0.16612432897090912, | |
| "learning_rate": 1.3909860071591278e-05, | |
| "loss": 0.0175, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 2.164009111617312, | |
| "grad_norm": 0.14543022215366364, | |
| "learning_rate": 1.382850634559063e-05, | |
| "loss": 0.0289, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.19248291571754, | |
| "grad_norm": 2.9880359172821045, | |
| "learning_rate": 1.374715261958998e-05, | |
| "loss": 0.0208, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 2.2209567198177678, | |
| "grad_norm": 9.878133773803711, | |
| "learning_rate": 1.3665798893589329e-05, | |
| "loss": 0.021, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.2494305239179955, | |
| "grad_norm": 9.684176445007324, | |
| "learning_rate": 1.3584445167588675e-05, | |
| "loss": 0.0267, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 2.277904328018223, | |
| "grad_norm": 33.15859603881836, | |
| "learning_rate": 1.3503091441588025e-05, | |
| "loss": 0.0295, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.306378132118451, | |
| "grad_norm": 2.559544086456299, | |
| "learning_rate": 1.3421737715587374e-05, | |
| "loss": 0.0292, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.334851936218679, | |
| "grad_norm": 0.07091552764177322, | |
| "learning_rate": 1.3340383989586724e-05, | |
| "loss": 0.0262, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.3633257403189067, | |
| "grad_norm": 1.8119585514068604, | |
| "learning_rate": 1.3259030263586072e-05, | |
| "loss": 0.03, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.3917995444191344, | |
| "grad_norm": 0.1476636379957199, | |
| "learning_rate": 1.3177676537585422e-05, | |
| "loss": 0.0196, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.420273348519362, | |
| "grad_norm": 0.04482650011777878, | |
| "learning_rate": 1.3096322811584771e-05, | |
| "loss": 0.0263, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.44874715261959, | |
| "grad_norm": 0.09201560169458389, | |
| "learning_rate": 1.3014969085584121e-05, | |
| "loss": 0.0172, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.477220956719818, | |
| "grad_norm": 0.02215876244008541, | |
| "learning_rate": 1.2933615359583469e-05, | |
| "loss": 0.0149, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.5056947608200457, | |
| "grad_norm": 0.6993932127952576, | |
| "learning_rate": 1.2852261633582819e-05, | |
| "loss": 0.0331, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.5341685649202734, | |
| "grad_norm": 0.030453965067863464, | |
| "learning_rate": 1.2770907907582168e-05, | |
| "loss": 0.0307, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.562642369020501, | |
| "grad_norm": 0.22211593389511108, | |
| "learning_rate": 1.2689554181581518e-05, | |
| "loss": 0.0183, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.591116173120729, | |
| "grad_norm": 1.2842738628387451, | |
| "learning_rate": 1.2608200455580866e-05, | |
| "loss": 0.0226, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.619589977220957, | |
| "grad_norm": 3.9048423767089844, | |
| "learning_rate": 1.2526846729580216e-05, | |
| "loss": 0.0239, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.6480637813211843, | |
| "grad_norm": 0.9887784719467163, | |
| "learning_rate": 1.2445493003579565e-05, | |
| "loss": 0.0271, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.6765375854214124, | |
| "grad_norm": 4.759728908538818, | |
| "learning_rate": 1.2364139277578915e-05, | |
| "loss": 0.0201, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.70501138952164, | |
| "grad_norm": 0.2671191096305847, | |
| "learning_rate": 1.2282785551578263e-05, | |
| "loss": 0.0212, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.733485193621868, | |
| "grad_norm": 0.036626849323511124, | |
| "learning_rate": 1.2201431825577612e-05, | |
| "loss": 0.0155, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.7619589977220955, | |
| "grad_norm": 6.877408027648926, | |
| "learning_rate": 1.2120078099576962e-05, | |
| "loss": 0.0266, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.7904328018223232, | |
| "grad_norm": 4.725896835327148, | |
| "learning_rate": 1.2038724373576312e-05, | |
| "loss": 0.0163, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.8189066059225514, | |
| "grad_norm": 0.11965059489011765, | |
| "learning_rate": 1.195737064757566e-05, | |
| "loss": 0.0246, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.847380410022779, | |
| "grad_norm": 0.6634079813957214, | |
| "learning_rate": 1.187601692157501e-05, | |
| "loss": 0.0304, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.875854214123007, | |
| "grad_norm": 3.952694892883301, | |
| "learning_rate": 1.1794663195574359e-05, | |
| "loss": 0.0183, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 2.9043280182232345, | |
| "grad_norm": 19.788942337036133, | |
| "learning_rate": 1.1713309469573709e-05, | |
| "loss": 0.0193, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 2.932801822323462, | |
| "grad_norm": 4.989261627197266, | |
| "learning_rate": 1.1631955743573055e-05, | |
| "loss": 0.023, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 2.9612756264236904, | |
| "grad_norm": 9.020853996276855, | |
| "learning_rate": 1.1550602017572406e-05, | |
| "loss": 0.0231, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 2.989749430523918, | |
| "grad_norm": 0.9151140451431274, | |
| "learning_rate": 1.1469248291571756e-05, | |
| "loss": 0.0314, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_f1": 0.9470420646891236, | |
| "eval_loss": 0.04870199039578438, | |
| "eval_precision": 0.943090787716956, | |
| "eval_recall": 0.9510265903736116, | |
| "eval_runtime": 3.7494, | |
| "eval_samples_per_second": 866.805, | |
| "eval_steps_per_second": 108.551, | |
| "step": 5268 | |
| }, | |
| { | |
| "epoch": 3.0182232346241458, | |
| "grad_norm": 8.62992000579834, | |
| "learning_rate": 1.1387894565571106e-05, | |
| "loss": 0.0194, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 3.0466970387243735, | |
| "grad_norm": 0.5615554451942444, | |
| "learning_rate": 1.1306540839570452e-05, | |
| "loss": 0.0114, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 3.075170842824601, | |
| "grad_norm": 2.5735936164855957, | |
| "learning_rate": 1.1225187113569801e-05, | |
| "loss": 0.0163, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 3.1036446469248293, | |
| "grad_norm": 6.504094123840332, | |
| "learning_rate": 1.1143833387569151e-05, | |
| "loss": 0.0135, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 3.132118451025057, | |
| "grad_norm": 14.477537155151367, | |
| "learning_rate": 1.10624796615685e-05, | |
| "loss": 0.0129, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 3.1605922551252847, | |
| "grad_norm": 0.04929841682314873, | |
| "learning_rate": 1.0981125935567849e-05, | |
| "loss": 0.0263, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 3.1890660592255125, | |
| "grad_norm": 0.08875144273042679, | |
| "learning_rate": 1.0899772209567198e-05, | |
| "loss": 0.0134, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 3.21753986332574, | |
| "grad_norm": 0.6238455772399902, | |
| "learning_rate": 1.0818418483566548e-05, | |
| "loss": 0.0107, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 3.2460136674259683, | |
| "grad_norm": 0.004948179703205824, | |
| "learning_rate": 1.0737064757565898e-05, | |
| "loss": 0.0167, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 3.274487471526196, | |
| "grad_norm": 0.017031218856573105, | |
| "learning_rate": 1.0655711031565246e-05, | |
| "loss": 0.0085, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 3.3029612756264237, | |
| "grad_norm": 2.9840469360351562, | |
| "learning_rate": 1.0574357305564595e-05, | |
| "loss": 0.0158, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 3.3314350797266514, | |
| "grad_norm": 1.7477540969848633, | |
| "learning_rate": 1.0493003579563945e-05, | |
| "loss": 0.0157, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 3.359908883826879, | |
| "grad_norm": 0.03963543102145195, | |
| "learning_rate": 1.0411649853563295e-05, | |
| "loss": 0.0201, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 3.3883826879271073, | |
| "grad_norm": 0.16669808328151703, | |
| "learning_rate": 1.0330296127562643e-05, | |
| "loss": 0.0159, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 3.416856492027335, | |
| "grad_norm": 0.01953568309545517, | |
| "learning_rate": 1.0248942401561992e-05, | |
| "loss": 0.0117, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.4453302961275627, | |
| "grad_norm": 0.07708246260881424, | |
| "learning_rate": 1.0167588675561342e-05, | |
| "loss": 0.0117, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 3.4738041002277904, | |
| "grad_norm": 2.423590898513794, | |
| "learning_rate": 1.0086234949560691e-05, | |
| "loss": 0.0111, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 3.502277904328018, | |
| "grad_norm": 0.006337775848805904, | |
| "learning_rate": 1.000488122356004e-05, | |
| "loss": 0.0052, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 3.5307517084282463, | |
| "grad_norm": 0.01373753696680069, | |
| "learning_rate": 9.923527497559389e-06, | |
| "loss": 0.009, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 3.559225512528474, | |
| "grad_norm": 0.11260247975587845, | |
| "learning_rate": 9.842173771558739e-06, | |
| "loss": 0.0208, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 3.5876993166287017, | |
| "grad_norm": 1.240822196006775, | |
| "learning_rate": 9.760820045558087e-06, | |
| "loss": 0.0157, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 3.6161731207289294, | |
| "grad_norm": 0.05271293595433235, | |
| "learning_rate": 9.679466319557436e-06, | |
| "loss": 0.0175, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 3.644646924829157, | |
| "grad_norm": 24.546663284301758, | |
| "learning_rate": 9.598112593556786e-06, | |
| "loss": 0.015, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 3.6731207289293852, | |
| "grad_norm": 5.120173931121826, | |
| "learning_rate": 9.516758867556136e-06, | |
| "loss": 0.017, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 3.7015945330296125, | |
| "grad_norm": 1.1922008991241455, | |
| "learning_rate": 9.435405141555484e-06, | |
| "loss": 0.0194, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 3.7300683371298406, | |
| "grad_norm": 0.039439987391233444, | |
| "learning_rate": 9.354051415554833e-06, | |
| "loss": 0.0309, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 3.7585421412300684, | |
| "grad_norm": 0.04055279493331909, | |
| "learning_rate": 9.272697689554181e-06, | |
| "loss": 0.0195, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 3.787015945330296, | |
| "grad_norm": 0.5971085429191589, | |
| "learning_rate": 9.191343963553533e-06, | |
| "loss": 0.0156, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 3.8154897494305238, | |
| "grad_norm": 0.2191866636276245, | |
| "learning_rate": 9.10999023755288e-06, | |
| "loss": 0.0174, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 3.8439635535307515, | |
| "grad_norm": 0.033559828996658325, | |
| "learning_rate": 9.02863651155223e-06, | |
| "loss": 0.0149, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 3.8724373576309796, | |
| "grad_norm": 3.7128634452819824, | |
| "learning_rate": 8.947282785551578e-06, | |
| "loss": 0.0221, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 3.9009111617312073, | |
| "grad_norm": 0.08547580987215042, | |
| "learning_rate": 8.865929059550928e-06, | |
| "loss": 0.0153, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 3.929384965831435, | |
| "grad_norm": 0.058124080300331116, | |
| "learning_rate": 8.784575333550277e-06, | |
| "loss": 0.0161, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 3.9578587699316627, | |
| "grad_norm": 0.015579139813780785, | |
| "learning_rate": 8.703221607549627e-06, | |
| "loss": 0.011, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 3.9863325740318905, | |
| "grad_norm": 0.12764935195446014, | |
| "learning_rate": 8.621867881548975e-06, | |
| "loss": 0.0166, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_f1": 0.9525329748802823, | |
| "eval_loss": 0.04983380436897278, | |
| "eval_precision": 0.9510149303808086, | |
| "eval_recall": 0.9540558734432851, | |
| "eval_runtime": 3.7495, | |
| "eval_samples_per_second": 866.79, | |
| "eval_steps_per_second": 108.549, | |
| "step": 7024 | |
| }, | |
| { | |
| "epoch": 4.014806378132119, | |
| "grad_norm": 0.04521024227142334, | |
| "learning_rate": 8.540514155548325e-06, | |
| "loss": 0.0102, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 4.043280182232346, | |
| "grad_norm": 0.015535669401288033, | |
| "learning_rate": 8.459160429547674e-06, | |
| "loss": 0.0068, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 4.071753986332574, | |
| "grad_norm": 12.566010475158691, | |
| "learning_rate": 8.377806703547024e-06, | |
| "loss": 0.0151, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 4.100227790432802, | |
| "grad_norm": 0.027761396020650864, | |
| "learning_rate": 8.296452977546372e-06, | |
| "loss": 0.0053, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 4.128701594533029, | |
| "grad_norm": 1.8846938610076904, | |
| "learning_rate": 8.215099251545722e-06, | |
| "loss": 0.0077, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 4.157175398633258, | |
| "grad_norm": 0.005013479385524988, | |
| "learning_rate": 8.13374552554507e-06, | |
| "loss": 0.0073, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 4.185649202733485, | |
| "grad_norm": 0.32696235179901123, | |
| "learning_rate": 8.052391799544421e-06, | |
| "loss": 0.0092, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 4.214123006833713, | |
| "grad_norm": 0.014357910491526127, | |
| "learning_rate": 7.971038073543769e-06, | |
| "loss": 0.0164, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 4.242596810933941, | |
| "grad_norm": 0.010845329612493515, | |
| "learning_rate": 7.889684347543118e-06, | |
| "loss": 0.0123, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 4.271070615034168, | |
| "grad_norm": 0.6969354748725891, | |
| "learning_rate": 7.808330621542466e-06, | |
| "loss": 0.0103, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 4.2995444191343966, | |
| "grad_norm": 0.3575742840766907, | |
| "learning_rate": 7.726976895541816e-06, | |
| "loss": 0.0161, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 4.328018223234624, | |
| "grad_norm": 0.008076228201389313, | |
| "learning_rate": 7.645623169541166e-06, | |
| "loss": 0.0122, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 4.356492027334852, | |
| "grad_norm": 0.029101597145199776, | |
| "learning_rate": 7.564269443540515e-06, | |
| "loss": 0.0071, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 4.38496583143508, | |
| "grad_norm": 0.06841142475605011, | |
| "learning_rate": 7.482915717539863e-06, | |
| "loss": 0.0091, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 4.413439635535307, | |
| "grad_norm": 0.04133535176515579, | |
| "learning_rate": 7.401561991539213e-06, | |
| "loss": 0.0073, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 4.4419134396355355, | |
| "grad_norm": 0.03571132943034172, | |
| "learning_rate": 7.320208265538562e-06, | |
| "loss": 0.0131, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 4.470387243735763, | |
| "grad_norm": 2.8286924362182617, | |
| "learning_rate": 7.2388545395379114e-06, | |
| "loss": 0.0043, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 4.498861047835991, | |
| "grad_norm": 1.3785158395767212, | |
| "learning_rate": 7.15750081353726e-06, | |
| "loss": 0.0103, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 4.527334851936219, | |
| "grad_norm": 0.05979786813259125, | |
| "learning_rate": 7.07614708753661e-06, | |
| "loss": 0.006, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 4.555808656036446, | |
| "grad_norm": 1.3434393405914307, | |
| "learning_rate": 6.994793361535959e-06, | |
| "loss": 0.0148, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 4.5842824601366745, | |
| "grad_norm": 0.019605603069067, | |
| "learning_rate": 6.913439635535308e-06, | |
| "loss": 0.0068, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 4.612756264236902, | |
| "grad_norm": 0.0028891051188111305, | |
| "learning_rate": 6.832085909534657e-06, | |
| "loss": 0.0075, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 4.64123006833713, | |
| "grad_norm": 0.014856363646686077, | |
| "learning_rate": 6.750732183534007e-06, | |
| "loss": 0.0073, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 4.669703872437358, | |
| "grad_norm": 0.0037149768322706223, | |
| "learning_rate": 6.669378457533356e-06, | |
| "loss": 0.0066, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 4.698177676537585, | |
| "grad_norm": 0.008534993045032024, | |
| "learning_rate": 6.588024731532705e-06, | |
| "loss": 0.0121, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 4.7266514806378135, | |
| "grad_norm": 0.013920712284743786, | |
| "learning_rate": 6.506671005532054e-06, | |
| "loss": 0.013, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 4.755125284738041, | |
| "grad_norm": 0.10121840238571167, | |
| "learning_rate": 6.425317279531404e-06, | |
| "loss": 0.0087, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 4.783599088838269, | |
| "grad_norm": 0.07937771081924438, | |
| "learning_rate": 6.343963553530752e-06, | |
| "loss": 0.0086, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 4.812072892938497, | |
| "grad_norm": 0.07674901187419891, | |
| "learning_rate": 6.262609827530101e-06, | |
| "loss": 0.0052, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 4.840546697038724, | |
| "grad_norm": 0.06683178246021271, | |
| "learning_rate": 6.18125610152945e-06, | |
| "loss": 0.015, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 4.8690205011389525, | |
| "grad_norm": 0.08861212432384491, | |
| "learning_rate": 6.0999023755288e-06, | |
| "loss": 0.0109, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 4.89749430523918, | |
| "grad_norm": 0.006198623217642307, | |
| "learning_rate": 6.018548649528149e-06, | |
| "loss": 0.0181, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 4.925968109339408, | |
| "grad_norm": 0.009029284119606018, | |
| "learning_rate": 5.937194923527498e-06, | |
| "loss": 0.0043, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 4.954441913439636, | |
| "grad_norm": 1.1225602626800537, | |
| "learning_rate": 5.855841197526847e-06, | |
| "loss": 0.0121, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 4.982915717539863, | |
| "grad_norm": 0.01218325924128294, | |
| "learning_rate": 5.774487471526197e-06, | |
| "loss": 0.0115, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_f1": 0.9533444816053511, | |
| "eval_loss": 0.04722925275564194, | |
| "eval_precision": 0.9473246925889, | |
| "eval_recall": 0.9594412655671491, | |
| "eval_runtime": 3.7358, | |
| "eval_samples_per_second": 869.968, | |
| "eval_steps_per_second": 108.947, | |
| "step": 8780 | |
| }, | |
| { | |
| "epoch": 5.011389521640091, | |
| "grad_norm": 0.8161097764968872, | |
| "learning_rate": 5.6931337455255455e-06, | |
| "loss": 0.0127, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 5.039863325740319, | |
| "grad_norm": 0.048089127987623215, | |
| "learning_rate": 5.611780019524895e-06, | |
| "loss": 0.0116, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 5.068337129840547, | |
| "grad_norm": 1.7576072216033936, | |
| "learning_rate": 5.530426293524244e-06, | |
| "loss": 0.0093, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 5.096810933940774, | |
| "grad_norm": 0.07270015776157379, | |
| "learning_rate": 5.449072567523594e-06, | |
| "loss": 0.0043, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 5.125284738041002, | |
| "grad_norm": 0.2580782175064087, | |
| "learning_rate": 5.367718841522942e-06, | |
| "loss": 0.0058, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 5.15375854214123, | |
| "grad_norm": 0.12649740278720856, | |
| "learning_rate": 5.286365115522292e-06, | |
| "loss": 0.0033, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 5.182232346241458, | |
| "grad_norm": 0.06850716471672058, | |
| "learning_rate": 5.20501138952164e-06, | |
| "loss": 0.003, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 5.210706150341686, | |
| "grad_norm": 0.01197959017008543, | |
| "learning_rate": 5.12365766352099e-06, | |
| "loss": 0.0029, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 5.239179954441913, | |
| "grad_norm": 0.10818086564540863, | |
| "learning_rate": 5.0423039375203385e-06, | |
| "loss": 0.0025, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 5.267653758542141, | |
| "grad_norm": 0.035460665822029114, | |
| "learning_rate": 4.960950211519688e-06, | |
| "loss": 0.0047, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 5.296127562642369, | |
| "grad_norm": 4.532562255859375, | |
| "learning_rate": 4.879596485519037e-06, | |
| "loss": 0.0086, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 5.324601366742597, | |
| "grad_norm": 0.013236219063401222, | |
| "learning_rate": 4.7982427595183866e-06, | |
| "loss": 0.0027, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 5.353075170842825, | |
| "grad_norm": 0.018059909343719482, | |
| "learning_rate": 4.716889033517735e-06, | |
| "loss": 0.0065, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 5.381548974943052, | |
| "grad_norm": 0.019096272066235542, | |
| "learning_rate": 4.635535307517084e-06, | |
| "loss": 0.0053, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 5.41002277904328, | |
| "grad_norm": 0.10389436781406403, | |
| "learning_rate": 4.554181581516434e-06, | |
| "loss": 0.0029, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 5.438496583143508, | |
| "grad_norm": 0.0029252381063997746, | |
| "learning_rate": 4.472827855515783e-06, | |
| "loss": 0.0045, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 5.466970387243736, | |
| "grad_norm": 0.004542892333120108, | |
| "learning_rate": 4.391474129515132e-06, | |
| "loss": 0.0102, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 5.495444191343964, | |
| "grad_norm": 0.130916565656662, | |
| "learning_rate": 4.310120403514481e-06, | |
| "loss": 0.0064, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 5.523917995444191, | |
| "grad_norm": 0.008013393729925156, | |
| "learning_rate": 4.22876667751383e-06, | |
| "loss": 0.005, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 5.552391799544419, | |
| "grad_norm": 0.012782514095306396, | |
| "learning_rate": 4.1474129515131795e-06, | |
| "loss": 0.0122, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 5.5808656036446465, | |
| "grad_norm": 0.9617053866386414, | |
| "learning_rate": 4.066059225512528e-06, | |
| "loss": 0.0063, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 5.609339407744875, | |
| "grad_norm": 0.014953136444091797, | |
| "learning_rate": 3.984705499511878e-06, | |
| "loss": 0.0046, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 5.637813211845103, | |
| "grad_norm": 1.0275633335113525, | |
| "learning_rate": 3.903351773511227e-06, | |
| "loss": 0.0075, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 5.66628701594533, | |
| "grad_norm": 0.007999264635145664, | |
| "learning_rate": 3.8219980475105764e-06, | |
| "loss": 0.0031, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 5.694760820045558, | |
| "grad_norm": 0.008241960778832436, | |
| "learning_rate": 3.7406443215099252e-06, | |
| "loss": 0.0065, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 5.723234624145785, | |
| "grad_norm": 2.578386068344116, | |
| "learning_rate": 3.6592905955092745e-06, | |
| "loss": 0.0057, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 5.751708428246014, | |
| "grad_norm": 2.8855443000793457, | |
| "learning_rate": 3.5779368695086237e-06, | |
| "loss": 0.0038, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 5.780182232346242, | |
| "grad_norm": 0.016262667253613472, | |
| "learning_rate": 3.496583143507973e-06, | |
| "loss": 0.0128, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 5.808656036446469, | |
| "grad_norm": 0.045168060809373856, | |
| "learning_rate": 3.415229417507322e-06, | |
| "loss": 0.0036, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 5.837129840546697, | |
| "grad_norm": 0.1192622259259224, | |
| "learning_rate": 3.3355027660266842e-06, | |
| "loss": 0.0111, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 5.865603644646924, | |
| "grad_norm": 0.08544190227985382, | |
| "learning_rate": 3.2541490400260335e-06, | |
| "loss": 0.0051, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 5.894077448747153, | |
| "grad_norm": 3.493161678314209, | |
| "learning_rate": 3.1727953140253827e-06, | |
| "loss": 0.0054, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 5.922551252847381, | |
| "grad_norm": 0.022789066657423973, | |
| "learning_rate": 3.0914415880247315e-06, | |
| "loss": 0.0033, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 5.951025056947608, | |
| "grad_norm": 12.131625175476074, | |
| "learning_rate": 3.0100878620240807e-06, | |
| "loss": 0.0023, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 5.979498861047836, | |
| "grad_norm": 0.03329641371965408, | |
| "learning_rate": 2.92873413602343e-06, | |
| "loss": 0.0058, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_f1": 0.9557566616390145, | |
| "eval_loss": 0.05410688370466232, | |
| "eval_precision": 0.9517690253671562, | |
| "eval_recall": 0.9597778525748906, | |
| "eval_runtime": 3.6572, | |
| "eval_samples_per_second": 888.662, | |
| "eval_steps_per_second": 111.288, | |
| "step": 10536 | |
| }, | |
| { | |
| "epoch": 6.007972665148063, | |
| "grad_norm": 0.008557640947401524, | |
| "learning_rate": 2.847380410022779e-06, | |
| "loss": 0.0152, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 6.0364464692482915, | |
| "grad_norm": 0.08105529844760895, | |
| "learning_rate": 2.7660266840221284e-06, | |
| "loss": 0.0025, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 6.06492027334852, | |
| "grad_norm": 1.100066065788269, | |
| "learning_rate": 2.6846729580214776e-06, | |
| "loss": 0.0015, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 6.093394077448747, | |
| "grad_norm": 1.96909761428833, | |
| "learning_rate": 2.603319232020827e-06, | |
| "loss": 0.0021, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 6.121867881548975, | |
| "grad_norm": 0.00583766121417284, | |
| "learning_rate": 2.5219655060201757e-06, | |
| "loss": 0.0127, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 6.150341685649202, | |
| "grad_norm": 0.026862677186727524, | |
| "learning_rate": 2.440611780019525e-06, | |
| "loss": 0.004, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 6.1788154897494305, | |
| "grad_norm": 0.010042566806077957, | |
| "learning_rate": 2.359258054018874e-06, | |
| "loss": 0.0033, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 6.207289293849659, | |
| "grad_norm": 0.8956929445266724, | |
| "learning_rate": 2.2779043280182233e-06, | |
| "loss": 0.0031, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 6.235763097949886, | |
| "grad_norm": 0.009118441492319107, | |
| "learning_rate": 2.1965506020175726e-06, | |
| "loss": 0.0039, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 6.264236902050114, | |
| "grad_norm": 0.22793921828269958, | |
| "learning_rate": 2.115196876016922e-06, | |
| "loss": 0.0013, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 6.292710706150341, | |
| "grad_norm": 0.015608682297170162, | |
| "learning_rate": 2.033843150016271e-06, | |
| "loss": 0.0021, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 6.3211845102505695, | |
| "grad_norm": 0.004031027667224407, | |
| "learning_rate": 1.95248942401562e-06, | |
| "loss": 0.0011, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 6.349658314350798, | |
| "grad_norm": 0.008949857205152512, | |
| "learning_rate": 1.8711356980149693e-06, | |
| "loss": 0.0049, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 6.378132118451025, | |
| "grad_norm": 0.018670039251446724, | |
| "learning_rate": 1.7897819720143183e-06, | |
| "loss": 0.003, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 6.406605922551253, | |
| "grad_norm": 0.032393742352724075, | |
| "learning_rate": 1.7084282460136675e-06, | |
| "loss": 0.0051, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 6.43507972665148, | |
| "grad_norm": 0.11851054430007935, | |
| "learning_rate": 1.6270745200130167e-06, | |
| "loss": 0.0046, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 6.4635535307517085, | |
| "grad_norm": 0.09247086197137833, | |
| "learning_rate": 1.5457207940123657e-06, | |
| "loss": 0.0049, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 6.492027334851937, | |
| "grad_norm": 0.002555207349359989, | |
| "learning_rate": 1.464367068011715e-06, | |
| "loss": 0.0026, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 6.520501138952164, | |
| "grad_norm": 3.7294840812683105, | |
| "learning_rate": 1.3830133420110642e-06, | |
| "loss": 0.0014, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 6.548974943052392, | |
| "grad_norm": 0.00703430688008666, | |
| "learning_rate": 1.3016596160104134e-06, | |
| "loss": 0.0035, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 6.577448747152619, | |
| "grad_norm": 12.061240196228027, | |
| "learning_rate": 1.2203058900097624e-06, | |
| "loss": 0.0078, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 6.605922551252847, | |
| "grad_norm": 0.011964640580117702, | |
| "learning_rate": 1.1389521640091117e-06, | |
| "loss": 0.0024, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 6.634396355353076, | |
| "grad_norm": 0.014603933319449425, | |
| "learning_rate": 1.057598438008461e-06, | |
| "loss": 0.0072, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 6.662870159453303, | |
| "grad_norm": 0.026642296463251114, | |
| "learning_rate": 9.7624471200781e-07, | |
| "loss": 0.0007, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 6.691343963553531, | |
| "grad_norm": 3.3646230697631836, | |
| "learning_rate": 8.948909860071591e-07, | |
| "loss": 0.0046, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 6.719817767653758, | |
| "grad_norm": 0.027053840458393097, | |
| "learning_rate": 8.135372600065084e-07, | |
| "loss": 0.0031, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 6.748291571753986, | |
| "grad_norm": 2.6146697998046875, | |
| "learning_rate": 7.321835340058575e-07, | |
| "loss": 0.003, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 6.776765375854215, | |
| "grad_norm": 8.964512825012207, | |
| "learning_rate": 6.508298080052067e-07, | |
| "loss": 0.0069, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 6.805239179954442, | |
| "grad_norm": 0.3086203932762146, | |
| "learning_rate": 5.694760820045558e-07, | |
| "loss": 0.0035, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 6.83371298405467, | |
| "grad_norm": 0.018831729888916016, | |
| "learning_rate": 4.88122356003905e-07, | |
| "loss": 0.0033, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 6.862186788154897, | |
| "grad_norm": 0.0018483272287994623, | |
| "learning_rate": 4.067686300032542e-07, | |
| "loss": 0.0022, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 6.890660592255125, | |
| "grad_norm": 0.003242627950385213, | |
| "learning_rate": 3.2541490400260336e-07, | |
| "loss": 0.0043, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 6.9191343963553535, | |
| "grad_norm": 0.002707740291953087, | |
| "learning_rate": 2.440611780019525e-07, | |
| "loss": 0.0017, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 6.947608200455581, | |
| "grad_norm": 0.21824024617671967, | |
| "learning_rate": 1.6270745200130168e-07, | |
| "loss": 0.0069, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 6.976082004555809, | |
| "grad_norm": 0.008130647242069244, | |
| "learning_rate": 8.135372600065084e-08, | |
| "loss": 0.0047, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_f1": 0.9563834240267894, | |
| "eval_loss": 0.052909377962350845, | |
| "eval_precision": 0.9515242378810594, | |
| "eval_recall": 0.9612924941097274, | |
| "eval_runtime": 3.7426, | |
| "eval_samples_per_second": 868.392, | |
| "eval_steps_per_second": 108.749, | |
| "step": 12292 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 12292, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 7, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2418024281348142.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |