diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,4759 @@ +{ + "best_metric": 0.2958612975391499, + "best_model_checkpoint": "./save/jonatasgrosman/wav2vec2-large-xlsr-53-chinese-zh-cn/checkpoint-35728", + "epoch": 30.0, + "global_step": 36960, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.04, + "learning_rate": 4.993235930735931e-05, + "loss": 3.1964, + "step": 50 + }, + { + "epoch": 0.08, + "learning_rate": 4.9864718614718616e-05, + "loss": 2.1923, + "step": 100 + }, + { + "epoch": 0.12, + "learning_rate": 4.979707792207792e-05, + "loss": 1.9162, + "step": 150 + }, + { + "epoch": 0.16, + "learning_rate": 4.972943722943723e-05, + "loss": 1.8832, + "step": 200 + }, + { + "epoch": 0.2, + "learning_rate": 4.966179653679654e-05, + "loss": 1.8216, + "step": 250 + }, + { + "epoch": 0.24, + "learning_rate": 4.959415584415585e-05, + "loss": 1.7282, + "step": 300 + }, + { + "epoch": 0.28, + "learning_rate": 4.952651515151515e-05, + "loss": 1.6915, + "step": 350 + }, + { + "epoch": 0.32, + "learning_rate": 4.945887445887446e-05, + "loss": 1.7018, + "step": 400 + }, + { + "epoch": 0.37, + "learning_rate": 4.939123376623377e-05, + "loss": 1.5646, + "step": 450 + }, + { + "epoch": 0.41, + "learning_rate": 4.932359307359308e-05, + "loss": 1.5312, + "step": 500 + }, + { + "epoch": 0.45, + "learning_rate": 4.925595238095238e-05, + "loss": 1.5983, + "step": 550 + }, + { + "epoch": 0.49, + "learning_rate": 4.918831168831169e-05, + "loss": 1.5388, + "step": 600 + }, + { + "epoch": 0.53, + "learning_rate": 4.9120670995671e-05, + "loss": 1.491, + "step": 650 + }, + { + "epoch": 0.57, + "learning_rate": 4.905303030303031e-05, + "loss": 1.4054, + "step": 700 + }, + { + "epoch": 0.61, + "learning_rate": 4.898538961038962e-05, + "loss": 1.4775, + "step": 750 + }, + { + "epoch": 0.65, + "learning_rate": 4.8917748917748915e-05, + "loss": 1.4369, + "step": 800 + }, + { + "epoch": 0.69, + "learning_rate": 4.8850108225108225e-05, + "loss": 1.4013, + "step": 850 + }, + { + "epoch": 0.73, + "learning_rate": 4.8782467532467536e-05, + "loss": 1.3465, + "step": 900 + }, + { + "epoch": 0.77, + "learning_rate": 4.8714826839826846e-05, + "loss": 1.4206, + "step": 950 + }, + { + "epoch": 0.81, + "learning_rate": 4.864718614718615e-05, + "loss": 1.2723, + "step": 1000 + }, + { + "epoch": 0.85, + "learning_rate": 4.857954545454545e-05, + "loss": 1.2339, + "step": 1050 + }, + { + "epoch": 0.89, + "learning_rate": 4.8511904761904764e-05, + "loss": 1.2496, + "step": 1100 + }, + { + "epoch": 0.93, + "learning_rate": 4.8444264069264074e-05, + "loss": 1.2337, + "step": 1150 + }, + { + "epoch": 0.97, + "learning_rate": 4.8376623376623384e-05, + "loss": 1.2815, + "step": 1200 + }, + { + "epoch": 1.0, + "eval_cer": 0.40496900379643425, + "eval_loss": 1.455051064491272, + "eval_mer": 0.42002237136465326, + "eval_runtime": 216.1491, + "eval_samples_per_second": 5.214, + "eval_steps_per_second": 0.652, + "step": 1232 + }, + { + "epoch": 1.01, + "learning_rate": 4.830898268398268e-05, + "loss": 1.1819, + "step": 1250 + }, + { + "epoch": 1.06, + "learning_rate": 4.824134199134199e-05, + "loss": 1.1662, + "step": 1300 + }, + { + "epoch": 1.1, + "learning_rate": 4.81737012987013e-05, + "loss": 1.1465, + "step": 1350 + }, + { + "epoch": 1.14, + "learning_rate": 4.810606060606061e-05, + "loss": 1.011, + "step": 1400 + }, + { + "epoch": 1.18, + "learning_rate": 4.8038419913419916e-05, + "loss": 1.0133, + "step": 1450 + }, + { + "epoch": 1.22, + "learning_rate": 4.797077922077922e-05, + "loss": 1.072, + "step": 1500 + }, + { + "epoch": 1.26, + "learning_rate": 4.790313852813853e-05, + "loss": 1.1135, + "step": 1550 + }, + { + "epoch": 1.3, + "learning_rate": 4.783549783549784e-05, + "loss": 0.9754, + "step": 1600 + }, + { + "epoch": 1.34, + "learning_rate": 4.7767857142857144e-05, + "loss": 1.052, + "step": 1650 + }, + { + "epoch": 1.38, + "learning_rate": 4.770021645021645e-05, + "loss": 1.0488, + "step": 1700 + }, + { + "epoch": 1.42, + "learning_rate": 4.763257575757576e-05, + "loss": 1.0345, + "step": 1750 + }, + { + "epoch": 1.46, + "learning_rate": 4.756493506493507e-05, + "loss": 1.0513, + "step": 1800 + }, + { + "epoch": 1.5, + "learning_rate": 4.749729437229438e-05, + "loss": 0.9665, + "step": 1850 + }, + { + "epoch": 1.54, + "learning_rate": 4.742965367965368e-05, + "loss": 1.0715, + "step": 1900 + }, + { + "epoch": 1.58, + "learning_rate": 4.7362012987012986e-05, + "loss": 1.0641, + "step": 1950 + }, + { + "epoch": 1.62, + "learning_rate": 4.72943722943723e-05, + "loss": 0.9736, + "step": 2000 + }, + { + "epoch": 1.66, + "learning_rate": 4.722673160173161e-05, + "loss": 0.9678, + "step": 2050 + }, + { + "epoch": 1.7, + "learning_rate": 4.715909090909091e-05, + "loss": 1.0556, + "step": 2100 + }, + { + "epoch": 1.75, + "learning_rate": 4.7091450216450214e-05, + "loss": 0.9854, + "step": 2150 + }, + { + "epoch": 1.79, + "learning_rate": 4.7023809523809525e-05, + "loss": 0.9546, + "step": 2200 + }, + { + "epoch": 1.83, + "learning_rate": 4.6956168831168835e-05, + "loss": 1.0219, + "step": 2250 + }, + { + "epoch": 1.87, + "learning_rate": 4.6888528138528146e-05, + "loss": 0.9491, + "step": 2300 + }, + { + "epoch": 1.91, + "learning_rate": 4.682088744588744e-05, + "loss": 1.0235, + "step": 2350 + }, + { + "epoch": 1.95, + "learning_rate": 4.675324675324675e-05, + "loss": 1.0252, + "step": 2400 + }, + { + "epoch": 1.99, + "learning_rate": 4.668560606060606e-05, + "loss": 0.9278, + "step": 2450 + }, + { + "epoch": 2.0, + "eval_cer": 0.3652265846508722, + "eval_loss": 1.3382420539855957, + "eval_mer": 0.3908596995845318, + "eval_runtime": 218.6583, + "eval_samples_per_second": 5.154, + "eval_steps_per_second": 0.645, + "step": 2464 + }, + { + "epoch": 2.03, + "learning_rate": 4.6617965367965374e-05, + "loss": 0.8101, + "step": 2500 + }, + { + "epoch": 2.07, + "learning_rate": 4.655032467532468e-05, + "loss": 0.7881, + "step": 2550 + }, + { + "epoch": 2.11, + "learning_rate": 4.648268398268398e-05, + "loss": 0.8176, + "step": 2600 + }, + { + "epoch": 2.15, + "learning_rate": 4.641504329004329e-05, + "loss": 0.8024, + "step": 2650 + }, + { + "epoch": 2.19, + "learning_rate": 4.63474025974026e-05, + "loss": 0.7591, + "step": 2700 + }, + { + "epoch": 2.23, + "learning_rate": 4.627976190476191e-05, + "loss": 0.86, + "step": 2750 + }, + { + "epoch": 2.27, + "learning_rate": 4.621212121212121e-05, + "loss": 0.7364, + "step": 2800 + }, + { + "epoch": 2.31, + "learning_rate": 4.614448051948052e-05, + "loss": 0.8359, + "step": 2850 + }, + { + "epoch": 2.35, + "learning_rate": 4.607683982683983e-05, + "loss": 0.8329, + "step": 2900 + }, + { + "epoch": 2.39, + "learning_rate": 4.600919913419914e-05, + "loss": 0.8302, + "step": 2950 + }, + { + "epoch": 2.44, + "learning_rate": 4.5941558441558444e-05, + "loss": 0.7797, + "step": 3000 + }, + { + "epoch": 2.48, + "learning_rate": 4.587391774891775e-05, + "loss": 0.8389, + "step": 3050 + }, + { + "epoch": 2.52, + "learning_rate": 4.580627705627706e-05, + "loss": 0.8095, + "step": 3100 + }, + { + "epoch": 2.56, + "learning_rate": 4.573863636363637e-05, + "loss": 0.7784, + "step": 3150 + }, + { + "epoch": 2.6, + "learning_rate": 4.567099567099568e-05, + "loss": 0.7606, + "step": 3200 + }, + { + "epoch": 2.64, + "learning_rate": 4.5603354978354976e-05, + "loss": 0.8153, + "step": 3250 + }, + { + "epoch": 2.68, + "learning_rate": 4.5535714285714286e-05, + "loss": 0.7842, + "step": 3300 + }, + { + "epoch": 2.72, + "learning_rate": 4.5468073593073596e-05, + "loss": 0.8033, + "step": 3350 + }, + { + "epoch": 2.76, + "learning_rate": 4.540043290043291e-05, + "loss": 0.7306, + "step": 3400 + }, + { + "epoch": 2.8, + "learning_rate": 4.533279220779221e-05, + "loss": 0.7315, + "step": 3450 + }, + { + "epoch": 2.84, + "learning_rate": 4.5265151515151514e-05, + "loss": 0.7986, + "step": 3500 + }, + { + "epoch": 2.88, + "learning_rate": 4.5197510822510825e-05, + "loss": 0.8575, + "step": 3550 + }, + { + "epoch": 2.92, + "learning_rate": 4.5129870129870135e-05, + "loss": 0.8591, + "step": 3600 + }, + { + "epoch": 2.96, + "learning_rate": 4.506222943722944e-05, + "loss": 0.7015, + "step": 3650 + }, + { + "epoch": 3.0, + "eval_cer": 0.3540295064635494, + "eval_loss": 1.530730128288269, + "eval_mer": 0.39429530201342283, + "eval_runtime": 217.1668, + "eval_samples_per_second": 5.19, + "eval_steps_per_second": 0.649, + "step": 3696 + }, + { + "epoch": 3.0, + "learning_rate": 4.499458874458874e-05, + "loss": 0.8337, + "step": 3700 + }, + { + "epoch": 3.04, + "learning_rate": 4.492694805194805e-05, + "loss": 0.6677, + "step": 3750 + }, + { + "epoch": 3.08, + "learning_rate": 4.485930735930736e-05, + "loss": 0.6976, + "step": 3800 + }, + { + "epoch": 3.12, + "learning_rate": 4.4791666666666673e-05, + "loss": 0.6131, + "step": 3850 + }, + { + "epoch": 3.17, + "learning_rate": 4.472402597402598e-05, + "loss": 0.6671, + "step": 3900 + }, + { + "epoch": 3.21, + "learning_rate": 4.465638528138528e-05, + "loss": 0.6474, + "step": 3950 + }, + { + "epoch": 3.25, + "learning_rate": 4.458874458874459e-05, + "loss": 0.6593, + "step": 4000 + }, + { + "epoch": 3.29, + "learning_rate": 4.45211038961039e-05, + "loss": 0.6617, + "step": 4050 + }, + { + "epoch": 3.33, + "learning_rate": 4.4453463203463205e-05, + "loss": 0.607, + "step": 4100 + }, + { + "epoch": 3.37, + "learning_rate": 4.438582251082251e-05, + "loss": 0.6391, + "step": 4150 + }, + { + "epoch": 3.41, + "learning_rate": 4.431818181818182e-05, + "loss": 0.5903, + "step": 4200 + }, + { + "epoch": 3.45, + "learning_rate": 4.425054112554113e-05, + "loss": 0.6669, + "step": 4250 + }, + { + "epoch": 3.49, + "learning_rate": 4.418290043290044e-05, + "loss": 0.6639, + "step": 4300 + }, + { + "epoch": 3.53, + "learning_rate": 4.411525974025974e-05, + "loss": 0.5948, + "step": 4350 + }, + { + "epoch": 3.57, + "learning_rate": 4.404761904761905e-05, + "loss": 0.6522, + "step": 4400 + }, + { + "epoch": 3.61, + "learning_rate": 4.397997835497836e-05, + "loss": 0.6048, + "step": 4450 + }, + { + "epoch": 3.65, + "learning_rate": 4.391233766233767e-05, + "loss": 0.5718, + "step": 4500 + }, + { + "epoch": 3.69, + "learning_rate": 4.384469696969697e-05, + "loss": 0.6911, + "step": 4550 + }, + { + "epoch": 3.73, + "learning_rate": 4.3777056277056275e-05, + "loss": 0.6172, + "step": 4600 + }, + { + "epoch": 3.77, + "learning_rate": 4.3709415584415586e-05, + "loss": 0.6573, + "step": 4650 + }, + { + "epoch": 3.81, + "learning_rate": 4.3641774891774896e-05, + "loss": 0.671, + "step": 4700 + }, + { + "epoch": 3.86, + "learning_rate": 4.357413419913421e-05, + "loss": 0.6021, + "step": 4750 + }, + { + "epoch": 3.9, + "learning_rate": 4.3506493506493503e-05, + "loss": 0.5882, + "step": 4800 + }, + { + "epoch": 3.94, + "learning_rate": 4.3438852813852814e-05, + "loss": 0.6395, + "step": 4850 + }, + { + "epoch": 3.98, + "learning_rate": 4.3371212121212124e-05, + "loss": 0.6266, + "step": 4900 + }, + { + "epoch": 4.0, + "eval_cer": 0.3161612763708011, + "eval_loss": 1.3133615255355835, + "eval_mer": 0.35802173218280603, + "eval_runtime": 216.4936, + "eval_samples_per_second": 5.206, + "eval_steps_per_second": 0.651, + "step": 4928 + }, + { + "epoch": 4.02, + "learning_rate": 4.3303571428571435e-05, + "loss": 0.6378, + "step": 4950 + }, + { + "epoch": 4.06, + "learning_rate": 4.323593073593074e-05, + "loss": 0.5454, + "step": 5000 + }, + { + "epoch": 4.1, + "learning_rate": 4.316829004329004e-05, + "loss": 0.4825, + "step": 5050 + }, + { + "epoch": 4.14, + "learning_rate": 4.310064935064935e-05, + "loss": 0.5549, + "step": 5100 + }, + { + "epoch": 4.18, + "learning_rate": 4.303300865800866e-05, + "loss": 0.515, + "step": 5150 + }, + { + "epoch": 4.22, + "learning_rate": 4.2965367965367966e-05, + "loss": 0.5566, + "step": 5200 + }, + { + "epoch": 4.26, + "learning_rate": 4.289772727272727e-05, + "loss": 0.5484, + "step": 5250 + }, + { + "epoch": 4.3, + "learning_rate": 4.283008658008658e-05, + "loss": 0.5033, + "step": 5300 + }, + { + "epoch": 4.34, + "learning_rate": 4.276244588744589e-05, + "loss": 0.4821, + "step": 5350 + }, + { + "epoch": 4.38, + "learning_rate": 4.26948051948052e-05, + "loss": 0.4676, + "step": 5400 + }, + { + "epoch": 4.42, + "learning_rate": 4.2627164502164505e-05, + "loss": 0.5176, + "step": 5450 + }, + { + "epoch": 4.46, + "learning_rate": 4.255952380952381e-05, + "loss": 0.5493, + "step": 5500 + }, + { + "epoch": 4.5, + "learning_rate": 4.249188311688312e-05, + "loss": 0.5177, + "step": 5550 + }, + { + "epoch": 4.55, + "learning_rate": 4.242424242424243e-05, + "loss": 0.4891, + "step": 5600 + }, + { + "epoch": 4.59, + "learning_rate": 4.235660173160173e-05, + "loss": 0.5674, + "step": 5650 + }, + { + "epoch": 4.63, + "learning_rate": 4.228896103896104e-05, + "loss": 0.4763, + "step": 5700 + }, + { + "epoch": 4.67, + "learning_rate": 4.222132034632035e-05, + "loss": 0.5431, + "step": 5750 + }, + { + "epoch": 4.71, + "learning_rate": 4.215367965367966e-05, + "loss": 0.5039, + "step": 5800 + }, + { + "epoch": 4.75, + "learning_rate": 4.208603896103897e-05, + "loss": 0.5064, + "step": 5850 + }, + { + "epoch": 4.79, + "learning_rate": 4.2018398268398265e-05, + "loss": 0.5018, + "step": 5900 + }, + { + "epoch": 4.83, + "learning_rate": 4.1950757575757575e-05, + "loss": 0.5066, + "step": 5950 + }, + { + "epoch": 4.87, + "learning_rate": 4.1883116883116886e-05, + "loss": 0.5357, + "step": 6000 + }, + { + "epoch": 4.91, + "learning_rate": 4.1815476190476196e-05, + "loss": 0.5285, + "step": 6050 + }, + { + "epoch": 4.95, + "learning_rate": 4.17478354978355e-05, + "loss": 0.5814, + "step": 6100 + }, + { + "epoch": 4.99, + "learning_rate": 4.16801948051948e-05, + "loss": 0.5216, + "step": 6150 + }, + { + "epoch": 5.0, + "eval_cer": 0.30030275361622377, + "eval_loss": 1.3027092218399048, + "eval_mer": 0.34603707254713967, + "eval_runtime": 217.8086, + "eval_samples_per_second": 5.174, + "eval_steps_per_second": 0.647, + "step": 6160 + }, + { + "epoch": 5.03, + "learning_rate": 4.1612554112554114e-05, + "loss": 0.4249, + "step": 6200 + }, + { + "epoch": 5.07, + "learning_rate": 4.1544913419913424e-05, + "loss": 0.3902, + "step": 6250 + }, + { + "epoch": 5.11, + "learning_rate": 4.1477272727272734e-05, + "loss": 0.399, + "step": 6300 + }, + { + "epoch": 5.15, + "learning_rate": 4.140963203463203e-05, + "loss": 0.4107, + "step": 6350 + }, + { + "epoch": 5.19, + "learning_rate": 4.134199134199134e-05, + "loss": 0.4767, + "step": 6400 + }, + { + "epoch": 5.24, + "learning_rate": 4.127435064935065e-05, + "loss": 0.4182, + "step": 6450 + }, + { + "epoch": 5.28, + "learning_rate": 4.120670995670996e-05, + "loss": 0.4235, + "step": 6500 + }, + { + "epoch": 5.32, + "learning_rate": 4.1139069264069266e-05, + "loss": 0.4028, + "step": 6550 + }, + { + "epoch": 5.36, + "learning_rate": 4.107142857142857e-05, + "loss": 0.4497, + "step": 6600 + }, + { + "epoch": 5.4, + "learning_rate": 4.100378787878788e-05, + "loss": 0.4645, + "step": 6650 + }, + { + "epoch": 5.44, + "learning_rate": 4.093614718614719e-05, + "loss": 0.3768, + "step": 6700 + }, + { + "epoch": 5.48, + "learning_rate": 4.08685064935065e-05, + "loss": 0.5213, + "step": 6750 + }, + { + "epoch": 5.52, + "learning_rate": 4.08008658008658e-05, + "loss": 0.4512, + "step": 6800 + }, + { + "epoch": 5.56, + "learning_rate": 4.073322510822511e-05, + "loss": 0.4031, + "step": 6850 + }, + { + "epoch": 5.6, + "learning_rate": 4.066558441558442e-05, + "loss": 0.4361, + "step": 6900 + }, + { + "epoch": 5.64, + "learning_rate": 4.059794372294373e-05, + "loss": 0.4232, + "step": 6950 + }, + { + "epoch": 5.68, + "learning_rate": 4.053030303030303e-05, + "loss": 0.4205, + "step": 7000 + }, + { + "epoch": 5.72, + "learning_rate": 4.0462662337662336e-05, + "loss": 0.4145, + "step": 7050 + }, + { + "epoch": 5.76, + "learning_rate": 4.039502164502165e-05, + "loss": 0.5018, + "step": 7100 + }, + { + "epoch": 5.8, + "learning_rate": 4.032738095238096e-05, + "loss": 0.4073, + "step": 7150 + }, + { + "epoch": 5.84, + "learning_rate": 4.025974025974026e-05, + "loss": 0.4543, + "step": 7200 + }, + { + "epoch": 5.88, + "learning_rate": 4.0192099567099564e-05, + "loss": 0.3707, + "step": 7250 + }, + { + "epoch": 5.93, + "learning_rate": 4.0124458874458875e-05, + "loss": 0.4212, + "step": 7300 + }, + { + "epoch": 5.97, + "learning_rate": 4.0056818181818185e-05, + "loss": 0.4556, + "step": 7350 + }, + { + "epoch": 6.0, + "eval_cer": 0.2909318083521553, + "eval_loss": 1.3074078559875488, + "eval_mer": 0.3376478108021732, + "eval_runtime": 217.3379, + "eval_samples_per_second": 5.185, + "eval_steps_per_second": 0.649, + "step": 7392 + }, + { + "epoch": 6.01, + "learning_rate": 3.9989177489177496e-05, + "loss": 0.4065, + "step": 7400 + }, + { + "epoch": 6.05, + "learning_rate": 3.99215367965368e-05, + "loss": 0.3537, + "step": 7450 + }, + { + "epoch": 6.09, + "learning_rate": 3.98538961038961e-05, + "loss": 0.3525, + "step": 7500 + }, + { + "epoch": 6.13, + "learning_rate": 3.978625541125541e-05, + "loss": 0.4335, + "step": 7550 + }, + { + "epoch": 6.17, + "learning_rate": 3.9718614718614724e-05, + "loss": 0.3785, + "step": 7600 + }, + { + "epoch": 6.21, + "learning_rate": 3.965097402597403e-05, + "loss": 0.3757, + "step": 7650 + }, + { + "epoch": 6.25, + "learning_rate": 3.958333333333333e-05, + "loss": 0.3563, + "step": 7700 + }, + { + "epoch": 6.29, + "learning_rate": 3.951569264069264e-05, + "loss": 0.3371, + "step": 7750 + }, + { + "epoch": 6.33, + "learning_rate": 3.944805194805195e-05, + "loss": 0.3632, + "step": 7800 + }, + { + "epoch": 6.37, + "learning_rate": 3.938041125541126e-05, + "loss": 0.365, + "step": 7850 + }, + { + "epoch": 6.41, + "learning_rate": 3.931277056277056e-05, + "loss": 0.355, + "step": 7900 + }, + { + "epoch": 6.45, + "learning_rate": 3.924512987012987e-05, + "loss": 0.3097, + "step": 7950 + }, + { + "epoch": 6.49, + "learning_rate": 3.917748917748918e-05, + "loss": 0.3188, + "step": 8000 + }, + { + "epoch": 6.53, + "learning_rate": 3.910984848484849e-05, + "loss": 0.3512, + "step": 8050 + }, + { + "epoch": 6.57, + "learning_rate": 3.9042207792207794e-05, + "loss": 0.3474, + "step": 8100 + }, + { + "epoch": 6.62, + "learning_rate": 3.89745670995671e-05, + "loss": 0.3851, + "step": 8150 + }, + { + "epoch": 6.66, + "learning_rate": 3.890692640692641e-05, + "loss": 0.3295, + "step": 8200 + }, + { + "epoch": 6.7, + "learning_rate": 3.883928571428572e-05, + "loss": 0.3445, + "step": 8250 + }, + { + "epoch": 6.74, + "learning_rate": 3.877164502164503e-05, + "loss": 0.3924, + "step": 8300 + }, + { + "epoch": 6.78, + "learning_rate": 3.8704004329004326e-05, + "loss": 0.3905, + "step": 8350 + }, + { + "epoch": 6.82, + "learning_rate": 3.8636363636363636e-05, + "loss": 0.3809, + "step": 8400 + }, + { + "epoch": 6.86, + "learning_rate": 3.8568722943722946e-05, + "loss": 0.3692, + "step": 8450 + }, + { + "epoch": 6.9, + "learning_rate": 3.850108225108226e-05, + "loss": 0.3243, + "step": 8500 + }, + { + "epoch": 6.94, + "learning_rate": 3.843344155844156e-05, + "loss": 0.3845, + "step": 8550 + }, + { + "epoch": 6.98, + "learning_rate": 3.8365800865800864e-05, + "loss": 0.3212, + "step": 8600 + }, + { + "epoch": 7.0, + "eval_cer": 0.284684511509443, + "eval_loss": 1.3443255424499512, + "eval_mer": 0.33237456056248005, + "eval_runtime": 216.9522, + "eval_samples_per_second": 5.195, + "eval_steps_per_second": 0.65, + "step": 8624 + }, + { + "epoch": 7.02, + "learning_rate": 3.8298160173160175e-05, + "loss": 0.3799, + "step": 8650 + }, + { + "epoch": 7.06, + "learning_rate": 3.8230519480519485e-05, + "loss": 0.31, + "step": 8700 + }, + { + "epoch": 7.1, + "learning_rate": 3.816287878787879e-05, + "loss": 0.2646, + "step": 8750 + }, + { + "epoch": 7.14, + "learning_rate": 3.809523809523809e-05, + "loss": 0.2783, + "step": 8800 + }, + { + "epoch": 7.18, + "learning_rate": 3.80275974025974e-05, + "loss": 0.364, + "step": 8850 + }, + { + "epoch": 7.22, + "learning_rate": 3.795995670995671e-05, + "loss": 0.2917, + "step": 8900 + }, + { + "epoch": 7.26, + "learning_rate": 3.7892316017316023e-05, + "loss": 0.3235, + "step": 8950 + }, + { + "epoch": 7.31, + "learning_rate": 3.782467532467533e-05, + "loss": 0.2982, + "step": 9000 + }, + { + "epoch": 7.35, + "learning_rate": 3.775703463203463e-05, + "loss": 0.323, + "step": 9050 + }, + { + "epoch": 7.39, + "learning_rate": 3.768939393939394e-05, + "loss": 0.2758, + "step": 9100 + }, + { + "epoch": 7.43, + "learning_rate": 3.762175324675325e-05, + "loss": 0.3276, + "step": 9150 + }, + { + "epoch": 7.47, + "learning_rate": 3.7554112554112555e-05, + "loss": 0.3555, + "step": 9200 + }, + { + "epoch": 7.51, + "learning_rate": 3.748647186147186e-05, + "loss": 0.2968, + "step": 9250 + }, + { + "epoch": 7.55, + "learning_rate": 3.741883116883117e-05, + "loss": 0.3656, + "step": 9300 + }, + { + "epoch": 7.59, + "learning_rate": 3.735119047619048e-05, + "loss": 0.3429, + "step": 9350 + }, + { + "epoch": 7.63, + "learning_rate": 3.728354978354979e-05, + "loss": 0.3144, + "step": 9400 + }, + { + "epoch": 7.67, + "learning_rate": 3.721590909090909e-05, + "loss": 0.3158, + "step": 9450 + }, + { + "epoch": 7.71, + "learning_rate": 3.71482683982684e-05, + "loss": 0.3139, + "step": 9500 + }, + { + "epoch": 7.75, + "learning_rate": 3.708062770562771e-05, + "loss": 0.2774, + "step": 9550 + }, + { + "epoch": 7.79, + "learning_rate": 3.701298701298702e-05, + "loss": 0.2942, + "step": 9600 + }, + { + "epoch": 7.83, + "learning_rate": 3.694534632034632e-05, + "loss": 0.3149, + "step": 9650 + }, + { + "epoch": 7.87, + "learning_rate": 3.6877705627705625e-05, + "loss": 0.2894, + "step": 9700 + }, + { + "epoch": 7.91, + "learning_rate": 3.6810064935064936e-05, + "loss": 0.286, + "step": 9750 + }, + { + "epoch": 7.95, + "learning_rate": 3.6742424242424246e-05, + "loss": 0.278, + "step": 9800 + }, + { + "epoch": 8.0, + "learning_rate": 3.6674783549783557e-05, + "loss": 0.2932, + "step": 9850 + }, + { + "epoch": 8.0, + "eval_cer": 0.28132058244029023, + "eval_loss": 1.3463859558105469, + "eval_mer": 0.33013742409715563, + "eval_runtime": 217.5907, + "eval_samples_per_second": 5.179, + "eval_steps_per_second": 0.648, + "step": 9856 + }, + { + "epoch": 8.04, + "learning_rate": 3.6607142857142853e-05, + "loss": 0.2679, + "step": 9900 + }, + { + "epoch": 8.08, + "learning_rate": 3.6539502164502164e-05, + "loss": 0.2465, + "step": 9950 + }, + { + "epoch": 8.12, + "learning_rate": 3.6471861471861474e-05, + "loss": 0.2562, + "step": 10000 + }, + { + "epoch": 8.16, + "learning_rate": 3.6404220779220785e-05, + "loss": 0.2457, + "step": 10050 + }, + { + "epoch": 8.2, + "learning_rate": 3.633658008658009e-05, + "loss": 0.2664, + "step": 10100 + }, + { + "epoch": 8.24, + "learning_rate": 3.626893939393939e-05, + "loss": 0.2926, + "step": 10150 + }, + { + "epoch": 8.28, + "learning_rate": 3.62012987012987e-05, + "loss": 0.2467, + "step": 10200 + }, + { + "epoch": 8.32, + "learning_rate": 3.613365800865801e-05, + "loss": 0.2339, + "step": 10250 + }, + { + "epoch": 8.36, + "learning_rate": 3.606601731601732e-05, + "loss": 0.2644, + "step": 10300 + }, + { + "epoch": 8.4, + "learning_rate": 3.599837662337662e-05, + "loss": 0.2132, + "step": 10350 + }, + { + "epoch": 8.44, + "learning_rate": 3.593073593073593e-05, + "loss": 0.2529, + "step": 10400 + }, + { + "epoch": 8.48, + "learning_rate": 3.586309523809524e-05, + "loss": 0.228, + "step": 10450 + }, + { + "epoch": 8.52, + "learning_rate": 3.579545454545455e-05, + "loss": 0.2362, + "step": 10500 + }, + { + "epoch": 8.56, + "learning_rate": 3.5727813852813855e-05, + "loss": 0.2834, + "step": 10550 + }, + { + "epoch": 8.6, + "learning_rate": 3.566017316017316e-05, + "loss": 0.2586, + "step": 10600 + }, + { + "epoch": 8.64, + "learning_rate": 3.559253246753247e-05, + "loss": 0.2667, + "step": 10650 + }, + { + "epoch": 8.69, + "learning_rate": 3.552489177489178e-05, + "loss": 0.264, + "step": 10700 + }, + { + "epoch": 8.73, + "learning_rate": 3.545725108225108e-05, + "loss": 0.2493, + "step": 10750 + }, + { + "epoch": 8.77, + "learning_rate": 3.5389610389610387e-05, + "loss": 0.2459, + "step": 10800 + }, + { + "epoch": 8.81, + "learning_rate": 3.53219696969697e-05, + "loss": 0.2584, + "step": 10850 + }, + { + "epoch": 8.85, + "learning_rate": 3.525432900432901e-05, + "loss": 0.2871, + "step": 10900 + }, + { + "epoch": 8.89, + "learning_rate": 3.518668831168832e-05, + "loss": 0.2366, + "step": 10950 + }, + { + "epoch": 8.93, + "learning_rate": 3.511904761904762e-05, + "loss": 0.283, + "step": 11000 + }, + { + "epoch": 8.97, + "learning_rate": 3.5051406926406925e-05, + "loss": 0.2787, + "step": 11050 + }, + { + "epoch": 9.0, + "eval_cer": 0.2752655101158153, + "eval_loss": 1.368951678276062, + "eval_mer": 0.3227069351230425, + "eval_runtime": 216.8467, + "eval_samples_per_second": 5.197, + "eval_steps_per_second": 0.65, + "step": 11088 + }, + { + "epoch": 9.01, + "learning_rate": 3.4983766233766235e-05, + "loss": 0.2791, + "step": 11100 + }, + { + "epoch": 9.05, + "learning_rate": 3.4916125541125546e-05, + "loss": 0.2467, + "step": 11150 + }, + { + "epoch": 9.09, + "learning_rate": 3.484848484848485e-05, + "loss": 0.2426, + "step": 11200 + }, + { + "epoch": 9.13, + "learning_rate": 3.478084415584415e-05, + "loss": 0.2166, + "step": 11250 + }, + { + "epoch": 9.17, + "learning_rate": 3.4713203463203464e-05, + "loss": 0.2115, + "step": 11300 + }, + { + "epoch": 9.21, + "learning_rate": 3.4645562770562774e-05, + "loss": 0.2068, + "step": 11350 + }, + { + "epoch": 9.25, + "learning_rate": 3.4577922077922084e-05, + "loss": 0.2428, + "step": 11400 + }, + { + "epoch": 9.29, + "learning_rate": 3.451028138528138e-05, + "loss": 0.2099, + "step": 11450 + }, + { + "epoch": 9.33, + "learning_rate": 3.444264069264069e-05, + "loss": 0.1878, + "step": 11500 + }, + { + "epoch": 9.38, + "learning_rate": 3.4375e-05, + "loss": 0.2063, + "step": 11550 + }, + { + "epoch": 9.42, + "learning_rate": 3.430735930735931e-05, + "loss": 0.2298, + "step": 11600 + }, + { + "epoch": 9.46, + "learning_rate": 3.4239718614718616e-05, + "loss": 0.2199, + "step": 11650 + }, + { + "epoch": 9.5, + "learning_rate": 3.417207792207792e-05, + "loss": 0.2313, + "step": 11700 + }, + { + "epoch": 9.54, + "learning_rate": 3.410443722943723e-05, + "loss": 0.1911, + "step": 11750 + }, + { + "epoch": 9.58, + "learning_rate": 3.403679653679654e-05, + "loss": 0.2218, + "step": 11800 + }, + { + "epoch": 9.62, + "learning_rate": 3.396915584415585e-05, + "loss": 0.2401, + "step": 11850 + }, + { + "epoch": 9.66, + "learning_rate": 3.390151515151515e-05, + "loss": 0.1973, + "step": 11900 + }, + { + "epoch": 9.7, + "learning_rate": 3.383387445887446e-05, + "loss": 0.2304, + "step": 11950 + }, + { + "epoch": 9.74, + "learning_rate": 3.376623376623377e-05, + "loss": 0.2131, + "step": 12000 + }, + { + "epoch": 9.78, + "learning_rate": 3.369859307359308e-05, + "loss": 0.2249, + "step": 12050 + }, + { + "epoch": 9.82, + "learning_rate": 3.363095238095238e-05, + "loss": 0.2094, + "step": 12100 + }, + { + "epoch": 9.86, + "learning_rate": 3.3563311688311686e-05, + "loss": 0.2466, + "step": 12150 + }, + { + "epoch": 9.9, + "learning_rate": 3.3495670995671e-05, + "loss": 0.2148, + "step": 12200 + }, + { + "epoch": 9.94, + "learning_rate": 3.342803030303031e-05, + "loss": 0.1822, + "step": 12250 + }, + { + "epoch": 9.98, + "learning_rate": 3.336038961038961e-05, + "loss": 0.2482, + "step": 12300 + }, + { + "epoch": 10.0, + "eval_cer": 0.27065212167812003, + "eval_loss": 1.430137276649475, + "eval_mer": 0.3196708213486737, + "eval_runtime": 217.321, + "eval_samples_per_second": 5.186, + "eval_steps_per_second": 0.649, + "step": 12320 + }, + { + "epoch": 10.02, + "learning_rate": 3.3292748917748914e-05, + "loss": 0.1891, + "step": 12350 + }, + { + "epoch": 10.06, + "learning_rate": 3.3225108225108225e-05, + "loss": 0.1518, + "step": 12400 + }, + { + "epoch": 10.11, + "learning_rate": 3.3157467532467535e-05, + "loss": 0.2368, + "step": 12450 + }, + { + "epoch": 10.15, + "learning_rate": 3.3089826839826846e-05, + "loss": 0.226, + "step": 12500 + }, + { + "epoch": 10.19, + "learning_rate": 3.302218614718615e-05, + "loss": 0.2228, + "step": 12550 + }, + { + "epoch": 10.23, + "learning_rate": 3.295454545454545e-05, + "loss": 0.1958, + "step": 12600 + }, + { + "epoch": 10.27, + "learning_rate": 3.288690476190476e-05, + "loss": 0.1967, + "step": 12650 + }, + { + "epoch": 10.31, + "learning_rate": 3.2819264069264074e-05, + "loss": 0.1856, + "step": 12700 + }, + { + "epoch": 10.35, + "learning_rate": 3.275162337662338e-05, + "loss": 0.1832, + "step": 12750 + }, + { + "epoch": 10.39, + "learning_rate": 3.268398268398268e-05, + "loss": 0.1739, + "step": 12800 + }, + { + "epoch": 10.43, + "learning_rate": 3.261634199134199e-05, + "loss": 0.1737, + "step": 12850 + }, + { + "epoch": 10.47, + "learning_rate": 3.25487012987013e-05, + "loss": 0.1694, + "step": 12900 + }, + { + "epoch": 10.51, + "learning_rate": 3.248106060606061e-05, + "loss": 0.1867, + "step": 12950 + }, + { + "epoch": 10.55, + "learning_rate": 3.241341991341991e-05, + "loss": 0.1907, + "step": 13000 + }, + { + "epoch": 10.59, + "learning_rate": 3.234577922077922e-05, + "loss": 0.1684, + "step": 13050 + }, + { + "epoch": 10.63, + "learning_rate": 3.227813852813853e-05, + "loss": 0.1795, + "step": 13100 + }, + { + "epoch": 10.67, + "learning_rate": 3.221049783549784e-05, + "loss": 0.1838, + "step": 13150 + }, + { + "epoch": 10.71, + "learning_rate": 3.2142857142857144e-05, + "loss": 0.198, + "step": 13200 + }, + { + "epoch": 10.75, + "learning_rate": 3.207521645021645e-05, + "loss": 0.19, + "step": 13250 + }, + { + "epoch": 10.8, + "learning_rate": 3.200757575757576e-05, + "loss": 0.1725, + "step": 13300 + }, + { + "epoch": 10.84, + "learning_rate": 3.193993506493507e-05, + "loss": 0.2693, + "step": 13350 + }, + { + "epoch": 10.88, + "learning_rate": 3.187229437229438e-05, + "loss": 0.2033, + "step": 13400 + }, + { + "epoch": 10.92, + "learning_rate": 3.1804653679653676e-05, + "loss": 0.2122, + "step": 13450 + }, + { + "epoch": 10.96, + "learning_rate": 3.1737012987012986e-05, + "loss": 0.1898, + "step": 13500 + }, + { + "epoch": 11.0, + "learning_rate": 3.1669372294372296e-05, + "loss": 0.2176, + "step": 13550 + }, + { + "epoch": 11.0, + "eval_cer": 0.26897015714354366, + "eval_loss": 1.4202213287353516, + "eval_mer": 0.3180728667305848, + "eval_runtime": 215.6805, + "eval_samples_per_second": 5.225, + "eval_steps_per_second": 0.654, + "step": 13552 + }, + { + "epoch": 11.04, + "learning_rate": 3.160173160173161e-05, + "loss": 0.1901, + "step": 13600 + }, + { + "epoch": 11.08, + "learning_rate": 3.153409090909091e-05, + "loss": 0.1398, + "step": 13650 + }, + { + "epoch": 11.12, + "learning_rate": 3.1466450216450214e-05, + "loss": 0.1427, + "step": 13700 + }, + { + "epoch": 11.16, + "learning_rate": 3.1398809523809525e-05, + "loss": 0.1461, + "step": 13750 + }, + { + "epoch": 11.2, + "learning_rate": 3.1331168831168835e-05, + "loss": 0.1296, + "step": 13800 + }, + { + "epoch": 11.24, + "learning_rate": 3.1263528138528145e-05, + "loss": 0.1695, + "step": 13850 + }, + { + "epoch": 11.28, + "learning_rate": 3.119588744588744e-05, + "loss": 0.1431, + "step": 13900 + }, + { + "epoch": 11.32, + "learning_rate": 3.112824675324675e-05, + "loss": 0.1482, + "step": 13950 + }, + { + "epoch": 11.36, + "learning_rate": 3.106060606060606e-05, + "loss": 0.155, + "step": 14000 + }, + { + "epoch": 11.4, + "learning_rate": 3.0992965367965373e-05, + "loss": 0.1611, + "step": 14050 + }, + { + "epoch": 11.44, + "learning_rate": 3.092532467532468e-05, + "loss": 0.1815, + "step": 14100 + }, + { + "epoch": 11.49, + "learning_rate": 3.085768398268398e-05, + "loss": 0.173, + "step": 14150 + }, + { + "epoch": 11.53, + "learning_rate": 3.079004329004329e-05, + "loss": 0.1693, + "step": 14200 + }, + { + "epoch": 11.57, + "learning_rate": 3.07224025974026e-05, + "loss": 0.1859, + "step": 14250 + }, + { + "epoch": 11.61, + "learning_rate": 3.0654761904761905e-05, + "loss": 0.1763, + "step": 14300 + }, + { + "epoch": 11.65, + "learning_rate": 3.058712121212121e-05, + "loss": 0.1747, + "step": 14350 + }, + { + "epoch": 11.69, + "learning_rate": 3.051948051948052e-05, + "loss": 0.176, + "step": 14400 + }, + { + "epoch": 11.73, + "learning_rate": 3.045183982683983e-05, + "loss": 0.1514, + "step": 14450 + }, + { + "epoch": 11.77, + "learning_rate": 3.0384199134199137e-05, + "loss": 0.2071, + "step": 14500 + }, + { + "epoch": 11.81, + "learning_rate": 3.031655844155844e-05, + "loss": 0.1684, + "step": 14550 + }, + { + "epoch": 11.85, + "learning_rate": 3.0248917748917747e-05, + "loss": 0.1552, + "step": 14600 + }, + { + "epoch": 11.89, + "learning_rate": 3.0181277056277058e-05, + "loss": 0.2061, + "step": 14650 + }, + { + "epoch": 11.93, + "learning_rate": 3.0113636363636365e-05, + "loss": 0.1661, + "step": 14700 + }, + { + "epoch": 11.97, + "learning_rate": 3.0045995670995675e-05, + "loss": 0.1789, + "step": 14750 + }, + { + "epoch": 12.0, + "eval_cer": 0.266951799702052, + "eval_loss": 1.5919291973114014, + "eval_mer": 0.3156759348034516, + "eval_runtime": 215.4469, + "eval_samples_per_second": 5.231, + "eval_steps_per_second": 0.654, + "step": 14784 + }, + { + "epoch": 12.01, + "learning_rate": 2.997835497835498e-05, + "loss": 0.175, + "step": 14800 + }, + { + "epoch": 12.05, + "learning_rate": 2.9910714285714286e-05, + "loss": 0.1368, + "step": 14850 + }, + { + "epoch": 12.09, + "learning_rate": 2.9843073593073596e-05, + "loss": 0.121, + "step": 14900 + }, + { + "epoch": 12.13, + "learning_rate": 2.9775432900432903e-05, + "loss": 0.1425, + "step": 14950 + }, + { + "epoch": 12.18, + "learning_rate": 2.9707792207792207e-05, + "loss": 0.1534, + "step": 15000 + }, + { + "epoch": 12.22, + "learning_rate": 2.9640151515151514e-05, + "loss": 0.1691, + "step": 15050 + }, + { + "epoch": 12.26, + "learning_rate": 2.9572510822510824e-05, + "loss": 0.1276, + "step": 15100 + }, + { + "epoch": 12.3, + "learning_rate": 2.950487012987013e-05, + "loss": 0.1304, + "step": 15150 + }, + { + "epoch": 12.34, + "learning_rate": 2.943722943722944e-05, + "loss": 0.1564, + "step": 15200 + }, + { + "epoch": 12.38, + "learning_rate": 2.9369588744588745e-05, + "loss": 0.1219, + "step": 15250 + }, + { + "epoch": 12.42, + "learning_rate": 2.9301948051948052e-05, + "loss": 0.1604, + "step": 15300 + }, + { + "epoch": 12.46, + "learning_rate": 2.9234307359307363e-05, + "loss": 0.145, + "step": 15350 + }, + { + "epoch": 12.5, + "learning_rate": 2.916666666666667e-05, + "loss": 0.136, + "step": 15400 + }, + { + "epoch": 12.54, + "learning_rate": 2.9099025974025973e-05, + "loss": 0.1394, + "step": 15450 + }, + { + "epoch": 12.58, + "learning_rate": 2.903138528138528e-05, + "loss": 0.1355, + "step": 15500 + }, + { + "epoch": 12.62, + "learning_rate": 2.896374458874459e-05, + "loss": 0.1293, + "step": 15550 + }, + { + "epoch": 12.66, + "learning_rate": 2.8896103896103898e-05, + "loss": 0.1454, + "step": 15600 + }, + { + "epoch": 12.7, + "learning_rate": 2.8828463203463208e-05, + "loss": 0.1378, + "step": 15650 + }, + { + "epoch": 12.74, + "learning_rate": 2.8760822510822512e-05, + "loss": 0.1405, + "step": 15700 + }, + { + "epoch": 12.78, + "learning_rate": 2.869318181818182e-05, + "loss": 0.1397, + "step": 15750 + }, + { + "epoch": 12.82, + "learning_rate": 2.862554112554113e-05, + "loss": 0.1307, + "step": 15800 + }, + { + "epoch": 12.87, + "learning_rate": 2.8557900432900436e-05, + "loss": 0.1331, + "step": 15850 + }, + { + "epoch": 12.91, + "learning_rate": 2.849025974025974e-05, + "loss": 0.1569, + "step": 15900 + }, + { + "epoch": 12.95, + "learning_rate": 2.8422619047619047e-05, + "loss": 0.1444, + "step": 15950 + }, + { + "epoch": 12.99, + "learning_rate": 2.8354978354978357e-05, + "loss": 0.1557, + "step": 16000 + }, + { + "epoch": 13.0, + "eval_cer": 0.26709596809073, + "eval_loss": 1.5367108583450317, + "eval_mer": 0.317913071268776, + "eval_runtime": 216.3585, + "eval_samples_per_second": 5.209, + "eval_steps_per_second": 0.652, + "step": 16016 + }, + { + "epoch": 13.03, + "learning_rate": 2.8287337662337664e-05, + "loss": 0.1467, + "step": 16050 + }, + { + "epoch": 13.07, + "learning_rate": 2.8219696969696975e-05, + "loss": 0.1197, + "step": 16100 + }, + { + "epoch": 13.11, + "learning_rate": 2.815205627705628e-05, + "loss": 0.1162, + "step": 16150 + }, + { + "epoch": 13.15, + "learning_rate": 2.8084415584415585e-05, + "loss": 0.093, + "step": 16200 + }, + { + "epoch": 13.19, + "learning_rate": 2.8016774891774892e-05, + "loss": 0.1373, + "step": 16250 + }, + { + "epoch": 13.23, + "learning_rate": 2.7949134199134203e-05, + "loss": 0.121, + "step": 16300 + }, + { + "epoch": 13.27, + "learning_rate": 2.7881493506493507e-05, + "loss": 0.1317, + "step": 16350 + }, + { + "epoch": 13.31, + "learning_rate": 2.7813852813852814e-05, + "loss": 0.1352, + "step": 16400 + }, + { + "epoch": 13.35, + "learning_rate": 2.7746212121212124e-05, + "loss": 0.1073, + "step": 16450 + }, + { + "epoch": 13.39, + "learning_rate": 2.767857142857143e-05, + "loss": 0.1212, + "step": 16500 + }, + { + "epoch": 13.43, + "learning_rate": 2.7610930735930735e-05, + "loss": 0.1542, + "step": 16550 + }, + { + "epoch": 13.47, + "learning_rate": 2.754329004329004e-05, + "loss": 0.1445, + "step": 16600 + }, + { + "epoch": 13.51, + "learning_rate": 2.7475649350649352e-05, + "loss": 0.106, + "step": 16650 + }, + { + "epoch": 13.56, + "learning_rate": 2.740800865800866e-05, + "loss": 0.1023, + "step": 16700 + }, + { + "epoch": 13.6, + "learning_rate": 2.734036796536797e-05, + "loss": 0.131, + "step": 16750 + }, + { + "epoch": 13.64, + "learning_rate": 2.7272727272727273e-05, + "loss": 0.1182, + "step": 16800 + }, + { + "epoch": 13.68, + "learning_rate": 2.720508658008658e-05, + "loss": 0.1086, + "step": 16850 + }, + { + "epoch": 13.72, + "learning_rate": 2.713744588744589e-05, + "loss": 0.1214, + "step": 16900 + }, + { + "epoch": 13.76, + "learning_rate": 2.7069805194805198e-05, + "loss": 0.1353, + "step": 16950 + }, + { + "epoch": 13.8, + "learning_rate": 2.70021645021645e-05, + "loss": 0.1273, + "step": 17000 + }, + { + "epoch": 13.84, + "learning_rate": 2.6934523809523808e-05, + "loss": 0.1184, + "step": 17050 + }, + { + "epoch": 13.88, + "learning_rate": 2.686688311688312e-05, + "loss": 0.1343, + "step": 17100 + }, + { + "epoch": 13.92, + "learning_rate": 2.6799242424242426e-05, + "loss": 0.1314, + "step": 17150 + }, + { + "epoch": 13.96, + "learning_rate": 2.6731601731601736e-05, + "loss": 0.1328, + "step": 17200 + }, + { + "epoch": 14.0, + "eval_cer": 0.2673843048680859, + "eval_loss": 1.5152249336242676, + "eval_mer": 0.3162352189197827, + "eval_runtime": 217.4486, + "eval_samples_per_second": 5.183, + "eval_steps_per_second": 0.648, + "step": 17248 + }, + { + "epoch": 14.0, + "learning_rate": 2.666396103896104e-05, + "loss": 0.1186, + "step": 17250 + }, + { + "epoch": 14.04, + "learning_rate": 2.6596320346320347e-05, + "loss": 0.0985, + "step": 17300 + }, + { + "epoch": 14.08, + "learning_rate": 2.6528679653679657e-05, + "loss": 0.11, + "step": 17350 + }, + { + "epoch": 14.12, + "learning_rate": 2.6461038961038964e-05, + "loss": 0.1185, + "step": 17400 + }, + { + "epoch": 14.16, + "learning_rate": 2.6393398268398268e-05, + "loss": 0.1099, + "step": 17450 + }, + { + "epoch": 14.2, + "learning_rate": 2.6325757575757575e-05, + "loss": 0.106, + "step": 17500 + }, + { + "epoch": 14.25, + "learning_rate": 2.6258116883116885e-05, + "loss": 0.0785, + "step": 17550 + }, + { + "epoch": 14.29, + "learning_rate": 2.6190476190476192e-05, + "loss": 0.1029, + "step": 17600 + }, + { + "epoch": 14.33, + "learning_rate": 2.6122835497835503e-05, + "loss": 0.1203, + "step": 17650 + }, + { + "epoch": 14.37, + "learning_rate": 2.6055194805194806e-05, + "loss": 0.1184, + "step": 17700 + }, + { + "epoch": 14.41, + "learning_rate": 2.5987554112554113e-05, + "loss": 0.1351, + "step": 17750 + }, + { + "epoch": 14.45, + "learning_rate": 2.591991341991342e-05, + "loss": 0.1235, + "step": 17800 + }, + { + "epoch": 14.49, + "learning_rate": 2.585227272727273e-05, + "loss": 0.1144, + "step": 17850 + }, + { + "epoch": 14.53, + "learning_rate": 2.5784632034632034e-05, + "loss": 0.147, + "step": 17900 + }, + { + "epoch": 14.57, + "learning_rate": 2.571699134199134e-05, + "loss": 0.1162, + "step": 17950 + }, + { + "epoch": 14.61, + "learning_rate": 2.5649350649350652e-05, + "loss": 0.1331, + "step": 18000 + }, + { + "epoch": 14.65, + "learning_rate": 2.558170995670996e-05, + "loss": 0.0867, + "step": 18050 + }, + { + "epoch": 14.69, + "learning_rate": 2.551406926406927e-05, + "loss": 0.1207, + "step": 18100 + }, + { + "epoch": 14.73, + "learning_rate": 2.544642857142857e-05, + "loss": 0.1016, + "step": 18150 + }, + { + "epoch": 14.77, + "learning_rate": 2.537878787878788e-05, + "loss": 0.1218, + "step": 18200 + }, + { + "epoch": 14.81, + "learning_rate": 2.5311147186147187e-05, + "loss": 0.1225, + "step": 18250 + }, + { + "epoch": 14.85, + "learning_rate": 2.5243506493506497e-05, + "loss": 0.1284, + "step": 18300 + }, + { + "epoch": 14.89, + "learning_rate": 2.51758658008658e-05, + "loss": 0.1163, + "step": 18350 + }, + { + "epoch": 14.94, + "learning_rate": 2.5108225108225108e-05, + "loss": 0.1368, + "step": 18400 + }, + { + "epoch": 14.98, + "learning_rate": 2.504058441558442e-05, + "loss": 0.1094, + "step": 18450 + }, + { + "epoch": 15.0, + "eval_cer": 0.261040895766255, + "eval_loss": 1.6068286895751953, + "eval_mer": 0.30864493448386066, + "eval_runtime": 217.2374, + "eval_samples_per_second": 5.188, + "eval_steps_per_second": 0.649, + "step": 18480 + }, + { + "epoch": 15.02, + "learning_rate": 2.4972943722943722e-05, + "loss": 0.122, + "step": 18500 + }, + { + "epoch": 15.06, + "learning_rate": 2.4905303030303032e-05, + "loss": 0.0876, + "step": 18550 + }, + { + "epoch": 15.1, + "learning_rate": 2.483766233766234e-05, + "loss": 0.0845, + "step": 18600 + }, + { + "epoch": 15.14, + "learning_rate": 2.4770021645021646e-05, + "loss": 0.079, + "step": 18650 + }, + { + "epoch": 15.18, + "learning_rate": 2.4702380952380953e-05, + "loss": 0.108, + "step": 18700 + }, + { + "epoch": 15.22, + "learning_rate": 2.463474025974026e-05, + "loss": 0.098, + "step": 18750 + }, + { + "epoch": 15.26, + "learning_rate": 2.4567099567099567e-05, + "loss": 0.0936, + "step": 18800 + }, + { + "epoch": 15.3, + "learning_rate": 2.4499458874458875e-05, + "loss": 0.0766, + "step": 18850 + }, + { + "epoch": 15.34, + "learning_rate": 2.4431818181818185e-05, + "loss": 0.0993, + "step": 18900 + }, + { + "epoch": 15.38, + "learning_rate": 2.436417748917749e-05, + "loss": 0.0851, + "step": 18950 + }, + { + "epoch": 15.42, + "learning_rate": 2.42965367965368e-05, + "loss": 0.1116, + "step": 19000 + }, + { + "epoch": 15.46, + "learning_rate": 2.4228896103896103e-05, + "loss": 0.0895, + "step": 19050 + }, + { + "epoch": 15.5, + "learning_rate": 2.4161255411255413e-05, + "loss": 0.0996, + "step": 19100 + }, + { + "epoch": 15.54, + "learning_rate": 2.409361471861472e-05, + "loss": 0.0965, + "step": 19150 + }, + { + "epoch": 15.58, + "learning_rate": 2.4025974025974027e-05, + "loss": 0.0833, + "step": 19200 + }, + { + "epoch": 15.62, + "learning_rate": 2.3958333333333334e-05, + "loss": 0.0886, + "step": 19250 + }, + { + "epoch": 15.67, + "learning_rate": 2.389069264069264e-05, + "loss": 0.1058, + "step": 19300 + }, + { + "epoch": 15.71, + "learning_rate": 2.382305194805195e-05, + "loss": 0.1254, + "step": 19350 + }, + { + "epoch": 15.75, + "learning_rate": 2.3755411255411255e-05, + "loss": 0.1088, + "step": 19400 + }, + { + "epoch": 15.79, + "learning_rate": 2.3687770562770566e-05, + "loss": 0.1121, + "step": 19450 + }, + { + "epoch": 15.83, + "learning_rate": 2.362012987012987e-05, + "loss": 0.1078, + "step": 19500 + }, + { + "epoch": 15.87, + "learning_rate": 2.355248917748918e-05, + "loss": 0.0855, + "step": 19550 + }, + { + "epoch": 15.91, + "learning_rate": 2.3484848484848487e-05, + "loss": 0.1041, + "step": 19600 + }, + { + "epoch": 15.95, + "learning_rate": 2.3417207792207794e-05, + "loss": 0.0947, + "step": 19650 + }, + { + "epoch": 15.99, + "learning_rate": 2.33495670995671e-05, + "loss": 0.204, + "step": 19700 + }, + { + "epoch": 16.0, + "eval_cer": 0.2614253448027296, + "eval_loss": 1.6343796253204346, + "eval_mer": 0.30752636625119845, + "eval_runtime": 215.5234, + "eval_samples_per_second": 5.229, + "eval_steps_per_second": 0.654, + "step": 19712 + }, + { + "epoch": 16.03, + "learning_rate": 2.3281926406926408e-05, + "loss": 0.0996, + "step": 19750 + }, + { + "epoch": 16.07, + "learning_rate": 2.3214285714285715e-05, + "loss": 0.0752, + "step": 19800 + }, + { + "epoch": 16.11, + "learning_rate": 2.314664502164502e-05, + "loss": 0.0983, + "step": 19850 + }, + { + "epoch": 16.15, + "learning_rate": 2.3079004329004332e-05, + "loss": 0.0698, + "step": 19900 + }, + { + "epoch": 16.19, + "learning_rate": 2.3011363636363636e-05, + "loss": 0.0724, + "step": 19950 + }, + { + "epoch": 16.23, + "learning_rate": 2.2943722943722946e-05, + "loss": 0.088, + "step": 20000 + }, + { + "epoch": 16.27, + "learning_rate": 2.287608225108225e-05, + "loss": 0.1087, + "step": 20050 + }, + { + "epoch": 16.31, + "learning_rate": 2.280844155844156e-05, + "loss": 0.084, + "step": 20100 + }, + { + "epoch": 16.36, + "learning_rate": 2.2740800865800867e-05, + "loss": 0.0737, + "step": 20150 + }, + { + "epoch": 16.4, + "learning_rate": 2.2673160173160174e-05, + "loss": 0.1058, + "step": 20200 + }, + { + "epoch": 16.44, + "learning_rate": 2.260551948051948e-05, + "loss": 0.0756, + "step": 20250 + }, + { + "epoch": 16.48, + "learning_rate": 2.2537878787878788e-05, + "loss": 0.0806, + "step": 20300 + }, + { + "epoch": 16.52, + "learning_rate": 2.2470238095238095e-05, + "loss": 0.0861, + "step": 20350 + }, + { + "epoch": 16.56, + "learning_rate": 2.2402597402597402e-05, + "loss": 0.0787, + "step": 20400 + }, + { + "epoch": 16.6, + "learning_rate": 2.2334956709956713e-05, + "loss": 0.079, + "step": 20450 + }, + { + "epoch": 16.64, + "learning_rate": 2.2267316017316016e-05, + "loss": 0.0798, + "step": 20500 + }, + { + "epoch": 16.68, + "learning_rate": 2.2199675324675327e-05, + "loss": 0.0894, + "step": 20550 + }, + { + "epoch": 16.72, + "learning_rate": 2.2132034632034634e-05, + "loss": 0.0994, + "step": 20600 + }, + { + "epoch": 16.76, + "learning_rate": 2.206439393939394e-05, + "loss": 0.0971, + "step": 20650 + }, + { + "epoch": 16.8, + "learning_rate": 2.1996753246753248e-05, + "loss": 0.0809, + "step": 20700 + }, + { + "epoch": 16.84, + "learning_rate": 2.1929112554112555e-05, + "loss": 0.0848, + "step": 20750 + }, + { + "epoch": 16.88, + "learning_rate": 2.1861471861471862e-05, + "loss": 0.0883, + "step": 20800 + }, + { + "epoch": 16.92, + "learning_rate": 2.179383116883117e-05, + "loss": 0.1039, + "step": 20850 + }, + { + "epoch": 16.96, + "learning_rate": 2.172619047619048e-05, + "loss": 0.1116, + "step": 20900 + }, + { + "epoch": 17.0, + "eval_cer": 0.2613772886731703, + "eval_loss": 1.701643466949463, + "eval_mer": 0.30824544582933844, + "eval_runtime": 216.1518, + "eval_samples_per_second": 5.214, + "eval_steps_per_second": 0.652, + "step": 20944 + }, + { + "epoch": 17.0, + "learning_rate": 2.1658549783549783e-05, + "loss": 0.0909, + "step": 20950 + }, + { + "epoch": 17.05, + "learning_rate": 2.1590909090909093e-05, + "loss": 0.0984, + "step": 21000 + }, + { + "epoch": 17.09, + "learning_rate": 2.1523268398268397e-05, + "loss": 0.0839, + "step": 21050 + }, + { + "epoch": 17.13, + "learning_rate": 2.1455627705627707e-05, + "loss": 0.0687, + "step": 21100 + }, + { + "epoch": 17.17, + "learning_rate": 2.1387987012987014e-05, + "loss": 0.0681, + "step": 21150 + }, + { + "epoch": 17.21, + "learning_rate": 2.132034632034632e-05, + "loss": 0.0737, + "step": 21200 + }, + { + "epoch": 17.25, + "learning_rate": 2.125270562770563e-05, + "loss": 0.0852, + "step": 21250 + }, + { + "epoch": 17.29, + "learning_rate": 2.1185064935064935e-05, + "loss": 0.0715, + "step": 21300 + }, + { + "epoch": 17.33, + "learning_rate": 2.1117424242424242e-05, + "loss": 0.0832, + "step": 21350 + }, + { + "epoch": 17.37, + "learning_rate": 2.104978354978355e-05, + "loss": 0.0956, + "step": 21400 + }, + { + "epoch": 17.41, + "learning_rate": 2.098214285714286e-05, + "loss": 0.075, + "step": 21450 + }, + { + "epoch": 17.45, + "learning_rate": 2.0914502164502164e-05, + "loss": 0.0763, + "step": 21500 + }, + { + "epoch": 17.49, + "learning_rate": 2.0846861471861474e-05, + "loss": 0.0774, + "step": 21550 + }, + { + "epoch": 17.53, + "learning_rate": 2.077922077922078e-05, + "loss": 0.0778, + "step": 21600 + }, + { + "epoch": 17.57, + "learning_rate": 2.0711580086580088e-05, + "loss": 0.0864, + "step": 21650 + }, + { + "epoch": 17.61, + "learning_rate": 2.0643939393939395e-05, + "loss": 0.082, + "step": 21700 + }, + { + "epoch": 17.65, + "learning_rate": 2.0576298701298702e-05, + "loss": 0.0764, + "step": 21750 + }, + { + "epoch": 17.69, + "learning_rate": 2.050865800865801e-05, + "loss": 0.0813, + "step": 21800 + }, + { + "epoch": 17.74, + "learning_rate": 2.0441017316017316e-05, + "loss": 0.0775, + "step": 21850 + }, + { + "epoch": 17.78, + "learning_rate": 2.0373376623376626e-05, + "loss": 0.0764, + "step": 21900 + }, + { + "epoch": 17.82, + "learning_rate": 2.030573593073593e-05, + "loss": 0.0711, + "step": 21950 + }, + { + "epoch": 17.86, + "learning_rate": 2.023809523809524e-05, + "loss": 0.0886, + "step": 22000 + }, + { + "epoch": 17.9, + "learning_rate": 2.0170454545454544e-05, + "loss": 0.0798, + "step": 22050 + }, + { + "epoch": 17.94, + "learning_rate": 2.0102813852813855e-05, + "loss": 0.086, + "step": 22100 + }, + { + "epoch": 17.98, + "learning_rate": 2.003517316017316e-05, + "loss": 0.0874, + "step": 22150 + }, + { + "epoch": 18.0, + "eval_cer": 0.26253063578259406, + "eval_loss": 1.7160123586654663, + "eval_mer": 0.3069670821348674, + "eval_runtime": 215.4465, + "eval_samples_per_second": 5.231, + "eval_steps_per_second": 0.654, + "step": 22176 + }, + { + "epoch": 18.02, + "learning_rate": 1.996753246753247e-05, + "loss": 0.0931, + "step": 22200 + }, + { + "epoch": 18.06, + "learning_rate": 1.9899891774891776e-05, + "loss": 0.069, + "step": 22250 + }, + { + "epoch": 18.1, + "learning_rate": 1.9832251082251083e-05, + "loss": 0.0586, + "step": 22300 + }, + { + "epoch": 18.14, + "learning_rate": 1.976461038961039e-05, + "loss": 0.0624, + "step": 22350 + }, + { + "epoch": 18.18, + "learning_rate": 1.9696969696969697e-05, + "loss": 0.0672, + "step": 22400 + }, + { + "epoch": 18.22, + "learning_rate": 1.9629329004329007e-05, + "loss": 0.0896, + "step": 22450 + }, + { + "epoch": 18.26, + "learning_rate": 1.956168831168831e-05, + "loss": 0.0876, + "step": 22500 + }, + { + "epoch": 18.3, + "learning_rate": 1.949404761904762e-05, + "loss": 0.0629, + "step": 22550 + }, + { + "epoch": 18.34, + "learning_rate": 1.9426406926406928e-05, + "loss": 0.0743, + "step": 22600 + }, + { + "epoch": 18.38, + "learning_rate": 1.9358766233766235e-05, + "loss": 0.0825, + "step": 22650 + }, + { + "epoch": 18.43, + "learning_rate": 1.9291125541125542e-05, + "loss": 0.0786, + "step": 22700 + }, + { + "epoch": 18.47, + "learning_rate": 1.922348484848485e-05, + "loss": 0.0519, + "step": 22750 + }, + { + "epoch": 18.51, + "learning_rate": 1.9155844155844156e-05, + "loss": 0.0624, + "step": 22800 + }, + { + "epoch": 18.55, + "learning_rate": 1.9088203463203463e-05, + "loss": 0.0635, + "step": 22850 + }, + { + "epoch": 18.59, + "learning_rate": 1.9020562770562774e-05, + "loss": 0.0611, + "step": 22900 + }, + { + "epoch": 18.63, + "learning_rate": 1.8952922077922077e-05, + "loss": 0.0815, + "step": 22950 + }, + { + "epoch": 18.67, + "learning_rate": 1.8885281385281388e-05, + "loss": 0.0776, + "step": 23000 + }, + { + "epoch": 18.71, + "learning_rate": 1.881764069264069e-05, + "loss": 0.0869, + "step": 23050 + }, + { + "epoch": 18.75, + "learning_rate": 1.8750000000000002e-05, + "loss": 0.06, + "step": 23100 + }, + { + "epoch": 18.79, + "learning_rate": 1.868235930735931e-05, + "loss": 0.0667, + "step": 23150 + }, + { + "epoch": 18.83, + "learning_rate": 1.8614718614718616e-05, + "loss": 0.0608, + "step": 23200 + }, + { + "epoch": 18.87, + "learning_rate": 1.8547077922077923e-05, + "loss": 0.0744, + "step": 23250 + }, + { + "epoch": 18.91, + "learning_rate": 1.847943722943723e-05, + "loss": 0.0843, + "step": 23300 + }, + { + "epoch": 18.95, + "learning_rate": 1.8411796536796537e-05, + "loss": 0.0877, + "step": 23350 + }, + { + "epoch": 18.99, + "learning_rate": 1.8344155844155844e-05, + "loss": 0.0495, + "step": 23400 + }, + { + "epoch": 19.0, + "eval_cer": 0.2569561247537123, + "eval_loss": 1.73832106590271, + "eval_mer": 0.30377117289868966, + "eval_runtime": 216.9476, + "eval_samples_per_second": 5.195, + "eval_steps_per_second": 0.65, + "step": 23408 + }, + { + "epoch": 19.03, + "learning_rate": 1.8276515151515154e-05, + "loss": 0.0714, + "step": 23450 + }, + { + "epoch": 19.07, + "learning_rate": 1.8208874458874458e-05, + "loss": 0.0916, + "step": 23500 + }, + { + "epoch": 19.12, + "learning_rate": 1.814123376623377e-05, + "loss": 0.0573, + "step": 23550 + }, + { + "epoch": 19.16, + "learning_rate": 1.8073593073593072e-05, + "loss": 0.0617, + "step": 23600 + }, + { + "epoch": 19.2, + "learning_rate": 1.8005952380952382e-05, + "loss": 0.0823, + "step": 23650 + }, + { + "epoch": 19.24, + "learning_rate": 1.793831168831169e-05, + "loss": 0.0498, + "step": 23700 + }, + { + "epoch": 19.28, + "learning_rate": 1.7870670995670996e-05, + "loss": 0.064, + "step": 23750 + }, + { + "epoch": 19.32, + "learning_rate": 1.7803030303030303e-05, + "loss": 0.0582, + "step": 23800 + }, + { + "epoch": 19.36, + "learning_rate": 1.773538961038961e-05, + "loss": 0.0841, + "step": 23850 + }, + { + "epoch": 19.4, + "learning_rate": 1.766774891774892e-05, + "loss": 0.0671, + "step": 23900 + }, + { + "epoch": 19.44, + "learning_rate": 1.7600108225108224e-05, + "loss": 0.0567, + "step": 23950 + }, + { + "epoch": 19.48, + "learning_rate": 1.7532467532467535e-05, + "loss": 0.0602, + "step": 24000 + }, + { + "epoch": 19.52, + "learning_rate": 1.746482683982684e-05, + "loss": 0.0696, + "step": 24050 + }, + { + "epoch": 19.56, + "learning_rate": 1.739718614718615e-05, + "loss": 0.0496, + "step": 24100 + }, + { + "epoch": 19.6, + "learning_rate": 1.7329545454545456e-05, + "loss": 0.0666, + "step": 24150 + }, + { + "epoch": 19.64, + "learning_rate": 1.7261904761904763e-05, + "loss": 0.0864, + "step": 24200 + }, + { + "epoch": 19.68, + "learning_rate": 1.719426406926407e-05, + "loss": 0.0633, + "step": 24250 + }, + { + "epoch": 19.72, + "learning_rate": 1.7126623376623377e-05, + "loss": 0.0577, + "step": 24300 + }, + { + "epoch": 19.76, + "learning_rate": 1.7058982683982684e-05, + "loss": 0.0527, + "step": 24350 + }, + { + "epoch": 19.81, + "learning_rate": 1.699134199134199e-05, + "loss": 0.0573, + "step": 24400 + }, + { + "epoch": 19.85, + "learning_rate": 1.69237012987013e-05, + "loss": 0.0662, + "step": 24450 + }, + { + "epoch": 19.89, + "learning_rate": 1.6856060606060605e-05, + "loss": 0.0595, + "step": 24500 + }, + { + "epoch": 19.93, + "learning_rate": 1.6788419913419916e-05, + "loss": 0.0667, + "step": 24550 + }, + { + "epoch": 19.97, + "learning_rate": 1.672077922077922e-05, + "loss": 0.0514, + "step": 24600 + }, + { + "epoch": 20.0, + "eval_cer": 0.2549377673122207, + "eval_loss": 1.7997801303863525, + "eval_mer": 0.3000958772770853, + "eval_runtime": 216.9181, + "eval_samples_per_second": 5.196, + "eval_steps_per_second": 0.65, + "step": 24640 + }, + { + "epoch": 20.01, + "learning_rate": 1.665313852813853e-05, + "loss": 0.064, + "step": 24650 + }, + { + "epoch": 20.05, + "learning_rate": 1.6585497835497837e-05, + "loss": 0.0531, + "step": 24700 + }, + { + "epoch": 20.09, + "learning_rate": 1.6517857142857144e-05, + "loss": 0.0659, + "step": 24750 + }, + { + "epoch": 20.13, + "learning_rate": 1.645021645021645e-05, + "loss": 0.0699, + "step": 24800 + }, + { + "epoch": 20.17, + "learning_rate": 1.6382575757575758e-05, + "loss": 0.073, + "step": 24850 + }, + { + "epoch": 20.21, + "learning_rate": 1.6314935064935065e-05, + "loss": 0.0483, + "step": 24900 + }, + { + "epoch": 20.25, + "learning_rate": 1.624729437229437e-05, + "loss": 0.0535, + "step": 24950 + }, + { + "epoch": 20.29, + "learning_rate": 1.6179653679653682e-05, + "loss": 0.0534, + "step": 25000 + }, + { + "epoch": 20.33, + "learning_rate": 1.6112012987012986e-05, + "loss": 0.0549, + "step": 25050 + }, + { + "epoch": 20.37, + "learning_rate": 1.6044372294372296e-05, + "loss": 0.0675, + "step": 25100 + }, + { + "epoch": 20.41, + "learning_rate": 1.5976731601731603e-05, + "loss": 0.0607, + "step": 25150 + }, + { + "epoch": 20.45, + "learning_rate": 1.590909090909091e-05, + "loss": 0.0469, + "step": 25200 + }, + { + "epoch": 20.5, + "learning_rate": 1.5841450216450217e-05, + "loss": 0.0484, + "step": 25250 + }, + { + "epoch": 20.54, + "learning_rate": 1.5773809523809524e-05, + "loss": 0.0506, + "step": 25300 + }, + { + "epoch": 20.58, + "learning_rate": 1.570616883116883e-05, + "loss": 0.0574, + "step": 25350 + }, + { + "epoch": 20.62, + "learning_rate": 1.5638528138528138e-05, + "loss": 0.0706, + "step": 25400 + }, + { + "epoch": 20.66, + "learning_rate": 1.557088744588745e-05, + "loss": 0.0419, + "step": 25450 + }, + { + "epoch": 20.7, + "learning_rate": 1.5503246753246752e-05, + "loss": 0.0583, + "step": 25500 + }, + { + "epoch": 20.74, + "learning_rate": 1.5435606060606063e-05, + "loss": 0.0766, + "step": 25550 + }, + { + "epoch": 20.78, + "learning_rate": 1.5367965367965366e-05, + "loss": 0.057, + "step": 25600 + }, + { + "epoch": 20.82, + "learning_rate": 1.5300324675324677e-05, + "loss": 0.0686, + "step": 25650 + }, + { + "epoch": 20.86, + "learning_rate": 1.5232683982683984e-05, + "loss": 0.0466, + "step": 25700 + }, + { + "epoch": 20.9, + "learning_rate": 1.516504329004329e-05, + "loss": 0.0614, + "step": 25750 + }, + { + "epoch": 20.94, + "learning_rate": 1.50974025974026e-05, + "loss": 0.0537, + "step": 25800 + }, + { + "epoch": 20.98, + "learning_rate": 1.5029761904761905e-05, + "loss": 0.0414, + "step": 25850 + }, + { + "epoch": 21.0, + "eval_cer": 0.25315969051852566, + "eval_loss": 1.8042283058166504, + "eval_mer": 0.2984979226589965, + "eval_runtime": 216.3821, + "eval_samples_per_second": 5.208, + "eval_steps_per_second": 0.652, + "step": 25872 + }, + { + "epoch": 21.02, + "learning_rate": 1.4962121212121214e-05, + "loss": 0.0628, + "step": 25900 + }, + { + "epoch": 21.06, + "learning_rate": 1.4894480519480519e-05, + "loss": 0.0408, + "step": 25950 + }, + { + "epoch": 21.1, + "learning_rate": 1.4826839826839828e-05, + "loss": 0.0497, + "step": 26000 + }, + { + "epoch": 21.14, + "learning_rate": 1.4759199134199133e-05, + "loss": 0.044, + "step": 26050 + }, + { + "epoch": 21.19, + "learning_rate": 1.4691558441558442e-05, + "loss": 0.0361, + "step": 26100 + }, + { + "epoch": 21.23, + "learning_rate": 1.462391774891775e-05, + "loss": 0.0498, + "step": 26150 + }, + { + "epoch": 21.27, + "learning_rate": 1.4556277056277057e-05, + "loss": 0.0485, + "step": 26200 + }, + { + "epoch": 21.31, + "learning_rate": 1.4488636363636366e-05, + "loss": 0.0438, + "step": 26250 + }, + { + "epoch": 21.35, + "learning_rate": 1.4420995670995671e-05, + "loss": 0.0749, + "step": 26300 + }, + { + "epoch": 21.39, + "learning_rate": 1.435335497835498e-05, + "loss": 0.0467, + "step": 26350 + }, + { + "epoch": 21.43, + "learning_rate": 1.4285714285714285e-05, + "loss": 0.0613, + "step": 26400 + }, + { + "epoch": 21.47, + "learning_rate": 1.4218073593073594e-05, + "loss": 0.0746, + "step": 26450 + }, + { + "epoch": 21.51, + "learning_rate": 1.41504329004329e-05, + "loss": 0.0516, + "step": 26500 + }, + { + "epoch": 21.55, + "learning_rate": 1.4082792207792208e-05, + "loss": 0.0478, + "step": 26550 + }, + { + "epoch": 21.59, + "learning_rate": 1.4015151515151515e-05, + "loss": 0.0412, + "step": 26600 + }, + { + "epoch": 21.63, + "learning_rate": 1.3947510822510824e-05, + "loss": 0.0495, + "step": 26650 + }, + { + "epoch": 21.67, + "learning_rate": 1.3879870129870131e-05, + "loss": 0.072, + "step": 26700 + }, + { + "epoch": 21.71, + "learning_rate": 1.3812229437229438e-05, + "loss": 0.054, + "step": 26750 + }, + { + "epoch": 21.75, + "learning_rate": 1.3744588744588747e-05, + "loss": 0.0664, + "step": 26800 + }, + { + "epoch": 21.79, + "learning_rate": 1.3676948051948052e-05, + "loss": 0.057, + "step": 26850 + }, + { + "epoch": 21.83, + "learning_rate": 1.360930735930736e-05, + "loss": 0.0503, + "step": 26900 + }, + { + "epoch": 21.88, + "learning_rate": 1.3541666666666666e-05, + "loss": 0.0741, + "step": 26950 + }, + { + "epoch": 21.92, + "learning_rate": 1.3474025974025975e-05, + "loss": 0.06, + "step": 27000 + }, + { + "epoch": 21.96, + "learning_rate": 1.340638528138528e-05, + "loss": 0.047, + "step": 27050 + }, + { + "epoch": 22.0, + "learning_rate": 1.3338744588744589e-05, + "loss": 0.0397, + "step": 27100 + }, + { + "epoch": 22.0, + "eval_cer": 0.256908068624153, + "eval_loss": 1.8340286016464233, + "eval_mer": 0.3004154682007031, + "eval_runtime": 216.2013, + "eval_samples_per_second": 5.213, + "eval_steps_per_second": 0.652, + "step": 27104 + }, + { + "epoch": 22.04, + "learning_rate": 1.3271103896103898e-05, + "loss": 0.055, + "step": 27150 + }, + { + "epoch": 22.08, + "learning_rate": 1.3203463203463205e-05, + "loss": 0.0488, + "step": 27200 + }, + { + "epoch": 22.12, + "learning_rate": 1.3135822510822513e-05, + "loss": 0.0452, + "step": 27250 + }, + { + "epoch": 22.16, + "learning_rate": 1.3068181818181819e-05, + "loss": 0.0625, + "step": 27300 + }, + { + "epoch": 22.2, + "learning_rate": 1.3000541125541127e-05, + "loss": 0.0555, + "step": 27350 + }, + { + "epoch": 22.24, + "learning_rate": 1.2932900432900433e-05, + "loss": 0.0482, + "step": 27400 + }, + { + "epoch": 22.28, + "learning_rate": 1.2865259740259741e-05, + "loss": 0.0447, + "step": 27450 + }, + { + "epoch": 22.32, + "learning_rate": 1.2797619047619047e-05, + "loss": 0.0539, + "step": 27500 + }, + { + "epoch": 22.36, + "learning_rate": 1.2729978354978355e-05, + "loss": 0.0593, + "step": 27550 + }, + { + "epoch": 22.4, + "learning_rate": 1.2662337662337662e-05, + "loss": 0.0418, + "step": 27600 + }, + { + "epoch": 22.44, + "learning_rate": 1.259469696969697e-05, + "loss": 0.0507, + "step": 27650 + }, + { + "epoch": 22.48, + "learning_rate": 1.2527056277056278e-05, + "loss": 0.0327, + "step": 27700 + }, + { + "epoch": 22.52, + "learning_rate": 1.2459415584415585e-05, + "loss": 0.0572, + "step": 27750 + }, + { + "epoch": 22.56, + "learning_rate": 1.2391774891774892e-05, + "loss": 0.07, + "step": 27800 + }, + { + "epoch": 22.61, + "learning_rate": 1.23241341991342e-05, + "loss": 0.0471, + "step": 27850 + }, + { + "epoch": 22.65, + "learning_rate": 1.2256493506493508e-05, + "loss": 0.059, + "step": 27900 + }, + { + "epoch": 22.69, + "learning_rate": 1.2188852813852815e-05, + "loss": 0.0594, + "step": 27950 + }, + { + "epoch": 22.73, + "learning_rate": 1.2121212121212122e-05, + "loss": 0.0368, + "step": 28000 + }, + { + "epoch": 22.77, + "learning_rate": 1.2053571428571429e-05, + "loss": 0.0663, + "step": 28050 + }, + { + "epoch": 22.81, + "learning_rate": 1.1985930735930736e-05, + "loss": 0.0508, + "step": 28100 + }, + { + "epoch": 22.85, + "learning_rate": 1.1918290043290043e-05, + "loss": 0.0621, + "step": 28150 + }, + { + "epoch": 22.89, + "learning_rate": 1.1850649350649352e-05, + "loss": 0.0455, + "step": 28200 + }, + { + "epoch": 22.93, + "learning_rate": 1.1783008658008659e-05, + "loss": 0.0606, + "step": 28250 + }, + { + "epoch": 22.97, + "learning_rate": 1.1715367965367966e-05, + "loss": 0.0303, + "step": 28300 + }, + { + "epoch": 23.0, + "eval_cer": 0.25364025181411887, + "eval_loss": 1.8699883222579956, + "eval_mer": 0.29841802492809205, + "eval_runtime": 216.7323, + "eval_samples_per_second": 5.2, + "eval_steps_per_second": 0.651, + "step": 28336 + }, + { + "epoch": 23.01, + "learning_rate": 1.1647727272727273e-05, + "loss": 0.0703, + "step": 28350 + }, + { + "epoch": 23.05, + "learning_rate": 1.1580086580086581e-05, + "loss": 0.0535, + "step": 28400 + }, + { + "epoch": 23.09, + "learning_rate": 1.1512445887445889e-05, + "loss": 0.0511, + "step": 28450 + }, + { + "epoch": 23.13, + "learning_rate": 1.1444805194805196e-05, + "loss": 0.0423, + "step": 28500 + }, + { + "epoch": 23.17, + "learning_rate": 1.1377164502164503e-05, + "loss": 0.0577, + "step": 28550 + }, + { + "epoch": 23.21, + "learning_rate": 1.130952380952381e-05, + "loss": 0.0388, + "step": 28600 + }, + { + "epoch": 23.25, + "learning_rate": 1.1241883116883117e-05, + "loss": 0.0426, + "step": 28650 + }, + { + "epoch": 23.3, + "learning_rate": 1.1174242424242425e-05, + "loss": 0.0597, + "step": 28700 + }, + { + "epoch": 23.34, + "learning_rate": 1.1106601731601732e-05, + "loss": 0.0409, + "step": 28750 + }, + { + "epoch": 23.38, + "learning_rate": 1.103896103896104e-05, + "loss": 0.0542, + "step": 28800 + }, + { + "epoch": 23.42, + "learning_rate": 1.0971320346320346e-05, + "loss": 0.0396, + "step": 28850 + }, + { + "epoch": 23.46, + "learning_rate": 1.0903679653679655e-05, + "loss": 0.0713, + "step": 28900 + }, + { + "epoch": 23.5, + "learning_rate": 1.0836038961038962e-05, + "loss": 0.0322, + "step": 28950 + }, + { + "epoch": 23.54, + "learning_rate": 1.0768398268398269e-05, + "loss": 0.0392, + "step": 29000 + }, + { + "epoch": 23.58, + "learning_rate": 1.0700757575757576e-05, + "loss": 0.0495, + "step": 29050 + }, + { + "epoch": 23.62, + "learning_rate": 1.0633116883116883e-05, + "loss": 0.0474, + "step": 29100 + }, + { + "epoch": 23.66, + "learning_rate": 1.056547619047619e-05, + "loss": 0.0647, + "step": 29150 + }, + { + "epoch": 23.7, + "learning_rate": 1.0497835497835499e-05, + "loss": 0.0423, + "step": 29200 + }, + { + "epoch": 23.74, + "learning_rate": 1.0430194805194806e-05, + "loss": 0.0563, + "step": 29250 + }, + { + "epoch": 23.78, + "learning_rate": 1.0362554112554113e-05, + "loss": 0.0581, + "step": 29300 + }, + { + "epoch": 23.82, + "learning_rate": 1.029491341991342e-05, + "loss": 0.0321, + "step": 29350 + }, + { + "epoch": 23.86, + "learning_rate": 1.0227272727272729e-05, + "loss": 0.0433, + "step": 29400 + }, + { + "epoch": 23.9, + "learning_rate": 1.0159632034632036e-05, + "loss": 0.0589, + "step": 29450 + }, + { + "epoch": 23.94, + "learning_rate": 1.0091991341991343e-05, + "loss": 0.0347, + "step": 29500 + }, + { + "epoch": 23.99, + "learning_rate": 1.002435064935065e-05, + "loss": 0.0552, + "step": 29550 + }, + { + "epoch": 24.0, + "eval_cer": 0.253207746648085, + "eval_loss": 1.9628108739852905, + "eval_mer": 0.29729945669542984, + "eval_runtime": 215.5186, + "eval_samples_per_second": 5.229, + "eval_steps_per_second": 0.654, + "step": 29568 + }, + { + "epoch": 24.03, + "learning_rate": 9.956709956709957e-06, + "loss": 0.0639, + "step": 29600 + }, + { + "epoch": 24.07, + "learning_rate": 9.889069264069264e-06, + "loss": 0.049, + "step": 29650 + }, + { + "epoch": 24.11, + "learning_rate": 9.821428571428573e-06, + "loss": 0.0438, + "step": 29700 + }, + { + "epoch": 24.15, + "learning_rate": 9.75378787878788e-06, + "loss": 0.0417, + "step": 29750 + }, + { + "epoch": 24.19, + "learning_rate": 9.686147186147187e-06, + "loss": 0.0402, + "step": 29800 + }, + { + "epoch": 24.23, + "learning_rate": 9.618506493506494e-06, + "loss": 0.0517, + "step": 29850 + }, + { + "epoch": 24.27, + "learning_rate": 9.550865800865802e-06, + "loss": 0.0439, + "step": 29900 + }, + { + "epoch": 24.31, + "learning_rate": 9.48322510822511e-06, + "loss": 0.0426, + "step": 29950 + }, + { + "epoch": 24.35, + "learning_rate": 9.415584415584416e-06, + "loss": 0.0567, + "step": 30000 + }, + { + "epoch": 24.39, + "learning_rate": 9.347943722943723e-06, + "loss": 0.0529, + "step": 30050 + }, + { + "epoch": 24.43, + "learning_rate": 9.28030303030303e-06, + "loss": 0.0479, + "step": 30100 + }, + { + "epoch": 24.47, + "learning_rate": 9.212662337662337e-06, + "loss": 0.0483, + "step": 30150 + }, + { + "epoch": 24.51, + "learning_rate": 9.145021645021646e-06, + "loss": 0.0385, + "step": 30200 + }, + { + "epoch": 24.55, + "learning_rate": 9.077380952380953e-06, + "loss": 0.048, + "step": 30250 + }, + { + "epoch": 24.59, + "learning_rate": 9.00974025974026e-06, + "loss": 0.0458, + "step": 30300 + }, + { + "epoch": 24.63, + "learning_rate": 8.942099567099567e-06, + "loss": 0.0519, + "step": 30350 + }, + { + "epoch": 24.68, + "learning_rate": 8.874458874458876e-06, + "loss": 0.0443, + "step": 30400 + }, + { + "epoch": 24.72, + "learning_rate": 8.806818181818183e-06, + "loss": 0.0636, + "step": 30450 + }, + { + "epoch": 24.76, + "learning_rate": 8.73917748917749e-06, + "loss": 0.0398, + "step": 30500 + }, + { + "epoch": 24.8, + "learning_rate": 8.671536796536797e-06, + "loss": 0.0587, + "step": 30550 + }, + { + "epoch": 24.84, + "learning_rate": 8.603896103896104e-06, + "loss": 0.05, + "step": 30600 + }, + { + "epoch": 24.88, + "learning_rate": 8.536255411255411e-06, + "loss": 0.0364, + "step": 30650 + }, + { + "epoch": 24.92, + "learning_rate": 8.468614718614718e-06, + "loss": 0.0454, + "step": 30700 + }, + { + "epoch": 24.96, + "learning_rate": 8.400974025974027e-06, + "loss": 0.0432, + "step": 30750 + }, + { + "epoch": 25.0, + "learning_rate": 8.333333333333334e-06, + "loss": 0.0401, + "step": 30800 + }, + { + "epoch": 25.0, + "eval_cer": 0.2528713537411697, + "eval_loss": 1.9063920974731445, + "eval_mer": 0.29825822946628316, + "eval_runtime": 216.7371, + "eval_samples_per_second": 5.2, + "eval_steps_per_second": 0.651, + "step": 30800 + }, + { + "epoch": 25.04, + "learning_rate": 8.26569264069264e-06, + "loss": 0.0462, + "step": 30850 + }, + { + "epoch": 25.08, + "learning_rate": 8.19805194805195e-06, + "loss": 0.0334, + "step": 30900 + }, + { + "epoch": 25.12, + "learning_rate": 8.130411255411256e-06, + "loss": 0.0404, + "step": 30950 + }, + { + "epoch": 25.16, + "learning_rate": 8.062770562770564e-06, + "loss": 0.0527, + "step": 31000 + }, + { + "epoch": 25.2, + "learning_rate": 7.99512987012987e-06, + "loss": 0.0458, + "step": 31050 + }, + { + "epoch": 25.24, + "learning_rate": 7.927489177489178e-06, + "loss": 0.0502, + "step": 31100 + }, + { + "epoch": 25.28, + "learning_rate": 7.859848484848485e-06, + "loss": 0.0313, + "step": 31150 + }, + { + "epoch": 25.32, + "learning_rate": 7.792207792207792e-06, + "loss": 0.0302, + "step": 31200 + }, + { + "epoch": 25.37, + "learning_rate": 7.7245670995671e-06, + "loss": 0.0432, + "step": 31250 + }, + { + "epoch": 25.41, + "learning_rate": 7.656926406926407e-06, + "loss": 0.0735, + "step": 31300 + }, + { + "epoch": 25.45, + "learning_rate": 7.589285714285714e-06, + "loss": 0.0408, + "step": 31350 + }, + { + "epoch": 25.49, + "learning_rate": 7.521645021645023e-06, + "loss": 0.0249, + "step": 31400 + }, + { + "epoch": 25.53, + "learning_rate": 7.45400432900433e-06, + "loss": 0.0527, + "step": 31450 + }, + { + "epoch": 25.57, + "learning_rate": 7.386363636363637e-06, + "loss": 0.0442, + "step": 31500 + }, + { + "epoch": 25.61, + "learning_rate": 7.318722943722944e-06, + "loss": 0.0394, + "step": 31550 + }, + { + "epoch": 25.65, + "learning_rate": 7.251082251082251e-06, + "loss": 0.0363, + "step": 31600 + }, + { + "epoch": 25.69, + "learning_rate": 7.183441558441559e-06, + "loss": 0.0359, + "step": 31650 + }, + { + "epoch": 25.73, + "learning_rate": 7.115800865800866e-06, + "loss": 0.0367, + "step": 31700 + }, + { + "epoch": 25.77, + "learning_rate": 7.048160173160173e-06, + "loss": 0.0451, + "step": 31750 + }, + { + "epoch": 25.81, + "learning_rate": 6.98051948051948e-06, + "loss": 0.0373, + "step": 31800 + }, + { + "epoch": 25.85, + "learning_rate": 6.912878787878788e-06, + "loss": 0.0452, + "step": 31850 + }, + { + "epoch": 25.89, + "learning_rate": 6.845238095238096e-06, + "loss": 0.046, + "step": 31900 + }, + { + "epoch": 25.93, + "learning_rate": 6.777597402597404e-06, + "loss": 0.0466, + "step": 31950 + }, + { + "epoch": 25.97, + "learning_rate": 6.709956709956711e-06, + "loss": 0.0417, + "step": 32000 + }, + { + "epoch": 26.0, + "eval_cer": 0.25364025181411887, + "eval_loss": 1.9402902126312256, + "eval_mer": 0.2979386385426654, + "eval_runtime": 217.1614, + "eval_samples_per_second": 5.19, + "eval_steps_per_second": 0.649, + "step": 32032 + }, + { + "epoch": 26.01, + "learning_rate": 6.642316017316018e-06, + "loss": 0.0395, + "step": 32050 + }, + { + "epoch": 26.06, + "learning_rate": 6.574675324675325e-06, + "loss": 0.046, + "step": 32100 + }, + { + "epoch": 26.1, + "learning_rate": 6.507034632034633e-06, + "loss": 0.0412, + "step": 32150 + }, + { + "epoch": 26.14, + "learning_rate": 6.43939393939394e-06, + "loss": 0.0322, + "step": 32200 + }, + { + "epoch": 26.18, + "learning_rate": 6.371753246753247e-06, + "loss": 0.0337, + "step": 32250 + }, + { + "epoch": 26.22, + "learning_rate": 6.304112554112554e-06, + "loss": 0.0366, + "step": 32300 + }, + { + "epoch": 26.26, + "learning_rate": 6.2364718614718615e-06, + "loss": 0.035, + "step": 32350 + }, + { + "epoch": 26.3, + "learning_rate": 6.168831168831169e-06, + "loss": 0.048, + "step": 32400 + }, + { + "epoch": 26.34, + "learning_rate": 6.101190476190476e-06, + "loss": 0.0188, + "step": 32450 + }, + { + "epoch": 26.38, + "learning_rate": 6.0335497835497834e-06, + "loss": 0.0469, + "step": 32500 + }, + { + "epoch": 26.42, + "learning_rate": 5.965909090909091e-06, + "loss": 0.03, + "step": 32550 + }, + { + "epoch": 26.46, + "learning_rate": 5.898268398268398e-06, + "loss": 0.0509, + "step": 32600 + }, + { + "epoch": 26.5, + "learning_rate": 5.830627705627706e-06, + "loss": 0.053, + "step": 32650 + }, + { + "epoch": 26.54, + "learning_rate": 5.762987012987013e-06, + "loss": 0.0413, + "step": 32700 + }, + { + "epoch": 26.58, + "learning_rate": 5.69534632034632e-06, + "loss": 0.0343, + "step": 32750 + }, + { + "epoch": 26.62, + "learning_rate": 5.627705627705628e-06, + "loss": 0.046, + "step": 32800 + }, + { + "epoch": 26.66, + "learning_rate": 5.560064935064935e-06, + "loss": 0.0529, + "step": 32850 + }, + { + "epoch": 26.7, + "learning_rate": 5.492424242424243e-06, + "loss": 0.0425, + "step": 32900 + }, + { + "epoch": 26.75, + "learning_rate": 5.42478354978355e-06, + "loss": 0.0322, + "step": 32950 + }, + { + "epoch": 26.79, + "learning_rate": 5.357142857142857e-06, + "loss": 0.044, + "step": 33000 + }, + { + "epoch": 26.83, + "learning_rate": 5.289502164502165e-06, + "loss": 0.0329, + "step": 33050 + }, + { + "epoch": 26.87, + "learning_rate": 5.221861471861472e-06, + "loss": 0.0385, + "step": 33100 + }, + { + "epoch": 26.91, + "learning_rate": 5.15422077922078e-06, + "loss": 0.0436, + "step": 33150 + }, + { + "epoch": 26.95, + "learning_rate": 5.086580086580087e-06, + "loss": 0.0524, + "step": 33200 + }, + { + "epoch": 26.99, + "learning_rate": 5.018939393939394e-06, + "loss": 0.0301, + "step": 33250 + }, + { + "epoch": 27.0, + "eval_cer": 0.2514296698543899, + "eval_loss": 1.915743350982666, + "eval_mer": 0.29658037711728985, + "eval_runtime": 215.5246, + "eval_samples_per_second": 5.229, + "eval_steps_per_second": 0.654, + "step": 33264 + }, + { + "epoch": 27.03, + "learning_rate": 4.951298701298702e-06, + "loss": 0.0274, + "step": 33300 + }, + { + "epoch": 27.07, + "learning_rate": 4.883658008658009e-06, + "loss": 0.0548, + "step": 33350 + }, + { + "epoch": 27.11, + "learning_rate": 4.816017316017317e-06, + "loss": 0.0354, + "step": 33400 + }, + { + "epoch": 27.15, + "learning_rate": 4.748376623376624e-06, + "loss": 0.0333, + "step": 33450 + }, + { + "epoch": 27.19, + "learning_rate": 4.680735930735931e-06, + "loss": 0.0361, + "step": 33500 + }, + { + "epoch": 27.23, + "learning_rate": 4.6130952380952385e-06, + "loss": 0.0292, + "step": 33550 + }, + { + "epoch": 27.27, + "learning_rate": 4.5454545454545455e-06, + "loss": 0.0312, + "step": 33600 + }, + { + "epoch": 27.31, + "learning_rate": 4.477813852813853e-06, + "loss": 0.0365, + "step": 33650 + }, + { + "epoch": 27.35, + "learning_rate": 4.41017316017316e-06, + "loss": 0.0383, + "step": 33700 + }, + { + "epoch": 27.39, + "learning_rate": 4.3425324675324674e-06, + "loss": 0.0331, + "step": 33750 + }, + { + "epoch": 27.44, + "learning_rate": 4.274891774891775e-06, + "loss": 0.024, + "step": 33800 + }, + { + "epoch": 27.48, + "learning_rate": 4.207251082251082e-06, + "loss": 0.0417, + "step": 33850 + }, + { + "epoch": 27.52, + "learning_rate": 4.13961038961039e-06, + "loss": 0.0447, + "step": 33900 + }, + { + "epoch": 27.56, + "learning_rate": 4.071969696969697e-06, + "loss": 0.0436, + "step": 33950 + }, + { + "epoch": 27.6, + "learning_rate": 4.004329004329004e-06, + "loss": 0.0597, + "step": 34000 + }, + { + "epoch": 27.64, + "learning_rate": 3.936688311688312e-06, + "loss": 0.0297, + "step": 34050 + }, + { + "epoch": 27.68, + "learning_rate": 3.869047619047619e-06, + "loss": 0.0383, + "step": 34100 + }, + { + "epoch": 27.72, + "learning_rate": 3.8014069264069266e-06, + "loss": 0.0443, + "step": 34150 + }, + { + "epoch": 27.76, + "learning_rate": 3.733766233766234e-06, + "loss": 0.0339, + "step": 34200 + }, + { + "epoch": 27.8, + "learning_rate": 3.666125541125541e-06, + "loss": 0.0338, + "step": 34250 + }, + { + "epoch": 27.84, + "learning_rate": 3.598484848484849e-06, + "loss": 0.0271, + "step": 34300 + }, + { + "epoch": 27.88, + "learning_rate": 3.5308441558441563e-06, + "loss": 0.0296, + "step": 34350 + }, + { + "epoch": 27.92, + "learning_rate": 3.4632034632034634e-06, + "loss": 0.0314, + "step": 34400 + }, + { + "epoch": 27.96, + "learning_rate": 3.3955627705627704e-06, + "loss": 0.0267, + "step": 34450 + }, + { + "epoch": 28.0, + "eval_cer": 0.2515738382430679, + "eval_loss": 1.982408881187439, + "eval_mer": 0.2975391498881432, + "eval_runtime": 217.1186, + "eval_samples_per_second": 5.191, + "eval_steps_per_second": 0.649, + "step": 34496 + }, + { + "epoch": 28.0, + "learning_rate": 3.327922077922078e-06, + "loss": 0.0364, + "step": 34500 + }, + { + "epoch": 28.04, + "learning_rate": 3.2602813852813857e-06, + "loss": 0.0339, + "step": 34550 + }, + { + "epoch": 28.08, + "learning_rate": 3.192640692640693e-06, + "loss": 0.0314, + "step": 34600 + }, + { + "epoch": 28.12, + "learning_rate": 3.125e-06, + "loss": 0.0226, + "step": 34650 + }, + { + "epoch": 28.17, + "learning_rate": 3.057359307359307e-06, + "loss": 0.0422, + "step": 34700 + }, + { + "epoch": 28.21, + "learning_rate": 2.989718614718615e-06, + "loss": 0.0367, + "step": 34750 + }, + { + "epoch": 28.25, + "learning_rate": 2.922077922077922e-06, + "loss": 0.0462, + "step": 34800 + }, + { + "epoch": 28.29, + "learning_rate": 2.85443722943723e-06, + "loss": 0.0377, + "step": 34850 + }, + { + "epoch": 28.33, + "learning_rate": 2.786796536796537e-06, + "loss": 0.0295, + "step": 34900 + }, + { + "epoch": 28.37, + "learning_rate": 2.719155844155844e-06, + "loss": 0.0313, + "step": 34950 + }, + { + "epoch": 28.41, + "learning_rate": 2.651515151515152e-06, + "loss": 0.0341, + "step": 35000 + }, + { + "epoch": 28.45, + "learning_rate": 2.583874458874459e-06, + "loss": 0.0451, + "step": 35050 + }, + { + "epoch": 28.49, + "learning_rate": 2.5162337662337663e-06, + "loss": 0.0432, + "step": 35100 + }, + { + "epoch": 28.53, + "learning_rate": 2.4485930735930737e-06, + "loss": 0.0531, + "step": 35150 + }, + { + "epoch": 28.57, + "learning_rate": 2.3809523809523808e-06, + "loss": 0.0283, + "step": 35200 + }, + { + "epoch": 28.61, + "learning_rate": 2.3133116883116886e-06, + "loss": 0.0256, + "step": 35250 + }, + { + "epoch": 28.65, + "learning_rate": 2.2456709956709957e-06, + "loss": 0.0289, + "step": 35300 + }, + { + "epoch": 28.69, + "learning_rate": 2.178030303030303e-06, + "loss": 0.0277, + "step": 35350 + }, + { + "epoch": 28.73, + "learning_rate": 2.1103896103896105e-06, + "loss": 0.0239, + "step": 35400 + }, + { + "epoch": 28.77, + "learning_rate": 2.0427489177489176e-06, + "loss": 0.0483, + "step": 35450 + }, + { + "epoch": 28.81, + "learning_rate": 1.9751082251082254e-06, + "loss": 0.0449, + "step": 35500 + }, + { + "epoch": 28.86, + "learning_rate": 1.9074675324675324e-06, + "loss": 0.0276, + "step": 35550 + }, + { + "epoch": 28.9, + "learning_rate": 1.83982683982684e-06, + "loss": 0.0276, + "step": 35600 + }, + { + "epoch": 28.94, + "learning_rate": 1.7721861471861473e-06, + "loss": 0.0466, + "step": 35650 + }, + { + "epoch": 28.98, + "learning_rate": 1.7045454545454546e-06, + "loss": 0.0448, + "step": 35700 + }, + { + "epoch": 29.0, + "eval_cer": 0.25046854726320344, + "eval_loss": 1.9711192846298218, + "eval_mer": 0.2958612975391499, + "eval_runtime": 217.3139, + "eval_samples_per_second": 5.186, + "eval_steps_per_second": 0.649, + "step": 35728 + }, + { + "epoch": 29.02, + "learning_rate": 1.636904761904762e-06, + "loss": 0.0394, + "step": 35750 + }, + { + "epoch": 29.06, + "learning_rate": 1.5692640692640692e-06, + "loss": 0.0343, + "step": 35800 + }, + { + "epoch": 29.1, + "learning_rate": 1.5016233766233767e-06, + "loss": 0.0333, + "step": 35850 + }, + { + "epoch": 29.14, + "learning_rate": 1.4339826839826841e-06, + "loss": 0.0462, + "step": 35900 + }, + { + "epoch": 29.18, + "learning_rate": 1.3663419913419914e-06, + "loss": 0.0377, + "step": 35950 + }, + { + "epoch": 29.22, + "learning_rate": 1.2987012987012988e-06, + "loss": 0.0233, + "step": 36000 + }, + { + "epoch": 29.26, + "learning_rate": 1.231060606060606e-06, + "loss": 0.031, + "step": 36050 + }, + { + "epoch": 29.3, + "learning_rate": 1.1634199134199135e-06, + "loss": 0.0302, + "step": 36100 + }, + { + "epoch": 29.34, + "learning_rate": 1.095779220779221e-06, + "loss": 0.0397, + "step": 36150 + }, + { + "epoch": 29.38, + "learning_rate": 1.0281385281385282e-06, + "loss": 0.0266, + "step": 36200 + }, + { + "epoch": 29.42, + "learning_rate": 9.604978354978354e-07, + "loss": 0.0242, + "step": 36250 + }, + { + "epoch": 29.46, + "learning_rate": 8.928571428571428e-07, + "loss": 0.0339, + "step": 36300 + }, + { + "epoch": 29.5, + "learning_rate": 8.252164502164503e-07, + "loss": 0.0364, + "step": 36350 + }, + { + "epoch": 29.55, + "learning_rate": 7.575757575757576e-07, + "loss": 0.0485, + "step": 36400 + }, + { + "epoch": 29.59, + "learning_rate": 6.89935064935065e-07, + "loss": 0.0299, + "step": 36450 + }, + { + "epoch": 29.63, + "learning_rate": 6.222943722943723e-07, + "loss": 0.0276, + "step": 36500 + }, + { + "epoch": 29.67, + "learning_rate": 5.546536796536797e-07, + "loss": 0.04, + "step": 36550 + }, + { + "epoch": 29.71, + "learning_rate": 4.87012987012987e-07, + "loss": 0.0426, + "step": 36600 + }, + { + "epoch": 29.75, + "learning_rate": 4.193722943722944e-07, + "loss": 0.0266, + "step": 36650 + }, + { + "epoch": 29.79, + "learning_rate": 3.5173160173160176e-07, + "loss": 0.0222, + "step": 36700 + }, + { + "epoch": 29.83, + "learning_rate": 2.840909090909091e-07, + "loss": 0.0347, + "step": 36750 + }, + { + "epoch": 29.87, + "learning_rate": 2.1645021645021646e-07, + "loss": 0.041, + "step": 36800 + }, + { + "epoch": 29.91, + "learning_rate": 1.488095238095238e-07, + "loss": 0.034, + "step": 36850 + }, + { + "epoch": 29.95, + "learning_rate": 8.116883116883118e-08, + "loss": 0.043, + "step": 36900 + }, + { + "epoch": 29.99, + "learning_rate": 1.3528138528138529e-08, + "loss": 0.0336, + "step": 36950 + }, + { + "epoch": 30.0, + "eval_cer": 0.2511893892065933, + "eval_loss": 1.9807147979736328, + "eval_mer": 0.29674017257909874, + "eval_runtime": 216.3415, + "eval_samples_per_second": 5.209, + "eval_steps_per_second": 0.652, + "step": 36960 + }, + { + "epoch": 30.0, + "step": 36960, + "total_flos": 3.3776644953273782e+19, + "train_loss": 0.25857088244661236, + "train_runtime": 51050.909, + "train_samples_per_second": 5.792, + "train_steps_per_second": 0.724 + } + ], + "max_steps": 36960, + "num_train_epochs": 30, + "total_flos": 3.3776644953273782e+19, + "trial_name": null, + "trial_params": null +}